475 files changed, 271278 insertions, 0 deletions

diff --git a/drivers/net/ethernet/mellanox/Kconfig b/drivers/net/ethernet/mellanox/Kconfig
new file mode 100644
index 000000000..b4f66eb9d
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/Kconfig
@@ -0,0 +1,27 @@
+# SPDX-License-Identifier: GPL-2.0-only
+#
+# Mellanox driver configuration
+#
+
+config NET_VENDOR_MELLANOX
+	bool "Mellanox devices"
+	default y
+	depends on PCI || I2C
+	help
+	  If you have a network (Ethernet or RDMA) device belonging to this
+	  class, say Y.
+
+	  Note that the answer to this question doesn't directly affect the
+	  kernel: saying N will just cause the configurator to skip all
+	  the questions about Mellanox cards. If you say Y, you will be asked
+	  for your specific card in the following questions.
+
+if NET_VENDOR_MELLANOX
+
+source "drivers/net/ethernet/mellanox/mlx4/Kconfig"
+source "drivers/net/ethernet/mellanox/mlx5/core/Kconfig"
+source "drivers/net/ethernet/mellanox/mlxsw/Kconfig"
+source "drivers/net/ethernet/mellanox/mlxfw/Kconfig"
+source "drivers/net/ethernet/mellanox/mlxbf_gige/Kconfig"
+
+endif # NET_VENDOR_MELLANOX
diff --git a/drivers/net/ethernet/mellanox/Makefile b/drivers/net/ethernet/mellanox/Makefile
new file mode 100644
index 000000000..d4b5f547a
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/Makefile
@@ -0,0 +1,10 @@
+# SPDX-License-Identifier: GPL-2.0-only
+#
+# Makefile for the Mellanox device drivers.
+#
+
+obj-$(CONFIG_MLX4_CORE) += mlx4/
+obj-$(CONFIG_MLX5_CORE) += mlx5/core/
+obj-$(CONFIG_MLXSW_CORE) += mlxsw/
+obj-$(CONFIG_MLXFW) += mlxfw/
+obj-$(CONFIG_MLXBF_GIGE) += mlxbf_gige/
diff --git a/drivers/net/ethernet/mellanox/mlx4/Kconfig b/drivers/net/ethernet/mellanox/mlx4/Kconfig
new file mode 100644
index 000000000..1b4b1f642
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx4/Kconfig
@@ -0,0 +1,49 @@
+# SPDX-License-Identifier: GPL-2.0-only
+#
+# Mellanox driver configuration
+#
+
+config MLX4_EN
+	tristate "Mellanox Technologies 1/10/40Gbit Ethernet support"
+	depends on PCI && NETDEVICES && ETHERNET && INET
+	depends on PTP_1588_CLOCK_OPTIONAL
+	select MLX4_CORE
+	help
+	  This driver supports Mellanox Technologies ConnectX Ethernet
+	  devices.
+
+config MLX4_EN_DCB
+	bool "Data Center Bridging (DCB) Support"
+	default y
+	depends on MLX4_EN && DCB
+	help
+	  Say Y here if you want to use Data Center Bridging (DCB) in the
+	  driver.
+	  If set to N, will not be able to configure QoS and ratelimit attributes.
+	  This flag is depended on the kernel's DCB support.
+
+	  If unsure, set to Y
+
+config MLX4_CORE
+	tristate
+	depends on PCI
+	select NET_DEVLINK
+	default n
+
+config MLX4_DEBUG
+	bool "Verbose debugging output" if (MLX4_CORE && EXPERT)
+	depends on MLX4_CORE
+	default y
+	help
+	  This option causes debugging code to be compiled into the
+	  mlx4_core driver.  The output can be turned on via the
+	  debug_level module parameter (which can also be set after
+	  the driver is loaded through sysfs).
+
+config MLX4_CORE_GEN2
+	bool "Support for old gen2 Mellanox PCI IDs" if (MLX4_CORE)
+	depends on MLX4_CORE
+	default y
+	help
+	  Say Y here if you want to use old gen2 Mellanox devices in the
+	  driver.
diff --git a/drivers/net/ethernet/mellanox/mlx4/Makefile b/drivers/net/ethernet/mellanox/mlx4/Makefile
new file mode 100644
index 000000000..3f400770f
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx4/Makefile
@@ -0,0 +1,12 @@
+# SPDX-License-Identifier: GPL-2.0
+obj-$(CONFIG_MLX4_CORE)		+= mlx4_core.o
+
+mlx4_core-y :=	alloc.o catas.o cmd.o cq.o eq.o fw.o fw_qos.o icm.o intf.o \
+		main.o mcg.o mr.o pd.o port.o profile.o qp.o reset.o sense.o \
+		srq.o resource_tracker.o crdump.o
+
+obj-$(CONFIG_MLX4_EN)               += mlx4_en.o
+
+mlx4_en-y := 	en_main.o en_tx.o en_rx.o en_ethtool.o en_port.o en_cq.o \
+		en_resources.o en_netdev.o en_selftest.o en_clock.o
+mlx4_en-$(CONFIG_MLX4_EN_DCB) += en_dcb_nl.o
diff --git a/drivers/net/ethernet/mellanox/mlx4/alloc.c b/drivers/net/ethernet/mellanox/mlx4/alloc.c
new file mode 100644
index 000000000..b330020dc
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx4/alloc.c
@@ -0,0 +1,818 @@
+/*
+ * Copyright (c) 2006, 2007 Cisco Systems, Inc.  All rights reserved.
+ * Copyright (c) 2007, 2008 Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/errno.h>
+#include <linux/slab.h>
+#include <linux/mm.h>
+#include <linux/export.h>
+#include <linux/bitmap.h>
+#include <linux/dma-mapping.h>
+#include <linux/vmalloc.h>
+
+#include "mlx4.h"
+
+u32 mlx4_bitmap_alloc(struct mlx4_bitmap *bitmap)
+{
+	u32 obj;
+
+	spin_lock(&bitmap->lock);
+
+	obj = find_next_zero_bit(bitmap->table, bitmap->max, bitmap->last);
+	if (obj >= bitmap->max) {
+		bitmap->top = (bitmap->top + bitmap->max + bitmap->reserved_top)
+				& bitmap->mask;
+		obj = find_first_zero_bit(bitmap->table, bitmap->max);
+	}
+
+	if (obj < bitmap->max) {
+		set_bit(obj, bitmap->table);
+		bitmap->last = (obj + 1);
+		if (bitmap->last == bitmap->max)
+			bitmap->last = 0;
+		obj |= bitmap->top;
+	} else
+		obj = -1;
+
+	if (obj != -1)
+		--bitmap->avail;
+
+	spin_unlock(&bitmap->lock);
+
+	return obj;
+}
+
+void mlx4_bitmap_free(struct mlx4_bitmap *bitmap, u32 obj, int use_rr)
+{
+	mlx4_bitmap_free_range(bitmap, obj, 1, use_rr);
+}
+
+static unsigned long find_aligned_range(unsigned long *bitmap,
+					u32 start, u32 nbits,
+					int len, int align, u32 skip_mask)
+{
+	unsigned long end, i;
+
+again:
+	start = ALIGN(start, align);
+
+	while ((start < nbits) && (test_bit(start, bitmap) ||
+				   (start & skip_mask)))
+		start += align;
+
+	if (start >= nbits)
+		return -1;
+
+	end = start+len;
+	if (end > nbits)
+		return -1;
+
+	for (i = start + 1; i < end; i++) {
+		if (test_bit(i, bitmap) || ((u32)i & skip_mask)) {
+			start = i + 1;
+			goto again;
+		}
+	}
+
+	return start;
+}
+
+u32 mlx4_bitmap_alloc_range(struct mlx4_bitmap *bitmap, int cnt,
+			    int align, u32 skip_mask)
+{
+	u32 obj;
+
+	if (likely(cnt == 1 && align == 1 && !skip_mask))
+		return mlx4_bitmap_alloc(bitmap);
+
+	spin_lock(&bitmap->lock);
+
+	obj = find_aligned_range(bitmap->table, bitmap->last,
+				 bitmap->max, cnt, align, skip_mask);
+	if (obj >= bitmap->max) {
+		bitmap->top = (bitmap->top + bitmap->max + bitmap->reserved_top)
+				& bitmap->mask;
+		obj = find_aligned_range(bitmap->table, 0, bitmap->max,
+					 cnt, align, skip_mask);
+	}
+
+	if (obj < bitmap->max) {
+		bitmap_set(bitmap->table, obj, cnt);
+		if (obj == bitmap->last) {
+			bitmap->last = (obj + cnt);
+			if (bitmap->last >= bitmap->max)
+				bitmap->last = 0;
+		}
+		obj |= bitmap->top;
+	} else
+		obj = -1;
+
+	if (obj != -1)
+		bitmap->avail -= cnt;
+
+	spin_unlock(&bitmap->lock);
+
+	return obj;
+}
+
+u32 mlx4_bitmap_avail(struct mlx4_bitmap *bitmap)
+{
+	return bitmap->avail;
+}
+
+static u32 mlx4_bitmap_masked_value(struct mlx4_bitmap *bitmap, u32 obj)
+{
+	return obj & (bitmap->max + bitmap->reserved_top - 1);
+}
+
+void mlx4_bitmap_free_range(struct mlx4_bitmap *bitmap, u32 obj, int cnt,
+			    int use_rr)
+{
+	obj &= bitmap->max + bitmap->reserved_top - 1;
+
+	spin_lock(&bitmap->lock);
+	if (!use_rr) {
+		bitmap->last = min(bitmap->last, obj);
+		bitmap->top = (bitmap->top + bitmap->max + bitmap->reserved_top)
+				& bitmap->mask;
+	}
+	bitmap_clear(bitmap->table, obj, cnt);
+	bitmap->avail += cnt;
+	spin_unlock(&bitmap->lock);
+}
+
+int mlx4_bitmap_init(struct mlx4_bitmap *bitmap, u32 num, u32 mask,
+		     u32 reserved_bot, u32 reserved_top)
+{
+	/* num must be a power of 2 */
+	if (num != roundup_pow_of_two(num))
+		return -EINVAL;
+
+	bitmap->last = 0;
+	bitmap->top  = 0;
+	bitmap->max  = num - reserved_top;
+	bitmap->mask = mask;
+	bitmap->reserved_top = reserved_top;
+	bitmap->avail = num - reserved_top - reserved_bot;
+	bitmap->effective_len = bitmap->avail;
+	spin_lock_init(&bitmap->lock);
+	bitmap->table = bitmap_zalloc(bitmap->max, GFP_KERNEL);
+	if (!bitmap->table)
+		return -ENOMEM;
+
+	bitmap_set(bitmap->table, 0, reserved_bot);
+
+	return 0;
+}
+
+void mlx4_bitmap_cleanup(struct mlx4_bitmap *bitmap)
+{
+	bitmap_free(bitmap->table);
+}
+
+struct mlx4_zone_allocator {
+	struct list_head		entries;
+	struct list_head		prios;
+	u32				last_uid;
+	u32				mask;
+	/* protect the zone_allocator from concurrent accesses */
+	spinlock_t			lock;
+	enum mlx4_zone_alloc_flags	flags;
+};
+
+struct mlx4_zone_entry {
+	struct list_head		list;
+	struct list_head		prio_list;
+	u32				uid;
+	struct mlx4_zone_allocator	*allocator;
+	struct mlx4_bitmap		*bitmap;
+	int				use_rr;
+	int				priority;
+	int				offset;
+	enum mlx4_zone_flags		flags;
+};
+
+struct mlx4_zone_allocator *mlx4_zone_allocator_create(enum mlx4_zone_alloc_flags flags)
+{
+	struct mlx4_zone_allocator *zones = kmalloc(sizeof(*zones), GFP_KERNEL);
+
+	if (NULL == zones)
+		return NULL;
+
+	INIT_LIST_HEAD(&zones->entries);
+	INIT_LIST_HEAD(&zones->prios);
+	spin_lock_init(&zones->lock);
+	zones->last_uid = 0;
+	zones->mask = 0;
+	zones->flags = flags;
+
+	return zones;
+}
+
+int mlx4_zone_add_one(struct mlx4_zone_allocator *zone_alloc,
+		      struct mlx4_bitmap *bitmap,
+		      u32 flags,
+		      int priority,
+		      int offset,
+		      u32 *puid)
+{
+	u32 mask = mlx4_bitmap_masked_value(bitmap, (u32)-1);
+	struct mlx4_zone_entry *it;
+	struct mlx4_zone_entry *zone = kmalloc(sizeof(*zone), GFP_KERNEL);
+
+	if (NULL == zone)
+		return -ENOMEM;
+
+	zone->flags = flags;
+	zone->bitmap = bitmap;
+	zone->use_rr = (flags & MLX4_ZONE_USE_RR) ? MLX4_USE_RR : 0;
+	zone->priority = priority;
+	zone->offset = offset;
+
+	spin_lock(&zone_alloc->lock);
+
+	zone->uid = zone_alloc->last_uid++;
+	zone->allocator = zone_alloc;
+
+	if (zone_alloc->mask < mask)
+		zone_alloc->mask = mask;
+
+	list_for_each_entry(it, &zone_alloc->prios, prio_list)
+		if (it->priority >= priority)
+			break;
+
+	if (&it->prio_list == &zone_alloc->prios || it->priority > priority)
+		list_add_tail(&zone->prio_list, &it->prio_list);
+	list_add_tail(&zone->list, &it->list);
+
+	spin_unlock(&zone_alloc->lock);
+
+	*puid = zone->uid;
+
+	return 0;
+}
+
+/* Should be called under a lock */
+static void __mlx4_zone_remove_one_entry(struct mlx4_zone_entry *entry)
+{
+	struct mlx4_zone_allocator *zone_alloc = entry->allocator;
+
+	if (!list_empty(&entry->prio_list)) {
+		/* Check if we need to add an alternative node to the prio list */
+		if (!list_is_last(&entry->list, &zone_alloc->entries)) {
+			struct mlx4_zone_entry *next = list_first_entry(&entry->list,
+									typeof(*next),
+									list);
+
+			if (next->priority == entry->priority)
+				list_add_tail(&next->prio_list, &entry->prio_list);
+		}
+
+		list_del(&entry->prio_list);
+	}
+
+	list_del(&entry->list);
+
+	if (zone_alloc->flags & MLX4_ZONE_ALLOC_FLAGS_NO_OVERLAP) {
+		u32 mask = 0;
+		struct mlx4_zone_entry *it;
+
+		list_for_each_entry(it, &zone_alloc->prios, prio_list) {
+			u32 cur_mask = mlx4_bitmap_masked_value(it->bitmap, (u32)-1);
+
+			if (mask < cur_mask)
+				mask = cur_mask;
+		}
+		zone_alloc->mask = mask;
+	}
+}
+
+void mlx4_zone_allocator_destroy(struct mlx4_zone_allocator *zone_alloc)
+{
+	struct mlx4_zone_entry *zone, *tmp;
+
+	spin_lock(&zone_alloc->lock);
+
+	list_for_each_entry_safe(zone, tmp, &zone_alloc->entries, list) {
+		list_del(&zone->list);
+		list_del(&zone->prio_list);
+		kfree(zone);
+	}
+
+	spin_unlock(&zone_alloc->lock);
+	kfree(zone_alloc);
+}
+
+/* Should be called under a lock */
+static u32 __mlx4_alloc_from_zone(struct mlx4_zone_entry *zone, int count,
+				  int align, u32 skip_mask, u32 *puid)
+{
+	u32 uid = 0;
+	u32 res;
+	struct mlx4_zone_allocator *zone_alloc = zone->allocator;
+	struct mlx4_zone_entry *curr_node;
+
+	res = mlx4_bitmap_alloc_range(zone->bitmap, count,
+				      align, skip_mask);
+
+	if (res != (u32)-1) {
+		res += zone->offset;
+		uid = zone->uid;
+		goto out;
+	}
+
+	list_for_each_entry(curr_node, &zone_alloc->prios, prio_list) {
+		if (unlikely(curr_node->priority == zone->priority))
+			break;
+	}
+
+	if (zone->flags & MLX4_ZONE_ALLOW_ALLOC_FROM_LOWER_PRIO) {
+		struct mlx4_zone_entry *it = curr_node;
+
+		list_for_each_entry_continue_reverse(it, &zone_alloc->entries, list) {
+			res = mlx4_bitmap_alloc_range(it->bitmap, count,
+						      align, skip_mask);
+			if (res != (u32)-1) {
+				res += it->offset;
+				uid = it->uid;
+				goto out;
+			}
+		}
+	}
+
+	if (zone->flags & MLX4_ZONE_ALLOW_ALLOC_FROM_EQ_PRIO) {
+		struct mlx4_zone_entry *it = curr_node;
+
+		list_for_each_entry_from(it, &zone_alloc->entries, list) {
+			if (unlikely(it == zone))
+				continue;
+
+			if (unlikely(it->priority != curr_node->priority))
+				break;
+
+			res = mlx4_bitmap_alloc_range(it->bitmap, count,
+						      align, skip_mask);
+			if (res != (u32)-1) {
+				res += it->offset;
+				uid = it->uid;
+				goto out;
+			}
+		}
+	}
+
+	if (zone->flags & MLX4_ZONE_FALLBACK_TO_HIGHER_PRIO) {
+		if (list_is_last(&curr_node->prio_list, &zone_alloc->prios))
+			goto out;
+
+		curr_node = list_first_entry(&curr_node->prio_list,
+					     typeof(*curr_node),
+					     prio_list);
+
+		list_for_each_entry_from(curr_node, &zone_alloc->entries, list) {
+			res = mlx4_bitmap_alloc_range(curr_node->bitmap, count,
+						      align, skip_mask);
+			if (res != (u32)-1) {
+				res += curr_node->offset;
+				uid = curr_node->uid;
+				goto out;
+			}
+		}
+	}
+
+out:
+	if (NULL != puid && res != (u32)-1)
+		*puid = uid;
+	return res;
+}
+
+/* Should be called under a lock */
+static void __mlx4_free_from_zone(struct mlx4_zone_entry *zone, u32 obj,
+				  u32 count)
+{
+	mlx4_bitmap_free_range(zone->bitmap, obj - zone->offset, count, zone->use_rr);
+}
+
+/* Should be called under a lock */
+static struct mlx4_zone_entry *__mlx4_find_zone_by_uid(
+		struct mlx4_zone_allocator *zones, u32 uid)
+{
+	struct mlx4_zone_entry *zone;
+
+	list_for_each_entry(zone, &zones->entries, list) {
+		if (zone->uid == uid)
+			return zone;
+	}
+
+	return NULL;
+}
+
+struct mlx4_bitmap *mlx4_zone_get_bitmap(struct mlx4_zone_allocator *zones, u32 uid)
+{
+	struct mlx4_zone_entry *zone;
+	struct mlx4_bitmap *bitmap;
+
+	spin_lock(&zones->lock);
+
+	zone = __mlx4_find_zone_by_uid(zones, uid);
+
+	bitmap = zone == NULL ? NULL : zone->bitmap;
+
+	spin_unlock(&zones->lock);
+
+	return bitmap;
+}
+
+int mlx4_zone_remove_one(struct mlx4_zone_allocator *zones, u32 uid)
+{
+	struct mlx4_zone_entry *zone;
+	int res = 0;
+
+	spin_lock(&zones->lock);
+
+	zone = __mlx4_find_zone_by_uid(zones, uid);
+
+	if (NULL == zone) {
+		res = -1;
+		goto out;
+	}
+
+	__mlx4_zone_remove_one_entry(zone);
+
+out:
+	spin_unlock(&zones->lock);
+	kfree(zone);
+
+	return res;
+}
+
+/* Should be called under a lock */
+static struct mlx4_zone_entry *__mlx4_find_zone_by_uid_unique(
+		struct mlx4_zone_allocator *zones, u32 obj)
+{
+	struct mlx4_zone_entry *zone, *zone_candidate = NULL;
+	u32 dist = (u32)-1;
+
+	/* Search for the smallest zone that this obj could be
+	 * allocated from. This is done in order to handle
+	 * situations when small bitmaps are allocated from bigger
+	 * bitmaps (and the allocated space is marked as reserved in
+	 * the bigger bitmap.
+	 */
+	list_for_each_entry(zone, &zones->entries, list) {
+		if (obj >= zone->offset) {
+			u32 mobj = (obj - zone->offset) & zones->mask;
+
+			if (mobj < zone->bitmap->max) {
+				u32 curr_dist = zone->bitmap->effective_len;
+
+				if (curr_dist < dist) {
+					dist = curr_dist;
+					zone_candidate = zone;
+				}
+			}
+		}
+	}
+
+	return zone_candidate;
+}
+
+u32 mlx4_zone_alloc_entries(struct mlx4_zone_allocator *zones, u32 uid, int count,
+			    int align, u32 skip_mask, u32 *puid)
+{
+	struct mlx4_zone_entry *zone;
+	int res = -1;
+
+	spin_lock(&zones->lock);
+
+	zone = __mlx4_find_zone_by_uid(zones, uid);
+
+	if (NULL == zone)
+		goto out;
+
+	res = __mlx4_alloc_from_zone(zone, count, align, skip_mask, puid);
+
+out:
+	spin_unlock(&zones->lock);
+
+	return res;
+}
+
+u32 mlx4_zone_free_entries(struct mlx4_zone_allocator *zones, u32 uid, u32 obj, u32 count)
+{
+	struct mlx4_zone_entry *zone;
+	int res = 0;
+
+	spin_lock(&zones->lock);
+
+	zone = __mlx4_find_zone_by_uid(zones, uid);
+
+	if (NULL == zone) {
+		res = -1;
+		goto out;
+	}
+
+	__mlx4_free_from_zone(zone, obj, count);
+
+out:
+	spin_unlock(&zones->lock);
+
+	return res;
+}
+
+u32 mlx4_zone_free_entries_unique(struct mlx4_zone_allocator *zones, u32 obj, u32 count)
+{
+	struct mlx4_zone_entry *zone;
+	int res;
+
+	if (!(zones->flags & MLX4_ZONE_ALLOC_FLAGS_NO_OVERLAP))
+		return -EFAULT;
+
+	spin_lock(&zones->lock);
+
+	zone = __mlx4_find_zone_by_uid_unique(zones, obj);
+
+	if (NULL == zone) {
+		res = -1;
+		goto out;
+	}
+
+	__mlx4_free_from_zone(zone, obj, count);
+	res = 0;
+
+out:
+	spin_unlock(&zones->lock);
+
+	return res;
+}
+
+static int mlx4_buf_direct_alloc(struct mlx4_dev *dev, int size,
+				 struct mlx4_buf *buf)
+{
+	dma_addr_t t;
+
+	buf->nbufs        = 1;
+	buf->npages       = 1;
+	buf->page_shift   = get_order(size) + PAGE_SHIFT;
+	buf->direct.buf   =
+		dma_alloc_coherent(&dev->persist->pdev->dev, size, &t,
+				   GFP_KERNEL);
+	if (!buf->direct.buf)
+		return -ENOMEM;
+
+	buf->direct.map = t;
+
+	while (t & ((1 << buf->page_shift) - 1)) {
+		--buf->page_shift;
+		buf->npages *= 2;
+	}
+
+	return 0;
+}
+
+/* Handling for queue buffers -- we allocate a bunch of memory and
+ * register it in a memory region at HCA virtual address 0. If the
+ *  requested size is > max_direct, we split the allocation into
+ *  multiple pages, so we don't require too much contiguous memory.
+ */
+int mlx4_buf_alloc(struct mlx4_dev *dev, int size, int max_direct,
+		   struct mlx4_buf *buf)
+{
+	if (size <= max_direct) {
+		return mlx4_buf_direct_alloc(dev, size, buf);
+	} else {
+		dma_addr_t t;
+		int i;
+
+		buf->direct.buf = NULL;
+		buf->nbufs      = DIV_ROUND_UP(size, PAGE_SIZE);
+		buf->npages	= buf->nbufs;
+		buf->page_shift  = PAGE_SHIFT;
+		buf->page_list   = kcalloc(buf->nbufs, sizeof(*buf->page_list),
+					   GFP_KERNEL);
+		if (!buf->page_list)
+			return -ENOMEM;
+
+		for (i = 0; i < buf->nbufs; ++i) {
+			buf->page_list[i].buf =
+				dma_alloc_coherent(&dev->persist->pdev->dev,
+						   PAGE_SIZE, &t, GFP_KERNEL);
+			if (!buf->page_list[i].buf)
+				goto err_free;
+
+			buf->page_list[i].map = t;
+		}
+	}
+
+	return 0;
+
+err_free:
+	mlx4_buf_free(dev, size, buf);
+
+	return -ENOMEM;
+}
+EXPORT_SYMBOL_GPL(mlx4_buf_alloc);
+
+void mlx4_buf_free(struct mlx4_dev *dev, int size, struct mlx4_buf *buf)
+{
+	if (buf->nbufs == 1) {
+		dma_free_coherent(&dev->persist->pdev->dev, size,
+				  buf->direct.buf, buf->direct.map);
+	} else {
+		int i;
+
+		for (i = 0; i < buf->nbufs; ++i)
+			if (buf->page_list[i].buf)
+				dma_free_coherent(&dev->persist->pdev->dev,
+						  PAGE_SIZE,
+						  buf->page_list[i].buf,
+						  buf->page_list[i].map);
+		kfree(buf->page_list);
+	}
+}
+EXPORT_SYMBOL_GPL(mlx4_buf_free);
+
+static struct mlx4_db_pgdir *mlx4_alloc_db_pgdir(struct device *dma_device)
+{
+	struct mlx4_db_pgdir *pgdir;
+
+	pgdir = kzalloc(sizeof(*pgdir), GFP_KERNEL);
+	if (!pgdir)
+		return NULL;
+
+	bitmap_fill(pgdir->order1, MLX4_DB_PER_PAGE / 2);
+	pgdir->bits[0] = pgdir->order0;
+	pgdir->bits[1] = pgdir->order1;
+	pgdir->db_page = dma_alloc_coherent(dma_device, PAGE_SIZE,
+					    &pgdir->db_dma, GFP_KERNEL);
+	if (!pgdir->db_page) {
+		kfree(pgdir);
+		return NULL;
+	}
+
+	return pgdir;
+}
+
+static int mlx4_alloc_db_from_pgdir(struct mlx4_db_pgdir *pgdir,
+				    struct mlx4_db *db, int order)
+{
+	int o;
+	int i;
+
+	for (o = order; o <= 1; ++o) {
+		i = find_first_bit(pgdir->bits[o], MLX4_DB_PER_PAGE >> o);
+		if (i < MLX4_DB_PER_PAGE >> o)
+			goto found;
+	}
+
+	return -ENOMEM;
+
+found:
+	clear_bit(i, pgdir->bits[o]);
+
+	i <<= o;
+
+	if (o > order)
+		set_bit(i ^ 1, pgdir->bits[order]);
+
+	db->u.pgdir = pgdir;
+	db->index   = i;
+	db->db      = pgdir->db_page + db->index;
+	db->dma     = pgdir->db_dma  + db->index * 4;
+	db->order   = order;
+
+	return 0;
+}
+
+int mlx4_db_alloc(struct mlx4_dev *dev, struct mlx4_db *db, int order)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_db_pgdir *pgdir;
+	int ret = 0;
+
+	mutex_lock(&priv->pgdir_mutex);
+
+	list_for_each_entry(pgdir, &priv->pgdir_list, list)
+		if (!mlx4_alloc_db_from_pgdir(pgdir, db, order))
+			goto out;
+
+	pgdir = mlx4_alloc_db_pgdir(&dev->persist->pdev->dev);
+	if (!pgdir) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	list_add(&pgdir->list, &priv->pgdir_list);
+
+	/* This should never fail -- we just allocated an empty page: */
+	WARN_ON(mlx4_alloc_db_from_pgdir(pgdir, db, order));
+
+out:
+	mutex_unlock(&priv->pgdir_mutex);
+
+	return ret;
+}
+EXPORT_SYMBOL_GPL(mlx4_db_alloc);
+
+void mlx4_db_free(struct mlx4_dev *dev, struct mlx4_db *db)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int o;
+	int i;
+
+	mutex_lock(&priv->pgdir_mutex);
+
+	o = db->order;
+	i = db->index;
+
+	if (db->order == 0 && test_bit(i ^ 1, db->u.pgdir->order0)) {
+		clear_bit(i ^ 1, db->u.pgdir->order0);
+		++o;
+	}
+	i >>= o;
+	set_bit(i, db->u.pgdir->bits[o]);
+
+	if (bitmap_full(db->u.pgdir->order1, MLX4_DB_PER_PAGE / 2)) {
+		dma_free_coherent(&dev->persist->pdev->dev, PAGE_SIZE,
+				  db->u.pgdir->db_page, db->u.pgdir->db_dma);
+		list_del(&db->u.pgdir->list);
+		kfree(db->u.pgdir);
+	}
+
+	mutex_unlock(&priv->pgdir_mutex);
+}
+EXPORT_SYMBOL_GPL(mlx4_db_free);
+
+int mlx4_alloc_hwq_res(struct mlx4_dev *dev, struct mlx4_hwq_resources *wqres,
+		       int size)
+{
+	int err;
+
+	err = mlx4_db_alloc(dev, &wqres->db, 1);
+	if (err)
+		return err;
+
+	*wqres->db.db = 0;
+
+	err = mlx4_buf_direct_alloc(dev, size, &wqres->buf);
+	if (err)
+		goto err_db;
+
+	err = mlx4_mtt_init(dev, wqres->buf.npages, wqres->buf.page_shift,
+			    &wqres->mtt);
+	if (err)
+		goto err_buf;
+
+	err = mlx4_buf_write_mtt(dev, &wqres->mtt, &wqres->buf);
+	if (err)
+		goto err_mtt;
+
+	return 0;
+
+err_mtt:
+	mlx4_mtt_cleanup(dev, &wqres->mtt);
+err_buf:
+	mlx4_buf_free(dev, size, &wqres->buf);
+err_db:
+	mlx4_db_free(dev, &wqres->db);
+
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx4_alloc_hwq_res);
+
+void mlx4_free_hwq_res(struct mlx4_dev *dev, struct mlx4_hwq_resources *wqres,
+		       int size)
+{
+	mlx4_mtt_cleanup(dev, &wqres->mtt);
+	mlx4_buf_free(dev, size, &wqres->buf);
+	mlx4_db_free(dev, &wqres->db);
+}
+EXPORT_SYMBOL_GPL(mlx4_free_hwq_res);
diff --git a/drivers/net/ethernet/mellanox/mlx4/catas.c b/drivers/net/ethernet/mellanox/mlx4/catas.c
new file mode 100644
index 000000000..0eb7b8363
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx4/catas.c
@@ -0,0 +1,335 @@
+/*
+ * Copyright (c) 2007 Cisco Systems, Inc. All rights reserved.
+ * Copyright (c) 2007, 2008 Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/workqueue.h>
+#include <linux/module.h>
+
+#include "mlx4.h"
+
+enum {
+	MLX4_CATAS_POLL_INTERVAL	= 5 * HZ,
+};
+
+
+
+int mlx4_internal_err_reset = 1;
+module_param_named(internal_err_reset, mlx4_internal_err_reset,  int, 0644);
+MODULE_PARM_DESC(internal_err_reset,
+		 "Reset device on internal errors if non-zero (default 1)");
+
+static int read_vendor_id(struct mlx4_dev *dev)
+{
+	u16 vendor_id = 0;
+	int ret;
+
+	ret = pci_read_config_word(dev->persist->pdev, 0, &vendor_id);
+	if (ret) {
+		mlx4_err(dev, "Failed to read vendor ID, ret=%d\n", ret);
+		return ret;
+	}
+
+	if (vendor_id == 0xffff) {
+		mlx4_err(dev, "PCI can't be accessed to read vendor id\n");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int mlx4_reset_master(struct mlx4_dev *dev)
+{
+	int err = 0;
+
+	if (mlx4_is_master(dev))
+		mlx4_report_internal_err_comm_event(dev);
+
+	if (!pci_channel_offline(dev->persist->pdev)) {
+		err = read_vendor_id(dev);
+		/* If PCI can't be accessed to read vendor ID we assume that its
+		 * link was disabled and chip was already reset.
+		 */
+		if (err)
+			return 0;
+
+		err = mlx4_reset(dev);
+		if (err)
+			mlx4_err(dev, "Fail to reset HCA\n");
+	}
+
+	return err;
+}
+
+static int mlx4_reset_slave(struct mlx4_dev *dev)
+{
+#define COM_CHAN_RST_REQ_OFFSET 0x10
+#define COM_CHAN_RST_ACK_OFFSET 0x08
+
+	u32 comm_flags;
+	u32 rst_req;
+	u32 rst_ack;
+	unsigned long end;
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	if (pci_channel_offline(dev->persist->pdev))
+		return 0;
+
+	comm_flags = swab32(readl((__iomem char *)priv->mfunc.comm +
+				  MLX4_COMM_CHAN_FLAGS));
+	if (comm_flags == 0xffffffff) {
+		mlx4_err(dev, "VF reset is not needed\n");
+		return 0;
+	}
+
+	if (!(dev->caps.vf_caps & MLX4_VF_CAP_FLAG_RESET)) {
+		mlx4_err(dev, "VF reset is not supported\n");
+		return -EOPNOTSUPP;
+	}
+
+	rst_req = (comm_flags & (u32)(1 << COM_CHAN_RST_REQ_OFFSET)) >>
+		COM_CHAN_RST_REQ_OFFSET;
+	rst_ack = (comm_flags & (u32)(1 << COM_CHAN_RST_ACK_OFFSET)) >>
+		COM_CHAN_RST_ACK_OFFSET;
+	if (rst_req != rst_ack) {
+		mlx4_err(dev, "Communication channel isn't sync, fail to send reset\n");
+		return -EIO;
+	}
+
+	rst_req ^= 1;
+	mlx4_warn(dev, "VF is sending reset request to Firmware\n");
+	comm_flags = rst_req << COM_CHAN_RST_REQ_OFFSET;
+	__raw_writel((__force u32)cpu_to_be32(comm_flags),
+		     (__iomem char *)priv->mfunc.comm + MLX4_COMM_CHAN_FLAGS);
+
+	end = msecs_to_jiffies(MLX4_COMM_TIME) + jiffies;
+	while (time_before(jiffies, end)) {
+		comm_flags = swab32(readl((__iomem char *)priv->mfunc.comm +
+					  MLX4_COMM_CHAN_FLAGS));
+		rst_ack = (comm_flags & (u32)(1 << COM_CHAN_RST_ACK_OFFSET)) >>
+			COM_CHAN_RST_ACK_OFFSET;
+
+		/* Reading rst_req again since the communication channel can
+		 * be reset at any time by the PF and all its bits will be
+		 * set to zero.
+		 */
+		rst_req = (comm_flags & (u32)(1 << COM_CHAN_RST_REQ_OFFSET)) >>
+			COM_CHAN_RST_REQ_OFFSET;
+
+		if (rst_ack == rst_req) {
+			mlx4_warn(dev, "VF Reset succeed\n");
+			return 0;
+		}
+		cond_resched();
+	}
+	mlx4_err(dev, "Fail to send reset over the communication channel\n");
+	return -ETIMEDOUT;
+}
+
+int mlx4_comm_internal_err(u32 slave_read)
+{
+	return (u32)COMM_CHAN_EVENT_INTERNAL_ERR ==
+		(slave_read & (u32)COMM_CHAN_EVENT_INTERNAL_ERR) ? 1 : 0;
+}
+
+void mlx4_enter_error_state(struct mlx4_dev_persistent *persist)
+{
+	int err;
+	struct mlx4_dev *dev;
+
+	if (!mlx4_internal_err_reset)
+		return;
+
+	mutex_lock(&persist->device_state_mutex);
+	if (persist->state & MLX4_DEVICE_STATE_INTERNAL_ERROR)
+		goto out;
+
+	dev = persist->dev;
+	mlx4_err(dev, "device is going to be reset\n");
+	if (mlx4_is_slave(dev)) {
+		err = mlx4_reset_slave(dev);
+	} else {
+		mlx4_crdump_collect(dev);
+		err = mlx4_reset_master(dev);
+	}
+
+	if (!err) {
+		mlx4_err(dev, "device was reset successfully\n");
+	} else {
+		/* EEH could have disabled the PCI channel during reset. That's
+		 * recoverable and the PCI error flow will handle it.
+		 */
+		if (!pci_channel_offline(dev->persist->pdev))
+			BUG_ON(1);
+	}
+	dev->persist->state |= MLX4_DEVICE_STATE_INTERNAL_ERROR;
+	mutex_unlock(&persist->device_state_mutex);
+
+	/* At that step HW was already reset, now notify clients */
+	mlx4_dispatch_event(dev, MLX4_DEV_EVENT_CATASTROPHIC_ERROR, 0);
+	mlx4_cmd_wake_completions(dev);
+	return;
+
+out:
+	mutex_unlock(&persist->device_state_mutex);
+}
+
+static void mlx4_handle_error_state(struct mlx4_dev_persistent *persist)
+{
+	struct mlx4_dev *dev = persist->dev;
+	struct devlink *devlink;
+	int err = 0;
+
+	mlx4_enter_error_state(persist);
+	devlink = priv_to_devlink(mlx4_priv(dev));
+	devl_lock(devlink);
+	mutex_lock(&persist->interface_state_mutex);
+	if (persist->interface_state & MLX4_INTERFACE_STATE_UP &&
+	    !(persist->interface_state & MLX4_INTERFACE_STATE_DELETION)) {
+		err = mlx4_restart_one(persist->pdev);
+		mlx4_info(persist->dev, "mlx4_restart_one was ended, ret=%d\n",
+			  err);
+	}
+	mutex_unlock(&persist->interface_state_mutex);
+	devl_unlock(devlink);
+}
+
+static void dump_err_buf(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	int i;
+
+	mlx4_err(dev, "Internal error detected:\n");
+	for (i = 0; i < priv->fw.catas_size; ++i)
+		mlx4_err(dev, "  buf[%02x]: %08x\n",
+			 i, swab32(readl(priv->catas_err.map + i)));
+}
+
+static void poll_catas(struct timer_list *t)
+{
+	struct mlx4_priv *priv = from_timer(priv, t, catas_err.timer);
+	struct mlx4_dev *dev = &priv->dev;
+	u32 slave_read;
+
+	if (mlx4_is_slave(dev)) {
+		slave_read = swab32(readl(&priv->mfunc.comm->slave_read));
+		if (mlx4_comm_internal_err(slave_read)) {
+			mlx4_warn(dev, "Internal error detected on the communication channel\n");
+			goto internal_err;
+		}
+	} else if (readl(priv->catas_err.map)) {
+		dump_err_buf(dev);
+		goto internal_err;
+	}
+
+	if (dev->persist->state & MLX4_DEVICE_STATE_INTERNAL_ERROR) {
+		mlx4_warn(dev, "Internal error mark was detected on device\n");
+		goto internal_err;
+	}
+
+	mod_timer(&priv->catas_err.timer,
+		  round_jiffies(jiffies + MLX4_CATAS_POLL_INTERVAL));
+	return;
+
+internal_err:
+	if (mlx4_internal_err_reset)
+		queue_work(dev->persist->catas_wq, &dev->persist->catas_work);
+}
+
+static void catas_reset(struct work_struct *work)
+{
+	struct mlx4_dev_persistent *persist =
+		container_of(work, struct mlx4_dev_persistent,
+			     catas_work);
+
+	mlx4_handle_error_state(persist);
+}
+
+void mlx4_start_catas_poll(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	phys_addr_t addr;
+
+	INIT_LIST_HEAD(&priv->catas_err.list);
+	timer_setup(&priv->catas_err.timer, poll_catas, 0);
+	priv->catas_err.map = NULL;
+
+	if (!mlx4_is_slave(dev)) {
+		addr = pci_resource_start(dev->persist->pdev,
+					  priv->fw.catas_bar) +
+					  priv->fw.catas_offset;
+
+		priv->catas_err.map = ioremap(addr, priv->fw.catas_size * 4);
+		if (!priv->catas_err.map) {
+			mlx4_warn(dev, "Failed to map internal error buffer at 0x%llx\n",
+				  (unsigned long long)addr);
+			return;
+		}
+	}
+
+	priv->catas_err.timer.expires  =
+		round_jiffies(jiffies + MLX4_CATAS_POLL_INTERVAL);
+	add_timer(&priv->catas_err.timer);
+}
+
+void mlx4_stop_catas_poll(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	del_timer_sync(&priv->catas_err.timer);
+
+	if (priv->catas_err.map) {
+		iounmap(priv->catas_err.map);
+		priv->catas_err.map = NULL;
+	}
+
+	if (dev->persist->interface_state & MLX4_INTERFACE_STATE_DELETION)
+		flush_workqueue(dev->persist->catas_wq);
+}
+
+int  mlx4_catas_init(struct mlx4_dev *dev)
+{
+	INIT_WORK(&dev->persist->catas_work, catas_reset);
+	dev->persist->catas_wq = create_singlethread_workqueue("mlx4_health");
+	if (!dev->persist->catas_wq)
+		return -ENOMEM;
+
+	return 0;
+}
+
+void mlx4_catas_end(struct mlx4_dev *dev)
+{
+	if (dev->persist->catas_wq) {
+		destroy_workqueue(dev->persist->catas_wq);
+		dev->persist->catas_wq = NULL;
+	}
+}
diff --git a/drivers/net/ethernet/mellanox/mlx4/cmd.c b/drivers/net/ethernet/mellanox/mlx4/cmd.c
new file mode 100644
index 000000000..c56d2194c
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx4/cmd.c
@@ -0,0 +1,3428 @@
+/*
+ * Copyright (c) 2004, 2005 Topspin Communications.  All rights reserved.
+ * Copyright (c) 2005, 2006, 2007, 2008 Mellanox Technologies. All rights reserved.
+ * Copyright (c) 2005, 2006, 2007 Cisco Systems, Inc.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/sched.h>
+#include <linux/slab.h>
+#include <linux/export.h>
+#include <linux/pci.h>
+#include <linux/errno.h>
+
+#include <linux/mlx4/cmd.h>
+#include <linux/mlx4/device.h>
+#include <linux/semaphore.h>
+#include <rdma/ib_smi.h>
+#include <linux/delay.h>
+#include <linux/etherdevice.h>
+
+#include <asm/io.h>
+
+#include "mlx4.h"
+#include "fw.h"
+#include "fw_qos.h"
+#include "mlx4_stats.h"
+
+#define CMD_POLL_TOKEN 0xffff
+#define INBOX_MASK	0xffffffffffffff00ULL
+
+#define CMD_CHAN_VER 1
+#define CMD_CHAN_IF_REV 1
+
+enum {
+	/* command completed successfully: */
+	CMD_STAT_OK		= 0x00,
+	/* Internal error (such as a bus error) occurred while processing command: */
+	CMD_STAT_INTERNAL_ERR	= 0x01,
+	/* Operation/command not supported or opcode modifier not supported: */
+	CMD_STAT_BAD_OP		= 0x02,
+	/* Parameter not supported or parameter out of range: */
+	CMD_STAT_BAD_PARAM	= 0x03,
+	/* System not enabled or bad system state: */
+	CMD_STAT_BAD_SYS_STATE	= 0x04,
+	/* Attempt to access reserved or unallocaterd resource: */
+	CMD_STAT_BAD_RESOURCE	= 0x05,
+	/* Requested resource is currently executing a command, or is otherwise busy: */
+	CMD_STAT_RESOURCE_BUSY	= 0x06,
+	/* Required capability exceeds device limits: */
+	CMD_STAT_EXCEED_LIM	= 0x08,
+	/* Resource is not in the appropriate state or ownership: */
+	CMD_STAT_BAD_RES_STATE	= 0x09,
+	/* Index out of range: */
+	CMD_STAT_BAD_INDEX	= 0x0a,
+	/* FW image corrupted: */
+	CMD_STAT_BAD_NVMEM	= 0x0b,
+	/* Error in ICM mapping (e.g. not enough auxiliary ICM pages to execute command): */
+	CMD_STAT_ICM_ERROR	= 0x0c,
+	/* Attempt to modify a QP/EE which is not in the presumed state: */
+	CMD_STAT_BAD_QP_STATE   = 0x10,
+	/* Bad segment parameters (Address/Size): */
+	CMD_STAT_BAD_SEG_PARAM	= 0x20,
+	/* Memory Region has Memory Windows bound to: */
+	CMD_STAT_REG_BOUND	= 0x21,
+	/* HCA local attached memory not present: */
+	CMD_STAT_LAM_NOT_PRE	= 0x22,
+	/* Bad management packet (silently discarded): */
+	CMD_STAT_BAD_PKT	= 0x30,
+	/* More outstanding CQEs in CQ than new CQ size: */
+	CMD_STAT_BAD_SIZE	= 0x40,
+	/* Multi Function device support required: */
+	CMD_STAT_MULTI_FUNC_REQ	= 0x50,
+};
+
+enum {
+	HCR_IN_PARAM_OFFSET	= 0x00,
+	HCR_IN_MODIFIER_OFFSET	= 0x08,
+	HCR_OUT_PARAM_OFFSET	= 0x0c,
+	HCR_TOKEN_OFFSET	= 0x14,
+	HCR_STATUS_OFFSET	= 0x18,
+
+	HCR_OPMOD_SHIFT		= 12,
+	HCR_T_BIT		= 21,
+	HCR_E_BIT		= 22,
+	HCR_GO_BIT		= 23
+};
+
+enum {
+	GO_BIT_TIMEOUT_MSECS	= 10000
+};
+
+enum mlx4_vlan_transition {
+	MLX4_VLAN_TRANSITION_VST_VST = 0,
+	MLX4_VLAN_TRANSITION_VST_VGT = 1,
+	MLX4_VLAN_TRANSITION_VGT_VST = 2,
+	MLX4_VLAN_TRANSITION_VGT_VGT = 3,
+};
+
+
+struct mlx4_cmd_context {
+	struct completion	done;
+	int			result;
+	int			next;
+	u64			out_param;
+	u16			token;
+	u8			fw_status;
+};
+
+static int mlx4_master_process_vhcr(struct mlx4_dev *dev, int slave,
+				    struct mlx4_vhcr_cmd *in_vhcr);
+
+static int mlx4_status_to_errno(u8 status)
+{
+	static const int trans_table[] = {
+		[CMD_STAT_INTERNAL_ERR]	  = -EIO,
+		[CMD_STAT_BAD_OP]	  = -EPERM,
+		[CMD_STAT_BAD_PARAM]	  = -EINVAL,
+		[CMD_STAT_BAD_SYS_STATE]  = -ENXIO,
+		[CMD_STAT_BAD_RESOURCE]	  = -EBADF,
+		[CMD_STAT_RESOURCE_BUSY]  = -EBUSY,
+		[CMD_STAT_EXCEED_LIM]	  = -ENOMEM,
+		[CMD_STAT_BAD_RES_STATE]  = -EBADF,
+		[CMD_STAT_BAD_INDEX]	  = -EBADF,
+		[CMD_STAT_BAD_NVMEM]	  = -EFAULT,
+		[CMD_STAT_ICM_ERROR]	  = -ENFILE,
+		[CMD_STAT_BAD_QP_STATE]   = -EINVAL,
+		[CMD_STAT_BAD_SEG_PARAM]  = -EFAULT,
+		[CMD_STAT_REG_BOUND]	  = -EBUSY,
+		[CMD_STAT_LAM_NOT_PRE]	  = -EAGAIN,
+		[CMD_STAT_BAD_PKT]	  = -EINVAL,
+		[CMD_STAT_BAD_SIZE]	  = -ENOMEM,
+		[CMD_STAT_MULTI_FUNC_REQ] = -EACCES,
+	};
+
+	if (status >= ARRAY_SIZE(trans_table) ||
+	    (status != CMD_STAT_OK && trans_table[status] == 0))
+		return -EIO;
+
+	return trans_table[status];
+}
+
+static u8 mlx4_errno_to_status(int errno)
+{
+	switch (errno) {
+	case -EPERM:
+		return CMD_STAT_BAD_OP;
+	case -EINVAL:
+		return CMD_STAT_BAD_PARAM;
+	case -ENXIO:
+		return CMD_STAT_BAD_SYS_STATE;
+	case -EBUSY:
+		return CMD_STAT_RESOURCE_BUSY;
+	case -ENOMEM:
+		return CMD_STAT_EXCEED_LIM;
+	case -ENFILE:
+		return CMD_STAT_ICM_ERROR;
+	default:
+		return CMD_STAT_INTERNAL_ERR;
+	}
+}
+
+static int mlx4_internal_err_ret_value(struct mlx4_dev *dev, u16 op,
+				       u8 op_modifier)
+{
+	switch (op) {
+	case MLX4_CMD_UNMAP_ICM:
+	case MLX4_CMD_UNMAP_ICM_AUX:
+	case MLX4_CMD_UNMAP_FA:
+	case MLX4_CMD_2RST_QP:
+	case MLX4_CMD_HW2SW_EQ:
+	case MLX4_CMD_HW2SW_CQ:
+	case MLX4_CMD_HW2SW_SRQ:
+	case MLX4_CMD_HW2SW_MPT:
+	case MLX4_CMD_CLOSE_HCA:
+	case MLX4_QP_FLOW_STEERING_DETACH:
+	case MLX4_CMD_FREE_RES:
+	case MLX4_CMD_CLOSE_PORT:
+		return CMD_STAT_OK;
+
+	case MLX4_CMD_QP_ATTACH:
+		/* On Detach case return success */
+		if (op_modifier == 0)
+			return CMD_STAT_OK;
+		return mlx4_status_to_errno(CMD_STAT_INTERNAL_ERR);
+
+	default:
+		return mlx4_status_to_errno(CMD_STAT_INTERNAL_ERR);
+	}
+}
+
+static int mlx4_closing_cmd_fatal_error(u16 op, u8 fw_status)
+{
+	/* Any error during the closing commands below is considered fatal */
+	if (op == MLX4_CMD_CLOSE_HCA ||
+	    op == MLX4_CMD_HW2SW_EQ ||
+	    op == MLX4_CMD_HW2SW_CQ ||
+	    op == MLX4_CMD_2RST_QP ||
+	    op == MLX4_CMD_HW2SW_SRQ ||
+	    op == MLX4_CMD_SYNC_TPT ||
+	    op == MLX4_CMD_UNMAP_ICM ||
+	    op == MLX4_CMD_UNMAP_ICM_AUX ||
+	    op == MLX4_CMD_UNMAP_FA)
+		return 1;
+	/* Error on MLX4_CMD_HW2SW_MPT is fatal except when fw status equals
+	  * CMD_STAT_REG_BOUND.
+	  * This status indicates that memory region has memory windows bound to it
+	  * which may result from invalid user space usage and is not fatal.
+	  */
+	if (op == MLX4_CMD_HW2SW_MPT && fw_status != CMD_STAT_REG_BOUND)
+		return 1;
+	return 0;
+}
+
+static int mlx4_cmd_reset_flow(struct mlx4_dev *dev, u16 op, u8 op_modifier,
+			       int err)
+{
+	/* Only if reset flow is really active return code is based on
+	  * command, otherwise current error code is returned.
+	  */
+	if (mlx4_internal_err_reset) {
+		mlx4_enter_error_state(dev->persist);
+		err = mlx4_internal_err_ret_value(dev, op, op_modifier);
+	}
+
+	return err;
+}
+
+static int comm_pending(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	u32 status = readl(&priv->mfunc.comm->slave_read);
+
+	return (swab32(status) >> 31) != priv->cmd.comm_toggle;
+}
+
+static int mlx4_comm_cmd_post(struct mlx4_dev *dev, u8 cmd, u16 param)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	u32 val;
+
+	/* To avoid writing to unknown addresses after the device state was
+	 * changed to internal error and the function was rest,
+	 * check the INTERNAL_ERROR flag which is updated under
+	 * device_state_mutex lock.
+	 */
+	mutex_lock(&dev->persist->device_state_mutex);
+
+	if (dev->persist->state & MLX4_DEVICE_STATE_INTERNAL_ERROR) {
+		mutex_unlock(&dev->persist->device_state_mutex);
+		return -EIO;
+	}
+
+	priv->cmd.comm_toggle ^= 1;
+	val = param | (cmd << 16) | (priv->cmd.comm_toggle << 31);
+	__raw_writel((__force u32) cpu_to_be32(val),
+		     &priv->mfunc.comm->slave_write);
+	mutex_unlock(&dev->persist->device_state_mutex);
+	return 0;
+}
+
+static int mlx4_comm_cmd_poll(struct mlx4_dev *dev, u8 cmd, u16 param,
+		       unsigned long timeout)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	unsigned long end;
+	int err = 0;
+	int ret_from_pending = 0;
+
+	/* First, verify that the master reports correct status */
+	if (comm_pending(dev)) {
+		mlx4_warn(dev, "Communication channel is not idle - my toggle is %d (cmd:0x%x)\n",
+			  priv->cmd.comm_toggle, cmd);
+		return -EAGAIN;
+	}
+
+	/* Write command */
+	down(&priv->cmd.poll_sem);
+	if (mlx4_comm_cmd_post(dev, cmd, param)) {
+		/* Only in case the device state is INTERNAL_ERROR,
+		 * mlx4_comm_cmd_post returns with an error
+		 */
+		err = mlx4_status_to_errno(CMD_STAT_INTERNAL_ERR);
+		goto out;
+	}
+
+	end = msecs_to_jiffies(timeout) + jiffies;
+	while (comm_pending(dev) && time_before(jiffies, end))
+		cond_resched();
+	ret_from_pending = comm_pending(dev);
+	if (ret_from_pending) {
+		/* check if the slave is trying to boot in the middle of
+		 * FLR process. The only non-zero result in the RESET command
+		 * is MLX4_DELAY_RESET_SLAVE*/
+		if ((MLX4_COMM_CMD_RESET == cmd)) {
+			err = MLX4_DELAY_RESET_SLAVE;
+			goto out;
+		} else {
+			mlx4_warn(dev, "Communication channel command 0x%x timed out\n",
+				  cmd);
+			err = mlx4_status_to_errno(CMD_STAT_INTERNAL_ERR);
+		}
+	}
+
+	if (err)
+		mlx4_enter_error_state(dev->persist);
+out:
+	up(&priv->cmd.poll_sem);
+	return err;
+}
+
+static int mlx4_comm_cmd_wait(struct mlx4_dev *dev, u8 vhcr_cmd,
+			      u16 param, u16 op, unsigned long timeout)
+{
+	struct mlx4_cmd *cmd = &mlx4_priv(dev)->cmd;
+	struct mlx4_cmd_context *context;
+	unsigned long end;
+	int err = 0;
+
+	down(&cmd->event_sem);
+
+	spin_lock(&cmd->context_lock);
+	BUG_ON(cmd->free_head < 0);
+	context = &cmd->context[cmd->free_head];
+	context->token += cmd->token_mask + 1;
+	cmd->free_head = context->next;
+	spin_unlock(&cmd->context_lock);
+
+	reinit_completion(&context->done);
+
+	if (mlx4_comm_cmd_post(dev, vhcr_cmd, param)) {
+		/* Only in case the device state is INTERNAL_ERROR,
+		 * mlx4_comm_cmd_post returns with an error
+		 */
+		err = mlx4_status_to_errno(CMD_STAT_INTERNAL_ERR);
+		goto out;
+	}
+
+	if (!wait_for_completion_timeout(&context->done,
+					 msecs_to_jiffies(timeout))) {
+		mlx4_warn(dev, "communication channel command 0x%x (op=0x%x) timed out\n",
+			  vhcr_cmd, op);
+		goto out_reset;
+	}
+
+	err = context->result;
+	if (err && context->fw_status != CMD_STAT_MULTI_FUNC_REQ) {
+		mlx4_err(dev, "command 0x%x failed: fw status = 0x%x\n",
+			 vhcr_cmd, context->fw_status);
+		if (mlx4_closing_cmd_fatal_error(op, context->fw_status))
+			goto out_reset;
+	}
+
+	/* wait for comm channel ready
+	 * this is necessary for prevention the race
+	 * when switching between event to polling mode
+	 * Skipping this section in case the device is in FATAL_ERROR state,
+	 * In this state, no commands are sent via the comm channel until
+	 * the device has returned from reset.
+	 */
+	if (!(dev->persist->state & MLX4_DEVICE_STATE_INTERNAL_ERROR)) {
+		end = msecs_to_jiffies(timeout) + jiffies;
+		while (comm_pending(dev) && time_before(jiffies, end))
+			cond_resched();
+	}
+	goto out;
+
+out_reset:
+	err = mlx4_status_to_errno(CMD_STAT_INTERNAL_ERR);
+	mlx4_enter_error_state(dev->persist);
+out:
+	spin_lock(&cmd->context_lock);
+	context->next = cmd->free_head;
+	cmd->free_head = context - cmd->context;
+	spin_unlock(&cmd->context_lock);
+
+	up(&cmd->event_sem);
+	return err;
+}
+
+int mlx4_comm_cmd(struct mlx4_dev *dev, u8 cmd, u16 param,
+		  u16 op, unsigned long timeout)
+{
+	if (dev->persist->state & MLX4_DEVICE_STATE_INTERNAL_ERROR)
+		return mlx4_status_to_errno(CMD_STAT_INTERNAL_ERR);
+
+	if (mlx4_priv(dev)->cmd.use_events)
+		return mlx4_comm_cmd_wait(dev, cmd, param, op, timeout);
+	return mlx4_comm_cmd_poll(dev, cmd, param, timeout);
+}
+
+static int cmd_pending(struct mlx4_dev *dev)
+{
+	u32 status;
+
+	if (pci_channel_offline(dev->persist->pdev))
+		return -EIO;
+
+	status = readl(mlx4_priv(dev)->cmd.hcr + HCR_STATUS_OFFSET);
+
+	return (status & swab32(1 << HCR_GO_BIT)) ||
+		(mlx4_priv(dev)->cmd.toggle ==
+		 !!(status & swab32(1 << HCR_T_BIT)));
+}
+
+static int mlx4_cmd_post(struct mlx4_dev *dev, u64 in_param, u64 out_param,
+			 u32 in_modifier, u8 op_modifier, u16 op, u16 token,
+			 int event)
+{
+	struct mlx4_cmd *cmd = &mlx4_priv(dev)->cmd;
+	u32 __iomem *hcr = cmd->hcr;
+	int ret = -EIO;
+	unsigned long end;
+
+	mutex_lock(&dev->persist->device_state_mutex);
+	/* To avoid writing to unknown addresses after the device state was
+	  * changed to internal error and the chip was reset,
+	  * check the INTERNAL_ERROR flag which is updated under
+	  * device_state_mutex lock.
+	  */
+	if (pci_channel_offline(dev->persist->pdev) ||
+	    (dev->persist->state & MLX4_DEVICE_STATE_INTERNAL_ERROR)) {
+		/*
+		 * Device is going through error recovery
+		 * and cannot accept commands.
+		 */
+		goto out;
+	}
+
+	end = jiffies;
+	if (event)
+		end += msecs_to_jiffies(GO_BIT_TIMEOUT_MSECS);
+
+	while (cmd_pending(dev)) {
+		if (pci_channel_offline(dev->persist->pdev)) {
+			/*
+			 * Device is going through error recovery
+			 * and cannot accept commands.
+			 */
+			goto out;
+		}
+
+		if (time_after_eq(jiffies, end)) {
+			mlx4_err(dev, "%s:cmd_pending failed\n", __func__);
+			goto out;
+		}
+		cond_resched();
+	}
+
+	/*
+	 * We use writel (instead of something like memcpy_toio)
+	 * because writes of less than 32 bits to the HCR don't work
+	 * (and some architectures such as ia64 implement memcpy_toio
+	 * in terms of writeb).
+	 */
+	__raw_writel((__force u32) cpu_to_be32(in_param >> 32),		  hcr + 0);
+	__raw_writel((__force u32) cpu_to_be32(in_param & 0xfffffffful),  hcr + 1);
+	__raw_writel((__force u32) cpu_to_be32(in_modifier),		  hcr + 2);
+	__raw_writel((__force u32) cpu_to_be32(out_param >> 32),	  hcr + 3);
+	__raw_writel((__force u32) cpu_to_be32(out_param & 0xfffffffful), hcr + 4);
+	__raw_writel((__force u32) cpu_to_be32(token << 16),		  hcr + 5);
+
+	/* __raw_writel may not order writes. */
+	wmb();
+
+	__raw_writel((__force u32) cpu_to_be32((1 << HCR_GO_BIT)		|
+					       (cmd->toggle << HCR_T_BIT)	|
+					       (event ? (1 << HCR_E_BIT) : 0)	|
+					       (op_modifier << HCR_OPMOD_SHIFT) |
+					       op), hcr + 6);
+
+	cmd->toggle = cmd->toggle ^ 1;
+
+	ret = 0;
+
+out:
+	if (ret)
+		mlx4_warn(dev, "Could not post command 0x%x: ret=%d, in_param=0x%llx, in_mod=0x%x, op_mod=0x%x\n",
+			  op, ret, in_param, in_modifier, op_modifier);
+	mutex_unlock(&dev->persist->device_state_mutex);
+
+	return ret;
+}
+
+static int mlx4_slave_cmd(struct mlx4_dev *dev, u64 in_param, u64 *out_param,
+			  int out_is_imm, u32 in_modifier, u8 op_modifier,
+			  u16 op, unsigned long timeout)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_vhcr_cmd *vhcr = priv->mfunc.vhcr;
+	int ret;
+
+	mutex_lock(&priv->cmd.slave_cmd_mutex);
+
+	vhcr->in_param = cpu_to_be64(in_param);
+	vhcr->out_param = out_param ? cpu_to_be64(*out_param) : 0;
+	vhcr->in_modifier = cpu_to_be32(in_modifier);
+	vhcr->opcode = cpu_to_be16((((u16) op_modifier) << 12) | (op & 0xfff));
+	vhcr->token = cpu_to_be16(CMD_POLL_TOKEN);
+	vhcr->status = 0;
+	vhcr->flags = !!(priv->cmd.use_events) << 6;
+
+	if (mlx4_is_master(dev)) {
+		ret = mlx4_master_process_vhcr(dev, dev->caps.function, vhcr);
+		if (!ret) {
+			if (out_is_imm) {
+				if (out_param)
+					*out_param =
+						be64_to_cpu(vhcr->out_param);
+				else {
+					mlx4_err(dev, "response expected while output mailbox is NULL for command 0x%x\n",
+						 op);
+					vhcr->status = CMD_STAT_BAD_PARAM;
+				}
+			}
+			ret = mlx4_status_to_errno(vhcr->status);
+		}
+		if (ret &&
+		    dev->persist->state & MLX4_DEVICE_STATE_INTERNAL_ERROR)
+			ret = mlx4_internal_err_ret_value(dev, op, op_modifier);
+	} else {
+		ret = mlx4_comm_cmd(dev, MLX4_COMM_CMD_VHCR_POST, 0, op,
+				    MLX4_COMM_TIME + timeout);
+		if (!ret) {
+			if (out_is_imm) {
+				if (out_param)
+					*out_param =
+						be64_to_cpu(vhcr->out_param);
+				else {
+					mlx4_err(dev, "response expected while output mailbox is NULL for command 0x%x\n",
+						 op);
+					vhcr->status = CMD_STAT_BAD_PARAM;
+				}
+			}
+			ret = mlx4_status_to_errno(vhcr->status);
+		} else {
+			if (dev->persist->state &
+			    MLX4_DEVICE_STATE_INTERNAL_ERROR)
+				ret = mlx4_internal_err_ret_value(dev, op,
+								  op_modifier);
+			else
+				mlx4_err(dev, "failed execution of VHCR_POST command opcode 0x%x\n", op);
+		}
+	}
+
+	mutex_unlock(&priv->cmd.slave_cmd_mutex);
+	return ret;
+}
+
+static int mlx4_cmd_poll(struct mlx4_dev *dev, u64 in_param, u64 *out_param,
+			 int out_is_imm, u32 in_modifier, u8 op_modifier,
+			 u16 op, unsigned long timeout)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	void __iomem *hcr = priv->cmd.hcr;
+	int err = 0;
+	unsigned long end;
+	u32 stat;
+
+	down(&priv->cmd.poll_sem);
+
+	if (dev->persist->state & MLX4_DEVICE_STATE_INTERNAL_ERROR) {
+		/*
+		 * Device is going through error recovery
+		 * and cannot accept commands.
+		 */
+		err = mlx4_internal_err_ret_value(dev, op, op_modifier);
+		goto out;
+	}
+
+	if (out_is_imm && !out_param) {
+		mlx4_err(dev, "response expected while output mailbox is NULL for command 0x%x\n",
+			 op);
+		err = -EINVAL;
+		goto out;
+	}
+
+	err = mlx4_cmd_post(dev, in_param, out_param ? *out_param : 0,
+			    in_modifier, op_modifier, op, CMD_POLL_TOKEN, 0);
+	if (err)
+		goto out_reset;
+
+	end = msecs_to_jiffies(timeout) + jiffies;
+	while (cmd_pending(dev) && time_before(jiffies, end)) {
+		if (pci_channel_offline(dev->persist->pdev)) {
+			/*
+			 * Device is going through error recovery
+			 * and cannot accept commands.
+			 */
+			err = -EIO;
+			goto out_reset;
+		}
+
+		if (dev->persist->state & MLX4_DEVICE_STATE_INTERNAL_ERROR) {
+			err = mlx4_internal_err_ret_value(dev, op, op_modifier);
+			goto out;
+		}
+
+		cond_resched();
+	}
+
+	if (cmd_pending(dev)) {
+		mlx4_warn(dev, "command 0x%x timed out (go bit not cleared)\n",
+			  op);
+		err = -EIO;
+		goto out_reset;
+	}
+
+	if (out_is_imm)
+		*out_param =
+			(u64) be32_to_cpu((__force __be32)
+					  __raw_readl(hcr + HCR_OUT_PARAM_OFFSET)) << 32 |
+			(u64) be32_to_cpu((__force __be32)
+					  __raw_readl(hcr + HCR_OUT_PARAM_OFFSET + 4));
+	stat = be32_to_cpu((__force __be32)
+			   __raw_readl(hcr + HCR_STATUS_OFFSET)) >> 24;
+	err = mlx4_status_to_errno(stat);
+	if (err) {
+		mlx4_err(dev, "command 0x%x failed: fw status = 0x%x\n",
+			 op, stat);
+		if (mlx4_closing_cmd_fatal_error(op, stat))
+			goto out_reset;
+		goto out;
+	}
+
+out_reset:
+	if (err)
+		err = mlx4_cmd_reset_flow(dev, op, op_modifier, err);
+out:
+	up(&priv->cmd.poll_sem);
+	return err;
+}
+
+void mlx4_cmd_event(struct mlx4_dev *dev, u16 token, u8 status, u64 out_param)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_cmd_context *context =
+		&priv->cmd.context[token & priv->cmd.token_mask];
+
+	/* previously timed out command completing at long last */
+	if (token != context->token)
+		return;
+
+	context->fw_status = status;
+	context->result    = mlx4_status_to_errno(status);
+	context->out_param = out_param;
+
+	complete(&context->done);
+}
+
+static int mlx4_cmd_wait(struct mlx4_dev *dev, u64 in_param, u64 *out_param,
+			 int out_is_imm, u32 in_modifier, u8 op_modifier,
+			 u16 op, unsigned long timeout)
+{
+	struct mlx4_cmd *cmd = &mlx4_priv(dev)->cmd;
+	struct mlx4_cmd_context *context;
+	long ret_wait;
+	int err = 0;
+
+	down(&cmd->event_sem);
+
+	spin_lock(&cmd->context_lock);
+	BUG_ON(cmd->free_head < 0);
+	context = &cmd->context[cmd->free_head];
+	context->token += cmd->token_mask + 1;
+	cmd->free_head = context->next;
+	spin_unlock(&cmd->context_lock);
+
+	if (out_is_imm && !out_param) {
+		mlx4_err(dev, "response expected while output mailbox is NULL for command 0x%x\n",
+			 op);
+		err = -EINVAL;
+		goto out;
+	}
+
+	reinit_completion(&context->done);
+
+	err = mlx4_cmd_post(dev, in_param, out_param ? *out_param : 0,
+			    in_modifier, op_modifier, op, context->token, 1);
+	if (err)
+		goto out_reset;
+
+	if (op == MLX4_CMD_SENSE_PORT) {
+		ret_wait =
+			wait_for_completion_interruptible_timeout(&context->done,
+								  msecs_to_jiffies(timeout));
+		if (ret_wait < 0) {
+			context->fw_status = 0;
+			context->out_param = 0;
+			context->result = 0;
+		}
+	} else {
+		ret_wait = (long)wait_for_completion_timeout(&context->done,
+							     msecs_to_jiffies(timeout));
+	}
+	if (!ret_wait) {
+		mlx4_warn(dev, "command 0x%x timed out (go bit not cleared)\n",
+			  op);
+		if (op == MLX4_CMD_NOP) {
+			err = -EBUSY;
+			goto out;
+		} else {
+			err = -EIO;
+			goto out_reset;
+		}
+	}
+
+	err = context->result;
+	if (err) {
+		/* Since we do not want to have this error message always
+		 * displayed at driver start when there are ConnectX2 HCAs
+		 * on the host, we deprecate the error message for this
+		 * specific command/input_mod/opcode_mod/fw-status to be debug.
+		 */
+		if (op == MLX4_CMD_SET_PORT &&
+		    (in_modifier == 1 || in_modifier == 2) &&
+		    op_modifier == MLX4_SET_PORT_IB_OPCODE &&
+		    context->fw_status == CMD_STAT_BAD_SIZE)
+			mlx4_dbg(dev, "command 0x%x failed: fw status = 0x%x\n",
+				 op, context->fw_status);
+		else
+			mlx4_err(dev, "command 0x%x failed: fw status = 0x%x\n",
+				 op, context->fw_status);
+		if (dev->persist->state & MLX4_DEVICE_STATE_INTERNAL_ERROR)
+			err = mlx4_internal_err_ret_value(dev, op, op_modifier);
+		else if (mlx4_closing_cmd_fatal_error(op, context->fw_status))
+			goto out_reset;
+
+		goto out;
+	}
+
+	if (out_is_imm)
+		*out_param = context->out_param;
+
+out_reset:
+	if (err)
+		err = mlx4_cmd_reset_flow(dev, op, op_modifier, err);
+out:
+	spin_lock(&cmd->context_lock);
+	context->next = cmd->free_head;
+	cmd->free_head = context - cmd->context;
+	spin_unlock(&cmd->context_lock);
+
+	up(&cmd->event_sem);
+	return err;
+}
+
+int __mlx4_cmd(struct mlx4_dev *dev, u64 in_param, u64 *out_param,
+	       int out_is_imm, u32 in_modifier, u8 op_modifier,
+	       u16 op, unsigned long timeout, int native)
+{
+	if (pci_channel_offline(dev->persist->pdev))
+		return mlx4_cmd_reset_flow(dev, op, op_modifier, -EIO);
+
+	if (!mlx4_is_mfunc(dev) || (native && mlx4_is_master(dev))) {
+		int ret;
+
+		if (dev->persist->state & MLX4_DEVICE_STATE_INTERNAL_ERROR)
+			return mlx4_internal_err_ret_value(dev, op,
+							  op_modifier);
+		down_read(&mlx4_priv(dev)->cmd.switch_sem);
+		if (mlx4_priv(dev)->cmd.use_events)
+			ret = mlx4_cmd_wait(dev, in_param, out_param,
+					    out_is_imm, in_modifier,
+					    op_modifier, op, timeout);
+		else
+			ret = mlx4_cmd_poll(dev, in_param, out_param,
+					    out_is_imm, in_modifier,
+					    op_modifier, op, timeout);
+
+		up_read(&mlx4_priv(dev)->cmd.switch_sem);
+		return ret;
+	}
+	return mlx4_slave_cmd(dev, in_param, out_param, out_is_imm,
+			      in_modifier, op_modifier, op, timeout);
+}
+EXPORT_SYMBOL_GPL(__mlx4_cmd);
+
+
+int mlx4_ARM_COMM_CHANNEL(struct mlx4_dev *dev)
+{
+	return mlx4_cmd(dev, 0, 0, 0, MLX4_CMD_ARM_COMM_CHANNEL,
+			MLX4_CMD_TIME_CLASS_B, MLX4_CMD_NATIVE);
+}
+
+static int mlx4_ACCESS_MEM(struct mlx4_dev *dev, u64 master_addr,
+			   int slave, u64 slave_addr,
+			   int size, int is_read)
+{
+	u64 in_param;
+	u64 out_param;
+
+	if ((slave_addr & 0xfff) | (master_addr & 0xfff) |
+	    (slave & ~0x7f) | (size & 0xff)) {
+		mlx4_err(dev, "Bad access mem params - slave_addr:0x%llx master_addr:0x%llx slave_id:%d size:%d\n",
+			 slave_addr, master_addr, slave, size);
+		return -EINVAL;
+	}
+
+	if (is_read) {
+		in_param = (u64) slave | slave_addr;
+		out_param = (u64) dev->caps.function | master_addr;
+	} else {
+		in_param = (u64) dev->caps.function | master_addr;
+		out_param = (u64) slave | slave_addr;
+	}
+
+	return mlx4_cmd_imm(dev, in_param, &out_param, size, 0,
+			    MLX4_CMD_ACCESS_MEM,
+			    MLX4_CMD_TIME_CLASS_A, MLX4_CMD_NATIVE);
+}
+
+static int query_pkey_block(struct mlx4_dev *dev, u8 port, u16 index, u16 *pkey,
+			       struct mlx4_cmd_mailbox *inbox,
+			       struct mlx4_cmd_mailbox *outbox)
+{
+	struct ib_smp *in_mad = (struct ib_smp *)(inbox->buf);
+	struct ib_smp *out_mad = (struct ib_smp *)(outbox->buf);
+	int err;
+	int i;
+
+	if (index & 0x1f)
+		return -EINVAL;
+
+	in_mad->attr_mod = cpu_to_be32(index / 32);
+
+	err = mlx4_cmd_box(dev, inbox->dma, outbox->dma, port, 3,
+			   MLX4_CMD_MAD_IFC, MLX4_CMD_TIME_CLASS_C,
+			   MLX4_CMD_NATIVE);
+	if (err)
+		return err;
+
+	for (i = 0; i < 32; ++i)
+		pkey[i] = be16_to_cpu(((__be16 *) out_mad->data)[i]);
+
+	return err;
+}
+
+static int get_full_pkey_table(struct mlx4_dev *dev, u8 port, u16 *table,
+			       struct mlx4_cmd_mailbox *inbox,
+			       struct mlx4_cmd_mailbox *outbox)
+{
+	int i;
+	int err;
+
+	for (i = 0; i < dev->caps.pkey_table_len[port]; i += 32) {
+		err = query_pkey_block(dev, port, i, table + i, inbox, outbox);
+		if (err)
+			return err;
+	}
+
+	return 0;
+}
+#define PORT_CAPABILITY_LOCATION_IN_SMP 20
+#define PORT_STATE_OFFSET 32
+
+static enum ib_port_state vf_port_state(struct mlx4_dev *dev, int port, int vf)
+{
+	if (mlx4_get_slave_port_state(dev, vf, port) == SLAVE_PORT_UP)
+		return IB_PORT_ACTIVE;
+	else
+		return IB_PORT_DOWN;
+}
+
+static int mlx4_MAD_IFC_wrapper(struct mlx4_dev *dev, int slave,
+				struct mlx4_vhcr *vhcr,
+				struct mlx4_cmd_mailbox *inbox,
+				struct mlx4_cmd_mailbox *outbox,
+				struct mlx4_cmd_info *cmd)
+{
+	struct ib_smp *smp = inbox->buf;
+	u32 index;
+	u8 port, slave_port;
+	u8 opcode_modifier;
+	u16 *table;
+	int err;
+	int vidx, pidx;
+	int network_view;
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct ib_smp *outsmp = outbox->buf;
+	__be16 *outtab = (__be16 *)(outsmp->data);
+	__be32 slave_cap_mask;
+	__be64 slave_node_guid;
+
+	slave_port = vhcr->in_modifier;
+	port = mlx4_slave_convert_port(dev, slave, slave_port);
+
+	/* network-view bit is for driver use only, and should not be passed to FW */
+	opcode_modifier = vhcr->op_modifier & ~0x8; /* clear netw view bit */
+	network_view = !!(vhcr->op_modifier & 0x8);
+
+	if (smp->base_version == 1 &&
+	    smp->mgmt_class == IB_MGMT_CLASS_SUBN_LID_ROUTED &&
+	    smp->class_version == 1) {
+		/* host view is paravirtualized */
+		if (!network_view && smp->method == IB_MGMT_METHOD_GET) {
+			if (smp->attr_id == IB_SMP_ATTR_PKEY_TABLE) {
+				index = be32_to_cpu(smp->attr_mod);
+				if (port < 1 || port > dev->caps.num_ports)
+					return -EINVAL;
+				table = kcalloc((dev->caps.pkey_table_len[port] / 32) + 1,
+						sizeof(*table) * 32, GFP_KERNEL);
+
+				if (!table)
+					return -ENOMEM;
+				/* need to get the full pkey table because the paravirtualized
+				 * pkeys may be scattered among several pkey blocks.
+				 */
+				err = get_full_pkey_table(dev, port, table, inbox, outbox);
+				if (!err) {
+					for (vidx = index * 32; vidx < (index + 1) * 32; ++vidx) {
+						pidx = priv->virt2phys_pkey[slave][port - 1][vidx];
+						outtab[vidx % 32] = cpu_to_be16(table[pidx]);
+					}
+				}
+				kfree(table);
+				return err;
+			}
+			if (smp->attr_id == IB_SMP_ATTR_PORT_INFO) {
+				/*get the slave specific caps:*/
+				/*do the command */
+				smp->attr_mod = cpu_to_be32(port);
+				err = mlx4_cmd_box(dev, inbox->dma, outbox->dma,
+					    port, opcode_modifier,
+					    vhcr->op, MLX4_CMD_TIME_CLASS_C, MLX4_CMD_NATIVE);
+				/* modify the response for slaves */
+				if (!err && slave != mlx4_master_func_num(dev)) {
+					u8 *state = outsmp->data + PORT_STATE_OFFSET;
+
+					*state = (*state & 0xf0) | vf_port_state(dev, port, slave);
+					slave_cap_mask = priv->mfunc.master.slave_state[slave].ib_cap_mask[port];
+					memcpy(outsmp->data + PORT_CAPABILITY_LOCATION_IN_SMP, &slave_cap_mask, 4);
+				}
+				return err;
+			}
+			if (smp->attr_id == IB_SMP_ATTR_GUID_INFO) {
+				__be64 guid = mlx4_get_admin_guid(dev, slave,
+								  port);
+
+				/* set the PF admin guid to the FW/HW burned
+				 * GUID, if it wasn't yet set
+				 */
+				if (slave == 0 && guid == 0) {
+					smp->attr_mod = 0;
+					err = mlx4_cmd_box(dev,
+							   inbox->dma,
+							   outbox->dma,
+							   vhcr->in_modifier,
+							   opcode_modifier,
+							   vhcr->op,
+							   MLX4_CMD_TIME_CLASS_C,
+							   MLX4_CMD_NATIVE);
+					if (err)
+						return err;
+					mlx4_set_admin_guid(dev,
+							    *(__be64 *)outsmp->
+							    data, slave, port);
+				} else {
+					memcpy(outsmp->data, &guid, 8);
+				}
+
+				/* clean all other gids */
+				memset(outsmp->data + 8, 0, 56);
+				return 0;
+			}
+			if (smp->attr_id == IB_SMP_ATTR_NODE_INFO) {
+				err = mlx4_cmd_box(dev, inbox->dma, outbox->dma,
+					     port, opcode_modifier,
+					     vhcr->op, MLX4_CMD_TIME_CLASS_C, MLX4_CMD_NATIVE);
+				if (!err) {
+					slave_node_guid =  mlx4_get_slave_node_guid(dev, slave);
+					memcpy(outsmp->data + 12, &slave_node_guid, 8);
+				}
+				return err;
+			}
+		}
+	}
+
+	/* Non-privileged VFs are only allowed "host" view LID-routed 'Get' MADs.
+	 * These are the MADs used by ib verbs (such as ib_query_gids).
+	 */
+	if (slave != mlx4_master_func_num(dev) &&
+	    !mlx4_vf_smi_enabled(dev, slave, port)) {
+		if (!(smp->mgmt_class == IB_MGMT_CLASS_SUBN_LID_ROUTED &&
+		      smp->method == IB_MGMT_METHOD_GET) || network_view) {
+			mlx4_err(dev, "Unprivileged slave %d is trying to execute a Subnet MGMT MAD, class 0x%x, method 0x%x, view=%s for attr 0x%x. Rejecting\n",
+				 slave, smp->mgmt_class, smp->method,
+				 network_view ? "Network" : "Host",
+				 be16_to_cpu(smp->attr_id));
+			return -EPERM;
+		}
+	}
+
+	return mlx4_cmd_box(dev, inbox->dma, outbox->dma,
+				    vhcr->in_modifier, opcode_modifier,
+				    vhcr->op, MLX4_CMD_TIME_CLASS_C, MLX4_CMD_NATIVE);
+}
+
+static int mlx4_CMD_EPERM_wrapper(struct mlx4_dev *dev, int slave,
+		     struct mlx4_vhcr *vhcr,
+		     struct mlx4_cmd_mailbox *inbox,
+		     struct mlx4_cmd_mailbox *outbox,
+		     struct mlx4_cmd_info *cmd)
+{
+	return -EPERM;
+}
+
+int mlx4_DMA_wrapper(struct mlx4_dev *dev, int slave,
+		     struct mlx4_vhcr *vhcr,
+		     struct mlx4_cmd_mailbox *inbox,
+		     struct mlx4_cmd_mailbox *outbox,
+		     struct mlx4_cmd_info *cmd)
+{
+	u64 in_param;
+	u64 out_param;
+	int err;
+
+	in_param = cmd->has_inbox ? (u64) inbox->dma : vhcr->in_param;
+	out_param = cmd->has_outbox ? (u64) outbox->dma : vhcr->out_param;
+	if (cmd->encode_slave_id) {
+		in_param &= 0xffffffffffffff00ll;
+		in_param |= slave;
+	}
+
+	err = __mlx4_cmd(dev, in_param, &out_param, cmd->out_is_imm,
+			 vhcr->in_modifier, vhcr->op_modifier, vhcr->op,
+			 MLX4_CMD_TIME_CLASS_A, MLX4_CMD_NATIVE);
+
+	if (cmd->out_is_imm)
+		vhcr->out_param = out_param;
+
+	return err;
+}
+
+static struct mlx4_cmd_info cmd_info[] = {
+	{
+		.opcode = MLX4_CMD_QUERY_FW,
+		.has_inbox = false,
+		.has_outbox = true,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_QUERY_FW_wrapper
+	},
+	{
+		.opcode = MLX4_CMD_QUERY_HCA,
+		.has_inbox = false,
+		.has_outbox = true,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = NULL
+	},
+	{
+		.opcode = MLX4_CMD_QUERY_DEV_CAP,
+		.has_inbox = false,
+		.has_outbox = true,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_QUERY_DEV_CAP_wrapper
+	},
+	{
+		.opcode = MLX4_CMD_QUERY_FUNC_CAP,
+		.has_inbox = false,
+		.has_outbox = true,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_QUERY_FUNC_CAP_wrapper
+	},
+	{
+		.opcode = MLX4_CMD_QUERY_ADAPTER,
+		.has_inbox = false,
+		.has_outbox = true,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = NULL
+	},
+	{
+		.opcode = MLX4_CMD_INIT_PORT,
+		.has_inbox = false,
+		.has_outbox = false,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_INIT_PORT_wrapper
+	},
+	{
+		.opcode = MLX4_CMD_CLOSE_PORT,
+		.has_inbox = false,
+		.has_outbox = false,
+		.out_is_imm  = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_CLOSE_PORT_wrapper
+	},
+	{
+		.opcode = MLX4_CMD_QUERY_PORT,
+		.has_inbox = false,
+		.has_outbox = true,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_QUERY_PORT_wrapper
+	},
+	{
+		.opcode = MLX4_CMD_SET_PORT,
+		.has_inbox = true,
+		.has_outbox = false,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_SET_PORT_wrapper
+	},
+	{
+		.opcode = MLX4_CMD_MAP_EQ,
+		.has_inbox = false,
+		.has_outbox = false,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_MAP_EQ_wrapper
+	},
+	{
+		.opcode = MLX4_CMD_SW2HW_EQ,
+		.has_inbox = true,
+		.has_outbox = false,
+		.out_is_imm = false,
+		.encode_slave_id = true,
+		.verify = NULL,
+		.wrapper = mlx4_SW2HW_EQ_wrapper
+	},
+	{
+		.opcode = MLX4_CMD_HW_HEALTH_CHECK,
+		.has_inbox = false,
+		.has_outbox = false,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = NULL
+	},
+	{
+		.opcode = MLX4_CMD_NOP,
+		.has_inbox = false,
+		.has_outbox = false,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = NULL
+	},
+	{
+		.opcode = MLX4_CMD_CONFIG_DEV,
+		.has_inbox = false,
+		.has_outbox = true,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_CONFIG_DEV_wrapper
+	},
+	{
+		.opcode = MLX4_CMD_ALLOC_RES,
+		.has_inbox = false,
+		.has_outbox = false,
+		.out_is_imm = true,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_ALLOC_RES_wrapper
+	},
+	{
+		.opcode = MLX4_CMD_FREE_RES,
+		.has_inbox = false,
+		.has_outbox = false,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_FREE_RES_wrapper
+	},
+	{
+		.opcode = MLX4_CMD_SW2HW_MPT,
+		.has_inbox = true,
+		.has_outbox = false,
+		.out_is_imm = false,
+		.encode_slave_id = true,
+		.verify = NULL,
+		.wrapper = mlx4_SW2HW_MPT_wrapper
+	},
+	{
+		.opcode = MLX4_CMD_QUERY_MPT,
+		.has_inbox = false,
+		.has_outbox = true,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_QUERY_MPT_wrapper
+	},
+	{
+		.opcode = MLX4_CMD_HW2SW_MPT,
+		.has_inbox = false,
+		.has_outbox = false,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_HW2SW_MPT_wrapper
+	},
+	{
+		.opcode = MLX4_CMD_READ_MTT,
+		.has_inbox = false,
+		.has_outbox = true,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = NULL
+	},
+	{
+		.opcode = MLX4_CMD_WRITE_MTT,
+		.has_inbox = true,
+		.has_outbox = false,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_WRITE_MTT_wrapper
+	},
+	{
+		.opcode = MLX4_CMD_SYNC_TPT,
+		.has_inbox = true,
+		.has_outbox = false,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = NULL
+	},
+	{
+		.opcode = MLX4_CMD_HW2SW_EQ,
+		.has_inbox = false,
+		.has_outbox = false,
+		.out_is_imm = false,
+		.encode_slave_id = true,
+		.verify = NULL,
+		.wrapper = mlx4_HW2SW_EQ_wrapper
+	},
+	{
+		.opcode = MLX4_CMD_QUERY_EQ,
+		.has_inbox = false,
+		.has_outbox = true,
+		.out_is_imm = false,
+		.encode_slave_id = true,
+		.verify = NULL,
+		.wrapper = mlx4_QUERY_EQ_wrapper
+	},
+	{
+		.opcode = MLX4_CMD_SW2HW_CQ,
+		.has_inbox = true,
+		.has_outbox = false,
+		.out_is_imm = false,
+		.encode_slave_id = true,
+		.verify = NULL,
+		.wrapper = mlx4_SW2HW_CQ_wrapper
+	},
+	{
+		.opcode = MLX4_CMD_HW2SW_CQ,
+		.has_inbox = false,
+		.has_outbox = false,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_HW2SW_CQ_wrapper
+	},
+	{
+		.opcode = MLX4_CMD_QUERY_CQ,
+		.has_inbox = false,
+		.has_outbox = true,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_QUERY_CQ_wrapper
+	},
+	{
+		.opcode = MLX4_CMD_MODIFY_CQ,
+		.has_inbox = true,
+		.has_outbox = false,
+		.out_is_imm = true,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_MODIFY_CQ_wrapper
+	},
+	{
+		.opcode = MLX4_CMD_SW2HW_SRQ,
+		.has_inbox = true,
+		.has_outbox = false,
+		.out_is_imm = false,
+		.encode_slave_id = true,
+		.verify = NULL,
+		.wrapper = mlx4_SW2HW_SRQ_wrapper
+	},
+	{
+		.opcode = MLX4_CMD_HW2SW_SRQ,
+		.has_inbox = false,
+		.has_outbox = false,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_HW2SW_SRQ_wrapper
+	},
+	{
+		.opcode = MLX4_CMD_QUERY_SRQ,
+		.has_inbox = false,
+		.has_outbox = true,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_QUERY_SRQ_wrapper
+	},
+	{
+		.opcode = MLX4_CMD_ARM_SRQ,
+		.has_inbox = false,
+		.has_outbox = false,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_ARM_SRQ_wrapper
+	},
+	{
+		.opcode = MLX4_CMD_RST2INIT_QP,
+		.has_inbox = true,
+		.has_outbox = false,
+		.out_is_imm = false,
+		.encode_slave_id = true,
+		.verify = NULL,
+		.wrapper = mlx4_RST2INIT_QP_wrapper
+	},
+	{
+		.opcode = MLX4_CMD_INIT2INIT_QP,
+		.has_inbox = true,
+		.has_outbox = false,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_INIT2INIT_QP_wrapper
+	},
+	{
+		.opcode = MLX4_CMD_INIT2RTR_QP,
+		.has_inbox = true,
+		.has_outbox = false,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_INIT2RTR_QP_wrapper
+	},
+	{
+		.opcode = MLX4_CMD_RTR2RTS_QP,
+		.has_inbox = true,
+		.has_outbox = false,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_RTR2RTS_QP_wrapper
+	},
+	{
+		.opcode = MLX4_CMD_RTS2RTS_QP,
+		.has_inbox = true,
+		.has_outbox = false,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_RTS2RTS_QP_wrapper
+	},
+	{
+		.opcode = MLX4_CMD_SQERR2RTS_QP,
+		.has_inbox = true,
+		.has_outbox = false,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_SQERR2RTS_QP_wrapper
+	},
+	{
+		.opcode = MLX4_CMD_2ERR_QP,
+		.has_inbox = false,
+		.has_outbox = false,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_GEN_QP_wrapper
+	},
+	{
+		.opcode = MLX4_CMD_RTS2SQD_QP,
+		.has_inbox = false,
+		.has_outbox = false,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_GEN_QP_wrapper
+	},
+	{
+		.opcode = MLX4_CMD_SQD2SQD_QP,
+		.has_inbox = true,
+		.has_outbox = false,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_SQD2SQD_QP_wrapper
+	},
+	{
+		.opcode = MLX4_CMD_SQD2RTS_QP,
+		.has_inbox = true,
+		.has_outbox = false,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_SQD2RTS_QP_wrapper
+	},
+	{
+		.opcode = MLX4_CMD_2RST_QP,
+		.has_inbox = false,
+		.has_outbox = false,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_2RST_QP_wrapper
+	},
+	{
+		.opcode = MLX4_CMD_QUERY_QP,
+		.has_inbox = false,
+		.has_outbox = true,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_GEN_QP_wrapper
+	},
+	{
+		.opcode = MLX4_CMD_SUSPEND_QP,
+		.has_inbox = false,
+		.has_outbox = false,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_GEN_QP_wrapper
+	},
+	{
+		.opcode = MLX4_CMD_UNSUSPEND_QP,
+		.has_inbox = false,
+		.has_outbox = false,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_GEN_QP_wrapper
+	},
+	{
+		.opcode = MLX4_CMD_UPDATE_QP,
+		.has_inbox = true,
+		.has_outbox = false,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_UPDATE_QP_wrapper
+	},
+	{
+		.opcode = MLX4_CMD_GET_OP_REQ,
+		.has_inbox = false,
+		.has_outbox = false,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_CMD_EPERM_wrapper,
+	},
+	{
+		.opcode = MLX4_CMD_ALLOCATE_VPP,
+		.has_inbox = false,
+		.has_outbox = true,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_CMD_EPERM_wrapper,
+	},
+	{
+		.opcode = MLX4_CMD_SET_VPORT_QOS,
+		.has_inbox = false,
+		.has_outbox = true,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_CMD_EPERM_wrapper,
+	},
+	{
+		.opcode = MLX4_CMD_CONF_SPECIAL_QP,
+		.has_inbox = false,
+		.has_outbox = false,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL, /* XXX verify: only demux can do this */
+		.wrapper = NULL
+	},
+	{
+		.opcode = MLX4_CMD_MAD_IFC,
+		.has_inbox = true,
+		.has_outbox = true,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_MAD_IFC_wrapper
+	},
+	{
+		.opcode = MLX4_CMD_MAD_DEMUX,
+		.has_inbox = false,
+		.has_outbox = false,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_CMD_EPERM_wrapper
+	},
+	{
+		.opcode = MLX4_CMD_QUERY_IF_STAT,
+		.has_inbox = false,
+		.has_outbox = true,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_QUERY_IF_STAT_wrapper
+	},
+	{
+		.opcode = MLX4_CMD_ACCESS_REG,
+		.has_inbox = true,
+		.has_outbox = true,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_ACCESS_REG_wrapper,
+	},
+	{
+		.opcode = MLX4_CMD_CONGESTION_CTRL_OPCODE,
+		.has_inbox = false,
+		.has_outbox = false,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_CMD_EPERM_wrapper,
+	},
+	/* Native multicast commands are not available for guests */
+	{
+		.opcode = MLX4_CMD_QP_ATTACH,
+		.has_inbox = true,
+		.has_outbox = false,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_QP_ATTACH_wrapper
+	},
+	{
+		.opcode = MLX4_CMD_PROMISC,
+		.has_inbox = false,
+		.has_outbox = false,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_PROMISC_wrapper
+	},
+	/* Ethernet specific commands */
+	{
+		.opcode = MLX4_CMD_SET_VLAN_FLTR,
+		.has_inbox = true,
+		.has_outbox = false,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_SET_VLAN_FLTR_wrapper
+	},
+	{
+		.opcode = MLX4_CMD_SET_MCAST_FLTR,
+		.has_inbox = false,
+		.has_outbox = false,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_SET_MCAST_FLTR_wrapper
+	},
+	{
+		.opcode = MLX4_CMD_DUMP_ETH_STATS,
+		.has_inbox = false,
+		.has_outbox = true,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_DUMP_ETH_STATS_wrapper
+	},
+	{
+		.opcode = MLX4_CMD_INFORM_FLR_DONE,
+		.has_inbox = false,
+		.has_outbox = false,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = NULL
+	},
+	/* flow steering commands */
+	{
+		.opcode = MLX4_QP_FLOW_STEERING_ATTACH,
+		.has_inbox = true,
+		.has_outbox = false,
+		.out_is_imm = true,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_QP_FLOW_STEERING_ATTACH_wrapper
+	},
+	{
+		.opcode = MLX4_QP_FLOW_STEERING_DETACH,
+		.has_inbox = false,
+		.has_outbox = false,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_QP_FLOW_STEERING_DETACH_wrapper
+	},
+	{
+		.opcode = MLX4_FLOW_STEERING_IB_UC_QP_RANGE,
+		.has_inbox = false,
+		.has_outbox = false,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_CMD_EPERM_wrapper
+	},
+	{
+		.opcode = MLX4_CMD_VIRT_PORT_MAP,
+		.has_inbox = false,
+		.has_outbox = false,
+		.out_is_imm = false,
+		.encode_slave_id = false,
+		.verify = NULL,
+		.wrapper = mlx4_CMD_EPERM_wrapper
+	},
+};
+
+static int mlx4_master_process_vhcr(struct mlx4_dev *dev, int slave,
+				    struct mlx4_vhcr_cmd *in_vhcr)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_cmd_info *cmd = NULL;
+	struct mlx4_vhcr_cmd *vhcr_cmd = in_vhcr ? in_vhcr : priv->mfunc.vhcr;
+	struct mlx4_vhcr *vhcr;
+	struct mlx4_cmd_mailbox *inbox = NULL;
+	struct mlx4_cmd_mailbox *outbox = NULL;
+	u64 in_param;
+	u64 out_param;
+	int ret = 0;
+	int i;
+	int err = 0;
+
+	/* Create sw representation of Virtual HCR */
+	vhcr = kzalloc(sizeof(struct mlx4_vhcr), GFP_KERNEL);
+	if (!vhcr)
+		return -ENOMEM;
+
+	/* DMA in the vHCR */
+	if (!in_vhcr) {
+		ret = mlx4_ACCESS_MEM(dev, priv->mfunc.vhcr_dma, slave,
+				      priv->mfunc.master.slave_state[slave].vhcr_dma,
+				      ALIGN(sizeof(struct mlx4_vhcr_cmd),
+					    MLX4_ACCESS_MEM_ALIGN), 1);
+		if (ret) {
+			if (!(dev->persist->state &
+			    MLX4_DEVICE_STATE_INTERNAL_ERROR))
+				mlx4_err(dev, "%s: Failed reading vhcr ret: 0x%x\n",
+					 __func__, ret);
+			kfree(vhcr);
+			return ret;
+		}
+	}
+
+	/* Fill SW VHCR fields */
+	vhcr->in_param = be64_to_cpu(vhcr_cmd->in_param);
+	vhcr->out_param = be64_to_cpu(vhcr_cmd->out_param);
+	vhcr->in_modifier = be32_to_cpu(vhcr_cmd->in_modifier);
+	vhcr->token = be16_to_cpu(vhcr_cmd->token);
+	vhcr->op = be16_to_cpu(vhcr_cmd->opcode) & 0xfff;
+	vhcr->op_modifier = (u8) (be16_to_cpu(vhcr_cmd->opcode) >> 12);
+	vhcr->e_bit = vhcr_cmd->flags & (1 << 6);
+
+	/* Lookup command */
+	for (i = 0; i < ARRAY_SIZE(cmd_info); ++i) {
+		if (vhcr->op == cmd_info[i].opcode) {
+			cmd = &cmd_info[i];
+			break;
+		}
+	}
+	if (!cmd) {
+		mlx4_err(dev, "Unknown command:0x%x accepted from slave:%d\n",
+			 vhcr->op, slave);
+		vhcr_cmd->status = CMD_STAT_BAD_PARAM;
+		goto out_status;
+	}
+
+	/* Read inbox */
+	if (cmd->has_inbox) {
+		vhcr->in_param &= INBOX_MASK;
+		inbox = mlx4_alloc_cmd_mailbox(dev);
+		if (IS_ERR(inbox)) {
+			vhcr_cmd->status = CMD_STAT_BAD_SIZE;
+			inbox = NULL;
+			goto out_status;
+		}
+
+		ret = mlx4_ACCESS_MEM(dev, inbox->dma, slave,
+				      vhcr->in_param,
+				      MLX4_MAILBOX_SIZE, 1);
+		if (ret) {
+			if (!(dev->persist->state &
+			    MLX4_DEVICE_STATE_INTERNAL_ERROR))
+				mlx4_err(dev, "%s: Failed reading inbox (cmd:0x%x)\n",
+					 __func__, cmd->opcode);
+			vhcr_cmd->status = CMD_STAT_INTERNAL_ERR;
+			goto out_status;
+		}
+	}
+
+	/* Apply permission and bound checks if applicable */
+	if (cmd->verify && cmd->verify(dev, slave, vhcr, inbox)) {
+		mlx4_warn(dev, "Command:0x%x from slave: %d failed protection checks for resource_id:%d\n",
+			  vhcr->op, slave, vhcr->in_modifier);
+		vhcr_cmd->status = CMD_STAT_BAD_OP;
+		goto out_status;
+	}
+
+	/* Allocate outbox */
+	if (cmd->has_outbox) {
+		outbox = mlx4_alloc_cmd_mailbox(dev);
+		if (IS_ERR(outbox)) {
+			vhcr_cmd->status = CMD_STAT_BAD_SIZE;
+			outbox = NULL;
+			goto out_status;
+		}
+	}
+
+	/* Execute the command! */
+	if (cmd->wrapper) {
+		err = cmd->wrapper(dev, slave, vhcr, inbox, outbox,
+				   cmd);
+		if (cmd->out_is_imm)
+			vhcr_cmd->out_param = cpu_to_be64(vhcr->out_param);
+	} else {
+		in_param = cmd->has_inbox ? (u64) inbox->dma :
+			vhcr->in_param;
+		out_param = cmd->has_outbox ? (u64) outbox->dma :
+			vhcr->out_param;
+		err = __mlx4_cmd(dev, in_param, &out_param,
+				 cmd->out_is_imm, vhcr->in_modifier,
+				 vhcr->op_modifier, vhcr->op,
+				 MLX4_CMD_TIME_CLASS_A,
+				 MLX4_CMD_NATIVE);
+
+		if (cmd->out_is_imm) {
+			vhcr->out_param = out_param;
+			vhcr_cmd->out_param = cpu_to_be64(vhcr->out_param);
+		}
+	}
+
+	if (err) {
+		if (!(dev->persist->state & MLX4_DEVICE_STATE_INTERNAL_ERROR)) {
+			if (vhcr->op == MLX4_CMD_ALLOC_RES &&
+			    (vhcr->in_modifier & 0xff) == RES_COUNTER &&
+			    err == -EDQUOT)
+				mlx4_dbg(dev,
+					 "Unable to allocate counter for slave %d (%d)\n",
+					 slave, err);
+			else
+				mlx4_warn(dev, "vhcr command:0x%x slave:%d failed with error:%d, status %d\n",
+					  vhcr->op, slave, vhcr->errno, err);
+		}
+		vhcr_cmd->status = mlx4_errno_to_status(err);
+		goto out_status;
+	}
+
+
+	/* Write outbox if command completed successfully */
+	if (cmd->has_outbox && !vhcr_cmd->status) {
+		ret = mlx4_ACCESS_MEM(dev, outbox->dma, slave,
+				      vhcr->out_param,
+				      MLX4_MAILBOX_SIZE, MLX4_CMD_WRAPPED);
+		if (ret) {
+			/* If we failed to write back the outbox after the
+			 *command was successfully executed, we must fail this
+			 * slave, as it is now in undefined state */
+			if (!(dev->persist->state &
+			    MLX4_DEVICE_STATE_INTERNAL_ERROR))
+				mlx4_err(dev, "%s:Failed writing outbox\n", __func__);
+			goto out;
+		}
+	}
+
+out_status:
+	/* DMA back vhcr result */
+	if (!in_vhcr) {
+		ret = mlx4_ACCESS_MEM(dev, priv->mfunc.vhcr_dma, slave,
+				      priv->mfunc.master.slave_state[slave].vhcr_dma,
+				      ALIGN(sizeof(struct mlx4_vhcr),
+					    MLX4_ACCESS_MEM_ALIGN),
+				      MLX4_CMD_WRAPPED);
+		if (ret)
+			mlx4_err(dev, "%s:Failed writing vhcr result\n",
+				 __func__);
+		else if (vhcr->e_bit &&
+			 mlx4_GEN_EQE(dev, slave, &priv->mfunc.master.cmd_eqe))
+				mlx4_warn(dev, "Failed to generate command completion eqe for slave %d\n",
+					  slave);
+	}
+
+out:
+	kfree(vhcr);
+	mlx4_free_cmd_mailbox(dev, inbox);
+	mlx4_free_cmd_mailbox(dev, outbox);
+	return ret;
+}
+
+static int mlx4_master_immediate_activate_vlan_qos(struct mlx4_priv *priv,
+					    int slave, int port)
+{
+	struct mlx4_vport_oper_state *vp_oper;
+	struct mlx4_vport_state *vp_admin;
+	struct mlx4_vf_immed_vlan_work *work;
+	struct mlx4_dev *dev = &(priv->dev);
+	int err;
+	int admin_vlan_ix = NO_INDX;
+
+	vp_oper = &priv->mfunc.master.vf_oper[slave].vport[port];
+	vp_admin = &priv->mfunc.master.vf_admin[slave].vport[port];
+
+	if (vp_oper->state.default_vlan == vp_admin->default_vlan &&
+	    vp_oper->state.default_qos == vp_admin->default_qos &&
+	    vp_oper->state.vlan_proto == vp_admin->vlan_proto &&
+	    vp_oper->state.link_state == vp_admin->link_state &&
+	    vp_oper->state.qos_vport == vp_admin->qos_vport)
+		return 0;
+
+	if (!(priv->mfunc.master.slave_state[slave].active &&
+	      dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_UPDATE_QP)) {
+		/* even if the UPDATE_QP command isn't supported, we still want
+		 * to set this VF link according to the admin directive
+		 */
+		vp_oper->state.link_state = vp_admin->link_state;
+		return -1;
+	}
+
+	mlx4_dbg(dev, "updating immediately admin params slave %d port %d\n",
+		 slave, port);
+	mlx4_dbg(dev, "vlan %d QoS %d link down %d\n",
+		 vp_admin->default_vlan, vp_admin->default_qos,
+		 vp_admin->link_state);
+
+	work = kzalloc(sizeof(*work), GFP_KERNEL);
+	if (!work)
+		return -ENOMEM;
+
+	if (vp_oper->state.default_vlan != vp_admin->default_vlan) {
+		if (MLX4_VGT != vp_admin->default_vlan) {
+			err = __mlx4_register_vlan(&priv->dev, port,
+						   vp_admin->default_vlan,
+						   &admin_vlan_ix);
+			if (err) {
+				kfree(work);
+				mlx4_warn(&priv->dev,
+					  "No vlan resources slave %d, port %d\n",
+					  slave, port);
+				return err;
+			}
+		} else {
+			admin_vlan_ix = NO_INDX;
+		}
+		work->flags |= MLX4_VF_IMMED_VLAN_FLAG_VLAN;
+		mlx4_dbg(&priv->dev,
+			 "alloc vlan %d idx  %d slave %d port %d\n",
+			 (int)(vp_admin->default_vlan),
+			 admin_vlan_ix, slave, port);
+	}
+
+	/* save original vlan ix and vlan id */
+	work->orig_vlan_id = vp_oper->state.default_vlan;
+	work->orig_vlan_ix = vp_oper->vlan_idx;
+
+	/* handle new qos */
+	if (vp_oper->state.default_qos != vp_admin->default_qos)
+		work->flags |= MLX4_VF_IMMED_VLAN_FLAG_QOS;
+
+	if (work->flags & MLX4_VF_IMMED_VLAN_FLAG_VLAN)
+		vp_oper->vlan_idx = admin_vlan_ix;
+
+	vp_oper->state.default_vlan = vp_admin->default_vlan;
+	vp_oper->state.default_qos = vp_admin->default_qos;
+	vp_oper->state.vlan_proto = vp_admin->vlan_proto;
+	vp_oper->state.link_state = vp_admin->link_state;
+	vp_oper->state.qos_vport = vp_admin->qos_vport;
+
+	if (vp_admin->link_state == IFLA_VF_LINK_STATE_DISABLE)
+		work->flags |= MLX4_VF_IMMED_VLAN_FLAG_LINK_DISABLE;
+
+	/* iterate over QPs owned by this slave, using UPDATE_QP */
+	work->port = port;
+	work->slave = slave;
+	work->qos = vp_oper->state.default_qos;
+	work->qos_vport = vp_oper->state.qos_vport;
+	work->vlan_id = vp_oper->state.default_vlan;
+	work->vlan_ix = vp_oper->vlan_idx;
+	work->vlan_proto = vp_oper->state.vlan_proto;
+	work->priv = priv;
+	INIT_WORK(&work->work, mlx4_vf_immed_vlan_work_handler);
+	queue_work(priv->mfunc.master.comm_wq, &work->work);
+
+	return 0;
+}
+
+static void mlx4_set_default_port_qos(struct mlx4_dev *dev, int port)
+{
+	struct mlx4_qos_manager *port_qos_ctl;
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	port_qos_ctl = &priv->mfunc.master.qos_ctl[port];
+	bitmap_zero(port_qos_ctl->priority_bm, MLX4_NUM_UP);
+
+	/* Enable only default prio at PF init routine */
+	set_bit(MLX4_DEFAULT_QOS_PRIO, port_qos_ctl->priority_bm);
+}
+
+static void mlx4_allocate_port_vpps(struct mlx4_dev *dev, int port)
+{
+	int i;
+	int err;
+	int num_vfs;
+	u16 available_vpp;
+	u8 vpp_param[MLX4_NUM_UP];
+	struct mlx4_qos_manager *port_qos;
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	err = mlx4_ALLOCATE_VPP_get(dev, port, &available_vpp, vpp_param);
+	if (err) {
+		mlx4_info(dev, "Failed query available VPPs\n");
+		return;
+	}
+
+	port_qos = &priv->mfunc.master.qos_ctl[port];
+	num_vfs = (available_vpp /
+		   bitmap_weight(port_qos->priority_bm, MLX4_NUM_UP));
+
+	for (i = 0; i < MLX4_NUM_UP; i++) {
+		if (test_bit(i, port_qos->priority_bm))
+			vpp_param[i] = num_vfs;
+	}
+
+	err = mlx4_ALLOCATE_VPP_set(dev, port, vpp_param);
+	if (err) {
+		mlx4_info(dev, "Failed allocating VPPs\n");
+		return;
+	}
+
+	/* Query actual allocated VPP, just to make sure */
+	err = mlx4_ALLOCATE_VPP_get(dev, port, &available_vpp, vpp_param);
+	if (err) {
+		mlx4_info(dev, "Failed query available VPPs\n");
+		return;
+	}
+
+	port_qos->num_of_qos_vfs = num_vfs;
+	mlx4_dbg(dev, "Port %d Available VPPs %d\n", port, available_vpp);
+
+	for (i = 0; i < MLX4_NUM_UP; i++)
+		mlx4_dbg(dev, "Port %d UP %d Allocated %d VPPs\n", port, i,
+			 vpp_param[i]);
+}
+
+static int mlx4_master_activate_admin_state(struct mlx4_priv *priv, int slave)
+{
+	int p, port, err;
+	struct mlx4_vport_state *vp_admin;
+	struct mlx4_vport_oper_state *vp_oper;
+	struct mlx4_slave_state *slave_state =
+		&priv->mfunc.master.slave_state[slave];
+	struct mlx4_active_ports actv_ports = mlx4_get_active_ports(
+			&priv->dev, slave);
+
+	for_each_set_bit(p, actv_ports.ports, priv->dev.caps.num_ports) {
+		port = p + 1;
+		priv->mfunc.master.vf_oper[slave].smi_enabled[port] =
+			priv->mfunc.master.vf_admin[slave].enable_smi[port];
+		vp_oper = &priv->mfunc.master.vf_oper[slave].vport[port];
+		vp_admin = &priv->mfunc.master.vf_admin[slave].vport[port];
+		if (vp_admin->vlan_proto != htons(ETH_P_8021AD) ||
+		    slave_state->vst_qinq_supported) {
+			vp_oper->state.vlan_proto   = vp_admin->vlan_proto;
+			vp_oper->state.default_vlan = vp_admin->default_vlan;
+			vp_oper->state.default_qos  = vp_admin->default_qos;
+		}
+		vp_oper->state.link_state = vp_admin->link_state;
+		vp_oper->state.mac        = vp_admin->mac;
+		vp_oper->state.spoofchk   = vp_admin->spoofchk;
+		vp_oper->state.tx_rate    = vp_admin->tx_rate;
+		vp_oper->state.qos_vport  = vp_admin->qos_vport;
+		vp_oper->state.guid       = vp_admin->guid;
+
+		if (MLX4_VGT != vp_admin->default_vlan) {
+			err = __mlx4_register_vlan(&priv->dev, port,
+						   vp_admin->default_vlan, &(vp_oper->vlan_idx));
+			if (err) {
+				vp_oper->vlan_idx = NO_INDX;
+				vp_oper->state.default_vlan = MLX4_VGT;
+				vp_oper->state.vlan_proto = htons(ETH_P_8021Q);
+				mlx4_warn(&priv->dev,
+					  "No vlan resources slave %d, port %d\n",
+					  slave, port);
+				return err;
+			}
+			mlx4_dbg(&priv->dev, "alloc vlan %d idx  %d slave %d port %d\n",
+				 (int)(vp_oper->state.default_vlan),
+				 vp_oper->vlan_idx, slave, port);
+		}
+		if (vp_admin->spoofchk) {
+			vp_oper->mac_idx = __mlx4_register_mac(&priv->dev,
+							       port,
+							       vp_admin->mac);
+			if (0 > vp_oper->mac_idx) {
+				err = vp_oper->mac_idx;
+				vp_oper->mac_idx = NO_INDX;
+				mlx4_warn(&priv->dev,
+					  "No mac resources slave %d, port %d\n",
+					  slave, port);
+				return err;
+			}
+			mlx4_dbg(&priv->dev, "alloc mac %llx idx  %d slave %d port %d\n",
+				 vp_oper->state.mac, vp_oper->mac_idx, slave, port);
+		}
+	}
+	return 0;
+}
+
+static void mlx4_master_deactivate_admin_state(struct mlx4_priv *priv, int slave)
+{
+	int p, port;
+	struct mlx4_vport_oper_state *vp_oper;
+	struct mlx4_active_ports actv_ports = mlx4_get_active_ports(
+			&priv->dev, slave);
+
+	for_each_set_bit(p, actv_ports.ports, priv->dev.caps.num_ports) {
+		port = p + 1;
+		priv->mfunc.master.vf_oper[slave].smi_enabled[port] =
+			MLX4_VF_SMI_DISABLED;
+		vp_oper = &priv->mfunc.master.vf_oper[slave].vport[port];
+		if (NO_INDX != vp_oper->vlan_idx) {
+			__mlx4_unregister_vlan(&priv->dev,
+					       port, vp_oper->state.default_vlan);
+			vp_oper->vlan_idx = NO_INDX;
+		}
+		if (NO_INDX != vp_oper->mac_idx) {
+			__mlx4_unregister_mac(&priv->dev, port, vp_oper->state.mac);
+			vp_oper->mac_idx = NO_INDX;
+		}
+	}
+	return;
+}
+
+static void mlx4_master_do_cmd(struct mlx4_dev *dev, int slave, u8 cmd,
+			       u16 param, u8 toggle)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_slave_state *slave_state = priv->mfunc.master.slave_state;
+	u32 reply;
+	u8 is_going_down = 0;
+	int i;
+	unsigned long flags;
+
+	slave_state[slave].comm_toggle ^= 1;
+	reply = (u32) slave_state[slave].comm_toggle << 31;
+	if (toggle != slave_state[slave].comm_toggle) {
+		mlx4_warn(dev, "Incorrect toggle %d from slave %d. *** MASTER STATE COMPROMISED ***\n",
+			  toggle, slave);
+		goto reset_slave;
+	}
+	if (cmd == MLX4_COMM_CMD_RESET) {
+		mlx4_warn(dev, "Received reset from slave:%d\n", slave);
+		slave_state[slave].active = false;
+		slave_state[slave].old_vlan_api = false;
+		slave_state[slave].vst_qinq_supported = false;
+		mlx4_master_deactivate_admin_state(priv, slave);
+		for (i = 0; i < MLX4_EVENT_TYPES_NUM; ++i) {
+				slave_state[slave].event_eq[i].eqn = -1;
+				slave_state[slave].event_eq[i].token = 0;
+		}
+		/*check if we are in the middle of FLR process,
+		if so return "retry" status to the slave*/
+		if (MLX4_COMM_CMD_FLR == slave_state[slave].last_cmd)
+			goto inform_slave_state;
+
+		mlx4_dispatch_event(dev, MLX4_DEV_EVENT_SLAVE_SHUTDOWN, slave);
+
+		/* write the version in the event field */
+		reply |= mlx4_comm_get_version();
+
+		goto reset_slave;
+	}
+	/*command from slave in the middle of FLR*/
+	if (cmd != MLX4_COMM_CMD_RESET &&
+	    MLX4_COMM_CMD_FLR == slave_state[slave].last_cmd) {
+		mlx4_warn(dev, "slave:%d is Trying to run cmd(0x%x) in the middle of FLR\n",
+			  slave, cmd);
+		return;
+	}
+
+	switch (cmd) {
+	case MLX4_COMM_CMD_VHCR0:
+		if (slave_state[slave].last_cmd != MLX4_COMM_CMD_RESET)
+			goto reset_slave;
+		slave_state[slave].vhcr_dma = ((u64) param) << 48;
+		priv->mfunc.master.slave_state[slave].cookie = 0;
+		break;
+	case MLX4_COMM_CMD_VHCR1:
+		if (slave_state[slave].last_cmd != MLX4_COMM_CMD_VHCR0)
+			goto reset_slave;
+		slave_state[slave].vhcr_dma |= ((u64) param) << 32;
+		break;
+	case MLX4_COMM_CMD_VHCR2:
+		if (slave_state[slave].last_cmd != MLX4_COMM_CMD_VHCR1)
+			goto reset_slave;
+		slave_state[slave].vhcr_dma |= ((u64) param) << 16;
+		break;
+	case MLX4_COMM_CMD_VHCR_EN:
+		if (slave_state[slave].last_cmd != MLX4_COMM_CMD_VHCR2)
+			goto reset_slave;
+		slave_state[slave].vhcr_dma |= param;
+		if (mlx4_master_activate_admin_state(priv, slave))
+				goto reset_slave;
+		slave_state[slave].active = true;
+		mlx4_dispatch_event(dev, MLX4_DEV_EVENT_SLAVE_INIT, slave);
+		break;
+	case MLX4_COMM_CMD_VHCR_POST:
+		if ((slave_state[slave].last_cmd != MLX4_COMM_CMD_VHCR_EN) &&
+		    (slave_state[slave].last_cmd != MLX4_COMM_CMD_VHCR_POST)) {
+			mlx4_warn(dev, "slave:%d is out of sync, cmd=0x%x, last command=0x%x, reset is needed\n",
+				  slave, cmd, slave_state[slave].last_cmd);
+			goto reset_slave;
+		}
+
+		mutex_lock(&priv->cmd.slave_cmd_mutex);
+		if (mlx4_master_process_vhcr(dev, slave, NULL)) {
+			mlx4_err(dev, "Failed processing vhcr for slave:%d, resetting slave\n",
+				 slave);
+			mutex_unlock(&priv->cmd.slave_cmd_mutex);
+			goto reset_slave;
+		}
+		mutex_unlock(&priv->cmd.slave_cmd_mutex);
+		break;
+	default:
+		mlx4_warn(dev, "Bad comm cmd:%d from slave:%d\n", cmd, slave);
+		goto reset_slave;
+	}
+	spin_lock_irqsave(&priv->mfunc.master.slave_state_lock, flags);
+	if (!slave_state[slave].is_slave_going_down)
+		slave_state[slave].last_cmd = cmd;
+	else
+		is_going_down = 1;
+	spin_unlock_irqrestore(&priv->mfunc.master.slave_state_lock, flags);
+	if (is_going_down) {
+		mlx4_warn(dev, "Slave is going down aborting command(%d) executing from slave:%d\n",
+			  cmd, slave);
+		return;
+	}
+	__raw_writel((__force u32) cpu_to_be32(reply),
+		     &priv->mfunc.comm[slave].slave_read);
+
+	return;
+
+reset_slave:
+	/* cleanup any slave resources */
+	if (dev->persist->interface_state & MLX4_INTERFACE_STATE_UP)
+		mlx4_delete_all_resources_for_slave(dev, slave);
+
+	if (cmd != MLX4_COMM_CMD_RESET) {
+		mlx4_warn(dev, "Turn on internal error to force reset, slave=%d, cmd=0x%x\n",
+			  slave, cmd);
+		/* Turn on internal error letting slave reset itself immeditaly,
+		 * otherwise it might take till timeout on command is passed
+		 */
+		reply |= ((u32)COMM_CHAN_EVENT_INTERNAL_ERR);
+	}
+
+	spin_lock_irqsave(&priv->mfunc.master.slave_state_lock, flags);
+	if (!slave_state[slave].is_slave_going_down)
+		slave_state[slave].last_cmd = MLX4_COMM_CMD_RESET;
+	spin_unlock_irqrestore(&priv->mfunc.master.slave_state_lock, flags);
+	/*with slave in the middle of flr, no need to clean resources again.*/
+inform_slave_state:
+	memset(&slave_state[slave].event_eq, 0,
+	       sizeof(struct mlx4_slave_event_eq_info));
+	__raw_writel((__force u32) cpu_to_be32(reply),
+		     &priv->mfunc.comm[slave].slave_read);
+	wmb();
+}
+
+/* master command processing */
+void mlx4_master_comm_channel(struct work_struct *work)
+{
+	struct mlx4_mfunc_master_ctx *master =
+		container_of(work,
+			     struct mlx4_mfunc_master_ctx,
+			     comm_work);
+	struct mlx4_mfunc *mfunc =
+		container_of(master, struct mlx4_mfunc, master);
+	struct mlx4_priv *priv =
+		container_of(mfunc, struct mlx4_priv, mfunc);
+	struct mlx4_dev *dev = &priv->dev;
+	u32 lbit_vec[COMM_CHANNEL_BIT_ARRAY_SIZE];
+	u32 nmbr_bits;
+	u32 comm_cmd;
+	int i, slave;
+	int toggle;
+	bool first = true;
+	int served = 0;
+	int reported = 0;
+	u32 slt;
+
+	for (i = 0; i < COMM_CHANNEL_BIT_ARRAY_SIZE; i++)
+		lbit_vec[i] = be32_to_cpu(master->comm_arm_bit_vector[i]);
+	nmbr_bits = dev->persist->num_vfs + 1;
+	if (++master->next_slave >= nmbr_bits)
+		master->next_slave = 0;
+	slave = master->next_slave;
+	while (true) {
+		slave = find_next_bit((const unsigned long *)&lbit_vec, nmbr_bits, slave);
+		if  (!first && slave >= master->next_slave)
+			break;
+		if (slave == nmbr_bits) {
+			if (!first)
+				break;
+			first = false;
+			slave = 0;
+			continue;
+		}
+		++reported;
+		comm_cmd = swab32(readl(&mfunc->comm[slave].slave_write));
+		slt = swab32(readl(&mfunc->comm[slave].slave_read)) >> 31;
+		toggle = comm_cmd >> 31;
+		if (toggle != slt) {
+			if (master->slave_state[slave].comm_toggle
+			    != slt) {
+				pr_info("slave %d out of sync. read toggle %d, state toggle %d. Resynching.\n",
+					slave, slt,
+					master->slave_state[slave].comm_toggle);
+				master->slave_state[slave].comm_toggle =
+					slt;
+			}
+			mlx4_master_do_cmd(dev, slave,
+					   comm_cmd >> 16 & 0xff,
+					   comm_cmd & 0xffff, toggle);
+			++served;
+		}
+		slave++;
+	}
+
+	if (reported && reported != served)
+		mlx4_warn(dev, "Got command event with bitmask from %d slaves but %d were served\n",
+			  reported, served);
+
+	if (mlx4_ARM_COMM_CHANNEL(dev))
+		mlx4_warn(dev, "Failed to arm comm channel events\n");
+}
+
+static int sync_toggles(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	u32 wr_toggle;
+	u32 rd_toggle;
+	unsigned long end;
+
+	wr_toggle = swab32(readl(&priv->mfunc.comm->slave_write));
+	if (wr_toggle == 0xffffffff)
+		end = jiffies + msecs_to_jiffies(30000);
+	else
+		end = jiffies + msecs_to_jiffies(5000);
+
+	while (time_before(jiffies, end)) {
+		rd_toggle = swab32(readl(&priv->mfunc.comm->slave_read));
+		if (wr_toggle == 0xffffffff || rd_toggle == 0xffffffff) {
+			/* PCI might be offline */
+
+			/* If device removal has been requested,
+			 * do not continue retrying.
+			 */
+			if (dev->persist->interface_state &
+			    MLX4_INTERFACE_STATE_NOWAIT) {
+				mlx4_warn(dev,
+					  "communication channel is offline\n");
+				return -EIO;
+			}
+
+			msleep(100);
+			wr_toggle = swab32(readl(&priv->mfunc.comm->
+					   slave_write));
+			continue;
+		}
+
+		if (rd_toggle >> 31 == wr_toggle >> 31) {
+			priv->cmd.comm_toggle = rd_toggle >> 31;
+			return 0;
+		}
+
+		cond_resched();
+	}
+
+	/*
+	 * we could reach here if for example the previous VM using this
+	 * function misbehaved and left the channel with unsynced state. We
+	 * should fix this here and give this VM a chance to use a properly
+	 * synced channel
+	 */
+	mlx4_warn(dev, "recovering from previously mis-behaved VM\n");
+	__raw_writel((__force u32) 0, &priv->mfunc.comm->slave_read);
+	__raw_writel((__force u32) 0, &priv->mfunc.comm->slave_write);
+	priv->cmd.comm_toggle = 0;
+
+	return 0;
+}
+
+int mlx4_multi_func_init(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_slave_state *s_state;
+	int i, j, err, port;
+
+	if (mlx4_is_master(dev))
+		priv->mfunc.comm =
+		ioremap(pci_resource_start(dev->persist->pdev,
+					   priv->fw.comm_bar) +
+			priv->fw.comm_base, MLX4_COMM_PAGESIZE);
+	else
+		priv->mfunc.comm =
+		ioremap(pci_resource_start(dev->persist->pdev, 2) +
+			MLX4_SLAVE_COMM_BASE, MLX4_COMM_PAGESIZE);
+	if (!priv->mfunc.comm) {
+		mlx4_err(dev, "Couldn't map communication vector\n");
+		goto err_vhcr;
+	}
+
+	if (mlx4_is_master(dev)) {
+		struct mlx4_vf_oper_state *vf_oper;
+		struct mlx4_vf_admin_state *vf_admin;
+
+		priv->mfunc.master.slave_state =
+			kcalloc(dev->num_slaves,
+				sizeof(struct mlx4_slave_state),
+				GFP_KERNEL);
+		if (!priv->mfunc.master.slave_state)
+			goto err_comm;
+
+		priv->mfunc.master.vf_admin =
+			kcalloc(dev->num_slaves,
+				sizeof(struct mlx4_vf_admin_state),
+				GFP_KERNEL);
+		if (!priv->mfunc.master.vf_admin)
+			goto err_comm_admin;
+
+		priv->mfunc.master.vf_oper =
+			kcalloc(dev->num_slaves,
+				sizeof(struct mlx4_vf_oper_state),
+				GFP_KERNEL);
+		if (!priv->mfunc.master.vf_oper)
+			goto err_comm_oper;
+
+		priv->mfunc.master.next_slave = 0;
+
+		for (i = 0; i < dev->num_slaves; ++i) {
+			vf_admin = &priv->mfunc.master.vf_admin[i];
+			vf_oper = &priv->mfunc.master.vf_oper[i];
+			s_state = &priv->mfunc.master.slave_state[i];
+			s_state->last_cmd = MLX4_COMM_CMD_RESET;
+			s_state->vst_qinq_supported = false;
+			mutex_init(&priv->mfunc.master.gen_eqe_mutex[i]);
+			for (j = 0; j < MLX4_EVENT_TYPES_NUM; ++j)
+				s_state->event_eq[j].eqn = -1;
+			__raw_writel((__force u32) 0,
+				     &priv->mfunc.comm[i].slave_write);
+			__raw_writel((__force u32) 0,
+				     &priv->mfunc.comm[i].slave_read);
+			for (port = 1; port <= MLX4_MAX_PORTS; port++) {
+				struct mlx4_vport_state *admin_vport;
+				struct mlx4_vport_state *oper_vport;
+
+				s_state->vlan_filter[port] =
+					kzalloc(sizeof(struct mlx4_vlan_fltr),
+						GFP_KERNEL);
+				if (!s_state->vlan_filter[port]) {
+					if (--port)
+						kfree(s_state->vlan_filter[port]);
+					goto err_slaves;
+				}
+
+				admin_vport = &vf_admin->vport[port];
+				oper_vport = &vf_oper->vport[port].state;
+				INIT_LIST_HEAD(&s_state->mcast_filters[port]);
+				admin_vport->default_vlan = MLX4_VGT;
+				oper_vport->default_vlan = MLX4_VGT;
+				admin_vport->qos_vport =
+						MLX4_VPP_DEFAULT_VPORT;
+				oper_vport->qos_vport = MLX4_VPP_DEFAULT_VPORT;
+				admin_vport->vlan_proto = htons(ETH_P_8021Q);
+				oper_vport->vlan_proto = htons(ETH_P_8021Q);
+				vf_oper->vport[port].vlan_idx = NO_INDX;
+				vf_oper->vport[port].mac_idx = NO_INDX;
+				mlx4_set_random_admin_guid(dev, i, port);
+			}
+			spin_lock_init(&s_state->lock);
+		}
+
+		if (dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_QOS_VPP) {
+			for (port = 1; port <= dev->caps.num_ports; port++) {
+				if (mlx4_is_eth(dev, port)) {
+					mlx4_set_default_port_qos(dev, port);
+					mlx4_allocate_port_vpps(dev, port);
+				}
+			}
+		}
+
+		memset(&priv->mfunc.master.cmd_eqe, 0, sizeof(struct mlx4_eqe));
+		priv->mfunc.master.cmd_eqe.type = MLX4_EVENT_TYPE_CMD;
+		INIT_WORK(&priv->mfunc.master.comm_work,
+			  mlx4_master_comm_channel);
+		INIT_WORK(&priv->mfunc.master.slave_event_work,
+			  mlx4_gen_slave_eqe);
+		INIT_WORK(&priv->mfunc.master.slave_flr_event_work,
+			  mlx4_master_handle_slave_flr);
+		spin_lock_init(&priv->mfunc.master.slave_state_lock);
+		spin_lock_init(&priv->mfunc.master.slave_eq.event_lock);
+		priv->mfunc.master.comm_wq =
+			create_singlethread_workqueue("mlx4_comm");
+		if (!priv->mfunc.master.comm_wq)
+			goto err_slaves;
+
+		if (mlx4_init_resource_tracker(dev))
+			goto err_thread;
+
+	} else {
+		err = sync_toggles(dev);
+		if (err) {
+			mlx4_err(dev, "Couldn't sync toggles\n");
+			goto err_comm;
+		}
+	}
+	return 0;
+
+err_thread:
+	destroy_workqueue(priv->mfunc.master.comm_wq);
+err_slaves:
+	while (i--) {
+		for (port = 1; port <= MLX4_MAX_PORTS; port++)
+			kfree(priv->mfunc.master.slave_state[i].vlan_filter[port]);
+	}
+	kfree(priv->mfunc.master.vf_oper);
+err_comm_oper:
+	kfree(priv->mfunc.master.vf_admin);
+err_comm_admin:
+	kfree(priv->mfunc.master.slave_state);
+err_comm:
+	iounmap(priv->mfunc.comm);
+	priv->mfunc.comm = NULL;
+err_vhcr:
+	dma_free_coherent(&dev->persist->pdev->dev, PAGE_SIZE,
+			  priv->mfunc.vhcr,
+			  priv->mfunc.vhcr_dma);
+	priv->mfunc.vhcr = NULL;
+	return -ENOMEM;
+}
+
+int mlx4_cmd_init(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int flags = 0;
+
+	if (!priv->cmd.initialized) {
+		init_rwsem(&priv->cmd.switch_sem);
+		mutex_init(&priv->cmd.slave_cmd_mutex);
+		sema_init(&priv->cmd.poll_sem, 1);
+		priv->cmd.use_events = 0;
+		priv->cmd.toggle     = 1;
+		priv->cmd.initialized = 1;
+		flags |= MLX4_CMD_CLEANUP_STRUCT;
+	}
+
+	if (!mlx4_is_slave(dev) && !priv->cmd.hcr) {
+		priv->cmd.hcr = ioremap(pci_resource_start(dev->persist->pdev,
+					0) + MLX4_HCR_BASE, MLX4_HCR_SIZE);
+		if (!priv->cmd.hcr) {
+			mlx4_err(dev, "Couldn't map command register\n");
+			goto err;
+		}
+		flags |= MLX4_CMD_CLEANUP_HCR;
+	}
+
+	if (mlx4_is_mfunc(dev) && !priv->mfunc.vhcr) {
+		priv->mfunc.vhcr = dma_alloc_coherent(&dev->persist->pdev->dev,
+						      PAGE_SIZE,
+						      &priv->mfunc.vhcr_dma,
+						      GFP_KERNEL);
+		if (!priv->mfunc.vhcr)
+			goto err;
+
+		flags |= MLX4_CMD_CLEANUP_VHCR;
+	}
+
+	if (!priv->cmd.pool) {
+		priv->cmd.pool = dma_pool_create("mlx4_cmd",
+						 &dev->persist->pdev->dev,
+						 MLX4_MAILBOX_SIZE,
+						 MLX4_MAILBOX_SIZE, 0);
+		if (!priv->cmd.pool)
+			goto err;
+
+		flags |= MLX4_CMD_CLEANUP_POOL;
+	}
+
+	return 0;
+
+err:
+	mlx4_cmd_cleanup(dev, flags);
+	return -ENOMEM;
+}
+
+void mlx4_report_internal_err_comm_event(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int slave;
+	u32 slave_read;
+
+	/* If the comm channel has not yet been initialized,
+	 * skip reporting the internal error event to all
+	 * the communication channels.
+	 */
+	if (!priv->mfunc.comm)
+		return;
+
+	/* Report an internal error event to all
+	 * communication channels.
+	 */
+	for (slave = 0; slave < dev->num_slaves; slave++) {
+		slave_read = swab32(readl(&priv->mfunc.comm[slave].slave_read));
+		slave_read |= (u32)COMM_CHAN_EVENT_INTERNAL_ERR;
+		__raw_writel((__force u32)cpu_to_be32(slave_read),
+			     &priv->mfunc.comm[slave].slave_read);
+	}
+}
+
+void mlx4_multi_func_cleanup(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int i, port;
+
+	if (mlx4_is_master(dev)) {
+		destroy_workqueue(priv->mfunc.master.comm_wq);
+		for (i = 0; i < dev->num_slaves; i++) {
+			for (port = 1; port <= MLX4_MAX_PORTS; port++)
+				kfree(priv->mfunc.master.slave_state[i].vlan_filter[port]);
+		}
+		kfree(priv->mfunc.master.slave_state);
+		kfree(priv->mfunc.master.vf_admin);
+		kfree(priv->mfunc.master.vf_oper);
+		dev->num_slaves = 0;
+	}
+
+	iounmap(priv->mfunc.comm);
+	priv->mfunc.comm = NULL;
+}
+
+void mlx4_cmd_cleanup(struct mlx4_dev *dev, int cleanup_mask)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	if (priv->cmd.pool && (cleanup_mask & MLX4_CMD_CLEANUP_POOL)) {
+		dma_pool_destroy(priv->cmd.pool);
+		priv->cmd.pool = NULL;
+	}
+
+	if (!mlx4_is_slave(dev) && priv->cmd.hcr &&
+	    (cleanup_mask & MLX4_CMD_CLEANUP_HCR)) {
+		iounmap(priv->cmd.hcr);
+		priv->cmd.hcr = NULL;
+	}
+	if (mlx4_is_mfunc(dev) && priv->mfunc.vhcr &&
+	    (cleanup_mask & MLX4_CMD_CLEANUP_VHCR)) {
+		dma_free_coherent(&dev->persist->pdev->dev, PAGE_SIZE,
+				  priv->mfunc.vhcr, priv->mfunc.vhcr_dma);
+		priv->mfunc.vhcr = NULL;
+	}
+	if (priv->cmd.initialized && (cleanup_mask & MLX4_CMD_CLEANUP_STRUCT))
+		priv->cmd.initialized = 0;
+}
+
+/*
+ * Switch to using events to issue FW commands (can only be called
+ * after event queue for command events has been initialized).
+ */
+int mlx4_cmd_use_events(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int i;
+	int err = 0;
+
+	priv->cmd.context = kmalloc_array(priv->cmd.max_cmds,
+					  sizeof(struct mlx4_cmd_context),
+					  GFP_KERNEL);
+	if (!priv->cmd.context)
+		return -ENOMEM;
+
+	if (mlx4_is_mfunc(dev))
+		mutex_lock(&priv->cmd.slave_cmd_mutex);
+	down_write(&priv->cmd.switch_sem);
+	for (i = 0; i < priv->cmd.max_cmds; ++i) {
+		priv->cmd.context[i].token = i;
+		priv->cmd.context[i].next  = i + 1;
+		/* To support fatal error flow, initialize all
+		 * cmd contexts to allow simulating completions
+		 * with complete() at any time.
+		 */
+		init_completion(&priv->cmd.context[i].done);
+	}
+
+	priv->cmd.context[priv->cmd.max_cmds - 1].next = -1;
+	priv->cmd.free_head = 0;
+
+	sema_init(&priv->cmd.event_sem, priv->cmd.max_cmds);
+
+	for (priv->cmd.token_mask = 1;
+	     priv->cmd.token_mask < priv->cmd.max_cmds;
+	     priv->cmd.token_mask <<= 1)
+		; /* nothing */
+	--priv->cmd.token_mask;
+
+	down(&priv->cmd.poll_sem);
+	priv->cmd.use_events = 1;
+	up_write(&priv->cmd.switch_sem);
+	if (mlx4_is_mfunc(dev))
+		mutex_unlock(&priv->cmd.slave_cmd_mutex);
+
+	return err;
+}
+
+/*
+ * Switch back to polling (used when shutting down the device)
+ */
+void mlx4_cmd_use_polling(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int i;
+
+	if (mlx4_is_mfunc(dev))
+		mutex_lock(&priv->cmd.slave_cmd_mutex);
+	down_write(&priv->cmd.switch_sem);
+	priv->cmd.use_events = 0;
+
+	for (i = 0; i < priv->cmd.max_cmds; ++i)
+		down(&priv->cmd.event_sem);
+
+	kfree(priv->cmd.context);
+	priv->cmd.context = NULL;
+
+	up(&priv->cmd.poll_sem);
+	up_write(&priv->cmd.switch_sem);
+	if (mlx4_is_mfunc(dev))
+		mutex_unlock(&priv->cmd.slave_cmd_mutex);
+}
+
+struct mlx4_cmd_mailbox *mlx4_alloc_cmd_mailbox(struct mlx4_dev *dev)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+
+	mailbox = kmalloc(sizeof(*mailbox), GFP_KERNEL);
+	if (!mailbox)
+		return ERR_PTR(-ENOMEM);
+
+	mailbox->buf = dma_pool_zalloc(mlx4_priv(dev)->cmd.pool, GFP_KERNEL,
+				       &mailbox->dma);
+	if (!mailbox->buf) {
+		kfree(mailbox);
+		return ERR_PTR(-ENOMEM);
+	}
+
+	return mailbox;
+}
+EXPORT_SYMBOL_GPL(mlx4_alloc_cmd_mailbox);
+
+void mlx4_free_cmd_mailbox(struct mlx4_dev *dev,
+			   struct mlx4_cmd_mailbox *mailbox)
+{
+	if (!mailbox)
+		return;
+
+	dma_pool_free(mlx4_priv(dev)->cmd.pool, mailbox->buf, mailbox->dma);
+	kfree(mailbox);
+}
+EXPORT_SYMBOL_GPL(mlx4_free_cmd_mailbox);
+
+u32 mlx4_comm_get_version(void)
+{
+	 return ((u32) CMD_CHAN_IF_REV << 8) | (u32) CMD_CHAN_VER;
+}
+
+static int mlx4_get_slave_indx(struct mlx4_dev *dev, int vf)
+{
+	if ((vf < 0) || (vf >= dev->persist->num_vfs)) {
+		mlx4_err(dev, "Bad vf number:%d (number of activated vf: %d)\n",
+			 vf, dev->persist->num_vfs);
+		return -EINVAL;
+	}
+
+	return vf+1;
+}
+
+int mlx4_get_vf_indx(struct mlx4_dev *dev, int slave)
+{
+	if (slave < 1 || slave > dev->persist->num_vfs) {
+		mlx4_err(dev,
+			 "Bad slave number:%d (number of activated slaves: %lu)\n",
+			 slave, dev->num_slaves);
+		return -EINVAL;
+	}
+	return slave - 1;
+}
+
+void mlx4_cmd_wake_completions(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_cmd_context *context;
+	int i;
+
+	spin_lock(&priv->cmd.context_lock);
+	if (priv->cmd.context) {
+		for (i = 0; i < priv->cmd.max_cmds; ++i) {
+			context = &priv->cmd.context[i];
+			context->fw_status = CMD_STAT_INTERNAL_ERR;
+			context->result    =
+				mlx4_status_to_errno(CMD_STAT_INTERNAL_ERR);
+			complete(&context->done);
+		}
+	}
+	spin_unlock(&priv->cmd.context_lock);
+}
+
+struct mlx4_active_ports mlx4_get_active_ports(struct mlx4_dev *dev, int slave)
+{
+	struct mlx4_active_ports actv_ports;
+	int vf;
+
+	bitmap_zero(actv_ports.ports, MLX4_MAX_PORTS);
+
+	if (slave == 0) {
+		bitmap_fill(actv_ports.ports, dev->caps.num_ports);
+		return actv_ports;
+	}
+
+	vf = mlx4_get_vf_indx(dev, slave);
+	if (vf < 0)
+		return actv_ports;
+
+	bitmap_set(actv_ports.ports, dev->dev_vfs[vf].min_port - 1,
+		   min((int)dev->dev_vfs[mlx4_get_vf_indx(dev, slave)].n_ports,
+		   dev->caps.num_ports));
+
+	return actv_ports;
+}
+EXPORT_SYMBOL_GPL(mlx4_get_active_ports);
+
+int mlx4_slave_convert_port(struct mlx4_dev *dev, int slave, int port)
+{
+	unsigned n;
+	struct mlx4_active_ports actv_ports = mlx4_get_active_ports(dev, slave);
+	unsigned m = bitmap_weight(actv_ports.ports, dev->caps.num_ports);
+
+	if (port <= 0 || port > m)
+		return -EINVAL;
+
+	n = find_first_bit(actv_ports.ports, dev->caps.num_ports);
+	if (port <= n)
+		port = n + 1;
+
+	return port;
+}
+EXPORT_SYMBOL_GPL(mlx4_slave_convert_port);
+
+int mlx4_phys_to_slave_port(struct mlx4_dev *dev, int slave, int port)
+{
+	struct mlx4_active_ports actv_ports = mlx4_get_active_ports(dev, slave);
+	if (test_bit(port - 1, actv_ports.ports))
+		return port -
+			find_first_bit(actv_ports.ports, dev->caps.num_ports);
+
+	return -1;
+}
+EXPORT_SYMBOL_GPL(mlx4_phys_to_slave_port);
+
+struct mlx4_slaves_pport mlx4_phys_to_slaves_pport(struct mlx4_dev *dev,
+						   int port)
+{
+	unsigned i;
+	struct mlx4_slaves_pport slaves_pport;
+
+	bitmap_zero(slaves_pport.slaves, MLX4_MFUNC_MAX);
+
+	if (port <= 0 || port > dev->caps.num_ports)
+		return slaves_pport;
+
+	for (i = 0; i < dev->persist->num_vfs + 1; i++) {
+		struct mlx4_active_ports actv_ports =
+			mlx4_get_active_ports(dev, i);
+		if (test_bit(port - 1, actv_ports.ports))
+			set_bit(i, slaves_pport.slaves);
+	}
+
+	return slaves_pport;
+}
+EXPORT_SYMBOL_GPL(mlx4_phys_to_slaves_pport);
+
+struct mlx4_slaves_pport mlx4_phys_to_slaves_pport_actv(
+		struct mlx4_dev *dev,
+		const struct mlx4_active_ports *crit_ports)
+{
+	unsigned i;
+	struct mlx4_slaves_pport slaves_pport;
+
+	bitmap_zero(slaves_pport.slaves, MLX4_MFUNC_MAX);
+
+	for (i = 0; i < dev->persist->num_vfs + 1; i++) {
+		struct mlx4_active_ports actv_ports =
+			mlx4_get_active_ports(dev, i);
+		if (bitmap_equal(crit_ports->ports, actv_ports.ports,
+				 dev->caps.num_ports))
+			set_bit(i, slaves_pport.slaves);
+	}
+
+	return slaves_pport;
+}
+EXPORT_SYMBOL_GPL(mlx4_phys_to_slaves_pport_actv);
+
+static int mlx4_slaves_closest_port(struct mlx4_dev *dev, int slave, int port)
+{
+	struct mlx4_active_ports actv_ports = mlx4_get_active_ports(dev, slave);
+	int min_port = find_first_bit(actv_ports.ports, dev->caps.num_ports)
+			+ 1;
+	int max_port = min_port +
+		bitmap_weight(actv_ports.ports, dev->caps.num_ports);
+
+	if (port < min_port)
+		port = min_port;
+	else if (port >= max_port)
+		port = max_port - 1;
+
+	return port;
+}
+
+static int mlx4_set_vport_qos(struct mlx4_priv *priv, int slave, int port,
+			      int max_tx_rate)
+{
+	int i;
+	int err;
+	struct mlx4_qos_manager *port_qos;
+	struct mlx4_dev *dev = &priv->dev;
+	struct mlx4_vport_qos_param vpp_qos[MLX4_NUM_UP];
+
+	port_qos = &priv->mfunc.master.qos_ctl[port];
+	memset(vpp_qos, 0, sizeof(struct mlx4_vport_qos_param) * MLX4_NUM_UP);
+
+	if (slave > port_qos->num_of_qos_vfs) {
+		mlx4_info(dev, "No available VPP resources for this VF\n");
+		return -EINVAL;
+	}
+
+	/* Query for default QoS values from Vport 0 is needed */
+	err = mlx4_SET_VPORT_QOS_get(dev, port, 0, vpp_qos);
+	if (err) {
+		mlx4_info(dev, "Failed to query Vport 0 QoS values\n");
+		return err;
+	}
+
+	for (i = 0; i < MLX4_NUM_UP; i++) {
+		if (test_bit(i, port_qos->priority_bm) && max_tx_rate) {
+			vpp_qos[i].max_avg_bw = max_tx_rate;
+			vpp_qos[i].enable = 1;
+		} else {
+			/* if user supplied tx_rate == 0, meaning no rate limit
+			 * configuration is required. so we are leaving the
+			 * value of max_avg_bw as queried from Vport 0.
+			 */
+			vpp_qos[i].enable = 0;
+		}
+	}
+
+	err = mlx4_SET_VPORT_QOS_set(dev, port, slave, vpp_qos);
+	if (err) {
+		mlx4_info(dev, "Failed to set Vport %d QoS values\n", slave);
+		return err;
+	}
+
+	return 0;
+}
+
+static bool mlx4_is_vf_vst_and_prio_qos(struct mlx4_dev *dev, int port,
+					struct mlx4_vport_state *vf_admin)
+{
+	struct mlx4_qos_manager *info;
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	if (!mlx4_is_master(dev) ||
+	    !(dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_QOS_VPP))
+		return false;
+
+	info = &priv->mfunc.master.qos_ctl[port];
+
+	if (vf_admin->default_vlan != MLX4_VGT &&
+	    test_bit(vf_admin->default_qos, info->priority_bm))
+		return true;
+
+	return false;
+}
+
+static bool mlx4_valid_vf_state_change(struct mlx4_dev *dev, int port,
+				       struct mlx4_vport_state *vf_admin,
+				       int vlan, int qos)
+{
+	struct mlx4_vport_state dummy_admin = {0};
+
+	if (!mlx4_is_vf_vst_and_prio_qos(dev, port, vf_admin) ||
+	    !vf_admin->tx_rate)
+		return true;
+
+	dummy_admin.default_qos = qos;
+	dummy_admin.default_vlan = vlan;
+
+	/* VF wants to move to other VST state which is valid with current
+	 * rate limit. Either differnt default vlan in VST or other
+	 * supported QoS priority. Otherwise we don't allow this change when
+	 * the TX rate is still configured.
+	 */
+	if (mlx4_is_vf_vst_and_prio_qos(dev, port, &dummy_admin))
+		return true;
+
+	mlx4_info(dev, "Cannot change VF state to %s while rate is set\n",
+		  (vlan == MLX4_VGT) ? "VGT" : "VST");
+
+	if (vlan != MLX4_VGT)
+		mlx4_info(dev, "VST priority %d not supported for QoS\n", qos);
+
+	mlx4_info(dev, "Please set rate to 0 prior to this VF state change\n");
+
+	return false;
+}
+
+int mlx4_set_vf_mac(struct mlx4_dev *dev, int port, int vf, u8 *mac)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_vport_state *s_info;
+	int slave;
+
+	if (!mlx4_is_master(dev))
+		return -EPROTONOSUPPORT;
+
+	if (is_multicast_ether_addr(mac))
+		return -EINVAL;
+
+	slave = mlx4_get_slave_indx(dev, vf);
+	if (slave < 0)
+		return -EINVAL;
+
+	port = mlx4_slaves_closest_port(dev, slave, port);
+	s_info = &priv->mfunc.master.vf_admin[slave].vport[port];
+
+	if (s_info->spoofchk && is_zero_ether_addr(mac)) {
+		mlx4_info(dev, "MAC invalidation is not allowed when spoofchk is on\n");
+		return -EPERM;
+	}
+
+	s_info->mac = ether_addr_to_u64(mac);
+	mlx4_info(dev, "default mac on vf %d port %d to %llX will take effect only after vf restart\n",
+		  vf, port, s_info->mac);
+	return 0;
+}
+EXPORT_SYMBOL_GPL(mlx4_set_vf_mac);
+
+
+int mlx4_set_vf_vlan(struct mlx4_dev *dev, int port, int vf, u16 vlan, u8 qos,
+		     __be16 proto)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_vport_state *vf_admin;
+	struct mlx4_slave_state *slave_state;
+	struct mlx4_vport_oper_state *vf_oper;
+	int slave;
+
+	if ((!mlx4_is_master(dev)) ||
+	    !(dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_VLAN_CONTROL))
+		return -EPROTONOSUPPORT;
+
+	if ((vlan > 4095) || (qos > 7))
+		return -EINVAL;
+
+	if (proto == htons(ETH_P_8021AD) &&
+	    !(dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_SVLAN_BY_QP))
+		return -EPROTONOSUPPORT;
+
+	if (proto != htons(ETH_P_8021Q) &&
+	    proto != htons(ETH_P_8021AD))
+		return -EINVAL;
+
+	if ((proto == htons(ETH_P_8021AD)) &&
+	    ((vlan == 0) || (vlan == MLX4_VGT)))
+		return -EINVAL;
+
+	slave = mlx4_get_slave_indx(dev, vf);
+	if (slave < 0)
+		return -EINVAL;
+
+	slave_state = &priv->mfunc.master.slave_state[slave];
+	if ((proto == htons(ETH_P_8021AD)) && (slave_state->active) &&
+	    (!slave_state->vst_qinq_supported)) {
+		mlx4_err(dev, "vf %d does not support VST QinQ mode\n", vf);
+		return -EPROTONOSUPPORT;
+	}
+	port = mlx4_slaves_closest_port(dev, slave, port);
+	vf_admin = &priv->mfunc.master.vf_admin[slave].vport[port];
+	vf_oper = &priv->mfunc.master.vf_oper[slave].vport[port];
+
+	if (!mlx4_valid_vf_state_change(dev, port, vf_admin, vlan, qos))
+		return -EPERM;
+
+	if ((0 == vlan) && (0 == qos))
+		vf_admin->default_vlan = MLX4_VGT;
+	else
+		vf_admin->default_vlan = vlan;
+	vf_admin->default_qos = qos;
+	vf_admin->vlan_proto = proto;
+
+	/* If rate was configured prior to VST, we saved the configured rate
+	 * in vf_admin->rate and now, if priority supported we enforce the QoS
+	 */
+	if (mlx4_is_vf_vst_and_prio_qos(dev, port, vf_admin) &&
+	    vf_admin->tx_rate)
+		vf_admin->qos_vport = slave;
+
+	/* Try to activate new vf state without restart,
+	 * this option is not supported while moving to VST QinQ mode.
+	 */
+	if ((proto == htons(ETH_P_8021AD) &&
+	     vf_oper->state.vlan_proto != proto) ||
+	    mlx4_master_immediate_activate_vlan_qos(priv, slave, port))
+		mlx4_info(dev,
+			  "updating vf %d port %d config will take effect on next VF restart\n",
+			  vf, port);
+	return 0;
+}
+EXPORT_SYMBOL_GPL(mlx4_set_vf_vlan);
+
+int mlx4_set_vf_rate(struct mlx4_dev *dev, int port, int vf, int min_tx_rate,
+		     int max_tx_rate)
+{
+	int err;
+	int slave;
+	struct mlx4_vport_state *vf_admin;
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	if (!mlx4_is_master(dev) ||
+	    !(dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_QOS_VPP))
+		return -EPROTONOSUPPORT;
+
+	if (min_tx_rate) {
+		mlx4_info(dev, "Minimum BW share not supported\n");
+		return -EPROTONOSUPPORT;
+	}
+
+	slave = mlx4_get_slave_indx(dev, vf);
+	if (slave < 0)
+		return -EINVAL;
+
+	port = mlx4_slaves_closest_port(dev, slave, port);
+	vf_admin = &priv->mfunc.master.vf_admin[slave].vport[port];
+
+	err = mlx4_set_vport_qos(priv, slave, port, max_tx_rate);
+	if (err) {
+		mlx4_info(dev, "vf %d failed to set rate %d\n", vf,
+			  max_tx_rate);
+		return err;
+	}
+
+	vf_admin->tx_rate = max_tx_rate;
+	/* if VF is not in supported mode (VST with supported prio),
+	 * we do not change vport configuration for its QPs, but save
+	 * the rate, so it will be enforced when it moves to supported
+	 * mode next time.
+	 */
+	if (!mlx4_is_vf_vst_and_prio_qos(dev, port, vf_admin)) {
+		mlx4_info(dev,
+			  "rate set for VF %d when not in valid state\n", vf);
+
+		if (vf_admin->default_vlan != MLX4_VGT)
+			mlx4_info(dev, "VST priority not supported by QoS\n");
+		else
+			mlx4_info(dev, "VF in VGT mode (needed VST)\n");
+
+		mlx4_info(dev,
+			  "rate %d take affect when VF moves to valid state\n",
+			  max_tx_rate);
+		return 0;
+	}
+
+	/* If user sets rate 0 assigning default vport for its QPs */
+	vf_admin->qos_vport = max_tx_rate ? slave : MLX4_VPP_DEFAULT_VPORT;
+
+	if (priv->mfunc.master.slave_state[slave].active &&
+	    dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_UPDATE_QP)
+		mlx4_master_immediate_activate_vlan_qos(priv, slave, port);
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(mlx4_set_vf_rate);
+
+ /* mlx4_get_slave_default_vlan -
+ * return true if VST ( default vlan)
+ * if VST, will return vlan & qos (if not NULL)
+ */
+bool mlx4_get_slave_default_vlan(struct mlx4_dev *dev, int port, int slave,
+				 u16 *vlan, u8 *qos)
+{
+	struct mlx4_vport_oper_state *vp_oper;
+	struct mlx4_priv *priv;
+
+	priv = mlx4_priv(dev);
+	port = mlx4_slaves_closest_port(dev, slave, port);
+	vp_oper = &priv->mfunc.master.vf_oper[slave].vport[port];
+
+	if (MLX4_VGT != vp_oper->state.default_vlan) {
+		if (vlan)
+			*vlan = vp_oper->state.default_vlan;
+		if (qos)
+			*qos = vp_oper->state.default_qos;
+		return true;
+	}
+	return false;
+}
+EXPORT_SYMBOL_GPL(mlx4_get_slave_default_vlan);
+
+int mlx4_set_vf_spoofchk(struct mlx4_dev *dev, int port, int vf, bool setting)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_vport_state *s_info;
+	int slave;
+	u8 mac[ETH_ALEN];
+
+	if ((!mlx4_is_master(dev)) ||
+	    !(dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_FSM))
+		return -EPROTONOSUPPORT;
+
+	slave = mlx4_get_slave_indx(dev, vf);
+	if (slave < 0)
+		return -EINVAL;
+
+	port = mlx4_slaves_closest_port(dev, slave, port);
+	s_info = &priv->mfunc.master.vf_admin[slave].vport[port];
+
+	u64_to_ether_addr(s_info->mac, mac);
+	if (setting && !is_valid_ether_addr(mac)) {
+		mlx4_info(dev, "Illegal MAC with spoofchk\n");
+		return -EPERM;
+	}
+
+	s_info->spoofchk = setting;
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(mlx4_set_vf_spoofchk);
+
+int mlx4_get_vf_config(struct mlx4_dev *dev, int port, int vf, struct ifla_vf_info *ivf)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_vport_state *s_info;
+	int slave;
+
+	if (!mlx4_is_master(dev))
+		return -EPROTONOSUPPORT;
+
+	slave = mlx4_get_slave_indx(dev, vf);
+	if (slave < 0)
+		return -EINVAL;
+
+	s_info = &priv->mfunc.master.vf_admin[slave].vport[port];
+	ivf->vf = vf;
+
+	/* need to convert it to a func */
+	ivf->mac[0] = ((s_info->mac >> (5*8)) & 0xff);
+	ivf->mac[1] = ((s_info->mac >> (4*8)) & 0xff);
+	ivf->mac[2] = ((s_info->mac >> (3*8)) & 0xff);
+	ivf->mac[3] = ((s_info->mac >> (2*8)) & 0xff);
+	ivf->mac[4] = ((s_info->mac >> (1*8)) & 0xff);
+	ivf->mac[5] = ((s_info->mac)  & 0xff);
+
+	ivf->vlan		= s_info->default_vlan;
+	ivf->qos		= s_info->default_qos;
+	ivf->vlan_proto		= s_info->vlan_proto;
+
+	if (mlx4_is_vf_vst_and_prio_qos(dev, port, s_info))
+		ivf->max_tx_rate = s_info->tx_rate;
+	else
+		ivf->max_tx_rate = 0;
+
+	ivf->min_tx_rate	= 0;
+	ivf->spoofchk		= s_info->spoofchk;
+	ivf->linkstate		= s_info->link_state;
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(mlx4_get_vf_config);
+
+int mlx4_set_vf_link_state(struct mlx4_dev *dev, int port, int vf, int link_state)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_vport_state *s_info;
+	int slave;
+	u8 link_stat_event;
+
+	slave = mlx4_get_slave_indx(dev, vf);
+	if (slave < 0)
+		return -EINVAL;
+
+	port = mlx4_slaves_closest_port(dev, slave, port);
+	switch (link_state) {
+	case IFLA_VF_LINK_STATE_AUTO:
+		/* get current link state */
+		if (!priv->sense.do_sense_port[port])
+			link_stat_event = MLX4_PORT_CHANGE_SUBTYPE_ACTIVE;
+		else
+			link_stat_event = MLX4_PORT_CHANGE_SUBTYPE_DOWN;
+	    break;
+
+	case IFLA_VF_LINK_STATE_ENABLE:
+		link_stat_event = MLX4_PORT_CHANGE_SUBTYPE_ACTIVE;
+	    break;
+
+	case IFLA_VF_LINK_STATE_DISABLE:
+		link_stat_event = MLX4_PORT_CHANGE_SUBTYPE_DOWN;
+	    break;
+
+	default:
+		mlx4_warn(dev, "unknown value for link_state %02x on slave %d port %d\n",
+			  link_state, slave, port);
+		return -EINVAL;
+	}
+	s_info = &priv->mfunc.master.vf_admin[slave].vport[port];
+	s_info->link_state = link_state;
+
+	/* send event */
+	mlx4_gen_port_state_change_eqe(dev, slave, port, link_stat_event);
+
+	if (mlx4_master_immediate_activate_vlan_qos(priv, slave, port))
+		mlx4_dbg(dev,
+			 "updating vf %d port %d no link state HW enforcement\n",
+			 vf, port);
+	return 0;
+}
+EXPORT_SYMBOL_GPL(mlx4_set_vf_link_state);
+
+int mlx4_get_counter_stats(struct mlx4_dev *dev, int counter_index,
+			   struct mlx4_counter *counter_stats, int reset)
+{
+	struct mlx4_cmd_mailbox *mailbox = NULL;
+	struct mlx4_counter *tmp_counter;
+	int err;
+	u32 if_stat_in_mod;
+
+	if (!counter_stats)
+		return -EINVAL;
+
+	if (counter_index == MLX4_SINK_COUNTER_INDEX(dev))
+		return 0;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+
+	memset(mailbox->buf, 0, sizeof(struct mlx4_counter));
+	if_stat_in_mod = counter_index;
+	if (reset)
+		if_stat_in_mod |= MLX4_QUERY_IF_STAT_RESET;
+	err = mlx4_cmd_box(dev, 0, mailbox->dma,
+			   if_stat_in_mod, 0,
+			   MLX4_CMD_QUERY_IF_STAT,
+			   MLX4_CMD_TIME_CLASS_C,
+			   MLX4_CMD_NATIVE);
+	if (err) {
+		mlx4_dbg(dev, "%s: failed to read statistics for counter index %d\n",
+			 __func__, counter_index);
+		goto if_stat_out;
+	}
+	tmp_counter = (struct mlx4_counter *)mailbox->buf;
+	counter_stats->counter_mode = tmp_counter->counter_mode;
+	if (counter_stats->counter_mode == 0) {
+		counter_stats->rx_frames =
+			cpu_to_be64(be64_to_cpu(counter_stats->rx_frames) +
+				    be64_to_cpu(tmp_counter->rx_frames));
+		counter_stats->tx_frames =
+			cpu_to_be64(be64_to_cpu(counter_stats->tx_frames) +
+				    be64_to_cpu(tmp_counter->tx_frames));
+		counter_stats->rx_bytes =
+			cpu_to_be64(be64_to_cpu(counter_stats->rx_bytes) +
+				    be64_to_cpu(tmp_counter->rx_bytes));
+		counter_stats->tx_bytes =
+			cpu_to_be64(be64_to_cpu(counter_stats->tx_bytes) +
+				    be64_to_cpu(tmp_counter->tx_bytes));
+	}
+
+if_stat_out:
+	mlx4_free_cmd_mailbox(dev, mailbox);
+
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx4_get_counter_stats);
+
+int mlx4_get_vf_stats(struct mlx4_dev *dev, int port, int vf_idx,
+		      struct ifla_vf_stats *vf_stats)
+{
+	struct mlx4_counter tmp_vf_stats;
+	int slave;
+	int err = 0;
+
+	if (!vf_stats)
+		return -EINVAL;
+
+	if (!mlx4_is_master(dev))
+		return -EPROTONOSUPPORT;
+
+	slave = mlx4_get_slave_indx(dev, vf_idx);
+	if (slave < 0)
+		return -EINVAL;
+
+	port = mlx4_slaves_closest_port(dev, slave, port);
+	err = mlx4_calc_vf_counters(dev, slave, port, &tmp_vf_stats);
+	if (!err && tmp_vf_stats.counter_mode == 0) {
+		vf_stats->rx_packets = be64_to_cpu(tmp_vf_stats.rx_frames);
+		vf_stats->tx_packets = be64_to_cpu(tmp_vf_stats.tx_frames);
+		vf_stats->rx_bytes = be64_to_cpu(tmp_vf_stats.rx_bytes);
+		vf_stats->tx_bytes = be64_to_cpu(tmp_vf_stats.tx_bytes);
+	}
+
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx4_get_vf_stats);
+
+int mlx4_vf_smi_enabled(struct mlx4_dev *dev, int slave, int port)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	if (slave < 1 || slave >= dev->num_slaves ||
+	    port < 1 || port > MLX4_MAX_PORTS)
+		return 0;
+
+	return priv->mfunc.master.vf_oper[slave].smi_enabled[port] ==
+		MLX4_VF_SMI_ENABLED;
+}
+EXPORT_SYMBOL_GPL(mlx4_vf_smi_enabled);
+
+int mlx4_vf_get_enable_smi_admin(struct mlx4_dev *dev, int slave, int port)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	if (slave == mlx4_master_func_num(dev))
+		return 1;
+
+	if (slave < 1 || slave >= dev->num_slaves ||
+	    port < 1 || port > MLX4_MAX_PORTS)
+		return 0;
+
+	return priv->mfunc.master.vf_admin[slave].enable_smi[port] ==
+		MLX4_VF_SMI_ENABLED;
+}
+EXPORT_SYMBOL_GPL(mlx4_vf_get_enable_smi_admin);
+
+int mlx4_vf_set_enable_smi_admin(struct mlx4_dev *dev, int slave, int port,
+				 int enabled)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_active_ports actv_ports = mlx4_get_active_ports(
+			&priv->dev, slave);
+	int min_port = find_first_bit(actv_ports.ports,
+				      priv->dev.caps.num_ports) + 1;
+	int max_port = min_port - 1 +
+		bitmap_weight(actv_ports.ports, priv->dev.caps.num_ports);
+
+	if (slave == mlx4_master_func_num(dev))
+		return 0;
+
+	if (slave < 1 || slave >= dev->num_slaves ||
+	    port < 1 || port > MLX4_MAX_PORTS ||
+	    enabled < 0 || enabled > 1)
+		return -EINVAL;
+
+	if (min_port == max_port && dev->caps.num_ports > 1) {
+		mlx4_info(dev, "SMI access disallowed for single ported VFs\n");
+		return -EPROTONOSUPPORT;
+	}
+
+	priv->mfunc.master.vf_admin[slave].enable_smi[port] = enabled;
+	return 0;
+}
+EXPORT_SYMBOL_GPL(mlx4_vf_set_enable_smi_admin);
diff --git a/drivers/net/ethernet/mellanox/mlx4/cq.c b/drivers/net/ethernet/mellanox/mlx4/cq.c
new file mode 100644
index 000000000..4d4f9cf9f
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx4/cq.c
@@ -0,0 +1,482 @@
+/*
+ * Copyright (c) 2004, 2005 Topspin Communications.  All rights reserved.
+ * Copyright (c) 2005 Sun Microsystems, Inc. All rights reserved.
+ * Copyright (c) 2005, 2006, 2007 Cisco Systems, Inc. All rights reserved.
+ * Copyright (c) 2005, 2006, 2007, 2008 Mellanox Technologies. All rights reserved.
+ * Copyright (c) 2004 Voltaire, Inc. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/hardirq.h>
+#include <linux/export.h>
+
+#include <linux/mlx4/cmd.h>
+#include <linux/mlx4/cq.h>
+
+#include "mlx4.h"
+#include "icm.h"
+
+#define MLX4_CQ_STATUS_OK		( 0 << 28)
+#define MLX4_CQ_STATUS_OVERFLOW		( 9 << 28)
+#define MLX4_CQ_STATUS_WRITE_FAIL	(10 << 28)
+#define MLX4_CQ_FLAG_CC			( 1 << 18)
+#define MLX4_CQ_FLAG_OI			( 1 << 17)
+#define MLX4_CQ_STATE_ARMED		( 9 <<  8)
+#define MLX4_CQ_STATE_ARMED_SOL		( 6 <<  8)
+#define MLX4_EQ_STATE_FIRED		(10 <<  8)
+
+#define TASKLET_MAX_TIME 2
+#define TASKLET_MAX_TIME_JIFFIES msecs_to_jiffies(TASKLET_MAX_TIME)
+
+void mlx4_cq_tasklet_cb(struct tasklet_struct *t)
+{
+	unsigned long flags;
+	unsigned long end = jiffies + TASKLET_MAX_TIME_JIFFIES;
+	struct mlx4_eq_tasklet *ctx = from_tasklet(ctx, t, task);
+	struct mlx4_cq *mcq, *temp;
+
+	spin_lock_irqsave(&ctx->lock, flags);
+	list_splice_tail_init(&ctx->list, &ctx->process_list);
+	spin_unlock_irqrestore(&ctx->lock, flags);
+
+	list_for_each_entry_safe(mcq, temp, &ctx->process_list, tasklet_ctx.list) {
+		list_del_init(&mcq->tasklet_ctx.list);
+		mcq->tasklet_ctx.comp(mcq);
+		if (refcount_dec_and_test(&mcq->refcount))
+			complete(&mcq->free);
+		if (time_after(jiffies, end))
+			break;
+	}
+
+	if (!list_empty(&ctx->process_list))
+		tasklet_schedule(&ctx->task);
+}
+
+static void mlx4_add_cq_to_tasklet(struct mlx4_cq *cq)
+{
+	struct mlx4_eq_tasklet *tasklet_ctx = cq->tasklet_ctx.priv;
+	unsigned long flags;
+	bool kick;
+
+	spin_lock_irqsave(&tasklet_ctx->lock, flags);
+	/* When migrating CQs between EQs will be implemented, please note
+	 * that you need to sync this point. It is possible that
+	 * while migrating a CQ, completions on the old EQs could
+	 * still arrive.
+	 */
+	if (list_empty_careful(&cq->tasklet_ctx.list)) {
+		refcount_inc(&cq->refcount);
+		kick = list_empty(&tasklet_ctx->list);
+		list_add_tail(&cq->tasklet_ctx.list, &tasklet_ctx->list);
+		if (kick)
+			tasklet_schedule(&tasklet_ctx->task);
+	}
+	spin_unlock_irqrestore(&tasklet_ctx->lock, flags);
+}
+
+void mlx4_cq_completion(struct mlx4_dev *dev, u32 cqn)
+{
+	struct mlx4_cq *cq;
+
+	rcu_read_lock();
+	cq = radix_tree_lookup(&mlx4_priv(dev)->cq_table.tree,
+			       cqn & (dev->caps.num_cqs - 1));
+	rcu_read_unlock();
+
+	if (!cq) {
+		mlx4_dbg(dev, "Completion event for bogus CQ %08x\n", cqn);
+		return;
+	}
+
+	/* Acessing the CQ outside of rcu_read_lock is safe, because
+	 * the CQ is freed only after interrupt handling is completed.
+	 */
+	++cq->arm_sn;
+
+	cq->comp(cq);
+}
+
+void mlx4_cq_event(struct mlx4_dev *dev, u32 cqn, int event_type)
+{
+	struct mlx4_cq_table *cq_table = &mlx4_priv(dev)->cq_table;
+	struct mlx4_cq *cq;
+
+	rcu_read_lock();
+	cq = radix_tree_lookup(&cq_table->tree, cqn & (dev->caps.num_cqs - 1));
+	rcu_read_unlock();
+
+	if (!cq) {
+		mlx4_dbg(dev, "Async event for bogus CQ %08x\n", cqn);
+		return;
+	}
+
+	/* Acessing the CQ outside of rcu_read_lock is safe, because
+	 * the CQ is freed only after interrupt handling is completed.
+	 */
+	cq->event(cq, event_type);
+}
+
+static int mlx4_SW2HW_CQ(struct mlx4_dev *dev, struct mlx4_cmd_mailbox *mailbox,
+			 int cq_num, u8 opmod)
+{
+	return mlx4_cmd(dev, mailbox->dma, cq_num, opmod,
+			MLX4_CMD_SW2HW_CQ, MLX4_CMD_TIME_CLASS_A,
+			MLX4_CMD_WRAPPED);
+}
+
+static int mlx4_MODIFY_CQ(struct mlx4_dev *dev, struct mlx4_cmd_mailbox *mailbox,
+			 int cq_num, u32 opmod)
+{
+	return mlx4_cmd(dev, mailbox->dma, cq_num, opmod, MLX4_CMD_MODIFY_CQ,
+			MLX4_CMD_TIME_CLASS_A, MLX4_CMD_WRAPPED);
+}
+
+static int mlx4_HW2SW_CQ(struct mlx4_dev *dev, struct mlx4_cmd_mailbox *mailbox,
+			 int cq_num)
+{
+	return mlx4_cmd_box(dev, 0, mailbox ? mailbox->dma : 0,
+			    cq_num, mailbox ? 0 : 1, MLX4_CMD_HW2SW_CQ,
+			    MLX4_CMD_TIME_CLASS_A, MLX4_CMD_WRAPPED);
+}
+
+int mlx4_cq_modify(struct mlx4_dev *dev, struct mlx4_cq *cq,
+		   u16 count, u16 period)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	struct mlx4_cq_context *cq_context;
+	int err;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+
+	cq_context = mailbox->buf;
+	cq_context->cq_max_count = cpu_to_be16(count);
+	cq_context->cq_period    = cpu_to_be16(period);
+
+	err = mlx4_MODIFY_CQ(dev, mailbox, cq->cqn, 1);
+
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx4_cq_modify);
+
+int mlx4_cq_resize(struct mlx4_dev *dev, struct mlx4_cq *cq,
+		   int entries, struct mlx4_mtt *mtt)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	struct mlx4_cq_context *cq_context;
+	u64 mtt_addr;
+	int err;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+
+	cq_context = mailbox->buf;
+	cq_context->logsize_usrpage = cpu_to_be32(ilog2(entries) << 24);
+	cq_context->log_page_size   = mtt->page_shift - 12;
+	mtt_addr = mlx4_mtt_addr(dev, mtt);
+	cq_context->mtt_base_addr_h = mtt_addr >> 32;
+	cq_context->mtt_base_addr_l = cpu_to_be32(mtt_addr & 0xffffffff);
+
+	err = mlx4_MODIFY_CQ(dev, mailbox, cq->cqn, 0);
+
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx4_cq_resize);
+
+int __mlx4_cq_alloc_icm(struct mlx4_dev *dev, int *cqn)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_cq_table *cq_table = &priv->cq_table;
+	int err;
+
+	*cqn = mlx4_bitmap_alloc(&cq_table->bitmap);
+	if (*cqn == -1)
+		return -ENOMEM;
+
+	err = mlx4_table_get(dev, &cq_table->table, *cqn);
+	if (err)
+		goto err_out;
+
+	err = mlx4_table_get(dev, &cq_table->cmpt_table, *cqn);
+	if (err)
+		goto err_put;
+	return 0;
+
+err_put:
+	mlx4_table_put(dev, &cq_table->table, *cqn);
+
+err_out:
+	mlx4_bitmap_free(&cq_table->bitmap, *cqn, MLX4_NO_RR);
+	return err;
+}
+
+static int mlx4_cq_alloc_icm(struct mlx4_dev *dev, int *cqn, u8 usage)
+{
+	u32 in_modifier = RES_CQ | (((u32)usage & 3) << 30);
+	u64 out_param;
+	int err;
+
+	if (mlx4_is_mfunc(dev)) {
+		err = mlx4_cmd_imm(dev, 0, &out_param, in_modifier,
+				   RES_OP_RESERVE_AND_MAP, MLX4_CMD_ALLOC_RES,
+				   MLX4_CMD_TIME_CLASS_A, MLX4_CMD_WRAPPED);
+		if (err)
+			return err;
+		else {
+			*cqn = get_param_l(&out_param);
+			return 0;
+		}
+	}
+	return __mlx4_cq_alloc_icm(dev, cqn);
+}
+
+void __mlx4_cq_free_icm(struct mlx4_dev *dev, int cqn)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_cq_table *cq_table = &priv->cq_table;
+
+	mlx4_table_put(dev, &cq_table->cmpt_table, cqn);
+	mlx4_table_put(dev, &cq_table->table, cqn);
+	mlx4_bitmap_free(&cq_table->bitmap, cqn, MLX4_NO_RR);
+}
+
+static void mlx4_cq_free_icm(struct mlx4_dev *dev, int cqn)
+{
+	u64 in_param = 0;
+	int err;
+
+	if (mlx4_is_mfunc(dev)) {
+		set_param_l(&in_param, cqn);
+		err = mlx4_cmd(dev, in_param, RES_CQ, RES_OP_RESERVE_AND_MAP,
+			       MLX4_CMD_FREE_RES,
+			       MLX4_CMD_TIME_CLASS_A, MLX4_CMD_WRAPPED);
+		if (err)
+			mlx4_warn(dev, "Failed freeing cq:%d\n", cqn);
+	} else
+		__mlx4_cq_free_icm(dev, cqn);
+}
+
+static int mlx4_init_user_cqes(void *buf, int entries, int cqe_size)
+{
+	int entries_per_copy = PAGE_SIZE / cqe_size;
+	void *init_ents;
+	int err = 0;
+	int i;
+
+	init_ents = kmalloc(PAGE_SIZE, GFP_KERNEL);
+	if (!init_ents)
+		return -ENOMEM;
+
+	/* Populate a list of CQ entries to reduce the number of
+	 * copy_to_user calls. 0xcc is the initialization value
+	 * required by the FW.
+	 */
+	memset(init_ents, 0xcc, PAGE_SIZE);
+
+	if (entries_per_copy < entries) {
+		for (i = 0; i < entries / entries_per_copy; i++) {
+			err = copy_to_user((void __user *)buf, init_ents, PAGE_SIZE) ?
+				-EFAULT : 0;
+			if (err)
+				goto out;
+
+			buf += PAGE_SIZE;
+		}
+	} else {
+		err = copy_to_user((void __user *)buf, init_ents,
+				   array_size(entries, cqe_size)) ?
+			-EFAULT : 0;
+	}
+
+out:
+	kfree(init_ents);
+
+	return err;
+}
+
+static void mlx4_init_kernel_cqes(struct mlx4_buf *buf,
+				  int entries,
+				  int cqe_size)
+{
+	int i;
+
+	if (buf->nbufs == 1)
+		memset(buf->direct.buf, 0xcc, entries * cqe_size);
+	else
+		for (i = 0; i < buf->npages; i++)
+			memset(buf->page_list[i].buf, 0xcc,
+			       1UL << buf->page_shift);
+}
+
+int mlx4_cq_alloc(struct mlx4_dev *dev, int nent,
+		  struct mlx4_mtt *mtt, struct mlx4_uar *uar, u64 db_rec,
+		  struct mlx4_cq *cq, unsigned vector, int collapsed,
+		  int timestamp_en, void *buf_addr, bool user_cq)
+{
+	bool sw_cq_init = dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_SW_CQ_INIT;
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_cq_table *cq_table = &priv->cq_table;
+	struct mlx4_cmd_mailbox *mailbox;
+	struct mlx4_cq_context *cq_context;
+	u64 mtt_addr;
+	int err;
+
+	if (vector >= dev->caps.num_comp_vectors)
+		return -EINVAL;
+
+	cq->vector = vector;
+
+	err = mlx4_cq_alloc_icm(dev, &cq->cqn, cq->usage);
+	if (err)
+		return err;
+
+	spin_lock(&cq_table->lock);
+	err = radix_tree_insert(&cq_table->tree, cq->cqn, cq);
+	spin_unlock(&cq_table->lock);
+	if (err)
+		goto err_icm;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox)) {
+		err = PTR_ERR(mailbox);
+		goto err_radix;
+	}
+
+	cq_context = mailbox->buf;
+	cq_context->flags	    = cpu_to_be32(!!collapsed << 18);
+	if (timestamp_en)
+		cq_context->flags  |= cpu_to_be32(1 << 19);
+
+	cq_context->logsize_usrpage =
+		cpu_to_be32((ilog2(nent) << 24) |
+			    mlx4_to_hw_uar_index(dev, uar->index));
+	cq_context->comp_eqn	    = priv->eq_table.eq[MLX4_CQ_TO_EQ_VECTOR(vector)].eqn;
+	cq_context->log_page_size   = mtt->page_shift - MLX4_ICM_PAGE_SHIFT;
+
+	mtt_addr = mlx4_mtt_addr(dev, mtt);
+	cq_context->mtt_base_addr_h = mtt_addr >> 32;
+	cq_context->mtt_base_addr_l = cpu_to_be32(mtt_addr & 0xffffffff);
+	cq_context->db_rec_addr     = cpu_to_be64(db_rec);
+
+	if (sw_cq_init) {
+		if (user_cq) {
+			err = mlx4_init_user_cqes(buf_addr, nent,
+						  dev->caps.cqe_size);
+			if (err)
+				sw_cq_init = false;
+		} else {
+			mlx4_init_kernel_cqes(buf_addr, nent,
+					      dev->caps.cqe_size);
+		}
+	}
+
+	err = mlx4_SW2HW_CQ(dev, mailbox, cq->cqn, sw_cq_init);
+
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	if (err)
+		goto err_radix;
+
+	cq->cons_index = 0;
+	cq->arm_sn     = 1;
+	cq->uar        = uar;
+	refcount_set(&cq->refcount, 1);
+	init_completion(&cq->free);
+	cq->comp = mlx4_add_cq_to_tasklet;
+	cq->tasklet_ctx.priv =
+		&priv->eq_table.eq[MLX4_CQ_TO_EQ_VECTOR(vector)].tasklet_ctx;
+	INIT_LIST_HEAD(&cq->tasklet_ctx.list);
+
+
+	cq->irq = priv->eq_table.eq[MLX4_CQ_TO_EQ_VECTOR(vector)].irq;
+	return 0;
+
+err_radix:
+	spin_lock(&cq_table->lock);
+	radix_tree_delete(&cq_table->tree, cq->cqn);
+	spin_unlock(&cq_table->lock);
+
+err_icm:
+	mlx4_cq_free_icm(dev, cq->cqn);
+
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx4_cq_alloc);
+
+void mlx4_cq_free(struct mlx4_dev *dev, struct mlx4_cq *cq)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_cq_table *cq_table = &priv->cq_table;
+	int err;
+
+	err = mlx4_HW2SW_CQ(dev, NULL, cq->cqn);
+	if (err)
+		mlx4_warn(dev, "HW2SW_CQ failed (%d) for CQN %06x\n", err, cq->cqn);
+
+	spin_lock(&cq_table->lock);
+	radix_tree_delete(&cq_table->tree, cq->cqn);
+	spin_unlock(&cq_table->lock);
+
+	synchronize_irq(priv->eq_table.eq[MLX4_CQ_TO_EQ_VECTOR(cq->vector)].irq);
+	if (priv->eq_table.eq[MLX4_CQ_TO_EQ_VECTOR(cq->vector)].irq !=
+	    priv->eq_table.eq[MLX4_EQ_ASYNC].irq)
+		synchronize_irq(priv->eq_table.eq[MLX4_EQ_ASYNC].irq);
+
+	if (refcount_dec_and_test(&cq->refcount))
+		complete(&cq->free);
+	wait_for_completion(&cq->free);
+
+	mlx4_cq_free_icm(dev, cq->cqn);
+}
+EXPORT_SYMBOL_GPL(mlx4_cq_free);
+
+int mlx4_init_cq_table(struct mlx4_dev *dev)
+{
+	struct mlx4_cq_table *cq_table = &mlx4_priv(dev)->cq_table;
+
+	spin_lock_init(&cq_table->lock);
+	INIT_RADIX_TREE(&cq_table->tree, GFP_ATOMIC);
+	if (mlx4_is_slave(dev))
+		return 0;
+
+	return mlx4_bitmap_init(&cq_table->bitmap, dev->caps.num_cqs,
+				dev->caps.num_cqs - 1, dev->caps.reserved_cqs, 0);
+}
+
+void mlx4_cleanup_cq_table(struct mlx4_dev *dev)
+{
+	if (mlx4_is_slave(dev))
+		return;
+	/* Nothing to do to clean up radix_tree */
+	mlx4_bitmap_cleanup(&mlx4_priv(dev)->cq_table.bitmap);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx4/crdump.c b/drivers/net/ethernet/mellanox/mlx4/crdump.c
new file mode 100644
index 000000000..82a07a31c
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx4/crdump.c
@@ -0,0 +1,258 @@
+/*
+ * Copyright (c) 2018, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "mlx4.h"
+
+#define BAD_ACCESS			0xBADACCE5
+#define HEALTH_BUFFER_SIZE		0x40
+#define CR_ENABLE_BIT			swab32(BIT(6))
+#define CR_ENABLE_BIT_OFFSET		0xF3F04
+#define MAX_NUM_OF_DUMPS_TO_STORE	(8)
+
+#define REGION_CR_SPACE "cr-space"
+#define REGION_FW_HEALTH "fw-health"
+
+static const char * const region_cr_space_str = REGION_CR_SPACE;
+static const char * const region_fw_health_str = REGION_FW_HEALTH;
+
+static const struct devlink_region_ops region_cr_space_ops = {
+	.name = REGION_CR_SPACE,
+	.destructor = &kvfree,
+};
+
+static const struct devlink_region_ops region_fw_health_ops = {
+	.name = REGION_FW_HEALTH,
+	.destructor = &kvfree,
+};
+
+/* Set to true in case cr enable bit was set to true before crdump */
+static bool crdump_enbale_bit_set;
+
+static void crdump_enable_crspace_access(struct mlx4_dev *dev,
+					 u8 __iomem *cr_space)
+{
+	/* Get current enable bit value */
+	crdump_enbale_bit_set =
+		readl(cr_space + CR_ENABLE_BIT_OFFSET) & CR_ENABLE_BIT;
+
+	/* Enable FW CR filter (set bit6 to 0) */
+	if (crdump_enbale_bit_set)
+		writel(readl(cr_space + CR_ENABLE_BIT_OFFSET) & ~CR_ENABLE_BIT,
+		       cr_space + CR_ENABLE_BIT_OFFSET);
+
+	/* Enable block volatile crspace accesses */
+	writel(swab32(1), cr_space + dev->caps.health_buffer_addrs +
+	       HEALTH_BUFFER_SIZE);
+}
+
+static void crdump_disable_crspace_access(struct mlx4_dev *dev,
+					  u8 __iomem *cr_space)
+{
+	/* Disable block volatile crspace accesses */
+	writel(0, cr_space + dev->caps.health_buffer_addrs +
+	       HEALTH_BUFFER_SIZE);
+
+	/* Restore FW CR filter value (set bit6 to original value) */
+	if (crdump_enbale_bit_set)
+		writel(readl(cr_space + CR_ENABLE_BIT_OFFSET) | CR_ENABLE_BIT,
+		       cr_space + CR_ENABLE_BIT_OFFSET);
+}
+
+static void mlx4_crdump_collect_crspace(struct mlx4_dev *dev,
+					u8 __iomem *cr_space,
+					u32 id)
+{
+	struct mlx4_fw_crdump *crdump = &dev->persist->crdump;
+	struct pci_dev *pdev = dev->persist->pdev;
+	unsigned long cr_res_size;
+	u8 *crspace_data;
+	int offset;
+	int err;
+
+	if (!crdump->region_crspace) {
+		mlx4_err(dev, "crdump: cr-space region is NULL\n");
+		return;
+	}
+
+	/* Try to collect CR space */
+	cr_res_size = pci_resource_len(pdev, 0);
+	crspace_data = kvmalloc(cr_res_size, GFP_KERNEL);
+	if (crspace_data) {
+		for (offset = 0; offset < cr_res_size; offset += 4)
+			*(u32 *)(crspace_data + offset) =
+					readl(cr_space + offset);
+
+		err = devlink_region_snapshot_create(crdump->region_crspace,
+						     crspace_data, id);
+		if (err) {
+			kvfree(crspace_data);
+			mlx4_warn(dev, "crdump: devlink create %s snapshot id %d err %d\n",
+				  region_cr_space_str, id, err);
+		} else {
+			mlx4_info(dev, "crdump: added snapshot %d to devlink region %s\n",
+				  id, region_cr_space_str);
+		}
+	} else {
+		mlx4_err(dev, "crdump: Failed to allocate crspace buffer\n");
+	}
+}
+
+static void mlx4_crdump_collect_fw_health(struct mlx4_dev *dev,
+					  u8 __iomem *cr_space,
+					  u32 id)
+{
+	struct mlx4_fw_crdump *crdump = &dev->persist->crdump;
+	u8 *health_data;
+	int offset;
+	int err;
+
+	if (!crdump->region_fw_health) {
+		mlx4_err(dev, "crdump: fw-health region is NULL\n");
+		return;
+	}
+
+	/* Try to collect health buffer */
+	health_data = kvmalloc(HEALTH_BUFFER_SIZE, GFP_KERNEL);
+	if (health_data) {
+		u8 __iomem *health_buf_start =
+				cr_space + dev->caps.health_buffer_addrs;
+
+		for (offset = 0; offset < HEALTH_BUFFER_SIZE; offset += 4)
+			*(u32 *)(health_data + offset) =
+					readl(health_buf_start + offset);
+
+		err = devlink_region_snapshot_create(crdump->region_fw_health,
+						     health_data, id);
+		if (err) {
+			kvfree(health_data);
+			mlx4_warn(dev, "crdump: devlink create %s snapshot id %d err %d\n",
+				  region_fw_health_str, id, err);
+		} else {
+			mlx4_info(dev, "crdump: added snapshot %d to devlink region %s\n",
+				  id, region_fw_health_str);
+		}
+	} else {
+		mlx4_err(dev, "crdump: Failed to allocate health buffer\n");
+	}
+}
+
+int mlx4_crdump_collect(struct mlx4_dev *dev)
+{
+	struct devlink *devlink = priv_to_devlink(mlx4_priv(dev));
+	struct mlx4_fw_crdump *crdump = &dev->persist->crdump;
+	struct pci_dev *pdev = dev->persist->pdev;
+	unsigned long cr_res_size;
+	u8 __iomem *cr_space;
+	int err;
+	u32 id;
+
+	if (!dev->caps.health_buffer_addrs) {
+		mlx4_info(dev, "crdump: FW doesn't support health buffer access, skipping\n");
+		return 0;
+	}
+
+	if (!crdump->snapshot_enable) {
+		mlx4_info(dev, "crdump: devlink snapshot disabled, skipping\n");
+		return 0;
+	}
+
+	cr_res_size = pci_resource_len(pdev, 0);
+
+	cr_space = ioremap(pci_resource_start(pdev, 0), cr_res_size);
+	if (!cr_space) {
+		mlx4_err(dev, "crdump: Failed to map pci cr region\n");
+		return -ENODEV;
+	}
+
+	/* Get the available snapshot ID for the dumps */
+	err = devlink_region_snapshot_id_get(devlink, &id);
+	if (err) {
+		mlx4_err(dev, "crdump: devlink get snapshot id err %d\n", err);
+		iounmap(cr_space);
+		return err;
+	}
+
+	crdump_enable_crspace_access(dev, cr_space);
+
+	/* Try to capture dumps */
+	mlx4_crdump_collect_crspace(dev, cr_space, id);
+	mlx4_crdump_collect_fw_health(dev, cr_space, id);
+
+	/* Release reference on the snapshot id */
+	devlink_region_snapshot_id_put(devlink, id);
+
+	crdump_disable_crspace_access(dev, cr_space);
+
+	iounmap(cr_space);
+	return 0;
+}
+
+int mlx4_crdump_init(struct mlx4_dev *dev)
+{
+	struct devlink *devlink = priv_to_devlink(mlx4_priv(dev));
+	struct mlx4_fw_crdump *crdump = &dev->persist->crdump;
+	struct pci_dev *pdev = dev->persist->pdev;
+
+	crdump->snapshot_enable = false;
+
+	/* Create cr-space region */
+	crdump->region_crspace =
+		devl_region_create(devlink,
+				   &region_cr_space_ops,
+				   MAX_NUM_OF_DUMPS_TO_STORE,
+				   pci_resource_len(pdev, 0));
+	if (IS_ERR(crdump->region_crspace))
+		mlx4_warn(dev, "crdump: create devlink region %s err %ld\n",
+			  region_cr_space_str,
+			  PTR_ERR(crdump->region_crspace));
+
+	/* Create fw-health region */
+	crdump->region_fw_health =
+		devl_region_create(devlink,
+				   &region_fw_health_ops,
+				   MAX_NUM_OF_DUMPS_TO_STORE,
+				   HEALTH_BUFFER_SIZE);
+	if (IS_ERR(crdump->region_fw_health))
+		mlx4_warn(dev, "crdump: create devlink region %s err %ld\n",
+			  region_fw_health_str,
+			  PTR_ERR(crdump->region_fw_health));
+
+	return 0;
+}
+
+void mlx4_crdump_end(struct mlx4_dev *dev)
+{
+	struct mlx4_fw_crdump *crdump = &dev->persist->crdump;
+
+	devl_region_destroy(crdump->region_fw_health);
+	devl_region_destroy(crdump->region_crspace);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx4/en_clock.c b/drivers/net/ethernet/mellanox/mlx4/en_clock.c
new file mode 100644
index 000000000..024788549
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx4/en_clock.c
@@ -0,0 +1,303 @@
+/*
+ * Copyright (c) 2012 Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+
+#include <linux/mlx4/device.h>
+#include <linux/clocksource.h>
+
+#include "mlx4_en.h"
+
+/* mlx4_en_read_clock - read raw cycle counter (to be used by time counter)
+ */
+static u64 mlx4_en_read_clock(const struct cyclecounter *tc)
+{
+	struct mlx4_en_dev *mdev =
+		container_of(tc, struct mlx4_en_dev, cycles);
+	struct mlx4_dev *dev = mdev->dev;
+
+	return mlx4_read_clock(dev) & tc->mask;
+}
+
+u64 mlx4_en_get_cqe_ts(struct mlx4_cqe *cqe)
+{
+	u64 hi, lo;
+	struct mlx4_ts_cqe *ts_cqe = (struct mlx4_ts_cqe *)cqe;
+
+	lo = (u64)be16_to_cpu(ts_cqe->timestamp_lo);
+	hi = ((u64)be32_to_cpu(ts_cqe->timestamp_hi) + !lo) << 16;
+
+	return hi | lo;
+}
+
+void mlx4_en_fill_hwtstamps(struct mlx4_en_dev *mdev,
+			    struct skb_shared_hwtstamps *hwts,
+			    u64 timestamp)
+{
+	unsigned int seq;
+	u64 nsec;
+
+	do {
+		seq = read_seqbegin(&mdev->clock_lock);
+		nsec = timecounter_cyc2time(&mdev->clock, timestamp);
+	} while (read_seqretry(&mdev->clock_lock, seq));
+
+	memset(hwts, 0, sizeof(struct skb_shared_hwtstamps));
+	hwts->hwtstamp = ns_to_ktime(nsec);
+}
+
+/**
+ * mlx4_en_remove_timestamp - disable PTP device
+ * @mdev: board private structure
+ *
+ * Stop the PTP support.
+ **/
+void mlx4_en_remove_timestamp(struct mlx4_en_dev *mdev)
+{
+	if (mdev->ptp_clock) {
+		ptp_clock_unregister(mdev->ptp_clock);
+		mdev->ptp_clock = NULL;
+		mlx4_info(mdev, "removed PHC\n");
+	}
+}
+
+#define MLX4_EN_WRAP_AROUND_SEC	10UL
+/* By scheduling the overflow check every 5 seconds, we have a reasonably
+ * good chance we wont miss a wrap around.
+ * TOTO: Use a timer instead of a work queue to increase the guarantee.
+ */
+#define MLX4_EN_OVERFLOW_PERIOD (MLX4_EN_WRAP_AROUND_SEC * HZ / 2)
+
+void mlx4_en_ptp_overflow_check(struct mlx4_en_dev *mdev)
+{
+	bool timeout = time_is_before_jiffies(mdev->last_overflow_check +
+					      MLX4_EN_OVERFLOW_PERIOD);
+	unsigned long flags;
+
+	if (timeout) {
+		write_seqlock_irqsave(&mdev->clock_lock, flags);
+		timecounter_read(&mdev->clock);
+		write_sequnlock_irqrestore(&mdev->clock_lock, flags);
+		mdev->last_overflow_check = jiffies;
+	}
+}
+
+/**
+ * mlx4_en_phc_adjfreq - adjust the frequency of the hardware clock
+ * @ptp: ptp clock structure
+ * @delta: Desired frequency change in parts per billion
+ *
+ * Adjust the frequency of the PHC cycle counter by the indicated delta from
+ * the base frequency.
+ **/
+static int mlx4_en_phc_adjfreq(struct ptp_clock_info *ptp, s32 delta)
+{
+	u64 adj;
+	u32 diff, mult;
+	int neg_adj = 0;
+	unsigned long flags;
+	struct mlx4_en_dev *mdev = container_of(ptp, struct mlx4_en_dev,
+						ptp_clock_info);
+
+	if (delta < 0) {
+		neg_adj = 1;
+		delta = -delta;
+	}
+	mult = mdev->nominal_c_mult;
+	adj = mult;
+	adj *= delta;
+	diff = div_u64(adj, 1000000000ULL);
+
+	write_seqlock_irqsave(&mdev->clock_lock, flags);
+	timecounter_read(&mdev->clock);
+	mdev->cycles.mult = neg_adj ? mult - diff : mult + diff;
+	write_sequnlock_irqrestore(&mdev->clock_lock, flags);
+
+	return 0;
+}
+
+/**
+ * mlx4_en_phc_adjtime - Shift the time of the hardware clock
+ * @ptp: ptp clock structure
+ * @delta: Desired change in nanoseconds
+ *
+ * Adjust the timer by resetting the timecounter structure.
+ **/
+static int mlx4_en_phc_adjtime(struct ptp_clock_info *ptp, s64 delta)
+{
+	struct mlx4_en_dev *mdev = container_of(ptp, struct mlx4_en_dev,
+						ptp_clock_info);
+	unsigned long flags;
+
+	write_seqlock_irqsave(&mdev->clock_lock, flags);
+	timecounter_adjtime(&mdev->clock, delta);
+	write_sequnlock_irqrestore(&mdev->clock_lock, flags);
+
+	return 0;
+}
+
+/**
+ * mlx4_en_phc_gettime - Reads the current time from the hardware clock
+ * @ptp: ptp clock structure
+ * @ts: timespec structure to hold the current time value
+ *
+ * Read the timecounter and return the correct value in ns after converting
+ * it into a struct timespec.
+ **/
+static int mlx4_en_phc_gettime(struct ptp_clock_info *ptp,
+			       struct timespec64 *ts)
+{
+	struct mlx4_en_dev *mdev = container_of(ptp, struct mlx4_en_dev,
+						ptp_clock_info);
+	unsigned long flags;
+	u64 ns;
+
+	write_seqlock_irqsave(&mdev->clock_lock, flags);
+	ns = timecounter_read(&mdev->clock);
+	write_sequnlock_irqrestore(&mdev->clock_lock, flags);
+
+	*ts = ns_to_timespec64(ns);
+
+	return 0;
+}
+
+/**
+ * mlx4_en_phc_settime - Set the current time on the hardware clock
+ * @ptp: ptp clock structure
+ * @ts: timespec containing the new time for the cycle counter
+ *
+ * Reset the timecounter to use a new base value instead of the kernel
+ * wall timer value.
+ **/
+static int mlx4_en_phc_settime(struct ptp_clock_info *ptp,
+			       const struct timespec64 *ts)
+{
+	struct mlx4_en_dev *mdev = container_of(ptp, struct mlx4_en_dev,
+						ptp_clock_info);
+	u64 ns = timespec64_to_ns(ts);
+	unsigned long flags;
+
+	/* reset the timecounter */
+	write_seqlock_irqsave(&mdev->clock_lock, flags);
+	timecounter_init(&mdev->clock, &mdev->cycles, ns);
+	write_sequnlock_irqrestore(&mdev->clock_lock, flags);
+
+	return 0;
+}
+
+/**
+ * mlx4_en_phc_enable - enable or disable an ancillary feature
+ * @ptp: ptp clock structure
+ * @request: Desired resource to enable or disable
+ * @on: Caller passes one to enable or zero to disable
+ *
+ * Enable (or disable) ancillary features of the PHC subsystem.
+ * Currently, no ancillary features are supported.
+ **/
+static int mlx4_en_phc_enable(struct ptp_clock_info __always_unused *ptp,
+			      struct ptp_clock_request __always_unused *request,
+			      int __always_unused on)
+{
+	return -EOPNOTSUPP;
+}
+
+static const struct ptp_clock_info mlx4_en_ptp_clock_info = {
+	.owner		= THIS_MODULE,
+	.max_adj	= 100000000,
+	.n_alarm	= 0,
+	.n_ext_ts	= 0,
+	.n_per_out	= 0,
+	.n_pins		= 0,
+	.pps		= 0,
+	.adjfreq	= mlx4_en_phc_adjfreq,
+	.adjtime	= mlx4_en_phc_adjtime,
+	.gettime64	= mlx4_en_phc_gettime,
+	.settime64	= mlx4_en_phc_settime,
+	.enable		= mlx4_en_phc_enable,
+};
+
+
+/* This function calculates the max shift that enables the user range
+ * of MLX4_EN_WRAP_AROUND_SEC values in the cycles register.
+ */
+static u32 freq_to_shift(u16 freq)
+{
+	u32 freq_khz = freq * 1000;
+	u64 max_val_cycles = freq_khz * 1000 * MLX4_EN_WRAP_AROUND_SEC;
+	u64 max_val_cycles_rounded = 1ULL << fls64(max_val_cycles - 1);
+	/* calculate max possible multiplier in order to fit in 64bit */
+	u64 max_mul = div64_u64(ULLONG_MAX, max_val_cycles_rounded);
+
+	/* This comes from the reverse of clocksource_khz2mult */
+	return ilog2(div_u64(max_mul * freq_khz, 1000000));
+}
+
+void mlx4_en_init_timestamp(struct mlx4_en_dev *mdev)
+{
+	struct mlx4_dev *dev = mdev->dev;
+	unsigned long flags;
+
+	/* mlx4_en_init_timestamp is called for each netdev.
+	 * mdev->ptp_clock is common for all ports, skip initialization if
+	 * was done for other port.
+	 */
+	if (mdev->ptp_clock)
+		return;
+
+	seqlock_init(&mdev->clock_lock);
+
+	memset(&mdev->cycles, 0, sizeof(mdev->cycles));
+	mdev->cycles.read = mlx4_en_read_clock;
+	mdev->cycles.mask = CLOCKSOURCE_MASK(48);
+	mdev->cycles.shift = freq_to_shift(dev->caps.hca_core_clock);
+	mdev->cycles.mult =
+		clocksource_khz2mult(1000 * dev->caps.hca_core_clock, mdev->cycles.shift);
+	mdev->nominal_c_mult = mdev->cycles.mult;
+
+	write_seqlock_irqsave(&mdev->clock_lock, flags);
+	timecounter_init(&mdev->clock, &mdev->cycles,
+			 ktime_to_ns(ktime_get_real()));
+	write_sequnlock_irqrestore(&mdev->clock_lock, flags);
+
+	/* Configure the PHC */
+	mdev->ptp_clock_info = mlx4_en_ptp_clock_info;
+	snprintf(mdev->ptp_clock_info.name, 16, "mlx4 ptp");
+
+	mdev->ptp_clock = ptp_clock_register(&mdev->ptp_clock_info,
+					     &mdev->pdev->dev);
+	if (IS_ERR(mdev->ptp_clock)) {
+		mdev->ptp_clock = NULL;
+		mlx4_err(mdev, "ptp_clock_register failed\n");
+	} else if (mdev->ptp_clock) {
+		mlx4_info(mdev, "registered PHC clock\n");
+	}
+
+}
diff --git a/drivers/net/ethernet/mellanox/mlx4/en_cq.c b/drivers/net/ethernet/mellanox/mlx4/en_cq.c
new file mode 100644
index 000000000..1184ac575
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx4/en_cq.c
@@ -0,0 +1,212 @@
+/*
+ * Copyright (c) 2007 Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+
+#include <linux/mlx4/cq.h>
+#include <linux/mlx4/qp.h>
+#include <linux/mlx4/cmd.h>
+
+#include "mlx4_en.h"
+
+static void mlx4_en_cq_event(struct mlx4_cq *cq, enum mlx4_event event)
+{
+	return;
+}
+
+
+int mlx4_en_create_cq(struct mlx4_en_priv *priv,
+		      struct mlx4_en_cq **pcq,
+		      int entries, int ring, enum cq_type mode,
+		      int node)
+{
+	struct mlx4_en_dev *mdev = priv->mdev;
+	struct mlx4_en_cq *cq;
+	int err;
+
+	cq = kzalloc_node(sizeof(*cq), GFP_KERNEL, node);
+	if (!cq) {
+		en_err(priv, "Failed to allocate CQ structure\n");
+		return -ENOMEM;
+	}
+
+	cq->size = entries;
+	cq->buf_size = cq->size * mdev->dev->caps.cqe_size;
+
+	cq->ring = ring;
+	cq->type = mode;
+	cq->vector = mdev->dev->caps.num_comp_vectors;
+
+	/* Allocate HW buffers on provided NUMA node.
+	 * dev->numa_node is used in mtt range allocation flow.
+	 */
+	set_dev_node(&mdev->dev->persist->pdev->dev, node);
+	err = mlx4_alloc_hwq_res(mdev->dev, &cq->wqres,
+				cq->buf_size);
+	set_dev_node(&mdev->dev->persist->pdev->dev, mdev->dev->numa_node);
+	if (err)
+		goto err_cq;
+
+	cq->buf = (struct mlx4_cqe *)cq->wqres.buf.direct.buf;
+	*pcq = cq;
+
+	return 0;
+
+err_cq:
+	kfree(cq);
+	*pcq = NULL;
+	return err;
+}
+
+int mlx4_en_activate_cq(struct mlx4_en_priv *priv, struct mlx4_en_cq *cq,
+			int cq_idx)
+{
+	struct mlx4_en_dev *mdev = priv->mdev;
+	int irq, err = 0;
+	int timestamp_en = 0;
+	bool assigned_eq = false;
+
+	cq->dev = mdev->pndev[priv->port];
+	cq->mcq.set_ci_db  = cq->wqres.db.db;
+	cq->mcq.arm_db     = cq->wqres.db.db + 1;
+	*cq->mcq.set_ci_db = 0;
+	*cq->mcq.arm_db    = 0;
+	memset(cq->buf, 0, cq->buf_size);
+
+	if (cq->type == RX) {
+		if (!mlx4_is_eq_vector_valid(mdev->dev, priv->port,
+					     cq->vector)) {
+			cq->vector = cpumask_first(priv->rx_ring[cq->ring]->affinity_mask);
+
+			err = mlx4_assign_eq(mdev->dev, priv->port,
+					     &cq->vector);
+			if (err) {
+				mlx4_err(mdev, "Failed assigning an EQ to CQ vector %d\n",
+					 cq->vector);
+				goto free_eq;
+			}
+
+			assigned_eq = true;
+		}
+		irq = mlx4_eq_get_irq(mdev->dev, cq->vector);
+		cq->aff_mask = irq_get_effective_affinity_mask(irq);
+	} else {
+		/* For TX we use the same irq per
+		ring we assigned for the RX    */
+		struct mlx4_en_cq *rx_cq;
+
+		cq_idx = cq_idx % priv->rx_ring_num;
+		rx_cq = priv->rx_cq[cq_idx];
+		cq->vector = rx_cq->vector;
+	}
+
+	if (cq->type == RX)
+		cq->size = priv->rx_ring[cq->ring]->actual_size;
+
+	if ((cq->type != RX && priv->hwtstamp_config.tx_type) ||
+	    (cq->type == RX && priv->hwtstamp_config.rx_filter))
+		timestamp_en = 1;
+
+	cq->mcq.usage = MLX4_RES_USAGE_DRIVER;
+	err = mlx4_cq_alloc(mdev->dev, cq->size, &cq->wqres.mtt,
+			    &mdev->priv_uar, cq->wqres.db.dma, &cq->mcq,
+			    cq->vector, 0, timestamp_en, &cq->wqres.buf, false);
+	if (err)
+		goto free_eq;
+
+	cq->mcq.event = mlx4_en_cq_event;
+
+	switch (cq->type) {
+	case TX:
+		cq->mcq.comp = mlx4_en_tx_irq;
+		netif_napi_add_tx(cq->dev, &cq->napi, mlx4_en_poll_tx_cq);
+		napi_enable(&cq->napi);
+		break;
+	case RX:
+		cq->mcq.comp = mlx4_en_rx_irq;
+		netif_napi_add(cq->dev, &cq->napi, mlx4_en_poll_rx_cq);
+		napi_enable(&cq->napi);
+		break;
+	case TX_XDP:
+		/* nothing regarding napi, it's shared with rx ring */
+		cq->xdp_busy = false;
+		break;
+	}
+
+	return 0;
+
+free_eq:
+	if (assigned_eq)
+		mlx4_release_eq(mdev->dev, cq->vector);
+	cq->vector = mdev->dev->caps.num_comp_vectors;
+	return err;
+}
+
+void mlx4_en_destroy_cq(struct mlx4_en_priv *priv, struct mlx4_en_cq **pcq)
+{
+	struct mlx4_en_dev *mdev = priv->mdev;
+	struct mlx4_en_cq *cq = *pcq;
+
+	mlx4_free_hwq_res(mdev->dev, &cq->wqres, cq->buf_size);
+	if (mlx4_is_eq_vector_valid(mdev->dev, priv->port, cq->vector) &&
+	    cq->type == RX)
+		mlx4_release_eq(priv->mdev->dev, cq->vector);
+	cq->vector = 0;
+	cq->buf_size = 0;
+	cq->buf = NULL;
+	kfree(cq);
+	*pcq = NULL;
+}
+
+void mlx4_en_deactivate_cq(struct mlx4_en_priv *priv, struct mlx4_en_cq *cq)
+{
+	if (cq->type != TX_XDP) {
+		napi_disable(&cq->napi);
+		netif_napi_del(&cq->napi);
+	}
+
+	mlx4_cq_free(priv->mdev->dev, &cq->mcq);
+}
+
+/* Set rx cq moderation parameters */
+int mlx4_en_set_cq_moder(struct mlx4_en_priv *priv, struct mlx4_en_cq *cq)
+{
+	return mlx4_cq_modify(priv->mdev->dev, &cq->mcq,
+			      cq->moder_cnt, cq->moder_time);
+}
+
+void mlx4_en_arm_cq(struct mlx4_en_priv *priv, struct mlx4_en_cq *cq)
+{
+	mlx4_cq_arm(&cq->mcq, MLX4_CQ_DB_REQ_NOT, priv->mdev->uar_map,
+		    &priv->mdev->uar_lock);
+}
+
+
diff --git a/drivers/net/ethernet/mellanox/mlx4/en_dcb_nl.c b/drivers/net/ethernet/mellanox/mlx4/en_dcb_nl.c
new file mode 100644
index 000000000..752a72499
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx4/en_dcb_nl.c
@@ -0,0 +1,753 @@
+/*
+ * Copyright (c) 2011 Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+
+#include <linux/dcbnl.h>
+#include <linux/math64.h>
+
+#include "mlx4_en.h"
+#include "fw_qos.h"
+
+enum {
+	MLX4_CEE_STATE_DOWN   = 0,
+	MLX4_CEE_STATE_UP     = 1,
+};
+
+/* Definitions for QCN
+ */
+
+struct mlx4_congestion_control_mb_prio_802_1_qau_params {
+	__be32 modify_enable_high;
+	__be32 modify_enable_low;
+	__be32 reserved1;
+	__be32 extended_enable;
+	__be32 rppp_max_rps;
+	__be32 rpg_time_reset;
+	__be32 rpg_byte_reset;
+	__be32 rpg_threshold;
+	__be32 rpg_max_rate;
+	__be32 rpg_ai_rate;
+	__be32 rpg_hai_rate;
+	__be32 rpg_gd;
+	__be32 rpg_min_dec_fac;
+	__be32 rpg_min_rate;
+	__be32 max_time_rise;
+	__be32 max_byte_rise;
+	__be32 max_qdelta;
+	__be32 min_qoffset;
+	__be32 gd_coefficient;
+	__be32 reserved2[5];
+	__be32 cp_sample_base;
+	__be32 reserved3[39];
+};
+
+struct mlx4_congestion_control_mb_prio_802_1_qau_statistics {
+	__be64 rppp_rp_centiseconds;
+	__be32 reserved1;
+	__be32 ignored_cnm;
+	__be32 rppp_created_rps;
+	__be32 estimated_total_rate;
+	__be32 max_active_rate_limiter_index;
+	__be32 dropped_cnms_busy_fw;
+	__be32 reserved2;
+	__be32 cnms_handled_successfully;
+	__be32 min_total_limiters_rate;
+	__be32 max_total_limiters_rate;
+	__be32 reserved3[4];
+};
+
+static u8 mlx4_en_dcbnl_getcap(struct net_device *dev, int capid, u8 *cap)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+
+	switch (capid) {
+	case DCB_CAP_ATTR_PFC:
+		*cap = true;
+		break;
+	case DCB_CAP_ATTR_DCBX:
+		*cap = priv->dcbx_cap;
+		break;
+	case DCB_CAP_ATTR_PFC_TCS:
+		*cap = 1 <<  mlx4_max_tc(priv->mdev->dev);
+		break;
+	default:
+		*cap = false;
+		break;
+	}
+
+	return 0;
+}
+
+static u8 mlx4_en_dcbnl_getpfcstate(struct net_device *netdev)
+{
+	struct mlx4_en_priv *priv = netdev_priv(netdev);
+
+	return priv->cee_config.pfc_state;
+}
+
+static void mlx4_en_dcbnl_setpfcstate(struct net_device *netdev, u8 state)
+{
+	struct mlx4_en_priv *priv = netdev_priv(netdev);
+
+	priv->cee_config.pfc_state = state;
+}
+
+static void mlx4_en_dcbnl_get_pfc_cfg(struct net_device *netdev, int priority,
+				      u8 *setting)
+{
+	struct mlx4_en_priv *priv = netdev_priv(netdev);
+
+	*setting = priv->cee_config.dcb_pfc[priority];
+}
+
+static void mlx4_en_dcbnl_set_pfc_cfg(struct net_device *netdev, int priority,
+				      u8 setting)
+{
+	struct mlx4_en_priv *priv = netdev_priv(netdev);
+
+	priv->cee_config.dcb_pfc[priority] = setting;
+	priv->cee_config.pfc_state = true;
+}
+
+static int mlx4_en_dcbnl_getnumtcs(struct net_device *netdev, int tcid, u8 *num)
+{
+	struct mlx4_en_priv *priv = netdev_priv(netdev);
+
+	if (!(priv->flags & MLX4_EN_FLAG_DCB_ENABLED))
+		return -EINVAL;
+
+	if (tcid == DCB_NUMTCS_ATTR_PFC)
+		*num = mlx4_max_tc(priv->mdev->dev);
+	else
+		*num = 0;
+
+	return 0;
+}
+
+static u8 mlx4_en_dcbnl_set_all(struct net_device *netdev)
+{
+	struct mlx4_en_priv *priv = netdev_priv(netdev);
+	struct mlx4_en_port_profile *prof = priv->prof;
+	struct mlx4_en_dev *mdev = priv->mdev;
+	u8 tx_pause, tx_ppp, rx_pause, rx_ppp;
+
+	if (!(priv->dcbx_cap & DCB_CAP_DCBX_VER_CEE))
+		return 1;
+
+	if (priv->cee_config.pfc_state) {
+		int tc;
+		rx_ppp = prof->rx_ppp;
+		tx_ppp = prof->tx_ppp;
+
+		for (tc = 0; tc < CEE_DCBX_MAX_PRIO; tc++) {
+			u8 tc_mask = 1 << tc;
+
+			switch (priv->cee_config.dcb_pfc[tc]) {
+			case pfc_disabled:
+				tx_ppp &= ~tc_mask;
+				rx_ppp &= ~tc_mask;
+				break;
+			case pfc_enabled_full:
+				tx_ppp |= tc_mask;
+				rx_ppp |= tc_mask;
+				break;
+			case pfc_enabled_tx:
+				tx_ppp |= tc_mask;
+				rx_ppp &= ~tc_mask;
+				break;
+			case pfc_enabled_rx:
+				tx_ppp &= ~tc_mask;
+				rx_ppp |= tc_mask;
+				break;
+			default:
+				break;
+			}
+		}
+		rx_pause = !!(rx_ppp || tx_ppp) ? 0 : prof->rx_pause;
+		tx_pause = !!(rx_ppp || tx_ppp) ? 0 : prof->tx_pause;
+	} else {
+		rx_ppp = 0;
+		tx_ppp = 0;
+		rx_pause = prof->rx_pause;
+		tx_pause = prof->tx_pause;
+	}
+
+	if (mlx4_SET_PORT_general(mdev->dev, priv->port,
+				  priv->rx_skb_size + ETH_FCS_LEN,
+				  tx_pause, tx_ppp, rx_pause, rx_ppp)) {
+		en_err(priv, "Failed setting pause params\n");
+		return 1;
+	}
+
+	prof->tx_ppp = tx_ppp;
+	prof->rx_ppp = rx_ppp;
+	prof->tx_pause = tx_pause;
+	prof->rx_pause = rx_pause;
+
+	return 0;
+}
+
+static u8 mlx4_en_dcbnl_get_state(struct net_device *dev)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+
+	if (priv->flags & MLX4_EN_FLAG_DCB_ENABLED)
+		return MLX4_CEE_STATE_UP;
+
+	return MLX4_CEE_STATE_DOWN;
+}
+
+static u8 mlx4_en_dcbnl_set_state(struct net_device *dev, u8 state)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	int num_tcs = 0;
+
+	if (!(priv->dcbx_cap & DCB_CAP_DCBX_VER_CEE))
+		return 1;
+
+	if (!!(state) == !!(priv->flags & MLX4_EN_FLAG_DCB_ENABLED))
+		return 0;
+
+	if (state) {
+		priv->flags |= MLX4_EN_FLAG_DCB_ENABLED;
+		num_tcs = IEEE_8021QAZ_MAX_TCS;
+	} else {
+		priv->flags &= ~MLX4_EN_FLAG_DCB_ENABLED;
+	}
+
+	if (mlx4_en_alloc_tx_queue_per_tc(dev, num_tcs))
+		return 1;
+
+	return 0;
+}
+
+/* On success returns a non-zero 802.1p user priority bitmap
+ * otherwise returns 0 as the invalid user priority bitmap to
+ * indicate an error.
+ */
+static int mlx4_en_dcbnl_getapp(struct net_device *netdev, u8 idtype, u16 id)
+{
+	struct mlx4_en_priv *priv = netdev_priv(netdev);
+	struct dcb_app app = {
+				.selector = idtype,
+				.protocol = id,
+			     };
+	if (!(priv->dcbx_cap & DCB_CAP_DCBX_VER_CEE))
+		return 0;
+
+	return dcb_getapp(netdev, &app);
+}
+
+static int mlx4_en_dcbnl_setapp(struct net_device *netdev, u8 idtype,
+				u16 id, u8 up)
+{
+	struct mlx4_en_priv *priv = netdev_priv(netdev);
+	struct dcb_app app;
+
+	if (!(priv->dcbx_cap & DCB_CAP_DCBX_VER_CEE))
+		return -EINVAL;
+
+	memset(&app, 0, sizeof(struct dcb_app));
+	app.selector = idtype;
+	app.protocol = id;
+	app.priority = up;
+
+	return dcb_setapp(netdev, &app);
+}
+
+static int mlx4_en_dcbnl_ieee_getets(struct net_device *dev,
+				   struct ieee_ets *ets)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	struct ieee_ets *my_ets = &priv->ets;
+
+	if (!my_ets)
+		return -EINVAL;
+
+	ets->ets_cap = IEEE_8021QAZ_MAX_TCS;
+	ets->cbs = my_ets->cbs;
+	memcpy(ets->tc_tx_bw, my_ets->tc_tx_bw, sizeof(ets->tc_tx_bw));
+	memcpy(ets->tc_tsa, my_ets->tc_tsa, sizeof(ets->tc_tsa));
+	memcpy(ets->prio_tc, my_ets->prio_tc, sizeof(ets->prio_tc));
+
+	return 0;
+}
+
+static int mlx4_en_ets_validate(struct mlx4_en_priv *priv, struct ieee_ets *ets)
+{
+	int i;
+	int total_ets_bw = 0;
+	int has_ets_tc = 0;
+
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) {
+		if (ets->prio_tc[i] >= MLX4_EN_NUM_UP_HIGH) {
+			en_err(priv, "Bad priority in UP <=> TC mapping. TC: %d, UP: %d\n",
+					i, ets->prio_tc[i]);
+			return -EINVAL;
+		}
+
+		switch (ets->tc_tsa[i]) {
+		case IEEE_8021QAZ_TSA_VENDOR:
+		case IEEE_8021QAZ_TSA_STRICT:
+			break;
+		case IEEE_8021QAZ_TSA_ETS:
+			has_ets_tc = 1;
+			total_ets_bw += ets->tc_tx_bw[i];
+			break;
+		default:
+			en_err(priv, "TC[%d]: Not supported TSA: %d\n",
+					i, ets->tc_tsa[i]);
+			return -EOPNOTSUPP;
+		}
+	}
+
+	if (has_ets_tc && total_ets_bw != MLX4_EN_BW_MAX) {
+		en_err(priv, "Bad ETS BW sum: %d. Should be exactly 100%%\n",
+				total_ets_bw);
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int mlx4_en_config_port_scheduler(struct mlx4_en_priv *priv,
+		struct ieee_ets *ets, u16 *ratelimit)
+{
+	struct mlx4_en_dev *mdev = priv->mdev;
+	int num_strict = 0;
+	int i;
+	__u8 tc_tx_bw[IEEE_8021QAZ_MAX_TCS] = { 0 };
+	__u8 pg[IEEE_8021QAZ_MAX_TCS] = { 0 };
+
+	ets = ets ?: &priv->ets;
+	ratelimit = ratelimit ?: priv->maxrate;
+
+	/* higher TC means higher priority => lower pg */
+	for (i = IEEE_8021QAZ_MAX_TCS - 1; i >= 0; i--) {
+		switch (ets->tc_tsa[i]) {
+		case IEEE_8021QAZ_TSA_VENDOR:
+			pg[i] = MLX4_EN_TC_VENDOR;
+			tc_tx_bw[i] = MLX4_EN_BW_MAX;
+			break;
+		case IEEE_8021QAZ_TSA_STRICT:
+			pg[i] = num_strict++;
+			tc_tx_bw[i] = MLX4_EN_BW_MAX;
+			break;
+		case IEEE_8021QAZ_TSA_ETS:
+			pg[i] = MLX4_EN_TC_ETS;
+			tc_tx_bw[i] = ets->tc_tx_bw[i] ?: MLX4_EN_BW_MIN;
+			break;
+		}
+	}
+
+	return mlx4_SET_PORT_SCHEDULER(mdev->dev, priv->port, tc_tx_bw, pg,
+			ratelimit);
+}
+
+static int
+mlx4_en_dcbnl_ieee_setets(struct net_device *dev, struct ieee_ets *ets)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	struct mlx4_en_dev *mdev = priv->mdev;
+	int err;
+
+	err = mlx4_en_ets_validate(priv, ets);
+	if (err)
+		return err;
+
+	err = mlx4_SET_PORT_PRIO2TC(mdev->dev, priv->port, ets->prio_tc);
+	if (err)
+		return err;
+
+	err = mlx4_en_config_port_scheduler(priv, ets, NULL);
+	if (err)
+		return err;
+
+	memcpy(&priv->ets, ets, sizeof(priv->ets));
+
+	return 0;
+}
+
+static int mlx4_en_dcbnl_ieee_getpfc(struct net_device *dev,
+		struct ieee_pfc *pfc)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+
+	pfc->pfc_cap = IEEE_8021QAZ_MAX_TCS;
+	pfc->pfc_en = priv->prof->tx_ppp;
+
+	return 0;
+}
+
+static int mlx4_en_dcbnl_ieee_setpfc(struct net_device *dev,
+		struct ieee_pfc *pfc)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	struct mlx4_en_port_profile *prof = priv->prof;
+	struct mlx4_en_dev *mdev = priv->mdev;
+	u32 tx_pause, tx_ppp, rx_pause, rx_ppp;
+	int err;
+
+	en_dbg(DRV, priv, "cap: 0x%x en: 0x%x mbc: 0x%x delay: %d\n",
+			pfc->pfc_cap,
+			pfc->pfc_en,
+			pfc->mbc,
+			pfc->delay);
+
+	rx_pause = prof->rx_pause && !pfc->pfc_en;
+	tx_pause = prof->tx_pause && !pfc->pfc_en;
+	rx_ppp = pfc->pfc_en;
+	tx_ppp = pfc->pfc_en;
+
+	err = mlx4_SET_PORT_general(mdev->dev, priv->port,
+				    priv->rx_skb_size + ETH_FCS_LEN,
+				    tx_pause, tx_ppp, rx_pause, rx_ppp);
+	if (err) {
+		en_err(priv, "Failed setting pause params\n");
+		return err;
+	}
+
+	mlx4_en_update_pfc_stats_bitmap(mdev->dev, &priv->stats_bitmap,
+					rx_ppp, rx_pause, tx_ppp, tx_pause);
+
+	prof->tx_ppp = tx_ppp;
+	prof->rx_ppp = rx_ppp;
+	prof->rx_pause = rx_pause;
+	prof->tx_pause = tx_pause;
+
+	return err;
+}
+
+static u8 mlx4_en_dcbnl_getdcbx(struct net_device *dev)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+
+	return priv->dcbx_cap;
+}
+
+static u8 mlx4_en_dcbnl_setdcbx(struct net_device *dev, u8 mode)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	struct ieee_ets ets = {0};
+	struct ieee_pfc pfc = {0};
+
+	if (mode == priv->dcbx_cap)
+		return 0;
+
+	if ((mode & DCB_CAP_DCBX_LLD_MANAGED) ||
+	    ((mode & DCB_CAP_DCBX_VER_IEEE) &&
+	     (mode & DCB_CAP_DCBX_VER_CEE)) ||
+	    !(mode & DCB_CAP_DCBX_HOST))
+		goto err;
+
+	priv->dcbx_cap = mode;
+
+	ets.ets_cap = IEEE_8021QAZ_MAX_TCS;
+	pfc.pfc_cap = IEEE_8021QAZ_MAX_TCS;
+
+	if (mode & DCB_CAP_DCBX_VER_IEEE) {
+		if (mlx4_en_dcbnl_ieee_setets(dev, &ets))
+			goto err;
+		if (mlx4_en_dcbnl_ieee_setpfc(dev, &pfc))
+			goto err;
+	} else if (mode & DCB_CAP_DCBX_VER_CEE) {
+		if (mlx4_en_dcbnl_set_all(dev))
+			goto err;
+	} else {
+		if (mlx4_en_dcbnl_ieee_setets(dev, &ets))
+			goto err;
+		if (mlx4_en_dcbnl_ieee_setpfc(dev, &pfc))
+			goto err;
+		if (mlx4_en_alloc_tx_queue_per_tc(dev, 0))
+			goto err;
+	}
+
+	return 0;
+err:
+	return 1;
+}
+
+#define MLX4_RATELIMIT_UNITS_IN_KB 100000 /* rate-limit HW unit in Kbps */
+static int mlx4_en_dcbnl_ieee_getmaxrate(struct net_device *dev,
+				   struct ieee_maxrate *maxrate)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	int i;
+
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++)
+		maxrate->tc_maxrate[i] =
+			priv->maxrate[i] * MLX4_RATELIMIT_UNITS_IN_KB;
+
+	return 0;
+}
+
+static int mlx4_en_dcbnl_ieee_setmaxrate(struct net_device *dev,
+		struct ieee_maxrate *maxrate)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	u16 tmp[IEEE_8021QAZ_MAX_TCS];
+	int i, err;
+
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) {
+		/* Convert from Kbps into HW units, rounding result up.
+		 * Setting to 0, means unlimited BW.
+		 */
+		tmp[i] = div_u64(maxrate->tc_maxrate[i] +
+				 MLX4_RATELIMIT_UNITS_IN_KB - 1,
+				 MLX4_RATELIMIT_UNITS_IN_KB);
+	}
+
+	err = mlx4_en_config_port_scheduler(priv, NULL, tmp);
+	if (err)
+		return err;
+
+	memcpy(priv->maxrate, tmp, sizeof(priv->maxrate));
+
+	return 0;
+}
+
+#define RPG_ENABLE_BIT	31
+#define CN_TAG_BIT	30
+
+static int mlx4_en_dcbnl_ieee_getqcn(struct net_device *dev,
+				     struct ieee_qcn *qcn)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	struct mlx4_congestion_control_mb_prio_802_1_qau_params *hw_qcn;
+	struct mlx4_cmd_mailbox *mailbox_out = NULL;
+	u64 mailbox_in_dma = 0;
+	u32 inmod = 0;
+	int i, err;
+
+	if (!(priv->mdev->dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_QCN))
+		return -EOPNOTSUPP;
+
+	mailbox_out = mlx4_alloc_cmd_mailbox(priv->mdev->dev);
+	if (IS_ERR(mailbox_out))
+		return -ENOMEM;
+	hw_qcn =
+	(struct mlx4_congestion_control_mb_prio_802_1_qau_params *)
+	mailbox_out->buf;
+
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) {
+		inmod = priv->port | ((1 << i) << 8) |
+			 (MLX4_CTRL_ALGO_802_1_QAU_REACTION_POINT << 16);
+		err = mlx4_cmd_box(priv->mdev->dev, mailbox_in_dma,
+				   mailbox_out->dma,
+				   inmod, MLX4_CONGESTION_CONTROL_GET_PARAMS,
+				   MLX4_CMD_CONGESTION_CTRL_OPCODE,
+				   MLX4_CMD_TIME_CLASS_C,
+				   MLX4_CMD_NATIVE);
+		if (err) {
+			mlx4_free_cmd_mailbox(priv->mdev->dev, mailbox_out);
+			return err;
+		}
+
+		qcn->rpg_enable[i] =
+			be32_to_cpu(hw_qcn->extended_enable) >> RPG_ENABLE_BIT;
+		qcn->rppp_max_rps[i] =
+			be32_to_cpu(hw_qcn->rppp_max_rps);
+		qcn->rpg_time_reset[i] =
+			be32_to_cpu(hw_qcn->rpg_time_reset);
+		qcn->rpg_byte_reset[i] =
+			be32_to_cpu(hw_qcn->rpg_byte_reset);
+		qcn->rpg_threshold[i] =
+			be32_to_cpu(hw_qcn->rpg_threshold);
+		qcn->rpg_max_rate[i] =
+			be32_to_cpu(hw_qcn->rpg_max_rate);
+		qcn->rpg_ai_rate[i] =
+			be32_to_cpu(hw_qcn->rpg_ai_rate);
+		qcn->rpg_hai_rate[i] =
+			be32_to_cpu(hw_qcn->rpg_hai_rate);
+		qcn->rpg_gd[i] =
+			be32_to_cpu(hw_qcn->rpg_gd);
+		qcn->rpg_min_dec_fac[i] =
+			be32_to_cpu(hw_qcn->rpg_min_dec_fac);
+		qcn->rpg_min_rate[i] =
+			be32_to_cpu(hw_qcn->rpg_min_rate);
+		qcn->cndd_state_machine[i] =
+			priv->cndd_state[i];
+	}
+	mlx4_free_cmd_mailbox(priv->mdev->dev, mailbox_out);
+	return 0;
+}
+
+static int mlx4_en_dcbnl_ieee_setqcn(struct net_device *dev,
+				     struct ieee_qcn *qcn)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	struct mlx4_congestion_control_mb_prio_802_1_qau_params *hw_qcn;
+	struct mlx4_cmd_mailbox *mailbox_in = NULL;
+	u64 mailbox_in_dma = 0;
+	u32 inmod = 0;
+	int i, err;
+#define MODIFY_ENABLE_HIGH_MASK 0xc0000000
+#define MODIFY_ENABLE_LOW_MASK 0xffc00000
+
+	if (!(priv->mdev->dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_QCN))
+		return -EOPNOTSUPP;
+
+	mailbox_in = mlx4_alloc_cmd_mailbox(priv->mdev->dev);
+	if (IS_ERR(mailbox_in))
+		return -ENOMEM;
+
+	mailbox_in_dma = mailbox_in->dma;
+	hw_qcn =
+	(struct mlx4_congestion_control_mb_prio_802_1_qau_params *)mailbox_in->buf;
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) {
+		inmod = priv->port | ((1 << i) << 8) |
+			 (MLX4_CTRL_ALGO_802_1_QAU_REACTION_POINT << 16);
+
+		/* Before updating QCN parameter,
+		 * need to set it's modify enable bit to 1
+		 */
+
+		hw_qcn->modify_enable_high = cpu_to_be32(
+						MODIFY_ENABLE_HIGH_MASK);
+		hw_qcn->modify_enable_low = cpu_to_be32(MODIFY_ENABLE_LOW_MASK);
+
+		hw_qcn->extended_enable = cpu_to_be32(qcn->rpg_enable[i] << RPG_ENABLE_BIT);
+		hw_qcn->rppp_max_rps = cpu_to_be32(qcn->rppp_max_rps[i]);
+		hw_qcn->rpg_time_reset = cpu_to_be32(qcn->rpg_time_reset[i]);
+		hw_qcn->rpg_byte_reset = cpu_to_be32(qcn->rpg_byte_reset[i]);
+		hw_qcn->rpg_threshold = cpu_to_be32(qcn->rpg_threshold[i]);
+		hw_qcn->rpg_max_rate = cpu_to_be32(qcn->rpg_max_rate[i]);
+		hw_qcn->rpg_ai_rate = cpu_to_be32(qcn->rpg_ai_rate[i]);
+		hw_qcn->rpg_hai_rate = cpu_to_be32(qcn->rpg_hai_rate[i]);
+		hw_qcn->rpg_gd = cpu_to_be32(qcn->rpg_gd[i]);
+		hw_qcn->rpg_min_dec_fac = cpu_to_be32(qcn->rpg_min_dec_fac[i]);
+		hw_qcn->rpg_min_rate = cpu_to_be32(qcn->rpg_min_rate[i]);
+		priv->cndd_state[i] = qcn->cndd_state_machine[i];
+		if (qcn->cndd_state_machine[i] == DCB_CNDD_INTERIOR_READY)
+			hw_qcn->extended_enable |= cpu_to_be32(1 << CN_TAG_BIT);
+
+		err = mlx4_cmd(priv->mdev->dev, mailbox_in_dma, inmod,
+			       MLX4_CONGESTION_CONTROL_SET_PARAMS,
+			       MLX4_CMD_CONGESTION_CTRL_OPCODE,
+			       MLX4_CMD_TIME_CLASS_C,
+			       MLX4_CMD_NATIVE);
+		if (err) {
+			mlx4_free_cmd_mailbox(priv->mdev->dev, mailbox_in);
+			return err;
+		}
+	}
+	mlx4_free_cmd_mailbox(priv->mdev->dev, mailbox_in);
+	return 0;
+}
+
+static int mlx4_en_dcbnl_ieee_getqcnstats(struct net_device *dev,
+					  struct ieee_qcn_stats *qcn_stats)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	struct mlx4_congestion_control_mb_prio_802_1_qau_statistics *hw_qcn_stats;
+	struct mlx4_cmd_mailbox *mailbox_out = NULL;
+	u64 mailbox_in_dma = 0;
+	u32 inmod = 0;
+	int i, err;
+
+	if (!(priv->mdev->dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_QCN))
+		return -EOPNOTSUPP;
+
+	mailbox_out = mlx4_alloc_cmd_mailbox(priv->mdev->dev);
+	if (IS_ERR(mailbox_out))
+		return -ENOMEM;
+
+	hw_qcn_stats =
+	(struct mlx4_congestion_control_mb_prio_802_1_qau_statistics *)
+	mailbox_out->buf;
+
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) {
+		inmod = priv->port | ((1 << i) << 8) |
+			 (MLX4_CTRL_ALGO_802_1_QAU_REACTION_POINT << 16);
+		err = mlx4_cmd_box(priv->mdev->dev, mailbox_in_dma,
+				   mailbox_out->dma, inmod,
+				   MLX4_CONGESTION_CONTROL_GET_STATISTICS,
+				   MLX4_CMD_CONGESTION_CTRL_OPCODE,
+				   MLX4_CMD_TIME_CLASS_C,
+				   MLX4_CMD_NATIVE);
+		if (err) {
+			mlx4_free_cmd_mailbox(priv->mdev->dev, mailbox_out);
+			return err;
+		}
+		qcn_stats->rppp_rp_centiseconds[i] =
+			be64_to_cpu(hw_qcn_stats->rppp_rp_centiseconds);
+		qcn_stats->rppp_created_rps[i] =
+			be32_to_cpu(hw_qcn_stats->rppp_created_rps);
+	}
+	mlx4_free_cmd_mailbox(priv->mdev->dev, mailbox_out);
+	return 0;
+}
+
+const struct dcbnl_rtnl_ops mlx4_en_dcbnl_ops = {
+	.ieee_getets		= mlx4_en_dcbnl_ieee_getets,
+	.ieee_setets		= mlx4_en_dcbnl_ieee_setets,
+	.ieee_getmaxrate	= mlx4_en_dcbnl_ieee_getmaxrate,
+	.ieee_setmaxrate	= mlx4_en_dcbnl_ieee_setmaxrate,
+	.ieee_getqcn		= mlx4_en_dcbnl_ieee_getqcn,
+	.ieee_setqcn		= mlx4_en_dcbnl_ieee_setqcn,
+	.ieee_getqcnstats	= mlx4_en_dcbnl_ieee_getqcnstats,
+	.ieee_getpfc		= mlx4_en_dcbnl_ieee_getpfc,
+	.ieee_setpfc		= mlx4_en_dcbnl_ieee_setpfc,
+
+	.getstate	= mlx4_en_dcbnl_get_state,
+	.setstate	= mlx4_en_dcbnl_set_state,
+	.getpfccfg	= mlx4_en_dcbnl_get_pfc_cfg,
+	.setpfccfg	= mlx4_en_dcbnl_set_pfc_cfg,
+	.setall		= mlx4_en_dcbnl_set_all,
+	.getcap		= mlx4_en_dcbnl_getcap,
+	.getnumtcs	= mlx4_en_dcbnl_getnumtcs,
+	.getpfcstate	= mlx4_en_dcbnl_getpfcstate,
+	.setpfcstate	= mlx4_en_dcbnl_setpfcstate,
+	.getapp		= mlx4_en_dcbnl_getapp,
+	.setapp		= mlx4_en_dcbnl_setapp,
+
+	.getdcbx	= mlx4_en_dcbnl_getdcbx,
+	.setdcbx	= mlx4_en_dcbnl_setdcbx,
+};
+
+const struct dcbnl_rtnl_ops mlx4_en_dcbnl_pfc_ops = {
+	.ieee_getpfc	= mlx4_en_dcbnl_ieee_getpfc,
+	.ieee_setpfc	= mlx4_en_dcbnl_ieee_setpfc,
+
+	.setstate	= mlx4_en_dcbnl_set_state,
+	.getpfccfg	= mlx4_en_dcbnl_get_pfc_cfg,
+	.setpfccfg	= mlx4_en_dcbnl_set_pfc_cfg,
+	.setall		= mlx4_en_dcbnl_set_all,
+	.getnumtcs	= mlx4_en_dcbnl_getnumtcs,
+	.getpfcstate	= mlx4_en_dcbnl_getpfcstate,
+	.setpfcstate	= mlx4_en_dcbnl_setpfcstate,
+	.getapp		= mlx4_en_dcbnl_getapp,
+	.setapp		= mlx4_en_dcbnl_setapp,
+
+	.getdcbx	= mlx4_en_dcbnl_getdcbx,
+	.setdcbx	= mlx4_en_dcbnl_setdcbx,
+};
diff --git a/drivers/net/ethernet/mellanox/mlx4/en_ethtool.c b/drivers/net/ethernet/mellanox/mlx4/en_ethtool.c
new file mode 100644
index 000000000..7d45f1d55
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx4/en_ethtool.c
@@ -0,0 +1,2193 @@
+/*
+ * Copyright (c) 2007 Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+
+#include <linux/kernel.h>
+#include <linux/ethtool.h>
+#include <linux/netdevice.h>
+#include <linux/mlx4/driver.h>
+#include <linux/mlx4/device.h>
+#include <linux/in.h>
+#include <net/ip.h>
+#include <linux/bitmap.h>
+#include <linux/mii.h>
+
+#include "mlx4_en.h"
+#include "en_port.h"
+
+#define EN_ETHTOOL_QP_ATTACH (1ull << 63)
+#define EN_ETHTOOL_SHORT_MASK cpu_to_be16(0xffff)
+#define EN_ETHTOOL_WORD_MASK  cpu_to_be32(0xffffffff)
+
+int mlx4_en_moderation_update(struct mlx4_en_priv *priv)
+{
+	int i, t;
+	int err = 0;
+
+	for (t = 0 ; t < MLX4_EN_NUM_TX_TYPES; t++) {
+		for (i = 0; i < priv->tx_ring_num[t]; i++) {
+			priv->tx_cq[t][i]->moder_cnt = priv->tx_frames;
+			priv->tx_cq[t][i]->moder_time = priv->tx_usecs;
+			if (priv->port_up) {
+				err = mlx4_en_set_cq_moder(priv,
+							   priv->tx_cq[t][i]);
+				if (err)
+					return err;
+			}
+		}
+	}
+
+	if (priv->adaptive_rx_coal)
+		return 0;
+
+	for (i = 0; i < priv->rx_ring_num; i++) {
+		priv->rx_cq[i]->moder_cnt = priv->rx_frames;
+		priv->rx_cq[i]->moder_time = priv->rx_usecs;
+		priv->last_moder_time[i] = MLX4_EN_AUTO_CONF;
+		if (priv->port_up) {
+			err = mlx4_en_set_cq_moder(priv, priv->rx_cq[i]);
+			if (err)
+				return err;
+		}
+	}
+
+	return err;
+}
+
+static void
+mlx4_en_get_drvinfo(struct net_device *dev, struct ethtool_drvinfo *drvinfo)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	struct mlx4_en_dev *mdev = priv->mdev;
+
+	strscpy(drvinfo->driver, DRV_NAME, sizeof(drvinfo->driver));
+	strscpy(drvinfo->version, DRV_VERSION,
+		sizeof(drvinfo->version));
+	snprintf(drvinfo->fw_version, sizeof(drvinfo->fw_version),
+		"%d.%d.%d",
+		(u16) (mdev->dev->caps.fw_ver >> 32),
+		(u16) ((mdev->dev->caps.fw_ver >> 16) & 0xffff),
+		(u16) (mdev->dev->caps.fw_ver & 0xffff));
+	strscpy(drvinfo->bus_info, pci_name(mdev->dev->persist->pdev),
+		sizeof(drvinfo->bus_info));
+}
+
+static const char mlx4_en_priv_flags[][ETH_GSTRING_LEN] = {
+	"blueflame",
+	"phv-bit"
+};
+
+static const char main_strings[][ETH_GSTRING_LEN] = {
+	/* main statistics */
+	"rx_packets", "tx_packets", "rx_bytes", "tx_bytes", "rx_errors",
+	"tx_errors", "rx_dropped", "tx_dropped", "multicast", "collisions",
+	"rx_length_errors", "rx_over_errors", "rx_crc_errors",
+	"rx_frame_errors", "rx_fifo_errors", "rx_missed_errors",
+	"tx_aborted_errors", "tx_carrier_errors", "tx_fifo_errors",
+	"tx_heartbeat_errors", "tx_window_errors",
+
+	/* port statistics */
+	"tso_packets",
+	"xmit_more",
+	"queue_stopped", "wake_queue", "tx_timeout", "rx_alloc_pages",
+	"rx_csum_good", "rx_csum_none", "rx_csum_complete", "tx_chksum_offload",
+
+	/* pf statistics */
+	"pf_rx_packets",
+	"pf_rx_bytes",
+	"pf_tx_packets",
+	"pf_tx_bytes",
+
+	/* priority flow control statistics rx */
+	"rx_pause_prio_0", "rx_pause_duration_prio_0",
+	"rx_pause_transition_prio_0",
+	"rx_pause_prio_1", "rx_pause_duration_prio_1",
+	"rx_pause_transition_prio_1",
+	"rx_pause_prio_2", "rx_pause_duration_prio_2",
+	"rx_pause_transition_prio_2",
+	"rx_pause_prio_3", "rx_pause_duration_prio_3",
+	"rx_pause_transition_prio_3",
+	"rx_pause_prio_4", "rx_pause_duration_prio_4",
+	"rx_pause_transition_prio_4",
+	"rx_pause_prio_5", "rx_pause_duration_prio_5",
+	"rx_pause_transition_prio_5",
+	"rx_pause_prio_6", "rx_pause_duration_prio_6",
+	"rx_pause_transition_prio_6",
+	"rx_pause_prio_7", "rx_pause_duration_prio_7",
+	"rx_pause_transition_prio_7",
+
+	/* flow control statistics rx */
+	"rx_pause", "rx_pause_duration", "rx_pause_transition",
+
+	/* priority flow control statistics tx */
+	"tx_pause_prio_0", "tx_pause_duration_prio_0",
+	"tx_pause_transition_prio_0",
+	"tx_pause_prio_1", "tx_pause_duration_prio_1",
+	"tx_pause_transition_prio_1",
+	"tx_pause_prio_2", "tx_pause_duration_prio_2",
+	"tx_pause_transition_prio_2",
+	"tx_pause_prio_3", "tx_pause_duration_prio_3",
+	"tx_pause_transition_prio_3",
+	"tx_pause_prio_4", "tx_pause_duration_prio_4",
+	"tx_pause_transition_prio_4",
+	"tx_pause_prio_5", "tx_pause_duration_prio_5",
+	"tx_pause_transition_prio_5",
+	"tx_pause_prio_6", "tx_pause_duration_prio_6",
+	"tx_pause_transition_prio_6",
+	"tx_pause_prio_7", "tx_pause_duration_prio_7",
+	"tx_pause_transition_prio_7",
+
+	/* flow control statistics tx */
+	"tx_pause", "tx_pause_duration", "tx_pause_transition",
+
+	/* packet statistics */
+	"rx_multicast_packets",
+	"rx_broadcast_packets",
+	"rx_jabbers",
+	"rx_in_range_length_error",
+	"rx_out_range_length_error",
+	"tx_multicast_packets",
+	"tx_broadcast_packets",
+	"rx_prio_0_packets", "rx_prio_0_bytes",
+	"rx_prio_1_packets", "rx_prio_1_bytes",
+	"rx_prio_2_packets", "rx_prio_2_bytes",
+	"rx_prio_3_packets", "rx_prio_3_bytes",
+	"rx_prio_4_packets", "rx_prio_4_bytes",
+	"rx_prio_5_packets", "rx_prio_5_bytes",
+	"rx_prio_6_packets", "rx_prio_6_bytes",
+	"rx_prio_7_packets", "rx_prio_7_bytes",
+	"rx_novlan_packets", "rx_novlan_bytes",
+	"tx_prio_0_packets", "tx_prio_0_bytes",
+	"tx_prio_1_packets", "tx_prio_1_bytes",
+	"tx_prio_2_packets", "tx_prio_2_bytes",
+	"tx_prio_3_packets", "tx_prio_3_bytes",
+	"tx_prio_4_packets", "tx_prio_4_bytes",
+	"tx_prio_5_packets", "tx_prio_5_bytes",
+	"tx_prio_6_packets", "tx_prio_6_bytes",
+	"tx_prio_7_packets", "tx_prio_7_bytes",
+	"tx_novlan_packets", "tx_novlan_bytes",
+
+	/* xdp statistics */
+	"rx_xdp_drop",
+	"rx_xdp_redirect",
+	"rx_xdp_redirect_fail",
+	"rx_xdp_tx",
+	"rx_xdp_tx_full",
+
+	/* phy statistics */
+	"rx_packets_phy", "rx_bytes_phy",
+	"tx_packets_phy", "tx_bytes_phy",
+};
+
+static const char mlx4_en_test_names[][ETH_GSTRING_LEN]= {
+	"Interrupt Test",
+	"Link Test",
+	"Speed Test",
+	"Register Test",
+	"Loopback Test",
+};
+
+static u32 mlx4_en_get_msglevel(struct net_device *dev)
+{
+	return ((struct mlx4_en_priv *) netdev_priv(dev))->msg_enable;
+}
+
+static void mlx4_en_set_msglevel(struct net_device *dev, u32 val)
+{
+	((struct mlx4_en_priv *) netdev_priv(dev))->msg_enable = val;
+}
+
+static void mlx4_en_get_wol(struct net_device *netdev,
+			    struct ethtool_wolinfo *wol)
+{
+	struct mlx4_en_priv *priv = netdev_priv(netdev);
+	struct mlx4_caps *caps = &priv->mdev->dev->caps;
+	int err = 0;
+	u64 config = 0;
+	u64 mask;
+
+	if ((priv->port < 1) || (priv->port > 2)) {
+		en_err(priv, "Failed to get WoL information\n");
+		return;
+	}
+
+	mask = (priv->port == 1) ? MLX4_DEV_CAP_FLAG_WOL_PORT1 :
+		MLX4_DEV_CAP_FLAG_WOL_PORT2;
+
+	if (!(caps->flags & mask)) {
+		wol->supported = 0;
+		wol->wolopts = 0;
+		return;
+	}
+
+	if (caps->wol_port[priv->port])
+		wol->supported = WAKE_MAGIC;
+	else
+		wol->supported = 0;
+
+	err = mlx4_wol_read(priv->mdev->dev, &config, priv->port);
+	if (err) {
+		en_err(priv, "Failed to get WoL information\n");
+		return;
+	}
+
+	if ((config & MLX4_EN_WOL_ENABLED) && (config & MLX4_EN_WOL_MAGIC))
+		wol->wolopts = WAKE_MAGIC;
+	else
+		wol->wolopts = 0;
+}
+
+static int mlx4_en_set_wol(struct net_device *netdev,
+			    struct ethtool_wolinfo *wol)
+{
+	struct mlx4_en_priv *priv = netdev_priv(netdev);
+	u64 config = 0;
+	int err = 0;
+	u64 mask;
+
+	if ((priv->port < 1) || (priv->port > 2))
+		return -EOPNOTSUPP;
+
+	mask = (priv->port == 1) ? MLX4_DEV_CAP_FLAG_WOL_PORT1 :
+		MLX4_DEV_CAP_FLAG_WOL_PORT2;
+
+	if (!(priv->mdev->dev->caps.flags & mask))
+		return -EOPNOTSUPP;
+
+	if (wol->supported & ~WAKE_MAGIC)
+		return -EINVAL;
+
+	err = mlx4_wol_read(priv->mdev->dev, &config, priv->port);
+	if (err) {
+		en_err(priv, "Failed to get WoL info, unable to modify\n");
+		return err;
+	}
+
+	if (wol->wolopts & WAKE_MAGIC) {
+		config |= MLX4_EN_WOL_DO_MODIFY | MLX4_EN_WOL_ENABLED |
+				MLX4_EN_WOL_MAGIC;
+	} else {
+		config &= ~(MLX4_EN_WOL_ENABLED | MLX4_EN_WOL_MAGIC);
+		config |= MLX4_EN_WOL_DO_MODIFY;
+	}
+
+	err = mlx4_wol_write(priv->mdev->dev, config, priv->port);
+	if (err)
+		en_err(priv, "Failed to set WoL information\n");
+
+	return err;
+}
+
+struct bitmap_iterator {
+	unsigned long *stats_bitmap;
+	unsigned int count;
+	unsigned int iterator;
+	bool advance_array; /* if set, force no increments */
+};
+
+static inline void bitmap_iterator_init(struct bitmap_iterator *h,
+					unsigned long *stats_bitmap,
+					int count)
+{
+	h->iterator = 0;
+	h->advance_array = !bitmap_empty(stats_bitmap, count);
+	h->count = h->advance_array ? bitmap_weight(stats_bitmap, count)
+		: count;
+	h->stats_bitmap = stats_bitmap;
+}
+
+static inline int bitmap_iterator_test(struct bitmap_iterator *h)
+{
+	return !h->advance_array ? 1 : test_bit(h->iterator, h->stats_bitmap);
+}
+
+static inline int bitmap_iterator_inc(struct bitmap_iterator *h)
+{
+	return h->iterator++;
+}
+
+static inline unsigned int
+bitmap_iterator_count(struct bitmap_iterator *h)
+{
+	return h->count;
+}
+
+static int mlx4_en_get_sset_count(struct net_device *dev, int sset)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	struct bitmap_iterator it;
+
+	bitmap_iterator_init(&it, priv->stats_bitmap.bitmap, NUM_ALL_STATS);
+
+	switch (sset) {
+	case ETH_SS_STATS:
+		return bitmap_iterator_count(&it) +
+			(priv->tx_ring_num[TX] * 2) +
+			(priv->rx_ring_num * (3 + NUM_XDP_STATS));
+	case ETH_SS_TEST:
+		return MLX4_EN_NUM_SELF_TEST - !(priv->mdev->dev->caps.flags
+					& MLX4_DEV_CAP_FLAG_UC_LOOPBACK) * 2;
+	case ETH_SS_PRIV_FLAGS:
+		return ARRAY_SIZE(mlx4_en_priv_flags);
+	default:
+		return -EOPNOTSUPP;
+	}
+}
+
+static void mlx4_en_get_ethtool_stats(struct net_device *dev,
+		struct ethtool_stats *stats, uint64_t *data)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	int index = 0;
+	int i;
+	struct bitmap_iterator it;
+
+	bitmap_iterator_init(&it, priv->stats_bitmap.bitmap, NUM_ALL_STATS);
+
+	spin_lock_bh(&priv->stats_lock);
+
+	mlx4_en_fold_software_stats(dev);
+
+	for (i = 0; i < NUM_MAIN_STATS; i++, bitmap_iterator_inc(&it))
+		if (bitmap_iterator_test(&it))
+			data[index++] = ((unsigned long *)&dev->stats)[i];
+
+	for (i = 0; i < NUM_PORT_STATS; i++, bitmap_iterator_inc(&it))
+		if (bitmap_iterator_test(&it))
+			data[index++] = ((unsigned long *)&priv->port_stats)[i];
+
+	for (i = 0; i < NUM_PF_STATS; i++, bitmap_iterator_inc(&it))
+		if (bitmap_iterator_test(&it))
+			data[index++] =
+				((unsigned long *)&priv->pf_stats)[i];
+
+	for (i = 0; i < NUM_FLOW_PRIORITY_STATS_RX;
+	     i++, bitmap_iterator_inc(&it))
+		if (bitmap_iterator_test(&it))
+			data[index++] =
+				((u64 *)&priv->rx_priority_flowstats)[i];
+
+	for (i = 0; i < NUM_FLOW_STATS_RX; i++, bitmap_iterator_inc(&it))
+		if (bitmap_iterator_test(&it))
+			data[index++] = ((u64 *)&priv->rx_flowstats)[i];
+
+	for (i = 0; i < NUM_FLOW_PRIORITY_STATS_TX;
+	     i++, bitmap_iterator_inc(&it))
+		if (bitmap_iterator_test(&it))
+			data[index++] =
+				((u64 *)&priv->tx_priority_flowstats)[i];
+
+	for (i = 0; i < NUM_FLOW_STATS_TX; i++, bitmap_iterator_inc(&it))
+		if (bitmap_iterator_test(&it))
+			data[index++] = ((u64 *)&priv->tx_flowstats)[i];
+
+	for (i = 0; i < NUM_PKT_STATS; i++, bitmap_iterator_inc(&it))
+		if (bitmap_iterator_test(&it))
+			data[index++] = ((unsigned long *)&priv->pkstats)[i];
+
+	for (i = 0; i < NUM_XDP_STATS; i++, bitmap_iterator_inc(&it))
+		if (bitmap_iterator_test(&it))
+			data[index++] = ((unsigned long *)&priv->xdp_stats)[i];
+
+	for (i = 0; i < NUM_PHY_STATS; i++, bitmap_iterator_inc(&it))
+		if (bitmap_iterator_test(&it))
+			data[index++] = ((unsigned long *)&priv->phy_stats)[i];
+
+	for (i = 0; i < priv->tx_ring_num[TX]; i++) {
+		data[index++] = priv->tx_ring[TX][i]->packets;
+		data[index++] = priv->tx_ring[TX][i]->bytes;
+	}
+	for (i = 0; i < priv->rx_ring_num; i++) {
+		data[index++] = priv->rx_ring[i]->packets;
+		data[index++] = priv->rx_ring[i]->bytes;
+		data[index++] = priv->rx_ring[i]->dropped;
+		data[index++] = priv->rx_ring[i]->xdp_drop;
+		data[index++] = priv->rx_ring[i]->xdp_redirect;
+		data[index++] = priv->rx_ring[i]->xdp_redirect_fail;
+		data[index++] = priv->rx_ring[i]->xdp_tx;
+		data[index++] = priv->rx_ring[i]->xdp_tx_full;
+	}
+	spin_unlock_bh(&priv->stats_lock);
+
+}
+
+static void mlx4_en_self_test(struct net_device *dev,
+			      struct ethtool_test *etest, u64 *buf)
+{
+	mlx4_en_ex_selftest(dev, &etest->flags, buf);
+}
+
+static void mlx4_en_get_strings(struct net_device *dev,
+				uint32_t stringset, uint8_t *data)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	int index = 0;
+	int i, strings = 0;
+	struct bitmap_iterator it;
+
+	bitmap_iterator_init(&it, priv->stats_bitmap.bitmap, NUM_ALL_STATS);
+
+	switch (stringset) {
+	case ETH_SS_TEST:
+		for (i = 0; i < MLX4_EN_NUM_SELF_TEST - 2; i++)
+			strcpy(data + i * ETH_GSTRING_LEN, mlx4_en_test_names[i]);
+		if (priv->mdev->dev->caps.flags & MLX4_DEV_CAP_FLAG_UC_LOOPBACK)
+			for (; i < MLX4_EN_NUM_SELF_TEST; i++)
+				strcpy(data + i * ETH_GSTRING_LEN, mlx4_en_test_names[i]);
+		break;
+
+	case ETH_SS_STATS:
+		/* Add main counters */
+		for (i = 0; i < NUM_MAIN_STATS; i++, strings++,
+		     bitmap_iterator_inc(&it))
+			if (bitmap_iterator_test(&it))
+				strcpy(data + (index++) * ETH_GSTRING_LEN,
+				       main_strings[strings]);
+
+		for (i = 0; i < NUM_PORT_STATS; i++, strings++,
+		     bitmap_iterator_inc(&it))
+			if (bitmap_iterator_test(&it))
+				strcpy(data + (index++) * ETH_GSTRING_LEN,
+				       main_strings[strings]);
+
+		for (i = 0; i < NUM_PF_STATS; i++, strings++,
+		     bitmap_iterator_inc(&it))
+			if (bitmap_iterator_test(&it))
+				strcpy(data + (index++) * ETH_GSTRING_LEN,
+				       main_strings[strings]);
+
+		for (i = 0; i < NUM_FLOW_STATS; i++, strings++,
+		     bitmap_iterator_inc(&it))
+			if (bitmap_iterator_test(&it))
+				strcpy(data + (index++) * ETH_GSTRING_LEN,
+				       main_strings[strings]);
+
+		for (i = 0; i < NUM_PKT_STATS; i++, strings++,
+		     bitmap_iterator_inc(&it))
+			if (bitmap_iterator_test(&it))
+				strcpy(data + (index++) * ETH_GSTRING_LEN,
+				       main_strings[strings]);
+
+		for (i = 0; i < NUM_XDP_STATS; i++, strings++,
+		     bitmap_iterator_inc(&it))
+			if (bitmap_iterator_test(&it))
+				strcpy(data + (index++) * ETH_GSTRING_LEN,
+				       main_strings[strings]);
+
+		for (i = 0; i < NUM_PHY_STATS; i++, strings++,
+		     bitmap_iterator_inc(&it))
+			if (bitmap_iterator_test(&it))
+				strcpy(data + (index++) * ETH_GSTRING_LEN,
+				       main_strings[strings]);
+
+		for (i = 0; i < priv->tx_ring_num[TX]; i++) {
+			sprintf(data + (index++) * ETH_GSTRING_LEN,
+				"tx%d_packets", i);
+			sprintf(data + (index++) * ETH_GSTRING_LEN,
+				"tx%d_bytes", i);
+		}
+		for (i = 0; i < priv->rx_ring_num; i++) {
+			sprintf(data + (index++) * ETH_GSTRING_LEN,
+				"rx%d_packets", i);
+			sprintf(data + (index++) * ETH_GSTRING_LEN,
+				"rx%d_bytes", i);
+			sprintf(data + (index++) * ETH_GSTRING_LEN,
+				"rx%d_dropped", i);
+			sprintf(data + (index++) * ETH_GSTRING_LEN,
+				"rx%d_xdp_drop", i);
+			sprintf(data + (index++) * ETH_GSTRING_LEN,
+				"rx%d_xdp_redirect", i);
+			sprintf(data + (index++) * ETH_GSTRING_LEN,
+				"rx%d_xdp_redirect_fail", i);
+			sprintf(data + (index++) * ETH_GSTRING_LEN,
+				"rx%d_xdp_tx", i);
+			sprintf(data + (index++) * ETH_GSTRING_LEN,
+				"rx%d_xdp_tx_full", i);
+		}
+		break;
+	case ETH_SS_PRIV_FLAGS:
+		for (i = 0; i < ARRAY_SIZE(mlx4_en_priv_flags); i++)
+			strcpy(data + i * ETH_GSTRING_LEN,
+			       mlx4_en_priv_flags[i]);
+		break;
+
+	}
+}
+
+static u32 mlx4_en_autoneg_get(struct net_device *dev)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	struct mlx4_en_dev *mdev = priv->mdev;
+	u32 autoneg = AUTONEG_DISABLE;
+
+	if ((mdev->dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_ETH_BACKPL_AN_REP) &&
+	    (priv->port_state.flags & MLX4_EN_PORT_ANE))
+		autoneg = AUTONEG_ENABLE;
+
+	return autoneg;
+}
+
+static void ptys2ethtool_update_supported_port(unsigned long *mask,
+					       struct mlx4_ptys_reg *ptys_reg)
+{
+	u32 eth_proto = be32_to_cpu(ptys_reg->eth_proto_cap);
+
+	if (eth_proto & (MLX4_PROT_MASK(MLX4_10GBASE_T)
+			 | MLX4_PROT_MASK(MLX4_1000BASE_T)
+			 | MLX4_PROT_MASK(MLX4_100BASE_TX))) {
+		__set_bit(ETHTOOL_LINK_MODE_TP_BIT, mask);
+	} else if (eth_proto & (MLX4_PROT_MASK(MLX4_10GBASE_CR)
+			 | MLX4_PROT_MASK(MLX4_10GBASE_SR)
+			 | MLX4_PROT_MASK(MLX4_56GBASE_SR4)
+			 | MLX4_PROT_MASK(MLX4_40GBASE_CR4)
+			 | MLX4_PROT_MASK(MLX4_40GBASE_SR4)
+			 | MLX4_PROT_MASK(MLX4_1000BASE_CX_SGMII))) {
+		__set_bit(ETHTOOL_LINK_MODE_FIBRE_BIT, mask);
+	} else if (eth_proto & (MLX4_PROT_MASK(MLX4_56GBASE_KR4)
+			 | MLX4_PROT_MASK(MLX4_40GBASE_KR4)
+			 | MLX4_PROT_MASK(MLX4_20GBASE_KR2)
+			 | MLX4_PROT_MASK(MLX4_10GBASE_KR)
+			 | MLX4_PROT_MASK(MLX4_10GBASE_KX4)
+			 | MLX4_PROT_MASK(MLX4_1000BASE_KX))) {
+		__set_bit(ETHTOOL_LINK_MODE_Backplane_BIT, mask);
+	}
+}
+
+static u32 ptys_get_active_port(struct mlx4_ptys_reg *ptys_reg)
+{
+	u32 eth_proto = be32_to_cpu(ptys_reg->eth_proto_oper);
+
+	if (!eth_proto) /* link down */
+		eth_proto = be32_to_cpu(ptys_reg->eth_proto_cap);
+
+	if (eth_proto & (MLX4_PROT_MASK(MLX4_10GBASE_T)
+			 | MLX4_PROT_MASK(MLX4_1000BASE_T)
+			 | MLX4_PROT_MASK(MLX4_100BASE_TX))) {
+			return PORT_TP;
+	}
+
+	if (eth_proto & (MLX4_PROT_MASK(MLX4_10GBASE_SR)
+			 | MLX4_PROT_MASK(MLX4_56GBASE_SR4)
+			 | MLX4_PROT_MASK(MLX4_40GBASE_SR4)
+			 | MLX4_PROT_MASK(MLX4_1000BASE_CX_SGMII))) {
+			return PORT_FIBRE;
+	}
+
+	if (eth_proto & (MLX4_PROT_MASK(MLX4_10GBASE_CR)
+			 | MLX4_PROT_MASK(MLX4_56GBASE_CR4)
+			 | MLX4_PROT_MASK(MLX4_40GBASE_CR4))) {
+			return PORT_DA;
+	}
+
+	if (eth_proto & (MLX4_PROT_MASK(MLX4_56GBASE_KR4)
+			 | MLX4_PROT_MASK(MLX4_40GBASE_KR4)
+			 | MLX4_PROT_MASK(MLX4_20GBASE_KR2)
+			 | MLX4_PROT_MASK(MLX4_10GBASE_KR)
+			 | MLX4_PROT_MASK(MLX4_10GBASE_KX4)
+			 | MLX4_PROT_MASK(MLX4_1000BASE_KX))) {
+			return PORT_NONE;
+	}
+	return PORT_OTHER;
+}
+
+#define MLX4_LINK_MODES_SZ \
+	(sizeof_field(struct mlx4_ptys_reg, eth_proto_cap) * 8)
+
+enum ethtool_report {
+	SUPPORTED = 0,
+	ADVERTISED = 1,
+};
+
+struct ptys2ethtool_config {
+	__ETHTOOL_DECLARE_LINK_MODE_MASK(supported);
+	__ETHTOOL_DECLARE_LINK_MODE_MASK(advertised);
+	u32 speed;
+};
+
+static unsigned long *ptys2ethtool_link_mode(struct ptys2ethtool_config *cfg,
+					     enum ethtool_report report)
+{
+	switch (report) {
+	case SUPPORTED:
+		return cfg->supported;
+	case ADVERTISED:
+		return cfg->advertised;
+	}
+	return NULL;
+}
+
+#define MLX4_BUILD_PTYS2ETHTOOL_CONFIG(reg_, speed_, ...)		\
+	({								\
+		struct ptys2ethtool_config *cfg;			\
+		static const unsigned int modes[] = { __VA_ARGS__ };	\
+		unsigned int i;						\
+		cfg = &ptys2ethtool_map[reg_];				\
+		cfg->speed = speed_;					\
+		linkmode_zero(cfg->supported);				\
+		linkmode_zero(cfg->advertised);				\
+		for (i = 0 ; i < ARRAY_SIZE(modes) ; ++i) {		\
+			__set_bit(modes[i], cfg->supported);		\
+			__set_bit(modes[i], cfg->advertised);		\
+		}							\
+	})
+
+/* Translates mlx4 link mode to equivalent ethtool Link modes/speed */
+static struct ptys2ethtool_config ptys2ethtool_map[MLX4_LINK_MODES_SZ];
+
+void __init mlx4_en_init_ptys2ethtool_map(void)
+{
+	MLX4_BUILD_PTYS2ETHTOOL_CONFIG(MLX4_100BASE_TX, SPEED_100,
+				       ETHTOOL_LINK_MODE_100baseT_Full_BIT);
+	MLX4_BUILD_PTYS2ETHTOOL_CONFIG(MLX4_1000BASE_T, SPEED_1000,
+				       ETHTOOL_LINK_MODE_1000baseT_Full_BIT);
+	MLX4_BUILD_PTYS2ETHTOOL_CONFIG(MLX4_1000BASE_CX_SGMII, SPEED_1000,
+				       ETHTOOL_LINK_MODE_1000baseX_Full_BIT);
+	MLX4_BUILD_PTYS2ETHTOOL_CONFIG(MLX4_1000BASE_KX, SPEED_1000,
+				       ETHTOOL_LINK_MODE_1000baseKX_Full_BIT);
+	MLX4_BUILD_PTYS2ETHTOOL_CONFIG(MLX4_10GBASE_T, SPEED_10000,
+				       ETHTOOL_LINK_MODE_10000baseT_Full_BIT);
+	MLX4_BUILD_PTYS2ETHTOOL_CONFIG(MLX4_10GBASE_CX4, SPEED_10000,
+				       ETHTOOL_LINK_MODE_10000baseKX4_Full_BIT);
+	MLX4_BUILD_PTYS2ETHTOOL_CONFIG(MLX4_10GBASE_KX4, SPEED_10000,
+				       ETHTOOL_LINK_MODE_10000baseKX4_Full_BIT);
+	MLX4_BUILD_PTYS2ETHTOOL_CONFIG(MLX4_10GBASE_KR, SPEED_10000,
+				       ETHTOOL_LINK_MODE_10000baseKR_Full_BIT);
+	MLX4_BUILD_PTYS2ETHTOOL_CONFIG(MLX4_10GBASE_CR, SPEED_10000,
+				       ETHTOOL_LINK_MODE_10000baseCR_Full_BIT);
+	MLX4_BUILD_PTYS2ETHTOOL_CONFIG(MLX4_10GBASE_SR, SPEED_10000,
+				       ETHTOOL_LINK_MODE_10000baseSR_Full_BIT);
+	MLX4_BUILD_PTYS2ETHTOOL_CONFIG(MLX4_20GBASE_KR2, SPEED_20000,
+				       ETHTOOL_LINK_MODE_20000baseMLD2_Full_BIT,
+				       ETHTOOL_LINK_MODE_20000baseKR2_Full_BIT);
+	MLX4_BUILD_PTYS2ETHTOOL_CONFIG(MLX4_40GBASE_CR4, SPEED_40000,
+				       ETHTOOL_LINK_MODE_40000baseCR4_Full_BIT);
+	MLX4_BUILD_PTYS2ETHTOOL_CONFIG(MLX4_40GBASE_KR4, SPEED_40000,
+				       ETHTOOL_LINK_MODE_40000baseKR4_Full_BIT);
+	MLX4_BUILD_PTYS2ETHTOOL_CONFIG(MLX4_40GBASE_SR4, SPEED_40000,
+				       ETHTOOL_LINK_MODE_40000baseSR4_Full_BIT);
+	MLX4_BUILD_PTYS2ETHTOOL_CONFIG(MLX4_56GBASE_KR4, SPEED_56000,
+				       ETHTOOL_LINK_MODE_56000baseKR4_Full_BIT);
+	MLX4_BUILD_PTYS2ETHTOOL_CONFIG(MLX4_56GBASE_CR4, SPEED_56000,
+				       ETHTOOL_LINK_MODE_56000baseCR4_Full_BIT);
+	MLX4_BUILD_PTYS2ETHTOOL_CONFIG(MLX4_56GBASE_SR4, SPEED_56000,
+				       ETHTOOL_LINK_MODE_56000baseSR4_Full_BIT);
+};
+
+static void ptys2ethtool_update_link_modes(unsigned long *link_modes,
+					   u32 eth_proto,
+					   enum ethtool_report report)
+{
+	int i;
+	for (i = 0; i < MLX4_LINK_MODES_SZ; i++) {
+		if (eth_proto & MLX4_PROT_MASK(i))
+			linkmode_or(link_modes, link_modes,
+				    ptys2ethtool_link_mode(&ptys2ethtool_map[i], report));
+	}
+}
+
+static u32 ethtool2ptys_link_modes(const unsigned long *link_modes,
+				   enum ethtool_report report)
+{
+	int i;
+	u32 ptys_modes = 0;
+
+	for (i = 0; i < MLX4_LINK_MODES_SZ; i++) {
+		ulong *map_mode = ptys2ethtool_link_mode(&ptys2ethtool_map[i],
+							 report);
+		if (linkmode_intersects(map_mode, link_modes))
+			ptys_modes |= 1 << i;
+	}
+	return ptys_modes;
+}
+
+/* Convert actual speed (SPEED_XXX) to ptys link modes */
+static u32 speed2ptys_link_modes(u32 speed)
+{
+	int i;
+	u32 ptys_modes = 0;
+
+	for (i = 0; i < MLX4_LINK_MODES_SZ; i++) {
+		if (ptys2ethtool_map[i].speed == speed)
+			ptys_modes |= 1 << i;
+	}
+	return ptys_modes;
+}
+
+static int
+ethtool_get_ptys_link_ksettings(struct net_device *dev,
+				struct ethtool_link_ksettings *link_ksettings)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	struct mlx4_ptys_reg ptys_reg;
+	u32 eth_proto;
+	int ret;
+
+	memset(&ptys_reg, 0, sizeof(ptys_reg));
+	ptys_reg.local_port = priv->port;
+	ptys_reg.proto_mask = MLX4_PTYS_EN;
+	ret = mlx4_ACCESS_PTYS_REG(priv->mdev->dev,
+				   MLX4_ACCESS_REG_QUERY, &ptys_reg);
+	if (ret) {
+		en_warn(priv, "Failed to run mlx4_ACCESS_PTYS_REG status(%x)",
+			ret);
+		return ret;
+	}
+	en_dbg(DRV, priv, "ptys_reg.proto_mask       %x\n",
+	       ptys_reg.proto_mask);
+	en_dbg(DRV, priv, "ptys_reg.eth_proto_cap    %x\n",
+	       be32_to_cpu(ptys_reg.eth_proto_cap));
+	en_dbg(DRV, priv, "ptys_reg.eth_proto_admin  %x\n",
+	       be32_to_cpu(ptys_reg.eth_proto_admin));
+	en_dbg(DRV, priv, "ptys_reg.eth_proto_oper   %x\n",
+	       be32_to_cpu(ptys_reg.eth_proto_oper));
+	en_dbg(DRV, priv, "ptys_reg.eth_proto_lp_adv %x\n",
+	       be32_to_cpu(ptys_reg.eth_proto_lp_adv));
+
+	/* reset supported/advertising masks */
+	ethtool_link_ksettings_zero_link_mode(link_ksettings, supported);
+	ethtool_link_ksettings_zero_link_mode(link_ksettings, advertising);
+
+	ptys2ethtool_update_supported_port(link_ksettings->link_modes.supported,
+					   &ptys_reg);
+
+	eth_proto = be32_to_cpu(ptys_reg.eth_proto_cap);
+	ptys2ethtool_update_link_modes(link_ksettings->link_modes.supported,
+				       eth_proto, SUPPORTED);
+
+	eth_proto = be32_to_cpu(ptys_reg.eth_proto_admin);
+	ptys2ethtool_update_link_modes(link_ksettings->link_modes.advertising,
+				       eth_proto, ADVERTISED);
+
+	ethtool_link_ksettings_add_link_mode(link_ksettings, supported,
+					     Pause);
+	ethtool_link_ksettings_add_link_mode(link_ksettings, supported,
+					     Asym_Pause);
+
+	if (priv->prof->tx_pause)
+		ethtool_link_ksettings_add_link_mode(link_ksettings,
+						     advertising, Pause);
+	if (priv->prof->tx_pause ^ priv->prof->rx_pause)
+		ethtool_link_ksettings_add_link_mode(link_ksettings,
+						     advertising, Asym_Pause);
+
+	link_ksettings->base.port = ptys_get_active_port(&ptys_reg);
+
+	if (mlx4_en_autoneg_get(dev)) {
+		ethtool_link_ksettings_add_link_mode(link_ksettings,
+						     supported, Autoneg);
+		ethtool_link_ksettings_add_link_mode(link_ksettings,
+						     advertising, Autoneg);
+	}
+
+	link_ksettings->base.autoneg
+		= (priv->port_state.flags & MLX4_EN_PORT_ANC) ?
+		AUTONEG_ENABLE : AUTONEG_DISABLE;
+
+	eth_proto = be32_to_cpu(ptys_reg.eth_proto_lp_adv);
+
+	ethtool_link_ksettings_zero_link_mode(link_ksettings, lp_advertising);
+	ptys2ethtool_update_link_modes(
+		link_ksettings->link_modes.lp_advertising,
+		eth_proto, ADVERTISED);
+	if (priv->port_state.flags & MLX4_EN_PORT_ANC)
+		ethtool_link_ksettings_add_link_mode(link_ksettings,
+						     lp_advertising, Autoneg);
+
+	link_ksettings->base.phy_address = 0;
+	link_ksettings->base.mdio_support = 0;
+	link_ksettings->base.eth_tp_mdix = ETH_TP_MDI_INVALID;
+	link_ksettings->base.eth_tp_mdix_ctrl = ETH_TP_MDI_AUTO;
+
+	return ret;
+}
+
+static void
+ethtool_get_default_link_ksettings(
+	struct net_device *dev, struct ethtool_link_ksettings *link_ksettings)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	int trans_type;
+
+	link_ksettings->base.autoneg = AUTONEG_DISABLE;
+
+	ethtool_link_ksettings_zero_link_mode(link_ksettings, supported);
+	ethtool_link_ksettings_add_link_mode(link_ksettings, supported,
+					     10000baseT_Full);
+
+	ethtool_link_ksettings_zero_link_mode(link_ksettings, advertising);
+	ethtool_link_ksettings_add_link_mode(link_ksettings, advertising,
+					     10000baseT_Full);
+
+	trans_type = priv->port_state.transceiver;
+	if (trans_type > 0 && trans_type <= 0xC) {
+		link_ksettings->base.port = PORT_FIBRE;
+		ethtool_link_ksettings_add_link_mode(link_ksettings,
+						     supported, FIBRE);
+		ethtool_link_ksettings_add_link_mode(link_ksettings,
+						     advertising, FIBRE);
+	} else if (trans_type == 0x80 || trans_type == 0) {
+		link_ksettings->base.port = PORT_TP;
+		ethtool_link_ksettings_add_link_mode(link_ksettings,
+						     supported, TP);
+		ethtool_link_ksettings_add_link_mode(link_ksettings,
+						     advertising, TP);
+	} else  {
+		link_ksettings->base.port = -1;
+	}
+}
+
+static int
+mlx4_en_get_link_ksettings(struct net_device *dev,
+			   struct ethtool_link_ksettings *link_ksettings)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	int ret = -EINVAL;
+
+	if (mlx4_en_QUERY_PORT(priv->mdev, priv->port))
+		return -ENOMEM;
+
+	en_dbg(DRV, priv, "query port state.flags ANC(%x) ANE(%x)\n",
+	       priv->port_state.flags & MLX4_EN_PORT_ANC,
+	       priv->port_state.flags & MLX4_EN_PORT_ANE);
+
+	if (priv->mdev->dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_ETH_PROT_CTRL)
+		ret = ethtool_get_ptys_link_ksettings(dev, link_ksettings);
+	if (ret) /* ETH PROT CRTL is not supported or PTYS CMD failed */
+		ethtool_get_default_link_ksettings(dev, link_ksettings);
+
+	if (netif_carrier_ok(dev)) {
+		link_ksettings->base.speed = priv->port_state.link_speed;
+		link_ksettings->base.duplex = DUPLEX_FULL;
+	} else {
+		link_ksettings->base.speed = SPEED_UNKNOWN;
+		link_ksettings->base.duplex = DUPLEX_UNKNOWN;
+	}
+	return 0;
+}
+
+/* Calculate PTYS admin according ethtool speed (SPEED_XXX) */
+static __be32 speed_set_ptys_admin(struct mlx4_en_priv *priv, u32 speed,
+				   __be32 proto_cap)
+{
+	__be32 proto_admin = 0;
+
+	if (!speed) { /* Speed = 0 ==> Reset Link modes */
+		proto_admin = proto_cap;
+		en_info(priv, "Speed was set to 0, Reset advertised Link Modes to default (%x)\n",
+			be32_to_cpu(proto_cap));
+	} else {
+		u32 ptys_link_modes = speed2ptys_link_modes(speed);
+
+		proto_admin = cpu_to_be32(ptys_link_modes) & proto_cap;
+		en_info(priv, "Setting Speed to %d\n", speed);
+	}
+	return proto_admin;
+}
+
+static int
+mlx4_en_set_link_ksettings(struct net_device *dev,
+			   const struct ethtool_link_ksettings *link_ksettings)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	struct mlx4_ptys_reg ptys_reg;
+	__be32 proto_admin;
+	u8 cur_autoneg;
+	int ret;
+
+	u32 ptys_adv = ethtool2ptys_link_modes(
+		link_ksettings->link_modes.advertising, ADVERTISED);
+	const int speed = link_ksettings->base.speed;
+
+	en_dbg(DRV, priv,
+	       "Set Speed=%d adv={%*pbl} autoneg=%d duplex=%d\n",
+	       speed, __ETHTOOL_LINK_MODE_MASK_NBITS,
+	       link_ksettings->link_modes.advertising,
+	       link_ksettings->base.autoneg,
+	       link_ksettings->base.duplex);
+
+	if (!(priv->mdev->dev->caps.flags2 &
+	      MLX4_DEV_CAP_FLAG2_ETH_PROT_CTRL) ||
+	    (link_ksettings->base.duplex == DUPLEX_HALF))
+		return -EINVAL;
+
+	memset(&ptys_reg, 0, sizeof(ptys_reg));
+	ptys_reg.local_port = priv->port;
+	ptys_reg.proto_mask = MLX4_PTYS_EN;
+	ret = mlx4_ACCESS_PTYS_REG(priv->mdev->dev,
+				   MLX4_ACCESS_REG_QUERY, &ptys_reg);
+	if (ret) {
+		en_warn(priv, "Failed to QUERY mlx4_ACCESS_PTYS_REG status(%x)\n",
+			ret);
+		return 0;
+	}
+
+	cur_autoneg = ptys_reg.flags & MLX4_PTYS_AN_DISABLE_ADMIN ?
+				AUTONEG_DISABLE : AUTONEG_ENABLE;
+
+	if (link_ksettings->base.autoneg == AUTONEG_DISABLE) {
+		proto_admin = speed_set_ptys_admin(priv, speed,
+						   ptys_reg.eth_proto_cap);
+		if ((be32_to_cpu(proto_admin) &
+		     (MLX4_PROT_MASK(MLX4_1000BASE_CX_SGMII) |
+		      MLX4_PROT_MASK(MLX4_1000BASE_KX))) &&
+		    (ptys_reg.flags & MLX4_PTYS_AN_DISABLE_CAP))
+			ptys_reg.flags |= MLX4_PTYS_AN_DISABLE_ADMIN;
+	} else {
+		proto_admin = cpu_to_be32(ptys_adv);
+		ptys_reg.flags &= ~MLX4_PTYS_AN_DISABLE_ADMIN;
+	}
+
+	proto_admin &= ptys_reg.eth_proto_cap;
+	if (!proto_admin) {
+		en_warn(priv, "Not supported link mode(s) requested, check supported link modes.\n");
+		return -EINVAL; /* nothing to change due to bad input */
+	}
+
+	if ((proto_admin == ptys_reg.eth_proto_admin) &&
+	    ((ptys_reg.flags & MLX4_PTYS_AN_DISABLE_CAP) &&
+	     (link_ksettings->base.autoneg == cur_autoneg)))
+		return 0; /* Nothing to change */
+
+	en_dbg(DRV, priv, "mlx4_ACCESS_PTYS_REG SET: ptys_reg.eth_proto_admin = 0x%x\n",
+	       be32_to_cpu(proto_admin));
+
+	ptys_reg.eth_proto_admin = proto_admin;
+	ret = mlx4_ACCESS_PTYS_REG(priv->mdev->dev, MLX4_ACCESS_REG_WRITE,
+				   &ptys_reg);
+	if (ret) {
+		en_warn(priv, "Failed to write mlx4_ACCESS_PTYS_REG eth_proto_admin(0x%x) status(0x%x)",
+			be32_to_cpu(ptys_reg.eth_proto_admin), ret);
+		return ret;
+	}
+
+	mutex_lock(&priv->mdev->state_lock);
+	if (priv->port_up) {
+		en_warn(priv, "Port link mode changed, restarting port...\n");
+		mlx4_en_stop_port(dev, 1);
+		if (mlx4_en_start_port(dev))
+			en_err(priv, "Failed restarting port %d\n", priv->port);
+	}
+	mutex_unlock(&priv->mdev->state_lock);
+	return 0;
+}
+
+static int mlx4_en_get_coalesce(struct net_device *dev,
+				struct ethtool_coalesce *coal,
+				struct kernel_ethtool_coalesce *kernel_coal,
+				struct netlink_ext_ack *extack)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+
+	coal->tx_coalesce_usecs = priv->tx_usecs;
+	coal->tx_max_coalesced_frames = priv->tx_frames;
+	coal->tx_max_coalesced_frames_irq = priv->tx_work_limit;
+
+	coal->rx_coalesce_usecs = priv->rx_usecs;
+	coal->rx_max_coalesced_frames = priv->rx_frames;
+
+	coal->pkt_rate_low = priv->pkt_rate_low;
+	coal->rx_coalesce_usecs_low = priv->rx_usecs_low;
+	coal->pkt_rate_high = priv->pkt_rate_high;
+	coal->rx_coalesce_usecs_high = priv->rx_usecs_high;
+	coal->rate_sample_interval = priv->sample_interval;
+	coal->use_adaptive_rx_coalesce = priv->adaptive_rx_coal;
+
+	return 0;
+}
+
+static int mlx4_en_set_coalesce(struct net_device *dev,
+				struct ethtool_coalesce *coal,
+				struct kernel_ethtool_coalesce *kernel_coal,
+				struct netlink_ext_ack *extack)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+
+	if (!coal->tx_max_coalesced_frames_irq)
+		return -EINVAL;
+
+	if (coal->tx_coalesce_usecs > MLX4_EN_MAX_COAL_TIME ||
+	    coal->rx_coalesce_usecs > MLX4_EN_MAX_COAL_TIME ||
+	    coal->rx_coalesce_usecs_low > MLX4_EN_MAX_COAL_TIME ||
+	    coal->rx_coalesce_usecs_high > MLX4_EN_MAX_COAL_TIME) {
+		netdev_info(dev, "%s: maximum coalesce time supported is %d usecs\n",
+			    __func__, MLX4_EN_MAX_COAL_TIME);
+		return -ERANGE;
+	}
+
+	if (coal->tx_max_coalesced_frames > MLX4_EN_MAX_COAL_PKTS ||
+	    coal->rx_max_coalesced_frames > MLX4_EN_MAX_COAL_PKTS) {
+		netdev_info(dev, "%s: maximum coalesced frames supported is %d\n",
+			    __func__, MLX4_EN_MAX_COAL_PKTS);
+		return -ERANGE;
+	}
+
+	priv->rx_frames = (coal->rx_max_coalesced_frames ==
+			   MLX4_EN_AUTO_CONF) ?
+				MLX4_EN_RX_COAL_TARGET :
+				coal->rx_max_coalesced_frames;
+	priv->rx_usecs = (coal->rx_coalesce_usecs ==
+			  MLX4_EN_AUTO_CONF) ?
+				MLX4_EN_RX_COAL_TIME :
+				coal->rx_coalesce_usecs;
+
+	/* Setting TX coalescing parameters */
+	if (coal->tx_coalesce_usecs != priv->tx_usecs ||
+	    coal->tx_max_coalesced_frames != priv->tx_frames) {
+		priv->tx_usecs = coal->tx_coalesce_usecs;
+		priv->tx_frames = coal->tx_max_coalesced_frames;
+	}
+
+	/* Set adaptive coalescing params */
+	priv->pkt_rate_low = coal->pkt_rate_low;
+	priv->rx_usecs_low = coal->rx_coalesce_usecs_low;
+	priv->pkt_rate_high = coal->pkt_rate_high;
+	priv->rx_usecs_high = coal->rx_coalesce_usecs_high;
+	priv->sample_interval = coal->rate_sample_interval;
+	priv->adaptive_rx_coal = coal->use_adaptive_rx_coalesce;
+	priv->tx_work_limit = coal->tx_max_coalesced_frames_irq;
+
+	return mlx4_en_moderation_update(priv);
+}
+
+static int mlx4_en_set_pauseparam(struct net_device *dev,
+				struct ethtool_pauseparam *pause)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	struct mlx4_en_dev *mdev = priv->mdev;
+	u8 tx_pause, tx_ppp, rx_pause, rx_ppp;
+	int err;
+
+	if (pause->autoneg)
+		return -EINVAL;
+
+	tx_pause = !!(pause->tx_pause);
+	rx_pause = !!(pause->rx_pause);
+	rx_ppp = (tx_pause || rx_pause) ? 0 : priv->prof->rx_ppp;
+	tx_ppp = (tx_pause || rx_pause) ? 0 : priv->prof->tx_ppp;
+
+	err = mlx4_SET_PORT_general(mdev->dev, priv->port,
+				    priv->rx_skb_size + ETH_FCS_LEN,
+				    tx_pause, tx_ppp, rx_pause, rx_ppp);
+	if (err) {
+		en_err(priv, "Failed setting pause params, err = %d\n", err);
+		return err;
+	}
+
+	mlx4_en_update_pfc_stats_bitmap(mdev->dev, &priv->stats_bitmap,
+					rx_ppp, rx_pause, tx_ppp, tx_pause);
+
+	priv->prof->tx_pause = tx_pause;
+	priv->prof->rx_pause = rx_pause;
+	priv->prof->tx_ppp = tx_ppp;
+	priv->prof->rx_ppp = rx_ppp;
+
+	return err;
+}
+
+static void mlx4_en_get_pause_stats(struct net_device *dev,
+				    struct ethtool_pause_stats *stats)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	struct bitmap_iterator it;
+
+	bitmap_iterator_init(&it, priv->stats_bitmap.bitmap, NUM_ALL_STATS);
+
+	spin_lock_bh(&priv->stats_lock);
+	if (test_bit(FLOW_PRIORITY_STATS_IDX_TX_FRAMES,
+		     priv->stats_bitmap.bitmap))
+		stats->tx_pause_frames = priv->tx_flowstats.tx_pause;
+	if (test_bit(FLOW_PRIORITY_STATS_IDX_RX_FRAMES,
+		     priv->stats_bitmap.bitmap))
+		stats->rx_pause_frames = priv->rx_flowstats.rx_pause;
+	spin_unlock_bh(&priv->stats_lock);
+}
+
+static void mlx4_en_get_pauseparam(struct net_device *dev,
+				 struct ethtool_pauseparam *pause)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+
+	pause->tx_pause = priv->prof->tx_pause;
+	pause->rx_pause = priv->prof->rx_pause;
+}
+
+static int mlx4_en_set_ringparam(struct net_device *dev,
+				 struct ethtool_ringparam *param,
+				 struct kernel_ethtool_ringparam *kernel_param,
+				 struct netlink_ext_ack *extack)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	struct mlx4_en_dev *mdev = priv->mdev;
+	struct mlx4_en_port_profile new_prof;
+	struct mlx4_en_priv *tmp;
+	u32 rx_size, tx_size;
+	int port_up = 0;
+	int err = 0;
+
+	if (param->rx_jumbo_pending || param->rx_mini_pending)
+		return -EINVAL;
+
+	if (param->rx_pending < MLX4_EN_MIN_RX_SIZE) {
+		en_warn(priv, "%s: rx_pending (%d) < min (%d)\n",
+			__func__, param->rx_pending,
+			MLX4_EN_MIN_RX_SIZE);
+		return -EINVAL;
+	}
+	if (param->tx_pending < MLX4_EN_MIN_TX_SIZE) {
+		en_warn(priv, "%s: tx_pending (%d) < min (%lu)\n",
+			__func__, param->tx_pending,
+			MLX4_EN_MIN_TX_SIZE);
+		return -EINVAL;
+	}
+
+	rx_size = roundup_pow_of_two(param->rx_pending);
+	tx_size = roundup_pow_of_two(param->tx_pending);
+
+	if (rx_size == (priv->port_up ? priv->rx_ring[0]->actual_size :
+					priv->rx_ring[0]->size) &&
+	    tx_size == priv->tx_ring[TX][0]->size)
+		return 0;
+
+	tmp = kzalloc(sizeof(*tmp), GFP_KERNEL);
+	if (!tmp)
+		return -ENOMEM;
+
+	mutex_lock(&mdev->state_lock);
+	memcpy(&new_prof, priv->prof, sizeof(struct mlx4_en_port_profile));
+	new_prof.tx_ring_size = tx_size;
+	new_prof.rx_ring_size = rx_size;
+	err = mlx4_en_try_alloc_resources(priv, tmp, &new_prof, true);
+	if (err)
+		goto out;
+
+	if (priv->port_up) {
+		port_up = 1;
+		mlx4_en_stop_port(dev, 1);
+	}
+
+	mlx4_en_safe_replace_resources(priv, tmp);
+
+	if (port_up) {
+		err = mlx4_en_start_port(dev);
+		if (err)
+			en_err(priv, "Failed starting port\n");
+	}
+
+	err = mlx4_en_moderation_update(priv);
+out:
+	kfree(tmp);
+	mutex_unlock(&mdev->state_lock);
+	return err;
+}
+
+static void mlx4_en_get_ringparam(struct net_device *dev,
+				  struct ethtool_ringparam *param,
+				  struct kernel_ethtool_ringparam *kernel_param,
+				  struct netlink_ext_ack *extack)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+
+	memset(param, 0, sizeof(*param));
+	param->rx_max_pending = MLX4_EN_MAX_RX_SIZE;
+	param->tx_max_pending = MLX4_EN_MAX_TX_SIZE;
+	param->rx_pending = priv->port_up ?
+		priv->rx_ring[0]->actual_size : priv->rx_ring[0]->size;
+	param->tx_pending = priv->tx_ring[TX][0]->size;
+}
+
+static u32 mlx4_en_get_rxfh_indir_size(struct net_device *dev)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+
+	return rounddown_pow_of_two(priv->rx_ring_num);
+}
+
+static u32 mlx4_en_get_rxfh_key_size(struct net_device *netdev)
+{
+	return MLX4_EN_RSS_KEY_SIZE;
+}
+
+static int mlx4_en_check_rxfh_func(struct net_device *dev, u8 hfunc)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+
+	/* check if requested function is supported by the device */
+	if (hfunc == ETH_RSS_HASH_TOP) {
+		if (!(priv->mdev->dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_RSS_TOP))
+			return -EINVAL;
+		if (!(dev->features & NETIF_F_RXHASH))
+			en_warn(priv, "Toeplitz hash function should be used in conjunction with RX hashing for optimal performance\n");
+		return 0;
+	} else if (hfunc == ETH_RSS_HASH_XOR) {
+		if (!(priv->mdev->dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_RSS_XOR))
+			return -EINVAL;
+		if (dev->features & NETIF_F_RXHASH)
+			en_warn(priv, "Enabling both XOR Hash function and RX Hashing can limit RPS functionality\n");
+		return 0;
+	}
+
+	return -EINVAL;
+}
+
+static int mlx4_en_get_rxfh(struct net_device *dev, u32 *ring_index, u8 *key,
+			    u8 *hfunc)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	u32 n = mlx4_en_get_rxfh_indir_size(dev);
+	u32 i, rss_rings;
+
+	rss_rings = priv->prof->rss_rings ?: n;
+	rss_rings = rounddown_pow_of_two(rss_rings);
+
+	for (i = 0; i < n; i++) {
+		if (!ring_index)
+			break;
+		ring_index[i] = i % rss_rings;
+	}
+	if (key)
+		memcpy(key, priv->rss_key, MLX4_EN_RSS_KEY_SIZE);
+	if (hfunc)
+		*hfunc = priv->rss_hash_fn;
+	return 0;
+}
+
+static int mlx4_en_set_rxfh(struct net_device *dev, const u32 *ring_index,
+			    const u8 *key, const u8 hfunc)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	u32 n = mlx4_en_get_rxfh_indir_size(dev);
+	struct mlx4_en_dev *mdev = priv->mdev;
+	int port_up = 0;
+	int err = 0;
+	int i;
+	int rss_rings = 0;
+
+	/* Calculate RSS table size and make sure flows are spread evenly
+	 * between rings
+	 */
+	for (i = 0; i < n; i++) {
+		if (!ring_index)
+			break;
+		if (i > 0 && !ring_index[i] && !rss_rings)
+			rss_rings = i;
+
+		if (ring_index[i] != (i % (rss_rings ?: n)))
+			return -EINVAL;
+	}
+
+	if (!rss_rings)
+		rss_rings = n;
+
+	/* RSS table size must be an order of 2 */
+	if (!is_power_of_2(rss_rings))
+		return -EINVAL;
+
+	if (hfunc != ETH_RSS_HASH_NO_CHANGE) {
+		err = mlx4_en_check_rxfh_func(dev, hfunc);
+		if (err)
+			return err;
+	}
+
+	mutex_lock(&mdev->state_lock);
+	if (priv->port_up) {
+		port_up = 1;
+		mlx4_en_stop_port(dev, 1);
+	}
+
+	if (ring_index)
+		priv->prof->rss_rings = rss_rings;
+	if (key)
+		memcpy(priv->rss_key, key, MLX4_EN_RSS_KEY_SIZE);
+	if (hfunc !=  ETH_RSS_HASH_NO_CHANGE)
+		priv->rss_hash_fn = hfunc;
+
+	if (port_up) {
+		err = mlx4_en_start_port(dev);
+		if (err)
+			en_err(priv, "Failed starting port\n");
+	}
+
+	mutex_unlock(&mdev->state_lock);
+	return err;
+}
+
+#define all_zeros_or_all_ones(field)		\
+	((field) == 0 || (field) == (__force typeof(field))-1)
+
+static int mlx4_en_validate_flow(struct net_device *dev,
+				 struct ethtool_rxnfc *cmd)
+{
+	struct ethtool_usrip4_spec *l3_mask;
+	struct ethtool_tcpip4_spec *l4_mask;
+	struct ethhdr *eth_mask;
+
+	if (cmd->fs.location >= MAX_NUM_OF_FS_RULES)
+		return -EINVAL;
+
+	if (cmd->fs.flow_type & FLOW_MAC_EXT) {
+		/* dest mac mask must be ff:ff:ff:ff:ff:ff */
+		if (!is_broadcast_ether_addr(cmd->fs.m_ext.h_dest))
+			return -EINVAL;
+	}
+
+	switch (cmd->fs.flow_type & ~(FLOW_EXT | FLOW_MAC_EXT)) {
+	case TCP_V4_FLOW:
+	case UDP_V4_FLOW:
+		if (cmd->fs.m_u.tcp_ip4_spec.tos)
+			return -EINVAL;
+		l4_mask = &cmd->fs.m_u.tcp_ip4_spec;
+		/* don't allow mask which isn't all 0 or 1 */
+		if (!all_zeros_or_all_ones(l4_mask->ip4src) ||
+		    !all_zeros_or_all_ones(l4_mask->ip4dst) ||
+		    !all_zeros_or_all_ones(l4_mask->psrc) ||
+		    !all_zeros_or_all_ones(l4_mask->pdst))
+			return -EINVAL;
+		break;
+	case IP_USER_FLOW:
+		l3_mask = &cmd->fs.m_u.usr_ip4_spec;
+		if (l3_mask->l4_4_bytes || l3_mask->tos || l3_mask->proto ||
+		    cmd->fs.h_u.usr_ip4_spec.ip_ver != ETH_RX_NFC_IP4 ||
+		    (!l3_mask->ip4src && !l3_mask->ip4dst) ||
+		    !all_zeros_or_all_ones(l3_mask->ip4src) ||
+		    !all_zeros_or_all_ones(l3_mask->ip4dst))
+			return -EINVAL;
+		break;
+	case ETHER_FLOW:
+		eth_mask = &cmd->fs.m_u.ether_spec;
+		/* source mac mask must not be set */
+		if (!is_zero_ether_addr(eth_mask->h_source))
+			return -EINVAL;
+
+		/* dest mac mask must be ff:ff:ff:ff:ff:ff */
+		if (!is_broadcast_ether_addr(eth_mask->h_dest))
+			return -EINVAL;
+
+		if (!all_zeros_or_all_ones(eth_mask->h_proto))
+			return -EINVAL;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	if ((cmd->fs.flow_type & FLOW_EXT)) {
+		if (cmd->fs.m_ext.vlan_etype ||
+		    !((cmd->fs.m_ext.vlan_tci & cpu_to_be16(VLAN_VID_MASK)) ==
+		      0 ||
+		      (cmd->fs.m_ext.vlan_tci & cpu_to_be16(VLAN_VID_MASK)) ==
+		      cpu_to_be16(VLAN_VID_MASK)))
+			return -EINVAL;
+
+		if (cmd->fs.m_ext.vlan_tci) {
+			if (be16_to_cpu(cmd->fs.h_ext.vlan_tci) >= VLAN_N_VID)
+				return -EINVAL;
+
+		}
+	}
+
+	return 0;
+}
+
+static int mlx4_en_ethtool_add_mac_rule(struct ethtool_rxnfc *cmd,
+					struct list_head *rule_list_h,
+					struct mlx4_spec_list *spec_l2,
+					unsigned char *mac)
+{
+	__be64 mac_msk = cpu_to_be64(MLX4_MAC_MASK << 16);
+
+	spec_l2->id = MLX4_NET_TRANS_RULE_ID_ETH;
+	memcpy(spec_l2->eth.dst_mac_msk, &mac_msk, ETH_ALEN);
+	memcpy(spec_l2->eth.dst_mac, mac, ETH_ALEN);
+
+	if ((cmd->fs.flow_type & FLOW_EXT) &&
+	    (cmd->fs.m_ext.vlan_tci & cpu_to_be16(VLAN_VID_MASK))) {
+		spec_l2->eth.vlan_id = cmd->fs.h_ext.vlan_tci;
+		spec_l2->eth.vlan_id_msk = cpu_to_be16(VLAN_VID_MASK);
+	}
+
+	list_add_tail(&spec_l2->list, rule_list_h);
+
+	return 0;
+}
+
+static int mlx4_en_ethtool_add_mac_rule_by_ipv4(struct mlx4_en_priv *priv,
+						struct ethtool_rxnfc *cmd,
+						struct list_head *rule_list_h,
+						struct mlx4_spec_list *spec_l2,
+						__be32 ipv4_dst)
+{
+#ifdef CONFIG_INET
+	unsigned char mac[ETH_ALEN];
+
+	if (!ipv4_is_multicast(ipv4_dst)) {
+		if (cmd->fs.flow_type & FLOW_MAC_EXT)
+			memcpy(&mac, cmd->fs.h_ext.h_dest, ETH_ALEN);
+		else
+			memcpy(&mac, priv->dev->dev_addr, ETH_ALEN);
+	} else {
+		ip_eth_mc_map(ipv4_dst, mac);
+	}
+
+	return mlx4_en_ethtool_add_mac_rule(cmd, rule_list_h, spec_l2, &mac[0]);
+#else
+	return -EINVAL;
+#endif
+}
+
+static int add_ip_rule(struct mlx4_en_priv *priv,
+		       struct ethtool_rxnfc *cmd,
+		       struct list_head *list_h)
+{
+	int err;
+	struct mlx4_spec_list *spec_l2 = NULL;
+	struct mlx4_spec_list *spec_l3 = NULL;
+	struct ethtool_usrip4_spec *l3_mask = &cmd->fs.m_u.usr_ip4_spec;
+
+	spec_l3 = kzalloc(sizeof(*spec_l3), GFP_KERNEL);
+	spec_l2 = kzalloc(sizeof(*spec_l2), GFP_KERNEL);
+	if (!spec_l2 || !spec_l3) {
+		err = -ENOMEM;
+		goto free_spec;
+	}
+
+	err = mlx4_en_ethtool_add_mac_rule_by_ipv4(priv, cmd, list_h, spec_l2,
+						   cmd->fs.h_u.
+						   usr_ip4_spec.ip4dst);
+	if (err)
+		goto free_spec;
+	spec_l3->id = MLX4_NET_TRANS_RULE_ID_IPV4;
+	spec_l3->ipv4.src_ip = cmd->fs.h_u.usr_ip4_spec.ip4src;
+	if (l3_mask->ip4src)
+		spec_l3->ipv4.src_ip_msk = EN_ETHTOOL_WORD_MASK;
+	spec_l3->ipv4.dst_ip = cmd->fs.h_u.usr_ip4_spec.ip4dst;
+	if (l3_mask->ip4dst)
+		spec_l3->ipv4.dst_ip_msk = EN_ETHTOOL_WORD_MASK;
+	list_add_tail(&spec_l3->list, list_h);
+
+	return 0;
+
+free_spec:
+	kfree(spec_l2);
+	kfree(spec_l3);
+	return err;
+}
+
+static int add_tcp_udp_rule(struct mlx4_en_priv *priv,
+			     struct ethtool_rxnfc *cmd,
+			     struct list_head *list_h, int proto)
+{
+	int err;
+	struct mlx4_spec_list *spec_l2 = NULL;
+	struct mlx4_spec_list *spec_l3 = NULL;
+	struct mlx4_spec_list *spec_l4 = NULL;
+	struct ethtool_tcpip4_spec *l4_mask = &cmd->fs.m_u.tcp_ip4_spec;
+
+	spec_l2 = kzalloc(sizeof(*spec_l2), GFP_KERNEL);
+	spec_l3 = kzalloc(sizeof(*spec_l3), GFP_KERNEL);
+	spec_l4 = kzalloc(sizeof(*spec_l4), GFP_KERNEL);
+	if (!spec_l2 || !spec_l3 || !spec_l4) {
+		err = -ENOMEM;
+		goto free_spec;
+	}
+
+	spec_l3->id = MLX4_NET_TRANS_RULE_ID_IPV4;
+
+	if (proto == TCP_V4_FLOW) {
+		err = mlx4_en_ethtool_add_mac_rule_by_ipv4(priv, cmd, list_h,
+							   spec_l2,
+							   cmd->fs.h_u.
+							   tcp_ip4_spec.ip4dst);
+		if (err)
+			goto free_spec;
+		spec_l4->id = MLX4_NET_TRANS_RULE_ID_TCP;
+		spec_l3->ipv4.src_ip = cmd->fs.h_u.tcp_ip4_spec.ip4src;
+		spec_l3->ipv4.dst_ip = cmd->fs.h_u.tcp_ip4_spec.ip4dst;
+		spec_l4->tcp_udp.src_port = cmd->fs.h_u.tcp_ip4_spec.psrc;
+		spec_l4->tcp_udp.dst_port = cmd->fs.h_u.tcp_ip4_spec.pdst;
+	} else {
+		err = mlx4_en_ethtool_add_mac_rule_by_ipv4(priv, cmd, list_h,
+							   spec_l2,
+							   cmd->fs.h_u.
+							   udp_ip4_spec.ip4dst);
+		if (err)
+			goto free_spec;
+		spec_l4->id = MLX4_NET_TRANS_RULE_ID_UDP;
+		spec_l3->ipv4.src_ip = cmd->fs.h_u.udp_ip4_spec.ip4src;
+		spec_l3->ipv4.dst_ip = cmd->fs.h_u.udp_ip4_spec.ip4dst;
+		spec_l4->tcp_udp.src_port = cmd->fs.h_u.udp_ip4_spec.psrc;
+		spec_l4->tcp_udp.dst_port = cmd->fs.h_u.udp_ip4_spec.pdst;
+	}
+
+	if (l4_mask->ip4src)
+		spec_l3->ipv4.src_ip_msk = EN_ETHTOOL_WORD_MASK;
+	if (l4_mask->ip4dst)
+		spec_l3->ipv4.dst_ip_msk = EN_ETHTOOL_WORD_MASK;
+
+	if (l4_mask->psrc)
+		spec_l4->tcp_udp.src_port_msk = EN_ETHTOOL_SHORT_MASK;
+	if (l4_mask->pdst)
+		spec_l4->tcp_udp.dst_port_msk = EN_ETHTOOL_SHORT_MASK;
+
+	list_add_tail(&spec_l3->list, list_h);
+	list_add_tail(&spec_l4->list, list_h);
+
+	return 0;
+
+free_spec:
+	kfree(spec_l2);
+	kfree(spec_l3);
+	kfree(spec_l4);
+	return err;
+}
+
+static int mlx4_en_ethtool_to_net_trans_rule(struct net_device *dev,
+					     struct ethtool_rxnfc *cmd,
+					     struct list_head *rule_list_h)
+{
+	int err;
+	struct ethhdr *eth_spec;
+	struct mlx4_spec_list *spec_l2;
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+
+	err = mlx4_en_validate_flow(dev, cmd);
+	if (err)
+		return err;
+
+	switch (cmd->fs.flow_type & ~(FLOW_EXT | FLOW_MAC_EXT)) {
+	case ETHER_FLOW:
+		spec_l2 = kzalloc(sizeof(*spec_l2), GFP_KERNEL);
+		if (!spec_l2)
+			return -ENOMEM;
+
+		eth_spec = &cmd->fs.h_u.ether_spec;
+		mlx4_en_ethtool_add_mac_rule(cmd, rule_list_h, spec_l2,
+					     &eth_spec->h_dest[0]);
+		spec_l2->eth.ether_type = eth_spec->h_proto;
+		if (eth_spec->h_proto)
+			spec_l2->eth.ether_type_enable = 1;
+		break;
+	case IP_USER_FLOW:
+		err = add_ip_rule(priv, cmd, rule_list_h);
+		break;
+	case TCP_V4_FLOW:
+		err = add_tcp_udp_rule(priv, cmd, rule_list_h, TCP_V4_FLOW);
+		break;
+	case UDP_V4_FLOW:
+		err = add_tcp_udp_rule(priv, cmd, rule_list_h, UDP_V4_FLOW);
+		break;
+	}
+
+	return err;
+}
+
+static int mlx4_en_flow_replace(struct net_device *dev,
+				struct ethtool_rxnfc *cmd)
+{
+	int err;
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	struct ethtool_flow_id *loc_rule;
+	struct mlx4_spec_list *spec, *tmp_spec;
+	u32 qpn;
+	u64 reg_id;
+
+	struct mlx4_net_trans_rule rule = {
+		.queue_mode = MLX4_NET_TRANS_Q_FIFO,
+		.exclusive = 0,
+		.allow_loopback = 1,
+		.promisc_mode = MLX4_FS_REGULAR,
+	};
+
+	rule.port = priv->port;
+	rule.priority = MLX4_DOMAIN_ETHTOOL | cmd->fs.location;
+	INIT_LIST_HEAD(&rule.list);
+
+	/* Allow direct QP attaches if the EN_ETHTOOL_QP_ATTACH flag is set */
+	if (cmd->fs.ring_cookie == RX_CLS_FLOW_DISC)
+		qpn = priv->drop_qp.qpn;
+	else if (cmd->fs.ring_cookie & EN_ETHTOOL_QP_ATTACH) {
+		qpn = cmd->fs.ring_cookie & (EN_ETHTOOL_QP_ATTACH - 1);
+	} else {
+		if (cmd->fs.ring_cookie >= priv->rx_ring_num) {
+			en_warn(priv, "rxnfc: RX ring (%llu) doesn't exist\n",
+				cmd->fs.ring_cookie);
+			return -EINVAL;
+		}
+		qpn = priv->rss_map.qps[cmd->fs.ring_cookie].qpn;
+		if (!qpn) {
+			en_warn(priv, "rxnfc: RX ring (%llu) is inactive\n",
+				cmd->fs.ring_cookie);
+			return -EINVAL;
+		}
+	}
+	rule.qpn = qpn;
+	err = mlx4_en_ethtool_to_net_trans_rule(dev, cmd, &rule.list);
+	if (err)
+		goto out_free_list;
+
+	loc_rule = &priv->ethtool_rules[cmd->fs.location];
+	if (loc_rule->id) {
+		err = mlx4_flow_detach(priv->mdev->dev, loc_rule->id);
+		if (err) {
+			en_err(priv, "Fail to detach network rule at location %d. registration id = %llx\n",
+			       cmd->fs.location, loc_rule->id);
+			goto out_free_list;
+		}
+		loc_rule->id = 0;
+		memset(&loc_rule->flow_spec, 0,
+		       sizeof(struct ethtool_rx_flow_spec));
+		list_del(&loc_rule->list);
+	}
+	err = mlx4_flow_attach(priv->mdev->dev, &rule, &reg_id);
+	if (err) {
+		en_err(priv, "Fail to attach network rule at location %d\n",
+		       cmd->fs.location);
+		goto out_free_list;
+	}
+	loc_rule->id = reg_id;
+	memcpy(&loc_rule->flow_spec, &cmd->fs,
+	       sizeof(struct ethtool_rx_flow_spec));
+	list_add_tail(&loc_rule->list, &priv->ethtool_list);
+
+out_free_list:
+	list_for_each_entry_safe(spec, tmp_spec, &rule.list, list) {
+		list_del(&spec->list);
+		kfree(spec);
+	}
+	return err;
+}
+
+static int mlx4_en_flow_detach(struct net_device *dev,
+			       struct ethtool_rxnfc *cmd)
+{
+	int err = 0;
+	struct ethtool_flow_id *rule;
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+
+	if (cmd->fs.location >= MAX_NUM_OF_FS_RULES)
+		return -EINVAL;
+
+	rule = &priv->ethtool_rules[cmd->fs.location];
+	if (!rule->id) {
+		err =  -ENOENT;
+		goto out;
+	}
+
+	err = mlx4_flow_detach(priv->mdev->dev, rule->id);
+	if (err) {
+		en_err(priv, "Fail to detach network rule at location %d. registration id = 0x%llx\n",
+		       cmd->fs.location, rule->id);
+		goto out;
+	}
+	rule->id = 0;
+	memset(&rule->flow_spec, 0, sizeof(struct ethtool_rx_flow_spec));
+	list_del(&rule->list);
+out:
+	return err;
+
+}
+
+static int mlx4_en_get_flow(struct net_device *dev, struct ethtool_rxnfc *cmd,
+			    int loc)
+{
+	int err = 0;
+	struct ethtool_flow_id *rule;
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+
+	if (loc < 0 || loc >= MAX_NUM_OF_FS_RULES)
+		return -EINVAL;
+
+	rule = &priv->ethtool_rules[loc];
+	if (rule->id)
+		memcpy(&cmd->fs, &rule->flow_spec,
+		       sizeof(struct ethtool_rx_flow_spec));
+	else
+		err = -ENOENT;
+
+	return err;
+}
+
+static int mlx4_en_get_num_flows(struct mlx4_en_priv *priv)
+{
+
+	int i, res = 0;
+	for (i = 0; i < MAX_NUM_OF_FS_RULES; i++) {
+		if (priv->ethtool_rules[i].id)
+			res++;
+	}
+	return res;
+
+}
+
+static int mlx4_en_get_rxnfc(struct net_device *dev, struct ethtool_rxnfc *cmd,
+			     u32 *rule_locs)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	struct mlx4_en_dev *mdev = priv->mdev;
+	int err = 0;
+	int i = 0, priority = 0;
+
+	if ((cmd->cmd == ETHTOOL_GRXCLSRLCNT ||
+	     cmd->cmd == ETHTOOL_GRXCLSRULE ||
+	     cmd->cmd == ETHTOOL_GRXCLSRLALL) &&
+	    (mdev->dev->caps.steering_mode !=
+	     MLX4_STEERING_MODE_DEVICE_MANAGED || !priv->port_up))
+		return -EINVAL;
+
+	switch (cmd->cmd) {
+	case ETHTOOL_GRXRINGS:
+		cmd->data = priv->rx_ring_num;
+		break;
+	case ETHTOOL_GRXCLSRLCNT:
+		cmd->rule_cnt = mlx4_en_get_num_flows(priv);
+		break;
+	case ETHTOOL_GRXCLSRULE:
+		err = mlx4_en_get_flow(dev, cmd, cmd->fs.location);
+		break;
+	case ETHTOOL_GRXCLSRLALL:
+		cmd->data = MAX_NUM_OF_FS_RULES;
+		while ((!err || err == -ENOENT) && priority < cmd->rule_cnt) {
+			err = mlx4_en_get_flow(dev, cmd, i);
+			if (!err)
+				rule_locs[priority++] = i;
+			i++;
+		}
+		err = 0;
+		break;
+	default:
+		err = -EOPNOTSUPP;
+		break;
+	}
+
+	return err;
+}
+
+static int mlx4_en_set_rxnfc(struct net_device *dev, struct ethtool_rxnfc *cmd)
+{
+	int err = 0;
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	struct mlx4_en_dev *mdev = priv->mdev;
+
+	if (mdev->dev->caps.steering_mode !=
+	    MLX4_STEERING_MODE_DEVICE_MANAGED || !priv->port_up)
+		return -EINVAL;
+
+	switch (cmd->cmd) {
+	case ETHTOOL_SRXCLSRLINS:
+		err = mlx4_en_flow_replace(dev, cmd);
+		break;
+	case ETHTOOL_SRXCLSRLDEL:
+		err = mlx4_en_flow_detach(dev, cmd);
+		break;
+	default:
+		en_warn(priv, "Unsupported ethtool command. (%d)\n", cmd->cmd);
+		return -EINVAL;
+	}
+
+	return err;
+}
+
+static int mlx4_en_get_max_num_rx_rings(struct net_device *dev)
+{
+	return min_t(int, num_online_cpus(), MAX_RX_RINGS);
+}
+
+static void mlx4_en_get_channels(struct net_device *dev,
+				 struct ethtool_channels *channel)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+
+	channel->max_rx = mlx4_en_get_max_num_rx_rings(dev);
+	channel->max_tx = priv->mdev->profile.max_num_tx_rings_p_up;
+
+	channel->rx_count = priv->rx_ring_num;
+	channel->tx_count = priv->tx_ring_num[TX] /
+			    priv->prof->num_up;
+}
+
+static int mlx4_en_set_channels(struct net_device *dev,
+				struct ethtool_channels *channel)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	struct mlx4_en_dev *mdev = priv->mdev;
+	struct mlx4_en_port_profile new_prof;
+	struct mlx4_en_priv *tmp;
+	int total_tx_count;
+	int port_up = 0;
+	int xdp_count;
+	int err = 0;
+	u8 up;
+
+	if (!channel->tx_count || !channel->rx_count)
+		return -EINVAL;
+
+	tmp = kzalloc(sizeof(*tmp), GFP_KERNEL);
+	if (!tmp)
+		return -ENOMEM;
+
+	mutex_lock(&mdev->state_lock);
+	xdp_count = priv->tx_ring_num[TX_XDP] ? channel->rx_count : 0;
+	total_tx_count = channel->tx_count * priv->prof->num_up + xdp_count;
+	if (total_tx_count > MAX_TX_RINGS) {
+		err = -EINVAL;
+		en_err(priv,
+		       "Total number of TX and XDP rings (%d) exceeds the maximum supported (%d)\n",
+		       total_tx_count, MAX_TX_RINGS);
+		goto out;
+	}
+
+	memcpy(&new_prof, priv->prof, sizeof(struct mlx4_en_port_profile));
+	new_prof.num_tx_rings_p_up = channel->tx_count;
+	new_prof.tx_ring_num[TX] = channel->tx_count * priv->prof->num_up;
+	new_prof.tx_ring_num[TX_XDP] = xdp_count;
+	new_prof.rx_ring_num = channel->rx_count;
+
+	err = mlx4_en_try_alloc_resources(priv, tmp, &new_prof, true);
+	if (err)
+		goto out;
+
+	if (priv->port_up) {
+		port_up = 1;
+		mlx4_en_stop_port(dev, 1);
+	}
+
+	mlx4_en_safe_replace_resources(priv, tmp);
+
+	netif_set_real_num_rx_queues(dev, priv->rx_ring_num);
+
+	up = (priv->prof->num_up == MLX4_EN_NUM_UP_LOW) ?
+				    0 : priv->prof->num_up;
+	mlx4_en_setup_tc(dev, up);
+
+	en_warn(priv, "Using %d TX rings\n", priv->tx_ring_num[TX]);
+	en_warn(priv, "Using %d RX rings\n", priv->rx_ring_num);
+
+	if (port_up) {
+		err = mlx4_en_start_port(dev);
+		if (err)
+			en_err(priv, "Failed starting port\n");
+	}
+
+	err = mlx4_en_moderation_update(priv);
+out:
+	mutex_unlock(&mdev->state_lock);
+	kfree(tmp);
+	return err;
+}
+
+static int mlx4_en_get_ts_info(struct net_device *dev,
+			       struct ethtool_ts_info *info)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	struct mlx4_en_dev *mdev = priv->mdev;
+	int ret;
+
+	ret = ethtool_op_get_ts_info(dev, info);
+	if (ret)
+		return ret;
+
+	if (mdev->dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_TS) {
+		info->so_timestamping |=
+			SOF_TIMESTAMPING_TX_HARDWARE |
+			SOF_TIMESTAMPING_RX_HARDWARE |
+			SOF_TIMESTAMPING_RAW_HARDWARE;
+
+		info->tx_types =
+			(1 << HWTSTAMP_TX_OFF) |
+			(1 << HWTSTAMP_TX_ON);
+
+		info->rx_filters =
+			(1 << HWTSTAMP_FILTER_NONE) |
+			(1 << HWTSTAMP_FILTER_ALL);
+
+		if (mdev->ptp_clock)
+			info->phc_index = ptp_clock_index(mdev->ptp_clock);
+	}
+
+	return ret;
+}
+
+static int mlx4_en_set_priv_flags(struct net_device *dev, u32 flags)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	struct mlx4_en_dev *mdev = priv->mdev;
+	bool bf_enabled_new = !!(flags & MLX4_EN_PRIV_FLAGS_BLUEFLAME);
+	bool bf_enabled_old = !!(priv->pflags & MLX4_EN_PRIV_FLAGS_BLUEFLAME);
+	bool phv_enabled_new = !!(flags & MLX4_EN_PRIV_FLAGS_PHV);
+	bool phv_enabled_old = !!(priv->pflags & MLX4_EN_PRIV_FLAGS_PHV);
+	int i;
+	int ret = 0;
+
+	if (bf_enabled_new != bf_enabled_old) {
+		int t;
+
+		if (bf_enabled_new) {
+			bool bf_supported = true;
+
+			for (t = 0; t < MLX4_EN_NUM_TX_TYPES; t++)
+				for (i = 0; i < priv->tx_ring_num[t]; i++)
+					bf_supported &=
+						priv->tx_ring[t][i]->bf_alloced;
+
+			if (!bf_supported) {
+				en_err(priv, "BlueFlame is not supported\n");
+				return -EINVAL;
+			}
+
+			priv->pflags |= MLX4_EN_PRIV_FLAGS_BLUEFLAME;
+		} else {
+			priv->pflags &= ~MLX4_EN_PRIV_FLAGS_BLUEFLAME;
+		}
+
+		for (t = 0; t < MLX4_EN_NUM_TX_TYPES; t++)
+			for (i = 0; i < priv->tx_ring_num[t]; i++)
+				priv->tx_ring[t][i]->bf_enabled =
+					bf_enabled_new;
+
+		en_info(priv, "BlueFlame %s\n",
+			bf_enabled_new ?  "Enabled" : "Disabled");
+	}
+
+	if (phv_enabled_new != phv_enabled_old) {
+		ret = set_phv_bit(mdev->dev, priv->port, (int)phv_enabled_new);
+		if (ret)
+			return ret;
+		else if (phv_enabled_new)
+			priv->pflags |= MLX4_EN_PRIV_FLAGS_PHV;
+		else
+			priv->pflags &= ~MLX4_EN_PRIV_FLAGS_PHV;
+		en_info(priv, "PHV bit %s\n",
+			phv_enabled_new ?  "Enabled" : "Disabled");
+	}
+	return 0;
+}
+
+static u32 mlx4_en_get_priv_flags(struct net_device *dev)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+
+	return priv->pflags;
+}
+
+static int mlx4_en_get_tunable(struct net_device *dev,
+			       const struct ethtool_tunable *tuna,
+			       void *data)
+{
+	const struct mlx4_en_priv *priv = netdev_priv(dev);
+	int ret = 0;
+
+	switch (tuna->id) {
+	case ETHTOOL_TX_COPYBREAK:
+		*(u32 *)data = priv->prof->inline_thold;
+		break;
+	default:
+		ret = -EINVAL;
+		break;
+	}
+
+	return ret;
+}
+
+static int mlx4_en_set_tunable(struct net_device *dev,
+			       const struct ethtool_tunable *tuna,
+			       const void *data)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	int val, ret = 0;
+
+	switch (tuna->id) {
+	case ETHTOOL_TX_COPYBREAK:
+		val = *(u32 *)data;
+		if (val < MIN_PKT_LEN || val > MAX_INLINE)
+			ret = -EINVAL;
+		else
+			priv->prof->inline_thold = val;
+		break;
+	default:
+		ret = -EINVAL;
+		break;
+	}
+
+	return ret;
+}
+
+static int mlx4_en_get_module_info(struct net_device *dev,
+				   struct ethtool_modinfo *modinfo)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	struct mlx4_en_dev *mdev = priv->mdev;
+	int ret;
+	u8 data[4];
+
+	/* Read first 2 bytes to get Module & REV ID */
+	ret = mlx4_get_module_info(mdev->dev, priv->port,
+				   0/*offset*/, 2/*size*/, data);
+	if (ret < 2)
+		return -EIO;
+
+	switch (data[0] /* identifier */) {
+	case MLX4_MODULE_ID_QSFP:
+		modinfo->type = ETH_MODULE_SFF_8436;
+		modinfo->eeprom_len = ETH_MODULE_SFF_8436_LEN;
+		break;
+	case MLX4_MODULE_ID_QSFP_PLUS:
+		if (data[1] >= 0x3) { /* revision id */
+			modinfo->type = ETH_MODULE_SFF_8636;
+			modinfo->eeprom_len = ETH_MODULE_SFF_8636_LEN;
+		} else {
+			modinfo->type = ETH_MODULE_SFF_8436;
+			modinfo->eeprom_len = ETH_MODULE_SFF_8436_LEN;
+		}
+		break;
+	case MLX4_MODULE_ID_QSFP28:
+		modinfo->type = ETH_MODULE_SFF_8636;
+		modinfo->eeprom_len = ETH_MODULE_SFF_8636_LEN;
+		break;
+	case MLX4_MODULE_ID_SFP:
+		modinfo->type = ETH_MODULE_SFF_8472;
+		modinfo->eeprom_len = ETH_MODULE_SFF_8472_LEN;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int mlx4_en_get_module_eeprom(struct net_device *dev,
+				     struct ethtool_eeprom *ee,
+				     u8 *data)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	struct mlx4_en_dev *mdev = priv->mdev;
+	int offset = ee->offset;
+	int i = 0, ret;
+
+	if (ee->len == 0)
+		return -EINVAL;
+
+	memset(data, 0, ee->len);
+
+	while (i < ee->len) {
+		en_dbg(DRV, priv,
+		       "mlx4_get_module_info i(%d) offset(%d) len(%d)\n",
+		       i, offset, ee->len - i);
+
+		ret = mlx4_get_module_info(mdev->dev, priv->port,
+					   offset, ee->len - i, data + i);
+
+		if (!ret) /* Done reading */
+			return 0;
+
+		if (ret < 0) {
+			en_err(priv,
+			       "mlx4_get_module_info i(%d) offset(%d) bytes_to_read(%d) - FAILED (0x%x)\n",
+			       i, offset, ee->len - i, ret);
+			return ret;
+		}
+
+		i += ret;
+		offset += ret;
+	}
+	return 0;
+}
+
+static int mlx4_en_set_phys_id(struct net_device *dev,
+			       enum ethtool_phys_id_state state)
+{
+	int err;
+	u16 beacon_duration;
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	struct mlx4_en_dev *mdev = priv->mdev;
+
+	if (!(mdev->dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_PORT_BEACON))
+		return -EOPNOTSUPP;
+
+	switch (state) {
+	case ETHTOOL_ID_ACTIVE:
+		beacon_duration = PORT_BEACON_MAX_LIMIT;
+		break;
+	case ETHTOOL_ID_INACTIVE:
+		beacon_duration = 0;
+		break;
+	default:
+		return -EOPNOTSUPP;
+	}
+
+	err = mlx4_SET_PORT_BEACON(mdev->dev, priv->port, beacon_duration);
+	return err;
+}
+
+const struct ethtool_ops mlx4_en_ethtool_ops = {
+	.supported_coalesce_params = ETHTOOL_COALESCE_USECS |
+				     ETHTOOL_COALESCE_MAX_FRAMES |
+				     ETHTOOL_COALESCE_TX_MAX_FRAMES_IRQ |
+				     ETHTOOL_COALESCE_PKT_RATE_RX_USECS,
+	.get_drvinfo = mlx4_en_get_drvinfo,
+	.get_link_ksettings = mlx4_en_get_link_ksettings,
+	.set_link_ksettings = mlx4_en_set_link_ksettings,
+	.get_link = ethtool_op_get_link,
+	.get_strings = mlx4_en_get_strings,
+	.get_sset_count = mlx4_en_get_sset_count,
+	.get_ethtool_stats = mlx4_en_get_ethtool_stats,
+	.self_test = mlx4_en_self_test,
+	.set_phys_id = mlx4_en_set_phys_id,
+	.get_wol = mlx4_en_get_wol,
+	.set_wol = mlx4_en_set_wol,
+	.get_msglevel = mlx4_en_get_msglevel,
+	.set_msglevel = mlx4_en_set_msglevel,
+	.get_coalesce = mlx4_en_get_coalesce,
+	.set_coalesce = mlx4_en_set_coalesce,
+	.get_pause_stats = mlx4_en_get_pause_stats,
+	.get_pauseparam = mlx4_en_get_pauseparam,
+	.set_pauseparam = mlx4_en_set_pauseparam,
+	.get_ringparam = mlx4_en_get_ringparam,
+	.set_ringparam = mlx4_en_set_ringparam,
+	.get_rxnfc = mlx4_en_get_rxnfc,
+	.set_rxnfc = mlx4_en_set_rxnfc,
+	.get_rxfh_indir_size = mlx4_en_get_rxfh_indir_size,
+	.get_rxfh_key_size = mlx4_en_get_rxfh_key_size,
+	.get_rxfh = mlx4_en_get_rxfh,
+	.set_rxfh = mlx4_en_set_rxfh,
+	.get_channels = mlx4_en_get_channels,
+	.set_channels = mlx4_en_set_channels,
+	.get_ts_info = mlx4_en_get_ts_info,
+	.set_priv_flags = mlx4_en_set_priv_flags,
+	.get_priv_flags = mlx4_en_get_priv_flags,
+	.get_tunable		= mlx4_en_get_tunable,
+	.set_tunable		= mlx4_en_set_tunable,
+	.get_module_info = mlx4_en_get_module_info,
+	.get_module_eeprom = mlx4_en_get_module_eeprom
+};
+
+
+
+
+
diff --git a/drivers/net/ethernet/mellanox/mlx4/en_main.c b/drivers/net/ethernet/mellanox/mlx4/en_main.c
new file mode 100644
index 000000000..f1259bdb1
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx4/en_main.c
@@ -0,0 +1,398 @@
+/*
+ * Copyright (c) 2007 Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+
+#include <linux/cpumask.h>
+#include <linux/module.h>
+#include <linux/delay.h>
+#include <linux/netdevice.h>
+#include <linux/slab.h>
+
+#include <linux/mlx4/driver.h>
+#include <linux/mlx4/device.h>
+#include <linux/mlx4/cmd.h>
+
+#include "mlx4_en.h"
+
+MODULE_AUTHOR("Liran Liss, Yevgeny Petrilin");
+MODULE_DESCRIPTION("Mellanox ConnectX HCA Ethernet driver");
+MODULE_LICENSE("Dual BSD/GPL");
+MODULE_VERSION(DRV_VERSION);
+
+static const char mlx4_en_version[] =
+	DRV_NAME ": Mellanox ConnectX HCA Ethernet driver v"
+	DRV_VERSION "\n";
+
+#define MLX4_EN_PARM_INT(X, def_val, desc) \
+	static unsigned int X = def_val;\
+	module_param(X , uint, 0444); \
+	MODULE_PARM_DESC(X, desc);
+
+
+/*
+ * Device scope module parameters
+ */
+
+/* Enable RSS UDP traffic */
+MLX4_EN_PARM_INT(udp_rss, 1,
+		 "Enable RSS for incoming UDP traffic or disabled (0)");
+
+/* Priority pausing */
+MLX4_EN_PARM_INT(pfctx, 0, "Priority based Flow Control policy on TX[7:0]."
+			   " Per priority bit mask");
+MLX4_EN_PARM_INT(pfcrx, 0, "Priority based Flow Control policy on RX[7:0]."
+			   " Per priority bit mask");
+
+MLX4_EN_PARM_INT(inline_thold, MAX_INLINE,
+		 "Threshold for using inline data (range: 17-104, default: 104)");
+
+#define MAX_PFC_TX     0xff
+#define MAX_PFC_RX     0xff
+
+void en_print(const char *level, const struct mlx4_en_priv *priv,
+	      const char *format, ...)
+{
+	va_list args;
+	struct va_format vaf;
+
+	va_start(args, format);
+
+	vaf.fmt = format;
+	vaf.va = &args;
+	if (priv->registered)
+		printk("%s%s: %s: %pV",
+		       level, DRV_NAME, priv->dev->name, &vaf);
+	else
+		printk("%s%s: %s: Port %d: %pV",
+		       level, DRV_NAME, dev_name(&priv->mdev->pdev->dev),
+		       priv->port, &vaf);
+	va_end(args);
+}
+
+void mlx4_en_update_loopback_state(struct net_device *dev,
+				   netdev_features_t features)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+
+	if (features & NETIF_F_LOOPBACK)
+		priv->ctrl_flags |= cpu_to_be32(MLX4_WQE_CTRL_FORCE_LOOPBACK);
+	else
+		priv->ctrl_flags &= cpu_to_be32(~MLX4_WQE_CTRL_FORCE_LOOPBACK);
+
+	priv->flags &= ~(MLX4_EN_FLAG_RX_FILTER_NEEDED|
+			MLX4_EN_FLAG_ENABLE_HW_LOOPBACK);
+
+	/* Drop the packet if SRIOV is not enabled
+	 * and not performing the selftest or flb disabled
+	 */
+	if (mlx4_is_mfunc(priv->mdev->dev) &&
+	    !(features & NETIF_F_LOOPBACK) && !priv->validate_loopback)
+		priv->flags |= MLX4_EN_FLAG_RX_FILTER_NEEDED;
+
+	/* Set dmac in Tx WQE if we are in SRIOV mode or if loopback selftest
+	 * is requested
+	 */
+	if (mlx4_is_mfunc(priv->mdev->dev) || priv->validate_loopback)
+		priv->flags |= MLX4_EN_FLAG_ENABLE_HW_LOOPBACK;
+
+	mutex_lock(&priv->mdev->state_lock);
+	if ((priv->mdev->dev->caps.flags2 &
+	     MLX4_DEV_CAP_FLAG2_UPDATE_QP_SRC_CHECK_LB) &&
+	    priv->rss_map.indir_qp && priv->rss_map.indir_qp->qpn) {
+		int i;
+		int err = 0;
+		int loopback = !!(features & NETIF_F_LOOPBACK);
+
+		for (i = 0; i < priv->rx_ring_num; i++) {
+			int ret;
+
+			ret = mlx4_en_change_mcast_lb(priv,
+						      &priv->rss_map.qps[i],
+						      loopback);
+			if (!err)
+				err = ret;
+		}
+		if (err)
+			mlx4_warn(priv->mdev, "failed to change mcast loopback\n");
+	}
+	mutex_unlock(&priv->mdev->state_lock);
+}
+
+static void mlx4_en_get_profile(struct mlx4_en_dev *mdev)
+{
+	struct mlx4_en_profile *params = &mdev->profile;
+	int i;
+
+	params->udp_rss = udp_rss;
+	params->max_num_tx_rings_p_up = mlx4_low_memory_profile() ?
+		MLX4_EN_MIN_TX_RING_P_UP :
+		min_t(int, num_online_cpus(), MLX4_EN_MAX_TX_RING_P_UP);
+
+	if (params->udp_rss && !(mdev->dev->caps.flags
+					& MLX4_DEV_CAP_FLAG_UDP_RSS)) {
+		mlx4_warn(mdev, "UDP RSS is not supported on this device\n");
+		params->udp_rss = 0;
+	}
+	for (i = 1; i <= MLX4_MAX_PORTS; i++) {
+		params->prof[i].rx_pause = !(pfcrx || pfctx);
+		params->prof[i].rx_ppp = pfcrx;
+		params->prof[i].tx_pause = !(pfcrx || pfctx);
+		params->prof[i].tx_ppp = pfctx;
+		if (mlx4_low_memory_profile()) {
+			params->prof[i].tx_ring_size = MLX4_EN_MIN_TX_SIZE;
+			params->prof[i].rx_ring_size = MLX4_EN_MIN_RX_SIZE;
+		} else {
+			params->prof[i].tx_ring_size = MLX4_EN_DEF_TX_RING_SIZE;
+			params->prof[i].rx_ring_size = MLX4_EN_DEF_RX_RING_SIZE;
+		}
+		params->prof[i].num_up = MLX4_EN_NUM_UP_LOW;
+		params->prof[i].num_tx_rings_p_up = params->max_num_tx_rings_p_up;
+		params->prof[i].tx_ring_num[TX] = params->max_num_tx_rings_p_up *
+			params->prof[i].num_up;
+		params->prof[i].rss_rings = 0;
+		params->prof[i].inline_thold = inline_thold;
+	}
+}
+
+static void *mlx4_en_get_netdev(struct mlx4_dev *dev, void *ctx, u8 port)
+{
+	struct mlx4_en_dev *endev = ctx;
+
+	return endev->pndev[port];
+}
+
+static void mlx4_en_event(struct mlx4_dev *dev, void *endev_ptr,
+			  enum mlx4_dev_event event, unsigned long port)
+{
+	struct mlx4_en_dev *mdev = (struct mlx4_en_dev *) endev_ptr;
+	struct mlx4_en_priv *priv;
+
+	switch (event) {
+	case MLX4_DEV_EVENT_PORT_UP:
+	case MLX4_DEV_EVENT_PORT_DOWN:
+		if (!mdev->pndev[port])
+			return;
+		priv = netdev_priv(mdev->pndev[port]);
+		/* To prevent races, we poll the link state in a separate
+		  task rather than changing it here */
+		priv->link_state = event;
+		queue_work(mdev->workqueue, &priv->linkstate_task);
+		break;
+
+	case MLX4_DEV_EVENT_CATASTROPHIC_ERROR:
+		mlx4_err(mdev, "Internal error detected, restarting device\n");
+		break;
+
+	case MLX4_DEV_EVENT_SLAVE_INIT:
+	case MLX4_DEV_EVENT_SLAVE_SHUTDOWN:
+		break;
+	default:
+		if (port < 1 || port > dev->caps.num_ports ||
+		    !mdev->pndev[port])
+			return;
+		mlx4_warn(mdev, "Unhandled event %d for port %d\n", event,
+			  (int) port);
+	}
+}
+
+static void mlx4_en_remove(struct mlx4_dev *dev, void *endev_ptr)
+{
+	struct mlx4_en_dev *mdev = endev_ptr;
+	int i;
+
+	mutex_lock(&mdev->state_lock);
+	mdev->device_up = false;
+	mutex_unlock(&mdev->state_lock);
+
+	mlx4_foreach_port(i, dev, MLX4_PORT_TYPE_ETH)
+		if (mdev->pndev[i])
+			mlx4_en_destroy_netdev(mdev->pndev[i]);
+
+	destroy_workqueue(mdev->workqueue);
+	(void) mlx4_mr_free(dev, &mdev->mr);
+	iounmap(mdev->uar_map);
+	mlx4_uar_free(dev, &mdev->priv_uar);
+	mlx4_pd_free(dev, mdev->priv_pdn);
+	if (mdev->nb.notifier_call)
+		unregister_netdevice_notifier(&mdev->nb);
+	kfree(mdev);
+}
+
+static void mlx4_en_activate(struct mlx4_dev *dev, void *ctx)
+{
+	int i;
+	struct mlx4_en_dev *mdev = ctx;
+
+	/* Create a netdev for each port */
+	mlx4_foreach_port(i, dev, MLX4_PORT_TYPE_ETH) {
+		mlx4_info(mdev, "Activating port:%d\n", i);
+		if (mlx4_en_init_netdev(mdev, i, &mdev->profile.prof[i]))
+			mdev->pndev[i] = NULL;
+	}
+
+	/* register notifier */
+	mdev->nb.notifier_call = mlx4_en_netdev_event;
+	if (register_netdevice_notifier(&mdev->nb)) {
+		mdev->nb.notifier_call = NULL;
+		mlx4_err(mdev, "Failed to create notifier\n");
+	}
+}
+
+static void *mlx4_en_add(struct mlx4_dev *dev)
+{
+	struct mlx4_en_dev *mdev;
+	int i;
+
+	printk_once(KERN_INFO "%s", mlx4_en_version);
+
+	mdev = kzalloc(sizeof(*mdev), GFP_KERNEL);
+	if (!mdev)
+		goto err_free_res;
+
+	if (mlx4_pd_alloc(dev, &mdev->priv_pdn))
+		goto err_free_dev;
+
+	if (mlx4_uar_alloc(dev, &mdev->priv_uar))
+		goto err_pd;
+
+	mdev->uar_map = ioremap((phys_addr_t) mdev->priv_uar.pfn << PAGE_SHIFT,
+				PAGE_SIZE);
+	if (!mdev->uar_map)
+		goto err_uar;
+	spin_lock_init(&mdev->uar_lock);
+
+	mdev->dev = dev;
+	mdev->dma_device = &dev->persist->pdev->dev;
+	mdev->pdev = dev->persist->pdev;
+	mdev->device_up = false;
+
+	mdev->LSO_support = !!(dev->caps.flags & (1 << 15));
+	if (!mdev->LSO_support)
+		mlx4_warn(mdev, "LSO not supported, please upgrade to later FW version to enable LSO\n");
+
+	if (mlx4_mr_alloc(mdev->dev, mdev->priv_pdn, 0, ~0ull,
+			 MLX4_PERM_LOCAL_WRITE |  MLX4_PERM_LOCAL_READ,
+			 0, 0, &mdev->mr)) {
+		mlx4_err(mdev, "Failed allocating memory region\n");
+		goto err_map;
+	}
+	if (mlx4_mr_enable(mdev->dev, &mdev->mr)) {
+		mlx4_err(mdev, "Failed enabling memory region\n");
+		goto err_mr;
+	}
+
+	/* Build device profile according to supplied module parameters */
+	mlx4_en_get_profile(mdev);
+
+	/* Configure which ports to start according to module parameters */
+	mdev->port_cnt = 0;
+	mlx4_foreach_port(i, dev, MLX4_PORT_TYPE_ETH)
+		mdev->port_cnt++;
+
+	/* Set default number of RX rings*/
+	mlx4_en_set_num_rx_rings(mdev);
+
+	/* Create our own workqueue for reset/multicast tasks
+	 * Note: we cannot use the shared workqueue because of deadlocks caused
+	 *       by the rtnl lock */
+	mdev->workqueue = create_singlethread_workqueue("mlx4_en");
+	if (!mdev->workqueue)
+		goto err_mr;
+
+	/* At this stage all non-port specific tasks are complete:
+	 * mark the card state as up */
+	mutex_init(&mdev->state_lock);
+	mdev->device_up = true;
+
+	return mdev;
+
+err_mr:
+	(void) mlx4_mr_free(dev, &mdev->mr);
+err_map:
+	if (mdev->uar_map)
+		iounmap(mdev->uar_map);
+err_uar:
+	mlx4_uar_free(dev, &mdev->priv_uar);
+err_pd:
+	mlx4_pd_free(dev, mdev->priv_pdn);
+err_free_dev:
+	kfree(mdev);
+err_free_res:
+	return NULL;
+}
+
+static struct mlx4_interface mlx4_en_interface = {
+	.add		= mlx4_en_add,
+	.remove		= mlx4_en_remove,
+	.event		= mlx4_en_event,
+	.get_dev	= mlx4_en_get_netdev,
+	.protocol	= MLX4_PROT_ETH,
+	.activate	= mlx4_en_activate,
+};
+
+static void mlx4_en_verify_params(void)
+{
+	if (pfctx > MAX_PFC_TX) {
+		pr_warn("mlx4_en: WARNING: illegal module parameter pfctx 0x%x - should be in range 0-0x%x, will be changed to default (0)\n",
+			pfctx, MAX_PFC_TX);
+		pfctx = 0;
+	}
+
+	if (pfcrx > MAX_PFC_RX) {
+		pr_warn("mlx4_en: WARNING: illegal module parameter pfcrx 0x%x - should be in range 0-0x%x, will be changed to default (0)\n",
+			pfcrx, MAX_PFC_RX);
+		pfcrx = 0;
+	}
+
+	if (inline_thold < MIN_PKT_LEN || inline_thold > MAX_INLINE) {
+		pr_warn("mlx4_en: WARNING: illegal module parameter inline_thold %d - should be in range %d-%d, will be changed to default (%d)\n",
+			inline_thold, MIN_PKT_LEN, MAX_INLINE, MAX_INLINE);
+		inline_thold = MAX_INLINE;
+	}
+}
+
+static int __init mlx4_en_init(void)
+{
+	mlx4_en_verify_params();
+	mlx4_en_init_ptys2ethtool_map();
+
+	return mlx4_register_interface(&mlx4_en_interface);
+}
+
+static void __exit mlx4_en_cleanup(void)
+{
+	mlx4_unregister_interface(&mlx4_en_interface);
+}
+
+module_init(mlx4_en_init);
+module_exit(mlx4_en_cleanup);
+
diff --git a/drivers/net/ethernet/mellanox/mlx4/en_netdev.c b/drivers/net/ethernet/mellanox/mlx4/en_netdev.c
new file mode 100644
index 000000000..ca4b93a01
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx4/en_netdev.c
@@ -0,0 +1,3584 @@
+/*
+ * Copyright (c) 2007 Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+
+#include <linux/bpf.h>
+#include <linux/etherdevice.h>
+#include <linux/filter.h>
+#include <linux/tcp.h>
+#include <linux/if_vlan.h>
+#include <linux/delay.h>
+#include <linux/slab.h>
+#include <linux/hash.h>
+#include <net/ip.h>
+#include <net/vxlan.h>
+#include <net/devlink.h>
+
+#include <linux/mlx4/driver.h>
+#include <linux/mlx4/device.h>
+#include <linux/mlx4/cmd.h>
+#include <linux/mlx4/cq.h>
+
+#include "mlx4_en.h"
+#include "en_port.h"
+
+#define MLX4_EN_MAX_XDP_MTU ((int)(PAGE_SIZE - ETH_HLEN - (2 * VLAN_HLEN) - \
+				XDP_PACKET_HEADROOM -			    \
+				SKB_DATA_ALIGN(sizeof(struct skb_shared_info))))
+
+int mlx4_en_setup_tc(struct net_device *dev, u8 up)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	int i;
+	unsigned int offset = 0;
+
+	if (up && up != MLX4_EN_NUM_UP_HIGH)
+		return -EINVAL;
+
+	netdev_set_num_tc(dev, up);
+	netif_set_real_num_tx_queues(dev, priv->tx_ring_num[TX]);
+	/* Partition Tx queues evenly amongst UP's */
+	for (i = 0; i < up; i++) {
+		netdev_set_tc_queue(dev, i, priv->num_tx_rings_p_up, offset);
+		offset += priv->num_tx_rings_p_up;
+	}
+
+#ifdef CONFIG_MLX4_EN_DCB
+	if (!mlx4_is_slave(priv->mdev->dev)) {
+		if (up) {
+			if (priv->dcbx_cap)
+				priv->flags |= MLX4_EN_FLAG_DCB_ENABLED;
+		} else {
+			priv->flags &= ~MLX4_EN_FLAG_DCB_ENABLED;
+			priv->cee_config.pfc_state = false;
+		}
+	}
+#endif /* CONFIG_MLX4_EN_DCB */
+
+	return 0;
+}
+
+int mlx4_en_alloc_tx_queue_per_tc(struct net_device *dev, u8 tc)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	struct mlx4_en_dev *mdev = priv->mdev;
+	struct mlx4_en_port_profile new_prof;
+	struct mlx4_en_priv *tmp;
+	int total_count;
+	int port_up = 0;
+	int err = 0;
+
+	tmp = kzalloc(sizeof(*tmp), GFP_KERNEL);
+	if (!tmp)
+		return -ENOMEM;
+
+	mutex_lock(&mdev->state_lock);
+	memcpy(&new_prof, priv->prof, sizeof(struct mlx4_en_port_profile));
+	new_prof.num_up = (tc == 0) ? MLX4_EN_NUM_UP_LOW :
+				      MLX4_EN_NUM_UP_HIGH;
+	new_prof.tx_ring_num[TX] = new_prof.num_tx_rings_p_up *
+				   new_prof.num_up;
+	total_count = new_prof.tx_ring_num[TX] + new_prof.tx_ring_num[TX_XDP];
+	if (total_count > MAX_TX_RINGS) {
+		err = -EINVAL;
+		en_err(priv,
+		       "Total number of TX and XDP rings (%d) exceeds the maximum supported (%d)\n",
+		       total_count, MAX_TX_RINGS);
+		goto out;
+	}
+	err = mlx4_en_try_alloc_resources(priv, tmp, &new_prof, true);
+	if (err)
+		goto out;
+
+	if (priv->port_up) {
+		port_up = 1;
+		mlx4_en_stop_port(dev, 1);
+	}
+
+	mlx4_en_safe_replace_resources(priv, tmp);
+	if (port_up) {
+		err = mlx4_en_start_port(dev);
+		if (err) {
+			en_err(priv, "Failed starting port for setup TC\n");
+			goto out;
+		}
+	}
+
+	err = mlx4_en_setup_tc(dev, tc);
+out:
+	mutex_unlock(&mdev->state_lock);
+	kfree(tmp);
+	return err;
+}
+
+static int __mlx4_en_setup_tc(struct net_device *dev, enum tc_setup_type type,
+			      void *type_data)
+{
+	struct tc_mqprio_qopt *mqprio = type_data;
+
+	if (type != TC_SETUP_QDISC_MQPRIO)
+		return -EOPNOTSUPP;
+
+	if (mqprio->num_tc && mqprio->num_tc != MLX4_EN_NUM_UP_HIGH)
+		return -EINVAL;
+
+	mqprio->hw = TC_MQPRIO_HW_OFFLOAD_TCS;
+
+	return mlx4_en_alloc_tx_queue_per_tc(dev, mqprio->num_tc);
+}
+
+#ifdef CONFIG_RFS_ACCEL
+
+struct mlx4_en_filter {
+	struct list_head next;
+	struct work_struct work;
+
+	u8     ip_proto;
+	__be32 src_ip;
+	__be32 dst_ip;
+	__be16 src_port;
+	__be16 dst_port;
+
+	int rxq_index;
+	struct mlx4_en_priv *priv;
+	u32 flow_id;			/* RFS infrastructure id */
+	int id;				/* mlx4_en driver id */
+	u64 reg_id;			/* Flow steering API id */
+	u8 activated;			/* Used to prevent expiry before filter
+					 * is attached
+					 */
+	struct hlist_node filter_chain;
+};
+
+static void mlx4_en_filter_rfs_expire(struct mlx4_en_priv *priv);
+
+static enum mlx4_net_trans_rule_id mlx4_ip_proto_to_trans_rule_id(u8 ip_proto)
+{
+	switch (ip_proto) {
+	case IPPROTO_UDP:
+		return MLX4_NET_TRANS_RULE_ID_UDP;
+	case IPPROTO_TCP:
+		return MLX4_NET_TRANS_RULE_ID_TCP;
+	default:
+		return MLX4_NET_TRANS_RULE_NUM;
+	}
+};
+
+/* Must not acquire state_lock, as its corresponding work_sync
+ * is done under it.
+ */
+static void mlx4_en_filter_work(struct work_struct *work)
+{
+	struct mlx4_en_filter *filter = container_of(work,
+						     struct mlx4_en_filter,
+						     work);
+	struct mlx4_en_priv *priv = filter->priv;
+	struct mlx4_spec_list spec_tcp_udp = {
+		.id = mlx4_ip_proto_to_trans_rule_id(filter->ip_proto),
+		{
+			.tcp_udp = {
+				.dst_port = filter->dst_port,
+				.dst_port_msk = (__force __be16)-1,
+				.src_port = filter->src_port,
+				.src_port_msk = (__force __be16)-1,
+			},
+		},
+	};
+	struct mlx4_spec_list spec_ip = {
+		.id = MLX4_NET_TRANS_RULE_ID_IPV4,
+		{
+			.ipv4 = {
+				.dst_ip = filter->dst_ip,
+				.dst_ip_msk = (__force __be32)-1,
+				.src_ip = filter->src_ip,
+				.src_ip_msk = (__force __be32)-1,
+			},
+		},
+	};
+	struct mlx4_spec_list spec_eth = {
+		.id = MLX4_NET_TRANS_RULE_ID_ETH,
+	};
+	struct mlx4_net_trans_rule rule = {
+		.list = LIST_HEAD_INIT(rule.list),
+		.queue_mode = MLX4_NET_TRANS_Q_LIFO,
+		.exclusive = 1,
+		.allow_loopback = 1,
+		.promisc_mode = MLX4_FS_REGULAR,
+		.port = priv->port,
+		.priority = MLX4_DOMAIN_RFS,
+	};
+	int rc;
+	__be64 mac_mask = cpu_to_be64(MLX4_MAC_MASK << 16);
+
+	if (spec_tcp_udp.id >= MLX4_NET_TRANS_RULE_NUM) {
+		en_warn(priv, "RFS: ignoring unsupported ip protocol (%d)\n",
+			filter->ip_proto);
+		goto ignore;
+	}
+	list_add_tail(&spec_eth.list, &rule.list);
+	list_add_tail(&spec_ip.list, &rule.list);
+	list_add_tail(&spec_tcp_udp.list, &rule.list);
+
+	rule.qpn = priv->rss_map.qps[filter->rxq_index].qpn;
+	memcpy(spec_eth.eth.dst_mac, priv->dev->dev_addr, ETH_ALEN);
+	memcpy(spec_eth.eth.dst_mac_msk, &mac_mask, ETH_ALEN);
+
+	filter->activated = 0;
+
+	if (filter->reg_id) {
+		rc = mlx4_flow_detach(priv->mdev->dev, filter->reg_id);
+		if (rc && rc != -ENOENT)
+			en_err(priv, "Error detaching flow. rc = %d\n", rc);
+	}
+
+	rc = mlx4_flow_attach(priv->mdev->dev, &rule, &filter->reg_id);
+	if (rc)
+		en_err(priv, "Error attaching flow. err = %d\n", rc);
+
+ignore:
+	mlx4_en_filter_rfs_expire(priv);
+
+	filter->activated = 1;
+}
+
+static inline struct hlist_head *
+filter_hash_bucket(struct mlx4_en_priv *priv, __be32 src_ip, __be32 dst_ip,
+		   __be16 src_port, __be16 dst_port)
+{
+	unsigned long l;
+	int bucket_idx;
+
+	l = (__force unsigned long)src_port |
+	    ((__force unsigned long)dst_port << 2);
+	l ^= (__force unsigned long)(src_ip ^ dst_ip);
+
+	bucket_idx = hash_long(l, MLX4_EN_FILTER_HASH_SHIFT);
+
+	return &priv->filter_hash[bucket_idx];
+}
+
+static struct mlx4_en_filter *
+mlx4_en_filter_alloc(struct mlx4_en_priv *priv, int rxq_index, __be32 src_ip,
+		     __be32 dst_ip, u8 ip_proto, __be16 src_port,
+		     __be16 dst_port, u32 flow_id)
+{
+	struct mlx4_en_filter *filter = NULL;
+
+	filter = kzalloc(sizeof(struct mlx4_en_filter), GFP_ATOMIC);
+	if (!filter)
+		return NULL;
+
+	filter->priv = priv;
+	filter->rxq_index = rxq_index;
+	INIT_WORK(&filter->work, mlx4_en_filter_work);
+
+	filter->src_ip = src_ip;
+	filter->dst_ip = dst_ip;
+	filter->ip_proto = ip_proto;
+	filter->src_port = src_port;
+	filter->dst_port = dst_port;
+
+	filter->flow_id = flow_id;
+
+	filter->id = priv->last_filter_id++ % RPS_NO_FILTER;
+
+	list_add_tail(&filter->next, &priv->filters);
+	hlist_add_head(&filter->filter_chain,
+		       filter_hash_bucket(priv, src_ip, dst_ip, src_port,
+					  dst_port));
+
+	return filter;
+}
+
+static void mlx4_en_filter_free(struct mlx4_en_filter *filter)
+{
+	struct mlx4_en_priv *priv = filter->priv;
+	int rc;
+
+	list_del(&filter->next);
+
+	rc = mlx4_flow_detach(priv->mdev->dev, filter->reg_id);
+	if (rc && rc != -ENOENT)
+		en_err(priv, "Error detaching flow. rc = %d\n", rc);
+
+	kfree(filter);
+}
+
+static inline struct mlx4_en_filter *
+mlx4_en_filter_find(struct mlx4_en_priv *priv, __be32 src_ip, __be32 dst_ip,
+		    u8 ip_proto, __be16 src_port, __be16 dst_port)
+{
+	struct mlx4_en_filter *filter;
+	struct mlx4_en_filter *ret = NULL;
+
+	hlist_for_each_entry(filter,
+			     filter_hash_bucket(priv, src_ip, dst_ip,
+						src_port, dst_port),
+			     filter_chain) {
+		if (filter->src_ip == src_ip &&
+		    filter->dst_ip == dst_ip &&
+		    filter->ip_proto == ip_proto &&
+		    filter->src_port == src_port &&
+		    filter->dst_port == dst_port) {
+			ret = filter;
+			break;
+		}
+	}
+
+	return ret;
+}
+
+static int
+mlx4_en_filter_rfs(struct net_device *net_dev, const struct sk_buff *skb,
+		   u16 rxq_index, u32 flow_id)
+{
+	struct mlx4_en_priv *priv = netdev_priv(net_dev);
+	struct mlx4_en_filter *filter;
+	const struct iphdr *ip;
+	const __be16 *ports;
+	u8 ip_proto;
+	__be32 src_ip;
+	__be32 dst_ip;
+	__be16 src_port;
+	__be16 dst_port;
+	int nhoff = skb_network_offset(skb);
+	int ret = 0;
+
+	if (skb->encapsulation)
+		return -EPROTONOSUPPORT;
+
+	if (skb->protocol != htons(ETH_P_IP))
+		return -EPROTONOSUPPORT;
+
+	ip = (const struct iphdr *)(skb->data + nhoff);
+	if (ip_is_fragment(ip))
+		return -EPROTONOSUPPORT;
+
+	if ((ip->protocol != IPPROTO_TCP) && (ip->protocol != IPPROTO_UDP))
+		return -EPROTONOSUPPORT;
+	ports = (const __be16 *)(skb->data + nhoff + 4 * ip->ihl);
+
+	ip_proto = ip->protocol;
+	src_ip = ip->saddr;
+	dst_ip = ip->daddr;
+	src_port = ports[0];
+	dst_port = ports[1];
+
+	spin_lock_bh(&priv->filters_lock);
+	filter = mlx4_en_filter_find(priv, src_ip, dst_ip, ip_proto,
+				     src_port, dst_port);
+	if (filter) {
+		if (filter->rxq_index == rxq_index)
+			goto out;
+
+		filter->rxq_index = rxq_index;
+	} else {
+		filter = mlx4_en_filter_alloc(priv, rxq_index,
+					      src_ip, dst_ip, ip_proto,
+					      src_port, dst_port, flow_id);
+		if (!filter) {
+			ret = -ENOMEM;
+			goto err;
+		}
+	}
+
+	queue_work(priv->mdev->workqueue, &filter->work);
+
+out:
+	ret = filter->id;
+err:
+	spin_unlock_bh(&priv->filters_lock);
+
+	return ret;
+}
+
+void mlx4_en_cleanup_filters(struct mlx4_en_priv *priv)
+{
+	struct mlx4_en_filter *filter, *tmp;
+	LIST_HEAD(del_list);
+
+	spin_lock_bh(&priv->filters_lock);
+	list_for_each_entry_safe(filter, tmp, &priv->filters, next) {
+		list_move(&filter->next, &del_list);
+		hlist_del(&filter->filter_chain);
+	}
+	spin_unlock_bh(&priv->filters_lock);
+
+	list_for_each_entry_safe(filter, tmp, &del_list, next) {
+		cancel_work_sync(&filter->work);
+		mlx4_en_filter_free(filter);
+	}
+}
+
+static void mlx4_en_filter_rfs_expire(struct mlx4_en_priv *priv)
+{
+	struct mlx4_en_filter *filter = NULL, *tmp, *last_filter = NULL;
+	LIST_HEAD(del_list);
+	int i = 0;
+
+	spin_lock_bh(&priv->filters_lock);
+	list_for_each_entry_safe(filter, tmp, &priv->filters, next) {
+		if (i > MLX4_EN_FILTER_EXPIRY_QUOTA)
+			break;
+
+		if (filter->activated &&
+		    !work_pending(&filter->work) &&
+		    rps_may_expire_flow(priv->dev,
+					filter->rxq_index, filter->flow_id,
+					filter->id)) {
+			list_move(&filter->next, &del_list);
+			hlist_del(&filter->filter_chain);
+		} else
+			last_filter = filter;
+
+		i++;
+	}
+
+	if (last_filter && (&last_filter->next != priv->filters.next))
+		list_move(&priv->filters, &last_filter->next);
+
+	spin_unlock_bh(&priv->filters_lock);
+
+	list_for_each_entry_safe(filter, tmp, &del_list, next)
+		mlx4_en_filter_free(filter);
+}
+#endif
+
+static int mlx4_en_vlan_rx_add_vid(struct net_device *dev,
+				   __be16 proto, u16 vid)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	struct mlx4_en_dev *mdev = priv->mdev;
+	int err;
+	int idx;
+
+	en_dbg(HW, priv, "adding VLAN:%d\n", vid);
+
+	set_bit(vid, priv->active_vlans);
+
+	/* Add VID to port VLAN filter */
+	mutex_lock(&mdev->state_lock);
+	if (mdev->device_up && priv->port_up) {
+		err = mlx4_SET_VLAN_FLTR(mdev->dev, priv);
+		if (err) {
+			en_err(priv, "Failed configuring VLAN filter\n");
+			goto out;
+		}
+	}
+	err = mlx4_register_vlan(mdev->dev, priv->port, vid, &idx);
+	if (err)
+		en_dbg(HW, priv, "Failed adding vlan %d\n", vid);
+
+out:
+	mutex_unlock(&mdev->state_lock);
+	return err;
+}
+
+static int mlx4_en_vlan_rx_kill_vid(struct net_device *dev,
+				    __be16 proto, u16 vid)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	struct mlx4_en_dev *mdev = priv->mdev;
+	int err = 0;
+
+	en_dbg(HW, priv, "Killing VID:%d\n", vid);
+
+	clear_bit(vid, priv->active_vlans);
+
+	/* Remove VID from port VLAN filter */
+	mutex_lock(&mdev->state_lock);
+	mlx4_unregister_vlan(mdev->dev, priv->port, vid);
+
+	if (mdev->device_up && priv->port_up) {
+		err = mlx4_SET_VLAN_FLTR(mdev->dev, priv);
+		if (err)
+			en_err(priv, "Failed configuring VLAN filter\n");
+	}
+	mutex_unlock(&mdev->state_lock);
+
+	return err;
+}
+
+static void mlx4_en_u64_to_mac(struct net_device *dev, u64 src_mac)
+{
+	u8 addr[ETH_ALEN];
+
+	u64_to_ether_addr(src_mac, addr);
+	eth_hw_addr_set(dev, addr);
+}
+
+
+static int mlx4_en_tunnel_steer_add(struct mlx4_en_priv *priv,
+				    const unsigned char *addr,
+				    int qpn, u64 *reg_id)
+{
+	int err;
+
+	if (priv->mdev->dev->caps.tunnel_offload_mode != MLX4_TUNNEL_OFFLOAD_MODE_VXLAN ||
+	    priv->mdev->dev->caps.dmfs_high_steer_mode == MLX4_STEERING_DMFS_A0_STATIC)
+		return 0; /* do nothing */
+
+	err = mlx4_tunnel_steer_add(priv->mdev->dev, addr, priv->port, qpn,
+				    MLX4_DOMAIN_NIC, reg_id);
+	if (err) {
+		en_err(priv, "failed to add vxlan steering rule, err %d\n", err);
+		return err;
+	}
+	en_dbg(DRV, priv, "added vxlan steering rule, mac %pM reg_id %llx\n", addr, *reg_id);
+	return 0;
+}
+
+
+static int mlx4_en_uc_steer_add(struct mlx4_en_priv *priv,
+				const unsigned char *mac, int *qpn, u64 *reg_id)
+{
+	struct mlx4_en_dev *mdev = priv->mdev;
+	struct mlx4_dev *dev = mdev->dev;
+	int err;
+
+	switch (dev->caps.steering_mode) {
+	case MLX4_STEERING_MODE_B0: {
+		struct mlx4_qp qp;
+		u8 gid[16] = {0};
+
+		qp.qpn = *qpn;
+		memcpy(&gid[10], mac, ETH_ALEN);
+		gid[5] = priv->port;
+
+		err = mlx4_unicast_attach(dev, &qp, gid, 0, MLX4_PROT_ETH);
+		break;
+	}
+	case MLX4_STEERING_MODE_DEVICE_MANAGED: {
+		struct mlx4_spec_list spec_eth = { {NULL} };
+		__be64 mac_mask = cpu_to_be64(MLX4_MAC_MASK << 16);
+
+		struct mlx4_net_trans_rule rule = {
+			.queue_mode = MLX4_NET_TRANS_Q_FIFO,
+			.exclusive = 0,
+			.allow_loopback = 1,
+			.promisc_mode = MLX4_FS_REGULAR,
+			.priority = MLX4_DOMAIN_NIC,
+		};
+
+		rule.port = priv->port;
+		rule.qpn = *qpn;
+		INIT_LIST_HEAD(&rule.list);
+
+		spec_eth.id = MLX4_NET_TRANS_RULE_ID_ETH;
+		memcpy(spec_eth.eth.dst_mac, mac, ETH_ALEN);
+		memcpy(spec_eth.eth.dst_mac_msk, &mac_mask, ETH_ALEN);
+		list_add_tail(&spec_eth.list, &rule.list);
+
+		err = mlx4_flow_attach(dev, &rule, reg_id);
+		break;
+	}
+	default:
+		return -EINVAL;
+	}
+	if (err)
+		en_warn(priv, "Failed Attaching Unicast\n");
+
+	return err;
+}
+
+static void mlx4_en_uc_steer_release(struct mlx4_en_priv *priv,
+				     const unsigned char *mac,
+				     int qpn, u64 reg_id)
+{
+	struct mlx4_en_dev *mdev = priv->mdev;
+	struct mlx4_dev *dev = mdev->dev;
+
+	switch (dev->caps.steering_mode) {
+	case MLX4_STEERING_MODE_B0: {
+		struct mlx4_qp qp;
+		u8 gid[16] = {0};
+
+		qp.qpn = qpn;
+		memcpy(&gid[10], mac, ETH_ALEN);
+		gid[5] = priv->port;
+
+		mlx4_unicast_detach(dev, &qp, gid, MLX4_PROT_ETH);
+		break;
+	}
+	case MLX4_STEERING_MODE_DEVICE_MANAGED: {
+		mlx4_flow_detach(dev, reg_id);
+		break;
+	}
+	default:
+		en_err(priv, "Invalid steering mode.\n");
+	}
+}
+
+static int mlx4_en_get_qp(struct mlx4_en_priv *priv)
+{
+	struct mlx4_en_dev *mdev = priv->mdev;
+	struct mlx4_dev *dev = mdev->dev;
+	int index = 0;
+	int err = 0;
+	int *qpn = &priv->base_qpn;
+	u64 mac = ether_addr_to_u64(priv->dev->dev_addr);
+
+	en_dbg(DRV, priv, "Registering MAC: %pM for adding\n",
+	       priv->dev->dev_addr);
+	index = mlx4_register_mac(dev, priv->port, mac);
+	if (index < 0) {
+		err = index;
+		en_err(priv, "Failed adding MAC: %pM\n",
+		       priv->dev->dev_addr);
+		return err;
+	}
+
+	en_info(priv, "Steering Mode %d\n", dev->caps.steering_mode);
+
+	if (dev->caps.steering_mode == MLX4_STEERING_MODE_A0) {
+		int base_qpn = mlx4_get_base_qpn(dev, priv->port);
+		*qpn = base_qpn + index;
+		return 0;
+	}
+
+	err = mlx4_qp_reserve_range(dev, 1, 1, qpn, MLX4_RESERVE_A0_QP,
+				    MLX4_RES_USAGE_DRIVER);
+	en_dbg(DRV, priv, "Reserved qp %d\n", *qpn);
+	if (err) {
+		en_err(priv, "Failed to reserve qp for mac registration\n");
+		mlx4_unregister_mac(dev, priv->port, mac);
+		return err;
+	}
+
+	return 0;
+}
+
+static void mlx4_en_put_qp(struct mlx4_en_priv *priv)
+{
+	struct mlx4_en_dev *mdev = priv->mdev;
+	struct mlx4_dev *dev = mdev->dev;
+	int qpn = priv->base_qpn;
+
+	if (dev->caps.steering_mode == MLX4_STEERING_MODE_A0) {
+		u64 mac = ether_addr_to_u64(priv->dev->dev_addr);
+		en_dbg(DRV, priv, "Registering MAC: %pM for deleting\n",
+		       priv->dev->dev_addr);
+		mlx4_unregister_mac(dev, priv->port, mac);
+	} else {
+		en_dbg(DRV, priv, "Releasing qp: port %d, qpn %d\n",
+		       priv->port, qpn);
+		mlx4_qp_release_range(dev, qpn, 1);
+		priv->flags &= ~MLX4_EN_FLAG_FORCE_PROMISC;
+	}
+}
+
+static int mlx4_en_replace_mac(struct mlx4_en_priv *priv, int qpn,
+			       unsigned char *new_mac, unsigned char *prev_mac)
+{
+	struct mlx4_en_dev *mdev = priv->mdev;
+	struct mlx4_dev *dev = mdev->dev;
+	int err = 0;
+	u64 new_mac_u64 = ether_addr_to_u64(new_mac);
+
+	if (dev->caps.steering_mode != MLX4_STEERING_MODE_A0) {
+		struct hlist_head *bucket;
+		unsigned int mac_hash;
+		struct mlx4_mac_entry *entry;
+		struct hlist_node *tmp;
+		u64 prev_mac_u64 = ether_addr_to_u64(prev_mac);
+
+		bucket = &priv->mac_hash[prev_mac[MLX4_EN_MAC_HASH_IDX]];
+		hlist_for_each_entry_safe(entry, tmp, bucket, hlist) {
+			if (ether_addr_equal_64bits(entry->mac, prev_mac)) {
+				mlx4_en_uc_steer_release(priv, entry->mac,
+							 qpn, entry->reg_id);
+				mlx4_unregister_mac(dev, priv->port,
+						    prev_mac_u64);
+				hlist_del_rcu(&entry->hlist);
+				synchronize_rcu();
+				memcpy(entry->mac, new_mac, ETH_ALEN);
+				entry->reg_id = 0;
+				mac_hash = new_mac[MLX4_EN_MAC_HASH_IDX];
+				hlist_add_head_rcu(&entry->hlist,
+						   &priv->mac_hash[mac_hash]);
+				mlx4_register_mac(dev, priv->port, new_mac_u64);
+				err = mlx4_en_uc_steer_add(priv, new_mac,
+							   &qpn,
+							   &entry->reg_id);
+				if (err)
+					return err;
+				if (priv->tunnel_reg_id) {
+					mlx4_flow_detach(priv->mdev->dev, priv->tunnel_reg_id);
+					priv->tunnel_reg_id = 0;
+				}
+				err = mlx4_en_tunnel_steer_add(priv, new_mac, qpn,
+							       &priv->tunnel_reg_id);
+				return err;
+			}
+		}
+		return -EINVAL;
+	}
+
+	return __mlx4_replace_mac(dev, priv->port, qpn, new_mac_u64);
+}
+
+static void mlx4_en_update_user_mac(struct mlx4_en_priv *priv,
+				    unsigned char new_mac[ETH_ALEN + 2])
+{
+	struct mlx4_en_dev *mdev = priv->mdev;
+	int err;
+
+	if (!(mdev->dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_USER_MAC_EN))
+		return;
+
+	err = mlx4_SET_PORT_user_mac(mdev->dev, priv->port, new_mac);
+	if (err)
+		en_err(priv, "Failed to pass user MAC(%pM) to Firmware for port %d, with error %d\n",
+		       new_mac, priv->port, err);
+}
+
+static int mlx4_en_do_set_mac(struct mlx4_en_priv *priv,
+			      unsigned char new_mac[ETH_ALEN + 2])
+{
+	int err = 0;
+
+	if (priv->port_up) {
+		/* Remove old MAC and insert the new one */
+		err = mlx4_en_replace_mac(priv, priv->base_qpn,
+					  new_mac, priv->current_mac);
+		if (err)
+			en_err(priv, "Failed changing HW MAC address\n");
+	} else
+		en_dbg(HW, priv, "Port is down while registering mac, exiting...\n");
+
+	if (!err)
+		memcpy(priv->current_mac, new_mac, sizeof(priv->current_mac));
+
+	return err;
+}
+
+static int mlx4_en_set_mac(struct net_device *dev, void *addr)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	struct mlx4_en_dev *mdev = priv->mdev;
+	struct sockaddr *saddr = addr;
+	unsigned char new_mac[ETH_ALEN + 2];
+	int err;
+
+	if (!is_valid_ether_addr(saddr->sa_data))
+		return -EADDRNOTAVAIL;
+
+	mutex_lock(&mdev->state_lock);
+	memcpy(new_mac, saddr->sa_data, ETH_ALEN);
+	err = mlx4_en_do_set_mac(priv, new_mac);
+	if (err)
+		goto out;
+
+	eth_hw_addr_set(dev, saddr->sa_data);
+	mlx4_en_update_user_mac(priv, new_mac);
+out:
+	mutex_unlock(&mdev->state_lock);
+
+	return err;
+}
+
+static void mlx4_en_clear_list(struct net_device *dev)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	struct mlx4_en_mc_list *tmp, *mc_to_del;
+
+	list_for_each_entry_safe(mc_to_del, tmp, &priv->mc_list, list) {
+		list_del(&mc_to_del->list);
+		kfree(mc_to_del);
+	}
+}
+
+static void mlx4_en_cache_mclist(struct net_device *dev)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	struct netdev_hw_addr *ha;
+	struct mlx4_en_mc_list *tmp;
+
+	mlx4_en_clear_list(dev);
+	netdev_for_each_mc_addr(ha, dev) {
+		tmp = kzalloc(sizeof(struct mlx4_en_mc_list), GFP_ATOMIC);
+		if (!tmp) {
+			mlx4_en_clear_list(dev);
+			return;
+		}
+		memcpy(tmp->addr, ha->addr, ETH_ALEN);
+		list_add_tail(&tmp->list, &priv->mc_list);
+	}
+}
+
+static void update_mclist_flags(struct mlx4_en_priv *priv,
+				struct list_head *dst,
+				struct list_head *src)
+{
+	struct mlx4_en_mc_list *dst_tmp, *src_tmp, *new_mc;
+	bool found;
+
+	/* Find all the entries that should be removed from dst,
+	 * These are the entries that are not found in src
+	 */
+	list_for_each_entry(dst_tmp, dst, list) {
+		found = false;
+		list_for_each_entry(src_tmp, src, list) {
+			if (ether_addr_equal(dst_tmp->addr, src_tmp->addr)) {
+				found = true;
+				break;
+			}
+		}
+		if (!found)
+			dst_tmp->action = MCLIST_REM;
+	}
+
+	/* Add entries that exist in src but not in dst
+	 * mark them as need to add
+	 */
+	list_for_each_entry(src_tmp, src, list) {
+		found = false;
+		list_for_each_entry(dst_tmp, dst, list) {
+			if (ether_addr_equal(dst_tmp->addr, src_tmp->addr)) {
+				dst_tmp->action = MCLIST_NONE;
+				found = true;
+				break;
+			}
+		}
+		if (!found) {
+			new_mc = kmemdup(src_tmp,
+					 sizeof(struct mlx4_en_mc_list),
+					 GFP_KERNEL);
+			if (!new_mc)
+				return;
+
+			new_mc->action = MCLIST_ADD;
+			list_add_tail(&new_mc->list, dst);
+		}
+	}
+}
+
+static void mlx4_en_set_rx_mode(struct net_device *dev)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+
+	if (!priv->port_up)
+		return;
+
+	queue_work(priv->mdev->workqueue, &priv->rx_mode_task);
+}
+
+static void mlx4_en_set_promisc_mode(struct mlx4_en_priv *priv,
+				     struct mlx4_en_dev *mdev)
+{
+	int err = 0;
+
+	if (!(priv->flags & MLX4_EN_FLAG_PROMISC)) {
+		if (netif_msg_rx_status(priv))
+			en_warn(priv, "Entering promiscuous mode\n");
+		priv->flags |= MLX4_EN_FLAG_PROMISC;
+
+		/* Enable promiscouos mode */
+		switch (mdev->dev->caps.steering_mode) {
+		case MLX4_STEERING_MODE_DEVICE_MANAGED:
+			err = mlx4_flow_steer_promisc_add(mdev->dev,
+							  priv->port,
+							  priv->base_qpn,
+							  MLX4_FS_ALL_DEFAULT);
+			if (err)
+				en_err(priv, "Failed enabling promiscuous mode\n");
+			priv->flags |= MLX4_EN_FLAG_MC_PROMISC;
+			break;
+
+		case MLX4_STEERING_MODE_B0:
+			err = mlx4_unicast_promisc_add(mdev->dev,
+						       priv->base_qpn,
+						       priv->port);
+			if (err)
+				en_err(priv, "Failed enabling unicast promiscuous mode\n");
+
+			/* Add the default qp number as multicast
+			 * promisc
+			 */
+			if (!(priv->flags & MLX4_EN_FLAG_MC_PROMISC)) {
+				err = mlx4_multicast_promisc_add(mdev->dev,
+								 priv->base_qpn,
+								 priv->port);
+				if (err)
+					en_err(priv, "Failed enabling multicast promiscuous mode\n");
+				priv->flags |= MLX4_EN_FLAG_MC_PROMISC;
+			}
+			break;
+
+		case MLX4_STEERING_MODE_A0:
+			err = mlx4_SET_PORT_qpn_calc(mdev->dev,
+						     priv->port,
+						     priv->base_qpn,
+						     1);
+			if (err)
+				en_err(priv, "Failed enabling promiscuous mode\n");
+			break;
+		}
+
+		/* Disable port multicast filter (unconditionally) */
+		err = mlx4_SET_MCAST_FLTR(mdev->dev, priv->port, 0,
+					  0, MLX4_MCAST_DISABLE);
+		if (err)
+			en_err(priv, "Failed disabling multicast filter\n");
+	}
+}
+
+static void mlx4_en_clear_promisc_mode(struct mlx4_en_priv *priv,
+				       struct mlx4_en_dev *mdev)
+{
+	int err = 0;
+
+	if (netif_msg_rx_status(priv))
+		en_warn(priv, "Leaving promiscuous mode\n");
+	priv->flags &= ~MLX4_EN_FLAG_PROMISC;
+
+	/* Disable promiscouos mode */
+	switch (mdev->dev->caps.steering_mode) {
+	case MLX4_STEERING_MODE_DEVICE_MANAGED:
+		err = mlx4_flow_steer_promisc_remove(mdev->dev,
+						     priv->port,
+						     MLX4_FS_ALL_DEFAULT);
+		if (err)
+			en_err(priv, "Failed disabling promiscuous mode\n");
+		priv->flags &= ~MLX4_EN_FLAG_MC_PROMISC;
+		break;
+
+	case MLX4_STEERING_MODE_B0:
+		err = mlx4_unicast_promisc_remove(mdev->dev,
+						  priv->base_qpn,
+						  priv->port);
+		if (err)
+			en_err(priv, "Failed disabling unicast promiscuous mode\n");
+		/* Disable Multicast promisc */
+		if (priv->flags & MLX4_EN_FLAG_MC_PROMISC) {
+			err = mlx4_multicast_promisc_remove(mdev->dev,
+							    priv->base_qpn,
+							    priv->port);
+			if (err)
+				en_err(priv, "Failed disabling multicast promiscuous mode\n");
+			priv->flags &= ~MLX4_EN_FLAG_MC_PROMISC;
+		}
+		break;
+
+	case MLX4_STEERING_MODE_A0:
+		err = mlx4_SET_PORT_qpn_calc(mdev->dev,
+					     priv->port,
+					     priv->base_qpn, 0);
+		if (err)
+			en_err(priv, "Failed disabling promiscuous mode\n");
+		break;
+	}
+}
+
+static void mlx4_en_do_multicast(struct mlx4_en_priv *priv,
+				 struct net_device *dev,
+				 struct mlx4_en_dev *mdev)
+{
+	struct mlx4_en_mc_list *mclist, *tmp;
+	u64 mcast_addr = 0;
+	u8 mc_list[16] = {0};
+	int err = 0;
+
+	/* Enable/disable the multicast filter according to IFF_ALLMULTI */
+	if (dev->flags & IFF_ALLMULTI) {
+		err = mlx4_SET_MCAST_FLTR(mdev->dev, priv->port, 0,
+					  0, MLX4_MCAST_DISABLE);
+		if (err)
+			en_err(priv, "Failed disabling multicast filter\n");
+
+		/* Add the default qp number as multicast promisc */
+		if (!(priv->flags & MLX4_EN_FLAG_MC_PROMISC)) {
+			switch (mdev->dev->caps.steering_mode) {
+			case MLX4_STEERING_MODE_DEVICE_MANAGED:
+				err = mlx4_flow_steer_promisc_add(mdev->dev,
+								  priv->port,
+								  priv->base_qpn,
+								  MLX4_FS_MC_DEFAULT);
+				break;
+
+			case MLX4_STEERING_MODE_B0:
+				err = mlx4_multicast_promisc_add(mdev->dev,
+								 priv->base_qpn,
+								 priv->port);
+				break;
+
+			case MLX4_STEERING_MODE_A0:
+				break;
+			}
+			if (err)
+				en_err(priv, "Failed entering multicast promisc mode\n");
+			priv->flags |= MLX4_EN_FLAG_MC_PROMISC;
+		}
+	} else {
+		/* Disable Multicast promisc */
+		if (priv->flags & MLX4_EN_FLAG_MC_PROMISC) {
+			switch (mdev->dev->caps.steering_mode) {
+			case MLX4_STEERING_MODE_DEVICE_MANAGED:
+				err = mlx4_flow_steer_promisc_remove(mdev->dev,
+								     priv->port,
+								     MLX4_FS_MC_DEFAULT);
+				break;
+
+			case MLX4_STEERING_MODE_B0:
+				err = mlx4_multicast_promisc_remove(mdev->dev,
+								    priv->base_qpn,
+								    priv->port);
+				break;
+
+			case MLX4_STEERING_MODE_A0:
+				break;
+			}
+			if (err)
+				en_err(priv, "Failed disabling multicast promiscuous mode\n");
+			priv->flags &= ~MLX4_EN_FLAG_MC_PROMISC;
+		}
+
+		err = mlx4_SET_MCAST_FLTR(mdev->dev, priv->port, 0,
+					  0, MLX4_MCAST_DISABLE);
+		if (err)
+			en_err(priv, "Failed disabling multicast filter\n");
+
+		/* Flush mcast filter and init it with broadcast address */
+		mlx4_SET_MCAST_FLTR(mdev->dev, priv->port, ETH_BCAST,
+				    1, MLX4_MCAST_CONFIG);
+
+		/* Update multicast list - we cache all addresses so they won't
+		 * change while HW is updated holding the command semaphor */
+		netif_addr_lock_bh(dev);
+		mlx4_en_cache_mclist(dev);
+		netif_addr_unlock_bh(dev);
+		list_for_each_entry(mclist, &priv->mc_list, list) {
+			mcast_addr = ether_addr_to_u64(mclist->addr);
+			mlx4_SET_MCAST_FLTR(mdev->dev, priv->port,
+					    mcast_addr, 0, MLX4_MCAST_CONFIG);
+		}
+		err = mlx4_SET_MCAST_FLTR(mdev->dev, priv->port, 0,
+					  0, MLX4_MCAST_ENABLE);
+		if (err)
+			en_err(priv, "Failed enabling multicast filter\n");
+
+		update_mclist_flags(priv, &priv->curr_list, &priv->mc_list);
+		list_for_each_entry_safe(mclist, tmp, &priv->curr_list, list) {
+			if (mclist->action == MCLIST_REM) {
+				/* detach this address and delete from list */
+				memcpy(&mc_list[10], mclist->addr, ETH_ALEN);
+				mc_list[5] = priv->port;
+				err = mlx4_multicast_detach(mdev->dev,
+							    priv->rss_map.indir_qp,
+							    mc_list,
+							    MLX4_PROT_ETH,
+							    mclist->reg_id);
+				if (err)
+					en_err(priv, "Fail to detach multicast address\n");
+
+				if (mclist->tunnel_reg_id) {
+					err = mlx4_flow_detach(priv->mdev->dev, mclist->tunnel_reg_id);
+					if (err)
+						en_err(priv, "Failed to detach multicast address\n");
+				}
+
+				/* remove from list */
+				list_del(&mclist->list);
+				kfree(mclist);
+			} else if (mclist->action == MCLIST_ADD) {
+				/* attach the address */
+				memcpy(&mc_list[10], mclist->addr, ETH_ALEN);
+				/* needed for B0 steering support */
+				mc_list[5] = priv->port;
+				err = mlx4_multicast_attach(mdev->dev,
+							    priv->rss_map.indir_qp,
+							    mc_list,
+							    priv->port, 0,
+							    MLX4_PROT_ETH,
+							    &mclist->reg_id);
+				if (err)
+					en_err(priv, "Fail to attach multicast address\n");
+
+				err = mlx4_en_tunnel_steer_add(priv, &mc_list[10], priv->base_qpn,
+							       &mclist->tunnel_reg_id);
+				if (err)
+					en_err(priv, "Failed to attach multicast address\n");
+			}
+		}
+	}
+}
+
+static void mlx4_en_do_uc_filter(struct mlx4_en_priv *priv,
+				 struct net_device *dev,
+				 struct mlx4_en_dev *mdev)
+{
+	struct netdev_hw_addr *ha;
+	struct mlx4_mac_entry *entry;
+	struct hlist_node *tmp;
+	bool found;
+	u64 mac;
+	int err = 0;
+	struct hlist_head *bucket;
+	unsigned int i;
+	int removed = 0;
+	u32 prev_flags;
+
+	/* Note that we do not need to protect our mac_hash traversal with rcu,
+	 * since all modification code is protected by mdev->state_lock
+	 */
+
+	/* find what to remove */
+	for (i = 0; i < MLX4_EN_MAC_HASH_SIZE; ++i) {
+		bucket = &priv->mac_hash[i];
+		hlist_for_each_entry_safe(entry, tmp, bucket, hlist) {
+			found = false;
+			netdev_for_each_uc_addr(ha, dev) {
+				if (ether_addr_equal_64bits(entry->mac,
+							    ha->addr)) {
+					found = true;
+					break;
+				}
+			}
+
+			/* MAC address of the port is not in uc list */
+			if (ether_addr_equal_64bits(entry->mac,
+						    priv->current_mac))
+				found = true;
+
+			if (!found) {
+				mac = ether_addr_to_u64(entry->mac);
+				mlx4_en_uc_steer_release(priv, entry->mac,
+							 priv->base_qpn,
+							 entry->reg_id);
+				mlx4_unregister_mac(mdev->dev, priv->port, mac);
+
+				hlist_del_rcu(&entry->hlist);
+				kfree_rcu(entry, rcu);
+				en_dbg(DRV, priv, "Removed MAC %pM on port:%d\n",
+				       entry->mac, priv->port);
+				++removed;
+			}
+		}
+	}
+
+	/* if we didn't remove anything, there is no use in trying to add
+	 * again once we are in a forced promisc mode state
+	 */
+	if ((priv->flags & MLX4_EN_FLAG_FORCE_PROMISC) && 0 == removed)
+		return;
+
+	prev_flags = priv->flags;
+	priv->flags &= ~MLX4_EN_FLAG_FORCE_PROMISC;
+
+	/* find what to add */
+	netdev_for_each_uc_addr(ha, dev) {
+		found = false;
+		bucket = &priv->mac_hash[ha->addr[MLX4_EN_MAC_HASH_IDX]];
+		hlist_for_each_entry(entry, bucket, hlist) {
+			if (ether_addr_equal_64bits(entry->mac, ha->addr)) {
+				found = true;
+				break;
+			}
+		}
+
+		if (!found) {
+			entry = kmalloc(sizeof(*entry), GFP_KERNEL);
+			if (!entry) {
+				en_err(priv, "Failed adding MAC %pM on port:%d (out of memory)\n",
+				       ha->addr, priv->port);
+				priv->flags |= MLX4_EN_FLAG_FORCE_PROMISC;
+				break;
+			}
+			mac = ether_addr_to_u64(ha->addr);
+			memcpy(entry->mac, ha->addr, ETH_ALEN);
+			err = mlx4_register_mac(mdev->dev, priv->port, mac);
+			if (err < 0) {
+				en_err(priv, "Failed registering MAC %pM on port %d: %d\n",
+				       ha->addr, priv->port, err);
+				kfree(entry);
+				priv->flags |= MLX4_EN_FLAG_FORCE_PROMISC;
+				break;
+			}
+			err = mlx4_en_uc_steer_add(priv, ha->addr,
+						   &priv->base_qpn,
+						   &entry->reg_id);
+			if (err) {
+				en_err(priv, "Failed adding MAC %pM on port %d: %d\n",
+				       ha->addr, priv->port, err);
+				mlx4_unregister_mac(mdev->dev, priv->port, mac);
+				kfree(entry);
+				priv->flags |= MLX4_EN_FLAG_FORCE_PROMISC;
+				break;
+			} else {
+				unsigned int mac_hash;
+				en_dbg(DRV, priv, "Added MAC %pM on port:%d\n",
+				       ha->addr, priv->port);
+				mac_hash = ha->addr[MLX4_EN_MAC_HASH_IDX];
+				bucket = &priv->mac_hash[mac_hash];
+				hlist_add_head_rcu(&entry->hlist, bucket);
+			}
+		}
+	}
+
+	if (priv->flags & MLX4_EN_FLAG_FORCE_PROMISC) {
+		en_warn(priv, "Forcing promiscuous mode on port:%d\n",
+			priv->port);
+	} else if (prev_flags & MLX4_EN_FLAG_FORCE_PROMISC) {
+		en_warn(priv, "Stop forcing promiscuous mode on port:%d\n",
+			priv->port);
+	}
+}
+
+static void mlx4_en_do_set_rx_mode(struct work_struct *work)
+{
+	struct mlx4_en_priv *priv = container_of(work, struct mlx4_en_priv,
+						 rx_mode_task);
+	struct mlx4_en_dev *mdev = priv->mdev;
+	struct net_device *dev = priv->dev;
+
+	mutex_lock(&mdev->state_lock);
+	if (!mdev->device_up) {
+		en_dbg(HW, priv, "Card is not up, ignoring rx mode change.\n");
+		goto out;
+	}
+	if (!priv->port_up) {
+		en_dbg(HW, priv, "Port is down, ignoring rx mode change.\n");
+		goto out;
+	}
+
+	if (!netif_carrier_ok(dev)) {
+		if (!mlx4_en_QUERY_PORT(mdev, priv->port)) {
+			if (priv->port_state.link_state) {
+				netif_carrier_on(dev);
+				en_dbg(LINK, priv, "Link Up\n");
+			}
+		}
+	}
+
+	if (dev->priv_flags & IFF_UNICAST_FLT)
+		mlx4_en_do_uc_filter(priv, dev, mdev);
+
+	/* Promsicuous mode: disable all filters */
+	if ((dev->flags & IFF_PROMISC) ||
+	    (priv->flags & MLX4_EN_FLAG_FORCE_PROMISC)) {
+		mlx4_en_set_promisc_mode(priv, mdev);
+		goto out;
+	}
+
+	/* Not in promiscuous mode */
+	if (priv->flags & MLX4_EN_FLAG_PROMISC)
+		mlx4_en_clear_promisc_mode(priv, mdev);
+
+	mlx4_en_do_multicast(priv, dev, mdev);
+out:
+	mutex_unlock(&mdev->state_lock);
+}
+
+static int mlx4_en_set_rss_steer_rules(struct mlx4_en_priv *priv)
+{
+	u64 reg_id;
+	int err = 0;
+	int *qpn = &priv->base_qpn;
+	struct mlx4_mac_entry *entry;
+
+	err = mlx4_en_uc_steer_add(priv, priv->dev->dev_addr, qpn, &reg_id);
+	if (err)
+		return err;
+
+	err = mlx4_en_tunnel_steer_add(priv, priv->dev->dev_addr, *qpn,
+				       &priv->tunnel_reg_id);
+	if (err)
+		goto tunnel_err;
+
+	entry = kmalloc(sizeof(*entry), GFP_KERNEL);
+	if (!entry) {
+		err = -ENOMEM;
+		goto alloc_err;
+	}
+
+	memcpy(entry->mac, priv->dev->dev_addr, sizeof(entry->mac));
+	memcpy(priv->current_mac, entry->mac, sizeof(priv->current_mac));
+	entry->reg_id = reg_id;
+	hlist_add_head_rcu(&entry->hlist,
+			   &priv->mac_hash[entry->mac[MLX4_EN_MAC_HASH_IDX]]);
+
+	return 0;
+
+alloc_err:
+	if (priv->tunnel_reg_id)
+		mlx4_flow_detach(priv->mdev->dev, priv->tunnel_reg_id);
+
+tunnel_err:
+	mlx4_en_uc_steer_release(priv, priv->dev->dev_addr, *qpn, reg_id);
+	return err;
+}
+
+static void mlx4_en_delete_rss_steer_rules(struct mlx4_en_priv *priv)
+{
+	u64 mac;
+	unsigned int i;
+	int qpn = priv->base_qpn;
+	struct hlist_head *bucket;
+	struct hlist_node *tmp;
+	struct mlx4_mac_entry *entry;
+
+	for (i = 0; i < MLX4_EN_MAC_HASH_SIZE; ++i) {
+		bucket = &priv->mac_hash[i];
+		hlist_for_each_entry_safe(entry, tmp, bucket, hlist) {
+			mac = ether_addr_to_u64(entry->mac);
+			en_dbg(DRV, priv, "Registering MAC:%pM for deleting\n",
+			       entry->mac);
+			mlx4_en_uc_steer_release(priv, entry->mac,
+						 qpn, entry->reg_id);
+
+			mlx4_unregister_mac(priv->mdev->dev, priv->port, mac);
+			hlist_del_rcu(&entry->hlist);
+			kfree_rcu(entry, rcu);
+		}
+	}
+
+	if (priv->tunnel_reg_id) {
+		mlx4_flow_detach(priv->mdev->dev, priv->tunnel_reg_id);
+		priv->tunnel_reg_id = 0;
+	}
+}
+
+static void mlx4_en_tx_timeout(struct net_device *dev, unsigned int txqueue)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	struct mlx4_en_dev *mdev = priv->mdev;
+	struct mlx4_en_tx_ring *tx_ring = priv->tx_ring[TX][txqueue];
+
+	if (netif_msg_timer(priv))
+		en_warn(priv, "Tx timeout called on port:%d\n", priv->port);
+
+	en_warn(priv, "TX timeout on queue: %d, QP: 0x%x, CQ: 0x%x, Cons: 0x%x, Prod: 0x%x\n",
+		txqueue, tx_ring->qpn, tx_ring->sp_cqn,
+		tx_ring->cons, tx_ring->prod);
+
+	priv->port_stats.tx_timeout++;
+	if (!test_and_set_bit(MLX4_EN_STATE_FLAG_RESTARTING, &priv->state)) {
+		en_dbg(DRV, priv, "Scheduling port restart\n");
+		queue_work(mdev->workqueue, &priv->restart_task);
+	}
+}
+
+
+static void
+mlx4_en_get_stats64(struct net_device *dev, struct rtnl_link_stats64 *stats)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+
+	spin_lock_bh(&priv->stats_lock);
+	mlx4_en_fold_software_stats(dev);
+	netdev_stats_to_stats64(stats, &dev->stats);
+	spin_unlock_bh(&priv->stats_lock);
+}
+
+static void mlx4_en_set_default_moderation(struct mlx4_en_priv *priv)
+{
+	struct mlx4_en_cq *cq;
+	int i, t;
+
+	/* If we haven't received a specific coalescing setting
+	 * (module param), we set the moderation parameters as follows:
+	 * - moder_cnt is set to the number of mtu sized packets to
+	 *   satisfy our coalescing target.
+	 * - moder_time is set to a fixed value.
+	 */
+	priv->rx_frames = MLX4_EN_RX_COAL_TARGET;
+	priv->rx_usecs = MLX4_EN_RX_COAL_TIME;
+	priv->tx_frames = MLX4_EN_TX_COAL_PKTS;
+	priv->tx_usecs = MLX4_EN_TX_COAL_TIME;
+	en_dbg(INTR, priv, "Default coalescing params for mtu:%d - rx_frames:%d rx_usecs:%d\n",
+	       priv->dev->mtu, priv->rx_frames, priv->rx_usecs);
+
+	/* Setup cq moderation params */
+	for (i = 0; i < priv->rx_ring_num; i++) {
+		cq = priv->rx_cq[i];
+		cq->moder_cnt = priv->rx_frames;
+		cq->moder_time = priv->rx_usecs;
+		priv->last_moder_time[i] = MLX4_EN_AUTO_CONF;
+		priv->last_moder_packets[i] = 0;
+		priv->last_moder_bytes[i] = 0;
+	}
+
+	for (t = 0 ; t < MLX4_EN_NUM_TX_TYPES; t++) {
+		for (i = 0; i < priv->tx_ring_num[t]; i++) {
+			cq = priv->tx_cq[t][i];
+			cq->moder_cnt = priv->tx_frames;
+			cq->moder_time = priv->tx_usecs;
+		}
+	}
+
+	/* Reset auto-moderation params */
+	priv->pkt_rate_low = MLX4_EN_RX_RATE_LOW;
+	priv->rx_usecs_low = MLX4_EN_RX_COAL_TIME_LOW;
+	priv->pkt_rate_high = MLX4_EN_RX_RATE_HIGH;
+	priv->rx_usecs_high = MLX4_EN_RX_COAL_TIME_HIGH;
+	priv->sample_interval = MLX4_EN_SAMPLE_INTERVAL;
+	priv->adaptive_rx_coal = 1;
+	priv->last_moder_jiffies = 0;
+	priv->last_moder_tx_packets = 0;
+}
+
+static void mlx4_en_auto_moderation(struct mlx4_en_priv *priv)
+{
+	unsigned long period = (unsigned long) (jiffies - priv->last_moder_jiffies);
+	u32 pkt_rate_high, pkt_rate_low;
+	struct mlx4_en_cq *cq;
+	unsigned long packets;
+	unsigned long rate;
+	unsigned long avg_pkt_size;
+	unsigned long rx_packets;
+	unsigned long rx_bytes;
+	unsigned long rx_pkt_diff;
+	int moder_time;
+	int ring, err;
+
+	if (!priv->adaptive_rx_coal || period < priv->sample_interval * HZ)
+		return;
+
+	pkt_rate_low = READ_ONCE(priv->pkt_rate_low);
+	pkt_rate_high = READ_ONCE(priv->pkt_rate_high);
+
+	for (ring = 0; ring < priv->rx_ring_num; ring++) {
+		rx_packets = READ_ONCE(priv->rx_ring[ring]->packets);
+		rx_bytes = READ_ONCE(priv->rx_ring[ring]->bytes);
+
+		rx_pkt_diff = rx_packets - priv->last_moder_packets[ring];
+		packets = rx_pkt_diff;
+		rate = packets * HZ / period;
+		avg_pkt_size = packets ? (rx_bytes -
+				priv->last_moder_bytes[ring]) / packets : 0;
+
+		/* Apply auto-moderation only when packet rate
+		 * exceeds a rate that it matters */
+		if (rate > (MLX4_EN_RX_RATE_THRESH / priv->rx_ring_num) &&
+		    avg_pkt_size > MLX4_EN_AVG_PKT_SMALL) {
+			if (rate <= pkt_rate_low)
+				moder_time = priv->rx_usecs_low;
+			else if (rate >= pkt_rate_high)
+				moder_time = priv->rx_usecs_high;
+			else
+				moder_time = (rate - pkt_rate_low) *
+					(priv->rx_usecs_high - priv->rx_usecs_low) /
+					(pkt_rate_high - pkt_rate_low) +
+					priv->rx_usecs_low;
+		} else {
+			moder_time = priv->rx_usecs_low;
+		}
+
+		cq = priv->rx_cq[ring];
+		if (moder_time != priv->last_moder_time[ring] ||
+		    cq->moder_cnt != priv->rx_frames) {
+			priv->last_moder_time[ring] = moder_time;
+			cq->moder_time = moder_time;
+			cq->moder_cnt = priv->rx_frames;
+			err = mlx4_en_set_cq_moder(priv, cq);
+			if (err)
+				en_err(priv, "Failed modifying moderation for cq:%d\n",
+				       ring);
+		}
+		priv->last_moder_packets[ring] = rx_packets;
+		priv->last_moder_bytes[ring] = rx_bytes;
+	}
+
+	priv->last_moder_jiffies = jiffies;
+}
+
+static void mlx4_en_do_get_stats(struct work_struct *work)
+{
+	struct delayed_work *delay = to_delayed_work(work);
+	struct mlx4_en_priv *priv = container_of(delay, struct mlx4_en_priv,
+						 stats_task);
+	struct mlx4_en_dev *mdev = priv->mdev;
+	int err;
+
+	mutex_lock(&mdev->state_lock);
+	if (mdev->device_up) {
+		if (priv->port_up) {
+			err = mlx4_en_DUMP_ETH_STATS(mdev, priv->port, 0);
+			if (err)
+				en_dbg(HW, priv, "Could not update stats\n");
+
+			mlx4_en_auto_moderation(priv);
+		}
+
+		queue_delayed_work(mdev->workqueue, &priv->stats_task, STATS_DELAY);
+	}
+	if (mdev->mac_removed[MLX4_MAX_PORTS + 1 - priv->port]) {
+		mlx4_en_do_set_mac(priv, priv->current_mac);
+		mdev->mac_removed[MLX4_MAX_PORTS + 1 - priv->port] = 0;
+	}
+	mutex_unlock(&mdev->state_lock);
+}
+
+/* mlx4_en_service_task - Run service task for tasks that needed to be done
+ * periodically
+ */
+static void mlx4_en_service_task(struct work_struct *work)
+{
+	struct delayed_work *delay = to_delayed_work(work);
+	struct mlx4_en_priv *priv = container_of(delay, struct mlx4_en_priv,
+						 service_task);
+	struct mlx4_en_dev *mdev = priv->mdev;
+
+	mutex_lock(&mdev->state_lock);
+	if (mdev->device_up) {
+		if (mdev->dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_TS)
+			mlx4_en_ptp_overflow_check(mdev);
+
+		mlx4_en_recover_from_oom(priv);
+		queue_delayed_work(mdev->workqueue, &priv->service_task,
+				   SERVICE_TASK_DELAY);
+	}
+	mutex_unlock(&mdev->state_lock);
+}
+
+static void mlx4_en_linkstate(struct mlx4_en_priv *priv)
+{
+	struct mlx4_en_port_state *port_state = &priv->port_state;
+	struct mlx4_en_dev *mdev = priv->mdev;
+	struct net_device *dev = priv->dev;
+	bool up;
+
+	if (mlx4_en_QUERY_PORT(mdev, priv->port))
+		port_state->link_state = MLX4_PORT_STATE_DEV_EVENT_PORT_DOWN;
+
+	up = port_state->link_state == MLX4_PORT_STATE_DEV_EVENT_PORT_UP;
+	if (up == netif_carrier_ok(dev))
+		netif_carrier_event(dev);
+	if (!up) {
+		en_info(priv, "Link Down\n");
+		netif_carrier_off(dev);
+	} else {
+		en_info(priv, "Link Up\n");
+		netif_carrier_on(dev);
+	}
+}
+
+static void mlx4_en_linkstate_work(struct work_struct *work)
+{
+	struct mlx4_en_priv *priv = container_of(work, struct mlx4_en_priv,
+						 linkstate_task);
+	struct mlx4_en_dev *mdev = priv->mdev;
+
+	mutex_lock(&mdev->state_lock);
+	mlx4_en_linkstate(priv);
+	mutex_unlock(&mdev->state_lock);
+}
+
+static int mlx4_en_init_affinity_hint(struct mlx4_en_priv *priv, int ring_idx)
+{
+	struct mlx4_en_rx_ring *ring = priv->rx_ring[ring_idx];
+	int numa_node = priv->mdev->dev->numa_node;
+
+	if (!zalloc_cpumask_var(&ring->affinity_mask, GFP_KERNEL))
+		return -ENOMEM;
+
+	cpumask_set_cpu(cpumask_local_spread(ring_idx, numa_node),
+			ring->affinity_mask);
+	return 0;
+}
+
+static void mlx4_en_free_affinity_hint(struct mlx4_en_priv *priv, int ring_idx)
+{
+	free_cpumask_var(priv->rx_ring[ring_idx]->affinity_mask);
+}
+
+static void mlx4_en_init_recycle_ring(struct mlx4_en_priv *priv,
+				      int tx_ring_idx)
+{
+	struct mlx4_en_tx_ring *tx_ring = priv->tx_ring[TX_XDP][tx_ring_idx];
+	int rr_index = tx_ring_idx;
+
+	tx_ring->free_tx_desc = mlx4_en_recycle_tx_desc;
+	tx_ring->recycle_ring = priv->rx_ring[rr_index];
+	en_dbg(DRV, priv, "Set tx_ring[%d][%d]->recycle_ring = rx_ring[%d]\n",
+	       TX_XDP, tx_ring_idx, rr_index);
+}
+
+int mlx4_en_start_port(struct net_device *dev)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	struct mlx4_en_dev *mdev = priv->mdev;
+	struct mlx4_en_cq *cq;
+	struct mlx4_en_tx_ring *tx_ring;
+	int rx_index = 0;
+	int err = 0;
+	int i, t;
+	int j;
+	u8 mc_list[16] = {0};
+
+	if (priv->port_up) {
+		en_dbg(DRV, priv, "start port called while port already up\n");
+		return 0;
+	}
+
+	INIT_LIST_HEAD(&priv->mc_list);
+	INIT_LIST_HEAD(&priv->curr_list);
+	INIT_LIST_HEAD(&priv->ethtool_list);
+	memset(&priv->ethtool_rules[0], 0,
+	       sizeof(struct ethtool_flow_id) * MAX_NUM_OF_FS_RULES);
+
+	/* Calculate Rx buf size */
+	dev->mtu = min(dev->mtu, priv->max_mtu);
+	mlx4_en_calc_rx_buf(dev);
+	en_dbg(DRV, priv, "Rx buf size:%d\n", priv->rx_skb_size);
+
+	/* Configure rx cq's and rings */
+	err = mlx4_en_activate_rx_rings(priv);
+	if (err) {
+		en_err(priv, "Failed to activate RX rings\n");
+		return err;
+	}
+	for (i = 0; i < priv->rx_ring_num; i++) {
+		cq = priv->rx_cq[i];
+
+		err = mlx4_en_init_affinity_hint(priv, i);
+		if (err) {
+			en_err(priv, "Failed preparing IRQ affinity hint\n");
+			goto cq_err;
+		}
+
+		err = mlx4_en_activate_cq(priv, cq, i);
+		if (err) {
+			en_err(priv, "Failed activating Rx CQ\n");
+			mlx4_en_free_affinity_hint(priv, i);
+			goto cq_err;
+		}
+
+		for (j = 0; j < cq->size; j++) {
+			struct mlx4_cqe *cqe = NULL;
+
+			cqe = mlx4_en_get_cqe(cq->buf, j, priv->cqe_size) +
+			      priv->cqe_factor;
+			cqe->owner_sr_opcode = MLX4_CQE_OWNER_MASK;
+		}
+
+		err = mlx4_en_set_cq_moder(priv, cq);
+		if (err) {
+			en_err(priv, "Failed setting cq moderation parameters\n");
+			mlx4_en_deactivate_cq(priv, cq);
+			mlx4_en_free_affinity_hint(priv, i);
+			goto cq_err;
+		}
+		mlx4_en_arm_cq(priv, cq);
+		priv->rx_ring[i]->cqn = cq->mcq.cqn;
+		++rx_index;
+	}
+
+	/* Set qp number */
+	en_dbg(DRV, priv, "Getting qp number for port %d\n", priv->port);
+	err = mlx4_en_get_qp(priv);
+	if (err) {
+		en_err(priv, "Failed getting eth qp\n");
+		goto cq_err;
+	}
+	mdev->mac_removed[priv->port] = 0;
+
+	priv->counter_index =
+			mlx4_get_default_counter_index(mdev->dev, priv->port);
+
+	err = mlx4_en_config_rss_steer(priv);
+	if (err) {
+		en_err(priv, "Failed configuring rss steering\n");
+		goto mac_err;
+	}
+
+	err = mlx4_en_create_drop_qp(priv);
+	if (err)
+		goto rss_err;
+
+	/* Configure tx cq's and rings */
+	for (t = 0 ; t < MLX4_EN_NUM_TX_TYPES; t++) {
+		u8 num_tx_rings_p_up = t == TX ?
+			priv->num_tx_rings_p_up : priv->tx_ring_num[t];
+
+		for (i = 0; i < priv->tx_ring_num[t]; i++) {
+			/* Configure cq */
+			cq = priv->tx_cq[t][i];
+			err = mlx4_en_activate_cq(priv, cq, i);
+			if (err) {
+				en_err(priv, "Failed allocating Tx CQ\n");
+				goto tx_err;
+			}
+			err = mlx4_en_set_cq_moder(priv, cq);
+			if (err) {
+				en_err(priv, "Failed setting cq moderation parameters\n");
+				mlx4_en_deactivate_cq(priv, cq);
+				goto tx_err;
+			}
+			en_dbg(DRV, priv,
+			       "Resetting index of collapsed CQ:%d to -1\n", i);
+			cq->buf->wqe_index = cpu_to_be16(0xffff);
+
+			/* Configure ring */
+			tx_ring = priv->tx_ring[t][i];
+			err = mlx4_en_activate_tx_ring(priv, tx_ring,
+						       cq->mcq.cqn,
+						       i / num_tx_rings_p_up);
+			if (err) {
+				en_err(priv, "Failed allocating Tx ring\n");
+				mlx4_en_deactivate_cq(priv, cq);
+				goto tx_err;
+			}
+			clear_bit(MLX4_EN_TX_RING_STATE_RECOVERING, &tx_ring->state);
+			if (t != TX_XDP) {
+				tx_ring->tx_queue = netdev_get_tx_queue(dev, i);
+				tx_ring->recycle_ring = NULL;
+
+				/* Arm CQ for TX completions */
+				mlx4_en_arm_cq(priv, cq);
+
+			} else {
+				mlx4_en_init_tx_xdp_ring_descs(priv, tx_ring);
+				mlx4_en_init_recycle_ring(priv, i);
+				/* XDP TX CQ should never be armed */
+			}
+
+			/* Set initial ownership of all Tx TXBBs to SW (1) */
+			for (j = 0; j < tx_ring->buf_size; j += STAMP_STRIDE)
+				*((u32 *)(tx_ring->buf + j)) = 0xffffffff;
+		}
+	}
+
+	/* Configure port */
+	err = mlx4_SET_PORT_general(mdev->dev, priv->port,
+				    priv->rx_skb_size + ETH_FCS_LEN,
+				    priv->prof->tx_pause,
+				    priv->prof->tx_ppp,
+				    priv->prof->rx_pause,
+				    priv->prof->rx_ppp);
+	if (err) {
+		en_err(priv, "Failed setting port general configurations for port %d, with error %d\n",
+		       priv->port, err);
+		goto tx_err;
+	}
+
+	err = mlx4_SET_PORT_user_mtu(mdev->dev, priv->port, dev->mtu);
+	if (err) {
+		en_err(priv, "Failed to pass user MTU(%d) to Firmware for port %d, with error %d\n",
+		       dev->mtu, priv->port, err);
+		goto tx_err;
+	}
+
+	/* Set default qp number */
+	err = mlx4_SET_PORT_qpn_calc(mdev->dev, priv->port, priv->base_qpn, 0);
+	if (err) {
+		en_err(priv, "Failed setting default qp numbers\n");
+		goto tx_err;
+	}
+
+	if (mdev->dev->caps.tunnel_offload_mode == MLX4_TUNNEL_OFFLOAD_MODE_VXLAN) {
+		err = mlx4_SET_PORT_VXLAN(mdev->dev, priv->port, VXLAN_STEER_BY_OUTER_MAC, 1);
+		if (err) {
+			en_err(priv, "Failed setting port L2 tunnel configuration, err %d\n",
+			       err);
+			goto tx_err;
+		}
+	}
+
+	/* Init port */
+	en_dbg(HW, priv, "Initializing port\n");
+	err = mlx4_INIT_PORT(mdev->dev, priv->port);
+	if (err) {
+		en_err(priv, "Failed Initializing port\n");
+		goto tx_err;
+	}
+
+	/* Set Unicast and VXLAN steering rules */
+	if (mdev->dev->caps.steering_mode != MLX4_STEERING_MODE_A0 &&
+	    mlx4_en_set_rss_steer_rules(priv))
+		mlx4_warn(mdev, "Failed setting steering rules\n");
+
+	/* Attach rx QP to bradcast address */
+	eth_broadcast_addr(&mc_list[10]);
+	mc_list[5] = priv->port; /* needed for B0 steering support */
+	if (mlx4_multicast_attach(mdev->dev, priv->rss_map.indir_qp, mc_list,
+				  priv->port, 0, MLX4_PROT_ETH,
+				  &priv->broadcast_id))
+		mlx4_warn(mdev, "Failed Attaching Broadcast\n");
+
+	/* Must redo promiscuous mode setup. */
+	priv->flags &= ~(MLX4_EN_FLAG_PROMISC | MLX4_EN_FLAG_MC_PROMISC);
+
+	/* Schedule multicast task to populate multicast list */
+	queue_work(mdev->workqueue, &priv->rx_mode_task);
+
+	if (priv->mdev->dev->caps.tunnel_offload_mode == MLX4_TUNNEL_OFFLOAD_MODE_VXLAN)
+		udp_tunnel_nic_reset_ntf(dev);
+
+	priv->port_up = true;
+
+	/* Process all completions if exist to prevent
+	 * the queues freezing if they are full
+	 */
+	for (i = 0; i < priv->rx_ring_num; i++) {
+		local_bh_disable();
+		napi_schedule(&priv->rx_cq[i]->napi);
+		local_bh_enable();
+	}
+
+	clear_bit(MLX4_EN_STATE_FLAG_RESTARTING, &priv->state);
+	netif_tx_start_all_queues(dev);
+	netif_device_attach(dev);
+
+	return 0;
+
+tx_err:
+	if (t == MLX4_EN_NUM_TX_TYPES) {
+		t--;
+		i = priv->tx_ring_num[t];
+	}
+	while (t >= 0) {
+		while (i--) {
+			mlx4_en_deactivate_tx_ring(priv, priv->tx_ring[t][i]);
+			mlx4_en_deactivate_cq(priv, priv->tx_cq[t][i]);
+		}
+		if (!t--)
+			break;
+		i = priv->tx_ring_num[t];
+	}
+	mlx4_en_destroy_drop_qp(priv);
+rss_err:
+	mlx4_en_release_rss_steer(priv);
+mac_err:
+	mlx4_en_put_qp(priv);
+cq_err:
+	while (rx_index--) {
+		mlx4_en_deactivate_cq(priv, priv->rx_cq[rx_index]);
+		mlx4_en_free_affinity_hint(priv, rx_index);
+	}
+	for (i = 0; i < priv->rx_ring_num; i++)
+		mlx4_en_deactivate_rx_ring(priv, priv->rx_ring[i]);
+
+	return err; /* need to close devices */
+}
+
+
+void mlx4_en_stop_port(struct net_device *dev, int detach)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	struct mlx4_en_dev *mdev = priv->mdev;
+	struct mlx4_en_mc_list *mclist, *tmp;
+	struct ethtool_flow_id *flow, *tmp_flow;
+	int i, t;
+	u8 mc_list[16] = {0};
+
+	if (!priv->port_up) {
+		en_dbg(DRV, priv, "stop port called while port already down\n");
+		return;
+	}
+
+	/* close port*/
+	mlx4_CLOSE_PORT(mdev->dev, priv->port);
+
+	/* Synchronize with tx routine */
+	netif_tx_lock_bh(dev);
+	if (detach)
+		netif_device_detach(dev);
+	netif_tx_stop_all_queues(dev);
+	netif_tx_unlock_bh(dev);
+
+	netif_tx_disable(dev);
+
+	spin_lock_bh(&priv->stats_lock);
+	mlx4_en_fold_software_stats(dev);
+	/* Set port as not active */
+	priv->port_up = false;
+	spin_unlock_bh(&priv->stats_lock);
+
+	priv->counter_index = MLX4_SINK_COUNTER_INDEX(mdev->dev);
+
+	/* Promsicuous mode */
+	if (mdev->dev->caps.steering_mode ==
+	    MLX4_STEERING_MODE_DEVICE_MANAGED) {
+		priv->flags &= ~(MLX4_EN_FLAG_PROMISC |
+				 MLX4_EN_FLAG_MC_PROMISC);
+		mlx4_flow_steer_promisc_remove(mdev->dev,
+					       priv->port,
+					       MLX4_FS_ALL_DEFAULT);
+		mlx4_flow_steer_promisc_remove(mdev->dev,
+					       priv->port,
+					       MLX4_FS_MC_DEFAULT);
+	} else if (priv->flags & MLX4_EN_FLAG_PROMISC) {
+		priv->flags &= ~MLX4_EN_FLAG_PROMISC;
+
+		/* Disable promiscouos mode */
+		mlx4_unicast_promisc_remove(mdev->dev, priv->base_qpn,
+					    priv->port);
+
+		/* Disable Multicast promisc */
+		if (priv->flags & MLX4_EN_FLAG_MC_PROMISC) {
+			mlx4_multicast_promisc_remove(mdev->dev, priv->base_qpn,
+						      priv->port);
+			priv->flags &= ~MLX4_EN_FLAG_MC_PROMISC;
+		}
+	}
+
+	/* Detach All multicasts */
+	eth_broadcast_addr(&mc_list[10]);
+	mc_list[5] = priv->port; /* needed for B0 steering support */
+	mlx4_multicast_detach(mdev->dev, priv->rss_map.indir_qp, mc_list,
+			      MLX4_PROT_ETH, priv->broadcast_id);
+	list_for_each_entry(mclist, &priv->curr_list, list) {
+		memcpy(&mc_list[10], mclist->addr, ETH_ALEN);
+		mc_list[5] = priv->port;
+		mlx4_multicast_detach(mdev->dev, priv->rss_map.indir_qp,
+				      mc_list, MLX4_PROT_ETH, mclist->reg_id);
+		if (mclist->tunnel_reg_id)
+			mlx4_flow_detach(mdev->dev, mclist->tunnel_reg_id);
+	}
+	mlx4_en_clear_list(dev);
+	list_for_each_entry_safe(mclist, tmp, &priv->curr_list, list) {
+		list_del(&mclist->list);
+		kfree(mclist);
+	}
+
+	/* Flush multicast filter */
+	mlx4_SET_MCAST_FLTR(mdev->dev, priv->port, 0, 1, MLX4_MCAST_CONFIG);
+
+	/* Remove flow steering rules for the port*/
+	if (mdev->dev->caps.steering_mode ==
+	    MLX4_STEERING_MODE_DEVICE_MANAGED) {
+		ASSERT_RTNL();
+		list_for_each_entry_safe(flow, tmp_flow,
+					 &priv->ethtool_list, list) {
+			mlx4_flow_detach(mdev->dev, flow->id);
+			list_del(&flow->list);
+		}
+	}
+
+	mlx4_en_destroy_drop_qp(priv);
+
+	/* Free TX Rings */
+	for (t = 0; t < MLX4_EN_NUM_TX_TYPES; t++) {
+		for (i = 0; i < priv->tx_ring_num[t]; i++) {
+			mlx4_en_deactivate_tx_ring(priv, priv->tx_ring[t][i]);
+			mlx4_en_deactivate_cq(priv, priv->tx_cq[t][i]);
+		}
+	}
+	msleep(10);
+
+	for (t = 0; t < MLX4_EN_NUM_TX_TYPES; t++)
+		for (i = 0; i < priv->tx_ring_num[t]; i++)
+			mlx4_en_free_tx_buf(dev, priv->tx_ring[t][i]);
+
+	if (mdev->dev->caps.steering_mode != MLX4_STEERING_MODE_A0)
+		mlx4_en_delete_rss_steer_rules(priv);
+
+	/* Free RSS qps */
+	mlx4_en_release_rss_steer(priv);
+
+	/* Unregister Mac address for the port */
+	mlx4_en_put_qp(priv);
+	if (!(mdev->dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_REASSIGN_MAC_EN))
+		mdev->mac_removed[priv->port] = 1;
+
+	/* Free RX Rings */
+	for (i = 0; i < priv->rx_ring_num; i++) {
+		struct mlx4_en_cq *cq = priv->rx_cq[i];
+
+		napi_synchronize(&cq->napi);
+		mlx4_en_deactivate_rx_ring(priv, priv->rx_ring[i]);
+		mlx4_en_deactivate_cq(priv, cq);
+
+		mlx4_en_free_affinity_hint(priv, i);
+	}
+}
+
+static void mlx4_en_restart(struct work_struct *work)
+{
+	struct mlx4_en_priv *priv = container_of(work, struct mlx4_en_priv,
+						 restart_task);
+	struct mlx4_en_dev *mdev = priv->mdev;
+	struct net_device *dev = priv->dev;
+
+	en_dbg(DRV, priv, "Watchdog task called for port %d\n", priv->port);
+
+	rtnl_lock();
+	mutex_lock(&mdev->state_lock);
+	if (priv->port_up) {
+		mlx4_en_stop_port(dev, 1);
+		if (mlx4_en_start_port(dev))
+			en_err(priv, "Failed restarting port %d\n", priv->port);
+	}
+	mutex_unlock(&mdev->state_lock);
+	rtnl_unlock();
+}
+
+static void mlx4_en_clear_stats(struct net_device *dev)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	struct mlx4_en_dev *mdev = priv->mdev;
+	struct mlx4_en_tx_ring **tx_ring;
+	int i;
+
+	if (!mlx4_is_slave(mdev->dev))
+		if (mlx4_en_DUMP_ETH_STATS(mdev, priv->port, 1))
+			en_dbg(HW, priv, "Failed dumping statistics\n");
+
+	memset(&priv->pkstats, 0, sizeof(priv->pkstats));
+	memset(&priv->port_stats, 0, sizeof(priv->port_stats));
+	memset(&priv->rx_flowstats, 0, sizeof(priv->rx_flowstats));
+	memset(&priv->tx_flowstats, 0, sizeof(priv->tx_flowstats));
+	memset(&priv->rx_priority_flowstats, 0,
+	       sizeof(priv->rx_priority_flowstats));
+	memset(&priv->tx_priority_flowstats, 0,
+	       sizeof(priv->tx_priority_flowstats));
+	memset(&priv->pf_stats, 0, sizeof(priv->pf_stats));
+
+	tx_ring = priv->tx_ring[TX];
+	for (i = 0; i < priv->tx_ring_num[TX]; i++) {
+		tx_ring[i]->bytes = 0;
+		tx_ring[i]->packets = 0;
+		tx_ring[i]->tx_csum = 0;
+		tx_ring[i]->tx_dropped = 0;
+		tx_ring[i]->queue_stopped = 0;
+		tx_ring[i]->wake_queue = 0;
+		tx_ring[i]->tso_packets = 0;
+		tx_ring[i]->xmit_more = 0;
+	}
+	for (i = 0; i < priv->rx_ring_num; i++) {
+		priv->rx_ring[i]->bytes = 0;
+		priv->rx_ring[i]->packets = 0;
+		priv->rx_ring[i]->csum_ok = 0;
+		priv->rx_ring[i]->csum_none = 0;
+		priv->rx_ring[i]->csum_complete = 0;
+	}
+}
+
+static int mlx4_en_open(struct net_device *dev)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	struct mlx4_en_dev *mdev = priv->mdev;
+	int err = 0;
+
+	mutex_lock(&mdev->state_lock);
+
+	if (!mdev->device_up) {
+		en_err(priv, "Cannot open - device down/disabled\n");
+		err = -EBUSY;
+		goto out;
+	}
+
+	/* Reset HW statistics and SW counters */
+	mlx4_en_clear_stats(dev);
+
+	err = mlx4_en_start_port(dev);
+	if (err) {
+		en_err(priv, "Failed starting port:%d\n", priv->port);
+		goto out;
+	}
+	mlx4_en_linkstate(priv);
+out:
+	mutex_unlock(&mdev->state_lock);
+	return err;
+}
+
+
+static int mlx4_en_close(struct net_device *dev)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	struct mlx4_en_dev *mdev = priv->mdev;
+
+	en_dbg(IFDOWN, priv, "Close port called\n");
+
+	mutex_lock(&mdev->state_lock);
+
+	mlx4_en_stop_port(dev, 0);
+	netif_carrier_off(dev);
+
+	mutex_unlock(&mdev->state_lock);
+	return 0;
+}
+
+static void mlx4_en_free_resources(struct mlx4_en_priv *priv)
+{
+	int i, t;
+
+#ifdef CONFIG_RFS_ACCEL
+	priv->dev->rx_cpu_rmap = NULL;
+#endif
+
+	for (t = 0; t < MLX4_EN_NUM_TX_TYPES; t++) {
+		for (i = 0; i < priv->tx_ring_num[t]; i++) {
+			if (priv->tx_ring[t] && priv->tx_ring[t][i])
+				mlx4_en_destroy_tx_ring(priv,
+							&priv->tx_ring[t][i]);
+			if (priv->tx_cq[t] && priv->tx_cq[t][i])
+				mlx4_en_destroy_cq(priv, &priv->tx_cq[t][i]);
+		}
+		kfree(priv->tx_ring[t]);
+		kfree(priv->tx_cq[t]);
+	}
+
+	for (i = 0; i < priv->rx_ring_num; i++) {
+		if (priv->rx_ring[i])
+			mlx4_en_destroy_rx_ring(priv, &priv->rx_ring[i],
+				priv->prof->rx_ring_size, priv->stride);
+		if (priv->rx_cq[i])
+			mlx4_en_destroy_cq(priv, &priv->rx_cq[i]);
+	}
+
+}
+
+static int mlx4_en_alloc_resources(struct mlx4_en_priv *priv)
+{
+	struct mlx4_en_port_profile *prof = priv->prof;
+	int i, t;
+	int node;
+
+	/* Create tx Rings */
+	for (t = 0; t < MLX4_EN_NUM_TX_TYPES; t++) {
+		for (i = 0; i < priv->tx_ring_num[t]; i++) {
+			node = cpu_to_node(i % num_online_cpus());
+			if (mlx4_en_create_cq(priv, &priv->tx_cq[t][i],
+					      prof->tx_ring_size, i, t, node))
+				goto err;
+
+			if (mlx4_en_create_tx_ring(priv, &priv->tx_ring[t][i],
+						   prof->tx_ring_size,
+						   TXBB_SIZE, node, i))
+				goto err;
+		}
+	}
+
+	/* Create rx Rings */
+	for (i = 0; i < priv->rx_ring_num; i++) {
+		node = cpu_to_node(i % num_online_cpus());
+		if (mlx4_en_create_cq(priv, &priv->rx_cq[i],
+				      prof->rx_ring_size, i, RX, node))
+			goto err;
+
+		if (mlx4_en_create_rx_ring(priv, &priv->rx_ring[i],
+					   prof->rx_ring_size, priv->stride,
+					   node, i))
+			goto err;
+
+	}
+
+#ifdef CONFIG_RFS_ACCEL
+	priv->dev->rx_cpu_rmap = mlx4_get_cpu_rmap(priv->mdev->dev, priv->port);
+#endif
+
+	return 0;
+
+err:
+	en_err(priv, "Failed to allocate NIC resources\n");
+	for (i = 0; i < priv->rx_ring_num; i++) {
+		if (priv->rx_ring[i])
+			mlx4_en_destroy_rx_ring(priv, &priv->rx_ring[i],
+						prof->rx_ring_size,
+						priv->stride);
+		if (priv->rx_cq[i])
+			mlx4_en_destroy_cq(priv, &priv->rx_cq[i]);
+	}
+	for (t = 0; t < MLX4_EN_NUM_TX_TYPES; t++) {
+		for (i = 0; i < priv->tx_ring_num[t]; i++) {
+			if (priv->tx_ring[t][i])
+				mlx4_en_destroy_tx_ring(priv,
+							&priv->tx_ring[t][i]);
+			if (priv->tx_cq[t][i])
+				mlx4_en_destroy_cq(priv, &priv->tx_cq[t][i]);
+		}
+	}
+	return -ENOMEM;
+}
+
+
+static int mlx4_en_copy_priv(struct mlx4_en_priv *dst,
+			     struct mlx4_en_priv *src,
+			     struct mlx4_en_port_profile *prof)
+{
+	int t;
+
+	memcpy(&dst->hwtstamp_config, &prof->hwtstamp_config,
+	       sizeof(dst->hwtstamp_config));
+	dst->num_tx_rings_p_up = prof->num_tx_rings_p_up;
+	dst->rx_ring_num = prof->rx_ring_num;
+	dst->flags = prof->flags;
+	dst->mdev = src->mdev;
+	dst->port = src->port;
+	dst->dev = src->dev;
+	dst->prof = prof;
+	dst->stride = roundup_pow_of_two(sizeof(struct mlx4_en_rx_desc) +
+					 DS_SIZE * MLX4_EN_MAX_RX_FRAGS);
+
+	for (t = 0; t < MLX4_EN_NUM_TX_TYPES; t++) {
+		dst->tx_ring_num[t] = prof->tx_ring_num[t];
+		if (!dst->tx_ring_num[t])
+			continue;
+
+		dst->tx_ring[t] = kcalloc(MAX_TX_RINGS,
+					  sizeof(struct mlx4_en_tx_ring *),
+					  GFP_KERNEL);
+		if (!dst->tx_ring[t])
+			goto err_free_tx;
+
+		dst->tx_cq[t] = kcalloc(MAX_TX_RINGS,
+					sizeof(struct mlx4_en_cq *),
+					GFP_KERNEL);
+		if (!dst->tx_cq[t]) {
+			kfree(dst->tx_ring[t]);
+			goto err_free_tx;
+		}
+	}
+
+	return 0;
+
+err_free_tx:
+	while (t--) {
+		kfree(dst->tx_ring[t]);
+		kfree(dst->tx_cq[t]);
+	}
+	return -ENOMEM;
+}
+
+static void mlx4_en_update_priv(struct mlx4_en_priv *dst,
+				struct mlx4_en_priv *src)
+{
+	int t;
+	memcpy(dst->rx_ring, src->rx_ring,
+	       sizeof(struct mlx4_en_rx_ring *) * src->rx_ring_num);
+	memcpy(dst->rx_cq, src->rx_cq,
+	       sizeof(struct mlx4_en_cq *) * src->rx_ring_num);
+	memcpy(&dst->hwtstamp_config, &src->hwtstamp_config,
+	       sizeof(dst->hwtstamp_config));
+	for (t = 0; t < MLX4_EN_NUM_TX_TYPES; t++) {
+		dst->tx_ring_num[t] = src->tx_ring_num[t];
+		dst->tx_ring[t] = src->tx_ring[t];
+		dst->tx_cq[t] = src->tx_cq[t];
+	}
+	dst->num_tx_rings_p_up = src->num_tx_rings_p_up;
+	dst->rx_ring_num = src->rx_ring_num;
+	memcpy(dst->prof, src->prof, sizeof(struct mlx4_en_port_profile));
+}
+
+int mlx4_en_try_alloc_resources(struct mlx4_en_priv *priv,
+				struct mlx4_en_priv *tmp,
+				struct mlx4_en_port_profile *prof,
+				bool carry_xdp_prog)
+{
+	struct bpf_prog *xdp_prog;
+	int i, t, ret;
+
+	ret = mlx4_en_copy_priv(tmp, priv, prof);
+	if (ret) {
+		en_warn(priv, "%s: mlx4_en_copy_priv() failed, return\n",
+			__func__);
+		return ret;
+	}
+
+	if (mlx4_en_alloc_resources(tmp)) {
+		en_warn(priv,
+			"%s: Resource allocation failed, using previous configuration\n",
+			__func__);
+		for (t = 0; t < MLX4_EN_NUM_TX_TYPES; t++) {
+			kfree(tmp->tx_ring[t]);
+			kfree(tmp->tx_cq[t]);
+		}
+		return -ENOMEM;
+	}
+
+	/* All rx_rings has the same xdp_prog.  Pick the first one. */
+	xdp_prog = rcu_dereference_protected(
+		priv->rx_ring[0]->xdp_prog,
+		lockdep_is_held(&priv->mdev->state_lock));
+
+	if (xdp_prog && carry_xdp_prog) {
+		bpf_prog_add(xdp_prog, tmp->rx_ring_num);
+		for (i = 0; i < tmp->rx_ring_num; i++)
+			rcu_assign_pointer(tmp->rx_ring[i]->xdp_prog,
+					   xdp_prog);
+	}
+
+	return 0;
+}
+
+void mlx4_en_safe_replace_resources(struct mlx4_en_priv *priv,
+				    struct mlx4_en_priv *tmp)
+{
+	mlx4_en_free_resources(priv);
+	mlx4_en_update_priv(priv, tmp);
+}
+
+void mlx4_en_destroy_netdev(struct net_device *dev)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	struct mlx4_en_dev *mdev = priv->mdev;
+
+	en_dbg(DRV, priv, "Destroying netdev on port:%d\n", priv->port);
+
+	/* Unregister device - this will close the port if it was up */
+	if (priv->registered) {
+		devlink_port_type_clear(mlx4_get_devlink_port(mdev->dev,
+							      priv->port));
+		unregister_netdev(dev);
+	}
+
+	if (priv->allocated)
+		mlx4_free_hwq_res(mdev->dev, &priv->res, MLX4_EN_PAGE_SIZE);
+
+	cancel_delayed_work(&priv->stats_task);
+	cancel_delayed_work(&priv->service_task);
+	/* flush any pending task for this netdev */
+	flush_workqueue(mdev->workqueue);
+
+	if (mdev->dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_TS)
+		mlx4_en_remove_timestamp(mdev);
+
+	/* Detach the netdev so tasks would not attempt to access it */
+	mutex_lock(&mdev->state_lock);
+	mdev->pndev[priv->port] = NULL;
+	mdev->upper[priv->port] = NULL;
+
+#ifdef CONFIG_RFS_ACCEL
+	mlx4_en_cleanup_filters(priv);
+#endif
+
+	mlx4_en_free_resources(priv);
+	mutex_unlock(&mdev->state_lock);
+
+	free_netdev(dev);
+}
+
+static bool mlx4_en_check_xdp_mtu(struct net_device *dev, int mtu)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+
+	if (mtu > MLX4_EN_MAX_XDP_MTU) {
+		en_err(priv, "mtu:%d > max:%d when XDP prog is attached\n",
+		       mtu, MLX4_EN_MAX_XDP_MTU);
+		return false;
+	}
+
+	return true;
+}
+
+static int mlx4_en_change_mtu(struct net_device *dev, int new_mtu)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	struct mlx4_en_dev *mdev = priv->mdev;
+	int err = 0;
+
+	en_dbg(DRV, priv, "Change MTU called - current:%d new:%d\n",
+		 dev->mtu, new_mtu);
+
+	if (priv->tx_ring_num[TX_XDP] &&
+	    !mlx4_en_check_xdp_mtu(dev, new_mtu))
+		return -EOPNOTSUPP;
+
+	dev->mtu = new_mtu;
+
+	if (netif_running(dev)) {
+		mutex_lock(&mdev->state_lock);
+		if (!mdev->device_up) {
+			/* NIC is probably restarting - let restart task reset
+			 * the port */
+			en_dbg(DRV, priv, "Change MTU called with card down!?\n");
+		} else {
+			mlx4_en_stop_port(dev, 1);
+			err = mlx4_en_start_port(dev);
+			if (err) {
+				en_err(priv, "Failed restarting port:%d\n",
+					 priv->port);
+				if (!test_and_set_bit(MLX4_EN_STATE_FLAG_RESTARTING,
+						      &priv->state))
+					queue_work(mdev->workqueue, &priv->restart_task);
+			}
+		}
+		mutex_unlock(&mdev->state_lock);
+	}
+	return 0;
+}
+
+static int mlx4_en_hwtstamp_set(struct net_device *dev, struct ifreq *ifr)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	struct mlx4_en_dev *mdev = priv->mdev;
+	struct hwtstamp_config config;
+
+	if (copy_from_user(&config, ifr->ifr_data, sizeof(config)))
+		return -EFAULT;
+
+	/* device doesn't support time stamping */
+	if (!(mdev->dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_TS))
+		return -EINVAL;
+
+	/* TX HW timestamp */
+	switch (config.tx_type) {
+	case HWTSTAMP_TX_OFF:
+	case HWTSTAMP_TX_ON:
+		break;
+	default:
+		return -ERANGE;
+	}
+
+	/* RX HW timestamp */
+	switch (config.rx_filter) {
+	case HWTSTAMP_FILTER_NONE:
+		break;
+	case HWTSTAMP_FILTER_ALL:
+	case HWTSTAMP_FILTER_SOME:
+	case HWTSTAMP_FILTER_PTP_V1_L4_EVENT:
+	case HWTSTAMP_FILTER_PTP_V1_L4_SYNC:
+	case HWTSTAMP_FILTER_PTP_V1_L4_DELAY_REQ:
+	case HWTSTAMP_FILTER_PTP_V2_L4_EVENT:
+	case HWTSTAMP_FILTER_PTP_V2_L4_SYNC:
+	case HWTSTAMP_FILTER_PTP_V2_L4_DELAY_REQ:
+	case HWTSTAMP_FILTER_PTP_V2_L2_EVENT:
+	case HWTSTAMP_FILTER_PTP_V2_L2_SYNC:
+	case HWTSTAMP_FILTER_PTP_V2_L2_DELAY_REQ:
+	case HWTSTAMP_FILTER_PTP_V2_EVENT:
+	case HWTSTAMP_FILTER_PTP_V2_SYNC:
+	case HWTSTAMP_FILTER_PTP_V2_DELAY_REQ:
+	case HWTSTAMP_FILTER_NTP_ALL:
+		config.rx_filter = HWTSTAMP_FILTER_ALL;
+		break;
+	default:
+		return -ERANGE;
+	}
+
+	if (mlx4_en_reset_config(dev, config, dev->features)) {
+		config.tx_type = HWTSTAMP_TX_OFF;
+		config.rx_filter = HWTSTAMP_FILTER_NONE;
+	}
+
+	return copy_to_user(ifr->ifr_data, &config,
+			    sizeof(config)) ? -EFAULT : 0;
+}
+
+static int mlx4_en_hwtstamp_get(struct net_device *dev, struct ifreq *ifr)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+
+	return copy_to_user(ifr->ifr_data, &priv->hwtstamp_config,
+			    sizeof(priv->hwtstamp_config)) ? -EFAULT : 0;
+}
+
+static int mlx4_en_ioctl(struct net_device *dev, struct ifreq *ifr, int cmd)
+{
+	switch (cmd) {
+	case SIOCSHWTSTAMP:
+		return mlx4_en_hwtstamp_set(dev, ifr);
+	case SIOCGHWTSTAMP:
+		return mlx4_en_hwtstamp_get(dev, ifr);
+	default:
+		return -EOPNOTSUPP;
+	}
+}
+
+static netdev_features_t mlx4_en_fix_features(struct net_device *netdev,
+					      netdev_features_t features)
+{
+	struct mlx4_en_priv *en_priv = netdev_priv(netdev);
+	struct mlx4_en_dev *mdev = en_priv->mdev;
+
+	/* Since there is no support for separate RX C-TAG/S-TAG vlan accel
+	 * enable/disable make sure S-TAG flag is always in same state as
+	 * C-TAG.
+	 */
+	if (features & NETIF_F_HW_VLAN_CTAG_RX &&
+	    !(mdev->dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_SKIP_OUTER_VLAN))
+		features |= NETIF_F_HW_VLAN_STAG_RX;
+	else
+		features &= ~NETIF_F_HW_VLAN_STAG_RX;
+
+	return features;
+}
+
+static int mlx4_en_set_features(struct net_device *netdev,
+		netdev_features_t features)
+{
+	struct mlx4_en_priv *priv = netdev_priv(netdev);
+	bool reset = false;
+	int ret = 0;
+
+	if (DEV_FEATURE_CHANGED(netdev, features, NETIF_F_RXFCS)) {
+		en_info(priv, "Turn %s RX-FCS\n",
+			(features & NETIF_F_RXFCS) ? "ON" : "OFF");
+		reset = true;
+	}
+
+	if (DEV_FEATURE_CHANGED(netdev, features, NETIF_F_RXALL)) {
+		u8 ignore_fcs_value = (features & NETIF_F_RXALL) ? 1 : 0;
+
+		en_info(priv, "Turn %s RX-ALL\n",
+			ignore_fcs_value ? "ON" : "OFF");
+		ret = mlx4_SET_PORT_fcs_check(priv->mdev->dev,
+					      priv->port, ignore_fcs_value);
+		if (ret)
+			return ret;
+	}
+
+	if (DEV_FEATURE_CHANGED(netdev, features, NETIF_F_HW_VLAN_CTAG_RX)) {
+		en_info(priv, "Turn %s RX vlan strip offload\n",
+			(features & NETIF_F_HW_VLAN_CTAG_RX) ? "ON" : "OFF");
+		reset = true;
+	}
+
+	if (DEV_FEATURE_CHANGED(netdev, features, NETIF_F_HW_VLAN_CTAG_TX))
+		en_info(priv, "Turn %s TX vlan strip offload\n",
+			(features & NETIF_F_HW_VLAN_CTAG_TX) ? "ON" : "OFF");
+
+	if (DEV_FEATURE_CHANGED(netdev, features, NETIF_F_HW_VLAN_STAG_TX))
+		en_info(priv, "Turn %s TX S-VLAN strip offload\n",
+			(features & NETIF_F_HW_VLAN_STAG_TX) ? "ON" : "OFF");
+
+	if (DEV_FEATURE_CHANGED(netdev, features, NETIF_F_LOOPBACK)) {
+		en_info(priv, "Turn %s loopback\n",
+			(features & NETIF_F_LOOPBACK) ? "ON" : "OFF");
+		mlx4_en_update_loopback_state(netdev, features);
+	}
+
+	if (reset) {
+		ret = mlx4_en_reset_config(netdev, priv->hwtstamp_config,
+					   features);
+		if (ret)
+			return ret;
+	}
+
+	return 0;
+}
+
+static int mlx4_en_set_vf_mac(struct net_device *dev, int queue, u8 *mac)
+{
+	struct mlx4_en_priv *en_priv = netdev_priv(dev);
+	struct mlx4_en_dev *mdev = en_priv->mdev;
+
+	return mlx4_set_vf_mac(mdev->dev, en_priv->port, queue, mac);
+}
+
+static int mlx4_en_set_vf_vlan(struct net_device *dev, int vf, u16 vlan, u8 qos,
+			       __be16 vlan_proto)
+{
+	struct mlx4_en_priv *en_priv = netdev_priv(dev);
+	struct mlx4_en_dev *mdev = en_priv->mdev;
+
+	return mlx4_set_vf_vlan(mdev->dev, en_priv->port, vf, vlan, qos,
+				vlan_proto);
+}
+
+static int mlx4_en_set_vf_rate(struct net_device *dev, int vf, int min_tx_rate,
+			       int max_tx_rate)
+{
+	struct mlx4_en_priv *en_priv = netdev_priv(dev);
+	struct mlx4_en_dev *mdev = en_priv->mdev;
+
+	return mlx4_set_vf_rate(mdev->dev, en_priv->port, vf, min_tx_rate,
+				max_tx_rate);
+}
+
+static int mlx4_en_set_vf_spoofchk(struct net_device *dev, int vf, bool setting)
+{
+	struct mlx4_en_priv *en_priv = netdev_priv(dev);
+	struct mlx4_en_dev *mdev = en_priv->mdev;
+
+	return mlx4_set_vf_spoofchk(mdev->dev, en_priv->port, vf, setting);
+}
+
+static int mlx4_en_get_vf_config(struct net_device *dev, int vf, struct ifla_vf_info *ivf)
+{
+	struct mlx4_en_priv *en_priv = netdev_priv(dev);
+	struct mlx4_en_dev *mdev = en_priv->mdev;
+
+	return mlx4_get_vf_config(mdev->dev, en_priv->port, vf, ivf);
+}
+
+static int mlx4_en_set_vf_link_state(struct net_device *dev, int vf, int link_state)
+{
+	struct mlx4_en_priv *en_priv = netdev_priv(dev);
+	struct mlx4_en_dev *mdev = en_priv->mdev;
+
+	return mlx4_set_vf_link_state(mdev->dev, en_priv->port, vf, link_state);
+}
+
+static int mlx4_en_get_vf_stats(struct net_device *dev, int vf,
+				struct ifla_vf_stats *vf_stats)
+{
+	struct mlx4_en_priv *en_priv = netdev_priv(dev);
+	struct mlx4_en_dev *mdev = en_priv->mdev;
+
+	return mlx4_get_vf_stats(mdev->dev, en_priv->port, vf, vf_stats);
+}
+
+#define PORT_ID_BYTE_LEN 8
+static int mlx4_en_get_phys_port_id(struct net_device *dev,
+				    struct netdev_phys_item_id *ppid)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	struct mlx4_dev *mdev = priv->mdev->dev;
+	int i;
+	u64 phys_port_id = mdev->caps.phys_port_id[priv->port];
+
+	if (!phys_port_id)
+		return -EOPNOTSUPP;
+
+	ppid->id_len = sizeof(phys_port_id);
+	for (i = PORT_ID_BYTE_LEN - 1; i >= 0; --i) {
+		ppid->id[i] =  phys_port_id & 0xff;
+		phys_port_id >>= 8;
+	}
+	return 0;
+}
+
+static int mlx4_udp_tunnel_sync(struct net_device *dev, unsigned int table)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	struct udp_tunnel_info ti;
+	int ret;
+
+	udp_tunnel_nic_get_port(dev, table, 0, &ti);
+	priv->vxlan_port = ti.port;
+
+	ret = mlx4_config_vxlan_port(priv->mdev->dev, priv->vxlan_port);
+	if (ret)
+		return ret;
+
+	return mlx4_SET_PORT_VXLAN(priv->mdev->dev, priv->port,
+				   VXLAN_STEER_BY_OUTER_MAC,
+				   !!priv->vxlan_port);
+}
+
+static const struct udp_tunnel_nic_info mlx4_udp_tunnels = {
+	.sync_table	= mlx4_udp_tunnel_sync,
+	.flags		= UDP_TUNNEL_NIC_INFO_MAY_SLEEP |
+			  UDP_TUNNEL_NIC_INFO_IPV4_ONLY,
+	.tables		= {
+		{ .n_entries = 1, .tunnel_types = UDP_TUNNEL_TYPE_VXLAN, },
+	},
+};
+
+static netdev_features_t mlx4_en_features_check(struct sk_buff *skb,
+						struct net_device *dev,
+						netdev_features_t features)
+{
+	features = vlan_features_check(skb, features);
+	features = vxlan_features_check(skb, features);
+
+	/* The ConnectX-3 doesn't support outer IPv6 checksums but it does
+	 * support inner IPv6 checksums and segmentation so  we need to
+	 * strip that feature if this is an IPv6 encapsulated frame.
+	 */
+	if (skb->encapsulation &&
+	    (skb->ip_summed == CHECKSUM_PARTIAL)) {
+		struct mlx4_en_priv *priv = netdev_priv(dev);
+
+		if (!priv->vxlan_port ||
+		    (ip_hdr(skb)->version != 4) ||
+		    (udp_hdr(skb)->dest != priv->vxlan_port))
+			features &= ~(NETIF_F_CSUM_MASK | NETIF_F_GSO_MASK);
+	}
+
+	return features;
+}
+
+static int mlx4_en_set_tx_maxrate(struct net_device *dev, int queue_index, u32 maxrate)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	struct mlx4_en_tx_ring *tx_ring = priv->tx_ring[TX][queue_index];
+	struct mlx4_update_qp_params params;
+	int err;
+
+	if (!(priv->mdev->dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_QP_RATE_LIMIT))
+		return -EOPNOTSUPP;
+
+	/* rate provided to us in Mbs, check if it fits into 12 bits, if not use Gbs */
+	if (maxrate >> 12) {
+		params.rate_unit = MLX4_QP_RATE_LIMIT_GBS;
+		params.rate_val  = maxrate / 1000;
+	} else if (maxrate) {
+		params.rate_unit = MLX4_QP_RATE_LIMIT_MBS;
+		params.rate_val  = maxrate;
+	} else { /* zero serves to revoke the QP rate-limitation */
+		params.rate_unit = 0;
+		params.rate_val  = 0;
+	}
+
+	err = mlx4_update_qp(priv->mdev->dev, tx_ring->qpn, MLX4_UPDATE_QP_RATE_LIMIT,
+			     &params);
+	return err;
+}
+
+static int mlx4_xdp_set(struct net_device *dev, struct bpf_prog *prog)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	struct mlx4_en_dev *mdev = priv->mdev;
+	struct mlx4_en_port_profile new_prof;
+	struct bpf_prog *old_prog;
+	struct mlx4_en_priv *tmp;
+	int tx_changed = 0;
+	int xdp_ring_num;
+	int port_up = 0;
+	int err;
+	int i;
+
+	xdp_ring_num = prog ? priv->rx_ring_num : 0;
+
+	/* No need to reconfigure buffers when simply swapping the
+	 * program for a new one.
+	 */
+	if (priv->tx_ring_num[TX_XDP] == xdp_ring_num) {
+		if (prog)
+			bpf_prog_add(prog, priv->rx_ring_num - 1);
+
+		mutex_lock(&mdev->state_lock);
+		for (i = 0; i < priv->rx_ring_num; i++) {
+			old_prog = rcu_dereference_protected(
+					priv->rx_ring[i]->xdp_prog,
+					lockdep_is_held(&mdev->state_lock));
+			rcu_assign_pointer(priv->rx_ring[i]->xdp_prog, prog);
+			if (old_prog)
+				bpf_prog_put(old_prog);
+		}
+		mutex_unlock(&mdev->state_lock);
+		return 0;
+	}
+
+	if (!mlx4_en_check_xdp_mtu(dev, dev->mtu))
+		return -EOPNOTSUPP;
+
+	tmp = kzalloc(sizeof(*tmp), GFP_KERNEL);
+	if (!tmp)
+		return -ENOMEM;
+
+	if (prog)
+		bpf_prog_add(prog, priv->rx_ring_num - 1);
+
+	mutex_lock(&mdev->state_lock);
+	memcpy(&new_prof, priv->prof, sizeof(struct mlx4_en_port_profile));
+	new_prof.tx_ring_num[TX_XDP] = xdp_ring_num;
+
+	if (priv->tx_ring_num[TX] + xdp_ring_num > MAX_TX_RINGS) {
+		tx_changed = 1;
+		new_prof.tx_ring_num[TX] =
+			MAX_TX_RINGS - ALIGN(xdp_ring_num, priv->prof->num_up);
+		en_warn(priv, "Reducing the number of TX rings, to not exceed the max total rings number.\n");
+	}
+
+	err = mlx4_en_try_alloc_resources(priv, tmp, &new_prof, false);
+	if (err) {
+		if (prog)
+			bpf_prog_sub(prog, priv->rx_ring_num - 1);
+		goto unlock_out;
+	}
+
+	if (priv->port_up) {
+		port_up = 1;
+		mlx4_en_stop_port(dev, 1);
+	}
+
+	mlx4_en_safe_replace_resources(priv, tmp);
+	if (tx_changed)
+		netif_set_real_num_tx_queues(dev, priv->tx_ring_num[TX]);
+
+	for (i = 0; i < priv->rx_ring_num; i++) {
+		old_prog = rcu_dereference_protected(
+					priv->rx_ring[i]->xdp_prog,
+					lockdep_is_held(&mdev->state_lock));
+		rcu_assign_pointer(priv->rx_ring[i]->xdp_prog, prog);
+		if (old_prog)
+			bpf_prog_put(old_prog);
+	}
+
+	if (port_up) {
+		err = mlx4_en_start_port(dev);
+		if (err) {
+			en_err(priv, "Failed starting port %d for XDP change\n",
+			       priv->port);
+			if (!test_and_set_bit(MLX4_EN_STATE_FLAG_RESTARTING, &priv->state))
+				queue_work(mdev->workqueue, &priv->restart_task);
+		}
+	}
+
+unlock_out:
+	mutex_unlock(&mdev->state_lock);
+	kfree(tmp);
+	return err;
+}
+
+static int mlx4_xdp(struct net_device *dev, struct netdev_bpf *xdp)
+{
+	switch (xdp->command) {
+	case XDP_SETUP_PROG:
+		return mlx4_xdp_set(dev, xdp->prog);
+	default:
+		return -EINVAL;
+	}
+}
+
+static const struct net_device_ops mlx4_netdev_ops = {
+	.ndo_open		= mlx4_en_open,
+	.ndo_stop		= mlx4_en_close,
+	.ndo_start_xmit		= mlx4_en_xmit,
+	.ndo_select_queue	= mlx4_en_select_queue,
+	.ndo_get_stats64	= mlx4_en_get_stats64,
+	.ndo_set_rx_mode	= mlx4_en_set_rx_mode,
+	.ndo_set_mac_address	= mlx4_en_set_mac,
+	.ndo_validate_addr	= eth_validate_addr,
+	.ndo_change_mtu		= mlx4_en_change_mtu,
+	.ndo_eth_ioctl		= mlx4_en_ioctl,
+	.ndo_tx_timeout		= mlx4_en_tx_timeout,
+	.ndo_vlan_rx_add_vid	= mlx4_en_vlan_rx_add_vid,
+	.ndo_vlan_rx_kill_vid	= mlx4_en_vlan_rx_kill_vid,
+	.ndo_set_features	= mlx4_en_set_features,
+	.ndo_fix_features	= mlx4_en_fix_features,
+	.ndo_setup_tc		= __mlx4_en_setup_tc,
+#ifdef CONFIG_RFS_ACCEL
+	.ndo_rx_flow_steer	= mlx4_en_filter_rfs,
+#endif
+	.ndo_get_phys_port_id	= mlx4_en_get_phys_port_id,
+	.ndo_features_check	= mlx4_en_features_check,
+	.ndo_set_tx_maxrate	= mlx4_en_set_tx_maxrate,
+	.ndo_bpf		= mlx4_xdp,
+};
+
+static const struct net_device_ops mlx4_netdev_ops_master = {
+	.ndo_open		= mlx4_en_open,
+	.ndo_stop		= mlx4_en_close,
+	.ndo_start_xmit		= mlx4_en_xmit,
+	.ndo_select_queue	= mlx4_en_select_queue,
+	.ndo_get_stats64	= mlx4_en_get_stats64,
+	.ndo_set_rx_mode	= mlx4_en_set_rx_mode,
+	.ndo_set_mac_address	= mlx4_en_set_mac,
+	.ndo_validate_addr	= eth_validate_addr,
+	.ndo_change_mtu		= mlx4_en_change_mtu,
+	.ndo_tx_timeout		= mlx4_en_tx_timeout,
+	.ndo_vlan_rx_add_vid	= mlx4_en_vlan_rx_add_vid,
+	.ndo_vlan_rx_kill_vid	= mlx4_en_vlan_rx_kill_vid,
+	.ndo_set_vf_mac		= mlx4_en_set_vf_mac,
+	.ndo_set_vf_vlan	= mlx4_en_set_vf_vlan,
+	.ndo_set_vf_rate	= mlx4_en_set_vf_rate,
+	.ndo_set_vf_spoofchk	= mlx4_en_set_vf_spoofchk,
+	.ndo_set_vf_link_state	= mlx4_en_set_vf_link_state,
+	.ndo_get_vf_stats       = mlx4_en_get_vf_stats,
+	.ndo_get_vf_config	= mlx4_en_get_vf_config,
+	.ndo_set_features	= mlx4_en_set_features,
+	.ndo_fix_features	= mlx4_en_fix_features,
+	.ndo_setup_tc		= __mlx4_en_setup_tc,
+#ifdef CONFIG_RFS_ACCEL
+	.ndo_rx_flow_steer	= mlx4_en_filter_rfs,
+#endif
+	.ndo_get_phys_port_id	= mlx4_en_get_phys_port_id,
+	.ndo_features_check	= mlx4_en_features_check,
+	.ndo_set_tx_maxrate	= mlx4_en_set_tx_maxrate,
+	.ndo_bpf		= mlx4_xdp,
+};
+
+struct mlx4_en_bond {
+	struct work_struct work;
+	struct mlx4_en_priv *priv;
+	int is_bonded;
+	struct mlx4_port_map port_map;
+};
+
+static void mlx4_en_bond_work(struct work_struct *work)
+{
+	struct mlx4_en_bond *bond = container_of(work,
+						     struct mlx4_en_bond,
+						     work);
+	int err = 0;
+	struct mlx4_dev *dev = bond->priv->mdev->dev;
+
+	if (bond->is_bonded) {
+		if (!mlx4_is_bonded(dev)) {
+			err = mlx4_bond(dev);
+			if (err)
+				en_err(bond->priv, "Fail to bond device\n");
+		}
+		if (!err) {
+			err = mlx4_port_map_set(dev, &bond->port_map);
+			if (err)
+				en_err(bond->priv, "Fail to set port map [%d][%d]: %d\n",
+				       bond->port_map.port1,
+				       bond->port_map.port2,
+				       err);
+		}
+	} else if (mlx4_is_bonded(dev)) {
+		err = mlx4_unbond(dev);
+		if (err)
+			en_err(bond->priv, "Fail to unbond device\n");
+	}
+	dev_put(bond->priv->dev);
+	kfree(bond);
+}
+
+static int mlx4_en_queue_bond_work(struct mlx4_en_priv *priv, int is_bonded,
+				   u8 v2p_p1, u8 v2p_p2)
+{
+	struct mlx4_en_bond *bond = NULL;
+
+	bond = kzalloc(sizeof(*bond), GFP_ATOMIC);
+	if (!bond)
+		return -ENOMEM;
+
+	INIT_WORK(&bond->work, mlx4_en_bond_work);
+	bond->priv = priv;
+	bond->is_bonded = is_bonded;
+	bond->port_map.port1 = v2p_p1;
+	bond->port_map.port2 = v2p_p2;
+	dev_hold(priv->dev);
+	queue_work(priv->mdev->workqueue, &bond->work);
+	return 0;
+}
+
+int mlx4_en_netdev_event(struct notifier_block *this,
+			 unsigned long event, void *ptr)
+{
+	struct net_device *ndev = netdev_notifier_info_to_dev(ptr);
+	u8 port = 0;
+	struct mlx4_en_dev *mdev;
+	struct mlx4_dev *dev;
+	int i, num_eth_ports = 0;
+	bool do_bond = true;
+	struct mlx4_en_priv *priv;
+	u8 v2p_port1 = 0;
+	u8 v2p_port2 = 0;
+
+	if (!net_eq(dev_net(ndev), &init_net))
+		return NOTIFY_DONE;
+
+	mdev = container_of(this, struct mlx4_en_dev, nb);
+	dev = mdev->dev;
+
+	/* Go into this mode only when two network devices set on two ports
+	 * of the same mlx4 device are slaves of the same bonding master
+	 */
+	mlx4_foreach_port(i, dev, MLX4_PORT_TYPE_ETH) {
+		++num_eth_ports;
+		if (!port && (mdev->pndev[i] == ndev))
+			port = i;
+		mdev->upper[i] = mdev->pndev[i] ?
+			netdev_master_upper_dev_get(mdev->pndev[i]) : NULL;
+		/* condition not met: network device is a slave */
+		if (!mdev->upper[i])
+			do_bond = false;
+		if (num_eth_ports < 2)
+			continue;
+		/* condition not met: same master */
+		if (mdev->upper[i] != mdev->upper[i-1])
+			do_bond = false;
+	}
+	/* condition not met: 2 salves */
+	do_bond = (num_eth_ports ==  2) ? do_bond : false;
+
+	/* handle only events that come with enough info */
+	if ((do_bond && (event != NETDEV_BONDING_INFO)) || !port)
+		return NOTIFY_DONE;
+
+	priv = netdev_priv(ndev);
+	if (do_bond) {
+		struct netdev_notifier_bonding_info *notifier_info = ptr;
+		struct netdev_bonding_info *bonding_info =
+			&notifier_info->bonding_info;
+
+		/* required mode 1, 2 or 4 */
+		if ((bonding_info->master.bond_mode != BOND_MODE_ACTIVEBACKUP) &&
+		    (bonding_info->master.bond_mode != BOND_MODE_XOR) &&
+		    (bonding_info->master.bond_mode != BOND_MODE_8023AD))
+			do_bond = false;
+
+		/* require exactly 2 slaves */
+		if (bonding_info->master.num_slaves != 2)
+			do_bond = false;
+
+		/* calc v2p */
+		if (do_bond) {
+			if (bonding_info->master.bond_mode ==
+			    BOND_MODE_ACTIVEBACKUP) {
+				/* in active-backup mode virtual ports are
+				 * mapped to the physical port of the active
+				 * slave */
+				if (bonding_info->slave.state ==
+				    BOND_STATE_BACKUP) {
+					if (port == 1) {
+						v2p_port1 = 2;
+						v2p_port2 = 2;
+					} else {
+						v2p_port1 = 1;
+						v2p_port2 = 1;
+					}
+				} else { /* BOND_STATE_ACTIVE */
+					if (port == 1) {
+						v2p_port1 = 1;
+						v2p_port2 = 1;
+					} else {
+						v2p_port1 = 2;
+						v2p_port2 = 2;
+					}
+				}
+			} else { /* Active-Active */
+				/* in active-active mode a virtual port is
+				 * mapped to the native physical port if and only
+				 * if the physical port is up */
+				__s8 link = bonding_info->slave.link;
+
+				if (port == 1)
+					v2p_port2 = 2;
+				else
+					v2p_port1 = 1;
+				if ((link == BOND_LINK_UP) ||
+				    (link == BOND_LINK_FAIL)) {
+					if (port == 1)
+						v2p_port1 = 1;
+					else
+						v2p_port2 = 2;
+				} else { /* BOND_LINK_DOWN || BOND_LINK_BACK */
+					if (port == 1)
+						v2p_port1 = 2;
+					else
+						v2p_port2 = 1;
+				}
+			}
+		}
+	}
+
+	mlx4_en_queue_bond_work(priv, do_bond,
+				v2p_port1, v2p_port2);
+
+	return NOTIFY_DONE;
+}
+
+void mlx4_en_update_pfc_stats_bitmap(struct mlx4_dev *dev,
+				     struct mlx4_en_stats_bitmap *stats_bitmap,
+				     u8 rx_ppp, u8 rx_pause,
+				     u8 tx_ppp, u8 tx_pause)
+{
+	int last_i = NUM_MAIN_STATS + NUM_PORT_STATS + NUM_PF_STATS;
+
+	if (!mlx4_is_slave(dev) &&
+	    (dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_FLOWSTATS_EN)) {
+		mutex_lock(&stats_bitmap->mutex);
+		bitmap_clear(stats_bitmap->bitmap, last_i, NUM_FLOW_STATS);
+
+		if (rx_ppp)
+			bitmap_set(stats_bitmap->bitmap, last_i,
+				   NUM_FLOW_PRIORITY_STATS_RX);
+		last_i += NUM_FLOW_PRIORITY_STATS_RX;
+
+		if (rx_pause && !(rx_ppp))
+			bitmap_set(stats_bitmap->bitmap, last_i,
+				   NUM_FLOW_STATS_RX);
+		last_i += NUM_FLOW_STATS_RX;
+
+		if (tx_ppp)
+			bitmap_set(stats_bitmap->bitmap, last_i,
+				   NUM_FLOW_PRIORITY_STATS_TX);
+		last_i += NUM_FLOW_PRIORITY_STATS_TX;
+
+		if (tx_pause && !(tx_ppp))
+			bitmap_set(stats_bitmap->bitmap, last_i,
+				   NUM_FLOW_STATS_TX);
+		last_i += NUM_FLOW_STATS_TX;
+
+		mutex_unlock(&stats_bitmap->mutex);
+	}
+}
+
+void mlx4_en_set_stats_bitmap(struct mlx4_dev *dev,
+			      struct mlx4_en_stats_bitmap *stats_bitmap,
+			      u8 rx_ppp, u8 rx_pause,
+			      u8 tx_ppp, u8 tx_pause)
+{
+	int last_i = 0;
+
+	mutex_init(&stats_bitmap->mutex);
+	bitmap_zero(stats_bitmap->bitmap, NUM_ALL_STATS);
+
+	if (mlx4_is_slave(dev)) {
+		bitmap_set(stats_bitmap->bitmap, last_i +
+					 MLX4_FIND_NETDEV_STAT(rx_packets), 1);
+		bitmap_set(stats_bitmap->bitmap, last_i +
+					 MLX4_FIND_NETDEV_STAT(tx_packets), 1);
+		bitmap_set(stats_bitmap->bitmap, last_i +
+					 MLX4_FIND_NETDEV_STAT(rx_bytes), 1);
+		bitmap_set(stats_bitmap->bitmap, last_i +
+					 MLX4_FIND_NETDEV_STAT(tx_bytes), 1);
+		bitmap_set(stats_bitmap->bitmap, last_i +
+					 MLX4_FIND_NETDEV_STAT(rx_dropped), 1);
+		bitmap_set(stats_bitmap->bitmap, last_i +
+					 MLX4_FIND_NETDEV_STAT(tx_dropped), 1);
+	} else {
+		bitmap_set(stats_bitmap->bitmap, last_i, NUM_MAIN_STATS);
+	}
+	last_i += NUM_MAIN_STATS;
+
+	bitmap_set(stats_bitmap->bitmap, last_i, NUM_PORT_STATS);
+	last_i += NUM_PORT_STATS;
+
+	if (mlx4_is_master(dev))
+		bitmap_set(stats_bitmap->bitmap, last_i,
+			   NUM_PF_STATS);
+	last_i += NUM_PF_STATS;
+
+	mlx4_en_update_pfc_stats_bitmap(dev, stats_bitmap,
+					rx_ppp, rx_pause,
+					tx_ppp, tx_pause);
+	last_i += NUM_FLOW_STATS;
+
+	if (!mlx4_is_slave(dev))
+		bitmap_set(stats_bitmap->bitmap, last_i, NUM_PKT_STATS);
+	last_i += NUM_PKT_STATS;
+
+	bitmap_set(stats_bitmap->bitmap, last_i, NUM_XDP_STATS);
+	last_i += NUM_XDP_STATS;
+
+	if (!mlx4_is_slave(dev))
+		bitmap_set(stats_bitmap->bitmap, last_i, NUM_PHY_STATS);
+	last_i += NUM_PHY_STATS;
+}
+
+int mlx4_en_init_netdev(struct mlx4_en_dev *mdev, int port,
+			struct mlx4_en_port_profile *prof)
+{
+	struct net_device *dev;
+	struct mlx4_en_priv *priv;
+	int i, t;
+	int err;
+
+	dev = alloc_etherdev_mqs(sizeof(struct mlx4_en_priv),
+				 MAX_TX_RINGS, MAX_RX_RINGS);
+	if (dev == NULL)
+		return -ENOMEM;
+
+	netif_set_real_num_tx_queues(dev, prof->tx_ring_num[TX]);
+	netif_set_real_num_rx_queues(dev, prof->rx_ring_num);
+
+	SET_NETDEV_DEV(dev, &mdev->dev->persist->pdev->dev);
+	dev->dev_port = port - 1;
+
+	/*
+	 * Initialize driver private data
+	 */
+
+	priv = netdev_priv(dev);
+	memset(priv, 0, sizeof(struct mlx4_en_priv));
+	priv->counter_index = MLX4_SINK_COUNTER_INDEX(mdev->dev);
+	spin_lock_init(&priv->stats_lock);
+	INIT_WORK(&priv->rx_mode_task, mlx4_en_do_set_rx_mode);
+	INIT_WORK(&priv->restart_task, mlx4_en_restart);
+	INIT_WORK(&priv->linkstate_task, mlx4_en_linkstate_work);
+	INIT_DELAYED_WORK(&priv->stats_task, mlx4_en_do_get_stats);
+	INIT_DELAYED_WORK(&priv->service_task, mlx4_en_service_task);
+#ifdef CONFIG_RFS_ACCEL
+	INIT_LIST_HEAD(&priv->filters);
+	spin_lock_init(&priv->filters_lock);
+#endif
+
+	priv->dev = dev;
+	priv->mdev = mdev;
+	priv->ddev = &mdev->pdev->dev;
+	priv->prof = prof;
+	priv->port = port;
+	priv->port_up = false;
+	priv->flags = prof->flags;
+	priv->pflags = MLX4_EN_PRIV_FLAGS_BLUEFLAME;
+	priv->ctrl_flags = cpu_to_be32(MLX4_WQE_CTRL_CQ_UPDATE |
+			MLX4_WQE_CTRL_SOLICITED);
+	priv->num_tx_rings_p_up = mdev->profile.max_num_tx_rings_p_up;
+	priv->tx_work_limit = MLX4_EN_DEFAULT_TX_WORK;
+	netdev_rss_key_fill(priv->rss_key, sizeof(priv->rss_key));
+
+	for (t = 0; t < MLX4_EN_NUM_TX_TYPES; t++) {
+		priv->tx_ring_num[t] = prof->tx_ring_num[t];
+		if (!priv->tx_ring_num[t])
+			continue;
+
+		priv->tx_ring[t] = kcalloc(MAX_TX_RINGS,
+					   sizeof(struct mlx4_en_tx_ring *),
+					   GFP_KERNEL);
+		if (!priv->tx_ring[t]) {
+			err = -ENOMEM;
+			goto out;
+		}
+		priv->tx_cq[t] = kcalloc(MAX_TX_RINGS,
+					 sizeof(struct mlx4_en_cq *),
+					 GFP_KERNEL);
+		if (!priv->tx_cq[t]) {
+			err = -ENOMEM;
+			goto out;
+		}
+	}
+	priv->rx_ring_num = prof->rx_ring_num;
+	priv->cqe_factor = (mdev->dev->caps.cqe_size == 64) ? 1 : 0;
+	priv->cqe_size = mdev->dev->caps.cqe_size;
+	priv->mac_index = -1;
+	priv->msg_enable = MLX4_EN_MSG_LEVEL;
+#ifdef CONFIG_MLX4_EN_DCB
+	if (!mlx4_is_slave(priv->mdev->dev)) {
+		u8 prio;
+
+		for (prio = 0; prio < IEEE_8021QAZ_MAX_TCS; ++prio) {
+			priv->ets.prio_tc[prio] = prio;
+			priv->ets.tc_tsa[prio]  = IEEE_8021QAZ_TSA_VENDOR;
+		}
+
+		priv->dcbx_cap = DCB_CAP_DCBX_VER_CEE | DCB_CAP_DCBX_HOST |
+			DCB_CAP_DCBX_VER_IEEE;
+		priv->flags |= MLX4_EN_DCB_ENABLED;
+		priv->cee_config.pfc_state = false;
+
+		for (i = 0; i < MLX4_EN_NUM_UP_HIGH; i++)
+			priv->cee_config.dcb_pfc[i] = pfc_disabled;
+
+		if (mdev->dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_ETS_CFG) {
+			dev->dcbnl_ops = &mlx4_en_dcbnl_ops;
+		} else {
+			en_info(priv, "enabling only PFC DCB ops\n");
+			dev->dcbnl_ops = &mlx4_en_dcbnl_pfc_ops;
+		}
+	}
+#endif
+
+	for (i = 0; i < MLX4_EN_MAC_HASH_SIZE; ++i)
+		INIT_HLIST_HEAD(&priv->mac_hash[i]);
+
+	/* Query for default mac and max mtu */
+	priv->max_mtu = mdev->dev->caps.eth_mtu_cap[priv->port];
+
+	if (mdev->dev->caps.rx_checksum_flags_port[priv->port] &
+	    MLX4_RX_CSUM_MODE_VAL_NON_TCP_UDP)
+		priv->flags |= MLX4_EN_FLAG_RX_CSUM_NON_TCP_UDP;
+
+	/* Set default MAC */
+	dev->addr_len = ETH_ALEN;
+	mlx4_en_u64_to_mac(dev, mdev->dev->caps.def_mac[priv->port]);
+	if (!is_valid_ether_addr(dev->dev_addr)) {
+		en_err(priv, "Port: %d, invalid mac burned: %pM, quitting\n",
+		       priv->port, dev->dev_addr);
+		err = -EINVAL;
+		goto out;
+	} else if (mlx4_is_slave(priv->mdev->dev) &&
+		   (priv->mdev->dev->port_random_macs & 1 << priv->port)) {
+		/* Random MAC was assigned in mlx4_slave_cap
+		 * in mlx4_core module
+		 */
+		dev->addr_assign_type |= NET_ADDR_RANDOM;
+		en_warn(priv, "Assigned random MAC address %pM\n", dev->dev_addr);
+	}
+
+	memcpy(priv->current_mac, dev->dev_addr, sizeof(priv->current_mac));
+
+	priv->stride = roundup_pow_of_two(sizeof(struct mlx4_en_rx_desc) +
+					  DS_SIZE * MLX4_EN_MAX_RX_FRAGS);
+	err = mlx4_en_alloc_resources(priv);
+	if (err)
+		goto out;
+
+	/* Initialize time stamping config */
+	priv->hwtstamp_config.flags = 0;
+	priv->hwtstamp_config.tx_type = HWTSTAMP_TX_OFF;
+	priv->hwtstamp_config.rx_filter = HWTSTAMP_FILTER_NONE;
+
+	/* Allocate page for receive rings */
+	err = mlx4_alloc_hwq_res(mdev->dev, &priv->res,
+				MLX4_EN_PAGE_SIZE);
+	if (err) {
+		en_err(priv, "Failed to allocate page for rx qps\n");
+		goto out;
+	}
+	priv->allocated = 1;
+
+	/*
+	 * Initialize netdev entry points
+	 */
+	if (mlx4_is_master(priv->mdev->dev))
+		dev->netdev_ops = &mlx4_netdev_ops_master;
+	else
+		dev->netdev_ops = &mlx4_netdev_ops;
+	dev->watchdog_timeo = MLX4_EN_WATCHDOG_TIMEOUT;
+	netif_set_real_num_tx_queues(dev, priv->tx_ring_num[TX]);
+	netif_set_real_num_rx_queues(dev, priv->rx_ring_num);
+
+	dev->ethtool_ops = &mlx4_en_ethtool_ops;
+
+	/*
+	 * Set driver features
+	 */
+	dev->hw_features = NETIF_F_SG | NETIF_F_IP_CSUM | NETIF_F_IPV6_CSUM;
+	if (mdev->LSO_support)
+		dev->hw_features |= NETIF_F_TSO | NETIF_F_TSO6;
+
+	if (mdev->dev->caps.tunnel_offload_mode ==
+	    MLX4_TUNNEL_OFFLOAD_MODE_VXLAN) {
+		dev->hw_features |= NETIF_F_GSO_UDP_TUNNEL |
+				    NETIF_F_GSO_UDP_TUNNEL_CSUM |
+				    NETIF_F_GSO_PARTIAL;
+		dev->features    |= NETIF_F_GSO_UDP_TUNNEL |
+				    NETIF_F_GSO_UDP_TUNNEL_CSUM |
+				    NETIF_F_GSO_PARTIAL;
+		dev->gso_partial_features = NETIF_F_GSO_UDP_TUNNEL_CSUM;
+		dev->hw_enc_features = NETIF_F_IP_CSUM | NETIF_F_IPV6_CSUM |
+				       NETIF_F_RXCSUM |
+				       NETIF_F_TSO | NETIF_F_TSO6 |
+				       NETIF_F_GSO_UDP_TUNNEL |
+				       NETIF_F_GSO_UDP_TUNNEL_CSUM |
+				       NETIF_F_GSO_PARTIAL;
+
+		dev->udp_tunnel_nic_info = &mlx4_udp_tunnels;
+	}
+
+	dev->vlan_features = dev->hw_features;
+
+	dev->hw_features |= NETIF_F_RXCSUM | NETIF_F_RXHASH;
+	dev->features = dev->hw_features | NETIF_F_HIGHDMA |
+			NETIF_F_HW_VLAN_CTAG_TX | NETIF_F_HW_VLAN_CTAG_RX |
+			NETIF_F_HW_VLAN_CTAG_FILTER;
+	dev->hw_features |= NETIF_F_LOOPBACK |
+			NETIF_F_HW_VLAN_CTAG_TX | NETIF_F_HW_VLAN_CTAG_RX;
+
+	if (!(mdev->dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_SKIP_OUTER_VLAN)) {
+		dev->features |= NETIF_F_HW_VLAN_STAG_RX |
+			NETIF_F_HW_VLAN_STAG_FILTER;
+		dev->hw_features |= NETIF_F_HW_VLAN_STAG_RX;
+	}
+
+	if (mlx4_is_slave(mdev->dev)) {
+		bool vlan_offload_disabled;
+		int phv;
+
+		err = get_phv_bit(mdev->dev, port, &phv);
+		if (!err && phv) {
+			dev->hw_features |= NETIF_F_HW_VLAN_STAG_TX;
+			priv->pflags |= MLX4_EN_PRIV_FLAGS_PHV;
+		}
+		err = mlx4_get_is_vlan_offload_disabled(mdev->dev, port,
+							&vlan_offload_disabled);
+		if (!err && vlan_offload_disabled) {
+			dev->hw_features &= ~(NETIF_F_HW_VLAN_CTAG_TX |
+					      NETIF_F_HW_VLAN_CTAG_RX |
+					      NETIF_F_HW_VLAN_STAG_TX |
+					      NETIF_F_HW_VLAN_STAG_RX);
+			dev->features &= ~(NETIF_F_HW_VLAN_CTAG_TX |
+					   NETIF_F_HW_VLAN_CTAG_RX |
+					   NETIF_F_HW_VLAN_STAG_TX |
+					   NETIF_F_HW_VLAN_STAG_RX);
+		}
+	} else {
+		if (mdev->dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_PHV_EN &&
+		    !(mdev->dev->caps.flags2 &
+		      MLX4_DEV_CAP_FLAG2_SKIP_OUTER_VLAN))
+			dev->hw_features |= NETIF_F_HW_VLAN_STAG_TX;
+	}
+
+	if (mdev->dev->caps.flags & MLX4_DEV_CAP_FLAG_FCS_KEEP)
+		dev->hw_features |= NETIF_F_RXFCS;
+
+	if (mdev->dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_IGNORE_FCS)
+		dev->hw_features |= NETIF_F_RXALL;
+
+	if (mdev->dev->caps.steering_mode ==
+	    MLX4_STEERING_MODE_DEVICE_MANAGED &&
+	    mdev->dev->caps.dmfs_high_steer_mode != MLX4_STEERING_DMFS_A0_STATIC)
+		dev->hw_features |= NETIF_F_NTUPLE;
+
+	if (mdev->dev->caps.steering_mode != MLX4_STEERING_MODE_A0)
+		dev->priv_flags |= IFF_UNICAST_FLT;
+
+	/* Setting a default hash function value */
+	if (mdev->dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_RSS_TOP) {
+		priv->rss_hash_fn = ETH_RSS_HASH_TOP;
+	} else if (mdev->dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_RSS_XOR) {
+		priv->rss_hash_fn = ETH_RSS_HASH_XOR;
+	} else {
+		en_warn(priv,
+			"No RSS hash capabilities exposed, using Toeplitz\n");
+		priv->rss_hash_fn = ETH_RSS_HASH_TOP;
+	}
+
+	/* MTU range: 68 - hw-specific max */
+	dev->min_mtu = ETH_MIN_MTU;
+	dev->max_mtu = priv->max_mtu;
+
+	/* supports LSOv2 packets. */
+	netif_set_tso_max_size(dev, GSO_MAX_SIZE);
+
+	mdev->pndev[port] = dev;
+	mdev->upper[port] = NULL;
+
+	netif_carrier_off(dev);
+	mlx4_en_set_default_moderation(priv);
+
+	en_warn(priv, "Using %d TX rings\n", prof->tx_ring_num[TX]);
+	en_warn(priv, "Using %d RX rings\n", prof->rx_ring_num);
+
+	mlx4_en_update_loopback_state(priv->dev, priv->dev->features);
+
+	/* Configure port */
+	mlx4_en_calc_rx_buf(dev);
+	err = mlx4_SET_PORT_general(mdev->dev, priv->port,
+				    priv->rx_skb_size + ETH_FCS_LEN,
+				    prof->tx_pause, prof->tx_ppp,
+				    prof->rx_pause, prof->rx_ppp);
+	if (err) {
+		en_err(priv, "Failed setting port general configurations for port %d, with error %d\n",
+		       priv->port, err);
+		goto out;
+	}
+
+	if (mdev->dev->caps.tunnel_offload_mode == MLX4_TUNNEL_OFFLOAD_MODE_VXLAN) {
+		err = mlx4_SET_PORT_VXLAN(mdev->dev, priv->port, VXLAN_STEER_BY_OUTER_MAC, 1);
+		if (err) {
+			en_err(priv, "Failed setting port L2 tunnel configuration, err %d\n",
+			       err);
+			goto out;
+		}
+	}
+
+	/* Init port */
+	en_warn(priv, "Initializing port\n");
+	err = mlx4_INIT_PORT(mdev->dev, priv->port);
+	if (err) {
+		en_err(priv, "Failed Initializing port\n");
+		goto out;
+	}
+	queue_delayed_work(mdev->workqueue, &priv->stats_task, STATS_DELAY);
+
+	/* Initialize time stamp mechanism */
+	if (mdev->dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_TS)
+		mlx4_en_init_timestamp(mdev);
+
+	queue_delayed_work(mdev->workqueue, &priv->service_task,
+			   SERVICE_TASK_DELAY);
+
+	mlx4_en_set_stats_bitmap(mdev->dev, &priv->stats_bitmap,
+				 mdev->profile.prof[priv->port].rx_ppp,
+				 mdev->profile.prof[priv->port].rx_pause,
+				 mdev->profile.prof[priv->port].tx_ppp,
+				 mdev->profile.prof[priv->port].tx_pause);
+
+	err = register_netdev(dev);
+	if (err) {
+		en_err(priv, "Netdev registration failed for port %d\n", port);
+		goto out;
+	}
+
+	priv->registered = 1;
+	devlink_port_type_eth_set(mlx4_get_devlink_port(mdev->dev, priv->port),
+				  dev);
+
+	return 0;
+
+out:
+	mlx4_en_destroy_netdev(dev);
+	return err;
+}
+
+int mlx4_en_reset_config(struct net_device *dev,
+			 struct hwtstamp_config ts_config,
+			 netdev_features_t features)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	struct mlx4_en_dev *mdev = priv->mdev;
+	struct mlx4_en_port_profile new_prof;
+	struct mlx4_en_priv *tmp;
+	int port_up = 0;
+	int err = 0;
+
+	if (priv->hwtstamp_config.tx_type == ts_config.tx_type &&
+	    priv->hwtstamp_config.rx_filter == ts_config.rx_filter &&
+	    !DEV_FEATURE_CHANGED(dev, features, NETIF_F_HW_VLAN_CTAG_RX) &&
+	    !DEV_FEATURE_CHANGED(dev, features, NETIF_F_RXFCS))
+		return 0; /* Nothing to change */
+
+	if (DEV_FEATURE_CHANGED(dev, features, NETIF_F_HW_VLAN_CTAG_RX) &&
+	    (features & NETIF_F_HW_VLAN_CTAG_RX) &&
+	    (priv->hwtstamp_config.rx_filter != HWTSTAMP_FILTER_NONE)) {
+		en_warn(priv, "Can't turn ON rx vlan offload while time-stamping rx filter is ON\n");
+		return -EINVAL;
+	}
+
+	tmp = kzalloc(sizeof(*tmp), GFP_KERNEL);
+	if (!tmp)
+		return -ENOMEM;
+
+	mutex_lock(&mdev->state_lock);
+
+	memcpy(&new_prof, priv->prof, sizeof(struct mlx4_en_port_profile));
+	memcpy(&new_prof.hwtstamp_config, &ts_config, sizeof(ts_config));
+
+	err = mlx4_en_try_alloc_resources(priv, tmp, &new_prof, true);
+	if (err)
+		goto out;
+
+	if (priv->port_up) {
+		port_up = 1;
+		mlx4_en_stop_port(dev, 1);
+	}
+
+	mlx4_en_safe_replace_resources(priv, tmp);
+
+	if (DEV_FEATURE_CHANGED(dev, features, NETIF_F_HW_VLAN_CTAG_RX)) {
+		if (features & NETIF_F_HW_VLAN_CTAG_RX)
+			dev->features |= NETIF_F_HW_VLAN_CTAG_RX;
+		else
+			dev->features &= ~NETIF_F_HW_VLAN_CTAG_RX;
+	} else if (ts_config.rx_filter == HWTSTAMP_FILTER_NONE) {
+		/* RX time-stamping is OFF, update the RX vlan offload
+		 * to the latest wanted state
+		 */
+		if (dev->wanted_features & NETIF_F_HW_VLAN_CTAG_RX)
+			dev->features |= NETIF_F_HW_VLAN_CTAG_RX;
+		else
+			dev->features &= ~NETIF_F_HW_VLAN_CTAG_RX;
+	}
+
+	if (DEV_FEATURE_CHANGED(dev, features, NETIF_F_RXFCS)) {
+		if (features & NETIF_F_RXFCS)
+			dev->features |= NETIF_F_RXFCS;
+		else
+			dev->features &= ~NETIF_F_RXFCS;
+	}
+
+	/* RX vlan offload and RX time-stamping can't co-exist !
+	 * Regardless of the caller's choice,
+	 * Turn Off RX vlan offload in case of time-stamping is ON
+	 */
+	if (ts_config.rx_filter != HWTSTAMP_FILTER_NONE) {
+		if (dev->features & NETIF_F_HW_VLAN_CTAG_RX)
+			en_warn(priv, "Turning off RX vlan offload since RX time-stamping is ON\n");
+		dev->features &= ~NETIF_F_HW_VLAN_CTAG_RX;
+	}
+
+	if (port_up) {
+		err = mlx4_en_start_port(dev);
+		if (err)
+			en_err(priv, "Failed starting port\n");
+	}
+
+	if (!err)
+		err = mlx4_en_moderation_update(priv);
+out:
+	mutex_unlock(&mdev->state_lock);
+	kfree(tmp);
+	if (!err)
+		netdev_features_change(dev);
+	return err;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx4/en_port.c b/drivers/net/ethernet/mellanox/mlx4/en_port.c
new file mode 100644
index 000000000..532997eba
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx4/en_port.c
@@ -0,0 +1,436 @@
+/*
+ * Copyright (c) 2007 Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+
+
+#include <linux/if_vlan.h>
+
+#include <linux/mlx4/device.h>
+#include <linux/mlx4/cmd.h>
+
+#include "en_port.h"
+#include "mlx4_en.h"
+
+
+int mlx4_SET_VLAN_FLTR(struct mlx4_dev *dev, struct mlx4_en_priv *priv)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	struct mlx4_set_vlan_fltr_mbox *filter;
+	int i;
+	int j;
+	int index = 0;
+	u32 entry;
+	int err = 0;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+
+	filter = mailbox->buf;
+	for (i = VLAN_FLTR_SIZE - 1; i >= 0; i--) {
+		entry = 0;
+		for (j = 0; j < 32; j++)
+			if (test_bit(index++, priv->active_vlans))
+				entry |= 1 << j;
+		filter->entry[i] = cpu_to_be32(entry);
+	}
+	err = mlx4_cmd(dev, mailbox->dma, priv->port, 0, MLX4_CMD_SET_VLAN_FLTR,
+		       MLX4_CMD_TIME_CLASS_B, MLX4_CMD_WRAPPED);
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return err;
+}
+
+int mlx4_en_QUERY_PORT(struct mlx4_en_dev *mdev, u8 port)
+{
+	struct mlx4_en_query_port_context *qport_context;
+	struct mlx4_en_priv *priv = netdev_priv(mdev->pndev[port]);
+	struct mlx4_en_port_state *state = &priv->port_state;
+	struct mlx4_cmd_mailbox *mailbox;
+	int err;
+
+	mailbox = mlx4_alloc_cmd_mailbox(mdev->dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+	err = mlx4_cmd_box(mdev->dev, 0, mailbox->dma, port, 0,
+			   MLX4_CMD_QUERY_PORT, MLX4_CMD_TIME_CLASS_B,
+			   MLX4_CMD_WRAPPED);
+	if (err)
+		goto out;
+	qport_context = mailbox->buf;
+
+	/* This command is always accessed from Ethtool context
+	 * already synchronized, no need in locking */
+	state->link_state = !!(qport_context->link_up & MLX4_EN_LINK_UP_MASK);
+	switch (qport_context->link_speed & MLX4_EN_SPEED_MASK) {
+	case MLX4_EN_100M_SPEED:
+		state->link_speed = SPEED_100;
+		break;
+	case MLX4_EN_1G_SPEED:
+		state->link_speed = SPEED_1000;
+		break;
+	case MLX4_EN_10G_SPEED_XAUI:
+	case MLX4_EN_10G_SPEED_XFI:
+		state->link_speed = SPEED_10000;
+		break;
+	case MLX4_EN_20G_SPEED:
+		state->link_speed = SPEED_20000;
+		break;
+	case MLX4_EN_40G_SPEED:
+		state->link_speed = SPEED_40000;
+		break;
+	case MLX4_EN_56G_SPEED:
+		state->link_speed = SPEED_56000;
+		break;
+	default:
+		state->link_speed = -1;
+		break;
+	}
+
+	state->transceiver = qport_context->transceiver;
+
+	state->flags = 0; /* Reset and recalculate the port flags */
+	state->flags |= (qport_context->link_up & MLX4_EN_ANC_MASK) ?
+		MLX4_EN_PORT_ANC : 0;
+	state->flags |= (qport_context->autoneg & MLX4_EN_AUTONEG_MASK) ?
+		MLX4_EN_PORT_ANE : 0;
+
+out:
+	mlx4_free_cmd_mailbox(mdev->dev, mailbox);
+	return err;
+}
+
+/* Each counter set is located in struct mlx4_en_stat_out_mbox
+ * with a const offset between its prio components.
+ * This function runs over a counter set and sum all of it's prio components.
+ */
+static unsigned long en_stats_adder(__be64 *start, __be64 *next, int num)
+{
+	__be64 *curr = start;
+	unsigned long ret = 0;
+	int i;
+	int offset = next - start;
+
+	for (i = 0; i < num; i++) {
+		ret += be64_to_cpu(*curr);
+		curr += offset;
+	}
+
+	return ret;
+}
+
+void mlx4_en_fold_software_stats(struct net_device *dev)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	struct mlx4_en_dev *mdev = priv->mdev;
+	unsigned long packets, bytes;
+	int i;
+
+	if (!priv->port_up || mlx4_is_master(mdev->dev))
+		return;
+
+	packets = 0;
+	bytes = 0;
+	for (i = 0; i < priv->rx_ring_num; i++) {
+		const struct mlx4_en_rx_ring *ring = priv->rx_ring[i];
+
+		packets += READ_ONCE(ring->packets);
+		bytes   += READ_ONCE(ring->bytes);
+	}
+	dev->stats.rx_packets = packets;
+	dev->stats.rx_bytes = bytes;
+
+	packets = 0;
+	bytes = 0;
+	for (i = 0; i < priv->tx_ring_num[TX]; i++) {
+		const struct mlx4_en_tx_ring *ring = priv->tx_ring[TX][i];
+
+		packets += READ_ONCE(ring->packets);
+		bytes   += READ_ONCE(ring->bytes);
+	}
+	dev->stats.tx_packets = packets;
+	dev->stats.tx_bytes = bytes;
+}
+
+int mlx4_en_DUMP_ETH_STATS(struct mlx4_en_dev *mdev, u8 port, u8 reset)
+{
+	struct mlx4_counter tmp_counter_stats;
+	struct mlx4_en_stat_out_mbox *mlx4_en_stats;
+	struct mlx4_en_stat_out_flow_control_mbox *flowstats;
+	struct net_device *dev = mdev->pndev[port];
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	struct net_device_stats *stats = &dev->stats;
+	struct mlx4_cmd_mailbox *mailbox, *mailbox_priority;
+	u64 in_mod = reset << 8 | port;
+	int err;
+	int i, counter_index;
+	unsigned long sw_tx_dropped = 0;
+	unsigned long sw_rx_dropped = 0;
+
+	mailbox = mlx4_alloc_cmd_mailbox(mdev->dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+
+	mailbox_priority = mlx4_alloc_cmd_mailbox(mdev->dev);
+	if (IS_ERR(mailbox_priority)) {
+		mlx4_free_cmd_mailbox(mdev->dev, mailbox);
+		return PTR_ERR(mailbox_priority);
+	}
+
+	err = mlx4_cmd_box(mdev->dev, 0, mailbox->dma, in_mod, 0,
+			   MLX4_CMD_DUMP_ETH_STATS, MLX4_CMD_TIME_CLASS_B,
+			   MLX4_CMD_NATIVE);
+	if (err)
+		goto out;
+
+	mlx4_en_stats = mailbox->buf;
+
+	memset(&tmp_counter_stats, 0, sizeof(tmp_counter_stats));
+	counter_index = mlx4_get_default_counter_index(mdev->dev, port);
+	err = mlx4_get_counter_stats(mdev->dev, counter_index,
+				     &tmp_counter_stats, reset);
+
+	/* 0xffs indicates invalid value */
+	memset(mailbox_priority->buf, 0xff,
+	       sizeof(*flowstats) * MLX4_NUM_PRIORITIES);
+
+	if (mdev->dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_FLOWSTATS_EN) {
+		memset(mailbox_priority->buf, 0,
+		       sizeof(*flowstats) * MLX4_NUM_PRIORITIES);
+		err = mlx4_cmd_box(mdev->dev, 0, mailbox_priority->dma,
+				   in_mod | MLX4_DUMP_ETH_STATS_FLOW_CONTROL,
+				   0, MLX4_CMD_DUMP_ETH_STATS,
+				   MLX4_CMD_TIME_CLASS_B, MLX4_CMD_NATIVE);
+		if (err)
+			goto out;
+	}
+
+	flowstats = mailbox_priority->buf;
+
+	spin_lock_bh(&priv->stats_lock);
+
+	mlx4_en_fold_software_stats(dev);
+
+	priv->port_stats.rx_chksum_good = 0;
+	priv->port_stats.rx_chksum_none = 0;
+	priv->port_stats.rx_chksum_complete = 0;
+	priv->port_stats.rx_alloc_pages = 0;
+	priv->xdp_stats.rx_xdp_drop    = 0;
+	priv->xdp_stats.rx_xdp_redirect = 0;
+	priv->xdp_stats.rx_xdp_redirect_fail = 0;
+	priv->xdp_stats.rx_xdp_tx      = 0;
+	priv->xdp_stats.rx_xdp_tx_full = 0;
+	for (i = 0; i < priv->rx_ring_num; i++) {
+		const struct mlx4_en_rx_ring *ring = priv->rx_ring[i];
+
+		sw_rx_dropped			+= READ_ONCE(ring->dropped);
+		priv->port_stats.rx_chksum_good += READ_ONCE(ring->csum_ok);
+		priv->port_stats.rx_chksum_none += READ_ONCE(ring->csum_none);
+		priv->port_stats.rx_chksum_complete += READ_ONCE(ring->csum_complete);
+		priv->port_stats.rx_alloc_pages += READ_ONCE(ring->rx_alloc_pages);
+		priv->xdp_stats.rx_xdp_drop	+= READ_ONCE(ring->xdp_drop);
+		priv->xdp_stats.rx_xdp_redirect += READ_ONCE(ring->xdp_redirect);
+		priv->xdp_stats.rx_xdp_redirect_fail += READ_ONCE(ring->xdp_redirect_fail);
+		priv->xdp_stats.rx_xdp_tx	+= READ_ONCE(ring->xdp_tx);
+		priv->xdp_stats.rx_xdp_tx_full	+= READ_ONCE(ring->xdp_tx_full);
+	}
+	priv->port_stats.tx_chksum_offload = 0;
+	priv->port_stats.queue_stopped = 0;
+	priv->port_stats.wake_queue = 0;
+	priv->port_stats.tso_packets = 0;
+	priv->port_stats.xmit_more = 0;
+
+	for (i = 0; i < priv->tx_ring_num[TX]; i++) {
+		const struct mlx4_en_tx_ring *ring = priv->tx_ring[TX][i];
+
+		sw_tx_dropped			   += READ_ONCE(ring->tx_dropped);
+		priv->port_stats.tx_chksum_offload += READ_ONCE(ring->tx_csum);
+		priv->port_stats.queue_stopped     += READ_ONCE(ring->queue_stopped);
+		priv->port_stats.wake_queue        += READ_ONCE(ring->wake_queue);
+		priv->port_stats.tso_packets       += READ_ONCE(ring->tso_packets);
+		priv->port_stats.xmit_more         += READ_ONCE(ring->xmit_more);
+	}
+
+	if (!mlx4_is_slave(mdev->dev)) {
+		struct mlx4_en_phy_stats *p_stats = &priv->phy_stats;
+
+		p_stats->rx_packets_phy =
+			en_stats_adder(&mlx4_en_stats->RTOT_prio_0,
+				       &mlx4_en_stats->RTOT_prio_1,
+				       NUM_PRIORITIES);
+		p_stats->tx_packets_phy =
+			en_stats_adder(&mlx4_en_stats->TTOT_prio_0,
+				       &mlx4_en_stats->TTOT_prio_1,
+				       NUM_PRIORITIES);
+		p_stats->rx_bytes_phy =
+			en_stats_adder(&mlx4_en_stats->ROCT_prio_0,
+				       &mlx4_en_stats->ROCT_prio_1,
+				       NUM_PRIORITIES);
+		p_stats->tx_bytes_phy =
+			en_stats_adder(&mlx4_en_stats->TOCT_prio_0,
+				       &mlx4_en_stats->TOCT_prio_1,
+				       NUM_PRIORITIES);
+		if (mlx4_is_master(mdev->dev)) {
+			stats->rx_packets = p_stats->rx_packets_phy;
+			stats->tx_packets = p_stats->tx_packets_phy;
+			stats->rx_bytes = p_stats->rx_bytes_phy;
+			stats->tx_bytes = p_stats->tx_bytes_phy;
+		}
+	}
+
+	/* net device stats */
+	stats->rx_errors = be64_to_cpu(mlx4_en_stats->PCS) +
+			   be32_to_cpu(mlx4_en_stats->RJBBR) +
+			   be32_to_cpu(mlx4_en_stats->RCRC) +
+			   be32_to_cpu(mlx4_en_stats->RRUNT) +
+			   be64_to_cpu(mlx4_en_stats->RInRangeLengthErr) +
+			   be64_to_cpu(mlx4_en_stats->ROutRangeLengthErr) +
+			   be32_to_cpu(mlx4_en_stats->RSHORT) +
+			   en_stats_adder(&mlx4_en_stats->RGIANT_prio_0,
+					  &mlx4_en_stats->RGIANT_prio_1,
+					  NUM_PRIORITIES);
+	stats->tx_errors = en_stats_adder(&mlx4_en_stats->TGIANT_prio_0,
+					  &mlx4_en_stats->TGIANT_prio_1,
+					  NUM_PRIORITIES);
+	stats->multicast = en_stats_adder(&mlx4_en_stats->MCAST_prio_0,
+					  &mlx4_en_stats->MCAST_prio_1,
+					  NUM_PRIORITIES);
+	stats->rx_dropped = be32_to_cpu(mlx4_en_stats->RDROP) +
+			    sw_rx_dropped;
+	stats->rx_length_errors = be32_to_cpu(mlx4_en_stats->RdropLength);
+	stats->rx_crc_errors = be32_to_cpu(mlx4_en_stats->RCRC);
+	stats->rx_fifo_errors = be32_to_cpu(mlx4_en_stats->RdropOvflw);
+	stats->tx_dropped = be32_to_cpu(mlx4_en_stats->TDROP) +
+			    sw_tx_dropped;
+
+	/* RX stats */
+	priv->pkstats.rx_multicast_packets = stats->multicast;
+	priv->pkstats.rx_broadcast_packets =
+			en_stats_adder(&mlx4_en_stats->RBCAST_prio_0,
+				       &mlx4_en_stats->RBCAST_prio_1,
+				       NUM_PRIORITIES);
+	priv->pkstats.rx_jabbers = be32_to_cpu(mlx4_en_stats->RJBBR);
+	priv->pkstats.rx_in_range_length_error =
+		be64_to_cpu(mlx4_en_stats->RInRangeLengthErr);
+	priv->pkstats.rx_out_range_length_error =
+		be64_to_cpu(mlx4_en_stats->ROutRangeLengthErr);
+
+	/* Tx stats */
+	priv->pkstats.tx_multicast_packets =
+		en_stats_adder(&mlx4_en_stats->TMCAST_prio_0,
+			       &mlx4_en_stats->TMCAST_prio_1,
+			       NUM_PRIORITIES);
+	priv->pkstats.tx_broadcast_packets =
+		en_stats_adder(&mlx4_en_stats->TBCAST_prio_0,
+			       &mlx4_en_stats->TBCAST_prio_1,
+			       NUM_PRIORITIES);
+
+	priv->pkstats.rx_prio[0][0] = be64_to_cpu(mlx4_en_stats->RTOT_prio_0);
+	priv->pkstats.rx_prio[0][1] = be64_to_cpu(mlx4_en_stats->ROCT_prio_0);
+	priv->pkstats.rx_prio[1][0] = be64_to_cpu(mlx4_en_stats->RTOT_prio_1);
+	priv->pkstats.rx_prio[1][1] = be64_to_cpu(mlx4_en_stats->ROCT_prio_1);
+	priv->pkstats.rx_prio[2][0] = be64_to_cpu(mlx4_en_stats->RTOT_prio_2);
+	priv->pkstats.rx_prio[2][1] = be64_to_cpu(mlx4_en_stats->ROCT_prio_2);
+	priv->pkstats.rx_prio[3][0] = be64_to_cpu(mlx4_en_stats->RTOT_prio_3);
+	priv->pkstats.rx_prio[3][1] = be64_to_cpu(mlx4_en_stats->ROCT_prio_3);
+	priv->pkstats.rx_prio[4][0] = be64_to_cpu(mlx4_en_stats->RTOT_prio_4);
+	priv->pkstats.rx_prio[4][1] = be64_to_cpu(mlx4_en_stats->ROCT_prio_4);
+	priv->pkstats.rx_prio[5][0] = be64_to_cpu(mlx4_en_stats->RTOT_prio_5);
+	priv->pkstats.rx_prio[5][1] = be64_to_cpu(mlx4_en_stats->ROCT_prio_5);
+	priv->pkstats.rx_prio[6][0] = be64_to_cpu(mlx4_en_stats->RTOT_prio_6);
+	priv->pkstats.rx_prio[6][1] = be64_to_cpu(mlx4_en_stats->ROCT_prio_6);
+	priv->pkstats.rx_prio[7][0] = be64_to_cpu(mlx4_en_stats->RTOT_prio_7);
+	priv->pkstats.rx_prio[7][1] = be64_to_cpu(mlx4_en_stats->ROCT_prio_7);
+	priv->pkstats.rx_prio[8][0] = be64_to_cpu(mlx4_en_stats->RTOT_novlan);
+	priv->pkstats.rx_prio[8][1] = be64_to_cpu(mlx4_en_stats->ROCT_novlan);
+	priv->pkstats.tx_prio[0][0] = be64_to_cpu(mlx4_en_stats->TTOT_prio_0);
+	priv->pkstats.tx_prio[0][1] = be64_to_cpu(mlx4_en_stats->TOCT_prio_0);
+	priv->pkstats.tx_prio[1][0] = be64_to_cpu(mlx4_en_stats->TTOT_prio_1);
+	priv->pkstats.tx_prio[1][1] = be64_to_cpu(mlx4_en_stats->TOCT_prio_1);
+	priv->pkstats.tx_prio[2][0] = be64_to_cpu(mlx4_en_stats->TTOT_prio_2);
+	priv->pkstats.tx_prio[2][1] = be64_to_cpu(mlx4_en_stats->TOCT_prio_2);
+	priv->pkstats.tx_prio[3][0] = be64_to_cpu(mlx4_en_stats->TTOT_prio_3);
+	priv->pkstats.tx_prio[3][1] = be64_to_cpu(mlx4_en_stats->TOCT_prio_3);
+	priv->pkstats.tx_prio[4][0] = be64_to_cpu(mlx4_en_stats->TTOT_prio_4);
+	priv->pkstats.tx_prio[4][1] = be64_to_cpu(mlx4_en_stats->TOCT_prio_4);
+	priv->pkstats.tx_prio[5][0] = be64_to_cpu(mlx4_en_stats->TTOT_prio_5);
+	priv->pkstats.tx_prio[5][1] = be64_to_cpu(mlx4_en_stats->TOCT_prio_5);
+	priv->pkstats.tx_prio[6][0] = be64_to_cpu(mlx4_en_stats->TTOT_prio_6);
+	priv->pkstats.tx_prio[6][1] = be64_to_cpu(mlx4_en_stats->TOCT_prio_6);
+	priv->pkstats.tx_prio[7][0] = be64_to_cpu(mlx4_en_stats->TTOT_prio_7);
+	priv->pkstats.tx_prio[7][1] = be64_to_cpu(mlx4_en_stats->TOCT_prio_7);
+	priv->pkstats.tx_prio[8][0] = be64_to_cpu(mlx4_en_stats->TTOT_novlan);
+	priv->pkstats.tx_prio[8][1] = be64_to_cpu(mlx4_en_stats->TOCT_novlan);
+
+	if (tmp_counter_stats.counter_mode == 0) {
+		priv->pf_stats.rx_bytes   = be64_to_cpu(tmp_counter_stats.rx_bytes);
+		priv->pf_stats.tx_bytes   = be64_to_cpu(tmp_counter_stats.tx_bytes);
+		priv->pf_stats.rx_packets = be64_to_cpu(tmp_counter_stats.rx_frames);
+		priv->pf_stats.tx_packets = be64_to_cpu(tmp_counter_stats.tx_frames);
+	}
+
+	for (i = 0; i < MLX4_NUM_PRIORITIES; i++)	{
+		priv->rx_priority_flowstats[i].rx_pause =
+			be64_to_cpu(flowstats[i].rx_pause);
+		priv->rx_priority_flowstats[i].rx_pause_duration =
+			be64_to_cpu(flowstats[i].rx_pause_duration);
+		priv->rx_priority_flowstats[i].rx_pause_transition =
+			be64_to_cpu(flowstats[i].rx_pause_transition);
+		priv->tx_priority_flowstats[i].tx_pause =
+			be64_to_cpu(flowstats[i].tx_pause);
+		priv->tx_priority_flowstats[i].tx_pause_duration =
+			be64_to_cpu(flowstats[i].tx_pause_duration);
+		priv->tx_priority_flowstats[i].tx_pause_transition =
+			be64_to_cpu(flowstats[i].tx_pause_transition);
+	}
+
+	/* if pfc is not in use, all priorities counters have the same value */
+	priv->rx_flowstats.rx_pause =
+		be64_to_cpu(flowstats[0].rx_pause);
+	priv->rx_flowstats.rx_pause_duration =
+		be64_to_cpu(flowstats[0].rx_pause_duration);
+	priv->rx_flowstats.rx_pause_transition =
+		be64_to_cpu(flowstats[0].rx_pause_transition);
+	priv->tx_flowstats.tx_pause =
+		be64_to_cpu(flowstats[0].tx_pause);
+	priv->tx_flowstats.tx_pause_duration =
+		be64_to_cpu(flowstats[0].tx_pause_duration);
+	priv->tx_flowstats.tx_pause_transition =
+		be64_to_cpu(flowstats[0].tx_pause_transition);
+
+	spin_unlock_bh(&priv->stats_lock);
+
+out:
+	mlx4_free_cmd_mailbox(mdev->dev, mailbox);
+	mlx4_free_cmd_mailbox(mdev->dev, mailbox_priority);
+	return err;
+}
+
diff --git a/drivers/net/ethernet/mellanox/mlx4/en_port.h b/drivers/net/ethernet/mellanox/mlx4/en_port.h
new file mode 100644
index 000000000..930f961fe
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx4/en_port.h
@@ -0,0 +1,586 @@
+/*
+ * Copyright (c) 2007 Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+
+#ifndef _MLX4_EN_PORT_H_
+#define _MLX4_EN_PORT_H_
+
+
+#define SET_PORT_PROMISC_SHIFT	31
+#define SET_PORT_MC_PROMISC_SHIFT	30
+
+#define MLX4_EN_NUM_TC		8
+
+#define VLAN_FLTR_SIZE	128
+struct mlx4_set_vlan_fltr_mbox {
+	__be32 entry[VLAN_FLTR_SIZE];
+};
+
+
+enum {
+	MLX4_MCAST_CONFIG       = 0,
+	MLX4_MCAST_DISABLE      = 1,
+	MLX4_MCAST_ENABLE       = 2,
+};
+
+enum mlx4_link_mode {
+	MLX4_1000BASE_CX_SGMII	 = 0,
+	MLX4_1000BASE_KX	 = 1,
+	MLX4_10GBASE_CX4	 = 2,
+	MLX4_10GBASE_KX4	 = 3,
+	MLX4_10GBASE_KR		 = 4,
+	MLX4_20GBASE_KR2	 = 5,
+	MLX4_40GBASE_CR4	 = 6,
+	MLX4_40GBASE_KR4	 = 7,
+	MLX4_56GBASE_KR4	 = 8,
+	MLX4_10GBASE_CR		 = 12,
+	MLX4_10GBASE_SR		 = 13,
+	MLX4_40GBASE_SR4	 = 15,
+	MLX4_56GBASE_CR4	 = 17,
+	MLX4_56GBASE_SR4	 = 18,
+	MLX4_100BASE_TX		 = 24,
+	MLX4_1000BASE_T		 = 25,
+	MLX4_10GBASE_T		 = 26,
+};
+
+#define MLX4_PROT_MASK(link_mode) (1<<link_mode)
+
+enum {
+	MLX4_EN_100M_SPEED	= 0x04,
+	MLX4_EN_10G_SPEED_XAUI	= 0x00,
+	MLX4_EN_10G_SPEED_XFI	= 0x01,
+	MLX4_EN_1G_SPEED	= 0x02,
+	MLX4_EN_20G_SPEED	= 0x08,
+	MLX4_EN_40G_SPEED	= 0x40,
+	MLX4_EN_56G_SPEED	= 0x20,
+	MLX4_EN_OTHER_SPEED	= 0x0f,
+};
+
+struct mlx4_en_query_port_context {
+	u8 link_up;
+#define MLX4_EN_LINK_UP_MASK	0x80
+#define MLX4_EN_ANC_MASK	0x40
+	u8 autoneg;
+#define MLX4_EN_AUTONEG_MASK	0x80
+	__be16 mtu;
+	u8 reserved2;
+	u8 link_speed;
+#define MLX4_EN_SPEED_MASK	0x6f
+	u16 reserved3[5];
+	__be64 mac;
+	u8 transceiver;
+};
+
+
+struct mlx4_en_stat_out_mbox {
+	/* Received frames with a length of 64 octets */
+	__be64 R64_prio_0;
+	__be64 R64_prio_1;
+	__be64 R64_prio_2;
+	__be64 R64_prio_3;
+	__be64 R64_prio_4;
+	__be64 R64_prio_5;
+	__be64 R64_prio_6;
+	__be64 R64_prio_7;
+	__be64 R64_novlan;
+	/* Received frames with a length of 127 octets */
+	__be64 R127_prio_0;
+	__be64 R127_prio_1;
+	__be64 R127_prio_2;
+	__be64 R127_prio_3;
+	__be64 R127_prio_4;
+	__be64 R127_prio_5;
+	__be64 R127_prio_6;
+	__be64 R127_prio_7;
+	__be64 R127_novlan;
+	/* Received frames with a length of 255 octets */
+	__be64 R255_prio_0;
+	__be64 R255_prio_1;
+	__be64 R255_prio_2;
+	__be64 R255_prio_3;
+	__be64 R255_prio_4;
+	__be64 R255_prio_5;
+	__be64 R255_prio_6;
+	__be64 R255_prio_7;
+	__be64 R255_novlan;
+	/* Received frames with a length of 511 octets */
+	__be64 R511_prio_0;
+	__be64 R511_prio_1;
+	__be64 R511_prio_2;
+	__be64 R511_prio_3;
+	__be64 R511_prio_4;
+	__be64 R511_prio_5;
+	__be64 R511_prio_6;
+	__be64 R511_prio_7;
+	__be64 R511_novlan;
+	/* Received frames with a length of 1023 octets */
+	__be64 R1023_prio_0;
+	__be64 R1023_prio_1;
+	__be64 R1023_prio_2;
+	__be64 R1023_prio_3;
+	__be64 R1023_prio_4;
+	__be64 R1023_prio_5;
+	__be64 R1023_prio_6;
+	__be64 R1023_prio_7;
+	__be64 R1023_novlan;
+	/* Received frames with a length of 1518 octets */
+	__be64 R1518_prio_0;
+	__be64 R1518_prio_1;
+	__be64 R1518_prio_2;
+	__be64 R1518_prio_3;
+	__be64 R1518_prio_4;
+	__be64 R1518_prio_5;
+	__be64 R1518_prio_6;
+	__be64 R1518_prio_7;
+	__be64 R1518_novlan;
+	/* Received frames with a length of 1522 octets */
+	__be64 R1522_prio_0;
+	__be64 R1522_prio_1;
+	__be64 R1522_prio_2;
+	__be64 R1522_prio_3;
+	__be64 R1522_prio_4;
+	__be64 R1522_prio_5;
+	__be64 R1522_prio_6;
+	__be64 R1522_prio_7;
+	__be64 R1522_novlan;
+	/* Received frames with a length of 1548 octets */
+	__be64 R1548_prio_0;
+	__be64 R1548_prio_1;
+	__be64 R1548_prio_2;
+	__be64 R1548_prio_3;
+	__be64 R1548_prio_4;
+	__be64 R1548_prio_5;
+	__be64 R1548_prio_6;
+	__be64 R1548_prio_7;
+	__be64 R1548_novlan;
+	/* Received frames with a length of 1548 < octets < MTU */
+	__be64 R2MTU_prio_0;
+	__be64 R2MTU_prio_1;
+	__be64 R2MTU_prio_2;
+	__be64 R2MTU_prio_3;
+	__be64 R2MTU_prio_4;
+	__be64 R2MTU_prio_5;
+	__be64 R2MTU_prio_6;
+	__be64 R2MTU_prio_7;
+	__be64 R2MTU_novlan;
+	/* Received frames with a length of MTU< octets and good CRC */
+	__be64 RGIANT_prio_0;
+	__be64 RGIANT_prio_1;
+	__be64 RGIANT_prio_2;
+	__be64 RGIANT_prio_3;
+	__be64 RGIANT_prio_4;
+	__be64 RGIANT_prio_5;
+	__be64 RGIANT_prio_6;
+	__be64 RGIANT_prio_7;
+	__be64 RGIANT_novlan;
+	/* Received broadcast frames with good CRC */
+	__be64 RBCAST_prio_0;
+	__be64 RBCAST_prio_1;
+	__be64 RBCAST_prio_2;
+	__be64 RBCAST_prio_3;
+	__be64 RBCAST_prio_4;
+	__be64 RBCAST_prio_5;
+	__be64 RBCAST_prio_6;
+	__be64 RBCAST_prio_7;
+	__be64 RBCAST_novlan;
+	/* Received multicast frames with good CRC */
+	__be64 MCAST_prio_0;
+	__be64 MCAST_prio_1;
+	__be64 MCAST_prio_2;
+	__be64 MCAST_prio_3;
+	__be64 MCAST_prio_4;
+	__be64 MCAST_prio_5;
+	__be64 MCAST_prio_6;
+	__be64 MCAST_prio_7;
+	__be64 MCAST_novlan;
+	/* Received unicast not short or GIANT frames with good CRC */
+	__be64 RTOTG_prio_0;
+	__be64 RTOTG_prio_1;
+	__be64 RTOTG_prio_2;
+	__be64 RTOTG_prio_3;
+	__be64 RTOTG_prio_4;
+	__be64 RTOTG_prio_5;
+	__be64 RTOTG_prio_6;
+	__be64 RTOTG_prio_7;
+	__be64 RTOTG_novlan;
+
+	/* Count of total octets of received frames, includes framing characters */
+	__be64 RTTLOCT_prio_0;
+	/* Count of total octets of received frames, not including framing
+	   characters */
+	__be64 RTTLOCT_NOFRM_prio_0;
+	/* Count of Total number of octets received
+	   (only for frames without errors) */
+	__be64 ROCT_prio_0;
+
+	__be64 RTTLOCT_prio_1;
+	__be64 RTTLOCT_NOFRM_prio_1;
+	__be64 ROCT_prio_1;
+
+	__be64 RTTLOCT_prio_2;
+	__be64 RTTLOCT_NOFRM_prio_2;
+	__be64 ROCT_prio_2;
+
+	__be64 RTTLOCT_prio_3;
+	__be64 RTTLOCT_NOFRM_prio_3;
+	__be64 ROCT_prio_3;
+
+	__be64 RTTLOCT_prio_4;
+	__be64 RTTLOCT_NOFRM_prio_4;
+	__be64 ROCT_prio_4;
+
+	__be64 RTTLOCT_prio_5;
+	__be64 RTTLOCT_NOFRM_prio_5;
+	__be64 ROCT_prio_5;
+
+	__be64 RTTLOCT_prio_6;
+	__be64 RTTLOCT_NOFRM_prio_6;
+	__be64 ROCT_prio_6;
+
+	__be64 RTTLOCT_prio_7;
+	__be64 RTTLOCT_NOFRM_prio_7;
+	__be64 ROCT_prio_7;
+
+	__be64 RTTLOCT_novlan;
+	__be64 RTTLOCT_NOFRM_novlan;
+	__be64 ROCT_novlan;
+
+	/* Count of Total received frames including bad frames */
+	__be64 RTOT_prio_0;
+	/* Count of  Total number of received frames with 802.1Q encapsulation */
+	__be64 R1Q_prio_0;
+	__be64 reserved1;
+
+	__be64 RTOT_prio_1;
+	__be64 R1Q_prio_1;
+	__be64 reserved2;
+
+	__be64 RTOT_prio_2;
+	__be64 R1Q_prio_2;
+	__be64 reserved3;
+
+	__be64 RTOT_prio_3;
+	__be64 R1Q_prio_3;
+	__be64 reserved4;
+
+	__be64 RTOT_prio_4;
+	__be64 R1Q_prio_4;
+	__be64 reserved5;
+
+	__be64 RTOT_prio_5;
+	__be64 R1Q_prio_5;
+	__be64 reserved6;
+
+	__be64 RTOT_prio_6;
+	__be64 R1Q_prio_6;
+	__be64 reserved7;
+
+	__be64 RTOT_prio_7;
+	__be64 R1Q_prio_7;
+	__be64 reserved8;
+
+	__be64 RTOT_novlan;
+	__be64 R1Q_novlan;
+	__be64 reserved9;
+
+	/* Total number of Successfully Received Control Frames */
+	__be64 RCNTL;
+	__be64 reserved10;
+	__be64 reserved11;
+	__be64 reserved12;
+	/* Count of received frames with a length/type field  value between 46
+	   (42 for VLANtagged frames) and 1500 (also 1500 for VLAN-tagged frames),
+	   inclusive */
+	__be64 RInRangeLengthErr;
+	/* Count of received frames with length/type field between 1501 and 1535
+	   decimal, inclusive */
+	__be64 ROutRangeLengthErr;
+	/* Count of received frames that are longer than max allowed size for
+	   802.3 frames (1518/1522) */
+	__be64 RFrmTooLong;
+	/* Count frames received with PCS error */
+	__be64 PCS;
+
+	/* Transmit frames with a length of 64 octets */
+	__be64 T64_prio_0;
+	__be64 T64_prio_1;
+	__be64 T64_prio_2;
+	__be64 T64_prio_3;
+	__be64 T64_prio_4;
+	__be64 T64_prio_5;
+	__be64 T64_prio_6;
+	__be64 T64_prio_7;
+	__be64 T64_novlan;
+	__be64 T64_loopbk;
+	/* Transmit frames with a length of 65 to 127 octets. */
+	__be64 T127_prio_0;
+	__be64 T127_prio_1;
+	__be64 T127_prio_2;
+	__be64 T127_prio_3;
+	__be64 T127_prio_4;
+	__be64 T127_prio_5;
+	__be64 T127_prio_6;
+	__be64 T127_prio_7;
+	__be64 T127_novlan;
+	__be64 T127_loopbk;
+	/* Transmit frames with a length of 128 to 255 octets */
+	__be64 T255_prio_0;
+	__be64 T255_prio_1;
+	__be64 T255_prio_2;
+	__be64 T255_prio_3;
+	__be64 T255_prio_4;
+	__be64 T255_prio_5;
+	__be64 T255_prio_6;
+	__be64 T255_prio_7;
+	__be64 T255_novlan;
+	__be64 T255_loopbk;
+	/* Transmit frames with a length of 256 to 511 octets */
+	__be64 T511_prio_0;
+	__be64 T511_prio_1;
+	__be64 T511_prio_2;
+	__be64 T511_prio_3;
+	__be64 T511_prio_4;
+	__be64 T511_prio_5;
+	__be64 T511_prio_6;
+	__be64 T511_prio_7;
+	__be64 T511_novlan;
+	__be64 T511_loopbk;
+	/* Transmit frames with a length of 512 to 1023 octets */
+	__be64 T1023_prio_0;
+	__be64 T1023_prio_1;
+	__be64 T1023_prio_2;
+	__be64 T1023_prio_3;
+	__be64 T1023_prio_4;
+	__be64 T1023_prio_5;
+	__be64 T1023_prio_6;
+	__be64 T1023_prio_7;
+	__be64 T1023_novlan;
+	__be64 T1023_loopbk;
+	/* Transmit frames with a length of 1024 to 1518 octets */
+	__be64 T1518_prio_0;
+	__be64 T1518_prio_1;
+	__be64 T1518_prio_2;
+	__be64 T1518_prio_3;
+	__be64 T1518_prio_4;
+	__be64 T1518_prio_5;
+	__be64 T1518_prio_6;
+	__be64 T1518_prio_7;
+	__be64 T1518_novlan;
+	__be64 T1518_loopbk;
+	/* Counts transmit frames with a length of 1519 to 1522 bytes */
+	__be64 T1522_prio_0;
+	__be64 T1522_prio_1;
+	__be64 T1522_prio_2;
+	__be64 T1522_prio_3;
+	__be64 T1522_prio_4;
+	__be64 T1522_prio_5;
+	__be64 T1522_prio_6;
+	__be64 T1522_prio_7;
+	__be64 T1522_novlan;
+	__be64 T1522_loopbk;
+	/* Transmit frames with a length of 1523 to 1548 octets */
+	__be64 T1548_prio_0;
+	__be64 T1548_prio_1;
+	__be64 T1548_prio_2;
+	__be64 T1548_prio_3;
+	__be64 T1548_prio_4;
+	__be64 T1548_prio_5;
+	__be64 T1548_prio_6;
+	__be64 T1548_prio_7;
+	__be64 T1548_novlan;
+	__be64 T1548_loopbk;
+	/* Counts transmit frames with a length of 1549 to MTU bytes */
+	__be64 T2MTU_prio_0;
+	__be64 T2MTU_prio_1;
+	__be64 T2MTU_prio_2;
+	__be64 T2MTU_prio_3;
+	__be64 T2MTU_prio_4;
+	__be64 T2MTU_prio_5;
+	__be64 T2MTU_prio_6;
+	__be64 T2MTU_prio_7;
+	__be64 T2MTU_novlan;
+	__be64 T2MTU_loopbk;
+	/* Transmit frames with a length greater than MTU octets and a good CRC. */
+	__be64 TGIANT_prio_0;
+	__be64 TGIANT_prio_1;
+	__be64 TGIANT_prio_2;
+	__be64 TGIANT_prio_3;
+	__be64 TGIANT_prio_4;
+	__be64 TGIANT_prio_5;
+	__be64 TGIANT_prio_6;
+	__be64 TGIANT_prio_7;
+	__be64 TGIANT_novlan;
+	__be64 TGIANT_loopbk;
+	/* Transmit broadcast frames with a good CRC */
+	__be64 TBCAST_prio_0;
+	__be64 TBCAST_prio_1;
+	__be64 TBCAST_prio_2;
+	__be64 TBCAST_prio_3;
+	__be64 TBCAST_prio_4;
+	__be64 TBCAST_prio_5;
+	__be64 TBCAST_prio_6;
+	__be64 TBCAST_prio_7;
+	__be64 TBCAST_novlan;
+	__be64 TBCAST_loopbk;
+	/* Transmit multicast frames with a good CRC */
+	__be64 TMCAST_prio_0;
+	__be64 TMCAST_prio_1;
+	__be64 TMCAST_prio_2;
+	__be64 TMCAST_prio_3;
+	__be64 TMCAST_prio_4;
+	__be64 TMCAST_prio_5;
+	__be64 TMCAST_prio_6;
+	__be64 TMCAST_prio_7;
+	__be64 TMCAST_novlan;
+	__be64 TMCAST_loopbk;
+	/* Transmit good frames that are neither broadcast nor multicast */
+	__be64 TTOTG_prio_0;
+	__be64 TTOTG_prio_1;
+	__be64 TTOTG_prio_2;
+	__be64 TTOTG_prio_3;
+	__be64 TTOTG_prio_4;
+	__be64 TTOTG_prio_5;
+	__be64 TTOTG_prio_6;
+	__be64 TTOTG_prio_7;
+	__be64 TTOTG_novlan;
+	__be64 TTOTG_loopbk;
+
+	/* total octets of transmitted frames, including framing characters */
+	__be64 TTTLOCT_prio_0;
+	/* total octets of transmitted frames, not including framing characters */
+	__be64 TTTLOCT_NOFRM_prio_0;
+	/* ifOutOctets */
+	__be64 TOCT_prio_0;
+
+	__be64 TTTLOCT_prio_1;
+	__be64 TTTLOCT_NOFRM_prio_1;
+	__be64 TOCT_prio_1;
+
+	__be64 TTTLOCT_prio_2;
+	__be64 TTTLOCT_NOFRM_prio_2;
+	__be64 TOCT_prio_2;
+
+	__be64 TTTLOCT_prio_3;
+	__be64 TTTLOCT_NOFRM_prio_3;
+	__be64 TOCT_prio_3;
+
+	__be64 TTTLOCT_prio_4;
+	__be64 TTTLOCT_NOFRM_prio_4;
+	__be64 TOCT_prio_4;
+
+	__be64 TTTLOCT_prio_5;
+	__be64 TTTLOCT_NOFRM_prio_5;
+	__be64 TOCT_prio_5;
+
+	__be64 TTTLOCT_prio_6;
+	__be64 TTTLOCT_NOFRM_prio_6;
+	__be64 TOCT_prio_6;
+
+	__be64 TTTLOCT_prio_7;
+	__be64 TTTLOCT_NOFRM_prio_7;
+	__be64 TOCT_prio_7;
+
+	__be64 TTTLOCT_novlan;
+	__be64 TTTLOCT_NOFRM_novlan;
+	__be64 TOCT_novlan;
+
+	__be64 TTTLOCT_loopbk;
+	__be64 TTTLOCT_NOFRM_loopbk;
+	__be64 TOCT_loopbk;
+
+	/* Total frames transmitted with a good CRC that are not aborted  */
+	__be64 TTOT_prio_0;
+	/* Total number of frames transmitted with 802.1Q encapsulation */
+	__be64 T1Q_prio_0;
+	__be64 reserved13;
+
+	__be64 TTOT_prio_1;
+	__be64 T1Q_prio_1;
+	__be64 reserved14;
+
+	__be64 TTOT_prio_2;
+	__be64 T1Q_prio_2;
+	__be64 reserved15;
+
+	__be64 TTOT_prio_3;
+	__be64 T1Q_prio_3;
+	__be64 reserved16;
+
+	__be64 TTOT_prio_4;
+	__be64 T1Q_prio_4;
+	__be64 reserved17;
+
+	__be64 TTOT_prio_5;
+	__be64 T1Q_prio_5;
+	__be64 reserved18;
+
+	__be64 TTOT_prio_6;
+	__be64 T1Q_prio_6;
+	__be64 reserved19;
+
+	__be64 TTOT_prio_7;
+	__be64 T1Q_prio_7;
+	__be64 reserved20;
+
+	__be64 TTOT_novlan;
+	__be64 T1Q_novlan;
+	__be64 reserved21;
+
+	__be64 TTOT_loopbk;
+	__be64 T1Q_loopbk;
+	__be64 reserved22;
+
+	/* Received frames with a length greater than MTU octets and a bad CRC */
+	__be32 RJBBR;
+	/* Received frames with a bad CRC that are not runts, jabbers,
+	   or alignment errors */
+	__be32 RCRC;
+	/* Received frames with SFD with a length of less than 64 octets and a
+	   bad CRC */
+	__be32 RRUNT;
+	/* Received frames with a length less than 64 octets and a good CRC */
+	__be32 RSHORT;
+	/* Total Number of Received Packets Dropped */
+	__be32 RDROP;
+	/* Drop due to overflow  */
+	__be32 RdropOvflw;
+	/* Drop due to overflow */
+	__be32 RdropLength;
+	/* Total of good frames. Does not include frames received with
+	   frame-too-long, FCS, or length errors */
+	__be32 RTOTFRMS;
+	/* Total dropped Xmited packets */
+	__be32 TDROP;
+};
+
+
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlx4/en_resources.c b/drivers/net/ethernet/mellanox/mlx4/en_resources.c
new file mode 100644
index 000000000..6883ac75d
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx4/en_resources.c
@@ -0,0 +1,115 @@
+/*
+ * Copyright (c) 2007 Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+
+#include <linux/slab.h>
+#include <linux/vmalloc.h>
+#include <linux/mlx4/qp.h>
+
+#include "mlx4_en.h"
+
+void mlx4_en_fill_qp_context(struct mlx4_en_priv *priv, int size, int stride,
+			     int is_tx, int rss, int qpn, int cqn,
+			     int user_prio, struct mlx4_qp_context *context)
+{
+	struct mlx4_en_dev *mdev = priv->mdev;
+	struct net_device *dev = priv->dev;
+
+	memset(context, 0, sizeof(*context));
+	context->flags = cpu_to_be32(7 << 16 | rss << MLX4_RSS_QPC_FLAG_OFFSET);
+	context->pd = cpu_to_be32(mdev->priv_pdn);
+	context->mtu_msgmax = 0xff;
+	if (!is_tx && !rss)
+		context->rq_size_stride = ilog2(size) << 3 | (ilog2(stride) - 4);
+	if (is_tx) {
+		context->sq_size_stride = ilog2(size) << 3 | (ilog2(stride) - 4);
+		if (mdev->dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_PORT_REMAP)
+			context->params2 |= cpu_to_be32(MLX4_QP_BIT_FPP);
+
+	} else {
+		context->sq_size_stride = ilog2(TXBB_SIZE) - 4;
+	}
+	context->usr_page = cpu_to_be32(mlx4_to_hw_uar_index(mdev->dev,
+					mdev->priv_uar.index));
+	context->local_qpn = cpu_to_be32(qpn);
+	context->pri_path.ackto = 1 & 0x07;
+	context->pri_path.sched_queue = 0x83 | (priv->port - 1) << 6;
+	/* force user priority per tx ring */
+	if (user_prio >= 0 && priv->prof->num_up == MLX4_EN_NUM_UP_HIGH) {
+		context->pri_path.sched_queue |= user_prio << 3;
+		context->pri_path.feup = MLX4_FEUP_FORCE_ETH_UP;
+	}
+	context->pri_path.counter_index = priv->counter_index;
+	context->cqn_send = cpu_to_be32(cqn);
+	context->cqn_recv = cpu_to_be32(cqn);
+	if (!rss &&
+	    (mdev->dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_LB_SRC_CHK) &&
+	    context->pri_path.counter_index !=
+			    MLX4_SINK_COUNTER_INDEX(mdev->dev)) {
+		/* disable multicast loopback to qp with same counter */
+		if (!(dev->features & NETIF_F_LOOPBACK))
+			context->pri_path.fl |= MLX4_FL_ETH_SRC_CHECK_MC_LB;
+		context->pri_path.control |= MLX4_CTRL_ETH_SRC_CHECK_IF_COUNTER;
+	}
+	context->db_rec_addr = cpu_to_be64(priv->res.db.dma << 2);
+	if (!(dev->features & NETIF_F_HW_VLAN_CTAG_RX))
+		context->param3 |= cpu_to_be32(1 << 30);
+
+	if (!is_tx && !rss &&
+	    (mdev->dev->caps.tunnel_offload_mode ==  MLX4_TUNNEL_OFFLOAD_MODE_VXLAN)) {
+		en_dbg(HW, priv, "Setting RX qp %x tunnel mode to RX tunneled & non-tunneled\n", qpn);
+		context->srqn = cpu_to_be32(7 << 28); /* this fills bits 30:28 */
+	}
+}
+
+int mlx4_en_change_mcast_lb(struct mlx4_en_priv *priv, struct mlx4_qp *qp,
+			    int loopback)
+{
+	int ret;
+	struct mlx4_update_qp_params qp_params;
+
+	memset(&qp_params, 0, sizeof(qp_params));
+	if (!loopback)
+		qp_params.flags = MLX4_UPDATE_QP_PARAMS_FLAGS_ETH_CHECK_MC_LB;
+
+	ret = mlx4_update_qp(priv->mdev->dev, qp->qpn,
+			     MLX4_UPDATE_QP_ETH_SRC_CHECK_MC_LB,
+			     &qp_params);
+
+	return ret;
+}
+
+void mlx4_en_sqp_event(struct mlx4_qp *qp, enum mlx4_event event)
+{
+    return;
+}
+
diff --git a/drivers/net/ethernet/mellanox/mlx4/en_rx.c b/drivers/net/ethernet/mellanox/mlx4/en_rx.c
new file mode 100644
index 000000000..8f762fc17
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx4/en_rx.c
@@ -0,0 +1,1285 @@
+/*
+ * Copyright (c) 2007 Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+
+#include <linux/bpf.h>
+#include <linux/bpf_trace.h>
+#include <linux/mlx4/cq.h>
+#include <linux/slab.h>
+#include <linux/mlx4/qp.h>
+#include <linux/skbuff.h>
+#include <linux/rculist.h>
+#include <linux/if_ether.h>
+#include <linux/if_vlan.h>
+#include <linux/vmalloc.h>
+#include <linux/irq.h>
+
+#include <net/ip.h>
+#if IS_ENABLED(CONFIG_IPV6)
+#include <net/ip6_checksum.h>
+#endif
+
+#include "mlx4_en.h"
+
+static int mlx4_alloc_page(struct mlx4_en_priv *priv,
+			   struct mlx4_en_rx_alloc *frag,
+			   gfp_t gfp)
+{
+	struct page *page;
+	dma_addr_t dma;
+
+	page = alloc_page(gfp);
+	if (unlikely(!page))
+		return -ENOMEM;
+	dma = dma_map_page(priv->ddev, page, 0, PAGE_SIZE, priv->dma_dir);
+	if (unlikely(dma_mapping_error(priv->ddev, dma))) {
+		__free_page(page);
+		return -ENOMEM;
+	}
+	frag->page = page;
+	frag->dma = dma;
+	frag->page_offset = priv->rx_headroom;
+	return 0;
+}
+
+static int mlx4_en_alloc_frags(struct mlx4_en_priv *priv,
+			       struct mlx4_en_rx_ring *ring,
+			       struct mlx4_en_rx_desc *rx_desc,
+			       struct mlx4_en_rx_alloc *frags,
+			       gfp_t gfp)
+{
+	int i;
+
+	for (i = 0; i < priv->num_frags; i++, frags++) {
+		if (!frags->page) {
+			if (mlx4_alloc_page(priv, frags, gfp))
+				return -ENOMEM;
+			ring->rx_alloc_pages++;
+		}
+		rx_desc->data[i].addr = cpu_to_be64(frags->dma +
+						    frags->page_offset);
+	}
+	return 0;
+}
+
+static void mlx4_en_free_frag(const struct mlx4_en_priv *priv,
+			      struct mlx4_en_rx_alloc *frag)
+{
+	if (frag->page) {
+		dma_unmap_page(priv->ddev, frag->dma,
+			       PAGE_SIZE, priv->dma_dir);
+		__free_page(frag->page);
+	}
+	/* We need to clear all fields, otherwise a change of priv->log_rx_info
+	 * could lead to see garbage later in frag->page.
+	 */
+	memset(frag, 0, sizeof(*frag));
+}
+
+static void mlx4_en_init_rx_desc(const struct mlx4_en_priv *priv,
+				 struct mlx4_en_rx_ring *ring, int index)
+{
+	struct mlx4_en_rx_desc *rx_desc = ring->buf + ring->stride * index;
+	int possible_frags;
+	int i;
+
+	/* Set size and memtype fields */
+	for (i = 0; i < priv->num_frags; i++) {
+		rx_desc->data[i].byte_count =
+			cpu_to_be32(priv->frag_info[i].frag_size);
+		rx_desc->data[i].lkey = cpu_to_be32(priv->mdev->mr.key);
+	}
+
+	/* If the number of used fragments does not fill up the ring stride,
+	 * remaining (unused) fragments must be padded with null address/size
+	 * and a special memory key */
+	possible_frags = (ring->stride - sizeof(struct mlx4_en_rx_desc)) / DS_SIZE;
+	for (i = priv->num_frags; i < possible_frags; i++) {
+		rx_desc->data[i].byte_count = 0;
+		rx_desc->data[i].lkey = cpu_to_be32(MLX4_EN_MEMTYPE_PAD);
+		rx_desc->data[i].addr = 0;
+	}
+}
+
+static int mlx4_en_prepare_rx_desc(struct mlx4_en_priv *priv,
+				   struct mlx4_en_rx_ring *ring, int index,
+				   gfp_t gfp)
+{
+	struct mlx4_en_rx_desc *rx_desc = ring->buf +
+		(index << ring->log_stride);
+	struct mlx4_en_rx_alloc *frags = ring->rx_info +
+					(index << priv->log_rx_info);
+	if (likely(ring->page_cache.index > 0)) {
+		/* XDP uses a single page per frame */
+		if (!frags->page) {
+			ring->page_cache.index--;
+			frags->page = ring->page_cache.buf[ring->page_cache.index].page;
+			frags->dma  = ring->page_cache.buf[ring->page_cache.index].dma;
+		}
+		frags->page_offset = XDP_PACKET_HEADROOM;
+		rx_desc->data[0].addr = cpu_to_be64(frags->dma +
+						    XDP_PACKET_HEADROOM);
+		return 0;
+	}
+
+	return mlx4_en_alloc_frags(priv, ring, rx_desc, frags, gfp);
+}
+
+static bool mlx4_en_is_ring_empty(const struct mlx4_en_rx_ring *ring)
+{
+	return ring->prod == ring->cons;
+}
+
+static inline void mlx4_en_update_rx_prod_db(struct mlx4_en_rx_ring *ring)
+{
+	*ring->wqres.db.db = cpu_to_be32(ring->prod & 0xffff);
+}
+
+/* slow path */
+static void mlx4_en_free_rx_desc(const struct mlx4_en_priv *priv,
+				 struct mlx4_en_rx_ring *ring,
+				 int index)
+{
+	struct mlx4_en_rx_alloc *frags;
+	int nr;
+
+	frags = ring->rx_info + (index << priv->log_rx_info);
+	for (nr = 0; nr < priv->num_frags; nr++) {
+		en_dbg(DRV, priv, "Freeing fragment:%d\n", nr);
+		mlx4_en_free_frag(priv, frags + nr);
+	}
+}
+
+/* Function not in fast-path */
+static int mlx4_en_fill_rx_buffers(struct mlx4_en_priv *priv)
+{
+	struct mlx4_en_rx_ring *ring;
+	int ring_ind;
+	int buf_ind;
+	int new_size;
+
+	for (buf_ind = 0; buf_ind < priv->prof->rx_ring_size; buf_ind++) {
+		for (ring_ind = 0; ring_ind < priv->rx_ring_num; ring_ind++) {
+			ring = priv->rx_ring[ring_ind];
+
+			if (mlx4_en_prepare_rx_desc(priv, ring,
+						    ring->actual_size,
+						    GFP_KERNEL)) {
+				if (ring->actual_size < MLX4_EN_MIN_RX_SIZE) {
+					en_err(priv, "Failed to allocate enough rx buffers\n");
+					return -ENOMEM;
+				} else {
+					new_size = rounddown_pow_of_two(ring->actual_size);
+					en_warn(priv, "Only %d buffers allocated reducing ring size to %d\n",
+						ring->actual_size, new_size);
+					goto reduce_rings;
+				}
+			}
+			ring->actual_size++;
+			ring->prod++;
+		}
+	}
+	return 0;
+
+reduce_rings:
+	for (ring_ind = 0; ring_ind < priv->rx_ring_num; ring_ind++) {
+		ring = priv->rx_ring[ring_ind];
+		while (ring->actual_size > new_size) {
+			ring->actual_size--;
+			ring->prod--;
+			mlx4_en_free_rx_desc(priv, ring, ring->actual_size);
+		}
+	}
+
+	return 0;
+}
+
+static void mlx4_en_free_rx_buf(struct mlx4_en_priv *priv,
+				struct mlx4_en_rx_ring *ring)
+{
+	int index;
+
+	en_dbg(DRV, priv, "Freeing Rx buf - cons:%d prod:%d\n",
+	       ring->cons, ring->prod);
+
+	/* Unmap and free Rx buffers */
+	for (index = 0; index < ring->size; index++) {
+		en_dbg(DRV, priv, "Processing descriptor:%d\n", index);
+		mlx4_en_free_rx_desc(priv, ring, index);
+	}
+	ring->cons = 0;
+	ring->prod = 0;
+}
+
+void mlx4_en_set_num_rx_rings(struct mlx4_en_dev *mdev)
+{
+	int i;
+	int num_of_eqs;
+	int num_rx_rings;
+	struct mlx4_dev *dev = mdev->dev;
+
+	mlx4_foreach_port(i, dev, MLX4_PORT_TYPE_ETH) {
+		num_of_eqs = max_t(int, MIN_RX_RINGS,
+				   min_t(int,
+					 mlx4_get_eqs_per_port(mdev->dev, i),
+					 DEF_RX_RINGS));
+
+		num_rx_rings = mlx4_low_memory_profile() ? MIN_RX_RINGS :
+			min_t(int, num_of_eqs, num_online_cpus());
+		mdev->profile.prof[i].rx_ring_num =
+			rounddown_pow_of_two(num_rx_rings);
+	}
+}
+
+int mlx4_en_create_rx_ring(struct mlx4_en_priv *priv,
+			   struct mlx4_en_rx_ring **pring,
+			   u32 size, u16 stride, int node, int queue_index)
+{
+	struct mlx4_en_dev *mdev = priv->mdev;
+	struct mlx4_en_rx_ring *ring;
+	int err = -ENOMEM;
+	int tmp;
+
+	ring = kzalloc_node(sizeof(*ring), GFP_KERNEL, node);
+	if (!ring) {
+		en_err(priv, "Failed to allocate RX ring structure\n");
+		return -ENOMEM;
+	}
+
+	ring->prod = 0;
+	ring->cons = 0;
+	ring->size = size;
+	ring->size_mask = size - 1;
+	ring->stride = stride;
+	ring->log_stride = ffs(ring->stride) - 1;
+	ring->buf_size = ring->size * ring->stride + TXBB_SIZE;
+
+	if (xdp_rxq_info_reg(&ring->xdp_rxq, priv->dev, queue_index, 0) < 0)
+		goto err_ring;
+
+	tmp = size * roundup_pow_of_two(MLX4_EN_MAX_RX_FRAGS *
+					sizeof(struct mlx4_en_rx_alloc));
+	ring->rx_info = kvzalloc_node(tmp, GFP_KERNEL, node);
+	if (!ring->rx_info) {
+		err = -ENOMEM;
+		goto err_xdp_info;
+	}
+
+	en_dbg(DRV, priv, "Allocated rx_info ring at addr:%p size:%d\n",
+		 ring->rx_info, tmp);
+
+	/* Allocate HW buffers on provided NUMA node */
+	set_dev_node(&mdev->dev->persist->pdev->dev, node);
+	err = mlx4_alloc_hwq_res(mdev->dev, &ring->wqres, ring->buf_size);
+	set_dev_node(&mdev->dev->persist->pdev->dev, mdev->dev->numa_node);
+	if (err)
+		goto err_info;
+
+	ring->buf = ring->wqres.buf.direct.buf;
+
+	ring->hwtstamp_rx_filter = priv->hwtstamp_config.rx_filter;
+
+	*pring = ring;
+	return 0;
+
+err_info:
+	kvfree(ring->rx_info);
+	ring->rx_info = NULL;
+err_xdp_info:
+	xdp_rxq_info_unreg(&ring->xdp_rxq);
+err_ring:
+	kfree(ring);
+	*pring = NULL;
+
+	return err;
+}
+
+int mlx4_en_activate_rx_rings(struct mlx4_en_priv *priv)
+{
+	struct mlx4_en_rx_ring *ring;
+	int i;
+	int ring_ind;
+	int err;
+	int stride = roundup_pow_of_two(sizeof(struct mlx4_en_rx_desc) +
+					DS_SIZE * priv->num_frags);
+
+	for (ring_ind = 0; ring_ind < priv->rx_ring_num; ring_ind++) {
+		ring = priv->rx_ring[ring_ind];
+
+		ring->prod = 0;
+		ring->cons = 0;
+		ring->actual_size = 0;
+		ring->cqn = priv->rx_cq[ring_ind]->mcq.cqn;
+
+		ring->stride = stride;
+		if (ring->stride <= TXBB_SIZE) {
+			/* Stamp first unused send wqe */
+			__be32 *ptr = (__be32 *)ring->buf;
+			__be32 stamp = cpu_to_be32(1 << STAMP_SHIFT);
+			*ptr = stamp;
+			/* Move pointer to start of rx section */
+			ring->buf += TXBB_SIZE;
+		}
+
+		ring->log_stride = ffs(ring->stride) - 1;
+		ring->buf_size = ring->size * ring->stride;
+
+		memset(ring->buf, 0, ring->buf_size);
+		mlx4_en_update_rx_prod_db(ring);
+
+		/* Initialize all descriptors */
+		for (i = 0; i < ring->size; i++)
+			mlx4_en_init_rx_desc(priv, ring, i);
+	}
+	err = mlx4_en_fill_rx_buffers(priv);
+	if (err)
+		goto err_buffers;
+
+	for (ring_ind = 0; ring_ind < priv->rx_ring_num; ring_ind++) {
+		ring = priv->rx_ring[ring_ind];
+
+		ring->size_mask = ring->actual_size - 1;
+		mlx4_en_update_rx_prod_db(ring);
+	}
+
+	return 0;
+
+err_buffers:
+	for (ring_ind = 0; ring_ind < priv->rx_ring_num; ring_ind++)
+		mlx4_en_free_rx_buf(priv, priv->rx_ring[ring_ind]);
+
+	ring_ind = priv->rx_ring_num - 1;
+	while (ring_ind >= 0) {
+		if (priv->rx_ring[ring_ind]->stride <= TXBB_SIZE)
+			priv->rx_ring[ring_ind]->buf -= TXBB_SIZE;
+		ring_ind--;
+	}
+	return err;
+}
+
+/* We recover from out of memory by scheduling our napi poll
+ * function (mlx4_en_process_cq), which tries to allocate
+ * all missing RX buffers (call to mlx4_en_refill_rx_buffers).
+ */
+void mlx4_en_recover_from_oom(struct mlx4_en_priv *priv)
+{
+	int ring;
+
+	if (!priv->port_up)
+		return;
+
+	for (ring = 0; ring < priv->rx_ring_num; ring++) {
+		if (mlx4_en_is_ring_empty(priv->rx_ring[ring])) {
+			local_bh_disable();
+			napi_reschedule(&priv->rx_cq[ring]->napi);
+			local_bh_enable();
+		}
+	}
+}
+
+/* When the rx ring is running in page-per-packet mode, a released frame can go
+ * directly into a small cache, to avoid unmapping or touching the page
+ * allocator. In bpf prog performance scenarios, buffers are either forwarded
+ * or dropped, never converted to skbs, so every page can come directly from
+ * this cache when it is sized to be a multiple of the napi budget.
+ */
+bool mlx4_en_rx_recycle(struct mlx4_en_rx_ring *ring,
+			struct mlx4_en_rx_alloc *frame)
+{
+	struct mlx4_en_page_cache *cache = &ring->page_cache;
+
+	if (cache->index >= MLX4_EN_CACHE_SIZE)
+		return false;
+
+	cache->buf[cache->index].page = frame->page;
+	cache->buf[cache->index].dma = frame->dma;
+	cache->index++;
+	return true;
+}
+
+void mlx4_en_destroy_rx_ring(struct mlx4_en_priv *priv,
+			     struct mlx4_en_rx_ring **pring,
+			     u32 size, u16 stride)
+{
+	struct mlx4_en_dev *mdev = priv->mdev;
+	struct mlx4_en_rx_ring *ring = *pring;
+	struct bpf_prog *old_prog;
+
+	old_prog = rcu_dereference_protected(
+					ring->xdp_prog,
+					lockdep_is_held(&mdev->state_lock));
+	if (old_prog)
+		bpf_prog_put(old_prog);
+	xdp_rxq_info_unreg(&ring->xdp_rxq);
+	mlx4_free_hwq_res(mdev->dev, &ring->wqres, size * stride + TXBB_SIZE);
+	kvfree(ring->rx_info);
+	ring->rx_info = NULL;
+	kfree(ring);
+	*pring = NULL;
+}
+
+void mlx4_en_deactivate_rx_ring(struct mlx4_en_priv *priv,
+				struct mlx4_en_rx_ring *ring)
+{
+	int i;
+
+	for (i = 0; i < ring->page_cache.index; i++) {
+		dma_unmap_page(priv->ddev, ring->page_cache.buf[i].dma,
+			       PAGE_SIZE, priv->dma_dir);
+		put_page(ring->page_cache.buf[i].page);
+	}
+	ring->page_cache.index = 0;
+	mlx4_en_free_rx_buf(priv, ring);
+	if (ring->stride <= TXBB_SIZE)
+		ring->buf -= TXBB_SIZE;
+}
+
+
+static int mlx4_en_complete_rx_desc(struct mlx4_en_priv *priv,
+				    struct mlx4_en_rx_alloc *frags,
+				    struct sk_buff *skb,
+				    int length)
+{
+	const struct mlx4_en_frag_info *frag_info = priv->frag_info;
+	unsigned int truesize = 0;
+	bool release = true;
+	int nr, frag_size;
+	struct page *page;
+	dma_addr_t dma;
+
+	/* Collect used fragments while replacing them in the HW descriptors */
+	for (nr = 0;; frags++) {
+		frag_size = min_t(int, length, frag_info->frag_size);
+
+		page = frags->page;
+		if (unlikely(!page))
+			goto fail;
+
+		dma = frags->dma;
+		dma_sync_single_range_for_cpu(priv->ddev, dma, frags->page_offset,
+					      frag_size, priv->dma_dir);
+
+		__skb_fill_page_desc(skb, nr, page, frags->page_offset,
+				     frag_size);
+
+		truesize += frag_info->frag_stride;
+		if (frag_info->frag_stride == PAGE_SIZE / 2) {
+			frags->page_offset ^= PAGE_SIZE / 2;
+			release = page_count(page) != 1 ||
+				  page_is_pfmemalloc(page) ||
+				  page_to_nid(page) != numa_mem_id();
+		} else if (!priv->rx_headroom) {
+			/* rx_headroom for non XDP setup is always 0.
+			 * When XDP is set, the above condition will
+			 * guarantee page is always released.
+			 */
+			u32 sz_align = ALIGN(frag_size, SMP_CACHE_BYTES);
+
+			frags->page_offset += sz_align;
+			release = frags->page_offset + frag_info->frag_size > PAGE_SIZE;
+		}
+		if (release) {
+			dma_unmap_page(priv->ddev, dma, PAGE_SIZE, priv->dma_dir);
+			frags->page = NULL;
+		} else {
+			page_ref_inc(page);
+		}
+
+		nr++;
+		length -= frag_size;
+		if (!length)
+			break;
+		frag_info++;
+	}
+	skb->truesize += truesize;
+	return nr;
+
+fail:
+	while (nr > 0) {
+		nr--;
+		__skb_frag_unref(skb_shinfo(skb)->frags + nr, false);
+	}
+	return 0;
+}
+
+static void validate_loopback(struct mlx4_en_priv *priv, void *va)
+{
+	const unsigned char *data = va + ETH_HLEN;
+	int i;
+
+	for (i = 0; i < MLX4_LOOPBACK_TEST_PAYLOAD; i++) {
+		if (data[i] != (unsigned char)i)
+			return;
+	}
+	/* Loopback found */
+	priv->loopback_ok = 1;
+}
+
+static void mlx4_en_refill_rx_buffers(struct mlx4_en_priv *priv,
+				      struct mlx4_en_rx_ring *ring)
+{
+	u32 missing = ring->actual_size - (ring->prod - ring->cons);
+
+	/* Try to batch allocations, but not too much. */
+	if (missing < 8)
+		return;
+	do {
+		if (mlx4_en_prepare_rx_desc(priv, ring,
+					    ring->prod & ring->size_mask,
+					    GFP_ATOMIC | __GFP_MEMALLOC))
+			break;
+		ring->prod++;
+	} while (likely(--missing));
+
+	mlx4_en_update_rx_prod_db(ring);
+}
+
+/* When hardware doesn't strip the vlan, we need to calculate the checksum
+ * over it and add it to the hardware's checksum calculation
+ */
+static inline __wsum get_fixed_vlan_csum(__wsum hw_checksum,
+					 struct vlan_hdr *vlanh)
+{
+	return csum_add(hw_checksum, *(__wsum *)vlanh);
+}
+
+/* Although the stack expects checksum which doesn't include the pseudo
+ * header, the HW adds it. To address that, we are subtracting the pseudo
+ * header checksum from the checksum value provided by the HW.
+ */
+static int get_fixed_ipv4_csum(__wsum hw_checksum, struct sk_buff *skb,
+			       struct iphdr *iph)
+{
+	__u16 length_for_csum = 0;
+	__wsum csum_pseudo_header = 0;
+	__u8 ipproto = iph->protocol;
+
+	if (unlikely(ipproto == IPPROTO_SCTP))
+		return -1;
+
+	length_for_csum = (be16_to_cpu(iph->tot_len) - (iph->ihl << 2));
+	csum_pseudo_header = csum_tcpudp_nofold(iph->saddr, iph->daddr,
+						length_for_csum, ipproto, 0);
+	skb->csum = csum_sub(hw_checksum, csum_pseudo_header);
+	return 0;
+}
+
+#if IS_ENABLED(CONFIG_IPV6)
+/* In IPv6 packets, hw_checksum lacks 6 bytes from IPv6 header:
+ * 4 first bytes : priority, version, flow_lbl
+ * and 2 additional bytes : nexthdr, hop_limit.
+ */
+static int get_fixed_ipv6_csum(__wsum hw_checksum, struct sk_buff *skb,
+			       struct ipv6hdr *ipv6h)
+{
+	__u8 nexthdr = ipv6h->nexthdr;
+	__wsum temp;
+
+	if (unlikely(nexthdr == IPPROTO_FRAGMENT ||
+		     nexthdr == IPPROTO_HOPOPTS ||
+		     nexthdr == IPPROTO_SCTP))
+		return -1;
+
+	/* priority, version, flow_lbl */
+	temp = csum_add(hw_checksum, *(__wsum *)ipv6h);
+	/* nexthdr and hop_limit */
+	skb->csum = csum_add(temp, (__force __wsum)*(__be16 *)&ipv6h->nexthdr);
+	return 0;
+}
+#endif
+
+#define short_frame(size) ((size) <= ETH_ZLEN + ETH_FCS_LEN)
+
+/* We reach this function only after checking that any of
+ * the (IPv4 | IPv6) bits are set in cqe->status.
+ */
+static int check_csum(struct mlx4_cqe *cqe, struct sk_buff *skb, void *va,
+		      netdev_features_t dev_features)
+{
+	__wsum hw_checksum = 0;
+	void *hdr;
+
+	/* CQE csum doesn't cover padding octets in short ethernet
+	 * frames. And the pad field is appended prior to calculating
+	 * and appending the FCS field.
+	 *
+	 * Detecting these padded frames requires to verify and parse
+	 * IP headers, so we simply force all those small frames to skip
+	 * checksum complete.
+	 */
+	if (short_frame(skb->len))
+		return -EINVAL;
+
+	hdr = (u8 *)va + sizeof(struct ethhdr);
+	hw_checksum = csum_unfold((__force __sum16)cqe->checksum);
+
+	if (cqe->vlan_my_qpn & cpu_to_be32(MLX4_CQE_CVLAN_PRESENT_MASK) &&
+	    !(dev_features & NETIF_F_HW_VLAN_CTAG_RX)) {
+		hw_checksum = get_fixed_vlan_csum(hw_checksum, hdr);
+		hdr += sizeof(struct vlan_hdr);
+	}
+
+#if IS_ENABLED(CONFIG_IPV6)
+	if (cqe->status & cpu_to_be16(MLX4_CQE_STATUS_IPV6))
+		return get_fixed_ipv6_csum(hw_checksum, skb, hdr);
+#endif
+	return get_fixed_ipv4_csum(hw_checksum, skb, hdr);
+}
+
+#if IS_ENABLED(CONFIG_IPV6)
+#define MLX4_CQE_STATUS_IP_ANY (MLX4_CQE_STATUS_IPV4 | MLX4_CQE_STATUS_IPV6)
+#else
+#define MLX4_CQE_STATUS_IP_ANY (MLX4_CQE_STATUS_IPV4)
+#endif
+
+int mlx4_en_process_rx_cq(struct net_device *dev, struct mlx4_en_cq *cq, int budget)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	int factor = priv->cqe_factor;
+	struct mlx4_en_rx_ring *ring;
+	struct bpf_prog *xdp_prog;
+	int cq_ring = cq->ring;
+	bool doorbell_pending;
+	bool xdp_redir_flush;
+	struct mlx4_cqe *cqe;
+	struct xdp_buff xdp;
+	int polled = 0;
+	int index;
+
+	if (unlikely(!priv->port_up || budget <= 0))
+		return 0;
+
+	ring = priv->rx_ring[cq_ring];
+
+	xdp_prog = rcu_dereference_bh(ring->xdp_prog);
+	xdp_init_buff(&xdp, priv->frag_info[0].frag_stride, &ring->xdp_rxq);
+	doorbell_pending = false;
+	xdp_redir_flush = false;
+
+	/* We assume a 1:1 mapping between CQEs and Rx descriptors, so Rx
+	 * descriptor offset can be deduced from the CQE index instead of
+	 * reading 'cqe->index' */
+	index = cq->mcq.cons_index & ring->size_mask;
+	cqe = mlx4_en_get_cqe(cq->buf, index, priv->cqe_size) + factor;
+
+	/* Process all completed CQEs */
+	while (XNOR(cqe->owner_sr_opcode & MLX4_CQE_OWNER_MASK,
+		    cq->mcq.cons_index & cq->size)) {
+		struct mlx4_en_rx_alloc *frags;
+		enum pkt_hash_types hash_type;
+		struct sk_buff *skb;
+		unsigned int length;
+		int ip_summed;
+		void *va;
+		int nr;
+
+		frags = ring->rx_info + (index << priv->log_rx_info);
+		va = page_address(frags[0].page) + frags[0].page_offset;
+		net_prefetchw(va);
+		/*
+		 * make sure we read the CQE after we read the ownership bit
+		 */
+		dma_rmb();
+
+		/* Drop packet on bad receive or bad checksum */
+		if (unlikely((cqe->owner_sr_opcode & MLX4_CQE_OPCODE_MASK) ==
+						MLX4_CQE_OPCODE_ERROR)) {
+			en_err(priv, "CQE completed in error - vendor syndrom:%d syndrom:%d\n",
+			       ((struct mlx4_err_cqe *)cqe)->vendor_err_syndrome,
+			       ((struct mlx4_err_cqe *)cqe)->syndrome);
+			goto next;
+		}
+		if (unlikely(cqe->badfcs_enc & MLX4_CQE_BAD_FCS)) {
+			en_dbg(RX_ERR, priv, "Accepted frame with bad FCS\n");
+			goto next;
+		}
+
+		/* Check if we need to drop the packet if SRIOV is not enabled
+		 * and not performing the selftest or flb disabled
+		 */
+		if (priv->flags & MLX4_EN_FLAG_RX_FILTER_NEEDED) {
+			const struct ethhdr *ethh = va;
+			dma_addr_t dma;
+			/* Get pointer to first fragment since we haven't
+			 * skb yet and cast it to ethhdr struct
+			 */
+			dma = frags[0].dma + frags[0].page_offset;
+			dma_sync_single_for_cpu(priv->ddev, dma, sizeof(*ethh),
+						DMA_FROM_DEVICE);
+
+			if (is_multicast_ether_addr(ethh->h_dest)) {
+				struct mlx4_mac_entry *entry;
+				struct hlist_head *bucket;
+				unsigned int mac_hash;
+
+				/* Drop the packet, since HW loopback-ed it */
+				mac_hash = ethh->h_source[MLX4_EN_MAC_HASH_IDX];
+				bucket = &priv->mac_hash[mac_hash];
+				hlist_for_each_entry_rcu_bh(entry, bucket, hlist) {
+					if (ether_addr_equal_64bits(entry->mac,
+								    ethh->h_source))
+						goto next;
+				}
+			}
+		}
+
+		if (unlikely(priv->validate_loopback)) {
+			validate_loopback(priv, va);
+			goto next;
+		}
+
+		/*
+		 * Packet is OK - process it.
+		 */
+		length = be32_to_cpu(cqe->byte_cnt);
+		length -= ring->fcs_del;
+
+		/* A bpf program gets first chance to drop the packet. It may
+		 * read bytes but not past the end of the frag.
+		 */
+		if (xdp_prog) {
+			dma_addr_t dma;
+			void *orig_data;
+			u32 act;
+
+			dma = frags[0].dma + frags[0].page_offset;
+			dma_sync_single_for_cpu(priv->ddev, dma,
+						priv->frag_info[0].frag_size,
+						DMA_FROM_DEVICE);
+
+			xdp_prepare_buff(&xdp, va - frags[0].page_offset,
+					 frags[0].page_offset, length, false);
+			orig_data = xdp.data;
+
+			act = bpf_prog_run_xdp(xdp_prog, &xdp);
+
+			length = xdp.data_end - xdp.data;
+			if (xdp.data != orig_data) {
+				frags[0].page_offset = xdp.data -
+					xdp.data_hard_start;
+				va = xdp.data;
+			}
+
+			switch (act) {
+			case XDP_PASS:
+				break;
+			case XDP_REDIRECT:
+				if (likely(!xdp_do_redirect(dev, &xdp, xdp_prog))) {
+					ring->xdp_redirect++;
+					xdp_redir_flush = true;
+					frags[0].page = NULL;
+					goto next;
+				}
+				ring->xdp_redirect_fail++;
+				trace_xdp_exception(dev, xdp_prog, act);
+				goto xdp_drop_no_cnt;
+			case XDP_TX:
+				if (likely(!mlx4_en_xmit_frame(ring, frags, priv,
+							length, cq_ring,
+							&doorbell_pending))) {
+					frags[0].page = NULL;
+					goto next;
+				}
+				trace_xdp_exception(dev, xdp_prog, act);
+				goto xdp_drop_no_cnt; /* Drop on xmit failure */
+			default:
+				bpf_warn_invalid_xdp_action(dev, xdp_prog, act);
+				fallthrough;
+			case XDP_ABORTED:
+				trace_xdp_exception(dev, xdp_prog, act);
+				fallthrough;
+			case XDP_DROP:
+				ring->xdp_drop++;
+xdp_drop_no_cnt:
+				goto next;
+			}
+		}
+
+		ring->bytes += length;
+		ring->packets++;
+
+		skb = napi_get_frags(&cq->napi);
+		if (unlikely(!skb))
+			goto next;
+
+		if (unlikely(ring->hwtstamp_rx_filter == HWTSTAMP_FILTER_ALL)) {
+			u64 timestamp = mlx4_en_get_cqe_ts(cqe);
+
+			mlx4_en_fill_hwtstamps(priv->mdev, skb_hwtstamps(skb),
+					       timestamp);
+		}
+		skb_record_rx_queue(skb, cq_ring);
+
+		if (likely(dev->features & NETIF_F_RXCSUM)) {
+			/* TODO: For IP non TCP/UDP packets when csum complete is
+			 * not an option (not supported or any other reason) we can
+			 * actually check cqe IPOK status bit and report
+			 * CHECKSUM_UNNECESSARY rather than CHECKSUM_NONE
+			 */
+			if ((cqe->status & cpu_to_be16(MLX4_CQE_STATUS_TCP |
+						       MLX4_CQE_STATUS_UDP)) &&
+			    (cqe->status & cpu_to_be16(MLX4_CQE_STATUS_IPOK)) &&
+			    cqe->checksum == cpu_to_be16(0xffff)) {
+				bool l2_tunnel;
+
+				l2_tunnel = (dev->hw_enc_features & NETIF_F_RXCSUM) &&
+					(cqe->vlan_my_qpn & cpu_to_be32(MLX4_CQE_L2_TUNNEL));
+				ip_summed = CHECKSUM_UNNECESSARY;
+				hash_type = PKT_HASH_TYPE_L4;
+				if (l2_tunnel)
+					skb->csum_level = 1;
+				ring->csum_ok++;
+			} else {
+				if (!(priv->flags & MLX4_EN_FLAG_RX_CSUM_NON_TCP_UDP &&
+				      (cqe->status & cpu_to_be16(MLX4_CQE_STATUS_IP_ANY))))
+					goto csum_none;
+				if (check_csum(cqe, skb, va, dev->features))
+					goto csum_none;
+				ip_summed = CHECKSUM_COMPLETE;
+				hash_type = PKT_HASH_TYPE_L3;
+				ring->csum_complete++;
+			}
+		} else {
+csum_none:
+			ip_summed = CHECKSUM_NONE;
+			hash_type = PKT_HASH_TYPE_L3;
+			ring->csum_none++;
+		}
+		skb->ip_summed = ip_summed;
+		if (dev->features & NETIF_F_RXHASH)
+			skb_set_hash(skb,
+				     be32_to_cpu(cqe->immed_rss_invalid),
+				     hash_type);
+
+		if ((cqe->vlan_my_qpn &
+		     cpu_to_be32(MLX4_CQE_CVLAN_PRESENT_MASK)) &&
+		    (dev->features & NETIF_F_HW_VLAN_CTAG_RX))
+			__vlan_hwaccel_put_tag(skb, htons(ETH_P_8021Q),
+					       be16_to_cpu(cqe->sl_vid));
+		else if ((cqe->vlan_my_qpn &
+			  cpu_to_be32(MLX4_CQE_SVLAN_PRESENT_MASK)) &&
+			 (dev->features & NETIF_F_HW_VLAN_STAG_RX))
+			__vlan_hwaccel_put_tag(skb, htons(ETH_P_8021AD),
+					       be16_to_cpu(cqe->sl_vid));
+
+		nr = mlx4_en_complete_rx_desc(priv, frags, skb, length);
+		if (likely(nr)) {
+			skb_shinfo(skb)->nr_frags = nr;
+			skb->len = length;
+			skb->data_len = length;
+			napi_gro_frags(&cq->napi);
+		} else {
+			__vlan_hwaccel_clear_tag(skb);
+			skb_clear_hash(skb);
+		}
+next:
+		++cq->mcq.cons_index;
+		index = (cq->mcq.cons_index) & ring->size_mask;
+		cqe = mlx4_en_get_cqe(cq->buf, index, priv->cqe_size) + factor;
+		if (unlikely(++polled == budget))
+			break;
+	}
+
+	if (xdp_redir_flush)
+		xdp_do_flush();
+
+	if (likely(polled)) {
+		if (doorbell_pending) {
+			priv->tx_cq[TX_XDP][cq_ring]->xdp_busy = true;
+			mlx4_en_xmit_doorbell(priv->tx_ring[TX_XDP][cq_ring]);
+		}
+
+		mlx4_cq_set_ci(&cq->mcq);
+		wmb(); /* ensure HW sees CQ consumer before we post new buffers */
+		ring->cons = cq->mcq.cons_index;
+	}
+
+	mlx4_en_refill_rx_buffers(priv, ring);
+
+	return polled;
+}
+
+
+void mlx4_en_rx_irq(struct mlx4_cq *mcq)
+{
+	struct mlx4_en_cq *cq = container_of(mcq, struct mlx4_en_cq, mcq);
+	struct mlx4_en_priv *priv = netdev_priv(cq->dev);
+
+	if (likely(priv->port_up))
+		napi_schedule_irqoff(&cq->napi);
+	else
+		mlx4_en_arm_cq(priv, cq);
+}
+
+/* Rx CQ polling - called by NAPI */
+int mlx4_en_poll_rx_cq(struct napi_struct *napi, int budget)
+{
+	struct mlx4_en_cq *cq = container_of(napi, struct mlx4_en_cq, napi);
+	struct net_device *dev = cq->dev;
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	struct mlx4_en_cq *xdp_tx_cq = NULL;
+	bool clean_complete = true;
+	int done;
+
+	if (!budget)
+		return 0;
+
+	if (priv->tx_ring_num[TX_XDP]) {
+		xdp_tx_cq = priv->tx_cq[TX_XDP][cq->ring];
+		if (xdp_tx_cq->xdp_busy) {
+			clean_complete = mlx4_en_process_tx_cq(dev, xdp_tx_cq,
+							       budget) < budget;
+			xdp_tx_cq->xdp_busy = !clean_complete;
+		}
+	}
+
+	done = mlx4_en_process_rx_cq(dev, cq, budget);
+
+	/* If we used up all the quota - we're probably not done yet... */
+	if (done == budget || !clean_complete) {
+		int cpu_curr;
+
+		/* in case we got here because of !clean_complete */
+		done = budget;
+
+		cpu_curr = smp_processor_id();
+
+		if (likely(cpumask_test_cpu(cpu_curr, cq->aff_mask)))
+			return budget;
+
+		/* Current cpu is not according to smp_irq_affinity -
+		 * probably affinity changed. Need to stop this NAPI
+		 * poll, and restart it on the right CPU.
+		 * Try to avoid returning a too small value (like 0),
+		 * to not fool net_rx_action() and its netdev_budget
+		 */
+		if (done)
+			done--;
+	}
+	/* Done for now */
+	if (likely(napi_complete_done(napi, done)))
+		mlx4_en_arm_cq(priv, cq);
+	return done;
+}
+
+void mlx4_en_calc_rx_buf(struct net_device *dev)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	int eff_mtu = MLX4_EN_EFF_MTU(dev->mtu);
+	int i = 0;
+
+	/* bpf requires buffers to be set up as 1 packet per page.
+	 * This only works when num_frags == 1.
+	 */
+	if (priv->tx_ring_num[TX_XDP]) {
+		priv->frag_info[0].frag_size = eff_mtu;
+		/* This will gain efficient xdp frame recycling at the
+		 * expense of more costly truesize accounting
+		 */
+		priv->frag_info[0].frag_stride = PAGE_SIZE;
+		priv->dma_dir = DMA_BIDIRECTIONAL;
+		priv->rx_headroom = XDP_PACKET_HEADROOM;
+		i = 1;
+	} else {
+		int frag_size_max = 2048, buf_size = 0;
+
+		/* should not happen, right ? */
+		if (eff_mtu > PAGE_SIZE + (MLX4_EN_MAX_RX_FRAGS - 1) * 2048)
+			frag_size_max = PAGE_SIZE;
+
+		while (buf_size < eff_mtu) {
+			int frag_stride, frag_size = eff_mtu - buf_size;
+			int pad, nb;
+
+			if (i < MLX4_EN_MAX_RX_FRAGS - 1)
+				frag_size = min(frag_size, frag_size_max);
+
+			priv->frag_info[i].frag_size = frag_size;
+			frag_stride = ALIGN(frag_size, SMP_CACHE_BYTES);
+			/* We can only pack 2 1536-bytes frames in on 4K page
+			 * Therefore, each frame would consume more bytes (truesize)
+			 */
+			nb = PAGE_SIZE / frag_stride;
+			pad = (PAGE_SIZE - nb * frag_stride) / nb;
+			pad &= ~(SMP_CACHE_BYTES - 1);
+			priv->frag_info[i].frag_stride = frag_stride + pad;
+
+			buf_size += frag_size;
+			i++;
+		}
+		priv->dma_dir = DMA_FROM_DEVICE;
+		priv->rx_headroom = 0;
+	}
+
+	priv->num_frags = i;
+	priv->rx_skb_size = eff_mtu;
+	priv->log_rx_info = ROUNDUP_LOG2(i * sizeof(struct mlx4_en_rx_alloc));
+
+	en_dbg(DRV, priv, "Rx buffer scatter-list (effective-mtu:%d num_frags:%d):\n",
+	       eff_mtu, priv->num_frags);
+	for (i = 0; i < priv->num_frags; i++) {
+		en_dbg(DRV,
+		       priv,
+		       "  frag:%d - size:%d stride:%d\n",
+		       i,
+		       priv->frag_info[i].frag_size,
+		       priv->frag_info[i].frag_stride);
+	}
+}
+
+/* RSS related functions */
+
+static int mlx4_en_config_rss_qp(struct mlx4_en_priv *priv, int qpn,
+				 struct mlx4_en_rx_ring *ring,
+				 enum mlx4_qp_state *state,
+				 struct mlx4_qp *qp)
+{
+	struct mlx4_en_dev *mdev = priv->mdev;
+	struct mlx4_qp_context *context;
+	int err = 0;
+
+	context = kzalloc(sizeof(*context), GFP_KERNEL);
+	if (!context)
+		return -ENOMEM;
+
+	err = mlx4_qp_alloc(mdev->dev, qpn, qp);
+	if (err) {
+		en_err(priv, "Failed to allocate qp #%x\n", qpn);
+		goto out;
+	}
+	qp->event = mlx4_en_sqp_event;
+
+	mlx4_en_fill_qp_context(priv, ring->actual_size, ring->stride, 0, 0,
+				qpn, ring->cqn, -1, context);
+	context->db_rec_addr = cpu_to_be64(ring->wqres.db.dma);
+
+	/* Cancel FCS removal if FW allows */
+	if (mdev->dev->caps.flags & MLX4_DEV_CAP_FLAG_FCS_KEEP) {
+		context->param3 |= cpu_to_be32(1 << 29);
+		if (priv->dev->features & NETIF_F_RXFCS)
+			ring->fcs_del = 0;
+		else
+			ring->fcs_del = ETH_FCS_LEN;
+	} else
+		ring->fcs_del = 0;
+
+	err = mlx4_qp_to_ready(mdev->dev, &ring->wqres.mtt, context, qp, state);
+	if (err) {
+		mlx4_qp_remove(mdev->dev, qp);
+		mlx4_qp_free(mdev->dev, qp);
+	}
+	mlx4_en_update_rx_prod_db(ring);
+out:
+	kfree(context);
+	return err;
+}
+
+int mlx4_en_create_drop_qp(struct mlx4_en_priv *priv)
+{
+	int err;
+	u32 qpn;
+
+	err = mlx4_qp_reserve_range(priv->mdev->dev, 1, 1, &qpn,
+				    MLX4_RESERVE_A0_QP,
+				    MLX4_RES_USAGE_DRIVER);
+	if (err) {
+		en_err(priv, "Failed reserving drop qpn\n");
+		return err;
+	}
+	err = mlx4_qp_alloc(priv->mdev->dev, qpn, &priv->drop_qp);
+	if (err) {
+		en_err(priv, "Failed allocating drop qp\n");
+		mlx4_qp_release_range(priv->mdev->dev, qpn, 1);
+		return err;
+	}
+
+	return 0;
+}
+
+void mlx4_en_destroy_drop_qp(struct mlx4_en_priv *priv)
+{
+	u32 qpn;
+
+	qpn = priv->drop_qp.qpn;
+	mlx4_qp_remove(priv->mdev->dev, &priv->drop_qp);
+	mlx4_qp_free(priv->mdev->dev, &priv->drop_qp);
+	mlx4_qp_release_range(priv->mdev->dev, qpn, 1);
+}
+
+/* Allocate rx qp's and configure them according to rss map */
+int mlx4_en_config_rss_steer(struct mlx4_en_priv *priv)
+{
+	struct mlx4_en_dev *mdev = priv->mdev;
+	struct mlx4_en_rss_map *rss_map = &priv->rss_map;
+	struct mlx4_qp_context context;
+	struct mlx4_rss_context *rss_context;
+	int rss_rings;
+	void *ptr;
+	u8 rss_mask = (MLX4_RSS_IPV4 | MLX4_RSS_TCP_IPV4 | MLX4_RSS_IPV6 |
+			MLX4_RSS_TCP_IPV6);
+	int i, qpn;
+	int err = 0;
+	int good_qps = 0;
+	u8 flags;
+
+	en_dbg(DRV, priv, "Configuring rss steering\n");
+
+	flags = priv->rx_ring_num == 1 ? MLX4_RESERVE_A0_QP : 0;
+	err = mlx4_qp_reserve_range(mdev->dev, priv->rx_ring_num,
+				    priv->rx_ring_num,
+				    &rss_map->base_qpn, flags,
+				    MLX4_RES_USAGE_DRIVER);
+	if (err) {
+		en_err(priv, "Failed reserving %d qps\n", priv->rx_ring_num);
+		return err;
+	}
+
+	for (i = 0; i < priv->rx_ring_num; i++) {
+		qpn = rss_map->base_qpn + i;
+		err = mlx4_en_config_rss_qp(priv, qpn, priv->rx_ring[i],
+					    &rss_map->state[i],
+					    &rss_map->qps[i]);
+		if (err)
+			goto rss_err;
+
+		++good_qps;
+	}
+
+	if (priv->rx_ring_num == 1) {
+		rss_map->indir_qp = &rss_map->qps[0];
+		priv->base_qpn = rss_map->indir_qp->qpn;
+		en_info(priv, "Optimized Non-RSS steering\n");
+		return 0;
+	}
+
+	rss_map->indir_qp = kzalloc(sizeof(*rss_map->indir_qp), GFP_KERNEL);
+	if (!rss_map->indir_qp) {
+		err = -ENOMEM;
+		goto rss_err;
+	}
+
+	/* Configure RSS indirection qp */
+	err = mlx4_qp_alloc(mdev->dev, priv->base_qpn, rss_map->indir_qp);
+	if (err) {
+		en_err(priv, "Failed to allocate RSS indirection QP\n");
+		goto qp_alloc_err;
+	}
+
+	rss_map->indir_qp->event = mlx4_en_sqp_event;
+	mlx4_en_fill_qp_context(priv, 0, 0, 0, 1, priv->base_qpn,
+				priv->rx_ring[0]->cqn, -1, &context);
+
+	if (!priv->prof->rss_rings || priv->prof->rss_rings > priv->rx_ring_num)
+		rss_rings = priv->rx_ring_num;
+	else
+		rss_rings = priv->prof->rss_rings;
+
+	ptr = ((void *) &context) + offsetof(struct mlx4_qp_context, pri_path)
+					+ MLX4_RSS_OFFSET_IN_QPC_PRI_PATH;
+	rss_context = ptr;
+	rss_context->base_qpn = cpu_to_be32(ilog2(rss_rings) << 24 |
+					    (rss_map->base_qpn));
+	rss_context->default_qpn = cpu_to_be32(rss_map->base_qpn);
+	if (priv->mdev->profile.udp_rss) {
+		rss_mask |=  MLX4_RSS_UDP_IPV4 | MLX4_RSS_UDP_IPV6;
+		rss_context->base_qpn_udp = rss_context->default_qpn;
+	}
+
+	if (mdev->dev->caps.tunnel_offload_mode == MLX4_TUNNEL_OFFLOAD_MODE_VXLAN) {
+		en_info(priv, "Setting RSS context tunnel type to RSS on inner headers\n");
+		rss_mask |= MLX4_RSS_BY_INNER_HEADERS;
+	}
+
+	rss_context->flags = rss_mask;
+	rss_context->hash_fn = MLX4_RSS_HASH_TOP;
+	if (priv->rss_hash_fn == ETH_RSS_HASH_XOR) {
+		rss_context->hash_fn = MLX4_RSS_HASH_XOR;
+	} else if (priv->rss_hash_fn == ETH_RSS_HASH_TOP) {
+		rss_context->hash_fn = MLX4_RSS_HASH_TOP;
+		memcpy(rss_context->rss_key, priv->rss_key,
+		       MLX4_EN_RSS_KEY_SIZE);
+	} else {
+		en_err(priv, "Unknown RSS hash function requested\n");
+		err = -EINVAL;
+		goto indir_err;
+	}
+
+	err = mlx4_qp_to_ready(mdev->dev, &priv->res.mtt, &context,
+			       rss_map->indir_qp, &rss_map->indir_state);
+	if (err)
+		goto indir_err;
+
+	return 0;
+
+indir_err:
+	mlx4_qp_modify(mdev->dev, NULL, rss_map->indir_state,
+		       MLX4_QP_STATE_RST, NULL, 0, 0, rss_map->indir_qp);
+	mlx4_qp_remove(mdev->dev, rss_map->indir_qp);
+	mlx4_qp_free(mdev->dev, rss_map->indir_qp);
+qp_alloc_err:
+	kfree(rss_map->indir_qp);
+	rss_map->indir_qp = NULL;
+rss_err:
+	for (i = 0; i < good_qps; i++) {
+		mlx4_qp_modify(mdev->dev, NULL, rss_map->state[i],
+			       MLX4_QP_STATE_RST, NULL, 0, 0, &rss_map->qps[i]);
+		mlx4_qp_remove(mdev->dev, &rss_map->qps[i]);
+		mlx4_qp_free(mdev->dev, &rss_map->qps[i]);
+	}
+	mlx4_qp_release_range(mdev->dev, rss_map->base_qpn, priv->rx_ring_num);
+	return err;
+}
+
+void mlx4_en_release_rss_steer(struct mlx4_en_priv *priv)
+{
+	struct mlx4_en_dev *mdev = priv->mdev;
+	struct mlx4_en_rss_map *rss_map = &priv->rss_map;
+	int i;
+
+	if (priv->rx_ring_num > 1) {
+		mlx4_qp_modify(mdev->dev, NULL, rss_map->indir_state,
+			       MLX4_QP_STATE_RST, NULL, 0, 0,
+			       rss_map->indir_qp);
+		mlx4_qp_remove(mdev->dev, rss_map->indir_qp);
+		mlx4_qp_free(mdev->dev, rss_map->indir_qp);
+		kfree(rss_map->indir_qp);
+		rss_map->indir_qp = NULL;
+	}
+
+	for (i = 0; i < priv->rx_ring_num; i++) {
+		mlx4_qp_modify(mdev->dev, NULL, rss_map->state[i],
+			       MLX4_QP_STATE_RST, NULL, 0, 0, &rss_map->qps[i]);
+		mlx4_qp_remove(mdev->dev, &rss_map->qps[i]);
+		mlx4_qp_free(mdev->dev, &rss_map->qps[i]);
+	}
+	mlx4_qp_release_range(mdev->dev, rss_map->base_qpn, priv->rx_ring_num);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx4/en_selftest.c b/drivers/net/ethernet/mellanox/mlx4/en_selftest.c
new file mode 100644
index 000000000..946d9db7c
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx4/en_selftest.c
@@ -0,0 +1,204 @@
+/*
+ * Copyright (c) 2007 Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+
+#include <linux/kernel.h>
+#include <linux/ethtool.h>
+#include <linux/netdevice.h>
+#include <linux/delay.h>
+#include <linux/mlx4/driver.h>
+
+#include "mlx4_en.h"
+
+
+static int mlx4_en_test_registers(struct mlx4_en_priv *priv)
+{
+	return mlx4_cmd(priv->mdev->dev, 0, 0, 0, MLX4_CMD_HW_HEALTH_CHECK,
+			MLX4_CMD_TIME_CLASS_A, MLX4_CMD_WRAPPED);
+}
+
+static int mlx4_en_test_loopback_xmit(struct mlx4_en_priv *priv)
+{
+	struct sk_buff *skb;
+	struct ethhdr *ethh;
+	unsigned char *packet;
+	unsigned int packet_size = MLX4_LOOPBACK_TEST_PAYLOAD;
+	unsigned int i;
+	int err;
+
+
+	/* build the pkt before xmit */
+	skb = netdev_alloc_skb(priv->dev, MLX4_LOOPBACK_TEST_PAYLOAD + ETH_HLEN + NET_IP_ALIGN);
+	if (!skb)
+		return -ENOMEM;
+
+	skb_reserve(skb, NET_IP_ALIGN);
+
+	ethh = skb_put(skb, sizeof(struct ethhdr));
+	packet = skb_put(skb, packet_size);
+	memcpy(ethh->h_dest, priv->dev->dev_addr, ETH_ALEN);
+	eth_zero_addr(ethh->h_source);
+	ethh->h_proto = htons(ETH_P_ARP);
+	skb_reset_mac_header(skb);
+	for (i = 0; i < packet_size; ++i)	/* fill our packet */
+		packet[i] = (unsigned char)(i & 0xff);
+
+	/* xmit the pkt */
+	err = mlx4_en_xmit(skb, priv->dev);
+	return err;
+}
+
+static int mlx4_en_test_loopback(struct mlx4_en_priv *priv)
+{
+	u32 loopback_ok = 0;
+	int i;
+
+        priv->loopback_ok = 0;
+	priv->validate_loopback = 1;
+
+	mlx4_en_update_loopback_state(priv->dev, priv->dev->features);
+
+	/* xmit */
+	if (mlx4_en_test_loopback_xmit(priv)) {
+		en_err(priv, "Transmitting loopback packet failed\n");
+		goto mlx4_en_test_loopback_exit;
+	}
+
+	/* polling for result */
+	for (i = 0; i < MLX4_EN_LOOPBACK_RETRIES; ++i) {
+		msleep(MLX4_EN_LOOPBACK_TIMEOUT);
+		if (priv->loopback_ok) {
+			loopback_ok = 1;
+			break;
+		}
+	}
+	if (!loopback_ok)
+		en_err(priv, "Loopback packet didn't arrive\n");
+
+mlx4_en_test_loopback_exit:
+
+	priv->validate_loopback = 0;
+
+	mlx4_en_update_loopback_state(priv->dev, priv->dev->features);
+	return !loopback_ok;
+}
+
+static int mlx4_en_test_interrupts(struct mlx4_en_priv *priv)
+{
+	struct mlx4_en_dev *mdev = priv->mdev;
+	int err = 0;
+	int i = 0;
+
+	err = mlx4_test_async(mdev->dev);
+	/* When not in MSI_X or slave, test only async */
+	if (!(mdev->dev->flags & MLX4_FLAG_MSI_X) || mlx4_is_slave(mdev->dev))
+		return err;
+
+	/* A loop over all completion vectors of current port,
+	 * for each vector check whether it works by mapping command
+	 * completions to that vector and performing a NOP command
+	 */
+	for (i = 0; i < priv->rx_ring_num; i++) {
+		err = mlx4_test_interrupt(mdev->dev, priv->rx_cq[i]->vector);
+		if (err)
+			break;
+	}
+
+	return err;
+}
+
+static int mlx4_en_test_link(struct mlx4_en_priv *priv)
+{
+	if (mlx4_en_QUERY_PORT(priv->mdev, priv->port))
+		return -ENOMEM;
+	if (priv->port_state.link_state == 1)
+		return 0;
+	else
+		return 1;
+}
+
+static int mlx4_en_test_speed(struct mlx4_en_priv *priv)
+{
+
+	if (mlx4_en_QUERY_PORT(priv->mdev, priv->port))
+		return -ENOMEM;
+
+	/* The device supports 100M, 1G, 10G, 20G, 40G and 56G speed */
+	if (priv->port_state.link_speed != SPEED_100 &&
+	    priv->port_state.link_speed != SPEED_1000 &&
+	    priv->port_state.link_speed != SPEED_10000 &&
+	    priv->port_state.link_speed != SPEED_20000 &&
+	    priv->port_state.link_speed != SPEED_40000 &&
+	    priv->port_state.link_speed != SPEED_56000)
+		return priv->port_state.link_speed;
+
+	return 0;
+}
+
+
+void mlx4_en_ex_selftest(struct net_device *dev, u32 *flags, u64 *buf)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	int i, carrier_ok;
+
+	memset(buf, 0, sizeof(u64) * MLX4_EN_NUM_SELF_TEST);
+
+	if (*flags & ETH_TEST_FL_OFFLINE) {
+		/* disable the interface */
+		carrier_ok = netif_carrier_ok(dev);
+
+		netif_carrier_off(dev);
+		/* Wait until all tx queues are empty.
+		 * there should not be any additional incoming traffic
+		 * since we turned the carrier off */
+		msleep(200);
+
+		if (priv->mdev->dev->caps.flags &
+					MLX4_DEV_CAP_FLAG_UC_LOOPBACK) {
+			buf[3] = mlx4_en_test_registers(priv);
+			if (priv->port_up && dev->mtu >= MLX4_SELFTEST_LB_MIN_MTU)
+				buf[4] = mlx4_en_test_loopback(priv);
+		}
+
+		if (carrier_ok)
+			netif_carrier_on(dev);
+
+	}
+	buf[0] = mlx4_en_test_interrupts(priv);
+	buf[1] = mlx4_en_test_link(priv);
+	buf[2] = mlx4_en_test_speed(priv);
+
+	for (i = 0; i < MLX4_EN_NUM_SELF_TEST; i++) {
+		if (buf[i])
+			*flags |= ETH_TEST_FL_FAILED;
+	}
+}
diff --git a/drivers/net/ethernet/mellanox/mlx4/en_tx.c b/drivers/net/ethernet/mellanox/mlx4/en_tx.c
new file mode 100644
index 000000000..7fccf1a79
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx4/en_tx.c
@@ -0,0 +1,1233 @@
+/*
+ * Copyright (c) 2007 Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+
+#include <asm/page.h>
+#include <linux/mlx4/cq.h>
+#include <linux/slab.h>
+#include <linux/mlx4/qp.h>
+#include <linux/skbuff.h>
+#include <linux/if_vlan.h>
+#include <linux/prefetch.h>
+#include <linux/vmalloc.h>
+#include <linux/tcp.h>
+#include <linux/ip.h>
+#include <linux/ipv6.h>
+#include <linux/indirect_call_wrapper.h>
+#include <net/ipv6.h>
+
+#include "mlx4_en.h"
+
+int mlx4_en_create_tx_ring(struct mlx4_en_priv *priv,
+			   struct mlx4_en_tx_ring **pring, u32 size,
+			   u16 stride, int node, int queue_index)
+{
+	struct mlx4_en_dev *mdev = priv->mdev;
+	struct mlx4_en_tx_ring *ring;
+	int tmp;
+	int err;
+
+	ring = kzalloc_node(sizeof(*ring), GFP_KERNEL, node);
+	if (!ring) {
+		en_err(priv, "Failed allocating TX ring\n");
+		return -ENOMEM;
+	}
+
+	ring->size = size;
+	ring->size_mask = size - 1;
+	ring->sp_stride = stride;
+	ring->full_size = ring->size - HEADROOM - MAX_DESC_TXBBS;
+
+	tmp = size * sizeof(struct mlx4_en_tx_info);
+	ring->tx_info = kvmalloc_node(tmp, GFP_KERNEL, node);
+	if (!ring->tx_info) {
+		err = -ENOMEM;
+		goto err_ring;
+	}
+
+	en_dbg(DRV, priv, "Allocated tx_info ring at addr:%p size:%d\n",
+		 ring->tx_info, tmp);
+
+	ring->bounce_buf = kmalloc_node(MAX_DESC_SIZE, GFP_KERNEL, node);
+	if (!ring->bounce_buf) {
+		ring->bounce_buf = kmalloc(MAX_DESC_SIZE, GFP_KERNEL);
+		if (!ring->bounce_buf) {
+			err = -ENOMEM;
+			goto err_info;
+		}
+	}
+	ring->buf_size = ALIGN(size * ring->sp_stride, MLX4_EN_PAGE_SIZE);
+
+	/* Allocate HW buffers on provided NUMA node */
+	set_dev_node(&mdev->dev->persist->pdev->dev, node);
+	err = mlx4_alloc_hwq_res(mdev->dev, &ring->sp_wqres, ring->buf_size);
+	set_dev_node(&mdev->dev->persist->pdev->dev, mdev->dev->numa_node);
+	if (err) {
+		en_err(priv, "Failed allocating hwq resources\n");
+		goto err_bounce;
+	}
+
+	ring->buf = ring->sp_wqres.buf.direct.buf;
+
+	en_dbg(DRV, priv, "Allocated TX ring (addr:%p) - buf:%p size:%d buf_size:%d dma:%llx\n",
+	       ring, ring->buf, ring->size, ring->buf_size,
+	       (unsigned long long) ring->sp_wqres.buf.direct.map);
+
+	err = mlx4_qp_reserve_range(mdev->dev, 1, 1, &ring->qpn,
+				    MLX4_RESERVE_ETH_BF_QP,
+				    MLX4_RES_USAGE_DRIVER);
+	if (err) {
+		en_err(priv, "failed reserving qp for TX ring\n");
+		goto err_hwq_res;
+	}
+
+	err = mlx4_qp_alloc(mdev->dev, ring->qpn, &ring->sp_qp);
+	if (err) {
+		en_err(priv, "Failed allocating qp %d\n", ring->qpn);
+		goto err_reserve;
+	}
+	ring->sp_qp.event = mlx4_en_sqp_event;
+
+	err = mlx4_bf_alloc(mdev->dev, &ring->bf, node);
+	if (err) {
+		en_dbg(DRV, priv, "working without blueflame (%d)\n", err);
+		ring->bf.uar = &mdev->priv_uar;
+		ring->bf.uar->map = mdev->uar_map;
+		ring->bf_enabled = false;
+		ring->bf_alloced = false;
+		priv->pflags &= ~MLX4_EN_PRIV_FLAGS_BLUEFLAME;
+	} else {
+		ring->bf_alloced = true;
+		ring->bf_enabled = !!(priv->pflags &
+				      MLX4_EN_PRIV_FLAGS_BLUEFLAME);
+	}
+	ring->doorbell_address = ring->bf.uar->map + MLX4_SEND_DOORBELL;
+
+	ring->hwtstamp_tx_type = priv->hwtstamp_config.tx_type;
+	ring->queue_index = queue_index;
+
+	if (queue_index < priv->num_tx_rings_p_up)
+		cpumask_set_cpu(cpumask_local_spread(queue_index,
+						     priv->mdev->dev->numa_node),
+				&ring->sp_affinity_mask);
+
+	*pring = ring;
+	return 0;
+
+err_reserve:
+	mlx4_qp_release_range(mdev->dev, ring->qpn, 1);
+err_hwq_res:
+	mlx4_free_hwq_res(mdev->dev, &ring->sp_wqres, ring->buf_size);
+err_bounce:
+	kfree(ring->bounce_buf);
+	ring->bounce_buf = NULL;
+err_info:
+	kvfree(ring->tx_info);
+	ring->tx_info = NULL;
+err_ring:
+	kfree(ring);
+	*pring = NULL;
+	return err;
+}
+
+void mlx4_en_destroy_tx_ring(struct mlx4_en_priv *priv,
+			     struct mlx4_en_tx_ring **pring)
+{
+	struct mlx4_en_dev *mdev = priv->mdev;
+	struct mlx4_en_tx_ring *ring = *pring;
+	en_dbg(DRV, priv, "Destroying tx ring, qpn: %d\n", ring->qpn);
+
+	if (ring->bf_alloced)
+		mlx4_bf_free(mdev->dev, &ring->bf);
+	mlx4_qp_remove(mdev->dev, &ring->sp_qp);
+	mlx4_qp_free(mdev->dev, &ring->sp_qp);
+	mlx4_qp_release_range(priv->mdev->dev, ring->qpn, 1);
+	mlx4_free_hwq_res(mdev->dev, &ring->sp_wqres, ring->buf_size);
+	kfree(ring->bounce_buf);
+	ring->bounce_buf = NULL;
+	kvfree(ring->tx_info);
+	ring->tx_info = NULL;
+	kfree(ring);
+	*pring = NULL;
+}
+
+int mlx4_en_activate_tx_ring(struct mlx4_en_priv *priv,
+			     struct mlx4_en_tx_ring *ring,
+			     int cq, int user_prio)
+{
+	struct mlx4_en_dev *mdev = priv->mdev;
+	int err;
+
+	ring->sp_cqn = cq;
+	ring->prod = 0;
+	ring->cons = 0xffffffff;
+	ring->last_nr_txbb = 1;
+	memset(ring->tx_info, 0, ring->size * sizeof(struct mlx4_en_tx_info));
+	memset(ring->buf, 0, ring->buf_size);
+	ring->free_tx_desc = mlx4_en_free_tx_desc;
+
+	ring->sp_qp_state = MLX4_QP_STATE_RST;
+	ring->doorbell_qpn = cpu_to_be32(ring->sp_qp.qpn << 8);
+	ring->mr_key = cpu_to_be32(mdev->mr.key);
+
+	mlx4_en_fill_qp_context(priv, ring->size, ring->sp_stride, 1, 0, ring->qpn,
+				ring->sp_cqn, user_prio, &ring->sp_context);
+	if (ring->bf_alloced)
+		ring->sp_context.usr_page =
+			cpu_to_be32(mlx4_to_hw_uar_index(mdev->dev,
+							 ring->bf.uar->index));
+
+	err = mlx4_qp_to_ready(mdev->dev, &ring->sp_wqres.mtt, &ring->sp_context,
+			       &ring->sp_qp, &ring->sp_qp_state);
+	if (!cpumask_empty(&ring->sp_affinity_mask))
+		netif_set_xps_queue(priv->dev, &ring->sp_affinity_mask,
+				    ring->queue_index);
+
+	return err;
+}
+
+void mlx4_en_deactivate_tx_ring(struct mlx4_en_priv *priv,
+				struct mlx4_en_tx_ring *ring)
+{
+	struct mlx4_en_dev *mdev = priv->mdev;
+
+	mlx4_qp_modify(mdev->dev, NULL, ring->sp_qp_state,
+		       MLX4_QP_STATE_RST, NULL, 0, 0, &ring->sp_qp);
+}
+
+static inline bool mlx4_en_is_tx_ring_full(struct mlx4_en_tx_ring *ring)
+{
+	return ring->prod - ring->cons > ring->full_size;
+}
+
+static void mlx4_en_stamp_wqe(struct mlx4_en_priv *priv,
+			      struct mlx4_en_tx_ring *ring, int index,
+			      u8 owner)
+{
+	__be32 stamp = cpu_to_be32(STAMP_VAL | (!!owner << STAMP_SHIFT));
+	struct mlx4_en_tx_desc *tx_desc = ring->buf + (index << LOG_TXBB_SIZE);
+	struct mlx4_en_tx_info *tx_info = &ring->tx_info[index];
+	void *end = ring->buf + ring->buf_size;
+	__be32 *ptr = (__be32 *)tx_desc;
+	int i;
+
+	/* Optimize the common case when there are no wraparounds */
+	if (likely((void *)tx_desc +
+		   (tx_info->nr_txbb << LOG_TXBB_SIZE) <= end)) {
+		/* Stamp the freed descriptor */
+		for (i = 0; i < tx_info->nr_txbb << LOG_TXBB_SIZE;
+		     i += STAMP_STRIDE) {
+			*ptr = stamp;
+			ptr += STAMP_DWORDS;
+		}
+	} else {
+		/* Stamp the freed descriptor */
+		for (i = 0; i < tx_info->nr_txbb << LOG_TXBB_SIZE;
+		     i += STAMP_STRIDE) {
+			*ptr = stamp;
+			ptr += STAMP_DWORDS;
+			if ((void *)ptr >= end) {
+				ptr = ring->buf;
+				stamp ^= cpu_to_be32(0x80000000);
+			}
+		}
+	}
+}
+
+INDIRECT_CALLABLE_DECLARE(u32 mlx4_en_free_tx_desc(struct mlx4_en_priv *priv,
+						   struct mlx4_en_tx_ring *ring,
+						   int index, u64 timestamp,
+						   int napi_mode));
+
+u32 mlx4_en_free_tx_desc(struct mlx4_en_priv *priv,
+			 struct mlx4_en_tx_ring *ring,
+			 int index, u64 timestamp,
+			 int napi_mode)
+{
+	struct mlx4_en_tx_info *tx_info = &ring->tx_info[index];
+	struct mlx4_en_tx_desc *tx_desc = ring->buf + (index << LOG_TXBB_SIZE);
+	struct mlx4_wqe_data_seg *data = (void *) tx_desc + tx_info->data_offset;
+	void *end = ring->buf + ring->buf_size;
+	struct sk_buff *skb = tx_info->skb;
+	int nr_maps = tx_info->nr_maps;
+	int i;
+
+	/* We do not touch skb here, so prefetch skb->users location
+	 * to speedup consume_skb()
+	 */
+	prefetchw(&skb->users);
+
+	if (unlikely(timestamp)) {
+		struct skb_shared_hwtstamps hwts;
+
+		mlx4_en_fill_hwtstamps(priv->mdev, &hwts, timestamp);
+		skb_tstamp_tx(skb, &hwts);
+	}
+
+	if (!tx_info->inl) {
+		if (tx_info->linear)
+			dma_unmap_single(priv->ddev,
+					 tx_info->map0_dma,
+					 tx_info->map0_byte_count,
+					 DMA_TO_DEVICE);
+		else
+			dma_unmap_page(priv->ddev,
+				       tx_info->map0_dma,
+				       tx_info->map0_byte_count,
+				       DMA_TO_DEVICE);
+		/* Optimize the common case when there are no wraparounds */
+		if (likely((void *)tx_desc +
+			   (tx_info->nr_txbb << LOG_TXBB_SIZE) <= end)) {
+			for (i = 1; i < nr_maps; i++) {
+				data++;
+				dma_unmap_page(priv->ddev,
+					(dma_addr_t)be64_to_cpu(data->addr),
+					be32_to_cpu(data->byte_count),
+					DMA_TO_DEVICE);
+			}
+		} else {
+			if ((void *)data >= end)
+				data = ring->buf + ((void *)data - end);
+
+			for (i = 1; i < nr_maps; i++) {
+				data++;
+				/* Check for wraparound before unmapping */
+				if ((void *) data >= end)
+					data = ring->buf;
+				dma_unmap_page(priv->ddev,
+					(dma_addr_t)be64_to_cpu(data->addr),
+					be32_to_cpu(data->byte_count),
+					DMA_TO_DEVICE);
+			}
+		}
+	}
+	napi_consume_skb(skb, napi_mode);
+
+	return tx_info->nr_txbb;
+}
+
+INDIRECT_CALLABLE_DECLARE(u32 mlx4_en_recycle_tx_desc(struct mlx4_en_priv *priv,
+						      struct mlx4_en_tx_ring *ring,
+						      int index, u64 timestamp,
+						      int napi_mode));
+
+u32 mlx4_en_recycle_tx_desc(struct mlx4_en_priv *priv,
+			    struct mlx4_en_tx_ring *ring,
+			    int index, u64 timestamp,
+			    int napi_mode)
+{
+	struct mlx4_en_tx_info *tx_info = &ring->tx_info[index];
+	struct mlx4_en_rx_alloc frame = {
+		.page = tx_info->page,
+		.dma = tx_info->map0_dma,
+	};
+
+	if (!napi_mode || !mlx4_en_rx_recycle(ring->recycle_ring, &frame)) {
+		dma_unmap_page(priv->ddev, tx_info->map0_dma,
+			       PAGE_SIZE, priv->dma_dir);
+		put_page(tx_info->page);
+	}
+
+	return tx_info->nr_txbb;
+}
+
+int mlx4_en_free_tx_buf(struct net_device *dev, struct mlx4_en_tx_ring *ring)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	int cnt = 0;
+
+	/* Skip last polled descriptor */
+	ring->cons += ring->last_nr_txbb;
+	en_dbg(DRV, priv, "Freeing Tx buf - cons:0x%x prod:0x%x\n",
+		 ring->cons, ring->prod);
+
+	if ((u32) (ring->prod - ring->cons) > ring->size) {
+		if (netif_msg_tx_err(priv))
+			en_warn(priv, "Tx consumer passed producer!\n");
+		return 0;
+	}
+
+	while (ring->cons != ring->prod) {
+		ring->last_nr_txbb = ring->free_tx_desc(priv, ring,
+						ring->cons & ring->size_mask,
+						0, 0 /* Non-NAPI caller */);
+		ring->cons += ring->last_nr_txbb;
+		cnt++;
+	}
+
+	if (ring->tx_queue)
+		netdev_tx_reset_queue(ring->tx_queue);
+
+	if (cnt)
+		en_dbg(DRV, priv, "Freed %d uncompleted tx descriptors\n", cnt);
+
+	return cnt;
+}
+
+static void mlx4_en_handle_err_cqe(struct mlx4_en_priv *priv, struct mlx4_err_cqe *err_cqe,
+				   u16 cqe_index, struct mlx4_en_tx_ring *ring)
+{
+	struct mlx4_en_dev *mdev = priv->mdev;
+	struct mlx4_en_tx_info *tx_info;
+	struct mlx4_en_tx_desc *tx_desc;
+	u16 wqe_index;
+	int desc_size;
+
+	en_err(priv, "CQE error - cqn 0x%x, ci 0x%x, vendor syndrome: 0x%x syndrome: 0x%x\n",
+	       ring->sp_cqn, cqe_index, err_cqe->vendor_err_syndrome, err_cqe->syndrome);
+	print_hex_dump(KERN_WARNING, "", DUMP_PREFIX_OFFSET, 16, 1, err_cqe, sizeof(*err_cqe),
+		       false);
+
+	wqe_index = be16_to_cpu(err_cqe->wqe_index) & ring->size_mask;
+	tx_info = &ring->tx_info[wqe_index];
+	desc_size = tx_info->nr_txbb << LOG_TXBB_SIZE;
+	en_err(priv, "Related WQE - qpn 0x%x, wqe index 0x%x, wqe size 0x%x\n", ring->qpn,
+	       wqe_index, desc_size);
+	tx_desc = ring->buf + (wqe_index << LOG_TXBB_SIZE);
+	print_hex_dump(KERN_WARNING, "", DUMP_PREFIX_OFFSET, 16, 1, tx_desc, desc_size, false);
+
+	if (test_and_set_bit(MLX4_EN_STATE_FLAG_RESTARTING, &priv->state))
+		return;
+
+	en_err(priv, "Scheduling port restart\n");
+	queue_work(mdev->workqueue, &priv->restart_task);
+}
+
+int mlx4_en_process_tx_cq(struct net_device *dev,
+			  struct mlx4_en_cq *cq, int napi_budget)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	struct mlx4_cq *mcq = &cq->mcq;
+	struct mlx4_en_tx_ring *ring = priv->tx_ring[cq->type][cq->ring];
+	struct mlx4_cqe *cqe;
+	u16 index, ring_index, stamp_index;
+	u32 txbbs_skipped = 0;
+	u32 txbbs_stamp = 0;
+	u32 cons_index = mcq->cons_index;
+	int size = cq->size;
+	u32 size_mask = ring->size_mask;
+	struct mlx4_cqe *buf = cq->buf;
+	u32 packets = 0;
+	u32 bytes = 0;
+	int factor = priv->cqe_factor;
+	int done = 0;
+	int budget = priv->tx_work_limit;
+	u32 last_nr_txbb;
+	u32 ring_cons;
+
+	if (unlikely(!priv->port_up))
+		return 0;
+
+	netdev_txq_bql_complete_prefetchw(ring->tx_queue);
+
+	index = cons_index & size_mask;
+	cqe = mlx4_en_get_cqe(buf, index, priv->cqe_size) + factor;
+	last_nr_txbb = READ_ONCE(ring->last_nr_txbb);
+	ring_cons = READ_ONCE(ring->cons);
+	ring_index = ring_cons & size_mask;
+	stamp_index = ring_index;
+
+	/* Process all completed CQEs */
+	while (XNOR(cqe->owner_sr_opcode & MLX4_CQE_OWNER_MASK,
+			cons_index & size) && (done < budget)) {
+		u16 new_index;
+
+		/*
+		 * make sure we read the CQE after we read the
+		 * ownership bit
+		 */
+		dma_rmb();
+
+		if (unlikely((cqe->owner_sr_opcode & MLX4_CQE_OPCODE_MASK) ==
+			     MLX4_CQE_OPCODE_ERROR))
+			if (!test_and_set_bit(MLX4_EN_TX_RING_STATE_RECOVERING, &ring->state))
+				mlx4_en_handle_err_cqe(priv, (struct mlx4_err_cqe *)cqe, index,
+						       ring);
+
+		/* Skip over last polled CQE */
+		new_index = be16_to_cpu(cqe->wqe_index) & size_mask;
+
+		do {
+			u64 timestamp = 0;
+
+			txbbs_skipped += last_nr_txbb;
+			ring_index = (ring_index + last_nr_txbb) & size_mask;
+
+			if (unlikely(ring->tx_info[ring_index].ts_requested))
+				timestamp = mlx4_en_get_cqe_ts(cqe);
+
+			/* free next descriptor */
+			last_nr_txbb = INDIRECT_CALL_2(ring->free_tx_desc,
+						       mlx4_en_free_tx_desc,
+						       mlx4_en_recycle_tx_desc,
+					priv, ring, ring_index,
+					timestamp, napi_budget);
+
+			mlx4_en_stamp_wqe(priv, ring, stamp_index,
+					  !!((ring_cons + txbbs_stamp) &
+						ring->size));
+			stamp_index = ring_index;
+			txbbs_stamp = txbbs_skipped;
+			packets++;
+			bytes += ring->tx_info[ring_index].nr_bytes;
+		} while ((++done < budget) && (ring_index != new_index));
+
+		++cons_index;
+		index = cons_index & size_mask;
+		cqe = mlx4_en_get_cqe(buf, index, priv->cqe_size) + factor;
+	}
+
+	/*
+	 * To prevent CQ overflow we first update CQ consumer and only then
+	 * the ring consumer.
+	 */
+	mcq->cons_index = cons_index;
+	mlx4_cq_set_ci(mcq);
+	wmb();
+
+	/* we want to dirty this cache line once */
+	WRITE_ONCE(ring->last_nr_txbb, last_nr_txbb);
+	WRITE_ONCE(ring->cons, ring_cons + txbbs_skipped);
+
+	if (cq->type == TX_XDP)
+		return done;
+
+	netdev_tx_completed_queue(ring->tx_queue, packets, bytes);
+
+	/* Wakeup Tx queue if this stopped, and ring is not full.
+	 */
+	if (netif_tx_queue_stopped(ring->tx_queue) &&
+	    !mlx4_en_is_tx_ring_full(ring)) {
+		netif_tx_wake_queue(ring->tx_queue);
+		ring->wake_queue++;
+	}
+
+	return done;
+}
+
+void mlx4_en_tx_irq(struct mlx4_cq *mcq)
+{
+	struct mlx4_en_cq *cq = container_of(mcq, struct mlx4_en_cq, mcq);
+	struct mlx4_en_priv *priv = netdev_priv(cq->dev);
+
+	if (likely(priv->port_up))
+		napi_schedule_irqoff(&cq->napi);
+	else
+		mlx4_en_arm_cq(priv, cq);
+}
+
+/* TX CQ polling - called by NAPI */
+int mlx4_en_poll_tx_cq(struct napi_struct *napi, int budget)
+{
+	struct mlx4_en_cq *cq = container_of(napi, struct mlx4_en_cq, napi);
+	struct net_device *dev = cq->dev;
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	int work_done;
+
+	work_done = mlx4_en_process_tx_cq(dev, cq, budget);
+	if (work_done >= budget)
+		return budget;
+
+	if (napi_complete_done(napi, work_done))
+		mlx4_en_arm_cq(priv, cq);
+
+	return 0;
+}
+
+static struct mlx4_en_tx_desc *mlx4_en_bounce_to_desc(struct mlx4_en_priv *priv,
+						      struct mlx4_en_tx_ring *ring,
+						      u32 index,
+						      unsigned int desc_size)
+{
+	u32 copy = (ring->size - index) << LOG_TXBB_SIZE;
+	int i;
+
+	for (i = desc_size - copy - 4; i >= 0; i -= 4) {
+		if ((i & (TXBB_SIZE - 1)) == 0)
+			wmb();
+
+		*((u32 *) (ring->buf + i)) =
+			*((u32 *) (ring->bounce_buf + copy + i));
+	}
+
+	for (i = copy - 4; i >= 4 ; i -= 4) {
+		if ((i & (TXBB_SIZE - 1)) == 0)
+			wmb();
+
+		*((u32 *)(ring->buf + (index << LOG_TXBB_SIZE) + i)) =
+			*((u32 *) (ring->bounce_buf + i));
+	}
+
+	/* Return real descriptor location */
+	return ring->buf + (index << LOG_TXBB_SIZE);
+}
+
+/* Decide if skb can be inlined in tx descriptor to avoid dma mapping
+ *
+ * It seems strange we do not simply use skb_copy_bits().
+ * This would allow to inline all skbs iff skb->len <= inline_thold
+ *
+ * Note that caller already checked skb was not a gso packet
+ */
+static bool is_inline(int inline_thold, const struct sk_buff *skb,
+		      const struct skb_shared_info *shinfo,
+		      void **pfrag)
+{
+	void *ptr;
+
+	if (skb->len > inline_thold || !inline_thold)
+		return false;
+
+	if (shinfo->nr_frags == 1) {
+		ptr = skb_frag_address_safe(&shinfo->frags[0]);
+		if (unlikely(!ptr))
+			return false;
+		*pfrag = ptr;
+		return true;
+	}
+	if (shinfo->nr_frags)
+		return false;
+	return true;
+}
+
+static int inline_size(const struct sk_buff *skb)
+{
+	if (skb->len + CTRL_SIZE + sizeof(struct mlx4_wqe_inline_seg)
+	    <= MLX4_INLINE_ALIGN)
+		return ALIGN(skb->len + CTRL_SIZE +
+			     sizeof(struct mlx4_wqe_inline_seg), 16);
+	else
+		return ALIGN(skb->len + CTRL_SIZE + 2 *
+			     sizeof(struct mlx4_wqe_inline_seg), 16);
+}
+
+static int get_real_size(const struct sk_buff *skb,
+			 const struct skb_shared_info *shinfo,
+			 struct net_device *dev,
+			 int *lso_header_size,
+			 bool *inline_ok,
+			 void **pfrag,
+			 int *hopbyhop)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	int real_size;
+
+	if (shinfo->gso_size) {
+		*inline_ok = false;
+		*hopbyhop = 0;
+		if (skb->encapsulation) {
+			*lso_header_size = skb_inner_tcp_all_headers(skb);
+		} else {
+			/* Detects large IPV6 TCP packets and prepares for removal of
+			 * HBH header that has been pushed by ip6_xmit(),
+			 * mainly so that tcpdump can dissect them.
+			 */
+			if (ipv6_has_hopopt_jumbo(skb))
+				*hopbyhop = sizeof(struct hop_jumbo_hdr);
+			*lso_header_size = skb_tcp_all_headers(skb);
+		}
+		real_size = CTRL_SIZE + shinfo->nr_frags * DS_SIZE +
+			ALIGN(*lso_header_size - *hopbyhop + 4, DS_SIZE);
+		if (unlikely(*lso_header_size != skb_headlen(skb))) {
+			/* We add a segment for the skb linear buffer only if
+			 * it contains data */
+			if (*lso_header_size < skb_headlen(skb))
+				real_size += DS_SIZE;
+			else {
+				if (netif_msg_tx_err(priv))
+					en_warn(priv, "Non-linear headers\n");
+				return 0;
+			}
+		}
+	} else {
+		*lso_header_size = 0;
+		*inline_ok = is_inline(priv->prof->inline_thold, skb,
+				       shinfo, pfrag);
+
+		if (*inline_ok)
+			real_size = inline_size(skb);
+		else
+			real_size = CTRL_SIZE +
+				    (shinfo->nr_frags + 1) * DS_SIZE;
+	}
+
+	return real_size;
+}
+
+static void build_inline_wqe(struct mlx4_en_tx_desc *tx_desc,
+			     const struct sk_buff *skb,
+			     const struct skb_shared_info *shinfo,
+			     void *fragptr)
+{
+	struct mlx4_wqe_inline_seg *inl = &tx_desc->inl;
+	int spc = MLX4_INLINE_ALIGN - CTRL_SIZE - sizeof(*inl);
+	unsigned int hlen = skb_headlen(skb);
+
+	if (skb->len <= spc) {
+		if (likely(skb->len >= MIN_PKT_LEN)) {
+			inl->byte_count = cpu_to_be32(1 << 31 | skb->len);
+		} else {
+			inl->byte_count = cpu_to_be32(1 << 31 | MIN_PKT_LEN);
+			memset(inl->data + skb->len, 0,
+			       MIN_PKT_LEN - skb->len);
+		}
+		skb_copy_from_linear_data(skb, inl->data, hlen);
+		if (shinfo->nr_frags)
+			memcpy(inl->data + hlen, fragptr,
+			       skb_frag_size(&shinfo->frags[0]));
+
+	} else {
+		inl->byte_count = cpu_to_be32(1 << 31 | spc);
+		if (hlen <= spc) {
+			skb_copy_from_linear_data(skb, inl->data, hlen);
+			if (hlen < spc) {
+				memcpy(inl->data + hlen,
+				       fragptr, spc - hlen);
+				fragptr +=  spc - hlen;
+			}
+			inl = (void *)inl->data + spc;
+			memcpy(inl->data, fragptr, skb->len - spc);
+		} else {
+			skb_copy_from_linear_data(skb, inl->data, spc);
+			inl = (void *)inl->data + spc;
+			skb_copy_from_linear_data_offset(skb, spc, inl->data,
+							 hlen - spc);
+			if (shinfo->nr_frags)
+				memcpy(inl->data + hlen - spc,
+				       fragptr,
+				       skb_frag_size(&shinfo->frags[0]));
+		}
+
+		dma_wmb();
+		inl->byte_count = cpu_to_be32(1 << 31 | (skb->len - spc));
+	}
+}
+
+u16 mlx4_en_select_queue(struct net_device *dev, struct sk_buff *skb,
+			 struct net_device *sb_dev)
+{
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	u16 rings_p_up = priv->num_tx_rings_p_up;
+
+	if (netdev_get_num_tc(dev))
+		return netdev_pick_tx(dev, skb, NULL);
+
+	return netdev_pick_tx(dev, skb, NULL) % rings_p_up;
+}
+
+static void mlx4_bf_copy(void __iomem *dst, const void *src,
+			 unsigned int bytecnt)
+{
+	__iowrite64_copy(dst, src, bytecnt / 8);
+}
+
+void mlx4_en_xmit_doorbell(struct mlx4_en_tx_ring *ring)
+{
+	wmb();
+	/* Since there is no iowrite*_native() that writes the
+	 * value as is, without byteswapping - using the one
+	 * the doesn't do byteswapping in the relevant arch
+	 * endianness.
+	 */
+#if defined(__LITTLE_ENDIAN)
+	iowrite32(
+#else
+	iowrite32be(
+#endif
+		  (__force u32)ring->doorbell_qpn, ring->doorbell_address);
+}
+
+static void mlx4_en_tx_write_desc(struct mlx4_en_tx_ring *ring,
+				  struct mlx4_en_tx_desc *tx_desc,
+				  union mlx4_wqe_qpn_vlan qpn_vlan,
+				  int desc_size, int bf_index,
+				  __be32 op_own, bool bf_ok,
+				  bool send_doorbell)
+{
+	tx_desc->ctrl.qpn_vlan = qpn_vlan;
+
+	if (bf_ok) {
+		op_own |= htonl((bf_index & 0xffff) << 8);
+		/* Ensure new descriptor hits memory
+		 * before setting ownership of this descriptor to HW
+		 */
+		dma_wmb();
+		tx_desc->ctrl.owner_opcode = op_own;
+
+		wmb();
+
+		mlx4_bf_copy(ring->bf.reg + ring->bf.offset, &tx_desc->ctrl,
+			     desc_size);
+
+		wmb();
+
+		ring->bf.offset ^= ring->bf.buf_size;
+	} else {
+		/* Ensure new descriptor hits memory
+		 * before setting ownership of this descriptor to HW
+		 */
+		dma_wmb();
+		tx_desc->ctrl.owner_opcode = op_own;
+		if (send_doorbell)
+			mlx4_en_xmit_doorbell(ring);
+		else
+			ring->xmit_more++;
+	}
+}
+
+static bool mlx4_en_build_dma_wqe(struct mlx4_en_priv *priv,
+				  struct skb_shared_info *shinfo,
+				  struct mlx4_wqe_data_seg *data,
+				  struct sk_buff *skb,
+				  int lso_header_size,
+				  __be32 mr_key,
+				  struct mlx4_en_tx_info *tx_info)
+{
+	struct device *ddev = priv->ddev;
+	dma_addr_t dma = 0;
+	u32 byte_count = 0;
+	int i_frag;
+
+	/* Map fragments if any */
+	for (i_frag = shinfo->nr_frags - 1; i_frag >= 0; i_frag--) {
+		const skb_frag_t *frag = &shinfo->frags[i_frag];
+		byte_count = skb_frag_size(frag);
+		dma = skb_frag_dma_map(ddev, frag,
+				       0, byte_count,
+				       DMA_TO_DEVICE);
+		if (dma_mapping_error(ddev, dma))
+			goto tx_drop_unmap;
+
+		data->addr = cpu_to_be64(dma);
+		data->lkey = mr_key;
+		dma_wmb();
+		data->byte_count = cpu_to_be32(byte_count);
+		--data;
+	}
+
+	/* Map linear part if needed */
+	if (tx_info->linear) {
+		byte_count = skb_headlen(skb) - lso_header_size;
+
+		dma = dma_map_single(ddev, skb->data +
+				     lso_header_size, byte_count,
+				     DMA_TO_DEVICE);
+		if (dma_mapping_error(ddev, dma))
+			goto tx_drop_unmap;
+
+		data->addr = cpu_to_be64(dma);
+		data->lkey = mr_key;
+		dma_wmb();
+		data->byte_count = cpu_to_be32(byte_count);
+	}
+	/* tx completion can avoid cache line miss for common cases */
+	tx_info->map0_dma = dma;
+	tx_info->map0_byte_count = byte_count;
+
+	return true;
+
+tx_drop_unmap:
+	en_err(priv, "DMA mapping error\n");
+
+	while (++i_frag < shinfo->nr_frags) {
+		++data;
+		dma_unmap_page(ddev, (dma_addr_t)be64_to_cpu(data->addr),
+			       be32_to_cpu(data->byte_count),
+			       DMA_TO_DEVICE);
+	}
+
+	return false;
+}
+
+netdev_tx_t mlx4_en_xmit(struct sk_buff *skb, struct net_device *dev)
+{
+	struct skb_shared_info *shinfo = skb_shinfo(skb);
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	union mlx4_wqe_qpn_vlan	qpn_vlan = {};
+	struct mlx4_en_tx_ring *ring;
+	struct mlx4_en_tx_desc *tx_desc;
+	struct mlx4_wqe_data_seg *data;
+	struct mlx4_en_tx_info *tx_info;
+	u32 __maybe_unused ring_cons;
+	int tx_ind;
+	int nr_txbb;
+	int desc_size;
+	int real_size;
+	u32 index, bf_index;
+	struct ipv6hdr *h6;
+	__be32 op_own;
+	int lso_header_size;
+	void *fragptr = NULL;
+	bool bounce = false;
+	bool send_doorbell;
+	bool stop_queue;
+	bool inline_ok;
+	u8 data_offset;
+	int hopbyhop;
+	bool bf_ok;
+
+	tx_ind = skb_get_queue_mapping(skb);
+	ring = priv->tx_ring[TX][tx_ind];
+
+	if (unlikely(!priv->port_up))
+		goto tx_drop;
+
+	real_size = get_real_size(skb, shinfo, dev, &lso_header_size,
+				  &inline_ok, &fragptr, &hopbyhop);
+	if (unlikely(!real_size))
+		goto tx_drop_count;
+
+	/* Align descriptor to TXBB size */
+	desc_size = ALIGN(real_size, TXBB_SIZE);
+	nr_txbb = desc_size >> LOG_TXBB_SIZE;
+	if (unlikely(nr_txbb > MAX_DESC_TXBBS)) {
+		if (netif_msg_tx_err(priv))
+			en_warn(priv, "Oversized header or SG list\n");
+		goto tx_drop_count;
+	}
+
+	bf_ok = ring->bf_enabled;
+	if (skb_vlan_tag_present(skb)) {
+		u16 vlan_proto;
+
+		qpn_vlan.vlan_tag = cpu_to_be16(skb_vlan_tag_get(skb));
+		vlan_proto = be16_to_cpu(skb->vlan_proto);
+		if (vlan_proto == ETH_P_8021AD)
+			qpn_vlan.ins_vlan = MLX4_WQE_CTRL_INS_SVLAN;
+		else if (vlan_proto == ETH_P_8021Q)
+			qpn_vlan.ins_vlan = MLX4_WQE_CTRL_INS_CVLAN;
+		else
+			qpn_vlan.ins_vlan = 0;
+		bf_ok = false;
+	}
+
+	netdev_txq_bql_enqueue_prefetchw(ring->tx_queue);
+
+	/* Packet is good - grab an index and transmit it */
+	index = ring->prod & ring->size_mask;
+	bf_index = ring->prod;
+
+	/* See if we have enough space for whole descriptor TXBB for setting
+	 * SW ownership on next descriptor; if not, use a bounce buffer. */
+	if (likely(index + nr_txbb <= ring->size))
+		tx_desc = ring->buf + (index << LOG_TXBB_SIZE);
+	else {
+		tx_desc = (struct mlx4_en_tx_desc *) ring->bounce_buf;
+		bounce = true;
+		bf_ok = false;
+	}
+
+	/* Save skb in tx_info ring */
+	tx_info = &ring->tx_info[index];
+	tx_info->skb = skb;
+	tx_info->nr_txbb = nr_txbb;
+
+	if (!lso_header_size) {
+		data = &tx_desc->data;
+		data_offset = offsetof(struct mlx4_en_tx_desc, data);
+	} else {
+		int lso_align = ALIGN(lso_header_size - hopbyhop + 4, DS_SIZE);
+
+		data = (void *)&tx_desc->lso + lso_align;
+		data_offset = offsetof(struct mlx4_en_tx_desc, lso) + lso_align;
+	}
+
+	/* valid only for none inline segments */
+	tx_info->data_offset = data_offset;
+
+	tx_info->inl = inline_ok;
+
+	tx_info->linear = lso_header_size < skb_headlen(skb) && !inline_ok;
+
+	tx_info->nr_maps = shinfo->nr_frags + tx_info->linear;
+	data += tx_info->nr_maps - 1;
+
+	if (!tx_info->inl)
+		if (!mlx4_en_build_dma_wqe(priv, shinfo, data, skb,
+					   lso_header_size, ring->mr_key,
+					   tx_info))
+			goto tx_drop_count;
+
+	/*
+	 * For timestamping add flag to skb_shinfo and
+	 * set flag for further reference
+	 */
+	tx_info->ts_requested = 0;
+	if (unlikely(ring->hwtstamp_tx_type == HWTSTAMP_TX_ON &&
+		     shinfo->tx_flags & SKBTX_HW_TSTAMP)) {
+		shinfo->tx_flags |= SKBTX_IN_PROGRESS;
+		tx_info->ts_requested = 1;
+	}
+
+	/* Prepare ctrl segement apart opcode+ownership, which depends on
+	 * whether LSO is used */
+	tx_desc->ctrl.srcrb_flags = priv->ctrl_flags;
+	if (likely(skb->ip_summed == CHECKSUM_PARTIAL)) {
+		if (!skb->encapsulation)
+			tx_desc->ctrl.srcrb_flags |= cpu_to_be32(MLX4_WQE_CTRL_IP_CSUM |
+								 MLX4_WQE_CTRL_TCP_UDP_CSUM);
+		else
+			tx_desc->ctrl.srcrb_flags |= cpu_to_be32(MLX4_WQE_CTRL_IP_CSUM);
+		ring->tx_csum++;
+	}
+
+	if (priv->flags & MLX4_EN_FLAG_ENABLE_HW_LOOPBACK) {
+		struct ethhdr *ethh;
+
+		/* Copy dst mac address to wqe. This allows loopback in eSwitch,
+		 * so that VFs and PF can communicate with each other
+		 */
+		ethh = (struct ethhdr *)skb->data;
+		tx_desc->ctrl.srcrb_flags16[0] = get_unaligned((__be16 *)ethh->h_dest);
+		tx_desc->ctrl.imm = get_unaligned((__be32 *)(ethh->h_dest + 2));
+	}
+
+	/* Handle LSO (TSO) packets */
+	if (lso_header_size) {
+		int i;
+
+		/* Mark opcode as LSO */
+		op_own = cpu_to_be32(MLX4_OPCODE_LSO | (1 << 6)) |
+			((ring->prod & ring->size) ?
+				cpu_to_be32(MLX4_EN_BIT_DESC_OWN) : 0);
+
+		lso_header_size -= hopbyhop;
+		/* Fill in the LSO prefix */
+		tx_desc->lso.mss_hdr_size = cpu_to_be32(
+			shinfo->gso_size << 16 | lso_header_size);
+
+
+		if (unlikely(hopbyhop)) {
+			/* remove the HBH header.
+			 * Layout: [Ethernet header][IPv6 header][HBH][TCP header]
+			 */
+			memcpy(tx_desc->lso.header, skb->data, ETH_HLEN + sizeof(*h6));
+			h6 = (struct ipv6hdr *)((char *)tx_desc->lso.header + ETH_HLEN);
+			h6->nexthdr = IPPROTO_TCP;
+			/* Copy the TCP header after the IPv6 one */
+			memcpy(h6 + 1,
+			       skb->data + ETH_HLEN + sizeof(*h6) +
+					sizeof(struct hop_jumbo_hdr),
+			       tcp_hdrlen(skb));
+			/* Leave ipv6 payload_len set to 0, as LSO v2 specs request. */
+		} else {
+			/* Copy headers;
+			 * note that we already verified that it is linear
+			 */
+			memcpy(tx_desc->lso.header, skb->data, lso_header_size);
+		}
+		ring->tso_packets++;
+
+		i = shinfo->gso_segs;
+		tx_info->nr_bytes = skb->len + (i - 1) * lso_header_size;
+		ring->packets += i;
+	} else {
+		/* Normal (Non LSO) packet */
+		op_own = cpu_to_be32(MLX4_OPCODE_SEND) |
+			((ring->prod & ring->size) ?
+			 cpu_to_be32(MLX4_EN_BIT_DESC_OWN) : 0);
+		tx_info->nr_bytes = max_t(unsigned int, skb->len, ETH_ZLEN);
+		ring->packets++;
+	}
+	ring->bytes += tx_info->nr_bytes;
+
+	if (tx_info->inl)
+		build_inline_wqe(tx_desc, skb, shinfo, fragptr);
+
+	if (skb->encapsulation) {
+		union {
+			struct iphdr *v4;
+			struct ipv6hdr *v6;
+			unsigned char *hdr;
+		} ip;
+		u8 proto;
+
+		ip.hdr = skb_inner_network_header(skb);
+		proto = (ip.v4->version == 4) ? ip.v4->protocol :
+						ip.v6->nexthdr;
+
+		if (proto == IPPROTO_TCP || proto == IPPROTO_UDP)
+			op_own |= cpu_to_be32(MLX4_WQE_CTRL_IIP | MLX4_WQE_CTRL_ILP);
+		else
+			op_own |= cpu_to_be32(MLX4_WQE_CTRL_IIP);
+	}
+
+	ring->prod += nr_txbb;
+
+	/* If we used a bounce buffer then copy descriptor back into place */
+	if (unlikely(bounce))
+		tx_desc = mlx4_en_bounce_to_desc(priv, ring, index, desc_size);
+
+	skb_tx_timestamp(skb);
+
+	/* Check available TXBBs And 2K spare for prefetch */
+	stop_queue = mlx4_en_is_tx_ring_full(ring);
+	if (unlikely(stop_queue)) {
+		netif_tx_stop_queue(ring->tx_queue);
+		ring->queue_stopped++;
+	}
+
+	send_doorbell = __netdev_tx_sent_queue(ring->tx_queue,
+					       tx_info->nr_bytes,
+					       netdev_xmit_more());
+
+	real_size = (real_size / 16) & 0x3f;
+
+	bf_ok &= desc_size <= MAX_BF && send_doorbell;
+
+	if (bf_ok)
+		qpn_vlan.bf_qpn = ring->doorbell_qpn | cpu_to_be32(real_size);
+	else
+		qpn_vlan.fence_size = real_size;
+
+	mlx4_en_tx_write_desc(ring, tx_desc, qpn_vlan, desc_size, bf_index,
+			      op_own, bf_ok, send_doorbell);
+
+	if (unlikely(stop_queue)) {
+		/* If queue was emptied after the if (stop_queue) , and before
+		 * the netif_tx_stop_queue() - need to wake the queue,
+		 * or else it will remain stopped forever.
+		 * Need a memory barrier to make sure ring->cons was not
+		 * updated before queue was stopped.
+		 */
+		smp_rmb();
+
+		if (unlikely(!mlx4_en_is_tx_ring_full(ring))) {
+			netif_tx_wake_queue(ring->tx_queue);
+			ring->wake_queue++;
+		}
+	}
+	return NETDEV_TX_OK;
+
+tx_drop_count:
+	ring->tx_dropped++;
+tx_drop:
+	dev_kfree_skb_any(skb);
+	return NETDEV_TX_OK;
+}
+
+#define MLX4_EN_XDP_TX_NRTXBB  1
+#define MLX4_EN_XDP_TX_REAL_SZ (((CTRL_SIZE + MLX4_EN_XDP_TX_NRTXBB * DS_SIZE) \
+				 / 16) & 0x3f)
+
+void mlx4_en_init_tx_xdp_ring_descs(struct mlx4_en_priv *priv,
+				    struct mlx4_en_tx_ring *ring)
+{
+	int i;
+
+	for (i = 0; i < ring->size; i++) {
+		struct mlx4_en_tx_info *tx_info = &ring->tx_info[i];
+		struct mlx4_en_tx_desc *tx_desc = ring->buf +
+			(i << LOG_TXBB_SIZE);
+
+		tx_info->map0_byte_count = PAGE_SIZE;
+		tx_info->nr_txbb = MLX4_EN_XDP_TX_NRTXBB;
+		tx_info->data_offset = offsetof(struct mlx4_en_tx_desc, data);
+		tx_info->ts_requested = 0;
+		tx_info->nr_maps = 1;
+		tx_info->linear = 1;
+		tx_info->inl = 0;
+
+		tx_desc->data.lkey = ring->mr_key;
+		tx_desc->ctrl.qpn_vlan.fence_size = MLX4_EN_XDP_TX_REAL_SZ;
+		tx_desc->ctrl.srcrb_flags = priv->ctrl_flags;
+	}
+}
+
+netdev_tx_t mlx4_en_xmit_frame(struct mlx4_en_rx_ring *rx_ring,
+			       struct mlx4_en_rx_alloc *frame,
+			       struct mlx4_en_priv *priv, unsigned int length,
+			       int tx_ind, bool *doorbell_pending)
+{
+	struct mlx4_en_tx_desc *tx_desc;
+	struct mlx4_en_tx_info *tx_info;
+	struct mlx4_wqe_data_seg *data;
+	struct mlx4_en_tx_ring *ring;
+	dma_addr_t dma;
+	__be32 op_own;
+	int index;
+
+	if (unlikely(!priv->port_up))
+		goto tx_drop;
+
+	ring = priv->tx_ring[TX_XDP][tx_ind];
+
+	if (unlikely(mlx4_en_is_tx_ring_full(ring)))
+		goto tx_drop_count;
+
+	index = ring->prod & ring->size_mask;
+	tx_info = &ring->tx_info[index];
+
+	tx_desc = ring->buf + (index << LOG_TXBB_SIZE);
+	data = &tx_desc->data;
+
+	dma = frame->dma;
+
+	tx_info->page = frame->page;
+	frame->page = NULL;
+	tx_info->map0_dma = dma;
+	tx_info->nr_bytes = max_t(unsigned int, length, ETH_ZLEN);
+
+	dma_sync_single_range_for_device(priv->ddev, dma, frame->page_offset,
+					 length, DMA_TO_DEVICE);
+
+	data->addr = cpu_to_be64(dma + frame->page_offset);
+	dma_wmb();
+	data->byte_count = cpu_to_be32(length);
+
+	/* tx completion can avoid cache line miss for common cases */
+
+	op_own = cpu_to_be32(MLX4_OPCODE_SEND) |
+		((ring->prod & ring->size) ?
+		 cpu_to_be32(MLX4_EN_BIT_DESC_OWN) : 0);
+
+	rx_ring->xdp_tx++;
+
+	ring->prod += MLX4_EN_XDP_TX_NRTXBB;
+
+	/* Ensure new descriptor hits memory
+	 * before setting ownership of this descriptor to HW
+	 */
+	dma_wmb();
+	tx_desc->ctrl.owner_opcode = op_own;
+	ring->xmit_more++;
+
+	*doorbell_pending = true;
+
+	return NETDEV_TX_OK;
+
+tx_drop_count:
+	rx_ring->xdp_tx_full++;
+	*doorbell_pending = true;
+tx_drop:
+	return NETDEV_TX_BUSY;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx4/eq.c b/drivers/net/ethernet/mellanox/mlx4/eq.c
new file mode 100644
index 000000000..414e390e6
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx4/eq.c
@@ -0,0 +1,1563 @@
+/*
+ * Copyright (c) 2005, 2006, 2007, 2008 Mellanox Technologies. All rights reserved.
+ * Copyright (c) 2005, 2006, 2007 Cisco Systems, Inc. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *	- Redistributions of source code must retain the above
+ *	  copyright notice, this list of conditions and the following
+ *	  disclaimer.
+ *
+ *	- Redistributions in binary form must reproduce the above
+ *	  copyright notice, this list of conditions and the following
+ *	  disclaimer in the documentation and/or other materials
+ *	  provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/interrupt.h>
+#include <linux/slab.h>
+#include <linux/export.h>
+#include <linux/mm.h>
+#include <linux/dma-mapping.h>
+
+#include <linux/mlx4/cmd.h>
+#include <linux/cpu_rmap.h>
+
+#include "mlx4.h"
+#include "fw.h"
+
+enum {
+	MLX4_IRQNAME_SIZE	= 32
+};
+
+enum {
+	MLX4_NUM_ASYNC_EQE	= 0x100,
+	MLX4_NUM_SPARE_EQE	= 0x80,
+	MLX4_EQ_ENTRY_SIZE	= 0x20
+};
+
+#define MLX4_EQ_STATUS_OK	   ( 0 << 28)
+#define MLX4_EQ_STATUS_WRITE_FAIL  (10 << 28)
+#define MLX4_EQ_OWNER_SW	   ( 0 << 24)
+#define MLX4_EQ_OWNER_HW	   ( 1 << 24)
+#define MLX4_EQ_FLAG_EC		   ( 1 << 18)
+#define MLX4_EQ_FLAG_OI		   ( 1 << 17)
+#define MLX4_EQ_STATE_ARMED	   ( 9 <<  8)
+#define MLX4_EQ_STATE_FIRED	   (10 <<  8)
+#define MLX4_EQ_STATE_ALWAYS_ARMED (11 <<  8)
+
+#define MLX4_ASYNC_EVENT_MASK ((1ull << MLX4_EVENT_TYPE_PATH_MIG)	    | \
+			       (1ull << MLX4_EVENT_TYPE_COMM_EST)	    | \
+			       (1ull << MLX4_EVENT_TYPE_SQ_DRAINED)	    | \
+			       (1ull << MLX4_EVENT_TYPE_CQ_ERROR)	    | \
+			       (1ull << MLX4_EVENT_TYPE_WQ_CATAS_ERROR)	    | \
+			       (1ull << MLX4_EVENT_TYPE_EEC_CATAS_ERROR)    | \
+			       (1ull << MLX4_EVENT_TYPE_PATH_MIG_FAILED)    | \
+			       (1ull << MLX4_EVENT_TYPE_WQ_INVAL_REQ_ERROR) | \
+			       (1ull << MLX4_EVENT_TYPE_WQ_ACCESS_ERROR)    | \
+			       (1ull << MLX4_EVENT_TYPE_PORT_CHANGE)	    | \
+			       (1ull << MLX4_EVENT_TYPE_ECC_DETECT)	    | \
+			       (1ull << MLX4_EVENT_TYPE_SRQ_CATAS_ERROR)    | \
+			       (1ull << MLX4_EVENT_TYPE_SRQ_QP_LAST_WQE)    | \
+			       (1ull << MLX4_EVENT_TYPE_SRQ_LIMIT)	    | \
+			       (1ull << MLX4_EVENT_TYPE_CMD)		    | \
+			       (1ull << MLX4_EVENT_TYPE_OP_REQUIRED)	    | \
+			       (1ull << MLX4_EVENT_TYPE_COMM_CHANNEL)       | \
+			       (1ull << MLX4_EVENT_TYPE_FLR_EVENT)	    | \
+			       (1ull << MLX4_EVENT_TYPE_FATAL_WARNING))
+
+static u64 get_async_ev_mask(struct mlx4_dev *dev)
+{
+	u64 async_ev_mask = MLX4_ASYNC_EVENT_MASK;
+	if (dev->caps.flags & MLX4_DEV_CAP_FLAG_PORT_MNG_CHG_EV)
+		async_ev_mask |= (1ull << MLX4_EVENT_TYPE_PORT_MNG_CHG_EVENT);
+	if (dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_RECOVERABLE_ERROR_EVENT)
+		async_ev_mask |= (1ull << MLX4_EVENT_TYPE_RECOVERABLE_ERROR_EVENT);
+
+	return async_ev_mask;
+}
+
+static void eq_set_ci(struct mlx4_eq *eq, int req_not)
+{
+	__raw_writel((__force u32) cpu_to_be32((eq->cons_index & 0xffffff) |
+					       req_not << 31),
+		     eq->doorbell);
+	/* We still want ordering, just not swabbing, so add a barrier */
+	wmb();
+}
+
+static struct mlx4_eqe *get_eqe(struct mlx4_eq *eq, u32 entry, u8 eqe_factor,
+				u8 eqe_size)
+{
+	/* (entry & (eq->nent - 1)) gives us a cyclic array */
+	unsigned long offset = (entry & (eq->nent - 1)) * eqe_size;
+	/* CX3 is capable of extending the EQE from 32 to 64 bytes with
+	 * strides of 64B,128B and 256B.
+	 * When 64B EQE is used, the first (in the lower addresses)
+	 * 32 bytes in the 64 byte EQE are reserved and the next 32 bytes
+	 * contain the legacy EQE information.
+	 * In all other cases, the first 32B contains the legacy EQE info.
+	 */
+	return eq->page_list[offset / PAGE_SIZE].buf + (offset + (eqe_factor ? MLX4_EQ_ENTRY_SIZE : 0)) % PAGE_SIZE;
+}
+
+static struct mlx4_eqe *next_eqe_sw(struct mlx4_eq *eq, u8 eqe_factor, u8 size)
+{
+	struct mlx4_eqe *eqe = get_eqe(eq, eq->cons_index, eqe_factor, size);
+	return !!(eqe->owner & 0x80) ^ !!(eq->cons_index & eq->nent) ? NULL : eqe;
+}
+
+static struct mlx4_eqe *next_slave_event_eqe(struct mlx4_slave_event_eq *slave_eq)
+{
+	struct mlx4_eqe *eqe =
+		&slave_eq->event_eqe[slave_eq->cons & (SLAVE_EVENT_EQ_SIZE - 1)];
+	return (!!(eqe->owner & 0x80) ^
+		!!(slave_eq->cons & SLAVE_EVENT_EQ_SIZE)) ?
+		eqe : NULL;
+}
+
+void mlx4_gen_slave_eqe(struct work_struct *work)
+{
+	struct mlx4_mfunc_master_ctx *master =
+		container_of(work, struct mlx4_mfunc_master_ctx,
+			     slave_event_work);
+	struct mlx4_mfunc *mfunc =
+		container_of(master, struct mlx4_mfunc, master);
+	struct mlx4_priv *priv = container_of(mfunc, struct mlx4_priv, mfunc);
+	struct mlx4_dev *dev = &priv->dev;
+	struct mlx4_slave_event_eq *slave_eq = &mfunc->master.slave_eq;
+	struct mlx4_eqe *eqe;
+	u8 slave;
+	int i, phys_port, slave_port;
+
+	for (eqe = next_slave_event_eqe(slave_eq); eqe;
+	      eqe = next_slave_event_eqe(slave_eq)) {
+		slave = eqe->slave_id;
+
+		if (eqe->type == MLX4_EVENT_TYPE_PORT_CHANGE &&
+		    eqe->subtype == MLX4_PORT_CHANGE_SUBTYPE_DOWN &&
+		    mlx4_is_bonded(dev)) {
+			struct mlx4_port_cap port_cap;
+
+			if (!mlx4_QUERY_PORT(dev, 1, &port_cap) && port_cap.link_state)
+				goto consume;
+
+			if (!mlx4_QUERY_PORT(dev, 2, &port_cap) && port_cap.link_state)
+				goto consume;
+		}
+		/* All active slaves need to receive the event */
+		if (slave == ALL_SLAVES) {
+			for (i = 0; i <= dev->persist->num_vfs; i++) {
+				phys_port = 0;
+				if (eqe->type == MLX4_EVENT_TYPE_PORT_MNG_CHG_EVENT &&
+				    eqe->subtype == MLX4_DEV_PMC_SUBTYPE_PORT_INFO) {
+					phys_port  = eqe->event.port_mgmt_change.port;
+					slave_port = mlx4_phys_to_slave_port(dev, i, phys_port);
+					if (slave_port < 0) /* VF doesn't have this port */
+						continue;
+					eqe->event.port_mgmt_change.port = slave_port;
+				}
+				if (mlx4_GEN_EQE(dev, i, eqe))
+					mlx4_warn(dev, "Failed to generate event for slave %d\n",
+						  i);
+				if (phys_port)
+					eqe->event.port_mgmt_change.port = phys_port;
+			}
+		} else {
+			if (mlx4_GEN_EQE(dev, slave, eqe))
+				mlx4_warn(dev, "Failed to generate event for slave %d\n",
+					  slave);
+		}
+consume:
+		++slave_eq->cons;
+	}
+}
+
+
+static void slave_event(struct mlx4_dev *dev, u8 slave, struct mlx4_eqe *eqe)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_slave_event_eq *slave_eq = &priv->mfunc.master.slave_eq;
+	struct mlx4_eqe *s_eqe;
+	unsigned long flags;
+
+	spin_lock_irqsave(&slave_eq->event_lock, flags);
+	s_eqe = &slave_eq->event_eqe[slave_eq->prod & (SLAVE_EVENT_EQ_SIZE - 1)];
+	if ((!!(s_eqe->owner & 0x80)) ^
+	    (!!(slave_eq->prod & SLAVE_EVENT_EQ_SIZE))) {
+		mlx4_warn(dev, "Master failed to generate an EQE for slave: %d. No free EQE on slave events queue\n",
+			  slave);
+		spin_unlock_irqrestore(&slave_eq->event_lock, flags);
+		return;
+	}
+
+	memcpy(s_eqe, eqe, sizeof(struct mlx4_eqe) - 1);
+	s_eqe->slave_id = slave;
+	/* ensure all information is written before setting the ownersip bit */
+	dma_wmb();
+	s_eqe->owner = !!(slave_eq->prod & SLAVE_EVENT_EQ_SIZE) ? 0x0 : 0x80;
+	++slave_eq->prod;
+
+	queue_work(priv->mfunc.master.comm_wq,
+		   &priv->mfunc.master.slave_event_work);
+	spin_unlock_irqrestore(&slave_eq->event_lock, flags);
+}
+
+static void mlx4_slave_event(struct mlx4_dev *dev, int slave,
+			     struct mlx4_eqe *eqe)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	if (slave < 0 || slave > dev->persist->num_vfs ||
+	    slave == dev->caps.function ||
+	    !priv->mfunc.master.slave_state[slave].active)
+		return;
+
+	slave_event(dev, slave, eqe);
+}
+
+#if defined(CONFIG_SMP)
+static void mlx4_set_eq_affinity_hint(struct mlx4_priv *priv, int vec)
+{
+	int hint_err;
+	struct mlx4_dev *dev = &priv->dev;
+	struct mlx4_eq *eq = &priv->eq_table.eq[vec];
+
+	if (!cpumask_available(eq->affinity_mask) ||
+	    cpumask_empty(eq->affinity_mask))
+		return;
+
+	hint_err = irq_update_affinity_hint(eq->irq, eq->affinity_mask);
+	if (hint_err)
+		mlx4_warn(dev, "irq_update_affinity_hint failed, err %d\n", hint_err);
+}
+#endif
+
+int mlx4_gen_pkey_eqe(struct mlx4_dev *dev, int slave, u8 port)
+{
+	struct mlx4_eqe eqe;
+
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_slave_state *s_slave = &priv->mfunc.master.slave_state[slave];
+
+	if (!s_slave->active)
+		return 0;
+
+	memset(&eqe, 0, sizeof(eqe));
+
+	eqe.type = MLX4_EVENT_TYPE_PORT_MNG_CHG_EVENT;
+	eqe.subtype = MLX4_DEV_PMC_SUBTYPE_PKEY_TABLE;
+	eqe.event.port_mgmt_change.port = mlx4_phys_to_slave_port(dev, slave, port);
+
+	return mlx4_GEN_EQE(dev, slave, &eqe);
+}
+EXPORT_SYMBOL(mlx4_gen_pkey_eqe);
+
+int mlx4_gen_guid_change_eqe(struct mlx4_dev *dev, int slave, u8 port)
+{
+	struct mlx4_eqe eqe;
+
+	/*don't send if we don't have the that slave */
+	if (dev->persist->num_vfs < slave)
+		return 0;
+	memset(&eqe, 0, sizeof(eqe));
+
+	eqe.type = MLX4_EVENT_TYPE_PORT_MNG_CHG_EVENT;
+	eqe.subtype = MLX4_DEV_PMC_SUBTYPE_GUID_INFO;
+	eqe.event.port_mgmt_change.port = mlx4_phys_to_slave_port(dev, slave, port);
+
+	return mlx4_GEN_EQE(dev, slave, &eqe);
+}
+EXPORT_SYMBOL(mlx4_gen_guid_change_eqe);
+
+int mlx4_gen_port_state_change_eqe(struct mlx4_dev *dev, int slave, u8 port,
+				   u8 port_subtype_change)
+{
+	struct mlx4_eqe eqe;
+	u8 slave_port = mlx4_phys_to_slave_port(dev, slave, port);
+
+	/*don't send if we don't have the that slave */
+	if (dev->persist->num_vfs < slave)
+		return 0;
+	memset(&eqe, 0, sizeof(eqe));
+
+	eqe.type = MLX4_EVENT_TYPE_PORT_CHANGE;
+	eqe.subtype = port_subtype_change;
+	eqe.event.port_change.port = cpu_to_be32(slave_port << 28);
+
+	mlx4_dbg(dev, "%s: sending: %d to slave: %d on port: %d\n", __func__,
+		 port_subtype_change, slave, port);
+	return mlx4_GEN_EQE(dev, slave, &eqe);
+}
+EXPORT_SYMBOL(mlx4_gen_port_state_change_eqe);
+
+enum slave_port_state mlx4_get_slave_port_state(struct mlx4_dev *dev, int slave, u8 port)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_slave_state *s_state = priv->mfunc.master.slave_state;
+	struct mlx4_active_ports actv_ports = mlx4_get_active_ports(dev, slave);
+
+	if (slave >= dev->num_slaves || port > dev->caps.num_ports ||
+	    port <= 0 || !test_bit(port - 1, actv_ports.ports)) {
+		pr_err("%s: Error: asking for slave:%d, port:%d\n",
+		       __func__, slave, port);
+		return SLAVE_PORT_DOWN;
+	}
+	return s_state[slave].port_state[port];
+}
+EXPORT_SYMBOL(mlx4_get_slave_port_state);
+
+static int mlx4_set_slave_port_state(struct mlx4_dev *dev, int slave, u8 port,
+				     enum slave_port_state state)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_slave_state *s_state = priv->mfunc.master.slave_state;
+	struct mlx4_active_ports actv_ports = mlx4_get_active_ports(dev, slave);
+
+	if (slave >= dev->num_slaves || port > dev->caps.num_ports ||
+	    port <= 0 || !test_bit(port - 1, actv_ports.ports)) {
+		pr_err("%s: Error: asking for slave:%d, port:%d\n",
+		       __func__, slave, port);
+		return -1;
+	}
+	s_state[slave].port_state[port] = state;
+
+	return 0;
+}
+
+static void set_all_slave_state(struct mlx4_dev *dev, u8 port, int event)
+{
+	int i;
+	enum slave_port_gen_event gen_event;
+	struct mlx4_slaves_pport slaves_pport = mlx4_phys_to_slaves_pport(dev,
+									  port);
+
+	for (i = 0; i < dev->persist->num_vfs + 1; i++)
+		if (test_bit(i, slaves_pport.slaves))
+			set_and_calc_slave_port_state(dev, i, port,
+						      event, &gen_event);
+}
+/**************************************************************************
+	The function get as input the new event to that port,
+	and according to the prev state change the slave's port state.
+	The events are:
+		MLX4_PORT_STATE_DEV_EVENT_PORT_DOWN,
+		MLX4_PORT_STATE_DEV_EVENT_PORT_UP
+		MLX4_PORT_STATE_IB_EVENT_GID_VALID
+		MLX4_PORT_STATE_IB_EVENT_GID_INVALID
+***************************************************************************/
+int set_and_calc_slave_port_state(struct mlx4_dev *dev, int slave,
+				  u8 port, int event,
+				  enum slave_port_gen_event *gen_event)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_slave_state *ctx = NULL;
+	unsigned long flags;
+	int ret = -1;
+	struct mlx4_active_ports actv_ports = mlx4_get_active_ports(dev, slave);
+	enum slave_port_state cur_state =
+		mlx4_get_slave_port_state(dev, slave, port);
+
+	*gen_event = SLAVE_PORT_GEN_EVENT_NONE;
+
+	if (slave >= dev->num_slaves || port > dev->caps.num_ports ||
+	    port <= 0 || !test_bit(port - 1, actv_ports.ports)) {
+		pr_err("%s: Error: asking for slave:%d, port:%d\n",
+		       __func__, slave, port);
+		return ret;
+	}
+
+	ctx = &priv->mfunc.master.slave_state[slave];
+	spin_lock_irqsave(&ctx->lock, flags);
+
+	switch (cur_state) {
+	case SLAVE_PORT_DOWN:
+		if (MLX4_PORT_STATE_DEV_EVENT_PORT_UP == event)
+			mlx4_set_slave_port_state(dev, slave, port,
+						  SLAVE_PENDING_UP);
+		break;
+	case SLAVE_PENDING_UP:
+		if (MLX4_PORT_STATE_DEV_EVENT_PORT_DOWN == event)
+			mlx4_set_slave_port_state(dev, slave, port,
+						  SLAVE_PORT_DOWN);
+		else if (MLX4_PORT_STATE_IB_PORT_STATE_EVENT_GID_VALID == event) {
+			mlx4_set_slave_port_state(dev, slave, port,
+						  SLAVE_PORT_UP);
+			*gen_event = SLAVE_PORT_GEN_EVENT_UP;
+		}
+		break;
+	case SLAVE_PORT_UP:
+		if (MLX4_PORT_STATE_DEV_EVENT_PORT_DOWN == event) {
+			mlx4_set_slave_port_state(dev, slave, port,
+						  SLAVE_PORT_DOWN);
+			*gen_event = SLAVE_PORT_GEN_EVENT_DOWN;
+		} else if (MLX4_PORT_STATE_IB_EVENT_GID_INVALID ==
+				event) {
+			mlx4_set_slave_port_state(dev, slave, port,
+						  SLAVE_PENDING_UP);
+			*gen_event = SLAVE_PORT_GEN_EVENT_DOWN;
+		}
+		break;
+	default:
+		pr_err("%s: BUG!!! UNKNOWN state: slave:%d, port:%d\n",
+		       __func__, slave, port);
+		goto out;
+	}
+	ret = mlx4_get_slave_port_state(dev, slave, port);
+
+out:
+	spin_unlock_irqrestore(&ctx->lock, flags);
+	return ret;
+}
+
+EXPORT_SYMBOL(set_and_calc_slave_port_state);
+
+int mlx4_gen_slaves_port_mgt_ev(struct mlx4_dev *dev, u8 port, int attr)
+{
+	struct mlx4_eqe eqe;
+
+	memset(&eqe, 0, sizeof(eqe));
+
+	eqe.type = MLX4_EVENT_TYPE_PORT_MNG_CHG_EVENT;
+	eqe.subtype = MLX4_DEV_PMC_SUBTYPE_PORT_INFO;
+	eqe.event.port_mgmt_change.port = port;
+	eqe.event.port_mgmt_change.params.port_info.changed_attr =
+		cpu_to_be32((u32) attr);
+
+	slave_event(dev, ALL_SLAVES, &eqe);
+	return 0;
+}
+EXPORT_SYMBOL(mlx4_gen_slaves_port_mgt_ev);
+
+void mlx4_master_handle_slave_flr(struct work_struct *work)
+{
+	struct mlx4_mfunc_master_ctx *master =
+		container_of(work, struct mlx4_mfunc_master_ctx,
+			     slave_flr_event_work);
+	struct mlx4_mfunc *mfunc =
+		container_of(master, struct mlx4_mfunc, master);
+	struct mlx4_priv *priv =
+		container_of(mfunc, struct mlx4_priv, mfunc);
+	struct mlx4_dev *dev = &priv->dev;
+	struct mlx4_slave_state *slave_state = priv->mfunc.master.slave_state;
+	int i;
+	int err;
+	unsigned long flags;
+
+	mlx4_dbg(dev, "mlx4_handle_slave_flr\n");
+
+	for (i = 0 ; i < dev->num_slaves; i++) {
+
+		if (MLX4_COMM_CMD_FLR == slave_state[i].last_cmd) {
+			mlx4_dbg(dev, "mlx4_handle_slave_flr: clean slave: %d\n",
+				 i);
+			/* In case of 'Reset flow' FLR can be generated for
+			 * a slave before mlx4_load_one is done.
+			 * make sure interface is up before trying to delete
+			 * slave resources which weren't allocated yet.
+			 */
+			if (dev->persist->interface_state &
+			    MLX4_INTERFACE_STATE_UP)
+				mlx4_delete_all_resources_for_slave(dev, i);
+			/*return the slave to running mode*/
+			spin_lock_irqsave(&priv->mfunc.master.slave_state_lock, flags);
+			slave_state[i].last_cmd = MLX4_COMM_CMD_RESET;
+			slave_state[i].is_slave_going_down = 0;
+			spin_unlock_irqrestore(&priv->mfunc.master.slave_state_lock, flags);
+			/*notify the FW:*/
+			err = mlx4_cmd(dev, 0, i, 0, MLX4_CMD_INFORM_FLR_DONE,
+				       MLX4_CMD_TIME_CLASS_A, MLX4_CMD_WRAPPED);
+			if (err)
+				mlx4_warn(dev, "Failed to notify FW on FLR done (slave:%d)\n",
+					  i);
+		}
+	}
+}
+
+static int mlx4_eq_int(struct mlx4_dev *dev, struct mlx4_eq *eq)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_eqe *eqe;
+	int cqn;
+	int eqes_found = 0;
+	int set_ci = 0;
+	int port;
+	int slave = 0;
+	int ret;
+	u32 flr_slave;
+	u8 update_slave_state;
+	int i;
+	enum slave_port_gen_event gen_event;
+	unsigned long flags;
+	struct mlx4_vport_state *s_info;
+	int eqe_size = dev->caps.eqe_size;
+
+	while ((eqe = next_eqe_sw(eq, dev->caps.eqe_factor, eqe_size))) {
+		/*
+		 * Make sure we read EQ entry contents after we've
+		 * checked the ownership bit.
+		 */
+		dma_rmb();
+
+		switch (eqe->type) {
+		case MLX4_EVENT_TYPE_COMP:
+			cqn = be32_to_cpu(eqe->event.comp.cqn) & 0xffffff;
+			mlx4_cq_completion(dev, cqn);
+			break;
+
+		case MLX4_EVENT_TYPE_PATH_MIG:
+		case MLX4_EVENT_TYPE_COMM_EST:
+		case MLX4_EVENT_TYPE_SQ_DRAINED:
+		case MLX4_EVENT_TYPE_SRQ_QP_LAST_WQE:
+		case MLX4_EVENT_TYPE_WQ_CATAS_ERROR:
+		case MLX4_EVENT_TYPE_PATH_MIG_FAILED:
+		case MLX4_EVENT_TYPE_WQ_INVAL_REQ_ERROR:
+		case MLX4_EVENT_TYPE_WQ_ACCESS_ERROR:
+			mlx4_dbg(dev, "event %d arrived\n", eqe->type);
+			if (mlx4_is_master(dev)) {
+				/* forward only to slave owning the QP */
+				ret = mlx4_get_slave_from_resource_id(dev,
+						RES_QP,
+						be32_to_cpu(eqe->event.qp.qpn)
+						& 0xffffff, &slave);
+				if (ret && ret != -ENOENT) {
+					mlx4_dbg(dev, "QP event %02x(%02x) on EQ %d at index %u: could not get slave id (%d)\n",
+						 eqe->type, eqe->subtype,
+						 eq->eqn, eq->cons_index, ret);
+					break;
+				}
+
+				if (!ret && slave != dev->caps.function) {
+					mlx4_slave_event(dev, slave, eqe);
+					break;
+				}
+
+			}
+			mlx4_qp_event(dev, be32_to_cpu(eqe->event.qp.qpn) &
+				      0xffffff, eqe->type);
+			break;
+
+		case MLX4_EVENT_TYPE_SRQ_LIMIT:
+			mlx4_dbg(dev, "%s: MLX4_EVENT_TYPE_SRQ_LIMIT. srq_no=0x%x, eq 0x%x\n",
+				 __func__, be32_to_cpu(eqe->event.srq.srqn),
+				 eq->eqn);
+			fallthrough;
+		case MLX4_EVENT_TYPE_SRQ_CATAS_ERROR:
+			if (mlx4_is_master(dev)) {
+				/* forward only to slave owning the SRQ */
+				ret = mlx4_get_slave_from_resource_id(dev,
+						RES_SRQ,
+						be32_to_cpu(eqe->event.srq.srqn)
+						& 0xffffff,
+						&slave);
+				if (ret && ret != -ENOENT) {
+					mlx4_warn(dev, "SRQ event %02x(%02x) on EQ %d at index %u: could not get slave id (%d)\n",
+						  eqe->type, eqe->subtype,
+						  eq->eqn, eq->cons_index, ret);
+					break;
+				}
+				if (eqe->type ==
+				    MLX4_EVENT_TYPE_SRQ_CATAS_ERROR)
+					mlx4_warn(dev, "%s: slave:%d, srq_no:0x%x, event: %02x(%02x)\n",
+						  __func__, slave,
+						  be32_to_cpu(eqe->event.srq.srqn),
+						  eqe->type, eqe->subtype);
+
+				if (!ret && slave != dev->caps.function) {
+					if (eqe->type ==
+					    MLX4_EVENT_TYPE_SRQ_CATAS_ERROR)
+						mlx4_warn(dev, "%s: sending event %02x(%02x) to slave:%d\n",
+							  __func__, eqe->type,
+							  eqe->subtype, slave);
+					mlx4_slave_event(dev, slave, eqe);
+					break;
+				}
+			}
+			mlx4_srq_event(dev, be32_to_cpu(eqe->event.srq.srqn) &
+				       0xffffff, eqe->type);
+			break;
+
+		case MLX4_EVENT_TYPE_CMD:
+			mlx4_cmd_event(dev,
+				       be16_to_cpu(eqe->event.cmd.token),
+				       eqe->event.cmd.status,
+				       be64_to_cpu(eqe->event.cmd.out_param));
+			break;
+
+		case MLX4_EVENT_TYPE_PORT_CHANGE: {
+			struct mlx4_slaves_pport slaves_port;
+			port = be32_to_cpu(eqe->event.port_change.port) >> 28;
+			slaves_port = mlx4_phys_to_slaves_pport(dev, port);
+			if (eqe->subtype == MLX4_PORT_CHANGE_SUBTYPE_DOWN) {
+				mlx4_dispatch_event(dev, MLX4_DEV_EVENT_PORT_DOWN,
+						    port);
+				mlx4_priv(dev)->sense.do_sense_port[port] = 1;
+				if (!mlx4_is_master(dev))
+					break;
+				for (i = 0; i < dev->persist->num_vfs + 1;
+				     i++) {
+					int reported_port = mlx4_is_bonded(dev) ? 1 : mlx4_phys_to_slave_port(dev, i, port);
+
+					if (!test_bit(i, slaves_port.slaves) && !mlx4_is_bonded(dev))
+						continue;
+					if (dev->caps.port_type[port] == MLX4_PORT_TYPE_ETH) {
+						if (i == mlx4_master_func_num(dev))
+							continue;
+						mlx4_dbg(dev, "%s: Sending MLX4_PORT_CHANGE_SUBTYPE_DOWN to slave: %d, port:%d\n",
+							 __func__, i, port);
+						s_info = &priv->mfunc.master.vf_oper[i].vport[port].state;
+						if (IFLA_VF_LINK_STATE_AUTO == s_info->link_state) {
+							eqe->event.port_change.port =
+								cpu_to_be32(
+								(be32_to_cpu(eqe->event.port_change.port) & 0xFFFFFFF)
+								| (reported_port << 28));
+							mlx4_slave_event(dev, i, eqe);
+						}
+					} else {  /* IB port */
+						set_and_calc_slave_port_state(dev, i, port,
+									      MLX4_PORT_STATE_DEV_EVENT_PORT_DOWN,
+									      &gen_event);
+						/*we can be in pending state, then do not send port_down event*/
+						if (SLAVE_PORT_GEN_EVENT_DOWN ==  gen_event) {
+							if (i == mlx4_master_func_num(dev))
+								continue;
+							eqe->event.port_change.port =
+								cpu_to_be32(
+								(be32_to_cpu(eqe->event.port_change.port) & 0xFFFFFFF)
+								| (mlx4_phys_to_slave_port(dev, i, port) << 28));
+							mlx4_slave_event(dev, i, eqe);
+						}
+					}
+				}
+			} else {
+				mlx4_dispatch_event(dev, MLX4_DEV_EVENT_PORT_UP, port);
+
+				mlx4_priv(dev)->sense.do_sense_port[port] = 0;
+
+				if (!mlx4_is_master(dev))
+					break;
+				if (dev->caps.port_type[port] == MLX4_PORT_TYPE_ETH)
+					for (i = 0;
+					     i < dev->persist->num_vfs + 1;
+					     i++) {
+						int reported_port = mlx4_is_bonded(dev) ? 1 : mlx4_phys_to_slave_port(dev, i, port);
+
+						if (!test_bit(i, slaves_port.slaves) && !mlx4_is_bonded(dev))
+							continue;
+						if (i == mlx4_master_func_num(dev))
+							continue;
+						s_info = &priv->mfunc.master.vf_oper[i].vport[port].state;
+						if (IFLA_VF_LINK_STATE_AUTO == s_info->link_state) {
+							eqe->event.port_change.port =
+								cpu_to_be32(
+								(be32_to_cpu(eqe->event.port_change.port) & 0xFFFFFFF)
+								| (reported_port << 28));
+							mlx4_slave_event(dev, i, eqe);
+						}
+					}
+				else /* IB port */
+					/* port-up event will be sent to a slave when the
+					 * slave's alias-guid is set. This is done in alias_GUID.c
+					 */
+					set_all_slave_state(dev, port, MLX4_DEV_EVENT_PORT_UP);
+			}
+			break;
+		}
+
+		case MLX4_EVENT_TYPE_CQ_ERROR:
+			mlx4_warn(dev, "CQ %s on CQN %06x\n",
+				  eqe->event.cq_err.syndrome == 1 ?
+				  "overrun" : "access violation",
+				  be32_to_cpu(eqe->event.cq_err.cqn) & 0xffffff);
+			if (mlx4_is_master(dev)) {
+				ret = mlx4_get_slave_from_resource_id(dev,
+					RES_CQ,
+					be32_to_cpu(eqe->event.cq_err.cqn)
+					& 0xffffff, &slave);
+				if (ret && ret != -ENOENT) {
+					mlx4_dbg(dev, "CQ event %02x(%02x) on EQ %d at index %u: could not get slave id (%d)\n",
+						 eqe->type, eqe->subtype,
+						 eq->eqn, eq->cons_index, ret);
+					break;
+				}
+
+				if (!ret && slave != dev->caps.function) {
+					mlx4_slave_event(dev, slave, eqe);
+					break;
+				}
+			}
+			mlx4_cq_event(dev,
+				      be32_to_cpu(eqe->event.cq_err.cqn)
+				      & 0xffffff,
+				      eqe->type);
+			break;
+
+		case MLX4_EVENT_TYPE_EQ_OVERFLOW:
+			mlx4_warn(dev, "EQ overrun on EQN %d\n", eq->eqn);
+			break;
+
+		case MLX4_EVENT_TYPE_OP_REQUIRED:
+			atomic_inc(&priv->opreq_count);
+			/* FW commands can't be executed from interrupt context
+			 * working in deferred task
+			 */
+			queue_work(mlx4_wq, &priv->opreq_task);
+			break;
+
+		case MLX4_EVENT_TYPE_COMM_CHANNEL:
+			if (!mlx4_is_master(dev)) {
+				mlx4_warn(dev, "Received comm channel event for non master device\n");
+				break;
+			}
+			memcpy(&priv->mfunc.master.comm_arm_bit_vector,
+			       eqe->event.comm_channel_arm.bit_vec,
+			       sizeof(eqe->event.comm_channel_arm.bit_vec));
+			queue_work(priv->mfunc.master.comm_wq,
+				   &priv->mfunc.master.comm_work);
+			break;
+
+		case MLX4_EVENT_TYPE_FLR_EVENT:
+			flr_slave = be32_to_cpu(eqe->event.flr_event.slave_id);
+			if (!mlx4_is_master(dev)) {
+				mlx4_warn(dev, "Non-master function received FLR event\n");
+				break;
+			}
+
+			mlx4_dbg(dev, "FLR event for slave: %d\n", flr_slave);
+
+			if (flr_slave >= dev->num_slaves) {
+				mlx4_warn(dev,
+					  "Got FLR for unknown function: %d\n",
+					  flr_slave);
+				update_slave_state = 0;
+			} else
+				update_slave_state = 1;
+
+			spin_lock_irqsave(&priv->mfunc.master.slave_state_lock, flags);
+			if (update_slave_state) {
+				priv->mfunc.master.slave_state[flr_slave].active = false;
+				priv->mfunc.master.slave_state[flr_slave].last_cmd = MLX4_COMM_CMD_FLR;
+				priv->mfunc.master.slave_state[flr_slave].is_slave_going_down = 1;
+			}
+			spin_unlock_irqrestore(&priv->mfunc.master.slave_state_lock, flags);
+			mlx4_dispatch_event(dev, MLX4_DEV_EVENT_SLAVE_SHUTDOWN,
+					    flr_slave);
+			queue_work(priv->mfunc.master.comm_wq,
+				   &priv->mfunc.master.slave_flr_event_work);
+			break;
+
+		case MLX4_EVENT_TYPE_FATAL_WARNING:
+			if (eqe->subtype == MLX4_FATAL_WARNING_SUBTYPE_WARMING) {
+				if (mlx4_is_master(dev))
+					for (i = 0; i < dev->num_slaves; i++) {
+						mlx4_dbg(dev, "%s: Sending MLX4_FATAL_WARNING_SUBTYPE_WARMING to slave: %d\n",
+							 __func__, i);
+						if (i == dev->caps.function)
+							continue;
+						mlx4_slave_event(dev, i, eqe);
+					}
+				mlx4_err(dev, "Temperature Threshold was reached! Threshold: %d celsius degrees; Current Temperature: %d\n",
+					 be16_to_cpu(eqe->event.warming.warning_threshold),
+					 be16_to_cpu(eqe->event.warming.current_temperature));
+			} else
+				mlx4_warn(dev, "Unhandled event FATAL WARNING (%02x), subtype %02x on EQ %d at index %u. owner=%x, nent=0x%x, slave=%x, ownership=%s\n",
+					  eqe->type, eqe->subtype, eq->eqn,
+					  eq->cons_index, eqe->owner, eq->nent,
+					  eqe->slave_id,
+					  !!(eqe->owner & 0x80) ^
+					  !!(eq->cons_index & eq->nent) ? "HW" : "SW");
+
+			break;
+
+		case MLX4_EVENT_TYPE_PORT_MNG_CHG_EVENT:
+			mlx4_dispatch_event(dev, MLX4_DEV_EVENT_PORT_MGMT_CHANGE,
+					    (unsigned long) eqe);
+			break;
+
+		case MLX4_EVENT_TYPE_RECOVERABLE_ERROR_EVENT:
+			switch (eqe->subtype) {
+			case MLX4_RECOVERABLE_ERROR_EVENT_SUBTYPE_BAD_CABLE:
+				mlx4_warn(dev, "Bad cable detected on port %u\n",
+					  eqe->event.bad_cable.port);
+				break;
+			case MLX4_RECOVERABLE_ERROR_EVENT_SUBTYPE_UNSUPPORTED_CABLE:
+				mlx4_warn(dev, "Unsupported cable detected\n");
+				break;
+			default:
+				mlx4_dbg(dev,
+					 "Unhandled recoverable error event detected: %02x(%02x) on EQ %d at index %u. owner=%x, nent=0x%x, ownership=%s\n",
+					 eqe->type, eqe->subtype, eq->eqn,
+					 eq->cons_index, eqe->owner, eq->nent,
+					 !!(eqe->owner & 0x80) ^
+					 !!(eq->cons_index & eq->nent) ? "HW" : "SW");
+				break;
+			}
+			break;
+
+		case MLX4_EVENT_TYPE_EEC_CATAS_ERROR:
+		case MLX4_EVENT_TYPE_ECC_DETECT:
+		default:
+			mlx4_warn(dev, "Unhandled event %02x(%02x) on EQ %d at index %u. owner=%x, nent=0x%x, slave=%x, ownership=%s\n",
+				  eqe->type, eqe->subtype, eq->eqn,
+				  eq->cons_index, eqe->owner, eq->nent,
+				  eqe->slave_id,
+				  !!(eqe->owner & 0x80) ^
+				  !!(eq->cons_index & eq->nent) ? "HW" : "SW");
+			break;
+		}
+
+		++eq->cons_index;
+		eqes_found = 1;
+		++set_ci;
+
+		/*
+		 * The HCA will think the queue has overflowed if we
+		 * don't tell it we've been processing events.  We
+		 * create our EQs with MLX4_NUM_SPARE_EQE extra
+		 * entries, so we must update our consumer index at
+		 * least that often.
+		 */
+		if (unlikely(set_ci >= MLX4_NUM_SPARE_EQE)) {
+			eq_set_ci(eq, 0);
+			set_ci = 0;
+		}
+	}
+
+	eq_set_ci(eq, 1);
+
+	return eqes_found;
+}
+
+static irqreturn_t mlx4_interrupt(int irq, void *dev_ptr)
+{
+	struct mlx4_dev *dev = dev_ptr;
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int work = 0;
+	int i;
+
+	writel(priv->eq_table.clr_mask, priv->eq_table.clr_int);
+
+	for (i = 0; i < dev->caps.num_comp_vectors + 1; ++i)
+		work |= mlx4_eq_int(dev, &priv->eq_table.eq[i]);
+
+	return IRQ_RETVAL(work);
+}
+
+static irqreturn_t mlx4_msi_x_interrupt(int irq, void *eq_ptr)
+{
+	struct mlx4_eq  *eq  = eq_ptr;
+	struct mlx4_dev *dev = eq->dev;
+
+	mlx4_eq_int(dev, eq);
+
+	/* MSI-X vectors always belong to us */
+	return IRQ_HANDLED;
+}
+
+int mlx4_MAP_EQ_wrapper(struct mlx4_dev *dev, int slave,
+			struct mlx4_vhcr *vhcr,
+			struct mlx4_cmd_mailbox *inbox,
+			struct mlx4_cmd_mailbox *outbox,
+			struct mlx4_cmd_info *cmd)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_slave_event_eq_info *event_eq =
+		priv->mfunc.master.slave_state[slave].event_eq;
+	u32 in_modifier = vhcr->in_modifier;
+	u32 eqn = in_modifier & 0x3FF;
+	u64 in_param =  vhcr->in_param;
+	int err = 0;
+	int i;
+
+	if (slave == dev->caps.function)
+		err = mlx4_cmd(dev, in_param, (in_modifier & 0x80000000) | eqn,
+			       0, MLX4_CMD_MAP_EQ, MLX4_CMD_TIME_CLASS_B,
+			       MLX4_CMD_NATIVE);
+	if (!err)
+		for (i = 0; i < MLX4_EVENT_TYPES_NUM; ++i)
+			if (in_param & (1LL << i))
+				event_eq[i].eqn = in_modifier >> 31 ? -1 : eqn;
+
+	return err;
+}
+
+static int mlx4_MAP_EQ(struct mlx4_dev *dev, u64 event_mask, int unmap,
+			int eq_num)
+{
+	return mlx4_cmd(dev, event_mask, (unmap << 31) | eq_num,
+			0, MLX4_CMD_MAP_EQ, MLX4_CMD_TIME_CLASS_B,
+			MLX4_CMD_WRAPPED);
+}
+
+static int mlx4_SW2HW_EQ(struct mlx4_dev *dev, struct mlx4_cmd_mailbox *mailbox,
+			 int eq_num)
+{
+	return mlx4_cmd(dev, mailbox->dma, eq_num, 0,
+			MLX4_CMD_SW2HW_EQ, MLX4_CMD_TIME_CLASS_A,
+			MLX4_CMD_WRAPPED);
+}
+
+static int mlx4_HW2SW_EQ(struct mlx4_dev *dev,  int eq_num)
+{
+	return mlx4_cmd(dev, 0, eq_num, 1, MLX4_CMD_HW2SW_EQ,
+			MLX4_CMD_TIME_CLASS_A, MLX4_CMD_WRAPPED);
+}
+
+static int mlx4_num_eq_uar(struct mlx4_dev *dev)
+{
+	/*
+	 * Each UAR holds 4 EQ doorbells.  To figure out how many UARs
+	 * we need to map, take the difference of highest index and
+	 * the lowest index we'll use and add 1.
+	 */
+	return (dev->caps.num_comp_vectors + 1 + dev->caps.reserved_eqs) / 4 -
+		dev->caps.reserved_eqs / 4 + 1;
+}
+
+static void __iomem *mlx4_get_eq_uar(struct mlx4_dev *dev, struct mlx4_eq *eq)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int index;
+
+	index = eq->eqn / 4 - dev->caps.reserved_eqs / 4;
+
+	if (!priv->eq_table.uar_map[index]) {
+		priv->eq_table.uar_map[index] =
+			ioremap(
+				pci_resource_start(dev->persist->pdev, 2) +
+				((eq->eqn / 4) << (dev->uar_page_shift)),
+				(1 << (dev->uar_page_shift)));
+		if (!priv->eq_table.uar_map[index]) {
+			mlx4_err(dev, "Couldn't map EQ doorbell for EQN 0x%06x\n",
+				 eq->eqn);
+			return NULL;
+		}
+	}
+
+	return priv->eq_table.uar_map[index] + 0x800 + 8 * (eq->eqn % 4);
+}
+
+static void mlx4_unmap_uar(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int i;
+
+	for (i = 0; i < mlx4_num_eq_uar(dev); ++i)
+		if (priv->eq_table.uar_map[i]) {
+			iounmap(priv->eq_table.uar_map[i]);
+			priv->eq_table.uar_map[i] = NULL;
+		}
+}
+
+static int mlx4_create_eq(struct mlx4_dev *dev, int nent,
+			  u8 intr, struct mlx4_eq *eq)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_cmd_mailbox *mailbox;
+	struct mlx4_eq_context *eq_context;
+	int npages;
+	u64 *dma_list = NULL;
+	dma_addr_t t;
+	u64 mtt_addr;
+	int err = -ENOMEM;
+	int i;
+
+	eq->dev   = dev;
+	eq->nent  = roundup_pow_of_two(max(nent, 2));
+	/* CX3 is capable of extending the CQE/EQE from 32 to 64 bytes, with
+	 * strides of 64B,128B and 256B.
+	 */
+	npages = PAGE_ALIGN(eq->nent * dev->caps.eqe_size) / PAGE_SIZE;
+
+	eq->page_list = kmalloc_array(npages, sizeof(*eq->page_list),
+				      GFP_KERNEL);
+	if (!eq->page_list)
+		goto err_out;
+
+	for (i = 0; i < npages; ++i)
+		eq->page_list[i].buf = NULL;
+
+	dma_list = kmalloc_array(npages, sizeof(*dma_list), GFP_KERNEL);
+	if (!dma_list)
+		goto err_out_free;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		goto err_out_free;
+	eq_context = mailbox->buf;
+
+	for (i = 0; i < npages; ++i) {
+		eq->page_list[i].buf = dma_alloc_coherent(&dev->persist->
+							  pdev->dev,
+							  PAGE_SIZE, &t,
+							  GFP_KERNEL);
+		if (!eq->page_list[i].buf)
+			goto err_out_free_pages;
+
+		dma_list[i] = t;
+		eq->page_list[i].map = t;
+	}
+
+	eq->eqn = mlx4_bitmap_alloc(&priv->eq_table.bitmap);
+	if (eq->eqn == -1)
+		goto err_out_free_pages;
+
+	eq->doorbell = mlx4_get_eq_uar(dev, eq);
+	if (!eq->doorbell) {
+		err = -ENOMEM;
+		goto err_out_free_eq;
+	}
+
+	err = mlx4_mtt_init(dev, npages, PAGE_SHIFT, &eq->mtt);
+	if (err)
+		goto err_out_free_eq;
+
+	err = mlx4_write_mtt(dev, &eq->mtt, 0, npages, dma_list);
+	if (err)
+		goto err_out_free_mtt;
+
+	eq_context->flags	  = cpu_to_be32(MLX4_EQ_STATUS_OK   |
+						MLX4_EQ_STATE_ARMED);
+	eq_context->log_eq_size	  = ilog2(eq->nent);
+	eq_context->intr	  = intr;
+	eq_context->log_page_size = PAGE_SHIFT - MLX4_ICM_PAGE_SHIFT;
+
+	mtt_addr = mlx4_mtt_addr(dev, &eq->mtt);
+	eq_context->mtt_base_addr_h = mtt_addr >> 32;
+	eq_context->mtt_base_addr_l = cpu_to_be32(mtt_addr & 0xffffffff);
+
+	err = mlx4_SW2HW_EQ(dev, mailbox, eq->eqn);
+	if (err) {
+		mlx4_warn(dev, "SW2HW_EQ failed (%d)\n", err);
+		goto err_out_free_mtt;
+	}
+
+	kfree(dma_list);
+	mlx4_free_cmd_mailbox(dev, mailbox);
+
+	eq->cons_index = 0;
+
+	INIT_LIST_HEAD(&eq->tasklet_ctx.list);
+	INIT_LIST_HEAD(&eq->tasklet_ctx.process_list);
+	spin_lock_init(&eq->tasklet_ctx.lock);
+	tasklet_setup(&eq->tasklet_ctx.task, mlx4_cq_tasklet_cb);
+
+	return err;
+
+err_out_free_mtt:
+	mlx4_mtt_cleanup(dev, &eq->mtt);
+
+err_out_free_eq:
+	mlx4_bitmap_free(&priv->eq_table.bitmap, eq->eqn, MLX4_USE_RR);
+
+err_out_free_pages:
+	for (i = 0; i < npages; ++i)
+		if (eq->page_list[i].buf)
+			dma_free_coherent(&dev->persist->pdev->dev, PAGE_SIZE,
+					  eq->page_list[i].buf,
+					  eq->page_list[i].map);
+
+	mlx4_free_cmd_mailbox(dev, mailbox);
+
+err_out_free:
+	kfree(eq->page_list);
+	kfree(dma_list);
+
+err_out:
+	return err;
+}
+
+static void mlx4_free_eq(struct mlx4_dev *dev,
+			 struct mlx4_eq *eq)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int err;
+	int i;
+	/* CX3 is capable of extending the CQE/EQE from 32 to 64 bytes, with
+	 * strides of 64B,128B and 256B
+	 */
+	int npages = PAGE_ALIGN(dev->caps.eqe_size  * eq->nent) / PAGE_SIZE;
+
+	err = mlx4_HW2SW_EQ(dev, eq->eqn);
+	if (err)
+		mlx4_warn(dev, "HW2SW_EQ failed (%d)\n", err);
+
+	synchronize_irq(eq->irq);
+	tasklet_disable(&eq->tasklet_ctx.task);
+
+	mlx4_mtt_cleanup(dev, &eq->mtt);
+	for (i = 0; i < npages; ++i)
+		dma_free_coherent(&dev->persist->pdev->dev, PAGE_SIZE,
+				  eq->page_list[i].buf,
+				  eq->page_list[i].map);
+
+	kfree(eq->page_list);
+	mlx4_bitmap_free(&priv->eq_table.bitmap, eq->eqn, MLX4_USE_RR);
+}
+
+static void mlx4_free_irqs(struct mlx4_dev *dev)
+{
+	struct mlx4_eq_table *eq_table = &mlx4_priv(dev)->eq_table;
+	int	i;
+
+	if (eq_table->have_irq)
+		free_irq(dev->persist->pdev->irq, dev);
+
+	for (i = 0; i < dev->caps.num_comp_vectors + 1; ++i)
+		if (eq_table->eq[i].have_irq) {
+			free_cpumask_var(eq_table->eq[i].affinity_mask);
+			irq_update_affinity_hint(eq_table->eq[i].irq, NULL);
+			free_irq(eq_table->eq[i].irq, eq_table->eq + i);
+			eq_table->eq[i].have_irq = 0;
+		}
+
+	kfree(eq_table->irq_names);
+}
+
+static int mlx4_map_clr_int(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	priv->clr_base = ioremap(pci_resource_start(dev->persist->pdev,
+				 priv->fw.clr_int_bar) +
+				 priv->fw.clr_int_base, MLX4_CLR_INT_SIZE);
+	if (!priv->clr_base) {
+		mlx4_err(dev, "Couldn't map interrupt clear register, aborting\n");
+		return -ENOMEM;
+	}
+
+	return 0;
+}
+
+static void mlx4_unmap_clr_int(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	iounmap(priv->clr_base);
+}
+
+int mlx4_alloc_eq_table(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	priv->eq_table.eq = kcalloc(dev->caps.num_eqs - dev->caps.reserved_eqs,
+				    sizeof(*priv->eq_table.eq), GFP_KERNEL);
+	if (!priv->eq_table.eq)
+		return -ENOMEM;
+
+	return 0;
+}
+
+void mlx4_free_eq_table(struct mlx4_dev *dev)
+{
+	kfree(mlx4_priv(dev)->eq_table.eq);
+}
+
+int mlx4_init_eq_table(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int err;
+	int i;
+
+	priv->eq_table.uar_map = kcalloc(mlx4_num_eq_uar(dev),
+					 sizeof(*priv->eq_table.uar_map),
+					 GFP_KERNEL);
+	if (!priv->eq_table.uar_map) {
+		err = -ENOMEM;
+		goto err_out_free;
+	}
+
+	err = mlx4_bitmap_init(&priv->eq_table.bitmap,
+			       roundup_pow_of_two(dev->caps.num_eqs),
+			       dev->caps.num_eqs - 1,
+			       dev->caps.reserved_eqs,
+			       roundup_pow_of_two(dev->caps.num_eqs) -
+			       dev->caps.num_eqs);
+	if (err)
+		goto err_out_free;
+
+	for (i = 0; i < mlx4_num_eq_uar(dev); ++i)
+		priv->eq_table.uar_map[i] = NULL;
+
+	if (!mlx4_is_slave(dev)) {
+		err = mlx4_map_clr_int(dev);
+		if (err)
+			goto err_out_bitmap;
+
+		priv->eq_table.clr_mask =
+			swab32(1 << (priv->eq_table.inta_pin & 31));
+		priv->eq_table.clr_int  = priv->clr_base +
+			(priv->eq_table.inta_pin < 32 ? 4 : 0);
+	}
+
+	priv->eq_table.irq_names =
+		kmalloc_array(MLX4_IRQNAME_SIZE,
+			      (dev->caps.num_comp_vectors + 1),
+			      GFP_KERNEL);
+	if (!priv->eq_table.irq_names) {
+		err = -ENOMEM;
+		goto err_out_clr_int;
+	}
+
+	for (i = 0; i < dev->caps.num_comp_vectors + 1; ++i) {
+		if (i == MLX4_EQ_ASYNC) {
+			err = mlx4_create_eq(dev,
+					     MLX4_NUM_ASYNC_EQE + MLX4_NUM_SPARE_EQE,
+					     0, &priv->eq_table.eq[MLX4_EQ_ASYNC]);
+		} else {
+			struct mlx4_eq	*eq = &priv->eq_table.eq[i];
+#ifdef CONFIG_RFS_ACCEL
+			int port = find_first_bit(eq->actv_ports.ports,
+						  dev->caps.num_ports) + 1;
+
+			if (port <= dev->caps.num_ports) {
+				struct mlx4_port_info *info =
+					&mlx4_priv(dev)->port[port];
+
+				if (!info->rmap) {
+					info->rmap = alloc_irq_cpu_rmap(
+						mlx4_get_eqs_per_port(dev, port));
+					if (!info->rmap) {
+						mlx4_warn(dev, "Failed to allocate cpu rmap\n");
+						err = -ENOMEM;
+						goto err_out_unmap;
+					}
+				}
+
+				err = irq_cpu_rmap_add(
+					info->rmap, eq->irq);
+				if (err)
+					mlx4_warn(dev, "Failed adding irq rmap\n");
+			}
+#endif
+			err = mlx4_create_eq(dev, dev->quotas.cq +
+					     MLX4_NUM_SPARE_EQE,
+					     (dev->flags & MLX4_FLAG_MSI_X) ?
+					     i + 1 - !!(i > MLX4_EQ_ASYNC) : 0,
+					     eq);
+		}
+		if (err)
+			goto err_out_unmap;
+	}
+
+	if (dev->flags & MLX4_FLAG_MSI_X) {
+		const char *eq_name;
+
+		snprintf(priv->eq_table.irq_names +
+			 MLX4_EQ_ASYNC * MLX4_IRQNAME_SIZE,
+			 MLX4_IRQNAME_SIZE,
+			 "mlx4-async@pci:%s",
+			 pci_name(dev->persist->pdev));
+		eq_name = priv->eq_table.irq_names +
+			MLX4_EQ_ASYNC * MLX4_IRQNAME_SIZE;
+
+		err = request_irq(priv->eq_table.eq[MLX4_EQ_ASYNC].irq,
+				  mlx4_msi_x_interrupt, 0, eq_name,
+				  priv->eq_table.eq + MLX4_EQ_ASYNC);
+		if (err)
+			goto err_out_unmap;
+
+		priv->eq_table.eq[MLX4_EQ_ASYNC].have_irq = 1;
+	} else {
+		snprintf(priv->eq_table.irq_names,
+			 MLX4_IRQNAME_SIZE,
+			 DRV_NAME "@pci:%s",
+			 pci_name(dev->persist->pdev));
+		err = request_irq(dev->persist->pdev->irq, mlx4_interrupt,
+				  IRQF_SHARED, priv->eq_table.irq_names, dev);
+		if (err)
+			goto err_out_unmap;
+
+		priv->eq_table.have_irq = 1;
+	}
+
+	err = mlx4_MAP_EQ(dev, get_async_ev_mask(dev), 0,
+			  priv->eq_table.eq[MLX4_EQ_ASYNC].eqn);
+	if (err)
+		mlx4_warn(dev, "MAP_EQ for async EQ %d failed (%d)\n",
+			   priv->eq_table.eq[MLX4_EQ_ASYNC].eqn, err);
+
+	/* arm ASYNC eq */
+	eq_set_ci(&priv->eq_table.eq[MLX4_EQ_ASYNC], 1);
+
+	return 0;
+
+err_out_unmap:
+	while (i > 0)
+		mlx4_free_eq(dev, &priv->eq_table.eq[--i]);
+#ifdef CONFIG_RFS_ACCEL
+	for (i = 1; i <= dev->caps.num_ports; i++) {
+		if (mlx4_priv(dev)->port[i].rmap) {
+			free_irq_cpu_rmap(mlx4_priv(dev)->port[i].rmap);
+			mlx4_priv(dev)->port[i].rmap = NULL;
+		}
+	}
+#endif
+	mlx4_free_irqs(dev);
+
+err_out_clr_int:
+	if (!mlx4_is_slave(dev))
+		mlx4_unmap_clr_int(dev);
+
+err_out_bitmap:
+	mlx4_unmap_uar(dev);
+	mlx4_bitmap_cleanup(&priv->eq_table.bitmap);
+
+err_out_free:
+	kfree(priv->eq_table.uar_map);
+
+	return err;
+}
+
+void mlx4_cleanup_eq_table(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int i;
+
+	mlx4_MAP_EQ(dev, get_async_ev_mask(dev), 1,
+		    priv->eq_table.eq[MLX4_EQ_ASYNC].eqn);
+
+#ifdef CONFIG_RFS_ACCEL
+	for (i = 1; i <= dev->caps.num_ports; i++) {
+		if (mlx4_priv(dev)->port[i].rmap) {
+			free_irq_cpu_rmap(mlx4_priv(dev)->port[i].rmap);
+			mlx4_priv(dev)->port[i].rmap = NULL;
+		}
+	}
+#endif
+	mlx4_free_irqs(dev);
+
+	for (i = 0; i < dev->caps.num_comp_vectors + 1; ++i)
+		mlx4_free_eq(dev, &priv->eq_table.eq[i]);
+
+	if (!mlx4_is_slave(dev))
+		mlx4_unmap_clr_int(dev);
+
+	mlx4_unmap_uar(dev);
+	mlx4_bitmap_cleanup(&priv->eq_table.bitmap);
+
+	kfree(priv->eq_table.uar_map);
+}
+
+/* A test that verifies that we can accept interrupts
+ * on the vector allocated for asynchronous events
+ */
+int mlx4_test_async(struct mlx4_dev *dev)
+{
+	return mlx4_NOP(dev);
+}
+EXPORT_SYMBOL(mlx4_test_async);
+
+/* A test that verifies that we can accept interrupts
+ * on the given irq vector of the tested port.
+ * Interrupts are checked using the NOP command.
+ */
+int mlx4_test_interrupt(struct mlx4_dev *dev, int vector)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int err;
+
+	/* Temporary use polling for command completions */
+	mlx4_cmd_use_polling(dev);
+
+	/* Map the new eq to handle all asynchronous events */
+	err = mlx4_MAP_EQ(dev, get_async_ev_mask(dev), 0,
+			  priv->eq_table.eq[MLX4_CQ_TO_EQ_VECTOR(vector)].eqn);
+	if (err) {
+		mlx4_warn(dev, "Failed mapping eq for interrupt test\n");
+		goto out;
+	}
+
+	/* Go back to using events */
+	mlx4_cmd_use_events(dev);
+	err = mlx4_NOP(dev);
+
+	/* Return to default */
+	mlx4_cmd_use_polling(dev);
+out:
+	mlx4_MAP_EQ(dev, get_async_ev_mask(dev), 0,
+		    priv->eq_table.eq[MLX4_EQ_ASYNC].eqn);
+	mlx4_cmd_use_events(dev);
+
+	return err;
+}
+EXPORT_SYMBOL(mlx4_test_interrupt);
+
+bool mlx4_is_eq_vector_valid(struct mlx4_dev *dev, u8 port, int vector)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	vector = MLX4_CQ_TO_EQ_VECTOR(vector);
+	if (vector < 0 || (vector >= dev->caps.num_comp_vectors + 1) ||
+	    (vector == MLX4_EQ_ASYNC))
+		return false;
+
+	return test_bit(port - 1, priv->eq_table.eq[vector].actv_ports.ports);
+}
+EXPORT_SYMBOL(mlx4_is_eq_vector_valid);
+
+u32 mlx4_get_eqs_per_port(struct mlx4_dev *dev, u8 port)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	unsigned int i;
+	unsigned int sum = 0;
+
+	for (i = 0; i < dev->caps.num_comp_vectors + 1; i++)
+		sum += !!test_bit(port - 1,
+				  priv->eq_table.eq[i].actv_ports.ports);
+
+	return sum;
+}
+EXPORT_SYMBOL(mlx4_get_eqs_per_port);
+
+int mlx4_is_eq_shared(struct mlx4_dev *dev, int vector)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	vector = MLX4_CQ_TO_EQ_VECTOR(vector);
+	if (vector <= 0 || (vector >= dev->caps.num_comp_vectors + 1))
+		return -EINVAL;
+
+	return !!(bitmap_weight(priv->eq_table.eq[vector].actv_ports.ports,
+				dev->caps.num_ports) > 1);
+}
+EXPORT_SYMBOL(mlx4_is_eq_shared);
+
+struct cpu_rmap *mlx4_get_cpu_rmap(struct mlx4_dev *dev, int port)
+{
+	return mlx4_priv(dev)->port[port].rmap;
+}
+EXPORT_SYMBOL(mlx4_get_cpu_rmap);
+
+int mlx4_assign_eq(struct mlx4_dev *dev, u8 port, int *vector)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int err = 0, i = 0;
+	u32 min_ref_count_val = (u32)-1;
+	int requested_vector = MLX4_CQ_TO_EQ_VECTOR(*vector);
+	int *prequested_vector = NULL;
+
+
+	mutex_lock(&priv->msix_ctl.pool_lock);
+	if (requested_vector < (dev->caps.num_comp_vectors + 1) &&
+	    (requested_vector >= 0) &&
+	    (requested_vector != MLX4_EQ_ASYNC)) {
+		if (test_bit(port - 1,
+			     priv->eq_table.eq[requested_vector].actv_ports.ports)) {
+			prequested_vector = &requested_vector;
+		} else {
+			struct mlx4_eq *eq;
+
+			for (i = 1; i < port;
+			     requested_vector += mlx4_get_eqs_per_port(dev, i++))
+				;
+
+			eq = &priv->eq_table.eq[requested_vector];
+			if (requested_vector < dev->caps.num_comp_vectors + 1 &&
+			    test_bit(port - 1, eq->actv_ports.ports)) {
+				prequested_vector = &requested_vector;
+			}
+		}
+	}
+
+	if  (!prequested_vector) {
+		requested_vector = -1;
+		for (i = 0; min_ref_count_val && i < dev->caps.num_comp_vectors + 1;
+		     i++) {
+			struct mlx4_eq *eq = &priv->eq_table.eq[i];
+
+			if (min_ref_count_val > eq->ref_count &&
+			    test_bit(port - 1, eq->actv_ports.ports)) {
+				min_ref_count_val = eq->ref_count;
+				requested_vector = i;
+			}
+		}
+
+		if (requested_vector < 0) {
+			err = -ENOSPC;
+			goto err_unlock;
+		}
+
+		prequested_vector = &requested_vector;
+	}
+
+	if (!test_bit(*prequested_vector, priv->msix_ctl.pool_bm) &&
+	    dev->flags & MLX4_FLAG_MSI_X) {
+		set_bit(*prequested_vector, priv->msix_ctl.pool_bm);
+		snprintf(priv->eq_table.irq_names +
+			 *prequested_vector * MLX4_IRQNAME_SIZE,
+			 MLX4_IRQNAME_SIZE, "mlx4-%d@%s",
+			 *prequested_vector, dev_name(&dev->persist->pdev->dev));
+
+		err = request_irq(priv->eq_table.eq[*prequested_vector].irq,
+				  mlx4_msi_x_interrupt, 0,
+				  &priv->eq_table.irq_names[*prequested_vector << 5],
+				  priv->eq_table.eq + *prequested_vector);
+
+		if (err) {
+			clear_bit(*prequested_vector, priv->msix_ctl.pool_bm);
+			*prequested_vector = -1;
+		} else {
+#if defined(CONFIG_SMP)
+			mlx4_set_eq_affinity_hint(priv, *prequested_vector);
+#endif
+			eq_set_ci(&priv->eq_table.eq[*prequested_vector], 1);
+			priv->eq_table.eq[*prequested_vector].have_irq = 1;
+		}
+	}
+
+	if (!err && *prequested_vector >= 0)
+		priv->eq_table.eq[*prequested_vector].ref_count++;
+
+err_unlock:
+	mutex_unlock(&priv->msix_ctl.pool_lock);
+
+	if (!err && *prequested_vector >= 0)
+		*vector = MLX4_EQ_TO_CQ_VECTOR(*prequested_vector);
+	else
+		*vector = 0;
+
+	return err;
+}
+EXPORT_SYMBOL(mlx4_assign_eq);
+
+int mlx4_eq_get_irq(struct mlx4_dev *dev, int cq_vec)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	return priv->eq_table.eq[MLX4_CQ_TO_EQ_VECTOR(cq_vec)].irq;
+}
+EXPORT_SYMBOL(mlx4_eq_get_irq);
+
+void mlx4_release_eq(struct mlx4_dev *dev, int vec)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int eq_vec = MLX4_CQ_TO_EQ_VECTOR(vec);
+
+	mutex_lock(&priv->msix_ctl.pool_lock);
+	priv->eq_table.eq[eq_vec].ref_count--;
+
+	/* once we allocated EQ, we don't release it because it might be binded
+	 * to cpu_rmap.
+	 */
+	mutex_unlock(&priv->msix_ctl.pool_lock);
+}
+EXPORT_SYMBOL(mlx4_release_eq);
+
diff --git a/drivers/net/ethernet/mellanox/mlx4/fw.c b/drivers/net/ethernet/mellanox/mlx4/fw.c
new file mode 100644
index 000000000..fe48d20d6
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx4/fw.c
@@ -0,0 +1,3111 @@
+/*
+ * Copyright (c) 2004, 2005 Topspin Communications.  All rights reserved.
+ * Copyright (c) 2005, 2006, 2007, 2008 Mellanox Technologies. All rights reserved.
+ * Copyright (c) 2005, 2006, 2007 Cisco Systems, Inc.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/etherdevice.h>
+#include <linux/mlx4/cmd.h>
+#include <linux/module.h>
+#include <linux/cache.h>
+#include <linux/kernel.h>
+#include <uapi/rdma/mlx4-abi.h>
+
+#include "fw.h"
+#include "icm.h"
+
+enum {
+	MLX4_COMMAND_INTERFACE_MIN_REV		= 2,
+	MLX4_COMMAND_INTERFACE_MAX_REV		= 3,
+	MLX4_COMMAND_INTERFACE_NEW_PORT_CMDS	= 3,
+};
+
+extern void __buggy_use_of_MLX4_GET(void);
+extern void __buggy_use_of_MLX4_PUT(void);
+
+static bool enable_qos;
+module_param(enable_qos, bool, 0444);
+MODULE_PARM_DESC(enable_qos, "Enable Enhanced QoS support (default: off)");
+
+#define MLX4_GET(dest, source, offset)				      \
+	do {							      \
+		void *__p = (char *) (source) + (offset);	      \
+		__be64 val;                                           \
+		switch (sizeof(dest)) {				      \
+		case 1: (dest) = *(u8 *) __p;	    break;	      \
+		case 2: (dest) = be16_to_cpup(__p); break;	      \
+		case 4: (dest) = be32_to_cpup(__p); break;	      \
+		case 8: val = get_unaligned((__be64 *)__p);           \
+			(dest) = be64_to_cpu(val);  break;            \
+		default: __buggy_use_of_MLX4_GET();		      \
+		}						      \
+	} while (0)
+
+#define MLX4_PUT(dest, source, offset)				      \
+	do {							      \
+		void *__d = ((char *) (dest) + (offset));	      \
+		switch (sizeof(source)) {			      \
+		case 1: *(u8 *) __d = (source);		       break; \
+		case 2:	*(__be16 *) __d = cpu_to_be16(source); break; \
+		case 4:	*(__be32 *) __d = cpu_to_be32(source); break; \
+		case 8:	*(__be64 *) __d = cpu_to_be64(source); break; \
+		default: __buggy_use_of_MLX4_PUT();		      \
+		}						      \
+	} while (0)
+
+static void dump_dev_cap_flags(struct mlx4_dev *dev, u64 flags)
+{
+	static const char *fname[] = {
+		[ 0] = "RC transport",
+		[ 1] = "UC transport",
+		[ 2] = "UD transport",
+		[ 3] = "XRC transport",
+		[ 6] = "SRQ support",
+		[ 7] = "IPoIB checksum offload",
+		[ 8] = "P_Key violation counter",
+		[ 9] = "Q_Key violation counter",
+		[12] = "Dual Port Different Protocol (DPDP) support",
+		[15] = "Big LSO headers",
+		[16] = "MW support",
+		[17] = "APM support",
+		[18] = "Atomic ops support",
+		[19] = "Raw multicast support",
+		[20] = "Address vector port checking support",
+		[21] = "UD multicast support",
+		[30] = "IBoE support",
+		[32] = "Unicast loopback support",
+		[34] = "FCS header control",
+		[37] = "Wake On LAN (port1) support",
+		[38] = "Wake On LAN (port2) support",
+		[40] = "UDP RSS support",
+		[41] = "Unicast VEP steering support",
+		[42] = "Multicast VEP steering support",
+		[48] = "Counters support",
+		[52] = "RSS IP fragments support",
+		[53] = "Port ETS Scheduler support",
+		[55] = "Port link type sensing support",
+		[59] = "Port management change event support",
+		[61] = "64 byte EQE support",
+		[62] = "64 byte CQE support",
+	};
+	int i;
+
+	mlx4_dbg(dev, "DEV_CAP flags:\n");
+	for (i = 0; i < ARRAY_SIZE(fname); ++i)
+		if (fname[i] && (flags & (1LL << i)))
+			mlx4_dbg(dev, "    %s\n", fname[i]);
+}
+
+static void dump_dev_cap_flags2(struct mlx4_dev *dev, u64 flags)
+{
+	static const char * const fname[] = {
+		[0] = "RSS support",
+		[1] = "RSS Toeplitz Hash Function support",
+		[2] = "RSS XOR Hash Function support",
+		[3] = "Device managed flow steering support",
+		[4] = "Automatic MAC reassignment support",
+		[5] = "Time stamping support",
+		[6] = "VST (control vlan insertion/stripping) support",
+		[7] = "FSM (MAC anti-spoofing) support",
+		[8] = "Dynamic QP updates support",
+		[9] = "Device managed flow steering IPoIB support",
+		[10] = "TCP/IP offloads/flow-steering for VXLAN support",
+		[11] = "MAD DEMUX (Secure-Host) support",
+		[12] = "Large cache line (>64B) CQE stride support",
+		[13] = "Large cache line (>64B) EQE stride support",
+		[14] = "Ethernet protocol control support",
+		[15] = "Ethernet Backplane autoneg support",
+		[16] = "CONFIG DEV support",
+		[17] = "Asymmetric EQs support",
+		[18] = "More than 80 VFs support",
+		[19] = "Performance optimized for limited rule configuration flow steering support",
+		[20] = "Recoverable error events support",
+		[21] = "Port Remap support",
+		[22] = "QCN support",
+		[23] = "QP rate limiting support",
+		[24] = "Ethernet Flow control statistics support",
+		[25] = "Granular QoS per VF support",
+		[26] = "Port ETS Scheduler support",
+		[27] = "Port beacon support",
+		[28] = "RX-ALL support",
+		[29] = "802.1ad offload support",
+		[31] = "Modifying loopback source checks using UPDATE_QP support",
+		[32] = "Loopback source checks support",
+		[33] = "RoCEv2 support",
+		[34] = "DMFS Sniffer support (UC & MC)",
+		[35] = "Diag counters per port",
+		[36] = "QinQ VST mode support",
+		[37] = "sl to vl mapping table change event support",
+		[38] = "user MAC support",
+		[39] = "Report driver version to FW support",
+		[40] = "SW CQ initialization support",
+	};
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(fname); ++i)
+		if (fname[i] && (flags & (1LL << i)))
+			mlx4_dbg(dev, "    %s\n", fname[i]);
+}
+
+int mlx4_MOD_STAT_CFG(struct mlx4_dev *dev, struct mlx4_mod_stat_cfg *cfg)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	u32 *inbox;
+	int err = 0;
+
+#define MOD_STAT_CFG_IN_SIZE		0x100
+
+#define MOD_STAT_CFG_PG_SZ_M_OFFSET	0x002
+#define MOD_STAT_CFG_PG_SZ_OFFSET	0x003
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+	inbox = mailbox->buf;
+
+	MLX4_PUT(inbox, cfg->log_pg_sz, MOD_STAT_CFG_PG_SZ_OFFSET);
+	MLX4_PUT(inbox, cfg->log_pg_sz_m, MOD_STAT_CFG_PG_SZ_M_OFFSET);
+
+	err = mlx4_cmd(dev, mailbox->dma, 0, 0, MLX4_CMD_MOD_STAT_CFG,
+			MLX4_CMD_TIME_CLASS_A, MLX4_CMD_NATIVE);
+
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return err;
+}
+
+int mlx4_QUERY_FUNC(struct mlx4_dev *dev, struct mlx4_func *func, int slave)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	u32 *outbox;
+	u8 in_modifier;
+	u8 field;
+	u16 field16;
+	int err;
+
+#define QUERY_FUNC_BUS_OFFSET			0x00
+#define QUERY_FUNC_DEVICE_OFFSET		0x01
+#define QUERY_FUNC_FUNCTION_OFFSET		0x01
+#define QUERY_FUNC_PHYSICAL_FUNCTION_OFFSET	0x03
+#define QUERY_FUNC_RSVD_EQS_OFFSET		0x04
+#define QUERY_FUNC_MAX_EQ_OFFSET		0x06
+#define QUERY_FUNC_RSVD_UARS_OFFSET		0x0b
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+	outbox = mailbox->buf;
+
+	in_modifier = slave;
+
+	err = mlx4_cmd_box(dev, 0, mailbox->dma, in_modifier, 0,
+			   MLX4_CMD_QUERY_FUNC,
+			   MLX4_CMD_TIME_CLASS_A,
+			   MLX4_CMD_NATIVE);
+	if (err)
+		goto out;
+
+	MLX4_GET(field, outbox, QUERY_FUNC_BUS_OFFSET);
+	func->bus = field & 0xf;
+	MLX4_GET(field, outbox, QUERY_FUNC_DEVICE_OFFSET);
+	func->device = field & 0xf1;
+	MLX4_GET(field, outbox, QUERY_FUNC_FUNCTION_OFFSET);
+	func->function = field & 0x7;
+	MLX4_GET(field, outbox, QUERY_FUNC_PHYSICAL_FUNCTION_OFFSET);
+	func->physical_function = field & 0xf;
+	MLX4_GET(field16, outbox, QUERY_FUNC_RSVD_EQS_OFFSET);
+	func->rsvd_eqs = field16 & 0xffff;
+	MLX4_GET(field16, outbox, QUERY_FUNC_MAX_EQ_OFFSET);
+	func->max_eq = field16 & 0xffff;
+	MLX4_GET(field, outbox, QUERY_FUNC_RSVD_UARS_OFFSET);
+	func->rsvd_uars = field & 0x0f;
+
+	mlx4_dbg(dev, "Bus: %d, Device: %d, Function: %d, Physical function: %d, Max EQs: %d, Reserved EQs: %d, Reserved UARs: %d\n",
+		 func->bus, func->device, func->function, func->physical_function,
+		 func->max_eq, func->rsvd_eqs, func->rsvd_uars);
+
+out:
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return err;
+}
+
+static int mlx4_activate_vst_qinq(struct mlx4_priv *priv, int slave, int port)
+{
+	struct mlx4_vport_oper_state *vp_oper;
+	struct mlx4_vport_state *vp_admin;
+	int err;
+
+	vp_oper = &priv->mfunc.master.vf_oper[slave].vport[port];
+	vp_admin = &priv->mfunc.master.vf_admin[slave].vport[port];
+
+	if (vp_admin->default_vlan != vp_oper->state.default_vlan) {
+		err = __mlx4_register_vlan(&priv->dev, port,
+					   vp_admin->default_vlan,
+					   &vp_oper->vlan_idx);
+		if (err) {
+			vp_oper->vlan_idx = NO_INDX;
+			mlx4_warn(&priv->dev,
+				  "No vlan resources slave %d, port %d\n",
+				  slave, port);
+			return err;
+		}
+		mlx4_dbg(&priv->dev, "alloc vlan %d idx  %d slave %d port %d\n",
+			 (int)(vp_oper->state.default_vlan),
+			 vp_oper->vlan_idx, slave, port);
+	}
+	vp_oper->state.vlan_proto   = vp_admin->vlan_proto;
+	vp_oper->state.default_vlan = vp_admin->default_vlan;
+	vp_oper->state.default_qos  = vp_admin->default_qos;
+
+	return 0;
+}
+
+static int mlx4_handle_vst_qinq(struct mlx4_priv *priv, int slave, int port)
+{
+	struct mlx4_vport_oper_state *vp_oper;
+	struct mlx4_slave_state *slave_state;
+	struct mlx4_vport_state *vp_admin;
+	int err;
+
+	vp_oper = &priv->mfunc.master.vf_oper[slave].vport[port];
+	vp_admin = &priv->mfunc.master.vf_admin[slave].vport[port];
+	slave_state = &priv->mfunc.master.slave_state[slave];
+
+	if ((vp_admin->vlan_proto != htons(ETH_P_8021AD)) ||
+	    (!slave_state->active))
+		return 0;
+
+	if (vp_oper->state.vlan_proto == vp_admin->vlan_proto &&
+	    vp_oper->state.default_vlan == vp_admin->default_vlan &&
+	    vp_oper->state.default_qos == vp_admin->default_qos)
+		return 0;
+
+	if (!slave_state->vst_qinq_supported) {
+		/* Warn and revert the request to set vst QinQ mode */
+		vp_admin->vlan_proto   = vp_oper->state.vlan_proto;
+		vp_admin->default_vlan = vp_oper->state.default_vlan;
+		vp_admin->default_qos  = vp_oper->state.default_qos;
+
+		mlx4_warn(&priv->dev,
+			  "Slave %d does not support VST QinQ mode\n", slave);
+		return 0;
+	}
+
+	err = mlx4_activate_vst_qinq(priv, slave, port);
+	return err;
+}
+
+int mlx4_QUERY_FUNC_CAP_wrapper(struct mlx4_dev *dev, int slave,
+				struct mlx4_vhcr *vhcr,
+				struct mlx4_cmd_mailbox *inbox,
+				struct mlx4_cmd_mailbox *outbox,
+				struct mlx4_cmd_info *cmd)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	u8	field, port;
+	u32	size, proxy_qp, qkey;
+	int	err = 0;
+	struct mlx4_func func;
+
+#define QUERY_FUNC_CAP_FLAGS_OFFSET		0x0
+#define QUERY_FUNC_CAP_NUM_PORTS_OFFSET		0x1
+#define QUERY_FUNC_CAP_PF_BHVR_OFFSET		0x4
+#define QUERY_FUNC_CAP_FMR_OFFSET		0x8
+#define QUERY_FUNC_CAP_QP_QUOTA_OFFSET_DEP	0x10
+#define QUERY_FUNC_CAP_CQ_QUOTA_OFFSET_DEP	0x14
+#define QUERY_FUNC_CAP_SRQ_QUOTA_OFFSET_DEP	0x18
+#define QUERY_FUNC_CAP_MPT_QUOTA_OFFSET_DEP	0x20
+#define QUERY_FUNC_CAP_MTT_QUOTA_OFFSET_DEP	0x24
+#define QUERY_FUNC_CAP_MCG_QUOTA_OFFSET_DEP	0x28
+#define QUERY_FUNC_CAP_MAX_EQ_OFFSET		0x2c
+#define QUERY_FUNC_CAP_RESERVED_EQ_OFFSET	0x30
+#define QUERY_FUNC_CAP_QP_RESD_LKEY_OFFSET	0x48
+
+#define QUERY_FUNC_CAP_QP_QUOTA_OFFSET		0x50
+#define QUERY_FUNC_CAP_CQ_QUOTA_OFFSET		0x54
+#define QUERY_FUNC_CAP_SRQ_QUOTA_OFFSET		0x58
+#define QUERY_FUNC_CAP_MPT_QUOTA_OFFSET		0x60
+#define QUERY_FUNC_CAP_MTT_QUOTA_OFFSET		0x64
+#define QUERY_FUNC_CAP_MCG_QUOTA_OFFSET		0x68
+
+#define QUERY_FUNC_CAP_EXTRA_FLAGS_OFFSET	0x6c
+
+#define QUERY_FUNC_CAP_FMR_FLAG			0x80
+#define QUERY_FUNC_CAP_FLAG_RDMA		0x40
+#define QUERY_FUNC_CAP_FLAG_ETH			0x80
+#define QUERY_FUNC_CAP_FLAG_QUOTAS		0x10
+#define QUERY_FUNC_CAP_FLAG_RESD_LKEY		0x08
+#define QUERY_FUNC_CAP_FLAG_VALID_MAILBOX	0x04
+
+#define QUERY_FUNC_CAP_EXTRA_FLAGS_BF_QP_ALLOC_FLAG	(1UL << 31)
+#define QUERY_FUNC_CAP_EXTRA_FLAGS_A0_QP_ALLOC_FLAG	(1UL << 30)
+
+/* when opcode modifier = 1 */
+#define QUERY_FUNC_CAP_PHYS_PORT_OFFSET		0x3
+#define QUERY_FUNC_CAP_PRIV_VF_QKEY_OFFSET	0x4
+#define QUERY_FUNC_CAP_FLAGS0_OFFSET		0x8
+#define QUERY_FUNC_CAP_FLAGS1_OFFSET		0xc
+
+#define QUERY_FUNC_CAP_QP0_TUNNEL		0x10
+#define QUERY_FUNC_CAP_QP0_PROXY		0x14
+#define QUERY_FUNC_CAP_QP1_TUNNEL		0x18
+#define QUERY_FUNC_CAP_QP1_PROXY		0x1c
+#define QUERY_FUNC_CAP_PHYS_PORT_ID		0x28
+
+#define QUERY_FUNC_CAP_FLAGS1_FORCE_MAC		0x40
+#define QUERY_FUNC_CAP_FLAGS1_FORCE_VLAN	0x80
+#define QUERY_FUNC_CAP_FLAGS1_NIC_INFO			0x10
+#define QUERY_FUNC_CAP_VF_ENABLE_QP0		0x08
+
+#define QUERY_FUNC_CAP_FLAGS0_FORCE_PHY_WQE_GID 0x80
+#define QUERY_FUNC_CAP_PHV_BIT			0x40
+#define QUERY_FUNC_CAP_VLAN_OFFLOAD_DISABLE	0x20
+
+#define QUERY_FUNC_CAP_SUPPORTS_VST_QINQ	BIT(30)
+#define QUERY_FUNC_CAP_SUPPORTS_NON_POWER_OF_2_NUM_EQS BIT(31)
+
+	if (vhcr->op_modifier == 1) {
+		struct mlx4_active_ports actv_ports =
+			mlx4_get_active_ports(dev, slave);
+		int converted_port = mlx4_slave_convert_port(
+				dev, slave, vhcr->in_modifier);
+		struct mlx4_vport_oper_state *vp_oper;
+
+		if (converted_port < 0)
+			return -EINVAL;
+
+		vhcr->in_modifier = converted_port;
+		/* phys-port = logical-port */
+		field = vhcr->in_modifier -
+			find_first_bit(actv_ports.ports, dev->caps.num_ports);
+		MLX4_PUT(outbox->buf, field, QUERY_FUNC_CAP_PHYS_PORT_OFFSET);
+
+		port = vhcr->in_modifier;
+		proxy_qp = dev->phys_caps.base_proxy_sqpn + 8 * slave + port - 1;
+
+		/* Set nic_info bit to mark new fields support */
+		field  = QUERY_FUNC_CAP_FLAGS1_NIC_INFO;
+
+		if (mlx4_vf_smi_enabled(dev, slave, port) &&
+		    !mlx4_get_parav_qkey(dev, proxy_qp, &qkey)) {
+			field |= QUERY_FUNC_CAP_VF_ENABLE_QP0;
+			MLX4_PUT(outbox->buf, qkey,
+				 QUERY_FUNC_CAP_PRIV_VF_QKEY_OFFSET);
+		}
+		MLX4_PUT(outbox->buf, field, QUERY_FUNC_CAP_FLAGS1_OFFSET);
+
+		/* size is now the QP number */
+		size = dev->phys_caps.base_tunnel_sqpn + 8 * slave + port - 1;
+		MLX4_PUT(outbox->buf, size, QUERY_FUNC_CAP_QP0_TUNNEL);
+
+		size += 2;
+		MLX4_PUT(outbox->buf, size, QUERY_FUNC_CAP_QP1_TUNNEL);
+
+		MLX4_PUT(outbox->buf, proxy_qp, QUERY_FUNC_CAP_QP0_PROXY);
+		proxy_qp += 2;
+		MLX4_PUT(outbox->buf, proxy_qp, QUERY_FUNC_CAP_QP1_PROXY);
+
+		MLX4_PUT(outbox->buf, dev->caps.phys_port_id[vhcr->in_modifier],
+			 QUERY_FUNC_CAP_PHYS_PORT_ID);
+
+		vp_oper = &priv->mfunc.master.vf_oper[slave].vport[port];
+		err = mlx4_handle_vst_qinq(priv, slave, port);
+		if (err)
+			return err;
+
+		field = 0;
+		if (dev->caps.phv_bit[port])
+			field |= QUERY_FUNC_CAP_PHV_BIT;
+		if (vp_oper->state.vlan_proto == htons(ETH_P_8021AD))
+			field |= QUERY_FUNC_CAP_VLAN_OFFLOAD_DISABLE;
+		MLX4_PUT(outbox->buf, field, QUERY_FUNC_CAP_FLAGS0_OFFSET);
+
+	} else if (vhcr->op_modifier == 0) {
+		struct mlx4_active_ports actv_ports =
+			mlx4_get_active_ports(dev, slave);
+		struct mlx4_slave_state *slave_state =
+			&priv->mfunc.master.slave_state[slave];
+
+		/* enable rdma and ethernet interfaces, new quota locations,
+		 * and reserved lkey
+		 */
+		field = (QUERY_FUNC_CAP_FLAG_ETH | QUERY_FUNC_CAP_FLAG_RDMA |
+			 QUERY_FUNC_CAP_FLAG_QUOTAS | QUERY_FUNC_CAP_FLAG_VALID_MAILBOX |
+			 QUERY_FUNC_CAP_FLAG_RESD_LKEY);
+		MLX4_PUT(outbox->buf, field, QUERY_FUNC_CAP_FLAGS_OFFSET);
+
+		field = min(
+			bitmap_weight(actv_ports.ports, dev->caps.num_ports),
+			(unsigned int) dev->caps.num_ports);
+		MLX4_PUT(outbox->buf, field, QUERY_FUNC_CAP_NUM_PORTS_OFFSET);
+
+		size = dev->caps.function_caps; /* set PF behaviours */
+		MLX4_PUT(outbox->buf, size, QUERY_FUNC_CAP_PF_BHVR_OFFSET);
+
+		field = 0; /* protected FMR support not available as yet */
+		MLX4_PUT(outbox->buf, field, QUERY_FUNC_CAP_FMR_OFFSET);
+
+		size = priv->mfunc.master.res_tracker.res_alloc[RES_QP].quota[slave];
+		MLX4_PUT(outbox->buf, size, QUERY_FUNC_CAP_QP_QUOTA_OFFSET);
+		size = dev->caps.num_qps;
+		MLX4_PUT(outbox->buf, size, QUERY_FUNC_CAP_QP_QUOTA_OFFSET_DEP);
+
+		size = priv->mfunc.master.res_tracker.res_alloc[RES_SRQ].quota[slave];
+		MLX4_PUT(outbox->buf, size, QUERY_FUNC_CAP_SRQ_QUOTA_OFFSET);
+		size = dev->caps.num_srqs;
+		MLX4_PUT(outbox->buf, size, QUERY_FUNC_CAP_SRQ_QUOTA_OFFSET_DEP);
+
+		size = priv->mfunc.master.res_tracker.res_alloc[RES_CQ].quota[slave];
+		MLX4_PUT(outbox->buf, size, QUERY_FUNC_CAP_CQ_QUOTA_OFFSET);
+		size = dev->caps.num_cqs;
+		MLX4_PUT(outbox->buf, size, QUERY_FUNC_CAP_CQ_QUOTA_OFFSET_DEP);
+
+		if (!(dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_SYS_EQS) ||
+		    mlx4_QUERY_FUNC(dev, &func, slave)) {
+			size = vhcr->in_modifier &
+				QUERY_FUNC_CAP_SUPPORTS_NON_POWER_OF_2_NUM_EQS ?
+				dev->caps.num_eqs :
+				rounddown_pow_of_two(dev->caps.num_eqs);
+			MLX4_PUT(outbox->buf, size, QUERY_FUNC_CAP_MAX_EQ_OFFSET);
+			size = dev->caps.reserved_eqs;
+			MLX4_PUT(outbox->buf, size, QUERY_FUNC_CAP_RESERVED_EQ_OFFSET);
+		} else {
+			size = vhcr->in_modifier &
+				QUERY_FUNC_CAP_SUPPORTS_NON_POWER_OF_2_NUM_EQS ?
+				func.max_eq :
+				rounddown_pow_of_two(func.max_eq);
+			MLX4_PUT(outbox->buf, size, QUERY_FUNC_CAP_MAX_EQ_OFFSET);
+			size = func.rsvd_eqs;
+			MLX4_PUT(outbox->buf, size, QUERY_FUNC_CAP_RESERVED_EQ_OFFSET);
+		}
+
+		size = priv->mfunc.master.res_tracker.res_alloc[RES_MPT].quota[slave];
+		MLX4_PUT(outbox->buf, size, QUERY_FUNC_CAP_MPT_QUOTA_OFFSET);
+		size = dev->caps.num_mpts;
+		MLX4_PUT(outbox->buf, size, QUERY_FUNC_CAP_MPT_QUOTA_OFFSET_DEP);
+
+		size = priv->mfunc.master.res_tracker.res_alloc[RES_MTT].quota[slave];
+		MLX4_PUT(outbox->buf, size, QUERY_FUNC_CAP_MTT_QUOTA_OFFSET);
+		size = dev->caps.num_mtts;
+		MLX4_PUT(outbox->buf, size, QUERY_FUNC_CAP_MTT_QUOTA_OFFSET_DEP);
+
+		size = dev->caps.num_mgms + dev->caps.num_amgms;
+		MLX4_PUT(outbox->buf, size, QUERY_FUNC_CAP_MCG_QUOTA_OFFSET);
+		MLX4_PUT(outbox->buf, size, QUERY_FUNC_CAP_MCG_QUOTA_OFFSET_DEP);
+
+		size = QUERY_FUNC_CAP_EXTRA_FLAGS_BF_QP_ALLOC_FLAG |
+			QUERY_FUNC_CAP_EXTRA_FLAGS_A0_QP_ALLOC_FLAG;
+		MLX4_PUT(outbox->buf, size, QUERY_FUNC_CAP_EXTRA_FLAGS_OFFSET);
+
+		size = dev->caps.reserved_lkey + ((slave << 8) & 0xFF00);
+		MLX4_PUT(outbox->buf, size, QUERY_FUNC_CAP_QP_RESD_LKEY_OFFSET);
+
+		if (vhcr->in_modifier & QUERY_FUNC_CAP_SUPPORTS_VST_QINQ)
+			slave_state->vst_qinq_supported = true;
+
+	} else
+		err = -EINVAL;
+
+	return err;
+}
+
+int mlx4_QUERY_FUNC_CAP(struct mlx4_dev *dev, u8 gen_or_port,
+			struct mlx4_func_cap *func_cap)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	u32			*outbox;
+	u8			field, op_modifier;
+	u32			size, qkey;
+	int			err = 0, quotas = 0;
+	u32                     in_modifier;
+	u32			slave_caps;
+
+	op_modifier = !!gen_or_port; /* 0 = general, 1 = logical port */
+	slave_caps = QUERY_FUNC_CAP_SUPPORTS_VST_QINQ |
+		QUERY_FUNC_CAP_SUPPORTS_NON_POWER_OF_2_NUM_EQS;
+	in_modifier = op_modifier ? gen_or_port : slave_caps;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+
+	err = mlx4_cmd_box(dev, 0, mailbox->dma, in_modifier, op_modifier,
+			   MLX4_CMD_QUERY_FUNC_CAP,
+			   MLX4_CMD_TIME_CLASS_A, MLX4_CMD_WRAPPED);
+	if (err)
+		goto out;
+
+	outbox = mailbox->buf;
+
+	if (!op_modifier) {
+		MLX4_GET(field, outbox, QUERY_FUNC_CAP_FLAGS_OFFSET);
+		if (!(field & (QUERY_FUNC_CAP_FLAG_ETH | QUERY_FUNC_CAP_FLAG_RDMA))) {
+			mlx4_err(dev, "The host supports neither eth nor rdma interfaces\n");
+			err = -EPROTONOSUPPORT;
+			goto out;
+		}
+		func_cap->flags = field;
+		quotas = !!(func_cap->flags & QUERY_FUNC_CAP_FLAG_QUOTAS);
+
+		MLX4_GET(field, outbox, QUERY_FUNC_CAP_NUM_PORTS_OFFSET);
+		func_cap->num_ports = field;
+
+		MLX4_GET(size, outbox, QUERY_FUNC_CAP_PF_BHVR_OFFSET);
+		func_cap->pf_context_behaviour = size;
+
+		if (quotas) {
+			MLX4_GET(size, outbox, QUERY_FUNC_CAP_QP_QUOTA_OFFSET);
+			func_cap->qp_quota = size & 0xFFFFFF;
+
+			MLX4_GET(size, outbox, QUERY_FUNC_CAP_SRQ_QUOTA_OFFSET);
+			func_cap->srq_quota = size & 0xFFFFFF;
+
+			MLX4_GET(size, outbox, QUERY_FUNC_CAP_CQ_QUOTA_OFFSET);
+			func_cap->cq_quota = size & 0xFFFFFF;
+
+			MLX4_GET(size, outbox, QUERY_FUNC_CAP_MPT_QUOTA_OFFSET);
+			func_cap->mpt_quota = size & 0xFFFFFF;
+
+			MLX4_GET(size, outbox, QUERY_FUNC_CAP_MTT_QUOTA_OFFSET);
+			func_cap->mtt_quota = size & 0xFFFFFF;
+
+			MLX4_GET(size, outbox, QUERY_FUNC_CAP_MCG_QUOTA_OFFSET);
+			func_cap->mcg_quota = size & 0xFFFFFF;
+
+		} else {
+			MLX4_GET(size, outbox, QUERY_FUNC_CAP_QP_QUOTA_OFFSET_DEP);
+			func_cap->qp_quota = size & 0xFFFFFF;
+
+			MLX4_GET(size, outbox, QUERY_FUNC_CAP_SRQ_QUOTA_OFFSET_DEP);
+			func_cap->srq_quota = size & 0xFFFFFF;
+
+			MLX4_GET(size, outbox, QUERY_FUNC_CAP_CQ_QUOTA_OFFSET_DEP);
+			func_cap->cq_quota = size & 0xFFFFFF;
+
+			MLX4_GET(size, outbox, QUERY_FUNC_CAP_MPT_QUOTA_OFFSET_DEP);
+			func_cap->mpt_quota = size & 0xFFFFFF;
+
+			MLX4_GET(size, outbox, QUERY_FUNC_CAP_MTT_QUOTA_OFFSET_DEP);
+			func_cap->mtt_quota = size & 0xFFFFFF;
+
+			MLX4_GET(size, outbox, QUERY_FUNC_CAP_MCG_QUOTA_OFFSET_DEP);
+			func_cap->mcg_quota = size & 0xFFFFFF;
+		}
+		MLX4_GET(size, outbox, QUERY_FUNC_CAP_MAX_EQ_OFFSET);
+		func_cap->max_eq = size & 0xFFFFFF;
+
+		MLX4_GET(size, outbox, QUERY_FUNC_CAP_RESERVED_EQ_OFFSET);
+		func_cap->reserved_eq = size & 0xFFFFFF;
+
+		if (func_cap->flags & QUERY_FUNC_CAP_FLAG_RESD_LKEY) {
+			MLX4_GET(size, outbox, QUERY_FUNC_CAP_QP_RESD_LKEY_OFFSET);
+			func_cap->reserved_lkey = size;
+		} else {
+			func_cap->reserved_lkey = 0;
+		}
+
+		func_cap->extra_flags = 0;
+
+		/* Mailbox data from 0x6c and onward should only be treated if
+		 * QUERY_FUNC_CAP_FLAG_VALID_MAILBOX is set in func_cap->flags
+		 */
+		if (func_cap->flags & QUERY_FUNC_CAP_FLAG_VALID_MAILBOX) {
+			MLX4_GET(size, outbox, QUERY_FUNC_CAP_EXTRA_FLAGS_OFFSET);
+			if (size & QUERY_FUNC_CAP_EXTRA_FLAGS_BF_QP_ALLOC_FLAG)
+				func_cap->extra_flags |= MLX4_QUERY_FUNC_FLAGS_BF_RES_QP;
+			if (size & QUERY_FUNC_CAP_EXTRA_FLAGS_A0_QP_ALLOC_FLAG)
+				func_cap->extra_flags |= MLX4_QUERY_FUNC_FLAGS_A0_RES_QP;
+		}
+
+		goto out;
+	}
+
+	/* logical port query */
+	if (gen_or_port > dev->caps.num_ports) {
+		err = -EINVAL;
+		goto out;
+	}
+
+	MLX4_GET(func_cap->flags1, outbox, QUERY_FUNC_CAP_FLAGS1_OFFSET);
+	if (dev->caps.port_type[gen_or_port] == MLX4_PORT_TYPE_ETH) {
+		if (func_cap->flags1 & QUERY_FUNC_CAP_FLAGS1_FORCE_VLAN) {
+			mlx4_err(dev, "VLAN is enforced on this port\n");
+			err = -EPROTONOSUPPORT;
+			goto out;
+		}
+
+		if (func_cap->flags1 & QUERY_FUNC_CAP_FLAGS1_FORCE_MAC) {
+			mlx4_err(dev, "Force mac is enabled on this port\n");
+			err = -EPROTONOSUPPORT;
+			goto out;
+		}
+	} else if (dev->caps.port_type[gen_or_port] == MLX4_PORT_TYPE_IB) {
+		MLX4_GET(field, outbox, QUERY_FUNC_CAP_FLAGS0_OFFSET);
+		if (field & QUERY_FUNC_CAP_FLAGS0_FORCE_PHY_WQE_GID) {
+			mlx4_err(dev, "phy_wqe_gid is enforced on this ib port\n");
+			err = -EPROTONOSUPPORT;
+			goto out;
+		}
+	}
+
+	MLX4_GET(field, outbox, QUERY_FUNC_CAP_PHYS_PORT_OFFSET);
+	func_cap->physical_port = field;
+	if (func_cap->physical_port != gen_or_port) {
+		err = -EINVAL;
+		goto out;
+	}
+
+	if (func_cap->flags1 & QUERY_FUNC_CAP_VF_ENABLE_QP0) {
+		MLX4_GET(qkey, outbox, QUERY_FUNC_CAP_PRIV_VF_QKEY_OFFSET);
+		func_cap->spec_qps.qp0_qkey = qkey;
+	} else {
+		func_cap->spec_qps.qp0_qkey = 0;
+	}
+
+	MLX4_GET(size, outbox, QUERY_FUNC_CAP_QP0_TUNNEL);
+	func_cap->spec_qps.qp0_tunnel = size & 0xFFFFFF;
+
+	MLX4_GET(size, outbox, QUERY_FUNC_CAP_QP0_PROXY);
+	func_cap->spec_qps.qp0_proxy = size & 0xFFFFFF;
+
+	MLX4_GET(size, outbox, QUERY_FUNC_CAP_QP1_TUNNEL);
+	func_cap->spec_qps.qp1_tunnel = size & 0xFFFFFF;
+
+	MLX4_GET(size, outbox, QUERY_FUNC_CAP_QP1_PROXY);
+	func_cap->spec_qps.qp1_proxy = size & 0xFFFFFF;
+
+	if (func_cap->flags1 & QUERY_FUNC_CAP_FLAGS1_NIC_INFO)
+		MLX4_GET(func_cap->phys_port_id, outbox,
+			 QUERY_FUNC_CAP_PHYS_PORT_ID);
+
+	MLX4_GET(func_cap->flags0, outbox, QUERY_FUNC_CAP_FLAGS0_OFFSET);
+
+	/* All other resources are allocated by the master, but we still report
+	 * 'num' and 'reserved' capabilities as follows:
+	 * - num remains the maximum resource index
+	 * - 'num - reserved' is the total available objects of a resource, but
+	 *   resource indices may be less than 'reserved'
+	 * TODO: set per-resource quotas */
+
+out:
+	mlx4_free_cmd_mailbox(dev, mailbox);
+
+	return err;
+}
+
+static void disable_unsupported_roce_caps(void *buf);
+
+int mlx4_QUERY_DEV_CAP(struct mlx4_dev *dev, struct mlx4_dev_cap *dev_cap)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	u32 *outbox;
+	u8 field;
+	u32 field32, flags, ext_flags;
+	u16 size;
+	u16 stat_rate;
+	int err;
+	int i;
+
+#define QUERY_DEV_CAP_OUT_SIZE		       0x100
+#define QUERY_DEV_CAP_MAX_SRQ_SZ_OFFSET		0x10
+#define QUERY_DEV_CAP_MAX_QP_SZ_OFFSET		0x11
+#define QUERY_DEV_CAP_RSVD_QP_OFFSET		0x12
+#define QUERY_DEV_CAP_MAX_QP_OFFSET		0x13
+#define QUERY_DEV_CAP_RSVD_SRQ_OFFSET		0x14
+#define QUERY_DEV_CAP_MAX_SRQ_OFFSET		0x15
+#define QUERY_DEV_CAP_RSVD_EEC_OFFSET		0x16
+#define QUERY_DEV_CAP_MAX_EEC_OFFSET		0x17
+#define QUERY_DEV_CAP_MAX_CQ_SZ_OFFSET		0x19
+#define QUERY_DEV_CAP_RSVD_CQ_OFFSET		0x1a
+#define QUERY_DEV_CAP_MAX_CQ_OFFSET		0x1b
+#define QUERY_DEV_CAP_MAX_MPT_OFFSET		0x1d
+#define QUERY_DEV_CAP_RSVD_EQ_OFFSET		0x1e
+#define QUERY_DEV_CAP_MAX_EQ_OFFSET		0x1f
+#define QUERY_DEV_CAP_RSVD_MTT_OFFSET		0x20
+#define QUERY_DEV_CAP_MAX_MRW_SZ_OFFSET		0x21
+#define QUERY_DEV_CAP_RSVD_MRW_OFFSET		0x22
+#define QUERY_DEV_CAP_MAX_MTT_SEG_OFFSET	0x23
+#define QUERY_DEV_CAP_NUM_SYS_EQ_OFFSET		0x26
+#define QUERY_DEV_CAP_MAX_AV_OFFSET		0x27
+#define QUERY_DEV_CAP_MAX_REQ_QP_OFFSET		0x29
+#define QUERY_DEV_CAP_MAX_RES_QP_OFFSET		0x2b
+#define QUERY_DEV_CAP_MAX_GSO_OFFSET		0x2d
+#define QUERY_DEV_CAP_RSS_OFFSET		0x2e
+#define QUERY_DEV_CAP_MAX_RDMA_OFFSET		0x2f
+#define QUERY_DEV_CAP_RSZ_SRQ_OFFSET		0x33
+#define QUERY_DEV_CAP_PORT_BEACON_OFFSET	0x34
+#define QUERY_DEV_CAP_ACK_DELAY_OFFSET		0x35
+#define QUERY_DEV_CAP_MTU_WIDTH_OFFSET		0x36
+#define QUERY_DEV_CAP_VL_PORT_OFFSET		0x37
+#define QUERY_DEV_CAP_MAX_MSG_SZ_OFFSET		0x38
+#define QUERY_DEV_CAP_MAX_GID_OFFSET		0x3b
+#define QUERY_DEV_CAP_RATE_SUPPORT_OFFSET	0x3c
+#define QUERY_DEV_CAP_CQ_TS_SUPPORT_OFFSET	0x3e
+#define QUERY_DEV_CAP_MAX_PKEY_OFFSET		0x3f
+#define QUERY_DEV_CAP_EXT_FLAGS_OFFSET		0x40
+#define QUERY_DEV_CAP_WOL_OFFSET		0x43
+#define QUERY_DEV_CAP_FLAGS_OFFSET		0x44
+#define QUERY_DEV_CAP_RSVD_UAR_OFFSET		0x48
+#define QUERY_DEV_CAP_UAR_SZ_OFFSET		0x49
+#define QUERY_DEV_CAP_PAGE_SZ_OFFSET		0x4b
+#define QUERY_DEV_CAP_BF_OFFSET			0x4c
+#define QUERY_DEV_CAP_LOG_BF_REG_SZ_OFFSET	0x4d
+#define QUERY_DEV_CAP_LOG_MAX_BF_REGS_PER_PAGE_OFFSET	0x4e
+#define QUERY_DEV_CAP_LOG_MAX_BF_PAGES_OFFSET	0x4f
+#define QUERY_DEV_CAP_MAX_SG_SQ_OFFSET		0x51
+#define QUERY_DEV_CAP_MAX_DESC_SZ_SQ_OFFSET	0x52
+#define QUERY_DEV_CAP_MAX_SG_RQ_OFFSET		0x55
+#define QUERY_DEV_CAP_MAX_DESC_SZ_RQ_OFFSET	0x56
+#define QUERY_DEV_CAP_USER_MAC_EN_OFFSET	0x5C
+#define QUERY_DEV_CAP_SVLAN_BY_QP_OFFSET	0x5D
+#define QUERY_DEV_CAP_MAX_QP_MCG_OFFSET		0x61
+#define QUERY_DEV_CAP_RSVD_MCG_OFFSET		0x62
+#define QUERY_DEV_CAP_MAX_MCG_OFFSET		0x63
+#define QUERY_DEV_CAP_RSVD_PD_OFFSET		0x64
+#define QUERY_DEV_CAP_MAX_PD_OFFSET		0x65
+#define QUERY_DEV_CAP_RSVD_XRC_OFFSET		0x66
+#define QUERY_DEV_CAP_MAX_XRC_OFFSET		0x67
+#define QUERY_DEV_CAP_MAX_COUNTERS_OFFSET	0x68
+#define QUERY_DEV_CAP_PORT_FLOWSTATS_COUNTERS_OFFSET	0x70
+#define QUERY_DEV_CAP_EXT_2_FLAGS_OFFSET	0x70
+#define QUERY_DEV_CAP_FLOW_STEERING_IPOIB_OFFSET	0x74
+#define QUERY_DEV_CAP_FLOW_STEERING_RANGE_EN_OFFSET	0x76
+#define QUERY_DEV_CAP_FLOW_STEERING_MAX_QP_OFFSET	0x77
+#define QUERY_DEV_CAP_SL2VL_EVENT_OFFSET	0x78
+#define QUERY_DEV_CAP_CQ_EQ_CACHE_LINE_STRIDE	0x7a
+#define QUERY_DEV_CAP_ECN_QCN_VER_OFFSET	0x7b
+#define QUERY_DEV_CAP_RDMARC_ENTRY_SZ_OFFSET	0x80
+#define QUERY_DEV_CAP_QPC_ENTRY_SZ_OFFSET	0x82
+#define QUERY_DEV_CAP_AUX_ENTRY_SZ_OFFSET	0x84
+#define QUERY_DEV_CAP_ALTC_ENTRY_SZ_OFFSET	0x86
+#define QUERY_DEV_CAP_EQC_ENTRY_SZ_OFFSET	0x88
+#define QUERY_DEV_CAP_CQC_ENTRY_SZ_OFFSET	0x8a
+#define QUERY_DEV_CAP_SRQ_ENTRY_SZ_OFFSET	0x8c
+#define QUERY_DEV_CAP_C_MPT_ENTRY_SZ_OFFSET	0x8e
+#define QUERY_DEV_CAP_MTT_ENTRY_SZ_OFFSET	0x90
+#define QUERY_DEV_CAP_D_MPT_ENTRY_SZ_OFFSET	0x92
+#define QUERY_DEV_CAP_BMME_FLAGS_OFFSET		0x94
+#define QUERY_DEV_CAP_CONFIG_DEV_OFFSET		0x94
+#define QUERY_DEV_CAP_PHV_EN_OFFSET		0x96
+#define QUERY_DEV_CAP_RSVD_LKEY_OFFSET		0x98
+#define QUERY_DEV_CAP_MAX_ICM_SZ_OFFSET		0xa0
+#define QUERY_DEV_CAP_ETH_BACKPL_OFFSET		0x9c
+#define QUERY_DEV_CAP_DIAG_RPRT_PER_PORT	0x9c
+#define QUERY_DEV_CAP_FW_REASSIGN_MAC		0x9d
+#define QUERY_DEV_CAP_VXLAN			0x9e
+#define QUERY_DEV_CAP_MAD_DEMUX_OFFSET		0xb0
+#define QUERY_DEV_CAP_DMFS_HIGH_RATE_QPN_BASE_OFFSET	0xa8
+#define QUERY_DEV_CAP_DMFS_HIGH_RATE_QPN_RANGE_OFFSET	0xac
+#define QUERY_DEV_CAP_MAP_CLOCK_TO_USER 0xc1
+#define QUERY_DEV_CAP_QP_RATE_LIMIT_NUM_OFFSET	0xcc
+#define QUERY_DEV_CAP_QP_RATE_LIMIT_MAX_OFFSET	0xd0
+#define QUERY_DEV_CAP_QP_RATE_LIMIT_MIN_OFFSET	0xd2
+#define QUERY_DEV_CAP_HEALTH_BUFFER_ADDRESS_OFFSET	0xe4
+
+	dev_cap->flags2 = 0;
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+	outbox = mailbox->buf;
+
+	err = mlx4_cmd_box(dev, 0, mailbox->dma, 0, 0, MLX4_CMD_QUERY_DEV_CAP,
+			   MLX4_CMD_TIME_CLASS_A, MLX4_CMD_NATIVE);
+	if (err)
+		goto out;
+
+	if (mlx4_is_mfunc(dev))
+		disable_unsupported_roce_caps(outbox);
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_MAP_CLOCK_TO_USER);
+	dev_cap->map_clock_to_user = field & 0x80;
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_RSVD_QP_OFFSET);
+	dev_cap->reserved_qps = 1 << (field & 0xf);
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_MAX_QP_OFFSET);
+	dev_cap->max_qps = 1 << (field & 0x1f);
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_RSVD_SRQ_OFFSET);
+	dev_cap->reserved_srqs = 1 << (field >> 4);
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_MAX_SRQ_OFFSET);
+	dev_cap->max_srqs = 1 << (field & 0x1f);
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_MAX_CQ_SZ_OFFSET);
+	dev_cap->max_cq_sz = 1 << field;
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_RSVD_CQ_OFFSET);
+	dev_cap->reserved_cqs = 1 << (field & 0xf);
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_MAX_CQ_OFFSET);
+	dev_cap->max_cqs = 1 << (field & 0x1f);
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_MAX_MPT_OFFSET);
+	dev_cap->max_mpts = 1 << (field & 0x3f);
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_RSVD_EQ_OFFSET);
+	dev_cap->reserved_eqs = 1 << (field & 0xf);
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_MAX_EQ_OFFSET);
+	dev_cap->max_eqs = 1 << (field & 0xf);
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_RSVD_MTT_OFFSET);
+	dev_cap->reserved_mtts = 1 << (field >> 4);
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_RSVD_MRW_OFFSET);
+	dev_cap->reserved_mrws = 1 << (field & 0xf);
+	MLX4_GET(size, outbox, QUERY_DEV_CAP_NUM_SYS_EQ_OFFSET);
+	dev_cap->num_sys_eqs = size & 0xfff;
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_MAX_REQ_QP_OFFSET);
+	dev_cap->max_requester_per_qp = 1 << (field & 0x3f);
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_MAX_RES_QP_OFFSET);
+	dev_cap->max_responder_per_qp = 1 << (field & 0x3f);
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_MAX_GSO_OFFSET);
+	field &= 0x1f;
+	if (!field)
+		dev_cap->max_gso_sz = 0;
+	else
+		dev_cap->max_gso_sz = 1 << field;
+
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_RSS_OFFSET);
+	if (field & 0x20)
+		dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_RSS_XOR;
+	if (field & 0x10)
+		dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_RSS_TOP;
+	field &= 0xf;
+	if (field) {
+		dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_RSS;
+		dev_cap->max_rss_tbl_sz = 1 << field;
+	} else
+		dev_cap->max_rss_tbl_sz = 0;
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_MAX_RDMA_OFFSET);
+	dev_cap->max_rdma_global = 1 << (field & 0x3f);
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_ACK_DELAY_OFFSET);
+	dev_cap->local_ca_ack_delay = field & 0x1f;
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_VL_PORT_OFFSET);
+	dev_cap->num_ports = field & 0xf;
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_MAX_MSG_SZ_OFFSET);
+	dev_cap->max_msg_sz = 1 << (field & 0x1f);
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_PORT_FLOWSTATS_COUNTERS_OFFSET);
+	if (field & 0x10)
+		dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_FLOWSTATS_EN;
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_FLOW_STEERING_RANGE_EN_OFFSET);
+	if (field & 0x80)
+		dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_FS_EN;
+	dev_cap->fs_log_max_ucast_qp_range_size = field & 0x1f;
+	if (field & 0x20)
+		dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_DMFS_UC_MC_SNIFFER;
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_PORT_BEACON_OFFSET);
+	if (field & 0x80)
+		dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_PORT_BEACON;
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_FLOW_STEERING_IPOIB_OFFSET);
+	if (field & 0x80)
+		dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_DMFS_IPOIB;
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_FLOW_STEERING_MAX_QP_OFFSET);
+	dev_cap->fs_max_num_qp_per_entry = field;
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_SL2VL_EVENT_OFFSET);
+	if (field & (1 << 5))
+		dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_SL_TO_VL_CHANGE_EVENT;
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_ECN_QCN_VER_OFFSET);
+	if (field & 0x1)
+		dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_QCN;
+	MLX4_GET(stat_rate, outbox, QUERY_DEV_CAP_RATE_SUPPORT_OFFSET);
+	dev_cap->stat_rate_support = stat_rate;
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_CQ_TS_SUPPORT_OFFSET);
+	if (field & 0x80)
+		dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_TS;
+	MLX4_GET(ext_flags, outbox, QUERY_DEV_CAP_EXT_FLAGS_OFFSET);
+	MLX4_GET(flags, outbox, QUERY_DEV_CAP_FLAGS_OFFSET);
+	dev_cap->flags = flags | (u64)ext_flags << 32;
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_WOL_OFFSET);
+	dev_cap->wol_port[1] = !!(field & 0x20);
+	dev_cap->wol_port[2] = !!(field & 0x40);
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_RSVD_UAR_OFFSET);
+	dev_cap->reserved_uars = field >> 4;
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_UAR_SZ_OFFSET);
+	dev_cap->uar_size = 1 << ((field & 0x3f) + 20);
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_PAGE_SZ_OFFSET);
+	dev_cap->min_page_sz = 1 << field;
+
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_BF_OFFSET);
+	if (field & 0x80) {
+		MLX4_GET(field, outbox, QUERY_DEV_CAP_LOG_BF_REG_SZ_OFFSET);
+		dev_cap->bf_reg_size = 1 << (field & 0x1f);
+		MLX4_GET(field, outbox, QUERY_DEV_CAP_LOG_MAX_BF_REGS_PER_PAGE_OFFSET);
+		if ((1 << (field & 0x3f)) > (PAGE_SIZE / dev_cap->bf_reg_size))
+			field = 3;
+		dev_cap->bf_regs_per_page = 1 << (field & 0x3f);
+	} else {
+		dev_cap->bf_reg_size = 0;
+	}
+
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_MAX_SG_SQ_OFFSET);
+	dev_cap->max_sq_sg = field;
+	MLX4_GET(size, outbox, QUERY_DEV_CAP_MAX_DESC_SZ_SQ_OFFSET);
+	dev_cap->max_sq_desc_sz = size;
+
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_USER_MAC_EN_OFFSET);
+	if (field & (1 << 2))
+		dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_USER_MAC_EN;
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_SVLAN_BY_QP_OFFSET);
+	if (field & 0x1)
+		dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_SVLAN_BY_QP;
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_MAX_QP_MCG_OFFSET);
+	dev_cap->max_qp_per_mcg = 1 << field;
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_RSVD_MCG_OFFSET);
+	dev_cap->reserved_mgms = field & 0xf;
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_MAX_MCG_OFFSET);
+	dev_cap->max_mcgs = 1 << field;
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_RSVD_PD_OFFSET);
+	dev_cap->reserved_pds = field >> 4;
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_MAX_PD_OFFSET);
+	dev_cap->max_pds = 1 << (field & 0x3f);
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_RSVD_XRC_OFFSET);
+	dev_cap->reserved_xrcds = field >> 4;
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_MAX_XRC_OFFSET);
+	dev_cap->max_xrcds = 1 << (field & 0x1f);
+
+	MLX4_GET(size, outbox, QUERY_DEV_CAP_RDMARC_ENTRY_SZ_OFFSET);
+	dev_cap->rdmarc_entry_sz = size;
+	MLX4_GET(size, outbox, QUERY_DEV_CAP_QPC_ENTRY_SZ_OFFSET);
+	dev_cap->qpc_entry_sz = size;
+	MLX4_GET(size, outbox, QUERY_DEV_CAP_AUX_ENTRY_SZ_OFFSET);
+	dev_cap->aux_entry_sz = size;
+	MLX4_GET(size, outbox, QUERY_DEV_CAP_ALTC_ENTRY_SZ_OFFSET);
+	dev_cap->altc_entry_sz = size;
+	MLX4_GET(size, outbox, QUERY_DEV_CAP_EQC_ENTRY_SZ_OFFSET);
+	dev_cap->eqc_entry_sz = size;
+	MLX4_GET(size, outbox, QUERY_DEV_CAP_CQC_ENTRY_SZ_OFFSET);
+	dev_cap->cqc_entry_sz = size;
+	MLX4_GET(size, outbox, QUERY_DEV_CAP_SRQ_ENTRY_SZ_OFFSET);
+	dev_cap->srq_entry_sz = size;
+	MLX4_GET(size, outbox, QUERY_DEV_CAP_C_MPT_ENTRY_SZ_OFFSET);
+	dev_cap->cmpt_entry_sz = size;
+	MLX4_GET(size, outbox, QUERY_DEV_CAP_MTT_ENTRY_SZ_OFFSET);
+	dev_cap->mtt_entry_sz = size;
+	MLX4_GET(size, outbox, QUERY_DEV_CAP_D_MPT_ENTRY_SZ_OFFSET);
+	dev_cap->dmpt_entry_sz = size;
+
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_MAX_SRQ_SZ_OFFSET);
+	dev_cap->max_srq_sz = 1 << field;
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_MAX_QP_SZ_OFFSET);
+	dev_cap->max_qp_sz = 1 << field;
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_RSZ_SRQ_OFFSET);
+	dev_cap->resize_srq = field & 1;
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_MAX_SG_RQ_OFFSET);
+	dev_cap->max_rq_sg = field;
+	MLX4_GET(size, outbox, QUERY_DEV_CAP_MAX_DESC_SZ_RQ_OFFSET);
+	dev_cap->max_rq_desc_sz = size;
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_CQ_EQ_CACHE_LINE_STRIDE);
+	if (field & (1 << 4))
+		dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_QOS_VPP;
+	if (field & (1 << 5))
+		dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_ETH_PROT_CTRL;
+	if (field & (1 << 6))
+		dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_CQE_STRIDE;
+	if (field & (1 << 7))
+		dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_EQE_STRIDE;
+	MLX4_GET(dev_cap->bmme_flags, outbox,
+		 QUERY_DEV_CAP_BMME_FLAGS_OFFSET);
+	if (dev_cap->bmme_flags & MLX4_FLAG_ROCE_V1_V2)
+		dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_ROCE_V1_V2;
+	if (dev_cap->bmme_flags & MLX4_FLAG_PORT_REMAP)
+		dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_PORT_REMAP;
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_CONFIG_DEV_OFFSET);
+	if (field & 0x20)
+		dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_CONFIG_DEV;
+	if (field & (1 << 2))
+		dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_IGNORE_FCS;
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_PHV_EN_OFFSET);
+	if (field & 0x80)
+		dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_PHV_EN;
+	if (field & 0x40)
+		dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_SKIP_OUTER_VLAN;
+
+	MLX4_GET(dev_cap->reserved_lkey, outbox,
+		 QUERY_DEV_CAP_RSVD_LKEY_OFFSET);
+	MLX4_GET(field32, outbox, QUERY_DEV_CAP_ETH_BACKPL_OFFSET);
+	if (field32 & (1 << 0))
+		dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_ETH_BACKPL_AN_REP;
+	if (field32 & (1 << 7))
+		dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_RECOVERABLE_ERROR_EVENT;
+	if (field32 & (1 << 8))
+		dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_DRIVER_VERSION_TO_FW;
+	MLX4_GET(field32, outbox, QUERY_DEV_CAP_DIAG_RPRT_PER_PORT);
+	if (field32 & (1 << 17))
+		dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_DIAG_PER_PORT;
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_FW_REASSIGN_MAC);
+	if (field & 1<<6)
+		dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_REASSIGN_MAC_EN;
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_VXLAN);
+	if (field & 1<<3)
+		dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_VXLAN_OFFLOADS;
+	if (field & (1 << 5))
+		dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_ETS_CFG;
+	MLX4_GET(dev_cap->max_icm_sz, outbox,
+		 QUERY_DEV_CAP_MAX_ICM_SZ_OFFSET);
+	if (dev_cap->flags & MLX4_DEV_CAP_FLAG_COUNTERS)
+		MLX4_GET(dev_cap->max_counters, outbox,
+			 QUERY_DEV_CAP_MAX_COUNTERS_OFFSET);
+
+	MLX4_GET(field32, outbox,
+		 QUERY_DEV_CAP_MAD_DEMUX_OFFSET);
+	if (field32 & (1 << 0))
+		dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_MAD_DEMUX;
+
+	MLX4_GET(dev_cap->dmfs_high_rate_qpn_base, outbox,
+		 QUERY_DEV_CAP_DMFS_HIGH_RATE_QPN_BASE_OFFSET);
+	dev_cap->dmfs_high_rate_qpn_base &= MGM_QPN_MASK;
+	MLX4_GET(dev_cap->dmfs_high_rate_qpn_range, outbox,
+		 QUERY_DEV_CAP_DMFS_HIGH_RATE_QPN_RANGE_OFFSET);
+	dev_cap->dmfs_high_rate_qpn_range &= MGM_QPN_MASK;
+
+	MLX4_GET(size, outbox, QUERY_DEV_CAP_QP_RATE_LIMIT_NUM_OFFSET);
+	dev_cap->rl_caps.num_rates = size;
+	if (dev_cap->rl_caps.num_rates) {
+		dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_QP_RATE_LIMIT;
+		MLX4_GET(size, outbox, QUERY_DEV_CAP_QP_RATE_LIMIT_MAX_OFFSET);
+		dev_cap->rl_caps.max_val  = size & 0xfff;
+		dev_cap->rl_caps.max_unit = size >> 14;
+		MLX4_GET(size, outbox, QUERY_DEV_CAP_QP_RATE_LIMIT_MIN_OFFSET);
+		dev_cap->rl_caps.min_val  = size & 0xfff;
+		dev_cap->rl_caps.min_unit = size >> 14;
+	}
+
+	MLX4_GET(dev_cap->health_buffer_addrs, outbox,
+		 QUERY_DEV_CAP_HEALTH_BUFFER_ADDRESS_OFFSET);
+
+	MLX4_GET(field32, outbox, QUERY_DEV_CAP_EXT_2_FLAGS_OFFSET);
+	if (field32 & (1 << 16))
+		dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_UPDATE_QP;
+	if (field32 & (1 << 18))
+		dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_UPDATE_QP_SRC_CHECK_LB;
+	if (field32 & (1 << 19))
+		dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_LB_SRC_CHK;
+	if (field32 & (1 << 26))
+		dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_VLAN_CONTROL;
+	if (field32 & (1 << 20))
+		dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_FSM;
+	if (field32 & (1 << 21))
+		dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_80_VFS;
+	if (field32 & (1 << 23))
+		dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_SW_CQ_INIT;
+
+	for (i = 1; i <= dev_cap->num_ports; i++) {
+		err = mlx4_QUERY_PORT(dev, i, dev_cap->port_cap + i);
+		if (err)
+			goto out;
+	}
+
+	/*
+	 * Each UAR has 4 EQ doorbells; so if a UAR is reserved, then
+	 * we can't use any EQs whose doorbell falls on that page,
+	 * even if the EQ itself isn't reserved.
+	 */
+	if (dev_cap->num_sys_eqs == 0)
+		dev_cap->reserved_eqs = max(dev_cap->reserved_uars * 4,
+					    dev_cap->reserved_eqs);
+	else
+		dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_SYS_EQS;
+
+out:
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return err;
+}
+
+void mlx4_dev_cap_dump(struct mlx4_dev *dev, struct mlx4_dev_cap *dev_cap)
+{
+	if (dev_cap->bf_reg_size > 0)
+		mlx4_dbg(dev, "BlueFlame available (reg size %d, regs/page %d)\n",
+			 dev_cap->bf_reg_size, dev_cap->bf_regs_per_page);
+	else
+		mlx4_dbg(dev, "BlueFlame not available\n");
+
+	mlx4_dbg(dev, "Base MM extensions: flags %08x, rsvd L_Key %08x\n",
+		 dev_cap->bmme_flags, dev_cap->reserved_lkey);
+	mlx4_dbg(dev, "Max ICM size %lld MB\n",
+		 (unsigned long long) dev_cap->max_icm_sz >> 20);
+	mlx4_dbg(dev, "Max QPs: %d, reserved QPs: %d, entry size: %d\n",
+		 dev_cap->max_qps, dev_cap->reserved_qps, dev_cap->qpc_entry_sz);
+	mlx4_dbg(dev, "Max SRQs: %d, reserved SRQs: %d, entry size: %d\n",
+		 dev_cap->max_srqs, dev_cap->reserved_srqs, dev_cap->srq_entry_sz);
+	mlx4_dbg(dev, "Max CQs: %d, reserved CQs: %d, entry size: %d\n",
+		 dev_cap->max_cqs, dev_cap->reserved_cqs, dev_cap->cqc_entry_sz);
+	mlx4_dbg(dev, "Num sys EQs: %d, max EQs: %d, reserved EQs: %d, entry size: %d\n",
+		 dev_cap->num_sys_eqs, dev_cap->max_eqs, dev_cap->reserved_eqs,
+		 dev_cap->eqc_entry_sz);
+	mlx4_dbg(dev, "reserved MPTs: %d, reserved MTTs: %d\n",
+		 dev_cap->reserved_mrws, dev_cap->reserved_mtts);
+	mlx4_dbg(dev, "Max PDs: %d, reserved PDs: %d, reserved UARs: %d\n",
+		 dev_cap->max_pds, dev_cap->reserved_pds, dev_cap->reserved_uars);
+	mlx4_dbg(dev, "Max QP/MCG: %d, reserved MGMs: %d\n",
+		 dev_cap->max_pds, dev_cap->reserved_mgms);
+	mlx4_dbg(dev, "Max CQEs: %d, max WQEs: %d, max SRQ WQEs: %d\n",
+		 dev_cap->max_cq_sz, dev_cap->max_qp_sz, dev_cap->max_srq_sz);
+	mlx4_dbg(dev, "Local CA ACK delay: %d, max MTU: %d, port width cap: %d\n",
+		 dev_cap->local_ca_ack_delay, 128 << dev_cap->port_cap[1].ib_mtu,
+		 dev_cap->port_cap[1].max_port_width);
+	mlx4_dbg(dev, "Max SQ desc size: %d, max SQ S/G: %d\n",
+		 dev_cap->max_sq_desc_sz, dev_cap->max_sq_sg);
+	mlx4_dbg(dev, "Max RQ desc size: %d, max RQ S/G: %d\n",
+		 dev_cap->max_rq_desc_sz, dev_cap->max_rq_sg);
+	mlx4_dbg(dev, "Max GSO size: %d\n", dev_cap->max_gso_sz);
+	mlx4_dbg(dev, "Max counters: %d\n", dev_cap->max_counters);
+	mlx4_dbg(dev, "Max RSS Table size: %d\n", dev_cap->max_rss_tbl_sz);
+	mlx4_dbg(dev, "DMFS high rate steer QPn base: %d\n",
+		 dev_cap->dmfs_high_rate_qpn_base);
+	mlx4_dbg(dev, "DMFS high rate steer QPn range: %d\n",
+		 dev_cap->dmfs_high_rate_qpn_range);
+
+	if (dev_cap->flags2 & MLX4_DEV_CAP_FLAG2_QP_RATE_LIMIT) {
+		struct mlx4_rate_limit_caps *rl_caps = &dev_cap->rl_caps;
+
+		mlx4_dbg(dev, "QP Rate-Limit: #rates %d, unit/val max %d/%d, min %d/%d\n",
+			 rl_caps->num_rates, rl_caps->max_unit, rl_caps->max_val,
+			 rl_caps->min_unit, rl_caps->min_val);
+	}
+
+	dump_dev_cap_flags(dev, dev_cap->flags);
+	dump_dev_cap_flags2(dev, dev_cap->flags2);
+}
+
+int mlx4_QUERY_PORT(struct mlx4_dev *dev, int port, struct mlx4_port_cap *port_cap)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	u32 *outbox;
+	u8 field;
+	u32 field32;
+	int err;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+	outbox = mailbox->buf;
+
+	if (dev->flags & MLX4_FLAG_OLD_PORT_CMDS) {
+		err = mlx4_cmd_box(dev, 0, mailbox->dma, 0, 0, MLX4_CMD_QUERY_DEV_CAP,
+				   MLX4_CMD_TIME_CLASS_A,
+				   MLX4_CMD_NATIVE);
+
+		if (err)
+			goto out;
+
+		MLX4_GET(field, outbox, QUERY_DEV_CAP_VL_PORT_OFFSET);
+		port_cap->max_vl	   = field >> 4;
+		MLX4_GET(field, outbox, QUERY_DEV_CAP_MTU_WIDTH_OFFSET);
+		port_cap->ib_mtu	   = field >> 4;
+		port_cap->max_port_width = field & 0xf;
+		MLX4_GET(field, outbox, QUERY_DEV_CAP_MAX_GID_OFFSET);
+		port_cap->max_gids	   = 1 << (field & 0xf);
+		MLX4_GET(field, outbox, QUERY_DEV_CAP_MAX_PKEY_OFFSET);
+		port_cap->max_pkeys	   = 1 << (field & 0xf);
+	} else {
+#define QUERY_PORT_SUPPORTED_TYPE_OFFSET	0x00
+#define QUERY_PORT_MTU_OFFSET			0x01
+#define QUERY_PORT_ETH_MTU_OFFSET		0x02
+#define QUERY_PORT_WIDTH_OFFSET			0x06
+#define QUERY_PORT_MAX_GID_PKEY_OFFSET		0x07
+#define QUERY_PORT_MAX_MACVLAN_OFFSET		0x0a
+#define QUERY_PORT_MAX_VL_OFFSET		0x0b
+#define QUERY_PORT_MAC_OFFSET			0x10
+#define QUERY_PORT_TRANS_VENDOR_OFFSET		0x18
+#define QUERY_PORT_WAVELENGTH_OFFSET		0x1c
+#define QUERY_PORT_TRANS_CODE_OFFSET		0x20
+
+		err = mlx4_cmd_box(dev, 0, mailbox->dma, port, 0, MLX4_CMD_QUERY_PORT,
+				   MLX4_CMD_TIME_CLASS_B, MLX4_CMD_NATIVE);
+		if (err)
+			goto out;
+
+		MLX4_GET(field, outbox, QUERY_PORT_SUPPORTED_TYPE_OFFSET);
+		port_cap->link_state = (field & 0x80) >> 7;
+		port_cap->supported_port_types = field & 3;
+		port_cap->suggested_type = (field >> 3) & 1;
+		port_cap->default_sense = (field >> 4) & 1;
+		port_cap->dmfs_optimized_state = (field >> 5) & 1;
+		MLX4_GET(field, outbox, QUERY_PORT_MTU_OFFSET);
+		port_cap->ib_mtu	   = field & 0xf;
+		MLX4_GET(field, outbox, QUERY_PORT_WIDTH_OFFSET);
+		port_cap->max_port_width = field & 0xf;
+		MLX4_GET(field, outbox, QUERY_PORT_MAX_GID_PKEY_OFFSET);
+		port_cap->max_gids	   = 1 << (field >> 4);
+		port_cap->max_pkeys	   = 1 << (field & 0xf);
+		MLX4_GET(field, outbox, QUERY_PORT_MAX_VL_OFFSET);
+		port_cap->max_vl	   = field & 0xf;
+		port_cap->max_tc_eth	   = field >> 4;
+		MLX4_GET(field, outbox, QUERY_PORT_MAX_MACVLAN_OFFSET);
+		port_cap->log_max_macs  = field & 0xf;
+		port_cap->log_max_vlans = field >> 4;
+		MLX4_GET(port_cap->eth_mtu, outbox, QUERY_PORT_ETH_MTU_OFFSET);
+		MLX4_GET(port_cap->def_mac, outbox, QUERY_PORT_MAC_OFFSET);
+		MLX4_GET(field32, outbox, QUERY_PORT_TRANS_VENDOR_OFFSET);
+		port_cap->trans_type = field32 >> 24;
+		port_cap->vendor_oui = field32 & 0xffffff;
+		MLX4_GET(port_cap->wavelength, outbox, QUERY_PORT_WAVELENGTH_OFFSET);
+		MLX4_GET(port_cap->trans_code, outbox, QUERY_PORT_TRANS_CODE_OFFSET);
+	}
+
+out:
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return err;
+}
+
+#define DEV_CAP_EXT_2_FLAG_PFC_COUNTERS	(1 << 28)
+#define DEV_CAP_EXT_2_FLAG_VLAN_CONTROL (1 << 26)
+#define DEV_CAP_EXT_2_FLAG_80_VFS	(1 << 21)
+#define DEV_CAP_EXT_2_FLAG_FSM		(1 << 20)
+
+int mlx4_QUERY_DEV_CAP_wrapper(struct mlx4_dev *dev, int slave,
+			       struct mlx4_vhcr *vhcr,
+			       struct mlx4_cmd_mailbox *inbox,
+			       struct mlx4_cmd_mailbox *outbox,
+			       struct mlx4_cmd_info *cmd)
+{
+	u64	flags;
+	int	err = 0;
+	u8	field;
+	u16	field16;
+	u32	bmme_flags, field32;
+	int	real_port;
+	int	slave_port;
+	int	first_port;
+	struct mlx4_active_ports actv_ports;
+
+	err = mlx4_cmd_box(dev, 0, outbox->dma, 0, 0, MLX4_CMD_QUERY_DEV_CAP,
+			   MLX4_CMD_TIME_CLASS_A, MLX4_CMD_NATIVE);
+	if (err)
+		return err;
+
+	disable_unsupported_roce_caps(outbox->buf);
+	/* add port mng change event capability and disable mw type 1
+	 * unconditionally to slaves
+	 */
+	MLX4_GET(flags, outbox->buf, QUERY_DEV_CAP_EXT_FLAGS_OFFSET);
+	flags |= MLX4_DEV_CAP_FLAG_PORT_MNG_CHG_EV;
+	flags &= ~MLX4_DEV_CAP_FLAG_MEM_WINDOW;
+	actv_ports = mlx4_get_active_ports(dev, slave);
+	first_port = find_first_bit(actv_ports.ports, dev->caps.num_ports);
+	for (slave_port = 0, real_port = first_port;
+	     real_port < first_port +
+	     bitmap_weight(actv_ports.ports, dev->caps.num_ports);
+	     ++real_port, ++slave_port) {
+		if (flags & (MLX4_DEV_CAP_FLAG_WOL_PORT1 << real_port))
+			flags |= MLX4_DEV_CAP_FLAG_WOL_PORT1 << slave_port;
+		else
+			flags &= ~(MLX4_DEV_CAP_FLAG_WOL_PORT1 << slave_port);
+	}
+	for (; slave_port < dev->caps.num_ports; ++slave_port)
+		flags &= ~(MLX4_DEV_CAP_FLAG_WOL_PORT1 << slave_port);
+
+	/* Not exposing RSS IP fragments to guests */
+	flags &= ~MLX4_DEV_CAP_FLAG_RSS_IP_FRAG;
+	MLX4_PUT(outbox->buf, flags, QUERY_DEV_CAP_EXT_FLAGS_OFFSET);
+
+	MLX4_GET(field, outbox->buf, QUERY_DEV_CAP_VL_PORT_OFFSET);
+	field &= ~0x0F;
+	field |= bitmap_weight(actv_ports.ports, dev->caps.num_ports) & 0x0F;
+	MLX4_PUT(outbox->buf, field, QUERY_DEV_CAP_VL_PORT_OFFSET);
+
+	/* For guests, disable timestamp */
+	MLX4_GET(field, outbox->buf, QUERY_DEV_CAP_CQ_TS_SUPPORT_OFFSET);
+	field &= 0x7f;
+	MLX4_PUT(outbox->buf, field, QUERY_DEV_CAP_CQ_TS_SUPPORT_OFFSET);
+
+	/* For guests, disable vxlan tunneling and QoS support */
+	MLX4_GET(field, outbox->buf, QUERY_DEV_CAP_VXLAN);
+	field &= 0xd7;
+	MLX4_PUT(outbox->buf, field, QUERY_DEV_CAP_VXLAN);
+
+	/* For guests, disable port BEACON */
+	MLX4_GET(field, outbox->buf, QUERY_DEV_CAP_PORT_BEACON_OFFSET);
+	field &= 0x7f;
+	MLX4_PUT(outbox->buf, field, QUERY_DEV_CAP_PORT_BEACON_OFFSET);
+
+	/* For guests, report Blueflame disabled */
+	MLX4_GET(field, outbox->buf, QUERY_DEV_CAP_BF_OFFSET);
+	field &= 0x7f;
+	MLX4_PUT(outbox->buf, field, QUERY_DEV_CAP_BF_OFFSET);
+
+	/* For guests, disable mw type 2 and port remap*/
+	MLX4_GET(bmme_flags, outbox->buf, QUERY_DEV_CAP_BMME_FLAGS_OFFSET);
+	bmme_flags &= ~MLX4_BMME_FLAG_TYPE_2_WIN;
+	bmme_flags &= ~MLX4_FLAG_PORT_REMAP;
+	MLX4_PUT(outbox->buf, bmme_flags, QUERY_DEV_CAP_BMME_FLAGS_OFFSET);
+
+	/* turn off device-managed steering capability if not enabled */
+	if (dev->caps.steering_mode != MLX4_STEERING_MODE_DEVICE_MANAGED) {
+		MLX4_GET(field, outbox->buf,
+			 QUERY_DEV_CAP_FLOW_STEERING_RANGE_EN_OFFSET);
+		field &= 0x7f;
+		MLX4_PUT(outbox->buf, field,
+			 QUERY_DEV_CAP_FLOW_STEERING_RANGE_EN_OFFSET);
+	}
+
+	/* turn off ipoib managed steering for guests */
+	MLX4_GET(field, outbox->buf, QUERY_DEV_CAP_FLOW_STEERING_IPOIB_OFFSET);
+	field &= ~0x80;
+	MLX4_PUT(outbox->buf, field, QUERY_DEV_CAP_FLOW_STEERING_IPOIB_OFFSET);
+
+	/* turn off host side virt features (VST, FSM, etc) for guests */
+	MLX4_GET(field32, outbox->buf, QUERY_DEV_CAP_EXT_2_FLAGS_OFFSET);
+	field32 &= ~(DEV_CAP_EXT_2_FLAG_VLAN_CONTROL | DEV_CAP_EXT_2_FLAG_80_VFS |
+		     DEV_CAP_EXT_2_FLAG_FSM | DEV_CAP_EXT_2_FLAG_PFC_COUNTERS);
+	MLX4_PUT(outbox->buf, field32, QUERY_DEV_CAP_EXT_2_FLAGS_OFFSET);
+
+	/* turn off QCN for guests */
+	MLX4_GET(field, outbox->buf, QUERY_DEV_CAP_ECN_QCN_VER_OFFSET);
+	field &= 0xfe;
+	MLX4_PUT(outbox->buf, field, QUERY_DEV_CAP_ECN_QCN_VER_OFFSET);
+
+	/* turn off QP max-rate limiting for guests */
+	field16 = 0;
+	MLX4_PUT(outbox->buf, field16, QUERY_DEV_CAP_QP_RATE_LIMIT_NUM_OFFSET);
+
+	/* turn off QoS per VF support for guests */
+	MLX4_GET(field, outbox->buf, QUERY_DEV_CAP_CQ_EQ_CACHE_LINE_STRIDE);
+	field &= 0xef;
+	MLX4_PUT(outbox->buf, field, QUERY_DEV_CAP_CQ_EQ_CACHE_LINE_STRIDE);
+
+	/* turn off ignore FCS feature for guests */
+	MLX4_GET(field, outbox->buf, QUERY_DEV_CAP_CONFIG_DEV_OFFSET);
+	field &= 0xfb;
+	MLX4_PUT(outbox->buf, field, QUERY_DEV_CAP_CONFIG_DEV_OFFSET);
+
+	return 0;
+}
+
+static void disable_unsupported_roce_caps(void *buf)
+{
+	u32 flags;
+
+	MLX4_GET(flags, buf, QUERY_DEV_CAP_EXT_FLAGS_OFFSET);
+	flags &= ~(1UL << 31);
+	MLX4_PUT(buf, flags, QUERY_DEV_CAP_EXT_FLAGS_OFFSET);
+	MLX4_GET(flags, buf, QUERY_DEV_CAP_EXT_2_FLAGS_OFFSET);
+	flags &= ~(1UL << 24);
+	MLX4_PUT(buf, flags, QUERY_DEV_CAP_EXT_2_FLAGS_OFFSET);
+	MLX4_GET(flags, buf, QUERY_DEV_CAP_BMME_FLAGS_OFFSET);
+	flags &= ~(MLX4_FLAG_ROCE_V1_V2);
+	MLX4_PUT(buf, flags, QUERY_DEV_CAP_BMME_FLAGS_OFFSET);
+}
+
+int mlx4_QUERY_PORT_wrapper(struct mlx4_dev *dev, int slave,
+			    struct mlx4_vhcr *vhcr,
+			    struct mlx4_cmd_mailbox *inbox,
+			    struct mlx4_cmd_mailbox *outbox,
+			    struct mlx4_cmd_info *cmd)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	u64 def_mac;
+	u8 port_type;
+	u16 short_field;
+	int err;
+	int admin_link_state;
+	int port = mlx4_slave_convert_port(dev, slave,
+					   vhcr->in_modifier & 0xFF);
+
+#define MLX4_VF_PORT_NO_LINK_SENSE_MASK	0xE0
+#define MLX4_PORT_LINK_UP_MASK		0x80
+#define QUERY_PORT_CUR_MAX_PKEY_OFFSET	0x0c
+#define QUERY_PORT_CUR_MAX_GID_OFFSET	0x0e
+
+	if (port < 0)
+		return -EINVAL;
+
+	/* Protect against untrusted guests: enforce that this is the
+	 * QUERY_PORT general query.
+	 */
+	if (vhcr->op_modifier || vhcr->in_modifier & ~0xFF)
+		return -EINVAL;
+
+	vhcr->in_modifier = port;
+
+	err = mlx4_cmd_box(dev, 0, outbox->dma, vhcr->in_modifier, 0,
+			   MLX4_CMD_QUERY_PORT, MLX4_CMD_TIME_CLASS_B,
+			   MLX4_CMD_NATIVE);
+
+	if (!err && dev->caps.function != slave) {
+		def_mac = priv->mfunc.master.vf_oper[slave].vport[vhcr->in_modifier].state.mac;
+		MLX4_PUT(outbox->buf, def_mac, QUERY_PORT_MAC_OFFSET);
+
+		/* get port type - currently only eth is enabled */
+		MLX4_GET(port_type, outbox->buf,
+			 QUERY_PORT_SUPPORTED_TYPE_OFFSET);
+
+		/* No link sensing allowed */
+		port_type &= MLX4_VF_PORT_NO_LINK_SENSE_MASK;
+		/* set port type to currently operating port type */
+		port_type |= (dev->caps.port_type[vhcr->in_modifier] & 0x3);
+
+		admin_link_state = priv->mfunc.master.vf_oper[slave].vport[vhcr->in_modifier].state.link_state;
+		if (IFLA_VF_LINK_STATE_ENABLE == admin_link_state)
+			port_type |= MLX4_PORT_LINK_UP_MASK;
+		else if (IFLA_VF_LINK_STATE_DISABLE == admin_link_state)
+			port_type &= ~MLX4_PORT_LINK_UP_MASK;
+		else if (IFLA_VF_LINK_STATE_AUTO == admin_link_state && mlx4_is_bonded(dev)) {
+			int other_port = (port == 1) ? 2 : 1;
+			struct mlx4_port_cap port_cap;
+
+			err = mlx4_QUERY_PORT(dev, other_port, &port_cap);
+			if (err)
+				goto out;
+			port_type |= (port_cap.link_state << 7);
+		}
+
+		MLX4_PUT(outbox->buf, port_type,
+			 QUERY_PORT_SUPPORTED_TYPE_OFFSET);
+
+		if (dev->caps.port_type[vhcr->in_modifier] == MLX4_PORT_TYPE_ETH)
+			short_field = mlx4_get_slave_num_gids(dev, slave, port);
+		else
+			short_field = 1; /* slave max gids */
+		MLX4_PUT(outbox->buf, short_field,
+			 QUERY_PORT_CUR_MAX_GID_OFFSET);
+
+		short_field = dev->caps.pkey_table_len[vhcr->in_modifier];
+		MLX4_PUT(outbox->buf, short_field,
+			 QUERY_PORT_CUR_MAX_PKEY_OFFSET);
+	}
+out:
+	return err;
+}
+
+int mlx4_get_slave_pkey_gid_tbl_len(struct mlx4_dev *dev, u8 port,
+				    int *gid_tbl_len, int *pkey_tbl_len)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	u32			*outbox;
+	u16			field;
+	int			err;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+
+	err =  mlx4_cmd_box(dev, 0, mailbox->dma, port, 0,
+			    MLX4_CMD_QUERY_PORT, MLX4_CMD_TIME_CLASS_B,
+			    MLX4_CMD_WRAPPED);
+	if (err)
+		goto out;
+
+	outbox = mailbox->buf;
+
+	MLX4_GET(field, outbox, QUERY_PORT_CUR_MAX_GID_OFFSET);
+	*gid_tbl_len = field;
+
+	MLX4_GET(field, outbox, QUERY_PORT_CUR_MAX_PKEY_OFFSET);
+	*pkey_tbl_len = field;
+
+out:
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return err;
+}
+EXPORT_SYMBOL(mlx4_get_slave_pkey_gid_tbl_len);
+
+int mlx4_map_cmd(struct mlx4_dev *dev, u16 op, struct mlx4_icm *icm, u64 virt)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	struct mlx4_icm_iter iter;
+	__be64 *pages;
+	int lg;
+	int nent = 0;
+	int i;
+	int err = 0;
+	int ts = 0, tc = 0;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+	pages = mailbox->buf;
+
+	for (mlx4_icm_first(icm, &iter);
+	     !mlx4_icm_last(&iter);
+	     mlx4_icm_next(&iter)) {
+		/*
+		 * We have to pass pages that are aligned to their
+		 * size, so find the least significant 1 in the
+		 * address or size and use that as our log2 size.
+		 */
+		lg = ffs(mlx4_icm_addr(&iter) | mlx4_icm_size(&iter)) - 1;
+		if (lg < MLX4_ICM_PAGE_SHIFT) {
+			mlx4_warn(dev, "Got FW area not aligned to %d (%llx/%lx)\n",
+				  MLX4_ICM_PAGE_SIZE,
+				  (unsigned long long) mlx4_icm_addr(&iter),
+				  mlx4_icm_size(&iter));
+			err = -EINVAL;
+			goto out;
+		}
+
+		for (i = 0; i < mlx4_icm_size(&iter) >> lg; ++i) {
+			if (virt != -1) {
+				pages[nent * 2] = cpu_to_be64(virt);
+				virt += 1ULL << lg;
+			}
+
+			pages[nent * 2 + 1] =
+				cpu_to_be64((mlx4_icm_addr(&iter) + (i << lg)) |
+					    (lg - MLX4_ICM_PAGE_SHIFT));
+			ts += 1 << (lg - 10);
+			++tc;
+
+			if (++nent == MLX4_MAILBOX_SIZE / 16) {
+				err = mlx4_cmd(dev, mailbox->dma, nent, 0, op,
+						MLX4_CMD_TIME_CLASS_B,
+						MLX4_CMD_NATIVE);
+				if (err)
+					goto out;
+				nent = 0;
+			}
+		}
+	}
+
+	if (nent)
+		err = mlx4_cmd(dev, mailbox->dma, nent, 0, op,
+			       MLX4_CMD_TIME_CLASS_B, MLX4_CMD_NATIVE);
+	if (err)
+		goto out;
+
+	switch (op) {
+	case MLX4_CMD_MAP_FA:
+		mlx4_dbg(dev, "Mapped %d chunks/%d KB for FW\n", tc, ts);
+		break;
+	case MLX4_CMD_MAP_ICM_AUX:
+		mlx4_dbg(dev, "Mapped %d chunks/%d KB for ICM aux\n", tc, ts);
+		break;
+	case MLX4_CMD_MAP_ICM:
+		mlx4_dbg(dev, "Mapped %d chunks/%d KB at %llx for ICM\n",
+			 tc, ts, (unsigned long long) virt - (ts << 10));
+		break;
+	}
+
+out:
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return err;
+}
+
+int mlx4_MAP_FA(struct mlx4_dev *dev, struct mlx4_icm *icm)
+{
+	return mlx4_map_cmd(dev, MLX4_CMD_MAP_FA, icm, -1);
+}
+
+int mlx4_UNMAP_FA(struct mlx4_dev *dev)
+{
+	return mlx4_cmd(dev, 0, 0, 0, MLX4_CMD_UNMAP_FA,
+			MLX4_CMD_TIME_CLASS_B, MLX4_CMD_NATIVE);
+}
+
+
+int mlx4_RUN_FW(struct mlx4_dev *dev)
+{
+	return mlx4_cmd(dev, 0, 0, 0, MLX4_CMD_RUN_FW,
+			MLX4_CMD_TIME_CLASS_A, MLX4_CMD_NATIVE);
+}
+
+int mlx4_QUERY_FW(struct mlx4_dev *dev)
+{
+	struct mlx4_fw  *fw  = &mlx4_priv(dev)->fw;
+	struct mlx4_cmd *cmd = &mlx4_priv(dev)->cmd;
+	struct mlx4_cmd_mailbox *mailbox;
+	u32 *outbox;
+	int err = 0;
+	u64 fw_ver;
+	u16 cmd_if_rev;
+	u8 lg;
+
+#define QUERY_FW_OUT_SIZE             0x100
+#define QUERY_FW_VER_OFFSET            0x00
+#define QUERY_FW_PPF_ID		       0x09
+#define QUERY_FW_CMD_IF_REV_OFFSET     0x0a
+#define QUERY_FW_MAX_CMD_OFFSET        0x0f
+#define QUERY_FW_ERR_START_OFFSET      0x30
+#define QUERY_FW_ERR_SIZE_OFFSET       0x38
+#define QUERY_FW_ERR_BAR_OFFSET        0x3c
+
+#define QUERY_FW_SIZE_OFFSET           0x00
+#define QUERY_FW_CLR_INT_BASE_OFFSET   0x20
+#define QUERY_FW_CLR_INT_BAR_OFFSET    0x28
+
+#define QUERY_FW_COMM_BASE_OFFSET      0x40
+#define QUERY_FW_COMM_BAR_OFFSET       0x48
+
+#define QUERY_FW_CLOCK_OFFSET	       0x50
+#define QUERY_FW_CLOCK_BAR	       0x58
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+	outbox = mailbox->buf;
+
+	err = mlx4_cmd_box(dev, 0, mailbox->dma, 0, 0, MLX4_CMD_QUERY_FW,
+			    MLX4_CMD_TIME_CLASS_A, MLX4_CMD_NATIVE);
+	if (err)
+		goto out;
+
+	MLX4_GET(fw_ver, outbox, QUERY_FW_VER_OFFSET);
+	/*
+	 * FW subminor version is at more significant bits than minor
+	 * version, so swap here.
+	 */
+	dev->caps.fw_ver = (fw_ver & 0xffff00000000ull) |
+		((fw_ver & 0xffff0000ull) >> 16) |
+		((fw_ver & 0x0000ffffull) << 16);
+
+	MLX4_GET(lg, outbox, QUERY_FW_PPF_ID);
+	dev->caps.function = lg;
+
+	if (mlx4_is_slave(dev))
+		goto out;
+
+
+	MLX4_GET(cmd_if_rev, outbox, QUERY_FW_CMD_IF_REV_OFFSET);
+	if (cmd_if_rev < MLX4_COMMAND_INTERFACE_MIN_REV ||
+	    cmd_if_rev > MLX4_COMMAND_INTERFACE_MAX_REV) {
+		mlx4_err(dev, "Installed FW has unsupported command interface revision %d\n",
+			 cmd_if_rev);
+		mlx4_err(dev, "(Installed FW version is %d.%d.%03d)\n",
+			 (int) (dev->caps.fw_ver >> 32),
+			 (int) (dev->caps.fw_ver >> 16) & 0xffff,
+			 (int) dev->caps.fw_ver & 0xffff);
+		mlx4_err(dev, "This driver version supports only revisions %d to %d\n",
+			 MLX4_COMMAND_INTERFACE_MIN_REV, MLX4_COMMAND_INTERFACE_MAX_REV);
+		err = -ENODEV;
+		goto out;
+	}
+
+	if (cmd_if_rev < MLX4_COMMAND_INTERFACE_NEW_PORT_CMDS)
+		dev->flags |= MLX4_FLAG_OLD_PORT_CMDS;
+
+	MLX4_GET(lg, outbox, QUERY_FW_MAX_CMD_OFFSET);
+	cmd->max_cmds = 1 << lg;
+
+	mlx4_dbg(dev, "FW version %d.%d.%03d (cmd intf rev %d), max commands %d\n",
+		 (int) (dev->caps.fw_ver >> 32),
+		 (int) (dev->caps.fw_ver >> 16) & 0xffff,
+		 (int) dev->caps.fw_ver & 0xffff,
+		 cmd_if_rev, cmd->max_cmds);
+
+	MLX4_GET(fw->catas_offset, outbox, QUERY_FW_ERR_START_OFFSET);
+	MLX4_GET(fw->catas_size,   outbox, QUERY_FW_ERR_SIZE_OFFSET);
+	MLX4_GET(fw->catas_bar,    outbox, QUERY_FW_ERR_BAR_OFFSET);
+	fw->catas_bar = (fw->catas_bar >> 6) * 2;
+
+	mlx4_dbg(dev, "Catastrophic error buffer at 0x%llx, size 0x%x, BAR %d\n",
+		 (unsigned long long) fw->catas_offset, fw->catas_size, fw->catas_bar);
+
+	MLX4_GET(fw->fw_pages,     outbox, QUERY_FW_SIZE_OFFSET);
+	MLX4_GET(fw->clr_int_base, outbox, QUERY_FW_CLR_INT_BASE_OFFSET);
+	MLX4_GET(fw->clr_int_bar,  outbox, QUERY_FW_CLR_INT_BAR_OFFSET);
+	fw->clr_int_bar = (fw->clr_int_bar >> 6) * 2;
+
+	MLX4_GET(fw->comm_base, outbox, QUERY_FW_COMM_BASE_OFFSET);
+	MLX4_GET(fw->comm_bar,  outbox, QUERY_FW_COMM_BAR_OFFSET);
+	fw->comm_bar = (fw->comm_bar >> 6) * 2;
+	mlx4_dbg(dev, "Communication vector bar:%d offset:0x%llx\n",
+		 fw->comm_bar, fw->comm_base);
+	mlx4_dbg(dev, "FW size %d KB\n", fw->fw_pages >> 2);
+
+	MLX4_GET(fw->clock_offset, outbox, QUERY_FW_CLOCK_OFFSET);
+	MLX4_GET(fw->clock_bar,    outbox, QUERY_FW_CLOCK_BAR);
+	fw->clock_bar = (fw->clock_bar >> 6) * 2;
+	mlx4_dbg(dev, "Internal clock bar:%d offset:0x%llx\n",
+		 fw->clock_bar, fw->clock_offset);
+
+	/*
+	 * Round up number of system pages needed in case
+	 * MLX4_ICM_PAGE_SIZE < PAGE_SIZE.
+	 */
+	fw->fw_pages =
+		ALIGN(fw->fw_pages, PAGE_SIZE / MLX4_ICM_PAGE_SIZE) >>
+		(PAGE_SHIFT - MLX4_ICM_PAGE_SHIFT);
+
+	mlx4_dbg(dev, "Clear int @ %llx, BAR %d\n",
+		 (unsigned long long) fw->clr_int_base, fw->clr_int_bar);
+
+out:
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return err;
+}
+
+int mlx4_QUERY_FW_wrapper(struct mlx4_dev *dev, int slave,
+			  struct mlx4_vhcr *vhcr,
+			  struct mlx4_cmd_mailbox *inbox,
+			  struct mlx4_cmd_mailbox *outbox,
+			  struct mlx4_cmd_info *cmd)
+{
+	u8 *outbuf;
+	int err;
+
+	outbuf = outbox->buf;
+	err = mlx4_cmd_box(dev, 0, outbox->dma, 0, 0, MLX4_CMD_QUERY_FW,
+			    MLX4_CMD_TIME_CLASS_A, MLX4_CMD_NATIVE);
+	if (err)
+		return err;
+
+	/* for slaves, set pci PPF ID to invalid and zero out everything
+	 * else except FW version */
+	outbuf[0] = outbuf[1] = 0;
+	memset(&outbuf[8], 0, QUERY_FW_OUT_SIZE - 8);
+	outbuf[QUERY_FW_PPF_ID] = MLX4_INVALID_SLAVE_ID;
+
+	return 0;
+}
+
+static void get_board_id(void *vsd, char *board_id)
+{
+	int i;
+
+#define VSD_OFFSET_SIG1		0x00
+#define VSD_OFFSET_SIG2		0xde
+#define VSD_OFFSET_MLX_BOARD_ID	0xd0
+#define VSD_OFFSET_TS_BOARD_ID	0x20
+
+#define VSD_SIGNATURE_TOPSPIN	0x5ad
+
+	memset(board_id, 0, MLX4_BOARD_ID_LEN);
+
+	if (be16_to_cpup(vsd + VSD_OFFSET_SIG1) == VSD_SIGNATURE_TOPSPIN &&
+	    be16_to_cpup(vsd + VSD_OFFSET_SIG2) == VSD_SIGNATURE_TOPSPIN) {
+		strscpy(board_id, vsd + VSD_OFFSET_TS_BOARD_ID, MLX4_BOARD_ID_LEN);
+	} else {
+		/*
+		 * The board ID is a string but the firmware byte
+		 * swaps each 4-byte word before passing it back to
+		 * us.  Therefore we need to swab it before printing.
+		 */
+		u32 *bid_u32 = (u32 *)board_id;
+
+		for (i = 0; i < 4; ++i) {
+			u32 *addr;
+			u32 val;
+
+			addr = (u32 *) (vsd + VSD_OFFSET_MLX_BOARD_ID + i * 4);
+			val = get_unaligned(addr);
+			val = swab32(val);
+			put_unaligned(val, &bid_u32[i]);
+		}
+	}
+}
+
+int mlx4_QUERY_ADAPTER(struct mlx4_dev *dev, struct mlx4_adapter *adapter)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	u32 *outbox;
+	int err;
+
+#define QUERY_ADAPTER_OUT_SIZE             0x100
+#define QUERY_ADAPTER_INTA_PIN_OFFSET      0x10
+#define QUERY_ADAPTER_VSD_OFFSET           0x20
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+	outbox = mailbox->buf;
+
+	err = mlx4_cmd_box(dev, 0, mailbox->dma, 0, 0, MLX4_CMD_QUERY_ADAPTER,
+			   MLX4_CMD_TIME_CLASS_A, MLX4_CMD_NATIVE);
+	if (err)
+		goto out;
+
+	MLX4_GET(adapter->inta_pin, outbox,    QUERY_ADAPTER_INTA_PIN_OFFSET);
+
+	get_board_id(outbox + QUERY_ADAPTER_VSD_OFFSET / 4,
+		     adapter->board_id);
+
+out:
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return err;
+}
+
+int mlx4_INIT_HCA(struct mlx4_dev *dev, struct mlx4_init_hca_param *param)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	__be32 *inbox;
+	int err;
+	static const u8 a0_dmfs_hw_steering[] =  {
+		[MLX4_STEERING_DMFS_A0_DEFAULT]		= 0,
+		[MLX4_STEERING_DMFS_A0_DYNAMIC]		= 1,
+		[MLX4_STEERING_DMFS_A0_STATIC]		= 2,
+		[MLX4_STEERING_DMFS_A0_DISABLE]		= 3
+	};
+
+#define INIT_HCA_IN_SIZE		 0x200
+#define INIT_HCA_VERSION_OFFSET		 0x000
+#define	 INIT_HCA_VERSION		 2
+#define INIT_HCA_VXLAN_OFFSET		 0x0c
+#define INIT_HCA_CACHELINE_SZ_OFFSET	 0x0e
+#define INIT_HCA_FLAGS_OFFSET		 0x014
+#define INIT_HCA_RECOVERABLE_ERROR_EVENT_OFFSET 0x018
+#define INIT_HCA_QPC_OFFSET		 0x020
+#define	 INIT_HCA_QPC_BASE_OFFSET	 (INIT_HCA_QPC_OFFSET + 0x10)
+#define	 INIT_HCA_LOG_QP_OFFSET		 (INIT_HCA_QPC_OFFSET + 0x17)
+#define	 INIT_HCA_SRQC_BASE_OFFSET	 (INIT_HCA_QPC_OFFSET + 0x28)
+#define	 INIT_HCA_LOG_SRQ_OFFSET	 (INIT_HCA_QPC_OFFSET + 0x2f)
+#define	 INIT_HCA_CQC_BASE_OFFSET	 (INIT_HCA_QPC_OFFSET + 0x30)
+#define	 INIT_HCA_LOG_CQ_OFFSET		 (INIT_HCA_QPC_OFFSET + 0x37)
+#define	 INIT_HCA_EQE_CQE_OFFSETS	 (INIT_HCA_QPC_OFFSET + 0x38)
+#define	 INIT_HCA_EQE_CQE_STRIDE_OFFSET  (INIT_HCA_QPC_OFFSET + 0x3b)
+#define	 INIT_HCA_ALTC_BASE_OFFSET	 (INIT_HCA_QPC_OFFSET + 0x40)
+#define	 INIT_HCA_AUXC_BASE_OFFSET	 (INIT_HCA_QPC_OFFSET + 0x50)
+#define	 INIT_HCA_EQC_BASE_OFFSET	 (INIT_HCA_QPC_OFFSET + 0x60)
+#define	 INIT_HCA_LOG_EQ_OFFSET		 (INIT_HCA_QPC_OFFSET + 0x67)
+#define	INIT_HCA_NUM_SYS_EQS_OFFSET	(INIT_HCA_QPC_OFFSET + 0x6a)
+#define	 INIT_HCA_RDMARC_BASE_OFFSET	 (INIT_HCA_QPC_OFFSET + 0x70)
+#define	 INIT_HCA_LOG_RD_OFFSET		 (INIT_HCA_QPC_OFFSET + 0x77)
+#define INIT_HCA_MCAST_OFFSET		 0x0c0
+#define	 INIT_HCA_MC_BASE_OFFSET	 (INIT_HCA_MCAST_OFFSET + 0x00)
+#define	 INIT_HCA_LOG_MC_ENTRY_SZ_OFFSET (INIT_HCA_MCAST_OFFSET + 0x13)
+#define	 INIT_HCA_LOG_MC_HASH_SZ_OFFSET	 (INIT_HCA_MCAST_OFFSET + 0x17)
+#define  INIT_HCA_UC_STEERING_OFFSET	 (INIT_HCA_MCAST_OFFSET + 0x18)
+#define	 INIT_HCA_LOG_MC_TABLE_SZ_OFFSET (INIT_HCA_MCAST_OFFSET + 0x1b)
+#define  INIT_HCA_DEVICE_MANAGED_FLOW_STEERING_EN	0x6
+#define  INIT_HCA_DRIVER_VERSION_OFFSET   0x140
+#define  INIT_HCA_DRIVER_VERSION_SZ       0x40
+#define  INIT_HCA_FS_PARAM_OFFSET         0x1d0
+#define  INIT_HCA_FS_BASE_OFFSET          (INIT_HCA_FS_PARAM_OFFSET + 0x00)
+#define  INIT_HCA_FS_LOG_ENTRY_SZ_OFFSET  (INIT_HCA_FS_PARAM_OFFSET + 0x13)
+#define  INIT_HCA_FS_A0_OFFSET		  (INIT_HCA_FS_PARAM_OFFSET + 0x18)
+#define  INIT_HCA_FS_LOG_TABLE_SZ_OFFSET  (INIT_HCA_FS_PARAM_OFFSET + 0x1b)
+#define  INIT_HCA_FS_ETH_BITS_OFFSET      (INIT_HCA_FS_PARAM_OFFSET + 0x21)
+#define  INIT_HCA_FS_ETH_NUM_ADDRS_OFFSET (INIT_HCA_FS_PARAM_OFFSET + 0x22)
+#define  INIT_HCA_FS_IB_BITS_OFFSET       (INIT_HCA_FS_PARAM_OFFSET + 0x25)
+#define  INIT_HCA_FS_IB_NUM_ADDRS_OFFSET  (INIT_HCA_FS_PARAM_OFFSET + 0x26)
+#define INIT_HCA_TPT_OFFSET		 0x0f0
+#define	 INIT_HCA_DMPT_BASE_OFFSET	 (INIT_HCA_TPT_OFFSET + 0x00)
+#define  INIT_HCA_TPT_MW_OFFSET		 (INIT_HCA_TPT_OFFSET + 0x08)
+#define	 INIT_HCA_LOG_MPT_SZ_OFFSET	 (INIT_HCA_TPT_OFFSET + 0x0b)
+#define	 INIT_HCA_MTT_BASE_OFFSET	 (INIT_HCA_TPT_OFFSET + 0x10)
+#define	 INIT_HCA_CMPT_BASE_OFFSET	 (INIT_HCA_TPT_OFFSET + 0x18)
+#define INIT_HCA_UAR_OFFSET		 0x120
+#define	 INIT_HCA_LOG_UAR_SZ_OFFSET	 (INIT_HCA_UAR_OFFSET + 0x0a)
+#define  INIT_HCA_UAR_PAGE_SZ_OFFSET     (INIT_HCA_UAR_OFFSET + 0x0b)
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+	inbox = mailbox->buf;
+
+	*((u8 *) mailbox->buf + INIT_HCA_VERSION_OFFSET) = INIT_HCA_VERSION;
+
+	*((u8 *) mailbox->buf + INIT_HCA_CACHELINE_SZ_OFFSET) =
+		((ilog2(cache_line_size()) - 4) << 5) | (1 << 4);
+
+#if defined(__LITTLE_ENDIAN)
+	*(inbox + INIT_HCA_FLAGS_OFFSET / 4) &= ~cpu_to_be32(1 << 1);
+#elif defined(__BIG_ENDIAN)
+	*(inbox + INIT_HCA_FLAGS_OFFSET / 4) |= cpu_to_be32(1 << 1);
+#else
+#error Host endianness not defined
+#endif
+	/* Check port for UD address vector: */
+	*(inbox + INIT_HCA_FLAGS_OFFSET / 4) |= cpu_to_be32(1);
+
+	/* Enable IPoIB checksumming if we can: */
+	if (dev->caps.flags & MLX4_DEV_CAP_FLAG_IPOIB_CSUM)
+		*(inbox + INIT_HCA_FLAGS_OFFSET / 4) |= cpu_to_be32(1 << 3);
+
+	/* Enable QoS support if module parameter set */
+	if (dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_ETS_CFG && enable_qos)
+		*(inbox + INIT_HCA_FLAGS_OFFSET / 4) |= cpu_to_be32(1 << 2);
+
+	/* enable counters */
+	if (dev->caps.flags & MLX4_DEV_CAP_FLAG_COUNTERS)
+		*(inbox + INIT_HCA_FLAGS_OFFSET / 4) |= cpu_to_be32(1 << 4);
+
+	/* Enable RSS spread to fragmented IP packets when supported */
+	if (dev->caps.flags & MLX4_DEV_CAP_FLAG_RSS_IP_FRAG)
+		*(inbox + INIT_HCA_FLAGS_OFFSET / 4) |= cpu_to_be32(1 << 13);
+
+	/* CX3 is capable of extending CQEs/EQEs from 32 to 64 bytes */
+	if (dev->caps.flags & MLX4_DEV_CAP_FLAG_64B_EQE) {
+		*(inbox + INIT_HCA_EQE_CQE_OFFSETS / 4) |= cpu_to_be32(1 << 29);
+		dev->caps.eqe_size   = 64;
+		dev->caps.eqe_factor = 1;
+	} else {
+		dev->caps.eqe_size   = 32;
+		dev->caps.eqe_factor = 0;
+	}
+
+	if (dev->caps.flags & MLX4_DEV_CAP_FLAG_64B_CQE) {
+		*(inbox + INIT_HCA_EQE_CQE_OFFSETS / 4) |= cpu_to_be32(1 << 30);
+		dev->caps.cqe_size   = 64;
+		dev->caps.userspace_caps |= MLX4_USER_DEV_CAP_LARGE_CQE;
+	} else {
+		dev->caps.cqe_size   = 32;
+	}
+
+	/* CX3 is capable of extending CQEs\EQEs to strides larger than 64B */
+	if ((dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_EQE_STRIDE) &&
+	    (dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_CQE_STRIDE)) {
+		dev->caps.eqe_size = cache_line_size();
+		dev->caps.cqe_size = cache_line_size();
+		dev->caps.eqe_factor = 0;
+		MLX4_PUT(inbox, (u8)((ilog2(dev->caps.eqe_size) - 5) << 4 |
+				      (ilog2(dev->caps.eqe_size) - 5)),
+			 INIT_HCA_EQE_CQE_STRIDE_OFFSET);
+
+		/* User still need to know to support CQE > 32B */
+		dev->caps.userspace_caps |= MLX4_USER_DEV_CAP_LARGE_CQE;
+	}
+
+	if (dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_RECOVERABLE_ERROR_EVENT)
+		*(inbox + INIT_HCA_RECOVERABLE_ERROR_EVENT_OFFSET / 4) |= cpu_to_be32(1 << 31);
+
+	if (dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_DRIVER_VERSION_TO_FW) {
+		u8 *dst = (u8 *)(inbox + INIT_HCA_DRIVER_VERSION_OFFSET / 4);
+
+		strncpy(dst, DRV_NAME_FOR_FW, INIT_HCA_DRIVER_VERSION_SZ - 1);
+		mlx4_dbg(dev, "Reporting Driver Version to FW: %s\n", dst);
+	}
+
+	/* QPC/EEC/CQC/EQC/RDMARC attributes */
+
+	MLX4_PUT(inbox, param->qpc_base,      INIT_HCA_QPC_BASE_OFFSET);
+	MLX4_PUT(inbox, param->log_num_qps,   INIT_HCA_LOG_QP_OFFSET);
+	MLX4_PUT(inbox, param->srqc_base,     INIT_HCA_SRQC_BASE_OFFSET);
+	MLX4_PUT(inbox, param->log_num_srqs,  INIT_HCA_LOG_SRQ_OFFSET);
+	MLX4_PUT(inbox, param->cqc_base,      INIT_HCA_CQC_BASE_OFFSET);
+	MLX4_PUT(inbox, param->log_num_cqs,   INIT_HCA_LOG_CQ_OFFSET);
+	MLX4_PUT(inbox, param->altc_base,     INIT_HCA_ALTC_BASE_OFFSET);
+	MLX4_PUT(inbox, param->auxc_base,     INIT_HCA_AUXC_BASE_OFFSET);
+	MLX4_PUT(inbox, param->eqc_base,      INIT_HCA_EQC_BASE_OFFSET);
+	MLX4_PUT(inbox, param->log_num_eqs,   INIT_HCA_LOG_EQ_OFFSET);
+	MLX4_PUT(inbox, param->num_sys_eqs,   INIT_HCA_NUM_SYS_EQS_OFFSET);
+	MLX4_PUT(inbox, param->rdmarc_base,   INIT_HCA_RDMARC_BASE_OFFSET);
+	MLX4_PUT(inbox, param->log_rd_per_qp, INIT_HCA_LOG_RD_OFFSET);
+
+	/* steering attributes */
+	if (dev->caps.steering_mode ==
+	    MLX4_STEERING_MODE_DEVICE_MANAGED) {
+		*(inbox + INIT_HCA_FLAGS_OFFSET / 4) |=
+			cpu_to_be32(1 <<
+				    INIT_HCA_DEVICE_MANAGED_FLOW_STEERING_EN);
+
+		MLX4_PUT(inbox, param->mc_base, INIT_HCA_FS_BASE_OFFSET);
+		MLX4_PUT(inbox, param->log_mc_entry_sz,
+			 INIT_HCA_FS_LOG_ENTRY_SZ_OFFSET);
+		MLX4_PUT(inbox, param->log_mc_table_sz,
+			 INIT_HCA_FS_LOG_TABLE_SZ_OFFSET);
+		/* Enable Ethernet flow steering
+		 * with udp unicast and tcp unicast
+		 */
+		if (dev->caps.dmfs_high_steer_mode !=
+		    MLX4_STEERING_DMFS_A0_STATIC)
+			MLX4_PUT(inbox,
+				 (u8)(MLX4_FS_UDP_UC_EN | MLX4_FS_TCP_UC_EN),
+				 INIT_HCA_FS_ETH_BITS_OFFSET);
+		MLX4_PUT(inbox, (u16) MLX4_FS_NUM_OF_L2_ADDR,
+			 INIT_HCA_FS_ETH_NUM_ADDRS_OFFSET);
+		/* Enable IPoIB flow steering
+		 * with udp unicast and tcp unicast
+		 */
+		MLX4_PUT(inbox, (u8) (MLX4_FS_UDP_UC_EN | MLX4_FS_TCP_UC_EN),
+			 INIT_HCA_FS_IB_BITS_OFFSET);
+		MLX4_PUT(inbox, (u16) MLX4_FS_NUM_OF_L2_ADDR,
+			 INIT_HCA_FS_IB_NUM_ADDRS_OFFSET);
+
+		if (dev->caps.dmfs_high_steer_mode !=
+		    MLX4_STEERING_DMFS_A0_NOT_SUPPORTED)
+			MLX4_PUT(inbox,
+				 ((u8)(a0_dmfs_hw_steering[dev->caps.dmfs_high_steer_mode]
+				       << 6)),
+				 INIT_HCA_FS_A0_OFFSET);
+	} else {
+		MLX4_PUT(inbox, param->mc_base,	INIT_HCA_MC_BASE_OFFSET);
+		MLX4_PUT(inbox, param->log_mc_entry_sz,
+			 INIT_HCA_LOG_MC_ENTRY_SZ_OFFSET);
+		MLX4_PUT(inbox, param->log_mc_hash_sz,
+			 INIT_HCA_LOG_MC_HASH_SZ_OFFSET);
+		MLX4_PUT(inbox, param->log_mc_table_sz,
+			 INIT_HCA_LOG_MC_TABLE_SZ_OFFSET);
+		if (dev->caps.steering_mode == MLX4_STEERING_MODE_B0)
+			MLX4_PUT(inbox, (u8) (1 << 3),
+				 INIT_HCA_UC_STEERING_OFFSET);
+	}
+
+	/* TPT attributes */
+
+	MLX4_PUT(inbox, param->dmpt_base,  INIT_HCA_DMPT_BASE_OFFSET);
+	MLX4_PUT(inbox, param->mw_enabled, INIT_HCA_TPT_MW_OFFSET);
+	MLX4_PUT(inbox, param->log_mpt_sz, INIT_HCA_LOG_MPT_SZ_OFFSET);
+	MLX4_PUT(inbox, param->mtt_base,   INIT_HCA_MTT_BASE_OFFSET);
+	MLX4_PUT(inbox, param->cmpt_base,  INIT_HCA_CMPT_BASE_OFFSET);
+
+	/* UAR attributes */
+
+	MLX4_PUT(inbox, param->uar_page_sz,	INIT_HCA_UAR_PAGE_SZ_OFFSET);
+	MLX4_PUT(inbox, param->log_uar_sz,      INIT_HCA_LOG_UAR_SZ_OFFSET);
+
+	/* set parser VXLAN attributes */
+	if (dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_VXLAN_OFFLOADS) {
+		u8 parser_params = 0;
+		MLX4_PUT(inbox, parser_params,	INIT_HCA_VXLAN_OFFSET);
+	}
+
+	err = mlx4_cmd(dev, mailbox->dma, 0, 0, MLX4_CMD_INIT_HCA,
+		       MLX4_CMD_TIME_CLASS_C, MLX4_CMD_NATIVE);
+
+	if (err)
+		mlx4_err(dev, "INIT_HCA returns %d\n", err);
+
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return err;
+}
+
+int mlx4_QUERY_HCA(struct mlx4_dev *dev,
+		   struct mlx4_init_hca_param *param)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	__be32 *outbox;
+	u64 qword_field;
+	u32 dword_field;
+	u16 word_field;
+	u8 byte_field;
+	int err;
+	static const u8 a0_dmfs_query_hw_steering[] =  {
+		[0] = MLX4_STEERING_DMFS_A0_DEFAULT,
+		[1] = MLX4_STEERING_DMFS_A0_DYNAMIC,
+		[2] = MLX4_STEERING_DMFS_A0_STATIC,
+		[3] = MLX4_STEERING_DMFS_A0_DISABLE
+	};
+
+#define QUERY_HCA_GLOBAL_CAPS_OFFSET	0x04
+#define QUERY_HCA_CORE_CLOCK_OFFSET	0x0c
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+	outbox = mailbox->buf;
+
+	err = mlx4_cmd_box(dev, 0, mailbox->dma, 0, 0,
+			   MLX4_CMD_QUERY_HCA,
+			   MLX4_CMD_TIME_CLASS_B,
+			   !mlx4_is_slave(dev));
+	if (err)
+		goto out;
+
+	MLX4_GET(param->global_caps, outbox, QUERY_HCA_GLOBAL_CAPS_OFFSET);
+	MLX4_GET(param->hca_core_clock, outbox, QUERY_HCA_CORE_CLOCK_OFFSET);
+
+	/* QPC/EEC/CQC/EQC/RDMARC attributes */
+
+	MLX4_GET(qword_field, outbox, INIT_HCA_QPC_BASE_OFFSET);
+	param->qpc_base = qword_field & ~((u64)0x1f);
+	MLX4_GET(byte_field, outbox, INIT_HCA_LOG_QP_OFFSET);
+	param->log_num_qps = byte_field & 0x1f;
+	MLX4_GET(qword_field, outbox, INIT_HCA_SRQC_BASE_OFFSET);
+	param->srqc_base = qword_field & ~((u64)0x1f);
+	MLX4_GET(byte_field, outbox, INIT_HCA_LOG_SRQ_OFFSET);
+	param->log_num_srqs = byte_field & 0x1f;
+	MLX4_GET(qword_field, outbox, INIT_HCA_CQC_BASE_OFFSET);
+	param->cqc_base = qword_field & ~((u64)0x1f);
+	MLX4_GET(byte_field, outbox, INIT_HCA_LOG_CQ_OFFSET);
+	param->log_num_cqs = byte_field & 0x1f;
+	MLX4_GET(qword_field, outbox, INIT_HCA_ALTC_BASE_OFFSET);
+	param->altc_base = qword_field;
+	MLX4_GET(qword_field, outbox, INIT_HCA_AUXC_BASE_OFFSET);
+	param->auxc_base = qword_field;
+	MLX4_GET(qword_field, outbox, INIT_HCA_EQC_BASE_OFFSET);
+	param->eqc_base = qword_field & ~((u64)0x1f);
+	MLX4_GET(byte_field, outbox, INIT_HCA_LOG_EQ_OFFSET);
+	param->log_num_eqs = byte_field & 0x1f;
+	MLX4_GET(word_field, outbox, INIT_HCA_NUM_SYS_EQS_OFFSET);
+	param->num_sys_eqs = word_field & 0xfff;
+	MLX4_GET(qword_field, outbox, INIT_HCA_RDMARC_BASE_OFFSET);
+	param->rdmarc_base = qword_field & ~((u64)0x1f);
+	MLX4_GET(byte_field, outbox, INIT_HCA_LOG_RD_OFFSET);
+	param->log_rd_per_qp = byte_field & 0x7;
+
+	MLX4_GET(dword_field, outbox, INIT_HCA_FLAGS_OFFSET);
+	if (dword_field & (1 << INIT_HCA_DEVICE_MANAGED_FLOW_STEERING_EN)) {
+		param->steering_mode = MLX4_STEERING_MODE_DEVICE_MANAGED;
+	} else {
+		MLX4_GET(byte_field, outbox, INIT_HCA_UC_STEERING_OFFSET);
+		if (byte_field & 0x8)
+			param->steering_mode = MLX4_STEERING_MODE_B0;
+		else
+			param->steering_mode = MLX4_STEERING_MODE_A0;
+	}
+
+	if (dword_field & (1 << 13))
+		param->rss_ip_frags = 1;
+
+	/* steering attributes */
+	if (param->steering_mode == MLX4_STEERING_MODE_DEVICE_MANAGED) {
+		MLX4_GET(param->mc_base, outbox, INIT_HCA_FS_BASE_OFFSET);
+		MLX4_GET(byte_field, outbox, INIT_HCA_FS_LOG_ENTRY_SZ_OFFSET);
+		param->log_mc_entry_sz = byte_field & 0x1f;
+		MLX4_GET(byte_field, outbox, INIT_HCA_FS_LOG_TABLE_SZ_OFFSET);
+		param->log_mc_table_sz = byte_field & 0x1f;
+		MLX4_GET(byte_field, outbox, INIT_HCA_FS_A0_OFFSET);
+		param->dmfs_high_steer_mode =
+			a0_dmfs_query_hw_steering[(byte_field >> 6) & 3];
+	} else {
+		MLX4_GET(param->mc_base, outbox, INIT_HCA_MC_BASE_OFFSET);
+		MLX4_GET(byte_field, outbox, INIT_HCA_LOG_MC_ENTRY_SZ_OFFSET);
+		param->log_mc_entry_sz = byte_field & 0x1f;
+		MLX4_GET(byte_field,  outbox, INIT_HCA_LOG_MC_HASH_SZ_OFFSET);
+		param->log_mc_hash_sz = byte_field & 0x1f;
+		MLX4_GET(byte_field, outbox, INIT_HCA_LOG_MC_TABLE_SZ_OFFSET);
+		param->log_mc_table_sz = byte_field & 0x1f;
+	}
+
+	/* CX3 is capable of extending CQEs/EQEs from 32 to 64 bytes */
+	MLX4_GET(byte_field, outbox, INIT_HCA_EQE_CQE_OFFSETS);
+	if (byte_field & 0x20) /* 64-bytes eqe enabled */
+		param->dev_cap_enabled |= MLX4_DEV_CAP_64B_EQE_ENABLED;
+	if (byte_field & 0x40) /* 64-bytes cqe enabled */
+		param->dev_cap_enabled |= MLX4_DEV_CAP_64B_CQE_ENABLED;
+
+	/* CX3 is capable of extending CQEs\EQEs to strides larger than 64B */
+	MLX4_GET(byte_field, outbox, INIT_HCA_EQE_CQE_STRIDE_OFFSET);
+	if (byte_field) {
+		param->dev_cap_enabled |= MLX4_DEV_CAP_EQE_STRIDE_ENABLED;
+		param->dev_cap_enabled |= MLX4_DEV_CAP_CQE_STRIDE_ENABLED;
+		param->cqe_size = 1 << ((byte_field &
+					 MLX4_CQE_SIZE_MASK_STRIDE) + 5);
+		param->eqe_size = 1 << (((byte_field &
+					  MLX4_EQE_SIZE_MASK_STRIDE) >> 4) + 5);
+	}
+
+	/* TPT attributes */
+
+	MLX4_GET(param->dmpt_base,  outbox, INIT_HCA_DMPT_BASE_OFFSET);
+	MLX4_GET(byte_field, outbox, INIT_HCA_TPT_MW_OFFSET);
+	param->mw_enabled = byte_field >> 7;
+	MLX4_GET(byte_field, outbox, INIT_HCA_LOG_MPT_SZ_OFFSET);
+	param->log_mpt_sz = byte_field & 0x3f;
+	MLX4_GET(param->mtt_base,   outbox, INIT_HCA_MTT_BASE_OFFSET);
+	MLX4_GET(param->cmpt_base,  outbox, INIT_HCA_CMPT_BASE_OFFSET);
+
+	/* UAR attributes */
+
+	MLX4_GET(param->uar_page_sz, outbox, INIT_HCA_UAR_PAGE_SZ_OFFSET);
+	MLX4_GET(byte_field, outbox, INIT_HCA_LOG_UAR_SZ_OFFSET);
+	param->log_uar_sz = byte_field & 0xf;
+
+	/* phv_check enable */
+	MLX4_GET(byte_field, outbox, INIT_HCA_CACHELINE_SZ_OFFSET);
+	if (byte_field & 0x2)
+		param->phv_check_en = 1;
+out:
+	mlx4_free_cmd_mailbox(dev, mailbox);
+
+	return err;
+}
+
+static int mlx4_hca_core_clock_update(struct mlx4_dev *dev)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	__be32 *outbox;
+	int err;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox)) {
+		mlx4_warn(dev, "hca_core_clock mailbox allocation failed\n");
+		return PTR_ERR(mailbox);
+	}
+	outbox = mailbox->buf;
+
+	err = mlx4_cmd_box(dev, 0, mailbox->dma, 0, 0,
+			   MLX4_CMD_QUERY_HCA,
+			   MLX4_CMD_TIME_CLASS_B,
+			   !mlx4_is_slave(dev));
+	if (err) {
+		mlx4_warn(dev, "hca_core_clock update failed\n");
+		goto out;
+	}
+
+	MLX4_GET(dev->caps.hca_core_clock, outbox, QUERY_HCA_CORE_CLOCK_OFFSET);
+
+out:
+	mlx4_free_cmd_mailbox(dev, mailbox);
+
+	return err;
+}
+
+/* for IB-type ports only in SRIOV mode. Checks that both proxy QP0
+ * and real QP0 are active, so that the paravirtualized QP0 is ready
+ * to operate */
+static int check_qp0_state(struct mlx4_dev *dev, int function, int port)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	/* irrelevant if not infiniband */
+	if (priv->mfunc.master.qp0_state[port].proxy_qp0_active &&
+	    priv->mfunc.master.qp0_state[port].qp0_active)
+		return 1;
+	return 0;
+}
+
+int mlx4_INIT_PORT_wrapper(struct mlx4_dev *dev, int slave,
+			   struct mlx4_vhcr *vhcr,
+			   struct mlx4_cmd_mailbox *inbox,
+			   struct mlx4_cmd_mailbox *outbox,
+			   struct mlx4_cmd_info *cmd)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int port = mlx4_slave_convert_port(dev, slave, vhcr->in_modifier);
+	int err;
+
+	if (port < 0)
+		return -EINVAL;
+
+	if (priv->mfunc.master.slave_state[slave].init_port_mask & (1 << port))
+		return 0;
+
+	if (dev->caps.port_mask[port] != MLX4_PORT_TYPE_IB) {
+		/* Enable port only if it was previously disabled */
+		if (!priv->mfunc.master.init_port_ref[port]) {
+			err = mlx4_cmd(dev, 0, port, 0, MLX4_CMD_INIT_PORT,
+				       MLX4_CMD_TIME_CLASS_A, MLX4_CMD_NATIVE);
+			if (err)
+				return err;
+		}
+		priv->mfunc.master.slave_state[slave].init_port_mask |= (1 << port);
+	} else {
+		if (slave == mlx4_master_func_num(dev)) {
+			if (check_qp0_state(dev, slave, port) &&
+			    !priv->mfunc.master.qp0_state[port].port_active) {
+				err = mlx4_cmd(dev, 0, port, 0, MLX4_CMD_INIT_PORT,
+					       MLX4_CMD_TIME_CLASS_A, MLX4_CMD_NATIVE);
+				if (err)
+					return err;
+				priv->mfunc.master.qp0_state[port].port_active = 1;
+				priv->mfunc.master.slave_state[slave].init_port_mask |= (1 << port);
+			}
+		} else
+			priv->mfunc.master.slave_state[slave].init_port_mask |= (1 << port);
+	}
+	++priv->mfunc.master.init_port_ref[port];
+	return 0;
+}
+
+int mlx4_INIT_PORT(struct mlx4_dev *dev, int port)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	u32 *inbox;
+	int err;
+	u32 flags;
+	u16 field;
+
+	if (dev->flags & MLX4_FLAG_OLD_PORT_CMDS) {
+#define INIT_PORT_IN_SIZE          256
+#define INIT_PORT_FLAGS_OFFSET     0x00
+#define INIT_PORT_FLAG_SIG         (1 << 18)
+#define INIT_PORT_FLAG_NG          (1 << 17)
+#define INIT_PORT_FLAG_G0          (1 << 16)
+#define INIT_PORT_VL_SHIFT         4
+#define INIT_PORT_PORT_WIDTH_SHIFT 8
+#define INIT_PORT_MTU_OFFSET       0x04
+#define INIT_PORT_MAX_GID_OFFSET   0x06
+#define INIT_PORT_MAX_PKEY_OFFSET  0x0a
+#define INIT_PORT_GUID0_OFFSET     0x10
+#define INIT_PORT_NODE_GUID_OFFSET 0x18
+#define INIT_PORT_SI_GUID_OFFSET   0x20
+
+		mailbox = mlx4_alloc_cmd_mailbox(dev);
+		if (IS_ERR(mailbox))
+			return PTR_ERR(mailbox);
+		inbox = mailbox->buf;
+
+		flags = 0;
+		flags |= (dev->caps.vl_cap[port] & 0xf) << INIT_PORT_VL_SHIFT;
+		flags |= (dev->caps.port_width_cap[port] & 0xf) << INIT_PORT_PORT_WIDTH_SHIFT;
+		MLX4_PUT(inbox, flags,		  INIT_PORT_FLAGS_OFFSET);
+
+		field = 128 << dev->caps.ib_mtu_cap[port];
+		MLX4_PUT(inbox, field, INIT_PORT_MTU_OFFSET);
+		field = dev->caps.gid_table_len[port];
+		MLX4_PUT(inbox, field, INIT_PORT_MAX_GID_OFFSET);
+		field = dev->caps.pkey_table_len[port];
+		MLX4_PUT(inbox, field, INIT_PORT_MAX_PKEY_OFFSET);
+
+		err = mlx4_cmd(dev, mailbox->dma, port, 0, MLX4_CMD_INIT_PORT,
+			       MLX4_CMD_TIME_CLASS_A, MLX4_CMD_NATIVE);
+
+		mlx4_free_cmd_mailbox(dev, mailbox);
+	} else
+		err = mlx4_cmd(dev, 0, port, 0, MLX4_CMD_INIT_PORT,
+			       MLX4_CMD_TIME_CLASS_A, MLX4_CMD_WRAPPED);
+
+	if (!err)
+		mlx4_hca_core_clock_update(dev);
+
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx4_INIT_PORT);
+
+int mlx4_CLOSE_PORT_wrapper(struct mlx4_dev *dev, int slave,
+			    struct mlx4_vhcr *vhcr,
+			    struct mlx4_cmd_mailbox *inbox,
+			    struct mlx4_cmd_mailbox *outbox,
+			    struct mlx4_cmd_info *cmd)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int port = mlx4_slave_convert_port(dev, slave, vhcr->in_modifier);
+	int err;
+
+	if (port < 0)
+		return -EINVAL;
+
+	if (!(priv->mfunc.master.slave_state[slave].init_port_mask &
+	    (1 << port)))
+		return 0;
+
+	if (dev->caps.port_mask[port] != MLX4_PORT_TYPE_IB) {
+		if (priv->mfunc.master.init_port_ref[port] == 1) {
+			err = mlx4_cmd(dev, 0, port, 0, MLX4_CMD_CLOSE_PORT,
+				       MLX4_CMD_TIME_CLASS_A, MLX4_CMD_NATIVE);
+			if (err)
+				return err;
+		}
+		priv->mfunc.master.slave_state[slave].init_port_mask &= ~(1 << port);
+	} else {
+		/* infiniband port */
+		if (slave == mlx4_master_func_num(dev)) {
+			if (!priv->mfunc.master.qp0_state[port].qp0_active &&
+			    priv->mfunc.master.qp0_state[port].port_active) {
+				err = mlx4_cmd(dev, 0, port, 0, MLX4_CMD_CLOSE_PORT,
+					       MLX4_CMD_TIME_CLASS_A, MLX4_CMD_NATIVE);
+				if (err)
+					return err;
+				priv->mfunc.master.slave_state[slave].init_port_mask &= ~(1 << port);
+				priv->mfunc.master.qp0_state[port].port_active = 0;
+			}
+		} else
+			priv->mfunc.master.slave_state[slave].init_port_mask &= ~(1 << port);
+	}
+	--priv->mfunc.master.init_port_ref[port];
+	return 0;
+}
+
+int mlx4_CLOSE_PORT(struct mlx4_dev *dev, int port)
+{
+	return mlx4_cmd(dev, 0, port, 0, MLX4_CMD_CLOSE_PORT,
+			MLX4_CMD_TIME_CLASS_A, MLX4_CMD_WRAPPED);
+}
+EXPORT_SYMBOL_GPL(mlx4_CLOSE_PORT);
+
+int mlx4_CLOSE_HCA(struct mlx4_dev *dev, int panic)
+{
+	return mlx4_cmd(dev, 0, 0, panic, MLX4_CMD_CLOSE_HCA,
+			MLX4_CMD_TIME_CLASS_C, MLX4_CMD_NATIVE);
+}
+
+struct mlx4_config_dev {
+	__be32	update_flags;
+	__be32	rsvd1[3];
+	__be16	vxlan_udp_dport;
+	__be16	rsvd2;
+	__be16  roce_v2_entropy;
+	__be16  roce_v2_udp_dport;
+	__be32	roce_flags;
+	__be32	rsvd4[25];
+	__be16	rsvd5;
+	u8	rsvd6;
+	u8	rx_checksum_val;
+};
+
+#define MLX4_VXLAN_UDP_DPORT (1 << 0)
+#define MLX4_ROCE_V2_UDP_DPORT BIT(3)
+#define MLX4_DISABLE_RX_PORT BIT(18)
+
+static int mlx4_CONFIG_DEV_set(struct mlx4_dev *dev, struct mlx4_config_dev *config_dev)
+{
+	int err;
+	struct mlx4_cmd_mailbox *mailbox;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+
+	memcpy(mailbox->buf, config_dev, sizeof(*config_dev));
+
+	err = mlx4_cmd(dev, mailbox->dma, 0, 0, MLX4_CMD_CONFIG_DEV,
+		       MLX4_CMD_TIME_CLASS_B, MLX4_CMD_NATIVE);
+
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return err;
+}
+
+static int mlx4_CONFIG_DEV_get(struct mlx4_dev *dev, struct mlx4_config_dev *config_dev)
+{
+	int err;
+	struct mlx4_cmd_mailbox *mailbox;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+
+	err = mlx4_cmd_box(dev, 0, mailbox->dma, 0, 1, MLX4_CMD_CONFIG_DEV,
+			   MLX4_CMD_TIME_CLASS_A, MLX4_CMD_NATIVE);
+
+	if (!err)
+		memcpy(config_dev, mailbox->buf, sizeof(*config_dev));
+
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return err;
+}
+
+/* Conversion between the HW values and the actual functionality.
+ * The value represented by the array index,
+ * and the functionality determined by the flags.
+ */
+static const u8 config_dev_csum_flags[] = {
+	[0] =	0,
+	[1] =	MLX4_RX_CSUM_MODE_VAL_NON_TCP_UDP,
+	[2] =	MLX4_RX_CSUM_MODE_VAL_NON_TCP_UDP	|
+		MLX4_RX_CSUM_MODE_L4,
+	[3] =	MLX4_RX_CSUM_MODE_L4			|
+		MLX4_RX_CSUM_MODE_IP_OK_IP_NON_TCP_UDP	|
+		MLX4_RX_CSUM_MODE_MULTI_VLAN
+};
+
+int mlx4_config_dev_retrieval(struct mlx4_dev *dev,
+			      struct mlx4_config_dev_params *params)
+{
+	struct mlx4_config_dev config_dev = {0};
+	int err;
+	u8 csum_mask;
+
+#define CONFIG_DEV_RX_CSUM_MODE_MASK			0x7
+#define CONFIG_DEV_RX_CSUM_MODE_PORT1_BIT_OFFSET	0
+#define CONFIG_DEV_RX_CSUM_MODE_PORT2_BIT_OFFSET	4
+
+	if (!(dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_CONFIG_DEV))
+		return -EOPNOTSUPP;
+
+	err = mlx4_CONFIG_DEV_get(dev, &config_dev);
+	if (err)
+		return err;
+
+	csum_mask = (config_dev.rx_checksum_val >> CONFIG_DEV_RX_CSUM_MODE_PORT1_BIT_OFFSET) &
+			CONFIG_DEV_RX_CSUM_MODE_MASK;
+
+	if (csum_mask >= ARRAY_SIZE(config_dev_csum_flags))
+		return -EINVAL;
+	params->rx_csum_flags_port_1 = config_dev_csum_flags[csum_mask];
+
+	csum_mask = (config_dev.rx_checksum_val >> CONFIG_DEV_RX_CSUM_MODE_PORT2_BIT_OFFSET) &
+			CONFIG_DEV_RX_CSUM_MODE_MASK;
+
+	if (csum_mask >= ARRAY_SIZE(config_dev_csum_flags))
+		return -EINVAL;
+	params->rx_csum_flags_port_2 = config_dev_csum_flags[csum_mask];
+
+	params->vxlan_udp_dport = be16_to_cpu(config_dev.vxlan_udp_dport);
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(mlx4_config_dev_retrieval);
+
+int mlx4_config_vxlan_port(struct mlx4_dev *dev, __be16 udp_port)
+{
+	struct mlx4_config_dev config_dev;
+
+	memset(&config_dev, 0, sizeof(config_dev));
+	config_dev.update_flags    = cpu_to_be32(MLX4_VXLAN_UDP_DPORT);
+	config_dev.vxlan_udp_dport = udp_port;
+
+	return mlx4_CONFIG_DEV_set(dev, &config_dev);
+}
+EXPORT_SYMBOL_GPL(mlx4_config_vxlan_port);
+
+#define CONFIG_DISABLE_RX_PORT BIT(15)
+int mlx4_disable_rx_port_check(struct mlx4_dev *dev, bool dis)
+{
+	struct mlx4_config_dev config_dev;
+
+	memset(&config_dev, 0, sizeof(config_dev));
+	config_dev.update_flags = cpu_to_be32(MLX4_DISABLE_RX_PORT);
+	if (dis)
+		config_dev.roce_flags =
+			cpu_to_be32(CONFIG_DISABLE_RX_PORT);
+
+	return mlx4_CONFIG_DEV_set(dev, &config_dev);
+}
+
+int mlx4_config_roce_v2_port(struct mlx4_dev *dev, u16 udp_port)
+{
+	struct mlx4_config_dev config_dev;
+
+	memset(&config_dev, 0, sizeof(config_dev));
+	config_dev.update_flags    = cpu_to_be32(MLX4_ROCE_V2_UDP_DPORT);
+	config_dev.roce_v2_udp_dport = cpu_to_be16(udp_port);
+
+	return mlx4_CONFIG_DEV_set(dev, &config_dev);
+}
+EXPORT_SYMBOL_GPL(mlx4_config_roce_v2_port);
+
+int mlx4_virt2phy_port_map(struct mlx4_dev *dev, u32 port1, u32 port2)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	struct {
+		__be32 v_port1;
+		__be32 v_port2;
+	} *v2p;
+	int err;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return -ENOMEM;
+
+	v2p = mailbox->buf;
+	v2p->v_port1 = cpu_to_be32(port1);
+	v2p->v_port2 = cpu_to_be32(port2);
+
+	err = mlx4_cmd(dev, mailbox->dma, 0,
+		       MLX4_SET_PORT_VIRT2PHY, MLX4_CMD_VIRT_PORT_MAP,
+		       MLX4_CMD_TIME_CLASS_B, MLX4_CMD_NATIVE);
+
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return err;
+}
+
+
+int mlx4_SET_ICM_SIZE(struct mlx4_dev *dev, u64 icm_size, u64 *aux_pages)
+{
+	int ret = mlx4_cmd_imm(dev, icm_size, aux_pages, 0, 0,
+			       MLX4_CMD_SET_ICM_SIZE,
+			       MLX4_CMD_TIME_CLASS_A, MLX4_CMD_NATIVE);
+	if (ret)
+		return ret;
+
+	/*
+	 * Round up number of system pages needed in case
+	 * MLX4_ICM_PAGE_SIZE < PAGE_SIZE.
+	 */
+	*aux_pages = ALIGN(*aux_pages, PAGE_SIZE / MLX4_ICM_PAGE_SIZE) >>
+		(PAGE_SHIFT - MLX4_ICM_PAGE_SHIFT);
+
+	return 0;
+}
+
+int mlx4_NOP(struct mlx4_dev *dev)
+{
+	/* Input modifier of 0x1f means "finish as soon as possible." */
+	return mlx4_cmd(dev, 0, 0x1f, 0, MLX4_CMD_NOP, MLX4_CMD_TIME_CLASS_A,
+			MLX4_CMD_NATIVE);
+}
+
+int mlx4_query_diag_counters(struct mlx4_dev *dev, u8 op_modifier,
+			     const u32 offset[],
+			     u32 value[], size_t array_len, u8 port)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	u32 *outbox;
+	size_t i;
+	int ret;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+
+	outbox = mailbox->buf;
+
+	ret = mlx4_cmd_box(dev, 0, mailbox->dma, port, op_modifier,
+			   MLX4_CMD_DIAG_RPRT, MLX4_CMD_TIME_CLASS_A,
+			   MLX4_CMD_NATIVE);
+	if (ret)
+		goto out;
+
+	for (i = 0; i < array_len; i++) {
+		if (offset[i] > MLX4_MAILBOX_SIZE) {
+			ret = -EINVAL;
+			goto out;
+		}
+
+		MLX4_GET(value[i], outbox, offset[i]);
+	}
+
+out:
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return ret;
+}
+EXPORT_SYMBOL(mlx4_query_diag_counters);
+
+int mlx4_get_phys_port_id(struct mlx4_dev *dev)
+{
+	u8 port;
+	u32 *outbox;
+	struct mlx4_cmd_mailbox *mailbox;
+	u32 in_mod;
+	u32 guid_hi, guid_lo;
+	int err, ret = 0;
+#define MOD_STAT_CFG_PORT_OFFSET 8
+#define MOD_STAT_CFG_GUID_H	 0X14
+#define MOD_STAT_CFG_GUID_L	 0X1c
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+	outbox = mailbox->buf;
+
+	for (port = 1; port <= dev->caps.num_ports; port++) {
+		in_mod = port << MOD_STAT_CFG_PORT_OFFSET;
+		err = mlx4_cmd_box(dev, 0, mailbox->dma, in_mod, 0x2,
+				   MLX4_CMD_MOD_STAT_CFG, MLX4_CMD_TIME_CLASS_A,
+				   MLX4_CMD_NATIVE);
+		if (err) {
+			mlx4_err(dev, "Fail to get port %d uplink guid\n",
+				 port);
+			ret = err;
+		} else {
+			MLX4_GET(guid_hi, outbox, MOD_STAT_CFG_GUID_H);
+			MLX4_GET(guid_lo, outbox, MOD_STAT_CFG_GUID_L);
+			dev->caps.phys_port_id[port] = (u64)guid_lo |
+						       (u64)guid_hi << 32;
+		}
+	}
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return ret;
+}
+
+#define MLX4_WOL_SETUP_MODE (5 << 28)
+int mlx4_wol_read(struct mlx4_dev *dev, u64 *config, int port)
+{
+	u32 in_mod = MLX4_WOL_SETUP_MODE | port << 8;
+
+	return mlx4_cmd_imm(dev, 0, config, in_mod, 0x3,
+			    MLX4_CMD_MOD_STAT_CFG, MLX4_CMD_TIME_CLASS_A,
+			    MLX4_CMD_NATIVE);
+}
+EXPORT_SYMBOL_GPL(mlx4_wol_read);
+
+int mlx4_wol_write(struct mlx4_dev *dev, u64 config, int port)
+{
+	u32 in_mod = MLX4_WOL_SETUP_MODE | port << 8;
+
+	return mlx4_cmd(dev, config, in_mod, 0x1, MLX4_CMD_MOD_STAT_CFG,
+			MLX4_CMD_TIME_CLASS_A, MLX4_CMD_NATIVE);
+}
+EXPORT_SYMBOL_GPL(mlx4_wol_write);
+
+enum {
+	ADD_TO_MCG = 0x26,
+};
+
+
+void mlx4_opreq_action(struct work_struct *work)
+{
+	struct mlx4_priv *priv = container_of(work, struct mlx4_priv,
+					      opreq_task);
+	struct mlx4_dev *dev = &priv->dev;
+	int num_tasks = atomic_read(&priv->opreq_count);
+	struct mlx4_cmd_mailbox *mailbox;
+	struct mlx4_mgm *mgm;
+	u32 *outbox;
+	u32 modifier;
+	u16 token;
+	u16 type;
+	int err;
+	u32 num_qps;
+	struct mlx4_qp qp;
+	int i;
+	u8 rem_mcg;
+	u8 prot;
+
+#define GET_OP_REQ_MODIFIER_OFFSET	0x08
+#define GET_OP_REQ_TOKEN_OFFSET		0x14
+#define GET_OP_REQ_TYPE_OFFSET		0x1a
+#define GET_OP_REQ_DATA_OFFSET		0x20
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox)) {
+		mlx4_err(dev, "Failed to allocate mailbox for GET_OP_REQ\n");
+		return;
+	}
+	outbox = mailbox->buf;
+
+	while (num_tasks) {
+		err = mlx4_cmd_box(dev, 0, mailbox->dma, 0, 0,
+				   MLX4_CMD_GET_OP_REQ, MLX4_CMD_TIME_CLASS_A,
+				   MLX4_CMD_NATIVE);
+		if (err) {
+			mlx4_err(dev, "Failed to retrieve required operation: %d\n",
+				 err);
+			goto out;
+		}
+		MLX4_GET(modifier, outbox, GET_OP_REQ_MODIFIER_OFFSET);
+		MLX4_GET(token, outbox, GET_OP_REQ_TOKEN_OFFSET);
+		MLX4_GET(type, outbox, GET_OP_REQ_TYPE_OFFSET);
+		type &= 0xfff;
+
+		switch (type) {
+		case ADD_TO_MCG:
+			if (dev->caps.steering_mode ==
+			    MLX4_STEERING_MODE_DEVICE_MANAGED) {
+				mlx4_warn(dev, "ADD MCG operation is not supported in DEVICE_MANAGED steering mode\n");
+				err = EPERM;
+				break;
+			}
+			mgm = (struct mlx4_mgm *)((u8 *)(outbox) +
+						  GET_OP_REQ_DATA_OFFSET);
+			num_qps = be32_to_cpu(mgm->members_count) &
+				  MGM_QPN_MASK;
+			rem_mcg = ((u8 *)(&mgm->members_count))[0] & 1;
+			prot = ((u8 *)(&mgm->members_count))[0] >> 6;
+
+			for (i = 0; i < num_qps; i++) {
+				qp.qpn = be32_to_cpu(mgm->qp[i]);
+				if (rem_mcg)
+					err = mlx4_multicast_detach(dev, &qp,
+								    mgm->gid,
+								    prot, 0);
+				else
+					err = mlx4_multicast_attach(dev, &qp,
+								    mgm->gid,
+								    mgm->gid[5]
+								    , 0, prot,
+								    NULL);
+				if (err)
+					break;
+			}
+			break;
+		default:
+			mlx4_warn(dev, "Bad type for required operation\n");
+			err = EINVAL;
+			break;
+		}
+		err = mlx4_cmd(dev, 0, ((u32) err |
+					(__force u32)cpu_to_be32(token) << 16),
+			       1, MLX4_CMD_GET_OP_REQ, MLX4_CMD_TIME_CLASS_A,
+			       MLX4_CMD_NATIVE);
+		if (err) {
+			mlx4_err(dev, "Failed to acknowledge required request: %d\n",
+				 err);
+			goto out;
+		}
+		memset(outbox, 0, 0xffc);
+		num_tasks = atomic_dec_return(&priv->opreq_count);
+	}
+
+out:
+	mlx4_free_cmd_mailbox(dev, mailbox);
+}
+
+static int mlx4_check_smp_firewall_active(struct mlx4_dev *dev,
+					  struct mlx4_cmd_mailbox *mailbox)
+{
+#define MLX4_CMD_MAD_DEMUX_SET_ATTR_OFFSET		0x10
+#define MLX4_CMD_MAD_DEMUX_GETRESP_ATTR_OFFSET		0x20
+#define MLX4_CMD_MAD_DEMUX_TRAP_ATTR_OFFSET		0x40
+#define MLX4_CMD_MAD_DEMUX_TRAP_REPRESS_ATTR_OFFSET	0x70
+
+	u32 set_attr_mask, getresp_attr_mask;
+	u32 trap_attr_mask, traprepress_attr_mask;
+
+	MLX4_GET(set_attr_mask, mailbox->buf,
+		 MLX4_CMD_MAD_DEMUX_SET_ATTR_OFFSET);
+	mlx4_dbg(dev, "SMP firewall set_attribute_mask = 0x%x\n",
+		 set_attr_mask);
+
+	MLX4_GET(getresp_attr_mask, mailbox->buf,
+		 MLX4_CMD_MAD_DEMUX_GETRESP_ATTR_OFFSET);
+	mlx4_dbg(dev, "SMP firewall getresp_attribute_mask = 0x%x\n",
+		 getresp_attr_mask);
+
+	MLX4_GET(trap_attr_mask, mailbox->buf,
+		 MLX4_CMD_MAD_DEMUX_TRAP_ATTR_OFFSET);
+	mlx4_dbg(dev, "SMP firewall trap_attribute_mask = 0x%x\n",
+		 trap_attr_mask);
+
+	MLX4_GET(traprepress_attr_mask, mailbox->buf,
+		 MLX4_CMD_MAD_DEMUX_TRAP_REPRESS_ATTR_OFFSET);
+	mlx4_dbg(dev, "SMP firewall traprepress_attribute_mask = 0x%x\n",
+		 traprepress_attr_mask);
+
+	if (set_attr_mask && getresp_attr_mask && trap_attr_mask &&
+	    traprepress_attr_mask)
+		return 1;
+
+	return 0;
+}
+
+int mlx4_config_mad_demux(struct mlx4_dev *dev)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	int err;
+
+	/* Check if mad_demux is supported */
+	if (!(dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_MAD_DEMUX))
+		return 0;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox)) {
+		mlx4_warn(dev, "Failed to allocate mailbox for cmd MAD_DEMUX");
+		return -ENOMEM;
+	}
+
+	/* Query mad_demux to find out which MADs are handled by internal sma */
+	err = mlx4_cmd_box(dev, 0, mailbox->dma, 0x01 /* subn mgmt class */,
+			   MLX4_CMD_MAD_DEMUX_QUERY_RESTR, MLX4_CMD_MAD_DEMUX,
+			   MLX4_CMD_TIME_CLASS_B, MLX4_CMD_NATIVE);
+	if (err) {
+		mlx4_warn(dev, "MLX4_CMD_MAD_DEMUX: query restrictions failed (%d)\n",
+			  err);
+		goto out;
+	}
+
+	if (mlx4_check_smp_firewall_active(dev, mailbox))
+		dev->flags |= MLX4_FLAG_SECURE_HOST;
+
+	/* Config mad_demux to handle all MADs returned by the query above */
+	err = mlx4_cmd(dev, mailbox->dma, 0x01 /* subn mgmt class */,
+		       MLX4_CMD_MAD_DEMUX_CONFIG, MLX4_CMD_MAD_DEMUX,
+		       MLX4_CMD_TIME_CLASS_B, MLX4_CMD_NATIVE);
+	if (err) {
+		mlx4_warn(dev, "MLX4_CMD_MAD_DEMUX: configure failed (%d)\n", err);
+		goto out;
+	}
+
+	if (dev->flags & MLX4_FLAG_SECURE_HOST)
+		mlx4_warn(dev, "HCA operating in secure-host mode. SMP firewall activated.\n");
+out:
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return err;
+}
+
+/* Access Reg commands */
+enum mlx4_access_reg_masks {
+	MLX4_ACCESS_REG_STATUS_MASK = 0x7f,
+	MLX4_ACCESS_REG_METHOD_MASK = 0x7f,
+	MLX4_ACCESS_REG_LEN_MASK = 0x7ff
+};
+
+struct mlx4_access_reg {
+	__be16 constant1;
+	u8 status;
+	u8 resrvd1;
+	__be16 reg_id;
+	u8 method;
+	u8 constant2;
+	__be32 resrvd2[2];
+	__be16 len_const;
+	__be16 resrvd3;
+#define MLX4_ACCESS_REG_HEADER_SIZE (20)
+	u8 reg_data[MLX4_MAILBOX_SIZE-MLX4_ACCESS_REG_HEADER_SIZE];
+} __attribute__((__packed__));
+
+/**
+ * mlx4_ACCESS_REG - Generic access reg command.
+ * @dev: mlx4_dev.
+ * @reg_id: register ID to access.
+ * @method: Access method Read/Write.
+ * @reg_len: register length to Read/Write in bytes.
+ * @reg_data: reg_data pointer to Read/Write From/To.
+ *
+ * Access ConnectX registers FW command.
+ * Returns 0 on success and copies outbox mlx4_access_reg data
+ * field into reg_data or a negative error code.
+ */
+static int mlx4_ACCESS_REG(struct mlx4_dev *dev, u16 reg_id,
+			   enum mlx4_access_reg_method method,
+			   u16 reg_len, void *reg_data)
+{
+	struct mlx4_cmd_mailbox *inbox, *outbox;
+	struct mlx4_access_reg *inbuf, *outbuf;
+	int err;
+
+	inbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(inbox))
+		return PTR_ERR(inbox);
+
+	outbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(outbox)) {
+		mlx4_free_cmd_mailbox(dev, inbox);
+		return PTR_ERR(outbox);
+	}
+
+	inbuf = inbox->buf;
+	outbuf = outbox->buf;
+
+	inbuf->constant1 = cpu_to_be16(0x1<<11 | 0x4);
+	inbuf->constant2 = 0x1;
+	inbuf->reg_id = cpu_to_be16(reg_id);
+	inbuf->method = method & MLX4_ACCESS_REG_METHOD_MASK;
+
+	reg_len = min(reg_len, (u16)(sizeof(inbuf->reg_data)));
+	inbuf->len_const =
+		cpu_to_be16(((reg_len/4 + 1) & MLX4_ACCESS_REG_LEN_MASK) |
+			    ((0x3) << 12));
+
+	memcpy(inbuf->reg_data, reg_data, reg_len);
+	err = mlx4_cmd_box(dev, inbox->dma, outbox->dma, 0, 0,
+			   MLX4_CMD_ACCESS_REG, MLX4_CMD_TIME_CLASS_C,
+			   MLX4_CMD_WRAPPED);
+	if (err)
+		goto out;
+
+	if (outbuf->status & MLX4_ACCESS_REG_STATUS_MASK) {
+		err = outbuf->status & MLX4_ACCESS_REG_STATUS_MASK;
+		mlx4_err(dev,
+			 "MLX4_CMD_ACCESS_REG(%x) returned REG status (%x)\n",
+			 reg_id, err);
+		goto out;
+	}
+
+	memcpy(reg_data, outbuf->reg_data, reg_len);
+out:
+	mlx4_free_cmd_mailbox(dev, inbox);
+	mlx4_free_cmd_mailbox(dev, outbox);
+	return err;
+}
+
+/* ConnectX registers IDs */
+enum mlx4_reg_id {
+	MLX4_REG_ID_PTYS = 0x5004,
+};
+
+/**
+ * mlx4_ACCESS_PTYS_REG - Access PTYs (Port Type and Speed)
+ * register
+ * @dev: mlx4_dev.
+ * @method: Access method Read/Write.
+ * @ptys_reg: PTYS register data pointer.
+ *
+ * Access ConnectX PTYS register, to Read/Write Port Type/Speed
+ * configuration
+ * Returns 0 on success or a negative error code.
+ */
+int mlx4_ACCESS_PTYS_REG(struct mlx4_dev *dev,
+			 enum mlx4_access_reg_method method,
+			 struct mlx4_ptys_reg *ptys_reg)
+{
+	return mlx4_ACCESS_REG(dev, MLX4_REG_ID_PTYS,
+			       method, sizeof(*ptys_reg), ptys_reg);
+}
+EXPORT_SYMBOL_GPL(mlx4_ACCESS_PTYS_REG);
+
+int mlx4_ACCESS_REG_wrapper(struct mlx4_dev *dev, int slave,
+			    struct mlx4_vhcr *vhcr,
+			    struct mlx4_cmd_mailbox *inbox,
+			    struct mlx4_cmd_mailbox *outbox,
+			    struct mlx4_cmd_info *cmd)
+{
+	struct mlx4_access_reg *inbuf = inbox->buf;
+	u8 method = inbuf->method & MLX4_ACCESS_REG_METHOD_MASK;
+	u16 reg_id = be16_to_cpu(inbuf->reg_id);
+
+	if (slave != mlx4_master_func_num(dev) &&
+	    method == MLX4_ACCESS_REG_WRITE)
+		return -EPERM;
+
+	if (reg_id == MLX4_REG_ID_PTYS) {
+		struct mlx4_ptys_reg *ptys_reg =
+			(struct mlx4_ptys_reg *)inbuf->reg_data;
+
+		ptys_reg->local_port =
+			mlx4_slave_convert_port(dev, slave,
+						ptys_reg->local_port);
+	}
+
+	return mlx4_cmd_box(dev, inbox->dma, outbox->dma, vhcr->in_modifier,
+			    0, MLX4_CMD_ACCESS_REG, MLX4_CMD_TIME_CLASS_C,
+			    MLX4_CMD_NATIVE);
+}
+
+static int mlx4_SET_PORT_phv_bit(struct mlx4_dev *dev, u8 port, u8 phv_bit)
+{
+#define SET_PORT_GEN_PHV_VALID	0x10
+#define SET_PORT_GEN_PHV_EN	0x80
+
+	struct mlx4_cmd_mailbox *mailbox;
+	struct mlx4_set_port_general_context *context;
+	u32 in_mod;
+	int err;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+	context = mailbox->buf;
+
+	context->flags2 |=  SET_PORT_GEN_PHV_VALID;
+	if (phv_bit)
+		context->phv_en |=  SET_PORT_GEN_PHV_EN;
+
+	in_mod = MLX4_SET_PORT_GENERAL << 8 | port;
+	err = mlx4_cmd(dev, mailbox->dma, in_mod, MLX4_SET_PORT_ETH_OPCODE,
+		       MLX4_CMD_SET_PORT, MLX4_CMD_TIME_CLASS_B,
+		       MLX4_CMD_NATIVE);
+
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return err;
+}
+
+int get_phv_bit(struct mlx4_dev *dev, u8 port, int *phv)
+{
+	int err;
+	struct mlx4_func_cap func_cap;
+
+	memset(&func_cap, 0, sizeof(func_cap));
+	err = mlx4_QUERY_FUNC_CAP(dev, port, &func_cap);
+	if (!err)
+		*phv = func_cap.flags0 & QUERY_FUNC_CAP_PHV_BIT;
+	return err;
+}
+EXPORT_SYMBOL(get_phv_bit);
+
+int set_phv_bit(struct mlx4_dev *dev, u8 port, int new_val)
+{
+	int ret;
+
+	if (mlx4_is_slave(dev))
+		return -EPERM;
+
+	if (dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_PHV_EN &&
+	    !(dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_SKIP_OUTER_VLAN)) {
+		ret = mlx4_SET_PORT_phv_bit(dev, port, new_val);
+		if (!ret)
+			dev->caps.phv_bit[port] = new_val;
+		return ret;
+	}
+
+	return -EOPNOTSUPP;
+}
+EXPORT_SYMBOL(set_phv_bit);
+
+int mlx4_get_is_vlan_offload_disabled(struct mlx4_dev *dev, u8 port,
+				      bool *vlan_offload_disabled)
+{
+	struct mlx4_func_cap func_cap;
+	int err;
+
+	memset(&func_cap, 0, sizeof(func_cap));
+	err = mlx4_QUERY_FUNC_CAP(dev, port, &func_cap);
+	if (!err)
+		*vlan_offload_disabled =
+			!!(func_cap.flags0 &
+			   QUERY_FUNC_CAP_VLAN_OFFLOAD_DISABLE);
+	return err;
+}
+EXPORT_SYMBOL(mlx4_get_is_vlan_offload_disabled);
+
+void mlx4_replace_zero_macs(struct mlx4_dev *dev)
+{
+	int i;
+	u8 mac_addr[ETH_ALEN];
+
+	dev->port_random_macs = 0;
+	for (i = 1; i <= dev->caps.num_ports; ++i)
+		if (!dev->caps.def_mac[i] &&
+		    dev->caps.port_type[i] == MLX4_PORT_TYPE_ETH) {
+			eth_random_addr(mac_addr);
+			dev->port_random_macs |= 1 << i;
+			dev->caps.def_mac[i] = ether_addr_to_u64(mac_addr);
+		}
+}
+EXPORT_SYMBOL_GPL(mlx4_replace_zero_macs);
diff --git a/drivers/net/ethernet/mellanox/mlx4/fw.h b/drivers/net/ethernet/mellanox/mlx4/fw.h
new file mode 100644
index 000000000..cf64e54ee
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx4/fw.h
@@ -0,0 +1,257 @@
+/*
+ * Copyright (c) 2004, 2005 Topspin Communications.  All rights reserved.
+ * Copyright (c) 2005, 2006, 2007, 2008 Mellanox Technologies. All rights reserved.
+ * Copyright (c) 2006, 2007 Cisco Systems.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef MLX4_FW_H
+#define MLX4_FW_H
+
+#include "mlx4.h"
+#include "icm.h"
+
+struct mlx4_mod_stat_cfg {
+	u8 log_pg_sz;
+	u8 log_pg_sz_m;
+};
+
+struct mlx4_port_cap {
+	u8  link_state;
+	u8  supported_port_types;
+	u8  suggested_type;
+	u8  default_sense;
+	u8  log_max_macs;
+	u8  log_max_vlans;
+	int ib_mtu;
+	int max_port_width;
+	int max_vl;
+	int max_tc_eth;
+	int max_gids;
+	int max_pkeys;
+	u64 def_mac;
+	u16 eth_mtu;
+	int trans_type;
+	int vendor_oui;
+	u16 wavelength;
+	u64 trans_code;
+	u8 dmfs_optimized_state;
+};
+
+struct mlx4_dev_cap {
+	int max_srq_sz;
+	int max_qp_sz;
+	int reserved_qps;
+	int max_qps;
+	int reserved_srqs;
+	int max_srqs;
+	int max_cq_sz;
+	int reserved_cqs;
+	int max_cqs;
+	int max_mpts;
+	int reserved_eqs;
+	int max_eqs;
+	int num_sys_eqs;
+	int reserved_mtts;
+	int reserved_mrws;
+	int max_requester_per_qp;
+	int max_responder_per_qp;
+	int max_rdma_global;
+	int local_ca_ack_delay;
+	int num_ports;
+	u32 max_msg_sz;
+	u16 stat_rate_support;
+	int fs_log_max_ucast_qp_range_size;
+	int fs_max_num_qp_per_entry;
+	u64 flags;
+	u64 flags2;
+	int reserved_uars;
+	int uar_size;
+	int min_page_sz;
+	int bf_reg_size;
+	int bf_regs_per_page;
+	int max_sq_sg;
+	int max_sq_desc_sz;
+	int max_rq_sg;
+	int max_rq_desc_sz;
+	int max_qp_per_mcg;
+	int reserved_mgms;
+	int max_mcgs;
+	int reserved_pds;
+	int max_pds;
+	int reserved_xrcds;
+	int max_xrcds;
+	int qpc_entry_sz;
+	int rdmarc_entry_sz;
+	int altc_entry_sz;
+	int aux_entry_sz;
+	int srq_entry_sz;
+	int cqc_entry_sz;
+	int eqc_entry_sz;
+	int dmpt_entry_sz;
+	int cmpt_entry_sz;
+	int mtt_entry_sz;
+	int resize_srq;
+	u32 bmme_flags;
+	u32 reserved_lkey;
+	u64 max_icm_sz;
+	int max_gso_sz;
+	int max_rss_tbl_sz;
+	u32 max_counters;
+	u32 dmfs_high_rate_qpn_base;
+	u32 dmfs_high_rate_qpn_range;
+	struct mlx4_rate_limit_caps rl_caps;
+	u32 health_buffer_addrs;
+	struct mlx4_port_cap port_cap[MLX4_MAX_PORTS + 1];
+	bool wol_port[MLX4_MAX_PORTS + 1];
+	bool map_clock_to_user;
+};
+
+struct mlx4_func_cap {
+	u8	num_ports;
+	u8	flags;
+	u32	pf_context_behaviour;
+	int	qp_quota;
+	int	cq_quota;
+	int	srq_quota;
+	int	mpt_quota;
+	int	mtt_quota;
+	int	max_eq;
+	int	reserved_eq;
+	int	mcg_quota;
+	struct mlx4_spec_qps spec_qps;
+	u32	reserved_lkey;
+	u8	physical_port;
+	u8	flags0;
+	u8	flags1;
+	u64	phys_port_id;
+	u32	extra_flags;
+};
+
+struct mlx4_func {
+	int	bus;
+	int	device;
+	int	function;
+	int	physical_function;
+	int	rsvd_eqs;
+	int	max_eq;
+	int	rsvd_uars;
+};
+
+struct mlx4_adapter {
+	char board_id[MLX4_BOARD_ID_LEN];
+	u8   inta_pin;
+};
+
+struct mlx4_init_hca_param {
+	u64 qpc_base;
+	u64 rdmarc_base;
+	u64 auxc_base;
+	u64 altc_base;
+	u64 srqc_base;
+	u64 cqc_base;
+	u64 eqc_base;
+	u64 mc_base;
+	u64 dmpt_base;
+	u64 cmpt_base;
+	u64 mtt_base;
+	u64 global_caps;
+	u8 log_mc_entry_sz;
+	u8 log_mc_hash_sz;
+	u16 hca_core_clock; /* Internal Clock Frequency (in MHz) */
+	u8  log_num_qps;
+	u8  log_num_srqs;
+	u8  log_num_cqs;
+	u8  log_num_eqs;
+	u16 num_sys_eqs;
+	u8  log_rd_per_qp;
+	u8  log_mc_table_sz;
+	u8  log_mpt_sz;
+	u8  log_uar_sz;
+	u8  mw_enabled;  /* Enable memory windows */
+	u8  uar_page_sz; /* log pg sz in 4k chunks */
+	u8  steering_mode; /* for QUERY_HCA */
+	u8  dmfs_high_steer_mode; /* for QUERY_HCA */
+	u64 dev_cap_enabled;
+	u16 cqe_size; /* For use only when CQE stride feature enabled */
+	u16 eqe_size; /* For use only when EQE stride feature enabled */
+	u8 rss_ip_frags;
+	u8 phv_check_en; /* for QUERY_HCA */
+};
+
+struct mlx4_init_ib_param {
+	int port_width;
+	int vl_cap;
+	int mtu_cap;
+	u16 gid_cap;
+	u16 pkey_cap;
+	int set_guid0;
+	u64 guid0;
+	int set_node_guid;
+	u64 node_guid;
+	int set_si_guid;
+	u64 si_guid;
+};
+
+struct mlx4_set_ib_param {
+	int set_si_guid;
+	int reset_qkey_viol;
+	u64 si_guid;
+	u32 cap_mask;
+};
+
+void mlx4_dev_cap_dump(struct mlx4_dev *dev, struct mlx4_dev_cap *dev_cap);
+int mlx4_QUERY_DEV_CAP(struct mlx4_dev *dev, struct mlx4_dev_cap *dev_cap);
+int mlx4_QUERY_PORT(struct mlx4_dev *dev, int port, struct mlx4_port_cap *port_cap);
+int mlx4_QUERY_FUNC_CAP(struct mlx4_dev *dev, u8 gen_or_port,
+			struct mlx4_func_cap *func_cap);
+int mlx4_QUERY_FUNC_CAP_wrapper(struct mlx4_dev *dev, int slave,
+				struct mlx4_vhcr *vhcr,
+				struct mlx4_cmd_mailbox *inbox,
+				struct mlx4_cmd_mailbox *outbox,
+				struct mlx4_cmd_info *cmd);
+int mlx4_QUERY_FUNC(struct mlx4_dev *dev, struct mlx4_func *func, int slave);
+int mlx4_MAP_FA(struct mlx4_dev *dev, struct mlx4_icm *icm);
+int mlx4_UNMAP_FA(struct mlx4_dev *dev);
+int mlx4_RUN_FW(struct mlx4_dev *dev);
+int mlx4_QUERY_FW(struct mlx4_dev *dev);
+int mlx4_QUERY_ADAPTER(struct mlx4_dev *dev, struct mlx4_adapter *adapter);
+int mlx4_INIT_HCA(struct mlx4_dev *dev, struct mlx4_init_hca_param *param);
+int mlx4_QUERY_HCA(struct mlx4_dev *dev, struct mlx4_init_hca_param *param);
+int mlx4_CLOSE_HCA(struct mlx4_dev *dev, int panic);
+int mlx4_map_cmd(struct mlx4_dev *dev, u16 op, struct mlx4_icm *icm, u64 virt);
+int mlx4_SET_ICM_SIZE(struct mlx4_dev *dev, u64 icm_size, u64 *aux_pages);
+int mlx4_MAP_ICM_AUX(struct mlx4_dev *dev, struct mlx4_icm *icm);
+int mlx4_UNMAP_ICM_AUX(struct mlx4_dev *dev);
+int mlx4_NOP(struct mlx4_dev *dev);
+int mlx4_MOD_STAT_CFG(struct mlx4_dev *dev, struct mlx4_mod_stat_cfg *cfg);
+void mlx4_opreq_action(struct work_struct *work);
+
+#endif /* MLX4_FW_H */
diff --git a/drivers/net/ethernet/mellanox/mlx4/fw_qos.c b/drivers/net/ethernet/mellanox/mlx4/fw_qos.c
new file mode 100644
index 000000000..3a09d7122
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx4/fw_qos.c
@@ -0,0 +1,289 @@
+/*
+ * Copyright (c) 2004, 2005 Topspin Communications.  All rights reserved.
+ * Copyright (c) 2005, 2006, 2007, 2008 Mellanox Technologies.
+ * All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/export.h>
+#include "fw_qos.h"
+#include "fw.h"
+
+enum {
+	/* allocate vpp opcode modifiers */
+	MLX4_ALLOCATE_VPP_ALLOCATE	= 0x0,
+	MLX4_ALLOCATE_VPP_QUERY		= 0x1
+};
+
+enum {
+	/* set vport qos opcode modifiers */
+	MLX4_SET_VPORT_QOS_SET		= 0x0,
+	MLX4_SET_VPORT_QOS_QUERY	= 0x1
+};
+
+struct mlx4_set_port_prio2tc_context {
+	u8 prio2tc[4];
+};
+
+struct mlx4_port_scheduler_tc_cfg_be {
+	__be16 pg;
+	__be16 bw_precentage;
+	__be16 max_bw_units; /* 3-100Mbps, 4-1Gbps, other values - reserved */
+	__be16 max_bw_value;
+};
+
+struct mlx4_set_port_scheduler_context {
+	struct mlx4_port_scheduler_tc_cfg_be tc[MLX4_NUM_TC];
+};
+
+/* Granular Qos (per VF) section */
+struct mlx4_alloc_vpp_param {
+	__be32 available_vpp;
+	__be32 vpp_p_up[MLX4_NUM_UP];
+};
+
+struct mlx4_prio_qos_param {
+	__be32 bw_share;
+	__be32 max_avg_bw;
+	__be32 reserved;
+	__be32 enable;
+	__be32 reserved1[4];
+};
+
+struct mlx4_set_vport_context {
+	__be32 reserved[8];
+	struct mlx4_prio_qos_param qos_p_up[MLX4_NUM_UP];
+};
+
+int mlx4_SET_PORT_PRIO2TC(struct mlx4_dev *dev, u8 port, u8 *prio2tc)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	struct mlx4_set_port_prio2tc_context *context;
+	int err;
+	u32 in_mod;
+	int i;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+
+	context = mailbox->buf;
+
+	for (i = 0; i < MLX4_NUM_UP; i += 2)
+		context->prio2tc[i >> 1] = prio2tc[i] << 4 | prio2tc[i + 1];
+
+	in_mod = MLX4_SET_PORT_PRIO2TC << 8 | port;
+	err = mlx4_cmd(dev, mailbox->dma, in_mod, 1, MLX4_CMD_SET_PORT,
+		       MLX4_CMD_TIME_CLASS_B, MLX4_CMD_NATIVE);
+
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return err;
+}
+EXPORT_SYMBOL(mlx4_SET_PORT_PRIO2TC);
+
+int mlx4_SET_PORT_SCHEDULER(struct mlx4_dev *dev, u8 port, u8 *tc_tx_bw,
+			    u8 *pg, u16 *ratelimit)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	struct mlx4_set_port_scheduler_context *context;
+	int err;
+	u32 in_mod;
+	int i;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+
+	context = mailbox->buf;
+
+	for (i = 0; i < MLX4_NUM_TC; i++) {
+		struct mlx4_port_scheduler_tc_cfg_be *tc = &context->tc[i];
+		u16 r;
+
+		if (ratelimit && ratelimit[i]) {
+			if (ratelimit[i] <= MLX4_MAX_100M_UNITS_VAL) {
+				r = ratelimit[i];
+				tc->max_bw_units =
+					htons(MLX4_RATELIMIT_100M_UNITS);
+			} else {
+				r = ratelimit[i] / 10;
+				tc->max_bw_units =
+					htons(MLX4_RATELIMIT_1G_UNITS);
+			}
+			tc->max_bw_value = htons(r);
+		} else {
+			tc->max_bw_value = htons(MLX4_RATELIMIT_DEFAULT);
+			tc->max_bw_units = htons(MLX4_RATELIMIT_1G_UNITS);
+		}
+
+		tc->pg = htons(pg[i]);
+		tc->bw_precentage = htons(tc_tx_bw[i]);
+	}
+
+	in_mod = MLX4_SET_PORT_SCHEDULER << 8 | port;
+	err = mlx4_cmd(dev, mailbox->dma, in_mod, 1, MLX4_CMD_SET_PORT,
+		       MLX4_CMD_TIME_CLASS_B, MLX4_CMD_NATIVE);
+
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return err;
+}
+EXPORT_SYMBOL(mlx4_SET_PORT_SCHEDULER);
+
+int mlx4_ALLOCATE_VPP_get(struct mlx4_dev *dev, u8 port,
+			  u16 *available_vpp, u8 *vpp_p_up)
+{
+	int i;
+	int err;
+	struct mlx4_cmd_mailbox *mailbox;
+	struct mlx4_alloc_vpp_param *out_param;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+
+	out_param = mailbox->buf;
+
+	err = mlx4_cmd_box(dev, 0, mailbox->dma, port,
+			   MLX4_ALLOCATE_VPP_QUERY,
+			   MLX4_CMD_ALLOCATE_VPP,
+			   MLX4_CMD_TIME_CLASS_A,
+			   MLX4_CMD_NATIVE);
+	if (err)
+		goto out;
+
+	/* Total number of supported VPPs */
+	*available_vpp = (u16)be32_to_cpu(out_param->available_vpp);
+
+	for (i = 0; i < MLX4_NUM_UP; i++)
+		vpp_p_up[i] = (u8)be32_to_cpu(out_param->vpp_p_up[i]);
+
+out:
+	mlx4_free_cmd_mailbox(dev, mailbox);
+
+	return err;
+}
+EXPORT_SYMBOL(mlx4_ALLOCATE_VPP_get);
+
+int mlx4_ALLOCATE_VPP_set(struct mlx4_dev *dev, u8 port, u8 *vpp_p_up)
+{
+	int i;
+	int err;
+	struct mlx4_cmd_mailbox *mailbox;
+	struct mlx4_alloc_vpp_param *in_param;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+
+	in_param = mailbox->buf;
+
+	for (i = 0; i < MLX4_NUM_UP; i++)
+		in_param->vpp_p_up[i] = cpu_to_be32(vpp_p_up[i]);
+
+	err = mlx4_cmd(dev, mailbox->dma, port,
+		       MLX4_ALLOCATE_VPP_ALLOCATE,
+		       MLX4_CMD_ALLOCATE_VPP,
+		       MLX4_CMD_TIME_CLASS_A,
+		       MLX4_CMD_NATIVE);
+
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return err;
+}
+EXPORT_SYMBOL(mlx4_ALLOCATE_VPP_set);
+
+int mlx4_SET_VPORT_QOS_get(struct mlx4_dev *dev, u8 port, u8 vport,
+			   struct mlx4_vport_qos_param *out_param)
+{
+	int i;
+	int err;
+	struct mlx4_cmd_mailbox *mailbox;
+	struct mlx4_set_vport_context *ctx;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+
+	ctx = mailbox->buf;
+
+	err = mlx4_cmd_box(dev, 0, mailbox->dma, (vport << 8) | port,
+			   MLX4_SET_VPORT_QOS_QUERY,
+			   MLX4_CMD_SET_VPORT_QOS,
+			   MLX4_CMD_TIME_CLASS_A,
+			   MLX4_CMD_NATIVE);
+	if (err)
+		goto out;
+
+	for (i = 0; i < MLX4_NUM_UP; i++) {
+		out_param[i].bw_share = be32_to_cpu(ctx->qos_p_up[i].bw_share);
+		out_param[i].max_avg_bw =
+			be32_to_cpu(ctx->qos_p_up[i].max_avg_bw);
+		out_param[i].enable =
+			!!(be32_to_cpu(ctx->qos_p_up[i].enable) & 31);
+	}
+
+out:
+	mlx4_free_cmd_mailbox(dev, mailbox);
+
+	return err;
+}
+EXPORT_SYMBOL(mlx4_SET_VPORT_QOS_get);
+
+int mlx4_SET_VPORT_QOS_set(struct mlx4_dev *dev, u8 port, u8 vport,
+			   struct mlx4_vport_qos_param *in_param)
+{
+	int i;
+	int err;
+	struct mlx4_cmd_mailbox *mailbox;
+	struct mlx4_set_vport_context *ctx;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+
+	ctx = mailbox->buf;
+
+	for (i = 0; i < MLX4_NUM_UP; i++) {
+		ctx->qos_p_up[i].bw_share = cpu_to_be32(in_param[i].bw_share);
+		ctx->qos_p_up[i].max_avg_bw =
+				cpu_to_be32(in_param[i].max_avg_bw);
+		ctx->qos_p_up[i].enable =
+				cpu_to_be32(in_param[i].enable << 31);
+	}
+
+	err = mlx4_cmd(dev, mailbox->dma, (vport << 8) | port,
+		       MLX4_SET_VPORT_QOS_SET,
+		       MLX4_CMD_SET_VPORT_QOS,
+		       MLX4_CMD_TIME_CLASS_A,
+		       MLX4_CMD_NATIVE);
+
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return err;
+}
+EXPORT_SYMBOL(mlx4_SET_VPORT_QOS_set);
diff --git a/drivers/net/ethernet/mellanox/mlx4/fw_qos.h b/drivers/net/ethernet/mellanox/mlx4/fw_qos.h
new file mode 100644
index 000000000..954b86faa
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx4/fw_qos.h
@@ -0,0 +1,145 @@
+/*
+ * Copyright (c) 2004, 2005 Topspin Communications.  All rights reserved.
+ * Copyright (c) 2005, 2006, 2007, 2008 Mellanox Technologies.
+ * All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef MLX4_FW_QOS_H
+#define MLX4_FW_QOS_H
+
+#include <linux/mlx4/cmd.h>
+#include <linux/mlx4/device.h>
+
+#define MLX4_NUM_UP 8
+#define MLX4_NUM_TC 8
+
+/* Default supported priorities for VPP allocation */
+#define MLX4_DEFAULT_QOS_PRIO (0)
+
+/* Derived from FW feature definition, 0 is the default vport fo all QPs */
+#define MLX4_VPP_DEFAULT_VPORT (0)
+
+struct mlx4_vport_qos_param {
+	u32 bw_share;
+	u32 max_avg_bw;
+	u8 enable;
+};
+
+/**
+ * mlx4_SET_PORT_PRIO2TC - This routine maps user priorities to traffic
+ * classes of a given port and device.
+ *
+ * @dev: mlx4_dev.
+ * @port: Physical port number.
+ * @prio2tc: Array of TC associated with each priorities.
+ *
+ * Returns 0 on success or a negative mlx4_core errno code.
+ **/
+int mlx4_SET_PORT_PRIO2TC(struct mlx4_dev *dev, u8 port, u8 *prio2tc);
+
+/**
+ * mlx4_SET_PORT_SCHEDULER - This routine configures the arbitration between
+ * traffic classes (ETS) and configured rate limit for traffic classes.
+ * tc_tx_bw, pg and ratelimit are arrays where each index represents a TC.
+ * The description for those parameters below refers to a single TC.
+ *
+ * @dev: mlx4_dev.
+ * @port: Physical port number.
+ * @tc_tx_bw: The percentage of the bandwidth allocated for traffic class
+ *  within a TC group. The sum of the bw_percentage of all the traffic
+ *  classes within a TC group must equal 100% for correct operation.
+ * @pg: The TC group the traffic class is associated with.
+ * @ratelimit: The maximal bandwidth allowed for the use by this traffic class.
+ *
+ * Returns 0 on success or a negative mlx4_core errno code.
+ **/
+int mlx4_SET_PORT_SCHEDULER(struct mlx4_dev *dev, u8 port, u8 *tc_tx_bw,
+			    u8 *pg, u16 *ratelimit);
+/**
+ * mlx4_ALLOCATE_VPP_get - Query port VPP available resources and allocation.
+ * Before distribution of VPPs to priorities, only available_vpp is returned.
+ * After initialization it returns the distribution of VPPs among priorities.
+ *
+ * @dev: mlx4_dev.
+ * @port: Physical port number.
+ * @available_vpp: Pointer to variable where number of available VPPs is stored
+ * @vpp_p_up: Distribution of VPPs to priorities is stored in this array
+ *
+ * Returns 0 on success or a negative mlx4_core errno code.
+ **/
+int mlx4_ALLOCATE_VPP_get(struct mlx4_dev *dev, u8 port,
+			  u16 *available_vpp, u8 *vpp_p_up);
+/**
+ * mlx4_ALLOCATE_VPP_set - Distribution of VPPs among differnt priorities.
+ * The total number of VPPs assigned to all for a port must not exceed
+ * the value reported by available_vpp in mlx4_ALLOCATE_VPP_get.
+ * VPP allocation is allowed only after the port type has been set,
+ * and while no QPs are open for this port.
+ *
+ * @dev: mlx4_dev.
+ * @port: Physical port number.
+ * @vpp_p_up: Allocation of VPPs to different priorities.
+ *
+ * Returns 0 on success or a negative mlx4_core errno code.
+ **/
+int mlx4_ALLOCATE_VPP_set(struct mlx4_dev *dev, u8 port, u8 *vpp_p_up);
+
+/**
+ * mlx4_SET_VPORT_QOS_get - Query QoS proporties of a Vport.
+ * Each priority allowed for the Vport is assigned with a share of the BW,
+ * and a BW limitation. This commands query the current QoS values.
+ *
+ * @dev: mlx4_dev.
+ * @port: Physical port number.
+ * @vport: Vport id.
+ * @out_param: Array of mlx4_vport_qos_param that will contain the values.
+ *
+ * Returns 0 on success or a negative mlx4_core errno code.
+ **/
+int mlx4_SET_VPORT_QOS_get(struct mlx4_dev *dev, u8 port, u8 vport,
+			   struct mlx4_vport_qos_param *out_param);
+
+/**
+ * mlx4_SET_VPORT_QOS_set - Set QoS proporties of a Vport.
+ * QoS parameters can be modified at any time, but must be initialized
+ * before any QP is associated with the VPort.
+ *
+ * @dev: mlx4_dev.
+ * @port: Physical port number.
+ * @vport: Vport id.
+ * @in_param: Array of mlx4_vport_qos_param which holds the requested values.
+ *
+ * Returns 0 on success or a negative mlx4_core errno code.
+ **/
+int mlx4_SET_VPORT_QOS_set(struct mlx4_dev *dev, u8 port, u8 vport,
+			   struct mlx4_vport_qos_param *in_param);
+
+#endif /* MLX4_FW_QOS_H */
diff --git a/drivers/net/ethernet/mellanox/mlx4/icm.c b/drivers/net/ethernet/mellanox/mlx4/icm.c
new file mode 100644
index 000000000..59b8b3c73
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx4/icm.c
@@ -0,0 +1,493 @@
+/*
+ * Copyright (c) 2005, 2006, 2007, 2008 Mellanox Technologies. All rights reserved.
+ * Copyright (c) 2006, 2007 Cisco Systems, Inc.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/errno.h>
+#include <linux/mm.h>
+#include <linux/scatterlist.h>
+#include <linux/slab.h>
+
+#include <linux/mlx4/cmd.h>
+
+#include "mlx4.h"
+#include "icm.h"
+#include "fw.h"
+
+/*
+ * We allocate in as big chunks as we can, up to a maximum of 256 KB
+ * per chunk. Note that the chunks are not necessarily in contiguous
+ * physical memory.
+ */
+enum {
+	MLX4_ICM_ALLOC_SIZE	= 1 << 18,
+	MLX4_TABLE_CHUNK_SIZE	= 1 << 18,
+};
+
+static void mlx4_free_icm_pages(struct mlx4_dev *dev, struct mlx4_icm_chunk *chunk)
+{
+	int i;
+
+	if (chunk->nsg > 0)
+		dma_unmap_sg(&dev->persist->pdev->dev, chunk->sg, chunk->npages,
+			     DMA_BIDIRECTIONAL);
+
+	for (i = 0; i < chunk->npages; ++i)
+		__free_pages(sg_page(&chunk->sg[i]),
+			     get_order(chunk->sg[i].length));
+}
+
+static void mlx4_free_icm_coherent(struct mlx4_dev *dev, struct mlx4_icm_chunk *chunk)
+{
+	int i;
+
+	for (i = 0; i < chunk->npages; ++i)
+		dma_free_coherent(&dev->persist->pdev->dev,
+				  chunk->buf[i].size,
+				  chunk->buf[i].addr,
+				  chunk->buf[i].dma_addr);
+}
+
+void mlx4_free_icm(struct mlx4_dev *dev, struct mlx4_icm *icm, int coherent)
+{
+	struct mlx4_icm_chunk *chunk, *tmp;
+
+	if (!icm)
+		return;
+
+	list_for_each_entry_safe(chunk, tmp, &icm->chunk_list, list) {
+		if (coherent)
+			mlx4_free_icm_coherent(dev, chunk);
+		else
+			mlx4_free_icm_pages(dev, chunk);
+
+		kfree(chunk);
+	}
+
+	kfree(icm);
+}
+
+static int mlx4_alloc_icm_pages(struct scatterlist *mem, int order,
+				gfp_t gfp_mask, int node)
+{
+	struct page *page;
+
+	page = alloc_pages_node(node, gfp_mask, order);
+	if (!page) {
+		page = alloc_pages(gfp_mask, order);
+		if (!page)
+			return -ENOMEM;
+	}
+
+	sg_set_page(mem, page, PAGE_SIZE << order, 0);
+	return 0;
+}
+
+static int mlx4_alloc_icm_coherent(struct device *dev, struct mlx4_icm_buf *buf,
+				   int order, gfp_t gfp_mask)
+{
+	buf->addr = dma_alloc_coherent(dev, PAGE_SIZE << order,
+				       &buf->dma_addr, gfp_mask);
+	if (!buf->addr)
+		return -ENOMEM;
+
+	if (offset_in_page(buf->addr)) {
+		dma_free_coherent(dev, PAGE_SIZE << order, buf->addr,
+				  buf->dma_addr);
+		return -ENOMEM;
+	}
+
+	buf->size = PAGE_SIZE << order;
+	return 0;
+}
+
+struct mlx4_icm *mlx4_alloc_icm(struct mlx4_dev *dev, int npages,
+				gfp_t gfp_mask, int coherent)
+{
+	struct mlx4_icm *icm;
+	struct mlx4_icm_chunk *chunk = NULL;
+	int cur_order;
+	gfp_t mask;
+	int ret;
+
+	/* We use sg_set_buf for coherent allocs, which assumes low memory */
+	BUG_ON(coherent && (gfp_mask & __GFP_HIGHMEM));
+
+	icm = kmalloc_node(sizeof(*icm),
+			   gfp_mask & ~(__GFP_HIGHMEM | __GFP_NOWARN),
+			   dev->numa_node);
+	if (!icm) {
+		icm = kmalloc(sizeof(*icm),
+			      gfp_mask & ~(__GFP_HIGHMEM | __GFP_NOWARN));
+		if (!icm)
+			return NULL;
+	}
+
+	icm->refcount = 0;
+	INIT_LIST_HEAD(&icm->chunk_list);
+
+	cur_order = get_order(MLX4_ICM_ALLOC_SIZE);
+
+	while (npages > 0) {
+		if (!chunk) {
+			chunk = kzalloc_node(sizeof(*chunk),
+					     gfp_mask & ~(__GFP_HIGHMEM |
+							  __GFP_NOWARN),
+					     dev->numa_node);
+			if (!chunk) {
+				chunk = kzalloc(sizeof(*chunk),
+						gfp_mask & ~(__GFP_HIGHMEM |
+							     __GFP_NOWARN));
+				if (!chunk)
+					goto fail;
+			}
+			chunk->coherent = coherent;
+
+			if (!coherent)
+				sg_init_table(chunk->sg, MLX4_ICM_CHUNK_LEN);
+			list_add_tail(&chunk->list, &icm->chunk_list);
+		}
+
+		while (1 << cur_order > npages)
+			--cur_order;
+
+		mask = gfp_mask;
+		if (cur_order)
+			mask &= ~__GFP_DIRECT_RECLAIM;
+
+		if (coherent)
+			ret = mlx4_alloc_icm_coherent(&dev->persist->pdev->dev,
+						&chunk->buf[chunk->npages],
+						cur_order, mask);
+		else
+			ret = mlx4_alloc_icm_pages(&chunk->sg[chunk->npages],
+						   cur_order, mask,
+						   dev->numa_node);
+
+		if (ret) {
+			if (--cur_order < 0)
+				goto fail;
+			else
+				continue;
+		}
+
+		++chunk->npages;
+
+		if (coherent)
+			++chunk->nsg;
+		else if (chunk->npages == MLX4_ICM_CHUNK_LEN) {
+			chunk->nsg = dma_map_sg(&dev->persist->pdev->dev,
+						chunk->sg, chunk->npages,
+						DMA_BIDIRECTIONAL);
+
+			if (!chunk->nsg)
+				goto fail;
+		}
+
+		if (chunk->npages == MLX4_ICM_CHUNK_LEN)
+			chunk = NULL;
+
+		npages -= 1 << cur_order;
+	}
+
+	if (!coherent && chunk) {
+		chunk->nsg = dma_map_sg(&dev->persist->pdev->dev, chunk->sg,
+					chunk->npages, DMA_BIDIRECTIONAL);
+
+		if (!chunk->nsg)
+			goto fail;
+	}
+
+	return icm;
+
+fail:
+	mlx4_free_icm(dev, icm, coherent);
+	return NULL;
+}
+
+static int mlx4_MAP_ICM(struct mlx4_dev *dev, struct mlx4_icm *icm, u64 virt)
+{
+	return mlx4_map_cmd(dev, MLX4_CMD_MAP_ICM, icm, virt);
+}
+
+static int mlx4_UNMAP_ICM(struct mlx4_dev *dev, u64 virt, u32 page_count)
+{
+	return mlx4_cmd(dev, virt, page_count, 0, MLX4_CMD_UNMAP_ICM,
+			MLX4_CMD_TIME_CLASS_B, MLX4_CMD_NATIVE);
+}
+
+int mlx4_MAP_ICM_AUX(struct mlx4_dev *dev, struct mlx4_icm *icm)
+{
+	return mlx4_map_cmd(dev, MLX4_CMD_MAP_ICM_AUX, icm, -1);
+}
+
+int mlx4_UNMAP_ICM_AUX(struct mlx4_dev *dev)
+{
+	return mlx4_cmd(dev, 0, 0, 0, MLX4_CMD_UNMAP_ICM_AUX,
+			MLX4_CMD_TIME_CLASS_B, MLX4_CMD_NATIVE);
+}
+
+int mlx4_table_get(struct mlx4_dev *dev, struct mlx4_icm_table *table, u32 obj)
+{
+	u32 i = (obj & (table->num_obj - 1)) /
+			(MLX4_TABLE_CHUNK_SIZE / table->obj_size);
+	int ret = 0;
+
+	mutex_lock(&table->mutex);
+
+	if (table->icm[i]) {
+		++table->icm[i]->refcount;
+		goto out;
+	}
+
+	table->icm[i] = mlx4_alloc_icm(dev, MLX4_TABLE_CHUNK_SIZE >> PAGE_SHIFT,
+				       (table->lowmem ? GFP_KERNEL : GFP_HIGHUSER) |
+				       __GFP_NOWARN, table->coherent);
+	if (!table->icm[i]) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	if (mlx4_MAP_ICM(dev, table->icm[i], table->virt +
+			 (u64) i * MLX4_TABLE_CHUNK_SIZE)) {
+		mlx4_free_icm(dev, table->icm[i], table->coherent);
+		table->icm[i] = NULL;
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	++table->icm[i]->refcount;
+
+out:
+	mutex_unlock(&table->mutex);
+	return ret;
+}
+
+void mlx4_table_put(struct mlx4_dev *dev, struct mlx4_icm_table *table, u32 obj)
+{
+	u32 i;
+	u64 offset;
+
+	i = (obj & (table->num_obj - 1)) / (MLX4_TABLE_CHUNK_SIZE / table->obj_size);
+
+	mutex_lock(&table->mutex);
+
+	if (--table->icm[i]->refcount == 0) {
+		offset = (u64) i * MLX4_TABLE_CHUNK_SIZE;
+		mlx4_UNMAP_ICM(dev, table->virt + offset,
+			       MLX4_TABLE_CHUNK_SIZE / MLX4_ICM_PAGE_SIZE);
+		mlx4_free_icm(dev, table->icm[i], table->coherent);
+		table->icm[i] = NULL;
+	}
+
+	mutex_unlock(&table->mutex);
+}
+
+void *mlx4_table_find(struct mlx4_icm_table *table, u32 obj,
+			dma_addr_t *dma_handle)
+{
+	int offset, dma_offset, i;
+	u64 idx;
+	struct mlx4_icm_chunk *chunk;
+	struct mlx4_icm *icm;
+	void *addr = NULL;
+
+	if (!table->lowmem)
+		return NULL;
+
+	mutex_lock(&table->mutex);
+
+	idx = (u64) (obj & (table->num_obj - 1)) * table->obj_size;
+	icm = table->icm[idx / MLX4_TABLE_CHUNK_SIZE];
+	dma_offset = offset = idx % MLX4_TABLE_CHUNK_SIZE;
+
+	if (!icm)
+		goto out;
+
+	list_for_each_entry(chunk, &icm->chunk_list, list) {
+		for (i = 0; i < chunk->npages; ++i) {
+			dma_addr_t dma_addr;
+			size_t len;
+
+			if (table->coherent) {
+				len = chunk->buf[i].size;
+				dma_addr = chunk->buf[i].dma_addr;
+				addr = chunk->buf[i].addr;
+			} else {
+				struct page *page;
+
+				len = sg_dma_len(&chunk->sg[i]);
+				dma_addr = sg_dma_address(&chunk->sg[i]);
+
+				/* XXX: we should never do this for highmem
+				 * allocation.  This function either needs
+				 * to be split, or the kernel virtual address
+				 * return needs to be made optional.
+				 */
+				page = sg_page(&chunk->sg[i]);
+				addr = lowmem_page_address(page);
+			}
+
+			if (dma_handle && dma_offset >= 0) {
+				if (len > dma_offset)
+					*dma_handle = dma_addr + dma_offset;
+				dma_offset -= len;
+			}
+
+			/*
+			 * DMA mapping can merge pages but not split them,
+			 * so if we found the page, dma_handle has already
+			 * been assigned to.
+			 */
+			if (len > offset)
+				goto out;
+			offset -= len;
+		}
+	}
+
+	addr = NULL;
+out:
+	mutex_unlock(&table->mutex);
+	return addr ? addr + offset : NULL;
+}
+
+int mlx4_table_get_range(struct mlx4_dev *dev, struct mlx4_icm_table *table,
+			 u32 start, u32 end)
+{
+	int inc = MLX4_TABLE_CHUNK_SIZE / table->obj_size;
+	int err;
+	u32 i;
+
+	for (i = start; i <= end; i += inc) {
+		err = mlx4_table_get(dev, table, i);
+		if (err)
+			goto fail;
+	}
+
+	return 0;
+
+fail:
+	while (i > start) {
+		i -= inc;
+		mlx4_table_put(dev, table, i);
+	}
+
+	return err;
+}
+
+void mlx4_table_put_range(struct mlx4_dev *dev, struct mlx4_icm_table *table,
+			  u32 start, u32 end)
+{
+	u32 i;
+
+	for (i = start; i <= end; i += MLX4_TABLE_CHUNK_SIZE / table->obj_size)
+		mlx4_table_put(dev, table, i);
+}
+
+int mlx4_init_icm_table(struct mlx4_dev *dev, struct mlx4_icm_table *table,
+			u64 virt, int obj_size,	u32 nobj, int reserved,
+			int use_lowmem, int use_coherent)
+{
+	int obj_per_chunk;
+	int num_icm;
+	unsigned chunk_size;
+	int i;
+	u64 size;
+
+	obj_per_chunk = MLX4_TABLE_CHUNK_SIZE / obj_size;
+	if (WARN_ON(!obj_per_chunk))
+		return -EINVAL;
+	num_icm = DIV_ROUND_UP(nobj, obj_per_chunk);
+
+	table->icm      = kvcalloc(num_icm, sizeof(*table->icm), GFP_KERNEL);
+	if (!table->icm)
+		return -ENOMEM;
+	table->virt     = virt;
+	table->num_icm  = num_icm;
+	table->num_obj  = nobj;
+	table->obj_size = obj_size;
+	table->lowmem   = use_lowmem;
+	table->coherent = use_coherent;
+	mutex_init(&table->mutex);
+
+	size = (u64) nobj * obj_size;
+	for (i = 0; i * MLX4_TABLE_CHUNK_SIZE < reserved * obj_size; ++i) {
+		chunk_size = MLX4_TABLE_CHUNK_SIZE;
+		if ((i + 1) * MLX4_TABLE_CHUNK_SIZE > size)
+			chunk_size = PAGE_ALIGN(size -
+					i * MLX4_TABLE_CHUNK_SIZE);
+
+		table->icm[i] = mlx4_alloc_icm(dev, chunk_size >> PAGE_SHIFT,
+					       (use_lowmem ? GFP_KERNEL : GFP_HIGHUSER) |
+					       __GFP_NOWARN, use_coherent);
+		if (!table->icm[i])
+			goto err;
+		if (mlx4_MAP_ICM(dev, table->icm[i], virt + i * MLX4_TABLE_CHUNK_SIZE)) {
+			mlx4_free_icm(dev, table->icm[i], use_coherent);
+			table->icm[i] = NULL;
+			goto err;
+		}
+
+		/*
+		 * Add a reference to this ICM chunk so that it never
+		 * gets freed (since it contains reserved firmware objects).
+		 */
+		++table->icm[i]->refcount;
+	}
+
+	return 0;
+
+err:
+	for (i = 0; i < num_icm; ++i)
+		if (table->icm[i]) {
+			mlx4_UNMAP_ICM(dev, virt + i * MLX4_TABLE_CHUNK_SIZE,
+				       MLX4_TABLE_CHUNK_SIZE / MLX4_ICM_PAGE_SIZE);
+			mlx4_free_icm(dev, table->icm[i], use_coherent);
+		}
+
+	kvfree(table->icm);
+
+	return -ENOMEM;
+}
+
+void mlx4_cleanup_icm_table(struct mlx4_dev *dev, struct mlx4_icm_table *table)
+{
+	int i;
+
+	for (i = 0; i < table->num_icm; ++i)
+		if (table->icm[i]) {
+			mlx4_UNMAP_ICM(dev, table->virt + i * MLX4_TABLE_CHUNK_SIZE,
+				       MLX4_TABLE_CHUNK_SIZE / MLX4_ICM_PAGE_SIZE);
+			mlx4_free_icm(dev, table->icm[i], table->coherent);
+		}
+
+	kvfree(table->icm);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx4/icm.h b/drivers/net/ethernet/mellanox/mlx4/icm.h
new file mode 100644
index 000000000..d199874b1
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx4/icm.h
@@ -0,0 +1,144 @@
+/*
+ * Copyright (c) 2005, 2006, 2007, 2008 Mellanox Technologies. All rights reserved.
+ * Copyright (c) 2006, 2007 Cisco Systems, Inc.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef MLX4_ICM_H
+#define MLX4_ICM_H
+
+#include <linux/list.h>
+#include <linux/pci.h>
+#include <linux/mutex.h>
+
+#define MLX4_ICM_CHUNK_LEN						\
+	((256 - sizeof(struct list_head) - 2 * sizeof(int)) /		\
+	 (sizeof(struct scatterlist)))
+
+enum {
+	MLX4_ICM_PAGE_SHIFT	= 12,
+	MLX4_ICM_PAGE_SIZE	= 1 << MLX4_ICM_PAGE_SHIFT,
+};
+
+struct mlx4_icm_buf {
+	void			*addr;
+	size_t			size;
+	dma_addr_t		dma_addr;
+};
+
+struct mlx4_icm_chunk {
+	struct list_head	list;
+	int			npages;
+	int			nsg;
+	bool			coherent;
+	union {
+		struct scatterlist	sg[MLX4_ICM_CHUNK_LEN];
+		struct mlx4_icm_buf	buf[MLX4_ICM_CHUNK_LEN];
+	};
+};
+
+struct mlx4_icm {
+	struct list_head	chunk_list;
+	int			refcount;
+};
+
+struct mlx4_icm_iter {
+	struct mlx4_icm	       *icm;
+	struct mlx4_icm_chunk  *chunk;
+	int			page_idx;
+};
+
+struct mlx4_dev;
+
+struct mlx4_icm *mlx4_alloc_icm(struct mlx4_dev *dev, int npages,
+				gfp_t gfp_mask, int coherent);
+void mlx4_free_icm(struct mlx4_dev *dev, struct mlx4_icm *icm, int coherent);
+
+int mlx4_table_get(struct mlx4_dev *dev, struct mlx4_icm_table *table, u32 obj);
+void mlx4_table_put(struct mlx4_dev *dev, struct mlx4_icm_table *table, u32 obj);
+int mlx4_table_get_range(struct mlx4_dev *dev, struct mlx4_icm_table *table,
+			 u32 start, u32 end);
+void mlx4_table_put_range(struct mlx4_dev *dev, struct mlx4_icm_table *table,
+			  u32 start, u32 end);
+int mlx4_init_icm_table(struct mlx4_dev *dev, struct mlx4_icm_table *table,
+			u64 virt, int obj_size,	u32 nobj, int reserved,
+			int use_lowmem, int use_coherent);
+void mlx4_cleanup_icm_table(struct mlx4_dev *dev, struct mlx4_icm_table *table);
+void *mlx4_table_find(struct mlx4_icm_table *table, u32 obj, dma_addr_t *dma_handle);
+
+static inline void mlx4_icm_first(struct mlx4_icm *icm,
+				  struct mlx4_icm_iter *iter)
+{
+	iter->icm      = icm;
+	iter->chunk    = list_empty(&icm->chunk_list) ?
+		NULL : list_entry(icm->chunk_list.next,
+				  struct mlx4_icm_chunk, list);
+	iter->page_idx = 0;
+}
+
+static inline int mlx4_icm_last(struct mlx4_icm_iter *iter)
+{
+	return !iter->chunk;
+}
+
+static inline void mlx4_icm_next(struct mlx4_icm_iter *iter)
+{
+	if (++iter->page_idx >= iter->chunk->nsg) {
+		if (iter->chunk->list.next == &iter->icm->chunk_list) {
+			iter->chunk = NULL;
+			return;
+		}
+
+		iter->chunk = list_entry(iter->chunk->list.next,
+					 struct mlx4_icm_chunk, list);
+		iter->page_idx = 0;
+	}
+}
+
+static inline dma_addr_t mlx4_icm_addr(struct mlx4_icm_iter *iter)
+{
+	if (iter->chunk->coherent)
+		return iter->chunk->buf[iter->page_idx].dma_addr;
+	else
+		return sg_dma_address(&iter->chunk->sg[iter->page_idx]);
+}
+
+static inline unsigned long mlx4_icm_size(struct mlx4_icm_iter *iter)
+{
+	if (iter->chunk->coherent)
+		return iter->chunk->buf[iter->page_idx].size;
+	else
+		return sg_dma_len(&iter->chunk->sg[iter->page_idx]);
+}
+
+int mlx4_MAP_ICM_AUX(struct mlx4_dev *dev, struct mlx4_icm *icm);
+int mlx4_UNMAP_ICM_AUX(struct mlx4_dev *dev);
+
+#endif /* MLX4_ICM_H */
diff --git a/drivers/net/ethernet/mellanox/mlx4/intf.c b/drivers/net/ethernet/mellanox/mlx4/intf.c
new file mode 100644
index 000000000..65482f004
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx4/intf.c
@@ -0,0 +1,275 @@
+/*
+ * Copyright (c) 2006, 2007 Cisco Systems, Inc. All rights reserved.
+ * Copyright (c) 2007, 2008 Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/slab.h>
+#include <linux/export.h>
+#include <linux/errno.h>
+#include <net/devlink.h>
+
+#include "mlx4.h"
+
+struct mlx4_device_context {
+	struct list_head	list;
+	struct list_head	bond_list;
+	struct mlx4_interface  *intf;
+	void		       *context;
+};
+
+static LIST_HEAD(intf_list);
+static LIST_HEAD(dev_list);
+static DEFINE_MUTEX(intf_mutex);
+
+static void mlx4_add_device(struct mlx4_interface *intf, struct mlx4_priv *priv)
+{
+	struct mlx4_device_context *dev_ctx;
+
+	dev_ctx = kmalloc(sizeof(*dev_ctx), GFP_KERNEL);
+	if (!dev_ctx)
+		return;
+
+	dev_ctx->intf    = intf;
+	dev_ctx->context = intf->add(&priv->dev);
+
+	if (dev_ctx->context) {
+		spin_lock_irq(&priv->ctx_lock);
+		list_add_tail(&dev_ctx->list, &priv->ctx_list);
+		spin_unlock_irq(&priv->ctx_lock);
+		if (intf->activate)
+			intf->activate(&priv->dev, dev_ctx->context);
+	} else
+		kfree(dev_ctx);
+
+}
+
+static void mlx4_remove_device(struct mlx4_interface *intf, struct mlx4_priv *priv)
+{
+	struct mlx4_device_context *dev_ctx;
+
+	list_for_each_entry(dev_ctx, &priv->ctx_list, list)
+		if (dev_ctx->intf == intf) {
+			spin_lock_irq(&priv->ctx_lock);
+			list_del(&dev_ctx->list);
+			spin_unlock_irq(&priv->ctx_lock);
+
+			intf->remove(&priv->dev, dev_ctx->context);
+			kfree(dev_ctx);
+			return;
+		}
+}
+
+int mlx4_register_interface(struct mlx4_interface *intf)
+{
+	struct mlx4_priv *priv;
+
+	if (!intf->add || !intf->remove)
+		return -EINVAL;
+
+	mutex_lock(&intf_mutex);
+
+	list_add_tail(&intf->list, &intf_list);
+	list_for_each_entry(priv, &dev_list, dev_list) {
+		if (mlx4_is_mfunc(&priv->dev) && (intf->flags & MLX4_INTFF_BONDING)) {
+			mlx4_dbg(&priv->dev,
+				 "SRIOV, disabling HA mode for intf proto %d\n", intf->protocol);
+			intf->flags &= ~MLX4_INTFF_BONDING;
+		}
+		mlx4_add_device(intf, priv);
+	}
+
+	mutex_unlock(&intf_mutex);
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(mlx4_register_interface);
+
+void mlx4_unregister_interface(struct mlx4_interface *intf)
+{
+	struct mlx4_priv *priv;
+
+	mutex_lock(&intf_mutex);
+
+	list_for_each_entry(priv, &dev_list, dev_list)
+		mlx4_remove_device(intf, priv);
+
+	list_del(&intf->list);
+
+	mutex_unlock(&intf_mutex);
+}
+EXPORT_SYMBOL_GPL(mlx4_unregister_interface);
+
+int mlx4_do_bond(struct mlx4_dev *dev, bool enable)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_device_context *dev_ctx = NULL, *temp_dev_ctx;
+	unsigned long flags;
+	int ret;
+	LIST_HEAD(bond_list);
+
+	if (!(dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_PORT_REMAP))
+		return -EOPNOTSUPP;
+
+	ret = mlx4_disable_rx_port_check(dev, enable);
+	if (ret) {
+		mlx4_err(dev, "Fail to %s rx port check\n",
+			 enable ? "enable" : "disable");
+		return ret;
+	}
+	if (enable) {
+		dev->flags |= MLX4_FLAG_BONDED;
+	} else {
+		ret = mlx4_virt2phy_port_map(dev, 1, 2);
+		if (ret) {
+			mlx4_err(dev, "Fail to reset port map\n");
+			return ret;
+		}
+		dev->flags &= ~MLX4_FLAG_BONDED;
+	}
+
+	spin_lock_irqsave(&priv->ctx_lock, flags);
+	list_for_each_entry_safe(dev_ctx, temp_dev_ctx, &priv->ctx_list, list) {
+		if (dev_ctx->intf->flags & MLX4_INTFF_BONDING) {
+			list_add_tail(&dev_ctx->bond_list, &bond_list);
+			list_del(&dev_ctx->list);
+		}
+	}
+	spin_unlock_irqrestore(&priv->ctx_lock, flags);
+
+	list_for_each_entry(dev_ctx, &bond_list, bond_list) {
+		dev_ctx->intf->remove(dev, dev_ctx->context);
+		dev_ctx->context =  dev_ctx->intf->add(dev);
+
+		spin_lock_irqsave(&priv->ctx_lock, flags);
+		list_add_tail(&dev_ctx->list, &priv->ctx_list);
+		spin_unlock_irqrestore(&priv->ctx_lock, flags);
+
+		mlx4_dbg(dev, "Interface for protocol %d restarted with bonded mode %s\n",
+			 dev_ctx->intf->protocol, enable ?
+			 "enabled" : "disabled");
+	}
+	return 0;
+}
+
+void mlx4_dispatch_event(struct mlx4_dev *dev, enum mlx4_dev_event type,
+			 unsigned long param)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_device_context *dev_ctx;
+	unsigned long flags;
+
+	spin_lock_irqsave(&priv->ctx_lock, flags);
+
+	list_for_each_entry(dev_ctx, &priv->ctx_list, list)
+		if (dev_ctx->intf->event)
+			dev_ctx->intf->event(dev, dev_ctx->context, type, param);
+
+	spin_unlock_irqrestore(&priv->ctx_lock, flags);
+}
+
+int mlx4_register_device(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_interface *intf;
+
+	mutex_lock(&intf_mutex);
+
+	dev->persist->interface_state |= MLX4_INTERFACE_STATE_UP;
+	list_add_tail(&priv->dev_list, &dev_list);
+	list_for_each_entry(intf, &intf_list, list)
+		mlx4_add_device(intf, priv);
+
+	mutex_unlock(&intf_mutex);
+	mlx4_start_catas_poll(dev);
+
+	return 0;
+}
+
+void mlx4_unregister_device(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_interface *intf;
+
+	if (!(dev->persist->interface_state & MLX4_INTERFACE_STATE_UP))
+		return;
+
+	mlx4_stop_catas_poll(dev);
+	if (dev->persist->interface_state & MLX4_INTERFACE_STATE_DELETION &&
+	    mlx4_is_slave(dev)) {
+		/* In mlx4_remove_one on a VF */
+		u32 slave_read =
+			swab32(readl(&mlx4_priv(dev)->mfunc.comm->slave_read));
+
+		if (mlx4_comm_internal_err(slave_read)) {
+			mlx4_dbg(dev, "%s: comm channel is down, entering error state.\n",
+				 __func__);
+			mlx4_enter_error_state(dev->persist);
+		}
+	}
+	mutex_lock(&intf_mutex);
+
+	list_for_each_entry(intf, &intf_list, list)
+		mlx4_remove_device(intf, priv);
+
+	list_del(&priv->dev_list);
+	dev->persist->interface_state &= ~MLX4_INTERFACE_STATE_UP;
+
+	mutex_unlock(&intf_mutex);
+}
+
+void *mlx4_get_protocol_dev(struct mlx4_dev *dev, enum mlx4_protocol proto, int port)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_device_context *dev_ctx;
+	unsigned long flags;
+	void *result = NULL;
+
+	spin_lock_irqsave(&priv->ctx_lock, flags);
+
+	list_for_each_entry(dev_ctx, &priv->ctx_list, list)
+		if (dev_ctx->intf->protocol == proto && dev_ctx->intf->get_dev) {
+			result = dev_ctx->intf->get_dev(dev, dev_ctx->context, port);
+			break;
+		}
+
+	spin_unlock_irqrestore(&priv->ctx_lock, flags);
+
+	return result;
+}
+EXPORT_SYMBOL_GPL(mlx4_get_protocol_dev);
+
+struct devlink_port *mlx4_get_devlink_port(struct mlx4_dev *dev, int port)
+{
+	struct mlx4_port_info *info = &mlx4_priv(dev)->port[port];
+
+	return &info->devlink_port;
+}
+EXPORT_SYMBOL_GPL(mlx4_get_devlink_port);
diff --git a/drivers/net/ethernet/mellanox/mlx4/main.c b/drivers/net/ethernet/mellanox/mlx4/main.c
new file mode 100644
index 000000000..d3fc86cd3
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx4/main.c
@@ -0,0 +1,4544 @@
+/*
+ * Copyright (c) 2004, 2005 Topspin Communications.  All rights reserved.
+ * Copyright (c) 2005 Sun Microsystems, Inc. All rights reserved.
+ * Copyright (c) 2005, 2006, 2007, 2008 Mellanox Technologies. All rights reserved.
+ * Copyright (c) 2006, 2007 Cisco Systems, Inc. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/module.h>
+#include <linux/kernel.h>
+#include <linux/init.h>
+#include <linux/errno.h>
+#include <linux/pci.h>
+#include <linux/dma-mapping.h>
+#include <linux/slab.h>
+#include <linux/io-mapping.h>
+#include <linux/delay.h>
+#include <linux/kmod.h>
+#include <linux/etherdevice.h>
+#include <net/devlink.h>
+
+#include <uapi/rdma/mlx4-abi.h>
+#include <linux/mlx4/device.h>
+#include <linux/mlx4/doorbell.h>
+
+#include "mlx4.h"
+#include "fw.h"
+#include "icm.h"
+
+MODULE_AUTHOR("Roland Dreier");
+MODULE_DESCRIPTION("Mellanox ConnectX HCA low-level driver");
+MODULE_LICENSE("Dual BSD/GPL");
+MODULE_VERSION(DRV_VERSION);
+
+struct workqueue_struct *mlx4_wq;
+
+#ifdef CONFIG_MLX4_DEBUG
+
+int mlx4_debug_level; /* 0 by default */
+module_param_named(debug_level, mlx4_debug_level, int, 0644);
+MODULE_PARM_DESC(debug_level, "Enable debug tracing if > 0");
+
+#endif /* CONFIG_MLX4_DEBUG */
+
+#ifdef CONFIG_PCI_MSI
+
+static int msi_x = 1;
+module_param(msi_x, int, 0444);
+MODULE_PARM_DESC(msi_x, "0 - don't use MSI-X, 1 - use MSI-X, >1 - limit number of MSI-X irqs to msi_x");
+
+#else /* CONFIG_PCI_MSI */
+
+#define msi_x (0)
+
+#endif /* CONFIG_PCI_MSI */
+
+static uint8_t num_vfs[3] = {0, 0, 0};
+static int num_vfs_argc;
+module_param_array(num_vfs, byte, &num_vfs_argc, 0444);
+MODULE_PARM_DESC(num_vfs, "enable #num_vfs functions if num_vfs > 0\n"
+			  "num_vfs=port1,port2,port1+2");
+
+static uint8_t probe_vf[3] = {0, 0, 0};
+static int probe_vfs_argc;
+module_param_array(probe_vf, byte, &probe_vfs_argc, 0444);
+MODULE_PARM_DESC(probe_vf, "number of vfs to probe by pf driver (num_vfs > 0)\n"
+			   "probe_vf=port1,port2,port1+2");
+
+static int mlx4_log_num_mgm_entry_size = MLX4_DEFAULT_MGM_LOG_ENTRY_SIZE;
+module_param_named(log_num_mgm_entry_size,
+			mlx4_log_num_mgm_entry_size, int, 0444);
+MODULE_PARM_DESC(log_num_mgm_entry_size, "log mgm size, that defines the num"
+					 " of qp per mcg, for example:"
+					 " 10 gives 248.range: 7 <="
+					 " log_num_mgm_entry_size <= 12."
+					 " To activate device managed"
+					 " flow steering when available, set to -1");
+
+static bool enable_64b_cqe_eqe = true;
+module_param(enable_64b_cqe_eqe, bool, 0444);
+MODULE_PARM_DESC(enable_64b_cqe_eqe,
+		 "Enable 64 byte CQEs/EQEs when the FW supports this (default: True)");
+
+static bool enable_4k_uar;
+module_param(enable_4k_uar, bool, 0444);
+MODULE_PARM_DESC(enable_4k_uar,
+		 "Enable using 4K UAR. Should not be enabled if have VFs which do not support 4K UARs (default: false)");
+
+#define PF_CONTEXT_BEHAVIOUR_MASK	(MLX4_FUNC_CAP_64B_EQE_CQE | \
+					 MLX4_FUNC_CAP_EQE_CQE_STRIDE | \
+					 MLX4_FUNC_CAP_DMFS_A0_STATIC)
+
+#define RESET_PERSIST_MASK_FLAGS	(MLX4_FLAG_SRIOV)
+
+static char mlx4_version[] =
+	DRV_NAME ": Mellanox ConnectX core driver v"
+	DRV_VERSION "\n";
+
+static const struct mlx4_profile default_profile = {
+	.num_qp		= 1 << 18,
+	.num_srq	= 1 << 16,
+	.rdmarc_per_qp	= 1 << 4,
+	.num_cq		= 1 << 16,
+	.num_mcg	= 1 << 13,
+	.num_mpt	= 1 << 19,
+	.num_mtt	= 1 << 20, /* It is really num mtt segements */
+};
+
+static const struct mlx4_profile low_mem_profile = {
+	.num_qp		= 1 << 17,
+	.num_srq	= 1 << 6,
+	.rdmarc_per_qp	= 1 << 4,
+	.num_cq		= 1 << 8,
+	.num_mcg	= 1 << 8,
+	.num_mpt	= 1 << 9,
+	.num_mtt	= 1 << 7,
+};
+
+static int log_num_mac = 7;
+module_param_named(log_num_mac, log_num_mac, int, 0444);
+MODULE_PARM_DESC(log_num_mac, "Log2 max number of MACs per ETH port (1-7)");
+
+static int log_num_vlan;
+module_param_named(log_num_vlan, log_num_vlan, int, 0444);
+MODULE_PARM_DESC(log_num_vlan, "Log2 max number of VLANs per ETH port (0-7)");
+/* Log2 max number of VLANs per ETH port (0-7) */
+#define MLX4_LOG_NUM_VLANS 7
+#define MLX4_MIN_LOG_NUM_VLANS 0
+#define MLX4_MIN_LOG_NUM_MAC 1
+
+static bool use_prio;
+module_param_named(use_prio, use_prio, bool, 0444);
+MODULE_PARM_DESC(use_prio, "Enable steering by VLAN priority on ETH ports (deprecated)");
+
+int log_mtts_per_seg = ilog2(1);
+module_param_named(log_mtts_per_seg, log_mtts_per_seg, int, 0444);
+MODULE_PARM_DESC(log_mtts_per_seg, "Log2 number of MTT entries per segment "
+		 "(0-7) (default: 0)");
+
+static int port_type_array[2] = {MLX4_PORT_TYPE_NONE, MLX4_PORT_TYPE_NONE};
+static int arr_argc = 2;
+module_param_array(port_type_array, int, &arr_argc, 0444);
+MODULE_PARM_DESC(port_type_array, "Array of port types: HW_DEFAULT (0) is default "
+				"1 for IB, 2 for Ethernet");
+
+struct mlx4_port_config {
+	struct list_head list;
+	enum mlx4_port_type port_type[MLX4_MAX_PORTS + 1];
+	struct pci_dev *pdev;
+};
+
+static atomic_t pf_loading = ATOMIC_INIT(0);
+
+static int mlx4_devlink_ierr_reset_get(struct devlink *devlink, u32 id,
+				       struct devlink_param_gset_ctx *ctx)
+{
+	ctx->val.vbool = !!mlx4_internal_err_reset;
+	return 0;
+}
+
+static int mlx4_devlink_ierr_reset_set(struct devlink *devlink, u32 id,
+				       struct devlink_param_gset_ctx *ctx)
+{
+	mlx4_internal_err_reset = ctx->val.vbool;
+	return 0;
+}
+
+static int mlx4_devlink_crdump_snapshot_get(struct devlink *devlink, u32 id,
+					    struct devlink_param_gset_ctx *ctx)
+{
+	struct mlx4_priv *priv = devlink_priv(devlink);
+	struct mlx4_dev *dev = &priv->dev;
+
+	ctx->val.vbool = dev->persist->crdump.snapshot_enable;
+	return 0;
+}
+
+static int mlx4_devlink_crdump_snapshot_set(struct devlink *devlink, u32 id,
+					    struct devlink_param_gset_ctx *ctx)
+{
+	struct mlx4_priv *priv = devlink_priv(devlink);
+	struct mlx4_dev *dev = &priv->dev;
+
+	dev->persist->crdump.snapshot_enable = ctx->val.vbool;
+	return 0;
+}
+
+static int
+mlx4_devlink_max_macs_validate(struct devlink *devlink, u32 id,
+			       union devlink_param_value val,
+			       struct netlink_ext_ack *extack)
+{
+	u32 value = val.vu32;
+
+	if (value < 1 || value > 128)
+		return -ERANGE;
+
+	if (!is_power_of_2(value)) {
+		NL_SET_ERR_MSG_MOD(extack, "max_macs supported must be power of 2");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+enum mlx4_devlink_param_id {
+	MLX4_DEVLINK_PARAM_ID_BASE = DEVLINK_PARAM_GENERIC_ID_MAX,
+	MLX4_DEVLINK_PARAM_ID_ENABLE_64B_CQE_EQE,
+	MLX4_DEVLINK_PARAM_ID_ENABLE_4K_UAR,
+};
+
+static const struct devlink_param mlx4_devlink_params[] = {
+	DEVLINK_PARAM_GENERIC(INT_ERR_RESET,
+			      BIT(DEVLINK_PARAM_CMODE_RUNTIME) |
+			      BIT(DEVLINK_PARAM_CMODE_DRIVERINIT),
+			      mlx4_devlink_ierr_reset_get,
+			      mlx4_devlink_ierr_reset_set, NULL),
+	DEVLINK_PARAM_GENERIC(MAX_MACS,
+			      BIT(DEVLINK_PARAM_CMODE_DRIVERINIT),
+			      NULL, NULL, mlx4_devlink_max_macs_validate),
+	DEVLINK_PARAM_GENERIC(REGION_SNAPSHOT,
+			      BIT(DEVLINK_PARAM_CMODE_RUNTIME) |
+			      BIT(DEVLINK_PARAM_CMODE_DRIVERINIT),
+			      mlx4_devlink_crdump_snapshot_get,
+			      mlx4_devlink_crdump_snapshot_set, NULL),
+	DEVLINK_PARAM_DRIVER(MLX4_DEVLINK_PARAM_ID_ENABLE_64B_CQE_EQE,
+			     "enable_64b_cqe_eqe", DEVLINK_PARAM_TYPE_BOOL,
+			     BIT(DEVLINK_PARAM_CMODE_DRIVERINIT),
+			     NULL, NULL, NULL),
+	DEVLINK_PARAM_DRIVER(MLX4_DEVLINK_PARAM_ID_ENABLE_4K_UAR,
+			     "enable_4k_uar", DEVLINK_PARAM_TYPE_BOOL,
+			     BIT(DEVLINK_PARAM_CMODE_DRIVERINIT),
+			     NULL, NULL, NULL),
+};
+
+static void mlx4_devlink_set_params_init_values(struct devlink *devlink)
+{
+	union devlink_param_value value;
+
+	value.vbool = !!mlx4_internal_err_reset;
+	devlink_param_driverinit_value_set(devlink,
+					   DEVLINK_PARAM_GENERIC_ID_INT_ERR_RESET,
+					   value);
+
+	value.vu32 = 1UL << log_num_mac;
+	devlink_param_driverinit_value_set(devlink,
+					   DEVLINK_PARAM_GENERIC_ID_MAX_MACS,
+					   value);
+
+	value.vbool = enable_64b_cqe_eqe;
+	devlink_param_driverinit_value_set(devlink,
+					   MLX4_DEVLINK_PARAM_ID_ENABLE_64B_CQE_EQE,
+					   value);
+
+	value.vbool = enable_4k_uar;
+	devlink_param_driverinit_value_set(devlink,
+					   MLX4_DEVLINK_PARAM_ID_ENABLE_4K_UAR,
+					   value);
+
+	value.vbool = false;
+	devlink_param_driverinit_value_set(devlink,
+					   DEVLINK_PARAM_GENERIC_ID_REGION_SNAPSHOT,
+					   value);
+}
+
+static inline void mlx4_set_num_reserved_uars(struct mlx4_dev *dev,
+					      struct mlx4_dev_cap *dev_cap)
+{
+	/* The reserved_uars is calculated by system page size unit.
+	 * Therefore, adjustment is added when the uar page size is less
+	 * than the system page size
+	 */
+	dev->caps.reserved_uars	=
+		max_t(int,
+		      mlx4_get_num_reserved_uar(dev),
+		      dev_cap->reserved_uars /
+			(1 << (PAGE_SHIFT - dev->uar_page_shift)));
+}
+
+int mlx4_check_port_params(struct mlx4_dev *dev,
+			   enum mlx4_port_type *port_type)
+{
+	int i;
+
+	if (!(dev->caps.flags & MLX4_DEV_CAP_FLAG_DPDP)) {
+		for (i = 0; i < dev->caps.num_ports - 1; i++) {
+			if (port_type[i] != port_type[i + 1]) {
+				mlx4_err(dev, "Only same port types supported on this HCA, aborting\n");
+				return -EOPNOTSUPP;
+			}
+		}
+	}
+
+	for (i = 0; i < dev->caps.num_ports; i++) {
+		if (!(port_type[i] & dev->caps.supported_type[i+1])) {
+			mlx4_err(dev, "Requested port type for port %d is not supported on this HCA\n",
+				 i + 1);
+			return -EOPNOTSUPP;
+		}
+	}
+	return 0;
+}
+
+static void mlx4_set_port_mask(struct mlx4_dev *dev)
+{
+	int i;
+
+	for (i = 1; i <= dev->caps.num_ports; ++i)
+		dev->caps.port_mask[i] = dev->caps.port_type[i];
+}
+
+enum {
+	MLX4_QUERY_FUNC_NUM_SYS_EQS = 1 << 0,
+};
+
+static int mlx4_query_func(struct mlx4_dev *dev, struct mlx4_dev_cap *dev_cap)
+{
+	int err = 0;
+	struct mlx4_func func;
+
+	if (dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_SYS_EQS) {
+		err = mlx4_QUERY_FUNC(dev, &func, 0);
+		if (err) {
+			mlx4_err(dev, "QUERY_DEV_CAP command failed, aborting.\n");
+			return err;
+		}
+		dev_cap->max_eqs = func.max_eq;
+		dev_cap->reserved_eqs = func.rsvd_eqs;
+		dev_cap->reserved_uars = func.rsvd_uars;
+		err |= MLX4_QUERY_FUNC_NUM_SYS_EQS;
+	}
+	return err;
+}
+
+static void mlx4_enable_cqe_eqe_stride(struct mlx4_dev *dev)
+{
+	struct mlx4_caps *dev_cap = &dev->caps;
+
+	/* FW not supporting or cancelled by user */
+	if (!(dev_cap->flags2 & MLX4_DEV_CAP_FLAG2_EQE_STRIDE) ||
+	    !(dev_cap->flags2 & MLX4_DEV_CAP_FLAG2_CQE_STRIDE))
+		return;
+
+	/* Must have 64B CQE_EQE enabled by FW to use bigger stride
+	 * When FW has NCSI it may decide not to report 64B CQE/EQEs
+	 */
+	if (!(dev_cap->flags & MLX4_DEV_CAP_FLAG_64B_EQE) ||
+	    !(dev_cap->flags & MLX4_DEV_CAP_FLAG_64B_CQE)) {
+		dev_cap->flags2 &= ~MLX4_DEV_CAP_FLAG2_CQE_STRIDE;
+		dev_cap->flags2 &= ~MLX4_DEV_CAP_FLAG2_EQE_STRIDE;
+		return;
+	}
+
+	if (cache_line_size() == 128 || cache_line_size() == 256) {
+		mlx4_dbg(dev, "Enabling CQE stride cacheLine supported\n");
+		/* Changing the real data inside CQE size to 32B */
+		dev_cap->flags &= ~MLX4_DEV_CAP_FLAG_64B_CQE;
+		dev_cap->flags &= ~MLX4_DEV_CAP_FLAG_64B_EQE;
+
+		if (mlx4_is_master(dev))
+			dev_cap->function_caps |= MLX4_FUNC_CAP_EQE_CQE_STRIDE;
+	} else {
+		if (cache_line_size() != 32  && cache_line_size() != 64)
+			mlx4_dbg(dev, "Disabling CQE stride, cacheLine size unsupported\n");
+		dev_cap->flags2 &= ~MLX4_DEV_CAP_FLAG2_CQE_STRIDE;
+		dev_cap->flags2 &= ~MLX4_DEV_CAP_FLAG2_EQE_STRIDE;
+	}
+}
+
+static int _mlx4_dev_port(struct mlx4_dev *dev, int port,
+			  struct mlx4_port_cap *port_cap)
+{
+	dev->caps.vl_cap[port]	    = port_cap->max_vl;
+	dev->caps.ib_mtu_cap[port]	    = port_cap->ib_mtu;
+	dev->phys_caps.gid_phys_table_len[port]  = port_cap->max_gids;
+	dev->phys_caps.pkey_phys_table_len[port] = port_cap->max_pkeys;
+	/* set gid and pkey table operating lengths by default
+	 * to non-sriov values
+	 */
+	dev->caps.gid_table_len[port]  = port_cap->max_gids;
+	dev->caps.pkey_table_len[port] = port_cap->max_pkeys;
+	dev->caps.port_width_cap[port] = port_cap->max_port_width;
+	dev->caps.eth_mtu_cap[port]    = port_cap->eth_mtu;
+	dev->caps.max_tc_eth	       = port_cap->max_tc_eth;
+	dev->caps.def_mac[port]        = port_cap->def_mac;
+	dev->caps.supported_type[port] = port_cap->supported_port_types;
+	dev->caps.suggested_type[port] = port_cap->suggested_type;
+	dev->caps.default_sense[port] = port_cap->default_sense;
+	dev->caps.trans_type[port]	    = port_cap->trans_type;
+	dev->caps.vendor_oui[port]     = port_cap->vendor_oui;
+	dev->caps.wavelength[port]     = port_cap->wavelength;
+	dev->caps.trans_code[port]     = port_cap->trans_code;
+
+	return 0;
+}
+
+static int mlx4_dev_port(struct mlx4_dev *dev, int port,
+			 struct mlx4_port_cap *port_cap)
+{
+	int err = 0;
+
+	err = mlx4_QUERY_PORT(dev, port, port_cap);
+
+	if (err)
+		mlx4_err(dev, "QUERY_PORT command failed.\n");
+
+	return err;
+}
+
+static inline void mlx4_enable_ignore_fcs(struct mlx4_dev *dev)
+{
+	if (!(dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_IGNORE_FCS))
+		return;
+
+	if (mlx4_is_mfunc(dev)) {
+		mlx4_dbg(dev, "SRIOV mode - Disabling Ignore FCS");
+		dev->caps.flags2 &= ~MLX4_DEV_CAP_FLAG2_IGNORE_FCS;
+		return;
+	}
+
+	if (!(dev->caps.flags & MLX4_DEV_CAP_FLAG_FCS_KEEP)) {
+		mlx4_dbg(dev,
+			 "Keep FCS is not supported - Disabling Ignore FCS");
+		dev->caps.flags2 &= ~MLX4_DEV_CAP_FLAG2_IGNORE_FCS;
+		return;
+	}
+}
+
+#define MLX4_A0_STEERING_TABLE_SIZE	256
+static int mlx4_dev_cap(struct mlx4_dev *dev, struct mlx4_dev_cap *dev_cap)
+{
+	int err;
+	int i;
+
+	err = mlx4_QUERY_DEV_CAP(dev, dev_cap);
+	if (err) {
+		mlx4_err(dev, "QUERY_DEV_CAP command failed, aborting\n");
+		return err;
+	}
+	mlx4_dev_cap_dump(dev, dev_cap);
+
+	if (dev_cap->min_page_sz > PAGE_SIZE) {
+		mlx4_err(dev, "HCA minimum page size of %d bigger than kernel PAGE_SIZE of %ld, aborting\n",
+			 dev_cap->min_page_sz, PAGE_SIZE);
+		return -ENODEV;
+	}
+	if (dev_cap->num_ports > MLX4_MAX_PORTS) {
+		mlx4_err(dev, "HCA has %d ports, but we only support %d, aborting\n",
+			 dev_cap->num_ports, MLX4_MAX_PORTS);
+		return -ENODEV;
+	}
+
+	if (dev_cap->uar_size > pci_resource_len(dev->persist->pdev, 2)) {
+		mlx4_err(dev, "HCA reported UAR size of 0x%x bigger than PCI resource 2 size of 0x%llx, aborting\n",
+			 dev_cap->uar_size,
+			 (unsigned long long)
+			 pci_resource_len(dev->persist->pdev, 2));
+		return -ENODEV;
+	}
+
+	dev->caps.num_ports	     = dev_cap->num_ports;
+	dev->caps.num_sys_eqs = dev_cap->num_sys_eqs;
+	dev->phys_caps.num_phys_eqs = dev_cap->flags2 & MLX4_DEV_CAP_FLAG2_SYS_EQS ?
+				      dev->caps.num_sys_eqs :
+				      MLX4_MAX_EQ_NUM;
+	for (i = 1; i <= dev->caps.num_ports; ++i) {
+		err = _mlx4_dev_port(dev, i, dev_cap->port_cap + i);
+		if (err) {
+			mlx4_err(dev, "QUERY_PORT command failed, aborting\n");
+			return err;
+		}
+	}
+
+	dev->caps.map_clock_to_user  = dev_cap->map_clock_to_user;
+	dev->caps.uar_page_size	     = PAGE_SIZE;
+	dev->caps.num_uars	     = dev_cap->uar_size / PAGE_SIZE;
+	dev->caps.local_ca_ack_delay = dev_cap->local_ca_ack_delay;
+	dev->caps.bf_reg_size	     = dev_cap->bf_reg_size;
+	dev->caps.bf_regs_per_page   = dev_cap->bf_regs_per_page;
+	dev->caps.max_sq_sg	     = dev_cap->max_sq_sg;
+	dev->caps.max_rq_sg	     = dev_cap->max_rq_sg;
+	dev->caps.max_wqes	     = dev_cap->max_qp_sz;
+	dev->caps.max_qp_init_rdma   = dev_cap->max_requester_per_qp;
+	dev->caps.max_srq_wqes	     = dev_cap->max_srq_sz;
+	dev->caps.max_srq_sge	     = dev_cap->max_rq_sg - 1;
+	dev->caps.reserved_srqs	     = dev_cap->reserved_srqs;
+	dev->caps.max_sq_desc_sz     = dev_cap->max_sq_desc_sz;
+	dev->caps.max_rq_desc_sz     = dev_cap->max_rq_desc_sz;
+	/*
+	 * Subtract 1 from the limit because we need to allocate a
+	 * spare CQE to enable resizing the CQ.
+	 */
+	dev->caps.max_cqes	     = dev_cap->max_cq_sz - 1;
+	dev->caps.reserved_cqs	     = dev_cap->reserved_cqs;
+	dev->caps.reserved_eqs	     = dev_cap->reserved_eqs;
+	dev->caps.reserved_mtts      = dev_cap->reserved_mtts;
+	dev->caps.reserved_mrws	     = dev_cap->reserved_mrws;
+
+	dev->caps.reserved_pds	     = dev_cap->reserved_pds;
+	dev->caps.reserved_xrcds     = (dev->caps.flags & MLX4_DEV_CAP_FLAG_XRC) ?
+					dev_cap->reserved_xrcds : 0;
+	dev->caps.max_xrcds          = (dev->caps.flags & MLX4_DEV_CAP_FLAG_XRC) ?
+					dev_cap->max_xrcds : 0;
+	dev->caps.mtt_entry_sz       = dev_cap->mtt_entry_sz;
+
+	dev->caps.max_msg_sz         = dev_cap->max_msg_sz;
+	dev->caps.page_size_cap	     = ~(u32) (dev_cap->min_page_sz - 1);
+	dev->caps.flags		     = dev_cap->flags;
+	dev->caps.flags2	     = dev_cap->flags2;
+	dev->caps.bmme_flags	     = dev_cap->bmme_flags;
+	dev->caps.reserved_lkey	     = dev_cap->reserved_lkey;
+	dev->caps.stat_rate_support  = dev_cap->stat_rate_support;
+	dev->caps.max_gso_sz	     = dev_cap->max_gso_sz;
+	dev->caps.max_rss_tbl_sz     = dev_cap->max_rss_tbl_sz;
+	dev->caps.wol_port[1]          = dev_cap->wol_port[1];
+	dev->caps.wol_port[2]          = dev_cap->wol_port[2];
+	dev->caps.health_buffer_addrs  = dev_cap->health_buffer_addrs;
+
+	/* Save uar page shift */
+	if (!mlx4_is_slave(dev)) {
+		/* Virtual PCI function needs to determine UAR page size from
+		 * firmware. Only master PCI function can set the uar page size
+		 */
+		if (enable_4k_uar || !dev->persist->num_vfs)
+			dev->uar_page_shift = DEFAULT_UAR_PAGE_SHIFT;
+		else
+			dev->uar_page_shift = PAGE_SHIFT;
+
+		mlx4_set_num_reserved_uars(dev, dev_cap);
+	}
+
+	if (dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_PHV_EN) {
+		struct mlx4_init_hca_param hca_param;
+
+		memset(&hca_param, 0, sizeof(hca_param));
+		err = mlx4_QUERY_HCA(dev, &hca_param);
+		/* Turn off PHV_EN flag in case phv_check_en is set.
+		 * phv_check_en is a HW check that parse the packet and verify
+		 * phv bit was reported correctly in the wqe. To allow QinQ
+		 * PHV_EN flag should be set and phv_check_en must be cleared
+		 * otherwise QinQ packets will be drop by the HW.
+		 */
+		if (err || hca_param.phv_check_en)
+			dev->caps.flags2 &= ~MLX4_DEV_CAP_FLAG2_PHV_EN;
+	}
+
+	/* Sense port always allowed on supported devices for ConnectX-1 and -2 */
+	if (mlx4_priv(dev)->pci_dev_data & MLX4_PCI_DEV_FORCE_SENSE_PORT)
+		dev->caps.flags |= MLX4_DEV_CAP_FLAG_SENSE_SUPPORT;
+	/* Don't do sense port on multifunction devices (for now at least) */
+	if (mlx4_is_mfunc(dev))
+		dev->caps.flags &= ~MLX4_DEV_CAP_FLAG_SENSE_SUPPORT;
+
+	if (mlx4_low_memory_profile()) {
+		dev->caps.log_num_macs  = MLX4_MIN_LOG_NUM_MAC;
+		dev->caps.log_num_vlans = MLX4_MIN_LOG_NUM_VLANS;
+	} else {
+		dev->caps.log_num_macs  = log_num_mac;
+		dev->caps.log_num_vlans = MLX4_LOG_NUM_VLANS;
+	}
+
+	for (i = 1; i <= dev->caps.num_ports; ++i) {
+		dev->caps.port_type[i] = MLX4_PORT_TYPE_NONE;
+		if (dev->caps.supported_type[i]) {
+			/* if only ETH is supported - assign ETH */
+			if (dev->caps.supported_type[i] == MLX4_PORT_TYPE_ETH)
+				dev->caps.port_type[i] = MLX4_PORT_TYPE_ETH;
+			/* if only IB is supported, assign IB */
+			else if (dev->caps.supported_type[i] ==
+				 MLX4_PORT_TYPE_IB)
+				dev->caps.port_type[i] = MLX4_PORT_TYPE_IB;
+			else {
+				/* if IB and ETH are supported, we set the port
+				 * type according to user selection of port type;
+				 * if user selected none, take the FW hint */
+				if (port_type_array[i - 1] == MLX4_PORT_TYPE_NONE)
+					dev->caps.port_type[i] = dev->caps.suggested_type[i] ?
+						MLX4_PORT_TYPE_ETH : MLX4_PORT_TYPE_IB;
+				else
+					dev->caps.port_type[i] = port_type_array[i - 1];
+			}
+		}
+		/*
+		 * Link sensing is allowed on the port if 3 conditions are true:
+		 * 1. Both protocols are supported on the port.
+		 * 2. Different types are supported on the port
+		 * 3. FW declared that it supports link sensing
+		 */
+		mlx4_priv(dev)->sense.sense_allowed[i] =
+			((dev->caps.supported_type[i] == MLX4_PORT_TYPE_AUTO) &&
+			 (dev->caps.flags & MLX4_DEV_CAP_FLAG_DPDP) &&
+			 (dev->caps.flags & MLX4_DEV_CAP_FLAG_SENSE_SUPPORT));
+
+		/*
+		 * If "default_sense" bit is set, we move the port to "AUTO" mode
+		 * and perform sense_port FW command to try and set the correct
+		 * port type from beginning
+		 */
+		if (mlx4_priv(dev)->sense.sense_allowed[i] && dev->caps.default_sense[i]) {
+			enum mlx4_port_type sensed_port = MLX4_PORT_TYPE_NONE;
+			dev->caps.possible_type[i] = MLX4_PORT_TYPE_AUTO;
+			mlx4_SENSE_PORT(dev, i, &sensed_port);
+			if (sensed_port != MLX4_PORT_TYPE_NONE)
+				dev->caps.port_type[i] = sensed_port;
+		} else {
+			dev->caps.possible_type[i] = dev->caps.port_type[i];
+		}
+
+		if (dev->caps.log_num_macs > dev_cap->port_cap[i].log_max_macs) {
+			dev->caps.log_num_macs = dev_cap->port_cap[i].log_max_macs;
+			mlx4_warn(dev, "Requested number of MACs is too much for port %d, reducing to %d\n",
+				  i, 1 << dev->caps.log_num_macs);
+		}
+		if (dev->caps.log_num_vlans > dev_cap->port_cap[i].log_max_vlans) {
+			dev->caps.log_num_vlans = dev_cap->port_cap[i].log_max_vlans;
+			mlx4_warn(dev, "Requested number of VLANs is too much for port %d, reducing to %d\n",
+				  i, 1 << dev->caps.log_num_vlans);
+		}
+	}
+
+	if (mlx4_is_master(dev) && (dev->caps.num_ports == 2) &&
+	    (port_type_array[0] == MLX4_PORT_TYPE_IB) &&
+	    (port_type_array[1] == MLX4_PORT_TYPE_ETH)) {
+		mlx4_warn(dev,
+			  "Granular QoS per VF not supported with IB/Eth configuration\n");
+		dev->caps.flags2 &= ~MLX4_DEV_CAP_FLAG2_QOS_VPP;
+	}
+
+	dev->caps.max_counters = dev_cap->max_counters;
+
+	dev->caps.reserved_qps_cnt[MLX4_QP_REGION_FW] = dev_cap->reserved_qps;
+	dev->caps.reserved_qps_cnt[MLX4_QP_REGION_ETH_ADDR] =
+		dev->caps.reserved_qps_cnt[MLX4_QP_REGION_FC_ADDR] =
+		(1 << dev->caps.log_num_macs) *
+		(1 << dev->caps.log_num_vlans) *
+		dev->caps.num_ports;
+	dev->caps.reserved_qps_cnt[MLX4_QP_REGION_FC_EXCH] = MLX4_NUM_FEXCH;
+
+	if (dev_cap->dmfs_high_rate_qpn_base > 0 &&
+	    dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_FS_EN)
+		dev->caps.dmfs_high_rate_qpn_base = dev_cap->dmfs_high_rate_qpn_base;
+	else
+		dev->caps.dmfs_high_rate_qpn_base =
+			dev->caps.reserved_qps_cnt[MLX4_QP_REGION_FW];
+
+	if (dev_cap->dmfs_high_rate_qpn_range > 0 &&
+	    dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_FS_EN) {
+		dev->caps.dmfs_high_rate_qpn_range = dev_cap->dmfs_high_rate_qpn_range;
+		dev->caps.dmfs_high_steer_mode = MLX4_STEERING_DMFS_A0_DEFAULT;
+		dev->caps.flags2 |= MLX4_DEV_CAP_FLAG2_FS_A0;
+	} else {
+		dev->caps.dmfs_high_steer_mode = MLX4_STEERING_DMFS_A0_NOT_SUPPORTED;
+		dev->caps.dmfs_high_rate_qpn_base =
+			dev->caps.reserved_qps_cnt[MLX4_QP_REGION_FW];
+		dev->caps.dmfs_high_rate_qpn_range = MLX4_A0_STEERING_TABLE_SIZE;
+	}
+
+	dev->caps.rl_caps = dev_cap->rl_caps;
+
+	dev->caps.reserved_qps_cnt[MLX4_QP_REGION_RSS_RAW_ETH] =
+		dev->caps.dmfs_high_rate_qpn_range;
+
+	dev->caps.reserved_qps = dev->caps.reserved_qps_cnt[MLX4_QP_REGION_FW] +
+		dev->caps.reserved_qps_cnt[MLX4_QP_REGION_ETH_ADDR] +
+		dev->caps.reserved_qps_cnt[MLX4_QP_REGION_FC_ADDR] +
+		dev->caps.reserved_qps_cnt[MLX4_QP_REGION_FC_EXCH];
+
+	dev->caps.sqp_demux = (mlx4_is_master(dev)) ? MLX4_MAX_NUM_SLAVES : 0;
+
+	if (!enable_64b_cqe_eqe && !mlx4_is_slave(dev)) {
+		if (dev_cap->flags &
+		    (MLX4_DEV_CAP_FLAG_64B_CQE | MLX4_DEV_CAP_FLAG_64B_EQE)) {
+			mlx4_warn(dev, "64B EQEs/CQEs supported by the device but not enabled\n");
+			dev->caps.flags &= ~MLX4_DEV_CAP_FLAG_64B_CQE;
+			dev->caps.flags &= ~MLX4_DEV_CAP_FLAG_64B_EQE;
+		}
+
+		if (dev_cap->flags2 &
+		    (MLX4_DEV_CAP_FLAG2_CQE_STRIDE |
+		     MLX4_DEV_CAP_FLAG2_EQE_STRIDE)) {
+			mlx4_warn(dev, "Disabling EQE/CQE stride per user request\n");
+			dev_cap->flags2 &= ~MLX4_DEV_CAP_FLAG2_CQE_STRIDE;
+			dev_cap->flags2 &= ~MLX4_DEV_CAP_FLAG2_EQE_STRIDE;
+		}
+	}
+
+	if ((dev->caps.flags &
+	    (MLX4_DEV_CAP_FLAG_64B_CQE | MLX4_DEV_CAP_FLAG_64B_EQE)) &&
+	    mlx4_is_master(dev))
+		dev->caps.function_caps |= MLX4_FUNC_CAP_64B_EQE_CQE;
+
+	if (!mlx4_is_slave(dev)) {
+		mlx4_enable_cqe_eqe_stride(dev);
+		dev->caps.alloc_res_qp_mask =
+			(dev->caps.bf_reg_size ? MLX4_RESERVE_ETH_BF_QP : 0) |
+			MLX4_RESERVE_A0_QP;
+
+		if (!(dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_ETS_CFG) &&
+		    dev->caps.flags & MLX4_DEV_CAP_FLAG_SET_ETH_SCHED) {
+			mlx4_warn(dev, "Old device ETS support detected\n");
+			mlx4_warn(dev, "Consider upgrading device FW.\n");
+			dev->caps.flags2 |= MLX4_DEV_CAP_FLAG2_ETS_CFG;
+		}
+
+	} else {
+		dev->caps.alloc_res_qp_mask = 0;
+	}
+
+	mlx4_enable_ignore_fcs(dev);
+
+	return 0;
+}
+
+/*The function checks if there are live vf, return the num of them*/
+static int mlx4_how_many_lives_vf(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_slave_state *s_state;
+	int i;
+	int ret = 0;
+
+	for (i = 1/*the ppf is 0*/; i < dev->num_slaves; ++i) {
+		s_state = &priv->mfunc.master.slave_state[i];
+		if (s_state->active && s_state->last_cmd !=
+		    MLX4_COMM_CMD_RESET) {
+			mlx4_warn(dev, "%s: slave: %d is still active\n",
+				  __func__, i);
+			ret++;
+		}
+	}
+	return ret;
+}
+
+int mlx4_get_parav_qkey(struct mlx4_dev *dev, u32 qpn, u32 *qkey)
+{
+	u32 qk = MLX4_RESERVED_QKEY_BASE;
+
+	if (qpn >= dev->phys_caps.base_tunnel_sqpn + 8 * MLX4_MFUNC_MAX ||
+	    qpn < dev->phys_caps.base_proxy_sqpn)
+		return -EINVAL;
+
+	if (qpn >= dev->phys_caps.base_tunnel_sqpn)
+		/* tunnel qp */
+		qk += qpn - dev->phys_caps.base_tunnel_sqpn;
+	else
+		qk += qpn - dev->phys_caps.base_proxy_sqpn;
+	*qkey = qk;
+	return 0;
+}
+EXPORT_SYMBOL(mlx4_get_parav_qkey);
+
+void mlx4_sync_pkey_table(struct mlx4_dev *dev, int slave, int port, int i, int val)
+{
+	struct mlx4_priv *priv = container_of(dev, struct mlx4_priv, dev);
+
+	if (!mlx4_is_master(dev))
+		return;
+
+	priv->virt2phys_pkey[slave][port - 1][i] = val;
+}
+EXPORT_SYMBOL(mlx4_sync_pkey_table);
+
+void mlx4_put_slave_node_guid(struct mlx4_dev *dev, int slave, __be64 guid)
+{
+	struct mlx4_priv *priv = container_of(dev, struct mlx4_priv, dev);
+
+	if (!mlx4_is_master(dev))
+		return;
+
+	priv->slave_node_guids[slave] = guid;
+}
+EXPORT_SYMBOL(mlx4_put_slave_node_guid);
+
+__be64 mlx4_get_slave_node_guid(struct mlx4_dev *dev, int slave)
+{
+	struct mlx4_priv *priv = container_of(dev, struct mlx4_priv, dev);
+
+	if (!mlx4_is_master(dev))
+		return 0;
+
+	return priv->slave_node_guids[slave];
+}
+EXPORT_SYMBOL(mlx4_get_slave_node_guid);
+
+int mlx4_is_slave_active(struct mlx4_dev *dev, int slave)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_slave_state *s_slave;
+
+	if (!mlx4_is_master(dev))
+		return 0;
+
+	s_slave = &priv->mfunc.master.slave_state[slave];
+	return !!s_slave->active;
+}
+EXPORT_SYMBOL(mlx4_is_slave_active);
+
+void mlx4_handle_eth_header_mcast_prio(struct mlx4_net_trans_rule_hw_ctrl *ctrl,
+				       struct _rule_hw *eth_header)
+{
+	if (is_multicast_ether_addr(eth_header->eth.dst_mac) ||
+	    is_broadcast_ether_addr(eth_header->eth.dst_mac)) {
+		struct mlx4_net_trans_rule_hw_eth *eth =
+			(struct mlx4_net_trans_rule_hw_eth *)eth_header;
+		struct _rule_hw *next_rule = (struct _rule_hw *)(eth + 1);
+		bool last_rule = next_rule->size == 0 && next_rule->id == 0 &&
+			next_rule->rsvd == 0;
+
+		if (last_rule)
+			ctrl->prio = cpu_to_be16(MLX4_DOMAIN_NIC);
+	}
+}
+EXPORT_SYMBOL(mlx4_handle_eth_header_mcast_prio);
+
+static void slave_adjust_steering_mode(struct mlx4_dev *dev,
+				       struct mlx4_dev_cap *dev_cap,
+				       struct mlx4_init_hca_param *hca_param)
+{
+	dev->caps.steering_mode = hca_param->steering_mode;
+	if (dev->caps.steering_mode == MLX4_STEERING_MODE_DEVICE_MANAGED) {
+		dev->caps.num_qp_per_mgm = dev_cap->fs_max_num_qp_per_entry;
+		dev->caps.fs_log_max_ucast_qp_range_size =
+			dev_cap->fs_log_max_ucast_qp_range_size;
+	} else
+		dev->caps.num_qp_per_mgm =
+			4 * ((1 << hca_param->log_mc_entry_sz)/16 - 2);
+
+	mlx4_dbg(dev, "Steering mode is: %s\n",
+		 mlx4_steering_mode_str(dev->caps.steering_mode));
+}
+
+static void mlx4_slave_destroy_special_qp_cap(struct mlx4_dev *dev)
+{
+	kfree(dev->caps.spec_qps);
+	dev->caps.spec_qps = NULL;
+}
+
+static int mlx4_slave_special_qp_cap(struct mlx4_dev *dev)
+{
+	struct mlx4_func_cap *func_cap = NULL;
+	struct mlx4_caps *caps = &dev->caps;
+	int i, err = 0;
+
+	func_cap = kzalloc(sizeof(*func_cap), GFP_KERNEL);
+	caps->spec_qps = kcalloc(caps->num_ports, sizeof(*caps->spec_qps), GFP_KERNEL);
+
+	if (!func_cap || !caps->spec_qps) {
+		mlx4_err(dev, "Failed to allocate memory for special qps cap\n");
+		err = -ENOMEM;
+		goto err_mem;
+	}
+
+	for (i = 1; i <= caps->num_ports; ++i) {
+		err = mlx4_QUERY_FUNC_CAP(dev, i, func_cap);
+		if (err) {
+			mlx4_err(dev, "QUERY_FUNC_CAP port command failed for port %d, aborting (%d)\n",
+				 i, err);
+			goto err_mem;
+		}
+		caps->spec_qps[i - 1] = func_cap->spec_qps;
+		caps->port_mask[i] = caps->port_type[i];
+		caps->phys_port_id[i] = func_cap->phys_port_id;
+		err = mlx4_get_slave_pkey_gid_tbl_len(dev, i,
+						      &caps->gid_table_len[i],
+						      &caps->pkey_table_len[i]);
+		if (err) {
+			mlx4_err(dev, "QUERY_PORT command failed for port %d, aborting (%d)\n",
+				 i, err);
+			goto err_mem;
+		}
+	}
+
+err_mem:
+	if (err)
+		mlx4_slave_destroy_special_qp_cap(dev);
+	kfree(func_cap);
+	return err;
+}
+
+static int mlx4_slave_cap(struct mlx4_dev *dev)
+{
+	int			   err;
+	u32			   page_size;
+	struct mlx4_dev_cap	   *dev_cap = NULL;
+	struct mlx4_func_cap	   *func_cap = NULL;
+	struct mlx4_init_hca_param *hca_param = NULL;
+
+	hca_param = kzalloc(sizeof(*hca_param), GFP_KERNEL);
+	func_cap = kzalloc(sizeof(*func_cap), GFP_KERNEL);
+	dev_cap = kzalloc(sizeof(*dev_cap), GFP_KERNEL);
+	if (!hca_param || !func_cap || !dev_cap) {
+		mlx4_err(dev, "Failed to allocate memory for slave_cap\n");
+		err = -ENOMEM;
+		goto free_mem;
+	}
+
+	err = mlx4_QUERY_HCA(dev, hca_param);
+	if (err) {
+		mlx4_err(dev, "QUERY_HCA command failed, aborting\n");
+		goto free_mem;
+	}
+
+	/* fail if the hca has an unknown global capability
+	 * at this time global_caps should be always zeroed
+	 */
+	if (hca_param->global_caps) {
+		mlx4_err(dev, "Unknown hca global capabilities\n");
+		err = -EINVAL;
+		goto free_mem;
+	}
+
+	dev->caps.hca_core_clock = hca_param->hca_core_clock;
+
+	dev->caps.max_qp_dest_rdma = 1 << hca_param->log_rd_per_qp;
+	err = mlx4_dev_cap(dev, dev_cap);
+	if (err) {
+		mlx4_err(dev, "QUERY_DEV_CAP command failed, aborting\n");
+		goto free_mem;
+	}
+
+	err = mlx4_QUERY_FW(dev);
+	if (err)
+		mlx4_err(dev, "QUERY_FW command failed: could not get FW version\n");
+
+	page_size = ~dev->caps.page_size_cap + 1;
+	mlx4_warn(dev, "HCA minimum page size:%d\n", page_size);
+	if (page_size > PAGE_SIZE) {
+		mlx4_err(dev, "HCA minimum page size of %d bigger than kernel PAGE_SIZE of %ld, aborting\n",
+			 page_size, PAGE_SIZE);
+		err = -ENODEV;
+		goto free_mem;
+	}
+
+	/* Set uar_page_shift for VF */
+	dev->uar_page_shift = hca_param->uar_page_sz + 12;
+
+	/* Make sure the master uar page size is valid */
+	if (dev->uar_page_shift > PAGE_SHIFT) {
+		mlx4_err(dev,
+			 "Invalid configuration: uar page size is larger than system page size\n");
+		err = -ENODEV;
+		goto free_mem;
+	}
+
+	/* Set reserved_uars based on the uar_page_shift */
+	mlx4_set_num_reserved_uars(dev, dev_cap);
+
+	/* Although uar page size in FW differs from system page size,
+	 * upper software layers (mlx4_ib, mlx4_en and part of mlx4_core)
+	 * still works with assumption that uar page size == system page size
+	 */
+	dev->caps.uar_page_size = PAGE_SIZE;
+
+	err = mlx4_QUERY_FUNC_CAP(dev, 0, func_cap);
+	if (err) {
+		mlx4_err(dev, "QUERY_FUNC_CAP general command failed, aborting (%d)\n",
+			 err);
+		goto free_mem;
+	}
+
+	if ((func_cap->pf_context_behaviour | PF_CONTEXT_BEHAVIOUR_MASK) !=
+	    PF_CONTEXT_BEHAVIOUR_MASK) {
+		mlx4_err(dev, "Unknown pf context behaviour %x known flags %x\n",
+			 func_cap->pf_context_behaviour,
+			 PF_CONTEXT_BEHAVIOUR_MASK);
+		err = -EINVAL;
+		goto free_mem;
+	}
+
+	dev->caps.num_ports		= func_cap->num_ports;
+	dev->quotas.qp			= func_cap->qp_quota;
+	dev->quotas.srq			= func_cap->srq_quota;
+	dev->quotas.cq			= func_cap->cq_quota;
+	dev->quotas.mpt			= func_cap->mpt_quota;
+	dev->quotas.mtt			= func_cap->mtt_quota;
+	dev->caps.num_qps		= 1 << hca_param->log_num_qps;
+	dev->caps.num_srqs		= 1 << hca_param->log_num_srqs;
+	dev->caps.num_cqs		= 1 << hca_param->log_num_cqs;
+	dev->caps.num_mpts		= 1 << hca_param->log_mpt_sz;
+	dev->caps.num_eqs		= func_cap->max_eq;
+	dev->caps.reserved_eqs		= func_cap->reserved_eq;
+	dev->caps.reserved_lkey		= func_cap->reserved_lkey;
+	dev->caps.num_pds               = MLX4_NUM_PDS;
+	dev->caps.num_mgms              = 0;
+	dev->caps.num_amgms             = 0;
+
+	if (dev->caps.num_ports > MLX4_MAX_PORTS) {
+		mlx4_err(dev, "HCA has %d ports, but we only support %d, aborting\n",
+			 dev->caps.num_ports, MLX4_MAX_PORTS);
+		err = -ENODEV;
+		goto free_mem;
+	}
+
+	mlx4_replace_zero_macs(dev);
+
+	err = mlx4_slave_special_qp_cap(dev);
+	if (err) {
+		mlx4_err(dev, "Set special QP caps failed. aborting\n");
+		goto free_mem;
+	}
+
+	if (dev->caps.uar_page_size * (dev->caps.num_uars -
+				       dev->caps.reserved_uars) >
+				       pci_resource_len(dev->persist->pdev,
+							2)) {
+		mlx4_err(dev, "HCA reported UAR region size of 0x%x bigger than PCI resource 2 size of 0x%llx, aborting\n",
+			 dev->caps.uar_page_size * dev->caps.num_uars,
+			 (unsigned long long)
+			 pci_resource_len(dev->persist->pdev, 2));
+		err = -ENOMEM;
+		goto err_mem;
+	}
+
+	if (hca_param->dev_cap_enabled & MLX4_DEV_CAP_64B_EQE_ENABLED) {
+		dev->caps.eqe_size   = 64;
+		dev->caps.eqe_factor = 1;
+	} else {
+		dev->caps.eqe_size   = 32;
+		dev->caps.eqe_factor = 0;
+	}
+
+	if (hca_param->dev_cap_enabled & MLX4_DEV_CAP_64B_CQE_ENABLED) {
+		dev->caps.cqe_size   = 64;
+		dev->caps.userspace_caps |= MLX4_USER_DEV_CAP_LARGE_CQE;
+	} else {
+		dev->caps.cqe_size   = 32;
+	}
+
+	if (hca_param->dev_cap_enabled & MLX4_DEV_CAP_EQE_STRIDE_ENABLED) {
+		dev->caps.eqe_size = hca_param->eqe_size;
+		dev->caps.eqe_factor = 0;
+	}
+
+	if (hca_param->dev_cap_enabled & MLX4_DEV_CAP_CQE_STRIDE_ENABLED) {
+		dev->caps.cqe_size = hca_param->cqe_size;
+		/* User still need to know when CQE > 32B */
+		dev->caps.userspace_caps |= MLX4_USER_DEV_CAP_LARGE_CQE;
+	}
+
+	dev->caps.flags2 &= ~MLX4_DEV_CAP_FLAG2_TS;
+	mlx4_warn(dev, "Timestamping is not supported in slave mode\n");
+
+	dev->caps.flags2 &= ~MLX4_DEV_CAP_FLAG2_USER_MAC_EN;
+	mlx4_dbg(dev, "User MAC FW update is not supported in slave mode\n");
+
+	slave_adjust_steering_mode(dev, dev_cap, hca_param);
+	mlx4_dbg(dev, "RSS support for IP fragments is %s\n",
+		 hca_param->rss_ip_frags ? "on" : "off");
+
+	if (func_cap->extra_flags & MLX4_QUERY_FUNC_FLAGS_BF_RES_QP &&
+	    dev->caps.bf_reg_size)
+		dev->caps.alloc_res_qp_mask |= MLX4_RESERVE_ETH_BF_QP;
+
+	if (func_cap->extra_flags & MLX4_QUERY_FUNC_FLAGS_A0_RES_QP)
+		dev->caps.alloc_res_qp_mask |= MLX4_RESERVE_A0_QP;
+
+err_mem:
+	if (err)
+		mlx4_slave_destroy_special_qp_cap(dev);
+free_mem:
+	kfree(hca_param);
+	kfree(func_cap);
+	kfree(dev_cap);
+	return err;
+}
+
+static void mlx4_request_modules(struct mlx4_dev *dev)
+{
+	int port;
+	int has_ib_port = false;
+	int has_eth_port = false;
+#define EN_DRV_NAME	"mlx4_en"
+#define IB_DRV_NAME	"mlx4_ib"
+
+	for (port = 1; port <= dev->caps.num_ports; port++) {
+		if (dev->caps.port_type[port] == MLX4_PORT_TYPE_IB)
+			has_ib_port = true;
+		else if (dev->caps.port_type[port] == MLX4_PORT_TYPE_ETH)
+			has_eth_port = true;
+	}
+
+	if (has_eth_port)
+		request_module_nowait(EN_DRV_NAME);
+	if (has_ib_port || (dev->caps.flags & MLX4_DEV_CAP_FLAG_IBOE))
+		request_module_nowait(IB_DRV_NAME);
+}
+
+/*
+ * Change the port configuration of the device.
+ * Every user of this function must hold the port mutex.
+ */
+int mlx4_change_port_types(struct mlx4_dev *dev,
+			   enum mlx4_port_type *port_types)
+{
+	int err = 0;
+	int change = 0;
+	int port;
+
+	for (port = 0; port <  dev->caps.num_ports; port++) {
+		/* Change the port type only if the new type is different
+		 * from the current, and not set to Auto */
+		if (port_types[port] != dev->caps.port_type[port + 1])
+			change = 1;
+	}
+	if (change) {
+		mlx4_unregister_device(dev);
+		for (port = 1; port <= dev->caps.num_ports; port++) {
+			mlx4_CLOSE_PORT(dev, port);
+			dev->caps.port_type[port] = port_types[port - 1];
+			err = mlx4_SET_PORT(dev, port, -1);
+			if (err) {
+				mlx4_err(dev, "Failed to set port %d, aborting\n",
+					 port);
+				goto out;
+			}
+		}
+		mlx4_set_port_mask(dev);
+		err = mlx4_register_device(dev);
+		if (err) {
+			mlx4_err(dev, "Failed to register device\n");
+			goto out;
+		}
+		mlx4_request_modules(dev);
+	}
+
+out:
+	return err;
+}
+
+static ssize_t show_port_type(struct device *dev,
+			      struct device_attribute *attr,
+			      char *buf)
+{
+	struct mlx4_port_info *info = container_of(attr, struct mlx4_port_info,
+						   port_attr);
+	struct mlx4_dev *mdev = info->dev;
+	char type[8];
+
+	sprintf(type, "%s",
+		(mdev->caps.port_type[info->port] == MLX4_PORT_TYPE_IB) ?
+		"ib" : "eth");
+	if (mdev->caps.possible_type[info->port] == MLX4_PORT_TYPE_AUTO)
+		sprintf(buf, "auto (%s)\n", type);
+	else
+		sprintf(buf, "%s\n", type);
+
+	return strlen(buf);
+}
+
+static int __set_port_type(struct mlx4_port_info *info,
+			   enum mlx4_port_type port_type)
+{
+	struct mlx4_dev *mdev = info->dev;
+	struct mlx4_priv *priv = mlx4_priv(mdev);
+	enum mlx4_port_type types[MLX4_MAX_PORTS];
+	enum mlx4_port_type new_types[MLX4_MAX_PORTS];
+	int i;
+	int err = 0;
+
+	if ((port_type & mdev->caps.supported_type[info->port]) != port_type) {
+		mlx4_err(mdev,
+			 "Requested port type for port %d is not supported on this HCA\n",
+			 info->port);
+		return -EOPNOTSUPP;
+	}
+
+	mlx4_stop_sense(mdev);
+	mutex_lock(&priv->port_mutex);
+	info->tmp_type = port_type;
+
+	/* Possible type is always the one that was delivered */
+	mdev->caps.possible_type[info->port] = info->tmp_type;
+
+	for (i = 0; i < mdev->caps.num_ports; i++) {
+		types[i] = priv->port[i+1].tmp_type ? priv->port[i+1].tmp_type :
+					mdev->caps.possible_type[i+1];
+		if (types[i] == MLX4_PORT_TYPE_AUTO)
+			types[i] = mdev->caps.port_type[i+1];
+	}
+
+	if (!(mdev->caps.flags & MLX4_DEV_CAP_FLAG_DPDP) &&
+	    !(mdev->caps.flags & MLX4_DEV_CAP_FLAG_SENSE_SUPPORT)) {
+		for (i = 1; i <= mdev->caps.num_ports; i++) {
+			if (mdev->caps.possible_type[i] == MLX4_PORT_TYPE_AUTO) {
+				mdev->caps.possible_type[i] = mdev->caps.port_type[i];
+				err = -EOPNOTSUPP;
+			}
+		}
+	}
+	if (err) {
+		mlx4_err(mdev, "Auto sensing is not supported on this HCA. Set only 'eth' or 'ib' for both ports (should be the same)\n");
+		goto out;
+	}
+
+	mlx4_do_sense_ports(mdev, new_types, types);
+
+	err = mlx4_check_port_params(mdev, new_types);
+	if (err)
+		goto out;
+
+	/* We are about to apply the changes after the configuration
+	 * was verified, no need to remember the temporary types
+	 * any more */
+	for (i = 0; i < mdev->caps.num_ports; i++)
+		priv->port[i + 1].tmp_type = 0;
+
+	err = mlx4_change_port_types(mdev, new_types);
+
+out:
+	mlx4_start_sense(mdev);
+	mutex_unlock(&priv->port_mutex);
+
+	return err;
+}
+
+static ssize_t set_port_type(struct device *dev,
+			     struct device_attribute *attr,
+			     const char *buf, size_t count)
+{
+	struct mlx4_port_info *info = container_of(attr, struct mlx4_port_info,
+						   port_attr);
+	struct mlx4_dev *mdev = info->dev;
+	enum mlx4_port_type port_type;
+	static DEFINE_MUTEX(set_port_type_mutex);
+	int err;
+
+	mutex_lock(&set_port_type_mutex);
+
+	if (!strcmp(buf, "ib\n")) {
+		port_type = MLX4_PORT_TYPE_IB;
+	} else if (!strcmp(buf, "eth\n")) {
+		port_type = MLX4_PORT_TYPE_ETH;
+	} else if (!strcmp(buf, "auto\n")) {
+		port_type = MLX4_PORT_TYPE_AUTO;
+	} else {
+		mlx4_err(mdev, "%s is not supported port type\n", buf);
+		err = -EINVAL;
+		goto err_out;
+	}
+
+	err = __set_port_type(info, port_type);
+
+err_out:
+	mutex_unlock(&set_port_type_mutex);
+
+	return err ? err : count;
+}
+
+enum ibta_mtu {
+	IB_MTU_256  = 1,
+	IB_MTU_512  = 2,
+	IB_MTU_1024 = 3,
+	IB_MTU_2048 = 4,
+	IB_MTU_4096 = 5
+};
+
+static inline int int_to_ibta_mtu(int mtu)
+{
+	switch (mtu) {
+	case 256:  return IB_MTU_256;
+	case 512:  return IB_MTU_512;
+	case 1024: return IB_MTU_1024;
+	case 2048: return IB_MTU_2048;
+	case 4096: return IB_MTU_4096;
+	default: return -1;
+	}
+}
+
+static inline int ibta_mtu_to_int(enum ibta_mtu mtu)
+{
+	switch (mtu) {
+	case IB_MTU_256:  return  256;
+	case IB_MTU_512:  return  512;
+	case IB_MTU_1024: return 1024;
+	case IB_MTU_2048: return 2048;
+	case IB_MTU_4096: return 4096;
+	default: return -1;
+	}
+}
+
+static ssize_t show_port_ib_mtu(struct device *dev,
+			     struct device_attribute *attr,
+			     char *buf)
+{
+	struct mlx4_port_info *info = container_of(attr, struct mlx4_port_info,
+						   port_mtu_attr);
+	struct mlx4_dev *mdev = info->dev;
+
+	if (mdev->caps.port_type[info->port] == MLX4_PORT_TYPE_ETH)
+		mlx4_warn(mdev, "port level mtu is only used for IB ports\n");
+
+	sprintf(buf, "%d\n",
+			ibta_mtu_to_int(mdev->caps.port_ib_mtu[info->port]));
+	return strlen(buf);
+}
+
+static ssize_t set_port_ib_mtu(struct device *dev,
+			     struct device_attribute *attr,
+			     const char *buf, size_t count)
+{
+	struct mlx4_port_info *info = container_of(attr, struct mlx4_port_info,
+						   port_mtu_attr);
+	struct mlx4_dev *mdev = info->dev;
+	struct mlx4_priv *priv = mlx4_priv(mdev);
+	int err, port, mtu, ibta_mtu = -1;
+
+	if (mdev->caps.port_type[info->port] == MLX4_PORT_TYPE_ETH) {
+		mlx4_warn(mdev, "port level mtu is only used for IB ports\n");
+		return -EINVAL;
+	}
+
+	err = kstrtoint(buf, 0, &mtu);
+	if (!err)
+		ibta_mtu = int_to_ibta_mtu(mtu);
+
+	if (err || ibta_mtu < 0) {
+		mlx4_err(mdev, "%s is invalid IBTA mtu\n", buf);
+		return -EINVAL;
+	}
+
+	mdev->caps.port_ib_mtu[info->port] = ibta_mtu;
+
+	mlx4_stop_sense(mdev);
+	mutex_lock(&priv->port_mutex);
+	mlx4_unregister_device(mdev);
+	for (port = 1; port <= mdev->caps.num_ports; port++) {
+		mlx4_CLOSE_PORT(mdev, port);
+		err = mlx4_SET_PORT(mdev, port, -1);
+		if (err) {
+			mlx4_err(mdev, "Failed to set port %d, aborting\n",
+				 port);
+			goto err_set_port;
+		}
+	}
+	err = mlx4_register_device(mdev);
+err_set_port:
+	mutex_unlock(&priv->port_mutex);
+	mlx4_start_sense(mdev);
+	return err ? err : count;
+}
+
+/* bond for multi-function device */
+#define MAX_MF_BOND_ALLOWED_SLAVES 63
+static int mlx4_mf_bond(struct mlx4_dev *dev)
+{
+	int err = 0;
+	int nvfs;
+	struct mlx4_slaves_pport slaves_port1;
+	struct mlx4_slaves_pport slaves_port2;
+	DECLARE_BITMAP(slaves_port_1_2, MLX4_MFUNC_MAX);
+
+	slaves_port1 = mlx4_phys_to_slaves_pport(dev, 1);
+	slaves_port2 = mlx4_phys_to_slaves_pport(dev, 2);
+	bitmap_and(slaves_port_1_2,
+		   slaves_port1.slaves, slaves_port2.slaves,
+		   dev->persist->num_vfs + 1);
+
+	/* only single port vfs are allowed */
+	if (bitmap_weight(slaves_port_1_2, dev->persist->num_vfs + 1) > 1) {
+		mlx4_warn(dev, "HA mode unsupported for dual ported VFs\n");
+		return -EINVAL;
+	}
+
+	/* number of virtual functions is number of total functions minus one
+	 * physical function for each port.
+	 */
+	nvfs = bitmap_weight(slaves_port1.slaves, dev->persist->num_vfs + 1) +
+		bitmap_weight(slaves_port2.slaves, dev->persist->num_vfs + 1) - 2;
+
+	/* limit on maximum allowed VFs */
+	if (nvfs > MAX_MF_BOND_ALLOWED_SLAVES) {
+		mlx4_warn(dev, "HA mode is not supported for %d VFs (max %d are allowed)\n",
+			  nvfs, MAX_MF_BOND_ALLOWED_SLAVES);
+		return -EINVAL;
+	}
+
+	if (dev->caps.steering_mode != MLX4_STEERING_MODE_DEVICE_MANAGED) {
+		mlx4_warn(dev, "HA mode unsupported for NON DMFS steering\n");
+		return -EINVAL;
+	}
+
+	err = mlx4_bond_mac_table(dev);
+	if (err)
+		return err;
+	err = mlx4_bond_vlan_table(dev);
+	if (err)
+		goto err1;
+	err = mlx4_bond_fs_rules(dev);
+	if (err)
+		goto err2;
+
+	return 0;
+err2:
+	(void)mlx4_unbond_vlan_table(dev);
+err1:
+	(void)mlx4_unbond_mac_table(dev);
+	return err;
+}
+
+static int mlx4_mf_unbond(struct mlx4_dev *dev)
+{
+	int ret, ret1;
+
+	ret = mlx4_unbond_fs_rules(dev);
+	if (ret)
+		mlx4_warn(dev, "multifunction unbond for flow rules failed (%d)\n", ret);
+	ret1 = mlx4_unbond_mac_table(dev);
+	if (ret1) {
+		mlx4_warn(dev, "multifunction unbond for MAC table failed (%d)\n", ret1);
+		ret = ret1;
+	}
+	ret1 = mlx4_unbond_vlan_table(dev);
+	if (ret1) {
+		mlx4_warn(dev, "multifunction unbond for VLAN table failed (%d)\n", ret1);
+		ret = ret1;
+	}
+	return ret;
+}
+
+int mlx4_bond(struct mlx4_dev *dev)
+{
+	int ret = 0;
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	mutex_lock(&priv->bond_mutex);
+
+	if (!mlx4_is_bonded(dev)) {
+		ret = mlx4_do_bond(dev, true);
+		if (ret)
+			mlx4_err(dev, "Failed to bond device: %d\n", ret);
+		if (!ret && mlx4_is_master(dev)) {
+			ret = mlx4_mf_bond(dev);
+			if (ret) {
+				mlx4_err(dev, "bond for multifunction failed\n");
+				mlx4_do_bond(dev, false);
+			}
+		}
+	}
+
+	mutex_unlock(&priv->bond_mutex);
+	if (!ret)
+		mlx4_dbg(dev, "Device is bonded\n");
+
+	return ret;
+}
+EXPORT_SYMBOL_GPL(mlx4_bond);
+
+int mlx4_unbond(struct mlx4_dev *dev)
+{
+	int ret = 0;
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	mutex_lock(&priv->bond_mutex);
+
+	if (mlx4_is_bonded(dev)) {
+		int ret2 = 0;
+
+		ret = mlx4_do_bond(dev, false);
+		if (ret)
+			mlx4_err(dev, "Failed to unbond device: %d\n", ret);
+		if (mlx4_is_master(dev))
+			ret2 = mlx4_mf_unbond(dev);
+		if (ret2) {
+			mlx4_warn(dev, "Failed to unbond device for multifunction (%d)\n", ret2);
+			ret = ret2;
+		}
+	}
+
+	mutex_unlock(&priv->bond_mutex);
+	if (!ret)
+		mlx4_dbg(dev, "Device is unbonded\n");
+
+	return ret;
+}
+EXPORT_SYMBOL_GPL(mlx4_unbond);
+
+
+int mlx4_port_map_set(struct mlx4_dev *dev, struct mlx4_port_map *v2p)
+{
+	u8 port1 = v2p->port1;
+	u8 port2 = v2p->port2;
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int err;
+
+	if (!(dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_PORT_REMAP))
+		return -EOPNOTSUPP;
+
+	mutex_lock(&priv->bond_mutex);
+
+	/* zero means keep current mapping for this port */
+	if (port1 == 0)
+		port1 = priv->v2p.port1;
+	if (port2 == 0)
+		port2 = priv->v2p.port2;
+
+	if ((port1 < 1) || (port1 > MLX4_MAX_PORTS) ||
+	    (port2 < 1) || (port2 > MLX4_MAX_PORTS) ||
+	    (port1 == 2 && port2 == 1)) {
+		/* besides boundary checks cross mapping makes
+		 * no sense and therefore not allowed */
+		err = -EINVAL;
+	} else if ((port1 == priv->v2p.port1) &&
+		 (port2 == priv->v2p.port2)) {
+		err = 0;
+	} else {
+		err = mlx4_virt2phy_port_map(dev, port1, port2);
+		if (!err) {
+			mlx4_dbg(dev, "port map changed: [%d][%d]\n",
+				 port1, port2);
+			priv->v2p.port1 = port1;
+			priv->v2p.port2 = port2;
+		} else {
+			mlx4_err(dev, "Failed to change port mape: %d\n", err);
+		}
+	}
+
+	mutex_unlock(&priv->bond_mutex);
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx4_port_map_set);
+
+static int mlx4_load_fw(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int err;
+
+	priv->fw.fw_icm = mlx4_alloc_icm(dev, priv->fw.fw_pages,
+					 GFP_HIGHUSER | __GFP_NOWARN, 0);
+	if (!priv->fw.fw_icm) {
+		mlx4_err(dev, "Couldn't allocate FW area, aborting\n");
+		return -ENOMEM;
+	}
+
+	err = mlx4_MAP_FA(dev, priv->fw.fw_icm);
+	if (err) {
+		mlx4_err(dev, "MAP_FA command failed, aborting\n");
+		goto err_free;
+	}
+
+	err = mlx4_RUN_FW(dev);
+	if (err) {
+		mlx4_err(dev, "RUN_FW command failed, aborting\n");
+		goto err_unmap_fa;
+	}
+
+	return 0;
+
+err_unmap_fa:
+	mlx4_UNMAP_FA(dev);
+
+err_free:
+	mlx4_free_icm(dev, priv->fw.fw_icm, 0);
+	return err;
+}
+
+static int mlx4_init_cmpt_table(struct mlx4_dev *dev, u64 cmpt_base,
+				int cmpt_entry_sz)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int err;
+	int num_eqs;
+
+	err = mlx4_init_icm_table(dev, &priv->qp_table.cmpt_table,
+				  cmpt_base +
+				  ((u64) (MLX4_CMPT_TYPE_QP *
+					  cmpt_entry_sz) << MLX4_CMPT_SHIFT),
+				  cmpt_entry_sz, dev->caps.num_qps,
+				  dev->caps.reserved_qps_cnt[MLX4_QP_REGION_FW],
+				  0, 0);
+	if (err)
+		goto err;
+
+	err = mlx4_init_icm_table(dev, &priv->srq_table.cmpt_table,
+				  cmpt_base +
+				  ((u64) (MLX4_CMPT_TYPE_SRQ *
+					  cmpt_entry_sz) << MLX4_CMPT_SHIFT),
+				  cmpt_entry_sz, dev->caps.num_srqs,
+				  dev->caps.reserved_srqs, 0, 0);
+	if (err)
+		goto err_qp;
+
+	err = mlx4_init_icm_table(dev, &priv->cq_table.cmpt_table,
+				  cmpt_base +
+				  ((u64) (MLX4_CMPT_TYPE_CQ *
+					  cmpt_entry_sz) << MLX4_CMPT_SHIFT),
+				  cmpt_entry_sz, dev->caps.num_cqs,
+				  dev->caps.reserved_cqs, 0, 0);
+	if (err)
+		goto err_srq;
+
+	num_eqs = dev->phys_caps.num_phys_eqs;
+	err = mlx4_init_icm_table(dev, &priv->eq_table.cmpt_table,
+				  cmpt_base +
+				  ((u64) (MLX4_CMPT_TYPE_EQ *
+					  cmpt_entry_sz) << MLX4_CMPT_SHIFT),
+				  cmpt_entry_sz, num_eqs, num_eqs, 0, 0);
+	if (err)
+		goto err_cq;
+
+	return 0;
+
+err_cq:
+	mlx4_cleanup_icm_table(dev, &priv->cq_table.cmpt_table);
+
+err_srq:
+	mlx4_cleanup_icm_table(dev, &priv->srq_table.cmpt_table);
+
+err_qp:
+	mlx4_cleanup_icm_table(dev, &priv->qp_table.cmpt_table);
+
+err:
+	return err;
+}
+
+static int mlx4_init_icm(struct mlx4_dev *dev, struct mlx4_dev_cap *dev_cap,
+			 struct mlx4_init_hca_param *init_hca, u64 icm_size)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	u64 aux_pages;
+	int num_eqs;
+	int err;
+
+	err = mlx4_SET_ICM_SIZE(dev, icm_size, &aux_pages);
+	if (err) {
+		mlx4_err(dev, "SET_ICM_SIZE command failed, aborting\n");
+		return err;
+	}
+
+	mlx4_dbg(dev, "%lld KB of HCA context requires %lld KB aux memory\n",
+		 (unsigned long long) icm_size >> 10,
+		 (unsigned long long) aux_pages << 2);
+
+	priv->fw.aux_icm = mlx4_alloc_icm(dev, aux_pages,
+					  GFP_HIGHUSER | __GFP_NOWARN, 0);
+	if (!priv->fw.aux_icm) {
+		mlx4_err(dev, "Couldn't allocate aux memory, aborting\n");
+		return -ENOMEM;
+	}
+
+	err = mlx4_MAP_ICM_AUX(dev, priv->fw.aux_icm);
+	if (err) {
+		mlx4_err(dev, "MAP_ICM_AUX command failed, aborting\n");
+		goto err_free_aux;
+	}
+
+	err = mlx4_init_cmpt_table(dev, init_hca->cmpt_base, dev_cap->cmpt_entry_sz);
+	if (err) {
+		mlx4_err(dev, "Failed to map cMPT context memory, aborting\n");
+		goto err_unmap_aux;
+	}
+
+
+	num_eqs = dev->phys_caps.num_phys_eqs;
+	err = mlx4_init_icm_table(dev, &priv->eq_table.table,
+				  init_hca->eqc_base, dev_cap->eqc_entry_sz,
+				  num_eqs, num_eqs, 0, 0);
+	if (err) {
+		mlx4_err(dev, "Failed to map EQ context memory, aborting\n");
+		goto err_unmap_cmpt;
+	}
+
+	/*
+	 * Reserved MTT entries must be aligned up to a cacheline
+	 * boundary, since the FW will write to them, while the driver
+	 * writes to all other MTT entries. (The variable
+	 * dev->caps.mtt_entry_sz below is really the MTT segment
+	 * size, not the raw entry size)
+	 */
+	dev->caps.reserved_mtts =
+		ALIGN(dev->caps.reserved_mtts * dev->caps.mtt_entry_sz,
+		      dma_get_cache_alignment()) / dev->caps.mtt_entry_sz;
+
+	err = mlx4_init_icm_table(dev, &priv->mr_table.mtt_table,
+				  init_hca->mtt_base,
+				  dev->caps.mtt_entry_sz,
+				  dev->caps.num_mtts,
+				  dev->caps.reserved_mtts, 1, 0);
+	if (err) {
+		mlx4_err(dev, "Failed to map MTT context memory, aborting\n");
+		goto err_unmap_eq;
+	}
+
+	err = mlx4_init_icm_table(dev, &priv->mr_table.dmpt_table,
+				  init_hca->dmpt_base,
+				  dev_cap->dmpt_entry_sz,
+				  dev->caps.num_mpts,
+				  dev->caps.reserved_mrws, 1, 1);
+	if (err) {
+		mlx4_err(dev, "Failed to map dMPT context memory, aborting\n");
+		goto err_unmap_mtt;
+	}
+
+	err = mlx4_init_icm_table(dev, &priv->qp_table.qp_table,
+				  init_hca->qpc_base,
+				  dev_cap->qpc_entry_sz,
+				  dev->caps.num_qps,
+				  dev->caps.reserved_qps_cnt[MLX4_QP_REGION_FW],
+				  0, 0);
+	if (err) {
+		mlx4_err(dev, "Failed to map QP context memory, aborting\n");
+		goto err_unmap_dmpt;
+	}
+
+	err = mlx4_init_icm_table(dev, &priv->qp_table.auxc_table,
+				  init_hca->auxc_base,
+				  dev_cap->aux_entry_sz,
+				  dev->caps.num_qps,
+				  dev->caps.reserved_qps_cnt[MLX4_QP_REGION_FW],
+				  0, 0);
+	if (err) {
+		mlx4_err(dev, "Failed to map AUXC context memory, aborting\n");
+		goto err_unmap_qp;
+	}
+
+	err = mlx4_init_icm_table(dev, &priv->qp_table.altc_table,
+				  init_hca->altc_base,
+				  dev_cap->altc_entry_sz,
+				  dev->caps.num_qps,
+				  dev->caps.reserved_qps_cnt[MLX4_QP_REGION_FW],
+				  0, 0);
+	if (err) {
+		mlx4_err(dev, "Failed to map ALTC context memory, aborting\n");
+		goto err_unmap_auxc;
+	}
+
+	err = mlx4_init_icm_table(dev, &priv->qp_table.rdmarc_table,
+				  init_hca->rdmarc_base,
+				  dev_cap->rdmarc_entry_sz << priv->qp_table.rdmarc_shift,
+				  dev->caps.num_qps,
+				  dev->caps.reserved_qps_cnt[MLX4_QP_REGION_FW],
+				  0, 0);
+	if (err) {
+		mlx4_err(dev, "Failed to map RDMARC context memory, aborting\n");
+		goto err_unmap_altc;
+	}
+
+	err = mlx4_init_icm_table(dev, &priv->cq_table.table,
+				  init_hca->cqc_base,
+				  dev_cap->cqc_entry_sz,
+				  dev->caps.num_cqs,
+				  dev->caps.reserved_cqs, 0, 0);
+	if (err) {
+		mlx4_err(dev, "Failed to map CQ context memory, aborting\n");
+		goto err_unmap_rdmarc;
+	}
+
+	err = mlx4_init_icm_table(dev, &priv->srq_table.table,
+				  init_hca->srqc_base,
+				  dev_cap->srq_entry_sz,
+				  dev->caps.num_srqs,
+				  dev->caps.reserved_srqs, 0, 0);
+	if (err) {
+		mlx4_err(dev, "Failed to map SRQ context memory, aborting\n");
+		goto err_unmap_cq;
+	}
+
+	/*
+	 * For flow steering device managed mode it is required to use
+	 * mlx4_init_icm_table. For B0 steering mode it's not strictly
+	 * required, but for simplicity just map the whole multicast
+	 * group table now.  The table isn't very big and it's a lot
+	 * easier than trying to track ref counts.
+	 */
+	err = mlx4_init_icm_table(dev, &priv->mcg_table.table,
+				  init_hca->mc_base,
+				  mlx4_get_mgm_entry_size(dev),
+				  dev->caps.num_mgms + dev->caps.num_amgms,
+				  dev->caps.num_mgms + dev->caps.num_amgms,
+				  0, 0);
+	if (err) {
+		mlx4_err(dev, "Failed to map MCG context memory, aborting\n");
+		goto err_unmap_srq;
+	}
+
+	return 0;
+
+err_unmap_srq:
+	mlx4_cleanup_icm_table(dev, &priv->srq_table.table);
+
+err_unmap_cq:
+	mlx4_cleanup_icm_table(dev, &priv->cq_table.table);
+
+err_unmap_rdmarc:
+	mlx4_cleanup_icm_table(dev, &priv->qp_table.rdmarc_table);
+
+err_unmap_altc:
+	mlx4_cleanup_icm_table(dev, &priv->qp_table.altc_table);
+
+err_unmap_auxc:
+	mlx4_cleanup_icm_table(dev, &priv->qp_table.auxc_table);
+
+err_unmap_qp:
+	mlx4_cleanup_icm_table(dev, &priv->qp_table.qp_table);
+
+err_unmap_dmpt:
+	mlx4_cleanup_icm_table(dev, &priv->mr_table.dmpt_table);
+
+err_unmap_mtt:
+	mlx4_cleanup_icm_table(dev, &priv->mr_table.mtt_table);
+
+err_unmap_eq:
+	mlx4_cleanup_icm_table(dev, &priv->eq_table.table);
+
+err_unmap_cmpt:
+	mlx4_cleanup_icm_table(dev, &priv->eq_table.cmpt_table);
+	mlx4_cleanup_icm_table(dev, &priv->cq_table.cmpt_table);
+	mlx4_cleanup_icm_table(dev, &priv->srq_table.cmpt_table);
+	mlx4_cleanup_icm_table(dev, &priv->qp_table.cmpt_table);
+
+err_unmap_aux:
+	mlx4_UNMAP_ICM_AUX(dev);
+
+err_free_aux:
+	mlx4_free_icm(dev, priv->fw.aux_icm, 0);
+
+	return err;
+}
+
+static void mlx4_free_icms(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	mlx4_cleanup_icm_table(dev, &priv->mcg_table.table);
+	mlx4_cleanup_icm_table(dev, &priv->srq_table.table);
+	mlx4_cleanup_icm_table(dev, &priv->cq_table.table);
+	mlx4_cleanup_icm_table(dev, &priv->qp_table.rdmarc_table);
+	mlx4_cleanup_icm_table(dev, &priv->qp_table.altc_table);
+	mlx4_cleanup_icm_table(dev, &priv->qp_table.auxc_table);
+	mlx4_cleanup_icm_table(dev, &priv->qp_table.qp_table);
+	mlx4_cleanup_icm_table(dev, &priv->mr_table.dmpt_table);
+	mlx4_cleanup_icm_table(dev, &priv->mr_table.mtt_table);
+	mlx4_cleanup_icm_table(dev, &priv->eq_table.table);
+	mlx4_cleanup_icm_table(dev, &priv->eq_table.cmpt_table);
+	mlx4_cleanup_icm_table(dev, &priv->cq_table.cmpt_table);
+	mlx4_cleanup_icm_table(dev, &priv->srq_table.cmpt_table);
+	mlx4_cleanup_icm_table(dev, &priv->qp_table.cmpt_table);
+
+	mlx4_UNMAP_ICM_AUX(dev);
+	mlx4_free_icm(dev, priv->fw.aux_icm, 0);
+}
+
+static void mlx4_slave_exit(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	mutex_lock(&priv->cmd.slave_cmd_mutex);
+	if (mlx4_comm_cmd(dev, MLX4_COMM_CMD_RESET, 0, MLX4_COMM_CMD_NA_OP,
+			  MLX4_COMM_TIME))
+		mlx4_warn(dev, "Failed to close slave function\n");
+	mutex_unlock(&priv->cmd.slave_cmd_mutex);
+}
+
+static int map_bf_area(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	resource_size_t bf_start;
+	resource_size_t bf_len;
+	int err = 0;
+
+	if (!dev->caps.bf_reg_size)
+		return -ENXIO;
+
+	bf_start = pci_resource_start(dev->persist->pdev, 2) +
+			(dev->caps.num_uars << PAGE_SHIFT);
+	bf_len = pci_resource_len(dev->persist->pdev, 2) -
+			(dev->caps.num_uars << PAGE_SHIFT);
+	priv->bf_mapping = io_mapping_create_wc(bf_start, bf_len);
+	if (!priv->bf_mapping)
+		err = -ENOMEM;
+
+	return err;
+}
+
+static void unmap_bf_area(struct mlx4_dev *dev)
+{
+	if (mlx4_priv(dev)->bf_mapping)
+		io_mapping_free(mlx4_priv(dev)->bf_mapping);
+}
+
+u64 mlx4_read_clock(struct mlx4_dev *dev)
+{
+	u32 clockhi, clocklo, clockhi1;
+	u64 cycles;
+	int i;
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	for (i = 0; i < 10; i++) {
+		clockhi = swab32(readl(priv->clock_mapping));
+		clocklo = swab32(readl(priv->clock_mapping + 4));
+		clockhi1 = swab32(readl(priv->clock_mapping));
+		if (clockhi == clockhi1)
+			break;
+	}
+
+	cycles = (u64) clockhi << 32 | (u64) clocklo;
+
+	return cycles;
+}
+EXPORT_SYMBOL_GPL(mlx4_read_clock);
+
+
+static int map_internal_clock(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	priv->clock_mapping =
+		ioremap(pci_resource_start(dev->persist->pdev,
+					   priv->fw.clock_bar) +
+			priv->fw.clock_offset, MLX4_CLOCK_SIZE);
+
+	if (!priv->clock_mapping)
+		return -ENOMEM;
+
+	return 0;
+}
+
+int mlx4_get_internal_clock_params(struct mlx4_dev *dev,
+				   struct mlx4_clock_params *params)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	if (mlx4_is_slave(dev))
+		return -EOPNOTSUPP;
+
+	if (!dev->caps.map_clock_to_user) {
+		mlx4_dbg(dev, "Map clock to user is not supported.\n");
+		return -EOPNOTSUPP;
+	}
+
+	if (!params)
+		return -EINVAL;
+
+	params->bar = priv->fw.clock_bar;
+	params->offset = priv->fw.clock_offset;
+	params->size = MLX4_CLOCK_SIZE;
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(mlx4_get_internal_clock_params);
+
+static void unmap_internal_clock(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	if (priv->clock_mapping)
+		iounmap(priv->clock_mapping);
+}
+
+static void mlx4_close_hca(struct mlx4_dev *dev)
+{
+	unmap_internal_clock(dev);
+	unmap_bf_area(dev);
+	if (mlx4_is_slave(dev))
+		mlx4_slave_exit(dev);
+	else {
+		mlx4_CLOSE_HCA(dev, 0);
+		mlx4_free_icms(dev);
+	}
+}
+
+static void mlx4_close_fw(struct mlx4_dev *dev)
+{
+	if (!mlx4_is_slave(dev)) {
+		mlx4_UNMAP_FA(dev);
+		mlx4_free_icm(dev, mlx4_priv(dev)->fw.fw_icm, 0);
+	}
+}
+
+static int mlx4_comm_check_offline(struct mlx4_dev *dev)
+{
+#define COMM_CHAN_OFFLINE_OFFSET 0x09
+
+	u32 comm_flags;
+	u32 offline_bit;
+	unsigned long end;
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	end = msecs_to_jiffies(MLX4_COMM_OFFLINE_TIME_OUT) + jiffies;
+	while (time_before(jiffies, end)) {
+		comm_flags = swab32(readl((__iomem char *)priv->mfunc.comm +
+					  MLX4_COMM_CHAN_FLAGS));
+		offline_bit = (comm_flags &
+			       (u32)(1 << COMM_CHAN_OFFLINE_OFFSET));
+		if (!offline_bit)
+			return 0;
+
+		/* If device removal has been requested,
+		 * do not continue retrying.
+		 */
+		if (dev->persist->interface_state &
+		    MLX4_INTERFACE_STATE_NOWAIT)
+			break;
+
+		/* There are cases as part of AER/Reset flow that PF needs
+		 * around 100 msec to load. We therefore sleep for 100 msec
+		 * to allow other tasks to make use of that CPU during this
+		 * time interval.
+		 */
+		msleep(100);
+	}
+	mlx4_err(dev, "Communication channel is offline.\n");
+	return -EIO;
+}
+
+static void mlx4_reset_vf_support(struct mlx4_dev *dev)
+{
+#define COMM_CHAN_RST_OFFSET 0x1e
+
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	u32 comm_rst;
+	u32 comm_caps;
+
+	comm_caps = swab32(readl((__iomem char *)priv->mfunc.comm +
+				 MLX4_COMM_CHAN_CAPS));
+	comm_rst = (comm_caps & (u32)(1 << COMM_CHAN_RST_OFFSET));
+
+	if (comm_rst)
+		dev->caps.vf_caps |= MLX4_VF_CAP_FLAG_RESET;
+}
+
+static int mlx4_init_slave(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	u64 dma = (u64) priv->mfunc.vhcr_dma;
+	int ret_from_reset = 0;
+	u32 slave_read;
+	u32 cmd_channel_ver;
+
+	if (atomic_read(&pf_loading)) {
+		mlx4_warn(dev, "PF is not ready - Deferring probe\n");
+		return -EPROBE_DEFER;
+	}
+
+	mutex_lock(&priv->cmd.slave_cmd_mutex);
+	priv->cmd.max_cmds = 1;
+	if (mlx4_comm_check_offline(dev)) {
+		mlx4_err(dev, "PF is not responsive, skipping initialization\n");
+		goto err_offline;
+	}
+
+	mlx4_reset_vf_support(dev);
+	mlx4_warn(dev, "Sending reset\n");
+	ret_from_reset = mlx4_comm_cmd(dev, MLX4_COMM_CMD_RESET, 0,
+				       MLX4_COMM_CMD_NA_OP, MLX4_COMM_TIME);
+	/* if we are in the middle of flr the slave will try
+	 * NUM_OF_RESET_RETRIES times before leaving.*/
+	if (ret_from_reset) {
+		if (MLX4_DELAY_RESET_SLAVE == ret_from_reset) {
+			mlx4_warn(dev, "slave is currently in the middle of FLR - Deferring probe\n");
+			mutex_unlock(&priv->cmd.slave_cmd_mutex);
+			return -EPROBE_DEFER;
+		} else
+			goto err;
+	}
+
+	/* check the driver version - the slave I/F revision
+	 * must match the master's */
+	slave_read = swab32(readl(&priv->mfunc.comm->slave_read));
+	cmd_channel_ver = mlx4_comm_get_version();
+
+	if (MLX4_COMM_GET_IF_REV(cmd_channel_ver) !=
+		MLX4_COMM_GET_IF_REV(slave_read)) {
+		mlx4_err(dev, "slave driver version is not supported by the master\n");
+		goto err;
+	}
+
+	mlx4_warn(dev, "Sending vhcr0\n");
+	if (mlx4_comm_cmd(dev, MLX4_COMM_CMD_VHCR0, dma >> 48,
+			     MLX4_COMM_CMD_NA_OP, MLX4_COMM_TIME))
+		goto err;
+	if (mlx4_comm_cmd(dev, MLX4_COMM_CMD_VHCR1, dma >> 32,
+			     MLX4_COMM_CMD_NA_OP, MLX4_COMM_TIME))
+		goto err;
+	if (mlx4_comm_cmd(dev, MLX4_COMM_CMD_VHCR2, dma >> 16,
+			     MLX4_COMM_CMD_NA_OP, MLX4_COMM_TIME))
+		goto err;
+	if (mlx4_comm_cmd(dev, MLX4_COMM_CMD_VHCR_EN, dma,
+			  MLX4_COMM_CMD_NA_OP, MLX4_COMM_TIME))
+		goto err;
+
+	mutex_unlock(&priv->cmd.slave_cmd_mutex);
+	return 0;
+
+err:
+	mlx4_comm_cmd(dev, MLX4_COMM_CMD_RESET, 0, MLX4_COMM_CMD_NA_OP, 0);
+err_offline:
+	mutex_unlock(&priv->cmd.slave_cmd_mutex);
+	return -EIO;
+}
+
+static void mlx4_parav_master_pf_caps(struct mlx4_dev *dev)
+{
+	int i;
+
+	for (i = 1; i <= dev->caps.num_ports; i++) {
+		if (dev->caps.port_type[i] == MLX4_PORT_TYPE_ETH)
+			dev->caps.gid_table_len[i] =
+				mlx4_get_slave_num_gids(dev, 0, i);
+		else
+			dev->caps.gid_table_len[i] = 1;
+		dev->caps.pkey_table_len[i] =
+			dev->phys_caps.pkey_phys_table_len[i] - 1;
+	}
+}
+
+static int choose_log_fs_mgm_entry_size(int qp_per_entry)
+{
+	int i = MLX4_MIN_MGM_LOG_ENTRY_SIZE;
+
+	for (i = MLX4_MIN_MGM_LOG_ENTRY_SIZE; i <= MLX4_MAX_MGM_LOG_ENTRY_SIZE;
+	      i++) {
+		if (qp_per_entry <= 4 * ((1 << i) / 16 - 2))
+			break;
+	}
+
+	return (i <= MLX4_MAX_MGM_LOG_ENTRY_SIZE) ? i : -1;
+}
+
+static const char *dmfs_high_rate_steering_mode_str(int dmfs_high_steer_mode)
+{
+	switch (dmfs_high_steer_mode) {
+	case MLX4_STEERING_DMFS_A0_DEFAULT:
+		return "default performance";
+
+	case MLX4_STEERING_DMFS_A0_DYNAMIC:
+		return "dynamic hybrid mode";
+
+	case MLX4_STEERING_DMFS_A0_STATIC:
+		return "performance optimized for limited rule configuration (static)";
+
+	case MLX4_STEERING_DMFS_A0_DISABLE:
+		return "disabled performance optimized steering";
+
+	case MLX4_STEERING_DMFS_A0_NOT_SUPPORTED:
+		return "performance optimized steering not supported";
+
+	default:
+		return "Unrecognized mode";
+	}
+}
+
+#define MLX4_DMFS_A0_STEERING			(1UL << 2)
+
+static void choose_steering_mode(struct mlx4_dev *dev,
+				 struct mlx4_dev_cap *dev_cap)
+{
+	if (mlx4_log_num_mgm_entry_size <= 0) {
+		if ((-mlx4_log_num_mgm_entry_size) & MLX4_DMFS_A0_STEERING) {
+			if (dev->caps.dmfs_high_steer_mode ==
+			    MLX4_STEERING_DMFS_A0_NOT_SUPPORTED)
+				mlx4_err(dev, "DMFS high rate mode not supported\n");
+			else
+				dev->caps.dmfs_high_steer_mode =
+					MLX4_STEERING_DMFS_A0_STATIC;
+		}
+	}
+
+	if (mlx4_log_num_mgm_entry_size <= 0 &&
+	    dev_cap->flags2 & MLX4_DEV_CAP_FLAG2_FS_EN &&
+	    (!mlx4_is_mfunc(dev) ||
+	     (dev_cap->fs_max_num_qp_per_entry >=
+	     (dev->persist->num_vfs + 1))) &&
+	    choose_log_fs_mgm_entry_size(dev_cap->fs_max_num_qp_per_entry) >=
+		MLX4_MIN_MGM_LOG_ENTRY_SIZE) {
+		dev->oper_log_mgm_entry_size =
+			choose_log_fs_mgm_entry_size(dev_cap->fs_max_num_qp_per_entry);
+		dev->caps.steering_mode = MLX4_STEERING_MODE_DEVICE_MANAGED;
+		dev->caps.num_qp_per_mgm = dev_cap->fs_max_num_qp_per_entry;
+		dev->caps.fs_log_max_ucast_qp_range_size =
+			dev_cap->fs_log_max_ucast_qp_range_size;
+	} else {
+		if (dev->caps.dmfs_high_steer_mode !=
+		    MLX4_STEERING_DMFS_A0_NOT_SUPPORTED)
+			dev->caps.dmfs_high_steer_mode = MLX4_STEERING_DMFS_A0_DISABLE;
+		if (dev->caps.flags & MLX4_DEV_CAP_FLAG_VEP_UC_STEER &&
+		    dev->caps.flags & MLX4_DEV_CAP_FLAG_VEP_MC_STEER)
+			dev->caps.steering_mode = MLX4_STEERING_MODE_B0;
+		else {
+			dev->caps.steering_mode = MLX4_STEERING_MODE_A0;
+
+			if (dev->caps.flags & MLX4_DEV_CAP_FLAG_VEP_UC_STEER ||
+			    dev->caps.flags & MLX4_DEV_CAP_FLAG_VEP_MC_STEER)
+				mlx4_warn(dev, "Must have both UC_STEER and MC_STEER flags set to use B0 steering - falling back to A0 steering mode\n");
+		}
+		dev->oper_log_mgm_entry_size =
+			mlx4_log_num_mgm_entry_size > 0 ?
+			mlx4_log_num_mgm_entry_size :
+			MLX4_DEFAULT_MGM_LOG_ENTRY_SIZE;
+		dev->caps.num_qp_per_mgm = mlx4_get_qp_per_mgm(dev);
+	}
+	mlx4_dbg(dev, "Steering mode is: %s, oper_log_mgm_entry_size = %d, modparam log_num_mgm_entry_size = %d\n",
+		 mlx4_steering_mode_str(dev->caps.steering_mode),
+		 dev->oper_log_mgm_entry_size,
+		 mlx4_log_num_mgm_entry_size);
+}
+
+static void choose_tunnel_offload_mode(struct mlx4_dev *dev,
+				       struct mlx4_dev_cap *dev_cap)
+{
+	if (dev->caps.steering_mode == MLX4_STEERING_MODE_DEVICE_MANAGED &&
+	    dev_cap->flags2 & MLX4_DEV_CAP_FLAG2_VXLAN_OFFLOADS)
+		dev->caps.tunnel_offload_mode = MLX4_TUNNEL_OFFLOAD_MODE_VXLAN;
+	else
+		dev->caps.tunnel_offload_mode = MLX4_TUNNEL_OFFLOAD_MODE_NONE;
+
+	mlx4_dbg(dev, "Tunneling offload mode is: %s\n",  (dev->caps.tunnel_offload_mode
+		 == MLX4_TUNNEL_OFFLOAD_MODE_VXLAN) ? "vxlan" : "none");
+}
+
+static int mlx4_validate_optimized_steering(struct mlx4_dev *dev)
+{
+	int i;
+	struct mlx4_port_cap port_cap;
+
+	if (dev->caps.dmfs_high_steer_mode == MLX4_STEERING_DMFS_A0_NOT_SUPPORTED)
+		return -EINVAL;
+
+	for (i = 1; i <= dev->caps.num_ports; i++) {
+		if (mlx4_dev_port(dev, i, &port_cap)) {
+			mlx4_err(dev,
+				 "QUERY_DEV_CAP command failed, can't verify DMFS high rate steering.\n");
+		} else if ((dev->caps.dmfs_high_steer_mode !=
+			    MLX4_STEERING_DMFS_A0_DEFAULT) &&
+			   (port_cap.dmfs_optimized_state ==
+			    !!(dev->caps.dmfs_high_steer_mode ==
+			    MLX4_STEERING_DMFS_A0_DISABLE))) {
+			mlx4_err(dev,
+				 "DMFS high rate steer mode differ, driver requested %s but %s in FW.\n",
+				 dmfs_high_rate_steering_mode_str(
+					dev->caps.dmfs_high_steer_mode),
+				 (port_cap.dmfs_optimized_state ?
+					"enabled" : "disabled"));
+		}
+	}
+
+	return 0;
+}
+
+static int mlx4_init_fw(struct mlx4_dev *dev)
+{
+	struct mlx4_mod_stat_cfg   mlx4_cfg;
+	int err = 0;
+
+	if (!mlx4_is_slave(dev)) {
+		err = mlx4_QUERY_FW(dev);
+		if (err) {
+			if (err == -EACCES)
+				mlx4_info(dev, "non-primary physical function, skipping\n");
+			else
+				mlx4_err(dev, "QUERY_FW command failed, aborting\n");
+			return err;
+		}
+
+		err = mlx4_load_fw(dev);
+		if (err) {
+			mlx4_err(dev, "Failed to start FW, aborting\n");
+			return err;
+		}
+
+		mlx4_cfg.log_pg_sz_m = 1;
+		mlx4_cfg.log_pg_sz = 0;
+		err = mlx4_MOD_STAT_CFG(dev, &mlx4_cfg);
+		if (err)
+			mlx4_warn(dev, "Failed to override log_pg_sz parameter\n");
+	}
+
+	return err;
+}
+
+static int mlx4_init_hca(struct mlx4_dev *dev)
+{
+	struct mlx4_priv	  *priv = mlx4_priv(dev);
+	struct mlx4_init_hca_param *init_hca = NULL;
+	struct mlx4_dev_cap	  *dev_cap = NULL;
+	struct mlx4_adapter	   adapter;
+	struct mlx4_profile	   profile;
+	u64 icm_size;
+	struct mlx4_config_dev_params params;
+	int err;
+
+	if (!mlx4_is_slave(dev)) {
+		dev_cap = kzalloc(sizeof(*dev_cap), GFP_KERNEL);
+		init_hca = kzalloc(sizeof(*init_hca), GFP_KERNEL);
+
+		if (!dev_cap || !init_hca) {
+			err = -ENOMEM;
+			goto out_free;
+		}
+
+		err = mlx4_dev_cap(dev, dev_cap);
+		if (err) {
+			mlx4_err(dev, "QUERY_DEV_CAP command failed, aborting\n");
+			goto out_free;
+		}
+
+		choose_steering_mode(dev, dev_cap);
+		choose_tunnel_offload_mode(dev, dev_cap);
+
+		if (dev->caps.dmfs_high_steer_mode == MLX4_STEERING_DMFS_A0_STATIC &&
+		    mlx4_is_master(dev))
+			dev->caps.function_caps |= MLX4_FUNC_CAP_DMFS_A0_STATIC;
+
+		err = mlx4_get_phys_port_id(dev);
+		if (err)
+			mlx4_err(dev, "Fail to get physical port id\n");
+
+		if (mlx4_is_master(dev))
+			mlx4_parav_master_pf_caps(dev);
+
+		if (mlx4_low_memory_profile()) {
+			mlx4_info(dev, "Running from within kdump kernel. Using low memory profile\n");
+			profile = low_mem_profile;
+		} else {
+			profile = default_profile;
+		}
+		if (dev->caps.steering_mode ==
+		    MLX4_STEERING_MODE_DEVICE_MANAGED)
+			profile.num_mcg = MLX4_FS_NUM_MCG;
+
+		icm_size = mlx4_make_profile(dev, &profile, dev_cap,
+					     init_hca);
+		if ((long long) icm_size < 0) {
+			err = icm_size;
+			goto out_free;
+		}
+
+		if (enable_4k_uar || !dev->persist->num_vfs) {
+			init_hca->log_uar_sz = ilog2(dev->caps.num_uars) +
+						    PAGE_SHIFT - DEFAULT_UAR_PAGE_SHIFT;
+			init_hca->uar_page_sz = DEFAULT_UAR_PAGE_SHIFT - 12;
+		} else {
+			init_hca->log_uar_sz = ilog2(dev->caps.num_uars);
+			init_hca->uar_page_sz = PAGE_SHIFT - 12;
+		}
+
+		init_hca->mw_enabled = 0;
+		if (dev->caps.flags & MLX4_DEV_CAP_FLAG_MEM_WINDOW ||
+		    dev->caps.bmme_flags & MLX4_BMME_FLAG_TYPE_2_WIN)
+			init_hca->mw_enabled = INIT_HCA_TPT_MW_ENABLE;
+
+		err = mlx4_init_icm(dev, dev_cap, init_hca, icm_size);
+		if (err)
+			goto out_free;
+
+		err = mlx4_INIT_HCA(dev, init_hca);
+		if (err) {
+			mlx4_err(dev, "INIT_HCA command failed, aborting\n");
+			goto err_free_icm;
+		}
+
+		if (dev_cap->flags2 & MLX4_DEV_CAP_FLAG2_SYS_EQS) {
+			err = mlx4_query_func(dev, dev_cap);
+			if (err < 0) {
+				mlx4_err(dev, "QUERY_FUNC command failed, aborting.\n");
+				goto err_close;
+			} else if (err & MLX4_QUERY_FUNC_NUM_SYS_EQS) {
+				dev->caps.num_eqs = dev_cap->max_eqs;
+				dev->caps.reserved_eqs = dev_cap->reserved_eqs;
+				dev->caps.reserved_uars = dev_cap->reserved_uars;
+			}
+		}
+
+		/*
+		 * If TS is supported by FW
+		 * read HCA frequency by QUERY_HCA command
+		 */
+		if (dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_TS) {
+			err = mlx4_QUERY_HCA(dev, init_hca);
+			if (err) {
+				mlx4_err(dev, "QUERY_HCA command failed, disable timestamp\n");
+				dev->caps.flags2 &= ~MLX4_DEV_CAP_FLAG2_TS;
+			} else {
+				dev->caps.hca_core_clock =
+					init_hca->hca_core_clock;
+			}
+
+			/* In case we got HCA frequency 0 - disable timestamping
+			 * to avoid dividing by zero
+			 */
+			if (!dev->caps.hca_core_clock) {
+				dev->caps.flags2 &= ~MLX4_DEV_CAP_FLAG2_TS;
+				mlx4_err(dev,
+					 "HCA frequency is 0 - timestamping is not supported\n");
+			} else if (map_internal_clock(dev)) {
+				/*
+				 * Map internal clock,
+				 * in case of failure disable timestamping
+				 */
+				dev->caps.flags2 &= ~MLX4_DEV_CAP_FLAG2_TS;
+				mlx4_err(dev, "Failed to map internal clock. Timestamping is not supported\n");
+			}
+		}
+
+		if (dev->caps.dmfs_high_steer_mode !=
+		    MLX4_STEERING_DMFS_A0_NOT_SUPPORTED) {
+			if (mlx4_validate_optimized_steering(dev))
+				mlx4_warn(dev, "Optimized steering validation failed\n");
+
+			if (dev->caps.dmfs_high_steer_mode ==
+			    MLX4_STEERING_DMFS_A0_DISABLE) {
+				dev->caps.dmfs_high_rate_qpn_base =
+					dev->caps.reserved_qps_cnt[MLX4_QP_REGION_FW];
+				dev->caps.dmfs_high_rate_qpn_range =
+					MLX4_A0_STEERING_TABLE_SIZE;
+			}
+
+			mlx4_info(dev, "DMFS high rate steer mode is: %s\n",
+				  dmfs_high_rate_steering_mode_str(
+					dev->caps.dmfs_high_steer_mode));
+		}
+	} else {
+		err = mlx4_init_slave(dev);
+		if (err) {
+			if (err != -EPROBE_DEFER)
+				mlx4_err(dev, "Failed to initialize slave\n");
+			return err;
+		}
+
+		err = mlx4_slave_cap(dev);
+		if (err) {
+			mlx4_err(dev, "Failed to obtain slave caps\n");
+			goto err_close;
+		}
+	}
+
+	if (map_bf_area(dev))
+		mlx4_dbg(dev, "Failed to map blue flame area\n");
+
+	/*Only the master set the ports, all the rest got it from it.*/
+	if (!mlx4_is_slave(dev))
+		mlx4_set_port_mask(dev);
+
+	err = mlx4_QUERY_ADAPTER(dev, &adapter);
+	if (err) {
+		mlx4_err(dev, "QUERY_ADAPTER command failed, aborting\n");
+		goto unmap_bf;
+	}
+
+	/* Query CONFIG_DEV parameters */
+	err = mlx4_config_dev_retrieval(dev, &params);
+	if (err && err != -EOPNOTSUPP) {
+		mlx4_err(dev, "Failed to query CONFIG_DEV parameters\n");
+	} else if (!err) {
+		dev->caps.rx_checksum_flags_port[1] = params.rx_csum_flags_port_1;
+		dev->caps.rx_checksum_flags_port[2] = params.rx_csum_flags_port_2;
+	}
+	priv->eq_table.inta_pin = adapter.inta_pin;
+	memcpy(dev->board_id, adapter.board_id, sizeof(dev->board_id));
+
+	err = 0;
+	goto out_free;
+
+unmap_bf:
+	unmap_internal_clock(dev);
+	unmap_bf_area(dev);
+
+	if (mlx4_is_slave(dev))
+		mlx4_slave_destroy_special_qp_cap(dev);
+
+err_close:
+	if (mlx4_is_slave(dev))
+		mlx4_slave_exit(dev);
+	else
+		mlx4_CLOSE_HCA(dev, 0);
+
+err_free_icm:
+	if (!mlx4_is_slave(dev))
+		mlx4_free_icms(dev);
+
+out_free:
+	kfree(dev_cap);
+	kfree(init_hca);
+
+	return err;
+}
+
+static int mlx4_init_counters_table(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int nent_pow2;
+
+	if (!(dev->caps.flags & MLX4_DEV_CAP_FLAG_COUNTERS))
+		return -ENOENT;
+
+	if (!dev->caps.max_counters)
+		return -ENOSPC;
+
+	nent_pow2 = roundup_pow_of_two(dev->caps.max_counters);
+	/* reserve last counter index for sink counter */
+	return mlx4_bitmap_init(&priv->counters_bitmap, nent_pow2,
+				nent_pow2 - 1, 0,
+				nent_pow2 - dev->caps.max_counters + 1);
+}
+
+static void mlx4_cleanup_counters_table(struct mlx4_dev *dev)
+{
+	if (!(dev->caps.flags & MLX4_DEV_CAP_FLAG_COUNTERS))
+		return;
+
+	if (!dev->caps.max_counters)
+		return;
+
+	mlx4_bitmap_cleanup(&mlx4_priv(dev)->counters_bitmap);
+}
+
+static void mlx4_cleanup_default_counters(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int port;
+
+	for (port = 0; port < dev->caps.num_ports; port++)
+		if (priv->def_counter[port] != -1)
+			mlx4_counter_free(dev,  priv->def_counter[port]);
+}
+
+static int mlx4_allocate_default_counters(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int port, err = 0;
+	u32 idx;
+
+	for (port = 0; port < dev->caps.num_ports; port++)
+		priv->def_counter[port] = -1;
+
+	for (port = 0; port < dev->caps.num_ports; port++) {
+		err = mlx4_counter_alloc(dev, &idx, MLX4_RES_USAGE_DRIVER);
+
+		if (!err || err == -ENOSPC) {
+			priv->def_counter[port] = idx;
+			err = 0;
+		} else if (err == -ENOENT) {
+			err = 0;
+			continue;
+		} else if (mlx4_is_slave(dev) && err == -EINVAL) {
+			priv->def_counter[port] = MLX4_SINK_COUNTER_INDEX(dev);
+			mlx4_warn(dev, "can't allocate counter from old PF driver, using index %d\n",
+				  MLX4_SINK_COUNTER_INDEX(dev));
+			err = 0;
+		} else {
+			mlx4_err(dev, "%s: failed to allocate default counter port %d err %d\n",
+				 __func__, port + 1, err);
+			mlx4_cleanup_default_counters(dev);
+			return err;
+		}
+
+		mlx4_dbg(dev, "%s: default counter index %d for port %d\n",
+			 __func__, priv->def_counter[port], port + 1);
+	}
+
+	return err;
+}
+
+int __mlx4_counter_alloc(struct mlx4_dev *dev, u32 *idx)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	if (!(dev->caps.flags & MLX4_DEV_CAP_FLAG_COUNTERS))
+		return -ENOENT;
+
+	*idx = mlx4_bitmap_alloc(&priv->counters_bitmap);
+	if (*idx == -1) {
+		*idx = MLX4_SINK_COUNTER_INDEX(dev);
+		return -ENOSPC;
+	}
+
+	return 0;
+}
+
+int mlx4_counter_alloc(struct mlx4_dev *dev, u32 *idx, u8 usage)
+{
+	u32 in_modifier = RES_COUNTER | (((u32)usage & 3) << 30);
+	u64 out_param;
+	int err;
+
+	if (mlx4_is_mfunc(dev)) {
+		err = mlx4_cmd_imm(dev, 0, &out_param, in_modifier,
+				   RES_OP_RESERVE, MLX4_CMD_ALLOC_RES,
+				   MLX4_CMD_TIME_CLASS_A, MLX4_CMD_WRAPPED);
+		if (!err)
+			*idx = get_param_l(&out_param);
+		if (WARN_ON(err == -ENOSPC))
+			err = -EINVAL;
+		return err;
+	}
+	return __mlx4_counter_alloc(dev, idx);
+}
+EXPORT_SYMBOL_GPL(mlx4_counter_alloc);
+
+static int __mlx4_clear_if_stat(struct mlx4_dev *dev,
+				u8 counter_index)
+{
+	struct mlx4_cmd_mailbox *if_stat_mailbox;
+	int err;
+	u32 if_stat_in_mod = (counter_index & 0xff) | MLX4_QUERY_IF_STAT_RESET;
+
+	if_stat_mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(if_stat_mailbox))
+		return PTR_ERR(if_stat_mailbox);
+
+	err = mlx4_cmd_box(dev, 0, if_stat_mailbox->dma, if_stat_in_mod, 0,
+			   MLX4_CMD_QUERY_IF_STAT, MLX4_CMD_TIME_CLASS_C,
+			   MLX4_CMD_NATIVE);
+
+	mlx4_free_cmd_mailbox(dev, if_stat_mailbox);
+	return err;
+}
+
+void __mlx4_counter_free(struct mlx4_dev *dev, u32 idx)
+{
+	if (!(dev->caps.flags & MLX4_DEV_CAP_FLAG_COUNTERS))
+		return;
+
+	if (idx == MLX4_SINK_COUNTER_INDEX(dev))
+		return;
+
+	__mlx4_clear_if_stat(dev, idx);
+
+	mlx4_bitmap_free(&mlx4_priv(dev)->counters_bitmap, idx, MLX4_USE_RR);
+	return;
+}
+
+void mlx4_counter_free(struct mlx4_dev *dev, u32 idx)
+{
+	u64 in_param = 0;
+
+	if (mlx4_is_mfunc(dev)) {
+		set_param_l(&in_param, idx);
+		mlx4_cmd(dev, in_param, RES_COUNTER, RES_OP_RESERVE,
+			 MLX4_CMD_FREE_RES, MLX4_CMD_TIME_CLASS_A,
+			 MLX4_CMD_WRAPPED);
+		return;
+	}
+	__mlx4_counter_free(dev, idx);
+}
+EXPORT_SYMBOL_GPL(mlx4_counter_free);
+
+int mlx4_get_default_counter_index(struct mlx4_dev *dev, int port)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	return priv->def_counter[port - 1];
+}
+EXPORT_SYMBOL_GPL(mlx4_get_default_counter_index);
+
+void mlx4_set_admin_guid(struct mlx4_dev *dev, __be64 guid, int entry, int port)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	priv->mfunc.master.vf_admin[entry].vport[port].guid = guid;
+}
+EXPORT_SYMBOL_GPL(mlx4_set_admin_guid);
+
+__be64 mlx4_get_admin_guid(struct mlx4_dev *dev, int entry, int port)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	return priv->mfunc.master.vf_admin[entry].vport[port].guid;
+}
+EXPORT_SYMBOL_GPL(mlx4_get_admin_guid);
+
+void mlx4_set_random_admin_guid(struct mlx4_dev *dev, int entry, int port)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	__be64 guid;
+
+	/* hw GUID */
+	if (entry == 0)
+		return;
+
+	get_random_bytes((char *)&guid, sizeof(guid));
+	guid &= ~(cpu_to_be64(1ULL << 56));
+	guid |= cpu_to_be64(1ULL << 57);
+	priv->mfunc.master.vf_admin[entry].vport[port].guid = guid;
+}
+
+static int mlx4_setup_hca(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int err;
+	int port;
+	__be32 ib_port_default_caps;
+
+	err = mlx4_init_uar_table(dev);
+	if (err) {
+		mlx4_err(dev, "Failed to initialize user access region table, aborting\n");
+		return err;
+	}
+
+	err = mlx4_uar_alloc(dev, &priv->driver_uar);
+	if (err) {
+		mlx4_err(dev, "Failed to allocate driver access region, aborting\n");
+		goto err_uar_table_free;
+	}
+
+	priv->kar = ioremap((phys_addr_t) priv->driver_uar.pfn << PAGE_SHIFT, PAGE_SIZE);
+	if (!priv->kar) {
+		mlx4_err(dev, "Couldn't map kernel access region, aborting\n");
+		err = -ENOMEM;
+		goto err_uar_free;
+	}
+
+	err = mlx4_init_pd_table(dev);
+	if (err) {
+		mlx4_err(dev, "Failed to initialize protection domain table, aborting\n");
+		goto err_kar_unmap;
+	}
+
+	err = mlx4_init_xrcd_table(dev);
+	if (err) {
+		mlx4_err(dev, "Failed to initialize reliable connection domain table, aborting\n");
+		goto err_pd_table_free;
+	}
+
+	err = mlx4_init_mr_table(dev);
+	if (err) {
+		mlx4_err(dev, "Failed to initialize memory region table, aborting\n");
+		goto err_xrcd_table_free;
+	}
+
+	if (!mlx4_is_slave(dev)) {
+		err = mlx4_init_mcg_table(dev);
+		if (err) {
+			mlx4_err(dev, "Failed to initialize multicast group table, aborting\n");
+			goto err_mr_table_free;
+		}
+		err = mlx4_config_mad_demux(dev);
+		if (err) {
+			mlx4_err(dev, "Failed in config_mad_demux, aborting\n");
+			goto err_mcg_table_free;
+		}
+	}
+
+	err = mlx4_init_eq_table(dev);
+	if (err) {
+		mlx4_err(dev, "Failed to initialize event queue table, aborting\n");
+		goto err_mcg_table_free;
+	}
+
+	err = mlx4_cmd_use_events(dev);
+	if (err) {
+		mlx4_err(dev, "Failed to switch to event-driven firmware commands, aborting\n");
+		goto err_eq_table_free;
+	}
+
+	err = mlx4_NOP(dev);
+	if (err) {
+		if (dev->flags & MLX4_FLAG_MSI_X) {
+			mlx4_warn(dev, "NOP command failed to generate MSI-X interrupt IRQ %d)\n",
+				  priv->eq_table.eq[MLX4_EQ_ASYNC].irq);
+			mlx4_warn(dev, "Trying again without MSI-X\n");
+		} else {
+			mlx4_err(dev, "NOP command failed to generate interrupt (IRQ %d), aborting\n",
+				 priv->eq_table.eq[MLX4_EQ_ASYNC].irq);
+			mlx4_err(dev, "BIOS or ACPI interrupt routing problem?\n");
+		}
+
+		goto err_cmd_poll;
+	}
+
+	mlx4_dbg(dev, "NOP command IRQ test passed\n");
+
+	err = mlx4_init_cq_table(dev);
+	if (err) {
+		mlx4_err(dev, "Failed to initialize completion queue table, aborting\n");
+		goto err_cmd_poll;
+	}
+
+	err = mlx4_init_srq_table(dev);
+	if (err) {
+		mlx4_err(dev, "Failed to initialize shared receive queue table, aborting\n");
+		goto err_cq_table_free;
+	}
+
+	err = mlx4_init_qp_table(dev);
+	if (err) {
+		mlx4_err(dev, "Failed to initialize queue pair table, aborting\n");
+		goto err_srq_table_free;
+	}
+
+	if (!mlx4_is_slave(dev)) {
+		err = mlx4_init_counters_table(dev);
+		if (err && err != -ENOENT) {
+			mlx4_err(dev, "Failed to initialize counters table, aborting\n");
+			goto err_qp_table_free;
+		}
+	}
+
+	err = mlx4_allocate_default_counters(dev);
+	if (err) {
+		mlx4_err(dev, "Failed to allocate default counters, aborting\n");
+		goto err_counters_table_free;
+	}
+
+	if (!mlx4_is_slave(dev)) {
+		for (port = 1; port <= dev->caps.num_ports; port++) {
+			ib_port_default_caps = 0;
+			err = mlx4_get_port_ib_caps(dev, port,
+						    &ib_port_default_caps);
+			if (err)
+				mlx4_warn(dev, "failed to get port %d default ib capabilities (%d). Continuing with caps = 0\n",
+					  port, err);
+			dev->caps.ib_port_def_cap[port] = ib_port_default_caps;
+
+			/* initialize per-slave default ib port capabilities */
+			if (mlx4_is_master(dev)) {
+				int i;
+				for (i = 0; i < dev->num_slaves; i++) {
+					if (i == mlx4_master_func_num(dev))
+						continue;
+					priv->mfunc.master.slave_state[i].ib_cap_mask[port] =
+						ib_port_default_caps;
+				}
+			}
+
+			if (mlx4_is_mfunc(dev))
+				dev->caps.port_ib_mtu[port] = IB_MTU_2048;
+			else
+				dev->caps.port_ib_mtu[port] = IB_MTU_4096;
+
+			err = mlx4_SET_PORT(dev, port, mlx4_is_master(dev) ?
+					    dev->caps.pkey_table_len[port] : -1);
+			if (err) {
+				mlx4_err(dev, "Failed to set port %d, aborting\n",
+					 port);
+				goto err_default_countes_free;
+			}
+		}
+	}
+
+	return 0;
+
+err_default_countes_free:
+	mlx4_cleanup_default_counters(dev);
+
+err_counters_table_free:
+	if (!mlx4_is_slave(dev))
+		mlx4_cleanup_counters_table(dev);
+
+err_qp_table_free:
+	mlx4_cleanup_qp_table(dev);
+
+err_srq_table_free:
+	mlx4_cleanup_srq_table(dev);
+
+err_cq_table_free:
+	mlx4_cleanup_cq_table(dev);
+
+err_cmd_poll:
+	mlx4_cmd_use_polling(dev);
+
+err_eq_table_free:
+	mlx4_cleanup_eq_table(dev);
+
+err_mcg_table_free:
+	if (!mlx4_is_slave(dev))
+		mlx4_cleanup_mcg_table(dev);
+
+err_mr_table_free:
+	mlx4_cleanup_mr_table(dev);
+
+err_xrcd_table_free:
+	mlx4_cleanup_xrcd_table(dev);
+
+err_pd_table_free:
+	mlx4_cleanup_pd_table(dev);
+
+err_kar_unmap:
+	iounmap(priv->kar);
+
+err_uar_free:
+	mlx4_uar_free(dev, &priv->driver_uar);
+
+err_uar_table_free:
+	mlx4_cleanup_uar_table(dev);
+	return err;
+}
+
+static int mlx4_init_affinity_hint(struct mlx4_dev *dev, int port, int eqn)
+{
+	int requested_cpu = 0;
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_eq *eq;
+	int off = 0;
+	int i;
+
+	if (eqn > dev->caps.num_comp_vectors)
+		return -EINVAL;
+
+	for (i = 1; i < port; i++)
+		off += mlx4_get_eqs_per_port(dev, i);
+
+	requested_cpu = eqn - off - !!(eqn > MLX4_EQ_ASYNC);
+
+	/* Meaning EQs are shared, and this call comes from the second port */
+	if (requested_cpu < 0)
+		return 0;
+
+	eq = &priv->eq_table.eq[eqn];
+
+	if (!zalloc_cpumask_var(&eq->affinity_mask, GFP_KERNEL))
+		return -ENOMEM;
+
+	cpumask_set_cpu(requested_cpu, eq->affinity_mask);
+
+	return 0;
+}
+
+static void mlx4_enable_msi_x(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct msix_entry *entries;
+	int i;
+	int port = 0;
+
+	if (msi_x) {
+		int nreq = min3(dev->caps.num_ports *
+				(int)num_online_cpus() + 1,
+				dev->caps.num_eqs - dev->caps.reserved_eqs,
+				MAX_MSIX);
+
+		if (msi_x > 1)
+			nreq = min_t(int, nreq, msi_x);
+
+		entries = kcalloc(nreq, sizeof(*entries), GFP_KERNEL);
+		if (!entries)
+			goto no_msi;
+
+		for (i = 0; i < nreq; ++i)
+			entries[i].entry = i;
+
+		nreq = pci_enable_msix_range(dev->persist->pdev, entries, 2,
+					     nreq);
+
+		if (nreq < 0 || nreq < MLX4_EQ_ASYNC) {
+			kfree(entries);
+			goto no_msi;
+		}
+		/* 1 is reserved for events (asyncrounous EQ) */
+		dev->caps.num_comp_vectors = nreq - 1;
+
+		priv->eq_table.eq[MLX4_EQ_ASYNC].irq = entries[0].vector;
+		bitmap_zero(priv->eq_table.eq[MLX4_EQ_ASYNC].actv_ports.ports,
+			    dev->caps.num_ports);
+
+		for (i = 0; i < dev->caps.num_comp_vectors + 1; i++) {
+			if (i == MLX4_EQ_ASYNC)
+				continue;
+
+			priv->eq_table.eq[i].irq =
+				entries[i + 1 - !!(i > MLX4_EQ_ASYNC)].vector;
+
+			if (MLX4_IS_LEGACY_EQ_MODE(dev->caps)) {
+				bitmap_fill(priv->eq_table.eq[i].actv_ports.ports,
+					    dev->caps.num_ports);
+				/* We don't set affinity hint when there
+				 * aren't enough EQs
+				 */
+			} else {
+				set_bit(port,
+					priv->eq_table.eq[i].actv_ports.ports);
+				if (mlx4_init_affinity_hint(dev, port + 1, i))
+					mlx4_warn(dev, "Couldn't init hint cpumask for EQ %d\n",
+						  i);
+			}
+			/* We divide the Eqs evenly between the two ports.
+			 * (dev->caps.num_comp_vectors / dev->caps.num_ports)
+			 * refers to the number of Eqs per port
+			 * (i.e eqs_per_port). Theoretically, we would like to
+			 * write something like (i + 1) % eqs_per_port == 0.
+			 * However, since there's an asynchronous Eq, we have
+			 * to skip over it by comparing this condition to
+			 * !!((i + 1) > MLX4_EQ_ASYNC).
+			 */
+			if ((dev->caps.num_comp_vectors > dev->caps.num_ports) &&
+			    ((i + 1) %
+			     (dev->caps.num_comp_vectors / dev->caps.num_ports)) ==
+			    !!((i + 1) > MLX4_EQ_ASYNC))
+				/* If dev->caps.num_comp_vectors < dev->caps.num_ports,
+				 * everything is shared anyway.
+				 */
+				port++;
+		}
+
+		dev->flags |= MLX4_FLAG_MSI_X;
+
+		kfree(entries);
+		return;
+	}
+
+no_msi:
+	dev->caps.num_comp_vectors = 1;
+
+	BUG_ON(MLX4_EQ_ASYNC >= 2);
+	for (i = 0; i < 2; ++i) {
+		priv->eq_table.eq[i].irq = dev->persist->pdev->irq;
+		if (i != MLX4_EQ_ASYNC) {
+			bitmap_fill(priv->eq_table.eq[i].actv_ports.ports,
+				    dev->caps.num_ports);
+		}
+	}
+}
+
+static int mlx4_init_port_info(struct mlx4_dev *dev, int port)
+{
+	struct devlink *devlink = priv_to_devlink(mlx4_priv(dev));
+	struct mlx4_port_info *info = &mlx4_priv(dev)->port[port];
+	int err;
+
+	err = devl_port_register(devlink, &info->devlink_port, port);
+	if (err)
+		return err;
+
+	/* Ethernet and IB drivers will normally set the port type,
+	 * but if they are not built set the type now to prevent
+	 * devlink_port_type_warn() from firing.
+	 */
+	if (!IS_ENABLED(CONFIG_MLX4_EN) &&
+	    dev->caps.port_type[port] == MLX4_PORT_TYPE_ETH)
+		devlink_port_type_eth_set(&info->devlink_port, NULL);
+	else if (!IS_ENABLED(CONFIG_MLX4_INFINIBAND) &&
+		 dev->caps.port_type[port] == MLX4_PORT_TYPE_IB)
+		devlink_port_type_ib_set(&info->devlink_port, NULL);
+
+	info->dev = dev;
+	info->port = port;
+	if (!mlx4_is_slave(dev)) {
+		mlx4_init_mac_table(dev, &info->mac_table);
+		mlx4_init_vlan_table(dev, &info->vlan_table);
+		mlx4_init_roce_gid_table(dev, &info->gid_table);
+		info->base_qpn = mlx4_get_base_qpn(dev, port);
+	}
+
+	sprintf(info->dev_name, "mlx4_port%d", port);
+	info->port_attr.attr.name = info->dev_name;
+	if (mlx4_is_mfunc(dev)) {
+		info->port_attr.attr.mode = 0444;
+	} else {
+		info->port_attr.attr.mode = 0644;
+		info->port_attr.store     = set_port_type;
+	}
+	info->port_attr.show      = show_port_type;
+	sysfs_attr_init(&info->port_attr.attr);
+
+	err = device_create_file(&dev->persist->pdev->dev, &info->port_attr);
+	if (err) {
+		mlx4_err(dev, "Failed to create file for port %d\n", port);
+		devlink_port_type_clear(&info->devlink_port);
+		devl_port_unregister(&info->devlink_port);
+		info->port = -1;
+		return err;
+	}
+
+	sprintf(info->dev_mtu_name, "mlx4_port%d_mtu", port);
+	info->port_mtu_attr.attr.name = info->dev_mtu_name;
+	if (mlx4_is_mfunc(dev)) {
+		info->port_mtu_attr.attr.mode = 0444;
+	} else {
+		info->port_mtu_attr.attr.mode = 0644;
+		info->port_mtu_attr.store     = set_port_ib_mtu;
+	}
+	info->port_mtu_attr.show      = show_port_ib_mtu;
+	sysfs_attr_init(&info->port_mtu_attr.attr);
+
+	err = device_create_file(&dev->persist->pdev->dev,
+				 &info->port_mtu_attr);
+	if (err) {
+		mlx4_err(dev, "Failed to create mtu file for port %d\n", port);
+		device_remove_file(&info->dev->persist->pdev->dev,
+				   &info->port_attr);
+		devlink_port_type_clear(&info->devlink_port);
+		devl_port_unregister(&info->devlink_port);
+		info->port = -1;
+		return err;
+	}
+
+	return 0;
+}
+
+static void mlx4_cleanup_port_info(struct mlx4_port_info *info)
+{
+	if (info->port < 0)
+		return;
+
+	device_remove_file(&info->dev->persist->pdev->dev, &info->port_attr);
+	device_remove_file(&info->dev->persist->pdev->dev,
+			   &info->port_mtu_attr);
+	devlink_port_type_clear(&info->devlink_port);
+	devl_port_unregister(&info->devlink_port);
+
+#ifdef CONFIG_RFS_ACCEL
+	free_irq_cpu_rmap(info->rmap);
+	info->rmap = NULL;
+#endif
+}
+
+static int mlx4_init_steering(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int num_entries = dev->caps.num_ports;
+	int i, j;
+
+	priv->steer = kcalloc(num_entries, sizeof(struct mlx4_steer),
+			      GFP_KERNEL);
+	if (!priv->steer)
+		return -ENOMEM;
+
+	for (i = 0; i < num_entries; i++)
+		for (j = 0; j < MLX4_NUM_STEERS; j++) {
+			INIT_LIST_HEAD(&priv->steer[i].promisc_qps[j]);
+			INIT_LIST_HEAD(&priv->steer[i].steer_entries[j]);
+		}
+	return 0;
+}
+
+static void mlx4_clear_steering(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_steer_index *entry, *tmp_entry;
+	struct mlx4_promisc_qp *pqp, *tmp_pqp;
+	int num_entries = dev->caps.num_ports;
+	int i, j;
+
+	for (i = 0; i < num_entries; i++) {
+		for (j = 0; j < MLX4_NUM_STEERS; j++) {
+			list_for_each_entry_safe(pqp, tmp_pqp,
+						 &priv->steer[i].promisc_qps[j],
+						 list) {
+				list_del(&pqp->list);
+				kfree(pqp);
+			}
+			list_for_each_entry_safe(entry, tmp_entry,
+						 &priv->steer[i].steer_entries[j],
+						 list) {
+				list_del(&entry->list);
+				list_for_each_entry_safe(pqp, tmp_pqp,
+							 &entry->duplicates,
+							 list) {
+					list_del(&pqp->list);
+					kfree(pqp);
+				}
+				kfree(entry);
+			}
+		}
+	}
+	kfree(priv->steer);
+}
+
+static int extended_func_num(struct pci_dev *pdev)
+{
+	return PCI_SLOT(pdev->devfn) * 8 + PCI_FUNC(pdev->devfn);
+}
+
+#define MLX4_OWNER_BASE	0x8069c
+#define MLX4_OWNER_SIZE	4
+
+static int mlx4_get_ownership(struct mlx4_dev *dev)
+{
+	void __iomem *owner;
+	u32 ret;
+
+	if (pci_channel_offline(dev->persist->pdev))
+		return -EIO;
+
+	owner = ioremap(pci_resource_start(dev->persist->pdev, 0) +
+			MLX4_OWNER_BASE,
+			MLX4_OWNER_SIZE);
+	if (!owner) {
+		mlx4_err(dev, "Failed to obtain ownership bit\n");
+		return -ENOMEM;
+	}
+
+	ret = readl(owner);
+	iounmap(owner);
+	return (int) !!ret;
+}
+
+static void mlx4_free_ownership(struct mlx4_dev *dev)
+{
+	void __iomem *owner;
+
+	if (pci_channel_offline(dev->persist->pdev))
+		return;
+
+	owner = ioremap(pci_resource_start(dev->persist->pdev, 0) +
+			MLX4_OWNER_BASE,
+			MLX4_OWNER_SIZE);
+	if (!owner) {
+		mlx4_err(dev, "Failed to obtain ownership bit\n");
+		return;
+	}
+	writel(0, owner);
+	msleep(1000);
+	iounmap(owner);
+}
+
+#define SRIOV_VALID_STATE(flags) (!!((flags) & MLX4_FLAG_SRIOV)	==\
+				  !!((flags) & MLX4_FLAG_MASTER))
+
+static u64 mlx4_enable_sriov(struct mlx4_dev *dev, struct pci_dev *pdev,
+			     u8 total_vfs, int existing_vfs, int reset_flow)
+{
+	u64 dev_flags = dev->flags;
+	int err = 0;
+	int fw_enabled_sriov_vfs = min(pci_sriov_get_totalvfs(pdev),
+					MLX4_MAX_NUM_VF);
+
+	if (reset_flow) {
+		dev->dev_vfs = kcalloc(total_vfs, sizeof(*dev->dev_vfs),
+				       GFP_KERNEL);
+		if (!dev->dev_vfs)
+			goto free_mem;
+		return dev_flags;
+	}
+
+	atomic_inc(&pf_loading);
+	if (dev->flags &  MLX4_FLAG_SRIOV) {
+		if (existing_vfs != total_vfs) {
+			mlx4_err(dev, "SR-IOV was already enabled, but with num_vfs (%d) different than requested (%d)\n",
+				 existing_vfs, total_vfs);
+			total_vfs = existing_vfs;
+		}
+	}
+
+	dev->dev_vfs = kcalloc(total_vfs, sizeof(*dev->dev_vfs), GFP_KERNEL);
+	if (NULL == dev->dev_vfs) {
+		mlx4_err(dev, "Failed to allocate memory for VFs\n");
+		goto disable_sriov;
+	}
+
+	if (!(dev->flags &  MLX4_FLAG_SRIOV)) {
+		if (total_vfs > fw_enabled_sriov_vfs) {
+			mlx4_err(dev, "requested vfs (%d) > available vfs (%d). Continuing without SR_IOV\n",
+				 total_vfs, fw_enabled_sriov_vfs);
+			err = -ENOMEM;
+			goto disable_sriov;
+		}
+		mlx4_warn(dev, "Enabling SR-IOV with %d VFs\n", total_vfs);
+		err = pci_enable_sriov(pdev, total_vfs);
+	}
+	if (err) {
+		mlx4_err(dev, "Failed to enable SR-IOV, continuing without SR-IOV (err = %d)\n",
+			 err);
+		goto disable_sriov;
+	} else {
+		mlx4_warn(dev, "Running in master mode\n");
+		dev_flags |= MLX4_FLAG_SRIOV |
+			MLX4_FLAG_MASTER;
+		dev_flags &= ~MLX4_FLAG_SLAVE;
+		dev->persist->num_vfs = total_vfs;
+	}
+	return dev_flags;
+
+disable_sriov:
+	atomic_dec(&pf_loading);
+free_mem:
+	dev->persist->num_vfs = 0;
+	kfree(dev->dev_vfs);
+	dev->dev_vfs = NULL;
+	return dev_flags & ~MLX4_FLAG_MASTER;
+}
+
+enum {
+	MLX4_DEV_CAP_CHECK_NUM_VFS_ABOVE_64 = -1,
+};
+
+static int mlx4_check_dev_cap(struct mlx4_dev *dev, struct mlx4_dev_cap *dev_cap,
+			      int *nvfs)
+{
+	int requested_vfs = nvfs[0] + nvfs[1] + nvfs[2];
+	/* Checking for 64 VFs as a limitation of CX2 */
+	if (!(dev_cap->flags2 & MLX4_DEV_CAP_FLAG2_80_VFS) &&
+	    requested_vfs >= 64) {
+		mlx4_err(dev, "Requested %d VFs, but FW does not support more than 64\n",
+			 requested_vfs);
+		return MLX4_DEV_CAP_CHECK_NUM_VFS_ABOVE_64;
+	}
+	return 0;
+}
+
+static int mlx4_pci_enable_device(struct mlx4_dev *dev)
+{
+	struct pci_dev *pdev = dev->persist->pdev;
+	int err = 0;
+
+	mutex_lock(&dev->persist->pci_status_mutex);
+	if (dev->persist->pci_status == MLX4_PCI_STATUS_DISABLED) {
+		err = pci_enable_device(pdev);
+		if (!err)
+			dev->persist->pci_status = MLX4_PCI_STATUS_ENABLED;
+	}
+	mutex_unlock(&dev->persist->pci_status_mutex);
+
+	return err;
+}
+
+static void mlx4_pci_disable_device(struct mlx4_dev *dev)
+{
+	struct pci_dev *pdev = dev->persist->pdev;
+
+	mutex_lock(&dev->persist->pci_status_mutex);
+	if (dev->persist->pci_status == MLX4_PCI_STATUS_ENABLED) {
+		pci_disable_device(pdev);
+		dev->persist->pci_status = MLX4_PCI_STATUS_DISABLED;
+	}
+	mutex_unlock(&dev->persist->pci_status_mutex);
+}
+
+static int mlx4_load_one(struct pci_dev *pdev, int pci_dev_data,
+			 int total_vfs, int *nvfs, struct mlx4_priv *priv,
+			 int reset_flow)
+{
+	struct devlink *devlink = priv_to_devlink(priv);
+	struct mlx4_dev *dev;
+	unsigned sum = 0;
+	int err;
+	int port;
+	int i;
+	struct mlx4_dev_cap *dev_cap = NULL;
+	int existing_vfs = 0;
+
+	devl_assert_locked(devlink);
+	dev = &priv->dev;
+
+	INIT_LIST_HEAD(&priv->ctx_list);
+	spin_lock_init(&priv->ctx_lock);
+
+	mutex_init(&priv->port_mutex);
+	mutex_init(&priv->bond_mutex);
+
+	INIT_LIST_HEAD(&priv->pgdir_list);
+	mutex_init(&priv->pgdir_mutex);
+	spin_lock_init(&priv->cmd.context_lock);
+
+	INIT_LIST_HEAD(&priv->bf_list);
+	mutex_init(&priv->bf_mutex);
+
+	dev->rev_id = pdev->revision;
+	dev->numa_node = dev_to_node(&pdev->dev);
+
+	/* Detect if this device is a virtual function */
+	if (pci_dev_data & MLX4_PCI_DEV_IS_VF) {
+		mlx4_warn(dev, "Detected virtual function - running in slave mode\n");
+		dev->flags |= MLX4_FLAG_SLAVE;
+	} else {
+		/* We reset the device and enable SRIOV only for physical
+		 * devices.  Try to claim ownership on the device;
+		 * if already taken, skip -- do not allow multiple PFs */
+		err = mlx4_get_ownership(dev);
+		if (err) {
+			if (err < 0)
+				return err;
+			else {
+				mlx4_warn(dev, "Multiple PFs not yet supported - Skipping PF\n");
+				return -EINVAL;
+			}
+		}
+
+		atomic_set(&priv->opreq_count, 0);
+		INIT_WORK(&priv->opreq_task, mlx4_opreq_action);
+
+		/*
+		 * Now reset the HCA before we touch the PCI capabilities or
+		 * attempt a firmware command, since a boot ROM may have left
+		 * the HCA in an undefined state.
+		 */
+		err = mlx4_reset(dev);
+		if (err) {
+			mlx4_err(dev, "Failed to reset HCA, aborting\n");
+			goto err_sriov;
+		}
+
+		if (total_vfs) {
+			dev->flags = MLX4_FLAG_MASTER;
+			existing_vfs = pci_num_vf(pdev);
+			if (existing_vfs)
+				dev->flags |= MLX4_FLAG_SRIOV;
+			dev->persist->num_vfs = total_vfs;
+		}
+	}
+
+	/* on load remove any previous indication of internal error,
+	 * device is up.
+	 */
+	dev->persist->state = MLX4_DEVICE_STATE_UP;
+
+slave_start:
+	err = mlx4_cmd_init(dev);
+	if (err) {
+		mlx4_err(dev, "Failed to init command interface, aborting\n");
+		goto err_sriov;
+	}
+
+	/* In slave functions, the communication channel must be initialized
+	 * before posting commands. Also, init num_slaves before calling
+	 * mlx4_init_hca */
+	if (mlx4_is_mfunc(dev)) {
+		if (mlx4_is_master(dev)) {
+			dev->num_slaves = MLX4_MAX_NUM_SLAVES;
+
+		} else {
+			dev->num_slaves = 0;
+			err = mlx4_multi_func_init(dev);
+			if (err) {
+				mlx4_err(dev, "Failed to init slave mfunc interface, aborting\n");
+				goto err_cmd;
+			}
+		}
+	}
+
+	err = mlx4_init_fw(dev);
+	if (err) {
+		mlx4_err(dev, "Failed to init fw, aborting.\n");
+		goto err_mfunc;
+	}
+
+	if (mlx4_is_master(dev)) {
+		/* when we hit the goto slave_start below, dev_cap already initialized */
+		if (!dev_cap) {
+			dev_cap = kzalloc(sizeof(*dev_cap), GFP_KERNEL);
+
+			if (!dev_cap) {
+				err = -ENOMEM;
+				goto err_fw;
+			}
+
+			err = mlx4_QUERY_DEV_CAP(dev, dev_cap);
+			if (err) {
+				mlx4_err(dev, "QUERY_DEV_CAP command failed, aborting.\n");
+				goto err_fw;
+			}
+
+			if (mlx4_check_dev_cap(dev, dev_cap, nvfs))
+				goto err_fw;
+
+			if (!(dev_cap->flags2 & MLX4_DEV_CAP_FLAG2_SYS_EQS)) {
+				u64 dev_flags = mlx4_enable_sriov(dev, pdev,
+								  total_vfs,
+								  existing_vfs,
+								  reset_flow);
+
+				mlx4_close_fw(dev);
+				mlx4_cmd_cleanup(dev, MLX4_CMD_CLEANUP_ALL);
+				dev->flags = dev_flags;
+				if (!SRIOV_VALID_STATE(dev->flags)) {
+					mlx4_err(dev, "Invalid SRIOV state\n");
+					goto err_sriov;
+				}
+				err = mlx4_reset(dev);
+				if (err) {
+					mlx4_err(dev, "Failed to reset HCA, aborting.\n");
+					goto err_sriov;
+				}
+				goto slave_start;
+			}
+		} else {
+			/* Legacy mode FW requires SRIOV to be enabled before
+			 * doing QUERY_DEV_CAP, since max_eq's value is different if
+			 * SRIOV is enabled.
+			 */
+			memset(dev_cap, 0, sizeof(*dev_cap));
+			err = mlx4_QUERY_DEV_CAP(dev, dev_cap);
+			if (err) {
+				mlx4_err(dev, "QUERY_DEV_CAP command failed, aborting.\n");
+				goto err_fw;
+			}
+
+			if (mlx4_check_dev_cap(dev, dev_cap, nvfs))
+				goto err_fw;
+		}
+	}
+
+	err = mlx4_init_hca(dev);
+	if (err) {
+		if (err == -EACCES) {
+			/* Not primary Physical function
+			 * Running in slave mode */
+			mlx4_cmd_cleanup(dev, MLX4_CMD_CLEANUP_ALL);
+			/* We're not a PF */
+			if (dev->flags & MLX4_FLAG_SRIOV) {
+				if (!existing_vfs)
+					pci_disable_sriov(pdev);
+				if (mlx4_is_master(dev) && !reset_flow)
+					atomic_dec(&pf_loading);
+				dev->flags &= ~MLX4_FLAG_SRIOV;
+			}
+			if (!mlx4_is_slave(dev))
+				mlx4_free_ownership(dev);
+			dev->flags |= MLX4_FLAG_SLAVE;
+			dev->flags &= ~MLX4_FLAG_MASTER;
+			goto slave_start;
+		} else
+			goto err_fw;
+	}
+
+	if (mlx4_is_master(dev) && (dev_cap->flags2 & MLX4_DEV_CAP_FLAG2_SYS_EQS)) {
+		u64 dev_flags = mlx4_enable_sriov(dev, pdev, total_vfs,
+						  existing_vfs, reset_flow);
+
+		if ((dev->flags ^ dev_flags) & (MLX4_FLAG_MASTER | MLX4_FLAG_SLAVE)) {
+			mlx4_cmd_cleanup(dev, MLX4_CMD_CLEANUP_VHCR);
+			dev->flags = dev_flags;
+			err = mlx4_cmd_init(dev);
+			if (err) {
+				/* Only VHCR is cleaned up, so could still
+				 * send FW commands
+				 */
+				mlx4_err(dev, "Failed to init VHCR command interface, aborting\n");
+				goto err_close;
+			}
+		} else {
+			dev->flags = dev_flags;
+		}
+
+		if (!SRIOV_VALID_STATE(dev->flags)) {
+			mlx4_err(dev, "Invalid SRIOV state\n");
+			err = -EINVAL;
+			goto err_close;
+		}
+	}
+
+	/* check if the device is functioning at its maximum possible speed.
+	 * No return code for this call, just warn the user in case of PCI
+	 * express device capabilities are under-satisfied by the bus.
+	 */
+	if (!mlx4_is_slave(dev))
+		pcie_print_link_status(dev->persist->pdev);
+
+	/* In master functions, the communication channel must be initialized
+	 * after obtaining its address from fw */
+	if (mlx4_is_master(dev)) {
+		if (dev->caps.num_ports < 2 &&
+		    num_vfs_argc > 1) {
+			err = -EINVAL;
+			mlx4_err(dev,
+				 "Error: Trying to configure VFs on port 2, but HCA has only %d physical ports\n",
+				 dev->caps.num_ports);
+			goto err_close;
+		}
+		memcpy(dev->persist->nvfs, nvfs, sizeof(dev->persist->nvfs));
+
+		for (i = 0;
+		     i < sizeof(dev->persist->nvfs)/
+		     sizeof(dev->persist->nvfs[0]); i++) {
+			unsigned j;
+
+			for (j = 0; j < dev->persist->nvfs[i]; ++sum, ++j) {
+				dev->dev_vfs[sum].min_port = i < 2 ? i + 1 : 1;
+				dev->dev_vfs[sum].n_ports = i < 2 ? 1 :
+					dev->caps.num_ports;
+			}
+		}
+
+		/* In master functions, the communication channel
+		 * must be initialized after obtaining its address from fw
+		 */
+		err = mlx4_multi_func_init(dev);
+		if (err) {
+			mlx4_err(dev, "Failed to init master mfunc interface, aborting.\n");
+			goto err_close;
+		}
+	}
+
+	err = mlx4_alloc_eq_table(dev);
+	if (err)
+		goto err_master_mfunc;
+
+	bitmap_zero(priv->msix_ctl.pool_bm, MAX_MSIX);
+	mutex_init(&priv->msix_ctl.pool_lock);
+
+	mlx4_enable_msi_x(dev);
+	if ((mlx4_is_mfunc(dev)) &&
+	    !(dev->flags & MLX4_FLAG_MSI_X)) {
+		err = -EOPNOTSUPP;
+		mlx4_err(dev, "INTx is not supported in multi-function mode, aborting\n");
+		goto err_free_eq;
+	}
+
+	if (!mlx4_is_slave(dev)) {
+		err = mlx4_init_steering(dev);
+		if (err)
+			goto err_disable_msix;
+	}
+
+	mlx4_init_quotas(dev);
+
+	err = mlx4_setup_hca(dev);
+	if (err == -EBUSY && (dev->flags & MLX4_FLAG_MSI_X) &&
+	    !mlx4_is_mfunc(dev)) {
+		dev->flags &= ~MLX4_FLAG_MSI_X;
+		dev->caps.num_comp_vectors = 1;
+		pci_disable_msix(pdev);
+		err = mlx4_setup_hca(dev);
+	}
+
+	if (err)
+		goto err_steer;
+
+	/* When PF resources are ready arm its comm channel to enable
+	 * getting commands
+	 */
+	if (mlx4_is_master(dev)) {
+		err = mlx4_ARM_COMM_CHANNEL(dev);
+		if (err) {
+			mlx4_err(dev, " Failed to arm comm channel eq: %x\n",
+				 err);
+			goto err_steer;
+		}
+	}
+
+	for (port = 1; port <= dev->caps.num_ports; port++) {
+		err = mlx4_init_port_info(dev, port);
+		if (err)
+			goto err_port;
+	}
+
+	priv->v2p.port1 = 1;
+	priv->v2p.port2 = 2;
+
+	err = mlx4_register_device(dev);
+	if (err)
+		goto err_port;
+
+	mlx4_request_modules(dev);
+
+	mlx4_sense_init(dev);
+	mlx4_start_sense(dev);
+
+	priv->removed = 0;
+
+	if (mlx4_is_master(dev) && dev->persist->num_vfs && !reset_flow)
+		atomic_dec(&pf_loading);
+
+	kfree(dev_cap);
+	return 0;
+
+err_port:
+	for (--port; port >= 1; --port)
+		mlx4_cleanup_port_info(&priv->port[port]);
+
+	mlx4_cleanup_default_counters(dev);
+	if (!mlx4_is_slave(dev))
+		mlx4_cleanup_counters_table(dev);
+	mlx4_cleanup_qp_table(dev);
+	mlx4_cleanup_srq_table(dev);
+	mlx4_cleanup_cq_table(dev);
+	mlx4_cmd_use_polling(dev);
+	mlx4_cleanup_eq_table(dev);
+	mlx4_cleanup_mcg_table(dev);
+	mlx4_cleanup_mr_table(dev);
+	mlx4_cleanup_xrcd_table(dev);
+	mlx4_cleanup_pd_table(dev);
+	mlx4_cleanup_uar_table(dev);
+
+err_steer:
+	if (!mlx4_is_slave(dev))
+		mlx4_clear_steering(dev);
+
+err_disable_msix:
+	if (dev->flags & MLX4_FLAG_MSI_X)
+		pci_disable_msix(pdev);
+
+err_free_eq:
+	mlx4_free_eq_table(dev);
+
+err_master_mfunc:
+	if (mlx4_is_master(dev)) {
+		mlx4_free_resource_tracker(dev, RES_TR_FREE_STRUCTS_ONLY);
+		mlx4_multi_func_cleanup(dev);
+	}
+
+	if (mlx4_is_slave(dev))
+		mlx4_slave_destroy_special_qp_cap(dev);
+
+err_close:
+	mlx4_close_hca(dev);
+
+err_fw:
+	mlx4_close_fw(dev);
+
+err_mfunc:
+	if (mlx4_is_slave(dev))
+		mlx4_multi_func_cleanup(dev);
+
+err_cmd:
+	mlx4_cmd_cleanup(dev, MLX4_CMD_CLEANUP_ALL);
+
+err_sriov:
+	if (dev->flags & MLX4_FLAG_SRIOV && !existing_vfs) {
+		pci_disable_sriov(pdev);
+		dev->flags &= ~MLX4_FLAG_SRIOV;
+	}
+
+	if (mlx4_is_master(dev) && dev->persist->num_vfs && !reset_flow)
+		atomic_dec(&pf_loading);
+
+	kfree(priv->dev.dev_vfs);
+
+	if (!mlx4_is_slave(dev))
+		mlx4_free_ownership(dev);
+
+	kfree(dev_cap);
+	return err;
+}
+
+static int __mlx4_init_one(struct pci_dev *pdev, int pci_dev_data,
+			   struct mlx4_priv *priv)
+{
+	int err;
+	int nvfs[MLX4_MAX_PORTS + 1] = {0, 0, 0};
+	int prb_vf[MLX4_MAX_PORTS + 1] = {0, 0, 0};
+	const int param_map[MLX4_MAX_PORTS + 1][MLX4_MAX_PORTS + 1] = {
+		{2, 0, 0}, {0, 1, 2}, {0, 1, 2} };
+	unsigned total_vfs = 0;
+	unsigned int i;
+
+	pr_info(DRV_NAME ": Initializing %s\n", pci_name(pdev));
+
+	err = mlx4_pci_enable_device(&priv->dev);
+	if (err) {
+		dev_err(&pdev->dev, "Cannot enable PCI device, aborting\n");
+		return err;
+	}
+
+	/* Due to requirement that all VFs and the PF are *guaranteed* 2 MACS
+	 * per port, we must limit the number of VFs to 63 (since their are
+	 * 128 MACs)
+	 */
+	for (i = 0; i < ARRAY_SIZE(nvfs) && i < num_vfs_argc;
+	     total_vfs += nvfs[param_map[num_vfs_argc - 1][i]], i++) {
+		nvfs[param_map[num_vfs_argc - 1][i]] = num_vfs[i];
+		if (nvfs[i] < 0) {
+			dev_err(&pdev->dev, "num_vfs module parameter cannot be negative\n");
+			err = -EINVAL;
+			goto err_disable_pdev;
+		}
+	}
+	for (i = 0; i < ARRAY_SIZE(prb_vf) && i < probe_vfs_argc;
+	     i++) {
+		prb_vf[param_map[probe_vfs_argc - 1][i]] = probe_vf[i];
+		if (prb_vf[i] < 0 || prb_vf[i] > nvfs[i]) {
+			dev_err(&pdev->dev, "probe_vf module parameter cannot be negative or greater than num_vfs\n");
+			err = -EINVAL;
+			goto err_disable_pdev;
+		}
+	}
+	if (total_vfs > MLX4_MAX_NUM_VF) {
+		dev_err(&pdev->dev,
+			"Requested more VF's (%d) than allowed by hw (%d)\n",
+			total_vfs, MLX4_MAX_NUM_VF);
+		err = -EINVAL;
+		goto err_disable_pdev;
+	}
+
+	for (i = 0; i < MLX4_MAX_PORTS; i++) {
+		if (nvfs[i] + nvfs[2] > MLX4_MAX_NUM_VF_P_PORT) {
+			dev_err(&pdev->dev,
+				"Requested more VF's (%d) for port (%d) than allowed by driver (%d)\n",
+				nvfs[i] + nvfs[2], i + 1,
+				MLX4_MAX_NUM_VF_P_PORT);
+			err = -EINVAL;
+			goto err_disable_pdev;
+		}
+	}
+
+	/* Check for BARs. */
+	if (!(pci_dev_data & MLX4_PCI_DEV_IS_VF) &&
+	    !(pci_resource_flags(pdev, 0) & IORESOURCE_MEM)) {
+		dev_err(&pdev->dev, "Missing DCS, aborting (driver_data: 0x%x, pci_resource_flags(pdev, 0):0x%lx)\n",
+			pci_dev_data, pci_resource_flags(pdev, 0));
+		err = -ENODEV;
+		goto err_disable_pdev;
+	}
+	if (!(pci_resource_flags(pdev, 2) & IORESOURCE_MEM)) {
+		dev_err(&pdev->dev, "Missing UAR, aborting\n");
+		err = -ENODEV;
+		goto err_disable_pdev;
+	}
+
+	err = pci_request_regions(pdev, DRV_NAME);
+	if (err) {
+		dev_err(&pdev->dev, "Couldn't get PCI resources, aborting\n");
+		goto err_disable_pdev;
+	}
+
+	pci_set_master(pdev);
+
+	err = dma_set_mask_and_coherent(&pdev->dev, DMA_BIT_MASK(64));
+	if (err) {
+		dev_warn(&pdev->dev, "Warning: couldn't set 64-bit PCI DMA mask\n");
+		err = dma_set_mask_and_coherent(&pdev->dev, DMA_BIT_MASK(32));
+		if (err) {
+			dev_err(&pdev->dev, "Can't set PCI DMA mask, aborting\n");
+			goto err_release_regions;
+		}
+	}
+
+	/* Allow large DMA segments, up to the firmware limit of 1 GB */
+	dma_set_max_seg_size(&pdev->dev, 1024 * 1024 * 1024);
+	/* Detect if this device is a virtual function */
+	if (pci_dev_data & MLX4_PCI_DEV_IS_VF) {
+		/* When acting as pf, we normally skip vfs unless explicitly
+		 * requested to probe them.
+		 */
+		if (total_vfs) {
+			unsigned vfs_offset = 0;
+
+			for (i = 0; i < ARRAY_SIZE(nvfs) &&
+			     vfs_offset + nvfs[i] < extended_func_num(pdev);
+			     vfs_offset += nvfs[i], i++)
+				;
+			if (i == ARRAY_SIZE(nvfs)) {
+				err = -ENODEV;
+				goto err_release_regions;
+			}
+			if ((extended_func_num(pdev) - vfs_offset)
+			    > prb_vf[i]) {
+				dev_warn(&pdev->dev, "Skipping virtual function:%d\n",
+					 extended_func_num(pdev));
+				err = -ENODEV;
+				goto err_release_regions;
+			}
+		}
+	}
+
+	err = mlx4_crdump_init(&priv->dev);
+	if (err)
+		goto err_release_regions;
+
+	err = mlx4_catas_init(&priv->dev);
+	if (err)
+		goto err_crdump;
+
+	err = mlx4_load_one(pdev, pci_dev_data, total_vfs, nvfs, priv, 0);
+	if (err)
+		goto err_catas;
+
+	return 0;
+
+err_catas:
+	mlx4_catas_end(&priv->dev);
+
+err_crdump:
+	mlx4_crdump_end(&priv->dev);
+
+err_release_regions:
+	pci_release_regions(pdev);
+
+err_disable_pdev:
+	mlx4_pci_disable_device(&priv->dev);
+	return err;
+}
+
+static int mlx4_devlink_port_type_set(struct devlink_port *devlink_port,
+				      enum devlink_port_type port_type)
+{
+	struct mlx4_port_info *info = container_of(devlink_port,
+						   struct mlx4_port_info,
+						   devlink_port);
+	enum mlx4_port_type mlx4_port_type;
+
+	switch (port_type) {
+	case DEVLINK_PORT_TYPE_AUTO:
+		mlx4_port_type = MLX4_PORT_TYPE_AUTO;
+		break;
+	case DEVLINK_PORT_TYPE_ETH:
+		mlx4_port_type = MLX4_PORT_TYPE_ETH;
+		break;
+	case DEVLINK_PORT_TYPE_IB:
+		mlx4_port_type = MLX4_PORT_TYPE_IB;
+		break;
+	default:
+		return -EOPNOTSUPP;
+	}
+
+	return __set_port_type(info, mlx4_port_type);
+}
+
+static void mlx4_devlink_param_load_driverinit_values(struct devlink *devlink)
+{
+	struct mlx4_priv *priv = devlink_priv(devlink);
+	struct mlx4_dev *dev = &priv->dev;
+	struct mlx4_fw_crdump *crdump = &dev->persist->crdump;
+	union devlink_param_value saved_value;
+	int err;
+
+	err = devlink_param_driverinit_value_get(devlink,
+						 DEVLINK_PARAM_GENERIC_ID_INT_ERR_RESET,
+						 &saved_value);
+	if (!err && mlx4_internal_err_reset != saved_value.vbool) {
+		mlx4_internal_err_reset = saved_value.vbool;
+		/* Notify on value changed on runtime configuration mode */
+		devlink_param_value_changed(devlink,
+					    DEVLINK_PARAM_GENERIC_ID_INT_ERR_RESET);
+	}
+	err = devlink_param_driverinit_value_get(devlink,
+						 DEVLINK_PARAM_GENERIC_ID_MAX_MACS,
+						 &saved_value);
+	if (!err)
+		log_num_mac = order_base_2(saved_value.vu32);
+	err = devlink_param_driverinit_value_get(devlink,
+						 MLX4_DEVLINK_PARAM_ID_ENABLE_64B_CQE_EQE,
+						 &saved_value);
+	if (!err)
+		enable_64b_cqe_eqe = saved_value.vbool;
+	err = devlink_param_driverinit_value_get(devlink,
+						 MLX4_DEVLINK_PARAM_ID_ENABLE_4K_UAR,
+						 &saved_value);
+	if (!err)
+		enable_4k_uar = saved_value.vbool;
+	err = devlink_param_driverinit_value_get(devlink,
+						 DEVLINK_PARAM_GENERIC_ID_REGION_SNAPSHOT,
+						 &saved_value);
+	if (!err && crdump->snapshot_enable != saved_value.vbool) {
+		crdump->snapshot_enable = saved_value.vbool;
+		devlink_param_value_changed(devlink,
+					    DEVLINK_PARAM_GENERIC_ID_REGION_SNAPSHOT);
+	}
+}
+
+static void mlx4_restart_one_down(struct pci_dev *pdev);
+static int mlx4_restart_one_up(struct pci_dev *pdev, bool reload,
+			       struct devlink *devlink);
+
+static int mlx4_devlink_reload_down(struct devlink *devlink, bool netns_change,
+				    enum devlink_reload_action action,
+				    enum devlink_reload_limit limit,
+				    struct netlink_ext_ack *extack)
+{
+	struct mlx4_priv *priv = devlink_priv(devlink);
+	struct mlx4_dev *dev = &priv->dev;
+	struct mlx4_dev_persistent *persist = dev->persist;
+
+	if (netns_change) {
+		NL_SET_ERR_MSG_MOD(extack, "Namespace change is not supported");
+		return -EOPNOTSUPP;
+	}
+	if (persist->num_vfs)
+		mlx4_warn(persist->dev, "Reload performed on PF, will cause reset on operating Virtual Functions\n");
+	mlx4_restart_one_down(persist->pdev);
+	return 0;
+}
+
+static int mlx4_devlink_reload_up(struct devlink *devlink, enum devlink_reload_action action,
+				  enum devlink_reload_limit limit, u32 *actions_performed,
+				  struct netlink_ext_ack *extack)
+{
+	struct mlx4_priv *priv = devlink_priv(devlink);
+	struct mlx4_dev *dev = &priv->dev;
+	struct mlx4_dev_persistent *persist = dev->persist;
+	int err;
+
+	*actions_performed = BIT(DEVLINK_RELOAD_ACTION_DRIVER_REINIT);
+	err = mlx4_restart_one_up(persist->pdev, true, devlink);
+	if (err)
+		mlx4_err(persist->dev, "mlx4_restart_one_up failed, ret=%d\n",
+			 err);
+
+	return err;
+}
+
+static const struct devlink_ops mlx4_devlink_ops = {
+	.port_type_set	= mlx4_devlink_port_type_set,
+	.reload_actions = BIT(DEVLINK_RELOAD_ACTION_DRIVER_REINIT),
+	.reload_down	= mlx4_devlink_reload_down,
+	.reload_up	= mlx4_devlink_reload_up,
+};
+
+static int mlx4_init_one(struct pci_dev *pdev, const struct pci_device_id *id)
+{
+	struct devlink *devlink;
+	struct mlx4_priv *priv;
+	struct mlx4_dev *dev;
+	int ret;
+
+	printk_once(KERN_INFO "%s", mlx4_version);
+
+	devlink = devlink_alloc(&mlx4_devlink_ops, sizeof(*priv), &pdev->dev);
+	if (!devlink)
+		return -ENOMEM;
+	devl_lock(devlink);
+	priv = devlink_priv(devlink);
+
+	dev       = &priv->dev;
+	dev->persist = kzalloc(sizeof(*dev->persist), GFP_KERNEL);
+	if (!dev->persist) {
+		ret = -ENOMEM;
+		goto err_devlink_free;
+	}
+	dev->persist->pdev = pdev;
+	dev->persist->dev = dev;
+	pci_set_drvdata(pdev, dev->persist);
+	priv->pci_dev_data = id->driver_data;
+	mutex_init(&dev->persist->device_state_mutex);
+	mutex_init(&dev->persist->interface_state_mutex);
+	mutex_init(&dev->persist->pci_status_mutex);
+
+	ret = devlink_params_register(devlink, mlx4_devlink_params,
+				      ARRAY_SIZE(mlx4_devlink_params));
+	if (ret)
+		goto err_devlink_unregister;
+	mlx4_devlink_set_params_init_values(devlink);
+	ret =  __mlx4_init_one(pdev, id->driver_data, priv);
+	if (ret)
+		goto err_params_unregister;
+
+	pci_save_state(pdev);
+	devlink_set_features(devlink, DEVLINK_F_RELOAD);
+	devl_unlock(devlink);
+	devlink_register(devlink);
+	return 0;
+
+err_params_unregister:
+	devlink_params_unregister(devlink, mlx4_devlink_params,
+				  ARRAY_SIZE(mlx4_devlink_params));
+err_devlink_unregister:
+	kfree(dev->persist);
+err_devlink_free:
+	devl_unlock(devlink);
+	devlink_free(devlink);
+	return ret;
+}
+
+static void mlx4_clean_dev(struct mlx4_dev *dev)
+{
+	struct mlx4_dev_persistent *persist = dev->persist;
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	unsigned long	flags = (dev->flags & RESET_PERSIST_MASK_FLAGS);
+
+	memset(priv, 0, sizeof(*priv));
+	priv->dev.persist = persist;
+	priv->dev.flags = flags;
+}
+
+static void mlx4_unload_one(struct pci_dev *pdev)
+{
+	struct mlx4_dev_persistent *persist = pci_get_drvdata(pdev);
+	struct mlx4_dev  *dev  = persist->dev;
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int               pci_dev_data;
+	struct devlink *devlink;
+	int p, i;
+
+	devlink = priv_to_devlink(priv);
+	devl_assert_locked(devlink);
+	if (priv->removed)
+		return;
+
+	/* saving current ports type for further use */
+	for (i = 0; i < dev->caps.num_ports; i++) {
+		dev->persist->curr_port_type[i] = dev->caps.port_type[i + 1];
+		dev->persist->curr_port_poss_type[i] = dev->caps.
+						       possible_type[i + 1];
+	}
+
+	pci_dev_data = priv->pci_dev_data;
+
+	mlx4_stop_sense(dev);
+	mlx4_unregister_device(dev);
+
+	for (p = 1; p <= dev->caps.num_ports; p++) {
+		mlx4_cleanup_port_info(&priv->port[p]);
+		mlx4_CLOSE_PORT(dev, p);
+	}
+
+	if (mlx4_is_master(dev))
+		mlx4_free_resource_tracker(dev,
+					   RES_TR_FREE_SLAVES_ONLY);
+
+	mlx4_cleanup_default_counters(dev);
+	if (!mlx4_is_slave(dev))
+		mlx4_cleanup_counters_table(dev);
+	mlx4_cleanup_qp_table(dev);
+	mlx4_cleanup_srq_table(dev);
+	mlx4_cleanup_cq_table(dev);
+	mlx4_cmd_use_polling(dev);
+	mlx4_cleanup_eq_table(dev);
+	mlx4_cleanup_mcg_table(dev);
+	mlx4_cleanup_mr_table(dev);
+	mlx4_cleanup_xrcd_table(dev);
+	mlx4_cleanup_pd_table(dev);
+
+	if (mlx4_is_master(dev))
+		mlx4_free_resource_tracker(dev,
+					   RES_TR_FREE_STRUCTS_ONLY);
+
+	iounmap(priv->kar);
+	mlx4_uar_free(dev, &priv->driver_uar);
+	mlx4_cleanup_uar_table(dev);
+	if (!mlx4_is_slave(dev))
+		mlx4_clear_steering(dev);
+	mlx4_free_eq_table(dev);
+	if (mlx4_is_master(dev))
+		mlx4_multi_func_cleanup(dev);
+	mlx4_close_hca(dev);
+	mlx4_close_fw(dev);
+	if (mlx4_is_slave(dev))
+		mlx4_multi_func_cleanup(dev);
+	mlx4_cmd_cleanup(dev, MLX4_CMD_CLEANUP_ALL);
+
+	if (dev->flags & MLX4_FLAG_MSI_X)
+		pci_disable_msix(pdev);
+
+	if (!mlx4_is_slave(dev))
+		mlx4_free_ownership(dev);
+
+	mlx4_slave_destroy_special_qp_cap(dev);
+	kfree(dev->dev_vfs);
+
+	mlx4_clean_dev(dev);
+	priv->pci_dev_data = pci_dev_data;
+	priv->removed = 1;
+}
+
+static void mlx4_remove_one(struct pci_dev *pdev)
+{
+	struct mlx4_dev_persistent *persist = pci_get_drvdata(pdev);
+	struct mlx4_dev  *dev  = persist->dev;
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct devlink *devlink = priv_to_devlink(priv);
+	int active_vfs = 0;
+
+	devlink_unregister(devlink);
+
+	devl_lock(devlink);
+	if (mlx4_is_slave(dev))
+		persist->interface_state |= MLX4_INTERFACE_STATE_NOWAIT;
+
+	mutex_lock(&persist->interface_state_mutex);
+	persist->interface_state |= MLX4_INTERFACE_STATE_DELETION;
+	mutex_unlock(&persist->interface_state_mutex);
+
+	/* Disabling SR-IOV is not allowed while there are active vf's */
+	if (mlx4_is_master(dev) && dev->flags & MLX4_FLAG_SRIOV) {
+		active_vfs = mlx4_how_many_lives_vf(dev);
+		if (active_vfs) {
+			pr_warn("Removing PF when there are active VF's !!\n");
+			pr_warn("Will not disable SR-IOV.\n");
+		}
+	}
+
+	/* device marked to be under deletion running now without the lock
+	 * letting other tasks to be terminated
+	 */
+	if (persist->interface_state & MLX4_INTERFACE_STATE_UP)
+		mlx4_unload_one(pdev);
+	else
+		mlx4_info(dev, "%s: interface is down\n", __func__);
+	mlx4_catas_end(dev);
+	mlx4_crdump_end(dev);
+	if (dev->flags & MLX4_FLAG_SRIOV && !active_vfs) {
+		mlx4_warn(dev, "Disabling SR-IOV\n");
+		pci_disable_sriov(pdev);
+	}
+
+	pci_release_regions(pdev);
+	mlx4_pci_disable_device(dev);
+	devlink_params_unregister(devlink, mlx4_devlink_params,
+				  ARRAY_SIZE(mlx4_devlink_params));
+	kfree(dev->persist);
+	devl_unlock(devlink);
+	devlink_free(devlink);
+}
+
+static int restore_current_port_types(struct mlx4_dev *dev,
+				      enum mlx4_port_type *types,
+				      enum mlx4_port_type *poss_types)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int err, i;
+
+	mlx4_stop_sense(dev);
+
+	mutex_lock(&priv->port_mutex);
+	for (i = 0; i < dev->caps.num_ports; i++)
+		dev->caps.possible_type[i + 1] = poss_types[i];
+	err = mlx4_change_port_types(dev, types);
+	mlx4_start_sense(dev);
+	mutex_unlock(&priv->port_mutex);
+
+	return err;
+}
+
+static void mlx4_restart_one_down(struct pci_dev *pdev)
+{
+	mlx4_unload_one(pdev);
+}
+
+static int mlx4_restart_one_up(struct pci_dev *pdev, bool reload,
+			       struct devlink *devlink)
+{
+	struct mlx4_dev_persistent *persist = pci_get_drvdata(pdev);
+	struct mlx4_dev	 *dev  = persist->dev;
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int nvfs[MLX4_MAX_PORTS + 1] = {0, 0, 0};
+	int pci_dev_data, err, total_vfs;
+
+	pci_dev_data = priv->pci_dev_data;
+	total_vfs = dev->persist->num_vfs;
+	memcpy(nvfs, dev->persist->nvfs, sizeof(dev->persist->nvfs));
+
+	if (reload)
+		mlx4_devlink_param_load_driverinit_values(devlink);
+	err = mlx4_load_one(pdev, pci_dev_data, total_vfs, nvfs, priv, 1);
+	if (err) {
+		mlx4_err(dev, "%s: ERROR: mlx4_load_one failed, pci_name=%s, err=%d\n",
+			 __func__, pci_name(pdev), err);
+		return err;
+	}
+
+	err = restore_current_port_types(dev, dev->persist->curr_port_type,
+					 dev->persist->curr_port_poss_type);
+	if (err)
+		mlx4_err(dev, "could not restore original port types (%d)\n",
+			 err);
+
+	return err;
+}
+
+int mlx4_restart_one(struct pci_dev *pdev)
+{
+	mlx4_restart_one_down(pdev);
+	return mlx4_restart_one_up(pdev, false, NULL);
+}
+
+#define MLX_SP(id) { PCI_VDEVICE(MELLANOX, id), MLX4_PCI_DEV_FORCE_SENSE_PORT }
+#define MLX_VF(id) { PCI_VDEVICE(MELLANOX, id), MLX4_PCI_DEV_IS_VF }
+#define MLX_GN(id) { PCI_VDEVICE(MELLANOX, id), 0 }
+
+static const struct pci_device_id mlx4_pci_table[] = {
+#ifdef CONFIG_MLX4_CORE_GEN2
+	/* MT25408 "Hermon" */
+	MLX_SP(PCI_DEVICE_ID_MELLANOX_HERMON_SDR),	/* SDR */
+	MLX_SP(PCI_DEVICE_ID_MELLANOX_HERMON_DDR),	/* DDR */
+	MLX_SP(PCI_DEVICE_ID_MELLANOX_HERMON_QDR),	/* QDR */
+	MLX_SP(PCI_DEVICE_ID_MELLANOX_HERMON_DDR_GEN2), /* DDR Gen2 */
+	MLX_SP(PCI_DEVICE_ID_MELLANOX_HERMON_QDR_GEN2),	/* QDR Gen2 */
+	MLX_SP(PCI_DEVICE_ID_MELLANOX_HERMON_EN),	/* EN 10GigE */
+	MLX_SP(PCI_DEVICE_ID_MELLANOX_HERMON_EN_GEN2),  /* EN 10GigE Gen2 */
+	/* MT25458 ConnectX EN 10GBASE-T */
+	MLX_SP(PCI_DEVICE_ID_MELLANOX_CONNECTX_EN),
+	MLX_SP(PCI_DEVICE_ID_MELLANOX_CONNECTX_EN_T_GEN2),	/* Gen2 */
+	/* MT26468 ConnectX EN 10GigE PCIe Gen2*/
+	MLX_SP(PCI_DEVICE_ID_MELLANOX_CONNECTX_EN_GEN2),
+	/* MT26438 ConnectX EN 40GigE PCIe Gen2 5GT/s */
+	MLX_SP(PCI_DEVICE_ID_MELLANOX_CONNECTX_EN_5_GEN2),
+	/* MT26478 ConnectX2 40GigE PCIe Gen2 */
+	MLX_SP(PCI_DEVICE_ID_MELLANOX_CONNECTX2),
+	/* MT25400 Family [ConnectX-2] */
+	MLX_VF(0x1002),					/* Virtual Function */
+#endif /* CONFIG_MLX4_CORE_GEN2 */
+	/* MT27500 Family [ConnectX-3] */
+	MLX_GN(PCI_DEVICE_ID_MELLANOX_CONNECTX3),
+	MLX_VF(0x1004),					/* Virtual Function */
+	MLX_GN(0x1005),					/* MT27510 Family */
+	MLX_GN(0x1006),					/* MT27511 Family */
+	MLX_GN(PCI_DEVICE_ID_MELLANOX_CONNECTX3_PRO),	/* MT27520 Family */
+	MLX_GN(0x1008),					/* MT27521 Family */
+	MLX_GN(0x1009),					/* MT27530 Family */
+	MLX_GN(0x100a),					/* MT27531 Family */
+	MLX_GN(0x100b),					/* MT27540 Family */
+	MLX_GN(0x100c),					/* MT27541 Family */
+	MLX_GN(0x100d),					/* MT27550 Family */
+	MLX_GN(0x100e),					/* MT27551 Family */
+	MLX_GN(0x100f),					/* MT27560 Family */
+	MLX_GN(0x1010),					/* MT27561 Family */
+
+	/*
+	 * See the mellanox_check_broken_intx_masking() quirk when
+	 * adding devices
+	 */
+
+	{ 0, }
+};
+
+MODULE_DEVICE_TABLE(pci, mlx4_pci_table);
+
+static pci_ers_result_t mlx4_pci_err_detected(struct pci_dev *pdev,
+					      pci_channel_state_t state)
+{
+	struct mlx4_dev_persistent *persist = pci_get_drvdata(pdev);
+	struct mlx4_dev *dev = persist->dev;
+	struct devlink *devlink;
+
+	mlx4_err(persist->dev, "mlx4_pci_err_detected was called\n");
+	mlx4_enter_error_state(persist);
+
+	devlink = priv_to_devlink(mlx4_priv(dev));
+	devl_lock(devlink);
+	mutex_lock(&persist->interface_state_mutex);
+	if (persist->interface_state & MLX4_INTERFACE_STATE_UP)
+		mlx4_unload_one(pdev);
+
+	mutex_unlock(&persist->interface_state_mutex);
+	devl_unlock(devlink);
+	if (state == pci_channel_io_perm_failure)
+		return PCI_ERS_RESULT_DISCONNECT;
+
+	mlx4_pci_disable_device(persist->dev);
+	return PCI_ERS_RESULT_NEED_RESET;
+}
+
+static pci_ers_result_t mlx4_pci_slot_reset(struct pci_dev *pdev)
+{
+	struct mlx4_dev_persistent *persist = pci_get_drvdata(pdev);
+	struct mlx4_dev	 *dev  = persist->dev;
+	int err;
+
+	mlx4_err(dev, "mlx4_pci_slot_reset was called\n");
+	err = mlx4_pci_enable_device(dev);
+	if (err) {
+		mlx4_err(dev, "Can not re-enable device, err=%d\n", err);
+		return PCI_ERS_RESULT_DISCONNECT;
+	}
+
+	pci_set_master(pdev);
+	pci_restore_state(pdev);
+	pci_save_state(pdev);
+	return PCI_ERS_RESULT_RECOVERED;
+}
+
+static void mlx4_pci_resume(struct pci_dev *pdev)
+{
+	struct mlx4_dev_persistent *persist = pci_get_drvdata(pdev);
+	struct mlx4_dev	 *dev  = persist->dev;
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int nvfs[MLX4_MAX_PORTS + 1] = {0, 0, 0};
+	struct devlink *devlink;
+	int total_vfs;
+	int err;
+
+	mlx4_err(dev, "%s was called\n", __func__);
+	total_vfs = dev->persist->num_vfs;
+	memcpy(nvfs, dev->persist->nvfs, sizeof(dev->persist->nvfs));
+
+	devlink = priv_to_devlink(priv);
+	devl_lock(devlink);
+	mutex_lock(&persist->interface_state_mutex);
+	if (!(persist->interface_state & MLX4_INTERFACE_STATE_UP)) {
+		err = mlx4_load_one(pdev, priv->pci_dev_data, total_vfs, nvfs,
+				    priv, 1);
+		if (err) {
+			mlx4_err(dev, "%s: mlx4_load_one failed, err=%d\n",
+				 __func__,  err);
+			goto end;
+		}
+
+		err = restore_current_port_types(dev, dev->persist->
+						 curr_port_type, dev->persist->
+						 curr_port_poss_type);
+		if (err)
+			mlx4_err(dev, "could not restore original port types (%d)\n", err);
+	}
+end:
+	mutex_unlock(&persist->interface_state_mutex);
+	devl_unlock(devlink);
+}
+
+static void mlx4_shutdown(struct pci_dev *pdev)
+{
+	struct mlx4_dev_persistent *persist = pci_get_drvdata(pdev);
+	struct mlx4_dev *dev = persist->dev;
+	struct devlink *devlink;
+
+	mlx4_info(persist->dev, "mlx4_shutdown was called\n");
+	devlink = priv_to_devlink(mlx4_priv(dev));
+	devl_lock(devlink);
+	mutex_lock(&persist->interface_state_mutex);
+	if (persist->interface_state & MLX4_INTERFACE_STATE_UP)
+		mlx4_unload_one(pdev);
+	mutex_unlock(&persist->interface_state_mutex);
+	devl_unlock(devlink);
+	mlx4_pci_disable_device(dev);
+}
+
+static const struct pci_error_handlers mlx4_err_handler = {
+	.error_detected = mlx4_pci_err_detected,
+	.slot_reset     = mlx4_pci_slot_reset,
+	.resume		= mlx4_pci_resume,
+};
+
+static int __maybe_unused mlx4_suspend(struct device *dev_d)
+{
+	struct pci_dev *pdev = to_pci_dev(dev_d);
+	struct mlx4_dev_persistent *persist = pci_get_drvdata(pdev);
+	struct mlx4_dev	*dev = persist->dev;
+	struct devlink *devlink;
+
+	mlx4_err(dev, "suspend was called\n");
+	devlink = priv_to_devlink(mlx4_priv(dev));
+	devl_lock(devlink);
+	mutex_lock(&persist->interface_state_mutex);
+	if (persist->interface_state & MLX4_INTERFACE_STATE_UP)
+		mlx4_unload_one(pdev);
+	mutex_unlock(&persist->interface_state_mutex);
+	devl_unlock(devlink);
+
+	return 0;
+}
+
+static int __maybe_unused mlx4_resume(struct device *dev_d)
+{
+	struct pci_dev *pdev = to_pci_dev(dev_d);
+	struct mlx4_dev_persistent *persist = pci_get_drvdata(pdev);
+	struct mlx4_dev	*dev = persist->dev;
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int nvfs[MLX4_MAX_PORTS + 1] = {0, 0, 0};
+	struct devlink *devlink;
+	int total_vfs;
+	int ret = 0;
+
+	mlx4_err(dev, "resume was called\n");
+	total_vfs = dev->persist->num_vfs;
+	memcpy(nvfs, dev->persist->nvfs, sizeof(dev->persist->nvfs));
+
+	devlink = priv_to_devlink(priv);
+	devl_lock(devlink);
+	mutex_lock(&persist->interface_state_mutex);
+	if (!(persist->interface_state & MLX4_INTERFACE_STATE_UP)) {
+		ret = mlx4_load_one(pdev, priv->pci_dev_data, total_vfs,
+				    nvfs, priv, 1);
+		if (!ret) {
+			ret = restore_current_port_types(dev,
+					dev->persist->curr_port_type,
+					dev->persist->curr_port_poss_type);
+			if (ret)
+				mlx4_err(dev, "resume: could not restore original port types (%d)\n", ret);
+		}
+	}
+	mutex_unlock(&persist->interface_state_mutex);
+	devl_unlock(devlink);
+
+	return ret;
+}
+
+static SIMPLE_DEV_PM_OPS(mlx4_pm_ops, mlx4_suspend, mlx4_resume);
+
+static struct pci_driver mlx4_driver = {
+	.name		= DRV_NAME,
+	.id_table	= mlx4_pci_table,
+	.probe		= mlx4_init_one,
+	.shutdown	= mlx4_shutdown,
+	.remove		= mlx4_remove_one,
+	.driver.pm	= &mlx4_pm_ops,
+	.err_handler    = &mlx4_err_handler,
+};
+
+static int __init mlx4_verify_params(void)
+{
+	if (msi_x < 0) {
+		pr_warn("mlx4_core: bad msi_x: %d\n", msi_x);
+		return -1;
+	}
+
+	if ((log_num_mac < 0) || (log_num_mac > 7)) {
+		pr_warn("mlx4_core: bad num_mac: %d\n", log_num_mac);
+		return -1;
+	}
+
+	if (log_num_vlan != 0)
+		pr_warn("mlx4_core: log_num_vlan - obsolete module param, using %d\n",
+			MLX4_LOG_NUM_VLANS);
+
+	if (use_prio != 0)
+		pr_warn("mlx4_core: use_prio - obsolete module param, ignored\n");
+
+	if ((log_mtts_per_seg < 0) || (log_mtts_per_seg > 7)) {
+		pr_warn("mlx4_core: bad log_mtts_per_seg: %d\n",
+			log_mtts_per_seg);
+		return -1;
+	}
+
+	/* Check if module param for ports type has legal combination */
+	if (port_type_array[0] == false && port_type_array[1] == true) {
+		pr_warn("Module parameter configuration ETH/IB is not supported. Switching to default configuration IB/IB\n");
+		port_type_array[0] = true;
+	}
+
+	if (mlx4_log_num_mgm_entry_size < -7 ||
+	    (mlx4_log_num_mgm_entry_size > 0 &&
+	     (mlx4_log_num_mgm_entry_size < MLX4_MIN_MGM_LOG_ENTRY_SIZE ||
+	      mlx4_log_num_mgm_entry_size > MLX4_MAX_MGM_LOG_ENTRY_SIZE))) {
+		pr_warn("mlx4_core: mlx4_log_num_mgm_entry_size (%d) not in legal range (-7..0 or %d..%d)\n",
+			mlx4_log_num_mgm_entry_size,
+			MLX4_MIN_MGM_LOG_ENTRY_SIZE,
+			MLX4_MAX_MGM_LOG_ENTRY_SIZE);
+		return -1;
+	}
+
+	return 0;
+}
+
+static int __init mlx4_init(void)
+{
+	int ret;
+
+	if (mlx4_verify_params())
+		return -EINVAL;
+
+
+	mlx4_wq = create_singlethread_workqueue("mlx4");
+	if (!mlx4_wq)
+		return -ENOMEM;
+
+	ret = pci_register_driver(&mlx4_driver);
+	if (ret < 0)
+		destroy_workqueue(mlx4_wq);
+	return ret < 0 ? ret : 0;
+}
+
+static void __exit mlx4_cleanup(void)
+{
+	pci_unregister_driver(&mlx4_driver);
+	destroy_workqueue(mlx4_wq);
+}
+
+module_init(mlx4_init);
+module_exit(mlx4_cleanup);
diff --git a/drivers/net/ethernet/mellanox/mlx4/mcg.c b/drivers/net/ethernet/mellanox/mlx4/mcg.c
new file mode 100644
index 000000000..f1716a83a
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx4/mcg.c
@@ -0,0 +1,1649 @@
+/*
+ * Copyright (c) 2006, 2007 Cisco Systems, Inc.  All rights reserved.
+ * Copyright (c) 2007, 2008 Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/string.h>
+#include <linux/etherdevice.h>
+
+#include <linux/mlx4/cmd.h>
+#include <linux/mlx4/qp.h>
+#include <linux/export.h>
+
+#include "mlx4.h"
+
+int mlx4_get_mgm_entry_size(struct mlx4_dev *dev)
+{
+	return 1 << dev->oper_log_mgm_entry_size;
+}
+
+int mlx4_get_qp_per_mgm(struct mlx4_dev *dev)
+{
+	return 4 * (mlx4_get_mgm_entry_size(dev) / 16 - 2);
+}
+
+static int mlx4_QP_FLOW_STEERING_ATTACH(struct mlx4_dev *dev,
+					struct mlx4_cmd_mailbox *mailbox,
+					u32 size,
+					u64 *reg_id)
+{
+	u64 imm;
+	int err = 0;
+
+	err = mlx4_cmd_imm(dev, mailbox->dma, &imm, size, 0,
+			   MLX4_QP_FLOW_STEERING_ATTACH, MLX4_CMD_TIME_CLASS_A,
+			   MLX4_CMD_NATIVE);
+	if (err)
+		return err;
+	*reg_id = imm;
+
+	return err;
+}
+
+static int mlx4_QP_FLOW_STEERING_DETACH(struct mlx4_dev *dev, u64 regid)
+{
+	int err = 0;
+
+	err = mlx4_cmd(dev, regid, 0, 0,
+		       MLX4_QP_FLOW_STEERING_DETACH, MLX4_CMD_TIME_CLASS_A,
+		       MLX4_CMD_NATIVE);
+
+	return err;
+}
+
+static int mlx4_READ_ENTRY(struct mlx4_dev *dev, int index,
+			   struct mlx4_cmd_mailbox *mailbox)
+{
+	return mlx4_cmd_box(dev, 0, mailbox->dma, index, 0, MLX4_CMD_READ_MCG,
+			    MLX4_CMD_TIME_CLASS_A, MLX4_CMD_NATIVE);
+}
+
+static int mlx4_WRITE_ENTRY(struct mlx4_dev *dev, int index,
+			    struct mlx4_cmd_mailbox *mailbox)
+{
+	return mlx4_cmd(dev, mailbox->dma, index, 0, MLX4_CMD_WRITE_MCG,
+			MLX4_CMD_TIME_CLASS_A, MLX4_CMD_NATIVE);
+}
+
+static int mlx4_WRITE_PROMISC(struct mlx4_dev *dev, u8 port, u8 steer,
+			      struct mlx4_cmd_mailbox *mailbox)
+{
+	u32 in_mod;
+
+	in_mod = (u32) port << 16 | steer << 1;
+	return mlx4_cmd(dev, mailbox->dma, in_mod, 0x1,
+			MLX4_CMD_WRITE_MCG, MLX4_CMD_TIME_CLASS_A,
+			MLX4_CMD_NATIVE);
+}
+
+static int mlx4_GID_HASH(struct mlx4_dev *dev, struct mlx4_cmd_mailbox *mailbox,
+			 u16 *hash, u8 op_mod)
+{
+	u64 imm;
+	int err;
+
+	err = mlx4_cmd_imm(dev, mailbox->dma, &imm, 0, op_mod,
+			   MLX4_CMD_MGID_HASH, MLX4_CMD_TIME_CLASS_A,
+			   MLX4_CMD_NATIVE);
+
+	if (!err)
+		*hash = imm;
+
+	return err;
+}
+
+static struct mlx4_promisc_qp *get_promisc_qp(struct mlx4_dev *dev, u8 port,
+					      enum mlx4_steer_type steer,
+					      u32 qpn)
+{
+	struct mlx4_steer *s_steer;
+	struct mlx4_promisc_qp *pqp;
+
+	if (port < 1 || port > dev->caps.num_ports)
+		return NULL;
+
+	s_steer = &mlx4_priv(dev)->steer[port - 1];
+
+	list_for_each_entry(pqp, &s_steer->promisc_qps[steer], list) {
+		if (pqp->qpn == qpn)
+			return pqp;
+	}
+	/* not found */
+	return NULL;
+}
+
+/*
+ * Add new entry to steering data structure.
+ * All promisc QPs should be added as well
+ */
+static int new_steering_entry(struct mlx4_dev *dev, u8 port,
+			      enum mlx4_steer_type steer,
+			      unsigned int index, u32 qpn)
+{
+	struct mlx4_steer *s_steer;
+	struct mlx4_cmd_mailbox *mailbox;
+	struct mlx4_mgm *mgm;
+	u32 members_count;
+	struct mlx4_steer_index *new_entry;
+	struct mlx4_promisc_qp *pqp;
+	struct mlx4_promisc_qp *dqp = NULL;
+	u32 prot;
+	int err;
+
+	if (port < 1 || port > dev->caps.num_ports)
+		return -EINVAL;
+
+	s_steer = &mlx4_priv(dev)->steer[port - 1];
+	new_entry = kzalloc(sizeof(*new_entry), GFP_KERNEL);
+	if (!new_entry)
+		return -ENOMEM;
+
+	INIT_LIST_HEAD(&new_entry->duplicates);
+	new_entry->index = index;
+	list_add_tail(&new_entry->list, &s_steer->steer_entries[steer]);
+
+	/* If the given qpn is also a promisc qp,
+	 * it should be inserted to duplicates list
+	 */
+	pqp = get_promisc_qp(dev, port, steer, qpn);
+	if (pqp) {
+		dqp = kmalloc(sizeof(*dqp), GFP_KERNEL);
+		if (!dqp) {
+			err = -ENOMEM;
+			goto out_alloc;
+		}
+		dqp->qpn = qpn;
+		list_add_tail(&dqp->list, &new_entry->duplicates);
+	}
+
+	/* if no promisc qps for this vep, we are done */
+	if (list_empty(&s_steer->promisc_qps[steer]))
+		return 0;
+
+	/* now need to add all the promisc qps to the new
+	 * steering entry, as they should also receive the packets
+	 * destined to this address */
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox)) {
+		err = -ENOMEM;
+		goto out_alloc;
+	}
+	mgm = mailbox->buf;
+
+	err = mlx4_READ_ENTRY(dev, index, mailbox);
+	if (err)
+		goto out_mailbox;
+
+	members_count = be32_to_cpu(mgm->members_count) & 0xffffff;
+	prot = be32_to_cpu(mgm->members_count) >> 30;
+	list_for_each_entry(pqp, &s_steer->promisc_qps[steer], list) {
+		/* don't add already existing qpn */
+		if (pqp->qpn == qpn)
+			continue;
+		if (members_count == dev->caps.num_qp_per_mgm) {
+			/* out of space */
+			err = -ENOMEM;
+			goto out_mailbox;
+		}
+
+		/* add the qpn */
+		mgm->qp[members_count++] = cpu_to_be32(pqp->qpn & MGM_QPN_MASK);
+	}
+	/* update the qps count and update the entry with all the promisc qps*/
+	mgm->members_count = cpu_to_be32(members_count | (prot << 30));
+	err = mlx4_WRITE_ENTRY(dev, index, mailbox);
+
+out_mailbox:
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	if (!err)
+		return 0;
+out_alloc:
+	if (dqp) {
+		list_del(&dqp->list);
+		kfree(dqp);
+	}
+	list_del(&new_entry->list);
+	kfree(new_entry);
+	return err;
+}
+
+/* update the data structures with existing steering entry */
+static int existing_steering_entry(struct mlx4_dev *dev, u8 port,
+				   enum mlx4_steer_type steer,
+				   unsigned int index, u32 qpn)
+{
+	struct mlx4_steer *s_steer;
+	struct mlx4_steer_index *tmp_entry, *entry = NULL;
+	struct mlx4_promisc_qp *pqp;
+	struct mlx4_promisc_qp *dqp;
+
+	if (port < 1 || port > dev->caps.num_ports)
+		return -EINVAL;
+
+	s_steer = &mlx4_priv(dev)->steer[port - 1];
+
+	pqp = get_promisc_qp(dev, port, steer, qpn);
+	if (!pqp)
+		return 0; /* nothing to do */
+
+	list_for_each_entry(tmp_entry, &s_steer->steer_entries[steer], list) {
+		if (tmp_entry->index == index) {
+			entry = tmp_entry;
+			break;
+		}
+	}
+	if (unlikely(!entry)) {
+		mlx4_warn(dev, "Steering entry at index %x is not registered\n", index);
+		return -EINVAL;
+	}
+
+	/* the given qpn is listed as a promisc qpn
+	 * we need to add it as a duplicate to this entry
+	 * for future references */
+	list_for_each_entry(dqp, &entry->duplicates, list) {
+		if (qpn == dqp->qpn)
+			return 0; /* qp is already duplicated */
+	}
+
+	/* add the qp as a duplicate on this index */
+	dqp = kmalloc(sizeof(*dqp), GFP_KERNEL);
+	if (!dqp)
+		return -ENOMEM;
+	dqp->qpn = qpn;
+	list_add_tail(&dqp->list, &entry->duplicates);
+
+	return 0;
+}
+
+/* Check whether a qpn is a duplicate on steering entry
+ * If so, it should not be removed from mgm */
+static bool check_duplicate_entry(struct mlx4_dev *dev, u8 port,
+				  enum mlx4_steer_type steer,
+				  unsigned int index, u32 qpn)
+{
+	struct mlx4_steer *s_steer;
+	struct mlx4_steer_index *tmp_entry, *entry = NULL;
+	struct mlx4_promisc_qp *dqp, *tmp_dqp;
+
+	if (port < 1 || port > dev->caps.num_ports)
+		return NULL;
+
+	s_steer = &mlx4_priv(dev)->steer[port - 1];
+
+	/* if qp is not promisc, it cannot be duplicated */
+	if (!get_promisc_qp(dev, port, steer, qpn))
+		return false;
+
+	/* The qp is promisc qp so it is a duplicate on this index
+	 * Find the index entry, and remove the duplicate */
+	list_for_each_entry(tmp_entry, &s_steer->steer_entries[steer], list) {
+		if (tmp_entry->index == index) {
+			entry = tmp_entry;
+			break;
+		}
+	}
+	if (unlikely(!entry)) {
+		mlx4_warn(dev, "Steering entry for index %x is not registered\n", index);
+		return false;
+	}
+	list_for_each_entry_safe(dqp, tmp_dqp, &entry->duplicates, list) {
+		if (dqp->qpn == qpn) {
+			list_del(&dqp->list);
+			kfree(dqp);
+		}
+	}
+	return true;
+}
+
+/* Returns true if all the QPs != tqpn contained in this entry
+ * are Promisc QPs. Returns false otherwise.
+ */
+static bool promisc_steering_entry(struct mlx4_dev *dev, u8 port,
+				   enum mlx4_steer_type steer,
+				   unsigned int index, u32 tqpn,
+				   u32 *members_count)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	struct mlx4_mgm *mgm;
+	u32 m_count;
+	bool ret = false;
+	int i;
+
+	if (port < 1 || port > dev->caps.num_ports)
+		return false;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return false;
+	mgm = mailbox->buf;
+
+	if (mlx4_READ_ENTRY(dev, index, mailbox))
+		goto out;
+	m_count = be32_to_cpu(mgm->members_count) & 0xffffff;
+	if (members_count)
+		*members_count = m_count;
+
+	for (i = 0;  i < m_count; i++) {
+		u32 qpn = be32_to_cpu(mgm->qp[i]) & MGM_QPN_MASK;
+		if (!get_promisc_qp(dev, port, steer, qpn) && qpn != tqpn) {
+			/* the qp is not promisc, the entry can't be removed */
+			goto out;
+		}
+	}
+	ret = true;
+out:
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return ret;
+}
+
+/* IF a steering entry contains only promisc QPs, it can be removed. */
+static bool can_remove_steering_entry(struct mlx4_dev *dev, u8 port,
+				      enum mlx4_steer_type steer,
+				      unsigned int index, u32 tqpn)
+{
+	struct mlx4_steer *s_steer;
+	struct mlx4_steer_index *entry = NULL, *tmp_entry;
+	u32 members_count;
+	bool ret = false;
+
+	if (port < 1 || port > dev->caps.num_ports)
+		return NULL;
+
+	s_steer = &mlx4_priv(dev)->steer[port - 1];
+
+	if (!promisc_steering_entry(dev, port, steer, index,
+				    tqpn, &members_count))
+		goto out;
+
+	/* All the qps currently registered for this entry are promiscuous,
+	  * Checking for duplicates */
+	ret = true;
+	list_for_each_entry_safe(entry, tmp_entry, &s_steer->steer_entries[steer], list) {
+		if (entry->index == index) {
+			if (list_empty(&entry->duplicates) ||
+			    members_count == 1) {
+				struct mlx4_promisc_qp *pqp, *tmp_pqp;
+				/* If there is only 1 entry in duplicates then
+				 * this is the QP we want to delete, going over
+				 * the list and deleting the entry.
+				 */
+				list_del(&entry->list);
+				list_for_each_entry_safe(pqp, tmp_pqp,
+							 &entry->duplicates,
+							 list) {
+					list_del(&pqp->list);
+					kfree(pqp);
+				}
+				kfree(entry);
+			} else {
+				/* This entry contains duplicates so it shouldn't be removed */
+				ret = false;
+				goto out;
+			}
+		}
+	}
+
+out:
+	return ret;
+}
+
+static int add_promisc_qp(struct mlx4_dev *dev, u8 port,
+			  enum mlx4_steer_type steer, u32 qpn)
+{
+	struct mlx4_steer *s_steer;
+	struct mlx4_cmd_mailbox *mailbox;
+	struct mlx4_mgm *mgm;
+	struct mlx4_steer_index *entry;
+	struct mlx4_promisc_qp *pqp;
+	struct mlx4_promisc_qp *dqp;
+	u32 members_count;
+	u32 prot;
+	int i;
+	bool found;
+	int err;
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	if (port < 1 || port > dev->caps.num_ports)
+		return -EINVAL;
+
+	s_steer = &mlx4_priv(dev)->steer[port - 1];
+
+	mutex_lock(&priv->mcg_table.mutex);
+
+	if (get_promisc_qp(dev, port, steer, qpn)) {
+		err = 0;  /* Noting to do, already exists */
+		goto out_mutex;
+	}
+
+	pqp = kmalloc(sizeof(*pqp), GFP_KERNEL);
+	if (!pqp) {
+		err = -ENOMEM;
+		goto out_mutex;
+	}
+	pqp->qpn = qpn;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox)) {
+		err = -ENOMEM;
+		goto out_alloc;
+	}
+	mgm = mailbox->buf;
+
+	if (!(mlx4_is_mfunc(dev) && steer == MLX4_UC_STEER)) {
+		/* The promisc QP needs to be added for each one of the steering
+		 * entries. If it already exists, needs to be added as
+		 * a duplicate for this entry.
+		 */
+		list_for_each_entry(entry,
+				    &s_steer->steer_entries[steer],
+				    list) {
+			err = mlx4_READ_ENTRY(dev, entry->index, mailbox);
+			if (err)
+				goto out_mailbox;
+
+			members_count = be32_to_cpu(mgm->members_count) &
+					0xffffff;
+			prot = be32_to_cpu(mgm->members_count) >> 30;
+			found = false;
+			for (i = 0; i < members_count; i++) {
+				if ((be32_to_cpu(mgm->qp[i]) &
+				     MGM_QPN_MASK) == qpn) {
+					/* Entry already exists.
+					 * Add to duplicates.
+					 */
+					dqp = kmalloc(sizeof(*dqp), GFP_KERNEL);
+					if (!dqp) {
+						err = -ENOMEM;
+						goto out_mailbox;
+					}
+					dqp->qpn = qpn;
+					list_add_tail(&dqp->list,
+						      &entry->duplicates);
+					found = true;
+				}
+			}
+			if (!found) {
+				/* Need to add the qpn to mgm */
+				if (members_count ==
+				    dev->caps.num_qp_per_mgm) {
+					/* entry is full */
+					err = -ENOMEM;
+					goto out_mailbox;
+				}
+				mgm->qp[members_count++] =
+					cpu_to_be32(qpn & MGM_QPN_MASK);
+				mgm->members_count =
+					cpu_to_be32(members_count |
+						    (prot << 30));
+				err = mlx4_WRITE_ENTRY(dev, entry->index,
+						       mailbox);
+				if (err)
+					goto out_mailbox;
+			}
+		}
+	}
+
+	/* add the new qpn to list of promisc qps */
+	list_add_tail(&pqp->list, &s_steer->promisc_qps[steer]);
+	/* now need to add all the promisc qps to default entry */
+	memset(mgm, 0, sizeof(*mgm));
+	members_count = 0;
+	list_for_each_entry(dqp, &s_steer->promisc_qps[steer], list) {
+		if (members_count == dev->caps.num_qp_per_mgm) {
+			/* entry is full */
+			err = -ENOMEM;
+			goto out_list;
+		}
+		mgm->qp[members_count++] = cpu_to_be32(dqp->qpn & MGM_QPN_MASK);
+	}
+	mgm->members_count = cpu_to_be32(members_count | MLX4_PROT_ETH << 30);
+
+	err = mlx4_WRITE_PROMISC(dev, port, steer, mailbox);
+	if (err)
+		goto out_list;
+
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	mutex_unlock(&priv->mcg_table.mutex);
+	return 0;
+
+out_list:
+	list_del(&pqp->list);
+out_mailbox:
+	mlx4_free_cmd_mailbox(dev, mailbox);
+out_alloc:
+	kfree(pqp);
+out_mutex:
+	mutex_unlock(&priv->mcg_table.mutex);
+	return err;
+}
+
+static int remove_promisc_qp(struct mlx4_dev *dev, u8 port,
+			     enum mlx4_steer_type steer, u32 qpn)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_steer *s_steer;
+	struct mlx4_cmd_mailbox *mailbox;
+	struct mlx4_mgm *mgm;
+	struct mlx4_steer_index *entry, *tmp_entry;
+	struct mlx4_promisc_qp *pqp;
+	struct mlx4_promisc_qp *dqp;
+	u32 members_count;
+	bool found;
+	bool back_to_list = false;
+	int i;
+	int err;
+
+	if (port < 1 || port > dev->caps.num_ports)
+		return -EINVAL;
+
+	s_steer = &mlx4_priv(dev)->steer[port - 1];
+	mutex_lock(&priv->mcg_table.mutex);
+
+	pqp = get_promisc_qp(dev, port, steer, qpn);
+	if (unlikely(!pqp)) {
+		mlx4_warn(dev, "QP %x is not promiscuous QP\n", qpn);
+		/* nothing to do */
+		err = 0;
+		goto out_mutex;
+	}
+
+	/*remove from list of promisc qps */
+	list_del(&pqp->list);
+
+	/* set the default entry not to include the removed one */
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox)) {
+		err = -ENOMEM;
+		back_to_list = true;
+		goto out_list;
+	}
+	mgm = mailbox->buf;
+	members_count = 0;
+	list_for_each_entry(dqp, &s_steer->promisc_qps[steer], list)
+		mgm->qp[members_count++] = cpu_to_be32(dqp->qpn & MGM_QPN_MASK);
+	mgm->members_count = cpu_to_be32(members_count | MLX4_PROT_ETH << 30);
+
+	err = mlx4_WRITE_PROMISC(dev, port, steer, mailbox);
+	if (err)
+		goto out_mailbox;
+
+	if (!(mlx4_is_mfunc(dev) && steer == MLX4_UC_STEER)) {
+		/* Remove the QP from all the steering entries */
+		list_for_each_entry_safe(entry, tmp_entry,
+					 &s_steer->steer_entries[steer],
+					 list) {
+			found = false;
+			list_for_each_entry(dqp, &entry->duplicates, list) {
+				if (dqp->qpn == qpn) {
+					found = true;
+					break;
+				}
+			}
+			if (found) {
+				/* A duplicate, no need to change the MGM,
+				 * only update the duplicates list
+				 */
+				list_del(&dqp->list);
+				kfree(dqp);
+			} else {
+				int loc = -1;
+
+				err = mlx4_READ_ENTRY(dev,
+						      entry->index,
+						      mailbox);
+				if (err)
+					goto out_mailbox;
+				members_count =
+					be32_to_cpu(mgm->members_count) &
+					0xffffff;
+				if (!members_count) {
+					mlx4_warn(dev, "QP %06x wasn't found in entry %x mcount=0. deleting entry...\n",
+						  qpn, entry->index);
+					list_del(&entry->list);
+					kfree(entry);
+					continue;
+				}
+
+				for (i = 0; i < members_count; ++i)
+					if ((be32_to_cpu(mgm->qp[i]) &
+					     MGM_QPN_MASK) == qpn) {
+						loc = i;
+						break;
+					}
+
+				if (loc < 0) {
+					mlx4_err(dev, "QP %06x wasn't found in entry %d\n",
+						 qpn, entry->index);
+					err = -EINVAL;
+					goto out_mailbox;
+				}
+
+				/* Copy the last QP in this MGM
+				 * over removed QP
+				 */
+				mgm->qp[loc] = mgm->qp[members_count - 1];
+				mgm->qp[members_count - 1] = 0;
+				mgm->members_count =
+					cpu_to_be32(--members_count |
+						    (MLX4_PROT_ETH << 30));
+
+				err = mlx4_WRITE_ENTRY(dev,
+						       entry->index,
+						       mailbox);
+				if (err)
+					goto out_mailbox;
+			}
+		}
+	}
+
+out_mailbox:
+	mlx4_free_cmd_mailbox(dev, mailbox);
+out_list:
+	if (back_to_list)
+		list_add_tail(&pqp->list, &s_steer->promisc_qps[steer]);
+	else
+		kfree(pqp);
+out_mutex:
+	mutex_unlock(&priv->mcg_table.mutex);
+	return err;
+}
+
+/*
+ * Caller must hold MCG table semaphore.  gid and mgm parameters must
+ * be properly aligned for command interface.
+ *
+ *  Returns 0 unless a firmware command error occurs.
+ *
+ * If GID is found in MGM or MGM is empty, *index = *hash, *prev = -1
+ * and *mgm holds MGM entry.
+ *
+ * if GID is found in AMGM, *index = index in AMGM, *prev = index of
+ * previous entry in hash chain and *mgm holds AMGM entry.
+ *
+ * If no AMGM exists for given gid, *index = -1, *prev = index of last
+ * entry in hash chain and *mgm holds end of hash chain.
+ */
+static int find_entry(struct mlx4_dev *dev, u8 port,
+		      u8 *gid, enum mlx4_protocol prot,
+		      struct mlx4_cmd_mailbox *mgm_mailbox,
+		      int *prev, int *index)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	struct mlx4_mgm *mgm = mgm_mailbox->buf;
+	u8 *mgid;
+	int err;
+	u16 hash;
+	u8 op_mod = (prot == MLX4_PROT_ETH) ?
+		!!(dev->caps.flags & MLX4_DEV_CAP_FLAG_VEP_MC_STEER) : 0;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return -ENOMEM;
+	mgid = mailbox->buf;
+
+	memcpy(mgid, gid, 16);
+
+	err = mlx4_GID_HASH(dev, mailbox, &hash, op_mod);
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	if (err)
+		return err;
+
+	if (0)
+		mlx4_dbg(dev, "Hash for %pI6 is %04x\n", gid, hash);
+
+	*index = hash;
+	*prev  = -1;
+
+	do {
+		err = mlx4_READ_ENTRY(dev, *index, mgm_mailbox);
+		if (err)
+			return err;
+
+		if (!(be32_to_cpu(mgm->members_count) & 0xffffff)) {
+			if (*index != hash) {
+				mlx4_err(dev, "Found zero MGID in AMGM\n");
+				err = -EINVAL;
+			}
+			return err;
+		}
+
+		if (!memcmp(mgm->gid, gid, 16) &&
+		    be32_to_cpu(mgm->members_count) >> 30 == prot)
+			return err;
+
+		*prev = *index;
+		*index = be32_to_cpu(mgm->next_gid_index) >> 6;
+	} while (*index);
+
+	*index = -1;
+	return err;
+}
+
+static const u8 __promisc_mode[] = {
+	[MLX4_FS_REGULAR]   = 0x0,
+	[MLX4_FS_ALL_DEFAULT] = 0x1,
+	[MLX4_FS_MC_DEFAULT] = 0x3,
+	[MLX4_FS_MIRROR_RX_PORT] = 0x4,
+	[MLX4_FS_MIRROR_SX_PORT] = 0x5,
+	[MLX4_FS_UC_SNIFFER] = 0x6,
+	[MLX4_FS_MC_SNIFFER] = 0x7,
+};
+
+int mlx4_map_sw_to_hw_steering_mode(struct mlx4_dev *dev,
+				    enum mlx4_net_trans_promisc_mode flow_type)
+{
+	if (flow_type >= MLX4_FS_MODE_NUM) {
+		mlx4_err(dev, "Invalid flow type. type = %d\n", flow_type);
+		return -EINVAL;
+	}
+	return __promisc_mode[flow_type];
+}
+EXPORT_SYMBOL_GPL(mlx4_map_sw_to_hw_steering_mode);
+
+static void trans_rule_ctrl_to_hw(struct mlx4_net_trans_rule *ctrl,
+				  struct mlx4_net_trans_rule_hw_ctrl *hw)
+{
+	u8 flags = 0;
+
+	flags = ctrl->queue_mode == MLX4_NET_TRANS_Q_LIFO ? 1 : 0;
+	flags |= ctrl->exclusive ? (1 << 2) : 0;
+	flags |= ctrl->allow_loopback ? (1 << 3) : 0;
+
+	hw->flags = flags;
+	hw->type = __promisc_mode[ctrl->promisc_mode];
+	hw->prio = cpu_to_be16(ctrl->priority);
+	hw->port = ctrl->port;
+	hw->qpn = cpu_to_be32(ctrl->qpn);
+}
+
+const u16 __sw_id_hw[] = {
+	[MLX4_NET_TRANS_RULE_ID_ETH]     = 0xE001,
+	[MLX4_NET_TRANS_RULE_ID_IB]      = 0xE005,
+	[MLX4_NET_TRANS_RULE_ID_IPV6]    = 0xE003,
+	[MLX4_NET_TRANS_RULE_ID_IPV4]    = 0xE002,
+	[MLX4_NET_TRANS_RULE_ID_TCP]     = 0xE004,
+	[MLX4_NET_TRANS_RULE_ID_UDP]     = 0xE006,
+	[MLX4_NET_TRANS_RULE_ID_VXLAN]	 = 0xE008
+};
+
+int mlx4_map_sw_to_hw_steering_id(struct mlx4_dev *dev,
+				  enum mlx4_net_trans_rule_id id)
+{
+	if (id >= MLX4_NET_TRANS_RULE_NUM) {
+		mlx4_err(dev, "Invalid network rule id. id = %d\n", id);
+		return -EINVAL;
+	}
+	return __sw_id_hw[id];
+}
+EXPORT_SYMBOL_GPL(mlx4_map_sw_to_hw_steering_id);
+
+static const int __rule_hw_sz[] = {
+	[MLX4_NET_TRANS_RULE_ID_ETH] =
+		sizeof(struct mlx4_net_trans_rule_hw_eth),
+	[MLX4_NET_TRANS_RULE_ID_IB] =
+		sizeof(struct mlx4_net_trans_rule_hw_ib),
+	[MLX4_NET_TRANS_RULE_ID_IPV6] = 0,
+	[MLX4_NET_TRANS_RULE_ID_IPV4] =
+		sizeof(struct mlx4_net_trans_rule_hw_ipv4),
+	[MLX4_NET_TRANS_RULE_ID_TCP] =
+		sizeof(struct mlx4_net_trans_rule_hw_tcp_udp),
+	[MLX4_NET_TRANS_RULE_ID_UDP] =
+		sizeof(struct mlx4_net_trans_rule_hw_tcp_udp),
+	[MLX4_NET_TRANS_RULE_ID_VXLAN] =
+		sizeof(struct mlx4_net_trans_rule_hw_vxlan)
+};
+
+int mlx4_hw_rule_sz(struct mlx4_dev *dev,
+	       enum mlx4_net_trans_rule_id id)
+{
+	if (id >= MLX4_NET_TRANS_RULE_NUM) {
+		mlx4_err(dev, "Invalid network rule id. id = %d\n", id);
+		return -EINVAL;
+	}
+
+	return __rule_hw_sz[id];
+}
+EXPORT_SYMBOL_GPL(mlx4_hw_rule_sz);
+
+static int parse_trans_rule(struct mlx4_dev *dev, struct mlx4_spec_list *spec,
+			    struct _rule_hw *rule_hw)
+{
+	if (mlx4_hw_rule_sz(dev, spec->id) < 0)
+		return -EINVAL;
+	memset(rule_hw, 0, mlx4_hw_rule_sz(dev, spec->id));
+	rule_hw->id = cpu_to_be16(__sw_id_hw[spec->id]);
+	rule_hw->size = mlx4_hw_rule_sz(dev, spec->id) >> 2;
+
+	switch (spec->id) {
+	case MLX4_NET_TRANS_RULE_ID_ETH:
+		memcpy(rule_hw->eth.dst_mac, spec->eth.dst_mac, ETH_ALEN);
+		memcpy(rule_hw->eth.dst_mac_msk, spec->eth.dst_mac_msk,
+		       ETH_ALEN);
+		memcpy(rule_hw->eth.src_mac, spec->eth.src_mac, ETH_ALEN);
+		memcpy(rule_hw->eth.src_mac_msk, spec->eth.src_mac_msk,
+		       ETH_ALEN);
+		if (spec->eth.ether_type_enable) {
+			rule_hw->eth.ether_type_enable = 1;
+			rule_hw->eth.ether_type = spec->eth.ether_type;
+		}
+		rule_hw->eth.vlan_tag = spec->eth.vlan_id;
+		rule_hw->eth.vlan_tag_msk = spec->eth.vlan_id_msk;
+		break;
+
+	case MLX4_NET_TRANS_RULE_ID_IB:
+		rule_hw->ib.l3_qpn = spec->ib.l3_qpn;
+		rule_hw->ib.qpn_mask = spec->ib.qpn_msk;
+		memcpy(&rule_hw->ib.dst_gid, &spec->ib.dst_gid, 16);
+		memcpy(&rule_hw->ib.dst_gid_msk, &spec->ib.dst_gid_msk, 16);
+		break;
+
+	case MLX4_NET_TRANS_RULE_ID_IPV6:
+		return -EOPNOTSUPP;
+
+	case MLX4_NET_TRANS_RULE_ID_IPV4:
+		rule_hw->ipv4.src_ip = spec->ipv4.src_ip;
+		rule_hw->ipv4.src_ip_msk = spec->ipv4.src_ip_msk;
+		rule_hw->ipv4.dst_ip = spec->ipv4.dst_ip;
+		rule_hw->ipv4.dst_ip_msk = spec->ipv4.dst_ip_msk;
+		break;
+
+	case MLX4_NET_TRANS_RULE_ID_TCP:
+	case MLX4_NET_TRANS_RULE_ID_UDP:
+		rule_hw->tcp_udp.dst_port = spec->tcp_udp.dst_port;
+		rule_hw->tcp_udp.dst_port_msk = spec->tcp_udp.dst_port_msk;
+		rule_hw->tcp_udp.src_port = spec->tcp_udp.src_port;
+		rule_hw->tcp_udp.src_port_msk = spec->tcp_udp.src_port_msk;
+		break;
+
+	case MLX4_NET_TRANS_RULE_ID_VXLAN:
+		rule_hw->vxlan.vni =
+			cpu_to_be32(be32_to_cpu(spec->vxlan.vni) << 8);
+		rule_hw->vxlan.vni_mask =
+			cpu_to_be32(be32_to_cpu(spec->vxlan.vni_mask) << 8);
+		break;
+
+	default:
+		return -EINVAL;
+	}
+
+	return __rule_hw_sz[spec->id];
+}
+
+static void mlx4_err_rule(struct mlx4_dev *dev, char *str,
+			  struct mlx4_net_trans_rule *rule)
+{
+#define BUF_SIZE 256
+	struct mlx4_spec_list *cur;
+	char buf[BUF_SIZE];
+	int len = 0;
+
+	mlx4_err(dev, "%s", str);
+	len += scnprintf(buf + len, BUF_SIZE - len,
+			 "port = %d prio = 0x%x qp = 0x%x ",
+			 rule->port, rule->priority, rule->qpn);
+
+	list_for_each_entry(cur, &rule->list, list) {
+		switch (cur->id) {
+		case MLX4_NET_TRANS_RULE_ID_ETH:
+			len += scnprintf(buf + len, BUF_SIZE - len,
+					 "dmac = %pM ", &cur->eth.dst_mac);
+			if (cur->eth.ether_type)
+				len += scnprintf(buf + len, BUF_SIZE - len,
+						 "ethertype = 0x%x ",
+						 be16_to_cpu(cur->eth.ether_type));
+			if (cur->eth.vlan_id)
+				len += scnprintf(buf + len, BUF_SIZE - len,
+						 "vlan-id = %d ",
+						 be16_to_cpu(cur->eth.vlan_id));
+			break;
+
+		case MLX4_NET_TRANS_RULE_ID_IPV4:
+			if (cur->ipv4.src_ip)
+				len += scnprintf(buf + len, BUF_SIZE - len,
+						 "src-ip = %pI4 ",
+						 &cur->ipv4.src_ip);
+			if (cur->ipv4.dst_ip)
+				len += scnprintf(buf + len, BUF_SIZE - len,
+						 "dst-ip = %pI4 ",
+						 &cur->ipv4.dst_ip);
+			break;
+
+		case MLX4_NET_TRANS_RULE_ID_TCP:
+		case MLX4_NET_TRANS_RULE_ID_UDP:
+			if (cur->tcp_udp.src_port)
+				len += scnprintf(buf + len, BUF_SIZE - len,
+						 "src-port = %d ",
+						 be16_to_cpu(cur->tcp_udp.src_port));
+			if (cur->tcp_udp.dst_port)
+				len += scnprintf(buf + len, BUF_SIZE - len,
+						 "dst-port = %d ",
+						 be16_to_cpu(cur->tcp_udp.dst_port));
+			break;
+
+		case MLX4_NET_TRANS_RULE_ID_IB:
+			len += scnprintf(buf + len, BUF_SIZE - len,
+					 "dst-gid = %pI6\n", cur->ib.dst_gid);
+			len += scnprintf(buf + len, BUF_SIZE - len,
+					 "dst-gid-mask = %pI6\n",
+					 cur->ib.dst_gid_msk);
+			break;
+
+		case MLX4_NET_TRANS_RULE_ID_VXLAN:
+			len += scnprintf(buf + len, BUF_SIZE - len,
+					 "VNID = %d ", be32_to_cpu(cur->vxlan.vni));
+			break;
+		case MLX4_NET_TRANS_RULE_ID_IPV6:
+			break;
+
+		default:
+			break;
+		}
+	}
+	len += scnprintf(buf + len, BUF_SIZE - len, "\n");
+	mlx4_err(dev, "%s", buf);
+
+	if (len >= BUF_SIZE)
+		mlx4_err(dev, "Network rule error message was truncated, print buffer is too small\n");
+}
+
+int mlx4_flow_attach(struct mlx4_dev *dev,
+		     struct mlx4_net_trans_rule *rule, u64 *reg_id)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	struct mlx4_spec_list *cur;
+	u32 size = 0;
+	int ret;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+
+	if (!mlx4_qp_lookup(dev, rule->qpn)) {
+		mlx4_err_rule(dev, "QP doesn't exist\n", rule);
+		ret = -EINVAL;
+		goto out;
+	}
+
+	trans_rule_ctrl_to_hw(rule, mailbox->buf);
+
+	size += sizeof(struct mlx4_net_trans_rule_hw_ctrl);
+
+	list_for_each_entry(cur, &rule->list, list) {
+		ret = parse_trans_rule(dev, cur, mailbox->buf + size);
+		if (ret < 0)
+			goto out;
+
+		size += ret;
+	}
+
+	ret = mlx4_QP_FLOW_STEERING_ATTACH(dev, mailbox, size >> 2, reg_id);
+	if (ret == -ENOMEM) {
+		mlx4_err_rule(dev,
+			      "mcg table is full. Fail to register network rule\n",
+			      rule);
+	} else if (ret) {
+		if (ret == -ENXIO) {
+			if (dev->caps.steering_mode != MLX4_STEERING_MODE_DEVICE_MANAGED)
+				mlx4_err_rule(dev,
+					      "DMFS is not enabled, "
+					      "failed to register network rule.\n",
+					      rule);
+			else
+				mlx4_err_rule(dev,
+					      "Rule exceeds the dmfs_high_rate_mode limitations, "
+					      "failed to register network rule.\n",
+					      rule);
+
+		} else {
+			mlx4_err_rule(dev, "Fail to register network rule.\n", rule);
+		}
+	}
+
+out:
+	mlx4_free_cmd_mailbox(dev, mailbox);
+
+	return ret;
+}
+EXPORT_SYMBOL_GPL(mlx4_flow_attach);
+
+int mlx4_flow_detach(struct mlx4_dev *dev, u64 reg_id)
+{
+	int err;
+
+	err = mlx4_QP_FLOW_STEERING_DETACH(dev, reg_id);
+	if (err)
+		mlx4_err(dev, "Fail to detach network rule. registration id = 0x%llx\n",
+			 reg_id);
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx4_flow_detach);
+
+int mlx4_tunnel_steer_add(struct mlx4_dev *dev, const unsigned char *addr,
+			  int port, int qpn, u16 prio, u64 *reg_id)
+{
+	int err;
+	struct mlx4_spec_list spec_eth_outer = { {NULL} };
+	struct mlx4_spec_list spec_vxlan     = { {NULL} };
+	struct mlx4_spec_list spec_eth_inner = { {NULL} };
+
+	struct mlx4_net_trans_rule rule = {
+		.queue_mode = MLX4_NET_TRANS_Q_FIFO,
+		.exclusive = 0,
+		.allow_loopback = 1,
+		.promisc_mode = MLX4_FS_REGULAR,
+	};
+
+	__be64 mac_mask = cpu_to_be64(MLX4_MAC_MASK << 16);
+
+	rule.port = port;
+	rule.qpn = qpn;
+	rule.priority = prio;
+	INIT_LIST_HEAD(&rule.list);
+
+	spec_eth_outer.id = MLX4_NET_TRANS_RULE_ID_ETH;
+	memcpy(spec_eth_outer.eth.dst_mac, addr, ETH_ALEN);
+	memcpy(spec_eth_outer.eth.dst_mac_msk, &mac_mask, ETH_ALEN);
+
+	spec_vxlan.id = MLX4_NET_TRANS_RULE_ID_VXLAN;    /* any vxlan header */
+	spec_eth_inner.id = MLX4_NET_TRANS_RULE_ID_ETH;	 /* any inner eth header */
+
+	list_add_tail(&spec_eth_outer.list, &rule.list);
+	list_add_tail(&spec_vxlan.list,     &rule.list);
+	list_add_tail(&spec_eth_inner.list, &rule.list);
+
+	err = mlx4_flow_attach(dev, &rule, reg_id);
+	return err;
+}
+EXPORT_SYMBOL(mlx4_tunnel_steer_add);
+
+int mlx4_FLOW_STEERING_IB_UC_QP_RANGE(struct mlx4_dev *dev, u32 min_range_qpn,
+				      u32 max_range_qpn)
+{
+	int err;
+	u64 in_param;
+
+	in_param = ((u64) min_range_qpn) << 32;
+	in_param |= ((u64) max_range_qpn) & 0xFFFFFFFF;
+
+	err = mlx4_cmd(dev, in_param, 0, 0,
+			MLX4_FLOW_STEERING_IB_UC_QP_RANGE,
+			MLX4_CMD_TIME_CLASS_A, MLX4_CMD_NATIVE);
+
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx4_FLOW_STEERING_IB_UC_QP_RANGE);
+
+int mlx4_qp_attach_common(struct mlx4_dev *dev, struct mlx4_qp *qp, u8 gid[16],
+			  int block_mcast_loopback, enum mlx4_protocol prot,
+			  enum mlx4_steer_type steer)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_cmd_mailbox *mailbox;
+	struct mlx4_mgm *mgm;
+	u32 members_count;
+	int index = -1, prev;
+	int link = 0;
+	int i;
+	int err;
+	u8 port = gid[5];
+	u8 new_entry = 0;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+	mgm = mailbox->buf;
+
+	mutex_lock(&priv->mcg_table.mutex);
+	err = find_entry(dev, port, gid, prot,
+			 mailbox, &prev, &index);
+	if (err)
+		goto out;
+
+	if (index != -1) {
+		if (!(be32_to_cpu(mgm->members_count) & 0xffffff)) {
+			new_entry = 1;
+			memcpy(mgm->gid, gid, 16);
+		}
+	} else {
+		link = 1;
+
+		index = mlx4_bitmap_alloc(&priv->mcg_table.bitmap);
+		if (index == -1) {
+			mlx4_err(dev, "No AMGM entries left\n");
+			err = -ENOMEM;
+			goto out;
+		}
+		index += dev->caps.num_mgms;
+
+		new_entry = 1;
+		memset(mgm, 0, sizeof(*mgm));
+		memcpy(mgm->gid, gid, 16);
+	}
+
+	members_count = be32_to_cpu(mgm->members_count) & 0xffffff;
+	if (members_count == dev->caps.num_qp_per_mgm) {
+		mlx4_err(dev, "MGM at index %x is full\n", index);
+		err = -ENOMEM;
+		goto out;
+	}
+
+	for (i = 0; i < members_count; ++i)
+		if ((be32_to_cpu(mgm->qp[i]) & MGM_QPN_MASK) == qp->qpn) {
+			mlx4_dbg(dev, "QP %06x already a member of MGM\n", qp->qpn);
+			err = 0;
+			goto out;
+		}
+
+	if (block_mcast_loopback)
+		mgm->qp[members_count++] = cpu_to_be32((qp->qpn & MGM_QPN_MASK) |
+						       (1U << MGM_BLCK_LB_BIT));
+	else
+		mgm->qp[members_count++] = cpu_to_be32(qp->qpn & MGM_QPN_MASK);
+
+	mgm->members_count = cpu_to_be32(members_count | (u32) prot << 30);
+
+	err = mlx4_WRITE_ENTRY(dev, index, mailbox);
+	if (err)
+		goto out;
+
+	if (!link)
+		goto out;
+
+	err = mlx4_READ_ENTRY(dev, prev, mailbox);
+	if (err)
+		goto out;
+
+	mgm->next_gid_index = cpu_to_be32(index << 6);
+
+	err = mlx4_WRITE_ENTRY(dev, prev, mailbox);
+	if (err)
+		goto out;
+
+out:
+	if (prot == MLX4_PROT_ETH && index != -1) {
+		/* manage the steering entry for promisc mode */
+		if (new_entry)
+			err = new_steering_entry(dev, port, steer,
+						 index, qp->qpn);
+		else
+			err = existing_steering_entry(dev, port, steer,
+						      index, qp->qpn);
+	}
+	if (err && link && index != -1) {
+		if (index < dev->caps.num_mgms)
+			mlx4_warn(dev, "Got AMGM index %d < %d\n",
+				  index, dev->caps.num_mgms);
+		else
+			mlx4_bitmap_free(&priv->mcg_table.bitmap,
+					 index - dev->caps.num_mgms, MLX4_USE_RR);
+	}
+	mutex_unlock(&priv->mcg_table.mutex);
+
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return err;
+}
+
+int mlx4_qp_detach_common(struct mlx4_dev *dev, struct mlx4_qp *qp, u8 gid[16],
+			  enum mlx4_protocol prot, enum mlx4_steer_type steer)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_cmd_mailbox *mailbox;
+	struct mlx4_mgm *mgm;
+	u32 members_count;
+	int prev, index;
+	int i, loc = -1;
+	int err;
+	u8 port = gid[5];
+	bool removed_entry = false;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+	mgm = mailbox->buf;
+
+	mutex_lock(&priv->mcg_table.mutex);
+
+	err = find_entry(dev, port, gid, prot,
+			 mailbox, &prev, &index);
+	if (err)
+		goto out;
+
+	if (index == -1) {
+		mlx4_err(dev, "MGID %pI6 not found\n", gid);
+		err = -EINVAL;
+		goto out;
+	}
+
+	/* If this QP is also a promisc QP, it shouldn't be removed only if
+	 * at least one none promisc QP is also attached to this MCG
+	 */
+	if (prot == MLX4_PROT_ETH &&
+	    check_duplicate_entry(dev, port, steer, index, qp->qpn) &&
+	    !promisc_steering_entry(dev, port, steer, index, qp->qpn, NULL))
+			goto out;
+
+	members_count = be32_to_cpu(mgm->members_count) & 0xffffff;
+	for (i = 0; i < members_count; ++i)
+		if ((be32_to_cpu(mgm->qp[i]) & MGM_QPN_MASK) == qp->qpn) {
+			loc = i;
+			break;
+		}
+
+	if (loc == -1) {
+		mlx4_err(dev, "QP %06x not found in MGM\n", qp->qpn);
+		err = -EINVAL;
+		goto out;
+	}
+
+	/* copy the last QP in this MGM over removed QP */
+	mgm->qp[loc] = mgm->qp[members_count - 1];
+	mgm->qp[members_count - 1] = 0;
+	mgm->members_count = cpu_to_be32(--members_count | (u32) prot << 30);
+
+	if (prot == MLX4_PROT_ETH)
+		removed_entry = can_remove_steering_entry(dev, port, steer,
+								index, qp->qpn);
+	if (members_count && (prot != MLX4_PROT_ETH || !removed_entry)) {
+		err = mlx4_WRITE_ENTRY(dev, index, mailbox);
+		goto out;
+	}
+
+	/* We are going to delete the entry, members count should be 0 */
+	mgm->members_count = cpu_to_be32((u32) prot << 30);
+
+	if (prev == -1) {
+		/* Remove entry from MGM */
+		int amgm_index = be32_to_cpu(mgm->next_gid_index) >> 6;
+		if (amgm_index) {
+			err = mlx4_READ_ENTRY(dev, amgm_index, mailbox);
+			if (err)
+				goto out;
+		} else
+			memset(mgm->gid, 0, 16);
+
+		err = mlx4_WRITE_ENTRY(dev, index, mailbox);
+		if (err)
+			goto out;
+
+		if (amgm_index) {
+			if (amgm_index < dev->caps.num_mgms)
+				mlx4_warn(dev, "MGM entry %d had AMGM index %d < %d\n",
+					  index, amgm_index, dev->caps.num_mgms);
+			else
+				mlx4_bitmap_free(&priv->mcg_table.bitmap,
+						 amgm_index - dev->caps.num_mgms, MLX4_USE_RR);
+		}
+	} else {
+		/* Remove entry from AMGM */
+		int cur_next_index = be32_to_cpu(mgm->next_gid_index) >> 6;
+		err = mlx4_READ_ENTRY(dev, prev, mailbox);
+		if (err)
+			goto out;
+
+		mgm->next_gid_index = cpu_to_be32(cur_next_index << 6);
+
+		err = mlx4_WRITE_ENTRY(dev, prev, mailbox);
+		if (err)
+			goto out;
+
+		if (index < dev->caps.num_mgms)
+			mlx4_warn(dev, "entry %d had next AMGM index %d < %d\n",
+				  prev, index, dev->caps.num_mgms);
+		else
+			mlx4_bitmap_free(&priv->mcg_table.bitmap,
+					 index - dev->caps.num_mgms, MLX4_USE_RR);
+	}
+
+out:
+	mutex_unlock(&priv->mcg_table.mutex);
+
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	if (err && dev->persist->state & MLX4_DEVICE_STATE_INTERNAL_ERROR)
+		/* In case device is under an error, return success as a closing command */
+		err = 0;
+	return err;
+}
+
+static int mlx4_QP_ATTACH(struct mlx4_dev *dev, struct mlx4_qp *qp,
+			  u8 gid[16], u8 attach, u8 block_loopback,
+			  enum mlx4_protocol prot)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	int err = 0;
+	int qpn;
+
+	if (!mlx4_is_mfunc(dev))
+		return -EBADF;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+
+	memcpy(mailbox->buf, gid, 16);
+	qpn = qp->qpn;
+	qpn |= (prot << 28);
+	if (attach && block_loopback)
+		qpn |= (1 << 31);
+
+	err = mlx4_cmd(dev, mailbox->dma, qpn, attach,
+		       MLX4_CMD_QP_ATTACH, MLX4_CMD_TIME_CLASS_A,
+		       MLX4_CMD_WRAPPED);
+
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	if (err && !attach &&
+	    dev->persist->state & MLX4_DEVICE_STATE_INTERNAL_ERROR)
+		err = 0;
+	return err;
+}
+
+int mlx4_trans_to_dmfs_attach(struct mlx4_dev *dev, struct mlx4_qp *qp,
+			      u8 gid[16], u8 port,
+			      int block_mcast_loopback,
+			      enum mlx4_protocol prot, u64 *reg_id)
+{
+		struct mlx4_spec_list spec = { {NULL} };
+		__be64 mac_mask = cpu_to_be64(MLX4_MAC_MASK << 16);
+
+		struct mlx4_net_trans_rule rule = {
+			.queue_mode = MLX4_NET_TRANS_Q_FIFO,
+			.exclusive = 0,
+			.promisc_mode = MLX4_FS_REGULAR,
+			.priority = MLX4_DOMAIN_NIC,
+		};
+
+		rule.allow_loopback = !block_mcast_loopback;
+		rule.port = port;
+		rule.qpn = qp->qpn;
+		INIT_LIST_HEAD(&rule.list);
+
+		switch (prot) {
+		case MLX4_PROT_ETH:
+			spec.id = MLX4_NET_TRANS_RULE_ID_ETH;
+			memcpy(spec.eth.dst_mac, &gid[10], ETH_ALEN);
+			memcpy(spec.eth.dst_mac_msk, &mac_mask, ETH_ALEN);
+			break;
+
+		case MLX4_PROT_IB_IPV6:
+			spec.id = MLX4_NET_TRANS_RULE_ID_IB;
+			memcpy(spec.ib.dst_gid, gid, 16);
+			memset(&spec.ib.dst_gid_msk, 0xff, 16);
+			break;
+		default:
+			return -EINVAL;
+		}
+		list_add_tail(&spec.list, &rule.list);
+
+		return mlx4_flow_attach(dev, &rule, reg_id);
+}
+
+int mlx4_multicast_attach(struct mlx4_dev *dev, struct mlx4_qp *qp, u8 gid[16],
+			  u8 port, int block_mcast_loopback,
+			  enum mlx4_protocol prot, u64 *reg_id)
+{
+	switch (dev->caps.steering_mode) {
+	case MLX4_STEERING_MODE_A0:
+		if (prot == MLX4_PROT_ETH)
+			return 0;
+		fallthrough;
+
+	case MLX4_STEERING_MODE_B0:
+		if (prot == MLX4_PROT_ETH)
+			gid[7] |= (MLX4_MC_STEER << 1);
+
+		if (mlx4_is_mfunc(dev))
+			return mlx4_QP_ATTACH(dev, qp, gid, 1,
+					      block_mcast_loopback, prot);
+		return mlx4_qp_attach_common(dev, qp, gid,
+					     block_mcast_loopback, prot,
+					     MLX4_MC_STEER);
+
+	case MLX4_STEERING_MODE_DEVICE_MANAGED:
+		return mlx4_trans_to_dmfs_attach(dev, qp, gid, port,
+						 block_mcast_loopback,
+						 prot, reg_id);
+	default:
+		return -EINVAL;
+	}
+}
+EXPORT_SYMBOL_GPL(mlx4_multicast_attach);
+
+int mlx4_multicast_detach(struct mlx4_dev *dev, struct mlx4_qp *qp, u8 gid[16],
+			  enum mlx4_protocol prot, u64 reg_id)
+{
+	switch (dev->caps.steering_mode) {
+	case MLX4_STEERING_MODE_A0:
+		if (prot == MLX4_PROT_ETH)
+			return 0;
+		fallthrough;
+
+	case MLX4_STEERING_MODE_B0:
+		if (prot == MLX4_PROT_ETH)
+			gid[7] |= (MLX4_MC_STEER << 1);
+
+		if (mlx4_is_mfunc(dev))
+			return mlx4_QP_ATTACH(dev, qp, gid, 0, 0, prot);
+
+		return mlx4_qp_detach_common(dev, qp, gid, prot,
+					     MLX4_MC_STEER);
+
+	case MLX4_STEERING_MODE_DEVICE_MANAGED:
+		return mlx4_flow_detach(dev, reg_id);
+
+	default:
+		return -EINVAL;
+	}
+}
+EXPORT_SYMBOL_GPL(mlx4_multicast_detach);
+
+int mlx4_flow_steer_promisc_add(struct mlx4_dev *dev, u8 port,
+				u32 qpn, enum mlx4_net_trans_promisc_mode mode)
+{
+	struct mlx4_net_trans_rule rule = {
+		.queue_mode = MLX4_NET_TRANS_Q_FIFO,
+		.exclusive = 0,
+		.allow_loopback = 1,
+	};
+
+	u64 *regid_p;
+
+	switch (mode) {
+	case MLX4_FS_ALL_DEFAULT:
+		regid_p = &dev->regid_promisc_array[port];
+		break;
+	case MLX4_FS_MC_DEFAULT:
+		regid_p = &dev->regid_allmulti_array[port];
+		break;
+	default:
+		return -1;
+	}
+
+	if (*regid_p != 0)
+		return -1;
+
+	rule.promisc_mode = mode;
+	rule.port = port;
+	rule.qpn = qpn;
+	INIT_LIST_HEAD(&rule.list);
+	mlx4_info(dev, "going promisc on %x\n", port);
+
+	return  mlx4_flow_attach(dev, &rule, regid_p);
+}
+EXPORT_SYMBOL_GPL(mlx4_flow_steer_promisc_add);
+
+int mlx4_flow_steer_promisc_remove(struct mlx4_dev *dev, u8 port,
+				   enum mlx4_net_trans_promisc_mode mode)
+{
+	int ret;
+	u64 *regid_p;
+
+	switch (mode) {
+	case MLX4_FS_ALL_DEFAULT:
+		regid_p = &dev->regid_promisc_array[port];
+		break;
+	case MLX4_FS_MC_DEFAULT:
+		regid_p = &dev->regid_allmulti_array[port];
+		break;
+	default:
+		return -1;
+	}
+
+	if (*regid_p == 0)
+		return -1;
+
+	ret =  mlx4_flow_detach(dev, *regid_p);
+	if (ret == 0)
+		*regid_p = 0;
+
+	return ret;
+}
+EXPORT_SYMBOL_GPL(mlx4_flow_steer_promisc_remove);
+
+int mlx4_unicast_attach(struct mlx4_dev *dev,
+			struct mlx4_qp *qp, u8 gid[16],
+			int block_mcast_loopback, enum mlx4_protocol prot)
+{
+	if (prot == MLX4_PROT_ETH)
+		gid[7] |= (MLX4_UC_STEER << 1);
+
+	if (mlx4_is_mfunc(dev))
+		return mlx4_QP_ATTACH(dev, qp, gid, 1,
+					block_mcast_loopback, prot);
+
+	return mlx4_qp_attach_common(dev, qp, gid, block_mcast_loopback,
+					prot, MLX4_UC_STEER);
+}
+EXPORT_SYMBOL_GPL(mlx4_unicast_attach);
+
+int mlx4_unicast_detach(struct mlx4_dev *dev, struct mlx4_qp *qp,
+			       u8 gid[16], enum mlx4_protocol prot)
+{
+	if (prot == MLX4_PROT_ETH)
+		gid[7] |= (MLX4_UC_STEER << 1);
+
+	if (mlx4_is_mfunc(dev))
+		return mlx4_QP_ATTACH(dev, qp, gid, 0, 0, prot);
+
+	return mlx4_qp_detach_common(dev, qp, gid, prot, MLX4_UC_STEER);
+}
+EXPORT_SYMBOL_GPL(mlx4_unicast_detach);
+
+int mlx4_PROMISC_wrapper(struct mlx4_dev *dev, int slave,
+			 struct mlx4_vhcr *vhcr,
+			 struct mlx4_cmd_mailbox *inbox,
+			 struct mlx4_cmd_mailbox *outbox,
+			 struct mlx4_cmd_info *cmd)
+{
+	u32 qpn = (u32) vhcr->in_param & 0xffffffff;
+	int port = mlx4_slave_convert_port(dev, slave, vhcr->in_param >> 62);
+	enum mlx4_steer_type steer = vhcr->in_modifier;
+
+	if (port < 0)
+		return -EINVAL;
+
+	/* Promiscuous unicast is not allowed in mfunc */
+	if (mlx4_is_mfunc(dev) && steer == MLX4_UC_STEER)
+		return 0;
+
+	if (vhcr->op_modifier)
+		return add_promisc_qp(dev, port, steer, qpn);
+	else
+		return remove_promisc_qp(dev, port, steer, qpn);
+}
+
+static int mlx4_PROMISC(struct mlx4_dev *dev, u32 qpn,
+			enum mlx4_steer_type steer, u8 add, u8 port)
+{
+	return mlx4_cmd(dev, (u64) qpn | (u64) port << 62, (u32) steer, add,
+			MLX4_CMD_PROMISC, MLX4_CMD_TIME_CLASS_A,
+			MLX4_CMD_WRAPPED);
+}
+
+int mlx4_multicast_promisc_add(struct mlx4_dev *dev, u32 qpn, u8 port)
+{
+	if (mlx4_is_mfunc(dev))
+		return mlx4_PROMISC(dev, qpn, MLX4_MC_STEER, 1, port);
+
+	return add_promisc_qp(dev, port, MLX4_MC_STEER, qpn);
+}
+EXPORT_SYMBOL_GPL(mlx4_multicast_promisc_add);
+
+int mlx4_multicast_promisc_remove(struct mlx4_dev *dev, u32 qpn, u8 port)
+{
+	if (mlx4_is_mfunc(dev))
+		return mlx4_PROMISC(dev, qpn, MLX4_MC_STEER, 0, port);
+
+	return remove_promisc_qp(dev, port, MLX4_MC_STEER, qpn);
+}
+EXPORT_SYMBOL_GPL(mlx4_multicast_promisc_remove);
+
+int mlx4_unicast_promisc_add(struct mlx4_dev *dev, u32 qpn, u8 port)
+{
+	if (mlx4_is_mfunc(dev))
+		return mlx4_PROMISC(dev, qpn, MLX4_UC_STEER, 1, port);
+
+	return add_promisc_qp(dev, port, MLX4_UC_STEER, qpn);
+}
+EXPORT_SYMBOL_GPL(mlx4_unicast_promisc_add);
+
+int mlx4_unicast_promisc_remove(struct mlx4_dev *dev, u32 qpn, u8 port)
+{
+	if (mlx4_is_mfunc(dev))
+		return mlx4_PROMISC(dev, qpn, MLX4_UC_STEER, 0, port);
+
+	return remove_promisc_qp(dev, port, MLX4_UC_STEER, qpn);
+}
+EXPORT_SYMBOL_GPL(mlx4_unicast_promisc_remove);
+
+int mlx4_init_mcg_table(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int err;
+
+	/* No need for mcg_table when fw managed the mcg table*/
+	if (dev->caps.steering_mode ==
+	    MLX4_STEERING_MODE_DEVICE_MANAGED)
+		return 0;
+	err = mlx4_bitmap_init(&priv->mcg_table.bitmap, dev->caps.num_amgms,
+			       dev->caps.num_amgms - 1, 0, 0);
+	if (err)
+		return err;
+
+	mutex_init(&priv->mcg_table.mutex);
+
+	return 0;
+}
+
+void mlx4_cleanup_mcg_table(struct mlx4_dev *dev)
+{
+	if (dev->caps.steering_mode !=
+	    MLX4_STEERING_MODE_DEVICE_MANAGED)
+		mlx4_bitmap_cleanup(&mlx4_priv(dev)->mcg_table.bitmap);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx4/mlx4.h b/drivers/net/ethernet/mellanox/mlx4/mlx4.h
new file mode 100644
index 000000000..6ccf34066
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx4/mlx4.h
@@ -0,0 +1,1486 @@
+/*
+ * Copyright (c) 2004, 2005 Topspin Communications.  All rights reserved.
+ * Copyright (c) 2005 Sun Microsystems, Inc. All rights reserved.
+ * Copyright (c) 2005, 2006, 2007 Cisco Systems.  All rights reserved.
+ * Copyright (c) 2005, 2006, 2007, 2008 Mellanox Technologies. All rights reserved.
+ * Copyright (c) 2004 Voltaire, Inc. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef MLX4_H
+#define MLX4_H
+
+#include <linux/mutex.h>
+#include <linux/radix-tree.h>
+#include <linux/rbtree.h>
+#include <linux/timer.h>
+#include <linux/semaphore.h>
+#include <linux/workqueue.h>
+#include <linux/interrupt.h>
+#include <linux/spinlock.h>
+#include <net/devlink.h>
+#include <linux/rwsem.h>
+
+#include <linux/mlx4/device.h>
+#include <linux/mlx4/driver.h>
+#include <linux/mlx4/doorbell.h>
+#include <linux/mlx4/cmd.h>
+#include "fw_qos.h"
+
+#define DRV_NAME	"mlx4_core"
+#define DRV_VERSION	"4.0-0"
+#define DRV_NAME_FOR_FW		"Linux," DRV_NAME "," DRV_VERSION
+
+#define MLX4_FS_UDP_UC_EN		(1 << 1)
+#define MLX4_FS_TCP_UC_EN		(1 << 2)
+#define MLX4_FS_NUM_OF_L2_ADDR		8
+#define MLX4_FS_MGM_LOG_ENTRY_SIZE	7
+#define MLX4_FS_NUM_MCG			(1 << 17)
+
+#define INIT_HCA_TPT_MW_ENABLE          (1 << 7)
+
+#define MLX4_QUERY_IF_STAT_RESET	BIT(31)
+
+enum {
+	MLX4_HCR_BASE		= 0x80680,
+	MLX4_HCR_SIZE		= 0x0001c,
+	MLX4_CLR_INT_SIZE	= 0x00008,
+	MLX4_SLAVE_COMM_BASE	= 0x0,
+	MLX4_COMM_PAGESIZE	= 0x1000,
+	MLX4_CLOCK_SIZE		= 0x00008,
+	MLX4_COMM_CHAN_CAPS	= 0x8,
+	MLX4_COMM_CHAN_FLAGS	= 0xc
+};
+
+enum {
+	MLX4_DEFAULT_MGM_LOG_ENTRY_SIZE = 10,
+	MLX4_MIN_MGM_LOG_ENTRY_SIZE = 7,
+	MLX4_MAX_MGM_LOG_ENTRY_SIZE = 12,
+	MLX4_MAX_QP_PER_MGM = 4 * ((1 << MLX4_MAX_MGM_LOG_ENTRY_SIZE) / 16 - 2),
+};
+
+enum {
+	MLX4_NUM_PDS		= 1 << 15
+};
+
+enum {
+	MLX4_CMPT_TYPE_QP	= 0,
+	MLX4_CMPT_TYPE_SRQ	= 1,
+	MLX4_CMPT_TYPE_CQ	= 2,
+	MLX4_CMPT_TYPE_EQ	= 3,
+	MLX4_CMPT_NUM_TYPE
+};
+
+enum {
+	MLX4_CMPT_SHIFT		= 24,
+	MLX4_NUM_CMPTS		= MLX4_CMPT_NUM_TYPE << MLX4_CMPT_SHIFT
+};
+
+enum mlx4_mpt_state {
+	MLX4_MPT_DISABLED = 0,
+	MLX4_MPT_EN_HW,
+	MLX4_MPT_EN_SW
+};
+
+#define MLX4_COMM_TIME		10000
+#define MLX4_COMM_OFFLINE_TIME_OUT 30000
+#define MLX4_COMM_CMD_NA_OP    0x0
+
+
+enum {
+	MLX4_COMM_CMD_RESET,
+	MLX4_COMM_CMD_VHCR0,
+	MLX4_COMM_CMD_VHCR1,
+	MLX4_COMM_CMD_VHCR2,
+	MLX4_COMM_CMD_VHCR_EN,
+	MLX4_COMM_CMD_VHCR_POST,
+	MLX4_COMM_CMD_FLR = 254
+};
+
+enum {
+	MLX4_VF_SMI_DISABLED,
+	MLX4_VF_SMI_ENABLED
+};
+
+/*The flag indicates that the slave should delay the RESET cmd*/
+#define MLX4_DELAY_RESET_SLAVE 0xbbbbbbb
+/*indicates how many retries will be done if we are in the middle of FLR*/
+#define NUM_OF_RESET_RETRIES	10
+#define SLEEP_TIME_IN_RESET	(2 * 1000)
+enum mlx4_resource {
+	RES_QP,
+	RES_CQ,
+	RES_SRQ,
+	RES_XRCD,
+	RES_MPT,
+	RES_MTT,
+	RES_MAC,
+	RES_VLAN,
+	RES_NPORT_ID,
+	RES_COUNTER,
+	RES_FS_RULE,
+	RES_EQ,
+	MLX4_NUM_OF_RESOURCE_TYPE
+};
+
+enum mlx4_alloc_mode {
+	RES_OP_RESERVE,
+	RES_OP_RESERVE_AND_MAP,
+	RES_OP_MAP_ICM,
+};
+
+enum mlx4_res_tracker_free_type {
+	RES_TR_FREE_ALL,
+	RES_TR_FREE_SLAVES_ONLY,
+	RES_TR_FREE_STRUCTS_ONLY,
+};
+
+/*
+ *Virtual HCR structures.
+ * mlx4_vhcr is the sw representation, in machine endianness
+ *
+ * mlx4_vhcr_cmd is the formalized structure, the one that is passed
+ * to FW to go through communication channel.
+ * It is big endian, and has the same structure as the physical HCR
+ * used by command interface
+ */
+struct mlx4_vhcr {
+	u64	in_param;
+	u64	out_param;
+	u32	in_modifier;
+	u32	errno;
+	u16	op;
+	u16	token;
+	u8	op_modifier;
+	u8	e_bit;
+};
+
+struct mlx4_vhcr_cmd {
+	__be64 in_param;
+	__be32 in_modifier;
+	u32 reserved1;
+	__be64 out_param;
+	__be16 token;
+	u16 reserved;
+	u8 status;
+	u8 flags;
+	__be16 opcode;
+};
+
+struct mlx4_cmd_info {
+	u16 opcode;
+	bool has_inbox;
+	bool has_outbox;
+	bool out_is_imm;
+	bool encode_slave_id;
+	int (*verify)(struct mlx4_dev *dev, int slave, struct mlx4_vhcr *vhcr,
+		      struct mlx4_cmd_mailbox *inbox);
+	int (*wrapper)(struct mlx4_dev *dev, int slave, struct mlx4_vhcr *vhcr,
+		       struct mlx4_cmd_mailbox *inbox,
+		       struct mlx4_cmd_mailbox *outbox,
+		       struct mlx4_cmd_info *cmd);
+};
+
+#ifdef CONFIG_MLX4_DEBUG
+extern int mlx4_debug_level;
+#else /* CONFIG_MLX4_DEBUG */
+#define mlx4_debug_level	(0)
+#endif /* CONFIG_MLX4_DEBUG */
+
+#define mlx4_dbg(mdev, format, ...)					\
+do {									\
+	if (mlx4_debug_level)						\
+		dev_printk(KERN_DEBUG,					\
+			   &(mdev)->persist->pdev->dev, format,		\
+			   ##__VA_ARGS__);				\
+} while (0)
+
+#define mlx4_err(mdev, format, ...)					\
+	dev_err(&(mdev)->persist->pdev->dev, format, ##__VA_ARGS__)
+#define mlx4_info(mdev, format, ...)					\
+	dev_info(&(mdev)->persist->pdev->dev, format, ##__VA_ARGS__)
+#define mlx4_warn(mdev, format, ...)					\
+	dev_warn(&(mdev)->persist->pdev->dev, format, ##__VA_ARGS__)
+
+extern int log_mtts_per_seg;
+extern int mlx4_internal_err_reset;
+
+#define MLX4_MAX_NUM_SLAVES	(min(MLX4_MAX_NUM_PF + MLX4_MAX_NUM_VF, \
+				     MLX4_MFUNC_MAX))
+#define ALL_SLAVES 0xff
+
+struct mlx4_bitmap {
+	u32			last;
+	u32			top;
+	u32			max;
+	u32                     reserved_top;
+	u32			mask;
+	u32			avail;
+	u32			effective_len;
+	spinlock_t		lock;
+	unsigned long	       *table;
+};
+
+struct mlx4_buddy {
+	unsigned long	      **bits;
+	unsigned int	       *num_free;
+	u32			max_order;
+	spinlock_t		lock;
+};
+
+struct mlx4_icm;
+
+struct mlx4_icm_table {
+	u64			virt;
+	int			num_icm;
+	u32			num_obj;
+	int			obj_size;
+	int			lowmem;
+	int			coherent;
+	struct mutex		mutex;
+	struct mlx4_icm	      **icm;
+};
+
+#define MLX4_MPT_FLAG_SW_OWNS	    (0xfUL << 28)
+#define MLX4_MPT_FLAG_FREE	    (0x3UL << 28)
+#define MLX4_MPT_FLAG_MIO	    (1 << 17)
+#define MLX4_MPT_FLAG_BIND_ENABLE   (1 << 15)
+#define MLX4_MPT_FLAG_PHYSICAL	    (1 <<  9)
+#define MLX4_MPT_FLAG_REGION	    (1 <<  8)
+
+#define MLX4_MPT_PD_MASK	    (0x1FFFFUL)
+#define MLX4_MPT_PD_VF_MASK	    (0xFE0000UL)
+#define MLX4_MPT_PD_FLAG_FAST_REG   (1 << 27)
+#define MLX4_MPT_PD_FLAG_RAE	    (1 << 28)
+#define MLX4_MPT_PD_FLAG_EN_INV	    (3 << 24)
+
+#define MLX4_MPT_QP_FLAG_BOUND_QP   (1 << 7)
+
+#define MLX4_MPT_STATUS_SW		0xF0
+#define MLX4_MPT_STATUS_HW		0x00
+
+#define MLX4_CQE_SIZE_MASK_STRIDE	0x3
+#define MLX4_EQE_SIZE_MASK_STRIDE	0x30
+
+#define MLX4_EQ_ASYNC			0
+#define MLX4_EQ_TO_CQ_VECTOR(vector)	((vector) - \
+					 !!((int)(vector) >= MLX4_EQ_ASYNC))
+#define MLX4_CQ_TO_EQ_VECTOR(vector)	((vector) + \
+					 !!((int)(vector) >= MLX4_EQ_ASYNC))
+
+/*
+ * Must be packed because mtt_seg is 64 bits but only aligned to 32 bits.
+ */
+struct mlx4_mpt_entry {
+	__be32 flags;
+	__be32 qpn;
+	__be32 key;
+	__be32 pd_flags;
+	__be64 start;
+	__be64 length;
+	__be32 lkey;
+	__be32 win_cnt;
+	u8	reserved1[3];
+	u8	mtt_rep;
+	__be64 mtt_addr;
+	__be32 mtt_sz;
+	__be32 entity_size;
+	__be32 first_byte_offset;
+} __packed;
+
+/*
+ * Must be packed because start is 64 bits but only aligned to 32 bits.
+ */
+struct mlx4_eq_context {
+	__be32			flags;
+	u16			reserved1[3];
+	__be16			page_offset;
+	u8			log_eq_size;
+	u8			reserved2[4];
+	u8			eq_period;
+	u8			reserved3;
+	u8			eq_max_count;
+	u8			reserved4[3];
+	u8			intr;
+	u8			log_page_size;
+	u8			reserved5[2];
+	u8			mtt_base_addr_h;
+	__be32			mtt_base_addr_l;
+	u32			reserved6[2];
+	__be32			consumer_index;
+	__be32			producer_index;
+	u32			reserved7[4];
+};
+
+struct mlx4_cq_context {
+	__be32			flags;
+	u16			reserved1[3];
+	__be16			page_offset;
+	__be32			logsize_usrpage;
+	__be16			cq_period;
+	__be16			cq_max_count;
+	u8			reserved2[3];
+	u8			comp_eqn;
+	u8			log_page_size;
+	u8			reserved3[2];
+	u8			mtt_base_addr_h;
+	__be32			mtt_base_addr_l;
+	__be32			last_notified_index;
+	__be32			solicit_producer_index;
+	__be32			consumer_index;
+	__be32			producer_index;
+	u32			reserved4[2];
+	__be64			db_rec_addr;
+};
+
+struct mlx4_srq_context {
+	__be32			state_logsize_srqn;
+	u8			logstride;
+	u8			reserved1;
+	__be16			xrcd;
+	__be32			pg_offset_cqn;
+	u32			reserved2;
+	u8			log_page_size;
+	u8			reserved3[2];
+	u8			mtt_base_addr_h;
+	__be32			mtt_base_addr_l;
+	__be32			pd;
+	__be16			limit_watermark;
+	__be16			wqe_cnt;
+	u16			reserved4;
+	__be16			wqe_counter;
+	u32			reserved5;
+	__be64			db_rec_addr;
+};
+
+struct mlx4_eq_tasklet {
+	struct list_head list;
+	struct list_head process_list;
+	struct tasklet_struct task;
+	/* lock on completion tasklet list */
+	spinlock_t lock;
+};
+
+struct mlx4_eq {
+	struct mlx4_dev	       *dev;
+	void __iomem	       *doorbell;
+	int			eqn;
+	u32			cons_index;
+	u16			irq;
+	u16			have_irq;
+	int			nent;
+	struct mlx4_buf_list   *page_list;
+	struct mlx4_mtt		mtt;
+	struct mlx4_eq_tasklet	tasklet_ctx;
+	struct mlx4_active_ports actv_ports;
+	u32			ref_count;
+	cpumask_var_t		affinity_mask;
+};
+
+struct mlx4_slave_eqe {
+	u8 type;
+	u8 port;
+	u32 param;
+};
+
+struct mlx4_slave_event_eq_info {
+	int eqn;
+	u16 token;
+};
+
+struct mlx4_profile {
+	int			num_qp;
+	int			rdmarc_per_qp;
+	int			num_srq;
+	int			num_cq;
+	int			num_mcg;
+	int			num_mpt;
+	unsigned		num_mtt;
+};
+
+struct mlx4_fw {
+	u64			clr_int_base;
+	u64			catas_offset;
+	u64			comm_base;
+	u64			clock_offset;
+	struct mlx4_icm	       *fw_icm;
+	struct mlx4_icm	       *aux_icm;
+	u32			catas_size;
+	u16			fw_pages;
+	u8			clr_int_bar;
+	u8			catas_bar;
+	u8			comm_bar;
+	u8			clock_bar;
+};
+
+struct mlx4_comm {
+	u32			slave_write;
+	u32			slave_read;
+};
+
+enum {
+	MLX4_MCAST_CONFIG       = 0,
+	MLX4_MCAST_DISABLE      = 1,
+	MLX4_MCAST_ENABLE       = 2,
+};
+
+#define VLAN_FLTR_SIZE	128
+
+struct mlx4_vlan_fltr {
+	__be32 entry[VLAN_FLTR_SIZE];
+};
+
+struct mlx4_mcast_entry {
+	struct list_head list;
+	u64 addr;
+};
+
+struct mlx4_promisc_qp {
+	struct list_head list;
+	u32 qpn;
+};
+
+struct mlx4_steer_index {
+	struct list_head list;
+	unsigned int index;
+	struct list_head duplicates;
+};
+
+#define MLX4_EVENT_TYPES_NUM 64
+
+struct mlx4_slave_state {
+	u8 comm_toggle;
+	u8 last_cmd;
+	u8 init_port_mask;
+	bool active;
+	bool old_vlan_api;
+	bool vst_qinq_supported;
+	u8 function;
+	dma_addr_t vhcr_dma;
+	u16 user_mtu[MLX4_MAX_PORTS + 1];
+	u16 mtu[MLX4_MAX_PORTS + 1];
+	__be32 ib_cap_mask[MLX4_MAX_PORTS + 1];
+	struct mlx4_slave_eqe eq[MLX4_MFUNC_MAX_EQES];
+	struct list_head mcast_filters[MLX4_MAX_PORTS + 1];
+	struct mlx4_vlan_fltr *vlan_filter[MLX4_MAX_PORTS + 1];
+	/* event type to eq number lookup */
+	struct mlx4_slave_event_eq_info event_eq[MLX4_EVENT_TYPES_NUM];
+	u16 eq_pi;
+	u16 eq_ci;
+	spinlock_t lock;
+	/*initialized via the kzalloc*/
+	u8 is_slave_going_down;
+	u32 cookie;
+	enum slave_port_state port_state[MLX4_MAX_PORTS + 1];
+};
+
+#define MLX4_VGT 4095
+#define NO_INDX  (-1)
+
+struct mlx4_vport_state {
+	u64 mac;
+	u16 default_vlan;
+	u8  default_qos;
+	__be16 vlan_proto;
+	u32 tx_rate;
+	bool spoofchk;
+	u32 link_state;
+	u8 qos_vport;
+	__be64 guid;
+};
+
+struct mlx4_vf_admin_state {
+	struct mlx4_vport_state vport[MLX4_MAX_PORTS + 1];
+	u8 enable_smi[MLX4_MAX_PORTS + 1];
+};
+
+struct mlx4_vport_oper_state {
+	struct mlx4_vport_state state;
+	int mac_idx;
+	int vlan_idx;
+};
+
+struct mlx4_vf_oper_state {
+	struct mlx4_vport_oper_state vport[MLX4_MAX_PORTS + 1];
+	u8 smi_enabled[MLX4_MAX_PORTS + 1];
+};
+
+struct slave_list {
+	struct mutex mutex;
+	struct list_head res_list[MLX4_NUM_OF_RESOURCE_TYPE];
+};
+
+struct resource_allocator {
+	spinlock_t alloc_lock; /* protect quotas */
+	union {
+		unsigned int res_reserved;
+		unsigned int res_port_rsvd[MLX4_MAX_PORTS];
+	};
+	union {
+		int res_free;
+		int res_port_free[MLX4_MAX_PORTS];
+	};
+	int *quota;
+	int *allocated;
+	int *guaranteed;
+};
+
+struct mlx4_resource_tracker {
+	spinlock_t lock;
+	/* tree for each resources */
+	struct rb_root res_tree[MLX4_NUM_OF_RESOURCE_TYPE];
+	/* num_of_slave's lists, one per slave */
+	struct slave_list *slave_list;
+	struct resource_allocator res_alloc[MLX4_NUM_OF_RESOURCE_TYPE];
+};
+
+#define SLAVE_EVENT_EQ_SIZE	128
+struct mlx4_slave_event_eq {
+	u32 eqn;
+	u32 cons;
+	u32 prod;
+	spinlock_t event_lock;
+	struct mlx4_eqe event_eqe[SLAVE_EVENT_EQ_SIZE];
+};
+
+struct mlx4_qos_manager {
+	int num_of_qos_vfs;
+	DECLARE_BITMAP(priority_bm, MLX4_NUM_UP);
+};
+
+struct mlx4_master_qp0_state {
+	int proxy_qp0_active;
+	int qp0_active;
+	int port_active;
+};
+
+struct mlx4_mfunc_master_ctx {
+	struct mlx4_slave_state *slave_state;
+	struct mlx4_vf_admin_state *vf_admin;
+	struct mlx4_vf_oper_state *vf_oper;
+	struct mlx4_master_qp0_state qp0_state[MLX4_MAX_PORTS + 1];
+	int			init_port_ref[MLX4_MAX_PORTS + 1];
+	u16			max_mtu[MLX4_MAX_PORTS + 1];
+	u16			max_user_mtu[MLX4_MAX_PORTS + 1];
+	u8			pptx;
+	u8			pprx;
+	int			disable_mcast_ref[MLX4_MAX_PORTS + 1];
+	struct mlx4_resource_tracker res_tracker;
+	struct workqueue_struct *comm_wq;
+	struct work_struct	comm_work;
+	struct work_struct	slave_event_work;
+	struct work_struct	slave_flr_event_work;
+	spinlock_t		slave_state_lock;
+	__be32			comm_arm_bit_vector[4];
+	struct mlx4_eqe		cmd_eqe;
+	struct mlx4_slave_event_eq slave_eq;
+	struct mutex		gen_eqe_mutex[MLX4_MFUNC_MAX];
+	struct mlx4_qos_manager qos_ctl[MLX4_MAX_PORTS + 1];
+	u32			next_slave; /* mlx4_master_comm_channel */
+};
+
+struct mlx4_mfunc {
+	struct mlx4_comm __iomem       *comm;
+	struct mlx4_vhcr_cmd	       *vhcr;
+	dma_addr_t			vhcr_dma;
+
+	struct mlx4_mfunc_master_ctx	master;
+};
+
+#define MGM_QPN_MASK       0x00FFFFFF
+#define MGM_BLCK_LB_BIT    30
+
+struct mlx4_mgm {
+	__be32			next_gid_index;
+	__be32			members_count;
+	u32			reserved[2];
+	u8			gid[16];
+	__be32			qp[MLX4_MAX_QP_PER_MGM];
+};
+
+struct mlx4_cmd {
+	struct dma_pool	       *pool;
+	void __iomem	       *hcr;
+	struct mutex		slave_cmd_mutex;
+	struct semaphore	poll_sem;
+	struct semaphore	event_sem;
+	struct rw_semaphore	switch_sem;
+	int			max_cmds;
+	spinlock_t		context_lock;
+	int			free_head;
+	struct mlx4_cmd_context *context;
+	u16			token_mask;
+	u8			use_events;
+	u8			toggle;
+	u8			comm_toggle;
+	u8			initialized;
+};
+
+enum {
+	MLX4_VF_IMMED_VLAN_FLAG_VLAN = 1 << 0,
+	MLX4_VF_IMMED_VLAN_FLAG_QOS = 1 << 1,
+	MLX4_VF_IMMED_VLAN_FLAG_LINK_DISABLE = 1 << 2,
+};
+struct mlx4_vf_immed_vlan_work {
+	struct work_struct	work;
+	struct mlx4_priv	*priv;
+	int			flags;
+	int			slave;
+	int			vlan_ix;
+	int			orig_vlan_ix;
+	u8			port;
+	u8			qos;
+	u8                      qos_vport;
+	u16			vlan_id;
+	u16			orig_vlan_id;
+	__be16			vlan_proto;
+};
+
+
+struct mlx4_uar_table {
+	struct mlx4_bitmap	bitmap;
+};
+
+struct mlx4_mr_table {
+	struct mlx4_bitmap	mpt_bitmap;
+	struct mlx4_buddy	mtt_buddy;
+	u64			mtt_base;
+	u64			mpt_base;
+	struct mlx4_icm_table	mtt_table;
+	struct mlx4_icm_table	dmpt_table;
+};
+
+struct mlx4_cq_table {
+	struct mlx4_bitmap	bitmap;
+	spinlock_t		lock;
+	struct radix_tree_root	tree;
+	struct mlx4_icm_table	table;
+	struct mlx4_icm_table	cmpt_table;
+};
+
+struct mlx4_eq_table {
+	struct mlx4_bitmap	bitmap;
+	char		       *irq_names;
+	void __iomem	       *clr_int;
+	void __iomem	      **uar_map;
+	u32			clr_mask;
+	struct mlx4_eq	       *eq;
+	struct mlx4_icm_table	table;
+	struct mlx4_icm_table	cmpt_table;
+	int			have_irq;
+	u8			inta_pin;
+};
+
+struct mlx4_srq_table {
+	struct mlx4_bitmap	bitmap;
+	spinlock_t		lock;
+	struct radix_tree_root	tree;
+	struct mlx4_icm_table	table;
+	struct mlx4_icm_table	cmpt_table;
+};
+
+enum mlx4_qp_table_zones {
+	MLX4_QP_TABLE_ZONE_GENERAL,
+	MLX4_QP_TABLE_ZONE_RSS,
+	MLX4_QP_TABLE_ZONE_RAW_ETH,
+	MLX4_QP_TABLE_ZONE_NUM
+};
+
+struct mlx4_qp_table {
+	struct mlx4_bitmap	*bitmap_gen;
+	struct mlx4_zone_allocator *zones;
+	u32			zones_uids[MLX4_QP_TABLE_ZONE_NUM];
+	u32			rdmarc_base;
+	int			rdmarc_shift;
+	spinlock_t		lock;
+	struct mlx4_icm_table	qp_table;
+	struct mlx4_icm_table	auxc_table;
+	struct mlx4_icm_table	altc_table;
+	struct mlx4_icm_table	rdmarc_table;
+	struct mlx4_icm_table	cmpt_table;
+};
+
+struct mlx4_mcg_table {
+	struct mutex		mutex;
+	struct mlx4_bitmap	bitmap;
+	struct mlx4_icm_table	table;
+};
+
+struct mlx4_catas_err {
+	u32 __iomem	       *map;
+	struct timer_list	timer;
+	struct list_head	list;
+};
+
+#define MLX4_MAX_MAC_NUM	128
+#define MLX4_MAC_TABLE_SIZE	(MLX4_MAX_MAC_NUM << 3)
+
+struct mlx4_mac_table {
+	__be64			entries[MLX4_MAX_MAC_NUM];
+	int			refs[MLX4_MAX_MAC_NUM];
+	bool			is_dup[MLX4_MAX_MAC_NUM];
+	struct mutex		mutex;
+	int			total;
+	int			max;
+};
+
+#define MLX4_ROCE_GID_ENTRY_SIZE	16
+
+struct mlx4_roce_gid_entry {
+	u8 raw[MLX4_ROCE_GID_ENTRY_SIZE];
+};
+
+struct mlx4_roce_gid_table {
+	struct mlx4_roce_gid_entry	roce_gids[MLX4_ROCE_MAX_GIDS];
+	struct mutex			mutex;
+};
+
+#define MLX4_MAX_VLAN_NUM	128
+#define MLX4_VLAN_TABLE_SIZE	(MLX4_MAX_VLAN_NUM << 2)
+
+struct mlx4_vlan_table {
+	__be32			entries[MLX4_MAX_VLAN_NUM];
+	int			refs[MLX4_MAX_VLAN_NUM];
+	int			is_dup[MLX4_MAX_VLAN_NUM];
+	struct mutex		mutex;
+	int			total;
+	int			max;
+};
+
+#define SET_PORT_GEN_ALL_VALID	(MLX4_FLAG_V_MTU_MASK	| \
+				 MLX4_FLAG_V_PPRX_MASK	| \
+				 MLX4_FLAG_V_PPTX_MASK)
+#define SET_PORT_PROMISC_SHIFT		31
+#define SET_PORT_MC_PROMISC_SHIFT	30
+
+enum {
+	MCAST_DIRECT_ONLY	= 0,
+	MCAST_DIRECT		= 1,
+	MCAST_DEFAULT		= 2
+};
+
+
+struct mlx4_set_port_general_context {
+	u16 reserved1;
+	u8 flags2;
+	u8 flags;
+	union {
+		u8 ignore_fcs;
+		u8 roce_mode;
+	};
+	u8 reserved2;
+	__be16 mtu;
+	u8 pptx;
+	u8 pfctx;
+	u16 reserved3;
+	u8 pprx;
+	u8 pfcrx;
+	u16 reserved4;
+	u32 reserved5;
+	u8 phv_en;
+	u8 reserved6[5];
+	__be16 user_mtu;
+	u16 reserved7;
+	u8 user_mac[6];
+};
+
+struct mlx4_set_port_rqp_calc_context {
+	__be32 base_qpn;
+	u8 rererved;
+	u8 n_mac;
+	u8 n_vlan;
+	u8 n_prio;
+	u8 reserved2[3];
+	u8 mac_miss;
+	u8 intra_no_vlan;
+	u8 no_vlan;
+	u8 intra_vlan_miss;
+	u8 vlan_miss;
+	u8 reserved3[3];
+	u8 no_vlan_prio;
+	__be32 promisc;
+	__be32 mcast;
+};
+
+struct mlx4_port_info {
+	struct mlx4_dev	       *dev;
+	int			port;
+	char			dev_name[16];
+	struct device_attribute port_attr;
+	enum mlx4_port_type	tmp_type;
+	char			dev_mtu_name[16];
+	struct device_attribute port_mtu_attr;
+	struct mlx4_mac_table	mac_table;
+	struct mlx4_vlan_table	vlan_table;
+	struct mlx4_roce_gid_table gid_table;
+	int			base_qpn;
+	struct cpu_rmap		*rmap;
+	struct devlink_port	devlink_port;
+};
+
+struct mlx4_sense {
+	struct mlx4_dev		*dev;
+	u8			do_sense_port[MLX4_MAX_PORTS + 1];
+	u8			sense_allowed[MLX4_MAX_PORTS + 1];
+	struct delayed_work	sense_poll;
+};
+
+struct mlx4_msix_ctl {
+	DECLARE_BITMAP(pool_bm, MAX_MSIX);
+	struct mutex	pool_lock;
+};
+
+struct mlx4_steer {
+	struct list_head promisc_qps[MLX4_NUM_STEERS];
+	struct list_head steer_entries[MLX4_NUM_STEERS];
+};
+
+enum {
+	MLX4_PCI_DEV_IS_VF		= 1 << 0,
+	MLX4_PCI_DEV_FORCE_SENSE_PORT	= 1 << 1,
+};
+
+enum {
+	MLX4_NO_RR	= 0,
+	MLX4_USE_RR	= 1,
+};
+
+struct mlx4_priv {
+	struct mlx4_dev		dev;
+
+	struct list_head	dev_list;
+	struct list_head	ctx_list;
+	spinlock_t		ctx_lock;
+
+	int			pci_dev_data;
+	int                     removed;
+
+	struct list_head        pgdir_list;
+	struct mutex            pgdir_mutex;
+
+	struct mlx4_fw		fw;
+	struct mlx4_cmd		cmd;
+	struct mlx4_mfunc	mfunc;
+
+	struct mlx4_bitmap	pd_bitmap;
+	struct mlx4_bitmap	xrcd_bitmap;
+	struct mlx4_uar_table	uar_table;
+	struct mlx4_mr_table	mr_table;
+	struct mlx4_cq_table	cq_table;
+	struct mlx4_eq_table	eq_table;
+	struct mlx4_srq_table	srq_table;
+	struct mlx4_qp_table	qp_table;
+	struct mlx4_mcg_table	mcg_table;
+	struct mlx4_bitmap	counters_bitmap;
+	int			def_counter[MLX4_MAX_PORTS];
+
+	struct mlx4_catas_err	catas_err;
+
+	void __iomem	       *clr_base;
+
+	struct mlx4_uar		driver_uar;
+	void __iomem	       *kar;
+	struct mlx4_port_info	port[MLX4_MAX_PORTS + 1];
+	struct mlx4_sense       sense;
+	struct mutex		port_mutex;
+	struct mlx4_msix_ctl	msix_ctl;
+	struct mlx4_steer	*steer;
+	struct list_head	bf_list;
+	struct mutex		bf_mutex;
+	struct io_mapping	*bf_mapping;
+	void __iomem            *clock_mapping;
+	int			reserved_mtts;
+	int			fs_hash_mode;
+	u8 virt2phys_pkey[MLX4_MFUNC_MAX][MLX4_MAX_PORTS][MLX4_MAX_PORT_PKEYS];
+	struct mlx4_port_map	v2p; /* cached port mapping configuration */
+	struct mutex		bond_mutex; /* for bond mode */
+	__be64			slave_node_guids[MLX4_MFUNC_MAX];
+
+	atomic_t		opreq_count;
+	struct work_struct	opreq_task;
+};
+
+static inline struct mlx4_priv *mlx4_priv(struct mlx4_dev *dev)
+{
+	return container_of(dev, struct mlx4_priv, dev);
+}
+
+#define MLX4_SENSE_RANGE	(HZ * 3)
+
+extern struct workqueue_struct *mlx4_wq;
+
+u32 mlx4_bitmap_alloc(struct mlx4_bitmap *bitmap);
+void mlx4_bitmap_free(struct mlx4_bitmap *bitmap, u32 obj, int use_rr);
+u32 mlx4_bitmap_alloc_range(struct mlx4_bitmap *bitmap, int cnt,
+			    int align, u32 skip_mask);
+void mlx4_bitmap_free_range(struct mlx4_bitmap *bitmap, u32 obj, int cnt,
+			    int use_rr);
+u32 mlx4_bitmap_avail(struct mlx4_bitmap *bitmap);
+int mlx4_bitmap_init(struct mlx4_bitmap *bitmap, u32 num, u32 mask,
+		     u32 reserved_bot, u32 resetrved_top);
+void mlx4_bitmap_cleanup(struct mlx4_bitmap *bitmap);
+
+int mlx4_reset(struct mlx4_dev *dev);
+
+int mlx4_alloc_eq_table(struct mlx4_dev *dev);
+void mlx4_free_eq_table(struct mlx4_dev *dev);
+
+int mlx4_init_pd_table(struct mlx4_dev *dev);
+int mlx4_init_xrcd_table(struct mlx4_dev *dev);
+int mlx4_init_uar_table(struct mlx4_dev *dev);
+int mlx4_init_mr_table(struct mlx4_dev *dev);
+int mlx4_init_eq_table(struct mlx4_dev *dev);
+int mlx4_init_cq_table(struct mlx4_dev *dev);
+int mlx4_init_qp_table(struct mlx4_dev *dev);
+int mlx4_init_srq_table(struct mlx4_dev *dev);
+int mlx4_init_mcg_table(struct mlx4_dev *dev);
+
+void mlx4_cleanup_pd_table(struct mlx4_dev *dev);
+void mlx4_cleanup_xrcd_table(struct mlx4_dev *dev);
+void mlx4_cleanup_uar_table(struct mlx4_dev *dev);
+void mlx4_cleanup_mr_table(struct mlx4_dev *dev);
+void mlx4_cleanup_eq_table(struct mlx4_dev *dev);
+void mlx4_cleanup_cq_table(struct mlx4_dev *dev);
+void mlx4_cleanup_qp_table(struct mlx4_dev *dev);
+void mlx4_cleanup_srq_table(struct mlx4_dev *dev);
+void mlx4_cleanup_mcg_table(struct mlx4_dev *dev);
+int __mlx4_qp_alloc_icm(struct mlx4_dev *dev, int qpn);
+void __mlx4_qp_free_icm(struct mlx4_dev *dev, int qpn);
+int __mlx4_cq_alloc_icm(struct mlx4_dev *dev, int *cqn);
+void __mlx4_cq_free_icm(struct mlx4_dev *dev, int cqn);
+int __mlx4_srq_alloc_icm(struct mlx4_dev *dev, int *srqn);
+void __mlx4_srq_free_icm(struct mlx4_dev *dev, int srqn);
+int __mlx4_mpt_reserve(struct mlx4_dev *dev);
+void __mlx4_mpt_release(struct mlx4_dev *dev, u32 index);
+int __mlx4_mpt_alloc_icm(struct mlx4_dev *dev, u32 index);
+void __mlx4_mpt_free_icm(struct mlx4_dev *dev, u32 index);
+u32 __mlx4_alloc_mtt_range(struct mlx4_dev *dev, int order);
+void __mlx4_free_mtt_range(struct mlx4_dev *dev, u32 first_seg, int order);
+
+int mlx4_WRITE_MTT_wrapper(struct mlx4_dev *dev, int slave,
+			   struct mlx4_vhcr *vhcr,
+			   struct mlx4_cmd_mailbox *inbox,
+			   struct mlx4_cmd_mailbox *outbox,
+			   struct mlx4_cmd_info *cmd);
+int mlx4_SYNC_TPT_wrapper(struct mlx4_dev *dev, int slave,
+			   struct mlx4_vhcr *vhcr,
+			   struct mlx4_cmd_mailbox *inbox,
+			   struct mlx4_cmd_mailbox *outbox,
+			   struct mlx4_cmd_info *cmd);
+int mlx4_SW2HW_MPT_wrapper(struct mlx4_dev *dev, int slave,
+			   struct mlx4_vhcr *vhcr,
+			   struct mlx4_cmd_mailbox *inbox,
+			   struct mlx4_cmd_mailbox *outbox,
+			   struct mlx4_cmd_info *cmd);
+int mlx4_HW2SW_MPT_wrapper(struct mlx4_dev *dev, int slave,
+			   struct mlx4_vhcr *vhcr,
+			   struct mlx4_cmd_mailbox *inbox,
+			   struct mlx4_cmd_mailbox *outbox,
+			   struct mlx4_cmd_info *cmd);
+int mlx4_QUERY_MPT_wrapper(struct mlx4_dev *dev, int slave,
+			   struct mlx4_vhcr *vhcr,
+			   struct mlx4_cmd_mailbox *inbox,
+			   struct mlx4_cmd_mailbox *outbox,
+			   struct mlx4_cmd_info *cmd);
+int mlx4_SW2HW_EQ_wrapper(struct mlx4_dev *dev, int slave,
+			  struct mlx4_vhcr *vhcr,
+			  struct mlx4_cmd_mailbox *inbox,
+			  struct mlx4_cmd_mailbox *outbox,
+			  struct mlx4_cmd_info *cmd);
+int mlx4_CONFIG_DEV_wrapper(struct mlx4_dev *dev, int slave,
+			    struct mlx4_vhcr *vhcr,
+			    struct mlx4_cmd_mailbox *inbox,
+			    struct mlx4_cmd_mailbox *outbox,
+			    struct mlx4_cmd_info *cmd);
+int mlx4_DMA_wrapper(struct mlx4_dev *dev, int slave,
+		     struct mlx4_vhcr *vhcr,
+		     struct mlx4_cmd_mailbox *inbox,
+		     struct mlx4_cmd_mailbox *outbox,
+		     struct mlx4_cmd_info *cmd);
+int __mlx4_qp_reserve_range(struct mlx4_dev *dev, int cnt, int align,
+			    int *base, u8 flags);
+void __mlx4_qp_release_range(struct mlx4_dev *dev, int base_qpn, int cnt);
+int __mlx4_register_mac(struct mlx4_dev *dev, u8 port, u64 mac);
+void __mlx4_unregister_mac(struct mlx4_dev *dev, u8 port, u64 mac);
+int __mlx4_write_mtt(struct mlx4_dev *dev, struct mlx4_mtt *mtt,
+		     int start_index, int npages, u64 *page_list);
+int __mlx4_counter_alloc(struct mlx4_dev *dev, u32 *idx);
+void __mlx4_counter_free(struct mlx4_dev *dev, u32 idx);
+int mlx4_calc_vf_counters(struct mlx4_dev *dev, int slave, int port,
+			  struct mlx4_counter *data);
+int __mlx4_xrcd_alloc(struct mlx4_dev *dev, u32 *xrcdn);
+void __mlx4_xrcd_free(struct mlx4_dev *dev, u32 xrcdn);
+
+void mlx4_start_catas_poll(struct mlx4_dev *dev);
+void mlx4_stop_catas_poll(struct mlx4_dev *dev);
+int mlx4_catas_init(struct mlx4_dev *dev);
+void mlx4_catas_end(struct mlx4_dev *dev);
+int mlx4_crdump_init(struct mlx4_dev *dev);
+void mlx4_crdump_end(struct mlx4_dev *dev);
+int mlx4_restart_one(struct pci_dev *pdev);
+int mlx4_register_device(struct mlx4_dev *dev);
+void mlx4_unregister_device(struct mlx4_dev *dev);
+void mlx4_dispatch_event(struct mlx4_dev *dev, enum mlx4_dev_event type,
+			 unsigned long param);
+
+struct mlx4_dev_cap;
+struct mlx4_init_hca_param;
+
+u64 mlx4_make_profile(struct mlx4_dev *dev,
+		      struct mlx4_profile *request,
+		      struct mlx4_dev_cap *dev_cap,
+		      struct mlx4_init_hca_param *init_hca);
+void mlx4_master_comm_channel(struct work_struct *work);
+void mlx4_gen_slave_eqe(struct work_struct *work);
+void mlx4_master_handle_slave_flr(struct work_struct *work);
+
+int mlx4_ALLOC_RES_wrapper(struct mlx4_dev *dev, int slave,
+			   struct mlx4_vhcr *vhcr,
+			   struct mlx4_cmd_mailbox *inbox,
+			   struct mlx4_cmd_mailbox *outbox,
+			   struct mlx4_cmd_info *cmd);
+int mlx4_FREE_RES_wrapper(struct mlx4_dev *dev, int slave,
+			  struct mlx4_vhcr *vhcr,
+			  struct mlx4_cmd_mailbox *inbox,
+			  struct mlx4_cmd_mailbox *outbox,
+			  struct mlx4_cmd_info *cmd);
+int mlx4_MAP_EQ_wrapper(struct mlx4_dev *dev, int slave,
+			struct mlx4_vhcr *vhcr, struct mlx4_cmd_mailbox *inbox,
+			struct mlx4_cmd_mailbox *outbox,
+			struct mlx4_cmd_info *cmd);
+int mlx4_COMM_INT_wrapper(struct mlx4_dev *dev, int slave,
+			  struct mlx4_vhcr *vhcr,
+			  struct mlx4_cmd_mailbox *inbox,
+			  struct mlx4_cmd_mailbox *outbox,
+			  struct mlx4_cmd_info *cmd);
+int mlx4_HW2SW_EQ_wrapper(struct mlx4_dev *dev, int slave,
+			    struct mlx4_vhcr *vhcr,
+			    struct mlx4_cmd_mailbox *inbox,
+			    struct mlx4_cmd_mailbox *outbox,
+			  struct mlx4_cmd_info *cmd);
+int mlx4_QUERY_EQ_wrapper(struct mlx4_dev *dev, int slave,
+			  struct mlx4_vhcr *vhcr,
+			  struct mlx4_cmd_mailbox *inbox,
+			  struct mlx4_cmd_mailbox *outbox,
+			  struct mlx4_cmd_info *cmd);
+int mlx4_SW2HW_CQ_wrapper(struct mlx4_dev *dev, int slave,
+			  struct mlx4_vhcr *vhcr,
+			  struct mlx4_cmd_mailbox *inbox,
+			  struct mlx4_cmd_mailbox *outbox,
+			  struct mlx4_cmd_info *cmd);
+int mlx4_HW2SW_CQ_wrapper(struct mlx4_dev *dev, int slave,
+			  struct mlx4_vhcr *vhcr,
+			  struct mlx4_cmd_mailbox *inbox,
+			  struct mlx4_cmd_mailbox *outbox,
+			  struct mlx4_cmd_info *cmd);
+int mlx4_QUERY_CQ_wrapper(struct mlx4_dev *dev, int slave,
+			  struct mlx4_vhcr *vhcr,
+			  struct mlx4_cmd_mailbox *inbox,
+			  struct mlx4_cmd_mailbox *outbox,
+			  struct mlx4_cmd_info *cmd);
+int mlx4_MODIFY_CQ_wrapper(struct mlx4_dev *dev, int slave,
+			  struct mlx4_vhcr *vhcr,
+			  struct mlx4_cmd_mailbox *inbox,
+			  struct mlx4_cmd_mailbox *outbox,
+			   struct mlx4_cmd_info *cmd);
+int mlx4_SW2HW_SRQ_wrapper(struct mlx4_dev *dev, int slave,
+			   struct mlx4_vhcr *vhcr,
+			   struct mlx4_cmd_mailbox *inbox,
+			   struct mlx4_cmd_mailbox *outbox,
+			   struct mlx4_cmd_info *cmd);
+int mlx4_HW2SW_SRQ_wrapper(struct mlx4_dev *dev, int slave,
+			   struct mlx4_vhcr *vhcr,
+			   struct mlx4_cmd_mailbox *inbox,
+			   struct mlx4_cmd_mailbox *outbox,
+			   struct mlx4_cmd_info *cmd);
+int mlx4_QUERY_SRQ_wrapper(struct mlx4_dev *dev, int slave,
+			   struct mlx4_vhcr *vhcr,
+			   struct mlx4_cmd_mailbox *inbox,
+			   struct mlx4_cmd_mailbox *outbox,
+			   struct mlx4_cmd_info *cmd);
+int mlx4_ARM_SRQ_wrapper(struct mlx4_dev *dev, int slave,
+			 struct mlx4_vhcr *vhcr,
+			 struct mlx4_cmd_mailbox *inbox,
+			 struct mlx4_cmd_mailbox *outbox,
+			 struct mlx4_cmd_info *cmd);
+int mlx4_GEN_QP_wrapper(struct mlx4_dev *dev, int slave,
+			struct mlx4_vhcr *vhcr,
+			struct mlx4_cmd_mailbox *inbox,
+			struct mlx4_cmd_mailbox *outbox,
+			struct mlx4_cmd_info *cmd);
+int mlx4_RST2INIT_QP_wrapper(struct mlx4_dev *dev, int slave,
+			     struct mlx4_vhcr *vhcr,
+			     struct mlx4_cmd_mailbox *inbox,
+			     struct mlx4_cmd_mailbox *outbox,
+			     struct mlx4_cmd_info *cmd);
+int mlx4_INIT2INIT_QP_wrapper(struct mlx4_dev *dev, int slave,
+			      struct mlx4_vhcr *vhcr,
+			      struct mlx4_cmd_mailbox *inbox,
+			      struct mlx4_cmd_mailbox *outbox,
+			      struct mlx4_cmd_info *cmd);
+int mlx4_INIT2RTR_QP_wrapper(struct mlx4_dev *dev, int slave,
+			     struct mlx4_vhcr *vhcr,
+			     struct mlx4_cmd_mailbox *inbox,
+			     struct mlx4_cmd_mailbox *outbox,
+			     struct mlx4_cmd_info *cmd);
+int mlx4_RTR2RTS_QP_wrapper(struct mlx4_dev *dev, int slave,
+			    struct mlx4_vhcr *vhcr,
+			    struct mlx4_cmd_mailbox *inbox,
+			    struct mlx4_cmd_mailbox *outbox,
+			    struct mlx4_cmd_info *cmd);
+int mlx4_RTS2RTS_QP_wrapper(struct mlx4_dev *dev, int slave,
+			    struct mlx4_vhcr *vhcr,
+			    struct mlx4_cmd_mailbox *inbox,
+			    struct mlx4_cmd_mailbox *outbox,
+			    struct mlx4_cmd_info *cmd);
+int mlx4_SQERR2RTS_QP_wrapper(struct mlx4_dev *dev, int slave,
+			      struct mlx4_vhcr *vhcr,
+			      struct mlx4_cmd_mailbox *inbox,
+			      struct mlx4_cmd_mailbox *outbox,
+			      struct mlx4_cmd_info *cmd);
+int mlx4_2ERR_QP_wrapper(struct mlx4_dev *dev, int slave,
+			 struct mlx4_vhcr *vhcr,
+			 struct mlx4_cmd_mailbox *inbox,
+			 struct mlx4_cmd_mailbox *outbox,
+			 struct mlx4_cmd_info *cmd);
+int mlx4_RTS2SQD_QP_wrapper(struct mlx4_dev *dev, int slave,
+			    struct mlx4_vhcr *vhcr,
+			    struct mlx4_cmd_mailbox *inbox,
+			    struct mlx4_cmd_mailbox *outbox,
+			    struct mlx4_cmd_info *cmd);
+int mlx4_SQD2SQD_QP_wrapper(struct mlx4_dev *dev, int slave,
+			    struct mlx4_vhcr *vhcr,
+			    struct mlx4_cmd_mailbox *inbox,
+			    struct mlx4_cmd_mailbox *outbox,
+			    struct mlx4_cmd_info *cmd);
+int mlx4_SQD2RTS_QP_wrapper(struct mlx4_dev *dev, int slave,
+			    struct mlx4_vhcr *vhcr,
+			    struct mlx4_cmd_mailbox *inbox,
+			    struct mlx4_cmd_mailbox *outbox,
+			    struct mlx4_cmd_info *cmd);
+int mlx4_2RST_QP_wrapper(struct mlx4_dev *dev, int slave,
+			 struct mlx4_vhcr *vhcr,
+			 struct mlx4_cmd_mailbox *inbox,
+			 struct mlx4_cmd_mailbox *outbox,
+			 struct mlx4_cmd_info *cmd);
+int mlx4_QUERY_QP_wrapper(struct mlx4_dev *dev, int slave,
+			  struct mlx4_vhcr *vhcr,
+			  struct mlx4_cmd_mailbox *inbox,
+			  struct mlx4_cmd_mailbox *outbox,
+			  struct mlx4_cmd_info *cmd);
+
+int mlx4_GEN_EQE(struct mlx4_dev *dev, int slave, struct mlx4_eqe *eqe);
+
+enum {
+	MLX4_CMD_CLEANUP_STRUCT = 1UL << 0,
+	MLX4_CMD_CLEANUP_POOL	= 1UL << 1,
+	MLX4_CMD_CLEANUP_HCR	= 1UL << 2,
+	MLX4_CMD_CLEANUP_VHCR	= 1UL << 3,
+	MLX4_CMD_CLEANUP_ALL	= (MLX4_CMD_CLEANUP_VHCR << 1) - 1
+};
+
+int mlx4_cmd_init(struct mlx4_dev *dev);
+void mlx4_cmd_cleanup(struct mlx4_dev *dev, int cleanup_mask);
+int mlx4_multi_func_init(struct mlx4_dev *dev);
+int mlx4_ARM_COMM_CHANNEL(struct mlx4_dev *dev);
+void mlx4_multi_func_cleanup(struct mlx4_dev *dev);
+void mlx4_cmd_event(struct mlx4_dev *dev, u16 token, u8 status, u64 out_param);
+int mlx4_cmd_use_events(struct mlx4_dev *dev);
+void mlx4_cmd_use_polling(struct mlx4_dev *dev);
+
+int mlx4_comm_cmd(struct mlx4_dev *dev, u8 cmd, u16 param,
+		  u16 op, unsigned long timeout);
+
+void mlx4_cq_tasklet_cb(struct tasklet_struct *t);
+void mlx4_cq_completion(struct mlx4_dev *dev, u32 cqn);
+void mlx4_cq_event(struct mlx4_dev *dev, u32 cqn, int event_type);
+
+void mlx4_qp_event(struct mlx4_dev *dev, u32 qpn, int event_type);
+
+void mlx4_srq_event(struct mlx4_dev *dev, u32 srqn, int event_type);
+
+void mlx4_enter_error_state(struct mlx4_dev_persistent *persist);
+int mlx4_comm_internal_err(u32 slave_read);
+
+int mlx4_crdump_collect(struct mlx4_dev *dev);
+
+int mlx4_SENSE_PORT(struct mlx4_dev *dev, int port,
+		    enum mlx4_port_type *type);
+void mlx4_do_sense_ports(struct mlx4_dev *dev,
+			 enum mlx4_port_type *stype,
+			 enum mlx4_port_type *defaults);
+void mlx4_start_sense(struct mlx4_dev *dev);
+void mlx4_stop_sense(struct mlx4_dev *dev);
+void mlx4_sense_init(struct mlx4_dev *dev);
+int mlx4_check_port_params(struct mlx4_dev *dev,
+			   enum mlx4_port_type *port_type);
+int mlx4_change_port_types(struct mlx4_dev *dev,
+			   enum mlx4_port_type *port_types);
+
+void mlx4_init_mac_table(struct mlx4_dev *dev, struct mlx4_mac_table *table);
+void mlx4_init_vlan_table(struct mlx4_dev *dev, struct mlx4_vlan_table *table);
+void mlx4_init_roce_gid_table(struct mlx4_dev *dev,
+			      struct mlx4_roce_gid_table *table);
+void __mlx4_unregister_vlan(struct mlx4_dev *dev, u8 port, u16 vlan);
+int __mlx4_register_vlan(struct mlx4_dev *dev, u8 port, u16 vlan, int *index);
+int mlx4_bond_vlan_table(struct mlx4_dev *dev);
+int mlx4_unbond_vlan_table(struct mlx4_dev *dev);
+int mlx4_bond_mac_table(struct mlx4_dev *dev);
+int mlx4_unbond_mac_table(struct mlx4_dev *dev);
+
+int mlx4_SET_PORT(struct mlx4_dev *dev, u8 port, int pkey_tbl_sz);
+/* resource tracker functions*/
+int mlx4_get_slave_from_resource_id(struct mlx4_dev *dev,
+				    enum mlx4_resource resource_type,
+				    u64 resource_id, int *slave);
+void mlx4_delete_all_resources_for_slave(struct mlx4_dev *dev, int slave_id);
+void mlx4_reset_roce_gids(struct mlx4_dev *dev, int slave);
+int mlx4_init_resource_tracker(struct mlx4_dev *dev);
+
+void mlx4_free_resource_tracker(struct mlx4_dev *dev,
+				enum mlx4_res_tracker_free_type type);
+
+int mlx4_QUERY_FW_wrapper(struct mlx4_dev *dev, int slave,
+			  struct mlx4_vhcr *vhcr,
+			  struct mlx4_cmd_mailbox *inbox,
+			  struct mlx4_cmd_mailbox *outbox,
+			  struct mlx4_cmd_info *cmd);
+int mlx4_SET_PORT_wrapper(struct mlx4_dev *dev, int slave,
+			  struct mlx4_vhcr *vhcr,
+			  struct mlx4_cmd_mailbox *inbox,
+			  struct mlx4_cmd_mailbox *outbox,
+			  struct mlx4_cmd_info *cmd);
+int mlx4_INIT_PORT_wrapper(struct mlx4_dev *dev, int slave,
+			   struct mlx4_vhcr *vhcr,
+			   struct mlx4_cmd_mailbox *inbox,
+			   struct mlx4_cmd_mailbox *outbox,
+			   struct mlx4_cmd_info *cmd);
+int mlx4_CLOSE_PORT_wrapper(struct mlx4_dev *dev, int slave,
+			    struct mlx4_vhcr *vhcr,
+			    struct mlx4_cmd_mailbox *inbox,
+			    struct mlx4_cmd_mailbox *outbox,
+			    struct mlx4_cmd_info *cmd);
+int mlx4_QUERY_DEV_CAP_wrapper(struct mlx4_dev *dev, int slave,
+			       struct mlx4_vhcr *vhcr,
+			       struct mlx4_cmd_mailbox *inbox,
+			       struct mlx4_cmd_mailbox *outbox,
+			       struct mlx4_cmd_info *cmd);
+int mlx4_QUERY_PORT_wrapper(struct mlx4_dev *dev, int slave,
+			    struct mlx4_vhcr *vhcr,
+			    struct mlx4_cmd_mailbox *inbox,
+			    struct mlx4_cmd_mailbox *outbox,
+			    struct mlx4_cmd_info *cmd);
+int mlx4_get_port_ib_caps(struct mlx4_dev *dev, u8 port, __be32 *caps);
+
+int mlx4_get_slave_pkey_gid_tbl_len(struct mlx4_dev *dev, u8 port,
+				    int *gid_tbl_len, int *pkey_tbl_len);
+
+int mlx4_QP_ATTACH_wrapper(struct mlx4_dev *dev, int slave,
+			   struct mlx4_vhcr *vhcr,
+			   struct mlx4_cmd_mailbox *inbox,
+			   struct mlx4_cmd_mailbox *outbox,
+			   struct mlx4_cmd_info *cmd);
+
+int mlx4_UPDATE_QP_wrapper(struct mlx4_dev *dev, int slave,
+			   struct mlx4_vhcr *vhcr,
+			   struct mlx4_cmd_mailbox *inbox,
+			   struct mlx4_cmd_mailbox *outbox,
+			   struct mlx4_cmd_info *cmd);
+
+int mlx4_PROMISC_wrapper(struct mlx4_dev *dev, int slave,
+			 struct mlx4_vhcr *vhcr,
+			 struct mlx4_cmd_mailbox *inbox,
+			 struct mlx4_cmd_mailbox *outbox,
+			 struct mlx4_cmd_info *cmd);
+int mlx4_qp_detach_common(struct mlx4_dev *dev, struct mlx4_qp *qp, u8 gid[16],
+			  enum mlx4_protocol prot, enum mlx4_steer_type steer);
+int mlx4_qp_attach_common(struct mlx4_dev *dev, struct mlx4_qp *qp, u8 gid[16],
+			  int block_mcast_loopback, enum mlx4_protocol prot,
+			  enum mlx4_steer_type steer);
+int mlx4_trans_to_dmfs_attach(struct mlx4_dev *dev, struct mlx4_qp *qp,
+			      u8 gid[16], u8 port,
+			      int block_mcast_loopback,
+			      enum mlx4_protocol prot, u64 *reg_id);
+int mlx4_SET_MCAST_FLTR_wrapper(struct mlx4_dev *dev, int slave,
+				struct mlx4_vhcr *vhcr,
+				struct mlx4_cmd_mailbox *inbox,
+				struct mlx4_cmd_mailbox *outbox,
+				struct mlx4_cmd_info *cmd);
+int mlx4_SET_VLAN_FLTR_wrapper(struct mlx4_dev *dev, int slave,
+			       struct mlx4_vhcr *vhcr,
+			       struct mlx4_cmd_mailbox *inbox,
+			       struct mlx4_cmd_mailbox *outbox,
+			       struct mlx4_cmd_info *cmd);
+int mlx4_common_set_vlan_fltr(struct mlx4_dev *dev, int function,
+				     int port, void *buf);
+int mlx4_DUMP_ETH_STATS_wrapper(struct mlx4_dev *dev, int slave,
+				   struct mlx4_vhcr *vhcr,
+				   struct mlx4_cmd_mailbox *inbox,
+				   struct mlx4_cmd_mailbox *outbox,
+				struct mlx4_cmd_info *cmd);
+int mlx4_PKEY_TABLE_wrapper(struct mlx4_dev *dev, int slave,
+			    struct mlx4_vhcr *vhcr,
+			    struct mlx4_cmd_mailbox *inbox,
+			    struct mlx4_cmd_mailbox *outbox,
+			    struct mlx4_cmd_info *cmd);
+int mlx4_QUERY_IF_STAT_wrapper(struct mlx4_dev *dev, int slave,
+			       struct mlx4_vhcr *vhcr,
+			       struct mlx4_cmd_mailbox *inbox,
+			       struct mlx4_cmd_mailbox *outbox,
+			       struct mlx4_cmd_info *cmd);
+int mlx4_QP_FLOW_STEERING_ATTACH_wrapper(struct mlx4_dev *dev, int slave,
+					 struct mlx4_vhcr *vhcr,
+					 struct mlx4_cmd_mailbox *inbox,
+					 struct mlx4_cmd_mailbox *outbox,
+					 struct mlx4_cmd_info *cmd);
+int mlx4_QP_FLOW_STEERING_DETACH_wrapper(struct mlx4_dev *dev, int slave,
+					 struct mlx4_vhcr *vhcr,
+					 struct mlx4_cmd_mailbox *inbox,
+					 struct mlx4_cmd_mailbox *outbox,
+					 struct mlx4_cmd_info *cmd);
+int mlx4_ACCESS_REG_wrapper(struct mlx4_dev *dev, int slave,
+			    struct mlx4_vhcr *vhcr,
+			    struct mlx4_cmd_mailbox *inbox,
+			    struct mlx4_cmd_mailbox *outbox,
+			    struct mlx4_cmd_info *cmd);
+
+int mlx4_get_mgm_entry_size(struct mlx4_dev *dev);
+int mlx4_get_qp_per_mgm(struct mlx4_dev *dev);
+
+static inline void set_param_l(u64 *arg, u32 val)
+{
+	*arg = (*arg & 0xffffffff00000000ULL) | (u64) val;
+}
+
+static inline void set_param_h(u64 *arg, u32 val)
+{
+	*arg = (*arg & 0xffffffff) | ((u64) val << 32);
+}
+
+static inline u32 get_param_l(u64 *arg)
+{
+	return (u32) (*arg & 0xffffffff);
+}
+
+static inline u32 get_param_h(u64 *arg)
+{
+	return (u32)(*arg >> 32);
+}
+
+static inline spinlock_t *mlx4_tlock(struct mlx4_dev *dev)
+{
+	return &mlx4_priv(dev)->mfunc.master.res_tracker.lock;
+}
+
+#define NOT_MASKED_PD_BITS 17
+
+void mlx4_vf_immed_vlan_work_handler(struct work_struct *_work);
+
+void mlx4_init_quotas(struct mlx4_dev *dev);
+
+/* for VFs, replace zero MACs with randomly-generated MACs at driver start */
+void mlx4_replace_zero_macs(struct mlx4_dev *dev);
+int mlx4_get_slave_num_gids(struct mlx4_dev *dev, int slave, int port);
+/* Returns the VF index of slave */
+int mlx4_get_vf_indx(struct mlx4_dev *dev, int slave);
+int mlx4_config_mad_demux(struct mlx4_dev *dev);
+int mlx4_do_bond(struct mlx4_dev *dev, bool enable);
+int mlx4_bond_fs_rules(struct mlx4_dev *dev);
+int mlx4_unbond_fs_rules(struct mlx4_dev *dev);
+
+enum mlx4_zone_flags {
+	MLX4_ZONE_ALLOW_ALLOC_FROM_LOWER_PRIO	= 1UL << 0,
+	MLX4_ZONE_ALLOW_ALLOC_FROM_EQ_PRIO	= 1UL << 1,
+	MLX4_ZONE_FALLBACK_TO_HIGHER_PRIO	= 1UL << 2,
+	MLX4_ZONE_USE_RR			= 1UL << 3,
+};
+
+enum mlx4_zone_alloc_flags {
+	/* No two objects could overlap between zones. UID
+	 * could be left unused. If this flag is given and
+	 * two overlapped zones are used, an object will be free'd
+	 * from the smallest possible matching zone.
+	 */
+	MLX4_ZONE_ALLOC_FLAGS_NO_OVERLAP	= 1UL << 0,
+};
+
+struct mlx4_zone_allocator;
+
+/* Create a new zone allocator */
+struct mlx4_zone_allocator *mlx4_zone_allocator_create(enum mlx4_zone_alloc_flags flags);
+
+/* Attach a mlx4_bitmap <bitmap> of priority <priority> to the zone allocator
+ * <zone_alloc>. Allocating an object from this zone adds an offset <offset>.
+ * Similarly, when searching for an object to free, this offset it taken into
+ * account. The use_rr mlx4_ib parameter for allocating objects from this <bitmap>
+ * is given through the MLX4_ZONE_USE_RR flag in <flags>.
+ * When an allocation fails, <zone_alloc> tries to allocate from other zones
+ * according to the policy set by <flags>. <puid> is the unique identifier
+ * received to this zone.
+ */
+int mlx4_zone_add_one(struct mlx4_zone_allocator *zone_alloc,
+		      struct mlx4_bitmap *bitmap,
+		      u32 flags,
+		      int priority,
+		      int offset,
+		      u32 *puid);
+
+/* Remove bitmap indicated by <uid> from <zone_alloc> */
+int mlx4_zone_remove_one(struct mlx4_zone_allocator *zone_alloc, u32 uid);
+
+/* Delete the zone allocator <zone_alloc. This function doesn't destroy
+ * the attached bitmaps.
+ */
+void mlx4_zone_allocator_destroy(struct mlx4_zone_allocator *zone_alloc);
+
+/* Allocate <count> objects with align <align> and skip_mask <skip_mask>
+ * from the mlx4_bitmap whose uid is <uid>. The bitmap which we actually
+ * allocated from is returned in <puid>. If the allocation fails, a negative
+ * number is returned. Otherwise, the offset of the first object is returned.
+ */
+u32 mlx4_zone_alloc_entries(struct mlx4_zone_allocator *zones, u32 uid, int count,
+			    int align, u32 skip_mask, u32 *puid);
+
+/* Free <count> objects, start from <obj> of the uid <uid> from zone_allocator
+ * <zones>.
+ */
+u32 mlx4_zone_free_entries(struct mlx4_zone_allocator *zones,
+			   u32 uid, u32 obj, u32 count);
+
+/* If <zones> was allocated with MLX4_ZONE_ALLOC_FLAGS_NO_OVERLAP, instead of
+ * specifying the uid when freeing an object, zone allocator could figure it by
+ * itself. Other parameters are similar to mlx4_zone_free.
+ */
+u32 mlx4_zone_free_entries_unique(struct mlx4_zone_allocator *zones, u32 obj, u32 count);
+
+/* Returns a pointer to mlx4_bitmap that was attached to <zones> with <uid> */
+struct mlx4_bitmap *mlx4_zone_get_bitmap(struct mlx4_zone_allocator *zones, u32 uid);
+
+#endif /* MLX4_H */
diff --git a/drivers/net/ethernet/mellanox/mlx4/mlx4_en.h b/drivers/net/ethernet/mellanox/mlx4/mlx4_en.h
new file mode 100644
index 000000000..e132ff4c8
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx4/mlx4_en.h
@@ -0,0 +1,837 @@
+/*
+ * Copyright (c) 2007 Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+
+#ifndef _MLX4_EN_H_
+#define _MLX4_EN_H_
+
+#include <linux/bitops.h>
+#include <linux/compiler.h>
+#include <linux/ethtool.h>
+#include <linux/list.h>
+#include <linux/mutex.h>
+#include <linux/netdevice.h>
+#include <linux/if_vlan.h>
+#include <linux/net_tstamp.h>
+#ifdef CONFIG_MLX4_EN_DCB
+#include <linux/dcbnl.h>
+#endif
+#include <linux/cpu_rmap.h>
+#include <linux/ptp_clock_kernel.h>
+#include <linux/irq.h>
+#include <net/xdp.h>
+
+#include <linux/mlx4/device.h>
+#include <linux/mlx4/qp.h>
+#include <linux/mlx4/cq.h>
+#include <linux/mlx4/srq.h>
+#include <linux/mlx4/doorbell.h>
+#include <linux/mlx4/cmd.h>
+
+#include "en_port.h"
+#include "mlx4_stats.h"
+
+#define DRV_NAME	"mlx4_en"
+#define DRV_VERSION	"4.0-0"
+
+#define MLX4_EN_MSG_LEVEL	(NETIF_MSG_LINK | NETIF_MSG_IFDOWN)
+
+/*
+ * Device constants
+ */
+
+
+#define MLX4_EN_PAGE_SHIFT	12
+#define MLX4_EN_PAGE_SIZE	(1 << MLX4_EN_PAGE_SHIFT)
+#define DEF_RX_RINGS		16
+#define MAX_RX_RINGS		128
+#define MIN_RX_RINGS		1
+#define LOG_TXBB_SIZE		6
+#define TXBB_SIZE		BIT(LOG_TXBB_SIZE)
+#define HEADROOM		(2048 / TXBB_SIZE + 1)
+#define STAMP_STRIDE		64
+#define STAMP_DWORDS		(STAMP_STRIDE / 4)
+#define STAMP_SHIFT		31
+#define STAMP_VAL		0x7fffffff
+#define STATS_DELAY		(HZ / 4)
+#define SERVICE_TASK_DELAY	(HZ / 4)
+#define MAX_NUM_OF_FS_RULES	256
+
+#define MLX4_EN_FILTER_HASH_SHIFT 4
+#define MLX4_EN_FILTER_EXPIRY_QUOTA 60
+
+/* Typical TSO descriptor with 16 gather entries is 352 bytes... */
+#define MAX_DESC_SIZE		512
+#define MAX_DESC_TXBBS		(MAX_DESC_SIZE / TXBB_SIZE)
+
+/*
+ * OS related constants and tunables
+ */
+
+#define MLX4_EN_PRIV_FLAGS_BLUEFLAME 1
+#define MLX4_EN_PRIV_FLAGS_PHV	     2
+
+#define MLX4_EN_WATCHDOG_TIMEOUT	(15 * HZ)
+
+/* Use the maximum between 16384 and a single page */
+#define MLX4_EN_ALLOC_SIZE	PAGE_ALIGN(16384)
+
+#define MLX4_EN_MAX_RX_FRAGS	4
+
+/* Maximum ring sizes */
+#define MLX4_EN_MAX_TX_SIZE	8192
+#define MLX4_EN_MAX_RX_SIZE	8192
+
+/* Minimum ring size for our page-allocation scheme to work */
+#define MLX4_EN_MIN_RX_SIZE	(MLX4_EN_ALLOC_SIZE / SMP_CACHE_BYTES)
+#define MLX4_EN_MIN_TX_SIZE	(4096 / TXBB_SIZE)
+
+#define MLX4_EN_SMALL_PKT_SIZE		64
+#define MLX4_EN_MIN_TX_RING_P_UP	1
+#define MLX4_EN_MAX_TX_RING_P_UP	32
+#define MLX4_EN_NUM_UP_LOW		1
+#define MLX4_EN_NUM_UP_HIGH		8
+#define MLX4_EN_DEF_RX_RING_SIZE  	1024
+#define MLX4_EN_DEF_TX_RING_SIZE	MLX4_EN_DEF_RX_RING_SIZE
+#define MAX_TX_RINGS			(MLX4_EN_MAX_TX_RING_P_UP * \
+					 MLX4_EN_NUM_UP_HIGH)
+
+#define MLX4_EN_DEFAULT_TX_WORK		256
+
+/* Target number of packets to coalesce with interrupt moderation */
+#define MLX4_EN_RX_COAL_TARGET	44
+#define MLX4_EN_RX_COAL_TIME	0x10
+
+#define MLX4_EN_TX_COAL_PKTS	16
+#define MLX4_EN_TX_COAL_TIME	0x10
+
+#define MLX4_EN_MAX_COAL_PKTS	U16_MAX
+#define MLX4_EN_MAX_COAL_TIME	U16_MAX
+
+#define MLX4_EN_RX_RATE_LOW		400000
+#define MLX4_EN_RX_COAL_TIME_LOW	0
+#define MLX4_EN_RX_RATE_HIGH		450000
+#define MLX4_EN_RX_COAL_TIME_HIGH	128
+#define MLX4_EN_RX_SIZE_THRESH		1024
+#define MLX4_EN_RX_RATE_THRESH		(1000000 / MLX4_EN_RX_COAL_TIME_HIGH)
+#define MLX4_EN_SAMPLE_INTERVAL		0
+#define MLX4_EN_AVG_PKT_SMALL		256
+
+#define MLX4_EN_AUTO_CONF	0xffff
+
+#define MLX4_EN_DEF_RX_PAUSE	1
+#define MLX4_EN_DEF_TX_PAUSE	1
+
+/* Interval between successive polls in the Tx routine when polling is used
+   instead of interrupts (in per-core Tx rings) - should be power of 2 */
+#define MLX4_EN_TX_POLL_MODER	16
+#define MLX4_EN_TX_POLL_TIMEOUT	(HZ / 4)
+
+#define SMALL_PACKET_SIZE      (256 - NET_IP_ALIGN)
+#define HEADER_COPY_SIZE       (128 - NET_IP_ALIGN)
+#define MLX4_LOOPBACK_TEST_PAYLOAD (HEADER_COPY_SIZE - ETH_HLEN)
+#define PREAMBLE_LEN           8
+#define MLX4_SELFTEST_LB_MIN_MTU (MLX4_LOOPBACK_TEST_PAYLOAD + NET_IP_ALIGN + \
+				  ETH_HLEN + PREAMBLE_LEN)
+
+/* VLAN_HLEN is added twice,to support skb vlan tagged with multiple
+ * headers. (For example: ETH_P_8021Q and ETH_P_8021AD).
+ */
+#define MLX4_EN_EFF_MTU(mtu)	((mtu) + ETH_HLEN + (2 * VLAN_HLEN))
+#define ETH_BCAST		0xffffffffffffULL
+
+#define MLX4_EN_LOOPBACK_RETRIES	5
+#define MLX4_EN_LOOPBACK_TIMEOUT	100
+
+/* Constants for TX flow */
+enum {
+	MAX_INLINE = 104, /* 128 - 16 - 4 - 4 */
+	MAX_BF = 256,
+	MIN_PKT_LEN = 17,
+};
+
+/*
+ * Configurables
+ */
+
+enum cq_type {
+	/* keep tx types first */
+	TX,
+	TX_XDP,
+#define MLX4_EN_NUM_TX_TYPES (TX_XDP + 1)
+	RX,
+};
+
+
+/*
+ * Useful macros
+ */
+#define ROUNDUP_LOG2(x)		ilog2(roundup_pow_of_two(x))
+#define XNOR(x, y)		(!(x) == !(y))
+
+
+struct mlx4_en_tx_info {
+	union {
+		struct sk_buff *skb;
+		struct page *page;
+	};
+	dma_addr_t	map0_dma;
+	u32		map0_byte_count;
+	u32		nr_txbb;
+	u32		nr_bytes;
+	u8		linear;
+	u8		data_offset;
+	u8		inl;
+	u8		ts_requested;
+	u8		nr_maps;
+} ____cacheline_aligned_in_smp;
+
+
+#define MLX4_EN_BIT_DESC_OWN	0x80000000
+#define CTRL_SIZE	sizeof(struct mlx4_wqe_ctrl_seg)
+#define MLX4_EN_MEMTYPE_PAD	0x100
+#define DS_SIZE		sizeof(struct mlx4_wqe_data_seg)
+
+
+struct mlx4_en_tx_desc {
+	struct mlx4_wqe_ctrl_seg ctrl;
+	union {
+		struct mlx4_wqe_data_seg data; /* at least one data segment */
+		struct mlx4_wqe_lso_seg lso;
+		struct mlx4_wqe_inline_seg inl;
+	};
+};
+
+#define MLX4_EN_USE_SRQ		0x01000000
+
+#define MLX4_EN_CX3_LOW_ID	0x1000
+#define MLX4_EN_CX3_HIGH_ID	0x1005
+
+struct mlx4_en_rx_alloc {
+	struct page	*page;
+	dma_addr_t	dma;
+	u32		page_offset;
+};
+
+#define MLX4_EN_CACHE_SIZE (2 * NAPI_POLL_WEIGHT)
+
+struct mlx4_en_page_cache {
+	u32 index;
+	struct {
+		struct page	*page;
+		dma_addr_t	dma;
+	} buf[MLX4_EN_CACHE_SIZE];
+};
+
+enum {
+	MLX4_EN_TX_RING_STATE_RECOVERING,
+};
+
+struct mlx4_en_priv;
+
+struct mlx4_en_tx_ring {
+	/* cache line used and dirtied in tx completion
+	 * (mlx4_en_free_tx_buf())
+	 */
+	u32			last_nr_txbb;
+	u32			cons;
+	unsigned long		wake_queue;
+	struct netdev_queue	*tx_queue;
+	u32			(*free_tx_desc)(struct mlx4_en_priv *priv,
+						struct mlx4_en_tx_ring *ring,
+						int index,
+						u64 timestamp, int napi_mode);
+	struct mlx4_en_rx_ring	*recycle_ring;
+
+	/* cache line used and dirtied in mlx4_en_xmit() */
+	u32			prod ____cacheline_aligned_in_smp;
+	unsigned int		tx_dropped;
+	unsigned long		bytes;
+	unsigned long		packets;
+	unsigned long		tx_csum;
+	unsigned long		tso_packets;
+	unsigned long		xmit_more;
+	struct mlx4_bf		bf;
+
+	/* Following part should be mostly read */
+	void __iomem		*doorbell_address;
+	__be32			doorbell_qpn;
+	__be32			mr_key;
+	u32			size; /* number of TXBBs */
+	u32			size_mask;
+	u32			full_size;
+	u32			buf_size;
+	void			*buf;
+	struct mlx4_en_tx_info	*tx_info;
+	int			qpn;
+	u8			queue_index;
+	bool			bf_enabled;
+	bool			bf_alloced;
+	u8			hwtstamp_tx_type;
+	u8			*bounce_buf;
+
+	/* Not used in fast path
+	 * Only queue_stopped might be used if BQL is not properly working.
+	 */
+	unsigned long		queue_stopped;
+	unsigned long		state;
+	struct mlx4_hwq_resources sp_wqres;
+	struct mlx4_qp		sp_qp;
+	struct mlx4_qp_context	sp_context;
+	cpumask_t		sp_affinity_mask;
+	enum mlx4_qp_state	sp_qp_state;
+	u16			sp_stride;
+	u16			sp_cqn;	/* index of port CQ associated with this ring */
+} ____cacheline_aligned_in_smp;
+
+struct mlx4_en_rx_desc {
+	/* actual number of entries depends on rx ring stride */
+	struct mlx4_wqe_data_seg data[0];
+};
+
+struct mlx4_en_rx_ring {
+	struct mlx4_hwq_resources wqres;
+	u32 size ;	/* number of Rx descs*/
+	u32 actual_size;
+	u32 size_mask;
+	u16 stride;
+	u16 log_stride;
+	u16 cqn;	/* index of port CQ associated with this ring */
+	u32 prod;
+	u32 cons;
+	u32 buf_size;
+	u8  fcs_del;
+	void *buf;
+	void *rx_info;
+	struct bpf_prog __rcu *xdp_prog;
+	struct mlx4_en_page_cache page_cache;
+	unsigned long bytes;
+	unsigned long packets;
+	unsigned long csum_ok;
+	unsigned long csum_none;
+	unsigned long csum_complete;
+	unsigned long rx_alloc_pages;
+	unsigned long xdp_drop;
+	unsigned long xdp_redirect;
+	unsigned long xdp_redirect_fail;
+	unsigned long xdp_tx;
+	unsigned long xdp_tx_full;
+	unsigned long dropped;
+	int hwtstamp_rx_filter;
+	cpumask_var_t affinity_mask;
+	struct xdp_rxq_info xdp_rxq;
+};
+
+struct mlx4_en_cq {
+	struct mlx4_cq          mcq;
+	struct mlx4_hwq_resources wqres;
+	int                     ring;
+	struct net_device      *dev;
+	union {
+		struct napi_struct napi;
+		bool               xdp_busy;
+	};
+	int size;
+	int buf_size;
+	int vector;
+	enum cq_type type;
+	u16 moder_time;
+	u16 moder_cnt;
+	struct mlx4_cqe *buf;
+#define MLX4_EN_OPCODE_ERROR	0x1e
+
+	const struct cpumask *aff_mask;
+};
+
+struct mlx4_en_port_profile {
+	u32 flags;
+	u32 tx_ring_num[MLX4_EN_NUM_TX_TYPES];
+	u32 rx_ring_num;
+	u32 tx_ring_size;
+	u32 rx_ring_size;
+	u8 num_tx_rings_p_up;
+	u8 rx_pause;
+	u8 rx_ppp;
+	u8 tx_pause;
+	u8 tx_ppp;
+	u8 num_up;
+	int rss_rings;
+	int inline_thold;
+	struct hwtstamp_config hwtstamp_config;
+};
+
+struct mlx4_en_profile {
+	int udp_rss;
+	u8 rss_mask;
+	u32 active_ports;
+	u32 small_pkt_int;
+	u8 no_reset;
+	u8 max_num_tx_rings_p_up;
+	struct mlx4_en_port_profile prof[MLX4_MAX_PORTS + 1];
+};
+
+struct mlx4_en_dev {
+	struct mlx4_dev         *dev;
+	struct pci_dev		*pdev;
+	struct mutex		state_lock;
+	struct net_device       *pndev[MLX4_MAX_PORTS + 1];
+	struct net_device       *upper[MLX4_MAX_PORTS + 1];
+	u32                     port_cnt;
+	bool			device_up;
+	struct mlx4_en_profile  profile;
+	u32			LSO_support;
+	struct workqueue_struct *workqueue;
+	struct device           *dma_device;
+	void __iomem            *uar_map;
+	struct mlx4_uar         priv_uar;
+	struct mlx4_mr		mr;
+	u32                     priv_pdn;
+	spinlock_t              uar_lock;
+	u8			mac_removed[MLX4_MAX_PORTS + 1];
+	u32			nominal_c_mult;
+	struct cyclecounter	cycles;
+	seqlock_t		clock_lock;
+	struct timecounter	clock;
+	unsigned long		last_overflow_check;
+	struct ptp_clock	*ptp_clock;
+	struct ptp_clock_info	ptp_clock_info;
+	struct notifier_block	nb;
+};
+
+
+struct mlx4_en_rss_map {
+	int base_qpn;
+	struct mlx4_qp qps[MAX_RX_RINGS];
+	enum mlx4_qp_state state[MAX_RX_RINGS];
+	struct mlx4_qp *indir_qp;
+	enum mlx4_qp_state indir_state;
+};
+
+enum mlx4_en_port_flag {
+	MLX4_EN_PORT_ANC = 1<<0, /* Auto-negotiation complete */
+	MLX4_EN_PORT_ANE = 1<<1, /* Auto-negotiation enabled */
+};
+
+struct mlx4_en_port_state {
+	int link_state;
+	int link_speed;
+	int transceiver;
+	u32 flags;
+};
+
+enum mlx4_en_mclist_act {
+	MCLIST_NONE,
+	MCLIST_REM,
+	MCLIST_ADD,
+};
+
+struct mlx4_en_mc_list {
+	struct list_head	list;
+	enum mlx4_en_mclist_act	action;
+	u8			addr[ETH_ALEN];
+	u64			reg_id;
+	u64			tunnel_reg_id;
+};
+
+struct mlx4_en_frag_info {
+	u16 frag_size;
+	u32 frag_stride;
+};
+
+#ifdef CONFIG_MLX4_EN_DCB
+/* Minimal TC BW - setting to 0 will block traffic */
+#define MLX4_EN_BW_MIN 1
+#define MLX4_EN_BW_MAX 100 /* Utilize 100% of the line */
+
+#define MLX4_EN_TC_VENDOR 0
+#define MLX4_EN_TC_ETS 7
+
+enum dcb_pfc_type {
+	pfc_disabled = 0,
+	pfc_enabled_full,
+	pfc_enabled_tx,
+	pfc_enabled_rx
+};
+
+struct mlx4_en_cee_config {
+	bool	pfc_state;
+	enum	dcb_pfc_type dcb_pfc[MLX4_EN_NUM_UP_HIGH];
+};
+#endif
+
+struct ethtool_flow_id {
+	struct list_head list;
+	struct ethtool_rx_flow_spec flow_spec;
+	u64 id;
+};
+
+enum {
+	MLX4_EN_FLAG_PROMISC		= (1 << 0),
+	MLX4_EN_FLAG_MC_PROMISC		= (1 << 1),
+	/* whether we need to enable hardware loopback by putting dmac
+	 * in Tx WQE
+	 */
+	MLX4_EN_FLAG_ENABLE_HW_LOOPBACK	= (1 << 2),
+	/* whether we need to drop packets that hardware loopback-ed */
+	MLX4_EN_FLAG_RX_FILTER_NEEDED	= (1 << 3),
+	MLX4_EN_FLAG_FORCE_PROMISC	= (1 << 4),
+	MLX4_EN_FLAG_RX_CSUM_NON_TCP_UDP	= (1 << 5),
+#ifdef CONFIG_MLX4_EN_DCB
+	MLX4_EN_FLAG_DCB_ENABLED        = (1 << 6),
+#endif
+};
+
+#define PORT_BEACON_MAX_LIMIT (65535)
+#define MLX4_EN_MAC_HASH_SIZE (1 << BITS_PER_BYTE)
+#define MLX4_EN_MAC_HASH_IDX 5
+
+struct mlx4_en_stats_bitmap {
+	DECLARE_BITMAP(bitmap, NUM_ALL_STATS);
+	struct mutex mutex; /* for mutual access to stats bitmap */
+};
+
+enum {
+	MLX4_EN_STATE_FLAG_RESTARTING,
+};
+
+struct mlx4_en_priv {
+	struct mlx4_en_dev *mdev;
+	struct mlx4_en_port_profile *prof;
+	struct net_device *dev;
+	unsigned long active_vlans[BITS_TO_LONGS(VLAN_N_VID)];
+	struct mlx4_en_port_state port_state;
+	spinlock_t stats_lock;
+	struct ethtool_flow_id ethtool_rules[MAX_NUM_OF_FS_RULES];
+	/* To allow rules removal while port is going down */
+	struct list_head ethtool_list;
+
+	unsigned long last_moder_packets[MAX_RX_RINGS];
+	unsigned long last_moder_tx_packets;
+	unsigned long last_moder_bytes[MAX_RX_RINGS];
+	unsigned long last_moder_jiffies;
+	int last_moder_time[MAX_RX_RINGS];
+	u16 rx_usecs;
+	u16 rx_frames;
+	u16 tx_usecs;
+	u16 tx_frames;
+	u32 pkt_rate_low;
+	u16 rx_usecs_low;
+	u32 pkt_rate_high;
+	u16 rx_usecs_high;
+	u32 sample_interval;
+	u32 adaptive_rx_coal;
+	u32 msg_enable;
+	u32 loopback_ok;
+	u32 validate_loopback;
+
+	struct mlx4_hwq_resources res;
+	int link_state;
+	bool port_up;
+	int port;
+	int registered;
+	int allocated;
+	int stride;
+	unsigned char current_mac[ETH_ALEN + 2];
+	int mac_index;
+	unsigned max_mtu;
+	int base_qpn;
+	int cqe_factor;
+	int cqe_size;
+
+	struct mlx4_en_rss_map rss_map;
+	__be32 ctrl_flags;
+	u32 flags;
+	u8 num_tx_rings_p_up;
+	u32 tx_work_limit;
+	u32 tx_ring_num[MLX4_EN_NUM_TX_TYPES];
+	u32 rx_ring_num;
+	u32 rx_skb_size;
+	struct mlx4_en_frag_info frag_info[MLX4_EN_MAX_RX_FRAGS];
+	u8 num_frags;
+	u8 log_rx_info;
+	u8 dma_dir;
+	u16 rx_headroom;
+
+	struct mlx4_en_tx_ring **tx_ring[MLX4_EN_NUM_TX_TYPES];
+	struct mlx4_en_rx_ring *rx_ring[MAX_RX_RINGS];
+	struct mlx4_en_cq **tx_cq[MLX4_EN_NUM_TX_TYPES];
+	struct mlx4_en_cq *rx_cq[MAX_RX_RINGS];
+	struct mlx4_qp drop_qp;
+	struct work_struct rx_mode_task;
+	struct work_struct restart_task;
+	struct work_struct linkstate_task;
+	struct delayed_work stats_task;
+	struct delayed_work service_task;
+	struct mlx4_en_pkt_stats pkstats;
+	struct mlx4_en_counter_stats pf_stats;
+	struct mlx4_en_flow_stats_rx rx_priority_flowstats[MLX4_NUM_PRIORITIES];
+	struct mlx4_en_flow_stats_tx tx_priority_flowstats[MLX4_NUM_PRIORITIES];
+	struct mlx4_en_flow_stats_rx rx_flowstats;
+	struct mlx4_en_flow_stats_tx tx_flowstats;
+	struct mlx4_en_port_stats port_stats;
+	struct mlx4_en_xdp_stats xdp_stats;
+	struct mlx4_en_phy_stats phy_stats;
+	struct mlx4_en_stats_bitmap stats_bitmap;
+	struct list_head mc_list;
+	struct list_head curr_list;
+	u64 broadcast_id;
+	struct mlx4_en_stat_out_mbox hw_stats;
+	int vids[128];
+	bool wol;
+	struct device *ddev;
+	struct hlist_head mac_hash[MLX4_EN_MAC_HASH_SIZE];
+	struct hwtstamp_config hwtstamp_config;
+	u32 counter_index;
+
+#ifdef CONFIG_MLX4_EN_DCB
+#define MLX4_EN_DCB_ENABLED	0x3
+	struct ieee_ets ets;
+	u16 maxrate[IEEE_8021QAZ_MAX_TCS];
+	enum dcbnl_cndd_states cndd_state[IEEE_8021QAZ_MAX_TCS];
+	struct mlx4_en_cee_config cee_config;
+	u8 dcbx_cap;
+#endif
+#ifdef CONFIG_RFS_ACCEL
+	spinlock_t filters_lock;
+	int last_filter_id;
+	struct list_head filters;
+	struct hlist_head filter_hash[1 << MLX4_EN_FILTER_HASH_SHIFT];
+#endif
+	u64 tunnel_reg_id;
+	__be16 vxlan_port;
+
+	u32 pflags;
+	u8 rss_key[MLX4_EN_RSS_KEY_SIZE];
+	u8 rss_hash_fn;
+	unsigned long state;
+};
+
+enum mlx4_en_wol {
+	MLX4_EN_WOL_MAGIC = (1ULL << 61),
+	MLX4_EN_WOL_ENABLED = (1ULL << 62),
+};
+
+struct mlx4_mac_entry {
+	struct hlist_node hlist;
+	unsigned char mac[ETH_ALEN + 2];
+	u64 reg_id;
+	struct rcu_head rcu;
+};
+
+static inline struct mlx4_cqe *mlx4_en_get_cqe(void *buf, int idx, int cqe_sz)
+{
+	return buf + idx * cqe_sz;
+}
+
+#define MLX4_EN_WOL_DO_MODIFY (1ULL << 63)
+
+void mlx4_en_init_ptys2ethtool_map(void);
+void mlx4_en_update_loopback_state(struct net_device *dev,
+				   netdev_features_t features);
+
+void mlx4_en_destroy_netdev(struct net_device *dev);
+int mlx4_en_init_netdev(struct mlx4_en_dev *mdev, int port,
+			struct mlx4_en_port_profile *prof);
+
+int mlx4_en_start_port(struct net_device *dev);
+void mlx4_en_stop_port(struct net_device *dev, int detach);
+
+void mlx4_en_set_stats_bitmap(struct mlx4_dev *dev,
+			      struct mlx4_en_stats_bitmap *stats_bitmap,
+			      u8 rx_ppp, u8 rx_pause,
+			      u8 tx_ppp, u8 tx_pause);
+
+int mlx4_en_try_alloc_resources(struct mlx4_en_priv *priv,
+				struct mlx4_en_priv *tmp,
+				struct mlx4_en_port_profile *prof,
+				bool carry_xdp_prog);
+void mlx4_en_safe_replace_resources(struct mlx4_en_priv *priv,
+				    struct mlx4_en_priv *tmp);
+
+int mlx4_en_create_cq(struct mlx4_en_priv *priv, struct mlx4_en_cq **pcq,
+		      int entries, int ring, enum cq_type mode, int node);
+void mlx4_en_destroy_cq(struct mlx4_en_priv *priv, struct mlx4_en_cq **pcq);
+int mlx4_en_activate_cq(struct mlx4_en_priv *priv, struct mlx4_en_cq *cq,
+			int cq_idx);
+void mlx4_en_deactivate_cq(struct mlx4_en_priv *priv, struct mlx4_en_cq *cq);
+int mlx4_en_set_cq_moder(struct mlx4_en_priv *priv, struct mlx4_en_cq *cq);
+void mlx4_en_arm_cq(struct mlx4_en_priv *priv, struct mlx4_en_cq *cq);
+
+void mlx4_en_tx_irq(struct mlx4_cq *mcq);
+u16 mlx4_en_select_queue(struct net_device *dev, struct sk_buff *skb,
+			 struct net_device *sb_dev);
+netdev_tx_t mlx4_en_xmit(struct sk_buff *skb, struct net_device *dev);
+netdev_tx_t mlx4_en_xmit_frame(struct mlx4_en_rx_ring *rx_ring,
+			       struct mlx4_en_rx_alloc *frame,
+			       struct mlx4_en_priv *priv, unsigned int length,
+			       int tx_ind, bool *doorbell_pending);
+void mlx4_en_xmit_doorbell(struct mlx4_en_tx_ring *ring);
+bool mlx4_en_rx_recycle(struct mlx4_en_rx_ring *ring,
+			struct mlx4_en_rx_alloc *frame);
+
+int mlx4_en_create_tx_ring(struct mlx4_en_priv *priv,
+			   struct mlx4_en_tx_ring **pring,
+			   u32 size, u16 stride,
+			   int node, int queue_index);
+void mlx4_en_destroy_tx_ring(struct mlx4_en_priv *priv,
+			     struct mlx4_en_tx_ring **pring);
+void mlx4_en_init_tx_xdp_ring_descs(struct mlx4_en_priv *priv,
+				    struct mlx4_en_tx_ring *ring);
+int mlx4_en_activate_tx_ring(struct mlx4_en_priv *priv,
+			     struct mlx4_en_tx_ring *ring,
+			     int cq, int user_prio);
+void mlx4_en_deactivate_tx_ring(struct mlx4_en_priv *priv,
+				struct mlx4_en_tx_ring *ring);
+void mlx4_en_set_num_rx_rings(struct mlx4_en_dev *mdev);
+void mlx4_en_recover_from_oom(struct mlx4_en_priv *priv);
+int mlx4_en_create_rx_ring(struct mlx4_en_priv *priv,
+			   struct mlx4_en_rx_ring **pring,
+			   u32 size, u16 stride, int node, int queue_index);
+void mlx4_en_destroy_rx_ring(struct mlx4_en_priv *priv,
+			     struct mlx4_en_rx_ring **pring,
+			     u32 size, u16 stride);
+int mlx4_en_activate_rx_rings(struct mlx4_en_priv *priv);
+void mlx4_en_deactivate_rx_ring(struct mlx4_en_priv *priv,
+				struct mlx4_en_rx_ring *ring);
+int mlx4_en_process_rx_cq(struct net_device *dev,
+			  struct mlx4_en_cq *cq,
+			  int budget);
+int mlx4_en_poll_rx_cq(struct napi_struct *napi, int budget);
+int mlx4_en_poll_tx_cq(struct napi_struct *napi, int budget);
+int mlx4_en_process_tx_cq(struct net_device *dev,
+			  struct mlx4_en_cq *cq, int napi_budget);
+u32 mlx4_en_free_tx_desc(struct mlx4_en_priv *priv,
+			 struct mlx4_en_tx_ring *ring,
+			 int index, u64 timestamp,
+			 int napi_mode);
+u32 mlx4_en_recycle_tx_desc(struct mlx4_en_priv *priv,
+			    struct mlx4_en_tx_ring *ring,
+			    int index, u64 timestamp,
+			    int napi_mode);
+void mlx4_en_fill_qp_context(struct mlx4_en_priv *priv, int size, int stride,
+		int is_tx, int rss, int qpn, int cqn, int user_prio,
+		struct mlx4_qp_context *context);
+void mlx4_en_sqp_event(struct mlx4_qp *qp, enum mlx4_event event);
+int mlx4_en_change_mcast_lb(struct mlx4_en_priv *priv, struct mlx4_qp *qp,
+			    int loopback);
+void mlx4_en_calc_rx_buf(struct net_device *dev);
+int mlx4_en_config_rss_steer(struct mlx4_en_priv *priv);
+void mlx4_en_release_rss_steer(struct mlx4_en_priv *priv);
+int mlx4_en_create_drop_qp(struct mlx4_en_priv *priv);
+void mlx4_en_destroy_drop_qp(struct mlx4_en_priv *priv);
+int mlx4_en_free_tx_buf(struct net_device *dev, struct mlx4_en_tx_ring *ring);
+void mlx4_en_rx_irq(struct mlx4_cq *mcq);
+
+int mlx4_SET_MCAST_FLTR(struct mlx4_dev *dev, u8 port, u64 mac, u64 clear, u8 mode);
+int mlx4_SET_VLAN_FLTR(struct mlx4_dev *dev, struct mlx4_en_priv *priv);
+
+void mlx4_en_fold_software_stats(struct net_device *dev);
+int mlx4_en_DUMP_ETH_STATS(struct mlx4_en_dev *mdev, u8 port, u8 reset);
+int mlx4_en_QUERY_PORT(struct mlx4_en_dev *mdev, u8 port);
+
+#ifdef CONFIG_MLX4_EN_DCB
+extern const struct dcbnl_rtnl_ops mlx4_en_dcbnl_ops;
+extern const struct dcbnl_rtnl_ops mlx4_en_dcbnl_pfc_ops;
+#endif
+
+int mlx4_en_setup_tc(struct net_device *dev, u8 up);
+int mlx4_en_alloc_tx_queue_per_tc(struct net_device *dev, u8 tc);
+
+#ifdef CONFIG_RFS_ACCEL
+void mlx4_en_cleanup_filters(struct mlx4_en_priv *priv);
+#endif
+
+#define MLX4_EN_NUM_SELF_TEST	5
+void mlx4_en_ex_selftest(struct net_device *dev, u32 *flags, u64 *buf);
+void mlx4_en_ptp_overflow_check(struct mlx4_en_dev *mdev);
+
+#define DEV_FEATURE_CHANGED(dev, new_features, feature) \
+	((dev->features & feature) ^ (new_features & feature))
+
+int mlx4_en_moderation_update(struct mlx4_en_priv *priv);
+int mlx4_en_reset_config(struct net_device *dev,
+			 struct hwtstamp_config ts_config,
+			 netdev_features_t new_features);
+void mlx4_en_update_pfc_stats_bitmap(struct mlx4_dev *dev,
+				     struct mlx4_en_stats_bitmap *stats_bitmap,
+				     u8 rx_ppp, u8 rx_pause,
+				     u8 tx_ppp, u8 tx_pause);
+int mlx4_en_netdev_event(struct notifier_block *this,
+			 unsigned long event, void *ptr);
+
+/*
+ * Functions for time stamping
+ */
+u64 mlx4_en_get_cqe_ts(struct mlx4_cqe *cqe);
+void mlx4_en_fill_hwtstamps(struct mlx4_en_dev *mdev,
+			    struct skb_shared_hwtstamps *hwts,
+			    u64 timestamp);
+void mlx4_en_init_timestamp(struct mlx4_en_dev *mdev);
+void mlx4_en_remove_timestamp(struct mlx4_en_dev *mdev);
+
+/* Globals
+ */
+extern const struct ethtool_ops mlx4_en_ethtool_ops;
+
+
+
+/*
+ * printk / logging functions
+ */
+
+__printf(3, 4)
+void en_print(const char *level, const struct mlx4_en_priv *priv,
+	      const char *format, ...);
+
+#define en_dbg(mlevel, priv, format, ...)				\
+do {									\
+	if (NETIF_MSG_##mlevel & (priv)->msg_enable)			\
+		en_print(KERN_DEBUG, priv, format, ##__VA_ARGS__);	\
+} while (0)
+#define en_warn(priv, format, ...)					\
+	en_print(KERN_WARNING, priv, format, ##__VA_ARGS__)
+#define en_err(priv, format, ...)					\
+	en_print(KERN_ERR, priv, format, ##__VA_ARGS__)
+#define en_info(priv, format, ...)					\
+	en_print(KERN_INFO, priv, format, ##__VA_ARGS__)
+
+#define mlx4_err(mdev, format, ...)					\
+	pr_err(DRV_NAME " %s: " format,					\
+	       dev_name(&(mdev)->pdev->dev), ##__VA_ARGS__)
+#define mlx4_info(mdev, format, ...)					\
+	pr_info(DRV_NAME " %s: " format,				\
+		dev_name(&(mdev)->pdev->dev), ##__VA_ARGS__)
+#define mlx4_warn(mdev, format, ...)					\
+	pr_warn(DRV_NAME " %s: " format,				\
+		dev_name(&(mdev)->pdev->dev), ##__VA_ARGS__)
+
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlx4/mlx4_stats.h b/drivers/net/ethernet/mellanox/mlx4/mlx4_stats.h
new file mode 100644
index 000000000..e9cd4bb6f
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx4/mlx4_stats.h
@@ -0,0 +1,130 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef _MLX4_STATS_
+#define _MLX4_STATS_
+
+#define NUM_PRIORITIES	9
+#define NUM_PRIORITY_STATS 2
+
+struct mlx4_en_pkt_stats {
+	unsigned long rx_multicast_packets;
+	unsigned long rx_broadcast_packets;
+	unsigned long rx_jabbers;
+	unsigned long rx_in_range_length_error;
+	unsigned long rx_out_range_length_error;
+	unsigned long tx_multicast_packets;
+	unsigned long tx_broadcast_packets;
+	unsigned long rx_prio[NUM_PRIORITIES][NUM_PRIORITY_STATS];
+	unsigned long tx_prio[NUM_PRIORITIES][NUM_PRIORITY_STATS];
+#define NUM_PKT_STATS		43
+};
+
+struct mlx4_en_counter_stats {
+	unsigned long rx_packets;
+	unsigned long rx_bytes;
+	unsigned long tx_packets;
+	unsigned long tx_bytes;
+#define NUM_PF_STATS      4
+};
+
+struct mlx4_en_port_stats {
+	unsigned long tso_packets;
+	unsigned long xmit_more;
+	unsigned long queue_stopped;
+	unsigned long wake_queue;
+	unsigned long tx_timeout;
+	unsigned long rx_alloc_pages;
+	unsigned long rx_chksum_good;
+	unsigned long rx_chksum_none;
+	unsigned long rx_chksum_complete;
+	unsigned long tx_chksum_offload;
+#define NUM_PORT_STATS		10
+};
+
+struct mlx4_en_xdp_stats {
+	unsigned long rx_xdp_drop;
+	unsigned long rx_xdp_redirect;
+	unsigned long rx_xdp_redirect_fail;
+	unsigned long rx_xdp_tx;
+	unsigned long rx_xdp_tx_full;
+#define NUM_XDP_STATS		5
+};
+
+struct mlx4_en_phy_stats {
+	unsigned long rx_packets_phy;
+	unsigned long rx_bytes_phy;
+	unsigned long tx_packets_phy;
+	unsigned long tx_bytes_phy;
+#define NUM_PHY_STATS		4
+};
+
+#define NUM_MAIN_STATS	21
+
+#define MLX4_NUM_PRIORITIES	8
+
+struct mlx4_en_flow_stats_rx {
+	u64 rx_pause;
+	u64 rx_pause_duration;
+	u64 rx_pause_transition;
+#define NUM_FLOW_STATS_RX	3
+#define NUM_FLOW_PRIORITY_STATS_RX	(NUM_FLOW_STATS_RX * \
+					 MLX4_NUM_PRIORITIES)
+};
+
+#define FLOW_PRIORITY_STATS_IDX_RX_FRAMES	(NUM_MAIN_STATS +	\
+						 NUM_PORT_STATS +	\
+						 NUM_PF_STATS +		\
+						 NUM_FLOW_PRIORITY_STATS_RX)
+
+struct mlx4_en_flow_stats_tx {
+	u64 tx_pause;
+	u64 tx_pause_duration;
+	u64 tx_pause_transition;
+#define NUM_FLOW_STATS_TX	3
+#define NUM_FLOW_PRIORITY_STATS_TX	(NUM_FLOW_STATS_TX * \
+					 MLX4_NUM_PRIORITIES)
+};
+
+#define FLOW_PRIORITY_STATS_IDX_TX_FRAMES	(NUM_MAIN_STATS +	\
+						 NUM_PORT_STATS +	\
+						 NUM_PF_STATS +		\
+						 NUM_FLOW_PRIORITY_STATS_RX + \
+						 NUM_FLOW_STATS_RX +	\
+						 NUM_FLOW_PRIORITY_STATS_TX)
+
+#define NUM_FLOW_STATS (NUM_FLOW_STATS_RX + NUM_FLOW_STATS_TX + \
+			NUM_FLOW_PRIORITY_STATS_TX + \
+			NUM_FLOW_PRIORITY_STATS_RX)
+
+struct mlx4_en_stat_out_flow_control_mbox {
+	/* Total number of PAUSE frames received from the far-end port */
+	__be64 rx_pause;
+	/* Total number of microseconds that far-end port requested to pause
+	* transmission of packets
+	*/
+	__be64 rx_pause_duration;
+	/* Number of received transmission from XOFF state to XON state */
+	__be64 rx_pause_transition;
+	/* Total number of PAUSE frames sent from the far-end port */
+	__be64 tx_pause;
+	/* Total time in microseconds that transmission of packets has been
+	* paused
+	*/
+	__be64 tx_pause_duration;
+	/* Number of transmitter transitions from XOFF state to XON state */
+	__be64 tx_pause_transition;
+	/* Reserverd */
+	__be64 reserved[2];
+};
+
+enum {
+	MLX4_DUMP_ETH_STATS_FLOW_CONTROL = 1 << 12
+};
+
+#define NUM_ALL_STATS	(NUM_MAIN_STATS + NUM_PORT_STATS + NUM_PKT_STATS + \
+			 NUM_FLOW_STATS + NUM_PF_STATS + \
+			 NUM_XDP_STATS + NUM_PHY_STATS)
+
+#define MLX4_FIND_NETDEV_STAT(n) (offsetof(struct net_device_stats, n) / \
+				  sizeof(((struct net_device_stats *)0)->n))
+
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlx4/mr.c b/drivers/net/ethernet/mellanox/mlx4/mr.c
new file mode 100644
index 000000000..d7444782b
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx4/mr.c
@@ -0,0 +1,974 @@
+/*
+ * Copyright (c) 2004 Topspin Communications.  All rights reserved.
+ * Copyright (c) 2005, 2006, 2007, 2008 Mellanox Technologies. All rights reserved.
+ * Copyright (c) 2006, 2007 Cisco Systems, Inc.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/errno.h>
+#include <linux/export.h>
+#include <linux/slab.h>
+#include <linux/kernel.h>
+#include <linux/vmalloc.h>
+
+#include <linux/mlx4/cmd.h>
+
+#include "mlx4.h"
+#include "icm.h"
+
+static u32 mlx4_buddy_alloc(struct mlx4_buddy *buddy, int order)
+{
+	int o;
+	int m;
+	u32 seg;
+
+	spin_lock(&buddy->lock);
+
+	for (o = order; o <= buddy->max_order; ++o)
+		if (buddy->num_free[o]) {
+			m = 1 << (buddy->max_order - o);
+			seg = find_first_bit(buddy->bits[o], m);
+			if (seg < m)
+				goto found;
+		}
+
+	spin_unlock(&buddy->lock);
+	return -1;
+
+ found:
+	clear_bit(seg, buddy->bits[o]);
+	--buddy->num_free[o];
+
+	while (o > order) {
+		--o;
+		seg <<= 1;
+		set_bit(seg ^ 1, buddy->bits[o]);
+		++buddy->num_free[o];
+	}
+
+	spin_unlock(&buddy->lock);
+
+	seg <<= order;
+
+	return seg;
+}
+
+static void mlx4_buddy_free(struct mlx4_buddy *buddy, u32 seg, int order)
+{
+	seg >>= order;
+
+	spin_lock(&buddy->lock);
+
+	while (test_bit(seg ^ 1, buddy->bits[order])) {
+		clear_bit(seg ^ 1, buddy->bits[order]);
+		--buddy->num_free[order];
+		seg >>= 1;
+		++order;
+	}
+
+	set_bit(seg, buddy->bits[order]);
+	++buddy->num_free[order];
+
+	spin_unlock(&buddy->lock);
+}
+
+static int mlx4_buddy_init(struct mlx4_buddy *buddy, int max_order)
+{
+	int i, s;
+
+	buddy->max_order = max_order;
+	spin_lock_init(&buddy->lock);
+
+	buddy->bits = kcalloc(buddy->max_order + 1, sizeof(long *),
+			      GFP_KERNEL);
+	buddy->num_free = kcalloc(buddy->max_order + 1, sizeof(*buddy->num_free),
+				  GFP_KERNEL);
+	if (!buddy->bits || !buddy->num_free)
+		goto err_out;
+
+	for (i = 0; i <= buddy->max_order; ++i) {
+		s = BITS_TO_LONGS(1UL << (buddy->max_order - i));
+		buddy->bits[i] = kvmalloc_array(s, sizeof(long), GFP_KERNEL | __GFP_ZERO);
+		if (!buddy->bits[i])
+			goto err_out_free;
+	}
+
+	set_bit(0, buddy->bits[buddy->max_order]);
+	buddy->num_free[buddy->max_order] = 1;
+
+	return 0;
+
+err_out_free:
+	for (i = 0; i <= buddy->max_order; ++i)
+		kvfree(buddy->bits[i]);
+
+err_out:
+	kfree(buddy->bits);
+	kfree(buddy->num_free);
+
+	return -ENOMEM;
+}
+
+static void mlx4_buddy_cleanup(struct mlx4_buddy *buddy)
+{
+	int i;
+
+	for (i = 0; i <= buddy->max_order; ++i)
+		kvfree(buddy->bits[i]);
+
+	kfree(buddy->bits);
+	kfree(buddy->num_free);
+}
+
+u32 __mlx4_alloc_mtt_range(struct mlx4_dev *dev, int order)
+{
+	struct mlx4_mr_table *mr_table = &mlx4_priv(dev)->mr_table;
+	u32 seg;
+	int seg_order;
+	u32 offset;
+
+	seg_order = max_t(int, order - log_mtts_per_seg, 0);
+
+	seg = mlx4_buddy_alloc(&mr_table->mtt_buddy, seg_order);
+	if (seg == -1)
+		return -1;
+
+	offset = seg * (1 << log_mtts_per_seg);
+
+	if (mlx4_table_get_range(dev, &mr_table->mtt_table, offset,
+				 offset + (1 << order) - 1)) {
+		mlx4_buddy_free(&mr_table->mtt_buddy, seg, seg_order);
+		return -1;
+	}
+
+	return offset;
+}
+
+static u32 mlx4_alloc_mtt_range(struct mlx4_dev *dev, int order)
+{
+	u64 in_param = 0;
+	u64 out_param;
+	int err;
+
+	if (mlx4_is_mfunc(dev)) {
+		set_param_l(&in_param, order);
+		err = mlx4_cmd_imm(dev, in_param, &out_param, RES_MTT,
+						       RES_OP_RESERVE_AND_MAP,
+						       MLX4_CMD_ALLOC_RES,
+						       MLX4_CMD_TIME_CLASS_A,
+						       MLX4_CMD_WRAPPED);
+		if (err)
+			return -1;
+		return get_param_l(&out_param);
+	}
+	return __mlx4_alloc_mtt_range(dev, order);
+}
+
+int mlx4_mtt_init(struct mlx4_dev *dev, int npages, int page_shift,
+		  struct mlx4_mtt *mtt)
+{
+	int i;
+
+	if (!npages) {
+		mtt->order      = -1;
+		mtt->page_shift = MLX4_ICM_PAGE_SHIFT;
+		return 0;
+	} else
+		mtt->page_shift = page_shift;
+
+	for (mtt->order = 0, i = 1; i < npages; i <<= 1)
+		++mtt->order;
+
+	mtt->offset = mlx4_alloc_mtt_range(dev, mtt->order);
+	if (mtt->offset == -1)
+		return -ENOMEM;
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(mlx4_mtt_init);
+
+void __mlx4_free_mtt_range(struct mlx4_dev *dev, u32 offset, int order)
+{
+	u32 first_seg;
+	int seg_order;
+	struct mlx4_mr_table *mr_table = &mlx4_priv(dev)->mr_table;
+
+	seg_order = max_t(int, order - log_mtts_per_seg, 0);
+	first_seg = offset / (1 << log_mtts_per_seg);
+
+	mlx4_buddy_free(&mr_table->mtt_buddy, first_seg, seg_order);
+	mlx4_table_put_range(dev, &mr_table->mtt_table, offset,
+			     offset + (1 << order) - 1);
+}
+
+static void mlx4_free_mtt_range(struct mlx4_dev *dev, u32 offset, int order)
+{
+	u64 in_param = 0;
+	int err;
+
+	if (mlx4_is_mfunc(dev)) {
+		set_param_l(&in_param, offset);
+		set_param_h(&in_param, order);
+		err = mlx4_cmd(dev, in_param, RES_MTT, RES_OP_RESERVE_AND_MAP,
+						       MLX4_CMD_FREE_RES,
+						       MLX4_CMD_TIME_CLASS_A,
+						       MLX4_CMD_WRAPPED);
+		if (err)
+			mlx4_warn(dev, "Failed to free mtt range at:%d order:%d\n",
+				  offset, order);
+		return;
+	}
+	__mlx4_free_mtt_range(dev, offset, order);
+}
+
+void mlx4_mtt_cleanup(struct mlx4_dev *dev, struct mlx4_mtt *mtt)
+{
+	if (mtt->order < 0)
+		return;
+
+	mlx4_free_mtt_range(dev, mtt->offset, mtt->order);
+}
+EXPORT_SYMBOL_GPL(mlx4_mtt_cleanup);
+
+u64 mlx4_mtt_addr(struct mlx4_dev *dev, struct mlx4_mtt *mtt)
+{
+	return (u64) mtt->offset * dev->caps.mtt_entry_sz;
+}
+EXPORT_SYMBOL_GPL(mlx4_mtt_addr);
+
+static u32 hw_index_to_key(u32 ind)
+{
+	return (ind >> 24) | (ind << 8);
+}
+
+static u32 key_to_hw_index(u32 key)
+{
+	return (key << 24) | (key >> 8);
+}
+
+static int mlx4_SW2HW_MPT(struct mlx4_dev *dev, struct mlx4_cmd_mailbox *mailbox,
+			  int mpt_index)
+{
+	return mlx4_cmd(dev, mailbox->dma, mpt_index,
+			0, MLX4_CMD_SW2HW_MPT, MLX4_CMD_TIME_CLASS_B,
+			MLX4_CMD_WRAPPED);
+}
+
+static int mlx4_HW2SW_MPT(struct mlx4_dev *dev, struct mlx4_cmd_mailbox *mailbox,
+			  int mpt_index)
+{
+	return mlx4_cmd_box(dev, 0, mailbox ? mailbox->dma : 0, mpt_index,
+			    !mailbox, MLX4_CMD_HW2SW_MPT,
+			    MLX4_CMD_TIME_CLASS_B, MLX4_CMD_WRAPPED);
+}
+
+/* Must protect against concurrent access */
+int mlx4_mr_hw_get_mpt(struct mlx4_dev *dev, struct mlx4_mr *mmr,
+		       struct mlx4_mpt_entry ***mpt_entry)
+{
+	int err;
+	int key = key_to_hw_index(mmr->key) & (dev->caps.num_mpts - 1);
+	struct mlx4_cmd_mailbox *mailbox = NULL;
+
+	if (mmr->enabled != MLX4_MPT_EN_HW)
+		return -EINVAL;
+
+	err = mlx4_HW2SW_MPT(dev, NULL, key);
+	if (err) {
+		mlx4_warn(dev, "HW2SW_MPT failed (%d).", err);
+		mlx4_warn(dev, "Most likely the MR has MWs bound to it.\n");
+		return err;
+	}
+
+	mmr->enabled = MLX4_MPT_EN_SW;
+
+	if (!mlx4_is_mfunc(dev)) {
+		**mpt_entry = mlx4_table_find(
+				&mlx4_priv(dev)->mr_table.dmpt_table,
+				key, NULL);
+	} else {
+		mailbox = mlx4_alloc_cmd_mailbox(dev);
+		if (IS_ERR(mailbox))
+			return PTR_ERR(mailbox);
+
+		err = mlx4_cmd_box(dev, 0, mailbox->dma, key,
+				   0, MLX4_CMD_QUERY_MPT,
+				   MLX4_CMD_TIME_CLASS_B,
+				   MLX4_CMD_WRAPPED);
+		if (err)
+			goto free_mailbox;
+
+		*mpt_entry = (struct mlx4_mpt_entry **)&mailbox->buf;
+	}
+
+	if (!(*mpt_entry) || !(**mpt_entry)) {
+		err = -ENOMEM;
+		goto free_mailbox;
+	}
+
+	return 0;
+
+free_mailbox:
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx4_mr_hw_get_mpt);
+
+int mlx4_mr_hw_write_mpt(struct mlx4_dev *dev, struct mlx4_mr *mmr,
+			 struct mlx4_mpt_entry **mpt_entry)
+{
+	int err;
+
+	if (!mlx4_is_mfunc(dev)) {
+		/* Make sure any changes to this entry are flushed */
+		wmb();
+
+		*(u8 *)(*mpt_entry) = MLX4_MPT_STATUS_HW;
+
+		/* Make sure the new status is written */
+		wmb();
+
+		err = mlx4_SYNC_TPT(dev);
+	} else {
+		int key = key_to_hw_index(mmr->key) & (dev->caps.num_mpts - 1);
+
+		struct mlx4_cmd_mailbox *mailbox =
+			container_of((void *)mpt_entry, struct mlx4_cmd_mailbox,
+				     buf);
+
+		(*mpt_entry)->lkey = 0;
+		err = mlx4_SW2HW_MPT(dev, mailbox, key);
+	}
+
+	if (!err) {
+		mmr->pd = be32_to_cpu((*mpt_entry)->pd_flags) & MLX4_MPT_PD_MASK;
+		mmr->enabled = MLX4_MPT_EN_HW;
+	}
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx4_mr_hw_write_mpt);
+
+void mlx4_mr_hw_put_mpt(struct mlx4_dev *dev,
+			struct mlx4_mpt_entry **mpt_entry)
+{
+	if (mlx4_is_mfunc(dev)) {
+		struct mlx4_cmd_mailbox *mailbox =
+			container_of((void *)mpt_entry, struct mlx4_cmd_mailbox,
+				     buf);
+		mlx4_free_cmd_mailbox(dev, mailbox);
+	}
+}
+EXPORT_SYMBOL_GPL(mlx4_mr_hw_put_mpt);
+
+int mlx4_mr_hw_change_pd(struct mlx4_dev *dev, struct mlx4_mpt_entry *mpt_entry,
+			 u32 pdn)
+{
+	u32 pd_flags = be32_to_cpu(mpt_entry->pd_flags) & ~MLX4_MPT_PD_MASK;
+	/* The wrapper function will put the slave's id here */
+	if (mlx4_is_mfunc(dev))
+		pd_flags &= ~MLX4_MPT_PD_VF_MASK;
+
+	mpt_entry->pd_flags = cpu_to_be32(pd_flags |
+					  (pdn & MLX4_MPT_PD_MASK)
+					  | MLX4_MPT_PD_FLAG_EN_INV);
+	return 0;
+}
+EXPORT_SYMBOL_GPL(mlx4_mr_hw_change_pd);
+
+int mlx4_mr_hw_change_access(struct mlx4_dev *dev,
+			     struct mlx4_mpt_entry *mpt_entry,
+			     u32 access)
+{
+	u32 flags = (be32_to_cpu(mpt_entry->flags) & ~MLX4_PERM_MASK) |
+		    (access & MLX4_PERM_MASK);
+
+	mpt_entry->flags = cpu_to_be32(flags);
+	return 0;
+}
+EXPORT_SYMBOL_GPL(mlx4_mr_hw_change_access);
+
+static int mlx4_mr_alloc_reserved(struct mlx4_dev *dev, u32 mridx, u32 pd,
+			   u64 iova, u64 size, u32 access, int npages,
+			   int page_shift, struct mlx4_mr *mr)
+{
+	mr->iova       = iova;
+	mr->size       = size;
+	mr->pd	       = pd;
+	mr->access     = access;
+	mr->enabled    = MLX4_MPT_DISABLED;
+	mr->key	       = hw_index_to_key(mridx);
+
+	return mlx4_mtt_init(dev, npages, page_shift, &mr->mtt);
+}
+
+static int mlx4_WRITE_MTT(struct mlx4_dev *dev,
+			  struct mlx4_cmd_mailbox *mailbox,
+			  int num_entries)
+{
+	return mlx4_cmd(dev, mailbox->dma, num_entries, 0, MLX4_CMD_WRITE_MTT,
+			MLX4_CMD_TIME_CLASS_A,  MLX4_CMD_WRAPPED);
+}
+
+int __mlx4_mpt_reserve(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	return mlx4_bitmap_alloc(&priv->mr_table.mpt_bitmap);
+}
+
+static int mlx4_mpt_reserve(struct mlx4_dev *dev)
+{
+	u64 out_param;
+
+	if (mlx4_is_mfunc(dev)) {
+		if (mlx4_cmd_imm(dev, 0, &out_param, RES_MPT, RES_OP_RESERVE,
+				   MLX4_CMD_ALLOC_RES,
+				   MLX4_CMD_TIME_CLASS_A, MLX4_CMD_WRAPPED))
+			return -1;
+		return get_param_l(&out_param);
+	}
+	return  __mlx4_mpt_reserve(dev);
+}
+
+void __mlx4_mpt_release(struct mlx4_dev *dev, u32 index)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	mlx4_bitmap_free(&priv->mr_table.mpt_bitmap, index, MLX4_NO_RR);
+}
+
+static void mlx4_mpt_release(struct mlx4_dev *dev, u32 index)
+{
+	u64 in_param = 0;
+
+	if (mlx4_is_mfunc(dev)) {
+		set_param_l(&in_param, index);
+		if (mlx4_cmd(dev, in_param, RES_MPT, RES_OP_RESERVE,
+			       MLX4_CMD_FREE_RES,
+			       MLX4_CMD_TIME_CLASS_A, MLX4_CMD_WRAPPED))
+			mlx4_warn(dev, "Failed to release mr index:%d\n",
+				  index);
+		return;
+	}
+	__mlx4_mpt_release(dev, index);
+}
+
+int __mlx4_mpt_alloc_icm(struct mlx4_dev *dev, u32 index)
+{
+	struct mlx4_mr_table *mr_table = &mlx4_priv(dev)->mr_table;
+
+	return mlx4_table_get(dev, &mr_table->dmpt_table, index);
+}
+
+static int mlx4_mpt_alloc_icm(struct mlx4_dev *dev, u32 index)
+{
+	u64 param = 0;
+
+	if (mlx4_is_mfunc(dev)) {
+		set_param_l(&param, index);
+		return mlx4_cmd_imm(dev, param, &param, RES_MPT, RES_OP_MAP_ICM,
+							MLX4_CMD_ALLOC_RES,
+							MLX4_CMD_TIME_CLASS_A,
+							MLX4_CMD_WRAPPED);
+	}
+	return __mlx4_mpt_alloc_icm(dev, index);
+}
+
+void __mlx4_mpt_free_icm(struct mlx4_dev *dev, u32 index)
+{
+	struct mlx4_mr_table *mr_table = &mlx4_priv(dev)->mr_table;
+
+	mlx4_table_put(dev, &mr_table->dmpt_table, index);
+}
+
+static void mlx4_mpt_free_icm(struct mlx4_dev *dev, u32 index)
+{
+	u64 in_param = 0;
+
+	if (mlx4_is_mfunc(dev)) {
+		set_param_l(&in_param, index);
+		if (mlx4_cmd(dev, in_param, RES_MPT, RES_OP_MAP_ICM,
+			     MLX4_CMD_FREE_RES, MLX4_CMD_TIME_CLASS_A,
+			     MLX4_CMD_WRAPPED))
+			mlx4_warn(dev, "Failed to free icm of mr index:%d\n",
+				  index);
+		return;
+	}
+	return __mlx4_mpt_free_icm(dev, index);
+}
+
+int mlx4_mr_alloc(struct mlx4_dev *dev, u32 pd, u64 iova, u64 size, u32 access,
+		  int npages, int page_shift, struct mlx4_mr *mr)
+{
+	u32 index;
+	int err;
+
+	index = mlx4_mpt_reserve(dev);
+	if (index == -1)
+		return -ENOMEM;
+
+	err = mlx4_mr_alloc_reserved(dev, index, pd, iova, size,
+				     access, npages, page_shift, mr);
+	if (err)
+		mlx4_mpt_release(dev, index);
+
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx4_mr_alloc);
+
+static int mlx4_mr_free_reserved(struct mlx4_dev *dev, struct mlx4_mr *mr)
+{
+	int err;
+
+	if (mr->enabled == MLX4_MPT_EN_HW) {
+		err = mlx4_HW2SW_MPT(dev, NULL,
+				     key_to_hw_index(mr->key) &
+				     (dev->caps.num_mpts - 1));
+		if (err) {
+			mlx4_warn(dev, "HW2SW_MPT failed (%d), MR has MWs bound to it\n",
+				  err);
+			return err;
+		}
+
+		mr->enabled = MLX4_MPT_EN_SW;
+	}
+	mlx4_mtt_cleanup(dev, &mr->mtt);
+
+	return 0;
+}
+
+int mlx4_mr_free(struct mlx4_dev *dev, struct mlx4_mr *mr)
+{
+	int ret;
+
+	ret = mlx4_mr_free_reserved(dev, mr);
+	if (ret)
+		return ret;
+	if (mr->enabled)
+		mlx4_mpt_free_icm(dev, key_to_hw_index(mr->key));
+	mlx4_mpt_release(dev, key_to_hw_index(mr->key));
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(mlx4_mr_free);
+
+void mlx4_mr_rereg_mem_cleanup(struct mlx4_dev *dev, struct mlx4_mr *mr)
+{
+	mlx4_mtt_cleanup(dev, &mr->mtt);
+	mr->mtt.order = -1;
+}
+EXPORT_SYMBOL_GPL(mlx4_mr_rereg_mem_cleanup);
+
+int mlx4_mr_rereg_mem_write(struct mlx4_dev *dev, struct mlx4_mr *mr,
+			    u64 iova, u64 size, int npages,
+			    int page_shift, struct mlx4_mpt_entry *mpt_entry)
+{
+	int err;
+
+	err = mlx4_mtt_init(dev, npages, page_shift, &mr->mtt);
+	if (err)
+		return err;
+
+	mpt_entry->start       = cpu_to_be64(iova);
+	mpt_entry->length      = cpu_to_be64(size);
+	mpt_entry->entity_size = cpu_to_be32(page_shift);
+	mpt_entry->flags    &= ~(cpu_to_be32(MLX4_MPT_FLAG_FREE |
+					   MLX4_MPT_FLAG_SW_OWNS));
+	if (mr->mtt.order < 0) {
+		mpt_entry->flags |= cpu_to_be32(MLX4_MPT_FLAG_PHYSICAL);
+		mpt_entry->mtt_addr = 0;
+	} else {
+		mpt_entry->mtt_addr = cpu_to_be64(mlx4_mtt_addr(dev,
+						  &mr->mtt));
+		if (mr->mtt.page_shift == 0)
+			mpt_entry->mtt_sz    = cpu_to_be32(1 << mr->mtt.order);
+	}
+	if (mr->mtt.order >= 0 && mr->mtt.page_shift == 0) {
+		/* fast register MR in free state */
+		mpt_entry->flags    |= cpu_to_be32(MLX4_MPT_FLAG_FREE);
+		mpt_entry->pd_flags |= cpu_to_be32(MLX4_MPT_PD_FLAG_FAST_REG |
+						   MLX4_MPT_PD_FLAG_RAE);
+	} else {
+		mpt_entry->flags    |= cpu_to_be32(MLX4_MPT_FLAG_SW_OWNS);
+	}
+	mr->enabled = MLX4_MPT_EN_SW;
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(mlx4_mr_rereg_mem_write);
+
+int mlx4_mr_enable(struct mlx4_dev *dev, struct mlx4_mr *mr)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	struct mlx4_mpt_entry *mpt_entry;
+	int err;
+
+	err = mlx4_mpt_alloc_icm(dev, key_to_hw_index(mr->key));
+	if (err)
+		return err;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox)) {
+		err = PTR_ERR(mailbox);
+		goto err_table;
+	}
+	mpt_entry = mailbox->buf;
+	mpt_entry->flags = cpu_to_be32(MLX4_MPT_FLAG_MIO	 |
+				       MLX4_MPT_FLAG_REGION	 |
+				       mr->access);
+
+	mpt_entry->key	       = cpu_to_be32(key_to_hw_index(mr->key));
+	mpt_entry->pd_flags    = cpu_to_be32(mr->pd | MLX4_MPT_PD_FLAG_EN_INV);
+	mpt_entry->start       = cpu_to_be64(mr->iova);
+	mpt_entry->length      = cpu_to_be64(mr->size);
+	mpt_entry->entity_size = cpu_to_be32(mr->mtt.page_shift);
+
+	if (mr->mtt.order < 0) {
+		mpt_entry->flags |= cpu_to_be32(MLX4_MPT_FLAG_PHYSICAL);
+		mpt_entry->mtt_addr = 0;
+	} else {
+		mpt_entry->mtt_addr = cpu_to_be64(mlx4_mtt_addr(dev,
+						  &mr->mtt));
+	}
+
+	if (mr->mtt.order >= 0 && mr->mtt.page_shift == 0) {
+		/* fast register MR in free state */
+		mpt_entry->flags    |= cpu_to_be32(MLX4_MPT_FLAG_FREE);
+		mpt_entry->pd_flags |= cpu_to_be32(MLX4_MPT_PD_FLAG_FAST_REG |
+						   MLX4_MPT_PD_FLAG_RAE);
+		mpt_entry->mtt_sz    = cpu_to_be32(1 << mr->mtt.order);
+	} else {
+		mpt_entry->flags    |= cpu_to_be32(MLX4_MPT_FLAG_SW_OWNS);
+	}
+
+	err = mlx4_SW2HW_MPT(dev, mailbox,
+			     key_to_hw_index(mr->key) & (dev->caps.num_mpts - 1));
+	if (err) {
+		mlx4_warn(dev, "SW2HW_MPT failed (%d)\n", err);
+		goto err_cmd;
+	}
+	mr->enabled = MLX4_MPT_EN_HW;
+
+	mlx4_free_cmd_mailbox(dev, mailbox);
+
+	return 0;
+
+err_cmd:
+	mlx4_free_cmd_mailbox(dev, mailbox);
+
+err_table:
+	mlx4_mpt_free_icm(dev, key_to_hw_index(mr->key));
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx4_mr_enable);
+
+static int mlx4_write_mtt_chunk(struct mlx4_dev *dev, struct mlx4_mtt *mtt,
+				int start_index, int npages, u64 *page_list)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	__be64 *mtts;
+	dma_addr_t dma_handle;
+	int i;
+
+	mtts = mlx4_table_find(&priv->mr_table.mtt_table, mtt->offset +
+			       start_index, &dma_handle);
+
+	if (!mtts)
+		return -ENOMEM;
+
+	dma_sync_single_for_cpu(&dev->persist->pdev->dev, dma_handle,
+				npages * sizeof(u64), DMA_TO_DEVICE);
+
+	for (i = 0; i < npages; ++i)
+		mtts[i] = cpu_to_be64(page_list[i] | MLX4_MTT_FLAG_PRESENT);
+
+	dma_sync_single_for_device(&dev->persist->pdev->dev, dma_handle,
+				   npages * sizeof(u64), DMA_TO_DEVICE);
+
+	return 0;
+}
+
+int __mlx4_write_mtt(struct mlx4_dev *dev, struct mlx4_mtt *mtt,
+		     int start_index, int npages, u64 *page_list)
+{
+	int err = 0;
+	int chunk;
+	int mtts_per_page;
+	int max_mtts_first_page;
+
+	/* compute how may mtts fit in the first page */
+	mtts_per_page = PAGE_SIZE / sizeof(u64);
+	max_mtts_first_page = mtts_per_page - (mtt->offset + start_index)
+			      % mtts_per_page;
+
+	chunk = min_t(int, max_mtts_first_page, npages);
+
+	while (npages > 0) {
+		err = mlx4_write_mtt_chunk(dev, mtt, start_index, chunk, page_list);
+		if (err)
+			return err;
+		npages      -= chunk;
+		start_index += chunk;
+		page_list   += chunk;
+
+		chunk = min_t(int, mtts_per_page, npages);
+	}
+	return err;
+}
+
+int mlx4_write_mtt(struct mlx4_dev *dev, struct mlx4_mtt *mtt,
+		   int start_index, int npages, u64 *page_list)
+{
+	struct mlx4_cmd_mailbox *mailbox = NULL;
+	__be64 *inbox = NULL;
+	int chunk;
+	int err = 0;
+	int i;
+
+	if (mtt->order < 0)
+		return -EINVAL;
+
+	if (mlx4_is_mfunc(dev)) {
+		mailbox = mlx4_alloc_cmd_mailbox(dev);
+		if (IS_ERR(mailbox))
+			return PTR_ERR(mailbox);
+		inbox = mailbox->buf;
+
+		while (npages > 0) {
+			chunk = min_t(int, MLX4_MAILBOX_SIZE / sizeof(u64) - 2,
+				      npages);
+			inbox[0] = cpu_to_be64(mtt->offset + start_index);
+			inbox[1] = 0;
+			for (i = 0; i < chunk; ++i)
+				inbox[i + 2] = cpu_to_be64(page_list[i] |
+					       MLX4_MTT_FLAG_PRESENT);
+			err = mlx4_WRITE_MTT(dev, mailbox, chunk);
+			if (err) {
+				mlx4_free_cmd_mailbox(dev, mailbox);
+				return err;
+			}
+
+			npages      -= chunk;
+			start_index += chunk;
+			page_list   += chunk;
+		}
+		mlx4_free_cmd_mailbox(dev, mailbox);
+		return err;
+	}
+
+	return __mlx4_write_mtt(dev, mtt, start_index, npages, page_list);
+}
+EXPORT_SYMBOL_GPL(mlx4_write_mtt);
+
+int mlx4_buf_write_mtt(struct mlx4_dev *dev, struct mlx4_mtt *mtt,
+		       struct mlx4_buf *buf)
+{
+	u64 *page_list;
+	int err;
+	int i;
+
+	page_list = kcalloc(buf->npages, sizeof(*page_list), GFP_KERNEL);
+	if (!page_list)
+		return -ENOMEM;
+
+	for (i = 0; i < buf->npages; ++i)
+		if (buf->nbufs == 1)
+			page_list[i] = buf->direct.map + (i << buf->page_shift);
+		else
+			page_list[i] = buf->page_list[i].map;
+
+	err = mlx4_write_mtt(dev, mtt, 0, buf->npages, page_list);
+
+	kfree(page_list);
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx4_buf_write_mtt);
+
+int mlx4_mw_alloc(struct mlx4_dev *dev, u32 pd, enum mlx4_mw_type type,
+		  struct mlx4_mw *mw)
+{
+	u32 index;
+
+	if ((type == MLX4_MW_TYPE_1 &&
+	     !(dev->caps.flags & MLX4_DEV_CAP_FLAG_MEM_WINDOW)) ||
+	     (type == MLX4_MW_TYPE_2 &&
+	     !(dev->caps.bmme_flags & MLX4_BMME_FLAG_TYPE_2_WIN)))
+		return -EOPNOTSUPP;
+
+	index = mlx4_mpt_reserve(dev);
+	if (index == -1)
+		return -ENOMEM;
+
+	mw->key	    = hw_index_to_key(index);
+	mw->pd      = pd;
+	mw->type    = type;
+	mw->enabled = MLX4_MPT_DISABLED;
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(mlx4_mw_alloc);
+
+int mlx4_mw_enable(struct mlx4_dev *dev, struct mlx4_mw *mw)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	struct mlx4_mpt_entry *mpt_entry;
+	int err;
+
+	err = mlx4_mpt_alloc_icm(dev, key_to_hw_index(mw->key));
+	if (err)
+		return err;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox)) {
+		err = PTR_ERR(mailbox);
+		goto err_table;
+	}
+	mpt_entry = mailbox->buf;
+
+	/* Note that the MLX4_MPT_FLAG_REGION bit in mpt_entry->flags is turned
+	 * off, thus creating a memory window and not a memory region.
+	 */
+	mpt_entry->key	       = cpu_to_be32(key_to_hw_index(mw->key));
+	mpt_entry->pd_flags    = cpu_to_be32(mw->pd);
+	if (mw->type == MLX4_MW_TYPE_2) {
+		mpt_entry->flags    |= cpu_to_be32(MLX4_MPT_FLAG_FREE);
+		mpt_entry->qpn       = cpu_to_be32(MLX4_MPT_QP_FLAG_BOUND_QP);
+		mpt_entry->pd_flags |= cpu_to_be32(MLX4_MPT_PD_FLAG_EN_INV);
+	}
+
+	err = mlx4_SW2HW_MPT(dev, mailbox,
+			     key_to_hw_index(mw->key) &
+			     (dev->caps.num_mpts - 1));
+	if (err) {
+		mlx4_warn(dev, "SW2HW_MPT failed (%d)\n", err);
+		goto err_cmd;
+	}
+	mw->enabled = MLX4_MPT_EN_HW;
+
+	mlx4_free_cmd_mailbox(dev, mailbox);
+
+	return 0;
+
+err_cmd:
+	mlx4_free_cmd_mailbox(dev, mailbox);
+
+err_table:
+	mlx4_mpt_free_icm(dev, key_to_hw_index(mw->key));
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx4_mw_enable);
+
+void mlx4_mw_free(struct mlx4_dev *dev, struct mlx4_mw *mw)
+{
+	int err;
+
+	if (mw->enabled == MLX4_MPT_EN_HW) {
+		err = mlx4_HW2SW_MPT(dev, NULL,
+				     key_to_hw_index(mw->key) &
+				     (dev->caps.num_mpts - 1));
+		if (err)
+			mlx4_warn(dev, "xxx HW2SW_MPT failed (%d)\n", err);
+
+		mw->enabled = MLX4_MPT_EN_SW;
+	}
+	if (mw->enabled)
+		mlx4_mpt_free_icm(dev, key_to_hw_index(mw->key));
+	mlx4_mpt_release(dev, key_to_hw_index(mw->key));
+}
+EXPORT_SYMBOL_GPL(mlx4_mw_free);
+
+int mlx4_init_mr_table(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_mr_table *mr_table = &priv->mr_table;
+	int err;
+
+	/* Nothing to do for slaves - all MR handling is forwarded
+	* to the master */
+	if (mlx4_is_slave(dev))
+		return 0;
+
+	if (!is_power_of_2(dev->caps.num_mpts))
+		return -EINVAL;
+
+	err = mlx4_bitmap_init(&mr_table->mpt_bitmap, dev->caps.num_mpts,
+			       ~0, dev->caps.reserved_mrws, 0);
+	if (err)
+		return err;
+
+	err = mlx4_buddy_init(&mr_table->mtt_buddy,
+			      ilog2((u32)dev->caps.num_mtts /
+			      (1 << log_mtts_per_seg)));
+	if (err)
+		goto err_buddy;
+
+	if (dev->caps.reserved_mtts) {
+		priv->reserved_mtts =
+			mlx4_alloc_mtt_range(dev,
+					     fls(dev->caps.reserved_mtts - 1));
+		if (priv->reserved_mtts < 0) {
+			mlx4_warn(dev, "MTT table of order %u is too small\n",
+				  mr_table->mtt_buddy.max_order);
+			err = -ENOMEM;
+			goto err_reserve_mtts;
+		}
+	}
+
+	return 0;
+
+err_reserve_mtts:
+	mlx4_buddy_cleanup(&mr_table->mtt_buddy);
+
+err_buddy:
+	mlx4_bitmap_cleanup(&mr_table->mpt_bitmap);
+
+	return err;
+}
+
+void mlx4_cleanup_mr_table(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_mr_table *mr_table = &priv->mr_table;
+
+	if (mlx4_is_slave(dev))
+		return;
+	if (priv->reserved_mtts >= 0)
+		mlx4_free_mtt_range(dev, priv->reserved_mtts,
+				    fls(dev->caps.reserved_mtts - 1));
+	mlx4_buddy_cleanup(&mr_table->mtt_buddy);
+	mlx4_bitmap_cleanup(&mr_table->mpt_bitmap);
+}
+
+int mlx4_SYNC_TPT(struct mlx4_dev *dev)
+{
+	return mlx4_cmd(dev, 0, 0, 0, MLX4_CMD_SYNC_TPT,
+			MLX4_CMD_TIME_CLASS_A, MLX4_CMD_NATIVE);
+}
+EXPORT_SYMBOL_GPL(mlx4_SYNC_TPT);
diff --git a/drivers/net/ethernet/mellanox/mlx4/pd.c b/drivers/net/ethernet/mellanox/mlx4/pd.c
new file mode 100644
index 000000000..6fc156a39
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx4/pd.c
@@ -0,0 +1,295 @@
+/*
+ * Copyright (c) 2006, 2007 Cisco Systems, Inc.  All rights reserved.
+ * Copyright (c) 2005 Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/errno.h>
+#include <linux/export.h>
+#include <linux/io-mapping.h>
+
+#include <asm/page.h>
+
+#include "mlx4.h"
+#include "icm.h"
+
+enum {
+	MLX4_NUM_RESERVED_UARS = 8
+};
+
+int mlx4_pd_alloc(struct mlx4_dev *dev, u32 *pdn)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	*pdn = mlx4_bitmap_alloc(&priv->pd_bitmap);
+	if (*pdn == -1)
+		return -ENOMEM;
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(mlx4_pd_alloc);
+
+void mlx4_pd_free(struct mlx4_dev *dev, u32 pdn)
+{
+	mlx4_bitmap_free(&mlx4_priv(dev)->pd_bitmap, pdn, MLX4_USE_RR);
+}
+EXPORT_SYMBOL_GPL(mlx4_pd_free);
+
+int __mlx4_xrcd_alloc(struct mlx4_dev *dev, u32 *xrcdn)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	*xrcdn = mlx4_bitmap_alloc(&priv->xrcd_bitmap);
+	if (*xrcdn == -1)
+		return -ENOMEM;
+
+	return 0;
+}
+
+int mlx4_xrcd_alloc(struct mlx4_dev *dev, u32 *xrcdn)
+{
+	u64 out_param;
+	int err;
+
+	if (mlx4_is_mfunc(dev)) {
+		err = mlx4_cmd_imm(dev, 0, &out_param,
+				   RES_XRCD, RES_OP_RESERVE,
+				   MLX4_CMD_ALLOC_RES,
+				   MLX4_CMD_TIME_CLASS_A, MLX4_CMD_WRAPPED);
+		if (err)
+			return err;
+
+		*xrcdn = get_param_l(&out_param);
+		return 0;
+	}
+	return __mlx4_xrcd_alloc(dev, xrcdn);
+}
+EXPORT_SYMBOL_GPL(mlx4_xrcd_alloc);
+
+void __mlx4_xrcd_free(struct mlx4_dev *dev, u32 xrcdn)
+{
+	mlx4_bitmap_free(&mlx4_priv(dev)->xrcd_bitmap, xrcdn, MLX4_USE_RR);
+}
+
+void mlx4_xrcd_free(struct mlx4_dev *dev, u32 xrcdn)
+{
+	u64 in_param = 0;
+	int err;
+
+	if (mlx4_is_mfunc(dev)) {
+		set_param_l(&in_param, xrcdn);
+		err = mlx4_cmd(dev, in_param, RES_XRCD,
+			       RES_OP_RESERVE, MLX4_CMD_FREE_RES,
+			       MLX4_CMD_TIME_CLASS_A, MLX4_CMD_WRAPPED);
+		if (err)
+			mlx4_warn(dev, "Failed to release xrcdn %d\n", xrcdn);
+	} else
+		__mlx4_xrcd_free(dev, xrcdn);
+}
+EXPORT_SYMBOL_GPL(mlx4_xrcd_free);
+
+int mlx4_init_pd_table(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	return mlx4_bitmap_init(&priv->pd_bitmap, dev->caps.num_pds,
+				(1 << NOT_MASKED_PD_BITS) - 1,
+				 dev->caps.reserved_pds, 0);
+}
+
+void mlx4_cleanup_pd_table(struct mlx4_dev *dev)
+{
+	mlx4_bitmap_cleanup(&mlx4_priv(dev)->pd_bitmap);
+}
+
+int mlx4_init_xrcd_table(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	return mlx4_bitmap_init(&priv->xrcd_bitmap, (1 << 16),
+				(1 << 16) - 1, dev->caps.reserved_xrcds + 1, 0);
+}
+
+void mlx4_cleanup_xrcd_table(struct mlx4_dev *dev)
+{
+	mlx4_bitmap_cleanup(&mlx4_priv(dev)->xrcd_bitmap);
+}
+
+int mlx4_uar_alloc(struct mlx4_dev *dev, struct mlx4_uar *uar)
+{
+	int offset;
+
+	uar->index = mlx4_bitmap_alloc(&mlx4_priv(dev)->uar_table.bitmap);
+	if (uar->index == -1)
+		return -ENOMEM;
+
+	if (mlx4_is_slave(dev))
+		offset = uar->index % ((int)pci_resource_len(dev->persist->pdev,
+							     2) /
+				       dev->caps.uar_page_size);
+	else
+		offset = uar->index;
+	uar->pfn = (pci_resource_start(dev->persist->pdev, 2) >> PAGE_SHIFT)
+		    + offset;
+	uar->map = NULL;
+	return 0;
+}
+EXPORT_SYMBOL_GPL(mlx4_uar_alloc);
+
+void mlx4_uar_free(struct mlx4_dev *dev, struct mlx4_uar *uar)
+{
+	mlx4_bitmap_free(&mlx4_priv(dev)->uar_table.bitmap, uar->index, MLX4_USE_RR);
+}
+EXPORT_SYMBOL_GPL(mlx4_uar_free);
+
+int mlx4_bf_alloc(struct mlx4_dev *dev, struct mlx4_bf *bf, int node)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_uar *uar;
+	int err = 0;
+	int idx;
+
+	if (!priv->bf_mapping)
+		return -ENOMEM;
+
+	mutex_lock(&priv->bf_mutex);
+	if (!list_empty(&priv->bf_list))
+		uar = list_entry(priv->bf_list.next, struct mlx4_uar, bf_list);
+	else {
+		if (mlx4_bitmap_avail(&priv->uar_table.bitmap) < MLX4_NUM_RESERVED_UARS) {
+			err = -ENOMEM;
+			goto out;
+		}
+		uar = kmalloc_node(sizeof(*uar), GFP_KERNEL, node);
+		if (!uar) {
+			uar = kmalloc(sizeof(*uar), GFP_KERNEL);
+			if (!uar) {
+				err = -ENOMEM;
+				goto out;
+			}
+		}
+		err = mlx4_uar_alloc(dev, uar);
+		if (err)
+			goto free_kmalloc;
+
+		uar->map = ioremap(uar->pfn << PAGE_SHIFT, PAGE_SIZE);
+		if (!uar->map) {
+			err = -ENOMEM;
+			goto free_uar;
+		}
+
+		uar->bf_map = io_mapping_map_wc(priv->bf_mapping,
+						uar->index << PAGE_SHIFT,
+						PAGE_SIZE);
+		if (!uar->bf_map) {
+			err = -ENOMEM;
+			goto unamp_uar;
+		}
+		uar->free_bf_bmap = 0;
+		list_add(&uar->bf_list, &priv->bf_list);
+	}
+
+	idx = ffz(uar->free_bf_bmap);
+	uar->free_bf_bmap |= 1 << idx;
+	bf->uar = uar;
+	bf->offset = 0;
+	bf->buf_size = dev->caps.bf_reg_size / 2;
+	bf->reg = uar->bf_map + idx * dev->caps.bf_reg_size;
+	if (uar->free_bf_bmap == (1 << dev->caps.bf_regs_per_page) - 1)
+		list_del_init(&uar->bf_list);
+
+	goto out;
+
+unamp_uar:
+	bf->uar = NULL;
+	iounmap(uar->map);
+
+free_uar:
+	mlx4_uar_free(dev, uar);
+
+free_kmalloc:
+	kfree(uar);
+
+out:
+	mutex_unlock(&priv->bf_mutex);
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx4_bf_alloc);
+
+void mlx4_bf_free(struct mlx4_dev *dev, struct mlx4_bf *bf)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int idx;
+
+	if (!bf->uar || !bf->uar->bf_map)
+		return;
+
+	mutex_lock(&priv->bf_mutex);
+	idx = (bf->reg - bf->uar->bf_map) / dev->caps.bf_reg_size;
+	bf->uar->free_bf_bmap &= ~(1 << idx);
+	if (!bf->uar->free_bf_bmap) {
+		if (!list_empty(&bf->uar->bf_list))
+			list_del(&bf->uar->bf_list);
+
+		io_mapping_unmap(bf->uar->bf_map);
+		iounmap(bf->uar->map);
+		mlx4_uar_free(dev, bf->uar);
+		kfree(bf->uar);
+	} else if (list_empty(&bf->uar->bf_list))
+		list_add(&bf->uar->bf_list, &priv->bf_list);
+
+	mutex_unlock(&priv->bf_mutex);
+}
+EXPORT_SYMBOL_GPL(mlx4_bf_free);
+
+int mlx4_init_uar_table(struct mlx4_dev *dev)
+{
+	int num_reserved_uar = mlx4_get_num_reserved_uar(dev);
+
+	mlx4_dbg(dev, "uar_page_shift = %d", dev->uar_page_shift);
+	mlx4_dbg(dev, "Effective reserved_uars=%d", dev->caps.reserved_uars);
+
+	if (dev->caps.num_uars <= num_reserved_uar) {
+		mlx4_err(
+			dev, "Only %d UAR pages (need more than %d)\n",
+			dev->caps.num_uars, num_reserved_uar);
+		mlx4_err(dev, "Increase firmware log2_uar_bar_megabytes?\n");
+		return -ENODEV;
+	}
+
+	return mlx4_bitmap_init(&mlx4_priv(dev)->uar_table.bitmap,
+				dev->caps.num_uars, dev->caps.num_uars - 1,
+				dev->caps.reserved_uars, 0);
+}
+
+void mlx4_cleanup_uar_table(struct mlx4_dev *dev)
+{
+	mlx4_bitmap_cleanup(&mlx4_priv(dev)->uar_table.bitmap);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx4/port.c b/drivers/net/ethernet/mellanox/mlx4/port.c
new file mode 100644
index 000000000..256a06b3c
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx4/port.c
@@ -0,0 +1,2229 @@
+/*
+ * Copyright (c) 2007 Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/errno.h>
+#include <linux/if_ether.h>
+#include <linux/if_vlan.h>
+#include <linux/export.h>
+
+#include <linux/mlx4/cmd.h>
+
+#include "mlx4.h"
+#include "mlx4_stats.h"
+
+#define MLX4_MAC_VALID		(1ull << 63)
+
+#define MLX4_VLAN_VALID		(1u << 31)
+#define MLX4_VLAN_MASK		0xfff
+
+#define MLX4_STATS_TRAFFIC_COUNTERS_MASK	0xfULL
+#define MLX4_STATS_TRAFFIC_DROPS_MASK		0xc0ULL
+#define MLX4_STATS_ERROR_COUNTERS_MASK		0x1ffc30ULL
+#define MLX4_STATS_PORT_COUNTERS_MASK		0x1fe00000ULL
+
+#define MLX4_FLAG2_V_IGNORE_FCS_MASK		BIT(1)
+#define MLX4_FLAG2_V_USER_MTU_MASK		BIT(5)
+#define MLX4_FLAG2_V_USER_MAC_MASK		BIT(6)
+#define MLX4_FLAG_V_MTU_MASK			BIT(0)
+#define MLX4_FLAG_V_PPRX_MASK			BIT(1)
+#define MLX4_FLAG_V_PPTX_MASK			BIT(2)
+#define MLX4_IGNORE_FCS_MASK			0x1
+#define MLX4_TC_MAX_NUMBER			8
+
+void mlx4_init_mac_table(struct mlx4_dev *dev, struct mlx4_mac_table *table)
+{
+	int i;
+
+	mutex_init(&table->mutex);
+	for (i = 0; i < MLX4_MAX_MAC_NUM; i++) {
+		table->entries[i] = 0;
+		table->refs[i]	 = 0;
+		table->is_dup[i] = false;
+	}
+	table->max   = 1 << dev->caps.log_num_macs;
+	table->total = 0;
+}
+
+void mlx4_init_vlan_table(struct mlx4_dev *dev, struct mlx4_vlan_table *table)
+{
+	int i;
+
+	mutex_init(&table->mutex);
+	for (i = 0; i < MLX4_MAX_VLAN_NUM; i++) {
+		table->entries[i] = 0;
+		table->refs[i]	 = 0;
+		table->is_dup[i] = false;
+	}
+	table->max   = (1 << dev->caps.log_num_vlans) - MLX4_VLAN_REGULAR;
+	table->total = 0;
+}
+
+void mlx4_init_roce_gid_table(struct mlx4_dev *dev,
+			      struct mlx4_roce_gid_table *table)
+{
+	int i;
+
+	mutex_init(&table->mutex);
+	for (i = 0; i < MLX4_ROCE_MAX_GIDS; i++)
+		memset(table->roce_gids[i].raw, 0, MLX4_ROCE_GID_ENTRY_SIZE);
+}
+
+static int validate_index(struct mlx4_dev *dev,
+			  struct mlx4_mac_table *table, int index)
+{
+	int err = 0;
+
+	if (index < 0 || index >= table->max || !table->entries[index]) {
+		mlx4_warn(dev, "No valid Mac entry for the given index\n");
+		err = -EINVAL;
+	}
+	return err;
+}
+
+static int find_index(struct mlx4_dev *dev,
+		      struct mlx4_mac_table *table, u64 mac)
+{
+	int i;
+
+	for (i = 0; i < MLX4_MAX_MAC_NUM; i++) {
+		if (table->refs[i] &&
+		    (MLX4_MAC_MASK & mac) ==
+		    (MLX4_MAC_MASK & be64_to_cpu(table->entries[i])))
+			return i;
+	}
+	/* Mac not found */
+	return -EINVAL;
+}
+
+static int mlx4_set_port_mac_table(struct mlx4_dev *dev, u8 port,
+				   __be64 *entries)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	u32 in_mod;
+	int err;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+
+	memcpy(mailbox->buf, entries, MLX4_MAC_TABLE_SIZE);
+
+	in_mod = MLX4_SET_PORT_MAC_TABLE << 8 | port;
+
+	err = mlx4_cmd(dev, mailbox->dma, in_mod, MLX4_SET_PORT_ETH_OPCODE,
+		       MLX4_CMD_SET_PORT, MLX4_CMD_TIME_CLASS_B,
+		       MLX4_CMD_NATIVE);
+
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return err;
+}
+
+int mlx4_find_cached_mac(struct mlx4_dev *dev, u8 port, u64 mac, int *idx)
+{
+	struct mlx4_port_info *info = &mlx4_priv(dev)->port[port];
+	struct mlx4_mac_table *table = &info->mac_table;
+	int i;
+
+	for (i = 0; i < MLX4_MAX_MAC_NUM; i++) {
+		if (!table->refs[i])
+			continue;
+
+		if (mac == (MLX4_MAC_MASK & be64_to_cpu(table->entries[i]))) {
+			*idx = i;
+			return 0;
+		}
+	}
+
+	return -ENOENT;
+}
+EXPORT_SYMBOL_GPL(mlx4_find_cached_mac);
+
+static bool mlx4_need_mf_bond(struct mlx4_dev *dev)
+{
+	int i, num_eth_ports = 0;
+
+	if (!mlx4_is_mfunc(dev))
+		return false;
+	mlx4_foreach_port(i, dev, MLX4_PORT_TYPE_ETH)
+		++num_eth_ports;
+
+	return (num_eth_ports ==  2) ? true : false;
+}
+
+int __mlx4_register_mac(struct mlx4_dev *dev, u8 port, u64 mac)
+{
+	struct mlx4_port_info *info = &mlx4_priv(dev)->port[port];
+	struct mlx4_mac_table *table = &info->mac_table;
+	int i, err = 0;
+	int free = -1;
+	int free_for_dup = -1;
+	bool dup = mlx4_is_mf_bonded(dev);
+	u8 dup_port = (port == 1) ? 2 : 1;
+	struct mlx4_mac_table *dup_table = &mlx4_priv(dev)->port[dup_port].mac_table;
+	bool need_mf_bond = mlx4_need_mf_bond(dev);
+	bool can_mf_bond = true;
+
+	mlx4_dbg(dev, "Registering MAC: 0x%llx for port %d %s duplicate\n",
+		 (unsigned long long)mac, port,
+		 dup ? "with" : "without");
+
+	if (need_mf_bond) {
+		if (port == 1) {
+			mutex_lock(&table->mutex);
+			mutex_lock_nested(&dup_table->mutex, SINGLE_DEPTH_NESTING);
+		} else {
+			mutex_lock(&dup_table->mutex);
+			mutex_lock_nested(&table->mutex, SINGLE_DEPTH_NESTING);
+		}
+	} else {
+		mutex_lock(&table->mutex);
+	}
+
+	if (need_mf_bond) {
+		int index_at_port = -1;
+		int index_at_dup_port = -1;
+
+		for (i = 0; i < MLX4_MAX_MAC_NUM; i++) {
+			if (((MLX4_MAC_MASK & mac) == (MLX4_MAC_MASK & be64_to_cpu(table->entries[i]))))
+				index_at_port = i;
+			if (((MLX4_MAC_MASK & mac) == (MLX4_MAC_MASK & be64_to_cpu(dup_table->entries[i]))))
+				index_at_dup_port = i;
+		}
+
+		/* check that same mac is not in the tables at different indices */
+		if ((index_at_port != index_at_dup_port) &&
+		    (index_at_port >= 0) &&
+		    (index_at_dup_port >= 0))
+			can_mf_bond = false;
+
+		/* If the mac is already in the primary table, the slot must be
+		 * available in the duplicate table as well.
+		 */
+		if (index_at_port >= 0 && index_at_dup_port < 0 &&
+		    dup_table->refs[index_at_port]) {
+			can_mf_bond = false;
+		}
+		/* If the mac is already in the duplicate table, check that the
+		 * corresponding index is not occupied in the primary table, or
+		 * the primary table already contains the mac at the same index.
+		 * Otherwise, you cannot bond (primary contains a different mac
+		 * at that index).
+		 */
+		if (index_at_dup_port >= 0) {
+			if (!table->refs[index_at_dup_port] ||
+			    ((MLX4_MAC_MASK & mac) == (MLX4_MAC_MASK & be64_to_cpu(table->entries[index_at_dup_port]))))
+				free_for_dup = index_at_dup_port;
+			else
+				can_mf_bond = false;
+		}
+	}
+
+	for (i = 0; i < MLX4_MAX_MAC_NUM; i++) {
+		if (!table->refs[i]) {
+			if (free < 0)
+				free = i;
+			if (free_for_dup < 0 && need_mf_bond && can_mf_bond) {
+				if (!dup_table->refs[i])
+					free_for_dup = i;
+			}
+			continue;
+		}
+
+		if ((MLX4_MAC_MASK & mac) ==
+		     (MLX4_MAC_MASK & be64_to_cpu(table->entries[i]))) {
+			/* MAC already registered, increment ref count */
+			err = i;
+			++table->refs[i];
+			if (dup) {
+				u64 dup_mac = MLX4_MAC_MASK & be64_to_cpu(dup_table->entries[i]);
+
+				if (dup_mac != mac || !dup_table->is_dup[i]) {
+					mlx4_warn(dev, "register mac: expect duplicate mac 0x%llx on port %d index %d\n",
+						  mac, dup_port, i);
+				}
+			}
+			goto out;
+		}
+	}
+
+	if (need_mf_bond && (free_for_dup < 0)) {
+		if (dup) {
+			mlx4_warn(dev, "Fail to allocate duplicate MAC table entry\n");
+			mlx4_warn(dev, "High Availability for virtual functions may not work as expected\n");
+			dup = false;
+		}
+		can_mf_bond = false;
+	}
+
+	if (need_mf_bond && can_mf_bond)
+		free = free_for_dup;
+
+	mlx4_dbg(dev, "Free MAC index is %d\n", free);
+
+	if (table->total == table->max) {
+		/* No free mac entries */
+		err = -ENOSPC;
+		goto out;
+	}
+
+	/* Register new MAC */
+	table->entries[free] = cpu_to_be64(mac | MLX4_MAC_VALID);
+
+	err = mlx4_set_port_mac_table(dev, port, table->entries);
+	if (unlikely(err)) {
+		mlx4_err(dev, "Failed adding MAC: 0x%llx\n",
+			 (unsigned long long) mac);
+		table->entries[free] = 0;
+		goto out;
+	}
+	table->refs[free] = 1;
+	table->is_dup[free] = false;
+	++table->total;
+	if (dup) {
+		dup_table->refs[free] = 0;
+		dup_table->is_dup[free] = true;
+		dup_table->entries[free] = cpu_to_be64(mac | MLX4_MAC_VALID);
+
+		err = mlx4_set_port_mac_table(dev, dup_port, dup_table->entries);
+		if (unlikely(err)) {
+			mlx4_warn(dev, "Failed adding duplicate mac: 0x%llx\n", mac);
+			dup_table->is_dup[free] = false;
+			dup_table->entries[free] = 0;
+			goto out;
+		}
+		++dup_table->total;
+	}
+	err = free;
+out:
+	if (need_mf_bond) {
+		if (port == 2) {
+			mutex_unlock(&table->mutex);
+			mutex_unlock(&dup_table->mutex);
+		} else {
+			mutex_unlock(&dup_table->mutex);
+			mutex_unlock(&table->mutex);
+		}
+	} else {
+		mutex_unlock(&table->mutex);
+	}
+	return err;
+}
+EXPORT_SYMBOL_GPL(__mlx4_register_mac);
+
+int mlx4_register_mac(struct mlx4_dev *dev, u8 port, u64 mac)
+{
+	u64 out_param = 0;
+	int err = -EINVAL;
+
+	if (mlx4_is_mfunc(dev)) {
+		if (!(dev->flags & MLX4_FLAG_OLD_REG_MAC)) {
+			err = mlx4_cmd_imm(dev, mac, &out_param,
+					   ((u32) port) << 8 | (u32) RES_MAC,
+					   RES_OP_RESERVE_AND_MAP, MLX4_CMD_ALLOC_RES,
+					   MLX4_CMD_TIME_CLASS_A, MLX4_CMD_WRAPPED);
+		}
+		if (err && err == -EINVAL && mlx4_is_slave(dev)) {
+			/* retry using old REG_MAC format */
+			set_param_l(&out_param, port);
+			err = mlx4_cmd_imm(dev, mac, &out_param, RES_MAC,
+					   RES_OP_RESERVE_AND_MAP, MLX4_CMD_ALLOC_RES,
+					   MLX4_CMD_TIME_CLASS_A, MLX4_CMD_WRAPPED);
+			if (!err)
+				dev->flags |= MLX4_FLAG_OLD_REG_MAC;
+		}
+		if (err)
+			return err;
+
+		return get_param_l(&out_param);
+	}
+	return __mlx4_register_mac(dev, port, mac);
+}
+EXPORT_SYMBOL_GPL(mlx4_register_mac);
+
+int mlx4_get_base_qpn(struct mlx4_dev *dev, u8 port)
+{
+	return dev->caps.reserved_qps_base[MLX4_QP_REGION_ETH_ADDR] +
+			(port - 1) * (1 << dev->caps.log_num_macs);
+}
+EXPORT_SYMBOL_GPL(mlx4_get_base_qpn);
+
+void __mlx4_unregister_mac(struct mlx4_dev *dev, u8 port, u64 mac)
+{
+	struct mlx4_port_info *info;
+	struct mlx4_mac_table *table;
+	int index;
+	bool dup = mlx4_is_mf_bonded(dev);
+	u8 dup_port = (port == 1) ? 2 : 1;
+	struct mlx4_mac_table *dup_table = &mlx4_priv(dev)->port[dup_port].mac_table;
+
+	if (port < 1 || port > dev->caps.num_ports) {
+		mlx4_warn(dev, "invalid port number (%d), aborting...\n", port);
+		return;
+	}
+	info = &mlx4_priv(dev)->port[port];
+	table = &info->mac_table;
+
+	if (dup) {
+		if (port == 1) {
+			mutex_lock(&table->mutex);
+			mutex_lock_nested(&dup_table->mutex, SINGLE_DEPTH_NESTING);
+		} else {
+			mutex_lock(&dup_table->mutex);
+			mutex_lock_nested(&table->mutex, SINGLE_DEPTH_NESTING);
+		}
+	} else {
+		mutex_lock(&table->mutex);
+	}
+
+	index = find_index(dev, table, mac);
+
+	if (validate_index(dev, table, index))
+		goto out;
+
+	if (--table->refs[index] || table->is_dup[index]) {
+		mlx4_dbg(dev, "Have more references for index %d, no need to modify mac table\n",
+			 index);
+		if (!table->refs[index])
+			dup_table->is_dup[index] = false;
+		goto out;
+	}
+
+	table->entries[index] = 0;
+	if (mlx4_set_port_mac_table(dev, port, table->entries))
+		mlx4_warn(dev, "Fail to set mac in port %d during unregister\n", port);
+	--table->total;
+
+	if (dup) {
+		dup_table->is_dup[index] = false;
+		if (dup_table->refs[index])
+			goto out;
+		dup_table->entries[index] = 0;
+		if (mlx4_set_port_mac_table(dev, dup_port, dup_table->entries))
+			mlx4_warn(dev, "Fail to set mac in duplicate port %d during unregister\n", dup_port);
+
+		--table->total;
+	}
+out:
+	if (dup) {
+		if (port == 2) {
+			mutex_unlock(&table->mutex);
+			mutex_unlock(&dup_table->mutex);
+		} else {
+			mutex_unlock(&dup_table->mutex);
+			mutex_unlock(&table->mutex);
+		}
+	} else {
+		mutex_unlock(&table->mutex);
+	}
+}
+EXPORT_SYMBOL_GPL(__mlx4_unregister_mac);
+
+void mlx4_unregister_mac(struct mlx4_dev *dev, u8 port, u64 mac)
+{
+	u64 out_param = 0;
+
+	if (mlx4_is_mfunc(dev)) {
+		if (!(dev->flags & MLX4_FLAG_OLD_REG_MAC)) {
+			(void) mlx4_cmd_imm(dev, mac, &out_param,
+					    ((u32) port) << 8 | (u32) RES_MAC,
+					    RES_OP_RESERVE_AND_MAP, MLX4_CMD_FREE_RES,
+					    MLX4_CMD_TIME_CLASS_A, MLX4_CMD_WRAPPED);
+		} else {
+			/* use old unregister mac format */
+			set_param_l(&out_param, port);
+			(void) mlx4_cmd_imm(dev, mac, &out_param, RES_MAC,
+					    RES_OP_RESERVE_AND_MAP, MLX4_CMD_FREE_RES,
+					    MLX4_CMD_TIME_CLASS_A, MLX4_CMD_WRAPPED);
+		}
+		return;
+	}
+	__mlx4_unregister_mac(dev, port, mac);
+	return;
+}
+EXPORT_SYMBOL_GPL(mlx4_unregister_mac);
+
+int __mlx4_replace_mac(struct mlx4_dev *dev, u8 port, int qpn, u64 new_mac)
+{
+	struct mlx4_port_info *info = &mlx4_priv(dev)->port[port];
+	struct mlx4_mac_table *table = &info->mac_table;
+	int index = qpn - info->base_qpn;
+	int err = 0;
+	bool dup = mlx4_is_mf_bonded(dev);
+	u8 dup_port = (port == 1) ? 2 : 1;
+	struct mlx4_mac_table *dup_table = &mlx4_priv(dev)->port[dup_port].mac_table;
+
+	/* CX1 doesn't support multi-functions */
+	if (dup) {
+		if (port == 1) {
+			mutex_lock(&table->mutex);
+			mutex_lock_nested(&dup_table->mutex, SINGLE_DEPTH_NESTING);
+		} else {
+			mutex_lock(&dup_table->mutex);
+			mutex_lock_nested(&table->mutex, SINGLE_DEPTH_NESTING);
+		}
+	} else {
+		mutex_lock(&table->mutex);
+	}
+
+	err = validate_index(dev, table, index);
+	if (err)
+		goto out;
+
+	table->entries[index] = cpu_to_be64(new_mac | MLX4_MAC_VALID);
+
+	err = mlx4_set_port_mac_table(dev, port, table->entries);
+	if (unlikely(err)) {
+		mlx4_err(dev, "Failed adding MAC: 0x%llx\n",
+			 (unsigned long long) new_mac);
+		table->entries[index] = 0;
+	} else {
+		if (dup) {
+			dup_table->entries[index] = cpu_to_be64(new_mac | MLX4_MAC_VALID);
+
+			err = mlx4_set_port_mac_table(dev, dup_port, dup_table->entries);
+			if (unlikely(err)) {
+				mlx4_err(dev, "Failed adding duplicate MAC: 0x%llx\n",
+					 (unsigned long long)new_mac);
+				dup_table->entries[index] = 0;
+			}
+		}
+	}
+out:
+	if (dup) {
+		if (port == 2) {
+			mutex_unlock(&table->mutex);
+			mutex_unlock(&dup_table->mutex);
+		} else {
+			mutex_unlock(&dup_table->mutex);
+			mutex_unlock(&table->mutex);
+		}
+	} else {
+		mutex_unlock(&table->mutex);
+	}
+	return err;
+}
+EXPORT_SYMBOL_GPL(__mlx4_replace_mac);
+
+static int mlx4_set_port_vlan_table(struct mlx4_dev *dev, u8 port,
+				    __be32 *entries)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	u32 in_mod;
+	int err;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+
+	memcpy(mailbox->buf, entries, MLX4_VLAN_TABLE_SIZE);
+	in_mod = MLX4_SET_PORT_VLAN_TABLE << 8 | port;
+	err = mlx4_cmd(dev, mailbox->dma, in_mod, MLX4_SET_PORT_ETH_OPCODE,
+		       MLX4_CMD_SET_PORT, MLX4_CMD_TIME_CLASS_B,
+		       MLX4_CMD_NATIVE);
+
+	mlx4_free_cmd_mailbox(dev, mailbox);
+
+	return err;
+}
+
+int mlx4_find_cached_vlan(struct mlx4_dev *dev, u8 port, u16 vid, int *idx)
+{
+	struct mlx4_vlan_table *table = &mlx4_priv(dev)->port[port].vlan_table;
+	int i;
+
+	for (i = 0; i < MLX4_MAX_VLAN_NUM; ++i) {
+		if (table->refs[i] &&
+		    (vid == (MLX4_VLAN_MASK &
+			      be32_to_cpu(table->entries[i])))) {
+			/* VLAN already registered, increase reference count */
+			*idx = i;
+			return 0;
+		}
+	}
+
+	return -ENOENT;
+}
+EXPORT_SYMBOL_GPL(mlx4_find_cached_vlan);
+
+int __mlx4_register_vlan(struct mlx4_dev *dev, u8 port, u16 vlan,
+				int *index)
+{
+	struct mlx4_vlan_table *table = &mlx4_priv(dev)->port[port].vlan_table;
+	int i, err = 0;
+	int free = -1;
+	int free_for_dup = -1;
+	bool dup = mlx4_is_mf_bonded(dev);
+	u8 dup_port = (port == 1) ? 2 : 1;
+	struct mlx4_vlan_table *dup_table = &mlx4_priv(dev)->port[dup_port].vlan_table;
+	bool need_mf_bond = mlx4_need_mf_bond(dev);
+	bool can_mf_bond = true;
+
+	mlx4_dbg(dev, "Registering VLAN: %d for port %d %s duplicate\n",
+		 vlan, port,
+		 dup ? "with" : "without");
+
+	if (need_mf_bond) {
+		if (port == 1) {
+			mutex_lock(&table->mutex);
+			mutex_lock_nested(&dup_table->mutex, SINGLE_DEPTH_NESTING);
+		} else {
+			mutex_lock(&dup_table->mutex);
+			mutex_lock_nested(&table->mutex, SINGLE_DEPTH_NESTING);
+		}
+	} else {
+		mutex_lock(&table->mutex);
+	}
+
+	if (table->total == table->max) {
+		/* No free vlan entries */
+		err = -ENOSPC;
+		goto out;
+	}
+
+	if (need_mf_bond) {
+		int index_at_port = -1;
+		int index_at_dup_port = -1;
+
+		for (i = MLX4_VLAN_REGULAR; i < MLX4_MAX_VLAN_NUM; i++) {
+			if (vlan == (MLX4_VLAN_MASK & be32_to_cpu(table->entries[i])))
+				index_at_port = i;
+			if (vlan == (MLX4_VLAN_MASK & be32_to_cpu(dup_table->entries[i])))
+				index_at_dup_port = i;
+		}
+		/* check that same vlan is not in the tables at different indices */
+		if ((index_at_port != index_at_dup_port) &&
+		    (index_at_port >= 0) &&
+		    (index_at_dup_port >= 0))
+			can_mf_bond = false;
+
+		/* If the vlan is already in the primary table, the slot must be
+		 * available in the duplicate table as well.
+		 */
+		if (index_at_port >= 0 && index_at_dup_port < 0 &&
+		    dup_table->refs[index_at_port]) {
+			can_mf_bond = false;
+		}
+		/* If the vlan is already in the duplicate table, check that the
+		 * corresponding index is not occupied in the primary table, or
+		 * the primary table already contains the vlan at the same index.
+		 * Otherwise, you cannot bond (primary contains a different vlan
+		 * at that index).
+		 */
+		if (index_at_dup_port >= 0) {
+			if (!table->refs[index_at_dup_port] ||
+			    (vlan == (MLX4_VLAN_MASK & be32_to_cpu(dup_table->entries[index_at_dup_port]))))
+				free_for_dup = index_at_dup_port;
+			else
+				can_mf_bond = false;
+		}
+	}
+
+	for (i = MLX4_VLAN_REGULAR; i < MLX4_MAX_VLAN_NUM; i++) {
+		if (!table->refs[i]) {
+			if (free < 0)
+				free = i;
+			if (free_for_dup < 0 && need_mf_bond && can_mf_bond) {
+				if (!dup_table->refs[i])
+					free_for_dup = i;
+			}
+		}
+
+		if ((table->refs[i] || table->is_dup[i]) &&
+		    (vlan == (MLX4_VLAN_MASK &
+			      be32_to_cpu(table->entries[i])))) {
+			/* Vlan already registered, increase references count */
+			mlx4_dbg(dev, "vlan %u is already registered.\n", vlan);
+			*index = i;
+			++table->refs[i];
+			if (dup) {
+				u16 dup_vlan = MLX4_VLAN_MASK & be32_to_cpu(dup_table->entries[i]);
+
+				if (dup_vlan != vlan || !dup_table->is_dup[i]) {
+					mlx4_warn(dev, "register vlan: expected duplicate vlan %u on port %d index %d\n",
+						  vlan, dup_port, i);
+				}
+			}
+			goto out;
+		}
+	}
+
+	if (need_mf_bond && (free_for_dup < 0)) {
+		if (dup) {
+			mlx4_warn(dev, "Fail to allocate duplicate VLAN table entry\n");
+			mlx4_warn(dev, "High Availability for virtual functions may not work as expected\n");
+			dup = false;
+		}
+		can_mf_bond = false;
+	}
+
+	if (need_mf_bond && can_mf_bond)
+		free = free_for_dup;
+
+	if (free < 0) {
+		err = -ENOMEM;
+		goto out;
+	}
+
+	/* Register new VLAN */
+	table->refs[free] = 1;
+	table->is_dup[free] = false;
+	table->entries[free] = cpu_to_be32(vlan | MLX4_VLAN_VALID);
+
+	err = mlx4_set_port_vlan_table(dev, port, table->entries);
+	if (unlikely(err)) {
+		mlx4_warn(dev, "Failed adding vlan: %u\n", vlan);
+		table->refs[free] = 0;
+		table->entries[free] = 0;
+		goto out;
+	}
+	++table->total;
+	if (dup) {
+		dup_table->refs[free] = 0;
+		dup_table->is_dup[free] = true;
+		dup_table->entries[free] = cpu_to_be32(vlan | MLX4_VLAN_VALID);
+
+		err = mlx4_set_port_vlan_table(dev, dup_port, dup_table->entries);
+		if (unlikely(err)) {
+			mlx4_warn(dev, "Failed adding duplicate vlan: %u\n", vlan);
+			dup_table->is_dup[free] = false;
+			dup_table->entries[free] = 0;
+			goto out;
+		}
+		++dup_table->total;
+	}
+
+	*index = free;
+out:
+	if (need_mf_bond) {
+		if (port == 2) {
+			mutex_unlock(&table->mutex);
+			mutex_unlock(&dup_table->mutex);
+		} else {
+			mutex_unlock(&dup_table->mutex);
+			mutex_unlock(&table->mutex);
+		}
+	} else {
+		mutex_unlock(&table->mutex);
+	}
+	return err;
+}
+
+int mlx4_register_vlan(struct mlx4_dev *dev, u8 port, u16 vlan, int *index)
+{
+	u64 out_param = 0;
+	int err;
+
+	if (vlan > 4095)
+		return -EINVAL;
+
+	if (mlx4_is_mfunc(dev)) {
+		err = mlx4_cmd_imm(dev, vlan, &out_param,
+				   ((u32) port) << 8 | (u32) RES_VLAN,
+				   RES_OP_RESERVE_AND_MAP, MLX4_CMD_ALLOC_RES,
+				   MLX4_CMD_TIME_CLASS_A, MLX4_CMD_WRAPPED);
+		if (!err)
+			*index = get_param_l(&out_param);
+
+		return err;
+	}
+	return __mlx4_register_vlan(dev, port, vlan, index);
+}
+EXPORT_SYMBOL_GPL(mlx4_register_vlan);
+
+void __mlx4_unregister_vlan(struct mlx4_dev *dev, u8 port, u16 vlan)
+{
+	struct mlx4_vlan_table *table = &mlx4_priv(dev)->port[port].vlan_table;
+	int index;
+	bool dup = mlx4_is_mf_bonded(dev);
+	u8 dup_port = (port == 1) ? 2 : 1;
+	struct mlx4_vlan_table *dup_table = &mlx4_priv(dev)->port[dup_port].vlan_table;
+
+	if (dup) {
+		if (port == 1) {
+			mutex_lock(&table->mutex);
+			mutex_lock_nested(&dup_table->mutex, SINGLE_DEPTH_NESTING);
+		} else {
+			mutex_lock(&dup_table->mutex);
+			mutex_lock_nested(&table->mutex, SINGLE_DEPTH_NESTING);
+		}
+	} else {
+		mutex_lock(&table->mutex);
+	}
+
+	if (mlx4_find_cached_vlan(dev, port, vlan, &index)) {
+		mlx4_warn(dev, "vlan 0x%x is not in the vlan table\n", vlan);
+		goto out;
+	}
+
+	if (index < MLX4_VLAN_REGULAR) {
+		mlx4_warn(dev, "Trying to free special vlan index %d\n", index);
+		goto out;
+	}
+
+	if (--table->refs[index] || table->is_dup[index]) {
+		mlx4_dbg(dev, "Have %d more references for index %d, no need to modify vlan table\n",
+			 table->refs[index], index);
+		if (!table->refs[index])
+			dup_table->is_dup[index] = false;
+		goto out;
+	}
+	table->entries[index] = 0;
+	if (mlx4_set_port_vlan_table(dev, port, table->entries))
+		mlx4_warn(dev, "Fail to set vlan in port %d during unregister\n", port);
+	--table->total;
+	if (dup) {
+		dup_table->is_dup[index] = false;
+		if (dup_table->refs[index])
+			goto out;
+		dup_table->entries[index] = 0;
+		if (mlx4_set_port_vlan_table(dev, dup_port, dup_table->entries))
+			mlx4_warn(dev, "Fail to set vlan in duplicate port %d during unregister\n", dup_port);
+		--dup_table->total;
+	}
+out:
+	if (dup) {
+		if (port == 2) {
+			mutex_unlock(&table->mutex);
+			mutex_unlock(&dup_table->mutex);
+		} else {
+			mutex_unlock(&dup_table->mutex);
+			mutex_unlock(&table->mutex);
+		}
+	} else {
+		mutex_unlock(&table->mutex);
+	}
+}
+
+void mlx4_unregister_vlan(struct mlx4_dev *dev, u8 port, u16 vlan)
+{
+	u64 out_param = 0;
+
+	if (mlx4_is_mfunc(dev)) {
+		(void) mlx4_cmd_imm(dev, vlan, &out_param,
+				    ((u32) port) << 8 | (u32) RES_VLAN,
+				    RES_OP_RESERVE_AND_MAP,
+				    MLX4_CMD_FREE_RES, MLX4_CMD_TIME_CLASS_A,
+				    MLX4_CMD_WRAPPED);
+		return;
+	}
+	__mlx4_unregister_vlan(dev, port, vlan);
+}
+EXPORT_SYMBOL_GPL(mlx4_unregister_vlan);
+
+int mlx4_bond_mac_table(struct mlx4_dev *dev)
+{
+	struct mlx4_mac_table *t1 = &mlx4_priv(dev)->port[1].mac_table;
+	struct mlx4_mac_table *t2 = &mlx4_priv(dev)->port[2].mac_table;
+	int ret = 0;
+	int i;
+	bool update1 = false;
+	bool update2 = false;
+
+	mutex_lock(&t1->mutex);
+	mutex_lock(&t2->mutex);
+	for (i = 0; i < MLX4_MAX_MAC_NUM; i++) {
+		if ((t1->entries[i] != t2->entries[i]) &&
+		    t1->entries[i] && t2->entries[i]) {
+			mlx4_warn(dev, "can't duplicate entry %d in mac table\n", i);
+			ret = -EINVAL;
+			goto unlock;
+		}
+	}
+
+	for (i = 0; i < MLX4_MAX_MAC_NUM; i++) {
+		if (t1->entries[i] && !t2->entries[i]) {
+			t2->entries[i] = t1->entries[i];
+			t2->is_dup[i] = true;
+			update2 = true;
+		} else if (!t1->entries[i] && t2->entries[i]) {
+			t1->entries[i] = t2->entries[i];
+			t1->is_dup[i] = true;
+			update1 = true;
+		} else if (t1->entries[i] && t2->entries[i]) {
+			t1->is_dup[i] = true;
+			t2->is_dup[i] = true;
+		}
+	}
+
+	if (update1) {
+		ret = mlx4_set_port_mac_table(dev, 1, t1->entries);
+		if (ret)
+			mlx4_warn(dev, "failed to set MAC table for port 1 (%d)\n", ret);
+	}
+	if (!ret && update2) {
+		ret = mlx4_set_port_mac_table(dev, 2, t2->entries);
+		if (ret)
+			mlx4_warn(dev, "failed to set MAC table for port 2 (%d)\n", ret);
+	}
+
+	if (ret)
+		mlx4_warn(dev, "failed to create mirror MAC tables\n");
+unlock:
+	mutex_unlock(&t2->mutex);
+	mutex_unlock(&t1->mutex);
+	return ret;
+}
+
+int mlx4_unbond_mac_table(struct mlx4_dev *dev)
+{
+	struct mlx4_mac_table *t1 = &mlx4_priv(dev)->port[1].mac_table;
+	struct mlx4_mac_table *t2 = &mlx4_priv(dev)->port[2].mac_table;
+	int ret = 0;
+	int ret1;
+	int i;
+	bool update1 = false;
+	bool update2 = false;
+
+	mutex_lock(&t1->mutex);
+	mutex_lock(&t2->mutex);
+	for (i = 0; i < MLX4_MAX_MAC_NUM; i++) {
+		if (t1->entries[i] != t2->entries[i]) {
+			mlx4_warn(dev, "mac table is in an unexpected state when trying to unbond\n");
+			ret = -EINVAL;
+			goto unlock;
+		}
+	}
+
+	for (i = 0; i < MLX4_MAX_MAC_NUM; i++) {
+		if (!t1->entries[i])
+			continue;
+		t1->is_dup[i] = false;
+		if (!t1->refs[i]) {
+			t1->entries[i] = 0;
+			update1 = true;
+		}
+		t2->is_dup[i] = false;
+		if (!t2->refs[i]) {
+			t2->entries[i] = 0;
+			update2 = true;
+		}
+	}
+
+	if (update1) {
+		ret = mlx4_set_port_mac_table(dev, 1, t1->entries);
+		if (ret)
+			mlx4_warn(dev, "failed to unmirror MAC tables for port 1(%d)\n", ret);
+	}
+	if (update2) {
+		ret1 = mlx4_set_port_mac_table(dev, 2, t2->entries);
+		if (ret1) {
+			mlx4_warn(dev, "failed to unmirror MAC tables for port 2(%d)\n", ret1);
+			ret = ret1;
+		}
+	}
+unlock:
+	mutex_unlock(&t2->mutex);
+	mutex_unlock(&t1->mutex);
+	return ret;
+}
+
+int mlx4_bond_vlan_table(struct mlx4_dev *dev)
+{
+	struct mlx4_vlan_table *t1 = &mlx4_priv(dev)->port[1].vlan_table;
+	struct mlx4_vlan_table *t2 = &mlx4_priv(dev)->port[2].vlan_table;
+	int ret = 0;
+	int i;
+	bool update1 = false;
+	bool update2 = false;
+
+	mutex_lock(&t1->mutex);
+	mutex_lock(&t2->mutex);
+	for (i = 0; i < MLX4_MAX_VLAN_NUM; i++) {
+		if ((t1->entries[i] != t2->entries[i]) &&
+		    t1->entries[i] && t2->entries[i]) {
+			mlx4_warn(dev, "can't duplicate entry %d in vlan table\n", i);
+			ret = -EINVAL;
+			goto unlock;
+		}
+	}
+
+	for (i = 0; i < MLX4_MAX_VLAN_NUM; i++) {
+		if (t1->entries[i] && !t2->entries[i]) {
+			t2->entries[i] = t1->entries[i];
+			t2->is_dup[i] = true;
+			update2 = true;
+		} else if (!t1->entries[i] && t2->entries[i]) {
+			t1->entries[i] = t2->entries[i];
+			t1->is_dup[i] = true;
+			update1 = true;
+		} else if (t1->entries[i] && t2->entries[i]) {
+			t1->is_dup[i] = true;
+			t2->is_dup[i] = true;
+		}
+	}
+
+	if (update1) {
+		ret = mlx4_set_port_vlan_table(dev, 1, t1->entries);
+		if (ret)
+			mlx4_warn(dev, "failed to set VLAN table for port 1 (%d)\n", ret);
+	}
+	if (!ret && update2) {
+		ret = mlx4_set_port_vlan_table(dev, 2, t2->entries);
+		if (ret)
+			mlx4_warn(dev, "failed to set VLAN table for port 2 (%d)\n", ret);
+	}
+
+	if (ret)
+		mlx4_warn(dev, "failed to create mirror VLAN tables\n");
+unlock:
+	mutex_unlock(&t2->mutex);
+	mutex_unlock(&t1->mutex);
+	return ret;
+}
+
+int mlx4_unbond_vlan_table(struct mlx4_dev *dev)
+{
+	struct mlx4_vlan_table *t1 = &mlx4_priv(dev)->port[1].vlan_table;
+	struct mlx4_vlan_table *t2 = &mlx4_priv(dev)->port[2].vlan_table;
+	int ret = 0;
+	int ret1;
+	int i;
+	bool update1 = false;
+	bool update2 = false;
+
+	mutex_lock(&t1->mutex);
+	mutex_lock(&t2->mutex);
+	for (i = 0; i < MLX4_MAX_VLAN_NUM; i++) {
+		if (t1->entries[i] != t2->entries[i]) {
+			mlx4_warn(dev, "vlan table is in an unexpected state when trying to unbond\n");
+			ret = -EINVAL;
+			goto unlock;
+		}
+	}
+
+	for (i = 0; i < MLX4_MAX_VLAN_NUM; i++) {
+		if (!t1->entries[i])
+			continue;
+		t1->is_dup[i] = false;
+		if (!t1->refs[i]) {
+			t1->entries[i] = 0;
+			update1 = true;
+		}
+		t2->is_dup[i] = false;
+		if (!t2->refs[i]) {
+			t2->entries[i] = 0;
+			update2 = true;
+		}
+	}
+
+	if (update1) {
+		ret = mlx4_set_port_vlan_table(dev, 1, t1->entries);
+		if (ret)
+			mlx4_warn(dev, "failed to unmirror VLAN tables for port 1(%d)\n", ret);
+	}
+	if (update2) {
+		ret1 = mlx4_set_port_vlan_table(dev, 2, t2->entries);
+		if (ret1) {
+			mlx4_warn(dev, "failed to unmirror VLAN tables for port 2(%d)\n", ret1);
+			ret = ret1;
+		}
+	}
+unlock:
+	mutex_unlock(&t2->mutex);
+	mutex_unlock(&t1->mutex);
+	return ret;
+}
+
+int mlx4_get_port_ib_caps(struct mlx4_dev *dev, u8 port, __be32 *caps)
+{
+	struct mlx4_cmd_mailbox *inmailbox, *outmailbox;
+	u8 *inbuf, *outbuf;
+	int err;
+
+	inmailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(inmailbox))
+		return PTR_ERR(inmailbox);
+
+	outmailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(outmailbox)) {
+		mlx4_free_cmd_mailbox(dev, inmailbox);
+		return PTR_ERR(outmailbox);
+	}
+
+	inbuf = inmailbox->buf;
+	outbuf = outmailbox->buf;
+	inbuf[0] = 1;
+	inbuf[1] = 1;
+	inbuf[2] = 1;
+	inbuf[3] = 1;
+	*(__be16 *) (&inbuf[16]) = cpu_to_be16(0x0015);
+	*(__be32 *) (&inbuf[20]) = cpu_to_be32(port);
+
+	err = mlx4_cmd_box(dev, inmailbox->dma, outmailbox->dma, port, 3,
+			   MLX4_CMD_MAD_IFC, MLX4_CMD_TIME_CLASS_C,
+			   MLX4_CMD_NATIVE);
+	if (!err)
+		*caps = *(__be32 *) (outbuf + 84);
+	mlx4_free_cmd_mailbox(dev, inmailbox);
+	mlx4_free_cmd_mailbox(dev, outmailbox);
+	return err;
+}
+static struct mlx4_roce_gid_entry zgid_entry;
+
+int mlx4_get_slave_num_gids(struct mlx4_dev *dev, int slave, int port)
+{
+	int vfs;
+	int slave_gid = slave;
+	unsigned i;
+	struct mlx4_slaves_pport slaves_pport;
+	struct mlx4_active_ports actv_ports;
+	unsigned max_port_p_one;
+
+	if (slave == 0)
+		return MLX4_ROCE_PF_GIDS;
+
+	/* Slave is a VF */
+	slaves_pport = mlx4_phys_to_slaves_pport(dev, port);
+	actv_ports = mlx4_get_active_ports(dev, slave);
+	max_port_p_one = find_first_bit(actv_ports.ports, dev->caps.num_ports) +
+		bitmap_weight(actv_ports.ports, dev->caps.num_ports) + 1;
+
+	for (i = 1; i < max_port_p_one; i++) {
+		struct mlx4_active_ports exclusive_ports;
+		struct mlx4_slaves_pport slaves_pport_actv;
+		bitmap_zero(exclusive_ports.ports, dev->caps.num_ports);
+		set_bit(i - 1, exclusive_ports.ports);
+		if (i == port)
+			continue;
+		slaves_pport_actv = mlx4_phys_to_slaves_pport_actv(
+				    dev, &exclusive_ports);
+		slave_gid -= bitmap_weight(slaves_pport_actv.slaves,
+					   dev->persist->num_vfs + 1);
+	}
+	vfs = bitmap_weight(slaves_pport.slaves, dev->persist->num_vfs + 1) - 1;
+	if (slave_gid <= ((MLX4_ROCE_MAX_GIDS - MLX4_ROCE_PF_GIDS) % vfs))
+		return ((MLX4_ROCE_MAX_GIDS - MLX4_ROCE_PF_GIDS) / vfs) + 1;
+	return (MLX4_ROCE_MAX_GIDS - MLX4_ROCE_PF_GIDS) / vfs;
+}
+
+int mlx4_get_base_gid_ix(struct mlx4_dev *dev, int slave, int port)
+{
+	int gids;
+	unsigned i;
+	int slave_gid = slave;
+	int vfs;
+
+	struct mlx4_slaves_pport slaves_pport;
+	struct mlx4_active_ports actv_ports;
+	unsigned max_port_p_one;
+
+	if (slave == 0)
+		return 0;
+
+	slaves_pport = mlx4_phys_to_slaves_pport(dev, port);
+	actv_ports = mlx4_get_active_ports(dev, slave);
+	max_port_p_one = find_first_bit(actv_ports.ports, dev->caps.num_ports) +
+		bitmap_weight(actv_ports.ports, dev->caps.num_ports) + 1;
+
+	for (i = 1; i < max_port_p_one; i++) {
+		struct mlx4_active_ports exclusive_ports;
+		struct mlx4_slaves_pport slaves_pport_actv;
+		bitmap_zero(exclusive_ports.ports, dev->caps.num_ports);
+		set_bit(i - 1, exclusive_ports.ports);
+		if (i == port)
+			continue;
+		slaves_pport_actv = mlx4_phys_to_slaves_pport_actv(
+				    dev, &exclusive_ports);
+		slave_gid -= bitmap_weight(slaves_pport_actv.slaves,
+					   dev->persist->num_vfs + 1);
+	}
+	gids = MLX4_ROCE_MAX_GIDS - MLX4_ROCE_PF_GIDS;
+	vfs = bitmap_weight(slaves_pport.slaves, dev->persist->num_vfs + 1) - 1;
+	if (slave_gid <= gids % vfs)
+		return MLX4_ROCE_PF_GIDS + ((gids / vfs) + 1) * (slave_gid - 1);
+
+	return MLX4_ROCE_PF_GIDS + (gids % vfs) +
+		((gids / vfs) * (slave_gid - 1));
+}
+EXPORT_SYMBOL_GPL(mlx4_get_base_gid_ix);
+
+static int mlx4_reset_roce_port_gids(struct mlx4_dev *dev, int slave,
+				     int port, struct mlx4_cmd_mailbox *mailbox)
+{
+	struct mlx4_roce_gid_entry *gid_entry_mbox;
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int num_gids, base, offset;
+	int i, err;
+
+	num_gids = mlx4_get_slave_num_gids(dev, slave, port);
+	base = mlx4_get_base_gid_ix(dev, slave, port);
+
+	memset(mailbox->buf, 0, MLX4_MAILBOX_SIZE);
+
+	mutex_lock(&(priv->port[port].gid_table.mutex));
+	/* Zero-out gids belonging to that slave in the port GID table */
+	for (i = 0, offset = base; i < num_gids; offset++, i++)
+		memcpy(priv->port[port].gid_table.roce_gids[offset].raw,
+		       zgid_entry.raw, MLX4_ROCE_GID_ENTRY_SIZE);
+
+	/* Now, copy roce port gids table to mailbox for passing to FW */
+	gid_entry_mbox = (struct mlx4_roce_gid_entry *)mailbox->buf;
+	for (i = 0; i < MLX4_ROCE_MAX_GIDS; gid_entry_mbox++, i++)
+		memcpy(gid_entry_mbox->raw,
+		       priv->port[port].gid_table.roce_gids[i].raw,
+		       MLX4_ROCE_GID_ENTRY_SIZE);
+
+	err = mlx4_cmd(dev, mailbox->dma,
+		       ((u32)port) | (MLX4_SET_PORT_GID_TABLE << 8),
+		       MLX4_SET_PORT_ETH_OPCODE, MLX4_CMD_SET_PORT,
+		       MLX4_CMD_TIME_CLASS_B, MLX4_CMD_NATIVE);
+	mutex_unlock(&(priv->port[port].gid_table.mutex));
+	return err;
+}
+
+
+void mlx4_reset_roce_gids(struct mlx4_dev *dev, int slave)
+{
+	struct mlx4_active_ports actv_ports;
+	struct mlx4_cmd_mailbox *mailbox;
+	int num_eth_ports, err;
+	int i;
+
+	if (slave < 0 || slave > dev->persist->num_vfs)
+		return;
+
+	actv_ports = mlx4_get_active_ports(dev, slave);
+
+	for (i = 0, num_eth_ports = 0; i < dev->caps.num_ports; i++) {
+		if (test_bit(i, actv_ports.ports)) {
+			if (dev->caps.port_type[i + 1] != MLX4_PORT_TYPE_ETH)
+				continue;
+			num_eth_ports++;
+		}
+	}
+
+	if (!num_eth_ports)
+		return;
+
+	/* have ETH ports.  Alloc mailbox for SET_PORT command */
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return;
+
+	for (i = 0; i < dev->caps.num_ports; i++) {
+		if (test_bit(i, actv_ports.ports)) {
+			if (dev->caps.port_type[i + 1] != MLX4_PORT_TYPE_ETH)
+				continue;
+			err = mlx4_reset_roce_port_gids(dev, slave, i + 1, mailbox);
+			if (err)
+				mlx4_warn(dev, "Could not reset ETH port GID table for slave %d, port %d (%d)\n",
+					  slave, i + 1, err);
+		}
+	}
+
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return;
+}
+
+static void
+mlx4_en_set_port_mtu(struct mlx4_dev *dev, int slave, int port,
+		     struct mlx4_set_port_general_context *gen_context)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_mfunc_master_ctx *master = &priv->mfunc.master;
+	struct mlx4_slave_state *slave_st = &master->slave_state[slave];
+	u16 mtu, prev_mtu;
+
+	/* Mtu is configured as the max USER_MTU among all
+	 * the functions on the port.
+	 */
+	mtu = be16_to_cpu(gen_context->mtu);
+	mtu = min_t(int, mtu, dev->caps.eth_mtu_cap[port] +
+		    ETH_HLEN + VLAN_HLEN + ETH_FCS_LEN);
+	prev_mtu = slave_st->mtu[port];
+	slave_st->mtu[port] = mtu;
+	if (mtu > master->max_mtu[port])
+		master->max_mtu[port] = mtu;
+	if (mtu < prev_mtu && prev_mtu == master->max_mtu[port]) {
+		int i;
+
+		slave_st->mtu[port] = mtu;
+		master->max_mtu[port] = mtu;
+		for (i = 0; i < dev->num_slaves; i++)
+			master->max_mtu[port] =
+				max_t(u16, master->max_mtu[port],
+				      master->slave_state[i].mtu[port]);
+	}
+	gen_context->mtu = cpu_to_be16(master->max_mtu[port]);
+}
+
+static void
+mlx4_en_set_port_user_mtu(struct mlx4_dev *dev, int slave, int port,
+			  struct mlx4_set_port_general_context *gen_context)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_mfunc_master_ctx *master = &priv->mfunc.master;
+	struct mlx4_slave_state *slave_st = &master->slave_state[slave];
+	u16 user_mtu, prev_user_mtu;
+
+	/* User Mtu is configured as the max USER_MTU among all
+	 * the functions on the port.
+	 */
+	user_mtu = be16_to_cpu(gen_context->user_mtu);
+	user_mtu = min_t(int, user_mtu, dev->caps.eth_mtu_cap[port]);
+	prev_user_mtu = slave_st->user_mtu[port];
+	slave_st->user_mtu[port] = user_mtu;
+	if (user_mtu > master->max_user_mtu[port])
+		master->max_user_mtu[port] = user_mtu;
+	if (user_mtu < prev_user_mtu &&
+	    prev_user_mtu == master->max_user_mtu[port]) {
+		int i;
+
+		slave_st->user_mtu[port] = user_mtu;
+		master->max_user_mtu[port] = user_mtu;
+		for (i = 0; i < dev->num_slaves; i++)
+			master->max_user_mtu[port] =
+				max_t(u16, master->max_user_mtu[port],
+				      master->slave_state[i].user_mtu[port]);
+	}
+	gen_context->user_mtu = cpu_to_be16(master->max_user_mtu[port]);
+}
+
+static void
+mlx4_en_set_port_global_pause(struct mlx4_dev *dev, int slave,
+			      struct mlx4_set_port_general_context *gen_context)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_mfunc_master_ctx *master = &priv->mfunc.master;
+
+	/* Slave cannot change Global Pause configuration */
+	if (slave != mlx4_master_func_num(dev) &&
+	    (gen_context->pptx != master->pptx ||
+	     gen_context->pprx != master->pprx)) {
+		gen_context->pptx = master->pptx;
+		gen_context->pprx = master->pprx;
+		mlx4_warn(dev, "denying Global Pause change for slave:%d\n",
+			  slave);
+	} else {
+		master->pptx = gen_context->pptx;
+		master->pprx = gen_context->pprx;
+	}
+}
+
+static int mlx4_common_set_port(struct mlx4_dev *dev, int slave, u32 in_mod,
+				u8 op_mod, struct mlx4_cmd_mailbox *inbox)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_port_info *port_info;
+	struct mlx4_set_port_rqp_calc_context *qpn_context;
+	struct mlx4_set_port_general_context *gen_context;
+	struct mlx4_roce_gid_entry *gid_entry_tbl, *gid_entry_mbox, *gid_entry_mb1;
+	int reset_qkey_viols;
+	int port;
+	int is_eth;
+	int num_gids;
+	int base;
+	u32 in_modifier;
+	u32 promisc;
+	int err;
+	int i, j;
+	int offset;
+	__be32 agg_cap_mask;
+	__be32 slave_cap_mask;
+	__be32 new_cap_mask;
+
+	port = in_mod & 0xff;
+	in_modifier = in_mod >> 8;
+	is_eth = op_mod;
+	port_info = &priv->port[port];
+
+	/* Slaves cannot perform SET_PORT operations,
+	 * except for changing MTU and USER_MTU.
+	 */
+	if (is_eth) {
+		if (slave != dev->caps.function &&
+		    in_modifier != MLX4_SET_PORT_GENERAL &&
+		    in_modifier != MLX4_SET_PORT_GID_TABLE) {
+			mlx4_warn(dev, "denying SET_PORT for slave:%d\n",
+					slave);
+			return -EINVAL;
+		}
+		switch (in_modifier) {
+		case MLX4_SET_PORT_RQP_CALC:
+			qpn_context = inbox->buf;
+			qpn_context->base_qpn =
+				cpu_to_be32(port_info->base_qpn);
+			qpn_context->n_mac = 0x7;
+			promisc = be32_to_cpu(qpn_context->promisc) >>
+				SET_PORT_PROMISC_SHIFT;
+			qpn_context->promisc = cpu_to_be32(
+				promisc << SET_PORT_PROMISC_SHIFT |
+				port_info->base_qpn);
+			promisc = be32_to_cpu(qpn_context->mcast) >>
+				SET_PORT_MC_PROMISC_SHIFT;
+			qpn_context->mcast = cpu_to_be32(
+				promisc << SET_PORT_MC_PROMISC_SHIFT |
+				port_info->base_qpn);
+			break;
+		case MLX4_SET_PORT_GENERAL:
+			gen_context = inbox->buf;
+
+			if (gen_context->flags & MLX4_FLAG_V_MTU_MASK)
+				mlx4_en_set_port_mtu(dev, slave, port,
+						     gen_context);
+
+			if (gen_context->flags2 & MLX4_FLAG2_V_USER_MTU_MASK)
+				mlx4_en_set_port_user_mtu(dev, slave, port,
+							  gen_context);
+
+			if (gen_context->flags &
+			    (MLX4_FLAG_V_PPRX_MASK | MLX4_FLAG_V_PPTX_MASK))
+				mlx4_en_set_port_global_pause(dev, slave,
+							      gen_context);
+
+			break;
+		case MLX4_SET_PORT_GID_TABLE:
+			/* change to MULTIPLE entries: number of guest's gids
+			 * need a FOR-loop here over number of gids the guest has.
+			 * 1. Check no duplicates in gids passed by slave
+			 */
+			num_gids = mlx4_get_slave_num_gids(dev, slave, port);
+			base = mlx4_get_base_gid_ix(dev, slave, port);
+			gid_entry_mbox = (struct mlx4_roce_gid_entry *)(inbox->buf);
+			for (i = 0; i < num_gids; gid_entry_mbox++, i++) {
+				if (!memcmp(gid_entry_mbox->raw, zgid_entry.raw,
+					    sizeof(zgid_entry)))
+					continue;
+				gid_entry_mb1 = gid_entry_mbox + 1;
+				for (j = i + 1; j < num_gids; gid_entry_mb1++, j++) {
+					if (!memcmp(gid_entry_mb1->raw,
+						    zgid_entry.raw, sizeof(zgid_entry)))
+						continue;
+					if (!memcmp(gid_entry_mb1->raw, gid_entry_mbox->raw,
+						    sizeof(gid_entry_mbox->raw))) {
+						/* found duplicate */
+						return -EINVAL;
+					}
+				}
+			}
+
+			/* 2. Check that do not have duplicates in OTHER
+			 *    entries in the port GID table
+			 */
+
+			mutex_lock(&(priv->port[port].gid_table.mutex));
+			for (i = 0; i < MLX4_ROCE_MAX_GIDS; i++) {
+				if (i >= base && i < base + num_gids)
+					continue; /* don't compare to slave's current gids */
+				gid_entry_tbl = &priv->port[port].gid_table.roce_gids[i];
+				if (!memcmp(gid_entry_tbl->raw, zgid_entry.raw, sizeof(zgid_entry)))
+					continue;
+				gid_entry_mbox = (struct mlx4_roce_gid_entry *)(inbox->buf);
+				for (j = 0; j < num_gids; gid_entry_mbox++, j++) {
+					if (!memcmp(gid_entry_mbox->raw, zgid_entry.raw,
+						    sizeof(zgid_entry)))
+						continue;
+					if (!memcmp(gid_entry_mbox->raw, gid_entry_tbl->raw,
+						    sizeof(gid_entry_tbl->raw))) {
+						/* found duplicate */
+						mlx4_warn(dev, "requested gid entry for slave:%d is a duplicate of gid at index %d\n",
+							  slave, i);
+						mutex_unlock(&(priv->port[port].gid_table.mutex));
+						return -EINVAL;
+					}
+				}
+			}
+
+			/* insert slave GIDs with memcpy, starting at slave's base index */
+			gid_entry_mbox = (struct mlx4_roce_gid_entry *)(inbox->buf);
+			for (i = 0, offset = base; i < num_gids; gid_entry_mbox++, offset++, i++)
+				memcpy(priv->port[port].gid_table.roce_gids[offset].raw,
+				       gid_entry_mbox->raw, MLX4_ROCE_GID_ENTRY_SIZE);
+
+			/* Now, copy roce port gids table to current mailbox for passing to FW */
+			gid_entry_mbox = (struct mlx4_roce_gid_entry *)(inbox->buf);
+			for (i = 0; i < MLX4_ROCE_MAX_GIDS; gid_entry_mbox++, i++)
+				memcpy(gid_entry_mbox->raw,
+				       priv->port[port].gid_table.roce_gids[i].raw,
+				       MLX4_ROCE_GID_ENTRY_SIZE);
+
+			err = mlx4_cmd(dev, inbox->dma, in_mod & 0xffff, op_mod,
+				       MLX4_CMD_SET_PORT, MLX4_CMD_TIME_CLASS_B,
+				       MLX4_CMD_NATIVE);
+			mutex_unlock(&(priv->port[port].gid_table.mutex));
+			return err;
+		}
+
+		return mlx4_cmd(dev, inbox->dma, in_mod & 0xffff, op_mod,
+				MLX4_CMD_SET_PORT, MLX4_CMD_TIME_CLASS_B,
+				MLX4_CMD_NATIVE);
+	}
+
+	/* Slaves are not allowed to SET_PORT beacon (LED) blink */
+	if (op_mod == MLX4_SET_PORT_BEACON_OPCODE) {
+		mlx4_warn(dev, "denying SET_PORT Beacon slave:%d\n", slave);
+		return -EPERM;
+	}
+
+	/* For IB, we only consider:
+	 * - The capability mask, which is set to the aggregate of all
+	 *   slave function capabilities
+	 * - The QKey violatin counter - reset according to each request.
+	 */
+
+	if (dev->flags & MLX4_FLAG_OLD_PORT_CMDS) {
+		reset_qkey_viols = (*(u8 *) inbox->buf) & 0x40;
+		new_cap_mask = ((__be32 *) inbox->buf)[2];
+	} else {
+		reset_qkey_viols = ((u8 *) inbox->buf)[3] & 0x1;
+		new_cap_mask = ((__be32 *) inbox->buf)[1];
+	}
+
+	/* slave may not set the IS_SM capability for the port */
+	if (slave != mlx4_master_func_num(dev) &&
+	    (be32_to_cpu(new_cap_mask) & MLX4_PORT_CAP_IS_SM))
+		return -EINVAL;
+
+	/* No DEV_MGMT in multifunc mode */
+	if (mlx4_is_mfunc(dev) &&
+	    (be32_to_cpu(new_cap_mask) & MLX4_PORT_CAP_DEV_MGMT_SUP))
+		return -EINVAL;
+
+	agg_cap_mask = 0;
+	slave_cap_mask =
+		priv->mfunc.master.slave_state[slave].ib_cap_mask[port];
+	priv->mfunc.master.slave_state[slave].ib_cap_mask[port] = new_cap_mask;
+	for (i = 0; i < dev->num_slaves; i++)
+		agg_cap_mask |=
+			priv->mfunc.master.slave_state[i].ib_cap_mask[port];
+
+	/* only clear mailbox for guests.  Master may be setting
+	* MTU or PKEY table size
+	*/
+	if (slave != dev->caps.function)
+		memset(inbox->buf, 0, 256);
+	if (dev->flags & MLX4_FLAG_OLD_PORT_CMDS) {
+		*(u8 *) inbox->buf	   |= !!reset_qkey_viols << 6;
+		((__be32 *) inbox->buf)[2] = agg_cap_mask;
+	} else {
+		((u8 *) inbox->buf)[3]     |= !!reset_qkey_viols;
+		((__be32 *) inbox->buf)[1] = agg_cap_mask;
+	}
+
+	err = mlx4_cmd(dev, inbox->dma, port, is_eth, MLX4_CMD_SET_PORT,
+		       MLX4_CMD_TIME_CLASS_B, MLX4_CMD_NATIVE);
+	if (err)
+		priv->mfunc.master.slave_state[slave].ib_cap_mask[port] =
+			slave_cap_mask;
+	return err;
+}
+
+int mlx4_SET_PORT_wrapper(struct mlx4_dev *dev, int slave,
+			  struct mlx4_vhcr *vhcr,
+			  struct mlx4_cmd_mailbox *inbox,
+			  struct mlx4_cmd_mailbox *outbox,
+			  struct mlx4_cmd_info *cmd)
+{
+	int port = mlx4_slave_convert_port(
+			dev, slave, vhcr->in_modifier & 0xFF);
+
+	if (port < 0)
+		return -EINVAL;
+
+	vhcr->in_modifier = (vhcr->in_modifier & ~0xFF) |
+			    (port & 0xFF);
+
+	return mlx4_common_set_port(dev, slave, vhcr->in_modifier,
+				    vhcr->op_modifier, inbox);
+}
+
+/* bit locations for set port command with zero op modifier */
+enum {
+	MLX4_SET_PORT_VL_CAP	 = 4, /* bits 7:4 */
+	MLX4_SET_PORT_MTU_CAP	 = 12, /* bits 15:12 */
+	MLX4_CHANGE_PORT_PKEY_TBL_SZ = 20,
+	MLX4_CHANGE_PORT_VL_CAP	 = 21,
+	MLX4_CHANGE_PORT_MTU_CAP = 22,
+};
+
+int mlx4_SET_PORT(struct mlx4_dev *dev, u8 port, int pkey_tbl_sz)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	int err, vl_cap, pkey_tbl_flag = 0;
+
+	if (dev->caps.port_type[port] == MLX4_PORT_TYPE_ETH)
+		return 0;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+
+	((__be32 *) mailbox->buf)[1] = dev->caps.ib_port_def_cap[port];
+
+	if (pkey_tbl_sz >= 0 && mlx4_is_master(dev)) {
+		pkey_tbl_flag = 1;
+		((__be16 *) mailbox->buf)[20] = cpu_to_be16(pkey_tbl_sz);
+	}
+
+	/* IB VL CAP enum isn't used by the firmware, just numerical values */
+	for (vl_cap = 8; vl_cap >= 1; vl_cap >>= 1) {
+		((__be32 *) mailbox->buf)[0] = cpu_to_be32(
+			(1 << MLX4_CHANGE_PORT_MTU_CAP) |
+			(1 << MLX4_CHANGE_PORT_VL_CAP)  |
+			(pkey_tbl_flag << MLX4_CHANGE_PORT_PKEY_TBL_SZ) |
+			(dev->caps.port_ib_mtu[port] << MLX4_SET_PORT_MTU_CAP) |
+			(vl_cap << MLX4_SET_PORT_VL_CAP));
+		err = mlx4_cmd(dev, mailbox->dma, port,
+			       MLX4_SET_PORT_IB_OPCODE, MLX4_CMD_SET_PORT,
+			       MLX4_CMD_TIME_CLASS_B, MLX4_CMD_WRAPPED);
+		if (err != -ENOMEM)
+			break;
+	}
+
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return err;
+}
+
+#define SET_PORT_ROCE_2_FLAGS          0x10
+#define MLX4_SET_PORT_ROCE_V1_V2       0x2
+int mlx4_SET_PORT_general(struct mlx4_dev *dev, u8 port, int mtu,
+			  u8 pptx, u8 pfctx, u8 pprx, u8 pfcrx)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	struct mlx4_set_port_general_context *context;
+	int err;
+	u32 in_mod;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+	context = mailbox->buf;
+	context->flags = SET_PORT_GEN_ALL_VALID;
+	context->mtu = cpu_to_be16(mtu);
+	context->pptx = (pptx * (!pfctx)) << 7;
+	context->pfctx = pfctx;
+	context->pprx = (pprx * (!pfcrx)) << 7;
+	context->pfcrx = pfcrx;
+
+	if (dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_ROCE_V1_V2) {
+		context->flags |= SET_PORT_ROCE_2_FLAGS;
+		context->roce_mode |=
+			MLX4_SET_PORT_ROCE_V1_V2 << 4;
+	}
+	in_mod = MLX4_SET_PORT_GENERAL << 8 | port;
+	err = mlx4_cmd(dev, mailbox->dma, in_mod, MLX4_SET_PORT_ETH_OPCODE,
+		       MLX4_CMD_SET_PORT, MLX4_CMD_TIME_CLASS_B,
+		       MLX4_CMD_WRAPPED);
+
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return err;
+}
+EXPORT_SYMBOL(mlx4_SET_PORT_general);
+
+int mlx4_SET_PORT_qpn_calc(struct mlx4_dev *dev, u8 port, u32 base_qpn,
+			   u8 promisc)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	struct mlx4_set_port_rqp_calc_context *context;
+	int err;
+	u32 in_mod;
+	u32 m_promisc = (dev->caps.flags & MLX4_DEV_CAP_FLAG_VEP_MC_STEER) ?
+		MCAST_DIRECT : MCAST_DEFAULT;
+
+	if (dev->caps.steering_mode != MLX4_STEERING_MODE_A0)
+		return 0;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+	context = mailbox->buf;
+	context->base_qpn = cpu_to_be32(base_qpn);
+	context->n_mac = dev->caps.log_num_macs;
+	context->promisc = cpu_to_be32(promisc << SET_PORT_PROMISC_SHIFT |
+				       base_qpn);
+	context->mcast = cpu_to_be32(m_promisc << SET_PORT_MC_PROMISC_SHIFT |
+				     base_qpn);
+	context->intra_no_vlan = 0;
+	context->no_vlan = MLX4_NO_VLAN_IDX;
+	context->intra_vlan_miss = 0;
+	context->vlan_miss = MLX4_VLAN_MISS_IDX;
+
+	in_mod = MLX4_SET_PORT_RQP_CALC << 8 | port;
+	err = mlx4_cmd(dev, mailbox->dma, in_mod, MLX4_SET_PORT_ETH_OPCODE,
+		       MLX4_CMD_SET_PORT, MLX4_CMD_TIME_CLASS_B,
+		       MLX4_CMD_WRAPPED);
+
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return err;
+}
+EXPORT_SYMBOL(mlx4_SET_PORT_qpn_calc);
+
+int mlx4_SET_PORT_user_mtu(struct mlx4_dev *dev, u8 port, u16 user_mtu)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	struct mlx4_set_port_general_context *context;
+	u32 in_mod;
+	int err;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+	context = mailbox->buf;
+	context->flags2 |= MLX4_FLAG2_V_USER_MTU_MASK;
+	context->user_mtu = cpu_to_be16(user_mtu);
+
+	in_mod = MLX4_SET_PORT_GENERAL << 8 | port;
+	err = mlx4_cmd(dev, mailbox->dma, in_mod, MLX4_SET_PORT_ETH_OPCODE,
+		       MLX4_CMD_SET_PORT, MLX4_CMD_TIME_CLASS_B,
+		       MLX4_CMD_WRAPPED);
+
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return err;
+}
+EXPORT_SYMBOL(mlx4_SET_PORT_user_mtu);
+
+int mlx4_SET_PORT_user_mac(struct mlx4_dev *dev, u8 port, u8 *user_mac)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	struct mlx4_set_port_general_context *context;
+	u32 in_mod;
+	int err;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+	context = mailbox->buf;
+	context->flags2 |= MLX4_FLAG2_V_USER_MAC_MASK;
+	memcpy(context->user_mac, user_mac, sizeof(context->user_mac));
+
+	in_mod = MLX4_SET_PORT_GENERAL << 8 | port;
+	err = mlx4_cmd(dev, mailbox->dma, in_mod, MLX4_SET_PORT_ETH_OPCODE,
+		       MLX4_CMD_SET_PORT, MLX4_CMD_TIME_CLASS_B,
+		       MLX4_CMD_NATIVE);
+
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return err;
+}
+EXPORT_SYMBOL(mlx4_SET_PORT_user_mac);
+
+int mlx4_SET_PORT_fcs_check(struct mlx4_dev *dev, u8 port, u8 ignore_fcs_value)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	struct mlx4_set_port_general_context *context;
+	u32 in_mod;
+	int err;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+	context = mailbox->buf;
+	context->flags2 |= MLX4_FLAG2_V_IGNORE_FCS_MASK;
+	if (ignore_fcs_value)
+		context->ignore_fcs |= MLX4_IGNORE_FCS_MASK;
+	else
+		context->ignore_fcs &= ~MLX4_IGNORE_FCS_MASK;
+
+	in_mod = MLX4_SET_PORT_GENERAL << 8 | port;
+	err = mlx4_cmd(dev, mailbox->dma, in_mod, 1, MLX4_CMD_SET_PORT,
+		       MLX4_CMD_TIME_CLASS_B, MLX4_CMD_NATIVE);
+
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return err;
+}
+EXPORT_SYMBOL(mlx4_SET_PORT_fcs_check);
+
+enum {
+	VXLAN_ENABLE_MODIFY	= 1 << 7,
+	VXLAN_STEERING_MODIFY	= 1 << 6,
+
+	VXLAN_ENABLE		= 1 << 7,
+};
+
+struct mlx4_set_port_vxlan_context {
+	u32	reserved1;
+	u8	modify_flags;
+	u8	reserved2;
+	u8	enable_flags;
+	u8	steering;
+};
+
+int mlx4_SET_PORT_VXLAN(struct mlx4_dev *dev, u8 port, u8 steering, int enable)
+{
+	int err;
+	u32 in_mod;
+	struct mlx4_cmd_mailbox *mailbox;
+	struct mlx4_set_port_vxlan_context  *context;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+	context = mailbox->buf;
+	memset(context, 0, sizeof(*context));
+
+	context->modify_flags = VXLAN_ENABLE_MODIFY | VXLAN_STEERING_MODIFY;
+	if (enable)
+		context->enable_flags = VXLAN_ENABLE;
+	context->steering  = steering;
+
+	in_mod = MLX4_SET_PORT_VXLAN << 8 | port;
+	err = mlx4_cmd(dev, mailbox->dma, in_mod, MLX4_SET_PORT_ETH_OPCODE,
+		       MLX4_CMD_SET_PORT, MLX4_CMD_TIME_CLASS_B,
+		       MLX4_CMD_NATIVE);
+
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return err;
+}
+EXPORT_SYMBOL(mlx4_SET_PORT_VXLAN);
+
+int mlx4_SET_PORT_BEACON(struct mlx4_dev *dev, u8 port, u16 time)
+{
+	int err;
+	struct mlx4_cmd_mailbox *mailbox;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+
+	*((__be32 *)mailbox->buf) = cpu_to_be32(time);
+
+	err = mlx4_cmd(dev, mailbox->dma, port, MLX4_SET_PORT_BEACON_OPCODE,
+		       MLX4_CMD_SET_PORT, MLX4_CMD_TIME_CLASS_B,
+		       MLX4_CMD_NATIVE);
+
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return err;
+}
+EXPORT_SYMBOL(mlx4_SET_PORT_BEACON);
+
+int mlx4_SET_MCAST_FLTR_wrapper(struct mlx4_dev *dev, int slave,
+				struct mlx4_vhcr *vhcr,
+				struct mlx4_cmd_mailbox *inbox,
+				struct mlx4_cmd_mailbox *outbox,
+				struct mlx4_cmd_info *cmd)
+{
+	int err = 0;
+
+	return err;
+}
+
+int mlx4_SET_MCAST_FLTR(struct mlx4_dev *dev, u8 port,
+			u64 mac, u64 clear, u8 mode)
+{
+	return mlx4_cmd(dev, (mac | (clear << 63)), port, mode,
+			MLX4_CMD_SET_MCAST_FLTR, MLX4_CMD_TIME_CLASS_B,
+			MLX4_CMD_WRAPPED);
+}
+EXPORT_SYMBOL(mlx4_SET_MCAST_FLTR);
+
+int mlx4_SET_VLAN_FLTR_wrapper(struct mlx4_dev *dev, int slave,
+			       struct mlx4_vhcr *vhcr,
+			       struct mlx4_cmd_mailbox *inbox,
+			       struct mlx4_cmd_mailbox *outbox,
+			       struct mlx4_cmd_info *cmd)
+{
+	int err = 0;
+
+	return err;
+}
+
+int mlx4_DUMP_ETH_STATS_wrapper(struct mlx4_dev *dev, int slave,
+				struct mlx4_vhcr *vhcr,
+				struct mlx4_cmd_mailbox *inbox,
+				struct mlx4_cmd_mailbox *outbox,
+				struct mlx4_cmd_info *cmd)
+{
+	return 0;
+}
+
+int mlx4_get_slave_from_roce_gid(struct mlx4_dev *dev, int port, u8 *gid,
+				 int *slave_id)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int i, found_ix = -1;
+	int vf_gids = MLX4_ROCE_MAX_GIDS - MLX4_ROCE_PF_GIDS;
+	struct mlx4_slaves_pport slaves_pport;
+	unsigned num_vfs;
+	int slave_gid;
+
+	if (!mlx4_is_mfunc(dev))
+		return -EINVAL;
+
+	slaves_pport = mlx4_phys_to_slaves_pport(dev, port);
+	num_vfs = bitmap_weight(slaves_pport.slaves,
+				dev->persist->num_vfs + 1) - 1;
+
+	for (i = 0; i < MLX4_ROCE_MAX_GIDS; i++) {
+		if (!memcmp(priv->port[port].gid_table.roce_gids[i].raw, gid,
+			    MLX4_ROCE_GID_ENTRY_SIZE)) {
+			found_ix = i;
+			break;
+		}
+	}
+
+	if (found_ix >= 0) {
+		/* Calculate a slave_gid which is the slave number in the gid
+		 * table and not a globally unique slave number.
+		 */
+		if (found_ix < MLX4_ROCE_PF_GIDS)
+			slave_gid = 0;
+		else if (found_ix < MLX4_ROCE_PF_GIDS + (vf_gids % num_vfs) *
+			 (vf_gids / num_vfs + 1))
+			slave_gid = ((found_ix - MLX4_ROCE_PF_GIDS) /
+				     (vf_gids / num_vfs + 1)) + 1;
+		else
+			slave_gid =
+			((found_ix - MLX4_ROCE_PF_GIDS -
+			  ((vf_gids % num_vfs) * ((vf_gids / num_vfs + 1)))) /
+			 (vf_gids / num_vfs)) + vf_gids % num_vfs + 1;
+
+		/* Calculate the globally unique slave id */
+		if (slave_gid) {
+			struct mlx4_active_ports exclusive_ports;
+			struct mlx4_active_ports actv_ports;
+			struct mlx4_slaves_pport slaves_pport_actv;
+			unsigned max_port_p_one;
+			int num_vfs_before = 0;
+			int candidate_slave_gid;
+
+			/* Calculate how many VFs are on the previous port, if exists */
+			for (i = 1; i < port; i++) {
+				bitmap_zero(exclusive_ports.ports, dev->caps.num_ports);
+				set_bit(i - 1, exclusive_ports.ports);
+				slaves_pport_actv =
+					mlx4_phys_to_slaves_pport_actv(
+							dev, &exclusive_ports);
+				num_vfs_before += bitmap_weight(
+						slaves_pport_actv.slaves,
+						dev->persist->num_vfs + 1);
+			}
+
+			/* candidate_slave_gid isn't necessarily the correct slave, but
+			 * it has the same number of ports and is assigned to the same
+			 * ports as the real slave we're looking for. On dual port VF,
+			 * slave_gid = [single port VFs on port <port>] +
+			 * [offset of the current slave from the first dual port VF] +
+			 * 1 (for the PF).
+			 */
+			candidate_slave_gid = slave_gid + num_vfs_before;
+
+			actv_ports = mlx4_get_active_ports(dev, candidate_slave_gid);
+			max_port_p_one = find_first_bit(
+				actv_ports.ports, dev->caps.num_ports) +
+				bitmap_weight(actv_ports.ports,
+					      dev->caps.num_ports) + 1;
+
+			/* Calculate the real slave number */
+			for (i = 1; i < max_port_p_one; i++) {
+				if (i == port)
+					continue;
+				bitmap_zero(exclusive_ports.ports,
+					    dev->caps.num_ports);
+				set_bit(i - 1, exclusive_ports.ports);
+				slaves_pport_actv =
+					mlx4_phys_to_slaves_pport_actv(
+						dev, &exclusive_ports);
+				slave_gid += bitmap_weight(
+						slaves_pport_actv.slaves,
+						dev->persist->num_vfs + 1);
+			}
+		}
+		*slave_id = slave_gid;
+	}
+
+	return (found_ix >= 0) ? 0 : -EINVAL;
+}
+EXPORT_SYMBOL(mlx4_get_slave_from_roce_gid);
+
+int mlx4_get_roce_gid_from_slave(struct mlx4_dev *dev, int port, int slave_id,
+				 u8 *gid)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	if (!mlx4_is_master(dev))
+		return -EINVAL;
+
+	memcpy(gid, priv->port[port].gid_table.roce_gids[slave_id].raw,
+	       MLX4_ROCE_GID_ENTRY_SIZE);
+	return 0;
+}
+EXPORT_SYMBOL(mlx4_get_roce_gid_from_slave);
+
+/* Cable Module Info */
+#define MODULE_INFO_MAX_READ 48
+
+#define I2C_ADDR_LOW  0x50
+#define I2C_ADDR_HIGH 0x51
+#define I2C_PAGE_SIZE 256
+#define I2C_HIGH_PAGE_SIZE 128
+
+/* Module Info Data */
+struct mlx4_cable_info {
+	u8	i2c_addr;
+	u8	page_num;
+	__be16	dev_mem_address;
+	__be16	reserved1;
+	__be16	size;
+	__be32	reserved2[2];
+	u8	data[MODULE_INFO_MAX_READ];
+};
+
+enum cable_info_err {
+	 CABLE_INF_INV_PORT      = 0x1,
+	 CABLE_INF_OP_NOSUP      = 0x2,
+	 CABLE_INF_NOT_CONN      = 0x3,
+	 CABLE_INF_NO_EEPRM      = 0x4,
+	 CABLE_INF_PAGE_ERR      = 0x5,
+	 CABLE_INF_INV_ADDR      = 0x6,
+	 CABLE_INF_I2C_ADDR      = 0x7,
+	 CABLE_INF_QSFP_VIO      = 0x8,
+	 CABLE_INF_I2C_BUSY      = 0x9,
+};
+
+#define MAD_STATUS_2_CABLE_ERR(mad_status) ((mad_status >> 8) & 0xFF)
+
+static inline const char *cable_info_mad_err_str(u16 mad_status)
+{
+	u8 err = MAD_STATUS_2_CABLE_ERR(mad_status);
+
+	switch (err) {
+	case CABLE_INF_INV_PORT:
+		return "invalid port selected";
+	case CABLE_INF_OP_NOSUP:
+		return "operation not supported for this port (the port is of type CX4 or internal)";
+	case CABLE_INF_NOT_CONN:
+		return "cable is not connected";
+	case CABLE_INF_NO_EEPRM:
+		return "the connected cable has no EPROM (passive copper cable)";
+	case CABLE_INF_PAGE_ERR:
+		return "page number is greater than 15";
+	case CABLE_INF_INV_ADDR:
+		return "invalid device_address or size (that is, size equals 0 or address+size is greater than 256)";
+	case CABLE_INF_I2C_ADDR:
+		return "invalid I2C slave address";
+	case CABLE_INF_QSFP_VIO:
+		return "at least one cable violates the QSFP specification and ignores the modsel signal";
+	case CABLE_INF_I2C_BUSY:
+		return "I2C bus is constantly busy";
+	}
+	return "Unknown Error";
+}
+
+static int mlx4_get_module_id(struct mlx4_dev *dev, u8 port, u8 *module_id)
+{
+	struct mlx4_cmd_mailbox *inbox, *outbox;
+	struct mlx4_mad_ifc *inmad, *outmad;
+	struct mlx4_cable_info *cable_info;
+	int ret;
+
+	inbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(inbox))
+		return PTR_ERR(inbox);
+
+	outbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(outbox)) {
+		mlx4_free_cmd_mailbox(dev, inbox);
+		return PTR_ERR(outbox);
+	}
+
+	inmad = (struct mlx4_mad_ifc *)(inbox->buf);
+	outmad = (struct mlx4_mad_ifc *)(outbox->buf);
+
+	inmad->method = 0x1; /* Get */
+	inmad->class_version = 0x1;
+	inmad->mgmt_class = 0x1;
+	inmad->base_version = 0x1;
+	inmad->attr_id = cpu_to_be16(0xFF60); /* Module Info */
+
+	cable_info = (struct mlx4_cable_info *)inmad->data;
+	cable_info->dev_mem_address = 0;
+	cable_info->page_num = 0;
+	cable_info->i2c_addr = I2C_ADDR_LOW;
+	cable_info->size = cpu_to_be16(1);
+
+	ret = mlx4_cmd_box(dev, inbox->dma, outbox->dma, port, 3,
+			   MLX4_CMD_MAD_IFC, MLX4_CMD_TIME_CLASS_C,
+			   MLX4_CMD_NATIVE);
+	if (ret)
+		goto out;
+
+	if (be16_to_cpu(outmad->status)) {
+		/* Mad returned with bad status */
+		ret = be16_to_cpu(outmad->status);
+		mlx4_warn(dev,
+			  "MLX4_CMD_MAD_IFC Get Module ID attr(%x) port(%d) i2c_addr(%x) offset(%d) size(%d): Response Mad Status(%x) - %s\n",
+			  0xFF60, port, I2C_ADDR_LOW, 0, 1, ret,
+			  cable_info_mad_err_str(ret));
+		ret = -ret;
+		goto out;
+	}
+	cable_info = (struct mlx4_cable_info *)outmad->data;
+	*module_id = cable_info->data[0];
+out:
+	mlx4_free_cmd_mailbox(dev, inbox);
+	mlx4_free_cmd_mailbox(dev, outbox);
+	return ret;
+}
+
+static void mlx4_sfp_eeprom_params_set(u8 *i2c_addr, u8 *page_num, u16 *offset)
+{
+	*i2c_addr = I2C_ADDR_LOW;
+	*page_num = 0;
+
+	if (*offset < I2C_PAGE_SIZE)
+		return;
+
+	*i2c_addr = I2C_ADDR_HIGH;
+	*offset -= I2C_PAGE_SIZE;
+}
+
+static void mlx4_qsfp_eeprom_params_set(u8 *i2c_addr, u8 *page_num, u16 *offset)
+{
+	/* Offsets 0-255 belong to page 0.
+	 * Offsets 256-639 belong to pages 01, 02, 03.
+	 * For example, offset 400 is page 02: 1 + (400 - 256) / 128 = 2
+	 */
+	if (*offset < I2C_PAGE_SIZE)
+		*page_num = 0;
+	else
+		*page_num = 1 + (*offset - I2C_PAGE_SIZE) / I2C_HIGH_PAGE_SIZE;
+	*i2c_addr = I2C_ADDR_LOW;
+	*offset -= *page_num * I2C_HIGH_PAGE_SIZE;
+}
+
+/**
+ * mlx4_get_module_info - Read cable module eeprom data
+ * @dev: mlx4_dev.
+ * @port: port number.
+ * @offset: byte offset in eeprom to start reading data from.
+ * @size: num of bytes to read.
+ * @data: output buffer to put the requested data into.
+ *
+ * Reads cable module eeprom data, puts the outcome data into
+ * data pointer paramer.
+ * Returns num of read bytes on success or a negative error
+ * code.
+ */
+int mlx4_get_module_info(struct mlx4_dev *dev, u8 port,
+			 u16 offset, u16 size, u8 *data)
+{
+	struct mlx4_cmd_mailbox *inbox, *outbox;
+	struct mlx4_mad_ifc *inmad, *outmad;
+	struct mlx4_cable_info *cable_info;
+	u8 module_id, i2c_addr, page_num;
+	int ret;
+
+	if (size > MODULE_INFO_MAX_READ)
+		size = MODULE_INFO_MAX_READ;
+
+	ret = mlx4_get_module_id(dev, port, &module_id);
+	if (ret)
+		return ret;
+
+	switch (module_id) {
+	case MLX4_MODULE_ID_SFP:
+		mlx4_sfp_eeprom_params_set(&i2c_addr, &page_num, &offset);
+		break;
+	case MLX4_MODULE_ID_QSFP:
+	case MLX4_MODULE_ID_QSFP_PLUS:
+	case MLX4_MODULE_ID_QSFP28:
+		mlx4_qsfp_eeprom_params_set(&i2c_addr, &page_num, &offset);
+		break;
+	default:
+		mlx4_err(dev, "Module ID not recognized: %#x\n", module_id);
+		return -EINVAL;
+	}
+
+	inbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(inbox))
+		return PTR_ERR(inbox);
+
+	outbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(outbox)) {
+		mlx4_free_cmd_mailbox(dev, inbox);
+		return PTR_ERR(outbox);
+	}
+
+	inmad = (struct mlx4_mad_ifc *)(inbox->buf);
+	outmad = (struct mlx4_mad_ifc *)(outbox->buf);
+
+	inmad->method = 0x1; /* Get */
+	inmad->class_version = 0x1;
+	inmad->mgmt_class = 0x1;
+	inmad->base_version = 0x1;
+	inmad->attr_id = cpu_to_be16(0xFF60); /* Module Info */
+
+	if (offset < I2C_PAGE_SIZE && offset + size > I2C_PAGE_SIZE)
+		/* Cross pages reads are not allowed
+		 * read until offset 256 in low page
+		 */
+		size -= offset + size - I2C_PAGE_SIZE;
+
+	cable_info = (struct mlx4_cable_info *)inmad->data;
+	cable_info->dev_mem_address = cpu_to_be16(offset);
+	cable_info->page_num = page_num;
+	cable_info->i2c_addr = i2c_addr;
+	cable_info->size = cpu_to_be16(size);
+
+	ret = mlx4_cmd_box(dev, inbox->dma, outbox->dma, port, 3,
+			   MLX4_CMD_MAD_IFC, MLX4_CMD_TIME_CLASS_C,
+			   MLX4_CMD_NATIVE);
+	if (ret)
+		goto out;
+
+	if (be16_to_cpu(outmad->status)) {
+		/* Mad returned with bad status */
+		ret = be16_to_cpu(outmad->status);
+		mlx4_warn(dev,
+			  "MLX4_CMD_MAD_IFC Get Module info attr(%x) port(%d) i2c_addr(%x) offset(%d) size(%d): Response Mad Status(%x) - %s\n",
+			  0xFF60, port, i2c_addr, offset, size,
+			  ret, cable_info_mad_err_str(ret));
+
+		if (i2c_addr == I2C_ADDR_HIGH &&
+		    MAD_STATUS_2_CABLE_ERR(ret) == CABLE_INF_I2C_ADDR)
+			/* Some SFP cables do not support i2c slave
+			 * address 0x51 (high page), abort silently.
+			 */
+			ret = 0;
+		else
+			ret = -ret;
+		goto out;
+	}
+	cable_info = (struct mlx4_cable_info *)outmad->data;
+	memcpy(data, cable_info->data, size);
+	ret = size;
+out:
+	mlx4_free_cmd_mailbox(dev, inbox);
+	mlx4_free_cmd_mailbox(dev, outbox);
+	return ret;
+}
+EXPORT_SYMBOL(mlx4_get_module_info);
+
+int mlx4_max_tc(struct mlx4_dev *dev)
+{
+	u8 num_tc = dev->caps.max_tc_eth;
+
+	if (!num_tc)
+		num_tc = MLX4_TC_MAX_NUMBER;
+
+	return num_tc;
+}
+EXPORT_SYMBOL(mlx4_max_tc);
diff --git a/drivers/net/ethernet/mellanox/mlx4/profile.c b/drivers/net/ethernet/mellanox/mlx4/profile.c
new file mode 100644
index 000000000..ba361c5fb
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx4/profile.c
@@ -0,0 +1,270 @@
+/*
+ * Copyright (c) 2004, 2005 Topspin Communications.  All rights reserved.
+ * Copyright (c) 2005 Mellanox Technologies. All rights reserved.
+ * Copyright (c) 2006, 2007 Cisco Systems, Inc. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/slab.h>
+
+#include "mlx4.h"
+#include "fw.h"
+
+enum {
+	MLX4_RES_QP,
+	MLX4_RES_RDMARC,
+	MLX4_RES_ALTC,
+	MLX4_RES_AUXC,
+	MLX4_RES_SRQ,
+	MLX4_RES_CQ,
+	MLX4_RES_EQ,
+	MLX4_RES_DMPT,
+	MLX4_RES_CMPT,
+	MLX4_RES_MTT,
+	MLX4_RES_MCG,
+	MLX4_RES_NUM
+};
+
+static const char *res_name[] = {
+	[MLX4_RES_QP]		= "QP",
+	[MLX4_RES_RDMARC]	= "RDMARC",
+	[MLX4_RES_ALTC]		= "ALTC",
+	[MLX4_RES_AUXC]		= "AUXC",
+	[MLX4_RES_SRQ]		= "SRQ",
+	[MLX4_RES_CQ]		= "CQ",
+	[MLX4_RES_EQ]		= "EQ",
+	[MLX4_RES_DMPT]		= "DMPT",
+	[MLX4_RES_CMPT]		= "CMPT",
+	[MLX4_RES_MTT]		= "MTT",
+	[MLX4_RES_MCG]		= "MCG",
+};
+
+u64 mlx4_make_profile(struct mlx4_dev *dev,
+		      struct mlx4_profile *request,
+		      struct mlx4_dev_cap *dev_cap,
+		      struct mlx4_init_hca_param *init_hca)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_resource {
+		u64 size;
+		u64 start;
+		int type;
+		u32 num;
+		int log_num;
+	};
+
+	u64 total_size = 0;
+	struct mlx4_resource *profile;
+	struct sysinfo si;
+	int i, j;
+
+	profile = kcalloc(MLX4_RES_NUM, sizeof(*profile), GFP_KERNEL);
+	if (!profile)
+		return -ENOMEM;
+
+	/*
+	 * We want to scale the number of MTTs with the size of the
+	 * system memory, since it makes sense to register a lot of
+	 * memory on a system with a lot of memory.  As a heuristic,
+	 * make sure we have enough MTTs to cover twice the system
+	 * memory (with PAGE_SIZE entries).
+	 *
+	 * This number has to be a power of two and fit into 32 bits
+	 * due to device limitations, so cap this at 2^31 as well.
+	 * That limits us to 8TB of memory registration per HCA with
+	 * 4KB pages, which is probably OK for the next few months.
+	 */
+	si_meminfo(&si);
+	request->num_mtt =
+		roundup_pow_of_two(max_t(unsigned, request->num_mtt,
+					 min(1UL << (31 - log_mtts_per_seg),
+					     (si.totalram << 1) >> log_mtts_per_seg)));
+
+
+	profile[MLX4_RES_QP].size     = dev_cap->qpc_entry_sz;
+	profile[MLX4_RES_RDMARC].size = dev_cap->rdmarc_entry_sz;
+	profile[MLX4_RES_ALTC].size   = dev_cap->altc_entry_sz;
+	profile[MLX4_RES_AUXC].size   = dev_cap->aux_entry_sz;
+	profile[MLX4_RES_SRQ].size    = dev_cap->srq_entry_sz;
+	profile[MLX4_RES_CQ].size     = dev_cap->cqc_entry_sz;
+	profile[MLX4_RES_EQ].size     = dev_cap->eqc_entry_sz;
+	profile[MLX4_RES_DMPT].size   = dev_cap->dmpt_entry_sz;
+	profile[MLX4_RES_CMPT].size   = dev_cap->cmpt_entry_sz;
+	profile[MLX4_RES_MTT].size    = dev_cap->mtt_entry_sz;
+	profile[MLX4_RES_MCG].size    = mlx4_get_mgm_entry_size(dev);
+
+	profile[MLX4_RES_QP].num      = request->num_qp;
+	profile[MLX4_RES_RDMARC].num  = request->num_qp * request->rdmarc_per_qp;
+	profile[MLX4_RES_ALTC].num    = request->num_qp;
+	profile[MLX4_RES_AUXC].num    = request->num_qp;
+	profile[MLX4_RES_SRQ].num     = request->num_srq;
+	profile[MLX4_RES_CQ].num      = request->num_cq;
+	profile[MLX4_RES_EQ].num = mlx4_is_mfunc(dev) ? dev->phys_caps.num_phys_eqs :
+					min_t(unsigned, dev_cap->max_eqs, MAX_MSIX);
+	profile[MLX4_RES_DMPT].num    = request->num_mpt;
+	profile[MLX4_RES_CMPT].num    = MLX4_NUM_CMPTS;
+	profile[MLX4_RES_MTT].num     = request->num_mtt * (1 << log_mtts_per_seg);
+	profile[MLX4_RES_MCG].num     = request->num_mcg;
+
+	for (i = 0; i < MLX4_RES_NUM; ++i) {
+		profile[i].type     = i;
+		profile[i].num      = roundup_pow_of_two(profile[i].num);
+		profile[i].log_num  = ilog2(profile[i].num);
+		profile[i].size    *= profile[i].num;
+		profile[i].size     = max(profile[i].size, (u64) PAGE_SIZE);
+	}
+
+	/*
+	 * Sort the resources in decreasing order of size.  Since they
+	 * all have sizes that are powers of 2, we'll be able to keep
+	 * resources aligned to their size and pack them without gaps
+	 * using the sorted order.
+	 */
+	for (i = MLX4_RES_NUM; i > 0; --i)
+		for (j = 1; j < i; ++j) {
+			if (profile[j].size > profile[j - 1].size)
+				swap(profile[j], profile[j - 1]);
+		}
+
+	for (i = 0; i < MLX4_RES_NUM; ++i) {
+		if (profile[i].size) {
+			profile[i].start = total_size;
+			total_size	+= profile[i].size;
+		}
+
+		if (total_size > dev_cap->max_icm_sz) {
+			mlx4_err(dev, "Profile requires 0x%llx bytes; won't fit in 0x%llx bytes of context memory\n",
+				 (unsigned long long) total_size,
+				 (unsigned long long) dev_cap->max_icm_sz);
+			kfree(profile);
+			return -ENOMEM;
+		}
+
+		if (profile[i].size)
+			mlx4_dbg(dev, "  profile[%2d] (%6s): 2^%02d entries @ 0x%10llx, size 0x%10llx\n",
+				 i, res_name[profile[i].type],
+				 profile[i].log_num,
+				 (unsigned long long) profile[i].start,
+				 (unsigned long long) profile[i].size);
+	}
+
+	mlx4_dbg(dev, "HCA context memory: reserving %d KB\n",
+		 (int) (total_size >> 10));
+
+	for (i = 0; i < MLX4_RES_NUM; ++i) {
+		switch (profile[i].type) {
+		case MLX4_RES_QP:
+			dev->caps.num_qps     = profile[i].num;
+			init_hca->qpc_base    = profile[i].start;
+			init_hca->log_num_qps = profile[i].log_num;
+			break;
+		case MLX4_RES_RDMARC:
+			for (priv->qp_table.rdmarc_shift = 0;
+			     request->num_qp << priv->qp_table.rdmarc_shift < profile[i].num;
+			     ++priv->qp_table.rdmarc_shift)
+				; /* nothing */
+			dev->caps.max_qp_dest_rdma = 1 << priv->qp_table.rdmarc_shift;
+			priv->qp_table.rdmarc_base   = (u32) profile[i].start;
+			init_hca->rdmarc_base	     = profile[i].start;
+			init_hca->log_rd_per_qp	     = priv->qp_table.rdmarc_shift;
+			break;
+		case MLX4_RES_ALTC:
+			init_hca->altc_base = profile[i].start;
+			break;
+		case MLX4_RES_AUXC:
+			init_hca->auxc_base = profile[i].start;
+			break;
+		case MLX4_RES_SRQ:
+			dev->caps.num_srqs     = profile[i].num;
+			init_hca->srqc_base    = profile[i].start;
+			init_hca->log_num_srqs = profile[i].log_num;
+			break;
+		case MLX4_RES_CQ:
+			dev->caps.num_cqs     = profile[i].num;
+			init_hca->cqc_base    = profile[i].start;
+			init_hca->log_num_cqs = profile[i].log_num;
+			break;
+		case MLX4_RES_EQ:
+			if (dev_cap->flags2 & MLX4_DEV_CAP_FLAG2_SYS_EQS) {
+				init_hca->log_num_eqs = 0x1f;
+				init_hca->eqc_base    = profile[i].start;
+				init_hca->num_sys_eqs = dev_cap->num_sys_eqs;
+			} else {
+				dev->caps.num_eqs     = roundup_pow_of_two(
+								min_t(unsigned,
+								      dev_cap->max_eqs,
+								      MAX_MSIX));
+				init_hca->eqc_base    = profile[i].start;
+				init_hca->log_num_eqs = ilog2(dev->caps.num_eqs);
+			}
+			break;
+		case MLX4_RES_DMPT:
+			dev->caps.num_mpts	= profile[i].num;
+			priv->mr_table.mpt_base = profile[i].start;
+			init_hca->dmpt_base	= profile[i].start;
+			init_hca->log_mpt_sz	= profile[i].log_num;
+			break;
+		case MLX4_RES_CMPT:
+			init_hca->cmpt_base	 = profile[i].start;
+			break;
+		case MLX4_RES_MTT:
+			dev->caps.num_mtts	 = profile[i].num;
+			priv->mr_table.mtt_base	 = profile[i].start;
+			init_hca->mtt_base	 = profile[i].start;
+			break;
+		case MLX4_RES_MCG:
+			init_hca->mc_base	  = profile[i].start;
+			init_hca->log_mc_entry_sz =
+					ilog2(mlx4_get_mgm_entry_size(dev));
+			init_hca->log_mc_table_sz = profile[i].log_num;
+			if (dev->caps.steering_mode ==
+			    MLX4_STEERING_MODE_DEVICE_MANAGED) {
+				dev->caps.num_mgms = profile[i].num;
+			} else {
+				init_hca->log_mc_hash_sz =
+						profile[i].log_num - 1;
+				dev->caps.num_mgms = profile[i].num >> 1;
+				dev->caps.num_amgms = profile[i].num >> 1;
+			}
+			break;
+		default:
+			break;
+		}
+	}
+
+	/*
+	 * PDs don't take any HCA memory, but we assign them as part
+	 * of the HCA profile anyway.
+	 */
+	dev->caps.num_pds = MLX4_NUM_PDS;
+
+	kfree(profile);
+	return total_size;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx4/qp.c b/drivers/net/ethernet/mellanox/mlx4/qp.c
new file mode 100644
index 000000000..48cfaa7ea
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx4/qp.c
@@ -0,0 +1,966 @@
+/*
+ * Copyright (c) 2004 Topspin Communications.  All rights reserved.
+ * Copyright (c) 2005, 2006, 2007 Cisco Systems, Inc. All rights reserved.
+ * Copyright (c) 2005, 2006, 2007, 2008 Mellanox Technologies. All rights reserved.
+ * Copyright (c) 2004 Voltaire, Inc. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/gfp.h>
+#include <linux/export.h>
+
+#include <linux/mlx4/cmd.h>
+#include <linux/mlx4/qp.h>
+
+#include "mlx4.h"
+#include "icm.h"
+
+/* QP to support BF should have bits 6,7 cleared */
+#define MLX4_BF_QP_SKIP_MASK	0xc0
+#define MLX4_MAX_BF_QP_RANGE	0x40
+
+void mlx4_qp_event(struct mlx4_dev *dev, u32 qpn, int event_type)
+{
+	struct mlx4_qp_table *qp_table = &mlx4_priv(dev)->qp_table;
+	struct mlx4_qp *qp;
+
+	spin_lock(&qp_table->lock);
+
+	qp = __mlx4_qp_lookup(dev, qpn);
+	if (qp)
+		refcount_inc(&qp->refcount);
+
+	spin_unlock(&qp_table->lock);
+
+	if (!qp) {
+		mlx4_dbg(dev, "Async event for none existent QP %08x\n", qpn);
+		return;
+	}
+
+	qp->event(qp, event_type);
+
+	if (refcount_dec_and_test(&qp->refcount))
+		complete(&qp->free);
+}
+
+/* used for INIT/CLOSE port logic */
+static int is_master_qp0(struct mlx4_dev *dev, struct mlx4_qp *qp, int *real_qp0, int *proxy_qp0)
+{
+	/* this procedure is called after we already know we are on the master */
+	/* qp0 is either the proxy qp0, or the real qp0 */
+	u32 pf_proxy_offset = dev->phys_caps.base_proxy_sqpn + 8 * mlx4_master_func_num(dev);
+	*proxy_qp0 = qp->qpn >= pf_proxy_offset && qp->qpn <= pf_proxy_offset + 1;
+
+	*real_qp0 = qp->qpn >= dev->phys_caps.base_sqpn &&
+		qp->qpn <= dev->phys_caps.base_sqpn + 1;
+
+	return *real_qp0 || *proxy_qp0;
+}
+
+static int __mlx4_qp_modify(struct mlx4_dev *dev, struct mlx4_mtt *mtt,
+		     enum mlx4_qp_state cur_state, enum mlx4_qp_state new_state,
+		     struct mlx4_qp_context *context,
+		     enum mlx4_qp_optpar optpar,
+		     int sqd_event, struct mlx4_qp *qp, int native)
+{
+	static const u16 op[MLX4_QP_NUM_STATE][MLX4_QP_NUM_STATE] = {
+		[MLX4_QP_STATE_RST] = {
+			[MLX4_QP_STATE_RST]	= MLX4_CMD_2RST_QP,
+			[MLX4_QP_STATE_ERR]	= MLX4_CMD_2ERR_QP,
+			[MLX4_QP_STATE_INIT]	= MLX4_CMD_RST2INIT_QP,
+		},
+		[MLX4_QP_STATE_INIT]  = {
+			[MLX4_QP_STATE_RST]	= MLX4_CMD_2RST_QP,
+			[MLX4_QP_STATE_ERR]	= MLX4_CMD_2ERR_QP,
+			[MLX4_QP_STATE_INIT]	= MLX4_CMD_INIT2INIT_QP,
+			[MLX4_QP_STATE_RTR]	= MLX4_CMD_INIT2RTR_QP,
+		},
+		[MLX4_QP_STATE_RTR]   = {
+			[MLX4_QP_STATE_RST]	= MLX4_CMD_2RST_QP,
+			[MLX4_QP_STATE_ERR]	= MLX4_CMD_2ERR_QP,
+			[MLX4_QP_STATE_RTS]	= MLX4_CMD_RTR2RTS_QP,
+		},
+		[MLX4_QP_STATE_RTS]   = {
+			[MLX4_QP_STATE_RST]	= MLX4_CMD_2RST_QP,
+			[MLX4_QP_STATE_ERR]	= MLX4_CMD_2ERR_QP,
+			[MLX4_QP_STATE_RTS]	= MLX4_CMD_RTS2RTS_QP,
+			[MLX4_QP_STATE_SQD]	= MLX4_CMD_RTS2SQD_QP,
+		},
+		[MLX4_QP_STATE_SQD] = {
+			[MLX4_QP_STATE_RST]	= MLX4_CMD_2RST_QP,
+			[MLX4_QP_STATE_ERR]	= MLX4_CMD_2ERR_QP,
+			[MLX4_QP_STATE_RTS]	= MLX4_CMD_SQD2RTS_QP,
+			[MLX4_QP_STATE_SQD]	= MLX4_CMD_SQD2SQD_QP,
+		},
+		[MLX4_QP_STATE_SQER] = {
+			[MLX4_QP_STATE_RST]	= MLX4_CMD_2RST_QP,
+			[MLX4_QP_STATE_ERR]	= MLX4_CMD_2ERR_QP,
+			[MLX4_QP_STATE_RTS]	= MLX4_CMD_SQERR2RTS_QP,
+		},
+		[MLX4_QP_STATE_ERR] = {
+			[MLX4_QP_STATE_RST]	= MLX4_CMD_2RST_QP,
+			[MLX4_QP_STATE_ERR]	= MLX4_CMD_2ERR_QP,
+		}
+	};
+
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_cmd_mailbox *mailbox;
+	int ret = 0;
+	int real_qp0 = 0;
+	int proxy_qp0 = 0;
+	u8 port;
+
+	if (cur_state >= MLX4_QP_NUM_STATE || new_state >= MLX4_QP_NUM_STATE ||
+	    !op[cur_state][new_state])
+		return -EINVAL;
+
+	if (op[cur_state][new_state] == MLX4_CMD_2RST_QP) {
+		ret = mlx4_cmd(dev, 0, qp->qpn, 2,
+			MLX4_CMD_2RST_QP, MLX4_CMD_TIME_CLASS_A, native);
+		if (mlx4_is_master(dev) && cur_state != MLX4_QP_STATE_ERR &&
+		    cur_state != MLX4_QP_STATE_RST &&
+		    is_master_qp0(dev, qp, &real_qp0, &proxy_qp0)) {
+			port = (qp->qpn & 1) + 1;
+			if (proxy_qp0)
+				priv->mfunc.master.qp0_state[port].proxy_qp0_active = 0;
+			else
+				priv->mfunc.master.qp0_state[port].qp0_active = 0;
+		}
+		return ret;
+	}
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+
+	if (cur_state == MLX4_QP_STATE_RST && new_state == MLX4_QP_STATE_INIT) {
+		u64 mtt_addr = mlx4_mtt_addr(dev, mtt);
+		context->mtt_base_addr_h = mtt_addr >> 32;
+		context->mtt_base_addr_l = cpu_to_be32(mtt_addr & 0xffffffff);
+		context->log_page_size   = mtt->page_shift - MLX4_ICM_PAGE_SHIFT;
+	}
+
+	if ((cur_state == MLX4_QP_STATE_RTR) &&
+	    (new_state == MLX4_QP_STATE_RTS) &&
+	    dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_ROCE_V1_V2)
+		context->roce_entropy =
+			cpu_to_be16(mlx4_qp_roce_entropy(dev, qp->qpn));
+
+	*(__be32 *) mailbox->buf = cpu_to_be32(optpar);
+	memcpy(mailbox->buf + 8, context, sizeof(*context));
+
+	((struct mlx4_qp_context *) (mailbox->buf + 8))->local_qpn =
+		cpu_to_be32(qp->qpn);
+
+	ret = mlx4_cmd(dev, mailbox->dma,
+		       qp->qpn | (!!sqd_event << 31),
+		       new_state == MLX4_QP_STATE_RST ? 2 : 0,
+		       op[cur_state][new_state], MLX4_CMD_TIME_CLASS_C, native);
+
+	if (mlx4_is_master(dev) && is_master_qp0(dev, qp, &real_qp0, &proxy_qp0)) {
+		port = (qp->qpn & 1) + 1;
+		if (cur_state != MLX4_QP_STATE_ERR &&
+		    cur_state != MLX4_QP_STATE_RST &&
+		    new_state == MLX4_QP_STATE_ERR) {
+			if (proxy_qp0)
+				priv->mfunc.master.qp0_state[port].proxy_qp0_active = 0;
+			else
+				priv->mfunc.master.qp0_state[port].qp0_active = 0;
+		} else if (new_state == MLX4_QP_STATE_RTR) {
+			if (proxy_qp0)
+				priv->mfunc.master.qp0_state[port].proxy_qp0_active = 1;
+			else
+				priv->mfunc.master.qp0_state[port].qp0_active = 1;
+		}
+	}
+
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return ret;
+}
+
+int mlx4_qp_modify(struct mlx4_dev *dev, struct mlx4_mtt *mtt,
+		   enum mlx4_qp_state cur_state, enum mlx4_qp_state new_state,
+		   struct mlx4_qp_context *context,
+		   enum mlx4_qp_optpar optpar,
+		   int sqd_event, struct mlx4_qp *qp)
+{
+	return __mlx4_qp_modify(dev, mtt, cur_state, new_state, context,
+				optpar, sqd_event, qp, 0);
+}
+EXPORT_SYMBOL_GPL(mlx4_qp_modify);
+
+int __mlx4_qp_reserve_range(struct mlx4_dev *dev, int cnt, int align,
+			    int *base, u8 flags)
+{
+	u32 uid;
+	int bf_qp = !!(flags & (u8)MLX4_RESERVE_ETH_BF_QP);
+
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_qp_table *qp_table = &priv->qp_table;
+
+	if (cnt > MLX4_MAX_BF_QP_RANGE && bf_qp)
+		return -ENOMEM;
+
+	uid = MLX4_QP_TABLE_ZONE_GENERAL;
+	if (flags & (u8)MLX4_RESERVE_A0_QP) {
+		if (bf_qp)
+			uid = MLX4_QP_TABLE_ZONE_RAW_ETH;
+		else
+			uid = MLX4_QP_TABLE_ZONE_RSS;
+	}
+
+	*base = mlx4_zone_alloc_entries(qp_table->zones, uid, cnt, align,
+					bf_qp ? MLX4_BF_QP_SKIP_MASK : 0, NULL);
+	if (*base == -1)
+		return -ENOMEM;
+
+	return 0;
+}
+
+int mlx4_qp_reserve_range(struct mlx4_dev *dev, int cnt, int align,
+			  int *base, u8 flags, u8 usage)
+{
+	u32 in_modifier = RES_QP | (((u32)usage & 3) << 30);
+	u64 in_param = 0;
+	u64 out_param;
+	int err;
+
+	/* Turn off all unsupported QP allocation flags */
+	flags &= dev->caps.alloc_res_qp_mask;
+
+	if (mlx4_is_mfunc(dev)) {
+		set_param_l(&in_param, (((u32)flags) << 24) | (u32)cnt);
+		set_param_h(&in_param, align);
+		err = mlx4_cmd_imm(dev, in_param, &out_param,
+				   in_modifier, RES_OP_RESERVE,
+				   MLX4_CMD_ALLOC_RES,
+				   MLX4_CMD_TIME_CLASS_A, MLX4_CMD_WRAPPED);
+		if (err)
+			return err;
+
+		*base = get_param_l(&out_param);
+		return 0;
+	}
+	return __mlx4_qp_reserve_range(dev, cnt, align, base, flags);
+}
+EXPORT_SYMBOL_GPL(mlx4_qp_reserve_range);
+
+void __mlx4_qp_release_range(struct mlx4_dev *dev, int base_qpn, int cnt)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_qp_table *qp_table = &priv->qp_table;
+
+	if (mlx4_is_qp_reserved(dev, (u32) base_qpn))
+		return;
+	mlx4_zone_free_entries_unique(qp_table->zones, base_qpn, cnt);
+}
+
+void mlx4_qp_release_range(struct mlx4_dev *dev, int base_qpn, int cnt)
+{
+	u64 in_param = 0;
+	int err;
+
+	if (!cnt)
+		return;
+
+	if (mlx4_is_mfunc(dev)) {
+		set_param_l(&in_param, base_qpn);
+		set_param_h(&in_param, cnt);
+		err = mlx4_cmd(dev, in_param, RES_QP, RES_OP_RESERVE,
+			       MLX4_CMD_FREE_RES,
+			       MLX4_CMD_TIME_CLASS_A, MLX4_CMD_WRAPPED);
+		if (err) {
+			mlx4_warn(dev, "Failed to release qp range base:%d cnt:%d\n",
+				  base_qpn, cnt);
+		}
+	} else
+		 __mlx4_qp_release_range(dev, base_qpn, cnt);
+}
+EXPORT_SYMBOL_GPL(mlx4_qp_release_range);
+
+int __mlx4_qp_alloc_icm(struct mlx4_dev *dev, int qpn)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_qp_table *qp_table = &priv->qp_table;
+	int err;
+
+	err = mlx4_table_get(dev, &qp_table->qp_table, qpn);
+	if (err)
+		goto err_out;
+
+	err = mlx4_table_get(dev, &qp_table->auxc_table, qpn);
+	if (err)
+		goto err_put_qp;
+
+	err = mlx4_table_get(dev, &qp_table->altc_table, qpn);
+	if (err)
+		goto err_put_auxc;
+
+	err = mlx4_table_get(dev, &qp_table->rdmarc_table, qpn);
+	if (err)
+		goto err_put_altc;
+
+	err = mlx4_table_get(dev, &qp_table->cmpt_table, qpn);
+	if (err)
+		goto err_put_rdmarc;
+
+	return 0;
+
+err_put_rdmarc:
+	mlx4_table_put(dev, &qp_table->rdmarc_table, qpn);
+
+err_put_altc:
+	mlx4_table_put(dev, &qp_table->altc_table, qpn);
+
+err_put_auxc:
+	mlx4_table_put(dev, &qp_table->auxc_table, qpn);
+
+err_put_qp:
+	mlx4_table_put(dev, &qp_table->qp_table, qpn);
+
+err_out:
+	return err;
+}
+
+static int mlx4_qp_alloc_icm(struct mlx4_dev *dev, int qpn)
+{
+	u64 param = 0;
+
+	if (mlx4_is_mfunc(dev)) {
+		set_param_l(&param, qpn);
+		return mlx4_cmd_imm(dev, param, &param, RES_QP, RES_OP_MAP_ICM,
+				    MLX4_CMD_ALLOC_RES, MLX4_CMD_TIME_CLASS_A,
+				    MLX4_CMD_WRAPPED);
+	}
+	return __mlx4_qp_alloc_icm(dev, qpn);
+}
+
+void __mlx4_qp_free_icm(struct mlx4_dev *dev, int qpn)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_qp_table *qp_table = &priv->qp_table;
+
+	mlx4_table_put(dev, &qp_table->cmpt_table, qpn);
+	mlx4_table_put(dev, &qp_table->rdmarc_table, qpn);
+	mlx4_table_put(dev, &qp_table->altc_table, qpn);
+	mlx4_table_put(dev, &qp_table->auxc_table, qpn);
+	mlx4_table_put(dev, &qp_table->qp_table, qpn);
+}
+
+static void mlx4_qp_free_icm(struct mlx4_dev *dev, int qpn)
+{
+	u64 in_param = 0;
+
+	if (mlx4_is_mfunc(dev)) {
+		set_param_l(&in_param, qpn);
+		if (mlx4_cmd(dev, in_param, RES_QP, RES_OP_MAP_ICM,
+			     MLX4_CMD_FREE_RES, MLX4_CMD_TIME_CLASS_A,
+			     MLX4_CMD_WRAPPED))
+			mlx4_warn(dev, "Failed to free icm of qp:%d\n", qpn);
+	} else
+		__mlx4_qp_free_icm(dev, qpn);
+}
+
+struct mlx4_qp *mlx4_qp_lookup(struct mlx4_dev *dev, u32 qpn)
+{
+	struct mlx4_qp_table *qp_table = &mlx4_priv(dev)->qp_table;
+	struct mlx4_qp *qp;
+
+	spin_lock_irq(&qp_table->lock);
+
+	qp = __mlx4_qp_lookup(dev, qpn);
+
+	spin_unlock_irq(&qp_table->lock);
+	return qp;
+}
+
+int mlx4_qp_alloc(struct mlx4_dev *dev, int qpn, struct mlx4_qp *qp)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_qp_table *qp_table = &priv->qp_table;
+	int err;
+
+	if (!qpn)
+		return -EINVAL;
+
+	qp->qpn = qpn;
+
+	err = mlx4_qp_alloc_icm(dev, qpn);
+	if (err)
+		return err;
+
+	spin_lock_irq(&qp_table->lock);
+	err = radix_tree_insert(&dev->qp_table_tree, qp->qpn &
+				(dev->caps.num_qps - 1), qp);
+	spin_unlock_irq(&qp_table->lock);
+	if (err)
+		goto err_icm;
+
+	refcount_set(&qp->refcount, 1);
+	init_completion(&qp->free);
+
+	return 0;
+
+err_icm:
+	mlx4_qp_free_icm(dev, qpn);
+	return err;
+}
+
+EXPORT_SYMBOL_GPL(mlx4_qp_alloc);
+
+int mlx4_update_qp(struct mlx4_dev *dev, u32 qpn,
+		   enum mlx4_update_qp_attr attr,
+		   struct mlx4_update_qp_params *params)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	struct mlx4_update_qp_context *cmd;
+	u64 pri_addr_path_mask = 0;
+	u64 qp_mask = 0;
+	int err = 0;
+
+	if (!attr || (attr & ~MLX4_UPDATE_QP_SUPPORTED_ATTRS))
+		return -EINVAL;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+
+	cmd = (struct mlx4_update_qp_context *)mailbox->buf;
+
+	if (attr & MLX4_UPDATE_QP_SMAC) {
+		pri_addr_path_mask |= 1ULL << MLX4_UPD_QP_PATH_MASK_MAC_INDEX;
+		cmd->qp_context.pri_path.grh_mylmc = params->smac_index;
+	}
+
+	if (attr & MLX4_UPDATE_QP_ETH_SRC_CHECK_MC_LB) {
+		if (!(dev->caps.flags2
+		      & MLX4_DEV_CAP_FLAG2_UPDATE_QP_SRC_CHECK_LB)) {
+			mlx4_warn(dev,
+				  "Trying to set src check LB, but it isn't supported\n");
+			err = -EOPNOTSUPP;
+			goto out;
+		}
+		pri_addr_path_mask |=
+			1ULL << MLX4_UPD_QP_PATH_MASK_ETH_SRC_CHECK_MC_LB;
+		if (params->flags &
+		    MLX4_UPDATE_QP_PARAMS_FLAGS_ETH_CHECK_MC_LB) {
+			cmd->qp_context.pri_path.fl |=
+				MLX4_FL_ETH_SRC_CHECK_MC_LB;
+		}
+	}
+
+	if (attr & MLX4_UPDATE_QP_VSD) {
+		qp_mask |= 1ULL << MLX4_UPD_QP_MASK_VSD;
+		if (params->flags & MLX4_UPDATE_QP_PARAMS_FLAGS_VSD_ENABLE)
+			cmd->qp_context.param3 |= cpu_to_be32(MLX4_STRIP_VLAN);
+	}
+
+	if (attr & MLX4_UPDATE_QP_RATE_LIMIT) {
+		qp_mask |= 1ULL << MLX4_UPD_QP_MASK_RATE_LIMIT;
+		cmd->qp_context.rate_limit_params = cpu_to_be16((params->rate_unit << 14) | params->rate_val);
+	}
+
+	if (attr & MLX4_UPDATE_QP_QOS_VPORT) {
+		if (!(dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_QOS_VPP)) {
+			mlx4_warn(dev, "Granular QoS per VF is not enabled\n");
+			err = -EOPNOTSUPP;
+			goto out;
+		}
+
+		qp_mask |= 1ULL << MLX4_UPD_QP_MASK_QOS_VPP;
+		cmd->qp_context.qos_vport = params->qos_vport;
+	}
+
+	cmd->primary_addr_path_mask = cpu_to_be64(pri_addr_path_mask);
+	cmd->qp_mask = cpu_to_be64(qp_mask);
+
+	err = mlx4_cmd(dev, mailbox->dma, qpn & 0xffffff, 0,
+		       MLX4_CMD_UPDATE_QP, MLX4_CMD_TIME_CLASS_A,
+		       MLX4_CMD_NATIVE);
+out:
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx4_update_qp);
+
+void mlx4_qp_remove(struct mlx4_dev *dev, struct mlx4_qp *qp)
+{
+	struct mlx4_qp_table *qp_table = &mlx4_priv(dev)->qp_table;
+	unsigned long flags;
+
+	spin_lock_irqsave(&qp_table->lock, flags);
+	radix_tree_delete(&dev->qp_table_tree, qp->qpn & (dev->caps.num_qps - 1));
+	spin_unlock_irqrestore(&qp_table->lock, flags);
+}
+EXPORT_SYMBOL_GPL(mlx4_qp_remove);
+
+void mlx4_qp_free(struct mlx4_dev *dev, struct mlx4_qp *qp)
+{
+	if (refcount_dec_and_test(&qp->refcount))
+		complete(&qp->free);
+	wait_for_completion(&qp->free);
+
+	mlx4_qp_free_icm(dev, qp->qpn);
+}
+EXPORT_SYMBOL_GPL(mlx4_qp_free);
+
+static int mlx4_CONF_SPECIAL_QP(struct mlx4_dev *dev, u32 base_qpn)
+{
+	return mlx4_cmd(dev, 0, base_qpn, 0, MLX4_CMD_CONF_SPECIAL_QP,
+			MLX4_CMD_TIME_CLASS_B, MLX4_CMD_NATIVE);
+}
+
+#define MLX4_QP_TABLE_RSS_ETH_PRIORITY 2
+#define MLX4_QP_TABLE_RAW_ETH_PRIORITY 1
+#define MLX4_QP_TABLE_RAW_ETH_SIZE     256
+
+static int mlx4_create_zones(struct mlx4_dev *dev,
+			     u32 reserved_bottom_general,
+			     u32 reserved_top_general,
+			     u32 reserved_bottom_rss,
+			     u32 start_offset_rss,
+			     u32 max_table_offset)
+{
+	struct mlx4_qp_table *qp_table = &mlx4_priv(dev)->qp_table;
+	struct mlx4_bitmap (*bitmap)[MLX4_QP_TABLE_ZONE_NUM] = NULL;
+	int bitmap_initialized = 0;
+	u32 last_offset;
+	int k;
+	int err;
+
+	qp_table->zones = mlx4_zone_allocator_create(MLX4_ZONE_ALLOC_FLAGS_NO_OVERLAP);
+
+	if (NULL == qp_table->zones)
+		return -ENOMEM;
+
+	bitmap = kmalloc(sizeof(*bitmap), GFP_KERNEL);
+
+	if (NULL == bitmap) {
+		err = -ENOMEM;
+		goto free_zone;
+	}
+
+	err = mlx4_bitmap_init(*bitmap + MLX4_QP_TABLE_ZONE_GENERAL, dev->caps.num_qps,
+			       (1 << 23) - 1, reserved_bottom_general,
+			       reserved_top_general);
+
+	if (err)
+		goto free_bitmap;
+
+	++bitmap_initialized;
+
+	err = mlx4_zone_add_one(qp_table->zones, *bitmap + MLX4_QP_TABLE_ZONE_GENERAL,
+				MLX4_ZONE_FALLBACK_TO_HIGHER_PRIO |
+				MLX4_ZONE_USE_RR, 0,
+				0, qp_table->zones_uids + MLX4_QP_TABLE_ZONE_GENERAL);
+
+	if (err)
+		goto free_bitmap;
+
+	err = mlx4_bitmap_init(*bitmap + MLX4_QP_TABLE_ZONE_RSS,
+			       reserved_bottom_rss,
+			       reserved_bottom_rss - 1,
+			       dev->caps.reserved_qps_cnt[MLX4_QP_REGION_FW],
+			       reserved_bottom_rss - start_offset_rss);
+
+	if (err)
+		goto free_bitmap;
+
+	++bitmap_initialized;
+
+	err = mlx4_zone_add_one(qp_table->zones, *bitmap + MLX4_QP_TABLE_ZONE_RSS,
+				MLX4_ZONE_ALLOW_ALLOC_FROM_LOWER_PRIO |
+				MLX4_ZONE_ALLOW_ALLOC_FROM_EQ_PRIO |
+				MLX4_ZONE_USE_RR, MLX4_QP_TABLE_RSS_ETH_PRIORITY,
+				0, qp_table->zones_uids + MLX4_QP_TABLE_ZONE_RSS);
+
+	if (err)
+		goto free_bitmap;
+
+	last_offset = dev->caps.reserved_qps_cnt[MLX4_QP_REGION_FW];
+	/*  We have a single zone for the A0 steering QPs area of the FW. This area
+	 *  needs to be split into subareas. One set of subareas is for RSS QPs
+	 *  (in which qp number bits 6 and/or 7 are set); the other set of subareas
+	 *  is for RAW_ETH QPs, which require that both bits 6 and 7 are zero.
+	 *  Currently, the values returned by the FW (A0 steering area starting qp number
+	 *  and A0 steering area size) are such that there are only two subareas -- one
+	 *  for RSS and one for RAW_ETH.
+	 */
+	for (k = MLX4_QP_TABLE_ZONE_RSS + 1; k < sizeof(*bitmap)/sizeof((*bitmap)[0]);
+	     k++) {
+		int size;
+		u32 offset = start_offset_rss;
+		u32 bf_mask;
+		u32 requested_size;
+
+		/* Assuming MLX4_BF_QP_SKIP_MASK is consecutive ones, this calculates
+		 * a mask of all LSB bits set until (and not including) the first
+		 * set bit of  MLX4_BF_QP_SKIP_MASK. For example, if MLX4_BF_QP_SKIP_MASK
+		 * is 0xc0, bf_mask will be 0x3f.
+		 */
+		bf_mask = (MLX4_BF_QP_SKIP_MASK & ~(MLX4_BF_QP_SKIP_MASK - 1)) - 1;
+		requested_size = min((u32)MLX4_QP_TABLE_RAW_ETH_SIZE, bf_mask + 1);
+
+		if (((last_offset & MLX4_BF_QP_SKIP_MASK) &&
+		     ((int)(max_table_offset - last_offset)) >=
+		     roundup_pow_of_two(MLX4_BF_QP_SKIP_MASK)) ||
+		    (!(last_offset & MLX4_BF_QP_SKIP_MASK) &&
+		     !((last_offset + requested_size - 1) &
+		       MLX4_BF_QP_SKIP_MASK)))
+			size = requested_size;
+		else {
+			u32 candidate_offset =
+				(last_offset | MLX4_BF_QP_SKIP_MASK | bf_mask) + 1;
+
+			if (last_offset & MLX4_BF_QP_SKIP_MASK)
+				last_offset = candidate_offset;
+
+			/* From this point, the BF bits are 0 */
+
+			if (last_offset > max_table_offset) {
+				/* need to skip */
+				size = -1;
+			} else {
+				size = min3(max_table_offset - last_offset,
+					    bf_mask - (last_offset & bf_mask),
+					    requested_size);
+				if (size < requested_size) {
+					int candidate_size;
+
+					candidate_size = min3(
+						max_table_offset - candidate_offset,
+						bf_mask - (last_offset & bf_mask),
+						requested_size);
+
+					/*  We will not take this path if last_offset was
+					 *  already set above to candidate_offset
+					 */
+					if (candidate_size > size) {
+						last_offset = candidate_offset;
+						size = candidate_size;
+					}
+				}
+			}
+		}
+
+		if (size > 0) {
+			/* mlx4_bitmap_alloc_range will find a contiguous range of "size"
+			 * QPs in which both bits 6 and 7 are zero, because we pass it the
+			 * MLX4_BF_SKIP_MASK).
+			 */
+			offset = mlx4_bitmap_alloc_range(
+					*bitmap + MLX4_QP_TABLE_ZONE_RSS,
+					size, 1,
+					MLX4_BF_QP_SKIP_MASK);
+
+			if (offset == (u32)-1) {
+				err = -ENOMEM;
+				break;
+			}
+
+			last_offset = offset + size;
+
+			err = mlx4_bitmap_init(*bitmap + k, roundup_pow_of_two(size),
+					       roundup_pow_of_two(size) - 1, 0,
+					       roundup_pow_of_two(size) - size);
+		} else {
+			/* Add an empty bitmap, we'll allocate from different zones (since
+			 * at least one is reserved)
+			 */
+			err = mlx4_bitmap_init(*bitmap + k, 1,
+					       MLX4_QP_TABLE_RAW_ETH_SIZE - 1, 0,
+					       0);
+			if (!err)
+				mlx4_bitmap_alloc_range(*bitmap + k, 1, 1, 0);
+		}
+
+		if (err)
+			break;
+
+		++bitmap_initialized;
+
+		err = mlx4_zone_add_one(qp_table->zones, *bitmap + k,
+					MLX4_ZONE_ALLOW_ALLOC_FROM_LOWER_PRIO |
+					MLX4_ZONE_ALLOW_ALLOC_FROM_EQ_PRIO |
+					MLX4_ZONE_USE_RR, MLX4_QP_TABLE_RAW_ETH_PRIORITY,
+					offset, qp_table->zones_uids + k);
+
+		if (err)
+			break;
+	}
+
+	if (err)
+		goto free_bitmap;
+
+	qp_table->bitmap_gen = *bitmap;
+
+	return err;
+
+free_bitmap:
+	for (k = 0; k < bitmap_initialized; k++)
+		mlx4_bitmap_cleanup(*bitmap + k);
+	kfree(bitmap);
+free_zone:
+	mlx4_zone_allocator_destroy(qp_table->zones);
+	return err;
+}
+
+static void mlx4_cleanup_qp_zones(struct mlx4_dev *dev)
+{
+	struct mlx4_qp_table *qp_table = &mlx4_priv(dev)->qp_table;
+
+	if (qp_table->zones) {
+		int i;
+
+		for (i = 0;
+		     i < ARRAY_SIZE(qp_table->zones_uids);
+		     i++) {
+			struct mlx4_bitmap *bitmap =
+				mlx4_zone_get_bitmap(qp_table->zones,
+						     qp_table->zones_uids[i]);
+
+			mlx4_zone_remove_one(qp_table->zones, qp_table->zones_uids[i]);
+			if (NULL == bitmap)
+				continue;
+
+			mlx4_bitmap_cleanup(bitmap);
+		}
+		mlx4_zone_allocator_destroy(qp_table->zones);
+		kfree(qp_table->bitmap_gen);
+		qp_table->bitmap_gen = NULL;
+		qp_table->zones = NULL;
+	}
+}
+
+int mlx4_init_qp_table(struct mlx4_dev *dev)
+{
+	struct mlx4_qp_table *qp_table = &mlx4_priv(dev)->qp_table;
+	int err;
+	int reserved_from_top = 0;
+	int reserved_from_bot;
+	int k;
+	int fixed_reserved_from_bot_rv = 0;
+	int bottom_reserved_for_rss_bitmap;
+	u32 max_table_offset = dev->caps.dmfs_high_rate_qpn_base +
+			dev->caps.dmfs_high_rate_qpn_range;
+
+	spin_lock_init(&qp_table->lock);
+	INIT_RADIX_TREE(&dev->qp_table_tree, GFP_ATOMIC);
+	if (mlx4_is_slave(dev))
+		return 0;
+
+	/* We reserve 2 extra QPs per port for the special QPs.  The
+	 * block of special QPs must be aligned to a multiple of 8, so
+	 * round up.
+	 *
+	 * We also reserve the MSB of the 24-bit QP number to indicate
+	 * that a QP is an XRC QP.
+	 */
+	for (k = 0; k <= MLX4_QP_REGION_BOTTOM; k++)
+		fixed_reserved_from_bot_rv += dev->caps.reserved_qps_cnt[k];
+
+	if (fixed_reserved_from_bot_rv < max_table_offset)
+		fixed_reserved_from_bot_rv = max_table_offset;
+
+	/* We reserve at least 1 extra for bitmaps that we don't have enough space for*/
+	bottom_reserved_for_rss_bitmap =
+		roundup_pow_of_two(fixed_reserved_from_bot_rv + 1);
+	dev->phys_caps.base_sqpn = ALIGN(bottom_reserved_for_rss_bitmap, 8);
+
+	{
+		int sort[MLX4_NUM_QP_REGION];
+		int i, j;
+		int last_base = dev->caps.num_qps;
+
+		for (i = 1; i < MLX4_NUM_QP_REGION; ++i)
+			sort[i] = i;
+
+		for (i = MLX4_NUM_QP_REGION; i > MLX4_QP_REGION_BOTTOM; --i) {
+			for (j = MLX4_QP_REGION_BOTTOM + 2; j < i; ++j) {
+				if (dev->caps.reserved_qps_cnt[sort[j]] >
+				    dev->caps.reserved_qps_cnt[sort[j - 1]])
+					swap(sort[j], sort[j - 1]);
+			}
+		}
+
+		for (i = MLX4_QP_REGION_BOTTOM + 1; i < MLX4_NUM_QP_REGION; ++i) {
+			last_base -= dev->caps.reserved_qps_cnt[sort[i]];
+			dev->caps.reserved_qps_base[sort[i]] = last_base;
+			reserved_from_top +=
+				dev->caps.reserved_qps_cnt[sort[i]];
+		}
+	}
+
+       /* Reserve 8 real SQPs in both native and SRIOV modes.
+	* In addition, in SRIOV mode, reserve 8 proxy SQPs per function
+	* (for all PFs and VFs), and 8 corresponding tunnel QPs.
+	* Each proxy SQP works opposite its own tunnel QP.
+	*
+	* The QPs are arranged as follows:
+	* a. 8 real SQPs
+	* b. All the proxy SQPs (8 per function)
+	* c. All the tunnel QPs (8 per function)
+	*/
+	reserved_from_bot = mlx4_num_reserved_sqps(dev);
+	if (reserved_from_bot + reserved_from_top > dev->caps.num_qps) {
+		mlx4_err(dev, "Number of reserved QPs is higher than number of QPs\n");
+		return -EINVAL;
+	}
+
+	err = mlx4_create_zones(dev, reserved_from_bot, reserved_from_bot,
+				bottom_reserved_for_rss_bitmap,
+				fixed_reserved_from_bot_rv,
+				max_table_offset);
+
+	if (err)
+		return err;
+
+	if (mlx4_is_mfunc(dev)) {
+		/* for PPF use */
+		dev->phys_caps.base_proxy_sqpn = dev->phys_caps.base_sqpn + 8;
+		dev->phys_caps.base_tunnel_sqpn = dev->phys_caps.base_sqpn + 8 + 8 * MLX4_MFUNC_MAX;
+
+		/* In mfunc, calculate proxy and tunnel qp offsets for the PF here,
+		 * since the PF does not call mlx4_slave_caps */
+		dev->caps.spec_qps = kcalloc(dev->caps.num_ports,
+					     sizeof(*dev->caps.spec_qps),
+					     GFP_KERNEL);
+		if (!dev->caps.spec_qps) {
+			err = -ENOMEM;
+			goto err_mem;
+		}
+
+		for (k = 0; k < dev->caps.num_ports; k++) {
+			dev->caps.spec_qps[k].qp0_proxy = dev->phys_caps.base_proxy_sqpn +
+				8 * mlx4_master_func_num(dev) + k;
+			dev->caps.spec_qps[k].qp0_tunnel = dev->caps.spec_qps[k].qp0_proxy + 8 * MLX4_MFUNC_MAX;
+			dev->caps.spec_qps[k].qp1_proxy = dev->phys_caps.base_proxy_sqpn +
+				8 * mlx4_master_func_num(dev) + MLX4_MAX_PORTS + k;
+			dev->caps.spec_qps[k].qp1_tunnel = dev->caps.spec_qps[k].qp1_proxy + 8 * MLX4_MFUNC_MAX;
+		}
+	}
+
+
+	err = mlx4_CONF_SPECIAL_QP(dev, dev->phys_caps.base_sqpn);
+	if (err)
+		goto err_mem;
+
+	return err;
+
+err_mem:
+	kfree(dev->caps.spec_qps);
+	dev->caps.spec_qps = NULL;
+	mlx4_cleanup_qp_zones(dev);
+	return err;
+}
+
+void mlx4_cleanup_qp_table(struct mlx4_dev *dev)
+{
+	if (mlx4_is_slave(dev))
+		return;
+
+	mlx4_CONF_SPECIAL_QP(dev, 0);
+
+	mlx4_cleanup_qp_zones(dev);
+}
+
+int mlx4_qp_query(struct mlx4_dev *dev, struct mlx4_qp *qp,
+		  struct mlx4_qp_context *context)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	int err;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+
+	err = mlx4_cmd_box(dev, 0, mailbox->dma, qp->qpn, 0,
+			   MLX4_CMD_QUERY_QP, MLX4_CMD_TIME_CLASS_A,
+			   MLX4_CMD_WRAPPED);
+	if (!err)
+		memcpy(context, mailbox->buf + 8, sizeof(*context));
+
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx4_qp_query);
+
+int mlx4_qp_to_ready(struct mlx4_dev *dev, struct mlx4_mtt *mtt,
+		     struct mlx4_qp_context *context,
+		     struct mlx4_qp *qp, enum mlx4_qp_state *qp_state)
+{
+	int err;
+	int i;
+	static const enum mlx4_qp_state states[] = {
+		MLX4_QP_STATE_RST,
+		MLX4_QP_STATE_INIT,
+		MLX4_QP_STATE_RTR,
+		MLX4_QP_STATE_RTS
+	};
+
+	for (i = 0; i < ARRAY_SIZE(states) - 1; i++) {
+		context->flags &= cpu_to_be32(~(0xf << 28));
+		context->flags |= cpu_to_be32(states[i + 1] << 28);
+		if (states[i + 1] != MLX4_QP_STATE_RTR)
+			context->params2 &= ~cpu_to_be32(MLX4_QP_BIT_FPP);
+		err = mlx4_qp_modify(dev, mtt, states[i], states[i + 1],
+				     context, 0, 0, qp);
+		if (err) {
+			mlx4_err(dev, "Failed to bring QP to state: %d with error: %d\n",
+				 states[i + 1], err);
+			return err;
+		}
+
+		*qp_state = states[i + 1];
+	}
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(mlx4_qp_to_ready);
+
+u16 mlx4_qp_roce_entropy(struct mlx4_dev *dev, u32 qpn)
+{
+	struct mlx4_qp_context context;
+	struct mlx4_qp qp;
+	int err;
+
+	qp.qpn = qpn;
+	err = mlx4_qp_query(dev, &qp, &context);
+	if (!err) {
+		u32 dest_qpn = be32_to_cpu(context.remote_qpn) & 0xffffff;
+		u16 folded_dst = folded_qp(dest_qpn);
+		u16 folded_src = folded_qp(qpn);
+
+		return (dest_qpn != qpn) ?
+			((folded_dst ^ folded_src) | 0xC000) :
+			folded_src | 0xC000;
+	}
+	return 0xdead;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx4/reset.c b/drivers/net/ethernet/mellanox/mlx4/reset.c
new file mode 100644
index 000000000..0076d8858
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx4/reset.c
@@ -0,0 +1,184 @@
+/*
+ * Copyright (c) 2006, 2007 Cisco Systems, Inc.  All rights reserved.
+ * Copyright (c) 2007, 2008 Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/errno.h>
+#include <linux/pci.h>
+#include <linux/delay.h>
+#include <linux/slab.h>
+#include <linux/jiffies.h>
+
+#include "mlx4.h"
+
+int mlx4_reset(struct mlx4_dev *dev)
+{
+	void __iomem *reset;
+	u32 *hca_header = NULL;
+	int pcie_cap;
+	u16 devctl;
+	u16 linkctl;
+	u16 vendor;
+	unsigned long end;
+	u32 sem;
+	int i;
+	int err = 0;
+
+#define MLX4_RESET_BASE		0xf0000
+#define MLX4_RESET_SIZE		  0x400
+#define MLX4_SEM_OFFSET		  0x3fc
+#define MLX4_RESET_OFFSET	   0x10
+#define MLX4_RESET_VALUE	swab32(1)
+
+#define MLX4_SEM_TIMEOUT_JIFFIES	(10 * HZ)
+#define MLX4_RESET_TIMEOUT_JIFFIES	(2 * HZ)
+
+	/*
+	 * Reset the chip.  This is somewhat ugly because we have to
+	 * save off the PCI header before reset and then restore it
+	 * after the chip reboots.  We skip config space offsets 22
+	 * and 23 since those have a special meaning.
+	 */
+
+	/* Do we need to save off the full 4K PCI Express header?? */
+	hca_header = kmalloc(256, GFP_KERNEL);
+	if (!hca_header) {
+		err = -ENOMEM;
+		mlx4_err(dev, "Couldn't allocate memory to save HCA PCI header, aborting\n");
+		goto out;
+	}
+
+	pcie_cap = pci_pcie_cap(dev->persist->pdev);
+
+	for (i = 0; i < 64; ++i) {
+		if (i == 22 || i == 23)
+			continue;
+		if (pci_read_config_dword(dev->persist->pdev, i * 4,
+					  hca_header + i)) {
+			err = -ENODEV;
+			mlx4_err(dev, "Couldn't save HCA PCI header, aborting\n");
+			goto out;
+		}
+	}
+
+	reset = ioremap(pci_resource_start(dev->persist->pdev, 0) +
+			MLX4_RESET_BASE,
+			MLX4_RESET_SIZE);
+	if (!reset) {
+		err = -ENOMEM;
+		mlx4_err(dev, "Couldn't map HCA reset register, aborting\n");
+		goto out;
+	}
+
+	/* grab HW semaphore to lock out flash updates */
+	end = jiffies + MLX4_SEM_TIMEOUT_JIFFIES;
+	do {
+		sem = readl(reset + MLX4_SEM_OFFSET);
+		if (!sem)
+			break;
+
+		msleep(1);
+	} while (time_before(jiffies, end));
+
+	if (sem) {
+		mlx4_err(dev, "Failed to obtain HW semaphore, aborting\n");
+		err = -EAGAIN;
+		iounmap(reset);
+		goto out;
+	}
+
+	/* actually hit reset */
+	writel(MLX4_RESET_VALUE, reset + MLX4_RESET_OFFSET);
+	iounmap(reset);
+
+	/* Docs say to wait one second before accessing device */
+	msleep(1000);
+
+	end = jiffies + MLX4_RESET_TIMEOUT_JIFFIES;
+	do {
+		if (!pci_read_config_word(dev->persist->pdev, PCI_VENDOR_ID,
+					  &vendor) && vendor != 0xffff)
+			break;
+
+		msleep(1);
+	} while (time_before(jiffies, end));
+
+	if (vendor == 0xffff) {
+		err = -ENODEV;
+		mlx4_err(dev, "PCI device did not come back after reset, aborting\n");
+		goto out;
+	}
+
+	/* Now restore the PCI headers */
+	if (pcie_cap) {
+		devctl = hca_header[(pcie_cap + PCI_EXP_DEVCTL) / 4];
+		if (pcie_capability_write_word(dev->persist->pdev,
+					       PCI_EXP_DEVCTL,
+					       devctl)) {
+			err = -ENODEV;
+			mlx4_err(dev, "Couldn't restore HCA PCI Express Device Control register, aborting\n");
+			goto out;
+		}
+		linkctl = hca_header[(pcie_cap + PCI_EXP_LNKCTL) / 4];
+		if (pcie_capability_write_word(dev->persist->pdev,
+					       PCI_EXP_LNKCTL,
+					       linkctl)) {
+			err = -ENODEV;
+			mlx4_err(dev, "Couldn't restore HCA PCI Express Link control register, aborting\n");
+			goto out;
+		}
+	}
+
+	for (i = 0; i < 16; ++i) {
+		if (i * 4 == PCI_COMMAND)
+			continue;
+
+		if (pci_write_config_dword(dev->persist->pdev, i * 4,
+					   hca_header[i])) {
+			err = -ENODEV;
+			mlx4_err(dev, "Couldn't restore HCA reg %x, aborting\n",
+				 i);
+			goto out;
+		}
+	}
+
+	if (pci_write_config_dword(dev->persist->pdev, PCI_COMMAND,
+				   hca_header[PCI_COMMAND / 4])) {
+		err = -ENODEV;
+		mlx4_err(dev, "Couldn't restore HCA COMMAND, aborting\n");
+		goto out;
+	}
+
+out:
+	kfree(hca_header);
+
+	return err;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c b/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c
new file mode 100644
index 000000000..771b92019
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c
@@ -0,0 +1,5423 @@
+/*
+ * Copyright (c) 2004, 2005 Topspin Communications.  All rights reserved.
+ * Copyright (c) 2005, 2006, 2007, 2008 Mellanox Technologies.
+ * All rights reserved.
+ * Copyright (c) 2005, 2006, 2007 Cisco Systems, Inc.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/sched.h>
+#include <linux/pci.h>
+#include <linux/errno.h>
+#include <linux/kernel.h>
+#include <linux/io.h>
+#include <linux/slab.h>
+#include <linux/mlx4/cmd.h>
+#include <linux/mlx4/qp.h>
+#include <linux/if_ether.h>
+#include <linux/etherdevice.h>
+
+#include "mlx4.h"
+#include "fw.h"
+#include "mlx4_stats.h"
+
+#define MLX4_MAC_VALID		(1ull << 63)
+#define MLX4_PF_COUNTERS_PER_PORT	2
+#define MLX4_VF_COUNTERS_PER_PORT	1
+
+struct mac_res {
+	struct list_head list;
+	u64 mac;
+	int ref_count;
+	u8 smac_index;
+	u8 port;
+};
+
+struct vlan_res {
+	struct list_head list;
+	u16 vlan;
+	int ref_count;
+	int vlan_index;
+	u8 port;
+};
+
+struct res_common {
+	struct list_head	list;
+	struct rb_node		node;
+	u64		        res_id;
+	int			owner;
+	int			state;
+	int			from_state;
+	int			to_state;
+	int			removing;
+	const char		*func_name;
+};
+
+enum {
+	RES_ANY_BUSY = 1
+};
+
+struct res_gid {
+	struct list_head	list;
+	u8			gid[16];
+	enum mlx4_protocol	prot;
+	enum mlx4_steer_type	steer;
+	u64			reg_id;
+};
+
+enum res_qp_states {
+	RES_QP_BUSY = RES_ANY_BUSY,
+
+	/* QP number was allocated */
+	RES_QP_RESERVED,
+
+	/* ICM memory for QP context was mapped */
+	RES_QP_MAPPED,
+
+	/* QP is in hw ownership */
+	RES_QP_HW
+};
+
+struct res_qp {
+	struct res_common	com;
+	struct res_mtt	       *mtt;
+	struct res_cq	       *rcq;
+	struct res_cq	       *scq;
+	struct res_srq	       *srq;
+	struct list_head	mcg_list;
+	spinlock_t		mcg_spl;
+	int			local_qpn;
+	atomic_t		ref_count;
+	u32			qpc_flags;
+	/* saved qp params before VST enforcement in order to restore on VGT */
+	u8			sched_queue;
+	__be32			param3;
+	u8			vlan_control;
+	u8			fvl_rx;
+	u8			pri_path_fl;
+	u8			vlan_index;
+	u8			feup;
+};
+
+enum res_mtt_states {
+	RES_MTT_BUSY = RES_ANY_BUSY,
+	RES_MTT_ALLOCATED,
+};
+
+static inline const char *mtt_states_str(enum res_mtt_states state)
+{
+	switch (state) {
+	case RES_MTT_BUSY: return "RES_MTT_BUSY";
+	case RES_MTT_ALLOCATED: return "RES_MTT_ALLOCATED";
+	default: return "Unknown";
+	}
+}
+
+struct res_mtt {
+	struct res_common	com;
+	int			order;
+	atomic_t		ref_count;
+};
+
+enum res_mpt_states {
+	RES_MPT_BUSY = RES_ANY_BUSY,
+	RES_MPT_RESERVED,
+	RES_MPT_MAPPED,
+	RES_MPT_HW,
+};
+
+struct res_mpt {
+	struct res_common	com;
+	struct res_mtt	       *mtt;
+	int			key;
+};
+
+enum res_eq_states {
+	RES_EQ_BUSY = RES_ANY_BUSY,
+	RES_EQ_RESERVED,
+	RES_EQ_HW,
+};
+
+struct res_eq {
+	struct res_common	com;
+	struct res_mtt	       *mtt;
+};
+
+enum res_cq_states {
+	RES_CQ_BUSY = RES_ANY_BUSY,
+	RES_CQ_ALLOCATED,
+	RES_CQ_HW,
+};
+
+struct res_cq {
+	struct res_common	com;
+	struct res_mtt	       *mtt;
+	atomic_t		ref_count;
+};
+
+enum res_srq_states {
+	RES_SRQ_BUSY = RES_ANY_BUSY,
+	RES_SRQ_ALLOCATED,
+	RES_SRQ_HW,
+};
+
+struct res_srq {
+	struct res_common	com;
+	struct res_mtt	       *mtt;
+	struct res_cq	       *cq;
+	atomic_t		ref_count;
+};
+
+enum res_counter_states {
+	RES_COUNTER_BUSY = RES_ANY_BUSY,
+	RES_COUNTER_ALLOCATED,
+};
+
+struct res_counter {
+	struct res_common	com;
+	int			port;
+};
+
+enum res_xrcdn_states {
+	RES_XRCD_BUSY = RES_ANY_BUSY,
+	RES_XRCD_ALLOCATED,
+};
+
+struct res_xrcdn {
+	struct res_common	com;
+	int			port;
+};
+
+enum res_fs_rule_states {
+	RES_FS_RULE_BUSY = RES_ANY_BUSY,
+	RES_FS_RULE_ALLOCATED,
+};
+
+struct res_fs_rule {
+	struct res_common	com;
+	int			qpn;
+	/* VF DMFS mbox with port flipped */
+	void			*mirr_mbox;
+	/* > 0 --> apply mirror when getting into HA mode      */
+	/* = 0 --> un-apply mirror when getting out of HA mode */
+	u32			mirr_mbox_size;
+	struct list_head	mirr_list;
+	u64			mirr_rule_id;
+};
+
+static void *res_tracker_lookup(struct rb_root *root, u64 res_id)
+{
+	struct rb_node *node = root->rb_node;
+
+	while (node) {
+		struct res_common *res = rb_entry(node, struct res_common,
+						  node);
+
+		if (res_id < res->res_id)
+			node = node->rb_left;
+		else if (res_id > res->res_id)
+			node = node->rb_right;
+		else
+			return res;
+	}
+	return NULL;
+}
+
+static int res_tracker_insert(struct rb_root *root, struct res_common *res)
+{
+	struct rb_node **new = &(root->rb_node), *parent = NULL;
+
+	/* Figure out where to put new node */
+	while (*new) {
+		struct res_common *this = rb_entry(*new, struct res_common,
+						   node);
+
+		parent = *new;
+		if (res->res_id < this->res_id)
+			new = &((*new)->rb_left);
+		else if (res->res_id > this->res_id)
+			new = &((*new)->rb_right);
+		else
+			return -EEXIST;
+	}
+
+	/* Add new node and rebalance tree. */
+	rb_link_node(&res->node, parent, new);
+	rb_insert_color(&res->node, root);
+
+	return 0;
+}
+
+enum qp_transition {
+	QP_TRANS_INIT2RTR,
+	QP_TRANS_RTR2RTS,
+	QP_TRANS_RTS2RTS,
+	QP_TRANS_SQERR2RTS,
+	QP_TRANS_SQD2SQD,
+	QP_TRANS_SQD2RTS
+};
+
+/* For Debug uses */
+static const char *resource_str(enum mlx4_resource rt)
+{
+	switch (rt) {
+	case RES_QP: return "RES_QP";
+	case RES_CQ: return "RES_CQ";
+	case RES_SRQ: return "RES_SRQ";
+	case RES_MPT: return "RES_MPT";
+	case RES_MTT: return "RES_MTT";
+	case RES_MAC: return  "RES_MAC";
+	case RES_VLAN: return  "RES_VLAN";
+	case RES_EQ: return "RES_EQ";
+	case RES_COUNTER: return "RES_COUNTER";
+	case RES_FS_RULE: return "RES_FS_RULE";
+	case RES_XRCD: return "RES_XRCD";
+	default: return "Unknown resource type !!!";
+	}
+}
+
+static void rem_slave_vlans(struct mlx4_dev *dev, int slave);
+static inline int mlx4_grant_resource(struct mlx4_dev *dev, int slave,
+				      enum mlx4_resource res_type, int count,
+				      int port)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct resource_allocator *res_alloc =
+		&priv->mfunc.master.res_tracker.res_alloc[res_type];
+	int err = -EDQUOT;
+	int allocated, free, reserved, guaranteed, from_free;
+	int from_rsvd;
+
+	if (slave > dev->persist->num_vfs)
+		return -EINVAL;
+
+	spin_lock(&res_alloc->alloc_lock);
+	allocated = (port > 0) ?
+		res_alloc->allocated[(port - 1) *
+		(dev->persist->num_vfs + 1) + slave] :
+		res_alloc->allocated[slave];
+	free = (port > 0) ? res_alloc->res_port_free[port - 1] :
+		res_alloc->res_free;
+	reserved = (port > 0) ? res_alloc->res_port_rsvd[port - 1] :
+		res_alloc->res_reserved;
+	guaranteed = res_alloc->guaranteed[slave];
+
+	if (allocated + count > res_alloc->quota[slave]) {
+		mlx4_warn(dev, "VF %d port %d res %s: quota exceeded, count %d alloc %d quota %d\n",
+			  slave, port, resource_str(res_type), count,
+			  allocated, res_alloc->quota[slave]);
+		goto out;
+	}
+
+	if (allocated + count <= guaranteed) {
+		err = 0;
+		from_rsvd = count;
+	} else {
+		/* portion may need to be obtained from free area */
+		if (guaranteed - allocated > 0)
+			from_free = count - (guaranteed - allocated);
+		else
+			from_free = count;
+
+		from_rsvd = count - from_free;
+
+		if (free - from_free >= reserved)
+			err = 0;
+		else
+			mlx4_warn(dev, "VF %d port %d res %s: free pool empty, free %d from_free %d rsvd %d\n",
+				  slave, port, resource_str(res_type), free,
+				  from_free, reserved);
+	}
+
+	if (!err) {
+		/* grant the request */
+		if (port > 0) {
+			res_alloc->allocated[(port - 1) *
+			(dev->persist->num_vfs + 1) + slave] += count;
+			res_alloc->res_port_free[port - 1] -= count;
+			res_alloc->res_port_rsvd[port - 1] -= from_rsvd;
+		} else {
+			res_alloc->allocated[slave] += count;
+			res_alloc->res_free -= count;
+			res_alloc->res_reserved -= from_rsvd;
+		}
+	}
+
+out:
+	spin_unlock(&res_alloc->alloc_lock);
+	return err;
+}
+
+static inline void mlx4_release_resource(struct mlx4_dev *dev, int slave,
+				    enum mlx4_resource res_type, int count,
+				    int port)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct resource_allocator *res_alloc =
+		&priv->mfunc.master.res_tracker.res_alloc[res_type];
+	int allocated, guaranteed, from_rsvd;
+
+	if (slave > dev->persist->num_vfs)
+		return;
+
+	spin_lock(&res_alloc->alloc_lock);
+
+	allocated = (port > 0) ?
+		res_alloc->allocated[(port - 1) *
+		(dev->persist->num_vfs + 1) + slave] :
+		res_alloc->allocated[slave];
+	guaranteed = res_alloc->guaranteed[slave];
+
+	if (allocated - count >= guaranteed) {
+		from_rsvd = 0;
+	} else {
+		/* portion may need to be returned to reserved area */
+		if (allocated - guaranteed > 0)
+			from_rsvd = count - (allocated - guaranteed);
+		else
+			from_rsvd = count;
+	}
+
+	if (port > 0) {
+		res_alloc->allocated[(port - 1) *
+		(dev->persist->num_vfs + 1) + slave] -= count;
+		res_alloc->res_port_free[port - 1] += count;
+		res_alloc->res_port_rsvd[port - 1] += from_rsvd;
+	} else {
+		res_alloc->allocated[slave] -= count;
+		res_alloc->res_free += count;
+		res_alloc->res_reserved += from_rsvd;
+	}
+
+	spin_unlock(&res_alloc->alloc_lock);
+	return;
+}
+
+static inline void initialize_res_quotas(struct mlx4_dev *dev,
+					 struct resource_allocator *res_alloc,
+					 enum mlx4_resource res_type,
+					 int vf, int num_instances)
+{
+	res_alloc->guaranteed[vf] = num_instances /
+				    (2 * (dev->persist->num_vfs + 1));
+	res_alloc->quota[vf] = (num_instances / 2) + res_alloc->guaranteed[vf];
+	if (vf == mlx4_master_func_num(dev)) {
+		res_alloc->res_free = num_instances;
+		if (res_type == RES_MTT) {
+			/* reserved mtts will be taken out of the PF allocation */
+			res_alloc->res_free += dev->caps.reserved_mtts;
+			res_alloc->guaranteed[vf] += dev->caps.reserved_mtts;
+			res_alloc->quota[vf] += dev->caps.reserved_mtts;
+		}
+	}
+}
+
+void mlx4_init_quotas(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int pf;
+
+	/* quotas for VFs are initialized in mlx4_slave_cap */
+	if (mlx4_is_slave(dev))
+		return;
+
+	if (!mlx4_is_mfunc(dev)) {
+		dev->quotas.qp = dev->caps.num_qps - dev->caps.reserved_qps -
+			mlx4_num_reserved_sqps(dev);
+		dev->quotas.cq = dev->caps.num_cqs - dev->caps.reserved_cqs;
+		dev->quotas.srq = dev->caps.num_srqs - dev->caps.reserved_srqs;
+		dev->quotas.mtt = dev->caps.num_mtts - dev->caps.reserved_mtts;
+		dev->quotas.mpt = dev->caps.num_mpts - dev->caps.reserved_mrws;
+		return;
+	}
+
+	pf = mlx4_master_func_num(dev);
+	dev->quotas.qp =
+		priv->mfunc.master.res_tracker.res_alloc[RES_QP].quota[pf];
+	dev->quotas.cq =
+		priv->mfunc.master.res_tracker.res_alloc[RES_CQ].quota[pf];
+	dev->quotas.srq =
+		priv->mfunc.master.res_tracker.res_alloc[RES_SRQ].quota[pf];
+	dev->quotas.mtt =
+		priv->mfunc.master.res_tracker.res_alloc[RES_MTT].quota[pf];
+	dev->quotas.mpt =
+		priv->mfunc.master.res_tracker.res_alloc[RES_MPT].quota[pf];
+}
+
+static int
+mlx4_calc_res_counter_guaranteed(struct mlx4_dev *dev,
+				 struct resource_allocator *res_alloc,
+				 int vf)
+{
+	struct mlx4_active_ports actv_ports;
+	int ports, counters_guaranteed;
+
+	/* For master, only allocate according to the number of phys ports */
+	if (vf == mlx4_master_func_num(dev))
+		return MLX4_PF_COUNTERS_PER_PORT * dev->caps.num_ports;
+
+	/* calculate real number of ports for the VF */
+	actv_ports = mlx4_get_active_ports(dev, vf);
+	ports = bitmap_weight(actv_ports.ports, dev->caps.num_ports);
+	counters_guaranteed = ports * MLX4_VF_COUNTERS_PER_PORT;
+
+	/* If we do not have enough counters for this VF, do not
+	 * allocate any for it. '-1' to reduce the sink counter.
+	 */
+	if ((res_alloc->res_reserved + counters_guaranteed) >
+	    (dev->caps.max_counters - 1))
+		return 0;
+
+	return counters_guaranteed;
+}
+
+int mlx4_init_resource_tracker(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int i, j;
+	int t;
+
+	priv->mfunc.master.res_tracker.slave_list =
+		kcalloc(dev->num_slaves, sizeof(struct slave_list),
+			GFP_KERNEL);
+	if (!priv->mfunc.master.res_tracker.slave_list)
+		return -ENOMEM;
+
+	for (i = 0 ; i < dev->num_slaves; i++) {
+		for (t = 0; t < MLX4_NUM_OF_RESOURCE_TYPE; ++t)
+			INIT_LIST_HEAD(&priv->mfunc.master.res_tracker.
+				       slave_list[i].res_list[t]);
+		mutex_init(&priv->mfunc.master.res_tracker.slave_list[i].mutex);
+	}
+
+	mlx4_dbg(dev, "Started init_resource_tracker: %ld slaves\n",
+		 dev->num_slaves);
+	for (i = 0 ; i < MLX4_NUM_OF_RESOURCE_TYPE; i++)
+		priv->mfunc.master.res_tracker.res_tree[i] = RB_ROOT;
+
+	for (i = 0; i < MLX4_NUM_OF_RESOURCE_TYPE; i++) {
+		struct resource_allocator *res_alloc =
+			&priv->mfunc.master.res_tracker.res_alloc[i];
+		res_alloc->quota = kmalloc_array(dev->persist->num_vfs + 1,
+						 sizeof(int),
+						 GFP_KERNEL);
+		res_alloc->guaranteed = kmalloc_array(dev->persist->num_vfs + 1,
+						      sizeof(int),
+						      GFP_KERNEL);
+		if (i == RES_MAC || i == RES_VLAN)
+			res_alloc->allocated =
+				kcalloc(MLX4_MAX_PORTS *
+						(dev->persist->num_vfs + 1),
+					sizeof(int), GFP_KERNEL);
+		else
+			res_alloc->allocated =
+				kcalloc(dev->persist->num_vfs + 1,
+					sizeof(int), GFP_KERNEL);
+		/* Reduce the sink counter */
+		if (i == RES_COUNTER)
+			res_alloc->res_free = dev->caps.max_counters - 1;
+
+		if (!res_alloc->quota || !res_alloc->guaranteed ||
+		    !res_alloc->allocated)
+			goto no_mem_err;
+
+		spin_lock_init(&res_alloc->alloc_lock);
+		for (t = 0; t < dev->persist->num_vfs + 1; t++) {
+			struct mlx4_active_ports actv_ports =
+				mlx4_get_active_ports(dev, t);
+			switch (i) {
+			case RES_QP:
+				initialize_res_quotas(dev, res_alloc, RES_QP,
+						      t, dev->caps.num_qps -
+						      dev->caps.reserved_qps -
+						      mlx4_num_reserved_sqps(dev));
+				break;
+			case RES_CQ:
+				initialize_res_quotas(dev, res_alloc, RES_CQ,
+						      t, dev->caps.num_cqs -
+						      dev->caps.reserved_cqs);
+				break;
+			case RES_SRQ:
+				initialize_res_quotas(dev, res_alloc, RES_SRQ,
+						      t, dev->caps.num_srqs -
+						      dev->caps.reserved_srqs);
+				break;
+			case RES_MPT:
+				initialize_res_quotas(dev, res_alloc, RES_MPT,
+						      t, dev->caps.num_mpts -
+						      dev->caps.reserved_mrws);
+				break;
+			case RES_MTT:
+				initialize_res_quotas(dev, res_alloc, RES_MTT,
+						      t, dev->caps.num_mtts -
+						      dev->caps.reserved_mtts);
+				break;
+			case RES_MAC:
+				if (t == mlx4_master_func_num(dev)) {
+					int max_vfs_pport = 0;
+					/* Calculate the max vfs per port for */
+					/* both ports.			      */
+					for (j = 0; j < dev->caps.num_ports;
+					     j++) {
+						struct mlx4_slaves_pport slaves_pport =
+							mlx4_phys_to_slaves_pport(dev, j + 1);
+						unsigned current_slaves =
+							bitmap_weight(slaves_pport.slaves,
+								      dev->caps.num_ports) - 1;
+						if (max_vfs_pport < current_slaves)
+							max_vfs_pport =
+								current_slaves;
+					}
+					res_alloc->quota[t] =
+						MLX4_MAX_MAC_NUM -
+						2 * max_vfs_pport;
+					res_alloc->guaranteed[t] = 2;
+					for (j = 0; j < MLX4_MAX_PORTS; j++)
+						res_alloc->res_port_free[j] =
+							MLX4_MAX_MAC_NUM;
+				} else {
+					res_alloc->quota[t] = MLX4_MAX_MAC_NUM;
+					res_alloc->guaranteed[t] = 2;
+				}
+				break;
+			case RES_VLAN:
+				if (t == mlx4_master_func_num(dev)) {
+					res_alloc->quota[t] = MLX4_MAX_VLAN_NUM;
+					res_alloc->guaranteed[t] = MLX4_MAX_VLAN_NUM / 2;
+					for (j = 0; j < MLX4_MAX_PORTS; j++)
+						res_alloc->res_port_free[j] =
+							res_alloc->quota[t];
+				} else {
+					res_alloc->quota[t] = MLX4_MAX_VLAN_NUM / 2;
+					res_alloc->guaranteed[t] = 0;
+				}
+				break;
+			case RES_COUNTER:
+				res_alloc->quota[t] = dev->caps.max_counters;
+				res_alloc->guaranteed[t] =
+					mlx4_calc_res_counter_guaranteed(dev, res_alloc, t);
+				break;
+			default:
+				break;
+			}
+			if (i == RES_MAC || i == RES_VLAN) {
+				for (j = 0; j < dev->caps.num_ports; j++)
+					if (test_bit(j, actv_ports.ports))
+						res_alloc->res_port_rsvd[j] +=
+							res_alloc->guaranteed[t];
+			} else {
+				res_alloc->res_reserved += res_alloc->guaranteed[t];
+			}
+		}
+	}
+	spin_lock_init(&priv->mfunc.master.res_tracker.lock);
+	return 0;
+
+no_mem_err:
+	for (i = 0; i < MLX4_NUM_OF_RESOURCE_TYPE; i++) {
+		kfree(priv->mfunc.master.res_tracker.res_alloc[i].allocated);
+		priv->mfunc.master.res_tracker.res_alloc[i].allocated = NULL;
+		kfree(priv->mfunc.master.res_tracker.res_alloc[i].guaranteed);
+		priv->mfunc.master.res_tracker.res_alloc[i].guaranteed = NULL;
+		kfree(priv->mfunc.master.res_tracker.res_alloc[i].quota);
+		priv->mfunc.master.res_tracker.res_alloc[i].quota = NULL;
+	}
+	return -ENOMEM;
+}
+
+void mlx4_free_resource_tracker(struct mlx4_dev *dev,
+				enum mlx4_res_tracker_free_type type)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int i;
+
+	if (priv->mfunc.master.res_tracker.slave_list) {
+		if (type != RES_TR_FREE_STRUCTS_ONLY) {
+			for (i = 0; i < dev->num_slaves; i++) {
+				if (type == RES_TR_FREE_ALL ||
+				    dev->caps.function != i)
+					mlx4_delete_all_resources_for_slave(dev, i);
+			}
+			/* free master's vlans */
+			i = dev->caps.function;
+			mlx4_reset_roce_gids(dev, i);
+			mutex_lock(&priv->mfunc.master.res_tracker.slave_list[i].mutex);
+			rem_slave_vlans(dev, i);
+			mutex_unlock(&priv->mfunc.master.res_tracker.slave_list[i].mutex);
+		}
+
+		if (type != RES_TR_FREE_SLAVES_ONLY) {
+			for (i = 0; i < MLX4_NUM_OF_RESOURCE_TYPE; i++) {
+				kfree(priv->mfunc.master.res_tracker.res_alloc[i].allocated);
+				priv->mfunc.master.res_tracker.res_alloc[i].allocated = NULL;
+				kfree(priv->mfunc.master.res_tracker.res_alloc[i].guaranteed);
+				priv->mfunc.master.res_tracker.res_alloc[i].guaranteed = NULL;
+				kfree(priv->mfunc.master.res_tracker.res_alloc[i].quota);
+				priv->mfunc.master.res_tracker.res_alloc[i].quota = NULL;
+			}
+			kfree(priv->mfunc.master.res_tracker.slave_list);
+			priv->mfunc.master.res_tracker.slave_list = NULL;
+		}
+	}
+}
+
+static void update_pkey_index(struct mlx4_dev *dev, int slave,
+			      struct mlx4_cmd_mailbox *inbox)
+{
+	u8 sched = *(u8 *)(inbox->buf + 64);
+	u8 orig_index = *(u8 *)(inbox->buf + 35);
+	u8 new_index;
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int port;
+
+	port = (sched >> 6 & 1) + 1;
+
+	new_index = priv->virt2phys_pkey[slave][port - 1][orig_index];
+	*(u8 *)(inbox->buf + 35) = new_index;
+}
+
+static void update_gid(struct mlx4_dev *dev, struct mlx4_cmd_mailbox *inbox,
+		       u8 slave)
+{
+	struct mlx4_qp_context	*qp_ctx = inbox->buf + 8;
+	enum mlx4_qp_optpar	optpar = be32_to_cpu(*(__be32 *) inbox->buf);
+	u32			ts = (be32_to_cpu(qp_ctx->flags) >> 16) & 0xff;
+	int port;
+
+	if (MLX4_QP_ST_UD == ts) {
+		port = (qp_ctx->pri_path.sched_queue >> 6 & 1) + 1;
+		if (mlx4_is_eth(dev, port))
+			qp_ctx->pri_path.mgid_index =
+				mlx4_get_base_gid_ix(dev, slave, port) | 0x80;
+		else
+			qp_ctx->pri_path.mgid_index = slave | 0x80;
+
+	} else if (MLX4_QP_ST_RC == ts || MLX4_QP_ST_XRC == ts || MLX4_QP_ST_UC == ts) {
+		if (optpar & MLX4_QP_OPTPAR_PRIMARY_ADDR_PATH) {
+			port = (qp_ctx->pri_path.sched_queue >> 6 & 1) + 1;
+			if (mlx4_is_eth(dev, port)) {
+				qp_ctx->pri_path.mgid_index +=
+					mlx4_get_base_gid_ix(dev, slave, port);
+				qp_ctx->pri_path.mgid_index &= 0x7f;
+			} else {
+				qp_ctx->pri_path.mgid_index = slave & 0x7F;
+			}
+		}
+		if (optpar & MLX4_QP_OPTPAR_ALT_ADDR_PATH) {
+			port = (qp_ctx->alt_path.sched_queue >> 6 & 1) + 1;
+			if (mlx4_is_eth(dev, port)) {
+				qp_ctx->alt_path.mgid_index +=
+					mlx4_get_base_gid_ix(dev, slave, port);
+				qp_ctx->alt_path.mgid_index &= 0x7f;
+			} else {
+				qp_ctx->alt_path.mgid_index = slave & 0x7F;
+			}
+		}
+	}
+}
+
+static int handle_counter(struct mlx4_dev *dev, struct mlx4_qp_context *qpc,
+			  u8 slave, int port);
+
+static int update_vport_qp_param(struct mlx4_dev *dev,
+				 struct mlx4_cmd_mailbox *inbox,
+				 u8 slave, u32 qpn)
+{
+	struct mlx4_qp_context	*qpc = inbox->buf + 8;
+	struct mlx4_vport_oper_state *vp_oper;
+	struct mlx4_priv *priv;
+	u32 qp_type;
+	int port, err = 0;
+
+	port = (qpc->pri_path.sched_queue & 0x40) ? 2 : 1;
+	priv = mlx4_priv(dev);
+	vp_oper = &priv->mfunc.master.vf_oper[slave].vport[port];
+	qp_type	= (be32_to_cpu(qpc->flags) >> 16) & 0xff;
+
+	err = handle_counter(dev, qpc, slave, port);
+	if (err)
+		goto out;
+
+	if (MLX4_VGT != vp_oper->state.default_vlan) {
+		/* the reserved QPs (special, proxy, tunnel)
+		 * do not operate over vlans
+		 */
+		if (mlx4_is_qp_reserved(dev, qpn))
+			return 0;
+
+		/* force strip vlan by clear vsd, MLX QP refers to Raw Ethernet */
+		if (qp_type == MLX4_QP_ST_UD ||
+		    (qp_type == MLX4_QP_ST_MLX && mlx4_is_eth(dev, port))) {
+			if (dev->caps.bmme_flags & MLX4_BMME_FLAG_VSD_INIT2RTR) {
+				*(__be32 *)inbox->buf =
+					cpu_to_be32(be32_to_cpu(*(__be32 *)inbox->buf) |
+					MLX4_QP_OPTPAR_VLAN_STRIPPING);
+				qpc->param3 &= ~cpu_to_be32(MLX4_STRIP_VLAN);
+			} else {
+				struct mlx4_update_qp_params params = {.flags = 0};
+
+				err = mlx4_update_qp(dev, qpn, MLX4_UPDATE_QP_VSD, &params);
+				if (err)
+					goto out;
+			}
+		}
+
+		/* preserve IF_COUNTER flag */
+		qpc->pri_path.vlan_control &=
+			MLX4_CTRL_ETH_SRC_CHECK_IF_COUNTER;
+		if (vp_oper->state.link_state == IFLA_VF_LINK_STATE_DISABLE &&
+		    dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_UPDATE_QP) {
+			qpc->pri_path.vlan_control |=
+				MLX4_VLAN_CTRL_ETH_TX_BLOCK_TAGGED |
+				MLX4_VLAN_CTRL_ETH_TX_BLOCK_PRIO_TAGGED |
+				MLX4_VLAN_CTRL_ETH_TX_BLOCK_UNTAGGED |
+				MLX4_VLAN_CTRL_ETH_RX_BLOCK_PRIO_TAGGED |
+				MLX4_VLAN_CTRL_ETH_RX_BLOCK_UNTAGGED |
+				MLX4_VLAN_CTRL_ETH_RX_BLOCK_TAGGED;
+		} else if (0 != vp_oper->state.default_vlan) {
+			if (vp_oper->state.vlan_proto == htons(ETH_P_8021AD)) {
+				/* vst QinQ should block untagged on TX,
+				 * but cvlan is in payload and phv is set so
+				 * hw see it as untagged. Block tagged instead.
+				 */
+				qpc->pri_path.vlan_control |=
+					MLX4_VLAN_CTRL_ETH_TX_BLOCK_PRIO_TAGGED |
+					MLX4_VLAN_CTRL_ETH_TX_BLOCK_TAGGED |
+					MLX4_VLAN_CTRL_ETH_RX_BLOCK_PRIO_TAGGED |
+					MLX4_VLAN_CTRL_ETH_RX_BLOCK_UNTAGGED;
+			} else { /* vst 802.1Q */
+				qpc->pri_path.vlan_control |=
+					MLX4_VLAN_CTRL_ETH_TX_BLOCK_TAGGED |
+					MLX4_VLAN_CTRL_ETH_RX_BLOCK_PRIO_TAGGED |
+					MLX4_VLAN_CTRL_ETH_RX_BLOCK_UNTAGGED;
+			}
+		} else { /* priority tagged */
+			qpc->pri_path.vlan_control |=
+				MLX4_VLAN_CTRL_ETH_TX_BLOCK_TAGGED |
+				MLX4_VLAN_CTRL_ETH_RX_BLOCK_TAGGED;
+		}
+
+		qpc->pri_path.fvl_rx |= MLX4_FVL_RX_FORCE_ETH_VLAN;
+		qpc->pri_path.vlan_index = vp_oper->vlan_idx;
+		qpc->pri_path.fl |= MLX4_FL_ETH_HIDE_CQE_VLAN;
+		if (vp_oper->state.vlan_proto == htons(ETH_P_8021AD))
+			qpc->pri_path.fl |= MLX4_FL_SV;
+		else
+			qpc->pri_path.fl |= MLX4_FL_CV;
+		qpc->pri_path.feup |= MLX4_FEUP_FORCE_ETH_UP | MLX4_FVL_FORCE_ETH_VLAN;
+		qpc->pri_path.sched_queue &= 0xC7;
+		qpc->pri_path.sched_queue |= (vp_oper->state.default_qos) << 3;
+		qpc->qos_vport = vp_oper->state.qos_vport;
+	}
+	if (vp_oper->state.spoofchk) {
+		qpc->pri_path.feup |= MLX4_FSM_FORCE_ETH_SRC_MAC;
+		qpc->pri_path.grh_mylmc = (0x80 & qpc->pri_path.grh_mylmc) + vp_oper->mac_idx;
+	}
+out:
+	return err;
+}
+
+static int mpt_mask(struct mlx4_dev *dev)
+{
+	return dev->caps.num_mpts - 1;
+}
+
+static const char *mlx4_resource_type_to_str(enum mlx4_resource t)
+{
+	switch (t) {
+	case RES_QP:
+		return "QP";
+	case RES_CQ:
+		return "CQ";
+	case RES_SRQ:
+		return "SRQ";
+	case RES_XRCD:
+		return "XRCD";
+	case RES_MPT:
+		return "MPT";
+	case RES_MTT:
+		return "MTT";
+	case RES_MAC:
+		return "MAC";
+	case RES_VLAN:
+		return "VLAN";
+	case RES_COUNTER:
+		return "COUNTER";
+	case RES_FS_RULE:
+		return "FS_RULE";
+	case RES_EQ:
+		return "EQ";
+	default:
+		return "INVALID RESOURCE";
+	}
+}
+
+static void *find_res(struct mlx4_dev *dev, u64 res_id,
+		      enum mlx4_resource type)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	return res_tracker_lookup(&priv->mfunc.master.res_tracker.res_tree[type],
+				  res_id);
+}
+
+static int _get_res(struct mlx4_dev *dev, int slave, u64 res_id,
+		    enum mlx4_resource type,
+		    void *res, const char *func_name)
+{
+	struct res_common *r;
+	int err = 0;
+
+	spin_lock_irq(mlx4_tlock(dev));
+	r = find_res(dev, res_id, type);
+	if (!r) {
+		err = -ENONET;
+		goto exit;
+	}
+
+	if (r->state == RES_ANY_BUSY) {
+		mlx4_warn(dev,
+			  "%s(%d) trying to get resource %llx of type %s, but it's already taken by %s\n",
+			  func_name, slave, res_id, mlx4_resource_type_to_str(type),
+			  r->func_name);
+		err = -EBUSY;
+		goto exit;
+	}
+
+	if (r->owner != slave) {
+		err = -EPERM;
+		goto exit;
+	}
+
+	r->from_state = r->state;
+	r->state = RES_ANY_BUSY;
+	r->func_name = func_name;
+
+	if (res)
+		*((struct res_common **)res) = r;
+
+exit:
+	spin_unlock_irq(mlx4_tlock(dev));
+	return err;
+}
+
+#define get_res(dev, slave, res_id, type, res) \
+	_get_res((dev), (slave), (res_id), (type), (res), __func__)
+
+int mlx4_get_slave_from_resource_id(struct mlx4_dev *dev,
+				    enum mlx4_resource type,
+				    u64 res_id, int *slave)
+{
+
+	struct res_common *r;
+	int err = -ENOENT;
+	int id = res_id;
+
+	if (type == RES_QP)
+		id &= 0x7fffff;
+	spin_lock(mlx4_tlock(dev));
+
+	r = find_res(dev, id, type);
+	if (r) {
+		*slave = r->owner;
+		err = 0;
+	}
+	spin_unlock(mlx4_tlock(dev));
+
+	return err;
+}
+
+static void put_res(struct mlx4_dev *dev, int slave, u64 res_id,
+		    enum mlx4_resource type)
+{
+	struct res_common *r;
+
+	spin_lock_irq(mlx4_tlock(dev));
+	r = find_res(dev, res_id, type);
+	if (r) {
+		r->state = r->from_state;
+		r->func_name = "";
+	}
+	spin_unlock_irq(mlx4_tlock(dev));
+}
+
+static int counter_alloc_res(struct mlx4_dev *dev, int slave, int op, int cmd,
+			     u64 in_param, u64 *out_param, int port);
+
+static int handle_existing_counter(struct mlx4_dev *dev, u8 slave, int port,
+				   int counter_index)
+{
+	struct res_common *r;
+	struct res_counter *counter;
+	int ret = 0;
+
+	if (counter_index == MLX4_SINK_COUNTER_INDEX(dev))
+		return ret;
+
+	spin_lock_irq(mlx4_tlock(dev));
+	r = find_res(dev, counter_index, RES_COUNTER);
+	if (!r || r->owner != slave) {
+		ret = -EINVAL;
+	} else {
+		counter = container_of(r, struct res_counter, com);
+		if (!counter->port)
+			counter->port = port;
+	}
+
+	spin_unlock_irq(mlx4_tlock(dev));
+	return ret;
+}
+
+static int handle_unexisting_counter(struct mlx4_dev *dev,
+				     struct mlx4_qp_context *qpc, u8 slave,
+				     int port)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_resource_tracker *tracker = &priv->mfunc.master.res_tracker;
+	struct res_common *tmp;
+	struct res_counter *counter;
+	u64 counter_idx = MLX4_SINK_COUNTER_INDEX(dev);
+	int err = 0;
+
+	spin_lock_irq(mlx4_tlock(dev));
+	list_for_each_entry(tmp,
+			    &tracker->slave_list[slave].res_list[RES_COUNTER],
+			    list) {
+		counter = container_of(tmp, struct res_counter, com);
+		if (port == counter->port) {
+			qpc->pri_path.counter_index  = counter->com.res_id;
+			spin_unlock_irq(mlx4_tlock(dev));
+			return 0;
+		}
+	}
+	spin_unlock_irq(mlx4_tlock(dev));
+
+	/* No existing counter, need to allocate a new counter */
+	err = counter_alloc_res(dev, slave, RES_OP_RESERVE, 0, 0, &counter_idx,
+				port);
+	if (err == -ENOENT) {
+		err = 0;
+	} else if (err && err != -ENOSPC) {
+		mlx4_err(dev, "%s: failed to create new counter for slave %d err %d\n",
+			 __func__, slave, err);
+	} else {
+		qpc->pri_path.counter_index = counter_idx;
+		mlx4_dbg(dev, "%s: alloc new counter for slave %d index %d\n",
+			 __func__, slave, qpc->pri_path.counter_index);
+		err = 0;
+	}
+
+	return err;
+}
+
+static int handle_counter(struct mlx4_dev *dev, struct mlx4_qp_context *qpc,
+			  u8 slave, int port)
+{
+	if (qpc->pri_path.counter_index != MLX4_SINK_COUNTER_INDEX(dev))
+		return handle_existing_counter(dev, slave, port,
+					       qpc->pri_path.counter_index);
+
+	return handle_unexisting_counter(dev, qpc, slave, port);
+}
+
+static struct res_common *alloc_qp_tr(int id)
+{
+	struct res_qp *ret;
+
+	ret = kzalloc(sizeof(*ret), GFP_KERNEL);
+	if (!ret)
+		return NULL;
+
+	ret->com.res_id = id;
+	ret->com.state = RES_QP_RESERVED;
+	ret->local_qpn = id;
+	INIT_LIST_HEAD(&ret->mcg_list);
+	spin_lock_init(&ret->mcg_spl);
+	atomic_set(&ret->ref_count, 0);
+
+	return &ret->com;
+}
+
+static struct res_common *alloc_mtt_tr(int id, int order)
+{
+	struct res_mtt *ret;
+
+	ret = kzalloc(sizeof(*ret), GFP_KERNEL);
+	if (!ret)
+		return NULL;
+
+	ret->com.res_id = id;
+	ret->order = order;
+	ret->com.state = RES_MTT_ALLOCATED;
+	atomic_set(&ret->ref_count, 0);
+
+	return &ret->com;
+}
+
+static struct res_common *alloc_mpt_tr(int id, int key)
+{
+	struct res_mpt *ret;
+
+	ret = kzalloc(sizeof(*ret), GFP_KERNEL);
+	if (!ret)
+		return NULL;
+
+	ret->com.res_id = id;
+	ret->com.state = RES_MPT_RESERVED;
+	ret->key = key;
+
+	return &ret->com;
+}
+
+static struct res_common *alloc_eq_tr(int id)
+{
+	struct res_eq *ret;
+
+	ret = kzalloc(sizeof(*ret), GFP_KERNEL);
+	if (!ret)
+		return NULL;
+
+	ret->com.res_id = id;
+	ret->com.state = RES_EQ_RESERVED;
+
+	return &ret->com;
+}
+
+static struct res_common *alloc_cq_tr(int id)
+{
+	struct res_cq *ret;
+
+	ret = kzalloc(sizeof(*ret), GFP_KERNEL);
+	if (!ret)
+		return NULL;
+
+	ret->com.res_id = id;
+	ret->com.state = RES_CQ_ALLOCATED;
+	atomic_set(&ret->ref_count, 0);
+
+	return &ret->com;
+}
+
+static struct res_common *alloc_srq_tr(int id)
+{
+	struct res_srq *ret;
+
+	ret = kzalloc(sizeof(*ret), GFP_KERNEL);
+	if (!ret)
+		return NULL;
+
+	ret->com.res_id = id;
+	ret->com.state = RES_SRQ_ALLOCATED;
+	atomic_set(&ret->ref_count, 0);
+
+	return &ret->com;
+}
+
+static struct res_common *alloc_counter_tr(int id, int port)
+{
+	struct res_counter *ret;
+
+	ret = kzalloc(sizeof(*ret), GFP_KERNEL);
+	if (!ret)
+		return NULL;
+
+	ret->com.res_id = id;
+	ret->com.state = RES_COUNTER_ALLOCATED;
+	ret->port = port;
+
+	return &ret->com;
+}
+
+static struct res_common *alloc_xrcdn_tr(int id)
+{
+	struct res_xrcdn *ret;
+
+	ret = kzalloc(sizeof(*ret), GFP_KERNEL);
+	if (!ret)
+		return NULL;
+
+	ret->com.res_id = id;
+	ret->com.state = RES_XRCD_ALLOCATED;
+
+	return &ret->com;
+}
+
+static struct res_common *alloc_fs_rule_tr(u64 id, int qpn)
+{
+	struct res_fs_rule *ret;
+
+	ret = kzalloc(sizeof(*ret), GFP_KERNEL);
+	if (!ret)
+		return NULL;
+
+	ret->com.res_id = id;
+	ret->com.state = RES_FS_RULE_ALLOCATED;
+	ret->qpn = qpn;
+	return &ret->com;
+}
+
+static struct res_common *alloc_tr(u64 id, enum mlx4_resource type, int slave,
+				   int extra)
+{
+	struct res_common *ret;
+
+	switch (type) {
+	case RES_QP:
+		ret = alloc_qp_tr(id);
+		break;
+	case RES_MPT:
+		ret = alloc_mpt_tr(id, extra);
+		break;
+	case RES_MTT:
+		ret = alloc_mtt_tr(id, extra);
+		break;
+	case RES_EQ:
+		ret = alloc_eq_tr(id);
+		break;
+	case RES_CQ:
+		ret = alloc_cq_tr(id);
+		break;
+	case RES_SRQ:
+		ret = alloc_srq_tr(id);
+		break;
+	case RES_MAC:
+		pr_err("implementation missing\n");
+		return NULL;
+	case RES_COUNTER:
+		ret = alloc_counter_tr(id, extra);
+		break;
+	case RES_XRCD:
+		ret = alloc_xrcdn_tr(id);
+		break;
+	case RES_FS_RULE:
+		ret = alloc_fs_rule_tr(id, extra);
+		break;
+	default:
+		return NULL;
+	}
+	if (ret)
+		ret->owner = slave;
+
+	return ret;
+}
+
+int mlx4_calc_vf_counters(struct mlx4_dev *dev, int slave, int port,
+			  struct mlx4_counter *data)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_resource_tracker *tracker = &priv->mfunc.master.res_tracker;
+	struct res_common *tmp;
+	struct res_counter *counter;
+	int *counters_arr;
+	int i = 0, err = 0;
+
+	memset(data, 0, sizeof(*data));
+
+	counters_arr = kmalloc_array(dev->caps.max_counters,
+				     sizeof(*counters_arr), GFP_KERNEL);
+	if (!counters_arr)
+		return -ENOMEM;
+
+	spin_lock_irq(mlx4_tlock(dev));
+	list_for_each_entry(tmp,
+			    &tracker->slave_list[slave].res_list[RES_COUNTER],
+			    list) {
+		counter = container_of(tmp, struct res_counter, com);
+		if (counter->port == port) {
+			counters_arr[i] = (int)tmp->res_id;
+			i++;
+		}
+	}
+	spin_unlock_irq(mlx4_tlock(dev));
+	counters_arr[i] = -1;
+
+	i = 0;
+
+	while (counters_arr[i] != -1) {
+		err = mlx4_get_counter_stats(dev, counters_arr[i], data,
+					     0);
+		if (err) {
+			memset(data, 0, sizeof(*data));
+			goto table_changed;
+		}
+		i++;
+	}
+
+table_changed:
+	kfree(counters_arr);
+	return 0;
+}
+
+static int add_res_range(struct mlx4_dev *dev, int slave, u64 base, int count,
+			 enum mlx4_resource type, int extra)
+{
+	int i;
+	int err;
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct res_common **res_arr;
+	struct mlx4_resource_tracker *tracker = &priv->mfunc.master.res_tracker;
+	struct rb_root *root = &tracker->res_tree[type];
+
+	res_arr = kcalloc(count, sizeof(*res_arr), GFP_KERNEL);
+	if (!res_arr)
+		return -ENOMEM;
+
+	for (i = 0; i < count; ++i) {
+		res_arr[i] = alloc_tr(base + i, type, slave, extra);
+		if (!res_arr[i]) {
+			for (--i; i >= 0; --i)
+				kfree(res_arr[i]);
+
+			kfree(res_arr);
+			return -ENOMEM;
+		}
+	}
+
+	spin_lock_irq(mlx4_tlock(dev));
+	for (i = 0; i < count; ++i) {
+		if (find_res(dev, base + i, type)) {
+			err = -EEXIST;
+			goto undo;
+		}
+		err = res_tracker_insert(root, res_arr[i]);
+		if (err)
+			goto undo;
+		list_add_tail(&res_arr[i]->list,
+			      &tracker->slave_list[slave].res_list[type]);
+	}
+	spin_unlock_irq(mlx4_tlock(dev));
+	kfree(res_arr);
+
+	return 0;
+
+undo:
+	for (--i; i >= 0; --i) {
+		rb_erase(&res_arr[i]->node, root);
+		list_del_init(&res_arr[i]->list);
+	}
+
+	spin_unlock_irq(mlx4_tlock(dev));
+
+	for (i = 0; i < count; ++i)
+		kfree(res_arr[i]);
+
+	kfree(res_arr);
+
+	return err;
+}
+
+static int remove_qp_ok(struct res_qp *res)
+{
+	if (res->com.state == RES_QP_BUSY || atomic_read(&res->ref_count) ||
+	    !list_empty(&res->mcg_list)) {
+		pr_err("resource tracker: fail to remove qp, state %d, ref_count %d\n",
+		       res->com.state, atomic_read(&res->ref_count));
+		return -EBUSY;
+	} else if (res->com.state != RES_QP_RESERVED) {
+		return -EPERM;
+	}
+
+	return 0;
+}
+
+static int remove_mtt_ok(struct res_mtt *res, int order)
+{
+	if (res->com.state == RES_MTT_BUSY ||
+	    atomic_read(&res->ref_count)) {
+		pr_devel("%s-%d: state %s, ref_count %d\n",
+			 __func__, __LINE__,
+			 mtt_states_str(res->com.state),
+			 atomic_read(&res->ref_count));
+		return -EBUSY;
+	} else if (res->com.state != RES_MTT_ALLOCATED)
+		return -EPERM;
+	else if (res->order != order)
+		return -EINVAL;
+
+	return 0;
+}
+
+static int remove_mpt_ok(struct res_mpt *res)
+{
+	if (res->com.state == RES_MPT_BUSY)
+		return -EBUSY;
+	else if (res->com.state != RES_MPT_RESERVED)
+		return -EPERM;
+
+	return 0;
+}
+
+static int remove_eq_ok(struct res_eq *res)
+{
+	if (res->com.state == RES_MPT_BUSY)
+		return -EBUSY;
+	else if (res->com.state != RES_MPT_RESERVED)
+		return -EPERM;
+
+	return 0;
+}
+
+static int remove_counter_ok(struct res_counter *res)
+{
+	if (res->com.state == RES_COUNTER_BUSY)
+		return -EBUSY;
+	else if (res->com.state != RES_COUNTER_ALLOCATED)
+		return -EPERM;
+
+	return 0;
+}
+
+static int remove_xrcdn_ok(struct res_xrcdn *res)
+{
+	if (res->com.state == RES_XRCD_BUSY)
+		return -EBUSY;
+	else if (res->com.state != RES_XRCD_ALLOCATED)
+		return -EPERM;
+
+	return 0;
+}
+
+static int remove_fs_rule_ok(struct res_fs_rule *res)
+{
+	if (res->com.state == RES_FS_RULE_BUSY)
+		return -EBUSY;
+	else if (res->com.state != RES_FS_RULE_ALLOCATED)
+		return -EPERM;
+
+	return 0;
+}
+
+static int remove_cq_ok(struct res_cq *res)
+{
+	if (res->com.state == RES_CQ_BUSY)
+		return -EBUSY;
+	else if (res->com.state != RES_CQ_ALLOCATED)
+		return -EPERM;
+
+	return 0;
+}
+
+static int remove_srq_ok(struct res_srq *res)
+{
+	if (res->com.state == RES_SRQ_BUSY)
+		return -EBUSY;
+	else if (res->com.state != RES_SRQ_ALLOCATED)
+		return -EPERM;
+
+	return 0;
+}
+
+static int remove_ok(struct res_common *res, enum mlx4_resource type, int extra)
+{
+	switch (type) {
+	case RES_QP:
+		return remove_qp_ok((struct res_qp *)res);
+	case RES_CQ:
+		return remove_cq_ok((struct res_cq *)res);
+	case RES_SRQ:
+		return remove_srq_ok((struct res_srq *)res);
+	case RES_MPT:
+		return remove_mpt_ok((struct res_mpt *)res);
+	case RES_MTT:
+		return remove_mtt_ok((struct res_mtt *)res, extra);
+	case RES_MAC:
+		return -EOPNOTSUPP;
+	case RES_EQ:
+		return remove_eq_ok((struct res_eq *)res);
+	case RES_COUNTER:
+		return remove_counter_ok((struct res_counter *)res);
+	case RES_XRCD:
+		return remove_xrcdn_ok((struct res_xrcdn *)res);
+	case RES_FS_RULE:
+		return remove_fs_rule_ok((struct res_fs_rule *)res);
+	default:
+		return -EINVAL;
+	}
+}
+
+static int rem_res_range(struct mlx4_dev *dev, int slave, u64 base, int count,
+			 enum mlx4_resource type, int extra)
+{
+	u64 i;
+	int err;
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_resource_tracker *tracker = &priv->mfunc.master.res_tracker;
+	struct res_common *r;
+
+	spin_lock_irq(mlx4_tlock(dev));
+	for (i = base; i < base + count; ++i) {
+		r = res_tracker_lookup(&tracker->res_tree[type], i);
+		if (!r) {
+			err = -ENOENT;
+			goto out;
+		}
+		if (r->owner != slave) {
+			err = -EPERM;
+			goto out;
+		}
+		err = remove_ok(r, type, extra);
+		if (err)
+			goto out;
+	}
+
+	for (i = base; i < base + count; ++i) {
+		r = res_tracker_lookup(&tracker->res_tree[type], i);
+		rb_erase(&r->node, &tracker->res_tree[type]);
+		list_del(&r->list);
+		kfree(r);
+	}
+	err = 0;
+
+out:
+	spin_unlock_irq(mlx4_tlock(dev));
+
+	return err;
+}
+
+static int qp_res_start_move_to(struct mlx4_dev *dev, int slave, int qpn,
+				enum res_qp_states state, struct res_qp **qp,
+				int alloc)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_resource_tracker *tracker = &priv->mfunc.master.res_tracker;
+	struct res_qp *r;
+	int err = 0;
+
+	spin_lock_irq(mlx4_tlock(dev));
+	r = res_tracker_lookup(&tracker->res_tree[RES_QP], qpn);
+	if (!r)
+		err = -ENOENT;
+	else if (r->com.owner != slave)
+		err = -EPERM;
+	else {
+		switch (state) {
+		case RES_QP_BUSY:
+			mlx4_dbg(dev, "%s: failed RES_QP, 0x%llx\n",
+				 __func__, r->com.res_id);
+			err = -EBUSY;
+			break;
+
+		case RES_QP_RESERVED:
+			if (r->com.state == RES_QP_MAPPED && !alloc)
+				break;
+
+			mlx4_dbg(dev, "failed RES_QP, 0x%llx\n", r->com.res_id);
+			err = -EINVAL;
+			break;
+
+		case RES_QP_MAPPED:
+			if ((r->com.state == RES_QP_RESERVED && alloc) ||
+			    r->com.state == RES_QP_HW)
+				break;
+			else {
+				mlx4_dbg(dev, "failed RES_QP, 0x%llx\n",
+					  r->com.res_id);
+				err = -EINVAL;
+			}
+
+			break;
+
+		case RES_QP_HW:
+			if (r->com.state != RES_QP_MAPPED)
+				err = -EINVAL;
+			break;
+		default:
+			err = -EINVAL;
+		}
+
+		if (!err) {
+			r->com.from_state = r->com.state;
+			r->com.to_state = state;
+			r->com.state = RES_QP_BUSY;
+			if (qp)
+				*qp = r;
+		}
+	}
+
+	spin_unlock_irq(mlx4_tlock(dev));
+
+	return err;
+}
+
+static int mr_res_start_move_to(struct mlx4_dev *dev, int slave, int index,
+				enum res_mpt_states state, struct res_mpt **mpt)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_resource_tracker *tracker = &priv->mfunc.master.res_tracker;
+	struct res_mpt *r;
+	int err = 0;
+
+	spin_lock_irq(mlx4_tlock(dev));
+	r = res_tracker_lookup(&tracker->res_tree[RES_MPT], index);
+	if (!r)
+		err = -ENOENT;
+	else if (r->com.owner != slave)
+		err = -EPERM;
+	else {
+		switch (state) {
+		case RES_MPT_BUSY:
+			err = -EINVAL;
+			break;
+
+		case RES_MPT_RESERVED:
+			if (r->com.state != RES_MPT_MAPPED)
+				err = -EINVAL;
+			break;
+
+		case RES_MPT_MAPPED:
+			if (r->com.state != RES_MPT_RESERVED &&
+			    r->com.state != RES_MPT_HW)
+				err = -EINVAL;
+			break;
+
+		case RES_MPT_HW:
+			if (r->com.state != RES_MPT_MAPPED)
+				err = -EINVAL;
+			break;
+		default:
+			err = -EINVAL;
+		}
+
+		if (!err) {
+			r->com.from_state = r->com.state;
+			r->com.to_state = state;
+			r->com.state = RES_MPT_BUSY;
+			if (mpt)
+				*mpt = r;
+		}
+	}
+
+	spin_unlock_irq(mlx4_tlock(dev));
+
+	return err;
+}
+
+static int eq_res_start_move_to(struct mlx4_dev *dev, int slave, int index,
+				enum res_eq_states state, struct res_eq **eq)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_resource_tracker *tracker = &priv->mfunc.master.res_tracker;
+	struct res_eq *r;
+	int err = 0;
+
+	spin_lock_irq(mlx4_tlock(dev));
+	r = res_tracker_lookup(&tracker->res_tree[RES_EQ], index);
+	if (!r)
+		err = -ENOENT;
+	else if (r->com.owner != slave)
+		err = -EPERM;
+	else {
+		switch (state) {
+		case RES_EQ_BUSY:
+			err = -EINVAL;
+			break;
+
+		case RES_EQ_RESERVED:
+			if (r->com.state != RES_EQ_HW)
+				err = -EINVAL;
+			break;
+
+		case RES_EQ_HW:
+			if (r->com.state != RES_EQ_RESERVED)
+				err = -EINVAL;
+			break;
+
+		default:
+			err = -EINVAL;
+		}
+
+		if (!err) {
+			r->com.from_state = r->com.state;
+			r->com.to_state = state;
+			r->com.state = RES_EQ_BUSY;
+		}
+	}
+
+	spin_unlock_irq(mlx4_tlock(dev));
+
+	if (!err && eq)
+		*eq = r;
+
+	return err;
+}
+
+static int cq_res_start_move_to(struct mlx4_dev *dev, int slave, int cqn,
+				enum res_cq_states state, struct res_cq **cq)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_resource_tracker *tracker = &priv->mfunc.master.res_tracker;
+	struct res_cq *r;
+	int err;
+
+	spin_lock_irq(mlx4_tlock(dev));
+	r = res_tracker_lookup(&tracker->res_tree[RES_CQ], cqn);
+	if (!r) {
+		err = -ENOENT;
+	} else if (r->com.owner != slave) {
+		err = -EPERM;
+	} else if (state == RES_CQ_ALLOCATED) {
+		if (r->com.state != RES_CQ_HW)
+			err = -EINVAL;
+		else if (atomic_read(&r->ref_count))
+			err = -EBUSY;
+		else
+			err = 0;
+	} else if (state != RES_CQ_HW || r->com.state != RES_CQ_ALLOCATED) {
+		err = -EINVAL;
+	} else {
+		err = 0;
+	}
+
+	if (!err) {
+		r->com.from_state = r->com.state;
+		r->com.to_state = state;
+		r->com.state = RES_CQ_BUSY;
+		if (cq)
+			*cq = r;
+	}
+
+	spin_unlock_irq(mlx4_tlock(dev));
+
+	return err;
+}
+
+static int srq_res_start_move_to(struct mlx4_dev *dev, int slave, int index,
+				 enum res_srq_states state, struct res_srq **srq)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_resource_tracker *tracker = &priv->mfunc.master.res_tracker;
+	struct res_srq *r;
+	int err = 0;
+
+	spin_lock_irq(mlx4_tlock(dev));
+	r = res_tracker_lookup(&tracker->res_tree[RES_SRQ], index);
+	if (!r) {
+		err = -ENOENT;
+	} else if (r->com.owner != slave) {
+		err = -EPERM;
+	} else if (state == RES_SRQ_ALLOCATED) {
+		if (r->com.state != RES_SRQ_HW)
+			err = -EINVAL;
+		else if (atomic_read(&r->ref_count))
+			err = -EBUSY;
+	} else if (state != RES_SRQ_HW || r->com.state != RES_SRQ_ALLOCATED) {
+		err = -EINVAL;
+	}
+
+	if (!err) {
+		r->com.from_state = r->com.state;
+		r->com.to_state = state;
+		r->com.state = RES_SRQ_BUSY;
+		if (srq)
+			*srq = r;
+	}
+
+	spin_unlock_irq(mlx4_tlock(dev));
+
+	return err;
+}
+
+static void res_abort_move(struct mlx4_dev *dev, int slave,
+			   enum mlx4_resource type, int id)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_resource_tracker *tracker = &priv->mfunc.master.res_tracker;
+	struct res_common *r;
+
+	spin_lock_irq(mlx4_tlock(dev));
+	r = res_tracker_lookup(&tracker->res_tree[type], id);
+	if (r && (r->owner == slave))
+		r->state = r->from_state;
+	spin_unlock_irq(mlx4_tlock(dev));
+}
+
+static void res_end_move(struct mlx4_dev *dev, int slave,
+			 enum mlx4_resource type, int id)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_resource_tracker *tracker = &priv->mfunc.master.res_tracker;
+	struct res_common *r;
+
+	spin_lock_irq(mlx4_tlock(dev));
+	r = res_tracker_lookup(&tracker->res_tree[type], id);
+	if (r && (r->owner == slave))
+		r->state = r->to_state;
+	spin_unlock_irq(mlx4_tlock(dev));
+}
+
+static int valid_reserved(struct mlx4_dev *dev, int slave, int qpn)
+{
+	return mlx4_is_qp_reserved(dev, qpn) &&
+		(mlx4_is_master(dev) || mlx4_is_guest_proxy(dev, slave, qpn));
+}
+
+static int fw_reserved(struct mlx4_dev *dev, int qpn)
+{
+	return qpn < dev->caps.reserved_qps_cnt[MLX4_QP_REGION_FW];
+}
+
+static int qp_alloc_res(struct mlx4_dev *dev, int slave, int op, int cmd,
+			u64 in_param, u64 *out_param)
+{
+	int err;
+	int count;
+	int align;
+	int base;
+	int qpn;
+	u8 flags;
+
+	switch (op) {
+	case RES_OP_RESERVE:
+		count = get_param_l(&in_param) & 0xffffff;
+		/* Turn off all unsupported QP allocation flags that the
+		 * slave tries to set.
+		 */
+		flags = (get_param_l(&in_param) >> 24) & dev->caps.alloc_res_qp_mask;
+		align = get_param_h(&in_param);
+		err = mlx4_grant_resource(dev, slave, RES_QP, count, 0);
+		if (err)
+			return err;
+
+		err = __mlx4_qp_reserve_range(dev, count, align, &base, flags);
+		if (err) {
+			mlx4_release_resource(dev, slave, RES_QP, count, 0);
+			return err;
+		}
+
+		err = add_res_range(dev, slave, base, count, RES_QP, 0);
+		if (err) {
+			mlx4_release_resource(dev, slave, RES_QP, count, 0);
+			__mlx4_qp_release_range(dev, base, count);
+			return err;
+		}
+		set_param_l(out_param, base);
+		break;
+	case RES_OP_MAP_ICM:
+		qpn = get_param_l(&in_param) & 0x7fffff;
+		if (valid_reserved(dev, slave, qpn)) {
+			err = add_res_range(dev, slave, qpn, 1, RES_QP, 0);
+			if (err)
+				return err;
+		}
+
+		err = qp_res_start_move_to(dev, slave, qpn, RES_QP_MAPPED,
+					   NULL, 1);
+		if (err)
+			return err;
+
+		if (!fw_reserved(dev, qpn)) {
+			err = __mlx4_qp_alloc_icm(dev, qpn);
+			if (err) {
+				res_abort_move(dev, slave, RES_QP, qpn);
+				return err;
+			}
+		}
+
+		res_end_move(dev, slave, RES_QP, qpn);
+		break;
+
+	default:
+		err = -EINVAL;
+		break;
+	}
+	return err;
+}
+
+static int mtt_alloc_res(struct mlx4_dev *dev, int slave, int op, int cmd,
+			 u64 in_param, u64 *out_param)
+{
+	int err = -EINVAL;
+	int base;
+	int order;
+
+	if (op != RES_OP_RESERVE_AND_MAP)
+		return err;
+
+	order = get_param_l(&in_param);
+
+	err = mlx4_grant_resource(dev, slave, RES_MTT, 1 << order, 0);
+	if (err)
+		return err;
+
+	base = __mlx4_alloc_mtt_range(dev, order);
+	if (base == -1) {
+		mlx4_release_resource(dev, slave, RES_MTT, 1 << order, 0);
+		return -ENOMEM;
+	}
+
+	err = add_res_range(dev, slave, base, 1, RES_MTT, order);
+	if (err) {
+		mlx4_release_resource(dev, slave, RES_MTT, 1 << order, 0);
+		__mlx4_free_mtt_range(dev, base, order);
+	} else {
+		set_param_l(out_param, base);
+	}
+
+	return err;
+}
+
+static int mpt_alloc_res(struct mlx4_dev *dev, int slave, int op, int cmd,
+			 u64 in_param, u64 *out_param)
+{
+	int err = -EINVAL;
+	int index;
+	int id;
+	struct res_mpt *mpt;
+
+	switch (op) {
+	case RES_OP_RESERVE:
+		err = mlx4_grant_resource(dev, slave, RES_MPT, 1, 0);
+		if (err)
+			break;
+
+		index = __mlx4_mpt_reserve(dev);
+		if (index == -1) {
+			mlx4_release_resource(dev, slave, RES_MPT, 1, 0);
+			break;
+		}
+		id = index & mpt_mask(dev);
+
+		err = add_res_range(dev, slave, id, 1, RES_MPT, index);
+		if (err) {
+			mlx4_release_resource(dev, slave, RES_MPT, 1, 0);
+			__mlx4_mpt_release(dev, index);
+			break;
+		}
+		set_param_l(out_param, index);
+		break;
+	case RES_OP_MAP_ICM:
+		index = get_param_l(&in_param);
+		id = index & mpt_mask(dev);
+		err = mr_res_start_move_to(dev, slave, id,
+					   RES_MPT_MAPPED, &mpt);
+		if (err)
+			return err;
+
+		err = __mlx4_mpt_alloc_icm(dev, mpt->key);
+		if (err) {
+			res_abort_move(dev, slave, RES_MPT, id);
+			return err;
+		}
+
+		res_end_move(dev, slave, RES_MPT, id);
+		break;
+	}
+	return err;
+}
+
+static int cq_alloc_res(struct mlx4_dev *dev, int slave, int op, int cmd,
+			u64 in_param, u64 *out_param)
+{
+	int cqn;
+	int err;
+
+	switch (op) {
+	case RES_OP_RESERVE_AND_MAP:
+		err = mlx4_grant_resource(dev, slave, RES_CQ, 1, 0);
+		if (err)
+			break;
+
+		err = __mlx4_cq_alloc_icm(dev, &cqn);
+		if (err) {
+			mlx4_release_resource(dev, slave, RES_CQ, 1, 0);
+			break;
+		}
+
+		err = add_res_range(dev, slave, cqn, 1, RES_CQ, 0);
+		if (err) {
+			mlx4_release_resource(dev, slave, RES_CQ, 1, 0);
+			__mlx4_cq_free_icm(dev, cqn);
+			break;
+		}
+
+		set_param_l(out_param, cqn);
+		break;
+
+	default:
+		err = -EINVAL;
+	}
+
+	return err;
+}
+
+static int srq_alloc_res(struct mlx4_dev *dev, int slave, int op, int cmd,
+			 u64 in_param, u64 *out_param)
+{
+	int srqn;
+	int err;
+
+	switch (op) {
+	case RES_OP_RESERVE_AND_MAP:
+		err = mlx4_grant_resource(dev, slave, RES_SRQ, 1, 0);
+		if (err)
+			break;
+
+		err = __mlx4_srq_alloc_icm(dev, &srqn);
+		if (err) {
+			mlx4_release_resource(dev, slave, RES_SRQ, 1, 0);
+			break;
+		}
+
+		err = add_res_range(dev, slave, srqn, 1, RES_SRQ, 0);
+		if (err) {
+			mlx4_release_resource(dev, slave, RES_SRQ, 1, 0);
+			__mlx4_srq_free_icm(dev, srqn);
+			break;
+		}
+
+		set_param_l(out_param, srqn);
+		break;
+
+	default:
+		err = -EINVAL;
+	}
+
+	return err;
+}
+
+static int mac_find_smac_ix_in_slave(struct mlx4_dev *dev, int slave, int port,
+				     u8 smac_index, u64 *mac)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_resource_tracker *tracker = &priv->mfunc.master.res_tracker;
+	struct list_head *mac_list =
+		&tracker->slave_list[slave].res_list[RES_MAC];
+	struct mac_res *res, *tmp;
+
+	list_for_each_entry_safe(res, tmp, mac_list, list) {
+		if (res->smac_index == smac_index && res->port == (u8) port) {
+			*mac = res->mac;
+			return 0;
+		}
+	}
+	return -ENOENT;
+}
+
+static int mac_add_to_slave(struct mlx4_dev *dev, int slave, u64 mac, int port, u8 smac_index)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_resource_tracker *tracker = &priv->mfunc.master.res_tracker;
+	struct list_head *mac_list =
+		&tracker->slave_list[slave].res_list[RES_MAC];
+	struct mac_res *res, *tmp;
+
+	list_for_each_entry_safe(res, tmp, mac_list, list) {
+		if (res->mac == mac && res->port == (u8) port) {
+			/* mac found. update ref count */
+			++res->ref_count;
+			return 0;
+		}
+	}
+
+	if (mlx4_grant_resource(dev, slave, RES_MAC, 1, port))
+		return -EINVAL;
+	res = kzalloc(sizeof(*res), GFP_KERNEL);
+	if (!res) {
+		mlx4_release_resource(dev, slave, RES_MAC, 1, port);
+		return -ENOMEM;
+	}
+	res->mac = mac;
+	res->port = (u8) port;
+	res->smac_index = smac_index;
+	res->ref_count = 1;
+	list_add_tail(&res->list,
+		      &tracker->slave_list[slave].res_list[RES_MAC]);
+	return 0;
+}
+
+static void mac_del_from_slave(struct mlx4_dev *dev, int slave, u64 mac,
+			       int port)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_resource_tracker *tracker = &priv->mfunc.master.res_tracker;
+	struct list_head *mac_list =
+		&tracker->slave_list[slave].res_list[RES_MAC];
+	struct mac_res *res, *tmp;
+
+	list_for_each_entry_safe(res, tmp, mac_list, list) {
+		if (res->mac == mac && res->port == (u8) port) {
+			if (!--res->ref_count) {
+				list_del(&res->list);
+				mlx4_release_resource(dev, slave, RES_MAC, 1, port);
+				kfree(res);
+			}
+			break;
+		}
+	}
+}
+
+static void rem_slave_macs(struct mlx4_dev *dev, int slave)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_resource_tracker *tracker = &priv->mfunc.master.res_tracker;
+	struct list_head *mac_list =
+		&tracker->slave_list[slave].res_list[RES_MAC];
+	struct mac_res *res, *tmp;
+	int i;
+
+	list_for_each_entry_safe(res, tmp, mac_list, list) {
+		list_del(&res->list);
+		/* dereference the mac the num times the slave referenced it */
+		for (i = 0; i < res->ref_count; i++)
+			__mlx4_unregister_mac(dev, res->port, res->mac);
+		mlx4_release_resource(dev, slave, RES_MAC, 1, res->port);
+		kfree(res);
+	}
+}
+
+static int mac_alloc_res(struct mlx4_dev *dev, int slave, int op, int cmd,
+			 u64 in_param, u64 *out_param, int in_port)
+{
+	int err = -EINVAL;
+	int port;
+	u64 mac;
+	u8 smac_index;
+
+	if (op != RES_OP_RESERVE_AND_MAP)
+		return err;
+
+	port = !in_port ? get_param_l(out_param) : in_port;
+	port = mlx4_slave_convert_port(
+			dev, slave, port);
+
+	if (port < 0)
+		return -EINVAL;
+	mac = in_param;
+
+	err = __mlx4_register_mac(dev, port, mac);
+	if (err >= 0) {
+		smac_index = err;
+		set_param_l(out_param, err);
+		err = 0;
+	}
+
+	if (!err) {
+		err = mac_add_to_slave(dev, slave, mac, port, smac_index);
+		if (err)
+			__mlx4_unregister_mac(dev, port, mac);
+	}
+	return err;
+}
+
+static int vlan_add_to_slave(struct mlx4_dev *dev, int slave, u16 vlan,
+			     int port, int vlan_index)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_resource_tracker *tracker = &priv->mfunc.master.res_tracker;
+	struct list_head *vlan_list =
+		&tracker->slave_list[slave].res_list[RES_VLAN];
+	struct vlan_res *res, *tmp;
+
+	list_for_each_entry_safe(res, tmp, vlan_list, list) {
+		if (res->vlan == vlan && res->port == (u8) port) {
+			/* vlan found. update ref count */
+			++res->ref_count;
+			return 0;
+		}
+	}
+
+	if (mlx4_grant_resource(dev, slave, RES_VLAN, 1, port))
+		return -EINVAL;
+	res = kzalloc(sizeof(*res), GFP_KERNEL);
+	if (!res) {
+		mlx4_release_resource(dev, slave, RES_VLAN, 1, port);
+		return -ENOMEM;
+	}
+	res->vlan = vlan;
+	res->port = (u8) port;
+	res->vlan_index = vlan_index;
+	res->ref_count = 1;
+	list_add_tail(&res->list,
+		      &tracker->slave_list[slave].res_list[RES_VLAN]);
+	return 0;
+}
+
+
+static void vlan_del_from_slave(struct mlx4_dev *dev, int slave, u16 vlan,
+				int port)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_resource_tracker *tracker = &priv->mfunc.master.res_tracker;
+	struct list_head *vlan_list =
+		&tracker->slave_list[slave].res_list[RES_VLAN];
+	struct vlan_res *res, *tmp;
+
+	list_for_each_entry_safe(res, tmp, vlan_list, list) {
+		if (res->vlan == vlan && res->port == (u8) port) {
+			if (!--res->ref_count) {
+				list_del(&res->list);
+				mlx4_release_resource(dev, slave, RES_VLAN,
+						      1, port);
+				kfree(res);
+			}
+			break;
+		}
+	}
+}
+
+static void rem_slave_vlans(struct mlx4_dev *dev, int slave)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_resource_tracker *tracker = &priv->mfunc.master.res_tracker;
+	struct list_head *vlan_list =
+		&tracker->slave_list[slave].res_list[RES_VLAN];
+	struct vlan_res *res, *tmp;
+	int i;
+
+	list_for_each_entry_safe(res, tmp, vlan_list, list) {
+		list_del(&res->list);
+		/* dereference the vlan the num times the slave referenced it */
+		for (i = 0; i < res->ref_count; i++)
+			__mlx4_unregister_vlan(dev, res->port, res->vlan);
+		mlx4_release_resource(dev, slave, RES_VLAN, 1, res->port);
+		kfree(res);
+	}
+}
+
+static int vlan_alloc_res(struct mlx4_dev *dev, int slave, int op, int cmd,
+			  u64 in_param, u64 *out_param, int in_port)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_slave_state *slave_state = priv->mfunc.master.slave_state;
+	int err;
+	u16 vlan;
+	int vlan_index;
+	int port;
+
+	port = !in_port ? get_param_l(out_param) : in_port;
+
+	if (!port || op != RES_OP_RESERVE_AND_MAP)
+		return -EINVAL;
+
+	port = mlx4_slave_convert_port(
+			dev, slave, port);
+
+	if (port < 0)
+		return -EINVAL;
+	/* upstream kernels had NOP for reg/unreg vlan. Continue this. */
+	if (!in_port && port > 0 && port <= dev->caps.num_ports) {
+		slave_state[slave].old_vlan_api = true;
+		return 0;
+	}
+
+	vlan = (u16) in_param;
+
+	err = __mlx4_register_vlan(dev, port, vlan, &vlan_index);
+	if (!err) {
+		set_param_l(out_param, (u32) vlan_index);
+		err = vlan_add_to_slave(dev, slave, vlan, port, vlan_index);
+		if (err)
+			__mlx4_unregister_vlan(dev, port, vlan);
+	}
+	return err;
+}
+
+static int counter_alloc_res(struct mlx4_dev *dev, int slave, int op, int cmd,
+			     u64 in_param, u64 *out_param, int port)
+{
+	u32 index;
+	int err;
+
+	if (op != RES_OP_RESERVE)
+		return -EINVAL;
+
+	err = mlx4_grant_resource(dev, slave, RES_COUNTER, 1, 0);
+	if (err)
+		return err;
+
+	err = __mlx4_counter_alloc(dev, &index);
+	if (err) {
+		mlx4_release_resource(dev, slave, RES_COUNTER, 1, 0);
+		return err;
+	}
+
+	err = add_res_range(dev, slave, index, 1, RES_COUNTER, port);
+	if (err) {
+		__mlx4_counter_free(dev, index);
+		mlx4_release_resource(dev, slave, RES_COUNTER, 1, 0);
+	} else {
+		set_param_l(out_param, index);
+	}
+
+	return err;
+}
+
+static int xrcdn_alloc_res(struct mlx4_dev *dev, int slave, int op, int cmd,
+			   u64 in_param, u64 *out_param)
+{
+	u32 xrcdn;
+	int err;
+
+	if (op != RES_OP_RESERVE)
+		return -EINVAL;
+
+	err = __mlx4_xrcd_alloc(dev, &xrcdn);
+	if (err)
+		return err;
+
+	err = add_res_range(dev, slave, xrcdn, 1, RES_XRCD, 0);
+	if (err)
+		__mlx4_xrcd_free(dev, xrcdn);
+	else
+		set_param_l(out_param, xrcdn);
+
+	return err;
+}
+
+int mlx4_ALLOC_RES_wrapper(struct mlx4_dev *dev, int slave,
+			   struct mlx4_vhcr *vhcr,
+			   struct mlx4_cmd_mailbox *inbox,
+			   struct mlx4_cmd_mailbox *outbox,
+			   struct mlx4_cmd_info *cmd)
+{
+	int err;
+	int alop = vhcr->op_modifier;
+
+	switch (vhcr->in_modifier & 0xFF) {
+	case RES_QP:
+		err = qp_alloc_res(dev, slave, vhcr->op_modifier, alop,
+				   vhcr->in_param, &vhcr->out_param);
+		break;
+
+	case RES_MTT:
+		err = mtt_alloc_res(dev, slave, vhcr->op_modifier, alop,
+				    vhcr->in_param, &vhcr->out_param);
+		break;
+
+	case RES_MPT:
+		err = mpt_alloc_res(dev, slave, vhcr->op_modifier, alop,
+				    vhcr->in_param, &vhcr->out_param);
+		break;
+
+	case RES_CQ:
+		err = cq_alloc_res(dev, slave, vhcr->op_modifier, alop,
+				   vhcr->in_param, &vhcr->out_param);
+		break;
+
+	case RES_SRQ:
+		err = srq_alloc_res(dev, slave, vhcr->op_modifier, alop,
+				    vhcr->in_param, &vhcr->out_param);
+		break;
+
+	case RES_MAC:
+		err = mac_alloc_res(dev, slave, vhcr->op_modifier, alop,
+				    vhcr->in_param, &vhcr->out_param,
+				    (vhcr->in_modifier >> 8) & 0xFF);
+		break;
+
+	case RES_VLAN:
+		err = vlan_alloc_res(dev, slave, vhcr->op_modifier, alop,
+				     vhcr->in_param, &vhcr->out_param,
+				     (vhcr->in_modifier >> 8) & 0xFF);
+		break;
+
+	case RES_COUNTER:
+		err = counter_alloc_res(dev, slave, vhcr->op_modifier, alop,
+					vhcr->in_param, &vhcr->out_param, 0);
+		break;
+
+	case RES_XRCD:
+		err = xrcdn_alloc_res(dev, slave, vhcr->op_modifier, alop,
+				      vhcr->in_param, &vhcr->out_param);
+		break;
+
+	default:
+		err = -EINVAL;
+		break;
+	}
+
+	return err;
+}
+
+static int qp_free_res(struct mlx4_dev *dev, int slave, int op, int cmd,
+		       u64 in_param)
+{
+	int err;
+	int count;
+	int base;
+	int qpn;
+
+	switch (op) {
+	case RES_OP_RESERVE:
+		base = get_param_l(&in_param) & 0x7fffff;
+		count = get_param_h(&in_param);
+		err = rem_res_range(dev, slave, base, count, RES_QP, 0);
+		if (err)
+			break;
+		mlx4_release_resource(dev, slave, RES_QP, count, 0);
+		__mlx4_qp_release_range(dev, base, count);
+		break;
+	case RES_OP_MAP_ICM:
+		qpn = get_param_l(&in_param) & 0x7fffff;
+		err = qp_res_start_move_to(dev, slave, qpn, RES_QP_RESERVED,
+					   NULL, 0);
+		if (err)
+			return err;
+
+		if (!fw_reserved(dev, qpn))
+			__mlx4_qp_free_icm(dev, qpn);
+
+		res_end_move(dev, slave, RES_QP, qpn);
+
+		if (valid_reserved(dev, slave, qpn))
+			err = rem_res_range(dev, slave, qpn, 1, RES_QP, 0);
+		break;
+	default:
+		err = -EINVAL;
+		break;
+	}
+	return err;
+}
+
+static int mtt_free_res(struct mlx4_dev *dev, int slave, int op, int cmd,
+			u64 in_param, u64 *out_param)
+{
+	int err = -EINVAL;
+	int base;
+	int order;
+
+	if (op != RES_OP_RESERVE_AND_MAP)
+		return err;
+
+	base = get_param_l(&in_param);
+	order = get_param_h(&in_param);
+	err = rem_res_range(dev, slave, base, 1, RES_MTT, order);
+	if (!err) {
+		mlx4_release_resource(dev, slave, RES_MTT, 1 << order, 0);
+		__mlx4_free_mtt_range(dev, base, order);
+	}
+	return err;
+}
+
+static int mpt_free_res(struct mlx4_dev *dev, int slave, int op, int cmd,
+			u64 in_param)
+{
+	int err = -EINVAL;
+	int index;
+	int id;
+	struct res_mpt *mpt;
+
+	switch (op) {
+	case RES_OP_RESERVE:
+		index = get_param_l(&in_param);
+		id = index & mpt_mask(dev);
+		err = get_res(dev, slave, id, RES_MPT, &mpt);
+		if (err)
+			break;
+		index = mpt->key;
+		put_res(dev, slave, id, RES_MPT);
+
+		err = rem_res_range(dev, slave, id, 1, RES_MPT, 0);
+		if (err)
+			break;
+		mlx4_release_resource(dev, slave, RES_MPT, 1, 0);
+		__mlx4_mpt_release(dev, index);
+		break;
+	case RES_OP_MAP_ICM:
+		index = get_param_l(&in_param);
+		id = index & mpt_mask(dev);
+		err = mr_res_start_move_to(dev, slave, id,
+					   RES_MPT_RESERVED, &mpt);
+		if (err)
+			return err;
+
+		__mlx4_mpt_free_icm(dev, mpt->key);
+		res_end_move(dev, slave, RES_MPT, id);
+		break;
+	default:
+		err = -EINVAL;
+		break;
+	}
+	return err;
+}
+
+static int cq_free_res(struct mlx4_dev *dev, int slave, int op, int cmd,
+		       u64 in_param, u64 *out_param)
+{
+	int cqn;
+	int err;
+
+	switch (op) {
+	case RES_OP_RESERVE_AND_MAP:
+		cqn = get_param_l(&in_param);
+		err = rem_res_range(dev, slave, cqn, 1, RES_CQ, 0);
+		if (err)
+			break;
+
+		mlx4_release_resource(dev, slave, RES_CQ, 1, 0);
+		__mlx4_cq_free_icm(dev, cqn);
+		break;
+
+	default:
+		err = -EINVAL;
+		break;
+	}
+
+	return err;
+}
+
+static int srq_free_res(struct mlx4_dev *dev, int slave, int op, int cmd,
+			u64 in_param, u64 *out_param)
+{
+	int srqn;
+	int err;
+
+	switch (op) {
+	case RES_OP_RESERVE_AND_MAP:
+		srqn = get_param_l(&in_param);
+		err = rem_res_range(dev, slave, srqn, 1, RES_SRQ, 0);
+		if (err)
+			break;
+
+		mlx4_release_resource(dev, slave, RES_SRQ, 1, 0);
+		__mlx4_srq_free_icm(dev, srqn);
+		break;
+
+	default:
+		err = -EINVAL;
+		break;
+	}
+
+	return err;
+}
+
+static int mac_free_res(struct mlx4_dev *dev, int slave, int op, int cmd,
+			    u64 in_param, u64 *out_param, int in_port)
+{
+	int port;
+	int err = 0;
+
+	switch (op) {
+	case RES_OP_RESERVE_AND_MAP:
+		port = !in_port ? get_param_l(out_param) : in_port;
+		port = mlx4_slave_convert_port(
+				dev, slave, port);
+
+		if (port < 0)
+			return -EINVAL;
+		mac_del_from_slave(dev, slave, in_param, port);
+		__mlx4_unregister_mac(dev, port, in_param);
+		break;
+	default:
+		err = -EINVAL;
+		break;
+	}
+
+	return err;
+
+}
+
+static int vlan_free_res(struct mlx4_dev *dev, int slave, int op, int cmd,
+			    u64 in_param, u64 *out_param, int port)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_slave_state *slave_state = priv->mfunc.master.slave_state;
+	int err = 0;
+
+	port = mlx4_slave_convert_port(
+			dev, slave, port);
+
+	if (port < 0)
+		return -EINVAL;
+	switch (op) {
+	case RES_OP_RESERVE_AND_MAP:
+		if (slave_state[slave].old_vlan_api)
+			return 0;
+		if (!port)
+			return -EINVAL;
+		vlan_del_from_slave(dev, slave, in_param, port);
+		__mlx4_unregister_vlan(dev, port, in_param);
+		break;
+	default:
+		err = -EINVAL;
+		break;
+	}
+
+	return err;
+}
+
+static int counter_free_res(struct mlx4_dev *dev, int slave, int op, int cmd,
+			    u64 in_param, u64 *out_param)
+{
+	int index;
+	int err;
+
+	if (op != RES_OP_RESERVE)
+		return -EINVAL;
+
+	index = get_param_l(&in_param);
+	if (index == MLX4_SINK_COUNTER_INDEX(dev))
+		return 0;
+
+	err = rem_res_range(dev, slave, index, 1, RES_COUNTER, 0);
+	if (err)
+		return err;
+
+	__mlx4_counter_free(dev, index);
+	mlx4_release_resource(dev, slave, RES_COUNTER, 1, 0);
+
+	return err;
+}
+
+static int xrcdn_free_res(struct mlx4_dev *dev, int slave, int op, int cmd,
+			  u64 in_param, u64 *out_param)
+{
+	int xrcdn;
+	int err;
+
+	if (op != RES_OP_RESERVE)
+		return -EINVAL;
+
+	xrcdn = get_param_l(&in_param);
+	err = rem_res_range(dev, slave, xrcdn, 1, RES_XRCD, 0);
+	if (err)
+		return err;
+
+	__mlx4_xrcd_free(dev, xrcdn);
+
+	return err;
+}
+
+int mlx4_FREE_RES_wrapper(struct mlx4_dev *dev, int slave,
+			  struct mlx4_vhcr *vhcr,
+			  struct mlx4_cmd_mailbox *inbox,
+			  struct mlx4_cmd_mailbox *outbox,
+			  struct mlx4_cmd_info *cmd)
+{
+	int err = -EINVAL;
+	int alop = vhcr->op_modifier;
+
+	switch (vhcr->in_modifier & 0xFF) {
+	case RES_QP:
+		err = qp_free_res(dev, slave, vhcr->op_modifier, alop,
+				  vhcr->in_param);
+		break;
+
+	case RES_MTT:
+		err = mtt_free_res(dev, slave, vhcr->op_modifier, alop,
+				   vhcr->in_param, &vhcr->out_param);
+		break;
+
+	case RES_MPT:
+		err = mpt_free_res(dev, slave, vhcr->op_modifier, alop,
+				   vhcr->in_param);
+		break;
+
+	case RES_CQ:
+		err = cq_free_res(dev, slave, vhcr->op_modifier, alop,
+				  vhcr->in_param, &vhcr->out_param);
+		break;
+
+	case RES_SRQ:
+		err = srq_free_res(dev, slave, vhcr->op_modifier, alop,
+				   vhcr->in_param, &vhcr->out_param);
+		break;
+
+	case RES_MAC:
+		err = mac_free_res(dev, slave, vhcr->op_modifier, alop,
+				   vhcr->in_param, &vhcr->out_param,
+				   (vhcr->in_modifier >> 8) & 0xFF);
+		break;
+
+	case RES_VLAN:
+		err = vlan_free_res(dev, slave, vhcr->op_modifier, alop,
+				    vhcr->in_param, &vhcr->out_param,
+				    (vhcr->in_modifier >> 8) & 0xFF);
+		break;
+
+	case RES_COUNTER:
+		err = counter_free_res(dev, slave, vhcr->op_modifier, alop,
+				       vhcr->in_param, &vhcr->out_param);
+		break;
+
+	case RES_XRCD:
+		err = xrcdn_free_res(dev, slave, vhcr->op_modifier, alop,
+				     vhcr->in_param, &vhcr->out_param);
+		break;
+
+	default:
+		break;
+	}
+	return err;
+}
+
+/* ugly but other choices are uglier */
+static int mr_phys_mpt(struct mlx4_mpt_entry *mpt)
+{
+	return (be32_to_cpu(mpt->flags) >> 9) & 1;
+}
+
+static int mr_get_mtt_addr(struct mlx4_mpt_entry *mpt)
+{
+	return (int)be64_to_cpu(mpt->mtt_addr) & 0xfffffff8;
+}
+
+static int mr_get_mtt_size(struct mlx4_mpt_entry *mpt)
+{
+	return be32_to_cpu(mpt->mtt_sz);
+}
+
+static u32 mr_get_pd(struct mlx4_mpt_entry *mpt)
+{
+	return be32_to_cpu(mpt->pd_flags) & 0x00ffffff;
+}
+
+static int mr_is_fmr(struct mlx4_mpt_entry *mpt)
+{
+	return be32_to_cpu(mpt->pd_flags) & MLX4_MPT_PD_FLAG_FAST_REG;
+}
+
+static int mr_is_bind_enabled(struct mlx4_mpt_entry *mpt)
+{
+	return be32_to_cpu(mpt->flags) & MLX4_MPT_FLAG_BIND_ENABLE;
+}
+
+static int mr_is_region(struct mlx4_mpt_entry *mpt)
+{
+	return be32_to_cpu(mpt->flags) & MLX4_MPT_FLAG_REGION;
+}
+
+static int qp_get_mtt_addr(struct mlx4_qp_context *qpc)
+{
+	return be32_to_cpu(qpc->mtt_base_addr_l) & 0xfffffff8;
+}
+
+static int srq_get_mtt_addr(struct mlx4_srq_context *srqc)
+{
+	return be32_to_cpu(srqc->mtt_base_addr_l) & 0xfffffff8;
+}
+
+static int qp_get_mtt_size(struct mlx4_qp_context *qpc)
+{
+	int page_shift = (qpc->log_page_size & 0x3f) + 12;
+	int log_sq_size = (qpc->sq_size_stride >> 3) & 0xf;
+	int log_sq_sride = qpc->sq_size_stride & 7;
+	int log_rq_size = (qpc->rq_size_stride >> 3) & 0xf;
+	int log_rq_stride = qpc->rq_size_stride & 7;
+	int srq = (be32_to_cpu(qpc->srqn) >> 24) & 1;
+	int rss = (be32_to_cpu(qpc->flags) >> 13) & 1;
+	u32 ts = (be32_to_cpu(qpc->flags) >> 16) & 0xff;
+	int xrc = (ts == MLX4_QP_ST_XRC) ? 1 : 0;
+	int sq_size;
+	int rq_size;
+	int total_pages;
+	int total_mem;
+	int page_offset = (be32_to_cpu(qpc->params2) >> 6) & 0x3f;
+	int tot;
+
+	sq_size = 1 << (log_sq_size + log_sq_sride + 4);
+	rq_size = (srq|rss|xrc) ? 0 : (1 << (log_rq_size + log_rq_stride + 4));
+	total_mem = sq_size + rq_size;
+	tot = (total_mem + (page_offset << 6)) >> page_shift;
+	total_pages = !tot ? 1 : roundup_pow_of_two(tot);
+
+	return total_pages;
+}
+
+static int check_mtt_range(struct mlx4_dev *dev, int slave, int start,
+			   int size, struct res_mtt *mtt)
+{
+	int res_start = mtt->com.res_id;
+	int res_size = (1 << mtt->order);
+
+	if (start < res_start || start + size > res_start + res_size)
+		return -EPERM;
+	return 0;
+}
+
+int mlx4_SW2HW_MPT_wrapper(struct mlx4_dev *dev, int slave,
+			   struct mlx4_vhcr *vhcr,
+			   struct mlx4_cmd_mailbox *inbox,
+			   struct mlx4_cmd_mailbox *outbox,
+			   struct mlx4_cmd_info *cmd)
+{
+	int err;
+	int index = vhcr->in_modifier;
+	struct res_mtt *mtt;
+	struct res_mpt *mpt = NULL;
+	int mtt_base = mr_get_mtt_addr(inbox->buf) / dev->caps.mtt_entry_sz;
+	int phys;
+	int id;
+	u32 pd;
+	int pd_slave;
+
+	id = index & mpt_mask(dev);
+	err = mr_res_start_move_to(dev, slave, id, RES_MPT_HW, &mpt);
+	if (err)
+		return err;
+
+	/* Disable memory windows for VFs. */
+	if (!mr_is_region(inbox->buf)) {
+		err = -EPERM;
+		goto ex_abort;
+	}
+
+	/* Make sure that the PD bits related to the slave id are zeros. */
+	pd = mr_get_pd(inbox->buf);
+	pd_slave = (pd >> 17) & 0x7f;
+	if (pd_slave != 0 && --pd_slave != slave) {
+		err = -EPERM;
+		goto ex_abort;
+	}
+
+	if (mr_is_fmr(inbox->buf)) {
+		/* FMR and Bind Enable are forbidden in slave devices. */
+		if (mr_is_bind_enabled(inbox->buf)) {
+			err = -EPERM;
+			goto ex_abort;
+		}
+		/* FMR and Memory Windows are also forbidden. */
+		if (!mr_is_region(inbox->buf)) {
+			err = -EPERM;
+			goto ex_abort;
+		}
+	}
+
+	phys = mr_phys_mpt(inbox->buf);
+	if (!phys) {
+		err = get_res(dev, slave, mtt_base, RES_MTT, &mtt);
+		if (err)
+			goto ex_abort;
+
+		err = check_mtt_range(dev, slave, mtt_base,
+				      mr_get_mtt_size(inbox->buf), mtt);
+		if (err)
+			goto ex_put;
+
+		mpt->mtt = mtt;
+	}
+
+	err = mlx4_DMA_wrapper(dev, slave, vhcr, inbox, outbox, cmd);
+	if (err)
+		goto ex_put;
+
+	if (!phys) {
+		atomic_inc(&mtt->ref_count);
+		put_res(dev, slave, mtt->com.res_id, RES_MTT);
+	}
+
+	res_end_move(dev, slave, RES_MPT, id);
+	return 0;
+
+ex_put:
+	if (!phys)
+		put_res(dev, slave, mtt->com.res_id, RES_MTT);
+ex_abort:
+	res_abort_move(dev, slave, RES_MPT, id);
+
+	return err;
+}
+
+int mlx4_HW2SW_MPT_wrapper(struct mlx4_dev *dev, int slave,
+			   struct mlx4_vhcr *vhcr,
+			   struct mlx4_cmd_mailbox *inbox,
+			   struct mlx4_cmd_mailbox *outbox,
+			   struct mlx4_cmd_info *cmd)
+{
+	int err;
+	int index = vhcr->in_modifier;
+	struct res_mpt *mpt;
+	int id;
+
+	id = index & mpt_mask(dev);
+	err = mr_res_start_move_to(dev, slave, id, RES_MPT_MAPPED, &mpt);
+	if (err)
+		return err;
+
+	err = mlx4_DMA_wrapper(dev, slave, vhcr, inbox, outbox, cmd);
+	if (err)
+		goto ex_abort;
+
+	if (mpt->mtt)
+		atomic_dec(&mpt->mtt->ref_count);
+
+	res_end_move(dev, slave, RES_MPT, id);
+	return 0;
+
+ex_abort:
+	res_abort_move(dev, slave, RES_MPT, id);
+
+	return err;
+}
+
+int mlx4_QUERY_MPT_wrapper(struct mlx4_dev *dev, int slave,
+			   struct mlx4_vhcr *vhcr,
+			   struct mlx4_cmd_mailbox *inbox,
+			   struct mlx4_cmd_mailbox *outbox,
+			   struct mlx4_cmd_info *cmd)
+{
+	int err;
+	int index = vhcr->in_modifier;
+	struct res_mpt *mpt;
+	int id;
+
+	id = index & mpt_mask(dev);
+	err = get_res(dev, slave, id, RES_MPT, &mpt);
+	if (err)
+		return err;
+
+	if (mpt->com.from_state == RES_MPT_MAPPED) {
+		/* In order to allow rereg in SRIOV, we need to alter the MPT entry. To do
+		 * that, the VF must read the MPT. But since the MPT entry memory is not
+		 * in the VF's virtual memory space, it must use QUERY_MPT to obtain the
+		 * entry contents. To guarantee that the MPT cannot be changed, the driver
+		 * must perform HW2SW_MPT before this query and return the MPT entry to HW
+		 * ownership fofollowing the change. The change here allows the VF to
+		 * perform QUERY_MPT also when the entry is in SW ownership.
+		 */
+		struct mlx4_mpt_entry *mpt_entry = mlx4_table_find(
+					&mlx4_priv(dev)->mr_table.dmpt_table,
+					mpt->key, NULL);
+
+		if (NULL == mpt_entry || NULL == outbox->buf) {
+			err = -EINVAL;
+			goto out;
+		}
+
+		memcpy(outbox->buf, mpt_entry, sizeof(*mpt_entry));
+
+		err = 0;
+	} else if (mpt->com.from_state == RES_MPT_HW) {
+		err = mlx4_DMA_wrapper(dev, slave, vhcr, inbox, outbox, cmd);
+	} else {
+		err = -EBUSY;
+		goto out;
+	}
+
+
+out:
+	put_res(dev, slave, id, RES_MPT);
+	return err;
+}
+
+static int qp_get_rcqn(struct mlx4_qp_context *qpc)
+{
+	return be32_to_cpu(qpc->cqn_recv) & 0xffffff;
+}
+
+static int qp_get_scqn(struct mlx4_qp_context *qpc)
+{
+	return be32_to_cpu(qpc->cqn_send) & 0xffffff;
+}
+
+static u32 qp_get_srqn(struct mlx4_qp_context *qpc)
+{
+	return be32_to_cpu(qpc->srqn) & 0x1ffffff;
+}
+
+static void adjust_proxy_tun_qkey(struct mlx4_dev *dev, struct mlx4_vhcr *vhcr,
+				  struct mlx4_qp_context *context)
+{
+	u32 qpn = vhcr->in_modifier & 0xffffff;
+	u32 qkey = 0;
+
+	if (mlx4_get_parav_qkey(dev, qpn, &qkey))
+		return;
+
+	/* adjust qkey in qp context */
+	context->qkey = cpu_to_be32(qkey);
+}
+
+static int adjust_qp_sched_queue(struct mlx4_dev *dev, int slave,
+				 struct mlx4_qp_context *qpc,
+				 struct mlx4_cmd_mailbox *inbox);
+
+int mlx4_RST2INIT_QP_wrapper(struct mlx4_dev *dev, int slave,
+			     struct mlx4_vhcr *vhcr,
+			     struct mlx4_cmd_mailbox *inbox,
+			     struct mlx4_cmd_mailbox *outbox,
+			     struct mlx4_cmd_info *cmd)
+{
+	int err;
+	int qpn = vhcr->in_modifier & 0x7fffff;
+	struct res_mtt *mtt;
+	struct res_qp *qp;
+	struct mlx4_qp_context *qpc = inbox->buf + 8;
+	int mtt_base = qp_get_mtt_addr(qpc) / dev->caps.mtt_entry_sz;
+	int mtt_size = qp_get_mtt_size(qpc);
+	struct res_cq *rcq;
+	struct res_cq *scq;
+	int rcqn = qp_get_rcqn(qpc);
+	int scqn = qp_get_scqn(qpc);
+	u32 srqn = qp_get_srqn(qpc) & 0xffffff;
+	int use_srq = (qp_get_srqn(qpc) >> 24) & 1;
+	struct res_srq *srq;
+	int local_qpn = vhcr->in_modifier & 0xffffff;
+
+	err = adjust_qp_sched_queue(dev, slave, qpc, inbox);
+	if (err)
+		return err;
+
+	err = qp_res_start_move_to(dev, slave, qpn, RES_QP_HW, &qp, 0);
+	if (err)
+		return err;
+	qp->local_qpn = local_qpn;
+	qp->sched_queue = 0;
+	qp->param3 = 0;
+	qp->vlan_control = 0;
+	qp->fvl_rx = 0;
+	qp->pri_path_fl = 0;
+	qp->vlan_index = 0;
+	qp->feup = 0;
+	qp->qpc_flags = be32_to_cpu(qpc->flags);
+
+	err = get_res(dev, slave, mtt_base, RES_MTT, &mtt);
+	if (err)
+		goto ex_abort;
+
+	err = check_mtt_range(dev, slave, mtt_base, mtt_size, mtt);
+	if (err)
+		goto ex_put_mtt;
+
+	err = get_res(dev, slave, rcqn, RES_CQ, &rcq);
+	if (err)
+		goto ex_put_mtt;
+
+	if (scqn != rcqn) {
+		err = get_res(dev, slave, scqn, RES_CQ, &scq);
+		if (err)
+			goto ex_put_rcq;
+	} else
+		scq = rcq;
+
+	if (use_srq) {
+		err = get_res(dev, slave, srqn, RES_SRQ, &srq);
+		if (err)
+			goto ex_put_scq;
+	}
+
+	adjust_proxy_tun_qkey(dev, vhcr, qpc);
+	update_pkey_index(dev, slave, inbox);
+	err = mlx4_DMA_wrapper(dev, slave, vhcr, inbox, outbox, cmd);
+	if (err)
+		goto ex_put_srq;
+	atomic_inc(&mtt->ref_count);
+	qp->mtt = mtt;
+	atomic_inc(&rcq->ref_count);
+	qp->rcq = rcq;
+	atomic_inc(&scq->ref_count);
+	qp->scq = scq;
+
+	if (scqn != rcqn)
+		put_res(dev, slave, scqn, RES_CQ);
+
+	if (use_srq) {
+		atomic_inc(&srq->ref_count);
+		put_res(dev, slave, srqn, RES_SRQ);
+		qp->srq = srq;
+	}
+
+	/* Save param3 for dynamic changes from VST back to VGT */
+	qp->param3 = qpc->param3;
+	put_res(dev, slave, rcqn, RES_CQ);
+	put_res(dev, slave, mtt_base, RES_MTT);
+	res_end_move(dev, slave, RES_QP, qpn);
+
+	return 0;
+
+ex_put_srq:
+	if (use_srq)
+		put_res(dev, slave, srqn, RES_SRQ);
+ex_put_scq:
+	if (scqn != rcqn)
+		put_res(dev, slave, scqn, RES_CQ);
+ex_put_rcq:
+	put_res(dev, slave, rcqn, RES_CQ);
+ex_put_mtt:
+	put_res(dev, slave, mtt_base, RES_MTT);
+ex_abort:
+	res_abort_move(dev, slave, RES_QP, qpn);
+
+	return err;
+}
+
+static int eq_get_mtt_addr(struct mlx4_eq_context *eqc)
+{
+	return be32_to_cpu(eqc->mtt_base_addr_l) & 0xfffffff8;
+}
+
+static int eq_get_mtt_size(struct mlx4_eq_context *eqc)
+{
+	int log_eq_size = eqc->log_eq_size & 0x1f;
+	int page_shift = (eqc->log_page_size & 0x3f) + 12;
+
+	if (log_eq_size + 5 < page_shift)
+		return 1;
+
+	return 1 << (log_eq_size + 5 - page_shift);
+}
+
+static int cq_get_mtt_addr(struct mlx4_cq_context *cqc)
+{
+	return be32_to_cpu(cqc->mtt_base_addr_l) & 0xfffffff8;
+}
+
+static int cq_get_mtt_size(struct mlx4_cq_context *cqc)
+{
+	int log_cq_size = (be32_to_cpu(cqc->logsize_usrpage) >> 24) & 0x1f;
+	int page_shift = (cqc->log_page_size & 0x3f) + 12;
+
+	if (log_cq_size + 5 < page_shift)
+		return 1;
+
+	return 1 << (log_cq_size + 5 - page_shift);
+}
+
+int mlx4_SW2HW_EQ_wrapper(struct mlx4_dev *dev, int slave,
+			  struct mlx4_vhcr *vhcr,
+			  struct mlx4_cmd_mailbox *inbox,
+			  struct mlx4_cmd_mailbox *outbox,
+			  struct mlx4_cmd_info *cmd)
+{
+	int err;
+	int eqn = vhcr->in_modifier;
+	int res_id = (slave << 10) | eqn;
+	struct mlx4_eq_context *eqc = inbox->buf;
+	int mtt_base = eq_get_mtt_addr(eqc) / dev->caps.mtt_entry_sz;
+	int mtt_size = eq_get_mtt_size(eqc);
+	struct res_eq *eq;
+	struct res_mtt *mtt;
+
+	err = add_res_range(dev, slave, res_id, 1, RES_EQ, 0);
+	if (err)
+		return err;
+	err = eq_res_start_move_to(dev, slave, res_id, RES_EQ_HW, &eq);
+	if (err)
+		goto out_add;
+
+	err = get_res(dev, slave, mtt_base, RES_MTT, &mtt);
+	if (err)
+		goto out_move;
+
+	err = check_mtt_range(dev, slave, mtt_base, mtt_size, mtt);
+	if (err)
+		goto out_put;
+
+	err = mlx4_DMA_wrapper(dev, slave, vhcr, inbox, outbox, cmd);
+	if (err)
+		goto out_put;
+
+	atomic_inc(&mtt->ref_count);
+	eq->mtt = mtt;
+	put_res(dev, slave, mtt->com.res_id, RES_MTT);
+	res_end_move(dev, slave, RES_EQ, res_id);
+	return 0;
+
+out_put:
+	put_res(dev, slave, mtt->com.res_id, RES_MTT);
+out_move:
+	res_abort_move(dev, slave, RES_EQ, res_id);
+out_add:
+	rem_res_range(dev, slave, res_id, 1, RES_EQ, 0);
+	return err;
+}
+
+int mlx4_CONFIG_DEV_wrapper(struct mlx4_dev *dev, int slave,
+			    struct mlx4_vhcr *vhcr,
+			    struct mlx4_cmd_mailbox *inbox,
+			    struct mlx4_cmd_mailbox *outbox,
+			    struct mlx4_cmd_info *cmd)
+{
+	int err;
+	u8 get = vhcr->op_modifier;
+
+	if (get != 1)
+		return -EPERM;
+
+	err = mlx4_DMA_wrapper(dev, slave, vhcr, inbox, outbox, cmd);
+
+	return err;
+}
+
+static int get_containing_mtt(struct mlx4_dev *dev, int slave, int start,
+			      int len, struct res_mtt **res)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_resource_tracker *tracker = &priv->mfunc.master.res_tracker;
+	struct res_mtt *mtt;
+	int err = -EINVAL;
+
+	spin_lock_irq(mlx4_tlock(dev));
+	list_for_each_entry(mtt, &tracker->slave_list[slave].res_list[RES_MTT],
+			    com.list) {
+		if (!check_mtt_range(dev, slave, start, len, mtt)) {
+			*res = mtt;
+			mtt->com.from_state = mtt->com.state;
+			mtt->com.state = RES_MTT_BUSY;
+			err = 0;
+			break;
+		}
+	}
+	spin_unlock_irq(mlx4_tlock(dev));
+
+	return err;
+}
+
+static int verify_qp_parameters(struct mlx4_dev *dev,
+				struct mlx4_vhcr *vhcr,
+				struct mlx4_cmd_mailbox *inbox,
+				enum qp_transition transition, u8 slave)
+{
+	u32			qp_type;
+	u32			qpn;
+	struct mlx4_qp_context	*qp_ctx;
+	enum mlx4_qp_optpar	optpar;
+	int port;
+	int num_gids;
+
+	qp_ctx  = inbox->buf + 8;
+	qp_type	= (be32_to_cpu(qp_ctx->flags) >> 16) & 0xff;
+	optpar	= be32_to_cpu(*(__be32 *) inbox->buf);
+
+	if (slave != mlx4_master_func_num(dev)) {
+		qp_ctx->params2 &= ~cpu_to_be32(MLX4_QP_BIT_FPP);
+		/* setting QP rate-limit is disallowed for VFs */
+		if (qp_ctx->rate_limit_params)
+			return -EPERM;
+	}
+
+	switch (qp_type) {
+	case MLX4_QP_ST_RC:
+	case MLX4_QP_ST_XRC:
+	case MLX4_QP_ST_UC:
+		switch (transition) {
+		case QP_TRANS_INIT2RTR:
+		case QP_TRANS_RTR2RTS:
+		case QP_TRANS_RTS2RTS:
+		case QP_TRANS_SQD2SQD:
+		case QP_TRANS_SQD2RTS:
+			if (slave != mlx4_master_func_num(dev)) {
+				if (optpar & MLX4_QP_OPTPAR_PRIMARY_ADDR_PATH) {
+					port = (qp_ctx->pri_path.sched_queue >> 6 & 1) + 1;
+					if (dev->caps.port_mask[port] != MLX4_PORT_TYPE_IB)
+						num_gids = mlx4_get_slave_num_gids(dev, slave, port);
+					else
+						num_gids = 1;
+					if (qp_ctx->pri_path.mgid_index >= num_gids)
+						return -EINVAL;
+				}
+				if (optpar & MLX4_QP_OPTPAR_ALT_ADDR_PATH) {
+					port = (qp_ctx->alt_path.sched_queue >> 6 & 1) + 1;
+					if (dev->caps.port_mask[port] != MLX4_PORT_TYPE_IB)
+						num_gids = mlx4_get_slave_num_gids(dev, slave, port);
+					else
+						num_gids = 1;
+					if (qp_ctx->alt_path.mgid_index >= num_gids)
+						return -EINVAL;
+				}
+			}
+			break;
+		default:
+			break;
+		}
+		break;
+
+	case MLX4_QP_ST_MLX:
+		qpn = vhcr->in_modifier & 0x7fffff;
+		port = (qp_ctx->pri_path.sched_queue >> 6 & 1) + 1;
+		if (transition == QP_TRANS_INIT2RTR &&
+		    slave != mlx4_master_func_num(dev) &&
+		    mlx4_is_qp_reserved(dev, qpn) &&
+		    !mlx4_vf_smi_enabled(dev, slave, port)) {
+			/* only enabled VFs may create MLX proxy QPs */
+			mlx4_err(dev, "%s: unprivileged slave %d attempting to create an MLX proxy special QP on port %d\n",
+				 __func__, slave, port);
+			return -EPERM;
+		}
+		break;
+
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+int mlx4_WRITE_MTT_wrapper(struct mlx4_dev *dev, int slave,
+			   struct mlx4_vhcr *vhcr,
+			   struct mlx4_cmd_mailbox *inbox,
+			   struct mlx4_cmd_mailbox *outbox,
+			   struct mlx4_cmd_info *cmd)
+{
+	struct mlx4_mtt mtt;
+	__be64 *page_list = inbox->buf;
+	u64 *pg_list = (u64 *)page_list;
+	int i;
+	struct res_mtt *rmtt = NULL;
+	int start = be64_to_cpu(page_list[0]);
+	int npages = vhcr->in_modifier;
+	int err;
+
+	err = get_containing_mtt(dev, slave, start, npages, &rmtt);
+	if (err)
+		return err;
+
+	/* Call the SW implementation of write_mtt:
+	 * - Prepare a dummy mtt struct
+	 * - Translate inbox contents to simple addresses in host endianness */
+	mtt.offset = 0;  /* TBD this is broken but I don't handle it since
+			    we don't really use it */
+	mtt.order = 0;
+	mtt.page_shift = 0;
+	for (i = 0; i < npages; ++i)
+		pg_list[i + 2] = (be64_to_cpu(page_list[i + 2]) & ~1ULL);
+
+	err = __mlx4_write_mtt(dev, &mtt, be64_to_cpu(page_list[0]), npages,
+			       ((u64 *)page_list + 2));
+
+	if (rmtt)
+		put_res(dev, slave, rmtt->com.res_id, RES_MTT);
+
+	return err;
+}
+
+int mlx4_HW2SW_EQ_wrapper(struct mlx4_dev *dev, int slave,
+			  struct mlx4_vhcr *vhcr,
+			  struct mlx4_cmd_mailbox *inbox,
+			  struct mlx4_cmd_mailbox *outbox,
+			  struct mlx4_cmd_info *cmd)
+{
+	int eqn = vhcr->in_modifier;
+	int res_id = eqn | (slave << 10);
+	struct res_eq *eq;
+	int err;
+
+	err = eq_res_start_move_to(dev, slave, res_id, RES_EQ_RESERVED, &eq);
+	if (err)
+		return err;
+
+	err = get_res(dev, slave, eq->mtt->com.res_id, RES_MTT, NULL);
+	if (err)
+		goto ex_abort;
+
+	err = mlx4_DMA_wrapper(dev, slave, vhcr, inbox, outbox, cmd);
+	if (err)
+		goto ex_put;
+
+	atomic_dec(&eq->mtt->ref_count);
+	put_res(dev, slave, eq->mtt->com.res_id, RES_MTT);
+	res_end_move(dev, slave, RES_EQ, res_id);
+	rem_res_range(dev, slave, res_id, 1, RES_EQ, 0);
+
+	return 0;
+
+ex_put:
+	put_res(dev, slave, eq->mtt->com.res_id, RES_MTT);
+ex_abort:
+	res_abort_move(dev, slave, RES_EQ, res_id);
+
+	return err;
+}
+
+int mlx4_GEN_EQE(struct mlx4_dev *dev, int slave, struct mlx4_eqe *eqe)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_slave_event_eq_info *event_eq;
+	struct mlx4_cmd_mailbox *mailbox;
+	u32 in_modifier = 0;
+	int err;
+	int res_id;
+	struct res_eq *req;
+
+	if (!priv->mfunc.master.slave_state)
+		return -EINVAL;
+
+	/* check for slave valid, slave not PF, and slave active */
+	if (slave < 0 || slave > dev->persist->num_vfs ||
+	    slave == dev->caps.function ||
+	    !priv->mfunc.master.slave_state[slave].active)
+		return 0;
+
+	event_eq = &priv->mfunc.master.slave_state[slave].event_eq[eqe->type];
+
+	/* Create the event only if the slave is registered */
+	if (event_eq->eqn < 0)
+		return 0;
+
+	mutex_lock(&priv->mfunc.master.gen_eqe_mutex[slave]);
+	res_id = (slave << 10) | event_eq->eqn;
+	err = get_res(dev, slave, res_id, RES_EQ, &req);
+	if (err)
+		goto unlock;
+
+	if (req->com.from_state != RES_EQ_HW) {
+		err = -EINVAL;
+		goto put;
+	}
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox)) {
+		err = PTR_ERR(mailbox);
+		goto put;
+	}
+
+	if (eqe->type == MLX4_EVENT_TYPE_CMD) {
+		++event_eq->token;
+		eqe->event.cmd.token = cpu_to_be16(event_eq->token);
+	}
+
+	memcpy(mailbox->buf, (u8 *) eqe, 28);
+
+	in_modifier = (slave & 0xff) | ((event_eq->eqn & 0x3ff) << 16);
+
+	err = mlx4_cmd(dev, mailbox->dma, in_modifier, 0,
+		       MLX4_CMD_GEN_EQE, MLX4_CMD_TIME_CLASS_B,
+		       MLX4_CMD_NATIVE);
+
+	put_res(dev, slave, res_id, RES_EQ);
+	mutex_unlock(&priv->mfunc.master.gen_eqe_mutex[slave]);
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return err;
+
+put:
+	put_res(dev, slave, res_id, RES_EQ);
+
+unlock:
+	mutex_unlock(&priv->mfunc.master.gen_eqe_mutex[slave]);
+	return err;
+}
+
+int mlx4_QUERY_EQ_wrapper(struct mlx4_dev *dev, int slave,
+			  struct mlx4_vhcr *vhcr,
+			  struct mlx4_cmd_mailbox *inbox,
+			  struct mlx4_cmd_mailbox *outbox,
+			  struct mlx4_cmd_info *cmd)
+{
+	int eqn = vhcr->in_modifier;
+	int res_id = eqn | (slave << 10);
+	struct res_eq *eq;
+	int err;
+
+	err = get_res(dev, slave, res_id, RES_EQ, &eq);
+	if (err)
+		return err;
+
+	if (eq->com.from_state != RES_EQ_HW) {
+		err = -EINVAL;
+		goto ex_put;
+	}
+
+	err = mlx4_DMA_wrapper(dev, slave, vhcr, inbox, outbox, cmd);
+
+ex_put:
+	put_res(dev, slave, res_id, RES_EQ);
+	return err;
+}
+
+int mlx4_SW2HW_CQ_wrapper(struct mlx4_dev *dev, int slave,
+			  struct mlx4_vhcr *vhcr,
+			  struct mlx4_cmd_mailbox *inbox,
+			  struct mlx4_cmd_mailbox *outbox,
+			  struct mlx4_cmd_info *cmd)
+{
+	int err;
+	int cqn = vhcr->in_modifier;
+	struct mlx4_cq_context *cqc = inbox->buf;
+	int mtt_base = cq_get_mtt_addr(cqc) / dev->caps.mtt_entry_sz;
+	struct res_cq *cq = NULL;
+	struct res_mtt *mtt;
+
+	err = cq_res_start_move_to(dev, slave, cqn, RES_CQ_HW, &cq);
+	if (err)
+		return err;
+	err = get_res(dev, slave, mtt_base, RES_MTT, &mtt);
+	if (err)
+		goto out_move;
+	err = check_mtt_range(dev, slave, mtt_base, cq_get_mtt_size(cqc), mtt);
+	if (err)
+		goto out_put;
+	err = mlx4_DMA_wrapper(dev, slave, vhcr, inbox, outbox, cmd);
+	if (err)
+		goto out_put;
+	atomic_inc(&mtt->ref_count);
+	cq->mtt = mtt;
+	put_res(dev, slave, mtt->com.res_id, RES_MTT);
+	res_end_move(dev, slave, RES_CQ, cqn);
+	return 0;
+
+out_put:
+	put_res(dev, slave, mtt->com.res_id, RES_MTT);
+out_move:
+	res_abort_move(dev, slave, RES_CQ, cqn);
+	return err;
+}
+
+int mlx4_HW2SW_CQ_wrapper(struct mlx4_dev *dev, int slave,
+			  struct mlx4_vhcr *vhcr,
+			  struct mlx4_cmd_mailbox *inbox,
+			  struct mlx4_cmd_mailbox *outbox,
+			  struct mlx4_cmd_info *cmd)
+{
+	int err;
+	int cqn = vhcr->in_modifier;
+	struct res_cq *cq = NULL;
+
+	err = cq_res_start_move_to(dev, slave, cqn, RES_CQ_ALLOCATED, &cq);
+	if (err)
+		return err;
+	err = mlx4_DMA_wrapper(dev, slave, vhcr, inbox, outbox, cmd);
+	if (err)
+		goto out_move;
+	atomic_dec(&cq->mtt->ref_count);
+	res_end_move(dev, slave, RES_CQ, cqn);
+	return 0;
+
+out_move:
+	res_abort_move(dev, slave, RES_CQ, cqn);
+	return err;
+}
+
+int mlx4_QUERY_CQ_wrapper(struct mlx4_dev *dev, int slave,
+			  struct mlx4_vhcr *vhcr,
+			  struct mlx4_cmd_mailbox *inbox,
+			  struct mlx4_cmd_mailbox *outbox,
+			  struct mlx4_cmd_info *cmd)
+{
+	int cqn = vhcr->in_modifier;
+	struct res_cq *cq;
+	int err;
+
+	err = get_res(dev, slave, cqn, RES_CQ, &cq);
+	if (err)
+		return err;
+
+	if (cq->com.from_state != RES_CQ_HW)
+		goto ex_put;
+
+	err = mlx4_DMA_wrapper(dev, slave, vhcr, inbox, outbox, cmd);
+ex_put:
+	put_res(dev, slave, cqn, RES_CQ);
+
+	return err;
+}
+
+static int handle_resize(struct mlx4_dev *dev, int slave,
+			 struct mlx4_vhcr *vhcr,
+			 struct mlx4_cmd_mailbox *inbox,
+			 struct mlx4_cmd_mailbox *outbox,
+			 struct mlx4_cmd_info *cmd,
+			 struct res_cq *cq)
+{
+	int err;
+	struct res_mtt *orig_mtt;
+	struct res_mtt *mtt;
+	struct mlx4_cq_context *cqc = inbox->buf;
+	int mtt_base = cq_get_mtt_addr(cqc) / dev->caps.mtt_entry_sz;
+
+	err = get_res(dev, slave, cq->mtt->com.res_id, RES_MTT, &orig_mtt);
+	if (err)
+		return err;
+
+	if (orig_mtt != cq->mtt) {
+		err = -EINVAL;
+		goto ex_put;
+	}
+
+	err = get_res(dev, slave, mtt_base, RES_MTT, &mtt);
+	if (err)
+		goto ex_put;
+
+	err = check_mtt_range(dev, slave, mtt_base, cq_get_mtt_size(cqc), mtt);
+	if (err)
+		goto ex_put1;
+	err = mlx4_DMA_wrapper(dev, slave, vhcr, inbox, outbox, cmd);
+	if (err)
+		goto ex_put1;
+	atomic_dec(&orig_mtt->ref_count);
+	put_res(dev, slave, orig_mtt->com.res_id, RES_MTT);
+	atomic_inc(&mtt->ref_count);
+	cq->mtt = mtt;
+	put_res(dev, slave, mtt->com.res_id, RES_MTT);
+	return 0;
+
+ex_put1:
+	put_res(dev, slave, mtt->com.res_id, RES_MTT);
+ex_put:
+	put_res(dev, slave, orig_mtt->com.res_id, RES_MTT);
+
+	return err;
+
+}
+
+int mlx4_MODIFY_CQ_wrapper(struct mlx4_dev *dev, int slave,
+			   struct mlx4_vhcr *vhcr,
+			   struct mlx4_cmd_mailbox *inbox,
+			   struct mlx4_cmd_mailbox *outbox,
+			   struct mlx4_cmd_info *cmd)
+{
+	int cqn = vhcr->in_modifier;
+	struct res_cq *cq;
+	int err;
+
+	err = get_res(dev, slave, cqn, RES_CQ, &cq);
+	if (err)
+		return err;
+
+	if (cq->com.from_state != RES_CQ_HW)
+		goto ex_put;
+
+	if (vhcr->op_modifier == 0) {
+		err = handle_resize(dev, slave, vhcr, inbox, outbox, cmd, cq);
+		goto ex_put;
+	}
+
+	err = mlx4_DMA_wrapper(dev, slave, vhcr, inbox, outbox, cmd);
+ex_put:
+	put_res(dev, slave, cqn, RES_CQ);
+
+	return err;
+}
+
+static int srq_get_mtt_size(struct mlx4_srq_context *srqc)
+{
+	int log_srq_size = (be32_to_cpu(srqc->state_logsize_srqn) >> 24) & 0xf;
+	int log_rq_stride = srqc->logstride & 7;
+	int page_shift = (srqc->log_page_size & 0x3f) + 12;
+
+	if (log_srq_size + log_rq_stride + 4 < page_shift)
+		return 1;
+
+	return 1 << (log_srq_size + log_rq_stride + 4 - page_shift);
+}
+
+int mlx4_SW2HW_SRQ_wrapper(struct mlx4_dev *dev, int slave,
+			   struct mlx4_vhcr *vhcr,
+			   struct mlx4_cmd_mailbox *inbox,
+			   struct mlx4_cmd_mailbox *outbox,
+			   struct mlx4_cmd_info *cmd)
+{
+	int err;
+	int srqn = vhcr->in_modifier;
+	struct res_mtt *mtt;
+	struct res_srq *srq = NULL;
+	struct mlx4_srq_context *srqc = inbox->buf;
+	int mtt_base = srq_get_mtt_addr(srqc) / dev->caps.mtt_entry_sz;
+
+	if (srqn != (be32_to_cpu(srqc->state_logsize_srqn) & 0xffffff))
+		return -EINVAL;
+
+	err = srq_res_start_move_to(dev, slave, srqn, RES_SRQ_HW, &srq);
+	if (err)
+		return err;
+	err = get_res(dev, slave, mtt_base, RES_MTT, &mtt);
+	if (err)
+		goto ex_abort;
+	err = check_mtt_range(dev, slave, mtt_base, srq_get_mtt_size(srqc),
+			      mtt);
+	if (err)
+		goto ex_put_mtt;
+
+	err = mlx4_DMA_wrapper(dev, slave, vhcr, inbox, outbox, cmd);
+	if (err)
+		goto ex_put_mtt;
+
+	atomic_inc(&mtt->ref_count);
+	srq->mtt = mtt;
+	put_res(dev, slave, mtt->com.res_id, RES_MTT);
+	res_end_move(dev, slave, RES_SRQ, srqn);
+	return 0;
+
+ex_put_mtt:
+	put_res(dev, slave, mtt->com.res_id, RES_MTT);
+ex_abort:
+	res_abort_move(dev, slave, RES_SRQ, srqn);
+
+	return err;
+}
+
+int mlx4_HW2SW_SRQ_wrapper(struct mlx4_dev *dev, int slave,
+			   struct mlx4_vhcr *vhcr,
+			   struct mlx4_cmd_mailbox *inbox,
+			   struct mlx4_cmd_mailbox *outbox,
+			   struct mlx4_cmd_info *cmd)
+{
+	int err;
+	int srqn = vhcr->in_modifier;
+	struct res_srq *srq = NULL;
+
+	err = srq_res_start_move_to(dev, slave, srqn, RES_SRQ_ALLOCATED, &srq);
+	if (err)
+		return err;
+	err = mlx4_DMA_wrapper(dev, slave, vhcr, inbox, outbox, cmd);
+	if (err)
+		goto ex_abort;
+	atomic_dec(&srq->mtt->ref_count);
+	if (srq->cq)
+		atomic_dec(&srq->cq->ref_count);
+	res_end_move(dev, slave, RES_SRQ, srqn);
+
+	return 0;
+
+ex_abort:
+	res_abort_move(dev, slave, RES_SRQ, srqn);
+
+	return err;
+}
+
+int mlx4_QUERY_SRQ_wrapper(struct mlx4_dev *dev, int slave,
+			   struct mlx4_vhcr *vhcr,
+			   struct mlx4_cmd_mailbox *inbox,
+			   struct mlx4_cmd_mailbox *outbox,
+			   struct mlx4_cmd_info *cmd)
+{
+	int err;
+	int srqn = vhcr->in_modifier;
+	struct res_srq *srq;
+
+	err = get_res(dev, slave, srqn, RES_SRQ, &srq);
+	if (err)
+		return err;
+	if (srq->com.from_state != RES_SRQ_HW) {
+		err = -EBUSY;
+		goto out;
+	}
+	err = mlx4_DMA_wrapper(dev, slave, vhcr, inbox, outbox, cmd);
+out:
+	put_res(dev, slave, srqn, RES_SRQ);
+	return err;
+}
+
+int mlx4_ARM_SRQ_wrapper(struct mlx4_dev *dev, int slave,
+			 struct mlx4_vhcr *vhcr,
+			 struct mlx4_cmd_mailbox *inbox,
+			 struct mlx4_cmd_mailbox *outbox,
+			 struct mlx4_cmd_info *cmd)
+{
+	int err;
+	int srqn = vhcr->in_modifier;
+	struct res_srq *srq;
+
+	err = get_res(dev, slave, srqn, RES_SRQ, &srq);
+	if (err)
+		return err;
+
+	if (srq->com.from_state != RES_SRQ_HW) {
+		err = -EBUSY;
+		goto out;
+	}
+
+	err = mlx4_DMA_wrapper(dev, slave, vhcr, inbox, outbox, cmd);
+out:
+	put_res(dev, slave, srqn, RES_SRQ);
+	return err;
+}
+
+int mlx4_GEN_QP_wrapper(struct mlx4_dev *dev, int slave,
+			struct mlx4_vhcr *vhcr,
+			struct mlx4_cmd_mailbox *inbox,
+			struct mlx4_cmd_mailbox *outbox,
+			struct mlx4_cmd_info *cmd)
+{
+	int err;
+	int qpn = vhcr->in_modifier & 0x7fffff;
+	struct res_qp *qp;
+
+	err = get_res(dev, slave, qpn, RES_QP, &qp);
+	if (err)
+		return err;
+	if (qp->com.from_state != RES_QP_HW) {
+		err = -EBUSY;
+		goto out;
+	}
+
+	err = mlx4_DMA_wrapper(dev, slave, vhcr, inbox, outbox, cmd);
+out:
+	put_res(dev, slave, qpn, RES_QP);
+	return err;
+}
+
+int mlx4_INIT2INIT_QP_wrapper(struct mlx4_dev *dev, int slave,
+			      struct mlx4_vhcr *vhcr,
+			      struct mlx4_cmd_mailbox *inbox,
+			      struct mlx4_cmd_mailbox *outbox,
+			      struct mlx4_cmd_info *cmd)
+{
+	struct mlx4_qp_context *context = inbox->buf + 8;
+	adjust_proxy_tun_qkey(dev, vhcr, context);
+	update_pkey_index(dev, slave, inbox);
+	return mlx4_GEN_QP_wrapper(dev, slave, vhcr, inbox, outbox, cmd);
+}
+
+static int adjust_qp_sched_queue(struct mlx4_dev *dev, int slave,
+				  struct mlx4_qp_context *qpc,
+				  struct mlx4_cmd_mailbox *inbox)
+{
+	enum mlx4_qp_optpar optpar = be32_to_cpu(*(__be32 *)inbox->buf);
+	u8 pri_sched_queue;
+	int port = mlx4_slave_convert_port(
+		   dev, slave, (qpc->pri_path.sched_queue >> 6 & 1) + 1) - 1;
+
+	if (port < 0)
+		return -EINVAL;
+
+	pri_sched_queue = (qpc->pri_path.sched_queue & ~(1 << 6)) |
+			  ((port & 1) << 6);
+
+	if (optpar & (MLX4_QP_OPTPAR_PRIMARY_ADDR_PATH | MLX4_QP_OPTPAR_SCHED_QUEUE) ||
+	    qpc->pri_path.sched_queue || mlx4_is_eth(dev, port + 1)) {
+		qpc->pri_path.sched_queue = pri_sched_queue;
+	}
+
+	if (optpar & MLX4_QP_OPTPAR_ALT_ADDR_PATH) {
+		port = mlx4_slave_convert_port(
+				dev, slave, (qpc->alt_path.sched_queue >> 6 & 1)
+				+ 1) - 1;
+		if (port < 0)
+			return -EINVAL;
+		qpc->alt_path.sched_queue =
+			(qpc->alt_path.sched_queue & ~(1 << 6)) |
+			(port & 1) << 6;
+	}
+	return 0;
+}
+
+static int roce_verify_mac(struct mlx4_dev *dev, int slave,
+				struct mlx4_qp_context *qpc,
+				struct mlx4_cmd_mailbox *inbox)
+{
+	u64 mac;
+	int port;
+	u32 ts = (be32_to_cpu(qpc->flags) >> 16) & 0xff;
+	u8 sched = *(u8 *)(inbox->buf + 64);
+	u8 smac_ix;
+
+	port = (sched >> 6 & 1) + 1;
+	if (mlx4_is_eth(dev, port) && (ts != MLX4_QP_ST_MLX)) {
+		smac_ix = qpc->pri_path.grh_mylmc & 0x7f;
+		if (mac_find_smac_ix_in_slave(dev, slave, port, smac_ix, &mac))
+			return -ENOENT;
+	}
+	return 0;
+}
+
+int mlx4_INIT2RTR_QP_wrapper(struct mlx4_dev *dev, int slave,
+			     struct mlx4_vhcr *vhcr,
+			     struct mlx4_cmd_mailbox *inbox,
+			     struct mlx4_cmd_mailbox *outbox,
+			     struct mlx4_cmd_info *cmd)
+{
+	int err;
+	struct mlx4_qp_context *qpc = inbox->buf + 8;
+	int qpn = vhcr->in_modifier & 0x7fffff;
+	struct res_qp *qp;
+	u8 orig_sched_queue;
+	u8 orig_vlan_control = qpc->pri_path.vlan_control;
+	u8 orig_fvl_rx = qpc->pri_path.fvl_rx;
+	u8 orig_pri_path_fl = qpc->pri_path.fl;
+	u8 orig_vlan_index = qpc->pri_path.vlan_index;
+	u8 orig_feup = qpc->pri_path.feup;
+
+	err = adjust_qp_sched_queue(dev, slave, qpc, inbox);
+	if (err)
+		return err;
+	err = verify_qp_parameters(dev, vhcr, inbox, QP_TRANS_INIT2RTR, slave);
+	if (err)
+		return err;
+
+	if (roce_verify_mac(dev, slave, qpc, inbox))
+		return -EINVAL;
+
+	update_pkey_index(dev, slave, inbox);
+	update_gid(dev, inbox, (u8)slave);
+	adjust_proxy_tun_qkey(dev, vhcr, qpc);
+	orig_sched_queue = qpc->pri_path.sched_queue;
+
+	err = get_res(dev, slave, qpn, RES_QP, &qp);
+	if (err)
+		return err;
+	if (qp->com.from_state != RES_QP_HW) {
+		err = -EBUSY;
+		goto out;
+	}
+
+	err = update_vport_qp_param(dev, inbox, slave, qpn);
+	if (err)
+		goto out;
+
+	err = mlx4_DMA_wrapper(dev, slave, vhcr, inbox, outbox, cmd);
+out:
+	/* if no error, save sched queue value passed in by VF. This is
+	 * essentially the QOS value provided by the VF. This will be useful
+	 * if we allow dynamic changes from VST back to VGT
+	 */
+	if (!err) {
+		qp->sched_queue = orig_sched_queue;
+		qp->vlan_control = orig_vlan_control;
+		qp->fvl_rx	=  orig_fvl_rx;
+		qp->pri_path_fl = orig_pri_path_fl;
+		qp->vlan_index  = orig_vlan_index;
+		qp->feup	= orig_feup;
+	}
+	put_res(dev, slave, qpn, RES_QP);
+	return err;
+}
+
+int mlx4_RTR2RTS_QP_wrapper(struct mlx4_dev *dev, int slave,
+			    struct mlx4_vhcr *vhcr,
+			    struct mlx4_cmd_mailbox *inbox,
+			    struct mlx4_cmd_mailbox *outbox,
+			    struct mlx4_cmd_info *cmd)
+{
+	int err;
+	struct mlx4_qp_context *context = inbox->buf + 8;
+
+	err = adjust_qp_sched_queue(dev, slave, context, inbox);
+	if (err)
+		return err;
+	err = verify_qp_parameters(dev, vhcr, inbox, QP_TRANS_RTR2RTS, slave);
+	if (err)
+		return err;
+
+	update_pkey_index(dev, slave, inbox);
+	update_gid(dev, inbox, (u8)slave);
+	adjust_proxy_tun_qkey(dev, vhcr, context);
+	return mlx4_GEN_QP_wrapper(dev, slave, vhcr, inbox, outbox, cmd);
+}
+
+int mlx4_RTS2RTS_QP_wrapper(struct mlx4_dev *dev, int slave,
+			    struct mlx4_vhcr *vhcr,
+			    struct mlx4_cmd_mailbox *inbox,
+			    struct mlx4_cmd_mailbox *outbox,
+			    struct mlx4_cmd_info *cmd)
+{
+	int err;
+	struct mlx4_qp_context *context = inbox->buf + 8;
+
+	err = adjust_qp_sched_queue(dev, slave, context, inbox);
+	if (err)
+		return err;
+	err = verify_qp_parameters(dev, vhcr, inbox, QP_TRANS_RTS2RTS, slave);
+	if (err)
+		return err;
+
+	update_pkey_index(dev, slave, inbox);
+	update_gid(dev, inbox, (u8)slave);
+	adjust_proxy_tun_qkey(dev, vhcr, context);
+	return mlx4_GEN_QP_wrapper(dev, slave, vhcr, inbox, outbox, cmd);
+}
+
+
+int mlx4_SQERR2RTS_QP_wrapper(struct mlx4_dev *dev, int slave,
+			      struct mlx4_vhcr *vhcr,
+			      struct mlx4_cmd_mailbox *inbox,
+			      struct mlx4_cmd_mailbox *outbox,
+			      struct mlx4_cmd_info *cmd)
+{
+	struct mlx4_qp_context *context = inbox->buf + 8;
+	int err = adjust_qp_sched_queue(dev, slave, context, inbox);
+	if (err)
+		return err;
+	adjust_proxy_tun_qkey(dev, vhcr, context);
+	return mlx4_GEN_QP_wrapper(dev, slave, vhcr, inbox, outbox, cmd);
+}
+
+int mlx4_SQD2SQD_QP_wrapper(struct mlx4_dev *dev, int slave,
+			    struct mlx4_vhcr *vhcr,
+			    struct mlx4_cmd_mailbox *inbox,
+			    struct mlx4_cmd_mailbox *outbox,
+			    struct mlx4_cmd_info *cmd)
+{
+	int err;
+	struct mlx4_qp_context *context = inbox->buf + 8;
+
+	err = adjust_qp_sched_queue(dev, slave, context, inbox);
+	if (err)
+		return err;
+	err = verify_qp_parameters(dev, vhcr, inbox, QP_TRANS_SQD2SQD, slave);
+	if (err)
+		return err;
+
+	adjust_proxy_tun_qkey(dev, vhcr, context);
+	update_gid(dev, inbox, (u8)slave);
+	update_pkey_index(dev, slave, inbox);
+	return mlx4_GEN_QP_wrapper(dev, slave, vhcr, inbox, outbox, cmd);
+}
+
+int mlx4_SQD2RTS_QP_wrapper(struct mlx4_dev *dev, int slave,
+			    struct mlx4_vhcr *vhcr,
+			    struct mlx4_cmd_mailbox *inbox,
+			    struct mlx4_cmd_mailbox *outbox,
+			    struct mlx4_cmd_info *cmd)
+{
+	int err;
+	struct mlx4_qp_context *context = inbox->buf + 8;
+
+	err = adjust_qp_sched_queue(dev, slave, context, inbox);
+	if (err)
+		return err;
+	err = verify_qp_parameters(dev, vhcr, inbox, QP_TRANS_SQD2RTS, slave);
+	if (err)
+		return err;
+
+	adjust_proxy_tun_qkey(dev, vhcr, context);
+	update_gid(dev, inbox, (u8)slave);
+	update_pkey_index(dev, slave, inbox);
+	return mlx4_GEN_QP_wrapper(dev, slave, vhcr, inbox, outbox, cmd);
+}
+
+int mlx4_2RST_QP_wrapper(struct mlx4_dev *dev, int slave,
+			 struct mlx4_vhcr *vhcr,
+			 struct mlx4_cmd_mailbox *inbox,
+			 struct mlx4_cmd_mailbox *outbox,
+			 struct mlx4_cmd_info *cmd)
+{
+	int err;
+	int qpn = vhcr->in_modifier & 0x7fffff;
+	struct res_qp *qp;
+
+	err = qp_res_start_move_to(dev, slave, qpn, RES_QP_MAPPED, &qp, 0);
+	if (err)
+		return err;
+	err = mlx4_DMA_wrapper(dev, slave, vhcr, inbox, outbox, cmd);
+	if (err)
+		goto ex_abort;
+
+	atomic_dec(&qp->mtt->ref_count);
+	atomic_dec(&qp->rcq->ref_count);
+	atomic_dec(&qp->scq->ref_count);
+	if (qp->srq)
+		atomic_dec(&qp->srq->ref_count);
+	res_end_move(dev, slave, RES_QP, qpn);
+	return 0;
+
+ex_abort:
+	res_abort_move(dev, slave, RES_QP, qpn);
+
+	return err;
+}
+
+static struct res_gid *find_gid(struct mlx4_dev *dev, int slave,
+				struct res_qp *rqp, u8 *gid)
+{
+	struct res_gid *res;
+
+	list_for_each_entry(res, &rqp->mcg_list, list) {
+		if (!memcmp(res->gid, gid, 16))
+			return res;
+	}
+	return NULL;
+}
+
+static int add_mcg_res(struct mlx4_dev *dev, int slave, struct res_qp *rqp,
+		       u8 *gid, enum mlx4_protocol prot,
+		       enum mlx4_steer_type steer, u64 reg_id)
+{
+	struct res_gid *res;
+	int err;
+
+	res = kzalloc(sizeof(*res), GFP_KERNEL);
+	if (!res)
+		return -ENOMEM;
+
+	spin_lock_irq(&rqp->mcg_spl);
+	if (find_gid(dev, slave, rqp, gid)) {
+		kfree(res);
+		err = -EEXIST;
+	} else {
+		memcpy(res->gid, gid, 16);
+		res->prot = prot;
+		res->steer = steer;
+		res->reg_id = reg_id;
+		list_add_tail(&res->list, &rqp->mcg_list);
+		err = 0;
+	}
+	spin_unlock_irq(&rqp->mcg_spl);
+
+	return err;
+}
+
+static int rem_mcg_res(struct mlx4_dev *dev, int slave, struct res_qp *rqp,
+		       u8 *gid, enum mlx4_protocol prot,
+		       enum mlx4_steer_type steer, u64 *reg_id)
+{
+	struct res_gid *res;
+	int err;
+
+	spin_lock_irq(&rqp->mcg_spl);
+	res = find_gid(dev, slave, rqp, gid);
+	if (!res || res->prot != prot || res->steer != steer)
+		err = -EINVAL;
+	else {
+		*reg_id = res->reg_id;
+		list_del(&res->list);
+		kfree(res);
+		err = 0;
+	}
+	spin_unlock_irq(&rqp->mcg_spl);
+
+	return err;
+}
+
+static int qp_attach(struct mlx4_dev *dev, int slave, struct mlx4_qp *qp,
+		     u8 gid[16], int block_loopback, enum mlx4_protocol prot,
+		     enum mlx4_steer_type type, u64 *reg_id)
+{
+	switch (dev->caps.steering_mode) {
+	case MLX4_STEERING_MODE_DEVICE_MANAGED: {
+		int port = mlx4_slave_convert_port(dev, slave, gid[5]);
+		if (port < 0)
+			return port;
+		return mlx4_trans_to_dmfs_attach(dev, qp, gid, port,
+						block_loopback, prot,
+						reg_id);
+	}
+	case MLX4_STEERING_MODE_B0:
+		if (prot == MLX4_PROT_ETH) {
+			int port = mlx4_slave_convert_port(dev, slave, gid[5]);
+			if (port < 0)
+				return port;
+			gid[5] = port;
+		}
+		return mlx4_qp_attach_common(dev, qp, gid,
+					    block_loopback, prot, type);
+	default:
+		return -EINVAL;
+	}
+}
+
+static int qp_detach(struct mlx4_dev *dev, struct mlx4_qp *qp,
+		     u8 gid[16], enum mlx4_protocol prot,
+		     enum mlx4_steer_type type, u64 reg_id)
+{
+	switch (dev->caps.steering_mode) {
+	case MLX4_STEERING_MODE_DEVICE_MANAGED:
+		return mlx4_flow_detach(dev, reg_id);
+	case MLX4_STEERING_MODE_B0:
+		return mlx4_qp_detach_common(dev, qp, gid, prot, type);
+	default:
+		return -EINVAL;
+	}
+}
+
+static int mlx4_adjust_port(struct mlx4_dev *dev, int slave,
+			    u8 *gid, enum mlx4_protocol prot)
+{
+	int real_port;
+
+	if (prot != MLX4_PROT_ETH)
+		return 0;
+
+	if (dev->caps.steering_mode == MLX4_STEERING_MODE_B0 ||
+	    dev->caps.steering_mode == MLX4_STEERING_MODE_DEVICE_MANAGED) {
+		real_port = mlx4_slave_convert_port(dev, slave, gid[5]);
+		if (real_port < 0)
+			return -EINVAL;
+		gid[5] = real_port;
+	}
+
+	return 0;
+}
+
+int mlx4_QP_ATTACH_wrapper(struct mlx4_dev *dev, int slave,
+			       struct mlx4_vhcr *vhcr,
+			       struct mlx4_cmd_mailbox *inbox,
+			       struct mlx4_cmd_mailbox *outbox,
+			       struct mlx4_cmd_info *cmd)
+{
+	struct mlx4_qp qp; /* dummy for calling attach/detach */
+	u8 *gid = inbox->buf;
+	enum mlx4_protocol prot = (vhcr->in_modifier >> 28) & 0x7;
+	int err;
+	int qpn;
+	struct res_qp *rqp;
+	u64 reg_id = 0;
+	int attach = vhcr->op_modifier;
+	int block_loopback = vhcr->in_modifier >> 31;
+	u8 steer_type_mask = 2;
+	enum mlx4_steer_type type = (gid[7] & steer_type_mask) >> 1;
+
+	qpn = vhcr->in_modifier & 0xffffff;
+	err = get_res(dev, slave, qpn, RES_QP, &rqp);
+	if (err)
+		return err;
+
+	qp.qpn = qpn;
+	if (attach) {
+		err = qp_attach(dev, slave, &qp, gid, block_loopback, prot,
+				type, &reg_id);
+		if (err) {
+			pr_err("Fail to attach rule to qp 0x%x\n", qpn);
+			goto ex_put;
+		}
+		err = add_mcg_res(dev, slave, rqp, gid, prot, type, reg_id);
+		if (err)
+			goto ex_detach;
+	} else {
+		err = mlx4_adjust_port(dev, slave, gid, prot);
+		if (err)
+			goto ex_put;
+
+		err = rem_mcg_res(dev, slave, rqp, gid, prot, type, &reg_id);
+		if (err)
+			goto ex_put;
+
+		err = qp_detach(dev, &qp, gid, prot, type, reg_id);
+		if (err)
+			pr_err("Fail to detach rule from qp 0x%x reg_id = 0x%llx\n",
+			       qpn, reg_id);
+	}
+	put_res(dev, slave, qpn, RES_QP);
+	return err;
+
+ex_detach:
+	qp_detach(dev, &qp, gid, prot, type, reg_id);
+ex_put:
+	put_res(dev, slave, qpn, RES_QP);
+	return err;
+}
+
+/*
+ * MAC validation for Flow Steering rules.
+ * VF can attach rules only with a mac address which is assigned to it.
+ */
+static int validate_eth_header_mac(int slave, struct _rule_hw *eth_header,
+				   struct list_head *rlist)
+{
+	struct mac_res *res, *tmp;
+	__be64 be_mac;
+
+	/* make sure it isn't multicast or broadcast mac*/
+	if (!is_multicast_ether_addr(eth_header->eth.dst_mac) &&
+	    !is_broadcast_ether_addr(eth_header->eth.dst_mac)) {
+		list_for_each_entry_safe(res, tmp, rlist, list) {
+			be_mac = cpu_to_be64(res->mac << 16);
+			if (ether_addr_equal((u8 *)&be_mac, eth_header->eth.dst_mac))
+				return 0;
+		}
+		pr_err("MAC %pM doesn't belong to VF %d, Steering rule rejected\n",
+		       eth_header->eth.dst_mac, slave);
+		return -EINVAL;
+	}
+	return 0;
+}
+
+/*
+ * In case of missing eth header, append eth header with a MAC address
+ * assigned to the VF.
+ */
+static int add_eth_header(struct mlx4_dev *dev, int slave,
+			  struct mlx4_cmd_mailbox *inbox,
+			  struct list_head *rlist, int header_id)
+{
+	struct mac_res *res, *tmp;
+	u8 port;
+	struct mlx4_net_trans_rule_hw_ctrl *ctrl;
+	struct mlx4_net_trans_rule_hw_eth *eth_header;
+	struct mlx4_net_trans_rule_hw_ipv4 *ip_header;
+	struct mlx4_net_trans_rule_hw_tcp_udp *l4_header;
+	__be64 be_mac = 0;
+	__be64 mac_msk = cpu_to_be64(MLX4_MAC_MASK << 16);
+
+	ctrl = (struct mlx4_net_trans_rule_hw_ctrl *)inbox->buf;
+	port = ctrl->port;
+	eth_header = (struct mlx4_net_trans_rule_hw_eth *)(ctrl + 1);
+
+	/* Clear a space in the inbox for eth header */
+	switch (header_id) {
+	case MLX4_NET_TRANS_RULE_ID_IPV4:
+		ip_header =
+			(struct mlx4_net_trans_rule_hw_ipv4 *)(eth_header + 1);
+		memmove(ip_header, eth_header,
+			sizeof(*ip_header) + sizeof(*l4_header));
+		break;
+	case MLX4_NET_TRANS_RULE_ID_TCP:
+	case MLX4_NET_TRANS_RULE_ID_UDP:
+		l4_header = (struct mlx4_net_trans_rule_hw_tcp_udp *)
+			    (eth_header + 1);
+		memmove(l4_header, eth_header, sizeof(*l4_header));
+		break;
+	default:
+		return -EINVAL;
+	}
+	list_for_each_entry_safe(res, tmp, rlist, list) {
+		if (port == res->port) {
+			be_mac = cpu_to_be64(res->mac << 16);
+			break;
+		}
+	}
+	if (!be_mac) {
+		pr_err("Failed adding eth header to FS rule, Can't find matching MAC for port %d\n",
+		       port);
+		return -EINVAL;
+	}
+
+	memset(eth_header, 0, sizeof(*eth_header));
+	eth_header->size = sizeof(*eth_header) >> 2;
+	eth_header->id = cpu_to_be16(__sw_id_hw[MLX4_NET_TRANS_RULE_ID_ETH]);
+	memcpy(eth_header->dst_mac, &be_mac, ETH_ALEN);
+	memcpy(eth_header->dst_mac_msk, &mac_msk, ETH_ALEN);
+
+	return 0;
+
+}
+
+#define MLX4_UPD_QP_PATH_MASK_SUPPORTED      (                                \
+	1ULL << MLX4_UPD_QP_PATH_MASK_MAC_INDEX                     |\
+	1ULL << MLX4_UPD_QP_PATH_MASK_ETH_SRC_CHECK_MC_LB)
+int mlx4_UPDATE_QP_wrapper(struct mlx4_dev *dev, int slave,
+			   struct mlx4_vhcr *vhcr,
+			   struct mlx4_cmd_mailbox *inbox,
+			   struct mlx4_cmd_mailbox *outbox,
+			   struct mlx4_cmd_info *cmd_info)
+{
+	int err;
+	u32 qpn = vhcr->in_modifier & 0xffffff;
+	struct res_qp *rqp;
+	u64 mac;
+	unsigned port;
+	u64 pri_addr_path_mask;
+	struct mlx4_update_qp_context *cmd;
+	int smac_index;
+
+	cmd = (struct mlx4_update_qp_context *)inbox->buf;
+
+	pri_addr_path_mask = be64_to_cpu(cmd->primary_addr_path_mask);
+	if (cmd->qp_mask || cmd->secondary_addr_path_mask ||
+	    (pri_addr_path_mask & ~MLX4_UPD_QP_PATH_MASK_SUPPORTED))
+		return -EPERM;
+
+	if ((pri_addr_path_mask &
+	     (1ULL << MLX4_UPD_QP_PATH_MASK_ETH_SRC_CHECK_MC_LB)) &&
+		!(dev->caps.flags2 &
+		  MLX4_DEV_CAP_FLAG2_UPDATE_QP_SRC_CHECK_LB)) {
+		mlx4_warn(dev, "Src check LB for slave %d isn't supported\n",
+			  slave);
+		return -EOPNOTSUPP;
+	}
+
+	/* Just change the smac for the QP */
+	err = get_res(dev, slave, qpn, RES_QP, &rqp);
+	if (err) {
+		mlx4_err(dev, "Updating qpn 0x%x for slave %d rejected\n", qpn, slave);
+		return err;
+	}
+
+	port = (rqp->sched_queue >> 6 & 1) + 1;
+
+	if (pri_addr_path_mask & (1ULL << MLX4_UPD_QP_PATH_MASK_MAC_INDEX)) {
+		smac_index = cmd->qp_context.pri_path.grh_mylmc;
+		err = mac_find_smac_ix_in_slave(dev, slave, port,
+						smac_index, &mac);
+
+		if (err) {
+			mlx4_err(dev, "Failed to update qpn 0x%x, MAC is invalid. smac_ix: %d\n",
+				 qpn, smac_index);
+			goto err_mac;
+		}
+	}
+
+	err = mlx4_cmd(dev, inbox->dma,
+		       vhcr->in_modifier, 0,
+		       MLX4_CMD_UPDATE_QP, MLX4_CMD_TIME_CLASS_A,
+		       MLX4_CMD_NATIVE);
+	if (err) {
+		mlx4_err(dev, "Failed to update qpn on qpn 0x%x, command failed\n", qpn);
+		goto err_mac;
+	}
+
+err_mac:
+	put_res(dev, slave, qpn, RES_QP);
+	return err;
+}
+
+static u32 qp_attach_mbox_size(void *mbox)
+{
+	u32 size = sizeof(struct mlx4_net_trans_rule_hw_ctrl);
+	struct _rule_hw  *rule_header;
+
+	rule_header = (struct _rule_hw *)(mbox + size);
+
+	while (rule_header->size) {
+		size += rule_header->size * sizeof(u32);
+		rule_header += 1;
+	}
+	return size;
+}
+
+static int mlx4_do_mirror_rule(struct mlx4_dev *dev, struct res_fs_rule *fs_rule);
+
+int mlx4_QP_FLOW_STEERING_ATTACH_wrapper(struct mlx4_dev *dev, int slave,
+					 struct mlx4_vhcr *vhcr,
+					 struct mlx4_cmd_mailbox *inbox,
+					 struct mlx4_cmd_mailbox *outbox,
+					 struct mlx4_cmd_info *cmd)
+{
+
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_resource_tracker *tracker = &priv->mfunc.master.res_tracker;
+	struct list_head *rlist = &tracker->slave_list[slave].res_list[RES_MAC];
+	int err;
+	int qpn;
+	struct res_qp *rqp;
+	struct mlx4_net_trans_rule_hw_ctrl *ctrl;
+	struct _rule_hw  *rule_header;
+	int header_id;
+	struct res_fs_rule *rrule;
+	u32 mbox_size;
+
+	if (dev->caps.steering_mode !=
+	    MLX4_STEERING_MODE_DEVICE_MANAGED)
+		return -EOPNOTSUPP;
+
+	ctrl = (struct mlx4_net_trans_rule_hw_ctrl *)inbox->buf;
+	err = mlx4_slave_convert_port(dev, slave, ctrl->port);
+	if (err <= 0)
+		return -EINVAL;
+	ctrl->port = err;
+	qpn = be32_to_cpu(ctrl->qpn) & 0xffffff;
+	err = get_res(dev, slave, qpn, RES_QP, &rqp);
+	if (err) {
+		pr_err("Steering rule with qpn 0x%x rejected\n", qpn);
+		return err;
+	}
+	rule_header = (struct _rule_hw *)(ctrl + 1);
+	header_id = map_hw_to_sw_id(be16_to_cpu(rule_header->id));
+
+	if (header_id == MLX4_NET_TRANS_RULE_ID_ETH)
+		mlx4_handle_eth_header_mcast_prio(ctrl, rule_header);
+
+	switch (header_id) {
+	case MLX4_NET_TRANS_RULE_ID_ETH:
+		if (validate_eth_header_mac(slave, rule_header, rlist)) {
+			err = -EINVAL;
+			goto err_put_qp;
+		}
+		break;
+	case MLX4_NET_TRANS_RULE_ID_IB:
+		break;
+	case MLX4_NET_TRANS_RULE_ID_IPV4:
+	case MLX4_NET_TRANS_RULE_ID_TCP:
+	case MLX4_NET_TRANS_RULE_ID_UDP:
+		pr_warn("Can't attach FS rule without L2 headers, adding L2 header\n");
+		if (add_eth_header(dev, slave, inbox, rlist, header_id)) {
+			err = -EINVAL;
+			goto err_put_qp;
+		}
+		vhcr->in_modifier +=
+			sizeof(struct mlx4_net_trans_rule_hw_eth) >> 2;
+		break;
+	default:
+		pr_err("Corrupted mailbox\n");
+		err = -EINVAL;
+		goto err_put_qp;
+	}
+
+	err = mlx4_cmd_imm(dev, inbox->dma, &vhcr->out_param,
+			   vhcr->in_modifier, 0,
+			   MLX4_QP_FLOW_STEERING_ATTACH, MLX4_CMD_TIME_CLASS_A,
+			   MLX4_CMD_NATIVE);
+	if (err)
+		goto err_put_qp;
+
+
+	err = add_res_range(dev, slave, vhcr->out_param, 1, RES_FS_RULE, qpn);
+	if (err) {
+		mlx4_err(dev, "Fail to add flow steering resources\n");
+		goto err_detach;
+	}
+
+	err = get_res(dev, slave, vhcr->out_param, RES_FS_RULE, &rrule);
+	if (err)
+		goto err_detach;
+
+	mbox_size = qp_attach_mbox_size(inbox->buf);
+	rrule->mirr_mbox = kmalloc(mbox_size, GFP_KERNEL);
+	if (!rrule->mirr_mbox) {
+		err = -ENOMEM;
+		goto err_put_rule;
+	}
+	rrule->mirr_mbox_size = mbox_size;
+	rrule->mirr_rule_id = 0;
+	memcpy(rrule->mirr_mbox, inbox->buf, mbox_size);
+
+	/* set different port */
+	ctrl = (struct mlx4_net_trans_rule_hw_ctrl *)rrule->mirr_mbox;
+	if (ctrl->port == 1)
+		ctrl->port = 2;
+	else
+		ctrl->port = 1;
+
+	if (mlx4_is_bonded(dev))
+		mlx4_do_mirror_rule(dev, rrule);
+
+	atomic_inc(&rqp->ref_count);
+
+err_put_rule:
+	put_res(dev, slave, vhcr->out_param, RES_FS_RULE);
+err_detach:
+	/* detach rule on error */
+	if (err)
+		mlx4_cmd(dev, vhcr->out_param, 0, 0,
+			 MLX4_QP_FLOW_STEERING_DETACH, MLX4_CMD_TIME_CLASS_A,
+			 MLX4_CMD_NATIVE);
+err_put_qp:
+	put_res(dev, slave, qpn, RES_QP);
+	return err;
+}
+
+static int mlx4_undo_mirror_rule(struct mlx4_dev *dev, struct res_fs_rule *fs_rule)
+{
+	int err;
+
+	err = rem_res_range(dev, fs_rule->com.owner, fs_rule->com.res_id, 1, RES_FS_RULE, 0);
+	if (err) {
+		mlx4_err(dev, "Fail to remove flow steering resources\n");
+		return err;
+	}
+
+	mlx4_cmd(dev, fs_rule->com.res_id, 0, 0, MLX4_QP_FLOW_STEERING_DETACH,
+		 MLX4_CMD_TIME_CLASS_A, MLX4_CMD_NATIVE);
+	return 0;
+}
+
+int mlx4_QP_FLOW_STEERING_DETACH_wrapper(struct mlx4_dev *dev, int slave,
+					 struct mlx4_vhcr *vhcr,
+					 struct mlx4_cmd_mailbox *inbox,
+					 struct mlx4_cmd_mailbox *outbox,
+					 struct mlx4_cmd_info *cmd)
+{
+	int err;
+	struct res_qp *rqp;
+	struct res_fs_rule *rrule;
+	u64 mirr_reg_id;
+	int qpn;
+
+	if (dev->caps.steering_mode !=
+	    MLX4_STEERING_MODE_DEVICE_MANAGED)
+		return -EOPNOTSUPP;
+
+	err = get_res(dev, slave, vhcr->in_param, RES_FS_RULE, &rrule);
+	if (err)
+		return err;
+
+	if (!rrule->mirr_mbox) {
+		mlx4_err(dev, "Mirror rules cannot be removed explicitly\n");
+		put_res(dev, slave, vhcr->in_param, RES_FS_RULE);
+		return -EINVAL;
+	}
+	mirr_reg_id = rrule->mirr_rule_id;
+	kfree(rrule->mirr_mbox);
+	qpn = rrule->qpn;
+
+	/* Release the rule form busy state before removal */
+	put_res(dev, slave, vhcr->in_param, RES_FS_RULE);
+	err = get_res(dev, slave, qpn, RES_QP, &rqp);
+	if (err)
+		return err;
+
+	if (mirr_reg_id && mlx4_is_bonded(dev)) {
+		err = get_res(dev, slave, mirr_reg_id, RES_FS_RULE, &rrule);
+		if (err) {
+			mlx4_err(dev, "Fail to get resource of mirror rule\n");
+		} else {
+			put_res(dev, slave, mirr_reg_id, RES_FS_RULE);
+			mlx4_undo_mirror_rule(dev, rrule);
+		}
+	}
+	err = rem_res_range(dev, slave, vhcr->in_param, 1, RES_FS_RULE, 0);
+	if (err) {
+		mlx4_err(dev, "Fail to remove flow steering resources\n");
+		goto out;
+	}
+
+	err = mlx4_cmd(dev, vhcr->in_param, 0, 0,
+		       MLX4_QP_FLOW_STEERING_DETACH, MLX4_CMD_TIME_CLASS_A,
+		       MLX4_CMD_NATIVE);
+	if (!err)
+		atomic_dec(&rqp->ref_count);
+out:
+	put_res(dev, slave, qpn, RES_QP);
+	return err;
+}
+
+enum {
+	BUSY_MAX_RETRIES = 10
+};
+
+int mlx4_QUERY_IF_STAT_wrapper(struct mlx4_dev *dev, int slave,
+			       struct mlx4_vhcr *vhcr,
+			       struct mlx4_cmd_mailbox *inbox,
+			       struct mlx4_cmd_mailbox *outbox,
+			       struct mlx4_cmd_info *cmd)
+{
+	int err;
+	int index = vhcr->in_modifier & 0xffff;
+
+	err = get_res(dev, slave, index, RES_COUNTER, NULL);
+	if (err)
+		return err;
+
+	err = mlx4_DMA_wrapper(dev, slave, vhcr, inbox, outbox, cmd);
+	put_res(dev, slave, index, RES_COUNTER);
+	return err;
+}
+
+static void detach_qp(struct mlx4_dev *dev, int slave, struct res_qp *rqp)
+{
+	struct res_gid *rgid;
+	struct res_gid *tmp;
+	struct mlx4_qp qp; /* dummy for calling attach/detach */
+
+	list_for_each_entry_safe(rgid, tmp, &rqp->mcg_list, list) {
+		switch (dev->caps.steering_mode) {
+		case MLX4_STEERING_MODE_DEVICE_MANAGED:
+			mlx4_flow_detach(dev, rgid->reg_id);
+			break;
+		case MLX4_STEERING_MODE_B0:
+			qp.qpn = rqp->local_qpn;
+			(void) mlx4_qp_detach_common(dev, &qp, rgid->gid,
+						     rgid->prot, rgid->steer);
+			break;
+		}
+		list_del(&rgid->list);
+		kfree(rgid);
+	}
+}
+
+static int _move_all_busy(struct mlx4_dev *dev, int slave,
+			  enum mlx4_resource type, int print)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_resource_tracker *tracker =
+		&priv->mfunc.master.res_tracker;
+	struct list_head *rlist = &tracker->slave_list[slave].res_list[type];
+	struct res_common *r;
+	struct res_common *tmp;
+	int busy;
+
+	busy = 0;
+	spin_lock_irq(mlx4_tlock(dev));
+	list_for_each_entry_safe(r, tmp, rlist, list) {
+		if (r->owner == slave) {
+			if (!r->removing) {
+				if (r->state == RES_ANY_BUSY) {
+					if (print)
+						mlx4_dbg(dev,
+							 "%s id 0x%llx is busy\n",
+							  resource_str(type),
+							  r->res_id);
+					++busy;
+				} else {
+					r->from_state = r->state;
+					r->state = RES_ANY_BUSY;
+					r->removing = 1;
+				}
+			}
+		}
+	}
+	spin_unlock_irq(mlx4_tlock(dev));
+
+	return busy;
+}
+
+static int move_all_busy(struct mlx4_dev *dev, int slave,
+			 enum mlx4_resource type)
+{
+	unsigned long begin;
+	int busy;
+
+	begin = jiffies;
+	do {
+		busy = _move_all_busy(dev, slave, type, 0);
+		if (time_after(jiffies, begin + 5 * HZ))
+			break;
+		if (busy)
+			cond_resched();
+	} while (busy);
+
+	if (busy)
+		busy = _move_all_busy(dev, slave, type, 1);
+
+	return busy;
+}
+static void rem_slave_qps(struct mlx4_dev *dev, int slave)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_resource_tracker *tracker = &priv->mfunc.master.res_tracker;
+	struct list_head *qp_list =
+		&tracker->slave_list[slave].res_list[RES_QP];
+	struct res_qp *qp;
+	struct res_qp *tmp;
+	int state;
+	u64 in_param;
+	int qpn;
+	int err;
+
+	err = move_all_busy(dev, slave, RES_QP);
+	if (err)
+		mlx4_warn(dev, "rem_slave_qps: Could not move all qps to busy for slave %d\n",
+			  slave);
+
+	spin_lock_irq(mlx4_tlock(dev));
+	list_for_each_entry_safe(qp, tmp, qp_list, com.list) {
+		spin_unlock_irq(mlx4_tlock(dev));
+		if (qp->com.owner == slave) {
+			qpn = qp->com.res_id;
+			detach_qp(dev, slave, qp);
+			state = qp->com.from_state;
+			while (state != 0) {
+				switch (state) {
+				case RES_QP_RESERVED:
+					spin_lock_irq(mlx4_tlock(dev));
+					rb_erase(&qp->com.node,
+						 &tracker->res_tree[RES_QP]);
+					list_del(&qp->com.list);
+					spin_unlock_irq(mlx4_tlock(dev));
+					if (!valid_reserved(dev, slave, qpn)) {
+						__mlx4_qp_release_range(dev, qpn, 1);
+						mlx4_release_resource(dev, slave,
+								      RES_QP, 1, 0);
+					}
+					kfree(qp);
+					state = 0;
+					break;
+				case RES_QP_MAPPED:
+					if (!valid_reserved(dev, slave, qpn))
+						__mlx4_qp_free_icm(dev, qpn);
+					state = RES_QP_RESERVED;
+					break;
+				case RES_QP_HW:
+					in_param = slave;
+					err = mlx4_cmd(dev, in_param,
+						       qp->local_qpn, 2,
+						       MLX4_CMD_2RST_QP,
+						       MLX4_CMD_TIME_CLASS_A,
+						       MLX4_CMD_NATIVE);
+					if (err)
+						mlx4_dbg(dev, "rem_slave_qps: failed to move slave %d qpn %d to reset\n",
+							 slave, qp->local_qpn);
+					atomic_dec(&qp->rcq->ref_count);
+					atomic_dec(&qp->scq->ref_count);
+					atomic_dec(&qp->mtt->ref_count);
+					if (qp->srq)
+						atomic_dec(&qp->srq->ref_count);
+					state = RES_QP_MAPPED;
+					break;
+				default:
+					state = 0;
+				}
+			}
+		}
+		spin_lock_irq(mlx4_tlock(dev));
+	}
+	spin_unlock_irq(mlx4_tlock(dev));
+}
+
+static void rem_slave_srqs(struct mlx4_dev *dev, int slave)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_resource_tracker *tracker = &priv->mfunc.master.res_tracker;
+	struct list_head *srq_list =
+		&tracker->slave_list[slave].res_list[RES_SRQ];
+	struct res_srq *srq;
+	struct res_srq *tmp;
+	int state;
+	u64 in_param;
+	int srqn;
+	int err;
+
+	err = move_all_busy(dev, slave, RES_SRQ);
+	if (err)
+		mlx4_warn(dev, "rem_slave_srqs: Could not move all srqs - too busy for slave %d\n",
+			  slave);
+
+	spin_lock_irq(mlx4_tlock(dev));
+	list_for_each_entry_safe(srq, tmp, srq_list, com.list) {
+		spin_unlock_irq(mlx4_tlock(dev));
+		if (srq->com.owner == slave) {
+			srqn = srq->com.res_id;
+			state = srq->com.from_state;
+			while (state != 0) {
+				switch (state) {
+				case RES_SRQ_ALLOCATED:
+					__mlx4_srq_free_icm(dev, srqn);
+					spin_lock_irq(mlx4_tlock(dev));
+					rb_erase(&srq->com.node,
+						 &tracker->res_tree[RES_SRQ]);
+					list_del(&srq->com.list);
+					spin_unlock_irq(mlx4_tlock(dev));
+					mlx4_release_resource(dev, slave,
+							      RES_SRQ, 1, 0);
+					kfree(srq);
+					state = 0;
+					break;
+
+				case RES_SRQ_HW:
+					in_param = slave;
+					err = mlx4_cmd(dev, in_param, srqn, 1,
+						       MLX4_CMD_HW2SW_SRQ,
+						       MLX4_CMD_TIME_CLASS_A,
+						       MLX4_CMD_NATIVE);
+					if (err)
+						mlx4_dbg(dev, "rem_slave_srqs: failed to move slave %d srq %d to SW ownership\n",
+							 slave, srqn);
+
+					atomic_dec(&srq->mtt->ref_count);
+					if (srq->cq)
+						atomic_dec(&srq->cq->ref_count);
+					state = RES_SRQ_ALLOCATED;
+					break;
+
+				default:
+					state = 0;
+				}
+			}
+		}
+		spin_lock_irq(mlx4_tlock(dev));
+	}
+	spin_unlock_irq(mlx4_tlock(dev));
+}
+
+static void rem_slave_cqs(struct mlx4_dev *dev, int slave)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_resource_tracker *tracker = &priv->mfunc.master.res_tracker;
+	struct list_head *cq_list =
+		&tracker->slave_list[slave].res_list[RES_CQ];
+	struct res_cq *cq;
+	struct res_cq *tmp;
+	int state;
+	u64 in_param;
+	int cqn;
+	int err;
+
+	err = move_all_busy(dev, slave, RES_CQ);
+	if (err)
+		mlx4_warn(dev, "rem_slave_cqs: Could not move all cqs - too busy for slave %d\n",
+			  slave);
+
+	spin_lock_irq(mlx4_tlock(dev));
+	list_for_each_entry_safe(cq, tmp, cq_list, com.list) {
+		spin_unlock_irq(mlx4_tlock(dev));
+		if (cq->com.owner == slave && !atomic_read(&cq->ref_count)) {
+			cqn = cq->com.res_id;
+			state = cq->com.from_state;
+			while (state != 0) {
+				switch (state) {
+				case RES_CQ_ALLOCATED:
+					__mlx4_cq_free_icm(dev, cqn);
+					spin_lock_irq(mlx4_tlock(dev));
+					rb_erase(&cq->com.node,
+						 &tracker->res_tree[RES_CQ]);
+					list_del(&cq->com.list);
+					spin_unlock_irq(mlx4_tlock(dev));
+					mlx4_release_resource(dev, slave,
+							      RES_CQ, 1, 0);
+					kfree(cq);
+					state = 0;
+					break;
+
+				case RES_CQ_HW:
+					in_param = slave;
+					err = mlx4_cmd(dev, in_param, cqn, 1,
+						       MLX4_CMD_HW2SW_CQ,
+						       MLX4_CMD_TIME_CLASS_A,
+						       MLX4_CMD_NATIVE);
+					if (err)
+						mlx4_dbg(dev, "rem_slave_cqs: failed to move slave %d cq %d to SW ownership\n",
+							 slave, cqn);
+					atomic_dec(&cq->mtt->ref_count);
+					state = RES_CQ_ALLOCATED;
+					break;
+
+				default:
+					state = 0;
+				}
+			}
+		}
+		spin_lock_irq(mlx4_tlock(dev));
+	}
+	spin_unlock_irq(mlx4_tlock(dev));
+}
+
+static void rem_slave_mrs(struct mlx4_dev *dev, int slave)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_resource_tracker *tracker = &priv->mfunc.master.res_tracker;
+	struct list_head *mpt_list =
+		&tracker->slave_list[slave].res_list[RES_MPT];
+	struct res_mpt *mpt;
+	struct res_mpt *tmp;
+	int state;
+	u64 in_param;
+	int mptn;
+	int err;
+
+	err = move_all_busy(dev, slave, RES_MPT);
+	if (err)
+		mlx4_warn(dev, "rem_slave_mrs: Could not move all mpts - too busy for slave %d\n",
+			  slave);
+
+	spin_lock_irq(mlx4_tlock(dev));
+	list_for_each_entry_safe(mpt, tmp, mpt_list, com.list) {
+		spin_unlock_irq(mlx4_tlock(dev));
+		if (mpt->com.owner == slave) {
+			mptn = mpt->com.res_id;
+			state = mpt->com.from_state;
+			while (state != 0) {
+				switch (state) {
+				case RES_MPT_RESERVED:
+					__mlx4_mpt_release(dev, mpt->key);
+					spin_lock_irq(mlx4_tlock(dev));
+					rb_erase(&mpt->com.node,
+						 &tracker->res_tree[RES_MPT]);
+					list_del(&mpt->com.list);
+					spin_unlock_irq(mlx4_tlock(dev));
+					mlx4_release_resource(dev, slave,
+							      RES_MPT, 1, 0);
+					kfree(mpt);
+					state = 0;
+					break;
+
+				case RES_MPT_MAPPED:
+					__mlx4_mpt_free_icm(dev, mpt->key);
+					state = RES_MPT_RESERVED;
+					break;
+
+				case RES_MPT_HW:
+					in_param = slave;
+					err = mlx4_cmd(dev, in_param, mptn, 0,
+						     MLX4_CMD_HW2SW_MPT,
+						     MLX4_CMD_TIME_CLASS_A,
+						     MLX4_CMD_NATIVE);
+					if (err)
+						mlx4_dbg(dev, "rem_slave_mrs: failed to move slave %d mpt %d to SW ownership\n",
+							 slave, mptn);
+					if (mpt->mtt)
+						atomic_dec(&mpt->mtt->ref_count);
+					state = RES_MPT_MAPPED;
+					break;
+				default:
+					state = 0;
+				}
+			}
+		}
+		spin_lock_irq(mlx4_tlock(dev));
+	}
+	spin_unlock_irq(mlx4_tlock(dev));
+}
+
+static void rem_slave_mtts(struct mlx4_dev *dev, int slave)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_resource_tracker *tracker =
+		&priv->mfunc.master.res_tracker;
+	struct list_head *mtt_list =
+		&tracker->slave_list[slave].res_list[RES_MTT];
+	struct res_mtt *mtt;
+	struct res_mtt *tmp;
+	int state;
+	int base;
+	int err;
+
+	err = move_all_busy(dev, slave, RES_MTT);
+	if (err)
+		mlx4_warn(dev, "rem_slave_mtts: Could not move all mtts  - too busy for slave %d\n",
+			  slave);
+
+	spin_lock_irq(mlx4_tlock(dev));
+	list_for_each_entry_safe(mtt, tmp, mtt_list, com.list) {
+		spin_unlock_irq(mlx4_tlock(dev));
+		if (mtt->com.owner == slave) {
+			base = mtt->com.res_id;
+			state = mtt->com.from_state;
+			while (state != 0) {
+				switch (state) {
+				case RES_MTT_ALLOCATED:
+					__mlx4_free_mtt_range(dev, base,
+							      mtt->order);
+					spin_lock_irq(mlx4_tlock(dev));
+					rb_erase(&mtt->com.node,
+						 &tracker->res_tree[RES_MTT]);
+					list_del(&mtt->com.list);
+					spin_unlock_irq(mlx4_tlock(dev));
+					mlx4_release_resource(dev, slave, RES_MTT,
+							      1 << mtt->order, 0);
+					kfree(mtt);
+					state = 0;
+					break;
+
+				default:
+					state = 0;
+				}
+			}
+		}
+		spin_lock_irq(mlx4_tlock(dev));
+	}
+	spin_unlock_irq(mlx4_tlock(dev));
+}
+
+static int mlx4_do_mirror_rule(struct mlx4_dev *dev, struct res_fs_rule *fs_rule)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	int err;
+	struct res_fs_rule *mirr_rule;
+	u64 reg_id;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+
+	if (!fs_rule->mirr_mbox) {
+		mlx4_err(dev, "rule mirroring mailbox is null\n");
+		mlx4_free_cmd_mailbox(dev, mailbox);
+		return -EINVAL;
+	}
+	memcpy(mailbox->buf, fs_rule->mirr_mbox, fs_rule->mirr_mbox_size);
+	err = mlx4_cmd_imm(dev, mailbox->dma, &reg_id, fs_rule->mirr_mbox_size >> 2, 0,
+			   MLX4_QP_FLOW_STEERING_ATTACH, MLX4_CMD_TIME_CLASS_A,
+			   MLX4_CMD_NATIVE);
+	mlx4_free_cmd_mailbox(dev, mailbox);
+
+	if (err)
+		goto err;
+
+	err = add_res_range(dev, fs_rule->com.owner, reg_id, 1, RES_FS_RULE, fs_rule->qpn);
+	if (err)
+		goto err_detach;
+
+	err = get_res(dev, fs_rule->com.owner, reg_id, RES_FS_RULE, &mirr_rule);
+	if (err)
+		goto err_rem;
+
+	fs_rule->mirr_rule_id = reg_id;
+	mirr_rule->mirr_rule_id = 0;
+	mirr_rule->mirr_mbox_size = 0;
+	mirr_rule->mirr_mbox = NULL;
+	put_res(dev, fs_rule->com.owner, reg_id, RES_FS_RULE);
+
+	return 0;
+err_rem:
+	rem_res_range(dev, fs_rule->com.owner, reg_id, 1, RES_FS_RULE, 0);
+err_detach:
+	mlx4_cmd(dev, reg_id, 0, 0, MLX4_QP_FLOW_STEERING_DETACH,
+		 MLX4_CMD_TIME_CLASS_A, MLX4_CMD_NATIVE);
+err:
+	return err;
+}
+
+static int mlx4_mirror_fs_rules(struct mlx4_dev *dev, bool bond)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_resource_tracker *tracker =
+		&priv->mfunc.master.res_tracker;
+	struct rb_root *root = &tracker->res_tree[RES_FS_RULE];
+	struct rb_node *p;
+	struct res_fs_rule *fs_rule;
+	int err = 0;
+	LIST_HEAD(mirr_list);
+
+	for (p = rb_first(root); p; p = rb_next(p)) {
+		fs_rule = rb_entry(p, struct res_fs_rule, com.node);
+		if ((bond && fs_rule->mirr_mbox_size) ||
+		    (!bond && !fs_rule->mirr_mbox_size))
+			list_add_tail(&fs_rule->mirr_list, &mirr_list);
+	}
+
+	list_for_each_entry(fs_rule, &mirr_list, mirr_list) {
+		if (bond)
+			err += mlx4_do_mirror_rule(dev, fs_rule);
+		else
+			err += mlx4_undo_mirror_rule(dev, fs_rule);
+	}
+	return err;
+}
+
+int mlx4_bond_fs_rules(struct mlx4_dev *dev)
+{
+	return mlx4_mirror_fs_rules(dev, true);
+}
+
+int mlx4_unbond_fs_rules(struct mlx4_dev *dev)
+{
+	return mlx4_mirror_fs_rules(dev, false);
+}
+
+static void rem_slave_fs_rule(struct mlx4_dev *dev, int slave)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_resource_tracker *tracker =
+		&priv->mfunc.master.res_tracker;
+	struct list_head *fs_rule_list =
+		&tracker->slave_list[slave].res_list[RES_FS_RULE];
+	struct res_fs_rule *fs_rule;
+	struct res_fs_rule *tmp;
+	int state;
+	u64 base;
+	int err;
+
+	err = move_all_busy(dev, slave, RES_FS_RULE);
+	if (err)
+		mlx4_warn(dev, "rem_slave_fs_rule: Could not move all mtts to busy for slave %d\n",
+			  slave);
+
+	spin_lock_irq(mlx4_tlock(dev));
+	list_for_each_entry_safe(fs_rule, tmp, fs_rule_list, com.list) {
+		spin_unlock_irq(mlx4_tlock(dev));
+		if (fs_rule->com.owner == slave) {
+			base = fs_rule->com.res_id;
+			state = fs_rule->com.from_state;
+			while (state != 0) {
+				switch (state) {
+				case RES_FS_RULE_ALLOCATED:
+					/* detach rule */
+					err = mlx4_cmd(dev, base, 0, 0,
+						       MLX4_QP_FLOW_STEERING_DETACH,
+						       MLX4_CMD_TIME_CLASS_A,
+						       MLX4_CMD_NATIVE);
+
+					spin_lock_irq(mlx4_tlock(dev));
+					rb_erase(&fs_rule->com.node,
+						 &tracker->res_tree[RES_FS_RULE]);
+					list_del(&fs_rule->com.list);
+					spin_unlock_irq(mlx4_tlock(dev));
+					kfree(fs_rule->mirr_mbox);
+					kfree(fs_rule);
+					state = 0;
+					break;
+
+				default:
+					state = 0;
+				}
+			}
+		}
+		spin_lock_irq(mlx4_tlock(dev));
+	}
+	spin_unlock_irq(mlx4_tlock(dev));
+}
+
+static void rem_slave_eqs(struct mlx4_dev *dev, int slave)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_resource_tracker *tracker = &priv->mfunc.master.res_tracker;
+	struct list_head *eq_list =
+		&tracker->slave_list[slave].res_list[RES_EQ];
+	struct res_eq *eq;
+	struct res_eq *tmp;
+	int err;
+	int state;
+	int eqn;
+
+	err = move_all_busy(dev, slave, RES_EQ);
+	if (err)
+		mlx4_warn(dev, "rem_slave_eqs: Could not move all eqs - too busy for slave %d\n",
+			  slave);
+
+	spin_lock_irq(mlx4_tlock(dev));
+	list_for_each_entry_safe(eq, tmp, eq_list, com.list) {
+		spin_unlock_irq(mlx4_tlock(dev));
+		if (eq->com.owner == slave) {
+			eqn = eq->com.res_id;
+			state = eq->com.from_state;
+			while (state != 0) {
+				switch (state) {
+				case RES_EQ_RESERVED:
+					spin_lock_irq(mlx4_tlock(dev));
+					rb_erase(&eq->com.node,
+						 &tracker->res_tree[RES_EQ]);
+					list_del(&eq->com.list);
+					spin_unlock_irq(mlx4_tlock(dev));
+					kfree(eq);
+					state = 0;
+					break;
+
+				case RES_EQ_HW:
+					err = mlx4_cmd(dev, slave, eqn & 0x3ff,
+						       1, MLX4_CMD_HW2SW_EQ,
+						       MLX4_CMD_TIME_CLASS_A,
+						       MLX4_CMD_NATIVE);
+					if (err)
+						mlx4_dbg(dev, "rem_slave_eqs: failed to move slave %d eqs %d to SW ownership\n",
+							 slave, eqn & 0x3ff);
+					atomic_dec(&eq->mtt->ref_count);
+					state = RES_EQ_RESERVED;
+					break;
+
+				default:
+					state = 0;
+				}
+			}
+		}
+		spin_lock_irq(mlx4_tlock(dev));
+	}
+	spin_unlock_irq(mlx4_tlock(dev));
+}
+
+static void rem_slave_counters(struct mlx4_dev *dev, int slave)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_resource_tracker *tracker = &priv->mfunc.master.res_tracker;
+	struct list_head *counter_list =
+		&tracker->slave_list[slave].res_list[RES_COUNTER];
+	struct res_counter *counter;
+	struct res_counter *tmp;
+	int err;
+	int *counters_arr = NULL;
+	int i, j;
+
+	err = move_all_busy(dev, slave, RES_COUNTER);
+	if (err)
+		mlx4_warn(dev, "rem_slave_counters: Could not move all counters - too busy for slave %d\n",
+			  slave);
+
+	counters_arr = kmalloc_array(dev->caps.max_counters,
+				     sizeof(*counters_arr), GFP_KERNEL);
+	if (!counters_arr)
+		return;
+
+	do {
+		i = 0;
+		j = 0;
+		spin_lock_irq(mlx4_tlock(dev));
+		list_for_each_entry_safe(counter, tmp, counter_list, com.list) {
+			if (counter->com.owner == slave) {
+				counters_arr[i++] = counter->com.res_id;
+				rb_erase(&counter->com.node,
+					 &tracker->res_tree[RES_COUNTER]);
+				list_del(&counter->com.list);
+				kfree(counter);
+			}
+		}
+		spin_unlock_irq(mlx4_tlock(dev));
+
+		while (j < i) {
+			__mlx4_counter_free(dev, counters_arr[j++]);
+			mlx4_release_resource(dev, slave, RES_COUNTER, 1, 0);
+		}
+	} while (i);
+
+	kfree(counters_arr);
+}
+
+static void rem_slave_xrcdns(struct mlx4_dev *dev, int slave)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_resource_tracker *tracker = &priv->mfunc.master.res_tracker;
+	struct list_head *xrcdn_list =
+		&tracker->slave_list[slave].res_list[RES_XRCD];
+	struct res_xrcdn *xrcd;
+	struct res_xrcdn *tmp;
+	int err;
+	int xrcdn;
+
+	err = move_all_busy(dev, slave, RES_XRCD);
+	if (err)
+		mlx4_warn(dev, "rem_slave_xrcdns: Could not move all xrcdns - too busy for slave %d\n",
+			  slave);
+
+	spin_lock_irq(mlx4_tlock(dev));
+	list_for_each_entry_safe(xrcd, tmp, xrcdn_list, com.list) {
+		if (xrcd->com.owner == slave) {
+			xrcdn = xrcd->com.res_id;
+			rb_erase(&xrcd->com.node, &tracker->res_tree[RES_XRCD]);
+			list_del(&xrcd->com.list);
+			kfree(xrcd);
+			__mlx4_xrcd_free(dev, xrcdn);
+		}
+	}
+	spin_unlock_irq(mlx4_tlock(dev));
+}
+
+void mlx4_delete_all_resources_for_slave(struct mlx4_dev *dev, int slave)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	mlx4_reset_roce_gids(dev, slave);
+	mutex_lock(&priv->mfunc.master.res_tracker.slave_list[slave].mutex);
+	rem_slave_vlans(dev, slave);
+	rem_slave_macs(dev, slave);
+	rem_slave_fs_rule(dev, slave);
+	rem_slave_qps(dev, slave);
+	rem_slave_srqs(dev, slave);
+	rem_slave_cqs(dev, slave);
+	rem_slave_mrs(dev, slave);
+	rem_slave_eqs(dev, slave);
+	rem_slave_mtts(dev, slave);
+	rem_slave_counters(dev, slave);
+	rem_slave_xrcdns(dev, slave);
+	mutex_unlock(&priv->mfunc.master.res_tracker.slave_list[slave].mutex);
+}
+
+static void update_qos_vpp(struct mlx4_update_qp_context *ctx,
+			   struct mlx4_vf_immed_vlan_work *work)
+{
+	ctx->qp_mask |= cpu_to_be64(1ULL << MLX4_UPD_QP_MASK_QOS_VPP);
+	ctx->qp_context.qos_vport = work->qos_vport;
+}
+
+void mlx4_vf_immed_vlan_work_handler(struct work_struct *_work)
+{
+	struct mlx4_vf_immed_vlan_work *work =
+		container_of(_work, struct mlx4_vf_immed_vlan_work, work);
+	struct mlx4_cmd_mailbox *mailbox;
+	struct mlx4_update_qp_context *upd_context;
+	struct mlx4_dev *dev = &work->priv->dev;
+	struct mlx4_resource_tracker *tracker =
+		&work->priv->mfunc.master.res_tracker;
+	struct list_head *qp_list =
+		&tracker->slave_list[work->slave].res_list[RES_QP];
+	struct res_qp *qp;
+	struct res_qp *tmp;
+	u64 qp_path_mask_vlan_ctrl =
+		       ((1ULL << MLX4_UPD_QP_PATH_MASK_ETH_TX_BLOCK_UNTAGGED) |
+		       (1ULL << MLX4_UPD_QP_PATH_MASK_ETH_TX_BLOCK_1P) |
+		       (1ULL << MLX4_UPD_QP_PATH_MASK_ETH_TX_BLOCK_TAGGED) |
+		       (1ULL << MLX4_UPD_QP_PATH_MASK_ETH_RX_BLOCK_UNTAGGED) |
+		       (1ULL << MLX4_UPD_QP_PATH_MASK_ETH_RX_BLOCK_1P) |
+		       (1ULL << MLX4_UPD_QP_PATH_MASK_ETH_RX_BLOCK_TAGGED));
+
+	u64 qp_path_mask = ((1ULL << MLX4_UPD_QP_PATH_MASK_VLAN_INDEX) |
+		       (1ULL << MLX4_UPD_QP_PATH_MASK_FVL) |
+		       (1ULL << MLX4_UPD_QP_PATH_MASK_CV) |
+		       (1ULL << MLX4_UPD_QP_PATH_MASK_SV) |
+		       (1ULL << MLX4_UPD_QP_PATH_MASK_ETH_HIDE_CQE_VLAN) |
+		       (1ULL << MLX4_UPD_QP_PATH_MASK_FEUP) |
+		       (1ULL << MLX4_UPD_QP_PATH_MASK_FVL_RX) |
+		       (1ULL << MLX4_UPD_QP_PATH_MASK_SCHED_QUEUE));
+
+	int err;
+	int port, errors = 0;
+	u8 vlan_control;
+
+	if (mlx4_is_slave(dev)) {
+		mlx4_warn(dev, "Trying to update-qp in slave %d\n",
+			  work->slave);
+		goto out;
+	}
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		goto out;
+	if (work->flags & MLX4_VF_IMMED_VLAN_FLAG_LINK_DISABLE) /* block all */
+		vlan_control = MLX4_VLAN_CTRL_ETH_TX_BLOCK_TAGGED |
+			MLX4_VLAN_CTRL_ETH_TX_BLOCK_PRIO_TAGGED |
+			MLX4_VLAN_CTRL_ETH_TX_BLOCK_UNTAGGED |
+			MLX4_VLAN_CTRL_ETH_RX_BLOCK_PRIO_TAGGED |
+			MLX4_VLAN_CTRL_ETH_RX_BLOCK_UNTAGGED |
+			MLX4_VLAN_CTRL_ETH_RX_BLOCK_TAGGED;
+	else if (!work->vlan_id)
+		vlan_control = MLX4_VLAN_CTRL_ETH_TX_BLOCK_TAGGED |
+			MLX4_VLAN_CTRL_ETH_RX_BLOCK_TAGGED;
+	else if (work->vlan_proto == htons(ETH_P_8021AD))
+		vlan_control = MLX4_VLAN_CTRL_ETH_TX_BLOCK_PRIO_TAGGED |
+			MLX4_VLAN_CTRL_ETH_TX_BLOCK_TAGGED |
+			MLX4_VLAN_CTRL_ETH_RX_BLOCK_PRIO_TAGGED |
+			MLX4_VLAN_CTRL_ETH_RX_BLOCK_UNTAGGED;
+	else  /* vst 802.1Q */
+		vlan_control = MLX4_VLAN_CTRL_ETH_TX_BLOCK_TAGGED |
+			MLX4_VLAN_CTRL_ETH_RX_BLOCK_PRIO_TAGGED |
+			MLX4_VLAN_CTRL_ETH_RX_BLOCK_UNTAGGED;
+
+	upd_context = mailbox->buf;
+	upd_context->qp_mask = cpu_to_be64(1ULL << MLX4_UPD_QP_MASK_VSD);
+
+	spin_lock_irq(mlx4_tlock(dev));
+	list_for_each_entry_safe(qp, tmp, qp_list, com.list) {
+		spin_unlock_irq(mlx4_tlock(dev));
+		if (qp->com.owner == work->slave) {
+			if (qp->com.from_state != RES_QP_HW ||
+			    !qp->sched_queue ||  /* no INIT2RTR trans yet */
+			    mlx4_is_qp_reserved(dev, qp->local_qpn) ||
+			    qp->qpc_flags & (1 << MLX4_RSS_QPC_FLAG_OFFSET)) {
+				spin_lock_irq(mlx4_tlock(dev));
+				continue;
+			}
+			port = (qp->sched_queue >> 6 & 1) + 1;
+			if (port != work->port) {
+				spin_lock_irq(mlx4_tlock(dev));
+				continue;
+			}
+			if (MLX4_QP_ST_RC == ((qp->qpc_flags >> 16) & 0xff))
+				upd_context->primary_addr_path_mask = cpu_to_be64(qp_path_mask);
+			else
+				upd_context->primary_addr_path_mask =
+					cpu_to_be64(qp_path_mask | qp_path_mask_vlan_ctrl);
+			if (work->vlan_id == MLX4_VGT) {
+				upd_context->qp_context.param3 = qp->param3;
+				upd_context->qp_context.pri_path.vlan_control = qp->vlan_control;
+				upd_context->qp_context.pri_path.fvl_rx = qp->fvl_rx;
+				upd_context->qp_context.pri_path.vlan_index = qp->vlan_index;
+				upd_context->qp_context.pri_path.fl = qp->pri_path_fl;
+				upd_context->qp_context.pri_path.feup = qp->feup;
+				upd_context->qp_context.pri_path.sched_queue =
+					qp->sched_queue;
+			} else {
+				upd_context->qp_context.param3 = qp->param3 & ~cpu_to_be32(MLX4_STRIP_VLAN);
+				upd_context->qp_context.pri_path.vlan_control = vlan_control;
+				upd_context->qp_context.pri_path.vlan_index = work->vlan_ix;
+				upd_context->qp_context.pri_path.fvl_rx =
+					qp->fvl_rx | MLX4_FVL_RX_FORCE_ETH_VLAN;
+				upd_context->qp_context.pri_path.fl =
+					qp->pri_path_fl | MLX4_FL_ETH_HIDE_CQE_VLAN;
+				if (work->vlan_proto == htons(ETH_P_8021AD))
+					upd_context->qp_context.pri_path.fl |= MLX4_FL_SV;
+				else
+					upd_context->qp_context.pri_path.fl |= MLX4_FL_CV;
+				upd_context->qp_context.pri_path.feup =
+					qp->feup | MLX4_FEUP_FORCE_ETH_UP | MLX4_FVL_FORCE_ETH_VLAN;
+				upd_context->qp_context.pri_path.sched_queue =
+					qp->sched_queue & 0xC7;
+				upd_context->qp_context.pri_path.sched_queue |=
+					((work->qos & 0x7) << 3);
+
+				if (dev->caps.flags2 &
+				    MLX4_DEV_CAP_FLAG2_QOS_VPP)
+					update_qos_vpp(upd_context, work);
+			}
+
+			err = mlx4_cmd(dev, mailbox->dma,
+				       qp->local_qpn & 0xffffff,
+				       0, MLX4_CMD_UPDATE_QP,
+				       MLX4_CMD_TIME_CLASS_C, MLX4_CMD_NATIVE);
+			if (err) {
+				mlx4_info(dev, "UPDATE_QP failed for slave %d, port %d, qpn %d (%d)\n",
+					  work->slave, port, qp->local_qpn, err);
+				errors++;
+			}
+		}
+		spin_lock_irq(mlx4_tlock(dev));
+	}
+	spin_unlock_irq(mlx4_tlock(dev));
+	mlx4_free_cmd_mailbox(dev, mailbox);
+
+	if (errors)
+		mlx4_err(dev, "%d UPDATE_QP failures for slave %d, port %d\n",
+			 errors, work->slave, work->port);
+
+	/* unregister previous vlan_id if needed and we had no errors
+	 * while updating the QPs
+	 */
+	if (work->flags & MLX4_VF_IMMED_VLAN_FLAG_VLAN && !errors &&
+	    NO_INDX != work->orig_vlan_ix)
+		__mlx4_unregister_vlan(&work->priv->dev, work->port,
+				       work->orig_vlan_id);
+out:
+	kfree(work);
+	return;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx4/sense.c b/drivers/net/ethernet/mellanox/mlx4/sense.c
new file mode 100644
index 000000000..094773d88
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx4/sense.c
@@ -0,0 +1,143 @@
+/*
+ * Copyright (c) 2007 Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+
+#include <linux/errno.h>
+#include <linux/if_ether.h>
+
+#include <linux/mlx4/cmd.h>
+
+#include "mlx4.h"
+
+int mlx4_SENSE_PORT(struct mlx4_dev *dev, int port,
+		    enum mlx4_port_type *type)
+{
+	u64 out_param;
+	int err = 0;
+
+	err = mlx4_cmd_imm(dev, 0, &out_param, port, 0,
+			   MLX4_CMD_SENSE_PORT, MLX4_CMD_TIME_CLASS_B,
+			   MLX4_CMD_WRAPPED);
+	if (err) {
+		mlx4_err(dev, "Sense command failed for port: %d\n", port);
+		return err;
+	}
+
+	if (out_param > 2) {
+		mlx4_err(dev, "Sense returned illegal value: 0x%llx\n", out_param);
+		return -EINVAL;
+	}
+
+	*type = out_param;
+	return 0;
+}
+
+void mlx4_do_sense_ports(struct mlx4_dev *dev,
+			 enum mlx4_port_type *stype,
+			 enum mlx4_port_type *defaults)
+{
+	struct mlx4_sense *sense = &mlx4_priv(dev)->sense;
+	int err;
+	int i;
+
+	for (i = 1; i <= dev->caps.num_ports; i++) {
+		stype[i - 1] = 0;
+		if (sense->do_sense_port[i] && sense->sense_allowed[i] &&
+		    dev->caps.possible_type[i] == MLX4_PORT_TYPE_AUTO) {
+			err = mlx4_SENSE_PORT(dev, i, &stype[i - 1]);
+			if (err)
+				stype[i - 1] = defaults[i - 1];
+		} else
+			stype[i - 1] = defaults[i - 1];
+	}
+
+	/*
+	 * If sensed nothing, remain in current configuration.
+	 */
+	for (i = 0; i < dev->caps.num_ports; i++)
+		stype[i] = stype[i] ? stype[i] : defaults[i];
+
+}
+
+static void mlx4_sense_port(struct work_struct *work)
+{
+	struct delayed_work *delay = to_delayed_work(work);
+	struct mlx4_sense *sense = container_of(delay, struct mlx4_sense,
+						sense_poll);
+	struct mlx4_dev *dev = sense->dev;
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	enum mlx4_port_type stype[MLX4_MAX_PORTS];
+
+	mutex_lock(&priv->port_mutex);
+	mlx4_do_sense_ports(dev, stype, &dev->caps.port_type[1]);
+
+	if (mlx4_check_port_params(dev, stype))
+		goto sense_again;
+
+	if (mlx4_change_port_types(dev, stype))
+		mlx4_err(dev, "Failed to change port_types\n");
+
+sense_again:
+	mutex_unlock(&priv->port_mutex);
+	queue_delayed_work(mlx4_wq , &sense->sense_poll,
+			   round_jiffies_relative(MLX4_SENSE_RANGE));
+}
+
+void mlx4_start_sense(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_sense *sense = &priv->sense;
+
+	if (!(dev->caps.flags & MLX4_DEV_CAP_FLAG_DPDP))
+		return;
+
+	queue_delayed_work(mlx4_wq , &sense->sense_poll,
+			   round_jiffies_relative(MLX4_SENSE_RANGE));
+}
+
+void mlx4_stop_sense(struct mlx4_dev *dev)
+{
+	cancel_delayed_work_sync(&mlx4_priv(dev)->sense.sense_poll);
+}
+
+void  mlx4_sense_init(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_sense *sense = &priv->sense;
+	int port;
+
+	sense->dev = dev;
+	for (port = 1; port <= dev->caps.num_ports; port++)
+		sense->do_sense_port[port] = 1;
+
+	INIT_DEFERRABLE_WORK(&sense->sense_poll, mlx4_sense_port);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx4/srq.c b/drivers/net/ethernet/mellanox/mlx4/srq.c
new file mode 100644
index 000000000..dd890f5d7
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx4/srq.c
@@ -0,0 +1,304 @@
+/*
+ * Copyright (c) 2006, 2007 Cisco Systems, Inc. All rights reserved.
+ * Copyright (c) 2007, 2008 Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+
+#include <linux/mlx4/cmd.h>
+#include <linux/mlx4/srq.h>
+#include <linux/export.h>
+#include <linux/gfp.h>
+
+#include "mlx4.h"
+#include "icm.h"
+
+void mlx4_srq_event(struct mlx4_dev *dev, u32 srqn, int event_type)
+{
+	struct mlx4_srq_table *srq_table = &mlx4_priv(dev)->srq_table;
+	struct mlx4_srq *srq;
+
+	rcu_read_lock();
+	srq = radix_tree_lookup(&srq_table->tree, srqn & (dev->caps.num_srqs - 1));
+	rcu_read_unlock();
+	if (srq)
+		refcount_inc(&srq->refcount);
+	else {
+		mlx4_warn(dev, "Async event for bogus SRQ %08x\n", srqn);
+		return;
+	}
+
+	srq->event(srq, event_type);
+
+	if (refcount_dec_and_test(&srq->refcount))
+		complete(&srq->free);
+}
+
+static int mlx4_SW2HW_SRQ(struct mlx4_dev *dev, struct mlx4_cmd_mailbox *mailbox,
+			  int srq_num)
+{
+	return mlx4_cmd(dev, mailbox->dma, srq_num, 0,
+			MLX4_CMD_SW2HW_SRQ, MLX4_CMD_TIME_CLASS_A,
+			MLX4_CMD_WRAPPED);
+}
+
+static int mlx4_HW2SW_SRQ(struct mlx4_dev *dev, struct mlx4_cmd_mailbox *mailbox,
+			  int srq_num)
+{
+	return mlx4_cmd_box(dev, 0, mailbox ? mailbox->dma : 0, srq_num,
+			    mailbox ? 0 : 1, MLX4_CMD_HW2SW_SRQ,
+			    MLX4_CMD_TIME_CLASS_A, MLX4_CMD_WRAPPED);
+}
+
+static int mlx4_ARM_SRQ(struct mlx4_dev *dev, int srq_num, int limit_watermark)
+{
+	return mlx4_cmd(dev, limit_watermark, srq_num, 0, MLX4_CMD_ARM_SRQ,
+			MLX4_CMD_TIME_CLASS_B, MLX4_CMD_WRAPPED);
+}
+
+static int mlx4_QUERY_SRQ(struct mlx4_dev *dev, struct mlx4_cmd_mailbox *mailbox,
+			  int srq_num)
+{
+	return mlx4_cmd_box(dev, 0, mailbox->dma, srq_num, 0, MLX4_CMD_QUERY_SRQ,
+			    MLX4_CMD_TIME_CLASS_A, MLX4_CMD_WRAPPED);
+}
+
+int __mlx4_srq_alloc_icm(struct mlx4_dev *dev, int *srqn)
+{
+	struct mlx4_srq_table *srq_table = &mlx4_priv(dev)->srq_table;
+	int err;
+
+
+	*srqn = mlx4_bitmap_alloc(&srq_table->bitmap);
+	if (*srqn == -1)
+		return -ENOMEM;
+
+	err = mlx4_table_get(dev, &srq_table->table, *srqn);
+	if (err)
+		goto err_out;
+
+	err = mlx4_table_get(dev, &srq_table->cmpt_table, *srqn);
+	if (err)
+		goto err_put;
+	return 0;
+
+err_put:
+	mlx4_table_put(dev, &srq_table->table, *srqn);
+
+err_out:
+	mlx4_bitmap_free(&srq_table->bitmap, *srqn, MLX4_NO_RR);
+	return err;
+}
+
+static int mlx4_srq_alloc_icm(struct mlx4_dev *dev, int *srqn)
+{
+	u64 out_param;
+	int err;
+
+	if (mlx4_is_mfunc(dev)) {
+		err = mlx4_cmd_imm(dev, 0, &out_param, RES_SRQ,
+				   RES_OP_RESERVE_AND_MAP,
+				   MLX4_CMD_ALLOC_RES,
+				   MLX4_CMD_TIME_CLASS_A, MLX4_CMD_WRAPPED);
+		if (!err)
+			*srqn = get_param_l(&out_param);
+
+		return err;
+	}
+	return __mlx4_srq_alloc_icm(dev, srqn);
+}
+
+void __mlx4_srq_free_icm(struct mlx4_dev *dev, int srqn)
+{
+	struct mlx4_srq_table *srq_table = &mlx4_priv(dev)->srq_table;
+
+	mlx4_table_put(dev, &srq_table->cmpt_table, srqn);
+	mlx4_table_put(dev, &srq_table->table, srqn);
+	mlx4_bitmap_free(&srq_table->bitmap, srqn, MLX4_NO_RR);
+}
+
+static void mlx4_srq_free_icm(struct mlx4_dev *dev, int srqn)
+{
+	u64 in_param = 0;
+
+	if (mlx4_is_mfunc(dev)) {
+		set_param_l(&in_param, srqn);
+		if (mlx4_cmd(dev, in_param, RES_SRQ, RES_OP_RESERVE_AND_MAP,
+			     MLX4_CMD_FREE_RES,
+			     MLX4_CMD_TIME_CLASS_A, MLX4_CMD_WRAPPED))
+			mlx4_warn(dev, "Failed freeing cq:%d\n", srqn);
+		return;
+	}
+	__mlx4_srq_free_icm(dev, srqn);
+}
+
+int mlx4_srq_alloc(struct mlx4_dev *dev, u32 pdn, u32 cqn, u16 xrcd,
+		   struct mlx4_mtt *mtt, u64 db_rec, struct mlx4_srq *srq)
+{
+	struct mlx4_srq_table *srq_table = &mlx4_priv(dev)->srq_table;
+	struct mlx4_cmd_mailbox *mailbox;
+	struct mlx4_srq_context *srq_context;
+	u64 mtt_addr;
+	int err;
+
+	err = mlx4_srq_alloc_icm(dev, &srq->srqn);
+	if (err)
+		return err;
+
+	spin_lock_irq(&srq_table->lock);
+	err = radix_tree_insert(&srq_table->tree, srq->srqn, srq);
+	spin_unlock_irq(&srq_table->lock);
+	if (err)
+		goto err_icm;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox)) {
+		err = PTR_ERR(mailbox);
+		goto err_radix;
+	}
+
+	srq_context = mailbox->buf;
+	srq_context->state_logsize_srqn = cpu_to_be32((ilog2(srq->max) << 24) |
+						      srq->srqn);
+	srq_context->logstride          = srq->wqe_shift - 4;
+	srq_context->xrcd		= cpu_to_be16(xrcd);
+	srq_context->pg_offset_cqn	= cpu_to_be32(cqn & 0xffffff);
+	srq_context->log_page_size      = mtt->page_shift - MLX4_ICM_PAGE_SHIFT;
+
+	mtt_addr = mlx4_mtt_addr(dev, mtt);
+	srq_context->mtt_base_addr_h    = mtt_addr >> 32;
+	srq_context->mtt_base_addr_l    = cpu_to_be32(mtt_addr & 0xffffffff);
+	srq_context->pd			= cpu_to_be32(pdn);
+	srq_context->db_rec_addr        = cpu_to_be64(db_rec);
+
+	err = mlx4_SW2HW_SRQ(dev, mailbox, srq->srqn);
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	if (err)
+		goto err_radix;
+
+	refcount_set(&srq->refcount, 1);
+	init_completion(&srq->free);
+
+	return 0;
+
+err_radix:
+	spin_lock_irq(&srq_table->lock);
+	radix_tree_delete(&srq_table->tree, srq->srqn);
+	spin_unlock_irq(&srq_table->lock);
+
+err_icm:
+	mlx4_srq_free_icm(dev, srq->srqn);
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx4_srq_alloc);
+
+void mlx4_srq_free(struct mlx4_dev *dev, struct mlx4_srq *srq)
+{
+	struct mlx4_srq_table *srq_table = &mlx4_priv(dev)->srq_table;
+	int err;
+
+	err = mlx4_HW2SW_SRQ(dev, NULL, srq->srqn);
+	if (err)
+		mlx4_warn(dev, "HW2SW_SRQ failed (%d) for SRQN %06x\n", err, srq->srqn);
+
+	spin_lock_irq(&srq_table->lock);
+	radix_tree_delete(&srq_table->tree, srq->srqn);
+	spin_unlock_irq(&srq_table->lock);
+
+	if (refcount_dec_and_test(&srq->refcount))
+		complete(&srq->free);
+	wait_for_completion(&srq->free);
+
+	mlx4_srq_free_icm(dev, srq->srqn);
+}
+EXPORT_SYMBOL_GPL(mlx4_srq_free);
+
+int mlx4_srq_arm(struct mlx4_dev *dev, struct mlx4_srq *srq, int limit_watermark)
+{
+	return mlx4_ARM_SRQ(dev, srq->srqn, limit_watermark);
+}
+EXPORT_SYMBOL_GPL(mlx4_srq_arm);
+
+int mlx4_srq_query(struct mlx4_dev *dev, struct mlx4_srq *srq, int *limit_watermark)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	struct mlx4_srq_context *srq_context;
+	int err;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+
+	srq_context = mailbox->buf;
+
+	err = mlx4_QUERY_SRQ(dev, mailbox, srq->srqn);
+	if (err)
+		goto err_out;
+	*limit_watermark = be16_to_cpu(srq_context->limit_watermark);
+
+err_out:
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx4_srq_query);
+
+int mlx4_init_srq_table(struct mlx4_dev *dev)
+{
+	struct mlx4_srq_table *srq_table = &mlx4_priv(dev)->srq_table;
+
+	spin_lock_init(&srq_table->lock);
+	INIT_RADIX_TREE(&srq_table->tree, GFP_ATOMIC);
+	if (mlx4_is_slave(dev))
+		return 0;
+
+	return mlx4_bitmap_init(&srq_table->bitmap, dev->caps.num_srqs,
+				dev->caps.num_srqs - 1, dev->caps.reserved_srqs, 0);
+}
+
+void mlx4_cleanup_srq_table(struct mlx4_dev *dev)
+{
+	if (mlx4_is_slave(dev))
+		return;
+	mlx4_bitmap_cleanup(&mlx4_priv(dev)->srq_table.bitmap);
+}
+
+struct mlx4_srq *mlx4_srq_lookup(struct mlx4_dev *dev, u32 srqn)
+{
+	struct mlx4_srq_table *srq_table = &mlx4_priv(dev)->srq_table;
+	struct mlx4_srq *srq;
+
+	rcu_read_lock();
+	srq = radix_tree_lookup(&srq_table->tree,
+				srqn & (dev->caps.num_srqs - 1));
+	rcu_read_unlock();
+
+	return srq;
+}
+EXPORT_SYMBOL_GPL(mlx4_srq_lookup);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/Kconfig b/drivers/net/ethernet/mellanox/mlx5/core/Kconfig
new file mode 100644
index 000000000..26685fd0f
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/Kconfig
@@ -0,0 +1,190 @@
+# SPDX-License-Identifier: GPL-2.0-only
+#
+# Mellanox driver configuration
+#
+
+config MLX5_CORE
+	tristate "Mellanox 5th generation network adapters (ConnectX series) core driver"
+	depends on PCI
+	select AUXILIARY_BUS
+	select NET_DEVLINK
+	depends on VXLAN || !VXLAN
+	depends on MLXFW || !MLXFW
+	depends on PTP_1588_CLOCK_OPTIONAL
+	depends on PCI_HYPERV_INTERFACE || !PCI_HYPERV_INTERFACE
+	help
+	  Core driver for low level functionality of the ConnectX-4 and
+	  Connect-IB cards by Mellanox Technologies.
+
+config MLX5_FPGA
+	bool "Mellanox Technologies Innova support"
+	depends on MLX5_CORE
+	help
+	  Build support for the Innova family of network cards by Mellanox
+	  Technologies. Innova network cards are comprised of a ConnectX chip
+	  and an FPGA chip on one board. If you select this option, the
+	  mlx5_core driver will include the Innova FPGA core and allow building
+	  sandbox-specific client drivers.
+
+config MLX5_CORE_EN
+	bool "Mellanox 5th generation network adapters (ConnectX series) Ethernet support"
+	depends on NETDEVICES && ETHERNET && INET && PCI && MLX5_CORE
+	select PAGE_POOL
+	select DIMLIB
+	help
+	  Ethernet support in Mellanox Technologies ConnectX-4 NIC.
+
+config MLX5_EN_ARFS
+	bool "Mellanox MLX5 ethernet accelerated receive flow steering (ARFS) support"
+	depends on MLX5_CORE_EN && RFS_ACCEL
+	default y
+	help
+	  Mellanox MLX5 ethernet hardware-accelerated receive flow steering support,
+	  Enables ethernet netdevice arfs support and ntuple filtering.
+
+config MLX5_EN_RXNFC
+	bool "Mellanox MLX5 ethernet rx nfc flow steering support"
+	depends on MLX5_CORE_EN
+	default y
+	help
+	  Mellanox MLX5 ethernet rx nfc flow steering support
+	  Enables ethtool receive network flow classification, which allows user defined
+	  flow rules to direct traffic into arbitrary rx queue via ethtool set/get_rxnfc
+	  API.
+
+config MLX5_MPFS
+	bool "Mellanox Technologies MLX5 MPFS support"
+	depends on MLX5_CORE_EN
+	default y
+	help
+	  Mellanox Technologies Ethernet Multi-Physical Function Switch (MPFS)
+	  support in ConnectX NIC. MPFs is required for when multi-PF configuration
+	  is enabled to allow passing user configured unicast MAC addresses to the
+	  requesting PF.
+
+config MLX5_ESWITCH
+	bool "Mellanox Technologies MLX5 SRIOV E-Switch support"
+	depends on MLX5_CORE_EN && NET_SWITCHDEV
+	default y
+	help
+	  Mellanox Technologies Ethernet SRIOV E-Switch support in ConnectX NIC.
+	  E-Switch provides internal SRIOV packet steering and switching for the
+	  enabled VFs and PF in two available modes:
+	        Legacy SRIOV mode (L2 mac vlan steering based).
+	        Switchdev mode (eswitch offloads).
+
+config MLX5_BRIDGE
+	bool
+	depends on MLX5_ESWITCH && BRIDGE
+	default y
+	help
+	  mlx5 ConnectX offloads support for Ethernet Bridging (BRIDGE).
+	  Enable adding representors of mlx5 uplink and VF ports to Bridge and
+	  offloading rules for traffic between such ports. Supports VLANs (trunk and
+	  access modes).
+
+config MLX5_CLS_ACT
+	bool "MLX5 TC classifier action support"
+	depends on MLX5_ESWITCH && NET_CLS_ACT
+	default y
+	help
+	  mlx5 ConnectX offloads support for TC classifier action (NET_CLS_ACT),
+	  works in both native NIC mode and Switchdev SRIOV mode.
+	  Actions get attached to a Hardware offloaded classifiers and are
+	  invoked after a successful classification. Actions are used to
+	  overwrite the classification result, instantly drop or redirect and/or
+	  reformat packets in wire speeds without involving the host cpu.
+
+	  If set to N, TC offloads in both NIC and switchdev modes will be disabled.
+	  If unsure, set to Y
+
+config MLX5_TC_CT
+	bool "MLX5 TC connection tracking offload support"
+	depends on MLX5_CLS_ACT && NF_FLOW_TABLE && NET_ACT_CT && NET_TC_SKB_EXT
+	default y
+	help
+	  Say Y here if you want to support offloading connection tracking rules
+	  via tc ct action.
+
+	  If unsure, set to Y
+
+config MLX5_TC_SAMPLE
+	bool "MLX5 TC sample offload support"
+	depends on MLX5_CLS_ACT
+	depends on PSAMPLE=y || PSAMPLE=n || MLX5_CORE=m
+	default y
+	help
+	  Say Y here if you want to support offloading sample rules via tc
+	  sample action.
+	  If set to N, will not be able to configure tc rules with sample
+	  action.
+
+	  If unsure, set to Y
+
+config MLX5_CORE_EN_DCB
+	bool "Data Center Bridging (DCB) Support"
+	default y
+	depends on MLX5_CORE_EN && DCB
+	help
+	  Say Y here if you want to use Data Center Bridging (DCB) in the
+	  driver.
+	  If set to N, will not be able to configure QoS and ratelimit attributes.
+	  This flag is depended on the kernel's DCB support.
+
+	  If unsure, set to Y
+
+config MLX5_CORE_IPOIB
+	bool "Mellanox 5th generation network adapters (connectX series) IPoIB offloads support"
+	depends on MLX5_CORE_EN
+	help
+	  MLX5 IPoIB offloads & acceleration support.
+
+config MLX5_EN_MACSEC
+	bool "Connect-X support for MACSec offload"
+	depends on MLX5_CORE_EN
+	depends on MACSEC
+	default n
+	help
+	  Build support for MACsec cryptography-offload acceleration in the NIC.
+
+config MLX5_EN_IPSEC
+	bool "Mellanox Technologies IPsec Connect-X support"
+	depends on MLX5_CORE_EN
+	depends on XFRM_OFFLOAD
+	depends on INET_ESP_OFFLOAD || INET6_ESP_OFFLOAD
+	help
+	  Build support for IPsec cryptography-offload acceleration in the NIC.
+
+config MLX5_EN_TLS
+	bool "Mellanox Technologies TLS Connect-X support"
+	depends on TLS_DEVICE
+	depends on TLS=y || MLX5_CORE=m
+	depends on MLX5_CORE_EN
+	help
+	Build support for TLS cryptography-offload acceleration in the NIC.
+
+config MLX5_SW_STEERING
+	bool "Mellanox Technologies software-managed steering"
+	depends on MLX5_CORE_EN && MLX5_ESWITCH
+	select CRC32
+	default y
+	help
+	Build support for software-managed steering in the NIC.
+
+config MLX5_SF
+	bool "Mellanox Technologies subfunction device support using auxiliary device"
+	depends on MLX5_CORE && MLX5_CORE_EN
+	help
+	Build support for subfuction device in the NIC. A Mellanox subfunction
+	device can support RDMA, netdevice and vdpa device.
+	It is similar to a SRIOV VF but it doesn't require SRIOV support.
+
+config MLX5_SF_MANAGER
+	bool
+	depends on MLX5_SF && MLX5_ESWITCH
+	default y
+	help
+	Build support for subfuction port in the NIC. A Mellanox subfunction
+	port is managed through devlink.  A subfunction supports RDMA, netdevice
+	and vdpa device. It is similar to a SRIOV VF but it doesn't require
+	SRIOV support.
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/Makefile b/drivers/net/ethernet/mellanox/mlx5/core/Makefile
new file mode 100644
index 000000000..a22c32aab
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/Makefile
@@ -0,0 +1,123 @@
+# SPDX-License-Identifier: GPL-2.0
+#
+# Makefile for Mellanox 5th generation network adapters
+# (ConnectX series) core & netdev driver
+#
+
+subdir-ccflags-y += -I$(src)
+
+obj-$(CONFIG_MLX5_CORE) += mlx5_core.o
+
+#
+# mlx5 core basic
+#
+mlx5_core-y :=	main.o cmd.o debugfs.o fw.o eq.o uar.o pagealloc.o \
+		health.o mcg.o cq.o alloc.o port.o mr.o pd.o \
+		transobj.o vport.o sriov.o fs_cmd.o fs_core.o pci_irq.o \
+		fs_counters.o fs_ft_pool.o rl.o lag/debugfs.o lag/lag.o dev.o events.o wq.o lib/gid.o \
+		lib/devcom.o lib/pci_vsc.o lib/dm.o lib/fs_ttc.o diag/fs_tracepoint.o \
+		diag/fw_tracer.o diag/crdump.o devlink.o diag/rsc_dump.o \
+		fw_reset.o qos.o lib/tout.o lib/aso.o
+
+#
+# Netdev basic
+#
+mlx5_core-$(CONFIG_MLX5_CORE_EN) += en/rqt.o en/tir.o en/rss.o en/rx_res.o \
+		en/channels.o en_main.o en_common.o en_fs.o en_ethtool.o \
+		en_tx.o en_rx.o en_dim.o en_txrx.o en/xdp.o en_stats.o \
+		en_selftest.o en/port.o en/monitor_stats.o en/health.o \
+		en/reporter_tx.o en/reporter_rx.o en/params.o en/xsk/pool.o \
+		en/xsk/setup.o en/xsk/rx.o en/xsk/tx.o en/devlink.o en/ptp.o \
+		en/qos.o en/htb.o en/trap.o en/fs_tt_redirect.o en/selq.o \
+		lib/crypto.o
+
+#
+# Netdev extra
+#
+mlx5_core-$(CONFIG_MLX5_EN_ARFS)     += en_arfs.o
+mlx5_core-$(CONFIG_MLX5_EN_RXNFC)    += en_fs_ethtool.o
+mlx5_core-$(CONFIG_MLX5_CORE_EN_DCB) += en_dcbnl.o en/port_buffer.o
+mlx5_core-$(CONFIG_PCI_HYPERV_INTERFACE) += en/hv_vhca_stats.o
+mlx5_core-$(CONFIG_MLX5_ESWITCH)     += lag/mp.o lag/port_sel.o lib/geneve.o lib/port_tun.o \
+					en_rep.o en/rep/bond.o en/mod_hdr.o \
+					en/mapping.o lag/mpesw.o
+mlx5_core-$(CONFIG_MLX5_CLS_ACT)     += en_tc.o en/rep/tc.o en/rep/neigh.o \
+					lib/fs_chains.o en/tc_tun.o \
+					esw/indir_table.o en/tc_tun_encap.o \
+					en/tc_tun_vxlan.o en/tc_tun_gre.o en/tc_tun_geneve.o \
+					en/tc_tun_mplsoudp.o diag/en_tc_tracepoint.o \
+					en/tc/post_act.o en/tc/int_port.o en/tc/meter.o \
+					en/tc/post_meter.o
+
+mlx5_core-$(CONFIG_MLX5_CLS_ACT)     += en/tc/act/act.o en/tc/act/drop.o en/tc/act/trap.o \
+					en/tc/act/accept.o en/tc/act/mark.o en/tc/act/goto.o \
+					en/tc/act/tun.o en/tc/act/csum.o en/tc/act/pedit.o \
+					en/tc/act/vlan.o en/tc/act/vlan_mangle.o en/tc/act/mpls.o \
+					en/tc/act/mirred.o en/tc/act/mirred_nic.o \
+					en/tc/act/ct.o en/tc/act/sample.o en/tc/act/ptype.o \
+					en/tc/act/redirect_ingress.o en/tc/act/police.o
+
+ifneq ($(CONFIG_MLX5_TC_CT),)
+	mlx5_core-y			     += en/tc_ct.o en/tc/ct_fs_dmfs.o
+	mlx5_core-$(CONFIG_MLX5_SW_STEERING) += en/tc/ct_fs_smfs.o
+endif
+
+mlx5_core-$(CONFIG_MLX5_TC_SAMPLE)   += en/tc/sample.o
+
+#
+# Core extra
+#
+mlx5_core-$(CONFIG_MLX5_ESWITCH)   += eswitch.o eswitch_offloads.o eswitch_offloads_termtbl.o \
+				      ecpf.o rdma.o esw/legacy.o \
+				      esw/debugfs.o esw/devlink_port.o esw/vporttbl.o esw/qos.o
+
+mlx5_core-$(CONFIG_MLX5_ESWITCH)   += esw/acl/helper.o \
+				      esw/acl/egress_lgcy.o esw/acl/egress_ofld.o \
+				      esw/acl/ingress_lgcy.o esw/acl/ingress_ofld.o
+
+mlx5_core-$(CONFIG_MLX5_BRIDGE)    += esw/bridge.o en/rep/bridge.o
+
+mlx5_core-$(CONFIG_MLX5_MPFS)      += lib/mpfs.o
+mlx5_core-$(CONFIG_VXLAN)          += lib/vxlan.o
+mlx5_core-$(CONFIG_PTP_1588_CLOCK) += lib/clock.o
+mlx5_core-$(CONFIG_PCI_HYPERV_INTERFACE) += lib/hv.o lib/hv_vhca.o
+
+#
+# Ipoib netdev
+#
+mlx5_core-$(CONFIG_MLX5_CORE_IPOIB) += ipoib/ipoib.o ipoib/ethtool.o ipoib/ipoib_vlan.o
+
+#
+# Accelerations & FPGA
+#
+mlx5_core-$(CONFIG_MLX5_FPGA) += fpga/cmd.o fpga/core.o fpga/conn.o fpga/sdk.o
+
+mlx5_core-$(CONFIG_MLX5_EN_MACSEC) += en_accel/macsec.o en_accel/macsec_fs.o \
+				      en_accel/macsec_stats.o
+
+mlx5_core-$(CONFIG_MLX5_EN_IPSEC) += en_accel/ipsec.o en_accel/ipsec_rxtx.o \
+				     en_accel/ipsec_stats.o en_accel/ipsec_fs.o \
+				     en_accel/ipsec_offload.o
+
+mlx5_core-$(CONFIG_MLX5_EN_TLS) += en_accel/ktls_stats.o \
+				   en_accel/fs_tcp.o en_accel/ktls.o en_accel/ktls_txrx.o \
+				   en_accel/ktls_tx.o en_accel/ktls_rx.o
+
+mlx5_core-$(CONFIG_MLX5_SW_STEERING) += steering/dr_domain.o steering/dr_table.o \
+					steering/dr_matcher.o steering/dr_rule.o \
+					steering/dr_icm_pool.o steering/dr_buddy.o \
+					steering/dr_ste.o steering/dr_send.o \
+					steering/dr_ste_v0.o steering/dr_ste_v1.o \
+					steering/dr_ste_v2.o \
+					steering/dr_cmd.o steering/dr_fw.o \
+					steering/dr_action.o steering/fs_dr.o \
+					steering/dr_dbg.o lib/smfs.o
+#
+# SF device
+#
+mlx5_core-$(CONFIG_MLX5_SF) += sf/vhca_event.o sf/dev/dev.o sf/dev/driver.o irq_affinity.o
+
+#
+# SF manager
+#
+mlx5_core-$(CONFIG_MLX5_SF_MANAGER) += sf/cmd.o sf/hw_table.o sf/devlink.o
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/alloc.c b/drivers/net/ethernet/mellanox/mlx5/core/alloc.c
new file mode 100644
index 000000000..6aca004e8
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/alloc.c
@@ -0,0 +1,250 @@
+/*
+ * Copyright (c) 2013-2015, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/errno.h>
+#include <linux/slab.h>
+#include <linux/mm.h>
+#include <linux/export.h>
+#include <linux/bitmap.h>
+#include <linux/dma-mapping.h>
+#include <linux/vmalloc.h>
+#include <linux/mlx5/driver.h>
+
+#include "mlx5_core.h"
+
+struct mlx5_db_pgdir {
+	struct list_head	list;
+	unsigned long	       *bitmap;
+	__be32		       *db_page;
+	dma_addr_t		db_dma;
+};
+
+/* Handling for queue buffers -- we allocate a bunch of memory and
+ * register it in a memory region at HCA virtual address 0.
+ */
+
+static void *mlx5_dma_zalloc_coherent_node(struct mlx5_core_dev *dev,
+					   size_t size, dma_addr_t *dma_handle,
+					   int node)
+{
+	struct device *device = mlx5_core_dma_dev(dev);
+	struct mlx5_priv *priv = &dev->priv;
+	int original_node;
+	void *cpu_handle;
+
+	mutex_lock(&priv->alloc_mutex);
+	original_node = dev_to_node(device);
+	set_dev_node(device, node);
+	cpu_handle = dma_alloc_coherent(device, size, dma_handle,
+					GFP_KERNEL);
+	set_dev_node(device, original_node);
+	mutex_unlock(&priv->alloc_mutex);
+	return cpu_handle;
+}
+
+int mlx5_frag_buf_alloc_node(struct mlx5_core_dev *dev, int size,
+			     struct mlx5_frag_buf *buf, int node)
+{
+	int i;
+
+	buf->size = size;
+	buf->npages = DIV_ROUND_UP(size, PAGE_SIZE);
+	buf->page_shift = PAGE_SHIFT;
+	buf->frags = kcalloc(buf->npages, sizeof(struct mlx5_buf_list),
+			     GFP_KERNEL);
+	if (!buf->frags)
+		goto err_out;
+
+	for (i = 0; i < buf->npages; i++) {
+		struct mlx5_buf_list *frag = &buf->frags[i];
+		int frag_sz = min_t(int, size, PAGE_SIZE);
+
+		frag->buf = mlx5_dma_zalloc_coherent_node(dev, frag_sz,
+							  &frag->map, node);
+		if (!frag->buf)
+			goto err_free_buf;
+		if (frag->map & ((1 << buf->page_shift) - 1)) {
+			dma_free_coherent(mlx5_core_dma_dev(dev), frag_sz,
+					  buf->frags[i].buf, buf->frags[i].map);
+			mlx5_core_warn(dev, "unexpected map alignment: %pad, page_shift=%d\n",
+				       &frag->map, buf->page_shift);
+			goto err_free_buf;
+		}
+		size -= frag_sz;
+	}
+
+	return 0;
+
+err_free_buf:
+	while (i--)
+		dma_free_coherent(mlx5_core_dma_dev(dev), PAGE_SIZE, buf->frags[i].buf,
+				  buf->frags[i].map);
+	kfree(buf->frags);
+err_out:
+	return -ENOMEM;
+}
+EXPORT_SYMBOL_GPL(mlx5_frag_buf_alloc_node);
+
+void mlx5_frag_buf_free(struct mlx5_core_dev *dev, struct mlx5_frag_buf *buf)
+{
+	int size = buf->size;
+	int i;
+
+	for (i = 0; i < buf->npages; i++) {
+		int frag_sz = min_t(int, size, PAGE_SIZE);
+
+		dma_free_coherent(mlx5_core_dma_dev(dev), frag_sz, buf->frags[i].buf,
+				  buf->frags[i].map);
+		size -= frag_sz;
+	}
+	kfree(buf->frags);
+}
+EXPORT_SYMBOL_GPL(mlx5_frag_buf_free);
+
+static struct mlx5_db_pgdir *mlx5_alloc_db_pgdir(struct mlx5_core_dev *dev,
+						 int node)
+{
+	u32 db_per_page = PAGE_SIZE / cache_line_size();
+	struct mlx5_db_pgdir *pgdir;
+
+	pgdir = kzalloc_node(sizeof(*pgdir), GFP_KERNEL, node);
+	if (!pgdir)
+		return NULL;
+
+	pgdir->bitmap = bitmap_zalloc_node(db_per_page, GFP_KERNEL, node);
+	if (!pgdir->bitmap) {
+		kfree(pgdir);
+		return NULL;
+	}
+
+	bitmap_fill(pgdir->bitmap, db_per_page);
+
+	pgdir->db_page = mlx5_dma_zalloc_coherent_node(dev, PAGE_SIZE,
+						       &pgdir->db_dma, node);
+	if (!pgdir->db_page) {
+		bitmap_free(pgdir->bitmap);
+		kfree(pgdir);
+		return NULL;
+	}
+
+	return pgdir;
+}
+
+static int mlx5_alloc_db_from_pgdir(struct mlx5_db_pgdir *pgdir,
+				    struct mlx5_db *db)
+{
+	u32 db_per_page = PAGE_SIZE / cache_line_size();
+	int offset;
+	int i;
+
+	i = find_first_bit(pgdir->bitmap, db_per_page);
+	if (i >= db_per_page)
+		return -ENOMEM;
+
+	__clear_bit(i, pgdir->bitmap);
+
+	db->u.pgdir = pgdir;
+	db->index   = i;
+	offset = db->index * cache_line_size();
+	db->db      = pgdir->db_page + offset / sizeof(*pgdir->db_page);
+	db->dma     = pgdir->db_dma  + offset;
+
+	db->db[0] = 0;
+	db->db[1] = 0;
+
+	return 0;
+}
+
+int mlx5_db_alloc_node(struct mlx5_core_dev *dev, struct mlx5_db *db, int node)
+{
+	struct mlx5_db_pgdir *pgdir;
+	int ret = 0;
+
+	mutex_lock(&dev->priv.pgdir_mutex);
+
+	list_for_each_entry(pgdir, &dev->priv.pgdir_list, list)
+		if (!mlx5_alloc_db_from_pgdir(pgdir, db))
+			goto out;
+
+	pgdir = mlx5_alloc_db_pgdir(dev, node);
+	if (!pgdir) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	list_add(&pgdir->list, &dev->priv.pgdir_list);
+
+	/* This should never fail -- we just allocated an empty page: */
+	WARN_ON(mlx5_alloc_db_from_pgdir(pgdir, db));
+
+out:
+	mutex_unlock(&dev->priv.pgdir_mutex);
+
+	return ret;
+}
+EXPORT_SYMBOL_GPL(mlx5_db_alloc_node);
+
+void mlx5_db_free(struct mlx5_core_dev *dev, struct mlx5_db *db)
+{
+	u32 db_per_page = PAGE_SIZE / cache_line_size();
+
+	mutex_lock(&dev->priv.pgdir_mutex);
+
+	__set_bit(db->index, db->u.pgdir->bitmap);
+
+	if (bitmap_full(db->u.pgdir->bitmap, db_per_page)) {
+		dma_free_coherent(mlx5_core_dma_dev(dev), PAGE_SIZE,
+				  db->u.pgdir->db_page, db->u.pgdir->db_dma);
+		list_del(&db->u.pgdir->list);
+		bitmap_free(db->u.pgdir->bitmap);
+		kfree(db->u.pgdir);
+	}
+
+	mutex_unlock(&dev->priv.pgdir_mutex);
+}
+EXPORT_SYMBOL_GPL(mlx5_db_free);
+
+void mlx5_fill_page_frag_array_perm(struct mlx5_frag_buf *buf, __be64 *pas, u8 perm)
+{
+	int i;
+
+	WARN_ON(perm & 0xfc);
+	for (i = 0; i < buf->npages; i++)
+		pas[i] = cpu_to_be64(buf->frags[i].map | perm);
+}
+EXPORT_SYMBOL_GPL(mlx5_fill_page_frag_array_perm);
+
+void mlx5_fill_page_frag_array(struct mlx5_frag_buf *buf, __be64 *pas)
+{
+	mlx5_fill_page_frag_array_perm(buf, pas, 0);
+}
+EXPORT_SYMBOL_GPL(mlx5_fill_page_frag_array);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/cmd.c b/drivers/net/ethernet/mellanox/mlx5/core/cmd.c
new file mode 100644
index 000000000..ac6a0785b
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/cmd.c
@@ -0,0 +1,2311 @@
+/*
+ * Copyright (c) 2013-2016, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/highmem.h>
+#include <linux/errno.h>
+#include <linux/pci.h>
+#include <linux/dma-mapping.h>
+#include <linux/slab.h>
+#include <linux/delay.h>
+#include <linux/random.h>
+#include <linux/io-mapping.h>
+#include <linux/mlx5/driver.h>
+#include <linux/mlx5/eq.h>
+#include <linux/debugfs.h>
+
+#include "mlx5_core.h"
+#include "lib/eq.h"
+#include "lib/tout.h"
+#define CREATE_TRACE_POINTS
+#include "diag/cmd_tracepoint.h"
+
+struct mlx5_ifc_mbox_out_bits {
+	u8         status[0x8];
+	u8         reserved_at_8[0x18];
+
+	u8         syndrome[0x20];
+
+	u8         reserved_at_40[0x40];
+};
+
+struct mlx5_ifc_mbox_in_bits {
+	u8         opcode[0x10];
+	u8         uid[0x10];
+
+	u8         reserved_at_20[0x10];
+	u8         op_mod[0x10];
+
+	u8         reserved_at_40[0x40];
+};
+
+enum {
+	CMD_IF_REV = 5,
+};
+
+enum {
+	CMD_MODE_POLLING,
+	CMD_MODE_EVENTS
+};
+
+enum {
+	MLX5_CMD_DELIVERY_STAT_OK			= 0x0,
+	MLX5_CMD_DELIVERY_STAT_SIGNAT_ERR		= 0x1,
+	MLX5_CMD_DELIVERY_STAT_TOK_ERR			= 0x2,
+	MLX5_CMD_DELIVERY_STAT_BAD_BLK_NUM_ERR		= 0x3,
+	MLX5_CMD_DELIVERY_STAT_OUT_PTR_ALIGN_ERR	= 0x4,
+	MLX5_CMD_DELIVERY_STAT_IN_PTR_ALIGN_ERR		= 0x5,
+	MLX5_CMD_DELIVERY_STAT_FW_ERR			= 0x6,
+	MLX5_CMD_DELIVERY_STAT_IN_LENGTH_ERR		= 0x7,
+	MLX5_CMD_DELIVERY_STAT_OUT_LENGTH_ERR		= 0x8,
+	MLX5_CMD_DELIVERY_STAT_RES_FLD_NOT_CLR_ERR	= 0x9,
+	MLX5_CMD_DELIVERY_STAT_CMD_DESCR_ERR		= 0x10,
+};
+
+static u16 in_to_opcode(void *in)
+{
+	return MLX5_GET(mbox_in, in, opcode);
+}
+
+/* Returns true for opcodes that might be triggered very frequently and throttle
+ * the command interface. Limit their command slots usage.
+ */
+static bool mlx5_cmd_is_throttle_opcode(u16 op)
+{
+	switch (op) {
+	case MLX5_CMD_OP_CREATE_GENERAL_OBJECT:
+	case MLX5_CMD_OP_DESTROY_GENERAL_OBJECT:
+	case MLX5_CMD_OP_MODIFY_GENERAL_OBJECT:
+	case MLX5_CMD_OP_QUERY_GENERAL_OBJECT:
+		return true;
+	}
+	return false;
+}
+
+static struct mlx5_cmd_work_ent *
+cmd_alloc_ent(struct mlx5_cmd *cmd, struct mlx5_cmd_msg *in,
+	      struct mlx5_cmd_msg *out, void *uout, int uout_size,
+	      mlx5_cmd_cbk_t cbk, void *context, int page_queue)
+{
+	gfp_t alloc_flags = cbk ? GFP_ATOMIC : GFP_KERNEL;
+	struct mlx5_cmd_work_ent *ent;
+
+	ent = kzalloc(sizeof(*ent), alloc_flags);
+	if (!ent)
+		return ERR_PTR(-ENOMEM);
+
+	ent->idx	= -EINVAL;
+	ent->in		= in;
+	ent->out	= out;
+	ent->uout	= uout;
+	ent->uout_size	= uout_size;
+	ent->callback	= cbk;
+	ent->context	= context;
+	ent->cmd	= cmd;
+	ent->page_queue = page_queue;
+	ent->op         = in_to_opcode(in->first.data);
+	refcount_set(&ent->refcnt, 1);
+
+	return ent;
+}
+
+static void cmd_free_ent(struct mlx5_cmd_work_ent *ent)
+{
+	kfree(ent);
+}
+
+static u8 alloc_token(struct mlx5_cmd *cmd)
+{
+	u8 token;
+
+	spin_lock(&cmd->token_lock);
+	cmd->token++;
+	if (cmd->token == 0)
+		cmd->token++;
+	token = cmd->token;
+	spin_unlock(&cmd->token_lock);
+
+	return token;
+}
+
+static int cmd_alloc_index(struct mlx5_cmd *cmd, struct mlx5_cmd_work_ent *ent)
+{
+	unsigned long flags;
+	int ret;
+
+	spin_lock_irqsave(&cmd->alloc_lock, flags);
+	ret = find_first_bit(&cmd->vars.bitmask, cmd->vars.max_reg_cmds);
+	if (ret < cmd->vars.max_reg_cmds) {
+		clear_bit(ret, &cmd->vars.bitmask);
+		ent->idx = ret;
+		cmd->ent_arr[ent->idx] = ent;
+	}
+	spin_unlock_irqrestore(&cmd->alloc_lock, flags);
+
+	return ret < cmd->vars.max_reg_cmds ? ret : -ENOMEM;
+}
+
+static void cmd_free_index(struct mlx5_cmd *cmd, int idx)
+{
+	lockdep_assert_held(&cmd->alloc_lock);
+	set_bit(idx, &cmd->vars.bitmask);
+}
+
+static void cmd_ent_get(struct mlx5_cmd_work_ent *ent)
+{
+	refcount_inc(&ent->refcnt);
+}
+
+static void cmd_ent_put(struct mlx5_cmd_work_ent *ent)
+{
+	struct mlx5_cmd *cmd = ent->cmd;
+	unsigned long flags;
+
+	spin_lock_irqsave(&cmd->alloc_lock, flags);
+	if (!refcount_dec_and_test(&ent->refcnt))
+		goto out;
+
+	if (ent->idx >= 0) {
+		cmd_free_index(cmd, ent->idx);
+		up(ent->page_queue ? &cmd->vars.pages_sem : &cmd->vars.sem);
+	}
+
+	cmd_free_ent(ent);
+out:
+	spin_unlock_irqrestore(&cmd->alloc_lock, flags);
+}
+
+static struct mlx5_cmd_layout *get_inst(struct mlx5_cmd *cmd, int idx)
+{
+	return cmd->cmd_buf + (idx << cmd->vars.log_stride);
+}
+
+static int mlx5_calc_cmd_blocks(struct mlx5_cmd_msg *msg)
+{
+	int size = msg->len;
+	int blen = size - min_t(int, sizeof(msg->first.data), size);
+
+	return DIV_ROUND_UP(blen, MLX5_CMD_DATA_BLOCK_SIZE);
+}
+
+static u8 xor8_buf(void *buf, size_t offset, int len)
+{
+	u8 *ptr = buf;
+	u8 sum = 0;
+	int i;
+	int end = len + offset;
+
+	for (i = offset; i < end; i++)
+		sum ^= ptr[i];
+
+	return sum;
+}
+
+static int verify_block_sig(struct mlx5_cmd_prot_block *block)
+{
+	size_t rsvd0_off = offsetof(struct mlx5_cmd_prot_block, rsvd0);
+	int xor_len = sizeof(*block) - sizeof(block->data) - 1;
+
+	if (xor8_buf(block, rsvd0_off, xor_len) != 0xff)
+		return -EHWPOISON;
+
+	if (xor8_buf(block, 0, sizeof(*block)) != 0xff)
+		return -EHWPOISON;
+
+	return 0;
+}
+
+static void calc_block_sig(struct mlx5_cmd_prot_block *block)
+{
+	int ctrl_xor_len = sizeof(*block) - sizeof(block->data) - 2;
+	size_t rsvd0_off = offsetof(struct mlx5_cmd_prot_block, rsvd0);
+
+	block->ctrl_sig = ~xor8_buf(block, rsvd0_off, ctrl_xor_len);
+	block->sig = ~xor8_buf(block, 0, sizeof(*block) - 1);
+}
+
+static void calc_chain_sig(struct mlx5_cmd_msg *msg)
+{
+	struct mlx5_cmd_mailbox *next = msg->next;
+	int n = mlx5_calc_cmd_blocks(msg);
+	int i = 0;
+
+	for (i = 0; i < n && next; i++)  {
+		calc_block_sig(next->buf);
+		next = next->next;
+	}
+}
+
+static void set_signature(struct mlx5_cmd_work_ent *ent, int csum)
+{
+	ent->lay->sig = ~xor8_buf(ent->lay, 0,  sizeof(*ent->lay));
+	if (csum) {
+		calc_chain_sig(ent->in);
+		calc_chain_sig(ent->out);
+	}
+}
+
+static void poll_timeout(struct mlx5_cmd_work_ent *ent)
+{
+	struct mlx5_core_dev *dev = container_of(ent->cmd, struct mlx5_core_dev, cmd);
+	u64 cmd_to_ms = mlx5_tout_ms(dev, CMD);
+	unsigned long poll_end;
+	u8 own;
+
+	poll_end = jiffies + msecs_to_jiffies(cmd_to_ms + 1000);
+
+	do {
+		own = READ_ONCE(ent->lay->status_own);
+		if (!(own & CMD_OWNER_HW)) {
+			ent->ret = 0;
+			return;
+		}
+		cond_resched();
+	} while (time_before(jiffies, poll_end));
+
+	ent->ret = -ETIMEDOUT;
+}
+
+static int verify_signature(struct mlx5_cmd_work_ent *ent)
+{
+	struct mlx5_cmd_mailbox *next = ent->out->next;
+	int n = mlx5_calc_cmd_blocks(ent->out);
+	int err;
+	u8 sig;
+	int i = 0;
+
+	sig = xor8_buf(ent->lay, 0, sizeof(*ent->lay));
+	if (sig != 0xff)
+		return -EHWPOISON;
+
+	for (i = 0; i < n && next; i++) {
+		err = verify_block_sig(next->buf);
+		if (err)
+			return -EHWPOISON;
+
+		next = next->next;
+	}
+
+	return 0;
+}
+
+static void dump_buf(void *buf, int size, int data_only, int offset, int idx)
+{
+	__be32 *p = buf;
+	int i;
+
+	for (i = 0; i < size; i += 16) {
+		pr_debug("cmd[%d]: %03x: %08x %08x %08x %08x\n", idx, offset,
+			 be32_to_cpu(p[0]), be32_to_cpu(p[1]),
+			 be32_to_cpu(p[2]), be32_to_cpu(p[3]));
+		p += 4;
+		offset += 16;
+	}
+	if (!data_only)
+		pr_debug("\n");
+}
+
+static int mlx5_internal_err_ret_value(struct mlx5_core_dev *dev, u16 op,
+				       u32 *synd, u8 *status)
+{
+	*synd = 0;
+	*status = 0;
+
+	switch (op) {
+	case MLX5_CMD_OP_TEARDOWN_HCA:
+	case MLX5_CMD_OP_DISABLE_HCA:
+	case MLX5_CMD_OP_MANAGE_PAGES:
+	case MLX5_CMD_OP_DESTROY_MKEY:
+	case MLX5_CMD_OP_DESTROY_EQ:
+	case MLX5_CMD_OP_DESTROY_CQ:
+	case MLX5_CMD_OP_DESTROY_QP:
+	case MLX5_CMD_OP_DESTROY_PSV:
+	case MLX5_CMD_OP_DESTROY_SRQ:
+	case MLX5_CMD_OP_DESTROY_XRC_SRQ:
+	case MLX5_CMD_OP_DESTROY_XRQ:
+	case MLX5_CMD_OP_DESTROY_DCT:
+	case MLX5_CMD_OP_DEALLOC_Q_COUNTER:
+	case MLX5_CMD_OP_DESTROY_SCHEDULING_ELEMENT:
+	case MLX5_CMD_OP_DESTROY_QOS_PARA_VPORT:
+	case MLX5_CMD_OP_DEALLOC_PD:
+	case MLX5_CMD_OP_DEALLOC_UAR:
+	case MLX5_CMD_OP_DETACH_FROM_MCG:
+	case MLX5_CMD_OP_DEALLOC_XRCD:
+	case MLX5_CMD_OP_DEALLOC_TRANSPORT_DOMAIN:
+	case MLX5_CMD_OP_DELETE_VXLAN_UDP_DPORT:
+	case MLX5_CMD_OP_DELETE_L2_TABLE_ENTRY:
+	case MLX5_CMD_OP_DESTROY_LAG:
+	case MLX5_CMD_OP_DESTROY_VPORT_LAG:
+	case MLX5_CMD_OP_DESTROY_TIR:
+	case MLX5_CMD_OP_DESTROY_SQ:
+	case MLX5_CMD_OP_DESTROY_RQ:
+	case MLX5_CMD_OP_DESTROY_RMP:
+	case MLX5_CMD_OP_DESTROY_TIS:
+	case MLX5_CMD_OP_DESTROY_RQT:
+	case MLX5_CMD_OP_DESTROY_FLOW_TABLE:
+	case MLX5_CMD_OP_DESTROY_FLOW_GROUP:
+	case MLX5_CMD_OP_DELETE_FLOW_TABLE_ENTRY:
+	case MLX5_CMD_OP_DEALLOC_FLOW_COUNTER:
+	case MLX5_CMD_OP_2ERR_QP:
+	case MLX5_CMD_OP_2RST_QP:
+	case MLX5_CMD_OP_MODIFY_NIC_VPORT_CONTEXT:
+	case MLX5_CMD_OP_MODIFY_FLOW_TABLE:
+	case MLX5_CMD_OP_SET_FLOW_TABLE_ENTRY:
+	case MLX5_CMD_OP_SET_FLOW_TABLE_ROOT:
+	case MLX5_CMD_OP_DEALLOC_PACKET_REFORMAT_CONTEXT:
+	case MLX5_CMD_OP_DEALLOC_MODIFY_HEADER_CONTEXT:
+	case MLX5_CMD_OP_FPGA_DESTROY_QP:
+	case MLX5_CMD_OP_DESTROY_GENERAL_OBJECT:
+	case MLX5_CMD_OP_DEALLOC_MEMIC:
+	case MLX5_CMD_OP_PAGE_FAULT_RESUME:
+	case MLX5_CMD_OP_QUERY_ESW_FUNCTIONS:
+	case MLX5_CMD_OP_DEALLOC_SF:
+	case MLX5_CMD_OP_DESTROY_UCTX:
+	case MLX5_CMD_OP_DESTROY_UMEM:
+	case MLX5_CMD_OP_MODIFY_RQT:
+		return MLX5_CMD_STAT_OK;
+
+	case MLX5_CMD_OP_QUERY_HCA_CAP:
+	case MLX5_CMD_OP_QUERY_ADAPTER:
+	case MLX5_CMD_OP_INIT_HCA:
+	case MLX5_CMD_OP_ENABLE_HCA:
+	case MLX5_CMD_OP_QUERY_PAGES:
+	case MLX5_CMD_OP_SET_HCA_CAP:
+	case MLX5_CMD_OP_QUERY_ISSI:
+	case MLX5_CMD_OP_SET_ISSI:
+	case MLX5_CMD_OP_CREATE_MKEY:
+	case MLX5_CMD_OP_QUERY_MKEY:
+	case MLX5_CMD_OP_QUERY_SPECIAL_CONTEXTS:
+	case MLX5_CMD_OP_CREATE_EQ:
+	case MLX5_CMD_OP_QUERY_EQ:
+	case MLX5_CMD_OP_GEN_EQE:
+	case MLX5_CMD_OP_CREATE_CQ:
+	case MLX5_CMD_OP_QUERY_CQ:
+	case MLX5_CMD_OP_MODIFY_CQ:
+	case MLX5_CMD_OP_CREATE_QP:
+	case MLX5_CMD_OP_RST2INIT_QP:
+	case MLX5_CMD_OP_INIT2RTR_QP:
+	case MLX5_CMD_OP_RTR2RTS_QP:
+	case MLX5_CMD_OP_RTS2RTS_QP:
+	case MLX5_CMD_OP_SQERR2RTS_QP:
+	case MLX5_CMD_OP_QUERY_QP:
+	case MLX5_CMD_OP_SQD_RTS_QP:
+	case MLX5_CMD_OP_INIT2INIT_QP:
+	case MLX5_CMD_OP_CREATE_PSV:
+	case MLX5_CMD_OP_CREATE_SRQ:
+	case MLX5_CMD_OP_QUERY_SRQ:
+	case MLX5_CMD_OP_ARM_RQ:
+	case MLX5_CMD_OP_CREATE_XRC_SRQ:
+	case MLX5_CMD_OP_QUERY_XRC_SRQ:
+	case MLX5_CMD_OP_ARM_XRC_SRQ:
+	case MLX5_CMD_OP_CREATE_XRQ:
+	case MLX5_CMD_OP_QUERY_XRQ:
+	case MLX5_CMD_OP_ARM_XRQ:
+	case MLX5_CMD_OP_CREATE_DCT:
+	case MLX5_CMD_OP_DRAIN_DCT:
+	case MLX5_CMD_OP_QUERY_DCT:
+	case MLX5_CMD_OP_ARM_DCT_FOR_KEY_VIOLATION:
+	case MLX5_CMD_OP_QUERY_VPORT_STATE:
+	case MLX5_CMD_OP_MODIFY_VPORT_STATE:
+	case MLX5_CMD_OP_QUERY_ESW_VPORT_CONTEXT:
+	case MLX5_CMD_OP_MODIFY_ESW_VPORT_CONTEXT:
+	case MLX5_CMD_OP_QUERY_NIC_VPORT_CONTEXT:
+	case MLX5_CMD_OP_QUERY_ROCE_ADDRESS:
+	case MLX5_CMD_OP_SET_ROCE_ADDRESS:
+	case MLX5_CMD_OP_QUERY_HCA_VPORT_CONTEXT:
+	case MLX5_CMD_OP_MODIFY_HCA_VPORT_CONTEXT:
+	case MLX5_CMD_OP_QUERY_HCA_VPORT_GID:
+	case MLX5_CMD_OP_QUERY_HCA_VPORT_PKEY:
+	case MLX5_CMD_OP_QUERY_VNIC_ENV:
+	case MLX5_CMD_OP_QUERY_VPORT_COUNTER:
+	case MLX5_CMD_OP_ALLOC_Q_COUNTER:
+	case MLX5_CMD_OP_QUERY_Q_COUNTER:
+	case MLX5_CMD_OP_SET_MONITOR_COUNTER:
+	case MLX5_CMD_OP_ARM_MONITOR_COUNTER:
+	case MLX5_CMD_OP_SET_PP_RATE_LIMIT:
+	case MLX5_CMD_OP_QUERY_RATE_LIMIT:
+	case MLX5_CMD_OP_CREATE_SCHEDULING_ELEMENT:
+	case MLX5_CMD_OP_QUERY_SCHEDULING_ELEMENT:
+	case MLX5_CMD_OP_MODIFY_SCHEDULING_ELEMENT:
+	case MLX5_CMD_OP_CREATE_QOS_PARA_VPORT:
+	case MLX5_CMD_OP_ALLOC_PD:
+	case MLX5_CMD_OP_ALLOC_UAR:
+	case MLX5_CMD_OP_CONFIG_INT_MODERATION:
+	case MLX5_CMD_OP_ACCESS_REG:
+	case MLX5_CMD_OP_ATTACH_TO_MCG:
+	case MLX5_CMD_OP_GET_DROPPED_PACKET_LOG:
+	case MLX5_CMD_OP_MAD_IFC:
+	case MLX5_CMD_OP_QUERY_MAD_DEMUX:
+	case MLX5_CMD_OP_SET_MAD_DEMUX:
+	case MLX5_CMD_OP_NOP:
+	case MLX5_CMD_OP_ALLOC_XRCD:
+	case MLX5_CMD_OP_ALLOC_TRANSPORT_DOMAIN:
+	case MLX5_CMD_OP_QUERY_CONG_STATUS:
+	case MLX5_CMD_OP_MODIFY_CONG_STATUS:
+	case MLX5_CMD_OP_QUERY_CONG_PARAMS:
+	case MLX5_CMD_OP_MODIFY_CONG_PARAMS:
+	case MLX5_CMD_OP_QUERY_CONG_STATISTICS:
+	case MLX5_CMD_OP_ADD_VXLAN_UDP_DPORT:
+	case MLX5_CMD_OP_SET_L2_TABLE_ENTRY:
+	case MLX5_CMD_OP_QUERY_L2_TABLE_ENTRY:
+	case MLX5_CMD_OP_CREATE_LAG:
+	case MLX5_CMD_OP_MODIFY_LAG:
+	case MLX5_CMD_OP_QUERY_LAG:
+	case MLX5_CMD_OP_CREATE_VPORT_LAG:
+	case MLX5_CMD_OP_CREATE_TIR:
+	case MLX5_CMD_OP_MODIFY_TIR:
+	case MLX5_CMD_OP_QUERY_TIR:
+	case MLX5_CMD_OP_CREATE_SQ:
+	case MLX5_CMD_OP_MODIFY_SQ:
+	case MLX5_CMD_OP_QUERY_SQ:
+	case MLX5_CMD_OP_CREATE_RQ:
+	case MLX5_CMD_OP_MODIFY_RQ:
+	case MLX5_CMD_OP_QUERY_RQ:
+	case MLX5_CMD_OP_CREATE_RMP:
+	case MLX5_CMD_OP_MODIFY_RMP:
+	case MLX5_CMD_OP_QUERY_RMP:
+	case MLX5_CMD_OP_CREATE_TIS:
+	case MLX5_CMD_OP_MODIFY_TIS:
+	case MLX5_CMD_OP_QUERY_TIS:
+	case MLX5_CMD_OP_CREATE_RQT:
+	case MLX5_CMD_OP_QUERY_RQT:
+
+	case MLX5_CMD_OP_CREATE_FLOW_TABLE:
+	case MLX5_CMD_OP_QUERY_FLOW_TABLE:
+	case MLX5_CMD_OP_CREATE_FLOW_GROUP:
+	case MLX5_CMD_OP_QUERY_FLOW_GROUP:
+	case MLX5_CMD_OP_QUERY_FLOW_TABLE_ENTRY:
+	case MLX5_CMD_OP_ALLOC_FLOW_COUNTER:
+	case MLX5_CMD_OP_QUERY_FLOW_COUNTER:
+	case MLX5_CMD_OP_ALLOC_PACKET_REFORMAT_CONTEXT:
+	case MLX5_CMD_OP_ALLOC_MODIFY_HEADER_CONTEXT:
+	case MLX5_CMD_OP_FPGA_CREATE_QP:
+	case MLX5_CMD_OP_FPGA_MODIFY_QP:
+	case MLX5_CMD_OP_FPGA_QUERY_QP:
+	case MLX5_CMD_OP_FPGA_QUERY_QP_COUNTERS:
+	case MLX5_CMD_OP_CREATE_GENERAL_OBJECT:
+	case MLX5_CMD_OP_MODIFY_GENERAL_OBJECT:
+	case MLX5_CMD_OP_QUERY_GENERAL_OBJECT:
+	case MLX5_CMD_OP_CREATE_UCTX:
+	case MLX5_CMD_OP_CREATE_UMEM:
+	case MLX5_CMD_OP_ALLOC_MEMIC:
+	case MLX5_CMD_OP_MODIFY_XRQ:
+	case MLX5_CMD_OP_RELEASE_XRQ_ERROR:
+	case MLX5_CMD_OP_QUERY_VHCA_STATE:
+	case MLX5_CMD_OP_MODIFY_VHCA_STATE:
+	case MLX5_CMD_OP_ALLOC_SF:
+	case MLX5_CMD_OP_SUSPEND_VHCA:
+	case MLX5_CMD_OP_RESUME_VHCA:
+	case MLX5_CMD_OP_QUERY_VHCA_MIGRATION_STATE:
+	case MLX5_CMD_OP_SAVE_VHCA_STATE:
+	case MLX5_CMD_OP_LOAD_VHCA_STATE:
+		*status = MLX5_DRIVER_STATUS_ABORTED;
+		*synd = MLX5_DRIVER_SYND;
+		return -ENOLINK;
+	default:
+		mlx5_core_err(dev, "Unknown FW command (%d)\n", op);
+		return -EINVAL;
+	}
+}
+
+const char *mlx5_command_str(int command)
+{
+#define MLX5_COMMAND_STR_CASE(__cmd) case MLX5_CMD_OP_ ## __cmd: return #__cmd
+
+	switch (command) {
+	MLX5_COMMAND_STR_CASE(QUERY_HCA_CAP);
+	MLX5_COMMAND_STR_CASE(QUERY_ADAPTER);
+	MLX5_COMMAND_STR_CASE(INIT_HCA);
+	MLX5_COMMAND_STR_CASE(TEARDOWN_HCA);
+	MLX5_COMMAND_STR_CASE(ENABLE_HCA);
+	MLX5_COMMAND_STR_CASE(DISABLE_HCA);
+	MLX5_COMMAND_STR_CASE(QUERY_PAGES);
+	MLX5_COMMAND_STR_CASE(MANAGE_PAGES);
+	MLX5_COMMAND_STR_CASE(SET_HCA_CAP);
+	MLX5_COMMAND_STR_CASE(QUERY_ISSI);
+	MLX5_COMMAND_STR_CASE(SET_ISSI);
+	MLX5_COMMAND_STR_CASE(SET_DRIVER_VERSION);
+	MLX5_COMMAND_STR_CASE(CREATE_MKEY);
+	MLX5_COMMAND_STR_CASE(QUERY_MKEY);
+	MLX5_COMMAND_STR_CASE(DESTROY_MKEY);
+	MLX5_COMMAND_STR_CASE(QUERY_SPECIAL_CONTEXTS);
+	MLX5_COMMAND_STR_CASE(PAGE_FAULT_RESUME);
+	MLX5_COMMAND_STR_CASE(CREATE_EQ);
+	MLX5_COMMAND_STR_CASE(DESTROY_EQ);
+	MLX5_COMMAND_STR_CASE(QUERY_EQ);
+	MLX5_COMMAND_STR_CASE(GEN_EQE);
+	MLX5_COMMAND_STR_CASE(CREATE_CQ);
+	MLX5_COMMAND_STR_CASE(DESTROY_CQ);
+	MLX5_COMMAND_STR_CASE(QUERY_CQ);
+	MLX5_COMMAND_STR_CASE(MODIFY_CQ);
+	MLX5_COMMAND_STR_CASE(CREATE_QP);
+	MLX5_COMMAND_STR_CASE(DESTROY_QP);
+	MLX5_COMMAND_STR_CASE(RST2INIT_QP);
+	MLX5_COMMAND_STR_CASE(INIT2RTR_QP);
+	MLX5_COMMAND_STR_CASE(RTR2RTS_QP);
+	MLX5_COMMAND_STR_CASE(RTS2RTS_QP);
+	MLX5_COMMAND_STR_CASE(SQERR2RTS_QP);
+	MLX5_COMMAND_STR_CASE(2ERR_QP);
+	MLX5_COMMAND_STR_CASE(2RST_QP);
+	MLX5_COMMAND_STR_CASE(QUERY_QP);
+	MLX5_COMMAND_STR_CASE(SQD_RTS_QP);
+	MLX5_COMMAND_STR_CASE(INIT2INIT_QP);
+	MLX5_COMMAND_STR_CASE(CREATE_PSV);
+	MLX5_COMMAND_STR_CASE(DESTROY_PSV);
+	MLX5_COMMAND_STR_CASE(CREATE_SRQ);
+	MLX5_COMMAND_STR_CASE(DESTROY_SRQ);
+	MLX5_COMMAND_STR_CASE(QUERY_SRQ);
+	MLX5_COMMAND_STR_CASE(ARM_RQ);
+	MLX5_COMMAND_STR_CASE(CREATE_XRC_SRQ);
+	MLX5_COMMAND_STR_CASE(DESTROY_XRC_SRQ);
+	MLX5_COMMAND_STR_CASE(QUERY_XRC_SRQ);
+	MLX5_COMMAND_STR_CASE(ARM_XRC_SRQ);
+	MLX5_COMMAND_STR_CASE(CREATE_DCT);
+	MLX5_COMMAND_STR_CASE(DESTROY_DCT);
+	MLX5_COMMAND_STR_CASE(DRAIN_DCT);
+	MLX5_COMMAND_STR_CASE(QUERY_DCT);
+	MLX5_COMMAND_STR_CASE(ARM_DCT_FOR_KEY_VIOLATION);
+	MLX5_COMMAND_STR_CASE(QUERY_VPORT_STATE);
+	MLX5_COMMAND_STR_CASE(MODIFY_VPORT_STATE);
+	MLX5_COMMAND_STR_CASE(QUERY_ESW_VPORT_CONTEXT);
+	MLX5_COMMAND_STR_CASE(MODIFY_ESW_VPORT_CONTEXT);
+	MLX5_COMMAND_STR_CASE(QUERY_NIC_VPORT_CONTEXT);
+	MLX5_COMMAND_STR_CASE(MODIFY_NIC_VPORT_CONTEXT);
+	MLX5_COMMAND_STR_CASE(QUERY_ROCE_ADDRESS);
+	MLX5_COMMAND_STR_CASE(SET_ROCE_ADDRESS);
+	MLX5_COMMAND_STR_CASE(QUERY_HCA_VPORT_CONTEXT);
+	MLX5_COMMAND_STR_CASE(MODIFY_HCA_VPORT_CONTEXT);
+	MLX5_COMMAND_STR_CASE(QUERY_HCA_VPORT_GID);
+	MLX5_COMMAND_STR_CASE(QUERY_HCA_VPORT_PKEY);
+	MLX5_COMMAND_STR_CASE(QUERY_VNIC_ENV);
+	MLX5_COMMAND_STR_CASE(QUERY_VPORT_COUNTER);
+	MLX5_COMMAND_STR_CASE(ALLOC_Q_COUNTER);
+	MLX5_COMMAND_STR_CASE(DEALLOC_Q_COUNTER);
+	MLX5_COMMAND_STR_CASE(QUERY_Q_COUNTER);
+	MLX5_COMMAND_STR_CASE(SET_MONITOR_COUNTER);
+	MLX5_COMMAND_STR_CASE(ARM_MONITOR_COUNTER);
+	MLX5_COMMAND_STR_CASE(SET_PP_RATE_LIMIT);
+	MLX5_COMMAND_STR_CASE(QUERY_RATE_LIMIT);
+	MLX5_COMMAND_STR_CASE(CREATE_SCHEDULING_ELEMENT);
+	MLX5_COMMAND_STR_CASE(DESTROY_SCHEDULING_ELEMENT);
+	MLX5_COMMAND_STR_CASE(QUERY_SCHEDULING_ELEMENT);
+	MLX5_COMMAND_STR_CASE(MODIFY_SCHEDULING_ELEMENT);
+	MLX5_COMMAND_STR_CASE(CREATE_QOS_PARA_VPORT);
+	MLX5_COMMAND_STR_CASE(DESTROY_QOS_PARA_VPORT);
+	MLX5_COMMAND_STR_CASE(ALLOC_PD);
+	MLX5_COMMAND_STR_CASE(DEALLOC_PD);
+	MLX5_COMMAND_STR_CASE(ALLOC_UAR);
+	MLX5_COMMAND_STR_CASE(DEALLOC_UAR);
+	MLX5_COMMAND_STR_CASE(CONFIG_INT_MODERATION);
+	MLX5_COMMAND_STR_CASE(ACCESS_REG);
+	MLX5_COMMAND_STR_CASE(ATTACH_TO_MCG);
+	MLX5_COMMAND_STR_CASE(DETACH_FROM_MCG);
+	MLX5_COMMAND_STR_CASE(GET_DROPPED_PACKET_LOG);
+	MLX5_COMMAND_STR_CASE(MAD_IFC);
+	MLX5_COMMAND_STR_CASE(QUERY_MAD_DEMUX);
+	MLX5_COMMAND_STR_CASE(SET_MAD_DEMUX);
+	MLX5_COMMAND_STR_CASE(NOP);
+	MLX5_COMMAND_STR_CASE(ALLOC_XRCD);
+	MLX5_COMMAND_STR_CASE(DEALLOC_XRCD);
+	MLX5_COMMAND_STR_CASE(ALLOC_TRANSPORT_DOMAIN);
+	MLX5_COMMAND_STR_CASE(DEALLOC_TRANSPORT_DOMAIN);
+	MLX5_COMMAND_STR_CASE(QUERY_CONG_STATUS);
+	MLX5_COMMAND_STR_CASE(MODIFY_CONG_STATUS);
+	MLX5_COMMAND_STR_CASE(QUERY_CONG_PARAMS);
+	MLX5_COMMAND_STR_CASE(MODIFY_CONG_PARAMS);
+	MLX5_COMMAND_STR_CASE(QUERY_CONG_STATISTICS);
+	MLX5_COMMAND_STR_CASE(ADD_VXLAN_UDP_DPORT);
+	MLX5_COMMAND_STR_CASE(DELETE_VXLAN_UDP_DPORT);
+	MLX5_COMMAND_STR_CASE(SET_L2_TABLE_ENTRY);
+	MLX5_COMMAND_STR_CASE(QUERY_L2_TABLE_ENTRY);
+	MLX5_COMMAND_STR_CASE(DELETE_L2_TABLE_ENTRY);
+	MLX5_COMMAND_STR_CASE(SET_WOL_ROL);
+	MLX5_COMMAND_STR_CASE(QUERY_WOL_ROL);
+	MLX5_COMMAND_STR_CASE(CREATE_LAG);
+	MLX5_COMMAND_STR_CASE(MODIFY_LAG);
+	MLX5_COMMAND_STR_CASE(QUERY_LAG);
+	MLX5_COMMAND_STR_CASE(DESTROY_LAG);
+	MLX5_COMMAND_STR_CASE(CREATE_VPORT_LAG);
+	MLX5_COMMAND_STR_CASE(DESTROY_VPORT_LAG);
+	MLX5_COMMAND_STR_CASE(CREATE_TIR);
+	MLX5_COMMAND_STR_CASE(MODIFY_TIR);
+	MLX5_COMMAND_STR_CASE(DESTROY_TIR);
+	MLX5_COMMAND_STR_CASE(QUERY_TIR);
+	MLX5_COMMAND_STR_CASE(CREATE_SQ);
+	MLX5_COMMAND_STR_CASE(MODIFY_SQ);
+	MLX5_COMMAND_STR_CASE(DESTROY_SQ);
+	MLX5_COMMAND_STR_CASE(QUERY_SQ);
+	MLX5_COMMAND_STR_CASE(CREATE_RQ);
+	MLX5_COMMAND_STR_CASE(MODIFY_RQ);
+	MLX5_COMMAND_STR_CASE(DESTROY_RQ);
+	MLX5_COMMAND_STR_CASE(QUERY_RQ);
+	MLX5_COMMAND_STR_CASE(CREATE_RMP);
+	MLX5_COMMAND_STR_CASE(MODIFY_RMP);
+	MLX5_COMMAND_STR_CASE(DESTROY_RMP);
+	MLX5_COMMAND_STR_CASE(QUERY_RMP);
+	MLX5_COMMAND_STR_CASE(CREATE_TIS);
+	MLX5_COMMAND_STR_CASE(MODIFY_TIS);
+	MLX5_COMMAND_STR_CASE(DESTROY_TIS);
+	MLX5_COMMAND_STR_CASE(QUERY_TIS);
+	MLX5_COMMAND_STR_CASE(CREATE_RQT);
+	MLX5_COMMAND_STR_CASE(MODIFY_RQT);
+	MLX5_COMMAND_STR_CASE(DESTROY_RQT);
+	MLX5_COMMAND_STR_CASE(QUERY_RQT);
+	MLX5_COMMAND_STR_CASE(SET_FLOW_TABLE_ROOT);
+	MLX5_COMMAND_STR_CASE(CREATE_FLOW_TABLE);
+	MLX5_COMMAND_STR_CASE(DESTROY_FLOW_TABLE);
+	MLX5_COMMAND_STR_CASE(QUERY_FLOW_TABLE);
+	MLX5_COMMAND_STR_CASE(CREATE_FLOW_GROUP);
+	MLX5_COMMAND_STR_CASE(DESTROY_FLOW_GROUP);
+	MLX5_COMMAND_STR_CASE(QUERY_FLOW_GROUP);
+	MLX5_COMMAND_STR_CASE(SET_FLOW_TABLE_ENTRY);
+	MLX5_COMMAND_STR_CASE(QUERY_FLOW_TABLE_ENTRY);
+	MLX5_COMMAND_STR_CASE(DELETE_FLOW_TABLE_ENTRY);
+	MLX5_COMMAND_STR_CASE(ALLOC_FLOW_COUNTER);
+	MLX5_COMMAND_STR_CASE(DEALLOC_FLOW_COUNTER);
+	MLX5_COMMAND_STR_CASE(QUERY_FLOW_COUNTER);
+	MLX5_COMMAND_STR_CASE(MODIFY_FLOW_TABLE);
+	MLX5_COMMAND_STR_CASE(ALLOC_PACKET_REFORMAT_CONTEXT);
+	MLX5_COMMAND_STR_CASE(DEALLOC_PACKET_REFORMAT_CONTEXT);
+	MLX5_COMMAND_STR_CASE(ALLOC_MODIFY_HEADER_CONTEXT);
+	MLX5_COMMAND_STR_CASE(DEALLOC_MODIFY_HEADER_CONTEXT);
+	MLX5_COMMAND_STR_CASE(FPGA_CREATE_QP);
+	MLX5_COMMAND_STR_CASE(FPGA_MODIFY_QP);
+	MLX5_COMMAND_STR_CASE(FPGA_QUERY_QP);
+	MLX5_COMMAND_STR_CASE(FPGA_QUERY_QP_COUNTERS);
+	MLX5_COMMAND_STR_CASE(FPGA_DESTROY_QP);
+	MLX5_COMMAND_STR_CASE(CREATE_XRQ);
+	MLX5_COMMAND_STR_CASE(DESTROY_XRQ);
+	MLX5_COMMAND_STR_CASE(QUERY_XRQ);
+	MLX5_COMMAND_STR_CASE(ARM_XRQ);
+	MLX5_COMMAND_STR_CASE(CREATE_GENERAL_OBJECT);
+	MLX5_COMMAND_STR_CASE(DESTROY_GENERAL_OBJECT);
+	MLX5_COMMAND_STR_CASE(MODIFY_GENERAL_OBJECT);
+	MLX5_COMMAND_STR_CASE(QUERY_GENERAL_OBJECT);
+	MLX5_COMMAND_STR_CASE(QUERY_MODIFY_HEADER_CONTEXT);
+	MLX5_COMMAND_STR_CASE(ALLOC_MEMIC);
+	MLX5_COMMAND_STR_CASE(DEALLOC_MEMIC);
+	MLX5_COMMAND_STR_CASE(QUERY_ESW_FUNCTIONS);
+	MLX5_COMMAND_STR_CASE(CREATE_UCTX);
+	MLX5_COMMAND_STR_CASE(DESTROY_UCTX);
+	MLX5_COMMAND_STR_CASE(CREATE_UMEM);
+	MLX5_COMMAND_STR_CASE(DESTROY_UMEM);
+	MLX5_COMMAND_STR_CASE(RELEASE_XRQ_ERROR);
+	MLX5_COMMAND_STR_CASE(MODIFY_XRQ);
+	MLX5_COMMAND_STR_CASE(QUERY_VHCA_STATE);
+	MLX5_COMMAND_STR_CASE(MODIFY_VHCA_STATE);
+	MLX5_COMMAND_STR_CASE(ALLOC_SF);
+	MLX5_COMMAND_STR_CASE(DEALLOC_SF);
+	MLX5_COMMAND_STR_CASE(SUSPEND_VHCA);
+	MLX5_COMMAND_STR_CASE(RESUME_VHCA);
+	MLX5_COMMAND_STR_CASE(QUERY_VHCA_MIGRATION_STATE);
+	MLX5_COMMAND_STR_CASE(SAVE_VHCA_STATE);
+	MLX5_COMMAND_STR_CASE(LOAD_VHCA_STATE);
+	default: return "unknown command opcode";
+	}
+}
+
+static const char *cmd_status_str(u8 status)
+{
+	switch (status) {
+	case MLX5_CMD_STAT_OK:
+		return "OK";
+	case MLX5_CMD_STAT_INT_ERR:
+		return "internal error";
+	case MLX5_CMD_STAT_BAD_OP_ERR:
+		return "bad operation";
+	case MLX5_CMD_STAT_BAD_PARAM_ERR:
+		return "bad parameter";
+	case MLX5_CMD_STAT_BAD_SYS_STATE_ERR:
+		return "bad system state";
+	case MLX5_CMD_STAT_BAD_RES_ERR:
+		return "bad resource";
+	case MLX5_CMD_STAT_RES_BUSY:
+		return "resource busy";
+	case MLX5_CMD_STAT_LIM_ERR:
+		return "limits exceeded";
+	case MLX5_CMD_STAT_BAD_RES_STATE_ERR:
+		return "bad resource state";
+	case MLX5_CMD_STAT_IX_ERR:
+		return "bad index";
+	case MLX5_CMD_STAT_NO_RES_ERR:
+		return "no resources";
+	case MLX5_CMD_STAT_BAD_INP_LEN_ERR:
+		return "bad input length";
+	case MLX5_CMD_STAT_BAD_OUTP_LEN_ERR:
+		return "bad output length";
+	case MLX5_CMD_STAT_BAD_QP_STATE_ERR:
+		return "bad QP state";
+	case MLX5_CMD_STAT_BAD_PKT_ERR:
+		return "bad packet (discarded)";
+	case MLX5_CMD_STAT_BAD_SIZE_OUTS_CQES_ERR:
+		return "bad size too many outstanding CQEs";
+	default:
+		return "unknown status";
+	}
+}
+
+static int cmd_status_to_err(u8 status)
+{
+	switch (status) {
+	case MLX5_CMD_STAT_OK:				return 0;
+	case MLX5_CMD_STAT_INT_ERR:			return -EIO;
+	case MLX5_CMD_STAT_BAD_OP_ERR:			return -EINVAL;
+	case MLX5_CMD_STAT_BAD_PARAM_ERR:		return -EINVAL;
+	case MLX5_CMD_STAT_BAD_SYS_STATE_ERR:		return -EIO;
+	case MLX5_CMD_STAT_BAD_RES_ERR:			return -EINVAL;
+	case MLX5_CMD_STAT_RES_BUSY:			return -EBUSY;
+	case MLX5_CMD_STAT_LIM_ERR:			return -ENOMEM;
+	case MLX5_CMD_STAT_BAD_RES_STATE_ERR:		return -EINVAL;
+	case MLX5_CMD_STAT_IX_ERR:			return -EINVAL;
+	case MLX5_CMD_STAT_NO_RES_ERR:			return -EAGAIN;
+	case MLX5_CMD_STAT_BAD_INP_LEN_ERR:		return -EIO;
+	case MLX5_CMD_STAT_BAD_OUTP_LEN_ERR:		return -EIO;
+	case MLX5_CMD_STAT_BAD_QP_STATE_ERR:		return -EINVAL;
+	case MLX5_CMD_STAT_BAD_PKT_ERR:			return -EINVAL;
+	case MLX5_CMD_STAT_BAD_SIZE_OUTS_CQES_ERR:	return -EINVAL;
+	default:					return -EIO;
+	}
+}
+
+void mlx5_cmd_out_err(struct mlx5_core_dev *dev, u16 opcode, u16 op_mod, void *out)
+{
+	u32 syndrome = MLX5_GET(mbox_out, out, syndrome);
+	u8 status = MLX5_GET(mbox_out, out, status);
+
+	mlx5_core_err_rl(dev,
+			 "%s(0x%x) op_mod(0x%x) failed, status %s(0x%x), syndrome (0x%x), err(%d)\n",
+			 mlx5_command_str(opcode), opcode, op_mod,
+			 cmd_status_str(status), status, syndrome, cmd_status_to_err(status));
+}
+EXPORT_SYMBOL(mlx5_cmd_out_err);
+
+static void cmd_status_print(struct mlx5_core_dev *dev, void *in, void *out)
+{
+	u16 opcode, op_mod;
+	u16 uid;
+
+	opcode = in_to_opcode(in);
+	op_mod = MLX5_GET(mbox_in, in, op_mod);
+	uid    = MLX5_GET(mbox_in, in, uid);
+
+	if (!uid && opcode != MLX5_CMD_OP_DESTROY_MKEY)
+		mlx5_cmd_out_err(dev, opcode, op_mod, out);
+}
+
+int mlx5_cmd_check(struct mlx5_core_dev *dev, int err, void *in, void *out)
+{
+	/* aborted due to PCI error or via reset flow mlx5_cmd_trigger_completions() */
+	if (err == -ENXIO) {
+		u16 opcode = in_to_opcode(in);
+		u32 syndrome;
+		u8 status;
+
+		/* PCI Error, emulate command return status, for smooth reset */
+		err = mlx5_internal_err_ret_value(dev, opcode, &syndrome, &status);
+		MLX5_SET(mbox_out, out, status, status);
+		MLX5_SET(mbox_out, out, syndrome, syndrome);
+		if (!err)
+			return 0;
+	}
+
+	/* driver or FW delivery error */
+	if (err != -EREMOTEIO && err)
+		return err;
+
+	/* check outbox status */
+	err = cmd_status_to_err(MLX5_GET(mbox_out, out, status));
+	if (err)
+		cmd_status_print(dev, in, out);
+
+	return err;
+}
+EXPORT_SYMBOL(mlx5_cmd_check);
+
+static void dump_command(struct mlx5_core_dev *dev,
+			 struct mlx5_cmd_work_ent *ent, int input)
+{
+	struct mlx5_cmd_msg *msg = input ? ent->in : ent->out;
+	struct mlx5_cmd_mailbox *next = msg->next;
+	int n = mlx5_calc_cmd_blocks(msg);
+	u16 op = ent->op;
+	int data_only;
+	u32 offset = 0;
+	int dump_len;
+	int i;
+
+	mlx5_core_dbg(dev, "cmd[%d]: start dump\n", ent->idx);
+	data_only = !!(mlx5_core_debug_mask & (1 << MLX5_CMD_DATA));
+
+	if (data_only)
+		mlx5_core_dbg_mask(dev, 1 << MLX5_CMD_DATA,
+				   "cmd[%d]: dump command data %s(0x%x) %s\n",
+				   ent->idx, mlx5_command_str(op), op,
+				   input ? "INPUT" : "OUTPUT");
+	else
+		mlx5_core_dbg(dev, "cmd[%d]: dump command %s(0x%x) %s\n",
+			      ent->idx, mlx5_command_str(op), op,
+			      input ? "INPUT" : "OUTPUT");
+
+	if (data_only) {
+		if (input) {
+			dump_buf(ent->lay->in, sizeof(ent->lay->in), 1, offset, ent->idx);
+			offset += sizeof(ent->lay->in);
+		} else {
+			dump_buf(ent->lay->out, sizeof(ent->lay->out), 1, offset, ent->idx);
+			offset += sizeof(ent->lay->out);
+		}
+	} else {
+		dump_buf(ent->lay, sizeof(*ent->lay), 0, offset, ent->idx);
+		offset += sizeof(*ent->lay);
+	}
+
+	for (i = 0; i < n && next; i++)  {
+		if (data_only) {
+			dump_len = min_t(int, MLX5_CMD_DATA_BLOCK_SIZE, msg->len - offset);
+			dump_buf(next->buf, dump_len, 1, offset, ent->idx);
+			offset += MLX5_CMD_DATA_BLOCK_SIZE;
+		} else {
+			mlx5_core_dbg(dev, "cmd[%d]: command block:\n", ent->idx);
+			dump_buf(next->buf, sizeof(struct mlx5_cmd_prot_block), 0, offset,
+				 ent->idx);
+			offset += sizeof(struct mlx5_cmd_prot_block);
+		}
+		next = next->next;
+	}
+
+	if (data_only)
+		pr_debug("\n");
+
+	mlx5_core_dbg(dev, "cmd[%d]: end dump\n", ent->idx);
+}
+
+static void mlx5_cmd_comp_handler(struct mlx5_core_dev *dev, u64 vec, bool forced);
+
+static void cb_timeout_handler(struct work_struct *work)
+{
+	struct delayed_work *dwork = container_of(work, struct delayed_work,
+						  work);
+	struct mlx5_cmd_work_ent *ent = container_of(dwork,
+						     struct mlx5_cmd_work_ent,
+						     cb_timeout_work);
+	struct mlx5_core_dev *dev = container_of(ent->cmd, struct mlx5_core_dev,
+						 cmd);
+
+	mlx5_cmd_eq_recover(dev);
+
+	/* Maybe got handled by eq recover ? */
+	if (!test_bit(MLX5_CMD_ENT_STATE_PENDING_COMP, &ent->state)) {
+		mlx5_core_warn(dev, "cmd[%d]: %s(0x%x) Async, recovered after timeout\n", ent->idx,
+			       mlx5_command_str(ent->op), ent->op);
+		goto out; /* phew, already handled */
+	}
+
+	ent->ret = -ETIMEDOUT;
+	mlx5_core_warn(dev, "cmd[%d]: %s(0x%x) Async, timeout. Will cause a leak of a command resource\n",
+		       ent->idx, mlx5_command_str(ent->op), ent->op);
+	mlx5_cmd_comp_handler(dev, 1ULL << ent->idx, true);
+
+out:
+	cmd_ent_put(ent); /* for the cmd_ent_get() took on schedule delayed work */
+}
+
+static void free_msg(struct mlx5_core_dev *dev, struct mlx5_cmd_msg *msg);
+static void mlx5_free_cmd_msg(struct mlx5_core_dev *dev,
+			      struct mlx5_cmd_msg *msg);
+
+static bool opcode_allowed(struct mlx5_cmd *cmd, u16 opcode)
+{
+	if (cmd->allowed_opcode == CMD_ALLOWED_OPCODE_ALL)
+		return true;
+
+	return cmd->allowed_opcode == opcode;
+}
+
+bool mlx5_cmd_is_down(struct mlx5_core_dev *dev)
+{
+	return pci_channel_offline(dev->pdev) ||
+	       dev->cmd.state != MLX5_CMDIF_STATE_UP ||
+	       dev->state == MLX5_DEVICE_STATE_INTERNAL_ERROR;
+}
+
+static void cmd_work_handler(struct work_struct *work)
+{
+	struct mlx5_cmd_work_ent *ent = container_of(work, struct mlx5_cmd_work_ent, work);
+	struct mlx5_cmd *cmd = ent->cmd;
+	bool poll_cmd = ent->polling;
+	struct mlx5_cmd_layout *lay;
+	struct mlx5_core_dev *dev;
+	unsigned long cb_timeout;
+	struct semaphore *sem;
+	unsigned long flags;
+	int alloc_ret;
+	int cmd_mode;
+
+	dev = container_of(cmd, struct mlx5_core_dev, cmd);
+	cb_timeout = msecs_to_jiffies(mlx5_tout_ms(dev, CMD));
+
+	complete(&ent->handling);
+	sem = ent->page_queue ? &cmd->vars.pages_sem : &cmd->vars.sem;
+	down(sem);
+	if (!ent->page_queue) {
+		alloc_ret = cmd_alloc_index(cmd, ent);
+		if (alloc_ret < 0) {
+			mlx5_core_err_rl(dev, "failed to allocate command entry\n");
+			if (ent->callback) {
+				ent->callback(-EAGAIN, ent->context);
+				mlx5_free_cmd_msg(dev, ent->out);
+				free_msg(dev, ent->in);
+				cmd_ent_put(ent);
+			} else {
+				ent->ret = -EAGAIN;
+				complete(&ent->done);
+			}
+			up(sem);
+			return;
+		}
+	} else {
+		ent->idx = cmd->vars.max_reg_cmds;
+		spin_lock_irqsave(&cmd->alloc_lock, flags);
+		clear_bit(ent->idx, &cmd->vars.bitmask);
+		cmd->ent_arr[ent->idx] = ent;
+		spin_unlock_irqrestore(&cmd->alloc_lock, flags);
+	}
+
+	lay = get_inst(cmd, ent->idx);
+	ent->lay = lay;
+	memset(lay, 0, sizeof(*lay));
+	memcpy(lay->in, ent->in->first.data, sizeof(lay->in));
+	if (ent->in->next)
+		lay->in_ptr = cpu_to_be64(ent->in->next->dma);
+	lay->inlen = cpu_to_be32(ent->in->len);
+	if (ent->out->next)
+		lay->out_ptr = cpu_to_be64(ent->out->next->dma);
+	lay->outlen = cpu_to_be32(ent->out->len);
+	lay->type = MLX5_PCI_CMD_XPORT;
+	lay->token = ent->token;
+	lay->status_own = CMD_OWNER_HW;
+	set_signature(ent, !cmd->checksum_disabled);
+	dump_command(dev, ent, 1);
+	ent->ts1 = ktime_get_ns();
+	cmd_mode = cmd->mode;
+
+	if (ent->callback && schedule_delayed_work(&ent->cb_timeout_work, cb_timeout))
+		cmd_ent_get(ent);
+	set_bit(MLX5_CMD_ENT_STATE_PENDING_COMP, &ent->state);
+
+	cmd_ent_get(ent); /* for the _real_ FW event on completion */
+	/* Skip sending command to fw if internal error */
+	if (mlx5_cmd_is_down(dev) || !opcode_allowed(&dev->cmd, ent->op)) {
+		ent->ret = -ENXIO;
+		mlx5_cmd_comp_handler(dev, 1ULL << ent->idx, true);
+		return;
+	}
+
+	/* ring doorbell after the descriptor is valid */
+	mlx5_core_dbg(dev, "writing 0x%x to command doorbell\n", 1 << ent->idx);
+	wmb();
+	iowrite32be(1 << ent->idx, &dev->iseg->cmd_dbell);
+	/* if not in polling don't use ent after this point */
+	if (cmd_mode == CMD_MODE_POLLING || poll_cmd) {
+		poll_timeout(ent);
+		/* make sure we read the descriptor after ownership is SW */
+		rmb();
+		mlx5_cmd_comp_handler(dev, 1ULL << ent->idx, (ent->ret == -ETIMEDOUT));
+	}
+}
+
+static int deliv_status_to_err(u8 status)
+{
+	switch (status) {
+	case MLX5_CMD_DELIVERY_STAT_OK:
+	case MLX5_DRIVER_STATUS_ABORTED:
+		return 0;
+	case MLX5_CMD_DELIVERY_STAT_SIGNAT_ERR:
+	case MLX5_CMD_DELIVERY_STAT_TOK_ERR:
+		return -EBADR;
+	case MLX5_CMD_DELIVERY_STAT_BAD_BLK_NUM_ERR:
+	case MLX5_CMD_DELIVERY_STAT_OUT_PTR_ALIGN_ERR:
+	case MLX5_CMD_DELIVERY_STAT_IN_PTR_ALIGN_ERR:
+		return -EFAULT; /* Bad address */
+	case MLX5_CMD_DELIVERY_STAT_IN_LENGTH_ERR:
+	case MLX5_CMD_DELIVERY_STAT_OUT_LENGTH_ERR:
+	case MLX5_CMD_DELIVERY_STAT_CMD_DESCR_ERR:
+	case MLX5_CMD_DELIVERY_STAT_RES_FLD_NOT_CLR_ERR:
+		return -ENOMSG;
+	case MLX5_CMD_DELIVERY_STAT_FW_ERR:
+		return -EIO;
+	default:
+		return -EINVAL;
+	}
+}
+
+static const char *deliv_status_to_str(u8 status)
+{
+	switch (status) {
+	case MLX5_CMD_DELIVERY_STAT_OK:
+		return "no errors";
+	case MLX5_CMD_DELIVERY_STAT_SIGNAT_ERR:
+		return "signature error";
+	case MLX5_CMD_DELIVERY_STAT_TOK_ERR:
+		return "token error";
+	case MLX5_CMD_DELIVERY_STAT_BAD_BLK_NUM_ERR:
+		return "bad block number";
+	case MLX5_CMD_DELIVERY_STAT_OUT_PTR_ALIGN_ERR:
+		return "output pointer not aligned to block size";
+	case MLX5_CMD_DELIVERY_STAT_IN_PTR_ALIGN_ERR:
+		return "input pointer not aligned to block size";
+	case MLX5_CMD_DELIVERY_STAT_FW_ERR:
+		return "firmware internal error";
+	case MLX5_CMD_DELIVERY_STAT_IN_LENGTH_ERR:
+		return "command input length error";
+	case MLX5_CMD_DELIVERY_STAT_OUT_LENGTH_ERR:
+		return "command output length error";
+	case MLX5_CMD_DELIVERY_STAT_RES_FLD_NOT_CLR_ERR:
+		return "reserved fields not cleared";
+	case MLX5_CMD_DELIVERY_STAT_CMD_DESCR_ERR:
+		return "bad command descriptor type";
+	default:
+		return "unknown status code";
+	}
+}
+
+enum {
+	MLX5_CMD_TIMEOUT_RECOVER_MSEC   = 5 * 1000,
+};
+
+static void wait_func_handle_exec_timeout(struct mlx5_core_dev *dev,
+					  struct mlx5_cmd_work_ent *ent)
+{
+	unsigned long timeout = msecs_to_jiffies(MLX5_CMD_TIMEOUT_RECOVER_MSEC);
+
+	mlx5_cmd_eq_recover(dev);
+
+	/* Re-wait on the ent->done after executing the recovery flow. If the
+	 * recovery flow (or any other recovery flow running simultaneously)
+	 * has recovered an EQE, it should cause the entry to be completed by
+	 * the command interface.
+	 */
+	if (wait_for_completion_timeout(&ent->done, timeout)) {
+		mlx5_core_warn(dev, "cmd[%d]: %s(0x%x) recovered after timeout\n", ent->idx,
+			       mlx5_command_str(ent->op), ent->op);
+		return;
+	}
+
+	mlx5_core_warn(dev, "cmd[%d]: %s(0x%x) No done completion\n", ent->idx,
+		       mlx5_command_str(ent->op), ent->op);
+
+	ent->ret = -ETIMEDOUT;
+	mlx5_cmd_comp_handler(dev, 1ULL << ent->idx, true);
+}
+
+static int wait_func(struct mlx5_core_dev *dev, struct mlx5_cmd_work_ent *ent)
+{
+	unsigned long timeout = msecs_to_jiffies(mlx5_tout_ms(dev, CMD));
+	struct mlx5_cmd *cmd = &dev->cmd;
+	int err;
+
+	if (!wait_for_completion_timeout(&ent->handling, timeout) &&
+	    cancel_work_sync(&ent->work)) {
+		ent->ret = -ECANCELED;
+		goto out_err;
+	}
+	if (cmd->mode == CMD_MODE_POLLING || ent->polling)
+		wait_for_completion(&ent->done);
+	else if (!wait_for_completion_timeout(&ent->done, timeout))
+		wait_func_handle_exec_timeout(dev, ent);
+
+out_err:
+	err = ent->ret;
+
+	if (err == -ETIMEDOUT) {
+		mlx5_core_warn(dev, "%s(0x%x) timeout. Will cause a leak of a command resource\n",
+			       mlx5_command_str(ent->op), ent->op);
+	} else if (err == -ECANCELED) {
+		mlx5_core_warn(dev, "%s(0x%x) canceled on out of queue timeout.\n",
+			       mlx5_command_str(ent->op), ent->op);
+	}
+	mlx5_core_dbg(dev, "err %d, delivery status %s(%d)\n",
+		      err, deliv_status_to_str(ent->status), ent->status);
+
+	return err;
+}
+
+/*  Notes:
+ *    1. Callback functions may not sleep
+ *    2. page queue commands do not support asynchrous completion
+ *
+ * return value in case (!callback):
+ *	ret < 0 : Command execution couldn't be submitted by driver
+ *	ret > 0 : Command execution couldn't be performed by firmware
+ *	ret == 0: Command was executed by FW, Caller must check FW outbox status.
+ *
+ * return value in case (callback):
+ *	ret < 0 : Command execution couldn't be submitted by driver
+ *	ret == 0: Command will be submitted to FW for execution
+ *		  and the callback will be called for further status updates
+ */
+static int mlx5_cmd_invoke(struct mlx5_core_dev *dev, struct mlx5_cmd_msg *in,
+			   struct mlx5_cmd_msg *out, void *uout, int uout_size,
+			   mlx5_cmd_cbk_t callback,
+			   void *context, int page_queue,
+			   u8 token, bool force_polling)
+{
+	struct mlx5_cmd *cmd = &dev->cmd;
+	struct mlx5_cmd_work_ent *ent;
+	struct mlx5_cmd_stats *stats;
+	u8 status = 0;
+	int err = 0;
+	s64 ds;
+
+	if (callback && page_queue)
+		return -EINVAL;
+
+	ent = cmd_alloc_ent(cmd, in, out, uout, uout_size,
+			    callback, context, page_queue);
+	if (IS_ERR(ent))
+		return PTR_ERR(ent);
+
+	/* put for this ent is when consumed, depending on the use case
+	 * 1) (!callback) blocking flow: by caller after wait_func completes
+	 * 2) (callback) flow: by mlx5_cmd_comp_handler() when ent is handled
+	 */
+
+	ent->token = token;
+	ent->polling = force_polling;
+
+	init_completion(&ent->handling);
+	if (!callback)
+		init_completion(&ent->done);
+
+	INIT_DELAYED_WORK(&ent->cb_timeout_work, cb_timeout_handler);
+	INIT_WORK(&ent->work, cmd_work_handler);
+	if (page_queue) {
+		cmd_work_handler(&ent->work);
+	} else if (!queue_work(cmd->wq, &ent->work)) {
+		mlx5_core_warn(dev, "failed to queue work\n");
+		err = -EALREADY;
+		goto out_free;
+	}
+
+	if (callback)
+		return 0; /* mlx5_cmd_comp_handler() will put(ent) */
+
+	err = wait_func(dev, ent);
+	if (err == -ETIMEDOUT || err == -ECANCELED)
+		goto out_free;
+
+	ds = ent->ts2 - ent->ts1;
+	if (ent->op < MLX5_CMD_OP_MAX) {
+		stats = &cmd->stats[ent->op];
+		spin_lock_irq(&stats->lock);
+		stats->sum += ds;
+		++stats->n;
+		spin_unlock_irq(&stats->lock);
+	}
+	mlx5_core_dbg_mask(dev, 1 << MLX5_CMD_TIME,
+			   "fw exec time for %s is %lld nsec\n",
+			   mlx5_command_str(ent->op), ds);
+
+out_free:
+	status = ent->status;
+	cmd_ent_put(ent);
+	return err ? : status;
+}
+
+static ssize_t dbg_write(struct file *filp, const char __user *buf,
+			 size_t count, loff_t *pos)
+{
+	struct mlx5_core_dev *dev = filp->private_data;
+	struct mlx5_cmd_debug *dbg = &dev->cmd.dbg;
+	char lbuf[3];
+	int err;
+
+	if (!dbg->in_msg || !dbg->out_msg)
+		return -ENOMEM;
+
+	if (count < sizeof(lbuf) - 1)
+		return -EINVAL;
+
+	if (copy_from_user(lbuf, buf, sizeof(lbuf) - 1))
+		return -EFAULT;
+
+	lbuf[sizeof(lbuf) - 1] = 0;
+
+	if (strcmp(lbuf, "go"))
+		return -EINVAL;
+
+	err = mlx5_cmd_exec(dev, dbg->in_msg, dbg->inlen, dbg->out_msg, dbg->outlen);
+
+	return err ? err : count;
+}
+
+static const struct file_operations fops = {
+	.owner	= THIS_MODULE,
+	.open	= simple_open,
+	.write	= dbg_write,
+};
+
+static int mlx5_copy_to_msg(struct mlx5_cmd_msg *to, void *from, int size,
+			    u8 token)
+{
+	struct mlx5_cmd_prot_block *block;
+	struct mlx5_cmd_mailbox *next;
+	int copy;
+
+	if (!to || !from)
+		return -ENOMEM;
+
+	copy = min_t(int, size, sizeof(to->first.data));
+	memcpy(to->first.data, from, copy);
+	size -= copy;
+	from += copy;
+
+	next = to->next;
+	while (size) {
+		if (!next) {
+			/* this is a BUG */
+			return -ENOMEM;
+		}
+
+		copy = min_t(int, size, MLX5_CMD_DATA_BLOCK_SIZE);
+		block = next->buf;
+		memcpy(block->data, from, copy);
+		from += copy;
+		size -= copy;
+		block->token = token;
+		next = next->next;
+	}
+
+	return 0;
+}
+
+static int mlx5_copy_from_msg(void *to, struct mlx5_cmd_msg *from, int size)
+{
+	struct mlx5_cmd_prot_block *block;
+	struct mlx5_cmd_mailbox *next;
+	int copy;
+
+	if (!to || !from)
+		return -ENOMEM;
+
+	copy = min_t(int, size, sizeof(from->first.data));
+	memcpy(to, from->first.data, copy);
+	size -= copy;
+	to += copy;
+
+	next = from->next;
+	while (size) {
+		if (!next) {
+			/* this is a BUG */
+			return -ENOMEM;
+		}
+
+		copy = min_t(int, size, MLX5_CMD_DATA_BLOCK_SIZE);
+		block = next->buf;
+
+		memcpy(to, block->data, copy);
+		to += copy;
+		size -= copy;
+		next = next->next;
+	}
+
+	return 0;
+}
+
+static struct mlx5_cmd_mailbox *alloc_cmd_box(struct mlx5_core_dev *dev,
+					      gfp_t flags)
+{
+	struct mlx5_cmd_mailbox *mailbox;
+
+	mailbox = kmalloc(sizeof(*mailbox), flags);
+	if (!mailbox)
+		return ERR_PTR(-ENOMEM);
+
+	mailbox->buf = dma_pool_zalloc(dev->cmd.pool, flags,
+				       &mailbox->dma);
+	if (!mailbox->buf) {
+		mlx5_core_dbg(dev, "failed allocation\n");
+		kfree(mailbox);
+		return ERR_PTR(-ENOMEM);
+	}
+	mailbox->next = NULL;
+
+	return mailbox;
+}
+
+static void free_cmd_box(struct mlx5_core_dev *dev,
+			 struct mlx5_cmd_mailbox *mailbox)
+{
+	dma_pool_free(dev->cmd.pool, mailbox->buf, mailbox->dma);
+	kfree(mailbox);
+}
+
+static struct mlx5_cmd_msg *mlx5_alloc_cmd_msg(struct mlx5_core_dev *dev,
+					       gfp_t flags, int size,
+					       u8 token)
+{
+	struct mlx5_cmd_mailbox *tmp, *head = NULL;
+	struct mlx5_cmd_prot_block *block;
+	struct mlx5_cmd_msg *msg;
+	int err;
+	int n;
+	int i;
+
+	msg = kzalloc(sizeof(*msg), flags);
+	if (!msg)
+		return ERR_PTR(-ENOMEM);
+
+	msg->len = size;
+	n = mlx5_calc_cmd_blocks(msg);
+
+	for (i = 0; i < n; i++) {
+		tmp = alloc_cmd_box(dev, flags);
+		if (IS_ERR(tmp)) {
+			mlx5_core_warn(dev, "failed allocating block\n");
+			err = PTR_ERR(tmp);
+			goto err_alloc;
+		}
+
+		block = tmp->buf;
+		tmp->next = head;
+		block->next = cpu_to_be64(tmp->next ? tmp->next->dma : 0);
+		block->block_num = cpu_to_be32(n - i - 1);
+		block->token = token;
+		head = tmp;
+	}
+	msg->next = head;
+	return msg;
+
+err_alloc:
+	while (head) {
+		tmp = head->next;
+		free_cmd_box(dev, head);
+		head = tmp;
+	}
+	kfree(msg);
+
+	return ERR_PTR(err);
+}
+
+static void mlx5_free_cmd_msg(struct mlx5_core_dev *dev,
+			      struct mlx5_cmd_msg *msg)
+{
+	struct mlx5_cmd_mailbox *head = msg->next;
+	struct mlx5_cmd_mailbox *next;
+
+	while (head) {
+		next = head->next;
+		free_cmd_box(dev, head);
+		head = next;
+	}
+	kfree(msg);
+}
+
+static ssize_t data_write(struct file *filp, const char __user *buf,
+			  size_t count, loff_t *pos)
+{
+	struct mlx5_core_dev *dev = filp->private_data;
+	struct mlx5_cmd_debug *dbg = &dev->cmd.dbg;
+	void *ptr;
+
+	if (*pos != 0)
+		return -EINVAL;
+
+	kfree(dbg->in_msg);
+	dbg->in_msg = NULL;
+	dbg->inlen = 0;
+	ptr = memdup_user(buf, count);
+	if (IS_ERR(ptr))
+		return PTR_ERR(ptr);
+	dbg->in_msg = ptr;
+	dbg->inlen = count;
+
+	*pos = count;
+
+	return count;
+}
+
+static ssize_t data_read(struct file *filp, char __user *buf, size_t count,
+			 loff_t *pos)
+{
+	struct mlx5_core_dev *dev = filp->private_data;
+	struct mlx5_cmd_debug *dbg = &dev->cmd.dbg;
+
+	if (!dbg->out_msg)
+		return -ENOMEM;
+
+	return simple_read_from_buffer(buf, count, pos, dbg->out_msg,
+				       dbg->outlen);
+}
+
+static const struct file_operations dfops = {
+	.owner	= THIS_MODULE,
+	.open	= simple_open,
+	.write	= data_write,
+	.read	= data_read,
+};
+
+static ssize_t outlen_read(struct file *filp, char __user *buf, size_t count,
+			   loff_t *pos)
+{
+	struct mlx5_core_dev *dev = filp->private_data;
+	struct mlx5_cmd_debug *dbg = &dev->cmd.dbg;
+	char outlen[8];
+	int err;
+
+	err = snprintf(outlen, sizeof(outlen), "%d", dbg->outlen);
+	if (err < 0)
+		return err;
+
+	return simple_read_from_buffer(buf, count, pos, outlen, err);
+}
+
+static ssize_t outlen_write(struct file *filp, const char __user *buf,
+			    size_t count, loff_t *pos)
+{
+	struct mlx5_core_dev *dev = filp->private_data;
+	struct mlx5_cmd_debug *dbg = &dev->cmd.dbg;
+	char outlen_str[8] = {0};
+	int outlen;
+	void *ptr;
+	int err;
+
+	if (*pos != 0 || count > 6)
+		return -EINVAL;
+
+	kfree(dbg->out_msg);
+	dbg->out_msg = NULL;
+	dbg->outlen = 0;
+
+	if (copy_from_user(outlen_str, buf, count))
+		return -EFAULT;
+
+	err = sscanf(outlen_str, "%d", &outlen);
+	if (err != 1)
+		return -EINVAL;
+
+	ptr = kzalloc(outlen, GFP_KERNEL);
+	if (!ptr)
+		return -ENOMEM;
+
+	dbg->out_msg = ptr;
+	dbg->outlen = outlen;
+
+	*pos = count;
+
+	return count;
+}
+
+static const struct file_operations olfops = {
+	.owner	= THIS_MODULE,
+	.open	= simple_open,
+	.write	= outlen_write,
+	.read	= outlen_read,
+};
+
+static void set_wqname(struct mlx5_core_dev *dev)
+{
+	struct mlx5_cmd *cmd = &dev->cmd;
+
+	snprintf(cmd->wq_name, sizeof(cmd->wq_name), "mlx5_cmd_%s",
+		 dev_name(dev->device));
+}
+
+static void clean_debug_files(struct mlx5_core_dev *dev)
+{
+	struct mlx5_cmd_debug *dbg = &dev->cmd.dbg;
+
+	if (!mlx5_debugfs_root)
+		return;
+
+	mlx5_cmdif_debugfs_cleanup(dev);
+	debugfs_remove_recursive(dbg->dbg_root);
+}
+
+static void create_debugfs_files(struct mlx5_core_dev *dev)
+{
+	struct mlx5_cmd_debug *dbg = &dev->cmd.dbg;
+
+	dbg->dbg_root = debugfs_create_dir("cmd", mlx5_debugfs_get_dev_root(dev));
+
+	debugfs_create_file("in", 0400, dbg->dbg_root, dev, &dfops);
+	debugfs_create_file("out", 0200, dbg->dbg_root, dev, &dfops);
+	debugfs_create_file("out_len", 0600, dbg->dbg_root, dev, &olfops);
+	debugfs_create_u8("status", 0600, dbg->dbg_root, &dbg->status);
+	debugfs_create_file("run", 0200, dbg->dbg_root, dev, &fops);
+
+	mlx5_cmdif_debugfs_init(dev);
+}
+
+void mlx5_cmd_allowed_opcode(struct mlx5_core_dev *dev, u16 opcode)
+{
+	struct mlx5_cmd *cmd = &dev->cmd;
+	int i;
+
+	for (i = 0; i < cmd->vars.max_reg_cmds; i++)
+		down(&cmd->vars.sem);
+	down(&cmd->vars.pages_sem);
+
+	cmd->allowed_opcode = opcode;
+
+	up(&cmd->vars.pages_sem);
+	for (i = 0; i < cmd->vars.max_reg_cmds; i++)
+		up(&cmd->vars.sem);
+}
+
+static void mlx5_cmd_change_mod(struct mlx5_core_dev *dev, int mode)
+{
+	struct mlx5_cmd *cmd = &dev->cmd;
+	int i;
+
+	for (i = 0; i < cmd->vars.max_reg_cmds; i++)
+		down(&cmd->vars.sem);
+	down(&cmd->vars.pages_sem);
+
+	cmd->mode = mode;
+
+	up(&cmd->vars.pages_sem);
+	for (i = 0; i < cmd->vars.max_reg_cmds; i++)
+		up(&cmd->vars.sem);
+}
+
+static int cmd_comp_notifier(struct notifier_block *nb,
+			     unsigned long type, void *data)
+{
+	struct mlx5_core_dev *dev;
+	struct mlx5_cmd *cmd;
+	struct mlx5_eqe *eqe;
+
+	cmd = mlx5_nb_cof(nb, struct mlx5_cmd, nb);
+	dev = container_of(cmd, struct mlx5_core_dev, cmd);
+	eqe = data;
+
+	mlx5_cmd_comp_handler(dev, be32_to_cpu(eqe->data.cmd.vector), false);
+
+	return NOTIFY_OK;
+}
+void mlx5_cmd_use_events(struct mlx5_core_dev *dev)
+{
+	MLX5_NB_INIT(&dev->cmd.nb, cmd_comp_notifier, CMD);
+	mlx5_eq_notifier_register(dev, &dev->cmd.nb);
+	mlx5_cmd_change_mod(dev, CMD_MODE_EVENTS);
+}
+
+void mlx5_cmd_use_polling(struct mlx5_core_dev *dev)
+{
+	mlx5_cmd_change_mod(dev, CMD_MODE_POLLING);
+	mlx5_eq_notifier_unregister(dev, &dev->cmd.nb);
+}
+
+static void free_msg(struct mlx5_core_dev *dev, struct mlx5_cmd_msg *msg)
+{
+	unsigned long flags;
+
+	if (msg->parent) {
+		spin_lock_irqsave(&msg->parent->lock, flags);
+		list_add_tail(&msg->list, &msg->parent->head);
+		spin_unlock_irqrestore(&msg->parent->lock, flags);
+	} else {
+		mlx5_free_cmd_msg(dev, msg);
+	}
+}
+
+static void mlx5_cmd_comp_handler(struct mlx5_core_dev *dev, u64 vec, bool forced)
+{
+	struct mlx5_cmd *cmd = &dev->cmd;
+	struct mlx5_cmd_work_ent *ent;
+	mlx5_cmd_cbk_t callback;
+	void *context;
+	int err;
+	int i;
+	s64 ds;
+	struct mlx5_cmd_stats *stats;
+	unsigned long flags;
+	unsigned long vector;
+
+	/* there can be at most 32 command queues */
+	vector = vec & 0xffffffff;
+	for (i = 0; i < (1 << cmd->vars.log_sz); i++) {
+		if (test_bit(i, &vector)) {
+			ent = cmd->ent_arr[i];
+
+			/* if we already completed the command, ignore it */
+			if (!test_and_clear_bit(MLX5_CMD_ENT_STATE_PENDING_COMP,
+						&ent->state)) {
+				/* only real completion can free the cmd slot */
+				if (!forced) {
+					mlx5_core_err(dev, "Command completion arrived after timeout (entry idx = %d).\n",
+						      ent->idx);
+					cmd_ent_put(ent);
+				}
+				continue;
+			}
+
+			if (ent->callback && cancel_delayed_work(&ent->cb_timeout_work))
+				cmd_ent_put(ent); /* timeout work was canceled */
+
+			if (!forced || /* Real FW completion */
+			     mlx5_cmd_is_down(dev) || /* No real FW completion is expected */
+			     !opcode_allowed(cmd, ent->op))
+				cmd_ent_put(ent);
+
+			ent->ts2 = ktime_get_ns();
+			memcpy(ent->out->first.data, ent->lay->out, sizeof(ent->lay->out));
+			dump_command(dev, ent, 0);
+
+			if (vec & MLX5_TRIGGERED_CMD_COMP)
+				ent->ret = -ENXIO;
+
+			if (!ent->ret) { /* Command completed by FW */
+				if (!cmd->checksum_disabled)
+					ent->ret = verify_signature(ent);
+
+				ent->status = ent->lay->status_own >> 1;
+
+				mlx5_core_dbg(dev, "command completed. ret 0x%x, delivery status %s(0x%x)\n",
+					      ent->ret, deliv_status_to_str(ent->status), ent->status);
+			}
+
+			if (ent->callback) {
+				ds = ent->ts2 - ent->ts1;
+				if (ent->op < MLX5_CMD_OP_MAX) {
+					stats = &cmd->stats[ent->op];
+					spin_lock_irqsave(&stats->lock, flags);
+					stats->sum += ds;
+					++stats->n;
+					spin_unlock_irqrestore(&stats->lock, flags);
+				}
+
+				callback = ent->callback;
+				context = ent->context;
+				err = ent->ret ? : ent->status;
+				if (err > 0) /* Failed in FW, command didn't execute */
+					err = deliv_status_to_err(err);
+
+				if (!err)
+					err = mlx5_copy_from_msg(ent->uout,
+								 ent->out,
+								 ent->uout_size);
+
+				mlx5_free_cmd_msg(dev, ent->out);
+				free_msg(dev, ent->in);
+
+				/* final consumer is done, release ent */
+				cmd_ent_put(ent);
+				callback(err, context);
+			} else {
+				/* release wait_func() so mlx5_cmd_invoke()
+				 * can make the final ent_put()
+				 */
+				complete(&ent->done);
+			}
+		}
+	}
+}
+
+static void mlx5_cmd_trigger_completions(struct mlx5_core_dev *dev)
+{
+	struct mlx5_cmd *cmd = &dev->cmd;
+	unsigned long bitmask;
+	unsigned long flags;
+	u64 vector;
+	int i;
+
+	/* wait for pending handlers to complete */
+	mlx5_eq_synchronize_cmd_irq(dev);
+	spin_lock_irqsave(&dev->cmd.alloc_lock, flags);
+	vector = ~dev->cmd.vars.bitmask & ((1ul << (1 << dev->cmd.vars.log_sz)) - 1);
+	if (!vector)
+		goto no_trig;
+
+	bitmask = vector;
+	/* we must increment the allocated entries refcount before triggering the completions
+	 * to guarantee pending commands will not get freed in the meanwhile.
+	 * For that reason, it also has to be done inside the alloc_lock.
+	 */
+	for_each_set_bit(i, &bitmask, (1 << cmd->vars.log_sz))
+		cmd_ent_get(cmd->ent_arr[i]);
+	vector |= MLX5_TRIGGERED_CMD_COMP;
+	spin_unlock_irqrestore(&dev->cmd.alloc_lock, flags);
+
+	mlx5_core_dbg(dev, "vector 0x%llx\n", vector);
+	mlx5_cmd_comp_handler(dev, vector, true);
+	for_each_set_bit(i, &bitmask, (1 << cmd->vars.log_sz))
+		cmd_ent_put(cmd->ent_arr[i]);
+	return;
+
+no_trig:
+	spin_unlock_irqrestore(&dev->cmd.alloc_lock, flags);
+}
+
+void mlx5_cmd_flush(struct mlx5_core_dev *dev)
+{
+	struct mlx5_cmd *cmd = &dev->cmd;
+	int i;
+
+	for (i = 0; i < cmd->vars.max_reg_cmds; i++) {
+		while (down_trylock(&cmd->vars.sem)) {
+			mlx5_cmd_trigger_completions(dev);
+			cond_resched();
+		}
+	}
+
+	while (down_trylock(&cmd->vars.pages_sem)) {
+		mlx5_cmd_trigger_completions(dev);
+		cond_resched();
+	}
+
+	/* Unlock cmdif */
+	up(&cmd->vars.pages_sem);
+	for (i = 0; i < cmd->vars.max_reg_cmds; i++)
+		up(&cmd->vars.sem);
+}
+
+static struct mlx5_cmd_msg *alloc_msg(struct mlx5_core_dev *dev, int in_size,
+				      gfp_t gfp)
+{
+	struct mlx5_cmd_msg *msg = ERR_PTR(-ENOMEM);
+	struct cmd_msg_cache *ch = NULL;
+	struct mlx5_cmd *cmd = &dev->cmd;
+	int i;
+
+	if (in_size <= 16)
+		goto cache_miss;
+
+	for (i = 0; i < MLX5_NUM_COMMAND_CACHES; i++) {
+		ch = &cmd->cache[i];
+		if (in_size > ch->max_inbox_size)
+			continue;
+		spin_lock_irq(&ch->lock);
+		if (list_empty(&ch->head)) {
+			spin_unlock_irq(&ch->lock);
+			continue;
+		}
+		msg = list_entry(ch->head.next, typeof(*msg), list);
+		/* For cached lists, we must explicitly state what is
+		 * the real size
+		 */
+		msg->len = in_size;
+		list_del(&msg->list);
+		spin_unlock_irq(&ch->lock);
+		break;
+	}
+
+	if (!IS_ERR(msg))
+		return msg;
+
+cache_miss:
+	msg = mlx5_alloc_cmd_msg(dev, gfp, in_size, 0);
+	return msg;
+}
+
+static int is_manage_pages(void *in)
+{
+	return in_to_opcode(in) == MLX5_CMD_OP_MANAGE_PAGES;
+}
+
+/*  Notes:
+ *    1. Callback functions may not sleep
+ *    2. Page queue commands do not support asynchrous completion
+ */
+static int cmd_exec(struct mlx5_core_dev *dev, void *in, int in_size, void *out,
+		    int out_size, mlx5_cmd_cbk_t callback, void *context,
+		    bool force_polling)
+{
+	struct mlx5_cmd_msg *inb, *outb;
+	u16 opcode = in_to_opcode(in);
+	bool throttle_op;
+	int pages_queue;
+	gfp_t gfp;
+	u8 token;
+	int err;
+
+	if (mlx5_cmd_is_down(dev) || !opcode_allowed(&dev->cmd, opcode))
+		return -ENXIO;
+
+	throttle_op = mlx5_cmd_is_throttle_opcode(opcode);
+	if (throttle_op) {
+		/* atomic context may not sleep */
+		if (callback)
+			return -EINVAL;
+		down(&dev->cmd.vars.throttle_sem);
+	}
+
+	pages_queue = is_manage_pages(in);
+	gfp = callback ? GFP_ATOMIC : GFP_KERNEL;
+
+	inb = alloc_msg(dev, in_size, gfp);
+	if (IS_ERR(inb)) {
+		err = PTR_ERR(inb);
+		goto out_up;
+	}
+
+	token = alloc_token(&dev->cmd);
+
+	err = mlx5_copy_to_msg(inb, in, in_size, token);
+	if (err) {
+		mlx5_core_warn(dev, "err %d\n", err);
+		goto out_in;
+	}
+
+	outb = mlx5_alloc_cmd_msg(dev, gfp, out_size, token);
+	if (IS_ERR(outb)) {
+		err = PTR_ERR(outb);
+		goto out_in;
+	}
+
+	err = mlx5_cmd_invoke(dev, inb, outb, out, out_size, callback, context,
+			      pages_queue, token, force_polling);
+	if (callback)
+		return err;
+
+	if (err > 0) /* Failed in FW, command didn't execute */
+		err = deliv_status_to_err(err);
+
+	if (err)
+		goto out_out;
+
+	/* command completed by FW */
+	err = mlx5_copy_from_msg(out, outb, out_size);
+out_out:
+	mlx5_free_cmd_msg(dev, outb);
+out_in:
+	free_msg(dev, inb);
+out_up:
+	if (throttle_op)
+		up(&dev->cmd.vars.throttle_sem);
+	return err;
+}
+
+static void mlx5_cmd_err_trace(struct mlx5_core_dev *dev, u16 opcode, u16 op_mod, void *out)
+{
+	u32 syndrome = MLX5_GET(mbox_out, out, syndrome);
+	u8 status = MLX5_GET(mbox_out, out, status);
+
+	trace_mlx5_cmd(mlx5_command_str(opcode), opcode, op_mod,
+		       cmd_status_str(status), status, syndrome,
+		       cmd_status_to_err(status));
+}
+
+static void cmd_status_log(struct mlx5_core_dev *dev, u16 opcode, u8 status,
+			   u32 syndrome, int err)
+{
+	const char *namep = mlx5_command_str(opcode);
+	struct mlx5_cmd_stats *stats;
+
+	if (!err || !(strcmp(namep, "unknown command opcode")))
+		return;
+
+	stats = &dev->cmd.stats[opcode];
+	spin_lock_irq(&stats->lock);
+	stats->failed++;
+	if (err < 0)
+		stats->last_failed_errno = -err;
+	if (err == -EREMOTEIO) {
+		stats->failed_mbox_status++;
+		stats->last_failed_mbox_status = status;
+		stats->last_failed_syndrome = syndrome;
+	}
+	spin_unlock_irq(&stats->lock);
+}
+
+/* preserve -EREMOTEIO for outbox.status != OK, otherwise return err as is */
+static int cmd_status_err(struct mlx5_core_dev *dev, int err, u16 opcode, u16 op_mod, void *out)
+{
+	u32 syndrome = MLX5_GET(mbox_out, out, syndrome);
+	u8 status = MLX5_GET(mbox_out, out, status);
+
+	if (err == -EREMOTEIO) /* -EREMOTEIO is preserved */
+		err = -EIO;
+
+	if (!err && status != MLX5_CMD_STAT_OK) {
+		err = -EREMOTEIO;
+		mlx5_cmd_err_trace(dev, opcode, op_mod, out);
+	}
+
+	cmd_status_log(dev, opcode, status, syndrome, err);
+	return err;
+}
+
+/**
+ * mlx5_cmd_do - Executes a fw command, wait for completion.
+ * Unlike mlx5_cmd_exec, this function will not translate or intercept
+ * outbox.status and will return -EREMOTEIO when
+ * outbox.status != MLX5_CMD_STAT_OK
+ *
+ * @dev: mlx5 core device
+ * @in: inbox mlx5_ifc command buffer
+ * @in_size: inbox buffer size
+ * @out: outbox mlx5_ifc buffer
+ * @out_size: outbox size
+ *
+ * @return:
+ * -EREMOTEIO : Command executed by FW, outbox.status != MLX5_CMD_STAT_OK.
+ *              Caller must check FW outbox status.
+ *   0 : Command execution successful, outbox.status == MLX5_CMD_STAT_OK.
+ * < 0 : Command execution couldn't be performed by firmware or driver
+ */
+int mlx5_cmd_do(struct mlx5_core_dev *dev, void *in, int in_size, void *out, int out_size)
+{
+	int err = cmd_exec(dev, in, in_size, out, out_size, NULL, NULL, false);
+	u16 op_mod = MLX5_GET(mbox_in, in, op_mod);
+	u16 opcode = in_to_opcode(in);
+
+	return cmd_status_err(dev, err, opcode, op_mod, out);
+}
+EXPORT_SYMBOL(mlx5_cmd_do);
+
+/**
+ * mlx5_cmd_exec - Executes a fw command, wait for completion
+ *
+ * @dev: mlx5 core device
+ * @in: inbox mlx5_ifc command buffer
+ * @in_size: inbox buffer size
+ * @out: outbox mlx5_ifc buffer
+ * @out_size: outbox size
+ *
+ * @return: 0 if no error, FW command execution was successful
+ *          and outbox status is ok.
+ */
+int mlx5_cmd_exec(struct mlx5_core_dev *dev, void *in, int in_size, void *out,
+		  int out_size)
+{
+	int err = mlx5_cmd_do(dev, in, in_size, out, out_size);
+
+	return mlx5_cmd_check(dev, err, in, out);
+}
+EXPORT_SYMBOL(mlx5_cmd_exec);
+
+/**
+ * mlx5_cmd_exec_polling - Executes a fw command, poll for completion
+ *	Needed for driver force teardown, when command completion EQ
+ *	will not be available to complete the command
+ *
+ * @dev: mlx5 core device
+ * @in: inbox mlx5_ifc command buffer
+ * @in_size: inbox buffer size
+ * @out: outbox mlx5_ifc buffer
+ * @out_size: outbox size
+ *
+ * @return: 0 if no error, FW command execution was successful
+ *          and outbox status is ok.
+ */
+int mlx5_cmd_exec_polling(struct mlx5_core_dev *dev, void *in, int in_size,
+			  void *out, int out_size)
+{
+	int err = cmd_exec(dev, in, in_size, out, out_size, NULL, NULL, true);
+	u16 op_mod = MLX5_GET(mbox_in, in, op_mod);
+	u16 opcode = in_to_opcode(in);
+
+	err = cmd_status_err(dev, err, opcode, op_mod, out);
+	return mlx5_cmd_check(dev, err, in, out);
+}
+EXPORT_SYMBOL(mlx5_cmd_exec_polling);
+
+void mlx5_cmd_init_async_ctx(struct mlx5_core_dev *dev,
+			     struct mlx5_async_ctx *ctx)
+{
+	ctx->dev = dev;
+	/* Starts at 1 to avoid doing wake_up if we are not cleaning up */
+	atomic_set(&ctx->num_inflight, 1);
+	init_completion(&ctx->inflight_done);
+}
+EXPORT_SYMBOL(mlx5_cmd_init_async_ctx);
+
+/**
+ * mlx5_cmd_cleanup_async_ctx - Clean up an async_ctx
+ * @ctx: The ctx to clean
+ *
+ * Upon return all callbacks given to mlx5_cmd_exec_cb() have been called. The
+ * caller must ensure that mlx5_cmd_exec_cb() is not called during or after
+ * the call mlx5_cleanup_async_ctx().
+ */
+void mlx5_cmd_cleanup_async_ctx(struct mlx5_async_ctx *ctx)
+{
+	if (!atomic_dec_and_test(&ctx->num_inflight))
+		wait_for_completion(&ctx->inflight_done);
+}
+EXPORT_SYMBOL(mlx5_cmd_cleanup_async_ctx);
+
+static void mlx5_cmd_exec_cb_handler(int status, void *_work)
+{
+	struct mlx5_async_work *work = _work;
+	struct mlx5_async_ctx *ctx;
+
+	ctx = work->ctx;
+	status = cmd_status_err(ctx->dev, status, work->opcode, work->op_mod, work->out);
+	work->user_callback(status, work);
+	if (atomic_dec_and_test(&ctx->num_inflight))
+		complete(&ctx->inflight_done);
+}
+
+int mlx5_cmd_exec_cb(struct mlx5_async_ctx *ctx, void *in, int in_size,
+		     void *out, int out_size, mlx5_async_cbk_t callback,
+		     struct mlx5_async_work *work)
+{
+	int ret;
+
+	work->ctx = ctx;
+	work->user_callback = callback;
+	work->opcode = in_to_opcode(in);
+	work->op_mod = MLX5_GET(mbox_in, in, op_mod);
+	work->out = out;
+	if (WARN_ON(!atomic_inc_not_zero(&ctx->num_inflight)))
+		return -EIO;
+	ret = cmd_exec(ctx->dev, in, in_size, out, out_size,
+		       mlx5_cmd_exec_cb_handler, work, false);
+	if (ret && atomic_dec_and_test(&ctx->num_inflight))
+		complete(&ctx->inflight_done);
+
+	return ret;
+}
+EXPORT_SYMBOL(mlx5_cmd_exec_cb);
+
+static void destroy_msg_cache(struct mlx5_core_dev *dev)
+{
+	struct cmd_msg_cache *ch;
+	struct mlx5_cmd_msg *msg;
+	struct mlx5_cmd_msg *n;
+	int i;
+
+	for (i = 0; i < MLX5_NUM_COMMAND_CACHES; i++) {
+		ch = &dev->cmd.cache[i];
+		list_for_each_entry_safe(msg, n, &ch->head, list) {
+			list_del(&msg->list);
+			mlx5_free_cmd_msg(dev, msg);
+		}
+	}
+}
+
+static unsigned cmd_cache_num_ent[MLX5_NUM_COMMAND_CACHES] = {
+	512, 32, 16, 8, 2
+};
+
+static unsigned cmd_cache_ent_size[MLX5_NUM_COMMAND_CACHES] = {
+	16 + MLX5_CMD_DATA_BLOCK_SIZE,
+	16 + MLX5_CMD_DATA_BLOCK_SIZE * 2,
+	16 + MLX5_CMD_DATA_BLOCK_SIZE * 16,
+	16 + MLX5_CMD_DATA_BLOCK_SIZE * 256,
+	16 + MLX5_CMD_DATA_BLOCK_SIZE * 512,
+};
+
+static void create_msg_cache(struct mlx5_core_dev *dev)
+{
+	struct mlx5_cmd *cmd = &dev->cmd;
+	struct cmd_msg_cache *ch;
+	struct mlx5_cmd_msg *msg;
+	int i;
+	int k;
+
+	/* Initialize and fill the caches with initial entries */
+	for (k = 0; k < MLX5_NUM_COMMAND_CACHES; k++) {
+		ch = &cmd->cache[k];
+		spin_lock_init(&ch->lock);
+		INIT_LIST_HEAD(&ch->head);
+		ch->num_ent = cmd_cache_num_ent[k];
+		ch->max_inbox_size = cmd_cache_ent_size[k];
+		for (i = 0; i < ch->num_ent; i++) {
+			msg = mlx5_alloc_cmd_msg(dev, GFP_KERNEL | __GFP_NOWARN,
+						 ch->max_inbox_size, 0);
+			if (IS_ERR(msg))
+				break;
+			msg->parent = ch;
+			list_add_tail(&msg->list, &ch->head);
+		}
+	}
+}
+
+static int alloc_cmd_page(struct mlx5_core_dev *dev, struct mlx5_cmd *cmd)
+{
+	cmd->cmd_alloc_buf = dma_alloc_coherent(mlx5_core_dma_dev(dev), MLX5_ADAPTER_PAGE_SIZE,
+						&cmd->alloc_dma, GFP_KERNEL);
+	if (!cmd->cmd_alloc_buf)
+		return -ENOMEM;
+
+	/* make sure it is aligned to 4K */
+	if (!((uintptr_t)cmd->cmd_alloc_buf & (MLX5_ADAPTER_PAGE_SIZE - 1))) {
+		cmd->cmd_buf = cmd->cmd_alloc_buf;
+		cmd->dma = cmd->alloc_dma;
+		cmd->alloc_size = MLX5_ADAPTER_PAGE_SIZE;
+		return 0;
+	}
+
+	dma_free_coherent(mlx5_core_dma_dev(dev), MLX5_ADAPTER_PAGE_SIZE, cmd->cmd_alloc_buf,
+			  cmd->alloc_dma);
+	cmd->cmd_alloc_buf = dma_alloc_coherent(mlx5_core_dma_dev(dev),
+						2 * MLX5_ADAPTER_PAGE_SIZE - 1,
+						&cmd->alloc_dma, GFP_KERNEL);
+	if (!cmd->cmd_alloc_buf)
+		return -ENOMEM;
+
+	cmd->cmd_buf = PTR_ALIGN(cmd->cmd_alloc_buf, MLX5_ADAPTER_PAGE_SIZE);
+	cmd->dma = ALIGN(cmd->alloc_dma, MLX5_ADAPTER_PAGE_SIZE);
+	cmd->alloc_size = 2 * MLX5_ADAPTER_PAGE_SIZE - 1;
+	return 0;
+}
+
+static void free_cmd_page(struct mlx5_core_dev *dev, struct mlx5_cmd *cmd)
+{
+	dma_free_coherent(mlx5_core_dma_dev(dev), cmd->alloc_size, cmd->cmd_alloc_buf,
+			  cmd->alloc_dma);
+}
+
+static u16 cmdif_rev(struct mlx5_core_dev *dev)
+{
+	return ioread32be(&dev->iseg->cmdif_rev_fw_sub) >> 16;
+}
+
+int mlx5_cmd_init(struct mlx5_core_dev *dev)
+{
+	int size = sizeof(struct mlx5_cmd_prot_block);
+	int align = roundup_pow_of_two(size);
+	struct mlx5_cmd *cmd = &dev->cmd;
+	u32 cmd_h, cmd_l;
+	u16 cmd_if_rev;
+	int err;
+	int i;
+
+	memset(cmd, 0, sizeof(*cmd));
+	cmd_if_rev = cmdif_rev(dev);
+	if (cmd_if_rev != CMD_IF_REV) {
+		mlx5_core_err(dev,
+			      "Driver cmdif rev(%d) differs from firmware's(%d)\n",
+			      CMD_IF_REV, cmd_if_rev);
+		return -EINVAL;
+	}
+
+	cmd->pool = dma_pool_create("mlx5_cmd", mlx5_core_dma_dev(dev), size, align, 0);
+	if (!cmd->pool)
+		return -ENOMEM;
+
+	err = alloc_cmd_page(dev, cmd);
+	if (err)
+		goto err_free_pool;
+
+	cmd_l = ioread32be(&dev->iseg->cmdq_addr_l_sz) & 0xff;
+	cmd->vars.log_sz = cmd_l >> 4 & 0xf;
+	cmd->vars.log_stride = cmd_l & 0xf;
+	if (1 << cmd->vars.log_sz > MLX5_MAX_COMMANDS) {
+		mlx5_core_err(dev, "firmware reports too many outstanding commands %d\n",
+			      1 << cmd->vars.log_sz);
+		err = -EINVAL;
+		goto err_free_page;
+	}
+
+	if (cmd->vars.log_sz + cmd->vars.log_stride > MLX5_ADAPTER_PAGE_SHIFT) {
+		mlx5_core_err(dev, "command queue size overflow\n");
+		err = -EINVAL;
+		goto err_free_page;
+	}
+
+	cmd->state = MLX5_CMDIF_STATE_DOWN;
+	cmd->checksum_disabled = 1;
+	cmd->vars.max_reg_cmds = (1 << cmd->vars.log_sz) - 1;
+	cmd->vars.bitmask = (1UL << cmd->vars.max_reg_cmds) - 1;
+
+	cmd->vars.cmdif_rev = ioread32be(&dev->iseg->cmdif_rev_fw_sub) >> 16;
+	if (cmd->vars.cmdif_rev > CMD_IF_REV) {
+		mlx5_core_err(dev, "driver does not support command interface version. driver %d, firmware %d\n",
+			      CMD_IF_REV, cmd->vars.cmdif_rev);
+		err = -EOPNOTSUPP;
+		goto err_free_page;
+	}
+
+	spin_lock_init(&cmd->alloc_lock);
+	spin_lock_init(&cmd->token_lock);
+	for (i = 0; i < MLX5_CMD_OP_MAX; i++)
+		spin_lock_init(&cmd->stats[i].lock);
+
+	sema_init(&cmd->vars.sem, cmd->vars.max_reg_cmds);
+	sema_init(&cmd->vars.pages_sem, 1);
+	sema_init(&cmd->vars.throttle_sem, DIV_ROUND_UP(cmd->vars.max_reg_cmds, 2));
+
+	cmd_h = (u32)((u64)(cmd->dma) >> 32);
+	cmd_l = (u32)(cmd->dma);
+	if (cmd_l & 0xfff) {
+		mlx5_core_err(dev, "invalid command queue address\n");
+		err = -ENOMEM;
+		goto err_free_page;
+	}
+
+	iowrite32be(cmd_h, &dev->iseg->cmdq_addr_h);
+	iowrite32be(cmd_l, &dev->iseg->cmdq_addr_l_sz);
+
+	/* Make sure firmware sees the complete address before we proceed */
+	wmb();
+
+	mlx5_core_dbg(dev, "descriptor at dma 0x%llx\n", (unsigned long long)(cmd->dma));
+
+	cmd->mode = CMD_MODE_POLLING;
+	cmd->allowed_opcode = CMD_ALLOWED_OPCODE_ALL;
+
+	create_msg_cache(dev);
+
+	set_wqname(dev);
+	cmd->wq = create_singlethread_workqueue(cmd->wq_name);
+	if (!cmd->wq) {
+		mlx5_core_err(dev, "failed to create command workqueue\n");
+		err = -ENOMEM;
+		goto err_cache;
+	}
+
+	create_debugfs_files(dev);
+
+	return 0;
+
+err_cache:
+	destroy_msg_cache(dev);
+
+err_free_page:
+	free_cmd_page(dev, cmd);
+
+err_free_pool:
+	dma_pool_destroy(cmd->pool);
+	return err;
+}
+
+void mlx5_cmd_cleanup(struct mlx5_core_dev *dev)
+{
+	struct mlx5_cmd *cmd = &dev->cmd;
+
+	clean_debug_files(dev);
+	destroy_workqueue(cmd->wq);
+	destroy_msg_cache(dev);
+	free_cmd_page(dev, cmd);
+	dma_pool_destroy(cmd->pool);
+}
+
+void mlx5_cmd_set_state(struct mlx5_core_dev *dev,
+			enum mlx5_cmdif_state cmdif_state)
+{
+	dev->cmd.state = cmdif_state;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/cq.c b/drivers/net/ethernet/mellanox/mlx5/core/cq.c
new file mode 100644
index 000000000..4caa1b6f4
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/cq.c
@@ -0,0 +1,228 @@
+/*
+ * Copyright (c) 2013-2015, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/kernel.h>
+#include <linux/hardirq.h>
+#include <linux/mlx5/driver.h>
+#include <rdma/ib_verbs.h>
+#include <linux/mlx5/cq.h>
+#include "mlx5_core.h"
+#include "lib/eq.h"
+
+#define TASKLET_MAX_TIME 2
+#define TASKLET_MAX_TIME_JIFFIES msecs_to_jiffies(TASKLET_MAX_TIME)
+
+void mlx5_cq_tasklet_cb(struct tasklet_struct *t)
+{
+	unsigned long flags;
+	unsigned long end = jiffies + TASKLET_MAX_TIME_JIFFIES;
+	struct mlx5_eq_tasklet *ctx = from_tasklet(ctx, t, task);
+	struct mlx5_core_cq *mcq;
+	struct mlx5_core_cq *temp;
+
+	spin_lock_irqsave(&ctx->lock, flags);
+	list_splice_tail_init(&ctx->list, &ctx->process_list);
+	spin_unlock_irqrestore(&ctx->lock, flags);
+
+	list_for_each_entry_safe(mcq, temp, &ctx->process_list,
+				 tasklet_ctx.list) {
+		list_del_init(&mcq->tasklet_ctx.list);
+		mcq->tasklet_ctx.comp(mcq, NULL);
+		mlx5_cq_put(mcq);
+		if (time_after(jiffies, end))
+			break;
+	}
+
+	if (!list_empty(&ctx->process_list))
+		tasklet_schedule(&ctx->task);
+}
+
+static void mlx5_add_cq_to_tasklet(struct mlx5_core_cq *cq,
+				   struct mlx5_eqe *eqe)
+{
+	unsigned long flags;
+	struct mlx5_eq_tasklet *tasklet_ctx = cq->tasklet_ctx.priv;
+
+	spin_lock_irqsave(&tasklet_ctx->lock, flags);
+	/* When migrating CQs between EQs will be implemented, please note
+	 * that you need to sync this point. It is possible that
+	 * while migrating a CQ, completions on the old EQs could
+	 * still arrive.
+	 */
+	if (list_empty_careful(&cq->tasklet_ctx.list)) {
+		mlx5_cq_hold(cq);
+		list_add_tail(&cq->tasklet_ctx.list, &tasklet_ctx->list);
+	}
+	spin_unlock_irqrestore(&tasklet_ctx->lock, flags);
+}
+
+/* Callers must verify outbox status in case of err */
+int mlx5_create_cq(struct mlx5_core_dev *dev, struct mlx5_core_cq *cq,
+		   u32 *in, int inlen, u32 *out, int outlen)
+{
+	int eqn = MLX5_GET(cqc, MLX5_ADDR_OF(create_cq_in, in, cq_context),
+			   c_eqn_or_apu_element);
+	u32 din[MLX5_ST_SZ_DW(destroy_cq_in)] = {};
+	struct mlx5_eq_comp *eq;
+	int err;
+
+	eq = mlx5_eqn2comp_eq(dev, eqn);
+	if (IS_ERR(eq))
+		return PTR_ERR(eq);
+
+	memset(out, 0, outlen);
+	MLX5_SET(create_cq_in, in, opcode, MLX5_CMD_OP_CREATE_CQ);
+	err = mlx5_cmd_do(dev, in, inlen, out, outlen);
+	if (err)
+		return err;
+
+	cq->cqn = MLX5_GET(create_cq_out, out, cqn);
+	cq->cons_index = 0;
+	cq->arm_sn     = 0;
+	cq->eq         = eq;
+	cq->uid = MLX5_GET(create_cq_in, in, uid);
+	refcount_set(&cq->refcount, 1);
+	init_completion(&cq->free);
+	if (!cq->comp)
+		cq->comp = mlx5_add_cq_to_tasklet;
+	/* assuming CQ will be deleted before the EQ */
+	cq->tasklet_ctx.priv = &eq->tasklet_ctx;
+	INIT_LIST_HEAD(&cq->tasklet_ctx.list);
+
+	/* Add to comp EQ CQ tree to recv comp events */
+	err = mlx5_eq_add_cq(&eq->core, cq);
+	if (err)
+		goto err_cmd;
+
+	/* Add to async EQ CQ tree to recv async events */
+	err = mlx5_eq_add_cq(mlx5_get_async_eq(dev), cq);
+	if (err)
+		goto err_cq_add;
+
+	cq->pid = current->pid;
+	err = mlx5_debug_cq_add(dev, cq);
+	if (err)
+		mlx5_core_dbg(dev, "failed adding CP 0x%x to debug file system\n",
+			      cq->cqn);
+
+	cq->uar = dev->priv.uar;
+	cq->irqn = eq->core.irqn;
+
+	return 0;
+
+err_cq_add:
+	mlx5_eq_del_cq(&eq->core, cq);
+err_cmd:
+	MLX5_SET(destroy_cq_in, din, opcode, MLX5_CMD_OP_DESTROY_CQ);
+	MLX5_SET(destroy_cq_in, din, cqn, cq->cqn);
+	MLX5_SET(destroy_cq_in, din, uid, cq->uid);
+	mlx5_cmd_exec_in(dev, destroy_cq, din);
+	return err;
+}
+EXPORT_SYMBOL(mlx5_create_cq);
+
+/* oubox is checked and err val is normalized */
+int mlx5_core_create_cq(struct mlx5_core_dev *dev, struct mlx5_core_cq *cq,
+			u32 *in, int inlen, u32 *out, int outlen)
+{
+	int err = mlx5_create_cq(dev, cq, in, inlen, out, outlen);
+
+	return mlx5_cmd_check(dev, err, in, out);
+}
+EXPORT_SYMBOL(mlx5_core_create_cq);
+
+int mlx5_core_destroy_cq(struct mlx5_core_dev *dev, struct mlx5_core_cq *cq)
+{
+	u32 in[MLX5_ST_SZ_DW(destroy_cq_in)] = {};
+	int err;
+
+	mlx5_debug_cq_remove(dev, cq);
+
+	mlx5_eq_del_cq(mlx5_get_async_eq(dev), cq);
+	mlx5_eq_del_cq(&cq->eq->core, cq);
+
+	MLX5_SET(destroy_cq_in, in, opcode, MLX5_CMD_OP_DESTROY_CQ);
+	MLX5_SET(destroy_cq_in, in, cqn, cq->cqn);
+	MLX5_SET(destroy_cq_in, in, uid, cq->uid);
+	err = mlx5_cmd_exec_in(dev, destroy_cq, in);
+	if (err)
+		return err;
+
+	synchronize_irq(cq->irqn);
+	mlx5_cq_put(cq);
+	wait_for_completion(&cq->free);
+
+	return 0;
+}
+EXPORT_SYMBOL(mlx5_core_destroy_cq);
+
+int mlx5_core_query_cq(struct mlx5_core_dev *dev, struct mlx5_core_cq *cq,
+		       u32 *out)
+{
+	u32 in[MLX5_ST_SZ_DW(query_cq_in)] = {};
+
+	MLX5_SET(query_cq_in, in, opcode, MLX5_CMD_OP_QUERY_CQ);
+	MLX5_SET(query_cq_in, in, cqn, cq->cqn);
+	return mlx5_cmd_exec_inout(dev, query_cq, in, out);
+}
+EXPORT_SYMBOL(mlx5_core_query_cq);
+
+int mlx5_core_modify_cq(struct mlx5_core_dev *dev, struct mlx5_core_cq *cq,
+			u32 *in, int inlen)
+{
+	u32 out[MLX5_ST_SZ_DW(modify_cq_out)] = {};
+
+	MLX5_SET(modify_cq_in, in, opcode, MLX5_CMD_OP_MODIFY_CQ);
+	MLX5_SET(modify_cq_in, in, uid, cq->uid);
+	return mlx5_cmd_exec(dev, in, inlen, out, sizeof(out));
+}
+EXPORT_SYMBOL(mlx5_core_modify_cq);
+
+int mlx5_core_modify_cq_moderation(struct mlx5_core_dev *dev,
+				   struct mlx5_core_cq *cq,
+				   u16 cq_period,
+				   u16 cq_max_count)
+{
+	u32 in[MLX5_ST_SZ_DW(modify_cq_in)] = {};
+	void *cqc;
+
+	MLX5_SET(modify_cq_in, in, cqn, cq->cqn);
+	cqc = MLX5_ADDR_OF(modify_cq_in, in, cq_context);
+	MLX5_SET(cqc, cqc, cq_period, cq_period);
+	MLX5_SET(cqc, cqc, cq_max_count, cq_max_count);
+	MLX5_SET(modify_cq_in, in,
+		 modify_field_select_resize_field_select.modify_field_select.modify_field_select,
+		 MLX5_CQ_MODIFY_PERIOD | MLX5_CQ_MODIFY_COUNT);
+
+	return mlx5_core_modify_cq(dev, cq, in, sizeof(in));
+}
+EXPORT_SYMBOL(mlx5_core_modify_cq_moderation);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/debugfs.c b/drivers/net/ethernet/mellanox/mlx5/core/debugfs.c
new file mode 100644
index 000000000..e0b0729e2
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/debugfs.c
@@ -0,0 +1,574 @@
+/*
+ * Copyright (c) 2013-2015, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/debugfs.h>
+#include <linux/mlx5/qp.h>
+#include <linux/mlx5/cq.h>
+#include <linux/mlx5/driver.h>
+#include "mlx5_core.h"
+#include "lib/eq.h"
+
+enum {
+	QP_PID,
+	QP_STATE,
+	QP_XPORT,
+	QP_MTU,
+	QP_N_RECV,
+	QP_RECV_SZ,
+	QP_N_SEND,
+	QP_LOG_PG_SZ,
+	QP_RQPN,
+};
+
+static char *qp_fields[] = {
+	[QP_PID]	= "pid",
+	[QP_STATE]	= "state",
+	[QP_XPORT]	= "transport",
+	[QP_MTU]	= "mtu",
+	[QP_N_RECV]	= "num_recv",
+	[QP_RECV_SZ]	= "rcv_wqe_sz",
+	[QP_N_SEND]	= "num_send",
+	[QP_LOG_PG_SZ]	= "log2_page_sz",
+	[QP_RQPN]	= "remote_qpn",
+};
+
+enum {
+	EQ_NUM_EQES,
+	EQ_INTR,
+	EQ_LOG_PG_SZ,
+};
+
+static char *eq_fields[] = {
+	[EQ_NUM_EQES]	= "num_eqes",
+	[EQ_INTR]	= "intr",
+	[EQ_LOG_PG_SZ]	= "log_page_size",
+};
+
+enum {
+	CQ_PID,
+	CQ_NUM_CQES,
+	CQ_LOG_PG_SZ,
+};
+
+static char *cq_fields[] = {
+	[CQ_PID]	= "pid",
+	[CQ_NUM_CQES]	= "num_cqes",
+	[CQ_LOG_PG_SZ]	= "log_page_size",
+};
+
+struct dentry *mlx5_debugfs_root;
+EXPORT_SYMBOL(mlx5_debugfs_root);
+
+void mlx5_register_debugfs(void)
+{
+	mlx5_debugfs_root = debugfs_create_dir("mlx5", NULL);
+}
+
+void mlx5_unregister_debugfs(void)
+{
+	debugfs_remove(mlx5_debugfs_root);
+}
+
+struct dentry *mlx5_debugfs_get_dev_root(struct mlx5_core_dev *dev)
+{
+	return dev->priv.dbg.dbg_root;
+}
+EXPORT_SYMBOL(mlx5_debugfs_get_dev_root);
+
+void mlx5_qp_debugfs_init(struct mlx5_core_dev *dev)
+{
+	dev->priv.dbg.qp_debugfs = debugfs_create_dir("QPs", dev->priv.dbg.dbg_root);
+}
+EXPORT_SYMBOL(mlx5_qp_debugfs_init);
+
+void mlx5_qp_debugfs_cleanup(struct mlx5_core_dev *dev)
+{
+	debugfs_remove_recursive(dev->priv.dbg.qp_debugfs);
+}
+EXPORT_SYMBOL(mlx5_qp_debugfs_cleanup);
+
+void mlx5_eq_debugfs_init(struct mlx5_core_dev *dev)
+{
+	dev->priv.dbg.eq_debugfs = debugfs_create_dir("EQs", dev->priv.dbg.dbg_root);
+}
+
+void mlx5_eq_debugfs_cleanup(struct mlx5_core_dev *dev)
+{
+	debugfs_remove_recursive(dev->priv.dbg.eq_debugfs);
+}
+
+static ssize_t average_read(struct file *filp, char __user *buf, size_t count,
+			    loff_t *pos)
+{
+	struct mlx5_cmd_stats *stats;
+	u64 field = 0;
+	int ret;
+	char tbuf[22];
+
+	stats = filp->private_data;
+	spin_lock_irq(&stats->lock);
+	if (stats->n)
+		field = div64_u64(stats->sum, stats->n);
+	spin_unlock_irq(&stats->lock);
+	ret = snprintf(tbuf, sizeof(tbuf), "%llu\n", field);
+	return simple_read_from_buffer(buf, count, pos, tbuf, ret);
+}
+
+static ssize_t average_write(struct file *filp, const char __user *buf,
+			     size_t count, loff_t *pos)
+{
+	struct mlx5_cmd_stats *stats;
+
+	stats = filp->private_data;
+	spin_lock_irq(&stats->lock);
+	stats->sum = 0;
+	stats->n = 0;
+	spin_unlock_irq(&stats->lock);
+
+	*pos += count;
+
+	return count;
+}
+
+static const struct file_operations stats_fops = {
+	.owner	= THIS_MODULE,
+	.open	= simple_open,
+	.read	= average_read,
+	.write	= average_write,
+};
+
+static ssize_t slots_read(struct file *filp, char __user *buf, size_t count,
+			  loff_t *pos)
+{
+	struct mlx5_cmd *cmd;
+	char tbuf[6];
+	int weight;
+	int field;
+	int ret;
+
+	cmd = filp->private_data;
+	weight = bitmap_weight(&cmd->vars.bitmask, cmd->vars.max_reg_cmds);
+	field = cmd->vars.max_reg_cmds - weight;
+	ret = snprintf(tbuf, sizeof(tbuf), "%d\n", field);
+	return simple_read_from_buffer(buf, count, pos, tbuf, ret);
+}
+
+static const struct file_operations slots_fops = {
+	.owner	= THIS_MODULE,
+	.open	= simple_open,
+	.read	= slots_read,
+};
+
+void mlx5_cmdif_debugfs_init(struct mlx5_core_dev *dev)
+{
+	struct mlx5_cmd_stats *stats;
+	struct dentry **cmd;
+	const char *namep;
+	int i;
+
+	cmd = &dev->priv.dbg.cmdif_debugfs;
+	*cmd = debugfs_create_dir("commands", dev->priv.dbg.dbg_root);
+
+	debugfs_create_file("slots_inuse", 0400, *cmd, &dev->cmd, &slots_fops);
+
+	for (i = 0; i < MLX5_CMD_OP_MAX; i++) {
+		stats = &dev->cmd.stats[i];
+		namep = mlx5_command_str(i);
+		if (strcmp(namep, "unknown command opcode")) {
+			stats->root = debugfs_create_dir(namep, *cmd);
+
+			debugfs_create_file("average", 0400, stats->root, stats,
+					    &stats_fops);
+			debugfs_create_u64("n", 0400, stats->root, &stats->n);
+			debugfs_create_u64("failed", 0400, stats->root, &stats->failed);
+			debugfs_create_u64("failed_mbox_status", 0400, stats->root,
+					   &stats->failed_mbox_status);
+			debugfs_create_u32("last_failed_errno", 0400, stats->root,
+					   &stats->last_failed_errno);
+			debugfs_create_u8("last_failed_mbox_status", 0400, stats->root,
+					  &stats->last_failed_mbox_status);
+			debugfs_create_x32("last_failed_syndrome", 0400, stats->root,
+					   &stats->last_failed_syndrome);
+		}
+	}
+}
+
+void mlx5_cmdif_debugfs_cleanup(struct mlx5_core_dev *dev)
+{
+	debugfs_remove_recursive(dev->priv.dbg.cmdif_debugfs);
+}
+
+void mlx5_cq_debugfs_init(struct mlx5_core_dev *dev)
+{
+	dev->priv.dbg.cq_debugfs = debugfs_create_dir("CQs", dev->priv.dbg.dbg_root);
+}
+
+void mlx5_cq_debugfs_cleanup(struct mlx5_core_dev *dev)
+{
+	debugfs_remove_recursive(dev->priv.dbg.cq_debugfs);
+}
+
+void mlx5_pages_debugfs_init(struct mlx5_core_dev *dev)
+{
+	struct dentry *pages;
+
+	dev->priv.dbg.pages_debugfs = debugfs_create_dir("pages", dev->priv.dbg.dbg_root);
+	pages = dev->priv.dbg.pages_debugfs;
+
+	debugfs_create_u32("fw_pages_total", 0400, pages, &dev->priv.fw_pages);
+	debugfs_create_u32("fw_pages_vfs", 0400, pages, &dev->priv.page_counters[MLX5_VF]);
+	debugfs_create_u32("fw_pages_sfs", 0400, pages, &dev->priv.page_counters[MLX5_SF]);
+	debugfs_create_u32("fw_pages_host_pf", 0400, pages, &dev->priv.page_counters[MLX5_HOST_PF]);
+	debugfs_create_u32("fw_pages_alloc_failed", 0400, pages, &dev->priv.fw_pages_alloc_failed);
+	debugfs_create_u32("fw_pages_give_dropped", 0400, pages, &dev->priv.give_pages_dropped);
+	debugfs_create_u32("fw_pages_reclaim_discard", 0400, pages,
+			   &dev->priv.reclaim_pages_discard);
+}
+
+void mlx5_pages_debugfs_cleanup(struct mlx5_core_dev *dev)
+{
+	debugfs_remove_recursive(dev->priv.dbg.pages_debugfs);
+}
+
+static u64 qp_read_field(struct mlx5_core_dev *dev, struct mlx5_core_qp *qp,
+			 int index, int *is_str)
+{
+	int outlen = MLX5_ST_SZ_BYTES(query_qp_out);
+	u32 in[MLX5_ST_SZ_DW(query_qp_in)] = {};
+	u64 param = 0;
+	u32 *out;
+	int state;
+	u32 *qpc;
+	int err;
+
+	out = kzalloc(outlen, GFP_KERNEL);
+	if (!out)
+		return 0;
+
+	MLX5_SET(query_qp_in, in, opcode, MLX5_CMD_OP_QUERY_QP);
+	MLX5_SET(query_qp_in, in, qpn, qp->qpn);
+	err = mlx5_cmd_exec_inout(dev, query_qp, in, out);
+	if (err)
+		goto out;
+
+	*is_str = 0;
+
+	qpc = MLX5_ADDR_OF(query_qp_out, out, qpc);
+	switch (index) {
+	case QP_PID:
+		param = qp->pid;
+		break;
+	case QP_STATE:
+		state = MLX5_GET(qpc, qpc, state);
+		param = (unsigned long)mlx5_qp_state_str(state);
+		*is_str = 1;
+		break;
+	case QP_XPORT:
+		param = (unsigned long)mlx5_qp_type_str(MLX5_GET(qpc, qpc, st));
+		*is_str = 1;
+		break;
+	case QP_MTU:
+		switch (MLX5_GET(qpc, qpc, mtu)) {
+		case IB_MTU_256:
+			param = 256;
+			break;
+		case IB_MTU_512:
+			param = 512;
+			break;
+		case IB_MTU_1024:
+			param = 1024;
+			break;
+		case IB_MTU_2048:
+			param = 2048;
+			break;
+		case IB_MTU_4096:
+			param = 4096;
+			break;
+		default:
+			param = 0;
+		}
+		break;
+	case QP_N_RECV:
+		param = 1 << MLX5_GET(qpc, qpc, log_rq_size);
+		break;
+	case QP_RECV_SZ:
+		param = 1 << (MLX5_GET(qpc, qpc, log_rq_stride) + 4);
+		break;
+	case QP_N_SEND:
+		if (!MLX5_GET(qpc, qpc, no_sq))
+			param = 1 << MLX5_GET(qpc, qpc, log_sq_size);
+		break;
+	case QP_LOG_PG_SZ:
+		param = MLX5_GET(qpc, qpc, log_page_size) + 12;
+		break;
+	case QP_RQPN:
+		param = MLX5_GET(qpc, qpc, remote_qpn);
+		break;
+	}
+out:
+	kfree(out);
+	return param;
+}
+
+static u64 eq_read_field(struct mlx5_core_dev *dev, struct mlx5_eq *eq,
+			 int index)
+{
+	int outlen = MLX5_ST_SZ_BYTES(query_eq_out);
+	u32 in[MLX5_ST_SZ_DW(query_eq_in)] = {};
+	u64 param = 0;
+	void *ctx;
+	u32 *out;
+	int err;
+
+	out = kzalloc(outlen, GFP_KERNEL);
+	if (!out)
+		return param;
+
+	MLX5_SET(query_eq_in, in, opcode, MLX5_CMD_OP_QUERY_EQ);
+	MLX5_SET(query_eq_in, in, eq_number, eq->eqn);
+	err = mlx5_cmd_exec_inout(dev, query_eq, in, out);
+	if (err) {
+		mlx5_core_warn(dev, "failed to query eq\n");
+		goto out;
+	}
+	ctx = MLX5_ADDR_OF(query_eq_out, out, eq_context_entry);
+
+	switch (index) {
+	case EQ_NUM_EQES:
+		param = 1 << MLX5_GET(eqc, ctx, log_eq_size);
+		break;
+	case EQ_INTR:
+		param = MLX5_GET(eqc, ctx, intr);
+		break;
+	case EQ_LOG_PG_SZ:
+		param = MLX5_GET(eqc, ctx, log_page_size) + 12;
+		break;
+	}
+
+out:
+	kfree(out);
+	return param;
+}
+
+static u64 cq_read_field(struct mlx5_core_dev *dev, struct mlx5_core_cq *cq,
+			 int index)
+{
+	int outlen = MLX5_ST_SZ_BYTES(query_cq_out);
+	u64 param = 0;
+	void *ctx;
+	u32 *out;
+	int err;
+
+	out = kvzalloc(outlen, GFP_KERNEL);
+	if (!out)
+		return param;
+
+	err = mlx5_core_query_cq(dev, cq, out);
+	if (err) {
+		mlx5_core_warn(dev, "failed to query cq\n");
+		goto out;
+	}
+	ctx = MLX5_ADDR_OF(query_cq_out, out, cq_context);
+
+	switch (index) {
+	case CQ_PID:
+		param = cq->pid;
+		break;
+	case CQ_NUM_CQES:
+		param = 1 << MLX5_GET(cqc, ctx, log_cq_size);
+		break;
+	case CQ_LOG_PG_SZ:
+		param = MLX5_GET(cqc, ctx, log_page_size);
+		break;
+	}
+
+out:
+	kvfree(out);
+	return param;
+}
+
+static ssize_t dbg_read(struct file *filp, char __user *buf, size_t count,
+			loff_t *pos)
+{
+	struct mlx5_field_desc *desc;
+	struct mlx5_rsc_debug *d;
+	char tbuf[18];
+	int is_str = 0;
+	u64 field;
+	int ret;
+
+	desc = filp->private_data;
+	d = (void *)(desc - desc->i) - sizeof(*d);
+	switch (d->type) {
+	case MLX5_DBG_RSC_QP:
+		field = qp_read_field(d->dev, d->object, desc->i, &is_str);
+		break;
+
+	case MLX5_DBG_RSC_EQ:
+		field = eq_read_field(d->dev, d->object, desc->i);
+		break;
+
+	case MLX5_DBG_RSC_CQ:
+		field = cq_read_field(d->dev, d->object, desc->i);
+		break;
+
+	default:
+		mlx5_core_warn(d->dev, "invalid resource type %d\n", d->type);
+		return -EINVAL;
+	}
+
+	if (is_str)
+		ret = snprintf(tbuf, sizeof(tbuf), "%s\n", (const char *)(unsigned long)field);
+	else
+		ret = snprintf(tbuf, sizeof(tbuf), "0x%llx\n", field);
+
+	return simple_read_from_buffer(buf, count, pos, tbuf, ret);
+}
+
+static const struct file_operations fops = {
+	.owner	= THIS_MODULE,
+	.open	= simple_open,
+	.read	= dbg_read,
+};
+
+static int add_res_tree(struct mlx5_core_dev *dev, enum dbg_rsc_type type,
+			struct dentry *root, struct mlx5_rsc_debug **dbg,
+			int rsn, char **field, int nfile, void *data)
+{
+	struct mlx5_rsc_debug *d;
+	char resn[32];
+	int i;
+
+	d = kzalloc(struct_size(d, fields, nfile), GFP_KERNEL);
+	if (!d)
+		return -ENOMEM;
+
+	d->dev = dev;
+	d->object = data;
+	d->type = type;
+	sprintf(resn, "0x%x", rsn);
+	d->root = debugfs_create_dir(resn,  root);
+
+	for (i = 0; i < nfile; i++) {
+		d->fields[i].i = i;
+		debugfs_create_file(field[i], 0400, d->root, &d->fields[i],
+				    &fops);
+	}
+	*dbg = d;
+
+	return 0;
+}
+
+static void rem_res_tree(struct mlx5_rsc_debug *d)
+{
+	debugfs_remove_recursive(d->root);
+	kfree(d);
+}
+
+int mlx5_debug_qp_add(struct mlx5_core_dev *dev, struct mlx5_core_qp *qp)
+{
+	int err;
+
+	if (!mlx5_debugfs_root)
+		return 0;
+
+	err = add_res_tree(dev, MLX5_DBG_RSC_QP, dev->priv.dbg.qp_debugfs,
+			   &qp->dbg, qp->qpn, qp_fields,
+			   ARRAY_SIZE(qp_fields), qp);
+	if (err)
+		qp->dbg = NULL;
+
+	return err;
+}
+EXPORT_SYMBOL(mlx5_debug_qp_add);
+
+void mlx5_debug_qp_remove(struct mlx5_core_dev *dev, struct mlx5_core_qp *qp)
+{
+	if (!mlx5_debugfs_root)
+		return;
+
+	if (qp->dbg)
+		rem_res_tree(qp->dbg);
+}
+EXPORT_SYMBOL(mlx5_debug_qp_remove);
+
+int mlx5_debug_eq_add(struct mlx5_core_dev *dev, struct mlx5_eq *eq)
+{
+	int err;
+
+	if (!mlx5_debugfs_root)
+		return 0;
+
+	err = add_res_tree(dev, MLX5_DBG_RSC_EQ, dev->priv.dbg.eq_debugfs,
+			   &eq->dbg, eq->eqn, eq_fields,
+			   ARRAY_SIZE(eq_fields), eq);
+	if (err)
+		eq->dbg = NULL;
+
+	return err;
+}
+
+void mlx5_debug_eq_remove(struct mlx5_core_dev *dev, struct mlx5_eq *eq)
+{
+	if (!mlx5_debugfs_root)
+		return;
+
+	if (eq->dbg)
+		rem_res_tree(eq->dbg);
+}
+
+int mlx5_debug_cq_add(struct mlx5_core_dev *dev, struct mlx5_core_cq *cq)
+{
+	int err;
+
+	if (!mlx5_debugfs_root)
+		return 0;
+
+	err = add_res_tree(dev, MLX5_DBG_RSC_CQ, dev->priv.dbg.cq_debugfs,
+			   &cq->dbg, cq->cqn, cq_fields,
+			   ARRAY_SIZE(cq_fields), cq);
+	if (err)
+		cq->dbg = NULL;
+
+	return err;
+}
+
+void mlx5_debug_cq_remove(struct mlx5_core_dev *dev, struct mlx5_core_cq *cq)
+{
+	if (!mlx5_debugfs_root)
+		return;
+
+	if (cq->dbg) {
+		rem_res_tree(cq->dbg);
+		cq->dbg = NULL;
+	}
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/dev.c b/drivers/net/ethernet/mellanox/mlx5/core/dev.c
new file mode 100644
index 000000000..02bb9d43f
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/dev.c
@@ -0,0 +1,648 @@
+/*
+ * Copyright (c) 2013-2015, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/mlx5/driver.h>
+#include <linux/mlx5/eswitch.h>
+#include <linux/mlx5/mlx5_ifc_vdpa.h>
+#include <linux/mlx5/vport.h>
+#include "mlx5_core.h"
+
+/* intf dev list mutex */
+static DEFINE_MUTEX(mlx5_intf_mutex);
+static DEFINE_IDA(mlx5_adev_ida);
+
+static bool is_eth_rep_supported(struct mlx5_core_dev *dev)
+{
+	if (!IS_ENABLED(CONFIG_MLX5_ESWITCH))
+		return false;
+
+	if (!MLX5_ESWITCH_MANAGER(dev))
+		return false;
+
+	if (!is_mdev_switchdev_mode(dev))
+		return false;
+
+	return true;
+}
+
+bool mlx5_eth_supported(struct mlx5_core_dev *dev)
+{
+	if (!IS_ENABLED(CONFIG_MLX5_CORE_EN))
+		return false;
+
+	if (MLX5_CAP_GEN(dev, port_type) != MLX5_CAP_PORT_TYPE_ETH)
+		return false;
+
+	if (!MLX5_CAP_GEN(dev, eth_net_offloads)) {
+		mlx5_core_warn(dev, "Missing eth_net_offloads capability\n");
+		return false;
+	}
+
+	if (!MLX5_CAP_GEN(dev, nic_flow_table)) {
+		mlx5_core_warn(dev, "Missing nic_flow_table capability\n");
+		return false;
+	}
+
+	if (!MLX5_CAP_ETH(dev, csum_cap)) {
+		mlx5_core_warn(dev, "Missing csum_cap capability\n");
+		return false;
+	}
+
+	if (!MLX5_CAP_ETH(dev, max_lso_cap)) {
+		mlx5_core_warn(dev, "Missing max_lso_cap capability\n");
+		return false;
+	}
+
+	if (!MLX5_CAP_ETH(dev, vlan_cap)) {
+		mlx5_core_warn(dev, "Missing vlan_cap capability\n");
+		return false;
+	}
+
+	if (!MLX5_CAP_ETH(dev, rss_ind_tbl_cap)) {
+		mlx5_core_warn(dev, "Missing rss_ind_tbl_cap capability\n");
+		return false;
+	}
+
+	if (MLX5_CAP_FLOWTABLE(dev,
+			       flow_table_properties_nic_receive.max_ft_level) < 3) {
+		mlx5_core_warn(dev, "max_ft_level < 3\n");
+		return false;
+	}
+
+	if (!MLX5_CAP_ETH(dev, self_lb_en_modifiable))
+		mlx5_core_warn(dev, "Self loop back prevention is not supported\n");
+	if (!MLX5_CAP_GEN(dev, cq_moderation))
+		mlx5_core_warn(dev, "CQ moderation is not supported\n");
+
+	return true;
+}
+
+static bool is_eth_enabled(struct mlx5_core_dev *dev)
+{
+	union devlink_param_value val;
+	int err;
+
+	err = devlink_param_driverinit_value_get(priv_to_devlink(dev),
+						 DEVLINK_PARAM_GENERIC_ID_ENABLE_ETH,
+						 &val);
+	return err ? false : val.vbool;
+}
+
+bool mlx5_vnet_supported(struct mlx5_core_dev *dev)
+{
+	if (!IS_ENABLED(CONFIG_MLX5_VDPA_NET))
+		return false;
+
+	if (mlx5_core_is_pf(dev))
+		return false;
+
+	if (!(MLX5_CAP_GEN_64(dev, general_obj_types) &
+	      MLX5_GENERAL_OBJ_TYPES_CAP_VIRTIO_NET_Q))
+		return false;
+
+	if (!(MLX5_CAP_DEV_VDPA_EMULATION(dev, event_mode) &
+	      MLX5_VIRTIO_Q_EVENT_MODE_QP_MODE))
+		return false;
+
+	if (!MLX5_CAP_DEV_VDPA_EMULATION(dev, eth_frame_offload_type))
+		return false;
+
+	return true;
+}
+
+static bool is_vnet_enabled(struct mlx5_core_dev *dev)
+{
+	union devlink_param_value val;
+	int err;
+
+	err = devlink_param_driverinit_value_get(priv_to_devlink(dev),
+						 DEVLINK_PARAM_GENERIC_ID_ENABLE_VNET,
+						 &val);
+	return err ? false : val.vbool;
+}
+
+static bool is_ib_rep_supported(struct mlx5_core_dev *dev)
+{
+	if (!IS_ENABLED(CONFIG_MLX5_INFINIBAND))
+		return false;
+
+	if (dev->priv.flags & MLX5_PRIV_FLAGS_DISABLE_IB_ADEV)
+		return false;
+
+	if (!is_eth_rep_supported(dev))
+		return false;
+
+	if (!MLX5_ESWITCH_MANAGER(dev))
+		return false;
+
+	if (!is_mdev_switchdev_mode(dev))
+		return false;
+
+	if (mlx5_core_mp_enabled(dev))
+		return false;
+
+	return true;
+}
+
+static bool is_mp_supported(struct mlx5_core_dev *dev)
+{
+	if (!IS_ENABLED(CONFIG_MLX5_INFINIBAND))
+		return false;
+
+	if (dev->priv.flags & MLX5_PRIV_FLAGS_DISABLE_IB_ADEV)
+		return false;
+
+	if (is_ib_rep_supported(dev))
+		return false;
+
+	if (MLX5_CAP_GEN(dev, port_type) != MLX5_CAP_PORT_TYPE_ETH)
+		return false;
+
+	if (!mlx5_core_is_mp_slave(dev))
+		return false;
+
+	return true;
+}
+
+bool mlx5_rdma_supported(struct mlx5_core_dev *dev)
+{
+	if (!IS_ENABLED(CONFIG_MLX5_INFINIBAND))
+		return false;
+
+	if (dev->priv.flags & MLX5_PRIV_FLAGS_DISABLE_IB_ADEV)
+		return false;
+
+	if (is_ib_rep_supported(dev))
+		return false;
+
+	if (is_mp_supported(dev))
+		return false;
+
+	return true;
+}
+
+static bool is_ib_enabled(struct mlx5_core_dev *dev)
+{
+	union devlink_param_value val;
+	int err;
+
+	err = devlink_param_driverinit_value_get(priv_to_devlink(dev),
+						 DEVLINK_PARAM_GENERIC_ID_ENABLE_RDMA,
+						 &val);
+	return err ? false : val.vbool;
+}
+
+enum {
+	MLX5_INTERFACE_PROTOCOL_ETH,
+	MLX5_INTERFACE_PROTOCOL_ETH_REP,
+
+	MLX5_INTERFACE_PROTOCOL_IB,
+	MLX5_INTERFACE_PROTOCOL_IB_REP,
+	MLX5_INTERFACE_PROTOCOL_MPIB,
+
+	MLX5_INTERFACE_PROTOCOL_VNET,
+};
+
+static const struct mlx5_adev_device {
+	const char *suffix;
+	bool (*is_supported)(struct mlx5_core_dev *dev);
+	bool (*is_enabled)(struct mlx5_core_dev *dev);
+} mlx5_adev_devices[] = {
+	[MLX5_INTERFACE_PROTOCOL_VNET] = { .suffix = "vnet",
+					   .is_supported = &mlx5_vnet_supported,
+					   .is_enabled = &is_vnet_enabled },
+	[MLX5_INTERFACE_PROTOCOL_IB] = { .suffix = "rdma",
+					 .is_supported = &mlx5_rdma_supported,
+					 .is_enabled = &is_ib_enabled },
+	[MLX5_INTERFACE_PROTOCOL_ETH] = { .suffix = "eth",
+					  .is_supported = &mlx5_eth_supported,
+					  .is_enabled = &is_eth_enabled },
+	[MLX5_INTERFACE_PROTOCOL_ETH_REP] = { .suffix = "eth-rep",
+					   .is_supported = &is_eth_rep_supported },
+	[MLX5_INTERFACE_PROTOCOL_IB_REP] = { .suffix = "rdma-rep",
+					   .is_supported = &is_ib_rep_supported },
+	[MLX5_INTERFACE_PROTOCOL_MPIB] = { .suffix = "multiport",
+					   .is_supported = &is_mp_supported },
+};
+
+int mlx5_adev_idx_alloc(void)
+{
+	return ida_alloc(&mlx5_adev_ida, GFP_KERNEL);
+}
+
+void mlx5_adev_idx_free(int idx)
+{
+	ida_free(&mlx5_adev_ida, idx);
+}
+
+int mlx5_adev_init(struct mlx5_core_dev *dev)
+{
+	struct mlx5_priv *priv = &dev->priv;
+
+	priv->adev = kcalloc(ARRAY_SIZE(mlx5_adev_devices),
+			     sizeof(struct mlx5_adev *), GFP_KERNEL);
+	if (!priv->adev)
+		return -ENOMEM;
+
+	return 0;
+}
+
+void mlx5_adev_cleanup(struct mlx5_core_dev *dev)
+{
+	struct mlx5_priv *priv = &dev->priv;
+
+	kfree(priv->adev);
+}
+
+static void adev_release(struct device *dev)
+{
+	struct mlx5_adev *mlx5_adev =
+		container_of(dev, struct mlx5_adev, adev.dev);
+	struct mlx5_priv *priv = &mlx5_adev->mdev->priv;
+	int idx = mlx5_adev->idx;
+
+	kfree(mlx5_adev);
+	priv->adev[idx] = NULL;
+}
+
+static struct mlx5_adev *add_adev(struct mlx5_core_dev *dev, int idx)
+{
+	const char *suffix = mlx5_adev_devices[idx].suffix;
+	struct auxiliary_device *adev;
+	struct mlx5_adev *madev;
+	int ret;
+
+	madev = kzalloc(sizeof(*madev), GFP_KERNEL);
+	if (!madev)
+		return ERR_PTR(-ENOMEM);
+
+	adev = &madev->adev;
+	adev->id = dev->priv.adev_idx;
+	adev->name = suffix;
+	adev->dev.parent = dev->device;
+	adev->dev.release = adev_release;
+	madev->mdev = dev;
+	madev->idx = idx;
+
+	ret = auxiliary_device_init(adev);
+	if (ret) {
+		kfree(madev);
+		return ERR_PTR(ret);
+	}
+
+	ret = auxiliary_device_add(adev);
+	if (ret) {
+		auxiliary_device_uninit(adev);
+		return ERR_PTR(ret);
+	}
+	return madev;
+}
+
+static void del_adev(struct auxiliary_device *adev)
+{
+	auxiliary_device_delete(adev);
+	auxiliary_device_uninit(adev);
+}
+
+int mlx5_attach_device(struct mlx5_core_dev *dev)
+{
+	struct mlx5_priv *priv = &dev->priv;
+	struct auxiliary_device *adev;
+	struct auxiliary_driver *adrv;
+	int ret = 0, i;
+
+	devl_assert_locked(priv_to_devlink(dev));
+	mutex_lock(&mlx5_intf_mutex);
+	priv->flags &= ~MLX5_PRIV_FLAGS_DETACH;
+	priv->flags |= MLX5_PRIV_FLAGS_MLX5E_LOCKED_FLOW;
+	for (i = 0; i < ARRAY_SIZE(mlx5_adev_devices); i++) {
+		if (!priv->adev[i]) {
+			bool is_supported = false;
+
+			if (mlx5_adev_devices[i].is_enabled) {
+				bool enabled;
+
+				enabled = mlx5_adev_devices[i].is_enabled(dev);
+				if (!enabled)
+					continue;
+			}
+
+			if (mlx5_adev_devices[i].is_supported)
+				is_supported = mlx5_adev_devices[i].is_supported(dev);
+
+			if (!is_supported)
+				continue;
+
+			priv->adev[i] = add_adev(dev, i);
+			if (IS_ERR(priv->adev[i])) {
+				ret = PTR_ERR(priv->adev[i]);
+				priv->adev[i] = NULL;
+			}
+		} else {
+			adev = &priv->adev[i]->adev;
+
+			/* Pay attention that this is not PCI driver that
+			 * mlx5_core_dev is connected, but auxiliary driver.
+			 *
+			 * Here we can race of module unload with devlink
+			 * reload, but we don't need to take extra lock because
+			 * we are holding global mlx5_intf_mutex.
+			 */
+			if (!adev->dev.driver)
+				continue;
+			adrv = to_auxiliary_drv(adev->dev.driver);
+
+			if (adrv->resume)
+				ret = adrv->resume(adev);
+		}
+		if (ret) {
+			mlx5_core_warn(dev, "Device[%d] (%s) failed to load\n",
+				       i, mlx5_adev_devices[i].suffix);
+
+			break;
+		}
+	}
+	priv->flags &= ~MLX5_PRIV_FLAGS_MLX5E_LOCKED_FLOW;
+	mutex_unlock(&mlx5_intf_mutex);
+	return ret;
+}
+
+void mlx5_detach_device(struct mlx5_core_dev *dev, bool suspend)
+{
+	struct mlx5_priv *priv = &dev->priv;
+	struct auxiliary_device *adev;
+	struct auxiliary_driver *adrv;
+	pm_message_t pm = {};
+	int i;
+
+	devl_assert_locked(priv_to_devlink(dev));
+	mutex_lock(&mlx5_intf_mutex);
+	priv->flags |= MLX5_PRIV_FLAGS_MLX5E_LOCKED_FLOW;
+	for (i = ARRAY_SIZE(mlx5_adev_devices) - 1; i >= 0; i--) {
+		if (!priv->adev[i])
+			continue;
+
+		if (mlx5_adev_devices[i].is_enabled) {
+			bool enabled;
+
+			enabled = mlx5_adev_devices[i].is_enabled(dev);
+			if (!enabled)
+				goto skip_suspend;
+		}
+
+		adev = &priv->adev[i]->adev;
+		/* Auxiliary driver was unbind manually through sysfs */
+		if (!adev->dev.driver)
+			goto skip_suspend;
+
+		adrv = to_auxiliary_drv(adev->dev.driver);
+
+		if (adrv->suspend && suspend) {
+			adrv->suspend(adev, pm);
+			continue;
+		}
+
+skip_suspend:
+		del_adev(&priv->adev[i]->adev);
+		priv->adev[i] = NULL;
+	}
+	priv->flags &= ~MLX5_PRIV_FLAGS_MLX5E_LOCKED_FLOW;
+	priv->flags |= MLX5_PRIV_FLAGS_DETACH;
+	mutex_unlock(&mlx5_intf_mutex);
+}
+
+int mlx5_register_device(struct mlx5_core_dev *dev)
+{
+	int ret;
+
+	devl_assert_locked(priv_to_devlink(dev));
+	mutex_lock(&mlx5_intf_mutex);
+	dev->priv.flags &= ~MLX5_PRIV_FLAGS_DISABLE_ALL_ADEV;
+	ret = mlx5_rescan_drivers_locked(dev);
+	mutex_unlock(&mlx5_intf_mutex);
+	if (ret)
+		mlx5_unregister_device(dev);
+
+	return ret;
+}
+
+void mlx5_unregister_device(struct mlx5_core_dev *dev)
+{
+	devl_assert_locked(priv_to_devlink(dev));
+	mutex_lock(&mlx5_intf_mutex);
+	dev->priv.flags = MLX5_PRIV_FLAGS_DISABLE_ALL_ADEV;
+	mlx5_rescan_drivers_locked(dev);
+	mutex_unlock(&mlx5_intf_mutex);
+}
+
+static int add_drivers(struct mlx5_core_dev *dev)
+{
+	struct mlx5_priv *priv = &dev->priv;
+	int i, ret = 0;
+
+	for (i = 0; i < ARRAY_SIZE(mlx5_adev_devices); i++) {
+		bool is_supported = false;
+
+		if (priv->adev[i])
+			continue;
+
+		if (mlx5_adev_devices[i].is_supported)
+			is_supported = mlx5_adev_devices[i].is_supported(dev);
+
+		if (!is_supported)
+			continue;
+
+		priv->adev[i] = add_adev(dev, i);
+		if (IS_ERR(priv->adev[i])) {
+			mlx5_core_warn(dev, "Device[%d] (%s) failed to load\n",
+				       i, mlx5_adev_devices[i].suffix);
+			/* We continue to rescan drivers and leave to the caller
+			 * to make decision if to release everything or continue.
+			 */
+			ret = PTR_ERR(priv->adev[i]);
+			priv->adev[i] = NULL;
+		}
+	}
+	return ret;
+}
+
+static void delete_drivers(struct mlx5_core_dev *dev)
+{
+	struct mlx5_priv *priv = &dev->priv;
+	bool delete_all;
+	int i;
+
+	delete_all = priv->flags & MLX5_PRIV_FLAGS_DISABLE_ALL_ADEV;
+
+	for (i = ARRAY_SIZE(mlx5_adev_devices) - 1; i >= 0; i--) {
+		bool is_supported = false;
+
+		if (!priv->adev[i])
+			continue;
+
+		if (mlx5_adev_devices[i].is_enabled) {
+			bool enabled;
+
+			enabled = mlx5_adev_devices[i].is_enabled(dev);
+			if (!enabled)
+				goto del_adev;
+		}
+
+		if (mlx5_adev_devices[i].is_supported && !delete_all)
+			is_supported = mlx5_adev_devices[i].is_supported(dev);
+
+		if (is_supported)
+			continue;
+
+del_adev:
+		del_adev(&priv->adev[i]->adev);
+		priv->adev[i] = NULL;
+	}
+}
+
+/* This function is used after mlx5_core_dev is reconfigured.
+ */
+int mlx5_rescan_drivers_locked(struct mlx5_core_dev *dev)
+{
+	struct mlx5_priv *priv = &dev->priv;
+	int err = 0;
+
+	lockdep_assert_held(&mlx5_intf_mutex);
+	if (priv->flags & MLX5_PRIV_FLAGS_DETACH)
+		return 0;
+
+	priv->flags |= MLX5_PRIV_FLAGS_MLX5E_LOCKED_FLOW;
+	delete_drivers(dev);
+	if (priv->flags & MLX5_PRIV_FLAGS_DISABLE_ALL_ADEV)
+		goto out;
+
+	err = add_drivers(dev);
+
+out:
+	priv->flags &= ~MLX5_PRIV_FLAGS_MLX5E_LOCKED_FLOW;
+	return err;
+}
+
+bool mlx5_same_hw_devs(struct mlx5_core_dev *dev, struct mlx5_core_dev *peer_dev)
+{
+	u64 fsystem_guid, psystem_guid;
+
+	fsystem_guid = mlx5_query_nic_system_image_guid(dev);
+	psystem_guid = mlx5_query_nic_system_image_guid(peer_dev);
+
+	return (fsystem_guid && psystem_guid && fsystem_guid == psystem_guid);
+}
+
+static u32 mlx5_gen_pci_id(const struct mlx5_core_dev *dev)
+{
+	return (u32)((pci_domain_nr(dev->pdev->bus) << 16) |
+		     (dev->pdev->bus->number << 8) |
+		     PCI_SLOT(dev->pdev->devfn));
+}
+
+static int _next_phys_dev(struct mlx5_core_dev *mdev,
+			  const struct mlx5_core_dev *curr)
+{
+	if (!mlx5_core_is_pf(mdev))
+		return 0;
+
+	if (mdev == curr)
+		return 0;
+
+	if (!mlx5_same_hw_devs(mdev, (struct mlx5_core_dev *)curr) &&
+	    mlx5_gen_pci_id(mdev) != mlx5_gen_pci_id(curr))
+		return 0;
+
+	return 1;
+}
+
+static void *pci_get_other_drvdata(struct device *this, struct device *other)
+{
+	if (this->driver != other->driver)
+		return NULL;
+
+	return pci_get_drvdata(to_pci_dev(other));
+}
+
+static int next_phys_dev_lag(struct device *dev, const void *data)
+{
+	struct mlx5_core_dev *mdev, *this = (struct mlx5_core_dev *)data;
+
+	mdev = pci_get_other_drvdata(this->device, dev);
+	if (!mdev)
+		return 0;
+
+	if (!MLX5_CAP_GEN(mdev, vport_group_manager) ||
+	    !MLX5_CAP_GEN(mdev, lag_master) ||
+	    (MLX5_CAP_GEN(mdev, num_lag_ports) > MLX5_MAX_PORTS ||
+	     MLX5_CAP_GEN(mdev, num_lag_ports) <= 1))
+		return 0;
+
+	return _next_phys_dev(mdev, data);
+}
+
+static struct mlx5_core_dev *mlx5_get_next_dev(struct mlx5_core_dev *dev,
+					       int (*match)(struct device *dev, const void *data))
+{
+	struct device *next;
+
+	if (!mlx5_core_is_pf(dev))
+		return NULL;
+
+	next = bus_find_device(&pci_bus_type, NULL, dev, match);
+	if (!next)
+		return NULL;
+
+	put_device(next);
+	return pci_get_drvdata(to_pci_dev(next));
+}
+
+/* Must be called with intf_mutex held */
+struct mlx5_core_dev *mlx5_get_next_phys_dev_lag(struct mlx5_core_dev *dev)
+{
+	lockdep_assert_held(&mlx5_intf_mutex);
+	return mlx5_get_next_dev(dev, &next_phys_dev_lag);
+}
+
+void mlx5_dev_list_lock(void)
+{
+	mutex_lock(&mlx5_intf_mutex);
+}
+void mlx5_dev_list_unlock(void)
+{
+	mutex_unlock(&mlx5_intf_mutex);
+}
+
+int mlx5_dev_list_trylock(void)
+{
+	return mutex_trylock(&mlx5_intf_mutex);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/devlink.c b/drivers/net/ethernet/mellanox/mlx5/core/devlink.c
new file mode 100644
index 000000000..3749eb83d
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/devlink.c
@@ -0,0 +1,918 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2019 Mellanox Technologies */
+
+#include <devlink.h>
+
+#include "mlx5_core.h"
+#include "fw_reset.h"
+#include "fs_core.h"
+#include "eswitch.h"
+#include "esw/qos.h"
+#include "sf/dev/dev.h"
+#include "sf/sf.h"
+
+static int mlx5_devlink_flash_update(struct devlink *devlink,
+				     struct devlink_flash_update_params *params,
+				     struct netlink_ext_ack *extack)
+{
+	struct mlx5_core_dev *dev = devlink_priv(devlink);
+
+	return mlx5_firmware_flash(dev, params->fw, extack);
+}
+
+static u8 mlx5_fw_ver_major(u32 version)
+{
+	return (version >> 24) & 0xff;
+}
+
+static u8 mlx5_fw_ver_minor(u32 version)
+{
+	return (version >> 16) & 0xff;
+}
+
+static u16 mlx5_fw_ver_subminor(u32 version)
+{
+	return version & 0xffff;
+}
+
+#define DEVLINK_FW_STRING_LEN 32
+
+static int
+mlx5_devlink_info_get(struct devlink *devlink, struct devlink_info_req *req,
+		      struct netlink_ext_ack *extack)
+{
+	struct mlx5_core_dev *dev = devlink_priv(devlink);
+	char version_str[DEVLINK_FW_STRING_LEN];
+	u32 running_fw, stored_fw;
+	int err;
+
+	err = devlink_info_driver_name_put(req, KBUILD_MODNAME);
+	if (err)
+		return err;
+
+	err = devlink_info_version_fixed_put(req, "fw.psid", dev->board_id);
+	if (err)
+		return err;
+
+	err = mlx5_fw_version_query(dev, &running_fw, &stored_fw);
+	if (err)
+		return err;
+
+	snprintf(version_str, sizeof(version_str), "%d.%d.%04d",
+		 mlx5_fw_ver_major(running_fw), mlx5_fw_ver_minor(running_fw),
+		 mlx5_fw_ver_subminor(running_fw));
+	err = devlink_info_version_running_put(req, "fw.version", version_str);
+	if (err)
+		return err;
+	err = devlink_info_version_running_put(req,
+					       DEVLINK_INFO_VERSION_GENERIC_FW,
+					       version_str);
+	if (err)
+		return err;
+
+	/* no pending version, return running (stored) version */
+	if (stored_fw == 0)
+		stored_fw = running_fw;
+
+	snprintf(version_str, sizeof(version_str), "%d.%d.%04d",
+		 mlx5_fw_ver_major(stored_fw), mlx5_fw_ver_minor(stored_fw),
+		 mlx5_fw_ver_subminor(stored_fw));
+	err = devlink_info_version_stored_put(req, "fw.version", version_str);
+	if (err)
+		return err;
+	return devlink_info_version_stored_put(req,
+					       DEVLINK_INFO_VERSION_GENERIC_FW,
+					       version_str);
+}
+
+static int mlx5_devlink_reload_fw_activate(struct devlink *devlink, struct netlink_ext_ack *extack)
+{
+	struct mlx5_core_dev *dev = devlink_priv(devlink);
+	u8 reset_level, reset_type, net_port_alive;
+	int err;
+
+	err = mlx5_fw_reset_query(dev, &reset_level, &reset_type);
+	if (err)
+		return err;
+	if (!(reset_level & MLX5_MFRL_REG_RESET_LEVEL3)) {
+		NL_SET_ERR_MSG_MOD(extack, "FW activate requires reboot");
+		return -EINVAL;
+	}
+
+	net_port_alive = !!(reset_type & MLX5_MFRL_REG_RESET_TYPE_NET_PORT_ALIVE);
+	err = mlx5_fw_reset_set_reset_sync(dev, net_port_alive, extack);
+	if (err)
+		return err;
+
+	err = mlx5_fw_reset_wait_reset_done(dev);
+	if (err)
+		return err;
+
+	mlx5_unload_one_devl_locked(dev, true);
+	err = mlx5_health_wait_pci_up(dev);
+	if (err)
+		NL_SET_ERR_MSG_MOD(extack, "FW activate aborted, PCI reads fail after reset");
+
+	return err;
+}
+
+static int mlx5_devlink_trigger_fw_live_patch(struct devlink *devlink,
+					      struct netlink_ext_ack *extack)
+{
+	struct mlx5_core_dev *dev = devlink_priv(devlink);
+	u8 reset_level;
+	int err;
+
+	err = mlx5_fw_reset_query(dev, &reset_level, NULL);
+	if (err)
+		return err;
+	if (!(reset_level & MLX5_MFRL_REG_RESET_LEVEL0)) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "FW upgrade to the stored FW can't be done by FW live patching");
+		return -EINVAL;
+	}
+
+	return mlx5_fw_reset_set_live_patch(dev);
+}
+
+static int mlx5_devlink_reload_down(struct devlink *devlink, bool netns_change,
+				    enum devlink_reload_action action,
+				    enum devlink_reload_limit limit,
+				    struct netlink_ext_ack *extack)
+{
+	struct mlx5_core_dev *dev = devlink_priv(devlink);
+	struct pci_dev *pdev = dev->pdev;
+	bool sf_dev_allocated;
+	int ret = 0;
+
+	sf_dev_allocated = mlx5_sf_dev_allocated(dev);
+	if (sf_dev_allocated) {
+		/* Reload results in deleting SF device which further results in
+		 * unregistering devlink instance while holding devlink_mutext.
+		 * Hence, do not support reload.
+		 */
+		NL_SET_ERR_MSG_MOD(extack, "reload is unsupported when SFs are allocated");
+		return -EOPNOTSUPP;
+	}
+
+	if (mlx5_lag_is_active(dev)) {
+		NL_SET_ERR_MSG_MOD(extack, "reload is unsupported in Lag mode");
+		return -EOPNOTSUPP;
+	}
+
+	if (pci_num_vf(pdev)) {
+		NL_SET_ERR_MSG_MOD(extack, "reload while VFs are present is unfavorable");
+	}
+
+	switch (action) {
+	case DEVLINK_RELOAD_ACTION_DRIVER_REINIT:
+		mlx5_unload_one_devl_locked(dev, false);
+		break;
+	case DEVLINK_RELOAD_ACTION_FW_ACTIVATE:
+		if (limit == DEVLINK_RELOAD_LIMIT_NO_RESET)
+			ret = mlx5_devlink_trigger_fw_live_patch(devlink, extack);
+		else
+			ret = mlx5_devlink_reload_fw_activate(devlink, extack);
+		break;
+	default:
+		/* Unsupported action should not get to this function */
+		WARN_ON(1);
+		ret = -EOPNOTSUPP;
+	}
+
+	return ret;
+}
+
+static int mlx5_devlink_reload_up(struct devlink *devlink, enum devlink_reload_action action,
+				  enum devlink_reload_limit limit, u32 *actions_performed,
+				  struct netlink_ext_ack *extack)
+{
+	struct mlx5_core_dev *dev = devlink_priv(devlink);
+	int ret = 0;
+
+	*actions_performed = BIT(action);
+	switch (action) {
+	case DEVLINK_RELOAD_ACTION_DRIVER_REINIT:
+		ret = mlx5_load_one_devl_locked(dev, false);
+		break;
+	case DEVLINK_RELOAD_ACTION_FW_ACTIVATE:
+		if (limit == DEVLINK_RELOAD_LIMIT_NO_RESET)
+			break;
+		/* On fw_activate action, also driver is reloaded and reinit performed */
+		*actions_performed |= BIT(DEVLINK_RELOAD_ACTION_DRIVER_REINIT);
+		ret = mlx5_load_one_devl_locked(dev, true);
+		break;
+	default:
+		/* Unsupported action should not get to this function */
+		WARN_ON(1);
+		ret = -EOPNOTSUPP;
+	}
+
+	return ret;
+}
+
+static struct mlx5_devlink_trap *mlx5_find_trap_by_id(struct mlx5_core_dev *dev, int trap_id)
+{
+	struct mlx5_devlink_trap *dl_trap;
+
+	list_for_each_entry(dl_trap, &dev->priv.traps, list)
+		if (dl_trap->trap.id == trap_id)
+			return dl_trap;
+
+	return NULL;
+}
+
+static int mlx5_devlink_trap_init(struct devlink *devlink, const struct devlink_trap *trap,
+				  void *trap_ctx)
+{
+	struct mlx5_core_dev *dev = devlink_priv(devlink);
+	struct mlx5_devlink_trap *dl_trap;
+
+	dl_trap = kzalloc(sizeof(*dl_trap), GFP_KERNEL);
+	if (!dl_trap)
+		return -ENOMEM;
+
+	dl_trap->trap.id = trap->id;
+	dl_trap->trap.action = DEVLINK_TRAP_ACTION_DROP;
+	dl_trap->item = trap_ctx;
+
+	if (mlx5_find_trap_by_id(dev, trap->id)) {
+		kfree(dl_trap);
+		mlx5_core_err(dev, "Devlink trap: Trap 0x%x already found", trap->id);
+		return -EEXIST;
+	}
+
+	list_add_tail(&dl_trap->list, &dev->priv.traps);
+	return 0;
+}
+
+static void mlx5_devlink_trap_fini(struct devlink *devlink, const struct devlink_trap *trap,
+				   void *trap_ctx)
+{
+	struct mlx5_core_dev *dev = devlink_priv(devlink);
+	struct mlx5_devlink_trap *dl_trap;
+
+	dl_trap = mlx5_find_trap_by_id(dev, trap->id);
+	if (!dl_trap) {
+		mlx5_core_err(dev, "Devlink trap: Missing trap id 0x%x", trap->id);
+		return;
+	}
+	list_del(&dl_trap->list);
+	kfree(dl_trap);
+}
+
+static int mlx5_devlink_trap_action_set(struct devlink *devlink,
+					const struct devlink_trap *trap,
+					enum devlink_trap_action action,
+					struct netlink_ext_ack *extack)
+{
+	struct mlx5_core_dev *dev = devlink_priv(devlink);
+	enum devlink_trap_action action_orig;
+	struct mlx5_devlink_trap *dl_trap;
+	int err = 0;
+
+	if (is_mdev_switchdev_mode(dev)) {
+		NL_SET_ERR_MSG_MOD(extack, "Devlink traps can't be set in switchdev mode");
+		return -EOPNOTSUPP;
+	}
+
+	dl_trap = mlx5_find_trap_by_id(dev, trap->id);
+	if (!dl_trap) {
+		mlx5_core_err(dev, "Devlink trap: Set action on invalid trap id 0x%x", trap->id);
+		err = -EINVAL;
+		goto out;
+	}
+
+	if (action != DEVLINK_TRAP_ACTION_DROP && action != DEVLINK_TRAP_ACTION_TRAP) {
+		err = -EOPNOTSUPP;
+		goto out;
+	}
+
+	if (action == dl_trap->trap.action)
+		goto out;
+
+	action_orig = dl_trap->trap.action;
+	dl_trap->trap.action = action;
+	err = mlx5_blocking_notifier_call_chain(dev, MLX5_DRIVER_EVENT_TYPE_TRAP,
+						&dl_trap->trap);
+	if (err)
+		dl_trap->trap.action = action_orig;
+out:
+	return err;
+}
+
+static const struct devlink_ops mlx5_devlink_ops = {
+#ifdef CONFIG_MLX5_ESWITCH
+	.eswitch_mode_set = mlx5_devlink_eswitch_mode_set,
+	.eswitch_mode_get = mlx5_devlink_eswitch_mode_get,
+	.eswitch_inline_mode_set = mlx5_devlink_eswitch_inline_mode_set,
+	.eswitch_inline_mode_get = mlx5_devlink_eswitch_inline_mode_get,
+	.eswitch_encap_mode_set = mlx5_devlink_eswitch_encap_mode_set,
+	.eswitch_encap_mode_get = mlx5_devlink_eswitch_encap_mode_get,
+	.port_function_hw_addr_get = mlx5_devlink_port_function_hw_addr_get,
+	.port_function_hw_addr_set = mlx5_devlink_port_function_hw_addr_set,
+	.rate_leaf_tx_share_set = mlx5_esw_devlink_rate_leaf_tx_share_set,
+	.rate_leaf_tx_max_set = mlx5_esw_devlink_rate_leaf_tx_max_set,
+	.rate_node_tx_share_set = mlx5_esw_devlink_rate_node_tx_share_set,
+	.rate_node_tx_max_set = mlx5_esw_devlink_rate_node_tx_max_set,
+	.rate_node_new = mlx5_esw_devlink_rate_node_new,
+	.rate_node_del = mlx5_esw_devlink_rate_node_del,
+	.rate_leaf_parent_set = mlx5_esw_devlink_rate_parent_set,
+#endif
+#ifdef CONFIG_MLX5_SF_MANAGER
+	.port_new = mlx5_devlink_sf_port_new,
+	.port_del = mlx5_devlink_sf_port_del,
+	.port_fn_state_get = mlx5_devlink_sf_port_fn_state_get,
+	.port_fn_state_set = mlx5_devlink_sf_port_fn_state_set,
+#endif
+	.flash_update = mlx5_devlink_flash_update,
+	.info_get = mlx5_devlink_info_get,
+	.reload_actions = BIT(DEVLINK_RELOAD_ACTION_DRIVER_REINIT) |
+			  BIT(DEVLINK_RELOAD_ACTION_FW_ACTIVATE),
+	.reload_limits = BIT(DEVLINK_RELOAD_LIMIT_NO_RESET),
+	.reload_down = mlx5_devlink_reload_down,
+	.reload_up = mlx5_devlink_reload_up,
+	.trap_init = mlx5_devlink_trap_init,
+	.trap_fini = mlx5_devlink_trap_fini,
+	.trap_action_set = mlx5_devlink_trap_action_set,
+};
+
+void mlx5_devlink_trap_report(struct mlx5_core_dev *dev, int trap_id, struct sk_buff *skb,
+			      struct devlink_port *dl_port)
+{
+	struct devlink *devlink = priv_to_devlink(dev);
+	struct mlx5_devlink_trap *dl_trap;
+
+	dl_trap = mlx5_find_trap_by_id(dev, trap_id);
+	if (!dl_trap) {
+		mlx5_core_err(dev, "Devlink trap: Report on invalid trap id 0x%x", trap_id);
+		return;
+	}
+
+	if (dl_trap->trap.action != DEVLINK_TRAP_ACTION_TRAP) {
+		mlx5_core_dbg(dev, "Devlink trap: Trap id %d has action %d", trap_id,
+			      dl_trap->trap.action);
+		return;
+	}
+	devlink_trap_report(devlink, skb, dl_trap->item, dl_port, NULL);
+}
+
+int mlx5_devlink_trap_get_num_active(struct mlx5_core_dev *dev)
+{
+	struct mlx5_devlink_trap *dl_trap;
+	int count = 0;
+
+	list_for_each_entry(dl_trap, &dev->priv.traps, list)
+		if (dl_trap->trap.action == DEVLINK_TRAP_ACTION_TRAP)
+			count++;
+
+	return count;
+}
+
+int mlx5_devlink_traps_get_action(struct mlx5_core_dev *dev, int trap_id,
+				  enum devlink_trap_action *action)
+{
+	struct mlx5_devlink_trap *dl_trap;
+
+	dl_trap = mlx5_find_trap_by_id(dev, trap_id);
+	if (!dl_trap) {
+		mlx5_core_err(dev, "Devlink trap: Get action on invalid trap id 0x%x",
+			      trap_id);
+		return -EINVAL;
+	}
+
+	*action = dl_trap->trap.action;
+	return 0;
+}
+
+struct devlink *mlx5_devlink_alloc(struct device *dev)
+{
+	return devlink_alloc(&mlx5_devlink_ops, sizeof(struct mlx5_core_dev),
+			     dev);
+}
+
+void mlx5_devlink_free(struct devlink *devlink)
+{
+	devlink_free(devlink);
+}
+
+static int mlx5_devlink_fs_mode_validate(struct devlink *devlink, u32 id,
+					 union devlink_param_value val,
+					 struct netlink_ext_ack *extack)
+{
+	struct mlx5_core_dev *dev = devlink_priv(devlink);
+	char *value = val.vstr;
+	int err = 0;
+
+	if (!strcmp(value, "dmfs")) {
+		return 0;
+	} else if (!strcmp(value, "smfs")) {
+		u8 eswitch_mode;
+		bool smfs_cap;
+
+		eswitch_mode = mlx5_eswitch_mode(dev);
+		smfs_cap = mlx5_fs_dr_is_supported(dev);
+
+		if (!smfs_cap) {
+			err = -EOPNOTSUPP;
+			NL_SET_ERR_MSG_MOD(extack,
+					   "Software managed steering is not supported by current device");
+		}
+
+		else if (eswitch_mode == MLX5_ESWITCH_OFFLOADS) {
+			NL_SET_ERR_MSG_MOD(extack,
+					   "Software managed steering is not supported when eswitch offloads enabled.");
+			err = -EOPNOTSUPP;
+		}
+	} else {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "Bad parameter: supported values are [\"dmfs\", \"smfs\"]");
+		err = -EINVAL;
+	}
+
+	return err;
+}
+
+static int mlx5_devlink_fs_mode_set(struct devlink *devlink, u32 id,
+				    struct devlink_param_gset_ctx *ctx)
+{
+	struct mlx5_core_dev *dev = devlink_priv(devlink);
+	enum mlx5_flow_steering_mode mode;
+
+	if (!strcmp(ctx->val.vstr, "smfs"))
+		mode = MLX5_FLOW_STEERING_MODE_SMFS;
+	else
+		mode = MLX5_FLOW_STEERING_MODE_DMFS;
+	dev->priv.steering->mode = mode;
+
+	return 0;
+}
+
+static int mlx5_devlink_fs_mode_get(struct devlink *devlink, u32 id,
+				    struct devlink_param_gset_ctx *ctx)
+{
+	struct mlx5_core_dev *dev = devlink_priv(devlink);
+
+	if (dev->priv.steering->mode == MLX5_FLOW_STEERING_MODE_SMFS)
+		strcpy(ctx->val.vstr, "smfs");
+	else
+		strcpy(ctx->val.vstr, "dmfs");
+	return 0;
+}
+
+static int mlx5_devlink_enable_roce_validate(struct devlink *devlink, u32 id,
+					     union devlink_param_value val,
+					     struct netlink_ext_ack *extack)
+{
+	struct mlx5_core_dev *dev = devlink_priv(devlink);
+	bool new_state = val.vbool;
+
+	if (new_state && !MLX5_CAP_GEN(dev, roce) &&
+	    !(MLX5_CAP_GEN(dev, roce_rw_supported) && MLX5_CAP_GEN_MAX(dev, roce))) {
+		NL_SET_ERR_MSG_MOD(extack, "Device doesn't support RoCE");
+		return -EOPNOTSUPP;
+	}
+	if (mlx5_core_is_mp_slave(dev) || mlx5_lag_is_active(dev)) {
+		NL_SET_ERR_MSG_MOD(extack, "Multi port slave/Lag device can't configure RoCE");
+		return -EOPNOTSUPP;
+	}
+
+	return 0;
+}
+
+#ifdef CONFIG_MLX5_ESWITCH
+static int mlx5_devlink_large_group_num_validate(struct devlink *devlink, u32 id,
+						 union devlink_param_value val,
+						 struct netlink_ext_ack *extack)
+{
+	int group_num = val.vu32;
+
+	if (group_num < 1 || group_num > 1024) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "Unsupported group number, supported range is 1-1024");
+		return -EOPNOTSUPP;
+	}
+
+	return 0;
+}
+
+static int mlx5_devlink_esw_port_metadata_set(struct devlink *devlink, u32 id,
+					      struct devlink_param_gset_ctx *ctx)
+{
+	struct mlx5_core_dev *dev = devlink_priv(devlink);
+
+	if (!MLX5_ESWITCH_MANAGER(dev))
+		return -EOPNOTSUPP;
+
+	return mlx5_esw_offloads_vport_metadata_set(dev->priv.eswitch, ctx->val.vbool);
+}
+
+static int mlx5_devlink_esw_port_metadata_get(struct devlink *devlink, u32 id,
+					      struct devlink_param_gset_ctx *ctx)
+{
+	struct mlx5_core_dev *dev = devlink_priv(devlink);
+
+	if (!MLX5_ESWITCH_MANAGER(dev))
+		return -EOPNOTSUPP;
+
+	ctx->val.vbool = mlx5_eswitch_vport_match_metadata_enabled(dev->priv.eswitch);
+	return 0;
+}
+
+static int mlx5_devlink_esw_port_metadata_validate(struct devlink *devlink, u32 id,
+						   union devlink_param_value val,
+						   struct netlink_ext_ack *extack)
+{
+	struct mlx5_core_dev *dev = devlink_priv(devlink);
+	u8 esw_mode;
+
+	if (!MLX5_ESWITCH_MANAGER(dev)) {
+		NL_SET_ERR_MSG_MOD(extack, "E-Switch is unsupported");
+		return -EOPNOTSUPP;
+	}
+	esw_mode = mlx5_eswitch_mode(dev);
+	if (esw_mode == MLX5_ESWITCH_OFFLOADS) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "E-Switch must either disabled or non switchdev mode");
+		return -EBUSY;
+	}
+	return 0;
+}
+
+#endif
+
+static int mlx5_devlink_enable_remote_dev_reset_set(struct devlink *devlink, u32 id,
+						    struct devlink_param_gset_ctx *ctx)
+{
+	struct mlx5_core_dev *dev = devlink_priv(devlink);
+
+	mlx5_fw_reset_enable_remote_dev_reset_set(dev, ctx->val.vbool);
+	return 0;
+}
+
+static int mlx5_devlink_enable_remote_dev_reset_get(struct devlink *devlink, u32 id,
+						    struct devlink_param_gset_ctx *ctx)
+{
+	struct mlx5_core_dev *dev = devlink_priv(devlink);
+
+	ctx->val.vbool = mlx5_fw_reset_enable_remote_dev_reset_get(dev);
+	return 0;
+}
+
+static int mlx5_devlink_eq_depth_validate(struct devlink *devlink, u32 id,
+					  union devlink_param_value val,
+					  struct netlink_ext_ack *extack)
+{
+	return (val.vu32 >= 64 && val.vu32 <= 4096) ? 0 : -EINVAL;
+}
+
+static const struct devlink_param mlx5_devlink_params[] = {
+	DEVLINK_PARAM_DRIVER(MLX5_DEVLINK_PARAM_ID_FLOW_STEERING_MODE,
+			     "flow_steering_mode", DEVLINK_PARAM_TYPE_STRING,
+			     BIT(DEVLINK_PARAM_CMODE_RUNTIME),
+			     mlx5_devlink_fs_mode_get, mlx5_devlink_fs_mode_set,
+			     mlx5_devlink_fs_mode_validate),
+	DEVLINK_PARAM_GENERIC(ENABLE_ROCE, BIT(DEVLINK_PARAM_CMODE_DRIVERINIT),
+			      NULL, NULL, mlx5_devlink_enable_roce_validate),
+#ifdef CONFIG_MLX5_ESWITCH
+	DEVLINK_PARAM_DRIVER(MLX5_DEVLINK_PARAM_ID_ESW_LARGE_GROUP_NUM,
+			     "fdb_large_groups", DEVLINK_PARAM_TYPE_U32,
+			     BIT(DEVLINK_PARAM_CMODE_DRIVERINIT),
+			     NULL, NULL,
+			     mlx5_devlink_large_group_num_validate),
+	DEVLINK_PARAM_DRIVER(MLX5_DEVLINK_PARAM_ID_ESW_PORT_METADATA,
+			     "esw_port_metadata", DEVLINK_PARAM_TYPE_BOOL,
+			     BIT(DEVLINK_PARAM_CMODE_RUNTIME),
+			     mlx5_devlink_esw_port_metadata_get,
+			     mlx5_devlink_esw_port_metadata_set,
+			     mlx5_devlink_esw_port_metadata_validate),
+#endif
+	DEVLINK_PARAM_GENERIC(ENABLE_REMOTE_DEV_RESET, BIT(DEVLINK_PARAM_CMODE_RUNTIME),
+			      mlx5_devlink_enable_remote_dev_reset_get,
+			      mlx5_devlink_enable_remote_dev_reset_set, NULL),
+	DEVLINK_PARAM_GENERIC(IO_EQ_SIZE, BIT(DEVLINK_PARAM_CMODE_DRIVERINIT),
+			      NULL, NULL, mlx5_devlink_eq_depth_validate),
+	DEVLINK_PARAM_GENERIC(EVENT_EQ_SIZE, BIT(DEVLINK_PARAM_CMODE_DRIVERINIT),
+			      NULL, NULL, mlx5_devlink_eq_depth_validate),
+};
+
+static void mlx5_devlink_set_params_init_values(struct devlink *devlink)
+{
+	struct mlx5_core_dev *dev = devlink_priv(devlink);
+	union devlink_param_value value;
+
+	value.vbool = MLX5_CAP_GEN(dev, roce);
+	devlink_param_driverinit_value_set(devlink,
+					   DEVLINK_PARAM_GENERIC_ID_ENABLE_ROCE,
+					   value);
+
+#ifdef CONFIG_MLX5_ESWITCH
+	value.vu32 = ESW_OFFLOADS_DEFAULT_NUM_GROUPS;
+	devlink_param_driverinit_value_set(devlink,
+					   MLX5_DEVLINK_PARAM_ID_ESW_LARGE_GROUP_NUM,
+					   value);
+#endif
+
+	value.vu32 = MLX5_COMP_EQ_SIZE;
+	devlink_param_driverinit_value_set(devlink,
+					   DEVLINK_PARAM_GENERIC_ID_IO_EQ_SIZE,
+					   value);
+
+	value.vu32 = MLX5_NUM_ASYNC_EQE;
+	devlink_param_driverinit_value_set(devlink,
+					   DEVLINK_PARAM_GENERIC_ID_EVENT_EQ_SIZE,
+					   value);
+}
+
+static const struct devlink_param enable_eth_param =
+	DEVLINK_PARAM_GENERIC(ENABLE_ETH, BIT(DEVLINK_PARAM_CMODE_DRIVERINIT),
+			      NULL, NULL, NULL);
+
+static int mlx5_devlink_eth_param_register(struct devlink *devlink)
+{
+	struct mlx5_core_dev *dev = devlink_priv(devlink);
+	union devlink_param_value value;
+	int err;
+
+	if (!mlx5_eth_supported(dev))
+		return 0;
+
+	err = devlink_param_register(devlink, &enable_eth_param);
+	if (err)
+		return err;
+
+	value.vbool = true;
+	devlink_param_driverinit_value_set(devlink,
+					   DEVLINK_PARAM_GENERIC_ID_ENABLE_ETH,
+					   value);
+	return 0;
+}
+
+static void mlx5_devlink_eth_param_unregister(struct devlink *devlink)
+{
+	struct mlx5_core_dev *dev = devlink_priv(devlink);
+
+	if (!mlx5_eth_supported(dev))
+		return;
+
+	devlink_param_unregister(devlink, &enable_eth_param);
+}
+
+static int mlx5_devlink_enable_rdma_validate(struct devlink *devlink, u32 id,
+					     union devlink_param_value val,
+					     struct netlink_ext_ack *extack)
+{
+	struct mlx5_core_dev *dev = devlink_priv(devlink);
+	bool new_state = val.vbool;
+
+	if (new_state && !mlx5_rdma_supported(dev))
+		return -EOPNOTSUPP;
+	return 0;
+}
+
+static const struct devlink_param enable_rdma_param =
+	DEVLINK_PARAM_GENERIC(ENABLE_RDMA, BIT(DEVLINK_PARAM_CMODE_DRIVERINIT),
+			      NULL, NULL, mlx5_devlink_enable_rdma_validate);
+
+static int mlx5_devlink_rdma_param_register(struct devlink *devlink)
+{
+	union devlink_param_value value;
+	int err;
+
+	if (!IS_ENABLED(CONFIG_MLX5_INFINIBAND))
+		return 0;
+
+	err = devlink_param_register(devlink, &enable_rdma_param);
+	if (err)
+		return err;
+
+	value.vbool = true;
+	devlink_param_driverinit_value_set(devlink,
+					   DEVLINK_PARAM_GENERIC_ID_ENABLE_RDMA,
+					   value);
+	return 0;
+}
+
+static void mlx5_devlink_rdma_param_unregister(struct devlink *devlink)
+{
+	if (!IS_ENABLED(CONFIG_MLX5_INFINIBAND))
+		return;
+
+	devlink_param_unregister(devlink, &enable_rdma_param);
+}
+
+static const struct devlink_param enable_vnet_param =
+	DEVLINK_PARAM_GENERIC(ENABLE_VNET, BIT(DEVLINK_PARAM_CMODE_DRIVERINIT),
+			      NULL, NULL, NULL);
+
+static int mlx5_devlink_vnet_param_register(struct devlink *devlink)
+{
+	struct mlx5_core_dev *dev = devlink_priv(devlink);
+	union devlink_param_value value;
+	int err;
+
+	if (!mlx5_vnet_supported(dev))
+		return 0;
+
+	err = devlink_param_register(devlink, &enable_vnet_param);
+	if (err)
+		return err;
+
+	value.vbool = true;
+	devlink_param_driverinit_value_set(devlink,
+					   DEVLINK_PARAM_GENERIC_ID_ENABLE_VNET,
+					   value);
+	return 0;
+}
+
+static void mlx5_devlink_vnet_param_unregister(struct devlink *devlink)
+{
+	struct mlx5_core_dev *dev = devlink_priv(devlink);
+
+	if (!mlx5_vnet_supported(dev))
+		return;
+
+	devlink_param_unregister(devlink, &enable_vnet_param);
+}
+
+static int mlx5_devlink_auxdev_params_register(struct devlink *devlink)
+{
+	int err;
+
+	err = mlx5_devlink_eth_param_register(devlink);
+	if (err)
+		return err;
+
+	err = mlx5_devlink_rdma_param_register(devlink);
+	if (err)
+		goto rdma_err;
+
+	err = mlx5_devlink_vnet_param_register(devlink);
+	if (err)
+		goto vnet_err;
+	return 0;
+
+vnet_err:
+	mlx5_devlink_rdma_param_unregister(devlink);
+rdma_err:
+	mlx5_devlink_eth_param_unregister(devlink);
+	return err;
+}
+
+static void mlx5_devlink_auxdev_params_unregister(struct devlink *devlink)
+{
+	mlx5_devlink_vnet_param_unregister(devlink);
+	mlx5_devlink_rdma_param_unregister(devlink);
+	mlx5_devlink_eth_param_unregister(devlink);
+}
+
+static int mlx5_devlink_max_uc_list_validate(struct devlink *devlink, u32 id,
+					     union devlink_param_value val,
+					     struct netlink_ext_ack *extack)
+{
+	struct mlx5_core_dev *dev = devlink_priv(devlink);
+
+	if (val.vu32 == 0) {
+		NL_SET_ERR_MSG_MOD(extack, "max_macs value must be greater than 0");
+		return -EINVAL;
+	}
+
+	if (!is_power_of_2(val.vu32)) {
+		NL_SET_ERR_MSG_MOD(extack, "Only power of 2 values are supported for max_macs");
+		return -EINVAL;
+	}
+
+	if (ilog2(val.vu32) >
+	    MLX5_CAP_GEN_MAX(dev, log_max_current_uc_list)) {
+		NL_SET_ERR_MSG_MOD(extack, "max_macs value is out of the supported range");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static const struct devlink_param max_uc_list_param =
+	DEVLINK_PARAM_GENERIC(MAX_MACS, BIT(DEVLINK_PARAM_CMODE_DRIVERINIT),
+			      NULL, NULL, mlx5_devlink_max_uc_list_validate);
+
+static int mlx5_devlink_max_uc_list_param_register(struct devlink *devlink)
+{
+	struct mlx5_core_dev *dev = devlink_priv(devlink);
+	union devlink_param_value value;
+	int err;
+
+	if (!MLX5_CAP_GEN_MAX(dev, log_max_current_uc_list_wr_supported))
+		return 0;
+
+	err = devlink_param_register(devlink, &max_uc_list_param);
+	if (err)
+		return err;
+
+	value.vu32 = 1 << MLX5_CAP_GEN(dev, log_max_current_uc_list);
+	devlink_param_driverinit_value_set(devlink,
+					   DEVLINK_PARAM_GENERIC_ID_MAX_MACS,
+					   value);
+	return 0;
+}
+
+static void
+mlx5_devlink_max_uc_list_param_unregister(struct devlink *devlink)
+{
+	struct mlx5_core_dev *dev = devlink_priv(devlink);
+
+	if (!MLX5_CAP_GEN_MAX(dev, log_max_current_uc_list_wr_supported))
+		return;
+
+	devlink_param_unregister(devlink, &max_uc_list_param);
+}
+
+#define MLX5_TRAP_DROP(_id, _group_id)					\
+	DEVLINK_TRAP_GENERIC(DROP, DROP, _id,				\
+			     DEVLINK_TRAP_GROUP_GENERIC_ID_##_group_id, \
+			     DEVLINK_TRAP_METADATA_TYPE_F_IN_PORT)
+
+static const struct devlink_trap mlx5_traps_arr[] = {
+	MLX5_TRAP_DROP(INGRESS_VLAN_FILTER, L2_DROPS),
+	MLX5_TRAP_DROP(DMAC_FILTER, L2_DROPS),
+};
+
+static const struct devlink_trap_group mlx5_trap_groups_arr[] = {
+	DEVLINK_TRAP_GROUP_GENERIC(L2_DROPS, 0),
+};
+
+static int mlx5_devlink_traps_register(struct devlink *devlink)
+{
+	struct mlx5_core_dev *core_dev = devlink_priv(devlink);
+	int err;
+
+	err = devl_trap_groups_register(devlink, mlx5_trap_groups_arr,
+					ARRAY_SIZE(mlx5_trap_groups_arr));
+	if (err)
+		return err;
+
+	err = devl_traps_register(devlink, mlx5_traps_arr, ARRAY_SIZE(mlx5_traps_arr),
+				  &core_dev->priv);
+	if (err)
+		goto err_trap_group;
+	return 0;
+
+err_trap_group:
+	devl_trap_groups_unregister(devlink, mlx5_trap_groups_arr,
+				    ARRAY_SIZE(mlx5_trap_groups_arr));
+	return err;
+}
+
+static void mlx5_devlink_traps_unregister(struct devlink *devlink)
+{
+	devl_traps_unregister(devlink, mlx5_traps_arr, ARRAY_SIZE(mlx5_traps_arr));
+	devl_trap_groups_unregister(devlink, mlx5_trap_groups_arr,
+				    ARRAY_SIZE(mlx5_trap_groups_arr));
+}
+
+int mlx5_devlink_register(struct devlink *devlink)
+{
+	struct mlx5_core_dev *dev = devlink_priv(devlink);
+	int err;
+
+	err = devlink_params_register(devlink, mlx5_devlink_params,
+				      ARRAY_SIZE(mlx5_devlink_params));
+	if (err)
+		return err;
+
+	mlx5_devlink_set_params_init_values(devlink);
+
+	err = mlx5_devlink_auxdev_params_register(devlink);
+	if (err)
+		goto auxdev_reg_err;
+
+	err = mlx5_devlink_max_uc_list_param_register(devlink);
+	if (err)
+		goto max_uc_list_err;
+
+	err = mlx5_devlink_traps_register(devlink);
+	if (err)
+		goto traps_reg_err;
+
+	if (!mlx5_core_is_mp_slave(dev))
+		devlink_set_features(devlink, DEVLINK_F_RELOAD);
+
+	return 0;
+
+traps_reg_err:
+	mlx5_devlink_max_uc_list_param_unregister(devlink);
+max_uc_list_err:
+	mlx5_devlink_auxdev_params_unregister(devlink);
+auxdev_reg_err:
+	devlink_params_unregister(devlink, mlx5_devlink_params,
+				  ARRAY_SIZE(mlx5_devlink_params));
+	return err;
+}
+
+void mlx5_devlink_unregister(struct devlink *devlink)
+{
+	mlx5_devlink_traps_unregister(devlink);
+	mlx5_devlink_max_uc_list_param_unregister(devlink);
+	mlx5_devlink_auxdev_params_unregister(devlink);
+	devlink_params_unregister(devlink, mlx5_devlink_params,
+				  ARRAY_SIZE(mlx5_devlink_params));
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/devlink.h b/drivers/net/ethernet/mellanox/mlx5/core/devlink.h
new file mode 100644
index 000000000..30bf48827
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/devlink.h
@@ -0,0 +1,39 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2019, Mellanox Technologies */
+
+#ifndef __MLX5_DEVLINK_H__
+#define __MLX5_DEVLINK_H__
+
+#include <net/devlink.h>
+
+enum mlx5_devlink_param_id {
+	MLX5_DEVLINK_PARAM_ID_BASE = DEVLINK_PARAM_GENERIC_ID_MAX,
+	MLX5_DEVLINK_PARAM_ID_FLOW_STEERING_MODE,
+	MLX5_DEVLINK_PARAM_ID_ESW_LARGE_GROUP_NUM,
+	MLX5_DEVLINK_PARAM_ID_ESW_PORT_METADATA,
+};
+
+struct mlx5_trap_ctx {
+	int id;
+	int action;
+};
+
+struct mlx5_devlink_trap {
+	struct mlx5_trap_ctx trap;
+	void *item;
+	struct list_head list;
+};
+
+struct mlx5_core_dev;
+void mlx5_devlink_trap_report(struct mlx5_core_dev *dev, int trap_id, struct sk_buff *skb,
+			      struct devlink_port *dl_port);
+int mlx5_devlink_trap_get_num_active(struct mlx5_core_dev *dev);
+int mlx5_devlink_traps_get_action(struct mlx5_core_dev *dev, int trap_id,
+				  enum devlink_trap_action *action);
+
+struct devlink *mlx5_devlink_alloc(struct device *dev);
+void mlx5_devlink_free(struct devlink *devlink);
+int mlx5_devlink_register(struct devlink *devlink);
+void mlx5_devlink_unregister(struct devlink *devlink);
+
+#endif /* __MLX5_DEVLINK_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/diag/cmd_tracepoint.h b/drivers/net/ethernet/mellanox/mlx5/core/diag/cmd_tracepoint.h
new file mode 100644
index 000000000..406ebe174
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/diag/cmd_tracepoint.h
@@ -0,0 +1,45 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2022, NVIDIA CORPORATION & AFFILIATES. All rights reserved. */
+
+#undef TRACE_SYSTEM
+#define TRACE_SYSTEM mlx5
+
+#if !defined(_MLX5_CMD_TP_H_) || defined(TRACE_HEADER_MULTI_READ)
+#define _MLX5_CMD_TP_H_
+
+#include <linux/tracepoint.h>
+#include <linux/trace_seq.h>
+
+TRACE_EVENT(mlx5_cmd,
+	    TP_PROTO(const char *command_str, u16 opcode, u16 op_mod,
+		     const char *status_str, u8 status, u32 syndrome, int err),
+	    TP_ARGS(command_str, opcode, op_mod, status_str, status, syndrome, err),
+	    TP_STRUCT__entry(__string(command_str, command_str)
+			     __field(u16, opcode)
+			     __field(u16, op_mod)
+			    __string(status_str, status_str)
+			    __field(u8, status)
+			    __field(u32, syndrome)
+			    __field(int, err)
+			    ),
+	    TP_fast_assign(__assign_str(command_str, command_str);
+			__entry->opcode = opcode;
+			__entry->op_mod = op_mod;
+			__assign_str(status_str, status_str);
+			__entry->status = status;
+			__entry->syndrome = syndrome;
+			__entry->err = err;
+	    ),
+	    TP_printk("%s(0x%x) op_mod(0x%x) failed, status %s(0x%x), syndrome (0x%x), err(%d)",
+		      __get_str(command_str), __entry->opcode, __entry->op_mod,
+		      __get_str(status_str), __entry->status, __entry->syndrome,
+		      __entry->err)
+);
+
+#endif /* _MLX5_CMD_TP_H_ */
+
+#undef TRACE_INCLUDE_PATH
+#define TRACE_INCLUDE_PATH ./diag
+#undef TRACE_INCLUDE_FILE
+#define TRACE_INCLUDE_FILE cmd_tracepoint
+#include <trace/define_trace.h>
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/diag/crdump.c b/drivers/net/ethernet/mellanox/mlx5/core/diag/crdump.c
new file mode 100644
index 000000000..28d02749d
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/diag/crdump.c
@@ -0,0 +1,115 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2019 Mellanox Technologies */
+
+#include <linux/mlx5/driver.h>
+#include "mlx5_core.h"
+#include "lib/pci_vsc.h"
+#include "lib/mlx5.h"
+
+#define BAD_ACCESS			0xBADACCE5
+#define MLX5_PROTECTED_CR_SCAN_CRSPACE	0x7
+
+static bool mlx5_crdump_enabled(struct mlx5_core_dev *dev)
+{
+	return !!dev->priv.health.crdump_size;
+}
+
+static int mlx5_crdump_fill(struct mlx5_core_dev *dev, u32 *cr_data)
+{
+	u32 crdump_size = dev->priv.health.crdump_size;
+	int i, ret;
+
+	for (i = 0; i < (crdump_size / 4); i++)
+		cr_data[i] = BAD_ACCESS;
+
+	ret = mlx5_vsc_gw_read_block_fast(dev, cr_data, crdump_size);
+	if (ret <= 0) {
+		if (ret == 0)
+			return -EIO;
+		return ret;
+	}
+
+	if (crdump_size != ret) {
+		mlx5_core_warn(dev, "failed to read full dump, read %d out of %u\n",
+			       ret, crdump_size);
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+int mlx5_crdump_collect(struct mlx5_core_dev *dev, u32 *cr_data)
+{
+	int ret;
+
+	if (!mlx5_crdump_enabled(dev))
+		return -ENODEV;
+
+	ret = mlx5_vsc_gw_lock(dev);
+	if (ret) {
+		mlx5_core_warn(dev, "crdump: failed to lock vsc gw err %d\n",
+			       ret);
+		return ret;
+	}
+	/* Verify no other PF is running cr-dump or sw reset */
+	ret = mlx5_vsc_sem_set_space(dev, MLX5_SEMAPHORE_SW_RESET,
+				     MLX5_VSC_LOCK);
+	if (ret) {
+		mlx5_core_warn(dev, "Failed to lock SW reset semaphore\n");
+		goto unlock_gw;
+	}
+
+	ret = mlx5_vsc_gw_set_space(dev, MLX5_VSC_SPACE_SCAN_CRSPACE, NULL);
+	if (ret)
+		goto unlock_sem;
+
+	ret = mlx5_crdump_fill(dev, cr_data);
+
+unlock_sem:
+	mlx5_vsc_sem_set_space(dev, MLX5_SEMAPHORE_SW_RESET, MLX5_VSC_UNLOCK);
+unlock_gw:
+	mlx5_vsc_gw_unlock(dev);
+	return ret;
+}
+
+int mlx5_crdump_enable(struct mlx5_core_dev *dev)
+{
+	struct mlx5_priv *priv = &dev->priv;
+	u32 space_size;
+	int ret;
+
+	if (!mlx5_core_is_pf(dev) || !mlx5_vsc_accessible(dev) ||
+	    mlx5_crdump_enabled(dev))
+		return 0;
+
+	ret = mlx5_vsc_gw_lock(dev);
+	if (ret)
+		return ret;
+
+	/* Check if space is supported and get space size */
+	ret = mlx5_vsc_gw_set_space(dev, MLX5_VSC_SPACE_SCAN_CRSPACE,
+				    &space_size);
+	if (ret) {
+		/* Unlock and mask error since space is not supported */
+		mlx5_vsc_gw_unlock(dev);
+		return 0;
+	}
+
+	if (!space_size) {
+		mlx5_core_warn(dev, "Invalid Crspace size, zero\n");
+		mlx5_vsc_gw_unlock(dev);
+		return -EINVAL;
+	}
+
+	ret = mlx5_vsc_gw_unlock(dev);
+	if (ret)
+		return ret;
+
+	priv->health.crdump_size = space_size;
+	return 0;
+}
+
+void mlx5_crdump_disable(struct mlx5_core_dev *dev)
+{
+	dev->priv.health.crdump_size = 0;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/diag/en_rep_tracepoint.h b/drivers/net/ethernet/mellanox/mlx5/core/diag/en_rep_tracepoint.h
new file mode 100644
index 000000000..f15718db5
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/diag/en_rep_tracepoint.h
@@ -0,0 +1,54 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2019 Mellanox Technologies. */
+
+#undef TRACE_SYSTEM
+#define TRACE_SYSTEM mlx5
+
+#if !defined(_MLX5_EN_REP_TP_) || defined(TRACE_HEADER_MULTI_READ)
+#define _MLX5_EN_REP_TP_
+
+#include <linux/tracepoint.h>
+#include <linux/trace_seq.h>
+#include "en_rep.h"
+
+TRACE_EVENT(mlx5e_rep_neigh_update,
+	    TP_PROTO(const struct mlx5e_neigh_hash_entry *nhe, const u8 *ha,
+		     bool neigh_connected),
+	    TP_ARGS(nhe, ha, neigh_connected),
+	    TP_STRUCT__entry(__string(devname, nhe->neigh_dev->name)
+			     __array(u8, ha, ETH_ALEN)
+			     __array(u8, v4, 4)
+			     __array(u8, v6, 16)
+			     __field(bool, neigh_connected)
+			     ),
+	    TP_fast_assign(const struct mlx5e_neigh *mn = &nhe->m_neigh;
+			struct in6_addr *pin6;
+			__be32 *p32;
+
+			__assign_str(devname, nhe->neigh_dev->name);
+			__entry->neigh_connected = neigh_connected;
+			memcpy(__entry->ha, ha, ETH_ALEN);
+
+			p32 = (__be32 *)__entry->v4;
+			pin6 = (struct in6_addr *)__entry->v6;
+			if (mn->family == AF_INET) {
+				*p32 = mn->dst_ip.v4;
+				ipv6_addr_set_v4mapped(*p32, pin6);
+			} else if (mn->family == AF_INET6) {
+				*pin6 = mn->dst_ip.v6;
+			}
+			),
+	    TP_printk("netdev: %s MAC: %pM IPv4: %pI4 IPv6: %pI6c neigh_connected=%d\n",
+		      __get_str(devname), __entry->ha,
+		      __entry->v4, __entry->v6, __entry->neigh_connected
+		      )
+);
+
+#endif /* _MLX5_EN_REP_TP_ */
+
+/* This part must be outside protection */
+#undef TRACE_INCLUDE_PATH
+#define TRACE_INCLUDE_PATH ./diag
+#undef TRACE_INCLUDE_FILE
+#define TRACE_INCLUDE_FILE en_rep_tracepoint
+#include <trace/define_trace.h>
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/diag/en_tc_tracepoint.c b/drivers/net/ethernet/mellanox/mlx5/core/diag/en_tc_tracepoint.c
new file mode 100644
index 000000000..c5dc6c50f
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/diag/en_tc_tracepoint.c
@@ -0,0 +1,58 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2019 Mellanox Technologies. */
+
+#define CREATE_TRACE_POINTS
+#include "en_tc_tracepoint.h"
+
+void put_ids_to_array(int *ids,
+		      const struct flow_action_entry *entries,
+		      unsigned int num)
+{
+	unsigned int i;
+
+	for (i = 0; i < num; i++)
+		ids[i] = entries[i].id;
+}
+
+#define NAME_SIZE 16
+
+static const char FLOWACT2STR[NUM_FLOW_ACTIONS][NAME_SIZE] = {
+	[FLOW_ACTION_ACCEPT]	= "ACCEPT",
+	[FLOW_ACTION_DROP]	= "DROP",
+	[FLOW_ACTION_TRAP]	= "TRAP",
+	[FLOW_ACTION_GOTO]	= "GOTO",
+	[FLOW_ACTION_REDIRECT]	= "REDIRECT",
+	[FLOW_ACTION_MIRRED]	= "MIRRED",
+	[FLOW_ACTION_VLAN_PUSH]	= "VLAN_PUSH",
+	[FLOW_ACTION_VLAN_POP]	= "VLAN_POP",
+	[FLOW_ACTION_VLAN_MANGLE]	= "VLAN_MANGLE",
+	[FLOW_ACTION_TUNNEL_ENCAP]	= "TUNNEL_ENCAP",
+	[FLOW_ACTION_TUNNEL_DECAP]	= "TUNNEL_DECAP",
+	[FLOW_ACTION_MANGLE]	= "MANGLE",
+	[FLOW_ACTION_ADD]	= "ADD",
+	[FLOW_ACTION_CSUM]	= "CSUM",
+	[FLOW_ACTION_MARK]	= "MARK",
+	[FLOW_ACTION_WAKE]	= "WAKE",
+	[FLOW_ACTION_QUEUE]	= "QUEUE",
+	[FLOW_ACTION_SAMPLE]	= "SAMPLE",
+	[FLOW_ACTION_POLICE]	= "POLICE",
+	[FLOW_ACTION_CT]	= "CT",
+};
+
+const char *parse_action(struct trace_seq *p,
+			 int *ids,
+			 unsigned int num)
+{
+	const char *ret = trace_seq_buffer_ptr(p);
+	unsigned int i;
+
+	for (i = 0; i < num; i++) {
+		if (ids[i] < NUM_FLOW_ACTIONS)
+			trace_seq_printf(p, "%s ", FLOWACT2STR[ids[i]]);
+		else
+			trace_seq_printf(p, "UNKNOWN ");
+	}
+
+	trace_seq_putc(p, 0);
+	return ret;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/diag/en_tc_tracepoint.h b/drivers/net/ethernet/mellanox/mlx5/core/diag/en_tc_tracepoint.h
new file mode 100644
index 000000000..ac52ef37f
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/diag/en_tc_tracepoint.h
@@ -0,0 +1,114 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2019 Mellanox Technologies. */
+
+#undef TRACE_SYSTEM
+#define TRACE_SYSTEM mlx5
+
+#if !defined(_MLX5_TC_TP_) || defined(TRACE_HEADER_MULTI_READ)
+#define _MLX5_TC_TP_
+
+#include <linux/tracepoint.h>
+#include <linux/trace_seq.h>
+#include <net/flow_offload.h>
+#include "en_rep.h"
+
+#define __parse_action(ids, num) parse_action(p, ids, num)
+
+void put_ids_to_array(int *ids,
+		      const struct flow_action_entry *entries,
+		      unsigned int num);
+
+const char *parse_action(struct trace_seq *p,
+			 int *ids,
+			 unsigned int num);
+
+DECLARE_EVENT_CLASS(mlx5e_flower_template,
+		    TP_PROTO(const struct flow_cls_offload *f),
+		    TP_ARGS(f),
+		    TP_STRUCT__entry(__field(void *, cookie)
+				     __field(unsigned int, num)
+				     __dynamic_array(int, ids, f->rule ?
+					     f->rule->action.num_entries : 0)
+				     ),
+		    TP_fast_assign(__entry->cookie = (void *)f->cookie;
+			__entry->num = (f->rule ?
+				f->rule->action.num_entries : 0);
+			if (__entry->num)
+				put_ids_to_array(__get_dynamic_array(ids),
+						 f->rule->action.entries,
+						 f->rule->action.num_entries);
+			),
+		    TP_printk("cookie=%p actions= %s\n",
+			      __entry->cookie, __entry->num ?
+				      __parse_action(__get_dynamic_array(ids),
+						     __entry->num) : "NULL"
+			      )
+);
+
+DEFINE_EVENT(mlx5e_flower_template, mlx5e_configure_flower,
+	     TP_PROTO(const struct flow_cls_offload *f),
+	     TP_ARGS(f)
+	     );
+
+DEFINE_EVENT(mlx5e_flower_template, mlx5e_delete_flower,
+	     TP_PROTO(const struct flow_cls_offload *f),
+	     TP_ARGS(f)
+	     );
+
+TRACE_EVENT(mlx5e_stats_flower,
+	    TP_PROTO(const struct flow_cls_offload *f),
+	    TP_ARGS(f),
+	    TP_STRUCT__entry(__field(void *, cookie)
+			     __field(u64, bytes)
+			     __field(u64, packets)
+			     __field(u64, lastused)
+			     ),
+	    TP_fast_assign(__entry->cookie = (void *)f->cookie;
+		__entry->bytes = f->stats.bytes;
+		__entry->packets = f->stats.pkts;
+		__entry->lastused = f->stats.lastused;
+		),
+	    TP_printk("cookie=%p bytes=%llu packets=%llu lastused=%llu\n",
+		      __entry->cookie, __entry->bytes,
+		      __entry->packets, __entry->lastused
+		      )
+);
+
+TRACE_EVENT(mlx5e_tc_update_neigh_used_value,
+	    TP_PROTO(const struct mlx5e_neigh_hash_entry *nhe, bool neigh_used),
+	    TP_ARGS(nhe, neigh_used),
+	    TP_STRUCT__entry(__string(devname, nhe->neigh_dev->name)
+			     __array(u8, v4, 4)
+			     __array(u8, v6, 16)
+			     __field(bool, neigh_used)
+			     ),
+	    TP_fast_assign(const struct mlx5e_neigh *mn = &nhe->m_neigh;
+			struct in6_addr *pin6;
+			__be32 *p32;
+
+			__assign_str(devname, nhe->neigh_dev->name);
+			__entry->neigh_used = neigh_used;
+
+			p32 = (__be32 *)__entry->v4;
+			pin6 = (struct in6_addr *)__entry->v6;
+			if (mn->family == AF_INET) {
+				*p32 = mn->dst_ip.v4;
+				ipv6_addr_set_v4mapped(*p32, pin6);
+			} else if (mn->family == AF_INET6) {
+				*pin6 = mn->dst_ip.v6;
+			}
+			),
+	    TP_printk("netdev: %s IPv4: %pI4 IPv6: %pI6c neigh_used=%d\n",
+		      __get_str(devname), __entry->v4, __entry->v6,
+		      __entry->neigh_used
+		      )
+);
+
+#endif /* _MLX5_TC_TP_ */
+
+/* This part must be outside protection */
+#undef TRACE_INCLUDE_PATH
+#define TRACE_INCLUDE_PATH ./diag
+#undef TRACE_INCLUDE_FILE
+#define TRACE_INCLUDE_FILE en_tc_tracepoint
+#include <trace/define_trace.h>
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/diag/fs_tracepoint.c b/drivers/net/ethernet/mellanox/mlx5/core/diag/fs_tracepoint.c
new file mode 100644
index 000000000..c5bb79a4f
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/diag/fs_tracepoint.c
@@ -0,0 +1,279 @@
+/*
+ * Copyright (c) 2017, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#define CREATE_TRACE_POINTS
+
+#include "fs_tracepoint.h"
+#include <linux/stringify.h>
+
+#define DECLARE_MASK_VAL(type, name) struct {type m; type v; } name
+#define MASK_VAL(type, spec, name, mask, val, fld)	\
+		DECLARE_MASK_VAL(type, name) =		\
+			{.m = MLX5_GET(spec, mask, fld),\
+			 .v = MLX5_GET(spec, val, fld)}
+#define MASK_VAL_BE(type, spec, name, mask, val, fld)	\
+		    DECLARE_MASK_VAL(type, name) =	\
+			{.m = MLX5_GET_BE(type, spec, mask, fld),\
+			 .v = MLX5_GET_BE(type, spec, val, fld)}
+#define GET_MASKED_VAL(name) (name.m & name.v)
+
+#define GET_MASK_VAL(name, type, mask, val, fld)	\
+		(name.m = MLX5_GET(type, mask, fld),	\
+		 name.v = MLX5_GET(type, val, fld),	\
+		 name.m & name.v)
+#define PRINT_MASKED_VAL(name, p, format) {		\
+	if (name.m)			\
+		trace_seq_printf(p, __stringify(name) "=" format " ", name.v); \
+	}
+#define PRINT_MASKED_VALP(name, cast, p, format) {	\
+	if (name.m)			\
+		trace_seq_printf(p, __stringify(name) "=" format " ",	       \
+				 (cast)&name.v);\
+	}
+
+static void print_lyr_2_4_hdrs(struct trace_seq *p,
+			       const u32 *mask, const u32 *value)
+{
+#define MASK_VAL_L2(type, name, fld) \
+	MASK_VAL(type, fte_match_set_lyr_2_4, name, mask, value, fld)
+	DECLARE_MASK_VAL(u64, smac) = {
+		.m = MLX5_GET(fte_match_set_lyr_2_4, mask, smac_47_16) << 16 |
+		     MLX5_GET(fte_match_set_lyr_2_4, mask, smac_15_0),
+		.v = MLX5_GET(fte_match_set_lyr_2_4, value, smac_47_16) << 16 |
+		     MLX5_GET(fte_match_set_lyr_2_4, value, smac_15_0)};
+	DECLARE_MASK_VAL(u64, dmac) = {
+		.m = MLX5_GET(fte_match_set_lyr_2_4, mask, dmac_47_16) << 16 |
+		     MLX5_GET(fte_match_set_lyr_2_4, mask, dmac_15_0),
+		.v = MLX5_GET(fte_match_set_lyr_2_4, value, dmac_47_16) << 16 |
+		     MLX5_GET(fte_match_set_lyr_2_4, value, dmac_15_0)};
+	MASK_VAL_L2(u16, ethertype, ethertype);
+	MASK_VAL_L2(u8, ip_version, ip_version);
+
+	PRINT_MASKED_VALP(smac, u8 *, p, "%pM");
+	PRINT_MASKED_VALP(dmac, u8 *, p, "%pM");
+	PRINT_MASKED_VAL(ethertype, p, "%04x");
+
+	if ((ethertype.m == 0xffff && ethertype.v == ETH_P_IP) ||
+	    (ip_version.m == 0xf && ip_version.v == 4)) {
+#define MASK_VAL_L2_BE(type, name, fld) \
+	MASK_VAL_BE(type, fte_match_set_lyr_2_4, name, mask, value, fld)
+		MASK_VAL_L2_BE(u32, src_ipv4,
+			       src_ipv4_src_ipv6.ipv4_layout.ipv4);
+		MASK_VAL_L2_BE(u32, dst_ipv4,
+			       dst_ipv4_dst_ipv6.ipv4_layout.ipv4);
+
+		PRINT_MASKED_VALP(src_ipv4, typeof(&src_ipv4.v), p,
+				  "%pI4");
+		PRINT_MASKED_VALP(dst_ipv4, typeof(&dst_ipv4.v), p,
+				  "%pI4");
+	} else if ((ethertype.m == 0xffff && ethertype.v == ETH_P_IPV6) ||
+		   (ip_version.m == 0xf && ip_version.v == 6)) {
+		static const struct in6_addr full_ones = {
+			.in6_u.u6_addr32 = {__constant_htonl(0xffffffff),
+					    __constant_htonl(0xffffffff),
+					    __constant_htonl(0xffffffff),
+					    __constant_htonl(0xffffffff)},
+		};
+		DECLARE_MASK_VAL(struct in6_addr, src_ipv6);
+		DECLARE_MASK_VAL(struct in6_addr, dst_ipv6);
+
+		memcpy(src_ipv6.m.in6_u.u6_addr8,
+		       MLX5_ADDR_OF(fte_match_set_lyr_2_4, mask,
+				    src_ipv4_src_ipv6.ipv6_layout.ipv6),
+		       sizeof(src_ipv6.m));
+		memcpy(dst_ipv6.m.in6_u.u6_addr8,
+		       MLX5_ADDR_OF(fte_match_set_lyr_2_4, mask,
+				    dst_ipv4_dst_ipv6.ipv6_layout.ipv6),
+		       sizeof(dst_ipv6.m));
+		memcpy(src_ipv6.v.in6_u.u6_addr8,
+		       MLX5_ADDR_OF(fte_match_set_lyr_2_4, value,
+				    src_ipv4_src_ipv6.ipv6_layout.ipv6),
+		       sizeof(src_ipv6.v));
+		memcpy(dst_ipv6.v.in6_u.u6_addr8,
+		       MLX5_ADDR_OF(fte_match_set_lyr_2_4, value,
+				    dst_ipv4_dst_ipv6.ipv6_layout.ipv6),
+		       sizeof(dst_ipv6.v));
+
+		if (!memcmp(&src_ipv6.m, &full_ones, sizeof(full_ones)))
+			trace_seq_printf(p, "src_ipv6=%pI6 ",
+					 &src_ipv6.v);
+		if (!memcmp(&dst_ipv6.m, &full_ones, sizeof(full_ones)))
+			trace_seq_printf(p, "dst_ipv6=%pI6 ",
+					 &dst_ipv6.v);
+	}
+
+#define PRINT_MASKED_VAL_L2(type, name, fld, p, format) {\
+	MASK_VAL_L2(type, name, fld);		         \
+	PRINT_MASKED_VAL(name, p, format);		 \
+}
+
+	PRINT_MASKED_VAL_L2(u8, ip_protocol, ip_protocol, p, "%02x");
+	PRINT_MASKED_VAL_L2(u16, tcp_flags, tcp_flags, p, "%x");
+	PRINT_MASKED_VAL_L2(u16, tcp_sport, tcp_sport, p, "%u");
+	PRINT_MASKED_VAL_L2(u16, tcp_dport, tcp_dport, p, "%u");
+	PRINT_MASKED_VAL_L2(u16, udp_sport, udp_sport, p, "%u");
+	PRINT_MASKED_VAL_L2(u16, udp_dport, udp_dport, p, "%u");
+	PRINT_MASKED_VAL_L2(u16, first_vid, first_vid, p, "%04x");
+	PRINT_MASKED_VAL_L2(u8, first_prio, first_prio, p, "%x");
+	PRINT_MASKED_VAL_L2(u8, first_cfi, first_cfi, p, "%d");
+	PRINT_MASKED_VAL_L2(u8, ip_dscp, ip_dscp, p, "%02x");
+	PRINT_MASKED_VAL_L2(u8, ip_ecn, ip_ecn, p, "%x");
+	PRINT_MASKED_VAL_L2(u8, cvlan_tag, cvlan_tag, p, "%d");
+	PRINT_MASKED_VAL_L2(u8, svlan_tag, svlan_tag, p, "%d");
+	PRINT_MASKED_VAL_L2(u8, frag, frag, p, "%d");
+}
+
+static void print_misc_parameters_hdrs(struct trace_seq *p,
+				       const u32 *mask, const u32 *value)
+{
+#define MASK_VAL_MISC(type, name, fld) \
+	MASK_VAL(type, fte_match_set_misc, name, mask, value, fld)
+#define PRINT_MASKED_VAL_MISC(type, name, fld, p, format) {\
+	MASK_VAL_MISC(type, name, fld);			   \
+	PRINT_MASKED_VAL(name, p, format);		   \
+}
+	DECLARE_MASK_VAL(u64, gre_key) = {
+		.m = MLX5_GET(fte_match_set_misc, mask, gre_key.nvgre.hi) << 8 |
+		     MLX5_GET(fte_match_set_misc, mask, gre_key.nvgre.lo),
+		.v = MLX5_GET(fte_match_set_misc, value, gre_key.nvgre.hi) << 8 |
+		     MLX5_GET(fte_match_set_misc, value, gre_key.nvgre.lo)};
+
+	PRINT_MASKED_VAL(gre_key, p, "%llu");
+	PRINT_MASKED_VAL_MISC(u32, source_sqn, source_sqn, p, "%u");
+	PRINT_MASKED_VAL_MISC(u16, source_port, source_port, p, "%u");
+	PRINT_MASKED_VAL_MISC(u8, outer_second_prio, outer_second_prio,
+			      p, "%u");
+	PRINT_MASKED_VAL_MISC(u8, outer_second_cfi, outer_second_cfi, p, "%u");
+	PRINT_MASKED_VAL_MISC(u16, outer_second_vid, outer_second_vid, p, "%u");
+	PRINT_MASKED_VAL_MISC(u8, inner_second_prio, inner_second_prio,
+			      p, "%u");
+	PRINT_MASKED_VAL_MISC(u8, inner_second_cfi, inner_second_cfi, p, "%u");
+	PRINT_MASKED_VAL_MISC(u16, inner_second_vid, inner_second_vid, p, "%u");
+
+	PRINT_MASKED_VAL_MISC(u8, outer_second_cvlan_tag,
+			      outer_second_cvlan_tag, p, "%u");
+	PRINT_MASKED_VAL_MISC(u8, inner_second_cvlan_tag,
+			      inner_second_cvlan_tag, p, "%u");
+	PRINT_MASKED_VAL_MISC(u8, outer_second_svlan_tag,
+			      outer_second_svlan_tag, p, "%u");
+	PRINT_MASKED_VAL_MISC(u8, inner_second_svlan_tag,
+			      inner_second_svlan_tag, p, "%u");
+
+	PRINT_MASKED_VAL_MISC(u8, gre_protocol, gre_protocol, p, "%u");
+
+	PRINT_MASKED_VAL_MISC(u32, vxlan_vni, vxlan_vni, p, "%u");
+	PRINT_MASKED_VAL_MISC(u32, outer_ipv6_flow_label, outer_ipv6_flow_label,
+			      p, "%x");
+	PRINT_MASKED_VAL_MISC(u32, inner_ipv6_flow_label, inner_ipv6_flow_label,
+			      p, "%x");
+}
+
+const char *parse_fs_hdrs(struct trace_seq *p,
+			  u8 match_criteria_enable,
+			  const u32 *mask_outer,
+			  const u32 *mask_misc,
+			  const u32 *mask_inner,
+			  const u32 *value_outer,
+			  const u32 *value_misc,
+			  const u32 *value_inner)
+{
+	const char *ret = trace_seq_buffer_ptr(p);
+
+	if (match_criteria_enable &
+	    1 << MLX5_CREATE_FLOW_GROUP_IN_MATCH_CRITERIA_ENABLE_OUTER_HEADERS) {
+		trace_seq_printf(p, "[outer] ");
+		print_lyr_2_4_hdrs(p, mask_outer, value_outer);
+	}
+
+	if (match_criteria_enable &
+	    1 << MLX5_CREATE_FLOW_GROUP_IN_MATCH_CRITERIA_ENABLE_MISC_PARAMETERS) {
+		trace_seq_printf(p, "[misc] ");
+		print_misc_parameters_hdrs(p, mask_misc, value_misc);
+	}
+	if (match_criteria_enable &
+	    1 << MLX5_CREATE_FLOW_GROUP_IN_MATCH_CRITERIA_ENABLE_INNER_HEADERS) {
+		trace_seq_printf(p, "[inner] ");
+		print_lyr_2_4_hdrs(p, mask_inner, value_inner);
+	}
+	trace_seq_putc(p, 0);
+	return ret;
+}
+
+const char *parse_fs_dst(struct trace_seq *p,
+			 const struct mlx5_flow_destination *dst,
+			 u32 counter_id)
+{
+	const char *ret = trace_seq_buffer_ptr(p);
+
+	switch (dst->type) {
+	case MLX5_FLOW_DESTINATION_TYPE_UPLINK:
+		trace_seq_printf(p, "uplink\n");
+		break;
+	case MLX5_FLOW_DESTINATION_TYPE_VPORT:
+		trace_seq_printf(p, "vport=%u\n", dst->vport.num);
+		break;
+	case MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE:
+		trace_seq_printf(p, "ft=%p\n", dst->ft);
+		break;
+	case MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE_NUM:
+		trace_seq_printf(p, "ft_num=%u\n", dst->ft_num);
+		break;
+	case MLX5_FLOW_DESTINATION_TYPE_TIR:
+		trace_seq_printf(p, "tir=%u\n", dst->tir_num);
+		break;
+	case MLX5_FLOW_DESTINATION_TYPE_FLOW_SAMPLER:
+		trace_seq_printf(p, "sampler_id=%u\n", dst->sampler_id);
+		break;
+	case MLX5_FLOW_DESTINATION_TYPE_COUNTER:
+		trace_seq_printf(p, "counter_id=%u\n", counter_id);
+		break;
+	case MLX5_FLOW_DESTINATION_TYPE_PORT:
+		trace_seq_printf(p, "port\n");
+		break;
+	case MLX5_FLOW_DESTINATION_TYPE_NONE:
+		trace_seq_printf(p, "none\n");
+		break;
+	}
+
+	trace_seq_putc(p, 0);
+	return ret;
+}
+
+EXPORT_TRACEPOINT_SYMBOL(mlx5_fs_add_ft);
+EXPORT_TRACEPOINT_SYMBOL(mlx5_fs_del_ft);
+EXPORT_TRACEPOINT_SYMBOL(mlx5_fs_add_fg);
+EXPORT_TRACEPOINT_SYMBOL(mlx5_fs_del_fg);
+EXPORT_TRACEPOINT_SYMBOL(mlx5_fs_set_fte);
+EXPORT_TRACEPOINT_SYMBOL(mlx5_fs_del_fte);
+EXPORT_TRACEPOINT_SYMBOL(mlx5_fs_add_rule);
+EXPORT_TRACEPOINT_SYMBOL(mlx5_fs_del_rule);
+
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/diag/fs_tracepoint.h b/drivers/net/ethernet/mellanox/mlx5/core/diag/fs_tracepoint.h
new file mode 100644
index 000000000..ddf1b87f1
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/diag/fs_tracepoint.h
@@ -0,0 +1,323 @@
+/*
+ * Copyright (c) 2017, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#if !defined(_MLX5_FS_TP_) || defined(TRACE_HEADER_MULTI_READ)
+#define _MLX5_FS_TP_
+
+#include <linux/tracepoint.h>
+#include <linux/trace_seq.h>
+#include "../fs_core.h"
+
+#undef TRACE_SYSTEM
+#define TRACE_SYSTEM mlx5
+
+#define __parse_fs_hdrs(match_criteria_enable, mouter, mmisc, minner, vouter, \
+			vinner, vmisc)					      \
+	parse_fs_hdrs(p, match_criteria_enable, mouter, mmisc, minner, vouter,\
+		      vinner, vmisc)
+
+const char *parse_fs_hdrs(struct trace_seq *p,
+			  u8 match_criteria_enable,
+			  const u32 *mask_outer,
+			  const u32 *mask_misc,
+			  const u32 *mask_inner,
+			  const u32 *value_outer,
+			  const u32 *value_misc,
+			  const u32 *value_inner);
+
+#define __parse_fs_dst(dst, counter_id) \
+	parse_fs_dst(p, (const struct mlx5_flow_destination *)dst, counter_id)
+
+const char *parse_fs_dst(struct trace_seq *p,
+			 const struct mlx5_flow_destination *dst,
+			 u32 counter_id);
+
+TRACE_EVENT(mlx5_fs_add_ft,
+	    TP_PROTO(const struct mlx5_flow_table *ft),
+	    TP_ARGS(ft),
+	    TP_STRUCT__entry(
+		__field(const struct mlx5_flow_table *, ft)
+		__field(u32, id)
+		__field(u32, level)
+		__field(u32, type)
+	    ),
+	    TP_fast_assign(
+			   __entry->ft = ft;
+			   __entry->id = ft->id;
+			   __entry->level = ft->level;
+			   __entry->type = ft->type;
+	    ),
+	    TP_printk("ft=%p id=%u level=%u type=%u \n",
+		      __entry->ft, __entry->id, __entry->level, __entry->type)
+	    );
+
+TRACE_EVENT(mlx5_fs_del_ft,
+	    TP_PROTO(const struct mlx5_flow_table *ft),
+	    TP_ARGS(ft),
+	    TP_STRUCT__entry(
+		__field(const struct mlx5_flow_table *, ft)
+		__field(u32, id)
+	    ),
+	    TP_fast_assign(
+			   __entry->ft = ft;
+			   __entry->id = ft->id;
+
+	    ),
+	    TP_printk("ft=%p id=%u\n",
+		      __entry->ft, __entry->id)
+	    );
+
+TRACE_EVENT(mlx5_fs_add_fg,
+	    TP_PROTO(const struct mlx5_flow_group *fg),
+	    TP_ARGS(fg),
+	    TP_STRUCT__entry(
+		__field(const struct mlx5_flow_group *, fg)
+		__field(const struct mlx5_flow_table *, ft)
+		__field(u32, start_index)
+		__field(u32, end_index)
+		__field(u32, id)
+		__field(u8, mask_enable)
+		__array(u32, mask_outer, MLX5_ST_SZ_DW(fte_match_set_lyr_2_4))
+		__array(u32, mask_inner, MLX5_ST_SZ_DW(fte_match_set_lyr_2_4))
+		__array(u32, mask_misc, MLX5_ST_SZ_DW(fte_match_set_misc))
+	    ),
+	    TP_fast_assign(
+			   __entry->fg = fg;
+			   fs_get_obj(__entry->ft, fg->node.parent);
+			   __entry->start_index = fg->start_index;
+			   __entry->end_index = fg->start_index + fg->max_ftes;
+			   __entry->id = fg->id;
+			   __entry->mask_enable = fg->mask.match_criteria_enable;
+			   memcpy(__entry->mask_outer,
+				  MLX5_ADDR_OF(fte_match_param,
+					       &fg->mask.match_criteria,
+					       outer_headers),
+				  sizeof(__entry->mask_outer));
+			   memcpy(__entry->mask_inner,
+				  MLX5_ADDR_OF(fte_match_param,
+					       &fg->mask.match_criteria,
+					       inner_headers),
+				  sizeof(__entry->mask_inner));
+			   memcpy(__entry->mask_misc,
+				  MLX5_ADDR_OF(fte_match_param,
+					       &fg->mask.match_criteria,
+					       misc_parameters),
+				  sizeof(__entry->mask_misc));
+
+	    ),
+	    TP_printk("fg=%p ft=%p id=%u start=%u end=%u bit_mask=%02x %s\n",
+		      __entry->fg, __entry->ft, __entry->id,
+		      __entry->start_index, __entry->end_index,
+		      __entry->mask_enable,
+		      __parse_fs_hdrs(__entry->mask_enable,
+				      __entry->mask_outer,
+				      __entry->mask_misc,
+				      __entry->mask_inner,
+				      __entry->mask_outer,
+				      __entry->mask_misc,
+				      __entry->mask_inner))
+	    );
+
+TRACE_EVENT(mlx5_fs_del_fg,
+	    TP_PROTO(const struct mlx5_flow_group *fg),
+	    TP_ARGS(fg),
+	    TP_STRUCT__entry(
+		__field(const struct mlx5_flow_group *, fg)
+		__field(u32, id)
+	    ),
+	    TP_fast_assign(
+			   __entry->fg = fg;
+			   __entry->id = fg->id;
+
+	    ),
+	    TP_printk("fg=%p id=%u\n",
+		      __entry->fg, __entry->id)
+	    );
+
+#define ACTION_FLAGS \
+	{MLX5_FLOW_CONTEXT_ACTION_ALLOW,	 "ALLOW"},\
+	{MLX5_FLOW_CONTEXT_ACTION_DROP,		 "DROP"},\
+	{MLX5_FLOW_CONTEXT_ACTION_FWD_DEST,	 "FWD"},\
+	{MLX5_FLOW_CONTEXT_ACTION_COUNT,	 "CNT"},\
+	{MLX5_FLOW_CONTEXT_ACTION_PACKET_REFORMAT, "REFORMAT"},\
+	{MLX5_FLOW_CONTEXT_ACTION_DECAP,	 "DECAP"},\
+	{MLX5_FLOW_CONTEXT_ACTION_MOD_HDR,	 "MOD_HDR"},\
+	{MLX5_FLOW_CONTEXT_ACTION_VLAN_PUSH,	 "VLAN_PUSH"},\
+	{MLX5_FLOW_CONTEXT_ACTION_VLAN_POP,	 "VLAN_POP"},\
+	{MLX5_FLOW_CONTEXT_ACTION_VLAN_PUSH_2,	 "VLAN_PUSH_2"},\
+	{MLX5_FLOW_CONTEXT_ACTION_VLAN_POP_2,	 "VLAN_POP_2"},\
+	{MLX5_FLOW_CONTEXT_ACTION_FWD_NEXT_PRIO, "NEXT_PRIO"}
+
+TRACE_EVENT(mlx5_fs_set_fte,
+	    TP_PROTO(const struct fs_fte *fte, int new_fte),
+	    TP_ARGS(fte, new_fte),
+	    TP_STRUCT__entry(
+		__field(const struct fs_fte *, fte)
+		__field(const struct mlx5_flow_group *, fg)
+		__field(u32, group_index)
+		__field(u32, index)
+		__field(u32, action)
+		__field(u32, flow_tag)
+		__field(u32, flow_source)
+		__field(u8,  mask_enable)
+		__field(int, new_fte)
+		__array(u32, mask_outer, MLX5_ST_SZ_DW(fte_match_set_lyr_2_4))
+		__array(u32, mask_inner, MLX5_ST_SZ_DW(fte_match_set_lyr_2_4))
+		__array(u32, mask_misc, MLX5_ST_SZ_DW(fte_match_set_misc))
+		__array(u32, value_outer, MLX5_ST_SZ_DW(fte_match_set_lyr_2_4))
+		__array(u32, value_inner, MLX5_ST_SZ_DW(fte_match_set_lyr_2_4))
+		__array(u32, value_misc, MLX5_ST_SZ_DW(fte_match_set_misc))
+	    ),
+	    TP_fast_assign(
+			   __entry->fte = fte;
+			   __entry->new_fte = new_fte;
+			   fs_get_obj(__entry->fg, fte->node.parent);
+			   __entry->group_index = __entry->fg->id;
+			   __entry->index = fte->index;
+			   __entry->action = fte->action.action;
+			   __entry->mask_enable = __entry->fg->mask.match_criteria_enable;
+			   __entry->flow_tag = fte->flow_context.flow_tag;
+			   __entry->flow_source = fte->flow_context.flow_source;
+			   memcpy(__entry->mask_outer,
+				  MLX5_ADDR_OF(fte_match_param,
+					       &__entry->fg->mask.match_criteria,
+					       outer_headers),
+				  sizeof(__entry->mask_outer));
+			   memcpy(__entry->mask_inner,
+				  MLX5_ADDR_OF(fte_match_param,
+					       &__entry->fg->mask.match_criteria,
+					       inner_headers),
+				  sizeof(__entry->mask_inner));
+			   memcpy(__entry->mask_misc,
+				  MLX5_ADDR_OF(fte_match_param,
+					       &__entry->fg->mask.match_criteria,
+					       misc_parameters),
+				  sizeof(__entry->mask_misc));
+			   memcpy(__entry->value_outer,
+				  MLX5_ADDR_OF(fte_match_param,
+					       &fte->val,
+					       outer_headers),
+				  sizeof(__entry->value_outer));
+			   memcpy(__entry->value_inner,
+				  MLX5_ADDR_OF(fte_match_param,
+					       &fte->val,
+					       inner_headers),
+				  sizeof(__entry->value_inner));
+			   memcpy(__entry->value_misc,
+				  MLX5_ADDR_OF(fte_match_param,
+					       &fte->val,
+					       misc_parameters),
+				  sizeof(__entry->value_misc));
+
+	    ),
+	    TP_printk("op=%s fte=%p fg=%p index=%u group_index=%u action=<%s> flow_tag=%x %s\n",
+		      __entry->new_fte ? "add" : "set",
+		      __entry->fte, __entry->fg, __entry->index,
+		      __entry->group_index, __print_flags(__entry->action, "|",
+							  ACTION_FLAGS),
+		      __entry->flow_tag,
+		      __parse_fs_hdrs(__entry->mask_enable,
+				      __entry->mask_outer,
+				      __entry->mask_misc,
+				      __entry->mask_inner,
+				      __entry->value_outer,
+				      __entry->value_misc,
+				      __entry->value_inner))
+	    );
+
+TRACE_EVENT(mlx5_fs_del_fte,
+	    TP_PROTO(const struct fs_fte *fte),
+	    TP_ARGS(fte),
+	    TP_STRUCT__entry(
+		__field(const struct fs_fte *, fte)
+		__field(u32, index)
+	    ),
+	    TP_fast_assign(
+			   __entry->fte = fte;
+			   __entry->index = fte->index;
+
+	    ),
+	    TP_printk("fte=%p index=%u\n",
+		      __entry->fte, __entry->index)
+	    );
+
+TRACE_EVENT(mlx5_fs_add_rule,
+	    TP_PROTO(const struct mlx5_flow_rule *rule),
+	    TP_ARGS(rule),
+	    TP_STRUCT__entry(
+		__field(const struct mlx5_flow_rule *, rule)
+		__field(const struct fs_fte *, fte)
+		__field(u32, sw_action)
+		__field(u32, index)
+		__field(u32, counter_id)
+		__array(u8, destination, sizeof(struct mlx5_flow_destination))
+	    ),
+	    TP_fast_assign(
+			   __entry->rule = rule;
+			   fs_get_obj(__entry->fte, rule->node.parent);
+			   __entry->index = __entry->fte->dests_size - 1;
+			   __entry->sw_action = rule->sw_action;
+			   memcpy(__entry->destination,
+				  &rule->dest_attr,
+				  sizeof(__entry->destination));
+			   if (rule->dest_attr.type &
+			       MLX5_FLOW_DESTINATION_TYPE_COUNTER)
+				__entry->counter_id =
+					rule->dest_attr.counter_id;
+	    ),
+	    TP_printk("rule=%p fte=%p index=%u sw_action=<%s> [dst] %s\n",
+		      __entry->rule, __entry->fte, __entry->index,
+		      __print_flags(__entry->sw_action, "|", ACTION_FLAGS),
+		      __parse_fs_dst(__entry->destination, __entry->counter_id))
+	    );
+
+TRACE_EVENT(mlx5_fs_del_rule,
+	    TP_PROTO(const struct mlx5_flow_rule *rule),
+	    TP_ARGS(rule),
+	    TP_STRUCT__entry(
+		__field(const struct mlx5_flow_rule *, rule)
+		__field(const struct fs_fte *, fte)
+	    ),
+	    TP_fast_assign(
+			   __entry->rule = rule;
+			   fs_get_obj(__entry->fte, rule->node.parent);
+	    ),
+	    TP_printk("rule=%p fte=%p\n",
+		      __entry->rule, __entry->fte)
+	    );
+#endif
+
+#undef TRACE_INCLUDE_PATH
+#define TRACE_INCLUDE_PATH ./diag
+#undef TRACE_INCLUDE_FILE
+#define TRACE_INCLUDE_FILE fs_tracepoint
+#include <trace/define_trace.h>
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/diag/fw_tracer.c b/drivers/net/ethernet/mellanox/mlx5/core/diag/fw_tracer.c
new file mode 100644
index 000000000..3ba54ffa5
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/diag/fw_tracer.c
@@ -0,0 +1,1154 @@
+/*
+ * Copyright (c) 2018, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+#define CREATE_TRACE_POINTS
+#include "lib/eq.h"
+#include "fw_tracer.h"
+#include "fw_tracer_tracepoint.h"
+
+static int mlx5_query_mtrc_caps(struct mlx5_fw_tracer *tracer)
+{
+	u32 *string_db_base_address_out = tracer->str_db.base_address_out;
+	u32 *string_db_size_out = tracer->str_db.size_out;
+	struct mlx5_core_dev *dev = tracer->dev;
+	u32 out[MLX5_ST_SZ_DW(mtrc_cap)] = {0};
+	u32 in[MLX5_ST_SZ_DW(mtrc_cap)] = {0};
+	void *mtrc_cap_sp;
+	int err, i;
+
+	err = mlx5_core_access_reg(dev, in, sizeof(in), out, sizeof(out),
+				   MLX5_REG_MTRC_CAP, 0, 0);
+	if (err) {
+		mlx5_core_warn(dev, "FWTracer: Error reading tracer caps %d\n",
+			       err);
+		return err;
+	}
+
+	if (!MLX5_GET(mtrc_cap, out, trace_to_memory)) {
+		mlx5_core_dbg(dev, "FWTracer: Device does not support logging traces to memory\n");
+		return -ENOTSUPP;
+	}
+
+	tracer->trc_ver = MLX5_GET(mtrc_cap, out, trc_ver);
+	tracer->str_db.first_string_trace =
+			MLX5_GET(mtrc_cap, out, first_string_trace);
+	tracer->str_db.num_string_trace =
+			MLX5_GET(mtrc_cap, out, num_string_trace);
+	tracer->str_db.num_string_db = MLX5_GET(mtrc_cap, out, num_string_db);
+	tracer->owner = !!MLX5_GET(mtrc_cap, out, trace_owner);
+	tracer->str_db.loaded = false;
+
+	for (i = 0; i < tracer->str_db.num_string_db; i++) {
+		mtrc_cap_sp = MLX5_ADDR_OF(mtrc_cap, out, string_db_param[i]);
+		string_db_base_address_out[i] = MLX5_GET(mtrc_string_db_param,
+							 mtrc_cap_sp,
+							 string_db_base_address);
+		string_db_size_out[i] = MLX5_GET(mtrc_string_db_param,
+						 mtrc_cap_sp, string_db_size);
+	}
+
+	return err;
+}
+
+static int mlx5_set_mtrc_caps_trace_owner(struct mlx5_fw_tracer *tracer,
+					  u32 *out, u32 out_size,
+					  u8 trace_owner)
+{
+	struct mlx5_core_dev *dev = tracer->dev;
+	u32 in[MLX5_ST_SZ_DW(mtrc_cap)] = {0};
+
+	MLX5_SET(mtrc_cap, in, trace_owner, trace_owner);
+
+	return mlx5_core_access_reg(dev, in, sizeof(in), out, out_size,
+				    MLX5_REG_MTRC_CAP, 0, 1);
+}
+
+static int mlx5_fw_tracer_ownership_acquire(struct mlx5_fw_tracer *tracer)
+{
+	struct mlx5_core_dev *dev = tracer->dev;
+	u32 out[MLX5_ST_SZ_DW(mtrc_cap)] = {0};
+	int err;
+
+	err = mlx5_set_mtrc_caps_trace_owner(tracer, out, sizeof(out),
+					     MLX5_FW_TRACER_ACQUIRE_OWNERSHIP);
+	if (err) {
+		mlx5_core_warn(dev, "FWTracer: Acquire tracer ownership failed %d\n",
+			       err);
+		return err;
+	}
+
+	tracer->owner = !!MLX5_GET(mtrc_cap, out, trace_owner);
+
+	if (!tracer->owner)
+		return -EBUSY;
+
+	return 0;
+}
+
+static void mlx5_fw_tracer_ownership_release(struct mlx5_fw_tracer *tracer)
+{
+	u32 out[MLX5_ST_SZ_DW(mtrc_cap)] = {0};
+
+	mlx5_set_mtrc_caps_trace_owner(tracer, out, sizeof(out),
+				       MLX5_FW_TRACER_RELEASE_OWNERSHIP);
+	tracer->owner = false;
+}
+
+static int mlx5_fw_tracer_create_log_buf(struct mlx5_fw_tracer *tracer)
+{
+	struct mlx5_core_dev *dev = tracer->dev;
+	struct device *ddev;
+	dma_addr_t dma;
+	void *buff;
+	gfp_t gfp;
+	int err;
+
+	tracer->buff.size = TRACE_BUFFER_SIZE_BYTE;
+
+	gfp = GFP_KERNEL | __GFP_ZERO;
+	buff = (void *)__get_free_pages(gfp,
+					get_order(tracer->buff.size));
+	if (!buff) {
+		err = -ENOMEM;
+		mlx5_core_warn(dev, "FWTracer: Failed to allocate pages, %d\n", err);
+		return err;
+	}
+	tracer->buff.log_buf = buff;
+
+	ddev = mlx5_core_dma_dev(dev);
+	dma = dma_map_single(ddev, buff, tracer->buff.size, DMA_FROM_DEVICE);
+	if (dma_mapping_error(ddev, dma)) {
+		mlx5_core_warn(dev, "FWTracer: Unable to map DMA: %d\n",
+			       dma_mapping_error(ddev, dma));
+		err = -ENOMEM;
+		goto free_pages;
+	}
+	tracer->buff.dma = dma;
+
+	return 0;
+
+free_pages:
+	free_pages((unsigned long)tracer->buff.log_buf, get_order(tracer->buff.size));
+
+	return err;
+}
+
+static void mlx5_fw_tracer_destroy_log_buf(struct mlx5_fw_tracer *tracer)
+{
+	struct mlx5_core_dev *dev = tracer->dev;
+	struct device *ddev;
+
+	if (!tracer->buff.log_buf)
+		return;
+
+	ddev = mlx5_core_dma_dev(dev);
+	dma_unmap_single(ddev, tracer->buff.dma, tracer->buff.size, DMA_FROM_DEVICE);
+	free_pages((unsigned long)tracer->buff.log_buf, get_order(tracer->buff.size));
+}
+
+static int mlx5_fw_tracer_create_mkey(struct mlx5_fw_tracer *tracer)
+{
+	struct mlx5_core_dev *dev = tracer->dev;
+	int err, inlen, i;
+	__be64 *mtt;
+	void *mkc;
+	u32 *in;
+
+	inlen = MLX5_ST_SZ_BYTES(create_mkey_in) +
+			sizeof(*mtt) * round_up(TRACER_BUFFER_PAGE_NUM, 2);
+
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in)
+		return -ENOMEM;
+
+	MLX5_SET(create_mkey_in, in, translations_octword_actual_size,
+		 DIV_ROUND_UP(TRACER_BUFFER_PAGE_NUM, 2));
+	mtt = (__be64 *)MLX5_ADDR_OF(create_mkey_in, in, klm_pas_mtt);
+	for (i = 0 ; i < TRACER_BUFFER_PAGE_NUM ; i++)
+		mtt[i] = cpu_to_be64(tracer->buff.dma + i * PAGE_SIZE);
+
+	mkc = MLX5_ADDR_OF(create_mkey_in, in, memory_key_mkey_entry);
+	MLX5_SET(mkc, mkc, access_mode_1_0, MLX5_MKC_ACCESS_MODE_MTT);
+	MLX5_SET(mkc, mkc, lr, 1);
+	MLX5_SET(mkc, mkc, lw, 1);
+	MLX5_SET(mkc, mkc, pd, tracer->buff.pdn);
+	MLX5_SET(mkc, mkc, bsf_octword_size, 0);
+	MLX5_SET(mkc, mkc, qpn, 0xffffff);
+	MLX5_SET(mkc, mkc, log_page_size, PAGE_SHIFT);
+	MLX5_SET(mkc, mkc, translations_octword_size,
+		 DIV_ROUND_UP(TRACER_BUFFER_PAGE_NUM, 2));
+	MLX5_SET64(mkc, mkc, start_addr, tracer->buff.dma);
+	MLX5_SET64(mkc, mkc, len, tracer->buff.size);
+	err = mlx5_core_create_mkey(dev, &tracer->buff.mkey, in, inlen);
+	if (err)
+		mlx5_core_warn(dev, "FWTracer: Failed to create mkey, %d\n", err);
+
+	kvfree(in);
+
+	return err;
+}
+
+static void mlx5_fw_tracer_free_strings_db(struct mlx5_fw_tracer *tracer)
+{
+	u32 num_string_db = tracer->str_db.num_string_db;
+	int i;
+
+	for (i = 0; i < num_string_db; i++) {
+		kfree(tracer->str_db.buffer[i]);
+		tracer->str_db.buffer[i] = NULL;
+	}
+}
+
+static int mlx5_fw_tracer_allocate_strings_db(struct mlx5_fw_tracer *tracer)
+{
+	u32 *string_db_size_out = tracer->str_db.size_out;
+	u32 num_string_db = tracer->str_db.num_string_db;
+	int i;
+
+	for (i = 0; i < num_string_db; i++) {
+		tracer->str_db.buffer[i] = kzalloc(string_db_size_out[i], GFP_KERNEL);
+		if (!tracer->str_db.buffer[i])
+			goto free_strings_db;
+	}
+
+	return 0;
+
+free_strings_db:
+	mlx5_fw_tracer_free_strings_db(tracer);
+	return -ENOMEM;
+}
+
+static void
+mlx5_fw_tracer_init_saved_traces_array(struct mlx5_fw_tracer *tracer)
+{
+	tracer->st_arr.saved_traces_index = 0;
+	mutex_init(&tracer->st_arr.lock);
+}
+
+static void
+mlx5_fw_tracer_clean_saved_traces_array(struct mlx5_fw_tracer *tracer)
+{
+	mutex_destroy(&tracer->st_arr.lock);
+}
+
+static void mlx5_tracer_read_strings_db(struct work_struct *work)
+{
+	struct mlx5_fw_tracer *tracer = container_of(work, struct mlx5_fw_tracer,
+						     read_fw_strings_work);
+	u32 num_of_reads, num_string_db = tracer->str_db.num_string_db;
+	struct mlx5_core_dev *dev = tracer->dev;
+	u32 in[MLX5_ST_SZ_DW(mtrc_cap)] = {0};
+	u32 leftovers, offset;
+	int err = 0, i, j;
+	u32 *out, outlen;
+	void *out_value;
+
+	outlen = MLX5_ST_SZ_BYTES(mtrc_stdb) + STRINGS_DB_READ_SIZE_BYTES;
+	out = kzalloc(outlen, GFP_KERNEL);
+	if (!out) {
+		err = -ENOMEM;
+		goto out;
+	}
+
+	for (i = 0; i < num_string_db; i++) {
+		offset = 0;
+		MLX5_SET(mtrc_stdb, in, string_db_index, i);
+		num_of_reads = tracer->str_db.size_out[i] /
+				STRINGS_DB_READ_SIZE_BYTES;
+		leftovers = (tracer->str_db.size_out[i] %
+				STRINGS_DB_READ_SIZE_BYTES) /
+					STRINGS_DB_LEFTOVER_SIZE_BYTES;
+
+		MLX5_SET(mtrc_stdb, in, read_size, STRINGS_DB_READ_SIZE_BYTES);
+		for (j = 0; j < num_of_reads; j++) {
+			MLX5_SET(mtrc_stdb, in, start_offset, offset);
+
+			err = mlx5_core_access_reg(dev, in, sizeof(in), out,
+						   outlen, MLX5_REG_MTRC_STDB,
+						   0, 1);
+			if (err) {
+				mlx5_core_dbg(dev, "FWTracer: Failed to read strings DB %d\n",
+					      err);
+				goto out_free;
+			}
+
+			out_value = MLX5_ADDR_OF(mtrc_stdb, out, string_db_data);
+			memcpy(tracer->str_db.buffer[i] + offset, out_value,
+			       STRINGS_DB_READ_SIZE_BYTES);
+			offset += STRINGS_DB_READ_SIZE_BYTES;
+		}
+
+		/* Strings database is aligned to 64, need to read leftovers*/
+		MLX5_SET(mtrc_stdb, in, read_size,
+			 STRINGS_DB_LEFTOVER_SIZE_BYTES);
+		for (j = 0; j < leftovers; j++) {
+			MLX5_SET(mtrc_stdb, in, start_offset, offset);
+
+			err = mlx5_core_access_reg(dev, in, sizeof(in), out,
+						   outlen, MLX5_REG_MTRC_STDB,
+						   0, 1);
+			if (err) {
+				mlx5_core_dbg(dev, "FWTracer: Failed to read strings DB %d\n",
+					      err);
+				goto out_free;
+			}
+
+			out_value = MLX5_ADDR_OF(mtrc_stdb, out, string_db_data);
+			memcpy(tracer->str_db.buffer[i] + offset, out_value,
+			       STRINGS_DB_LEFTOVER_SIZE_BYTES);
+			offset += STRINGS_DB_LEFTOVER_SIZE_BYTES;
+		}
+	}
+
+	tracer->str_db.loaded = true;
+
+out_free:
+	kfree(out);
+out:
+	return;
+}
+
+static void mlx5_fw_tracer_arm(struct mlx5_core_dev *dev)
+{
+	u32 out[MLX5_ST_SZ_DW(mtrc_ctrl)] = {0};
+	u32 in[MLX5_ST_SZ_DW(mtrc_ctrl)] = {0};
+	int err;
+
+	MLX5_SET(mtrc_ctrl, in, arm_event, 1);
+
+	err = mlx5_core_access_reg(dev, in, sizeof(in), out, sizeof(out),
+				   MLX5_REG_MTRC_CTRL, 0, 1);
+	if (err)
+		mlx5_core_warn(dev, "FWTracer: Failed to arm tracer event %d\n", err);
+}
+
+static const char *VAL_PARM		= "%llx";
+static const char *REPLACE_64_VAL_PARM	= "%x%x";
+static const char *PARAM_CHAR		= "%";
+
+static int mlx5_tracer_message_hash(u32 message_id)
+{
+	return jhash_1word(message_id, 0) & (MESSAGE_HASH_SIZE - 1);
+}
+
+static struct tracer_string_format *mlx5_tracer_message_insert(struct mlx5_fw_tracer *tracer,
+							       struct tracer_event *tracer_event)
+{
+	struct hlist_head *head =
+		&tracer->hash[mlx5_tracer_message_hash(tracer_event->string_event.tmsn)];
+	struct tracer_string_format *cur_string;
+
+	cur_string = kzalloc(sizeof(*cur_string), GFP_KERNEL);
+	if (!cur_string)
+		return NULL;
+
+	hlist_add_head(&cur_string->hlist, head);
+
+	return cur_string;
+}
+
+static struct tracer_string_format *mlx5_tracer_get_string(struct mlx5_fw_tracer *tracer,
+							   struct tracer_event *tracer_event)
+{
+	struct tracer_string_format *cur_string;
+	u32 str_ptr, offset;
+	int i;
+
+	str_ptr = tracer_event->string_event.string_param;
+
+	for (i = 0; i < tracer->str_db.num_string_db; i++) {
+		if (str_ptr > tracer->str_db.base_address_out[i] &&
+		    str_ptr < tracer->str_db.base_address_out[i] +
+		    tracer->str_db.size_out[i]) {
+			offset = str_ptr - tracer->str_db.base_address_out[i];
+			/* add it to the hash */
+			cur_string = mlx5_tracer_message_insert(tracer, tracer_event);
+			if (!cur_string)
+				return NULL;
+			cur_string->string = (char *)(tracer->str_db.buffer[i] +
+							offset);
+			return cur_string;
+		}
+	}
+
+	return NULL;
+}
+
+static void mlx5_tracer_clean_message(struct tracer_string_format *str_frmt)
+{
+	hlist_del(&str_frmt->hlist);
+	kfree(str_frmt);
+}
+
+static int mlx5_tracer_get_num_of_params(char *str)
+{
+	char *substr, *pstr = str;
+	int num_of_params = 0;
+
+	/* replace %llx with %x%x */
+	substr = strstr(pstr, VAL_PARM);
+	while (substr) {
+		memcpy(substr, REPLACE_64_VAL_PARM, 4);
+		pstr = substr;
+		substr = strstr(pstr, VAL_PARM);
+	}
+
+	/* count all the % characters */
+	substr = strstr(str, PARAM_CHAR);
+	while (substr) {
+		num_of_params += 1;
+		str = substr + 1;
+		substr = strstr(str, PARAM_CHAR);
+	}
+
+	return num_of_params;
+}
+
+static struct tracer_string_format *mlx5_tracer_message_find(struct hlist_head *head,
+							     u8 event_id, u32 tmsn)
+{
+	struct tracer_string_format *message;
+
+	hlist_for_each_entry(message, head, hlist)
+		if (message->event_id == event_id && message->tmsn == tmsn)
+			return message;
+
+	return NULL;
+}
+
+static struct tracer_string_format *mlx5_tracer_message_get(struct mlx5_fw_tracer *tracer,
+							    struct tracer_event *tracer_event)
+{
+	struct hlist_head *head =
+		&tracer->hash[mlx5_tracer_message_hash(tracer_event->string_event.tmsn)];
+
+	return mlx5_tracer_message_find(head, tracer_event->event_id, tracer_event->string_event.tmsn);
+}
+
+static void poll_trace(struct mlx5_fw_tracer *tracer,
+		       struct tracer_event *tracer_event, u64 *trace)
+{
+	u32 timestamp_low, timestamp_mid, timestamp_high, urts;
+
+	tracer_event->event_id = MLX5_GET(tracer_event, trace, event_id);
+	tracer_event->lost_event = MLX5_GET(tracer_event, trace, lost);
+
+	switch (tracer_event->event_id) {
+	case TRACER_EVENT_TYPE_TIMESTAMP:
+		tracer_event->type = TRACER_EVENT_TYPE_TIMESTAMP;
+		urts = MLX5_GET(tracer_timestamp_event, trace, urts);
+		if (tracer->trc_ver == 0)
+			tracer_event->timestamp_event.unreliable = !!(urts >> 2);
+		else
+			tracer_event->timestamp_event.unreliable = !!(urts & 1);
+
+		timestamp_low = MLX5_GET(tracer_timestamp_event,
+					 trace, timestamp7_0);
+		timestamp_mid = MLX5_GET(tracer_timestamp_event,
+					 trace, timestamp39_8);
+		timestamp_high = MLX5_GET(tracer_timestamp_event,
+					  trace, timestamp52_40);
+
+		tracer_event->timestamp_event.timestamp =
+				((u64)timestamp_high << 40) |
+				((u64)timestamp_mid << 8) |
+				(u64)timestamp_low;
+		break;
+	default:
+		if (tracer_event->event_id >= tracer->str_db.first_string_trace &&
+		    tracer_event->event_id <= tracer->str_db.first_string_trace +
+					      tracer->str_db.num_string_trace) {
+			tracer_event->type = TRACER_EVENT_TYPE_STRING;
+			tracer_event->string_event.timestamp =
+				MLX5_GET(tracer_string_event, trace, timestamp);
+			tracer_event->string_event.string_param =
+				MLX5_GET(tracer_string_event, trace, string_param);
+			tracer_event->string_event.tmsn =
+				MLX5_GET(tracer_string_event, trace, tmsn);
+			tracer_event->string_event.tdsn =
+				MLX5_GET(tracer_string_event, trace, tdsn);
+		} else {
+			tracer_event->type = TRACER_EVENT_TYPE_UNRECOGNIZED;
+		}
+		break;
+	}
+}
+
+static u64 get_block_timestamp(struct mlx5_fw_tracer *tracer, u64 *ts_event)
+{
+	struct tracer_event tracer_event;
+	u8 event_id;
+
+	event_id = MLX5_GET(tracer_event, ts_event, event_id);
+
+	if (event_id == TRACER_EVENT_TYPE_TIMESTAMP)
+		poll_trace(tracer, &tracer_event, ts_event);
+	else
+		tracer_event.timestamp_event.timestamp = 0;
+
+	return tracer_event.timestamp_event.timestamp;
+}
+
+static void mlx5_fw_tracer_clean_print_hash(struct mlx5_fw_tracer *tracer)
+{
+	struct tracer_string_format *str_frmt;
+	struct hlist_node *n;
+	int i;
+
+	for (i = 0; i < MESSAGE_HASH_SIZE; i++) {
+		hlist_for_each_entry_safe(str_frmt, n, &tracer->hash[i], hlist)
+			mlx5_tracer_clean_message(str_frmt);
+	}
+}
+
+static void mlx5_fw_tracer_clean_ready_list(struct mlx5_fw_tracer *tracer)
+{
+	struct tracer_string_format *str_frmt, *tmp_str;
+
+	list_for_each_entry_safe(str_frmt, tmp_str, &tracer->ready_strings_list,
+				 list)
+		list_del(&str_frmt->list);
+}
+
+static void mlx5_fw_tracer_save_trace(struct mlx5_fw_tracer *tracer,
+				      u64 timestamp, bool lost,
+				      u8 event_id, char *msg)
+{
+	struct mlx5_fw_trace_data *trace_data;
+
+	mutex_lock(&tracer->st_arr.lock);
+	trace_data = &tracer->st_arr.straces[tracer->st_arr.saved_traces_index];
+	trace_data->timestamp = timestamp;
+	trace_data->lost = lost;
+	trace_data->event_id = event_id;
+	strscpy_pad(trace_data->msg, msg, TRACE_STR_MSG);
+
+	tracer->st_arr.saved_traces_index =
+		(tracer->st_arr.saved_traces_index + 1) & (SAVED_TRACES_NUM - 1);
+	mutex_unlock(&tracer->st_arr.lock);
+}
+
+static noinline
+void mlx5_tracer_print_trace(struct tracer_string_format *str_frmt,
+			     struct mlx5_core_dev *dev,
+			     u64 trace_timestamp)
+{
+	char	tmp[512];
+
+	snprintf(tmp, sizeof(tmp), str_frmt->string,
+		 str_frmt->params[0],
+		 str_frmt->params[1],
+		 str_frmt->params[2],
+		 str_frmt->params[3],
+		 str_frmt->params[4],
+		 str_frmt->params[5],
+		 str_frmt->params[6]);
+
+	trace_mlx5_fw(dev->tracer, trace_timestamp, str_frmt->lost,
+		      str_frmt->event_id, tmp);
+
+	mlx5_fw_tracer_save_trace(dev->tracer, trace_timestamp,
+				  str_frmt->lost, str_frmt->event_id, tmp);
+
+	/* remove it from hash */
+	mlx5_tracer_clean_message(str_frmt);
+}
+
+static int mlx5_tracer_handle_string_trace(struct mlx5_fw_tracer *tracer,
+					   struct tracer_event *tracer_event)
+{
+	struct tracer_string_format *cur_string;
+
+	if (tracer_event->string_event.tdsn == 0) {
+		cur_string = mlx5_tracer_get_string(tracer, tracer_event);
+		if (!cur_string)
+			return -1;
+
+		cur_string->num_of_params = mlx5_tracer_get_num_of_params(cur_string->string);
+		cur_string->last_param_num = 0;
+		cur_string->event_id = tracer_event->event_id;
+		cur_string->tmsn = tracer_event->string_event.tmsn;
+		cur_string->timestamp = tracer_event->string_event.timestamp;
+		cur_string->lost = tracer_event->lost_event;
+		if (cur_string->num_of_params == 0) /* trace with no params */
+			list_add_tail(&cur_string->list, &tracer->ready_strings_list);
+	} else {
+		cur_string = mlx5_tracer_message_get(tracer, tracer_event);
+		if (!cur_string) {
+			pr_debug("%s Got string event for unknown string tmsn: %d\n",
+				 __func__, tracer_event->string_event.tmsn);
+			return -1;
+		}
+		cur_string->last_param_num += 1;
+		if (cur_string->last_param_num > TRACER_MAX_PARAMS) {
+			pr_debug("%s Number of params exceeds the max (%d)\n",
+				 __func__, TRACER_MAX_PARAMS);
+			list_add_tail(&cur_string->list, &tracer->ready_strings_list);
+			return 0;
+		}
+		/* keep the new parameter */
+		cur_string->params[cur_string->last_param_num - 1] =
+			tracer_event->string_event.string_param;
+		if (cur_string->last_param_num == cur_string->num_of_params)
+			list_add_tail(&cur_string->list, &tracer->ready_strings_list);
+	}
+
+	return 0;
+}
+
+static void mlx5_tracer_handle_timestamp_trace(struct mlx5_fw_tracer *tracer,
+					       struct tracer_event *tracer_event)
+{
+	struct tracer_timestamp_event timestamp_event =
+						tracer_event->timestamp_event;
+	struct tracer_string_format *str_frmt, *tmp_str;
+	struct mlx5_core_dev *dev = tracer->dev;
+	u64 trace_timestamp;
+
+	list_for_each_entry_safe(str_frmt, tmp_str, &tracer->ready_strings_list, list) {
+		list_del(&str_frmt->list);
+		if (str_frmt->timestamp < (timestamp_event.timestamp & MASK_6_0))
+			trace_timestamp = (timestamp_event.timestamp & MASK_52_7) |
+					  (str_frmt->timestamp & MASK_6_0);
+		else
+			trace_timestamp = ((timestamp_event.timestamp - 1) & MASK_52_7) |
+					  (str_frmt->timestamp & MASK_6_0);
+
+		mlx5_tracer_print_trace(str_frmt, dev, trace_timestamp);
+	}
+}
+
+static int mlx5_tracer_handle_trace(struct mlx5_fw_tracer *tracer,
+				    struct tracer_event *tracer_event)
+{
+	if (tracer_event->type == TRACER_EVENT_TYPE_STRING) {
+		mlx5_tracer_handle_string_trace(tracer, tracer_event);
+	} else if (tracer_event->type == TRACER_EVENT_TYPE_TIMESTAMP) {
+		if (!tracer_event->timestamp_event.unreliable)
+			mlx5_tracer_handle_timestamp_trace(tracer, tracer_event);
+	} else {
+		pr_debug("%s Got unrecognised type %d for parsing, exiting..\n",
+			 __func__, tracer_event->type);
+	}
+	return 0;
+}
+
+static void mlx5_fw_tracer_handle_traces(struct work_struct *work)
+{
+	struct mlx5_fw_tracer *tracer =
+			container_of(work, struct mlx5_fw_tracer, handle_traces_work);
+	u64 block_timestamp, last_block_timestamp, tmp_trace_block[TRACES_PER_BLOCK];
+	u32 block_count, start_offset, prev_start_offset, prev_consumer_index;
+	u32 trace_event_size = MLX5_ST_SZ_BYTES(tracer_event);
+	struct mlx5_core_dev *dev = tracer->dev;
+	struct tracer_event tracer_event;
+	int i;
+
+	mlx5_core_dbg(dev, "FWTracer: Handle Trace event, owner=(%d)\n", tracer->owner);
+	if (!tracer->owner)
+		return;
+
+	if (unlikely(!tracer->str_db.loaded))
+		goto arm;
+
+	block_count = tracer->buff.size / TRACER_BLOCK_SIZE_BYTE;
+	start_offset = tracer->buff.consumer_index * TRACER_BLOCK_SIZE_BYTE;
+
+	/* Copy the block to local buffer to avoid HW override while being processed */
+	memcpy(tmp_trace_block, tracer->buff.log_buf + start_offset,
+	       TRACER_BLOCK_SIZE_BYTE);
+
+	block_timestamp =
+		get_block_timestamp(tracer, &tmp_trace_block[TRACES_PER_BLOCK - 1]);
+
+	while (block_timestamp > tracer->last_timestamp) {
+		/* Check block override if it's not the first block */
+		if (tracer->last_timestamp) {
+			u64 *ts_event;
+			/* To avoid block override be the HW in case of buffer
+			 * wraparound, the time stamp of the previous block
+			 * should be compared to the last timestamp handled
+			 * by the driver.
+			 */
+			prev_consumer_index =
+				(tracer->buff.consumer_index - 1) & (block_count - 1);
+			prev_start_offset = prev_consumer_index * TRACER_BLOCK_SIZE_BYTE;
+
+			ts_event = tracer->buff.log_buf + prev_start_offset +
+				   (TRACES_PER_BLOCK - 1) * trace_event_size;
+			last_block_timestamp = get_block_timestamp(tracer, ts_event);
+			/* If previous timestamp different from last stored
+			 * timestamp then there is a good chance that the
+			 * current buffer is overwritten and therefore should
+			 * not be parsed.
+			 */
+			if (tracer->last_timestamp != last_block_timestamp) {
+				mlx5_core_warn(dev, "FWTracer: Events were lost\n");
+				tracer->last_timestamp = block_timestamp;
+				tracer->buff.consumer_index =
+					(tracer->buff.consumer_index + 1) & (block_count - 1);
+				break;
+			}
+		}
+
+		/* Parse events */
+		for (i = 0; i < TRACES_PER_BLOCK ; i++) {
+			poll_trace(tracer, &tracer_event, &tmp_trace_block[i]);
+			mlx5_tracer_handle_trace(tracer, &tracer_event);
+		}
+
+		tracer->buff.consumer_index =
+			(tracer->buff.consumer_index + 1) & (block_count - 1);
+
+		tracer->last_timestamp = block_timestamp;
+		start_offset = tracer->buff.consumer_index * TRACER_BLOCK_SIZE_BYTE;
+		memcpy(tmp_trace_block, tracer->buff.log_buf + start_offset,
+		       TRACER_BLOCK_SIZE_BYTE);
+		block_timestamp = get_block_timestamp(tracer,
+						      &tmp_trace_block[TRACES_PER_BLOCK - 1]);
+	}
+
+arm:
+	mlx5_fw_tracer_arm(dev);
+}
+
+static int mlx5_fw_tracer_set_mtrc_conf(struct mlx5_fw_tracer *tracer)
+{
+	struct mlx5_core_dev *dev = tracer->dev;
+	u32 out[MLX5_ST_SZ_DW(mtrc_conf)] = {0};
+	u32 in[MLX5_ST_SZ_DW(mtrc_conf)] = {0};
+	int err;
+
+	MLX5_SET(mtrc_conf, in, trace_mode, TRACE_TO_MEMORY);
+	MLX5_SET(mtrc_conf, in, log_trace_buffer_size,
+		 ilog2(TRACER_BUFFER_PAGE_NUM));
+	MLX5_SET(mtrc_conf, in, trace_mkey, tracer->buff.mkey);
+
+	err = mlx5_core_access_reg(dev, in, sizeof(in), out, sizeof(out),
+				   MLX5_REG_MTRC_CONF, 0, 1);
+	if (err)
+		mlx5_core_warn(dev, "FWTracer: Failed to set tracer configurations %d\n", err);
+
+	tracer->buff.consumer_index = 0;
+	return err;
+}
+
+static int mlx5_fw_tracer_set_mtrc_ctrl(struct mlx5_fw_tracer *tracer, u8 status, u8 arm)
+{
+	struct mlx5_core_dev *dev = tracer->dev;
+	u32 out[MLX5_ST_SZ_DW(mtrc_ctrl)] = {0};
+	u32 in[MLX5_ST_SZ_DW(mtrc_ctrl)] = {0};
+	int err;
+
+	MLX5_SET(mtrc_ctrl, in, modify_field_select, TRACE_STATUS);
+	MLX5_SET(mtrc_ctrl, in, trace_status, status);
+	MLX5_SET(mtrc_ctrl, in, arm_event, arm);
+
+	err = mlx5_core_access_reg(dev, in, sizeof(in), out, sizeof(out),
+				   MLX5_REG_MTRC_CTRL, 0, 1);
+
+	if (!err && status)
+		tracer->last_timestamp = 0;
+
+	return err;
+}
+
+static int mlx5_fw_tracer_start(struct mlx5_fw_tracer *tracer)
+{
+	struct mlx5_core_dev *dev = tracer->dev;
+	int err;
+
+	err = mlx5_fw_tracer_ownership_acquire(tracer);
+	if (err) {
+		mlx5_core_dbg(dev, "FWTracer: Ownership was not granted %d\n", err);
+		/* Don't fail since ownership can be acquired on a later FW event */
+		return 0;
+	}
+
+	err = mlx5_fw_tracer_set_mtrc_conf(tracer);
+	if (err) {
+		mlx5_core_warn(dev, "FWTracer: Failed to set tracer configuration %d\n", err);
+		goto release_ownership;
+	}
+
+	/* enable tracer & trace events */
+	err = mlx5_fw_tracer_set_mtrc_ctrl(tracer, 1, 1);
+	if (err) {
+		mlx5_core_warn(dev, "FWTracer: Failed to enable tracer %d\n", err);
+		goto release_ownership;
+	}
+
+	mlx5_core_dbg(dev, "FWTracer: Ownership granted and active\n");
+	return 0;
+
+release_ownership:
+	mlx5_fw_tracer_ownership_release(tracer);
+	return err;
+}
+
+static void mlx5_fw_tracer_ownership_change(struct work_struct *work)
+{
+	struct mlx5_fw_tracer *tracer =
+		container_of(work, struct mlx5_fw_tracer, ownership_change_work);
+
+	mlx5_core_dbg(tracer->dev, "FWTracer: ownership changed, current=(%d)\n", tracer->owner);
+	if (tracer->owner) {
+		mlx5_fw_tracer_ownership_acquire(tracer);
+		return;
+	}
+
+	mlx5_fw_tracer_start(tracer);
+}
+
+static int mlx5_fw_tracer_set_core_dump_reg(struct mlx5_core_dev *dev,
+					    u32 *in, int size_in)
+{
+	u32 out[MLX5_ST_SZ_DW(core_dump_reg)] = {};
+
+	if (!MLX5_CAP_DEBUG(dev, core_dump_general) &&
+	    !MLX5_CAP_DEBUG(dev, core_dump_qp))
+		return -EOPNOTSUPP;
+
+	return mlx5_core_access_reg(dev, in, size_in, out, sizeof(out),
+				    MLX5_REG_CORE_DUMP, 0, 1);
+}
+
+int mlx5_fw_tracer_trigger_core_dump_general(struct mlx5_core_dev *dev)
+{
+	struct mlx5_fw_tracer *tracer = dev->tracer;
+	u32 in[MLX5_ST_SZ_DW(core_dump_reg)] = {};
+	int err;
+
+	if (!MLX5_CAP_DEBUG(dev, core_dump_general) || !tracer)
+		return -EOPNOTSUPP;
+	if (!tracer->owner)
+		return -EPERM;
+
+	MLX5_SET(core_dump_reg, in, core_dump_type, 0x0);
+
+	err =  mlx5_fw_tracer_set_core_dump_reg(dev, in, sizeof(in));
+	if (err)
+		return err;
+	queue_work(tracer->work_queue, &tracer->handle_traces_work);
+	flush_workqueue(tracer->work_queue);
+	return 0;
+}
+
+static int
+mlx5_devlink_fmsg_fill_trace(struct devlink_fmsg *fmsg,
+			     struct mlx5_fw_trace_data *trace_data)
+{
+	int err;
+
+	err = devlink_fmsg_obj_nest_start(fmsg);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u64_pair_put(fmsg, "timestamp", trace_data->timestamp);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_bool_pair_put(fmsg, "lost", trace_data->lost);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u8_pair_put(fmsg, "event_id", trace_data->event_id);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_string_pair_put(fmsg, "msg", trace_data->msg);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_obj_nest_end(fmsg);
+	if (err)
+		return err;
+	return 0;
+}
+
+int mlx5_fw_tracer_get_saved_traces_objects(struct mlx5_fw_tracer *tracer,
+					    struct devlink_fmsg *fmsg)
+{
+	struct mlx5_fw_trace_data *straces = tracer->st_arr.straces;
+	u32 index, start_index, end_index;
+	u32 saved_traces_index;
+	int err;
+
+	if (!straces[0].timestamp)
+		return -ENOMSG;
+
+	mutex_lock(&tracer->st_arr.lock);
+	saved_traces_index = tracer->st_arr.saved_traces_index;
+	if (straces[saved_traces_index].timestamp)
+		start_index = saved_traces_index;
+	else
+		start_index = 0;
+	end_index = (saved_traces_index - 1) & (SAVED_TRACES_NUM - 1);
+
+	err = devlink_fmsg_arr_pair_nest_start(fmsg, "dump fw traces");
+	if (err)
+		goto unlock;
+	index = start_index;
+	while (index != end_index) {
+		err = mlx5_devlink_fmsg_fill_trace(fmsg, &straces[index]);
+		if (err)
+			goto unlock;
+
+		index = (index + 1) & (SAVED_TRACES_NUM - 1);
+	}
+
+	err = devlink_fmsg_arr_pair_nest_end(fmsg);
+unlock:
+	mutex_unlock(&tracer->st_arr.lock);
+	return err;
+}
+
+/* Create software resources (Buffers, etc ..) */
+struct mlx5_fw_tracer *mlx5_fw_tracer_create(struct mlx5_core_dev *dev)
+{
+	struct mlx5_fw_tracer *tracer = NULL;
+	int err;
+
+	if (!MLX5_CAP_MCAM_REG(dev, tracer_registers)) {
+		mlx5_core_dbg(dev, "FWTracer: Tracer capability not present\n");
+		return NULL;
+	}
+
+	tracer = kvzalloc(sizeof(*tracer), GFP_KERNEL);
+	if (!tracer)
+		return ERR_PTR(-ENOMEM);
+
+	tracer->work_queue = create_singlethread_workqueue("mlx5_fw_tracer");
+	if (!tracer->work_queue) {
+		err = -ENOMEM;
+		goto free_tracer;
+	}
+
+	tracer->dev = dev;
+
+	INIT_LIST_HEAD(&tracer->ready_strings_list);
+	INIT_WORK(&tracer->ownership_change_work, mlx5_fw_tracer_ownership_change);
+	INIT_WORK(&tracer->read_fw_strings_work, mlx5_tracer_read_strings_db);
+	INIT_WORK(&tracer->handle_traces_work, mlx5_fw_tracer_handle_traces);
+
+
+	err = mlx5_query_mtrc_caps(tracer);
+	if (err) {
+		mlx5_core_dbg(dev, "FWTracer: Failed to query capabilities %d\n", err);
+		goto destroy_workqueue;
+	}
+
+	err = mlx5_fw_tracer_create_log_buf(tracer);
+	if (err) {
+		mlx5_core_warn(dev, "FWTracer: Create log buffer failed %d\n", err);
+		goto destroy_workqueue;
+	}
+
+	err = mlx5_fw_tracer_allocate_strings_db(tracer);
+	if (err) {
+		mlx5_core_warn(dev, "FWTracer: Allocate strings database failed %d\n", err);
+		goto free_log_buf;
+	}
+
+	mlx5_fw_tracer_init_saved_traces_array(tracer);
+	mlx5_core_dbg(dev, "FWTracer: Tracer created\n");
+
+	return tracer;
+
+free_log_buf:
+	mlx5_fw_tracer_destroy_log_buf(tracer);
+destroy_workqueue:
+	tracer->dev = NULL;
+	destroy_workqueue(tracer->work_queue);
+free_tracer:
+	kvfree(tracer);
+	return ERR_PTR(err);
+}
+
+static int fw_tracer_event(struct notifier_block *nb, unsigned long action, void *data);
+
+/* Create HW resources + start tracer */
+int mlx5_fw_tracer_init(struct mlx5_fw_tracer *tracer)
+{
+	struct mlx5_core_dev *dev;
+	int err;
+
+	if (IS_ERR_OR_NULL(tracer))
+		return 0;
+
+	dev = tracer->dev;
+
+	if (!tracer->str_db.loaded)
+		queue_work(tracer->work_queue, &tracer->read_fw_strings_work);
+
+	err = mlx5_core_alloc_pd(dev, &tracer->buff.pdn);
+	if (err) {
+		mlx5_core_warn(dev, "FWTracer: Failed to allocate PD %d\n", err);
+		goto err_cancel_work;
+	}
+
+	err = mlx5_fw_tracer_create_mkey(tracer);
+	if (err) {
+		mlx5_core_warn(dev, "FWTracer: Failed to create mkey %d\n", err);
+		goto err_dealloc_pd;
+	}
+
+	MLX5_NB_INIT(&tracer->nb, fw_tracer_event, DEVICE_TRACER);
+	mlx5_eq_notifier_register(dev, &tracer->nb);
+
+	err = mlx5_fw_tracer_start(tracer);
+	if (err) {
+		mlx5_core_warn(dev, "FWTracer: Failed to start tracer %d\n", err);
+		goto err_notifier_unregister;
+	}
+	return 0;
+
+err_notifier_unregister:
+	mlx5_eq_notifier_unregister(dev, &tracer->nb);
+	mlx5_core_destroy_mkey(dev, tracer->buff.mkey);
+err_dealloc_pd:
+	mlx5_core_dealloc_pd(dev, tracer->buff.pdn);
+err_cancel_work:
+	cancel_work_sync(&tracer->read_fw_strings_work);
+	return err;
+}
+
+/* Stop tracer + Cleanup HW resources */
+void mlx5_fw_tracer_cleanup(struct mlx5_fw_tracer *tracer)
+{
+	if (IS_ERR_OR_NULL(tracer))
+		return;
+
+	mlx5_core_dbg(tracer->dev, "FWTracer: Cleanup, is owner ? (%d)\n",
+		      tracer->owner);
+	mlx5_eq_notifier_unregister(tracer->dev, &tracer->nb);
+	cancel_work_sync(&tracer->ownership_change_work);
+	cancel_work_sync(&tracer->handle_traces_work);
+
+	if (tracer->owner)
+		mlx5_fw_tracer_ownership_release(tracer);
+
+	mlx5_core_destroy_mkey(tracer->dev, tracer->buff.mkey);
+	mlx5_core_dealloc_pd(tracer->dev, tracer->buff.pdn);
+}
+
+/* Free software resources (Buffers, etc ..) */
+void mlx5_fw_tracer_destroy(struct mlx5_fw_tracer *tracer)
+{
+	if (IS_ERR_OR_NULL(tracer))
+		return;
+
+	mlx5_core_dbg(tracer->dev, "FWTracer: Destroy\n");
+
+	cancel_work_sync(&tracer->read_fw_strings_work);
+	mlx5_fw_tracer_clean_ready_list(tracer);
+	mlx5_fw_tracer_clean_print_hash(tracer);
+	mlx5_fw_tracer_clean_saved_traces_array(tracer);
+	mlx5_fw_tracer_free_strings_db(tracer);
+	mlx5_fw_tracer_destroy_log_buf(tracer);
+	destroy_workqueue(tracer->work_queue);
+	kvfree(tracer);
+}
+
+static int mlx5_fw_tracer_recreate_strings_db(struct mlx5_fw_tracer *tracer)
+{
+	struct mlx5_core_dev *dev;
+	int err;
+
+	cancel_work_sync(&tracer->read_fw_strings_work);
+	mlx5_fw_tracer_clean_ready_list(tracer);
+	mlx5_fw_tracer_clean_print_hash(tracer);
+	mlx5_fw_tracer_clean_saved_traces_array(tracer);
+	mlx5_fw_tracer_free_strings_db(tracer);
+
+	dev = tracer->dev;
+	err = mlx5_query_mtrc_caps(tracer);
+	if (err) {
+		mlx5_core_dbg(dev, "FWTracer: Failed to query capabilities %d\n", err);
+		return err;
+	}
+
+	err = mlx5_fw_tracer_allocate_strings_db(tracer);
+	if (err) {
+		mlx5_core_warn(dev, "FWTracer: Allocate strings DB failed %d\n", err);
+		return err;
+	}
+	mlx5_fw_tracer_init_saved_traces_array(tracer);
+
+	return 0;
+}
+
+int mlx5_fw_tracer_reload(struct mlx5_fw_tracer *tracer)
+{
+	struct mlx5_core_dev *dev;
+	int err;
+
+	if (IS_ERR_OR_NULL(tracer))
+		return 0;
+
+	dev = tracer->dev;
+	mlx5_fw_tracer_cleanup(tracer);
+	err = mlx5_fw_tracer_recreate_strings_db(tracer);
+	if (err) {
+		mlx5_core_warn(dev, "Failed to recreate FW tracer strings DB\n");
+		return err;
+	}
+	err = mlx5_fw_tracer_init(tracer);
+	if (err) {
+		mlx5_core_warn(dev, "Failed to re-initialize FW tracer\n");
+		return err;
+	}
+
+	return 0;
+}
+
+static int fw_tracer_event(struct notifier_block *nb, unsigned long action, void *data)
+{
+	struct mlx5_fw_tracer *tracer = mlx5_nb_cof(nb, struct mlx5_fw_tracer, nb);
+	struct mlx5_core_dev *dev = tracer->dev;
+	struct mlx5_eqe *eqe = data;
+
+	switch (eqe->sub_type) {
+	case MLX5_TRACER_SUBTYPE_OWNERSHIP_CHANGE:
+		queue_work(tracer->work_queue, &tracer->ownership_change_work);
+		break;
+	case MLX5_TRACER_SUBTYPE_TRACES_AVAILABLE:
+		queue_work(tracer->work_queue, &tracer->handle_traces_work);
+		break;
+	default:
+		mlx5_core_dbg(dev, "FWTracer: Event with unrecognized subtype: sub_type %d\n",
+			      eqe->sub_type);
+	}
+
+	return NOTIFY_OK;
+}
+
+EXPORT_TRACEPOINT_SYMBOL(mlx5_fw);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/diag/fw_tracer.h b/drivers/net/ethernet/mellanox/mlx5/core/diag/fw_tracer.h
new file mode 100644
index 000000000..4762b55b0
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/diag/fw_tracer.h
@@ -0,0 +1,196 @@
+/*
+ * Copyright (c) 2018, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef __LIB_TRACER_H__
+#define __LIB_TRACER_H__
+
+#include <linux/mlx5/driver.h>
+#include "mlx5_core.h"
+
+#define STRINGS_DB_SECTIONS_NUM 8
+#define STRINGS_DB_READ_SIZE_BYTES 256
+#define STRINGS_DB_LEFTOVER_SIZE_BYTES 64
+#define TRACER_BUFFER_PAGE_NUM 64
+#define TRACER_BUFFER_CHUNK 4096
+#define TRACE_BUFFER_SIZE_BYTE (TRACER_BUFFER_PAGE_NUM * TRACER_BUFFER_CHUNK)
+
+#define TRACER_BLOCK_SIZE_BYTE 256
+#define TRACES_PER_BLOCK 32
+
+#define TRACE_STR_MSG 256
+#define SAVED_TRACES_NUM 8192
+
+#define TRACER_MAX_PARAMS 7
+#define MESSAGE_HASH_BITS 6
+#define MESSAGE_HASH_SIZE BIT(MESSAGE_HASH_BITS)
+
+#define MASK_52_7 (0x1FFFFFFFFFFF80)
+#define MASK_6_0  (0x7F)
+
+struct mlx5_fw_trace_data {
+	u64 timestamp;
+	bool lost;
+	u8 event_id;
+	char msg[TRACE_STR_MSG];
+};
+
+struct mlx5_fw_tracer {
+	struct mlx5_core_dev *dev;
+	struct mlx5_nb        nb;
+	bool owner;
+	u8   trc_ver;
+	struct workqueue_struct *work_queue;
+	struct work_struct ownership_change_work;
+	struct work_struct read_fw_strings_work;
+
+	/* Strings DB */
+	struct {
+		u8 first_string_trace;
+		u8 num_string_trace;
+		u32 num_string_db;
+		u32 base_address_out[STRINGS_DB_SECTIONS_NUM];
+		u32 size_out[STRINGS_DB_SECTIONS_NUM];
+		void *buffer[STRINGS_DB_SECTIONS_NUM];
+		bool loaded;
+	} str_db;
+
+	/* Log Buffer */
+	struct {
+		u32 pdn;
+		void *log_buf;
+		dma_addr_t dma;
+		u32 size;
+		u32 mkey;
+		u32 consumer_index;
+	} buff;
+
+	/* Saved Traces Array */
+	struct {
+		struct mlx5_fw_trace_data straces[SAVED_TRACES_NUM];
+		u32 saved_traces_index;
+		struct mutex lock; /* Protect st_arr access */
+	} st_arr;
+
+	u64 last_timestamp;
+	struct work_struct handle_traces_work;
+	struct hlist_head hash[MESSAGE_HASH_SIZE];
+	struct list_head ready_strings_list;
+};
+
+struct tracer_string_format {
+	char *string;
+	int params[TRACER_MAX_PARAMS];
+	int num_of_params;
+	int last_param_num;
+	u8 event_id;
+	u32 tmsn;
+	struct hlist_node hlist;
+	struct list_head list;
+	u32 timestamp;
+	bool lost;
+};
+
+enum mlx5_fw_tracer_ownership_state {
+	MLX5_FW_TRACER_RELEASE_OWNERSHIP,
+	MLX5_FW_TRACER_ACQUIRE_OWNERSHIP,
+};
+
+enum tracer_ctrl_fields_select {
+	TRACE_STATUS = 1 << 0,
+};
+
+enum tracer_event_type {
+	TRACER_EVENT_TYPE_STRING,
+	TRACER_EVENT_TYPE_TIMESTAMP = 0xFF,
+	TRACER_EVENT_TYPE_UNRECOGNIZED,
+};
+
+enum tracing_mode {
+	TRACE_TO_MEMORY = 1 << 0,
+};
+
+struct tracer_timestamp_event {
+	u64        timestamp;
+	u8         unreliable;
+};
+
+struct tracer_string_event {
+	u32        timestamp;
+	u32        tmsn;
+	u32        tdsn;
+	u32        string_param;
+};
+
+struct tracer_event {
+	bool      lost_event;
+	u32       type;
+	u8        event_id;
+	union {
+		struct tracer_string_event string_event;
+		struct tracer_timestamp_event timestamp_event;
+	};
+};
+
+struct mlx5_ifc_tracer_event_bits {
+	u8         lost[0x1];
+	u8         timestamp[0x7];
+	u8         event_id[0x8];
+	u8         event_data[0x30];
+};
+
+struct mlx5_ifc_tracer_string_event_bits {
+	u8         lost[0x1];
+	u8         timestamp[0x7];
+	u8         event_id[0x8];
+	u8         tmsn[0xd];
+	u8         tdsn[0x3];
+	u8         string_param[0x20];
+};
+
+struct mlx5_ifc_tracer_timestamp_event_bits {
+	u8         timestamp7_0[0x8];
+	u8         event_id[0x8];
+	u8         urts[0x3];
+	u8         timestamp52_40[0xd];
+	u8         timestamp39_8[0x20];
+};
+
+struct mlx5_fw_tracer *mlx5_fw_tracer_create(struct mlx5_core_dev *dev);
+int mlx5_fw_tracer_init(struct mlx5_fw_tracer *tracer);
+void mlx5_fw_tracer_cleanup(struct mlx5_fw_tracer *tracer);
+void mlx5_fw_tracer_destroy(struct mlx5_fw_tracer *tracer);
+int mlx5_fw_tracer_trigger_core_dump_general(struct mlx5_core_dev *dev);
+int mlx5_fw_tracer_get_saved_traces_objects(struct mlx5_fw_tracer *tracer,
+					    struct devlink_fmsg *fmsg);
+int mlx5_fw_tracer_reload(struct mlx5_fw_tracer *tracer);
+
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/diag/fw_tracer_tracepoint.h b/drivers/net/ethernet/mellanox/mlx5/core/diag/fw_tracer_tracepoint.h
new file mode 100644
index 000000000..3038be575
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/diag/fw_tracer_tracepoint.h
@@ -0,0 +1,79 @@
+/*
+ * Copyright (c) 2018, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#if !defined(__LIB_TRACER_TRACEPOINT_H__) || defined(TRACE_HEADER_MULTI_READ)
+#define __LIB_TRACER_TRACEPOINT_H__
+
+#include <linux/tracepoint.h>
+#include "fw_tracer.h"
+
+#undef TRACE_SYSTEM
+#define TRACE_SYSTEM mlx5
+
+/* Tracepoint for FWTracer messages: */
+TRACE_EVENT(mlx5_fw,
+	TP_PROTO(const struct mlx5_fw_tracer *tracer, u64 trace_timestamp,
+		 bool lost, u8 event_id, const char *msg),
+
+	TP_ARGS(tracer, trace_timestamp, lost, event_id, msg),
+
+	TP_STRUCT__entry(
+		__string(dev_name, dev_name(tracer->dev->device))
+		__field(u64, trace_timestamp)
+		__field(bool, lost)
+		__field(u8, event_id)
+		__string(msg, msg)
+	),
+
+	TP_fast_assign(
+		__assign_str(dev_name,
+			     dev_name(tracer->dev->device));
+		__entry->trace_timestamp = trace_timestamp;
+		__entry->lost = lost;
+		__entry->event_id = event_id;
+		__assign_str(msg, msg);
+	),
+
+	TP_printk("%s [0x%llx] %d [0x%x] %s",
+		  __get_str(dev_name),
+		  __entry->trace_timestamp,
+		  __entry->lost, __entry->event_id,
+		  __get_str(msg))
+);
+
+#endif
+
+#undef TRACE_INCLUDE_PATH
+#undef TRACE_INCLUDE_FILE
+#define TRACE_INCLUDE_PATH ./diag
+#define TRACE_INCLUDE_FILE fw_tracer_tracepoint
+#include <trace/define_trace.h>
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/diag/rsc_dump.c b/drivers/net/ethernet/mellanox/mlx5/core/diag/rsc_dump.c
new file mode 100644
index 000000000..c5b560a8b
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/diag/rsc_dump.c
@@ -0,0 +1,311 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2019 Mellanox Technologies. */
+
+#include "rsc_dump.h"
+#include "lib/mlx5.h"
+
+#define MLX5_SGMT_TYPE(SGMT) MLX5_SGMT_TYPE_##SGMT
+#define MLX5_SGMT_STR_ASSING(SGMT)[MLX5_SGMT_TYPE(SGMT)] = #SGMT
+static const char *const mlx5_rsc_sgmt_name[] = {
+	MLX5_SGMT_STR_ASSING(HW_CQPC),
+	MLX5_SGMT_STR_ASSING(HW_SQPC),
+	MLX5_SGMT_STR_ASSING(HW_RQPC),
+	MLX5_SGMT_STR_ASSING(FULL_SRQC),
+	MLX5_SGMT_STR_ASSING(FULL_CQC),
+	MLX5_SGMT_STR_ASSING(FULL_EQC),
+	MLX5_SGMT_STR_ASSING(FULL_QPC),
+	MLX5_SGMT_STR_ASSING(SND_BUFF),
+	MLX5_SGMT_STR_ASSING(RCV_BUFF),
+	MLX5_SGMT_STR_ASSING(SRQ_BUFF),
+	MLX5_SGMT_STR_ASSING(CQ_BUFF),
+	MLX5_SGMT_STR_ASSING(EQ_BUFF),
+	MLX5_SGMT_STR_ASSING(SX_SLICE),
+	MLX5_SGMT_STR_ASSING(SX_SLICE_ALL),
+	MLX5_SGMT_STR_ASSING(RDB),
+	MLX5_SGMT_STR_ASSING(RX_SLICE_ALL),
+	MLX5_SGMT_STR_ASSING(PRM_QUERY_QP),
+	MLX5_SGMT_STR_ASSING(PRM_QUERY_CQ),
+	MLX5_SGMT_STR_ASSING(PRM_QUERY_MKEY),
+};
+
+struct mlx5_rsc_dump {
+	u32 pdn;
+	u32 mkey;
+	u32 number_of_menu_items;
+	u16 fw_segment_type[MLX5_SGMT_TYPE_NUM];
+};
+
+struct mlx5_rsc_dump_cmd {
+	u64 mem_size;
+	u8 cmd[MLX5_ST_SZ_BYTES(resource_dump)];
+};
+
+static int mlx5_rsc_dump_sgmt_get_by_name(char *name)
+{
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(mlx5_rsc_sgmt_name); i++)
+		if (!strcmp(name, mlx5_rsc_sgmt_name[i]))
+			return i;
+
+	return -EINVAL;
+}
+
+#define MLX5_RSC_DUMP_MENU_HEADER_SIZE (MLX5_ST_SZ_BYTES(resource_dump_info_segment) + \
+					MLX5_ST_SZ_BYTES(resource_dump_command_segment) + \
+					MLX5_ST_SZ_BYTES(resource_dump_menu_segment))
+
+static int mlx5_rsc_dump_read_menu_sgmt(struct mlx5_rsc_dump *rsc_dump, struct page *page,
+					int read_size, int start_idx)
+{
+	void *data = page_address(page);
+	enum mlx5_sgmt_type sgmt_idx;
+	int num_of_items;
+	char *sgmt_name;
+	void *member;
+	int size = 0;
+	void *menu;
+	int i;
+
+	if (!start_idx) {
+		menu = MLX5_ADDR_OF(menu_resource_dump_response, data, menu);
+		rsc_dump->number_of_menu_items = MLX5_GET(resource_dump_menu_segment, menu,
+							  num_of_records);
+		size = MLX5_RSC_DUMP_MENU_HEADER_SIZE;
+		data += size;
+	}
+	num_of_items = rsc_dump->number_of_menu_items;
+
+	for (i = 0; start_idx + i < num_of_items; i++) {
+		size += MLX5_ST_SZ_BYTES(resource_dump_menu_record);
+		if (size >= read_size)
+			return start_idx + i;
+
+		member = data + MLX5_ST_SZ_BYTES(resource_dump_menu_record) * i;
+		sgmt_name =  MLX5_ADDR_OF(resource_dump_menu_record, member, segment_name);
+		sgmt_idx = mlx5_rsc_dump_sgmt_get_by_name(sgmt_name);
+		if (sgmt_idx == -EINVAL)
+			continue;
+		rsc_dump->fw_segment_type[sgmt_idx] = MLX5_GET(resource_dump_menu_record,
+							       member, segment_type);
+	}
+	return 0;
+}
+
+static int mlx5_rsc_dump_trigger(struct mlx5_core_dev *dev, struct mlx5_rsc_dump_cmd *cmd,
+				 struct page *page)
+{
+	struct mlx5_rsc_dump *rsc_dump = dev->rsc_dump;
+	struct device *ddev = mlx5_core_dma_dev(dev);
+	u32 out_seq_num;
+	u32 in_seq_num;
+	dma_addr_t dma;
+	int err;
+
+	dma = dma_map_page(ddev, page, 0, cmd->mem_size, DMA_FROM_DEVICE);
+	if (unlikely(dma_mapping_error(ddev, dma)))
+		return -ENOMEM;
+
+	in_seq_num = MLX5_GET(resource_dump, cmd->cmd, seq_num);
+	MLX5_SET(resource_dump, cmd->cmd, mkey, rsc_dump->mkey);
+	MLX5_SET64(resource_dump, cmd->cmd, address, dma);
+
+	err = mlx5_core_access_reg(dev, cmd->cmd, sizeof(cmd->cmd), cmd->cmd,
+				   sizeof(cmd->cmd), MLX5_REG_RESOURCE_DUMP, 0, 1);
+	if (err) {
+		mlx5_core_err(dev, "Resource dump: Failed to access err %d\n", err);
+		goto out;
+	}
+	out_seq_num = MLX5_GET(resource_dump, cmd->cmd, seq_num);
+	if (out_seq_num && (in_seq_num + 1 != out_seq_num))
+		err = -EIO;
+out:
+	dma_unmap_page(ddev, dma, cmd->mem_size, DMA_FROM_DEVICE);
+	return err;
+}
+
+struct mlx5_rsc_dump_cmd *mlx5_rsc_dump_cmd_create(struct mlx5_core_dev *dev,
+						   struct mlx5_rsc_key *key)
+{
+	struct mlx5_rsc_dump_cmd *cmd;
+	int sgmt_type;
+
+	if (IS_ERR_OR_NULL(dev->rsc_dump))
+		return ERR_PTR(-EOPNOTSUPP);
+
+	sgmt_type = dev->rsc_dump->fw_segment_type[key->rsc];
+	if (!sgmt_type && key->rsc != MLX5_SGMT_TYPE_MENU)
+		return ERR_PTR(-EOPNOTSUPP);
+
+	cmd = kzalloc(sizeof(*cmd), GFP_KERNEL);
+	if (!cmd) {
+		mlx5_core_err(dev, "Resource dump: Failed to allocate command\n");
+		return ERR_PTR(-ENOMEM);
+	}
+	MLX5_SET(resource_dump, cmd->cmd, segment_type, sgmt_type);
+	MLX5_SET(resource_dump, cmd->cmd, index1, key->index1);
+	MLX5_SET(resource_dump, cmd->cmd, index2, key->index2);
+	MLX5_SET(resource_dump, cmd->cmd, num_of_obj1, key->num_of_obj1);
+	MLX5_SET(resource_dump, cmd->cmd, num_of_obj2, key->num_of_obj2);
+	MLX5_SET(resource_dump, cmd->cmd, size, key->size);
+	cmd->mem_size = key->size;
+	return cmd;
+}
+EXPORT_SYMBOL(mlx5_rsc_dump_cmd_create);
+
+void mlx5_rsc_dump_cmd_destroy(struct mlx5_rsc_dump_cmd *cmd)
+{
+	kfree(cmd);
+}
+EXPORT_SYMBOL(mlx5_rsc_dump_cmd_destroy);
+
+int mlx5_rsc_dump_next(struct mlx5_core_dev *dev, struct mlx5_rsc_dump_cmd *cmd,
+		       struct page *page, int *size)
+{
+	bool more_dump;
+	int err;
+
+	if (IS_ERR_OR_NULL(dev->rsc_dump))
+		return -EOPNOTSUPP;
+
+	err = mlx5_rsc_dump_trigger(dev, cmd, page);
+	if (err) {
+		mlx5_core_err(dev, "Resource dump: Failed to trigger dump, %d\n", err);
+		return err;
+	}
+	*size = MLX5_GET(resource_dump, cmd->cmd, size);
+	more_dump = MLX5_GET(resource_dump, cmd->cmd, more_dump);
+
+	return more_dump;
+}
+EXPORT_SYMBOL(mlx5_rsc_dump_next);
+
+#define MLX5_RSC_DUMP_MENU_SEGMENT 0xffff
+static int mlx5_rsc_dump_menu(struct mlx5_core_dev *dev)
+{
+	struct mlx5_rsc_dump_cmd *cmd = NULL;
+	struct mlx5_rsc_key key = {};
+	struct page *page;
+	int start_idx = 0;
+	int size;
+	int err;
+
+	page = alloc_page(GFP_KERNEL);
+	if (!page)
+		return -ENOMEM;
+
+	key.rsc = MLX5_SGMT_TYPE_MENU;
+	key.size = PAGE_SIZE;
+	cmd  = mlx5_rsc_dump_cmd_create(dev, &key);
+	if (IS_ERR(cmd)) {
+		err = PTR_ERR(cmd);
+		goto free_page;
+	}
+	MLX5_SET(resource_dump, cmd->cmd, segment_type, MLX5_RSC_DUMP_MENU_SEGMENT);
+
+	do {
+		err = mlx5_rsc_dump_next(dev, cmd, page, &size);
+		if (err < 0)
+			goto destroy_cmd;
+
+		start_idx = mlx5_rsc_dump_read_menu_sgmt(dev->rsc_dump, page, size, start_idx);
+
+	} while (err > 0);
+
+destroy_cmd:
+	mlx5_rsc_dump_cmd_destroy(cmd);
+free_page:
+	__free_page(page);
+
+	return err;
+}
+
+static int mlx5_rsc_dump_create_mkey(struct mlx5_core_dev *mdev, u32 pdn,
+				     u32 *mkey)
+{
+	int inlen = MLX5_ST_SZ_BYTES(create_mkey_in);
+	void *mkc;
+	u32 *in;
+	int err;
+
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in)
+		return -ENOMEM;
+
+	mkc = MLX5_ADDR_OF(create_mkey_in, in, memory_key_mkey_entry);
+	MLX5_SET(mkc, mkc, access_mode_1_0, MLX5_MKC_ACCESS_MODE_PA);
+	MLX5_SET(mkc, mkc, lw, 1);
+	MLX5_SET(mkc, mkc, lr, 1);
+
+	MLX5_SET(mkc, mkc, pd, pdn);
+	MLX5_SET(mkc, mkc, length64, 1);
+	MLX5_SET(mkc, mkc, qpn, 0xffffff);
+
+	err = mlx5_core_create_mkey(mdev, mkey, in, inlen);
+
+	kvfree(in);
+	return err;
+}
+
+struct mlx5_rsc_dump *mlx5_rsc_dump_create(struct mlx5_core_dev *dev)
+{
+	struct mlx5_rsc_dump *rsc_dump;
+
+	if (!MLX5_CAP_DEBUG(dev, resource_dump)) {
+		mlx5_core_dbg(dev, "Resource dump: capability not present\n");
+		return NULL;
+	}
+	rsc_dump = kzalloc(sizeof(*rsc_dump), GFP_KERNEL);
+	if (!rsc_dump)
+		return ERR_PTR(-ENOMEM);
+
+	return rsc_dump;
+}
+
+void mlx5_rsc_dump_destroy(struct mlx5_core_dev *dev)
+{
+	if (IS_ERR_OR_NULL(dev->rsc_dump))
+		return;
+	kfree(dev->rsc_dump);
+}
+
+int mlx5_rsc_dump_init(struct mlx5_core_dev *dev)
+{
+	struct mlx5_rsc_dump *rsc_dump = dev->rsc_dump;
+	int err;
+
+	if (IS_ERR_OR_NULL(dev->rsc_dump))
+		return 0;
+
+	err = mlx5_core_alloc_pd(dev, &rsc_dump->pdn);
+	if (err) {
+		mlx5_core_warn(dev, "Resource dump: Failed to allocate PD %d\n", err);
+		return err;
+	}
+	err = mlx5_rsc_dump_create_mkey(dev, rsc_dump->pdn, &rsc_dump->mkey);
+	if (err) {
+		mlx5_core_err(dev, "Resource dump: Failed to create mkey, %d\n", err);
+		goto free_pd;
+	}
+	err = mlx5_rsc_dump_menu(dev);
+	if (err) {
+		mlx5_core_err(dev, "Resource dump: Failed to read menu, %d\n", err);
+		goto destroy_mkey;
+	}
+	return err;
+
+destroy_mkey:
+	mlx5_core_destroy_mkey(dev, rsc_dump->mkey);
+free_pd:
+	mlx5_core_dealloc_pd(dev, rsc_dump->pdn);
+	return err;
+}
+
+void mlx5_rsc_dump_cleanup(struct mlx5_core_dev *dev)
+{
+	if (IS_ERR_OR_NULL(dev->rsc_dump))
+		return;
+
+	mlx5_core_destroy_mkey(dev, dev->rsc_dump->mkey);
+	mlx5_core_dealloc_pd(dev, dev->rsc_dump->pdn);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/diag/rsc_dump.h b/drivers/net/ethernet/mellanox/mlx5/core/diag/rsc_dump.h
new file mode 100644
index 000000000..64c4956db
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/diag/rsc_dump.h
@@ -0,0 +1,27 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/* Copyright (c) 2019 Mellanox Technologies. */
+
+#ifndef __MLX5_RSC_DUMP_H
+#define __MLX5_RSC_DUMP_H
+
+#include <linux/mlx5/rsc_dump.h>
+#include <linux/mlx5/driver.h>
+#include "mlx5_core.h"
+
+#define MLX5_RSC_DUMP_ALL 0xFFFF
+struct mlx5_rsc_dump_cmd;
+struct mlx5_rsc_dump;
+
+struct mlx5_rsc_dump *mlx5_rsc_dump_create(struct mlx5_core_dev *dev);
+void mlx5_rsc_dump_destroy(struct mlx5_core_dev *dev);
+
+int mlx5_rsc_dump_init(struct mlx5_core_dev *dev);
+void mlx5_rsc_dump_cleanup(struct mlx5_core_dev *dev);
+
+struct mlx5_rsc_dump_cmd *mlx5_rsc_dump_cmd_create(struct mlx5_core_dev *dev,
+						   struct mlx5_rsc_key *key);
+void mlx5_rsc_dump_cmd_destroy(struct mlx5_rsc_dump_cmd *cmd);
+
+int mlx5_rsc_dump_next(struct mlx5_core_dev *dev, struct mlx5_rsc_dump_cmd *cmd,
+		       struct page *page, int *size);
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/ecpf.c b/drivers/net/ethernet/mellanox/mlx5/core/ecpf.c
new file mode 100644
index 000000000..d000236dd
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/ecpf.c
@@ -0,0 +1,97 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2019 Mellanox Technologies. */
+
+#include "ecpf.h"
+
+bool mlx5_read_embedded_cpu(struct mlx5_core_dev *dev)
+{
+	return (ioread32be(&dev->iseg->initializing) >> MLX5_ECPU_BIT_NUM) & 1;
+}
+
+static bool mlx5_ecpf_esw_admins_host_pf(const struct mlx5_core_dev *dev)
+{
+	/* In separate host mode, PF enables itself.
+	 * When ECPF is eswitch manager, eswitch enables host PF after
+	 * eswitch is setup.
+	 */
+	return mlx5_core_is_ecpf_esw_manager(dev);
+}
+
+int mlx5_cmd_host_pf_enable_hca(struct mlx5_core_dev *dev)
+{
+	u32 out[MLX5_ST_SZ_DW(enable_hca_out)] = {};
+	u32 in[MLX5_ST_SZ_DW(enable_hca_in)]   = {};
+
+	MLX5_SET(enable_hca_in, in, opcode, MLX5_CMD_OP_ENABLE_HCA);
+	MLX5_SET(enable_hca_in, in, function_id, 0);
+	MLX5_SET(enable_hca_in, in, embedded_cpu_function, 0);
+	return mlx5_cmd_exec(dev, &in, sizeof(in), &out, sizeof(out));
+}
+
+int mlx5_cmd_host_pf_disable_hca(struct mlx5_core_dev *dev)
+{
+	u32 out[MLX5_ST_SZ_DW(disable_hca_out)] = {};
+	u32 in[MLX5_ST_SZ_DW(disable_hca_in)]   = {};
+
+	MLX5_SET(disable_hca_in, in, opcode, MLX5_CMD_OP_DISABLE_HCA);
+	MLX5_SET(disable_hca_in, in, function_id, 0);
+	MLX5_SET(disable_hca_in, in, embedded_cpu_function, 0);
+	return mlx5_cmd_exec(dev, in, sizeof(in), out, sizeof(out));
+}
+
+static int mlx5_host_pf_init(struct mlx5_core_dev *dev)
+{
+	int err;
+
+	if (mlx5_ecpf_esw_admins_host_pf(dev))
+		return 0;
+
+	/* ECPF shall enable HCA for host PF in the same way a PF
+	 * does this for its VFs when ECPF is not a eswitch manager.
+	 */
+	err = mlx5_cmd_host_pf_enable_hca(dev);
+	if (err)
+		mlx5_core_err(dev, "Failed to enable external host PF HCA err(%d)\n", err);
+
+	return err;
+}
+
+static void mlx5_host_pf_cleanup(struct mlx5_core_dev *dev)
+{
+	int err;
+
+	if (mlx5_ecpf_esw_admins_host_pf(dev))
+		return;
+
+	err = mlx5_cmd_host_pf_disable_hca(dev);
+	if (err) {
+		mlx5_core_err(dev, "Failed to disable external host PF HCA err(%d)\n", err);
+		return;
+	}
+}
+
+int mlx5_ec_init(struct mlx5_core_dev *dev)
+{
+	if (!mlx5_core_is_ecpf(dev))
+		return 0;
+
+	return mlx5_host_pf_init(dev);
+}
+
+void mlx5_ec_cleanup(struct mlx5_core_dev *dev)
+{
+	int err;
+
+	if (!mlx5_core_is_ecpf(dev))
+		return;
+
+	mlx5_host_pf_cleanup(dev);
+
+	err = mlx5_wait_for_pages(dev, &dev->priv.page_counters[MLX5_HOST_PF]);
+	if (err)
+		mlx5_core_warn(dev, "Timeout reclaiming external host PF pages err(%d)\n", err);
+
+	err = mlx5_wait_for_pages(dev, &dev->priv.page_counters[MLX5_VF]);
+	if (err)
+		mlx5_core_warn(dev, "Timeout reclaiming external host VFs pages err(%d)\n", err);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/ecpf.h b/drivers/net/ethernet/mellanox/mlx5/core/ecpf.h
new file mode 100644
index 000000000..40b6ad76d
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/ecpf.h
@@ -0,0 +1,32 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2019 Mellanox Technologies. */
+
+#ifndef __MLX5_ECPF_H__
+#define __MLX5_ECPF_H__
+
+#include <linux/mlx5/driver.h>
+#include "mlx5_core.h"
+
+#ifdef CONFIG_MLX5_ESWITCH
+
+enum {
+	MLX5_ECPU_BIT_NUM = 23,
+};
+
+bool mlx5_read_embedded_cpu(struct mlx5_core_dev *dev);
+int mlx5_ec_init(struct mlx5_core_dev *dev);
+void mlx5_ec_cleanup(struct mlx5_core_dev *dev);
+
+int mlx5_cmd_host_pf_enable_hca(struct mlx5_core_dev *dev);
+int mlx5_cmd_host_pf_disable_hca(struct mlx5_core_dev *dev);
+
+#else  /* CONFIG_MLX5_ESWITCH */
+
+static inline bool
+mlx5_read_embedded_cpu(struct mlx5_core_dev *dev) { return false; }
+static inline int mlx5_ec_init(struct mlx5_core_dev *dev) { return 0; }
+static inline void mlx5_ec_cleanup(struct mlx5_core_dev *dev) {}
+
+#endif /* CONFIG_MLX5_ESWITCH */
+
+#endif /* __MLX5_ECPF_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en.h b/drivers/net/ethernet/mellanox/mlx5/core/en.h
new file mode 100644
index 000000000..0ee456480
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en.h
@@ -0,0 +1,1246 @@
+/*
+ * Copyright (c) 2015-2016, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+#ifndef __MLX5_EN_H__
+#define __MLX5_EN_H__
+
+#include <linux/if_vlan.h>
+#include <linux/etherdevice.h>
+#include <linux/timecounter.h>
+#include <linux/net_tstamp.h>
+#include <linux/crash_dump.h>
+#include <linux/mlx5/driver.h>
+#include <linux/mlx5/qp.h>
+#include <linux/mlx5/cq.h>
+#include <linux/mlx5/port.h>
+#include <linux/mlx5/vport.h>
+#include <linux/mlx5/transobj.h>
+#include <linux/mlx5/fs.h>
+#include <linux/rhashtable.h>
+#include <net/udp_tunnel.h>
+#include <net/switchdev.h>
+#include <net/xdp.h>
+#include <linux/dim.h>
+#include <linux/bits.h>
+#include "wq.h"
+#include "mlx5_core.h"
+#include "en_stats.h"
+#include "en/dcbnl.h"
+#include "en/fs.h"
+#include "en/qos.h"
+#include "lib/hv_vhca.h"
+#include "lib/clock.h"
+#include "en/rx_res.h"
+#include "en/selq.h"
+
+extern const struct net_device_ops mlx5e_netdev_ops;
+struct page_pool;
+
+#define MLX5E_METADATA_ETHER_TYPE (0x8CE4)
+#define MLX5E_METADATA_ETHER_LEN 8
+
+#define MLX5E_ETH_HARD_MTU (ETH_HLEN + VLAN_HLEN + ETH_FCS_LEN)
+
+#define MLX5E_HW2SW_MTU(params, hwmtu) ((hwmtu) - ((params)->hard_mtu))
+#define MLX5E_SW2HW_MTU(params, swmtu) ((swmtu) + ((params)->hard_mtu))
+
+#define MLX5E_MAX_NUM_TC	8
+#define MLX5E_MAX_NUM_MQPRIO_CH_TC TC_QOPT_MAX_QUEUE
+
+#define MLX5_RX_HEADROOM NET_SKB_PAD
+#define MLX5_SKB_FRAG_SZ(len)	(SKB_DATA_ALIGN(len) +	\
+				 SKB_DATA_ALIGN(sizeof(struct skb_shared_info)))
+
+#define MLX5E_RX_MAX_HEAD (256)
+#define MLX5E_SHAMPO_LOG_MAX_HEADER_ENTRY_SIZE (9)
+#define MLX5E_SHAMPO_WQ_HEADER_PER_PAGE (PAGE_SIZE >> MLX5E_SHAMPO_LOG_MAX_HEADER_ENTRY_SIZE)
+#define MLX5E_SHAMPO_WQ_BASE_HEAD_ENTRY_SIZE (64)
+#define MLX5E_SHAMPO_WQ_RESRV_SIZE (64 * 1024)
+#define MLX5E_SHAMPO_WQ_BASE_RESRV_SIZE (4096)
+
+#define MLX5_MPWRQ_MIN_LOG_STRIDE_SZ(mdev) \
+	(6 + MLX5_CAP_GEN(mdev, cache_line_128byte)) /* HW restriction */
+#define MLX5_MPWRQ_LOG_STRIDE_SZ(mdev, req) \
+	max_t(u32, MLX5_MPWRQ_MIN_LOG_STRIDE_SZ(mdev), req)
+#define MLX5_MPWRQ_DEF_LOG_STRIDE_SZ(mdev) \
+	MLX5_MPWRQ_LOG_STRIDE_SZ(mdev, order_base_2(MLX5E_RX_MAX_HEAD))
+
+#define MLX5_MPWRQ_MAX_LOG_WQE_SZ 18
+
+/* Keep in sync with mlx5e_mpwrq_log_wqe_sz.
+ * These are theoretical maximums, which can be further restricted by
+ * capabilities. These values are used for static resource allocations and
+ * sanity checks.
+ * MLX5_SEND_WQE_MAX_SIZE is a bit bigger than the maximum cacheline-aligned WQE
+ * size actually used at runtime, but it's not a problem when calculating static
+ * array sizes.
+ */
+#define MLX5_UMR_MAX_MTT_SPACE \
+	(ALIGN_DOWN(MLX5_SEND_WQE_MAX_SIZE - sizeof(struct mlx5e_umr_wqe), \
+		    MLX5_UMR_MTT_ALIGNMENT))
+#define MLX5_MPWRQ_MAX_PAGES_PER_WQE \
+	rounddown_pow_of_two(MLX5_UMR_MAX_MTT_SPACE / sizeof(struct mlx5_mtt))
+
+#define MLX5E_MAX_RQ_NUM_MTTS	\
+	(ALIGN_DOWN(U16_MAX, 4) * 2) /* Fits into u16 and aligned by WQEBB. */
+#define MLX5E_MAX_RQ_NUM_KSMS (U16_MAX - 1) /* So that num_ksms fits into u16. */
+#define MLX5E_ORDER2_MAX_PACKET_MTU (order_base_2(10 * 1024))
+
+#define MLX5E_MIN_SKB_FRAG_SZ		(MLX5_SKB_FRAG_SZ(MLX5_RX_HEADROOM))
+#define MLX5E_LOG_MAX_RX_WQE_BULK	\
+	(ilog2(PAGE_SIZE / roundup_pow_of_two(MLX5E_MIN_SKB_FRAG_SZ)))
+
+#define MLX5E_PARAMS_MINIMUM_LOG_SQ_SIZE                0x6
+#define MLX5E_PARAMS_DEFAULT_LOG_SQ_SIZE                0xa
+#define MLX5E_PARAMS_MAXIMUM_LOG_SQ_SIZE                0xd
+
+#define MLX5E_PARAMS_MINIMUM_LOG_RQ_SIZE (1 + MLX5E_LOG_MAX_RX_WQE_BULK)
+#define MLX5E_PARAMS_DEFAULT_LOG_RQ_SIZE                0xa
+#define MLX5E_PARAMS_MAXIMUM_LOG_RQ_SIZE		0xd
+
+#define MLX5E_PARAMS_MINIMUM_LOG_RQ_SIZE_MPW            0x2
+
+#define MLX5E_DEFAULT_LRO_TIMEOUT                       32
+#define MLX5E_LRO_TIMEOUT_ARR_SIZE                      4
+
+#define MLX5E_PARAMS_DEFAULT_RX_CQ_MODERATION_USEC      0x10
+#define MLX5E_PARAMS_DEFAULT_RX_CQ_MODERATION_USEC_FROM_CQE 0x3
+#define MLX5E_PARAMS_DEFAULT_RX_CQ_MODERATION_PKTS      0x20
+#define MLX5E_PARAMS_DEFAULT_TX_CQ_MODERATION_USEC      0x10
+#define MLX5E_PARAMS_DEFAULT_TX_CQ_MODERATION_USEC_FROM_CQE 0x10
+#define MLX5E_PARAMS_DEFAULT_TX_CQ_MODERATION_PKTS      0x20
+#define MLX5E_PARAMS_DEFAULT_MIN_RX_WQES                0x80
+#define MLX5E_PARAMS_DEFAULT_MIN_RX_WQES_MPW            0x2
+
+#define MLX5E_MIN_NUM_CHANNELS         0x1
+#define MLX5E_MAX_NUM_CHANNELS         (MLX5E_INDIR_RQT_SIZE / 2)
+#define MLX5E_TX_CQ_POLL_BUDGET        128
+#define MLX5E_TX_XSK_POLL_BUDGET       64
+#define MLX5E_SQ_RECOVER_MIN_INTERVAL  500 /* msecs */
+
+#define MLX5E_KLM_UMR_WQE_SZ(sgl_len)\
+	(sizeof(struct mlx5e_umr_wqe) +\
+	(sizeof(struct mlx5_klm) * (sgl_len)))
+
+#define MLX5E_KLM_UMR_WQEBBS(klm_entries) \
+	(DIV_ROUND_UP(MLX5E_KLM_UMR_WQE_SZ(klm_entries), MLX5_SEND_WQE_BB))
+
+#define MLX5E_KLM_UMR_DS_CNT(klm_entries)\
+	(DIV_ROUND_UP(MLX5E_KLM_UMR_WQE_SZ(klm_entries), MLX5_SEND_WQE_DS))
+
+#define MLX5E_KLM_MAX_ENTRIES_PER_WQE(wqe_size)\
+	(((wqe_size) - sizeof(struct mlx5e_umr_wqe)) / sizeof(struct mlx5_klm))
+
+#define MLX5E_KLM_ENTRIES_PER_WQE(wqe_size)\
+	ALIGN_DOWN(MLX5E_KLM_MAX_ENTRIES_PER_WQE(wqe_size), MLX5_UMR_KLM_ALIGNMENT)
+
+#define MLX5E_MAX_KLM_PER_WQE(mdev) \
+	MLX5E_KLM_ENTRIES_PER_WQE(MLX5_SEND_WQE_BB * mlx5e_get_max_sq_aligned_wqebbs(mdev))
+
+#define MLX5E_MSG_LEVEL			NETIF_MSG_LINK
+
+#define mlx5e_dbg(mlevel, priv, format, ...)                    \
+do {                                                            \
+	if (NETIF_MSG_##mlevel & (priv)->msglevel)              \
+		netdev_warn(priv->netdev, format,               \
+			    ##__VA_ARGS__);                     \
+} while (0)
+
+#define mlx5e_state_dereference(priv, p) \
+	rcu_dereference_protected((p), lockdep_is_held(&(priv)->state_lock))
+
+static inline u8 mlx5e_get_num_lag_ports(struct mlx5_core_dev *mdev)
+{
+	if (mlx5_lag_is_lacp_owner(mdev))
+		return 1;
+
+	return clamp_t(u8, MLX5_CAP_GEN(mdev, num_lag_ports), 1, MLX5_MAX_PORTS);
+}
+
+static inline u16 mlx5_min_rx_wqes(int wq_type, u32 wq_size)
+{
+	switch (wq_type) {
+	case MLX5_WQ_TYPE_LINKED_LIST_STRIDING_RQ:
+		return min_t(u16, MLX5E_PARAMS_DEFAULT_MIN_RX_WQES_MPW,
+			     wq_size / 2);
+	default:
+		return min_t(u16, MLX5E_PARAMS_DEFAULT_MIN_RX_WQES,
+			     wq_size / 2);
+	}
+}
+
+/* Use this function to get max num channels (rxqs/txqs) only to create netdev */
+static inline int mlx5e_get_max_num_channels(struct mlx5_core_dev *mdev)
+{
+	return is_kdump_kernel() ?
+		MLX5E_MIN_NUM_CHANNELS :
+		min_t(int, mlx5_comp_vectors_count(mdev), MLX5E_MAX_NUM_CHANNELS);
+}
+
+/* The maximum WQE size can be retrieved by max_wqe_sz_sq in
+ * bytes units. Driver hardens the limitation to 1KB (16
+ * WQEBBs), unless firmware capability is stricter.
+ */
+static inline u8 mlx5e_get_max_sq_wqebbs(struct mlx5_core_dev *mdev)
+{
+	BUILD_BUG_ON(MLX5_SEND_WQE_MAX_WQEBBS > U8_MAX);
+
+	return (u8)min_t(u16, MLX5_SEND_WQE_MAX_WQEBBS,
+			 MLX5_CAP_GEN(mdev, max_wqe_sz_sq) / MLX5_SEND_WQE_BB);
+}
+
+static inline u8 mlx5e_get_max_sq_aligned_wqebbs(struct mlx5_core_dev *mdev)
+{
+/* The return value will be multiplied by MLX5_SEND_WQEBB_NUM_DS.
+ * Since max_sq_wqebbs may be up to MLX5_SEND_WQE_MAX_WQEBBS == 16,
+ * see mlx5e_get_max_sq_wqebbs(), the multiplication (16 * 4 == 64)
+ * overflows the 6-bit DS field of Ctrl Segment. Use a bound lower
+ * than MLX5_SEND_WQE_MAX_WQEBBS to let a full-session WQE be
+ * cache-aligned.
+ */
+	u8 wqebbs = mlx5e_get_max_sq_wqebbs(mdev);
+
+	wqebbs = min_t(u8, wqebbs, MLX5_SEND_WQE_MAX_WQEBBS - 1);
+#if L1_CACHE_BYTES >= 128
+	wqebbs = ALIGN_DOWN(wqebbs, 2);
+#endif
+	return wqebbs;
+}
+
+struct mlx5e_tx_wqe {
+	struct mlx5_wqe_ctrl_seg ctrl;
+	struct mlx5_wqe_eth_seg  eth;
+	struct mlx5_wqe_data_seg data[];
+};
+
+struct mlx5e_rx_wqe_ll {
+	struct mlx5_wqe_srq_next_seg  next;
+	struct mlx5_wqe_data_seg      data[];
+};
+
+struct mlx5e_rx_wqe_cyc {
+	struct mlx5_wqe_data_seg      data[0];
+};
+
+struct mlx5e_umr_wqe {
+	struct mlx5_wqe_ctrl_seg       ctrl;
+	struct mlx5_wqe_umr_ctrl_seg   uctrl;
+	struct mlx5_mkey_seg           mkc;
+	union {
+		DECLARE_FLEX_ARRAY(struct mlx5_mtt, inline_mtts);
+		DECLARE_FLEX_ARRAY(struct mlx5_klm, inline_klms);
+		DECLARE_FLEX_ARRAY(struct mlx5_ksm, inline_ksms);
+	};
+};
+
+enum mlx5e_priv_flag {
+	MLX5E_PFLAG_RX_CQE_BASED_MODER,
+	MLX5E_PFLAG_TX_CQE_BASED_MODER,
+	MLX5E_PFLAG_RX_CQE_COMPRESS,
+	MLX5E_PFLAG_RX_STRIDING_RQ,
+	MLX5E_PFLAG_RX_NO_CSUM_COMPLETE,
+	MLX5E_PFLAG_XDP_TX_MPWQE,
+	MLX5E_PFLAG_SKB_TX_MPWQE,
+	MLX5E_PFLAG_TX_PORT_TS,
+	MLX5E_NUM_PFLAGS, /* Keep last */
+};
+
+#define MLX5E_SET_PFLAG(params, pflag, enable)			\
+	do {							\
+		if (enable)					\
+			(params)->pflags |= BIT(pflag);		\
+		else						\
+			(params)->pflags &= ~(BIT(pflag));	\
+	} while (0)
+
+#define MLX5E_GET_PFLAG(params, pflag) (!!((params)->pflags & (BIT(pflag))))
+
+enum packet_merge {
+	MLX5E_PACKET_MERGE_NONE,
+	MLX5E_PACKET_MERGE_LRO,
+	MLX5E_PACKET_MERGE_SHAMPO,
+};
+
+struct mlx5e_packet_merge_param {
+	enum packet_merge type;
+	u32 timeout;
+	struct {
+		u8 match_criteria_type;
+		u8 alignment_granularity;
+	} shampo;
+};
+
+struct mlx5e_params {
+	u8  log_sq_size;
+	u8  rq_wq_type;
+	u8  log_rq_mtu_frames;
+	u16 num_channels;
+	struct {
+		u16 mode;
+		u8 num_tc;
+		struct netdev_tc_txq tc_to_txq[TC_MAX_QUEUE];
+		struct {
+			u64 max_rate[TC_MAX_QUEUE];
+			u32 hw_id[TC_MAX_QUEUE];
+		} channel;
+	} mqprio;
+	bool rx_cqe_compress_def;
+	struct dim_cq_moder rx_cq_moderation;
+	struct dim_cq_moder tx_cq_moderation;
+	struct mlx5e_packet_merge_param packet_merge;
+	u8  tx_min_inline_mode;
+	bool vlan_strip_disable;
+	bool scatter_fcs_en;
+	bool rx_dim_enabled;
+	bool tx_dim_enabled;
+	u32 pflags;
+	struct bpf_prog *xdp_prog;
+	struct mlx5e_xsk *xsk;
+	unsigned int sw_mtu;
+	int hard_mtu;
+	bool ptp_rx;
+};
+
+static inline u8 mlx5e_get_dcb_num_tc(struct mlx5e_params *params)
+{
+	return params->mqprio.mode == TC_MQPRIO_MODE_DCB ?
+		params->mqprio.num_tc : 1;
+}
+
+enum {
+	MLX5E_RQ_STATE_ENABLED,
+	MLX5E_RQ_STATE_RECOVERING,
+	MLX5E_RQ_STATE_AM,
+	MLX5E_RQ_STATE_NO_CSUM_COMPLETE,
+	MLX5E_RQ_STATE_CSUM_FULL, /* cqe_csum_full hw bit is set */
+	MLX5E_RQ_STATE_MINI_CQE_HW_STRIDX, /* set when mini_cqe_resp_stride_index cap is used */
+	MLX5E_RQ_STATE_SHAMPO, /* set when SHAMPO cap is used */
+};
+
+struct mlx5e_cq {
+	/* data path - accessed per cqe */
+	struct mlx5_cqwq           wq;
+
+	/* data path - accessed per napi poll */
+	u16                        event_ctr;
+	struct napi_struct        *napi;
+	struct mlx5_core_cq        mcq;
+	struct mlx5e_ch_stats     *ch_stats;
+
+	/* control */
+	struct net_device         *netdev;
+	struct mlx5_core_dev      *mdev;
+	struct mlx5e_priv         *priv;
+	struct mlx5_wq_ctrl        wq_ctrl;
+} ____cacheline_aligned_in_smp;
+
+struct mlx5e_cq_decomp {
+	/* cqe decompression */
+	struct mlx5_cqe64          title;
+	struct mlx5_mini_cqe8      mini_arr[MLX5_MINI_CQE_ARRAY_SIZE];
+	u8                         mini_arr_idx;
+	u16                        left;
+	u16                        wqe_counter;
+} ____cacheline_aligned_in_smp;
+
+enum mlx5e_dma_map_type {
+	MLX5E_DMA_MAP_SINGLE,
+	MLX5E_DMA_MAP_PAGE
+};
+
+struct mlx5e_sq_dma {
+	dma_addr_t              addr;
+	u32                     size;
+	enum mlx5e_dma_map_type type;
+};
+
+enum {
+	MLX5E_SQ_STATE_ENABLED,
+	MLX5E_SQ_STATE_MPWQE,
+	MLX5E_SQ_STATE_RECOVERING,
+	MLX5E_SQ_STATE_IPSEC,
+	MLX5E_SQ_STATE_AM,
+	MLX5E_SQ_STATE_VLAN_NEED_L2_INLINE,
+	MLX5E_SQ_STATE_PENDING_XSK_TX,
+	MLX5E_SQ_STATE_PENDING_TLS_RX_RESYNC,
+	MLX5E_SQ_STATE_XDP_MULTIBUF,
+};
+
+struct mlx5e_tx_mpwqe {
+	/* Current MPWQE session */
+	struct mlx5e_tx_wqe *wqe;
+	u32 bytes_count;
+	u8 ds_count;
+	u8 pkt_count;
+	u8 inline_on;
+};
+
+struct mlx5e_skb_fifo {
+	struct sk_buff **fifo;
+	u16 *pc;
+	u16 *cc;
+	u16 mask;
+};
+
+struct mlx5e_ptpsq;
+
+struct mlx5e_txqsq {
+	/* data path */
+
+	/* dirtied @completion */
+	u16                        cc;
+	u16                        skb_fifo_cc;
+	u32                        dma_fifo_cc;
+	struct dim                 dim; /* Adaptive Moderation */
+
+	/* dirtied @xmit */
+	u16                        pc ____cacheline_aligned_in_smp;
+	u16                        skb_fifo_pc;
+	u32                        dma_fifo_pc;
+	struct mlx5e_tx_mpwqe      mpwqe;
+
+	struct mlx5e_cq            cq;
+
+	/* read only */
+	struct mlx5_wq_cyc         wq;
+	u32                        dma_fifo_mask;
+	struct mlx5e_sq_stats     *stats;
+	struct {
+		struct mlx5e_sq_dma       *dma_fifo;
+		struct mlx5e_skb_fifo      skb_fifo;
+		struct mlx5e_tx_wqe_info  *wqe_info;
+	} db;
+	void __iomem              *uar_map;
+	struct netdev_queue       *txq;
+	u32                        sqn;
+	u16                        stop_room;
+	u8                         max_sq_mpw_wqebbs;
+	u8                         min_inline_mode;
+	struct device             *pdev;
+	__be32                     mkey_be;
+	unsigned long              state;
+	unsigned int               hw_mtu;
+	struct mlx5_clock         *clock;
+	struct net_device         *netdev;
+	struct mlx5_core_dev      *mdev;
+	struct mlx5e_priv         *priv;
+
+	/* control path */
+	struct mlx5_wq_ctrl        wq_ctrl;
+	int                        ch_ix;
+	int                        txq_ix;
+	u32                        rate_limit;
+	struct work_struct         recover_work;
+	struct mlx5e_ptpsq        *ptpsq;
+	cqe_ts_to_ns               ptp_cyc2time;
+} ____cacheline_aligned_in_smp;
+
+union mlx5e_alloc_unit {
+	struct page *page;
+	struct xdp_buff *xsk;
+};
+
+/* XDP packets can be transmitted in different ways. On completion, we need to
+ * distinguish between them to clean up things in a proper way.
+ */
+enum mlx5e_xdp_xmit_mode {
+	/* An xdp_frame was transmitted due to either XDP_REDIRECT from another
+	 * device or XDP_TX from an XSK RQ. The frame has to be unmapped and
+	 * returned.
+	 */
+	MLX5E_XDP_XMIT_MODE_FRAME,
+
+	/* The xdp_frame was created in place as a result of XDP_TX from a
+	 * regular RQ. No DMA remapping happened, and the page belongs to us.
+	 */
+	MLX5E_XDP_XMIT_MODE_PAGE,
+
+	/* No xdp_frame was created at all, the transmit happened from a UMEM
+	 * page. The UMEM Completion Ring producer pointer has to be increased.
+	 */
+	MLX5E_XDP_XMIT_MODE_XSK,
+};
+
+struct mlx5e_xdp_info {
+	enum mlx5e_xdp_xmit_mode mode;
+	union {
+		struct {
+			struct xdp_frame *xdpf;
+			dma_addr_t dma_addr;
+		} frame;
+		struct {
+			struct mlx5e_rq *rq;
+			struct page *page;
+		} page;
+	};
+};
+
+struct mlx5e_xmit_data {
+	dma_addr_t  dma_addr;
+	void       *data;
+	u32         len;
+};
+
+struct mlx5e_xdp_info_fifo {
+	struct mlx5e_xdp_info *xi;
+	u32 *cc;
+	u32 *pc;
+	u32 mask;
+};
+
+struct mlx5e_xdpsq;
+typedef int (*mlx5e_fp_xmit_xdp_frame_check)(struct mlx5e_xdpsq *);
+typedef bool (*mlx5e_fp_xmit_xdp_frame)(struct mlx5e_xdpsq *,
+					struct mlx5e_xmit_data *,
+					struct skb_shared_info *,
+					int);
+
+struct mlx5e_xdpsq {
+	/* data path */
+
+	/* dirtied @completion */
+	u32                        xdpi_fifo_cc;
+	u16                        cc;
+
+	/* dirtied @xmit */
+	u32                        xdpi_fifo_pc ____cacheline_aligned_in_smp;
+	u16                        pc;
+	struct mlx5_wqe_ctrl_seg   *doorbell_cseg;
+	struct mlx5e_tx_mpwqe      mpwqe;
+
+	struct mlx5e_cq            cq;
+
+	/* read only */
+	struct xsk_buff_pool      *xsk_pool;
+	struct mlx5_wq_cyc         wq;
+	struct mlx5e_xdpsq_stats  *stats;
+	mlx5e_fp_xmit_xdp_frame_check xmit_xdp_frame_check;
+	mlx5e_fp_xmit_xdp_frame    xmit_xdp_frame;
+	struct {
+		struct mlx5e_xdp_wqe_info *wqe_info;
+		struct mlx5e_xdp_info_fifo xdpi_fifo;
+	} db;
+	void __iomem              *uar_map;
+	u32                        sqn;
+	struct device             *pdev;
+	__be32                     mkey_be;
+	u16                        stop_room;
+	u8                         max_sq_mpw_wqebbs;
+	u8                         min_inline_mode;
+	unsigned long              state;
+	unsigned int               hw_mtu;
+
+	/* control path */
+	struct mlx5_wq_ctrl        wq_ctrl;
+	struct mlx5e_channel      *channel;
+} ____cacheline_aligned_in_smp;
+
+struct mlx5e_ktls_resync_resp;
+
+struct mlx5e_icosq {
+	/* data path */
+	u16                        cc;
+	u16                        pc;
+
+	struct mlx5_wqe_ctrl_seg  *doorbell_cseg;
+	struct mlx5e_cq            cq;
+
+	/* write@xmit, read@completion */
+	struct {
+		struct mlx5e_icosq_wqe_info *wqe_info;
+	} db;
+
+	/* read only */
+	struct mlx5_wq_cyc         wq;
+	void __iomem              *uar_map;
+	u32                        sqn;
+	u16                        reserved_room;
+	unsigned long              state;
+	struct mlx5e_ktls_resync_resp *ktls_resync;
+
+	/* control path */
+	struct mlx5_wq_ctrl        wq_ctrl;
+	struct mlx5e_channel      *channel;
+
+	struct work_struct         recover_work;
+} ____cacheline_aligned_in_smp;
+
+struct mlx5e_wqe_frag_info {
+	union mlx5e_alloc_unit *au;
+	u32 offset;
+	bool last_in_page;
+};
+
+struct mlx5e_mpw_info {
+	u16 consumed_strides;
+	DECLARE_BITMAP(xdp_xmit_bitmap, MLX5_MPWRQ_MAX_PAGES_PER_WQE);
+	union mlx5e_alloc_unit alloc_units[];
+};
+
+#define MLX5E_MAX_RX_FRAGS 4
+
+/* a single cache unit is capable to serve one napi call (for non-striding rq)
+ * or a MPWQE (for striding rq).
+ */
+#define MLX5E_CACHE_UNIT (MLX5_MPWRQ_MAX_PAGES_PER_WQE > NAPI_POLL_WEIGHT ? \
+			  MLX5_MPWRQ_MAX_PAGES_PER_WQE : NAPI_POLL_WEIGHT)
+#define MLX5E_CACHE_SIZE	(4 * roundup_pow_of_two(MLX5E_CACHE_UNIT))
+struct mlx5e_page_cache {
+	u32 head;
+	u32 tail;
+	struct page *page_cache[MLX5E_CACHE_SIZE];
+};
+
+struct mlx5e_rq;
+typedef void (*mlx5e_fp_handle_rx_cqe)(struct mlx5e_rq*, struct mlx5_cqe64*);
+typedef struct sk_buff *
+(*mlx5e_fp_skb_from_cqe_mpwrq)(struct mlx5e_rq *rq, struct mlx5e_mpw_info *wi,
+			       u16 cqe_bcnt, u32 head_offset, u32 page_idx);
+typedef struct sk_buff *
+(*mlx5e_fp_skb_from_cqe)(struct mlx5e_rq *rq, struct mlx5e_wqe_frag_info *wi,
+			 u32 cqe_bcnt);
+typedef bool (*mlx5e_fp_post_rx_wqes)(struct mlx5e_rq *rq);
+typedef void (*mlx5e_fp_dealloc_wqe)(struct mlx5e_rq*, u16);
+typedef void (*mlx5e_fp_shampo_dealloc_hd)(struct mlx5e_rq*, u16, u16, bool);
+
+int mlx5e_rq_set_handlers(struct mlx5e_rq *rq, struct mlx5e_params *params, bool xsk);
+void mlx5e_rq_set_trap_handlers(struct mlx5e_rq *rq, struct mlx5e_params *params);
+
+enum mlx5e_rq_flag {
+	MLX5E_RQ_FLAG_XDP_XMIT,
+	MLX5E_RQ_FLAG_XDP_REDIRECT,
+};
+
+struct mlx5e_rq_frag_info {
+	int frag_size;
+	int frag_stride;
+};
+
+struct mlx5e_rq_frags_info {
+	struct mlx5e_rq_frag_info arr[MLX5E_MAX_RX_FRAGS];
+	u8 num_frags;
+	u8 log_num_frags;
+	u8 wqe_bulk;
+	u8 wqe_index_mask;
+};
+
+struct mlx5e_dma_info {
+	dma_addr_t addr;
+	struct page *page;
+};
+
+struct mlx5e_shampo_hd {
+	u32 mkey;
+	struct mlx5e_dma_info *info;
+	struct page *last_page;
+	u16 hd_per_wq;
+	u16 hd_per_wqe;
+	unsigned long *bitmap;
+	u16 pi;
+	u16 ci;
+	__be32 key;
+	u64 last_addr;
+};
+
+struct mlx5e_hw_gro_data {
+	struct sk_buff *skb;
+	struct flow_keys fk;
+	int second_ip_id;
+};
+
+enum mlx5e_mpwrq_umr_mode {
+	MLX5E_MPWRQ_UMR_MODE_ALIGNED,
+	MLX5E_MPWRQ_UMR_MODE_UNALIGNED,
+	MLX5E_MPWRQ_UMR_MODE_OVERSIZED,
+	MLX5E_MPWRQ_UMR_MODE_TRIPLE,
+};
+
+struct mlx5e_rq {
+	/* data path */
+	union {
+		struct {
+			struct mlx5_wq_cyc          wq;
+			struct mlx5e_wqe_frag_info *frags;
+			union mlx5e_alloc_unit     *alloc_units;
+			struct mlx5e_rq_frags_info  info;
+			mlx5e_fp_skb_from_cqe       skb_from_cqe;
+		} wqe;
+		struct {
+			struct mlx5_wq_ll      wq;
+			struct mlx5e_umr_wqe   umr_wqe;
+			struct mlx5e_mpw_info *info;
+			mlx5e_fp_skb_from_cqe_mpwrq skb_from_cqe_mpwrq;
+			__be32                 umr_mkey_be;
+			u16                    num_strides;
+			u16                    actual_wq_head;
+			u8                     log_stride_sz;
+			u8                     umr_in_progress;
+			u8                     umr_last_bulk;
+			u8                     umr_completed;
+			u8                     min_wqe_bulk;
+			u8                     page_shift;
+			u8                     pages_per_wqe;
+			u8                     umr_wqebbs;
+			u8                     mtts_per_wqe;
+			u8                     umr_mode;
+			struct mlx5e_shampo_hd *shampo;
+		} mpwqe;
+	};
+	struct {
+		u16            headroom;
+		u32            frame0_sz;
+		u8             map_dir;   /* dma map direction */
+	} buff;
+
+	struct device         *pdev;
+	struct net_device     *netdev;
+	struct mlx5e_rq_stats *stats;
+	struct mlx5e_cq        cq;
+	struct mlx5e_cq_decomp cqd;
+	struct mlx5e_page_cache page_cache;
+	struct hwtstamp_config *tstamp;
+	struct mlx5_clock      *clock;
+	struct mlx5e_icosq    *icosq;
+	struct mlx5e_priv     *priv;
+
+	struct mlx5e_hw_gro_data *hw_gro_data;
+
+	mlx5e_fp_handle_rx_cqe handle_rx_cqe;
+	mlx5e_fp_post_rx_wqes  post_wqes;
+	mlx5e_fp_dealloc_wqe   dealloc_wqe;
+
+	unsigned long          state;
+	int                    ix;
+	unsigned int           hw_mtu;
+
+	struct dim         dim; /* Dynamic Interrupt Moderation */
+
+	/* XDP */
+	struct bpf_prog __rcu *xdp_prog;
+	struct mlx5e_xdpsq    *xdpsq;
+	DECLARE_BITMAP(flags, 8);
+	struct page_pool      *page_pool;
+
+	/* AF_XDP zero-copy */
+	struct xsk_buff_pool  *xsk_pool;
+
+	struct work_struct     recover_work;
+
+	/* control */
+	struct mlx5_wq_ctrl    wq_ctrl;
+	__be32                 mkey_be;
+	u8                     wq_type;
+	u32                    rqn;
+	struct mlx5_core_dev  *mdev;
+	struct mlx5e_channel  *channel;
+	struct mlx5e_dma_info  wqe_overflow;
+
+	/* XDP read-mostly */
+	struct xdp_rxq_info    xdp_rxq;
+	cqe_ts_to_ns           ptp_cyc2time;
+} ____cacheline_aligned_in_smp;
+
+enum mlx5e_channel_state {
+	MLX5E_CHANNEL_STATE_XSK,
+	MLX5E_CHANNEL_NUM_STATES
+};
+
+struct mlx5e_channel {
+	/* data path */
+	struct mlx5e_rq            rq;
+	struct mlx5e_xdpsq         rq_xdpsq;
+	struct mlx5e_txqsq         sq[MLX5E_MAX_NUM_TC];
+	struct mlx5e_icosq         icosq;   /* internal control operations */
+	struct mlx5e_txqsq __rcu * __rcu *qos_sqs;
+	bool                       xdp;
+	struct napi_struct         napi;
+	struct device             *pdev;
+	struct net_device         *netdev;
+	__be32                     mkey_be;
+	u16                        qos_sqs_size;
+	u8                         num_tc;
+	u8                         lag_port;
+
+	/* XDP_REDIRECT */
+	struct mlx5e_xdpsq         xdpsq;
+
+	/* AF_XDP zero-copy */
+	struct mlx5e_rq            xskrq;
+	struct mlx5e_xdpsq         xsksq;
+
+	/* Async ICOSQ */
+	struct mlx5e_icosq         async_icosq;
+	/* async_icosq can be accessed from any CPU - the spinlock protects it. */
+	spinlock_t                 async_icosq_lock;
+
+	/* data path - accessed per napi poll */
+	const struct cpumask	  *aff_mask;
+	struct mlx5e_ch_stats     *stats;
+
+	/* control */
+	struct mlx5e_priv         *priv;
+	struct mlx5_core_dev      *mdev;
+	struct hwtstamp_config    *tstamp;
+	DECLARE_BITMAP(state, MLX5E_CHANNEL_NUM_STATES);
+	int                        ix;
+	int                        cpu;
+	/* Sync between icosq recovery and XSK enable/disable. */
+	struct mutex               icosq_recovery_lock;
+};
+
+struct mlx5e_ptp;
+
+struct mlx5e_channels {
+	struct mlx5e_channel **c;
+	struct mlx5e_ptp      *ptp;
+	unsigned int           num;
+	struct mlx5e_params    params;
+};
+
+struct mlx5e_channel_stats {
+	struct mlx5e_ch_stats ch;
+	struct mlx5e_sq_stats sq[MLX5E_MAX_NUM_TC];
+	struct mlx5e_rq_stats rq;
+	struct mlx5e_rq_stats xskrq;
+	struct mlx5e_xdpsq_stats rq_xdpsq;
+	struct mlx5e_xdpsq_stats xdpsq;
+	struct mlx5e_xdpsq_stats xsksq;
+} ____cacheline_aligned_in_smp;
+
+struct mlx5e_ptp_stats {
+	struct mlx5e_ch_stats ch;
+	struct mlx5e_sq_stats sq[MLX5E_MAX_NUM_TC];
+	struct mlx5e_ptp_cq_stats cq[MLX5E_MAX_NUM_TC];
+	struct mlx5e_rq_stats rq;
+} ____cacheline_aligned_in_smp;
+
+enum {
+	MLX5E_STATE_OPENED,
+	MLX5E_STATE_DESTROYING,
+	MLX5E_STATE_XDP_TX_ENABLED,
+	MLX5E_STATE_XDP_ACTIVE,
+	MLX5E_STATE_CHANNELS_ACTIVE,
+};
+
+struct mlx5e_modify_sq_param {
+	int curr_state;
+	int next_state;
+	int rl_update;
+	int rl_index;
+	bool qos_update;
+	u16 qos_queue_group_id;
+};
+
+#if IS_ENABLED(CONFIG_PCI_HYPERV_INTERFACE)
+struct mlx5e_hv_vhca_stats_agent {
+	struct mlx5_hv_vhca_agent *agent;
+	struct delayed_work        work;
+	u16                        delay;
+	void                      *buf;
+};
+#endif
+
+struct mlx5e_xsk {
+	/* XSK buffer pools are stored separately from channels,
+	 * because we don't want to lose them when channels are
+	 * recreated. The kernel also stores buffer pool, but it doesn't
+	 * distinguish between zero-copy and non-zero-copy UMEMs, so
+	 * rely on our mechanism.
+	 */
+	struct xsk_buff_pool **pools;
+	u16 refcnt;
+	bool ever_used;
+};
+
+/* Temporary storage for variables that are allocated when struct mlx5e_priv is
+ * initialized, and used where we can't allocate them because that functions
+ * must not fail. Use with care and make sure the same variable is not used
+ * simultaneously by multiple users.
+ */
+struct mlx5e_scratchpad {
+	cpumask_var_t cpumask;
+};
+
+struct mlx5e_trap;
+struct mlx5e_htb;
+
+struct mlx5e_priv {
+	/* priv data path fields - start */
+	struct mlx5e_selq selq;
+	struct mlx5e_txqsq **txq2sq;
+#ifdef CONFIG_MLX5_CORE_EN_DCB
+	struct mlx5e_dcbx_dp       dcbx_dp;
+#endif
+	/* priv data path fields - end */
+
+	u32                        msglevel;
+	unsigned long              state;
+	struct mutex               state_lock; /* Protects Interface state */
+	struct mlx5e_rq            drop_rq;
+
+	struct mlx5e_channels      channels;
+	u32                        tisn[MLX5_MAX_PORTS][MLX5E_MAX_NUM_TC];
+	struct mlx5e_rx_res       *rx_res;
+	u32                       *tx_rates;
+
+	struct mlx5e_flow_steering *fs;
+
+	struct workqueue_struct    *wq;
+	struct work_struct         update_carrier_work;
+	struct work_struct         set_rx_mode_work;
+	struct work_struct         tx_timeout_work;
+	struct work_struct         update_stats_work;
+	struct work_struct         monitor_counters_work;
+	struct mlx5_nb             monitor_counters_nb;
+
+	struct mlx5_core_dev      *mdev;
+	struct net_device         *netdev;
+	struct mlx5e_trap         *en_trap;
+	struct mlx5e_stats         stats;
+	struct mlx5e_channel_stats **channel_stats;
+	struct mlx5e_channel_stats trap_stats;
+	struct mlx5e_ptp_stats     ptp_stats;
+	struct mlx5e_sq_stats      **htb_qos_sq_stats;
+	u16                        htb_max_qos_sqs;
+	u16                        stats_nch;
+	u16                        max_nch;
+	u8                         max_opened_tc;
+	bool                       tx_ptp_opened;
+	bool                       rx_ptp_opened;
+	struct hwtstamp_config     tstamp;
+	u16                        q_counter;
+	u16                        drop_rq_q_counter;
+	struct notifier_block      events_nb;
+	struct notifier_block      blocking_events_nb;
+
+	struct udp_tunnel_nic_info nic_info;
+#ifdef CONFIG_MLX5_CORE_EN_DCB
+	struct mlx5e_dcbx          dcbx;
+#endif
+
+	const struct mlx5e_profile *profile;
+	void                      *ppriv;
+#ifdef CONFIG_MLX5_EN_MACSEC
+	struct mlx5e_macsec       *macsec;
+#endif
+#ifdef CONFIG_MLX5_EN_IPSEC
+	struct mlx5e_ipsec        *ipsec;
+#endif
+#ifdef CONFIG_MLX5_EN_TLS
+	struct mlx5e_tls          *tls;
+#endif
+	struct devlink_health_reporter *tx_reporter;
+	struct devlink_health_reporter *rx_reporter;
+	struct mlx5e_xsk           xsk;
+#if IS_ENABLED(CONFIG_PCI_HYPERV_INTERFACE)
+	struct mlx5e_hv_vhca_stats_agent stats_agent;
+#endif
+	struct mlx5e_scratchpad    scratchpad;
+	struct mlx5e_htb          *htb;
+	struct mlx5e_mqprio_rl    *mqprio_rl;
+};
+
+struct mlx5e_rx_handlers {
+	mlx5e_fp_handle_rx_cqe handle_rx_cqe;
+	mlx5e_fp_handle_rx_cqe handle_rx_cqe_mpwqe;
+	mlx5e_fp_handle_rx_cqe handle_rx_cqe_mpwqe_shampo;
+};
+
+extern const struct mlx5e_rx_handlers mlx5e_rx_handlers_nic;
+
+enum mlx5e_profile_feature {
+	MLX5E_PROFILE_FEATURE_PTP_RX,
+	MLX5E_PROFILE_FEATURE_PTP_TX,
+	MLX5E_PROFILE_FEATURE_QOS_HTB,
+	MLX5E_PROFILE_FEATURE_FS_VLAN,
+	MLX5E_PROFILE_FEATURE_FS_TC,
+};
+
+struct mlx5e_profile {
+	int	(*init)(struct mlx5_core_dev *mdev,
+			struct net_device *netdev);
+	void	(*cleanup)(struct mlx5e_priv *priv);
+	int	(*init_rx)(struct mlx5e_priv *priv);
+	void	(*cleanup_rx)(struct mlx5e_priv *priv);
+	int	(*init_tx)(struct mlx5e_priv *priv);
+	void	(*cleanup_tx)(struct mlx5e_priv *priv);
+	void	(*enable)(struct mlx5e_priv *priv);
+	void	(*disable)(struct mlx5e_priv *priv);
+	int	(*update_rx)(struct mlx5e_priv *priv);
+	void	(*update_stats)(struct mlx5e_priv *priv);
+	void	(*update_carrier)(struct mlx5e_priv *priv);
+	int	(*max_nch_limit)(struct mlx5_core_dev *mdev);
+	unsigned int (*stats_grps_num)(struct mlx5e_priv *priv);
+	mlx5e_stats_grp_t *stats_grps;
+	const struct mlx5e_rx_handlers *rx_handlers;
+	int	max_tc;
+	u32     features;
+};
+
+#define mlx5e_profile_feature_cap(profile, feature)	\
+	((profile)->features & BIT(MLX5E_PROFILE_FEATURE_##feature))
+
+void mlx5e_build_ptys2ethtool_map(void);
+
+bool mlx5e_check_fragmented_striding_rq_cap(struct mlx5_core_dev *mdev, u8 page_shift,
+					    enum mlx5e_mpwrq_umr_mode umr_mode);
+
+void mlx5e_shampo_dealloc_hd(struct mlx5e_rq *rq, u16 len, u16 start, bool close);
+void mlx5e_get_stats(struct net_device *dev, struct rtnl_link_stats64 *stats);
+void mlx5e_fold_sw_stats64(struct mlx5e_priv *priv, struct rtnl_link_stats64 *s);
+
+int mlx5e_self_test_num(struct mlx5e_priv *priv);
+int mlx5e_self_test_fill_strings(struct mlx5e_priv *priv, u8 *data);
+void mlx5e_self_test(struct net_device *ndev, struct ethtool_test *etest,
+		     u64 *buf);
+void mlx5e_set_rx_mode_work(struct work_struct *work);
+
+int mlx5e_hwstamp_set(struct mlx5e_priv *priv, struct ifreq *ifr);
+int mlx5e_hwstamp_get(struct mlx5e_priv *priv, struct ifreq *ifr);
+int mlx5e_modify_rx_cqe_compression_locked(struct mlx5e_priv *priv, bool val, bool rx_filter);
+
+int mlx5e_vlan_rx_add_vid(struct net_device *dev, __always_unused __be16 proto,
+			  u16 vid);
+int mlx5e_vlan_rx_kill_vid(struct net_device *dev, __always_unused __be16 proto,
+			   u16 vid);
+void mlx5e_timestamp_init(struct mlx5e_priv *priv);
+
+struct mlx5e_xsk_param;
+
+struct mlx5e_rq_param;
+int mlx5e_open_rq(struct mlx5e_params *params, struct mlx5e_rq_param *param,
+		  struct mlx5e_xsk_param *xsk, int node,
+		  struct mlx5e_rq *rq);
+#define MLX5E_RQ_WQES_TIMEOUT 20000 /* msecs */
+int mlx5e_wait_for_min_rx_wqes(struct mlx5e_rq *rq, int wait_time);
+void mlx5e_close_rq(struct mlx5e_rq *rq);
+int mlx5e_create_rq(struct mlx5e_rq *rq, struct mlx5e_rq_param *param);
+void mlx5e_destroy_rq(struct mlx5e_rq *rq);
+
+struct mlx5e_sq_param;
+int mlx5e_open_xdpsq(struct mlx5e_channel *c, struct mlx5e_params *params,
+		     struct mlx5e_sq_param *param, struct xsk_buff_pool *xsk_pool,
+		     struct mlx5e_xdpsq *sq, bool is_redirect);
+void mlx5e_close_xdpsq(struct mlx5e_xdpsq *sq);
+
+struct mlx5e_create_cq_param {
+	struct napi_struct *napi;
+	struct mlx5e_ch_stats *ch_stats;
+	int node;
+	int ix;
+};
+
+struct mlx5e_cq_param;
+int mlx5e_open_cq(struct mlx5e_priv *priv, struct dim_cq_moder moder,
+		  struct mlx5e_cq_param *param, struct mlx5e_create_cq_param *ccp,
+		  struct mlx5e_cq *cq);
+void mlx5e_close_cq(struct mlx5e_cq *cq);
+
+int mlx5e_open_locked(struct net_device *netdev);
+int mlx5e_close_locked(struct net_device *netdev);
+
+void mlx5e_trigger_napi_icosq(struct mlx5e_channel *c);
+void mlx5e_trigger_napi_sched(struct napi_struct *napi);
+
+int mlx5e_open_channels(struct mlx5e_priv *priv,
+			struct mlx5e_channels *chs);
+void mlx5e_close_channels(struct mlx5e_channels *chs);
+
+/* Function pointer to be used to modify HW or kernel settings while
+ * switching channels
+ */
+typedef int (*mlx5e_fp_preactivate)(struct mlx5e_priv *priv, void *context);
+#define MLX5E_DEFINE_PREACTIVATE_WRAPPER_CTX(fn) \
+int fn##_ctx(struct mlx5e_priv *priv, void *context) \
+{ \
+	return fn(priv); \
+}
+int mlx5e_safe_reopen_channels(struct mlx5e_priv *priv);
+int mlx5e_safe_switch_params(struct mlx5e_priv *priv,
+			     struct mlx5e_params *new_params,
+			     mlx5e_fp_preactivate preactivate,
+			     void *context, bool reset);
+int mlx5e_update_tx_netdev_queues(struct mlx5e_priv *priv);
+int mlx5e_num_channels_changed_ctx(struct mlx5e_priv *priv, void *context);
+void mlx5e_activate_priv_channels(struct mlx5e_priv *priv);
+void mlx5e_deactivate_priv_channels(struct mlx5e_priv *priv);
+int mlx5e_ptp_rx_manage_fs_ctx(struct mlx5e_priv *priv, void *ctx);
+
+int mlx5e_flush_rq(struct mlx5e_rq *rq, int curr_state);
+void mlx5e_activate_rq(struct mlx5e_rq *rq);
+void mlx5e_deactivate_rq(struct mlx5e_rq *rq);
+void mlx5e_activate_icosq(struct mlx5e_icosq *icosq);
+void mlx5e_deactivate_icosq(struct mlx5e_icosq *icosq);
+
+int mlx5e_modify_sq(struct mlx5_core_dev *mdev, u32 sqn,
+		    struct mlx5e_modify_sq_param *p);
+int mlx5e_open_txqsq(struct mlx5e_channel *c, u32 tisn, int txq_ix,
+		     struct mlx5e_params *params, struct mlx5e_sq_param *param,
+		     struct mlx5e_txqsq *sq, int tc, u16 qos_queue_group_id,
+		     struct mlx5e_sq_stats *sq_stats);
+void mlx5e_activate_txqsq(struct mlx5e_txqsq *sq);
+void mlx5e_deactivate_txqsq(struct mlx5e_txqsq *sq);
+void mlx5e_free_txqsq(struct mlx5e_txqsq *sq);
+void mlx5e_tx_disable_queue(struct netdev_queue *txq);
+int mlx5e_alloc_txqsq_db(struct mlx5e_txqsq *sq, int numa);
+void mlx5e_free_txqsq_db(struct mlx5e_txqsq *sq);
+struct mlx5e_create_sq_param;
+int mlx5e_create_sq_rdy(struct mlx5_core_dev *mdev,
+			struct mlx5e_sq_param *param,
+			struct mlx5e_create_sq_param *csp,
+			u16 qos_queue_group_id,
+			u32 *sqn);
+void mlx5e_tx_err_cqe_work(struct work_struct *recover_work);
+void mlx5e_close_txqsq(struct mlx5e_txqsq *sq);
+
+static inline bool mlx5_tx_swp_supported(struct mlx5_core_dev *mdev)
+{
+	return MLX5_CAP_ETH(mdev, swp) &&
+		MLX5_CAP_ETH(mdev, swp_csum) && MLX5_CAP_ETH(mdev, swp_lso);
+}
+
+extern const struct ethtool_ops mlx5e_ethtool_ops;
+
+int mlx5e_create_mkey(struct mlx5_core_dev *mdev, u32 pdn, u32 *mkey);
+int mlx5e_create_mdev_resources(struct mlx5_core_dev *mdev);
+void mlx5e_destroy_mdev_resources(struct mlx5_core_dev *mdev);
+int mlx5e_refresh_tirs(struct mlx5e_priv *priv, bool enable_uc_lb,
+		       bool enable_mc_lb);
+void mlx5e_mkey_set_relaxed_ordering(struct mlx5_core_dev *mdev, void *mkc);
+
+/* common netdev helpers */
+void mlx5e_create_q_counters(struct mlx5e_priv *priv);
+void mlx5e_destroy_q_counters(struct mlx5e_priv *priv);
+int mlx5e_open_drop_rq(struct mlx5e_priv *priv,
+		       struct mlx5e_rq *drop_rq);
+void mlx5e_close_drop_rq(struct mlx5e_rq *drop_rq);
+
+int mlx5e_create_tis(struct mlx5_core_dev *mdev, void *in, u32 *tisn);
+void mlx5e_destroy_tis(struct mlx5_core_dev *mdev, u32 tisn);
+
+int mlx5e_create_tises(struct mlx5e_priv *priv);
+void mlx5e_destroy_tises(struct mlx5e_priv *priv);
+int mlx5e_update_nic_rx(struct mlx5e_priv *priv);
+void mlx5e_update_carrier(struct mlx5e_priv *priv);
+int mlx5e_close(struct net_device *netdev);
+int mlx5e_open(struct net_device *netdev);
+
+void mlx5e_queue_update_stats(struct mlx5e_priv *priv);
+
+int mlx5e_set_dev_port_mtu(struct mlx5e_priv *priv);
+int mlx5e_set_dev_port_mtu_ctx(struct mlx5e_priv *priv, void *context);
+int mlx5e_change_mtu(struct net_device *netdev, int new_mtu,
+		     mlx5e_fp_preactivate preactivate);
+void mlx5e_vxlan_set_netdev_info(struct mlx5e_priv *priv);
+
+/* ethtool helpers */
+void mlx5e_ethtool_get_drvinfo(struct mlx5e_priv *priv,
+			       struct ethtool_drvinfo *drvinfo);
+void mlx5e_ethtool_get_strings(struct mlx5e_priv *priv,
+			       uint32_t stringset, uint8_t *data);
+int mlx5e_ethtool_get_sset_count(struct mlx5e_priv *priv, int sset);
+void mlx5e_ethtool_get_ethtool_stats(struct mlx5e_priv *priv,
+				     struct ethtool_stats *stats, u64 *data);
+void mlx5e_ethtool_get_ringparam(struct mlx5e_priv *priv,
+				 struct ethtool_ringparam *param,
+				 struct kernel_ethtool_ringparam *kernel_param);
+int mlx5e_ethtool_set_ringparam(struct mlx5e_priv *priv,
+				struct ethtool_ringparam *param);
+void mlx5e_ethtool_get_channels(struct mlx5e_priv *priv,
+				struct ethtool_channels *ch);
+int mlx5e_ethtool_set_channels(struct mlx5e_priv *priv,
+			       struct ethtool_channels *ch);
+int mlx5e_ethtool_get_coalesce(struct mlx5e_priv *priv,
+			       struct ethtool_coalesce *coal,
+			       struct kernel_ethtool_coalesce *kernel_coal);
+int mlx5e_ethtool_set_coalesce(struct mlx5e_priv *priv,
+			       struct ethtool_coalesce *coal,
+			       struct kernel_ethtool_coalesce *kernel_coal,
+			       struct netlink_ext_ack *extack);
+int mlx5e_ethtool_get_link_ksettings(struct mlx5e_priv *priv,
+				     struct ethtool_link_ksettings *link_ksettings);
+int mlx5e_ethtool_set_link_ksettings(struct mlx5e_priv *priv,
+				     const struct ethtool_link_ksettings *link_ksettings);
+int mlx5e_get_rxfh(struct net_device *netdev, u32 *indir, u8 *key, u8 *hfunc);
+int mlx5e_set_rxfh(struct net_device *dev, const u32 *indir, const u8 *key,
+		   const u8 hfunc);
+int mlx5e_get_rxnfc(struct net_device *dev, struct ethtool_rxnfc *info,
+		    u32 *rule_locs);
+int mlx5e_set_rxnfc(struct net_device *dev, struct ethtool_rxnfc *cmd);
+u32 mlx5e_ethtool_get_rxfh_key_size(struct mlx5e_priv *priv);
+u32 mlx5e_ethtool_get_rxfh_indir_size(struct mlx5e_priv *priv);
+int mlx5e_ethtool_get_ts_info(struct mlx5e_priv *priv,
+			      struct ethtool_ts_info *info);
+int mlx5e_ethtool_flash_device(struct mlx5e_priv *priv,
+			       struct ethtool_flash *flash);
+void mlx5e_ethtool_get_pauseparam(struct mlx5e_priv *priv,
+				  struct ethtool_pauseparam *pauseparam);
+int mlx5e_ethtool_set_pauseparam(struct mlx5e_priv *priv,
+				 struct ethtool_pauseparam *pauseparam);
+
+/* mlx5e generic netdev management API */
+static inline bool
+mlx5e_tx_mpwqe_supported(struct mlx5_core_dev *mdev)
+{
+	return !is_kdump_kernel() &&
+		MLX5_CAP_ETH(mdev, enhanced_multi_pkt_send_wqe);
+}
+
+int mlx5e_get_pf_num_tirs(struct mlx5_core_dev *mdev);
+int mlx5e_priv_init(struct mlx5e_priv *priv,
+		    const struct mlx5e_profile *profile,
+		    struct net_device *netdev,
+		    struct mlx5_core_dev *mdev);
+void mlx5e_priv_cleanup(struct mlx5e_priv *priv);
+struct net_device *
+mlx5e_create_netdev(struct mlx5_core_dev *mdev, const struct mlx5e_profile *profile);
+int mlx5e_attach_netdev(struct mlx5e_priv *priv);
+void mlx5e_detach_netdev(struct mlx5e_priv *priv);
+void mlx5e_destroy_netdev(struct mlx5e_priv *priv);
+int mlx5e_netdev_change_profile(struct mlx5e_priv *priv,
+				const struct mlx5e_profile *new_profile, void *new_ppriv);
+void mlx5e_netdev_attach_nic_profile(struct mlx5e_priv *priv);
+void mlx5e_set_netdev_mtu_boundaries(struct mlx5e_priv *priv);
+void mlx5e_build_nic_params(struct mlx5e_priv *priv, struct mlx5e_xsk *xsk, u16 mtu);
+void mlx5e_rx_dim_work(struct work_struct *work);
+void mlx5e_tx_dim_work(struct work_struct *work);
+
+netdev_features_t mlx5e_features_check(struct sk_buff *skb,
+				       struct net_device *netdev,
+				       netdev_features_t features);
+int mlx5e_set_features(struct net_device *netdev, netdev_features_t features);
+#ifdef CONFIG_MLX5_ESWITCH
+int mlx5e_set_vf_mac(struct net_device *dev, int vf, u8 *mac);
+int mlx5e_set_vf_rate(struct net_device *dev, int vf, int min_tx_rate, int max_tx_rate);
+int mlx5e_get_vf_config(struct net_device *dev, int vf, struct ifla_vf_info *ivi);
+int mlx5e_get_vf_stats(struct net_device *dev, int vf, struct ifla_vf_stats *vf_stats);
+#endif
+#endif /* __MLX5_EN_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/channels.c b/drivers/net/ethernet/mellanox/mlx5/core/en/channels.c
new file mode 100644
index 000000000..48581ea3a
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/channels.c
@@ -0,0 +1,51 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2021, Mellanox Technologies inc. All rights reserved. */
+
+#include "channels.h"
+#include "en.h"
+#include "en/ptp.h"
+
+unsigned int mlx5e_channels_get_num(struct mlx5e_channels *chs)
+{
+	return chs->num;
+}
+
+static struct mlx5e_channel *mlx5e_channels_get(struct mlx5e_channels *chs, unsigned int ix)
+{
+	WARN_ON_ONCE(ix >= mlx5e_channels_get_num(chs));
+	return chs->c[ix];
+}
+
+bool mlx5e_channels_is_xsk(struct mlx5e_channels *chs, unsigned int ix)
+{
+	struct mlx5e_channel *c = mlx5e_channels_get(chs, ix);
+
+	return test_bit(MLX5E_CHANNEL_STATE_XSK, c->state);
+}
+
+void mlx5e_channels_get_regular_rqn(struct mlx5e_channels *chs, unsigned int ix, u32 *rqn)
+{
+	struct mlx5e_channel *c = mlx5e_channels_get(chs, ix);
+
+	*rqn = c->rq.rqn;
+}
+
+void mlx5e_channels_get_xsk_rqn(struct mlx5e_channels *chs, unsigned int ix, u32 *rqn)
+{
+	struct mlx5e_channel *c = mlx5e_channels_get(chs, ix);
+
+	WARN_ON_ONCE(!test_bit(MLX5E_CHANNEL_STATE_XSK, c->state));
+
+	*rqn = c->xskrq.rqn;
+}
+
+bool mlx5e_channels_get_ptp_rqn(struct mlx5e_channels *chs, u32 *rqn)
+{
+	struct mlx5e_ptp *c = chs->ptp;
+
+	if (!c || !test_bit(MLX5E_PTP_STATE_RX, c->state))
+		return false;
+
+	*rqn = c->rq.rqn;
+	return true;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/channels.h b/drivers/net/ethernet/mellanox/mlx5/core/en/channels.h
new file mode 100644
index 000000000..637ca90da
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/channels.h
@@ -0,0 +1,17 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2021, Mellanox Technologies inc. All rights reserved. */
+
+#ifndef __MLX5_EN_CHANNELS_H__
+#define __MLX5_EN_CHANNELS_H__
+
+#include <linux/kernel.h>
+
+struct mlx5e_channels;
+
+unsigned int mlx5e_channels_get_num(struct mlx5e_channels *chs);
+bool mlx5e_channels_is_xsk(struct mlx5e_channels *chs, unsigned int ix);
+void mlx5e_channels_get_regular_rqn(struct mlx5e_channels *chs, unsigned int ix, u32 *rqn);
+void mlx5e_channels_get_xsk_rqn(struct mlx5e_channels *chs, unsigned int ix, u32 *rqn);
+bool mlx5e_channels_get_ptp_rqn(struct mlx5e_channels *chs, u32 *rqn);
+
+#endif /* __MLX5_EN_CHANNELS_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/dcbnl.h b/drivers/net/ethernet/mellanox/mlx5/core/en/dcbnl.h
new file mode 100644
index 000000000..b59aee75d
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/dcbnl.h
@@ -0,0 +1,53 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2020 Mellanox Technologies. */
+
+#ifndef __MLX5E_DCBNL_H__
+#define __MLX5E_DCBNL_H__
+
+#ifdef CONFIG_MLX5_CORE_EN_DCB
+
+#define MLX5E_MAX_PRIORITY (8)
+
+struct mlx5e_cee_config {
+	/* bw pct for priority group */
+	u8                         pg_bw_pct[CEE_DCBX_MAX_PGS];
+	u8                         prio_to_pg_map[CEE_DCBX_MAX_PRIO];
+	bool                       pfc_setting[CEE_DCBX_MAX_PRIO];
+	bool                       pfc_enable;
+};
+
+struct mlx5e_dcbx {
+	enum mlx5_dcbx_oper_mode   mode;
+	struct mlx5e_cee_config    cee_cfg; /* pending configuration */
+	u8                         dscp_app_cnt;
+
+	/* The only setting that cannot be read from FW */
+	u8                         tc_tsa[IEEE_8021QAZ_MAX_TCS];
+	u8                         cap;
+
+	/* Buffer configuration */
+	bool                       manual_buffer;
+	u32                        cable_len;
+	u32                        xoff;
+	u16                        port_buff_cell_sz;
+};
+
+#define MLX5E_MAX_DSCP (64)
+
+struct mlx5e_dcbx_dp {
+	u8                         dscp2prio[MLX5E_MAX_DSCP];
+	u8                         trust_state;
+};
+
+void mlx5e_dcbnl_build_netdev(struct net_device *netdev);
+void mlx5e_dcbnl_initialize(struct mlx5e_priv *priv);
+void mlx5e_dcbnl_init_app(struct mlx5e_priv *priv);
+void mlx5e_dcbnl_delete_app(struct mlx5e_priv *priv);
+#else
+static inline void mlx5e_dcbnl_build_netdev(struct net_device *netdev) {}
+static inline void mlx5e_dcbnl_initialize(struct mlx5e_priv *priv) {}
+static inline void mlx5e_dcbnl_init_app(struct mlx5e_priv *priv) {}
+static inline void mlx5e_dcbnl_delete_app(struct mlx5e_priv *priv) {}
+#endif
+
+#endif /* __MLX5E_DCBNL_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/devlink.c b/drivers/net/ethernet/mellanox/mlx5/core/en/devlink.c
new file mode 100644
index 000000000..b69f9d10c
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/devlink.c
@@ -0,0 +1,81 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2020, Mellanox Technologies inc.  All rights reserved. */
+
+#include "en/devlink.h"
+#include "eswitch.h"
+
+static void
+mlx5e_devlink_get_port_parent_id(struct mlx5_core_dev *dev, struct netdev_phys_item_id *ppid)
+{
+	u64 parent_id;
+
+	parent_id = mlx5_query_nic_system_image_guid(dev);
+	ppid->id_len = sizeof(parent_id);
+	memcpy(ppid->id, &parent_id, sizeof(parent_id));
+}
+
+int mlx5e_devlink_port_register(struct mlx5e_priv *priv)
+{
+	struct devlink *devlink = priv_to_devlink(priv->mdev);
+	struct devlink_port_attrs attrs = {};
+	struct netdev_phys_item_id ppid = {};
+	struct devlink_port *dl_port;
+	unsigned int dl_port_index;
+	int ret;
+
+	if (mlx5_core_is_pf(priv->mdev)) {
+		attrs.flavour = DEVLINK_PORT_FLAVOUR_PHYSICAL;
+		attrs.phys.port_number = mlx5_get_dev_index(priv->mdev);
+		if (MLX5_ESWITCH_MANAGER(priv->mdev)) {
+			mlx5e_devlink_get_port_parent_id(priv->mdev, &ppid);
+			memcpy(attrs.switch_id.id, ppid.id, ppid.id_len);
+			attrs.switch_id.id_len = ppid.id_len;
+		}
+		dl_port_index = mlx5_esw_vport_to_devlink_port_index(priv->mdev,
+								     MLX5_VPORT_UPLINK);
+	} else {
+		attrs.flavour = DEVLINK_PORT_FLAVOUR_VIRTUAL;
+		dl_port_index = mlx5_esw_vport_to_devlink_port_index(priv->mdev, 0);
+	}
+
+	dl_port = mlx5e_devlink_get_dl_port(priv);
+	memset(dl_port, 0, sizeof(*dl_port));
+	devlink_port_attrs_set(dl_port, &attrs);
+
+	if (!(priv->mdev->priv.flags & MLX5_PRIV_FLAGS_MLX5E_LOCKED_FLOW))
+		devl_lock(devlink);
+	ret = devl_port_register(devlink, dl_port, dl_port_index);
+	if (!(priv->mdev->priv.flags & MLX5_PRIV_FLAGS_MLX5E_LOCKED_FLOW))
+		devl_unlock(devlink);
+
+	return ret;
+}
+
+void mlx5e_devlink_port_type_eth_set(struct mlx5e_priv *priv)
+{
+	struct devlink_port *dl_port = mlx5e_devlink_get_dl_port(priv);
+
+	devlink_port_type_eth_set(dl_port, priv->netdev);
+}
+
+void mlx5e_devlink_port_unregister(struct mlx5e_priv *priv)
+{
+	struct devlink_port *dl_port = mlx5e_devlink_get_dl_port(priv);
+	struct devlink *devlink = priv_to_devlink(priv->mdev);
+
+	if (!(priv->mdev->priv.flags & MLX5_PRIV_FLAGS_MLX5E_LOCKED_FLOW))
+		devl_lock(devlink);
+	devl_port_unregister(dl_port);
+	if (!(priv->mdev->priv.flags & MLX5_PRIV_FLAGS_MLX5E_LOCKED_FLOW))
+		devl_unlock(devlink);
+}
+
+struct devlink_port *mlx5e_get_devlink_port(struct net_device *dev)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+
+	if (!netif_device_present(dev))
+		return NULL;
+
+	return mlx5e_devlink_get_dl_port(priv);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/devlink.h b/drivers/net/ethernet/mellanox/mlx5/core/en/devlink.h
new file mode 100644
index 000000000..10b50feb9
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/devlink.h
@@ -0,0 +1,21 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2020, Mellanox Technologies inc.  All rights reserved. */
+
+#ifndef __MLX5E_EN_DEVLINK_H
+#define __MLX5E_EN_DEVLINK_H
+
+#include <net/devlink.h>
+#include "en.h"
+
+int mlx5e_devlink_port_register(struct mlx5e_priv *priv);
+void mlx5e_devlink_port_unregister(struct mlx5e_priv *priv);
+void mlx5e_devlink_port_type_eth_set(struct mlx5e_priv *priv);
+struct devlink_port *mlx5e_get_devlink_port(struct net_device *dev);
+
+static inline struct devlink_port *
+mlx5e_devlink_get_dl_port(struct mlx5e_priv *priv)
+{
+	return &priv->mdev->mlx5e_res.dl_port;
+}
+
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/fs.h b/drivers/net/ethernet/mellanox/mlx5/core/en/fs.h
new file mode 100644
index 000000000..bf2741eb7
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/fs.h
@@ -0,0 +1,204 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2018 Mellanox Technologies. */
+
+#ifndef __MLX5E_FLOW_STEER_H__
+#define __MLX5E_FLOW_STEER_H__
+
+#include "mod_hdr.h"
+#include "lib/fs_ttc.h"
+
+struct mlx5e_post_act;
+struct mlx5e_tc_table;
+
+enum {
+	MLX5E_TC_FT_LEVEL = 0,
+	MLX5E_TC_TTC_FT_LEVEL,
+	MLX5E_TC_MISS_LEVEL,
+};
+
+enum {
+	MLX5E_TC_PRIO = 0,
+	MLX5E_NIC_PRIO
+};
+
+struct mlx5e_flow_table {
+	int num_groups;
+	struct mlx5_flow_table *t;
+	struct mlx5_flow_group **g;
+};
+
+struct mlx5e_l2_rule {
+	u8  addr[ETH_ALEN + 2];
+	struct mlx5_flow_handle *rule;
+};
+
+#define MLX5E_L2_ADDR_HASH_SIZE BIT(BITS_PER_BYTE)
+
+struct mlx5e_promisc_table {
+	struct mlx5e_flow_table	ft;
+	struct mlx5_flow_handle	*rule;
+};
+
+/* Forward declaration and APIs to get private fields of vlan_table */
+struct mlx5e_vlan_table;
+unsigned long *mlx5e_vlan_get_active_svlans(struct mlx5e_vlan_table *vlan);
+struct mlx5_flow_table *mlx5e_vlan_get_flowtable(struct mlx5e_vlan_table *vlan);
+
+struct mlx5e_l2_table {
+	struct mlx5e_flow_table    ft;
+	struct hlist_head          netdev_uc[MLX5E_L2_ADDR_HASH_SIZE];
+	struct hlist_head          netdev_mc[MLX5E_L2_ADDR_HASH_SIZE];
+	struct mlx5e_l2_rule	   broadcast;
+	struct mlx5e_l2_rule	   allmulti;
+	struct mlx5_flow_handle    *trap_rule;
+	bool                       broadcast_enabled;
+	bool                       allmulti_enabled;
+	bool                       promisc_enabled;
+};
+
+#define MLX5E_NUM_INDIR_TIRS (MLX5_NUM_TT - 1)
+
+#define MLX5_HASH_IP		(MLX5_HASH_FIELD_SEL_SRC_IP   |\
+				 MLX5_HASH_FIELD_SEL_DST_IP)
+#define MLX5_HASH_IP_L4PORTS	(MLX5_HASH_FIELD_SEL_SRC_IP   |\
+				 MLX5_HASH_FIELD_SEL_DST_IP   |\
+				 MLX5_HASH_FIELD_SEL_L4_SPORT |\
+				 MLX5_HASH_FIELD_SEL_L4_DPORT)
+#define MLX5_HASH_IP_IPSEC_SPI	(MLX5_HASH_FIELD_SEL_SRC_IP   |\
+				 MLX5_HASH_FIELD_SEL_DST_IP   |\
+				 MLX5_HASH_FIELD_SEL_IPSEC_SPI)
+
+/* NIC prio FTS */
+enum {
+	MLX5E_PROMISC_FT_LEVEL,
+	MLX5E_VLAN_FT_LEVEL,
+	MLX5E_L2_FT_LEVEL,
+	MLX5E_TTC_FT_LEVEL,
+	MLX5E_INNER_TTC_FT_LEVEL,
+	MLX5E_FS_TT_UDP_FT_LEVEL = MLX5E_INNER_TTC_FT_LEVEL + 1,
+	MLX5E_FS_TT_ANY_FT_LEVEL = MLX5E_INNER_TTC_FT_LEVEL + 1,
+#ifdef CONFIG_MLX5_EN_TLS
+	MLX5E_ACCEL_FS_TCP_FT_LEVEL = MLX5E_INNER_TTC_FT_LEVEL + 1,
+#endif
+#ifdef CONFIG_MLX5_EN_ARFS
+	MLX5E_ARFS_FT_LEVEL = MLX5E_INNER_TTC_FT_LEVEL + 1,
+#endif
+#ifdef CONFIG_MLX5_EN_IPSEC
+	MLX5E_ACCEL_FS_ESP_FT_LEVEL = MLX5E_INNER_TTC_FT_LEVEL + 1,
+	MLX5E_ACCEL_FS_ESP_FT_ERR_LEVEL,
+#endif
+};
+
+struct mlx5e_flow_steering;
+struct mlx5e_rx_res;
+
+#ifdef CONFIG_MLX5_EN_ARFS
+struct mlx5e_arfs_tables;
+
+int mlx5e_arfs_create_tables(struct mlx5e_flow_steering *fs,
+			     struct mlx5e_rx_res *rx_res, bool ntuple);
+void mlx5e_arfs_destroy_tables(struct mlx5e_flow_steering *fs, bool ntuple);
+int mlx5e_arfs_enable(struct mlx5e_flow_steering *fs);
+int mlx5e_arfs_disable(struct mlx5e_flow_steering *fs);
+int mlx5e_rx_flow_steer(struct net_device *dev, const struct sk_buff *skb,
+			u16 rxq_index, u32 flow_id);
+#else
+static inline int mlx5e_arfs_create_tables(struct mlx5e_flow_steering *fs,
+					   struct mlx5e_rx_res *rx_res, bool ntuple)
+{ return 0; }
+static inline void mlx5e_arfs_destroy_tables(struct mlx5e_flow_steering *fs, bool ntuple) {}
+static inline int mlx5e_arfs_enable(struct mlx5e_flow_steering *fs)
+{ return -EOPNOTSUPP; }
+static inline int mlx5e_arfs_disable(struct mlx5e_flow_steering *fs)
+{ return -EOPNOTSUPP; }
+#endif
+
+#ifdef CONFIG_MLX5_EN_TLS
+struct mlx5e_accel_fs_tcp;
+#endif
+
+struct mlx5e_profile;
+struct mlx5e_fs_udp;
+struct mlx5e_fs_any;
+struct mlx5e_ptp_fs;
+
+void mlx5e_set_ttc_params(struct mlx5e_flow_steering *fs,
+			  struct mlx5e_rx_res *rx_res,
+			  struct ttc_params *ttc_params, bool tunnel);
+
+void mlx5e_destroy_ttc_table(struct mlx5e_flow_steering *fs);
+int mlx5e_create_ttc_table(struct mlx5e_flow_steering  *fs,
+			   struct mlx5e_rx_res *rx_res);
+
+void mlx5e_destroy_flow_table(struct mlx5e_flow_table *ft);
+
+void mlx5e_enable_cvlan_filter(struct mlx5e_flow_steering *fs, bool promisc);
+void mlx5e_disable_cvlan_filter(struct mlx5e_flow_steering *fs, bool promisc);
+
+int mlx5e_create_flow_steering(struct mlx5e_flow_steering *fs,
+			       struct mlx5e_rx_res *rx_res,
+			       const struct mlx5e_profile *profile,
+			       struct net_device *netdev);
+void mlx5e_destroy_flow_steering(struct mlx5e_flow_steering *fs, bool ntuple,
+				 const struct mlx5e_profile *profile);
+
+struct mlx5e_flow_steering *mlx5e_fs_init(const struct mlx5e_profile *profile,
+					  struct mlx5_core_dev *mdev,
+					  bool state_destroy);
+void mlx5e_fs_cleanup(struct mlx5e_flow_steering *fs);
+struct mlx5e_vlan_table *mlx5e_fs_get_vlan(struct mlx5e_flow_steering *fs);
+void mlx5e_fs_set_tc(struct mlx5e_flow_steering *fs, struct mlx5e_tc_table *tc);
+struct mlx5e_tc_table *mlx5e_fs_get_tc(struct mlx5e_flow_steering *fs);
+struct mlx5e_l2_table *mlx5e_fs_get_l2(struct mlx5e_flow_steering *fs);
+struct mlx5_flow_namespace *mlx5e_fs_get_ns(struct mlx5e_flow_steering *fs, bool egress);
+void mlx5e_fs_set_ns(struct mlx5e_flow_steering *fs, struct mlx5_flow_namespace *ns, bool egress);
+#ifdef CONFIG_MLX5_EN_RXNFC
+struct mlx5e_ethtool_steering *mlx5e_fs_get_ethtool(struct mlx5e_flow_steering *fs);
+#endif
+struct mlx5_ttc_table *mlx5e_fs_get_ttc(struct mlx5e_flow_steering *fs, bool inner);
+void mlx5e_fs_set_ttc(struct mlx5e_flow_steering *fs, struct mlx5_ttc_table *ttc, bool inner);
+#ifdef CONFIG_MLX5_EN_ARFS
+struct mlx5e_arfs_tables *mlx5e_fs_get_arfs(struct mlx5e_flow_steering *fs);
+void mlx5e_fs_set_arfs(struct mlx5e_flow_steering *fs, struct mlx5e_arfs_tables *arfs);
+#endif
+struct mlx5e_ptp_fs *mlx5e_fs_get_ptp(struct mlx5e_flow_steering *fs);
+void mlx5e_fs_set_ptp(struct mlx5e_flow_steering *fs, struct mlx5e_ptp_fs *ptp_fs);
+struct mlx5e_fs_any *mlx5e_fs_get_any(struct mlx5e_flow_steering *fs);
+void mlx5e_fs_set_any(struct mlx5e_flow_steering *fs, struct mlx5e_fs_any *any);
+struct mlx5e_fs_udp *mlx5e_fs_get_udp(struct mlx5e_flow_steering *fs);
+void mlx5e_fs_set_udp(struct mlx5e_flow_steering *fs, struct mlx5e_fs_udp *udp);
+#ifdef CONFIG_MLX5_EN_TLS
+struct mlx5e_accel_fs_tcp *mlx5e_fs_get_accel_tcp(struct mlx5e_flow_steering *fs);
+void mlx5e_fs_set_accel_tcp(struct mlx5e_flow_steering *fs, struct mlx5e_accel_fs_tcp *accel_tcp);
+#endif
+void mlx5e_fs_set_state_destroy(struct mlx5e_flow_steering *fs, bool state_destroy);
+void mlx5e_fs_set_vlan_strip_disable(struct mlx5e_flow_steering *fs, bool vlan_strip_disable);
+
+struct mlx5_core_dev *mlx5e_fs_get_mdev(struct mlx5e_flow_steering *fs);
+int mlx5e_add_vlan_trap(struct mlx5e_flow_steering *fs, int  trap_id, int tir_num);
+void mlx5e_remove_vlan_trap(struct mlx5e_flow_steering *fs);
+int mlx5e_add_mac_trap(struct mlx5e_flow_steering *fs, int  trap_id, int tir_num);
+void mlx5e_remove_mac_trap(struct mlx5e_flow_steering *fs);
+void mlx5e_fs_set_rx_mode_work(struct mlx5e_flow_steering *fs, struct net_device *netdev);
+int mlx5e_fs_vlan_rx_add_vid(struct mlx5e_flow_steering *fs,
+			     struct net_device *netdev,
+			     __be16 proto, u16 vid);
+int mlx5e_fs_vlan_rx_kill_vid(struct mlx5e_flow_steering *fs,
+			      struct net_device *netdev,
+			      __be16 proto, u16 vid);
+void mlx5e_fs_init_l2_addr(struct mlx5e_flow_steering *fs, struct net_device *netdev);
+
+#define fs_err(fs, fmt, ...) \
+	mlx5_core_err(mlx5e_fs_get_mdev(fs), fmt, ##__VA_ARGS__)
+
+#define fs_dbg(fs, fmt, ...) \
+	mlx5_core_dbg(mlx5e_fs_get_mdev(fs), fmt, ##__VA_ARGS__)
+
+#define fs_warn(fs, fmt, ...) \
+	mlx5_core_warn(mlx5e_fs_get_mdev(fs), fmt, ##__VA_ARGS__)
+
+#define fs_warn_once(fs, fmt, ...) \
+	mlx5_core_warn_once(mlx5e_fs_get_mdev(fs), fmt, ##__VA_ARGS__)
+
+#endif /* __MLX5E_FLOW_STEER_H__ */
+
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/fs_ethtool.h b/drivers/net/ethernet/mellanox/mlx5/core/en/fs_ethtool.h
new file mode 100644
index 000000000..9e276fd3c
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/fs_ethtool.h
@@ -0,0 +1,29 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2022, NVIDIA CORPORATION & AFFILIATES. */
+
+#ifndef __MLX5E_FS_ETHTOOL_H__
+#define __MLX5E_FS_ETHTOOL_H__
+
+struct mlx5e_priv;
+struct mlx5e_ethtool_steering;
+#ifdef CONFIG_MLX5_EN_RXNFC
+int mlx5e_ethtool_alloc(struct mlx5e_ethtool_steering **ethtool);
+void mlx5e_ethtool_free(struct mlx5e_ethtool_steering *ethtool);
+void mlx5e_ethtool_init_steering(struct mlx5e_flow_steering *fs);
+void mlx5e_ethtool_cleanup_steering(struct mlx5e_flow_steering *fs);
+int mlx5e_ethtool_set_rxnfc(struct mlx5e_priv *priv, struct ethtool_rxnfc *cmd);
+int mlx5e_ethtool_get_rxnfc(struct mlx5e_priv *priv,
+			    struct ethtool_rxnfc *info, u32 *rule_locs);
+#else
+static inline int mlx5e_ethtool_alloc(struct mlx5e_ethtool_steering **ethtool)
+{ return 0; }
+static inline void mlx5e_ethtool_free(struct mlx5e_ethtool_steering *ethtool) { }
+static inline void mlx5e_ethtool_init_steering(struct mlx5e_flow_steering *fs) { }
+static inline void mlx5e_ethtool_cleanup_steering(struct mlx5e_flow_steering *fs) { }
+static inline int mlx5e_ethtool_set_rxnfc(struct mlx5e_priv *priv, struct ethtool_rxnfc *cmd)
+{ return -EOPNOTSUPP; }
+static inline int mlx5e_ethtool_get_rxnfc(struct mlx5e_priv *priv,
+					  struct ethtool_rxnfc *info, u32 *rule_locs)
+{ return -EOPNOTSUPP; }
+#endif
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/fs_tt_redirect.c b/drivers/net/ethernet/mellanox/mlx5/core/en/fs_tt_redirect.c
new file mode 100644
index 000000000..671adbad0
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/fs_tt_redirect.c
@@ -0,0 +1,615 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2021, Mellanox Technologies inc. All rights reserved. */
+
+#include "en/fs_tt_redirect.h"
+#include "fs_core.h"
+#include "mlx5_core.h"
+
+enum fs_udp_type {
+	FS_IPV4_UDP,
+	FS_IPV6_UDP,
+	FS_UDP_NUM_TYPES,
+};
+
+struct mlx5e_fs_udp {
+	struct mlx5e_flow_table tables[FS_UDP_NUM_TYPES];
+	struct mlx5_flow_handle *default_rules[FS_UDP_NUM_TYPES];
+	int ref_cnt;
+};
+
+struct mlx5e_fs_any {
+	struct mlx5e_flow_table table;
+	struct mlx5_flow_handle *default_rule;
+	int ref_cnt;
+};
+
+static char *fs_udp_type2str(enum fs_udp_type i)
+{
+	switch (i) {
+	case FS_IPV4_UDP:
+		return "UDP v4";
+	default: /* FS_IPV6_UDP */
+		return "UDP v6";
+	}
+}
+
+static enum mlx5_traffic_types fs_udp2tt(enum fs_udp_type i)
+{
+	switch (i) {
+	case FS_IPV4_UDP:
+		return MLX5_TT_IPV4_UDP;
+	default: /* FS_IPV6_UDP */
+		return MLX5_TT_IPV6_UDP;
+	}
+}
+
+static enum fs_udp_type tt2fs_udp(enum mlx5_traffic_types i)
+{
+	switch (i) {
+	case MLX5_TT_IPV4_UDP:
+		return FS_IPV4_UDP;
+	case MLX5_TT_IPV6_UDP:
+		return FS_IPV6_UDP;
+	default:
+		return FS_UDP_NUM_TYPES;
+	}
+}
+
+void mlx5e_fs_tt_redirect_del_rule(struct mlx5_flow_handle *rule)
+{
+	mlx5_del_flow_rules(rule);
+}
+
+static void fs_udp_set_dport_flow(struct mlx5_flow_spec *spec, enum fs_udp_type type,
+				  u16 udp_dport)
+{
+	spec->match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
+	MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria, outer_headers.ip_protocol);
+	MLX5_SET(fte_match_param, spec->match_value, outer_headers.ip_protocol, IPPROTO_UDP);
+	MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria, outer_headers.ip_version);
+	MLX5_SET(fte_match_param, spec->match_value, outer_headers.ip_version,
+		 type == FS_IPV4_UDP ? 4 : 6);
+	MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria, outer_headers.udp_dport);
+	MLX5_SET(fte_match_param, spec->match_value, outer_headers.udp_dport, udp_dport);
+}
+
+struct mlx5_flow_handle *
+mlx5e_fs_tt_redirect_udp_add_rule(struct mlx5e_flow_steering *fs,
+				  enum mlx5_traffic_types ttc_type,
+				  u32 tir_num, u16 d_port)
+{
+	struct mlx5e_fs_udp *fs_udp = mlx5e_fs_get_udp(fs);
+	enum fs_udp_type type = tt2fs_udp(ttc_type);
+	struct mlx5_flow_destination dest = {};
+	struct mlx5_flow_table *ft = NULL;
+	MLX5_DECLARE_FLOW_ACT(flow_act);
+	struct mlx5_flow_handle *rule;
+	struct mlx5_flow_spec *spec;
+	int err;
+
+	if (type == FS_UDP_NUM_TYPES)
+		return ERR_PTR(-EINVAL);
+
+	spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
+	if (!spec)
+		return ERR_PTR(-ENOMEM);
+
+	ft = fs_udp->tables[type].t;
+
+	fs_udp_set_dport_flow(spec, type, d_port);
+	dest.type = MLX5_FLOW_DESTINATION_TYPE_TIR;
+	dest.tir_num = tir_num;
+
+	rule = mlx5_add_flow_rules(ft, spec, &flow_act, &dest, 1);
+	kvfree(spec);
+
+	if (IS_ERR(rule)) {
+		err = PTR_ERR(rule);
+		fs_err(fs, "%s: add %s rule failed, err %d\n",
+		       __func__, fs_udp_type2str(type), err);
+	}
+	return rule;
+}
+
+static int fs_udp_add_default_rule(struct mlx5e_flow_steering *fs, enum fs_udp_type type)
+{
+	struct mlx5_ttc_table *ttc = mlx5e_fs_get_ttc(fs, false);
+	struct mlx5e_fs_udp *fs_udp = mlx5e_fs_get_udp(fs);
+	struct mlx5e_flow_table *fs_udp_t;
+	struct mlx5_flow_destination dest;
+	MLX5_DECLARE_FLOW_ACT(flow_act);
+	struct mlx5_flow_handle *rule;
+	int err;
+
+	fs_udp_t = &fs_udp->tables[type];
+
+	dest = mlx5_ttc_get_default_dest(ttc, fs_udp2tt(type));
+	rule = mlx5_add_flow_rules(fs_udp_t->t, NULL, &flow_act, &dest, 1);
+	if (IS_ERR(rule)) {
+		err = PTR_ERR(rule);
+		fs_err(fs, "%s: add default rule failed, fs type=%d, err %d\n",
+		       __func__, type, err);
+		return err;
+	}
+
+	fs_udp->default_rules[type] = rule;
+	return 0;
+}
+
+#define MLX5E_FS_UDP_NUM_GROUPS	(2)
+#define MLX5E_FS_UDP_GROUP1_SIZE	(BIT(16))
+#define MLX5E_FS_UDP_GROUP2_SIZE	(BIT(0))
+#define MLX5E_FS_UDP_TABLE_SIZE		(MLX5E_FS_UDP_GROUP1_SIZE +\
+					 MLX5E_FS_UDP_GROUP2_SIZE)
+static int fs_udp_create_groups(struct mlx5e_flow_table *ft, enum fs_udp_type type)
+{
+	int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
+	void *outer_headers_c;
+	int ix = 0;
+	u32 *in;
+	int err;
+	u8 *mc;
+
+	ft->g = kcalloc(MLX5E_FS_UDP_NUM_GROUPS, sizeof(*ft->g), GFP_KERNEL);
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if  (!in || !ft->g) {
+		kfree(ft->g);
+		ft->g = NULL;
+		kvfree(in);
+		return -ENOMEM;
+	}
+
+	mc = MLX5_ADDR_OF(create_flow_group_in, in, match_criteria);
+	outer_headers_c = MLX5_ADDR_OF(fte_match_param, mc, outer_headers);
+	MLX5_SET_TO_ONES(fte_match_set_lyr_2_4, outer_headers_c, ip_protocol);
+	MLX5_SET_TO_ONES(fte_match_set_lyr_2_4, outer_headers_c, ip_version);
+
+	switch (type) {
+	case FS_IPV4_UDP:
+	case FS_IPV6_UDP:
+		MLX5_SET_TO_ONES(fte_match_set_lyr_2_4, outer_headers_c, udp_dport);
+		break;
+	default:
+		err = -EINVAL;
+		goto out;
+	}
+	/* Match on udp protocol, Ipv4/6 and dport */
+	MLX5_SET_CFG(in, match_criteria_enable, MLX5_MATCH_OUTER_HEADERS);
+	MLX5_SET_CFG(in, start_flow_index, ix);
+	ix += MLX5E_FS_UDP_GROUP1_SIZE;
+	MLX5_SET_CFG(in, end_flow_index, ix - 1);
+	ft->g[ft->num_groups] = mlx5_create_flow_group(ft->t, in);
+	if (IS_ERR(ft->g[ft->num_groups]))
+		goto err;
+	ft->num_groups++;
+
+	/* Default Flow Group */
+	memset(in, 0, inlen);
+	MLX5_SET_CFG(in, start_flow_index, ix);
+	ix += MLX5E_FS_UDP_GROUP2_SIZE;
+	MLX5_SET_CFG(in, end_flow_index, ix - 1);
+	ft->g[ft->num_groups] = mlx5_create_flow_group(ft->t, in);
+	if (IS_ERR(ft->g[ft->num_groups]))
+		goto err;
+	ft->num_groups++;
+
+	kvfree(in);
+	return 0;
+
+err:
+	err = PTR_ERR(ft->g[ft->num_groups]);
+	ft->g[ft->num_groups] = NULL;
+out:
+	kvfree(in);
+
+	return err;
+}
+
+static int fs_udp_create_table(struct mlx5e_flow_steering *fs, enum fs_udp_type type)
+{
+	struct mlx5_flow_namespace *ns = mlx5e_fs_get_ns(fs, false);
+	struct mlx5e_fs_udp *fs_udp = mlx5e_fs_get_udp(fs);
+	struct mlx5_flow_table_attr ft_attr = {};
+	struct mlx5e_flow_table *ft;
+	int err;
+
+	ft = &fs_udp->tables[type];
+	ft->num_groups = 0;
+
+	ft_attr.max_fte = MLX5E_FS_UDP_TABLE_SIZE;
+	ft_attr.level = MLX5E_FS_TT_UDP_FT_LEVEL;
+	ft_attr.prio = MLX5E_NIC_PRIO;
+
+	ft->t = mlx5_create_flow_table(ns, &ft_attr);
+	if (IS_ERR(ft->t)) {
+		err = PTR_ERR(ft->t);
+		ft->t = NULL;
+		return err;
+	}
+
+	mlx5_core_dbg(mlx5e_fs_get_mdev(fs), "Created fs %s table id %u level %u\n",
+		      fs_udp_type2str(type), ft->t->id, ft->t->level);
+
+	err = fs_udp_create_groups(ft, type);
+	if (err)
+		goto err;
+
+	err = fs_udp_add_default_rule(fs, type);
+	if (err)
+		goto err;
+
+	return 0;
+
+err:
+	mlx5e_destroy_flow_table(ft);
+	return err;
+}
+
+static void fs_udp_destroy_table(struct mlx5e_fs_udp *fs_udp, int i)
+{
+	if (IS_ERR_OR_NULL(fs_udp->tables[i].t))
+		return;
+
+	mlx5_del_flow_rules(fs_udp->default_rules[i]);
+	mlx5e_destroy_flow_table(&fs_udp->tables[i]);
+	fs_udp->tables[i].t = NULL;
+}
+
+static int fs_udp_disable(struct mlx5e_flow_steering *fs)
+{
+	struct mlx5_ttc_table *ttc = mlx5e_fs_get_ttc(fs, false);
+	int err, i;
+
+	for (i = 0; i < FS_UDP_NUM_TYPES; i++) {
+		/* Modify ttc rules destination to point back to the indir TIRs */
+		err = mlx5_ttc_fwd_default_dest(ttc, fs_udp2tt(i));
+		if (err) {
+			fs_err(fs, "%s: modify ttc[%d] default destination failed, err(%d)\n",
+			       __func__, fs_udp2tt(i), err);
+			return err;
+		}
+	}
+
+	return 0;
+}
+
+static int fs_udp_enable(struct mlx5e_flow_steering *fs)
+{
+	struct mlx5_ttc_table *ttc = mlx5e_fs_get_ttc(fs, false);
+	struct mlx5e_fs_udp *udp = mlx5e_fs_get_udp(fs);
+	struct mlx5_flow_destination dest = {};
+	int err, i;
+
+	dest.type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
+	for (i = 0; i < FS_UDP_NUM_TYPES; i++) {
+		dest.ft = udp->tables[i].t;
+
+		/* Modify ttc rules destination to point on the accel_fs FTs */
+		err = mlx5_ttc_fwd_dest(ttc, fs_udp2tt(i), &dest);
+		if (err) {
+			fs_err(fs, "%s: modify ttc[%d] destination to accel failed, err(%d)\n",
+			       __func__, fs_udp2tt(i), err);
+			return err;
+		}
+	}
+	return 0;
+}
+
+void mlx5e_fs_tt_redirect_udp_destroy(struct mlx5e_flow_steering *fs)
+{
+	struct mlx5e_fs_udp *fs_udp = mlx5e_fs_get_udp(fs);
+	int i;
+
+	if (!fs_udp)
+		return;
+
+	if (--fs_udp->ref_cnt)
+		return;
+
+	fs_udp_disable(fs);
+
+	for (i = 0; i < FS_UDP_NUM_TYPES; i++)
+		fs_udp_destroy_table(fs_udp, i);
+
+	kfree(fs_udp);
+	mlx5e_fs_set_udp(fs, NULL);
+}
+
+int mlx5e_fs_tt_redirect_udp_create(struct mlx5e_flow_steering *fs)
+{
+	struct mlx5e_fs_udp *udp = mlx5e_fs_get_udp(fs);
+	int i, err;
+
+	if (udp) {
+		udp->ref_cnt++;
+		return 0;
+	}
+
+	udp = kzalloc(sizeof(*udp), GFP_KERNEL);
+	if (!udp)
+		return -ENOMEM;
+	mlx5e_fs_set_udp(fs, udp);
+
+	for (i = 0; i < FS_UDP_NUM_TYPES; i++) {
+		err = fs_udp_create_table(fs, i);
+		if (err)
+			goto err_destroy_tables;
+	}
+
+	err = fs_udp_enable(fs);
+	if (err)
+		goto err_destroy_tables;
+
+	udp->ref_cnt = 1;
+
+	return 0;
+
+err_destroy_tables:
+	while (--i >= 0)
+		fs_udp_destroy_table(udp, i);
+
+	kfree(udp);
+	mlx5e_fs_set_udp(fs, NULL);
+	return err;
+}
+
+static void fs_any_set_ethertype_flow(struct mlx5_flow_spec *spec, u16 ether_type)
+{
+	spec->match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
+	MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria, outer_headers.ethertype);
+	MLX5_SET(fte_match_param, spec->match_value, outer_headers.ethertype, ether_type);
+}
+
+struct mlx5_flow_handle *
+mlx5e_fs_tt_redirect_any_add_rule(struct mlx5e_flow_steering *fs,
+				  u32 tir_num, u16 ether_type)
+{
+	struct mlx5e_fs_any *fs_any = mlx5e_fs_get_any(fs);
+	struct mlx5_flow_destination dest = {};
+	struct mlx5_flow_table *ft = NULL;
+	MLX5_DECLARE_FLOW_ACT(flow_act);
+	struct mlx5_flow_handle *rule;
+	struct mlx5_flow_spec *spec;
+	int err;
+
+	spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
+	if (!spec)
+		return ERR_PTR(-ENOMEM);
+
+	ft = fs_any->table.t;
+
+	fs_any_set_ethertype_flow(spec, ether_type);
+	dest.type = MLX5_FLOW_DESTINATION_TYPE_TIR;
+	dest.tir_num = tir_num;
+
+	rule = mlx5_add_flow_rules(ft, spec, &flow_act, &dest, 1);
+	kvfree(spec);
+
+	if (IS_ERR(rule)) {
+		err = PTR_ERR(rule);
+		fs_err(fs, "%s: add ANY rule failed, err %d\n",
+		       __func__, err);
+	}
+	return rule;
+}
+
+static int fs_any_add_default_rule(struct mlx5e_flow_steering *fs)
+{
+	struct mlx5_ttc_table *ttc = mlx5e_fs_get_ttc(fs, false);
+	struct mlx5e_fs_any *fs_any = mlx5e_fs_get_any(fs);
+	struct mlx5e_flow_table *fs_any_t;
+	struct mlx5_flow_destination dest;
+	MLX5_DECLARE_FLOW_ACT(flow_act);
+	struct mlx5_flow_handle *rule;
+	int err;
+
+	fs_any_t = &fs_any->table;
+	dest = mlx5_ttc_get_default_dest(ttc, MLX5_TT_ANY);
+	rule = mlx5_add_flow_rules(fs_any_t->t, NULL, &flow_act, &dest, 1);
+	if (IS_ERR(rule)) {
+		err = PTR_ERR(rule);
+		fs_err(fs, "%s: add default rule failed, fs type=ANY, err %d\n",
+		       __func__, err);
+		return err;
+	}
+
+	fs_any->default_rule = rule;
+	return 0;
+}
+
+#define MLX5E_FS_ANY_NUM_GROUPS	(2)
+#define MLX5E_FS_ANY_GROUP1_SIZE	(BIT(16))
+#define MLX5E_FS_ANY_GROUP2_SIZE	(BIT(0))
+#define MLX5E_FS_ANY_TABLE_SIZE		(MLX5E_FS_ANY_GROUP1_SIZE +\
+					 MLX5E_FS_ANY_GROUP2_SIZE)
+
+static int fs_any_create_groups(struct mlx5e_flow_table *ft)
+{
+	int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
+	void *outer_headers_c;
+	int ix = 0;
+	u32 *in;
+	int err;
+	u8 *mc;
+
+	ft->g = kcalloc(MLX5E_FS_UDP_NUM_GROUPS, sizeof(*ft->g), GFP_KERNEL);
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if  (!in || !ft->g) {
+		kfree(ft->g);
+		ft->g = NULL;
+		kvfree(in);
+		return -ENOMEM;
+	}
+
+	/* Match on ethertype */
+	mc = MLX5_ADDR_OF(create_flow_group_in, in, match_criteria);
+	outer_headers_c = MLX5_ADDR_OF(fte_match_param, mc, outer_headers);
+	MLX5_SET_TO_ONES(fte_match_set_lyr_2_4, outer_headers_c, ethertype);
+	MLX5_SET_CFG(in, match_criteria_enable, MLX5_MATCH_OUTER_HEADERS);
+	MLX5_SET_CFG(in, start_flow_index, ix);
+	ix += MLX5E_FS_ANY_GROUP1_SIZE;
+	MLX5_SET_CFG(in, end_flow_index, ix - 1);
+	ft->g[ft->num_groups] = mlx5_create_flow_group(ft->t, in);
+	if (IS_ERR(ft->g[ft->num_groups]))
+		goto err;
+	ft->num_groups++;
+
+	/* Default Flow Group */
+	memset(in, 0, inlen);
+	MLX5_SET_CFG(in, start_flow_index, ix);
+	ix += MLX5E_FS_ANY_GROUP2_SIZE;
+	MLX5_SET_CFG(in, end_flow_index, ix - 1);
+	ft->g[ft->num_groups] = mlx5_create_flow_group(ft->t, in);
+	if (IS_ERR(ft->g[ft->num_groups]))
+		goto err;
+	ft->num_groups++;
+
+	kvfree(in);
+	return 0;
+
+err:
+	err = PTR_ERR(ft->g[ft->num_groups]);
+	ft->g[ft->num_groups] = NULL;
+	kvfree(in);
+
+	return err;
+}
+
+static int fs_any_create_table(struct mlx5e_flow_steering *fs)
+{
+	struct mlx5_flow_namespace *ns = mlx5e_fs_get_ns(fs, false);
+	struct mlx5e_fs_any *fs_any = mlx5e_fs_get_any(fs);
+	struct mlx5e_flow_table *ft = &fs_any->table;
+	struct mlx5_flow_table_attr ft_attr = {};
+	int err;
+
+	ft->num_groups = 0;
+
+	ft_attr.max_fte = MLX5E_FS_UDP_TABLE_SIZE;
+	ft_attr.level = MLX5E_FS_TT_ANY_FT_LEVEL;
+	ft_attr.prio = MLX5E_NIC_PRIO;
+
+	ft->t = mlx5_create_flow_table(ns, &ft_attr);
+	if (IS_ERR(ft->t)) {
+		err = PTR_ERR(ft->t);
+		ft->t = NULL;
+		return err;
+	}
+
+	mlx5_core_dbg(mlx5e_fs_get_mdev(fs), "Created fs ANY table id %u level %u\n",
+		      ft->t->id, ft->t->level);
+
+	err = fs_any_create_groups(ft);
+	if (err)
+		goto err;
+
+	err = fs_any_add_default_rule(fs);
+	if (err)
+		goto err;
+
+	return 0;
+
+err:
+	mlx5e_destroy_flow_table(ft);
+	return err;
+}
+
+static int fs_any_disable(struct mlx5e_flow_steering *fs)
+{
+	struct mlx5_ttc_table *ttc = mlx5e_fs_get_ttc(fs, false);
+	int err;
+
+	/* Modify ttc rules destination to point back to the indir TIRs */
+	err = mlx5_ttc_fwd_default_dest(ttc, MLX5_TT_ANY);
+	if (err) {
+		fs_err(fs,
+		       "%s: modify ttc[%d] default destination failed, err(%d)\n",
+		       __func__, MLX5_TT_ANY, err);
+		return err;
+	}
+	return 0;
+}
+
+static int fs_any_enable(struct mlx5e_flow_steering *fs)
+{
+	struct mlx5_ttc_table *ttc = mlx5e_fs_get_ttc(fs, false);
+	struct mlx5e_fs_any *any = mlx5e_fs_get_any(fs);
+	struct mlx5_flow_destination dest = {};
+	int err;
+
+	dest.type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
+	dest.ft = any->table.t;
+
+	/* Modify ttc rules destination to point on the accel_fs FTs */
+	err = mlx5_ttc_fwd_dest(ttc, MLX5_TT_ANY, &dest);
+	if (err) {
+		fs_err(fs,
+		       "%s: modify ttc[%d] destination to accel failed, err(%d)\n",
+		       __func__, MLX5_TT_ANY, err);
+		return err;
+	}
+	return 0;
+}
+
+static void fs_any_destroy_table(struct mlx5e_fs_any *fs_any)
+{
+	if (IS_ERR_OR_NULL(fs_any->table.t))
+		return;
+
+	mlx5_del_flow_rules(fs_any->default_rule);
+	mlx5e_destroy_flow_table(&fs_any->table);
+	fs_any->table.t = NULL;
+}
+
+void mlx5e_fs_tt_redirect_any_destroy(struct mlx5e_flow_steering *fs)
+{
+	struct mlx5e_fs_any *fs_any = mlx5e_fs_get_any(fs);
+
+	if (!fs_any)
+		return;
+
+	if (--fs_any->ref_cnt)
+		return;
+
+	fs_any_disable(fs);
+
+	fs_any_destroy_table(fs_any);
+
+	kfree(fs_any);
+	mlx5e_fs_set_any(fs, NULL);
+}
+
+int mlx5e_fs_tt_redirect_any_create(struct mlx5e_flow_steering *fs)
+{
+	struct mlx5e_fs_any *fs_any = mlx5e_fs_get_any(fs);
+	int err;
+
+	if (fs_any) {
+		fs_any->ref_cnt++;
+		return 0;
+	}
+
+	fs_any = kzalloc(sizeof(*fs_any), GFP_KERNEL);
+	if (!fs_any)
+		return -ENOMEM;
+	mlx5e_fs_set_any(fs, fs_any);
+
+	err = fs_any_create_table(fs);
+	if (err)
+		goto err_free_any;
+
+	err = fs_any_enable(fs);
+	if (err)
+		goto err_destroy_table;
+
+	fs_any->ref_cnt = 1;
+
+	return 0;
+
+err_destroy_table:
+	fs_any_destroy_table(fs_any);
+err_free_any:
+	mlx5e_fs_set_any(fs, NULL);
+	kfree(fs_any);
+	return err;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/fs_tt_redirect.h b/drivers/net/ethernet/mellanox/mlx5/core/en/fs_tt_redirect.h
new file mode 100644
index 000000000..5780fd7ad
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/fs_tt_redirect.h
@@ -0,0 +1,25 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2021 Mellanox Technologies. */
+
+#ifndef __MLX5E_FS_TT_REDIRECT_H__
+#define __MLX5E_FS_TT_REDIRECT_H__
+
+#include "en/fs.h"
+
+void mlx5e_fs_tt_redirect_del_rule(struct mlx5_flow_handle *rule);
+
+/* UDP traffic type redirect */
+struct mlx5_flow_handle *
+mlx5e_fs_tt_redirect_udp_add_rule(struct mlx5e_flow_steering *fs,
+				  enum mlx5_traffic_types ttc_type,
+				  u32 tir_num, u16 d_port);
+void mlx5e_fs_tt_redirect_udp_destroy(struct mlx5e_flow_steering *fs);
+int mlx5e_fs_tt_redirect_udp_create(struct mlx5e_flow_steering *fs);
+
+/* ANY traffic type redirect*/
+struct mlx5_flow_handle *
+mlx5e_fs_tt_redirect_any_add_rule(struct mlx5e_flow_steering *fs,
+				  u32 tir_num, u16 ether_type);
+void mlx5e_fs_tt_redirect_any_destroy(struct mlx5e_flow_steering *fs);
+int mlx5e_fs_tt_redirect_any_create(struct mlx5e_flow_steering *fs);
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/health.c b/drivers/net/ethernet/mellanox/mlx5/core/en/health.c
new file mode 100644
index 000000000..6f4e6c34b
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/health.c
@@ -0,0 +1,339 @@
+// SPDX-License-Identifier: GPL-2.0
+// Copyright (c) 2019 Mellanox Technologies.
+
+#include "health.h"
+#include "lib/eq.h"
+#include "lib/mlx5.h"
+
+int mlx5e_health_fmsg_named_obj_nest_start(struct devlink_fmsg *fmsg, char *name)
+{
+	int err;
+
+	err = devlink_fmsg_pair_nest_start(fmsg, name);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_obj_nest_start(fmsg);
+	if (err)
+		return err;
+
+	return 0;
+}
+
+int mlx5e_health_fmsg_named_obj_nest_end(struct devlink_fmsg *fmsg)
+{
+	int err;
+
+	err = devlink_fmsg_obj_nest_end(fmsg);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_pair_nest_end(fmsg);
+	if (err)
+		return err;
+
+	return 0;
+}
+
+int mlx5e_health_cq_diag_fmsg(struct mlx5e_cq *cq, struct devlink_fmsg *fmsg)
+{
+	u32 out[MLX5_ST_SZ_DW(query_cq_out)] = {};
+	u8 hw_status;
+	void *cqc;
+	int err;
+
+	err = mlx5_core_query_cq(cq->mdev, &cq->mcq, out);
+	if (err)
+		return err;
+
+	cqc = MLX5_ADDR_OF(query_cq_out, out, cq_context);
+	hw_status = MLX5_GET(cqc, cqc, status);
+
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, "CQ");
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u32_pair_put(fmsg, "cqn", cq->mcq.cqn);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u8_pair_put(fmsg, "HW status", hw_status);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u32_pair_put(fmsg, "ci", mlx5_cqwq_get_ci(&cq->wq));
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u32_pair_put(fmsg, "size", mlx5_cqwq_get_size(&cq->wq));
+	if (err)
+		return err;
+
+	err = mlx5e_health_fmsg_named_obj_nest_end(fmsg);
+	if (err)
+		return err;
+
+	return 0;
+}
+
+int mlx5e_health_cq_common_diag_fmsg(struct mlx5e_cq *cq, struct devlink_fmsg *fmsg)
+{
+	u8 cq_log_stride;
+	u32 cq_sz;
+	int err;
+
+	cq_sz = mlx5_cqwq_get_size(&cq->wq);
+	cq_log_stride = mlx5_cqwq_get_log_stride_size(&cq->wq);
+
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, "CQ");
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u64_pair_put(fmsg, "stride size", BIT(cq_log_stride));
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u32_pair_put(fmsg, "size", cq_sz);
+	if (err)
+		return err;
+
+	err = mlx5e_health_fmsg_named_obj_nest_end(fmsg);
+	if (err)
+		return err;
+
+	return 0;
+}
+
+int mlx5e_health_eq_diag_fmsg(struct mlx5_eq_comp *eq, struct devlink_fmsg *fmsg)
+{
+	int err;
+
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, "EQ");
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u8_pair_put(fmsg, "eqn", eq->core.eqn);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u32_pair_put(fmsg, "irqn", eq->core.irqn);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u32_pair_put(fmsg, "vecidx", eq->core.vecidx);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u32_pair_put(fmsg, "ci", eq->core.cons_index);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u32_pair_put(fmsg, "size", eq_get_size(&eq->core));
+	if (err)
+		return err;
+
+	return mlx5e_health_fmsg_named_obj_nest_end(fmsg);
+}
+
+void mlx5e_health_create_reporters(struct mlx5e_priv *priv)
+{
+	mlx5e_reporter_tx_create(priv);
+	mlx5e_reporter_rx_create(priv);
+}
+
+void mlx5e_health_destroy_reporters(struct mlx5e_priv *priv)
+{
+	mlx5e_reporter_rx_destroy(priv);
+	mlx5e_reporter_tx_destroy(priv);
+}
+
+void mlx5e_health_channels_update(struct mlx5e_priv *priv)
+{
+	if (priv->tx_reporter)
+		devlink_health_reporter_state_update(priv->tx_reporter,
+						     DEVLINK_HEALTH_REPORTER_STATE_HEALTHY);
+	if (priv->rx_reporter)
+		devlink_health_reporter_state_update(priv->rx_reporter,
+						     DEVLINK_HEALTH_REPORTER_STATE_HEALTHY);
+}
+
+int mlx5e_health_sq_to_ready(struct mlx5_core_dev *mdev, struct net_device *dev, u32 sqn)
+{
+	struct mlx5e_modify_sq_param msp = {};
+	int err;
+
+	msp.curr_state = MLX5_SQC_STATE_ERR;
+	msp.next_state = MLX5_SQC_STATE_RST;
+
+	err = mlx5e_modify_sq(mdev, sqn, &msp);
+	if (err) {
+		netdev_err(dev, "Failed to move sq 0x%x to reset\n", sqn);
+		return err;
+	}
+
+	memset(&msp, 0, sizeof(msp));
+	msp.curr_state = MLX5_SQC_STATE_RST;
+	msp.next_state = MLX5_SQC_STATE_RDY;
+
+	err = mlx5e_modify_sq(mdev, sqn, &msp);
+	if (err) {
+		netdev_err(dev, "Failed to move sq 0x%x to ready\n", sqn);
+		return err;
+	}
+
+	return 0;
+}
+
+int mlx5e_health_recover_channels(struct mlx5e_priv *priv)
+{
+	int err = 0;
+
+	rtnl_lock();
+	mutex_lock(&priv->state_lock);
+
+	if (!test_bit(MLX5E_STATE_OPENED, &priv->state))
+		goto out;
+
+	err = mlx5e_safe_reopen_channels(priv);
+
+out:
+	mutex_unlock(&priv->state_lock);
+	rtnl_unlock();
+
+	return err;
+}
+
+int mlx5e_health_channel_eq_recover(struct net_device *dev, struct mlx5_eq_comp *eq,
+				    struct mlx5e_ch_stats *stats)
+{
+	u32 eqe_count;
+
+	netdev_err(dev, "EQ 0x%x: Cons = 0x%x, irqn = 0x%x\n",
+		   eq->core.eqn, eq->core.cons_index, eq->core.irqn);
+
+	eqe_count = mlx5_eq_poll_irq_disabled(eq);
+	if (!eqe_count)
+		return -EIO;
+
+	netdev_err(dev, "Recovered %d eqes on EQ 0x%x\n",
+		   eqe_count, eq->core.eqn);
+
+	stats->eq_rearm++;
+	return 0;
+}
+
+int mlx5e_health_report(struct mlx5e_priv *priv,
+			struct devlink_health_reporter *reporter, char *err_str,
+			struct mlx5e_err_ctx *err_ctx)
+{
+	netdev_err(priv->netdev, "%s\n", err_str);
+
+	if (!reporter)
+		return err_ctx->recover(err_ctx->ctx);
+
+	return devlink_health_report(reporter, err_str, err_ctx);
+}
+
+#define MLX5_HEALTH_DEVLINK_MAX_SIZE 1024
+static int mlx5e_health_rsc_fmsg_binary(struct devlink_fmsg *fmsg,
+					const void *value, u32 value_len)
+
+{
+	u32 data_size;
+	int err = 0;
+	u32 offset;
+
+	for (offset = 0; offset < value_len; offset += data_size) {
+		data_size = value_len - offset;
+		if (data_size > MLX5_HEALTH_DEVLINK_MAX_SIZE)
+			data_size = MLX5_HEALTH_DEVLINK_MAX_SIZE;
+		err = devlink_fmsg_binary_put(fmsg, value + offset, data_size);
+		if (err)
+			break;
+	}
+	return err;
+}
+
+int mlx5e_health_rsc_fmsg_dump(struct mlx5e_priv *priv, struct mlx5_rsc_key *key,
+			       struct devlink_fmsg *fmsg)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+	struct mlx5_rsc_dump_cmd *cmd;
+	struct page *page;
+	int cmd_err, err;
+	int end_err;
+	int size;
+
+	if (IS_ERR_OR_NULL(mdev->rsc_dump))
+		return -EOPNOTSUPP;
+
+	page = alloc_page(GFP_KERNEL);
+	if (!page)
+		return -ENOMEM;
+
+	err = devlink_fmsg_binary_pair_nest_start(fmsg, "data");
+	if (err)
+		goto free_page;
+
+	cmd = mlx5_rsc_dump_cmd_create(mdev, key);
+	if (IS_ERR(cmd)) {
+		err = PTR_ERR(cmd);
+		goto free_page;
+	}
+
+	do {
+		cmd_err = mlx5_rsc_dump_next(mdev, cmd, page, &size);
+		if (cmd_err < 0) {
+			err = cmd_err;
+			goto destroy_cmd;
+		}
+
+		err = mlx5e_health_rsc_fmsg_binary(fmsg, page_address(page), size);
+		if (err)
+			goto destroy_cmd;
+
+	} while (cmd_err > 0);
+
+destroy_cmd:
+	mlx5_rsc_dump_cmd_destroy(cmd);
+	end_err = devlink_fmsg_binary_pair_nest_end(fmsg);
+	if (end_err)
+		err = end_err;
+free_page:
+	__free_page(page);
+	return err;
+}
+
+int mlx5e_health_queue_dump(struct mlx5e_priv *priv, struct devlink_fmsg *fmsg,
+			    int queue_idx, char *lbl)
+{
+	struct mlx5_rsc_key key = {};
+	int err;
+
+	key.rsc = MLX5_SGMT_TYPE_FULL_QPC;
+	key.index1 = queue_idx;
+	key.size = PAGE_SIZE;
+	key.num_of_obj1 = 1;
+
+	err = devlink_fmsg_obj_nest_start(fmsg);
+	if (err)
+		return err;
+
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, lbl);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u32_pair_put(fmsg, "index", queue_idx);
+	if (err)
+		return err;
+
+	err = mlx5e_health_rsc_fmsg_dump(priv, &key, fmsg);
+	if (err)
+		return err;
+
+	err = mlx5e_health_fmsg_named_obj_nest_end(fmsg);
+	if (err)
+		return err;
+
+	return devlink_fmsg_obj_nest_end(fmsg);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/health.h b/drivers/net/ethernet/mellanox/mlx5/core/en/health.h
new file mode 100644
index 000000000..0107e4e73
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/health.h
@@ -0,0 +1,58 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/* Copyright (c) 2019 Mellanox Technologies. */
+
+#ifndef __MLX5E_EN_HEALTH_H
+#define __MLX5E_EN_HEALTH_H
+
+#include "en.h"
+#include "diag/rsc_dump.h"
+
+static inline bool cqe_syndrome_needs_recover(u8 syndrome)
+{
+	return syndrome == MLX5_CQE_SYNDROME_LOCAL_QP_OP_ERR ||
+	       syndrome == MLX5_CQE_SYNDROME_LOCAL_PROT_ERR ||
+	       syndrome == MLX5_CQE_SYNDROME_WR_FLUSH_ERR;
+}
+
+void mlx5e_reporter_tx_create(struct mlx5e_priv *priv);
+void mlx5e_reporter_tx_destroy(struct mlx5e_priv *priv);
+void mlx5e_reporter_tx_err_cqe(struct mlx5e_txqsq *sq);
+int mlx5e_reporter_tx_timeout(struct mlx5e_txqsq *sq);
+
+int mlx5e_health_cq_diag_fmsg(struct mlx5e_cq *cq, struct devlink_fmsg *fmsg);
+int mlx5e_health_cq_common_diag_fmsg(struct mlx5e_cq *cq, struct devlink_fmsg *fmsg);
+int mlx5e_health_eq_diag_fmsg(struct mlx5_eq_comp *eq, struct devlink_fmsg *fmsg);
+int mlx5e_health_fmsg_named_obj_nest_start(struct devlink_fmsg *fmsg, char *name);
+int mlx5e_health_fmsg_named_obj_nest_end(struct devlink_fmsg *fmsg);
+
+void mlx5e_reporter_rx_create(struct mlx5e_priv *priv);
+void mlx5e_reporter_rx_destroy(struct mlx5e_priv *priv);
+void mlx5e_reporter_icosq_cqe_err(struct mlx5e_icosq *icosq);
+void mlx5e_reporter_rq_cqe_err(struct mlx5e_rq *rq);
+void mlx5e_reporter_rx_timeout(struct mlx5e_rq *rq);
+void mlx5e_reporter_icosq_suspend_recovery(struct mlx5e_channel *c);
+void mlx5e_reporter_icosq_resume_recovery(struct mlx5e_channel *c);
+
+#define MLX5E_REPORTER_PER_Q_MAX_LEN 256
+
+struct mlx5e_err_ctx {
+	int (*recover)(void *ctx);
+	int (*dump)(struct mlx5e_priv *priv, struct devlink_fmsg *fmsg, void *ctx);
+	void *ctx;
+};
+
+int mlx5e_health_sq_to_ready(struct mlx5_core_dev *mdev, struct net_device *dev, u32 sqn);
+int mlx5e_health_channel_eq_recover(struct net_device *dev, struct mlx5_eq_comp *eq,
+				    struct mlx5e_ch_stats *stats);
+int mlx5e_health_recover_channels(struct mlx5e_priv *priv);
+int mlx5e_health_report(struct mlx5e_priv *priv,
+			struct devlink_health_reporter *reporter, char *err_str,
+			struct mlx5e_err_ctx *err_ctx);
+void mlx5e_health_create_reporters(struct mlx5e_priv *priv);
+void mlx5e_health_destroy_reporters(struct mlx5e_priv *priv);
+void mlx5e_health_channels_update(struct mlx5e_priv *priv);
+int mlx5e_health_rsc_fmsg_dump(struct mlx5e_priv *priv, struct mlx5_rsc_key *key,
+			       struct devlink_fmsg *fmsg);
+int mlx5e_health_queue_dump(struct mlx5e_priv *priv, struct devlink_fmsg *fmsg,
+			    int queue_idx, char *lbl);
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/htb.c b/drivers/net/ethernet/mellanox/mlx5/core/en/htb.c
new file mode 100644
index 000000000..09d441ecb
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/htb.c
@@ -0,0 +1,722 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2022, NVIDIA CORPORATION & AFFILIATES. All rights reserved. */
+
+#include <net/pkt_cls.h>
+#include "htb.h"
+#include "en.h"
+#include "../qos.h"
+
+struct mlx5e_qos_node {
+	struct hlist_node hnode;
+	struct mlx5e_qos_node *parent;
+	u64 rate;
+	u32 bw_share;
+	u32 max_average_bw;
+	u32 hw_id;
+	u32 classid; /* 16-bit, except root. */
+	u16 qid;
+};
+
+struct mlx5e_htb {
+	DECLARE_HASHTABLE(qos_tc2node, order_base_2(MLX5E_QOS_MAX_LEAF_NODES));
+	DECLARE_BITMAP(qos_used_qids, MLX5E_QOS_MAX_LEAF_NODES);
+	struct mlx5_core_dev *mdev;
+	struct net_device *netdev;
+	struct mlx5e_priv *priv;
+	struct mlx5e_selq *selq;
+};
+
+#define MLX5E_QOS_QID_INNER 0xffff
+#define MLX5E_HTB_CLASSID_ROOT 0xffffffff
+
+/* Software representation of the QoS tree */
+
+int mlx5e_htb_enumerate_leaves(struct mlx5e_htb *htb, mlx5e_fp_htb_enumerate callback, void *data)
+{
+	struct mlx5e_qos_node *node = NULL;
+	int bkt, err;
+
+	hash_for_each(htb->qos_tc2node, bkt, node, hnode) {
+		if (node->qid == MLX5E_QOS_QID_INNER)
+			continue;
+		err = callback(data, node->qid, node->hw_id);
+		if (err)
+			return err;
+	}
+	return 0;
+}
+
+int mlx5e_htb_cur_leaf_nodes(struct mlx5e_htb *htb)
+{
+	int last;
+
+	last = find_last_bit(htb->qos_used_qids, mlx5e_qos_max_leaf_nodes(htb->mdev));
+	return last == mlx5e_qos_max_leaf_nodes(htb->mdev) ? 0 : last + 1;
+}
+
+static int mlx5e_htb_find_unused_qos_qid(struct mlx5e_htb *htb)
+{
+	int size = mlx5e_qos_max_leaf_nodes(htb->mdev);
+	struct mlx5e_priv *priv = htb->priv;
+	int res;
+
+	WARN_ONCE(!mutex_is_locked(&priv->state_lock), "%s: state_lock is not held\n", __func__);
+	res = find_first_zero_bit(htb->qos_used_qids, size);
+
+	return res == size ? -ENOSPC : res;
+}
+
+static struct mlx5e_qos_node *
+mlx5e_htb_node_create_leaf(struct mlx5e_htb *htb, u16 classid, u16 qid,
+			   struct mlx5e_qos_node *parent)
+{
+	struct mlx5e_qos_node *node;
+
+	node = kzalloc(sizeof(*node), GFP_KERNEL);
+	if (!node)
+		return ERR_PTR(-ENOMEM);
+
+	node->parent = parent;
+
+	node->qid = qid;
+	__set_bit(qid, htb->qos_used_qids);
+
+	node->classid = classid;
+	hash_add_rcu(htb->qos_tc2node, &node->hnode, classid);
+
+	mlx5e_update_tx_netdev_queues(htb->priv);
+
+	return node;
+}
+
+static struct mlx5e_qos_node *mlx5e_htb_node_create_root(struct mlx5e_htb *htb)
+{
+	struct mlx5e_qos_node *node;
+
+	node = kzalloc(sizeof(*node), GFP_KERNEL);
+	if (!node)
+		return ERR_PTR(-ENOMEM);
+
+	node->qid = MLX5E_QOS_QID_INNER;
+	node->classid = MLX5E_HTB_CLASSID_ROOT;
+	hash_add_rcu(htb->qos_tc2node, &node->hnode, node->classid);
+
+	return node;
+}
+
+static struct mlx5e_qos_node *mlx5e_htb_node_find(struct mlx5e_htb *htb, u32 classid)
+{
+	struct mlx5e_qos_node *node = NULL;
+
+	hash_for_each_possible(htb->qos_tc2node, node, hnode, classid) {
+		if (node->classid == classid)
+			break;
+	}
+
+	return node;
+}
+
+static struct mlx5e_qos_node *mlx5e_htb_node_find_rcu(struct mlx5e_htb *htb, u32 classid)
+{
+	struct mlx5e_qos_node *node = NULL;
+
+	hash_for_each_possible_rcu(htb->qos_tc2node, node, hnode, classid) {
+		if (node->classid == classid)
+			break;
+	}
+
+	return node;
+}
+
+static void mlx5e_htb_node_delete(struct mlx5e_htb *htb, struct mlx5e_qos_node *node)
+{
+	hash_del_rcu(&node->hnode);
+	if (node->qid != MLX5E_QOS_QID_INNER) {
+		__clear_bit(node->qid, htb->qos_used_qids);
+		mlx5e_update_tx_netdev_queues(htb->priv);
+	}
+	/* Make sure this qid is no longer selected by mlx5e_select_queue, so
+	 * that mlx5e_reactivate_qos_sq can safely restart the netdev TX queue.
+	 */
+	synchronize_net();
+	kfree(node);
+}
+
+/* TX datapath API */
+
+int mlx5e_htb_get_txq_by_classid(struct mlx5e_htb *htb, u16 classid)
+{
+	struct mlx5e_qos_node *node;
+	u16 qid;
+	int res;
+
+	rcu_read_lock();
+
+	node = mlx5e_htb_node_find_rcu(htb, classid);
+	if (!node) {
+		res = -ENOENT;
+		goto out;
+	}
+	qid = READ_ONCE(node->qid);
+	if (qid == MLX5E_QOS_QID_INNER) {
+		res = -EINVAL;
+		goto out;
+	}
+	res = mlx5e_qid_from_qos(&htb->priv->channels, qid);
+
+out:
+	rcu_read_unlock();
+	return res;
+}
+
+/* HTB TC handlers */
+
+static int
+mlx5e_htb_root_add(struct mlx5e_htb *htb, u16 htb_maj_id, u16 htb_defcls,
+		   struct netlink_ext_ack *extack)
+{
+	struct mlx5e_priv *priv = htb->priv;
+	struct mlx5e_qos_node *root;
+	bool opened;
+	int err;
+
+	qos_dbg(htb->mdev, "TC_HTB_CREATE handle %04x:, default :%04x\n", htb_maj_id, htb_defcls);
+
+	mlx5e_selq_prepare_htb(htb->selq, htb_maj_id, htb_defcls);
+
+	opened = test_bit(MLX5E_STATE_OPENED, &priv->state);
+	if (opened) {
+		err = mlx5e_qos_alloc_queues(priv, &priv->channels);
+		if (err)
+			goto err_cancel_selq;
+	}
+
+	root = mlx5e_htb_node_create_root(htb);
+	if (IS_ERR(root)) {
+		err = PTR_ERR(root);
+		goto err_free_queues;
+	}
+
+	err = mlx5_qos_create_root_node(htb->mdev, &root->hw_id);
+	if (err) {
+		NL_SET_ERR_MSG_MOD(extack, "Firmware error. Try upgrading firmware.");
+		goto err_sw_node_delete;
+	}
+
+	mlx5e_selq_apply(htb->selq);
+
+	return 0;
+
+err_sw_node_delete:
+	mlx5e_htb_node_delete(htb, root);
+
+err_free_queues:
+	if (opened)
+		mlx5e_qos_close_all_queues(&priv->channels);
+err_cancel_selq:
+	mlx5e_selq_cancel(htb->selq);
+	return err;
+}
+
+static int mlx5e_htb_root_del(struct mlx5e_htb *htb)
+{
+	struct mlx5e_priv *priv = htb->priv;
+	struct mlx5e_qos_node *root;
+	int err;
+
+	qos_dbg(htb->mdev, "TC_HTB_DESTROY\n");
+
+	/* Wait until real_num_tx_queues is updated for mlx5e_select_queue,
+	 * so that we can safely switch to its non-HTB non-PTP fastpath.
+	 */
+	synchronize_net();
+
+	mlx5e_selq_prepare_htb(htb->selq, 0, 0);
+	mlx5e_selq_apply(htb->selq);
+
+	root = mlx5e_htb_node_find(htb, MLX5E_HTB_CLASSID_ROOT);
+	if (!root) {
+		qos_err(htb->mdev, "Failed to find the root node in the QoS tree\n");
+		return -ENOENT;
+	}
+	err = mlx5_qos_destroy_node(htb->mdev, root->hw_id);
+	if (err)
+		qos_err(htb->mdev, "Failed to destroy root node %u, err = %d\n",
+			root->hw_id, err);
+	mlx5e_htb_node_delete(htb, root);
+
+	mlx5e_qos_deactivate_all_queues(&priv->channels);
+	mlx5e_qos_close_all_queues(&priv->channels);
+
+	return err;
+}
+
+static int mlx5e_htb_convert_rate(struct mlx5e_htb *htb, u64 rate,
+				  struct mlx5e_qos_node *parent, u32 *bw_share)
+{
+	u64 share = 0;
+
+	while (parent->classid != MLX5E_HTB_CLASSID_ROOT && !parent->max_average_bw)
+		parent = parent->parent;
+
+	if (parent->max_average_bw)
+		share = div64_u64(div_u64(rate * 100, BYTES_IN_MBIT),
+				  parent->max_average_bw);
+	else
+		share = 101;
+
+	*bw_share = share == 0 ? 1 : share > 100 ? 0 : share;
+
+	qos_dbg(htb->mdev, "Convert: rate %llu, parent ceil %llu -> bw_share %u\n",
+		rate, (u64)parent->max_average_bw * BYTES_IN_MBIT, *bw_share);
+
+	return 0;
+}
+
+static void mlx5e_htb_convert_ceil(struct mlx5e_htb *htb, u64 ceil, u32 *max_average_bw)
+{
+	/* Hardware treats 0 as "unlimited", set at least 1. */
+	*max_average_bw = max_t(u32, div_u64(ceil, BYTES_IN_MBIT), 1);
+
+	qos_dbg(htb->mdev, "Convert: ceil %llu -> max_average_bw %u\n",
+		ceil, *max_average_bw);
+}
+
+int
+mlx5e_htb_leaf_alloc_queue(struct mlx5e_htb *htb, u16 classid,
+			   u32 parent_classid, u64 rate, u64 ceil,
+			   struct netlink_ext_ack *extack)
+{
+	struct mlx5e_qos_node *node, *parent;
+	struct mlx5e_priv *priv = htb->priv;
+	int qid;
+	int err;
+
+	qos_dbg(htb->mdev, "TC_HTB_LEAF_ALLOC_QUEUE classid %04x, parent %04x, rate %llu, ceil %llu\n",
+		classid, parent_classid, rate, ceil);
+
+	qid = mlx5e_htb_find_unused_qos_qid(htb);
+	if (qid < 0) {
+		NL_SET_ERR_MSG_MOD(extack, "Maximum amount of leaf classes is reached.");
+		return qid;
+	}
+
+	parent = mlx5e_htb_node_find(htb, parent_classid);
+	if (!parent)
+		return -EINVAL;
+
+	node = mlx5e_htb_node_create_leaf(htb, classid, qid, parent);
+	if (IS_ERR(node))
+		return PTR_ERR(node);
+
+	node->rate = rate;
+	mlx5e_htb_convert_rate(htb, rate, node->parent, &node->bw_share);
+	mlx5e_htb_convert_ceil(htb, ceil, &node->max_average_bw);
+
+	err = mlx5_qos_create_leaf_node(htb->mdev, node->parent->hw_id,
+					node->bw_share, node->max_average_bw,
+					&node->hw_id);
+	if (err) {
+		NL_SET_ERR_MSG_MOD(extack, "Firmware error when creating a leaf node.");
+		qos_err(htb->mdev, "Failed to create a leaf node (class %04x), err = %d\n",
+			classid, err);
+		mlx5e_htb_node_delete(htb, node);
+		return err;
+	}
+
+	if (test_bit(MLX5E_STATE_OPENED, &priv->state)) {
+		err = mlx5e_open_qos_sq(priv, &priv->channels, node->qid, node->hw_id);
+		if (err) {
+			NL_SET_ERR_MSG_MOD(extack, "Error creating an SQ.");
+			qos_warn(htb->mdev, "Failed to create a QoS SQ (class %04x), err = %d\n",
+				 classid, err);
+		} else {
+			mlx5e_activate_qos_sq(priv, node->qid, node->hw_id);
+		}
+	}
+
+	return mlx5e_qid_from_qos(&priv->channels, node->qid);
+}
+
+int
+mlx5e_htb_leaf_to_inner(struct mlx5e_htb *htb, u16 classid, u16 child_classid,
+			u64 rate, u64 ceil, struct netlink_ext_ack *extack)
+{
+	struct mlx5e_qos_node *node, *child;
+	struct mlx5e_priv *priv = htb->priv;
+	int err, tmp_err;
+	u32 new_hw_id;
+	u16 qid;
+
+	qos_dbg(htb->mdev, "TC_HTB_LEAF_TO_INNER classid %04x, upcoming child %04x, rate %llu, ceil %llu\n",
+		classid, child_classid, rate, ceil);
+
+	node = mlx5e_htb_node_find(htb, classid);
+	if (!node)
+		return -ENOENT;
+
+	err = mlx5_qos_create_inner_node(htb->mdev, node->parent->hw_id,
+					 node->bw_share, node->max_average_bw,
+					 &new_hw_id);
+	if (err) {
+		NL_SET_ERR_MSG_MOD(extack, "Firmware error when creating an inner node.");
+		qos_err(htb->mdev, "Failed to create an inner node (class %04x), err = %d\n",
+			classid, err);
+		return err;
+	}
+
+	/* Intentionally reuse the qid for the upcoming first child. */
+	child = mlx5e_htb_node_create_leaf(htb, child_classid, node->qid, node);
+	if (IS_ERR(child)) {
+		err = PTR_ERR(child);
+		goto err_destroy_hw_node;
+	}
+
+	child->rate = rate;
+	mlx5e_htb_convert_rate(htb, rate, node, &child->bw_share);
+	mlx5e_htb_convert_ceil(htb, ceil, &child->max_average_bw);
+
+	err = mlx5_qos_create_leaf_node(htb->mdev, new_hw_id, child->bw_share,
+					child->max_average_bw, &child->hw_id);
+	if (err) {
+		NL_SET_ERR_MSG_MOD(extack, "Firmware error when creating a leaf node.");
+		qos_err(htb->mdev, "Failed to create a leaf node (class %04x), err = %d\n",
+			classid, err);
+		goto err_delete_sw_node;
+	}
+
+	/* No fail point. */
+
+	qid = node->qid;
+	/* Pairs with mlx5e_htb_get_txq_by_classid. */
+	WRITE_ONCE(node->qid, MLX5E_QOS_QID_INNER);
+
+	if (test_bit(MLX5E_STATE_OPENED, &priv->state)) {
+		mlx5e_deactivate_qos_sq(priv, qid);
+		mlx5e_close_qos_sq(priv, qid);
+	}
+
+	err = mlx5_qos_destroy_node(htb->mdev, node->hw_id);
+	if (err) /* Not fatal. */
+		qos_warn(htb->mdev, "Failed to destroy leaf node %u (class %04x), err = %d\n",
+			 node->hw_id, classid, err);
+
+	node->hw_id = new_hw_id;
+
+	if (test_bit(MLX5E_STATE_OPENED, &priv->state)) {
+		err = mlx5e_open_qos_sq(priv, &priv->channels, child->qid, child->hw_id);
+		if (err) {
+			NL_SET_ERR_MSG_MOD(extack, "Error creating an SQ.");
+			qos_warn(htb->mdev, "Failed to create a QoS SQ (class %04x), err = %d\n",
+				 classid, err);
+		} else {
+			mlx5e_activate_qos_sq(priv, child->qid, child->hw_id);
+		}
+	}
+
+	return 0;
+
+err_delete_sw_node:
+	child->qid = MLX5E_QOS_QID_INNER;
+	mlx5e_htb_node_delete(htb, child);
+
+err_destroy_hw_node:
+	tmp_err = mlx5_qos_destroy_node(htb->mdev, new_hw_id);
+	if (tmp_err) /* Not fatal. */
+		qos_warn(htb->mdev, "Failed to roll back creation of an inner node %u (class %04x), err = %d\n",
+			 new_hw_id, classid, tmp_err);
+	return err;
+}
+
+static struct mlx5e_qos_node *mlx5e_htb_node_find_by_qid(struct mlx5e_htb *htb, u16 qid)
+{
+	struct mlx5e_qos_node *node = NULL;
+	int bkt;
+
+	hash_for_each(htb->qos_tc2node, bkt, node, hnode)
+		if (node->qid == qid)
+			break;
+
+	return node;
+}
+
+int mlx5e_htb_leaf_del(struct mlx5e_htb *htb, u16 *classid,
+		       struct netlink_ext_ack *extack)
+{
+	struct mlx5e_priv *priv = htb->priv;
+	struct mlx5e_qos_node *node;
+	struct netdev_queue *txq;
+	u16 qid, moved_qid;
+	bool opened;
+	int err;
+
+	qos_dbg(htb->mdev, "TC_HTB_LEAF_DEL classid %04x\n", *classid);
+
+	node = mlx5e_htb_node_find(htb, *classid);
+	if (!node)
+		return -ENOENT;
+
+	/* Store qid for reuse. */
+	qid = node->qid;
+
+	opened = test_bit(MLX5E_STATE_OPENED, &priv->state);
+	if (opened) {
+		txq = netdev_get_tx_queue(htb->netdev,
+					  mlx5e_qid_from_qos(&priv->channels, qid));
+		mlx5e_deactivate_qos_sq(priv, qid);
+		mlx5e_close_qos_sq(priv, qid);
+	}
+
+	err = mlx5_qos_destroy_node(htb->mdev, node->hw_id);
+	if (err) /* Not fatal. */
+		qos_warn(htb->mdev, "Failed to destroy leaf node %u (class %04x), err = %d\n",
+			 node->hw_id, *classid, err);
+
+	mlx5e_htb_node_delete(htb, node);
+
+	moved_qid = mlx5e_htb_cur_leaf_nodes(htb);
+
+	if (moved_qid == 0) {
+		/* The last QoS SQ was just destroyed. */
+		if (opened)
+			mlx5e_reactivate_qos_sq(priv, qid, txq);
+		return 0;
+	}
+	moved_qid--;
+
+	if (moved_qid < qid) {
+		/* The highest QoS SQ was just destroyed. */
+		WARN(moved_qid != qid - 1, "Gaps in queue numeration: destroyed queue %u, the highest queue is %u",
+		     qid, moved_qid);
+		if (opened)
+			mlx5e_reactivate_qos_sq(priv, qid, txq);
+		return 0;
+	}
+
+	WARN(moved_qid == qid, "Can't move node with qid %u to itself", qid);
+	qos_dbg(htb->mdev, "Moving QoS SQ %u to %u\n", moved_qid, qid);
+
+	node = mlx5e_htb_node_find_by_qid(htb, moved_qid);
+	WARN(!node, "Could not find a node with qid %u to move to queue %u",
+	     moved_qid, qid);
+
+	/* Stop traffic to the old queue. */
+	WRITE_ONCE(node->qid, MLX5E_QOS_QID_INNER);
+	__clear_bit(moved_qid, priv->htb->qos_used_qids);
+
+	if (opened) {
+		txq = netdev_get_tx_queue(htb->netdev,
+					  mlx5e_qid_from_qos(&priv->channels, moved_qid));
+		mlx5e_deactivate_qos_sq(priv, moved_qid);
+		mlx5e_close_qos_sq(priv, moved_qid);
+	}
+
+	/* Prevent packets from the old class from getting into the new one. */
+	mlx5e_reset_qdisc(htb->netdev, moved_qid);
+
+	__set_bit(qid, htb->qos_used_qids);
+	WRITE_ONCE(node->qid, qid);
+
+	if (test_bit(MLX5E_STATE_OPENED, &priv->state)) {
+		err = mlx5e_open_qos_sq(priv, &priv->channels, node->qid, node->hw_id);
+		if (err) {
+			NL_SET_ERR_MSG_MOD(extack, "Error creating an SQ.");
+			qos_warn(htb->mdev, "Failed to create a QoS SQ (class %04x) while moving qid %u to %u, err = %d\n",
+				 node->classid, moved_qid, qid, err);
+		} else {
+			mlx5e_activate_qos_sq(priv, node->qid, node->hw_id);
+		}
+	}
+
+	mlx5e_update_tx_netdev_queues(priv);
+	if (opened)
+		mlx5e_reactivate_qos_sq(priv, moved_qid, txq);
+
+	*classid = node->classid;
+	return 0;
+}
+
+int
+mlx5e_htb_leaf_del_last(struct mlx5e_htb *htb, u16 classid, bool force,
+			struct netlink_ext_ack *extack)
+{
+	struct mlx5e_qos_node *node, *parent;
+	struct mlx5e_priv *priv = htb->priv;
+	u32 old_hw_id, new_hw_id;
+	int err, saved_err = 0;
+	u16 qid;
+
+	qos_dbg(htb->mdev, "TC_HTB_LEAF_DEL_LAST%s classid %04x\n",
+		force ? "_FORCE" : "", classid);
+
+	node = mlx5e_htb_node_find(htb, classid);
+	if (!node)
+		return -ENOENT;
+
+	err = mlx5_qos_create_leaf_node(htb->mdev, node->parent->parent->hw_id,
+					node->parent->bw_share,
+					node->parent->max_average_bw,
+					&new_hw_id);
+	if (err) {
+		NL_SET_ERR_MSG_MOD(extack, "Firmware error when creating a leaf node.");
+		qos_err(htb->mdev, "Failed to create a leaf node (class %04x), err = %d\n",
+			classid, err);
+		if (!force)
+			return err;
+		saved_err = err;
+	}
+
+	/* Store qid for reuse and prevent clearing the bit. */
+	qid = node->qid;
+	/* Pairs with mlx5e_htb_get_txq_by_classid. */
+	WRITE_ONCE(node->qid, MLX5E_QOS_QID_INNER);
+
+	if (test_bit(MLX5E_STATE_OPENED, &priv->state)) {
+		mlx5e_deactivate_qos_sq(priv, qid);
+		mlx5e_close_qos_sq(priv, qid);
+	}
+
+	/* Prevent packets from the old class from getting into the new one. */
+	mlx5e_reset_qdisc(htb->netdev, qid);
+
+	err = mlx5_qos_destroy_node(htb->mdev, node->hw_id);
+	if (err) /* Not fatal. */
+		qos_warn(htb->mdev, "Failed to destroy leaf node %u (class %04x), err = %d\n",
+			 node->hw_id, classid, err);
+
+	parent = node->parent;
+	mlx5e_htb_node_delete(htb, node);
+
+	node = parent;
+	WRITE_ONCE(node->qid, qid);
+
+	/* Early return on error in force mode. Parent will still be an inner
+	 * node to be deleted by a following delete operation.
+	 */
+	if (saved_err)
+		return saved_err;
+
+	old_hw_id = node->hw_id;
+	node->hw_id = new_hw_id;
+
+	if (test_bit(MLX5E_STATE_OPENED, &priv->state)) {
+		err = mlx5e_open_qos_sq(priv, &priv->channels, node->qid, node->hw_id);
+		if (err) {
+			NL_SET_ERR_MSG_MOD(extack, "Error creating an SQ.");
+			qos_warn(htb->mdev, "Failed to create a QoS SQ (class %04x), err = %d\n",
+				 classid, err);
+		} else {
+			mlx5e_activate_qos_sq(priv, node->qid, node->hw_id);
+		}
+	}
+
+	err = mlx5_qos_destroy_node(htb->mdev, old_hw_id);
+	if (err) /* Not fatal. */
+		qos_warn(htb->mdev, "Failed to destroy leaf node %u (class %04x), err = %d\n",
+			 node->hw_id, classid, err);
+
+	return 0;
+}
+
+static int
+mlx5e_htb_update_children(struct mlx5e_htb *htb, struct mlx5e_qos_node *node,
+			  struct netlink_ext_ack *extack)
+{
+	struct mlx5e_qos_node *child;
+	int err = 0;
+	int bkt;
+
+	hash_for_each(htb->qos_tc2node, bkt, child, hnode) {
+		u32 old_bw_share = child->bw_share;
+		int err_one;
+
+		if (child->parent != node)
+			continue;
+
+		mlx5e_htb_convert_rate(htb, child->rate, node, &child->bw_share);
+		if (child->bw_share == old_bw_share)
+			continue;
+
+		err_one = mlx5_qos_update_node(htb->mdev, child->bw_share,
+					       child->max_average_bw, child->hw_id);
+		if (!err && err_one) {
+			err = err_one;
+
+			NL_SET_ERR_MSG_MOD(extack, "Firmware error when modifying a child node.");
+			qos_err(htb->mdev, "Failed to modify a child node (class %04x), err = %d\n",
+				node->classid, err);
+		}
+	}
+
+	return err;
+}
+
+int
+mlx5e_htb_node_modify(struct mlx5e_htb *htb, u16 classid, u64 rate, u64 ceil,
+		      struct netlink_ext_ack *extack)
+{
+	u32 bw_share, max_average_bw;
+	struct mlx5e_qos_node *node;
+	bool ceil_changed = false;
+	int err;
+
+	qos_dbg(htb->mdev, "TC_HTB_LEAF_MODIFY classid %04x, rate %llu, ceil %llu\n",
+		classid, rate, ceil);
+
+	node = mlx5e_htb_node_find(htb, classid);
+	if (!node)
+		return -ENOENT;
+
+	node->rate = rate;
+	mlx5e_htb_convert_rate(htb, rate, node->parent, &bw_share);
+	mlx5e_htb_convert_ceil(htb, ceil, &max_average_bw);
+
+	err = mlx5_qos_update_node(htb->mdev, bw_share,
+				   max_average_bw, node->hw_id);
+	if (err) {
+		NL_SET_ERR_MSG_MOD(extack, "Firmware error when modifying a node.");
+		qos_err(htb->mdev, "Failed to modify a node (class %04x), err = %d\n",
+			classid, err);
+		return err;
+	}
+
+	if (max_average_bw != node->max_average_bw)
+		ceil_changed = true;
+
+	node->bw_share = bw_share;
+	node->max_average_bw = max_average_bw;
+
+	if (ceil_changed)
+		err = mlx5e_htb_update_children(htb, node, extack);
+
+	return err;
+}
+
+struct mlx5e_htb *mlx5e_htb_alloc(void)
+{
+	return kvzalloc(sizeof(struct mlx5e_htb), GFP_KERNEL);
+}
+
+void mlx5e_htb_free(struct mlx5e_htb *htb)
+{
+	kvfree(htb);
+}
+
+int mlx5e_htb_init(struct mlx5e_htb *htb, struct tc_htb_qopt_offload *htb_qopt,
+		   struct net_device *netdev, struct mlx5_core_dev *mdev,
+		   struct mlx5e_selq *selq, struct mlx5e_priv *priv)
+{
+	htb->mdev = mdev;
+	htb->netdev = netdev;
+	htb->selq = selq;
+	htb->priv = priv;
+	hash_init(htb->qos_tc2node);
+	return mlx5e_htb_root_add(htb, htb_qopt->parent_classid, htb_qopt->classid,
+				  htb_qopt->extack);
+}
+
+void mlx5e_htb_cleanup(struct mlx5e_htb *htb)
+{
+	mlx5e_htb_root_del(htb);
+}
+
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/htb.h b/drivers/net/ethernet/mellanox/mlx5/core/en/htb.h
new file mode 100644
index 000000000..8386f1ea4
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/htb.h
@@ -0,0 +1,46 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2022, NVIDIA CORPORATION & AFFILIATES. All rights reserved. */
+
+#ifndef __MLX5E_EN_HTB_H_
+#define __MLX5E_EN_HTB_H_
+
+#include "qos.h"
+
+#define MLX5E_QOS_MAX_LEAF_NODES 256
+
+struct mlx5e_selq;
+struct mlx5e_htb;
+
+typedef int (*mlx5e_fp_htb_enumerate)(void *data, u16 qid, u32 hw_id);
+int mlx5e_htb_enumerate_leaves(struct mlx5e_htb *htb, mlx5e_fp_htb_enumerate callback, void *data);
+
+int mlx5e_htb_cur_leaf_nodes(struct mlx5e_htb *htb);
+
+/* TX datapath API */
+int mlx5e_htb_get_txq_by_classid(struct mlx5e_htb *htb, u16 classid);
+
+/* HTB TC handlers */
+
+int
+mlx5e_htb_leaf_alloc_queue(struct mlx5e_htb *htb, u16 classid,
+			   u32 parent_classid, u64 rate, u64 ceil,
+			   struct netlink_ext_ack *extack);
+int
+mlx5e_htb_leaf_to_inner(struct mlx5e_htb *htb, u16 classid, u16 child_classid,
+			u64 rate, u64 ceil, struct netlink_ext_ack *extack);
+int mlx5e_htb_leaf_del(struct mlx5e_htb *htb, u16 *classid,
+		       struct netlink_ext_ack *extack);
+int
+mlx5e_htb_leaf_del_last(struct mlx5e_htb *htb, u16 classid, bool force,
+			struct netlink_ext_ack *extack);
+int
+mlx5e_htb_node_modify(struct mlx5e_htb *htb, u16 classid, u64 rate, u64 ceil,
+		      struct netlink_ext_ack *extack);
+struct mlx5e_htb *mlx5e_htb_alloc(void);
+void mlx5e_htb_free(struct mlx5e_htb *htb);
+int mlx5e_htb_init(struct mlx5e_htb *htb, struct tc_htb_qopt_offload *htb_qopt,
+		   struct net_device *netdev, struct mlx5_core_dev *mdev,
+		   struct mlx5e_selq *selq, struct mlx5e_priv *priv);
+void mlx5e_htb_cleanup(struct mlx5e_htb *htb);
+#endif
+
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/hv_vhca_stats.c b/drivers/net/ethernet/mellanox/mlx5/core/en/hv_vhca_stats.c
new file mode 100644
index 000000000..b4f3bd7d3
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/hv_vhca_stats.c
@@ -0,0 +1,159 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+// Copyright (c) 2018 Mellanox Technologies
+
+#include "en.h"
+#include "en/hv_vhca_stats.h"
+#include "lib/hv_vhca.h"
+#include "lib/hv.h"
+
+struct mlx5e_hv_vhca_per_ring_stats {
+	u64     rx_packets;
+	u64     rx_bytes;
+	u64     tx_packets;
+	u64     tx_bytes;
+};
+
+static void
+mlx5e_hv_vhca_fill_ring_stats(struct mlx5e_priv *priv, int ch,
+			      struct mlx5e_hv_vhca_per_ring_stats *data)
+{
+	struct mlx5e_channel_stats *stats;
+	int tc;
+
+	stats = priv->channel_stats[ch];
+	data->rx_packets = stats->rq.packets;
+	data->rx_bytes   = stats->rq.bytes;
+
+	for (tc = 0; tc < priv->max_opened_tc; tc++) {
+		data->tx_packets += stats->sq[tc].packets;
+		data->tx_bytes   += stats->sq[tc].bytes;
+	}
+}
+
+static void mlx5e_hv_vhca_fill_stats(struct mlx5e_priv *priv, void *data,
+				     int buf_len)
+{
+	int ch, i = 0;
+
+	for (ch = 0; ch < priv->stats_nch; ch++) {
+		void *buf = data + i;
+
+		if (WARN_ON_ONCE(buf +
+				 sizeof(struct mlx5e_hv_vhca_per_ring_stats) >
+				 data + buf_len))
+			return;
+
+		mlx5e_hv_vhca_fill_ring_stats(priv, ch, buf);
+		i += sizeof(struct mlx5e_hv_vhca_per_ring_stats);
+	}
+}
+
+static int mlx5e_hv_vhca_stats_buf_size(struct mlx5e_priv *priv)
+{
+	return (sizeof(struct mlx5e_hv_vhca_per_ring_stats) *
+		priv->stats_nch);
+}
+
+static void mlx5e_hv_vhca_stats_work(struct work_struct *work)
+{
+	struct mlx5e_hv_vhca_stats_agent *sagent;
+	struct mlx5_hv_vhca_agent *agent;
+	struct delayed_work *dwork;
+	struct mlx5e_priv *priv;
+	int buf_len, rc;
+	void *buf;
+
+	dwork = to_delayed_work(work);
+	sagent = container_of(dwork, struct mlx5e_hv_vhca_stats_agent, work);
+	priv = container_of(sagent, struct mlx5e_priv, stats_agent);
+	buf_len = mlx5e_hv_vhca_stats_buf_size(priv);
+	agent = sagent->agent;
+	buf = sagent->buf;
+
+	memset(buf, 0, buf_len);
+	mlx5e_hv_vhca_fill_stats(priv, buf, buf_len);
+
+	rc = mlx5_hv_vhca_agent_write(agent, buf, buf_len);
+	if (rc) {
+		mlx5_core_err(priv->mdev,
+			      "%s: Failed to write stats, err = %d\n",
+			      __func__, rc);
+		return;
+	}
+
+	if (sagent->delay)
+		queue_delayed_work(priv->wq, &sagent->work, sagent->delay);
+}
+
+enum {
+	MLX5_HV_VHCA_STATS_VERSION     = 1,
+	MLX5_HV_VHCA_STATS_UPDATE_ONCE = 0xFFFF,
+};
+
+static void mlx5e_hv_vhca_stats_control(struct mlx5_hv_vhca_agent *agent,
+					struct mlx5_hv_vhca_control_block *block)
+{
+	struct mlx5e_hv_vhca_stats_agent *sagent;
+	struct mlx5e_priv *priv;
+
+	priv = mlx5_hv_vhca_agent_priv(agent);
+	sagent = &priv->stats_agent;
+
+	block->version = MLX5_HV_VHCA_STATS_VERSION;
+	block->rings   = priv->stats_nch;
+
+	if (!block->command) {
+		cancel_delayed_work_sync(&priv->stats_agent.work);
+		return;
+	}
+
+	sagent->delay = block->command == MLX5_HV_VHCA_STATS_UPDATE_ONCE ? 0 :
+			msecs_to_jiffies(block->command * 100);
+
+	queue_delayed_work(priv->wq, &sagent->work, sagent->delay);
+}
+
+static void mlx5e_hv_vhca_stats_cleanup(struct mlx5_hv_vhca_agent *agent)
+{
+	struct mlx5e_priv *priv = mlx5_hv_vhca_agent_priv(agent);
+
+	cancel_delayed_work_sync(&priv->stats_agent.work);
+}
+
+void mlx5e_hv_vhca_stats_create(struct mlx5e_priv *priv)
+{
+	int buf_len = mlx5e_hv_vhca_stats_buf_size(priv);
+	struct mlx5_hv_vhca_agent *agent;
+
+	priv->stats_agent.buf = kvzalloc(buf_len, GFP_KERNEL);
+	if (!priv->stats_agent.buf)
+		return;
+
+	agent = mlx5_hv_vhca_agent_create(priv->mdev->hv_vhca,
+					  MLX5_HV_VHCA_AGENT_STATS,
+					  mlx5e_hv_vhca_stats_control, NULL,
+					  mlx5e_hv_vhca_stats_cleanup,
+					  priv);
+
+	if (IS_ERR_OR_NULL(agent)) {
+		if (IS_ERR(agent))
+			netdev_warn(priv->netdev,
+				    "Failed to create hv vhca stats agent, err = %ld\n",
+				    PTR_ERR(agent));
+
+		kvfree(priv->stats_agent.buf);
+		return;
+	}
+
+	priv->stats_agent.agent = agent;
+	INIT_DELAYED_WORK(&priv->stats_agent.work, mlx5e_hv_vhca_stats_work);
+}
+
+void mlx5e_hv_vhca_stats_destroy(struct mlx5e_priv *priv)
+{
+	if (IS_ERR_OR_NULL(priv->stats_agent.agent))
+		return;
+
+	mlx5_hv_vhca_agent_destroy(priv->stats_agent.agent);
+	kvfree(priv->stats_agent.buf);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/hv_vhca_stats.h b/drivers/net/ethernet/mellanox/mlx5/core/en/hv_vhca_stats.h
new file mode 100644
index 000000000..29c8c6d32
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/hv_vhca_stats.h
@@ -0,0 +1,18 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2019 Mellanox Technologies. */
+
+#ifndef __MLX5_EN_STATS_VHCA_H__
+#define __MLX5_EN_STATS_VHCA_H__
+#include "en.h"
+
+#if IS_ENABLED(CONFIG_PCI_HYPERV_INTERFACE)
+
+void mlx5e_hv_vhca_stats_create(struct mlx5e_priv *priv);
+void mlx5e_hv_vhca_stats_destroy(struct mlx5e_priv *priv);
+
+#else
+static inline void mlx5e_hv_vhca_stats_create(struct mlx5e_priv *priv) {}
+static inline void mlx5e_hv_vhca_stats_destroy(struct mlx5e_priv *priv) {}
+#endif
+
+#endif /* __MLX5_EN_STATS_VHCA_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/mapping.c b/drivers/net/ethernet/mellanox/mlx5/core/en/mapping.c
new file mode 100644
index 000000000..4e72ca807
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/mapping.c
@@ -0,0 +1,263 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2018 Mellanox Technologies */
+
+#include <linux/jhash.h>
+#include <linux/slab.h>
+#include <linux/xarray.h>
+#include <linux/hashtable.h>
+#include <linux/refcount.h>
+
+#include "mapping.h"
+
+#define MAPPING_GRACE_PERIOD 2000
+
+static LIST_HEAD(shared_ctx_list);
+static DEFINE_MUTEX(shared_ctx_lock);
+
+struct mapping_ctx {
+	struct xarray xarray;
+	DECLARE_HASHTABLE(ht, 8);
+	struct mutex lock; /* Guards hashtable and xarray */
+	unsigned long max_id;
+	size_t data_size;
+	bool delayed_removal;
+	struct delayed_work dwork;
+	struct list_head pending_list;
+	spinlock_t pending_list_lock; /* Guards pending list */
+	u64 id;
+	u8 type;
+	struct list_head list;
+	refcount_t refcount;
+};
+
+struct mapping_item {
+	struct rcu_head rcu;
+	struct list_head list;
+	unsigned long timeout;
+	struct hlist_node node;
+	int cnt;
+	u32 id;
+	char data[];
+};
+
+int mapping_add(struct mapping_ctx *ctx, void *data, u32 *id)
+{
+	struct mapping_item *mi;
+	int err = -ENOMEM;
+	u32 hash_key;
+
+	mutex_lock(&ctx->lock);
+
+	hash_key = jhash(data, ctx->data_size, 0);
+	hash_for_each_possible(ctx->ht, mi, node, hash_key) {
+		if (!memcmp(data, mi->data, ctx->data_size))
+			goto attach;
+	}
+
+	mi = kzalloc(sizeof(*mi) + ctx->data_size, GFP_KERNEL);
+	if (!mi)
+		goto err_alloc;
+
+	memcpy(mi->data, data, ctx->data_size);
+	hash_add(ctx->ht, &mi->node, hash_key);
+
+	err = xa_alloc(&ctx->xarray, &mi->id, mi, XA_LIMIT(1, ctx->max_id),
+		       GFP_KERNEL);
+	if (err)
+		goto err_assign;
+attach:
+	++mi->cnt;
+	*id = mi->id;
+
+	mutex_unlock(&ctx->lock);
+
+	return 0;
+
+err_assign:
+	hash_del(&mi->node);
+	kfree(mi);
+err_alloc:
+	mutex_unlock(&ctx->lock);
+
+	return err;
+}
+
+static void mapping_remove_and_free(struct mapping_ctx *ctx,
+				    struct mapping_item *mi)
+{
+	xa_erase(&ctx->xarray, mi->id);
+	kfree_rcu(mi, rcu);
+}
+
+static void mapping_free_item(struct mapping_ctx *ctx,
+			      struct mapping_item *mi)
+{
+	if (!ctx->delayed_removal) {
+		mapping_remove_and_free(ctx, mi);
+		return;
+	}
+
+	mi->timeout = jiffies + msecs_to_jiffies(MAPPING_GRACE_PERIOD);
+
+	spin_lock(&ctx->pending_list_lock);
+	list_add_tail(&mi->list, &ctx->pending_list);
+	spin_unlock(&ctx->pending_list_lock);
+
+	schedule_delayed_work(&ctx->dwork, MAPPING_GRACE_PERIOD);
+}
+
+int mapping_remove(struct mapping_ctx *ctx, u32 id)
+{
+	unsigned long index = id;
+	struct mapping_item *mi;
+	int err = -ENOENT;
+
+	mutex_lock(&ctx->lock);
+	mi = xa_load(&ctx->xarray, index);
+	if (!mi)
+		goto out;
+	err = 0;
+
+	if (--mi->cnt > 0)
+		goto out;
+
+	hash_del(&mi->node);
+	mapping_free_item(ctx, mi);
+out:
+	mutex_unlock(&ctx->lock);
+
+	return err;
+}
+
+int mapping_find(struct mapping_ctx *ctx, u32 id, void *data)
+{
+	unsigned long index = id;
+	struct mapping_item *mi;
+	int err = -ENOENT;
+
+	rcu_read_lock();
+	mi = xa_load(&ctx->xarray, index);
+	if (!mi)
+		goto err_find;
+
+	memcpy(data, mi->data, ctx->data_size);
+	err = 0;
+
+err_find:
+	rcu_read_unlock();
+	return err;
+}
+
+static void
+mapping_remove_and_free_list(struct mapping_ctx *ctx, struct list_head *list)
+{
+	struct mapping_item *mi;
+
+	list_for_each_entry(mi, list, list)
+		mapping_remove_and_free(ctx, mi);
+}
+
+static void mapping_work_handler(struct work_struct *work)
+{
+	unsigned long min_timeout = 0, now = jiffies;
+	struct mapping_item *mi, *next;
+	LIST_HEAD(pending_items);
+	struct mapping_ctx *ctx;
+
+	ctx = container_of(work, struct mapping_ctx, dwork.work);
+
+	spin_lock(&ctx->pending_list_lock);
+	list_for_each_entry_safe(mi, next, &ctx->pending_list, list) {
+		if (time_after(now, mi->timeout))
+			list_move(&mi->list, &pending_items);
+		else if (!min_timeout ||
+			 time_before(mi->timeout, min_timeout))
+			min_timeout = mi->timeout;
+	}
+	spin_unlock(&ctx->pending_list_lock);
+
+	mapping_remove_and_free_list(ctx, &pending_items);
+
+	if (min_timeout)
+		schedule_delayed_work(&ctx->dwork, abs(min_timeout - now));
+}
+
+static void mapping_flush_work(struct mapping_ctx *ctx)
+{
+	if (!ctx->delayed_removal)
+		return;
+
+	cancel_delayed_work_sync(&ctx->dwork);
+	mapping_remove_and_free_list(ctx, &ctx->pending_list);
+}
+
+struct mapping_ctx *
+mapping_create(size_t data_size, u32 max_id, bool delayed_removal)
+{
+	struct mapping_ctx *ctx;
+
+	ctx = kzalloc(sizeof(*ctx), GFP_KERNEL);
+	if (!ctx)
+		return ERR_PTR(-ENOMEM);
+
+	ctx->max_id = max_id ? max_id : UINT_MAX;
+	ctx->data_size = data_size;
+
+	if (delayed_removal) {
+		INIT_DELAYED_WORK(&ctx->dwork, mapping_work_handler);
+		INIT_LIST_HEAD(&ctx->pending_list);
+		spin_lock_init(&ctx->pending_list_lock);
+		ctx->delayed_removal = true;
+	}
+
+	mutex_init(&ctx->lock);
+	xa_init_flags(&ctx->xarray, XA_FLAGS_ALLOC1);
+
+	refcount_set(&ctx->refcount, 1);
+	INIT_LIST_HEAD(&ctx->list);
+
+	return ctx;
+}
+
+struct mapping_ctx *
+mapping_create_for_id(u64 id, u8 type, size_t data_size, u32 max_id, bool delayed_removal)
+{
+	struct mapping_ctx *ctx;
+
+	mutex_lock(&shared_ctx_lock);
+	list_for_each_entry(ctx, &shared_ctx_list, list) {
+		if (ctx->id == id && ctx->type == type) {
+			if (refcount_inc_not_zero(&ctx->refcount))
+				goto unlock;
+			break;
+		}
+	}
+
+	ctx = mapping_create(data_size, max_id, delayed_removal);
+	if (IS_ERR(ctx))
+		goto unlock;
+
+	ctx->id = id;
+	ctx->type = type;
+	list_add(&ctx->list, &shared_ctx_list);
+
+unlock:
+	mutex_unlock(&shared_ctx_lock);
+	return ctx;
+}
+
+void mapping_destroy(struct mapping_ctx *ctx)
+{
+	if (!refcount_dec_and_test(&ctx->refcount))
+		return;
+
+	mutex_lock(&shared_ctx_lock);
+	list_del(&ctx->list);
+	mutex_unlock(&shared_ctx_lock);
+
+	mapping_flush_work(ctx);
+	xa_destroy(&ctx->xarray);
+	mutex_destroy(&ctx->lock);
+
+	kfree(ctx);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/mapping.h b/drivers/net/ethernet/mellanox/mlx5/core/en/mapping.h
new file mode 100644
index 000000000..4e2119f0f
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/mapping.h
@@ -0,0 +1,32 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2019 Mellanox Technologies */
+
+#ifndef __MLX5_MAPPING_H__
+#define __MLX5_MAPPING_H__
+
+struct mapping_ctx;
+
+int mapping_add(struct mapping_ctx *ctx, void *data, u32 *id);
+int mapping_remove(struct mapping_ctx *ctx, u32 id);
+int mapping_find(struct mapping_ctx *ctx, u32 id, void *data);
+
+/* mapping uses an xarray to map data to ids in add(), and for find().
+ * For locking, it uses a internal xarray spin lock for add()/remove(),
+ * find() uses rcu_read_lock().
+ * Choosing delayed_removal postpones the removal of a previously mapped
+ * id by MAPPING_GRACE_PERIOD milliseconds.
+ * This is to avoid races against hardware, where we mark the packet in
+ * hardware with a previous id, and quick remove() and add() reusing the same
+ * previous id. Then find() will get the new mapping instead of the old
+ * which was used to mark the packet.
+ */
+struct mapping_ctx *mapping_create(size_t data_size, u32 max_id,
+				   bool delayed_removal);
+void mapping_destroy(struct mapping_ctx *ctx);
+
+/* adds mapping with an id or get an existing mapping with the same id
+ */
+struct mapping_ctx *
+mapping_create_for_id(u64 id, u8 type, size_t data_size, u32 max_id, bool delayed_removal);
+
+#endif /* __MLX5_MAPPING_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/mod_hdr.c b/drivers/net/ethernet/mellanox/mlx5/core/en/mod_hdr.c
new file mode 100644
index 000000000..17325c5d6
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/mod_hdr.c
@@ -0,0 +1,215 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+// Copyright (c) 2020 Mellanox Technologies
+
+#include <linux/jhash.h>
+#include "mod_hdr.h"
+
+#define MLX5_MH_ACT_SZ MLX5_UN_SZ_BYTES(set_add_copy_action_in_auto)
+
+struct mod_hdr_key {
+	int num_actions;
+	void *actions;
+};
+
+struct mlx5e_mod_hdr_handle {
+	/* a node of a hash table which keeps all the mod_hdr entries */
+	struct hlist_node mod_hdr_hlist;
+
+	struct mod_hdr_key key;
+
+	struct mlx5_modify_hdr *modify_hdr;
+
+	refcount_t refcnt;
+	struct completion res_ready;
+	int compl_result;
+};
+
+static u32 hash_mod_hdr_info(struct mod_hdr_key *key)
+{
+	return jhash(key->actions,
+		     key->num_actions * MLX5_MH_ACT_SZ, 0);
+}
+
+static int cmp_mod_hdr_info(struct mod_hdr_key *a, struct mod_hdr_key *b)
+{
+	if (a->num_actions != b->num_actions)
+		return 1;
+
+	return memcmp(a->actions, b->actions,
+		      a->num_actions * MLX5_MH_ACT_SZ);
+}
+
+void mlx5e_mod_hdr_tbl_init(struct mod_hdr_tbl *tbl)
+{
+	mutex_init(&tbl->lock);
+	hash_init(tbl->hlist);
+}
+
+void mlx5e_mod_hdr_tbl_destroy(struct mod_hdr_tbl *tbl)
+{
+	mutex_destroy(&tbl->lock);
+}
+
+static struct mlx5e_mod_hdr_handle *mod_hdr_get(struct mod_hdr_tbl *tbl,
+						struct mod_hdr_key *key,
+						u32 hash_key)
+{
+	struct mlx5e_mod_hdr_handle *mh, *found = NULL;
+
+	hash_for_each_possible(tbl->hlist, mh, mod_hdr_hlist, hash_key) {
+		if (!cmp_mod_hdr_info(&mh->key, key)) {
+			refcount_inc(&mh->refcnt);
+			found = mh;
+			break;
+		}
+	}
+
+	return found;
+}
+
+struct mlx5e_mod_hdr_handle *
+mlx5e_mod_hdr_attach(struct mlx5_core_dev *mdev,
+		     struct mod_hdr_tbl *tbl,
+		     enum mlx5_flow_namespace_type namespace,
+		     struct mlx5e_tc_mod_hdr_acts *mod_hdr_acts)
+{
+	int num_actions, actions_size, err;
+	struct mlx5e_mod_hdr_handle *mh;
+	struct mod_hdr_key key;
+	u32 hash_key;
+
+	num_actions  = mod_hdr_acts->num_actions;
+	actions_size = MLX5_MH_ACT_SZ * num_actions;
+
+	key.actions = mod_hdr_acts->actions;
+	key.num_actions = num_actions;
+
+	hash_key = hash_mod_hdr_info(&key);
+
+	mutex_lock(&tbl->lock);
+	mh = mod_hdr_get(tbl, &key, hash_key);
+	if (mh) {
+		mutex_unlock(&tbl->lock);
+		wait_for_completion(&mh->res_ready);
+
+		if (mh->compl_result < 0) {
+			err = -EREMOTEIO;
+			goto attach_header_err;
+		}
+		goto attach_header;
+	}
+
+	mh = kzalloc(sizeof(*mh) + actions_size, GFP_KERNEL);
+	if (!mh) {
+		mutex_unlock(&tbl->lock);
+		return ERR_PTR(-ENOMEM);
+	}
+
+	mh->key.actions = (void *)mh + sizeof(*mh);
+	memcpy(mh->key.actions, key.actions, actions_size);
+	mh->key.num_actions = num_actions;
+	refcount_set(&mh->refcnt, 1);
+	init_completion(&mh->res_ready);
+
+	hash_add(tbl->hlist, &mh->mod_hdr_hlist, hash_key);
+	mutex_unlock(&tbl->lock);
+
+	mh->modify_hdr = mlx5_modify_header_alloc(mdev, namespace,
+						  mh->key.num_actions,
+						  mh->key.actions);
+	if (IS_ERR(mh->modify_hdr)) {
+		err = PTR_ERR(mh->modify_hdr);
+		mh->compl_result = err;
+		goto alloc_header_err;
+	}
+	mh->compl_result = 1;
+	complete_all(&mh->res_ready);
+
+attach_header:
+	return mh;
+
+alloc_header_err:
+	complete_all(&mh->res_ready);
+attach_header_err:
+	mlx5e_mod_hdr_detach(mdev, tbl, mh);
+	return ERR_PTR(err);
+}
+
+void mlx5e_mod_hdr_detach(struct mlx5_core_dev *mdev,
+			  struct mod_hdr_tbl *tbl,
+			  struct mlx5e_mod_hdr_handle *mh)
+{
+	if (!refcount_dec_and_mutex_lock(&mh->refcnt, &tbl->lock))
+		return;
+	hash_del(&mh->mod_hdr_hlist);
+	mutex_unlock(&tbl->lock);
+
+	if (mh->compl_result > 0)
+		mlx5_modify_header_dealloc(mdev, mh->modify_hdr);
+
+	kfree(mh);
+}
+
+struct mlx5_modify_hdr *mlx5e_mod_hdr_get(struct mlx5e_mod_hdr_handle *mh)
+{
+	return mh->modify_hdr;
+}
+
+char *
+mlx5e_mod_hdr_alloc(struct mlx5_core_dev *mdev, int namespace,
+		    struct mlx5e_tc_mod_hdr_acts *mod_hdr_acts)
+{
+	int new_num_actions, max_hw_actions;
+	size_t new_sz, old_sz;
+	void *ret;
+
+	if (mod_hdr_acts->num_actions < mod_hdr_acts->max_actions)
+		goto out;
+
+	max_hw_actions = mlx5e_mod_hdr_max_actions(mdev, namespace);
+	new_num_actions = min(max_hw_actions,
+			      mod_hdr_acts->actions ?
+			      mod_hdr_acts->max_actions * 2 : 1);
+	if (mod_hdr_acts->max_actions == new_num_actions)
+		return ERR_PTR(-ENOSPC);
+
+	new_sz = MLX5_MH_ACT_SZ * new_num_actions;
+	old_sz = mod_hdr_acts->max_actions * MLX5_MH_ACT_SZ;
+
+	if (mod_hdr_acts->is_static) {
+		ret = kzalloc(new_sz, GFP_KERNEL);
+		if (ret) {
+			memcpy(ret, mod_hdr_acts->actions, old_sz);
+			mod_hdr_acts->is_static = false;
+		}
+	} else {
+		ret = krealloc(mod_hdr_acts->actions, new_sz, GFP_KERNEL);
+		if (ret)
+			memset(ret + old_sz, 0, new_sz - old_sz);
+	}
+	if (!ret)
+		return ERR_PTR(-ENOMEM);
+
+	mod_hdr_acts->actions = ret;
+	mod_hdr_acts->max_actions = new_num_actions;
+
+out:
+	return mod_hdr_acts->actions + (mod_hdr_acts->num_actions * MLX5_MH_ACT_SZ);
+}
+
+void
+mlx5e_mod_hdr_dealloc(struct mlx5e_tc_mod_hdr_acts *mod_hdr_acts)
+{
+	if (!mod_hdr_acts->is_static)
+		kfree(mod_hdr_acts->actions);
+
+	mod_hdr_acts->actions = NULL;
+	mod_hdr_acts->num_actions = 0;
+	mod_hdr_acts->max_actions = 0;
+}
+
+char *
+mlx5e_mod_hdr_get_item(struct mlx5e_tc_mod_hdr_acts *mod_hdr_acts, int pos)
+{
+	return mod_hdr_acts->actions + (pos * MLX5_MH_ACT_SZ);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/mod_hdr.h b/drivers/net/ethernet/mellanox/mlx5/core/en/mod_hdr.h
new file mode 100644
index 000000000..b8dac418d
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/mod_hdr.h
@@ -0,0 +1,57 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2020 Mellanox Technologies */
+
+#ifndef __MLX5E_EN_MOD_HDR_H__
+#define __MLX5E_EN_MOD_HDR_H__
+
+#include <linux/hashtable.h>
+#include <linux/mlx5/fs.h>
+
+#define MLX5_MH_ACT_SZ MLX5_UN_SZ_BYTES(set_add_copy_action_in_auto)
+
+struct mlx5e_mod_hdr_handle;
+
+struct mlx5e_tc_mod_hdr_acts {
+	int num_actions;
+	int max_actions;
+	bool is_static;
+	void *actions;
+};
+
+#define DECLARE_MOD_HDR_ACTS_ACTIONS(name, len) \
+	u8 name[len][MLX5_MH_ACT_SZ] = {}
+
+#define DECLARE_MOD_HDR_ACTS(name, acts_arr) \
+	struct mlx5e_tc_mod_hdr_acts name = { \
+		.max_actions = ARRAY_SIZE(acts_arr), \
+		.is_static = true, \
+		.actions = acts_arr, \
+	}
+
+char *mlx5e_mod_hdr_alloc(struct mlx5_core_dev *mdev, int namespace,
+			  struct mlx5e_tc_mod_hdr_acts *mod_hdr_acts);
+void mlx5e_mod_hdr_dealloc(struct mlx5e_tc_mod_hdr_acts *mod_hdr_acts);
+char *mlx5e_mod_hdr_get_item(struct mlx5e_tc_mod_hdr_acts *mod_hdr_acts, int pos);
+
+struct mlx5e_mod_hdr_handle *
+mlx5e_mod_hdr_attach(struct mlx5_core_dev *mdev,
+		     struct mod_hdr_tbl *tbl,
+		     enum mlx5_flow_namespace_type namespace,
+		     struct mlx5e_tc_mod_hdr_acts *mod_hdr_acts);
+void mlx5e_mod_hdr_detach(struct mlx5_core_dev *mdev,
+			  struct mod_hdr_tbl *tbl,
+			  struct mlx5e_mod_hdr_handle *mh);
+struct mlx5_modify_hdr *mlx5e_mod_hdr_get(struct mlx5e_mod_hdr_handle *mh);
+
+void mlx5e_mod_hdr_tbl_init(struct mod_hdr_tbl *tbl);
+void mlx5e_mod_hdr_tbl_destroy(struct mod_hdr_tbl *tbl);
+
+static inline int mlx5e_mod_hdr_max_actions(struct mlx5_core_dev *mdev, int namespace)
+{
+	if (namespace == MLX5_FLOW_NAMESPACE_FDB) /* FDB offloading */
+		return MLX5_CAP_ESW_FLOWTABLE_FDB(mdev, max_modify_header_actions);
+	else /* namespace is MLX5_FLOW_NAMESPACE_KERNEL - NIC offloading */
+		return MLX5_CAP_FLOWTABLE_NIC_RX(mdev, max_modify_header_actions);
+}
+
+#endif /* __MLX5E_EN_MOD_HDR_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/monitor_stats.c b/drivers/net/ethernet/mellanox/mlx5/core/en/monitor_stats.c
new file mode 100644
index 000000000..254c84739
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/monitor_stats.c
@@ -0,0 +1,151 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/* Copyright (c) 2018 Mellanox Technologies. */
+
+#include "en.h"
+#include "monitor_stats.h"
+#include "lib/eq.h"
+
+/* Driver will set the following watch counters list:
+ * Ppcnt.802_3:
+ * a_in_range_length_errors      Type: 0x0, Counter:  0x0, group_id = N/A
+ * a_out_of_range_length_field   Type: 0x0, Counter:  0x1, group_id = N/A
+ * a_frame_too_long_errors       Type: 0x0, Counter:  0x2, group_id = N/A
+ * a_frame_check_sequence_errors Type: 0x0, Counter:  0x3, group_id = N/A
+ * a_alignment_errors            Type: 0x0, Counter:  0x4, group_id = N/A
+ * if_out_discards               Type: 0x0, Counter:  0x5, group_id = N/A
+ * Q_Counters:
+ * Q[index].rx_out_of_buffer   Type: 0x1, Counter:  0x4, group_id = counter_ix
+ */
+
+#define NUM_REQ_PPCNT_COUNTER_S1 MLX5_CMD_SET_MONITOR_NUM_PPCNT_COUNTER_SET1
+#define NUM_REQ_Q_COUNTERS_S1    MLX5_CMD_SET_MONITOR_NUM_Q_COUNTERS_SET1
+
+int mlx5e_monitor_counter_supported(struct mlx5e_priv *priv)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+
+	if (!MLX5_CAP_GEN(mdev, max_num_of_monitor_counters))
+		return false;
+	if (MLX5_CAP_PCAM_REG(mdev, ppcnt) &&
+	    MLX5_CAP_GEN(mdev, num_ppcnt_monitor_counters) <
+	    NUM_REQ_PPCNT_COUNTER_S1)
+		return false;
+	if (MLX5_CAP_GEN(mdev, num_q_monitor_counters) <
+	    NUM_REQ_Q_COUNTERS_S1)
+		return false;
+	return true;
+}
+
+void mlx5e_monitor_counter_arm(struct mlx5e_priv *priv)
+{
+	u32 in[MLX5_ST_SZ_DW(arm_monitor_counter_in)] = {};
+
+	MLX5_SET(arm_monitor_counter_in, in, opcode,
+		 MLX5_CMD_OP_ARM_MONITOR_COUNTER);
+	mlx5_cmd_exec_in(priv->mdev, arm_monitor_counter, in);
+}
+
+static void mlx5e_monitor_counters_work(struct work_struct *work)
+{
+	struct mlx5e_priv *priv = container_of(work, struct mlx5e_priv,
+					       monitor_counters_work);
+
+	mutex_lock(&priv->state_lock);
+	mlx5e_stats_update_ndo_stats(priv);
+	mutex_unlock(&priv->state_lock);
+	mlx5e_monitor_counter_arm(priv);
+}
+
+static int mlx5e_monitor_event_handler(struct notifier_block *nb,
+				       unsigned long event, void *eqe)
+{
+	struct mlx5e_priv *priv = mlx5_nb_cof(nb, struct mlx5e_priv,
+					      monitor_counters_nb);
+	queue_work(priv->wq, &priv->monitor_counters_work);
+	return NOTIFY_OK;
+}
+
+static int fill_monitor_counter_ppcnt_set1(int cnt, u32 *in)
+{
+	enum mlx5_monitor_counter_ppcnt ppcnt_cnt;
+
+	for (ppcnt_cnt = 0;
+	     ppcnt_cnt < NUM_REQ_PPCNT_COUNTER_S1;
+	     ppcnt_cnt++, cnt++) {
+		MLX5_SET(set_monitor_counter_in, in,
+			 monitor_counter[cnt].type,
+			 MLX5_QUERY_MONITOR_CNT_TYPE_PPCNT);
+		MLX5_SET(set_monitor_counter_in, in,
+			 monitor_counter[cnt].counter,
+			 ppcnt_cnt);
+	}
+	return ppcnt_cnt;
+}
+
+static int fill_monitor_counter_q_counter_set1(int cnt, int q_counter, u32 *in)
+{
+	MLX5_SET(set_monitor_counter_in, in,
+		 monitor_counter[cnt].type,
+		 MLX5_QUERY_MONITOR_CNT_TYPE_Q_COUNTER);
+	MLX5_SET(set_monitor_counter_in, in,
+		 monitor_counter[cnt].counter,
+		 MLX5_QUERY_MONITOR_Q_COUNTER_RX_OUT_OF_BUFFER);
+	MLX5_SET(set_monitor_counter_in, in,
+		 monitor_counter[cnt].counter_group_id,
+		 q_counter);
+	return 1;
+}
+
+/* check if mlx5e_monitor_counter_supported before calling this function*/
+static void mlx5e_set_monitor_counter(struct mlx5e_priv *priv)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+	int max_num_of_counters = MLX5_CAP_GEN(mdev, max_num_of_monitor_counters);
+	int num_q_counters      = MLX5_CAP_GEN(mdev, num_q_monitor_counters);
+	int num_ppcnt_counters  = !MLX5_CAP_PCAM_REG(mdev, ppcnt) ? 0 :
+				  MLX5_CAP_GEN(mdev, num_ppcnt_monitor_counters);
+	u32 in[MLX5_ST_SZ_DW(set_monitor_counter_in)] = {};
+	int q_counter = priv->q_counter;
+	int cnt	= 0;
+
+	if (num_ppcnt_counters  >=  NUM_REQ_PPCNT_COUNTER_S1 &&
+	    max_num_of_counters >= (NUM_REQ_PPCNT_COUNTER_S1 + cnt))
+		cnt += fill_monitor_counter_ppcnt_set1(cnt, in);
+
+	if (num_q_counters      >=  NUM_REQ_Q_COUNTERS_S1 &&
+	    max_num_of_counters >= (NUM_REQ_Q_COUNTERS_S1 + cnt) &&
+	    q_counter)
+		cnt += fill_monitor_counter_q_counter_set1(cnt, q_counter, in);
+
+	MLX5_SET(set_monitor_counter_in, in, num_of_counters, cnt);
+	MLX5_SET(set_monitor_counter_in, in, opcode,
+		 MLX5_CMD_OP_SET_MONITOR_COUNTER);
+
+	mlx5_cmd_exec_in(mdev, set_monitor_counter, in);
+}
+
+/* check if mlx5e_monitor_counter_supported before calling this function*/
+void mlx5e_monitor_counter_init(struct mlx5e_priv *priv)
+{
+	INIT_WORK(&priv->monitor_counters_work, mlx5e_monitor_counters_work);
+	MLX5_NB_INIT(&priv->monitor_counters_nb, mlx5e_monitor_event_handler,
+		     MONITOR_COUNTER);
+	mlx5_eq_notifier_register(priv->mdev, &priv->monitor_counters_nb);
+
+	mlx5e_set_monitor_counter(priv);
+	mlx5e_monitor_counter_arm(priv);
+	queue_work(priv->wq, &priv->update_stats_work);
+}
+
+/* check if mlx5e_monitor_counter_supported before calling this function*/
+void mlx5e_monitor_counter_cleanup(struct mlx5e_priv *priv)
+{
+	u32 in[MLX5_ST_SZ_DW(set_monitor_counter_in)] = {};
+
+	MLX5_SET(set_monitor_counter_in, in, opcode,
+		 MLX5_CMD_OP_SET_MONITOR_COUNTER);
+
+	mlx5_cmd_exec_in(priv->mdev, set_monitor_counter, in);
+	mlx5_eq_notifier_unregister(priv->mdev, &priv->monitor_counters_nb);
+	cancel_work_sync(&priv->monitor_counters_work);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/monitor_stats.h b/drivers/net/ethernet/mellanox/mlx5/core/en/monitor_stats.h
new file mode 100644
index 000000000..e1ac4b3d2
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/monitor_stats.h
@@ -0,0 +1,12 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2018 Mellanox Technologies. */
+
+#ifndef __MLX5_MONITOR_H__
+#define __MLX5_MONITOR_H__
+
+int  mlx5e_monitor_counter_supported(struct mlx5e_priv *priv);
+void mlx5e_monitor_counter_init(struct mlx5e_priv *priv);
+void mlx5e_monitor_counter_cleanup(struct mlx5e_priv *priv);
+void mlx5e_monitor_counter_arm(struct mlx5e_priv *priv);
+
+#endif /* __MLX5_MONITOR_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/params.c b/drivers/net/ethernet/mellanox/mlx5/core/en/params.c
new file mode 100644
index 000000000..d3de1b7a8
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/params.c
@@ -0,0 +1,1240 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2019 Mellanox Technologies. */
+
+#include "en/params.h"
+#include "en/txrx.h"
+#include "en/port.h"
+#include "en_accel/en_accel.h"
+#include "en_accel/ipsec.h"
+#include <net/xdp_sock_drv.h>
+
+static u8 mlx5e_mpwrq_min_page_shift(struct mlx5_core_dev *mdev)
+{
+	u8 min_page_shift = MLX5_CAP_GEN_2(mdev, log_min_mkey_entity_size);
+
+	return min_page_shift ? : 12;
+}
+
+u8 mlx5e_mpwrq_page_shift(struct mlx5_core_dev *mdev, struct mlx5e_xsk_param *xsk)
+{
+	u8 req_page_shift = xsk ? order_base_2(xsk->chunk_size) : PAGE_SHIFT;
+	u8 min_page_shift = mlx5e_mpwrq_min_page_shift(mdev);
+
+	/* Regular RQ uses order-0 pages, the NIC must be able to map them. */
+	if (WARN_ON_ONCE(!xsk && req_page_shift < min_page_shift))
+		min_page_shift = req_page_shift;
+
+	return max(req_page_shift, min_page_shift);
+}
+
+enum mlx5e_mpwrq_umr_mode
+mlx5e_mpwrq_umr_mode(struct mlx5_core_dev *mdev, struct mlx5e_xsk_param *xsk)
+{
+	/* Different memory management schemes use different mechanisms to map
+	 * user-mode memory. The stricter guarantees we have, the faster
+	 * mechanisms we use:
+	 * 1. MTT - direct mapping in page granularity.
+	 * 2. KSM - indirect mapping to another MKey to arbitrary addresses, but
+	 *    all mappings have the same size.
+	 * 3. KLM - indirect mapping to another MKey to arbitrary addresses, and
+	 *    mappings can have different sizes.
+	 */
+	u8 page_shift = mlx5e_mpwrq_page_shift(mdev, xsk);
+	bool unaligned = xsk ? xsk->unaligned : false;
+	bool oversized = false;
+
+	if (xsk) {
+		oversized = xsk->chunk_size < (1 << page_shift);
+		WARN_ON_ONCE(xsk->chunk_size > (1 << page_shift));
+	}
+
+	/* XSK frame size doesn't match the UMR page size, either because the
+	 * frame size is not a power of two, or it's smaller than the minimal
+	 * page size supported by the firmware.
+	 * It's possible to receive packets bigger than MTU in certain setups.
+	 * To avoid writing over the XSK frame boundary, the top region of each
+	 * stride is mapped to a garbage page, resulting in two mappings of
+	 * different sizes per frame.
+	 */
+	if (oversized) {
+		/* An optimization for frame sizes equal to 3 * power_of_two.
+		 * 3 KSMs point to the frame, and one KSM points to the garbage
+		 * page, which works faster than KLM.
+		 */
+		if (xsk->chunk_size % 3 == 0 && is_power_of_2(xsk->chunk_size / 3))
+			return MLX5E_MPWRQ_UMR_MODE_TRIPLE;
+
+		return MLX5E_MPWRQ_UMR_MODE_OVERSIZED;
+	}
+
+	/* XSK frames can start at arbitrary unaligned locations, but they all
+	 * have the same size which is a power of two. It allows to optimize to
+	 * one KSM per frame.
+	 */
+	if (unaligned)
+		return MLX5E_MPWRQ_UMR_MODE_UNALIGNED;
+
+	/* XSK: frames are naturally aligned, MTT can be used.
+	 * Non-XSK: Allocations happen in units of CPU pages, therefore, the
+	 * mappings are naturally aligned.
+	 */
+	return MLX5E_MPWRQ_UMR_MODE_ALIGNED;
+}
+
+u8 mlx5e_mpwrq_umr_entry_size(enum mlx5e_mpwrq_umr_mode mode)
+{
+	switch (mode) {
+	case MLX5E_MPWRQ_UMR_MODE_ALIGNED:
+		return sizeof(struct mlx5_mtt);
+	case MLX5E_MPWRQ_UMR_MODE_UNALIGNED:
+		return sizeof(struct mlx5_ksm);
+	case MLX5E_MPWRQ_UMR_MODE_OVERSIZED:
+		return sizeof(struct mlx5_klm) * 2;
+	case MLX5E_MPWRQ_UMR_MODE_TRIPLE:
+		return sizeof(struct mlx5_ksm) * 4;
+	}
+	WARN_ONCE(1, "MPWRQ UMR mode %d is not known\n", mode);
+	return 0;
+}
+
+u8 mlx5e_mpwrq_log_wqe_sz(struct mlx5_core_dev *mdev, u8 page_shift,
+			  enum mlx5e_mpwrq_umr_mode umr_mode)
+{
+	u8 umr_entry_size = mlx5e_mpwrq_umr_entry_size(umr_mode);
+	u8 max_pages_per_wqe, max_log_mpwqe_size;
+	u16 max_wqe_size;
+
+	/* Keep in sync with MLX5_MPWRQ_MAX_PAGES_PER_WQE. */
+	max_wqe_size = mlx5e_get_max_sq_aligned_wqebbs(mdev) * MLX5_SEND_WQE_BB;
+	max_pages_per_wqe = ALIGN_DOWN(max_wqe_size - sizeof(struct mlx5e_umr_wqe),
+				       MLX5_UMR_MTT_ALIGNMENT) / umr_entry_size;
+	max_log_mpwqe_size = ilog2(max_pages_per_wqe) + page_shift;
+
+	WARN_ON_ONCE(max_log_mpwqe_size < MLX5E_ORDER2_MAX_PACKET_MTU);
+
+	return min_t(u8, max_log_mpwqe_size, MLX5_MPWRQ_MAX_LOG_WQE_SZ);
+}
+
+u8 mlx5e_mpwrq_pages_per_wqe(struct mlx5_core_dev *mdev, u8 page_shift,
+			     enum mlx5e_mpwrq_umr_mode umr_mode)
+{
+	u8 log_wqe_sz = mlx5e_mpwrq_log_wqe_sz(mdev, page_shift, umr_mode);
+	u8 pages_per_wqe;
+
+	pages_per_wqe = log_wqe_sz > page_shift ? (1 << (log_wqe_sz - page_shift)) : 1;
+
+	/* Two MTTs are needed to form an octword. The number of MTTs is encoded
+	 * in octwords in a UMR WQE, so we need at least two to avoid mapping
+	 * garbage addresses.
+	 */
+	if (WARN_ON_ONCE(pages_per_wqe < 2 && umr_mode == MLX5E_MPWRQ_UMR_MODE_ALIGNED))
+		pages_per_wqe = 2;
+
+	/* Sanity check for further calculations to succeed. */
+	BUILD_BUG_ON(MLX5_MPWRQ_MAX_PAGES_PER_WQE > 64);
+	if (WARN_ON_ONCE(pages_per_wqe > MLX5_MPWRQ_MAX_PAGES_PER_WQE))
+		return MLX5_MPWRQ_MAX_PAGES_PER_WQE;
+
+	return pages_per_wqe;
+}
+
+u16 mlx5e_mpwrq_umr_wqe_sz(struct mlx5_core_dev *mdev, u8 page_shift,
+			   enum mlx5e_mpwrq_umr_mode umr_mode)
+{
+	u8 pages_per_wqe = mlx5e_mpwrq_pages_per_wqe(mdev, page_shift, umr_mode);
+	u8 umr_entry_size = mlx5e_mpwrq_umr_entry_size(umr_mode);
+	u16 umr_wqe_sz;
+
+	umr_wqe_sz = sizeof(struct mlx5e_umr_wqe) +
+		ALIGN(pages_per_wqe * umr_entry_size, MLX5_UMR_MTT_ALIGNMENT);
+
+	WARN_ON_ONCE(DIV_ROUND_UP(umr_wqe_sz, MLX5_SEND_WQE_DS) > MLX5_WQE_CTRL_DS_MASK);
+
+	return umr_wqe_sz;
+}
+
+u8 mlx5e_mpwrq_umr_wqebbs(struct mlx5_core_dev *mdev, u8 page_shift,
+			  enum mlx5e_mpwrq_umr_mode umr_mode)
+{
+	return DIV_ROUND_UP(mlx5e_mpwrq_umr_wqe_sz(mdev, page_shift, umr_mode),
+			    MLX5_SEND_WQE_BB);
+}
+
+u8 mlx5e_mpwrq_mtts_per_wqe(struct mlx5_core_dev *mdev, u8 page_shift,
+			    enum mlx5e_mpwrq_umr_mode umr_mode)
+{
+	u8 pages_per_wqe = mlx5e_mpwrq_pages_per_wqe(mdev, page_shift, umr_mode);
+
+	/* Add another page as a buffer between WQEs. This page will absorb
+	 * write overflow by the hardware, when receiving packets larger than
+	 * MTU. These oversize packets are dropped by the driver at a later
+	 * stage.
+	 */
+	return ALIGN(pages_per_wqe + 1,
+		     MLX5_SEND_WQE_BB / mlx5e_mpwrq_umr_entry_size(umr_mode));
+}
+
+u32 mlx5e_mpwrq_max_num_entries(struct mlx5_core_dev *mdev,
+				enum mlx5e_mpwrq_umr_mode umr_mode)
+{
+	/* Same limits apply to KSMs and KLMs. */
+	u32 klm_limit = min(MLX5E_MAX_RQ_NUM_KSMS,
+			    1 << MLX5_CAP_GEN(mdev, log_max_klm_list_size));
+
+	switch (umr_mode) {
+	case MLX5E_MPWRQ_UMR_MODE_ALIGNED:
+		return MLX5E_MAX_RQ_NUM_MTTS;
+	case MLX5E_MPWRQ_UMR_MODE_UNALIGNED:
+		return klm_limit;
+	case MLX5E_MPWRQ_UMR_MODE_OVERSIZED:
+		/* Each entry is two KLMs. */
+		return klm_limit / 2;
+	case MLX5E_MPWRQ_UMR_MODE_TRIPLE:
+		/* Each entry is four KSMs. */
+		return klm_limit / 4;
+	}
+	WARN_ONCE(1, "MPWRQ UMR mode %d is not known\n", umr_mode);
+	return 0;
+}
+
+static u8 mlx5e_mpwrq_max_log_rq_size(struct mlx5_core_dev *mdev, u8 page_shift,
+				      enum mlx5e_mpwrq_umr_mode umr_mode)
+{
+	u8 mtts_per_wqe = mlx5e_mpwrq_mtts_per_wqe(mdev, page_shift, umr_mode);
+	u32 max_entries = mlx5e_mpwrq_max_num_entries(mdev, umr_mode);
+
+	return ilog2(max_entries / mtts_per_wqe);
+}
+
+u8 mlx5e_mpwrq_max_log_rq_pkts(struct mlx5_core_dev *mdev, u8 page_shift,
+			       enum mlx5e_mpwrq_umr_mode umr_mode)
+{
+	return mlx5e_mpwrq_max_log_rq_size(mdev, page_shift, umr_mode) +
+		mlx5e_mpwrq_log_wqe_sz(mdev, page_shift, umr_mode) -
+		MLX5E_ORDER2_MAX_PACKET_MTU;
+}
+
+u16 mlx5e_get_linear_rq_headroom(struct mlx5e_params *params,
+				 struct mlx5e_xsk_param *xsk)
+{
+	u16 headroom;
+
+	if (xsk)
+		return xsk->headroom;
+
+	headroom = NET_IP_ALIGN;
+	if (params->xdp_prog)
+		headroom += XDP_PACKET_HEADROOM;
+	else
+		headroom += MLX5_RX_HEADROOM;
+
+	return headroom;
+}
+
+static u32 mlx5e_rx_get_linear_sz_xsk(struct mlx5e_params *params,
+				      struct mlx5e_xsk_param *xsk)
+{
+	u32 hw_mtu = MLX5E_SW2HW_MTU(params, params->sw_mtu);
+
+	return xsk->headroom + hw_mtu;
+}
+
+static u32 mlx5e_rx_get_linear_sz_skb(struct mlx5e_params *params, bool xsk)
+{
+	/* SKBs built on XDP_PASS on XSK RQs don't have headroom. */
+	u16 headroom = xsk ? 0 : mlx5e_get_linear_rq_headroom(params, NULL);
+	u32 hw_mtu = MLX5E_SW2HW_MTU(params, params->sw_mtu);
+
+	return MLX5_SKB_FRAG_SZ(headroom + hw_mtu);
+}
+
+static u32 mlx5e_rx_get_linear_stride_sz(struct mlx5_core_dev *mdev,
+					 struct mlx5e_params *params,
+					 struct mlx5e_xsk_param *xsk,
+					 bool mpwqe)
+{
+	/* XSK frames are mapped as individual pages, because frames may come in
+	 * an arbitrary order from random locations in the UMEM.
+	 */
+	if (xsk)
+		return mpwqe ? 1 << mlx5e_mpwrq_page_shift(mdev, xsk) : PAGE_SIZE;
+
+	/* XDP in mlx5e doesn't support multiple packets per page. */
+	if (params->xdp_prog)
+		return PAGE_SIZE;
+
+	return roundup_pow_of_two(mlx5e_rx_get_linear_sz_skb(params, false));
+}
+
+static u8 mlx5e_mpwqe_log_pkts_per_wqe(struct mlx5_core_dev *mdev,
+				       struct mlx5e_params *params,
+				       struct mlx5e_xsk_param *xsk)
+{
+	u32 linear_stride_sz = mlx5e_rx_get_linear_stride_sz(mdev, params, xsk, true);
+	enum mlx5e_mpwrq_umr_mode umr_mode = mlx5e_mpwrq_umr_mode(mdev, xsk);
+	u8 page_shift = mlx5e_mpwrq_page_shift(mdev, xsk);
+
+	return mlx5e_mpwrq_log_wqe_sz(mdev, page_shift, umr_mode) -
+		order_base_2(linear_stride_sz);
+}
+
+bool mlx5e_rx_is_linear_skb(struct mlx5_core_dev *mdev,
+			    struct mlx5e_params *params,
+			    struct mlx5e_xsk_param *xsk)
+{
+	if (params->packet_merge.type != MLX5E_PACKET_MERGE_NONE)
+		return false;
+
+	/* Both XSK and non-XSK cases allocate an SKB on XDP_PASS. Packet data
+	 * must fit into a CPU page.
+	 */
+	if (mlx5e_rx_get_linear_sz_skb(params, xsk) > PAGE_SIZE)
+		return false;
+
+	/* XSK frames must be big enough to hold the packet data. */
+	if (xsk && mlx5e_rx_get_linear_sz_xsk(params, xsk) > xsk->chunk_size)
+		return false;
+
+	return true;
+}
+
+static bool mlx5e_verify_rx_mpwqe_strides(struct mlx5_core_dev *mdev,
+					  u8 log_stride_sz, u8 log_num_strides,
+					  u8 page_shift,
+					  enum mlx5e_mpwrq_umr_mode umr_mode)
+{
+	if (log_stride_sz + log_num_strides !=
+	    mlx5e_mpwrq_log_wqe_sz(mdev, page_shift, umr_mode))
+		return false;
+
+	if (log_stride_sz < MLX5_MPWQE_LOG_STRIDE_SZ_BASE ||
+	    log_stride_sz > MLX5_MPWQE_LOG_STRIDE_SZ_MAX)
+		return false;
+
+	if (log_num_strides > MLX5_MPWQE_LOG_NUM_STRIDES_MAX)
+		return false;
+
+	if (MLX5_CAP_GEN(mdev, ext_stride_num_range))
+		return log_num_strides >= MLX5_MPWQE_LOG_NUM_STRIDES_EXT_BASE;
+
+	return log_num_strides >= MLX5_MPWQE_LOG_NUM_STRIDES_BASE;
+}
+
+bool mlx5e_rx_mpwqe_is_linear_skb(struct mlx5_core_dev *mdev,
+				  struct mlx5e_params *params,
+				  struct mlx5e_xsk_param *xsk)
+{
+	enum mlx5e_mpwrq_umr_mode umr_mode = mlx5e_mpwrq_umr_mode(mdev, xsk);
+	u8 page_shift = mlx5e_mpwrq_page_shift(mdev, xsk);
+	u8 log_num_strides;
+	u8 log_stride_sz;
+	u8 log_wqe_sz;
+
+	if (!mlx5e_rx_is_linear_skb(mdev, params, xsk))
+		return false;
+
+	log_stride_sz = order_base_2(mlx5e_rx_get_linear_stride_sz(mdev, params, xsk, true));
+	log_wqe_sz = mlx5e_mpwrq_log_wqe_sz(mdev, page_shift, umr_mode);
+
+	if (log_wqe_sz < log_stride_sz)
+		return false;
+
+	log_num_strides = log_wqe_sz - log_stride_sz;
+
+	return mlx5e_verify_rx_mpwqe_strides(mdev, log_stride_sz,
+					     log_num_strides, page_shift,
+					     umr_mode);
+}
+
+u8 mlx5e_mpwqe_get_log_rq_size(struct mlx5_core_dev *mdev,
+			       struct mlx5e_params *params,
+			       struct mlx5e_xsk_param *xsk)
+{
+	enum mlx5e_mpwrq_umr_mode umr_mode = mlx5e_mpwrq_umr_mode(mdev, xsk);
+	u8 log_pkts_per_wqe, page_shift, max_log_rq_size;
+
+	log_pkts_per_wqe = mlx5e_mpwqe_log_pkts_per_wqe(mdev, params, xsk);
+	page_shift = mlx5e_mpwrq_page_shift(mdev, xsk);
+	max_log_rq_size = mlx5e_mpwrq_max_log_rq_size(mdev, page_shift, umr_mode);
+
+	/* Numbers are unsigned, don't subtract to avoid underflow. */
+	if (params->log_rq_mtu_frames <
+	    log_pkts_per_wqe + MLX5E_PARAMS_MINIMUM_LOG_RQ_SIZE_MPW)
+		return MLX5E_PARAMS_MINIMUM_LOG_RQ_SIZE_MPW;
+
+	/* Ethtool's rx_max_pending is calculated for regular RQ, that uses
+	 * pages of PAGE_SIZE. Max length of an XSK RQ might differ if it uses a
+	 * frame size not equal to PAGE_SIZE.
+	 * A stricter condition is checked in mlx5e_mpwrq_validate_xsk, WARN on
+	 * unexpected failure.
+	 */
+	if (WARN_ON_ONCE(params->log_rq_mtu_frames > log_pkts_per_wqe + max_log_rq_size))
+		return max_log_rq_size;
+
+	return params->log_rq_mtu_frames - log_pkts_per_wqe;
+}
+
+u8 mlx5e_shampo_get_log_hd_entry_size(struct mlx5_core_dev *mdev,
+				      struct mlx5e_params *params)
+{
+	return order_base_2(DIV_ROUND_UP(MLX5E_RX_MAX_HEAD, MLX5E_SHAMPO_WQ_BASE_HEAD_ENTRY_SIZE));
+}
+
+u8 mlx5e_shampo_get_log_rsrv_size(struct mlx5_core_dev *mdev,
+				  struct mlx5e_params *params)
+{
+	return order_base_2(MLX5E_SHAMPO_WQ_RESRV_SIZE / MLX5E_SHAMPO_WQ_BASE_RESRV_SIZE);
+}
+
+u8 mlx5e_shampo_get_log_pkt_per_rsrv(struct mlx5_core_dev *mdev,
+				     struct mlx5e_params *params)
+{
+	u32 resrv_size = BIT(mlx5e_shampo_get_log_rsrv_size(mdev, params)) *
+			 PAGE_SIZE;
+
+	return order_base_2(DIV_ROUND_UP(resrv_size, params->sw_mtu));
+}
+
+u8 mlx5e_mpwqe_get_log_stride_size(struct mlx5_core_dev *mdev,
+				   struct mlx5e_params *params,
+				   struct mlx5e_xsk_param *xsk)
+{
+	if (mlx5e_rx_mpwqe_is_linear_skb(mdev, params, xsk))
+		return order_base_2(mlx5e_rx_get_linear_stride_sz(mdev, params, xsk, true));
+
+	return MLX5_MPWRQ_DEF_LOG_STRIDE_SZ(mdev);
+}
+
+u8 mlx5e_mpwqe_get_log_num_strides(struct mlx5_core_dev *mdev,
+				   struct mlx5e_params *params,
+				   struct mlx5e_xsk_param *xsk)
+{
+	enum mlx5e_mpwrq_umr_mode umr_mode = mlx5e_mpwrq_umr_mode(mdev, xsk);
+	u8 page_shift = mlx5e_mpwrq_page_shift(mdev, xsk);
+
+	return mlx5e_mpwrq_log_wqe_sz(mdev, page_shift, umr_mode) -
+		mlx5e_mpwqe_get_log_stride_size(mdev, params, xsk);
+}
+
+u8 mlx5e_mpwqe_get_min_wqe_bulk(unsigned int wq_sz)
+{
+#define UMR_WQE_BULK (2)
+	return min_t(unsigned int, UMR_WQE_BULK, wq_sz / 2 - 1);
+}
+
+u16 mlx5e_get_rq_headroom(struct mlx5_core_dev *mdev,
+			  struct mlx5e_params *params,
+			  struct mlx5e_xsk_param *xsk)
+{
+	u16 linear_headroom = mlx5e_get_linear_rq_headroom(params, xsk);
+
+	if (params->rq_wq_type == MLX5_WQ_TYPE_CYCLIC)
+		return linear_headroom;
+
+	if (mlx5e_rx_mpwqe_is_linear_skb(mdev, params, xsk))
+		return linear_headroom;
+
+	if (params->packet_merge.type == MLX5E_PACKET_MERGE_SHAMPO)
+		return linear_headroom;
+
+	return 0;
+}
+
+u16 mlx5e_calc_sq_stop_room(struct mlx5_core_dev *mdev, struct mlx5e_params *params)
+{
+	bool is_mpwqe = MLX5E_GET_PFLAG(params, MLX5E_PFLAG_SKB_TX_MPWQE);
+	u16 stop_room;
+
+	stop_room  = mlx5e_ktls_get_stop_room(mdev, params);
+	stop_room += mlx5e_stop_room_for_max_wqe(mdev);
+	if (is_mpwqe)
+		/* A MPWQE can take up to the maximum cacheline-aligned WQE +
+		 * all the normal stop room can be taken if a new packet breaks
+		 * the active MPWQE session and allocates its WQEs right away.
+		 */
+		stop_room += mlx5e_stop_room_for_mpwqe(mdev);
+
+	return stop_room;
+}
+
+int mlx5e_validate_params(struct mlx5_core_dev *mdev, struct mlx5e_params *params)
+{
+	size_t sq_size = 1 << params->log_sq_size;
+	u16 stop_room;
+
+	stop_room = mlx5e_calc_sq_stop_room(mdev, params);
+	if (stop_room >= sq_size) {
+		mlx5_core_err(mdev, "Stop room %u is bigger than the SQ size %zu\n",
+			      stop_room, sq_size);
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static struct dim_cq_moder mlx5e_get_def_tx_moderation(u8 cq_period_mode)
+{
+	struct dim_cq_moder moder = {};
+
+	moder.cq_period_mode = cq_period_mode;
+	moder.pkts = MLX5E_PARAMS_DEFAULT_TX_CQ_MODERATION_PKTS;
+	moder.usec = MLX5E_PARAMS_DEFAULT_TX_CQ_MODERATION_USEC;
+	if (cq_period_mode == MLX5_CQ_PERIOD_MODE_START_FROM_CQE)
+		moder.usec = MLX5E_PARAMS_DEFAULT_TX_CQ_MODERATION_USEC_FROM_CQE;
+
+	return moder;
+}
+
+static struct dim_cq_moder mlx5e_get_def_rx_moderation(u8 cq_period_mode)
+{
+	struct dim_cq_moder moder = {};
+
+	moder.cq_period_mode = cq_period_mode;
+	moder.pkts = MLX5E_PARAMS_DEFAULT_RX_CQ_MODERATION_PKTS;
+	moder.usec = MLX5E_PARAMS_DEFAULT_RX_CQ_MODERATION_USEC;
+	if (cq_period_mode == MLX5_CQ_PERIOD_MODE_START_FROM_CQE)
+		moder.usec = MLX5E_PARAMS_DEFAULT_RX_CQ_MODERATION_USEC_FROM_CQE;
+
+	return moder;
+}
+
+static u8 mlx5_to_net_dim_cq_period_mode(u8 cq_period_mode)
+{
+	return cq_period_mode == MLX5_CQ_PERIOD_MODE_START_FROM_CQE ?
+		DIM_CQ_PERIOD_MODE_START_FROM_CQE :
+		DIM_CQ_PERIOD_MODE_START_FROM_EQE;
+}
+
+void mlx5e_reset_tx_moderation(struct mlx5e_params *params, u8 cq_period_mode)
+{
+	if (params->tx_dim_enabled) {
+		u8 dim_period_mode = mlx5_to_net_dim_cq_period_mode(cq_period_mode);
+
+		params->tx_cq_moderation = net_dim_get_def_tx_moderation(dim_period_mode);
+	} else {
+		params->tx_cq_moderation = mlx5e_get_def_tx_moderation(cq_period_mode);
+	}
+}
+
+void mlx5e_reset_rx_moderation(struct mlx5e_params *params, u8 cq_period_mode)
+{
+	if (params->rx_dim_enabled) {
+		u8 dim_period_mode = mlx5_to_net_dim_cq_period_mode(cq_period_mode);
+
+		params->rx_cq_moderation = net_dim_get_def_rx_moderation(dim_period_mode);
+	} else {
+		params->rx_cq_moderation = mlx5e_get_def_rx_moderation(cq_period_mode);
+	}
+}
+
+void mlx5e_set_tx_cq_mode_params(struct mlx5e_params *params, u8 cq_period_mode)
+{
+	mlx5e_reset_tx_moderation(params, cq_period_mode);
+	MLX5E_SET_PFLAG(params, MLX5E_PFLAG_TX_CQE_BASED_MODER,
+			params->tx_cq_moderation.cq_period_mode ==
+				MLX5_CQ_PERIOD_MODE_START_FROM_CQE);
+}
+
+void mlx5e_set_rx_cq_mode_params(struct mlx5e_params *params, u8 cq_period_mode)
+{
+	mlx5e_reset_rx_moderation(params, cq_period_mode);
+	MLX5E_SET_PFLAG(params, MLX5E_PFLAG_RX_CQE_BASED_MODER,
+			params->rx_cq_moderation.cq_period_mode ==
+				MLX5_CQ_PERIOD_MODE_START_FROM_CQE);
+}
+
+bool slow_pci_heuristic(struct mlx5_core_dev *mdev)
+{
+	u32 link_speed = 0;
+	u32 pci_bw = 0;
+
+	mlx5e_port_max_linkspeed(mdev, &link_speed);
+	pci_bw = pcie_bandwidth_available(mdev->pdev, NULL, NULL, NULL);
+	mlx5_core_dbg_once(mdev, "Max link speed = %d, PCI BW = %d\n",
+			   link_speed, pci_bw);
+
+#define MLX5E_SLOW_PCI_RATIO (2)
+
+	return link_speed && pci_bw &&
+		link_speed > MLX5E_SLOW_PCI_RATIO * pci_bw;
+}
+
+int mlx5e_mpwrq_validate_regular(struct mlx5_core_dev *mdev, struct mlx5e_params *params)
+{
+	enum mlx5e_mpwrq_umr_mode umr_mode = mlx5e_mpwrq_umr_mode(mdev, NULL);
+	u8 page_shift = mlx5e_mpwrq_page_shift(mdev, NULL);
+
+	if (!mlx5e_check_fragmented_striding_rq_cap(mdev, page_shift, umr_mode))
+		return -EOPNOTSUPP;
+
+	if (params->xdp_prog && !mlx5e_rx_mpwqe_is_linear_skb(mdev, params, NULL))
+		return -EINVAL;
+
+	return 0;
+}
+
+int mlx5e_mpwrq_validate_xsk(struct mlx5_core_dev *mdev, struct mlx5e_params *params,
+			     struct mlx5e_xsk_param *xsk)
+{
+	enum mlx5e_mpwrq_umr_mode umr_mode = mlx5e_mpwrq_umr_mode(mdev, xsk);
+	u8 page_shift = mlx5e_mpwrq_page_shift(mdev, xsk);
+	bool unaligned = xsk ? xsk->unaligned : false;
+	u16 max_mtu_pkts;
+
+	if (!mlx5e_check_fragmented_striding_rq_cap(mdev, page_shift, umr_mode))
+		return -EOPNOTSUPP;
+
+	if (!mlx5e_rx_mpwqe_is_linear_skb(mdev, params, xsk))
+		return -EINVAL;
+
+	/* Current RQ length is too big for the given frame size, the
+	 * needed number of WQEs exceeds the maximum.
+	 */
+	max_mtu_pkts = min_t(u8, MLX5E_PARAMS_MAXIMUM_LOG_RQ_SIZE,
+			     mlx5e_mpwrq_max_log_rq_pkts(mdev, page_shift, unaligned));
+	if (params->log_rq_mtu_frames > max_mtu_pkts) {
+		mlx5_core_err(mdev, "Current RQ length %d is too big for XSK with given frame size %u\n",
+			      1 << params->log_rq_mtu_frames, xsk->chunk_size);
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+void mlx5e_init_rq_type_params(struct mlx5_core_dev *mdev,
+			       struct mlx5e_params *params)
+{
+	params->log_rq_mtu_frames = is_kdump_kernel() ?
+		MLX5E_PARAMS_MINIMUM_LOG_RQ_SIZE :
+		MLX5E_PARAMS_DEFAULT_LOG_RQ_SIZE;
+
+	mlx5_core_info(mdev, "MLX5E: StrdRq(%d) RqSz(%ld) StrdSz(%ld) RxCqeCmprss(%d)\n",
+		       params->rq_wq_type == MLX5_WQ_TYPE_LINKED_LIST_STRIDING_RQ,
+		       params->rq_wq_type == MLX5_WQ_TYPE_LINKED_LIST_STRIDING_RQ ?
+		       BIT(mlx5e_mpwqe_get_log_rq_size(mdev, params, NULL)) :
+		       BIT(params->log_rq_mtu_frames),
+		       BIT(mlx5e_mpwqe_get_log_stride_size(mdev, params, NULL)),
+		       MLX5E_GET_PFLAG(params, MLX5E_PFLAG_RX_CQE_COMPRESS));
+}
+
+void mlx5e_set_rq_type(struct mlx5_core_dev *mdev, struct mlx5e_params *params)
+{
+	params->rq_wq_type = MLX5E_GET_PFLAG(params, MLX5E_PFLAG_RX_STRIDING_RQ) ?
+		MLX5_WQ_TYPE_LINKED_LIST_STRIDING_RQ :
+		MLX5_WQ_TYPE_CYCLIC;
+}
+
+void mlx5e_build_rq_params(struct mlx5_core_dev *mdev,
+			   struct mlx5e_params *params)
+{
+	/* Prefer Striding RQ, unless any of the following holds:
+	 * - Striding RQ configuration is not possible/supported.
+	 * - CQE compression is ON, and stride_index mini_cqe layout is not supported.
+	 * - Legacy RQ would use linear SKB while Striding RQ would use non-linear.
+	 *
+	 * No XSK params: checking the availability of striding RQ in general.
+	 */
+	if ((!MLX5E_GET_PFLAG(params, MLX5E_PFLAG_RX_CQE_COMPRESS) ||
+	     MLX5_CAP_GEN(mdev, mini_cqe_resp_stride_index)) &&
+	    !mlx5e_mpwrq_validate_regular(mdev, params) &&
+	    (mlx5e_rx_mpwqe_is_linear_skb(mdev, params, NULL) ||
+	     !mlx5e_rx_is_linear_skb(mdev, params, NULL)))
+		MLX5E_SET_PFLAG(params, MLX5E_PFLAG_RX_STRIDING_RQ, true);
+	mlx5e_set_rq_type(mdev, params);
+	mlx5e_init_rq_type_params(mdev, params);
+}
+
+/* Build queue parameters */
+
+void mlx5e_build_create_cq_param(struct mlx5e_create_cq_param *ccp, struct mlx5e_channel *c)
+{
+	*ccp = (struct mlx5e_create_cq_param) {
+		.napi = &c->napi,
+		.ch_stats = c->stats,
+		.node = cpu_to_node(c->cpu),
+		.ix = c->ix,
+	};
+}
+
+static int mlx5e_max_nonlinear_mtu(int first_frag_size, int frag_size, bool xdp)
+{
+	if (xdp)
+		/* XDP requires all fragments to be of the same size. */
+		return first_frag_size + (MLX5E_MAX_RX_FRAGS - 1) * frag_size;
+
+	/* Optimization for small packets: the last fragment is bigger than the others. */
+	return first_frag_size + (MLX5E_MAX_RX_FRAGS - 2) * frag_size + PAGE_SIZE;
+}
+
+#define DEFAULT_FRAG_SIZE (2048)
+
+static int mlx5e_build_rq_frags_info(struct mlx5_core_dev *mdev,
+				     struct mlx5e_params *params,
+				     struct mlx5e_xsk_param *xsk,
+				     struct mlx5e_rq_frags_info *info)
+{
+	u32 byte_count = MLX5E_SW2HW_MTU(params, params->sw_mtu);
+	int frag_size_max = DEFAULT_FRAG_SIZE;
+	int first_frag_size_max;
+	u32 buf_size = 0;
+	u16 headroom;
+	int max_mtu;
+	int i;
+
+	if (mlx5e_rx_is_linear_skb(mdev, params, xsk)) {
+		int frag_stride;
+
+		frag_stride = mlx5e_rx_get_linear_stride_sz(mdev, params, xsk, false);
+
+		info->arr[0].frag_size = byte_count;
+		info->arr[0].frag_stride = frag_stride;
+		info->num_frags = 1;
+
+		/* N WQEs share the same page, N = PAGE_SIZE / frag_stride. The
+		 * first WQE in the page is responsible for allocation of this
+		 * page, this WQE's index is k*N. If WQEs [k*N+1; k*N+N-1] are
+		 * still not completed, the allocation must stop before k*N.
+		 */
+		info->wqe_index_mask = (PAGE_SIZE / frag_stride) - 1;
+
+		goto out;
+	}
+
+	headroom = mlx5e_get_linear_rq_headroom(params, xsk);
+	first_frag_size_max = SKB_WITH_OVERHEAD(frag_size_max - headroom);
+
+	max_mtu = mlx5e_max_nonlinear_mtu(first_frag_size_max, frag_size_max,
+					  params->xdp_prog);
+	if (byte_count > max_mtu || params->xdp_prog) {
+		frag_size_max = PAGE_SIZE;
+		first_frag_size_max = SKB_WITH_OVERHEAD(frag_size_max - headroom);
+
+		max_mtu = mlx5e_max_nonlinear_mtu(first_frag_size_max, frag_size_max,
+						  params->xdp_prog);
+		if (byte_count > max_mtu) {
+			mlx5_core_err(mdev, "MTU %u is too big for non-linear legacy RQ (max %d)\n",
+				      params->sw_mtu, max_mtu);
+			return -EINVAL;
+		}
+	}
+
+	i = 0;
+	while (buf_size < byte_count) {
+		int frag_size = byte_count - buf_size;
+
+		if (i == 0)
+			frag_size = min(frag_size, first_frag_size_max);
+		else if (i < MLX5E_MAX_RX_FRAGS - 1)
+			frag_size = min(frag_size, frag_size_max);
+
+		info->arr[i].frag_size = frag_size;
+		buf_size += frag_size;
+
+		if (params->xdp_prog) {
+			/* XDP multi buffer expects fragments of the same size. */
+			info->arr[i].frag_stride = frag_size_max;
+		} else {
+			if (i == 0) {
+				/* Ensure that headroom and tailroom are included. */
+				frag_size += headroom;
+				frag_size += SKB_DATA_ALIGN(sizeof(struct skb_shared_info));
+			}
+			info->arr[i].frag_stride = roundup_pow_of_two(frag_size);
+		}
+
+		i++;
+	}
+	info->num_frags = i;
+
+	/* The last fragment of WQE with index 2*N may share the page with the
+	 * first fragment of WQE with index 2*N+1 in certain cases. If WQE 2*N+1
+	 * is not completed yet, WQE 2*N must not be allocated, as it's
+	 * responsible for allocating a new page.
+	 */
+	if (frag_size_max == PAGE_SIZE) {
+		/* No WQE can start in the middle of a page. */
+		info->wqe_index_mask = 0;
+	} else {
+		/* PAGE_SIZEs starting from 8192 don't use 2K-sized fragments,
+		 * because there would be more than MLX5E_MAX_RX_FRAGS of them.
+		 */
+		WARN_ON(PAGE_SIZE != 2 * DEFAULT_FRAG_SIZE);
+
+		/* Odd number of fragments allows to pack the last fragment of
+		 * the previous WQE and the first fragment of the next WQE into
+		 * the same page.
+		 * As long as DEFAULT_FRAG_SIZE is 2048, and MLX5E_MAX_RX_FRAGS
+		 * is 4, the last fragment can be bigger than the rest only if
+		 * it's the fourth one, so WQEs consisting of 3 fragments will
+		 * always share a page.
+		 * When a page is shared, WQE bulk size is 2, otherwise just 1.
+		 */
+		info->wqe_index_mask = info->num_frags % 2;
+	}
+
+out:
+	/* Bulking optimization to skip allocation until at least 8 WQEs can be
+	 * allocated in a row. At the same time, never start allocation when
+	 * the page is still used by older WQEs.
+	 */
+	info->wqe_bulk = max_t(u8, info->wqe_index_mask + 1, 8);
+
+	info->log_num_frags = order_base_2(info->num_frags);
+
+	return 0;
+}
+
+static u8 mlx5e_get_rqwq_log_stride(u8 wq_type, int ndsegs)
+{
+	int sz = sizeof(struct mlx5_wqe_data_seg) * ndsegs;
+
+	switch (wq_type) {
+	case MLX5_WQ_TYPE_LINKED_LIST_STRIDING_RQ:
+		sz += sizeof(struct mlx5e_rx_wqe_ll);
+		break;
+	default: /* MLX5_WQ_TYPE_CYCLIC */
+		sz += sizeof(struct mlx5e_rx_wqe_cyc);
+	}
+
+	return order_base_2(sz);
+}
+
+static void mlx5e_build_common_cq_param(struct mlx5_core_dev *mdev,
+					struct mlx5e_cq_param *param)
+{
+	void *cqc = param->cqc;
+
+	MLX5_SET(cqc, cqc, uar_page, mdev->priv.uar->index);
+	if (MLX5_CAP_GEN(mdev, cqe_128_always) && cache_line_size() >= 128)
+		MLX5_SET(cqc, cqc, cqe_sz, CQE_STRIDE_128_PAD);
+}
+
+static u32 mlx5e_shampo_get_log_cq_size(struct mlx5_core_dev *mdev,
+					struct mlx5e_params *params,
+					struct mlx5e_xsk_param *xsk)
+{
+	int rsrv_size = BIT(mlx5e_shampo_get_log_rsrv_size(mdev, params)) * PAGE_SIZE;
+	u16 num_strides = BIT(mlx5e_mpwqe_get_log_num_strides(mdev, params, xsk));
+	int pkt_per_rsrv = BIT(mlx5e_shampo_get_log_pkt_per_rsrv(mdev, params));
+	u8 log_stride_sz = mlx5e_mpwqe_get_log_stride_size(mdev, params, xsk);
+	int wq_size = BIT(mlx5e_mpwqe_get_log_rq_size(mdev, params, xsk));
+	int wqe_size = BIT(log_stride_sz) * num_strides;
+
+	/* +1 is for the case that the pkt_per_rsrv dont consume the reservation
+	 * so we get a filler cqe for the rest of the reservation.
+	 */
+	return order_base_2((wqe_size / rsrv_size) * wq_size * (pkt_per_rsrv + 1));
+}
+
+static void mlx5e_build_rx_cq_param(struct mlx5_core_dev *mdev,
+				    struct mlx5e_params *params,
+				    struct mlx5e_xsk_param *xsk,
+				    struct mlx5e_cq_param *param)
+{
+	bool hw_stridx = false;
+	void *cqc = param->cqc;
+	u8 log_cq_size;
+
+	switch (params->rq_wq_type) {
+	case MLX5_WQ_TYPE_LINKED_LIST_STRIDING_RQ:
+		hw_stridx = MLX5_CAP_GEN(mdev, mini_cqe_resp_stride_index);
+		if (params->packet_merge.type == MLX5E_PACKET_MERGE_SHAMPO)
+			log_cq_size = mlx5e_shampo_get_log_cq_size(mdev, params, xsk);
+		else
+			log_cq_size = mlx5e_mpwqe_get_log_rq_size(mdev, params, xsk) +
+				mlx5e_mpwqe_get_log_num_strides(mdev, params, xsk);
+		break;
+	default: /* MLX5_WQ_TYPE_CYCLIC */
+		log_cq_size = params->log_rq_mtu_frames;
+	}
+
+	MLX5_SET(cqc, cqc, log_cq_size, log_cq_size);
+	if (MLX5E_GET_PFLAG(params, MLX5E_PFLAG_RX_CQE_COMPRESS)) {
+		MLX5_SET(cqc, cqc, mini_cqe_res_format, hw_stridx ?
+			 MLX5_CQE_FORMAT_CSUM_STRIDX : MLX5_CQE_FORMAT_CSUM);
+		MLX5_SET(cqc, cqc, cqe_comp_en, 1);
+	}
+
+	mlx5e_build_common_cq_param(mdev, param);
+	param->cq_period_mode = params->rx_cq_moderation.cq_period_mode;
+}
+
+static u8 rq_end_pad_mode(struct mlx5_core_dev *mdev, struct mlx5e_params *params)
+{
+	bool lro_en = params->packet_merge.type == MLX5E_PACKET_MERGE_LRO;
+	bool ro = pcie_relaxed_ordering_enabled(mdev->pdev) &&
+		MLX5_CAP_GEN(mdev, relaxed_ordering_write);
+
+	return ro && lro_en ?
+		MLX5_WQ_END_PAD_MODE_NONE : MLX5_WQ_END_PAD_MODE_ALIGN;
+}
+
+int mlx5e_build_rq_param(struct mlx5_core_dev *mdev,
+			 struct mlx5e_params *params,
+			 struct mlx5e_xsk_param *xsk,
+			 u16 q_counter,
+			 struct mlx5e_rq_param *param)
+{
+	void *rqc = param->rqc;
+	void *wq = MLX5_ADDR_OF(rqc, rqc, wq);
+	int ndsegs = 1;
+	int err;
+
+	switch (params->rq_wq_type) {
+	case MLX5_WQ_TYPE_LINKED_LIST_STRIDING_RQ: {
+		u8 log_wqe_num_of_strides = mlx5e_mpwqe_get_log_num_strides(mdev, params, xsk);
+		u8 log_wqe_stride_size = mlx5e_mpwqe_get_log_stride_size(mdev, params, xsk);
+		enum mlx5e_mpwrq_umr_mode umr_mode = mlx5e_mpwrq_umr_mode(mdev, xsk);
+		u8 page_shift = mlx5e_mpwrq_page_shift(mdev, xsk);
+
+		if (!mlx5e_verify_rx_mpwqe_strides(mdev, log_wqe_stride_size,
+						   log_wqe_num_of_strides,
+						   page_shift, umr_mode)) {
+			mlx5_core_err(mdev,
+				      "Bad RX MPWQE params: log_stride_size %u, log_num_strides %u, umr_mode %d\n",
+				      log_wqe_stride_size, log_wqe_num_of_strides,
+				      umr_mode);
+			return -EINVAL;
+		}
+
+		MLX5_SET(wq, wq, log_wqe_num_of_strides,
+			 log_wqe_num_of_strides - MLX5_MPWQE_LOG_NUM_STRIDES_BASE);
+		MLX5_SET(wq, wq, log_wqe_stride_size,
+			 log_wqe_stride_size - MLX5_MPWQE_LOG_STRIDE_SZ_BASE);
+		MLX5_SET(wq, wq, log_wq_sz, mlx5e_mpwqe_get_log_rq_size(mdev, params, xsk));
+		if (params->packet_merge.type == MLX5E_PACKET_MERGE_SHAMPO) {
+			MLX5_SET(wq, wq, shampo_enable, true);
+			MLX5_SET(wq, wq, log_reservation_size,
+				 mlx5e_shampo_get_log_rsrv_size(mdev, params));
+			MLX5_SET(wq, wq,
+				 log_max_num_of_packets_per_reservation,
+				 mlx5e_shampo_get_log_pkt_per_rsrv(mdev, params));
+			MLX5_SET(wq, wq, log_headers_entry_size,
+				 mlx5e_shampo_get_log_hd_entry_size(mdev, params));
+			MLX5_SET(rqc, rqc, reservation_timeout,
+				 params->packet_merge.timeout);
+			MLX5_SET(rqc, rqc, shampo_match_criteria_type,
+				 params->packet_merge.shampo.match_criteria_type);
+			MLX5_SET(rqc, rqc, shampo_no_match_alignment_granularity,
+				 params->packet_merge.shampo.alignment_granularity);
+		}
+		break;
+	}
+	default: /* MLX5_WQ_TYPE_CYCLIC */
+		MLX5_SET(wq, wq, log_wq_sz, params->log_rq_mtu_frames);
+		err = mlx5e_build_rq_frags_info(mdev, params, xsk, &param->frags_info);
+		if (err)
+			return err;
+		ndsegs = param->frags_info.num_frags;
+	}
+
+	MLX5_SET(wq, wq, wq_type,          params->rq_wq_type);
+	MLX5_SET(wq, wq, end_padding_mode, rq_end_pad_mode(mdev, params));
+	MLX5_SET(wq, wq, log_wq_stride,
+		 mlx5e_get_rqwq_log_stride(params->rq_wq_type, ndsegs));
+	MLX5_SET(wq, wq, pd,               mdev->mlx5e_res.hw_objs.pdn);
+	MLX5_SET(rqc, rqc, counter_set_id, q_counter);
+	MLX5_SET(rqc, rqc, vsd,            params->vlan_strip_disable);
+	MLX5_SET(rqc, rqc, scatter_fcs,    params->scatter_fcs_en);
+
+	param->wq.buf_numa_node = dev_to_node(mlx5_core_dma_dev(mdev));
+	mlx5e_build_rx_cq_param(mdev, params, xsk, &param->cqp);
+
+	return 0;
+}
+
+void mlx5e_build_drop_rq_param(struct mlx5_core_dev *mdev,
+			       u16 q_counter,
+			       struct mlx5e_rq_param *param)
+{
+	void *rqc = param->rqc;
+	void *wq = MLX5_ADDR_OF(rqc, rqc, wq);
+
+	MLX5_SET(wq, wq, wq_type, MLX5_WQ_TYPE_CYCLIC);
+	MLX5_SET(wq, wq, log_wq_stride,
+		 mlx5e_get_rqwq_log_stride(MLX5_WQ_TYPE_CYCLIC, 1));
+	MLX5_SET(rqc, rqc, counter_set_id, q_counter);
+
+	param->wq.buf_numa_node = dev_to_node(mlx5_core_dma_dev(mdev));
+}
+
+void mlx5e_build_tx_cq_param(struct mlx5_core_dev *mdev,
+			     struct mlx5e_params *params,
+			     struct mlx5e_cq_param *param)
+{
+	void *cqc = param->cqc;
+
+	MLX5_SET(cqc, cqc, log_cq_size, params->log_sq_size);
+
+	mlx5e_build_common_cq_param(mdev, param);
+	param->cq_period_mode = params->tx_cq_moderation.cq_period_mode;
+}
+
+void mlx5e_build_sq_param_common(struct mlx5_core_dev *mdev,
+				 struct mlx5e_sq_param *param)
+{
+	void *sqc = param->sqc;
+	void *wq = MLX5_ADDR_OF(sqc, sqc, wq);
+
+	MLX5_SET(wq, wq, log_wq_stride, ilog2(MLX5_SEND_WQE_BB));
+	MLX5_SET(wq, wq, pd,            mdev->mlx5e_res.hw_objs.pdn);
+
+	param->wq.buf_numa_node = dev_to_node(mlx5_core_dma_dev(mdev));
+}
+
+void mlx5e_build_sq_param(struct mlx5_core_dev *mdev,
+			  struct mlx5e_params *params,
+			  struct mlx5e_sq_param *param)
+{
+	void *sqc = param->sqc;
+	void *wq = MLX5_ADDR_OF(sqc, sqc, wq);
+	bool allow_swp;
+
+	allow_swp = mlx5_geneve_tx_allowed(mdev) ||
+		    (mlx5_ipsec_device_caps(mdev) & MLX5_IPSEC_CAP_CRYPTO);
+	mlx5e_build_sq_param_common(mdev, param);
+	MLX5_SET(wq, wq, log_wq_sz, params->log_sq_size);
+	MLX5_SET(sqc, sqc, allow_swp, allow_swp);
+	param->is_mpw = MLX5E_GET_PFLAG(params, MLX5E_PFLAG_SKB_TX_MPWQE);
+	param->stop_room = mlx5e_calc_sq_stop_room(mdev, params);
+	mlx5e_build_tx_cq_param(mdev, params, &param->cqp);
+}
+
+static void mlx5e_build_ico_cq_param(struct mlx5_core_dev *mdev,
+				     u8 log_wq_size,
+				     struct mlx5e_cq_param *param)
+{
+	void *cqc = param->cqc;
+
+	MLX5_SET(cqc, cqc, log_cq_size, log_wq_size);
+
+	mlx5e_build_common_cq_param(mdev, param);
+
+	param->cq_period_mode = DIM_CQ_PERIOD_MODE_START_FROM_EQE;
+}
+
+/* This function calculates the maximum number of headers entries that are needed
+ * per WQE, the formula is based on the size of the reservations and the
+ * restriction we have about max packets for reservation that is equal to max
+ * headers per reservation.
+ */
+u32 mlx5e_shampo_hd_per_wqe(struct mlx5_core_dev *mdev,
+			    struct mlx5e_params *params,
+			    struct mlx5e_rq_param *rq_param)
+{
+	int resv_size = BIT(mlx5e_shampo_get_log_rsrv_size(mdev, params)) * PAGE_SIZE;
+	u16 num_strides = BIT(mlx5e_mpwqe_get_log_num_strides(mdev, params, NULL));
+	int pkt_per_resv = BIT(mlx5e_shampo_get_log_pkt_per_rsrv(mdev, params));
+	u8 log_stride_sz = mlx5e_mpwqe_get_log_stride_size(mdev, params, NULL);
+	int wqe_size = BIT(log_stride_sz) * num_strides;
+	u32 hd_per_wqe;
+
+	/* Assumption: hd_per_wqe % 8 == 0. */
+	hd_per_wqe = (wqe_size / resv_size) * pkt_per_resv;
+	mlx5_core_dbg(mdev, "%s hd_per_wqe = %d rsrv_size = %d wqe_size = %d pkt_per_resv = %d\n",
+		      __func__, hd_per_wqe, resv_size, wqe_size, pkt_per_resv);
+	return hd_per_wqe;
+}
+
+/* This function calculates the maximum number of headers entries that are needed
+ * for the WQ, this value is uesed to allocate the header buffer in HW, thus
+ * must be a pow of 2.
+ */
+u32 mlx5e_shampo_hd_per_wq(struct mlx5_core_dev *mdev,
+			   struct mlx5e_params *params,
+			   struct mlx5e_rq_param *rq_param)
+{
+	void *wqc = MLX5_ADDR_OF(rqc, rq_param->rqc, wq);
+	int wq_size = BIT(MLX5_GET(wq, wqc, log_wq_sz));
+	u32 hd_per_wqe, hd_per_wq;
+
+	hd_per_wqe = mlx5e_shampo_hd_per_wqe(mdev, params, rq_param);
+	hd_per_wq = roundup_pow_of_two(hd_per_wqe * wq_size);
+	return hd_per_wq;
+}
+
+static u32 mlx5e_shampo_icosq_sz(struct mlx5_core_dev *mdev,
+				 struct mlx5e_params *params,
+				 struct mlx5e_rq_param *rq_param)
+{
+	int max_num_of_umr_per_wqe, max_hd_per_wqe, max_klm_per_umr, rest;
+	void *wqc = MLX5_ADDR_OF(rqc, rq_param->rqc, wq);
+	int wq_size = BIT(MLX5_GET(wq, wqc, log_wq_sz));
+	u32 wqebbs;
+
+	max_klm_per_umr = MLX5E_MAX_KLM_PER_WQE(mdev);
+	max_hd_per_wqe = mlx5e_shampo_hd_per_wqe(mdev, params, rq_param);
+	max_num_of_umr_per_wqe = max_hd_per_wqe / max_klm_per_umr;
+	rest = max_hd_per_wqe % max_klm_per_umr;
+	wqebbs = MLX5E_KLM_UMR_WQEBBS(max_klm_per_umr) * max_num_of_umr_per_wqe;
+	if (rest)
+		wqebbs += MLX5E_KLM_UMR_WQEBBS(rest);
+	wqebbs *= wq_size;
+	return wqebbs;
+}
+
+static u32 mlx5e_mpwrq_total_umr_wqebbs(struct mlx5_core_dev *mdev,
+					struct mlx5e_params *params,
+					struct mlx5e_xsk_param *xsk)
+{
+	enum mlx5e_mpwrq_umr_mode umr_mode = mlx5e_mpwrq_umr_mode(mdev, xsk);
+	u8 page_shift = mlx5e_mpwrq_page_shift(mdev, xsk);
+	u8 umr_wqebbs;
+
+	umr_wqebbs = mlx5e_mpwrq_umr_wqebbs(mdev, page_shift, umr_mode);
+
+	return umr_wqebbs * (1 << mlx5e_mpwqe_get_log_rq_size(mdev, params, xsk));
+}
+
+static u8 mlx5e_build_icosq_log_wq_sz(struct mlx5_core_dev *mdev,
+				      struct mlx5e_params *params,
+				      struct mlx5e_rq_param *rqp)
+{
+	u32 wqebbs, total_pages, useful_space;
+
+	/* MLX5_WQ_TYPE_CYCLIC */
+	if (params->rq_wq_type != MLX5_WQ_TYPE_LINKED_LIST_STRIDING_RQ)
+		return MLX5E_PARAMS_MINIMUM_LOG_SQ_SIZE;
+
+	/* UMR WQEs for the regular RQ. */
+	wqebbs = mlx5e_mpwrq_total_umr_wqebbs(mdev, params, NULL);
+
+	/* If XDP program is attached, XSK may be turned on at any time without
+	 * restarting the channel. ICOSQ must be big enough to fit UMR WQEs of
+	 * both regular RQ and XSK RQ.
+	 *
+	 * XSK uses different values of page_shift, and the total number of UMR
+	 * WQEBBs depends on it. This dependency is complex and not monotonic,
+	 * especially taking into consideration that some of the parameters come
+	 * from capabilities. Hence, we have to try all valid values of XSK
+	 * frame size (and page_shift) to find the maximum.
+	 */
+	if (params->xdp_prog) {
+		u32 max_xsk_wqebbs = 0;
+		u8 frame_shift;
+
+		for (frame_shift = XDP_UMEM_MIN_CHUNK_SHIFT;
+		     frame_shift <= PAGE_SHIFT; frame_shift++) {
+			/* The headroom doesn't affect the calculation. */
+			struct mlx5e_xsk_param xsk = {
+				.chunk_size = 1 << frame_shift,
+				.unaligned = false,
+			};
+
+			/* XSK aligned mode. */
+			max_xsk_wqebbs = max(max_xsk_wqebbs,
+				mlx5e_mpwrq_total_umr_wqebbs(mdev, params, &xsk));
+
+			/* XSK unaligned mode, frame size is a power of two. */
+			xsk.unaligned = true;
+			max_xsk_wqebbs = max(max_xsk_wqebbs,
+				mlx5e_mpwrq_total_umr_wqebbs(mdev, params, &xsk));
+
+			/* XSK unaligned mode, frame size is not equal to stride size. */
+			xsk.chunk_size -= 1;
+			max_xsk_wqebbs = max(max_xsk_wqebbs,
+				mlx5e_mpwrq_total_umr_wqebbs(mdev, params, &xsk));
+
+			/* XSK unaligned mode, frame size is a triple power of two. */
+			xsk.chunk_size = (1 << frame_shift) / 4 * 3;
+			max_xsk_wqebbs = max(max_xsk_wqebbs,
+				mlx5e_mpwrq_total_umr_wqebbs(mdev, params, &xsk));
+		}
+
+		wqebbs += max_xsk_wqebbs;
+	}
+
+	if (params->packet_merge.type == MLX5E_PACKET_MERGE_SHAMPO)
+		wqebbs += mlx5e_shampo_icosq_sz(mdev, params, rqp);
+
+	/* UMR WQEs don't cross the page boundary, they are padded with NOPs.
+	 * This padding is always smaller than the max WQE size. That gives us
+	 * at least (PAGE_SIZE - (max WQE size - MLX5_SEND_WQE_BB)) useful bytes
+	 * per page. The number of pages is estimated as the total size of WQEs
+	 * divided by the useful space in page, rounding up. If some WQEs don't
+	 * fully fit into the useful space, they can occupy part of the padding,
+	 * which proves this estimation to be correct (reserve enough space).
+	 */
+	useful_space = PAGE_SIZE - mlx5e_get_max_sq_wqebbs(mdev) + MLX5_SEND_WQE_BB;
+	total_pages = DIV_ROUND_UP(wqebbs * MLX5_SEND_WQE_BB, useful_space);
+	wqebbs = total_pages * (PAGE_SIZE / MLX5_SEND_WQE_BB);
+
+	return max_t(u8, MLX5E_PARAMS_MINIMUM_LOG_SQ_SIZE, order_base_2(wqebbs));
+}
+
+static u8 mlx5e_build_async_icosq_log_wq_sz(struct mlx5_core_dev *mdev)
+{
+	if (mlx5e_is_ktls_rx(mdev))
+		return MLX5E_PARAMS_DEFAULT_LOG_SQ_SIZE;
+
+	return MLX5E_PARAMS_MINIMUM_LOG_SQ_SIZE;
+}
+
+static void mlx5e_build_icosq_param(struct mlx5_core_dev *mdev,
+				    u8 log_wq_size,
+				    struct mlx5e_sq_param *param)
+{
+	void *sqc = param->sqc;
+	void *wq = MLX5_ADDR_OF(sqc, sqc, wq);
+
+	mlx5e_build_sq_param_common(mdev, param);
+
+	MLX5_SET(wq, wq, log_wq_sz, log_wq_size);
+	MLX5_SET(sqc, sqc, reg_umr, MLX5_CAP_ETH(mdev, reg_umr_sq));
+	mlx5e_build_ico_cq_param(mdev, log_wq_size, &param->cqp);
+}
+
+static void mlx5e_build_async_icosq_param(struct mlx5_core_dev *mdev,
+					  u8 log_wq_size,
+					  struct mlx5e_sq_param *param)
+{
+	void *sqc = param->sqc;
+	void *wq = MLX5_ADDR_OF(sqc, sqc, wq);
+
+	mlx5e_build_sq_param_common(mdev, param);
+	param->stop_room = mlx5e_stop_room_for_wqe(mdev, 1); /* for XSK NOP */
+	param->is_tls = mlx5e_is_ktls_rx(mdev);
+	if (param->is_tls)
+		param->stop_room += mlx5e_stop_room_for_wqe(mdev, 1); /* for TLS RX resync NOP */
+	MLX5_SET(sqc, sqc, reg_umr, MLX5_CAP_ETH(mdev, reg_umr_sq));
+	MLX5_SET(wq, wq, log_wq_sz, log_wq_size);
+	mlx5e_build_ico_cq_param(mdev, log_wq_size, &param->cqp);
+}
+
+void mlx5e_build_xdpsq_param(struct mlx5_core_dev *mdev,
+			     struct mlx5e_params *params,
+			     struct mlx5e_xsk_param *xsk,
+			     struct mlx5e_sq_param *param)
+{
+	void *sqc = param->sqc;
+	void *wq = MLX5_ADDR_OF(sqc, sqc, wq);
+
+	mlx5e_build_sq_param_common(mdev, param);
+	MLX5_SET(wq, wq, log_wq_sz, params->log_sq_size);
+	param->is_mpw = MLX5E_GET_PFLAG(params, MLX5E_PFLAG_XDP_TX_MPWQE);
+	param->is_xdp_mb = !mlx5e_rx_is_linear_skb(mdev, params, xsk);
+	mlx5e_build_tx_cq_param(mdev, params, &param->cqp);
+}
+
+int mlx5e_build_channel_param(struct mlx5_core_dev *mdev,
+			      struct mlx5e_params *params,
+			      u16 q_counter,
+			      struct mlx5e_channel_param *cparam)
+{
+	u8 icosq_log_wq_sz, async_icosq_log_wq_sz;
+	int err;
+
+	err = mlx5e_build_rq_param(mdev, params, NULL, q_counter, &cparam->rq);
+	if (err)
+		return err;
+
+	icosq_log_wq_sz = mlx5e_build_icosq_log_wq_sz(mdev, params, &cparam->rq);
+	async_icosq_log_wq_sz = mlx5e_build_async_icosq_log_wq_sz(mdev);
+
+	mlx5e_build_sq_param(mdev, params, &cparam->txq_sq);
+	mlx5e_build_xdpsq_param(mdev, params, NULL, &cparam->xdp_sq);
+	mlx5e_build_icosq_param(mdev, icosq_log_wq_sz, &cparam->icosq);
+	mlx5e_build_async_icosq_param(mdev, async_icosq_log_wq_sz, &cparam->async_icosq);
+
+	return 0;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/params.h b/drivers/net/ethernet/mellanox/mlx5/core/en/params.h
new file mode 100644
index 000000000..034debd14
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/params.h
@@ -0,0 +1,157 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2019 Mellanox Technologies. */
+
+#ifndef __MLX5_EN_PARAMS_H__
+#define __MLX5_EN_PARAMS_H__
+
+#include "en.h"
+
+struct mlx5e_xsk_param {
+	u16 headroom;
+	u16 chunk_size;
+	bool unaligned;
+};
+
+struct mlx5e_cq_param {
+	u32                        cqc[MLX5_ST_SZ_DW(cqc)];
+	struct mlx5_wq_param       wq;
+	u16                        eq_ix;
+	u8                         cq_period_mode;
+};
+
+struct mlx5e_rq_param {
+	struct mlx5e_cq_param      cqp;
+	u32                        rqc[MLX5_ST_SZ_DW(rqc)];
+	struct mlx5_wq_param       wq;
+	struct mlx5e_rq_frags_info frags_info;
+};
+
+struct mlx5e_sq_param {
+	struct mlx5e_cq_param      cqp;
+	u32                        sqc[MLX5_ST_SZ_DW(sqc)];
+	struct mlx5_wq_param       wq;
+	bool                       is_mpw;
+	bool                       is_tls;
+	bool                       is_xdp_mb;
+	u16                        stop_room;
+};
+
+struct mlx5e_channel_param {
+	struct mlx5e_rq_param      rq;
+	struct mlx5e_sq_param      txq_sq;
+	struct mlx5e_sq_param      xdp_sq;
+	struct mlx5e_sq_param      icosq;
+	struct mlx5e_sq_param      async_icosq;
+};
+
+struct mlx5e_create_sq_param {
+	struct mlx5_wq_ctrl        *wq_ctrl;
+	u32                         cqn;
+	u32                         ts_cqe_to_dest_cqn;
+	u32                         tisn;
+	u8                          tis_lst_sz;
+	u8                          min_inline_mode;
+};
+
+/* Striding RQ dynamic parameters */
+
+u8 mlx5e_mpwrq_page_shift(struct mlx5_core_dev *mdev, struct mlx5e_xsk_param *xsk);
+enum mlx5e_mpwrq_umr_mode
+mlx5e_mpwrq_umr_mode(struct mlx5_core_dev *mdev, struct mlx5e_xsk_param *xsk);
+u8 mlx5e_mpwrq_umr_entry_size(enum mlx5e_mpwrq_umr_mode mode);
+u8 mlx5e_mpwrq_log_wqe_sz(struct mlx5_core_dev *mdev, u8 page_shift,
+			  enum mlx5e_mpwrq_umr_mode umr_mode);
+u8 mlx5e_mpwrq_pages_per_wqe(struct mlx5_core_dev *mdev, u8 page_shift,
+			     enum mlx5e_mpwrq_umr_mode umr_mode);
+u16 mlx5e_mpwrq_umr_wqe_sz(struct mlx5_core_dev *mdev, u8 page_shift,
+			   enum mlx5e_mpwrq_umr_mode umr_mode);
+u8 mlx5e_mpwrq_umr_wqebbs(struct mlx5_core_dev *mdev, u8 page_shift,
+			  enum mlx5e_mpwrq_umr_mode umr_mode);
+u8 mlx5e_mpwrq_mtts_per_wqe(struct mlx5_core_dev *mdev, u8 page_shift,
+			    enum mlx5e_mpwrq_umr_mode umr_mode);
+u32 mlx5e_mpwrq_max_num_entries(struct mlx5_core_dev *mdev,
+				enum mlx5e_mpwrq_umr_mode umr_mode);
+u8 mlx5e_mpwrq_max_log_rq_pkts(struct mlx5_core_dev *mdev, u8 page_shift,
+			       enum mlx5e_mpwrq_umr_mode umr_mode);
+
+/* Parameter calculations */
+
+void mlx5e_reset_tx_moderation(struct mlx5e_params *params, u8 cq_period_mode);
+void mlx5e_reset_rx_moderation(struct mlx5e_params *params, u8 cq_period_mode);
+void mlx5e_set_tx_cq_mode_params(struct mlx5e_params *params, u8 cq_period_mode);
+void mlx5e_set_rx_cq_mode_params(struct mlx5e_params *params, u8 cq_period_mode);
+
+bool slow_pci_heuristic(struct mlx5_core_dev *mdev);
+int mlx5e_mpwrq_validate_regular(struct mlx5_core_dev *mdev, struct mlx5e_params *params);
+int mlx5e_mpwrq_validate_xsk(struct mlx5_core_dev *mdev, struct mlx5e_params *params,
+			     struct mlx5e_xsk_param *xsk);
+void mlx5e_build_rq_params(struct mlx5_core_dev *mdev, struct mlx5e_params *params);
+void mlx5e_set_rq_type(struct mlx5_core_dev *mdev, struct mlx5e_params *params);
+void mlx5e_init_rq_type_params(struct mlx5_core_dev *mdev, struct mlx5e_params *params);
+
+u16 mlx5e_get_linear_rq_headroom(struct mlx5e_params *params,
+				 struct mlx5e_xsk_param *xsk);
+bool mlx5e_rx_is_linear_skb(struct mlx5_core_dev *mdev,
+			    struct mlx5e_params *params,
+			    struct mlx5e_xsk_param *xsk);
+bool mlx5e_rx_mpwqe_is_linear_skb(struct mlx5_core_dev *mdev,
+				  struct mlx5e_params *params,
+				  struct mlx5e_xsk_param *xsk);
+u8 mlx5e_mpwqe_get_log_rq_size(struct mlx5_core_dev *mdev,
+			       struct mlx5e_params *params,
+			       struct mlx5e_xsk_param *xsk);
+u8 mlx5e_shampo_get_log_hd_entry_size(struct mlx5_core_dev *mdev,
+				      struct mlx5e_params *params);
+u8 mlx5e_shampo_get_log_rsrv_size(struct mlx5_core_dev *mdev,
+				  struct mlx5e_params *params);
+u8 mlx5e_shampo_get_log_pkt_per_rsrv(struct mlx5_core_dev *mdev,
+				     struct mlx5e_params *params);
+u32 mlx5e_shampo_hd_per_wqe(struct mlx5_core_dev *mdev,
+			    struct mlx5e_params *params,
+			    struct mlx5e_rq_param *rq_param);
+u32 mlx5e_shampo_hd_per_wq(struct mlx5_core_dev *mdev,
+			   struct mlx5e_params *params,
+			   struct mlx5e_rq_param *rq_param);
+u8 mlx5e_mpwqe_get_log_stride_size(struct mlx5_core_dev *mdev,
+				   struct mlx5e_params *params,
+				   struct mlx5e_xsk_param *xsk);
+u8 mlx5e_mpwqe_get_log_num_strides(struct mlx5_core_dev *mdev,
+				   struct mlx5e_params *params,
+				   struct mlx5e_xsk_param *xsk);
+u8 mlx5e_mpwqe_get_min_wqe_bulk(unsigned int wq_sz);
+u16 mlx5e_get_rq_headroom(struct mlx5_core_dev *mdev,
+			  struct mlx5e_params *params,
+			  struct mlx5e_xsk_param *xsk);
+
+/* Build queue parameters */
+
+void mlx5e_build_create_cq_param(struct mlx5e_create_cq_param *ccp, struct mlx5e_channel *c);
+int mlx5e_build_rq_param(struct mlx5_core_dev *mdev,
+			 struct mlx5e_params *params,
+			 struct mlx5e_xsk_param *xsk,
+			 u16 q_counter,
+			 struct mlx5e_rq_param *param);
+void mlx5e_build_drop_rq_param(struct mlx5_core_dev *mdev,
+			       u16 q_counter,
+			       struct mlx5e_rq_param *param);
+void mlx5e_build_sq_param_common(struct mlx5_core_dev *mdev,
+				 struct mlx5e_sq_param *param);
+void mlx5e_build_sq_param(struct mlx5_core_dev *mdev,
+			  struct mlx5e_params *params,
+			  struct mlx5e_sq_param *param);
+void mlx5e_build_tx_cq_param(struct mlx5_core_dev *mdev,
+			     struct mlx5e_params *params,
+			     struct mlx5e_cq_param *param);
+void mlx5e_build_xdpsq_param(struct mlx5_core_dev *mdev,
+			     struct mlx5e_params *params,
+			     struct mlx5e_xsk_param *xsk,
+			     struct mlx5e_sq_param *param);
+int mlx5e_build_channel_param(struct mlx5_core_dev *mdev,
+			      struct mlx5e_params *params,
+			      u16 q_counter,
+			      struct mlx5e_channel_param *cparam);
+
+u16 mlx5e_calc_sq_stop_room(struct mlx5_core_dev *mdev, struct mlx5e_params *params);
+int mlx5e_validate_params(struct mlx5_core_dev *mdev, struct mlx5e_params *params);
+
+#endif /* __MLX5_EN_PARAMS_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/port.c b/drivers/net/ethernet/mellanox/mlx5/core/en/port.c
new file mode 100644
index 000000000..89510cac4
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/port.c
@@ -0,0 +1,594 @@
+/*
+ * Copyright (c) 2018, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "port.h"
+
+/* speed in units of 1Mb */
+static const u32 mlx5e_link_speed[MLX5E_LINK_MODES_NUMBER] = {
+	[MLX5E_1000BASE_CX_SGMII] = 1000,
+	[MLX5E_1000BASE_KX]       = 1000,
+	[MLX5E_10GBASE_CX4]       = 10000,
+	[MLX5E_10GBASE_KX4]       = 10000,
+	[MLX5E_10GBASE_KR]        = 10000,
+	[MLX5E_20GBASE_KR2]       = 20000,
+	[MLX5E_40GBASE_CR4]       = 40000,
+	[MLX5E_40GBASE_KR4]       = 40000,
+	[MLX5E_56GBASE_R4]        = 56000,
+	[MLX5E_10GBASE_CR]        = 10000,
+	[MLX5E_10GBASE_SR]        = 10000,
+	[MLX5E_10GBASE_ER]        = 10000,
+	[MLX5E_40GBASE_SR4]       = 40000,
+	[MLX5E_40GBASE_LR4]       = 40000,
+	[MLX5E_50GBASE_SR2]       = 50000,
+	[MLX5E_100GBASE_CR4]      = 100000,
+	[MLX5E_100GBASE_SR4]      = 100000,
+	[MLX5E_100GBASE_KR4]      = 100000,
+	[MLX5E_100GBASE_LR4]      = 100000,
+	[MLX5E_100BASE_TX]        = 100,
+	[MLX5E_1000BASE_T]        = 1000,
+	[MLX5E_10GBASE_T]         = 10000,
+	[MLX5E_25GBASE_CR]        = 25000,
+	[MLX5E_25GBASE_KR]        = 25000,
+	[MLX5E_25GBASE_SR]        = 25000,
+	[MLX5E_50GBASE_CR2]       = 50000,
+	[MLX5E_50GBASE_KR2]       = 50000,
+};
+
+static const u32 mlx5e_ext_link_speed[MLX5E_EXT_LINK_MODES_NUMBER] = {
+	[MLX5E_SGMII_100M]			= 100,
+	[MLX5E_1000BASE_X_SGMII]		= 1000,
+	[MLX5E_5GBASE_R]			= 5000,
+	[MLX5E_10GBASE_XFI_XAUI_1]		= 10000,
+	[MLX5E_40GBASE_XLAUI_4_XLPPI_4]		= 40000,
+	[MLX5E_25GAUI_1_25GBASE_CR_KR]		= 25000,
+	[MLX5E_50GAUI_2_LAUI_2_50GBASE_CR2_KR2]	= 50000,
+	[MLX5E_50GAUI_1_LAUI_1_50GBASE_CR_KR]	= 50000,
+	[MLX5E_CAUI_4_100GBASE_CR4_KR4]		= 100000,
+	[MLX5E_100GAUI_2_100GBASE_CR2_KR2]	= 100000,
+	[MLX5E_200GAUI_4_200GBASE_CR4_KR4]	= 200000,
+	[MLX5E_400GAUI_8]			= 400000,
+	[MLX5E_100GAUI_1_100GBASE_CR_KR]	= 100000,
+	[MLX5E_200GAUI_2_200GBASE_CR2_KR2]	= 200000,
+	[MLX5E_400GAUI_4_400GBASE_CR4_KR4]	= 400000,
+};
+
+bool mlx5e_ptys_ext_supported(struct mlx5_core_dev *mdev)
+{
+	struct mlx5e_port_eth_proto eproto;
+	int err;
+
+	if (MLX5_CAP_PCAM_FEATURE(mdev, ptys_extended_ethernet))
+		return true;
+
+	err = mlx5_port_query_eth_proto(mdev, 1, true, &eproto);
+	if (err)
+		return false;
+
+	return !!eproto.cap;
+}
+
+static void mlx5e_port_get_speed_arr(struct mlx5_core_dev *mdev,
+				     const u32 **arr, u32 *size,
+				     bool force_legacy)
+{
+	bool ext = force_legacy ? false : mlx5e_ptys_ext_supported(mdev);
+
+	*size = ext ? ARRAY_SIZE(mlx5e_ext_link_speed) :
+		      ARRAY_SIZE(mlx5e_link_speed);
+	*arr  = ext ? mlx5e_ext_link_speed : mlx5e_link_speed;
+}
+
+int mlx5_port_query_eth_proto(struct mlx5_core_dev *dev, u8 port, bool ext,
+			      struct mlx5e_port_eth_proto *eproto)
+{
+	u32 out[MLX5_ST_SZ_DW(ptys_reg)];
+	int err;
+
+	if (!eproto)
+		return -EINVAL;
+
+	err = mlx5_query_port_ptys(dev, out, sizeof(out), MLX5_PTYS_EN, port);
+	if (err)
+		return err;
+
+	eproto->cap   = MLX5_GET_ETH_PROTO(ptys_reg, out, ext,
+					   eth_proto_capability);
+	eproto->admin = MLX5_GET_ETH_PROTO(ptys_reg, out, ext, eth_proto_admin);
+	eproto->oper  = MLX5_GET_ETH_PROTO(ptys_reg, out, ext, eth_proto_oper);
+	return 0;
+}
+
+void mlx5_port_query_eth_autoneg(struct mlx5_core_dev *dev, u8 *an_status,
+				 u8 *an_disable_cap, u8 *an_disable_admin)
+{
+	u32 out[MLX5_ST_SZ_DW(ptys_reg)];
+
+	*an_status = 0;
+	*an_disable_cap = 0;
+	*an_disable_admin = 0;
+
+	if (mlx5_query_port_ptys(dev, out, sizeof(out), MLX5_PTYS_EN, 1))
+		return;
+
+	*an_status = MLX5_GET(ptys_reg, out, an_status);
+	*an_disable_cap = MLX5_GET(ptys_reg, out, an_disable_cap);
+	*an_disable_admin = MLX5_GET(ptys_reg, out, an_disable_admin);
+}
+
+int mlx5_port_set_eth_ptys(struct mlx5_core_dev *dev, bool an_disable,
+			   u32 proto_admin, bool ext)
+{
+	u32 out[MLX5_ST_SZ_DW(ptys_reg)];
+	u32 in[MLX5_ST_SZ_DW(ptys_reg)];
+	u8 an_disable_admin;
+	u8 an_disable_cap;
+	u8 an_status;
+
+	mlx5_port_query_eth_autoneg(dev, &an_status, &an_disable_cap,
+				    &an_disable_admin);
+	if (!an_disable_cap && an_disable)
+		return -EPERM;
+
+	memset(in, 0, sizeof(in));
+
+	MLX5_SET(ptys_reg, in, local_port, 1);
+	MLX5_SET(ptys_reg, in, an_disable_admin, an_disable);
+	MLX5_SET(ptys_reg, in, proto_mask, MLX5_PTYS_EN);
+	if (ext)
+		MLX5_SET(ptys_reg, in, ext_eth_proto_admin, proto_admin);
+	else
+		MLX5_SET(ptys_reg, in, eth_proto_admin, proto_admin);
+
+	return mlx5_core_access_reg(dev, in, sizeof(in), out,
+			    sizeof(out), MLX5_REG_PTYS, 0, 1);
+}
+
+u32 mlx5e_port_ptys2speed(struct mlx5_core_dev *mdev, u32 eth_proto_oper,
+			  bool force_legacy)
+{
+	unsigned long temp = eth_proto_oper;
+	const u32 *table;
+	u32 speed = 0;
+	u32 max_size;
+	int i;
+
+	mlx5e_port_get_speed_arr(mdev, &table, &max_size, force_legacy);
+	i = find_first_bit(&temp, max_size);
+	if (i < max_size)
+		speed = table[i];
+	return speed;
+}
+
+int mlx5e_port_linkspeed(struct mlx5_core_dev *mdev, u32 *speed)
+{
+	struct mlx5e_port_eth_proto eproto;
+	bool force_legacy = false;
+	bool ext;
+	int err;
+
+	ext = mlx5e_ptys_ext_supported(mdev);
+	err = mlx5_port_query_eth_proto(mdev, 1, ext, &eproto);
+	if (err)
+		goto out;
+	if (ext && !eproto.admin) {
+		force_legacy = true;
+		err = mlx5_port_query_eth_proto(mdev, 1, false, &eproto);
+		if (err)
+			goto out;
+	}
+	*speed = mlx5e_port_ptys2speed(mdev, eproto.oper, force_legacy);
+	if (!(*speed))
+		err = -EINVAL;
+
+out:
+	return err;
+}
+
+int mlx5e_port_max_linkspeed(struct mlx5_core_dev *mdev, u32 *speed)
+{
+	struct mlx5e_port_eth_proto eproto;
+	u32 max_speed = 0;
+	const u32 *table;
+	u32 max_size;
+	bool ext;
+	int err;
+	int i;
+
+	ext = mlx5e_ptys_ext_supported(mdev);
+	err = mlx5_port_query_eth_proto(mdev, 1, ext, &eproto);
+	if (err)
+		return err;
+
+	mlx5e_port_get_speed_arr(mdev, &table, &max_size, false);
+	for (i = 0; i < max_size; ++i)
+		if (eproto.cap & MLX5E_PROT_MASK(i))
+			max_speed = max(max_speed, table[i]);
+
+	*speed = max_speed;
+	return 0;
+}
+
+u32 mlx5e_port_speed2linkmodes(struct mlx5_core_dev *mdev, u32 speed,
+			       bool force_legacy)
+{
+	u32 link_modes = 0;
+	const u32 *table;
+	u32 max_size;
+	int i;
+
+	mlx5e_port_get_speed_arr(mdev, &table, &max_size, force_legacy);
+	for (i = 0; i < max_size; ++i) {
+		if (table[i] == speed)
+			link_modes |= MLX5E_PROT_MASK(i);
+	}
+	return link_modes;
+}
+
+int mlx5e_port_query_pbmc(struct mlx5_core_dev *mdev, void *out)
+{
+	int sz = MLX5_ST_SZ_BYTES(pbmc_reg);
+	void *in;
+	int err;
+
+	in = kzalloc(sz, GFP_KERNEL);
+	if (!in)
+		return -ENOMEM;
+
+	MLX5_SET(pbmc_reg, in, local_port, 1);
+	err = mlx5_core_access_reg(mdev, in, sz, out, sz, MLX5_REG_PBMC, 0, 0);
+
+	kfree(in);
+	return err;
+}
+
+int mlx5e_port_set_pbmc(struct mlx5_core_dev *mdev, void *in)
+{
+	int sz = MLX5_ST_SZ_BYTES(pbmc_reg);
+	void *out;
+	int err;
+
+	out = kzalloc(sz, GFP_KERNEL);
+	if (!out)
+		return -ENOMEM;
+
+	MLX5_SET(pbmc_reg, in, local_port, 1);
+	err = mlx5_core_access_reg(mdev, in, sz, out, sz, MLX5_REG_PBMC, 0, 1);
+
+	kfree(out);
+	return err;
+}
+
+/* buffer[i]: buffer that priority i mapped to */
+int mlx5e_port_query_priority2buffer(struct mlx5_core_dev *mdev, u8 *buffer)
+{
+	int sz = MLX5_ST_SZ_BYTES(pptb_reg);
+	u32 prio_x_buff;
+	void *out;
+	void *in;
+	int prio;
+	int err;
+
+	in = kzalloc(sz, GFP_KERNEL);
+	out = kzalloc(sz, GFP_KERNEL);
+	if (!in || !out) {
+		err = -ENOMEM;
+		goto out;
+	}
+
+	MLX5_SET(pptb_reg, in, local_port, 1);
+	err = mlx5_core_access_reg(mdev, in, sz, out, sz, MLX5_REG_PPTB, 0, 0);
+	if (err)
+		goto out;
+
+	prio_x_buff = MLX5_GET(pptb_reg, out, prio_x_buff);
+	for (prio = 0; prio < 8; prio++) {
+		buffer[prio] = (u8)(prio_x_buff >> (4 * prio)) & 0xF;
+		mlx5_core_dbg(mdev, "prio %d, buffer %d\n", prio, buffer[prio]);
+	}
+out:
+	kfree(in);
+	kfree(out);
+	return err;
+}
+
+int mlx5e_port_set_priority2buffer(struct mlx5_core_dev *mdev, u8 *buffer)
+{
+	int sz = MLX5_ST_SZ_BYTES(pptb_reg);
+	u32 prio_x_buff;
+	void *out;
+	void *in;
+	int prio;
+	int err;
+
+	in = kzalloc(sz, GFP_KERNEL);
+	out = kzalloc(sz, GFP_KERNEL);
+	if (!in || !out) {
+		err = -ENOMEM;
+		goto out;
+	}
+
+	/* First query the pptb register */
+	MLX5_SET(pptb_reg, in, local_port, 1);
+	err = mlx5_core_access_reg(mdev, in, sz, out, sz, MLX5_REG_PPTB, 0, 0);
+	if (err)
+		goto out;
+
+	memcpy(in, out, sz);
+	MLX5_SET(pptb_reg, in, local_port, 1);
+
+	/* Update the pm and prio_x_buff */
+	MLX5_SET(pptb_reg, in, pm, 0xFF);
+
+	prio_x_buff = 0;
+	for (prio = 0; prio < 8; prio++)
+		prio_x_buff |= (buffer[prio] << (4 * prio));
+	MLX5_SET(pptb_reg, in, prio_x_buff, prio_x_buff);
+
+	err = mlx5_core_access_reg(mdev, in, sz, out, sz, MLX5_REG_PPTB, 0, 1);
+
+out:
+	kfree(in);
+	kfree(out);
+	return err;
+}
+
+enum mlx5e_fec_supported_link_mode {
+	MLX5E_FEC_SUPPORTED_LINK_MODES_10G_40G,
+	MLX5E_FEC_SUPPORTED_LINK_MODES_25G,
+	MLX5E_FEC_SUPPORTED_LINK_MODES_50G,
+	MLX5E_FEC_SUPPORTED_LINK_MODES_56G,
+	MLX5E_FEC_SUPPORTED_LINK_MODES_100G,
+	MLX5E_FEC_SUPPORTED_LINK_MODE_50G_1X,
+	MLX5E_FEC_SUPPORTED_LINK_MODE_100G_2X,
+	MLX5E_FEC_SUPPORTED_LINK_MODE_200G_4X,
+	MLX5E_FEC_SUPPORTED_LINK_MODE_400G_8X,
+	MLX5E_MAX_FEC_SUPPORTED_LINK_MODE,
+};
+
+#define MLX5E_FEC_FIRST_50G_PER_LANE_MODE MLX5E_FEC_SUPPORTED_LINK_MODE_50G_1X
+
+#define MLX5E_FEC_OVERRIDE_ADMIN_POLICY(buf, policy, write, link)			\
+	do {										\
+		u16 *_policy = &(policy);						\
+		u32 *_buf = buf;							\
+											\
+		if (write)								\
+			MLX5_SET(pplm_reg, _buf, fec_override_admin_##link, *_policy);	\
+		else									\
+			*_policy = MLX5_GET(pplm_reg, _buf, fec_override_admin_##link);	\
+	} while (0)
+
+/* get/set FEC admin field for a given speed */
+static int mlx5e_fec_admin_field(u32 *pplm, u16 *fec_policy, bool write,
+				 enum mlx5e_fec_supported_link_mode link_mode)
+{
+	switch (link_mode) {
+	case MLX5E_FEC_SUPPORTED_LINK_MODES_10G_40G:
+		MLX5E_FEC_OVERRIDE_ADMIN_POLICY(pplm, *fec_policy, write, 10g_40g);
+		break;
+	case MLX5E_FEC_SUPPORTED_LINK_MODES_25G:
+		MLX5E_FEC_OVERRIDE_ADMIN_POLICY(pplm, *fec_policy, write, 25g);
+		break;
+	case MLX5E_FEC_SUPPORTED_LINK_MODES_50G:
+		MLX5E_FEC_OVERRIDE_ADMIN_POLICY(pplm, *fec_policy, write, 50g);
+		break;
+	case MLX5E_FEC_SUPPORTED_LINK_MODES_56G:
+		MLX5E_FEC_OVERRIDE_ADMIN_POLICY(pplm, *fec_policy, write, 56g);
+		break;
+	case MLX5E_FEC_SUPPORTED_LINK_MODES_100G:
+		MLX5E_FEC_OVERRIDE_ADMIN_POLICY(pplm, *fec_policy, write, 100g);
+		break;
+	case MLX5E_FEC_SUPPORTED_LINK_MODE_50G_1X:
+		MLX5E_FEC_OVERRIDE_ADMIN_POLICY(pplm, *fec_policy, write, 50g_1x);
+		break;
+	case MLX5E_FEC_SUPPORTED_LINK_MODE_100G_2X:
+		MLX5E_FEC_OVERRIDE_ADMIN_POLICY(pplm, *fec_policy, write, 100g_2x);
+		break;
+	case MLX5E_FEC_SUPPORTED_LINK_MODE_200G_4X:
+		MLX5E_FEC_OVERRIDE_ADMIN_POLICY(pplm, *fec_policy, write, 200g_4x);
+		break;
+	case MLX5E_FEC_SUPPORTED_LINK_MODE_400G_8X:
+		MLX5E_FEC_OVERRIDE_ADMIN_POLICY(pplm, *fec_policy, write, 400g_8x);
+		break;
+	default:
+		return -EINVAL;
+	}
+	return 0;
+}
+
+#define MLX5E_GET_FEC_OVERRIDE_CAP(buf, link)  \
+	MLX5_GET(pplm_reg, buf, fec_override_cap_##link)
+
+/* returns FEC capabilities for a given speed */
+static int mlx5e_get_fec_cap_field(u32 *pplm, u16 *fec_cap,
+				   enum mlx5e_fec_supported_link_mode link_mode)
+{
+	switch (link_mode) {
+	case MLX5E_FEC_SUPPORTED_LINK_MODES_10G_40G:
+		*fec_cap = MLX5E_GET_FEC_OVERRIDE_CAP(pplm, 10g_40g);
+		break;
+	case MLX5E_FEC_SUPPORTED_LINK_MODES_25G:
+		*fec_cap = MLX5E_GET_FEC_OVERRIDE_CAP(pplm, 25g);
+		break;
+	case MLX5E_FEC_SUPPORTED_LINK_MODES_50G:
+		*fec_cap = MLX5E_GET_FEC_OVERRIDE_CAP(pplm, 50g);
+		break;
+	case MLX5E_FEC_SUPPORTED_LINK_MODES_56G:
+		*fec_cap = MLX5E_GET_FEC_OVERRIDE_CAP(pplm, 56g);
+		break;
+	case MLX5E_FEC_SUPPORTED_LINK_MODES_100G:
+		*fec_cap = MLX5E_GET_FEC_OVERRIDE_CAP(pplm, 100g);
+		break;
+	case MLX5E_FEC_SUPPORTED_LINK_MODE_50G_1X:
+		*fec_cap = MLX5E_GET_FEC_OVERRIDE_CAP(pplm, 50g_1x);
+		break;
+	case MLX5E_FEC_SUPPORTED_LINK_MODE_100G_2X:
+		*fec_cap = MLX5E_GET_FEC_OVERRIDE_CAP(pplm, 100g_2x);
+		break;
+	case MLX5E_FEC_SUPPORTED_LINK_MODE_200G_4X:
+		*fec_cap = MLX5E_GET_FEC_OVERRIDE_CAP(pplm, 200g_4x);
+		break;
+	case MLX5E_FEC_SUPPORTED_LINK_MODE_400G_8X:
+		*fec_cap = MLX5E_GET_FEC_OVERRIDE_CAP(pplm, 400g_8x);
+		break;
+	default:
+		return -EINVAL;
+	}
+	return 0;
+}
+
+bool mlx5e_fec_in_caps(struct mlx5_core_dev *dev, int fec_policy)
+{
+	bool fec_50g_per_lane = MLX5_CAP_PCAM_FEATURE(dev, fec_50G_per_lane_in_pplm);
+	u32 out[MLX5_ST_SZ_DW(pplm_reg)] = {};
+	u32 in[MLX5_ST_SZ_DW(pplm_reg)] = {};
+	int sz = MLX5_ST_SZ_BYTES(pplm_reg);
+	int err;
+	int i;
+
+	if (!MLX5_CAP_GEN(dev, pcam_reg) || !MLX5_CAP_PCAM_REG(dev, pplm))
+		return false;
+
+	MLX5_SET(pplm_reg, in, local_port, 1);
+	err =  mlx5_core_access_reg(dev, in, sz, out, sz, MLX5_REG_PPLM, 0, 0);
+	if (err)
+		return false;
+
+	for (i = 0; i < MLX5E_MAX_FEC_SUPPORTED_LINK_MODE; i++) {
+		u16 fec_caps;
+
+		if (i >= MLX5E_FEC_FIRST_50G_PER_LANE_MODE && !fec_50g_per_lane)
+			break;
+
+		mlx5e_get_fec_cap_field(out, &fec_caps, i);
+		if (fec_caps & fec_policy)
+			return true;
+	}
+	return false;
+}
+
+int mlx5e_get_fec_mode(struct mlx5_core_dev *dev, u32 *fec_mode_active,
+		       u16 *fec_configured_mode)
+{
+	bool fec_50g_per_lane = MLX5_CAP_PCAM_FEATURE(dev, fec_50G_per_lane_in_pplm);
+	u32 out[MLX5_ST_SZ_DW(pplm_reg)] = {};
+	u32 in[MLX5_ST_SZ_DW(pplm_reg)] = {};
+	int sz = MLX5_ST_SZ_BYTES(pplm_reg);
+	int err;
+	int i;
+
+	if (!MLX5_CAP_GEN(dev, pcam_reg))
+		return -EOPNOTSUPP;
+
+	if (!MLX5_CAP_PCAM_REG(dev, pplm))
+		return -EOPNOTSUPP;
+
+	MLX5_SET(pplm_reg, in, local_port, 1);
+	err =  mlx5_core_access_reg(dev, in, sz, out, sz, MLX5_REG_PPLM, 0, 0);
+	if (err)
+		return err;
+
+	*fec_mode_active = MLX5_GET(pplm_reg, out, fec_mode_active);
+
+	if (!fec_configured_mode)
+		goto out;
+
+	*fec_configured_mode = 0;
+	for (i = 0; i < MLX5E_MAX_FEC_SUPPORTED_LINK_MODE; i++) {
+		if (i >= MLX5E_FEC_FIRST_50G_PER_LANE_MODE && !fec_50g_per_lane)
+			break;
+
+		mlx5e_fec_admin_field(out, fec_configured_mode, 0, i);
+		if (*fec_configured_mode != 0)
+			goto out;
+	}
+out:
+	return 0;
+}
+
+int mlx5e_set_fec_mode(struct mlx5_core_dev *dev, u16 fec_policy)
+{
+	bool fec_50g_per_lane = MLX5_CAP_PCAM_FEATURE(dev, fec_50G_per_lane_in_pplm);
+	u32 out[MLX5_ST_SZ_DW(pplm_reg)] = {};
+	u32 in[MLX5_ST_SZ_DW(pplm_reg)] = {};
+	int sz = MLX5_ST_SZ_BYTES(pplm_reg);
+	u16 fec_policy_auto = 0;
+	int err;
+	int i;
+
+	if (!MLX5_CAP_GEN(dev, pcam_reg))
+		return -EOPNOTSUPP;
+
+	if (!MLX5_CAP_PCAM_REG(dev, pplm))
+		return -EOPNOTSUPP;
+
+	if (fec_policy >= (1 << MLX5E_FEC_LLRS_272_257_1) && !fec_50g_per_lane)
+		return -EOPNOTSUPP;
+
+	if (fec_policy && !mlx5e_fec_in_caps(dev, fec_policy))
+		return -EOPNOTSUPP;
+
+	MLX5_SET(pplm_reg, in, local_port, 1);
+	err = mlx5_core_access_reg(dev, in, sz, out, sz, MLX5_REG_PPLM, 0, 0);
+	if (err)
+		return err;
+
+	MLX5_SET(pplm_reg, out, local_port, 1);
+
+	for (i = 0; i < MLX5E_MAX_FEC_SUPPORTED_LINK_MODE; i++) {
+		u16 conf_fec = fec_policy;
+		u16 fec_caps = 0;
+
+		if (i >= MLX5E_FEC_FIRST_50G_PER_LANE_MODE && !fec_50g_per_lane)
+			break;
+
+		/* RS fec in ethtool is mapped to MLX5E_FEC_RS_528_514
+		 * to link modes up to 25G per lane and to
+		 * MLX5E_FEC_RS_544_514 in the new link modes based on
+		 * 50 G per lane
+		 */
+		if (conf_fec == (1 << MLX5E_FEC_RS_528_514) &&
+		    i >= MLX5E_FEC_FIRST_50G_PER_LANE_MODE)
+			conf_fec = (1 << MLX5E_FEC_RS_544_514);
+
+		mlx5e_get_fec_cap_field(out, &fec_caps, i);
+
+		/* policy supported for link speed */
+		if (fec_caps & conf_fec)
+			mlx5e_fec_admin_field(out, &conf_fec, 1, i);
+		else
+			/* set FEC to auto*/
+			mlx5e_fec_admin_field(out, &fec_policy_auto, 1, i);
+	}
+
+	return mlx5_core_access_reg(dev, out, sz, out, sz, MLX5_REG_PPLM, 0, 1);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/port.h b/drivers/net/ethernet/mellanox/mlx5/core/en/port.h
new file mode 100644
index 000000000..7a7defe60
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/port.h
@@ -0,0 +1,76 @@
+/*
+ * Copyright (c) 2018, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef __MLX5E_EN_PORT_H
+#define __MLX5E_EN_PORT_H
+
+#include <linux/mlx5/driver.h>
+#include "en.h"
+
+struct mlx5e_port_eth_proto {
+	u32 cap;
+	u32 admin;
+	u32 oper;
+};
+
+int mlx5_port_query_eth_proto(struct mlx5_core_dev *dev, u8 port, bool ext,
+			      struct mlx5e_port_eth_proto *eproto);
+void mlx5_port_query_eth_autoneg(struct mlx5_core_dev *dev, u8 *an_status,
+				 u8 *an_disable_cap, u8 *an_disable_admin);
+int mlx5_port_set_eth_ptys(struct mlx5_core_dev *dev, bool an_disable,
+			   u32 proto_admin, bool ext);
+u32 mlx5e_port_ptys2speed(struct mlx5_core_dev *mdev, u32 eth_proto_oper,
+			  bool force_legacy);
+int mlx5e_port_linkspeed(struct mlx5_core_dev *mdev, u32 *speed);
+int mlx5e_port_max_linkspeed(struct mlx5_core_dev *mdev, u32 *speed);
+u32 mlx5e_port_speed2linkmodes(struct mlx5_core_dev *mdev, u32 speed,
+			       bool force_legacy);
+bool mlx5e_ptys_ext_supported(struct mlx5_core_dev *mdev);
+int mlx5e_port_query_pbmc(struct mlx5_core_dev *mdev, void *out);
+int mlx5e_port_set_pbmc(struct mlx5_core_dev *mdev, void *in);
+int mlx5e_port_query_priority2buffer(struct mlx5_core_dev *mdev, u8 *buffer);
+int mlx5e_port_set_priority2buffer(struct mlx5_core_dev *mdev, u8 *buffer);
+
+bool mlx5e_fec_in_caps(struct mlx5_core_dev *dev, int fec_policy);
+int mlx5e_get_fec_mode(struct mlx5_core_dev *dev, u32 *fec_mode_active,
+		       u16 *fec_configured_mode);
+int mlx5e_set_fec_mode(struct mlx5_core_dev *dev, u16 fec_policy);
+
+enum {
+	MLX5E_FEC_NOFEC,
+	MLX5E_FEC_FIRECODE,
+	MLX5E_FEC_RS_528_514,
+	MLX5E_FEC_RS_544_514 = 7,
+	MLX5E_FEC_LLRS_272_257_1 = 9,
+};
+
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/port_buffer.c b/drivers/net/ethernet/mellanox/mlx5/core/en/port_buffer.c
new file mode 100644
index 000000000..c9d5d8d93
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/port_buffer.c
@@ -0,0 +1,362 @@
+/*
+ * Copyright (c) 2018, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+#include "port_buffer.h"
+
+int mlx5e_port_query_buffer(struct mlx5e_priv *priv,
+			    struct mlx5e_port_buffer *port_buffer)
+{
+	u16 port_buff_cell_sz = priv->dcbx.port_buff_cell_sz;
+	struct mlx5_core_dev *mdev = priv->mdev;
+	int sz = MLX5_ST_SZ_BYTES(pbmc_reg);
+	u32 total_used = 0;
+	void *buffer;
+	void *out;
+	int err;
+	int i;
+
+	out = kzalloc(sz, GFP_KERNEL);
+	if (!out)
+		return -ENOMEM;
+
+	err = mlx5e_port_query_pbmc(mdev, out);
+	if (err)
+		goto out;
+
+	for (i = 0; i < MLX5E_MAX_BUFFER; i++) {
+		buffer = MLX5_ADDR_OF(pbmc_reg, out, buffer[i]);
+		port_buffer->buffer[i].lossy =
+			MLX5_GET(bufferx_reg, buffer, lossy);
+		port_buffer->buffer[i].epsb =
+			MLX5_GET(bufferx_reg, buffer, epsb);
+		port_buffer->buffer[i].size =
+			MLX5_GET(bufferx_reg, buffer, size) * port_buff_cell_sz;
+		port_buffer->buffer[i].xon =
+			MLX5_GET(bufferx_reg, buffer, xon_threshold) * port_buff_cell_sz;
+		port_buffer->buffer[i].xoff =
+			MLX5_GET(bufferx_reg, buffer, xoff_threshold) * port_buff_cell_sz;
+		total_used += port_buffer->buffer[i].size;
+
+		mlx5e_dbg(HW, priv, "buffer %d: size=%d, xon=%d, xoff=%d, epsb=%d, lossy=%d\n", i,
+			  port_buffer->buffer[i].size,
+			  port_buffer->buffer[i].xon,
+			  port_buffer->buffer[i].xoff,
+			  port_buffer->buffer[i].epsb,
+			  port_buffer->buffer[i].lossy);
+	}
+
+	port_buffer->port_buffer_size =
+		MLX5_GET(pbmc_reg, out, port_buffer_size) * port_buff_cell_sz;
+	port_buffer->spare_buffer_size =
+		port_buffer->port_buffer_size - total_used;
+
+	mlx5e_dbg(HW, priv, "total buffer size=%d, spare buffer size=%d\n",
+		  port_buffer->port_buffer_size,
+		  port_buffer->spare_buffer_size);
+out:
+	kfree(out);
+	return err;
+}
+
+static int port_set_buffer(struct mlx5e_priv *priv,
+			   struct mlx5e_port_buffer *port_buffer)
+{
+	u16 port_buff_cell_sz = priv->dcbx.port_buff_cell_sz;
+	struct mlx5_core_dev *mdev = priv->mdev;
+	int sz = MLX5_ST_SZ_BYTES(pbmc_reg);
+	void *in;
+	int err;
+	int i;
+
+	in = kzalloc(sz, GFP_KERNEL);
+	if (!in)
+		return -ENOMEM;
+
+	err = mlx5e_port_query_pbmc(mdev, in);
+	if (err)
+		goto out;
+
+	for (i = 0; i < MLX5E_MAX_BUFFER; i++) {
+		void *buffer = MLX5_ADDR_OF(pbmc_reg, in, buffer[i]);
+		u64 size = port_buffer->buffer[i].size;
+		u64 xoff = port_buffer->buffer[i].xoff;
+		u64 xon = port_buffer->buffer[i].xon;
+
+		do_div(size, port_buff_cell_sz);
+		do_div(xoff, port_buff_cell_sz);
+		do_div(xon, port_buff_cell_sz);
+		MLX5_SET(bufferx_reg, buffer, size, size);
+		MLX5_SET(bufferx_reg, buffer, lossy, port_buffer->buffer[i].lossy);
+		MLX5_SET(bufferx_reg, buffer, xoff_threshold, xoff);
+		MLX5_SET(bufferx_reg, buffer, xon_threshold, xon);
+	}
+
+	err = mlx5e_port_set_pbmc(mdev, in);
+out:
+	kfree(in);
+	return err;
+}
+
+/* xoff = ((301+2.16 * len [m]) * speed [Gbps] + 2.72 MTU [B])
+ * minimum speed value is 40Gbps
+ */
+static u32 calculate_xoff(struct mlx5e_priv *priv, unsigned int mtu)
+{
+	u32 speed;
+	u32 xoff;
+	int err;
+
+	err = mlx5e_port_linkspeed(priv->mdev, &speed);
+	if (err)
+		speed = SPEED_40000;
+	speed = max_t(u32, speed, SPEED_40000);
+
+	xoff = (301 + 216 * priv->dcbx.cable_len / 100) * speed / 1000 + 272 * mtu / 100;
+
+	mlx5e_dbg(HW, priv, "%s: xoff=%d\n", __func__, xoff);
+	return xoff;
+}
+
+static int update_xoff_threshold(struct mlx5e_port_buffer *port_buffer,
+				 u32 xoff, unsigned int max_mtu, u16 port_buff_cell_sz)
+{
+	int i;
+
+	for (i = 0; i < MLX5E_MAX_BUFFER; i++) {
+		if (port_buffer->buffer[i].lossy) {
+			port_buffer->buffer[i].xoff = 0;
+			port_buffer->buffer[i].xon  = 0;
+			continue;
+		}
+
+		if (port_buffer->buffer[i].size <
+		    (xoff + max_mtu + port_buff_cell_sz)) {
+			pr_err("buffer_size[%d]=%d is not enough for lossless buffer\n",
+			       i, port_buffer->buffer[i].size);
+			return -ENOMEM;
+		}
+
+		port_buffer->buffer[i].xoff = port_buffer->buffer[i].size - xoff;
+		port_buffer->buffer[i].xon  =
+			port_buffer->buffer[i].xoff - max_mtu;
+	}
+
+	return 0;
+}
+
+/**
+ *	update_buffer_lossy	- Update buffer configuration based on pfc
+ *	@max_mtu: netdev's max_mtu
+ *	@pfc_en: <input> current pfc configuration
+ *	@buffer: <input> current prio to buffer mapping
+ *	@xoff:   <input> xoff value
+ *	@port_buff_cell_sz: <input> port buffer cell_size
+ *	@port_buffer: <output> port receive buffer configuration
+ *	@change: <output>
+ *
+ *	Update buffer configuration based on pfc configuration and
+ *	priority to buffer mapping.
+ *	Buffer's lossy bit is changed to:
+ *		lossless if there is at least one PFC enabled priority
+ *		mapped to this buffer lossy if all priorities mapped to
+ *		this buffer are PFC disabled
+ *
+ *	@return: 0 if no error,
+ *	sets change to true if buffer configuration was modified.
+ */
+static int update_buffer_lossy(unsigned int max_mtu,
+			       u8 pfc_en, u8 *buffer, u32 xoff, u16 port_buff_cell_sz,
+			       struct mlx5e_port_buffer *port_buffer,
+			       bool *change)
+{
+	bool changed = false;
+	u8 lossy_count;
+	u8 prio_count;
+	u8 lossy;
+	int prio;
+	int err;
+	int i;
+
+	for (i = 0; i < MLX5E_MAX_BUFFER; i++) {
+		prio_count = 0;
+		lossy_count = 0;
+
+		for (prio = 0; prio < MLX5E_MAX_PRIORITY; prio++) {
+			if (buffer[prio] != i)
+				continue;
+
+			prio_count++;
+			lossy_count += !(pfc_en & (1 << prio));
+		}
+
+		if (lossy_count == prio_count)
+			lossy = 1;
+		else /* lossy_count < prio_count */
+			lossy = 0;
+
+		if (lossy != port_buffer->buffer[i].lossy) {
+			port_buffer->buffer[i].lossy = lossy;
+			changed = true;
+		}
+	}
+
+	if (changed) {
+		err = update_xoff_threshold(port_buffer, xoff, max_mtu, port_buff_cell_sz);
+		if (err)
+			return err;
+
+		*change = true;
+	}
+
+	return 0;
+}
+
+static int fill_pfc_en(struct mlx5_core_dev *mdev, u8 *pfc_en)
+{
+	u32 g_rx_pause, g_tx_pause;
+	int err;
+
+	err = mlx5_query_port_pause(mdev, &g_rx_pause, &g_tx_pause);
+	if (err)
+		return err;
+
+	/* If global pause enabled, set all active buffers to lossless.
+	 * Otherwise, check PFC setting.
+	 */
+	if (g_rx_pause || g_tx_pause)
+		*pfc_en = 0xff;
+	else
+		err = mlx5_query_port_pfc(mdev, pfc_en, NULL);
+
+	return err;
+}
+
+#define MINIMUM_MAX_MTU 9216
+int mlx5e_port_manual_buffer_config(struct mlx5e_priv *priv,
+				    u32 change, unsigned int mtu,
+				    struct ieee_pfc *pfc,
+				    u32 *buffer_size,
+				    u8 *prio2buffer)
+{
+	u16 port_buff_cell_sz = priv->dcbx.port_buff_cell_sz;
+	struct mlx5e_port_buffer port_buffer;
+	u32 xoff = calculate_xoff(priv, mtu);
+	bool update_prio2buffer = false;
+	u8 buffer[MLX5E_MAX_PRIORITY];
+	bool update_buffer = false;
+	unsigned int max_mtu;
+	u32 total_used = 0;
+	u8 curr_pfc_en;
+	int err;
+	int i;
+
+	mlx5e_dbg(HW, priv, "%s: change=%x\n", __func__, change);
+	max_mtu = max_t(unsigned int, priv->netdev->max_mtu, MINIMUM_MAX_MTU);
+
+	err = mlx5e_port_query_buffer(priv, &port_buffer);
+	if (err)
+		return err;
+
+	if (change & MLX5E_PORT_BUFFER_CABLE_LEN) {
+		update_buffer = true;
+		err = update_xoff_threshold(&port_buffer, xoff, max_mtu, port_buff_cell_sz);
+		if (err)
+			return err;
+	}
+
+	if (change & MLX5E_PORT_BUFFER_PFC) {
+		err = mlx5e_port_query_priority2buffer(priv->mdev, buffer);
+		if (err)
+			return err;
+
+		err = update_buffer_lossy(max_mtu, pfc->pfc_en, buffer, xoff, port_buff_cell_sz,
+					  &port_buffer, &update_buffer);
+		if (err)
+			return err;
+	}
+
+	if (change & MLX5E_PORT_BUFFER_PRIO2BUFFER) {
+		update_prio2buffer = true;
+		err = fill_pfc_en(priv->mdev, &curr_pfc_en);
+		if (err)
+			return err;
+
+		err = update_buffer_lossy(max_mtu, curr_pfc_en, prio2buffer, xoff,
+					  port_buff_cell_sz, &port_buffer, &update_buffer);
+		if (err)
+			return err;
+	}
+
+	if (change & MLX5E_PORT_BUFFER_SIZE) {
+		for (i = 0; i < MLX5E_MAX_BUFFER; i++) {
+			mlx5e_dbg(HW, priv, "%s: buffer[%d]=%d\n", __func__, i, buffer_size[i]);
+			if (!port_buffer.buffer[i].lossy && !buffer_size[i]) {
+				mlx5e_dbg(HW, priv, "%s: lossless buffer[%d] size cannot be zero\n",
+					  __func__, i);
+				return -EINVAL;
+			}
+
+			port_buffer.buffer[i].size = buffer_size[i];
+			total_used += buffer_size[i];
+		}
+
+		mlx5e_dbg(HW, priv, "%s: total buffer requested=%d\n", __func__, total_used);
+
+		if (total_used > port_buffer.port_buffer_size)
+			return -EINVAL;
+
+		update_buffer = true;
+		err = update_xoff_threshold(&port_buffer, xoff, max_mtu, port_buff_cell_sz);
+		if (err)
+			return err;
+	}
+
+	/* Need to update buffer configuration if xoff value is changed */
+	if (!update_buffer && xoff != priv->dcbx.xoff) {
+		update_buffer = true;
+		err = update_xoff_threshold(&port_buffer, xoff, max_mtu, port_buff_cell_sz);
+		if (err)
+			return err;
+	}
+	priv->dcbx.xoff = xoff;
+
+	/* Apply the settings */
+	if (update_buffer) {
+		err = port_set_buffer(priv, &port_buffer);
+		if (err)
+			return err;
+	}
+
+	if (update_prio2buffer)
+		err = mlx5e_port_set_priority2buffer(priv->mdev, prio2buffer);
+
+	return err;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/port_buffer.h b/drivers/net/ethernet/mellanox/mlx5/core/en/port_buffer.h
new file mode 100644
index 000000000..80af7a5ac
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/port_buffer.h
@@ -0,0 +1,74 @@
+/*
+ * Copyright (c) 2018, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+#ifndef __MLX5_EN_PORT_BUFFER_H__
+#define __MLX5_EN_PORT_BUFFER_H__
+
+#include "en.h"
+#include "port.h"
+
+#define MLX5E_MAX_BUFFER 8
+#define MLX5E_DEFAULT_CABLE_LEN 7 /* 7 meters */
+
+#define MLX5_BUFFER_SUPPORTED(mdev) (MLX5_CAP_GEN(mdev, pcam_reg) && \
+				     MLX5_CAP_PCAM_REG(mdev, pbmc) && \
+				     MLX5_CAP_PCAM_REG(mdev, pptb))
+
+enum {
+	MLX5E_PORT_BUFFER_CABLE_LEN   = BIT(0),
+	MLX5E_PORT_BUFFER_PFC         = BIT(1),
+	MLX5E_PORT_BUFFER_PRIO2BUFFER = BIT(2),
+	MLX5E_PORT_BUFFER_SIZE        = BIT(3),
+};
+
+struct mlx5e_bufferx_reg {
+	u8   lossy;
+	u8   epsb;
+	u32  size;
+	u32  xoff;
+	u32  xon;
+};
+
+struct mlx5e_port_buffer {
+	u32                       port_buffer_size;
+	u32                       spare_buffer_size;
+	struct mlx5e_bufferx_reg  buffer[MLX5E_MAX_BUFFER];
+};
+
+int mlx5e_port_manual_buffer_config(struct mlx5e_priv *priv,
+				    u32 change, unsigned int mtu,
+				    struct ieee_pfc *pfc,
+				    u32 *buffer_size,
+				    u8 *prio2buffer);
+
+int mlx5e_port_query_buffer(struct mlx5e_priv *priv,
+			    struct mlx5e_port_buffer *port_buffer);
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/ptp.c b/drivers/net/ethernet/mellanox/mlx5/core/en/ptp.c
new file mode 100644
index 000000000..72b4781f0
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/ptp.c
@@ -0,0 +1,877 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+// Copyright (c) 2020 Mellanox Technologies
+
+#include "en/ptp.h"
+#include "en/txrx.h"
+#include "en/params.h"
+#include "en/fs_tt_redirect.h"
+
+struct mlx5e_ptp_fs {
+	struct mlx5_flow_handle *l2_rule;
+	struct mlx5_flow_handle *udp_v4_rule;
+	struct mlx5_flow_handle *udp_v6_rule;
+	bool valid;
+};
+
+struct mlx5e_ptp_params {
+	struct mlx5e_params params;
+	struct mlx5e_sq_param txq_sq_param;
+	struct mlx5e_rq_param rq_param;
+};
+
+struct mlx5e_skb_cb_hwtstamp {
+	ktime_t cqe_hwtstamp;
+	ktime_t port_hwtstamp;
+};
+
+void mlx5e_skb_cb_hwtstamp_init(struct sk_buff *skb)
+{
+	memset(skb->cb, 0, sizeof(struct mlx5e_skb_cb_hwtstamp));
+}
+
+static struct mlx5e_skb_cb_hwtstamp *mlx5e_skb_cb_get_hwts(struct sk_buff *skb)
+{
+	BUILD_BUG_ON(sizeof(struct mlx5e_skb_cb_hwtstamp) > sizeof(skb->cb));
+	return (struct mlx5e_skb_cb_hwtstamp *)skb->cb;
+}
+
+static void mlx5e_skb_cb_hwtstamp_tx(struct sk_buff *skb,
+				     struct mlx5e_ptp_cq_stats *cq_stats)
+{
+	struct skb_shared_hwtstamps hwts = {};
+	ktime_t diff;
+
+	diff = abs(mlx5e_skb_cb_get_hwts(skb)->port_hwtstamp -
+		   mlx5e_skb_cb_get_hwts(skb)->cqe_hwtstamp);
+
+	/* Maximal allowed diff is 1 / 128 second */
+	if (diff > (NSEC_PER_SEC >> 7)) {
+		cq_stats->abort++;
+		cq_stats->abort_abs_diff_ns += diff;
+		return;
+	}
+
+	hwts.hwtstamp = mlx5e_skb_cb_get_hwts(skb)->port_hwtstamp;
+	skb_tstamp_tx(skb, &hwts);
+}
+
+void mlx5e_skb_cb_hwtstamp_handler(struct sk_buff *skb, int hwtstamp_type,
+				   ktime_t hwtstamp,
+				   struct mlx5e_ptp_cq_stats *cq_stats)
+{
+	switch (hwtstamp_type) {
+	case (MLX5E_SKB_CB_CQE_HWTSTAMP):
+		mlx5e_skb_cb_get_hwts(skb)->cqe_hwtstamp = hwtstamp;
+		break;
+	case (MLX5E_SKB_CB_PORT_HWTSTAMP):
+		mlx5e_skb_cb_get_hwts(skb)->port_hwtstamp = hwtstamp;
+		break;
+	}
+
+	/* If both CQEs arrive, check and report the port tstamp, and clear skb cb as
+	 * skb soon to be released.
+	 */
+	if (!mlx5e_skb_cb_get_hwts(skb)->cqe_hwtstamp ||
+	    !mlx5e_skb_cb_get_hwts(skb)->port_hwtstamp)
+		return;
+
+	mlx5e_skb_cb_hwtstamp_tx(skb, cq_stats);
+	memset(skb->cb, 0, sizeof(struct mlx5e_skb_cb_hwtstamp));
+}
+
+#define PTP_WQE_CTR2IDX(val) ((val) & ptpsq->ts_cqe_ctr_mask)
+
+static bool mlx5e_ptp_ts_cqe_drop(struct mlx5e_ptpsq *ptpsq, u16 skb_cc, u16 skb_id)
+{
+	return (ptpsq->ts_cqe_ctr_mask && (skb_cc != skb_id));
+}
+
+static bool mlx5e_ptp_ts_cqe_ooo(struct mlx5e_ptpsq *ptpsq, u16 skb_id)
+{
+	u16 skb_cc = PTP_WQE_CTR2IDX(ptpsq->skb_fifo_cc);
+	u16 skb_pc = PTP_WQE_CTR2IDX(ptpsq->skb_fifo_pc);
+
+	if (PTP_WQE_CTR2IDX(skb_id - skb_cc) >= PTP_WQE_CTR2IDX(skb_pc - skb_cc))
+		return true;
+
+	return false;
+}
+
+static void mlx5e_ptp_skb_fifo_ts_cqe_resync(struct mlx5e_ptpsq *ptpsq, u16 skb_cc,
+					     u16 skb_id, int budget)
+{
+	struct skb_shared_hwtstamps hwts = {};
+	struct sk_buff *skb;
+
+	ptpsq->cq_stats->resync_event++;
+
+	while (skb_cc != skb_id) {
+		skb = mlx5e_skb_fifo_pop(&ptpsq->skb_fifo);
+		hwts.hwtstamp = mlx5e_skb_cb_get_hwts(skb)->cqe_hwtstamp;
+		skb_tstamp_tx(skb, &hwts);
+		ptpsq->cq_stats->resync_cqe++;
+		napi_consume_skb(skb, budget);
+		skb_cc = PTP_WQE_CTR2IDX(ptpsq->skb_fifo_cc);
+	}
+}
+
+static void mlx5e_ptp_handle_ts_cqe(struct mlx5e_ptpsq *ptpsq,
+				    struct mlx5_cqe64 *cqe,
+				    int budget)
+{
+	u16 skb_id = PTP_WQE_CTR2IDX(be16_to_cpu(cqe->wqe_counter));
+	u16 skb_cc = PTP_WQE_CTR2IDX(ptpsq->skb_fifo_cc);
+	struct mlx5e_txqsq *sq = &ptpsq->txqsq;
+	struct sk_buff *skb;
+	ktime_t hwtstamp;
+
+	if (unlikely(MLX5E_RX_ERR_CQE(cqe))) {
+		skb = mlx5e_skb_fifo_pop(&ptpsq->skb_fifo);
+		ptpsq->cq_stats->err_cqe++;
+		goto out;
+	}
+
+	if (mlx5e_ptp_ts_cqe_drop(ptpsq, skb_cc, skb_id)) {
+		if (mlx5e_ptp_ts_cqe_ooo(ptpsq, skb_id)) {
+			/* already handled by a previous resync */
+			ptpsq->cq_stats->ooo_cqe_drop++;
+			return;
+		}
+		mlx5e_ptp_skb_fifo_ts_cqe_resync(ptpsq, skb_cc, skb_id, budget);
+	}
+
+	skb = mlx5e_skb_fifo_pop(&ptpsq->skb_fifo);
+	hwtstamp = mlx5e_cqe_ts_to_ns(sq->ptp_cyc2time, sq->clock, get_cqe_ts(cqe));
+	mlx5e_skb_cb_hwtstamp_handler(skb, MLX5E_SKB_CB_PORT_HWTSTAMP,
+				      hwtstamp, ptpsq->cq_stats);
+	ptpsq->cq_stats->cqe++;
+
+out:
+	napi_consume_skb(skb, budget);
+}
+
+static bool mlx5e_ptp_poll_ts_cq(struct mlx5e_cq *cq, int budget)
+{
+	struct mlx5e_ptpsq *ptpsq = container_of(cq, struct mlx5e_ptpsq, ts_cq);
+	struct mlx5_cqwq *cqwq = &cq->wq;
+	struct mlx5_cqe64 *cqe;
+	int work_done = 0;
+
+	if (unlikely(!test_bit(MLX5E_SQ_STATE_ENABLED, &ptpsq->txqsq.state)))
+		return false;
+
+	cqe = mlx5_cqwq_get_cqe(cqwq);
+	if (!cqe)
+		return false;
+
+	do {
+		mlx5_cqwq_pop(cqwq);
+
+		mlx5e_ptp_handle_ts_cqe(ptpsq, cqe, budget);
+	} while ((++work_done < budget) && (cqe = mlx5_cqwq_get_cqe(cqwq)));
+
+	mlx5_cqwq_update_db_record(cqwq);
+
+	/* ensure cq space is freed before enabling more cqes */
+	wmb();
+
+	mlx5e_txqsq_wake(&ptpsq->txqsq);
+
+	return work_done == budget;
+}
+
+static int mlx5e_ptp_napi_poll(struct napi_struct *napi, int budget)
+{
+	struct mlx5e_ptp *c = container_of(napi, struct mlx5e_ptp, napi);
+	struct mlx5e_ch_stats *ch_stats = c->stats;
+	struct mlx5e_rq *rq = &c->rq;
+	bool busy = false;
+	int work_done = 0;
+	int i;
+
+	rcu_read_lock();
+
+	ch_stats->poll++;
+
+	if (test_bit(MLX5E_PTP_STATE_TX, c->state)) {
+		for (i = 0; i < c->num_tc; i++) {
+			busy |= mlx5e_poll_tx_cq(&c->ptpsq[i].txqsq.cq, budget);
+			busy |= mlx5e_ptp_poll_ts_cq(&c->ptpsq[i].ts_cq, budget);
+		}
+	}
+	if (test_bit(MLX5E_PTP_STATE_RX, c->state) && likely(budget)) {
+		work_done = mlx5e_poll_rx_cq(&rq->cq, budget);
+		busy |= work_done == budget;
+		busy |= INDIRECT_CALL_2(rq->post_wqes,
+					mlx5e_post_rx_mpwqes,
+					mlx5e_post_rx_wqes,
+					rq);
+	}
+
+	if (busy) {
+		work_done = budget;
+		goto out;
+	}
+
+	if (unlikely(!napi_complete_done(napi, work_done)))
+		goto out;
+
+	ch_stats->arm++;
+
+	if (test_bit(MLX5E_PTP_STATE_TX, c->state)) {
+		for (i = 0; i < c->num_tc; i++) {
+			mlx5e_cq_arm(&c->ptpsq[i].txqsq.cq);
+			mlx5e_cq_arm(&c->ptpsq[i].ts_cq);
+		}
+	}
+	if (test_bit(MLX5E_PTP_STATE_RX, c->state))
+		mlx5e_cq_arm(&rq->cq);
+
+out:
+	rcu_read_unlock();
+
+	return work_done;
+}
+
+static int mlx5e_ptp_alloc_txqsq(struct mlx5e_ptp *c, int txq_ix,
+				 struct mlx5e_params *params,
+				 struct mlx5e_sq_param *param,
+				 struct mlx5e_txqsq *sq, int tc,
+				 struct mlx5e_ptpsq *ptpsq)
+{
+	void *sqc_wq               = MLX5_ADDR_OF(sqc, param->sqc, wq);
+	struct mlx5_core_dev *mdev = c->mdev;
+	struct mlx5_wq_cyc *wq = &sq->wq;
+	int err;
+	int node;
+
+	sq->pdev      = c->pdev;
+	sq->clock     = &mdev->clock;
+	sq->mkey_be   = c->mkey_be;
+	sq->netdev    = c->netdev;
+	sq->priv      = c->priv;
+	sq->mdev      = mdev;
+	sq->ch_ix     = MLX5E_PTP_CHANNEL_IX;
+	sq->txq_ix    = txq_ix;
+	sq->uar_map   = mdev->mlx5e_res.hw_objs.bfreg.map;
+	sq->min_inline_mode = params->tx_min_inline_mode;
+	sq->hw_mtu    = MLX5E_SW2HW_MTU(params, params->sw_mtu);
+	sq->stats     = &c->priv->ptp_stats.sq[tc];
+	sq->ptpsq     = ptpsq;
+	INIT_WORK(&sq->recover_work, mlx5e_tx_err_cqe_work);
+	if (!MLX5_CAP_ETH(mdev, wqe_vlan_insert))
+		set_bit(MLX5E_SQ_STATE_VLAN_NEED_L2_INLINE, &sq->state);
+	sq->stop_room = param->stop_room;
+	sq->ptp_cyc2time = mlx5_sq_ts_translator(mdev);
+
+	node = dev_to_node(mlx5_core_dma_dev(mdev));
+
+	param->wq.db_numa_node = node;
+	err = mlx5_wq_cyc_create(mdev, &param->wq, sqc_wq, wq, &sq->wq_ctrl);
+	if (err)
+		return err;
+	wq->db    = &wq->db[MLX5_SND_DBR];
+
+	err = mlx5e_alloc_txqsq_db(sq, node);
+	if (err)
+		goto err_sq_wq_destroy;
+
+	return 0;
+
+err_sq_wq_destroy:
+	mlx5_wq_destroy(&sq->wq_ctrl);
+
+	return err;
+}
+
+static void mlx5e_ptp_destroy_sq(struct mlx5_core_dev *mdev, u32 sqn)
+{
+	mlx5_core_destroy_sq(mdev, sqn);
+}
+
+static int mlx5e_ptp_alloc_traffic_db(struct mlx5e_ptpsq *ptpsq, int numa)
+{
+	int wq_sz = mlx5_wq_cyc_get_size(&ptpsq->txqsq.wq);
+	struct mlx5_core_dev *mdev = ptpsq->txqsq.mdev;
+
+	ptpsq->skb_fifo.fifo = kvzalloc_node(array_size(wq_sz, sizeof(*ptpsq->skb_fifo.fifo)),
+					     GFP_KERNEL, numa);
+	if (!ptpsq->skb_fifo.fifo)
+		return -ENOMEM;
+
+	ptpsq->skb_fifo.pc   = &ptpsq->skb_fifo_pc;
+	ptpsq->skb_fifo.cc   = &ptpsq->skb_fifo_cc;
+	ptpsq->skb_fifo.mask = wq_sz - 1;
+	if (MLX5_CAP_GEN_2(mdev, ts_cqe_metadata_size2wqe_counter))
+		ptpsq->ts_cqe_ctr_mask =
+			(1 << MLX5_CAP_GEN_2(mdev, ts_cqe_metadata_size2wqe_counter)) - 1;
+	return 0;
+}
+
+static void mlx5e_ptp_drain_skb_fifo(struct mlx5e_skb_fifo *skb_fifo)
+{
+	while (*skb_fifo->pc != *skb_fifo->cc) {
+		struct sk_buff *skb = mlx5e_skb_fifo_pop(skb_fifo);
+
+		dev_kfree_skb_any(skb);
+	}
+}
+
+static void mlx5e_ptp_free_traffic_db(struct mlx5e_skb_fifo *skb_fifo)
+{
+	mlx5e_ptp_drain_skb_fifo(skb_fifo);
+	kvfree(skb_fifo->fifo);
+}
+
+static int mlx5e_ptp_open_txqsq(struct mlx5e_ptp *c, u32 tisn,
+				int txq_ix, struct mlx5e_ptp_params *cparams,
+				int tc, struct mlx5e_ptpsq *ptpsq)
+{
+	struct mlx5e_sq_param *sqp = &cparams->txq_sq_param;
+	struct mlx5e_txqsq *txqsq = &ptpsq->txqsq;
+	struct mlx5e_create_sq_param csp = {};
+	int err;
+
+	err = mlx5e_ptp_alloc_txqsq(c, txq_ix, &cparams->params, sqp,
+				    txqsq, tc, ptpsq);
+	if (err)
+		return err;
+
+	csp.tisn            = tisn;
+	csp.tis_lst_sz      = 1;
+	csp.cqn             = txqsq->cq.mcq.cqn;
+	csp.wq_ctrl         = &txqsq->wq_ctrl;
+	csp.min_inline_mode = txqsq->min_inline_mode;
+	csp.ts_cqe_to_dest_cqn = ptpsq->ts_cq.mcq.cqn;
+
+	err = mlx5e_create_sq_rdy(c->mdev, sqp, &csp, 0, &txqsq->sqn);
+	if (err)
+		goto err_free_txqsq;
+
+	err = mlx5e_ptp_alloc_traffic_db(ptpsq,
+					 dev_to_node(mlx5_core_dma_dev(c->mdev)));
+	if (err)
+		goto err_free_txqsq;
+
+	return 0;
+
+err_free_txqsq:
+	mlx5e_free_txqsq(txqsq);
+
+	return err;
+}
+
+static void mlx5e_ptp_close_txqsq(struct mlx5e_ptpsq *ptpsq)
+{
+	struct mlx5e_txqsq *sq = &ptpsq->txqsq;
+	struct mlx5_core_dev *mdev = sq->mdev;
+
+	mlx5e_ptp_free_traffic_db(&ptpsq->skb_fifo);
+	cancel_work_sync(&sq->recover_work);
+	mlx5e_ptp_destroy_sq(mdev, sq->sqn);
+	mlx5e_free_txqsq_descs(sq);
+	mlx5e_free_txqsq(sq);
+}
+
+static int mlx5e_ptp_open_txqsqs(struct mlx5e_ptp *c,
+				 struct mlx5e_ptp_params *cparams)
+{
+	struct mlx5e_params *params = &cparams->params;
+	u8 num_tc = mlx5e_get_dcb_num_tc(params);
+	int ix_base;
+	int err;
+	int tc;
+
+	ix_base = num_tc * params->num_channels;
+
+	for (tc = 0; tc < num_tc; tc++) {
+		int txq_ix = ix_base + tc;
+
+		err = mlx5e_ptp_open_txqsq(c, c->priv->tisn[c->lag_port][tc], txq_ix,
+					   cparams, tc, &c->ptpsq[tc]);
+		if (err)
+			goto close_txqsq;
+	}
+
+	return 0;
+
+close_txqsq:
+	for (--tc; tc >= 0; tc--)
+		mlx5e_ptp_close_txqsq(&c->ptpsq[tc]);
+
+	return err;
+}
+
+static void mlx5e_ptp_close_txqsqs(struct mlx5e_ptp *c)
+{
+	int tc;
+
+	for (tc = 0; tc < c->num_tc; tc++)
+		mlx5e_ptp_close_txqsq(&c->ptpsq[tc]);
+}
+
+static int mlx5e_ptp_open_tx_cqs(struct mlx5e_ptp *c,
+				 struct mlx5e_ptp_params *cparams)
+{
+	struct mlx5e_params *params = &cparams->params;
+	struct mlx5e_create_cq_param ccp = {};
+	struct dim_cq_moder ptp_moder = {};
+	struct mlx5e_cq_param *cq_param;
+	u8 num_tc;
+	int err;
+	int tc;
+
+	num_tc = mlx5e_get_dcb_num_tc(params);
+
+	ccp.node     = dev_to_node(mlx5_core_dma_dev(c->mdev));
+	ccp.ch_stats = c->stats;
+	ccp.napi     = &c->napi;
+	ccp.ix       = MLX5E_PTP_CHANNEL_IX;
+
+	cq_param = &cparams->txq_sq_param.cqp;
+
+	for (tc = 0; tc < num_tc; tc++) {
+		struct mlx5e_cq *cq = &c->ptpsq[tc].txqsq.cq;
+
+		err = mlx5e_open_cq(c->priv, ptp_moder, cq_param, &ccp, cq);
+		if (err)
+			goto out_err_txqsq_cq;
+	}
+
+	for (tc = 0; tc < num_tc; tc++) {
+		struct mlx5e_cq *cq = &c->ptpsq[tc].ts_cq;
+		struct mlx5e_ptpsq *ptpsq = &c->ptpsq[tc];
+
+		err = mlx5e_open_cq(c->priv, ptp_moder, cq_param, &ccp, cq);
+		if (err)
+			goto out_err_ts_cq;
+
+		ptpsq->cq_stats = &c->priv->ptp_stats.cq[tc];
+	}
+
+	return 0;
+
+out_err_ts_cq:
+	for (--tc; tc >= 0; tc--)
+		mlx5e_close_cq(&c->ptpsq[tc].ts_cq);
+	tc = num_tc;
+out_err_txqsq_cq:
+	for (--tc; tc >= 0; tc--)
+		mlx5e_close_cq(&c->ptpsq[tc].txqsq.cq);
+
+	return err;
+}
+
+static int mlx5e_ptp_open_rx_cq(struct mlx5e_ptp *c,
+				struct mlx5e_ptp_params *cparams)
+{
+	struct mlx5e_create_cq_param ccp = {};
+	struct dim_cq_moder ptp_moder = {};
+	struct mlx5e_cq_param *cq_param;
+	struct mlx5e_cq *cq = &c->rq.cq;
+
+	ccp.node     = dev_to_node(mlx5_core_dma_dev(c->mdev));
+	ccp.ch_stats = c->stats;
+	ccp.napi     = &c->napi;
+	ccp.ix       = MLX5E_PTP_CHANNEL_IX;
+
+	cq_param = &cparams->rq_param.cqp;
+
+	return mlx5e_open_cq(c->priv, ptp_moder, cq_param, &ccp, cq);
+}
+
+static void mlx5e_ptp_close_tx_cqs(struct mlx5e_ptp *c)
+{
+	int tc;
+
+	for (tc = 0; tc < c->num_tc; tc++)
+		mlx5e_close_cq(&c->ptpsq[tc].ts_cq);
+
+	for (tc = 0; tc < c->num_tc; tc++)
+		mlx5e_close_cq(&c->ptpsq[tc].txqsq.cq);
+}
+
+static void mlx5e_ptp_build_sq_param(struct mlx5_core_dev *mdev,
+				     struct mlx5e_params *params,
+				     struct mlx5e_sq_param *param)
+{
+	void *sqc = param->sqc;
+	void *wq;
+
+	mlx5e_build_sq_param_common(mdev, param);
+
+	wq = MLX5_ADDR_OF(sqc, sqc, wq);
+	MLX5_SET(wq, wq, log_wq_sz, params->log_sq_size);
+	param->stop_room = mlx5e_stop_room_for_max_wqe(mdev);
+	mlx5e_build_tx_cq_param(mdev, params, &param->cqp);
+}
+
+static void mlx5e_ptp_build_rq_param(struct mlx5_core_dev *mdev,
+				     struct net_device *netdev,
+				     u16 q_counter,
+				     struct mlx5e_ptp_params *ptp_params)
+{
+	struct mlx5e_rq_param *rq_params = &ptp_params->rq_param;
+	struct mlx5e_params *params = &ptp_params->params;
+
+	params->rq_wq_type = MLX5_WQ_TYPE_CYCLIC;
+	mlx5e_init_rq_type_params(mdev, params);
+	params->sw_mtu = netdev->max_mtu;
+	mlx5e_build_rq_param(mdev, params, NULL, q_counter, rq_params);
+}
+
+static void mlx5e_ptp_build_params(struct mlx5e_ptp *c,
+				   struct mlx5e_ptp_params *cparams,
+				   struct mlx5e_params *orig)
+{
+	struct mlx5e_params *params = &cparams->params;
+
+	params->tx_min_inline_mode = orig->tx_min_inline_mode;
+	params->num_channels = orig->num_channels;
+	params->hard_mtu = orig->hard_mtu;
+	params->sw_mtu = orig->sw_mtu;
+	params->mqprio = orig->mqprio;
+
+	/* SQ */
+	if (test_bit(MLX5E_PTP_STATE_TX, c->state)) {
+		params->log_sq_size = orig->log_sq_size;
+		mlx5e_ptp_build_sq_param(c->mdev, params, &cparams->txq_sq_param);
+	}
+	/* RQ */
+	if (test_bit(MLX5E_PTP_STATE_RX, c->state)) {
+		params->vlan_strip_disable = orig->vlan_strip_disable;
+		mlx5e_ptp_build_rq_param(c->mdev, c->netdev, c->priv->q_counter, cparams);
+	}
+}
+
+static int mlx5e_init_ptp_rq(struct mlx5e_ptp *c, struct mlx5e_params *params,
+			     struct mlx5e_rq *rq)
+{
+	struct mlx5_core_dev *mdev = c->mdev;
+	struct mlx5e_priv *priv = c->priv;
+	int err;
+
+	rq->wq_type      = params->rq_wq_type;
+	rq->pdev         = c->pdev;
+	rq->netdev       = priv->netdev;
+	rq->priv         = priv;
+	rq->clock        = &mdev->clock;
+	rq->tstamp       = &priv->tstamp;
+	rq->mdev         = mdev;
+	rq->hw_mtu       = MLX5E_SW2HW_MTU(params, params->sw_mtu);
+	rq->stats        = &c->priv->ptp_stats.rq;
+	rq->ix           = MLX5E_PTP_CHANNEL_IX;
+	rq->ptp_cyc2time = mlx5_rq_ts_translator(mdev);
+	err = mlx5e_rq_set_handlers(rq, params, false);
+	if (err)
+		return err;
+
+	return xdp_rxq_info_reg(&rq->xdp_rxq, rq->netdev, rq->ix, 0);
+}
+
+static int mlx5e_ptp_open_rq(struct mlx5e_ptp *c, struct mlx5e_params *params,
+			     struct mlx5e_rq_param *rq_param)
+{
+	int node = dev_to_node(c->mdev->device);
+	int err;
+
+	err = mlx5e_init_ptp_rq(c, params, &c->rq);
+	if (err)
+		return err;
+
+	return mlx5e_open_rq(params, rq_param, NULL, node, &c->rq);
+}
+
+static int mlx5e_ptp_open_queues(struct mlx5e_ptp *c,
+				 struct mlx5e_ptp_params *cparams)
+{
+	int err;
+
+	if (test_bit(MLX5E_PTP_STATE_TX, c->state)) {
+		err = mlx5e_ptp_open_tx_cqs(c, cparams);
+		if (err)
+			return err;
+
+		err = mlx5e_ptp_open_txqsqs(c, cparams);
+		if (err)
+			goto close_tx_cqs;
+	}
+	if (test_bit(MLX5E_PTP_STATE_RX, c->state)) {
+		err = mlx5e_ptp_open_rx_cq(c, cparams);
+		if (err)
+			goto close_txqsq;
+
+		err = mlx5e_ptp_open_rq(c, &cparams->params, &cparams->rq_param);
+		if (err)
+			goto close_rx_cq;
+	}
+	return 0;
+
+close_rx_cq:
+	if (test_bit(MLX5E_PTP_STATE_RX, c->state))
+		mlx5e_close_cq(&c->rq.cq);
+close_txqsq:
+	if (test_bit(MLX5E_PTP_STATE_TX, c->state))
+		mlx5e_ptp_close_txqsqs(c);
+close_tx_cqs:
+	if (test_bit(MLX5E_PTP_STATE_TX, c->state))
+		mlx5e_ptp_close_tx_cqs(c);
+
+	return err;
+}
+
+static void mlx5e_ptp_close_queues(struct mlx5e_ptp *c)
+{
+	if (test_bit(MLX5E_PTP_STATE_RX, c->state)) {
+		mlx5e_close_rq(&c->rq);
+		mlx5e_close_cq(&c->rq.cq);
+	}
+	if (test_bit(MLX5E_PTP_STATE_TX, c->state)) {
+		mlx5e_ptp_close_txqsqs(c);
+		mlx5e_ptp_close_tx_cqs(c);
+	}
+}
+
+static int mlx5e_ptp_set_state(struct mlx5e_ptp *c, struct mlx5e_params *params)
+{
+	if (MLX5E_GET_PFLAG(params, MLX5E_PFLAG_TX_PORT_TS))
+		__set_bit(MLX5E_PTP_STATE_TX, c->state);
+
+	if (params->ptp_rx)
+		__set_bit(MLX5E_PTP_STATE_RX, c->state);
+
+	return bitmap_empty(c->state, MLX5E_PTP_STATE_NUM_STATES) ? -EINVAL : 0;
+}
+
+static void mlx5e_ptp_rx_unset_fs(struct mlx5e_flow_steering *fs)
+{
+	struct mlx5e_ptp_fs *ptp_fs = mlx5e_fs_get_ptp(fs);
+
+	if (!ptp_fs->valid)
+		return;
+
+	mlx5e_fs_tt_redirect_del_rule(ptp_fs->l2_rule);
+	mlx5e_fs_tt_redirect_any_destroy(fs);
+
+	mlx5e_fs_tt_redirect_del_rule(ptp_fs->udp_v6_rule);
+	mlx5e_fs_tt_redirect_del_rule(ptp_fs->udp_v4_rule);
+	mlx5e_fs_tt_redirect_udp_destroy(fs);
+	ptp_fs->valid = false;
+}
+
+static int mlx5e_ptp_rx_set_fs(struct mlx5e_priv *priv)
+{
+	u32 tirn = mlx5e_rx_res_get_tirn_ptp(priv->rx_res);
+	struct mlx5e_flow_steering *fs = priv->fs;
+	struct mlx5_flow_handle *rule;
+	struct mlx5e_ptp_fs *ptp_fs;
+	int err;
+
+	ptp_fs = mlx5e_fs_get_ptp(fs);
+	if (ptp_fs->valid)
+		return 0;
+
+	err = mlx5e_fs_tt_redirect_udp_create(fs);
+	if (err)
+		goto out_free;
+
+	rule = mlx5e_fs_tt_redirect_udp_add_rule(fs, MLX5_TT_IPV4_UDP,
+						 tirn, PTP_EV_PORT);
+	if (IS_ERR(rule)) {
+		err = PTR_ERR(rule);
+		goto out_destroy_fs_udp;
+	}
+	ptp_fs->udp_v4_rule = rule;
+
+	rule = mlx5e_fs_tt_redirect_udp_add_rule(fs, MLX5_TT_IPV6_UDP,
+						 tirn, PTP_EV_PORT);
+	if (IS_ERR(rule)) {
+		err = PTR_ERR(rule);
+		goto out_destroy_udp_v4_rule;
+	}
+	ptp_fs->udp_v6_rule = rule;
+
+	err = mlx5e_fs_tt_redirect_any_create(fs);
+	if (err)
+		goto out_destroy_udp_v6_rule;
+
+	rule = mlx5e_fs_tt_redirect_any_add_rule(fs, tirn, ETH_P_1588);
+	if (IS_ERR(rule)) {
+		err = PTR_ERR(rule);
+		goto out_destroy_fs_any;
+	}
+	ptp_fs->l2_rule = rule;
+	ptp_fs->valid = true;
+
+	return 0;
+
+out_destroy_fs_any:
+	mlx5e_fs_tt_redirect_any_destroy(fs);
+out_destroy_udp_v6_rule:
+	mlx5e_fs_tt_redirect_del_rule(ptp_fs->udp_v6_rule);
+out_destroy_udp_v4_rule:
+	mlx5e_fs_tt_redirect_del_rule(ptp_fs->udp_v4_rule);
+out_destroy_fs_udp:
+	mlx5e_fs_tt_redirect_udp_destroy(fs);
+out_free:
+	return err;
+}
+
+int mlx5e_ptp_open(struct mlx5e_priv *priv, struct mlx5e_params *params,
+		   u8 lag_port, struct mlx5e_ptp **cp)
+{
+	struct net_device *netdev = priv->netdev;
+	struct mlx5_core_dev *mdev = priv->mdev;
+	struct mlx5e_ptp_params *cparams;
+	struct mlx5e_ptp *c;
+	int err;
+
+
+	c = kvzalloc_node(sizeof(*c), GFP_KERNEL, dev_to_node(mlx5_core_dma_dev(mdev)));
+	cparams = kvzalloc(sizeof(*cparams), GFP_KERNEL);
+	if (!c || !cparams) {
+		err = -ENOMEM;
+		goto err_free;
+	}
+
+	c->priv     = priv;
+	c->mdev     = priv->mdev;
+	c->tstamp   = &priv->tstamp;
+	c->pdev     = mlx5_core_dma_dev(priv->mdev);
+	c->netdev   = priv->netdev;
+	c->mkey_be  = cpu_to_be32(priv->mdev->mlx5e_res.hw_objs.mkey);
+	c->num_tc   = mlx5e_get_dcb_num_tc(params);
+	c->stats    = &priv->ptp_stats.ch;
+	c->lag_port = lag_port;
+
+	err = mlx5e_ptp_set_state(c, params);
+	if (err)
+		goto err_free;
+
+	netif_napi_add(netdev, &c->napi, mlx5e_ptp_napi_poll);
+
+	mlx5e_ptp_build_params(c, cparams, params);
+
+	err = mlx5e_ptp_open_queues(c, cparams);
+	if (unlikely(err))
+		goto err_napi_del;
+
+	if (test_bit(MLX5E_PTP_STATE_RX, c->state))
+		priv->rx_ptp_opened = true;
+
+	*cp = c;
+
+	kvfree(cparams);
+
+	return 0;
+
+err_napi_del:
+	netif_napi_del(&c->napi);
+err_free:
+	kvfree(cparams);
+	kvfree(c);
+	return err;
+}
+
+void mlx5e_ptp_close(struct mlx5e_ptp *c)
+{
+	mlx5e_ptp_close_queues(c);
+	netif_napi_del(&c->napi);
+
+	kvfree(c);
+}
+
+void mlx5e_ptp_activate_channel(struct mlx5e_ptp *c)
+{
+	int tc;
+
+	napi_enable(&c->napi);
+
+	if (test_bit(MLX5E_PTP_STATE_TX, c->state)) {
+		for (tc = 0; tc < c->num_tc; tc++)
+			mlx5e_activate_txqsq(&c->ptpsq[tc].txqsq);
+	}
+	if (test_bit(MLX5E_PTP_STATE_RX, c->state)) {
+		mlx5e_ptp_rx_set_fs(c->priv);
+		mlx5e_activate_rq(&c->rq);
+		mlx5e_trigger_napi_sched(&c->napi);
+	}
+}
+
+void mlx5e_ptp_deactivate_channel(struct mlx5e_ptp *c)
+{
+	int tc;
+
+	if (test_bit(MLX5E_PTP_STATE_RX, c->state))
+		mlx5e_deactivate_rq(&c->rq);
+
+	if (test_bit(MLX5E_PTP_STATE_TX, c->state)) {
+		for (tc = 0; tc < c->num_tc; tc++)
+			mlx5e_deactivate_txqsq(&c->ptpsq[tc].txqsq);
+	}
+
+	napi_disable(&c->napi);
+}
+
+int mlx5e_ptp_get_rqn(struct mlx5e_ptp *c, u32 *rqn)
+{
+	if (!c || !test_bit(MLX5E_PTP_STATE_RX, c->state))
+		return -EINVAL;
+
+	*rqn = c->rq.rqn;
+	return 0;
+}
+
+int mlx5e_ptp_alloc_rx_fs(struct mlx5e_flow_steering *fs,
+			  const struct mlx5e_profile *profile)
+{
+	struct mlx5e_ptp_fs *ptp_fs;
+
+	if (!mlx5e_profile_feature_cap(profile, PTP_RX))
+		return 0;
+
+	ptp_fs = kzalloc(sizeof(*ptp_fs), GFP_KERNEL);
+	if (!ptp_fs)
+		return -ENOMEM;
+	mlx5e_fs_set_ptp(fs, ptp_fs);
+
+	return 0;
+}
+
+void mlx5e_ptp_free_rx_fs(struct mlx5e_flow_steering *fs,
+			  const struct mlx5e_profile *profile)
+{
+	struct mlx5e_ptp_fs *ptp_fs = mlx5e_fs_get_ptp(fs);
+
+	if (!mlx5e_profile_feature_cap(profile, PTP_RX))
+		return;
+
+	mlx5e_ptp_rx_unset_fs(fs);
+	kfree(ptp_fs);
+}
+
+int mlx5e_ptp_rx_manage_fs(struct mlx5e_priv *priv, bool set)
+{
+	struct mlx5e_ptp *c = priv->channels.ptp;
+
+	if (!mlx5e_profile_feature_cap(priv->profile, PTP_RX))
+		return 0;
+
+	if (!test_bit(MLX5E_STATE_OPENED, &priv->state))
+		return 0;
+
+	if (set) {
+		if (!c || !test_bit(MLX5E_PTP_STATE_RX, c->state)) {
+			netdev_WARN_ONCE(priv->netdev, "Don't try to add PTP RX-FS rules");
+			return -EINVAL;
+		}
+		return mlx5e_ptp_rx_set_fs(priv);
+	}
+	/* set == false */
+	if (c && test_bit(MLX5E_PTP_STATE_RX, c->state)) {
+		netdev_WARN_ONCE(priv->netdev, "Don't try to remove PTP RX-FS rules");
+		return -EINVAL;
+	}
+	mlx5e_ptp_rx_unset_fs(priv->fs);
+	return 0;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/ptp.h b/drivers/net/ethernet/mellanox/mlx5/core/en/ptp.h
new file mode 100644
index 000000000..cc7efde88
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/ptp.h
@@ -0,0 +1,102 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2020 Mellanox Technologies. */
+
+#ifndef __MLX5_EN_PTP_H__
+#define __MLX5_EN_PTP_H__
+
+#include "en.h"
+#include "en_stats.h"
+#include "en/txrx.h"
+#include <linux/ptp_classify.h>
+
+#define MLX5E_PTP_CHANNEL_IX 0
+
+struct mlx5e_ptpsq {
+	struct mlx5e_txqsq       txqsq;
+	struct mlx5e_cq          ts_cq;
+	u16                      skb_fifo_cc;
+	u16                      skb_fifo_pc;
+	struct mlx5e_skb_fifo    skb_fifo;
+	struct mlx5e_ptp_cq_stats *cq_stats;
+	u16                      ts_cqe_ctr_mask;
+};
+
+enum {
+	MLX5E_PTP_STATE_TX,
+	MLX5E_PTP_STATE_RX,
+	MLX5E_PTP_STATE_NUM_STATES,
+};
+
+struct mlx5e_ptp {
+	/* data path */
+	struct mlx5e_ptpsq         ptpsq[MLX5E_MAX_NUM_TC];
+	struct mlx5e_rq            rq;
+	struct napi_struct         napi;
+	struct device             *pdev;
+	struct net_device         *netdev;
+	__be32                     mkey_be;
+	u8                         num_tc;
+	u8                         lag_port;
+
+	/* data path - accessed per napi poll */
+	struct mlx5e_ch_stats     *stats;
+
+	/* control */
+	struct mlx5e_priv         *priv;
+	struct mlx5_core_dev      *mdev;
+	struct hwtstamp_config    *tstamp;
+	DECLARE_BITMAP(state, MLX5E_PTP_STATE_NUM_STATES);
+};
+
+static inline bool mlx5e_use_ptpsq(struct sk_buff *skb)
+{
+	struct flow_keys fk;
+
+	if (!(skb_shinfo(skb)->tx_flags & SKBTX_HW_TSTAMP))
+		return false;
+
+	if (!skb_flow_dissect_flow_keys(skb, &fk, 0))
+		return false;
+
+	if (fk.basic.n_proto == htons(ETH_P_1588))
+		return true;
+
+	if (fk.basic.n_proto != htons(ETH_P_IP) &&
+	    fk.basic.n_proto != htons(ETH_P_IPV6))
+		return false;
+
+	return (fk.basic.ip_proto == IPPROTO_UDP &&
+		fk.ports.dst == htons(PTP_EV_PORT));
+}
+
+static inline bool mlx5e_ptpsq_fifo_has_room(struct mlx5e_txqsq *sq)
+{
+	if (!sq->ptpsq)
+		return true;
+
+	return mlx5e_skb_fifo_has_room(&sq->ptpsq->skb_fifo);
+}
+
+int mlx5e_ptp_open(struct mlx5e_priv *priv, struct mlx5e_params *params,
+		   u8 lag_port, struct mlx5e_ptp **cp);
+void mlx5e_ptp_close(struct mlx5e_ptp *c);
+void mlx5e_ptp_activate_channel(struct mlx5e_ptp *c);
+void mlx5e_ptp_deactivate_channel(struct mlx5e_ptp *c);
+int mlx5e_ptp_get_rqn(struct mlx5e_ptp *c, u32 *rqn);
+int mlx5e_ptp_alloc_rx_fs(struct mlx5e_flow_steering *fs,
+			  const struct mlx5e_profile *profile);
+void mlx5e_ptp_free_rx_fs(struct mlx5e_flow_steering *fs,
+			  const struct mlx5e_profile *profile);
+int mlx5e_ptp_rx_manage_fs(struct mlx5e_priv *priv, bool set);
+
+enum {
+	MLX5E_SKB_CB_CQE_HWTSTAMP  = BIT(0),
+	MLX5E_SKB_CB_PORT_HWTSTAMP = BIT(1),
+};
+
+void mlx5e_skb_cb_hwtstamp_handler(struct sk_buff *skb, int hwtstamp_type,
+				   ktime_t hwtstamp,
+				   struct mlx5e_ptp_cq_stats *cq_stats);
+
+void mlx5e_skb_cb_hwtstamp_init(struct sk_buff *skb);
+#endif /* __MLX5_EN_PTP_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/qos.c b/drivers/net/ethernet/mellanox/mlx5/core/en/qos.c
new file mode 100644
index 000000000..2842195ee
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/qos.c
@@ -0,0 +1,518 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2020, Mellanox Technologies inc. All rights reserved. */
+#include <net/sch_generic.h>
+
+#include <net/pkt_cls.h>
+#include "en.h"
+#include "params.h"
+#include "../qos.h"
+#include "en/htb.h"
+
+struct qos_sq_callback_params {
+	struct mlx5e_priv *priv;
+	struct mlx5e_channels *chs;
+};
+
+int mlx5e_qos_bytes_rate_check(struct mlx5_core_dev *mdev, u64 nbytes)
+{
+	if (nbytes < BYTES_IN_MBIT) {
+		qos_warn(mdev, "Input rate (%llu Bytes/sec) below minimum supported (%u Bytes/sec)\n",
+			 nbytes, BYTES_IN_MBIT);
+		return -EINVAL;
+	}
+	return 0;
+}
+
+static u32 mlx5e_qos_bytes2mbits(struct mlx5_core_dev *mdev, u64 nbytes)
+{
+	return div_u64(nbytes, BYTES_IN_MBIT);
+}
+
+int mlx5e_qos_max_leaf_nodes(struct mlx5_core_dev *mdev)
+{
+	return min(MLX5E_QOS_MAX_LEAF_NODES, mlx5_qos_max_leaf_nodes(mdev));
+}
+
+/* TX datapath API */
+
+u16 mlx5e_qid_from_qos(struct mlx5e_channels *chs, u16 qid)
+{
+	/* These channel params are safe to access from the datapath, because:
+	 * 1. This function is called only after checking selq->htb_maj_id != 0,
+	 *    and the number of queues can't change while HTB offload is active.
+	 * 2. When selq->htb_maj_id becomes 0, synchronize_rcu waits for
+	 *    mlx5e_select_queue to finish while holding priv->state_lock,
+	 *    preventing other code from changing the number of queues.
+	 */
+	bool is_ptp = MLX5E_GET_PFLAG(&chs->params, MLX5E_PFLAG_TX_PORT_TS);
+
+	return (chs->params.num_channels + is_ptp) * mlx5e_get_dcb_num_tc(&chs->params) + qid;
+}
+
+/* SQ lifecycle */
+
+static struct mlx5e_txqsq *mlx5e_get_qos_sq(struct mlx5e_priv *priv, int qid)
+{
+	struct mlx5e_params *params = &priv->channels.params;
+	struct mlx5e_txqsq __rcu **qos_sqs;
+	struct mlx5e_channel *c;
+	int ix;
+
+	ix = qid % params->num_channels;
+	qid /= params->num_channels;
+	c = priv->channels.c[ix];
+
+	qos_sqs = mlx5e_state_dereference(priv, c->qos_sqs);
+	return mlx5e_state_dereference(priv, qos_sqs[qid]);
+}
+
+int mlx5e_open_qos_sq(struct mlx5e_priv *priv, struct mlx5e_channels *chs,
+		      u16 node_qid, u32 hw_id)
+{
+	struct mlx5e_create_cq_param ccp = {};
+	struct mlx5e_txqsq __rcu **qos_sqs;
+	struct mlx5e_sq_param param_sq;
+	struct mlx5e_cq_param param_cq;
+	int txq_ix, ix, qid, err = 0;
+	struct mlx5e_params *params;
+	struct mlx5e_channel *c;
+	struct mlx5e_txqsq *sq;
+
+	params = &chs->params;
+
+	txq_ix = mlx5e_qid_from_qos(chs, node_qid);
+
+	WARN_ON(node_qid > priv->htb_max_qos_sqs);
+	if (node_qid == priv->htb_max_qos_sqs) {
+		struct mlx5e_sq_stats *stats, **stats_list = NULL;
+
+		if (priv->htb_max_qos_sqs == 0) {
+			stats_list = kvcalloc(mlx5e_qos_max_leaf_nodes(priv->mdev),
+					      sizeof(*stats_list),
+					      GFP_KERNEL);
+			if (!stats_list)
+				return -ENOMEM;
+		}
+		stats = kzalloc(sizeof(*stats), GFP_KERNEL);
+		if (!stats) {
+			kvfree(stats_list);
+			return -ENOMEM;
+		}
+		if (stats_list)
+			WRITE_ONCE(priv->htb_qos_sq_stats, stats_list);
+		WRITE_ONCE(priv->htb_qos_sq_stats[node_qid], stats);
+		/* Order htb_max_qos_sqs increment after writing the array pointer.
+		 * Pairs with smp_load_acquire in en_stats.c.
+		 */
+		smp_store_release(&priv->htb_max_qos_sqs, priv->htb_max_qos_sqs + 1);
+	}
+
+	ix = node_qid % params->num_channels;
+	qid = node_qid / params->num_channels;
+	c = chs->c[ix];
+
+	qos_sqs = mlx5e_state_dereference(priv, c->qos_sqs);
+	sq = kzalloc(sizeof(*sq), GFP_KERNEL);
+
+	if (!sq)
+		return -ENOMEM;
+
+	mlx5e_build_create_cq_param(&ccp, c);
+
+	memset(&param_sq, 0, sizeof(param_sq));
+	memset(&param_cq, 0, sizeof(param_cq));
+	mlx5e_build_sq_param(priv->mdev, params, &param_sq);
+	mlx5e_build_tx_cq_param(priv->mdev, params, &param_cq);
+	err = mlx5e_open_cq(priv, params->tx_cq_moderation, &param_cq, &ccp, &sq->cq);
+	if (err)
+		goto err_free_sq;
+	err = mlx5e_open_txqsq(c, priv->tisn[c->lag_port][0], txq_ix, params,
+			       &param_sq, sq, 0, hw_id,
+			       priv->htb_qos_sq_stats[node_qid]);
+	if (err)
+		goto err_close_cq;
+
+	rcu_assign_pointer(qos_sqs[qid], sq);
+
+	return 0;
+
+err_close_cq:
+	mlx5e_close_cq(&sq->cq);
+err_free_sq:
+	kfree(sq);
+	return err;
+}
+
+static int mlx5e_open_qos_sq_cb_wrapper(void *data, u16 node_qid, u32 hw_id)
+{
+	struct qos_sq_callback_params *cb_params = data;
+
+	return mlx5e_open_qos_sq(cb_params->priv, cb_params->chs, node_qid, hw_id);
+}
+
+int mlx5e_activate_qos_sq(void *data, u16 node_qid, u32 hw_id)
+{
+	struct mlx5e_priv *priv = data;
+	struct mlx5e_txqsq *sq;
+	u16 qid;
+
+	sq = mlx5e_get_qos_sq(priv, node_qid);
+
+	qid = mlx5e_qid_from_qos(&priv->channels, node_qid);
+
+	/* If it's a new queue, it will be marked as started at this point.
+	 * Stop it before updating txq2sq.
+	 */
+	mlx5e_tx_disable_queue(netdev_get_tx_queue(priv->netdev, qid));
+
+	priv->txq2sq[qid] = sq;
+
+	/* Make the change to txq2sq visible before the queue is started.
+	 * As mlx5e_xmit runs under a spinlock, there is an implicit ACQUIRE,
+	 * which pairs with this barrier.
+	 */
+	smp_wmb();
+
+	qos_dbg(priv->mdev, "Activate QoS SQ qid %u\n", node_qid);
+	mlx5e_activate_txqsq(sq);
+
+	return 0;
+}
+
+void mlx5e_deactivate_qos_sq(struct mlx5e_priv *priv, u16 qid)
+{
+	struct mlx5e_txqsq *sq;
+
+	sq = mlx5e_get_qos_sq(priv, qid);
+	if (!sq) /* Handle the case when the SQ failed to open. */
+		return;
+
+	qos_dbg(priv->mdev, "Deactivate QoS SQ qid %u\n", qid);
+	mlx5e_deactivate_txqsq(sq);
+
+	priv->txq2sq[mlx5e_qid_from_qos(&priv->channels, qid)] = NULL;
+
+	/* Make the change to txq2sq visible before the queue is started again.
+	 * As mlx5e_xmit runs under a spinlock, there is an implicit ACQUIRE,
+	 * which pairs with this barrier.
+	 */
+	smp_wmb();
+}
+
+void mlx5e_close_qos_sq(struct mlx5e_priv *priv, u16 qid)
+{
+	struct mlx5e_txqsq __rcu **qos_sqs;
+	struct mlx5e_params *params;
+	struct mlx5e_channel *c;
+	struct mlx5e_txqsq *sq;
+	int ix;
+
+	params = &priv->channels.params;
+
+	ix = qid % params->num_channels;
+	qid /= params->num_channels;
+	c = priv->channels.c[ix];
+	qos_sqs = mlx5e_state_dereference(priv, c->qos_sqs);
+	sq = rcu_replace_pointer(qos_sqs[qid], NULL, lockdep_is_held(&priv->state_lock));
+	if (!sq) /* Handle the case when the SQ failed to open. */
+		return;
+
+	synchronize_rcu(); /* Sync with NAPI. */
+
+	mlx5e_close_txqsq(sq);
+	mlx5e_close_cq(&sq->cq);
+	kfree(sq);
+}
+
+void mlx5e_qos_close_queues(struct mlx5e_channel *c)
+{
+	struct mlx5e_txqsq __rcu **qos_sqs;
+	int i;
+
+	qos_sqs = rcu_replace_pointer(c->qos_sqs, NULL, lockdep_is_held(&c->priv->state_lock));
+	if (!qos_sqs)
+		return;
+	synchronize_rcu(); /* Sync with NAPI. */
+
+	for (i = 0; i < c->qos_sqs_size; i++) {
+		struct mlx5e_txqsq *sq;
+
+		sq = mlx5e_state_dereference(c->priv, qos_sqs[i]);
+		if (!sq) /* Handle the case when the SQ failed to open. */
+			continue;
+
+		mlx5e_close_txqsq(sq);
+		mlx5e_close_cq(&sq->cq);
+		kfree(sq);
+	}
+
+	kvfree(qos_sqs);
+}
+
+void mlx5e_qos_close_all_queues(struct mlx5e_channels *chs)
+{
+	int i;
+
+	for (i = 0; i < chs->num; i++)
+		mlx5e_qos_close_queues(chs->c[i]);
+}
+
+int mlx5e_qos_alloc_queues(struct mlx5e_priv *priv, struct mlx5e_channels *chs)
+{
+	u16 qos_sqs_size;
+	int i;
+
+	qos_sqs_size = DIV_ROUND_UP(mlx5e_qos_max_leaf_nodes(priv->mdev), chs->num);
+
+	for (i = 0; i < chs->num; i++) {
+		struct mlx5e_txqsq **sqs;
+
+		sqs = kvcalloc(qos_sqs_size, sizeof(struct mlx5e_txqsq *), GFP_KERNEL);
+		if (!sqs)
+			goto err_free;
+
+		WRITE_ONCE(chs->c[i]->qos_sqs_size, qos_sqs_size);
+		smp_wmb(); /* Pairs with mlx5e_napi_poll. */
+		rcu_assign_pointer(chs->c[i]->qos_sqs, sqs);
+	}
+
+	return 0;
+
+err_free:
+	while (--i >= 0) {
+		struct mlx5e_txqsq **sqs;
+
+		sqs = rcu_replace_pointer(chs->c[i]->qos_sqs, NULL,
+					  lockdep_is_held(&priv->state_lock));
+
+		synchronize_rcu(); /* Sync with NAPI. */
+		kvfree(sqs);
+	}
+	return -ENOMEM;
+}
+
+int mlx5e_qos_open_queues(struct mlx5e_priv *priv, struct mlx5e_channels *chs)
+{
+	struct qos_sq_callback_params callback_params;
+	int err;
+
+	err = mlx5e_qos_alloc_queues(priv, chs);
+	if (err)
+		return err;
+
+	callback_params.priv = priv;
+	callback_params.chs = chs;
+
+	err = mlx5e_htb_enumerate_leaves(priv->htb, mlx5e_open_qos_sq_cb_wrapper, &callback_params);
+	if (err) {
+		mlx5e_qos_close_all_queues(chs);
+		return err;
+	}
+
+	return 0;
+}
+
+void mlx5e_qos_activate_queues(struct mlx5e_priv *priv)
+{
+	mlx5e_htb_enumerate_leaves(priv->htb, mlx5e_activate_qos_sq, priv);
+}
+
+void mlx5e_qos_deactivate_queues(struct mlx5e_channel *c)
+{
+	struct mlx5e_params *params = &c->priv->channels.params;
+	struct mlx5e_txqsq __rcu **qos_sqs;
+	int i;
+
+	qos_sqs = mlx5e_state_dereference(c->priv, c->qos_sqs);
+	if (!qos_sqs)
+		return;
+
+	for (i = 0; i < c->qos_sqs_size; i++) {
+		u16 qid = params->num_channels * i + c->ix;
+		struct mlx5e_txqsq *sq;
+
+		sq = mlx5e_state_dereference(c->priv, qos_sqs[i]);
+		if (!sq) /* Handle the case when the SQ failed to open. */
+			continue;
+
+		qos_dbg(c->mdev, "Deactivate QoS SQ qid %u\n", qid);
+		mlx5e_deactivate_txqsq(sq);
+
+		/* The queue is disabled, no synchronization with datapath is needed. */
+		c->priv->txq2sq[mlx5e_qid_from_qos(&c->priv->channels, qid)] = NULL;
+	}
+}
+
+void mlx5e_qos_deactivate_all_queues(struct mlx5e_channels *chs)
+{
+	int i;
+
+	for (i = 0; i < chs->num; i++)
+		mlx5e_qos_deactivate_queues(chs->c[i]);
+}
+
+void mlx5e_reactivate_qos_sq(struct mlx5e_priv *priv, u16 qid, struct netdev_queue *txq)
+{
+	qos_dbg(priv->mdev, "Reactivate QoS SQ qid %u\n", qid);
+	netdev_tx_reset_queue(txq);
+	netif_tx_start_queue(txq);
+}
+
+void mlx5e_reset_qdisc(struct net_device *dev, u16 qid)
+{
+	struct netdev_queue *dev_queue = netdev_get_tx_queue(dev, qid);
+	struct Qdisc *qdisc = dev_queue->qdisc_sleeping;
+
+	if (!qdisc)
+		return;
+
+	spin_lock_bh(qdisc_lock(qdisc));
+	qdisc_reset(qdisc);
+	spin_unlock_bh(qdisc_lock(qdisc));
+}
+
+int mlx5e_htb_setup_tc(struct mlx5e_priv *priv, struct tc_htb_qopt_offload *htb_qopt)
+{
+	struct mlx5e_htb *htb = priv->htb;
+	int res;
+
+	if (!htb && htb_qopt->command != TC_HTB_CREATE)
+		return -EINVAL;
+
+	switch (htb_qopt->command) {
+	case TC_HTB_CREATE:
+		if (!mlx5_qos_is_supported(priv->mdev)) {
+			NL_SET_ERR_MSG_MOD(htb_qopt->extack,
+					   "Missing QoS capabilities. Try disabling SRIOV or use a supported device.");
+			return -EOPNOTSUPP;
+		}
+		priv->htb = mlx5e_htb_alloc();
+		htb = priv->htb;
+		if (!htb)
+			return -ENOMEM;
+		res = mlx5e_htb_init(htb, htb_qopt, priv->netdev, priv->mdev, &priv->selq, priv);
+		if (res) {
+			mlx5e_htb_free(htb);
+			priv->htb = NULL;
+		}
+		return res;
+	case TC_HTB_DESTROY:
+		mlx5e_htb_cleanup(htb);
+		mlx5e_htb_free(htb);
+		priv->htb = NULL;
+		return 0;
+	case TC_HTB_LEAF_ALLOC_QUEUE:
+		res = mlx5e_htb_leaf_alloc_queue(htb, htb_qopt->classid, htb_qopt->parent_classid,
+						 htb_qopt->rate, htb_qopt->ceil, htb_qopt->extack);
+		if (res < 0)
+			return res;
+		htb_qopt->qid = res;
+		return 0;
+	case TC_HTB_LEAF_TO_INNER:
+		return mlx5e_htb_leaf_to_inner(htb, htb_qopt->parent_classid, htb_qopt->classid,
+					       htb_qopt->rate, htb_qopt->ceil, htb_qopt->extack);
+	case TC_HTB_LEAF_DEL:
+		return mlx5e_htb_leaf_del(htb, &htb_qopt->classid, htb_qopt->extack);
+	case TC_HTB_LEAF_DEL_LAST:
+	case TC_HTB_LEAF_DEL_LAST_FORCE:
+		return mlx5e_htb_leaf_del_last(htb, htb_qopt->classid,
+					       htb_qopt->command == TC_HTB_LEAF_DEL_LAST_FORCE,
+					       htb_qopt->extack);
+	case TC_HTB_NODE_MODIFY:
+		return mlx5e_htb_node_modify(htb, htb_qopt->classid, htb_qopt->rate, htb_qopt->ceil,
+					     htb_qopt->extack);
+	case TC_HTB_LEAF_QUERY_QUEUE:
+		res = mlx5e_htb_get_txq_by_classid(htb, htb_qopt->classid);
+		if (res < 0)
+			return res;
+		htb_qopt->qid = res;
+		return 0;
+	default:
+		return -EOPNOTSUPP;
+	}
+}
+
+struct mlx5e_mqprio_rl {
+	struct mlx5_core_dev *mdev;
+	u32 root_id;
+	u32 *leaves_id;
+	u8 num_tc;
+};
+
+struct mlx5e_mqprio_rl *mlx5e_mqprio_rl_alloc(void)
+{
+	return kvzalloc(sizeof(struct mlx5e_mqprio_rl), GFP_KERNEL);
+}
+
+void mlx5e_mqprio_rl_free(struct mlx5e_mqprio_rl *rl)
+{
+	kvfree(rl);
+}
+
+int mlx5e_mqprio_rl_init(struct mlx5e_mqprio_rl *rl, struct mlx5_core_dev *mdev, u8 num_tc,
+			 u64 max_rate[])
+{
+	int err;
+	int tc;
+
+	if (!mlx5_qos_is_supported(mdev)) {
+		qos_warn(mdev, "Missing QoS capabilities. Try disabling SRIOV or use a supported device.");
+		return -EOPNOTSUPP;
+	}
+	if (num_tc > mlx5e_qos_max_leaf_nodes(mdev))
+		return -EINVAL;
+
+	rl->mdev = mdev;
+	rl->num_tc = num_tc;
+	rl->leaves_id = kvcalloc(num_tc, sizeof(*rl->leaves_id), GFP_KERNEL);
+	if (!rl->leaves_id)
+		return -ENOMEM;
+
+	err = mlx5_qos_create_root_node(mdev, &rl->root_id);
+	if (err)
+		goto err_free_leaves;
+
+	qos_dbg(mdev, "Root created, id %#x\n", rl->root_id);
+
+	for (tc = 0; tc < num_tc; tc++) {
+		u32 max_average_bw;
+
+		max_average_bw = mlx5e_qos_bytes2mbits(mdev, max_rate[tc]);
+		err = mlx5_qos_create_leaf_node(mdev, rl->root_id, 0, max_average_bw,
+						&rl->leaves_id[tc]);
+		if (err)
+			goto err_destroy_leaves;
+
+		qos_dbg(mdev, "Leaf[%d] created, id %#x, max average bw %u Mbits/sec\n",
+			tc, rl->leaves_id[tc], max_average_bw);
+	}
+	return 0;
+
+err_destroy_leaves:
+	while (--tc >= 0)
+		mlx5_qos_destroy_node(mdev, rl->leaves_id[tc]);
+	mlx5_qos_destroy_node(mdev, rl->root_id);
+err_free_leaves:
+	kvfree(rl->leaves_id);
+	return err;
+}
+
+void mlx5e_mqprio_rl_cleanup(struct mlx5e_mqprio_rl *rl)
+{
+	int tc;
+
+	for (tc = 0; tc < rl->num_tc; tc++)
+		mlx5_qos_destroy_node(rl->mdev, rl->leaves_id[tc]);
+	mlx5_qos_destroy_node(rl->mdev, rl->root_id);
+	kvfree(rl->leaves_id);
+}
+
+int mlx5e_mqprio_rl_get_node_hw_id(struct mlx5e_mqprio_rl *rl, int tc, u32 *hw_id)
+{
+	if (tc >= rl->num_tc)
+		return -EINVAL;
+
+	*hw_id = rl->leaves_id[tc];
+	return 0;
+}
+
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/qos.h b/drivers/net/ethernet/mellanox/mlx5/core/en/qos.h
new file mode 100644
index 000000000..4947afa23
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/qos.h
@@ -0,0 +1,51 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2020, Mellanox Technologies inc. All rights reserved. */
+
+#ifndef __MLX5E_EN_QOS_H
+#define __MLX5E_EN_QOS_H
+
+#include <linux/mlx5/driver.h>
+
+#define BYTES_IN_MBIT 125000
+
+struct mlx5e_priv;
+struct mlx5e_htb;
+struct mlx5e_channels;
+struct mlx5e_channel;
+struct tc_htb_qopt_offload;
+
+int mlx5e_qos_bytes_rate_check(struct mlx5_core_dev *mdev, u64 nbytes);
+int mlx5e_qos_max_leaf_nodes(struct mlx5_core_dev *mdev);
+
+/* SQ lifecycle */
+int mlx5e_open_qos_sq(struct mlx5e_priv *priv, struct mlx5e_channels *chs,
+		      u16 node_qid, u32 hw_id);
+int mlx5e_activate_qos_sq(void *data, u16 node_qid, u32 hw_id);
+void mlx5e_deactivate_qos_sq(struct mlx5e_priv *priv, u16 qid);
+void mlx5e_close_qos_sq(struct mlx5e_priv *priv, u16 qid);
+void mlx5e_reactivate_qos_sq(struct mlx5e_priv *priv, u16 qid, struct netdev_queue *txq);
+void mlx5e_reset_qdisc(struct net_device *dev, u16 qid);
+
+int mlx5e_qos_open_queues(struct mlx5e_priv *priv, struct mlx5e_channels *chs);
+void mlx5e_qos_activate_queues(struct mlx5e_priv *priv);
+void mlx5e_qos_deactivate_queues(struct mlx5e_channel *c);
+void mlx5e_qos_deactivate_all_queues(struct mlx5e_channels *chs);
+void mlx5e_qos_close_queues(struct mlx5e_channel *c);
+void mlx5e_qos_close_all_queues(struct mlx5e_channels *chs);
+int mlx5e_qos_alloc_queues(struct mlx5e_priv *priv, struct mlx5e_channels *chs);
+
+/* TX datapath API */
+u16 mlx5e_qid_from_qos(struct mlx5e_channels *chs, u16 qid);
+
+/* HTB API */
+int mlx5e_htb_setup_tc(struct mlx5e_priv *priv, struct tc_htb_qopt_offload *htb);
+
+/* MQPRIO TX rate limit */
+struct mlx5e_mqprio_rl;
+struct mlx5e_mqprio_rl *mlx5e_mqprio_rl_alloc(void);
+void mlx5e_mqprio_rl_free(struct mlx5e_mqprio_rl *rl);
+int mlx5e_mqprio_rl_init(struct mlx5e_mqprio_rl *rl, struct mlx5_core_dev *mdev, u8 num_tc,
+			 u64 max_rate[]);
+void mlx5e_mqprio_rl_cleanup(struct mlx5e_mqprio_rl *rl);
+int mlx5e_mqprio_rl_get_node_hw_id(struct mlx5e_mqprio_rl *rl, int tc, u32 *hw_id);
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/rep/bond.c b/drivers/net/ethernet/mellanox/mlx5/core/en/rep/bond.c
new file mode 100644
index 000000000..b6f5c1bcd
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/rep/bond.c
@@ -0,0 +1,351 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2020 Mellanox Technologies Inc. All rights reserved. */
+
+#include <linux/netdevice.h>
+#include <linux/list.h>
+#include <net/lag.h>
+
+#include "mlx5_core.h"
+#include "eswitch.h"
+#include "esw/acl/ofld.h"
+#include "en_rep.h"
+
+struct mlx5e_rep_bond {
+	struct notifier_block nb;
+	struct netdev_net_notifier nn;
+	struct list_head metadata_list;
+};
+
+struct mlx5e_rep_bond_slave_entry {
+	struct list_head list;
+	struct net_device *netdev;
+};
+
+struct mlx5e_rep_bond_metadata {
+	struct list_head list; /* link to global list of rep_bond_metadata */
+	struct mlx5_eswitch *esw;
+	 /* private of uplink holding rep bond metadata list */
+	struct net_device *lag_dev;
+	u32 metadata_reg_c_0;
+
+	struct list_head slaves_list; /* slaves list */
+	int slaves;
+};
+
+static struct mlx5e_rep_bond_metadata *
+mlx5e_lookup_rep_bond_metadata(struct mlx5_rep_uplink_priv *uplink_priv,
+			       const struct net_device *lag_dev)
+{
+	struct mlx5e_rep_bond_metadata *found = NULL;
+	struct mlx5e_rep_bond_metadata *cur;
+
+	list_for_each_entry(cur, &uplink_priv->bond->metadata_list, list) {
+		if (cur->lag_dev == lag_dev) {
+			found = cur;
+			break;
+		}
+	}
+
+	return found;
+}
+
+static struct mlx5e_rep_bond_slave_entry *
+mlx5e_lookup_rep_bond_slave_entry(struct mlx5e_rep_bond_metadata *mdata,
+				  const struct net_device *netdev)
+{
+	struct mlx5e_rep_bond_slave_entry *found = NULL;
+	struct mlx5e_rep_bond_slave_entry *cur;
+
+	list_for_each_entry(cur, &mdata->slaves_list, list) {
+		if (cur->netdev == netdev) {
+			found = cur;
+			break;
+		}
+	}
+
+	return found;
+}
+
+static void mlx5e_rep_bond_metadata_release(struct mlx5e_rep_bond_metadata *mdata)
+{
+	netdev_dbg(mdata->lag_dev, "destroy rep_bond_metadata(%d)\n",
+		   mdata->metadata_reg_c_0);
+	list_del(&mdata->list);
+	mlx5_esw_match_metadata_free(mdata->esw, mdata->metadata_reg_c_0);
+	WARN_ON(!list_empty(&mdata->slaves_list));
+	kfree(mdata);
+}
+
+/* This must be called under rtnl_lock */
+int mlx5e_rep_bond_enslave(struct mlx5_eswitch *esw, struct net_device *netdev,
+			   struct net_device *lag_dev)
+{
+	struct mlx5e_rep_bond_slave_entry *s_entry;
+	struct mlx5e_rep_bond_metadata *mdata;
+	struct mlx5e_rep_priv *rpriv;
+	struct mlx5e_priv *priv;
+	int err;
+
+	ASSERT_RTNL();
+
+	rpriv = mlx5_eswitch_get_uplink_priv(esw, REP_ETH);
+	mdata = mlx5e_lookup_rep_bond_metadata(&rpriv->uplink_priv, lag_dev);
+	if (!mdata) {
+		/* First netdev becomes slave, no metadata presents the lag_dev. Create one */
+		mdata = kzalloc(sizeof(*mdata), GFP_KERNEL);
+		if (!mdata)
+			return -ENOMEM;
+
+		mdata->lag_dev = lag_dev;
+		mdata->esw = esw;
+		INIT_LIST_HEAD(&mdata->slaves_list);
+		mdata->metadata_reg_c_0 = mlx5_esw_match_metadata_alloc(esw);
+		if (!mdata->metadata_reg_c_0) {
+			kfree(mdata);
+			return -ENOSPC;
+		}
+		list_add(&mdata->list, &rpriv->uplink_priv.bond->metadata_list);
+
+		netdev_dbg(lag_dev, "create rep_bond_metadata(%d)\n",
+			   mdata->metadata_reg_c_0);
+	}
+
+	s_entry = kzalloc(sizeof(*s_entry), GFP_KERNEL);
+	if (!s_entry) {
+		err = -ENOMEM;
+		goto entry_alloc_err;
+	}
+
+	s_entry->netdev = netdev;
+	priv = netdev_priv(netdev);
+	rpriv = priv->ppriv;
+
+	err = mlx5_esw_acl_ingress_vport_bond_update(esw, rpriv->rep->vport,
+						     mdata->metadata_reg_c_0);
+	if (err)
+		goto ingress_err;
+
+	mdata->slaves++;
+	list_add_tail(&s_entry->list, &mdata->slaves_list);
+	netdev_dbg(netdev, "enslave rep vport(%d) lag_dev(%s) metadata(0x%x)\n",
+		   rpriv->rep->vport, lag_dev->name, mdata->metadata_reg_c_0);
+
+	return 0;
+
+ingress_err:
+	kfree(s_entry);
+entry_alloc_err:
+	if (!mdata->slaves)
+		mlx5e_rep_bond_metadata_release(mdata);
+	return err;
+}
+
+/* This must be called under rtnl_lock */
+void mlx5e_rep_bond_unslave(struct mlx5_eswitch *esw,
+			    const struct net_device *netdev,
+			    const struct net_device *lag_dev)
+{
+	struct mlx5e_rep_bond_slave_entry *s_entry;
+	struct mlx5e_rep_bond_metadata *mdata;
+	struct mlx5e_rep_priv *rpriv;
+	struct mlx5e_priv *priv;
+
+	ASSERT_RTNL();
+
+	rpriv = mlx5_eswitch_get_uplink_priv(esw, REP_ETH);
+	mdata = mlx5e_lookup_rep_bond_metadata(&rpriv->uplink_priv, lag_dev);
+	if (!mdata)
+		return;
+
+	s_entry = mlx5e_lookup_rep_bond_slave_entry(mdata, netdev);
+	if (!s_entry)
+		return;
+
+	priv = netdev_priv(netdev);
+	rpriv = priv->ppriv;
+
+	/* Reset bond_metadata to zero first then reset all ingress/egress
+	 * acls and rx rules of unslave representor's vport
+	 */
+	mlx5_esw_acl_ingress_vport_bond_update(esw, rpriv->rep->vport, 0);
+	mlx5_esw_acl_egress_vport_unbond(esw, rpriv->rep->vport);
+	mlx5e_rep_bond_update(priv, false);
+
+	list_del(&s_entry->list);
+
+	netdev_dbg(netdev, "unslave rep vport(%d) lag_dev(%s) metadata(0x%x)\n",
+		   rpriv->rep->vport, lag_dev->name, mdata->metadata_reg_c_0);
+
+	if (--mdata->slaves == 0)
+		mlx5e_rep_bond_metadata_release(mdata);
+	kfree(s_entry);
+}
+
+static bool mlx5e_rep_is_lag_netdev(struct net_device *netdev)
+{
+	return netif_is_lag_port(netdev) && mlx5e_eswitch_vf_rep(netdev);
+}
+
+static void mlx5e_rep_changelowerstate_event(struct net_device *netdev, void *ptr)
+{
+	struct netdev_notifier_changelowerstate_info *info;
+	struct netdev_lag_lower_state_info *lag_info;
+	struct mlx5e_rep_priv *rpriv;
+	struct net_device *lag_dev;
+	struct mlx5e_priv *priv;
+	struct list_head *iter;
+	struct net_device *dev;
+	u16 acl_vport_num;
+	u16 fwd_vport_num;
+	int err;
+
+	info = ptr;
+	lag_info = info->lower_state_info;
+	/* This is not an event of a representor becoming active slave */
+	if (!lag_info->tx_enabled)
+		return;
+
+	priv = netdev_priv(netdev);
+	rpriv = priv->ppriv;
+	fwd_vport_num = rpriv->rep->vport;
+	lag_dev = netdev_master_upper_dev_get(netdev);
+	if (!lag_dev)
+		return;
+
+	netdev_dbg(netdev, "lag_dev(%s)'s slave vport(%d) is txable(%d)\n",
+		   lag_dev->name, fwd_vport_num, net_lag_port_dev_txable(netdev));
+
+	/* Point everyone's egress acl to the vport of the active representor */
+	netdev_for_each_lower_dev(lag_dev, dev, iter) {
+		priv = netdev_priv(dev);
+		rpriv = priv->ppriv;
+		acl_vport_num = rpriv->rep->vport;
+		if (acl_vport_num != fwd_vport_num) {
+			/* Only single rx_rule for unique bond_metadata should be
+			 * present, delete it if it's saved as passive vport's
+			 * rx_rule with destination as passive vport's root_ft
+			 */
+			mlx5e_rep_bond_update(priv, true);
+			err = mlx5_esw_acl_egress_vport_bond(priv->mdev->priv.eswitch,
+							     fwd_vport_num,
+							     acl_vport_num);
+			if (err)
+				netdev_warn(dev,
+					    "configure slave vport(%d) egress fwd, err(%d)",
+					    acl_vport_num, err);
+		}
+	}
+
+	/* Insert new rx_rule for unique bond_metadata, save it as active vport's
+	 * rx_rule with new destination as active vport's root_ft
+	 */
+	err = mlx5e_rep_bond_update(netdev_priv(netdev), false);
+	if (err)
+		netdev_warn(netdev, "configure active slave vport(%d) rx_rule, err(%d)",
+			    fwd_vport_num, err);
+}
+
+static void mlx5e_rep_changeupper_event(struct net_device *netdev, void *ptr)
+{
+	struct netdev_notifier_changeupper_info *info = ptr;
+	struct mlx5e_rep_priv *rpriv;
+	struct net_device *lag_dev;
+	struct mlx5e_priv *priv;
+
+	priv = netdev_priv(netdev);
+	rpriv = priv->ppriv;
+	lag_dev = info->upper_dev;
+
+	netdev_dbg(netdev, "%sslave vport(%d) lag(%s)\n",
+		   info->linking ? "en" : "un", rpriv->rep->vport, lag_dev->name);
+
+	if (info->linking)
+		mlx5e_rep_bond_enslave(priv->mdev->priv.eswitch, netdev, lag_dev);
+	else
+		mlx5e_rep_bond_unslave(priv->mdev->priv.eswitch, netdev, lag_dev);
+}
+
+/* Bond device of representors and netdev events are used here in specific way
+ * to support eswitch vports bonding and to perform failover of eswitch vport
+ * by modifying the vport's egress acl of lower dev representors. Thus this
+ * also change the traditional behavior of lower dev under bond device.
+ * All non-representor netdevs or representors of other vendors as lower dev
+ * of bond device are not supported.
+ */
+static int mlx5e_rep_esw_bond_netevent(struct notifier_block *nb,
+				       unsigned long event, void *ptr)
+{
+	struct net_device *netdev = netdev_notifier_info_to_dev(ptr);
+	struct mlx5e_rep_priv *rpriv;
+	struct mlx5e_rep_bond *bond;
+	struct mlx5e_priv *priv;
+
+	if (!mlx5e_rep_is_lag_netdev(netdev))
+		return NOTIFY_DONE;
+
+	bond = container_of(nb, struct mlx5e_rep_bond, nb);
+	priv = netdev_priv(netdev);
+	rpriv = mlx5_eswitch_get_uplink_priv(priv->mdev->priv.eswitch, REP_ETH);
+	/* Verify VF representor is on the same device of the bond handling the netevent. */
+	if (rpriv->uplink_priv.bond != bond)
+		return NOTIFY_DONE;
+
+	switch (event) {
+	case NETDEV_CHANGELOWERSTATE:
+		mlx5e_rep_changelowerstate_event(netdev, ptr);
+		break;
+	case NETDEV_CHANGEUPPER:
+		mlx5e_rep_changeupper_event(netdev, ptr);
+		break;
+	}
+	return NOTIFY_DONE;
+}
+
+/* If HW support eswitch vports bonding, register a specific notifier to
+ * handle it when two or more representors are bonded
+ */
+int mlx5e_rep_bond_init(struct mlx5e_rep_priv *rpriv)
+{
+	struct mlx5_rep_uplink_priv *uplink_priv = &rpriv->uplink_priv;
+	struct net_device *netdev = rpriv->netdev;
+	struct mlx5e_priv *priv;
+	int ret = 0;
+
+	priv = netdev_priv(netdev);
+	if (!mlx5_esw_acl_egress_fwd2vport_supported(priv->mdev->priv.eswitch))
+		goto out;
+
+	uplink_priv->bond = kvzalloc(sizeof(*uplink_priv->bond), GFP_KERNEL);
+	if (!uplink_priv->bond) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	INIT_LIST_HEAD(&uplink_priv->bond->metadata_list);
+	uplink_priv->bond->nb.notifier_call = mlx5e_rep_esw_bond_netevent;
+	ret = register_netdevice_notifier_dev_net(netdev,
+						  &uplink_priv->bond->nb,
+						  &uplink_priv->bond->nn);
+	if (ret) {
+		netdev_err(netdev, "register bonding netevent notifier, err(%d)\n", ret);
+		kvfree(uplink_priv->bond);
+		uplink_priv->bond = NULL;
+	}
+
+out:
+	return ret;
+}
+
+void mlx5e_rep_bond_cleanup(struct mlx5e_rep_priv *rpriv)
+{
+	struct mlx5e_priv *priv = netdev_priv(rpriv->netdev);
+
+	if (!mlx5_esw_acl_egress_fwd2vport_supported(priv->mdev->priv.eswitch) ||
+	    !rpriv->uplink_priv.bond)
+		return;
+
+	unregister_netdevice_notifier_dev_net(rpriv->netdev,
+					      &rpriv->uplink_priv.bond->nb,
+					      &rpriv->uplink_priv.bond->nn);
+	kvfree(rpriv->uplink_priv.bond);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/rep/bridge.c b/drivers/net/ethernet/mellanox/mlx5/core/en/rep/bridge.c
new file mode 100644
index 000000000..ce85b48d3
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/rep/bridge.c
@@ -0,0 +1,569 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2021 Mellanox Technologies. */
+
+#include <linux/netdevice.h>
+#include <linux/if_bridge.h>
+#include <net/netevent.h>
+#include <net/switchdev.h>
+#include "bridge.h"
+#include "esw/bridge.h"
+#include "en_rep.h"
+
+#define MLX5_ESW_BRIDGE_UPDATE_INTERVAL 1000
+
+struct mlx5_bridge_switchdev_fdb_work {
+	struct work_struct work;
+	struct switchdev_notifier_fdb_info fdb_info;
+	struct net_device *dev;
+	struct mlx5_esw_bridge_offloads *br_offloads;
+	bool add;
+};
+
+static bool mlx5_esw_bridge_dev_same_esw(struct net_device *dev, struct mlx5_eswitch *esw)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+
+	return esw == priv->mdev->priv.eswitch;
+}
+
+static bool mlx5_esw_bridge_dev_same_hw(struct net_device *dev, struct mlx5_eswitch *esw)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	struct mlx5_core_dev *mdev, *esw_mdev;
+	u64 system_guid, esw_system_guid;
+
+	mdev = priv->mdev;
+	esw_mdev = esw->dev;
+
+	system_guid = mlx5_query_nic_system_image_guid(mdev);
+	esw_system_guid = mlx5_query_nic_system_image_guid(esw_mdev);
+
+	return system_guid == esw_system_guid;
+}
+
+static struct net_device *
+mlx5_esw_bridge_lag_rep_get(struct net_device *dev, struct mlx5_eswitch *esw)
+{
+	struct net_device *lower;
+	struct list_head *iter;
+
+	netdev_for_each_lower_dev(dev, lower, iter) {
+		struct mlx5_core_dev *mdev;
+		struct mlx5e_priv *priv;
+
+		if (!mlx5e_eswitch_rep(lower))
+			continue;
+
+		priv = netdev_priv(lower);
+		mdev = priv->mdev;
+		if (mlx5_lag_is_shared_fdb(mdev) && mlx5_esw_bridge_dev_same_esw(lower, esw))
+			return lower;
+	}
+
+	return NULL;
+}
+
+static struct net_device *
+mlx5_esw_bridge_rep_vport_num_vhca_id_get(struct net_device *dev, struct mlx5_eswitch *esw,
+					  u16 *vport_num, u16 *esw_owner_vhca_id)
+{
+	struct mlx5e_rep_priv *rpriv;
+	struct mlx5e_priv *priv;
+
+	if (netif_is_lag_master(dev))
+		dev = mlx5_esw_bridge_lag_rep_get(dev, esw);
+
+	if (!dev || !mlx5e_eswitch_rep(dev) || !mlx5_esw_bridge_dev_same_hw(dev, esw))
+		return NULL;
+
+	priv = netdev_priv(dev);
+	rpriv = priv->ppriv;
+	*vport_num = rpriv->rep->vport;
+	*esw_owner_vhca_id = MLX5_CAP_GEN(priv->mdev, vhca_id);
+	return dev;
+}
+
+static struct net_device *
+mlx5_esw_bridge_lower_rep_vport_num_vhca_id_get(struct net_device *dev, struct mlx5_eswitch *esw,
+						u16 *vport_num, u16 *esw_owner_vhca_id)
+{
+	struct net_device *lower_dev;
+	struct list_head *iter;
+
+	if (netif_is_lag_master(dev) || mlx5e_eswitch_rep(dev))
+		return mlx5_esw_bridge_rep_vport_num_vhca_id_get(dev, esw, vport_num,
+								 esw_owner_vhca_id);
+
+	netdev_for_each_lower_dev(dev, lower_dev, iter) {
+		struct net_device *rep;
+
+		if (netif_is_bridge_master(lower_dev))
+			continue;
+
+		rep = mlx5_esw_bridge_lower_rep_vport_num_vhca_id_get(lower_dev, esw, vport_num,
+								      esw_owner_vhca_id);
+		if (rep)
+			return rep;
+	}
+
+	return NULL;
+}
+
+static bool mlx5_esw_bridge_is_local(struct net_device *dev, struct net_device *rep,
+				     struct mlx5_eswitch *esw)
+{
+	struct mlx5_core_dev *mdev;
+	struct mlx5e_priv *priv;
+
+	if (!mlx5_esw_bridge_dev_same_esw(rep, esw))
+		return false;
+
+	priv = netdev_priv(rep);
+	mdev = priv->mdev;
+	if (netif_is_lag_master(dev))
+		return mlx5_lag_is_shared_fdb(mdev) && mlx5_lag_is_master(mdev);
+	return true;
+}
+
+static int mlx5_esw_bridge_port_changeupper(struct notifier_block *nb, void *ptr)
+{
+	struct mlx5_esw_bridge_offloads *br_offloads = container_of(nb,
+								    struct mlx5_esw_bridge_offloads,
+								    netdev_nb);
+	struct net_device *dev = netdev_notifier_info_to_dev(ptr);
+	struct netdev_notifier_changeupper_info *info = ptr;
+	struct net_device *upper = info->upper_dev, *rep;
+	struct mlx5_eswitch *esw = br_offloads->esw;
+	u16 vport_num, esw_owner_vhca_id;
+	struct netlink_ext_ack *extack;
+	int ifindex = upper->ifindex;
+	int err = 0;
+
+	if (!netif_is_bridge_master(upper))
+		return 0;
+
+	rep = mlx5_esw_bridge_rep_vport_num_vhca_id_get(dev, esw, &vport_num, &esw_owner_vhca_id);
+	if (!rep)
+		return 0;
+
+	extack = netdev_notifier_info_to_extack(&info->info);
+
+	if (mlx5_esw_bridge_is_local(dev, rep, esw))
+		err = info->linking ?
+			mlx5_esw_bridge_vport_link(ifindex, vport_num, esw_owner_vhca_id,
+						   br_offloads, extack) :
+			mlx5_esw_bridge_vport_unlink(ifindex, vport_num, esw_owner_vhca_id,
+						     br_offloads, extack);
+	else if (mlx5_esw_bridge_dev_same_hw(rep, esw))
+		err = info->linking ?
+			mlx5_esw_bridge_vport_peer_link(ifindex, vport_num, esw_owner_vhca_id,
+							br_offloads, extack) :
+			mlx5_esw_bridge_vport_peer_unlink(ifindex, vport_num, esw_owner_vhca_id,
+							  br_offloads, extack);
+
+	return err;
+}
+
+static int
+mlx5_esw_bridge_changeupper_validate_netdev(void *ptr)
+{
+	struct net_device *dev = netdev_notifier_info_to_dev(ptr);
+	struct netdev_notifier_changeupper_info *info = ptr;
+	struct net_device *upper = info->upper_dev;
+	struct net_device *lower;
+	struct list_head *iter;
+
+	if (!netif_is_bridge_master(upper) || !netif_is_lag_master(dev))
+		return 0;
+
+	netdev_for_each_lower_dev(dev, lower, iter) {
+		struct mlx5_core_dev *mdev;
+		struct mlx5e_priv *priv;
+
+		if (!mlx5e_eswitch_rep(lower))
+			continue;
+
+		priv = netdev_priv(lower);
+		mdev = priv->mdev;
+		if (!mlx5_lag_is_active(mdev))
+			return -EAGAIN;
+		if (!mlx5_lag_is_shared_fdb(mdev))
+			return -EOPNOTSUPP;
+	}
+
+	return 0;
+}
+
+static int mlx5_esw_bridge_switchdev_port_event(struct notifier_block *nb,
+						unsigned long event, void *ptr)
+{
+	int err = 0;
+
+	switch (event) {
+	case NETDEV_PRECHANGEUPPER:
+		err = mlx5_esw_bridge_changeupper_validate_netdev(ptr);
+		break;
+
+	case NETDEV_CHANGEUPPER:
+		err = mlx5_esw_bridge_port_changeupper(nb, ptr);
+		break;
+	}
+
+	return notifier_from_errno(err);
+}
+
+static int
+mlx5_esw_bridge_port_obj_add(struct net_device *dev,
+			     struct switchdev_notifier_port_obj_info *port_obj_info,
+			     struct mlx5_esw_bridge_offloads *br_offloads)
+{
+	struct netlink_ext_ack *extack = switchdev_notifier_info_to_extack(&port_obj_info->info);
+	const struct switchdev_obj *obj = port_obj_info->obj;
+	const struct switchdev_obj_port_vlan *vlan;
+	u16 vport_num, esw_owner_vhca_id;
+	int err;
+
+	if (!mlx5_esw_bridge_rep_vport_num_vhca_id_get(dev, br_offloads->esw, &vport_num,
+						       &esw_owner_vhca_id))
+		return 0;
+
+	port_obj_info->handled = true;
+
+	switch (obj->id) {
+	case SWITCHDEV_OBJ_ID_PORT_VLAN:
+		vlan = SWITCHDEV_OBJ_PORT_VLAN(obj);
+		err = mlx5_esw_bridge_port_vlan_add(vport_num, esw_owner_vhca_id, vlan->vid,
+						    vlan->flags, br_offloads, extack);
+		break;
+	default:
+		return -EOPNOTSUPP;
+	}
+	return err;
+}
+
+static int
+mlx5_esw_bridge_port_obj_del(struct net_device *dev,
+			     struct switchdev_notifier_port_obj_info *port_obj_info,
+			     struct mlx5_esw_bridge_offloads *br_offloads)
+{
+	const struct switchdev_obj *obj = port_obj_info->obj;
+	const struct switchdev_obj_port_vlan *vlan;
+	u16 vport_num, esw_owner_vhca_id;
+
+	if (!mlx5_esw_bridge_rep_vport_num_vhca_id_get(dev, br_offloads->esw, &vport_num,
+						       &esw_owner_vhca_id))
+		return 0;
+
+	port_obj_info->handled = true;
+
+	switch (obj->id) {
+	case SWITCHDEV_OBJ_ID_PORT_VLAN:
+		vlan = SWITCHDEV_OBJ_PORT_VLAN(obj);
+		mlx5_esw_bridge_port_vlan_del(vport_num, esw_owner_vhca_id, vlan->vid, br_offloads);
+		break;
+	default:
+		return -EOPNOTSUPP;
+	}
+	return 0;
+}
+
+static int
+mlx5_esw_bridge_port_obj_attr_set(struct net_device *dev,
+				  struct switchdev_notifier_port_attr_info *port_attr_info,
+				  struct mlx5_esw_bridge_offloads *br_offloads)
+{
+	struct netlink_ext_ack *extack = switchdev_notifier_info_to_extack(&port_attr_info->info);
+	const struct switchdev_attr *attr = port_attr_info->attr;
+	u16 vport_num, esw_owner_vhca_id;
+	int err = 0;
+
+	if (!mlx5_esw_bridge_lower_rep_vport_num_vhca_id_get(dev, br_offloads->esw, &vport_num,
+							     &esw_owner_vhca_id))
+		return 0;
+
+	port_attr_info->handled = true;
+
+	switch (attr->id) {
+	case SWITCHDEV_ATTR_ID_PORT_PRE_BRIDGE_FLAGS:
+		if (attr->u.brport_flags.mask & ~(BR_LEARNING | BR_FLOOD | BR_MCAST_FLOOD)) {
+			NL_SET_ERR_MSG_MOD(extack, "Flag is not supported");
+			err = -EINVAL;
+		}
+		break;
+	case SWITCHDEV_ATTR_ID_PORT_BRIDGE_FLAGS:
+		break;
+	case SWITCHDEV_ATTR_ID_BRIDGE_AGEING_TIME:
+		err = mlx5_esw_bridge_ageing_time_set(vport_num, esw_owner_vhca_id,
+						      attr->u.ageing_time, br_offloads);
+		break;
+	case SWITCHDEV_ATTR_ID_BRIDGE_VLAN_FILTERING:
+		err = mlx5_esw_bridge_vlan_filtering_set(vport_num, esw_owner_vhca_id,
+							 attr->u.vlan_filtering, br_offloads);
+		break;
+	case SWITCHDEV_ATTR_ID_BRIDGE_VLAN_PROTOCOL:
+		err = mlx5_esw_bridge_vlan_proto_set(vport_num,
+						     esw_owner_vhca_id,
+						     attr->u.vlan_protocol,
+						     br_offloads);
+		break;
+	default:
+		err = -EOPNOTSUPP;
+	}
+
+	return err;
+}
+
+static int mlx5_esw_bridge_event_blocking(struct notifier_block *nb,
+					  unsigned long event, void *ptr)
+{
+	struct mlx5_esw_bridge_offloads *br_offloads = container_of(nb,
+								    struct mlx5_esw_bridge_offloads,
+								    nb_blk);
+	struct net_device *dev = switchdev_notifier_info_to_dev(ptr);
+	int err;
+
+	switch (event) {
+	case SWITCHDEV_PORT_OBJ_ADD:
+		err = mlx5_esw_bridge_port_obj_add(dev, ptr, br_offloads);
+		break;
+	case SWITCHDEV_PORT_OBJ_DEL:
+		err = mlx5_esw_bridge_port_obj_del(dev, ptr, br_offloads);
+		break;
+	case SWITCHDEV_PORT_ATTR_SET:
+		err = mlx5_esw_bridge_port_obj_attr_set(dev, ptr, br_offloads);
+		break;
+	default:
+		err = 0;
+	}
+
+	return notifier_from_errno(err);
+}
+
+static void
+mlx5_esw_bridge_cleanup_switchdev_fdb_work(struct mlx5_bridge_switchdev_fdb_work *fdb_work)
+{
+	dev_put(fdb_work->dev);
+	kfree(fdb_work->fdb_info.addr);
+	kfree(fdb_work);
+}
+
+static void mlx5_esw_bridge_switchdev_fdb_event_work(struct work_struct *work)
+{
+	struct mlx5_bridge_switchdev_fdb_work *fdb_work =
+		container_of(work, struct mlx5_bridge_switchdev_fdb_work, work);
+	struct switchdev_notifier_fdb_info *fdb_info =
+		&fdb_work->fdb_info;
+	struct mlx5_esw_bridge_offloads *br_offloads =
+		fdb_work->br_offloads;
+	struct net_device *dev = fdb_work->dev;
+	u16 vport_num, esw_owner_vhca_id;
+
+	rtnl_lock();
+
+	if (!mlx5_esw_bridge_rep_vport_num_vhca_id_get(dev, br_offloads->esw, &vport_num,
+						       &esw_owner_vhca_id))
+		goto out;
+
+	if (fdb_work->add)
+		mlx5_esw_bridge_fdb_create(dev, vport_num, esw_owner_vhca_id, br_offloads,
+					   fdb_info);
+	else
+		mlx5_esw_bridge_fdb_remove(dev, vport_num, esw_owner_vhca_id, br_offloads,
+					   fdb_info);
+
+out:
+	rtnl_unlock();
+	mlx5_esw_bridge_cleanup_switchdev_fdb_work(fdb_work);
+}
+
+static struct mlx5_bridge_switchdev_fdb_work *
+mlx5_esw_bridge_init_switchdev_fdb_work(struct net_device *dev, bool add,
+					struct switchdev_notifier_fdb_info *fdb_info,
+					struct mlx5_esw_bridge_offloads *br_offloads)
+{
+	struct mlx5_bridge_switchdev_fdb_work *work;
+	u8 *addr;
+
+	work = kzalloc(sizeof(*work), GFP_ATOMIC);
+	if (!work)
+		return ERR_PTR(-ENOMEM);
+
+	INIT_WORK(&work->work, mlx5_esw_bridge_switchdev_fdb_event_work);
+	memcpy(&work->fdb_info, fdb_info, sizeof(work->fdb_info));
+
+	addr = kzalloc(ETH_ALEN, GFP_ATOMIC);
+	if (!addr) {
+		kfree(work);
+		return ERR_PTR(-ENOMEM);
+	}
+	ether_addr_copy(addr, fdb_info->addr);
+	work->fdb_info.addr = addr;
+
+	dev_hold(dev);
+	work->dev = dev;
+	work->br_offloads = br_offloads;
+	work->add = add;
+	return work;
+}
+
+static int mlx5_esw_bridge_switchdev_event(struct notifier_block *nb,
+					   unsigned long event, void *ptr)
+{
+	struct mlx5_esw_bridge_offloads *br_offloads = container_of(nb,
+								    struct mlx5_esw_bridge_offloads,
+								    nb);
+	struct net_device *dev = switchdev_notifier_info_to_dev(ptr);
+	struct switchdev_notifier_fdb_info *fdb_info;
+	struct mlx5_bridge_switchdev_fdb_work *work;
+	struct mlx5_eswitch *esw = br_offloads->esw;
+	struct switchdev_notifier_info *info = ptr;
+	u16 vport_num, esw_owner_vhca_id;
+	struct net_device *upper, *rep;
+
+	if (event == SWITCHDEV_PORT_ATTR_SET) {
+		int err = mlx5_esw_bridge_port_obj_attr_set(dev, ptr, br_offloads);
+
+		return notifier_from_errno(err);
+	}
+
+	upper = netdev_master_upper_dev_get_rcu(dev);
+	if (!upper)
+		return NOTIFY_DONE;
+	if (!netif_is_bridge_master(upper))
+		return NOTIFY_DONE;
+
+	rep = mlx5_esw_bridge_rep_vport_num_vhca_id_get(dev, esw, &vport_num, &esw_owner_vhca_id);
+	if (!rep)
+		return NOTIFY_DONE;
+
+	switch (event) {
+	case SWITCHDEV_FDB_ADD_TO_BRIDGE:
+		fdb_info = container_of(info,
+					struct switchdev_notifier_fdb_info,
+					info);
+		mlx5_esw_bridge_fdb_update_used(dev, vport_num, esw_owner_vhca_id, br_offloads,
+						fdb_info);
+		break;
+	case SWITCHDEV_FDB_DEL_TO_BRIDGE:
+		/* only handle the event on peers */
+		if (mlx5_esw_bridge_is_local(dev, rep, esw))
+			break;
+		fallthrough;
+	case SWITCHDEV_FDB_ADD_TO_DEVICE:
+	case SWITCHDEV_FDB_DEL_TO_DEVICE:
+		fdb_info = container_of(info,
+					struct switchdev_notifier_fdb_info,
+					info);
+
+		work = mlx5_esw_bridge_init_switchdev_fdb_work(dev,
+							       event == SWITCHDEV_FDB_ADD_TO_DEVICE,
+							       fdb_info,
+							       br_offloads);
+		if (IS_ERR(work)) {
+			WARN_ONCE(1, "Failed to init switchdev work, err=%ld",
+				  PTR_ERR(work));
+			return notifier_from_errno(PTR_ERR(work));
+		}
+
+		queue_work(br_offloads->wq, &work->work);
+		break;
+	default:
+		break;
+	}
+	return NOTIFY_DONE;
+}
+
+static void mlx5_esw_bridge_update_work(struct work_struct *work)
+{
+	struct mlx5_esw_bridge_offloads *br_offloads = container_of(work,
+								    struct mlx5_esw_bridge_offloads,
+								    update_work.work);
+
+	rtnl_lock();
+	mlx5_esw_bridge_update(br_offloads);
+	rtnl_unlock();
+
+	queue_delayed_work(br_offloads->wq, &br_offloads->update_work,
+			   msecs_to_jiffies(MLX5_ESW_BRIDGE_UPDATE_INTERVAL));
+}
+
+void mlx5e_rep_bridge_init(struct mlx5e_priv *priv)
+{
+	struct mlx5_esw_bridge_offloads *br_offloads;
+	struct mlx5_core_dev *mdev = priv->mdev;
+	struct mlx5_eswitch *esw =
+		mdev->priv.eswitch;
+	int err;
+
+	rtnl_lock();
+	br_offloads = mlx5_esw_bridge_init(esw);
+	rtnl_unlock();
+	if (IS_ERR(br_offloads)) {
+		esw_warn(mdev, "Failed to init esw bridge (err=%ld)\n", PTR_ERR(br_offloads));
+		return;
+	}
+
+	br_offloads->wq = alloc_ordered_workqueue("mlx5_bridge_wq", 0);
+	if (!br_offloads->wq) {
+		esw_warn(mdev, "Failed to allocate bridge offloads workqueue\n");
+		goto err_alloc_wq;
+	}
+
+	br_offloads->nb.notifier_call = mlx5_esw_bridge_switchdev_event;
+	err = register_switchdev_notifier(&br_offloads->nb);
+	if (err) {
+		esw_warn(mdev, "Failed to register switchdev notifier (err=%d)\n", err);
+		goto err_register_swdev;
+	}
+
+	br_offloads->nb_blk.notifier_call = mlx5_esw_bridge_event_blocking;
+	err = register_switchdev_blocking_notifier(&br_offloads->nb_blk);
+	if (err) {
+		esw_warn(mdev, "Failed to register blocking switchdev notifier (err=%d)\n", err);
+		goto err_register_swdev_blk;
+	}
+
+	br_offloads->netdev_nb.notifier_call = mlx5_esw_bridge_switchdev_port_event;
+	err = register_netdevice_notifier_net(&init_net, &br_offloads->netdev_nb);
+	if (err) {
+		esw_warn(mdev, "Failed to register bridge offloads netdevice notifier (err=%d)\n",
+			 err);
+		goto err_register_netdev;
+	}
+	INIT_DELAYED_WORK(&br_offloads->update_work, mlx5_esw_bridge_update_work);
+	queue_delayed_work(br_offloads->wq, &br_offloads->update_work,
+			   msecs_to_jiffies(MLX5_ESW_BRIDGE_UPDATE_INTERVAL));
+	return;
+
+err_register_netdev:
+	unregister_switchdev_blocking_notifier(&br_offloads->nb_blk);
+err_register_swdev_blk:
+	unregister_switchdev_notifier(&br_offloads->nb);
+err_register_swdev:
+	destroy_workqueue(br_offloads->wq);
+err_alloc_wq:
+	rtnl_lock();
+	mlx5_esw_bridge_cleanup(esw);
+	rtnl_unlock();
+}
+
+void mlx5e_rep_bridge_cleanup(struct mlx5e_priv *priv)
+{
+	struct mlx5_esw_bridge_offloads *br_offloads;
+	struct mlx5_core_dev *mdev = priv->mdev;
+	struct mlx5_eswitch *esw =
+		mdev->priv.eswitch;
+
+	br_offloads = esw->br_offloads;
+	if (!br_offloads)
+		return;
+
+	cancel_delayed_work_sync(&br_offloads->update_work);
+	unregister_netdevice_notifier_net(&init_net, &br_offloads->netdev_nb);
+	unregister_switchdev_blocking_notifier(&br_offloads->nb_blk);
+	unregister_switchdev_notifier(&br_offloads->nb);
+	destroy_workqueue(br_offloads->wq);
+	rtnl_lock();
+	mlx5_esw_bridge_cleanup(esw);
+	rtnl_unlock();
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/rep/bridge.h b/drivers/net/ethernet/mellanox/mlx5/core/en/rep/bridge.h
new file mode 100644
index 000000000..fbeb64242
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/rep/bridge.h
@@ -0,0 +1,21 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2021 Mellanox Technologies. */
+
+#ifndef __MLX5_EN_REP_BRIDGE__
+#define __MLX5_EN_REP_BRIDGE__
+
+#include "en.h"
+
+#if IS_ENABLED(CONFIG_MLX5_BRIDGE)
+
+void mlx5e_rep_bridge_init(struct mlx5e_priv *priv);
+void mlx5e_rep_bridge_cleanup(struct mlx5e_priv *priv);
+
+#else /* CONFIG_MLX5_BRIDGE */
+
+static inline void mlx5e_rep_bridge_init(struct mlx5e_priv *priv) {}
+static inline void mlx5e_rep_bridge_cleanup(struct mlx5e_priv *priv) {}
+
+#endif /* CONFIG_MLX5_BRIDGE */
+
+#endif /* __MLX5_EN_REP_BRIDGE__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/rep/neigh.c b/drivers/net/ethernet/mellanox/mlx5/core/en/rep/neigh.c
new file mode 100644
index 000000000..2e9bee4e5
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/rep/neigh.c
@@ -0,0 +1,398 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2020 Mellanox Technologies. */
+
+#include <linux/refcount.h>
+#include <linux/list.h>
+#include <linux/rculist.h>
+#include <linux/rtnetlink.h>
+#include <linux/workqueue.h>
+#include <linux/spinlock.h>
+#include <linux/notifier.h>
+#include <net/netevent.h>
+#include <net/arp.h>
+#include "neigh.h"
+#include "tc.h"
+#include "en_rep.h"
+#include "fs_core.h"
+#include "diag/en_rep_tracepoint.h"
+
+static unsigned long mlx5e_rep_ipv6_interval(void)
+{
+	if (IS_ENABLED(CONFIG_IPV6) && ipv6_stub->nd_tbl)
+		return NEIGH_VAR(&ipv6_stub->nd_tbl->parms, DELAY_PROBE_TIME);
+
+	return ~0UL;
+}
+
+static void mlx5e_rep_neigh_update_init_interval(struct mlx5e_rep_priv *rpriv)
+{
+	unsigned long ipv4_interval = NEIGH_VAR(&arp_tbl.parms, DELAY_PROBE_TIME);
+	unsigned long ipv6_interval = mlx5e_rep_ipv6_interval();
+	struct net_device *netdev = rpriv->netdev;
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+
+	rpriv->neigh_update.min_interval = min_t(unsigned long, ipv6_interval, ipv4_interval);
+	mlx5_fc_update_sampling_interval(priv->mdev, rpriv->neigh_update.min_interval);
+}
+
+void mlx5e_rep_queue_neigh_stats_work(struct mlx5e_priv *priv)
+{
+	struct mlx5e_rep_priv *rpriv = priv->ppriv;
+	struct mlx5e_neigh_update_table *neigh_update = &rpriv->neigh_update;
+
+	mlx5_fc_queue_stats_work(priv->mdev,
+				 &neigh_update->neigh_stats_work,
+				 neigh_update->min_interval);
+}
+
+static bool mlx5e_rep_neigh_entry_hold(struct mlx5e_neigh_hash_entry *nhe)
+{
+	return refcount_inc_not_zero(&nhe->refcnt);
+}
+
+static void mlx5e_rep_neigh_entry_remove(struct mlx5e_neigh_hash_entry *nhe);
+
+void mlx5e_rep_neigh_entry_release(struct mlx5e_neigh_hash_entry *nhe)
+{
+	if (refcount_dec_and_test(&nhe->refcnt)) {
+		mlx5e_rep_neigh_entry_remove(nhe);
+		kfree_rcu(nhe, rcu);
+	}
+}
+
+static struct mlx5e_neigh_hash_entry *
+mlx5e_get_next_nhe(struct mlx5e_rep_priv *rpriv,
+		   struct mlx5e_neigh_hash_entry *nhe)
+{
+	struct mlx5e_neigh_hash_entry *next = NULL;
+
+	rcu_read_lock();
+
+	for (next = nhe ?
+		     list_next_or_null_rcu(&rpriv->neigh_update.neigh_list,
+					   &nhe->neigh_list,
+					   struct mlx5e_neigh_hash_entry,
+					   neigh_list) :
+		     list_first_or_null_rcu(&rpriv->neigh_update.neigh_list,
+					    struct mlx5e_neigh_hash_entry,
+					    neigh_list);
+	     next;
+	     next = list_next_or_null_rcu(&rpriv->neigh_update.neigh_list,
+					  &next->neigh_list,
+					  struct mlx5e_neigh_hash_entry,
+					  neigh_list))
+		if (mlx5e_rep_neigh_entry_hold(next))
+			break;
+
+	rcu_read_unlock();
+
+	if (nhe)
+		mlx5e_rep_neigh_entry_release(nhe);
+
+	return next;
+}
+
+static void mlx5e_rep_neigh_stats_work(struct work_struct *work)
+{
+	struct mlx5e_rep_priv *rpriv = container_of(work, struct mlx5e_rep_priv,
+						    neigh_update.neigh_stats_work.work);
+	struct net_device *netdev = rpriv->netdev;
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	struct mlx5e_neigh_hash_entry *nhe = NULL;
+
+	rtnl_lock();
+	if (!list_empty(&rpriv->neigh_update.neigh_list))
+		mlx5e_rep_queue_neigh_stats_work(priv);
+
+	while ((nhe = mlx5e_get_next_nhe(rpriv, nhe)) != NULL)
+		mlx5e_tc_update_neigh_used_value(nhe);
+
+	rtnl_unlock();
+}
+
+struct neigh_update_work {
+	struct work_struct work;
+	struct neighbour *n;
+	struct mlx5e_neigh_hash_entry *nhe;
+};
+
+static void mlx5e_release_neigh_update_work(struct neigh_update_work *update_work)
+{
+	neigh_release(update_work->n);
+	mlx5e_rep_neigh_entry_release(update_work->nhe);
+	kfree(update_work);
+}
+
+static void mlx5e_rep_neigh_update(struct work_struct *work)
+{
+	struct neigh_update_work *update_work = container_of(work, struct neigh_update_work,
+							     work);
+	struct mlx5e_neigh_hash_entry *nhe = update_work->nhe;
+	struct neighbour *n = update_work->n;
+	struct mlx5e_encap_entry *e = NULL;
+	bool neigh_connected, same_dev;
+	unsigned char ha[ETH_ALEN];
+	u8 nud_state, dead;
+
+	rtnl_lock();
+
+	/* If these parameters are changed after we release the lock,
+	 * we'll receive another event letting us know about it.
+	 * We use this lock to avoid inconsistency between the neigh validity
+	 * and it's hw address.
+	 */
+	read_lock_bh(&n->lock);
+	memcpy(ha, n->ha, ETH_ALEN);
+	nud_state = n->nud_state;
+	dead = n->dead;
+	same_dev = READ_ONCE(nhe->neigh_dev) == n->dev;
+	read_unlock_bh(&n->lock);
+
+	neigh_connected = (nud_state & NUD_VALID) && !dead;
+
+	trace_mlx5e_rep_neigh_update(nhe, ha, neigh_connected);
+
+	if (!same_dev)
+		goto out;
+
+	/* mlx5e_get_next_init_encap() releases previous encap before returning
+	 * the next one.
+	 */
+	while ((e = mlx5e_get_next_init_encap(nhe, e)) != NULL)
+		mlx5e_rep_update_flows(netdev_priv(e->out_dev), e, neigh_connected, ha);
+
+out:
+	rtnl_unlock();
+	mlx5e_release_neigh_update_work(update_work);
+}
+
+static struct neigh_update_work *mlx5e_alloc_neigh_update_work(struct mlx5e_priv *priv,
+							       struct neighbour *n)
+{
+	struct neigh_update_work *update_work;
+	struct mlx5e_neigh_hash_entry *nhe;
+	struct mlx5e_neigh m_neigh = {};
+
+	update_work = kzalloc(sizeof(*update_work), GFP_ATOMIC);
+	if (WARN_ON(!update_work))
+		return NULL;
+
+	m_neigh.family = n->ops->family;
+	memcpy(&m_neigh.dst_ip, n->primary_key, n->tbl->key_len);
+
+	/* Obtain reference to nhe as last step in order not to release it in
+	 * atomic context.
+	 */
+	rcu_read_lock();
+	nhe = mlx5e_rep_neigh_entry_lookup(priv, &m_neigh);
+	rcu_read_unlock();
+	if (!nhe) {
+		kfree(update_work);
+		return NULL;
+	}
+
+	INIT_WORK(&update_work->work, mlx5e_rep_neigh_update);
+	neigh_hold(n);
+	update_work->n = n;
+	update_work->nhe = nhe;
+
+	return update_work;
+}
+
+static int mlx5e_rep_netevent_event(struct notifier_block *nb,
+				    unsigned long event, void *ptr)
+{
+	struct mlx5e_rep_priv *rpriv = container_of(nb, struct mlx5e_rep_priv,
+						    neigh_update.netevent_nb);
+	struct mlx5e_neigh_update_table *neigh_update = &rpriv->neigh_update;
+	struct net_device *netdev = rpriv->netdev;
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	struct mlx5e_neigh_hash_entry *nhe = NULL;
+	struct neigh_update_work *update_work;
+	struct neigh_parms *p;
+	struct neighbour *n;
+	bool found = false;
+
+	switch (event) {
+	case NETEVENT_NEIGH_UPDATE:
+		n = ptr;
+#if IS_ENABLED(CONFIG_IPV6)
+		if (n->tbl != ipv6_stub->nd_tbl && n->tbl != &arp_tbl)
+#else
+		if (n->tbl != &arp_tbl)
+#endif
+			return NOTIFY_DONE;
+
+		update_work = mlx5e_alloc_neigh_update_work(priv, n);
+		if (!update_work)
+			return NOTIFY_DONE;
+
+		queue_work(priv->wq, &update_work->work);
+		break;
+
+	case NETEVENT_DELAY_PROBE_TIME_UPDATE:
+		p = ptr;
+
+		/* We check the device is present since we don't care about
+		 * changes in the default table, we only care about changes
+		 * done per device delay prob time parameter.
+		 */
+#if IS_ENABLED(CONFIG_IPV6)
+		if (!p->dev || (p->tbl != ipv6_stub->nd_tbl && p->tbl != &arp_tbl))
+#else
+		if (!p->dev || p->tbl != &arp_tbl)
+#endif
+			return NOTIFY_DONE;
+
+		rcu_read_lock();
+		list_for_each_entry_rcu(nhe, &neigh_update->neigh_list,
+					neigh_list) {
+			if (p->dev == READ_ONCE(nhe->neigh_dev)) {
+				found = true;
+				break;
+			}
+		}
+		rcu_read_unlock();
+		if (!found)
+			return NOTIFY_DONE;
+
+		neigh_update->min_interval = min_t(unsigned long,
+						   NEIGH_VAR(p, DELAY_PROBE_TIME),
+						   neigh_update->min_interval);
+		mlx5_fc_update_sampling_interval(priv->mdev,
+						 neigh_update->min_interval);
+		break;
+	}
+	return NOTIFY_DONE;
+}
+
+static const struct rhashtable_params mlx5e_neigh_ht_params = {
+	.head_offset = offsetof(struct mlx5e_neigh_hash_entry, rhash_node),
+	.key_offset = offsetof(struct mlx5e_neigh_hash_entry, m_neigh),
+	.key_len = sizeof(struct mlx5e_neigh),
+	.automatic_shrinking = true,
+};
+
+int mlx5e_rep_neigh_init(struct mlx5e_rep_priv *rpriv)
+{
+	struct mlx5e_neigh_update_table *neigh_update = &rpriv->neigh_update;
+	int err;
+
+	err = rhashtable_init(&neigh_update->neigh_ht, &mlx5e_neigh_ht_params);
+	if (err)
+		goto out_err;
+
+	INIT_LIST_HEAD(&neigh_update->neigh_list);
+	mutex_init(&neigh_update->encap_lock);
+	INIT_DELAYED_WORK(&neigh_update->neigh_stats_work,
+			  mlx5e_rep_neigh_stats_work);
+	mlx5e_rep_neigh_update_init_interval(rpriv);
+
+	neigh_update->netevent_nb.notifier_call = mlx5e_rep_netevent_event;
+	err = register_netevent_notifier(&neigh_update->netevent_nb);
+	if (err)
+		goto out_notifier;
+	return 0;
+
+out_notifier:
+	neigh_update->netevent_nb.notifier_call = NULL;
+	rhashtable_destroy(&neigh_update->neigh_ht);
+out_err:
+	netdev_warn(rpriv->netdev,
+		    "Failed to initialize neighbours handling for vport %d\n",
+		    rpriv->rep->vport);
+	return err;
+}
+
+void mlx5e_rep_neigh_cleanup(struct mlx5e_rep_priv *rpriv)
+{
+	struct mlx5e_neigh_update_table *neigh_update = &rpriv->neigh_update;
+	struct mlx5e_priv *priv = netdev_priv(rpriv->netdev);
+
+	if (!rpriv->neigh_update.netevent_nb.notifier_call)
+		return;
+
+	unregister_netevent_notifier(&neigh_update->netevent_nb);
+
+	flush_workqueue(priv->wq); /* flush neigh update works */
+
+	cancel_delayed_work_sync(&rpriv->neigh_update.neigh_stats_work);
+
+	mutex_destroy(&neigh_update->encap_lock);
+	rhashtable_destroy(&neigh_update->neigh_ht);
+}
+
+static int mlx5e_rep_neigh_entry_insert(struct mlx5e_priv *priv,
+					struct mlx5e_neigh_hash_entry *nhe)
+{
+	struct mlx5e_rep_priv *rpriv = priv->ppriv;
+	int err;
+
+	err = rhashtable_insert_fast(&rpriv->neigh_update.neigh_ht,
+				     &nhe->rhash_node,
+				     mlx5e_neigh_ht_params);
+	if (err)
+		return err;
+
+	list_add_rcu(&nhe->neigh_list, &rpriv->neigh_update.neigh_list);
+
+	return err;
+}
+
+static void mlx5e_rep_neigh_entry_remove(struct mlx5e_neigh_hash_entry *nhe)
+{
+	struct mlx5e_rep_priv *rpriv = nhe->priv->ppriv;
+
+	mutex_lock(&rpriv->neigh_update.encap_lock);
+
+	list_del_rcu(&nhe->neigh_list);
+
+	rhashtable_remove_fast(&rpriv->neigh_update.neigh_ht,
+			       &nhe->rhash_node,
+			       mlx5e_neigh_ht_params);
+	mutex_unlock(&rpriv->neigh_update.encap_lock);
+}
+
+/* This function must only be called under the representor's encap_lock or
+ * inside rcu read lock section.
+ */
+struct mlx5e_neigh_hash_entry *
+mlx5e_rep_neigh_entry_lookup(struct mlx5e_priv *priv,
+			     struct mlx5e_neigh *m_neigh)
+{
+	struct mlx5e_rep_priv *rpriv = priv->ppriv;
+	struct mlx5e_neigh_update_table *neigh_update = &rpriv->neigh_update;
+	struct mlx5e_neigh_hash_entry *nhe;
+
+	nhe = rhashtable_lookup_fast(&neigh_update->neigh_ht, m_neigh,
+				     mlx5e_neigh_ht_params);
+	return nhe && mlx5e_rep_neigh_entry_hold(nhe) ? nhe : NULL;
+}
+
+int mlx5e_rep_neigh_entry_create(struct mlx5e_priv *priv,
+				 struct mlx5e_neigh *m_neigh,
+				 struct net_device *neigh_dev,
+				 struct mlx5e_neigh_hash_entry **nhe)
+{
+	int err;
+
+	*nhe = kzalloc(sizeof(**nhe), GFP_KERNEL);
+	if (!*nhe)
+		return -ENOMEM;
+
+	(*nhe)->priv = priv;
+	memcpy(&(*nhe)->m_neigh, m_neigh, sizeof(*m_neigh));
+	spin_lock_init(&(*nhe)->encap_list_lock);
+	INIT_LIST_HEAD(&(*nhe)->encap_list);
+	refcount_set(&(*nhe)->refcnt, 1);
+	WRITE_ONCE((*nhe)->neigh_dev, neigh_dev);
+
+	err = mlx5e_rep_neigh_entry_insert(priv, *nhe);
+	if (err)
+		goto out_free;
+	return 0;
+
+out_free:
+	kfree(*nhe);
+	return err;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/rep/neigh.h b/drivers/net/ethernet/mellanox/mlx5/core/en/rep/neigh.h
new file mode 100644
index 000000000..6fe0ab970
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/rep/neigh.h
@@ -0,0 +1,35 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2020 Mellanox Technologies. */
+
+#ifndef __MLX5_EN_REP_NEIGH__
+#define __MLX5_EN_REP_NEIGH__
+
+#include "en.h"
+#include "en_rep.h"
+
+#if IS_ENABLED(CONFIG_MLX5_CLS_ACT)
+
+int mlx5e_rep_neigh_init(struct mlx5e_rep_priv *rpriv);
+void mlx5e_rep_neigh_cleanup(struct mlx5e_rep_priv *rpriv);
+
+struct mlx5e_neigh_hash_entry *
+mlx5e_rep_neigh_entry_lookup(struct mlx5e_priv *priv,
+			     struct mlx5e_neigh *m_neigh);
+int mlx5e_rep_neigh_entry_create(struct mlx5e_priv *priv,
+				 struct mlx5e_neigh *m_neigh,
+				 struct net_device *neigh_dev,
+				 struct mlx5e_neigh_hash_entry **nhe);
+void mlx5e_rep_neigh_entry_release(struct mlx5e_neigh_hash_entry *nhe);
+
+void mlx5e_rep_queue_neigh_stats_work(struct mlx5e_priv *priv);
+
+#else /* CONFIG_MLX5_CLS_ACT */
+
+static inline int
+mlx5e_rep_neigh_init(struct mlx5e_rep_priv *rpriv) { return 0; }
+static inline void
+mlx5e_rep_neigh_cleanup(struct mlx5e_rep_priv *rpriv) {}
+
+#endif /* CONFIG_MLX5_CLS_ACT */
+
+#endif /* __MLX5_EN_REP_NEIGH__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/rep/tc.c b/drivers/net/ethernet/mellanox/mlx5/core/en/rep/tc.c
new file mode 100644
index 000000000..fac7e3ff2
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/rep/tc.c
@@ -0,0 +1,900 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2020 Mellanox Technologies. */
+
+#include <net/dst_metadata.h>
+#include <linux/netdevice.h>
+#include <linux/if_macvlan.h>
+#include <linux/list.h>
+#include <linux/rculist.h>
+#include <linux/rtnetlink.h>
+#include <linux/workqueue.h>
+#include <linux/spinlock.h>
+#include "tc.h"
+#include "neigh.h"
+#include "en_rep.h"
+#include "eswitch.h"
+#include "lib/fs_chains.h"
+#include "en/tc_ct.h"
+#include "en/mapping.h"
+#include "en/tc_tun.h"
+#include "lib/port_tun.h"
+#include "en/tc/sample.h"
+#include "en_accel/ipsec_rxtx.h"
+#include "en/tc/int_port.h"
+#include "en/tc/act/act.h"
+
+struct mlx5e_rep_indr_block_priv {
+	struct net_device *netdev;
+	struct mlx5e_rep_priv *rpriv;
+	enum flow_block_binder_type binder_type;
+
+	struct list_head list;
+};
+
+int mlx5e_rep_encap_entry_attach(struct mlx5e_priv *priv,
+				 struct mlx5e_encap_entry *e,
+				 struct mlx5e_neigh *m_neigh,
+				 struct net_device *neigh_dev)
+{
+	struct mlx5e_rep_priv *rpriv = priv->ppriv;
+	struct mlx5_rep_uplink_priv *uplink_priv = &rpriv->uplink_priv;
+	struct mlx5_tun_entropy *tun_entropy = &uplink_priv->tun_entropy;
+	struct mlx5e_neigh_hash_entry *nhe;
+	int err;
+
+	err = mlx5_tun_entropy_refcount_inc(tun_entropy, e->reformat_type);
+	if (err)
+		return err;
+
+	mutex_lock(&rpriv->neigh_update.encap_lock);
+	nhe = mlx5e_rep_neigh_entry_lookup(priv, m_neigh);
+	if (!nhe) {
+		err = mlx5e_rep_neigh_entry_create(priv, m_neigh, neigh_dev, &nhe);
+		if (err) {
+			mutex_unlock(&rpriv->neigh_update.encap_lock);
+			mlx5_tun_entropy_refcount_dec(tun_entropy,
+						      e->reformat_type);
+			return err;
+		}
+	}
+
+	e->nhe = nhe;
+	spin_lock(&nhe->encap_list_lock);
+	list_add_rcu(&e->encap_list, &nhe->encap_list);
+	spin_unlock(&nhe->encap_list_lock);
+
+	mutex_unlock(&rpriv->neigh_update.encap_lock);
+
+	return 0;
+}
+
+void mlx5e_rep_encap_entry_detach(struct mlx5e_priv *priv,
+				  struct mlx5e_encap_entry *e)
+{
+	struct mlx5e_rep_priv *rpriv = priv->ppriv;
+	struct mlx5_rep_uplink_priv *uplink_priv = &rpriv->uplink_priv;
+	struct mlx5_tun_entropy *tun_entropy = &uplink_priv->tun_entropy;
+
+	if (!e->nhe)
+		return;
+
+	spin_lock(&e->nhe->encap_list_lock);
+	list_del_rcu(&e->encap_list);
+	spin_unlock(&e->nhe->encap_list_lock);
+
+	mlx5e_rep_neigh_entry_release(e->nhe);
+	e->nhe = NULL;
+	mlx5_tun_entropy_refcount_dec(tun_entropy, e->reformat_type);
+}
+
+void mlx5e_rep_update_flows(struct mlx5e_priv *priv,
+			    struct mlx5e_encap_entry *e,
+			    bool neigh_connected,
+			    unsigned char ha[ETH_ALEN])
+{
+	struct ethhdr *eth = (struct ethhdr *)e->encap_header;
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	bool encap_connected;
+	LIST_HEAD(flow_list);
+
+	ASSERT_RTNL();
+
+	mutex_lock(&esw->offloads.encap_tbl_lock);
+	encap_connected = !!(e->flags & MLX5_ENCAP_ENTRY_VALID);
+	if (encap_connected == neigh_connected && ether_addr_equal(e->h_dest, ha))
+		goto unlock;
+
+	mlx5e_take_all_encap_flows(e, &flow_list);
+
+	if ((e->flags & MLX5_ENCAP_ENTRY_VALID) &&
+	    (!neigh_connected || !ether_addr_equal(e->h_dest, ha)))
+		mlx5e_tc_encap_flows_del(priv, e, &flow_list);
+
+	if (neigh_connected && !(e->flags & MLX5_ENCAP_ENTRY_VALID)) {
+		struct net_device *route_dev;
+
+		ether_addr_copy(e->h_dest, ha);
+		ether_addr_copy(eth->h_dest, ha);
+		/* Update the encap source mac, in case that we delete
+		 * the flows when encap source mac changed.
+		 */
+		route_dev = __dev_get_by_index(dev_net(priv->netdev), e->route_dev_ifindex);
+		if (route_dev)
+			ether_addr_copy(eth->h_source, route_dev->dev_addr);
+
+		mlx5e_tc_encap_flows_add(priv, e, &flow_list);
+	}
+unlock:
+	mutex_unlock(&esw->offloads.encap_tbl_lock);
+	mlx5e_put_flow_list(priv, &flow_list);
+}
+
+static int
+mlx5e_rep_setup_tc_cls_flower(struct mlx5e_priv *priv,
+			      struct flow_cls_offload *cls_flower, int flags)
+{
+	switch (cls_flower->command) {
+	case FLOW_CLS_REPLACE:
+		return mlx5e_configure_flower(priv->netdev, priv, cls_flower,
+					      flags);
+	case FLOW_CLS_DESTROY:
+		return mlx5e_delete_flower(priv->netdev, priv, cls_flower,
+					   flags);
+	case FLOW_CLS_STATS:
+		return mlx5e_stats_flower(priv->netdev, priv, cls_flower,
+					  flags);
+	default:
+		return -EOPNOTSUPP;
+	}
+}
+
+static
+int mlx5e_rep_setup_tc_cls_matchall(struct mlx5e_priv *priv,
+				    struct tc_cls_matchall_offload *ma)
+{
+	switch (ma->command) {
+	case TC_CLSMATCHALL_REPLACE:
+		return mlx5e_tc_configure_matchall(priv, ma);
+	case TC_CLSMATCHALL_DESTROY:
+		return mlx5e_tc_delete_matchall(priv, ma);
+	case TC_CLSMATCHALL_STATS:
+		mlx5e_tc_stats_matchall(priv, ma);
+		return 0;
+	default:
+		return -EOPNOTSUPP;
+	}
+}
+
+static int mlx5e_rep_setup_tc_cb(enum tc_setup_type type, void *type_data,
+				 void *cb_priv)
+{
+	unsigned long flags = MLX5_TC_FLAG(INGRESS) | MLX5_TC_FLAG(ESW_OFFLOAD);
+	struct mlx5e_priv *priv = cb_priv;
+
+	if (!priv->netdev || !netif_device_present(priv->netdev))
+		return -EOPNOTSUPP;
+
+	switch (type) {
+	case TC_SETUP_CLSFLOWER:
+		return mlx5e_rep_setup_tc_cls_flower(priv, type_data, flags);
+	case TC_SETUP_CLSMATCHALL:
+		return mlx5e_rep_setup_tc_cls_matchall(priv, type_data);
+	default:
+		return -EOPNOTSUPP;
+	}
+}
+
+static int mlx5e_rep_setup_ft_cb(enum tc_setup_type type, void *type_data,
+				 void *cb_priv)
+{
+	struct flow_cls_offload tmp, *f = type_data;
+	struct mlx5e_priv *priv = cb_priv;
+	struct mlx5_eswitch *esw;
+	unsigned long flags;
+	int err;
+
+	flags = MLX5_TC_FLAG(INGRESS) |
+		MLX5_TC_FLAG(ESW_OFFLOAD) |
+		MLX5_TC_FLAG(FT_OFFLOAD);
+	esw = priv->mdev->priv.eswitch;
+
+	switch (type) {
+	case TC_SETUP_CLSFLOWER:
+		memcpy(&tmp, f, sizeof(*f));
+
+		if (!mlx5_chains_prios_supported(esw_chains(esw)))
+			return -EOPNOTSUPP;
+
+		/* Re-use tc offload path by moving the ft flow to the
+		 * reserved ft chain.
+		 *
+		 * FT offload can use prio range [0, INT_MAX], so we normalize
+		 * it to range [1, mlx5_esw_chains_get_prio_range(esw)]
+		 * as with tc, where prio 0 isn't supported.
+		 *
+		 * We only support chain 0 of FT offload.
+		 */
+		if (tmp.common.prio >= mlx5_chains_get_prio_range(esw_chains(esw)))
+			return -EOPNOTSUPP;
+		if (tmp.common.chain_index != 0)
+			return -EOPNOTSUPP;
+
+		tmp.common.chain_index = mlx5_chains_get_nf_ft_chain(esw_chains(esw));
+		tmp.common.prio++;
+		err = mlx5e_rep_setup_tc_cls_flower(priv, &tmp, flags);
+		memcpy(&f->stats, &tmp.stats, sizeof(f->stats));
+		return err;
+	default:
+		return -EOPNOTSUPP;
+	}
+}
+
+static LIST_HEAD(mlx5e_rep_block_tc_cb_list);
+static LIST_HEAD(mlx5e_rep_block_ft_cb_list);
+int mlx5e_rep_setup_tc(struct net_device *dev, enum tc_setup_type type,
+		       void *type_data)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	struct flow_block_offload *f = type_data;
+
+	f->unlocked_driver_cb = true;
+
+	switch (type) {
+	case TC_SETUP_BLOCK:
+		return flow_block_cb_setup_simple(type_data,
+						  &mlx5e_rep_block_tc_cb_list,
+						  mlx5e_rep_setup_tc_cb,
+						  priv, priv, true);
+	case TC_SETUP_FT:
+		return flow_block_cb_setup_simple(type_data,
+						  &mlx5e_rep_block_ft_cb_list,
+						  mlx5e_rep_setup_ft_cb,
+						  priv, priv, true);
+	default:
+		return -EOPNOTSUPP;
+	}
+}
+
+int mlx5e_rep_tc_init(struct mlx5e_rep_priv *rpriv)
+{
+	struct mlx5_rep_uplink_priv *uplink_priv = &rpriv->uplink_priv;
+	int err;
+
+	mutex_init(&uplink_priv->unready_flows_lock);
+	INIT_LIST_HEAD(&uplink_priv->unready_flows);
+
+	/* init shared tc flow table */
+	err = mlx5e_tc_esw_init(uplink_priv);
+	return err;
+}
+
+void mlx5e_rep_tc_cleanup(struct mlx5e_rep_priv *rpriv)
+{
+	/* delete shared tc flow table */
+	mlx5e_tc_esw_cleanup(&rpriv->uplink_priv);
+	mutex_destroy(&rpriv->uplink_priv.unready_flows_lock);
+}
+
+void mlx5e_rep_tc_enable(struct mlx5e_priv *priv)
+{
+	struct mlx5e_rep_priv *rpriv = priv->ppriv;
+
+	INIT_WORK(&rpriv->uplink_priv.reoffload_flows_work,
+		  mlx5e_tc_reoffload_flows_work);
+}
+
+void mlx5e_rep_tc_disable(struct mlx5e_priv *priv)
+{
+	struct mlx5e_rep_priv *rpriv = priv->ppriv;
+
+	cancel_work_sync(&rpriv->uplink_priv.reoffload_flows_work);
+}
+
+int mlx5e_rep_tc_event_port_affinity(struct mlx5e_priv *priv)
+{
+	struct mlx5e_rep_priv *rpriv = priv->ppriv;
+
+	queue_work(priv->wq, &rpriv->uplink_priv.reoffload_flows_work);
+
+	return NOTIFY_OK;
+}
+
+static struct mlx5e_rep_indr_block_priv *
+mlx5e_rep_indr_block_priv_lookup(struct mlx5e_rep_priv *rpriv,
+				 struct net_device *netdev,
+				 enum flow_block_binder_type binder_type)
+{
+	struct mlx5e_rep_indr_block_priv *cb_priv;
+
+	list_for_each_entry(cb_priv,
+			    &rpriv->uplink_priv.tc_indr_block_priv_list,
+			    list)
+		if (cb_priv->netdev == netdev &&
+		    cb_priv->binder_type == binder_type)
+			return cb_priv;
+
+	return NULL;
+}
+
+static int
+mlx5e_rep_indr_offload(struct net_device *netdev,
+		       struct flow_cls_offload *flower,
+		       struct mlx5e_rep_indr_block_priv *indr_priv,
+		       unsigned long flags)
+{
+	struct mlx5e_priv *priv = netdev_priv(indr_priv->rpriv->netdev);
+	int err = 0;
+
+	if (!netif_device_present(indr_priv->rpriv->netdev))
+		return -EOPNOTSUPP;
+
+	switch (flower->command) {
+	case FLOW_CLS_REPLACE:
+		err = mlx5e_configure_flower(netdev, priv, flower, flags);
+		break;
+	case FLOW_CLS_DESTROY:
+		err = mlx5e_delete_flower(netdev, priv, flower, flags);
+		break;
+	case FLOW_CLS_STATS:
+		err = mlx5e_stats_flower(netdev, priv, flower, flags);
+		break;
+	default:
+		err = -EOPNOTSUPP;
+	}
+
+	return err;
+}
+
+static int mlx5e_rep_indr_setup_tc_cb(enum tc_setup_type type,
+				      void *type_data, void *indr_priv)
+{
+	unsigned long flags = MLX5_TC_FLAG(ESW_OFFLOAD);
+	struct mlx5e_rep_indr_block_priv *priv = indr_priv;
+
+	flags |= (priv->binder_type == FLOW_BLOCK_BINDER_TYPE_CLSACT_EGRESS) ?
+		MLX5_TC_FLAG(EGRESS) :
+		MLX5_TC_FLAG(INGRESS);
+
+	switch (type) {
+	case TC_SETUP_CLSFLOWER:
+		return mlx5e_rep_indr_offload(priv->netdev, type_data, priv,
+					      flags);
+	default:
+		return -EOPNOTSUPP;
+	}
+}
+
+static int mlx5e_rep_indr_setup_ft_cb(enum tc_setup_type type,
+				      void *type_data, void *indr_priv)
+{
+	struct mlx5e_rep_indr_block_priv *priv = indr_priv;
+	struct flow_cls_offload *f = type_data;
+	struct flow_cls_offload tmp;
+	struct mlx5e_priv *mpriv;
+	struct mlx5_eswitch *esw;
+	unsigned long flags;
+	int err;
+
+	mpriv = netdev_priv(priv->rpriv->netdev);
+	esw = mpriv->mdev->priv.eswitch;
+
+	flags = MLX5_TC_FLAG(EGRESS) |
+		MLX5_TC_FLAG(ESW_OFFLOAD) |
+		MLX5_TC_FLAG(FT_OFFLOAD);
+
+	switch (type) {
+	case TC_SETUP_CLSFLOWER:
+		memcpy(&tmp, f, sizeof(*f));
+
+		/* Re-use tc offload path by moving the ft flow to the
+		 * reserved ft chain.
+		 *
+		 * FT offload can use prio range [0, INT_MAX], so we normalize
+		 * it to range [1, mlx5_esw_chains_get_prio_range(esw)]
+		 * as with tc, where prio 0 isn't supported.
+		 *
+		 * We only support chain 0 of FT offload.
+		 */
+		if (!mlx5_chains_prios_supported(esw_chains(esw)) ||
+		    tmp.common.prio >= mlx5_chains_get_prio_range(esw_chains(esw)) ||
+		    tmp.common.chain_index)
+			return -EOPNOTSUPP;
+
+		tmp.common.chain_index = mlx5_chains_get_nf_ft_chain(esw_chains(esw));
+		tmp.common.prio++;
+		err = mlx5e_rep_indr_offload(priv->netdev, &tmp, priv, flags);
+		memcpy(&f->stats, &tmp.stats, sizeof(f->stats));
+		return err;
+	default:
+		return -EOPNOTSUPP;
+	}
+}
+
+static void mlx5e_rep_indr_block_unbind(void *cb_priv)
+{
+	struct mlx5e_rep_indr_block_priv *indr_priv = cb_priv;
+
+	list_del(&indr_priv->list);
+	kfree(indr_priv);
+}
+
+static LIST_HEAD(mlx5e_block_cb_list);
+
+static bool mlx5e_rep_macvlan_mode_supported(const struct net_device *dev)
+{
+	struct macvlan_dev *macvlan = netdev_priv(dev);
+
+	return macvlan->mode == MACVLAN_MODE_PASSTHRU;
+}
+
+static int
+mlx5e_rep_indr_setup_block(struct net_device *netdev, struct Qdisc *sch,
+			   struct mlx5e_rep_priv *rpriv,
+			   struct flow_block_offload *f,
+			   flow_setup_cb_t *setup_cb,
+			   void *data,
+			   void (*cleanup)(struct flow_block_cb *block_cb))
+{
+	struct mlx5e_priv *priv = netdev_priv(rpriv->netdev);
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	bool is_ovs_int_port = netif_is_ovs_master(netdev);
+	struct mlx5e_rep_indr_block_priv *indr_priv;
+	struct flow_block_cb *block_cb;
+
+	if (!mlx5e_tc_tun_device_to_offload(priv, netdev) &&
+	    !(is_vlan_dev(netdev) && vlan_dev_real_dev(netdev) == rpriv->netdev) &&
+	    !is_ovs_int_port) {
+		if (!(netif_is_macvlan(netdev) && macvlan_dev_real_dev(netdev) == rpriv->netdev))
+			return -EOPNOTSUPP;
+		if (!mlx5e_rep_macvlan_mode_supported(netdev)) {
+			netdev_warn(netdev, "Offloading ingress filter is supported only with macvlan passthru mode");
+			return -EOPNOTSUPP;
+		}
+	}
+
+	if (f->binder_type != FLOW_BLOCK_BINDER_TYPE_CLSACT_INGRESS &&
+	    f->binder_type != FLOW_BLOCK_BINDER_TYPE_CLSACT_EGRESS)
+		return -EOPNOTSUPP;
+
+	if (f->binder_type == FLOW_BLOCK_BINDER_TYPE_CLSACT_EGRESS && !is_ovs_int_port)
+		return -EOPNOTSUPP;
+
+	if (is_ovs_int_port && !mlx5e_tc_int_port_supported(esw))
+		return -EOPNOTSUPP;
+
+	f->unlocked_driver_cb = true;
+	f->driver_block_list = &mlx5e_block_cb_list;
+
+	switch (f->command) {
+	case FLOW_BLOCK_BIND:
+		indr_priv = mlx5e_rep_indr_block_priv_lookup(rpriv, netdev, f->binder_type);
+		if (indr_priv)
+			return -EEXIST;
+
+		indr_priv = kmalloc(sizeof(*indr_priv), GFP_KERNEL);
+		if (!indr_priv)
+			return -ENOMEM;
+
+		indr_priv->netdev = netdev;
+		indr_priv->rpriv = rpriv;
+		indr_priv->binder_type = f->binder_type;
+		list_add(&indr_priv->list,
+			 &rpriv->uplink_priv.tc_indr_block_priv_list);
+
+		block_cb = flow_indr_block_cb_alloc(setup_cb, indr_priv, indr_priv,
+						    mlx5e_rep_indr_block_unbind,
+						    f, netdev, sch, data, rpriv,
+						    cleanup);
+		if (IS_ERR(block_cb)) {
+			list_del(&indr_priv->list);
+			kfree(indr_priv);
+			return PTR_ERR(block_cb);
+		}
+		flow_block_cb_add(block_cb, f);
+		list_add_tail(&block_cb->driver_list, &mlx5e_block_cb_list);
+
+		return 0;
+	case FLOW_BLOCK_UNBIND:
+		indr_priv = mlx5e_rep_indr_block_priv_lookup(rpriv, netdev, f->binder_type);
+		if (!indr_priv)
+			return -ENOENT;
+
+		block_cb = flow_block_cb_lookup(f->block, setup_cb, indr_priv);
+		if (!block_cb)
+			return -ENOENT;
+
+		flow_indr_block_cb_remove(block_cb, f);
+		list_del(&block_cb->driver_list);
+		return 0;
+	default:
+		return -EOPNOTSUPP;
+	}
+	return 0;
+}
+
+static int
+mlx5e_rep_indr_replace_act(struct mlx5e_rep_priv *rpriv,
+			   struct flow_offload_action *fl_act)
+
+{
+	struct mlx5e_priv *priv = netdev_priv(rpriv->netdev);
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	enum mlx5_flow_namespace_type ns_type;
+	struct flow_action_entry *action;
+	struct mlx5e_tc_act *act;
+	bool add = false;
+	int i;
+
+	/* There is no use case currently for more than one action (e.g. pedit).
+	 * when there will be, need to handle cleaning multiple actions on err.
+	 */
+	if (!flow_offload_has_one_action(&fl_act->action))
+		return -EOPNOTSUPP;
+
+	if (esw && esw->mode == MLX5_ESWITCH_OFFLOADS)
+		ns_type = MLX5_FLOW_NAMESPACE_FDB;
+	else
+		ns_type = MLX5_FLOW_NAMESPACE_KERNEL;
+
+	flow_action_for_each(i, action, &fl_act->action) {
+		act = mlx5e_tc_act_get(action->id, ns_type);
+		if (!act)
+			continue;
+
+		if (!act->offload_action)
+			continue;
+
+		if (!act->offload_action(priv, fl_act, action))
+			add = true;
+	}
+
+	return add ? 0 : -EOPNOTSUPP;
+}
+
+static int
+mlx5e_rep_indr_destroy_act(struct mlx5e_rep_priv *rpriv,
+			   struct flow_offload_action *fl_act)
+{
+	struct mlx5e_priv *priv = netdev_priv(rpriv->netdev);
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	enum mlx5_flow_namespace_type ns_type;
+	struct mlx5e_tc_act *act;
+
+	if (esw && esw->mode == MLX5_ESWITCH_OFFLOADS)
+		ns_type = MLX5_FLOW_NAMESPACE_FDB;
+	else
+		ns_type = MLX5_FLOW_NAMESPACE_KERNEL;
+
+	act = mlx5e_tc_act_get(fl_act->id, ns_type);
+	if (!act || !act->destroy_action)
+		return -EOPNOTSUPP;
+
+	return act->destroy_action(priv, fl_act);
+}
+
+static int
+mlx5e_rep_indr_stats_act(struct mlx5e_rep_priv *rpriv,
+			 struct flow_offload_action *fl_act)
+
+{
+	struct mlx5e_priv *priv = netdev_priv(rpriv->netdev);
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	enum mlx5_flow_namespace_type ns_type;
+	struct mlx5e_tc_act *act;
+
+	if (esw && esw->mode == MLX5_ESWITCH_OFFLOADS)
+		ns_type = MLX5_FLOW_NAMESPACE_FDB;
+	else
+		ns_type = MLX5_FLOW_NAMESPACE_KERNEL;
+
+	act = mlx5e_tc_act_get(fl_act->id, ns_type);
+	if (!act || !act->stats_action)
+		return -EOPNOTSUPP;
+
+	return act->stats_action(priv, fl_act);
+}
+
+static int
+mlx5e_rep_indr_setup_act(struct mlx5e_rep_priv *rpriv,
+			 struct flow_offload_action *fl_act)
+{
+	switch (fl_act->command) {
+	case FLOW_ACT_REPLACE:
+		return mlx5e_rep_indr_replace_act(rpriv, fl_act);
+	case FLOW_ACT_DESTROY:
+		return mlx5e_rep_indr_destroy_act(rpriv, fl_act);
+	case FLOW_ACT_STATS:
+		return mlx5e_rep_indr_stats_act(rpriv, fl_act);
+	default:
+		return -EOPNOTSUPP;
+	}
+}
+
+static int
+mlx5e_rep_indr_no_dev_setup(struct mlx5e_rep_priv *rpriv,
+			    enum tc_setup_type type,
+			    void *data)
+{
+	if (!data)
+		return -EOPNOTSUPP;
+
+	switch (type) {
+	case TC_SETUP_ACT:
+		return mlx5e_rep_indr_setup_act(rpriv, data);
+	default:
+		return -EOPNOTSUPP;
+	}
+}
+
+static
+int mlx5e_rep_indr_setup_cb(struct net_device *netdev, struct Qdisc *sch, void *cb_priv,
+			    enum tc_setup_type type, void *type_data,
+			    void *data,
+			    void (*cleanup)(struct flow_block_cb *block_cb))
+{
+	if (!netdev)
+		return mlx5e_rep_indr_no_dev_setup(cb_priv, type, data);
+
+	switch (type) {
+	case TC_SETUP_BLOCK:
+		return mlx5e_rep_indr_setup_block(netdev, sch, cb_priv, type_data,
+						  mlx5e_rep_indr_setup_tc_cb,
+						  data, cleanup);
+	case TC_SETUP_FT:
+		return mlx5e_rep_indr_setup_block(netdev, sch, cb_priv, type_data,
+						  mlx5e_rep_indr_setup_ft_cb,
+						  data, cleanup);
+	default:
+		return -EOPNOTSUPP;
+	}
+}
+
+int mlx5e_rep_tc_netdevice_event_register(struct mlx5e_rep_priv *rpriv)
+{
+	struct mlx5_rep_uplink_priv *uplink_priv = &rpriv->uplink_priv;
+
+	/* init indirect block notifications */
+	INIT_LIST_HEAD(&uplink_priv->tc_indr_block_priv_list);
+
+	return flow_indr_dev_register(mlx5e_rep_indr_setup_cb, rpriv);
+}
+
+void mlx5e_rep_tc_netdevice_event_unregister(struct mlx5e_rep_priv *rpriv)
+{
+	flow_indr_dev_unregister(mlx5e_rep_indr_setup_cb, rpriv,
+				 mlx5e_rep_indr_block_unbind);
+}
+
+static bool mlx5e_restore_tunnel(struct mlx5e_priv *priv, struct sk_buff *skb,
+				 struct mlx5e_tc_update_priv *tc_priv,
+				 u32 tunnel_id)
+{
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	struct tunnel_match_enc_opts enc_opts = {};
+	struct mlx5_rep_uplink_priv *uplink_priv;
+	struct mlx5e_rep_priv *uplink_rpriv;
+	struct metadata_dst *tun_dst;
+	struct tunnel_match_key key;
+	u32 tun_id, enc_opts_id;
+	struct net_device *dev;
+	int err;
+
+	enc_opts_id = tunnel_id & ENC_OPTS_BITS_MASK;
+	tun_id = tunnel_id >> ENC_OPTS_BITS;
+
+	if (!tun_id)
+		return true;
+
+	uplink_rpriv = mlx5_eswitch_get_uplink_priv(esw, REP_ETH);
+	uplink_priv = &uplink_rpriv->uplink_priv;
+
+	err = mapping_find(uplink_priv->tunnel_mapping, tun_id, &key);
+	if (err) {
+		WARN_ON_ONCE(true);
+		netdev_dbg(priv->netdev,
+			   "Couldn't find tunnel for tun_id: %d, err: %d\n",
+			   tun_id, err);
+		return false;
+	}
+
+	if (enc_opts_id) {
+		err = mapping_find(uplink_priv->tunnel_enc_opts_mapping,
+				   enc_opts_id, &enc_opts);
+		if (err) {
+			netdev_dbg(priv->netdev,
+				   "Couldn't find tunnel (opts) for tun_id: %d, err: %d\n",
+				   enc_opts_id, err);
+			return false;
+		}
+	}
+
+	if (key.enc_control.addr_type == FLOW_DISSECTOR_KEY_IPV4_ADDRS) {
+		tun_dst = __ip_tun_set_dst(key.enc_ipv4.src, key.enc_ipv4.dst,
+					   key.enc_ip.tos, key.enc_ip.ttl,
+					   key.enc_tp.dst, TUNNEL_KEY,
+					   key32_to_tunnel_id(key.enc_key_id.keyid),
+					   enc_opts.key.len);
+	} else if (key.enc_control.addr_type == FLOW_DISSECTOR_KEY_IPV6_ADDRS) {
+		tun_dst = __ipv6_tun_set_dst(&key.enc_ipv6.src, &key.enc_ipv6.dst,
+					     key.enc_ip.tos, key.enc_ip.ttl,
+					     key.enc_tp.dst, 0, TUNNEL_KEY,
+					     key32_to_tunnel_id(key.enc_key_id.keyid),
+					     enc_opts.key.len);
+	} else {
+		netdev_dbg(priv->netdev,
+			   "Couldn't restore tunnel, unsupported addr_type: %d\n",
+			   key.enc_control.addr_type);
+		return false;
+	}
+
+	if (!tun_dst) {
+		netdev_dbg(priv->netdev, "Couldn't restore tunnel, no tun_dst\n");
+		return false;
+	}
+
+	tun_dst->u.tun_info.key.tp_src = key.enc_tp.src;
+
+	if (enc_opts.key.len)
+		ip_tunnel_info_opts_set(&tun_dst->u.tun_info,
+					enc_opts.key.data,
+					enc_opts.key.len,
+					enc_opts.key.dst_opt_type);
+
+	skb_dst_set(skb, (struct dst_entry *)tun_dst);
+	dev = dev_get_by_index(&init_net, key.filter_ifindex);
+	if (!dev) {
+		netdev_dbg(priv->netdev,
+			   "Couldn't find tunnel device with ifindex: %d\n",
+			   key.filter_ifindex);
+		return false;
+	}
+
+	/* Set fwd_dev so we do dev_put() after datapath */
+	tc_priv->fwd_dev = dev;
+
+	skb->dev = dev;
+
+	return true;
+}
+
+static bool mlx5e_restore_skb_chain(struct sk_buff *skb, u32 chain, u32 reg_c1,
+				    struct mlx5e_tc_update_priv *tc_priv)
+{
+	struct mlx5e_priv *priv = netdev_priv(skb->dev);
+	u32 tunnel_id = (reg_c1 >> ESW_TUN_OFFSET) & TUNNEL_ID_MASK;
+
+#if IS_ENABLED(CONFIG_NET_TC_SKB_EXT)
+	if (chain) {
+		struct mlx5_rep_uplink_priv *uplink_priv;
+		struct mlx5e_rep_priv *uplink_rpriv;
+		struct tc_skb_ext *tc_skb_ext;
+		struct mlx5_eswitch *esw;
+		u32 zone_restore_id;
+
+		tc_skb_ext = tc_skb_ext_alloc(skb);
+		if (!tc_skb_ext) {
+			WARN_ON(1);
+			return false;
+		}
+		tc_skb_ext->chain = chain;
+		zone_restore_id = reg_c1 & ESW_ZONE_ID_MASK;
+		esw = priv->mdev->priv.eswitch;
+		uplink_rpriv = mlx5_eswitch_get_uplink_priv(esw, REP_ETH);
+		uplink_priv = &uplink_rpriv->uplink_priv;
+		if (!mlx5e_tc_ct_restore_flow(uplink_priv->ct_priv, skb,
+					      zone_restore_id))
+			return false;
+	}
+#endif /* CONFIG_NET_TC_SKB_EXT */
+
+	return mlx5e_restore_tunnel(priv, skb, tc_priv, tunnel_id);
+}
+
+static void mlx5_rep_tc_post_napi_receive(struct mlx5e_tc_update_priv *tc_priv)
+{
+	if (tc_priv->fwd_dev)
+		dev_put(tc_priv->fwd_dev);
+}
+
+static void mlx5e_restore_skb_sample(struct mlx5e_priv *priv, struct sk_buff *skb,
+				     struct mlx5_mapped_obj *mapped_obj,
+				     struct mlx5e_tc_update_priv *tc_priv)
+{
+	if (!mlx5e_restore_tunnel(priv, skb, tc_priv, mapped_obj->sample.tunnel_id)) {
+		netdev_dbg(priv->netdev,
+			   "Failed to restore tunnel info for sampled packet\n");
+		return;
+	}
+	mlx5e_tc_sample_skb(skb, mapped_obj);
+	mlx5_rep_tc_post_napi_receive(tc_priv);
+}
+
+static bool mlx5e_restore_skb_int_port(struct mlx5e_priv *priv, struct sk_buff *skb,
+				       struct mlx5_mapped_obj *mapped_obj,
+				       struct mlx5e_tc_update_priv *tc_priv,
+				       bool *forward_tx,
+				       u32 reg_c1)
+{
+	u32 tunnel_id = (reg_c1 >> ESW_TUN_OFFSET) & TUNNEL_ID_MASK;
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	struct mlx5_rep_uplink_priv *uplink_priv;
+	struct mlx5e_rep_priv *uplink_rpriv;
+
+	/* Tunnel restore takes precedence over int port restore */
+	if (tunnel_id)
+		return mlx5e_restore_tunnel(priv, skb, tc_priv, tunnel_id);
+
+	uplink_rpriv = mlx5_eswitch_get_uplink_priv(esw, REP_ETH);
+	uplink_priv = &uplink_rpriv->uplink_priv;
+
+	if (mlx5e_tc_int_port_dev_fwd(uplink_priv->int_port_priv, skb,
+				      mapped_obj->int_port_metadata, forward_tx)) {
+		/* Set fwd_dev for future dev_put */
+		tc_priv->fwd_dev = skb->dev;
+
+		return true;
+	}
+
+	return false;
+}
+
+void mlx5e_rep_tc_receive(struct mlx5_cqe64 *cqe, struct mlx5e_rq *rq,
+			  struct sk_buff *skb)
+{
+	u32 reg_c1 = be32_to_cpu(cqe->ft_metadata);
+	struct mlx5e_tc_update_priv tc_priv = {};
+	struct mlx5_mapped_obj mapped_obj;
+	struct mlx5_eswitch *esw;
+	bool forward_tx = false;
+	struct mlx5e_priv *priv;
+	u32 reg_c0;
+	int err;
+
+	reg_c0 = (be32_to_cpu(cqe->sop_drop_qpn) & MLX5E_TC_FLOW_ID_MASK);
+	if (!reg_c0 || reg_c0 == MLX5_FS_DEFAULT_FLOW_TAG)
+		goto forward;
+
+	/* If reg_c0 is not equal to the default flow tag then skb->mark
+	 * is not supported and must be reset back to 0.
+	 */
+	skb->mark = 0;
+
+	priv = netdev_priv(skb->dev);
+	esw = priv->mdev->priv.eswitch;
+	err = mapping_find(esw->offloads.reg_c0_obj_pool, reg_c0, &mapped_obj);
+	if (err) {
+		netdev_dbg(priv->netdev,
+			   "Couldn't find mapped object for reg_c0: %d, err: %d\n",
+			   reg_c0, err);
+		goto free_skb;
+	}
+
+	if (mapped_obj.type == MLX5_MAPPED_OBJ_CHAIN) {
+		if (!mlx5e_restore_skb_chain(skb, mapped_obj.chain, reg_c1, &tc_priv) &&
+		    !mlx5_ipsec_is_rx_flow(cqe))
+			goto free_skb;
+	} else if (mapped_obj.type == MLX5_MAPPED_OBJ_SAMPLE) {
+		mlx5e_restore_skb_sample(priv, skb, &mapped_obj, &tc_priv);
+		goto free_skb;
+	} else if (mapped_obj.type == MLX5_MAPPED_OBJ_INT_PORT_METADATA) {
+		if (!mlx5e_restore_skb_int_port(priv, skb, &mapped_obj, &tc_priv,
+						&forward_tx, reg_c1))
+			goto free_skb;
+	} else {
+		netdev_dbg(priv->netdev, "Invalid mapped object type: %d\n", mapped_obj.type);
+		goto free_skb;
+	}
+
+forward:
+	if (forward_tx)
+		dev_queue_xmit(skb);
+	else
+		napi_gro_receive(rq->cq.napi, skb);
+
+	mlx5_rep_tc_post_napi_receive(&tc_priv);
+
+	return;
+
+free_skb:
+	dev_kfree_skb_any(skb);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/rep/tc.h b/drivers/net/ethernet/mellanox/mlx5/core/en/rep/tc.h
new file mode 100644
index 000000000..7c9dd3a75
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/rep/tc.h
@@ -0,0 +1,73 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2020 Mellanox Technologies. */
+
+#ifndef __MLX5_EN_REP_TC_H__
+#define __MLX5_EN_REP_TC_H__
+
+#include <linux/skbuff.h>
+#include "en_tc.h"
+#include "en_rep.h"
+
+#if IS_ENABLED(CONFIG_MLX5_CLS_ACT)
+
+int mlx5e_rep_tc_init(struct mlx5e_rep_priv *rpriv);
+void mlx5e_rep_tc_cleanup(struct mlx5e_rep_priv *rpriv);
+
+int mlx5e_rep_tc_netdevice_event_register(struct mlx5e_rep_priv *rpriv);
+void mlx5e_rep_tc_netdevice_event_unregister(struct mlx5e_rep_priv *rpriv);
+
+void mlx5e_rep_tc_enable(struct mlx5e_priv *priv);
+void mlx5e_rep_tc_disable(struct mlx5e_priv *priv);
+
+int mlx5e_rep_tc_event_port_affinity(struct mlx5e_priv *priv);
+
+void mlx5e_rep_update_flows(struct mlx5e_priv *priv,
+			    struct mlx5e_encap_entry *e,
+			    bool neigh_connected,
+			    unsigned char ha[ETH_ALEN]);
+
+int mlx5e_rep_encap_entry_attach(struct mlx5e_priv *priv,
+				 struct mlx5e_encap_entry *e,
+				 struct mlx5e_neigh *m_neigh,
+				 struct net_device *neigh_dev);
+void mlx5e_rep_encap_entry_detach(struct mlx5e_priv *priv,
+				  struct mlx5e_encap_entry *e);
+
+int mlx5e_rep_setup_tc(struct net_device *dev, enum tc_setup_type type,
+		       void *type_data);
+
+void mlx5e_rep_tc_receive(struct mlx5_cqe64 *cqe, struct mlx5e_rq *rq,
+			  struct sk_buff *skb);
+
+#else /* CONFIG_MLX5_CLS_ACT */
+
+struct mlx5e_rep_priv;
+static inline int
+mlx5e_rep_tc_init(struct mlx5e_rep_priv *rpriv) { return 0; }
+static inline void
+mlx5e_rep_tc_cleanup(struct mlx5e_rep_priv *rpriv) {}
+
+static inline int
+mlx5e_rep_tc_netdevice_event_register(struct mlx5e_rep_priv *rpriv) { return 0; }
+static inline void
+mlx5e_rep_tc_netdevice_event_unregister(struct mlx5e_rep_priv *rpriv) {}
+
+static inline void
+mlx5e_rep_tc_enable(struct mlx5e_priv *priv) {}
+static inline void
+mlx5e_rep_tc_disable(struct mlx5e_priv *priv) {}
+
+static inline int
+mlx5e_rep_tc_event_port_affinity(struct mlx5e_priv *priv) { return NOTIFY_DONE; }
+
+static inline int
+mlx5e_rep_setup_tc(struct net_device *dev, enum tc_setup_type type,
+		   void *type_data) { return -EOPNOTSUPP; }
+
+static inline void
+mlx5e_rep_tc_receive(struct mlx5_cqe64 *cqe, struct mlx5e_rq *rq,
+		     struct sk_buff *skb) { napi_gro_receive(rq->cq.napi, skb); }
+
+#endif /* CONFIG_MLX5_CLS_ACT */
+
+#endif /* __MLX5_EN_REP_TC_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/reporter_rx.c b/drivers/net/ethernet/mellanox/mlx5/core/en/reporter_rx.c
new file mode 100644
index 000000000..9b1f1369a
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/reporter_rx.c
@@ -0,0 +1,759 @@
+// SPDX-License-Identifier: GPL-2.0
+// Copyright (c) 2019 Mellanox Technologies.
+
+#include "health.h"
+#include "params.h"
+#include "txrx.h"
+#include "devlink.h"
+#include "ptp.h"
+#include "lib/tout.h"
+
+static int mlx5e_query_rq_state(struct mlx5_core_dev *dev, u32 rqn, u8 *state)
+{
+	int outlen = MLX5_ST_SZ_BYTES(query_rq_out);
+	void *out;
+	void *rqc;
+	int err;
+
+	out = kvzalloc(outlen, GFP_KERNEL);
+	if (!out)
+		return -ENOMEM;
+
+	err = mlx5_core_query_rq(dev, rqn, out);
+	if (err)
+		goto out;
+
+	rqc = MLX5_ADDR_OF(query_rq_out, out, rq_context);
+	*state = MLX5_GET(rqc, rqc, state);
+
+out:
+	kvfree(out);
+	return err;
+}
+
+static int mlx5e_wait_for_icosq_flush(struct mlx5e_icosq *icosq)
+{
+	struct mlx5_core_dev *dev = icosq->channel->mdev;
+	unsigned long exp_time;
+
+	exp_time = jiffies + msecs_to_jiffies(mlx5_tout_ms(dev, FLUSH_ON_ERROR));
+
+	while (time_before(jiffies, exp_time)) {
+		if (icosq->cc == icosq->pc)
+			return 0;
+
+		msleep(20);
+	}
+
+	netdev_err(icosq->channel->netdev,
+		   "Wait for ICOSQ 0x%x flush timeout (cc = 0x%x, pc = 0x%x)\n",
+		   icosq->sqn, icosq->cc, icosq->pc);
+
+	return -ETIMEDOUT;
+}
+
+static void mlx5e_reset_icosq_cc_pc(struct mlx5e_icosq *icosq)
+{
+	WARN_ONCE(icosq->cc != icosq->pc, "ICOSQ 0x%x: cc (0x%x) != pc (0x%x)\n",
+		  icosq->sqn, icosq->cc, icosq->pc);
+	icosq->cc = 0;
+	icosq->pc = 0;
+}
+
+static int mlx5e_rx_reporter_err_icosq_cqe_recover(void *ctx)
+{
+	struct mlx5e_rq *xskrq = NULL;
+	struct mlx5_core_dev *mdev;
+	struct mlx5e_icosq *icosq;
+	struct net_device *dev;
+	struct mlx5e_rq *rq;
+	u8 state;
+	int err;
+
+	icosq = ctx;
+
+	mutex_lock(&icosq->channel->icosq_recovery_lock);
+
+	/* mlx5e_close_rq cancels this work before RQ and ICOSQ are killed. */
+	rq = &icosq->channel->rq;
+	if (test_bit(MLX5E_RQ_STATE_ENABLED, &icosq->channel->xskrq.state))
+		xskrq = &icosq->channel->xskrq;
+	mdev = icosq->channel->mdev;
+	dev = icosq->channel->netdev;
+	err = mlx5_core_query_sq_state(mdev, icosq->sqn, &state);
+	if (err) {
+		netdev_err(dev, "Failed to query ICOSQ 0x%x state. err = %d\n",
+			   icosq->sqn, err);
+		goto out;
+	}
+
+	if (state != MLX5_SQC_STATE_ERR)
+		goto out;
+
+	mlx5e_deactivate_rq(rq);
+	if (xskrq)
+		mlx5e_deactivate_rq(xskrq);
+
+	err = mlx5e_wait_for_icosq_flush(icosq);
+	if (err)
+		goto out;
+
+	mlx5e_deactivate_icosq(icosq);
+
+	/* At this point, both the rq and the icosq are disabled */
+
+	err = mlx5e_health_sq_to_ready(mdev, dev, icosq->sqn);
+	if (err)
+		goto out;
+
+	mlx5e_reset_icosq_cc_pc(icosq);
+
+	mlx5e_free_rx_in_progress_descs(rq);
+	if (xskrq)
+		mlx5e_free_rx_in_progress_descs(xskrq);
+
+	clear_bit(MLX5E_SQ_STATE_RECOVERING, &icosq->state);
+	mlx5e_activate_icosq(icosq);
+
+	mlx5e_activate_rq(rq);
+	rq->stats->recover++;
+
+	if (xskrq) {
+		mlx5e_activate_rq(xskrq);
+		xskrq->stats->recover++;
+	}
+
+	mlx5e_trigger_napi_icosq(icosq->channel);
+
+	mutex_unlock(&icosq->channel->icosq_recovery_lock);
+
+	return 0;
+out:
+	clear_bit(MLX5E_SQ_STATE_RECOVERING, &icosq->state);
+	mutex_unlock(&icosq->channel->icosq_recovery_lock);
+	return err;
+}
+
+static int mlx5e_rx_reporter_err_rq_cqe_recover(void *ctx)
+{
+	struct mlx5e_rq *rq = ctx;
+	int err;
+
+	mlx5e_deactivate_rq(rq);
+	err = mlx5e_flush_rq(rq, MLX5_RQC_STATE_ERR);
+	clear_bit(MLX5E_RQ_STATE_RECOVERING, &rq->state);
+	if (err)
+		return err;
+
+	mlx5e_activate_rq(rq);
+	rq->stats->recover++;
+	if (rq->channel)
+		mlx5e_trigger_napi_icosq(rq->channel);
+	else
+		mlx5e_trigger_napi_sched(rq->cq.napi);
+	return 0;
+}
+
+static int mlx5e_rx_reporter_timeout_recover(void *ctx)
+{
+	struct mlx5_eq_comp *eq;
+	struct mlx5e_rq *rq;
+	int err;
+
+	rq = ctx;
+	eq = rq->cq.mcq.eq;
+
+	err = mlx5e_health_channel_eq_recover(rq->netdev, eq, rq->cq.ch_stats);
+	if (err && rq->icosq)
+		clear_bit(MLX5E_SQ_STATE_ENABLED, &rq->icosq->state);
+
+	return err;
+}
+
+static int mlx5e_rx_reporter_recover_from_ctx(struct mlx5e_err_ctx *err_ctx)
+{
+	return err_ctx->recover(err_ctx->ctx);
+}
+
+static int mlx5e_rx_reporter_recover(struct devlink_health_reporter *reporter,
+				     void *context,
+				     struct netlink_ext_ack *extack)
+{
+	struct mlx5e_priv *priv = devlink_health_reporter_priv(reporter);
+	struct mlx5e_err_ctx *err_ctx = context;
+
+	return err_ctx ? mlx5e_rx_reporter_recover_from_ctx(err_ctx) :
+			 mlx5e_health_recover_channels(priv);
+}
+
+static int mlx5e_reporter_icosq_diagnose(struct mlx5e_icosq *icosq, u8 hw_state,
+					 struct devlink_fmsg *fmsg)
+{
+	int err;
+
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, "ICOSQ");
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u32_pair_put(fmsg, "sqn", icosq->sqn);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u8_pair_put(fmsg, "HW state", hw_state);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u32_pair_put(fmsg, "cc", icosq->cc);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u32_pair_put(fmsg, "pc", icosq->pc);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u32_pair_put(fmsg, "WQE size",
+					mlx5_wq_cyc_get_size(&icosq->wq));
+	if (err)
+		return err;
+
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, "CQ");
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u32_pair_put(fmsg, "cqn", icosq->cq.mcq.cqn);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u32_pair_put(fmsg, "cc", icosq->cq.wq.cc);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u32_pair_put(fmsg, "size", mlx5_cqwq_get_size(&icosq->cq.wq));
+	if (err)
+		return err;
+
+	err = mlx5e_health_fmsg_named_obj_nest_end(fmsg);
+	if (err)
+		return err;
+
+	return mlx5e_health_fmsg_named_obj_nest_end(fmsg);
+}
+
+static int
+mlx5e_rx_reporter_build_diagnose_output_rq_common(struct mlx5e_rq *rq,
+						  struct devlink_fmsg *fmsg)
+{
+	u16 wqe_counter;
+	int wqes_sz;
+	u8 hw_state;
+	u16 wq_head;
+	int err;
+
+	err = mlx5e_query_rq_state(rq->mdev, rq->rqn, &hw_state);
+	if (err)
+		return err;
+
+	wqes_sz = mlx5e_rqwq_get_cur_sz(rq);
+	wq_head = mlx5e_rqwq_get_head(rq);
+	wqe_counter = mlx5e_rqwq_get_wqe_counter(rq);
+
+	err = devlink_fmsg_u32_pair_put(fmsg, "rqn", rq->rqn);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u8_pair_put(fmsg, "HW state", hw_state);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u8_pair_put(fmsg, "SW state", rq->state);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u32_pair_put(fmsg, "WQE counter", wqe_counter);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u32_pair_put(fmsg, "posted WQEs", wqes_sz);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u32_pair_put(fmsg, "cc", wq_head);
+	if (err)
+		return err;
+
+	err = mlx5e_health_cq_diag_fmsg(&rq->cq, fmsg);
+	if (err)
+		return err;
+
+	err = mlx5e_health_eq_diag_fmsg(rq->cq.mcq.eq, fmsg);
+	if (err)
+		return err;
+
+	if (rq->icosq) {
+		struct mlx5e_icosq *icosq = rq->icosq;
+		u8 icosq_hw_state;
+
+		err = mlx5_core_query_sq_state(rq->mdev, icosq->sqn, &icosq_hw_state);
+		if (err)
+			return err;
+
+		err = mlx5e_reporter_icosq_diagnose(icosq, icosq_hw_state, fmsg);
+		if (err)
+			return err;
+	}
+
+	return 0;
+}
+
+static int mlx5e_rx_reporter_build_diagnose_output(struct mlx5e_rq *rq,
+						   struct devlink_fmsg *fmsg)
+{
+	int err;
+
+	err = devlink_fmsg_obj_nest_start(fmsg);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u32_pair_put(fmsg, "channel ix", rq->ix);
+	if (err)
+		return err;
+
+	err = mlx5e_rx_reporter_build_diagnose_output_rq_common(rq, fmsg);
+	if (err)
+		return err;
+
+	return devlink_fmsg_obj_nest_end(fmsg);
+}
+
+static int mlx5e_rx_reporter_diagnose_generic_rq(struct mlx5e_rq *rq,
+						 struct devlink_fmsg *fmsg)
+{
+	struct mlx5e_priv *priv = rq->priv;
+	struct mlx5e_params *params;
+	u32 rq_stride, rq_sz;
+	bool real_time;
+	int err;
+
+	params = &priv->channels.params;
+	rq_sz = mlx5e_rqwq_get_size(rq);
+	real_time =  mlx5_is_real_time_rq(priv->mdev);
+	rq_stride = BIT(mlx5e_mpwqe_get_log_stride_size(priv->mdev, params, NULL));
+
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, "RQ");
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u8_pair_put(fmsg, "type", params->rq_wq_type);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u64_pair_put(fmsg, "stride size", rq_stride);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u32_pair_put(fmsg, "size", rq_sz);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_string_pair_put(fmsg, "ts_format", real_time ? "RT" : "FRC");
+	if (err)
+		return err;
+
+	err = mlx5e_health_cq_common_diag_fmsg(&rq->cq, fmsg);
+	if (err)
+		return err;
+
+	return mlx5e_health_fmsg_named_obj_nest_end(fmsg);
+}
+
+static int
+mlx5e_rx_reporter_diagnose_common_ptp_config(struct mlx5e_priv *priv, struct mlx5e_ptp *ptp_ch,
+					     struct devlink_fmsg *fmsg)
+{
+	int err;
+
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, "PTP");
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u32_pair_put(fmsg, "filter_type", priv->tstamp.rx_filter);
+	if (err)
+		return err;
+
+	err = mlx5e_rx_reporter_diagnose_generic_rq(&ptp_ch->rq, fmsg);
+	if (err)
+		return err;
+
+	return mlx5e_health_fmsg_named_obj_nest_end(fmsg);
+}
+
+static int
+mlx5e_rx_reporter_diagnose_common_config(struct devlink_health_reporter *reporter,
+					 struct devlink_fmsg *fmsg)
+{
+	struct mlx5e_priv *priv = devlink_health_reporter_priv(reporter);
+	struct mlx5e_rq *generic_rq = &priv->channels.c[0]->rq;
+	struct mlx5e_ptp *ptp_ch = priv->channels.ptp;
+	int err;
+
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, "Common config");
+	if (err)
+		return err;
+
+	err = mlx5e_rx_reporter_diagnose_generic_rq(generic_rq, fmsg);
+	if (err)
+		return err;
+
+	if (ptp_ch && test_bit(MLX5E_PTP_STATE_RX, ptp_ch->state)) {
+		err = mlx5e_rx_reporter_diagnose_common_ptp_config(priv, ptp_ch, fmsg);
+		if (err)
+			return err;
+	}
+
+	return mlx5e_health_fmsg_named_obj_nest_end(fmsg);
+}
+
+static int mlx5e_rx_reporter_build_diagnose_output_ptp_rq(struct mlx5e_rq *rq,
+							  struct devlink_fmsg *fmsg)
+{
+	int err;
+
+	err = devlink_fmsg_obj_nest_start(fmsg);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_string_pair_put(fmsg, "channel", "ptp");
+	if (err)
+		return err;
+
+	err = mlx5e_rx_reporter_build_diagnose_output_rq_common(rq, fmsg);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_obj_nest_end(fmsg);
+	if (err)
+		return err;
+
+	return 0;
+}
+
+static int mlx5e_rx_reporter_diagnose(struct devlink_health_reporter *reporter,
+				      struct devlink_fmsg *fmsg,
+				      struct netlink_ext_ack *extack)
+{
+	struct mlx5e_priv *priv = devlink_health_reporter_priv(reporter);
+	struct mlx5e_ptp *ptp_ch = priv->channels.ptp;
+	int i, err = 0;
+
+	mutex_lock(&priv->state_lock);
+
+	if (!test_bit(MLX5E_STATE_OPENED, &priv->state))
+		goto unlock;
+
+	err = mlx5e_rx_reporter_diagnose_common_config(reporter, fmsg);
+	if (err)
+		goto unlock;
+
+	err = devlink_fmsg_arr_pair_nest_start(fmsg, "RQs");
+	if (err)
+		goto unlock;
+
+	for (i = 0; i < priv->channels.num; i++) {
+		struct mlx5e_channel *c = priv->channels.c[i];
+		struct mlx5e_rq *rq;
+
+		rq = test_bit(MLX5E_CHANNEL_STATE_XSK, c->state) ?
+			&c->xskrq : &c->rq;
+
+		err = mlx5e_rx_reporter_build_diagnose_output(rq, fmsg);
+		if (err)
+			goto unlock;
+	}
+	if (ptp_ch && test_bit(MLX5E_PTP_STATE_RX, ptp_ch->state)) {
+		err = mlx5e_rx_reporter_build_diagnose_output_ptp_rq(&ptp_ch->rq, fmsg);
+		if (err)
+			goto unlock;
+	}
+	err = devlink_fmsg_arr_pair_nest_end(fmsg);
+unlock:
+	mutex_unlock(&priv->state_lock);
+	return err;
+}
+
+static int mlx5e_rx_reporter_dump_icosq(struct mlx5e_priv *priv, struct devlink_fmsg *fmsg,
+					void *ctx)
+{
+	struct mlx5e_txqsq *icosq = ctx;
+	struct mlx5_rsc_key key = {};
+	int err;
+
+	if (!test_bit(MLX5E_STATE_OPENED, &priv->state))
+		return 0;
+
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, "SX Slice");
+	if (err)
+		return err;
+
+	key.size = PAGE_SIZE;
+	key.rsc = MLX5_SGMT_TYPE_SX_SLICE_ALL;
+	err = mlx5e_health_rsc_fmsg_dump(priv, &key, fmsg);
+	if (err)
+		return err;
+
+	err = mlx5e_health_fmsg_named_obj_nest_end(fmsg);
+	if (err)
+		return err;
+
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, "ICOSQ");
+	if (err)
+		return err;
+
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, "QPC");
+	if (err)
+		return err;
+
+	key.rsc = MLX5_SGMT_TYPE_FULL_QPC;
+	key.index1 = icosq->sqn;
+	key.num_of_obj1 = 1;
+
+	err = mlx5e_health_rsc_fmsg_dump(priv, &key, fmsg);
+	if (err)
+		return err;
+
+	err = mlx5e_health_fmsg_named_obj_nest_end(fmsg);
+	if (err)
+		return err;
+
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, "send_buff");
+	if (err)
+		return err;
+
+	key.rsc = MLX5_SGMT_TYPE_SND_BUFF;
+	key.num_of_obj2 = MLX5_RSC_DUMP_ALL;
+
+	err = mlx5e_health_rsc_fmsg_dump(priv, &key, fmsg);
+	if (err)
+		return err;
+
+	err = mlx5e_health_fmsg_named_obj_nest_end(fmsg);
+	if (err)
+		return err;
+
+	return mlx5e_health_fmsg_named_obj_nest_end(fmsg);
+}
+
+static int mlx5e_rx_reporter_dump_rq(struct mlx5e_priv *priv, struct devlink_fmsg *fmsg,
+				     void *ctx)
+{
+	struct mlx5_rsc_key key = {};
+	struct mlx5e_rq *rq = ctx;
+	int err;
+
+	if (!test_bit(MLX5E_STATE_OPENED, &priv->state))
+		return 0;
+
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, "RX Slice");
+	if (err)
+		return err;
+
+	key.size = PAGE_SIZE;
+	key.rsc = MLX5_SGMT_TYPE_RX_SLICE_ALL;
+	err = mlx5e_health_rsc_fmsg_dump(priv, &key, fmsg);
+	if (err)
+		return err;
+
+	err = mlx5e_health_fmsg_named_obj_nest_end(fmsg);
+	if (err)
+		return err;
+
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, "RQ");
+	if (err)
+		return err;
+
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, "QPC");
+	if (err)
+		return err;
+
+	key.rsc = MLX5_SGMT_TYPE_FULL_QPC;
+	key.index1 = rq->rqn;
+	key.num_of_obj1 = 1;
+
+	err = mlx5e_health_rsc_fmsg_dump(priv, &key, fmsg);
+	if (err)
+		return err;
+
+	err = mlx5e_health_fmsg_named_obj_nest_end(fmsg);
+	if (err)
+		return err;
+
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, "receive_buff");
+	if (err)
+		return err;
+
+	key.rsc = MLX5_SGMT_TYPE_RCV_BUFF;
+	key.num_of_obj2 = MLX5_RSC_DUMP_ALL;
+	err = mlx5e_health_rsc_fmsg_dump(priv, &key, fmsg);
+	if (err)
+		return err;
+
+	err = mlx5e_health_fmsg_named_obj_nest_end(fmsg);
+	if (err)
+		return err;
+
+	return mlx5e_health_fmsg_named_obj_nest_end(fmsg);
+}
+
+static int mlx5e_rx_reporter_dump_all_rqs(struct mlx5e_priv *priv,
+					  struct devlink_fmsg *fmsg)
+{
+	struct mlx5e_ptp *ptp_ch = priv->channels.ptp;
+	struct mlx5_rsc_key key = {};
+	int i, err;
+
+	if (!test_bit(MLX5E_STATE_OPENED, &priv->state))
+		return 0;
+
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, "RX Slice");
+	if (err)
+		return err;
+
+	key.size = PAGE_SIZE;
+	key.rsc = MLX5_SGMT_TYPE_RX_SLICE_ALL;
+	err = mlx5e_health_rsc_fmsg_dump(priv, &key, fmsg);
+	if (err)
+		return err;
+
+	err = mlx5e_health_fmsg_named_obj_nest_end(fmsg);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_arr_pair_nest_start(fmsg, "RQs");
+	if (err)
+		return err;
+
+	for (i = 0; i < priv->channels.num; i++) {
+		struct mlx5e_rq *rq = &priv->channels.c[i]->rq;
+
+		err = mlx5e_health_queue_dump(priv, fmsg, rq->rqn, "RQ");
+		if (err)
+			return err;
+	}
+
+	if (ptp_ch && test_bit(MLX5E_PTP_STATE_RX, ptp_ch->state)) {
+		err = mlx5e_health_queue_dump(priv, fmsg, ptp_ch->rq.rqn, "PTP RQ");
+		if (err)
+			return err;
+	}
+
+	return devlink_fmsg_arr_pair_nest_end(fmsg);
+}
+
+static int mlx5e_rx_reporter_dump_from_ctx(struct mlx5e_priv *priv,
+					   struct mlx5e_err_ctx *err_ctx,
+					   struct devlink_fmsg *fmsg)
+{
+	return err_ctx->dump(priv, fmsg, err_ctx->ctx);
+}
+
+static int mlx5e_rx_reporter_dump(struct devlink_health_reporter *reporter,
+				  struct devlink_fmsg *fmsg, void *context,
+				  struct netlink_ext_ack *extack)
+{
+	struct mlx5e_priv *priv = devlink_health_reporter_priv(reporter);
+	struct mlx5e_err_ctx *err_ctx = context;
+
+	return err_ctx ? mlx5e_rx_reporter_dump_from_ctx(priv, err_ctx, fmsg) :
+			 mlx5e_rx_reporter_dump_all_rqs(priv, fmsg);
+}
+
+void mlx5e_reporter_rx_timeout(struct mlx5e_rq *rq)
+{
+	char err_str[MLX5E_REPORTER_PER_Q_MAX_LEN];
+	struct mlx5e_icosq *icosq = rq->icosq;
+	struct mlx5e_priv *priv = rq->priv;
+	struct mlx5e_err_ctx err_ctx = {};
+	char icosq_str[32] = {};
+
+	err_ctx.ctx = rq;
+	err_ctx.recover = mlx5e_rx_reporter_timeout_recover;
+	err_ctx.dump = mlx5e_rx_reporter_dump_rq;
+
+	if (icosq)
+		snprintf(icosq_str, sizeof(icosq_str), "ICOSQ: 0x%x, ", icosq->sqn);
+	snprintf(err_str, sizeof(err_str),
+		 "RX timeout on channel: %d, %s RQ: 0x%x, CQ: 0x%x",
+		 rq->ix, icosq_str, rq->rqn, rq->cq.mcq.cqn);
+
+	mlx5e_health_report(priv, priv->rx_reporter, err_str, &err_ctx);
+}
+
+void mlx5e_reporter_rq_cqe_err(struct mlx5e_rq *rq)
+{
+	char err_str[MLX5E_REPORTER_PER_Q_MAX_LEN];
+	struct mlx5e_priv *priv = rq->priv;
+	struct mlx5e_err_ctx err_ctx = {};
+
+	err_ctx.ctx = rq;
+	err_ctx.recover = mlx5e_rx_reporter_err_rq_cqe_recover;
+	err_ctx.dump = mlx5e_rx_reporter_dump_rq;
+	snprintf(err_str, sizeof(err_str), "ERR CQE on RQ: 0x%x", rq->rqn);
+
+	mlx5e_health_report(priv, priv->rx_reporter, err_str, &err_ctx);
+}
+
+void mlx5e_reporter_icosq_cqe_err(struct mlx5e_icosq *icosq)
+{
+	struct mlx5e_priv *priv = icosq->channel->priv;
+	char err_str[MLX5E_REPORTER_PER_Q_MAX_LEN];
+	struct mlx5e_err_ctx err_ctx = {};
+
+	err_ctx.ctx = icosq;
+	err_ctx.recover = mlx5e_rx_reporter_err_icosq_cqe_recover;
+	err_ctx.dump = mlx5e_rx_reporter_dump_icosq;
+	snprintf(err_str, sizeof(err_str), "ERR CQE on ICOSQ: 0x%x", icosq->sqn);
+
+	mlx5e_health_report(priv, priv->rx_reporter, err_str, &err_ctx);
+}
+
+void mlx5e_reporter_icosq_suspend_recovery(struct mlx5e_channel *c)
+{
+	mutex_lock(&c->icosq_recovery_lock);
+}
+
+void mlx5e_reporter_icosq_resume_recovery(struct mlx5e_channel *c)
+{
+	mutex_unlock(&c->icosq_recovery_lock);
+}
+
+static const struct devlink_health_reporter_ops mlx5_rx_reporter_ops = {
+	.name = "rx",
+	.recover = mlx5e_rx_reporter_recover,
+	.diagnose = mlx5e_rx_reporter_diagnose,
+	.dump = mlx5e_rx_reporter_dump,
+};
+
+#define MLX5E_REPORTER_RX_GRACEFUL_PERIOD 500
+
+void mlx5e_reporter_rx_create(struct mlx5e_priv *priv)
+{
+	struct devlink_port *dl_port = mlx5e_devlink_get_dl_port(priv);
+	struct devlink_health_reporter *reporter;
+
+	reporter = devlink_port_health_reporter_create(dl_port, &mlx5_rx_reporter_ops,
+						       MLX5E_REPORTER_RX_GRACEFUL_PERIOD, priv);
+	if (IS_ERR(reporter)) {
+		netdev_warn(priv->netdev, "Failed to create rx reporter, err = %ld\n",
+			    PTR_ERR(reporter));
+		return;
+	}
+	priv->rx_reporter = reporter;
+}
+
+void mlx5e_reporter_rx_destroy(struct mlx5e_priv *priv)
+{
+	if (!priv->rx_reporter)
+		return;
+
+	devlink_port_health_reporter_destroy(priv->rx_reporter);
+	priv->rx_reporter = NULL;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/reporter_tx.c b/drivers/net/ethernet/mellanox/mlx5/core/en/reporter_tx.c
new file mode 100644
index 000000000..60bc5b577
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/reporter_tx.c
@@ -0,0 +1,614 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/* Copyright (c) 2019 Mellanox Technologies. */
+
+#include "health.h"
+#include "en/ptp.h"
+#include "en/devlink.h"
+#include "lib/tout.h"
+
+static int mlx5e_wait_for_sq_flush(struct mlx5e_txqsq *sq)
+{
+	struct mlx5_core_dev *dev = sq->mdev;
+	unsigned long exp_time;
+
+	exp_time = jiffies + msecs_to_jiffies(mlx5_tout_ms(dev, FLUSH_ON_ERROR));
+
+	while (time_before(jiffies, exp_time)) {
+		if (sq->cc == sq->pc)
+			return 0;
+
+		msleep(20);
+	}
+
+	netdev_err(sq->netdev,
+		   "Wait for SQ 0x%x flush timeout (sq cc = 0x%x, sq pc = 0x%x)\n",
+		   sq->sqn, sq->cc, sq->pc);
+
+	return -ETIMEDOUT;
+}
+
+static void mlx5e_reset_txqsq_cc_pc(struct mlx5e_txqsq *sq)
+{
+	WARN_ONCE(sq->cc != sq->pc,
+		  "SQ 0x%x: cc (0x%x) != pc (0x%x)\n",
+		  sq->sqn, sq->cc, sq->pc);
+	sq->cc = 0;
+	sq->dma_fifo_cc = 0;
+	sq->pc = 0;
+}
+
+static int mlx5e_tx_reporter_err_cqe_recover(void *ctx)
+{
+	struct mlx5_core_dev *mdev;
+	struct net_device *dev;
+	struct mlx5e_txqsq *sq;
+	u8 state;
+	int err;
+
+	sq = ctx;
+	mdev = sq->mdev;
+	dev = sq->netdev;
+
+	if (!test_bit(MLX5E_SQ_STATE_RECOVERING, &sq->state))
+		return 0;
+
+	err = mlx5_core_query_sq_state(mdev, sq->sqn, &state);
+	if (err) {
+		netdev_err(dev, "Failed to query SQ 0x%x state. err = %d\n",
+			   sq->sqn, err);
+		goto out;
+	}
+
+	if (state != MLX5_SQC_STATE_ERR)
+		goto out;
+
+	mlx5e_tx_disable_queue(sq->txq);
+
+	err = mlx5e_wait_for_sq_flush(sq);
+	if (err)
+		goto out;
+
+	/* At this point, no new packets will arrive from the stack as TXQ is
+	 * marked with QUEUE_STATE_DRV_XOFF. In addition, NAPI cleared all
+	 * pending WQEs. SQ can safely reset the SQ.
+	 */
+
+	err = mlx5e_health_sq_to_ready(mdev, dev, sq->sqn);
+	if (err)
+		goto out;
+
+	mlx5e_reset_txqsq_cc_pc(sq);
+	sq->stats->recover++;
+	clear_bit(MLX5E_SQ_STATE_RECOVERING, &sq->state);
+	mlx5e_activate_txqsq(sq);
+
+	return 0;
+out:
+	clear_bit(MLX5E_SQ_STATE_RECOVERING, &sq->state);
+	return err;
+}
+
+struct mlx5e_tx_timeout_ctx {
+	struct mlx5e_txqsq *sq;
+	signed int status;
+};
+
+static int mlx5e_tx_reporter_timeout_recover(void *ctx)
+{
+	struct mlx5e_tx_timeout_ctx *to_ctx;
+	struct mlx5e_priv *priv;
+	struct mlx5_eq_comp *eq;
+	struct mlx5e_txqsq *sq;
+	int err;
+
+	to_ctx = ctx;
+	sq = to_ctx->sq;
+	eq = sq->cq.mcq.eq;
+	priv = sq->priv;
+	err = mlx5e_health_channel_eq_recover(sq->netdev, eq, sq->cq.ch_stats);
+	if (!err) {
+		to_ctx->status = 0; /* this sq recovered */
+		return err;
+	}
+
+	err = mlx5e_safe_reopen_channels(priv);
+	if (!err) {
+		to_ctx->status = 1; /* all channels recovered */
+		return err;
+	}
+
+	to_ctx->status = err;
+	clear_bit(MLX5E_SQ_STATE_ENABLED, &sq->state);
+	netdev_err(priv->netdev,
+		   "mlx5e_safe_reopen_channels failed recovering from a tx_timeout, err(%d).\n",
+		   err);
+
+	return err;
+}
+
+/* state lock cannot be grabbed within this function.
+ * It can cause a dead lock or a read-after-free.
+ */
+static int mlx5e_tx_reporter_recover_from_ctx(struct mlx5e_err_ctx *err_ctx)
+{
+	return err_ctx->recover(err_ctx->ctx);
+}
+
+static int mlx5e_tx_reporter_recover(struct devlink_health_reporter *reporter,
+				     void *context,
+				     struct netlink_ext_ack *extack)
+{
+	struct mlx5e_priv *priv = devlink_health_reporter_priv(reporter);
+	struct mlx5e_err_ctx *err_ctx = context;
+
+	return err_ctx ? mlx5e_tx_reporter_recover_from_ctx(err_ctx) :
+			 mlx5e_health_recover_channels(priv);
+}
+
+static int
+mlx5e_tx_reporter_build_diagnose_output_sq_common(struct devlink_fmsg *fmsg,
+						  struct mlx5e_txqsq *sq, int tc)
+{
+	bool stopped = netif_xmit_stopped(sq->txq);
+	struct mlx5e_priv *priv = sq->priv;
+	u8 state;
+	int err;
+
+	err = mlx5_core_query_sq_state(priv->mdev, sq->sqn, &state);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u32_pair_put(fmsg, "tc", tc);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u32_pair_put(fmsg, "txq ix", sq->txq_ix);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u32_pair_put(fmsg, "sqn", sq->sqn);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u8_pair_put(fmsg, "HW state", state);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_bool_pair_put(fmsg, "stopped", stopped);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u32_pair_put(fmsg, "cc", sq->cc);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u32_pair_put(fmsg, "pc", sq->pc);
+	if (err)
+		return err;
+
+	err = mlx5e_health_cq_diag_fmsg(&sq->cq, fmsg);
+	if (err)
+		return err;
+
+	return mlx5e_health_eq_diag_fmsg(sq->cq.mcq.eq, fmsg);
+}
+
+static int
+mlx5e_tx_reporter_build_diagnose_output(struct devlink_fmsg *fmsg,
+					struct mlx5e_txqsq *sq, int tc)
+{
+	int err;
+
+	err = devlink_fmsg_obj_nest_start(fmsg);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u32_pair_put(fmsg, "channel ix", sq->ch_ix);
+	if (err)
+		return err;
+
+	err = mlx5e_tx_reporter_build_diagnose_output_sq_common(fmsg, sq, tc);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_obj_nest_end(fmsg);
+	if (err)
+		return err;
+
+	return 0;
+}
+
+static int
+mlx5e_tx_reporter_build_diagnose_output_ptpsq(struct devlink_fmsg *fmsg,
+					      struct mlx5e_ptpsq *ptpsq, int tc)
+{
+	int err;
+
+	err = devlink_fmsg_obj_nest_start(fmsg);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_string_pair_put(fmsg, "channel", "ptp");
+	if (err)
+		return err;
+
+	err = mlx5e_tx_reporter_build_diagnose_output_sq_common(fmsg, &ptpsq->txqsq, tc);
+	if (err)
+		return err;
+
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, "Port TS");
+	if (err)
+		return err;
+
+	err = mlx5e_health_cq_diag_fmsg(&ptpsq->ts_cq, fmsg);
+	if (err)
+		return err;
+
+	err = mlx5e_health_fmsg_named_obj_nest_end(fmsg);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_obj_nest_end(fmsg);
+	if (err)
+		return err;
+
+	return 0;
+}
+
+static int
+mlx5e_tx_reporter_diagnose_generic_txqsq(struct devlink_fmsg *fmsg,
+					 struct mlx5e_txqsq *txqsq)
+{
+	u32 sq_stride, sq_sz;
+	bool real_time;
+	int err;
+
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, "SQ");
+	if (err)
+		return err;
+
+	real_time =  mlx5_is_real_time_sq(txqsq->mdev);
+	sq_sz = mlx5_wq_cyc_get_size(&txqsq->wq);
+	sq_stride = MLX5_SEND_WQE_BB;
+
+	err = devlink_fmsg_u64_pair_put(fmsg, "stride size", sq_stride);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_u32_pair_put(fmsg, "size", sq_sz);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_string_pair_put(fmsg, "ts_format", real_time ? "RT" : "FRC");
+	if (err)
+		return err;
+
+	err = mlx5e_health_cq_common_diag_fmsg(&txqsq->cq, fmsg);
+	if (err)
+		return err;
+
+	return mlx5e_health_fmsg_named_obj_nest_end(fmsg);
+}
+
+static int
+mlx5e_tx_reporter_diagnose_generic_tx_port_ts(struct devlink_fmsg *fmsg,
+					      struct mlx5e_ptpsq *ptpsq)
+{
+	int err;
+
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, "Port TS");
+	if (err)
+		return err;
+
+	err = mlx5e_health_cq_common_diag_fmsg(&ptpsq->ts_cq, fmsg);
+	if (err)
+		return err;
+
+	return mlx5e_health_fmsg_named_obj_nest_end(fmsg);
+}
+
+static int
+mlx5e_tx_reporter_diagnose_common_config(struct devlink_health_reporter *reporter,
+					 struct devlink_fmsg *fmsg)
+{
+	struct mlx5e_priv *priv = devlink_health_reporter_priv(reporter);
+	struct mlx5e_txqsq *generic_sq = priv->txq2sq[0];
+	struct mlx5e_ptp *ptp_ch = priv->channels.ptp;
+	struct mlx5e_ptpsq *generic_ptpsq;
+	int err;
+
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, "Common Config");
+	if (err)
+		return err;
+
+	err = mlx5e_tx_reporter_diagnose_generic_txqsq(fmsg, generic_sq);
+	if (err)
+		return err;
+
+	if (!ptp_ch || !test_bit(MLX5E_PTP_STATE_TX, ptp_ch->state))
+		goto out;
+
+	generic_ptpsq = &ptp_ch->ptpsq[0];
+
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, "PTP");
+	if (err)
+		return err;
+
+	err = mlx5e_tx_reporter_diagnose_generic_txqsq(fmsg, &generic_ptpsq->txqsq);
+	if (err)
+		return err;
+
+	err = mlx5e_tx_reporter_diagnose_generic_tx_port_ts(fmsg, generic_ptpsq);
+	if (err)
+		return err;
+
+	err = mlx5e_health_fmsg_named_obj_nest_end(fmsg);
+	if (err)
+		return err;
+
+out:
+	return mlx5e_health_fmsg_named_obj_nest_end(fmsg);
+}
+
+static int mlx5e_tx_reporter_diagnose(struct devlink_health_reporter *reporter,
+				      struct devlink_fmsg *fmsg,
+				      struct netlink_ext_ack *extack)
+{
+	struct mlx5e_priv *priv = devlink_health_reporter_priv(reporter);
+	struct mlx5e_ptp *ptp_ch = priv->channels.ptp;
+
+	int i, tc, err = 0;
+
+	mutex_lock(&priv->state_lock);
+
+	if (!test_bit(MLX5E_STATE_OPENED, &priv->state))
+		goto unlock;
+
+	err = mlx5e_tx_reporter_diagnose_common_config(reporter, fmsg);
+	if (err)
+		goto unlock;
+
+	err = devlink_fmsg_arr_pair_nest_start(fmsg, "SQs");
+	if (err)
+		goto unlock;
+
+	for (i = 0; i < priv->channels.num; i++) {
+		struct mlx5e_channel *c = priv->channels.c[i];
+
+		for (tc = 0; tc < mlx5e_get_dcb_num_tc(&priv->channels.params); tc++) {
+			struct mlx5e_txqsq *sq = &c->sq[tc];
+
+			err = mlx5e_tx_reporter_build_diagnose_output(fmsg, sq, tc);
+			if (err)
+				goto unlock;
+		}
+	}
+
+	if (!ptp_ch || !test_bit(MLX5E_PTP_STATE_TX, ptp_ch->state))
+		goto close_sqs_nest;
+
+	for (tc = 0; tc < mlx5e_get_dcb_num_tc(&priv->channels.params); tc++) {
+		err = mlx5e_tx_reporter_build_diagnose_output_ptpsq(fmsg,
+								    &ptp_ch->ptpsq[tc],
+								    tc);
+		if (err)
+			goto unlock;
+	}
+
+close_sqs_nest:
+	err = devlink_fmsg_arr_pair_nest_end(fmsg);
+	if (err)
+		goto unlock;
+
+unlock:
+	mutex_unlock(&priv->state_lock);
+	return err;
+}
+
+static int mlx5e_tx_reporter_dump_sq(struct mlx5e_priv *priv, struct devlink_fmsg *fmsg,
+				     void *ctx)
+{
+	struct mlx5_rsc_key key = {};
+	struct mlx5e_txqsq *sq = ctx;
+	int err;
+
+	if (!test_bit(MLX5E_STATE_OPENED, &priv->state))
+		return 0;
+
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, "SX Slice");
+	if (err)
+		return err;
+
+	key.size = PAGE_SIZE;
+	key.rsc = MLX5_SGMT_TYPE_SX_SLICE_ALL;
+	err = mlx5e_health_rsc_fmsg_dump(priv, &key, fmsg);
+	if (err)
+		return err;
+
+	err = mlx5e_health_fmsg_named_obj_nest_end(fmsg);
+	if (err)
+		return err;
+
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, "SQ");
+	if (err)
+		return err;
+
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, "QPC");
+	if (err)
+		return err;
+
+	key.rsc = MLX5_SGMT_TYPE_FULL_QPC;
+	key.index1 = sq->sqn;
+	key.num_of_obj1 = 1;
+
+	err = mlx5e_health_rsc_fmsg_dump(priv, &key, fmsg);
+	if (err)
+		return err;
+
+	err = mlx5e_health_fmsg_named_obj_nest_end(fmsg);
+	if (err)
+		return err;
+
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, "send_buff");
+	if (err)
+		return err;
+
+	key.rsc = MLX5_SGMT_TYPE_SND_BUFF;
+	key.num_of_obj2 = MLX5_RSC_DUMP_ALL;
+	err = mlx5e_health_rsc_fmsg_dump(priv, &key, fmsg);
+	if (err)
+		return err;
+
+	err = mlx5e_health_fmsg_named_obj_nest_end(fmsg);
+	if (err)
+		return err;
+
+	return mlx5e_health_fmsg_named_obj_nest_end(fmsg);
+}
+
+static int mlx5e_tx_reporter_timeout_dump(struct mlx5e_priv *priv, struct devlink_fmsg *fmsg,
+					  void *ctx)
+{
+	struct mlx5e_tx_timeout_ctx *to_ctx = ctx;
+
+	return mlx5e_tx_reporter_dump_sq(priv, fmsg, to_ctx->sq);
+}
+
+static int mlx5e_tx_reporter_dump_all_sqs(struct mlx5e_priv *priv,
+					  struct devlink_fmsg *fmsg)
+{
+	struct mlx5e_ptp *ptp_ch = priv->channels.ptp;
+	struct mlx5_rsc_key key = {};
+	int i, tc, err;
+
+	if (!test_bit(MLX5E_STATE_OPENED, &priv->state))
+		return 0;
+
+	err = mlx5e_health_fmsg_named_obj_nest_start(fmsg, "SX Slice");
+	if (err)
+		return err;
+
+	key.size = PAGE_SIZE;
+	key.rsc = MLX5_SGMT_TYPE_SX_SLICE_ALL;
+	err = mlx5e_health_rsc_fmsg_dump(priv, &key, fmsg);
+	if (err)
+		return err;
+
+	err = mlx5e_health_fmsg_named_obj_nest_end(fmsg);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_arr_pair_nest_start(fmsg, "SQs");
+	if (err)
+		return err;
+
+	for (i = 0; i < priv->channels.num; i++) {
+		struct mlx5e_channel *c = priv->channels.c[i];
+
+		for (tc = 0; tc < mlx5e_get_dcb_num_tc(&priv->channels.params); tc++) {
+			struct mlx5e_txqsq *sq = &c->sq[tc];
+
+			err = mlx5e_health_queue_dump(priv, fmsg, sq->sqn, "SQ");
+			if (err)
+				return err;
+		}
+	}
+
+	if (ptp_ch && test_bit(MLX5E_PTP_STATE_TX, ptp_ch->state)) {
+		for (tc = 0; tc < mlx5e_get_dcb_num_tc(&priv->channels.params); tc++) {
+			struct mlx5e_txqsq *sq = &ptp_ch->ptpsq[tc].txqsq;
+
+			err = mlx5e_health_queue_dump(priv, fmsg, sq->sqn, "PTP SQ");
+			if (err)
+				return err;
+		}
+	}
+
+	return devlink_fmsg_arr_pair_nest_end(fmsg);
+}
+
+static int mlx5e_tx_reporter_dump_from_ctx(struct mlx5e_priv *priv,
+					   struct mlx5e_err_ctx *err_ctx,
+					   struct devlink_fmsg *fmsg)
+{
+	return err_ctx->dump(priv, fmsg, err_ctx->ctx);
+}
+
+static int mlx5e_tx_reporter_dump(struct devlink_health_reporter *reporter,
+				  struct devlink_fmsg *fmsg, void *context,
+				  struct netlink_ext_ack *extack)
+{
+	struct mlx5e_priv *priv = devlink_health_reporter_priv(reporter);
+	struct mlx5e_err_ctx *err_ctx = context;
+
+	return err_ctx ? mlx5e_tx_reporter_dump_from_ctx(priv, err_ctx, fmsg) :
+			 mlx5e_tx_reporter_dump_all_sqs(priv, fmsg);
+}
+
+void mlx5e_reporter_tx_err_cqe(struct mlx5e_txqsq *sq)
+{
+	char err_str[MLX5E_REPORTER_PER_Q_MAX_LEN];
+	struct mlx5e_priv *priv = sq->priv;
+	struct mlx5e_err_ctx err_ctx = {};
+
+	err_ctx.ctx = sq;
+	err_ctx.recover = mlx5e_tx_reporter_err_cqe_recover;
+	err_ctx.dump = mlx5e_tx_reporter_dump_sq;
+	snprintf(err_str, sizeof(err_str), "ERR CQE on SQ: 0x%x", sq->sqn);
+
+	mlx5e_health_report(priv, priv->tx_reporter, err_str, &err_ctx);
+}
+
+int mlx5e_reporter_tx_timeout(struct mlx5e_txqsq *sq)
+{
+	char err_str[MLX5E_REPORTER_PER_Q_MAX_LEN];
+	struct mlx5e_tx_timeout_ctx to_ctx = {};
+	struct mlx5e_priv *priv = sq->priv;
+	struct mlx5e_err_ctx err_ctx = {};
+
+	to_ctx.sq = sq;
+	err_ctx.ctx = &to_ctx;
+	err_ctx.recover = mlx5e_tx_reporter_timeout_recover;
+	err_ctx.dump = mlx5e_tx_reporter_timeout_dump;
+	snprintf(err_str, sizeof(err_str),
+		 "TX timeout on queue: %d, SQ: 0x%x, CQ: 0x%x, SQ Cons: 0x%x SQ Prod: 0x%x, usecs since last trans: %u",
+		 sq->ch_ix, sq->sqn, sq->cq.mcq.cqn, sq->cc, sq->pc,
+		 jiffies_to_usecs(jiffies - READ_ONCE(sq->txq->trans_start)));
+
+	mlx5e_health_report(priv, priv->tx_reporter, err_str, &err_ctx);
+	return to_ctx.status;
+}
+
+static const struct devlink_health_reporter_ops mlx5_tx_reporter_ops = {
+		.name = "tx",
+		.recover = mlx5e_tx_reporter_recover,
+		.diagnose = mlx5e_tx_reporter_diagnose,
+		.dump = mlx5e_tx_reporter_dump,
+};
+
+#define MLX5_REPORTER_TX_GRACEFUL_PERIOD 500
+
+void mlx5e_reporter_tx_create(struct mlx5e_priv *priv)
+{
+	struct devlink_port *dl_port = mlx5e_devlink_get_dl_port(priv);
+	struct devlink_health_reporter *reporter;
+
+	reporter = devlink_port_health_reporter_create(dl_port, &mlx5_tx_reporter_ops,
+						       MLX5_REPORTER_TX_GRACEFUL_PERIOD, priv);
+	if (IS_ERR(reporter)) {
+		netdev_warn(priv->netdev,
+			    "Failed to create tx reporter, err = %ld\n",
+			    PTR_ERR(reporter));
+		return;
+	}
+	priv->tx_reporter = reporter;
+}
+
+void mlx5e_reporter_tx_destroy(struct mlx5e_priv *priv)
+{
+	if (!priv->tx_reporter)
+		return;
+
+	devlink_port_health_reporter_destroy(priv->tx_reporter);
+	priv->tx_reporter = NULL;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/rqt.c b/drivers/net/ethernet/mellanox/mlx5/core/en/rqt.c
new file mode 100644
index 000000000..b915fb29d
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/rqt.c
@@ -0,0 +1,170 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2021, Mellanox Technologies inc. All rights reserved. */
+
+#include "rqt.h"
+#include <linux/mlx5/transobj.h>
+
+void mlx5e_rss_params_indir_init_uniform(struct mlx5e_rss_params_indir *indir,
+					 unsigned int num_channels)
+{
+	unsigned int i;
+
+	for (i = 0; i < MLX5E_INDIR_RQT_SIZE; i++)
+		indir->table[i] = i % num_channels;
+}
+
+static int mlx5e_rqt_init(struct mlx5e_rqt *rqt, struct mlx5_core_dev *mdev,
+			  u16 max_size, u32 *init_rqns, u16 init_size)
+{
+	void *rqtc;
+	int inlen;
+	int err;
+	u32 *in;
+	int i;
+
+	rqt->mdev = mdev;
+	rqt->size = max_size;
+
+	inlen = MLX5_ST_SZ_BYTES(create_rqt_in) + sizeof(u32) * init_size;
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in)
+		return -ENOMEM;
+
+	rqtc = MLX5_ADDR_OF(create_rqt_in, in, rqt_context);
+
+	MLX5_SET(rqtc, rqtc, rqt_max_size, rqt->size);
+
+	MLX5_SET(rqtc, rqtc, rqt_actual_size, init_size);
+	for (i = 0; i < init_size; i++)
+		MLX5_SET(rqtc, rqtc, rq_num[i], init_rqns[i]);
+
+	err = mlx5_core_create_rqt(rqt->mdev, in, inlen, &rqt->rqtn);
+
+	kvfree(in);
+	return err;
+}
+
+int mlx5e_rqt_init_direct(struct mlx5e_rqt *rqt, struct mlx5_core_dev *mdev,
+			  bool indir_enabled, u32 init_rqn)
+{
+	u16 max_size = indir_enabled ? MLX5E_INDIR_RQT_SIZE : 1;
+
+	return mlx5e_rqt_init(rqt, mdev, max_size, &init_rqn, 1);
+}
+
+static int mlx5e_bits_invert(unsigned long a, int size)
+{
+	int inv = 0;
+	int i;
+
+	for (i = 0; i < size; i++)
+		inv |= (test_bit(size - i - 1, &a) ? 1 : 0) << i;
+
+	return inv;
+}
+
+static int mlx5e_calc_indir_rqns(u32 *rss_rqns, u32 *rqns, unsigned int num_rqns,
+				 u8 hfunc, struct mlx5e_rss_params_indir *indir)
+{
+	unsigned int i;
+
+	for (i = 0; i < MLX5E_INDIR_RQT_SIZE; i++) {
+		unsigned int ix = i;
+
+		if (hfunc == ETH_RSS_HASH_XOR)
+			ix = mlx5e_bits_invert(ix, ilog2(MLX5E_INDIR_RQT_SIZE));
+
+		ix = indir->table[ix];
+
+		if (WARN_ON(ix >= num_rqns))
+			/* Could be a bug in the driver or in the kernel part of
+			 * ethtool: indir table refers to non-existent RQs.
+			 */
+			return -EINVAL;
+		rss_rqns[i] = rqns[ix];
+	}
+
+	return 0;
+}
+
+int mlx5e_rqt_init_indir(struct mlx5e_rqt *rqt, struct mlx5_core_dev *mdev,
+			 u32 *rqns, unsigned int num_rqns,
+			 u8 hfunc, struct mlx5e_rss_params_indir *indir)
+{
+	u32 *rss_rqns;
+	int err;
+
+	rss_rqns = kvmalloc_array(MLX5E_INDIR_RQT_SIZE, sizeof(*rss_rqns), GFP_KERNEL);
+	if (!rss_rqns)
+		return -ENOMEM;
+
+	err = mlx5e_calc_indir_rqns(rss_rqns, rqns, num_rqns, hfunc, indir);
+	if (err)
+		goto out;
+
+	err = mlx5e_rqt_init(rqt, mdev, MLX5E_INDIR_RQT_SIZE, rss_rqns, MLX5E_INDIR_RQT_SIZE);
+
+out:
+	kvfree(rss_rqns);
+	return err;
+}
+
+void mlx5e_rqt_destroy(struct mlx5e_rqt *rqt)
+{
+	mlx5_core_destroy_rqt(rqt->mdev, rqt->rqtn);
+}
+
+static int mlx5e_rqt_redirect(struct mlx5e_rqt *rqt, u32 *rqns, unsigned int size)
+{
+	unsigned int i;
+	void *rqtc;
+	int inlen;
+	u32 *in;
+	int err;
+
+	inlen = MLX5_ST_SZ_BYTES(modify_rqt_in) + sizeof(u32) * size;
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in)
+		return -ENOMEM;
+
+	rqtc = MLX5_ADDR_OF(modify_rqt_in, in, ctx);
+
+	MLX5_SET(modify_rqt_in, in, bitmask.rqn_list, 1);
+	MLX5_SET(rqtc, rqtc, rqt_actual_size, size);
+	for (i = 0; i < size; i++)
+		MLX5_SET(rqtc, rqtc, rq_num[i], rqns[i]);
+
+	err = mlx5_core_modify_rqt(rqt->mdev, rqt->rqtn, in, inlen);
+
+	kvfree(in);
+	return err;
+}
+
+int mlx5e_rqt_redirect_direct(struct mlx5e_rqt *rqt, u32 rqn)
+{
+	return mlx5e_rqt_redirect(rqt, &rqn, 1);
+}
+
+int mlx5e_rqt_redirect_indir(struct mlx5e_rqt *rqt, u32 *rqns, unsigned int num_rqns,
+			     u8 hfunc, struct mlx5e_rss_params_indir *indir)
+{
+	u32 *rss_rqns;
+	int err;
+
+	if (WARN_ON(rqt->size != MLX5E_INDIR_RQT_SIZE))
+		return -EINVAL;
+
+	rss_rqns = kvmalloc_array(MLX5E_INDIR_RQT_SIZE, sizeof(*rss_rqns), GFP_KERNEL);
+	if (!rss_rqns)
+		return -ENOMEM;
+
+	err = mlx5e_calc_indir_rqns(rss_rqns, rqns, num_rqns, hfunc, indir);
+	if (err)
+		goto out;
+
+	err = mlx5e_rqt_redirect(rqt, rss_rqns, MLX5E_INDIR_RQT_SIZE);
+
+out:
+	kvfree(rss_rqns);
+	return err;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/rqt.h b/drivers/net/ethernet/mellanox/mlx5/core/en/rqt.h
new file mode 100644
index 000000000..60c985a12
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/rqt.h
@@ -0,0 +1,42 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2021, Mellanox Technologies inc. All rights reserved. */
+
+#ifndef __MLX5_EN_RQT_H__
+#define __MLX5_EN_RQT_H__
+
+#include <linux/kernel.h>
+
+#define MLX5E_INDIR_RQT_SIZE (1 << 8)
+
+struct mlx5_core_dev;
+
+struct mlx5e_rss_params_indir {
+	u32 table[MLX5E_INDIR_RQT_SIZE];
+};
+
+void mlx5e_rss_params_indir_init_uniform(struct mlx5e_rss_params_indir *indir,
+					 unsigned int num_channels);
+
+struct mlx5e_rqt {
+	struct mlx5_core_dev *mdev;
+	u32 rqtn;
+	u16 size;
+};
+
+int mlx5e_rqt_init_direct(struct mlx5e_rqt *rqt, struct mlx5_core_dev *mdev,
+			  bool indir_enabled, u32 init_rqn);
+int mlx5e_rqt_init_indir(struct mlx5e_rqt *rqt, struct mlx5_core_dev *mdev,
+			 u32 *rqns, unsigned int num_rqns,
+			 u8 hfunc, struct mlx5e_rss_params_indir *indir);
+void mlx5e_rqt_destroy(struct mlx5e_rqt *rqt);
+
+static inline u32 mlx5e_rqt_get_rqtn(struct mlx5e_rqt *rqt)
+{
+	return rqt->rqtn;
+}
+
+int mlx5e_rqt_redirect_direct(struct mlx5e_rqt *rqt, u32 rqn);
+int mlx5e_rqt_redirect_indir(struct mlx5e_rqt *rqt, u32 *rqns, unsigned int num_rqns,
+			     u8 hfunc, struct mlx5e_rss_params_indir *indir);
+
+#endif /* __MLX5_EN_RQT_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/rss.c b/drivers/net/ethernet/mellanox/mlx5/core/en/rss.c
new file mode 100644
index 000000000..7f93426b8
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/rss.c
@@ -0,0 +1,606 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+// Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES.
+
+#include "rss.h"
+
+#define mlx5e_rss_warn(__dev, format, ...)			\
+	dev_warn((__dev)->device, "%s:%d:(pid %d): " format,	\
+		 __func__, __LINE__, current->pid,		\
+		 ##__VA_ARGS__)
+
+static const struct mlx5e_rss_params_traffic_type rss_default_config[MLX5E_NUM_INDIR_TIRS] = {
+	[MLX5_TT_IPV4_TCP] = {
+		.l3_prot_type = MLX5_L3_PROT_TYPE_IPV4,
+		.l4_prot_type = MLX5_L4_PROT_TYPE_TCP,
+		.rx_hash_fields = MLX5_HASH_IP_L4PORTS,
+	},
+	[MLX5_TT_IPV6_TCP] = {
+		.l3_prot_type = MLX5_L3_PROT_TYPE_IPV6,
+		.l4_prot_type = MLX5_L4_PROT_TYPE_TCP,
+		.rx_hash_fields = MLX5_HASH_IP_L4PORTS,
+	},
+	[MLX5_TT_IPV4_UDP] = {
+		.l3_prot_type = MLX5_L3_PROT_TYPE_IPV4,
+		.l4_prot_type = MLX5_L4_PROT_TYPE_UDP,
+		.rx_hash_fields = MLX5_HASH_IP_L4PORTS,
+	},
+	[MLX5_TT_IPV6_UDP] = {
+		.l3_prot_type = MLX5_L3_PROT_TYPE_IPV6,
+		.l4_prot_type = MLX5_L4_PROT_TYPE_UDP,
+		.rx_hash_fields = MLX5_HASH_IP_L4PORTS,
+	},
+	[MLX5_TT_IPV4_IPSEC_AH] = {
+		.l3_prot_type = MLX5_L3_PROT_TYPE_IPV4,
+		.l4_prot_type = 0,
+		.rx_hash_fields = MLX5_HASH_IP_IPSEC_SPI,
+	},
+	[MLX5_TT_IPV6_IPSEC_AH] = {
+		.l3_prot_type = MLX5_L3_PROT_TYPE_IPV6,
+		.l4_prot_type = 0,
+		.rx_hash_fields = MLX5_HASH_IP_IPSEC_SPI,
+	},
+	[MLX5_TT_IPV4_IPSEC_ESP] = {
+		.l3_prot_type = MLX5_L3_PROT_TYPE_IPV4,
+		.l4_prot_type = 0,
+		.rx_hash_fields = MLX5_HASH_IP_IPSEC_SPI,
+	},
+	[MLX5_TT_IPV6_IPSEC_ESP] = {
+		.l3_prot_type = MLX5_L3_PROT_TYPE_IPV6,
+		.l4_prot_type = 0,
+		.rx_hash_fields = MLX5_HASH_IP_IPSEC_SPI,
+	},
+	[MLX5_TT_IPV4] = {
+		.l3_prot_type = MLX5_L3_PROT_TYPE_IPV4,
+		.l4_prot_type = 0,
+		.rx_hash_fields = MLX5_HASH_IP,
+	},
+	[MLX5_TT_IPV6] = {
+		.l3_prot_type = MLX5_L3_PROT_TYPE_IPV6,
+		.l4_prot_type = 0,
+		.rx_hash_fields = MLX5_HASH_IP,
+	},
+};
+
+struct mlx5e_rss_params_traffic_type
+mlx5e_rss_get_default_tt_config(enum mlx5_traffic_types tt)
+{
+	return rss_default_config[tt];
+}
+
+struct mlx5e_rss {
+	struct mlx5e_rss_params_hash hash;
+	struct mlx5e_rss_params_indir indir;
+	u32 rx_hash_fields[MLX5E_NUM_INDIR_TIRS];
+	struct mlx5e_tir *tir[MLX5E_NUM_INDIR_TIRS];
+	struct mlx5e_tir *inner_tir[MLX5E_NUM_INDIR_TIRS];
+	struct mlx5e_rqt rqt;
+	struct mlx5_core_dev *mdev;
+	u32 drop_rqn;
+	bool inner_ft_support;
+	bool enabled;
+	refcount_t refcnt;
+};
+
+struct mlx5e_rss *mlx5e_rss_alloc(void)
+{
+	return kvzalloc(sizeof(struct mlx5e_rss), GFP_KERNEL);
+}
+
+void mlx5e_rss_free(struct mlx5e_rss *rss)
+{
+	kvfree(rss);
+}
+
+static void mlx5e_rss_params_init(struct mlx5e_rss *rss)
+{
+	enum mlx5_traffic_types tt;
+
+	rss->hash.hfunc = ETH_RSS_HASH_TOP;
+	netdev_rss_key_fill(rss->hash.toeplitz_hash_key,
+			    sizeof(rss->hash.toeplitz_hash_key));
+	for (tt = 0; tt < MLX5E_NUM_INDIR_TIRS; tt++)
+		rss->rx_hash_fields[tt] =
+			mlx5e_rss_get_default_tt_config(tt).rx_hash_fields;
+}
+
+static struct mlx5e_tir **rss_get_tirp(struct mlx5e_rss *rss, enum mlx5_traffic_types tt,
+				       bool inner)
+{
+	return inner ? &rss->inner_tir[tt] : &rss->tir[tt];
+}
+
+static struct mlx5e_tir *rss_get_tir(struct mlx5e_rss *rss, enum mlx5_traffic_types tt,
+				     bool inner)
+{
+	return *rss_get_tirp(rss, tt, inner);
+}
+
+static struct mlx5e_rss_params_traffic_type
+mlx5e_rss_get_tt_config(struct mlx5e_rss *rss, enum mlx5_traffic_types tt)
+{
+	struct mlx5e_rss_params_traffic_type rss_tt;
+
+	rss_tt = mlx5e_rss_get_default_tt_config(tt);
+	rss_tt.rx_hash_fields = rss->rx_hash_fields[tt];
+	return rss_tt;
+}
+
+static int mlx5e_rss_create_tir(struct mlx5e_rss *rss,
+				enum mlx5_traffic_types tt,
+				const struct mlx5e_packet_merge_param *init_pkt_merge_param,
+				bool inner)
+{
+	struct mlx5e_rss_params_traffic_type rss_tt;
+	struct mlx5e_tir_builder *builder;
+	struct mlx5e_tir **tir_p;
+	struct mlx5e_tir *tir;
+	u32 rqtn;
+	int err;
+
+	if (inner && !rss->inner_ft_support) {
+		mlx5e_rss_warn(rss->mdev,
+			       "Cannot create inner indirect TIR[%d], RSS inner FT is not supported.\n",
+			       tt);
+		return -EINVAL;
+	}
+
+	tir_p = rss_get_tirp(rss, tt, inner);
+	if (*tir_p)
+		return -EINVAL;
+
+	tir = kvzalloc(sizeof(*tir), GFP_KERNEL);
+	if (!tir)
+		return -ENOMEM;
+
+	builder = mlx5e_tir_builder_alloc(false);
+	if (!builder) {
+		err = -ENOMEM;
+		goto free_tir;
+	}
+
+	rqtn = mlx5e_rqt_get_rqtn(&rss->rqt);
+	mlx5e_tir_builder_build_rqt(builder, rss->mdev->mlx5e_res.hw_objs.td.tdn,
+				    rqtn, rss->inner_ft_support);
+	mlx5e_tir_builder_build_packet_merge(builder, init_pkt_merge_param);
+	rss_tt = mlx5e_rss_get_tt_config(rss, tt);
+	mlx5e_tir_builder_build_rss(builder, &rss->hash, &rss_tt, inner);
+
+	err = mlx5e_tir_init(tir, builder, rss->mdev, true);
+	mlx5e_tir_builder_free(builder);
+	if (err) {
+		mlx5e_rss_warn(rss->mdev, "Failed to create %sindirect TIR: err = %d, tt = %d\n",
+			       inner ? "inner " : "", err, tt);
+		goto free_tir;
+	}
+
+	*tir_p = tir;
+	return 0;
+
+free_tir:
+	kvfree(tir);
+	return err;
+}
+
+static void mlx5e_rss_destroy_tir(struct mlx5e_rss *rss, enum mlx5_traffic_types tt,
+				  bool inner)
+{
+	struct mlx5e_tir **tir_p;
+	struct mlx5e_tir *tir;
+
+	tir_p = rss_get_tirp(rss, tt, inner);
+	if (!*tir_p)
+		return;
+
+	tir = *tir_p;
+	mlx5e_tir_destroy(tir);
+	kvfree(tir);
+	*tir_p = NULL;
+}
+
+static int mlx5e_rss_create_tirs(struct mlx5e_rss *rss,
+				 const struct mlx5e_packet_merge_param *init_pkt_merge_param,
+				 bool inner)
+{
+	enum mlx5_traffic_types tt, max_tt;
+	int err;
+
+	for (tt = 0; tt < MLX5E_NUM_INDIR_TIRS; tt++) {
+		err = mlx5e_rss_create_tir(rss, tt, init_pkt_merge_param, inner);
+		if (err)
+			goto err_destroy_tirs;
+	}
+
+	return 0;
+
+err_destroy_tirs:
+	max_tt = tt;
+	for (tt = 0; tt < max_tt; tt++)
+		mlx5e_rss_destroy_tir(rss, tt, inner);
+	return err;
+}
+
+static void mlx5e_rss_destroy_tirs(struct mlx5e_rss *rss, bool inner)
+{
+	enum mlx5_traffic_types tt;
+
+	for (tt = 0; tt < MLX5E_NUM_INDIR_TIRS; tt++)
+		mlx5e_rss_destroy_tir(rss, tt, inner);
+}
+
+static int mlx5e_rss_update_tir(struct mlx5e_rss *rss, enum mlx5_traffic_types tt,
+				bool inner)
+{
+	struct mlx5e_rss_params_traffic_type rss_tt;
+	struct mlx5e_tir_builder *builder;
+	struct mlx5e_tir *tir;
+	int err;
+
+	tir = rss_get_tir(rss, tt, inner);
+	if (!tir)
+		return 0;
+
+	builder = mlx5e_tir_builder_alloc(true);
+	if (!builder)
+		return -ENOMEM;
+
+	rss_tt = mlx5e_rss_get_tt_config(rss, tt);
+
+	mlx5e_tir_builder_build_rss(builder, &rss->hash, &rss_tt, inner);
+	err = mlx5e_tir_modify(tir, builder);
+
+	mlx5e_tir_builder_free(builder);
+	return err;
+}
+
+static int mlx5e_rss_update_tirs(struct mlx5e_rss *rss)
+{
+	enum mlx5_traffic_types tt;
+	int err, retval;
+
+	retval = 0;
+
+	for (tt = 0; tt < MLX5E_NUM_INDIR_TIRS; tt++) {
+		err = mlx5e_rss_update_tir(rss, tt, false);
+		if (err) {
+			retval = retval ? : err;
+			mlx5e_rss_warn(rss->mdev,
+				       "Failed to update RSS hash of indirect TIR for traffic type %d: err = %d\n",
+				       tt, err);
+		}
+
+		if (!rss->inner_ft_support)
+			continue;
+
+		err = mlx5e_rss_update_tir(rss, tt, true);
+		if (err) {
+			retval = retval ? : err;
+			mlx5e_rss_warn(rss->mdev,
+				       "Failed to update RSS hash of inner indirect TIR for traffic type %d: err = %d\n",
+				       tt, err);
+		}
+	}
+	return retval;
+}
+
+int mlx5e_rss_init_no_tirs(struct mlx5e_rss *rss, struct mlx5_core_dev *mdev,
+			   bool inner_ft_support, u32 drop_rqn)
+{
+	rss->mdev = mdev;
+	rss->inner_ft_support = inner_ft_support;
+	rss->drop_rqn = drop_rqn;
+
+	mlx5e_rss_params_init(rss);
+	refcount_set(&rss->refcnt, 1);
+
+	return mlx5e_rqt_init_direct(&rss->rqt, mdev, true, drop_rqn);
+}
+
+int mlx5e_rss_init(struct mlx5e_rss *rss, struct mlx5_core_dev *mdev,
+		   bool inner_ft_support, u32 drop_rqn,
+		   const struct mlx5e_packet_merge_param *init_pkt_merge_param)
+{
+	int err;
+
+	err = mlx5e_rss_init_no_tirs(rss, mdev, inner_ft_support, drop_rqn);
+	if (err)
+		goto err_out;
+
+	err = mlx5e_rss_create_tirs(rss, init_pkt_merge_param, false);
+	if (err)
+		goto err_destroy_rqt;
+
+	if (inner_ft_support) {
+		err = mlx5e_rss_create_tirs(rss, init_pkt_merge_param, true);
+		if (err)
+			goto err_destroy_tirs;
+	}
+
+	return 0;
+
+err_destroy_tirs:
+	mlx5e_rss_destroy_tirs(rss, false);
+err_destroy_rqt:
+	mlx5e_rqt_destroy(&rss->rqt);
+err_out:
+	return err;
+}
+
+int mlx5e_rss_cleanup(struct mlx5e_rss *rss)
+{
+	if (!refcount_dec_if_one(&rss->refcnt))
+		return -EBUSY;
+
+	mlx5e_rss_destroy_tirs(rss, false);
+
+	if (rss->inner_ft_support)
+		mlx5e_rss_destroy_tirs(rss, true);
+
+	mlx5e_rqt_destroy(&rss->rqt);
+
+	return 0;
+}
+
+void mlx5e_rss_refcnt_inc(struct mlx5e_rss *rss)
+{
+	refcount_inc(&rss->refcnt);
+}
+
+void mlx5e_rss_refcnt_dec(struct mlx5e_rss *rss)
+{
+	refcount_dec(&rss->refcnt);
+}
+
+unsigned int mlx5e_rss_refcnt_read(struct mlx5e_rss *rss)
+{
+	return refcount_read(&rss->refcnt);
+}
+
+u32 mlx5e_rss_get_tirn(struct mlx5e_rss *rss, enum mlx5_traffic_types tt,
+		       bool inner)
+{
+	struct mlx5e_tir *tir;
+
+	WARN_ON(inner && !rss->inner_ft_support);
+	tir = rss_get_tir(rss, tt, inner);
+	WARN_ON(!tir);
+
+	return mlx5e_tir_get_tirn(tir);
+}
+
+/* Fill the "tirn" output parameter.
+ * Create the requested TIR if it's its first usage.
+ */
+int mlx5e_rss_obtain_tirn(struct mlx5e_rss *rss,
+			  enum mlx5_traffic_types tt,
+			  const struct mlx5e_packet_merge_param *init_pkt_merge_param,
+			  bool inner, u32 *tirn)
+{
+	struct mlx5e_tir *tir;
+
+	tir = rss_get_tir(rss, tt, inner);
+	if (!tir) { /* TIR doesn't exist, create one */
+		int err;
+
+		err = mlx5e_rss_create_tir(rss, tt, init_pkt_merge_param, inner);
+		if (err)
+			return err;
+		tir = rss_get_tir(rss, tt, inner);
+	}
+
+	*tirn = mlx5e_tir_get_tirn(tir);
+	return 0;
+}
+
+static int mlx5e_rss_apply(struct mlx5e_rss *rss, u32 *rqns, unsigned int num_rqns)
+{
+	int err;
+
+	err = mlx5e_rqt_redirect_indir(&rss->rqt, rqns, num_rqns, rss->hash.hfunc, &rss->indir);
+	if (err)
+		mlx5e_rss_warn(rss->mdev, "Failed to redirect RQT %#x to channels: err = %d\n",
+			       mlx5e_rqt_get_rqtn(&rss->rqt), err);
+	return err;
+}
+
+void mlx5e_rss_enable(struct mlx5e_rss *rss, u32 *rqns, unsigned int num_rqns)
+{
+	rss->enabled = true;
+	mlx5e_rss_apply(rss, rqns, num_rqns);
+}
+
+void mlx5e_rss_disable(struct mlx5e_rss *rss)
+{
+	int err;
+
+	rss->enabled = false;
+	err = mlx5e_rqt_redirect_direct(&rss->rqt, rss->drop_rqn);
+	if (err)
+		mlx5e_rss_warn(rss->mdev, "Failed to redirect RQT %#x to drop RQ %#x: err = %d\n",
+			       mlx5e_rqt_get_rqtn(&rss->rqt), rss->drop_rqn, err);
+}
+
+int mlx5e_rss_packet_merge_set_param(struct mlx5e_rss *rss,
+				     struct mlx5e_packet_merge_param *pkt_merge_param)
+{
+	struct mlx5e_tir_builder *builder;
+	enum mlx5_traffic_types tt;
+	int err, final_err;
+
+	builder = mlx5e_tir_builder_alloc(true);
+	if (!builder)
+		return -ENOMEM;
+
+	mlx5e_tir_builder_build_packet_merge(builder, pkt_merge_param);
+
+	final_err = 0;
+
+	for (tt = 0; tt < MLX5E_NUM_INDIR_TIRS; tt++) {
+		struct mlx5e_tir *tir;
+
+		tir = rss_get_tir(rss, tt, false);
+		if (!tir)
+			goto inner_tir;
+		err = mlx5e_tir_modify(tir, builder);
+		if (err) {
+			mlx5e_rss_warn(rss->mdev, "Failed to update packet merge state of indirect TIR %#x for traffic type %d: err = %d\n",
+				       mlx5e_tir_get_tirn(tir), tt, err);
+			if (!final_err)
+				final_err = err;
+		}
+
+inner_tir:
+		if (!rss->inner_ft_support)
+			continue;
+
+		tir = rss_get_tir(rss, tt, true);
+		if (!tir)
+			continue;
+		err = mlx5e_tir_modify(tir, builder);
+		if (err) {
+			mlx5e_rss_warn(rss->mdev, "Failed to update packet merge state of inner indirect TIR %#x for traffic type %d: err = %d\n",
+				       mlx5e_tir_get_tirn(tir), tt, err);
+			if (!final_err)
+				final_err = err;
+		}
+	}
+
+	mlx5e_tir_builder_free(builder);
+	return final_err;
+}
+
+int mlx5e_rss_get_rxfh(struct mlx5e_rss *rss, u32 *indir, u8 *key, u8 *hfunc)
+{
+	unsigned int i;
+
+	if (indir)
+		for (i = 0; i < MLX5E_INDIR_RQT_SIZE; i++)
+			indir[i] = rss->indir.table[i];
+
+	if (key)
+		memcpy(key, rss->hash.toeplitz_hash_key,
+		       sizeof(rss->hash.toeplitz_hash_key));
+
+	if (hfunc)
+		*hfunc = rss->hash.hfunc;
+
+	return 0;
+}
+
+int mlx5e_rss_set_rxfh(struct mlx5e_rss *rss, const u32 *indir,
+		       const u8 *key, const u8 *hfunc,
+		       u32 *rqns, unsigned int num_rqns)
+{
+	bool changed_indir = false;
+	bool changed_hash = false;
+	struct mlx5e_rss *old_rss;
+	int err = 0;
+
+	old_rss = mlx5e_rss_alloc();
+	if (!old_rss)
+		return -ENOMEM;
+
+	*old_rss = *rss;
+
+	if (hfunc && *hfunc != rss->hash.hfunc) {
+		switch (*hfunc) {
+		case ETH_RSS_HASH_XOR:
+		case ETH_RSS_HASH_TOP:
+			break;
+		default:
+			err = -EINVAL;
+			goto out;
+		}
+		changed_hash = true;
+		changed_indir = true;
+		rss->hash.hfunc = *hfunc;
+	}
+
+	if (key) {
+		if (rss->hash.hfunc == ETH_RSS_HASH_TOP)
+			changed_hash = true;
+		memcpy(rss->hash.toeplitz_hash_key, key,
+		       sizeof(rss->hash.toeplitz_hash_key));
+	}
+
+	if (indir) {
+		unsigned int i;
+
+		changed_indir = true;
+
+		for (i = 0; i < MLX5E_INDIR_RQT_SIZE; i++)
+			rss->indir.table[i] = indir[i];
+	}
+
+	if (changed_indir && rss->enabled) {
+		err = mlx5e_rss_apply(rss, rqns, num_rqns);
+		if (err) {
+			*rss = *old_rss;
+			goto out;
+		}
+	}
+
+	if (changed_hash)
+		mlx5e_rss_update_tirs(rss);
+
+out:
+	mlx5e_rss_free(old_rss);
+	return err;
+}
+
+struct mlx5e_rss_params_hash mlx5e_rss_get_hash(struct mlx5e_rss *rss)
+{
+	return rss->hash;
+}
+
+u8 mlx5e_rss_get_hash_fields(struct mlx5e_rss *rss, enum mlx5_traffic_types tt)
+{
+	return rss->rx_hash_fields[tt];
+}
+
+int mlx5e_rss_set_hash_fields(struct mlx5e_rss *rss, enum mlx5_traffic_types tt,
+			      u8 rx_hash_fields)
+{
+	u8 old_rx_hash_fields;
+	int err;
+
+	old_rx_hash_fields = rss->rx_hash_fields[tt];
+
+	if (old_rx_hash_fields == rx_hash_fields)
+		return 0;
+
+	rss->rx_hash_fields[tt] = rx_hash_fields;
+
+	err = mlx5e_rss_update_tir(rss, tt, false);
+	if (err) {
+		rss->rx_hash_fields[tt] = old_rx_hash_fields;
+		mlx5e_rss_warn(rss->mdev,
+			       "Failed to update RSS hash fields of indirect TIR for traffic type %d: err = %d\n",
+			       tt, err);
+		return err;
+	}
+
+	if (!(rss->inner_ft_support))
+		return 0;
+
+	err = mlx5e_rss_update_tir(rss, tt, true);
+	if (err) {
+		/* Partial update happened. Try to revert - it may fail too, but
+		 * there is nothing more we can do.
+		 */
+		rss->rx_hash_fields[tt] = old_rx_hash_fields;
+		mlx5e_rss_warn(rss->mdev,
+			       "Failed to update RSS hash fields of inner indirect TIR for traffic type %d: err = %d\n",
+			       tt, err);
+		if (mlx5e_rss_update_tir(rss, tt, false))
+			mlx5e_rss_warn(rss->mdev,
+				       "Partial update of RSS hash fields happened: failed to revert indirect TIR for traffic type %d to the old values\n",
+				       tt);
+	}
+
+	return err;
+}
+
+void mlx5e_rss_set_indir_uniform(struct mlx5e_rss *rss, unsigned int nch)
+{
+	mlx5e_rss_params_indir_init_uniform(&rss->indir, nch);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/rss.h b/drivers/net/ethernet/mellanox/mlx5/core/en/rss.h
new file mode 100644
index 000000000..c6b216416
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/rss.h
@@ -0,0 +1,50 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. */
+
+#ifndef __MLX5_EN_RSS_H__
+#define __MLX5_EN_RSS_H__
+
+#include "rqt.h"
+#include "tir.h"
+#include "fs.h"
+
+struct mlx5e_rss_params_traffic_type
+mlx5e_rss_get_default_tt_config(enum mlx5_traffic_types tt);
+
+struct mlx5e_rss;
+
+struct mlx5e_rss *mlx5e_rss_alloc(void);
+void mlx5e_rss_free(struct mlx5e_rss *rss);
+int mlx5e_rss_init(struct mlx5e_rss *rss, struct mlx5_core_dev *mdev,
+		   bool inner_ft_support, u32 drop_rqn,
+		   const struct mlx5e_packet_merge_param *init_pkt_merge_param);
+int mlx5e_rss_init_no_tirs(struct mlx5e_rss *rss, struct mlx5_core_dev *mdev,
+			   bool inner_ft_support, u32 drop_rqn);
+int mlx5e_rss_cleanup(struct mlx5e_rss *rss);
+
+void mlx5e_rss_refcnt_inc(struct mlx5e_rss *rss);
+void mlx5e_rss_refcnt_dec(struct mlx5e_rss *rss);
+unsigned int mlx5e_rss_refcnt_read(struct mlx5e_rss *rss);
+
+u32 mlx5e_rss_get_tirn(struct mlx5e_rss *rss, enum mlx5_traffic_types tt,
+		       bool inner);
+int mlx5e_rss_obtain_tirn(struct mlx5e_rss *rss,
+			  enum mlx5_traffic_types tt,
+			  const struct mlx5e_packet_merge_param *init_pkt_merge_param,
+			  bool inner, u32 *tirn);
+
+void mlx5e_rss_enable(struct mlx5e_rss *rss, u32 *rqns, unsigned int num_rqns);
+void mlx5e_rss_disable(struct mlx5e_rss *rss);
+
+int mlx5e_rss_packet_merge_set_param(struct mlx5e_rss *rss,
+				     struct mlx5e_packet_merge_param *pkt_merge_param);
+int mlx5e_rss_get_rxfh(struct mlx5e_rss *rss, u32 *indir, u8 *key, u8 *hfunc);
+int mlx5e_rss_set_rxfh(struct mlx5e_rss *rss, const u32 *indir,
+		       const u8 *key, const u8 *hfunc,
+		       u32 *rqns, unsigned int num_rqns);
+struct mlx5e_rss_params_hash mlx5e_rss_get_hash(struct mlx5e_rss *rss);
+u8 mlx5e_rss_get_hash_fields(struct mlx5e_rss *rss, enum mlx5_traffic_types tt);
+int mlx5e_rss_set_hash_fields(struct mlx5e_rss *rss, enum mlx5_traffic_types tt,
+			      u8 rx_hash_fields);
+void mlx5e_rss_set_indir_uniform(struct mlx5e_rss *rss, unsigned int nch);
+#endif /* __MLX5_EN_RSS_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/rx_res.c b/drivers/net/ethernet/mellanox/mlx5/core/en/rx_res.c
new file mode 100644
index 000000000..e1095bc36
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/rx_res.c
@@ -0,0 +1,640 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2021, Mellanox Technologies inc. All rights reserved. */
+
+#include "rx_res.h"
+#include "channels.h"
+#include "params.h"
+
+#define MLX5E_MAX_NUM_RSS 16
+
+struct mlx5e_rx_res {
+	struct mlx5_core_dev *mdev;
+	enum mlx5e_rx_res_features features;
+	unsigned int max_nch;
+	u32 drop_rqn;
+
+	struct mlx5e_packet_merge_param pkt_merge_param;
+	struct rw_semaphore pkt_merge_param_sem;
+
+	struct mlx5e_rss *rss[MLX5E_MAX_NUM_RSS];
+	bool rss_active;
+	u32 rss_rqns[MLX5E_INDIR_RQT_SIZE];
+	unsigned int rss_nch;
+
+	struct {
+		struct mlx5e_rqt direct_rqt;
+		struct mlx5e_tir direct_tir;
+	} *channels;
+
+	struct {
+		struct mlx5e_rqt rqt;
+		struct mlx5e_tir tir;
+	} ptp;
+};
+
+/* API for rx_res_rss_* */
+
+static int mlx5e_rx_res_rss_init_def(struct mlx5e_rx_res *res,
+				     unsigned int init_nch)
+{
+	bool inner_ft_support = res->features & MLX5E_RX_RES_FEATURE_INNER_FT;
+	struct mlx5e_rss *rss;
+	int err;
+
+	if (WARN_ON(res->rss[0]))
+		return -EINVAL;
+
+	rss = mlx5e_rss_alloc();
+	if (!rss)
+		return -ENOMEM;
+
+	err = mlx5e_rss_init(rss, res->mdev, inner_ft_support, res->drop_rqn,
+			     &res->pkt_merge_param);
+	if (err)
+		goto err_rss_free;
+
+	mlx5e_rss_set_indir_uniform(rss, init_nch);
+
+	res->rss[0] = rss;
+
+	return 0;
+
+err_rss_free:
+	mlx5e_rss_free(rss);
+	return err;
+}
+
+int mlx5e_rx_res_rss_init(struct mlx5e_rx_res *res, u32 *rss_idx, unsigned int init_nch)
+{
+	bool inner_ft_support = res->features & MLX5E_RX_RES_FEATURE_INNER_FT;
+	struct mlx5e_rss *rss;
+	int err, i;
+
+	for (i = 1; i < MLX5E_MAX_NUM_RSS; i++)
+		if (!res->rss[i])
+			break;
+
+	if (i == MLX5E_MAX_NUM_RSS)
+		return -ENOSPC;
+
+	rss = mlx5e_rss_alloc();
+	if (!rss)
+		return -ENOMEM;
+
+	err = mlx5e_rss_init_no_tirs(rss, res->mdev, inner_ft_support, res->drop_rqn);
+	if (err)
+		goto err_rss_free;
+
+	mlx5e_rss_set_indir_uniform(rss, init_nch);
+	if (res->rss_active)
+		mlx5e_rss_enable(rss, res->rss_rqns, res->rss_nch);
+
+	res->rss[i] = rss;
+	*rss_idx = i;
+
+	return 0;
+
+err_rss_free:
+	mlx5e_rss_free(rss);
+	return err;
+}
+
+static int __mlx5e_rx_res_rss_destroy(struct mlx5e_rx_res *res, u32 rss_idx)
+{
+	struct mlx5e_rss *rss = res->rss[rss_idx];
+	int err;
+
+	err = mlx5e_rss_cleanup(rss);
+	if (err)
+		return err;
+
+	mlx5e_rss_free(rss);
+	res->rss[rss_idx] = NULL;
+
+	return 0;
+}
+
+int mlx5e_rx_res_rss_destroy(struct mlx5e_rx_res *res, u32 rss_idx)
+{
+	struct mlx5e_rss *rss;
+
+	if (rss_idx >= MLX5E_MAX_NUM_RSS)
+		return -EINVAL;
+
+	rss = res->rss[rss_idx];
+	if (!rss)
+		return -EINVAL;
+
+	return __mlx5e_rx_res_rss_destroy(res, rss_idx);
+}
+
+static void mlx5e_rx_res_rss_destroy_all(struct mlx5e_rx_res *res)
+{
+	int i;
+
+	for (i = 0; i < MLX5E_MAX_NUM_RSS; i++) {
+		struct mlx5e_rss *rss = res->rss[i];
+		int err;
+
+		if (!rss)
+			continue;
+
+		err = __mlx5e_rx_res_rss_destroy(res, i);
+		if (err) {
+			unsigned int refcount;
+
+			refcount = mlx5e_rss_refcnt_read(rss);
+			mlx5_core_warn(res->mdev,
+				       "Failed to destroy RSS context %d, refcount = %u, err = %d\n",
+				       i, refcount, err);
+		}
+	}
+}
+
+static void mlx5e_rx_res_rss_enable(struct mlx5e_rx_res *res)
+{
+	int i;
+
+	res->rss_active = true;
+
+	for (i = 0; i < MLX5E_MAX_NUM_RSS; i++) {
+		struct mlx5e_rss *rss = res->rss[i];
+
+		if (!rss)
+			continue;
+		mlx5e_rss_enable(rss, res->rss_rqns, res->rss_nch);
+	}
+}
+
+static void mlx5e_rx_res_rss_disable(struct mlx5e_rx_res *res)
+{
+	int i;
+
+	res->rss_active = false;
+
+	for (i = 0; i < MLX5E_MAX_NUM_RSS; i++) {
+		struct mlx5e_rss *rss = res->rss[i];
+
+		if (!rss)
+			continue;
+		mlx5e_rss_disable(rss);
+	}
+}
+
+/* Updates the indirection table SW shadow, does not update the HW resources yet */
+void mlx5e_rx_res_rss_set_indir_uniform(struct mlx5e_rx_res *res, unsigned int nch)
+{
+	WARN_ON_ONCE(res->rss_active);
+	mlx5e_rss_set_indir_uniform(res->rss[0], nch);
+}
+
+int mlx5e_rx_res_rss_get_rxfh(struct mlx5e_rx_res *res, u32 rss_idx,
+			      u32 *indir, u8 *key, u8 *hfunc)
+{
+	struct mlx5e_rss *rss;
+
+	if (rss_idx >= MLX5E_MAX_NUM_RSS)
+		return -EINVAL;
+
+	rss = res->rss[rss_idx];
+	if (!rss)
+		return -ENOENT;
+
+	return mlx5e_rss_get_rxfh(rss, indir, key, hfunc);
+}
+
+int mlx5e_rx_res_rss_set_rxfh(struct mlx5e_rx_res *res, u32 rss_idx,
+			      const u32 *indir, const u8 *key, const u8 *hfunc)
+{
+	struct mlx5e_rss *rss;
+
+	if (rss_idx >= MLX5E_MAX_NUM_RSS)
+		return -EINVAL;
+
+	rss = res->rss[rss_idx];
+	if (!rss)
+		return -ENOENT;
+
+	return mlx5e_rss_set_rxfh(rss, indir, key, hfunc, res->rss_rqns, res->rss_nch);
+}
+
+u8 mlx5e_rx_res_rss_get_hash_fields(struct mlx5e_rx_res *res, enum mlx5_traffic_types tt)
+{
+	struct mlx5e_rss *rss = res->rss[0];
+
+	return mlx5e_rss_get_hash_fields(rss, tt);
+}
+
+int mlx5e_rx_res_rss_set_hash_fields(struct mlx5e_rx_res *res, enum mlx5_traffic_types tt,
+				     u8 rx_hash_fields)
+{
+	struct mlx5e_rss *rss = res->rss[0];
+
+	return mlx5e_rss_set_hash_fields(rss, tt, rx_hash_fields);
+}
+
+int mlx5e_rx_res_rss_cnt(struct mlx5e_rx_res *res)
+{
+	int i, cnt;
+
+	cnt = 0;
+	for (i = 0; i < MLX5E_MAX_NUM_RSS; i++)
+		if (res->rss[i])
+			cnt++;
+
+	return cnt;
+}
+
+int mlx5e_rx_res_rss_index(struct mlx5e_rx_res *res, struct mlx5e_rss *rss)
+{
+	int i;
+
+	if (!rss)
+		return -EINVAL;
+
+	for (i = 0; i < MLX5E_MAX_NUM_RSS; i++)
+		if (rss == res->rss[i])
+			return i;
+
+	return -ENOENT;
+}
+
+struct mlx5e_rss *mlx5e_rx_res_rss_get(struct mlx5e_rx_res *res, u32 rss_idx)
+{
+	if (rss_idx >= MLX5E_MAX_NUM_RSS)
+		return NULL;
+
+	return res->rss[rss_idx];
+}
+
+/* End of API rx_res_rss_* */
+
+struct mlx5e_rx_res *mlx5e_rx_res_alloc(void)
+{
+	return kvzalloc(sizeof(struct mlx5e_rx_res), GFP_KERNEL);
+}
+
+static int mlx5e_rx_res_channels_init(struct mlx5e_rx_res *res)
+{
+	bool inner_ft_support = res->features & MLX5E_RX_RES_FEATURE_INNER_FT;
+	struct mlx5e_tir_builder *builder;
+	int err = 0;
+	int ix;
+
+	builder = mlx5e_tir_builder_alloc(false);
+	if (!builder)
+		return -ENOMEM;
+
+	res->channels = kvcalloc(res->max_nch, sizeof(*res->channels), GFP_KERNEL);
+	if (!res->channels) {
+		err = -ENOMEM;
+		goto out;
+	}
+
+	for (ix = 0; ix < res->max_nch; ix++) {
+		err = mlx5e_rqt_init_direct(&res->channels[ix].direct_rqt,
+					    res->mdev, false, res->drop_rqn);
+		if (err) {
+			mlx5_core_warn(res->mdev, "Failed to create a direct RQT: err = %d, ix = %u\n",
+				       err, ix);
+			goto err_destroy_direct_rqts;
+		}
+	}
+
+	for (ix = 0; ix < res->max_nch; ix++) {
+		mlx5e_tir_builder_build_rqt(builder, res->mdev->mlx5e_res.hw_objs.td.tdn,
+					    mlx5e_rqt_get_rqtn(&res->channels[ix].direct_rqt),
+					    inner_ft_support);
+		mlx5e_tir_builder_build_packet_merge(builder, &res->pkt_merge_param);
+		mlx5e_tir_builder_build_direct(builder);
+
+		err = mlx5e_tir_init(&res->channels[ix].direct_tir, builder, res->mdev, true);
+		if (err) {
+			mlx5_core_warn(res->mdev, "Failed to create a direct TIR: err = %d, ix = %u\n",
+				       err, ix);
+			goto err_destroy_direct_tirs;
+		}
+
+		mlx5e_tir_builder_clear(builder);
+	}
+
+	goto out;
+
+err_destroy_direct_tirs:
+	while (--ix >= 0)
+		mlx5e_tir_destroy(&res->channels[ix].direct_tir);
+
+	ix = res->max_nch;
+err_destroy_direct_rqts:
+	while (--ix >= 0)
+		mlx5e_rqt_destroy(&res->channels[ix].direct_rqt);
+
+	kvfree(res->channels);
+
+out:
+	mlx5e_tir_builder_free(builder);
+
+	return err;
+}
+
+static int mlx5e_rx_res_ptp_init(struct mlx5e_rx_res *res)
+{
+	bool inner_ft_support = res->features & MLX5E_RX_RES_FEATURE_INNER_FT;
+	struct mlx5e_tir_builder *builder;
+	int err;
+
+	builder = mlx5e_tir_builder_alloc(false);
+	if (!builder)
+		return -ENOMEM;
+
+	err = mlx5e_rqt_init_direct(&res->ptp.rqt, res->mdev, false, res->drop_rqn);
+	if (err)
+		goto out;
+
+	/* Separated from the channels RQs, does not share pkt_merge state with them */
+	mlx5e_tir_builder_build_rqt(builder, res->mdev->mlx5e_res.hw_objs.td.tdn,
+				    mlx5e_rqt_get_rqtn(&res->ptp.rqt),
+				    inner_ft_support);
+	mlx5e_tir_builder_build_direct(builder);
+
+	err = mlx5e_tir_init(&res->ptp.tir, builder, res->mdev, true);
+	if (err)
+		goto err_destroy_ptp_rqt;
+
+	goto out;
+
+err_destroy_ptp_rqt:
+	mlx5e_rqt_destroy(&res->ptp.rqt);
+
+out:
+	mlx5e_tir_builder_free(builder);
+	return err;
+}
+
+static void mlx5e_rx_res_channels_destroy(struct mlx5e_rx_res *res)
+{
+	unsigned int ix;
+
+	for (ix = 0; ix < res->max_nch; ix++) {
+		mlx5e_tir_destroy(&res->channels[ix].direct_tir);
+		mlx5e_rqt_destroy(&res->channels[ix].direct_rqt);
+	}
+
+	kvfree(res->channels);
+}
+
+static void mlx5e_rx_res_ptp_destroy(struct mlx5e_rx_res *res)
+{
+	mlx5e_tir_destroy(&res->ptp.tir);
+	mlx5e_rqt_destroy(&res->ptp.rqt);
+}
+
+int mlx5e_rx_res_init(struct mlx5e_rx_res *res, struct mlx5_core_dev *mdev,
+		      enum mlx5e_rx_res_features features, unsigned int max_nch,
+		      u32 drop_rqn, const struct mlx5e_packet_merge_param *init_pkt_merge_param,
+		      unsigned int init_nch)
+{
+	int err;
+
+	res->mdev = mdev;
+	res->features = features;
+	res->max_nch = max_nch;
+	res->drop_rqn = drop_rqn;
+
+	res->pkt_merge_param = *init_pkt_merge_param;
+	init_rwsem(&res->pkt_merge_param_sem);
+
+	err = mlx5e_rx_res_rss_init_def(res, init_nch);
+	if (err)
+		goto err_out;
+
+	err = mlx5e_rx_res_channels_init(res);
+	if (err)
+		goto err_rss_destroy;
+
+	err = mlx5e_rx_res_ptp_init(res);
+	if (err)
+		goto err_channels_destroy;
+
+	return 0;
+
+err_channels_destroy:
+	mlx5e_rx_res_channels_destroy(res);
+err_rss_destroy:
+	__mlx5e_rx_res_rss_destroy(res, 0);
+err_out:
+	return err;
+}
+
+void mlx5e_rx_res_destroy(struct mlx5e_rx_res *res)
+{
+	mlx5e_rx_res_ptp_destroy(res);
+	mlx5e_rx_res_channels_destroy(res);
+	mlx5e_rx_res_rss_destroy_all(res);
+}
+
+void mlx5e_rx_res_free(struct mlx5e_rx_res *res)
+{
+	kvfree(res);
+}
+
+u32 mlx5e_rx_res_get_tirn_direct(struct mlx5e_rx_res *res, unsigned int ix)
+{
+	return mlx5e_tir_get_tirn(&res->channels[ix].direct_tir);
+}
+
+u32 mlx5e_rx_res_get_tirn_rss(struct mlx5e_rx_res *res, enum mlx5_traffic_types tt)
+{
+	struct mlx5e_rss *rss = res->rss[0];
+
+	return mlx5e_rss_get_tirn(rss, tt, false);
+}
+
+u32 mlx5e_rx_res_get_tirn_rss_inner(struct mlx5e_rx_res *res, enum mlx5_traffic_types tt)
+{
+	struct mlx5e_rss *rss = res->rss[0];
+
+	return mlx5e_rss_get_tirn(rss, tt, true);
+}
+
+u32 mlx5e_rx_res_get_tirn_ptp(struct mlx5e_rx_res *res)
+{
+	WARN_ON(!(res->features & MLX5E_RX_RES_FEATURE_PTP));
+	return mlx5e_tir_get_tirn(&res->ptp.tir);
+}
+
+static u32 mlx5e_rx_res_get_rqtn_direct(struct mlx5e_rx_res *res, unsigned int ix)
+{
+	return mlx5e_rqt_get_rqtn(&res->channels[ix].direct_rqt);
+}
+
+static void mlx5e_rx_res_channel_activate_direct(struct mlx5e_rx_res *res,
+						 struct mlx5e_channels *chs,
+						 unsigned int ix)
+{
+	u32 rqn = res->rss_rqns[ix];
+	int err;
+
+	err = mlx5e_rqt_redirect_direct(&res->channels[ix].direct_rqt, rqn);
+	if (err)
+		mlx5_core_warn(res->mdev, "Failed to redirect direct RQT %#x to RQ %#x (channel %u): err = %d\n",
+			       mlx5e_rqt_get_rqtn(&res->channels[ix].direct_rqt),
+			       rqn, ix, err);
+}
+
+static void mlx5e_rx_res_channel_deactivate_direct(struct mlx5e_rx_res *res,
+						   unsigned int ix)
+{
+	int err;
+
+	err = mlx5e_rqt_redirect_direct(&res->channels[ix].direct_rqt, res->drop_rqn);
+	if (err)
+		mlx5_core_warn(res->mdev, "Failed to redirect direct RQT %#x to drop RQ %#x (channel %u): err = %d\n",
+			       mlx5e_rqt_get_rqtn(&res->channels[ix].direct_rqt),
+			       res->drop_rqn, ix, err);
+}
+
+void mlx5e_rx_res_channels_activate(struct mlx5e_rx_res *res, struct mlx5e_channels *chs)
+{
+	unsigned int nch, ix;
+	int err;
+
+	nch = mlx5e_channels_get_num(chs);
+
+	for (ix = 0; ix < chs->num; ix++) {
+		if (mlx5e_channels_is_xsk(chs, ix))
+			mlx5e_channels_get_xsk_rqn(chs, ix, &res->rss_rqns[ix]);
+		else
+			mlx5e_channels_get_regular_rqn(chs, ix, &res->rss_rqns[ix]);
+	}
+	res->rss_nch = chs->num;
+
+	mlx5e_rx_res_rss_enable(res);
+
+	for (ix = 0; ix < nch; ix++)
+		mlx5e_rx_res_channel_activate_direct(res, chs, ix);
+	for (ix = nch; ix < res->max_nch; ix++)
+		mlx5e_rx_res_channel_deactivate_direct(res, ix);
+
+	if (res->features & MLX5E_RX_RES_FEATURE_PTP) {
+		u32 rqn;
+
+		if (!mlx5e_channels_get_ptp_rqn(chs, &rqn))
+			rqn = res->drop_rqn;
+
+		err = mlx5e_rqt_redirect_direct(&res->ptp.rqt, rqn);
+		if (err)
+			mlx5_core_warn(res->mdev, "Failed to redirect direct RQT %#x to RQ %#x (PTP): err = %d\n",
+				       mlx5e_rqt_get_rqtn(&res->ptp.rqt),
+				       rqn, err);
+	}
+}
+
+void mlx5e_rx_res_channels_deactivate(struct mlx5e_rx_res *res)
+{
+	unsigned int ix;
+	int err;
+
+	mlx5e_rx_res_rss_disable(res);
+
+	for (ix = 0; ix < res->max_nch; ix++)
+		mlx5e_rx_res_channel_deactivate_direct(res, ix);
+
+	if (res->features & MLX5E_RX_RES_FEATURE_PTP) {
+		err = mlx5e_rqt_redirect_direct(&res->ptp.rqt, res->drop_rqn);
+		if (err)
+			mlx5_core_warn(res->mdev, "Failed to redirect direct RQT %#x to drop RQ %#x (PTP): err = %d\n",
+				       mlx5e_rqt_get_rqtn(&res->ptp.rqt),
+				       res->drop_rqn, err);
+	}
+}
+
+void mlx5e_rx_res_xsk_update(struct mlx5e_rx_res *res, struct mlx5e_channels *chs,
+			     unsigned int ix, bool xsk)
+{
+	if (xsk)
+		mlx5e_channels_get_xsk_rqn(chs, ix, &res->rss_rqns[ix]);
+	else
+		mlx5e_channels_get_regular_rqn(chs, ix, &res->rss_rqns[ix]);
+
+	mlx5e_rx_res_rss_enable(res);
+
+	mlx5e_rx_res_channel_activate_direct(res, chs, ix);
+}
+
+int mlx5e_rx_res_packet_merge_set_param(struct mlx5e_rx_res *res,
+					struct mlx5e_packet_merge_param *pkt_merge_param)
+{
+	struct mlx5e_tir_builder *builder;
+	int err, final_err;
+	unsigned int ix;
+
+	builder = mlx5e_tir_builder_alloc(true);
+	if (!builder)
+		return -ENOMEM;
+
+	down_write(&res->pkt_merge_param_sem);
+	res->pkt_merge_param = *pkt_merge_param;
+
+	mlx5e_tir_builder_build_packet_merge(builder, pkt_merge_param);
+
+	final_err = 0;
+
+	for (ix = 0; ix < MLX5E_MAX_NUM_RSS; ix++) {
+		struct mlx5e_rss *rss = res->rss[ix];
+
+		if (!rss)
+			continue;
+
+		err = mlx5e_rss_packet_merge_set_param(rss, pkt_merge_param);
+		if (err)
+			final_err = final_err ? : err;
+	}
+
+	for (ix = 0; ix < res->max_nch; ix++) {
+		err = mlx5e_tir_modify(&res->channels[ix].direct_tir, builder);
+		if (err) {
+			mlx5_core_warn(res->mdev, "Failed to update packet merge state of direct TIR %#x for channel %u: err = %d\n",
+				       mlx5e_tir_get_tirn(&res->channels[ix].direct_tir), ix, err);
+			if (!final_err)
+				final_err = err;
+		}
+	}
+
+	up_write(&res->pkt_merge_param_sem);
+	mlx5e_tir_builder_free(builder);
+	return final_err;
+}
+
+struct mlx5e_rss_params_hash mlx5e_rx_res_get_current_hash(struct mlx5e_rx_res *res)
+{
+	return mlx5e_rss_get_hash(res->rss[0]);
+}
+
+int mlx5e_rx_res_tls_tir_create(struct mlx5e_rx_res *res, unsigned int rxq,
+				struct mlx5e_tir *tir)
+{
+	bool inner_ft_support = res->features & MLX5E_RX_RES_FEATURE_INNER_FT;
+	struct mlx5e_tir_builder *builder;
+	u32 rqtn;
+	int err;
+
+	builder = mlx5e_tir_builder_alloc(false);
+	if (!builder)
+		return -ENOMEM;
+
+	rqtn = mlx5e_rx_res_get_rqtn_direct(res, rxq);
+
+	mlx5e_tir_builder_build_rqt(builder, res->mdev->mlx5e_res.hw_objs.td.tdn, rqtn,
+				    inner_ft_support);
+	mlx5e_tir_builder_build_direct(builder);
+	mlx5e_tir_builder_build_tls(builder);
+	down_read(&res->pkt_merge_param_sem);
+	mlx5e_tir_builder_build_packet_merge(builder, &res->pkt_merge_param);
+	err = mlx5e_tir_init(tir, builder, res->mdev, false);
+	up_read(&res->pkt_merge_param_sem);
+
+	mlx5e_tir_builder_free(builder);
+
+	return err;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/rx_res.h b/drivers/net/ethernet/mellanox/mlx5/core/en/rx_res.h
new file mode 100644
index 000000000..5d5f64fab
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/rx_res.h
@@ -0,0 +1,69 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2021, Mellanox Technologies inc. All rights reserved. */
+
+#ifndef __MLX5_EN_RX_RES_H__
+#define __MLX5_EN_RX_RES_H__
+
+#include <linux/kernel.h>
+#include "rqt.h"
+#include "tir.h"
+#include "fs.h"
+#include "rss.h"
+
+struct mlx5e_rx_res;
+
+struct mlx5e_channels;
+struct mlx5e_rss_params_hash;
+
+enum mlx5e_rx_res_features {
+	MLX5E_RX_RES_FEATURE_INNER_FT = BIT(0),
+	MLX5E_RX_RES_FEATURE_PTP = BIT(1),
+};
+
+/* Setup */
+struct mlx5e_rx_res *mlx5e_rx_res_alloc(void);
+int mlx5e_rx_res_init(struct mlx5e_rx_res *res, struct mlx5_core_dev *mdev,
+		      enum mlx5e_rx_res_features features, unsigned int max_nch,
+		      u32 drop_rqn, const struct mlx5e_packet_merge_param *init_pkt_merge_param,
+		      unsigned int init_nch);
+void mlx5e_rx_res_destroy(struct mlx5e_rx_res *res);
+void mlx5e_rx_res_free(struct mlx5e_rx_res *res);
+
+/* TIRN getters for flow steering */
+u32 mlx5e_rx_res_get_tirn_direct(struct mlx5e_rx_res *res, unsigned int ix);
+u32 mlx5e_rx_res_get_tirn_rss(struct mlx5e_rx_res *res, enum mlx5_traffic_types tt);
+u32 mlx5e_rx_res_get_tirn_rss_inner(struct mlx5e_rx_res *res, enum mlx5_traffic_types tt);
+u32 mlx5e_rx_res_get_tirn_ptp(struct mlx5e_rx_res *res);
+
+/* Activate/deactivate API */
+void mlx5e_rx_res_channels_activate(struct mlx5e_rx_res *res, struct mlx5e_channels *chs);
+void mlx5e_rx_res_channels_deactivate(struct mlx5e_rx_res *res);
+void mlx5e_rx_res_xsk_update(struct mlx5e_rx_res *res, struct mlx5e_channels *chs,
+			     unsigned int ix, bool xsk);
+
+/* Configuration API */
+void mlx5e_rx_res_rss_set_indir_uniform(struct mlx5e_rx_res *res, unsigned int nch);
+int mlx5e_rx_res_rss_get_rxfh(struct mlx5e_rx_res *res, u32 rss_idx,
+			      u32 *indir, u8 *key, u8 *hfunc);
+int mlx5e_rx_res_rss_set_rxfh(struct mlx5e_rx_res *res, u32 rss_idx,
+			      const u32 *indir, const u8 *key, const u8 *hfunc);
+
+u8 mlx5e_rx_res_rss_get_hash_fields(struct mlx5e_rx_res *res, enum mlx5_traffic_types tt);
+int mlx5e_rx_res_rss_set_hash_fields(struct mlx5e_rx_res *res, enum mlx5_traffic_types tt,
+				     u8 rx_hash_fields);
+int mlx5e_rx_res_packet_merge_set_param(struct mlx5e_rx_res *res,
+					struct mlx5e_packet_merge_param *pkt_merge_param);
+
+int mlx5e_rx_res_rss_init(struct mlx5e_rx_res *res, u32 *rss_idx, unsigned int init_nch);
+int mlx5e_rx_res_rss_destroy(struct mlx5e_rx_res *res, u32 rss_idx);
+int mlx5e_rx_res_rss_cnt(struct mlx5e_rx_res *res);
+int mlx5e_rx_res_rss_index(struct mlx5e_rx_res *res, struct mlx5e_rss *rss);
+struct mlx5e_rss *mlx5e_rx_res_rss_get(struct mlx5e_rx_res *res, u32 rss_idx);
+
+/* Workaround for hairpin */
+struct mlx5e_rss_params_hash mlx5e_rx_res_get_current_hash(struct mlx5e_rx_res *res);
+
+/* Accel TIRs */
+int mlx5e_rx_res_tls_tir_create(struct mlx5e_rx_res *res, unsigned int rxq,
+				struct mlx5e_tir *tir);
+#endif /* __MLX5_EN_RX_RES_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/selq.c b/drivers/net/ethernet/mellanox/mlx5/core/en/selq.c
new file mode 100644
index 000000000..f675b1926
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/selq.c
@@ -0,0 +1,266 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. All rights reserved. */
+
+#include "selq.h"
+#include <linux/slab.h>
+#include <linux/netdevice.h>
+#include <linux/rcupdate.h>
+#include "en.h"
+#include "en/ptp.h"
+#include "en/htb.h"
+
+struct mlx5e_selq_params {
+	unsigned int num_regular_queues;
+	unsigned int num_channels;
+	unsigned int num_tcs;
+	union {
+		u8 is_special_queues;
+		struct {
+			bool is_htb : 1;
+			bool is_ptp : 1;
+		};
+	};
+	u16 htb_maj_id;
+	u16 htb_defcls;
+};
+
+int mlx5e_selq_init(struct mlx5e_selq *selq, struct mutex *state_lock)
+{
+	struct mlx5e_selq_params *init_params;
+
+	selq->state_lock = state_lock;
+
+	selq->standby = kvzalloc(sizeof(*selq->standby), GFP_KERNEL);
+	if (!selq->standby)
+		return -ENOMEM;
+
+	init_params = kvzalloc(sizeof(*selq->active), GFP_KERNEL);
+	if (!init_params) {
+		kvfree(selq->standby);
+		selq->standby = NULL;
+		return -ENOMEM;
+	}
+	/* Assign dummy values, so that mlx5e_select_queue won't crash. */
+	*init_params = (struct mlx5e_selq_params) {
+		.num_regular_queues = 1,
+		.num_channels = 1,
+		.num_tcs = 1,
+		.is_htb = false,
+		.is_ptp = false,
+		.htb_maj_id = 0,
+		.htb_defcls = 0,
+	};
+	rcu_assign_pointer(selq->active, init_params);
+
+	return 0;
+}
+
+void mlx5e_selq_cleanup(struct mlx5e_selq *selq)
+{
+	WARN_ON_ONCE(selq->is_prepared);
+
+	kvfree(selq->standby);
+	selq->standby = NULL;
+	selq->is_prepared = true;
+
+	mlx5e_selq_apply(selq);
+
+	kvfree(selq->standby);
+	selq->standby = NULL;
+}
+
+void mlx5e_selq_prepare_params(struct mlx5e_selq *selq, struct mlx5e_params *params)
+{
+	struct mlx5e_selq_params *selq_active;
+
+	lockdep_assert_held(selq->state_lock);
+	WARN_ON_ONCE(selq->is_prepared);
+
+	selq->is_prepared = true;
+
+	selq_active = rcu_dereference_protected(selq->active,
+						lockdep_is_held(selq->state_lock));
+	*selq->standby = *selq_active;
+	selq->standby->num_channels = params->num_channels;
+	selq->standby->num_tcs = mlx5e_get_dcb_num_tc(params);
+	selq->standby->num_regular_queues =
+		selq->standby->num_channels * selq->standby->num_tcs;
+	selq->standby->is_ptp = MLX5E_GET_PFLAG(params, MLX5E_PFLAG_TX_PORT_TS);
+}
+
+bool mlx5e_selq_is_htb_enabled(struct mlx5e_selq *selq)
+{
+	struct mlx5e_selq_params *selq_active =
+		rcu_dereference_protected(selq->active, lockdep_is_held(selq->state_lock));
+
+	return selq_active->htb_maj_id;
+}
+
+void mlx5e_selq_prepare_htb(struct mlx5e_selq *selq, u16 htb_maj_id, u16 htb_defcls)
+{
+	struct mlx5e_selq_params *selq_active;
+
+	lockdep_assert_held(selq->state_lock);
+	WARN_ON_ONCE(selq->is_prepared);
+
+	selq->is_prepared = true;
+
+	selq_active = rcu_dereference_protected(selq->active,
+						lockdep_is_held(selq->state_lock));
+	*selq->standby = *selq_active;
+	selq->standby->is_htb = htb_maj_id;
+	selq->standby->htb_maj_id = htb_maj_id;
+	selq->standby->htb_defcls = htb_defcls;
+}
+
+void mlx5e_selq_apply(struct mlx5e_selq *selq)
+{
+	struct mlx5e_selq_params *old_params;
+
+	WARN_ON_ONCE(!selq->is_prepared);
+
+	selq->is_prepared = false;
+
+	old_params = rcu_replace_pointer(selq->active, selq->standby,
+					 lockdep_is_held(selq->state_lock));
+	synchronize_net(); /* Wait until ndo_select_queue starts emitting correct values. */
+	selq->standby = old_params;
+}
+
+void mlx5e_selq_cancel(struct mlx5e_selq *selq)
+{
+	lockdep_assert_held(selq->state_lock);
+	WARN_ON_ONCE(!selq->is_prepared);
+
+	selq->is_prepared = false;
+}
+
+#ifdef CONFIG_MLX5_CORE_EN_DCB
+static int mlx5e_get_dscp_up(struct mlx5e_priv *priv, struct sk_buff *skb)
+{
+	int dscp_cp = 0;
+
+	if (skb->protocol == htons(ETH_P_IP))
+		dscp_cp = ipv4_get_dsfield(ip_hdr(skb)) >> 2;
+	else if (skb->protocol == htons(ETH_P_IPV6))
+		dscp_cp = ipv6_get_dsfield(ipv6_hdr(skb)) >> 2;
+
+	return priv->dcbx_dp.dscp2prio[dscp_cp];
+}
+#endif
+
+static int mlx5e_get_up(struct mlx5e_priv *priv, struct sk_buff *skb)
+{
+#ifdef CONFIG_MLX5_CORE_EN_DCB
+	if (READ_ONCE(priv->dcbx_dp.trust_state) == MLX5_QPTS_TRUST_DSCP)
+		return mlx5e_get_dscp_up(priv, skb);
+#endif
+	if (skb_vlan_tag_present(skb))
+		return skb_vlan_tag_get_prio(skb);
+	return 0;
+}
+
+static u16 mlx5e_select_ptpsq(struct net_device *dev, struct sk_buff *skb,
+			      struct mlx5e_selq_params *selq)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	int up;
+
+	up = selq->num_tcs > 1 ? mlx5e_get_up(priv, skb) : 0;
+
+	return selq->num_regular_queues + up;
+}
+
+static int mlx5e_select_htb_queue(struct mlx5e_priv *priv, struct sk_buff *skb,
+				  struct mlx5e_selq_params *selq)
+{
+	u16 classid;
+
+	/* Order maj_id before defcls - pairs with mlx5e_htb_root_add. */
+	if ((TC_H_MAJ(skb->priority) >> 16) == selq->htb_maj_id)
+		classid = TC_H_MIN(skb->priority);
+	else
+		classid = selq->htb_defcls;
+
+	if (!classid)
+		return 0;
+
+	return mlx5e_htb_get_txq_by_classid(priv->htb, classid);
+}
+
+u16 mlx5e_select_queue(struct net_device *dev, struct sk_buff *skb,
+		       struct net_device *sb_dev)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	struct mlx5e_selq_params *selq;
+	int txq_ix, up;
+
+	selq = rcu_dereference_bh(priv->selq.active);
+
+	/* This is a workaround needed only for the mlx5e_netdev_change_profile
+	 * flow that zeroes out the whole priv without unregistering the netdev
+	 * and without preventing ndo_select_queue from being called.
+	 */
+	if (unlikely(!selq))
+		return 0;
+
+	if (likely(!selq->is_special_queues)) {
+		/* No special queues, netdev_pick_tx returns one of the regular ones. */
+
+		txq_ix = netdev_pick_tx(dev, skb, NULL);
+
+		if (selq->num_tcs <= 1)
+			return txq_ix;
+
+		up = mlx5e_get_up(priv, skb);
+
+		/* Normalize any picked txq_ix to [0, num_channels),
+		 * So we can return a txq_ix that matches the channel and
+		 * packet UP.
+		 */
+		return mlx5e_txq_to_ch_ix(txq_ix, selq->num_channels) +
+			up * selq->num_channels;
+	}
+
+	if (unlikely(selq->htb_maj_id)) {
+		/* num_tcs == 1, shortcut for PTP */
+
+		txq_ix = mlx5e_select_htb_queue(priv, skb, selq);
+		if (txq_ix > 0)
+			return txq_ix;
+
+		if (unlikely(selq->is_ptp && mlx5e_use_ptpsq(skb)))
+			return selq->num_channels;
+
+		txq_ix = netdev_pick_tx(dev, skb, NULL);
+
+		/* Fix netdev_pick_tx() not to choose ptp_channel and HTB txqs.
+		 * If they are selected, switch to regular queues.
+		 * Driver to select these queues only at mlx5e_select_ptpsq()
+		 * and mlx5e_select_htb_queue().
+		 */
+		return mlx5e_txq_to_ch_ix_htb(txq_ix, selq->num_channels);
+	}
+
+	/* PTP is enabled */
+
+	if (mlx5e_use_ptpsq(skb))
+		return mlx5e_select_ptpsq(dev, skb, selq);
+
+	txq_ix = netdev_pick_tx(dev, skb, NULL);
+
+	/* Normalize any picked txq_ix to [0, num_channels). Queues in range
+	 * [0, num_regular_queues) will be mapped to the corresponding channel
+	 * index, so that we can apply the packet's UP (if num_tcs > 1).
+	 * If netdev_pick_tx() picks ptp_channel, switch to a regular queue,
+	 * because driver should select the PTP only at mlx5e_select_ptpsq().
+	 */
+	txq_ix = mlx5e_txq_to_ch_ix(txq_ix, selq->num_channels);
+
+	if (selq->num_tcs <= 1)
+		return txq_ix;
+
+	up = mlx5e_get_up(priv, skb);
+
+	return txq_ix + up * selq->num_channels;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/selq.h b/drivers/net/ethernet/mellanox/mlx5/core/en/selq.h
new file mode 100644
index 000000000..fd590f80e
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/selq.h
@@ -0,0 +1,53 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. All rights reserved. */
+
+#ifndef __MLX5_EN_SELQ_H__
+#define __MLX5_EN_SELQ_H__
+
+#include <linux/kernel.h>
+
+struct mlx5e_selq_params;
+
+struct mlx5e_selq {
+	struct mlx5e_selq_params __rcu *active;
+	struct mlx5e_selq_params *standby;
+	struct mutex *state_lock; /* points to priv->state_lock */
+	bool is_prepared;
+};
+
+struct mlx5e_params;
+struct net_device;
+struct sk_buff;
+
+int mlx5e_selq_init(struct mlx5e_selq *selq, struct mutex *state_lock);
+void mlx5e_selq_cleanup(struct mlx5e_selq *selq);
+void mlx5e_selq_prepare_params(struct mlx5e_selq *selq, struct mlx5e_params *params);
+void mlx5e_selq_prepare_htb(struct mlx5e_selq *selq, u16 htb_maj_id, u16 htb_defcls);
+bool mlx5e_selq_is_htb_enabled(struct mlx5e_selq *selq);
+void mlx5e_selq_apply(struct mlx5e_selq *selq);
+void mlx5e_selq_cancel(struct mlx5e_selq *selq);
+
+static inline u16 mlx5e_txq_to_ch_ix(u16 txq, u16 num_channels)
+{
+	while (unlikely(txq >= num_channels))
+		txq -= num_channels;
+	return txq;
+}
+
+static inline u16 mlx5e_txq_to_ch_ix_htb(u16 txq, u16 num_channels)
+{
+	if (unlikely(txq >= num_channels)) {
+		if (unlikely(txq >= num_channels << 3))
+			txq %= num_channels;
+		else
+			do
+				txq -= num_channels;
+			while (txq >= num_channels);
+	}
+	return txq;
+}
+
+u16 mlx5e_select_queue(struct net_device *dev, struct sk_buff *skb,
+		       struct net_device *sb_dev);
+
+#endif /* __MLX5_EN_SELQ_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/accept.c b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/accept.c
new file mode 100644
index 000000000..21aab9635
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/accept.c
@@ -0,0 +1,31 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+// Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+
+#include "act.h"
+#include "en/tc_priv.h"
+
+static bool
+tc_act_can_offload_accept(struct mlx5e_tc_act_parse_state *parse_state,
+			  const struct flow_action_entry *act,
+			  int act_index,
+			  struct mlx5_flow_attr *attr)
+{
+	return true;
+}
+
+static int
+tc_act_parse_accept(struct mlx5e_tc_act_parse_state *parse_state,
+		    const struct flow_action_entry *act,
+		    struct mlx5e_priv *priv,
+		    struct mlx5_flow_attr *attr)
+{
+	attr->action |= MLX5_FLOW_CONTEXT_ACTION_FWD_DEST;
+	attr->flags |= MLX5_ATTR_FLAG_ACCEPT;
+
+	return 0;
+}
+
+struct mlx5e_tc_act mlx5e_tc_act_accept = {
+	.can_offload = tc_act_can_offload_accept,
+	.parse_action = tc_act_parse_accept,
+};
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/act.c b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/act.c
new file mode 100644
index 000000000..3337241cf
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/act.c
@@ -0,0 +1,154 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+// Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+
+#include "act.h"
+#include "en/tc/post_act.h"
+#include "en/tc_priv.h"
+#include "mlx5_core.h"
+
+static struct mlx5e_tc_act *tc_acts_fdb[NUM_FLOW_ACTIONS] = {
+	[FLOW_ACTION_ACCEPT] = &mlx5e_tc_act_accept,
+	[FLOW_ACTION_DROP] = &mlx5e_tc_act_drop,
+	[FLOW_ACTION_TRAP] = &mlx5e_tc_act_trap,
+	[FLOW_ACTION_GOTO] = &mlx5e_tc_act_goto,
+	[FLOW_ACTION_REDIRECT] = &mlx5e_tc_act_mirred,
+	[FLOW_ACTION_MIRRED] = &mlx5e_tc_act_mirred,
+	[FLOW_ACTION_REDIRECT_INGRESS] = &mlx5e_tc_act_redirect_ingress,
+	[FLOW_ACTION_VLAN_PUSH] = &mlx5e_tc_act_vlan,
+	[FLOW_ACTION_VLAN_POP] = &mlx5e_tc_act_vlan,
+	[FLOW_ACTION_VLAN_MANGLE] = &mlx5e_tc_act_vlan_mangle,
+	[FLOW_ACTION_TUNNEL_ENCAP] = &mlx5e_tc_act_tun_encap,
+	[FLOW_ACTION_TUNNEL_DECAP] = &mlx5e_tc_act_tun_decap,
+	[FLOW_ACTION_MANGLE] = &mlx5e_tc_act_pedit,
+	[FLOW_ACTION_ADD] = &mlx5e_tc_act_pedit,
+	[FLOW_ACTION_CSUM] = &mlx5e_tc_act_csum,
+	[FLOW_ACTION_PTYPE] = &mlx5e_tc_act_ptype,
+	[FLOW_ACTION_SAMPLE] = &mlx5e_tc_act_sample,
+	[FLOW_ACTION_POLICE] = &mlx5e_tc_act_police,
+	[FLOW_ACTION_CT] = &mlx5e_tc_act_ct,
+	[FLOW_ACTION_MPLS_PUSH] = &mlx5e_tc_act_mpls_push,
+	[FLOW_ACTION_MPLS_POP] = &mlx5e_tc_act_mpls_pop,
+	[FLOW_ACTION_VLAN_PUSH_ETH] = &mlx5e_tc_act_vlan,
+	[FLOW_ACTION_VLAN_POP_ETH] = &mlx5e_tc_act_vlan,
+};
+
+static struct mlx5e_tc_act *tc_acts_nic[NUM_FLOW_ACTIONS] = {
+	[FLOW_ACTION_ACCEPT] = &mlx5e_tc_act_accept,
+	[FLOW_ACTION_DROP] = &mlx5e_tc_act_drop,
+	[FLOW_ACTION_GOTO] = &mlx5e_tc_act_goto,
+	[FLOW_ACTION_REDIRECT] = &mlx5e_tc_act_mirred_nic,
+	[FLOW_ACTION_MANGLE] = &mlx5e_tc_act_pedit,
+	[FLOW_ACTION_ADD] = &mlx5e_tc_act_pedit,
+	[FLOW_ACTION_CSUM] = &mlx5e_tc_act_csum,
+	[FLOW_ACTION_MARK] = &mlx5e_tc_act_mark,
+	[FLOW_ACTION_CT] = &mlx5e_tc_act_ct,
+};
+
+/**
+ * mlx5e_tc_act_get() - Get an action parser for an action id.
+ * @act_id: Flow action id.
+ * @ns_type: flow namespace type.
+ */
+struct mlx5e_tc_act *
+mlx5e_tc_act_get(enum flow_action_id act_id,
+		 enum mlx5_flow_namespace_type ns_type)
+{
+	struct mlx5e_tc_act **tc_acts;
+
+	tc_acts = ns_type == MLX5_FLOW_NAMESPACE_FDB ? tc_acts_fdb : tc_acts_nic;
+
+	return tc_acts[act_id];
+}
+
+/**
+ * mlx5e_tc_act_init_parse_state() - Init a new parse_state.
+ * @parse_state: Parsing state.
+ * @flow:        mlx5e tc flow being handled.
+ * @flow_action: flow action to parse.
+ * @extack:      to set an error msg.
+ *
+ * The same parse_state should be passed to action parsers
+ * for tracking the current parsing state.
+ */
+void
+mlx5e_tc_act_init_parse_state(struct mlx5e_tc_act_parse_state *parse_state,
+			      struct mlx5e_tc_flow *flow,
+			      struct flow_action *flow_action,
+			      struct netlink_ext_ack *extack)
+{
+	memset(parse_state, 0, sizeof(*parse_state));
+	parse_state->flow = flow;
+	parse_state->extack = extack;
+	parse_state->flow_action = flow_action;
+}
+
+void
+mlx5e_tc_act_reorder_flow_actions(struct flow_action *flow_action,
+				  struct mlx5e_tc_flow_action *flow_action_reorder)
+{
+	struct flow_action_entry *act;
+	int i, j = 0;
+
+	flow_action_for_each(i, act, flow_action) {
+		/* Add CT action to be first. */
+		if (act->id == FLOW_ACTION_CT)
+			flow_action_reorder->entries[j++] = act;
+	}
+
+	flow_action_for_each(i, act, flow_action) {
+		if (act->id == FLOW_ACTION_CT)
+			continue;
+		flow_action_reorder->entries[j++] = act;
+	}
+}
+
+int
+mlx5e_tc_act_post_parse(struct mlx5e_tc_act_parse_state *parse_state,
+			struct flow_action *flow_action,
+			struct mlx5_flow_attr *attr,
+			enum mlx5_flow_namespace_type ns_type)
+{
+	struct flow_action_entry *act;
+	struct mlx5e_tc_act *tc_act;
+	struct mlx5e_priv *priv;
+	int err = 0, i;
+
+	priv = parse_state->flow->priv;
+
+	flow_action_for_each(i, act, flow_action) {
+		tc_act = mlx5e_tc_act_get(act->id, ns_type);
+		if (!tc_act || !tc_act->post_parse)
+			continue;
+
+		err = tc_act->post_parse(parse_state, priv, attr);
+		if (err)
+			goto out;
+	}
+
+out:
+	return err;
+}
+
+int
+mlx5e_tc_act_set_next_post_act(struct mlx5e_tc_flow *flow,
+			       struct mlx5_flow_attr *attr,
+			       struct mlx5_flow_attr *next_attr)
+{
+	struct mlx5_core_dev *mdev = flow->priv->mdev;
+	struct mlx5e_tc_mod_hdr_acts *mod_acts;
+	int err;
+
+	mod_acts = &attr->parse_attr->mod_hdr_acts;
+
+	/* Set handle on current post act rule to next post act rule. */
+	err = mlx5e_tc_post_act_set_handle(mdev, next_attr->post_act_handle, mod_acts);
+	if (err) {
+		mlx5_core_warn(mdev, "Failed setting post action handle");
+		return err;
+	}
+
+	attr->action |= MLX5_FLOW_CONTEXT_ACTION_FWD_DEST |
+			MLX5_FLOW_CONTEXT_ACTION_MOD_HDR;
+
+	return 0;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/act.h b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/act.h
new file mode 100644
index 000000000..e1570ff05
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/act.h
@@ -0,0 +1,116 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. All rights reserved. */
+
+#ifndef __MLX5_EN_TC_ACT_H__
+#define __MLX5_EN_TC_ACT_H__
+
+#include <net/tc_act/tc_pedit.h>
+#include <net/flow_offload.h>
+#include <linux/netlink.h>
+#include "eswitch.h"
+#include "pedit.h"
+
+struct mlx5_flow_attr;
+
+struct mlx5e_tc_act_parse_state {
+	struct flow_action *flow_action;
+	struct mlx5e_tc_flow *flow;
+	struct netlink_ext_ack *extack;
+	u32 actions;
+	bool ct;
+	bool ct_clear;
+	bool encap;
+	bool decap;
+	bool mpls_push;
+	bool eth_push;
+	bool eth_pop;
+	bool ptype_host;
+	const struct ip_tunnel_info *tun_info;
+	struct mlx5e_mpls_info mpls_info;
+	int ifindexes[MLX5_MAX_FLOW_FWD_VPORTS];
+	int if_count;
+	struct mlx5_tc_ct_priv *ct_priv;
+};
+
+struct mlx5e_tc_act {
+	bool (*can_offload)(struct mlx5e_tc_act_parse_state *parse_state,
+			    const struct flow_action_entry *act,
+			    int act_index,
+			    struct mlx5_flow_attr *attr);
+
+	int (*parse_action)(struct mlx5e_tc_act_parse_state *parse_state,
+			    const struct flow_action_entry *act,
+			    struct mlx5e_priv *priv,
+			    struct mlx5_flow_attr *attr);
+
+	int (*post_parse)(struct mlx5e_tc_act_parse_state *parse_state,
+			  struct mlx5e_priv *priv,
+			  struct mlx5_flow_attr *attr);
+
+	bool (*is_multi_table_act)(struct mlx5e_priv *priv,
+				   const struct flow_action_entry *act,
+				   struct mlx5_flow_attr *attr);
+
+	int (*offload_action)(struct mlx5e_priv *priv,
+			      struct flow_offload_action *fl_act,
+			      struct flow_action_entry *act);
+
+	int (*destroy_action)(struct mlx5e_priv *priv,
+			      struct flow_offload_action *fl_act);
+
+	int (*stats_action)(struct mlx5e_priv *priv,
+			    struct flow_offload_action *fl_act);
+};
+
+struct mlx5e_tc_flow_action {
+	unsigned int num_entries;
+	struct flow_action_entry **entries;
+};
+
+extern struct mlx5e_tc_act mlx5e_tc_act_drop;
+extern struct mlx5e_tc_act mlx5e_tc_act_trap;
+extern struct mlx5e_tc_act mlx5e_tc_act_accept;
+extern struct mlx5e_tc_act mlx5e_tc_act_mark;
+extern struct mlx5e_tc_act mlx5e_tc_act_goto;
+extern struct mlx5e_tc_act mlx5e_tc_act_tun_encap;
+extern struct mlx5e_tc_act mlx5e_tc_act_tun_decap;
+extern struct mlx5e_tc_act mlx5e_tc_act_csum;
+extern struct mlx5e_tc_act mlx5e_tc_act_pedit;
+extern struct mlx5e_tc_act mlx5e_tc_act_vlan;
+extern struct mlx5e_tc_act mlx5e_tc_act_vlan_mangle;
+extern struct mlx5e_tc_act mlx5e_tc_act_mpls_push;
+extern struct mlx5e_tc_act mlx5e_tc_act_mpls_pop;
+extern struct mlx5e_tc_act mlx5e_tc_act_mirred;
+extern struct mlx5e_tc_act mlx5e_tc_act_mirred_nic;
+extern struct mlx5e_tc_act mlx5e_tc_act_ct;
+extern struct mlx5e_tc_act mlx5e_tc_act_sample;
+extern struct mlx5e_tc_act mlx5e_tc_act_ptype;
+extern struct mlx5e_tc_act mlx5e_tc_act_redirect_ingress;
+extern struct mlx5e_tc_act mlx5e_tc_act_police;
+
+struct mlx5e_tc_act *
+mlx5e_tc_act_get(enum flow_action_id act_id,
+		 enum mlx5_flow_namespace_type ns_type);
+
+void
+mlx5e_tc_act_init_parse_state(struct mlx5e_tc_act_parse_state *parse_state,
+			      struct mlx5e_tc_flow *flow,
+			      struct flow_action *flow_action,
+			      struct netlink_ext_ack *extack);
+
+void
+mlx5e_tc_act_reorder_flow_actions(struct flow_action *flow_action,
+				  struct mlx5e_tc_flow_action *flow_action_reorder);
+
+int
+mlx5e_tc_act_post_parse(struct mlx5e_tc_act_parse_state *parse_state,
+			struct flow_action *flow_action,
+			struct mlx5_flow_attr *attr,
+			enum mlx5_flow_namespace_type ns_type);
+
+int
+mlx5e_tc_act_set_next_post_act(struct mlx5e_tc_flow *flow,
+			       struct mlx5_flow_attr *attr,
+			       struct mlx5_flow_attr *next_attr);
+
+#endif /* __MLX5_EN_TC_ACT_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/csum.c b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/csum.c
new file mode 100644
index 000000000..c0f08ae6a
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/csum.c
@@ -0,0 +1,62 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+// Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+
+#include <linux/tc_act/tc_csum.h>
+#include "act.h"
+#include "en/tc_priv.h"
+
+static bool
+csum_offload_supported(struct mlx5e_priv *priv,
+		       u32 action,
+		       u32 update_flags,
+		       struct netlink_ext_ack *extack)
+{
+	u32 prot_flags = TCA_CSUM_UPDATE_FLAG_IPV4HDR | TCA_CSUM_UPDATE_FLAG_TCP |
+			 TCA_CSUM_UPDATE_FLAG_UDP;
+
+	/*  The HW recalcs checksums only if re-writing headers */
+	if (!(action & MLX5_FLOW_CONTEXT_ACTION_MOD_HDR)) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "TC csum action is only offloaded with pedit");
+		netdev_warn(priv->netdev,
+			    "TC csum action is only offloaded with pedit\n");
+		return false;
+	}
+
+	if (update_flags & ~prot_flags) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "can't offload TC csum action for some header/s");
+		netdev_warn(priv->netdev,
+			    "can't offload TC csum action for some header/s - flags %#x\n",
+			    update_flags);
+		return false;
+	}
+
+	return true;
+}
+
+static bool
+tc_act_can_offload_csum(struct mlx5e_tc_act_parse_state *parse_state,
+			const struct flow_action_entry *act,
+			int act_index,
+			struct mlx5_flow_attr *attr)
+{
+	struct mlx5e_tc_flow *flow = parse_state->flow;
+
+	return csum_offload_supported(flow->priv, attr->action,
+				      act->csum_flags, parse_state->extack);
+}
+
+static int
+tc_act_parse_csum(struct mlx5e_tc_act_parse_state *parse_state,
+		  const struct flow_action_entry *act,
+		  struct mlx5e_priv *priv,
+		  struct mlx5_flow_attr *attr)
+{
+	return 0;
+}
+
+struct mlx5e_tc_act mlx5e_tc_act_csum = {
+	.can_offload = tc_act_can_offload_csum,
+	.parse_action = tc_act_parse_csum,
+};
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/ct.c b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/ct.c
new file mode 100644
index 000000000..a829c9428
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/ct.c
@@ -0,0 +1,104 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+// Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+
+#include "act.h"
+#include "en/tc_priv.h"
+#include "en/tc_ct.h"
+
+static bool
+tc_act_can_offload_ct(struct mlx5e_tc_act_parse_state *parse_state,
+		      const struct flow_action_entry *act,
+		      int act_index,
+		      struct mlx5_flow_attr *attr)
+{
+	bool clear_action = act->ct.action & TCA_CT_ACT_CLEAR;
+	struct netlink_ext_ack *extack = parse_state->extack;
+
+	if (parse_state->ct && !clear_action) {
+		NL_SET_ERR_MSG_MOD(extack, "Multiple CT actions are not supported");
+		return false;
+	}
+
+	return true;
+}
+
+static int
+tc_act_parse_ct(struct mlx5e_tc_act_parse_state *parse_state,
+		const struct flow_action_entry *act,
+		struct mlx5e_priv *priv,
+		struct mlx5_flow_attr *attr)
+{
+	bool clear_action = act->ct.action & TCA_CT_ACT_CLEAR;
+	int err;
+
+	/* It's redundant to do ct clear more than once. */
+	if (clear_action && parse_state->ct_clear)
+		return 0;
+
+	err = mlx5_tc_ct_parse_action(parse_state->ct_priv, attr,
+				      &attr->parse_attr->mod_hdr_acts,
+				      act, parse_state->extack);
+	if (err)
+		return err;
+
+
+	if (mlx5e_is_eswitch_flow(parse_state->flow))
+		attr->esw_attr->split_count = attr->esw_attr->out_count;
+
+	if (clear_action) {
+		parse_state->ct_clear = true;
+	} else {
+		attr->flags |= MLX5_ATTR_FLAG_CT;
+		flow_flag_set(parse_state->flow, CT);
+		parse_state->ct = true;
+	}
+
+	return 0;
+}
+
+static int
+tc_act_post_parse_ct(struct mlx5e_tc_act_parse_state *parse_state,
+		     struct mlx5e_priv *priv,
+		     struct mlx5_flow_attr *attr)
+{
+	struct mlx5e_tc_mod_hdr_acts *mod_acts = &attr->parse_attr->mod_hdr_acts;
+	int err;
+
+	/* If ct action exist, we can ignore previous ct_clear actions */
+	if (parse_state->ct)
+		return 0;
+
+	if (parse_state->ct_clear) {
+		err = mlx5_tc_ct_set_ct_clear_regs(parse_state->ct_priv, mod_acts);
+		if (err) {
+			NL_SET_ERR_MSG_MOD(parse_state->extack,
+					   "Failed to set registers for ct clear");
+			return err;
+		}
+		attr->action |= MLX5_FLOW_CONTEXT_ACTION_MOD_HDR;
+
+		/* Prevent handling of additional, redundant clear actions */
+		parse_state->ct_clear = false;
+	}
+
+	return 0;
+}
+
+static bool
+tc_act_is_multi_table_act_ct(struct mlx5e_priv *priv,
+			     const struct flow_action_entry *act,
+			     struct mlx5_flow_attr *attr)
+{
+	if (act->ct.action & TCA_CT_ACT_CLEAR)
+		return false;
+
+	return true;
+}
+
+struct mlx5e_tc_act mlx5e_tc_act_ct = {
+	.can_offload = tc_act_can_offload_ct,
+	.parse_action = tc_act_parse_ct,
+	.is_multi_table_act = tc_act_is_multi_table_act_ct,
+	.post_parse = tc_act_post_parse_ct,
+};
+
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/drop.c b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/drop.c
new file mode 100644
index 000000000..dd025a95c
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/drop.c
@@ -0,0 +1,30 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+// Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+
+#include "act.h"
+#include "en/tc_priv.h"
+
+static bool
+tc_act_can_offload_drop(struct mlx5e_tc_act_parse_state *parse_state,
+			const struct flow_action_entry *act,
+			int act_index,
+			struct mlx5_flow_attr *attr)
+{
+	return true;
+}
+
+static int
+tc_act_parse_drop(struct mlx5e_tc_act_parse_state *parse_state,
+		  const struct flow_action_entry *act,
+		  struct mlx5e_priv *priv,
+		  struct mlx5_flow_attr *attr)
+{
+	attr->action |= MLX5_FLOW_CONTEXT_ACTION_DROP;
+
+	return 0;
+}
+
+struct mlx5e_tc_act mlx5e_tc_act_drop = {
+	.can_offload = tc_act_can_offload_drop,
+	.parse_action = tc_act_parse_drop,
+};
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/goto.c b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/goto.c
new file mode 100644
index 000000000..25174f686
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/goto.c
@@ -0,0 +1,124 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+// Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+
+#include "act.h"
+#include "en/tc_priv.h"
+#include "eswitch.h"
+
+static int
+validate_goto_chain(struct mlx5e_priv *priv,
+		    struct mlx5e_tc_flow *flow,
+		    struct mlx5_flow_attr *attr,
+		    const struct flow_action_entry *act,
+		    struct netlink_ext_ack *extack)
+{
+	struct mlx5e_tc_table *tc = mlx5e_fs_get_tc(priv->fs);
+	bool is_esw = mlx5e_is_eswitch_flow(flow);
+	bool ft_flow = mlx5e_is_ft_flow(flow);
+	u32 dest_chain = act->chain_index;
+	struct mlx5_fs_chains *chains;
+	struct mlx5_eswitch *esw;
+	u32 reformat_and_fwd;
+	u32 max_chain;
+
+	esw = priv->mdev->priv.eswitch;
+	chains = is_esw ? esw_chains(esw) : mlx5e_nic_chains(tc);
+	max_chain = mlx5_chains_get_chain_range(chains);
+	reformat_and_fwd = is_esw ?
+			   MLX5_CAP_ESW_FLOWTABLE_FDB(priv->mdev, reformat_and_fwd_to_table) :
+			   MLX5_CAP_FLOWTABLE_NIC_RX(priv->mdev, reformat_and_fwd_to_table);
+
+	if (ft_flow) {
+		NL_SET_ERR_MSG_MOD(extack, "Goto action is not supported");
+		return -EOPNOTSUPP;
+	}
+
+	if (!mlx5_chains_backwards_supported(chains) &&
+	    dest_chain <= attr->chain) {
+		NL_SET_ERR_MSG_MOD(extack, "Goto lower numbered chain isn't supported");
+		return -EOPNOTSUPP;
+	}
+
+	if (dest_chain > max_chain) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "Requested destination chain is out of supported range");
+		return -EOPNOTSUPP;
+	}
+
+	if (attr->action & (MLX5_FLOW_CONTEXT_ACTION_PACKET_REFORMAT |
+			    MLX5_FLOW_CONTEXT_ACTION_DECAP) &&
+	    !reformat_and_fwd) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "Goto chain is not allowed if action has reformat or decap");
+		return -EOPNOTSUPP;
+	}
+
+	return 0;
+}
+
+static bool
+tc_act_can_offload_goto(struct mlx5e_tc_act_parse_state *parse_state,
+			const struct flow_action_entry *act,
+			int act_index,
+			struct mlx5_flow_attr *attr)
+{
+	struct netlink_ext_ack *extack = parse_state->extack;
+	struct mlx5e_tc_flow *flow = parse_state->flow;
+
+	if (validate_goto_chain(flow->priv, flow, attr, act, extack))
+		return false;
+
+	return true;
+}
+
+static int
+tc_act_parse_goto(struct mlx5e_tc_act_parse_state *parse_state,
+		  const struct flow_action_entry *act,
+		  struct mlx5e_priv *priv,
+		  struct mlx5_flow_attr *attr)
+{
+	attr->action |= MLX5_FLOW_CONTEXT_ACTION_FWD_DEST;
+	attr->dest_chain = act->chain_index;
+
+	return 0;
+}
+
+static int
+tc_act_post_parse_goto(struct mlx5e_tc_act_parse_state *parse_state,
+		       struct mlx5e_priv *priv,
+		       struct mlx5_flow_attr *attr)
+{
+	struct mlx5e_tc_flow_parse_attr *parse_attr = attr->parse_attr;
+	struct netlink_ext_ack *extack = parse_state->extack;
+	struct mlx5e_tc_flow *flow = parse_state->flow;
+
+	if (!attr->dest_chain)
+		return 0;
+
+	if (parse_state->decap) {
+		/* It can be supported if we'll create a mapping for
+		 * the tunnel device only (without tunnel), and set
+		 * this tunnel id with this decap flow.
+		 *
+		 * On restore (miss), we'll just set this saved tunnel
+		 * device.
+		 */
+
+		NL_SET_ERR_MSG_MOD(extack, "Decap with goto isn't supported");
+		netdev_warn(priv->netdev, "Decap with goto isn't supported");
+		return -EOPNOTSUPP;
+	}
+
+	if (!mlx5e_is_eswitch_flow(flow) && parse_attr->mirred_ifindex[0]) {
+		NL_SET_ERR_MSG_MOD(extack, "Mirroring goto chain rules isn't supported");
+		return -EOPNOTSUPP;
+	}
+
+	return 0;
+}
+
+struct mlx5e_tc_act mlx5e_tc_act_goto = {
+	.can_offload = tc_act_can_offload_goto,
+	.parse_action = tc_act_parse_goto,
+	.post_parse = tc_act_post_parse_goto,
+};
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/mark.c b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/mark.c
new file mode 100644
index 000000000..e8d227595
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/mark.c
@@ -0,0 +1,36 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+// Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+
+#include "act.h"
+#include "en_tc.h"
+
+static bool
+tc_act_can_offload_mark(struct mlx5e_tc_act_parse_state *parse_state,
+			const struct flow_action_entry *act,
+			int act_index,
+			struct mlx5_flow_attr *attr)
+{
+	if (act->mark & ~MLX5E_TC_FLOW_ID_MASK) {
+		NL_SET_ERR_MSG_MOD(parse_state->extack, "Bad flow mark, only 16 bit supported");
+		return false;
+	}
+
+	return true;
+}
+
+static int
+tc_act_parse_mark(struct mlx5e_tc_act_parse_state *parse_state,
+		  const struct flow_action_entry *act,
+		  struct mlx5e_priv *priv,
+		  struct mlx5_flow_attr *attr)
+{
+	attr->nic_attr->flow_tag = act->mark;
+	attr->action |= MLX5_FLOW_CONTEXT_ACTION_FWD_DEST;
+
+	return 0;
+}
+
+struct mlx5e_tc_act mlx5e_tc_act_mark = {
+	.can_offload = tc_act_can_offload_mark,
+	.parse_action = tc_act_parse_mark,
+};
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/mirred.c b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/mirred.c
new file mode 100644
index 000000000..4ac7de3f6
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/mirred.c
@@ -0,0 +1,337 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+// Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+
+#include <linux/if_macvlan.h>
+#include <linux/if_vlan.h>
+#include <net/bareudp.h>
+#include <net/bonding.h>
+#include "act.h"
+#include "vlan.h"
+#include "en/tc_tun_encap.h"
+#include "en/tc_priv.h"
+#include "en_rep.h"
+#include "lag/lag.h"
+
+static bool
+same_vf_reps(struct mlx5e_priv *priv, struct net_device *out_dev)
+{
+	return mlx5e_eswitch_vf_rep(priv->netdev) &&
+	       priv->netdev == out_dev;
+}
+
+static int
+verify_uplink_forwarding(struct mlx5e_priv *priv,
+			 struct mlx5_flow_attr *attr,
+			 struct net_device *out_dev,
+			 struct netlink_ext_ack *extack)
+{
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	struct mlx5e_rep_priv *rep_priv;
+
+	/* Forwarding non encapsulated traffic between
+	 * uplink ports is allowed only if
+	 * termination_table_raw_traffic cap is set.
+	 *
+	 * Input vport was stored attr->in_rep.
+	 * In LAG case, *priv* is the private data of
+	 * uplink which may be not the input vport.
+	 */
+	rep_priv = mlx5e_rep_to_rep_priv(attr->esw_attr->in_rep);
+
+	if (!(mlx5e_eswitch_uplink_rep(rep_priv->netdev) &&
+	      mlx5e_eswitch_uplink_rep(out_dev)))
+		return 0;
+
+	if (!MLX5_CAP_ESW_FLOWTABLE_FDB(esw->dev,
+					termination_table_raw_traffic)) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "devices are both uplink, can't offload forwarding");
+			return -EOPNOTSUPP;
+	} else if (out_dev != rep_priv->netdev) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "devices are not the same uplink, can't offload forwarding");
+		return -EOPNOTSUPP;
+	}
+	return 0;
+}
+
+static bool
+is_duplicated_output_device(struct net_device *dev,
+			    struct net_device *out_dev,
+			    int *ifindexes, int if_count,
+			    struct netlink_ext_ack *extack)
+{
+	int i;
+
+	for (i = 0; i < if_count; i++) {
+		if (ifindexes[i] == out_dev->ifindex) {
+			NL_SET_ERR_MSG_MOD(extack, "can't duplicate output to same device");
+			netdev_err(dev, "can't duplicate output to same device: %s\n",
+				   out_dev->name);
+			return true;
+		}
+	}
+
+	return false;
+}
+
+static struct net_device *
+get_fdb_out_dev(struct net_device *uplink_dev, struct net_device *out_dev)
+{
+	struct net_device *fdb_out_dev = out_dev;
+	struct net_device *uplink_upper;
+
+	rcu_read_lock();
+	uplink_upper = netdev_master_upper_dev_get_rcu(uplink_dev);
+	if (uplink_upper && netif_is_lag_master(uplink_upper) &&
+	    uplink_upper == out_dev) {
+		fdb_out_dev = uplink_dev;
+	} else if (netif_is_lag_master(out_dev)) {
+		fdb_out_dev = bond_option_active_slave_get_rcu(netdev_priv(out_dev));
+		if (fdb_out_dev &&
+		    (!mlx5e_eswitch_rep(fdb_out_dev) ||
+		     !netdev_port_same_parent_id(fdb_out_dev, uplink_dev)))
+			fdb_out_dev = NULL;
+	}
+	rcu_read_unlock();
+	return fdb_out_dev;
+}
+
+static bool
+tc_act_can_offload_mirred(struct mlx5e_tc_act_parse_state *parse_state,
+			  const struct flow_action_entry *act,
+			  int act_index,
+			  struct mlx5_flow_attr *attr)
+{
+	struct netlink_ext_ack *extack = parse_state->extack;
+	struct mlx5e_tc_flow *flow = parse_state->flow;
+	struct mlx5e_tc_flow_parse_attr *parse_attr;
+	struct net_device *out_dev = act->dev;
+	struct mlx5e_priv *priv = flow->priv;
+	struct mlx5_esw_flow_attr *esw_attr;
+
+	parse_attr = attr->parse_attr;
+	esw_attr = attr->esw_attr;
+
+	if (!out_dev) {
+		/* out_dev is NULL when filters with
+		 * non-existing mirred device are replayed to
+		 * the driver.
+		 */
+		return false;
+	}
+
+	if (parse_state->mpls_push && !netif_is_bareudp(out_dev)) {
+		NL_SET_ERR_MSG_MOD(extack, "mpls is supported only through a bareudp device");
+		return false;
+	}
+
+	if (parse_state->eth_pop && !parse_state->mpls_push) {
+		NL_SET_ERR_MSG_MOD(extack, "vlan pop eth is supported only with mpls push");
+		return false;
+	}
+
+	if (flow_flag_test(parse_state->flow, L3_TO_L2_DECAP) && !parse_state->eth_push) {
+		NL_SET_ERR_MSG_MOD(extack, "mpls pop is only supported with vlan eth push");
+		return false;
+	}
+
+	if (mlx5e_is_ft_flow(flow) && out_dev == priv->netdev) {
+		/* Ignore forward to self rules generated
+		 * by adding both mlx5 devs to the flow table
+		 * block on a normal nft offload setup.
+		 */
+		return false;
+	}
+
+	if (esw_attr->out_count >= MLX5_MAX_FLOW_FWD_VPORTS) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "can't support more output ports, can't offload forwarding");
+		netdev_warn(priv->netdev,
+			    "can't support more than %d output ports, can't offload forwarding\n",
+			    esw_attr->out_count);
+		return false;
+	}
+
+	if (parse_state->encap ||
+	    netdev_port_same_parent_id(priv->netdev, out_dev) ||
+	    netif_is_ovs_master(out_dev))
+		return true;
+
+	if (parse_attr->filter_dev != priv->netdev) {
+		/* All mlx5 devices are called to configure
+		 * high level device filters. Therefore, the
+		 * *attempt* to  install a filter on invalid
+		 * eswitch should not trigger an explicit error
+		 */
+		return false;
+	}
+
+	NL_SET_ERR_MSG_MOD(extack, "devices are not on same switch HW, can't offload forwarding");
+
+	return false;
+}
+
+static int
+parse_mirred_encap(struct mlx5e_tc_act_parse_state *parse_state,
+		   const struct flow_action_entry *act,
+		   struct mlx5_flow_attr *attr)
+{
+	struct mlx5e_tc_flow_parse_attr *parse_attr = attr->parse_attr;
+	struct mlx5_esw_flow_attr *esw_attr = attr->esw_attr;
+	struct net_device *out_dev = act->dev;
+
+	parse_attr->mirred_ifindex[esw_attr->out_count] = out_dev->ifindex;
+	parse_attr->tun_info[esw_attr->out_count] =
+		mlx5e_dup_tun_info(parse_state->tun_info);
+
+	if (!parse_attr->tun_info[esw_attr->out_count])
+		return -ENOMEM;
+
+	parse_state->encap = false;
+
+	if (parse_state->mpls_push) {
+		memcpy(&parse_attr->mpls_info[esw_attr->out_count],
+		       &parse_state->mpls_info, sizeof(parse_state->mpls_info));
+		parse_state->mpls_push = false;
+	}
+	esw_attr->dests[esw_attr->out_count].flags |= MLX5_ESW_DEST_ENCAP;
+	esw_attr->out_count++;
+	/* attr->dests[].rep is resolved when we handle encap */
+
+	return 0;
+}
+
+static int
+parse_mirred(struct mlx5e_tc_act_parse_state *parse_state,
+	     const struct flow_action_entry *act,
+	     struct mlx5e_priv *priv,
+	     struct mlx5_flow_attr *attr)
+{
+	struct mlx5e_tc_flow_parse_attr *parse_attr = attr->parse_attr;
+	struct mlx5_esw_flow_attr *esw_attr = attr->esw_attr;
+	struct netlink_ext_ack *extack = parse_state->extack;
+	struct mlx5e_rep_priv *rpriv = priv->ppriv;
+	struct net_device *out_dev = act->dev;
+	struct net_device *uplink_dev;
+	struct mlx5e_priv *out_priv;
+	struct mlx5_eswitch *esw;
+	bool is_uplink_rep;
+	int *ifindexes;
+	int if_count;
+	int err;
+
+	esw = priv->mdev->priv.eswitch;
+	uplink_dev = mlx5_eswitch_uplink_get_proto_dev(esw, REP_ETH);
+	ifindexes = parse_state->ifindexes;
+	if_count = parse_state->if_count;
+
+	if (is_duplicated_output_device(priv->netdev, out_dev, ifindexes, if_count, extack))
+		return -EOPNOTSUPP;
+
+	parse_state->ifindexes[if_count] = out_dev->ifindex;
+	parse_state->if_count++;
+	is_uplink_rep = mlx5e_eswitch_uplink_rep(out_dev);
+	err = mlx5_lag_do_mirred(priv->mdev, out_dev);
+	if (err)
+		return err;
+
+	out_dev = get_fdb_out_dev(uplink_dev, out_dev);
+	if (!out_dev)
+		return -ENODEV;
+
+	if (is_vlan_dev(out_dev)) {
+		err = mlx5e_tc_act_vlan_add_push_action(priv, attr, &out_dev, extack);
+		if (err)
+			return err;
+	}
+
+	if (is_vlan_dev(parse_attr->filter_dev)) {
+		err = mlx5e_tc_act_vlan_add_pop_action(priv, attr, extack);
+		if (err)
+			return err;
+	}
+
+	if (netif_is_macvlan(out_dev))
+		out_dev = macvlan_dev_real_dev(out_dev);
+
+	err = verify_uplink_forwarding(priv, attr, out_dev, extack);
+	if (err)
+		return err;
+
+	if (!mlx5e_is_valid_eswitch_fwd_dev(priv, out_dev)) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "devices are not on same switch HW, can't offload forwarding");
+		return -EOPNOTSUPP;
+	}
+
+	if (same_vf_reps(priv, out_dev)) {
+		NL_SET_ERR_MSG_MOD(extack, "can't forward from a VF to itself");
+		return -EOPNOTSUPP;
+	}
+
+	out_priv = netdev_priv(out_dev);
+	rpriv = out_priv->ppriv;
+	esw_attr->dests[esw_attr->out_count].rep = rpriv->rep;
+	esw_attr->dests[esw_attr->out_count].mdev = out_priv->mdev;
+
+	/* If output device is bond master then rules are not explicit
+	 * so we don't attempt to count them.
+	 */
+	if (is_uplink_rep && MLX5_CAP_PORT_SELECTION(priv->mdev, port_select_flow_table) &&
+	    MLX5_CAP_GEN(priv->mdev, create_lag_when_not_master_up))
+		attr->lag.count = true;
+
+	esw_attr->out_count++;
+
+	return 0;
+}
+
+static int
+parse_mirred_ovs_master(struct mlx5e_tc_act_parse_state *parse_state,
+			const struct flow_action_entry *act,
+			struct mlx5e_priv *priv,
+			struct mlx5_flow_attr *attr)
+{
+	struct mlx5_esw_flow_attr *esw_attr = attr->esw_attr;
+	struct net_device *out_dev = act->dev;
+	int err;
+
+	err = mlx5e_set_fwd_to_int_port_actions(priv, attr, out_dev->ifindex,
+						MLX5E_TC_INT_PORT_EGRESS,
+						&attr->action, esw_attr->out_count);
+	if (err)
+		return err;
+
+	esw_attr->out_count++;
+	return 0;
+}
+
+static int
+tc_act_parse_mirred(struct mlx5e_tc_act_parse_state *parse_state,
+		    const struct flow_action_entry *act,
+		    struct mlx5e_priv *priv,
+		    struct mlx5_flow_attr *attr)
+{
+	struct net_device *out_dev = act->dev;
+	int err = -EOPNOTSUPP;
+
+	if (parse_state->encap)
+		err = parse_mirred_encap(parse_state, act, attr);
+	else if (netdev_port_same_parent_id(priv->netdev, out_dev))
+		err = parse_mirred(parse_state, act, priv, attr);
+	else if (netif_is_ovs_master(out_dev))
+		err = parse_mirred_ovs_master(parse_state, act, priv, attr);
+
+	if (err)
+		return err;
+
+	attr->action |= MLX5_FLOW_CONTEXT_ACTION_FWD_DEST;
+
+	return 0;
+}
+
+struct mlx5e_tc_act mlx5e_tc_act_mirred = {
+	.can_offload = tc_act_can_offload_mirred,
+	.parse_action = tc_act_parse_mirred,
+};
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/mirred_nic.c b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/mirred_nic.c
new file mode 100644
index 000000000..90b4c1b34
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/mirred_nic.c
@@ -0,0 +1,51 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+// Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+
+#include "act.h"
+#include "en/tc_priv.h"
+
+static bool
+tc_act_can_offload_mirred_nic(struct mlx5e_tc_act_parse_state *parse_state,
+			      const struct flow_action_entry *act,
+			      int act_index,
+			      struct mlx5_flow_attr *attr)
+{
+	struct netlink_ext_ack *extack = parse_state->extack;
+	struct mlx5e_tc_flow *flow = parse_state->flow;
+	struct net_device *out_dev = act->dev;
+	struct mlx5e_priv *priv = flow->priv;
+
+	if (act->id != FLOW_ACTION_REDIRECT)
+		return false;
+
+	if (priv->netdev->netdev_ops != out_dev->netdev_ops ||
+	    !mlx5e_same_hw_devs(priv, netdev_priv(out_dev))) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "devices are not on same switch HW, can't offload forwarding");
+		netdev_warn(priv->netdev,
+			    "devices %s %s not on same switch HW, can't offload forwarding\n",
+			    netdev_name(priv->netdev),
+			    out_dev->name);
+		return false;
+	}
+
+	return true;
+}
+
+static int
+tc_act_parse_mirred_nic(struct mlx5e_tc_act_parse_state *parse_state,
+			const struct flow_action_entry *act,
+			struct mlx5e_priv *priv,
+			struct mlx5_flow_attr *attr)
+{
+	attr->parse_attr->mirred_ifindex[0] = act->dev->ifindex;
+	flow_flag_set(parse_state->flow, HAIRPIN);
+	attr->action |= MLX5_FLOW_CONTEXT_ACTION_FWD_DEST;
+
+	return 0;
+}
+
+struct mlx5e_tc_act mlx5e_tc_act_mirred_nic = {
+	.can_offload = tc_act_can_offload_mirred_nic,
+	.parse_action = tc_act_parse_mirred_nic,
+};
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/mpls.c b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/mpls.c
new file mode 100644
index 000000000..f106190bf
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/mpls.c
@@ -0,0 +1,99 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+// Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+
+#include <net/bareudp.h>
+#include "act.h"
+#include "en/tc_priv.h"
+
+static bool
+tc_act_can_offload_mpls_push(struct mlx5e_tc_act_parse_state *parse_state,
+			     const struct flow_action_entry *act,
+			     int act_index,
+			     struct mlx5_flow_attr *attr)
+{
+	struct netlink_ext_ack *extack = parse_state->extack;
+	struct mlx5e_priv *priv = parse_state->flow->priv;
+
+	if (!MLX5_CAP_ESW_FLOWTABLE_FDB(priv->mdev, reformat_l2_to_l3_tunnel) ||
+	    act->mpls_push.proto != htons(ETH_P_MPLS_UC)) {
+		NL_SET_ERR_MSG_MOD(extack, "mpls push is supported only for mpls_uc protocol");
+		return false;
+	}
+
+	return true;
+}
+
+static void
+copy_mpls_info(struct mlx5e_mpls_info *mpls_info,
+	       const struct flow_action_entry *act)
+{
+	mpls_info->label = act->mpls_push.label;
+	mpls_info->tc = act->mpls_push.tc;
+	mpls_info->bos = act->mpls_push.bos;
+	mpls_info->ttl = act->mpls_push.ttl;
+}
+
+static int
+tc_act_parse_mpls_push(struct mlx5e_tc_act_parse_state *parse_state,
+		       const struct flow_action_entry *act,
+		       struct mlx5e_priv *priv,
+		       struct mlx5_flow_attr *attr)
+{
+	parse_state->mpls_push = true;
+	copy_mpls_info(&parse_state->mpls_info, act);
+
+	return 0;
+}
+
+static bool
+tc_act_can_offload_mpls_pop(struct mlx5e_tc_act_parse_state *parse_state,
+			    const struct flow_action_entry *act,
+			    int act_index,
+			    struct mlx5_flow_attr *attr)
+{
+	struct netlink_ext_ack *extack = parse_state->extack;
+	struct net_device *filter_dev;
+
+	filter_dev = attr->parse_attr->filter_dev;
+
+	/* we only support mpls pop if it is the first action
+	 * or it is second action after tunnel key unset
+	 * and the filter net device is bareudp. Subsequent
+	 * actions can be pedit and the last can be mirred
+	 * egress redirect.
+	 */
+	if ((act_index == 1 && !parse_state->decap) || act_index > 1) {
+		NL_SET_ERR_MSG_MOD(extack, "mpls pop supported only as first action or with decap");
+		return false;
+	}
+
+	if (!netif_is_bareudp(filter_dev)) {
+		NL_SET_ERR_MSG_MOD(extack, "mpls pop supported only on bareudp devices");
+		return false;
+	}
+
+	return true;
+}
+
+static int
+tc_act_parse_mpls_pop(struct mlx5e_tc_act_parse_state *parse_state,
+		      const struct flow_action_entry *act,
+		      struct mlx5e_priv *priv,
+		      struct mlx5_flow_attr *attr)
+{
+	attr->esw_attr->eth.h_proto = act->mpls_pop.proto;
+	attr->action |= MLX5_FLOW_CONTEXT_ACTION_PACKET_REFORMAT;
+	flow_flag_set(parse_state->flow, L3_TO_L2_DECAP);
+
+	return 0;
+}
+
+struct mlx5e_tc_act mlx5e_tc_act_mpls_push = {
+	.can_offload = tc_act_can_offload_mpls_push,
+	.parse_action = tc_act_parse_mpls_push,
+};
+
+struct mlx5e_tc_act mlx5e_tc_act_mpls_pop = {
+	.can_offload = tc_act_can_offload_mpls_pop,
+	.parse_action = tc_act_parse_mpls_pop,
+};
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/pedit.c b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/pedit.c
new file mode 100644
index 000000000..47597c524
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/pedit.c
@@ -0,0 +1,119 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+// Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+
+#include <linux/if_vlan.h>
+#include "act.h"
+#include "pedit.h"
+#include "en/tc_priv.h"
+#include "en/mod_hdr.h"
+
+static int pedit_header_offsets[] = {
+	[FLOW_ACT_MANGLE_HDR_TYPE_ETH] = offsetof(struct pedit_headers, eth),
+	[FLOW_ACT_MANGLE_HDR_TYPE_IP4] = offsetof(struct pedit_headers, ip4),
+	[FLOW_ACT_MANGLE_HDR_TYPE_IP6] = offsetof(struct pedit_headers, ip6),
+	[FLOW_ACT_MANGLE_HDR_TYPE_TCP] = offsetof(struct pedit_headers, tcp),
+	[FLOW_ACT_MANGLE_HDR_TYPE_UDP] = offsetof(struct pedit_headers, udp),
+};
+
+#define pedit_header(_ph, _htype) ((void *)(_ph) + pedit_header_offsets[_htype])
+
+static int
+set_pedit_val(u8 hdr_type, u32 mask, u32 val, u32 offset,
+	      struct pedit_headers_action *hdrs,
+	      struct netlink_ext_ack *extack)
+{
+	u32 *curr_pmask, *curr_pval;
+
+	curr_pmask = (u32 *)(pedit_header(&hdrs->masks, hdr_type) + offset);
+	curr_pval  = (u32 *)(pedit_header(&hdrs->vals, hdr_type) + offset);
+
+	if (*curr_pmask & mask) { /* disallow acting twice on the same location */
+		NL_SET_ERR_MSG_MOD(extack,
+				   "curr_pmask and new mask same. Acting twice on same location");
+		goto out_err;
+	}
+
+	*curr_pmask |= mask;
+	*curr_pval  |= (val & mask);
+
+	return 0;
+
+out_err:
+	return -EOPNOTSUPP;
+}
+
+int
+mlx5e_tc_act_pedit_parse_action(struct mlx5e_priv *priv,
+				const struct flow_action_entry *act, int namespace,
+				struct pedit_headers_action *hdrs,
+				struct netlink_ext_ack *extack)
+{
+	u8 cmd = (act->id == FLOW_ACTION_MANGLE) ? 0 : 1;
+	u8 htype = act->mangle.htype;
+	int err = -EOPNOTSUPP;
+	u32 mask, val, offset;
+
+	if (htype == FLOW_ACT_MANGLE_UNSPEC) {
+		NL_SET_ERR_MSG_MOD(extack, "legacy pedit isn't offloaded");
+		goto out_err;
+	}
+
+	if (!mlx5e_mod_hdr_max_actions(priv->mdev, namespace)) {
+		NL_SET_ERR_MSG_MOD(extack, "The pedit offload action is not supported");
+		goto out_err;
+	}
+
+	mask = act->mangle.mask;
+	val = act->mangle.val;
+	offset = act->mangle.offset;
+
+	err = set_pedit_val(htype, ~mask, val, offset, &hdrs[cmd], extack);
+	if (err)
+		goto out_err;
+
+	hdrs[cmd].pedits++;
+
+	return 0;
+out_err:
+	return err;
+}
+
+static bool
+tc_act_can_offload_pedit(struct mlx5e_tc_act_parse_state *parse_state,
+			 const struct flow_action_entry *act,
+			 int act_index,
+			 struct mlx5_flow_attr *attr)
+{
+	return true;
+}
+
+static int
+tc_act_parse_pedit(struct mlx5e_tc_act_parse_state *parse_state,
+		   const struct flow_action_entry *act,
+		   struct mlx5e_priv *priv,
+		   struct mlx5_flow_attr *attr)
+{
+	struct mlx5_esw_flow_attr *esw_attr = attr->esw_attr;
+	struct mlx5e_tc_flow *flow = parse_state->flow;
+	enum mlx5_flow_namespace_type ns_type;
+	int err;
+
+	ns_type = mlx5e_get_flow_namespace(flow);
+
+	err = mlx5e_tc_act_pedit_parse_action(flow->priv, act, ns_type, attr->parse_attr->hdrs,
+					      parse_state->extack);
+	if (err)
+		return err;
+
+	attr->action |= MLX5_FLOW_CONTEXT_ACTION_MOD_HDR;
+
+	if (ns_type == MLX5_FLOW_NAMESPACE_FDB)
+		esw_attr->split_count = esw_attr->out_count;
+
+	return 0;
+}
+
+struct mlx5e_tc_act mlx5e_tc_act_pedit = {
+	.can_offload = tc_act_can_offload_pedit,
+	.parse_action = tc_act_parse_pedit,
+};
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/pedit.h b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/pedit.h
new file mode 100644
index 000000000..434c8bd71
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/pedit.h
@@ -0,0 +1,30 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. All rights reserved. */
+
+#ifndef __MLX5_EN_TC_ACT_PEDIT_H__
+#define __MLX5_EN_TC_ACT_PEDIT_H__
+
+#include "en_tc.h"
+
+struct pedit_headers {
+	struct ethhdr   eth;
+	struct vlan_hdr vlan;
+	struct iphdr    ip4;
+	struct ipv6hdr  ip6;
+	struct tcphdr   tcp;
+	struct udphdr   udp;
+};
+
+struct pedit_headers_action {
+	struct pedit_headers vals;
+	struct pedit_headers masks;
+	u32 pedits;
+};
+
+int
+mlx5e_tc_act_pedit_parse_action(struct mlx5e_priv *priv,
+				const struct flow_action_entry *act, int namespace,
+				struct pedit_headers_action *hdrs,
+				struct netlink_ext_ack *extack);
+
+#endif /* __MLX5_EN_TC_ACT_PEDIT_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/police.c b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/police.c
new file mode 100644
index 000000000..c8e5ca65b
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/police.c
@@ -0,0 +1,157 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+// Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+
+#include "act.h"
+#include "en/tc_priv.h"
+
+static bool
+tc_act_can_offload_police(struct mlx5e_tc_act_parse_state *parse_state,
+			  const struct flow_action_entry *act,
+			  int act_index,
+			  struct mlx5_flow_attr *attr)
+{
+	if (act->police.notexceed.act_id != FLOW_ACTION_PIPE &&
+	    act->police.notexceed.act_id != FLOW_ACTION_ACCEPT) {
+		NL_SET_ERR_MSG_MOD(parse_state->extack,
+				   "Offload not supported when conform action is not pipe or ok");
+		return false;
+	}
+	if (mlx5e_policer_validate(parse_state->flow_action, act,
+				   parse_state->extack))
+		return false;
+
+	return !!mlx5e_get_flow_meters(parse_state->flow->priv->mdev);
+}
+
+static int
+fill_meter_params_from_act(const struct flow_action_entry *act,
+			   struct mlx5e_flow_meter_params *params)
+{
+	params->index = act->hw_index;
+	if (act->police.rate_bytes_ps) {
+		params->mode = MLX5_RATE_LIMIT_BPS;
+		/* change rate to bits per second */
+		params->rate = act->police.rate_bytes_ps << 3;
+		params->burst = act->police.burst;
+	} else if (act->police.rate_pkt_ps) {
+		params->mode = MLX5_RATE_LIMIT_PPS;
+		params->rate = act->police.rate_pkt_ps;
+		params->burst = act->police.burst_pkt;
+	} else {
+		return -EOPNOTSUPP;
+	}
+
+	return 0;
+}
+
+static int
+tc_act_parse_police(struct mlx5e_tc_act_parse_state *parse_state,
+		    const struct flow_action_entry *act,
+		    struct mlx5e_priv *priv,
+		    struct mlx5_flow_attr *attr)
+{
+	int err;
+
+	err = fill_meter_params_from_act(act, &attr->meter_attr.params);
+	if (err)
+		return err;
+
+	attr->action |= MLX5_FLOW_CONTEXT_ACTION_EXECUTE_ASO;
+	attr->exe_aso_type = MLX5_EXE_ASO_FLOW_METER;
+
+	return 0;
+}
+
+static bool
+tc_act_is_multi_table_act_police(struct mlx5e_priv *priv,
+				 const struct flow_action_entry *act,
+				 struct mlx5_flow_attr *attr)
+{
+	return true;
+}
+
+static int
+tc_act_police_offload(struct mlx5e_priv *priv,
+		      struct flow_offload_action *fl_act,
+		      struct flow_action_entry *act)
+{
+	struct mlx5e_flow_meter_params params = {};
+	struct mlx5e_flow_meter_handle *meter;
+	int err = 0;
+
+	err = mlx5e_policer_validate(&fl_act->action, act, fl_act->extack);
+	if (err)
+		return err;
+
+	err = fill_meter_params_from_act(act, &params);
+	if (err)
+		return err;
+
+	meter = mlx5e_tc_meter_get(priv->mdev, &params);
+	if (IS_ERR(meter) && PTR_ERR(meter) == -ENOENT) {
+		meter = mlx5e_tc_meter_replace(priv->mdev, &params);
+	} else if (!IS_ERR(meter)) {
+		err = mlx5e_tc_meter_update(meter, &params);
+		mlx5e_tc_meter_put(meter);
+	}
+
+	if (IS_ERR(meter)) {
+		NL_SET_ERR_MSG_MOD(fl_act->extack, "Failed to get flow meter");
+		mlx5_core_err(priv->mdev, "Failed to get flow meter %d\n", params.index);
+		err = PTR_ERR(meter);
+	}
+
+	return err;
+}
+
+static int
+tc_act_police_destroy(struct mlx5e_priv *priv,
+		      struct flow_offload_action *fl_act)
+{
+	struct mlx5e_flow_meter_params params = {};
+	struct mlx5e_flow_meter_handle *meter;
+
+	params.index = fl_act->index;
+	meter = mlx5e_tc_meter_get(priv->mdev, &params);
+	if (IS_ERR(meter)) {
+		NL_SET_ERR_MSG_MOD(fl_act->extack, "Failed to get flow meter");
+		mlx5_core_err(priv->mdev, "Failed to get flow meter %d\n", params.index);
+		return PTR_ERR(meter);
+	}
+	/* first put for the get and second for cleanup */
+	mlx5e_tc_meter_put(meter);
+	mlx5e_tc_meter_put(meter);
+	return 0;
+}
+
+static int
+tc_act_police_stats(struct mlx5e_priv *priv,
+		    struct flow_offload_action *fl_act)
+{
+	struct mlx5e_flow_meter_params params = {};
+	struct mlx5e_flow_meter_handle *meter;
+	u64 bytes, packets, drops, lastuse;
+
+	params.index = fl_act->index;
+	meter = mlx5e_tc_meter_get(priv->mdev, &params);
+	if (IS_ERR(meter)) {
+		NL_SET_ERR_MSG_MOD(fl_act->extack, "Failed to get flow meter");
+		mlx5_core_err(priv->mdev, "Failed to get flow meter %d\n", params.index);
+		return PTR_ERR(meter);
+	}
+
+	mlx5e_tc_meter_get_stats(meter, &bytes, &packets, &drops, &lastuse);
+	flow_stats_update(&fl_act->stats, bytes, packets, drops, lastuse,
+			  FLOW_ACTION_HW_STATS_DELAYED);
+	mlx5e_tc_meter_put(meter);
+	return 0;
+}
+
+struct mlx5e_tc_act mlx5e_tc_act_police = {
+	.can_offload = tc_act_can_offload_police,
+	.parse_action = tc_act_parse_police,
+	.is_multi_table_act = tc_act_is_multi_table_act_police,
+	.offload_action = tc_act_police_offload,
+	.destroy_action = tc_act_police_destroy,
+	.stats_action = tc_act_police_stats,
+};
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/ptype.c b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/ptype.c
new file mode 100644
index 000000000..6454b031f
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/ptype.c
@@ -0,0 +1,36 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+// Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+
+#include "act.h"
+#include "en/tc_priv.h"
+
+static bool
+tc_act_can_offload_ptype(struct mlx5e_tc_act_parse_state *parse_state,
+			 const struct flow_action_entry *act,
+			 int act_index,
+			 struct mlx5_flow_attr *attr)
+{
+	return true;
+}
+
+static int
+tc_act_parse_ptype(struct mlx5e_tc_act_parse_state *parse_state,
+		   const struct flow_action_entry *act,
+		   struct mlx5e_priv *priv,
+		   struct mlx5_flow_attr *attr)
+{
+	struct netlink_ext_ack *extack = parse_state->extack;
+
+	if (act->ptype != PACKET_HOST) {
+		NL_SET_ERR_MSG_MOD(extack, "skbedit ptype is only supported with type host");
+		return -EOPNOTSUPP;
+	}
+
+	parse_state->ptype_host = true;
+	return 0;
+}
+
+struct mlx5e_tc_act mlx5e_tc_act_ptype = {
+	.can_offload = tc_act_can_offload_ptype,
+	.parse_action = tc_act_parse_ptype,
+};
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/redirect_ingress.c b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/redirect_ingress.c
new file mode 100644
index 000000000..ad09a8a5f
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/redirect_ingress.c
@@ -0,0 +1,78 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+// Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+
+#include "act.h"
+#include "en/tc_priv.h"
+
+static bool
+tc_act_can_offload_redirect_ingress(struct mlx5e_tc_act_parse_state *parse_state,
+				    const struct flow_action_entry *act,
+				    int act_index,
+				    struct mlx5_flow_attr *attr)
+{
+	struct netlink_ext_ack *extack = parse_state->extack;
+	struct mlx5e_tc_flow_parse_attr *parse_attr;
+	struct net_device *out_dev = act->dev;
+	struct mlx5_esw_flow_attr *esw_attr;
+
+	parse_attr = attr->parse_attr;
+	esw_attr = attr->esw_attr;
+
+	if (!out_dev)
+		return false;
+
+	if (!netif_is_ovs_master(out_dev)) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "redirect to ingress is supported only for OVS internal ports");
+		return false;
+	}
+
+	if (netif_is_ovs_master(parse_attr->filter_dev)) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "redirect to ingress is not supported from internal port");
+		return false;
+	}
+
+	if (!parse_state->ptype_host) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "redirect to int port ingress requires ptype=host action");
+		return false;
+	}
+
+	if (esw_attr->out_count) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "redirect to int port ingress is supported only as single destination");
+		return false;
+	}
+
+	return true;
+}
+
+static int
+tc_act_parse_redirect_ingress(struct mlx5e_tc_act_parse_state *parse_state,
+			      const struct flow_action_entry *act,
+			      struct mlx5e_priv *priv,
+			      struct mlx5_flow_attr *attr)
+{
+	struct mlx5_esw_flow_attr *esw_attr = attr->esw_attr;
+	struct net_device *out_dev = act->dev;
+	int err;
+
+	attr->action |= MLX5_FLOW_CONTEXT_ACTION_FWD_DEST;
+
+	err = mlx5e_set_fwd_to_int_port_actions(priv, attr, out_dev->ifindex,
+						MLX5E_TC_INT_PORT_INGRESS,
+						&attr->action, esw_attr->out_count);
+	if (err)
+		return err;
+
+	esw_attr->out_count++;
+
+	return 0;
+}
+
+struct mlx5e_tc_act mlx5e_tc_act_redirect_ingress = {
+	.can_offload = tc_act_can_offload_redirect_ingress,
+	.parse_action = tc_act_parse_redirect_ingress,
+};
+
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/sample.c b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/sample.c
new file mode 100644
index 000000000..2c0196431
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/sample.c
@@ -0,0 +1,71 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+// Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+
+#include <net/psample.h>
+#include "act.h"
+#include "en/tc_priv.h"
+#include "en/tc/act/sample.h"
+
+static bool
+tc_act_can_offload_sample(struct mlx5e_tc_act_parse_state *parse_state,
+			  const struct flow_action_entry *act,
+			  int act_index,
+			  struct mlx5_flow_attr *attr)
+{
+	struct netlink_ext_ack *extack = parse_state->extack;
+	bool ct_nat;
+
+	ct_nat = attr->ct_attr.ct_action & TCA_CT_ACT_NAT;
+
+	if (flow_flag_test(parse_state->flow, CT) && ct_nat) {
+		NL_SET_ERR_MSG_MOD(extack, "Sample action with CT NAT is not supported");
+		return false;
+	}
+
+	return true;
+}
+
+static int
+tc_act_parse_sample(struct mlx5e_tc_act_parse_state *parse_state,
+		    const struct flow_action_entry *act,
+		    struct mlx5e_priv *priv,
+		    struct mlx5_flow_attr *attr)
+{
+	struct mlx5e_sample_attr *sample_attr = &attr->sample_attr;
+
+	sample_attr->rate = act->sample.rate;
+	sample_attr->group_num = act->sample.psample_group->group_num;
+
+	if (act->sample.truncate)
+		sample_attr->trunc_size = act->sample.trunc_size;
+
+	attr->flags |= MLX5_ATTR_FLAG_SAMPLE;
+	flow_flag_set(parse_state->flow, SAMPLE);
+
+	return 0;
+}
+
+bool
+mlx5e_tc_act_sample_is_multi_table(struct mlx5_core_dev *mdev,
+				   struct mlx5_flow_attr *attr)
+{
+	if (MLX5_CAP_GEN(mdev, reg_c_preserve) ||
+	    attr->action & MLX5_FLOW_CONTEXT_ACTION_DECAP)
+		return true;
+
+	return false;
+}
+
+static bool
+tc_act_is_multi_table_act_sample(struct mlx5e_priv *priv,
+				 const struct flow_action_entry *act,
+				 struct mlx5_flow_attr *attr)
+{
+	return mlx5e_tc_act_sample_is_multi_table(priv->mdev, attr);
+}
+
+struct mlx5e_tc_act mlx5e_tc_act_sample = {
+	.can_offload = tc_act_can_offload_sample,
+	.parse_action = tc_act_parse_sample,
+	.is_multi_table_act = tc_act_is_multi_table_act_sample,
+};
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/sample.h b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/sample.h
new file mode 100644
index 000000000..3efb3a15c
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/sample.h
@@ -0,0 +1,14 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. All rights reserved. */
+
+#ifndef __MLX5_EN_TC_ACT_SAMPLE_H__
+#define __MLX5_EN_TC_ACT_SAMPLE_H__
+
+#include <net/flow_offload.h>
+#include "en/tc_priv.h"
+
+bool
+mlx5e_tc_act_sample_is_multi_table(struct mlx5_core_dev *mdev,
+				   struct mlx5_flow_attr *attr);
+
+#endif /* __MLX5_EN_TC_ACT_SAMPLE_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/trap.c b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/trap.c
new file mode 100644
index 000000000..53b270f65
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/trap.c
@@ -0,0 +1,38 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+// Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+
+#include "act.h"
+#include "en/tc_priv.h"
+
+static bool
+tc_act_can_offload_trap(struct mlx5e_tc_act_parse_state *parse_state,
+			const struct flow_action_entry *act,
+			int act_index,
+			struct mlx5_flow_attr *attr)
+{
+	struct netlink_ext_ack *extack = parse_state->extack;
+
+	if (parse_state->flow_action->num_entries != 1) {
+		NL_SET_ERR_MSG_MOD(extack, "action trap is supported as a sole action only");
+		return false;
+	}
+
+	return true;
+}
+
+static int
+tc_act_parse_trap(struct mlx5e_tc_act_parse_state *parse_state,
+		  const struct flow_action_entry *act,
+		  struct mlx5e_priv *priv,
+		  struct mlx5_flow_attr *attr)
+{
+	attr->action |= MLX5_FLOW_CONTEXT_ACTION_FWD_DEST;
+	attr->flags |= MLX5_ATTR_FLAG_SLOW_PATH;
+
+	return 0;
+}
+
+struct mlx5e_tc_act mlx5e_tc_act_trap = {
+	.can_offload = tc_act_can_offload_trap,
+	.parse_action = tc_act_parse_trap,
+};
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/tun.c b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/tun.c
new file mode 100644
index 000000000..b4fa2de97
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/tun.c
@@ -0,0 +1,63 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+// Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+
+#include "act.h"
+#include "en/tc_tun_encap.h"
+#include "en/tc_priv.h"
+
+static bool
+tc_act_can_offload_tun_encap(struct mlx5e_tc_act_parse_state *parse_state,
+			     const struct flow_action_entry *act,
+			     int act_index,
+			     struct mlx5_flow_attr *attr)
+{
+	if (!act->tunnel) {
+		NL_SET_ERR_MSG_MOD(parse_state->extack,
+				   "Zero tunnel attributes is not supported");
+		return false;
+	}
+
+	return true;
+}
+
+static int
+tc_act_parse_tun_encap(struct mlx5e_tc_act_parse_state *parse_state,
+		       const struct flow_action_entry *act,
+		       struct mlx5e_priv *priv,
+		       struct mlx5_flow_attr *attr)
+{
+	parse_state->tun_info = act->tunnel;
+	parse_state->encap = true;
+
+	return 0;
+}
+
+static bool
+tc_act_can_offload_tun_decap(struct mlx5e_tc_act_parse_state *parse_state,
+			     const struct flow_action_entry *act,
+			     int act_index,
+			     struct mlx5_flow_attr *attr)
+{
+	return true;
+}
+
+static int
+tc_act_parse_tun_decap(struct mlx5e_tc_act_parse_state *parse_state,
+		       const struct flow_action_entry *act,
+		       struct mlx5e_priv *priv,
+		       struct mlx5_flow_attr *attr)
+{
+	parse_state->decap = true;
+
+	return 0;
+}
+
+struct mlx5e_tc_act mlx5e_tc_act_tun_encap = {
+	.can_offload = tc_act_can_offload_tun_encap,
+	.parse_action = tc_act_parse_tun_encap,
+};
+
+struct mlx5e_tc_act mlx5e_tc_act_tun_decap = {
+	.can_offload = tc_act_can_offload_tun_decap,
+	.parse_action = tc_act_parse_tun_decap,
+};
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/vlan.c b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/vlan.c
new file mode 100644
index 000000000..b86ac604d
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/vlan.c
@@ -0,0 +1,228 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+// Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+
+#include <linux/if_vlan.h>
+#include "act.h"
+#include "vlan.h"
+#include "en/tc_priv.h"
+
+static int
+add_vlan_prio_tag_rewrite_action(struct mlx5e_priv *priv,
+				 struct mlx5e_tc_flow_parse_attr *parse_attr,
+				 u32 *action, struct netlink_ext_ack *extack)
+{
+	const struct flow_action_entry prio_tag_act = {
+		.vlan.vid = 0,
+		.vlan.prio =
+			MLX5_GET(fte_match_set_lyr_2_4,
+				 mlx5e_get_match_headers_value(*action,
+							       &parse_attr->spec),
+				 first_prio) &
+			MLX5_GET(fte_match_set_lyr_2_4,
+				 mlx5e_get_match_headers_criteria(*action,
+								  &parse_attr->spec),
+				 first_prio),
+	};
+
+	return mlx5e_tc_act_vlan_add_rewrite_action(priv, MLX5_FLOW_NAMESPACE_FDB,
+						    &prio_tag_act, parse_attr, action,
+						    extack);
+}
+
+static int
+parse_tc_vlan_action(struct mlx5e_priv *priv,
+		     const struct flow_action_entry *act,
+		     struct mlx5_esw_flow_attr *attr,
+		     u32 *action,
+		     struct netlink_ext_ack *extack,
+		     struct mlx5e_tc_act_parse_state *parse_state)
+{
+	u8 vlan_idx = attr->total_vlan;
+
+	if (vlan_idx >= MLX5_FS_VLAN_DEPTH) {
+		NL_SET_ERR_MSG_MOD(extack, "Total vlans used is greater than supported");
+		return -EOPNOTSUPP;
+	}
+
+	switch (act->id) {
+	case FLOW_ACTION_VLAN_POP:
+		if (vlan_idx) {
+			if (!mlx5_eswitch_vlan_actions_supported(priv->mdev,
+								 MLX5_FS_VLAN_DEPTH)) {
+				NL_SET_ERR_MSG_MOD(extack, "vlan pop action is not supported");
+				return -EOPNOTSUPP;
+			}
+
+			*action |= MLX5_FLOW_CONTEXT_ACTION_VLAN_POP_2;
+		} else {
+			*action |= MLX5_FLOW_CONTEXT_ACTION_VLAN_POP;
+		}
+		break;
+	case FLOW_ACTION_VLAN_PUSH:
+		attr->vlan_vid[vlan_idx] = act->vlan.vid;
+		attr->vlan_prio[vlan_idx] = act->vlan.prio;
+		attr->vlan_proto[vlan_idx] = act->vlan.proto;
+		if (!attr->vlan_proto[vlan_idx])
+			attr->vlan_proto[vlan_idx] = htons(ETH_P_8021Q);
+
+		if (vlan_idx) {
+			if (!mlx5_eswitch_vlan_actions_supported(priv->mdev,
+								 MLX5_FS_VLAN_DEPTH)) {
+				NL_SET_ERR_MSG_MOD(extack,
+						   "vlan push action is not supported for vlan depth > 1");
+				return -EOPNOTSUPP;
+			}
+
+			*action |= MLX5_FLOW_CONTEXT_ACTION_VLAN_PUSH_2;
+		} else {
+			if (!mlx5_eswitch_vlan_actions_supported(priv->mdev, 1) &&
+			    (act->vlan.proto != htons(ETH_P_8021Q) ||
+			     act->vlan.prio)) {
+				NL_SET_ERR_MSG_MOD(extack, "vlan push action is not supported");
+				return -EOPNOTSUPP;
+			}
+
+			*action |= MLX5_FLOW_CONTEXT_ACTION_VLAN_PUSH;
+		}
+		break;
+	case FLOW_ACTION_VLAN_POP_ETH:
+		parse_state->eth_pop = true;
+		break;
+	case FLOW_ACTION_VLAN_PUSH_ETH:
+		if (!flow_flag_test(parse_state->flow, L3_TO_L2_DECAP))
+			return -EOPNOTSUPP;
+		parse_state->eth_push = true;
+		memcpy(attr->eth.h_dest, act->vlan_push_eth.dst, ETH_ALEN);
+		memcpy(attr->eth.h_source, act->vlan_push_eth.src, ETH_ALEN);
+		break;
+	default:
+		NL_SET_ERR_MSG_MOD(extack, "Unexpected action id for VLAN");
+		return -EINVAL;
+	}
+
+	attr->total_vlan = vlan_idx + 1;
+
+	return 0;
+}
+
+int
+mlx5e_tc_act_vlan_add_push_action(struct mlx5e_priv *priv,
+				  struct mlx5_flow_attr *attr,
+				  struct net_device **out_dev,
+				  struct netlink_ext_ack *extack)
+{
+	struct net_device *vlan_dev = *out_dev;
+	struct flow_action_entry vlan_act = {
+		.id = FLOW_ACTION_VLAN_PUSH,
+		.vlan.vid = vlan_dev_vlan_id(vlan_dev),
+		.vlan.proto = vlan_dev_vlan_proto(vlan_dev),
+		.vlan.prio = 0,
+	};
+	int err;
+
+	err = parse_tc_vlan_action(priv, &vlan_act, attr->esw_attr, &attr->action, extack, NULL);
+	if (err)
+		return err;
+
+	rcu_read_lock();
+	*out_dev = dev_get_by_index_rcu(dev_net(vlan_dev), dev_get_iflink(vlan_dev));
+	rcu_read_unlock();
+	if (!*out_dev)
+		return -ENODEV;
+
+	if (is_vlan_dev(*out_dev))
+		err = mlx5e_tc_act_vlan_add_push_action(priv, attr, out_dev, extack);
+
+	return err;
+}
+
+int
+mlx5e_tc_act_vlan_add_pop_action(struct mlx5e_priv *priv,
+				 struct mlx5_flow_attr *attr,
+				 struct netlink_ext_ack *extack)
+{
+	struct flow_action_entry vlan_act = {
+		.id = FLOW_ACTION_VLAN_POP,
+	};
+	int nest_level, err = 0;
+
+	nest_level = attr->parse_attr->filter_dev->lower_level -
+						priv->netdev->lower_level;
+	while (nest_level--) {
+		err = parse_tc_vlan_action(priv, &vlan_act, attr->esw_attr, &attr->action,
+					   extack, NULL);
+		if (err)
+			return err;
+	}
+
+	return err;
+}
+
+static bool
+tc_act_can_offload_vlan(struct mlx5e_tc_act_parse_state *parse_state,
+			const struct flow_action_entry *act,
+			int act_index,
+			struct mlx5_flow_attr *attr)
+{
+	return true;
+}
+
+static int
+tc_act_parse_vlan(struct mlx5e_tc_act_parse_state *parse_state,
+		  const struct flow_action_entry *act,
+		  struct mlx5e_priv *priv,
+		  struct mlx5_flow_attr *attr)
+{
+	struct mlx5_esw_flow_attr *esw_attr = attr->esw_attr;
+	int err;
+
+	if (act->id == FLOW_ACTION_VLAN_PUSH &&
+	    (attr->action & MLX5_FLOW_CONTEXT_ACTION_VLAN_POP)) {
+		/* Replace vlan pop+push with vlan modify */
+		attr->action &= ~MLX5_FLOW_CONTEXT_ACTION_VLAN_POP;
+		err = mlx5e_tc_act_vlan_add_rewrite_action(priv, MLX5_FLOW_NAMESPACE_FDB, act,
+							   attr->parse_attr, &attr->action,
+							   parse_state->extack);
+	} else {
+		err = parse_tc_vlan_action(priv, act, esw_attr, &attr->action,
+					   parse_state->extack, parse_state);
+	}
+
+	if (err)
+		return err;
+
+	esw_attr->split_count = esw_attr->out_count;
+
+	return 0;
+}
+
+static int
+tc_act_post_parse_vlan(struct mlx5e_tc_act_parse_state *parse_state,
+		       struct mlx5e_priv *priv,
+		       struct mlx5_flow_attr *attr)
+{
+	struct mlx5e_tc_flow_parse_attr *parse_attr = attr->parse_attr;
+	struct netlink_ext_ack *extack = parse_state->extack;
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	int err;
+
+	if (MLX5_CAP_GEN(esw->dev, prio_tag_required) &&
+	    attr->action & MLX5_FLOW_CONTEXT_ACTION_VLAN_POP) {
+		/* For prio tag mode, replace vlan pop with rewrite vlan prio
+		 * tag rewrite.
+		 */
+		attr->action &= ~MLX5_FLOW_CONTEXT_ACTION_VLAN_POP;
+		err = add_vlan_prio_tag_rewrite_action(priv, parse_attr,
+						       &attr->action, extack);
+		if (err)
+			return err;
+	}
+
+	return 0;
+}
+
+struct mlx5e_tc_act mlx5e_tc_act_vlan = {
+	.can_offload = tc_act_can_offload_vlan,
+	.parse_action = tc_act_parse_vlan,
+	.post_parse = tc_act_post_parse_vlan,
+};
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/vlan.h b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/vlan.h
new file mode 100644
index 000000000..2fa58c6f4
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/vlan.h
@@ -0,0 +1,29 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. All rights reserved. */
+
+#ifndef __MLX5_EN_TC_ACT_VLAN_H__
+#define __MLX5_EN_TC_ACT_VLAN_H__
+
+#include <net/flow_offload.h>
+#include "en/tc_priv.h"
+
+struct pedit_headers_action;
+
+int
+mlx5e_tc_act_vlan_add_push_action(struct mlx5e_priv *priv,
+				  struct mlx5_flow_attr *attr,
+				  struct net_device **out_dev,
+				  struct netlink_ext_ack *extack);
+
+int
+mlx5e_tc_act_vlan_add_pop_action(struct mlx5e_priv *priv,
+				 struct mlx5_flow_attr *attr,
+				 struct netlink_ext_ack *extack);
+
+int
+mlx5e_tc_act_vlan_add_rewrite_action(struct mlx5e_priv *priv, int namespace,
+				     const struct flow_action_entry *act,
+				     struct mlx5e_tc_flow_parse_attr *parse_attr,
+				     u32 *action, struct netlink_ext_ack *extack);
+
+#endif /* __MLX5_EN_TC_ACT_VLAN_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/vlan_mangle.c b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/vlan_mangle.c
new file mode 100644
index 000000000..9a8a1a6bd
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/vlan_mangle.c
@@ -0,0 +1,86 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+// Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+
+#include <linux/if_vlan.h>
+#include "act.h"
+#include "vlan.h"
+#include "en/tc_priv.h"
+
+struct pedit_headers_action;
+
+int
+mlx5e_tc_act_vlan_add_rewrite_action(struct mlx5e_priv *priv, int namespace,
+				     const struct flow_action_entry *act,
+				     struct mlx5e_tc_flow_parse_attr *parse_attr,
+				     u32 *action, struct netlink_ext_ack *extack)
+{
+	u16 mask16 = VLAN_VID_MASK;
+	u16 val16 = act->vlan.vid & VLAN_VID_MASK;
+	const struct flow_action_entry pedit_act = {
+		.id = FLOW_ACTION_MANGLE,
+		.mangle.htype = FLOW_ACT_MANGLE_HDR_TYPE_ETH,
+		.mangle.offset = offsetof(struct vlan_ethhdr, h_vlan_TCI),
+		.mangle.mask = ~(u32)be16_to_cpu(*(__be16 *)&mask16),
+		.mangle.val = (u32)be16_to_cpu(*(__be16 *)&val16),
+	};
+	u8 match_prio_mask, match_prio_val;
+	void *headers_c, *headers_v;
+	int err;
+
+	headers_c = mlx5e_get_match_headers_criteria(*action, &parse_attr->spec);
+	headers_v = mlx5e_get_match_headers_value(*action, &parse_attr->spec);
+
+	if (!(MLX5_GET(fte_match_set_lyr_2_4, headers_c, cvlan_tag) &&
+	      MLX5_GET(fte_match_set_lyr_2_4, headers_v, cvlan_tag))) {
+		NL_SET_ERR_MSG_MOD(extack, "VLAN rewrite action must have VLAN protocol match");
+		return -EOPNOTSUPP;
+	}
+
+	match_prio_mask = MLX5_GET(fte_match_set_lyr_2_4, headers_c, first_prio);
+	match_prio_val = MLX5_GET(fte_match_set_lyr_2_4, headers_v, first_prio);
+	if (act->vlan.prio != (match_prio_val & match_prio_mask)) {
+		NL_SET_ERR_MSG_MOD(extack, "Changing VLAN prio is not supported");
+		return -EOPNOTSUPP;
+	}
+
+	err = mlx5e_tc_act_pedit_parse_action(priv, &pedit_act, namespace, parse_attr->hdrs,
+					      extack);
+	*action |= MLX5_FLOW_CONTEXT_ACTION_MOD_HDR;
+
+	return err;
+}
+
+static bool
+tc_act_can_offload_vlan_mangle(struct mlx5e_tc_act_parse_state *parse_state,
+			       const struct flow_action_entry *act,
+			       int act_index,
+			       struct mlx5_flow_attr *attr)
+{
+	return true;
+}
+
+static int
+tc_act_parse_vlan_mangle(struct mlx5e_tc_act_parse_state *parse_state,
+			 const struct flow_action_entry *act,
+			 struct mlx5e_priv *priv,
+			 struct mlx5_flow_attr *attr)
+{
+	enum mlx5_flow_namespace_type ns_type;
+	int err;
+
+	ns_type = mlx5e_get_flow_namespace(parse_state->flow);
+	err = mlx5e_tc_act_vlan_add_rewrite_action(priv, ns_type, act, attr->parse_attr,
+						   &attr->action, parse_state->extack);
+	if (err)
+		return err;
+
+	if (ns_type == MLX5_FLOW_NAMESPACE_FDB)
+		attr->esw_attr->split_count = attr->esw_attr->out_count;
+
+	return 0;
+}
+
+struct mlx5e_tc_act mlx5e_tc_act_vlan_mangle = {
+	.can_offload = tc_act_can_offload_vlan_mangle,
+	.parse_action = tc_act_parse_vlan_mangle,
+};
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc/ct_fs.h b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/ct_fs.h
new file mode 100644
index 000000000..bb6b1a979
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/ct_fs.h
@@ -0,0 +1,49 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. */
+
+#ifndef __MLX5_EN_TC_CT_FS_H__
+#define __MLX5_EN_TC_CT_FS_H__
+
+struct mlx5_ct_fs {
+	const struct net_device *netdev;
+	struct mlx5_core_dev *dev;
+
+	/* private data */
+	void *priv_data[];
+};
+
+struct mlx5_ct_fs_rule {
+};
+
+struct mlx5_ct_fs_ops {
+	int (*init)(struct mlx5_ct_fs *fs, struct mlx5_flow_table *ct,
+		    struct mlx5_flow_table *ct_nat, struct mlx5_flow_table *post_ct);
+	void (*destroy)(struct mlx5_ct_fs *fs);
+
+	struct mlx5_ct_fs_rule * (*ct_rule_add)(struct mlx5_ct_fs *fs,
+						struct mlx5_flow_spec *spec,
+						struct mlx5_flow_attr *attr,
+						struct flow_rule *flow_rule);
+	void (*ct_rule_del)(struct mlx5_ct_fs *fs, struct mlx5_ct_fs_rule *fs_rule);
+
+	size_t priv_size;
+};
+
+static inline void *mlx5_ct_fs_priv(struct mlx5_ct_fs *fs)
+{
+	return &fs->priv_data;
+}
+
+struct mlx5_ct_fs_ops *mlx5_ct_fs_dmfs_ops_get(void);
+
+#if IS_ENABLED(CONFIG_MLX5_SW_STEERING)
+struct mlx5_ct_fs_ops *mlx5_ct_fs_smfs_ops_get(void);
+#else
+static inline struct mlx5_ct_fs_ops *
+mlx5_ct_fs_smfs_ops_get(void)
+{
+	return NULL;
+}
+#endif /* IS_ENABLED(CONFIG_MLX5_SW_STEERING) */
+
+#endif /* __MLX5_EN_TC_CT_FS_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc/ct_fs_dmfs.c b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/ct_fs_dmfs.c
new file mode 100644
index 000000000..ae4f55be4
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/ct_fs_dmfs.c
@@ -0,0 +1,79 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. */
+
+#include "en_tc.h"
+#include "en/tc_ct.h"
+#include "en/tc/ct_fs.h"
+
+#define ct_dbg(fmt, args...)\
+	netdev_dbg(fs->netdev, "ct_fs_dmfs debug: " fmt "\n", ##args)
+
+struct mlx5_ct_fs_dmfs_rule {
+	struct mlx5_ct_fs_rule fs_rule;
+	struct mlx5_flow_handle *rule;
+	struct mlx5_flow_attr *attr;
+};
+
+static int
+mlx5_ct_fs_dmfs_init(struct mlx5_ct_fs *fs, struct mlx5_flow_table *ct,
+		     struct mlx5_flow_table *ct_nat, struct mlx5_flow_table *post_ct)
+{
+	return 0;
+}
+
+static void
+mlx5_ct_fs_dmfs_destroy(struct mlx5_ct_fs *fs)
+{
+}
+
+static struct mlx5_ct_fs_rule *
+mlx5_ct_fs_dmfs_ct_rule_add(struct mlx5_ct_fs *fs, struct mlx5_flow_spec *spec,
+			    struct mlx5_flow_attr *attr, struct flow_rule *flow_rule)
+{
+	struct mlx5e_priv *priv = netdev_priv(fs->netdev);
+	struct mlx5_ct_fs_dmfs_rule *dmfs_rule;
+	int err;
+
+	dmfs_rule = kzalloc(sizeof(*dmfs_rule), GFP_KERNEL);
+	if (!dmfs_rule)
+		return ERR_PTR(-ENOMEM);
+
+	dmfs_rule->rule = mlx5_tc_rule_insert(priv, spec, attr);
+	if (IS_ERR(dmfs_rule->rule)) {
+		err = PTR_ERR(dmfs_rule->rule);
+		ct_dbg("Failed to add ct entry fs rule");
+		goto err_insert;
+	}
+
+	dmfs_rule->attr = attr;
+
+	return &dmfs_rule->fs_rule;
+
+err_insert:
+	kfree(dmfs_rule);
+	return ERR_PTR(err);
+}
+
+static void
+mlx5_ct_fs_dmfs_ct_rule_del(struct mlx5_ct_fs *fs, struct mlx5_ct_fs_rule *fs_rule)
+{
+	struct mlx5_ct_fs_dmfs_rule *dmfs_rule = container_of(fs_rule,
+							      struct mlx5_ct_fs_dmfs_rule,
+							      fs_rule);
+
+	mlx5_tc_rule_delete(netdev_priv(fs->netdev), dmfs_rule->rule, dmfs_rule->attr);
+	kfree(dmfs_rule);
+}
+
+static struct mlx5_ct_fs_ops dmfs_ops = {
+	.ct_rule_add = mlx5_ct_fs_dmfs_ct_rule_add,
+	.ct_rule_del = mlx5_ct_fs_dmfs_ct_rule_del,
+
+	.init = mlx5_ct_fs_dmfs_init,
+	.destroy = mlx5_ct_fs_dmfs_destroy,
+};
+
+struct mlx5_ct_fs_ops *mlx5_ct_fs_dmfs_ops_get(void)
+{
+	return &dmfs_ops;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc/ct_fs_smfs.c b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/ct_fs_smfs.c
new file mode 100644
index 000000000..2b80fe735
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/ct_fs_smfs.c
@@ -0,0 +1,380 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. */
+
+#include <linux/refcount.h>
+
+#include "en_tc.h"
+#include "en/tc_priv.h"
+#include "en/tc_ct.h"
+#include "en/tc/ct_fs.h"
+
+#include "lib/smfs.h"
+
+#define INIT_ERR_PREFIX "ct_fs_smfs init failed"
+#define ct_dbg(fmt, args...)\
+	netdev_dbg(fs->netdev, "ct_fs_smfs debug: " fmt "\n", ##args)
+#define MLX5_CT_TCP_FLAGS_MASK cpu_to_be16(be32_to_cpu(TCP_FLAG_RST | TCP_FLAG_FIN) >> 16)
+
+struct mlx5_ct_fs_smfs_matcher {
+	struct mlx5dr_matcher *dr_matcher;
+	struct list_head list;
+	int prio;
+	refcount_t ref;
+};
+
+struct mlx5_ct_fs_smfs_matchers {
+	struct mlx5_ct_fs_smfs_matcher smfs_matchers[6];
+	struct list_head used;
+};
+
+struct mlx5_ct_fs_smfs {
+	struct mlx5dr_table *ct_tbl, *ct_nat_tbl;
+	struct mlx5_ct_fs_smfs_matchers matchers;
+	struct mlx5_ct_fs_smfs_matchers matchers_nat;
+	struct mlx5dr_action *fwd_action;
+	struct mlx5_flow_table *ct_nat;
+	struct mutex lock; /* Guards matchers */
+};
+
+struct mlx5_ct_fs_smfs_rule {
+	struct mlx5_ct_fs_rule fs_rule;
+	struct mlx5dr_rule *rule;
+	struct mlx5dr_action *count_action;
+	struct mlx5_ct_fs_smfs_matcher *smfs_matcher;
+};
+
+static inline void
+mlx5_ct_fs_smfs_fill_mask(struct mlx5_ct_fs *fs, struct mlx5_flow_spec *spec, bool ipv4, bool tcp,
+			  bool gre)
+{
+	void *headers_c = MLX5_ADDR_OF(fte_match_param, spec->match_criteria, outer_headers);
+
+	if (likely(MLX5_CAP_FLOWTABLE_NIC_RX(fs->dev, ft_field_support.outer_ip_version)))
+		MLX5_SET_TO_ONES(fte_match_set_lyr_2_4, headers_c, ip_version);
+	else
+		MLX5_SET_TO_ONES(fte_match_set_lyr_2_4, headers_c, ethertype);
+
+	MLX5_SET_TO_ONES(fte_match_set_lyr_2_4, headers_c, ip_protocol);
+	if (likely(ipv4)) {
+		MLX5_SET_TO_ONES(fte_match_set_lyr_2_4, headers_c,
+				 src_ipv4_src_ipv6.ipv4_layout.ipv4);
+		MLX5_SET_TO_ONES(fte_match_set_lyr_2_4, headers_c,
+				 dst_ipv4_dst_ipv6.ipv4_layout.ipv4);
+	} else {
+		memset(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_c,
+				    dst_ipv4_dst_ipv6.ipv6_layout.ipv6),
+		       0xFF,
+		       MLX5_FLD_SZ_BYTES(fte_match_set_lyr_2_4,
+					 dst_ipv4_dst_ipv6.ipv6_layout.ipv6));
+		memset(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_c,
+				    src_ipv4_src_ipv6.ipv6_layout.ipv6),
+		       0xFF,
+		       MLX5_FLD_SZ_BYTES(fte_match_set_lyr_2_4,
+					 src_ipv4_src_ipv6.ipv6_layout.ipv6));
+	}
+
+	if (likely(tcp)) {
+		MLX5_SET_TO_ONES(fte_match_set_lyr_2_4, headers_c, tcp_sport);
+		MLX5_SET_TO_ONES(fte_match_set_lyr_2_4, headers_c, tcp_dport);
+		MLX5_SET(fte_match_set_lyr_2_4, headers_c, tcp_flags,
+			 ntohs(MLX5_CT_TCP_FLAGS_MASK));
+	} else if (!gre) {
+		MLX5_SET_TO_ONES(fte_match_set_lyr_2_4, headers_c, udp_sport);
+		MLX5_SET_TO_ONES(fte_match_set_lyr_2_4, headers_c, udp_dport);
+	}
+
+	mlx5e_tc_match_to_reg_match(spec, ZONE_TO_REG, 0, MLX5_CT_ZONE_MASK);
+}
+
+static struct mlx5dr_matcher *
+mlx5_ct_fs_smfs_matcher_create(struct mlx5_ct_fs *fs, struct mlx5dr_table *tbl, bool ipv4,
+			       bool tcp, bool gre, u32 priority)
+{
+	struct mlx5dr_matcher *dr_matcher;
+	struct mlx5_flow_spec *spec;
+
+	spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
+	if (!spec)
+		return ERR_PTR(-ENOMEM);
+
+	mlx5_ct_fs_smfs_fill_mask(fs, spec, ipv4, tcp, gre);
+	spec->match_criteria_enable = MLX5_MATCH_MISC_PARAMETERS_2 | MLX5_MATCH_OUTER_HEADERS;
+
+	dr_matcher = mlx5_smfs_matcher_create(tbl, priority, spec);
+	kvfree(spec);
+	if (!dr_matcher)
+		return ERR_PTR(-EINVAL);
+
+	return dr_matcher;
+}
+
+static struct mlx5_ct_fs_smfs_matcher *
+mlx5_ct_fs_smfs_matcher_get(struct mlx5_ct_fs *fs, bool nat, bool ipv4, bool tcp, bool gre)
+{
+	struct mlx5_ct_fs_smfs *fs_smfs = mlx5_ct_fs_priv(fs);
+	struct mlx5_ct_fs_smfs_matcher *m, *smfs_matcher;
+	struct mlx5_ct_fs_smfs_matchers *matchers;
+	struct mlx5dr_matcher *dr_matcher;
+	struct mlx5dr_table *tbl;
+	struct list_head *prev;
+	int prio;
+
+	matchers = nat ? &fs_smfs->matchers_nat : &fs_smfs->matchers;
+	smfs_matcher = &matchers->smfs_matchers[ipv4 * 3 + tcp * 2 + gre];
+
+	if (refcount_inc_not_zero(&smfs_matcher->ref))
+		return smfs_matcher;
+
+	mutex_lock(&fs_smfs->lock);
+
+	/* Retry with lock, as another thread might have already created the relevant matcher
+	 * till we acquired the lock
+	 */
+	if (refcount_inc_not_zero(&smfs_matcher->ref))
+		goto out_unlock;
+
+	// Find next available priority in sorted used list
+	prio = 0;
+	prev = &matchers->used;
+	list_for_each_entry(m, &matchers->used, list) {
+		prev = &m->list;
+
+		if (m->prio == prio)
+			prio = m->prio + 1;
+		else
+			break;
+	}
+
+	tbl = nat ? fs_smfs->ct_nat_tbl : fs_smfs->ct_tbl;
+	dr_matcher = mlx5_ct_fs_smfs_matcher_create(fs, tbl, ipv4, tcp, gre, prio);
+	if (IS_ERR(dr_matcher)) {
+		netdev_warn(fs->netdev,
+			    "ct_fs_smfs: failed to create matcher (nat %d, ipv4 %d, tcp %d, gre %d), err: %ld\n",
+			    nat, ipv4, tcp, gre, PTR_ERR(dr_matcher));
+
+		smfs_matcher = ERR_CAST(dr_matcher);
+		goto out_unlock;
+	}
+
+	smfs_matcher->dr_matcher = dr_matcher;
+	smfs_matcher->prio = prio;
+	list_add(&smfs_matcher->list, prev);
+	refcount_set(&smfs_matcher->ref, 1);
+
+out_unlock:
+	mutex_unlock(&fs_smfs->lock);
+	return smfs_matcher;
+}
+
+static void
+mlx5_ct_fs_smfs_matcher_put(struct mlx5_ct_fs *fs, struct mlx5_ct_fs_smfs_matcher *smfs_matcher)
+{
+	struct mlx5_ct_fs_smfs *fs_smfs = mlx5_ct_fs_priv(fs);
+
+	if (!refcount_dec_and_mutex_lock(&smfs_matcher->ref, &fs_smfs->lock))
+		return;
+
+	mlx5_smfs_matcher_destroy(smfs_matcher->dr_matcher);
+	list_del(&smfs_matcher->list);
+	mutex_unlock(&fs_smfs->lock);
+}
+
+static int
+mlx5_ct_fs_smfs_init(struct mlx5_ct_fs *fs, struct mlx5_flow_table *ct,
+		     struct mlx5_flow_table *ct_nat, struct mlx5_flow_table *post_ct)
+{
+	struct mlx5dr_table *ct_tbl, *ct_nat_tbl, *post_ct_tbl;
+	struct mlx5_ct_fs_smfs *fs_smfs = mlx5_ct_fs_priv(fs);
+
+	post_ct_tbl = mlx5_smfs_table_get_from_fs_ft(post_ct);
+	ct_nat_tbl = mlx5_smfs_table_get_from_fs_ft(ct_nat);
+	ct_tbl = mlx5_smfs_table_get_from_fs_ft(ct);
+	fs_smfs->ct_nat = ct_nat;
+
+	if (!ct_tbl || !ct_nat_tbl || !post_ct_tbl) {
+		netdev_warn(fs->netdev, "ct_fs_smfs: failed to init, missing backing dr tables");
+		return -EOPNOTSUPP;
+	}
+
+	ct_dbg("using smfs steering");
+
+	fs_smfs->fwd_action = mlx5_smfs_action_create_dest_table(post_ct_tbl);
+	if (!fs_smfs->fwd_action) {
+		return -EINVAL;
+	}
+
+	fs_smfs->ct_tbl = ct_tbl;
+	fs_smfs->ct_nat_tbl = ct_nat_tbl;
+	mutex_init(&fs_smfs->lock);
+	INIT_LIST_HEAD(&fs_smfs->matchers.used);
+	INIT_LIST_HEAD(&fs_smfs->matchers_nat.used);
+
+	return 0;
+}
+
+static void
+mlx5_ct_fs_smfs_destroy(struct mlx5_ct_fs *fs)
+{
+	struct mlx5_ct_fs_smfs *fs_smfs = mlx5_ct_fs_priv(fs);
+
+	mlx5_smfs_action_destroy(fs_smfs->fwd_action);
+}
+
+static inline bool
+mlx5_tc_ct_valid_used_dissector_keys(const u32 used_keys)
+{
+#define DISS_BIT(name) BIT(FLOW_DISSECTOR_KEY_ ## name)
+	const u32 basic_keys = DISS_BIT(BASIC) | DISS_BIT(CONTROL) | DISS_BIT(META);
+	const u32 ipv4_tcp = basic_keys | DISS_BIT(IPV4_ADDRS) | DISS_BIT(PORTS) | DISS_BIT(TCP);
+	const u32 ipv6_tcp = basic_keys | DISS_BIT(IPV6_ADDRS) | DISS_BIT(PORTS) | DISS_BIT(TCP);
+	const u32 ipv4_udp = basic_keys | DISS_BIT(IPV4_ADDRS) | DISS_BIT(PORTS);
+	const u32 ipv6_udp = basic_keys | DISS_BIT(IPV6_ADDRS) | DISS_BIT(PORTS);
+	const u32 ipv4_gre = basic_keys | DISS_BIT(IPV4_ADDRS);
+	const u32 ipv6_gre = basic_keys | DISS_BIT(IPV6_ADDRS);
+
+	return (used_keys == ipv4_tcp || used_keys == ipv4_udp || used_keys == ipv6_tcp ||
+		used_keys == ipv6_udp || used_keys == ipv4_gre || used_keys == ipv6_gre);
+}
+
+static bool
+mlx5_ct_fs_smfs_ct_validate_flow_rule(struct mlx5_ct_fs *fs, struct flow_rule *flow_rule)
+{
+	struct flow_match_ipv4_addrs ipv4_addrs;
+	struct flow_match_ipv6_addrs ipv6_addrs;
+	struct flow_match_control control;
+	struct flow_match_basic basic;
+	struct flow_match_ports ports;
+	struct flow_match_tcp tcp;
+
+	if (!mlx5_tc_ct_valid_used_dissector_keys(flow_rule->match.dissector->used_keys)) {
+		ct_dbg("rule uses unexpected dissectors (0x%08x)",
+		       flow_rule->match.dissector->used_keys);
+		return false;
+	}
+
+	flow_rule_match_basic(flow_rule, &basic);
+	flow_rule_match_control(flow_rule, &control);
+	flow_rule_match_ipv4_addrs(flow_rule, &ipv4_addrs);
+	flow_rule_match_ipv6_addrs(flow_rule, &ipv6_addrs);
+	if (basic.key->ip_proto != IPPROTO_GRE)
+		flow_rule_match_ports(flow_rule, &ports);
+	if (basic.key->ip_proto == IPPROTO_TCP)
+		flow_rule_match_tcp(flow_rule, &tcp);
+
+	if (basic.mask->n_proto != htons(0xFFFF) ||
+	    (basic.key->n_proto != htons(ETH_P_IP) && basic.key->n_proto != htons(ETH_P_IPV6)) ||
+	    basic.mask->ip_proto != 0xFF ||
+	    (basic.key->ip_proto != IPPROTO_UDP && basic.key->ip_proto != IPPROTO_TCP &&
+	     basic.key->ip_proto != IPPROTO_GRE)) {
+		ct_dbg("rule uses unexpected basic match (n_proto 0x%04x/0x%04x, ip_proto 0x%02x/0x%02x)",
+		       ntohs(basic.key->n_proto), ntohs(basic.mask->n_proto),
+		       basic.key->ip_proto, basic.mask->ip_proto);
+		return false;
+	}
+
+	if (basic.key->ip_proto != IPPROTO_GRE &&
+	    (ports.mask->src != htons(0xFFFF) || ports.mask->dst != htons(0xFFFF))) {
+		ct_dbg("rule uses ports match (src 0x%04x, dst 0x%04x)",
+		       ports.mask->src, ports.mask->dst);
+		return false;
+	}
+
+	if (basic.key->ip_proto == IPPROTO_TCP && tcp.mask->flags != MLX5_CT_TCP_FLAGS_MASK) {
+		ct_dbg("rule uses unexpected tcp match (flags 0x%02x)", tcp.mask->flags);
+		return false;
+	}
+
+	return true;
+}
+
+static struct mlx5_ct_fs_rule *
+mlx5_ct_fs_smfs_ct_rule_add(struct mlx5_ct_fs *fs, struct mlx5_flow_spec *spec,
+			    struct mlx5_flow_attr *attr, struct flow_rule *flow_rule)
+{
+	struct mlx5_ct_fs_smfs *fs_smfs = mlx5_ct_fs_priv(fs);
+	struct mlx5_ct_fs_smfs_matcher *smfs_matcher;
+	struct mlx5_ct_fs_smfs_rule *smfs_rule;
+	struct mlx5dr_action *actions[5];
+	struct mlx5dr_rule *rule;
+	int num_actions = 0, err;
+	bool nat, tcp, ipv4, gre;
+
+	if (!mlx5_ct_fs_smfs_ct_validate_flow_rule(fs, flow_rule))
+		return ERR_PTR(-EOPNOTSUPP);
+
+	smfs_rule = kzalloc(sizeof(*smfs_rule), GFP_KERNEL);
+	if (!smfs_rule)
+		return ERR_PTR(-ENOMEM);
+
+	smfs_rule->count_action = mlx5_smfs_action_create_flow_counter(mlx5_fc_id(attr->counter));
+	if (!smfs_rule->count_action) {
+		err = -EINVAL;
+		goto err_count;
+	}
+
+	actions[num_actions++] = smfs_rule->count_action;
+	actions[num_actions++] = attr->modify_hdr->action.dr_action;
+	actions[num_actions++] = fs_smfs->fwd_action;
+
+	nat = (attr->ft == fs_smfs->ct_nat);
+	ipv4 = mlx5e_tc_get_ip_version(spec, true) == 4;
+	tcp = MLX5_GET(fte_match_param, spec->match_value,
+		       outer_headers.ip_protocol) == IPPROTO_TCP;
+	gre = MLX5_GET(fte_match_param, spec->match_value,
+		       outer_headers.ip_protocol) == IPPROTO_GRE;
+
+	smfs_matcher = mlx5_ct_fs_smfs_matcher_get(fs, nat, ipv4, tcp, gre);
+	if (IS_ERR(smfs_matcher)) {
+		err = PTR_ERR(smfs_matcher);
+		goto err_matcher;
+	}
+
+	rule = mlx5_smfs_rule_create(smfs_matcher->dr_matcher, spec, num_actions, actions,
+				     spec->flow_context.flow_source);
+	if (!rule) {
+		err = -EINVAL;
+		goto err_create;
+	}
+
+	smfs_rule->rule = rule;
+	smfs_rule->smfs_matcher = smfs_matcher;
+
+	return &smfs_rule->fs_rule;
+
+err_create:
+	mlx5_ct_fs_smfs_matcher_put(fs, smfs_matcher);
+err_matcher:
+	mlx5_smfs_action_destroy(smfs_rule->count_action);
+err_count:
+	kfree(smfs_rule);
+	return ERR_PTR(err);
+}
+
+static void
+mlx5_ct_fs_smfs_ct_rule_del(struct mlx5_ct_fs *fs, struct mlx5_ct_fs_rule *fs_rule)
+{
+	struct mlx5_ct_fs_smfs_rule *smfs_rule = container_of(fs_rule,
+							      struct mlx5_ct_fs_smfs_rule,
+							      fs_rule);
+
+	mlx5_smfs_rule_destroy(smfs_rule->rule);
+	mlx5_ct_fs_smfs_matcher_put(fs, smfs_rule->smfs_matcher);
+	mlx5_smfs_action_destroy(smfs_rule->count_action);
+	kfree(smfs_rule);
+}
+
+static struct mlx5_ct_fs_ops fs_smfs_ops = {
+	.ct_rule_add = mlx5_ct_fs_smfs_ct_rule_add,
+	.ct_rule_del = mlx5_ct_fs_smfs_ct_rule_del,
+
+	.init = mlx5_ct_fs_smfs_init,
+	.destroy = mlx5_ct_fs_smfs_destroy,
+
+	.priv_size = sizeof(struct mlx5_ct_fs_smfs),
+};
+
+struct mlx5_ct_fs_ops *
+mlx5_ct_fs_smfs_ops_get(void)
+{
+	return &fs_smfs_ops;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc/int_port.c b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/int_port.c
new file mode 100644
index 000000000..ca834bbcb
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/int_port.c
@@ -0,0 +1,457 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. All rights reserved. */
+
+#include <linux/mlx5/fs.h>
+#include "en/mapping.h"
+#include "en/tc/int_port.h"
+#include "en.h"
+#include "en_rep.h"
+#include "en_tc.h"
+
+struct mlx5e_tc_int_port {
+	enum mlx5e_tc_int_port_type type;
+	int ifindex;
+	u32 match_metadata;
+	u32 mapping;
+	struct list_head list;
+	struct mlx5_flow_handle *rx_rule;
+	refcount_t refcnt;
+	struct rcu_head rcu_head;
+};
+
+struct mlx5e_tc_int_port_priv {
+	struct mlx5_core_dev *dev;
+	struct mutex int_ports_lock; /* Protects int ports list */
+	struct list_head int_ports; /* Uses int_ports_lock */
+	u16 num_ports;
+	bool ul_rep_rx_ready; /* Set when uplink is performing teardown */
+	struct mapping_ctx *metadata_mapping; /* Metadata for source port rewrite and matching */
+};
+
+bool mlx5e_tc_int_port_supported(const struct mlx5_eswitch *esw)
+{
+	return mlx5_eswitch_vport_match_metadata_enabled(esw) &&
+	       MLX5_CAP_GEN(esw->dev, reg_c_preserve);
+}
+
+u32 mlx5e_tc_int_port_get_metadata(struct mlx5e_tc_int_port *int_port)
+{
+	return int_port->match_metadata;
+}
+
+int mlx5e_tc_int_port_get_flow_source(struct mlx5e_tc_int_port *int_port)
+{
+	/* For egress forwarding we can have the case
+	 * where the packet came from a vport and redirected
+	 * to int port or it came from the uplink, going
+	 * via internal port and hairpinned back to uplink
+	 * so we set the source to any port in this case.
+	 */
+	return int_port->type == MLX5E_TC_INT_PORT_EGRESS ?
+		MLX5_FLOW_CONTEXT_FLOW_SOURCE_ANY_VPORT :
+		MLX5_FLOW_CONTEXT_FLOW_SOURCE_UPLINK;
+}
+
+u32 mlx5e_tc_int_port_get_metadata_for_match(struct mlx5e_tc_int_port *int_port)
+{
+	return int_port->match_metadata << (32 - ESW_SOURCE_PORT_METADATA_BITS);
+}
+
+static struct mlx5_flow_handle *
+mlx5e_int_port_create_rx_rule(struct mlx5_eswitch *esw,
+			      struct mlx5e_tc_int_port *int_port,
+			      struct mlx5_flow_destination *dest)
+
+{
+	struct mlx5_flow_context *flow_context;
+	struct mlx5_flow_act flow_act = {};
+	struct mlx5_flow_handle *flow_rule;
+	struct mlx5_flow_spec *spec;
+	void *misc;
+
+	spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
+	if (!spec)
+		return ERR_PTR(-ENOMEM);
+
+	misc = MLX5_ADDR_OF(fte_match_param, spec->match_value, misc_parameters_2);
+	MLX5_SET(fte_match_set_misc2, misc, metadata_reg_c_0,
+		 mlx5e_tc_int_port_get_metadata_for_match(int_port));
+
+	misc = MLX5_ADDR_OF(fte_match_param, spec->match_criteria, misc_parameters_2);
+	MLX5_SET(fte_match_set_misc2, misc, metadata_reg_c_0,
+		 mlx5_eswitch_get_vport_metadata_mask());
+
+	spec->match_criteria_enable = MLX5_MATCH_MISC_PARAMETERS_2;
+
+	/* Overwrite flow tag with the int port metadata mapping
+	 * instead of the chain mapping.
+	 */
+	flow_context = &spec->flow_context;
+	flow_context->flags |= FLOW_CONTEXT_HAS_TAG;
+	flow_context->flow_tag = int_port->mapping;
+	flow_act.action = MLX5_FLOW_CONTEXT_ACTION_FWD_DEST;
+	flow_rule = mlx5_add_flow_rules(esw->offloads.ft_offloads, spec,
+					&flow_act, dest, 1);
+	if (IS_ERR(flow_rule))
+		mlx5_core_warn(esw->dev, "ft offloads: Failed to add internal vport rx rule err %ld\n",
+			       PTR_ERR(flow_rule));
+
+	kvfree(spec);
+
+	return flow_rule;
+}
+
+static struct mlx5e_tc_int_port *
+mlx5e_int_port_lookup(struct mlx5e_tc_int_port_priv *priv,
+		      int ifindex,
+		      enum mlx5e_tc_int_port_type type)
+{
+	struct mlx5e_tc_int_port *int_port;
+
+	if (!priv->ul_rep_rx_ready)
+		goto not_found;
+
+	list_for_each_entry(int_port, &priv->int_ports, list)
+		if (int_port->ifindex == ifindex && int_port->type == type) {
+			refcount_inc(&int_port->refcnt);
+			return int_port;
+		}
+
+not_found:
+	return NULL;
+}
+
+static int mlx5e_int_port_metadata_alloc(struct mlx5e_tc_int_port_priv *priv,
+					 int ifindex, enum mlx5e_tc_int_port_type type,
+					 u32 *id)
+{
+	u32 mapped_key[2] = {type, ifindex};
+	int err;
+
+	err = mapping_add(priv->metadata_mapping, mapped_key, id);
+	if (err)
+		return err;
+
+	/* Fill upper 4 bits of PFNUM with reserved value */
+	*id |= 0xf << ESW_VPORT_BITS;
+
+	return 0;
+}
+
+static void mlx5e_int_port_metadata_free(struct mlx5e_tc_int_port_priv *priv,
+					 u32 id)
+{
+	id &= (1 << ESW_VPORT_BITS) - 1;
+	mapping_remove(priv->metadata_mapping, id);
+}
+
+/* Must be called with priv->int_ports_lock held */
+static struct mlx5e_tc_int_port *
+mlx5e_int_port_add(struct mlx5e_tc_int_port_priv *priv,
+		   int ifindex,
+		   enum mlx5e_tc_int_port_type type)
+{
+	struct mlx5_eswitch *esw = priv->dev->priv.eswitch;
+	struct mlx5_mapped_obj mapped_obj = {};
+	struct mlx5e_rep_priv *uplink_rpriv;
+	struct mlx5e_tc_int_port *int_port;
+	struct mlx5_flow_destination dest;
+	struct mapping_ctx *ctx;
+	u32 match_metadata;
+	u32 mapping;
+	int err;
+
+	if (priv->num_ports == MLX5E_TC_MAX_INT_PORT_NUM) {
+		mlx5_core_dbg(priv->dev, "Cannot add a new int port, max supported %d",
+			      MLX5E_TC_MAX_INT_PORT_NUM);
+		return ERR_PTR(-ENOSPC);
+	}
+
+	int_port = kzalloc(sizeof(*int_port), GFP_KERNEL);
+	if (!int_port)
+		return ERR_PTR(-ENOMEM);
+
+	err = mlx5e_int_port_metadata_alloc(priv, ifindex, type, &match_metadata);
+	if (err) {
+		mlx5_core_warn(esw->dev, "Cannot add a new internal port, metadata allocation failed for ifindex %d",
+			       ifindex);
+		goto err_metadata;
+	}
+
+	/* map metadata to reg_c0 object for miss handling */
+	ctx = esw->offloads.reg_c0_obj_pool;
+	mapped_obj.type = MLX5_MAPPED_OBJ_INT_PORT_METADATA;
+	mapped_obj.int_port_metadata = match_metadata;
+	err = mapping_add(ctx, &mapped_obj, &mapping);
+	if (err)
+		goto err_map;
+
+	int_port->type = type;
+	int_port->ifindex = ifindex;
+	int_port->match_metadata = match_metadata;
+	int_port->mapping = mapping;
+
+	/* Create a match on internal vport metadata in vport table */
+	uplink_rpriv = mlx5_eswitch_get_uplink_priv(esw, REP_ETH);
+
+	dest.type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
+	dest.ft = uplink_rpriv->root_ft;
+
+	int_port->rx_rule = mlx5e_int_port_create_rx_rule(esw, int_port, &dest);
+	if (IS_ERR(int_port->rx_rule)) {
+		err = PTR_ERR(int_port->rx_rule);
+		mlx5_core_warn(esw->dev, "Can't add internal port rx rule, err %d", err);
+		goto err_rx_rule;
+	}
+
+	refcount_set(&int_port->refcnt, 1);
+	list_add_rcu(&int_port->list, &priv->int_ports);
+	priv->num_ports++;
+
+	return int_port;
+
+err_rx_rule:
+	mapping_remove(ctx, int_port->mapping);
+
+err_map:
+	mlx5e_int_port_metadata_free(priv, match_metadata);
+
+err_metadata:
+	kfree(int_port);
+
+	return ERR_PTR(err);
+}
+
+/* Must be called with priv->int_ports_lock held */
+static void
+mlx5e_int_port_remove(struct mlx5e_tc_int_port_priv *priv,
+		      struct mlx5e_tc_int_port *int_port)
+{
+	struct mlx5_eswitch *esw = priv->dev->priv.eswitch;
+	struct mapping_ctx *ctx;
+
+	ctx = esw->offloads.reg_c0_obj_pool;
+
+	list_del_rcu(&int_port->list);
+
+	/* The following parameters are not used by the
+	 * rcu readers of this int_port object so it is
+	 * safe to release them.
+	 */
+	if (int_port->rx_rule)
+		mlx5_del_flow_rules(int_port->rx_rule);
+	mapping_remove(ctx, int_port->mapping);
+	mlx5e_int_port_metadata_free(priv, int_port->match_metadata);
+	kfree_rcu(int_port);
+	priv->num_ports--;
+}
+
+/* Must be called with rcu_read_lock held */
+static struct mlx5e_tc_int_port *
+mlx5e_int_port_get_from_metadata(struct mlx5e_tc_int_port_priv *priv,
+				 u32 metadata)
+{
+	struct mlx5e_tc_int_port *int_port;
+
+	list_for_each_entry_rcu(int_port, &priv->int_ports, list)
+		if (int_port->match_metadata == metadata)
+			return int_port;
+
+	return NULL;
+}
+
+struct mlx5e_tc_int_port *
+mlx5e_tc_int_port_get(struct mlx5e_tc_int_port_priv *priv,
+		      int ifindex,
+		      enum mlx5e_tc_int_port_type type)
+{
+	struct mlx5e_tc_int_port *int_port;
+
+	if (!priv)
+		return ERR_PTR(-EOPNOTSUPP);
+
+	mutex_lock(&priv->int_ports_lock);
+
+	/* Reject request if ul rep not ready */
+	if (!priv->ul_rep_rx_ready) {
+		int_port = ERR_PTR(-EOPNOTSUPP);
+		goto done;
+	}
+
+	int_port = mlx5e_int_port_lookup(priv, ifindex, type);
+	if (int_port)
+		goto done;
+
+	/* Alloc and add new int port to list */
+	int_port = mlx5e_int_port_add(priv, ifindex, type);
+
+done:
+	mutex_unlock(&priv->int_ports_lock);
+
+	return int_port;
+}
+
+void
+mlx5e_tc_int_port_put(struct mlx5e_tc_int_port_priv *priv,
+		      struct mlx5e_tc_int_port *int_port)
+{
+	if (!refcount_dec_and_mutex_lock(&int_port->refcnt, &priv->int_ports_lock))
+		return;
+
+	mlx5e_int_port_remove(priv, int_port);
+	mutex_unlock(&priv->int_ports_lock);
+}
+
+struct mlx5e_tc_int_port_priv *
+mlx5e_tc_int_port_init(struct mlx5e_priv *priv)
+{
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	struct mlx5e_tc_int_port_priv *int_port_priv;
+	u64 mapping_id;
+
+	if (!mlx5e_tc_int_port_supported(esw))
+		return NULL;
+
+	int_port_priv = kzalloc(sizeof(*int_port_priv), GFP_KERNEL);
+	if (!int_port_priv)
+		return NULL;
+
+	mapping_id = mlx5_query_nic_system_image_guid(priv->mdev);
+
+	int_port_priv->metadata_mapping = mapping_create_for_id(mapping_id, MAPPING_TYPE_INT_PORT,
+								sizeof(u32) * 2,
+								(1 << ESW_VPORT_BITS) - 1, true);
+	if (IS_ERR(int_port_priv->metadata_mapping)) {
+		mlx5_core_warn(priv->mdev, "Can't allocate metadata mapping of int port offload, err=%ld\n",
+			       PTR_ERR(int_port_priv->metadata_mapping));
+		goto err_mapping;
+	}
+
+	int_port_priv->dev = priv->mdev;
+	mutex_init(&int_port_priv->int_ports_lock);
+	INIT_LIST_HEAD(&int_port_priv->int_ports);
+
+	return int_port_priv;
+
+err_mapping:
+	kfree(int_port_priv);
+
+	return NULL;
+}
+
+void
+mlx5e_tc_int_port_cleanup(struct mlx5e_tc_int_port_priv *priv)
+{
+	if (!priv)
+		return;
+
+	mutex_destroy(&priv->int_ports_lock);
+	mapping_destroy(priv->metadata_mapping);
+	kfree(priv);
+}
+
+/* Int port rx rules reside in ul rep rx tables.
+ * It is possible the ul rep will go down while there are
+ * still int port rules in its rx table so proper cleanup
+ * is required to free resources.
+ */
+void mlx5e_tc_int_port_init_rep_rx(struct mlx5e_priv *priv)
+{
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	struct mlx5_rep_uplink_priv *uplink_priv;
+	struct mlx5e_tc_int_port_priv *ppriv;
+	struct mlx5e_rep_priv *uplink_rpriv;
+
+	uplink_rpriv = mlx5_eswitch_get_uplink_priv(esw, REP_ETH);
+	uplink_priv = &uplink_rpriv->uplink_priv;
+
+	ppriv = uplink_priv->int_port_priv;
+
+	if (!ppriv)
+		return;
+
+	mutex_lock(&ppriv->int_ports_lock);
+	ppriv->ul_rep_rx_ready = true;
+	mutex_unlock(&ppriv->int_ports_lock);
+}
+
+void mlx5e_tc_int_port_cleanup_rep_rx(struct mlx5e_priv *priv)
+{
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	struct mlx5_rep_uplink_priv *uplink_priv;
+	struct mlx5e_tc_int_port_priv *ppriv;
+	struct mlx5e_rep_priv *uplink_rpriv;
+	struct mlx5e_tc_int_port *int_port;
+
+	uplink_rpriv = mlx5_eswitch_get_uplink_priv(esw, REP_ETH);
+	uplink_priv = &uplink_rpriv->uplink_priv;
+
+	ppriv = uplink_priv->int_port_priv;
+
+	if (!ppriv)
+		return;
+
+	mutex_lock(&ppriv->int_ports_lock);
+
+	ppriv->ul_rep_rx_ready = false;
+
+	list_for_each_entry(int_port, &ppriv->int_ports, list) {
+		if (!IS_ERR_OR_NULL(int_port->rx_rule))
+			mlx5_del_flow_rules(int_port->rx_rule);
+
+		int_port->rx_rule = NULL;
+	}
+
+	mutex_unlock(&ppriv->int_ports_lock);
+}
+
+bool
+mlx5e_tc_int_port_dev_fwd(struct mlx5e_tc_int_port_priv *priv,
+			  struct sk_buff *skb, u32 int_vport_metadata,
+			  bool *forward_tx)
+{
+	enum mlx5e_tc_int_port_type fwd_type;
+	struct mlx5e_tc_int_port *int_port;
+	struct net_device *dev;
+	int ifindex;
+
+	if (!priv)
+		return false;
+
+	rcu_read_lock();
+	int_port = mlx5e_int_port_get_from_metadata(priv, int_vport_metadata);
+	if (!int_port) {
+		rcu_read_unlock();
+		mlx5_core_dbg(priv->dev, "Unable to find int port with metadata 0x%.8x\n",
+			      int_vport_metadata);
+		return false;
+	}
+
+	ifindex = int_port->ifindex;
+	fwd_type = int_port->type;
+	rcu_read_unlock();
+
+	dev = dev_get_by_index(&init_net, ifindex);
+	if (!dev) {
+		mlx5_core_dbg(priv->dev,
+			      "Couldn't find internal port device with ifindex: %d\n",
+			      ifindex);
+		return false;
+	}
+
+	skb->skb_iif = dev->ifindex;
+	skb->dev = dev;
+
+	if (fwd_type == MLX5E_TC_INT_PORT_INGRESS) {
+		skb->pkt_type = PACKET_HOST;
+		skb_set_redirected(skb, true);
+		*forward_tx = false;
+	} else {
+		skb_reset_network_header(skb);
+		skb_push_rcsum(skb, skb->mac_len);
+		skb_set_redirected(skb, false);
+		*forward_tx = true;
+	}
+
+	return true;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc/int_port.h b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/int_port.h
new file mode 100644
index 000000000..e72c79d30
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/int_port.h
@@ -0,0 +1,65 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. All rights reserved. */
+
+#ifndef __MLX5_EN_TC_INT_PORT_H__
+#define __MLX5_EN_TC_INT_PORT_H__
+
+#include "en.h"
+
+struct mlx5e_tc_int_port;
+struct mlx5e_tc_int_port_priv;
+
+enum mlx5e_tc_int_port_type {
+	MLX5E_TC_INT_PORT_INGRESS,
+	MLX5E_TC_INT_PORT_EGRESS,
+};
+
+#if IS_ENABLED(CONFIG_MLX5_CLS_ACT)
+bool mlx5e_tc_int_port_supported(const struct mlx5_eswitch *esw);
+
+struct mlx5e_tc_int_port_priv *
+mlx5e_tc_int_port_init(struct mlx5e_priv *priv);
+void
+mlx5e_tc_int_port_cleanup(struct mlx5e_tc_int_port_priv *priv);
+
+void mlx5e_tc_int_port_init_rep_rx(struct mlx5e_priv *priv);
+void mlx5e_tc_int_port_cleanup_rep_rx(struct mlx5e_priv *priv);
+
+bool
+mlx5e_tc_int_port_dev_fwd(struct mlx5e_tc_int_port_priv *priv,
+			  struct sk_buff *skb, u32 int_vport_metadata,
+			  bool *forward_tx);
+struct mlx5e_tc_int_port *
+mlx5e_tc_int_port_get(struct mlx5e_tc_int_port_priv *priv,
+		      int ifindex,
+		      enum mlx5e_tc_int_port_type type);
+void
+mlx5e_tc_int_port_put(struct mlx5e_tc_int_port_priv *priv,
+		      struct mlx5e_tc_int_port *int_port);
+
+u32 mlx5e_tc_int_port_get_metadata(struct mlx5e_tc_int_port *int_port);
+u32 mlx5e_tc_int_port_get_metadata_for_match(struct mlx5e_tc_int_port *int_port);
+int mlx5e_tc_int_port_get_flow_source(struct mlx5e_tc_int_port *int_port);
+#else /* CONFIG_MLX5_CLS_ACT */
+static inline u32
+mlx5e_tc_int_port_get_metadata_for_match(struct mlx5e_tc_int_port *int_port)
+{
+		return 0;
+}
+
+static inline int
+mlx5e_tc_int_port_get_flow_source(struct mlx5e_tc_int_port *int_port)
+{
+		return 0;
+}
+
+static inline bool mlx5e_tc_int_port_supported(const struct mlx5_eswitch *esw)
+{
+	return false;
+}
+
+static inline void mlx5e_tc_int_port_init_rep_rx(struct mlx5e_priv *priv) {}
+static inline void mlx5e_tc_int_port_cleanup_rep_rx(struct mlx5e_priv *priv) {}
+
+#endif /* CONFIG_MLX5_CLS_ACT */
+#endif /* __MLX5_EN_TC_INT_PORT_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc/meter.c b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/meter.c
new file mode 100644
index 000000000..be74e1403
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/meter.c
@@ -0,0 +1,585 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+// Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+
+#include <linux/math64.h>
+#include "lib/aso.h"
+#include "en/tc/post_act.h"
+#include "meter.h"
+#include "en/tc_priv.h"
+
+#define MLX5_START_COLOR_SHIFT 28
+#define MLX5_METER_MODE_SHIFT 24
+#define MLX5_CBS_EXP_SHIFT 24
+#define MLX5_CBS_MAN_SHIFT 16
+#define MLX5_CIR_EXP_SHIFT 8
+
+/* cir = 8*(10^9)*cir_mantissa/(2^cir_exponent)) bits/s */
+#define MLX5_CONST_CIR 8000000000ULL
+#define MLX5_CALC_CIR(m, e)  ((MLX5_CONST_CIR * (m)) >> (e))
+#define MLX5_MAX_CIR ((MLX5_CONST_CIR * 0x100) - 1)
+
+/* cbs = cbs_mantissa*2^cbs_exponent */
+#define MLX5_CALC_CBS(m, e)  ((m) << (e))
+#define MLX5_MAX_CBS ((0x100ULL << 0x1F) - 1)
+#define MLX5_MAX_HW_CBS 0x7FFFFFFF
+
+struct mlx5e_flow_meter_aso_obj {
+	struct list_head entry;
+	int base_id;
+	int total_meters;
+
+	unsigned long meters_map[0]; /* must be at the end of this struct */
+};
+
+struct mlx5e_flow_meters {
+	enum mlx5_flow_namespace_type ns_type;
+	struct mlx5_aso *aso;
+	struct mutex aso_lock; /* Protects aso operations */
+	int log_granularity;
+	u32 pdn;
+
+	DECLARE_HASHTABLE(hashtbl, 8);
+
+	struct mutex sync_lock; /* protect flow meter operations */
+	struct list_head partial_list;
+	struct list_head full_list;
+
+	struct mlx5_core_dev *mdev;
+	struct mlx5e_post_act *post_act;
+};
+
+static void
+mlx5e_flow_meter_cir_calc(u64 cir, u8 *man, u8 *exp)
+{
+	s64 _cir, _delta, delta = S64_MAX;
+	u8 e, _man = 0, _exp = 0;
+	u64 m;
+
+	for (e = 0; e <= 0x1F; e++) { /* exp width 5bit */
+		m = cir << e;
+		if ((s64)m < 0) /* overflow */
+			break;
+		m = div64_u64(m, MLX5_CONST_CIR);
+		if (m > 0xFF) /* man width 8 bit */
+			continue;
+		_cir = MLX5_CALC_CIR(m, e);
+		_delta = cir - _cir;
+		if (_delta < delta) {
+			_man = m;
+			_exp = e;
+			if (!_delta)
+				goto found;
+			delta = _delta;
+		}
+	}
+
+found:
+	*man = _man;
+	*exp = _exp;
+}
+
+static void
+mlx5e_flow_meter_cbs_calc(u64 cbs, u8 *man, u8 *exp)
+{
+	s64 _cbs, _delta, delta = S64_MAX;
+	u8 e, _man = 0, _exp = 0;
+	u64 m;
+
+	for (e = 0; e <= 0x1F; e++) { /* exp width 5bit */
+		m = cbs >> e;
+		if (m > 0xFF) /* man width 8 bit */
+			continue;
+		_cbs = MLX5_CALC_CBS(m, e);
+		_delta = cbs - _cbs;
+		if (_delta < delta) {
+			_man = m;
+			_exp = e;
+			if (!_delta)
+				goto found;
+			delta = _delta;
+		}
+	}
+
+found:
+	*man = _man;
+	*exp = _exp;
+}
+
+int
+mlx5e_tc_meter_modify(struct mlx5_core_dev *mdev,
+		      struct mlx5e_flow_meter_handle *meter,
+		      struct mlx5e_flow_meter_params *meter_params)
+{
+	struct mlx5_wqe_aso_ctrl_seg *aso_ctrl;
+	struct mlx5_wqe_aso_data_seg *aso_data;
+	struct mlx5e_flow_meters *flow_meters;
+	u8 cir_man, cir_exp, cbs_man, cbs_exp;
+	struct mlx5_aso_wqe *aso_wqe;
+	unsigned long expires;
+	struct mlx5_aso *aso;
+	u64 rate, burst;
+	u8 ds_cnt;
+	int err;
+
+	rate = meter_params->rate;
+	burst = meter_params->burst;
+
+	/* HW treats each packet as 128 bytes in PPS mode */
+	if (meter_params->mode == MLX5_RATE_LIMIT_PPS) {
+		rate <<= 10;
+		burst <<= 7;
+	}
+
+	if (!rate || rate > MLX5_MAX_CIR || !burst || burst > MLX5_MAX_CBS)
+		return -EINVAL;
+
+	/* HW has limitation of total 31 bits for cbs */
+	if (burst > MLX5_MAX_HW_CBS) {
+		mlx5_core_warn(mdev,
+			       "burst(%lld) is too large, use HW allowed value(%d)\n",
+			       burst, MLX5_MAX_HW_CBS);
+		burst = MLX5_MAX_HW_CBS;
+	}
+
+	mlx5_core_dbg(mdev, "meter mode=%d\n", meter_params->mode);
+	mlx5e_flow_meter_cir_calc(rate, &cir_man, &cir_exp);
+	mlx5_core_dbg(mdev, "rate=%lld, cir=%lld, exp=%d, man=%d\n",
+		      rate, MLX5_CALC_CIR(cir_man, cir_exp), cir_exp, cir_man);
+	mlx5e_flow_meter_cbs_calc(burst, &cbs_man, &cbs_exp);
+	mlx5_core_dbg(mdev, "burst=%lld, cbs=%lld, exp=%d, man=%d\n",
+		      burst, MLX5_CALC_CBS((u64)cbs_man, cbs_exp), cbs_exp, cbs_man);
+
+	if (!cir_man || !cbs_man)
+		return -EINVAL;
+
+	flow_meters = meter->flow_meters;
+	aso = flow_meters->aso;
+
+	mutex_lock(&flow_meters->aso_lock);
+	aso_wqe = mlx5_aso_get_wqe(aso);
+	ds_cnt = DIV_ROUND_UP(sizeof(struct mlx5_aso_wqe_data), MLX5_SEND_WQE_DS);
+	mlx5_aso_build_wqe(aso, ds_cnt, aso_wqe, meter->obj_id,
+			   MLX5_ACCESS_ASO_OPC_MOD_FLOW_METER);
+
+	aso_ctrl = &aso_wqe->aso_ctrl;
+	memset(aso_ctrl, 0, sizeof(*aso_ctrl));
+	aso_ctrl->data_mask_mode = MLX5_ASO_DATA_MASK_MODE_BYTEWISE_64BYTE << 6;
+	aso_ctrl->condition_1_0_operand = MLX5_ASO_ALWAYS_TRUE |
+					  MLX5_ASO_ALWAYS_TRUE << 4;
+	aso_ctrl->data_offset_condition_operand = MLX5_ASO_LOGICAL_OR << 6;
+	aso_ctrl->data_mask = cpu_to_be64(0x80FFFFFFULL << (meter->idx ? 0 : 32));
+
+	aso_data = (struct mlx5_wqe_aso_data_seg *)(aso_wqe + 1);
+	memset(aso_data, 0, sizeof(*aso_data));
+	aso_data->bytewise_data[meter->idx * 8] = cpu_to_be32((0x1 << 31) | /* valid */
+					(MLX5_FLOW_METER_COLOR_GREEN << MLX5_START_COLOR_SHIFT));
+	if (meter_params->mode == MLX5_RATE_LIMIT_PPS)
+		aso_data->bytewise_data[meter->idx * 8] |=
+			cpu_to_be32(MLX5_FLOW_METER_MODE_NUM_PACKETS << MLX5_METER_MODE_SHIFT);
+	else
+		aso_data->bytewise_data[meter->idx * 8] |=
+			cpu_to_be32(MLX5_FLOW_METER_MODE_BYTES_IP_LENGTH << MLX5_METER_MODE_SHIFT);
+
+	aso_data->bytewise_data[meter->idx * 8 + 2] = cpu_to_be32((cbs_exp << MLX5_CBS_EXP_SHIFT) |
+								  (cbs_man << MLX5_CBS_MAN_SHIFT) |
+								  (cir_exp << MLX5_CIR_EXP_SHIFT) |
+								  cir_man);
+
+	mlx5_aso_post_wqe(aso, true, &aso_wqe->ctrl);
+
+	/* With newer FW, the wait for the first ASO WQE is more than 2us, put the wait 10ms. */
+	expires = jiffies + msecs_to_jiffies(10);
+	do {
+		err = mlx5_aso_poll_cq(aso, true);
+		if (err)
+			usleep_range(2, 10);
+	} while (err && time_is_after_jiffies(expires));
+	mutex_unlock(&flow_meters->aso_lock);
+
+	return err;
+}
+
+static int
+mlx5e_flow_meter_create_aso_obj(struct mlx5e_flow_meters *flow_meters, int *obj_id)
+{
+	u32 in[MLX5_ST_SZ_DW(create_flow_meter_aso_obj_in)] = {};
+	u32 out[MLX5_ST_SZ_DW(general_obj_out_cmd_hdr)];
+	struct mlx5_core_dev *mdev = flow_meters->mdev;
+	void *obj;
+	int err;
+
+	MLX5_SET(general_obj_in_cmd_hdr, in, opcode, MLX5_CMD_OP_CREATE_GENERAL_OBJECT);
+	MLX5_SET(general_obj_in_cmd_hdr, in, obj_type,
+		 MLX5_GENERAL_OBJECT_TYPES_FLOW_METER_ASO);
+	MLX5_SET(general_obj_in_cmd_hdr, in, log_obj_range, flow_meters->log_granularity);
+
+	obj = MLX5_ADDR_OF(create_flow_meter_aso_obj_in, in, flow_meter_aso_obj);
+	MLX5_SET(flow_meter_aso_obj, obj, meter_aso_access_pd, flow_meters->pdn);
+
+	err = mlx5_cmd_exec(mdev, in, sizeof(in), out, sizeof(out));
+	if (!err) {
+		*obj_id = MLX5_GET(general_obj_out_cmd_hdr, out, obj_id);
+		mlx5_core_dbg(mdev, "flow meter aso obj(0x%x) created\n", *obj_id);
+	}
+
+	return err;
+}
+
+static void
+mlx5e_flow_meter_destroy_aso_obj(struct mlx5_core_dev *mdev, u32 obj_id)
+{
+	u32 in[MLX5_ST_SZ_DW(general_obj_in_cmd_hdr)] = {};
+	u32 out[MLX5_ST_SZ_DW(general_obj_out_cmd_hdr)];
+
+	MLX5_SET(general_obj_in_cmd_hdr, in, opcode, MLX5_CMD_OP_DESTROY_GENERAL_OBJECT);
+	MLX5_SET(general_obj_in_cmd_hdr, in, obj_type,
+		 MLX5_GENERAL_OBJECT_TYPES_FLOW_METER_ASO);
+	MLX5_SET(general_obj_in_cmd_hdr, in, obj_id, obj_id);
+
+	mlx5_cmd_exec(mdev, in, sizeof(in), out, sizeof(out));
+	mlx5_core_dbg(mdev, "flow meter aso obj(0x%x) destroyed\n", obj_id);
+}
+
+static struct mlx5e_flow_meter_handle *
+__mlx5e_flow_meter_alloc(struct mlx5e_flow_meters *flow_meters)
+{
+	struct mlx5_core_dev *mdev = flow_meters->mdev;
+	struct mlx5e_flow_meter_aso_obj *meters_obj;
+	struct mlx5e_flow_meter_handle *meter;
+	struct mlx5_fc *counter;
+	int err, pos, total;
+	u32 id;
+
+	meter = kzalloc(sizeof(*meter), GFP_KERNEL);
+	if (!meter)
+		return ERR_PTR(-ENOMEM);
+
+	counter = mlx5_fc_create(mdev, true);
+	if (IS_ERR(counter)) {
+		err = PTR_ERR(counter);
+		goto err_red_counter;
+	}
+	meter->red_counter = counter;
+
+	counter = mlx5_fc_create(mdev, true);
+	if (IS_ERR(counter)) {
+		err = PTR_ERR(counter);
+		goto err_green_counter;
+	}
+	meter->green_counter = counter;
+
+	meters_obj = list_first_entry_or_null(&flow_meters->partial_list,
+					      struct mlx5e_flow_meter_aso_obj,
+					      entry);
+	/* 2 meters in one object */
+	total = 1 << (flow_meters->log_granularity + 1);
+	if (!meters_obj) {
+		err = mlx5e_flow_meter_create_aso_obj(flow_meters, &id);
+		if (err) {
+			mlx5_core_err(mdev, "Failed to create flow meter ASO object\n");
+			goto err_create;
+		}
+
+		meters_obj = kzalloc(sizeof(*meters_obj) + BITS_TO_BYTES(total),
+				     GFP_KERNEL);
+		if (!meters_obj) {
+			err = -ENOMEM;
+			goto err_mem;
+		}
+
+		meters_obj->base_id = id;
+		meters_obj->total_meters = total;
+		list_add(&meters_obj->entry, &flow_meters->partial_list);
+		pos = 0;
+	} else {
+		pos = find_first_zero_bit(meters_obj->meters_map, total);
+		if (bitmap_weight(meters_obj->meters_map, total) == total - 1) {
+			list_del(&meters_obj->entry);
+			list_add(&meters_obj->entry, &flow_meters->full_list);
+		}
+	}
+
+	bitmap_set(meters_obj->meters_map, pos, 1);
+	meter->flow_meters = flow_meters;
+	meter->meters_obj = meters_obj;
+	meter->obj_id = meters_obj->base_id + pos / 2;
+	meter->idx = pos % 2;
+
+	mlx5_core_dbg(mdev, "flow meter allocated, obj_id=0x%x, index=%d\n",
+		      meter->obj_id, meter->idx);
+
+	return meter;
+
+err_mem:
+	mlx5e_flow_meter_destroy_aso_obj(mdev, id);
+err_create:
+	mlx5_fc_destroy(mdev, meter->green_counter);
+err_green_counter:
+	mlx5_fc_destroy(mdev, meter->red_counter);
+err_red_counter:
+	kfree(meter);
+	return ERR_PTR(err);
+}
+
+static void
+__mlx5e_flow_meter_free(struct mlx5e_flow_meter_handle *meter)
+{
+	struct mlx5e_flow_meters *flow_meters = meter->flow_meters;
+	struct mlx5_core_dev *mdev = flow_meters->mdev;
+	struct mlx5e_flow_meter_aso_obj *meters_obj;
+	int n, pos;
+
+	mlx5_fc_destroy(mdev, meter->green_counter);
+	mlx5_fc_destroy(mdev, meter->red_counter);
+
+	meters_obj = meter->meters_obj;
+	pos = (meter->obj_id - meters_obj->base_id) * 2 + meter->idx;
+	bitmap_clear(meters_obj->meters_map, pos, 1);
+	n = bitmap_weight(meters_obj->meters_map, meters_obj->total_meters);
+	if (n == 0) {
+		list_del(&meters_obj->entry);
+		mlx5e_flow_meter_destroy_aso_obj(mdev, meters_obj->base_id);
+		kfree(meters_obj);
+	} else if (n == meters_obj->total_meters - 1) {
+		list_del(&meters_obj->entry);
+		list_add(&meters_obj->entry, &flow_meters->partial_list);
+	}
+
+	mlx5_core_dbg(mdev, "flow meter freed, obj_id=0x%x, index=%d\n",
+		      meter->obj_id, meter->idx);
+	kfree(meter);
+}
+
+static struct mlx5e_flow_meter_handle *
+__mlx5e_tc_meter_get(struct mlx5e_flow_meters *flow_meters, u32 index)
+{
+	struct mlx5e_flow_meter_handle *meter;
+
+	hash_for_each_possible(flow_meters->hashtbl, meter, hlist, index)
+		if (meter->params.index == index)
+			goto add_ref;
+
+	return ERR_PTR(-ENOENT);
+
+add_ref:
+	meter->refcnt++;
+
+	return meter;
+}
+
+struct mlx5e_flow_meter_handle *
+mlx5e_tc_meter_get(struct mlx5_core_dev *mdev, struct mlx5e_flow_meter_params *params)
+{
+	struct mlx5e_flow_meters *flow_meters;
+	struct mlx5e_flow_meter_handle *meter;
+
+	flow_meters = mlx5e_get_flow_meters(mdev);
+	if (!flow_meters)
+		return ERR_PTR(-EOPNOTSUPP);
+
+	mutex_lock(&flow_meters->sync_lock);
+	meter = __mlx5e_tc_meter_get(flow_meters, params->index);
+	mutex_unlock(&flow_meters->sync_lock);
+
+	return meter;
+}
+
+static void
+__mlx5e_tc_meter_put(struct mlx5e_flow_meter_handle *meter)
+{
+	if (--meter->refcnt == 0) {
+		hash_del(&meter->hlist);
+		__mlx5e_flow_meter_free(meter);
+	}
+}
+
+void
+mlx5e_tc_meter_put(struct mlx5e_flow_meter_handle *meter)
+{
+	struct mlx5e_flow_meters *flow_meters = meter->flow_meters;
+
+	mutex_lock(&flow_meters->sync_lock);
+	__mlx5e_tc_meter_put(meter);
+	mutex_unlock(&flow_meters->sync_lock);
+}
+
+static struct mlx5e_flow_meter_handle *
+mlx5e_tc_meter_alloc(struct mlx5e_flow_meters *flow_meters,
+		     struct mlx5e_flow_meter_params *params)
+{
+	struct mlx5e_flow_meter_handle *meter;
+
+	meter = __mlx5e_flow_meter_alloc(flow_meters);
+	if (IS_ERR(meter))
+		return meter;
+
+	hash_add(flow_meters->hashtbl, &meter->hlist, params->index);
+	meter->params.index = params->index;
+	meter->refcnt++;
+
+	return meter;
+}
+
+static int
+__mlx5e_tc_meter_update(struct mlx5e_flow_meter_handle *meter,
+			struct mlx5e_flow_meter_params *params)
+{
+	struct mlx5_core_dev *mdev = meter->flow_meters->mdev;
+	int err = 0;
+
+	if (meter->params.mode != params->mode || meter->params.rate != params->rate ||
+	    meter->params.burst != params->burst) {
+		err = mlx5e_tc_meter_modify(mdev, meter, params);
+		if (err)
+			goto out;
+
+		meter->params.mode = params->mode;
+		meter->params.rate = params->rate;
+		meter->params.burst = params->burst;
+	}
+
+out:
+	return err;
+}
+
+int
+mlx5e_tc_meter_update(struct mlx5e_flow_meter_handle *meter,
+		      struct mlx5e_flow_meter_params *params)
+{
+	struct mlx5_core_dev *mdev = meter->flow_meters->mdev;
+	struct mlx5e_flow_meters *flow_meters;
+	int err;
+
+	flow_meters = mlx5e_get_flow_meters(mdev);
+	if (!flow_meters)
+		return -EOPNOTSUPP;
+
+	mutex_lock(&flow_meters->sync_lock);
+	err = __mlx5e_tc_meter_update(meter, params);
+	mutex_unlock(&flow_meters->sync_lock);
+	return err;
+}
+
+struct mlx5e_flow_meter_handle *
+mlx5e_tc_meter_replace(struct mlx5_core_dev *mdev, struct mlx5e_flow_meter_params *params)
+{
+	struct mlx5e_flow_meters *flow_meters;
+	struct mlx5e_flow_meter_handle *meter;
+	int err;
+
+	flow_meters = mlx5e_get_flow_meters(mdev);
+	if (!flow_meters)
+		return ERR_PTR(-EOPNOTSUPP);
+
+	mutex_lock(&flow_meters->sync_lock);
+	meter = __mlx5e_tc_meter_get(flow_meters, params->index);
+	if (IS_ERR(meter)) {
+		meter = mlx5e_tc_meter_alloc(flow_meters, params);
+		if (IS_ERR(meter)) {
+			err = PTR_ERR(meter);
+			goto err_get;
+		}
+	}
+
+	err = __mlx5e_tc_meter_update(meter, params);
+	if (err)
+		goto err_update;
+
+	mutex_unlock(&flow_meters->sync_lock);
+	return meter;
+
+err_update:
+	__mlx5e_tc_meter_put(meter);
+err_get:
+	mutex_unlock(&flow_meters->sync_lock);
+	return ERR_PTR(err);
+}
+
+enum mlx5_flow_namespace_type
+mlx5e_tc_meter_get_namespace(struct mlx5e_flow_meters *flow_meters)
+{
+	return flow_meters->ns_type;
+}
+
+struct mlx5e_flow_meters *
+mlx5e_flow_meters_init(struct mlx5e_priv *priv,
+		       enum mlx5_flow_namespace_type ns_type,
+		       struct mlx5e_post_act *post_act)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+	struct mlx5e_flow_meters *flow_meters;
+	int err;
+
+	if (!(MLX5_CAP_GEN_64(mdev, general_obj_types) &
+	      MLX5_HCA_CAP_GENERAL_OBJECT_TYPES_FLOW_METER_ASO))
+		return ERR_PTR(-EOPNOTSUPP);
+
+	if (IS_ERR_OR_NULL(post_act)) {
+		netdev_dbg(priv->netdev,
+			   "flow meter offload is not supported, post action is missing\n");
+		return ERR_PTR(-EOPNOTSUPP);
+	}
+
+	flow_meters = kzalloc(sizeof(*flow_meters), GFP_KERNEL);
+	if (!flow_meters)
+		return ERR_PTR(-ENOMEM);
+
+	err = mlx5_core_alloc_pd(mdev, &flow_meters->pdn);
+	if (err) {
+		mlx5_core_err(mdev, "Failed to alloc pd for flow meter aso, err=%d\n", err);
+		goto err_out;
+	}
+
+	flow_meters->aso = mlx5_aso_create(mdev, flow_meters->pdn);
+	if (IS_ERR(flow_meters->aso)) {
+		mlx5_core_warn(mdev, "Failed to create aso wqe for flow meter\n");
+		err = PTR_ERR(flow_meters->aso);
+		goto err_sq;
+	}
+
+	mutex_init(&flow_meters->sync_lock);
+	INIT_LIST_HEAD(&flow_meters->partial_list);
+	INIT_LIST_HEAD(&flow_meters->full_list);
+
+	flow_meters->ns_type = ns_type;
+	flow_meters->mdev = mdev;
+	flow_meters->post_act = post_act;
+	mutex_init(&flow_meters->aso_lock);
+	flow_meters->log_granularity = min_t(int, 6,
+					     MLX5_CAP_QOS(mdev, log_meter_aso_max_alloc));
+
+	return flow_meters;
+
+err_sq:
+	mlx5_core_dealloc_pd(mdev, flow_meters->pdn);
+err_out:
+	kfree(flow_meters);
+	return ERR_PTR(err);
+}
+
+void
+mlx5e_flow_meters_cleanup(struct mlx5e_flow_meters *flow_meters)
+{
+	if (IS_ERR_OR_NULL(flow_meters))
+		return;
+
+	mlx5_aso_destroy(flow_meters->aso);
+	mlx5_core_dealloc_pd(flow_meters->mdev, flow_meters->pdn);
+	kfree(flow_meters);
+}
+
+void
+mlx5e_tc_meter_get_stats(struct mlx5e_flow_meter_handle *meter,
+			 u64 *bytes, u64 *packets, u64 *drops, u64 *lastuse)
+{
+	u64 bytes1, packets1, lastuse1;
+	u64 bytes2, packets2, lastuse2;
+
+	mlx5_fc_query_cached(meter->green_counter, &bytes1, &packets1, &lastuse1);
+	mlx5_fc_query_cached(meter->red_counter, &bytes2, &packets2, &lastuse2);
+
+	*bytes = bytes1 + bytes2;
+	*packets = packets1 + packets2;
+	*drops = packets2;
+	*lastuse = max_t(u64, lastuse1, lastuse2);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc/meter.h b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/meter.h
new file mode 100644
index 000000000..6de6e8a16
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/meter.h
@@ -0,0 +1,74 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. All rights reserved. */
+
+#ifndef __MLX5_EN_FLOW_METER_H__
+#define __MLX5_EN_FLOW_METER_H__
+
+struct mlx5e_post_meter_priv;
+struct mlx5e_flow_meter_aso_obj;
+struct mlx5e_flow_meters;
+struct mlx5_flow_attr;
+
+enum mlx5e_flow_meter_mode {
+	MLX5_RATE_LIMIT_BPS,
+	MLX5_RATE_LIMIT_PPS,
+};
+
+struct mlx5e_flow_meter_params {
+	enum mlx5e_flow_meter_mode mode;
+	 /* police action index */
+	u32 index;
+	u64 rate;
+	u64 burst;
+};
+
+struct mlx5e_flow_meter_handle {
+	struct mlx5e_flow_meters *flow_meters;
+	struct mlx5e_flow_meter_aso_obj *meters_obj;
+	u32 obj_id;
+	u8 idx;
+
+	int refcnt;
+	struct hlist_node hlist;
+	struct mlx5e_flow_meter_params params;
+
+	struct mlx5_fc *green_counter;
+	struct mlx5_fc *red_counter;
+};
+
+struct mlx5e_meter_attr {
+	struct mlx5e_flow_meter_params params;
+	struct mlx5e_flow_meter_handle *meter;
+	struct mlx5e_post_meter_priv *post_meter;
+};
+
+int
+mlx5e_tc_meter_modify(struct mlx5_core_dev *mdev,
+		      struct mlx5e_flow_meter_handle *meter,
+		      struct mlx5e_flow_meter_params *meter_params);
+
+struct mlx5e_flow_meter_handle *
+mlx5e_tc_meter_get(struct mlx5_core_dev *mdev, struct mlx5e_flow_meter_params *params);
+void
+mlx5e_tc_meter_put(struct mlx5e_flow_meter_handle *meter);
+int
+mlx5e_tc_meter_update(struct mlx5e_flow_meter_handle *meter,
+		      struct mlx5e_flow_meter_params *params);
+struct mlx5e_flow_meter_handle *
+mlx5e_tc_meter_replace(struct mlx5_core_dev *mdev, struct mlx5e_flow_meter_params *params);
+
+enum mlx5_flow_namespace_type
+mlx5e_tc_meter_get_namespace(struct mlx5e_flow_meters *flow_meters);
+
+struct mlx5e_flow_meters *
+mlx5e_flow_meters_init(struct mlx5e_priv *priv,
+		       enum mlx5_flow_namespace_type ns_type,
+		       struct mlx5e_post_act *post_action);
+void
+mlx5e_flow_meters_cleanup(struct mlx5e_flow_meters *flow_meters);
+
+void
+mlx5e_tc_meter_get_stats(struct mlx5e_flow_meter_handle *meter,
+			 u64 *bytes, u64 *packets, u64 *drops, u64 *lastuse);
+
+#endif /* __MLX5_EN_FLOW_METER_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc/post_act.c b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/post_act.c
new file mode 100644
index 000000000..0290e0dea
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/post_act.c
@@ -0,0 +1,177 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+// Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+
+#include "en/tc_priv.h"
+#include "en_tc.h"
+#include "post_act.h"
+#include "mlx5_core.h"
+#include "fs_core.h"
+
+struct mlx5e_post_act {
+	enum mlx5_flow_namespace_type ns_type;
+	struct mlx5_fs_chains *chains;
+	struct mlx5_flow_table *ft;
+	struct mlx5e_priv *priv;
+	struct xarray ids;
+};
+
+struct mlx5e_post_act_handle {
+	enum mlx5_flow_namespace_type ns_type;
+	struct mlx5_flow_attr *attr;
+	struct mlx5_flow_handle *rule;
+	u32 id;
+};
+
+#define MLX5_POST_ACTION_BITS MLX5_REG_MAPPING_MBITS(FTEID_TO_REG)
+#define MLX5_POST_ACTION_MASK MLX5_REG_MAPPING_MASK(FTEID_TO_REG)
+#define MLX5_POST_ACTION_MAX MLX5_POST_ACTION_MASK
+
+struct mlx5e_post_act *
+mlx5e_tc_post_act_init(struct mlx5e_priv *priv, struct mlx5_fs_chains *chains,
+		       enum mlx5_flow_namespace_type ns_type)
+{
+	enum fs_flow_table_type table_type = ns_type == MLX5_FLOW_NAMESPACE_FDB ?
+					     FS_FT_FDB : FS_FT_NIC_RX;
+	struct mlx5e_post_act *post_act;
+	int err;
+
+	if (!MLX5_CAP_FLOWTABLE_TYPE(priv->mdev, ignore_flow_level, table_type)) {
+		if (priv->mdev->coredev_type == MLX5_COREDEV_PF)
+			mlx5_core_warn(priv->mdev, "firmware level support is missing\n");
+		err = -EOPNOTSUPP;
+		goto err_check;
+	}
+
+	post_act = kzalloc(sizeof(*post_act), GFP_KERNEL);
+	if (!post_act) {
+		err = -ENOMEM;
+		goto err_check;
+	}
+	post_act->ft = mlx5_chains_create_global_table(chains);
+	if (IS_ERR(post_act->ft)) {
+		err = PTR_ERR(post_act->ft);
+		mlx5_core_warn(priv->mdev, "failed to create post action table, err: %d\n", err);
+		goto err_ft;
+	}
+	post_act->chains = chains;
+	post_act->ns_type = ns_type;
+	post_act->priv = priv;
+	xa_init_flags(&post_act->ids, XA_FLAGS_ALLOC1);
+	return post_act;
+
+err_ft:
+	kfree(post_act);
+err_check:
+	return ERR_PTR(err);
+}
+
+void
+mlx5e_tc_post_act_destroy(struct mlx5e_post_act *post_act)
+{
+	if (IS_ERR_OR_NULL(post_act))
+		return;
+
+	xa_destroy(&post_act->ids);
+	mlx5_chains_destroy_global_table(post_act->chains, post_act->ft);
+	kfree(post_act);
+}
+
+int
+mlx5e_tc_post_act_offload(struct mlx5e_post_act *post_act,
+			  struct mlx5e_post_act_handle *handle)
+{
+	struct mlx5_flow_spec *spec;
+	int err;
+
+	spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
+	if (!spec)
+		return -ENOMEM;
+
+	/* Post action rule matches on fte_id and executes original rule's tc rule action */
+	mlx5e_tc_match_to_reg_match(spec, FTEID_TO_REG, handle->id, MLX5_POST_ACTION_MASK);
+
+	handle->rule = mlx5e_tc_rule_offload(post_act->priv, spec, handle->attr);
+	if (IS_ERR(handle->rule)) {
+		err = PTR_ERR(handle->rule);
+		netdev_warn(post_act->priv->netdev, "Failed to add post action rule");
+		goto err_rule;
+	}
+
+	kvfree(spec);
+	return 0;
+
+err_rule:
+	kvfree(spec);
+	return err;
+}
+
+struct mlx5e_post_act_handle *
+mlx5e_tc_post_act_add(struct mlx5e_post_act *post_act, struct mlx5_flow_attr *post_attr)
+{
+	struct mlx5e_post_act_handle *handle;
+	int err;
+
+	handle = kzalloc(sizeof(*handle), GFP_KERNEL);
+	if (!handle) {
+		kfree(handle);
+		return ERR_PTR(-ENOMEM);
+	}
+
+	post_attr->chain = 0;
+	post_attr->prio = 0;
+	post_attr->ft = post_act->ft;
+	post_attr->inner_match_level = MLX5_MATCH_NONE;
+	post_attr->outer_match_level = MLX5_MATCH_NONE;
+	post_attr->action &= ~MLX5_FLOW_CONTEXT_ACTION_DECAP;
+	post_attr->flags |= MLX5_ATTR_FLAG_NO_IN_PORT;
+
+	handle->ns_type = post_act->ns_type;
+	/* Splits were handled before post action */
+	if (handle->ns_type == MLX5_FLOW_NAMESPACE_FDB)
+		post_attr->esw_attr->split_count = 0;
+
+	err = xa_alloc(&post_act->ids, &handle->id, post_attr,
+		       XA_LIMIT(1, MLX5_POST_ACTION_MAX), GFP_KERNEL);
+	if (err)
+		goto err_xarray;
+
+	handle->attr = post_attr;
+
+	return handle;
+
+err_xarray:
+	kfree(handle);
+	return ERR_PTR(err);
+}
+
+void
+mlx5e_tc_post_act_unoffload(struct mlx5e_post_act *post_act,
+			    struct mlx5e_post_act_handle *handle)
+{
+	mlx5e_tc_rule_unoffload(post_act->priv, handle->rule, handle->attr);
+	handle->rule = NULL;
+}
+
+void
+mlx5e_tc_post_act_del(struct mlx5e_post_act *post_act, struct mlx5e_post_act_handle *handle)
+{
+	if (!IS_ERR_OR_NULL(handle->rule))
+		mlx5e_tc_post_act_unoffload(post_act, handle);
+	xa_erase(&post_act->ids, handle->id);
+	kfree(handle);
+}
+
+struct mlx5_flow_table *
+mlx5e_tc_post_act_get_ft(struct mlx5e_post_act *post_act)
+{
+	return post_act->ft;
+}
+
+/* Allocate a header modify action to write the post action handle fte id to a register. */
+int
+mlx5e_tc_post_act_set_handle(struct mlx5_core_dev *dev,
+			     struct mlx5e_post_act_handle *handle,
+			     struct mlx5e_tc_mod_hdr_acts *acts)
+{
+	return mlx5e_tc_match_to_reg_set(dev, acts, handle->ns_type, FTEID_TO_REG, handle->id);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc/post_act.h b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/post_act.h
new file mode 100644
index 000000000..40b8df184
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/post_act.h
@@ -0,0 +1,43 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. All rights reserved. */
+
+#ifndef __MLX5_POST_ACTION_H__
+#define __MLX5_POST_ACTION_H__
+
+#include "en.h"
+#include "lib/fs_chains.h"
+
+struct mlx5_flow_attr;
+struct mlx5e_priv;
+struct mlx5e_tc_mod_hdr_acts;
+
+struct mlx5e_post_act *
+mlx5e_tc_post_act_init(struct mlx5e_priv *priv, struct mlx5_fs_chains *chains,
+		       enum mlx5_flow_namespace_type ns_type);
+
+void
+mlx5e_tc_post_act_destroy(struct mlx5e_post_act *post_act);
+
+struct mlx5e_post_act_handle *
+mlx5e_tc_post_act_add(struct mlx5e_post_act *post_act, struct mlx5_flow_attr *post_attr);
+
+void
+mlx5e_tc_post_act_del(struct mlx5e_post_act *post_act, struct mlx5e_post_act_handle *handle);
+
+int
+mlx5e_tc_post_act_offload(struct mlx5e_post_act *post_act,
+			  struct mlx5e_post_act_handle *handle);
+
+void
+mlx5e_tc_post_act_unoffload(struct mlx5e_post_act *post_act,
+			    struct mlx5e_post_act_handle *handle);
+
+struct mlx5_flow_table *
+mlx5e_tc_post_act_get_ft(struct mlx5e_post_act *post_act);
+
+int
+mlx5e_tc_post_act_set_handle(struct mlx5_core_dev *dev,
+			     struct mlx5e_post_act_handle *handle,
+			     struct mlx5e_tc_mod_hdr_acts *acts);
+
+#endif /* __MLX5_POST_ACTION_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc/post_meter.c b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/post_meter.c
new file mode 100644
index 000000000..8b77e8228
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/post_meter.c
@@ -0,0 +1,209 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+// Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+
+#include "en/tc_priv.h"
+#include "post_meter.h"
+#include "en/tc/post_act.h"
+
+#define MLX5_PACKET_COLOR_BITS MLX5_REG_MAPPING_MBITS(PACKET_COLOR_TO_REG)
+#define MLX5_PACKET_COLOR_MASK MLX5_REG_MAPPING_MASK(PACKET_COLOR_TO_REG)
+
+struct mlx5e_post_meter_priv {
+	struct mlx5_flow_table *ft;
+	struct mlx5_flow_group *fg;
+	struct mlx5_flow_handle *fwd_green_rule;
+	struct mlx5_flow_handle *drop_red_rule;
+};
+
+struct mlx5_flow_table *
+mlx5e_post_meter_get_ft(struct mlx5e_post_meter_priv *post_meter)
+{
+	return post_meter->ft;
+}
+
+static int
+mlx5e_post_meter_table_create(struct mlx5e_priv *priv,
+			      enum mlx5_flow_namespace_type ns_type,
+			      struct mlx5e_post_meter_priv *post_meter)
+{
+	struct mlx5_flow_table_attr ft_attr = {};
+	struct mlx5_flow_namespace *root_ns;
+
+	root_ns = mlx5_get_flow_namespace(priv->mdev, ns_type);
+	if (!root_ns) {
+		mlx5_core_warn(priv->mdev, "Failed to get namespace for flow meter\n");
+		return -EOPNOTSUPP;
+	}
+
+	ft_attr.flags = MLX5_FLOW_TABLE_UNMANAGED;
+	ft_attr.prio = FDB_SLOW_PATH;
+	ft_attr.max_fte = 2;
+	ft_attr.level = 1;
+
+	post_meter->ft = mlx5_create_flow_table(root_ns, &ft_attr);
+	if (IS_ERR(post_meter->ft)) {
+		mlx5_core_warn(priv->mdev, "Failed to create post_meter table\n");
+		return PTR_ERR(post_meter->ft);
+	}
+
+	return 0;
+}
+
+static int
+mlx5e_post_meter_fg_create(struct mlx5e_priv *priv,
+			   struct mlx5e_post_meter_priv *post_meter)
+{
+	int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
+	void *misc2, *match_criteria;
+	u32 *flow_group_in;
+	int err = 0;
+
+	flow_group_in = kvzalloc(inlen, GFP_KERNEL);
+	if (!flow_group_in)
+		return -ENOMEM;
+
+	MLX5_SET(create_flow_group_in, flow_group_in, match_criteria_enable,
+		 MLX5_MATCH_MISC_PARAMETERS_2);
+	match_criteria = MLX5_ADDR_OF(create_flow_group_in, flow_group_in,
+				      match_criteria);
+	misc2 = MLX5_ADDR_OF(fte_match_param, match_criteria, misc_parameters_2);
+	MLX5_SET(fte_match_set_misc2, misc2, metadata_reg_c_5, MLX5_PACKET_COLOR_MASK);
+	MLX5_SET(create_flow_group_in, flow_group_in, start_flow_index, 0);
+	MLX5_SET(create_flow_group_in, flow_group_in, end_flow_index, 1);
+
+	post_meter->fg = mlx5_create_flow_group(post_meter->ft, flow_group_in);
+	if (IS_ERR(post_meter->fg)) {
+		mlx5_core_warn(priv->mdev, "Failed to create post_meter flow group\n");
+		err = PTR_ERR(post_meter->fg);
+	}
+
+	kvfree(flow_group_in);
+	return err;
+}
+
+static int
+mlx5e_post_meter_rules_create(struct mlx5e_priv *priv,
+			      struct mlx5e_post_meter_priv *post_meter,
+			      struct mlx5e_post_act *post_act,
+			      struct mlx5_fc *green_counter,
+			      struct mlx5_fc *red_counter)
+{
+	struct mlx5_flow_destination dest[2] = {};
+	struct mlx5_flow_act flow_act = {};
+	struct mlx5_flow_handle *rule;
+	struct mlx5_flow_spec *spec;
+	int err;
+
+	spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
+	if (!spec)
+		return -ENOMEM;
+
+	mlx5e_tc_match_to_reg_match(spec, PACKET_COLOR_TO_REG,
+				    MLX5_FLOW_METER_COLOR_RED, MLX5_PACKET_COLOR_MASK);
+	flow_act.action = MLX5_FLOW_CONTEXT_ACTION_DROP |
+			  MLX5_FLOW_CONTEXT_ACTION_COUNT;
+	flow_act.flags |= FLOW_ACT_IGNORE_FLOW_LEVEL;
+	dest[0].type = MLX5_FLOW_DESTINATION_TYPE_COUNTER;
+	dest[0].counter_id = mlx5_fc_id(red_counter);
+
+	rule = mlx5_add_flow_rules(post_meter->ft, spec, &flow_act, dest, 1);
+	if (IS_ERR(rule)) {
+		mlx5_core_warn(priv->mdev, "Failed to create post_meter flow drop rule\n");
+		err = PTR_ERR(rule);
+		goto err_red;
+	}
+	post_meter->drop_red_rule = rule;
+
+	mlx5e_tc_match_to_reg_match(spec, PACKET_COLOR_TO_REG,
+				    MLX5_FLOW_METER_COLOR_GREEN, MLX5_PACKET_COLOR_MASK);
+	flow_act.action = MLX5_FLOW_CONTEXT_ACTION_FWD_DEST |
+			  MLX5_FLOW_CONTEXT_ACTION_COUNT;
+	dest[0].type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
+	dest[0].ft = mlx5e_tc_post_act_get_ft(post_act);
+	dest[1].type = MLX5_FLOW_DESTINATION_TYPE_COUNTER;
+	dest[1].counter_id = mlx5_fc_id(green_counter);
+
+	rule = mlx5_add_flow_rules(post_meter->ft, spec, &flow_act, dest, 2);
+	if (IS_ERR(rule)) {
+		mlx5_core_warn(priv->mdev, "Failed to create post_meter flow fwd rule\n");
+		err = PTR_ERR(rule);
+		goto err_green;
+	}
+	post_meter->fwd_green_rule = rule;
+
+	kvfree(spec);
+	return 0;
+
+err_green:
+	mlx5_del_flow_rules(post_meter->drop_red_rule);
+err_red:
+	kvfree(spec);
+	return err;
+}
+
+static void
+mlx5e_post_meter_rules_destroy(struct mlx5e_post_meter_priv *post_meter)
+{
+	mlx5_del_flow_rules(post_meter->drop_red_rule);
+	mlx5_del_flow_rules(post_meter->fwd_green_rule);
+}
+
+static void
+mlx5e_post_meter_fg_destroy(struct mlx5e_post_meter_priv *post_meter)
+{
+	mlx5_destroy_flow_group(post_meter->fg);
+}
+
+static void
+mlx5e_post_meter_table_destroy(struct mlx5e_post_meter_priv *post_meter)
+{
+	mlx5_destroy_flow_table(post_meter->ft);
+}
+
+struct mlx5e_post_meter_priv *
+mlx5e_post_meter_init(struct mlx5e_priv *priv,
+		      enum mlx5_flow_namespace_type ns_type,
+		      struct mlx5e_post_act *post_act,
+		      struct mlx5_fc *green_counter,
+		      struct mlx5_fc *red_counter)
+{
+	struct mlx5e_post_meter_priv *post_meter;
+	int err;
+
+	post_meter = kzalloc(sizeof(*post_meter), GFP_KERNEL);
+	if (!post_meter)
+		return ERR_PTR(-ENOMEM);
+
+	err = mlx5e_post_meter_table_create(priv, ns_type, post_meter);
+	if (err)
+		goto err_ft;
+
+	err = mlx5e_post_meter_fg_create(priv, post_meter);
+	if (err)
+		goto err_fg;
+
+	err = mlx5e_post_meter_rules_create(priv, post_meter, post_act, green_counter,
+					    red_counter);
+	if (err)
+		goto err_rules;
+
+	return post_meter;
+
+err_rules:
+	mlx5e_post_meter_fg_destroy(post_meter);
+err_fg:
+	mlx5e_post_meter_table_destroy(post_meter);
+err_ft:
+	kfree(post_meter);
+	return ERR_PTR(err);
+}
+
+void
+mlx5e_post_meter_cleanup(struct mlx5e_post_meter_priv *post_meter)
+{
+	mlx5e_post_meter_rules_destroy(post_meter);
+	mlx5e_post_meter_fg_destroy(post_meter);
+	mlx5e_post_meter_table_destroy(post_meter);
+	kfree(post_meter);
+}
+
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc/post_meter.h b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/post_meter.h
new file mode 100644
index 000000000..34d0e4b9f
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/post_meter.h
@@ -0,0 +1,29 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. All rights reserved. */
+
+#ifndef __MLX5_EN_POST_METER_H__
+#define __MLX5_EN_POST_METER_H__
+
+#define packet_color_to_reg { \
+	.mfield = MLX5_ACTION_IN_FIELD_METADATA_REG_C_5, \
+	.moffset = 0, \
+	.mlen = 8, \
+	.soffset = MLX5_BYTE_OFF(fte_match_param, \
+				 misc_parameters_2.metadata_reg_c_5), \
+}
+
+struct mlx5e_post_meter_priv;
+
+struct mlx5_flow_table *
+mlx5e_post_meter_get_ft(struct mlx5e_post_meter_priv *post_meter);
+
+struct mlx5e_post_meter_priv *
+mlx5e_post_meter_init(struct mlx5e_priv *priv,
+		      enum mlx5_flow_namespace_type ns_type,
+		      struct mlx5e_post_act *post_act,
+		      struct mlx5_fc *green_counter,
+		      struct mlx5_fc *red_counter);
+void
+mlx5e_post_meter_cleanup(struct mlx5e_post_meter_priv *post_meter);
+
+#endif /* __MLX5_EN_POST_METER_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc/sample.c b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/sample.c
new file mode 100644
index 000000000..c57b09727
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/sample.c
@@ -0,0 +1,655 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2021 Mellanox Technologies. */
+
+#include <linux/skbuff.h>
+#include <net/psample.h>
+#include "en/mapping.h"
+#include "en/tc/post_act.h"
+#include "en/tc/act/sample.h"
+#include "en/mod_hdr.h"
+#include "sample.h"
+#include "eswitch.h"
+#include "en_tc.h"
+#include "fs_core.h"
+
+#define MLX5_ESW_VPORT_TBL_SIZE_SAMPLE (64 * 1024)
+
+static struct esw_vport_tbl_namespace mlx5_esw_vport_tbl_sample_ns = {
+	.max_fte = MLX5_ESW_VPORT_TBL_SIZE_SAMPLE,
+	.max_num_groups = 0,    /* default num of groups */
+	.flags = 0,
+};
+
+struct mlx5e_tc_psample {
+	struct mlx5_eswitch *esw;
+	struct mlx5_flow_table *termtbl;
+	struct mlx5_flow_handle *termtbl_rule;
+	DECLARE_HASHTABLE(hashtbl, 8);
+	struct mutex ht_lock; /* protect hashtbl */
+	DECLARE_HASHTABLE(restore_hashtbl, 8);
+	struct mutex restore_lock; /* protect restore_hashtbl */
+	struct mlx5e_post_act *post_act;
+};
+
+struct mlx5e_sampler {
+	struct hlist_node hlist;
+	u32 sampler_id;
+	u32 sample_ratio;
+	u32 sample_table_id;
+	u32 default_table_id;
+	int count;
+};
+
+struct mlx5e_sample_flow {
+	struct mlx5e_sampler *sampler;
+	struct mlx5e_sample_restore *restore;
+	struct mlx5_flow_attr *pre_attr;
+	struct mlx5_flow_handle *pre_rule;
+	struct mlx5_flow_attr *post_attr;
+	struct mlx5_flow_handle *post_rule;
+};
+
+struct mlx5e_sample_restore {
+	struct hlist_node hlist;
+	struct mlx5_modify_hdr *modify_hdr;
+	struct mlx5_flow_handle *rule;
+	u32 obj_id;
+	int count;
+};
+
+static int
+sampler_termtbl_create(struct mlx5e_tc_psample *tc_psample)
+{
+	struct mlx5_eswitch *esw = tc_psample->esw;
+	struct mlx5_flow_table_attr ft_attr = {};
+	struct mlx5_flow_destination dest = {};
+	struct mlx5_core_dev *dev = esw->dev;
+	struct mlx5_flow_namespace *root_ns;
+	struct mlx5_flow_act act = {};
+	int err;
+
+	if (!MLX5_CAP_ESW_FLOWTABLE_FDB(dev, termination_table))  {
+		mlx5_core_warn(dev, "termination table is not supported\n");
+		return -EOPNOTSUPP;
+	}
+
+	root_ns = mlx5_get_flow_namespace(dev, MLX5_FLOW_NAMESPACE_FDB);
+	if (!root_ns) {
+		mlx5_core_warn(dev, "failed to get FDB flow namespace\n");
+		return -EOPNOTSUPP;
+	}
+
+	ft_attr.flags = MLX5_FLOW_TABLE_TERMINATION | MLX5_FLOW_TABLE_UNMANAGED;
+	ft_attr.autogroup.max_num_groups = 1;
+	ft_attr.prio = FDB_SLOW_PATH;
+	ft_attr.max_fte = 1;
+	ft_attr.level = 1;
+	tc_psample->termtbl = mlx5_create_auto_grouped_flow_table(root_ns, &ft_attr);
+	if (IS_ERR(tc_psample->termtbl)) {
+		err = PTR_ERR(tc_psample->termtbl);
+		mlx5_core_warn(dev, "failed to create termtbl, err: %d\n", err);
+		return err;
+	}
+
+	act.action = MLX5_FLOW_CONTEXT_ACTION_FWD_DEST;
+	dest.vport.num = esw->manager_vport;
+	dest.type = MLX5_FLOW_DESTINATION_TYPE_VPORT;
+	tc_psample->termtbl_rule = mlx5_add_flow_rules(tc_psample->termtbl, NULL, &act, &dest, 1);
+	if (IS_ERR(tc_psample->termtbl_rule)) {
+		err = PTR_ERR(tc_psample->termtbl_rule);
+		mlx5_core_warn(dev, "failed to create termtbl rule, err: %d\n", err);
+		mlx5_destroy_flow_table(tc_psample->termtbl);
+		return err;
+	}
+
+	return 0;
+}
+
+static void
+sampler_termtbl_destroy(struct mlx5e_tc_psample *tc_psample)
+{
+	mlx5_del_flow_rules(tc_psample->termtbl_rule);
+	mlx5_destroy_flow_table(tc_psample->termtbl);
+}
+
+static int
+sampler_obj_create(struct mlx5_core_dev *mdev, struct mlx5e_sampler *sampler)
+{
+	u32 in[MLX5_ST_SZ_DW(create_sampler_obj_in)] = {};
+	u32 out[MLX5_ST_SZ_DW(general_obj_out_cmd_hdr)];
+	u64 general_obj_types;
+	void *obj;
+	int err;
+
+	general_obj_types = MLX5_CAP_GEN_64(mdev, general_obj_types);
+	if (!(general_obj_types & MLX5_HCA_CAP_GENERAL_OBJECT_TYPES_SAMPLER))
+		return -EOPNOTSUPP;
+	if (!MLX5_CAP_ESW_FLOWTABLE_FDB(mdev, ignore_flow_level))
+		return -EOPNOTSUPP;
+
+	obj = MLX5_ADDR_OF(create_sampler_obj_in, in, sampler_object);
+	MLX5_SET(sampler_obj, obj, table_type, FS_FT_FDB);
+	MLX5_SET(sampler_obj, obj, ignore_flow_level, 1);
+	MLX5_SET(sampler_obj, obj, level, 1);
+	MLX5_SET(sampler_obj, obj, sample_ratio, sampler->sample_ratio);
+	MLX5_SET(sampler_obj, obj, sample_table_id, sampler->sample_table_id);
+	MLX5_SET(sampler_obj, obj, default_table_id, sampler->default_table_id);
+	MLX5_SET(general_obj_in_cmd_hdr, in, opcode, MLX5_CMD_OP_CREATE_GENERAL_OBJECT);
+	MLX5_SET(general_obj_in_cmd_hdr, in, obj_type, MLX5_GENERAL_OBJECT_TYPES_SAMPLER);
+
+	err = mlx5_cmd_exec(mdev, in, sizeof(in), out, sizeof(out));
+	if (!err)
+		sampler->sampler_id = MLX5_GET(general_obj_out_cmd_hdr, out, obj_id);
+
+	return err;
+}
+
+static void
+sampler_obj_destroy(struct mlx5_core_dev *mdev, u32 sampler_id)
+{
+	u32 in[MLX5_ST_SZ_DW(general_obj_in_cmd_hdr)] = {};
+	u32 out[MLX5_ST_SZ_DW(general_obj_out_cmd_hdr)];
+
+	MLX5_SET(general_obj_in_cmd_hdr, in, opcode, MLX5_CMD_OP_DESTROY_GENERAL_OBJECT);
+	MLX5_SET(general_obj_in_cmd_hdr, in, obj_type, MLX5_GENERAL_OBJECT_TYPES_SAMPLER);
+	MLX5_SET(general_obj_in_cmd_hdr, in, obj_id, sampler_id);
+
+	mlx5_cmd_exec(mdev, in, sizeof(in), out, sizeof(out));
+}
+
+static u32
+sampler_hash(u32 sample_ratio, u32 default_table_id)
+{
+	return jhash_2words(sample_ratio, default_table_id, 0);
+}
+
+static int
+sampler_cmp(u32 sample_ratio1, u32 default_table_id1, u32 sample_ratio2, u32 default_table_id2)
+{
+	return sample_ratio1 != sample_ratio2 || default_table_id1 != default_table_id2;
+}
+
+static struct mlx5e_sampler *
+sampler_get(struct mlx5e_tc_psample *tc_psample, u32 sample_ratio, u32 default_table_id)
+{
+	struct mlx5e_sampler *sampler;
+	u32 hash_key;
+	int err;
+
+	mutex_lock(&tc_psample->ht_lock);
+	hash_key = sampler_hash(sample_ratio, default_table_id);
+	hash_for_each_possible(tc_psample->hashtbl, sampler, hlist, hash_key)
+		if (!sampler_cmp(sampler->sample_ratio, sampler->default_table_id,
+				 sample_ratio, default_table_id))
+			goto add_ref;
+
+	sampler = kzalloc(sizeof(*sampler), GFP_KERNEL);
+	if (!sampler) {
+		err = -ENOMEM;
+		goto err_alloc;
+	}
+
+	sampler->sample_table_id = tc_psample->termtbl->id;
+	sampler->default_table_id = default_table_id;
+	sampler->sample_ratio = sample_ratio;
+
+	err = sampler_obj_create(tc_psample->esw->dev, sampler);
+	if (err)
+		goto err_create;
+
+	hash_add(tc_psample->hashtbl, &sampler->hlist, hash_key);
+
+add_ref:
+	sampler->count++;
+	mutex_unlock(&tc_psample->ht_lock);
+	return sampler;
+
+err_create:
+	kfree(sampler);
+err_alloc:
+	mutex_unlock(&tc_psample->ht_lock);
+	return ERR_PTR(err);
+}
+
+static void
+sampler_put(struct mlx5e_tc_psample *tc_psample, struct mlx5e_sampler *sampler)
+{
+	mutex_lock(&tc_psample->ht_lock);
+	if (--sampler->count == 0) {
+		hash_del(&sampler->hlist);
+		sampler_obj_destroy(tc_psample->esw->dev, sampler->sampler_id);
+		kfree(sampler);
+	}
+	mutex_unlock(&tc_psample->ht_lock);
+}
+
+/* obj_id is used to restore the sample parameters.
+ * Set fte_id in original flow table, then match it in the default table.
+ * Only set it for NICs can preserve reg_c or decap action. For other cases,
+ * use the same match in the default table.
+ * Use one header rewrite for both obj_id and fte_id.
+ */
+static struct mlx5_modify_hdr *
+sample_modify_hdr_get(struct mlx5_core_dev *mdev, u32 obj_id,
+		      struct mlx5e_tc_mod_hdr_acts *mod_acts)
+{
+	struct mlx5_modify_hdr *modify_hdr;
+	int err;
+
+	err = mlx5e_tc_match_to_reg_set(mdev, mod_acts, MLX5_FLOW_NAMESPACE_FDB,
+					CHAIN_TO_REG, obj_id);
+	if (err)
+		goto err_set_regc0;
+
+	modify_hdr = mlx5_modify_header_alloc(mdev, MLX5_FLOW_NAMESPACE_FDB,
+					      mod_acts->num_actions,
+					      mod_acts->actions);
+	if (IS_ERR(modify_hdr)) {
+		err = PTR_ERR(modify_hdr);
+		goto err_modify_hdr;
+	}
+
+	mlx5e_mod_hdr_dealloc(mod_acts);
+	return modify_hdr;
+
+err_modify_hdr:
+	mlx5e_mod_hdr_dealloc(mod_acts);
+err_set_regc0:
+	return ERR_PTR(err);
+}
+
+static struct mlx5e_sample_restore *
+sample_restore_get(struct mlx5e_tc_psample *tc_psample, u32 obj_id,
+		   struct mlx5e_tc_mod_hdr_acts *mod_acts)
+{
+	struct mlx5_eswitch *esw = tc_psample->esw;
+	struct mlx5_core_dev *mdev = esw->dev;
+	struct mlx5e_sample_restore *restore;
+	struct mlx5_modify_hdr *modify_hdr;
+	int err;
+
+	mutex_lock(&tc_psample->restore_lock);
+	hash_for_each_possible(tc_psample->restore_hashtbl, restore, hlist, obj_id)
+		if (restore->obj_id == obj_id)
+			goto add_ref;
+
+	restore = kzalloc(sizeof(*restore), GFP_KERNEL);
+	if (!restore) {
+		err = -ENOMEM;
+		goto err_alloc;
+	}
+	restore->obj_id = obj_id;
+
+	modify_hdr = sample_modify_hdr_get(mdev, obj_id, mod_acts);
+	if (IS_ERR(modify_hdr)) {
+		err = PTR_ERR(modify_hdr);
+		goto err_modify_hdr;
+	}
+	restore->modify_hdr = modify_hdr;
+
+	restore->rule = esw_add_restore_rule(esw, obj_id);
+	if (IS_ERR(restore->rule)) {
+		err = PTR_ERR(restore->rule);
+		goto err_restore;
+	}
+
+	hash_add(tc_psample->restore_hashtbl, &restore->hlist, obj_id);
+add_ref:
+	restore->count++;
+	mutex_unlock(&tc_psample->restore_lock);
+	return restore;
+
+err_restore:
+	mlx5_modify_header_dealloc(mdev, restore->modify_hdr);
+err_modify_hdr:
+	kfree(restore);
+err_alloc:
+	mutex_unlock(&tc_psample->restore_lock);
+	return ERR_PTR(err);
+}
+
+static void
+sample_restore_put(struct mlx5e_tc_psample *tc_psample, struct mlx5e_sample_restore *restore)
+{
+	mutex_lock(&tc_psample->restore_lock);
+	if (--restore->count == 0)
+		hash_del(&restore->hlist);
+	mutex_unlock(&tc_psample->restore_lock);
+
+	if (!restore->count) {
+		mlx5_del_flow_rules(restore->rule);
+		mlx5_modify_header_dealloc(tc_psample->esw->dev, restore->modify_hdr);
+		kfree(restore);
+	}
+}
+
+void mlx5e_tc_sample_skb(struct sk_buff *skb, struct mlx5_mapped_obj *mapped_obj)
+{
+	u32 trunc_size = mapped_obj->sample.trunc_size;
+	struct psample_group psample_group = {};
+	struct psample_metadata md = {};
+
+	md.trunc_size = trunc_size ? min(trunc_size, skb->len) : skb->len;
+	md.in_ifindex = skb->dev->ifindex;
+	psample_group.group_num = mapped_obj->sample.group_id;
+	psample_group.net = &init_net;
+	skb_push(skb, skb->mac_len);
+
+	psample_sample_packet(&psample_group, skb, mapped_obj->sample.rate, &md);
+}
+
+static int
+add_post_rule(struct mlx5_eswitch *esw, struct mlx5e_sample_flow *sample_flow,
+	      struct mlx5_flow_spec *spec, struct mlx5_flow_attr *attr,
+	      u32 *default_tbl_id)
+{
+	struct mlx5_esw_flow_attr *esw_attr = attr->esw_attr;
+	u32 attr_sz = ns_to_attr_sz(MLX5_FLOW_NAMESPACE_FDB);
+	struct mlx5_vport_tbl_attr per_vport_tbl_attr;
+	struct mlx5_flow_table *default_tbl;
+	struct mlx5_flow_attr *post_attr;
+	int err;
+
+	/* Allocate default table per vport, chain and prio. Otherwise, there is
+	 * only one default table for the same sampler object. Rules with different
+	 * prio and chain may overlap. For CT sample action, per vport default
+	 * table is needed to resotre the metadata.
+	 */
+	per_vport_tbl_attr.chain = attr->chain;
+	per_vport_tbl_attr.prio = attr->prio;
+	per_vport_tbl_attr.vport = esw_attr->in_rep->vport;
+	per_vport_tbl_attr.vport_ns = &mlx5_esw_vport_tbl_sample_ns;
+	default_tbl = mlx5_esw_vporttbl_get(esw, &per_vport_tbl_attr);
+	if (IS_ERR(default_tbl)) {
+		err = PTR_ERR(default_tbl);
+		goto err_default_tbl;
+	}
+	*default_tbl_id = default_tbl->id;
+
+	post_attr = mlx5_alloc_flow_attr(MLX5_FLOW_NAMESPACE_FDB);
+	if (!post_attr) {
+		err = -ENOMEM;
+		goto err_attr;
+	}
+	sample_flow->post_attr = post_attr;
+	memcpy(post_attr, attr, attr_sz);
+	/* Perform the original matches on the default table.
+	 * Offload all actions except the sample action.
+	 */
+	post_attr->chain = 0;
+	post_attr->prio = 0;
+	post_attr->ft = default_tbl;
+	post_attr->flags = MLX5_ATTR_FLAG_NO_IN_PORT;
+
+	/* When offloading sample and encap action, if there is no valid
+	 * neigh data struct, a slow path rule is offloaded first. Source
+	 * port metadata match is set at that time. A per vport table is
+	 * already allocated. No need to match it again. So clear the source
+	 * port metadata match.
+	 */
+	mlx5_eswitch_clear_rule_source_port(esw, spec);
+	sample_flow->post_rule = mlx5_eswitch_add_offloaded_rule(esw, spec, post_attr);
+	if (IS_ERR(sample_flow->post_rule)) {
+		err = PTR_ERR(sample_flow->post_rule);
+		goto err_rule;
+	}
+	return 0;
+
+err_rule:
+	kfree(post_attr);
+err_attr:
+	mlx5_esw_vporttbl_put(esw, &per_vport_tbl_attr);
+err_default_tbl:
+	return err;
+}
+
+static void
+del_post_rule(struct mlx5_eswitch *esw, struct mlx5e_sample_flow *sample_flow,
+	      struct mlx5_flow_attr *attr)
+{
+	struct mlx5_esw_flow_attr *esw_attr = attr->esw_attr;
+	struct mlx5_vport_tbl_attr tbl_attr;
+
+	mlx5_eswitch_del_offloaded_rule(esw, sample_flow->post_rule, sample_flow->post_attr);
+	kfree(sample_flow->post_attr);
+	tbl_attr.chain = attr->chain;
+	tbl_attr.prio = attr->prio;
+	tbl_attr.vport = esw_attr->in_rep->vport;
+	tbl_attr.vport_ns = &mlx5_esw_vport_tbl_sample_ns;
+	mlx5_esw_vporttbl_put(esw, &tbl_attr);
+}
+
+/* For the following typical flow table:
+ *
+ * +-------------------------------+
+ * +       original flow table     +
+ * +-------------------------------+
+ * +         original match        +
+ * +-------------------------------+
+ * + sample action + other actions +
+ * +-------------------------------+
+ *
+ * We translate the tc filter with sample action to the following HW model:
+ *
+ *         +---------------------+
+ *         + original flow table +
+ *         +---------------------+
+ *         +   original match    +
+ *         +---------------------+
+ *               | set fte_id (if reg_c preserve cap)
+ *               | do decap (if required)
+ *               v
+ * +------------------------------------------------+
+ * +                Flow Sampler Object             +
+ * +------------------------------------------------+
+ * +                    sample ratio                +
+ * +------------------------------------------------+
+ * +    sample table id    |    default table id    +
+ * +------------------------------------------------+
+ *            |                            |
+ *            v                            v
+ * +-----------------------------+  +-------------------+
+ * +        sample table         +  +   default table   +
+ * +-----------------------------+  +-------------------+
+ * + forward to management vport +             |
+ * +-----------------------------+             |
+ *                                     +-------+------+
+ *                                     |              |reg_c preserve cap
+ *                                     |              |or decap action
+ *                                     v              v
+ *                        +-----------------+   +-------------+
+ *                        + per vport table +   + post action +
+ *                        +-----------------+   +-------------+
+ *                        + original match  +
+ *                        +-----------------+
+ *                        + other actions   +
+ *                        +-----------------+
+ */
+struct mlx5_flow_handle *
+mlx5e_tc_sample_offload(struct mlx5e_tc_psample *tc_psample,
+			struct mlx5_flow_spec *spec,
+			struct mlx5_flow_attr *attr)
+{
+	struct mlx5_esw_flow_attr *esw_attr = attr->esw_attr;
+	struct mlx5_esw_flow_attr *pre_esw_attr;
+	struct mlx5_mapped_obj restore_obj = {};
+	struct mlx5e_tc_mod_hdr_acts *mod_acts;
+	struct mlx5e_sample_flow *sample_flow;
+	struct mlx5e_sample_attr *sample_attr;
+	struct mlx5_flow_attr *pre_attr;
+	struct mlx5_eswitch *esw;
+	u32 default_tbl_id;
+	u32 obj_id;
+	int err;
+
+	if (IS_ERR_OR_NULL(tc_psample))
+		return ERR_PTR(-EOPNOTSUPP);
+
+	sample_flow = kzalloc(sizeof(*sample_flow), GFP_KERNEL);
+	if (!sample_flow)
+		return ERR_PTR(-ENOMEM);
+	sample_attr = &attr->sample_attr;
+	sample_attr->sample_flow = sample_flow;
+
+	/* For NICs with reg_c_preserve support or decap action, use
+	 * post action instead of the per vport, chain and prio table.
+	 * Only match the fte id instead of the same match in the
+	 * original flow table.
+	 */
+	esw = tc_psample->esw;
+	if (mlx5e_tc_act_sample_is_multi_table(esw->dev, attr)) {
+		struct mlx5_flow_table *ft;
+
+		ft = mlx5e_tc_post_act_get_ft(tc_psample->post_act);
+		default_tbl_id = ft->id;
+	} else {
+		err = add_post_rule(esw, sample_flow, spec, attr, &default_tbl_id);
+		if (err)
+			goto err_post_rule;
+	}
+
+	/* Create sampler object. */
+	sample_flow->sampler = sampler_get(tc_psample, sample_attr->rate, default_tbl_id);
+	if (IS_ERR(sample_flow->sampler)) {
+		err = PTR_ERR(sample_flow->sampler);
+		goto err_sampler;
+	}
+	sample_attr->sampler_id = sample_flow->sampler->sampler_id;
+
+	/* Create an id mapping reg_c0 value to sample object. */
+	restore_obj.type = MLX5_MAPPED_OBJ_SAMPLE;
+	restore_obj.sample.group_id = sample_attr->group_num;
+	restore_obj.sample.rate = sample_attr->rate;
+	restore_obj.sample.trunc_size = sample_attr->trunc_size;
+	restore_obj.sample.tunnel_id = attr->tunnel_id;
+	err = mapping_add(esw->offloads.reg_c0_obj_pool, &restore_obj, &obj_id);
+	if (err)
+		goto err_obj_id;
+	sample_attr->restore_obj_id = obj_id;
+
+	/* Create sample restore context. */
+	mod_acts = &attr->parse_attr->mod_hdr_acts;
+	sample_flow->restore = sample_restore_get(tc_psample, obj_id, mod_acts);
+	if (IS_ERR(sample_flow->restore)) {
+		err = PTR_ERR(sample_flow->restore);
+		goto err_sample_restore;
+	}
+
+	/* Perform the original matches on the original table. Offload the
+	 * sample action. The destination is the sampler object.
+	 */
+	pre_attr = mlx5_alloc_flow_attr(MLX5_FLOW_NAMESPACE_FDB);
+	if (!pre_attr) {
+		err = -ENOMEM;
+		goto err_alloc_pre_flow_attr;
+	}
+	pre_attr->action = MLX5_FLOW_CONTEXT_ACTION_FWD_DEST | MLX5_FLOW_CONTEXT_ACTION_MOD_HDR;
+	/* For decap action, do decap in the original flow table instead of the
+	 * default flow table.
+	 */
+	if (attr->action & MLX5_FLOW_CONTEXT_ACTION_DECAP)
+		pre_attr->action |= MLX5_FLOW_CONTEXT_ACTION_DECAP;
+	pre_attr->modify_hdr = sample_flow->restore->modify_hdr;
+	pre_attr->flags = MLX5_ATTR_FLAG_SAMPLE;
+	pre_attr->inner_match_level = attr->inner_match_level;
+	pre_attr->outer_match_level = attr->outer_match_level;
+	pre_attr->chain = attr->chain;
+	pre_attr->prio = attr->prio;
+	pre_attr->ft = attr->ft;
+	pre_attr->sample_attr = *sample_attr;
+	pre_esw_attr = pre_attr->esw_attr;
+	pre_esw_attr->in_mdev = esw_attr->in_mdev;
+	pre_esw_attr->in_rep = esw_attr->in_rep;
+	sample_flow->pre_rule = mlx5_eswitch_add_offloaded_rule(esw, spec, pre_attr);
+	if (IS_ERR(sample_flow->pre_rule)) {
+		err = PTR_ERR(sample_flow->pre_rule);
+		goto err_pre_offload_rule;
+	}
+	sample_flow->pre_attr = pre_attr;
+
+	return sample_flow->pre_rule;
+
+err_pre_offload_rule:
+	kfree(pre_attr);
+err_alloc_pre_flow_attr:
+	sample_restore_put(tc_psample, sample_flow->restore);
+err_sample_restore:
+	mapping_remove(esw->offloads.reg_c0_obj_pool, obj_id);
+err_obj_id:
+	sampler_put(tc_psample, sample_flow->sampler);
+err_sampler:
+	if (sample_flow->post_rule)
+		del_post_rule(esw, sample_flow, attr);
+err_post_rule:
+	kfree(sample_flow);
+	return ERR_PTR(err);
+}
+
+void
+mlx5e_tc_sample_unoffload(struct mlx5e_tc_psample *tc_psample,
+			  struct mlx5_flow_handle *rule,
+			  struct mlx5_flow_attr *attr)
+{
+	struct mlx5e_sample_flow *sample_flow;
+	struct mlx5_eswitch *esw;
+
+	if (IS_ERR_OR_NULL(tc_psample))
+		return;
+
+	/* The following delete order can't be changed, otherwise,
+	 * will hit fw syndromes.
+	 */
+	esw = tc_psample->esw;
+	sample_flow = attr->sample_attr.sample_flow;
+	mlx5_eswitch_del_offloaded_rule(esw, sample_flow->pre_rule, sample_flow->pre_attr);
+
+	sample_restore_put(tc_psample, sample_flow->restore);
+	mapping_remove(esw->offloads.reg_c0_obj_pool, attr->sample_attr.restore_obj_id);
+	sampler_put(tc_psample, sample_flow->sampler);
+	if (sample_flow->post_rule)
+		del_post_rule(esw, sample_flow, attr);
+
+	kfree(sample_flow->pre_attr);
+	kfree(sample_flow);
+}
+
+struct mlx5e_tc_psample *
+mlx5e_tc_sample_init(struct mlx5_eswitch *esw, struct mlx5e_post_act *post_act)
+{
+	struct mlx5e_tc_psample *tc_psample;
+	int err;
+
+	tc_psample = kzalloc(sizeof(*tc_psample), GFP_KERNEL);
+	if (!tc_psample)
+		return ERR_PTR(-ENOMEM);
+	if (IS_ERR_OR_NULL(post_act)) {
+		err = PTR_ERR(post_act);
+		goto err_post_act;
+	}
+	tc_psample->post_act = post_act;
+	tc_psample->esw = esw;
+	err = sampler_termtbl_create(tc_psample);
+	if (err)
+		goto err_post_act;
+
+	mutex_init(&tc_psample->ht_lock);
+	mutex_init(&tc_psample->restore_lock);
+
+	return tc_psample;
+
+err_post_act:
+	kfree(tc_psample);
+	return ERR_PTR(err);
+}
+
+void
+mlx5e_tc_sample_cleanup(struct mlx5e_tc_psample *tc_psample)
+{
+	if (IS_ERR_OR_NULL(tc_psample))
+		return;
+
+	mutex_destroy(&tc_psample->restore_lock);
+	mutex_destroy(&tc_psample->ht_lock);
+	sampler_termtbl_destroy(tc_psample);
+	kfree(tc_psample);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc/sample.h b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/sample.h
new file mode 100644
index 000000000..a569367ea
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/sample.h
@@ -0,0 +1,66 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2021 Mellanox Technologies. */
+
+#ifndef __MLX5_EN_TC_SAMPLE_H__
+#define __MLX5_EN_TC_SAMPLE_H__
+
+#include "eswitch.h"
+
+struct mlx5_flow_attr;
+struct mlx5e_tc_psample;
+struct mlx5e_post_act;
+
+struct mlx5e_sample_attr {
+	u32 group_num;
+	u32 rate;
+	u32 trunc_size;
+	u32 restore_obj_id;
+	u32 sampler_id;
+	struct mlx5e_sample_flow *sample_flow;
+};
+
+#if IS_ENABLED(CONFIG_MLX5_TC_SAMPLE)
+
+void mlx5e_tc_sample_skb(struct sk_buff *skb, struct mlx5_mapped_obj *mapped_obj);
+
+struct mlx5_flow_handle *
+mlx5e_tc_sample_offload(struct mlx5e_tc_psample *sample_priv,
+			struct mlx5_flow_spec *spec,
+			struct mlx5_flow_attr *attr);
+
+void
+mlx5e_tc_sample_unoffload(struct mlx5e_tc_psample *sample_priv,
+			  struct mlx5_flow_handle *rule,
+			  struct mlx5_flow_attr *attr);
+
+struct mlx5e_tc_psample *
+mlx5e_tc_sample_init(struct mlx5_eswitch *esw, struct mlx5e_post_act *post_act);
+
+void
+mlx5e_tc_sample_cleanup(struct mlx5e_tc_psample *tc_psample);
+
+#else /* CONFIG_MLX5_TC_SAMPLE */
+
+static inline struct mlx5_flow_handle *
+mlx5e_tc_sample_offload(struct mlx5e_tc_psample *tc_psample,
+			struct mlx5_flow_spec *spec,
+			struct mlx5_flow_attr *attr)
+{ return ERR_PTR(-EOPNOTSUPP); }
+
+static inline void
+mlx5e_tc_sample_unoffload(struct mlx5e_tc_psample *tc_psample,
+			  struct mlx5_flow_handle *rule,
+			  struct mlx5_flow_attr *attr) {}
+
+static inline struct mlx5e_tc_psample *
+mlx5e_tc_sample_init(struct mlx5_eswitch *esw, struct mlx5e_post_act *post_act)
+{ return ERR_PTR(-EOPNOTSUPP); }
+
+static inline void
+mlx5e_tc_sample_cleanup(struct mlx5e_tc_psample *tc_psample) {}
+
+static inline void
+mlx5e_tc_sample_skb(struct sk_buff *skb, struct mlx5_mapped_obj *mapped_obj) {}
+
+#endif /* CONFIG_MLX5_TC_SAMPLE */
+#endif /* __MLX5_EN_TC_SAMPLE_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc_ct.c b/drivers/net/ethernet/mellanox/mlx5/core/en/tc_ct.c
new file mode 100644
index 000000000..f01f7dfdb
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc_ct.c
@@ -0,0 +1,2272 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2019 Mellanox Technologies. */
+
+#include <net/netfilter/nf_conntrack.h>
+#include <net/netfilter/nf_conntrack_core.h>
+#include <net/netfilter/nf_conntrack_zones.h>
+#include <net/netfilter/nf_conntrack_labels.h>
+#include <net/netfilter/nf_conntrack_helper.h>
+#include <net/netfilter/nf_conntrack_acct.h>
+#include <uapi/linux/tc_act/tc_pedit.h>
+#include <net/tc_act/tc_ct.h>
+#include <net/flow_offload.h>
+#include <net/netfilter/nf_flow_table.h>
+#include <linux/workqueue.h>
+#include <linux/refcount.h>
+#include <linux/xarray.h>
+#include <linux/if_macvlan.h>
+#include <linux/debugfs.h>
+
+#include "lib/fs_chains.h"
+#include "en/tc_ct.h"
+#include "en/tc/ct_fs.h"
+#include "en/tc_priv.h"
+#include "en/mod_hdr.h"
+#include "en/mapping.h"
+#include "en/tc/post_act.h"
+#include "en.h"
+#include "en_tc.h"
+#include "en_rep.h"
+#include "fs_core.h"
+
+#define MLX5_CT_STATE_ESTABLISHED_BIT BIT(1)
+#define MLX5_CT_STATE_TRK_BIT BIT(2)
+#define MLX5_CT_STATE_NAT_BIT BIT(3)
+#define MLX5_CT_STATE_REPLY_BIT BIT(4)
+#define MLX5_CT_STATE_RELATED_BIT BIT(5)
+#define MLX5_CT_STATE_INVALID_BIT BIT(6)
+
+#define MLX5_CT_LABELS_BITS MLX5_REG_MAPPING_MBITS(LABELS_TO_REG)
+#define MLX5_CT_LABELS_MASK MLX5_REG_MAPPING_MASK(LABELS_TO_REG)
+
+/* Statically allocate modify actions for
+ * ipv6 and port nat (5) + tuple fields (4) + nic mode zone restore (1) = 10.
+ * This will be increased dynamically if needed (for the ipv6 snat + dnat).
+ */
+#define MLX5_CT_MIN_MOD_ACTS 10
+
+#define ct_dbg(fmt, args...)\
+	netdev_dbg(ct_priv->netdev, "ct_debug: " fmt "\n", ##args)
+
+struct mlx5_tc_ct_debugfs {
+	struct {
+		atomic_t offloaded;
+		atomic_t rx_dropped;
+	} stats;
+
+	struct dentry *root;
+};
+
+struct mlx5_tc_ct_priv {
+	struct mlx5_core_dev *dev;
+	const struct net_device *netdev;
+	struct mod_hdr_tbl *mod_hdr_tbl;
+	struct xarray tuple_ids;
+	struct rhashtable zone_ht;
+	struct rhashtable ct_tuples_ht;
+	struct rhashtable ct_tuples_nat_ht;
+	struct mlx5_flow_table *ct;
+	struct mlx5_flow_table *ct_nat;
+	struct mlx5e_post_act *post_act;
+	struct mutex control_lock; /* guards parallel adds/dels */
+	struct mapping_ctx *zone_mapping;
+	struct mapping_ctx *labels_mapping;
+	enum mlx5_flow_namespace_type ns_type;
+	struct mlx5_fs_chains *chains;
+	struct mlx5_ct_fs *fs;
+	struct mlx5_ct_fs_ops *fs_ops;
+	spinlock_t ht_lock; /* protects ft entries */
+	struct workqueue_struct *wq;
+
+	struct mlx5_tc_ct_debugfs debugfs;
+};
+
+struct mlx5_ct_flow {
+	struct mlx5_flow_attr *pre_ct_attr;
+	struct mlx5_flow_handle *pre_ct_rule;
+	struct mlx5_ct_ft *ft;
+	u32 chain_mapping;
+};
+
+struct mlx5_ct_zone_rule {
+	struct mlx5_ct_fs_rule *rule;
+	struct mlx5e_mod_hdr_handle *mh;
+	struct mlx5_flow_attr *attr;
+	bool nat;
+};
+
+struct mlx5_tc_ct_pre {
+	struct mlx5_flow_table *ft;
+	struct mlx5_flow_group *flow_grp;
+	struct mlx5_flow_group *miss_grp;
+	struct mlx5_flow_handle *flow_rule;
+	struct mlx5_flow_handle *miss_rule;
+	struct mlx5_modify_hdr *modify_hdr;
+};
+
+struct mlx5_ct_ft {
+	struct rhash_head node;
+	u16 zone;
+	u32 zone_restore_id;
+	refcount_t refcount;
+	struct nf_flowtable *nf_ft;
+	struct mlx5_tc_ct_priv *ct_priv;
+	struct rhashtable ct_entries_ht;
+	struct mlx5_tc_ct_pre pre_ct;
+	struct mlx5_tc_ct_pre pre_ct_nat;
+};
+
+struct mlx5_ct_tuple {
+	u16 addr_type;
+	__be16 n_proto;
+	u8 ip_proto;
+	struct {
+		union {
+			__be32 src_v4;
+			struct in6_addr src_v6;
+		};
+		union {
+			__be32 dst_v4;
+			struct in6_addr dst_v6;
+		};
+	} ip;
+	struct {
+		__be16 src;
+		__be16 dst;
+	} port;
+
+	u16 zone;
+};
+
+struct mlx5_ct_counter {
+	struct mlx5_fc *counter;
+	refcount_t refcount;
+	bool is_shared;
+};
+
+enum {
+	MLX5_CT_ENTRY_FLAG_VALID,
+};
+
+struct mlx5_ct_entry {
+	struct rhash_head node;
+	struct rhash_head tuple_node;
+	struct rhash_head tuple_nat_node;
+	struct mlx5_ct_counter *counter;
+	unsigned long cookie;
+	unsigned long restore_cookie;
+	struct mlx5_ct_tuple tuple;
+	struct mlx5_ct_tuple tuple_nat;
+	struct mlx5_ct_zone_rule zone_rules[2];
+
+	struct mlx5_tc_ct_priv *ct_priv;
+	struct work_struct work;
+
+	refcount_t refcnt;
+	unsigned long flags;
+};
+
+static void
+mlx5_tc_ct_entry_destroy_mod_hdr(struct mlx5_tc_ct_priv *ct_priv,
+				 struct mlx5_flow_attr *attr,
+				 struct mlx5e_mod_hdr_handle *mh);
+
+static const struct rhashtable_params cts_ht_params = {
+	.head_offset = offsetof(struct mlx5_ct_entry, node),
+	.key_offset = offsetof(struct mlx5_ct_entry, cookie),
+	.key_len = sizeof(((struct mlx5_ct_entry *)0)->cookie),
+	.automatic_shrinking = true,
+	.min_size = 16 * 1024,
+};
+
+static const struct rhashtable_params zone_params = {
+	.head_offset = offsetof(struct mlx5_ct_ft, node),
+	.key_offset = offsetof(struct mlx5_ct_ft, zone),
+	.key_len = sizeof(((struct mlx5_ct_ft *)0)->zone),
+	.automatic_shrinking = true,
+};
+
+static const struct rhashtable_params tuples_ht_params = {
+	.head_offset = offsetof(struct mlx5_ct_entry, tuple_node),
+	.key_offset = offsetof(struct mlx5_ct_entry, tuple),
+	.key_len = sizeof(((struct mlx5_ct_entry *)0)->tuple),
+	.automatic_shrinking = true,
+	.min_size = 16 * 1024,
+};
+
+static const struct rhashtable_params tuples_nat_ht_params = {
+	.head_offset = offsetof(struct mlx5_ct_entry, tuple_nat_node),
+	.key_offset = offsetof(struct mlx5_ct_entry, tuple_nat),
+	.key_len = sizeof(((struct mlx5_ct_entry *)0)->tuple_nat),
+	.automatic_shrinking = true,
+	.min_size = 16 * 1024,
+};
+
+static bool
+mlx5_tc_ct_entry_has_nat(struct mlx5_ct_entry *entry)
+{
+	return !!(entry->tuple_nat_node.next);
+}
+
+static int
+mlx5_get_label_mapping(struct mlx5_tc_ct_priv *ct_priv,
+		       u32 *labels, u32 *id)
+{
+	if (!memchr_inv(labels, 0, sizeof(u32) * 4)) {
+		*id = 0;
+		return 0;
+	}
+
+	if (mapping_add(ct_priv->labels_mapping, labels, id))
+		return -EOPNOTSUPP;
+
+	return 0;
+}
+
+static void
+mlx5_put_label_mapping(struct mlx5_tc_ct_priv *ct_priv, u32 id)
+{
+	if (id)
+		mapping_remove(ct_priv->labels_mapping, id);
+}
+
+static int
+mlx5_tc_ct_rule_to_tuple(struct mlx5_ct_tuple *tuple, struct flow_rule *rule)
+{
+	struct flow_match_control control;
+	struct flow_match_basic basic;
+
+	flow_rule_match_basic(rule, &basic);
+	flow_rule_match_control(rule, &control);
+
+	tuple->n_proto = basic.key->n_proto;
+	tuple->ip_proto = basic.key->ip_proto;
+	tuple->addr_type = control.key->addr_type;
+
+	if (tuple->addr_type == FLOW_DISSECTOR_KEY_IPV4_ADDRS) {
+		struct flow_match_ipv4_addrs match;
+
+		flow_rule_match_ipv4_addrs(rule, &match);
+		tuple->ip.src_v4 = match.key->src;
+		tuple->ip.dst_v4 = match.key->dst;
+	} else if (tuple->addr_type == FLOW_DISSECTOR_KEY_IPV6_ADDRS) {
+		struct flow_match_ipv6_addrs match;
+
+		flow_rule_match_ipv6_addrs(rule, &match);
+		tuple->ip.src_v6 = match.key->src;
+		tuple->ip.dst_v6 = match.key->dst;
+	} else {
+		return -EOPNOTSUPP;
+	}
+
+	if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_PORTS)) {
+		struct flow_match_ports match;
+
+		flow_rule_match_ports(rule, &match);
+		switch (tuple->ip_proto) {
+		case IPPROTO_TCP:
+		case IPPROTO_UDP:
+			tuple->port.src = match.key->src;
+			tuple->port.dst = match.key->dst;
+			break;
+		default:
+			return -EOPNOTSUPP;
+		}
+	} else {
+		if (tuple->ip_proto != IPPROTO_GRE)
+			return -EOPNOTSUPP;
+	}
+
+	return 0;
+}
+
+static int
+mlx5_tc_ct_rule_to_tuple_nat(struct mlx5_ct_tuple *tuple,
+			     struct flow_rule *rule)
+{
+	struct flow_action *flow_action = &rule->action;
+	struct flow_action_entry *act;
+	u32 offset, val, ip6_offset;
+	int i;
+
+	flow_action_for_each(i, act, flow_action) {
+		if (act->id != FLOW_ACTION_MANGLE)
+			continue;
+
+		offset = act->mangle.offset;
+		val = act->mangle.val;
+		switch (act->mangle.htype) {
+		case FLOW_ACT_MANGLE_HDR_TYPE_IP4:
+			if (offset == offsetof(struct iphdr, saddr))
+				tuple->ip.src_v4 = cpu_to_be32(val);
+			else if (offset == offsetof(struct iphdr, daddr))
+				tuple->ip.dst_v4 = cpu_to_be32(val);
+			else
+				return -EOPNOTSUPP;
+			break;
+
+		case FLOW_ACT_MANGLE_HDR_TYPE_IP6:
+			ip6_offset = (offset - offsetof(struct ipv6hdr, saddr));
+			ip6_offset /= 4;
+			if (ip6_offset < 4)
+				tuple->ip.src_v6.s6_addr32[ip6_offset] = cpu_to_be32(val);
+			else if (ip6_offset < 8)
+				tuple->ip.dst_v6.s6_addr32[ip6_offset - 4] = cpu_to_be32(val);
+			else
+				return -EOPNOTSUPP;
+			break;
+
+		case FLOW_ACT_MANGLE_HDR_TYPE_TCP:
+			if (offset == offsetof(struct tcphdr, source))
+				tuple->port.src = cpu_to_be16(val);
+			else if (offset == offsetof(struct tcphdr, dest))
+				tuple->port.dst = cpu_to_be16(val);
+			else
+				return -EOPNOTSUPP;
+			break;
+
+		case FLOW_ACT_MANGLE_HDR_TYPE_UDP:
+			if (offset == offsetof(struct udphdr, source))
+				tuple->port.src = cpu_to_be16(val);
+			else if (offset == offsetof(struct udphdr, dest))
+				tuple->port.dst = cpu_to_be16(val);
+			else
+				return -EOPNOTSUPP;
+			break;
+
+		default:
+			return -EOPNOTSUPP;
+		}
+	}
+
+	return 0;
+}
+
+static int
+mlx5_tc_ct_get_flow_source_match(struct mlx5_tc_ct_priv *ct_priv,
+				 struct net_device *ndev)
+{
+	struct mlx5e_priv *other_priv = netdev_priv(ndev);
+	struct mlx5_core_dev *mdev = ct_priv->dev;
+	bool vf_rep, uplink_rep;
+
+	vf_rep = mlx5e_eswitch_vf_rep(ndev) && mlx5_same_hw_devs(mdev, other_priv->mdev);
+	uplink_rep = mlx5e_eswitch_uplink_rep(ndev) && mlx5_same_hw_devs(mdev, other_priv->mdev);
+
+	if (vf_rep)
+		return MLX5_FLOW_CONTEXT_FLOW_SOURCE_LOCAL_VPORT;
+	if (uplink_rep)
+		return MLX5_FLOW_CONTEXT_FLOW_SOURCE_UPLINK;
+	if (is_vlan_dev(ndev))
+		return mlx5_tc_ct_get_flow_source_match(ct_priv, vlan_dev_real_dev(ndev));
+	if (netif_is_macvlan(ndev))
+		return mlx5_tc_ct_get_flow_source_match(ct_priv, macvlan_dev_real_dev(ndev));
+	if (mlx5e_get_tc_tun(ndev) || netif_is_lag_master(ndev))
+		return MLX5_FLOW_CONTEXT_FLOW_SOURCE_UPLINK;
+
+	return MLX5_FLOW_CONTEXT_FLOW_SOURCE_ANY_VPORT;
+}
+
+static int
+mlx5_tc_ct_set_tuple_match(struct mlx5_tc_ct_priv *ct_priv,
+			   struct mlx5_flow_spec *spec,
+			   struct flow_rule *rule)
+{
+	void *headers_c = MLX5_ADDR_OF(fte_match_param, spec->match_criteria,
+				       outer_headers);
+	void *headers_v = MLX5_ADDR_OF(fte_match_param, spec->match_value,
+				       outer_headers);
+	u16 addr_type = 0;
+	u8 ip_proto = 0;
+
+	if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_BASIC)) {
+		struct flow_match_basic match;
+
+		flow_rule_match_basic(rule, &match);
+
+		mlx5e_tc_set_ethertype(ct_priv->dev, &match, true, headers_c, headers_v);
+		MLX5_SET(fte_match_set_lyr_2_4, headers_c, ip_protocol,
+			 match.mask->ip_proto);
+		MLX5_SET(fte_match_set_lyr_2_4, headers_v, ip_protocol,
+			 match.key->ip_proto);
+
+		ip_proto = match.key->ip_proto;
+	}
+
+	if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_CONTROL)) {
+		struct flow_match_control match;
+
+		flow_rule_match_control(rule, &match);
+		addr_type = match.key->addr_type;
+	}
+
+	if (addr_type == FLOW_DISSECTOR_KEY_IPV4_ADDRS) {
+		struct flow_match_ipv4_addrs match;
+
+		flow_rule_match_ipv4_addrs(rule, &match);
+		memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_c,
+				    src_ipv4_src_ipv6.ipv4_layout.ipv4),
+		       &match.mask->src, sizeof(match.mask->src));
+		memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_v,
+				    src_ipv4_src_ipv6.ipv4_layout.ipv4),
+		       &match.key->src, sizeof(match.key->src));
+		memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_c,
+				    dst_ipv4_dst_ipv6.ipv4_layout.ipv4),
+		       &match.mask->dst, sizeof(match.mask->dst));
+		memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_v,
+				    dst_ipv4_dst_ipv6.ipv4_layout.ipv4),
+		       &match.key->dst, sizeof(match.key->dst));
+	}
+
+	if (addr_type == FLOW_DISSECTOR_KEY_IPV6_ADDRS) {
+		struct flow_match_ipv6_addrs match;
+
+		flow_rule_match_ipv6_addrs(rule, &match);
+		memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_c,
+				    src_ipv4_src_ipv6.ipv6_layout.ipv6),
+		       &match.mask->src, sizeof(match.mask->src));
+		memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_v,
+				    src_ipv4_src_ipv6.ipv6_layout.ipv6),
+		       &match.key->src, sizeof(match.key->src));
+
+		memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_c,
+				    dst_ipv4_dst_ipv6.ipv6_layout.ipv6),
+		       &match.mask->dst, sizeof(match.mask->dst));
+		memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_v,
+				    dst_ipv4_dst_ipv6.ipv6_layout.ipv6),
+		       &match.key->dst, sizeof(match.key->dst));
+	}
+
+	if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_PORTS)) {
+		struct flow_match_ports match;
+
+		flow_rule_match_ports(rule, &match);
+		switch (ip_proto) {
+		case IPPROTO_TCP:
+			MLX5_SET(fte_match_set_lyr_2_4, headers_c,
+				 tcp_sport, ntohs(match.mask->src));
+			MLX5_SET(fte_match_set_lyr_2_4, headers_v,
+				 tcp_sport, ntohs(match.key->src));
+
+			MLX5_SET(fte_match_set_lyr_2_4, headers_c,
+				 tcp_dport, ntohs(match.mask->dst));
+			MLX5_SET(fte_match_set_lyr_2_4, headers_v,
+				 tcp_dport, ntohs(match.key->dst));
+			break;
+
+		case IPPROTO_UDP:
+			MLX5_SET(fte_match_set_lyr_2_4, headers_c,
+				 udp_sport, ntohs(match.mask->src));
+			MLX5_SET(fte_match_set_lyr_2_4, headers_v,
+				 udp_sport, ntohs(match.key->src));
+
+			MLX5_SET(fte_match_set_lyr_2_4, headers_c,
+				 udp_dport, ntohs(match.mask->dst));
+			MLX5_SET(fte_match_set_lyr_2_4, headers_v,
+				 udp_dport, ntohs(match.key->dst));
+			break;
+		default:
+			break;
+		}
+	}
+
+	if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_TCP)) {
+		struct flow_match_tcp match;
+
+		flow_rule_match_tcp(rule, &match);
+		MLX5_SET(fte_match_set_lyr_2_4, headers_c, tcp_flags,
+			 ntohs(match.mask->flags));
+		MLX5_SET(fte_match_set_lyr_2_4, headers_v, tcp_flags,
+			 ntohs(match.key->flags));
+	}
+
+	if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_META)) {
+		struct flow_match_meta match;
+
+		flow_rule_match_meta(rule, &match);
+
+		if (match.key->ingress_ifindex & match.mask->ingress_ifindex) {
+			struct net_device *dev;
+
+			dev = dev_get_by_index(&init_net, match.key->ingress_ifindex);
+			if (dev && MLX5_CAP_ESW_FLOWTABLE(ct_priv->dev, flow_source))
+				spec->flow_context.flow_source =
+					mlx5_tc_ct_get_flow_source_match(ct_priv, dev);
+
+			dev_put(dev);
+		}
+	}
+
+	return 0;
+}
+
+static void
+mlx5_tc_ct_counter_put(struct mlx5_tc_ct_priv *ct_priv, struct mlx5_ct_entry *entry)
+{
+	if (entry->counter->is_shared &&
+	    !refcount_dec_and_test(&entry->counter->refcount))
+		return;
+
+	mlx5_fc_destroy(ct_priv->dev, entry->counter->counter);
+	kfree(entry->counter);
+}
+
+static void
+mlx5_tc_ct_entry_del_rule(struct mlx5_tc_ct_priv *ct_priv,
+			  struct mlx5_ct_entry *entry,
+			  bool nat)
+{
+	struct mlx5_ct_zone_rule *zone_rule = &entry->zone_rules[nat];
+	struct mlx5_flow_attr *attr = zone_rule->attr;
+
+	ct_dbg("Deleting ct entry rule in zone %d", entry->tuple.zone);
+
+	ct_priv->fs_ops->ct_rule_del(ct_priv->fs, zone_rule->rule);
+	mlx5_tc_ct_entry_destroy_mod_hdr(ct_priv, zone_rule->attr, zone_rule->mh);
+	mlx5_put_label_mapping(ct_priv, attr->ct_attr.ct_labels_id);
+	kfree(attr);
+}
+
+static void
+mlx5_tc_ct_entry_del_rules(struct mlx5_tc_ct_priv *ct_priv,
+			   struct mlx5_ct_entry *entry)
+{
+	mlx5_tc_ct_entry_del_rule(ct_priv, entry, true);
+	mlx5_tc_ct_entry_del_rule(ct_priv, entry, false);
+
+	atomic_dec(&ct_priv->debugfs.stats.offloaded);
+}
+
+static struct flow_action_entry *
+mlx5_tc_ct_get_ct_metadata_action(struct flow_rule *flow_rule)
+{
+	struct flow_action *flow_action = &flow_rule->action;
+	struct flow_action_entry *act;
+	int i;
+
+	flow_action_for_each(i, act, flow_action) {
+		if (act->id == FLOW_ACTION_CT_METADATA)
+			return act;
+	}
+
+	return NULL;
+}
+
+static int
+mlx5_tc_ct_entry_set_registers(struct mlx5_tc_ct_priv *ct_priv,
+			       struct mlx5e_tc_mod_hdr_acts *mod_acts,
+			       u8 ct_state,
+			       u32 mark,
+			       u32 labels_id,
+			       u8 zone_restore_id)
+{
+	enum mlx5_flow_namespace_type ns = ct_priv->ns_type;
+	struct mlx5_core_dev *dev = ct_priv->dev;
+	int err;
+
+	err = mlx5e_tc_match_to_reg_set(dev, mod_acts, ns,
+					CTSTATE_TO_REG, ct_state);
+	if (err)
+		return err;
+
+	err = mlx5e_tc_match_to_reg_set(dev, mod_acts, ns,
+					MARK_TO_REG, mark);
+	if (err)
+		return err;
+
+	err = mlx5e_tc_match_to_reg_set(dev, mod_acts, ns,
+					LABELS_TO_REG, labels_id);
+	if (err)
+		return err;
+
+	err = mlx5e_tc_match_to_reg_set(dev, mod_acts, ns,
+					ZONE_RESTORE_TO_REG, zone_restore_id);
+	if (err)
+		return err;
+
+	/* Make another copy of zone id in reg_b for
+	 * NIC rx flows since we don't copy reg_c1 to
+	 * reg_b upon miss.
+	 */
+	if (ns != MLX5_FLOW_NAMESPACE_FDB) {
+		err = mlx5e_tc_match_to_reg_set(dev, mod_acts, ns,
+						NIC_ZONE_RESTORE_TO_REG, zone_restore_id);
+		if (err)
+			return err;
+	}
+	return 0;
+}
+
+int mlx5_tc_ct_set_ct_clear_regs(struct mlx5_tc_ct_priv *priv,
+				 struct mlx5e_tc_mod_hdr_acts *mod_acts)
+{
+		return mlx5_tc_ct_entry_set_registers(priv, mod_acts, 0, 0, 0, 0);
+}
+
+static int
+mlx5_tc_ct_parse_mangle_to_mod_act(struct flow_action_entry *act,
+				   char *modact)
+{
+	u32 offset = act->mangle.offset, field;
+
+	switch (act->mangle.htype) {
+	case FLOW_ACT_MANGLE_HDR_TYPE_IP4:
+		MLX5_SET(set_action_in, modact, length, 0);
+		if (offset == offsetof(struct iphdr, saddr))
+			field = MLX5_ACTION_IN_FIELD_OUT_SIPV4;
+		else if (offset == offsetof(struct iphdr, daddr))
+			field = MLX5_ACTION_IN_FIELD_OUT_DIPV4;
+		else
+			return -EOPNOTSUPP;
+		break;
+
+	case FLOW_ACT_MANGLE_HDR_TYPE_IP6:
+		MLX5_SET(set_action_in, modact, length, 0);
+		if (offset == offsetof(struct ipv6hdr, saddr) + 12)
+			field = MLX5_ACTION_IN_FIELD_OUT_SIPV6_31_0;
+		else if (offset == offsetof(struct ipv6hdr, saddr) + 8)
+			field = MLX5_ACTION_IN_FIELD_OUT_SIPV6_63_32;
+		else if (offset == offsetof(struct ipv6hdr, saddr) + 4)
+			field = MLX5_ACTION_IN_FIELD_OUT_SIPV6_95_64;
+		else if (offset == offsetof(struct ipv6hdr, saddr))
+			field = MLX5_ACTION_IN_FIELD_OUT_SIPV6_127_96;
+		else if (offset == offsetof(struct ipv6hdr, daddr) + 12)
+			field = MLX5_ACTION_IN_FIELD_OUT_DIPV6_31_0;
+		else if (offset == offsetof(struct ipv6hdr, daddr) + 8)
+			field = MLX5_ACTION_IN_FIELD_OUT_DIPV6_63_32;
+		else if (offset == offsetof(struct ipv6hdr, daddr) + 4)
+			field = MLX5_ACTION_IN_FIELD_OUT_DIPV6_95_64;
+		else if (offset == offsetof(struct ipv6hdr, daddr))
+			field = MLX5_ACTION_IN_FIELD_OUT_DIPV6_127_96;
+		else
+			return -EOPNOTSUPP;
+		break;
+
+	case FLOW_ACT_MANGLE_HDR_TYPE_TCP:
+		MLX5_SET(set_action_in, modact, length, 16);
+		if (offset == offsetof(struct tcphdr, source))
+			field = MLX5_ACTION_IN_FIELD_OUT_TCP_SPORT;
+		else if (offset == offsetof(struct tcphdr, dest))
+			field = MLX5_ACTION_IN_FIELD_OUT_TCP_DPORT;
+		else
+			return -EOPNOTSUPP;
+		break;
+
+	case FLOW_ACT_MANGLE_HDR_TYPE_UDP:
+		MLX5_SET(set_action_in, modact, length, 16);
+		if (offset == offsetof(struct udphdr, source))
+			field = MLX5_ACTION_IN_FIELD_OUT_UDP_SPORT;
+		else if (offset == offsetof(struct udphdr, dest))
+			field = MLX5_ACTION_IN_FIELD_OUT_UDP_DPORT;
+		else
+			return -EOPNOTSUPP;
+		break;
+
+	default:
+		return -EOPNOTSUPP;
+	}
+
+	MLX5_SET(set_action_in, modact, action_type, MLX5_ACTION_TYPE_SET);
+	MLX5_SET(set_action_in, modact, offset, 0);
+	MLX5_SET(set_action_in, modact, field, field);
+	MLX5_SET(set_action_in, modact, data, act->mangle.val);
+
+	return 0;
+}
+
+static int
+mlx5_tc_ct_entry_create_nat(struct mlx5_tc_ct_priv *ct_priv,
+			    struct flow_rule *flow_rule,
+			    struct mlx5e_tc_mod_hdr_acts *mod_acts)
+{
+	struct flow_action *flow_action = &flow_rule->action;
+	struct mlx5_core_dev *mdev = ct_priv->dev;
+	struct flow_action_entry *act;
+	char *modact;
+	int err, i;
+
+	flow_action_for_each(i, act, flow_action) {
+		switch (act->id) {
+		case FLOW_ACTION_MANGLE: {
+			modact = mlx5e_mod_hdr_alloc(mdev, ct_priv->ns_type, mod_acts);
+			if (IS_ERR(modact))
+				return PTR_ERR(modact);
+
+			err = mlx5_tc_ct_parse_mangle_to_mod_act(act, modact);
+			if (err)
+				return err;
+
+			mod_acts->num_actions++;
+		}
+		break;
+
+		case FLOW_ACTION_CT_METADATA:
+			/* Handled earlier */
+			continue;
+		default:
+			return -EOPNOTSUPP;
+		}
+	}
+
+	return 0;
+}
+
+static int
+mlx5_tc_ct_entry_create_mod_hdr(struct mlx5_tc_ct_priv *ct_priv,
+				struct mlx5_flow_attr *attr,
+				struct flow_rule *flow_rule,
+				struct mlx5e_mod_hdr_handle **mh,
+				u8 zone_restore_id, bool nat_table, bool has_nat)
+{
+	DECLARE_MOD_HDR_ACTS_ACTIONS(actions_arr, MLX5_CT_MIN_MOD_ACTS);
+	DECLARE_MOD_HDR_ACTS(mod_acts, actions_arr);
+	struct flow_action_entry *meta;
+	u16 ct_state = 0;
+	int err;
+
+	meta = mlx5_tc_ct_get_ct_metadata_action(flow_rule);
+	if (!meta)
+		return -EOPNOTSUPP;
+
+	err = mlx5_get_label_mapping(ct_priv, meta->ct_metadata.labels,
+				     &attr->ct_attr.ct_labels_id);
+	if (err)
+		return -EOPNOTSUPP;
+	if (nat_table) {
+		if (has_nat) {
+			err = mlx5_tc_ct_entry_create_nat(ct_priv, flow_rule, &mod_acts);
+			if (err)
+				goto err_mapping;
+		}
+
+		ct_state |= MLX5_CT_STATE_NAT_BIT;
+	}
+
+	ct_state |= MLX5_CT_STATE_ESTABLISHED_BIT | MLX5_CT_STATE_TRK_BIT;
+	ct_state |= meta->ct_metadata.orig_dir ? 0 : MLX5_CT_STATE_REPLY_BIT;
+	err = mlx5_tc_ct_entry_set_registers(ct_priv, &mod_acts,
+					     ct_state,
+					     meta->ct_metadata.mark,
+					     attr->ct_attr.ct_labels_id,
+					     zone_restore_id);
+	if (err)
+		goto err_mapping;
+
+	if (nat_table && has_nat) {
+		attr->modify_hdr = mlx5_modify_header_alloc(ct_priv->dev, ct_priv->ns_type,
+							    mod_acts.num_actions,
+							    mod_acts.actions);
+		if (IS_ERR(attr->modify_hdr)) {
+			err = PTR_ERR(attr->modify_hdr);
+			goto err_mapping;
+		}
+
+		*mh = NULL;
+	} else {
+		*mh = mlx5e_mod_hdr_attach(ct_priv->dev,
+					   ct_priv->mod_hdr_tbl,
+					   ct_priv->ns_type,
+					   &mod_acts);
+		if (IS_ERR(*mh)) {
+			err = PTR_ERR(*mh);
+			goto err_mapping;
+		}
+		attr->modify_hdr = mlx5e_mod_hdr_get(*mh);
+	}
+
+	mlx5e_mod_hdr_dealloc(&mod_acts);
+	return 0;
+
+err_mapping:
+	mlx5e_mod_hdr_dealloc(&mod_acts);
+	mlx5_put_label_mapping(ct_priv, attr->ct_attr.ct_labels_id);
+	return err;
+}
+
+static void
+mlx5_tc_ct_entry_destroy_mod_hdr(struct mlx5_tc_ct_priv *ct_priv,
+				 struct mlx5_flow_attr *attr,
+				 struct mlx5e_mod_hdr_handle *mh)
+{
+	if (mh)
+		mlx5e_mod_hdr_detach(ct_priv->dev, ct_priv->mod_hdr_tbl, mh);
+	else
+		mlx5_modify_header_dealloc(ct_priv->dev, attr->modify_hdr);
+}
+
+static int
+mlx5_tc_ct_entry_add_rule(struct mlx5_tc_ct_priv *ct_priv,
+			  struct flow_rule *flow_rule,
+			  struct mlx5_ct_entry *entry,
+			  bool nat, u8 zone_restore_id)
+{
+	struct mlx5_ct_zone_rule *zone_rule = &entry->zone_rules[nat];
+	struct mlx5e_priv *priv = netdev_priv(ct_priv->netdev);
+	struct mlx5_flow_spec *spec = NULL;
+	struct mlx5_flow_attr *attr;
+	int err;
+
+	zone_rule->nat = nat;
+
+	spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
+	if (!spec)
+		return -ENOMEM;
+
+	attr = mlx5_alloc_flow_attr(ct_priv->ns_type);
+	if (!attr) {
+		err = -ENOMEM;
+		goto err_attr;
+	}
+
+	err = mlx5_tc_ct_entry_create_mod_hdr(ct_priv, attr, flow_rule,
+					      &zone_rule->mh,
+					      zone_restore_id,
+					      nat,
+					      mlx5_tc_ct_entry_has_nat(entry));
+	if (err) {
+		ct_dbg("Failed to create ct entry mod hdr");
+		goto err_mod_hdr;
+	}
+
+	attr->action = MLX5_FLOW_CONTEXT_ACTION_MOD_HDR |
+		       MLX5_FLOW_CONTEXT_ACTION_FWD_DEST |
+		       MLX5_FLOW_CONTEXT_ACTION_COUNT;
+	attr->dest_chain = 0;
+	attr->dest_ft = mlx5e_tc_post_act_get_ft(ct_priv->post_act);
+	attr->ft = nat ? ct_priv->ct_nat : ct_priv->ct;
+	if (entry->tuple.ip_proto == IPPROTO_TCP ||
+	    entry->tuple.ip_proto == IPPROTO_UDP)
+		attr->outer_match_level = MLX5_MATCH_L4;
+	else
+		attr->outer_match_level = MLX5_MATCH_L3;
+	attr->counter = entry->counter->counter;
+	attr->flags |= MLX5_ATTR_FLAG_NO_IN_PORT;
+	if (ct_priv->ns_type == MLX5_FLOW_NAMESPACE_FDB)
+		attr->esw_attr->in_mdev = priv->mdev;
+
+	mlx5_tc_ct_set_tuple_match(ct_priv, spec, flow_rule);
+	mlx5e_tc_match_to_reg_match(spec, ZONE_TO_REG, entry->tuple.zone, MLX5_CT_ZONE_MASK);
+
+	zone_rule->rule = ct_priv->fs_ops->ct_rule_add(ct_priv->fs, spec, attr, flow_rule);
+	if (IS_ERR(zone_rule->rule)) {
+		err = PTR_ERR(zone_rule->rule);
+		ct_dbg("Failed to add ct entry rule, nat: %d", nat);
+		goto err_rule;
+	}
+
+	zone_rule->attr = attr;
+
+	kvfree(spec);
+	ct_dbg("Offloaded ct entry rule in zone %d", entry->tuple.zone);
+
+	return 0;
+
+err_rule:
+	mlx5_tc_ct_entry_destroy_mod_hdr(ct_priv, zone_rule->attr, zone_rule->mh);
+	mlx5_put_label_mapping(ct_priv, attr->ct_attr.ct_labels_id);
+err_mod_hdr:
+	kfree(attr);
+err_attr:
+	kvfree(spec);
+	return err;
+}
+
+static bool
+mlx5_tc_ct_entry_valid(struct mlx5_ct_entry *entry)
+{
+	return test_bit(MLX5_CT_ENTRY_FLAG_VALID, &entry->flags);
+}
+
+static struct mlx5_ct_entry *
+mlx5_tc_ct_entry_get(struct mlx5_tc_ct_priv *ct_priv, struct mlx5_ct_tuple *tuple)
+{
+	struct mlx5_ct_entry *entry;
+
+	entry = rhashtable_lookup_fast(&ct_priv->ct_tuples_ht, tuple,
+				       tuples_ht_params);
+	if (entry && mlx5_tc_ct_entry_valid(entry) &&
+	    refcount_inc_not_zero(&entry->refcnt)) {
+		return entry;
+	} else if (!entry) {
+		entry = rhashtable_lookup_fast(&ct_priv->ct_tuples_nat_ht,
+					       tuple, tuples_nat_ht_params);
+		if (entry && mlx5_tc_ct_entry_valid(entry) &&
+		    refcount_inc_not_zero(&entry->refcnt))
+			return entry;
+	}
+
+	return entry ? ERR_PTR(-EINVAL) : NULL;
+}
+
+static void mlx5_tc_ct_entry_remove_from_tuples(struct mlx5_ct_entry *entry)
+{
+	struct mlx5_tc_ct_priv *ct_priv = entry->ct_priv;
+
+	rhashtable_remove_fast(&ct_priv->ct_tuples_nat_ht,
+			       &entry->tuple_nat_node,
+			       tuples_nat_ht_params);
+	rhashtable_remove_fast(&ct_priv->ct_tuples_ht, &entry->tuple_node,
+			       tuples_ht_params);
+}
+
+static void mlx5_tc_ct_entry_del(struct mlx5_ct_entry *entry)
+{
+	struct mlx5_tc_ct_priv *ct_priv = entry->ct_priv;
+
+	mlx5_tc_ct_entry_del_rules(ct_priv, entry);
+
+	spin_lock_bh(&ct_priv->ht_lock);
+	mlx5_tc_ct_entry_remove_from_tuples(entry);
+	spin_unlock_bh(&ct_priv->ht_lock);
+
+	mlx5_tc_ct_counter_put(ct_priv, entry);
+	kfree(entry);
+}
+
+static void
+mlx5_tc_ct_entry_put(struct mlx5_ct_entry *entry)
+{
+	if (!refcount_dec_and_test(&entry->refcnt))
+		return;
+
+	mlx5_tc_ct_entry_del(entry);
+}
+
+static void mlx5_tc_ct_entry_del_work(struct work_struct *work)
+{
+	struct mlx5_ct_entry *entry = container_of(work, struct mlx5_ct_entry, work);
+
+	mlx5_tc_ct_entry_del(entry);
+}
+
+static void
+__mlx5_tc_ct_entry_put(struct mlx5_ct_entry *entry)
+{
+	if (!refcount_dec_and_test(&entry->refcnt))
+		return;
+
+	INIT_WORK(&entry->work, mlx5_tc_ct_entry_del_work);
+	queue_work(entry->ct_priv->wq, &entry->work);
+}
+
+static struct mlx5_ct_counter *
+mlx5_tc_ct_counter_create(struct mlx5_tc_ct_priv *ct_priv)
+{
+	struct mlx5_ct_counter *counter;
+	int ret;
+
+	counter = kzalloc(sizeof(*counter), GFP_KERNEL);
+	if (!counter)
+		return ERR_PTR(-ENOMEM);
+
+	counter->is_shared = false;
+	counter->counter = mlx5_fc_create_ex(ct_priv->dev, true);
+	if (IS_ERR(counter->counter)) {
+		ct_dbg("Failed to create counter for ct entry");
+		ret = PTR_ERR(counter->counter);
+		kfree(counter);
+		return ERR_PTR(ret);
+	}
+
+	return counter;
+}
+
+static struct mlx5_ct_counter *
+mlx5_tc_ct_shared_counter_get(struct mlx5_tc_ct_priv *ct_priv,
+			      struct mlx5_ct_entry *entry)
+{
+	struct mlx5_ct_tuple rev_tuple = entry->tuple;
+	struct mlx5_ct_counter *shared_counter;
+	struct mlx5_ct_entry *rev_entry;
+
+	/* get the reversed tuple */
+	swap(rev_tuple.port.src, rev_tuple.port.dst);
+
+	if (rev_tuple.addr_type == FLOW_DISSECTOR_KEY_IPV4_ADDRS) {
+		__be32 tmp_addr = rev_tuple.ip.src_v4;
+
+		rev_tuple.ip.src_v4 = rev_tuple.ip.dst_v4;
+		rev_tuple.ip.dst_v4 = tmp_addr;
+	} else if (rev_tuple.addr_type == FLOW_DISSECTOR_KEY_IPV6_ADDRS) {
+		struct in6_addr tmp_addr = rev_tuple.ip.src_v6;
+
+		rev_tuple.ip.src_v6 = rev_tuple.ip.dst_v6;
+		rev_tuple.ip.dst_v6 = tmp_addr;
+	} else {
+		return ERR_PTR(-EOPNOTSUPP);
+	}
+
+	/* Use the same counter as the reverse direction */
+	spin_lock_bh(&ct_priv->ht_lock);
+	rev_entry = mlx5_tc_ct_entry_get(ct_priv, &rev_tuple);
+
+	if (IS_ERR(rev_entry)) {
+		spin_unlock_bh(&ct_priv->ht_lock);
+		goto create_counter;
+	}
+
+	if (rev_entry && refcount_inc_not_zero(&rev_entry->counter->refcount)) {
+		ct_dbg("Using shared counter entry=0x%p rev=0x%p", entry, rev_entry);
+		shared_counter = rev_entry->counter;
+		spin_unlock_bh(&ct_priv->ht_lock);
+
+		mlx5_tc_ct_entry_put(rev_entry);
+		return shared_counter;
+	}
+
+	spin_unlock_bh(&ct_priv->ht_lock);
+
+create_counter:
+
+	shared_counter = mlx5_tc_ct_counter_create(ct_priv);
+	if (IS_ERR(shared_counter))
+		return shared_counter;
+
+	shared_counter->is_shared = true;
+	refcount_set(&shared_counter->refcount, 1);
+	return shared_counter;
+}
+
+static int
+mlx5_tc_ct_entry_add_rules(struct mlx5_tc_ct_priv *ct_priv,
+			   struct flow_rule *flow_rule,
+			   struct mlx5_ct_entry *entry,
+			   u8 zone_restore_id)
+{
+	int err;
+
+	if (nf_ct_acct_enabled(dev_net(ct_priv->netdev)))
+		entry->counter = mlx5_tc_ct_counter_create(ct_priv);
+	else
+		entry->counter = mlx5_tc_ct_shared_counter_get(ct_priv, entry);
+
+	if (IS_ERR(entry->counter)) {
+		err = PTR_ERR(entry->counter);
+		return err;
+	}
+
+	err = mlx5_tc_ct_entry_add_rule(ct_priv, flow_rule, entry, false,
+					zone_restore_id);
+	if (err)
+		goto err_orig;
+
+	err = mlx5_tc_ct_entry_add_rule(ct_priv, flow_rule, entry, true,
+					zone_restore_id);
+	if (err)
+		goto err_nat;
+
+	atomic_inc(&ct_priv->debugfs.stats.offloaded);
+	return 0;
+
+err_nat:
+	mlx5_tc_ct_entry_del_rule(ct_priv, entry, false);
+err_orig:
+	mlx5_tc_ct_counter_put(ct_priv, entry);
+	return err;
+}
+
+static int
+mlx5_tc_ct_block_flow_offload_add(struct mlx5_ct_ft *ft,
+				  struct flow_cls_offload *flow)
+{
+	struct flow_rule *flow_rule = flow_cls_offload_flow_rule(flow);
+	struct mlx5_tc_ct_priv *ct_priv = ft->ct_priv;
+	struct flow_action_entry *meta_action;
+	unsigned long cookie = flow->cookie;
+	struct mlx5_ct_entry *entry;
+	int err;
+
+	meta_action = mlx5_tc_ct_get_ct_metadata_action(flow_rule);
+	if (!meta_action)
+		return -EOPNOTSUPP;
+
+	spin_lock_bh(&ct_priv->ht_lock);
+	entry = rhashtable_lookup_fast(&ft->ct_entries_ht, &cookie, cts_ht_params);
+	if (entry && refcount_inc_not_zero(&entry->refcnt)) {
+		spin_unlock_bh(&ct_priv->ht_lock);
+		mlx5_tc_ct_entry_put(entry);
+		return -EEXIST;
+	}
+	spin_unlock_bh(&ct_priv->ht_lock);
+
+	entry = kzalloc(sizeof(*entry), GFP_KERNEL);
+	if (!entry)
+		return -ENOMEM;
+
+	entry->tuple.zone = ft->zone;
+	entry->cookie = flow->cookie;
+	entry->restore_cookie = meta_action->ct_metadata.cookie;
+	refcount_set(&entry->refcnt, 2);
+	entry->ct_priv = ct_priv;
+
+	err = mlx5_tc_ct_rule_to_tuple(&entry->tuple, flow_rule);
+	if (err)
+		goto err_set;
+
+	memcpy(&entry->tuple_nat, &entry->tuple, sizeof(entry->tuple));
+	err = mlx5_tc_ct_rule_to_tuple_nat(&entry->tuple_nat, flow_rule);
+	if (err)
+		goto err_set;
+
+	spin_lock_bh(&ct_priv->ht_lock);
+
+	err = rhashtable_lookup_insert_fast(&ft->ct_entries_ht, &entry->node,
+					    cts_ht_params);
+	if (err)
+		goto err_entries;
+
+	err = rhashtable_lookup_insert_fast(&ct_priv->ct_tuples_ht,
+					    &entry->tuple_node,
+					    tuples_ht_params);
+	if (err)
+		goto err_tuple;
+
+	if (memcmp(&entry->tuple, &entry->tuple_nat, sizeof(entry->tuple))) {
+		err = rhashtable_lookup_insert_fast(&ct_priv->ct_tuples_nat_ht,
+						    &entry->tuple_nat_node,
+						    tuples_nat_ht_params);
+		if (err)
+			goto err_tuple_nat;
+	}
+	spin_unlock_bh(&ct_priv->ht_lock);
+
+	err = mlx5_tc_ct_entry_add_rules(ct_priv, flow_rule, entry,
+					 ft->zone_restore_id);
+	if (err)
+		goto err_rules;
+
+	set_bit(MLX5_CT_ENTRY_FLAG_VALID, &entry->flags);
+	mlx5_tc_ct_entry_put(entry); /* this function reference */
+
+	return 0;
+
+err_rules:
+	spin_lock_bh(&ct_priv->ht_lock);
+	if (mlx5_tc_ct_entry_has_nat(entry))
+		rhashtable_remove_fast(&ct_priv->ct_tuples_nat_ht,
+				       &entry->tuple_nat_node, tuples_nat_ht_params);
+err_tuple_nat:
+	rhashtable_remove_fast(&ct_priv->ct_tuples_ht,
+			       &entry->tuple_node,
+			       tuples_ht_params);
+err_tuple:
+	rhashtable_remove_fast(&ft->ct_entries_ht,
+			       &entry->node,
+			       cts_ht_params);
+err_entries:
+	spin_unlock_bh(&ct_priv->ht_lock);
+err_set:
+	kfree(entry);
+	if (err != -EEXIST)
+		netdev_warn(ct_priv->netdev, "Failed to offload ct entry, err: %d\n", err);
+	return err;
+}
+
+static int
+mlx5_tc_ct_block_flow_offload_del(struct mlx5_ct_ft *ft,
+				  struct flow_cls_offload *flow)
+{
+	struct mlx5_tc_ct_priv *ct_priv = ft->ct_priv;
+	unsigned long cookie = flow->cookie;
+	struct mlx5_ct_entry *entry;
+
+	spin_lock_bh(&ct_priv->ht_lock);
+	entry = rhashtable_lookup_fast(&ft->ct_entries_ht, &cookie, cts_ht_params);
+	if (!entry) {
+		spin_unlock_bh(&ct_priv->ht_lock);
+		return -ENOENT;
+	}
+
+	if (!mlx5_tc_ct_entry_valid(entry)) {
+		spin_unlock_bh(&ct_priv->ht_lock);
+		return -EINVAL;
+	}
+
+	rhashtable_remove_fast(&ft->ct_entries_ht, &entry->node, cts_ht_params);
+	spin_unlock_bh(&ct_priv->ht_lock);
+
+	mlx5_tc_ct_entry_put(entry);
+
+	return 0;
+}
+
+static int
+mlx5_tc_ct_block_flow_offload_stats(struct mlx5_ct_ft *ft,
+				    struct flow_cls_offload *f)
+{
+	struct mlx5_tc_ct_priv *ct_priv = ft->ct_priv;
+	unsigned long cookie = f->cookie;
+	struct mlx5_ct_entry *entry;
+	u64 lastuse, packets, bytes;
+
+	spin_lock_bh(&ct_priv->ht_lock);
+	entry = rhashtable_lookup_fast(&ft->ct_entries_ht, &cookie, cts_ht_params);
+	if (!entry) {
+		spin_unlock_bh(&ct_priv->ht_lock);
+		return -ENOENT;
+	}
+
+	if (!mlx5_tc_ct_entry_valid(entry) || !refcount_inc_not_zero(&entry->refcnt)) {
+		spin_unlock_bh(&ct_priv->ht_lock);
+		return -EINVAL;
+	}
+
+	spin_unlock_bh(&ct_priv->ht_lock);
+
+	mlx5_fc_query_cached(entry->counter->counter, &bytes, &packets, &lastuse);
+	flow_stats_update(&f->stats, bytes, packets, 0, lastuse,
+			  FLOW_ACTION_HW_STATS_DELAYED);
+
+	mlx5_tc_ct_entry_put(entry);
+	return 0;
+}
+
+static int
+mlx5_tc_ct_block_flow_offload(enum tc_setup_type type, void *type_data,
+			      void *cb_priv)
+{
+	struct flow_cls_offload *f = type_data;
+	struct mlx5_ct_ft *ft = cb_priv;
+
+	if (type != TC_SETUP_CLSFLOWER)
+		return -EOPNOTSUPP;
+
+	switch (f->command) {
+	case FLOW_CLS_REPLACE:
+		return mlx5_tc_ct_block_flow_offload_add(ft, f);
+	case FLOW_CLS_DESTROY:
+		return mlx5_tc_ct_block_flow_offload_del(ft, f);
+	case FLOW_CLS_STATS:
+		return mlx5_tc_ct_block_flow_offload_stats(ft, f);
+	default:
+		break;
+	}
+
+	return -EOPNOTSUPP;
+}
+
+static bool
+mlx5_tc_ct_skb_to_tuple(struct sk_buff *skb, struct mlx5_ct_tuple *tuple,
+			u16 zone)
+{
+	struct flow_keys flow_keys;
+
+	skb_reset_network_header(skb);
+	skb_flow_dissect_flow_keys(skb, &flow_keys, FLOW_DISSECTOR_F_STOP_BEFORE_ENCAP);
+
+	tuple->zone = zone;
+
+	if (flow_keys.basic.ip_proto != IPPROTO_TCP &&
+	    flow_keys.basic.ip_proto != IPPROTO_UDP &&
+	    flow_keys.basic.ip_proto != IPPROTO_GRE)
+		return false;
+
+	if (flow_keys.basic.ip_proto == IPPROTO_TCP ||
+	    flow_keys.basic.ip_proto == IPPROTO_UDP) {
+		tuple->port.src = flow_keys.ports.src;
+		tuple->port.dst = flow_keys.ports.dst;
+	}
+	tuple->n_proto = flow_keys.basic.n_proto;
+	tuple->ip_proto = flow_keys.basic.ip_proto;
+
+	switch (flow_keys.basic.n_proto) {
+	case htons(ETH_P_IP):
+		tuple->addr_type = FLOW_DISSECTOR_KEY_IPV4_ADDRS;
+		tuple->ip.src_v4 = flow_keys.addrs.v4addrs.src;
+		tuple->ip.dst_v4 = flow_keys.addrs.v4addrs.dst;
+		break;
+
+	case htons(ETH_P_IPV6):
+		tuple->addr_type = FLOW_DISSECTOR_KEY_IPV6_ADDRS;
+		tuple->ip.src_v6 = flow_keys.addrs.v6addrs.src;
+		tuple->ip.dst_v6 = flow_keys.addrs.v6addrs.dst;
+		break;
+	default:
+		goto out;
+	}
+
+	return true;
+
+out:
+	return false;
+}
+
+int mlx5_tc_ct_add_no_trk_match(struct mlx5_flow_spec *spec)
+{
+	u32 ctstate = 0, ctstate_mask = 0;
+
+	mlx5e_tc_match_to_reg_get_match(spec, CTSTATE_TO_REG,
+					&ctstate, &ctstate_mask);
+
+	if ((ctstate & ctstate_mask) == MLX5_CT_STATE_TRK_BIT)
+		return -EOPNOTSUPP;
+
+	ctstate_mask |= MLX5_CT_STATE_TRK_BIT;
+	mlx5e_tc_match_to_reg_match(spec, CTSTATE_TO_REG,
+				    ctstate, ctstate_mask);
+
+	return 0;
+}
+
+void mlx5_tc_ct_match_del(struct mlx5_tc_ct_priv *priv, struct mlx5_ct_attr *ct_attr)
+{
+	if (!priv || !ct_attr->ct_labels_id)
+		return;
+
+	mlx5_put_label_mapping(priv, ct_attr->ct_labels_id);
+}
+
+int
+mlx5_tc_ct_match_add(struct mlx5_tc_ct_priv *priv,
+		     struct mlx5_flow_spec *spec,
+		     struct flow_cls_offload *f,
+		     struct mlx5_ct_attr *ct_attr,
+		     struct netlink_ext_ack *extack)
+{
+	bool trk, est, untrk, unest, new, rpl, unrpl, rel, unrel, inv, uninv;
+	struct flow_rule *rule = flow_cls_offload_flow_rule(f);
+	struct flow_dissector_key_ct *mask, *key;
+	u32 ctstate = 0, ctstate_mask = 0;
+	u16 ct_state_on, ct_state_off;
+	u16 ct_state, ct_state_mask;
+	struct flow_match_ct match;
+	u32 ct_labels[4];
+
+	if (!flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_CT))
+		return 0;
+
+	if (!priv) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "offload of ct matching isn't available");
+		return -EOPNOTSUPP;
+	}
+
+	flow_rule_match_ct(rule, &match);
+
+	key = match.key;
+	mask = match.mask;
+
+	ct_state = key->ct_state;
+	ct_state_mask = mask->ct_state;
+
+	if (ct_state_mask & ~(TCA_FLOWER_KEY_CT_FLAGS_TRACKED |
+			      TCA_FLOWER_KEY_CT_FLAGS_ESTABLISHED |
+			      TCA_FLOWER_KEY_CT_FLAGS_NEW |
+			      TCA_FLOWER_KEY_CT_FLAGS_REPLY |
+			      TCA_FLOWER_KEY_CT_FLAGS_RELATED |
+			      TCA_FLOWER_KEY_CT_FLAGS_INVALID)) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "only ct_state trk, est, new and rpl are supported for offload");
+		return -EOPNOTSUPP;
+	}
+
+	ct_state_on = ct_state & ct_state_mask;
+	ct_state_off = (ct_state & ct_state_mask) ^ ct_state_mask;
+	trk = ct_state_on & TCA_FLOWER_KEY_CT_FLAGS_TRACKED;
+	new = ct_state_on & TCA_FLOWER_KEY_CT_FLAGS_NEW;
+	est = ct_state_on & TCA_FLOWER_KEY_CT_FLAGS_ESTABLISHED;
+	rpl = ct_state_on & TCA_FLOWER_KEY_CT_FLAGS_REPLY;
+	rel = ct_state_on & TCA_FLOWER_KEY_CT_FLAGS_RELATED;
+	inv = ct_state_on & TCA_FLOWER_KEY_CT_FLAGS_INVALID;
+	untrk = ct_state_off & TCA_FLOWER_KEY_CT_FLAGS_TRACKED;
+	unest = ct_state_off & TCA_FLOWER_KEY_CT_FLAGS_ESTABLISHED;
+	unrpl = ct_state_off & TCA_FLOWER_KEY_CT_FLAGS_REPLY;
+	unrel = ct_state_off & TCA_FLOWER_KEY_CT_FLAGS_RELATED;
+	uninv = ct_state_off & TCA_FLOWER_KEY_CT_FLAGS_INVALID;
+
+	ctstate |= trk ? MLX5_CT_STATE_TRK_BIT : 0;
+	ctstate |= est ? MLX5_CT_STATE_ESTABLISHED_BIT : 0;
+	ctstate |= rpl ? MLX5_CT_STATE_REPLY_BIT : 0;
+	ctstate_mask |= (untrk || trk) ? MLX5_CT_STATE_TRK_BIT : 0;
+	ctstate_mask |= (unest || est) ? MLX5_CT_STATE_ESTABLISHED_BIT : 0;
+	ctstate_mask |= (unrpl || rpl) ? MLX5_CT_STATE_REPLY_BIT : 0;
+	ctstate_mask |= unrel ? MLX5_CT_STATE_RELATED_BIT : 0;
+	ctstate_mask |= uninv ? MLX5_CT_STATE_INVALID_BIT : 0;
+
+	if (rel) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "matching on ct_state +rel isn't supported");
+		return -EOPNOTSUPP;
+	}
+
+	if (inv) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "matching on ct_state +inv isn't supported");
+		return -EOPNOTSUPP;
+	}
+
+	if (new) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "matching on ct_state +new isn't supported");
+		return -EOPNOTSUPP;
+	}
+
+	if (mask->ct_zone)
+		mlx5e_tc_match_to_reg_match(spec, ZONE_TO_REG,
+					    key->ct_zone, MLX5_CT_ZONE_MASK);
+	if (ctstate_mask)
+		mlx5e_tc_match_to_reg_match(spec, CTSTATE_TO_REG,
+					    ctstate, ctstate_mask);
+	if (mask->ct_mark)
+		mlx5e_tc_match_to_reg_match(spec, MARK_TO_REG,
+					    key->ct_mark, mask->ct_mark);
+	if (mask->ct_labels[0] || mask->ct_labels[1] || mask->ct_labels[2] ||
+	    mask->ct_labels[3]) {
+		ct_labels[0] = key->ct_labels[0] & mask->ct_labels[0];
+		ct_labels[1] = key->ct_labels[1] & mask->ct_labels[1];
+		ct_labels[2] = key->ct_labels[2] & mask->ct_labels[2];
+		ct_labels[3] = key->ct_labels[3] & mask->ct_labels[3];
+		if (mlx5_get_label_mapping(priv, ct_labels, &ct_attr->ct_labels_id))
+			return -EOPNOTSUPP;
+		mlx5e_tc_match_to_reg_match(spec, LABELS_TO_REG, ct_attr->ct_labels_id,
+					    MLX5_CT_LABELS_MASK);
+	}
+
+	return 0;
+}
+
+int
+mlx5_tc_ct_parse_action(struct mlx5_tc_ct_priv *priv,
+			struct mlx5_flow_attr *attr,
+			struct mlx5e_tc_mod_hdr_acts *mod_acts,
+			const struct flow_action_entry *act,
+			struct netlink_ext_ack *extack)
+{
+	if (!priv) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "offload of ct action isn't available");
+		return -EOPNOTSUPP;
+	}
+
+	attr->ct_attr.zone = act->ct.zone;
+	attr->ct_attr.ct_action = act->ct.action;
+	attr->ct_attr.nf_ft = act->ct.flow_table;
+
+	return 0;
+}
+
+static int tc_ct_pre_ct_add_rules(struct mlx5_ct_ft *ct_ft,
+				  struct mlx5_tc_ct_pre *pre_ct,
+				  bool nat)
+{
+	struct mlx5_tc_ct_priv *ct_priv = ct_ft->ct_priv;
+	struct mlx5e_tc_mod_hdr_acts pre_mod_acts = {};
+	struct mlx5_core_dev *dev = ct_priv->dev;
+	struct mlx5_flow_table *ft = pre_ct->ft;
+	struct mlx5_flow_destination dest = {};
+	struct mlx5_flow_act flow_act = {};
+	struct mlx5_modify_hdr *mod_hdr;
+	struct mlx5_flow_handle *rule;
+	struct mlx5_flow_spec *spec;
+	u32 ctstate;
+	u16 zone;
+	int err;
+
+	spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
+	if (!spec)
+		return -ENOMEM;
+
+	zone = ct_ft->zone & MLX5_CT_ZONE_MASK;
+	err = mlx5e_tc_match_to_reg_set(dev, &pre_mod_acts, ct_priv->ns_type,
+					ZONE_TO_REG, zone);
+	if (err) {
+		ct_dbg("Failed to set zone register mapping");
+		goto err_mapping;
+	}
+
+	mod_hdr = mlx5_modify_header_alloc(dev, ct_priv->ns_type,
+					   pre_mod_acts.num_actions,
+					   pre_mod_acts.actions);
+
+	if (IS_ERR(mod_hdr)) {
+		err = PTR_ERR(mod_hdr);
+		ct_dbg("Failed to create pre ct mod hdr");
+		goto err_mapping;
+	}
+	pre_ct->modify_hdr = mod_hdr;
+
+	flow_act.action = MLX5_FLOW_CONTEXT_ACTION_FWD_DEST |
+			  MLX5_FLOW_CONTEXT_ACTION_MOD_HDR;
+	flow_act.flags |= FLOW_ACT_IGNORE_FLOW_LEVEL;
+	flow_act.modify_hdr = mod_hdr;
+	dest.type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
+
+	/* add flow rule */
+	mlx5e_tc_match_to_reg_match(spec, ZONE_TO_REG,
+				    zone, MLX5_CT_ZONE_MASK);
+	ctstate = MLX5_CT_STATE_TRK_BIT;
+	if (nat)
+		ctstate |= MLX5_CT_STATE_NAT_BIT;
+	mlx5e_tc_match_to_reg_match(spec, CTSTATE_TO_REG, ctstate, ctstate);
+
+	dest.ft = mlx5e_tc_post_act_get_ft(ct_priv->post_act);
+	rule = mlx5_add_flow_rules(ft, spec, &flow_act, &dest, 1);
+	if (IS_ERR(rule)) {
+		err = PTR_ERR(rule);
+		ct_dbg("Failed to add pre ct flow rule zone %d", zone);
+		goto err_flow_rule;
+	}
+	pre_ct->flow_rule = rule;
+
+	/* add miss rule */
+	dest.ft = nat ? ct_priv->ct_nat : ct_priv->ct;
+	rule = mlx5_add_flow_rules(ft, NULL, &flow_act, &dest, 1);
+	if (IS_ERR(rule)) {
+		err = PTR_ERR(rule);
+		ct_dbg("Failed to add pre ct miss rule zone %d", zone);
+		goto err_miss_rule;
+	}
+	pre_ct->miss_rule = rule;
+
+	mlx5e_mod_hdr_dealloc(&pre_mod_acts);
+	kvfree(spec);
+	return 0;
+
+err_miss_rule:
+	mlx5_del_flow_rules(pre_ct->flow_rule);
+err_flow_rule:
+	mlx5_modify_header_dealloc(dev, pre_ct->modify_hdr);
+err_mapping:
+	mlx5e_mod_hdr_dealloc(&pre_mod_acts);
+	kvfree(spec);
+	return err;
+}
+
+static void
+tc_ct_pre_ct_del_rules(struct mlx5_ct_ft *ct_ft,
+		       struct mlx5_tc_ct_pre *pre_ct)
+{
+	struct mlx5_tc_ct_priv *ct_priv = ct_ft->ct_priv;
+	struct mlx5_core_dev *dev = ct_priv->dev;
+
+	mlx5_del_flow_rules(pre_ct->flow_rule);
+	mlx5_del_flow_rules(pre_ct->miss_rule);
+	mlx5_modify_header_dealloc(dev, pre_ct->modify_hdr);
+}
+
+static int
+mlx5_tc_ct_alloc_pre_ct(struct mlx5_ct_ft *ct_ft,
+			struct mlx5_tc_ct_pre *pre_ct,
+			bool nat)
+{
+	int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
+	struct mlx5_tc_ct_priv *ct_priv = ct_ft->ct_priv;
+	struct mlx5_core_dev *dev = ct_priv->dev;
+	struct mlx5_flow_table_attr ft_attr = {};
+	struct mlx5_flow_namespace *ns;
+	struct mlx5_flow_table *ft;
+	struct mlx5_flow_group *g;
+	u32 metadata_reg_c_2_mask;
+	u32 *flow_group_in;
+	void *misc;
+	int err;
+
+	ns = mlx5_get_flow_namespace(dev, ct_priv->ns_type);
+	if (!ns) {
+		err = -EOPNOTSUPP;
+		ct_dbg("Failed to get flow namespace");
+		return err;
+	}
+
+	flow_group_in = kvzalloc(inlen, GFP_KERNEL);
+	if (!flow_group_in)
+		return -ENOMEM;
+
+	ft_attr.flags = MLX5_FLOW_TABLE_UNMANAGED;
+	ft_attr.prio =  ct_priv->ns_type ==  MLX5_FLOW_NAMESPACE_FDB ?
+			FDB_TC_OFFLOAD : MLX5E_TC_PRIO;
+	ft_attr.max_fte = 2;
+	ft_attr.level = 1;
+	ft = mlx5_create_flow_table(ns, &ft_attr);
+	if (IS_ERR(ft)) {
+		err = PTR_ERR(ft);
+		ct_dbg("Failed to create pre ct table");
+		goto out_free;
+	}
+	pre_ct->ft = ft;
+
+	/* create flow group */
+	MLX5_SET(create_flow_group_in, flow_group_in, start_flow_index, 0);
+	MLX5_SET(create_flow_group_in, flow_group_in, end_flow_index, 0);
+	MLX5_SET(create_flow_group_in, flow_group_in, match_criteria_enable,
+		 MLX5_MATCH_MISC_PARAMETERS_2);
+
+	misc = MLX5_ADDR_OF(create_flow_group_in, flow_group_in,
+			    match_criteria.misc_parameters_2);
+
+	metadata_reg_c_2_mask = MLX5_CT_ZONE_MASK;
+	metadata_reg_c_2_mask |= (MLX5_CT_STATE_TRK_BIT << 16);
+	if (nat)
+		metadata_reg_c_2_mask |= (MLX5_CT_STATE_NAT_BIT << 16);
+
+	MLX5_SET(fte_match_set_misc2, misc, metadata_reg_c_2,
+		 metadata_reg_c_2_mask);
+
+	g = mlx5_create_flow_group(ft, flow_group_in);
+	if (IS_ERR(g)) {
+		err = PTR_ERR(g);
+		ct_dbg("Failed to create pre ct group");
+		goto err_flow_grp;
+	}
+	pre_ct->flow_grp = g;
+
+	/* create miss group */
+	memset(flow_group_in, 0, inlen);
+	MLX5_SET(create_flow_group_in, flow_group_in, start_flow_index, 1);
+	MLX5_SET(create_flow_group_in, flow_group_in, end_flow_index, 1);
+	g = mlx5_create_flow_group(ft, flow_group_in);
+	if (IS_ERR(g)) {
+		err = PTR_ERR(g);
+		ct_dbg("Failed to create pre ct miss group");
+		goto err_miss_grp;
+	}
+	pre_ct->miss_grp = g;
+
+	err = tc_ct_pre_ct_add_rules(ct_ft, pre_ct, nat);
+	if (err)
+		goto err_add_rules;
+
+	kvfree(flow_group_in);
+	return 0;
+
+err_add_rules:
+	mlx5_destroy_flow_group(pre_ct->miss_grp);
+err_miss_grp:
+	mlx5_destroy_flow_group(pre_ct->flow_grp);
+err_flow_grp:
+	mlx5_destroy_flow_table(ft);
+out_free:
+	kvfree(flow_group_in);
+	return err;
+}
+
+static void
+mlx5_tc_ct_free_pre_ct(struct mlx5_ct_ft *ct_ft,
+		       struct mlx5_tc_ct_pre *pre_ct)
+{
+	tc_ct_pre_ct_del_rules(ct_ft, pre_ct);
+	mlx5_destroy_flow_group(pre_ct->miss_grp);
+	mlx5_destroy_flow_group(pre_ct->flow_grp);
+	mlx5_destroy_flow_table(pre_ct->ft);
+}
+
+static int
+mlx5_tc_ct_alloc_pre_ct_tables(struct mlx5_ct_ft *ft)
+{
+	int err;
+
+	err = mlx5_tc_ct_alloc_pre_ct(ft, &ft->pre_ct, false);
+	if (err)
+		return err;
+
+	err = mlx5_tc_ct_alloc_pre_ct(ft, &ft->pre_ct_nat, true);
+	if (err)
+		goto err_pre_ct_nat;
+
+	return 0;
+
+err_pre_ct_nat:
+	mlx5_tc_ct_free_pre_ct(ft, &ft->pre_ct);
+	return err;
+}
+
+static void
+mlx5_tc_ct_free_pre_ct_tables(struct mlx5_ct_ft *ft)
+{
+	mlx5_tc_ct_free_pre_ct(ft, &ft->pre_ct_nat);
+	mlx5_tc_ct_free_pre_ct(ft, &ft->pre_ct);
+}
+
+/* To avoid false lock dependency warning set the ct_entries_ht lock
+ * class different than the lock class of the ht being used when deleting
+ * last flow from a group and then deleting a group, we get into del_sw_flow_group()
+ * which call rhashtable_destroy on fg->ftes_hash which will take ht->mutex but
+ * it's different than the ht->mutex here.
+ */
+static struct lock_class_key ct_entries_ht_lock_key;
+
+static struct mlx5_ct_ft *
+mlx5_tc_ct_add_ft_cb(struct mlx5_tc_ct_priv *ct_priv, u16 zone,
+		     struct nf_flowtable *nf_ft)
+{
+	struct mlx5_ct_ft *ft;
+	int err;
+
+	ft = rhashtable_lookup_fast(&ct_priv->zone_ht, &zone, zone_params);
+	if (ft) {
+		refcount_inc(&ft->refcount);
+		return ft;
+	}
+
+	ft = kzalloc(sizeof(*ft), GFP_KERNEL);
+	if (!ft)
+		return ERR_PTR(-ENOMEM);
+
+	err = mapping_add(ct_priv->zone_mapping, &zone, &ft->zone_restore_id);
+	if (err)
+		goto err_mapping;
+
+	ft->zone = zone;
+	ft->nf_ft = nf_ft;
+	ft->ct_priv = ct_priv;
+	refcount_set(&ft->refcount, 1);
+
+	err = mlx5_tc_ct_alloc_pre_ct_tables(ft);
+	if (err)
+		goto err_alloc_pre_ct;
+
+	err = rhashtable_init(&ft->ct_entries_ht, &cts_ht_params);
+	if (err)
+		goto err_init;
+
+	lockdep_set_class(&ft->ct_entries_ht.mutex, &ct_entries_ht_lock_key);
+
+	err = rhashtable_insert_fast(&ct_priv->zone_ht, &ft->node,
+				     zone_params);
+	if (err)
+		goto err_insert;
+
+	err = nf_flow_table_offload_add_cb(ft->nf_ft,
+					   mlx5_tc_ct_block_flow_offload, ft);
+	if (err)
+		goto err_add_cb;
+
+	return ft;
+
+err_add_cb:
+	rhashtable_remove_fast(&ct_priv->zone_ht, &ft->node, zone_params);
+err_insert:
+	rhashtable_destroy(&ft->ct_entries_ht);
+err_init:
+	mlx5_tc_ct_free_pre_ct_tables(ft);
+err_alloc_pre_ct:
+	mapping_remove(ct_priv->zone_mapping, ft->zone_restore_id);
+err_mapping:
+	kfree(ft);
+	return ERR_PTR(err);
+}
+
+static void
+mlx5_tc_ct_flush_ft_entry(void *ptr, void *arg)
+{
+	struct mlx5_ct_entry *entry = ptr;
+
+	mlx5_tc_ct_entry_put(entry);
+}
+
+static void
+mlx5_tc_ct_del_ft_cb(struct mlx5_tc_ct_priv *ct_priv, struct mlx5_ct_ft *ft)
+{
+	if (!refcount_dec_and_test(&ft->refcount))
+		return;
+
+	flush_workqueue(ct_priv->wq);
+	nf_flow_table_offload_del_cb(ft->nf_ft,
+				     mlx5_tc_ct_block_flow_offload, ft);
+	rhashtable_remove_fast(&ct_priv->zone_ht, &ft->node, zone_params);
+	rhashtable_free_and_destroy(&ft->ct_entries_ht,
+				    mlx5_tc_ct_flush_ft_entry,
+				    ct_priv);
+	mlx5_tc_ct_free_pre_ct_tables(ft);
+	mapping_remove(ct_priv->zone_mapping, ft->zone_restore_id);
+	kfree(ft);
+}
+
+/* We translate the tc filter with CT action to the following HW model:
+ *
+ * +---------------------+
+ * + ft prio (tc chain)  +
+ * + original match      +
+ * +---------------------+
+ *      | set chain miss mapping
+ *      | set fte_id
+ *      | set tunnel_id
+ *      | do decap
+ *      v
+ * +---------------------+
+ * + pre_ct/pre_ct_nat   +  if matches     +-------------------------+
+ * + zone+nat match      +---------------->+ post_act (see below)    +
+ * +---------------------+  set zone       +-------------------------+
+ *      | set zone
+ *      v
+ * +--------------------+
+ * + CT (nat or no nat) +
+ * + tuple + zone match +
+ * +--------------------+
+ *      | set mark
+ *      | set labels_id
+ *      | set established
+ *	| set zone_restore
+ *      | do nat (if needed)
+ *      v
+ * +--------------+
+ * + post_act     + original filter actions
+ * + fte_id match +------------------------>
+ * +--------------+
+ */
+static struct mlx5_flow_handle *
+__mlx5_tc_ct_flow_offload(struct mlx5_tc_ct_priv *ct_priv,
+			  struct mlx5_flow_spec *orig_spec,
+			  struct mlx5_flow_attr *attr)
+{
+	bool nat = attr->ct_attr.ct_action & TCA_CT_ACT_NAT;
+	struct mlx5e_priv *priv = netdev_priv(ct_priv->netdev);
+	struct mlx5e_tc_mod_hdr_acts *pre_mod_acts;
+	u32 attr_sz = ns_to_attr_sz(ct_priv->ns_type);
+	struct mlx5_flow_attr *pre_ct_attr;
+	struct mlx5_modify_hdr *mod_hdr;
+	struct mlx5_ct_flow *ct_flow;
+	int chain_mapping = 0, err;
+	struct mlx5_ct_ft *ft;
+
+	ct_flow = kzalloc(sizeof(*ct_flow), GFP_KERNEL);
+	if (!ct_flow) {
+		return ERR_PTR(-ENOMEM);
+	}
+
+	/* Register for CT established events */
+	ft = mlx5_tc_ct_add_ft_cb(ct_priv, attr->ct_attr.zone,
+				  attr->ct_attr.nf_ft);
+	if (IS_ERR(ft)) {
+		err = PTR_ERR(ft);
+		ct_dbg("Failed to register to ft callback");
+		goto err_ft;
+	}
+	ct_flow->ft = ft;
+
+	/* Base flow attributes of both rules on original rule attribute */
+	ct_flow->pre_ct_attr = mlx5_alloc_flow_attr(ct_priv->ns_type);
+	if (!ct_flow->pre_ct_attr) {
+		err = -ENOMEM;
+		goto err_alloc_pre;
+	}
+
+	pre_ct_attr = ct_flow->pre_ct_attr;
+	memcpy(pre_ct_attr, attr, attr_sz);
+	pre_mod_acts = &pre_ct_attr->parse_attr->mod_hdr_acts;
+
+	/* Modify the original rule's action to fwd and modify, leave decap */
+	pre_ct_attr->action = attr->action & MLX5_FLOW_CONTEXT_ACTION_DECAP;
+	pre_ct_attr->action |= MLX5_FLOW_CONTEXT_ACTION_FWD_DEST |
+			       MLX5_FLOW_CONTEXT_ACTION_MOD_HDR;
+
+	/* Write chain miss tag for miss in ct table as we
+	 * don't go though all prios of this chain as normal tc rules
+	 * miss.
+	 */
+	err = mlx5_chains_get_chain_mapping(ct_priv->chains, attr->chain,
+					    &chain_mapping);
+	if (err) {
+		ct_dbg("Failed to get chain register mapping for chain");
+		goto err_get_chain;
+	}
+	ct_flow->chain_mapping = chain_mapping;
+
+	err = mlx5e_tc_match_to_reg_set(priv->mdev, pre_mod_acts, ct_priv->ns_type,
+					CHAIN_TO_REG, chain_mapping);
+	if (err) {
+		ct_dbg("Failed to set chain register mapping");
+		goto err_mapping;
+	}
+
+	/* If original flow is decap, we do it before going into ct table
+	 * so add a rewrite for the tunnel match_id.
+	 */
+	if ((pre_ct_attr->action & MLX5_FLOW_CONTEXT_ACTION_DECAP) &&
+	    attr->chain == 0) {
+		err = mlx5e_tc_match_to_reg_set(priv->mdev, pre_mod_acts,
+						ct_priv->ns_type,
+						TUNNEL_TO_REG,
+						attr->tunnel_id);
+		if (err) {
+			ct_dbg("Failed to set tunnel register mapping");
+			goto err_mapping;
+		}
+	}
+
+	mod_hdr = mlx5_modify_header_alloc(priv->mdev, ct_priv->ns_type,
+					   pre_mod_acts->num_actions,
+					   pre_mod_acts->actions);
+	if (IS_ERR(mod_hdr)) {
+		err = PTR_ERR(mod_hdr);
+		ct_dbg("Failed to create pre ct mod hdr");
+		goto err_mapping;
+	}
+	pre_ct_attr->modify_hdr = mod_hdr;
+
+	/* Change original rule point to ct table */
+	pre_ct_attr->dest_chain = 0;
+	pre_ct_attr->dest_ft = nat ? ft->pre_ct_nat.ft : ft->pre_ct.ft;
+	ct_flow->pre_ct_rule = mlx5_tc_rule_insert(priv, orig_spec,
+						   pre_ct_attr);
+	if (IS_ERR(ct_flow->pre_ct_rule)) {
+		err = PTR_ERR(ct_flow->pre_ct_rule);
+		ct_dbg("Failed to add pre ct rule");
+		goto err_insert_orig;
+	}
+
+	attr->ct_attr.ct_flow = ct_flow;
+	mlx5e_mod_hdr_dealloc(pre_mod_acts);
+
+	return ct_flow->pre_ct_rule;
+
+err_insert_orig:
+	mlx5_modify_header_dealloc(priv->mdev, pre_ct_attr->modify_hdr);
+err_mapping:
+	mlx5e_mod_hdr_dealloc(pre_mod_acts);
+	mlx5_chains_put_chain_mapping(ct_priv->chains, ct_flow->chain_mapping);
+err_get_chain:
+	kfree(ct_flow->pre_ct_attr);
+err_alloc_pre:
+	mlx5_tc_ct_del_ft_cb(ct_priv, ft);
+err_ft:
+	kfree(ct_flow);
+	netdev_warn(priv->netdev, "Failed to offload ct flow, err %d\n", err);
+	return ERR_PTR(err);
+}
+
+struct mlx5_flow_handle *
+mlx5_tc_ct_flow_offload(struct mlx5_tc_ct_priv *priv,
+			struct mlx5_flow_spec *spec,
+			struct mlx5_flow_attr *attr,
+			struct mlx5e_tc_mod_hdr_acts *mod_hdr_acts)
+{
+	struct mlx5_flow_handle *rule;
+
+	if (!priv)
+		return ERR_PTR(-EOPNOTSUPP);
+
+	mutex_lock(&priv->control_lock);
+	rule = __mlx5_tc_ct_flow_offload(priv, spec, attr);
+	mutex_unlock(&priv->control_lock);
+
+	return rule;
+}
+
+static void
+__mlx5_tc_ct_delete_flow(struct mlx5_tc_ct_priv *ct_priv,
+			 struct mlx5_ct_flow *ct_flow,
+			 struct mlx5_flow_attr *attr)
+{
+	struct mlx5_flow_attr *pre_ct_attr = ct_flow->pre_ct_attr;
+	struct mlx5e_priv *priv = netdev_priv(ct_priv->netdev);
+
+	mlx5_tc_rule_delete(priv, ct_flow->pre_ct_rule, pre_ct_attr);
+	mlx5_modify_header_dealloc(priv->mdev, pre_ct_attr->modify_hdr);
+
+	mlx5_chains_put_chain_mapping(ct_priv->chains, ct_flow->chain_mapping);
+	mlx5_tc_ct_del_ft_cb(ct_priv, ct_flow->ft);
+
+	kfree(ct_flow->pre_ct_attr);
+	kfree(ct_flow);
+}
+
+void
+mlx5_tc_ct_delete_flow(struct mlx5_tc_ct_priv *priv,
+		       struct mlx5_flow_attr *attr)
+{
+	struct mlx5_ct_flow *ct_flow = attr->ct_attr.ct_flow;
+
+	/* We are called on error to clean up stuff from parsing
+	 * but we don't have anything for now
+	 */
+	if (!ct_flow)
+		return;
+
+	mutex_lock(&priv->control_lock);
+	__mlx5_tc_ct_delete_flow(priv, ct_flow, attr);
+	mutex_unlock(&priv->control_lock);
+}
+
+static int
+mlx5_tc_ct_fs_init(struct mlx5_tc_ct_priv *ct_priv)
+{
+	struct mlx5_flow_table *post_ct = mlx5e_tc_post_act_get_ft(ct_priv->post_act);
+	struct mlx5_ct_fs_ops *fs_ops = mlx5_ct_fs_dmfs_ops_get();
+	int err;
+
+	if (ct_priv->ns_type == MLX5_FLOW_NAMESPACE_FDB &&
+	    ct_priv->dev->priv.steering->mode == MLX5_FLOW_STEERING_MODE_SMFS) {
+		ct_dbg("Using SMFS ct flow steering provider");
+		fs_ops = mlx5_ct_fs_smfs_ops_get();
+	}
+
+	ct_priv->fs = kzalloc(sizeof(*ct_priv->fs) + fs_ops->priv_size, GFP_KERNEL);
+	if (!ct_priv->fs)
+		return -ENOMEM;
+
+	ct_priv->fs->netdev = ct_priv->netdev;
+	ct_priv->fs->dev = ct_priv->dev;
+	ct_priv->fs_ops = fs_ops;
+
+	err = ct_priv->fs_ops->init(ct_priv->fs, ct_priv->ct, ct_priv->ct_nat, post_ct);
+	if (err)
+		goto err_init;
+
+	return 0;
+
+err_init:
+	kfree(ct_priv->fs);
+	return err;
+}
+
+static int
+mlx5_tc_ct_init_check_esw_support(struct mlx5_eswitch *esw,
+				  const char **err_msg)
+{
+	if (!mlx5_eswitch_vlan_actions_supported(esw->dev, 1)) {
+		/* vlan workaround should be avoided for multi chain rules.
+		 * This is just a sanity check as pop vlan action should
+		 * be supported by any FW that supports ignore_flow_level
+		 */
+
+		*err_msg = "firmware vlan actions support is missing";
+		return -EOPNOTSUPP;
+	}
+
+	if (!MLX5_CAP_ESW_FLOWTABLE(esw->dev,
+				    fdb_modify_header_fwd_to_table)) {
+		/* CT always writes to registers which are mod header actions.
+		 * Therefore, mod header and goto is required
+		 */
+
+		*err_msg = "firmware fwd and modify support is missing";
+		return -EOPNOTSUPP;
+	}
+
+	if (!mlx5_eswitch_reg_c1_loopback_enabled(esw)) {
+		*err_msg = "register loopback isn't supported";
+		return -EOPNOTSUPP;
+	}
+
+	return 0;
+}
+
+static int
+mlx5_tc_ct_init_check_support(struct mlx5e_priv *priv,
+			      enum mlx5_flow_namespace_type ns_type,
+			      struct mlx5e_post_act *post_act)
+{
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	const char *err_msg = NULL;
+	int err = 0;
+
+#if !IS_ENABLED(CONFIG_NET_TC_SKB_EXT)
+	/* cannot restore chain ID on HW miss */
+
+	err_msg = "tc skb extension missing";
+	err = -EOPNOTSUPP;
+	goto out_err;
+#endif
+	if (IS_ERR_OR_NULL(post_act)) {
+		/* Ignore_flow_level support isn't supported by default for VFs and so post_act
+		 * won't be supported. Skip showing error msg.
+		 */
+		if (priv->mdev->coredev_type == MLX5_COREDEV_PF)
+			err_msg = "post action is missing";
+		err = -EOPNOTSUPP;
+		goto out_err;
+	}
+
+	if (ns_type == MLX5_FLOW_NAMESPACE_FDB)
+		err = mlx5_tc_ct_init_check_esw_support(esw, &err_msg);
+
+out_err:
+	if (err && err_msg)
+		netdev_dbg(priv->netdev, "tc ct offload not supported, %s\n", err_msg);
+	return err;
+}
+
+static void
+mlx5_ct_tc_create_dbgfs(struct mlx5_tc_ct_priv *ct_priv)
+{
+	struct mlx5_tc_ct_debugfs *ct_dbgfs = &ct_priv->debugfs;
+
+	ct_dbgfs->root = debugfs_create_dir("ct", mlx5_debugfs_get_dev_root(ct_priv->dev));
+	debugfs_create_atomic_t("offloaded", 0400, ct_dbgfs->root,
+				&ct_dbgfs->stats.offloaded);
+	debugfs_create_atomic_t("rx_dropped", 0400, ct_dbgfs->root,
+				&ct_dbgfs->stats.rx_dropped);
+}
+
+static void
+mlx5_ct_tc_remove_dbgfs(struct mlx5_tc_ct_priv *ct_priv)
+{
+	debugfs_remove_recursive(ct_priv->debugfs.root);
+}
+
+#define INIT_ERR_PREFIX "tc ct offload init failed"
+
+struct mlx5_tc_ct_priv *
+mlx5_tc_ct_init(struct mlx5e_priv *priv, struct mlx5_fs_chains *chains,
+		struct mod_hdr_tbl *mod_hdr,
+		enum mlx5_flow_namespace_type ns_type,
+		struct mlx5e_post_act *post_act)
+{
+	struct mlx5_tc_ct_priv *ct_priv;
+	struct mlx5_core_dev *dev;
+	u64 mapping_id;
+	int err;
+
+	dev = priv->mdev;
+	err = mlx5_tc_ct_init_check_support(priv, ns_type, post_act);
+	if (err)
+		goto err_support;
+
+	ct_priv = kzalloc(sizeof(*ct_priv), GFP_KERNEL);
+	if (!ct_priv)
+		goto err_alloc;
+
+	mapping_id = mlx5_query_nic_system_image_guid(dev);
+
+	ct_priv->zone_mapping = mapping_create_for_id(mapping_id, MAPPING_TYPE_ZONE,
+						      sizeof(u16), 0, true);
+	if (IS_ERR(ct_priv->zone_mapping)) {
+		err = PTR_ERR(ct_priv->zone_mapping);
+		goto err_mapping_zone;
+	}
+
+	ct_priv->labels_mapping = mapping_create_for_id(mapping_id, MAPPING_TYPE_LABELS,
+							sizeof(u32) * 4, 0, true);
+	if (IS_ERR(ct_priv->labels_mapping)) {
+		err = PTR_ERR(ct_priv->labels_mapping);
+		goto err_mapping_labels;
+	}
+
+	spin_lock_init(&ct_priv->ht_lock);
+	ct_priv->ns_type = ns_type;
+	ct_priv->chains = chains;
+	ct_priv->netdev = priv->netdev;
+	ct_priv->dev = priv->mdev;
+	ct_priv->mod_hdr_tbl = mod_hdr;
+	ct_priv->ct = mlx5_chains_create_global_table(chains);
+	if (IS_ERR(ct_priv->ct)) {
+		err = PTR_ERR(ct_priv->ct);
+		mlx5_core_warn(dev,
+			       "%s, failed to create ct table err: %d\n",
+			       INIT_ERR_PREFIX, err);
+		goto err_ct_tbl;
+	}
+
+	ct_priv->ct_nat = mlx5_chains_create_global_table(chains);
+	if (IS_ERR(ct_priv->ct_nat)) {
+		err = PTR_ERR(ct_priv->ct_nat);
+		mlx5_core_warn(dev,
+			       "%s, failed to create ct nat table err: %d\n",
+			       INIT_ERR_PREFIX, err);
+		goto err_ct_nat_tbl;
+	}
+
+	ct_priv->post_act = post_act;
+	mutex_init(&ct_priv->control_lock);
+	if (rhashtable_init(&ct_priv->zone_ht, &zone_params))
+		goto err_ct_zone_ht;
+	if (rhashtable_init(&ct_priv->ct_tuples_ht, &tuples_ht_params))
+		goto err_ct_tuples_ht;
+	if (rhashtable_init(&ct_priv->ct_tuples_nat_ht, &tuples_nat_ht_params))
+		goto err_ct_tuples_nat_ht;
+
+	ct_priv->wq = alloc_ordered_workqueue("mlx5e_ct_priv_wq", 0);
+	if (!ct_priv->wq) {
+		err = -ENOMEM;
+		goto err_wq;
+	}
+
+	err = mlx5_tc_ct_fs_init(ct_priv);
+	if (err)
+		goto err_init_fs;
+
+	mlx5_ct_tc_create_dbgfs(ct_priv);
+	return ct_priv;
+
+err_init_fs:
+	destroy_workqueue(ct_priv->wq);
+err_wq:
+	rhashtable_destroy(&ct_priv->ct_tuples_nat_ht);
+err_ct_tuples_nat_ht:
+	rhashtable_destroy(&ct_priv->ct_tuples_ht);
+err_ct_tuples_ht:
+	rhashtable_destroy(&ct_priv->zone_ht);
+err_ct_zone_ht:
+	mlx5_chains_destroy_global_table(chains, ct_priv->ct_nat);
+err_ct_nat_tbl:
+	mlx5_chains_destroy_global_table(chains, ct_priv->ct);
+err_ct_tbl:
+	mapping_destroy(ct_priv->labels_mapping);
+err_mapping_labels:
+	mapping_destroy(ct_priv->zone_mapping);
+err_mapping_zone:
+	kfree(ct_priv);
+err_alloc:
+err_support:
+
+	return NULL;
+}
+
+void
+mlx5_tc_ct_clean(struct mlx5_tc_ct_priv *ct_priv)
+{
+	struct mlx5_fs_chains *chains;
+
+	if (!ct_priv)
+		return;
+
+	destroy_workqueue(ct_priv->wq);
+	mlx5_ct_tc_remove_dbgfs(ct_priv);
+	chains = ct_priv->chains;
+
+	ct_priv->fs_ops->destroy(ct_priv->fs);
+	kfree(ct_priv->fs);
+
+	mlx5_chains_destroy_global_table(chains, ct_priv->ct_nat);
+	mlx5_chains_destroy_global_table(chains, ct_priv->ct);
+	mapping_destroy(ct_priv->zone_mapping);
+	mapping_destroy(ct_priv->labels_mapping);
+
+	rhashtable_destroy(&ct_priv->ct_tuples_ht);
+	rhashtable_destroy(&ct_priv->ct_tuples_nat_ht);
+	rhashtable_destroy(&ct_priv->zone_ht);
+	mutex_destroy(&ct_priv->control_lock);
+	kfree(ct_priv);
+}
+
+bool
+mlx5e_tc_ct_restore_flow(struct mlx5_tc_ct_priv *ct_priv,
+			 struct sk_buff *skb, u8 zone_restore_id)
+{
+	struct mlx5_ct_tuple tuple = {};
+	struct mlx5_ct_entry *entry;
+	u16 zone;
+
+	if (!ct_priv || !zone_restore_id)
+		return true;
+
+	if (mapping_find(ct_priv->zone_mapping, zone_restore_id, &zone))
+		goto out_inc_drop;
+
+	if (!mlx5_tc_ct_skb_to_tuple(skb, &tuple, zone))
+		goto out_inc_drop;
+
+	spin_lock(&ct_priv->ht_lock);
+
+	entry = mlx5_tc_ct_entry_get(ct_priv, &tuple);
+	if (!entry) {
+		spin_unlock(&ct_priv->ht_lock);
+		goto out_inc_drop;
+	}
+
+	if (IS_ERR(entry)) {
+		spin_unlock(&ct_priv->ht_lock);
+		goto out_inc_drop;
+	}
+	spin_unlock(&ct_priv->ht_lock);
+
+	tcf_ct_flow_table_restore_skb(skb, entry->restore_cookie);
+	__mlx5_tc_ct_entry_put(entry);
+
+	return true;
+
+out_inc_drop:
+	atomic_inc(&ct_priv->debugfs.stats.rx_dropped);
+	return false;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc_ct.h b/drivers/net/ethernet/mellanox/mlx5/core/en/tc_ct.h
new file mode 100644
index 000000000..5bbd6b928
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc_ct.h
@@ -0,0 +1,220 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2018 Mellanox Technologies. */
+
+#ifndef __MLX5_EN_TC_CT_H__
+#define __MLX5_EN_TC_CT_H__
+
+#include <net/pkt_cls.h>
+#include <linux/mlx5/fs.h>
+#include <net/tc_act/tc_ct.h>
+
+#include "en.h"
+
+struct mlx5_flow_attr;
+struct mlx5e_tc_mod_hdr_acts;
+struct mlx5_rep_uplink_priv;
+struct mlx5e_tc_flow;
+struct mlx5e_priv;
+
+struct mlx5_fs_chains;
+struct mlx5_tc_ct_priv;
+struct mlx5_ct_flow;
+
+struct nf_flowtable;
+
+struct mlx5_ct_attr {
+	u16 zone;
+	u16 ct_action;
+	struct mlx5_ct_flow *ct_flow;
+	struct nf_flowtable *nf_ft;
+	u32 ct_labels_id;
+};
+
+#define zone_to_reg_ct {\
+	.mfield = MLX5_ACTION_IN_FIELD_METADATA_REG_C_2,\
+	.moffset = 0,\
+	.mlen = 16,\
+	.soffset = MLX5_BYTE_OFF(fte_match_param,\
+				 misc_parameters_2.metadata_reg_c_2),\
+}
+
+#define ctstate_to_reg_ct {\
+	.mfield = MLX5_ACTION_IN_FIELD_METADATA_REG_C_2,\
+	.moffset = 16,\
+	.mlen = 16,\
+	.soffset = MLX5_BYTE_OFF(fte_match_param,\
+				 misc_parameters_2.metadata_reg_c_2),\
+}
+
+#define mark_to_reg_ct {\
+	.mfield = MLX5_ACTION_IN_FIELD_METADATA_REG_C_3,\
+	.moffset = 0,\
+	.mlen = 32,\
+	.soffset = MLX5_BYTE_OFF(fte_match_param,\
+				 misc_parameters_2.metadata_reg_c_3),\
+}
+
+#define labels_to_reg_ct {\
+	.mfield = MLX5_ACTION_IN_FIELD_METADATA_REG_C_4,\
+	.moffset = 0,\
+	.mlen = 32,\
+	.soffset = MLX5_BYTE_OFF(fte_match_param,\
+				 misc_parameters_2.metadata_reg_c_4),\
+}
+
+/* 8 LSB of metadata C5 are reserved for packet color */
+#define fteid_to_reg_ct {\
+	.mfield = MLX5_ACTION_IN_FIELD_METADATA_REG_C_5,\
+	.moffset = 8,\
+	.mlen = 24,\
+	.soffset = MLX5_BYTE_OFF(fte_match_param,\
+				 misc_parameters_2.metadata_reg_c_5),\
+}
+
+#define zone_restore_to_reg_ct {\
+	.mfield = MLX5_ACTION_IN_FIELD_METADATA_REG_C_1,\
+	.moffset = 0,\
+	.mlen = ESW_ZONE_ID_BITS,\
+	.soffset = MLX5_BYTE_OFF(fte_match_param,\
+				 misc_parameters_2.metadata_reg_c_1),\
+}
+
+#define nic_zone_restore_to_reg_ct {\
+	.mfield = MLX5_ACTION_IN_FIELD_METADATA_REG_B,\
+	.moffset = 16,\
+	.mlen = ESW_ZONE_ID_BITS,\
+}
+
+#define MLX5_CT_ZONE_BITS MLX5_REG_MAPPING_MBITS(ZONE_TO_REG)
+#define MLX5_CT_ZONE_MASK MLX5_REG_MAPPING_MASK(ZONE_TO_REG)
+
+#if IS_ENABLED(CONFIG_MLX5_TC_CT)
+
+struct mlx5_tc_ct_priv *
+mlx5_tc_ct_init(struct mlx5e_priv *priv, struct mlx5_fs_chains *chains,
+		struct mod_hdr_tbl *mod_hdr,
+		enum mlx5_flow_namespace_type ns_type,
+		struct mlx5e_post_act *post_act);
+void
+mlx5_tc_ct_clean(struct mlx5_tc_ct_priv *ct_priv);
+
+void
+mlx5_tc_ct_match_del(struct mlx5_tc_ct_priv *priv, struct mlx5_ct_attr *ct_attr);
+
+int
+mlx5_tc_ct_match_add(struct mlx5_tc_ct_priv *priv,
+		     struct mlx5_flow_spec *spec,
+		     struct flow_cls_offload *f,
+		     struct mlx5_ct_attr *ct_attr,
+		     struct netlink_ext_ack *extack);
+int mlx5_tc_ct_add_no_trk_match(struct mlx5_flow_spec *spec);
+int
+mlx5_tc_ct_parse_action(struct mlx5_tc_ct_priv *priv,
+			struct mlx5_flow_attr *attr,
+			struct mlx5e_tc_mod_hdr_acts *mod_acts,
+			const struct flow_action_entry *act,
+			struct netlink_ext_ack *extack);
+
+struct mlx5_flow_handle *
+mlx5_tc_ct_flow_offload(struct mlx5_tc_ct_priv *priv,
+			struct mlx5_flow_spec *spec,
+			struct mlx5_flow_attr *attr,
+			struct mlx5e_tc_mod_hdr_acts *mod_hdr_acts);
+void
+mlx5_tc_ct_delete_flow(struct mlx5_tc_ct_priv *priv,
+		       struct mlx5_flow_attr *attr);
+
+bool
+mlx5e_tc_ct_restore_flow(struct mlx5_tc_ct_priv *ct_priv,
+			 struct sk_buff *skb, u8 zone_restore_id);
+
+int
+mlx5_tc_ct_set_ct_clear_regs(struct mlx5_tc_ct_priv *priv,
+			     struct mlx5e_tc_mod_hdr_acts *mod_acts);
+
+#else /* CONFIG_MLX5_TC_CT */
+
+static inline struct mlx5_tc_ct_priv *
+mlx5_tc_ct_init(struct mlx5e_priv *priv, struct mlx5_fs_chains *chains,
+		struct mod_hdr_tbl *mod_hdr,
+		enum mlx5_flow_namespace_type ns_type,
+		struct mlx5e_post_act *post_act)
+{
+	return NULL;
+}
+
+static inline void
+mlx5_tc_ct_clean(struct mlx5_tc_ct_priv *ct_priv)
+{
+}
+
+static inline void
+mlx5_tc_ct_match_del(struct mlx5_tc_ct_priv *priv, struct mlx5_ct_attr *ct_attr) {}
+
+static inline int
+mlx5_tc_ct_match_add(struct mlx5_tc_ct_priv *priv,
+		     struct mlx5_flow_spec *spec,
+		     struct flow_cls_offload *f,
+		     struct mlx5_ct_attr *ct_attr,
+		     struct netlink_ext_ack *extack)
+{
+	struct flow_rule *rule = flow_cls_offload_flow_rule(f);
+
+	if (!flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_CT))
+		return 0;
+
+	NL_SET_ERR_MSG_MOD(extack, "mlx5 tc ct offload isn't enabled.");
+	return -EOPNOTSUPP;
+}
+
+static inline int
+mlx5_tc_ct_add_no_trk_match(struct mlx5_flow_spec *spec)
+{
+	return 0;
+}
+
+static inline int
+mlx5_tc_ct_set_ct_clear_regs(struct mlx5_tc_ct_priv *priv,
+			     struct mlx5e_tc_mod_hdr_acts *mod_acts)
+{
+	return -EOPNOTSUPP;
+}
+
+static inline int
+mlx5_tc_ct_parse_action(struct mlx5_tc_ct_priv *priv,
+			struct mlx5_flow_attr *attr,
+			struct mlx5e_tc_mod_hdr_acts *mod_acts,
+			const struct flow_action_entry *act,
+			struct netlink_ext_ack *extack)
+{
+	NL_SET_ERR_MSG_MOD(extack, "mlx5 tc ct offload isn't enabled.");
+	return -EOPNOTSUPP;
+}
+
+static inline struct mlx5_flow_handle *
+mlx5_tc_ct_flow_offload(struct mlx5_tc_ct_priv *priv,
+			struct mlx5_flow_spec *spec,
+			struct mlx5_flow_attr *attr,
+			struct mlx5e_tc_mod_hdr_acts *mod_hdr_acts)
+{
+	return ERR_PTR(-EOPNOTSUPP);
+}
+
+static inline void
+mlx5_tc_ct_delete_flow(struct mlx5_tc_ct_priv *priv,
+		       struct mlx5_flow_attr *attr)
+{
+}
+
+static inline bool
+mlx5e_tc_ct_restore_flow(struct mlx5_tc_ct_priv *ct_priv,
+			 struct sk_buff *skb, u8 zone_restore_id)
+{
+	if (!zone_restore_id)
+		return true;
+
+	return false;
+}
+
+#endif /* !IS_ENABLED(CONFIG_MLX5_TC_CT) */
+#endif /* __MLX5_EN_TC_CT_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc_priv.h b/drivers/net/ethernet/mellanox/mlx5/core/en/tc_priv.h
new file mode 100644
index 000000000..2e42d7c54
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc_priv.h
@@ -0,0 +1,218 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2021 Mellanox Technologies. */
+
+#ifndef __MLX5_EN_TC_PRIV_H__
+#define __MLX5_EN_TC_PRIV_H__
+
+#include "en_tc.h"
+#include "en/tc/act/act.h"
+
+#define MLX5E_TC_FLOW_BASE (MLX5E_TC_FLAG_LAST_EXPORTED_BIT + 1)
+
+#define MLX5E_TC_MAX_SPLITS 1
+
+
+enum {
+	MLX5E_TC_FLOW_FLAG_INGRESS               = MLX5E_TC_FLAG_INGRESS_BIT,
+	MLX5E_TC_FLOW_FLAG_EGRESS                = MLX5E_TC_FLAG_EGRESS_BIT,
+	MLX5E_TC_FLOW_FLAG_ESWITCH               = MLX5E_TC_FLAG_ESW_OFFLOAD_BIT,
+	MLX5E_TC_FLOW_FLAG_FT                    = MLX5E_TC_FLAG_FT_OFFLOAD_BIT,
+	MLX5E_TC_FLOW_FLAG_NIC                   = MLX5E_TC_FLAG_NIC_OFFLOAD_BIT,
+	MLX5E_TC_FLOW_FLAG_OFFLOADED             = MLX5E_TC_FLOW_BASE,
+	MLX5E_TC_FLOW_FLAG_HAIRPIN               = MLX5E_TC_FLOW_BASE + 1,
+	MLX5E_TC_FLOW_FLAG_HAIRPIN_RSS           = MLX5E_TC_FLOW_BASE + 2,
+	MLX5E_TC_FLOW_FLAG_SLOW                  = MLX5E_TC_FLOW_BASE + 3,
+	MLX5E_TC_FLOW_FLAG_DUP                   = MLX5E_TC_FLOW_BASE + 4,
+	MLX5E_TC_FLOW_FLAG_NOT_READY             = MLX5E_TC_FLOW_BASE + 5,
+	MLX5E_TC_FLOW_FLAG_DELETED               = MLX5E_TC_FLOW_BASE + 6,
+	MLX5E_TC_FLOW_FLAG_CT                    = MLX5E_TC_FLOW_BASE + 7,
+	MLX5E_TC_FLOW_FLAG_L3_TO_L2_DECAP        = MLX5E_TC_FLOW_BASE + 8,
+	MLX5E_TC_FLOW_FLAG_TUN_RX                = MLX5E_TC_FLOW_BASE + 9,
+	MLX5E_TC_FLOW_FLAG_FAILED                = MLX5E_TC_FLOW_BASE + 10,
+	MLX5E_TC_FLOW_FLAG_SAMPLE                = MLX5E_TC_FLOW_BASE + 11,
+};
+
+struct mlx5e_tc_flow_parse_attr {
+	const struct ip_tunnel_info *tun_info[MLX5_MAX_FLOW_FWD_VPORTS];
+	struct mlx5e_mpls_info mpls_info[MLX5_MAX_FLOW_FWD_VPORTS];
+	struct net_device *filter_dev;
+	struct mlx5_flow_spec spec;
+	struct pedit_headers_action hdrs[__PEDIT_CMD_MAX];
+	struct mlx5e_tc_mod_hdr_acts mod_hdr_acts;
+	int mirred_ifindex[MLX5_MAX_FLOW_FWD_VPORTS];
+	struct mlx5e_tc_act_parse_state parse_state;
+};
+
+struct mlx5_fs_chains *mlx5e_nic_chains(struct mlx5e_tc_table *tc);
+
+/* Helper struct for accessing a struct containing list_head array.
+ * Containing struct
+ *   |- Helper array
+ *      [0] Helper item 0
+ *          |- list_head item 0
+ *          |- index (0)
+ *      [1] Helper item 1
+ *          |- list_head item 1
+ *          |- index (1)
+ * To access the containing struct from one of the list_head items:
+ * 1. Get the helper item from the list_head item using
+ *    helper item =
+ *        container_of(list_head item, helper struct type, list_head field)
+ * 2. Get the contining struct from the helper item and its index in the array:
+ *    containing struct =
+ *        container_of(helper item, containing struct type, helper field[index])
+ */
+struct encap_flow_item {
+	struct mlx5e_encap_entry *e; /* attached encap instance */
+	struct list_head list;
+	int index;
+};
+
+struct encap_route_flow_item {
+	struct mlx5e_route_entry *r; /* attached route instance */
+	int index;
+};
+
+struct mlx5e_tc_flow {
+	struct rhash_head node;
+	struct mlx5e_priv *priv;
+	u64 cookie;
+	unsigned long flags;
+	struct mlx5_flow_handle *rule[MLX5E_TC_MAX_SPLITS + 1];
+
+	/* flows sharing the same reformat object - currently mpls decap */
+	struct list_head l3_to_l2_reformat;
+	struct mlx5e_decap_entry *decap_reformat;
+
+	/* flows sharing same route entry */
+	struct list_head decap_routes;
+	struct mlx5e_route_entry *decap_route;
+	struct encap_route_flow_item encap_routes[MLX5_MAX_FLOW_FWD_VPORTS];
+
+	/* Flow can be associated with multiple encap IDs.
+	 * The number of encaps is bounded by the number of supported
+	 * destinations.
+	 */
+	struct encap_flow_item encaps[MLX5_MAX_FLOW_FWD_VPORTS];
+	struct mlx5e_tc_flow *peer_flow;
+	struct mlx5e_mod_hdr_handle *mh; /* attached mod header instance */
+	struct mlx5e_mod_hdr_handle *slow_mh; /* attached mod header instance for slow path */
+	struct mlx5e_hairpin_entry *hpe; /* attached hairpin instance */
+	struct list_head hairpin; /* flows sharing the same hairpin */
+	struct list_head peer;    /* flows with peer flow */
+	struct list_head unready; /* flows not ready to be offloaded (e.g
+				   * due to missing route)
+				   */
+	struct net_device *orig_dev; /* netdev adding flow first */
+	int tmp_entry_index;
+	struct list_head tmp_list; /* temporary flow list used by neigh update */
+	refcount_t refcnt;
+	struct rcu_head rcu_head;
+	struct completion init_done;
+	struct completion del_hw_done;
+	struct mlx5_flow_attr *attr;
+	struct list_head attrs;
+	u32 chain_mapping;
+};
+
+struct mlx5_flow_handle *
+mlx5e_tc_rule_offload(struct mlx5e_priv *priv,
+		      struct mlx5_flow_spec *spec,
+		      struct mlx5_flow_attr *attr);
+
+void
+mlx5e_tc_rule_unoffload(struct mlx5e_priv *priv,
+			struct mlx5_flow_handle *rule,
+			struct mlx5_flow_attr *attr);
+
+u8 mlx5e_tc_get_ip_version(struct mlx5_flow_spec *spec, bool outer);
+
+struct mlx5_flow_handle *
+mlx5e_tc_offload_fdb_rules(struct mlx5_eswitch *esw,
+			   struct mlx5e_tc_flow *flow,
+			   struct mlx5_flow_spec *spec,
+			   struct mlx5_flow_attr *attr);
+
+struct mlx5_flow_attr *
+mlx5e_tc_get_encap_attr(struct mlx5e_tc_flow *flow);
+
+void mlx5e_tc_unoffload_flow_post_acts(struct mlx5e_tc_flow *flow);
+int mlx5e_tc_offload_flow_post_acts(struct mlx5e_tc_flow *flow);
+
+bool mlx5e_is_eswitch_flow(struct mlx5e_tc_flow *flow);
+bool mlx5e_is_ft_flow(struct mlx5e_tc_flow *flow);
+bool mlx5e_is_offloaded_flow(struct mlx5e_tc_flow *flow);
+int mlx5e_get_flow_namespace(struct mlx5e_tc_flow *flow);
+bool mlx5e_same_hw_devs(struct mlx5e_priv *priv, struct mlx5e_priv *peer_priv);
+
+static inline void __flow_flag_set(struct mlx5e_tc_flow *flow, unsigned long flag)
+{
+	/* Complete all memory stores before setting bit. */
+	smp_mb__before_atomic();
+	set_bit(flag, &flow->flags);
+}
+
+#define flow_flag_set(flow, flag) __flow_flag_set(flow, MLX5E_TC_FLOW_FLAG_##flag)
+
+static inline bool __flow_flag_test_and_set(struct mlx5e_tc_flow *flow,
+					    unsigned long flag)
+{
+	/* test_and_set_bit() provides all necessary barriers */
+	return test_and_set_bit(flag, &flow->flags);
+}
+
+#define flow_flag_test_and_set(flow, flag)			\
+	__flow_flag_test_and_set(flow,				\
+				 MLX5E_TC_FLOW_FLAG_##flag)
+
+static inline void __flow_flag_clear(struct mlx5e_tc_flow *flow, unsigned long flag)
+{
+	/* Complete all memory stores before clearing bit. */
+	smp_mb__before_atomic();
+	clear_bit(flag, &flow->flags);
+}
+
+#define flow_flag_clear(flow, flag) __flow_flag_clear(flow,		\
+						      MLX5E_TC_FLOW_FLAG_##flag)
+
+static inline bool __flow_flag_test(struct mlx5e_tc_flow *flow, unsigned long flag)
+{
+	bool ret = test_bit(flag, &flow->flags);
+
+	/* Read fields of flow structure only after checking flags. */
+	smp_mb__after_atomic();
+	return ret;
+}
+
+#define flow_flag_test(flow, flag) __flow_flag_test(flow,		\
+						    MLX5E_TC_FLOW_FLAG_##flag)
+
+void mlx5e_tc_unoffload_from_slow_path(struct mlx5_eswitch *esw,
+				       struct mlx5e_tc_flow *flow);
+struct mlx5_flow_handle *
+mlx5e_tc_offload_to_slow_path(struct mlx5_eswitch *esw,
+			      struct mlx5e_tc_flow *flow,
+			      struct mlx5_flow_spec *spec);
+
+void mlx5e_tc_unoffload_fdb_rules(struct mlx5_eswitch *esw,
+				  struct mlx5e_tc_flow *flow,
+				  struct mlx5_flow_attr *attr);
+
+struct mlx5e_tc_flow *mlx5e_flow_get(struct mlx5e_tc_flow *flow);
+void mlx5e_flow_put(struct mlx5e_priv *priv, struct mlx5e_tc_flow *flow);
+
+struct mlx5_fc *mlx5e_tc_get_counter(struct mlx5e_tc_flow *flow);
+
+struct mlx5e_tc_int_port_priv *
+mlx5e_get_int_port_priv(struct mlx5e_priv *priv);
+
+struct mlx5e_flow_meters *mlx5e_get_flow_meters(struct mlx5_core_dev *dev);
+
+void *mlx5e_get_match_headers_value(u32 flags, struct mlx5_flow_spec *spec);
+void *mlx5e_get_match_headers_criteria(u32 flags, struct mlx5_flow_spec *spec);
+
+int mlx5e_policer_validate(const struct flow_action *action,
+			   const struct flow_action_entry *act,
+			   struct netlink_ext_ack *extack);
+
+#endif /* __MLX5_EN_TC_PRIV_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun.c b/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun.c
new file mode 100644
index 000000000..83bb0811e
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun.c
@@ -0,0 +1,991 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2018 Mellanox Technologies. */
+
+#include <net/inet_ecn.h>
+#include <net/vxlan.h>
+#include <net/gre.h>
+#include <net/geneve.h>
+#include <net/bareudp.h>
+#include "en/tc_tun.h"
+#include "en/tc_priv.h"
+#include "en_tc.h"
+#include "rep/tc.h"
+#include "rep/neigh.h"
+#include "lag/lag.h"
+#include "lag/mp.h"
+
+struct mlx5e_tc_tun_route_attr {
+	struct net_device *out_dev;
+	struct net_device *route_dev;
+	union {
+		struct flowi4 fl4;
+		struct flowi6 fl6;
+	} fl;
+	struct neighbour *n;
+	u8 ttl;
+};
+
+#define TC_TUN_ROUTE_ATTR_INIT(name) struct mlx5e_tc_tun_route_attr name = {}
+
+static void mlx5e_tc_tun_route_attr_cleanup(struct mlx5e_tc_tun_route_attr *attr)
+{
+	if (attr->n)
+		neigh_release(attr->n);
+	if (attr->route_dev)
+		dev_put(attr->route_dev);
+}
+
+struct mlx5e_tc_tunnel *mlx5e_get_tc_tun(struct net_device *tunnel_dev)
+{
+	if (netif_is_vxlan(tunnel_dev))
+		return &vxlan_tunnel;
+	else if (netif_is_geneve(tunnel_dev))
+		return &geneve_tunnel;
+	else if (netif_is_gretap(tunnel_dev) ||
+		 netif_is_ip6gretap(tunnel_dev))
+		return &gre_tunnel;
+	else if (netif_is_bareudp(tunnel_dev))
+		return &mplsoudp_tunnel;
+	else
+		return NULL;
+}
+
+static int get_route_and_out_devs(struct mlx5e_priv *priv,
+				  struct net_device *dev,
+				  struct net_device **route_dev,
+				  struct net_device **out_dev)
+{
+	struct net_device *uplink_dev, *uplink_upper, *real_dev;
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	bool dst_is_lag_dev;
+
+	real_dev = is_vlan_dev(dev) ? vlan_dev_real_dev(dev) : dev;
+	uplink_dev = mlx5_eswitch_uplink_get_proto_dev(esw, REP_ETH);
+
+	rcu_read_lock();
+	uplink_upper = netdev_master_upper_dev_get_rcu(uplink_dev);
+	/* mlx5_lag_is_sriov() is a blocking function which can't be called
+	 * while holding rcu read lock. Take the net_device for correctness
+	 * sake.
+	 */
+	if (uplink_upper)
+		dev_hold(uplink_upper);
+	rcu_read_unlock();
+
+	dst_is_lag_dev = (uplink_upper &&
+			  netif_is_lag_master(uplink_upper) &&
+			  real_dev == uplink_upper &&
+			  mlx5_lag_is_sriov(priv->mdev));
+	if (uplink_upper)
+		dev_put(uplink_upper);
+
+	/* if the egress device isn't on the same HW e-switch or
+	 * it's a LAG device, use the uplink
+	 */
+	*route_dev = dev;
+	if (!netdev_port_same_parent_id(priv->netdev, real_dev) ||
+	    dst_is_lag_dev || is_vlan_dev(*route_dev) ||
+	    netif_is_ovs_master(*route_dev))
+		*out_dev = uplink_dev;
+	else if (mlx5e_eswitch_rep(dev) &&
+		 mlx5e_is_valid_eswitch_fwd_dev(priv, dev))
+		*out_dev = *route_dev;
+	else
+		return -EOPNOTSUPP;
+
+	if (!(mlx5e_eswitch_rep(*out_dev) &&
+	      mlx5e_is_uplink_rep(netdev_priv(*out_dev))))
+		return -EOPNOTSUPP;
+
+	if (mlx5e_eswitch_uplink_rep(priv->netdev) && *out_dev != priv->netdev)
+		return -EOPNOTSUPP;
+
+	return 0;
+}
+
+static int mlx5e_route_lookup_ipv4_get(struct mlx5e_priv *priv,
+				       struct net_device *dev,
+				       struct mlx5e_tc_tun_route_attr *attr)
+{
+	struct net_device *route_dev;
+	struct net_device *out_dev;
+	struct neighbour *n;
+	struct rtable *rt;
+
+#if IS_ENABLED(CONFIG_INET)
+	struct mlx5_core_dev *mdev = priv->mdev;
+	struct net_device *uplink_dev;
+	int ret;
+
+	if (mlx5_lag_is_multipath(mdev)) {
+		struct mlx5_eswitch *esw = mdev->priv.eswitch;
+
+		uplink_dev = mlx5_eswitch_uplink_get_proto_dev(esw, REP_ETH);
+		attr->fl.fl4.flowi4_oif = uplink_dev->ifindex;
+	} else {
+		struct mlx5e_tc_tunnel *tunnel = mlx5e_get_tc_tun(dev);
+
+		if (tunnel && tunnel->get_remote_ifindex)
+			attr->fl.fl4.flowi4_oif = tunnel->get_remote_ifindex(dev);
+	}
+
+	rt = ip_route_output_key(dev_net(dev), &attr->fl.fl4);
+	if (IS_ERR(rt))
+		return PTR_ERR(rt);
+
+	if (rt->rt_type != RTN_UNICAST) {
+		ret = -ENETUNREACH;
+		goto err_rt_release;
+	}
+
+	if (mlx5_lag_is_multipath(mdev) && rt->rt_gw_family != AF_INET) {
+		ret = -ENETUNREACH;
+		goto err_rt_release;
+	}
+#else
+	return -EOPNOTSUPP;
+#endif
+
+	ret = get_route_and_out_devs(priv, rt->dst.dev, &route_dev, &out_dev);
+	if (ret < 0)
+		goto err_rt_release;
+	dev_hold(route_dev);
+
+	if (!attr->ttl)
+		attr->ttl = ip4_dst_hoplimit(&rt->dst);
+	n = dst_neigh_lookup(&rt->dst, &attr->fl.fl4.daddr);
+	if (!n) {
+		ret = -ENOMEM;
+		goto err_dev_release;
+	}
+
+	ip_rt_put(rt);
+	attr->route_dev = route_dev;
+	attr->out_dev = out_dev;
+	attr->n = n;
+	return 0;
+
+err_dev_release:
+	dev_put(route_dev);
+err_rt_release:
+	ip_rt_put(rt);
+	return ret;
+}
+
+static void mlx5e_route_lookup_ipv4_put(struct mlx5e_tc_tun_route_attr *attr)
+{
+	mlx5e_tc_tun_route_attr_cleanup(attr);
+}
+
+static const char *mlx5e_netdev_kind(struct net_device *dev)
+{
+	if (dev->rtnl_link_ops)
+		return dev->rtnl_link_ops->kind;
+	else
+		return "unknown";
+}
+
+static int mlx5e_gen_ip_tunnel_header(char buf[], __u8 *ip_proto,
+				      struct mlx5e_encap_entry *e)
+{
+	if (!e->tunnel) {
+		pr_warn("mlx5: Cannot generate tunnel header for this tunnel\n");
+		return -EOPNOTSUPP;
+	}
+
+	return e->tunnel->generate_ip_tun_hdr(buf, ip_proto, e);
+}
+
+static char *gen_eth_tnl_hdr(char *buf, struct net_device *dev,
+			     struct mlx5e_encap_entry *e,
+			     u16 proto)
+{
+	struct ethhdr *eth = (struct ethhdr *)buf;
+	char *ip;
+
+	ether_addr_copy(eth->h_dest, e->h_dest);
+	ether_addr_copy(eth->h_source, dev->dev_addr);
+	if (is_vlan_dev(dev)) {
+		struct vlan_hdr *vlan = (struct vlan_hdr *)
+					((char *)eth + ETH_HLEN);
+		ip = (char *)vlan + VLAN_HLEN;
+		eth->h_proto = vlan_dev_vlan_proto(dev);
+		vlan->h_vlan_TCI = htons(vlan_dev_vlan_id(dev));
+		vlan->h_vlan_encapsulated_proto = htons(proto);
+	} else {
+		eth->h_proto = htons(proto);
+		ip = (char *)eth + ETH_HLEN;
+	}
+
+	return ip;
+}
+
+int mlx5e_tc_tun_create_header_ipv4(struct mlx5e_priv *priv,
+				    struct net_device *mirred_dev,
+				    struct mlx5e_encap_entry *e)
+{
+	int max_encap_size = MLX5_CAP_ESW(priv->mdev, max_encap_header_size);
+	const struct ip_tunnel_key *tun_key = &e->tun_info->key;
+	struct mlx5_pkt_reformat_params reformat_params;
+	struct mlx5e_neigh m_neigh = {};
+	TC_TUN_ROUTE_ATTR_INIT(attr);
+	int ipv4_encap_size;
+	char *encap_header;
+	struct iphdr *ip;
+	u8 nud_state;
+	int err;
+
+	/* add the IP fields */
+	attr.fl.fl4.flowi4_tos = tun_key->tos & ~INET_ECN_MASK;
+	attr.fl.fl4.daddr = tun_key->u.ipv4.dst;
+	attr.fl.fl4.saddr = tun_key->u.ipv4.src;
+	attr.ttl = tun_key->ttl;
+
+	err = mlx5e_route_lookup_ipv4_get(priv, mirred_dev, &attr);
+	if (err)
+		return err;
+
+	ipv4_encap_size =
+		(is_vlan_dev(attr.route_dev) ? VLAN_ETH_HLEN : ETH_HLEN) +
+		sizeof(struct iphdr) +
+		e->tunnel->calc_hlen(e);
+
+	if (max_encap_size < ipv4_encap_size) {
+		mlx5_core_warn(priv->mdev, "encap size %d too big, max supported is %d\n",
+			       ipv4_encap_size, max_encap_size);
+		err = -EOPNOTSUPP;
+		goto release_neigh;
+	}
+
+	encap_header = kzalloc(ipv4_encap_size, GFP_KERNEL);
+	if (!encap_header) {
+		err = -ENOMEM;
+		goto release_neigh;
+	}
+
+	m_neigh.family = attr.n->ops->family;
+	memcpy(&m_neigh.dst_ip, attr.n->primary_key, attr.n->tbl->key_len);
+	e->out_dev = attr.out_dev;
+	e->route_dev_ifindex = attr.route_dev->ifindex;
+
+	/* It's important to add the neigh to the hash table before checking
+	 * the neigh validity state. So if we'll get a notification, in case the
+	 * neigh changes it's validity state, we would find the relevant neigh
+	 * in the hash.
+	 */
+	err = mlx5e_rep_encap_entry_attach(netdev_priv(attr.out_dev), e, &m_neigh, attr.n->dev);
+	if (err)
+		goto free_encap;
+
+	read_lock_bh(&attr.n->lock);
+	nud_state = attr.n->nud_state;
+	ether_addr_copy(e->h_dest, attr.n->ha);
+	read_unlock_bh(&attr.n->lock);
+
+	/* add ethernet header */
+	ip = (struct iphdr *)gen_eth_tnl_hdr(encap_header, attr.route_dev, e,
+					     ETH_P_IP);
+
+	/* add ip header */
+	ip->tos = tun_key->tos;
+	ip->version = 0x4;
+	ip->ihl = 0x5;
+	ip->ttl = attr.ttl;
+	ip->daddr = attr.fl.fl4.daddr;
+	ip->saddr = attr.fl.fl4.saddr;
+
+	/* add tunneling protocol header */
+	err = mlx5e_gen_ip_tunnel_header((char *)ip + sizeof(struct iphdr),
+					 &ip->protocol, e);
+	if (err)
+		goto destroy_neigh_entry;
+
+	e->encap_size = ipv4_encap_size;
+	e->encap_header = encap_header;
+
+	if (!(nud_state & NUD_VALID)) {
+		neigh_event_send(attr.n, NULL);
+		/* the encap entry will be made valid on neigh update event
+		 * and not used before that.
+		 */
+		goto release_neigh;
+	}
+
+	memset(&reformat_params, 0, sizeof(reformat_params));
+	reformat_params.type = e->reformat_type;
+	reformat_params.size = ipv4_encap_size;
+	reformat_params.data = encap_header;
+	e->pkt_reformat = mlx5_packet_reformat_alloc(priv->mdev, &reformat_params,
+						     MLX5_FLOW_NAMESPACE_FDB);
+	if (IS_ERR(e->pkt_reformat)) {
+		err = PTR_ERR(e->pkt_reformat);
+		goto destroy_neigh_entry;
+	}
+
+	e->flags |= MLX5_ENCAP_ENTRY_VALID;
+	mlx5e_rep_queue_neigh_stats_work(netdev_priv(attr.out_dev));
+	mlx5e_route_lookup_ipv4_put(&attr);
+	return err;
+
+destroy_neigh_entry:
+	mlx5e_rep_encap_entry_detach(netdev_priv(e->out_dev), e);
+free_encap:
+	kfree(encap_header);
+release_neigh:
+	mlx5e_route_lookup_ipv4_put(&attr);
+	return err;
+}
+
+int mlx5e_tc_tun_update_header_ipv4(struct mlx5e_priv *priv,
+				    struct net_device *mirred_dev,
+				    struct mlx5e_encap_entry *e)
+{
+	int max_encap_size = MLX5_CAP_ESW(priv->mdev, max_encap_header_size);
+	const struct ip_tunnel_key *tun_key = &e->tun_info->key;
+	struct mlx5_pkt_reformat_params reformat_params;
+	TC_TUN_ROUTE_ATTR_INIT(attr);
+	int ipv4_encap_size;
+	char *encap_header;
+	struct iphdr *ip;
+	u8 nud_state;
+	int err;
+
+	/* add the IP fields */
+	attr.fl.fl4.flowi4_tos = tun_key->tos & ~INET_ECN_MASK;
+	attr.fl.fl4.daddr = tun_key->u.ipv4.dst;
+	attr.fl.fl4.saddr = tun_key->u.ipv4.src;
+	attr.ttl = tun_key->ttl;
+
+	err = mlx5e_route_lookup_ipv4_get(priv, mirred_dev, &attr);
+	if (err)
+		return err;
+
+	ipv4_encap_size =
+		(is_vlan_dev(attr.route_dev) ? VLAN_ETH_HLEN : ETH_HLEN) +
+		sizeof(struct iphdr) +
+		e->tunnel->calc_hlen(e);
+
+	if (max_encap_size < ipv4_encap_size) {
+		mlx5_core_warn(priv->mdev, "encap size %d too big, max supported is %d\n",
+			       ipv4_encap_size, max_encap_size);
+		err = -EOPNOTSUPP;
+		goto release_neigh;
+	}
+
+	encap_header = kzalloc(ipv4_encap_size, GFP_KERNEL);
+	if (!encap_header) {
+		err = -ENOMEM;
+		goto release_neigh;
+	}
+
+	e->route_dev_ifindex = attr.route_dev->ifindex;
+
+	read_lock_bh(&attr.n->lock);
+	nud_state = attr.n->nud_state;
+	ether_addr_copy(e->h_dest, attr.n->ha);
+	WRITE_ONCE(e->nhe->neigh_dev, attr.n->dev);
+	read_unlock_bh(&attr.n->lock);
+
+	/* add ethernet header */
+	ip = (struct iphdr *)gen_eth_tnl_hdr(encap_header, attr.route_dev, e,
+					     ETH_P_IP);
+
+	/* add ip header */
+	ip->tos = tun_key->tos;
+	ip->version = 0x4;
+	ip->ihl = 0x5;
+	ip->ttl = attr.ttl;
+	ip->daddr = attr.fl.fl4.daddr;
+	ip->saddr = attr.fl.fl4.saddr;
+
+	/* add tunneling protocol header */
+	err = mlx5e_gen_ip_tunnel_header((char *)ip + sizeof(struct iphdr),
+					 &ip->protocol, e);
+	if (err)
+		goto free_encap;
+
+	e->encap_size = ipv4_encap_size;
+	kfree(e->encap_header);
+	e->encap_header = encap_header;
+
+	if (!(nud_state & NUD_VALID)) {
+		neigh_event_send(attr.n, NULL);
+		/* the encap entry will be made valid on neigh update event
+		 * and not used before that.
+		 */
+		goto release_neigh;
+	}
+
+	memset(&reformat_params, 0, sizeof(reformat_params));
+	reformat_params.type = e->reformat_type;
+	reformat_params.size = ipv4_encap_size;
+	reformat_params.data = encap_header;
+	e->pkt_reformat = mlx5_packet_reformat_alloc(priv->mdev, &reformat_params,
+						     MLX5_FLOW_NAMESPACE_FDB);
+	if (IS_ERR(e->pkt_reformat)) {
+		err = PTR_ERR(e->pkt_reformat);
+		goto free_encap;
+	}
+
+	e->flags |= MLX5_ENCAP_ENTRY_VALID;
+	mlx5e_rep_queue_neigh_stats_work(netdev_priv(attr.out_dev));
+	mlx5e_route_lookup_ipv4_put(&attr);
+	return err;
+
+free_encap:
+	kfree(encap_header);
+release_neigh:
+	mlx5e_route_lookup_ipv4_put(&attr);
+	return err;
+}
+
+#if IS_ENABLED(CONFIG_INET) && IS_ENABLED(CONFIG_IPV6)
+static int mlx5e_route_lookup_ipv6_get(struct mlx5e_priv *priv,
+				       struct net_device *dev,
+				       struct mlx5e_tc_tun_route_attr *attr)
+{
+	struct mlx5e_tc_tunnel *tunnel = mlx5e_get_tc_tun(dev);
+	struct net_device *route_dev;
+	struct net_device *out_dev;
+	struct dst_entry *dst;
+	struct neighbour *n;
+	int ret;
+
+	if (tunnel && tunnel->get_remote_ifindex)
+		attr->fl.fl6.flowi6_oif = tunnel->get_remote_ifindex(dev);
+	dst = ipv6_stub->ipv6_dst_lookup_flow(dev_net(dev), NULL, &attr->fl.fl6,
+					      NULL);
+	if (IS_ERR(dst))
+		return PTR_ERR(dst);
+
+	if (!attr->ttl)
+		attr->ttl = ip6_dst_hoplimit(dst);
+
+	ret = get_route_and_out_devs(priv, dst->dev, &route_dev, &out_dev);
+	if (ret < 0)
+		goto err_dst_release;
+
+	dev_hold(route_dev);
+	n = dst_neigh_lookup(dst, &attr->fl.fl6.daddr);
+	if (!n) {
+		ret = -ENOMEM;
+		goto err_dev_release;
+	}
+
+	dst_release(dst);
+	attr->out_dev = out_dev;
+	attr->route_dev = route_dev;
+	attr->n = n;
+	return 0;
+
+err_dev_release:
+	dev_put(route_dev);
+err_dst_release:
+	dst_release(dst);
+	return ret;
+}
+
+static void mlx5e_route_lookup_ipv6_put(struct mlx5e_tc_tun_route_attr *attr)
+{
+	mlx5e_tc_tun_route_attr_cleanup(attr);
+}
+
+int mlx5e_tc_tun_create_header_ipv6(struct mlx5e_priv *priv,
+				    struct net_device *mirred_dev,
+				    struct mlx5e_encap_entry *e)
+{
+	int max_encap_size = MLX5_CAP_ESW(priv->mdev, max_encap_header_size);
+	const struct ip_tunnel_key *tun_key = &e->tun_info->key;
+	struct mlx5_pkt_reformat_params reformat_params;
+	struct mlx5e_neigh m_neigh = {};
+	TC_TUN_ROUTE_ATTR_INIT(attr);
+	struct ipv6hdr *ip6h;
+	int ipv6_encap_size;
+	char *encap_header;
+	u8 nud_state;
+	int err;
+
+	attr.ttl = tun_key->ttl;
+	attr.fl.fl6.flowlabel = ip6_make_flowinfo(tun_key->tos, tun_key->label);
+	attr.fl.fl6.daddr = tun_key->u.ipv6.dst;
+	attr.fl.fl6.saddr = tun_key->u.ipv6.src;
+
+	err = mlx5e_route_lookup_ipv6_get(priv, mirred_dev, &attr);
+	if (err)
+		return err;
+
+	ipv6_encap_size =
+		(is_vlan_dev(attr.route_dev) ? VLAN_ETH_HLEN : ETH_HLEN) +
+		sizeof(struct ipv6hdr) +
+		e->tunnel->calc_hlen(e);
+
+	if (max_encap_size < ipv6_encap_size) {
+		mlx5_core_warn(priv->mdev, "encap size %d too big, max supported is %d\n",
+			       ipv6_encap_size, max_encap_size);
+		err = -EOPNOTSUPP;
+		goto release_neigh;
+	}
+
+	encap_header = kzalloc(ipv6_encap_size, GFP_KERNEL);
+	if (!encap_header) {
+		err = -ENOMEM;
+		goto release_neigh;
+	}
+
+	m_neigh.family = attr.n->ops->family;
+	memcpy(&m_neigh.dst_ip, attr.n->primary_key, attr.n->tbl->key_len);
+	e->out_dev = attr.out_dev;
+	e->route_dev_ifindex = attr.route_dev->ifindex;
+
+	/* It's important to add the neigh to the hash table before checking
+	 * the neigh validity state. So if we'll get a notification, in case the
+	 * neigh changes it's validity state, we would find the relevant neigh
+	 * in the hash.
+	 */
+	err = mlx5e_rep_encap_entry_attach(netdev_priv(attr.out_dev), e, &m_neigh, attr.n->dev);
+	if (err)
+		goto free_encap;
+
+	read_lock_bh(&attr.n->lock);
+	nud_state = attr.n->nud_state;
+	ether_addr_copy(e->h_dest, attr.n->ha);
+	read_unlock_bh(&attr.n->lock);
+
+	/* add ethernet header */
+	ip6h = (struct ipv6hdr *)gen_eth_tnl_hdr(encap_header, attr.route_dev, e,
+						 ETH_P_IPV6);
+
+	/* add ip header */
+	ip6_flow_hdr(ip6h, tun_key->tos, 0);
+	/* the HW fills up ipv6 payload len */
+	ip6h->hop_limit   = attr.ttl;
+	ip6h->daddr	  = attr.fl.fl6.daddr;
+	ip6h->saddr	  = attr.fl.fl6.saddr;
+
+	/* add tunneling protocol header */
+	err = mlx5e_gen_ip_tunnel_header((char *)ip6h + sizeof(struct ipv6hdr),
+					 &ip6h->nexthdr, e);
+	if (err)
+		goto destroy_neigh_entry;
+
+	e->encap_size = ipv6_encap_size;
+	e->encap_header = encap_header;
+
+	if (!(nud_state & NUD_VALID)) {
+		neigh_event_send(attr.n, NULL);
+		/* the encap entry will be made valid on neigh update event
+		 * and not used before that.
+		 */
+		goto release_neigh;
+	}
+
+	memset(&reformat_params, 0, sizeof(reformat_params));
+	reformat_params.type = e->reformat_type;
+	reformat_params.size = ipv6_encap_size;
+	reformat_params.data = encap_header;
+	e->pkt_reformat = mlx5_packet_reformat_alloc(priv->mdev, &reformat_params,
+						     MLX5_FLOW_NAMESPACE_FDB);
+	if (IS_ERR(e->pkt_reformat)) {
+		err = PTR_ERR(e->pkt_reformat);
+		goto destroy_neigh_entry;
+	}
+
+	e->flags |= MLX5_ENCAP_ENTRY_VALID;
+	mlx5e_rep_queue_neigh_stats_work(netdev_priv(attr.out_dev));
+	mlx5e_route_lookup_ipv6_put(&attr);
+	return err;
+
+destroy_neigh_entry:
+	mlx5e_rep_encap_entry_detach(netdev_priv(e->out_dev), e);
+free_encap:
+	kfree(encap_header);
+release_neigh:
+	mlx5e_route_lookup_ipv6_put(&attr);
+	return err;
+}
+
+int mlx5e_tc_tun_update_header_ipv6(struct mlx5e_priv *priv,
+				    struct net_device *mirred_dev,
+				    struct mlx5e_encap_entry *e)
+{
+	int max_encap_size = MLX5_CAP_ESW(priv->mdev, max_encap_header_size);
+	const struct ip_tunnel_key *tun_key = &e->tun_info->key;
+	struct mlx5_pkt_reformat_params reformat_params;
+	TC_TUN_ROUTE_ATTR_INIT(attr);
+	struct ipv6hdr *ip6h;
+	int ipv6_encap_size;
+	char *encap_header;
+	u8 nud_state;
+	int err;
+
+	attr.ttl = tun_key->ttl;
+
+	attr.fl.fl6.flowlabel = ip6_make_flowinfo(tun_key->tos, tun_key->label);
+	attr.fl.fl6.daddr = tun_key->u.ipv6.dst;
+	attr.fl.fl6.saddr = tun_key->u.ipv6.src;
+
+	err = mlx5e_route_lookup_ipv6_get(priv, mirred_dev, &attr);
+	if (err)
+		return err;
+
+	ipv6_encap_size =
+		(is_vlan_dev(attr.route_dev) ? VLAN_ETH_HLEN : ETH_HLEN) +
+		sizeof(struct ipv6hdr) +
+		e->tunnel->calc_hlen(e);
+
+	if (max_encap_size < ipv6_encap_size) {
+		mlx5_core_warn(priv->mdev, "encap size %d too big, max supported is %d\n",
+			       ipv6_encap_size, max_encap_size);
+		err = -EOPNOTSUPP;
+		goto release_neigh;
+	}
+
+	encap_header = kzalloc(ipv6_encap_size, GFP_KERNEL);
+	if (!encap_header) {
+		err = -ENOMEM;
+		goto release_neigh;
+	}
+
+	e->route_dev_ifindex = attr.route_dev->ifindex;
+
+	read_lock_bh(&attr.n->lock);
+	nud_state = attr.n->nud_state;
+	ether_addr_copy(e->h_dest, attr.n->ha);
+	WRITE_ONCE(e->nhe->neigh_dev, attr.n->dev);
+	read_unlock_bh(&attr.n->lock);
+
+	/* add ethernet header */
+	ip6h = (struct ipv6hdr *)gen_eth_tnl_hdr(encap_header, attr.route_dev, e,
+						 ETH_P_IPV6);
+
+	/* add ip header */
+	ip6_flow_hdr(ip6h, tun_key->tos, 0);
+	/* the HW fills up ipv6 payload len */
+	ip6h->hop_limit   = attr.ttl;
+	ip6h->daddr	  = attr.fl.fl6.daddr;
+	ip6h->saddr	  = attr.fl.fl6.saddr;
+
+	/* add tunneling protocol header */
+	err = mlx5e_gen_ip_tunnel_header((char *)ip6h + sizeof(struct ipv6hdr),
+					 &ip6h->nexthdr, e);
+	if (err)
+		goto free_encap;
+
+	e->encap_size = ipv6_encap_size;
+	kfree(e->encap_header);
+	e->encap_header = encap_header;
+
+	if (!(nud_state & NUD_VALID)) {
+		neigh_event_send(attr.n, NULL);
+		/* the encap entry will be made valid on neigh update event
+		 * and not used before that.
+		 */
+		goto release_neigh;
+	}
+
+	memset(&reformat_params, 0, sizeof(reformat_params));
+	reformat_params.type = e->reformat_type;
+	reformat_params.size = ipv6_encap_size;
+	reformat_params.data = encap_header;
+	e->pkt_reformat = mlx5_packet_reformat_alloc(priv->mdev, &reformat_params,
+						     MLX5_FLOW_NAMESPACE_FDB);
+	if (IS_ERR(e->pkt_reformat)) {
+		err = PTR_ERR(e->pkt_reformat);
+		goto free_encap;
+	}
+
+	e->flags |= MLX5_ENCAP_ENTRY_VALID;
+	mlx5e_rep_queue_neigh_stats_work(netdev_priv(attr.out_dev));
+	mlx5e_route_lookup_ipv6_put(&attr);
+	return err;
+
+free_encap:
+	kfree(encap_header);
+release_neigh:
+	mlx5e_route_lookup_ipv6_put(&attr);
+	return err;
+}
+#endif
+
+int mlx5e_tc_tun_route_lookup(struct mlx5e_priv *priv,
+			      struct mlx5_flow_spec *spec,
+			      struct mlx5_flow_attr *flow_attr,
+			      struct net_device *filter_dev)
+{
+	struct mlx5_esw_flow_attr *esw_attr = flow_attr->esw_attr;
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	struct mlx5e_tc_int_port *int_port;
+	TC_TUN_ROUTE_ATTR_INIT(attr);
+	u16 vport_num;
+	int err = 0;
+
+	if (flow_attr->tun_ip_version == 4) {
+		/* Addresses are swapped for decap */
+		attr.fl.fl4.saddr = esw_attr->rx_tun_attr->dst_ip.v4;
+		attr.fl.fl4.daddr = esw_attr->rx_tun_attr->src_ip.v4;
+		err = mlx5e_route_lookup_ipv4_get(priv, filter_dev, &attr);
+	}
+#if IS_ENABLED(CONFIG_INET) && IS_ENABLED(CONFIG_IPV6)
+	else if (flow_attr->tun_ip_version == 6) {
+		/* Addresses are swapped for decap */
+		attr.fl.fl6.saddr = esw_attr->rx_tun_attr->dst_ip.v6;
+		attr.fl.fl6.daddr = esw_attr->rx_tun_attr->src_ip.v6;
+		err = mlx5e_route_lookup_ipv6_get(priv, filter_dev, &attr);
+	}
+#endif
+	else
+		return 0;
+
+	if (err)
+		return err;
+
+	if (attr.route_dev->netdev_ops == &mlx5e_netdev_ops &&
+	    mlx5e_tc_is_vf_tunnel(attr.out_dev, attr.route_dev)) {
+		err = mlx5e_tc_query_route_vport(attr.out_dev, attr.route_dev, &vport_num);
+		if (err)
+			goto out;
+
+		esw_attr->rx_tun_attr->decap_vport = vport_num;
+	} else if (netif_is_ovs_master(attr.route_dev) && mlx5e_tc_int_port_supported(esw)) {
+		int_port = mlx5e_tc_int_port_get(mlx5e_get_int_port_priv(priv),
+						 attr.route_dev->ifindex,
+						 MLX5E_TC_INT_PORT_INGRESS);
+		if (IS_ERR(int_port)) {
+			err = PTR_ERR(int_port);
+			goto out;
+		}
+		esw_attr->int_port = int_port;
+	}
+
+out:
+	if (flow_attr->tun_ip_version == 4)
+		mlx5e_route_lookup_ipv4_put(&attr);
+#if IS_ENABLED(CONFIG_INET) && IS_ENABLED(CONFIG_IPV6)
+	else if (flow_attr->tun_ip_version == 6)
+		mlx5e_route_lookup_ipv6_put(&attr);
+#endif
+	return err;
+}
+
+bool mlx5e_tc_tun_device_to_offload(struct mlx5e_priv *priv,
+				    struct net_device *netdev)
+{
+	struct mlx5e_tc_tunnel *tunnel = mlx5e_get_tc_tun(netdev);
+
+	if (tunnel && tunnel->can_offload(priv))
+		return true;
+	else
+		return false;
+}
+
+int mlx5e_tc_tun_init_encap_attr(struct net_device *tunnel_dev,
+				 struct mlx5e_priv *priv,
+				 struct mlx5e_encap_entry *e,
+				 struct netlink_ext_ack *extack)
+{
+	struct mlx5e_tc_tunnel *tunnel = mlx5e_get_tc_tun(tunnel_dev);
+
+	if (!tunnel) {
+		e->reformat_type = -1;
+		return -EOPNOTSUPP;
+	}
+
+	return tunnel->init_encap_attr(tunnel_dev, priv, e, extack);
+}
+
+int mlx5e_tc_tun_parse(struct net_device *filter_dev,
+		       struct mlx5e_priv *priv,
+		       struct mlx5_flow_spec *spec,
+		       struct flow_cls_offload *f,
+		       u8 *match_level)
+{
+	struct mlx5e_tc_tunnel *tunnel = mlx5e_get_tc_tun(filter_dev);
+	struct flow_rule *rule = flow_cls_offload_flow_rule(f);
+	void *headers_c = MLX5_ADDR_OF(fte_match_param, spec->match_criteria,
+				       outer_headers);
+	void *headers_v = MLX5_ADDR_OF(fte_match_param, spec->match_value,
+				       outer_headers);
+	struct netlink_ext_ack *extack = f->common.extack;
+	int err = 0;
+
+	if (!tunnel) {
+		netdev_warn(priv->netdev,
+			    "decapsulation offload is not supported for %s net device\n",
+			    mlx5e_netdev_kind(filter_dev));
+		err = -EOPNOTSUPP;
+		goto out;
+	}
+
+	*match_level = tunnel->match_level;
+
+	if (tunnel->parse_udp_ports) {
+		err = tunnel->parse_udp_ports(priv, spec, f,
+					      headers_c, headers_v);
+		if (err)
+			goto out;
+	}
+
+	if (tunnel->parse_tunnel) {
+		err = tunnel->parse_tunnel(priv, spec, f,
+					   headers_c, headers_v);
+		if (err)
+			goto out;
+	}
+
+	if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_ENC_CONTROL)) {
+		struct flow_dissector_key_basic key_basic = {};
+		struct flow_dissector_key_basic mask_basic = {
+			.n_proto = htons(0xFFFF),
+		};
+		struct flow_match_basic match_basic = {
+			.key = &key_basic, .mask = &mask_basic,
+		};
+		struct flow_match_control match;
+		u16 addr_type;
+
+		flow_rule_match_enc_control(rule, &match);
+		addr_type = match.key->addr_type;
+
+		/* For tunnel addr_type used same key id`s as for non-tunnel */
+		if (addr_type == FLOW_DISSECTOR_KEY_IPV4_ADDRS) {
+			struct flow_match_ipv4_addrs match;
+
+			flow_rule_match_enc_ipv4_addrs(rule, &match);
+			MLX5_SET(fte_match_set_lyr_2_4, headers_c,
+				 src_ipv4_src_ipv6.ipv4_layout.ipv4,
+				 ntohl(match.mask->src));
+			MLX5_SET(fte_match_set_lyr_2_4, headers_v,
+				 src_ipv4_src_ipv6.ipv4_layout.ipv4,
+				 ntohl(match.key->src));
+
+			MLX5_SET(fte_match_set_lyr_2_4, headers_c,
+				 dst_ipv4_dst_ipv6.ipv4_layout.ipv4,
+				 ntohl(match.mask->dst));
+			MLX5_SET(fte_match_set_lyr_2_4, headers_v,
+				 dst_ipv4_dst_ipv6.ipv4_layout.ipv4,
+				 ntohl(match.key->dst));
+
+			key_basic.n_proto = htons(ETH_P_IP);
+			mlx5e_tc_set_ethertype(priv->mdev, &match_basic, true,
+					       headers_c, headers_v);
+		} else if (addr_type == FLOW_DISSECTOR_KEY_IPV6_ADDRS) {
+			struct flow_match_ipv6_addrs match;
+
+			flow_rule_match_enc_ipv6_addrs(rule, &match);
+			memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_c,
+					    src_ipv4_src_ipv6.ipv6_layout.ipv6),
+			       &match.mask->src, MLX5_FLD_SZ_BYTES(ipv6_layout,
+								   ipv6));
+			memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_v,
+					    src_ipv4_src_ipv6.ipv6_layout.ipv6),
+			       &match.key->src, MLX5_FLD_SZ_BYTES(ipv6_layout,
+								  ipv6));
+
+			memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_c,
+					    dst_ipv4_dst_ipv6.ipv6_layout.ipv6),
+			       &match.mask->dst, MLX5_FLD_SZ_BYTES(ipv6_layout,
+								   ipv6));
+			memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_v,
+					    dst_ipv4_dst_ipv6.ipv6_layout.ipv6),
+			       &match.key->dst, MLX5_FLD_SZ_BYTES(ipv6_layout,
+								  ipv6));
+
+			key_basic.n_proto = htons(ETH_P_IPV6);
+			mlx5e_tc_set_ethertype(priv->mdev, &match_basic, true,
+					       headers_c, headers_v);
+		}
+	}
+
+	if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_ENC_IP)) {
+		struct flow_match_ip match;
+
+		flow_rule_match_enc_ip(rule, &match);
+		MLX5_SET(fte_match_set_lyr_2_4, headers_c, ip_ecn,
+			 match.mask->tos & 0x3);
+		MLX5_SET(fte_match_set_lyr_2_4, headers_v, ip_ecn,
+			 match.key->tos & 0x3);
+
+		MLX5_SET(fte_match_set_lyr_2_4, headers_c, ip_dscp,
+			 match.mask->tos >> 2);
+		MLX5_SET(fte_match_set_lyr_2_4, headers_v, ip_dscp,
+			 match.key->tos  >> 2);
+
+		MLX5_SET(fte_match_set_lyr_2_4, headers_c, ttl_hoplimit,
+			 match.mask->ttl);
+		MLX5_SET(fte_match_set_lyr_2_4, headers_v, ttl_hoplimit,
+			 match.key->ttl);
+
+		if (match.mask->ttl &&
+		    !MLX5_CAP_ESW_FLOWTABLE_FDB
+			(priv->mdev,
+			 ft_field_support.outer_ipv4_ttl)) {
+			NL_SET_ERR_MSG_MOD(extack,
+					   "Matching on TTL is not supported");
+			err = -EOPNOTSUPP;
+			goto out;
+		}
+	}
+
+	/* let software handle IP fragments */
+	MLX5_SET(fte_match_set_lyr_2_4, headers_c, frag, 1);
+	MLX5_SET(fte_match_set_lyr_2_4, headers_v, frag, 0);
+
+	return 0;
+
+out:
+	return err;
+}
+
+int mlx5e_tc_tun_parse_udp_ports(struct mlx5e_priv *priv,
+				 struct mlx5_flow_spec *spec,
+				 struct flow_cls_offload *f,
+				 void *headers_c,
+				 void *headers_v)
+{
+	struct flow_rule *rule = flow_cls_offload_flow_rule(f);
+	struct netlink_ext_ack *extack = f->common.extack;
+	struct flow_match_ports enc_ports;
+
+	/* Full udp dst port must be given */
+
+	if (!flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_ENC_PORTS)) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "UDP tunnel decap filter must include enc_dst_port condition");
+		netdev_warn(priv->netdev,
+			    "UDP tunnel decap filter must include enc_dst_port condition\n");
+		return -EOPNOTSUPP;
+	}
+
+	flow_rule_match_enc_ports(rule, &enc_ports);
+
+	if (memchr_inv(&enc_ports.mask->dst, 0xff,
+		       sizeof(enc_ports.mask->dst))) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "UDP tunnel decap filter must match enc_dst_port fully");
+		netdev_warn(priv->netdev,
+			    "UDP tunnel decap filter must match enc_dst_port fully\n");
+		return -EOPNOTSUPP;
+	}
+
+	/* match on UDP protocol and dst port number */
+
+	MLX5_SET_TO_ONES(fte_match_set_lyr_2_4, headers_c, ip_protocol);
+	MLX5_SET(fte_match_set_lyr_2_4, headers_v, ip_protocol, IPPROTO_UDP);
+
+	MLX5_SET(fte_match_set_lyr_2_4, headers_c, udp_dport,
+		 ntohs(enc_ports.mask->dst));
+	MLX5_SET(fte_match_set_lyr_2_4, headers_v, udp_dport,
+		 ntohs(enc_ports.key->dst));
+
+	/* UDP src port on outer header is generated by HW,
+	 * so it is probably a bad idea to request matching it.
+	 * Nonetheless, it is allowed.
+	 */
+
+	MLX5_SET(fte_match_set_lyr_2_4, headers_c, udp_sport,
+		 ntohs(enc_ports.mask->src));
+	MLX5_SET(fte_match_set_lyr_2_4, headers_v, udp_sport,
+		 ntohs(enc_ports.key->src));
+
+	return 0;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun.h b/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun.h
new file mode 100644
index 000000000..b38f693bb
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun.h
@@ -0,0 +1,120 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2018 Mellanox Technologies. */
+
+#ifndef __MLX5_EN_TC_TUNNEL_H__
+#define __MLX5_EN_TC_TUNNEL_H__
+
+#include <linux/netdevice.h>
+#include <linux/mlx5/fs.h>
+#include <net/pkt_cls.h>
+#include <linux/netlink.h>
+#include "en.h"
+#include "en_rep.h"
+
+#ifdef CONFIG_MLX5_ESWITCH
+
+enum {
+	MLX5E_TC_TUNNEL_TYPE_UNKNOWN,
+	MLX5E_TC_TUNNEL_TYPE_VXLAN,
+	MLX5E_TC_TUNNEL_TYPE_GENEVE,
+	MLX5E_TC_TUNNEL_TYPE_GRETAP,
+	MLX5E_TC_TUNNEL_TYPE_MPLSOUDP,
+};
+
+struct mlx5e_encap_key {
+	const struct ip_tunnel_key *ip_tun_key;
+	struct mlx5e_tc_tunnel     *tc_tunnel;
+};
+
+struct mlx5e_tc_tunnel {
+	int tunnel_type;
+	enum mlx5_flow_match_level match_level;
+
+	bool (*can_offload)(struct mlx5e_priv *priv);
+	int (*calc_hlen)(struct mlx5e_encap_entry *e);
+	int (*init_encap_attr)(struct net_device *tunnel_dev,
+			       struct mlx5e_priv *priv,
+			       struct mlx5e_encap_entry *e,
+			       struct netlink_ext_ack *extack);
+	int (*generate_ip_tun_hdr)(char buf[],
+				   __u8 *ip_proto,
+				   struct mlx5e_encap_entry *e);
+	int (*parse_udp_ports)(struct mlx5e_priv *priv,
+			       struct mlx5_flow_spec *spec,
+			       struct flow_cls_offload *f,
+			       void *headers_c,
+			       void *headers_v);
+	int (*parse_tunnel)(struct mlx5e_priv *priv,
+			    struct mlx5_flow_spec *spec,
+			    struct flow_cls_offload *f,
+			    void *headers_c,
+			    void *headers_v);
+	bool (*encap_info_equal)(struct mlx5e_encap_key *a,
+				 struct mlx5e_encap_key *b);
+	int (*get_remote_ifindex)(struct net_device *mirred_dev);
+};
+
+extern struct mlx5e_tc_tunnel vxlan_tunnel;
+extern struct mlx5e_tc_tunnel geneve_tunnel;
+extern struct mlx5e_tc_tunnel gre_tunnel;
+extern struct mlx5e_tc_tunnel mplsoudp_tunnel;
+
+struct mlx5e_tc_tunnel *mlx5e_get_tc_tun(struct net_device *tunnel_dev);
+
+int mlx5e_tc_tun_init_encap_attr(struct net_device *tunnel_dev,
+				 struct mlx5e_priv *priv,
+				 struct mlx5e_encap_entry *e,
+				 struct netlink_ext_ack *extack);
+
+int mlx5e_tc_tun_create_header_ipv4(struct mlx5e_priv *priv,
+				    struct net_device *mirred_dev,
+				    struct mlx5e_encap_entry *e);
+int mlx5e_tc_tun_update_header_ipv4(struct mlx5e_priv *priv,
+				    struct net_device *mirred_dev,
+				    struct mlx5e_encap_entry *e);
+
+#if IS_ENABLED(CONFIG_INET) && IS_ENABLED(CONFIG_IPV6)
+int mlx5e_tc_tun_create_header_ipv6(struct mlx5e_priv *priv,
+				    struct net_device *mirred_dev,
+				    struct mlx5e_encap_entry *e);
+int mlx5e_tc_tun_update_header_ipv6(struct mlx5e_priv *priv,
+				    struct net_device *mirred_dev,
+				    struct mlx5e_encap_entry *e);
+#else
+static inline int
+mlx5e_tc_tun_create_header_ipv6(struct mlx5e_priv *priv,
+				struct net_device *mirred_dev,
+				struct mlx5e_encap_entry *e)
+{ return -EOPNOTSUPP; }
+static inline int
+mlx5e_tc_tun_update_header_ipv6(struct mlx5e_priv *priv,
+				struct net_device *mirred_dev,
+				struct mlx5e_encap_entry *e)
+{ return -EOPNOTSUPP; }
+#endif
+int mlx5e_tc_tun_route_lookup(struct mlx5e_priv *priv,
+			      struct mlx5_flow_spec *spec,
+			      struct mlx5_flow_attr *attr,
+			      struct net_device *filter_dev);
+
+bool mlx5e_tc_tun_device_to_offload(struct mlx5e_priv *priv,
+				    struct net_device *netdev);
+
+int mlx5e_tc_tun_parse(struct net_device *filter_dev,
+		       struct mlx5e_priv *priv,
+		       struct mlx5_flow_spec *spec,
+		       struct flow_cls_offload *f,
+		       u8 *match_level);
+
+int mlx5e_tc_tun_parse_udp_ports(struct mlx5e_priv *priv,
+				 struct mlx5_flow_spec *spec,
+				 struct flow_cls_offload *f,
+				 void *headers_c,
+				 void *headers_v);
+
+bool mlx5e_tc_tun_encap_info_equal_generic(struct mlx5e_encap_key *a,
+					   struct mlx5e_encap_key *b);
+
+#endif /* CONFIG_MLX5_ESWITCH */
+
+#endif //__MLX5_EN_TC_TUNNEL_H__
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun_encap.c b/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun_encap.c
new file mode 100644
index 000000000..907ad6ffe
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun_encap.c
@@ -0,0 +1,1766 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2021 Mellanox Technologies. */
+
+#include <net/fib_notifier.h>
+#include <net/nexthop.h>
+#include "tc_tun_encap.h"
+#include "en_tc.h"
+#include "tc_tun.h"
+#include "rep/tc.h"
+#include "diag/en_tc_tracepoint.h"
+
+enum {
+	MLX5E_ROUTE_ENTRY_VALID     = BIT(0),
+};
+
+static int mlx5e_set_int_port_tunnel(struct mlx5e_priv *priv,
+				     struct mlx5_flow_attr *attr,
+				     struct mlx5e_encap_entry *e,
+				     int out_index)
+{
+	struct net_device *route_dev;
+	int err = 0;
+
+	route_dev = dev_get_by_index(dev_net(e->out_dev), e->route_dev_ifindex);
+
+	if (!route_dev || !netif_is_ovs_master(route_dev) ||
+	    attr->parse_attr->filter_dev == e->out_dev)
+		goto out;
+
+	err = mlx5e_set_fwd_to_int_port_actions(priv, attr, e->route_dev_ifindex,
+						MLX5E_TC_INT_PORT_EGRESS,
+						&attr->action, out_index);
+
+out:
+	if (route_dev)
+		dev_put(route_dev);
+
+	return err;
+}
+
+struct mlx5e_route_key {
+	int ip_version;
+	union {
+		__be32 v4;
+		struct in6_addr v6;
+	} endpoint_ip;
+};
+
+struct mlx5e_route_entry {
+	struct mlx5e_route_key key;
+	struct list_head encap_entries;
+	struct list_head decap_flows;
+	u32 flags;
+	struct hlist_node hlist;
+	refcount_t refcnt;
+	int tunnel_dev_index;
+	struct rcu_head rcu;
+};
+
+struct mlx5e_tc_tun_encap {
+	struct mlx5e_priv *priv;
+	struct notifier_block fib_nb;
+	spinlock_t route_lock; /* protects route_tbl */
+	unsigned long route_tbl_last_update;
+	DECLARE_HASHTABLE(route_tbl, 8);
+};
+
+static bool mlx5e_route_entry_valid(struct mlx5e_route_entry *r)
+{
+	return r->flags & MLX5E_ROUTE_ENTRY_VALID;
+}
+
+int mlx5e_tc_set_attr_rx_tun(struct mlx5e_tc_flow *flow,
+			     struct mlx5_flow_spec *spec)
+{
+	struct mlx5_esw_flow_attr *esw_attr = flow->attr->esw_attr;
+	struct mlx5_rx_tun_attr *tun_attr;
+	void *daddr, *saddr;
+	u8 ip_version;
+
+	tun_attr = kvzalloc(sizeof(*tun_attr), GFP_KERNEL);
+	if (!tun_attr)
+		return -ENOMEM;
+
+	esw_attr->rx_tun_attr = tun_attr;
+	ip_version = mlx5e_tc_get_ip_version(spec, true);
+
+	if (ip_version == 4) {
+		daddr = MLX5_ADDR_OF(fte_match_param, spec->match_value,
+				     outer_headers.dst_ipv4_dst_ipv6.ipv4_layout.ipv4);
+		saddr = MLX5_ADDR_OF(fte_match_param, spec->match_value,
+				     outer_headers.src_ipv4_src_ipv6.ipv4_layout.ipv4);
+		tun_attr->dst_ip.v4 = *(__be32 *)daddr;
+		tun_attr->src_ip.v4 = *(__be32 *)saddr;
+		if (!tun_attr->dst_ip.v4 || !tun_attr->src_ip.v4)
+			return 0;
+	}
+#if IS_ENABLED(CONFIG_INET) && IS_ENABLED(CONFIG_IPV6)
+	else if (ip_version == 6) {
+		int ipv6_size = MLX5_FLD_SZ_BYTES(ipv6_layout, ipv6);
+
+		daddr = MLX5_ADDR_OF(fte_match_param, spec->match_value,
+				     outer_headers.dst_ipv4_dst_ipv6.ipv6_layout.ipv6);
+		saddr = MLX5_ADDR_OF(fte_match_param, spec->match_value,
+				     outer_headers.src_ipv4_src_ipv6.ipv6_layout.ipv6);
+		memcpy(&tun_attr->dst_ip.v6, daddr, ipv6_size);
+		memcpy(&tun_attr->src_ip.v6, saddr, ipv6_size);
+		if (ipv6_addr_any(&tun_attr->dst_ip.v6) ||
+		    ipv6_addr_any(&tun_attr->src_ip.v6))
+			return 0;
+	}
+#endif
+	/* Only set the flag if both src and dst ip addresses exist. They are
+	 * required to establish routing.
+	 */
+	flow_flag_set(flow, TUN_RX);
+	flow->attr->tun_ip_version = ip_version;
+	return 0;
+}
+
+static bool mlx5e_tc_flow_all_encaps_valid(struct mlx5_esw_flow_attr *esw_attr)
+{
+	bool all_flow_encaps_valid = true;
+	int i;
+
+	/* Flow can be associated with multiple encap entries.
+	 * Before offloading the flow verify that all of them have
+	 * a valid neighbour.
+	 */
+	for (i = 0; i < MLX5_MAX_FLOW_FWD_VPORTS; i++) {
+		if (!(esw_attr->dests[i].flags & MLX5_ESW_DEST_ENCAP))
+			continue;
+		if (!(esw_attr->dests[i].flags & MLX5_ESW_DEST_ENCAP_VALID)) {
+			all_flow_encaps_valid = false;
+			break;
+		}
+	}
+
+	return all_flow_encaps_valid;
+}
+
+void mlx5e_tc_encap_flows_add(struct mlx5e_priv *priv,
+			      struct mlx5e_encap_entry *e,
+			      struct list_head *flow_list)
+{
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	struct mlx5_pkt_reformat_params reformat_params;
+	struct mlx5_esw_flow_attr *esw_attr;
+	struct mlx5_flow_handle *rule;
+	struct mlx5_flow_attr *attr;
+	struct mlx5_flow_spec *spec;
+	struct mlx5e_tc_flow *flow;
+	int err;
+
+	if (e->flags & MLX5_ENCAP_ENTRY_NO_ROUTE)
+		return;
+
+	memset(&reformat_params, 0, sizeof(reformat_params));
+	reformat_params.type = e->reformat_type;
+	reformat_params.size = e->encap_size;
+	reformat_params.data = e->encap_header;
+	e->pkt_reformat = mlx5_packet_reformat_alloc(priv->mdev,
+						     &reformat_params,
+						     MLX5_FLOW_NAMESPACE_FDB);
+	if (IS_ERR(e->pkt_reformat)) {
+		mlx5_core_warn(priv->mdev, "Failed to offload cached encapsulation header, %lu\n",
+			       PTR_ERR(e->pkt_reformat));
+		return;
+	}
+	e->flags |= MLX5_ENCAP_ENTRY_VALID;
+	mlx5e_rep_queue_neigh_stats_work(priv);
+
+	list_for_each_entry(flow, flow_list, tmp_list) {
+		if (!mlx5e_is_offloaded_flow(flow) || !flow_flag_test(flow, SLOW))
+			continue;
+
+		spec = &flow->attr->parse_attr->spec;
+
+		attr = mlx5e_tc_get_encap_attr(flow);
+		esw_attr = attr->esw_attr;
+		esw_attr->dests[flow->tmp_entry_index].pkt_reformat = e->pkt_reformat;
+		esw_attr->dests[flow->tmp_entry_index].flags |= MLX5_ESW_DEST_ENCAP_VALID;
+
+		/* Do not offload flows with unresolved neighbors */
+		if (!mlx5e_tc_flow_all_encaps_valid(esw_attr))
+			continue;
+
+		err = mlx5e_tc_offload_flow_post_acts(flow);
+		if (err) {
+			mlx5_core_warn(priv->mdev, "Failed to update flow post acts, %d\n",
+				       err);
+			continue;
+		}
+
+		/* update from slow path rule to encap rule */
+		rule = mlx5e_tc_offload_fdb_rules(esw, flow, spec, flow->attr);
+		if (IS_ERR(rule)) {
+			mlx5e_tc_unoffload_flow_post_acts(flow);
+			err = PTR_ERR(rule);
+			mlx5_core_warn(priv->mdev, "Failed to update cached encapsulation flow, %d\n",
+				       err);
+			continue;
+		}
+
+		mlx5e_tc_unoffload_from_slow_path(esw, flow);
+		flow->rule[0] = rule;
+		/* was unset when slow path rule removed */
+		flow_flag_set(flow, OFFLOADED);
+	}
+}
+
+void mlx5e_tc_encap_flows_del(struct mlx5e_priv *priv,
+			      struct mlx5e_encap_entry *e,
+			      struct list_head *flow_list)
+{
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	struct mlx5_esw_flow_attr *esw_attr;
+	struct mlx5_flow_handle *rule;
+	struct mlx5_flow_attr *attr;
+	struct mlx5_flow_spec *spec;
+	struct mlx5e_tc_flow *flow;
+	int err;
+
+	list_for_each_entry(flow, flow_list, tmp_list) {
+		if (!mlx5e_is_offloaded_flow(flow))
+			continue;
+
+		attr = mlx5e_tc_get_encap_attr(flow);
+		esw_attr = attr->esw_attr;
+		/* mark the flow's encap dest as non-valid */
+		esw_attr->dests[flow->tmp_entry_index].flags &= ~MLX5_ESW_DEST_ENCAP_VALID;
+		esw_attr->dests[flow->tmp_entry_index].pkt_reformat = NULL;
+
+		/* Clear pkt_reformat before checking slow path flag. Because
+		 * in next iteration, the same flow is already set slow path
+		 * flag, but still need to clear the pkt_reformat.
+		 */
+		if (flow_flag_test(flow, SLOW))
+			continue;
+
+		/* update from encap rule to slow path rule */
+		spec = &flow->attr->parse_attr->spec;
+		rule = mlx5e_tc_offload_to_slow_path(esw, flow, spec);
+
+		if (IS_ERR(rule)) {
+			err = PTR_ERR(rule);
+			mlx5_core_warn(priv->mdev, "Failed to update slow path (encap) flow, %d\n",
+				       err);
+			continue;
+		}
+
+		mlx5e_tc_unoffload_fdb_rules(esw, flow, flow->attr);
+		mlx5e_tc_unoffload_flow_post_acts(flow);
+		flow->rule[0] = rule;
+		/* was unset when fast path rule removed */
+		flow_flag_set(flow, OFFLOADED);
+	}
+
+	/* we know that the encap is valid */
+	e->flags &= ~MLX5_ENCAP_ENTRY_VALID;
+	mlx5_packet_reformat_dealloc(priv->mdev, e->pkt_reformat);
+	e->pkt_reformat = NULL;
+}
+
+static void mlx5e_take_tmp_flow(struct mlx5e_tc_flow *flow,
+				struct list_head *flow_list,
+				int index)
+{
+	if (IS_ERR(mlx5e_flow_get(flow))) {
+		/* Flow is being deleted concurrently. Wait for it to be
+		 * unoffloaded from hardware, otherwise deleting encap will
+		 * fail.
+		 */
+		wait_for_completion(&flow->del_hw_done);
+		return;
+	}
+	wait_for_completion(&flow->init_done);
+
+	flow->tmp_entry_index = index;
+	list_add(&flow->tmp_list, flow_list);
+}
+
+/* Takes reference to all flows attached to encap and adds the flows to
+ * flow_list using 'tmp_list' list_head in mlx5e_tc_flow.
+ */
+void mlx5e_take_all_encap_flows(struct mlx5e_encap_entry *e, struct list_head *flow_list)
+{
+	struct encap_flow_item *efi;
+	struct mlx5e_tc_flow *flow;
+
+	list_for_each_entry(efi, &e->flows, list) {
+		flow = container_of(efi, struct mlx5e_tc_flow, encaps[efi->index]);
+		mlx5e_take_tmp_flow(flow, flow_list, efi->index);
+	}
+}
+
+/* Takes reference to all flows attached to route and adds the flows to
+ * flow_list using 'tmp_list' list_head in mlx5e_tc_flow.
+ */
+static void mlx5e_take_all_route_decap_flows(struct mlx5e_route_entry *r,
+					     struct list_head *flow_list)
+{
+	struct mlx5e_tc_flow *flow;
+
+	list_for_each_entry(flow, &r->decap_flows, decap_routes)
+		mlx5e_take_tmp_flow(flow, flow_list, 0);
+}
+
+typedef bool (match_cb)(struct mlx5e_encap_entry *);
+
+static struct mlx5e_encap_entry *
+mlx5e_get_next_matching_encap(struct mlx5e_neigh_hash_entry *nhe,
+			      struct mlx5e_encap_entry *e,
+			      match_cb match)
+{
+	struct mlx5e_encap_entry *next = NULL;
+
+retry:
+	rcu_read_lock();
+
+	/* find encap with non-zero reference counter value */
+	for (next = e ?
+		     list_next_or_null_rcu(&nhe->encap_list,
+					   &e->encap_list,
+					   struct mlx5e_encap_entry,
+					   encap_list) :
+		     list_first_or_null_rcu(&nhe->encap_list,
+					    struct mlx5e_encap_entry,
+					    encap_list);
+	     next;
+	     next = list_next_or_null_rcu(&nhe->encap_list,
+					  &next->encap_list,
+					  struct mlx5e_encap_entry,
+					  encap_list))
+		if (mlx5e_encap_take(next))
+			break;
+
+	rcu_read_unlock();
+
+	/* release starting encap */
+	if (e)
+		mlx5e_encap_put(netdev_priv(e->out_dev), e);
+	if (!next)
+		return next;
+
+	/* wait for encap to be fully initialized */
+	wait_for_completion(&next->res_ready);
+	/* continue searching if encap entry is not in valid state after completion */
+	if (!match(next)) {
+		e = next;
+		goto retry;
+	}
+
+	return next;
+}
+
+static bool mlx5e_encap_valid(struct mlx5e_encap_entry *e)
+{
+	return e->flags & MLX5_ENCAP_ENTRY_VALID;
+}
+
+static struct mlx5e_encap_entry *
+mlx5e_get_next_valid_encap(struct mlx5e_neigh_hash_entry *nhe,
+			   struct mlx5e_encap_entry *e)
+{
+	return mlx5e_get_next_matching_encap(nhe, e, mlx5e_encap_valid);
+}
+
+static bool mlx5e_encap_initialized(struct mlx5e_encap_entry *e)
+{
+	return e->compl_result >= 0;
+}
+
+struct mlx5e_encap_entry *
+mlx5e_get_next_init_encap(struct mlx5e_neigh_hash_entry *nhe,
+			  struct mlx5e_encap_entry *e)
+{
+	return mlx5e_get_next_matching_encap(nhe, e, mlx5e_encap_initialized);
+}
+
+void mlx5e_tc_update_neigh_used_value(struct mlx5e_neigh_hash_entry *nhe)
+{
+	struct mlx5e_neigh *m_neigh = &nhe->m_neigh;
+	struct mlx5e_encap_entry *e = NULL;
+	struct mlx5e_tc_flow *flow;
+	struct mlx5_fc *counter;
+	struct neigh_table *tbl;
+	bool neigh_used = false;
+	struct neighbour *n;
+	u64 lastuse;
+
+	if (m_neigh->family == AF_INET)
+		tbl = &arp_tbl;
+#if IS_ENABLED(CONFIG_IPV6)
+	else if (m_neigh->family == AF_INET6)
+		tbl = ipv6_stub->nd_tbl;
+#endif
+	else
+		return;
+
+	/* mlx5e_get_next_valid_encap() releases previous encap before returning
+	 * next one.
+	 */
+	while ((e = mlx5e_get_next_valid_encap(nhe, e)) != NULL) {
+		struct mlx5e_priv *priv = netdev_priv(e->out_dev);
+		struct encap_flow_item *efi, *tmp;
+		struct mlx5_eswitch *esw;
+		LIST_HEAD(flow_list);
+
+		esw = priv->mdev->priv.eswitch;
+		mutex_lock(&esw->offloads.encap_tbl_lock);
+		list_for_each_entry_safe(efi, tmp, &e->flows, list) {
+			flow = container_of(efi, struct mlx5e_tc_flow,
+					    encaps[efi->index]);
+			if (IS_ERR(mlx5e_flow_get(flow)))
+				continue;
+			list_add(&flow->tmp_list, &flow_list);
+
+			if (mlx5e_is_offloaded_flow(flow)) {
+				counter = mlx5e_tc_get_counter(flow);
+				lastuse = mlx5_fc_query_lastuse(counter);
+				if (time_after((unsigned long)lastuse, nhe->reported_lastuse)) {
+					neigh_used = true;
+					break;
+				}
+			}
+		}
+		mutex_unlock(&esw->offloads.encap_tbl_lock);
+
+		mlx5e_put_flow_list(priv, &flow_list);
+		if (neigh_used) {
+			/* release current encap before breaking the loop */
+			mlx5e_encap_put(priv, e);
+			break;
+		}
+	}
+
+	trace_mlx5e_tc_update_neigh_used_value(nhe, neigh_used);
+
+	if (neigh_used) {
+		nhe->reported_lastuse = jiffies;
+
+		/* find the relevant neigh according to the cached device and
+		 * dst ip pair
+		 */
+		n = neigh_lookup(tbl, &m_neigh->dst_ip, READ_ONCE(nhe->neigh_dev));
+		if (!n)
+			return;
+
+		neigh_event_send(n, NULL);
+		neigh_release(n);
+	}
+}
+
+static void mlx5e_encap_dealloc(struct mlx5e_priv *priv, struct mlx5e_encap_entry *e)
+{
+	WARN_ON(!list_empty(&e->flows));
+
+	if (e->compl_result > 0) {
+		mlx5e_rep_encap_entry_detach(netdev_priv(e->out_dev), e);
+
+		if (e->flags & MLX5_ENCAP_ENTRY_VALID)
+			mlx5_packet_reformat_dealloc(priv->mdev, e->pkt_reformat);
+	}
+
+	kfree(e->tun_info);
+	kfree(e->encap_header);
+	kfree_rcu(e, rcu);
+}
+
+static void mlx5e_decap_dealloc(struct mlx5e_priv *priv,
+				struct mlx5e_decap_entry *d)
+{
+	WARN_ON(!list_empty(&d->flows));
+
+	if (!d->compl_result)
+		mlx5_packet_reformat_dealloc(priv->mdev, d->pkt_reformat);
+
+	kfree_rcu(d, rcu);
+}
+
+void mlx5e_encap_put(struct mlx5e_priv *priv, struct mlx5e_encap_entry *e)
+{
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+
+	if (!refcount_dec_and_mutex_lock(&e->refcnt, &esw->offloads.encap_tbl_lock))
+		return;
+	list_del(&e->route_list);
+	hash_del_rcu(&e->encap_hlist);
+	mutex_unlock(&esw->offloads.encap_tbl_lock);
+
+	mlx5e_encap_dealloc(priv, e);
+}
+
+static void mlx5e_decap_put(struct mlx5e_priv *priv, struct mlx5e_decap_entry *d)
+{
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+
+	if (!refcount_dec_and_mutex_lock(&d->refcnt, &esw->offloads.decap_tbl_lock))
+		return;
+	hash_del_rcu(&d->hlist);
+	mutex_unlock(&esw->offloads.decap_tbl_lock);
+
+	mlx5e_decap_dealloc(priv, d);
+}
+
+static void mlx5e_detach_encap_route(struct mlx5e_priv *priv,
+				     struct mlx5e_tc_flow *flow,
+				     int out_index);
+
+void mlx5e_detach_encap(struct mlx5e_priv *priv,
+			struct mlx5e_tc_flow *flow,
+			struct mlx5_flow_attr *attr,
+			int out_index)
+{
+	struct mlx5e_encap_entry *e = flow->encaps[out_index].e;
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+
+	if (!mlx5e_is_eswitch_flow(flow))
+		return;
+
+	if (attr->esw_attr->dests[out_index].flags &
+	    MLX5_ESW_DEST_CHAIN_WITH_SRC_PORT_CHANGE)
+		mlx5e_detach_encap_route(priv, flow, out_index);
+
+	/* flow wasn't fully initialized */
+	if (!e)
+		return;
+
+	mutex_lock(&esw->offloads.encap_tbl_lock);
+	list_del(&flow->encaps[out_index].list);
+	flow->encaps[out_index].e = NULL;
+	if (!refcount_dec_and_test(&e->refcnt)) {
+		mutex_unlock(&esw->offloads.encap_tbl_lock);
+		return;
+	}
+	list_del(&e->route_list);
+	hash_del_rcu(&e->encap_hlist);
+	mutex_unlock(&esw->offloads.encap_tbl_lock);
+
+	mlx5e_encap_dealloc(priv, e);
+}
+
+void mlx5e_detach_decap(struct mlx5e_priv *priv,
+			struct mlx5e_tc_flow *flow)
+{
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	struct mlx5e_decap_entry *d = flow->decap_reformat;
+
+	if (!d)
+		return;
+
+	mutex_lock(&esw->offloads.decap_tbl_lock);
+	list_del(&flow->l3_to_l2_reformat);
+	flow->decap_reformat = NULL;
+
+	if (!refcount_dec_and_test(&d->refcnt)) {
+		mutex_unlock(&esw->offloads.decap_tbl_lock);
+		return;
+	}
+	hash_del_rcu(&d->hlist);
+	mutex_unlock(&esw->offloads.decap_tbl_lock);
+
+	mlx5e_decap_dealloc(priv, d);
+}
+
+bool mlx5e_tc_tun_encap_info_equal_generic(struct mlx5e_encap_key *a,
+					   struct mlx5e_encap_key *b)
+{
+	return memcmp(a->ip_tun_key, b->ip_tun_key, sizeof(*a->ip_tun_key)) == 0 &&
+		a->tc_tunnel->tunnel_type == b->tc_tunnel->tunnel_type;
+}
+
+static int cmp_decap_info(struct mlx5e_decap_key *a,
+			  struct mlx5e_decap_key *b)
+{
+	return memcmp(&a->key, &b->key, sizeof(b->key));
+}
+
+static int hash_encap_info(struct mlx5e_encap_key *key)
+{
+	return jhash(key->ip_tun_key, sizeof(*key->ip_tun_key),
+		     key->tc_tunnel->tunnel_type);
+}
+
+static int hash_decap_info(struct mlx5e_decap_key *key)
+{
+	return jhash(&key->key, sizeof(key->key), 0);
+}
+
+bool mlx5e_encap_take(struct mlx5e_encap_entry *e)
+{
+	return refcount_inc_not_zero(&e->refcnt);
+}
+
+static bool mlx5e_decap_take(struct mlx5e_decap_entry *e)
+{
+	return refcount_inc_not_zero(&e->refcnt);
+}
+
+static struct mlx5e_encap_entry *
+mlx5e_encap_get(struct mlx5e_priv *priv, struct mlx5e_encap_key *key,
+		uintptr_t hash_key)
+{
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	struct mlx5e_encap_key e_key;
+	struct mlx5e_encap_entry *e;
+
+	hash_for_each_possible_rcu(esw->offloads.encap_tbl, e,
+				   encap_hlist, hash_key) {
+		e_key.ip_tun_key = &e->tun_info->key;
+		e_key.tc_tunnel = e->tunnel;
+		if (e->tunnel->encap_info_equal(&e_key, key) &&
+		    mlx5e_encap_take(e))
+			return e;
+	}
+
+	return NULL;
+}
+
+static struct mlx5e_decap_entry *
+mlx5e_decap_get(struct mlx5e_priv *priv, struct mlx5e_decap_key *key,
+		uintptr_t hash_key)
+{
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	struct mlx5e_decap_key r_key;
+	struct mlx5e_decap_entry *e;
+
+	hash_for_each_possible_rcu(esw->offloads.decap_tbl, e,
+				   hlist, hash_key) {
+		r_key = e->key;
+		if (!cmp_decap_info(&r_key, key) &&
+		    mlx5e_decap_take(e))
+			return e;
+	}
+	return NULL;
+}
+
+struct ip_tunnel_info *mlx5e_dup_tun_info(const struct ip_tunnel_info *tun_info)
+{
+	size_t tun_size = sizeof(*tun_info) + tun_info->options_len;
+
+	return kmemdup(tun_info, tun_size, GFP_KERNEL);
+}
+
+static bool is_duplicated_encap_entry(struct mlx5e_priv *priv,
+				      struct mlx5e_tc_flow *flow,
+				      int out_index,
+				      struct mlx5e_encap_entry *e,
+				      struct netlink_ext_ack *extack)
+{
+	int i;
+
+	for (i = 0; i < out_index; i++) {
+		if (flow->encaps[i].e != e)
+			continue;
+		NL_SET_ERR_MSG_MOD(extack, "can't duplicate encap action");
+		netdev_err(priv->netdev, "can't duplicate encap action\n");
+		return true;
+	}
+
+	return false;
+}
+
+static int mlx5e_set_vf_tunnel(struct mlx5_eswitch *esw,
+			       struct mlx5_flow_attr *attr,
+			       struct mlx5e_tc_mod_hdr_acts *mod_hdr_acts,
+			       struct net_device *out_dev,
+			       int route_dev_ifindex,
+			       int out_index)
+{
+	struct mlx5_esw_flow_attr *esw_attr = attr->esw_attr;
+	struct net_device *route_dev;
+	u16 vport_num;
+	int err = 0;
+	u32 data;
+
+	route_dev = dev_get_by_index(dev_net(out_dev), route_dev_ifindex);
+
+	if (!route_dev || route_dev->netdev_ops != &mlx5e_netdev_ops ||
+	    !mlx5e_tc_is_vf_tunnel(out_dev, route_dev))
+		goto out;
+
+	err = mlx5e_tc_query_route_vport(out_dev, route_dev, &vport_num);
+	if (err)
+		goto out;
+
+	attr->dest_chain = 0;
+	attr->action |= MLX5_FLOW_CONTEXT_ACTION_MOD_HDR;
+	esw_attr->dests[out_index].flags |= MLX5_ESW_DEST_CHAIN_WITH_SRC_PORT_CHANGE;
+	data = mlx5_eswitch_get_vport_metadata_for_set(esw_attr->in_mdev->priv.eswitch,
+						       vport_num);
+	err = mlx5e_tc_match_to_reg_set_and_get_id(esw->dev, mod_hdr_acts,
+						   MLX5_FLOW_NAMESPACE_FDB,
+						   VPORT_TO_REG, data);
+	if (err >= 0) {
+		esw_attr->dests[out_index].src_port_rewrite_act_id = err;
+		err = 0;
+	}
+
+out:
+	if (route_dev)
+		dev_put(route_dev);
+	return err;
+}
+
+static int mlx5e_update_vf_tunnel(struct mlx5_eswitch *esw,
+				  struct mlx5_esw_flow_attr *attr,
+				  struct mlx5e_tc_mod_hdr_acts *mod_hdr_acts,
+				  struct net_device *out_dev,
+				  int route_dev_ifindex,
+				  int out_index)
+{
+	int act_id = attr->dests[out_index].src_port_rewrite_act_id;
+	struct net_device *route_dev;
+	u16 vport_num;
+	int err = 0;
+	u32 data;
+
+	route_dev = dev_get_by_index(dev_net(out_dev), route_dev_ifindex);
+
+	if (!route_dev || route_dev->netdev_ops != &mlx5e_netdev_ops ||
+	    !mlx5e_tc_is_vf_tunnel(out_dev, route_dev)) {
+		err = -ENODEV;
+		goto out;
+	}
+
+	err = mlx5e_tc_query_route_vport(out_dev, route_dev, &vport_num);
+	if (err)
+		goto out;
+
+	data = mlx5_eswitch_get_vport_metadata_for_set(attr->in_mdev->priv.eswitch,
+						       vport_num);
+	mlx5e_tc_match_to_reg_mod_hdr_change(esw->dev, mod_hdr_acts, VPORT_TO_REG, act_id, data);
+
+out:
+	if (route_dev)
+		dev_put(route_dev);
+	return err;
+}
+
+static unsigned int mlx5e_route_tbl_get_last_update(struct mlx5e_priv *priv)
+{
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	struct mlx5_rep_uplink_priv *uplink_priv;
+	struct mlx5e_rep_priv *uplink_rpriv;
+	struct mlx5e_tc_tun_encap *encap;
+	unsigned int ret;
+
+	uplink_rpriv = mlx5_eswitch_get_uplink_priv(esw, REP_ETH);
+	uplink_priv = &uplink_rpriv->uplink_priv;
+	encap = uplink_priv->encap;
+
+	spin_lock_bh(&encap->route_lock);
+	ret = encap->route_tbl_last_update;
+	spin_unlock_bh(&encap->route_lock);
+	return ret;
+}
+
+static int mlx5e_attach_encap_route(struct mlx5e_priv *priv,
+				    struct mlx5e_tc_flow *flow,
+				    struct mlx5_flow_attr *attr,
+				    struct mlx5e_encap_entry *e,
+				    bool new_encap_entry,
+				    unsigned long tbl_time_before,
+				    int out_index);
+
+int mlx5e_attach_encap(struct mlx5e_priv *priv,
+		       struct mlx5e_tc_flow *flow,
+		       struct mlx5_flow_attr *attr,
+		       struct net_device *mirred_dev,
+		       int out_index,
+		       struct netlink_ext_ack *extack,
+		       struct net_device **encap_dev)
+{
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	struct mlx5e_tc_flow_parse_attr *parse_attr;
+	const struct ip_tunnel_info *tun_info;
+	const struct mlx5e_mpls_info *mpls_info;
+	unsigned long tbl_time_before = 0;
+	struct mlx5e_encap_entry *e;
+	struct mlx5e_encap_key key;
+	bool entry_created = false;
+	unsigned short family;
+	uintptr_t hash_key;
+	int err = 0;
+
+	parse_attr = attr->parse_attr;
+	tun_info = parse_attr->tun_info[out_index];
+	mpls_info = &parse_attr->mpls_info[out_index];
+	family = ip_tunnel_info_af(tun_info);
+	key.ip_tun_key = &tun_info->key;
+	key.tc_tunnel = mlx5e_get_tc_tun(mirred_dev);
+	if (!key.tc_tunnel) {
+		NL_SET_ERR_MSG_MOD(extack, "Unsupported tunnel");
+		return -EOPNOTSUPP;
+	}
+
+	hash_key = hash_encap_info(&key);
+
+	mutex_lock(&esw->offloads.encap_tbl_lock);
+	e = mlx5e_encap_get(priv, &key, hash_key);
+
+	/* must verify if encap is valid or not */
+	if (e) {
+		/* Check that entry was not already attached to this flow */
+		if (is_duplicated_encap_entry(priv, flow, out_index, e, extack)) {
+			err = -EOPNOTSUPP;
+			goto out_err;
+		}
+
+		mutex_unlock(&esw->offloads.encap_tbl_lock);
+		wait_for_completion(&e->res_ready);
+
+		/* Protect against concurrent neigh update. */
+		mutex_lock(&esw->offloads.encap_tbl_lock);
+		if (e->compl_result < 0) {
+			err = -EREMOTEIO;
+			goto out_err;
+		}
+		goto attach_flow;
+	}
+
+	e = kzalloc(sizeof(*e), GFP_KERNEL);
+	if (!e) {
+		err = -ENOMEM;
+		goto out_err;
+	}
+
+	refcount_set(&e->refcnt, 1);
+	init_completion(&e->res_ready);
+	entry_created = true;
+	INIT_LIST_HEAD(&e->route_list);
+
+	tun_info = mlx5e_dup_tun_info(tun_info);
+	if (!tun_info) {
+		err = -ENOMEM;
+		goto out_err_init;
+	}
+	e->tun_info = tun_info;
+	memcpy(&e->mpls_info, mpls_info, sizeof(*mpls_info));
+	err = mlx5e_tc_tun_init_encap_attr(mirred_dev, priv, e, extack);
+	if (err)
+		goto out_err_init;
+
+	INIT_LIST_HEAD(&e->flows);
+	hash_add_rcu(esw->offloads.encap_tbl, &e->encap_hlist, hash_key);
+	tbl_time_before = mlx5e_route_tbl_get_last_update(priv);
+	mutex_unlock(&esw->offloads.encap_tbl_lock);
+
+	if (family == AF_INET)
+		err = mlx5e_tc_tun_create_header_ipv4(priv, mirred_dev, e);
+	else if (family == AF_INET6)
+		err = mlx5e_tc_tun_create_header_ipv6(priv, mirred_dev, e);
+
+	/* Protect against concurrent neigh update. */
+	mutex_lock(&esw->offloads.encap_tbl_lock);
+	complete_all(&e->res_ready);
+	if (err) {
+		e->compl_result = err;
+		goto out_err;
+	}
+	e->compl_result = 1;
+
+attach_flow:
+	err = mlx5e_attach_encap_route(priv, flow, attr, e, entry_created,
+				       tbl_time_before, out_index);
+	if (err)
+		goto out_err;
+
+	err = mlx5e_set_int_port_tunnel(priv, attr, e, out_index);
+	if (err == -EOPNOTSUPP) {
+		/* If device doesn't support int port offload,
+		 * redirect to uplink vport.
+		 */
+		mlx5_core_dbg(priv->mdev, "attaching int port as encap dev not supported, using uplink\n");
+		err = 0;
+	} else if (err) {
+		goto out_err;
+	}
+
+	flow->encaps[out_index].e = e;
+	list_add(&flow->encaps[out_index].list, &e->flows);
+	flow->encaps[out_index].index = out_index;
+	*encap_dev = e->out_dev;
+	if (e->flags & MLX5_ENCAP_ENTRY_VALID) {
+		attr->esw_attr->dests[out_index].pkt_reformat = e->pkt_reformat;
+		attr->esw_attr->dests[out_index].flags |= MLX5_ESW_DEST_ENCAP_VALID;
+	} else {
+		flow_flag_set(flow, SLOW);
+	}
+	mutex_unlock(&esw->offloads.encap_tbl_lock);
+
+	return err;
+
+out_err:
+	mutex_unlock(&esw->offloads.encap_tbl_lock);
+	if (e)
+		mlx5e_encap_put(priv, e);
+	return err;
+
+out_err_init:
+	mutex_unlock(&esw->offloads.encap_tbl_lock);
+	kfree(tun_info);
+	kfree(e);
+	return err;
+}
+
+int mlx5e_attach_decap(struct mlx5e_priv *priv,
+		       struct mlx5e_tc_flow *flow,
+		       struct netlink_ext_ack *extack)
+{
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	struct mlx5_esw_flow_attr *attr = flow->attr->esw_attr;
+	struct mlx5_pkt_reformat_params reformat_params;
+	struct mlx5e_decap_entry *d;
+	struct mlx5e_decap_key key;
+	uintptr_t hash_key;
+	int err = 0;
+
+	if (sizeof(attr->eth) > MLX5_CAP_ESW(priv->mdev, max_encap_header_size)) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "encap header larger than max supported");
+		return -EOPNOTSUPP;
+	}
+
+	key.key = attr->eth;
+	hash_key = hash_decap_info(&key);
+	mutex_lock(&esw->offloads.decap_tbl_lock);
+	d = mlx5e_decap_get(priv, &key, hash_key);
+	if (d) {
+		mutex_unlock(&esw->offloads.decap_tbl_lock);
+		wait_for_completion(&d->res_ready);
+		mutex_lock(&esw->offloads.decap_tbl_lock);
+		if (d->compl_result) {
+			err = -EREMOTEIO;
+			goto out_free;
+		}
+		goto found;
+	}
+
+	d = kzalloc(sizeof(*d), GFP_KERNEL);
+	if (!d) {
+		err = -ENOMEM;
+		goto out_err;
+	}
+
+	d->key = key;
+	refcount_set(&d->refcnt, 1);
+	init_completion(&d->res_ready);
+	INIT_LIST_HEAD(&d->flows);
+	hash_add_rcu(esw->offloads.decap_tbl, &d->hlist, hash_key);
+	mutex_unlock(&esw->offloads.decap_tbl_lock);
+
+	memset(&reformat_params, 0, sizeof(reformat_params));
+	reformat_params.type = MLX5_REFORMAT_TYPE_L3_TUNNEL_TO_L2;
+	reformat_params.size = sizeof(attr->eth);
+	reformat_params.data = &attr->eth;
+	d->pkt_reformat = mlx5_packet_reformat_alloc(priv->mdev,
+						     &reformat_params,
+						     MLX5_FLOW_NAMESPACE_FDB);
+	if (IS_ERR(d->pkt_reformat)) {
+		err = PTR_ERR(d->pkt_reformat);
+		d->compl_result = err;
+	}
+	mutex_lock(&esw->offloads.decap_tbl_lock);
+	complete_all(&d->res_ready);
+	if (err)
+		goto out_free;
+
+found:
+	flow->decap_reformat = d;
+	attr->decap_pkt_reformat = d->pkt_reformat;
+	list_add(&flow->l3_to_l2_reformat, &d->flows);
+	mutex_unlock(&esw->offloads.decap_tbl_lock);
+	return 0;
+
+out_free:
+	mutex_unlock(&esw->offloads.decap_tbl_lock);
+	mlx5e_decap_put(priv, d);
+	return err;
+
+out_err:
+	mutex_unlock(&esw->offloads.decap_tbl_lock);
+	return err;
+}
+
+static int cmp_route_info(struct mlx5e_route_key *a,
+			  struct mlx5e_route_key *b)
+{
+	if (a->ip_version == 4 && b->ip_version == 4)
+		return memcmp(&a->endpoint_ip.v4, &b->endpoint_ip.v4,
+			      sizeof(a->endpoint_ip.v4));
+	else if (a->ip_version == 6 && b->ip_version == 6)
+		return memcmp(&a->endpoint_ip.v6, &b->endpoint_ip.v6,
+			      sizeof(a->endpoint_ip.v6));
+	return 1;
+}
+
+static u32 hash_route_info(struct mlx5e_route_key *key)
+{
+	if (key->ip_version == 4)
+		return jhash(&key->endpoint_ip.v4, sizeof(key->endpoint_ip.v4), 0);
+	return jhash(&key->endpoint_ip.v6, sizeof(key->endpoint_ip.v6), 0);
+}
+
+static void mlx5e_route_dealloc(struct mlx5e_priv *priv,
+				struct mlx5e_route_entry *r)
+{
+	WARN_ON(!list_empty(&r->decap_flows));
+	WARN_ON(!list_empty(&r->encap_entries));
+
+	kfree_rcu(r, rcu);
+}
+
+static void mlx5e_route_put(struct mlx5e_priv *priv, struct mlx5e_route_entry *r)
+{
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+
+	if (!refcount_dec_and_mutex_lock(&r->refcnt, &esw->offloads.encap_tbl_lock))
+		return;
+
+	hash_del_rcu(&r->hlist);
+	mutex_unlock(&esw->offloads.encap_tbl_lock);
+
+	mlx5e_route_dealloc(priv, r);
+}
+
+static void mlx5e_route_put_locked(struct mlx5e_priv *priv, struct mlx5e_route_entry *r)
+{
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+
+	lockdep_assert_held(&esw->offloads.encap_tbl_lock);
+
+	if (!refcount_dec_and_test(&r->refcnt))
+		return;
+	hash_del_rcu(&r->hlist);
+	mlx5e_route_dealloc(priv, r);
+}
+
+static struct mlx5e_route_entry *
+mlx5e_route_get(struct mlx5e_tc_tun_encap *encap, struct mlx5e_route_key *key,
+		u32 hash_key)
+{
+	struct mlx5e_route_key r_key;
+	struct mlx5e_route_entry *r;
+
+	hash_for_each_possible(encap->route_tbl, r, hlist, hash_key) {
+		r_key = r->key;
+		if (!cmp_route_info(&r_key, key) &&
+		    refcount_inc_not_zero(&r->refcnt))
+			return r;
+	}
+	return NULL;
+}
+
+static struct mlx5e_route_entry *
+mlx5e_route_get_create(struct mlx5e_priv *priv,
+		       struct mlx5e_route_key *key,
+		       int tunnel_dev_index,
+		       unsigned long *route_tbl_change_time)
+{
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	struct mlx5_rep_uplink_priv *uplink_priv;
+	struct mlx5e_rep_priv *uplink_rpriv;
+	struct mlx5e_tc_tun_encap *encap;
+	struct mlx5e_route_entry *r;
+	u32 hash_key;
+
+	uplink_rpriv = mlx5_eswitch_get_uplink_priv(esw, REP_ETH);
+	uplink_priv = &uplink_rpriv->uplink_priv;
+	encap = uplink_priv->encap;
+
+	hash_key = hash_route_info(key);
+	spin_lock_bh(&encap->route_lock);
+	r = mlx5e_route_get(encap, key, hash_key);
+	spin_unlock_bh(&encap->route_lock);
+	if (r) {
+		if (!mlx5e_route_entry_valid(r)) {
+			mlx5e_route_put_locked(priv, r);
+			return ERR_PTR(-EINVAL);
+		}
+		return r;
+	}
+
+	r = kzalloc(sizeof(*r), GFP_KERNEL);
+	if (!r)
+		return ERR_PTR(-ENOMEM);
+
+	r->key = *key;
+	r->flags |= MLX5E_ROUTE_ENTRY_VALID;
+	r->tunnel_dev_index = tunnel_dev_index;
+	refcount_set(&r->refcnt, 1);
+	INIT_LIST_HEAD(&r->decap_flows);
+	INIT_LIST_HEAD(&r->encap_entries);
+
+	spin_lock_bh(&encap->route_lock);
+	*route_tbl_change_time = encap->route_tbl_last_update;
+	hash_add(encap->route_tbl, &r->hlist, hash_key);
+	spin_unlock_bh(&encap->route_lock);
+
+	return r;
+}
+
+static struct mlx5e_route_entry *
+mlx5e_route_lookup_for_update(struct mlx5e_tc_tun_encap *encap, struct mlx5e_route_key *key)
+{
+	u32 hash_key = hash_route_info(key);
+	struct mlx5e_route_entry *r;
+
+	spin_lock_bh(&encap->route_lock);
+	encap->route_tbl_last_update = jiffies;
+	r = mlx5e_route_get(encap, key, hash_key);
+	spin_unlock_bh(&encap->route_lock);
+
+	return r;
+}
+
+struct mlx5e_tc_fib_event_data {
+	struct work_struct work;
+	unsigned long event;
+	struct mlx5e_route_entry *r;
+	struct net_device *ul_dev;
+};
+
+static void mlx5e_tc_fib_event_work(struct work_struct *work);
+static struct mlx5e_tc_fib_event_data *
+mlx5e_tc_init_fib_work(unsigned long event, struct net_device *ul_dev, gfp_t flags)
+{
+	struct mlx5e_tc_fib_event_data *fib_work;
+
+	fib_work = kzalloc(sizeof(*fib_work), flags);
+	if (WARN_ON(!fib_work))
+		return NULL;
+
+	INIT_WORK(&fib_work->work, mlx5e_tc_fib_event_work);
+	fib_work->event = event;
+	fib_work->ul_dev = ul_dev;
+
+	return fib_work;
+}
+
+static int
+mlx5e_route_enqueue_update(struct mlx5e_priv *priv,
+			   struct mlx5e_route_entry *r,
+			   unsigned long event)
+{
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	struct mlx5e_tc_fib_event_data *fib_work;
+	struct mlx5e_rep_priv *uplink_rpriv;
+	struct net_device *ul_dev;
+
+	uplink_rpriv = mlx5_eswitch_get_uplink_priv(esw, REP_ETH);
+	ul_dev = uplink_rpriv->netdev;
+
+	fib_work = mlx5e_tc_init_fib_work(event, ul_dev, GFP_KERNEL);
+	if (!fib_work)
+		return -ENOMEM;
+
+	dev_hold(ul_dev);
+	refcount_inc(&r->refcnt);
+	fib_work->r = r;
+	queue_work(priv->wq, &fib_work->work);
+
+	return 0;
+}
+
+int mlx5e_attach_decap_route(struct mlx5e_priv *priv,
+			     struct mlx5e_tc_flow *flow)
+{
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	unsigned long tbl_time_before, tbl_time_after;
+	struct mlx5e_tc_flow_parse_attr *parse_attr;
+	struct mlx5_flow_attr *attr = flow->attr;
+	struct mlx5_esw_flow_attr *esw_attr;
+	struct mlx5e_route_entry *r;
+	struct mlx5e_route_key key;
+	int err = 0;
+
+	esw_attr = attr->esw_attr;
+	parse_attr = attr->parse_attr;
+	mutex_lock(&esw->offloads.encap_tbl_lock);
+	if (!esw_attr->rx_tun_attr)
+		goto out;
+
+	tbl_time_before = mlx5e_route_tbl_get_last_update(priv);
+	tbl_time_after = tbl_time_before;
+	err = mlx5e_tc_tun_route_lookup(priv, &parse_attr->spec, attr, parse_attr->filter_dev);
+	if (err || !esw_attr->rx_tun_attr->decap_vport)
+		goto out;
+
+	key.ip_version = attr->tun_ip_version;
+	if (key.ip_version == 4)
+		key.endpoint_ip.v4 = esw_attr->rx_tun_attr->dst_ip.v4;
+	else
+		key.endpoint_ip.v6 = esw_attr->rx_tun_attr->dst_ip.v6;
+
+	r = mlx5e_route_get_create(priv, &key, parse_attr->filter_dev->ifindex,
+				   &tbl_time_after);
+	if (IS_ERR(r)) {
+		err = PTR_ERR(r);
+		goto out;
+	}
+	/* Routing changed concurrently. FIB event handler might have missed new
+	 * entry, schedule update.
+	 */
+	if (tbl_time_before != tbl_time_after) {
+		err = mlx5e_route_enqueue_update(priv, r, FIB_EVENT_ENTRY_REPLACE);
+		if (err) {
+			mlx5e_route_put_locked(priv, r);
+			goto out;
+		}
+	}
+
+	flow->decap_route = r;
+	list_add(&flow->decap_routes, &r->decap_flows);
+	mutex_unlock(&esw->offloads.encap_tbl_lock);
+	return 0;
+
+out:
+	mutex_unlock(&esw->offloads.encap_tbl_lock);
+	return err;
+}
+
+static int mlx5e_attach_encap_route(struct mlx5e_priv *priv,
+				    struct mlx5e_tc_flow *flow,
+				    struct mlx5_flow_attr *attr,
+				    struct mlx5e_encap_entry *e,
+				    bool new_encap_entry,
+				    unsigned long tbl_time_before,
+				    int out_index)
+{
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	unsigned long tbl_time_after = tbl_time_before;
+	struct mlx5e_tc_flow_parse_attr *parse_attr;
+	const struct ip_tunnel_info *tun_info;
+	struct mlx5_esw_flow_attr *esw_attr;
+	struct mlx5e_route_entry *r;
+	struct mlx5e_route_key key;
+	unsigned short family;
+	int err = 0;
+
+	esw_attr = attr->esw_attr;
+	parse_attr = attr->parse_attr;
+	tun_info = parse_attr->tun_info[out_index];
+	family = ip_tunnel_info_af(tun_info);
+
+	if (family == AF_INET) {
+		key.endpoint_ip.v4 = tun_info->key.u.ipv4.src;
+		key.ip_version = 4;
+	} else if (family == AF_INET6) {
+		key.endpoint_ip.v6 = tun_info->key.u.ipv6.src;
+		key.ip_version = 6;
+	}
+
+	err = mlx5e_set_vf_tunnel(esw, attr, &parse_attr->mod_hdr_acts, e->out_dev,
+				  e->route_dev_ifindex, out_index);
+	if (err || !(esw_attr->dests[out_index].flags &
+		     MLX5_ESW_DEST_CHAIN_WITH_SRC_PORT_CHANGE))
+		return err;
+
+	r = mlx5e_route_get_create(priv, &key, parse_attr->mirred_ifindex[out_index],
+				   &tbl_time_after);
+	if (IS_ERR(r))
+		return PTR_ERR(r);
+	/* Routing changed concurrently. FIB event handler might have missed new
+	 * entry, schedule update.
+	 */
+	if (tbl_time_before != tbl_time_after) {
+		err = mlx5e_route_enqueue_update(priv, r, FIB_EVENT_ENTRY_REPLACE);
+		if (err) {
+			mlx5e_route_put_locked(priv, r);
+			return err;
+		}
+	}
+
+	flow->encap_routes[out_index].r = r;
+	if (new_encap_entry)
+		list_add(&e->route_list, &r->encap_entries);
+	flow->encap_routes[out_index].index = out_index;
+	return 0;
+}
+
+void mlx5e_detach_decap_route(struct mlx5e_priv *priv,
+			      struct mlx5e_tc_flow *flow)
+{
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	struct mlx5e_route_entry *r = flow->decap_route;
+
+	if (!r)
+		return;
+
+	mutex_lock(&esw->offloads.encap_tbl_lock);
+	list_del(&flow->decap_routes);
+	flow->decap_route = NULL;
+
+	if (!refcount_dec_and_test(&r->refcnt)) {
+		mutex_unlock(&esw->offloads.encap_tbl_lock);
+		return;
+	}
+	hash_del_rcu(&r->hlist);
+	mutex_unlock(&esw->offloads.encap_tbl_lock);
+
+	mlx5e_route_dealloc(priv, r);
+}
+
+static void mlx5e_detach_encap_route(struct mlx5e_priv *priv,
+				     struct mlx5e_tc_flow *flow,
+				     int out_index)
+{
+	struct mlx5e_route_entry *r = flow->encap_routes[out_index].r;
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	struct mlx5e_encap_entry *e, *tmp;
+
+	if (!r)
+		return;
+
+	mutex_lock(&esw->offloads.encap_tbl_lock);
+	flow->encap_routes[out_index].r = NULL;
+
+	if (!refcount_dec_and_test(&r->refcnt)) {
+		mutex_unlock(&esw->offloads.encap_tbl_lock);
+		return;
+	}
+	list_for_each_entry_safe(e, tmp, &r->encap_entries, route_list)
+		list_del_init(&e->route_list);
+	hash_del_rcu(&r->hlist);
+	mutex_unlock(&esw->offloads.encap_tbl_lock);
+
+	mlx5e_route_dealloc(priv, r);
+}
+
+static void mlx5e_invalidate_encap(struct mlx5e_priv *priv,
+				   struct mlx5e_encap_entry *e,
+				   struct list_head *encap_flows)
+{
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	struct mlx5e_tc_flow *flow;
+
+	list_for_each_entry(flow, encap_flows, tmp_list) {
+		struct mlx5_esw_flow_attr *esw_attr;
+		struct mlx5_flow_attr *attr;
+
+		if (!mlx5e_is_offloaded_flow(flow))
+			continue;
+
+		attr = mlx5e_tc_get_encap_attr(flow);
+		esw_attr = attr->esw_attr;
+
+		if (flow_flag_test(flow, SLOW))
+			mlx5e_tc_unoffload_from_slow_path(esw, flow);
+		else
+			mlx5e_tc_unoffload_fdb_rules(esw, flow, flow->attr);
+		mlx5_modify_header_dealloc(priv->mdev, attr->modify_hdr);
+		attr->modify_hdr = NULL;
+
+		esw_attr->dests[flow->tmp_entry_index].flags &=
+			~MLX5_ESW_DEST_ENCAP_VALID;
+		esw_attr->dests[flow->tmp_entry_index].pkt_reformat = NULL;
+	}
+
+	e->flags |= MLX5_ENCAP_ENTRY_NO_ROUTE;
+	if (e->flags & MLX5_ENCAP_ENTRY_VALID) {
+		e->flags &= ~MLX5_ENCAP_ENTRY_VALID;
+		mlx5_packet_reformat_dealloc(priv->mdev, e->pkt_reformat);
+		e->pkt_reformat = NULL;
+	}
+}
+
+static void mlx5e_reoffload_encap(struct mlx5e_priv *priv,
+				  struct net_device *tunnel_dev,
+				  struct mlx5e_encap_entry *e,
+				  struct list_head *encap_flows)
+{
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	struct mlx5e_tc_flow *flow;
+	int err;
+
+	err = ip_tunnel_info_af(e->tun_info) == AF_INET ?
+		mlx5e_tc_tun_update_header_ipv4(priv, tunnel_dev, e) :
+		mlx5e_tc_tun_update_header_ipv6(priv, tunnel_dev, e);
+	if (err)
+		mlx5_core_warn(priv->mdev, "Failed to update encap header, %d", err);
+	e->flags &= ~MLX5_ENCAP_ENTRY_NO_ROUTE;
+
+	list_for_each_entry(flow, encap_flows, tmp_list) {
+		struct mlx5e_tc_flow_parse_attr *parse_attr;
+		struct mlx5_esw_flow_attr *esw_attr;
+		struct mlx5_flow_handle *rule;
+		struct mlx5_flow_attr *attr;
+		struct mlx5_flow_spec *spec;
+
+		if (flow_flag_test(flow, FAILED))
+			continue;
+
+		spec = &flow->attr->parse_attr->spec;
+
+		attr = mlx5e_tc_get_encap_attr(flow);
+		esw_attr = attr->esw_attr;
+		parse_attr = attr->parse_attr;
+
+		err = mlx5e_update_vf_tunnel(esw, esw_attr, &parse_attr->mod_hdr_acts,
+					     e->out_dev, e->route_dev_ifindex,
+					     flow->tmp_entry_index);
+		if (err) {
+			mlx5_core_warn(priv->mdev, "Failed to update VF tunnel err=%d", err);
+			continue;
+		}
+
+		err = mlx5e_tc_add_flow_mod_hdr(priv, flow, attr);
+		if (err) {
+			mlx5_core_warn(priv->mdev, "Failed to update flow mod_hdr err=%d",
+				       err);
+			continue;
+		}
+
+		if (e->flags & MLX5_ENCAP_ENTRY_VALID) {
+			esw_attr->dests[flow->tmp_entry_index].pkt_reformat = e->pkt_reformat;
+			esw_attr->dests[flow->tmp_entry_index].flags |= MLX5_ESW_DEST_ENCAP_VALID;
+			if (!mlx5e_tc_flow_all_encaps_valid(esw_attr))
+				goto offload_to_slow_path;
+
+			err = mlx5e_tc_offload_flow_post_acts(flow);
+			if (err) {
+				mlx5_core_warn(priv->mdev, "Failed to update flow post acts, %d\n",
+					       err);
+				goto offload_to_slow_path;
+			}
+
+			/* update from slow path rule to encap rule */
+			rule = mlx5e_tc_offload_fdb_rules(esw, flow, spec, flow->attr);
+			if (IS_ERR(rule)) {
+				mlx5e_tc_unoffload_flow_post_acts(flow);
+				err = PTR_ERR(rule);
+				mlx5_core_warn(priv->mdev, "Failed to update cached encapsulation flow, %d\n",
+					       err);
+			} else {
+				flow->rule[0] = rule;
+			}
+		} else {
+offload_to_slow_path:
+			rule = mlx5e_tc_offload_to_slow_path(esw, flow, spec);
+			/* mark the flow's encap dest as non-valid */
+			esw_attr->dests[flow->tmp_entry_index].flags &=
+				~MLX5_ESW_DEST_ENCAP_VALID;
+
+			if (IS_ERR(rule)) {
+				err = PTR_ERR(rule);
+				mlx5_core_warn(priv->mdev, "Failed to update slow path (encap) flow, %d\n",
+					       err);
+			} else {
+				flow->rule[0] = rule;
+			}
+		}
+		flow_flag_set(flow, OFFLOADED);
+	}
+}
+
+static int mlx5e_update_route_encaps(struct mlx5e_priv *priv,
+				     struct mlx5e_route_entry *r,
+				     struct list_head *flow_list,
+				     bool replace)
+{
+	struct net_device *tunnel_dev;
+	struct mlx5e_encap_entry *e;
+
+	tunnel_dev = __dev_get_by_index(dev_net(priv->netdev), r->tunnel_dev_index);
+	if (!tunnel_dev)
+		return -ENODEV;
+
+	list_for_each_entry(e, &r->encap_entries, route_list) {
+		LIST_HEAD(encap_flows);
+
+		mlx5e_take_all_encap_flows(e, &encap_flows);
+		if (list_empty(&encap_flows))
+			continue;
+
+		if (mlx5e_route_entry_valid(r))
+			mlx5e_invalidate_encap(priv, e, &encap_flows);
+
+		if (!replace) {
+			list_splice(&encap_flows, flow_list);
+			continue;
+		}
+
+		mlx5e_reoffload_encap(priv, tunnel_dev, e, &encap_flows);
+		list_splice(&encap_flows, flow_list);
+	}
+
+	return 0;
+}
+
+static void mlx5e_unoffload_flow_list(struct mlx5e_priv *priv,
+				      struct list_head *flow_list)
+{
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	struct mlx5e_tc_flow *flow;
+
+	list_for_each_entry(flow, flow_list, tmp_list)
+		if (mlx5e_is_offloaded_flow(flow))
+			mlx5e_tc_unoffload_fdb_rules(esw, flow, flow->attr);
+}
+
+static void mlx5e_reoffload_decap(struct mlx5e_priv *priv,
+				  struct list_head *decap_flows)
+{
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	struct mlx5e_tc_flow *flow;
+
+	list_for_each_entry(flow, decap_flows, tmp_list) {
+		struct mlx5e_tc_flow_parse_attr *parse_attr;
+		struct mlx5_flow_attr *attr = flow->attr;
+		struct mlx5_flow_handle *rule;
+		struct mlx5_flow_spec *spec;
+		int err;
+
+		if (flow_flag_test(flow, FAILED))
+			continue;
+
+		parse_attr = attr->parse_attr;
+		spec = &parse_attr->spec;
+		err = mlx5e_tc_tun_route_lookup(priv, spec, attr, parse_attr->filter_dev);
+		if (err) {
+			mlx5_core_warn(priv->mdev, "Failed to lookup route for flow, %d\n",
+				       err);
+			continue;
+		}
+
+		rule = mlx5e_tc_offload_fdb_rules(esw, flow, spec, attr);
+		if (IS_ERR(rule)) {
+			err = PTR_ERR(rule);
+			mlx5_core_warn(priv->mdev, "Failed to update cached decap flow, %d\n",
+				       err);
+		} else {
+			flow->rule[0] = rule;
+			flow_flag_set(flow, OFFLOADED);
+		}
+	}
+}
+
+static int mlx5e_update_route_decap_flows(struct mlx5e_priv *priv,
+					  struct mlx5e_route_entry *r,
+					  struct list_head *flow_list,
+					  bool replace)
+{
+	struct net_device *tunnel_dev;
+	LIST_HEAD(decap_flows);
+
+	tunnel_dev = __dev_get_by_index(dev_net(priv->netdev), r->tunnel_dev_index);
+	if (!tunnel_dev)
+		return -ENODEV;
+
+	mlx5e_take_all_route_decap_flows(r, &decap_flows);
+	if (mlx5e_route_entry_valid(r))
+		mlx5e_unoffload_flow_list(priv, &decap_flows);
+	if (replace)
+		mlx5e_reoffload_decap(priv, &decap_flows);
+
+	list_splice(&decap_flows, flow_list);
+
+	return 0;
+}
+
+static void mlx5e_tc_fib_event_work(struct work_struct *work)
+{
+	struct mlx5e_tc_fib_event_data *event_data =
+		container_of(work, struct mlx5e_tc_fib_event_data, work);
+	struct net_device *ul_dev = event_data->ul_dev;
+	struct mlx5e_priv *priv = netdev_priv(ul_dev);
+	struct mlx5e_route_entry *r = event_data->r;
+	struct mlx5_eswitch *esw;
+	LIST_HEAD(flow_list);
+	bool replace;
+	int err;
+
+	/* sync with concurrent neigh updates */
+	rtnl_lock();
+	esw = priv->mdev->priv.eswitch;
+	mutex_lock(&esw->offloads.encap_tbl_lock);
+	replace = event_data->event == FIB_EVENT_ENTRY_REPLACE;
+
+	if (!mlx5e_route_entry_valid(r) && !replace)
+		goto out;
+
+	err = mlx5e_update_route_encaps(priv, r, &flow_list, replace);
+	if (err)
+		mlx5_core_warn(priv->mdev, "Failed to update route encaps, %d\n",
+			       err);
+
+	err = mlx5e_update_route_decap_flows(priv, r, &flow_list, replace);
+	if (err)
+		mlx5_core_warn(priv->mdev, "Failed to update route decap flows, %d\n",
+			       err);
+
+	if (replace)
+		r->flags |= MLX5E_ROUTE_ENTRY_VALID;
+out:
+	mutex_unlock(&esw->offloads.encap_tbl_lock);
+	rtnl_unlock();
+
+	mlx5e_put_flow_list(priv, &flow_list);
+	mlx5e_route_put(priv, event_data->r);
+	dev_put(event_data->ul_dev);
+	kfree(event_data);
+}
+
+static struct mlx5e_tc_fib_event_data *
+mlx5e_init_fib_work_ipv4(struct mlx5e_priv *priv,
+			 struct net_device *ul_dev,
+			 struct mlx5e_tc_tun_encap *encap,
+			 unsigned long event,
+			 struct fib_notifier_info *info)
+{
+	struct fib_entry_notifier_info *fen_info;
+	struct mlx5e_tc_fib_event_data *fib_work;
+	struct mlx5e_route_entry *r;
+	struct mlx5e_route_key key;
+	struct net_device *fib_dev;
+
+	fen_info = container_of(info, struct fib_entry_notifier_info, info);
+	if (fen_info->fi->nh)
+		return NULL;
+	fib_dev = fib_info_nh(fen_info->fi, 0)->fib_nh_dev;
+	if (!fib_dev || fib_dev->netdev_ops != &mlx5e_netdev_ops ||
+	    fen_info->dst_len != 32)
+		return NULL;
+
+	fib_work = mlx5e_tc_init_fib_work(event, ul_dev, GFP_ATOMIC);
+	if (!fib_work)
+		return ERR_PTR(-ENOMEM);
+
+	key.endpoint_ip.v4 = htonl(fen_info->dst);
+	key.ip_version = 4;
+
+	/* Can't fail after this point because releasing reference to r
+	 * requires obtaining sleeping mutex which we can't do in atomic
+	 * context.
+	 */
+	r = mlx5e_route_lookup_for_update(encap, &key);
+	if (!r)
+		goto out;
+	fib_work->r = r;
+	dev_hold(ul_dev);
+
+	return fib_work;
+
+out:
+	kfree(fib_work);
+	return NULL;
+}
+
+static struct mlx5e_tc_fib_event_data *
+mlx5e_init_fib_work_ipv6(struct mlx5e_priv *priv,
+			 struct net_device *ul_dev,
+			 struct mlx5e_tc_tun_encap *encap,
+			 unsigned long event,
+			 struct fib_notifier_info *info)
+{
+	struct fib6_entry_notifier_info *fen_info;
+	struct mlx5e_tc_fib_event_data *fib_work;
+	struct mlx5e_route_entry *r;
+	struct mlx5e_route_key key;
+	struct net_device *fib_dev;
+
+	fen_info = container_of(info, struct fib6_entry_notifier_info, info);
+	fib_dev = fib6_info_nh_dev(fen_info->rt);
+	if (fib_dev->netdev_ops != &mlx5e_netdev_ops ||
+	    fen_info->rt->fib6_dst.plen != 128)
+		return NULL;
+
+	fib_work = mlx5e_tc_init_fib_work(event, ul_dev, GFP_ATOMIC);
+	if (!fib_work)
+		return ERR_PTR(-ENOMEM);
+
+	memcpy(&key.endpoint_ip.v6, &fen_info->rt->fib6_dst.addr,
+	       sizeof(fen_info->rt->fib6_dst.addr));
+	key.ip_version = 6;
+
+	/* Can't fail after this point because releasing reference to r
+	 * requires obtaining sleeping mutex which we can't do in atomic
+	 * context.
+	 */
+	r = mlx5e_route_lookup_for_update(encap, &key);
+	if (!r)
+		goto out;
+	fib_work->r = r;
+	dev_hold(ul_dev);
+
+	return fib_work;
+
+out:
+	kfree(fib_work);
+	return NULL;
+}
+
+static int mlx5e_tc_tun_fib_event(struct notifier_block *nb, unsigned long event, void *ptr)
+{
+	struct mlx5e_tc_fib_event_data *fib_work;
+	struct fib_notifier_info *info = ptr;
+	struct mlx5e_tc_tun_encap *encap;
+	struct net_device *ul_dev;
+	struct mlx5e_priv *priv;
+
+	encap = container_of(nb, struct mlx5e_tc_tun_encap, fib_nb);
+	priv = encap->priv;
+	ul_dev = priv->netdev;
+	priv = netdev_priv(ul_dev);
+
+	switch (event) {
+	case FIB_EVENT_ENTRY_REPLACE:
+	case FIB_EVENT_ENTRY_DEL:
+		if (info->family == AF_INET)
+			fib_work = mlx5e_init_fib_work_ipv4(priv, ul_dev, encap, event, info);
+		else if (info->family == AF_INET6)
+			fib_work = mlx5e_init_fib_work_ipv6(priv, ul_dev, encap, event, info);
+		else
+			return NOTIFY_DONE;
+
+		if (!IS_ERR_OR_NULL(fib_work)) {
+			queue_work(priv->wq, &fib_work->work);
+		} else if (IS_ERR(fib_work)) {
+			NL_SET_ERR_MSG_MOD(info->extack, "Failed to init fib work");
+			mlx5_core_warn(priv->mdev, "Failed to init fib work, %ld\n",
+				       PTR_ERR(fib_work));
+		}
+
+		break;
+	default:
+		return NOTIFY_DONE;
+	}
+
+	return NOTIFY_DONE;
+}
+
+struct mlx5e_tc_tun_encap *mlx5e_tc_tun_init(struct mlx5e_priv *priv)
+{
+	struct mlx5e_tc_tun_encap *encap;
+	int err;
+
+	encap = kvzalloc(sizeof(*encap), GFP_KERNEL);
+	if (!encap)
+		return ERR_PTR(-ENOMEM);
+
+	encap->priv = priv;
+	encap->fib_nb.notifier_call = mlx5e_tc_tun_fib_event;
+	spin_lock_init(&encap->route_lock);
+	hash_init(encap->route_tbl);
+	err = register_fib_notifier(dev_net(priv->netdev), &encap->fib_nb,
+				    NULL, NULL);
+	if (err) {
+		kvfree(encap);
+		return ERR_PTR(err);
+	}
+
+	return encap;
+}
+
+void mlx5e_tc_tun_cleanup(struct mlx5e_tc_tun_encap *encap)
+{
+	if (!encap)
+		return;
+
+	unregister_fib_notifier(dev_net(encap->priv->netdev), &encap->fib_nb);
+	flush_workqueue(encap->priv->wq); /* flush fib event works */
+	kvfree(encap);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun_encap.h b/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun_encap.h
new file mode 100644
index 000000000..8ad273dde
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun_encap.h
@@ -0,0 +1,41 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2021 Mellanox Technologies. */
+
+#ifndef __MLX5_EN_TC_TUN_ENCAP_H__
+#define __MLX5_EN_TC_TUN_ENCAP_H__
+
+#include "tc_priv.h"
+
+void mlx5e_detach_encap(struct mlx5e_priv *priv,
+			struct mlx5e_tc_flow *flow,
+			struct mlx5_flow_attr *attr,
+			int out_index);
+
+int mlx5e_attach_encap(struct mlx5e_priv *priv,
+		       struct mlx5e_tc_flow *flow,
+		       struct mlx5_flow_attr *attr,
+		       struct net_device *mirred_dev,
+		       int out_index,
+		       struct netlink_ext_ack *extack,
+		       struct net_device **encap_dev);
+
+int mlx5e_attach_decap(struct mlx5e_priv *priv,
+		       struct mlx5e_tc_flow *flow,
+		       struct netlink_ext_ack *extack);
+void mlx5e_detach_decap(struct mlx5e_priv *priv,
+			struct mlx5e_tc_flow *flow);
+
+int mlx5e_attach_decap_route(struct mlx5e_priv *priv,
+			     struct mlx5e_tc_flow *flow);
+void mlx5e_detach_decap_route(struct mlx5e_priv *priv,
+			      struct mlx5e_tc_flow *flow);
+
+struct ip_tunnel_info *mlx5e_dup_tun_info(const struct ip_tunnel_info *tun_info);
+
+int mlx5e_tc_set_attr_rx_tun(struct mlx5e_tc_flow *flow,
+			     struct mlx5_flow_spec *spec);
+
+struct mlx5e_tc_tun_encap *mlx5e_tc_tun_init(struct mlx5e_priv *priv);
+void mlx5e_tc_tun_cleanup(struct mlx5e_tc_tun_encap *encap);
+
+#endif /* __MLX5_EN_TC_TUN_ENCAP_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun_geneve.c b/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun_geneve.c
new file mode 100644
index 000000000..054d80c4e
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun_geneve.c
@@ -0,0 +1,375 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2018 Mellanox Technologies. */
+
+#include <net/geneve.h>
+#include "lib/geneve.h"
+#include "en/tc_tun.h"
+
+#define MLX5E_GENEVE_VER 0
+
+static bool mlx5e_tc_tun_can_offload_geneve(struct mlx5e_priv *priv)
+{
+	return !!(MLX5_CAP_GEN(priv->mdev, flex_parser_protocols) & MLX5_FLEX_PROTO_GENEVE);
+}
+
+static int mlx5e_tc_tun_calc_hlen_geneve(struct mlx5e_encap_entry *e)
+{
+	return sizeof(struct udphdr) +
+	       sizeof(struct genevehdr) +
+	       e->tun_info->options_len;
+}
+
+static int mlx5e_tc_tun_check_udp_dport_geneve(struct mlx5e_priv *priv,
+					       struct flow_cls_offload *f)
+{
+	struct flow_rule *rule = flow_cls_offload_flow_rule(f);
+	struct netlink_ext_ack *extack = f->common.extack;
+	struct flow_match_ports enc_ports;
+
+	if (!flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_ENC_PORTS))
+		return -EOPNOTSUPP;
+
+	flow_rule_match_enc_ports(rule, &enc_ports);
+
+	/* Currently we support only default GENEVE
+	 * port, so udp dst port must match.
+	 */
+	if (be16_to_cpu(enc_ports.key->dst) != GENEVE_UDP_PORT) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "Matched UDP dst port is not registered as a GENEVE port");
+		netdev_warn(priv->netdev,
+			    "UDP port %d is not registered as a GENEVE port\n",
+			    be16_to_cpu(enc_ports.key->dst));
+		return -EOPNOTSUPP;
+	}
+
+	return 0;
+}
+
+static int mlx5e_tc_tun_parse_udp_ports_geneve(struct mlx5e_priv *priv,
+					       struct mlx5_flow_spec *spec,
+					       struct flow_cls_offload *f,
+					       void *headers_c,
+					       void *headers_v)
+{
+	int err;
+
+	err = mlx5e_tc_tun_parse_udp_ports(priv, spec, f, headers_c, headers_v);
+	if (err)
+		return err;
+
+	return mlx5e_tc_tun_check_udp_dport_geneve(priv, f);
+}
+
+static int mlx5e_tc_tun_init_encap_attr_geneve(struct net_device *tunnel_dev,
+					       struct mlx5e_priv *priv,
+					       struct mlx5e_encap_entry *e,
+					       struct netlink_ext_ack *extack)
+{
+	e->tunnel = &geneve_tunnel;
+
+	/* Reformat type for GENEVE encap is similar to VXLAN:
+	 * in both cases the HW adds in the same place a
+	 * defined encapsulation header that the SW provides.
+	 */
+	e->reformat_type = MLX5_REFORMAT_TYPE_L2_TO_VXLAN;
+	return 0;
+}
+
+static void mlx5e_tunnel_id_to_vni(__be64 tun_id, __u8 *vni)
+{
+#ifdef __BIG_ENDIAN
+	vni[0] = (__force __u8)(tun_id >> 16);
+	vni[1] = (__force __u8)(tun_id >> 8);
+	vni[2] = (__force __u8)tun_id;
+#else
+	vni[0] = (__force __u8)((__force u64)tun_id >> 40);
+	vni[1] = (__force __u8)((__force u64)tun_id >> 48);
+	vni[2] = (__force __u8)((__force u64)tun_id >> 56);
+#endif
+}
+
+static int mlx5e_gen_ip_tunnel_header_geneve(char buf[],
+					     __u8 *ip_proto,
+					     struct mlx5e_encap_entry *e)
+{
+	const struct ip_tunnel_info *tun_info = e->tun_info;
+	struct udphdr *udp = (struct udphdr *)(buf);
+	struct genevehdr *geneveh;
+
+	geneveh = (struct genevehdr *)((char *)udp + sizeof(struct udphdr));
+
+	*ip_proto = IPPROTO_UDP;
+
+	udp->dest = tun_info->key.tp_dst;
+
+	memset(geneveh, 0, sizeof(*geneveh));
+	geneveh->ver = MLX5E_GENEVE_VER;
+	geneveh->opt_len = tun_info->options_len / 4;
+	geneveh->oam = !!(tun_info->key.tun_flags & TUNNEL_OAM);
+	geneveh->critical = !!(tun_info->key.tun_flags & TUNNEL_CRIT_OPT);
+	mlx5e_tunnel_id_to_vni(tun_info->key.tun_id, geneveh->vni);
+	geneveh->proto_type = htons(ETH_P_TEB);
+
+	if (tun_info->key.tun_flags & TUNNEL_GENEVE_OPT) {
+		if (!geneveh->opt_len)
+			return -EOPNOTSUPP;
+		ip_tunnel_info_opts_get(geneveh->options, tun_info);
+	}
+
+	return 0;
+}
+
+static int mlx5e_tc_tun_parse_geneve_vni(struct mlx5e_priv *priv,
+					 struct mlx5_flow_spec *spec,
+					 struct flow_cls_offload *f)
+{
+	struct flow_rule *rule = flow_cls_offload_flow_rule(f);
+	struct netlink_ext_ack *extack = f->common.extack;
+	struct flow_match_enc_keyid enc_keyid;
+	void *misc_c, *misc_v;
+
+	misc_c = MLX5_ADDR_OF(fte_match_param, spec->match_criteria, misc_parameters);
+	misc_v = MLX5_ADDR_OF(fte_match_param, spec->match_value, misc_parameters);
+
+	if (!flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_ENC_KEYID))
+		return 0;
+
+	flow_rule_match_enc_keyid(rule, &enc_keyid);
+
+	if (!enc_keyid.mask->keyid)
+		return 0;
+
+	if (!MLX5_CAP_ESW_FLOWTABLE_FDB(priv->mdev, ft_field_support.outer_geneve_vni)) {
+		NL_SET_ERR_MSG_MOD(extack, "Matching on GENEVE VNI is not supported");
+		netdev_warn(priv->netdev, "Matching on GENEVE VNI is not supported\n");
+		return -EOPNOTSUPP;
+	}
+
+	MLX5_SET(fte_match_set_misc, misc_c, geneve_vni, be32_to_cpu(enc_keyid.mask->keyid));
+	MLX5_SET(fte_match_set_misc, misc_v, geneve_vni, be32_to_cpu(enc_keyid.key->keyid));
+
+	return 0;
+}
+
+static int mlx5e_tc_tun_parse_geneve_options(struct mlx5e_priv *priv,
+					     struct mlx5_flow_spec *spec,
+					     struct flow_cls_offload *f)
+{
+	u8 max_tlv_option_data_len = MLX5_CAP_GEN(priv->mdev, max_geneve_tlv_option_data_len);
+	u8 max_tlv_options = MLX5_CAP_GEN(priv->mdev, max_geneve_tlv_options);
+	struct flow_rule *rule = flow_cls_offload_flow_rule(f);
+	struct netlink_ext_ack *extack = f->common.extack;
+	void *misc_c, *misc_v, *misc_3_c, *misc_3_v;
+	struct geneve_opt *option_key, *option_mask;
+	__be32 opt_data_key = 0, opt_data_mask = 0;
+	struct flow_match_enc_opts enc_opts;
+	int res = 0;
+
+	misc_c = MLX5_ADDR_OF(fte_match_param, spec->match_criteria, misc_parameters);
+	misc_v = MLX5_ADDR_OF(fte_match_param, spec->match_value, misc_parameters);
+	misc_3_c = MLX5_ADDR_OF(fte_match_param, spec->match_criteria, misc_parameters_3);
+	misc_3_v = MLX5_ADDR_OF(fte_match_param, spec->match_value, misc_parameters_3);
+
+	if (!flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_ENC_OPTS))
+		return 0;
+
+	flow_rule_match_enc_opts(rule, &enc_opts);
+
+	if (memchr_inv(&enc_opts.mask->data, 0, sizeof(enc_opts.mask->data)) &&
+	    !MLX5_CAP_ESW_FLOWTABLE_FDB(priv->mdev,
+					ft_field_support.geneve_tlv_option_0_data)) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "Matching on GENEVE options is not supported");
+		netdev_warn(priv->netdev,
+			    "Matching on GENEVE options is not supported\n");
+		return -EOPNOTSUPP;
+	}
+
+	/* make sure that we're talking about GENEVE options */
+
+	if (enc_opts.key->dst_opt_type != TUNNEL_GENEVE_OPT) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "Matching on GENEVE options: option type is not GENEVE");
+		netdev_warn(priv->netdev,
+			    "Matching on GENEVE options: option type is not GENEVE\n");
+		return -EOPNOTSUPP;
+	}
+
+	if (enc_opts.mask->len &&
+	    !MLX5_CAP_ESW_FLOWTABLE_FDB(priv->mdev,
+					ft_field_support.outer_geneve_opt_len)) {
+		NL_SET_ERR_MSG_MOD(extack, "Matching on GENEVE options len is not supported");
+		netdev_warn(priv->netdev,
+			    "Matching on GENEVE options len is not supported\n");
+		return -EOPNOTSUPP;
+	}
+
+	/* max_geneve_tlv_option_data_len comes in multiples of 4 bytes, and it
+	 * doesn't include the TLV option header. 'geneve_opt_len' is a total
+	 * len of all the options, including the headers, also multiples of 4
+	 * bytes. Len that comes from the dissector is in bytes.
+	 */
+
+	if ((enc_opts.key->len / 4) > ((max_tlv_option_data_len + 1) * max_tlv_options)) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "Matching on GENEVE options: unsupported options len");
+		netdev_warn(priv->netdev,
+			    "Matching on GENEVE options: unsupported options len (len=%d)\n",
+			    enc_opts.key->len);
+		return -EOPNOTSUPP;
+	}
+
+	MLX5_SET(fte_match_set_misc, misc_c, geneve_opt_len, enc_opts.mask->len / 4);
+	MLX5_SET(fte_match_set_misc, misc_v, geneve_opt_len, enc_opts.key->len / 4);
+
+	/* we support matching on one option only, so just get it */
+	option_key = (struct geneve_opt *)&enc_opts.key->data[0];
+	option_mask = (struct geneve_opt *)&enc_opts.mask->data[0];
+
+	if (option_mask->opt_class == 0 && option_mask->type == 0 &&
+	    !memchr_inv(option_mask->opt_data, 0, option_mask->length * 4))
+		return 0;
+
+	if (option_key->length > max_tlv_option_data_len) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "Matching on GENEVE options: unsupported option len");
+		netdev_warn(priv->netdev,
+			    "Matching on GENEVE options: unsupported option len (key=%d, mask=%d)\n",
+			    option_key->length, option_mask->length);
+		return -EOPNOTSUPP;
+	}
+
+	/* data can't be all 0 - fail to offload such rule */
+	if (!memchr_inv(option_key->opt_data, 0, option_key->length * 4)) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "Matching on GENEVE options: can't match on 0 data field");
+		netdev_warn(priv->netdev,
+			    "Matching on GENEVE options: can't match on 0 data field\n");
+		return -EOPNOTSUPP;
+	}
+
+	/* add new GENEVE TLV options object */
+	res = mlx5_geneve_tlv_option_add(priv->mdev->geneve, option_key);
+	if (res) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "Matching on GENEVE options: failed creating TLV opt object");
+		netdev_warn(priv->netdev,
+			    "Matching on GENEVE options: failed creating TLV opt object (class:type:len = 0x%x:0x%x:%d)\n",
+			    be16_to_cpu(option_key->opt_class),
+			    option_key->type, option_key->length);
+		return res;
+	}
+
+	/* In general, after creating the object, need to query it
+	 * in order to check which option data to set in misc3.
+	 * But we support only geneve_tlv_option_0_data, so no
+	 * point querying at this stage.
+	 */
+
+	memcpy(&opt_data_key, option_key->opt_data, option_key->length * 4);
+	memcpy(&opt_data_mask, option_mask->opt_data, option_mask->length * 4);
+	MLX5_SET(fte_match_set_misc3, misc_3_v,
+		 geneve_tlv_option_0_data, be32_to_cpu(opt_data_key));
+	MLX5_SET(fte_match_set_misc3, misc_3_c,
+		 geneve_tlv_option_0_data, be32_to_cpu(opt_data_mask));
+	if (MLX5_CAP_ESW_FLOWTABLE_FDB(priv->mdev,
+				       ft_field_support.geneve_tlv_option_0_exist)) {
+		MLX5_SET_TO_ONES(fte_match_set_misc, misc_c, geneve_tlv_option_0_exist);
+		MLX5_SET_TO_ONES(fte_match_set_misc, misc_v, geneve_tlv_option_0_exist);
+	}
+
+	spec->match_criteria_enable |= MLX5_MATCH_MISC_PARAMETERS_3;
+
+	return 0;
+}
+
+static int mlx5e_tc_tun_parse_geneve_params(struct mlx5e_priv *priv,
+					    struct mlx5_flow_spec *spec,
+					    struct flow_cls_offload *f)
+{
+	void *misc_c = MLX5_ADDR_OF(fte_match_param, spec->match_criteria, misc_parameters);
+	void *misc_v = MLX5_ADDR_OF(fte_match_param, spec->match_value,  misc_parameters);
+	struct netlink_ext_ack *extack = f->common.extack;
+
+	/* match on OAM - packets with OAM bit on should NOT be offloaded */
+
+	if (!MLX5_CAP_ESW_FLOWTABLE_FDB(priv->mdev, ft_field_support.outer_geneve_oam)) {
+		NL_SET_ERR_MSG_MOD(extack, "Matching on GENEVE OAM is not supported");
+		netdev_warn(priv->netdev, "Matching on GENEVE OAM is not supported\n");
+		return -EOPNOTSUPP;
+	}
+	MLX5_SET_TO_ONES(fte_match_set_misc, misc_c, geneve_oam);
+	MLX5_SET(fte_match_set_misc, misc_v, geneve_oam, 0);
+
+	/* Match on GENEVE protocol. We support only Transparent Eth Bridge. */
+
+	if (MLX5_CAP_ESW_FLOWTABLE_FDB(priv->mdev,
+				       ft_field_support.outer_geneve_protocol_type)) {
+		MLX5_SET_TO_ONES(fte_match_set_misc, misc_c, geneve_protocol_type);
+		MLX5_SET(fte_match_set_misc, misc_v, geneve_protocol_type, ETH_P_TEB);
+	}
+
+	spec->match_criteria_enable |= MLX5_MATCH_MISC_PARAMETERS;
+
+	return 0;
+}
+
+static int mlx5e_tc_tun_parse_geneve(struct mlx5e_priv *priv,
+				     struct mlx5_flow_spec *spec,
+				     struct flow_cls_offload *f,
+				     void *headers_c,
+				     void *headers_v)
+{
+	int err;
+
+	err = mlx5e_tc_tun_parse_geneve_params(priv, spec, f);
+	if (err)
+		return err;
+
+	err = mlx5e_tc_tun_parse_geneve_vni(priv, spec, f);
+	if (err)
+		return err;
+
+	return mlx5e_tc_tun_parse_geneve_options(priv, spec, f);
+}
+
+static bool mlx5e_tc_tun_encap_info_equal_geneve(struct mlx5e_encap_key *a,
+						 struct mlx5e_encap_key *b)
+{
+	struct ip_tunnel_info *a_info;
+	struct ip_tunnel_info *b_info;
+	bool a_has_opts, b_has_opts;
+
+	if (!mlx5e_tc_tun_encap_info_equal_generic(a, b))
+		return false;
+
+	a_has_opts = !!(a->ip_tun_key->tun_flags & TUNNEL_GENEVE_OPT);
+	b_has_opts = !!(b->ip_tun_key->tun_flags & TUNNEL_GENEVE_OPT);
+
+	/* keys are equal when both don't have any options attached */
+	if (!a_has_opts && !b_has_opts)
+		return true;
+
+	if (a_has_opts != b_has_opts)
+		return false;
+
+	/* geneve options stored in memory next to ip_tunnel_info struct */
+	a_info = container_of(a->ip_tun_key, struct ip_tunnel_info, key);
+	b_info = container_of(b->ip_tun_key, struct ip_tunnel_info, key);
+
+	return a_info->options_len == b_info->options_len &&
+		memcmp(a_info + 1, b_info + 1, a_info->options_len) == 0;
+}
+
+struct mlx5e_tc_tunnel geneve_tunnel = {
+	.tunnel_type          = MLX5E_TC_TUNNEL_TYPE_GENEVE,
+	.match_level          = MLX5_MATCH_L4,
+	.can_offload          = mlx5e_tc_tun_can_offload_geneve,
+	.calc_hlen            = mlx5e_tc_tun_calc_hlen_geneve,
+	.init_encap_attr      = mlx5e_tc_tun_init_encap_attr_geneve,
+	.generate_ip_tun_hdr  = mlx5e_gen_ip_tunnel_header_geneve,
+	.parse_udp_ports      = mlx5e_tc_tun_parse_udp_ports_geneve,
+	.parse_tunnel         = mlx5e_tc_tun_parse_geneve,
+	.encap_info_equal     = mlx5e_tc_tun_encap_info_equal_geneve,
+};
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun_gre.c b/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun_gre.c
new file mode 100644
index 000000000..ada14f057
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun_gre.c
@@ -0,0 +1,98 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2018 Mellanox Technologies. */
+
+#include <net/gre.h>
+#include "en/tc_tun.h"
+
+static bool mlx5e_tc_tun_can_offload_gretap(struct mlx5e_priv *priv)
+{
+	return !!MLX5_CAP_ESW(priv->mdev, nvgre_encap_decap);
+}
+
+static int mlx5e_tc_tun_calc_hlen_gretap(struct mlx5e_encap_entry *e)
+{
+	return gre_calc_hlen(e->tun_info->key.tun_flags);
+}
+
+static int mlx5e_tc_tun_init_encap_attr_gretap(struct net_device *tunnel_dev,
+					       struct mlx5e_priv *priv,
+					       struct mlx5e_encap_entry *e,
+					       struct netlink_ext_ack *extack)
+{
+	e->tunnel = &gre_tunnel;
+	e->reformat_type = MLX5_REFORMAT_TYPE_L2_TO_NVGRE;
+	return 0;
+}
+
+static int mlx5e_gen_ip_tunnel_header_gretap(char buf[],
+					     __u8 *ip_proto,
+					     struct mlx5e_encap_entry *e)
+{
+	const struct ip_tunnel_key *tun_key  = &e->tun_info->key;
+	struct gre_base_hdr *greh = (struct gre_base_hdr *)(buf);
+	__be32 tun_id = tunnel_id_to_key32(tun_key->tun_id);
+	int hdr_len;
+
+	*ip_proto = IPPROTO_GRE;
+
+	/* the HW does not calculate GRE csum or sequences */
+	if (tun_key->tun_flags & (TUNNEL_CSUM | TUNNEL_SEQ))
+		return -EOPNOTSUPP;
+
+	greh->protocol = htons(ETH_P_TEB);
+
+	/* GRE key */
+	hdr_len	= mlx5e_tc_tun_calc_hlen_gretap(e);
+	greh->flags = gre_tnl_flags_to_gre_flags(tun_key->tun_flags);
+	if (tun_key->tun_flags & TUNNEL_KEY) {
+		__be32 *ptr = (__be32 *)(((u8 *)greh) + hdr_len - 4);
+		*ptr = tun_id;
+	}
+
+	return 0;
+}
+
+static int mlx5e_tc_tun_parse_gretap(struct mlx5e_priv *priv,
+				     struct mlx5_flow_spec *spec,
+				     struct flow_cls_offload *f,
+				     void *headers_c,
+				     void *headers_v)
+{
+	void *misc_c = MLX5_ADDR_OF(fte_match_param, spec->match_criteria, misc_parameters);
+	void *misc_v = MLX5_ADDR_OF(fte_match_param, spec->match_value, misc_parameters);
+	struct flow_rule *rule = flow_cls_offload_flow_rule(f);
+
+	MLX5_SET_TO_ONES(fte_match_set_lyr_2_4, headers_c, ip_protocol);
+	MLX5_SET(fte_match_set_lyr_2_4, headers_v, ip_protocol, IPPROTO_GRE);
+
+	/* gre protocol */
+	MLX5_SET_TO_ONES(fte_match_set_misc, misc_c, gre_protocol);
+	MLX5_SET(fte_match_set_misc, misc_v, gre_protocol, ETH_P_TEB);
+
+	/* gre key */
+	if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_ENC_KEYID)) {
+		struct flow_match_enc_keyid enc_keyid;
+
+		flow_rule_match_enc_keyid(rule, &enc_keyid);
+		MLX5_SET(fte_match_set_misc, misc_c,
+			 gre_key.key, be32_to_cpu(enc_keyid.mask->keyid));
+		MLX5_SET(fte_match_set_misc, misc_v,
+			 gre_key.key, be32_to_cpu(enc_keyid.key->keyid));
+	}
+
+	spec->match_criteria_enable |= MLX5_MATCH_MISC_PARAMETERS;
+
+	return 0;
+}
+
+struct mlx5e_tc_tunnel gre_tunnel = {
+	.tunnel_type          = MLX5E_TC_TUNNEL_TYPE_GRETAP,
+	.match_level          = MLX5_MATCH_L3,
+	.can_offload          = mlx5e_tc_tun_can_offload_gretap,
+	.calc_hlen            = mlx5e_tc_tun_calc_hlen_gretap,
+	.init_encap_attr      = mlx5e_tc_tun_init_encap_attr_gretap,
+	.generate_ip_tun_hdr  = mlx5e_gen_ip_tunnel_header_gretap,
+	.parse_udp_ports      = NULL,
+	.parse_tunnel         = mlx5e_tc_tun_parse_gretap,
+	.encap_info_equal     = mlx5e_tc_tun_encap_info_equal_generic,
+};
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun_mplsoudp.c b/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun_mplsoudp.c
new file mode 100644
index 000000000..c5b1617d5
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun_mplsoudp.c
@@ -0,0 +1,128 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2018 Mellanox Technologies. */
+
+#include <net/bareudp.h>
+#include <net/mpls.h>
+#include "en/tc_tun.h"
+
+static bool can_offload(struct mlx5e_priv *priv)
+{
+	return MLX5_CAP_ESW_FLOWTABLE_FDB(priv->mdev, reformat_l3_tunnel_to_l2);
+}
+
+static int calc_hlen(struct mlx5e_encap_entry *e)
+{
+	return sizeof(struct udphdr) + MPLS_HLEN;
+}
+
+static int init_encap_attr(struct net_device *tunnel_dev,
+			   struct mlx5e_priv *priv,
+			   struct mlx5e_encap_entry *e,
+			   struct netlink_ext_ack *extack)
+{
+	e->tunnel = &mplsoudp_tunnel;
+	e->reformat_type = MLX5_REFORMAT_TYPE_L2_TO_L3_TUNNEL;
+	return 0;
+}
+
+static int generate_ip_tun_hdr(char buf[],
+			       __u8 *ip_proto,
+			       struct mlx5e_encap_entry *r)
+{
+	const struct ip_tunnel_key *tun_key = &r->tun_info->key;
+	const struct mlx5e_mpls_info *mpls_info = &r->mpls_info;
+	struct udphdr *udp = (struct udphdr *)(buf);
+	struct mpls_shim_hdr *mpls;
+
+	mpls = (struct mpls_shim_hdr *)(udp + 1);
+	*ip_proto = IPPROTO_UDP;
+
+	udp->dest = tun_key->tp_dst;
+	*mpls = mpls_entry_encode(mpls_info->label, mpls_info->ttl, mpls_info->tc, mpls_info->bos);
+
+	return 0;
+}
+
+static int parse_udp_ports(struct mlx5e_priv *priv,
+			   struct mlx5_flow_spec *spec,
+			   struct flow_cls_offload *f,
+			   void *headers_c,
+			   void *headers_v)
+{
+	return mlx5e_tc_tun_parse_udp_ports(priv, spec, f, headers_c, headers_v);
+}
+
+static int parse_tunnel(struct mlx5e_priv *priv,
+			struct mlx5_flow_spec *spec,
+			struct flow_cls_offload *f,
+			void *headers_c,
+			void *headers_v)
+{
+	struct flow_rule *rule = flow_cls_offload_flow_rule(f);
+	struct flow_match_mpls match;
+	void *misc2_c;
+	void *misc2_v;
+
+	if (!MLX5_CAP_ETH(priv->mdev, tunnel_stateless_mpls_over_udp) &&
+	    !(MLX5_CAP_GEN(priv->mdev, flex_parser_protocols) & MLX5_FLEX_PROTO_CW_MPLS_UDP))
+		return -EOPNOTSUPP;
+
+	if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_ENC_KEYID))
+		return -EOPNOTSUPP;
+
+	if (!flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_MPLS))
+		return 0;
+
+	flow_rule_match_mpls(rule, &match);
+
+	/* Only support matching the first LSE */
+	if (match.mask->used_lses != 1)
+		return -EOPNOTSUPP;
+
+	misc2_c = MLX5_ADDR_OF(fte_match_param, spec->match_criteria,
+			       misc_parameters_2);
+	misc2_v = MLX5_ADDR_OF(fte_match_param, spec->match_value,
+			       misc_parameters_2);
+
+	MLX5_SET(fte_match_set_misc2, misc2_c,
+		 outer_first_mpls_over_udp.mpls_label,
+		 match.mask->ls[0].mpls_label);
+	MLX5_SET(fte_match_set_misc2, misc2_v,
+		 outer_first_mpls_over_udp.mpls_label,
+		 match.key->ls[0].mpls_label);
+
+	MLX5_SET(fte_match_set_misc2, misc2_c,
+		 outer_first_mpls_over_udp.mpls_exp,
+		 match.mask->ls[0].mpls_tc);
+	MLX5_SET(fte_match_set_misc2, misc2_v,
+		 outer_first_mpls_over_udp.mpls_exp, match.key->ls[0].mpls_tc);
+
+	MLX5_SET(fte_match_set_misc2, misc2_c,
+		 outer_first_mpls_over_udp.mpls_s_bos,
+		 match.mask->ls[0].mpls_bos);
+	MLX5_SET(fte_match_set_misc2, misc2_v,
+		 outer_first_mpls_over_udp.mpls_s_bos,
+		 match.key->ls[0].mpls_bos);
+
+	MLX5_SET(fte_match_set_misc2, misc2_c,
+		 outer_first_mpls_over_udp.mpls_ttl,
+		 match.mask->ls[0].mpls_ttl);
+	MLX5_SET(fte_match_set_misc2, misc2_v,
+		 outer_first_mpls_over_udp.mpls_ttl,
+		 match.key->ls[0].mpls_ttl);
+	spec->match_criteria_enable |= MLX5_MATCH_MISC_PARAMETERS_2;
+
+	return 0;
+}
+
+struct mlx5e_tc_tunnel mplsoudp_tunnel = {
+	.tunnel_type          = MLX5E_TC_TUNNEL_TYPE_MPLSOUDP,
+	.match_level          = MLX5_MATCH_L4,
+	.can_offload          = can_offload,
+	.calc_hlen            = calc_hlen,
+	.init_encap_attr      = init_encap_attr,
+	.generate_ip_tun_hdr  = generate_ip_tun_hdr,
+	.parse_udp_ports      = parse_udp_ports,
+	.parse_tunnel         = parse_tunnel,
+	.encap_info_equal     = mlx5e_tc_tun_encap_info_equal_generic,
+};
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun_vxlan.c b/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun_vxlan.c
new file mode 100644
index 000000000..1f62c702b
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun_vxlan.c
@@ -0,0 +1,165 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2018 Mellanox Technologies. */
+
+#include <net/vxlan.h>
+#include "lib/vxlan.h"
+#include "en/tc_tun.h"
+
+static bool mlx5e_tc_tun_can_offload_vxlan(struct mlx5e_priv *priv)
+{
+	return !!MLX5_CAP_ESW(priv->mdev, vxlan_encap_decap);
+}
+
+static int mlx5e_tc_tun_calc_hlen_vxlan(struct mlx5e_encap_entry *e)
+{
+	return VXLAN_HLEN;
+}
+
+static int mlx5e_tc_tun_check_udp_dport_vxlan(struct mlx5e_priv *priv,
+					      struct flow_cls_offload *f)
+{
+	struct flow_rule *rule = flow_cls_offload_flow_rule(f);
+	struct netlink_ext_ack *extack = f->common.extack;
+	struct flow_match_ports enc_ports;
+
+	if (!flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_ENC_PORTS))
+		return -EOPNOTSUPP;
+
+	flow_rule_match_enc_ports(rule, &enc_ports);
+
+	/* check the UDP destination port validity */
+
+	if (!mlx5_vxlan_lookup_port(priv->mdev->vxlan,
+				    be16_to_cpu(enc_ports.key->dst))) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "Matched UDP dst port is not registered as a VXLAN port");
+		netdev_warn(priv->netdev,
+			    "UDP port %d is not registered as a VXLAN port\n",
+			    be16_to_cpu(enc_ports.key->dst));
+		return -EOPNOTSUPP;
+	}
+
+	return 0;
+}
+
+static int mlx5e_tc_tun_parse_udp_ports_vxlan(struct mlx5e_priv *priv,
+					      struct mlx5_flow_spec *spec,
+					      struct flow_cls_offload *f,
+					      void *headers_c,
+					      void *headers_v)
+{
+	int err = 0;
+
+	err = mlx5e_tc_tun_parse_udp_ports(priv, spec, f, headers_c, headers_v);
+	if (err)
+		return err;
+
+	return mlx5e_tc_tun_check_udp_dport_vxlan(priv, f);
+}
+
+static int mlx5e_tc_tun_init_encap_attr_vxlan(struct net_device *tunnel_dev,
+					      struct mlx5e_priv *priv,
+					      struct mlx5e_encap_entry *e,
+					      struct netlink_ext_ack *extack)
+{
+	int dst_port = be16_to_cpu(e->tun_info->key.tp_dst);
+
+	e->tunnel = &vxlan_tunnel;
+
+	if (!mlx5_vxlan_lookup_port(priv->mdev->vxlan, dst_port)) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "vxlan udp dport was not registered with the HW");
+		netdev_warn(priv->netdev,
+			    "%d isn't an offloaded vxlan udp dport\n",
+			    dst_port);
+		return -EOPNOTSUPP;
+	}
+
+	e->reformat_type = MLX5_REFORMAT_TYPE_L2_TO_VXLAN;
+	return 0;
+}
+
+static int mlx5e_gen_ip_tunnel_header_vxlan(char buf[],
+					    __u8 *ip_proto,
+					    struct mlx5e_encap_entry *e)
+{
+	const struct ip_tunnel_key *tun_key = &e->tun_info->key;
+	__be32 tun_id = tunnel_id_to_key32(tun_key->tun_id);
+	struct udphdr *udp = (struct udphdr *)(buf);
+	struct vxlanhdr *vxh;
+
+	if (tun_key->tun_flags & TUNNEL_VXLAN_OPT)
+		return -EOPNOTSUPP;
+	vxh = (struct vxlanhdr *)((char *)udp + sizeof(struct udphdr));
+	*ip_proto = IPPROTO_UDP;
+
+	udp->dest = tun_key->tp_dst;
+	vxh->vx_flags = VXLAN_HF_VNI;
+	vxh->vx_vni = vxlan_vni_field(tun_id);
+
+	return 0;
+}
+
+static int mlx5e_tc_tun_parse_vxlan(struct mlx5e_priv *priv,
+				    struct mlx5_flow_spec *spec,
+				    struct flow_cls_offload *f,
+				    void *headers_c,
+				    void *headers_v)
+{
+	struct flow_rule *rule = flow_cls_offload_flow_rule(f);
+	struct netlink_ext_ack *extack = f->common.extack;
+	struct flow_match_enc_keyid enc_keyid;
+	void *misc_c, *misc_v;
+
+	misc_c = MLX5_ADDR_OF(fte_match_param, spec->match_criteria, misc_parameters);
+	misc_v = MLX5_ADDR_OF(fte_match_param, spec->match_value, misc_parameters);
+
+	if (!flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_ENC_KEYID))
+		return 0;
+
+	flow_rule_match_enc_keyid(rule, &enc_keyid);
+
+	if (!enc_keyid.mask->keyid)
+		return 0;
+
+	/* match on VNI is required */
+
+	if (!MLX5_CAP_ESW_FLOWTABLE_FDB(priv->mdev,
+					ft_field_support.outer_vxlan_vni)) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "Matching on VXLAN VNI is not supported");
+		netdev_warn(priv->netdev,
+			    "Matching on VXLAN VNI is not supported\n");
+		return -EOPNOTSUPP;
+	}
+
+	MLX5_SET(fte_match_set_misc, misc_c, vxlan_vni,
+		 be32_to_cpu(enc_keyid.mask->keyid));
+	MLX5_SET(fte_match_set_misc, misc_v, vxlan_vni,
+		 be32_to_cpu(enc_keyid.key->keyid));
+
+	spec->match_criteria_enable |= MLX5_MATCH_MISC_PARAMETERS;
+
+	return 0;
+}
+
+static int mlx5e_tc_tun_get_remote_ifindex(struct net_device *mirred_dev)
+{
+	const struct vxlan_dev *vxlan = netdev_priv(mirred_dev);
+	const struct vxlan_rdst *dst = &vxlan->default_dst;
+
+	return dst->remote_ifindex;
+}
+
+struct mlx5e_tc_tunnel vxlan_tunnel = {
+	.tunnel_type          = MLX5E_TC_TUNNEL_TYPE_VXLAN,
+	.match_level          = MLX5_MATCH_L4,
+	.can_offload          = mlx5e_tc_tun_can_offload_vxlan,
+	.calc_hlen            = mlx5e_tc_tun_calc_hlen_vxlan,
+	.init_encap_attr      = mlx5e_tc_tun_init_encap_attr_vxlan,
+	.generate_ip_tun_hdr  = mlx5e_gen_ip_tunnel_header_vxlan,
+	.parse_udp_ports      = mlx5e_tc_tun_parse_udp_ports_vxlan,
+	.parse_tunnel         = mlx5e_tc_tun_parse_vxlan,
+	.encap_info_equal     = mlx5e_tc_tun_encap_info_equal_generic,
+	.get_remote_ifindex   = mlx5e_tc_tun_get_remote_ifindex,
+};
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tir.c b/drivers/net/ethernet/mellanox/mlx5/core/en/tir.c
new file mode 100644
index 000000000..d4239e3b3
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tir.c
@@ -0,0 +1,203 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2021, Mellanox Technologies inc. All rights reserved. */
+
+#include "tir.h"
+#include "params.h"
+#include <linux/mlx5/transobj.h>
+
+#define MLX5E_PARAMS_DEFAULT_LRO_WQE_SZ (64 * 1024)
+
+/* max() doesn't work inside square brackets. */
+#define MLX5E_TIR_CMD_IN_SZ_DW ( \
+	MLX5_ST_SZ_DW(create_tir_in) > MLX5_ST_SZ_DW(modify_tir_in) ? \
+	MLX5_ST_SZ_DW(create_tir_in) : MLX5_ST_SZ_DW(modify_tir_in) \
+)
+
+struct mlx5e_tir_builder {
+	u32 in[MLX5E_TIR_CMD_IN_SZ_DW];
+	bool modify;
+};
+
+struct mlx5e_tir_builder *mlx5e_tir_builder_alloc(bool modify)
+{
+	struct mlx5e_tir_builder *builder;
+
+	builder = kvzalloc(sizeof(*builder), GFP_KERNEL);
+	builder->modify = modify;
+
+	return builder;
+}
+
+void mlx5e_tir_builder_free(struct mlx5e_tir_builder *builder)
+{
+	kvfree(builder);
+}
+
+void mlx5e_tir_builder_clear(struct mlx5e_tir_builder *builder)
+{
+	memset(builder->in, 0, sizeof(builder->in));
+}
+
+static void *mlx5e_tir_builder_get_tirc(struct mlx5e_tir_builder *builder)
+{
+	if (builder->modify)
+		return MLX5_ADDR_OF(modify_tir_in, builder->in, ctx);
+	return MLX5_ADDR_OF(create_tir_in, builder->in, ctx);
+}
+
+void mlx5e_tir_builder_build_inline(struct mlx5e_tir_builder *builder, u32 tdn, u32 rqn)
+{
+	void *tirc = mlx5e_tir_builder_get_tirc(builder);
+
+	WARN_ON(builder->modify);
+
+	MLX5_SET(tirc, tirc, transport_domain, tdn);
+	MLX5_SET(tirc, tirc, disp_type, MLX5_TIRC_DISP_TYPE_DIRECT);
+	MLX5_SET(tirc, tirc, rx_hash_fn, MLX5_RX_HASH_FN_NONE);
+	MLX5_SET(tirc, tirc, inline_rqn, rqn);
+}
+
+void mlx5e_tir_builder_build_rqt(struct mlx5e_tir_builder *builder, u32 tdn,
+				 u32 rqtn, bool inner_ft_support)
+{
+	void *tirc = mlx5e_tir_builder_get_tirc(builder);
+
+	WARN_ON(builder->modify);
+
+	MLX5_SET(tirc, tirc, transport_domain, tdn);
+	MLX5_SET(tirc, tirc, disp_type, MLX5_TIRC_DISP_TYPE_INDIRECT);
+	MLX5_SET(tirc, tirc, indirect_table, rqtn);
+	MLX5_SET(tirc, tirc, tunneled_offload_en, inner_ft_support);
+}
+
+void mlx5e_tir_builder_build_packet_merge(struct mlx5e_tir_builder *builder,
+					  const struct mlx5e_packet_merge_param *pkt_merge_param)
+{
+	void *tirc = mlx5e_tir_builder_get_tirc(builder);
+	const unsigned int rough_max_l2_l3_hdr_sz = 256;
+
+	if (builder->modify)
+		MLX5_SET(modify_tir_in, builder->in, bitmask.packet_merge, 1);
+
+	switch (pkt_merge_param->type) {
+	case MLX5E_PACKET_MERGE_LRO:
+		MLX5_SET(tirc, tirc, packet_merge_mask,
+			 MLX5_TIRC_PACKET_MERGE_MASK_IPV4_LRO |
+			 MLX5_TIRC_PACKET_MERGE_MASK_IPV6_LRO);
+		MLX5_SET(tirc, tirc, lro_max_ip_payload_size,
+			 (MLX5E_PARAMS_DEFAULT_LRO_WQE_SZ - rough_max_l2_l3_hdr_sz) >> 8);
+		MLX5_SET(tirc, tirc, lro_timeout_period_usecs, pkt_merge_param->timeout);
+		break;
+	default:
+		break;
+	}
+}
+
+static int mlx5e_hfunc_to_hw(u8 hfunc)
+{
+	switch (hfunc) {
+	case ETH_RSS_HASH_TOP:
+		return MLX5_RX_HASH_FN_TOEPLITZ;
+	case ETH_RSS_HASH_XOR:
+		return MLX5_RX_HASH_FN_INVERTED_XOR8;
+	default:
+		return MLX5_RX_HASH_FN_NONE;
+	}
+}
+
+void mlx5e_tir_builder_build_rss(struct mlx5e_tir_builder *builder,
+				 const struct mlx5e_rss_params_hash *rss_hash,
+				 const struct mlx5e_rss_params_traffic_type *rss_tt,
+				 bool inner)
+{
+	void *tirc = mlx5e_tir_builder_get_tirc(builder);
+	void *hfso;
+
+	if (builder->modify)
+		MLX5_SET(modify_tir_in, builder->in, bitmask.hash, 1);
+
+	MLX5_SET(tirc, tirc, rx_hash_fn, mlx5e_hfunc_to_hw(rss_hash->hfunc));
+	if (rss_hash->hfunc == ETH_RSS_HASH_TOP) {
+		const size_t len = MLX5_FLD_SZ_BYTES(tirc, rx_hash_toeplitz_key);
+		void *rss_key = MLX5_ADDR_OF(tirc, tirc, rx_hash_toeplitz_key);
+
+		MLX5_SET(tirc, tirc, rx_hash_symmetric, 1);
+		memcpy(rss_key, rss_hash->toeplitz_hash_key, len);
+	}
+
+	if (inner)
+		hfso = MLX5_ADDR_OF(tirc, tirc, rx_hash_field_selector_inner);
+	else
+		hfso = MLX5_ADDR_OF(tirc, tirc, rx_hash_field_selector_outer);
+	MLX5_SET(rx_hash_field_select, hfso, l3_prot_type, rss_tt->l3_prot_type);
+	MLX5_SET(rx_hash_field_select, hfso, l4_prot_type, rss_tt->l4_prot_type);
+	MLX5_SET(rx_hash_field_select, hfso, selected_fields, rss_tt->rx_hash_fields);
+}
+
+void mlx5e_tir_builder_build_direct(struct mlx5e_tir_builder *builder)
+{
+	void *tirc = mlx5e_tir_builder_get_tirc(builder);
+
+	WARN_ON(builder->modify);
+
+	MLX5_SET(tirc, tirc, rx_hash_fn, MLX5_RX_HASH_FN_INVERTED_XOR8);
+}
+
+void mlx5e_tir_builder_build_tls(struct mlx5e_tir_builder *builder)
+{
+	void *tirc = mlx5e_tir_builder_get_tirc(builder);
+
+	WARN_ON(builder->modify);
+
+	MLX5_SET(tirc, tirc, tls_en, 1);
+	MLX5_SET(tirc, tirc, self_lb_block,
+		 MLX5_TIRC_SELF_LB_BLOCK_BLOCK_UNICAST |
+		 MLX5_TIRC_SELF_LB_BLOCK_BLOCK_MULTICAST);
+}
+
+int mlx5e_tir_init(struct mlx5e_tir *tir, struct mlx5e_tir_builder *builder,
+		   struct mlx5_core_dev *mdev, bool reg)
+{
+	int err;
+
+	tir->mdev = mdev;
+
+	err = mlx5_core_create_tir(tir->mdev, builder->in, &tir->tirn);
+	if (err)
+		return err;
+
+	if (reg) {
+		struct mlx5e_hw_objs *res = &tir->mdev->mlx5e_res.hw_objs;
+
+		mutex_lock(&res->td.list_lock);
+		list_add(&tir->list, &res->td.tirs_list);
+		mutex_unlock(&res->td.list_lock);
+	} else {
+		INIT_LIST_HEAD(&tir->list);
+	}
+
+	return 0;
+}
+
+void mlx5e_tir_destroy(struct mlx5e_tir *tir)
+{
+	struct mlx5e_hw_objs *res = &tir->mdev->mlx5e_res.hw_objs;
+
+	/* Skip mutex if list_del is no-op (the TIR wasn't registered in the
+	 * list). list_empty will never return true for an item of tirs_list,
+	 * and READ_ONCE/WRITE_ONCE in list_empty/list_del guarantee consistency
+	 * of the list->next value.
+	 */
+	if (!list_empty(&tir->list)) {
+		mutex_lock(&res->td.list_lock);
+		list_del(&tir->list);
+		mutex_unlock(&res->td.list_lock);
+	}
+
+	mlx5_core_destroy_tir(tir->mdev, tir->tirn);
+}
+
+int mlx5e_tir_modify(struct mlx5e_tir *tir, struct mlx5e_tir_builder *builder)
+{
+	return mlx5_core_modify_tir(tir->mdev, tir->tirn, builder->in);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tir.h b/drivers/net/ethernet/mellanox/mlx5/core/en/tir.h
new file mode 100644
index 000000000..857a84bcd
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tir.h
@@ -0,0 +1,58 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2021, Mellanox Technologies inc. All rights reserved. */
+
+#ifndef __MLX5_EN_TIR_H__
+#define __MLX5_EN_TIR_H__
+
+#include <linux/kernel.h>
+
+struct mlx5e_rss_params_hash {
+	u8 hfunc;
+	u8 toeplitz_hash_key[40];
+};
+
+struct mlx5e_rss_params_traffic_type {
+	u8 l3_prot_type;
+	u8 l4_prot_type;
+	u32 rx_hash_fields;
+};
+
+struct mlx5e_tir_builder;
+struct mlx5e_packet_merge_param;
+
+struct mlx5e_tir_builder *mlx5e_tir_builder_alloc(bool modify);
+void mlx5e_tir_builder_free(struct mlx5e_tir_builder *builder);
+void mlx5e_tir_builder_clear(struct mlx5e_tir_builder *builder);
+
+void mlx5e_tir_builder_build_inline(struct mlx5e_tir_builder *builder, u32 tdn, u32 rqn);
+void mlx5e_tir_builder_build_rqt(struct mlx5e_tir_builder *builder, u32 tdn,
+				 u32 rqtn, bool inner_ft_support);
+void mlx5e_tir_builder_build_packet_merge(struct mlx5e_tir_builder *builder,
+					  const struct mlx5e_packet_merge_param *pkt_merge_param);
+void mlx5e_tir_builder_build_rss(struct mlx5e_tir_builder *builder,
+				 const struct mlx5e_rss_params_hash *rss_hash,
+				 const struct mlx5e_rss_params_traffic_type *rss_tt,
+				 bool inner);
+void mlx5e_tir_builder_build_direct(struct mlx5e_tir_builder *builder);
+void mlx5e_tir_builder_build_tls(struct mlx5e_tir_builder *builder);
+
+struct mlx5_core_dev;
+
+struct mlx5e_tir {
+	struct mlx5_core_dev *mdev;
+	u32 tirn;
+	struct list_head list;
+};
+
+int mlx5e_tir_init(struct mlx5e_tir *tir, struct mlx5e_tir_builder *builder,
+		   struct mlx5_core_dev *mdev, bool reg);
+void mlx5e_tir_destroy(struct mlx5e_tir *tir);
+
+static inline u32 mlx5e_tir_get_tirn(struct mlx5e_tir *tir)
+{
+	return tir->tirn;
+}
+
+int mlx5e_tir_modify(struct mlx5e_tir *tir, struct mlx5e_tir_builder *builder);
+
+#endif /* __MLX5_EN_TIR_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/trap.c b/drivers/net/ethernet/mellanox/mlx5/core/en/trap.c
new file mode 100644
index 000000000..201ac7dd3
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/trap.c
@@ -0,0 +1,331 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2020 Mellanox Technologies */
+
+#include <net/page_pool.h>
+#include "en/txrx.h"
+#include "en/params.h"
+#include "en/trap.h"
+
+static int mlx5e_trap_napi_poll(struct napi_struct *napi, int budget)
+{
+	struct mlx5e_trap *trap_ctx = container_of(napi, struct mlx5e_trap, napi);
+	struct mlx5e_ch_stats *ch_stats = trap_ctx->stats;
+	struct mlx5e_rq *rq = &trap_ctx->rq;
+	bool busy = false;
+	int work_done = 0;
+
+	rcu_read_lock();
+
+	ch_stats->poll++;
+
+	work_done = mlx5e_poll_rx_cq(&rq->cq, budget);
+	busy |= work_done == budget;
+	busy |= rq->post_wqes(rq);
+
+	if (busy) {
+		work_done = budget;
+		goto out;
+	}
+
+	if (unlikely(!napi_complete_done(napi, work_done)))
+		goto out;
+
+	mlx5e_cq_arm(&rq->cq);
+
+out:
+	rcu_read_unlock();
+	return work_done;
+}
+
+static void mlx5e_init_trap_rq(struct mlx5e_trap *t, struct mlx5e_params *params,
+			       struct mlx5e_rq *rq)
+{
+	struct mlx5_core_dev *mdev = t->mdev;
+	struct mlx5e_priv *priv = t->priv;
+
+	rq->wq_type      = params->rq_wq_type;
+	rq->pdev         = t->pdev;
+	rq->netdev       = priv->netdev;
+	rq->priv         = priv;
+	rq->clock        = &mdev->clock;
+	rq->tstamp       = &priv->tstamp;
+	rq->mdev         = mdev;
+	rq->hw_mtu       = MLX5E_SW2HW_MTU(params, params->sw_mtu);
+	rq->stats        = &priv->trap_stats.rq;
+	rq->ptp_cyc2time = mlx5_rq_ts_translator(mdev);
+	xdp_rxq_info_unused(&rq->xdp_rxq);
+	mlx5e_rq_set_trap_handlers(rq, params);
+}
+
+static int mlx5e_open_trap_rq(struct mlx5e_priv *priv, struct mlx5e_trap *t)
+{
+	struct mlx5e_rq_param *rq_param = &t->rq_param;
+	struct mlx5_core_dev *mdev = priv->mdev;
+	struct mlx5e_create_cq_param ccp = {};
+	struct dim_cq_moder trap_moder = {};
+	struct mlx5e_rq *rq = &t->rq;
+	int node;
+	int err;
+
+	node = dev_to_node(mdev->device);
+
+	ccp.node     = node;
+	ccp.ch_stats = t->stats;
+	ccp.napi     = &t->napi;
+	ccp.ix       = 0;
+	err = mlx5e_open_cq(priv, trap_moder, &rq_param->cqp, &ccp, &rq->cq);
+	if (err)
+		return err;
+
+	mlx5e_init_trap_rq(t, &t->params, rq);
+	err = mlx5e_open_rq(&t->params, rq_param, NULL, node, rq);
+	if (err)
+		goto err_destroy_cq;
+
+	return 0;
+
+err_destroy_cq:
+	mlx5e_close_cq(&rq->cq);
+
+	return err;
+}
+
+static void mlx5e_close_trap_rq(struct mlx5e_rq *rq)
+{
+	mlx5e_close_rq(rq);
+	mlx5e_close_cq(&rq->cq);
+}
+
+static int mlx5e_create_trap_direct_rq_tir(struct mlx5_core_dev *mdev, struct mlx5e_tir *tir,
+					   u32 rqn)
+{
+	struct mlx5e_tir_builder *builder;
+	int err;
+
+	builder = mlx5e_tir_builder_alloc(false);
+	if (!builder)
+		return -ENOMEM;
+
+	mlx5e_tir_builder_build_inline(builder, mdev->mlx5e_res.hw_objs.td.tdn, rqn);
+	err = mlx5e_tir_init(tir, builder, mdev, true);
+
+	mlx5e_tir_builder_free(builder);
+
+	return err;
+}
+
+static void mlx5e_build_trap_params(struct mlx5_core_dev *mdev,
+				    int max_mtu, u16 q_counter,
+				    struct mlx5e_trap *t)
+{
+	struct mlx5e_params *params = &t->params;
+
+	params->rq_wq_type = MLX5_WQ_TYPE_CYCLIC;
+	mlx5e_init_rq_type_params(mdev, params);
+	params->sw_mtu = max_mtu;
+	mlx5e_build_rq_param(mdev, params, NULL, q_counter, &t->rq_param);
+}
+
+static struct mlx5e_trap *mlx5e_open_trap(struct mlx5e_priv *priv)
+{
+	int cpu = cpumask_first(mlx5_comp_irq_get_affinity_mask(priv->mdev, 0));
+	struct net_device *netdev = priv->netdev;
+	struct mlx5e_trap *t;
+	int err;
+
+	t = kvzalloc_node(sizeof(*t), GFP_KERNEL, cpu_to_node(cpu));
+	if (!t)
+		return ERR_PTR(-ENOMEM);
+
+	mlx5e_build_trap_params(priv->mdev, netdev->max_mtu, priv->q_counter, t);
+
+	t->priv     = priv;
+	t->mdev     = priv->mdev;
+	t->tstamp   = &priv->tstamp;
+	t->pdev     = mlx5_core_dma_dev(priv->mdev);
+	t->netdev   = priv->netdev;
+	t->mkey_be  = cpu_to_be32(priv->mdev->mlx5e_res.hw_objs.mkey);
+	t->stats    = &priv->trap_stats.ch;
+
+	netif_napi_add(netdev, &t->napi, mlx5e_trap_napi_poll);
+
+	err = mlx5e_open_trap_rq(priv, t);
+	if (unlikely(err))
+		goto err_napi_del;
+
+	err = mlx5e_create_trap_direct_rq_tir(t->mdev, &t->tir, t->rq.rqn);
+	if (err)
+		goto err_close_trap_rq;
+
+	return t;
+
+err_close_trap_rq:
+	mlx5e_close_trap_rq(&t->rq);
+err_napi_del:
+	netif_napi_del(&t->napi);
+	kvfree(t);
+	return ERR_PTR(err);
+}
+
+void mlx5e_close_trap(struct mlx5e_trap *trap)
+{
+	mlx5e_tir_destroy(&trap->tir);
+	mlx5e_close_trap_rq(&trap->rq);
+	netif_napi_del(&trap->napi);
+	kvfree(trap);
+}
+
+static void mlx5e_activate_trap(struct mlx5e_trap *trap)
+{
+	napi_enable(&trap->napi);
+	mlx5e_activate_rq(&trap->rq);
+	mlx5e_trigger_napi_sched(&trap->napi);
+}
+
+void mlx5e_deactivate_trap(struct mlx5e_priv *priv)
+{
+	struct mlx5e_trap *trap = priv->en_trap;
+
+	mlx5e_deactivate_rq(&trap->rq);
+	napi_disable(&trap->napi);
+}
+
+static struct mlx5e_trap *mlx5e_add_trap_queue(struct mlx5e_priv *priv)
+{
+	struct mlx5e_trap *trap;
+
+	trap = mlx5e_open_trap(priv);
+	if (IS_ERR(trap))
+		goto out;
+
+	mlx5e_activate_trap(trap);
+out:
+	return trap;
+}
+
+static void mlx5e_del_trap_queue(struct mlx5e_priv *priv)
+{
+	mlx5e_deactivate_trap(priv);
+	mlx5e_close_trap(priv->en_trap);
+	priv->en_trap = NULL;
+}
+
+static int mlx5e_trap_get_tirn(struct mlx5e_trap *en_trap)
+{
+	return en_trap->tir.tirn;
+}
+
+static int mlx5e_handle_action_trap(struct mlx5e_priv *priv, int trap_id)
+{
+	bool open_queue = !priv->en_trap;
+	struct mlx5e_trap *trap;
+	int err;
+
+	if (open_queue) {
+		trap = mlx5e_add_trap_queue(priv);
+		if (IS_ERR(trap))
+			return PTR_ERR(trap);
+		priv->en_trap = trap;
+	}
+
+	switch (trap_id) {
+	case DEVLINK_TRAP_GENERIC_ID_INGRESS_VLAN_FILTER:
+		err = mlx5e_add_vlan_trap(priv->fs, trap_id, mlx5e_trap_get_tirn(priv->en_trap));
+		if (err)
+			goto err_out;
+		break;
+	case DEVLINK_TRAP_GENERIC_ID_DMAC_FILTER:
+		err = mlx5e_add_mac_trap(priv->fs, trap_id, mlx5e_trap_get_tirn(priv->en_trap));
+		if (err)
+			goto err_out;
+		break;
+	default:
+		netdev_warn(priv->netdev, "%s: Unknown trap id %d\n", __func__, trap_id);
+		err = -EINVAL;
+		goto err_out;
+	}
+	return 0;
+
+err_out:
+	if (open_queue)
+		mlx5e_del_trap_queue(priv);
+	return err;
+}
+
+static int mlx5e_handle_action_drop(struct mlx5e_priv *priv, int trap_id)
+{
+	switch (trap_id) {
+	case DEVLINK_TRAP_GENERIC_ID_INGRESS_VLAN_FILTER:
+		mlx5e_remove_vlan_trap(priv->fs);
+		break;
+	case DEVLINK_TRAP_GENERIC_ID_DMAC_FILTER:
+		mlx5e_remove_mac_trap(priv->fs);
+		break;
+	default:
+		netdev_warn(priv->netdev, "%s: Unknown trap id %d\n", __func__, trap_id);
+		return -EINVAL;
+	}
+	if (priv->en_trap && !mlx5_devlink_trap_get_num_active(priv->mdev))
+		mlx5e_del_trap_queue(priv);
+
+	return 0;
+}
+
+int mlx5e_handle_trap_event(struct mlx5e_priv *priv, struct mlx5_trap_ctx *trap_ctx)
+{
+	int err = 0;
+
+	/* Traps are unarmed when interface is down, no need to update
+	 * them. The configuration is saved in the core driver,
+	 * queried and applied upon interface up operation in
+	 * mlx5e_open_locked().
+	 */
+	if (!test_bit(MLX5E_STATE_OPENED, &priv->state))
+		return 0;
+
+	switch (trap_ctx->action) {
+	case DEVLINK_TRAP_ACTION_TRAP:
+		err = mlx5e_handle_action_trap(priv, trap_ctx->id);
+		break;
+	case DEVLINK_TRAP_ACTION_DROP:
+		err = mlx5e_handle_action_drop(priv, trap_ctx->id);
+		break;
+	default:
+		netdev_warn(priv->netdev, "%s: Unsupported action %d\n", __func__,
+			    trap_ctx->action);
+		err = -EINVAL;
+	}
+	return err;
+}
+
+static int mlx5e_apply_trap(struct mlx5e_priv *priv, int trap_id, bool enable)
+{
+	enum devlink_trap_action action;
+	int err;
+
+	err = mlx5_devlink_traps_get_action(priv->mdev, trap_id, &action);
+	if (err)
+		return err;
+	if (action == DEVLINK_TRAP_ACTION_TRAP)
+		err = enable ? mlx5e_handle_action_trap(priv, trap_id) :
+			       mlx5e_handle_action_drop(priv, trap_id);
+	return err;
+}
+
+static const int mlx5e_traps_arr[] = {
+	DEVLINK_TRAP_GENERIC_ID_INGRESS_VLAN_FILTER,
+	DEVLINK_TRAP_GENERIC_ID_DMAC_FILTER,
+};
+
+int mlx5e_apply_traps(struct mlx5e_priv *priv, bool enable)
+{
+	int err;
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(mlx5e_traps_arr); i++) {
+		err = mlx5e_apply_trap(priv, mlx5e_traps_arr[i], enable);
+		if (err)
+			return err;
+	}
+	return 0;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/trap.h b/drivers/net/ethernet/mellanox/mlx5/core/en/trap.h
new file mode 100644
index 000000000..aa3f17658
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/trap.h
@@ -0,0 +1,37 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2020, Mellanox Technologies */
+
+#ifndef __MLX5E_TRAP_H__
+#define __MLX5E_TRAP_H__
+
+#include "../en.h"
+#include "../devlink.h"
+
+struct mlx5e_trap {
+	/* data path */
+	struct mlx5e_rq            rq;
+	struct mlx5e_tir           tir;
+	struct napi_struct         napi;
+	struct device             *pdev;
+	struct net_device         *netdev;
+	__be32                     mkey_be;
+
+	/* data path - accessed per napi poll */
+	struct mlx5e_ch_stats     *stats;
+
+	/* control */
+	struct mlx5e_priv         *priv;
+	struct mlx5_core_dev      *mdev;
+	struct hwtstamp_config    *tstamp;
+	DECLARE_BITMAP(state, MLX5E_CHANNEL_NUM_STATES);
+
+	struct mlx5e_params        params;
+	struct mlx5e_rq_param      rq_param;
+};
+
+void mlx5e_close_trap(struct mlx5e_trap *trap);
+void mlx5e_deactivate_trap(struct mlx5e_priv *priv);
+int mlx5e_handle_trap_event(struct mlx5e_priv *priv, struct mlx5_trap_ctx *trap_ctx);
+int mlx5e_apply_traps(struct mlx5e_priv *priv, bool enable);
+
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/txrx.h b/drivers/net/ethernet/mellanox/mlx5/core/en/txrx.h
new file mode 100644
index 000000000..344245c01
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/txrx.h
@@ -0,0 +1,494 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2019 Mellanox Technologies. */
+
+#ifndef __MLX5_EN_TXRX_H___
+#define __MLX5_EN_TXRX_H___
+
+#include "en.h"
+#include <linux/indirect_call_wrapper.h>
+
+#define MLX5E_TX_WQE_EMPTY_DS_COUNT (sizeof(struct mlx5e_tx_wqe) / MLX5_SEND_WQE_DS)
+
+#define INL_HDR_START_SZ (sizeof(((struct mlx5_wqe_eth_seg *)NULL)->inline_hdr.start))
+
+/* IPSEC inline data includes:
+ * 1. ESP trailer: up to 255 bytes of padding, 1 byte for pad length, 1 byte for
+ *    next header.
+ * 2. ESP authentication data: 16 bytes for ICV.
+ */
+#define MLX5E_MAX_TX_IPSEC_DS DIV_ROUND_UP(sizeof(struct mlx5_wqe_inline_seg) + \
+					   255 + 1 + 1 + 16, MLX5_SEND_WQE_DS)
+
+/* 366 should be big enough to cover all L2, L3 and L4 headers with possible
+ * encapsulations.
+ */
+#define MLX5E_MAX_TX_INLINE_DS DIV_ROUND_UP(366 - INL_HDR_START_SZ + VLAN_HLEN, \
+					    MLX5_SEND_WQE_DS)
+
+/* Sync the calculation with mlx5e_sq_calc_wqe_attr. */
+#define MLX5E_MAX_TX_WQEBBS DIV_ROUND_UP(MLX5E_TX_WQE_EMPTY_DS_COUNT + \
+					 MLX5E_MAX_TX_INLINE_DS + \
+					 MLX5E_MAX_TX_IPSEC_DS + \
+					 MAX_SKB_FRAGS + 1, \
+					 MLX5_SEND_WQEBB_NUM_DS)
+
+#define MLX5E_RX_ERR_CQE(cqe) (get_cqe_opcode(cqe) != MLX5_CQE_RESP_SEND)
+
+static inline
+ktime_t mlx5e_cqe_ts_to_ns(cqe_ts_to_ns func, struct mlx5_clock *clock, u64 cqe_ts)
+{
+	return INDIRECT_CALL_2(func, mlx5_real_time_cyc2time, mlx5_timecounter_cyc2time,
+			       clock, cqe_ts);
+}
+
+enum mlx5e_icosq_wqe_type {
+	MLX5E_ICOSQ_WQE_NOP,
+	MLX5E_ICOSQ_WQE_UMR_RX,
+	MLX5E_ICOSQ_WQE_SHAMPO_HD_UMR,
+#ifdef CONFIG_MLX5_EN_TLS
+	MLX5E_ICOSQ_WQE_UMR_TLS,
+	MLX5E_ICOSQ_WQE_SET_PSV_TLS,
+	MLX5E_ICOSQ_WQE_GET_PSV_TLS,
+#endif
+};
+
+/* General */
+static inline bool mlx5e_skb_is_multicast(struct sk_buff *skb)
+{
+	return skb->pkt_type == PACKET_MULTICAST || skb->pkt_type == PACKET_BROADCAST;
+}
+
+void mlx5e_trigger_irq(struct mlx5e_icosq *sq);
+void mlx5e_completion_event(struct mlx5_core_cq *mcq, struct mlx5_eqe *eqe);
+void mlx5e_cq_error_event(struct mlx5_core_cq *mcq, enum mlx5_event event);
+int mlx5e_napi_poll(struct napi_struct *napi, int budget);
+int mlx5e_poll_ico_cq(struct mlx5e_cq *cq);
+
+/* RX */
+void mlx5e_page_dma_unmap(struct mlx5e_rq *rq, struct page *page);
+void mlx5e_page_release_dynamic(struct mlx5e_rq *rq, struct page *page, bool recycle);
+INDIRECT_CALLABLE_DECLARE(bool mlx5e_post_rx_wqes(struct mlx5e_rq *rq));
+INDIRECT_CALLABLE_DECLARE(bool mlx5e_post_rx_mpwqes(struct mlx5e_rq *rq));
+int mlx5e_poll_rx_cq(struct mlx5e_cq *cq, int budget);
+void mlx5e_free_rx_descs(struct mlx5e_rq *rq);
+void mlx5e_free_rx_in_progress_descs(struct mlx5e_rq *rq);
+
+/* TX */
+netdev_tx_t mlx5e_xmit(struct sk_buff *skb, struct net_device *dev);
+bool mlx5e_poll_tx_cq(struct mlx5e_cq *cq, int napi_budget);
+void mlx5e_free_txqsq_descs(struct mlx5e_txqsq *sq);
+
+static inline bool
+mlx5e_skb_fifo_has_room(struct mlx5e_skb_fifo *fifo)
+{
+	return (u16)(*fifo->pc - *fifo->cc) < fifo->mask;
+}
+
+static inline bool
+mlx5e_wqc_has_room_for(struct mlx5_wq_cyc *wq, u16 cc, u16 pc, u16 n)
+{
+	return (mlx5_wq_cyc_ctr2ix(wq, cc - pc) >= n) || (cc == pc);
+}
+
+static inline void *mlx5e_fetch_wqe(struct mlx5_wq_cyc *wq, u16 pi, size_t wqe_size)
+{
+	void *wqe;
+
+	wqe = mlx5_wq_cyc_get_wqe(wq, pi);
+	memset(wqe, 0, wqe_size);
+
+	return wqe;
+}
+
+#define MLX5E_TX_FETCH_WQE(sq, pi) \
+	((struct mlx5e_tx_wqe *)mlx5e_fetch_wqe(&(sq)->wq, pi, sizeof(struct mlx5e_tx_wqe)))
+
+static inline struct mlx5e_tx_wqe *
+mlx5e_post_nop(struct mlx5_wq_cyc *wq, u32 sqn, u16 *pc)
+{
+	u16                         pi   = mlx5_wq_cyc_ctr2ix(wq, *pc);
+	struct mlx5e_tx_wqe        *wqe  = mlx5_wq_cyc_get_wqe(wq, pi);
+	struct mlx5_wqe_ctrl_seg   *cseg = &wqe->ctrl;
+
+	memset(cseg, 0, sizeof(*cseg));
+
+	cseg->opmod_idx_opcode = cpu_to_be32((*pc << 8) | MLX5_OPCODE_NOP);
+	cseg->qpn_ds           = cpu_to_be32((sqn << 8) | 0x01);
+
+	(*pc)++;
+
+	return wqe;
+}
+
+static inline struct mlx5e_tx_wqe *
+mlx5e_post_nop_fence(struct mlx5_wq_cyc *wq, u32 sqn, u16 *pc)
+{
+	u16                         pi   = mlx5_wq_cyc_ctr2ix(wq, *pc);
+	struct mlx5e_tx_wqe        *wqe  = mlx5_wq_cyc_get_wqe(wq, pi);
+	struct mlx5_wqe_ctrl_seg   *cseg = &wqe->ctrl;
+
+	memset(cseg, 0, sizeof(*cseg));
+
+	cseg->opmod_idx_opcode = cpu_to_be32((*pc << 8) | MLX5_OPCODE_NOP);
+	cseg->qpn_ds           = cpu_to_be32((sqn << 8) | 0x01);
+	cseg->fm_ce_se         = MLX5_FENCE_MODE_INITIATOR_SMALL;
+
+	(*pc)++;
+
+	return wqe;
+}
+
+struct mlx5e_tx_wqe_info {
+	struct sk_buff *skb;
+	u32 num_bytes;
+	u8 num_wqebbs;
+	u8 num_dma;
+	u8 num_fifo_pkts;
+#ifdef CONFIG_MLX5_EN_TLS
+	struct page *resync_dump_frag_page;
+#endif
+};
+
+static inline u16 mlx5e_txqsq_get_next_pi(struct mlx5e_txqsq *sq, u16 size)
+{
+	struct mlx5_wq_cyc *wq = &sq->wq;
+	u16 pi, contig_wqebbs;
+
+	pi = mlx5_wq_cyc_ctr2ix(wq, sq->pc);
+	contig_wqebbs = mlx5_wq_cyc_get_contig_wqebbs(wq, pi);
+	if (unlikely(contig_wqebbs < size)) {
+		struct mlx5e_tx_wqe_info *wi, *edge_wi;
+
+		wi = &sq->db.wqe_info[pi];
+		edge_wi = wi + contig_wqebbs;
+
+		/* Fill SQ frag edge with NOPs to avoid WQE wrapping two pages. */
+		for (; wi < edge_wi; wi++) {
+			*wi = (struct mlx5e_tx_wqe_info) {
+				.num_wqebbs = 1,
+			};
+			mlx5e_post_nop(wq, sq->sqn, &sq->pc);
+		}
+		sq->stats->nop += contig_wqebbs;
+
+		pi = mlx5_wq_cyc_ctr2ix(wq, sq->pc);
+	}
+
+	return pi;
+}
+
+void mlx5e_txqsq_wake(struct mlx5e_txqsq *sq);
+
+static inline u16 mlx5e_shampo_get_cqe_header_index(struct mlx5e_rq *rq, struct mlx5_cqe64 *cqe)
+{
+	return be16_to_cpu(cqe->shampo.header_entry_index) & (rq->mpwqe.shampo->hd_per_wq - 1);
+}
+
+struct mlx5e_shampo_umr {
+	u16 len;
+};
+
+struct mlx5e_icosq_wqe_info {
+	u8 wqe_type;
+	u8 num_wqebbs;
+
+	/* Auxiliary data for different wqe types. */
+	union {
+		struct {
+			struct mlx5e_rq *rq;
+		} umr;
+		struct mlx5e_shampo_umr shampo;
+#ifdef CONFIG_MLX5_EN_TLS
+		struct {
+			struct mlx5e_ktls_offload_context_rx *priv_rx;
+		} tls_set_params;
+		struct {
+			struct mlx5e_ktls_rx_resync_buf *buf;
+		} tls_get_params;
+#endif
+	};
+};
+
+void mlx5e_free_icosq_descs(struct mlx5e_icosq *sq);
+
+static inline u16 mlx5e_icosq_get_next_pi(struct mlx5e_icosq *sq, u16 size)
+{
+	struct mlx5_wq_cyc *wq = &sq->wq;
+	u16 pi, contig_wqebbs;
+
+	pi = mlx5_wq_cyc_ctr2ix(wq, sq->pc);
+	contig_wqebbs = mlx5_wq_cyc_get_contig_wqebbs(wq, pi);
+	if (unlikely(contig_wqebbs < size)) {
+		struct mlx5e_icosq_wqe_info *wi, *edge_wi;
+
+		wi = &sq->db.wqe_info[pi];
+		edge_wi = wi + contig_wqebbs;
+
+		/* Fill SQ frag edge with NOPs to avoid WQE wrapping two pages. */
+		for (; wi < edge_wi; wi++) {
+			*wi = (struct mlx5e_icosq_wqe_info) {
+				.wqe_type   = MLX5E_ICOSQ_WQE_NOP,
+				.num_wqebbs = 1,
+			};
+			mlx5e_post_nop(wq, sq->sqn, &sq->pc);
+		}
+
+		pi = mlx5_wq_cyc_ctr2ix(wq, sq->pc);
+	}
+
+	return pi;
+}
+
+static inline void
+mlx5e_notify_hw(struct mlx5_wq_cyc *wq, u16 pc, void __iomem *uar_map,
+		struct mlx5_wqe_ctrl_seg *ctrl)
+{
+	ctrl->fm_ce_se |= MLX5_WQE_CTRL_CQ_UPDATE;
+	/* ensure wqe is visible to device before updating doorbell record */
+	dma_wmb();
+
+	*wq->db = cpu_to_be32(pc);
+
+	/* ensure doorbell record is visible to device before ringing the
+	 * doorbell
+	 */
+	wmb();
+
+	mlx5_write64((__be32 *)ctrl, uar_map);
+}
+
+static inline void mlx5e_cq_arm(struct mlx5e_cq *cq)
+{
+	struct mlx5_core_cq *mcq;
+
+	mcq = &cq->mcq;
+	mlx5_cq_arm(mcq, MLX5_CQ_DB_REQ_NOT, mcq->uar->map, cq->wq.cc);
+}
+
+static inline struct mlx5e_sq_dma *
+mlx5e_dma_get(struct mlx5e_txqsq *sq, u32 i)
+{
+	return &sq->db.dma_fifo[i & sq->dma_fifo_mask];
+}
+
+static inline void
+mlx5e_dma_push(struct mlx5e_txqsq *sq, dma_addr_t addr, u32 size,
+	       enum mlx5e_dma_map_type map_type)
+{
+	struct mlx5e_sq_dma *dma = mlx5e_dma_get(sq, sq->dma_fifo_pc++);
+
+	dma->addr = addr;
+	dma->size = size;
+	dma->type = map_type;
+}
+
+static inline
+struct sk_buff **mlx5e_skb_fifo_get(struct mlx5e_skb_fifo *fifo, u16 i)
+{
+	return &fifo->fifo[i & fifo->mask];
+}
+
+static inline
+void mlx5e_skb_fifo_push(struct mlx5e_skb_fifo *fifo, struct sk_buff *skb)
+{
+	struct sk_buff **skb_item = mlx5e_skb_fifo_get(fifo, (*fifo->pc)++);
+
+	*skb_item = skb;
+}
+
+static inline
+struct sk_buff *mlx5e_skb_fifo_pop(struct mlx5e_skb_fifo *fifo)
+{
+	WARN_ON_ONCE(*fifo->pc == *fifo->cc);
+
+	return *mlx5e_skb_fifo_get(fifo, (*fifo->cc)++);
+}
+
+static inline void
+mlx5e_tx_dma_unmap(struct device *pdev, struct mlx5e_sq_dma *dma)
+{
+	switch (dma->type) {
+	case MLX5E_DMA_MAP_SINGLE:
+		dma_unmap_single(pdev, dma->addr, dma->size, DMA_TO_DEVICE);
+		break;
+	case MLX5E_DMA_MAP_PAGE:
+		dma_unmap_page(pdev, dma->addr, dma->size, DMA_TO_DEVICE);
+		break;
+	default:
+		WARN_ONCE(true, "mlx5e_tx_dma_unmap unknown DMA type!\n");
+	}
+}
+
+void mlx5e_sq_xmit_simple(struct mlx5e_txqsq *sq, struct sk_buff *skb, bool xmit_more);
+void mlx5e_tx_mpwqe_ensure_complete(struct mlx5e_txqsq *sq);
+
+static inline bool mlx5e_tx_mpwqe_is_full(struct mlx5e_tx_mpwqe *session, u8 max_sq_mpw_wqebbs)
+{
+	return session->ds_count == max_sq_mpw_wqebbs * MLX5_SEND_WQEBB_NUM_DS;
+}
+
+static inline void mlx5e_rqwq_reset(struct mlx5e_rq *rq)
+{
+	if (rq->wq_type == MLX5_WQ_TYPE_LINKED_LIST_STRIDING_RQ) {
+		mlx5_wq_ll_reset(&rq->mpwqe.wq);
+		rq->mpwqe.actual_wq_head = 0;
+	} else {
+		mlx5_wq_cyc_reset(&rq->wqe.wq);
+	}
+}
+
+static inline void mlx5e_dump_error_cqe(struct mlx5e_cq *cq, u32 qn,
+					struct mlx5_err_cqe *err_cqe)
+{
+	struct mlx5_cqwq *wq = &cq->wq;
+	u32 ci;
+
+	ci = mlx5_cqwq_ctr2ix(wq, wq->cc - 1);
+
+	netdev_err(cq->netdev,
+		   "Error cqe on cqn 0x%x, ci 0x%x, qn 0x%x, opcode 0x%x, syndrome 0x%x, vendor syndrome 0x%x\n",
+		   cq->mcq.cqn, ci, qn,
+		   get_cqe_opcode((struct mlx5_cqe64 *)err_cqe),
+		   err_cqe->syndrome, err_cqe->vendor_err_synd);
+	mlx5_dump_err_cqe(cq->mdev, err_cqe);
+}
+
+static inline u32 mlx5e_rqwq_get_size(struct mlx5e_rq *rq)
+{
+	switch (rq->wq_type) {
+	case MLX5_WQ_TYPE_LINKED_LIST_STRIDING_RQ:
+		return mlx5_wq_ll_get_size(&rq->mpwqe.wq);
+	default:
+		return mlx5_wq_cyc_get_size(&rq->wqe.wq);
+	}
+}
+
+static inline u32 mlx5e_rqwq_get_cur_sz(struct mlx5e_rq *rq)
+{
+	switch (rq->wq_type) {
+	case MLX5_WQ_TYPE_LINKED_LIST_STRIDING_RQ:
+		return rq->mpwqe.wq.cur_sz;
+	default:
+		return rq->wqe.wq.cur_sz;
+	}
+}
+
+static inline u16 mlx5e_rqwq_get_head(struct mlx5e_rq *rq)
+{
+	switch (rq->wq_type) {
+	case MLX5_WQ_TYPE_LINKED_LIST_STRIDING_RQ:
+		return mlx5_wq_ll_get_head(&rq->mpwqe.wq);
+	default:
+		return mlx5_wq_cyc_get_head(&rq->wqe.wq);
+	}
+}
+
+static inline u16 mlx5e_rqwq_get_wqe_counter(struct mlx5e_rq *rq)
+{
+	switch (rq->wq_type) {
+	case MLX5_WQ_TYPE_LINKED_LIST_STRIDING_RQ:
+		return mlx5_wq_ll_get_counter(&rq->mpwqe.wq);
+	default:
+		return mlx5_wq_cyc_get_counter(&rq->wqe.wq);
+	}
+}
+
+/* SW parser related functions */
+
+struct mlx5e_swp_spec {
+	__be16 l3_proto;
+	u8 l4_proto;
+	u8 is_tun;
+	__be16 tun_l3_proto;
+	u8 tun_l4_proto;
+};
+
+static inline void mlx5e_eseg_swp_offsets_add_vlan(struct mlx5_wqe_eth_seg *eseg)
+{
+	/* SWP offsets are in 2-bytes words */
+	eseg->swp_outer_l3_offset += VLAN_HLEN / 2;
+	eseg->swp_outer_l4_offset += VLAN_HLEN / 2;
+	eseg->swp_inner_l3_offset += VLAN_HLEN / 2;
+	eseg->swp_inner_l4_offset += VLAN_HLEN / 2;
+}
+
+static inline void
+mlx5e_set_eseg_swp(struct sk_buff *skb, struct mlx5_wqe_eth_seg *eseg,
+		   struct mlx5e_swp_spec *swp_spec)
+{
+	/* SWP offsets are in 2-bytes words */
+	eseg->swp_outer_l3_offset = skb_network_offset(skb) / 2;
+	if (swp_spec->l3_proto == htons(ETH_P_IPV6))
+		eseg->swp_flags |= MLX5_ETH_WQE_SWP_OUTER_L3_IPV6;
+	if (swp_spec->l4_proto) {
+		eseg->swp_outer_l4_offset = skb_transport_offset(skb) / 2;
+		if (swp_spec->l4_proto == IPPROTO_UDP)
+			eseg->swp_flags |= MLX5_ETH_WQE_SWP_OUTER_L4_UDP;
+	}
+
+	if (swp_spec->is_tun) {
+		eseg->swp_inner_l3_offset = skb_inner_network_offset(skb) / 2;
+		if (swp_spec->tun_l3_proto == htons(ETH_P_IPV6))
+			eseg->swp_flags |= MLX5_ETH_WQE_SWP_INNER_L3_IPV6;
+	} else { /* typically for ipsec when xfrm mode != XFRM_MODE_TUNNEL */
+		eseg->swp_inner_l3_offset = skb_network_offset(skb) / 2;
+		if (swp_spec->l3_proto == htons(ETH_P_IPV6))
+			eseg->swp_flags |= MLX5_ETH_WQE_SWP_INNER_L3_IPV6;
+	}
+	switch (swp_spec->tun_l4_proto) {
+	case IPPROTO_UDP:
+		eseg->swp_flags |= MLX5_ETH_WQE_SWP_INNER_L4_UDP;
+		fallthrough;
+	case IPPROTO_TCP:
+		eseg->swp_inner_l4_offset = skb_inner_transport_offset(skb) / 2;
+		break;
+	}
+}
+
+#define MLX5E_STOP_ROOM(wqebbs) ((wqebbs) * 2 - 1)
+
+static inline u16 mlx5e_stop_room_for_wqe(struct mlx5_core_dev *mdev, u16 wqe_size)
+{
+	WARN_ON_ONCE(PAGE_SIZE / MLX5_SEND_WQE_BB < mlx5e_get_max_sq_wqebbs(mdev));
+
+	/* A WQE must not cross the page boundary, hence two conditions:
+	 * 1. Its size must not exceed the page size.
+	 * 2. If the WQE size is X, and the space remaining in a page is less
+	 *    than X, this space needs to be padded with NOPs. So, one WQE of
+	 *    size X may require up to X-1 WQEBBs of padding, which makes the
+	 *    stop room of X-1 + X.
+	 * WQE size is also limited by the hardware limit.
+	 */
+	WARN_ONCE(wqe_size > mlx5e_get_max_sq_wqebbs(mdev),
+		  "wqe_size %u is greater than max SQ WQEBBs %u",
+		  wqe_size, mlx5e_get_max_sq_wqebbs(mdev));
+
+	return MLX5E_STOP_ROOM(wqe_size);
+}
+
+static inline u16 mlx5e_stop_room_for_max_wqe(struct mlx5_core_dev *mdev)
+{
+	return MLX5E_STOP_ROOM(mlx5e_get_max_sq_wqebbs(mdev));
+}
+
+static inline u16 mlx5e_stop_room_for_mpwqe(struct mlx5_core_dev *mdev)
+{
+	u8 mpwqe_wqebbs = mlx5e_get_max_sq_aligned_wqebbs(mdev);
+
+	return mlx5e_stop_room_for_wqe(mdev, mpwqe_wqebbs);
+}
+
+static inline bool mlx5e_icosq_can_post_wqe(struct mlx5e_icosq *sq, u16 wqe_size)
+{
+	u16 room = sq->reserved_room + MLX5E_STOP_ROOM(wqe_size);
+
+	return mlx5e_wqc_has_room_for(&sq->wq, sq->cc, sq->pc, room);
+}
+
+static inline struct mlx5e_mpw_info *mlx5e_get_mpw_info(struct mlx5e_rq *rq, int i)
+{
+	size_t isz = struct_size(rq->mpwqe.info, alloc_units, rq->mpwqe.pages_per_wqe);
+
+	return (struct mlx5e_mpw_info *)((char *)rq->mpwqe.info + array_size(i, isz));
+}
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/xdp.c b/drivers/net/ethernet/mellanox/mlx5/core/en/xdp.c
new file mode 100644
index 000000000..20507ef2f
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/xdp.c
@@ -0,0 +1,691 @@
+/*
+ * Copyright (c) 2018, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/bpf_trace.h>
+#include <net/xdp_sock_drv.h>
+#include "en/xdp.h"
+#include "en/params.h"
+
+int mlx5e_xdp_max_mtu(struct mlx5e_params *params, struct mlx5e_xsk_param *xsk)
+{
+	int hr = mlx5e_get_linear_rq_headroom(params, xsk);
+
+	/* Let S := SKB_DATA_ALIGN(sizeof(struct skb_shared_info)).
+	 * The condition checked in mlx5e_rx_is_linear_skb is:
+	 *   SKB_DATA_ALIGN(sw_mtu + hard_mtu + hr) + S <= PAGE_SIZE         (1)
+	 *   (Note that hw_mtu == sw_mtu + hard_mtu.)
+	 * What is returned from this function is:
+	 *   max_mtu = PAGE_SIZE - S - hr - hard_mtu                         (2)
+	 * After assigning sw_mtu := max_mtu, the left side of (1) turns to
+	 * SKB_DATA_ALIGN(PAGE_SIZE - S) + S, which is equal to PAGE_SIZE,
+	 * because both PAGE_SIZE and S are already aligned. Any number greater
+	 * than max_mtu would make the left side of (1) greater than PAGE_SIZE,
+	 * so max_mtu is the maximum MTU allowed.
+	 */
+
+	return MLX5E_HW2SW_MTU(params, SKB_MAX_HEAD(hr));
+}
+
+static inline bool
+mlx5e_xmit_xdp_buff(struct mlx5e_xdpsq *sq, struct mlx5e_rq *rq,
+		    struct page *page, struct xdp_buff *xdp)
+{
+	struct skb_shared_info *sinfo = NULL;
+	struct mlx5e_xmit_data xdptxd;
+	struct mlx5e_xdp_info xdpi;
+	struct xdp_frame *xdpf;
+	dma_addr_t dma_addr;
+	int i;
+
+	xdpf = xdp_convert_buff_to_frame(xdp);
+	if (unlikely(!xdpf))
+		return false;
+
+	xdptxd.data = xdpf->data;
+	xdptxd.len  = xdpf->len;
+
+	if (xdp->rxq->mem.type == MEM_TYPE_XSK_BUFF_POOL) {
+		/* The xdp_buff was in the UMEM and was copied into a newly
+		 * allocated page. The UMEM page was returned via the ZCA, and
+		 * this new page has to be mapped at this point and has to be
+		 * unmapped and returned via xdp_return_frame on completion.
+		 */
+
+		/* Prevent double recycling of the UMEM page. Even in case this
+		 * function returns false, the xdp_buff shouldn't be recycled,
+		 * as it was already done in xdp_convert_zc_to_xdp_frame.
+		 */
+		__set_bit(MLX5E_RQ_FLAG_XDP_XMIT, rq->flags); /* non-atomic */
+
+		xdpi.mode = MLX5E_XDP_XMIT_MODE_FRAME;
+
+		dma_addr = dma_map_single(sq->pdev, xdptxd.data, xdptxd.len,
+					  DMA_TO_DEVICE);
+		if (dma_mapping_error(sq->pdev, dma_addr)) {
+			xdp_return_frame(xdpf);
+			return false;
+		}
+
+		xdptxd.dma_addr     = dma_addr;
+		xdpi.frame.xdpf     = xdpf;
+		xdpi.frame.dma_addr = dma_addr;
+
+		if (unlikely(!INDIRECT_CALL_2(sq->xmit_xdp_frame, mlx5e_xmit_xdp_frame_mpwqe,
+					      mlx5e_xmit_xdp_frame, sq, &xdptxd, NULL, 0)))
+			return false;
+
+		mlx5e_xdpi_fifo_push(&sq->db.xdpi_fifo, &xdpi);
+		return true;
+	}
+
+	/* Driver assumes that xdp_convert_buff_to_frame returns an xdp_frame
+	 * that points to the same memory region as the original xdp_buff. It
+	 * allows to map the memory only once and to use the DMA_BIDIRECTIONAL
+	 * mode.
+	 */
+
+	xdpi.mode = MLX5E_XDP_XMIT_MODE_PAGE;
+	xdpi.page.rq = rq;
+
+	dma_addr = page_pool_get_dma_addr(page) + (xdpf->data - (void *)xdpf);
+	dma_sync_single_for_device(sq->pdev, dma_addr, xdptxd.len, DMA_BIDIRECTIONAL);
+
+	if (unlikely(xdp_frame_has_frags(xdpf))) {
+		sinfo = xdp_get_shared_info_from_frame(xdpf);
+
+		for (i = 0; i < sinfo->nr_frags; i++) {
+			skb_frag_t *frag = &sinfo->frags[i];
+			dma_addr_t addr;
+			u32 len;
+
+			addr = page_pool_get_dma_addr(skb_frag_page(frag)) +
+				skb_frag_off(frag);
+			len = skb_frag_size(frag);
+			dma_sync_single_for_device(sq->pdev, addr, len,
+						   DMA_BIDIRECTIONAL);
+		}
+	}
+
+	xdptxd.dma_addr = dma_addr;
+
+	if (unlikely(!INDIRECT_CALL_2(sq->xmit_xdp_frame, mlx5e_xmit_xdp_frame_mpwqe,
+				      mlx5e_xmit_xdp_frame, sq, &xdptxd, sinfo, 0)))
+		return false;
+
+	xdpi.page.page = page;
+	mlx5e_xdpi_fifo_push(&sq->db.xdpi_fifo, &xdpi);
+
+	if (unlikely(xdp_frame_has_frags(xdpf))) {
+		for (i = 0; i < sinfo->nr_frags; i++) {
+			skb_frag_t *frag = &sinfo->frags[i];
+
+			xdpi.page.page = skb_frag_page(frag);
+			mlx5e_xdpi_fifo_push(&sq->db.xdpi_fifo, &xdpi);
+		}
+	}
+
+	return true;
+}
+
+/* returns true if packet was consumed by xdp */
+bool mlx5e_xdp_handle(struct mlx5e_rq *rq, struct page *page,
+		      struct bpf_prog *prog, struct xdp_buff *xdp)
+{
+	u32 act;
+	int err;
+
+	act = bpf_prog_run_xdp(prog, xdp);
+	switch (act) {
+	case XDP_PASS:
+		return false;
+	case XDP_TX:
+		if (unlikely(!mlx5e_xmit_xdp_buff(rq->xdpsq, rq, page, xdp)))
+			goto xdp_abort;
+		__set_bit(MLX5E_RQ_FLAG_XDP_XMIT, rq->flags); /* non-atomic */
+		return true;
+	case XDP_REDIRECT:
+		/* When XDP enabled then page-refcnt==1 here */
+		err = xdp_do_redirect(rq->netdev, xdp, prog);
+		if (unlikely(err))
+			goto xdp_abort;
+		__set_bit(MLX5E_RQ_FLAG_XDP_XMIT, rq->flags);
+		__set_bit(MLX5E_RQ_FLAG_XDP_REDIRECT, rq->flags);
+		if (xdp->rxq->mem.type != MEM_TYPE_XSK_BUFF_POOL)
+			mlx5e_page_dma_unmap(rq, page);
+		rq->stats->xdp_redirect++;
+		return true;
+	default:
+		bpf_warn_invalid_xdp_action(rq->netdev, prog, act);
+		fallthrough;
+	case XDP_ABORTED:
+xdp_abort:
+		trace_xdp_exception(rq->netdev, prog, act);
+		fallthrough;
+	case XDP_DROP:
+		rq->stats->xdp_drop++;
+		return true;
+	}
+}
+
+static u16 mlx5e_xdpsq_get_next_pi(struct mlx5e_xdpsq *sq, u16 size)
+{
+	struct mlx5_wq_cyc *wq = &sq->wq;
+	u16 pi, contig_wqebbs;
+
+	pi = mlx5_wq_cyc_ctr2ix(wq, sq->pc);
+	contig_wqebbs = mlx5_wq_cyc_get_contig_wqebbs(wq, pi);
+	if (unlikely(contig_wqebbs < size)) {
+		struct mlx5e_xdp_wqe_info *wi, *edge_wi;
+
+		wi = &sq->db.wqe_info[pi];
+		edge_wi = wi + contig_wqebbs;
+
+		/* Fill SQ frag edge with NOPs to avoid WQE wrapping two pages. */
+		for (; wi < edge_wi; wi++) {
+			*wi = (struct mlx5e_xdp_wqe_info) {
+				.num_wqebbs = 1,
+				.num_pkts = 0,
+			};
+			mlx5e_post_nop(wq, sq->sqn, &sq->pc);
+		}
+		sq->stats->nops += contig_wqebbs;
+
+		pi = mlx5_wq_cyc_ctr2ix(wq, sq->pc);
+	}
+
+	return pi;
+}
+
+static void mlx5e_xdp_mpwqe_session_start(struct mlx5e_xdpsq *sq)
+{
+	struct mlx5e_tx_mpwqe *session = &sq->mpwqe;
+	struct mlx5e_xdpsq_stats *stats = sq->stats;
+	struct mlx5e_tx_wqe *wqe;
+	u16 pi;
+
+	pi = mlx5e_xdpsq_get_next_pi(sq, sq->max_sq_mpw_wqebbs);
+	wqe = MLX5E_TX_FETCH_WQE(sq, pi);
+	net_prefetchw(wqe->data);
+
+	*session = (struct mlx5e_tx_mpwqe) {
+		.wqe = wqe,
+		.bytes_count = 0,
+		.ds_count = MLX5E_TX_WQE_EMPTY_DS_COUNT,
+		.pkt_count = 0,
+		.inline_on = mlx5e_xdp_get_inline_state(sq, session->inline_on),
+	};
+
+	stats->mpwqe++;
+}
+
+void mlx5e_xdp_mpwqe_complete(struct mlx5e_xdpsq *sq)
+{
+	struct mlx5_wq_cyc       *wq    = &sq->wq;
+	struct mlx5e_tx_mpwqe *session = &sq->mpwqe;
+	struct mlx5_wqe_ctrl_seg *cseg = &session->wqe->ctrl;
+	u16 ds_count = session->ds_count;
+	u16 pi = mlx5_wq_cyc_ctr2ix(wq, sq->pc);
+	struct mlx5e_xdp_wqe_info *wi = &sq->db.wqe_info[pi];
+
+	cseg->opmod_idx_opcode =
+		cpu_to_be32((sq->pc << 8) | MLX5_OPCODE_ENHANCED_MPSW);
+	cseg->qpn_ds = cpu_to_be32((sq->sqn << 8) | ds_count);
+
+	wi->num_wqebbs = DIV_ROUND_UP(ds_count, MLX5_SEND_WQEBB_NUM_DS);
+	wi->num_pkts   = session->pkt_count;
+
+	sq->pc += wi->num_wqebbs;
+
+	sq->doorbell_cseg = cseg;
+
+	session->wqe = NULL; /* Close session */
+}
+
+enum {
+	MLX5E_XDP_CHECK_OK = 1,
+	MLX5E_XDP_CHECK_START_MPWQE = 2,
+};
+
+INDIRECT_CALLABLE_SCOPE int mlx5e_xmit_xdp_frame_check_mpwqe(struct mlx5e_xdpsq *sq)
+{
+	if (unlikely(!sq->mpwqe.wqe)) {
+		if (unlikely(!mlx5e_wqc_has_room_for(&sq->wq, sq->cc, sq->pc,
+						     sq->stop_room))) {
+			/* SQ is full, ring doorbell */
+			mlx5e_xmit_xdp_doorbell(sq);
+			sq->stats->full++;
+			return -EBUSY;
+		}
+
+		return MLX5E_XDP_CHECK_START_MPWQE;
+	}
+
+	return MLX5E_XDP_CHECK_OK;
+}
+
+INDIRECT_CALLABLE_SCOPE bool
+mlx5e_xmit_xdp_frame(struct mlx5e_xdpsq *sq, struct mlx5e_xmit_data *xdptxd,
+		     struct skb_shared_info *sinfo, int check_result);
+
+INDIRECT_CALLABLE_SCOPE bool
+mlx5e_xmit_xdp_frame_mpwqe(struct mlx5e_xdpsq *sq, struct mlx5e_xmit_data *xdptxd,
+			   struct skb_shared_info *sinfo, int check_result)
+{
+	struct mlx5e_tx_mpwqe *session = &sq->mpwqe;
+	struct mlx5e_xdpsq_stats *stats = sq->stats;
+
+	if (unlikely(sinfo)) {
+		/* MPWQE is enabled, but a multi-buffer packet is queued for
+		 * transmission. MPWQE can't send fragmented packets, so close
+		 * the current session and fall back to a regular WQE.
+		 */
+		if (unlikely(sq->mpwqe.wqe))
+			mlx5e_xdp_mpwqe_complete(sq);
+		return mlx5e_xmit_xdp_frame(sq, xdptxd, sinfo, 0);
+	}
+
+	if (unlikely(xdptxd->len > sq->hw_mtu)) {
+		stats->err++;
+		return false;
+	}
+
+	if (!check_result)
+		check_result = mlx5e_xmit_xdp_frame_check_mpwqe(sq);
+	if (unlikely(check_result < 0))
+		return false;
+
+	if (check_result == MLX5E_XDP_CHECK_START_MPWQE) {
+		/* Start the session when nothing can fail, so it's guaranteed
+		 * that if there is an active session, it has at least one dseg,
+		 * and it's safe to complete it at any time.
+		 */
+		mlx5e_xdp_mpwqe_session_start(sq);
+	}
+
+	mlx5e_xdp_mpwqe_add_dseg(sq, xdptxd, stats);
+
+	if (unlikely(mlx5e_xdp_mpwqe_is_full(session, sq->max_sq_mpw_wqebbs)))
+		mlx5e_xdp_mpwqe_complete(sq);
+
+	stats->xmit++;
+	return true;
+}
+
+static int mlx5e_xmit_xdp_frame_check_stop_room(struct mlx5e_xdpsq *sq, int stop_room)
+{
+	if (unlikely(!mlx5e_wqc_has_room_for(&sq->wq, sq->cc, sq->pc, stop_room))) {
+		/* SQ is full, ring doorbell */
+		mlx5e_xmit_xdp_doorbell(sq);
+		sq->stats->full++;
+		return -EBUSY;
+	}
+
+	return MLX5E_XDP_CHECK_OK;
+}
+
+INDIRECT_CALLABLE_SCOPE int mlx5e_xmit_xdp_frame_check(struct mlx5e_xdpsq *sq)
+{
+	return mlx5e_xmit_xdp_frame_check_stop_room(sq, 1);
+}
+
+INDIRECT_CALLABLE_SCOPE bool
+mlx5e_xmit_xdp_frame(struct mlx5e_xdpsq *sq, struct mlx5e_xmit_data *xdptxd,
+		     struct skb_shared_info *sinfo, int check_result)
+{
+	struct mlx5_wq_cyc       *wq   = &sq->wq;
+	struct mlx5_wqe_ctrl_seg *cseg;
+	struct mlx5_wqe_data_seg *dseg;
+	struct mlx5_wqe_eth_seg *eseg;
+	struct mlx5e_tx_wqe *wqe;
+
+	dma_addr_t dma_addr = xdptxd->dma_addr;
+	u32 dma_len = xdptxd->len;
+	u16 ds_cnt, inline_hdr_sz;
+	u8 num_wqebbs = 1;
+	int num_frags = 0;
+	u16 pi;
+
+	struct mlx5e_xdpsq_stats *stats = sq->stats;
+
+	if (unlikely(dma_len < MLX5E_XDP_MIN_INLINE || sq->hw_mtu < dma_len)) {
+		stats->err++;
+		return false;
+	}
+
+	ds_cnt = MLX5E_TX_WQE_EMPTY_DS_COUNT + 1;
+	if (sq->min_inline_mode != MLX5_INLINE_MODE_NONE)
+		ds_cnt++;
+
+	/* check_result must be 0 if sinfo is passed. */
+	if (!check_result) {
+		int stop_room = 1;
+
+		if (unlikely(sinfo)) {
+			ds_cnt += sinfo->nr_frags;
+			num_frags = sinfo->nr_frags;
+			num_wqebbs = DIV_ROUND_UP(ds_cnt, MLX5_SEND_WQEBB_NUM_DS);
+			/* Assuming MLX5_CAP_GEN(mdev, max_wqe_sz_sq) is big
+			 * enough to hold all fragments.
+			 */
+			stop_room = MLX5E_STOP_ROOM(num_wqebbs);
+		}
+
+		check_result = mlx5e_xmit_xdp_frame_check_stop_room(sq, stop_room);
+	}
+	if (unlikely(check_result < 0))
+		return false;
+
+	pi = mlx5e_xdpsq_get_next_pi(sq, num_wqebbs);
+	wqe = mlx5_wq_cyc_get_wqe(wq, pi);
+	net_prefetchw(wqe);
+
+	cseg = &wqe->ctrl;
+	eseg = &wqe->eth;
+	dseg = wqe->data;
+
+	inline_hdr_sz = 0;
+
+	/* copy the inline part if required */
+	if (sq->min_inline_mode != MLX5_INLINE_MODE_NONE) {
+		memcpy(eseg->inline_hdr.start, xdptxd->data, sizeof(eseg->inline_hdr.start));
+		memcpy(dseg, xdptxd->data + sizeof(eseg->inline_hdr.start),
+		       MLX5E_XDP_MIN_INLINE - sizeof(eseg->inline_hdr.start));
+		dma_len  -= MLX5E_XDP_MIN_INLINE;
+		dma_addr += MLX5E_XDP_MIN_INLINE;
+		inline_hdr_sz = MLX5E_XDP_MIN_INLINE;
+		dseg++;
+	}
+
+	/* write the dma part */
+	dseg->addr       = cpu_to_be64(dma_addr);
+	dseg->byte_count = cpu_to_be32(dma_len);
+
+	cseg->opmod_idx_opcode = cpu_to_be32((sq->pc << 8) | MLX5_OPCODE_SEND);
+
+	if (unlikely(test_bit(MLX5E_SQ_STATE_XDP_MULTIBUF, &sq->state))) {
+		u8 num_pkts = 1 + num_frags;
+		int i;
+
+		memset(&cseg->trailer, 0, sizeof(cseg->trailer));
+		memset(eseg, 0, sizeof(*eseg) - sizeof(eseg->trailer));
+
+		eseg->inline_hdr.sz = cpu_to_be16(inline_hdr_sz);
+		dseg->lkey = sq->mkey_be;
+
+		for (i = 0; i < num_frags; i++) {
+			skb_frag_t *frag = &sinfo->frags[i];
+			dma_addr_t addr;
+
+			addr = page_pool_get_dma_addr(skb_frag_page(frag)) +
+				skb_frag_off(frag);
+
+			dseg++;
+			dseg->addr = cpu_to_be64(addr);
+			dseg->byte_count = cpu_to_be32(skb_frag_size(frag));
+			dseg->lkey = sq->mkey_be;
+		}
+
+		cseg->qpn_ds = cpu_to_be32((sq->sqn << 8) | ds_cnt);
+
+		sq->db.wqe_info[pi] = (struct mlx5e_xdp_wqe_info) {
+			.num_wqebbs = num_wqebbs,
+			.num_pkts = num_pkts,
+		};
+
+		sq->pc += num_wqebbs;
+	} else {
+		cseg->fm_ce_se = 0;
+
+		sq->pc++;
+	}
+
+	sq->doorbell_cseg = cseg;
+
+	stats->xmit++;
+	return true;
+}
+
+static void mlx5e_free_xdpsq_desc(struct mlx5e_xdpsq *sq,
+				  struct mlx5e_xdp_wqe_info *wi,
+				  u32 *xsk_frames,
+				  bool recycle,
+				  struct xdp_frame_bulk *bq)
+{
+	struct mlx5e_xdp_info_fifo *xdpi_fifo = &sq->db.xdpi_fifo;
+	u16 i;
+
+	for (i = 0; i < wi->num_pkts; i++) {
+		struct mlx5e_xdp_info xdpi = mlx5e_xdpi_fifo_pop(xdpi_fifo);
+
+		switch (xdpi.mode) {
+		case MLX5E_XDP_XMIT_MODE_FRAME:
+			/* XDP_TX from the XSK RQ and XDP_REDIRECT */
+			dma_unmap_single(sq->pdev, xdpi.frame.dma_addr,
+					 xdpi.frame.xdpf->len, DMA_TO_DEVICE);
+			xdp_return_frame_bulk(xdpi.frame.xdpf, bq);
+			break;
+		case MLX5E_XDP_XMIT_MODE_PAGE:
+			/* XDP_TX from the regular RQ */
+			mlx5e_page_release_dynamic(xdpi.page.rq, xdpi.page.page, recycle);
+			break;
+		case MLX5E_XDP_XMIT_MODE_XSK:
+			/* AF_XDP send */
+			(*xsk_frames)++;
+			break;
+		default:
+			WARN_ON_ONCE(true);
+		}
+	}
+}
+
+bool mlx5e_poll_xdpsq_cq(struct mlx5e_cq *cq)
+{
+	struct xdp_frame_bulk bq;
+	struct mlx5e_xdpsq *sq;
+	struct mlx5_cqe64 *cqe;
+	u32 xsk_frames = 0;
+	u16 sqcc;
+	int i;
+
+	xdp_frame_bulk_init(&bq);
+
+	sq = container_of(cq, struct mlx5e_xdpsq, cq);
+
+	if (unlikely(!test_bit(MLX5E_SQ_STATE_ENABLED, &sq->state)))
+		return false;
+
+	cqe = mlx5_cqwq_get_cqe(&cq->wq);
+	if (!cqe)
+		return false;
+
+	/* sq->cc must be updated only after mlx5_cqwq_update_db_record(),
+	 * otherwise a cq overrun may occur
+	 */
+	sqcc = sq->cc;
+
+	i = 0;
+	do {
+		struct mlx5e_xdp_wqe_info *wi;
+		u16 wqe_counter, ci;
+		bool last_wqe;
+
+		mlx5_cqwq_pop(&cq->wq);
+
+		wqe_counter = be16_to_cpu(cqe->wqe_counter);
+
+		do {
+			last_wqe = (sqcc == wqe_counter);
+			ci = mlx5_wq_cyc_ctr2ix(&sq->wq, sqcc);
+			wi = &sq->db.wqe_info[ci];
+
+			sqcc += wi->num_wqebbs;
+
+			mlx5e_free_xdpsq_desc(sq, wi, &xsk_frames, true, &bq);
+		} while (!last_wqe);
+
+		if (unlikely(get_cqe_opcode(cqe) != MLX5_CQE_REQ)) {
+			netdev_WARN_ONCE(sq->channel->netdev,
+					 "Bad OP in XDPSQ CQE: 0x%x\n",
+					 get_cqe_opcode(cqe));
+			mlx5e_dump_error_cqe(&sq->cq, sq->sqn,
+					     (struct mlx5_err_cqe *)cqe);
+			mlx5_wq_cyc_wqe_dump(&sq->wq, ci, wi->num_wqebbs);
+		}
+	} while ((++i < MLX5E_TX_CQ_POLL_BUDGET) && (cqe = mlx5_cqwq_get_cqe(&cq->wq)));
+
+	xdp_flush_frame_bulk(&bq);
+
+	if (xsk_frames)
+		xsk_tx_completed(sq->xsk_pool, xsk_frames);
+
+	sq->stats->cqes += i;
+
+	mlx5_cqwq_update_db_record(&cq->wq);
+
+	/* ensure cq space is freed before enabling more cqes */
+	wmb();
+
+	sq->cc = sqcc;
+	return (i == MLX5E_TX_CQ_POLL_BUDGET);
+}
+
+void mlx5e_free_xdpsq_descs(struct mlx5e_xdpsq *sq)
+{
+	struct xdp_frame_bulk bq;
+	u32 xsk_frames = 0;
+
+	xdp_frame_bulk_init(&bq);
+
+	rcu_read_lock(); /* need for xdp_return_frame_bulk */
+
+	while (sq->cc != sq->pc) {
+		struct mlx5e_xdp_wqe_info *wi;
+		u16 ci;
+
+		ci = mlx5_wq_cyc_ctr2ix(&sq->wq, sq->cc);
+		wi = &sq->db.wqe_info[ci];
+
+		sq->cc += wi->num_wqebbs;
+
+		mlx5e_free_xdpsq_desc(sq, wi, &xsk_frames, false, &bq);
+	}
+
+	xdp_flush_frame_bulk(&bq);
+	rcu_read_unlock();
+
+	if (xsk_frames)
+		xsk_tx_completed(sq->xsk_pool, xsk_frames);
+}
+
+int mlx5e_xdp_xmit(struct net_device *dev, int n, struct xdp_frame **frames,
+		   u32 flags)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	struct mlx5e_xdpsq *sq;
+	int nxmit = 0;
+	int sq_num;
+	int i;
+
+	/* this flag is sufficient, no need to test internal sq state */
+	if (unlikely(!mlx5e_xdp_tx_is_enabled(priv)))
+		return -ENETDOWN;
+
+	if (unlikely(flags & ~XDP_XMIT_FLAGS_MASK))
+		return -EINVAL;
+
+	sq_num = smp_processor_id();
+
+	if (unlikely(sq_num >= priv->channels.num))
+		return -ENXIO;
+
+	sq = &priv->channels.c[sq_num]->xdpsq;
+
+	for (i = 0; i < n; i++) {
+		struct xdp_frame *xdpf = frames[i];
+		struct mlx5e_xmit_data xdptxd;
+		struct mlx5e_xdp_info xdpi;
+		bool ret;
+
+		xdptxd.data = xdpf->data;
+		xdptxd.len = xdpf->len;
+		xdptxd.dma_addr = dma_map_single(sq->pdev, xdptxd.data,
+						 xdptxd.len, DMA_TO_DEVICE);
+
+		if (unlikely(dma_mapping_error(sq->pdev, xdptxd.dma_addr)))
+			break;
+
+		xdpi.mode           = MLX5E_XDP_XMIT_MODE_FRAME;
+		xdpi.frame.xdpf     = xdpf;
+		xdpi.frame.dma_addr = xdptxd.dma_addr;
+
+		ret = INDIRECT_CALL_2(sq->xmit_xdp_frame, mlx5e_xmit_xdp_frame_mpwqe,
+				      mlx5e_xmit_xdp_frame, sq, &xdptxd, NULL, 0);
+		if (unlikely(!ret)) {
+			dma_unmap_single(sq->pdev, xdptxd.dma_addr,
+					 xdptxd.len, DMA_TO_DEVICE);
+			break;
+		}
+		mlx5e_xdpi_fifo_push(&sq->db.xdpi_fifo, &xdpi);
+		nxmit++;
+	}
+
+	if (flags & XDP_XMIT_FLUSH) {
+		if (sq->mpwqe.wqe)
+			mlx5e_xdp_mpwqe_complete(sq);
+		mlx5e_xmit_xdp_doorbell(sq);
+	}
+
+	return nxmit;
+}
+
+void mlx5e_xdp_rx_poll_complete(struct mlx5e_rq *rq)
+{
+	struct mlx5e_xdpsq *xdpsq = rq->xdpsq;
+
+	if (xdpsq->mpwqe.wqe)
+		mlx5e_xdp_mpwqe_complete(xdpsq);
+
+	mlx5e_xmit_xdp_doorbell(xdpsq);
+
+	if (test_bit(MLX5E_RQ_FLAG_XDP_REDIRECT, rq->flags)) {
+		xdp_do_flush_map();
+		__clear_bit(MLX5E_RQ_FLAG_XDP_REDIRECT, rq->flags);
+	}
+}
+
+void mlx5e_set_xmit_fp(struct mlx5e_xdpsq *sq, bool is_mpw)
+{
+	sq->xmit_xdp_frame_check = is_mpw ?
+		mlx5e_xmit_xdp_frame_check_mpwqe : mlx5e_xmit_xdp_frame_check;
+	sq->xmit_xdp_frame = is_mpw ?
+		mlx5e_xmit_xdp_frame_mpwqe : mlx5e_xmit_xdp_frame;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/xdp.h b/drivers/net/ethernet/mellanox/mlx5/core/en/xdp.h
new file mode 100644
index 000000000..bc2d9034a
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/xdp.h
@@ -0,0 +1,186 @@
+/*
+ * Copyright (c) 2018, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+#ifndef __MLX5_EN_XDP_H__
+#define __MLX5_EN_XDP_H__
+
+#include <linux/indirect_call_wrapper.h>
+
+#include "en.h"
+#include "en/txrx.h"
+
+#define MLX5E_XDP_MIN_INLINE (ETH_HLEN + VLAN_HLEN)
+
+#define MLX5E_XDP_INLINE_WQE_MAX_DS_CNT 16
+#define MLX5E_XDP_INLINE_WQE_SZ_THRSD \
+	(MLX5E_XDP_INLINE_WQE_MAX_DS_CNT * MLX5_SEND_WQE_DS - \
+	 sizeof(struct mlx5_wqe_inline_seg))
+
+struct mlx5e_xsk_param;
+int mlx5e_xdp_max_mtu(struct mlx5e_params *params, struct mlx5e_xsk_param *xsk);
+bool mlx5e_xdp_handle(struct mlx5e_rq *rq, struct page *page,
+		      struct bpf_prog *prog, struct xdp_buff *xdp);
+void mlx5e_xdp_mpwqe_complete(struct mlx5e_xdpsq *sq);
+bool mlx5e_poll_xdpsq_cq(struct mlx5e_cq *cq);
+void mlx5e_free_xdpsq_descs(struct mlx5e_xdpsq *sq);
+void mlx5e_set_xmit_fp(struct mlx5e_xdpsq *sq, bool is_mpw);
+void mlx5e_xdp_rx_poll_complete(struct mlx5e_rq *rq);
+int mlx5e_xdp_xmit(struct net_device *dev, int n, struct xdp_frame **frames,
+		   u32 flags);
+
+INDIRECT_CALLABLE_DECLARE(bool mlx5e_xmit_xdp_frame_mpwqe(struct mlx5e_xdpsq *sq,
+							  struct mlx5e_xmit_data *xdptxd,
+							  struct skb_shared_info *sinfo,
+							  int check_result));
+INDIRECT_CALLABLE_DECLARE(bool mlx5e_xmit_xdp_frame(struct mlx5e_xdpsq *sq,
+						    struct mlx5e_xmit_data *xdptxd,
+						    struct skb_shared_info *sinfo,
+						    int check_result));
+INDIRECT_CALLABLE_DECLARE(int mlx5e_xmit_xdp_frame_check_mpwqe(struct mlx5e_xdpsq *sq));
+INDIRECT_CALLABLE_DECLARE(int mlx5e_xmit_xdp_frame_check(struct mlx5e_xdpsq *sq));
+
+static inline void mlx5e_xdp_tx_enable(struct mlx5e_priv *priv)
+{
+	set_bit(MLX5E_STATE_XDP_TX_ENABLED, &priv->state);
+
+	if (priv->channels.params.xdp_prog)
+		set_bit(MLX5E_STATE_XDP_ACTIVE, &priv->state);
+}
+
+static inline void mlx5e_xdp_tx_disable(struct mlx5e_priv *priv)
+{
+	if (priv->channels.params.xdp_prog)
+		clear_bit(MLX5E_STATE_XDP_ACTIVE, &priv->state);
+
+	clear_bit(MLX5E_STATE_XDP_TX_ENABLED, &priv->state);
+	/* Let other device's napi(s) and XSK wakeups see our new state. */
+	synchronize_net();
+}
+
+static inline bool mlx5e_xdp_tx_is_enabled(struct mlx5e_priv *priv)
+{
+	return test_bit(MLX5E_STATE_XDP_TX_ENABLED, &priv->state);
+}
+
+static inline bool mlx5e_xdp_is_active(struct mlx5e_priv *priv)
+{
+	return test_bit(MLX5E_STATE_XDP_ACTIVE, &priv->state);
+}
+
+static inline void mlx5e_xmit_xdp_doorbell(struct mlx5e_xdpsq *sq)
+{
+	if (sq->doorbell_cseg) {
+		mlx5e_notify_hw(&sq->wq, sq->pc, sq->uar_map, sq->doorbell_cseg);
+		sq->doorbell_cseg = NULL;
+	}
+}
+
+/* Enable inline WQEs to shift some load from a congested HCA (HW) to
+ * a less congested cpu (SW).
+ */
+static inline bool mlx5e_xdp_get_inline_state(struct mlx5e_xdpsq *sq, bool cur)
+{
+	u16 outstanding = sq->xdpi_fifo_pc - sq->xdpi_fifo_cc;
+
+#define MLX5E_XDP_INLINE_WATERMARK_LOW	10
+#define MLX5E_XDP_INLINE_WATERMARK_HIGH 128
+
+	if (cur && outstanding <= MLX5E_XDP_INLINE_WATERMARK_LOW)
+		return false;
+
+	if (!cur && outstanding >= MLX5E_XDP_INLINE_WATERMARK_HIGH)
+		return true;
+
+	return cur;
+}
+
+static inline bool mlx5e_xdp_mpwqe_is_full(struct mlx5e_tx_mpwqe *session, u8 max_sq_mpw_wqebbs)
+{
+	if (session->inline_on)
+		return session->ds_count + MLX5E_XDP_INLINE_WQE_MAX_DS_CNT >
+		       max_sq_mpw_wqebbs * MLX5_SEND_WQEBB_NUM_DS;
+
+	return mlx5e_tx_mpwqe_is_full(session, max_sq_mpw_wqebbs);
+}
+
+struct mlx5e_xdp_wqe_info {
+	u8 num_wqebbs;
+	u8 num_pkts;
+};
+
+static inline void
+mlx5e_xdp_mpwqe_add_dseg(struct mlx5e_xdpsq *sq,
+			 struct mlx5e_xmit_data *xdptxd,
+			 struct mlx5e_xdpsq_stats *stats)
+{
+	struct mlx5e_tx_mpwqe *session = &sq->mpwqe;
+	struct mlx5_wqe_data_seg *dseg =
+		(struct mlx5_wqe_data_seg *)session->wqe + session->ds_count;
+	u32 dma_len = xdptxd->len;
+
+	session->pkt_count++;
+	session->bytes_count += dma_len;
+
+	if (session->inline_on && dma_len <= MLX5E_XDP_INLINE_WQE_SZ_THRSD) {
+		struct mlx5_wqe_inline_seg *inline_dseg =
+			(struct mlx5_wqe_inline_seg *)dseg;
+		u16 ds_len = sizeof(*inline_dseg) + dma_len;
+		u16 ds_cnt = DIV_ROUND_UP(ds_len, MLX5_SEND_WQE_DS);
+
+		inline_dseg->byte_count = cpu_to_be32(dma_len | MLX5_INLINE_SEG);
+		memcpy(inline_dseg->data, xdptxd->data, dma_len);
+
+		session->ds_count += ds_cnt;
+		stats->inlnw++;
+		return;
+	}
+
+	dseg->addr       = cpu_to_be64(xdptxd->dma_addr);
+	dseg->byte_count = cpu_to_be32(dma_len);
+	dseg->lkey       = sq->mkey_be;
+	session->ds_count++;
+}
+
+static inline void
+mlx5e_xdpi_fifo_push(struct mlx5e_xdp_info_fifo *fifo,
+		     struct mlx5e_xdp_info *xi)
+{
+	u32 i = (*fifo->pc)++ & fifo->mask;
+
+	fifo->xi[i] = *xi;
+}
+
+static inline struct mlx5e_xdp_info
+mlx5e_xdpi_fifo_pop(struct mlx5e_xdp_info_fifo *fifo)
+{
+	return fifo->xi[(*fifo->cc)++ & fifo->mask];
+}
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/xsk/pool.c b/drivers/net/ethernet/mellanox/mlx5/core/en/xsk/pool.c
new file mode 100644
index 000000000..ebada0c5a
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/xsk/pool.c
@@ -0,0 +1,230 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2019-2020, Mellanox Technologies inc. All rights reserved. */
+
+#include <net/xdp_sock_drv.h>
+#include "pool.h"
+#include "setup.h"
+#include "en/params.h"
+
+static int mlx5e_xsk_map_pool(struct mlx5e_priv *priv,
+			      struct xsk_buff_pool *pool)
+{
+	struct device *dev = mlx5_core_dma_dev(priv->mdev);
+
+	return xsk_pool_dma_map(pool, dev, DMA_ATTR_SKIP_CPU_SYNC);
+}
+
+static void mlx5e_xsk_unmap_pool(struct mlx5e_priv *priv,
+				 struct xsk_buff_pool *pool)
+{
+	return xsk_pool_dma_unmap(pool, DMA_ATTR_SKIP_CPU_SYNC);
+}
+
+static int mlx5e_xsk_get_pools(struct mlx5e_xsk *xsk)
+{
+	if (!xsk->pools) {
+		xsk->pools = kcalloc(MLX5E_MAX_NUM_CHANNELS,
+				     sizeof(*xsk->pools), GFP_KERNEL);
+		if (unlikely(!xsk->pools))
+			return -ENOMEM;
+	}
+
+	xsk->refcnt++;
+	xsk->ever_used = true;
+
+	return 0;
+}
+
+static void mlx5e_xsk_put_pools(struct mlx5e_xsk *xsk)
+{
+	if (!--xsk->refcnt) {
+		kfree(xsk->pools);
+		xsk->pools = NULL;
+	}
+}
+
+static int mlx5e_xsk_add_pool(struct mlx5e_xsk *xsk, struct xsk_buff_pool *pool, u16 ix)
+{
+	int err;
+
+	err = mlx5e_xsk_get_pools(xsk);
+	if (unlikely(err))
+		return err;
+
+	xsk->pools[ix] = pool;
+	return 0;
+}
+
+static void mlx5e_xsk_remove_pool(struct mlx5e_xsk *xsk, u16 ix)
+{
+	xsk->pools[ix] = NULL;
+
+	mlx5e_xsk_put_pools(xsk);
+}
+
+static bool mlx5e_xsk_is_pool_sane(struct xsk_buff_pool *pool)
+{
+	return xsk_pool_get_headroom(pool) <= 0xffff &&
+		xsk_pool_get_chunk_size(pool) <= 0xffff;
+}
+
+void mlx5e_build_xsk_param(struct xsk_buff_pool *pool, struct mlx5e_xsk_param *xsk)
+{
+	xsk->headroom = xsk_pool_get_headroom(pool);
+	xsk->chunk_size = xsk_pool_get_chunk_size(pool);
+	xsk->unaligned = pool->unaligned;
+}
+
+static int mlx5e_xsk_enable_locked(struct mlx5e_priv *priv,
+				   struct xsk_buff_pool *pool, u16 ix)
+{
+	struct mlx5e_params *params = &priv->channels.params;
+	struct mlx5e_xsk_param xsk;
+	struct mlx5e_channel *c;
+	int err;
+
+	if (unlikely(mlx5e_xsk_get_pool(&priv->channels.params, &priv->xsk, ix)))
+		return -EBUSY;
+
+	if (unlikely(!mlx5e_xsk_is_pool_sane(pool)))
+		return -EINVAL;
+
+	err = mlx5e_xsk_map_pool(priv, pool);
+	if (unlikely(err))
+		return err;
+
+	err = mlx5e_xsk_add_pool(&priv->xsk, pool, ix);
+	if (unlikely(err))
+		goto err_unmap_pool;
+
+	mlx5e_build_xsk_param(pool, &xsk);
+
+	if (priv->channels.params.rq_wq_type == MLX5_WQ_TYPE_LINKED_LIST_STRIDING_RQ &&
+	    mlx5e_mpwrq_umr_mode(priv->mdev, &xsk) == MLX5E_MPWRQ_UMR_MODE_OVERSIZED) {
+		const char *recommendation = is_power_of_2(xsk.chunk_size) ?
+			"Upgrade firmware" : "Disable striding RQ";
+
+		mlx5_core_warn(priv->mdev, "Expected slowdown with XSK frame size %u. %s for better performance.\n",
+			       xsk.chunk_size, recommendation);
+	}
+
+	if (!test_bit(MLX5E_STATE_OPENED, &priv->state)) {
+		/* XSK objects will be created on open. */
+		goto validate_closed;
+	}
+
+	if (!params->xdp_prog) {
+		/* XSK objects will be created when an XDP program is set,
+		 * and the channels are reopened.
+		 */
+		goto validate_closed;
+	}
+
+	c = priv->channels.c[ix];
+
+	err = mlx5e_open_xsk(priv, params, &xsk, pool, c);
+	if (unlikely(err))
+		goto err_remove_pool;
+
+	mlx5e_activate_xsk(c);
+	mlx5e_trigger_napi_icosq(c);
+
+	/* Don't wait for WQEs, because the newer xdpsock sample doesn't provide
+	 * any Fill Ring entries at the setup stage.
+	 */
+
+	mlx5e_rx_res_xsk_update(priv->rx_res, &priv->channels, ix, true);
+
+	mlx5e_deactivate_rq(&c->rq);
+	mlx5e_flush_rq(&c->rq, MLX5_RQC_STATE_RDY);
+
+	return 0;
+
+err_remove_pool:
+	mlx5e_xsk_remove_pool(&priv->xsk, ix);
+
+err_unmap_pool:
+	mlx5e_xsk_unmap_pool(priv, pool);
+
+	return err;
+
+validate_closed:
+	/* Check the configuration in advance, rather than fail at a later stage
+	 * (in mlx5e_xdp_set or on open) and end up with no channels.
+	 */
+	if (!mlx5e_validate_xsk_param(params, &xsk, priv->mdev)) {
+		err = -EINVAL;
+		goto err_remove_pool;
+	}
+
+	return 0;
+}
+
+static int mlx5e_xsk_disable_locked(struct mlx5e_priv *priv, u16 ix)
+{
+	struct xsk_buff_pool *pool = mlx5e_xsk_get_pool(&priv->channels.params,
+						   &priv->xsk, ix);
+	struct mlx5e_channel *c;
+
+	if (unlikely(!pool))
+		return -EINVAL;
+
+	if (!test_bit(MLX5E_STATE_OPENED, &priv->state))
+		goto remove_pool;
+
+	/* XSK RQ and SQ are only created if XDP program is set. */
+	if (!priv->channels.params.xdp_prog)
+		goto remove_pool;
+
+	c = priv->channels.c[ix];
+
+	mlx5e_activate_rq(&c->rq);
+	mlx5e_trigger_napi_icosq(c);
+	mlx5e_wait_for_min_rx_wqes(&c->rq, MLX5E_RQ_WQES_TIMEOUT);
+
+	mlx5e_rx_res_xsk_update(priv->rx_res, &priv->channels, ix, false);
+
+	mlx5e_deactivate_xsk(c);
+	mlx5e_close_xsk(c);
+
+remove_pool:
+	mlx5e_xsk_remove_pool(&priv->xsk, ix);
+	mlx5e_xsk_unmap_pool(priv, pool);
+
+	return 0;
+}
+
+static int mlx5e_xsk_enable_pool(struct mlx5e_priv *priv, struct xsk_buff_pool *pool,
+				 u16 ix)
+{
+	int err;
+
+	mutex_lock(&priv->state_lock);
+	err = mlx5e_xsk_enable_locked(priv, pool, ix);
+	mutex_unlock(&priv->state_lock);
+
+	return err;
+}
+
+static int mlx5e_xsk_disable_pool(struct mlx5e_priv *priv, u16 ix)
+{
+	int err;
+
+	mutex_lock(&priv->state_lock);
+	err = mlx5e_xsk_disable_locked(priv, ix);
+	mutex_unlock(&priv->state_lock);
+
+	return err;
+}
+
+int mlx5e_xsk_setup_pool(struct net_device *dev, struct xsk_buff_pool *pool, u16 qid)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	struct mlx5e_params *params = &priv->channels.params;
+
+	if (unlikely(qid >= params->num_channels))
+		return -EINVAL;
+
+	return pool ? mlx5e_xsk_enable_pool(priv, pool, qid) :
+		      mlx5e_xsk_disable_pool(priv, qid);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/xsk/pool.h b/drivers/net/ethernet/mellanox/mlx5/core/en/xsk/pool.h
new file mode 100644
index 000000000..dca0010a0
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/xsk/pool.h
@@ -0,0 +1,27 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2019-2020, Mellanox Technologies inc. All rights reserved. */
+
+#ifndef __MLX5_EN_XSK_POOL_H__
+#define __MLX5_EN_XSK_POOL_H__
+
+#include "en.h"
+
+static inline struct xsk_buff_pool *mlx5e_xsk_get_pool(struct mlx5e_params *params,
+						       struct mlx5e_xsk *xsk, u16 ix)
+{
+	if (!xsk || !xsk->pools)
+		return NULL;
+
+	if (unlikely(ix >= params->num_channels))
+		return NULL;
+
+	return xsk->pools[ix];
+}
+
+struct mlx5e_xsk_param;
+void mlx5e_build_xsk_param(struct xsk_buff_pool *pool, struct mlx5e_xsk_param *xsk);
+
+/* .ndo_bpf callback. */
+int mlx5e_xsk_setup_pool(struct net_device *dev, struct xsk_buff_pool *pool, u16 qid);
+
+#endif /* __MLX5_EN_XSK_POOL_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/xsk/rx.c b/drivers/net/ethernet/mellanox/mlx5/core/en/xsk/rx.c
new file mode 100644
index 000000000..c91b54d9f
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/xsk/rx.c
@@ -0,0 +1,311 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2019 Mellanox Technologies. */
+
+#include "rx.h"
+#include "en/xdp.h"
+#include <net/xdp_sock_drv.h>
+#include <linux/filter.h>
+
+/* RX data path */
+
+int mlx5e_xsk_alloc_rx_mpwqe(struct mlx5e_rq *rq, u16 ix)
+{
+	struct mlx5e_mpw_info *wi = mlx5e_get_mpw_info(rq, ix);
+	struct mlx5e_icosq *icosq = rq->icosq;
+	struct mlx5_wq_cyc *wq = &icosq->wq;
+	struct mlx5e_umr_wqe *umr_wqe;
+	int batch, i;
+	u32 offset; /* 17-bit value with MTT. */
+	u16 pi;
+
+	if (unlikely(!xsk_buff_can_alloc(rq->xsk_pool, rq->mpwqe.pages_per_wqe)))
+		goto err;
+
+	BUILD_BUG_ON(sizeof(wi->alloc_units[0]) != sizeof(wi->alloc_units[0].xsk));
+	batch = xsk_buff_alloc_batch(rq->xsk_pool, (struct xdp_buff **)wi->alloc_units,
+				     rq->mpwqe.pages_per_wqe);
+
+	/* If batch < pages_per_wqe, either:
+	 * 1. Some (or all) descriptors were invalid.
+	 * 2. dma_need_sync is true, and it fell back to allocating one frame.
+	 * In either case, try to continue allocating frames one by one, until
+	 * the first error, which will mean there are no more valid descriptors.
+	 */
+	for (; batch < rq->mpwqe.pages_per_wqe; batch++) {
+		wi->alloc_units[batch].xsk = xsk_buff_alloc(rq->xsk_pool);
+		if (unlikely(!wi->alloc_units[batch].xsk))
+			goto err_reuse_batch;
+	}
+
+	pi = mlx5e_icosq_get_next_pi(icosq, rq->mpwqe.umr_wqebbs);
+	umr_wqe = mlx5_wq_cyc_get_wqe(wq, pi);
+	memcpy(umr_wqe, &rq->mpwqe.umr_wqe, sizeof(struct mlx5e_umr_wqe));
+
+	if (likely(rq->mpwqe.umr_mode == MLX5E_MPWRQ_UMR_MODE_ALIGNED)) {
+		for (i = 0; i < batch; i++) {
+			dma_addr_t addr = xsk_buff_xdp_get_frame_dma(wi->alloc_units[i].xsk);
+
+			umr_wqe->inline_mtts[i] = (struct mlx5_mtt) {
+				.ptag = cpu_to_be64(addr | MLX5_EN_WR),
+			};
+		}
+	} else if (unlikely(rq->mpwqe.umr_mode == MLX5E_MPWRQ_UMR_MODE_UNALIGNED)) {
+		for (i = 0; i < batch; i++) {
+			dma_addr_t addr = xsk_buff_xdp_get_frame_dma(wi->alloc_units[i].xsk);
+
+			umr_wqe->inline_ksms[i] = (struct mlx5_ksm) {
+				.key = rq->mkey_be,
+				.va = cpu_to_be64(addr),
+			};
+		}
+	} else if (likely(rq->mpwqe.umr_mode == MLX5E_MPWRQ_UMR_MODE_TRIPLE)) {
+		u32 mapping_size = 1 << (rq->mpwqe.page_shift - 2);
+
+		for (i = 0; i < batch; i++) {
+			dma_addr_t addr = xsk_buff_xdp_get_frame_dma(wi->alloc_units[i].xsk);
+
+			umr_wqe->inline_ksms[i << 2] = (struct mlx5_ksm) {
+				.key = rq->mkey_be,
+				.va = cpu_to_be64(addr),
+			};
+			umr_wqe->inline_ksms[(i << 2) + 1] = (struct mlx5_ksm) {
+				.key = rq->mkey_be,
+				.va = cpu_to_be64(addr + mapping_size),
+			};
+			umr_wqe->inline_ksms[(i << 2) + 2] = (struct mlx5_ksm) {
+				.key = rq->mkey_be,
+				.va = cpu_to_be64(addr + mapping_size * 2),
+			};
+			umr_wqe->inline_ksms[(i << 2) + 3] = (struct mlx5_ksm) {
+				.key = rq->mkey_be,
+				.va = cpu_to_be64(rq->wqe_overflow.addr),
+			};
+		}
+	} else {
+		__be32 pad_size = cpu_to_be32((1 << rq->mpwqe.page_shift) -
+					      rq->xsk_pool->chunk_size);
+		__be32 frame_size = cpu_to_be32(rq->xsk_pool->chunk_size);
+
+		for (i = 0; i < batch; i++) {
+			dma_addr_t addr = xsk_buff_xdp_get_frame_dma(wi->alloc_units[i].xsk);
+
+			umr_wqe->inline_klms[i << 1] = (struct mlx5_klm) {
+				.key = rq->mkey_be,
+				.va = cpu_to_be64(addr),
+				.bcount = frame_size,
+			};
+			umr_wqe->inline_klms[(i << 1) + 1] = (struct mlx5_klm) {
+				.key = rq->mkey_be,
+				.va = cpu_to_be64(rq->wqe_overflow.addr),
+				.bcount = pad_size,
+			};
+		}
+	}
+
+	bitmap_zero(wi->xdp_xmit_bitmap, rq->mpwqe.pages_per_wqe);
+	wi->consumed_strides = 0;
+
+	umr_wqe->ctrl.opmod_idx_opcode =
+		cpu_to_be32((icosq->pc << MLX5_WQE_CTRL_WQE_INDEX_SHIFT) | MLX5_OPCODE_UMR);
+
+	/* Optimized for speed: keep in sync with mlx5e_mpwrq_umr_entry_size. */
+	offset = ix * rq->mpwqe.mtts_per_wqe;
+	if (likely(rq->mpwqe.umr_mode == MLX5E_MPWRQ_UMR_MODE_ALIGNED))
+		offset = offset * sizeof(struct mlx5_mtt) / MLX5_OCTWORD;
+	else if (unlikely(rq->mpwqe.umr_mode == MLX5E_MPWRQ_UMR_MODE_OVERSIZED))
+		offset = offset * sizeof(struct mlx5_klm) * 2 / MLX5_OCTWORD;
+	else if (unlikely(rq->mpwqe.umr_mode == MLX5E_MPWRQ_UMR_MODE_TRIPLE))
+		offset = offset * sizeof(struct mlx5_ksm) * 4 / MLX5_OCTWORD;
+	umr_wqe->uctrl.xlt_offset = cpu_to_be16(offset);
+
+	icosq->db.wqe_info[pi] = (struct mlx5e_icosq_wqe_info) {
+		.wqe_type = MLX5E_ICOSQ_WQE_UMR_RX,
+		.num_wqebbs = rq->mpwqe.umr_wqebbs,
+		.umr.rq = rq,
+	};
+
+	icosq->pc += rq->mpwqe.umr_wqebbs;
+
+	icosq->doorbell_cseg = &umr_wqe->ctrl;
+
+	return 0;
+
+err_reuse_batch:
+	while (--batch >= 0)
+		xsk_buff_free(wi->alloc_units[batch].xsk);
+
+err:
+	rq->stats->buff_alloc_err++;
+	return -ENOMEM;
+}
+
+int mlx5e_xsk_alloc_rx_wqes_batched(struct mlx5e_rq *rq, u16 ix, int wqe_bulk)
+{
+	struct mlx5_wq_cyc *wq = &rq->wqe.wq;
+	struct xdp_buff **buffs;
+	u32 contig, alloc;
+	int i;
+
+	/* mlx5e_init_frags_partition creates a 1:1 mapping between
+	 * rq->wqe.frags and rq->wqe.alloc_units, which allows us to
+	 * allocate XDP buffers straight into alloc_units.
+	 */
+	BUILD_BUG_ON(sizeof(rq->wqe.alloc_units[0]) !=
+		     sizeof(rq->wqe.alloc_units[0].xsk));
+	buffs = (struct xdp_buff **)rq->wqe.alloc_units;
+	contig = mlx5_wq_cyc_get_size(wq) - ix;
+	if (wqe_bulk <= contig) {
+		alloc = xsk_buff_alloc_batch(rq->xsk_pool, buffs + ix, wqe_bulk);
+	} else {
+		alloc = xsk_buff_alloc_batch(rq->xsk_pool, buffs + ix, contig);
+		if (likely(alloc == contig))
+			alloc += xsk_buff_alloc_batch(rq->xsk_pool, buffs, wqe_bulk - contig);
+	}
+
+	for (i = 0; i < alloc; i++) {
+		int j = mlx5_wq_cyc_ctr2ix(wq, ix + i);
+		struct mlx5e_wqe_frag_info *frag;
+		struct mlx5e_rx_wqe_cyc *wqe;
+		dma_addr_t addr;
+
+		wqe = mlx5_wq_cyc_get_wqe(wq, j);
+		/* Assumes log_num_frags == 0. */
+		frag = &rq->wqe.frags[j];
+
+		addr = xsk_buff_xdp_get_frame_dma(frag->au->xsk);
+		wqe->data[0].addr = cpu_to_be64(addr + rq->buff.headroom);
+	}
+
+	return alloc;
+}
+
+int mlx5e_xsk_alloc_rx_wqes(struct mlx5e_rq *rq, u16 ix, int wqe_bulk)
+{
+	struct mlx5_wq_cyc *wq = &rq->wqe.wq;
+	int i;
+
+	for (i = 0; i < wqe_bulk; i++) {
+		int j = mlx5_wq_cyc_ctr2ix(wq, ix + i);
+		struct mlx5e_wqe_frag_info *frag;
+		struct mlx5e_rx_wqe_cyc *wqe;
+		dma_addr_t addr;
+
+		wqe = mlx5_wq_cyc_get_wqe(wq, j);
+		/* Assumes log_num_frags == 0. */
+		frag = &rq->wqe.frags[j];
+
+		frag->au->xsk = xsk_buff_alloc(rq->xsk_pool);
+		if (unlikely(!frag->au->xsk))
+			return i;
+
+		addr = xsk_buff_xdp_get_frame_dma(frag->au->xsk);
+		wqe->data[0].addr = cpu_to_be64(addr + rq->buff.headroom);
+	}
+
+	return wqe_bulk;
+}
+
+static struct sk_buff *mlx5e_xsk_construct_skb(struct mlx5e_rq *rq, struct xdp_buff *xdp)
+{
+	u32 totallen = xdp->data_end - xdp->data_meta;
+	u32 metalen = xdp->data - xdp->data_meta;
+	struct sk_buff *skb;
+
+	skb = napi_alloc_skb(rq->cq.napi, totallen);
+	if (unlikely(!skb)) {
+		rq->stats->buff_alloc_err++;
+		return NULL;
+	}
+
+	skb_put_data(skb, xdp->data_meta, totallen);
+
+	if (metalen) {
+		skb_metadata_set(skb, metalen);
+		__skb_pull(skb, metalen);
+	}
+
+	return skb;
+}
+
+struct sk_buff *mlx5e_xsk_skb_from_cqe_mpwrq_linear(struct mlx5e_rq *rq,
+						    struct mlx5e_mpw_info *wi,
+						    u16 cqe_bcnt,
+						    u32 head_offset,
+						    u32 page_idx)
+{
+	struct xdp_buff *xdp = wi->alloc_units[page_idx].xsk;
+	struct bpf_prog *prog;
+
+	/* Check packet size. Note LRO doesn't use linear SKB */
+	if (unlikely(cqe_bcnt > rq->hw_mtu)) {
+		rq->stats->oversize_pkts_sw_drop++;
+		return NULL;
+	}
+
+	/* head_offset is not used in this function, because xdp->data and the
+	 * DMA address point directly to the necessary place. Furthermore, in
+	 * the current implementation, UMR pages are mapped to XSK frames, so
+	 * head_offset should always be 0.
+	 */
+	WARN_ON_ONCE(head_offset);
+
+	xsk_buff_set_size(xdp, cqe_bcnt);
+	xsk_buff_dma_sync_for_cpu(xdp, rq->xsk_pool);
+	net_prefetch(xdp->data);
+
+	/* Possible flows:
+	 * - XDP_REDIRECT to XSKMAP:
+	 *   The page is owned by the userspace from now.
+	 * - XDP_TX and other XDP_REDIRECTs:
+	 *   The page was returned by ZCA and recycled.
+	 * - XDP_DROP:
+	 *   Recycle the page.
+	 * - XDP_PASS:
+	 *   Allocate an SKB, copy the data and recycle the page.
+	 *
+	 * Pages to be recycled go to the Reuse Ring on MPWQE deallocation. Its
+	 * size is the same as the Driver RX Ring's size, and pages for WQEs are
+	 * allocated first from the Reuse Ring, so it has enough space.
+	 */
+
+	prog = rcu_dereference(rq->xdp_prog);
+	if (likely(prog && mlx5e_xdp_handle(rq, NULL, prog, xdp))) {
+		if (likely(__test_and_clear_bit(MLX5E_RQ_FLAG_XDP_XMIT, rq->flags)))
+			__set_bit(page_idx, wi->xdp_xmit_bitmap); /* non-atomic */
+		return NULL; /* page/packet was consumed by XDP */
+	}
+
+	/* XDP_PASS: copy the data from the UMEM to a new SKB and reuse the
+	 * frame. On SKB allocation failure, NULL is returned.
+	 */
+	return mlx5e_xsk_construct_skb(rq, xdp);
+}
+
+struct sk_buff *mlx5e_xsk_skb_from_cqe_linear(struct mlx5e_rq *rq,
+					      struct mlx5e_wqe_frag_info *wi,
+					      u32 cqe_bcnt)
+{
+	struct xdp_buff *xdp = wi->au->xsk;
+	struct bpf_prog *prog;
+
+	/* wi->offset is not used in this function, because xdp->data and the
+	 * DMA address point directly to the necessary place. Furthermore, the
+	 * XSK allocator allocates frames per packet, instead of pages, so
+	 * wi->offset should always be 0.
+	 */
+	WARN_ON_ONCE(wi->offset);
+
+	xsk_buff_set_size(xdp, cqe_bcnt);
+	xsk_buff_dma_sync_for_cpu(xdp, rq->xsk_pool);
+	net_prefetch(xdp->data);
+
+	prog = rcu_dereference(rq->xdp_prog);
+	if (likely(prog && mlx5e_xdp_handle(rq, NULL, prog, xdp)))
+		return NULL; /* page/packet was consumed by XDP */
+
+	/* XDP_PASS: copy the data from the UMEM to a new SKB. The frame reuse
+	 * will be handled by mlx5e_free_rx_wqe.
+	 * On SKB allocation failure, NULL is returned.
+	 */
+	return mlx5e_xsk_construct_skb(rq, xdp);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/xsk/rx.h b/drivers/net/ethernet/mellanox/mlx5/core/en/xsk/rx.h
new file mode 100644
index 000000000..087c943bd
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/xsk/rx.h
@@ -0,0 +1,23 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2019 Mellanox Technologies. */
+
+#ifndef __MLX5_EN_XSK_RX_H__
+#define __MLX5_EN_XSK_RX_H__
+
+#include "en.h"
+
+/* RX data path */
+
+int mlx5e_xsk_alloc_rx_mpwqe(struct mlx5e_rq *rq, u16 ix);
+int mlx5e_xsk_alloc_rx_wqes_batched(struct mlx5e_rq *rq, u16 ix, int wqe_bulk);
+int mlx5e_xsk_alloc_rx_wqes(struct mlx5e_rq *rq, u16 ix, int wqe_bulk);
+struct sk_buff *mlx5e_xsk_skb_from_cqe_mpwrq_linear(struct mlx5e_rq *rq,
+						    struct mlx5e_mpw_info *wi,
+						    u16 cqe_bcnt,
+						    u32 head_offset,
+						    u32 page_idx);
+struct sk_buff *mlx5e_xsk_skb_from_cqe_linear(struct mlx5e_rq *rq,
+					      struct mlx5e_wqe_frag_info *wi,
+					      u32 cqe_bcnt);
+
+#endif /* __MLX5_EN_XSK_RX_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/xsk/setup.c b/drivers/net/ethernet/mellanox/mlx5/core/en/xsk/setup.c
new file mode 100644
index 000000000..ff03c4383
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/xsk/setup.c
@@ -0,0 +1,191 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2019 Mellanox Technologies. */
+
+#include "setup.h"
+#include "en/params.h"
+#include "en/txrx.h"
+#include "en/health.h"
+#include <net/xdp_sock_drv.h>
+
+/* The limitation of 2048 can be altered, but shouldn't go beyond the minimal
+ * stride size of striding RQ.
+ */
+#define MLX5E_MIN_XSK_CHUNK_SIZE max(2048, XDP_UMEM_MIN_CHUNK_SIZE)
+
+bool mlx5e_validate_xsk_param(struct mlx5e_params *params,
+			      struct mlx5e_xsk_param *xsk,
+			      struct mlx5_core_dev *mdev)
+{
+	/* AF_XDP doesn't support frames larger than PAGE_SIZE. */
+	if (xsk->chunk_size > PAGE_SIZE || xsk->chunk_size < MLX5E_MIN_XSK_CHUNK_SIZE)
+		return false;
+
+	/* frag_sz is different for regular and XSK RQs, so ensure that linear
+	 * SKB mode is possible.
+	 */
+	switch (params->rq_wq_type) {
+	case MLX5_WQ_TYPE_LINKED_LIST_STRIDING_RQ:
+		return !mlx5e_mpwrq_validate_xsk(mdev, params, xsk);
+	default: /* MLX5_WQ_TYPE_CYCLIC */
+		return mlx5e_rx_is_linear_skb(mdev, params, xsk);
+	}
+}
+
+static void mlx5e_build_xsk_cparam(struct mlx5_core_dev *mdev,
+				   struct mlx5e_params *params,
+				   struct mlx5e_xsk_param *xsk,
+				   u16 q_counter,
+				   struct mlx5e_channel_param *cparam)
+{
+	mlx5e_build_rq_param(mdev, params, xsk, q_counter, &cparam->rq);
+	mlx5e_build_xdpsq_param(mdev, params, xsk, &cparam->xdp_sq);
+}
+
+static int mlx5e_init_xsk_rq(struct mlx5e_channel *c,
+			     struct mlx5e_params *params,
+			     struct xsk_buff_pool *pool,
+			     struct mlx5e_xsk_param *xsk,
+			     struct mlx5e_rq *rq)
+{
+	struct mlx5_core_dev *mdev = c->mdev;
+	int rq_xdp_ix;
+	int err;
+
+	rq->wq_type      = params->rq_wq_type;
+	rq->pdev         = c->pdev;
+	rq->netdev       = c->netdev;
+	rq->priv         = c->priv;
+	rq->tstamp       = c->tstamp;
+	rq->clock        = &mdev->clock;
+	rq->icosq        = &c->icosq;
+	rq->ix           = c->ix;
+	rq->channel      = c;
+	rq->mdev         = mdev;
+	rq->hw_mtu       = MLX5E_SW2HW_MTU(params, params->sw_mtu);
+	rq->xdpsq        = &c->rq_xdpsq;
+	rq->xsk_pool     = pool;
+	rq->stats        = &c->priv->channel_stats[c->ix]->xskrq;
+	rq->ptp_cyc2time = mlx5_rq_ts_translator(mdev);
+	rq_xdp_ix        = c->ix;
+	err = mlx5e_rq_set_handlers(rq, params, xsk);
+	if (err)
+		return err;
+
+	return  xdp_rxq_info_reg(&rq->xdp_rxq, rq->netdev, rq_xdp_ix, 0);
+}
+
+static int mlx5e_open_xsk_rq(struct mlx5e_channel *c, struct mlx5e_params *params,
+			     struct mlx5e_rq_param *rq_params, struct xsk_buff_pool *pool,
+			     struct mlx5e_xsk_param *xsk)
+{
+	int err;
+
+	err = mlx5e_init_xsk_rq(c, params, pool, xsk, &c->xskrq);
+	if (err)
+		return err;
+
+	return mlx5e_open_rq(params, rq_params, xsk, cpu_to_node(c->cpu), &c->xskrq);
+}
+
+int mlx5e_open_xsk(struct mlx5e_priv *priv, struct mlx5e_params *params,
+		   struct mlx5e_xsk_param *xsk, struct xsk_buff_pool *pool,
+		   struct mlx5e_channel *c)
+{
+	struct mlx5e_channel_param *cparam;
+	struct mlx5e_create_cq_param ccp;
+	int err;
+
+	mlx5e_build_create_cq_param(&ccp, c);
+
+	if (!mlx5e_validate_xsk_param(params, xsk, priv->mdev))
+		return -EINVAL;
+
+	cparam = kvzalloc(sizeof(*cparam), GFP_KERNEL);
+	if (!cparam)
+		return -ENOMEM;
+
+	mlx5e_build_xsk_cparam(priv->mdev, params, xsk, priv->q_counter, cparam);
+
+	err = mlx5e_open_cq(c->priv, params->rx_cq_moderation, &cparam->rq.cqp, &ccp,
+			    &c->xskrq.cq);
+	if (unlikely(err))
+		goto err_free_cparam;
+
+	err = mlx5e_open_xsk_rq(c, params, &cparam->rq, pool, xsk);
+	if (unlikely(err))
+		goto err_close_rx_cq;
+
+	err = mlx5e_open_cq(c->priv, params->tx_cq_moderation, &cparam->xdp_sq.cqp, &ccp,
+			    &c->xsksq.cq);
+	if (unlikely(err))
+		goto err_close_rq;
+
+	/* Create a separate SQ, so that when the buff pool is disabled, we could
+	 * close this SQ safely and stop receiving CQEs. In other case, e.g., if
+	 * the XDPSQ was used instead, we might run into trouble when the buff pool
+	 * is disabled and then re-enabled, but the SQ continues receiving CQEs
+	 * from the old buff pool.
+	 */
+	err = mlx5e_open_xdpsq(c, params, &cparam->xdp_sq, pool, &c->xsksq, true);
+	if (unlikely(err))
+		goto err_close_tx_cq;
+
+	kvfree(cparam);
+
+	set_bit(MLX5E_CHANNEL_STATE_XSK, c->state);
+
+	return 0;
+
+err_close_tx_cq:
+	mlx5e_close_cq(&c->xsksq.cq);
+
+err_close_rq:
+	mlx5e_close_rq(&c->xskrq);
+
+err_close_rx_cq:
+	mlx5e_close_cq(&c->xskrq.cq);
+
+err_free_cparam:
+	kvfree(cparam);
+
+	return err;
+}
+
+void mlx5e_close_xsk(struct mlx5e_channel *c)
+{
+	clear_bit(MLX5E_CHANNEL_STATE_XSK, c->state);
+	synchronize_net(); /* Sync with NAPI. */
+
+	mlx5e_close_rq(&c->xskrq);
+	mlx5e_close_cq(&c->xskrq.cq);
+	mlx5e_close_xdpsq(&c->xsksq);
+	mlx5e_close_cq(&c->xsksq.cq);
+
+	memset(&c->xskrq, 0, sizeof(c->xskrq));
+	memset(&c->xsksq, 0, sizeof(c->xsksq));
+}
+
+void mlx5e_activate_xsk(struct mlx5e_channel *c)
+{
+	/* ICOSQ recovery deactivates RQs. Suspend the recovery to avoid
+	 * activating XSKRQ in the middle of recovery.
+	 */
+	mlx5e_reporter_icosq_suspend_recovery(c);
+	set_bit(MLX5E_RQ_STATE_ENABLED, &c->xskrq.state);
+	mlx5e_reporter_icosq_resume_recovery(c);
+
+	/* TX queue is created active. */
+}
+
+void mlx5e_deactivate_xsk(struct mlx5e_channel *c)
+{
+	/* ICOSQ recovery may reactivate XSKRQ if clear_bit is called in the
+	 * middle of recovery. Suspend the recovery to avoid it.
+	 */
+	mlx5e_reporter_icosq_suspend_recovery(c);
+	clear_bit(MLX5E_RQ_STATE_ENABLED, &c->xskrq.state);
+	mlx5e_reporter_icosq_resume_recovery(c);
+	synchronize_net(); /* Sync with NAPI to prevent mlx5e_post_rx_wqes. */
+
+	/* TX queue is disabled on close. */
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/xsk/setup.h b/drivers/net/ethernet/mellanox/mlx5/core/en/xsk/setup.h
new file mode 100644
index 000000000..50e111b85
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/xsk/setup.h
@@ -0,0 +1,21 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2019 Mellanox Technologies. */
+
+#ifndef __MLX5_EN_XSK_SETUP_H__
+#define __MLX5_EN_XSK_SETUP_H__
+
+#include "en.h"
+
+struct mlx5e_xsk_param;
+
+bool mlx5e_validate_xsk_param(struct mlx5e_params *params,
+			      struct mlx5e_xsk_param *xsk,
+			      struct mlx5_core_dev *mdev);
+int mlx5e_open_xsk(struct mlx5e_priv *priv, struct mlx5e_params *params,
+		   struct mlx5e_xsk_param *xsk, struct xsk_buff_pool *pool,
+		   struct mlx5e_channel *c);
+void mlx5e_close_xsk(struct mlx5e_channel *c);
+void mlx5e_activate_xsk(struct mlx5e_channel *c);
+void mlx5e_deactivate_xsk(struct mlx5e_channel *c);
+
+#endif /* __MLX5_EN_XSK_SETUP_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/xsk/tx.c b/drivers/net/ethernet/mellanox/mlx5/core/en/xsk/tx.c
new file mode 100644
index 000000000..367a9505c
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/xsk/tx.c
@@ -0,0 +1,123 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2019 Mellanox Technologies. */
+
+#include "tx.h"
+#include "pool.h"
+#include "en/xdp.h"
+#include "en/params.h"
+#include <net/xdp_sock_drv.h>
+
+int mlx5e_xsk_wakeup(struct net_device *dev, u32 qid, u32 flags)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	struct mlx5e_params *params = &priv->channels.params;
+	struct mlx5e_channel *c;
+
+	if (unlikely(!mlx5e_xdp_is_active(priv)))
+		return -ENETDOWN;
+
+	if (unlikely(qid >= params->num_channels))
+		return -EINVAL;
+
+	c = priv->channels.c[qid];
+
+	if (!napi_if_scheduled_mark_missed(&c->napi)) {
+		/* To avoid WQE overrun, don't post a NOP if async_icosq is not
+		 * active and not polled by NAPI. Return 0, because the upcoming
+		 * activate will trigger the IRQ for us.
+		 */
+		if (unlikely(!test_bit(MLX5E_SQ_STATE_ENABLED, &c->async_icosq.state)))
+			return 0;
+
+		if (test_and_set_bit(MLX5E_SQ_STATE_PENDING_XSK_TX, &c->async_icosq.state))
+			return 0;
+
+		mlx5e_trigger_napi_icosq(c);
+	}
+
+	return 0;
+}
+
+/* When TX fails (because of the size of the packet), we need to get completions
+ * in order, so post a NOP to get a CQE. Since AF_XDP doesn't distinguish
+ * between successful TX and errors, handling in mlx5e_poll_xdpsq_cq is the
+ * same.
+ */
+static void mlx5e_xsk_tx_post_err(struct mlx5e_xdpsq *sq,
+				  struct mlx5e_xdp_info *xdpi)
+{
+	u16 pi = mlx5_wq_cyc_ctr2ix(&sq->wq, sq->pc);
+	struct mlx5e_xdp_wqe_info *wi = &sq->db.wqe_info[pi];
+	struct mlx5e_tx_wqe *nopwqe;
+
+	wi->num_wqebbs = 1;
+	wi->num_pkts = 1;
+
+	nopwqe = mlx5e_post_nop(&sq->wq, sq->sqn, &sq->pc);
+	mlx5e_xdpi_fifo_push(&sq->db.xdpi_fifo, xdpi);
+	sq->doorbell_cseg = &nopwqe->ctrl;
+}
+
+bool mlx5e_xsk_tx(struct mlx5e_xdpsq *sq, unsigned int budget)
+{
+	struct xsk_buff_pool *pool = sq->xsk_pool;
+	struct mlx5e_xmit_data xdptxd;
+	struct mlx5e_xdp_info xdpi;
+	bool work_done = true;
+	bool flush = false;
+
+	xdpi.mode = MLX5E_XDP_XMIT_MODE_XSK;
+
+	for (; budget; budget--) {
+		int check_result = INDIRECT_CALL_2(sq->xmit_xdp_frame_check,
+						   mlx5e_xmit_xdp_frame_check_mpwqe,
+						   mlx5e_xmit_xdp_frame_check,
+						   sq);
+		struct xdp_desc desc;
+		bool ret;
+
+		if (unlikely(check_result < 0)) {
+			work_done = false;
+			break;
+		}
+
+		if (!xsk_tx_peek_desc(pool, &desc)) {
+			/* TX will get stuck until something wakes it up by
+			 * triggering NAPI. Currently it's expected that the
+			 * application calls sendto() if there are consumed, but
+			 * not completed frames.
+			 */
+			break;
+		}
+
+		xdptxd.dma_addr = xsk_buff_raw_get_dma(pool, desc.addr);
+		xdptxd.data = xsk_buff_raw_get_data(pool, desc.addr);
+		xdptxd.len = desc.len;
+
+		xsk_buff_raw_dma_sync_for_device(pool, xdptxd.dma_addr, xdptxd.len);
+
+		ret = INDIRECT_CALL_2(sq->xmit_xdp_frame, mlx5e_xmit_xdp_frame_mpwqe,
+				      mlx5e_xmit_xdp_frame, sq, &xdptxd, NULL,
+				      check_result);
+		if (unlikely(!ret)) {
+			if (sq->mpwqe.wqe)
+				mlx5e_xdp_mpwqe_complete(sq);
+
+			mlx5e_xsk_tx_post_err(sq, &xdpi);
+		} else {
+			mlx5e_xdpi_fifo_push(&sq->db.xdpi_fifo, &xdpi);
+		}
+
+		flush = true;
+	}
+
+	if (flush) {
+		if (sq->mpwqe.wqe)
+			mlx5e_xdp_mpwqe_complete(sq);
+		mlx5e_xmit_xdp_doorbell(sq);
+
+		xsk_tx_release(pool);
+	}
+
+	return !(budget && work_done);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/xsk/tx.h b/drivers/net/ethernet/mellanox/mlx5/core/en/xsk/tx.h
new file mode 100644
index 000000000..9c505158b
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/xsk/tx.h
@@ -0,0 +1,15 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2019 Mellanox Technologies. */
+
+#ifndef __MLX5_EN_XSK_TX_H__
+#define __MLX5_EN_XSK_TX_H__
+
+#include "en.h"
+
+/* TX data path */
+
+int mlx5e_xsk_wakeup(struct net_device *dev, u32 qid, u32 flags);
+
+bool mlx5e_xsk_tx(struct mlx5e_xdpsq *sq, unsigned int budget);
+
+#endif /* __MLX5_EN_XSK_TX_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/en_accel.h b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/en_accel.h
new file mode 100644
index 000000000..07187028f
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/en_accel.h
@@ -0,0 +1,222 @@
+/*
+ * Copyright (c) 2018 Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+
+#ifndef __MLX5E_EN_ACCEL_H__
+#define __MLX5E_EN_ACCEL_H__
+
+#include <linux/skbuff.h>
+#include <linux/netdevice.h>
+#include "en_accel/ipsec_rxtx.h"
+#include "en_accel/ktls.h"
+#include "en_accel/ktls_txrx.h"
+#include <en_accel/macsec.h>
+#include "en.h"
+#include "en/txrx.h"
+
+#if IS_ENABLED(CONFIG_GENEVE)
+#include <net/geneve.h>
+
+static inline bool mlx5_geneve_tx_allowed(struct mlx5_core_dev *mdev)
+{
+	return mlx5_tx_swp_supported(mdev);
+}
+
+static inline void
+mlx5e_tx_tunnel_accel(struct sk_buff *skb, struct mlx5_wqe_eth_seg *eseg, u16 ihs)
+{
+	struct mlx5e_swp_spec swp_spec = {};
+	unsigned int offset = 0;
+	__be16 l3_proto;
+	u8 l4_proto;
+
+	l3_proto = vlan_get_protocol(skb);
+	switch (l3_proto) {
+	case htons(ETH_P_IP):
+		l4_proto = ip_hdr(skb)->protocol;
+		break;
+	case htons(ETH_P_IPV6):
+		l4_proto = ipv6_find_hdr(skb, &offset, -1, NULL, NULL);
+		break;
+	default:
+		return;
+	}
+
+	if (l4_proto != IPPROTO_UDP ||
+	    udp_hdr(skb)->dest != cpu_to_be16(GENEVE_UDP_PORT))
+		return;
+	swp_spec.l3_proto = l3_proto;
+	swp_spec.l4_proto = l4_proto;
+	swp_spec.is_tun = true;
+	if (inner_ip_hdr(skb)->version == 6) {
+		swp_spec.tun_l3_proto = htons(ETH_P_IPV6);
+		swp_spec.tun_l4_proto = inner_ipv6_hdr(skb)->nexthdr;
+	} else {
+		swp_spec.tun_l3_proto = htons(ETH_P_IP);
+		swp_spec.tun_l4_proto = inner_ip_hdr(skb)->protocol;
+	}
+
+	mlx5e_set_eseg_swp(skb, eseg, &swp_spec);
+	if (skb_vlan_tag_present(skb) && ihs)
+		mlx5e_eseg_swp_offsets_add_vlan(eseg);
+}
+
+#else
+static inline bool mlx5_geneve_tx_allowed(struct mlx5_core_dev *mdev)
+{
+	return false;
+}
+
+#endif /* CONFIG_GENEVE */
+
+static inline void
+mlx5e_udp_gso_handle_tx_skb(struct sk_buff *skb)
+{
+	int payload_len = skb_shinfo(skb)->gso_size + sizeof(struct udphdr);
+
+	udp_hdr(skb)->len = htons(payload_len);
+}
+
+struct mlx5e_accel_tx_state {
+#ifdef CONFIG_MLX5_EN_TLS
+	struct mlx5e_accel_tx_tls_state tls;
+#endif
+#ifdef CONFIG_MLX5_EN_IPSEC
+	struct mlx5e_accel_tx_ipsec_state ipsec;
+#endif
+};
+
+static inline bool mlx5e_accel_tx_begin(struct net_device *dev,
+					struct mlx5e_txqsq *sq,
+					struct sk_buff *skb,
+					struct mlx5e_accel_tx_state *state)
+{
+	if (skb_is_gso(skb) && skb_shinfo(skb)->gso_type & SKB_GSO_UDP_L4)
+		mlx5e_udp_gso_handle_tx_skb(skb);
+
+#ifdef CONFIG_MLX5_EN_TLS
+	/* May send SKBs and WQEs. */
+	if (mlx5e_ktls_skb_offloaded(skb))
+		if (unlikely(!mlx5e_ktls_handle_tx_skb(dev, sq, skb,
+						       &state->tls)))
+			return false;
+#endif
+
+#ifdef CONFIG_MLX5_EN_IPSEC
+	if (test_bit(MLX5E_SQ_STATE_IPSEC, &sq->state) && xfrm_offload(skb)) {
+		if (unlikely(!mlx5e_ipsec_handle_tx_skb(dev, skb, &state->ipsec)))
+			return false;
+	}
+#endif
+
+#ifdef CONFIG_MLX5_EN_MACSEC
+	if (unlikely(mlx5e_macsec_skb_is_offload(skb))) {
+		struct mlx5e_priv *priv = netdev_priv(dev);
+
+		if (unlikely(!mlx5e_macsec_handle_tx_skb(priv->macsec, skb)))
+			return false;
+	}
+#endif
+
+	return true;
+}
+
+static inline unsigned int mlx5e_accel_tx_ids_len(struct mlx5e_txqsq *sq,
+						  struct mlx5e_accel_tx_state *state)
+{
+#ifdef CONFIG_MLX5_EN_IPSEC
+	if (test_bit(MLX5E_SQ_STATE_IPSEC, &sq->state))
+		return mlx5e_ipsec_tx_ids_len(&state->ipsec);
+#endif
+
+	return 0;
+}
+
+/* Part of the eseg touched by TX offloads */
+#define MLX5E_ACCEL_ESEG_LEN offsetof(struct mlx5_wqe_eth_seg, mss)
+
+static inline void mlx5e_accel_tx_eseg(struct mlx5e_priv *priv,
+				       struct sk_buff *skb,
+				       struct mlx5_wqe_eth_seg *eseg, u16 ihs)
+{
+#ifdef CONFIG_MLX5_EN_IPSEC
+	if (xfrm_offload(skb))
+		mlx5e_ipsec_tx_build_eseg(priv, skb, eseg);
+#endif
+
+#ifdef CONFIG_MLX5_EN_MACSEC
+	if (unlikely(mlx5e_macsec_skb_is_offload(skb)))
+		mlx5e_macsec_tx_build_eseg(priv->macsec, skb, eseg);
+#endif
+
+#if IS_ENABLED(CONFIG_GENEVE)
+	if (skb->encapsulation && skb->ip_summed == CHECKSUM_PARTIAL)
+		mlx5e_tx_tunnel_accel(skb, eseg, ihs);
+#endif
+}
+
+static inline void mlx5e_accel_tx_finish(struct mlx5e_txqsq *sq,
+					 struct mlx5e_tx_wqe *wqe,
+					 struct mlx5e_accel_tx_state *state,
+					 struct mlx5_wqe_inline_seg *inlseg)
+{
+#ifdef CONFIG_MLX5_EN_TLS
+	mlx5e_ktls_handle_tx_wqe(&wqe->ctrl, &state->tls);
+#endif
+
+#ifdef CONFIG_MLX5_EN_IPSEC
+	if (test_bit(MLX5E_SQ_STATE_IPSEC, &sq->state) &&
+	    state->ipsec.xo && state->ipsec.tailen)
+		mlx5e_ipsec_handle_tx_wqe(wqe, &state->ipsec, inlseg);
+#endif
+}
+
+static inline int mlx5e_accel_init_rx(struct mlx5e_priv *priv)
+{
+	return mlx5e_ktls_init_rx(priv);
+}
+
+static inline void mlx5e_accel_cleanup_rx(struct mlx5e_priv *priv)
+{
+	mlx5e_ktls_cleanup_rx(priv);
+}
+
+static inline int mlx5e_accel_init_tx(struct mlx5e_priv *priv)
+{
+	return mlx5e_ktls_init_tx(priv);
+}
+
+static inline void mlx5e_accel_cleanup_tx(struct mlx5e_priv *priv)
+{
+	mlx5e_ktls_cleanup_tx(priv);
+}
+#endif /* __MLX5E_EN_ACCEL_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/fs_tcp.c b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/fs_tcp.c
new file mode 100644
index 000000000..06c474049
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/fs_tcp.c
@@ -0,0 +1,404 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2020, Mellanox Technologies inc. All rights reserved. */
+
+#include <mlx5_core.h>
+#include "en_accel/fs_tcp.h"
+#include "fs_core.h"
+
+enum accel_fs_tcp_type {
+	ACCEL_FS_IPV4_TCP,
+	ACCEL_FS_IPV6_TCP,
+	ACCEL_FS_TCP_NUM_TYPES,
+};
+
+struct mlx5e_accel_fs_tcp {
+	struct mlx5e_flow_table tables[ACCEL_FS_TCP_NUM_TYPES];
+	struct mlx5_flow_handle *default_rules[ACCEL_FS_TCP_NUM_TYPES];
+};
+
+static enum mlx5_traffic_types fs_accel2tt(enum accel_fs_tcp_type i)
+{
+	switch (i) {
+	case ACCEL_FS_IPV4_TCP:
+		return MLX5_TT_IPV4_TCP;
+	default: /* ACCEL_FS_IPV6_TCP */
+		return MLX5_TT_IPV6_TCP;
+	}
+}
+
+static void accel_fs_tcp_set_ipv4_flow(struct mlx5_flow_spec *spec, struct sock *sk)
+{
+	MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria, outer_headers.ip_protocol);
+	MLX5_SET(fte_match_param, spec->match_value, outer_headers.ip_protocol, IPPROTO_TCP);
+	MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria, outer_headers.ip_version);
+	MLX5_SET(fte_match_param, spec->match_value, outer_headers.ip_version, 4);
+	memcpy(MLX5_ADDR_OF(fte_match_param, spec->match_value,
+			    outer_headers.src_ipv4_src_ipv6.ipv4_layout.ipv4),
+	       &inet_sk(sk)->inet_daddr, 4);
+	memcpy(MLX5_ADDR_OF(fte_match_param, spec->match_value,
+			    outer_headers.dst_ipv4_dst_ipv6.ipv4_layout.ipv4),
+	       &inet_sk(sk)->inet_rcv_saddr, 4);
+	MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria,
+			 outer_headers.src_ipv4_src_ipv6.ipv4_layout.ipv4);
+	MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria,
+			 outer_headers.dst_ipv4_dst_ipv6.ipv4_layout.ipv4);
+}
+
+#if IS_ENABLED(CONFIG_IPV6)
+static void accel_fs_tcp_set_ipv6_flow(struct mlx5_flow_spec *spec, struct sock *sk)
+{
+	MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria, outer_headers.ip_protocol);
+	MLX5_SET(fte_match_param, spec->match_value, outer_headers.ip_protocol, IPPROTO_TCP);
+	MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria, outer_headers.ip_version);
+	MLX5_SET(fte_match_param, spec->match_value, outer_headers.ip_version, 6);
+	memcpy(MLX5_ADDR_OF(fte_match_param, spec->match_value,
+			    outer_headers.src_ipv4_src_ipv6.ipv6_layout.ipv6),
+	       &sk->sk_v6_daddr, 16);
+	memcpy(MLX5_ADDR_OF(fte_match_param, spec->match_value,
+			    outer_headers.dst_ipv4_dst_ipv6.ipv6_layout.ipv6),
+	       &inet6_sk(sk)->saddr, 16);
+	memset(MLX5_ADDR_OF(fte_match_param, spec->match_criteria,
+			    outer_headers.src_ipv4_src_ipv6.ipv6_layout.ipv6),
+	       0xff, 16);
+	memset(MLX5_ADDR_OF(fte_match_param, spec->match_criteria,
+			    outer_headers.dst_ipv4_dst_ipv6.ipv6_layout.ipv6),
+	       0xff, 16);
+}
+#endif
+
+void mlx5e_accel_fs_del_sk(struct mlx5_flow_handle *rule)
+{
+	mlx5_del_flow_rules(rule);
+}
+
+struct mlx5_flow_handle *mlx5e_accel_fs_add_sk(struct mlx5e_flow_steering *fs,
+					       struct sock *sk, u32 tirn,
+					       uint32_t flow_tag)
+{
+	struct mlx5e_accel_fs_tcp *fs_tcp = mlx5e_fs_get_accel_tcp(fs);
+	struct mlx5_flow_destination dest = {};
+	struct mlx5e_flow_table *ft = NULL;
+	MLX5_DECLARE_FLOW_ACT(flow_act);
+	struct mlx5_flow_handle *flow;
+	struct mlx5_flow_spec *spec;
+
+	spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
+	if (!spec)
+		return ERR_PTR(-ENOMEM);
+
+	spec->match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
+
+	switch (sk->sk_family) {
+	case AF_INET:
+		accel_fs_tcp_set_ipv4_flow(spec, sk);
+		ft = &fs_tcp->tables[ACCEL_FS_IPV4_TCP];
+		fs_dbg(fs, "%s flow is %pI4:%d -> %pI4:%d\n", __func__,
+		       &inet_sk(sk)->inet_rcv_saddr,
+		       inet_sk(sk)->inet_sport,
+		       &inet_sk(sk)->inet_daddr,
+		       inet_sk(sk)->inet_dport);
+		break;
+#if IS_ENABLED(CONFIG_IPV6)
+	case AF_INET6:
+		if (!ipv6_only_sock(sk) &&
+		    ipv6_addr_type(&sk->sk_v6_daddr) == IPV6_ADDR_MAPPED) {
+			accel_fs_tcp_set_ipv4_flow(spec, sk);
+			ft = &fs_tcp->tables[ACCEL_FS_IPV4_TCP];
+		} else {
+			accel_fs_tcp_set_ipv6_flow(spec, sk);
+			ft = &fs_tcp->tables[ACCEL_FS_IPV6_TCP];
+		}
+		break;
+#endif
+	default:
+		break;
+	}
+
+	if (!ft) {
+		flow = ERR_PTR(-EINVAL);
+		goto out;
+	}
+
+	MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria,
+			 outer_headers.tcp_dport);
+	MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria,
+			 outer_headers.tcp_sport);
+	MLX5_SET(fte_match_param, spec->match_value, outer_headers.tcp_dport,
+		 ntohs(inet_sk(sk)->inet_sport));
+	MLX5_SET(fte_match_param, spec->match_value, outer_headers.tcp_sport,
+		 ntohs(inet_sk(sk)->inet_dport));
+
+	dest.type = MLX5_FLOW_DESTINATION_TYPE_TIR;
+	dest.tir_num = tirn;
+	if (flow_tag != MLX5_FS_DEFAULT_FLOW_TAG) {
+		spec->flow_context.flow_tag = flow_tag;
+		spec->flow_context.flags = FLOW_CONTEXT_HAS_TAG;
+	}
+
+	flow = mlx5_add_flow_rules(ft->t, spec, &flow_act, &dest, 1);
+
+	if (IS_ERR(flow))
+		fs_err(fs, "mlx5_add_flow_rules() failed, flow is %ld\n", PTR_ERR(flow));
+
+out:
+	kvfree(spec);
+	return flow;
+}
+
+static int accel_fs_tcp_add_default_rule(struct mlx5e_flow_steering *fs,
+					 enum accel_fs_tcp_type type)
+{
+	struct mlx5e_accel_fs_tcp *fs_tcp = mlx5e_fs_get_accel_tcp(fs);
+	struct mlx5_ttc_table *ttc = mlx5e_fs_get_ttc(fs, false);
+	struct mlx5e_flow_table *accel_fs_t;
+	struct mlx5_flow_destination dest;
+	MLX5_DECLARE_FLOW_ACT(flow_act);
+	struct mlx5_flow_handle *rule;
+	int err = 0;
+
+	accel_fs_t = &fs_tcp->tables[type];
+
+	dest = mlx5_ttc_get_default_dest(ttc, fs_accel2tt(type));
+	rule = mlx5_add_flow_rules(accel_fs_t->t, NULL, &flow_act, &dest, 1);
+	if (IS_ERR(rule)) {
+		err = PTR_ERR(rule);
+		fs_err(fs, "%s: add default rule failed, accel_fs type=%d, err %d\n",
+		       __func__, type, err);
+		return err;
+	}
+
+	fs_tcp->default_rules[type] = rule;
+	return 0;
+}
+
+#define MLX5E_ACCEL_FS_TCP_NUM_GROUPS	(2)
+#define MLX5E_ACCEL_FS_TCP_GROUP1_SIZE	(BIT(16) - 1)
+#define MLX5E_ACCEL_FS_TCP_GROUP2_SIZE	(BIT(0))
+#define MLX5E_ACCEL_FS_TCP_TABLE_SIZE	(MLX5E_ACCEL_FS_TCP_GROUP1_SIZE +\
+					 MLX5E_ACCEL_FS_TCP_GROUP2_SIZE)
+static int accel_fs_tcp_create_groups(struct mlx5e_flow_table *ft,
+				      enum accel_fs_tcp_type type)
+{
+	int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
+	void *outer_headers_c;
+	int ix = 0;
+	u32 *in;
+	int err;
+	u8 *mc;
+
+	ft->g = kcalloc(MLX5E_ACCEL_FS_TCP_NUM_GROUPS, sizeof(*ft->g), GFP_KERNEL);
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if  (!in || !ft->g) {
+		kfree(ft->g);
+		ft->g = NULL;
+		kvfree(in);
+		return -ENOMEM;
+	}
+
+	mc = MLX5_ADDR_OF(create_flow_group_in, in, match_criteria);
+	outer_headers_c = MLX5_ADDR_OF(fte_match_param, mc, outer_headers);
+	MLX5_SET_TO_ONES(fte_match_set_lyr_2_4, outer_headers_c, ip_protocol);
+	MLX5_SET_TO_ONES(fte_match_set_lyr_2_4, outer_headers_c, ip_version);
+
+	switch (type) {
+	case ACCEL_FS_IPV4_TCP:
+	case ACCEL_FS_IPV6_TCP:
+		MLX5_SET_TO_ONES(fte_match_set_lyr_2_4, outer_headers_c, tcp_dport);
+		MLX5_SET_TO_ONES(fte_match_set_lyr_2_4, outer_headers_c, tcp_sport);
+		break;
+	default:
+		err = -EINVAL;
+		goto out;
+	}
+
+	switch (type) {
+	case ACCEL_FS_IPV4_TCP:
+		MLX5_SET_TO_ONES(fte_match_set_lyr_2_4, outer_headers_c,
+				 src_ipv4_src_ipv6.ipv4_layout.ipv4);
+		MLX5_SET_TO_ONES(fte_match_set_lyr_2_4, outer_headers_c,
+				 dst_ipv4_dst_ipv6.ipv4_layout.ipv4);
+		break;
+	case ACCEL_FS_IPV6_TCP:
+		memset(MLX5_ADDR_OF(fte_match_set_lyr_2_4, outer_headers_c,
+				    src_ipv4_src_ipv6.ipv6_layout.ipv6),
+		       0xff, 16);
+		memset(MLX5_ADDR_OF(fte_match_set_lyr_2_4, outer_headers_c,
+				    dst_ipv4_dst_ipv6.ipv6_layout.ipv6),
+		       0xff, 16);
+		break;
+	default:
+		err = -EINVAL;
+		goto out;
+	}
+
+	MLX5_SET_CFG(in, match_criteria_enable, MLX5_MATCH_OUTER_HEADERS);
+	MLX5_SET_CFG(in, start_flow_index, ix);
+	ix += MLX5E_ACCEL_FS_TCP_GROUP1_SIZE;
+	MLX5_SET_CFG(in, end_flow_index, ix - 1);
+	ft->g[ft->num_groups] = mlx5_create_flow_group(ft->t, in);
+	if (IS_ERR(ft->g[ft->num_groups]))
+		goto err;
+	ft->num_groups++;
+
+	/* Default Flow Group */
+	memset(in, 0, inlen);
+	MLX5_SET_CFG(in, start_flow_index, ix);
+	ix += MLX5E_ACCEL_FS_TCP_GROUP2_SIZE;
+	MLX5_SET_CFG(in, end_flow_index, ix - 1);
+	ft->g[ft->num_groups] = mlx5_create_flow_group(ft->t, in);
+	if (IS_ERR(ft->g[ft->num_groups]))
+		goto err;
+	ft->num_groups++;
+
+	kvfree(in);
+	return 0;
+
+err:
+	err = PTR_ERR(ft->g[ft->num_groups]);
+	ft->g[ft->num_groups] = NULL;
+out:
+	kvfree(in);
+
+	return err;
+}
+
+static int accel_fs_tcp_create_table(struct mlx5e_flow_steering *fs, enum accel_fs_tcp_type type)
+{
+	struct mlx5e_accel_fs_tcp *accel_tcp = mlx5e_fs_get_accel_tcp(fs);
+	struct mlx5_flow_namespace *ns = mlx5e_fs_get_ns(fs, false);
+	struct mlx5e_flow_table *ft = &accel_tcp->tables[type];
+	struct mlx5_flow_table_attr ft_attr = {};
+	int err;
+
+	ft->num_groups = 0;
+
+	ft_attr.max_fte = MLX5E_ACCEL_FS_TCP_TABLE_SIZE;
+	ft_attr.level = MLX5E_ACCEL_FS_TCP_FT_LEVEL;
+	ft_attr.prio = MLX5E_NIC_PRIO;
+
+	ft->t = mlx5_create_flow_table(ns, &ft_attr);
+	if (IS_ERR(ft->t)) {
+		err = PTR_ERR(ft->t);
+		ft->t = NULL;
+		return err;
+	}
+
+	fs_dbg(fs, "Created fs accel table id %u level %u\n",
+	       ft->t->id, ft->t->level);
+
+	err = accel_fs_tcp_create_groups(ft, type);
+	if (err)
+		goto err;
+
+	err = accel_fs_tcp_add_default_rule(fs, type);
+	if (err)
+		goto err;
+
+	return 0;
+err:
+	mlx5e_destroy_flow_table(ft);
+	return err;
+}
+
+static int accel_fs_tcp_disable(struct mlx5e_flow_steering *fs)
+{
+	struct mlx5_ttc_table *ttc = mlx5e_fs_get_ttc(fs, false);
+	int err, i;
+
+	for (i = 0; i < ACCEL_FS_TCP_NUM_TYPES; i++) {
+		/* Modify ttc rules destination to point back to the indir TIRs */
+		err = mlx5_ttc_fwd_default_dest(ttc, fs_accel2tt(i));
+		if (err) {
+			fs_err(fs,
+			       "%s: modify ttc[%d] default destination failed, err(%d)\n",
+			       __func__, fs_accel2tt(i), err);
+			return err;
+		}
+	}
+
+	return 0;
+}
+
+static int accel_fs_tcp_enable(struct mlx5e_flow_steering *fs)
+{
+	struct mlx5e_accel_fs_tcp *accel_tcp = mlx5e_fs_get_accel_tcp(fs);
+	struct mlx5_ttc_table *ttc = mlx5e_fs_get_ttc(fs, false);
+	struct mlx5_flow_destination dest = {};
+	int err, i;
+
+	dest.type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
+	for (i = 0; i < ACCEL_FS_TCP_NUM_TYPES; i++) {
+		dest.ft = accel_tcp->tables[i].t;
+
+		/* Modify ttc rules destination to point on the accel_fs FTs */
+		err = mlx5_ttc_fwd_dest(ttc, fs_accel2tt(i), &dest);
+		if (err) {
+			fs_err(fs, "%s: modify ttc[%d] destination to accel failed, err(%d)\n",
+			       __func__, fs_accel2tt(i), err);
+			return err;
+		}
+	}
+	return 0;
+}
+
+static void accel_fs_tcp_destroy_table(struct mlx5e_flow_steering *fs, int i)
+{
+	struct mlx5e_accel_fs_tcp *fs_tcp = mlx5e_fs_get_accel_tcp(fs);
+
+	if (IS_ERR_OR_NULL(fs_tcp->tables[i].t))
+		return;
+
+	mlx5_del_flow_rules(fs_tcp->default_rules[i]);
+	mlx5e_destroy_flow_table(&fs_tcp->tables[i]);
+	fs_tcp->tables[i].t = NULL;
+}
+
+void mlx5e_accel_fs_tcp_destroy(struct mlx5e_flow_steering *fs)
+{
+	struct mlx5e_accel_fs_tcp *accel_tcp = mlx5e_fs_get_accel_tcp(fs);
+	int i;
+
+	if (!accel_tcp)
+		return;
+
+	accel_fs_tcp_disable(fs);
+
+	for (i = 0; i < ACCEL_FS_TCP_NUM_TYPES; i++)
+		accel_fs_tcp_destroy_table(fs, i);
+
+	kvfree(accel_tcp);
+	mlx5e_fs_set_accel_tcp(fs, NULL);
+}
+
+int mlx5e_accel_fs_tcp_create(struct mlx5e_flow_steering *fs)
+{
+	struct mlx5e_accel_fs_tcp *accel_tcp;
+	int i, err;
+
+	if (!MLX5_CAP_FLOWTABLE_NIC_RX(mlx5e_fs_get_mdev(fs), ft_field_support.outer_ip_version))
+		return -EOPNOTSUPP;
+
+	accel_tcp = kvzalloc(sizeof(*accel_tcp), GFP_KERNEL);
+	if (!accel_tcp)
+		return -ENOMEM;
+	mlx5e_fs_set_accel_tcp(fs, accel_tcp);
+
+	for (i = 0; i < ACCEL_FS_TCP_NUM_TYPES; i++) {
+		err = accel_fs_tcp_create_table(fs, i);
+		if (err)
+			goto err_destroy_tables;
+	}
+
+	err = accel_fs_tcp_enable(fs);
+	if (err)
+		goto err_destroy_tables;
+
+	return 0;
+
+err_destroy_tables:
+	while (--i >= 0)
+		accel_fs_tcp_destroy_table(fs, i);
+	kvfree(accel_tcp);
+	mlx5e_fs_set_accel_tcp(fs, NULL);
+	return err;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/fs_tcp.h b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/fs_tcp.h
new file mode 100644
index 000000000..a032bff48
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/fs_tcp.h
@@ -0,0 +1,27 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2020, Mellanox Technologies inc. All rights reserved. */
+
+#ifndef __MLX5E_ACCEL_FS_TCP_H__
+#define __MLX5E_ACCEL_FS_TCP_H__
+
+#include "en/fs.h"
+
+#ifdef CONFIG_MLX5_EN_TLS
+int mlx5e_accel_fs_tcp_create(struct mlx5e_flow_steering *fs);
+void mlx5e_accel_fs_tcp_destroy(struct mlx5e_flow_steering *fs);
+struct mlx5_flow_handle *mlx5e_accel_fs_add_sk(struct mlx5e_flow_steering *fs,
+					       struct sock *sk, u32 tirn,
+					       uint32_t flow_tag);
+void mlx5e_accel_fs_del_sk(struct mlx5_flow_handle *rule);
+#else
+static inline int mlx5e_accel_fs_tcp_create(struct mlx5e_flow_steering *fs) { return 0; }
+static inline void mlx5e_accel_fs_tcp_destroy(struct mlx5e_flow_steering *fs) {}
+static inline struct mlx5_flow_handle *mlx5e_accel_fs_add_sk(struct mlx5e_flow_steering *fs,
+							     struct sock *sk, u32 tirn,
+							     uint32_t flow_tag)
+{ return ERR_PTR(-EOPNOTSUPP); }
+static inline void mlx5e_accel_fs_del_sk(struct mlx5_flow_handle *rule) {}
+#endif
+
+#endif /* __MLX5E_ACCEL_FS_TCP_H__ */
+
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec.c b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec.c
new file mode 100644
index 000000000..a71560186
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec.c
@@ -0,0 +1,469 @@
+/*
+ * Copyright (c) 2017 Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+
+#include <crypto/internal/geniv.h>
+#include <crypto/aead.h>
+#include <linux/inetdevice.h>
+#include <linux/netdevice.h>
+
+#include "en.h"
+#include "ipsec.h"
+#include "ipsec_rxtx.h"
+
+static struct mlx5e_ipsec_sa_entry *to_ipsec_sa_entry(struct xfrm_state *x)
+{
+	return (struct mlx5e_ipsec_sa_entry *)x->xso.offload_handle;
+}
+
+struct xfrm_state *mlx5e_ipsec_sadb_rx_lookup(struct mlx5e_ipsec *ipsec,
+					      unsigned int handle)
+{
+	struct mlx5e_ipsec_sa_entry *sa_entry;
+	struct xfrm_state *ret = NULL;
+
+	rcu_read_lock();
+	hash_for_each_possible_rcu(ipsec->sadb_rx, sa_entry, hlist, handle)
+		if (sa_entry->handle == handle) {
+			ret = sa_entry->x;
+			xfrm_state_hold(ret);
+			break;
+		}
+	rcu_read_unlock();
+
+	return ret;
+}
+
+static int mlx5e_ipsec_sadb_rx_add(struct mlx5e_ipsec_sa_entry *sa_entry)
+{
+	unsigned int handle = sa_entry->ipsec_obj_id;
+	struct mlx5e_ipsec *ipsec = sa_entry->ipsec;
+	struct mlx5e_ipsec_sa_entry *_sa_entry;
+	unsigned long flags;
+
+	rcu_read_lock();
+	hash_for_each_possible_rcu(ipsec->sadb_rx, _sa_entry, hlist, handle)
+		if (_sa_entry->handle == handle) {
+			rcu_read_unlock();
+			return  -EEXIST;
+		}
+	rcu_read_unlock();
+
+	spin_lock_irqsave(&ipsec->sadb_rx_lock, flags);
+	sa_entry->handle = handle;
+	hash_add_rcu(ipsec->sadb_rx, &sa_entry->hlist, sa_entry->handle);
+	spin_unlock_irqrestore(&ipsec->sadb_rx_lock, flags);
+
+	return 0;
+}
+
+static void mlx5e_ipsec_sadb_rx_del(struct mlx5e_ipsec_sa_entry *sa_entry)
+{
+	struct mlx5e_ipsec *ipsec = sa_entry->ipsec;
+	unsigned long flags;
+
+	spin_lock_irqsave(&ipsec->sadb_rx_lock, flags);
+	hash_del_rcu(&sa_entry->hlist);
+	spin_unlock_irqrestore(&ipsec->sadb_rx_lock, flags);
+}
+
+static bool mlx5e_ipsec_update_esn_state(struct mlx5e_ipsec_sa_entry *sa_entry)
+{
+	struct xfrm_replay_state_esn *replay_esn;
+	u32 seq_bottom = 0;
+	u8 overlap;
+
+	if (!(sa_entry->x->props.flags & XFRM_STATE_ESN)) {
+		sa_entry->esn_state.trigger = 0;
+		return false;
+	}
+
+	replay_esn = sa_entry->x->replay_esn;
+	if (replay_esn->seq >= replay_esn->replay_window)
+		seq_bottom = replay_esn->seq - replay_esn->replay_window + 1;
+
+	overlap = sa_entry->esn_state.overlap;
+
+	sa_entry->esn_state.esn = xfrm_replay_seqhi(sa_entry->x,
+						    htonl(seq_bottom));
+
+	sa_entry->esn_state.trigger = 1;
+	if (unlikely(overlap && seq_bottom < MLX5E_IPSEC_ESN_SCOPE_MID)) {
+		sa_entry->esn_state.overlap = 0;
+		return true;
+	} else if (unlikely(!overlap &&
+			    (seq_bottom >= MLX5E_IPSEC_ESN_SCOPE_MID))) {
+		sa_entry->esn_state.overlap = 1;
+		return true;
+	}
+
+	return false;
+}
+
+static void
+mlx5e_ipsec_build_accel_xfrm_attrs(struct mlx5e_ipsec_sa_entry *sa_entry,
+				   struct mlx5_accel_esp_xfrm_attrs *attrs)
+{
+	struct xfrm_state *x = sa_entry->x;
+	struct aes_gcm_keymat *aes_gcm = &attrs->aes_gcm;
+	struct aead_geniv_ctx *geniv_ctx;
+	struct crypto_aead *aead;
+	unsigned int crypto_data_len, key_len;
+	int ivsize;
+
+	memset(attrs, 0, sizeof(*attrs));
+
+	/* key */
+	crypto_data_len = (x->aead->alg_key_len + 7) / 8;
+	key_len = crypto_data_len - 4; /* 4 bytes salt at end */
+
+	memcpy(aes_gcm->aes_key, x->aead->alg_key, key_len);
+	aes_gcm->key_len = key_len * 8;
+
+	/* salt and seq_iv */
+	aead = x->data;
+	geniv_ctx = crypto_aead_ctx(aead);
+	ivsize = crypto_aead_ivsize(aead);
+	memcpy(&aes_gcm->seq_iv, &geniv_ctx->salt, ivsize);
+	memcpy(&aes_gcm->salt, x->aead->alg_key + key_len,
+	       sizeof(aes_gcm->salt));
+
+	/* iv len */
+	aes_gcm->icv_len = x->aead->alg_icv_len;
+
+	/* esn */
+	if (sa_entry->esn_state.trigger) {
+		attrs->flags |= MLX5_ACCEL_ESP_FLAGS_ESN_TRIGGERED;
+		attrs->esn = sa_entry->esn_state.esn;
+		if (sa_entry->esn_state.overlap)
+			attrs->flags |= MLX5_ACCEL_ESP_FLAGS_ESN_STATE_OVERLAP;
+	}
+
+	/* action */
+	attrs->action = (x->xso.dir == XFRM_DEV_OFFLOAD_OUT) ?
+				MLX5_ACCEL_ESP_ACTION_ENCRYPT :
+				      MLX5_ACCEL_ESP_ACTION_DECRYPT;
+	/* flags */
+	attrs->flags |= (x->props.mode == XFRM_MODE_TRANSPORT) ?
+			MLX5_ACCEL_ESP_FLAGS_TRANSPORT :
+			MLX5_ACCEL_ESP_FLAGS_TUNNEL;
+
+	/* spi */
+	attrs->spi = be32_to_cpu(x->id.spi);
+
+	/* source , destination ips */
+	memcpy(&attrs->saddr, x->props.saddr.a6, sizeof(attrs->saddr));
+	memcpy(&attrs->daddr, x->id.daddr.a6, sizeof(attrs->daddr));
+	attrs->is_ipv6 = (x->props.family != AF_INET);
+}
+
+static inline int mlx5e_xfrm_validate_state(struct xfrm_state *x)
+{
+	struct net_device *netdev = x->xso.real_dev;
+	struct mlx5e_priv *priv;
+
+	priv = netdev_priv(netdev);
+
+	if (x->props.aalgo != SADB_AALG_NONE) {
+		netdev_info(netdev, "Cannot offload authenticated xfrm states\n");
+		return -EINVAL;
+	}
+	if (x->props.ealgo != SADB_X_EALG_AES_GCM_ICV16) {
+		netdev_info(netdev, "Only AES-GCM-ICV16 xfrm state may be offloaded\n");
+		return -EINVAL;
+	}
+	if (x->props.calgo != SADB_X_CALG_NONE) {
+		netdev_info(netdev, "Cannot offload compressed xfrm states\n");
+		return -EINVAL;
+	}
+	if (x->props.flags & XFRM_STATE_ESN &&
+	    !(mlx5_ipsec_device_caps(priv->mdev) & MLX5_IPSEC_CAP_ESN)) {
+		netdev_info(netdev, "Cannot offload ESN xfrm states\n");
+		return -EINVAL;
+	}
+	if (x->props.family != AF_INET &&
+	    x->props.family != AF_INET6) {
+		netdev_info(netdev, "Only IPv4/6 xfrm states may be offloaded\n");
+		return -EINVAL;
+	}
+	if (x->props.mode != XFRM_MODE_TRANSPORT &&
+	    x->props.mode != XFRM_MODE_TUNNEL) {
+		dev_info(&netdev->dev, "Only transport and tunnel xfrm states may be offloaded\n");
+		return -EINVAL;
+	}
+	if (x->id.proto != IPPROTO_ESP) {
+		netdev_info(netdev, "Only ESP xfrm state may be offloaded\n");
+		return -EINVAL;
+	}
+	if (x->encap) {
+		netdev_info(netdev, "Encapsulated xfrm state may not be offloaded\n");
+		return -EINVAL;
+	}
+	if (!x->aead) {
+		netdev_info(netdev, "Cannot offload xfrm states without aead\n");
+		return -EINVAL;
+	}
+	if (x->aead->alg_icv_len != 128) {
+		netdev_info(netdev, "Cannot offload xfrm states with AEAD ICV length other than 128bit\n");
+		return -EINVAL;
+	}
+	if ((x->aead->alg_key_len != 128 + 32) &&
+	    (x->aead->alg_key_len != 256 + 32)) {
+		netdev_info(netdev, "Cannot offload xfrm states with AEAD key length other than 128/256 bit\n");
+		return -EINVAL;
+	}
+	if (x->tfcpad) {
+		netdev_info(netdev, "Cannot offload xfrm states with tfc padding\n");
+		return -EINVAL;
+	}
+	if (!x->geniv) {
+		netdev_info(netdev, "Cannot offload xfrm states without geniv\n");
+		return -EINVAL;
+	}
+	if (strcmp(x->geniv, "seqiv")) {
+		netdev_info(netdev, "Cannot offload xfrm states with geniv other than seqiv\n");
+		return -EINVAL;
+	}
+	return 0;
+}
+
+static void _update_xfrm_state(struct work_struct *work)
+{
+	struct mlx5e_ipsec_modify_state_work *modify_work =
+		container_of(work, struct mlx5e_ipsec_modify_state_work, work);
+	struct mlx5e_ipsec_sa_entry *sa_entry = container_of(
+		modify_work, struct mlx5e_ipsec_sa_entry, modify_work);
+
+	mlx5_accel_esp_modify_xfrm(sa_entry, &modify_work->attrs);
+}
+
+static int mlx5e_xfrm_add_state(struct xfrm_state *x)
+{
+	struct mlx5e_ipsec_sa_entry *sa_entry = NULL;
+	struct net_device *netdev = x->xso.real_dev;
+	struct mlx5e_priv *priv;
+	int err;
+
+	priv = netdev_priv(netdev);
+	if (!priv->ipsec)
+		return -EOPNOTSUPP;
+
+	err = mlx5e_xfrm_validate_state(x);
+	if (err)
+		return err;
+
+	sa_entry = kzalloc(sizeof(*sa_entry), GFP_KERNEL);
+	if (!sa_entry) {
+		err = -ENOMEM;
+		goto out;
+	}
+
+	sa_entry->x = x;
+	sa_entry->ipsec = priv->ipsec;
+
+	/* check esn */
+	mlx5e_ipsec_update_esn_state(sa_entry);
+
+	mlx5e_ipsec_build_accel_xfrm_attrs(sa_entry, &sa_entry->attrs);
+	/* create hw context */
+	err = mlx5_ipsec_create_sa_ctx(sa_entry);
+	if (err)
+		goto err_xfrm;
+
+	err = mlx5e_accel_ipsec_fs_add_rule(priv, sa_entry);
+	if (err)
+		goto err_hw_ctx;
+
+	if (x->xso.dir == XFRM_DEV_OFFLOAD_IN) {
+		err = mlx5e_ipsec_sadb_rx_add(sa_entry);
+		if (err)
+			goto err_add_rule;
+	} else {
+		sa_entry->set_iv_op = (x->props.flags & XFRM_STATE_ESN) ?
+				mlx5e_ipsec_set_iv_esn : mlx5e_ipsec_set_iv;
+	}
+
+	INIT_WORK(&sa_entry->modify_work.work, _update_xfrm_state);
+	x->xso.offload_handle = (unsigned long)sa_entry;
+	goto out;
+
+err_add_rule:
+	mlx5e_accel_ipsec_fs_del_rule(priv, sa_entry);
+err_hw_ctx:
+	mlx5_ipsec_free_sa_ctx(sa_entry);
+err_xfrm:
+	kfree(sa_entry);
+out:
+	return err;
+}
+
+static void mlx5e_xfrm_del_state(struct xfrm_state *x)
+{
+	struct mlx5e_ipsec_sa_entry *sa_entry = to_ipsec_sa_entry(x);
+
+	if (x->xso.dir == XFRM_DEV_OFFLOAD_IN)
+		mlx5e_ipsec_sadb_rx_del(sa_entry);
+}
+
+static void mlx5e_xfrm_free_state(struct xfrm_state *x)
+{
+	struct mlx5e_ipsec_sa_entry *sa_entry = to_ipsec_sa_entry(x);
+	struct mlx5e_priv *priv = netdev_priv(x->xso.dev);
+
+	cancel_work_sync(&sa_entry->modify_work.work);
+	mlx5e_accel_ipsec_fs_del_rule(priv, sa_entry);
+	mlx5_ipsec_free_sa_ctx(sa_entry);
+	kfree(sa_entry);
+}
+
+int mlx5e_ipsec_init(struct mlx5e_priv *priv)
+{
+	struct mlx5e_ipsec *ipsec;
+	int ret;
+
+	if (!mlx5_ipsec_device_caps(priv->mdev)) {
+		netdev_dbg(priv->netdev, "Not an IPSec offload device\n");
+		return 0;
+	}
+
+	ipsec = kzalloc(sizeof(*ipsec), GFP_KERNEL);
+	if (!ipsec)
+		return -ENOMEM;
+
+	hash_init(ipsec->sadb_rx);
+	spin_lock_init(&ipsec->sadb_rx_lock);
+	ipsec->mdev = priv->mdev;
+	ipsec->wq = alloc_ordered_workqueue("mlx5e_ipsec: %s", 0,
+					    priv->netdev->name);
+	if (!ipsec->wq) {
+		ret = -ENOMEM;
+		goto err_wq;
+	}
+
+	ret = mlx5e_accel_ipsec_fs_init(ipsec);
+	if (ret)
+		goto err_fs_init;
+
+	priv->ipsec = ipsec;
+	netdev_dbg(priv->netdev, "IPSec attached to netdevice\n");
+	return 0;
+
+err_fs_init:
+	destroy_workqueue(ipsec->wq);
+err_wq:
+	kfree(ipsec);
+	return (ret != -EOPNOTSUPP) ? ret : 0;
+}
+
+void mlx5e_ipsec_cleanup(struct mlx5e_priv *priv)
+{
+	struct mlx5e_ipsec *ipsec = priv->ipsec;
+
+	if (!ipsec)
+		return;
+
+	mlx5e_accel_ipsec_fs_cleanup(ipsec);
+	destroy_workqueue(ipsec->wq);
+	kfree(ipsec);
+	priv->ipsec = NULL;
+}
+
+static bool mlx5e_ipsec_offload_ok(struct sk_buff *skb, struct xfrm_state *x)
+{
+	if (x->props.family == AF_INET) {
+		/* Offload with IPv4 options is not supported yet */
+		if (ip_hdr(skb)->ihl > 5)
+			return false;
+	} else {
+		/* Offload with IPv6 extension headers is not support yet */
+		if (ipv6_ext_hdr(ipv6_hdr(skb)->nexthdr))
+			return false;
+	}
+
+	return true;
+}
+
+static void mlx5e_xfrm_advance_esn_state(struct xfrm_state *x)
+{
+	struct mlx5e_ipsec_sa_entry *sa_entry = to_ipsec_sa_entry(x);
+	struct mlx5e_ipsec_modify_state_work *modify_work =
+		&sa_entry->modify_work;
+	bool need_update;
+
+	need_update = mlx5e_ipsec_update_esn_state(sa_entry);
+	if (!need_update)
+		return;
+
+	mlx5e_ipsec_build_accel_xfrm_attrs(sa_entry, &modify_work->attrs);
+	queue_work(sa_entry->ipsec->wq, &modify_work->work);
+}
+
+static const struct xfrmdev_ops mlx5e_ipsec_xfrmdev_ops = {
+	.xdo_dev_state_add	= mlx5e_xfrm_add_state,
+	.xdo_dev_state_delete	= mlx5e_xfrm_del_state,
+	.xdo_dev_state_free	= mlx5e_xfrm_free_state,
+	.xdo_dev_offload_ok	= mlx5e_ipsec_offload_ok,
+	.xdo_dev_state_advance_esn = mlx5e_xfrm_advance_esn_state,
+};
+
+void mlx5e_ipsec_build_netdev(struct mlx5e_priv *priv)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+	struct net_device *netdev = priv->netdev;
+
+	if (!mlx5_ipsec_device_caps(mdev))
+		return;
+
+	mlx5_core_info(mdev, "mlx5e: IPSec ESP acceleration enabled\n");
+	netdev->xfrmdev_ops = &mlx5e_ipsec_xfrmdev_ops;
+	netdev->features |= NETIF_F_HW_ESP;
+	netdev->hw_enc_features |= NETIF_F_HW_ESP;
+
+	if (!MLX5_CAP_ETH(mdev, swp_csum)) {
+		mlx5_core_dbg(mdev, "mlx5e: SWP checksum not supported\n");
+		return;
+	}
+
+	netdev->features |= NETIF_F_HW_ESP_TX_CSUM;
+	netdev->hw_enc_features |= NETIF_F_HW_ESP_TX_CSUM;
+
+	if (!MLX5_CAP_ETH(mdev, swp_lso)) {
+		mlx5_core_dbg(mdev, "mlx5e: ESP LSO not supported\n");
+		return;
+	}
+
+	netdev->gso_partial_features |= NETIF_F_GSO_ESP;
+	mlx5_core_dbg(mdev, "mlx5e: ESP GSO capability turned on\n");
+	netdev->features |= NETIF_F_GSO_ESP;
+	netdev->hw_features |= NETIF_F_GSO_ESP;
+	netdev->hw_enc_features |= NETIF_F_GSO_ESP;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec.h b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec.h
new file mode 100644
index 000000000..785f18814
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec.h
@@ -0,0 +1,196 @@
+/*
+ * Copyright (c) 2017 Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+
+#ifndef __MLX5E_IPSEC_H__
+#define __MLX5E_IPSEC_H__
+
+
+#include <linux/mlx5/device.h>
+#include <net/xfrm.h>
+#include <linux/idr.h>
+
+#define MLX5E_IPSEC_SADB_RX_BITS 10
+#define MLX5E_IPSEC_ESN_SCOPE_MID 0x80000000L
+
+enum mlx5_accel_esp_flags {
+	MLX5_ACCEL_ESP_FLAGS_TUNNEL            = 0,    /* Default */
+	MLX5_ACCEL_ESP_FLAGS_TRANSPORT         = 1UL << 0,
+	MLX5_ACCEL_ESP_FLAGS_ESN_TRIGGERED     = 1UL << 1,
+	MLX5_ACCEL_ESP_FLAGS_ESN_STATE_OVERLAP = 1UL << 2,
+};
+
+enum mlx5_accel_esp_action {
+	MLX5_ACCEL_ESP_ACTION_DECRYPT,
+	MLX5_ACCEL_ESP_ACTION_ENCRYPT,
+};
+
+struct aes_gcm_keymat {
+	u64   seq_iv;
+
+	u32   salt;
+	u32   icv_len;
+
+	u32   key_len;
+	u32   aes_key[256 / 32];
+};
+
+struct mlx5_accel_esp_xfrm_attrs {
+	enum mlx5_accel_esp_action action;
+	u32   esn;
+	u32   spi;
+	u32   flags;
+	struct aes_gcm_keymat aes_gcm;
+
+	union {
+		__be32 a4;
+		__be32 a6[4];
+	} saddr;
+
+	union {
+		__be32 a4;
+		__be32 a6[4];
+	} daddr;
+
+	u8 is_ipv6;
+};
+
+enum mlx5_ipsec_cap {
+	MLX5_IPSEC_CAP_CRYPTO		= 1 << 0,
+	MLX5_IPSEC_CAP_ESN		= 1 << 1,
+};
+
+struct mlx5e_priv;
+
+struct mlx5e_ipsec_sw_stats {
+	atomic64_t ipsec_rx_drop_sp_alloc;
+	atomic64_t ipsec_rx_drop_sadb_miss;
+	atomic64_t ipsec_rx_drop_syndrome;
+	atomic64_t ipsec_tx_drop_bundle;
+	atomic64_t ipsec_tx_drop_no_state;
+	atomic64_t ipsec_tx_drop_not_ip;
+	atomic64_t ipsec_tx_drop_trailer;
+};
+
+struct mlx5e_accel_fs_esp;
+struct mlx5e_ipsec_tx;
+
+struct mlx5e_ipsec {
+	struct mlx5_core_dev *mdev;
+	DECLARE_HASHTABLE(sadb_rx, MLX5E_IPSEC_SADB_RX_BITS);
+	spinlock_t sadb_rx_lock; /* Protects sadb_rx */
+	struct mlx5e_ipsec_sw_stats sw_stats;
+	struct workqueue_struct *wq;
+	struct mlx5e_accel_fs_esp *rx_fs;
+	struct mlx5e_ipsec_tx *tx_fs;
+};
+
+struct mlx5e_ipsec_esn_state {
+	u32 esn;
+	u8 trigger: 1;
+	u8 overlap: 1;
+};
+
+struct mlx5e_ipsec_rule {
+	struct mlx5_flow_handle *rule;
+	struct mlx5_modify_hdr *set_modify_hdr;
+};
+
+struct mlx5e_ipsec_modify_state_work {
+	struct work_struct		work;
+	struct mlx5_accel_esp_xfrm_attrs attrs;
+};
+
+struct mlx5e_ipsec_sa_entry {
+	struct hlist_node hlist; /* Item in SADB_RX hashtable */
+	struct mlx5e_ipsec_esn_state esn_state;
+	unsigned int handle; /* Handle in SADB_RX */
+	struct xfrm_state *x;
+	struct mlx5e_ipsec *ipsec;
+	struct mlx5_accel_esp_xfrm_attrs attrs;
+	void (*set_iv_op)(struct sk_buff *skb, struct xfrm_state *x,
+			  struct xfrm_offload *xo);
+	u32 ipsec_obj_id;
+	u32 enc_key_id;
+	struct mlx5e_ipsec_rule ipsec_rule;
+	struct mlx5e_ipsec_modify_state_work modify_work;
+};
+
+#ifdef CONFIG_MLX5_EN_IPSEC
+int mlx5e_ipsec_init(struct mlx5e_priv *priv);
+void mlx5e_ipsec_cleanup(struct mlx5e_priv *priv);
+void mlx5e_ipsec_build_netdev(struct mlx5e_priv *priv);
+
+struct xfrm_state *mlx5e_ipsec_sadb_rx_lookup(struct mlx5e_ipsec *dev,
+					      unsigned int handle);
+
+void mlx5e_accel_ipsec_fs_cleanup(struct mlx5e_ipsec *ipsec);
+int mlx5e_accel_ipsec_fs_init(struct mlx5e_ipsec *ipsec);
+int mlx5e_accel_ipsec_fs_add_rule(struct mlx5e_priv *priv,
+				  struct mlx5e_ipsec_sa_entry *sa_entry);
+void mlx5e_accel_ipsec_fs_del_rule(struct mlx5e_priv *priv,
+				   struct mlx5e_ipsec_sa_entry *sa_entry);
+
+int mlx5_ipsec_create_sa_ctx(struct mlx5e_ipsec_sa_entry *sa_entry);
+void mlx5_ipsec_free_sa_ctx(struct mlx5e_ipsec_sa_entry *sa_entry);
+
+u32 mlx5_ipsec_device_caps(struct mlx5_core_dev *mdev);
+
+void mlx5_accel_esp_modify_xfrm(struct mlx5e_ipsec_sa_entry *sa_entry,
+				const struct mlx5_accel_esp_xfrm_attrs *attrs);
+
+static inline struct mlx5_core_dev *
+mlx5e_ipsec_sa2dev(struct mlx5e_ipsec_sa_entry *sa_entry)
+{
+	return sa_entry->ipsec->mdev;
+}
+#else
+static inline int mlx5e_ipsec_init(struct mlx5e_priv *priv)
+{
+	return 0;
+}
+
+static inline void mlx5e_ipsec_cleanup(struct mlx5e_priv *priv)
+{
+}
+
+static inline void mlx5e_ipsec_build_netdev(struct mlx5e_priv *priv)
+{
+}
+
+static inline u32 mlx5_ipsec_device_caps(struct mlx5_core_dev *mdev)
+{
+	return 0;
+}
+#endif
+
+#endif	/* __MLX5E_IPSEC_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_fs.c b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_fs.c
new file mode 100644
index 000000000..b859e4a4c
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_fs.c
@@ -0,0 +1,606 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2020, Mellanox Technologies inc. All rights reserved. */
+
+#include <linux/netdevice.h>
+#include "en.h"
+#include "en/fs.h"
+#include "ipsec.h"
+#include "fs_core.h"
+
+#define NUM_IPSEC_FTE BIT(15)
+
+enum accel_fs_esp_type {
+	ACCEL_FS_ESP4,
+	ACCEL_FS_ESP6,
+	ACCEL_FS_ESP_NUM_TYPES,
+};
+
+struct mlx5e_ipsec_rx_err {
+	struct mlx5_flow_table *ft;
+	struct mlx5_flow_handle *rule;
+	struct mlx5_modify_hdr *copy_modify_hdr;
+};
+
+struct mlx5e_accel_fs_esp_prot {
+	struct mlx5_flow_table *ft;
+	struct mlx5_flow_group *miss_group;
+	struct mlx5_flow_handle *miss_rule;
+	struct mlx5_flow_destination default_dest;
+	struct mlx5e_ipsec_rx_err rx_err;
+	u32 refcnt;
+	struct mutex prot_mutex; /* protect ESP4/ESP6 protocol */
+};
+
+struct mlx5e_accel_fs_esp {
+	struct mlx5e_accel_fs_esp_prot fs_prot[ACCEL_FS_ESP_NUM_TYPES];
+};
+
+struct mlx5e_ipsec_tx {
+	struct mlx5_flow_namespace *ns;
+	struct mlx5_flow_table *ft;
+	struct mutex mutex; /* Protect IPsec TX steering */
+	u32 refcnt;
+};
+
+/* IPsec RX flow steering */
+static enum mlx5_traffic_types fs_esp2tt(enum accel_fs_esp_type i)
+{
+	if (i == ACCEL_FS_ESP4)
+		return MLX5_TT_IPV4_IPSEC_ESP;
+	return MLX5_TT_IPV6_IPSEC_ESP;
+}
+
+static int rx_err_add_rule(struct mlx5e_priv *priv,
+			   struct mlx5e_accel_fs_esp_prot *fs_prot,
+			   struct mlx5e_ipsec_rx_err *rx_err)
+{
+	u8 action[MLX5_UN_SZ_BYTES(set_add_copy_action_in_auto)] = {};
+	struct mlx5_core_dev *mdev = priv->mdev;
+	struct mlx5_flow_act flow_act = {};
+	struct mlx5_modify_hdr *modify_hdr;
+	struct mlx5_flow_handle *fte;
+	struct mlx5_flow_spec *spec;
+	int err;
+
+	spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
+	if (!spec)
+		return -ENOMEM;
+
+	/* Action to copy 7 bit ipsec_syndrome to regB[24:30] */
+	MLX5_SET(copy_action_in, action, action_type, MLX5_ACTION_TYPE_COPY);
+	MLX5_SET(copy_action_in, action, src_field, MLX5_ACTION_IN_FIELD_IPSEC_SYNDROME);
+	MLX5_SET(copy_action_in, action, src_offset, 0);
+	MLX5_SET(copy_action_in, action, length, 7);
+	MLX5_SET(copy_action_in, action, dst_field, MLX5_ACTION_IN_FIELD_METADATA_REG_B);
+	MLX5_SET(copy_action_in, action, dst_offset, 24);
+
+	modify_hdr = mlx5_modify_header_alloc(mdev, MLX5_FLOW_NAMESPACE_KERNEL,
+					      1, action);
+
+	if (IS_ERR(modify_hdr)) {
+		err = PTR_ERR(modify_hdr);
+		netdev_err(priv->netdev,
+			   "fail to alloc ipsec copy modify_header_id err=%d\n", err);
+		goto out_spec;
+	}
+
+	/* create fte */
+	flow_act.action = MLX5_FLOW_CONTEXT_ACTION_MOD_HDR |
+			  MLX5_FLOW_CONTEXT_ACTION_FWD_DEST;
+	flow_act.modify_hdr = modify_hdr;
+	fte = mlx5_add_flow_rules(rx_err->ft, spec, &flow_act,
+				  &fs_prot->default_dest, 1);
+	if (IS_ERR(fte)) {
+		err = PTR_ERR(fte);
+		netdev_err(priv->netdev, "fail to add ipsec rx err copy rule err=%d\n", err);
+		goto out;
+	}
+
+	kvfree(spec);
+	rx_err->rule = fte;
+	rx_err->copy_modify_hdr = modify_hdr;
+	return 0;
+
+out:
+	mlx5_modify_header_dealloc(mdev, modify_hdr);
+out_spec:
+	kvfree(spec);
+	return err;
+}
+
+static int rx_fs_create(struct mlx5e_priv *priv,
+			struct mlx5e_accel_fs_esp_prot *fs_prot)
+{
+	int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
+	struct mlx5_flow_table *ft = fs_prot->ft;
+	struct mlx5_flow_group *miss_group;
+	struct mlx5_flow_handle *miss_rule;
+	MLX5_DECLARE_FLOW_ACT(flow_act);
+	struct mlx5_flow_spec *spec;
+	u32 *flow_group_in;
+	int err = 0;
+
+	flow_group_in = kvzalloc(inlen, GFP_KERNEL);
+	spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
+	if (!flow_group_in || !spec) {
+		err = -ENOMEM;
+		goto out;
+	}
+
+	/* Create miss_group */
+	MLX5_SET(create_flow_group_in, flow_group_in, start_flow_index, ft->max_fte - 1);
+	MLX5_SET(create_flow_group_in, flow_group_in, end_flow_index, ft->max_fte - 1);
+	miss_group = mlx5_create_flow_group(ft, flow_group_in);
+	if (IS_ERR(miss_group)) {
+		err = PTR_ERR(miss_group);
+		netdev_err(priv->netdev, "fail to create ipsec rx miss_group err=%d\n", err);
+		goto out;
+	}
+	fs_prot->miss_group = miss_group;
+
+	/* Create miss rule */
+	miss_rule = mlx5_add_flow_rules(ft, spec, &flow_act, &fs_prot->default_dest, 1);
+	if (IS_ERR(miss_rule)) {
+		mlx5_destroy_flow_group(fs_prot->miss_group);
+		err = PTR_ERR(miss_rule);
+		netdev_err(priv->netdev, "fail to create ipsec rx miss_rule err=%d\n", err);
+		goto out;
+	}
+	fs_prot->miss_rule = miss_rule;
+out:
+	kvfree(flow_group_in);
+	kvfree(spec);
+	return err;
+}
+
+static void rx_destroy(struct mlx5e_priv *priv, enum accel_fs_esp_type type)
+{
+	struct mlx5e_accel_fs_esp_prot *fs_prot;
+	struct mlx5e_accel_fs_esp *accel_esp;
+
+	accel_esp = priv->ipsec->rx_fs;
+
+	/* The netdev unreg already happened, so all offloaded rule are already removed */
+	fs_prot = &accel_esp->fs_prot[type];
+
+	mlx5_del_flow_rules(fs_prot->miss_rule);
+	mlx5_destroy_flow_group(fs_prot->miss_group);
+	mlx5_destroy_flow_table(fs_prot->ft);
+
+	mlx5_del_flow_rules(fs_prot->rx_err.rule);
+	mlx5_modify_header_dealloc(priv->mdev, fs_prot->rx_err.copy_modify_hdr);
+	mlx5_destroy_flow_table(fs_prot->rx_err.ft);
+}
+
+static int rx_create(struct mlx5e_priv *priv, enum accel_fs_esp_type type)
+{
+	struct mlx5_flow_namespace *ns = mlx5e_fs_get_ns(priv->fs, false);
+	struct mlx5_ttc_table *ttc = mlx5e_fs_get_ttc(priv->fs, false);
+	struct mlx5_flow_table_attr ft_attr = {};
+	struct mlx5e_accel_fs_esp_prot *fs_prot;
+	struct mlx5e_accel_fs_esp *accel_esp;
+	struct mlx5_flow_table *ft;
+	int err;
+
+	accel_esp = priv->ipsec->rx_fs;
+	fs_prot = &accel_esp->fs_prot[type];
+	fs_prot->default_dest =
+		mlx5_ttc_get_default_dest(ttc, fs_esp2tt(type));
+
+	ft_attr.max_fte = 1;
+	ft_attr.autogroup.max_num_groups = 1;
+	ft_attr.level = MLX5E_ACCEL_FS_ESP_FT_ERR_LEVEL;
+	ft_attr.prio = MLX5E_NIC_PRIO;
+	ft = mlx5_create_auto_grouped_flow_table(ns, &ft_attr);
+	if (IS_ERR(ft))
+		return PTR_ERR(ft);
+
+	fs_prot->rx_err.ft = ft;
+	err = rx_err_add_rule(priv, fs_prot, &fs_prot->rx_err);
+	if (err)
+		goto err_add;
+
+	/* Create FT */
+	ft_attr.max_fte = NUM_IPSEC_FTE;
+	ft_attr.level = MLX5E_ACCEL_FS_ESP_FT_LEVEL;
+	ft_attr.prio = MLX5E_NIC_PRIO;
+	ft_attr.autogroup.num_reserved_entries = 1;
+	ft_attr.autogroup.max_num_groups = 1;
+	ft = mlx5_create_auto_grouped_flow_table(ns, &ft_attr);
+	if (IS_ERR(ft)) {
+		err = PTR_ERR(ft);
+		goto err_fs_ft;
+	}
+	fs_prot->ft = ft;
+
+	err = rx_fs_create(priv, fs_prot);
+	if (err)
+		goto err_fs;
+
+	return 0;
+
+err_fs:
+	mlx5_destroy_flow_table(fs_prot->ft);
+err_fs_ft:
+	mlx5_del_flow_rules(fs_prot->rx_err.rule);
+	mlx5_modify_header_dealloc(priv->mdev, fs_prot->rx_err.copy_modify_hdr);
+err_add:
+	mlx5_destroy_flow_table(fs_prot->rx_err.ft);
+	return err;
+}
+
+static int rx_ft_get(struct mlx5e_priv *priv, enum accel_fs_esp_type type)
+{
+	struct mlx5_ttc_table *ttc = mlx5e_fs_get_ttc(priv->fs, false);
+	struct mlx5e_accel_fs_esp_prot *fs_prot;
+	struct mlx5_flow_destination dest = {};
+	struct mlx5e_accel_fs_esp *accel_esp;
+	int err = 0;
+
+	accel_esp = priv->ipsec->rx_fs;
+	fs_prot = &accel_esp->fs_prot[type];
+	mutex_lock(&fs_prot->prot_mutex);
+	if (fs_prot->refcnt)
+		goto skip;
+
+	/* create FT */
+	err = rx_create(priv, type);
+	if (err)
+		goto out;
+
+	/* connect */
+	dest.type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
+	dest.ft = fs_prot->ft;
+	mlx5_ttc_fwd_dest(ttc, fs_esp2tt(type), &dest);
+
+skip:
+	fs_prot->refcnt++;
+out:
+	mutex_unlock(&fs_prot->prot_mutex);
+	return err;
+}
+
+static void rx_ft_put(struct mlx5e_priv *priv, enum accel_fs_esp_type type)
+{
+	struct mlx5_ttc_table *ttc = mlx5e_fs_get_ttc(priv->fs, false);
+	struct mlx5e_accel_fs_esp_prot *fs_prot;
+	struct mlx5e_accel_fs_esp *accel_esp;
+
+	accel_esp = priv->ipsec->rx_fs;
+	fs_prot = &accel_esp->fs_prot[type];
+	mutex_lock(&fs_prot->prot_mutex);
+	fs_prot->refcnt--;
+	if (fs_prot->refcnt)
+		goto out;
+
+	/* disconnect */
+	mlx5_ttc_fwd_default_dest(ttc, fs_esp2tt(type));
+
+	/* remove FT */
+	rx_destroy(priv, type);
+
+out:
+	mutex_unlock(&fs_prot->prot_mutex);
+}
+
+/* IPsec TX flow steering */
+static int tx_create(struct mlx5e_priv *priv)
+{
+	struct mlx5_flow_table_attr ft_attr = {};
+	struct mlx5e_ipsec *ipsec = priv->ipsec;
+	struct mlx5_flow_table *ft;
+	int err;
+
+	ft_attr.max_fte = NUM_IPSEC_FTE;
+	ft_attr.autogroup.max_num_groups = 1;
+	ft = mlx5_create_auto_grouped_flow_table(ipsec->tx_fs->ns, &ft_attr);
+	if (IS_ERR(ft)) {
+		err = PTR_ERR(ft);
+		netdev_err(priv->netdev, "fail to create ipsec tx ft err=%d\n", err);
+		return err;
+	}
+	ipsec->tx_fs->ft = ft;
+	return 0;
+}
+
+static int tx_ft_get(struct mlx5e_priv *priv)
+{
+	struct mlx5e_ipsec_tx *tx_fs = priv->ipsec->tx_fs;
+	int err = 0;
+
+	mutex_lock(&tx_fs->mutex);
+	if (tx_fs->refcnt)
+		goto skip;
+
+	err = tx_create(priv);
+	if (err)
+		goto out;
+skip:
+	tx_fs->refcnt++;
+out:
+	mutex_unlock(&tx_fs->mutex);
+	return err;
+}
+
+static void tx_ft_put(struct mlx5e_priv *priv)
+{
+	struct mlx5e_ipsec_tx *tx_fs = priv->ipsec->tx_fs;
+
+	mutex_lock(&tx_fs->mutex);
+	tx_fs->refcnt--;
+	if (tx_fs->refcnt)
+		goto out;
+
+	mlx5_destroy_flow_table(tx_fs->ft);
+out:
+	mutex_unlock(&tx_fs->mutex);
+}
+
+static void setup_fte_common(struct mlx5_accel_esp_xfrm_attrs *attrs,
+			     u32 ipsec_obj_id,
+			     struct mlx5_flow_spec *spec,
+			     struct mlx5_flow_act *flow_act)
+{
+	u8 ip_version = attrs->is_ipv6 ? 6 : 4;
+
+	spec->match_criteria_enable = MLX5_MATCH_OUTER_HEADERS | MLX5_MATCH_MISC_PARAMETERS;
+
+	/* ip_version */
+	MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria, outer_headers.ip_version);
+	MLX5_SET(fte_match_param, spec->match_value, outer_headers.ip_version, ip_version);
+
+	/* Non fragmented */
+	MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria, outer_headers.frag);
+	MLX5_SET(fte_match_param, spec->match_value, outer_headers.frag, 0);
+
+	/* ESP header */
+	MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria, outer_headers.ip_protocol);
+	MLX5_SET(fte_match_param, spec->match_value, outer_headers.ip_protocol, IPPROTO_ESP);
+
+	/* SPI number */
+	MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria, misc_parameters.outer_esp_spi);
+	MLX5_SET(fte_match_param, spec->match_value,
+		 misc_parameters.outer_esp_spi, attrs->spi);
+
+	if (ip_version == 4) {
+		memcpy(MLX5_ADDR_OF(fte_match_param, spec->match_value,
+				    outer_headers.src_ipv4_src_ipv6.ipv4_layout.ipv4),
+		       &attrs->saddr.a4, 4);
+		memcpy(MLX5_ADDR_OF(fte_match_param, spec->match_value,
+				    outer_headers.dst_ipv4_dst_ipv6.ipv4_layout.ipv4),
+		       &attrs->daddr.a4, 4);
+		MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria,
+				 outer_headers.src_ipv4_src_ipv6.ipv4_layout.ipv4);
+		MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria,
+				 outer_headers.dst_ipv4_dst_ipv6.ipv4_layout.ipv4);
+	} else {
+		memcpy(MLX5_ADDR_OF(fte_match_param, spec->match_value,
+				    outer_headers.src_ipv4_src_ipv6.ipv6_layout.ipv6),
+		       &attrs->saddr.a6, 16);
+		memcpy(MLX5_ADDR_OF(fte_match_param, spec->match_value,
+				    outer_headers.dst_ipv4_dst_ipv6.ipv6_layout.ipv6),
+		       &attrs->daddr.a6, 16);
+		memset(MLX5_ADDR_OF(fte_match_param, spec->match_criteria,
+				    outer_headers.src_ipv4_src_ipv6.ipv6_layout.ipv6),
+		       0xff, 16);
+		memset(MLX5_ADDR_OF(fte_match_param, spec->match_criteria,
+				    outer_headers.dst_ipv4_dst_ipv6.ipv6_layout.ipv6),
+		       0xff, 16);
+	}
+
+	flow_act->crypto.type = MLX5_FLOW_CONTEXT_ENCRYPT_DECRYPT_TYPE_IPSEC;
+	flow_act->crypto.obj_id = ipsec_obj_id;
+	flow_act->flags |= FLOW_ACT_NO_APPEND;
+}
+
+static int rx_add_rule(struct mlx5e_priv *priv,
+		       struct mlx5e_ipsec_sa_entry *sa_entry)
+{
+	u8 action[MLX5_UN_SZ_BYTES(set_add_copy_action_in_auto)] = {};
+	struct mlx5e_ipsec_rule *ipsec_rule = &sa_entry->ipsec_rule;
+	struct mlx5_accel_esp_xfrm_attrs *attrs = &sa_entry->attrs;
+	u32 ipsec_obj_id = sa_entry->ipsec_obj_id;
+	struct mlx5_modify_hdr *modify_hdr = NULL;
+	struct mlx5e_accel_fs_esp_prot *fs_prot;
+	struct mlx5_flow_destination dest = {};
+	struct mlx5e_accel_fs_esp *accel_esp;
+	struct mlx5_flow_act flow_act = {};
+	struct mlx5_flow_handle *rule;
+	enum accel_fs_esp_type type;
+	struct mlx5_flow_spec *spec;
+	int err = 0;
+
+	accel_esp = priv->ipsec->rx_fs;
+	type = attrs->is_ipv6 ? ACCEL_FS_ESP6 : ACCEL_FS_ESP4;
+	fs_prot = &accel_esp->fs_prot[type];
+
+	err = rx_ft_get(priv, type);
+	if (err)
+		return err;
+
+	spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
+	if (!spec) {
+		err = -ENOMEM;
+		goto out_err;
+	}
+
+	setup_fte_common(attrs, ipsec_obj_id, spec, &flow_act);
+
+	/* Set bit[31] ipsec marker */
+	/* Set bit[23-0] ipsec_obj_id */
+	MLX5_SET(set_action_in, action, action_type, MLX5_ACTION_TYPE_SET);
+	MLX5_SET(set_action_in, action, field, MLX5_ACTION_IN_FIELD_METADATA_REG_B);
+	MLX5_SET(set_action_in, action, data, (ipsec_obj_id | BIT(31)));
+	MLX5_SET(set_action_in, action, offset, 0);
+	MLX5_SET(set_action_in, action, length, 32);
+
+	modify_hdr = mlx5_modify_header_alloc(priv->mdev, MLX5_FLOW_NAMESPACE_KERNEL,
+					      1, action);
+	if (IS_ERR(modify_hdr)) {
+		err = PTR_ERR(modify_hdr);
+		netdev_err(priv->netdev,
+			   "fail to alloc ipsec set modify_header_id err=%d\n", err);
+		modify_hdr = NULL;
+		goto out_err;
+	}
+
+	flow_act.action = MLX5_FLOW_CONTEXT_ACTION_FWD_DEST |
+			  MLX5_FLOW_CONTEXT_ACTION_CRYPTO_DECRYPT |
+			  MLX5_FLOW_CONTEXT_ACTION_MOD_HDR;
+	dest.type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
+	flow_act.modify_hdr = modify_hdr;
+	dest.ft = fs_prot->rx_err.ft;
+	rule = mlx5_add_flow_rules(fs_prot->ft, spec, &flow_act, &dest, 1);
+	if (IS_ERR(rule)) {
+		err = PTR_ERR(rule);
+		netdev_err(priv->netdev, "fail to add ipsec rule attrs->action=0x%x, err=%d\n",
+			   attrs->action, err);
+		goto out_err;
+	}
+
+	ipsec_rule->rule = rule;
+	ipsec_rule->set_modify_hdr = modify_hdr;
+	goto out;
+
+out_err:
+	if (modify_hdr)
+		mlx5_modify_header_dealloc(priv->mdev, modify_hdr);
+	rx_ft_put(priv, type);
+
+out:
+	kvfree(spec);
+	return err;
+}
+
+static int tx_add_rule(struct mlx5e_priv *priv,
+		       struct mlx5e_ipsec_sa_entry *sa_entry)
+{
+	struct mlx5_flow_act flow_act = {};
+	struct mlx5_flow_handle *rule;
+	struct mlx5_flow_spec *spec;
+	int err = 0;
+
+	err = tx_ft_get(priv);
+	if (err)
+		return err;
+
+	spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
+	if (!spec) {
+		err = -ENOMEM;
+		goto out;
+	}
+
+	setup_fte_common(&sa_entry->attrs, sa_entry->ipsec_obj_id, spec,
+			 &flow_act);
+
+	/* Add IPsec indicator in metadata_reg_a */
+	spec->match_criteria_enable |= MLX5_MATCH_MISC_PARAMETERS_2;
+	MLX5_SET(fte_match_param, spec->match_criteria, misc_parameters_2.metadata_reg_a,
+		 MLX5_ETH_WQE_FT_META_IPSEC);
+	MLX5_SET(fte_match_param, spec->match_value, misc_parameters_2.metadata_reg_a,
+		 MLX5_ETH_WQE_FT_META_IPSEC);
+
+	flow_act.action = MLX5_FLOW_CONTEXT_ACTION_ALLOW |
+			  MLX5_FLOW_CONTEXT_ACTION_CRYPTO_ENCRYPT;
+	rule = mlx5_add_flow_rules(priv->ipsec->tx_fs->ft, spec, &flow_act, NULL, 0);
+	if (IS_ERR(rule)) {
+		err = PTR_ERR(rule);
+		netdev_err(priv->netdev, "fail to add ipsec rule attrs->action=0x%x, err=%d\n",
+				sa_entry->attrs.action, err);
+		goto out;
+	}
+
+	sa_entry->ipsec_rule.rule = rule;
+
+out:
+	kvfree(spec);
+	if (err)
+		tx_ft_put(priv);
+	return err;
+}
+
+int mlx5e_accel_ipsec_fs_add_rule(struct mlx5e_priv *priv,
+				  struct mlx5e_ipsec_sa_entry *sa_entry)
+{
+	if (sa_entry->attrs.action == MLX5_ACCEL_ESP_ACTION_ENCRYPT)
+		return tx_add_rule(priv, sa_entry);
+
+	return rx_add_rule(priv, sa_entry);
+}
+
+void mlx5e_accel_ipsec_fs_del_rule(struct mlx5e_priv *priv,
+				   struct mlx5e_ipsec_sa_entry *sa_entry)
+{
+	struct mlx5e_ipsec_rule *ipsec_rule = &sa_entry->ipsec_rule;
+	struct mlx5_core_dev *mdev = mlx5e_ipsec_sa2dev(sa_entry);
+
+	mlx5_del_flow_rules(ipsec_rule->rule);
+
+	if (sa_entry->attrs.action == MLX5_ACCEL_ESP_ACTION_ENCRYPT) {
+		tx_ft_put(priv);
+		return;
+	}
+
+	mlx5_modify_header_dealloc(mdev, ipsec_rule->set_modify_hdr);
+	rx_ft_put(priv,
+		  sa_entry->attrs.is_ipv6 ? ACCEL_FS_ESP6 : ACCEL_FS_ESP4);
+}
+
+void mlx5e_accel_ipsec_fs_cleanup(struct mlx5e_ipsec *ipsec)
+{
+	struct mlx5e_accel_fs_esp_prot *fs_prot;
+	struct mlx5e_accel_fs_esp *accel_esp;
+	enum accel_fs_esp_type i;
+
+	if (!ipsec->rx_fs)
+		return;
+
+	mutex_destroy(&ipsec->tx_fs->mutex);
+	WARN_ON(ipsec->tx_fs->refcnt);
+	kfree(ipsec->tx_fs);
+
+	accel_esp = ipsec->rx_fs;
+	for (i = 0; i < ACCEL_FS_ESP_NUM_TYPES; i++) {
+		fs_prot = &accel_esp->fs_prot[i];
+		mutex_destroy(&fs_prot->prot_mutex);
+		WARN_ON(fs_prot->refcnt);
+	}
+	kfree(ipsec->rx_fs);
+}
+
+int mlx5e_accel_ipsec_fs_init(struct mlx5e_ipsec *ipsec)
+{
+	struct mlx5e_accel_fs_esp_prot *fs_prot;
+	struct mlx5e_accel_fs_esp *accel_esp;
+	struct mlx5_flow_namespace *ns;
+	enum accel_fs_esp_type i;
+	int err = -ENOMEM;
+
+	ns = mlx5_get_flow_namespace(ipsec->mdev,
+				     MLX5_FLOW_NAMESPACE_EGRESS_IPSEC);
+	if (!ns)
+		return -EOPNOTSUPP;
+
+	ipsec->tx_fs = kzalloc(sizeof(*ipsec->tx_fs), GFP_KERNEL);
+	if (!ipsec->tx_fs)
+		return -ENOMEM;
+
+	ipsec->rx_fs = kzalloc(sizeof(*ipsec->rx_fs), GFP_KERNEL);
+	if (!ipsec->rx_fs)
+		goto err_rx;
+
+	mutex_init(&ipsec->tx_fs->mutex);
+	ipsec->tx_fs->ns = ns;
+
+	accel_esp = ipsec->rx_fs;
+	for (i = 0; i < ACCEL_FS_ESP_NUM_TYPES; i++) {
+		fs_prot = &accel_esp->fs_prot[i];
+		mutex_init(&fs_prot->prot_mutex);
+	}
+
+	return 0;
+
+err_rx:
+	kfree(ipsec->tx_fs);
+	return err;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_offload.c b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_offload.c
new file mode 100644
index 000000000..792724ce7
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_offload.c
@@ -0,0 +1,205 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2017, Mellanox Technologies inc. All rights reserved. */
+
+#include "mlx5_core.h"
+#include "ipsec.h"
+#include "lib/mlx5.h"
+
+u32 mlx5_ipsec_device_caps(struct mlx5_core_dev *mdev)
+{
+	u32 caps = 0;
+
+	if (!MLX5_CAP_GEN(mdev, ipsec_offload))
+		return 0;
+
+	if (!MLX5_CAP_GEN(mdev, log_max_dek))
+		return 0;
+
+	if (!(MLX5_CAP_GEN_64(mdev, general_obj_types) &
+	    MLX5_HCA_CAP_GENERAL_OBJECT_TYPES_IPSEC))
+		return 0;
+
+	if (!MLX5_CAP_FLOWTABLE_NIC_TX(mdev, ipsec_encrypt) ||
+	    !MLX5_CAP_FLOWTABLE_NIC_RX(mdev, ipsec_decrypt))
+		return 0;
+
+	if (!MLX5_CAP_IPSEC(mdev, ipsec_crypto_esp_aes_gcm_128_encrypt) ||
+	    !MLX5_CAP_IPSEC(mdev, ipsec_crypto_esp_aes_gcm_128_decrypt))
+		return 0;
+
+	if (MLX5_CAP_IPSEC(mdev, ipsec_crypto_offload) &&
+	    MLX5_CAP_ETH(mdev, insert_trailer) && MLX5_CAP_ETH(mdev, swp))
+		caps |= MLX5_IPSEC_CAP_CRYPTO;
+
+	if (!caps)
+		return 0;
+
+	if (MLX5_CAP_IPSEC(mdev, ipsec_esn))
+		caps |= MLX5_IPSEC_CAP_ESN;
+
+	/* We can accommodate up to 2^24 different IPsec objects
+	 * because we use up to 24 bit in flow table metadata
+	 * to hold the IPsec Object unique handle.
+	 */
+	WARN_ON_ONCE(MLX5_CAP_IPSEC(mdev, log_max_ipsec_offload) > 24);
+	return caps;
+}
+EXPORT_SYMBOL_GPL(mlx5_ipsec_device_caps);
+
+static int mlx5_create_ipsec_obj(struct mlx5e_ipsec_sa_entry *sa_entry)
+{
+	struct mlx5_accel_esp_xfrm_attrs *attrs = &sa_entry->attrs;
+	struct mlx5_core_dev *mdev = mlx5e_ipsec_sa2dev(sa_entry);
+	struct aes_gcm_keymat *aes_gcm = &attrs->aes_gcm;
+	u32 out[MLX5_ST_SZ_DW(general_obj_out_cmd_hdr)];
+	u32 in[MLX5_ST_SZ_DW(create_ipsec_obj_in)] = {};
+	void *obj, *salt_p, *salt_iv_p;
+	int err;
+
+	obj = MLX5_ADDR_OF(create_ipsec_obj_in, in, ipsec_object);
+
+	/* salt and seq_iv */
+	salt_p = MLX5_ADDR_OF(ipsec_obj, obj, salt);
+	memcpy(salt_p, &aes_gcm->salt, sizeof(aes_gcm->salt));
+
+	MLX5_SET(ipsec_obj, obj, icv_length, MLX5_IPSEC_OBJECT_ICV_LEN_16B);
+	salt_iv_p = MLX5_ADDR_OF(ipsec_obj, obj, implicit_iv);
+	memcpy(salt_iv_p, &aes_gcm->seq_iv, sizeof(aes_gcm->seq_iv));
+	/* esn */
+	if (attrs->flags & MLX5_ACCEL_ESP_FLAGS_ESN_TRIGGERED) {
+		MLX5_SET(ipsec_obj, obj, esn_en, 1);
+		MLX5_SET(ipsec_obj, obj, esn_msb, attrs->esn);
+		if (attrs->flags & MLX5_ACCEL_ESP_FLAGS_ESN_STATE_OVERLAP)
+			MLX5_SET(ipsec_obj, obj, esn_overlap, 1);
+	}
+
+	MLX5_SET(ipsec_obj, obj, dekn, sa_entry->enc_key_id);
+
+	/* general object fields set */
+	MLX5_SET(general_obj_in_cmd_hdr, in, opcode,
+		 MLX5_CMD_OP_CREATE_GENERAL_OBJECT);
+	MLX5_SET(general_obj_in_cmd_hdr, in, obj_type,
+		 MLX5_GENERAL_OBJECT_TYPES_IPSEC);
+
+	err = mlx5_cmd_exec(mdev, in, sizeof(in), out, sizeof(out));
+	if (!err)
+		sa_entry->ipsec_obj_id =
+			MLX5_GET(general_obj_out_cmd_hdr, out, obj_id);
+
+	return err;
+}
+
+static void mlx5_destroy_ipsec_obj(struct mlx5e_ipsec_sa_entry *sa_entry)
+{
+	struct mlx5_core_dev *mdev = mlx5e_ipsec_sa2dev(sa_entry);
+	u32 in[MLX5_ST_SZ_DW(general_obj_in_cmd_hdr)] = {};
+	u32 out[MLX5_ST_SZ_DW(general_obj_out_cmd_hdr)];
+
+	MLX5_SET(general_obj_in_cmd_hdr, in, opcode,
+		 MLX5_CMD_OP_DESTROY_GENERAL_OBJECT);
+	MLX5_SET(general_obj_in_cmd_hdr, in, obj_type,
+		 MLX5_GENERAL_OBJECT_TYPES_IPSEC);
+	MLX5_SET(general_obj_in_cmd_hdr, in, obj_id, sa_entry->ipsec_obj_id);
+
+	mlx5_cmd_exec(mdev, in, sizeof(in), out, sizeof(out));
+}
+
+int mlx5_ipsec_create_sa_ctx(struct mlx5e_ipsec_sa_entry *sa_entry)
+{
+	struct aes_gcm_keymat *aes_gcm = &sa_entry->attrs.aes_gcm;
+	struct mlx5_core_dev *mdev = mlx5e_ipsec_sa2dev(sa_entry);
+	int err;
+
+	/* key */
+	err = mlx5_create_encryption_key(mdev, aes_gcm->aes_key,
+					 aes_gcm->key_len / BITS_PER_BYTE,
+					 MLX5_ACCEL_OBJ_IPSEC_KEY,
+					 &sa_entry->enc_key_id);
+	if (err) {
+		mlx5_core_dbg(mdev, "Failed to create encryption key (err = %d)\n", err);
+		return err;
+	}
+
+	err = mlx5_create_ipsec_obj(sa_entry);
+	if (err) {
+		mlx5_core_dbg(mdev, "Failed to create IPsec object (err = %d)\n", err);
+		goto err_enc_key;
+	}
+
+	return 0;
+
+err_enc_key:
+	mlx5_destroy_encryption_key(mdev, sa_entry->enc_key_id);
+	return err;
+}
+
+void mlx5_ipsec_free_sa_ctx(struct mlx5e_ipsec_sa_entry *sa_entry)
+{
+	struct mlx5_core_dev *mdev = mlx5e_ipsec_sa2dev(sa_entry);
+
+	mlx5_destroy_ipsec_obj(sa_entry);
+	mlx5_destroy_encryption_key(mdev, sa_entry->enc_key_id);
+}
+
+static int mlx5_modify_ipsec_obj(struct mlx5e_ipsec_sa_entry *sa_entry,
+				 const struct mlx5_accel_esp_xfrm_attrs *attrs)
+{
+	struct mlx5_core_dev *mdev = mlx5e_ipsec_sa2dev(sa_entry);
+	u32 in[MLX5_ST_SZ_DW(modify_ipsec_obj_in)] = {};
+	u32 out[MLX5_ST_SZ_DW(query_ipsec_obj_out)];
+	u64 modify_field_select = 0;
+	u64 general_obj_types;
+	void *obj;
+	int err;
+
+	if (!(attrs->flags & MLX5_ACCEL_ESP_FLAGS_ESN_TRIGGERED))
+		return 0;
+
+	general_obj_types = MLX5_CAP_GEN_64(mdev, general_obj_types);
+	if (!(general_obj_types & MLX5_HCA_CAP_GENERAL_OBJECT_TYPES_IPSEC))
+		return -EINVAL;
+
+	/* general object fields set */
+	MLX5_SET(general_obj_in_cmd_hdr, in, opcode, MLX5_CMD_OP_QUERY_GENERAL_OBJECT);
+	MLX5_SET(general_obj_in_cmd_hdr, in, obj_type, MLX5_GENERAL_OBJECT_TYPES_IPSEC);
+	MLX5_SET(general_obj_in_cmd_hdr, in, obj_id, sa_entry->ipsec_obj_id);
+	err = mlx5_cmd_exec(mdev, in, sizeof(in), out, sizeof(out));
+	if (err) {
+		mlx5_core_err(mdev, "Query IPsec object failed (Object id %d), err = %d\n",
+			      sa_entry->ipsec_obj_id, err);
+		return err;
+	}
+
+	obj = MLX5_ADDR_OF(query_ipsec_obj_out, out, ipsec_object);
+	modify_field_select = MLX5_GET64(ipsec_obj, obj, modify_field_select);
+
+	/* esn */
+	if (!(modify_field_select & MLX5_MODIFY_IPSEC_BITMASK_ESN_OVERLAP) ||
+	    !(modify_field_select & MLX5_MODIFY_IPSEC_BITMASK_ESN_MSB))
+		return -EOPNOTSUPP;
+
+	obj = MLX5_ADDR_OF(modify_ipsec_obj_in, in, ipsec_object);
+	MLX5_SET64(ipsec_obj, obj, modify_field_select,
+		   MLX5_MODIFY_IPSEC_BITMASK_ESN_OVERLAP |
+			   MLX5_MODIFY_IPSEC_BITMASK_ESN_MSB);
+	MLX5_SET(ipsec_obj, obj, esn_msb, attrs->esn);
+	if (attrs->flags & MLX5_ACCEL_ESP_FLAGS_ESN_STATE_OVERLAP)
+		MLX5_SET(ipsec_obj, obj, esn_overlap, 1);
+
+	/* general object fields set */
+	MLX5_SET(general_obj_in_cmd_hdr, in, opcode, MLX5_CMD_OP_MODIFY_GENERAL_OBJECT);
+
+	return mlx5_cmd_exec(mdev, in, sizeof(in), out, sizeof(out));
+}
+
+void mlx5_accel_esp_modify_xfrm(struct mlx5e_ipsec_sa_entry *sa_entry,
+				const struct mlx5_accel_esp_xfrm_attrs *attrs)
+{
+	int err;
+
+	err = mlx5_modify_ipsec_obj(sa_entry, attrs);
+	if (err)
+		return;
+
+	memcpy(&sa_entry->attrs, attrs, sizeof(sa_entry->attrs));
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_rxtx.c b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_rxtx.c
new file mode 100644
index 000000000..c4a84f0a3
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_rxtx.c
@@ -0,0 +1,356 @@
+/*
+ * Copyright (c) 2017 Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+
+#include <crypto/aead.h>
+#include <net/xfrm.h>
+#include <net/esp.h>
+#include "ipsec.h"
+#include "ipsec_rxtx.h"
+#include "en.h"
+
+enum {
+	MLX5E_IPSEC_TX_SYNDROME_OFFLOAD = 0x8,
+	MLX5E_IPSEC_TX_SYNDROME_OFFLOAD_WITH_LSO_TCP = 0x9,
+};
+
+static int mlx5e_ipsec_remove_trailer(struct sk_buff *skb, struct xfrm_state *x)
+{
+	unsigned int alen = crypto_aead_authsize(x->data);
+	struct ipv6hdr *ipv6hdr = ipv6_hdr(skb);
+	struct iphdr *ipv4hdr = ip_hdr(skb);
+	unsigned int trailer_len;
+	u8 plen;
+	int ret;
+
+	ret = skb_copy_bits(skb, skb->len - alen - 2, &plen, 1);
+	if (unlikely(ret))
+		return ret;
+
+	trailer_len = alen + plen + 2;
+
+	ret = pskb_trim(skb, skb->len - trailer_len);
+	if (unlikely(ret))
+		return ret;
+	if (skb->protocol == htons(ETH_P_IP)) {
+		ipv4hdr->tot_len = htons(ntohs(ipv4hdr->tot_len) - trailer_len);
+		ip_send_check(ipv4hdr);
+	} else {
+		ipv6hdr->payload_len = htons(ntohs(ipv6hdr->payload_len) -
+					     trailer_len);
+	}
+	return 0;
+}
+
+static void mlx5e_ipsec_set_swp(struct sk_buff *skb,
+				struct mlx5_wqe_eth_seg *eseg, u8 mode,
+				struct xfrm_offload *xo)
+{
+	/* Tunnel Mode:
+	 * SWP:      OutL3       InL3  InL4
+	 * Pkt: MAC  IP     ESP  IP    L4
+	 *
+	 * Transport Mode:
+	 * SWP:      OutL3       OutL4
+	 * Pkt: MAC  IP     ESP  L4
+	 *
+	 * Tunnel(VXLAN TCP/UDP) over Transport Mode
+	 * SWP:      OutL3                   InL3  InL4
+	 * Pkt: MAC  IP     ESP  UDP  VXLAN  IP    L4
+	 */
+
+	/* Shared settings */
+	eseg->swp_outer_l3_offset = skb_network_offset(skb) / 2;
+	if (skb->protocol == htons(ETH_P_IPV6))
+		eseg->swp_flags |= MLX5_ETH_WQE_SWP_OUTER_L3_IPV6;
+
+	/* Tunnel mode */
+	if (mode == XFRM_MODE_TUNNEL) {
+		eseg->swp_inner_l3_offset = skb_inner_network_offset(skb) / 2;
+		if (xo->proto == IPPROTO_IPV6)
+			eseg->swp_flags |= MLX5_ETH_WQE_SWP_INNER_L3_IPV6;
+
+		switch (xo->inner_ipproto) {
+		case IPPROTO_UDP:
+			eseg->swp_flags |= MLX5_ETH_WQE_SWP_INNER_L4_UDP;
+			fallthrough;
+		case IPPROTO_TCP:
+			/* IP | ESP | IP | [TCP | UDP] */
+			eseg->swp_inner_l4_offset = skb_inner_transport_offset(skb) / 2;
+			break;
+		default:
+			break;
+		}
+		return;
+	}
+
+	/* Transport mode */
+	if (mode != XFRM_MODE_TRANSPORT)
+		return;
+
+	if (!xo->inner_ipproto) {
+		switch (xo->proto) {
+		case IPPROTO_UDP:
+			eseg->swp_flags |= MLX5_ETH_WQE_SWP_OUTER_L4_UDP;
+			fallthrough;
+		case IPPROTO_TCP:
+			/* IP | ESP | TCP */
+			eseg->swp_outer_l4_offset = skb_inner_transport_offset(skb) / 2;
+			break;
+		default:
+			break;
+		}
+	} else {
+		/* Tunnel(VXLAN TCP/UDP) over Transport Mode */
+		switch (xo->inner_ipproto) {
+		case IPPROTO_UDP:
+			eseg->swp_flags |= MLX5_ETH_WQE_SWP_INNER_L4_UDP;
+			fallthrough;
+		case IPPROTO_TCP:
+			eseg->swp_inner_l3_offset = skb_inner_network_offset(skb) / 2;
+			eseg->swp_inner_l4_offset =
+				(skb->csum_start + skb->head - skb->data) / 2;
+			if (inner_ip_hdr(skb)->version == 6)
+				eseg->swp_flags |= MLX5_ETH_WQE_SWP_INNER_L3_IPV6;
+			break;
+		default:
+			break;
+		}
+	}
+
+}
+
+void mlx5e_ipsec_set_iv_esn(struct sk_buff *skb, struct xfrm_state *x,
+			    struct xfrm_offload *xo)
+{
+	struct xfrm_replay_state_esn *replay_esn = x->replay_esn;
+	__u32 oseq = replay_esn->oseq;
+	int iv_offset;
+	__be64 seqno;
+	u32 seq_hi;
+
+	if (unlikely(skb_is_gso(skb) && oseq < MLX5E_IPSEC_ESN_SCOPE_MID &&
+		     MLX5E_IPSEC_ESN_SCOPE_MID < (oseq - skb_shinfo(skb)->gso_segs))) {
+		seq_hi = xo->seq.hi - 1;
+	} else {
+		seq_hi = xo->seq.hi;
+	}
+
+	/* Place the SN in the IV field */
+	seqno = cpu_to_be64(xo->seq.low + ((u64)seq_hi << 32));
+	iv_offset = skb_transport_offset(skb) + sizeof(struct ip_esp_hdr);
+	skb_store_bits(skb, iv_offset, &seqno, 8);
+}
+
+void mlx5e_ipsec_set_iv(struct sk_buff *skb, struct xfrm_state *x,
+			struct xfrm_offload *xo)
+{
+	int iv_offset;
+	__be64 seqno;
+
+	/* Place the SN in the IV field */
+	seqno = cpu_to_be64(xo->seq.low + ((u64)xo->seq.hi << 32));
+	iv_offset = skb_transport_offset(skb) + sizeof(struct ip_esp_hdr);
+	skb_store_bits(skb, iv_offset, &seqno, 8);
+}
+
+void mlx5e_ipsec_handle_tx_wqe(struct mlx5e_tx_wqe *wqe,
+			       struct mlx5e_accel_tx_ipsec_state *ipsec_st,
+			       struct mlx5_wqe_inline_seg *inlseg)
+{
+	inlseg->byte_count = cpu_to_be32(ipsec_st->tailen | MLX5_INLINE_SEG);
+	esp_output_fill_trailer((u8 *)inlseg->data, 0, ipsec_st->plen, ipsec_st->xo->proto);
+}
+
+static int mlx5e_ipsec_set_state(struct mlx5e_priv *priv,
+				 struct sk_buff *skb,
+				 struct xfrm_state *x,
+				 struct xfrm_offload *xo,
+				 struct mlx5e_accel_tx_ipsec_state *ipsec_st)
+{
+	unsigned int blksize, clen, alen, plen;
+	struct crypto_aead *aead;
+	unsigned int tailen;
+
+	ipsec_st->x = x;
+	ipsec_st->xo = xo;
+	aead = x->data;
+	alen = crypto_aead_authsize(aead);
+	blksize = ALIGN(crypto_aead_blocksize(aead), 4);
+	clen = ALIGN(skb->len + 2, blksize);
+	plen = max_t(u32, clen - skb->len, 4);
+	tailen = plen + alen;
+	ipsec_st->plen = plen;
+	ipsec_st->tailen = tailen;
+
+	return 0;
+}
+
+void mlx5e_ipsec_tx_build_eseg(struct mlx5e_priv *priv, struct sk_buff *skb,
+			       struct mlx5_wqe_eth_seg *eseg)
+{
+	struct xfrm_offload *xo = xfrm_offload(skb);
+	struct xfrm_encap_tmpl  *encap;
+	struct xfrm_state *x;
+	struct sec_path *sp;
+	u8 l3_proto;
+
+	sp = skb_sec_path(skb);
+	if (unlikely(sp->len != 1))
+		return;
+
+	x = xfrm_input_state(skb);
+	if (unlikely(!x))
+		return;
+
+	if (unlikely(!x->xso.offload_handle ||
+		     (skb->protocol != htons(ETH_P_IP) &&
+		      skb->protocol != htons(ETH_P_IPV6))))
+		return;
+
+	mlx5e_ipsec_set_swp(skb, eseg, x->props.mode, xo);
+
+	l3_proto = (x->props.family == AF_INET) ?
+		   ((struct iphdr *)skb_network_header(skb))->protocol :
+		   ((struct ipv6hdr *)skb_network_header(skb))->nexthdr;
+
+	eseg->flow_table_metadata |= cpu_to_be32(MLX5_ETH_WQE_FT_META_IPSEC);
+	eseg->trailer |= cpu_to_be32(MLX5_ETH_WQE_INSERT_TRAILER);
+	encap = x->encap;
+	if (!encap) {
+		eseg->trailer |= (l3_proto == IPPROTO_ESP) ?
+			cpu_to_be32(MLX5_ETH_WQE_TRAILER_HDR_OUTER_IP_ASSOC) :
+			cpu_to_be32(MLX5_ETH_WQE_TRAILER_HDR_OUTER_L4_ASSOC);
+	} else if (encap->encap_type == UDP_ENCAP_ESPINUDP) {
+		eseg->trailer |= (l3_proto == IPPROTO_ESP) ?
+			cpu_to_be32(MLX5_ETH_WQE_TRAILER_HDR_INNER_IP_ASSOC) :
+			cpu_to_be32(MLX5_ETH_WQE_TRAILER_HDR_INNER_L4_ASSOC);
+	}
+}
+
+bool mlx5e_ipsec_handle_tx_skb(struct net_device *netdev,
+			       struct sk_buff *skb,
+			       struct mlx5e_accel_tx_ipsec_state *ipsec_st)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	struct xfrm_offload *xo = xfrm_offload(skb);
+	struct mlx5e_ipsec_sa_entry *sa_entry;
+	struct xfrm_state *x;
+	struct sec_path *sp;
+
+	sp = skb_sec_path(skb);
+	if (unlikely(sp->len != 1)) {
+		atomic64_inc(&priv->ipsec->sw_stats.ipsec_tx_drop_bundle);
+		goto drop;
+	}
+
+	x = xfrm_input_state(skb);
+	if (unlikely(!x)) {
+		atomic64_inc(&priv->ipsec->sw_stats.ipsec_tx_drop_no_state);
+		goto drop;
+	}
+
+	if (unlikely(!x->xso.offload_handle ||
+		     (skb->protocol != htons(ETH_P_IP) &&
+		      skb->protocol != htons(ETH_P_IPV6)))) {
+		atomic64_inc(&priv->ipsec->sw_stats.ipsec_tx_drop_not_ip);
+		goto drop;
+	}
+
+	if (!skb_is_gso(skb))
+		if (unlikely(mlx5e_ipsec_remove_trailer(skb, x))) {
+			atomic64_inc(&priv->ipsec->sw_stats.ipsec_tx_drop_trailer);
+			goto drop;
+		}
+
+	sa_entry = (struct mlx5e_ipsec_sa_entry *)x->xso.offload_handle;
+	sa_entry->set_iv_op(skb, x, xo);
+	mlx5e_ipsec_set_state(priv, skb, x, xo, ipsec_st);
+
+	return true;
+
+drop:
+	kfree_skb(skb);
+	return false;
+}
+
+enum {
+	MLX5E_IPSEC_OFFLOAD_RX_SYNDROME_DECRYPTED,
+	MLX5E_IPSEC_OFFLOAD_RX_SYNDROME_AUTH_FAILED,
+	MLX5E_IPSEC_OFFLOAD_RX_SYNDROME_BAD_TRAILER,
+};
+
+void mlx5e_ipsec_offload_handle_rx_skb(struct net_device *netdev,
+				       struct sk_buff *skb,
+				       struct mlx5_cqe64 *cqe)
+{
+	u32 ipsec_meta_data = be32_to_cpu(cqe->ft_metadata);
+	struct mlx5e_priv *priv;
+	struct xfrm_offload *xo;
+	struct xfrm_state *xs;
+	struct sec_path *sp;
+	u32  sa_handle;
+
+	sa_handle = MLX5_IPSEC_METADATA_HANDLE(ipsec_meta_data);
+	priv = netdev_priv(netdev);
+	sp = secpath_set(skb);
+	if (unlikely(!sp)) {
+		atomic64_inc(&priv->ipsec->sw_stats.ipsec_rx_drop_sp_alloc);
+		return;
+	}
+
+	xs = mlx5e_ipsec_sadb_rx_lookup(priv->ipsec, sa_handle);
+	if (unlikely(!xs)) {
+		atomic64_inc(&priv->ipsec->sw_stats.ipsec_rx_drop_sadb_miss);
+		return;
+	}
+
+	sp->xvec[sp->len++] = xs;
+	sp->olen++;
+
+	xo = xfrm_offload(skb);
+	xo->flags = CRYPTO_DONE;
+
+	switch (MLX5_IPSEC_METADATA_SYNDROM(ipsec_meta_data)) {
+	case MLX5E_IPSEC_OFFLOAD_RX_SYNDROME_DECRYPTED:
+		xo->status = CRYPTO_SUCCESS;
+		break;
+	case MLX5E_IPSEC_OFFLOAD_RX_SYNDROME_AUTH_FAILED:
+		xo->status = CRYPTO_TUNNEL_ESP_AUTH_FAILED;
+		break;
+	case MLX5E_IPSEC_OFFLOAD_RX_SYNDROME_BAD_TRAILER:
+		xo->status = CRYPTO_INVALID_PACKET_SYNTAX;
+		break;
+	default:
+		atomic64_inc(&priv->ipsec->sw_stats.ipsec_rx_drop_syndrome);
+	}
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_rxtx.h b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_rxtx.h
new file mode 100644
index 000000000..1878a70b9
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_rxtx.h
@@ -0,0 +1,169 @@
+/*
+ * Copyright (c) 2017 Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+
+#ifndef __MLX5E_IPSEC_RXTX_H__
+#define __MLX5E_IPSEC_RXTX_H__
+
+#include <linux/skbuff.h>
+#include <net/xfrm.h>
+#include "en.h"
+#include "en/txrx.h"
+
+/* Bit31: IPsec marker, Bit30: reserved, Bit29-24: IPsec syndrome, Bit23-0: IPsec obj id */
+#define MLX5_IPSEC_METADATA_MARKER(metadata)  (((metadata) >> 31) & 0x1)
+#define MLX5_IPSEC_METADATA_SYNDROM(metadata) (((metadata) >> 24) & GENMASK(5, 0))
+#define MLX5_IPSEC_METADATA_HANDLE(metadata)  ((metadata) & GENMASK(23, 0))
+
+struct mlx5e_accel_tx_ipsec_state {
+	struct xfrm_offload *xo;
+	struct xfrm_state *x;
+	u32 tailen;
+	u32 plen;
+};
+
+#ifdef CONFIG_MLX5_EN_IPSEC
+
+void mlx5e_ipsec_inverse_table_init(void);
+void mlx5e_ipsec_set_iv_esn(struct sk_buff *skb, struct xfrm_state *x,
+			    struct xfrm_offload *xo);
+void mlx5e_ipsec_set_iv(struct sk_buff *skb, struct xfrm_state *x,
+			struct xfrm_offload *xo);
+bool mlx5e_ipsec_handle_tx_skb(struct net_device *netdev,
+			       struct sk_buff *skb,
+			       struct mlx5e_accel_tx_ipsec_state *ipsec_st);
+void mlx5e_ipsec_handle_tx_wqe(struct mlx5e_tx_wqe *wqe,
+			       struct mlx5e_accel_tx_ipsec_state *ipsec_st,
+			       struct mlx5_wqe_inline_seg *inlseg);
+void mlx5e_ipsec_offload_handle_rx_skb(struct net_device *netdev,
+				       struct sk_buff *skb,
+				       struct mlx5_cqe64 *cqe);
+static inline unsigned int mlx5e_ipsec_tx_ids_len(struct mlx5e_accel_tx_ipsec_state *ipsec_st)
+{
+	return ipsec_st->tailen;
+}
+
+static inline bool mlx5_ipsec_is_rx_flow(struct mlx5_cqe64 *cqe)
+{
+	return MLX5_IPSEC_METADATA_MARKER(be32_to_cpu(cqe->ft_metadata));
+}
+
+static inline bool mlx5e_ipsec_eseg_meta(struct mlx5_wqe_eth_seg *eseg)
+{
+	return eseg->flow_table_metadata & cpu_to_be32(MLX5_ETH_WQE_FT_META_IPSEC);
+}
+
+void mlx5e_ipsec_tx_build_eseg(struct mlx5e_priv *priv, struct sk_buff *skb,
+			       struct mlx5_wqe_eth_seg *eseg);
+
+static inline netdev_features_t
+mlx5e_ipsec_feature_check(struct sk_buff *skb, netdev_features_t features)
+{
+	struct xfrm_offload *xo = xfrm_offload(skb);
+	struct sec_path *sp = skb_sec_path(skb);
+
+	if (sp && sp->len && xo) {
+		struct xfrm_state *x = sp->xvec[0];
+
+		if (!x || !x->xso.offload_handle)
+			goto out_disable;
+
+		if (xo->inner_ipproto) {
+			/* Cannot support tunnel packet over IPsec tunnel mode
+			 * because we cannot offload three IP header csum
+			 */
+			if (x->props.mode == XFRM_MODE_TUNNEL)
+				goto out_disable;
+
+			/* Only support UDP or TCP L4 checksum */
+			if (xo->inner_ipproto != IPPROTO_UDP &&
+			    xo->inner_ipproto != IPPROTO_TCP)
+				goto out_disable;
+		}
+
+		return features;
+
+	}
+
+	/* Disable CSUM and GSO for software IPsec */
+out_disable:
+	return features & ~(NETIF_F_CSUM_MASK | NETIF_F_GSO_MASK);
+}
+
+static inline bool
+mlx5e_ipsec_txwqe_build_eseg_csum(struct mlx5e_txqsq *sq, struct sk_buff *skb,
+				  struct mlx5_wqe_eth_seg *eseg)
+{
+	u8 inner_ipproto;
+
+	if (!mlx5e_ipsec_eseg_meta(eseg))
+		return false;
+
+	eseg->cs_flags = MLX5_ETH_WQE_L3_CSUM;
+	inner_ipproto = xfrm_offload(skb)->inner_ipproto;
+	if (inner_ipproto) {
+		eseg->cs_flags |= MLX5_ETH_WQE_L3_INNER_CSUM;
+		if (inner_ipproto == IPPROTO_TCP || inner_ipproto == IPPROTO_UDP)
+			eseg->cs_flags |= MLX5_ETH_WQE_L4_INNER_CSUM;
+	} else if (likely(skb->ip_summed == CHECKSUM_PARTIAL)) {
+		eseg->cs_flags |= MLX5_ETH_WQE_L4_CSUM;
+		sq->stats->csum_partial_inner++;
+	}
+
+	return true;
+}
+#else
+static inline
+void mlx5e_ipsec_offload_handle_rx_skb(struct net_device *netdev,
+				       struct sk_buff *skb,
+				       struct mlx5_cqe64 *cqe)
+{}
+
+static inline bool mlx5e_ipsec_eseg_meta(struct mlx5_wqe_eth_seg *eseg)
+{
+	return false;
+}
+
+static inline bool mlx5_ipsec_is_rx_flow(struct mlx5_cqe64 *cqe) { return false; }
+static inline netdev_features_t
+mlx5e_ipsec_feature_check(struct sk_buff *skb, netdev_features_t features)
+{ return features & ~(NETIF_F_CSUM_MASK | NETIF_F_GSO_MASK); }
+
+static inline bool
+mlx5e_ipsec_txwqe_build_eseg_csum(struct mlx5e_txqsq *sq, struct sk_buff *skb,
+				  struct mlx5_wqe_eth_seg *eseg)
+{
+	return false;
+}
+#endif /* CONFIG_MLX5_EN_IPSEC */
+
+#endif /* __MLX5E_IPSEC_RXTX_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_stats.c b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_stats.c
new file mode 100644
index 000000000..9de84821d
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_stats.c
@@ -0,0 +1,84 @@
+/*
+ * Copyright (c) 2017 Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+
+#include <linux/ethtool.h>
+#include <net/sock.h>
+
+#include "en.h"
+#include "ipsec.h"
+
+static const struct counter_desc mlx5e_ipsec_sw_stats_desc[] = {
+	{ MLX5E_DECLARE_STAT(struct mlx5e_ipsec_sw_stats, ipsec_rx_drop_sp_alloc) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_ipsec_sw_stats, ipsec_rx_drop_sadb_miss) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_ipsec_sw_stats, ipsec_rx_drop_syndrome) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_ipsec_sw_stats, ipsec_tx_drop_bundle) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_ipsec_sw_stats, ipsec_tx_drop_no_state) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_ipsec_sw_stats, ipsec_tx_drop_not_ip) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_ipsec_sw_stats, ipsec_tx_drop_trailer) },
+};
+
+#define MLX5E_READ_CTR_ATOMIC64(ptr, dsc, i) \
+	atomic64_read((atomic64_t *)((char *)(ptr) + (dsc)[i].offset))
+
+#define NUM_IPSEC_SW_COUNTERS ARRAY_SIZE(mlx5e_ipsec_sw_stats_desc)
+
+static MLX5E_DECLARE_STATS_GRP_OP_NUM_STATS(ipsec_sw)
+{
+	return priv->ipsec ? NUM_IPSEC_SW_COUNTERS : 0;
+}
+
+static inline MLX5E_DECLARE_STATS_GRP_OP_UPDATE_STATS(ipsec_sw) {}
+
+static MLX5E_DECLARE_STATS_GRP_OP_FILL_STRS(ipsec_sw)
+{
+	unsigned int i;
+
+	if (priv->ipsec)
+		for (i = 0; i < NUM_IPSEC_SW_COUNTERS; i++)
+			strcpy(data + (idx++) * ETH_GSTRING_LEN,
+			       mlx5e_ipsec_sw_stats_desc[i].format);
+	return idx;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_FILL_STATS(ipsec_sw)
+{
+	int i;
+
+	if (priv->ipsec)
+		for (i = 0; i < NUM_IPSEC_SW_COUNTERS; i++)
+			data[idx++] = MLX5E_READ_CTR_ATOMIC64(&priv->ipsec->sw_stats,
+							      mlx5e_ipsec_sw_stats_desc, i);
+	return idx;
+}
+
+MLX5E_DEFINE_STATS_GRP(ipsec_sw, 0);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls.c b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls.c
new file mode 100644
index 000000000..da2184c94
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls.c
@@ -0,0 +1,199 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+// Copyright (c) 2019 Mellanox Technologies.
+
+#include "en.h"
+#include "lib/mlx5.h"
+#include "en_accel/ktls.h"
+#include "en_accel/ktls_utils.h"
+#include "en_accel/fs_tcp.h"
+
+int mlx5_ktls_create_key(struct mlx5_core_dev *mdev,
+			 struct tls_crypto_info *crypto_info,
+			 u32 *p_key_id)
+{
+	u32 sz_bytes;
+	void *key;
+
+	switch (crypto_info->cipher_type) {
+	case TLS_CIPHER_AES_GCM_128: {
+		struct tls12_crypto_info_aes_gcm_128 *info =
+			(struct tls12_crypto_info_aes_gcm_128 *)crypto_info;
+
+		key      = info->key;
+		sz_bytes = sizeof(info->key);
+		break;
+	}
+	case TLS_CIPHER_AES_GCM_256: {
+		struct tls12_crypto_info_aes_gcm_256 *info =
+			(struct tls12_crypto_info_aes_gcm_256 *)crypto_info;
+
+		key      = info->key;
+		sz_bytes = sizeof(info->key);
+		break;
+	}
+	default:
+		return -EINVAL;
+	}
+
+	return mlx5_create_encryption_key(mdev, key, sz_bytes,
+					  MLX5_ACCEL_OBJ_TLS_KEY,
+					  p_key_id);
+}
+
+void mlx5_ktls_destroy_key(struct mlx5_core_dev *mdev, u32 key_id)
+{
+	mlx5_destroy_encryption_key(mdev, key_id);
+}
+
+static int mlx5e_ktls_add(struct net_device *netdev, struct sock *sk,
+			  enum tls_offload_ctx_dir direction,
+			  struct tls_crypto_info *crypto_info,
+			  u32 start_offload_tcp_sn)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+	int err;
+
+	if (!mlx5e_ktls_type_check(mdev, crypto_info))
+		return -EOPNOTSUPP;
+
+	if (direction == TLS_OFFLOAD_CTX_DIR_TX)
+		err = mlx5e_ktls_add_tx(netdev, sk, crypto_info, start_offload_tcp_sn);
+	else
+		err = mlx5e_ktls_add_rx(netdev, sk, crypto_info, start_offload_tcp_sn);
+
+	return err;
+}
+
+static void mlx5e_ktls_del(struct net_device *netdev,
+			   struct tls_context *tls_ctx,
+			   enum tls_offload_ctx_dir direction)
+{
+	if (direction == TLS_OFFLOAD_CTX_DIR_TX)
+		mlx5e_ktls_del_tx(netdev, tls_ctx);
+	else
+		mlx5e_ktls_del_rx(netdev, tls_ctx);
+}
+
+static int mlx5e_ktls_resync(struct net_device *netdev,
+			     struct sock *sk, u32 seq, u8 *rcd_sn,
+			     enum tls_offload_ctx_dir direction)
+{
+	if (unlikely(direction != TLS_OFFLOAD_CTX_DIR_RX))
+		return -EOPNOTSUPP;
+
+	mlx5e_ktls_rx_resync(netdev, sk, seq, rcd_sn);
+	return 0;
+}
+
+static const struct tlsdev_ops mlx5e_ktls_ops = {
+	.tls_dev_add = mlx5e_ktls_add,
+	.tls_dev_del = mlx5e_ktls_del,
+	.tls_dev_resync = mlx5e_ktls_resync,
+};
+
+bool mlx5e_is_ktls_rx(struct mlx5_core_dev *mdev)
+{
+	u8 max_sq_wqebbs = mlx5e_get_max_sq_wqebbs(mdev);
+
+	if (is_kdump_kernel() || !MLX5_CAP_GEN(mdev, tls_rx))
+		return false;
+
+	/* Check the possibility to post the required ICOSQ WQEs. */
+	if (WARN_ON_ONCE(max_sq_wqebbs < MLX5E_TLS_SET_STATIC_PARAMS_WQEBBS))
+		return false;
+	if (WARN_ON_ONCE(max_sq_wqebbs < MLX5E_TLS_SET_PROGRESS_PARAMS_WQEBBS))
+		return false;
+	if (WARN_ON_ONCE(max_sq_wqebbs < MLX5E_KTLS_GET_PROGRESS_WQEBBS))
+		return false;
+
+	return true;
+}
+
+void mlx5e_ktls_build_netdev(struct mlx5e_priv *priv)
+{
+	struct net_device *netdev = priv->netdev;
+	struct mlx5_core_dev *mdev = priv->mdev;
+
+	if (!mlx5e_is_ktls_tx(mdev) && !mlx5e_is_ktls_rx(mdev))
+		return;
+
+	if (mlx5e_is_ktls_tx(mdev)) {
+		netdev->hw_features |= NETIF_F_HW_TLS_TX;
+		netdev->features    |= NETIF_F_HW_TLS_TX;
+	}
+
+	if (mlx5e_is_ktls_rx(mdev))
+		netdev->hw_features |= NETIF_F_HW_TLS_RX;
+
+	netdev->tlsdev_ops = &mlx5e_ktls_ops;
+}
+
+int mlx5e_ktls_set_feature_rx(struct net_device *netdev, bool enable)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	int err = 0;
+
+	mutex_lock(&priv->state_lock);
+	if (enable)
+		err = mlx5e_accel_fs_tcp_create(priv->fs);
+	else
+		mlx5e_accel_fs_tcp_destroy(priv->fs);
+	mutex_unlock(&priv->state_lock);
+
+	return err;
+}
+
+int mlx5e_ktls_init_rx(struct mlx5e_priv *priv)
+{
+	int err;
+
+	if (!mlx5e_is_ktls_rx(priv->mdev))
+		return 0;
+
+	priv->tls->rx_wq = create_singlethread_workqueue("mlx5e_tls_rx");
+	if (!priv->tls->rx_wq)
+		return -ENOMEM;
+
+	if (priv->netdev->features & NETIF_F_HW_TLS_RX) {
+		err = mlx5e_accel_fs_tcp_create(priv->fs);
+		if (err) {
+			destroy_workqueue(priv->tls->rx_wq);
+			return err;
+		}
+	}
+
+	return 0;
+}
+
+void mlx5e_ktls_cleanup_rx(struct mlx5e_priv *priv)
+{
+	if (!mlx5e_is_ktls_rx(priv->mdev))
+		return;
+
+	if (priv->netdev->features & NETIF_F_HW_TLS_RX)
+		mlx5e_accel_fs_tcp_destroy(priv->fs);
+
+	destroy_workqueue(priv->tls->rx_wq);
+}
+
+int mlx5e_ktls_init(struct mlx5e_priv *priv)
+{
+	struct mlx5e_tls *tls;
+
+	if (!mlx5e_is_ktls_device(priv->mdev))
+		return 0;
+
+	tls = kzalloc(sizeof(*tls), GFP_KERNEL);
+	if (!tls)
+		return -ENOMEM;
+
+	priv->tls = tls;
+	return 0;
+}
+
+void mlx5e_ktls_cleanup(struct mlx5e_priv *priv)
+{
+	kfree(priv->tls);
+	priv->tls = NULL;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls.h b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls.h
new file mode 100644
index 000000000..1c35045e4
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls.h
@@ -0,0 +1,145 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2019 Mellanox Technologies. */
+
+#ifndef __MLX5E_KTLS_H__
+#define __MLX5E_KTLS_H__
+
+#include <linux/tls.h>
+#include <net/tls.h>
+#include "en.h"
+
+#ifdef CONFIG_MLX5_EN_TLS
+int mlx5_ktls_create_key(struct mlx5_core_dev *mdev,
+			 struct tls_crypto_info *crypto_info,
+			 u32 *p_key_id);
+void mlx5_ktls_destroy_key(struct mlx5_core_dev *mdev, u32 key_id);
+
+static inline bool mlx5e_is_ktls_device(struct mlx5_core_dev *mdev)
+{
+	if (is_kdump_kernel())
+		return false;
+
+	if (!MLX5_CAP_GEN(mdev, tls_tx) && !MLX5_CAP_GEN(mdev, tls_rx))
+		return false;
+
+	if (!MLX5_CAP_GEN(mdev, log_max_dek))
+		return false;
+
+	return (MLX5_CAP_TLS(mdev, tls_1_2_aes_gcm_128) ||
+		MLX5_CAP_TLS(mdev, tls_1_2_aes_gcm_256));
+}
+
+static inline bool mlx5e_ktls_type_check(struct mlx5_core_dev *mdev,
+					 struct tls_crypto_info *crypto_info)
+{
+	switch (crypto_info->cipher_type) {
+	case TLS_CIPHER_AES_GCM_128:
+		if (crypto_info->version == TLS_1_2_VERSION)
+			return MLX5_CAP_TLS(mdev,  tls_1_2_aes_gcm_128);
+		break;
+	case TLS_CIPHER_AES_GCM_256:
+		if (crypto_info->version == TLS_1_2_VERSION)
+			return MLX5_CAP_TLS(mdev,  tls_1_2_aes_gcm_256);
+		break;
+	}
+
+	return false;
+}
+
+void mlx5e_ktls_build_netdev(struct mlx5e_priv *priv);
+int mlx5e_ktls_init_tx(struct mlx5e_priv *priv);
+void mlx5e_ktls_cleanup_tx(struct mlx5e_priv *priv);
+int mlx5e_ktls_init_rx(struct mlx5e_priv *priv);
+void mlx5e_ktls_cleanup_rx(struct mlx5e_priv *priv);
+int mlx5e_ktls_set_feature_rx(struct net_device *netdev, bool enable);
+struct mlx5e_ktls_resync_resp *
+mlx5e_ktls_rx_resync_create_resp_list(void);
+void mlx5e_ktls_rx_resync_destroy_resp_list(struct mlx5e_ktls_resync_resp *resp_list);
+
+static inline bool mlx5e_is_ktls_tx(struct mlx5_core_dev *mdev)
+{
+	return !is_kdump_kernel() && MLX5_CAP_GEN(mdev, tls_tx);
+}
+
+bool mlx5e_is_ktls_rx(struct mlx5_core_dev *mdev);
+
+struct mlx5e_tls_sw_stats {
+	atomic64_t tx_tls_ctx;
+	atomic64_t tx_tls_del;
+	atomic64_t tx_tls_pool_alloc;
+	atomic64_t tx_tls_pool_free;
+	atomic64_t rx_tls_ctx;
+	atomic64_t rx_tls_del;
+};
+
+struct mlx5e_tls {
+	struct mlx5e_tls_sw_stats sw_stats;
+	struct workqueue_struct *rx_wq;
+	struct mlx5e_tls_tx_pool *tx_pool;
+};
+
+int mlx5e_ktls_init(struct mlx5e_priv *priv);
+void mlx5e_ktls_cleanup(struct mlx5e_priv *priv);
+
+int mlx5e_ktls_get_count(struct mlx5e_priv *priv);
+int mlx5e_ktls_get_strings(struct mlx5e_priv *priv, uint8_t *data);
+int mlx5e_ktls_get_stats(struct mlx5e_priv *priv, u64 *data);
+
+#else
+static inline void mlx5e_ktls_build_netdev(struct mlx5e_priv *priv)
+{
+}
+
+static inline int mlx5e_ktls_init_tx(struct mlx5e_priv *priv)
+{
+	return 0;
+}
+
+static inline void mlx5e_ktls_cleanup_tx(struct mlx5e_priv *priv)
+{
+}
+
+static inline int mlx5e_ktls_init_rx(struct mlx5e_priv *priv)
+{
+	return 0;
+}
+
+static inline void mlx5e_ktls_cleanup_rx(struct mlx5e_priv *priv)
+{
+}
+
+static inline int mlx5e_ktls_set_feature_rx(struct net_device *netdev, bool enable)
+{
+	netdev_warn(netdev, "kTLS is not supported\n");
+	return -EOPNOTSUPP;
+}
+
+static inline struct mlx5e_ktls_resync_resp *
+mlx5e_ktls_rx_resync_create_resp_list(void)
+{
+	return ERR_PTR(-EOPNOTSUPP);
+}
+
+static inline void
+mlx5e_ktls_rx_resync_destroy_resp_list(struct mlx5e_ktls_resync_resp *resp_list) {}
+
+static inline bool mlx5e_is_ktls_rx(struct mlx5_core_dev *mdev)
+{
+	return false;
+}
+
+static inline int mlx5e_ktls_init(struct mlx5e_priv *priv) { return 0; }
+static inline void mlx5e_ktls_cleanup(struct mlx5e_priv *priv) { }
+static inline int mlx5e_ktls_get_count(struct mlx5e_priv *priv) { return 0; }
+static inline int mlx5e_ktls_get_strings(struct mlx5e_priv *priv, uint8_t *data)
+{
+	return 0;
+}
+
+static inline int mlx5e_ktls_get_stats(struct mlx5e_priv *priv, u64 *data)
+{
+	return 0;
+}
+#endif
+
+#endif /* __MLX5E_TLS_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_rx.c b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_rx.c
new file mode 100644
index 000000000..3e5483474
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_rx.c
@@ -0,0 +1,782 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+// Copyright (c) 2019 Mellanox Technologies.
+
+#include <net/inet6_hashtables.h>
+#include "en_accel/en_accel.h"
+#include "en_accel/ktls.h"
+#include "en_accel/ktls_txrx.h"
+#include "en_accel/ktls_utils.h"
+#include "en_accel/fs_tcp.h"
+
+struct accel_rule {
+	struct work_struct work;
+	struct mlx5e_priv *priv;
+	struct mlx5_flow_handle *rule;
+};
+
+#define PROGRESS_PARAMS_WRITE_UNIT	64
+#define PROGRESS_PARAMS_PADDED_SIZE	\
+		(ALIGN(sizeof(struct mlx5_wqe_tls_progress_params_seg), \
+		       PROGRESS_PARAMS_WRITE_UNIT))
+
+struct mlx5e_ktls_rx_resync_buf {
+	union {
+		struct mlx5_wqe_tls_progress_params_seg progress;
+		u8 pad[PROGRESS_PARAMS_PADDED_SIZE];
+	} ____cacheline_aligned_in_smp;
+	dma_addr_t dma_addr;
+	struct mlx5e_ktls_offload_context_rx *priv_rx;
+};
+
+enum {
+	MLX5E_PRIV_RX_FLAG_DELETING,
+	MLX5E_NUM_PRIV_RX_FLAGS,
+};
+
+struct mlx5e_ktls_rx_resync_ctx {
+	struct tls_offload_resync_async core;
+	struct work_struct work;
+	struct mlx5e_priv *priv;
+	refcount_t refcnt;
+	__be64 sw_rcd_sn_be;
+	u32 seq;
+};
+
+struct mlx5e_ktls_offload_context_rx {
+	union mlx5e_crypto_info crypto_info;
+	struct accel_rule rule;
+	struct sock *sk;
+	struct mlx5e_rq_stats *rq_stats;
+	struct mlx5e_tls_sw_stats *sw_stats;
+	struct completion add_ctx;
+	struct mlx5e_tir tir;
+	u32 key_id;
+	u32 rxq;
+	DECLARE_BITMAP(flags, MLX5E_NUM_PRIV_RX_FLAGS);
+
+	/* resync */
+	spinlock_t lock; /* protects resync fields */
+	struct mlx5e_ktls_rx_resync_ctx resync;
+	struct list_head list;
+};
+
+static bool mlx5e_ktls_priv_rx_put(struct mlx5e_ktls_offload_context_rx *priv_rx)
+{
+	if (!refcount_dec_and_test(&priv_rx->resync.refcnt))
+		return false;
+
+	kfree(priv_rx);
+	return true;
+}
+
+static void mlx5e_ktls_priv_rx_get(struct mlx5e_ktls_offload_context_rx *priv_rx)
+{
+	refcount_inc(&priv_rx->resync.refcnt);
+}
+
+struct mlx5e_ktls_resync_resp {
+	/* protects list changes */
+	spinlock_t lock;
+	struct list_head list;
+};
+
+void mlx5e_ktls_rx_resync_destroy_resp_list(struct mlx5e_ktls_resync_resp *resp_list)
+{
+	kvfree(resp_list);
+}
+
+struct mlx5e_ktls_resync_resp *
+mlx5e_ktls_rx_resync_create_resp_list(void)
+{
+	struct mlx5e_ktls_resync_resp *resp_list;
+
+	resp_list = kvzalloc(sizeof(*resp_list), GFP_KERNEL);
+	if (!resp_list)
+		return ERR_PTR(-ENOMEM);
+
+	INIT_LIST_HEAD(&resp_list->list);
+	spin_lock_init(&resp_list->lock);
+
+	return resp_list;
+}
+
+static void accel_rule_handle_work(struct work_struct *work)
+{
+	struct mlx5e_ktls_offload_context_rx *priv_rx;
+	struct accel_rule *accel_rule;
+	struct mlx5_flow_handle *rule;
+
+	accel_rule = container_of(work, struct accel_rule, work);
+	priv_rx = container_of(accel_rule, struct mlx5e_ktls_offload_context_rx, rule);
+	if (unlikely(test_bit(MLX5E_PRIV_RX_FLAG_DELETING, priv_rx->flags)))
+		goto out;
+
+	rule = mlx5e_accel_fs_add_sk(accel_rule->priv->fs, priv_rx->sk,
+				     mlx5e_tir_get_tirn(&priv_rx->tir),
+				     MLX5_FS_DEFAULT_FLOW_TAG);
+	if (!IS_ERR_OR_NULL(rule))
+		accel_rule->rule = rule;
+out:
+	complete(&priv_rx->add_ctx);
+}
+
+static void accel_rule_init(struct accel_rule *rule, struct mlx5e_priv *priv)
+{
+	INIT_WORK(&rule->work, accel_rule_handle_work);
+	rule->priv = priv;
+}
+
+static void icosq_fill_wi(struct mlx5e_icosq *sq, u16 pi,
+			  struct mlx5e_icosq_wqe_info *wi)
+{
+	sq->db.wqe_info[pi] = *wi;
+}
+
+static struct mlx5_wqe_ctrl_seg *
+post_static_params(struct mlx5e_icosq *sq,
+		   struct mlx5e_ktls_offload_context_rx *priv_rx)
+{
+	struct mlx5e_set_tls_static_params_wqe *wqe;
+	struct mlx5e_icosq_wqe_info wi;
+	u16 pi, num_wqebbs;
+
+	num_wqebbs = MLX5E_TLS_SET_STATIC_PARAMS_WQEBBS;
+	if (unlikely(!mlx5e_icosq_can_post_wqe(sq, num_wqebbs)))
+		return ERR_PTR(-ENOSPC);
+
+	pi = mlx5e_icosq_get_next_pi(sq, num_wqebbs);
+	wqe = MLX5E_TLS_FETCH_SET_STATIC_PARAMS_WQE(sq, pi);
+	mlx5e_ktls_build_static_params(wqe, sq->pc, sq->sqn, &priv_rx->crypto_info,
+				       mlx5e_tir_get_tirn(&priv_rx->tir),
+				       priv_rx->key_id, priv_rx->resync.seq, false,
+				       TLS_OFFLOAD_CTX_DIR_RX);
+	wi = (struct mlx5e_icosq_wqe_info) {
+		.wqe_type = MLX5E_ICOSQ_WQE_UMR_TLS,
+		.num_wqebbs = num_wqebbs,
+		.tls_set_params.priv_rx = priv_rx,
+	};
+	icosq_fill_wi(sq, pi, &wi);
+	sq->pc += num_wqebbs;
+
+	return &wqe->ctrl;
+}
+
+static struct mlx5_wqe_ctrl_seg *
+post_progress_params(struct mlx5e_icosq *sq,
+		     struct mlx5e_ktls_offload_context_rx *priv_rx,
+		     u32 next_record_tcp_sn)
+{
+	struct mlx5e_set_tls_progress_params_wqe *wqe;
+	struct mlx5e_icosq_wqe_info wi;
+	u16 pi, num_wqebbs;
+
+	num_wqebbs = MLX5E_TLS_SET_PROGRESS_PARAMS_WQEBBS;
+	if (unlikely(!mlx5e_icosq_can_post_wqe(sq, num_wqebbs)))
+		return ERR_PTR(-ENOSPC);
+
+	pi = mlx5e_icosq_get_next_pi(sq, num_wqebbs);
+	wqe = MLX5E_TLS_FETCH_SET_PROGRESS_PARAMS_WQE(sq, pi);
+	mlx5e_ktls_build_progress_params(wqe, sq->pc, sq->sqn,
+					 mlx5e_tir_get_tirn(&priv_rx->tir),
+					 false, next_record_tcp_sn,
+					 TLS_OFFLOAD_CTX_DIR_RX);
+	wi = (struct mlx5e_icosq_wqe_info) {
+		.wqe_type = MLX5E_ICOSQ_WQE_SET_PSV_TLS,
+		.num_wqebbs = num_wqebbs,
+		.tls_set_params.priv_rx = priv_rx,
+	};
+
+	icosq_fill_wi(sq, pi, &wi);
+	sq->pc += num_wqebbs;
+
+	return &wqe->ctrl;
+}
+
+static int post_rx_param_wqes(struct mlx5e_channel *c,
+			      struct mlx5e_ktls_offload_context_rx *priv_rx,
+			      u32 next_record_tcp_sn)
+{
+	struct mlx5_wqe_ctrl_seg *cseg;
+	struct mlx5e_icosq *sq;
+	int err;
+
+	err = 0;
+	sq = &c->async_icosq;
+	spin_lock_bh(&c->async_icosq_lock);
+
+	cseg = post_static_params(sq, priv_rx);
+	if (IS_ERR(cseg))
+		goto err_out;
+	cseg = post_progress_params(sq, priv_rx, next_record_tcp_sn);
+	if (IS_ERR(cseg))
+		goto err_out;
+
+	mlx5e_notify_hw(&sq->wq, sq->pc, sq->uar_map, cseg);
+unlock:
+	spin_unlock_bh(&c->async_icosq_lock);
+
+	return err;
+
+err_out:
+	priv_rx->rq_stats->tls_resync_req_skip++;
+	err = PTR_ERR(cseg);
+	complete(&priv_rx->add_ctx);
+	goto unlock;
+}
+
+static void
+mlx5e_set_ktls_rx_priv_ctx(struct tls_context *tls_ctx,
+			   struct mlx5e_ktls_offload_context_rx *priv_rx)
+{
+	struct mlx5e_ktls_offload_context_rx **ctx =
+		__tls_driver_ctx(tls_ctx, TLS_OFFLOAD_CTX_DIR_RX);
+
+	BUILD_BUG_ON(sizeof(priv_rx) > TLS_DRIVER_STATE_SIZE_RX);
+
+	*ctx = priv_rx;
+}
+
+static struct mlx5e_ktls_offload_context_rx *
+mlx5e_get_ktls_rx_priv_ctx(struct tls_context *tls_ctx)
+{
+	struct mlx5e_ktls_offload_context_rx **ctx =
+		__tls_driver_ctx(tls_ctx, TLS_OFFLOAD_CTX_DIR_RX);
+
+	return *ctx;
+}
+
+/* Re-sync */
+/* Runs in work context */
+static int
+resync_post_get_progress_params(struct mlx5e_icosq *sq,
+				struct mlx5e_ktls_offload_context_rx *priv_rx)
+{
+	struct mlx5e_get_tls_progress_params_wqe *wqe;
+	struct mlx5e_ktls_rx_resync_buf *buf;
+	struct mlx5e_icosq_wqe_info wi;
+	struct mlx5_wqe_ctrl_seg *cseg;
+	struct mlx5_seg_get_psv *psv;
+	struct device *pdev;
+	int err;
+	u16 pi;
+
+	buf = kzalloc(sizeof(*buf), GFP_KERNEL);
+	if (unlikely(!buf)) {
+		err = -ENOMEM;
+		goto err_out;
+	}
+
+	pdev = mlx5_core_dma_dev(sq->channel->priv->mdev);
+	buf->dma_addr = dma_map_single(pdev, &buf->progress,
+				       PROGRESS_PARAMS_PADDED_SIZE, DMA_FROM_DEVICE);
+	if (unlikely(dma_mapping_error(pdev, buf->dma_addr))) {
+		err = -ENOMEM;
+		goto err_free;
+	}
+
+	buf->priv_rx = priv_rx;
+
+	spin_lock_bh(&sq->channel->async_icosq_lock);
+
+	if (unlikely(!mlx5e_icosq_can_post_wqe(sq, MLX5E_KTLS_GET_PROGRESS_WQEBBS))) {
+		spin_unlock_bh(&sq->channel->async_icosq_lock);
+		err = -ENOSPC;
+		goto err_dma_unmap;
+	}
+
+	pi = mlx5e_icosq_get_next_pi(sq, MLX5E_KTLS_GET_PROGRESS_WQEBBS);
+	wqe = MLX5E_TLS_FETCH_GET_PROGRESS_PARAMS_WQE(sq, pi);
+
+#define GET_PSV_DS_CNT (DIV_ROUND_UP(sizeof(*wqe), MLX5_SEND_WQE_DS))
+
+	cseg = &wqe->ctrl;
+	cseg->opmod_idx_opcode =
+		cpu_to_be32((sq->pc << 8) | MLX5_OPCODE_GET_PSV |
+			    (MLX5_OPC_MOD_TLS_TIR_PROGRESS_PARAMS << 24));
+	cseg->qpn_ds =
+		cpu_to_be32((sq->sqn << MLX5_WQE_CTRL_QPN_SHIFT) | GET_PSV_DS_CNT);
+
+	psv = &wqe->psv;
+	psv->num_psv      = 1 << 4;
+	psv->l_key        = sq->channel->mkey_be;
+	psv->psv_index[0] = cpu_to_be32(mlx5e_tir_get_tirn(&priv_rx->tir));
+	psv->va           = cpu_to_be64(buf->dma_addr);
+
+	wi = (struct mlx5e_icosq_wqe_info) {
+		.wqe_type = MLX5E_ICOSQ_WQE_GET_PSV_TLS,
+		.num_wqebbs = MLX5E_KTLS_GET_PROGRESS_WQEBBS,
+		.tls_get_params.buf = buf,
+	};
+	icosq_fill_wi(sq, pi, &wi);
+	sq->pc++;
+	mlx5e_notify_hw(&sq->wq, sq->pc, sq->uar_map, cseg);
+	spin_unlock_bh(&sq->channel->async_icosq_lock);
+
+	return 0;
+
+err_dma_unmap:
+	dma_unmap_single(pdev, buf->dma_addr, PROGRESS_PARAMS_PADDED_SIZE, DMA_FROM_DEVICE);
+err_free:
+	kfree(buf);
+err_out:
+	priv_rx->rq_stats->tls_resync_req_skip++;
+	return err;
+}
+
+/* Function is called with elevated refcount.
+ * It decreases it only if no WQE is posted.
+ */
+static void resync_handle_work(struct work_struct *work)
+{
+	struct mlx5e_ktls_offload_context_rx *priv_rx;
+	struct mlx5e_ktls_rx_resync_ctx *resync;
+	struct mlx5e_channel *c;
+	struct mlx5e_icosq *sq;
+
+	resync = container_of(work, struct mlx5e_ktls_rx_resync_ctx, work);
+	priv_rx = container_of(resync, struct mlx5e_ktls_offload_context_rx, resync);
+
+	if (unlikely(test_bit(MLX5E_PRIV_RX_FLAG_DELETING, priv_rx->flags))) {
+		mlx5e_ktls_priv_rx_put(priv_rx);
+		return;
+	}
+
+	c = resync->priv->channels.c[priv_rx->rxq];
+	sq = &c->async_icosq;
+
+	if (resync_post_get_progress_params(sq, priv_rx))
+		mlx5e_ktls_priv_rx_put(priv_rx);
+}
+
+static void resync_init(struct mlx5e_ktls_rx_resync_ctx *resync,
+			struct mlx5e_priv *priv)
+{
+	INIT_WORK(&resync->work, resync_handle_work);
+	resync->priv = priv;
+	refcount_set(&resync->refcnt, 1);
+}
+
+/* Function can be called with the refcount being either elevated or not.
+ * It does not affect the refcount.
+ */
+static void resync_handle_seq_match(struct mlx5e_ktls_offload_context_rx *priv_rx,
+				    struct mlx5e_channel *c)
+{
+	struct mlx5e_ktls_resync_resp *ktls_resync;
+	struct mlx5e_icosq *sq;
+	bool trigger_poll;
+
+	sq = &c->async_icosq;
+	ktls_resync = sq->ktls_resync;
+	trigger_poll = false;
+
+	spin_lock_bh(&ktls_resync->lock);
+	spin_lock_bh(&priv_rx->lock);
+	switch (priv_rx->crypto_info.crypto_info.cipher_type) {
+	case TLS_CIPHER_AES_GCM_128: {
+		struct tls12_crypto_info_aes_gcm_128 *info =
+			&priv_rx->crypto_info.crypto_info_128;
+
+		memcpy(info->rec_seq, &priv_rx->resync.sw_rcd_sn_be,
+		       sizeof(info->rec_seq));
+		break;
+	}
+	case TLS_CIPHER_AES_GCM_256: {
+		struct tls12_crypto_info_aes_gcm_256 *info =
+			&priv_rx->crypto_info.crypto_info_256;
+
+		memcpy(info->rec_seq, &priv_rx->resync.sw_rcd_sn_be,
+		       sizeof(info->rec_seq));
+		break;
+	}
+	default:
+		WARN_ONCE(1, "Unsupported cipher type %u\n",
+			  priv_rx->crypto_info.crypto_info.cipher_type);
+		spin_unlock_bh(&priv_rx->lock);
+		spin_unlock_bh(&ktls_resync->lock);
+		return;
+	}
+
+	if (list_empty(&priv_rx->list)) {
+		list_add_tail(&priv_rx->list, &ktls_resync->list);
+		trigger_poll = !test_and_set_bit(MLX5E_SQ_STATE_PENDING_TLS_RX_RESYNC, &sq->state);
+	}
+	spin_unlock_bh(&priv_rx->lock);
+	spin_unlock_bh(&ktls_resync->lock);
+
+	if (!trigger_poll)
+		return;
+
+	if (!napi_if_scheduled_mark_missed(&c->napi)) {
+		spin_lock_bh(&c->async_icosq_lock);
+		mlx5e_trigger_irq(sq);
+		spin_unlock_bh(&c->async_icosq_lock);
+	}
+}
+
+/* Function can be called with the refcount being either elevated or not.
+ * It decreases the refcount and may free the kTLS priv context.
+ * Refcount is not elevated only if tls_dev_del has been called, but GET_PSV was
+ * already in flight.
+ */
+void mlx5e_ktls_handle_get_psv_completion(struct mlx5e_icosq_wqe_info *wi,
+					  struct mlx5e_icosq *sq)
+{
+	struct mlx5e_ktls_rx_resync_buf *buf = wi->tls_get_params.buf;
+	struct mlx5e_ktls_offload_context_rx *priv_rx;
+	struct mlx5e_ktls_rx_resync_ctx *resync;
+	u8 tracker_state, auth_state, *ctx;
+	struct device *dev;
+	u32 hw_seq;
+
+	priv_rx = buf->priv_rx;
+	resync = &priv_rx->resync;
+	dev = mlx5_core_dma_dev(resync->priv->mdev);
+	if (unlikely(test_bit(MLX5E_PRIV_RX_FLAG_DELETING, priv_rx->flags)))
+		goto out;
+
+	dma_sync_single_for_cpu(dev, buf->dma_addr, PROGRESS_PARAMS_PADDED_SIZE,
+				DMA_FROM_DEVICE);
+
+	ctx = buf->progress.ctx;
+	tracker_state = MLX5_GET(tls_progress_params, ctx, record_tracker_state);
+	auth_state = MLX5_GET(tls_progress_params, ctx, auth_state);
+	if (tracker_state != MLX5E_TLS_PROGRESS_PARAMS_RECORD_TRACKER_STATE_TRACKING ||
+	    auth_state != MLX5E_TLS_PROGRESS_PARAMS_AUTH_STATE_NO_OFFLOAD) {
+		priv_rx->rq_stats->tls_resync_req_skip++;
+		goto out;
+	}
+
+	hw_seq = MLX5_GET(tls_progress_params, ctx, hw_resync_tcp_sn);
+	tls_offload_rx_resync_async_request_end(priv_rx->sk, cpu_to_be32(hw_seq));
+	priv_rx->rq_stats->tls_resync_req_end++;
+out:
+	mlx5e_ktls_priv_rx_put(priv_rx);
+	dma_unmap_single(dev, buf->dma_addr, PROGRESS_PARAMS_PADDED_SIZE, DMA_FROM_DEVICE);
+	kfree(buf);
+}
+
+/* Runs in NAPI.
+ * Function elevates the refcount, unless no work is queued.
+ */
+static bool resync_queue_get_psv(struct sock *sk)
+{
+	struct mlx5e_ktls_offload_context_rx *priv_rx;
+	struct mlx5e_ktls_rx_resync_ctx *resync;
+
+	priv_rx = mlx5e_get_ktls_rx_priv_ctx(tls_get_ctx(sk));
+	if (unlikely(!priv_rx))
+		return false;
+
+	if (unlikely(test_bit(MLX5E_PRIV_RX_FLAG_DELETING, priv_rx->flags)))
+		return false;
+
+	resync = &priv_rx->resync;
+	mlx5e_ktls_priv_rx_get(priv_rx);
+	if (unlikely(!queue_work(resync->priv->tls->rx_wq, &resync->work)))
+		mlx5e_ktls_priv_rx_put(priv_rx);
+
+	return true;
+}
+
+/* Runs in NAPI */
+static void resync_update_sn(struct mlx5e_rq *rq, struct sk_buff *skb)
+{
+	struct ethhdr *eth = (struct ethhdr *)(skb->data);
+	struct net_device *netdev = rq->netdev;
+	struct net *net = dev_net(netdev);
+	struct sock *sk = NULL;
+	unsigned int datalen;
+	struct iphdr *iph;
+	struct tcphdr *th;
+	__be32 seq;
+	int depth = 0;
+
+	__vlan_get_protocol(skb, eth->h_proto, &depth);
+	iph = (struct iphdr *)(skb->data + depth);
+
+	if (iph->version == 4) {
+		depth += sizeof(struct iphdr);
+		th = (void *)iph + sizeof(struct iphdr);
+
+		sk = inet_lookup_established(net, net->ipv4.tcp_death_row.hashinfo,
+					     iph->saddr, th->source, iph->daddr,
+					     th->dest, netdev->ifindex);
+#if IS_ENABLED(CONFIG_IPV6)
+	} else {
+		struct ipv6hdr *ipv6h = (struct ipv6hdr *)iph;
+
+		depth += sizeof(struct ipv6hdr);
+		th = (void *)ipv6h + sizeof(struct ipv6hdr);
+
+		sk = __inet6_lookup_established(net, net->ipv4.tcp_death_row.hashinfo,
+						&ipv6h->saddr, th->source,
+						&ipv6h->daddr, ntohs(th->dest),
+						netdev->ifindex, 0);
+#endif
+	}
+
+	depth += sizeof(struct tcphdr);
+
+	if (unlikely(!sk))
+		return;
+
+	if (unlikely(sk->sk_state == TCP_TIME_WAIT))
+		goto unref;
+
+	if (unlikely(!resync_queue_get_psv(sk)))
+		goto unref;
+
+	seq = th->seq;
+	datalen = skb->len - depth;
+	tls_offload_rx_resync_async_request_start(sk, seq, datalen);
+	rq->stats->tls_resync_req_start++;
+
+unref:
+	sock_gen_put(sk);
+}
+
+void mlx5e_ktls_rx_resync(struct net_device *netdev, struct sock *sk,
+			  u32 seq, u8 *rcd_sn)
+{
+	struct mlx5e_ktls_offload_context_rx *priv_rx;
+	struct mlx5e_ktls_rx_resync_ctx *resync;
+	struct mlx5e_priv *priv;
+	struct mlx5e_channel *c;
+
+	priv_rx = mlx5e_get_ktls_rx_priv_ctx(tls_get_ctx(sk));
+	if (unlikely(!priv_rx))
+		return;
+
+	resync = &priv_rx->resync;
+	resync->sw_rcd_sn_be = *(__be64 *)rcd_sn;
+	resync->seq = seq;
+
+	priv = netdev_priv(netdev);
+	c = priv->channels.c[priv_rx->rxq];
+
+	resync_handle_seq_match(priv_rx, c);
+}
+
+/* End of resync section */
+
+void mlx5e_ktls_handle_rx_skb(struct mlx5e_rq *rq, struct sk_buff *skb,
+			      struct mlx5_cqe64 *cqe, u32 *cqe_bcnt)
+{
+	struct mlx5e_rq_stats *stats = rq->stats;
+
+	switch (get_cqe_tls_offload(cqe)) {
+	case CQE_TLS_OFFLOAD_DECRYPTED:
+		skb->decrypted = 1;
+		stats->tls_decrypted_packets++;
+		stats->tls_decrypted_bytes += *cqe_bcnt;
+		break;
+	case CQE_TLS_OFFLOAD_RESYNC:
+		stats->tls_resync_req_pkt++;
+		resync_update_sn(rq, skb);
+		break;
+	default: /* CQE_TLS_OFFLOAD_ERROR: */
+		stats->tls_err++;
+		break;
+	}
+}
+
+void mlx5e_ktls_handle_ctx_completion(struct mlx5e_icosq_wqe_info *wi)
+{
+	struct mlx5e_ktls_offload_context_rx *priv_rx = wi->tls_set_params.priv_rx;
+	struct accel_rule *rule = &priv_rx->rule;
+
+	if (unlikely(test_bit(MLX5E_PRIV_RX_FLAG_DELETING, priv_rx->flags))) {
+		complete(&priv_rx->add_ctx);
+		return;
+	}
+	queue_work(rule->priv->tls->rx_wq, &rule->work);
+}
+
+static int mlx5e_ktls_sk_get_rxq(struct sock *sk)
+{
+	int rxq = sk_rx_queue_get(sk);
+
+	if (unlikely(rxq == -1))
+		rxq = 0;
+
+	return rxq;
+}
+
+int mlx5e_ktls_add_rx(struct net_device *netdev, struct sock *sk,
+		      struct tls_crypto_info *crypto_info,
+		      u32 start_offload_tcp_sn)
+{
+	struct mlx5e_ktls_offload_context_rx *priv_rx;
+	struct mlx5e_ktls_rx_resync_ctx *resync;
+	struct tls_context *tls_ctx;
+	struct mlx5_core_dev *mdev;
+	struct mlx5e_priv *priv;
+	int rxq, err;
+
+	tls_ctx = tls_get_ctx(sk);
+	priv = netdev_priv(netdev);
+	mdev = priv->mdev;
+	priv_rx = kzalloc(sizeof(*priv_rx), GFP_KERNEL);
+	if (unlikely(!priv_rx))
+		return -ENOMEM;
+
+	err = mlx5_ktls_create_key(mdev, crypto_info, &priv_rx->key_id);
+	if (err)
+		goto err_create_key;
+
+	INIT_LIST_HEAD(&priv_rx->list);
+	spin_lock_init(&priv_rx->lock);
+	switch (crypto_info->cipher_type) {
+	case TLS_CIPHER_AES_GCM_128:
+		priv_rx->crypto_info.crypto_info_128 =
+			*(struct tls12_crypto_info_aes_gcm_128 *)crypto_info;
+		break;
+	case TLS_CIPHER_AES_GCM_256:
+		priv_rx->crypto_info.crypto_info_256 =
+			*(struct tls12_crypto_info_aes_gcm_256 *)crypto_info;
+		break;
+	default:
+		WARN_ONCE(1, "Unsupported cipher type %u\n",
+			  crypto_info->cipher_type);
+		return -EOPNOTSUPP;
+	}
+
+	rxq = mlx5e_ktls_sk_get_rxq(sk);
+	priv_rx->rxq = rxq;
+	priv_rx->sk = sk;
+
+	priv_rx->rq_stats = &priv->channel_stats[rxq]->rq;
+	priv_rx->sw_stats = &priv->tls->sw_stats;
+	mlx5e_set_ktls_rx_priv_ctx(tls_ctx, priv_rx);
+
+	err = mlx5e_rx_res_tls_tir_create(priv->rx_res, rxq, &priv_rx->tir);
+	if (err)
+		goto err_create_tir;
+
+	init_completion(&priv_rx->add_ctx);
+
+	accel_rule_init(&priv_rx->rule, priv);
+	resync = &priv_rx->resync;
+	resync_init(resync, priv);
+	tls_offload_ctx_rx(tls_ctx)->resync_async = &resync->core;
+	tls_offload_rx_resync_set_type(sk, TLS_OFFLOAD_SYNC_TYPE_DRIVER_REQ_ASYNC);
+
+	err = post_rx_param_wqes(priv->channels.c[rxq], priv_rx, start_offload_tcp_sn);
+	if (err)
+		goto err_post_wqes;
+
+	atomic64_inc(&priv_rx->sw_stats->rx_tls_ctx);
+
+	return 0;
+
+err_post_wqes:
+	mlx5e_tir_destroy(&priv_rx->tir);
+err_create_tir:
+	mlx5_ktls_destroy_key(mdev, priv_rx->key_id);
+err_create_key:
+	kfree(priv_rx);
+	return err;
+}
+
+void mlx5e_ktls_del_rx(struct net_device *netdev, struct tls_context *tls_ctx)
+{
+	struct mlx5e_ktls_offload_context_rx *priv_rx;
+	struct mlx5e_ktls_rx_resync_ctx *resync;
+	struct mlx5_core_dev *mdev;
+	struct mlx5e_priv *priv;
+
+	priv = netdev_priv(netdev);
+	mdev = priv->mdev;
+
+	priv_rx = mlx5e_get_ktls_rx_priv_ctx(tls_ctx);
+	set_bit(MLX5E_PRIV_RX_FLAG_DELETING, priv_rx->flags);
+	mlx5e_set_ktls_rx_priv_ctx(tls_ctx, NULL);
+	synchronize_net(); /* Sync with NAPI */
+	if (!cancel_work_sync(&priv_rx->rule.work))
+		/* completion is needed, as the priv_rx in the add flow
+		 * is maintained on the wqe info (wi), not on the socket.
+		 */
+		wait_for_completion(&priv_rx->add_ctx);
+	resync = &priv_rx->resync;
+	if (cancel_work_sync(&resync->work))
+		mlx5e_ktls_priv_rx_put(priv_rx);
+
+	atomic64_inc(&priv_rx->sw_stats->rx_tls_del);
+	if (priv_rx->rule.rule)
+		mlx5e_accel_fs_del_sk(priv_rx->rule.rule);
+
+	mlx5e_tir_destroy(&priv_rx->tir);
+	mlx5_ktls_destroy_key(mdev, priv_rx->key_id);
+	/* priv_rx should normally be freed here, but if there is an outstanding
+	 * GET_PSV, deallocation will be delayed until the CQE for GET_PSV is
+	 * processed.
+	 */
+	mlx5e_ktls_priv_rx_put(priv_rx);
+}
+
+bool mlx5e_ktls_rx_handle_resync_list(struct mlx5e_channel *c, int budget)
+{
+	struct mlx5e_ktls_offload_context_rx *priv_rx, *tmp;
+	struct mlx5e_ktls_resync_resp *ktls_resync;
+	struct mlx5_wqe_ctrl_seg *db_cseg;
+	struct mlx5e_icosq *sq;
+	LIST_HEAD(local_list);
+	int i, j;
+
+	sq = &c->async_icosq;
+
+	if (unlikely(!test_bit(MLX5E_SQ_STATE_ENABLED, &sq->state)))
+		return false;
+
+	ktls_resync = sq->ktls_resync;
+	db_cseg = NULL;
+	i = 0;
+
+	spin_lock(&ktls_resync->lock);
+	list_for_each_entry_safe(priv_rx, tmp, &ktls_resync->list, list) {
+		list_move(&priv_rx->list, &local_list);
+		if (++i == budget)
+			break;
+	}
+	if (list_empty(&ktls_resync->list))
+		clear_bit(MLX5E_SQ_STATE_PENDING_TLS_RX_RESYNC, &sq->state);
+	spin_unlock(&ktls_resync->lock);
+
+	spin_lock(&c->async_icosq_lock);
+	for (j = 0; j < i; j++) {
+		struct mlx5_wqe_ctrl_seg *cseg;
+
+		priv_rx = list_first_entry(&local_list,
+					   struct mlx5e_ktls_offload_context_rx,
+					   list);
+		spin_lock(&priv_rx->lock);
+		cseg = post_static_params(sq, priv_rx);
+		if (IS_ERR(cseg)) {
+			spin_unlock(&priv_rx->lock);
+			break;
+		}
+		list_del_init(&priv_rx->list);
+		spin_unlock(&priv_rx->lock);
+		db_cseg = cseg;
+	}
+	if (db_cseg)
+		mlx5e_notify_hw(&sq->wq, sq->pc, sq->uar_map, db_cseg);
+	spin_unlock(&c->async_icosq_lock);
+
+	priv_rx->rq_stats->tls_resync_res_ok += j;
+
+	if (!list_empty(&local_list)) {
+		/* This happens only if ICOSQ is full.
+		 * There is no need to mark busy or explicitly ask for a NAPI cycle,
+		 * it will be triggered by the outstanding ICOSQ completions.
+		 */
+		spin_lock(&ktls_resync->lock);
+		list_splice(&local_list, &ktls_resync->list);
+		set_bit(MLX5E_SQ_STATE_PENDING_TLS_RX_RESYNC, &sq->state);
+		spin_unlock(&ktls_resync->lock);
+		priv_rx->rq_stats->tls_resync_res_retry++;
+	}
+
+	return i == budget;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_stats.c b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_stats.c
new file mode 100644
index 000000000..7c1c0eb16
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_stats.c
@@ -0,0 +1,92 @@
+/*
+ * Copyright (c) 2018 Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+
+#include <linux/ethtool.h>
+#include <net/sock.h>
+
+#include "en.h"
+#include "fpga/sdk.h"
+#include "en_accel/ktls.h"
+
+static const struct counter_desc mlx5e_ktls_sw_stats_desc[] = {
+	{ MLX5E_DECLARE_STAT(struct mlx5e_tls_sw_stats, tx_tls_ctx) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_tls_sw_stats, tx_tls_del) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_tls_sw_stats, tx_tls_pool_alloc) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_tls_sw_stats, tx_tls_pool_free) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_tls_sw_stats, rx_tls_ctx) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_tls_sw_stats, rx_tls_del) },
+};
+
+#define MLX5E_READ_CTR_ATOMIC64(ptr, dsc, i) \
+	atomic64_read((atomic64_t *)((char *)(ptr) + (dsc)[i].offset))
+
+int mlx5e_ktls_get_count(struct mlx5e_priv *priv)
+{
+	if (!priv->tls)
+		return 0;
+
+	return ARRAY_SIZE(mlx5e_ktls_sw_stats_desc);
+}
+
+int mlx5e_ktls_get_strings(struct mlx5e_priv *priv, uint8_t *data)
+{
+	unsigned int i, n, idx = 0;
+
+	if (!priv->tls)
+		return 0;
+
+	n = mlx5e_ktls_get_count(priv);
+
+	for (i = 0; i < n; i++)
+		strcpy(data + (idx++) * ETH_GSTRING_LEN,
+		       mlx5e_ktls_sw_stats_desc[i].format);
+
+	return n;
+}
+
+int mlx5e_ktls_get_stats(struct mlx5e_priv *priv, u64 *data)
+{
+	unsigned int i, n, idx = 0;
+
+	if (!priv->tls)
+		return 0;
+
+	n = mlx5e_ktls_get_count(priv);
+
+	for (i = 0; i < n; i++)
+		data[idx++] = MLX5E_READ_CTR_ATOMIC64(&priv->tls->sw_stats,
+						      mlx5e_ktls_sw_stats_desc,
+						      i);
+
+	return n;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_tx.c b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_tx.c
new file mode 100644
index 000000000..2e0335246
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_tx.c
@@ -0,0 +1,921 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+// Copyright (c) 2019 Mellanox Technologies.
+
+#include "en_accel/ktls.h"
+#include "en_accel/ktls_txrx.h"
+#include "en_accel/ktls_utils.h"
+
+struct mlx5e_dump_wqe {
+	struct mlx5_wqe_ctrl_seg ctrl;
+	struct mlx5_wqe_data_seg data;
+};
+
+#define MLX5E_KTLS_DUMP_WQEBBS \
+	(DIV_ROUND_UP(sizeof(struct mlx5e_dump_wqe), MLX5_SEND_WQE_BB))
+
+static u8
+mlx5e_ktls_dumps_num_wqes(struct mlx5e_params *params, unsigned int nfrags,
+			  unsigned int sync_len)
+{
+	/* Given the MTU and sync_len, calculates an upper bound for the
+	 * number of DUMP WQEs needed for the TX resync of a record.
+	 */
+	return nfrags + DIV_ROUND_UP(sync_len, MLX5E_SW2HW_MTU(params, params->sw_mtu));
+}
+
+u16 mlx5e_ktls_get_stop_room(struct mlx5_core_dev *mdev, struct mlx5e_params *params)
+{
+	u16 num_dumps, stop_room = 0;
+
+	if (!mlx5e_is_ktls_tx(mdev))
+		return 0;
+
+	num_dumps = mlx5e_ktls_dumps_num_wqes(params, MAX_SKB_FRAGS, TLS_MAX_PAYLOAD_SIZE);
+
+	stop_room += mlx5e_stop_room_for_wqe(mdev, MLX5E_TLS_SET_STATIC_PARAMS_WQEBBS);
+	stop_room += mlx5e_stop_room_for_wqe(mdev, MLX5E_TLS_SET_PROGRESS_PARAMS_WQEBBS);
+	stop_room += num_dumps * mlx5e_stop_room_for_wqe(mdev, MLX5E_KTLS_DUMP_WQEBBS);
+	stop_room += 1; /* fence nop */
+
+	return stop_room;
+}
+
+static void mlx5e_ktls_set_tisc(struct mlx5_core_dev *mdev, void *tisc)
+{
+	MLX5_SET(tisc, tisc, tls_en, 1);
+	MLX5_SET(tisc, tisc, pd, mdev->mlx5e_res.hw_objs.pdn);
+	MLX5_SET(tisc, tisc, transport_domain, mdev->mlx5e_res.hw_objs.td.tdn);
+}
+
+static int mlx5e_ktls_create_tis(struct mlx5_core_dev *mdev, u32 *tisn)
+{
+	u32 in[MLX5_ST_SZ_DW(create_tis_in)] = {};
+
+	mlx5e_ktls_set_tisc(mdev, MLX5_ADDR_OF(create_tis_in, in, ctx));
+
+	return mlx5_core_create_tis(mdev, in, tisn);
+}
+
+static int mlx5e_ktls_create_tis_cb(struct mlx5_core_dev *mdev,
+				    struct mlx5_async_ctx *async_ctx,
+				    u32 *out, int outlen,
+				    mlx5_async_cbk_t callback,
+				    struct mlx5_async_work *context)
+{
+	u32 in[MLX5_ST_SZ_DW(create_tis_in)] = {};
+
+	mlx5e_ktls_set_tisc(mdev, MLX5_ADDR_OF(create_tis_in, in, ctx));
+	MLX5_SET(create_tis_in, in, opcode, MLX5_CMD_OP_CREATE_TIS);
+
+	return mlx5_cmd_exec_cb(async_ctx, in, sizeof(in),
+				out, outlen, callback, context);
+}
+
+static int mlx5e_ktls_destroy_tis_cb(struct mlx5_core_dev *mdev, u32 tisn,
+				     struct mlx5_async_ctx *async_ctx,
+				     u32 *out, int outlen,
+				     mlx5_async_cbk_t callback,
+				     struct mlx5_async_work *context)
+{
+	u32 in[MLX5_ST_SZ_DW(destroy_tis_in)] = {};
+
+	MLX5_SET(destroy_tis_in, in, opcode, MLX5_CMD_OP_DESTROY_TIS);
+	MLX5_SET(destroy_tis_in, in, tisn, tisn);
+
+	return mlx5_cmd_exec_cb(async_ctx, in, sizeof(in),
+				out, outlen, callback, context);
+}
+
+struct mlx5e_ktls_offload_context_tx {
+	/* fast path */
+	u32 expected_seq;
+	u32 tisn;
+	bool ctx_post_pending;
+	/* control / resync */
+	struct list_head list_node; /* member of the pool */
+	union mlx5e_crypto_info crypto_info;
+	struct tls_offload_context_tx *tx_ctx;
+	struct mlx5_core_dev *mdev;
+	struct mlx5e_tls_sw_stats *sw_stats;
+	u32 key_id;
+	u8 create_err : 1;
+};
+
+static void
+mlx5e_set_ktls_tx_priv_ctx(struct tls_context *tls_ctx,
+			   struct mlx5e_ktls_offload_context_tx *priv_tx)
+{
+	struct mlx5e_ktls_offload_context_tx **ctx =
+		__tls_driver_ctx(tls_ctx, TLS_OFFLOAD_CTX_DIR_TX);
+
+	BUILD_BUG_ON(sizeof(priv_tx) > TLS_DRIVER_STATE_SIZE_TX);
+
+	*ctx = priv_tx;
+}
+
+static struct mlx5e_ktls_offload_context_tx *
+mlx5e_get_ktls_tx_priv_ctx(struct tls_context *tls_ctx)
+{
+	struct mlx5e_ktls_offload_context_tx **ctx =
+		__tls_driver_ctx(tls_ctx, TLS_OFFLOAD_CTX_DIR_TX);
+
+	return *ctx;
+}
+
+/* struct for callback API management */
+struct mlx5e_async_ctx {
+	struct mlx5_async_work context;
+	struct mlx5_async_ctx async_ctx;
+	struct work_struct work;
+	struct mlx5e_ktls_offload_context_tx *priv_tx;
+	struct completion complete;
+	int err;
+	union {
+		u32 out_create[MLX5_ST_SZ_DW(create_tis_out)];
+		u32 out_destroy[MLX5_ST_SZ_DW(destroy_tis_out)];
+	};
+};
+
+static struct mlx5e_async_ctx *mlx5e_bulk_async_init(struct mlx5_core_dev *mdev, int n)
+{
+	struct mlx5e_async_ctx *bulk_async;
+	int i;
+
+	bulk_async = kvcalloc(n, sizeof(struct mlx5e_async_ctx), GFP_KERNEL);
+	if (!bulk_async)
+		return NULL;
+
+	for (i = 0; i < n; i++) {
+		struct mlx5e_async_ctx *async = &bulk_async[i];
+
+		mlx5_cmd_init_async_ctx(mdev, &async->async_ctx);
+		init_completion(&async->complete);
+	}
+
+	return bulk_async;
+}
+
+static void mlx5e_bulk_async_cleanup(struct mlx5e_async_ctx *bulk_async, int n)
+{
+	int i;
+
+	for (i = 0; i < n; i++) {
+		struct mlx5e_async_ctx *async = &bulk_async[i];
+
+		mlx5_cmd_cleanup_async_ctx(&async->async_ctx);
+	}
+	kvfree(bulk_async);
+}
+
+static void create_tis_callback(int status, struct mlx5_async_work *context)
+{
+	struct mlx5e_async_ctx *async =
+		container_of(context, struct mlx5e_async_ctx, context);
+	struct mlx5e_ktls_offload_context_tx *priv_tx = async->priv_tx;
+
+	if (status) {
+		async->err = status;
+		priv_tx->create_err = 1;
+		goto out;
+	}
+
+	priv_tx->tisn = MLX5_GET(create_tis_out, async->out_create, tisn);
+out:
+	complete(&async->complete);
+}
+
+static void destroy_tis_callback(int status, struct mlx5_async_work *context)
+{
+	struct mlx5e_async_ctx *async =
+		container_of(context, struct mlx5e_async_ctx, context);
+	struct mlx5e_ktls_offload_context_tx *priv_tx = async->priv_tx;
+
+	complete(&async->complete);
+	kfree(priv_tx);
+}
+
+static struct mlx5e_ktls_offload_context_tx *
+mlx5e_tls_priv_tx_init(struct mlx5_core_dev *mdev, struct mlx5e_tls_sw_stats *sw_stats,
+		       struct mlx5e_async_ctx *async)
+{
+	struct mlx5e_ktls_offload_context_tx *priv_tx;
+	int err;
+
+	priv_tx = kzalloc(sizeof(*priv_tx), GFP_KERNEL);
+	if (!priv_tx)
+		return ERR_PTR(-ENOMEM);
+
+	priv_tx->mdev = mdev;
+	priv_tx->sw_stats = sw_stats;
+
+	if (!async) {
+		err = mlx5e_ktls_create_tis(mdev, &priv_tx->tisn);
+		if (err)
+			goto err_out;
+	} else {
+		async->priv_tx = priv_tx;
+		err = mlx5e_ktls_create_tis_cb(mdev, &async->async_ctx,
+					       async->out_create, sizeof(async->out_create),
+					       create_tis_callback, &async->context);
+		if (err)
+			goto err_out;
+	}
+
+	return priv_tx;
+
+err_out:
+	kfree(priv_tx);
+	return ERR_PTR(err);
+}
+
+static void mlx5e_tls_priv_tx_cleanup(struct mlx5e_ktls_offload_context_tx *priv_tx,
+				      struct mlx5e_async_ctx *async)
+{
+	if (priv_tx->create_err) {
+		complete(&async->complete);
+		kfree(priv_tx);
+		return;
+	}
+	async->priv_tx = priv_tx;
+	mlx5e_ktls_destroy_tis_cb(priv_tx->mdev, priv_tx->tisn,
+				  &async->async_ctx,
+				  async->out_destroy, sizeof(async->out_destroy),
+				  destroy_tis_callback, &async->context);
+}
+
+static void mlx5e_tls_priv_tx_list_cleanup(struct mlx5_core_dev *mdev,
+					   struct list_head *list, int size)
+{
+	struct mlx5e_ktls_offload_context_tx *obj, *n;
+	struct mlx5e_async_ctx *bulk_async;
+	int i;
+
+	bulk_async = mlx5e_bulk_async_init(mdev, size);
+	if (!bulk_async)
+		return;
+
+	i = 0;
+	list_for_each_entry_safe(obj, n, list, list_node) {
+		mlx5e_tls_priv_tx_cleanup(obj, &bulk_async[i]);
+		i++;
+	}
+
+	for (i = 0; i < size; i++) {
+		struct mlx5e_async_ctx *async = &bulk_async[i];
+
+		wait_for_completion(&async->complete);
+	}
+	mlx5e_bulk_async_cleanup(bulk_async, size);
+}
+
+/* Recycling pool API */
+
+#define MLX5E_TLS_TX_POOL_BULK (16)
+#define MLX5E_TLS_TX_POOL_HIGH (4 * 1024)
+#define MLX5E_TLS_TX_POOL_LOW (MLX5E_TLS_TX_POOL_HIGH / 4)
+
+struct mlx5e_tls_tx_pool {
+	struct mlx5_core_dev *mdev;
+	struct mlx5e_tls_sw_stats *sw_stats;
+	struct mutex lock; /* Protects access to the pool */
+	struct list_head list;
+	size_t size;
+
+	struct workqueue_struct *wq;
+	struct work_struct create_work;
+	struct work_struct destroy_work;
+};
+
+static void create_work(struct work_struct *work)
+{
+	struct mlx5e_tls_tx_pool *pool =
+		container_of(work, struct mlx5e_tls_tx_pool, create_work);
+	struct mlx5e_ktls_offload_context_tx *obj;
+	struct mlx5e_async_ctx *bulk_async;
+	LIST_HEAD(local_list);
+	int i, j, err = 0;
+
+	bulk_async = mlx5e_bulk_async_init(pool->mdev, MLX5E_TLS_TX_POOL_BULK);
+	if (!bulk_async)
+		return;
+
+	for (i = 0; i < MLX5E_TLS_TX_POOL_BULK; i++) {
+		obj = mlx5e_tls_priv_tx_init(pool->mdev, pool->sw_stats, &bulk_async[i]);
+		if (IS_ERR(obj)) {
+			err = PTR_ERR(obj);
+			break;
+		}
+		list_add(&obj->list_node, &local_list);
+	}
+
+	for (j = 0; j < i; j++) {
+		struct mlx5e_async_ctx *async = &bulk_async[j];
+
+		wait_for_completion(&async->complete);
+		if (!err && async->err)
+			err = async->err;
+	}
+	atomic64_add(i, &pool->sw_stats->tx_tls_pool_alloc);
+	mlx5e_bulk_async_cleanup(bulk_async, MLX5E_TLS_TX_POOL_BULK);
+	if (err)
+		goto err_out;
+
+	mutex_lock(&pool->lock);
+	if (pool->size + MLX5E_TLS_TX_POOL_BULK >= MLX5E_TLS_TX_POOL_HIGH) {
+		mutex_unlock(&pool->lock);
+		goto err_out;
+	}
+	list_splice(&local_list, &pool->list);
+	pool->size += MLX5E_TLS_TX_POOL_BULK;
+	if (pool->size <= MLX5E_TLS_TX_POOL_LOW)
+		queue_work(pool->wq, work);
+	mutex_unlock(&pool->lock);
+	return;
+
+err_out:
+	mlx5e_tls_priv_tx_list_cleanup(pool->mdev, &local_list, i);
+	atomic64_add(i, &pool->sw_stats->tx_tls_pool_free);
+}
+
+static void destroy_work(struct work_struct *work)
+{
+	struct mlx5e_tls_tx_pool *pool =
+		container_of(work, struct mlx5e_tls_tx_pool, destroy_work);
+	struct mlx5e_ktls_offload_context_tx *obj;
+	LIST_HEAD(local_list);
+	int i = 0;
+
+	mutex_lock(&pool->lock);
+	if (pool->size < MLX5E_TLS_TX_POOL_HIGH) {
+		mutex_unlock(&pool->lock);
+		return;
+	}
+
+	list_for_each_entry(obj, &pool->list, list_node)
+		if (++i == MLX5E_TLS_TX_POOL_BULK)
+			break;
+
+	list_cut_position(&local_list, &pool->list, &obj->list_node);
+	pool->size -= MLX5E_TLS_TX_POOL_BULK;
+	if (pool->size >= MLX5E_TLS_TX_POOL_HIGH)
+		queue_work(pool->wq, work);
+	mutex_unlock(&pool->lock);
+
+	mlx5e_tls_priv_tx_list_cleanup(pool->mdev, &local_list, MLX5E_TLS_TX_POOL_BULK);
+	atomic64_add(MLX5E_TLS_TX_POOL_BULK, &pool->sw_stats->tx_tls_pool_free);
+}
+
+static struct mlx5e_tls_tx_pool *mlx5e_tls_tx_pool_init(struct mlx5_core_dev *mdev,
+							struct mlx5e_tls_sw_stats *sw_stats)
+{
+	struct mlx5e_tls_tx_pool *pool;
+
+	BUILD_BUG_ON(MLX5E_TLS_TX_POOL_LOW + MLX5E_TLS_TX_POOL_BULK >= MLX5E_TLS_TX_POOL_HIGH);
+
+	pool = kvzalloc(sizeof(*pool), GFP_KERNEL);
+	if (!pool)
+		return NULL;
+
+	pool->wq = create_singlethread_workqueue("mlx5e_tls_tx_pool");
+	if (!pool->wq)
+		goto err_free;
+
+	INIT_LIST_HEAD(&pool->list);
+	mutex_init(&pool->lock);
+
+	INIT_WORK(&pool->create_work, create_work);
+	INIT_WORK(&pool->destroy_work, destroy_work);
+
+	pool->mdev = mdev;
+	pool->sw_stats = sw_stats;
+
+	return pool;
+
+err_free:
+	kvfree(pool);
+	return NULL;
+}
+
+static void mlx5e_tls_tx_pool_list_cleanup(struct mlx5e_tls_tx_pool *pool)
+{
+	while (pool->size > MLX5E_TLS_TX_POOL_BULK) {
+		struct mlx5e_ktls_offload_context_tx *obj;
+		LIST_HEAD(local_list);
+		int i = 0;
+
+		list_for_each_entry(obj, &pool->list, list_node)
+			if (++i == MLX5E_TLS_TX_POOL_BULK)
+				break;
+
+		list_cut_position(&local_list, &pool->list, &obj->list_node);
+		mlx5e_tls_priv_tx_list_cleanup(pool->mdev, &local_list, MLX5E_TLS_TX_POOL_BULK);
+		atomic64_add(MLX5E_TLS_TX_POOL_BULK, &pool->sw_stats->tx_tls_pool_free);
+		pool->size -= MLX5E_TLS_TX_POOL_BULK;
+	}
+	if (pool->size) {
+		mlx5e_tls_priv_tx_list_cleanup(pool->mdev, &pool->list, pool->size);
+		atomic64_add(pool->size, &pool->sw_stats->tx_tls_pool_free);
+	}
+}
+
+static void mlx5e_tls_tx_pool_cleanup(struct mlx5e_tls_tx_pool *pool)
+{
+	mlx5e_tls_tx_pool_list_cleanup(pool);
+	destroy_workqueue(pool->wq);
+	kvfree(pool);
+}
+
+static void pool_push(struct mlx5e_tls_tx_pool *pool, struct mlx5e_ktls_offload_context_tx *obj)
+{
+	mutex_lock(&pool->lock);
+	list_add(&obj->list_node, &pool->list);
+	if (++pool->size == MLX5E_TLS_TX_POOL_HIGH)
+		queue_work(pool->wq, &pool->destroy_work);
+	mutex_unlock(&pool->lock);
+}
+
+static struct mlx5e_ktls_offload_context_tx *pool_pop(struct mlx5e_tls_tx_pool *pool)
+{
+	struct mlx5e_ktls_offload_context_tx *obj;
+
+	mutex_lock(&pool->lock);
+	if (unlikely(pool->size == 0)) {
+		/* pool is empty:
+		 * - trigger the populating work, and
+		 * - serve the current context via the regular blocking api.
+		 */
+		queue_work(pool->wq, &pool->create_work);
+		mutex_unlock(&pool->lock);
+		obj = mlx5e_tls_priv_tx_init(pool->mdev, pool->sw_stats, NULL);
+		if (!IS_ERR(obj))
+			atomic64_inc(&pool->sw_stats->tx_tls_pool_alloc);
+		return obj;
+	}
+
+	obj = list_first_entry(&pool->list, struct mlx5e_ktls_offload_context_tx,
+			       list_node);
+	list_del(&obj->list_node);
+	if (--pool->size == MLX5E_TLS_TX_POOL_LOW)
+		queue_work(pool->wq, &pool->create_work);
+	mutex_unlock(&pool->lock);
+	return obj;
+}
+
+/* End of pool API */
+
+int mlx5e_ktls_add_tx(struct net_device *netdev, struct sock *sk,
+		      struct tls_crypto_info *crypto_info, u32 start_offload_tcp_sn)
+{
+	struct mlx5e_ktls_offload_context_tx *priv_tx;
+	struct mlx5e_tls_tx_pool *pool;
+	struct tls_context *tls_ctx;
+	struct mlx5e_priv *priv;
+	int err;
+
+	tls_ctx = tls_get_ctx(sk);
+	priv = netdev_priv(netdev);
+	pool = priv->tls->tx_pool;
+
+	priv_tx = pool_pop(pool);
+	if (IS_ERR(priv_tx))
+		return PTR_ERR(priv_tx);
+
+	err = mlx5_ktls_create_key(pool->mdev, crypto_info, &priv_tx->key_id);
+	if (err)
+		goto err_create_key;
+
+	priv_tx->expected_seq = start_offload_tcp_sn;
+	switch (crypto_info->cipher_type) {
+	case TLS_CIPHER_AES_GCM_128:
+		priv_tx->crypto_info.crypto_info_128 =
+			*(struct tls12_crypto_info_aes_gcm_128 *)crypto_info;
+		break;
+	case TLS_CIPHER_AES_GCM_256:
+		priv_tx->crypto_info.crypto_info_256 =
+			*(struct tls12_crypto_info_aes_gcm_256 *)crypto_info;
+		break;
+	default:
+		WARN_ONCE(1, "Unsupported cipher type %u\n",
+			  crypto_info->cipher_type);
+		return -EOPNOTSUPP;
+	}
+	priv_tx->tx_ctx = tls_offload_ctx_tx(tls_ctx);
+
+	mlx5e_set_ktls_tx_priv_ctx(tls_ctx, priv_tx);
+
+	priv_tx->ctx_post_pending = true;
+	atomic64_inc(&priv_tx->sw_stats->tx_tls_ctx);
+
+	return 0;
+
+err_create_key:
+	pool_push(pool, priv_tx);
+	return err;
+}
+
+void mlx5e_ktls_del_tx(struct net_device *netdev, struct tls_context *tls_ctx)
+{
+	struct mlx5e_ktls_offload_context_tx *priv_tx;
+	struct mlx5e_tls_tx_pool *pool;
+	struct mlx5e_priv *priv;
+
+	priv_tx = mlx5e_get_ktls_tx_priv_ctx(tls_ctx);
+	priv = netdev_priv(netdev);
+	pool = priv->tls->tx_pool;
+
+	atomic64_inc(&priv_tx->sw_stats->tx_tls_del);
+	mlx5_ktls_destroy_key(priv_tx->mdev, priv_tx->key_id);
+	pool_push(pool, priv_tx);
+}
+
+static void tx_fill_wi(struct mlx5e_txqsq *sq,
+		       u16 pi, u8 num_wqebbs, u32 num_bytes,
+		       struct page *page)
+{
+	struct mlx5e_tx_wqe_info *wi = &sq->db.wqe_info[pi];
+
+	*wi = (struct mlx5e_tx_wqe_info) {
+		.num_wqebbs = num_wqebbs,
+		.num_bytes  = num_bytes,
+		.resync_dump_frag_page = page,
+	};
+}
+
+static bool
+mlx5e_ktls_tx_offload_test_and_clear_pending(struct mlx5e_ktls_offload_context_tx *priv_tx)
+{
+	bool ret = priv_tx->ctx_post_pending;
+
+	priv_tx->ctx_post_pending = false;
+
+	return ret;
+}
+
+static void
+post_static_params(struct mlx5e_txqsq *sq,
+		   struct mlx5e_ktls_offload_context_tx *priv_tx,
+		   bool fence)
+{
+	struct mlx5e_set_tls_static_params_wqe *wqe;
+	u16 pi, num_wqebbs;
+
+	num_wqebbs = MLX5E_TLS_SET_STATIC_PARAMS_WQEBBS;
+	pi = mlx5e_txqsq_get_next_pi(sq, num_wqebbs);
+	wqe = MLX5E_TLS_FETCH_SET_STATIC_PARAMS_WQE(sq, pi);
+	mlx5e_ktls_build_static_params(wqe, sq->pc, sq->sqn, &priv_tx->crypto_info,
+				       priv_tx->tisn, priv_tx->key_id, 0, fence,
+				       TLS_OFFLOAD_CTX_DIR_TX);
+	tx_fill_wi(sq, pi, num_wqebbs, 0, NULL);
+	sq->pc += num_wqebbs;
+}
+
+static void
+post_progress_params(struct mlx5e_txqsq *sq,
+		     struct mlx5e_ktls_offload_context_tx *priv_tx,
+		     bool fence)
+{
+	struct mlx5e_set_tls_progress_params_wqe *wqe;
+	u16 pi, num_wqebbs;
+
+	num_wqebbs = MLX5E_TLS_SET_PROGRESS_PARAMS_WQEBBS;
+	pi = mlx5e_txqsq_get_next_pi(sq, num_wqebbs);
+	wqe = MLX5E_TLS_FETCH_SET_PROGRESS_PARAMS_WQE(sq, pi);
+	mlx5e_ktls_build_progress_params(wqe, sq->pc, sq->sqn, priv_tx->tisn, fence, 0,
+					 TLS_OFFLOAD_CTX_DIR_TX);
+	tx_fill_wi(sq, pi, num_wqebbs, 0, NULL);
+	sq->pc += num_wqebbs;
+}
+
+static void tx_post_fence_nop(struct mlx5e_txqsq *sq)
+{
+	struct mlx5_wq_cyc *wq = &sq->wq;
+	u16 pi = mlx5_wq_cyc_ctr2ix(wq, sq->pc);
+
+	tx_fill_wi(sq, pi, 1, 0, NULL);
+
+	mlx5e_post_nop_fence(wq, sq->sqn, &sq->pc);
+}
+
+static void
+mlx5e_ktls_tx_post_param_wqes(struct mlx5e_txqsq *sq,
+			      struct mlx5e_ktls_offload_context_tx *priv_tx,
+			      bool skip_static_post, bool fence_first_post)
+{
+	bool progress_fence = skip_static_post || !fence_first_post;
+
+	if (!skip_static_post)
+		post_static_params(sq, priv_tx, fence_first_post);
+
+	post_progress_params(sq, priv_tx, progress_fence);
+	tx_post_fence_nop(sq);
+}
+
+struct tx_sync_info {
+	u64 rcd_sn;
+	u32 sync_len;
+	int nr_frags;
+	skb_frag_t frags[MAX_SKB_FRAGS];
+};
+
+enum mlx5e_ktls_sync_retval {
+	MLX5E_KTLS_SYNC_DONE,
+	MLX5E_KTLS_SYNC_FAIL,
+	MLX5E_KTLS_SYNC_SKIP_NO_DATA,
+};
+
+static enum mlx5e_ktls_sync_retval
+tx_sync_info_get(struct mlx5e_ktls_offload_context_tx *priv_tx,
+		 u32 tcp_seq, int datalen, struct tx_sync_info *info)
+{
+	struct tls_offload_context_tx *tx_ctx = priv_tx->tx_ctx;
+	enum mlx5e_ktls_sync_retval ret = MLX5E_KTLS_SYNC_DONE;
+	struct tls_record_info *record;
+	int remaining, i = 0;
+	unsigned long flags;
+	bool ends_before;
+
+	spin_lock_irqsave(&tx_ctx->lock, flags);
+	record = tls_get_record(tx_ctx, tcp_seq, &info->rcd_sn);
+
+	if (unlikely(!record)) {
+		ret = MLX5E_KTLS_SYNC_FAIL;
+		goto out;
+	}
+
+	/* There are the following cases:
+	 * 1. packet ends before start marker: bypass offload.
+	 * 2. packet starts before start marker and ends after it: drop,
+	 *    not supported, breaks contract with kernel.
+	 * 3. packet ends before tls record info starts: drop,
+	 *    this packet was already acknowledged and its record info
+	 *    was released.
+	 */
+	ends_before = before(tcp_seq + datalen - 1, tls_record_start_seq(record));
+
+	if (unlikely(tls_record_is_start_marker(record))) {
+		ret = ends_before ? MLX5E_KTLS_SYNC_SKIP_NO_DATA : MLX5E_KTLS_SYNC_FAIL;
+		goto out;
+	} else if (ends_before) {
+		ret = MLX5E_KTLS_SYNC_FAIL;
+		goto out;
+	}
+
+	info->sync_len = tcp_seq - tls_record_start_seq(record);
+	remaining = info->sync_len;
+	while (remaining > 0) {
+		skb_frag_t *frag = &record->frags[i];
+
+		get_page(skb_frag_page(frag));
+		remaining -= skb_frag_size(frag);
+		info->frags[i++] = *frag;
+	}
+	/* reduce the part which will be sent with the original SKB */
+	if (remaining < 0)
+		skb_frag_size_add(&info->frags[i - 1], remaining);
+	info->nr_frags = i;
+out:
+	spin_unlock_irqrestore(&tx_ctx->lock, flags);
+	return ret;
+}
+
+static void
+tx_post_resync_params(struct mlx5e_txqsq *sq,
+		      struct mlx5e_ktls_offload_context_tx *priv_tx,
+		      u64 rcd_sn)
+{
+	__be64 rn_be = cpu_to_be64(rcd_sn);
+	bool skip_static_post;
+	u16 rec_seq_sz;
+	char *rec_seq;
+
+	switch (priv_tx->crypto_info.crypto_info.cipher_type) {
+	case TLS_CIPHER_AES_GCM_128: {
+		struct tls12_crypto_info_aes_gcm_128 *info = &priv_tx->crypto_info.crypto_info_128;
+
+		rec_seq = info->rec_seq;
+		rec_seq_sz = sizeof(info->rec_seq);
+		break;
+	}
+	case TLS_CIPHER_AES_GCM_256: {
+		struct tls12_crypto_info_aes_gcm_256 *info = &priv_tx->crypto_info.crypto_info_256;
+
+		rec_seq = info->rec_seq;
+		rec_seq_sz = sizeof(info->rec_seq);
+		break;
+	}
+	default:
+		WARN_ONCE(1, "Unsupported cipher type %u\n",
+			  priv_tx->crypto_info.crypto_info.cipher_type);
+		return;
+	}
+
+	skip_static_post = !memcmp(rec_seq, &rn_be, rec_seq_sz);
+	if (!skip_static_post)
+		memcpy(rec_seq, &rn_be, rec_seq_sz);
+
+	mlx5e_ktls_tx_post_param_wqes(sq, priv_tx, skip_static_post, true);
+}
+
+static int
+tx_post_resync_dump(struct mlx5e_txqsq *sq, skb_frag_t *frag, u32 tisn)
+{
+	struct mlx5_wqe_ctrl_seg *cseg;
+	struct mlx5_wqe_data_seg *dseg;
+	struct mlx5e_dump_wqe *wqe;
+	dma_addr_t dma_addr = 0;
+	u16 ds_cnt;
+	int fsz;
+	u16 pi;
+
+	BUILD_BUG_ON(MLX5E_KTLS_DUMP_WQEBBS != 1);
+	pi = mlx5_wq_cyc_ctr2ix(&sq->wq, sq->pc);
+	wqe = MLX5E_TLS_FETCH_DUMP_WQE(sq, pi);
+
+	ds_cnt = sizeof(*wqe) / MLX5_SEND_WQE_DS;
+
+	cseg = &wqe->ctrl;
+	dseg = &wqe->data;
+
+	cseg->opmod_idx_opcode = cpu_to_be32((sq->pc << 8)  | MLX5_OPCODE_DUMP);
+	cseg->qpn_ds           = cpu_to_be32((sq->sqn << 8) | ds_cnt);
+	cseg->tis_tir_num      = cpu_to_be32(tisn << 8);
+
+	fsz = skb_frag_size(frag);
+	dma_addr = skb_frag_dma_map(sq->pdev, frag, 0, fsz,
+				    DMA_TO_DEVICE);
+	if (unlikely(dma_mapping_error(sq->pdev, dma_addr)))
+		return -ENOMEM;
+
+	dseg->addr       = cpu_to_be64(dma_addr);
+	dseg->lkey       = sq->mkey_be;
+	dseg->byte_count = cpu_to_be32(fsz);
+	mlx5e_dma_push(sq, dma_addr, fsz, MLX5E_DMA_MAP_PAGE);
+
+	tx_fill_wi(sq, pi, MLX5E_KTLS_DUMP_WQEBBS, fsz, skb_frag_page(frag));
+	sq->pc += MLX5E_KTLS_DUMP_WQEBBS;
+
+	return 0;
+}
+
+void mlx5e_ktls_tx_handle_resync_dump_comp(struct mlx5e_txqsq *sq,
+					   struct mlx5e_tx_wqe_info *wi,
+					   u32 *dma_fifo_cc)
+{
+	struct mlx5e_sq_stats *stats;
+	struct mlx5e_sq_dma *dma;
+
+	dma = mlx5e_dma_get(sq, (*dma_fifo_cc)++);
+	stats = sq->stats;
+
+	mlx5e_tx_dma_unmap(sq->pdev, dma);
+	put_page(wi->resync_dump_frag_page);
+	stats->tls_dump_packets++;
+	stats->tls_dump_bytes += wi->num_bytes;
+}
+
+static enum mlx5e_ktls_sync_retval
+mlx5e_ktls_tx_handle_ooo(struct mlx5e_ktls_offload_context_tx *priv_tx,
+			 struct mlx5e_txqsq *sq,
+			 int datalen,
+			 u32 seq)
+{
+	enum mlx5e_ktls_sync_retval ret;
+	struct tx_sync_info info = {};
+	int i;
+
+	ret = tx_sync_info_get(priv_tx, seq, datalen, &info);
+	if (unlikely(ret != MLX5E_KTLS_SYNC_DONE))
+		/* We might get here with ret == FAIL if a retransmission
+		 * reaches the driver after the relevant record is acked.
+		 * It should be safe to drop the packet in this case
+		 */
+		return ret;
+
+	tx_post_resync_params(sq, priv_tx, info.rcd_sn);
+
+	for (i = 0; i < info.nr_frags; i++) {
+		unsigned int orig_fsz, frag_offset = 0, n = 0;
+		skb_frag_t *f = &info.frags[i];
+
+		orig_fsz = skb_frag_size(f);
+
+		do {
+			unsigned int fsz;
+
+			n++;
+			fsz = min_t(unsigned int, sq->hw_mtu, orig_fsz - frag_offset);
+			skb_frag_size_set(f, fsz);
+			if (tx_post_resync_dump(sq, f, priv_tx->tisn)) {
+				page_ref_add(skb_frag_page(f), n - 1);
+				goto err_out;
+			}
+
+			skb_frag_off_add(f, fsz);
+			frag_offset += fsz;
+		} while (frag_offset < orig_fsz);
+
+		page_ref_add(skb_frag_page(f), n - 1);
+	}
+
+	return MLX5E_KTLS_SYNC_DONE;
+
+err_out:
+	for (; i < info.nr_frags; i++)
+		/* The put_page() here undoes the page ref obtained in tx_sync_info_get().
+		 * Page refs obtained for the DUMP WQEs above (by page_ref_add) will be
+		 * released only upon their completions (or in mlx5e_free_txqsq_descs,
+		 * if channel closes).
+		 */
+		put_page(skb_frag_page(&info.frags[i]));
+
+	return MLX5E_KTLS_SYNC_FAIL;
+}
+
+bool mlx5e_ktls_handle_tx_skb(struct net_device *netdev, struct mlx5e_txqsq *sq,
+			      struct sk_buff *skb,
+			      struct mlx5e_accel_tx_tls_state *state)
+{
+	struct mlx5e_ktls_offload_context_tx *priv_tx;
+	struct mlx5e_sq_stats *stats = sq->stats;
+	struct net_device *tls_netdev;
+	struct tls_context *tls_ctx;
+	int datalen;
+	u32 seq;
+
+	datalen = skb->len - skb_tcp_all_headers(skb);
+	if (!datalen)
+		return true;
+
+	mlx5e_tx_mpwqe_ensure_complete(sq);
+
+	tls_ctx = tls_get_ctx(skb->sk);
+	tls_netdev = rcu_dereference_bh(tls_ctx->netdev);
+	/* Don't WARN on NULL: if tls_device_down is running in parallel,
+	 * netdev might become NULL, even if tls_is_sk_tx_device_offloaded was
+	 * true. Rather continue processing this packet.
+	 */
+	if (WARN_ON_ONCE(tls_netdev && tls_netdev != netdev))
+		goto err_out;
+
+	priv_tx = mlx5e_get_ktls_tx_priv_ctx(tls_ctx);
+
+	if (unlikely(mlx5e_ktls_tx_offload_test_and_clear_pending(priv_tx)))
+		mlx5e_ktls_tx_post_param_wqes(sq, priv_tx, false, false);
+
+	seq = ntohl(tcp_hdr(skb)->seq);
+	if (unlikely(priv_tx->expected_seq != seq)) {
+		enum mlx5e_ktls_sync_retval ret =
+			mlx5e_ktls_tx_handle_ooo(priv_tx, sq, datalen, seq);
+
+		stats->tls_ooo++;
+
+		switch (ret) {
+		case MLX5E_KTLS_SYNC_DONE:
+			break;
+		case MLX5E_KTLS_SYNC_SKIP_NO_DATA:
+			stats->tls_skip_no_sync_data++;
+			if (likely(!skb->decrypted))
+				goto out;
+			WARN_ON_ONCE(1);
+			goto err_out;
+		case MLX5E_KTLS_SYNC_FAIL:
+			stats->tls_drop_no_sync_data++;
+			goto err_out;
+		}
+	}
+
+	priv_tx->expected_seq = seq + datalen;
+
+	state->tls_tisn = priv_tx->tisn;
+
+	stats->tls_encrypted_packets += skb_is_gso(skb) ? skb_shinfo(skb)->gso_segs : 1;
+	stats->tls_encrypted_bytes   += datalen;
+
+out:
+	return true;
+
+err_out:
+	dev_kfree_skb_any(skb);
+	return false;
+}
+
+int mlx5e_ktls_init_tx(struct mlx5e_priv *priv)
+{
+	if (!mlx5e_is_ktls_tx(priv->mdev))
+		return 0;
+
+	priv->tls->tx_pool = mlx5e_tls_tx_pool_init(priv->mdev, &priv->tls->sw_stats);
+	if (!priv->tls->tx_pool)
+		return -ENOMEM;
+
+	return 0;
+}
+
+void mlx5e_ktls_cleanup_tx(struct mlx5e_priv *priv)
+{
+	if (!mlx5e_is_ktls_tx(priv->mdev))
+		return;
+
+	mlx5e_tls_tx_pool_cleanup(priv->tls->tx_pool);
+	priv->tls->tx_pool = NULL;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_txrx.c b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_txrx.c
new file mode 100644
index 000000000..570a912dd
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_txrx.c
@@ -0,0 +1,138 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2020, Mellanox Technologies inc. All rights reserved. */
+
+#include "en_accel/ktls_txrx.h"
+#include "en_accel/ktls_utils.h"
+
+enum {
+	MLX5E_STATIC_PARAMS_CONTEXT_TLS_1_2 = 0x2,
+};
+
+enum {
+	MLX5E_ENCRYPTION_STANDARD_TLS = 0x1,
+};
+
+#define EXTRACT_INFO_FIELDS do { \
+	salt    = info->salt;    \
+	rec_seq = info->rec_seq; \
+	salt_sz    = sizeof(info->salt);    \
+	rec_seq_sz = sizeof(info->rec_seq); \
+} while (0)
+
+static void
+fill_static_params(struct mlx5_wqe_tls_static_params_seg *params,
+		   union mlx5e_crypto_info *crypto_info,
+		   u32 key_id, u32 resync_tcp_sn)
+{
+	char *initial_rn, *gcm_iv;
+	u16 salt_sz, rec_seq_sz;
+	char *salt, *rec_seq;
+	u8 tls_version;
+	u8 *ctx;
+
+	ctx = params->ctx;
+
+	switch (crypto_info->crypto_info.cipher_type) {
+	case TLS_CIPHER_AES_GCM_128: {
+		struct tls12_crypto_info_aes_gcm_128 *info =
+			&crypto_info->crypto_info_128;
+
+		EXTRACT_INFO_FIELDS;
+		break;
+	}
+	case TLS_CIPHER_AES_GCM_256: {
+		struct tls12_crypto_info_aes_gcm_256 *info =
+			&crypto_info->crypto_info_256;
+
+		EXTRACT_INFO_FIELDS;
+		break;
+	}
+	default:
+		WARN_ONCE(1, "Unsupported cipher type %u\n",
+			  crypto_info->crypto_info.cipher_type);
+		return;
+	}
+
+	gcm_iv      = MLX5_ADDR_OF(tls_static_params, ctx, gcm_iv);
+	initial_rn  = MLX5_ADDR_OF(tls_static_params, ctx, initial_record_number);
+
+	memcpy(gcm_iv,      salt,    salt_sz);
+	memcpy(initial_rn,  rec_seq, rec_seq_sz);
+
+	tls_version = MLX5E_STATIC_PARAMS_CONTEXT_TLS_1_2;
+
+	MLX5_SET(tls_static_params, ctx, tls_version, tls_version);
+	MLX5_SET(tls_static_params, ctx, const_1, 1);
+	MLX5_SET(tls_static_params, ctx, const_2, 2);
+	MLX5_SET(tls_static_params, ctx, encryption_standard,
+		 MLX5E_ENCRYPTION_STANDARD_TLS);
+	MLX5_SET(tls_static_params, ctx, resync_tcp_sn, resync_tcp_sn);
+	MLX5_SET(tls_static_params, ctx, dek_index, key_id);
+}
+
+void
+mlx5e_ktls_build_static_params(struct mlx5e_set_tls_static_params_wqe *wqe,
+			       u16 pc, u32 sqn,
+			       union mlx5e_crypto_info *crypto_info,
+			       u32 tis_tir_num, u32 key_id, u32 resync_tcp_sn,
+			       bool fence, enum tls_offload_ctx_dir direction)
+{
+	struct mlx5_wqe_umr_ctrl_seg *ucseg = &wqe->uctrl;
+	struct mlx5_wqe_ctrl_seg     *cseg  = &wqe->ctrl;
+	u8 opmod = direction == TLS_OFFLOAD_CTX_DIR_TX ?
+		MLX5_OPC_MOD_TLS_TIS_STATIC_PARAMS :
+		MLX5_OPC_MOD_TLS_TIR_STATIC_PARAMS;
+
+#define STATIC_PARAMS_DS_CNT DIV_ROUND_UP(sizeof(*wqe), MLX5_SEND_WQE_DS)
+
+	cseg->opmod_idx_opcode = cpu_to_be32((pc << 8) | MLX5_OPCODE_UMR | (opmod << 24));
+	cseg->qpn_ds           = cpu_to_be32((sqn << MLX5_WQE_CTRL_QPN_SHIFT) |
+					     STATIC_PARAMS_DS_CNT);
+	cseg->fm_ce_se         = fence ? MLX5_FENCE_MODE_INITIATOR_SMALL : 0;
+	cseg->tis_tir_num      = cpu_to_be32(tis_tir_num << 8);
+
+	ucseg->flags = MLX5_UMR_INLINE;
+	ucseg->bsf_octowords = cpu_to_be16(MLX5_ST_SZ_BYTES(tls_static_params) / 16);
+
+	fill_static_params(&wqe->params, crypto_info, key_id, resync_tcp_sn);
+}
+
+static void
+fill_progress_params(struct mlx5_wqe_tls_progress_params_seg *params, u32 tis_tir_num,
+		     u32 next_record_tcp_sn)
+{
+	u8 *ctx = params->ctx;
+
+	params->tis_tir_num = cpu_to_be32(tis_tir_num);
+
+	MLX5_SET(tls_progress_params, ctx, next_record_tcp_sn,
+		 next_record_tcp_sn);
+	MLX5_SET(tls_progress_params, ctx, record_tracker_state,
+		 MLX5E_TLS_PROGRESS_PARAMS_RECORD_TRACKER_STATE_START);
+	MLX5_SET(tls_progress_params, ctx, auth_state,
+		 MLX5E_TLS_PROGRESS_PARAMS_AUTH_STATE_NO_OFFLOAD);
+}
+
+void
+mlx5e_ktls_build_progress_params(struct mlx5e_set_tls_progress_params_wqe *wqe,
+				 u16 pc, u32 sqn,
+				 u32 tis_tir_num, bool fence,
+				 u32 next_record_tcp_sn,
+				 enum tls_offload_ctx_dir direction)
+{
+	struct mlx5_wqe_ctrl_seg *cseg = &wqe->ctrl;
+	u8 opmod = direction == TLS_OFFLOAD_CTX_DIR_TX ?
+		MLX5_OPC_MOD_TLS_TIS_PROGRESS_PARAMS :
+		MLX5_OPC_MOD_TLS_TIR_PROGRESS_PARAMS;
+
+#define PROGRESS_PARAMS_DS_CNT DIV_ROUND_UP(sizeof(*wqe), MLX5_SEND_WQE_DS)
+
+	cseg->opmod_idx_opcode =
+		cpu_to_be32((pc << 8) | MLX5_OPCODE_SET_PSV | (opmod << 24));
+	cseg->qpn_ds           = cpu_to_be32((sqn << MLX5_WQE_CTRL_QPN_SHIFT) |
+					     PROGRESS_PARAMS_DS_CNT);
+	cseg->fm_ce_se         = fence ? MLX5_FENCE_MODE_INITIATOR_SMALL : 0;
+
+	fill_progress_params(&wqe->params, tis_tir_num, next_record_tcp_sn);
+}
+
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_txrx.h b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_txrx.h
new file mode 100644
index 000000000..2dd78dd4a
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_txrx.h
@@ -0,0 +1,98 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2020, Mellanox Technologies inc. All rights reserved. */
+
+#ifndef __MLX5E_KTLS_TXRX_H__
+#define __MLX5E_KTLS_TXRX_H__
+
+#ifdef CONFIG_MLX5_EN_TLS
+
+#include <net/tls.h>
+#include "en.h"
+#include "en/txrx.h"
+
+struct mlx5e_accel_tx_tls_state {
+	u32 tls_tisn;
+};
+
+u16 mlx5e_ktls_get_stop_room(struct mlx5_core_dev *mdev, struct mlx5e_params *params);
+
+bool mlx5e_ktls_handle_tx_skb(struct net_device *netdev, struct mlx5e_txqsq *sq,
+			      struct sk_buff *skb,
+			      struct mlx5e_accel_tx_tls_state *state);
+void mlx5e_ktls_handle_rx_skb(struct mlx5e_rq *rq, struct sk_buff *skb,
+			      struct mlx5_cqe64 *cqe, u32 *cqe_bcnt);
+
+void mlx5e_ktls_handle_ctx_completion(struct mlx5e_icosq_wqe_info *wi);
+void mlx5e_ktls_handle_get_psv_completion(struct mlx5e_icosq_wqe_info *wi,
+					  struct mlx5e_icosq *sq);
+
+void mlx5e_ktls_tx_handle_resync_dump_comp(struct mlx5e_txqsq *sq,
+					   struct mlx5e_tx_wqe_info *wi,
+					   u32 *dma_fifo_cc);
+static inline bool
+mlx5e_ktls_tx_try_handle_resync_dump_comp(struct mlx5e_txqsq *sq,
+					  struct mlx5e_tx_wqe_info *wi,
+					  u32 *dma_fifo_cc)
+{
+	if (unlikely(wi->resync_dump_frag_page)) {
+		mlx5e_ktls_tx_handle_resync_dump_comp(sq, wi, dma_fifo_cc);
+		return true;
+	}
+	return false;
+}
+
+bool mlx5e_ktls_rx_handle_resync_list(struct mlx5e_channel *c, int budget);
+
+static inline bool
+mlx5e_ktls_rx_pending_resync_list(struct mlx5e_channel *c, int budget)
+{
+	return budget && test_bit(MLX5E_SQ_STATE_PENDING_TLS_RX_RESYNC, &c->async_icosq.state);
+}
+
+static inline bool mlx5e_ktls_skb_offloaded(struct sk_buff *skb)
+{
+	return skb->sk && tls_is_sk_tx_device_offloaded(skb->sk);
+}
+
+static inline void
+mlx5e_ktls_handle_tx_wqe(struct mlx5_wqe_ctrl_seg *cseg,
+			 struct mlx5e_accel_tx_tls_state *state)
+{
+	cseg->tis_tir_num = cpu_to_be32(state->tls_tisn << 8);
+}
+#else
+static inline bool
+mlx5e_ktls_tx_try_handle_resync_dump_comp(struct mlx5e_txqsq *sq,
+					  struct mlx5e_tx_wqe_info *wi,
+					  u32 *dma_fifo_cc)
+{
+	return false;
+}
+
+static inline bool
+mlx5e_ktls_rx_handle_resync_list(struct mlx5e_channel *c, int budget)
+{
+	return false;
+}
+
+static inline bool
+mlx5e_ktls_rx_pending_resync_list(struct mlx5e_channel *c, int budget)
+{
+	return false;
+}
+
+static inline u16 mlx5e_ktls_get_stop_room(struct mlx5_core_dev *mdev,
+					   struct mlx5e_params *params)
+{
+	return 0;
+}
+
+static inline void mlx5e_ktls_handle_rx_skb(struct mlx5e_rq *rq,
+					    struct sk_buff *skb,
+					    struct mlx5_cqe64 *cqe,
+					    u32 *cqe_bcnt)
+{
+}
+#endif /* CONFIG_MLX5_EN_TLS */
+
+#endif /* __MLX5E_TLS_TXRX_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_utils.h b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_utils.h
new file mode 100644
index 000000000..3d79cd379
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_utils.h
@@ -0,0 +1,91 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2020, Mellanox Technologies inc. All rights reserved. */
+
+#ifndef __MLX5E_KTLS_UTILS_H__
+#define __MLX5E_KTLS_UTILS_H__
+
+#include <net/tls.h>
+#include "en.h"
+
+enum {
+	MLX5E_TLS_PROGRESS_PARAMS_AUTH_STATE_NO_OFFLOAD     = 0,
+	MLX5E_TLS_PROGRESS_PARAMS_AUTH_STATE_OFFLOAD        = 1,
+	MLX5E_TLS_PROGRESS_PARAMS_AUTH_STATE_AUTHENTICATION = 2,
+};
+
+enum {
+	MLX5E_TLS_PROGRESS_PARAMS_RECORD_TRACKER_STATE_START     = 0,
+	MLX5E_TLS_PROGRESS_PARAMS_RECORD_TRACKER_STATE_TRACKING  = 1,
+	MLX5E_TLS_PROGRESS_PARAMS_RECORD_TRACKER_STATE_SEARCHING = 2,
+};
+
+int mlx5e_ktls_add_tx(struct net_device *netdev, struct sock *sk,
+		      struct tls_crypto_info *crypto_info, u32 start_offload_tcp_sn);
+void mlx5e_ktls_del_tx(struct net_device *netdev, struct tls_context *tls_ctx);
+int mlx5e_ktls_add_rx(struct net_device *netdev, struct sock *sk,
+		      struct tls_crypto_info *crypto_info, u32 start_offload_tcp_sn);
+void mlx5e_ktls_del_rx(struct net_device *netdev, struct tls_context *tls_ctx);
+void mlx5e_ktls_rx_resync(struct net_device *netdev, struct sock *sk, u32 seq, u8 *rcd_sn);
+
+union mlx5e_crypto_info {
+	struct tls_crypto_info crypto_info;
+	struct tls12_crypto_info_aes_gcm_128 crypto_info_128;
+	struct tls12_crypto_info_aes_gcm_256 crypto_info_256;
+};
+
+struct mlx5e_set_tls_static_params_wqe {
+	struct mlx5_wqe_ctrl_seg ctrl;
+	struct mlx5_wqe_umr_ctrl_seg uctrl;
+	struct mlx5_mkey_seg mkc;
+	struct mlx5_wqe_tls_static_params_seg params;
+};
+
+struct mlx5e_set_tls_progress_params_wqe {
+	struct mlx5_wqe_ctrl_seg ctrl;
+	struct mlx5_wqe_tls_progress_params_seg params;
+};
+
+struct mlx5e_get_tls_progress_params_wqe {
+	struct mlx5_wqe_ctrl_seg ctrl;
+	struct mlx5_seg_get_psv  psv;
+};
+
+#define MLX5E_TLS_SET_STATIC_PARAMS_WQEBBS \
+	(DIV_ROUND_UP(sizeof(struct mlx5e_set_tls_static_params_wqe), MLX5_SEND_WQE_BB))
+
+#define MLX5E_TLS_SET_PROGRESS_PARAMS_WQEBBS \
+	(DIV_ROUND_UP(sizeof(struct mlx5e_set_tls_progress_params_wqe), MLX5_SEND_WQE_BB))
+
+#define MLX5E_KTLS_GET_PROGRESS_WQEBBS \
+	(DIV_ROUND_UP(sizeof(struct mlx5e_get_tls_progress_params_wqe), MLX5_SEND_WQE_BB))
+
+#define MLX5E_TLS_FETCH_SET_STATIC_PARAMS_WQE(sq, pi) \
+	((struct mlx5e_set_tls_static_params_wqe *)\
+	 mlx5e_fetch_wqe(&(sq)->wq, pi, sizeof(struct mlx5e_set_tls_static_params_wqe)))
+
+#define MLX5E_TLS_FETCH_SET_PROGRESS_PARAMS_WQE(sq, pi) \
+	((struct mlx5e_set_tls_progress_params_wqe *)\
+	 mlx5e_fetch_wqe(&(sq)->wq, pi, sizeof(struct mlx5e_set_tls_progress_params_wqe)))
+
+#define MLX5E_TLS_FETCH_GET_PROGRESS_PARAMS_WQE(sq, pi) \
+	((struct mlx5e_get_tls_progress_params_wqe *)\
+	 mlx5e_fetch_wqe(&(sq)->wq, pi, sizeof(struct mlx5e_get_tls_progress_params_wqe)))
+
+#define MLX5E_TLS_FETCH_DUMP_WQE(sq, pi) \
+	((struct mlx5e_dump_wqe *)\
+	 mlx5e_fetch_wqe(&(sq)->wq, pi, sizeof(struct mlx5e_dump_wqe)))
+
+void
+mlx5e_ktls_build_static_params(struct mlx5e_set_tls_static_params_wqe *wqe,
+			       u16 pc, u32 sqn,
+			       union mlx5e_crypto_info *crypto_info,
+			       u32 tis_tir_num, u32 key_id, u32 resync_tcp_sn,
+			       bool fence, enum tls_offload_ctx_dir direction);
+void
+mlx5e_ktls_build_progress_params(struct mlx5e_set_tls_progress_params_wqe *wqe,
+				 u16 pc, u32 sqn,
+				 u32 tis_tir_num, bool fence,
+				 u32 next_record_tcp_sn,
+				 enum tls_offload_ctx_dir direction);
+
+#endif /* __MLX5E_TLS_UTILS_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec.c b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec.c
new file mode 100644
index 000000000..a7832a018
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec.c
@@ -0,0 +1,1866 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2022, NVIDIA CORPORATION & AFFILIATES. All rights reserved. */
+
+#include <linux/mlx5/device.h>
+#include <linux/mlx5/mlx5_ifc.h>
+#include <linux/xarray.h>
+
+#include "en.h"
+#include "lib/aso.h"
+#include "lib/mlx5.h"
+#include "en_accel/macsec.h"
+#include "en_accel/macsec_fs.h"
+
+#define MLX5_MACSEC_EPN_SCOPE_MID 0x80000000L
+#define MLX5E_MACSEC_ASO_CTX_SZ MLX5_ST_SZ_BYTES(macsec_aso)
+
+enum mlx5_macsec_aso_event_arm {
+	MLX5E_ASO_EPN_ARM = BIT(0),
+};
+
+enum {
+	MLX5_MACSEC_ASO_REMOVE_FLOW_PKT_CNT_OFFSET,
+};
+
+struct mlx5e_macsec_handle {
+	struct mlx5e_macsec *macsec;
+	u32 obj_id;
+	u8 idx;
+};
+
+enum {
+	MLX5_MACSEC_EPN,
+};
+
+struct mlx5e_macsec_aso_out {
+	u8 event_arm;
+	u32 mode_param;
+};
+
+struct mlx5e_macsec_aso_in {
+	u8 mode;
+	u32 obj_id;
+};
+
+struct mlx5e_macsec_epn_state {
+	u32 epn_msb;
+	u8 epn_enabled;
+	u8 overlap;
+};
+
+struct mlx5e_macsec_async_work {
+	struct mlx5e_macsec *macsec;
+	struct mlx5_core_dev *mdev;
+	struct work_struct work;
+	u32 obj_id;
+};
+
+struct mlx5e_macsec_sa {
+	bool active;
+	u8  assoc_num;
+	u32 macsec_obj_id;
+	u32 enc_key_id;
+	u32 next_pn;
+	sci_t sci;
+	ssci_t ssci;
+	salt_t salt;
+
+	struct rhash_head hash;
+	u32 fs_id;
+	union mlx5e_macsec_rule *macsec_rule;
+	struct rcu_head rcu_head;
+	struct mlx5e_macsec_epn_state epn_state;
+};
+
+struct mlx5e_macsec_rx_sc;
+struct mlx5e_macsec_rx_sc_xarray_element {
+	u32 fs_id;
+	struct mlx5e_macsec_rx_sc *rx_sc;
+};
+
+struct mlx5e_macsec_rx_sc {
+	bool active;
+	sci_t sci;
+	struct mlx5e_macsec_sa *rx_sa[MACSEC_NUM_AN];
+	struct list_head rx_sc_list_element;
+	struct mlx5e_macsec_rx_sc_xarray_element *sc_xarray_element;
+	struct metadata_dst *md_dst;
+	struct rcu_head rcu_head;
+};
+
+struct mlx5e_macsec_umr {
+	u8 __aligned(64) ctx[MLX5_ST_SZ_BYTES(macsec_aso)];
+	dma_addr_t dma_addr;
+	u32 mkey;
+};
+
+struct mlx5e_macsec_aso {
+	/* ASO */
+	struct mlx5_aso *maso;
+	/* Protects macsec ASO */
+	struct mutex aso_lock;
+	/* UMR */
+	struct mlx5e_macsec_umr *umr;
+
+	u32 pdn;
+};
+
+static const struct rhashtable_params rhash_sci = {
+	.key_len = sizeof_field(struct mlx5e_macsec_sa, sci),
+	.key_offset = offsetof(struct mlx5e_macsec_sa, sci),
+	.head_offset = offsetof(struct mlx5e_macsec_sa, hash),
+	.automatic_shrinking = true,
+	.min_size = 1,
+};
+
+struct mlx5e_macsec_device {
+	const struct net_device *netdev;
+	struct mlx5e_macsec_sa *tx_sa[MACSEC_NUM_AN];
+	struct list_head macsec_rx_sc_list_head;
+	unsigned char *dev_addr;
+	struct list_head macsec_device_list_element;
+};
+
+struct mlx5e_macsec {
+	struct list_head macsec_device_list_head;
+	int num_of_devices;
+	struct mlx5e_macsec_fs *macsec_fs;
+	struct mutex lock; /* Protects mlx5e_macsec internal contexts */
+
+	/* Tx sci -> fs id mapping handling */
+	struct rhashtable sci_hash;      /* sci -> mlx5e_macsec_sa */
+
+	/* Rx fs_id -> rx_sc mapping */
+	struct xarray sc_xarray;
+
+	struct mlx5_core_dev *mdev;
+
+	/* Stats manage */
+	struct mlx5e_macsec_stats stats;
+
+	/* ASO */
+	struct mlx5e_macsec_aso aso;
+
+	struct notifier_block nb;
+	struct workqueue_struct *wq;
+};
+
+struct mlx5_macsec_obj_attrs {
+	u32 aso_pdn;
+	u32 next_pn;
+	__be64 sci;
+	u32 enc_key_id;
+	bool encrypt;
+	struct mlx5e_macsec_epn_state epn_state;
+	salt_t salt;
+	__be32 ssci;
+	bool replay_protect;
+	u32 replay_window;
+};
+
+struct mlx5_aso_ctrl_param {
+	u8   data_mask_mode;
+	u8   condition_0_operand;
+	u8   condition_1_operand;
+	u8   condition_0_offset;
+	u8   condition_1_offset;
+	u8   data_offset;
+	u8   condition_operand;
+	u32  condition_0_data;
+	u32  condition_0_mask;
+	u32  condition_1_data;
+	u32  condition_1_mask;
+	u64  bitwise_data;
+	u64  data_mask;
+};
+
+static int mlx5e_macsec_aso_reg_mr(struct mlx5_core_dev *mdev, struct mlx5e_macsec_aso *aso)
+{
+	struct mlx5e_macsec_umr *umr;
+	struct device *dma_device;
+	dma_addr_t dma_addr;
+	int err;
+
+	umr = kzalloc(sizeof(*umr), GFP_KERNEL);
+	if (!umr) {
+		err = -ENOMEM;
+		return err;
+	}
+
+	dma_device = &mdev->pdev->dev;
+	dma_addr = dma_map_single(dma_device, umr->ctx, sizeof(umr->ctx), DMA_BIDIRECTIONAL);
+	err = dma_mapping_error(dma_device, dma_addr);
+	if (err) {
+		mlx5_core_err(mdev, "Can't map dma device, err=%d\n", err);
+		goto out_dma;
+	}
+
+	err = mlx5e_create_mkey(mdev, aso->pdn, &umr->mkey);
+	if (err) {
+		mlx5_core_err(mdev, "Can't create mkey, err=%d\n", err);
+		goto out_mkey;
+	}
+
+	umr->dma_addr = dma_addr;
+
+	aso->umr = umr;
+
+	return 0;
+
+out_mkey:
+	dma_unmap_single(dma_device, dma_addr, sizeof(umr->ctx), DMA_BIDIRECTIONAL);
+out_dma:
+	kfree(umr);
+	return err;
+}
+
+static void mlx5e_macsec_aso_dereg_mr(struct mlx5_core_dev *mdev, struct mlx5e_macsec_aso *aso)
+{
+	struct mlx5e_macsec_umr *umr = aso->umr;
+
+	mlx5_core_destroy_mkey(mdev, umr->mkey);
+	dma_unmap_single(&mdev->pdev->dev, umr->dma_addr, sizeof(umr->ctx), DMA_BIDIRECTIONAL);
+	kfree(umr);
+}
+
+static int macsec_set_replay_protection(struct mlx5_macsec_obj_attrs *attrs, void *aso_ctx)
+{
+	u8 window_sz;
+
+	if (!attrs->replay_protect)
+		return 0;
+
+	switch (attrs->replay_window) {
+	case 256:
+		window_sz = MLX5_MACSEC_ASO_REPLAY_WIN_256BIT;
+		break;
+	case 128:
+		window_sz = MLX5_MACSEC_ASO_REPLAY_WIN_128BIT;
+		break;
+	case 64:
+		window_sz = MLX5_MACSEC_ASO_REPLAY_WIN_64BIT;
+		break;
+	case 32:
+		window_sz = MLX5_MACSEC_ASO_REPLAY_WIN_32BIT;
+		break;
+	default:
+		return -EINVAL;
+	}
+	MLX5_SET(macsec_aso, aso_ctx, window_size, window_sz);
+	MLX5_SET(macsec_aso, aso_ctx, mode, MLX5_MACSEC_ASO_REPLAY_PROTECTION);
+
+	return 0;
+}
+
+static int mlx5e_macsec_create_object(struct mlx5_core_dev *mdev,
+				      struct mlx5_macsec_obj_attrs *attrs,
+				      bool is_tx,
+				      u32 *macsec_obj_id)
+{
+	u32 in[MLX5_ST_SZ_DW(create_macsec_obj_in)] = {};
+	u32 out[MLX5_ST_SZ_DW(general_obj_out_cmd_hdr)];
+	void *aso_ctx;
+	void *obj;
+	int err;
+
+	obj = MLX5_ADDR_OF(create_macsec_obj_in, in, macsec_object);
+	aso_ctx = MLX5_ADDR_OF(macsec_offload_obj, obj, macsec_aso);
+
+	MLX5_SET(macsec_offload_obj, obj, confidentiality_en, attrs->encrypt);
+	MLX5_SET(macsec_offload_obj, obj, dekn, attrs->enc_key_id);
+	MLX5_SET(macsec_offload_obj, obj, aso_return_reg, MLX5_MACSEC_ASO_REG_C_4_5);
+	MLX5_SET(macsec_offload_obj, obj, macsec_aso_access_pd, attrs->aso_pdn);
+	MLX5_SET(macsec_aso, aso_ctx, mode_parameter, attrs->next_pn);
+
+	/* Epn */
+	if (attrs->epn_state.epn_enabled) {
+		void *salt_p;
+		int i;
+
+		MLX5_SET(macsec_aso, aso_ctx, epn_event_arm, 1);
+		MLX5_SET(macsec_offload_obj, obj, epn_en, 1);
+		MLX5_SET(macsec_offload_obj, obj, epn_msb, attrs->epn_state.epn_msb);
+		MLX5_SET(macsec_offload_obj, obj, epn_overlap, attrs->epn_state.overlap);
+		MLX5_SET64(macsec_offload_obj, obj, sci, (__force u64)attrs->ssci);
+		salt_p = MLX5_ADDR_OF(macsec_offload_obj, obj, salt);
+		for (i = 0; i < 3 ; i++)
+			memcpy((u32 *)salt_p + i, &attrs->salt.bytes[4 * (2 - i)], 4);
+	} else {
+		MLX5_SET64(macsec_offload_obj, obj, sci, (__force u64)(attrs->sci));
+	}
+
+	MLX5_SET(macsec_aso, aso_ctx, valid, 0x1);
+	if (is_tx) {
+		MLX5_SET(macsec_aso, aso_ctx, mode, MLX5_MACSEC_ASO_INC_SN);
+	} else {
+		err = macsec_set_replay_protection(attrs, aso_ctx);
+		if (err)
+			return err;
+	}
+
+	/* general object fields set */
+	MLX5_SET(general_obj_in_cmd_hdr, in, opcode, MLX5_CMD_OP_CREATE_GENERAL_OBJECT);
+	MLX5_SET(general_obj_in_cmd_hdr, in, obj_type, MLX5_GENERAL_OBJECT_TYPES_MACSEC);
+
+	err = mlx5_cmd_exec(mdev, in, sizeof(in), out, sizeof(out));
+	if (err) {
+		mlx5_core_err(mdev,
+			      "MACsec offload: Failed to create MACsec object (err = %d)\n",
+			      err);
+		return err;
+	}
+
+	*macsec_obj_id = MLX5_GET(general_obj_out_cmd_hdr, out, obj_id);
+
+	return err;
+}
+
+static void mlx5e_macsec_destroy_object(struct mlx5_core_dev *mdev, u32 macsec_obj_id)
+{
+	u32 in[MLX5_ST_SZ_DW(general_obj_in_cmd_hdr)] = {};
+	u32 out[MLX5_ST_SZ_DW(general_obj_out_cmd_hdr)];
+
+	MLX5_SET(general_obj_in_cmd_hdr, in, opcode, MLX5_CMD_OP_DESTROY_GENERAL_OBJECT);
+	MLX5_SET(general_obj_in_cmd_hdr, in, obj_type, MLX5_GENERAL_OBJECT_TYPES_MACSEC);
+	MLX5_SET(general_obj_in_cmd_hdr, in, obj_id, macsec_obj_id);
+
+	mlx5_cmd_exec(mdev, in, sizeof(in), out, sizeof(out));
+}
+
+static void mlx5e_macsec_cleanup_sa(struct mlx5e_macsec *macsec,
+				    struct mlx5e_macsec_sa *sa,
+				    bool is_tx)
+{
+	int action =  (is_tx) ?  MLX5_ACCEL_MACSEC_ACTION_ENCRYPT :
+				 MLX5_ACCEL_MACSEC_ACTION_DECRYPT;
+
+	if ((is_tx) && sa->fs_id) {
+		/* Make sure ongoing datapath readers sees a valid SA */
+		rhashtable_remove_fast(&macsec->sci_hash, &sa->hash, rhash_sci);
+		sa->fs_id = 0;
+	}
+
+	if (!sa->macsec_rule)
+		return;
+
+	mlx5e_macsec_fs_del_rule(macsec->macsec_fs, sa->macsec_rule, action);
+	mlx5e_macsec_destroy_object(macsec->mdev, sa->macsec_obj_id);
+	sa->macsec_rule = NULL;
+}
+
+static int mlx5e_macsec_init_sa(struct macsec_context *ctx,
+				struct mlx5e_macsec_sa *sa,
+				bool encrypt,
+				bool is_tx)
+{
+	struct mlx5e_priv *priv = netdev_priv(ctx->netdev);
+	struct mlx5e_macsec *macsec = priv->macsec;
+	struct mlx5_macsec_rule_attrs rule_attrs;
+	struct mlx5_core_dev *mdev = priv->mdev;
+	struct mlx5_macsec_obj_attrs obj_attrs;
+	union mlx5e_macsec_rule *macsec_rule;
+	int err;
+
+	obj_attrs.next_pn = sa->next_pn;
+	obj_attrs.sci = cpu_to_be64((__force u64)sa->sci);
+	obj_attrs.enc_key_id = sa->enc_key_id;
+	obj_attrs.encrypt = encrypt;
+	obj_attrs.aso_pdn = macsec->aso.pdn;
+	obj_attrs.epn_state = sa->epn_state;
+
+	if (sa->epn_state.epn_enabled) {
+		obj_attrs.ssci = cpu_to_be32((__force u32)sa->ssci);
+		memcpy(&obj_attrs.salt, &sa->salt, sizeof(sa->salt));
+	}
+
+	obj_attrs.replay_window = ctx->secy->replay_window;
+	obj_attrs.replay_protect = ctx->secy->replay_protect;
+
+	err = mlx5e_macsec_create_object(mdev, &obj_attrs, is_tx, &sa->macsec_obj_id);
+	if (err)
+		return err;
+
+	rule_attrs.macsec_obj_id = sa->macsec_obj_id;
+	rule_attrs.sci = sa->sci;
+	rule_attrs.assoc_num = sa->assoc_num;
+	rule_attrs.action = (is_tx) ? MLX5_ACCEL_MACSEC_ACTION_ENCRYPT :
+				      MLX5_ACCEL_MACSEC_ACTION_DECRYPT;
+
+	macsec_rule = mlx5e_macsec_fs_add_rule(macsec->macsec_fs, ctx, &rule_attrs, &sa->fs_id);
+	if (!macsec_rule) {
+		err = -ENOMEM;
+		goto destroy_macsec_object;
+	}
+
+	sa->macsec_rule = macsec_rule;
+
+	if (is_tx) {
+		err = rhashtable_insert_fast(&macsec->sci_hash, &sa->hash, rhash_sci);
+		if (err)
+			goto destroy_macsec_object_and_rule;
+	}
+
+	return 0;
+
+destroy_macsec_object_and_rule:
+	mlx5e_macsec_cleanup_sa(macsec, sa, is_tx);
+destroy_macsec_object:
+	mlx5e_macsec_destroy_object(mdev, sa->macsec_obj_id);
+
+	return err;
+}
+
+static struct mlx5e_macsec_rx_sc *
+mlx5e_macsec_get_rx_sc_from_sc_list(const struct list_head *list, sci_t sci)
+{
+	struct mlx5e_macsec_rx_sc *iter;
+
+	list_for_each_entry_rcu(iter, list, rx_sc_list_element) {
+		if (iter->sci == sci)
+			return iter;
+	}
+
+	return NULL;
+}
+
+static int macsec_rx_sa_active_update(struct macsec_context *ctx,
+				      struct mlx5e_macsec_sa *rx_sa,
+				      bool active)
+{
+	struct mlx5e_priv *priv = netdev_priv(ctx->netdev);
+	struct mlx5e_macsec *macsec = priv->macsec;
+	int err = 0;
+
+	if (rx_sa->active == active)
+		return 0;
+
+	rx_sa->active = active;
+	if (!active) {
+		mlx5e_macsec_cleanup_sa(macsec, rx_sa, false);
+		return 0;
+	}
+
+	err = mlx5e_macsec_init_sa(ctx, rx_sa, true, false);
+	if (err)
+		rx_sa->active = false;
+
+	return err;
+}
+
+static bool mlx5e_macsec_secy_features_validate(struct macsec_context *ctx)
+{
+	const struct net_device *netdev = ctx->netdev;
+	const struct macsec_secy *secy = ctx->secy;
+
+	if (secy->validate_frames != MACSEC_VALIDATE_STRICT) {
+		netdev_err(netdev,
+			   "MACsec offload is supported only when validate_frame is in strict mode\n");
+		return false;
+	}
+
+	if (secy->icv_len != MACSEC_DEFAULT_ICV_LEN) {
+		netdev_err(netdev, "MACsec offload is supported only when icv_len is %d\n",
+			   MACSEC_DEFAULT_ICV_LEN);
+		return false;
+	}
+
+	if (!secy->protect_frames) {
+		netdev_err(netdev,
+			   "MACsec offload is supported only when protect_frames is set\n");
+		return false;
+	}
+
+	if (!ctx->secy->tx_sc.encrypt) {
+		netdev_err(netdev, "MACsec offload: encrypt off isn't supported\n");
+		return false;
+	}
+
+	return true;
+}
+
+static struct mlx5e_macsec_device *
+mlx5e_macsec_get_macsec_device_context(const struct mlx5e_macsec *macsec,
+				       const struct macsec_context *ctx)
+{
+	struct mlx5e_macsec_device *iter;
+	const struct list_head *list;
+
+	list = &macsec->macsec_device_list_head;
+	list_for_each_entry_rcu(iter, list, macsec_device_list_element) {
+		if (iter->netdev == ctx->secy->netdev)
+			return iter;
+	}
+
+	return NULL;
+}
+
+static void update_macsec_epn(struct mlx5e_macsec_sa *sa, const struct macsec_key *key,
+			      const pn_t *next_pn_halves, ssci_t ssci)
+{
+	struct mlx5e_macsec_epn_state *epn_state = &sa->epn_state;
+
+	sa->ssci = ssci;
+	sa->salt = key->salt;
+	epn_state->epn_enabled = 1;
+	epn_state->epn_msb = next_pn_halves->upper;
+	epn_state->overlap = next_pn_halves->lower < MLX5_MACSEC_EPN_SCOPE_MID ? 0 : 1;
+}
+
+static int mlx5e_macsec_add_txsa(struct macsec_context *ctx)
+{
+	const struct macsec_tx_sc *tx_sc = &ctx->secy->tx_sc;
+	const struct macsec_tx_sa *ctx_tx_sa = ctx->sa.tx_sa;
+	struct mlx5e_priv *priv = netdev_priv(ctx->netdev);
+	const struct macsec_secy *secy = ctx->secy;
+	struct mlx5e_macsec_device *macsec_device;
+	struct mlx5_core_dev *mdev = priv->mdev;
+	u8 assoc_num = ctx->sa.assoc_num;
+	struct mlx5e_macsec_sa *tx_sa;
+	struct mlx5e_macsec *macsec;
+	int err = 0;
+
+	mutex_lock(&priv->macsec->lock);
+
+	macsec = priv->macsec;
+	macsec_device = mlx5e_macsec_get_macsec_device_context(macsec, ctx);
+	if (!macsec_device) {
+		netdev_err(ctx->netdev, "MACsec offload: Failed to find device context\n");
+		err = -EEXIST;
+		goto out;
+	}
+
+	if (macsec_device->tx_sa[assoc_num]) {
+		netdev_err(ctx->netdev, "MACsec offload tx_sa: %d already exist\n", assoc_num);
+		err = -EEXIST;
+		goto out;
+	}
+
+	tx_sa = kzalloc(sizeof(*tx_sa), GFP_KERNEL);
+	if (!tx_sa) {
+		err = -ENOMEM;
+		goto out;
+	}
+
+	tx_sa->active = ctx_tx_sa->active;
+	tx_sa->next_pn = ctx_tx_sa->next_pn_halves.lower;
+	tx_sa->sci = secy->sci;
+	tx_sa->assoc_num = assoc_num;
+
+	if (secy->xpn)
+		update_macsec_epn(tx_sa, &ctx_tx_sa->key, &ctx_tx_sa->next_pn_halves,
+				  ctx_tx_sa->ssci);
+
+	err = mlx5_create_encryption_key(mdev, ctx->sa.key, secy->key_len,
+					 MLX5_ACCEL_OBJ_MACSEC_KEY,
+					 &tx_sa->enc_key_id);
+	if (err)
+		goto destroy_sa;
+
+	macsec_device->tx_sa[assoc_num] = tx_sa;
+	if (!secy->operational ||
+	    assoc_num != tx_sc->encoding_sa ||
+	    !tx_sa->active)
+		goto out;
+
+	err = mlx5e_macsec_init_sa(ctx, tx_sa, tx_sc->encrypt, true);
+	if (err)
+		goto destroy_encryption_key;
+
+	mutex_unlock(&macsec->lock);
+
+	return 0;
+
+destroy_encryption_key:
+	macsec_device->tx_sa[assoc_num] = NULL;
+	mlx5_destroy_encryption_key(mdev, tx_sa->enc_key_id);
+destroy_sa:
+	kfree(tx_sa);
+out:
+	mutex_unlock(&macsec->lock);
+
+	return err;
+}
+
+static int mlx5e_macsec_upd_txsa(struct macsec_context *ctx)
+{
+	const struct macsec_tx_sc *tx_sc = &ctx->secy->tx_sc;
+	const struct macsec_tx_sa *ctx_tx_sa = ctx->sa.tx_sa;
+	struct mlx5e_priv *priv = netdev_priv(ctx->netdev);
+	struct mlx5e_macsec_device *macsec_device;
+	u8 assoc_num = ctx->sa.assoc_num;
+	struct mlx5e_macsec_sa *tx_sa;
+	struct mlx5e_macsec *macsec;
+	struct net_device *netdev;
+	int err = 0;
+
+	mutex_lock(&priv->macsec->lock);
+
+	macsec = priv->macsec;
+	netdev = ctx->netdev;
+	macsec_device = mlx5e_macsec_get_macsec_device_context(macsec, ctx);
+	if (!macsec_device) {
+		netdev_err(netdev, "MACsec offload: Failed to find device context\n");
+		err = -EINVAL;
+		goto out;
+	}
+
+	tx_sa = macsec_device->tx_sa[assoc_num];
+	if (!tx_sa) {
+		netdev_err(netdev, "MACsec offload: TX sa 0x%x doesn't exist\n", assoc_num);
+		err = -EEXIST;
+		goto out;
+	}
+
+	if (ctx->sa.update_pn) {
+		netdev_err(netdev, "MACsec offload: update TX sa %d PN isn't supported\n",
+			   assoc_num);
+		err = -EINVAL;
+		goto out;
+	}
+
+	if (tx_sa->active == ctx_tx_sa->active)
+		goto out;
+
+	tx_sa->active = ctx_tx_sa->active;
+	if (tx_sa->assoc_num != tx_sc->encoding_sa)
+		goto out;
+
+	if (ctx_tx_sa->active) {
+		err = mlx5e_macsec_init_sa(ctx, tx_sa, tx_sc->encrypt, true);
+		if (err)
+			goto out;
+	} else {
+		if (!tx_sa->macsec_rule) {
+			err = -EINVAL;
+			goto out;
+		}
+
+		mlx5e_macsec_cleanup_sa(macsec, tx_sa, true);
+	}
+out:
+	mutex_unlock(&macsec->lock);
+
+	return err;
+}
+
+static int mlx5e_macsec_del_txsa(struct macsec_context *ctx)
+{
+	struct mlx5e_priv *priv = netdev_priv(ctx->netdev);
+	struct mlx5e_macsec_device *macsec_device;
+	u8 assoc_num = ctx->sa.assoc_num;
+	struct mlx5e_macsec_sa *tx_sa;
+	struct mlx5e_macsec *macsec;
+	int err = 0;
+
+	mutex_lock(&priv->macsec->lock);
+	macsec = priv->macsec;
+	macsec_device = mlx5e_macsec_get_macsec_device_context(macsec, ctx);
+	if (!macsec_device) {
+		netdev_err(ctx->netdev, "MACsec offload: Failed to find device context\n");
+		err = -EINVAL;
+		goto out;
+	}
+
+	tx_sa = macsec_device->tx_sa[assoc_num];
+	if (!tx_sa) {
+		netdev_err(ctx->netdev, "MACsec offload: TX sa 0x%x doesn't exist\n", assoc_num);
+		err = -EEXIST;
+		goto out;
+	}
+
+	mlx5e_macsec_cleanup_sa(macsec, tx_sa, true);
+	mlx5_destroy_encryption_key(macsec->mdev, tx_sa->enc_key_id);
+	kfree_rcu(tx_sa);
+	macsec_device->tx_sa[assoc_num] = NULL;
+
+out:
+	mutex_unlock(&macsec->lock);
+
+	return err;
+}
+
+static u32 mlx5e_macsec_get_sa_from_hashtable(struct rhashtable *sci_hash, sci_t *sci)
+{
+	struct mlx5e_macsec_sa *macsec_sa;
+	u32 fs_id = 0;
+
+	rcu_read_lock();
+	macsec_sa = rhashtable_lookup(sci_hash, sci, rhash_sci);
+	if (macsec_sa)
+		fs_id = macsec_sa->fs_id;
+	rcu_read_unlock();
+
+	return fs_id;
+}
+
+static int mlx5e_macsec_add_rxsc(struct macsec_context *ctx)
+{
+	struct mlx5e_macsec_rx_sc_xarray_element *sc_xarray_element;
+	struct mlx5e_priv *priv = netdev_priv(ctx->netdev);
+	const struct macsec_rx_sc *ctx_rx_sc = ctx->rx_sc;
+	struct mlx5e_macsec_device *macsec_device;
+	struct mlx5e_macsec_rx_sc *rx_sc;
+	struct list_head *rx_sc_list;
+	struct mlx5e_macsec *macsec;
+	int err = 0;
+
+	mutex_lock(&priv->macsec->lock);
+	macsec = priv->macsec;
+	macsec_device = mlx5e_macsec_get_macsec_device_context(macsec, ctx);
+	if (!macsec_device) {
+		netdev_err(ctx->netdev, "MACsec offload: Failed to find device context\n");
+		err = -EINVAL;
+		goto out;
+	}
+
+	rx_sc_list = &macsec_device->macsec_rx_sc_list_head;
+	rx_sc = mlx5e_macsec_get_rx_sc_from_sc_list(rx_sc_list, ctx_rx_sc->sci);
+	if (rx_sc) {
+		netdev_err(ctx->netdev, "MACsec offload: rx_sc (sci %lld) already exists\n",
+			   ctx_rx_sc->sci);
+		err = -EEXIST;
+		goto out;
+	}
+
+	rx_sc = kzalloc(sizeof(*rx_sc), GFP_KERNEL);
+	if (!rx_sc) {
+		err = -ENOMEM;
+		goto out;
+	}
+
+	sc_xarray_element = kzalloc(sizeof(*sc_xarray_element), GFP_KERNEL);
+	if (!sc_xarray_element) {
+		err = -ENOMEM;
+		goto destroy_rx_sc;
+	}
+
+	sc_xarray_element->rx_sc = rx_sc;
+	err = xa_alloc(&macsec->sc_xarray, &sc_xarray_element->fs_id, sc_xarray_element,
+		       XA_LIMIT(1, MLX5_MACEC_RX_FS_ID_MAX), GFP_KERNEL);
+	if (err) {
+		if (err == -EBUSY)
+			netdev_err(ctx->netdev,
+				   "MACsec offload: unable to create entry for RX SC (%d Rx SCs already allocated)\n",
+				   MLX5_MACEC_RX_FS_ID_MAX);
+		goto destroy_sc_xarray_elemenet;
+	}
+
+	rx_sc->md_dst = metadata_dst_alloc(0, METADATA_MACSEC, GFP_KERNEL);
+	if (!rx_sc->md_dst) {
+		err = -ENOMEM;
+		goto erase_xa_alloc;
+	}
+
+	rx_sc->sci = ctx_rx_sc->sci;
+	rx_sc->active = ctx_rx_sc->active;
+	list_add_rcu(&rx_sc->rx_sc_list_element, rx_sc_list);
+
+	rx_sc->sc_xarray_element = sc_xarray_element;
+	rx_sc->md_dst->u.macsec_info.sci = rx_sc->sci;
+	mutex_unlock(&macsec->lock);
+
+	return 0;
+
+erase_xa_alloc:
+	xa_erase(&macsec->sc_xarray, sc_xarray_element->fs_id);
+destroy_sc_xarray_elemenet:
+	kfree(sc_xarray_element);
+destroy_rx_sc:
+	kfree(rx_sc);
+
+out:
+	mutex_unlock(&macsec->lock);
+
+	return err;
+}
+
+static int mlx5e_macsec_upd_rxsc(struct macsec_context *ctx)
+{
+	struct mlx5e_priv *priv = netdev_priv(ctx->netdev);
+	const struct macsec_rx_sc *ctx_rx_sc = ctx->rx_sc;
+	struct mlx5e_macsec_device *macsec_device;
+	struct mlx5e_macsec_rx_sc *rx_sc;
+	struct mlx5e_macsec_sa *rx_sa;
+	struct mlx5e_macsec *macsec;
+	struct list_head *list;
+	int i;
+	int err = 0;
+
+	mutex_lock(&priv->macsec->lock);
+
+	macsec = priv->macsec;
+	macsec_device = mlx5e_macsec_get_macsec_device_context(macsec, ctx);
+	if (!macsec_device) {
+		netdev_err(ctx->netdev, "MACsec offload: Failed to find device context\n");
+		err = -EINVAL;
+		goto out;
+	}
+
+	list = &macsec_device->macsec_rx_sc_list_head;
+	rx_sc = mlx5e_macsec_get_rx_sc_from_sc_list(list, ctx_rx_sc->sci);
+	if (!rx_sc) {
+		err = -EINVAL;
+		goto out;
+	}
+
+	if (rx_sc->active == ctx_rx_sc->active)
+		goto out;
+
+	rx_sc->active = ctx_rx_sc->active;
+	for (i = 0; i < MACSEC_NUM_AN; ++i) {
+		rx_sa = rx_sc->rx_sa[i];
+		if (!rx_sa)
+			continue;
+
+		err = macsec_rx_sa_active_update(ctx, rx_sa, rx_sa->active && ctx_rx_sc->active);
+		if (err)
+			goto out;
+	}
+
+out:
+	mutex_unlock(&macsec->lock);
+
+	return err;
+}
+
+static void macsec_del_rxsc_ctx(struct mlx5e_macsec *macsec, struct mlx5e_macsec_rx_sc *rx_sc)
+{
+	struct mlx5e_macsec_sa *rx_sa;
+	int i;
+
+	for (i = 0; i < MACSEC_NUM_AN; ++i) {
+		rx_sa = rx_sc->rx_sa[i];
+		if (!rx_sa)
+			continue;
+
+		mlx5e_macsec_cleanup_sa(macsec, rx_sa, false);
+		mlx5_destroy_encryption_key(macsec->mdev, rx_sa->enc_key_id);
+
+		kfree(rx_sa);
+		rx_sc->rx_sa[i] = NULL;
+	}
+
+	/* At this point the relevant MACsec offload Rx rule already removed at
+	 * mlx5e_macsec_cleanup_sa need to wait for datapath to finish current
+	 * Rx related data propagating using xa_erase which uses rcu to sync,
+	 * once fs_id is erased then this rx_sc is hidden from datapath.
+	 */
+	list_del_rcu(&rx_sc->rx_sc_list_element);
+	xa_erase(&macsec->sc_xarray, rx_sc->sc_xarray_element->fs_id);
+	metadata_dst_free(rx_sc->md_dst);
+	kfree(rx_sc->sc_xarray_element);
+	kfree_rcu(rx_sc);
+}
+
+static int mlx5e_macsec_del_rxsc(struct macsec_context *ctx)
+{
+	struct mlx5e_priv *priv = netdev_priv(ctx->netdev);
+	struct mlx5e_macsec_device *macsec_device;
+	struct mlx5e_macsec_rx_sc *rx_sc;
+	struct mlx5e_macsec *macsec;
+	struct list_head *list;
+	int err = 0;
+
+	mutex_lock(&priv->macsec->lock);
+
+	macsec = priv->macsec;
+	macsec_device = mlx5e_macsec_get_macsec_device_context(macsec, ctx);
+	if (!macsec_device) {
+		netdev_err(ctx->netdev, "MACsec offload: Failed to find device context\n");
+		err = -EINVAL;
+		goto out;
+	}
+
+	list = &macsec_device->macsec_rx_sc_list_head;
+	rx_sc = mlx5e_macsec_get_rx_sc_from_sc_list(list, ctx->rx_sc->sci);
+	if (!rx_sc) {
+		netdev_err(ctx->netdev,
+			   "MACsec offload rx_sc sci %lld doesn't exist\n",
+			   ctx->sa.rx_sa->sc->sci);
+		err = -EINVAL;
+		goto out;
+	}
+
+	macsec_del_rxsc_ctx(macsec, rx_sc);
+out:
+	mutex_unlock(&macsec->lock);
+
+	return err;
+}
+
+static int mlx5e_macsec_add_rxsa(struct macsec_context *ctx)
+{
+	const struct macsec_rx_sa *ctx_rx_sa = ctx->sa.rx_sa;
+	struct mlx5e_priv *priv = netdev_priv(ctx->netdev);
+	struct mlx5e_macsec_device *macsec_device;
+	struct mlx5_core_dev *mdev = priv->mdev;
+	u8 assoc_num = ctx->sa.assoc_num;
+	struct mlx5e_macsec_rx_sc *rx_sc;
+	sci_t sci = ctx_rx_sa->sc->sci;
+	struct mlx5e_macsec_sa *rx_sa;
+	struct mlx5e_macsec *macsec;
+	struct list_head *list;
+	int err = 0;
+
+	mutex_lock(&priv->macsec->lock);
+
+	macsec = priv->macsec;
+	macsec_device = mlx5e_macsec_get_macsec_device_context(macsec, ctx);
+	if (!macsec_device) {
+		netdev_err(ctx->netdev, "MACsec offload: Failed to find device context\n");
+		err = -EINVAL;
+		goto out;
+	}
+
+	list = &macsec_device->macsec_rx_sc_list_head;
+	rx_sc = mlx5e_macsec_get_rx_sc_from_sc_list(list, sci);
+	if (!rx_sc) {
+		netdev_err(ctx->netdev,
+			   "MACsec offload rx_sc sci %lld doesn't exist\n",
+			   ctx->sa.rx_sa->sc->sci);
+		err = -EINVAL;
+		goto out;
+	}
+
+	if (rx_sc->rx_sa[assoc_num]) {
+		netdev_err(ctx->netdev,
+			   "MACsec offload rx_sc sci %lld rx_sa %d already exist\n",
+			   sci, assoc_num);
+		err = -EEXIST;
+		goto out;
+	}
+
+	rx_sa = kzalloc(sizeof(*rx_sa), GFP_KERNEL);
+	if (!rx_sa) {
+		err = -ENOMEM;
+		goto out;
+	}
+
+	rx_sa->active = ctx_rx_sa->active;
+	rx_sa->next_pn = ctx_rx_sa->next_pn;
+	rx_sa->sci = sci;
+	rx_sa->assoc_num = assoc_num;
+	rx_sa->fs_id = rx_sc->sc_xarray_element->fs_id;
+
+	if (ctx->secy->xpn)
+		update_macsec_epn(rx_sa, &ctx_rx_sa->key, &ctx_rx_sa->next_pn_halves,
+				  ctx_rx_sa->ssci);
+
+	err = mlx5_create_encryption_key(mdev, ctx->sa.key, ctx->secy->key_len,
+					 MLX5_ACCEL_OBJ_MACSEC_KEY,
+					 &rx_sa->enc_key_id);
+	if (err)
+		goto destroy_sa;
+
+	rx_sc->rx_sa[assoc_num] = rx_sa;
+	if (!rx_sa->active)
+		goto out;
+
+	//TODO - add support for both authentication and encryption flows
+	err = mlx5e_macsec_init_sa(ctx, rx_sa, true, false);
+	if (err)
+		goto destroy_encryption_key;
+
+	goto out;
+
+destroy_encryption_key:
+	rx_sc->rx_sa[assoc_num] = NULL;
+	mlx5_destroy_encryption_key(mdev, rx_sa->enc_key_id);
+destroy_sa:
+	kfree(rx_sa);
+out:
+	mutex_unlock(&macsec->lock);
+
+	return err;
+}
+
+static int mlx5e_macsec_upd_rxsa(struct macsec_context *ctx)
+{
+	const struct macsec_rx_sa *ctx_rx_sa = ctx->sa.rx_sa;
+	struct mlx5e_priv *priv = netdev_priv(ctx->netdev);
+	struct mlx5e_macsec_device *macsec_device;
+	u8 assoc_num = ctx->sa.assoc_num;
+	struct mlx5e_macsec_rx_sc *rx_sc;
+	sci_t sci = ctx_rx_sa->sc->sci;
+	struct mlx5e_macsec_sa *rx_sa;
+	struct mlx5e_macsec *macsec;
+	struct list_head *list;
+	int err = 0;
+
+	mutex_lock(&priv->macsec->lock);
+
+	macsec = priv->macsec;
+	macsec_device = mlx5e_macsec_get_macsec_device_context(macsec, ctx);
+	if (!macsec_device) {
+		netdev_err(ctx->netdev, "MACsec offload: Failed to find device context\n");
+		err = -EINVAL;
+		goto out;
+	}
+
+	list = &macsec_device->macsec_rx_sc_list_head;
+	rx_sc = mlx5e_macsec_get_rx_sc_from_sc_list(list, sci);
+	if (!rx_sc) {
+		netdev_err(ctx->netdev,
+			   "MACsec offload rx_sc sci %lld doesn't exist\n",
+			   ctx->sa.rx_sa->sc->sci);
+		err = -EINVAL;
+		goto out;
+	}
+
+	rx_sa = rx_sc->rx_sa[assoc_num];
+	if (!rx_sa) {
+		netdev_err(ctx->netdev,
+			   "MACsec offload rx_sc sci %lld rx_sa %d doesn't exist\n",
+			   sci, assoc_num);
+		err = -EINVAL;
+		goto out;
+	}
+
+	if (ctx->sa.update_pn) {
+		netdev_err(ctx->netdev,
+			   "MACsec offload update RX sa %d PN isn't supported\n",
+			   assoc_num);
+		err = -EINVAL;
+		goto out;
+	}
+
+	err = macsec_rx_sa_active_update(ctx, rx_sa, ctx_rx_sa->active);
+out:
+	mutex_unlock(&macsec->lock);
+
+	return err;
+}
+
+static int mlx5e_macsec_del_rxsa(struct macsec_context *ctx)
+{
+	struct mlx5e_priv *priv = netdev_priv(ctx->netdev);
+	struct mlx5e_macsec_device *macsec_device;
+	sci_t sci = ctx->sa.rx_sa->sc->sci;
+	struct mlx5e_macsec_rx_sc *rx_sc;
+	u8 assoc_num = ctx->sa.assoc_num;
+	struct mlx5e_macsec_sa *rx_sa;
+	struct mlx5e_macsec *macsec;
+	struct list_head *list;
+	int err = 0;
+
+	mutex_lock(&priv->macsec->lock);
+
+	macsec = priv->macsec;
+	macsec_device = mlx5e_macsec_get_macsec_device_context(macsec, ctx);
+	if (!macsec_device) {
+		netdev_err(ctx->netdev, "MACsec offload: Failed to find device context\n");
+		err = -EINVAL;
+		goto out;
+	}
+
+	list = &macsec_device->macsec_rx_sc_list_head;
+	rx_sc = mlx5e_macsec_get_rx_sc_from_sc_list(list, sci);
+	if (!rx_sc) {
+		netdev_err(ctx->netdev,
+			   "MACsec offload rx_sc sci %lld doesn't exist\n",
+			   ctx->sa.rx_sa->sc->sci);
+		err = -EINVAL;
+		goto out;
+	}
+
+	rx_sa = rx_sc->rx_sa[assoc_num];
+	if (!rx_sa) {
+		netdev_err(ctx->netdev,
+			   "MACsec offload rx_sc sci %lld rx_sa %d doesn't exist\n",
+			   sci, assoc_num);
+		err = -EINVAL;
+		goto out;
+	}
+
+	mlx5e_macsec_cleanup_sa(macsec, rx_sa, false);
+	mlx5_destroy_encryption_key(macsec->mdev, rx_sa->enc_key_id);
+	kfree(rx_sa);
+	rx_sc->rx_sa[assoc_num] = NULL;
+
+out:
+	mutex_unlock(&macsec->lock);
+
+	return err;
+}
+
+static int mlx5e_macsec_add_secy(struct macsec_context *ctx)
+{
+	struct mlx5e_priv *priv = netdev_priv(ctx->netdev);
+	const struct net_device *dev = ctx->secy->netdev;
+	const struct net_device *netdev = ctx->netdev;
+	struct mlx5e_macsec_device *macsec_device;
+	struct mlx5e_macsec *macsec;
+	int err = 0;
+
+	if (!mlx5e_macsec_secy_features_validate(ctx))
+		return -EINVAL;
+
+	mutex_lock(&priv->macsec->lock);
+	macsec = priv->macsec;
+	if (mlx5e_macsec_get_macsec_device_context(macsec, ctx)) {
+		netdev_err(netdev, "MACsec offload: MACsec net_device already exist\n");
+		goto out;
+	}
+
+	if (macsec->num_of_devices >= MLX5_MACSEC_NUM_OF_SUPPORTED_INTERFACES) {
+		netdev_err(netdev, "Currently, only %d MACsec offload devices can be set\n",
+			   MLX5_MACSEC_NUM_OF_SUPPORTED_INTERFACES);
+		err = -EBUSY;
+		goto out;
+	}
+
+	macsec_device = kzalloc(sizeof(*macsec_device), GFP_KERNEL);
+	if (!macsec_device) {
+		err = -ENOMEM;
+		goto out;
+	}
+
+	macsec_device->dev_addr = kmemdup(dev->dev_addr, dev->addr_len, GFP_KERNEL);
+	if (!macsec_device->dev_addr) {
+		kfree(macsec_device);
+		err = -ENOMEM;
+		goto out;
+	}
+
+	macsec_device->netdev = dev;
+
+	INIT_LIST_HEAD_RCU(&macsec_device->macsec_rx_sc_list_head);
+	list_add_rcu(&macsec_device->macsec_device_list_element, &macsec->macsec_device_list_head);
+
+	++macsec->num_of_devices;
+out:
+	mutex_unlock(&macsec->lock);
+
+	return err;
+}
+
+static int macsec_upd_secy_hw_address(struct macsec_context *ctx,
+				      struct mlx5e_macsec_device *macsec_device)
+{
+	struct mlx5e_priv *priv = netdev_priv(ctx->netdev);
+	const struct net_device *dev = ctx->secy->netdev;
+	struct mlx5e_macsec *macsec = priv->macsec;
+	struct mlx5e_macsec_rx_sc *rx_sc, *tmp;
+	struct mlx5e_macsec_sa *rx_sa;
+	struct list_head *list;
+	int i, err = 0;
+
+
+	list = &macsec_device->macsec_rx_sc_list_head;
+	list_for_each_entry_safe(rx_sc, tmp, list, rx_sc_list_element) {
+		for (i = 0; i < MACSEC_NUM_AN; ++i) {
+			rx_sa = rx_sc->rx_sa[i];
+			if (!rx_sa || !rx_sa->macsec_rule)
+				continue;
+
+			mlx5e_macsec_cleanup_sa(macsec, rx_sa, false);
+		}
+	}
+
+	list_for_each_entry_safe(rx_sc, tmp, list, rx_sc_list_element) {
+		for (i = 0; i < MACSEC_NUM_AN; ++i) {
+			rx_sa = rx_sc->rx_sa[i];
+			if (!rx_sa)
+				continue;
+
+			if (rx_sa->active) {
+				err = mlx5e_macsec_init_sa(ctx, rx_sa, true, false);
+				if (err)
+					goto out;
+			}
+		}
+	}
+
+	memcpy(macsec_device->dev_addr, dev->dev_addr, dev->addr_len);
+out:
+	return err;
+}
+
+/* this function is called from 2 macsec ops functions:
+ *  macsec_set_mac_address – MAC address was changed, therefore we need to destroy
+ *  and create new Tx contexts(macsec object + steering).
+ *  macsec_changelink – in this case the tx SC or SecY may be changed, therefore need to
+ *  destroy Tx and Rx contexts(macsec object + steering)
+ */
+static int mlx5e_macsec_upd_secy(struct macsec_context *ctx)
+{
+	const struct macsec_tx_sc *tx_sc = &ctx->secy->tx_sc;
+	struct mlx5e_priv *priv = netdev_priv(ctx->netdev);
+	const struct net_device *dev = ctx->secy->netdev;
+	struct mlx5e_macsec_device *macsec_device;
+	struct mlx5e_macsec_sa *tx_sa;
+	struct mlx5e_macsec *macsec;
+	int i, err = 0;
+
+	if (!mlx5e_macsec_secy_features_validate(ctx))
+		return -EINVAL;
+
+	mutex_lock(&priv->macsec->lock);
+
+	macsec = priv->macsec;
+	macsec_device = mlx5e_macsec_get_macsec_device_context(macsec, ctx);
+	if (!macsec_device) {
+		netdev_err(ctx->netdev, "MACsec offload: Failed to find device context\n");
+		err = -EINVAL;
+		goto out;
+	}
+
+	/* if the dev_addr hasn't change, it mean the callback is from macsec_changelink */
+	if (!memcmp(macsec_device->dev_addr, dev->dev_addr, dev->addr_len)) {
+		err = macsec_upd_secy_hw_address(ctx, macsec_device);
+		if (err)
+			goto out;
+	}
+
+	for (i = 0; i < MACSEC_NUM_AN; ++i) {
+		tx_sa = macsec_device->tx_sa[i];
+		if (!tx_sa)
+			continue;
+
+		mlx5e_macsec_cleanup_sa(macsec, tx_sa, true);
+	}
+
+	for (i = 0; i < MACSEC_NUM_AN; ++i) {
+		tx_sa = macsec_device->tx_sa[i];
+		if (!tx_sa)
+			continue;
+
+		if (tx_sa->assoc_num == tx_sc->encoding_sa && tx_sa->active) {
+			err = mlx5e_macsec_init_sa(ctx, tx_sa, tx_sc->encrypt, true);
+			if (err)
+				goto out;
+		}
+	}
+
+out:
+	mutex_unlock(&macsec->lock);
+
+	return err;
+}
+
+static int mlx5e_macsec_del_secy(struct macsec_context *ctx)
+{
+	struct mlx5e_priv *priv = netdev_priv(ctx->netdev);
+	struct mlx5e_macsec_device *macsec_device;
+	struct mlx5e_macsec_rx_sc *rx_sc, *tmp;
+	struct mlx5e_macsec_sa *tx_sa;
+	struct mlx5e_macsec *macsec;
+	struct list_head *list;
+	int err = 0;
+	int i;
+
+	mutex_lock(&priv->macsec->lock);
+	macsec = priv->macsec;
+	macsec_device = mlx5e_macsec_get_macsec_device_context(macsec, ctx);
+	if (!macsec_device) {
+		netdev_err(ctx->netdev, "MACsec offload: Failed to find device context\n");
+		err = -EINVAL;
+
+		goto out;
+	}
+
+	for (i = 0; i < MACSEC_NUM_AN; ++i) {
+		tx_sa = macsec_device->tx_sa[i];
+		if (!tx_sa)
+			continue;
+
+		mlx5e_macsec_cleanup_sa(macsec, tx_sa, true);
+		mlx5_destroy_encryption_key(macsec->mdev, tx_sa->enc_key_id);
+		kfree(tx_sa);
+		macsec_device->tx_sa[i] = NULL;
+	}
+
+	list = &macsec_device->macsec_rx_sc_list_head;
+	list_for_each_entry_safe(rx_sc, tmp, list, rx_sc_list_element)
+		macsec_del_rxsc_ctx(macsec, rx_sc);
+
+	kfree(macsec_device->dev_addr);
+	macsec_device->dev_addr = NULL;
+
+	list_del_rcu(&macsec_device->macsec_device_list_element);
+	--macsec->num_of_devices;
+	kfree(macsec_device);
+
+out:
+	mutex_unlock(&macsec->lock);
+
+	return err;
+}
+
+static void macsec_build_accel_attrs(struct mlx5e_macsec_sa *sa,
+				     struct mlx5_macsec_obj_attrs *attrs)
+{
+	attrs->epn_state.epn_msb = sa->epn_state.epn_msb;
+	attrs->epn_state.overlap = sa->epn_state.overlap;
+}
+
+static void macsec_aso_build_wqe_ctrl_seg(struct mlx5e_macsec_aso *macsec_aso,
+					  struct mlx5_wqe_aso_ctrl_seg *aso_ctrl,
+					  struct mlx5_aso_ctrl_param *param)
+{
+	memset(aso_ctrl, 0, sizeof(*aso_ctrl));
+	if (macsec_aso->umr->dma_addr) {
+		aso_ctrl->va_l  = cpu_to_be32(macsec_aso->umr->dma_addr | ASO_CTRL_READ_EN);
+		aso_ctrl->va_h  = cpu_to_be32((u64)macsec_aso->umr->dma_addr >> 32);
+		aso_ctrl->l_key = cpu_to_be32(macsec_aso->umr->mkey);
+	}
+
+	if (!param)
+		return;
+
+	aso_ctrl->data_mask_mode = param->data_mask_mode << 6;
+	aso_ctrl->condition_1_0_operand = param->condition_1_operand |
+						param->condition_0_operand << 4;
+	aso_ctrl->condition_1_0_offset = param->condition_1_offset |
+						param->condition_0_offset << 4;
+	aso_ctrl->data_offset_condition_operand = param->data_offset |
+						param->condition_operand << 6;
+	aso_ctrl->condition_0_data = cpu_to_be32(param->condition_0_data);
+	aso_ctrl->condition_0_mask = cpu_to_be32(param->condition_0_mask);
+	aso_ctrl->condition_1_data = cpu_to_be32(param->condition_1_data);
+	aso_ctrl->condition_1_mask = cpu_to_be32(param->condition_1_mask);
+	aso_ctrl->bitwise_data = cpu_to_be64(param->bitwise_data);
+	aso_ctrl->data_mask = cpu_to_be64(param->data_mask);
+}
+
+static int mlx5e_macsec_modify_obj(struct mlx5_core_dev *mdev, struct mlx5_macsec_obj_attrs *attrs,
+				   u32 macsec_id)
+{
+	u32 in[MLX5_ST_SZ_DW(modify_macsec_obj_in)] = {};
+	u32 out[MLX5_ST_SZ_DW(query_macsec_obj_out)];
+	u64 modify_field_select = 0;
+	void *obj;
+	int err;
+
+	/* General object fields set */
+	MLX5_SET(general_obj_in_cmd_hdr, in, opcode, MLX5_CMD_OP_QUERY_GENERAL_OBJECT);
+	MLX5_SET(general_obj_in_cmd_hdr, in, obj_type, MLX5_GENERAL_OBJECT_TYPES_MACSEC);
+	MLX5_SET(general_obj_in_cmd_hdr, in, obj_id, macsec_id);
+	err = mlx5_cmd_exec(mdev, in, sizeof(in), out, sizeof(out));
+	if (err) {
+		mlx5_core_err(mdev, "Query MACsec object failed (Object id %d), err = %d\n",
+			      macsec_id, err);
+		return err;
+	}
+
+	obj = MLX5_ADDR_OF(query_macsec_obj_out, out, macsec_object);
+	modify_field_select = MLX5_GET64(macsec_offload_obj, obj, modify_field_select);
+
+	/* EPN */
+	if (!(modify_field_select & MLX5_MODIFY_MACSEC_BITMASK_EPN_OVERLAP) ||
+	    !(modify_field_select & MLX5_MODIFY_MACSEC_BITMASK_EPN_MSB)) {
+		mlx5_core_dbg(mdev, "MACsec object field is not modifiable (Object id %d)\n",
+			      macsec_id);
+		return -EOPNOTSUPP;
+	}
+
+	obj = MLX5_ADDR_OF(modify_macsec_obj_in, in, macsec_object);
+	MLX5_SET64(macsec_offload_obj, obj, modify_field_select,
+		   MLX5_MODIFY_MACSEC_BITMASK_EPN_OVERLAP | MLX5_MODIFY_MACSEC_BITMASK_EPN_MSB);
+	MLX5_SET(macsec_offload_obj, obj, epn_msb, attrs->epn_state.epn_msb);
+	MLX5_SET(macsec_offload_obj, obj, epn_overlap, attrs->epn_state.overlap);
+
+	/* General object fields set */
+	MLX5_SET(general_obj_in_cmd_hdr, in, opcode, MLX5_CMD_OP_MODIFY_GENERAL_OBJECT);
+
+	return mlx5_cmd_exec(mdev, in, sizeof(in), out, sizeof(out));
+}
+
+static void macsec_aso_build_ctrl(struct mlx5e_macsec_aso *aso,
+				  struct mlx5_wqe_aso_ctrl_seg *aso_ctrl,
+				  struct mlx5e_macsec_aso_in *in)
+{
+	struct mlx5_aso_ctrl_param param = {};
+
+	param.data_mask_mode = MLX5_ASO_DATA_MASK_MODE_BITWISE_64BIT;
+	param.condition_0_operand = MLX5_ASO_ALWAYS_TRUE;
+	param.condition_1_operand = MLX5_ASO_ALWAYS_TRUE;
+	if (in->mode == MLX5_MACSEC_EPN) {
+		param.data_offset = MLX5_MACSEC_ASO_REMOVE_FLOW_PKT_CNT_OFFSET;
+		param.bitwise_data = BIT_ULL(54);
+		param.data_mask = param.bitwise_data;
+	}
+	macsec_aso_build_wqe_ctrl_seg(aso, aso_ctrl, &param);
+}
+
+static int macsec_aso_set_arm_event(struct mlx5_core_dev *mdev, struct mlx5e_macsec *macsec,
+				    struct mlx5e_macsec_aso_in *in)
+{
+	struct mlx5e_macsec_aso *aso;
+	struct mlx5_aso_wqe *aso_wqe;
+	struct mlx5_aso *maso;
+	int err;
+
+	aso = &macsec->aso;
+	maso = aso->maso;
+
+	mutex_lock(&aso->aso_lock);
+	aso_wqe = mlx5_aso_get_wqe(maso);
+	mlx5_aso_build_wqe(maso, MLX5_MACSEC_ASO_DS_CNT, aso_wqe, in->obj_id,
+			   MLX5_ACCESS_ASO_OPC_MOD_MACSEC);
+	macsec_aso_build_ctrl(aso, &aso_wqe->aso_ctrl, in);
+	mlx5_aso_post_wqe(maso, false, &aso_wqe->ctrl);
+	err = mlx5_aso_poll_cq(maso, false);
+	mutex_unlock(&aso->aso_lock);
+
+	return err;
+}
+
+static int macsec_aso_query(struct mlx5_core_dev *mdev, struct mlx5e_macsec *macsec,
+			    struct mlx5e_macsec_aso_in *in, struct mlx5e_macsec_aso_out *out)
+{
+	struct mlx5e_macsec_aso *aso;
+	struct mlx5_aso_wqe *aso_wqe;
+	struct mlx5_aso *maso;
+	unsigned long expires;
+	int err;
+
+	aso = &macsec->aso;
+	maso = aso->maso;
+
+	mutex_lock(&aso->aso_lock);
+
+	aso_wqe = mlx5_aso_get_wqe(maso);
+	mlx5_aso_build_wqe(maso, MLX5_MACSEC_ASO_DS_CNT, aso_wqe, in->obj_id,
+			   MLX5_ACCESS_ASO_OPC_MOD_MACSEC);
+	macsec_aso_build_wqe_ctrl_seg(aso, &aso_wqe->aso_ctrl, NULL);
+
+	mlx5_aso_post_wqe(maso, false, &aso_wqe->ctrl);
+	expires = jiffies + msecs_to_jiffies(10);
+	do {
+		err = mlx5_aso_poll_cq(maso, false);
+		if (err)
+			usleep_range(2, 10);
+	} while (err && time_is_after_jiffies(expires));
+
+	if (err)
+		goto err_out;
+
+	if (MLX5_GET(macsec_aso, aso->umr->ctx, epn_event_arm))
+		out->event_arm |= MLX5E_ASO_EPN_ARM;
+
+	out->mode_param = MLX5_GET(macsec_aso, aso->umr->ctx, mode_parameter);
+
+err_out:
+	mutex_unlock(&aso->aso_lock);
+	return err;
+}
+
+static struct mlx5e_macsec_sa *get_macsec_tx_sa_from_obj_id(const struct mlx5e_macsec *macsec,
+							    const u32 obj_id)
+{
+	const struct list_head *device_list;
+	struct mlx5e_macsec_sa *macsec_sa;
+	struct mlx5e_macsec_device *iter;
+	int i;
+
+	device_list = &macsec->macsec_device_list_head;
+
+	list_for_each_entry(iter, device_list, macsec_device_list_element) {
+		for (i = 0; i < MACSEC_NUM_AN; ++i) {
+			macsec_sa = iter->tx_sa[i];
+			if (!macsec_sa || !macsec_sa->active)
+				continue;
+			if (macsec_sa->macsec_obj_id == obj_id)
+				return macsec_sa;
+		}
+	}
+
+	return NULL;
+}
+
+static struct mlx5e_macsec_sa *get_macsec_rx_sa_from_obj_id(const struct mlx5e_macsec *macsec,
+							    const u32 obj_id)
+{
+	const struct list_head *device_list, *sc_list;
+	struct mlx5e_macsec_rx_sc *mlx5e_rx_sc;
+	struct mlx5e_macsec_sa *macsec_sa;
+	struct mlx5e_macsec_device *iter;
+	int i;
+
+	device_list = &macsec->macsec_device_list_head;
+
+	list_for_each_entry(iter, device_list, macsec_device_list_element) {
+		sc_list = &iter->macsec_rx_sc_list_head;
+		list_for_each_entry(mlx5e_rx_sc, sc_list, rx_sc_list_element) {
+			for (i = 0; i < MACSEC_NUM_AN; ++i) {
+				macsec_sa = mlx5e_rx_sc->rx_sa[i];
+				if (!macsec_sa || !macsec_sa->active)
+					continue;
+				if (macsec_sa->macsec_obj_id == obj_id)
+					return macsec_sa;
+			}
+		}
+	}
+
+	return NULL;
+}
+
+static void macsec_epn_update(struct mlx5e_macsec *macsec, struct mlx5_core_dev *mdev,
+			      struct mlx5e_macsec_sa *sa, u32 obj_id, u32 mode_param)
+{
+	struct mlx5_macsec_obj_attrs attrs = {};
+	struct mlx5e_macsec_aso_in in = {};
+
+	/* When the bottom of the replay protection window (mode_param) crosses 2^31 (half sequence
+	 * number wraparound) hence mode_param > MLX5_MACSEC_EPN_SCOPE_MID the SW should update the
+	 * esn_overlap to OLD (1).
+	 * When the bottom of the replay protection window (mode_param) crosses 2^32 (full sequence
+	 * number wraparound) hence mode_param < MLX5_MACSEC_EPN_SCOPE_MID since it did a
+	 * wraparound, the SW should update the esn_overlap to NEW (0), and increment the esn_msb.
+	 */
+
+	if (mode_param < MLX5_MACSEC_EPN_SCOPE_MID) {
+		sa->epn_state.epn_msb++;
+		sa->epn_state.overlap = 0;
+	} else {
+		sa->epn_state.overlap = 1;
+	}
+
+	macsec_build_accel_attrs(sa, &attrs);
+	mlx5e_macsec_modify_obj(mdev, &attrs, obj_id);
+
+	/* Re-set EPN arm event */
+	in.obj_id = obj_id;
+	in.mode = MLX5_MACSEC_EPN;
+	macsec_aso_set_arm_event(mdev, macsec, &in);
+}
+
+static void macsec_async_event(struct work_struct *work)
+{
+	struct mlx5e_macsec_async_work *async_work;
+	struct mlx5e_macsec_aso_out out = {};
+	struct mlx5e_macsec_aso_in in = {};
+	struct mlx5e_macsec_sa *macsec_sa;
+	struct mlx5e_macsec *macsec;
+	struct mlx5_core_dev *mdev;
+	u32 obj_id;
+
+	async_work = container_of(work, struct mlx5e_macsec_async_work, work);
+	macsec = async_work->macsec;
+	mutex_lock(&macsec->lock);
+
+	mdev = async_work->mdev;
+	obj_id = async_work->obj_id;
+	macsec_sa = get_macsec_tx_sa_from_obj_id(macsec, obj_id);
+	if (!macsec_sa) {
+		macsec_sa = get_macsec_rx_sa_from_obj_id(macsec, obj_id);
+		if (!macsec_sa) {
+			mlx5_core_dbg(mdev, "MACsec SA is not found (SA object id %d)\n", obj_id);
+			goto out_async_work;
+		}
+	}
+
+	/* Query MACsec ASO context */
+	in.obj_id = obj_id;
+	macsec_aso_query(mdev, macsec, &in, &out);
+
+	/* EPN case */
+	if (macsec_sa->epn_state.epn_enabled && !(out.event_arm & MLX5E_ASO_EPN_ARM))
+		macsec_epn_update(macsec, mdev, macsec_sa, obj_id, out.mode_param);
+
+out_async_work:
+	kfree(async_work);
+	mutex_unlock(&macsec->lock);
+}
+
+static int macsec_obj_change_event(struct notifier_block *nb, unsigned long event, void *data)
+{
+	struct mlx5e_macsec *macsec = container_of(nb, struct mlx5e_macsec, nb);
+	struct mlx5e_macsec_async_work *async_work;
+	struct mlx5_eqe_obj_change *obj_change;
+	struct mlx5_eqe *eqe = data;
+	u16 obj_type;
+	u32 obj_id;
+
+	if (event != MLX5_EVENT_TYPE_OBJECT_CHANGE)
+		return NOTIFY_DONE;
+
+	obj_change = &eqe->data.obj_change;
+	obj_type = be16_to_cpu(obj_change->obj_type);
+	obj_id = be32_to_cpu(obj_change->obj_id);
+
+	if (obj_type != MLX5_GENERAL_OBJECT_TYPES_MACSEC)
+		return NOTIFY_DONE;
+
+	async_work = kzalloc(sizeof(*async_work), GFP_ATOMIC);
+	if (!async_work)
+		return NOTIFY_DONE;
+
+	async_work->macsec = macsec;
+	async_work->mdev = macsec->mdev;
+	async_work->obj_id = obj_id;
+
+	INIT_WORK(&async_work->work, macsec_async_event);
+
+	WARN_ON(!queue_work(macsec->wq, &async_work->work));
+
+	return NOTIFY_OK;
+}
+
+static int mlx5e_macsec_aso_init(struct mlx5e_macsec_aso *aso, struct mlx5_core_dev *mdev)
+{
+	struct mlx5_aso *maso;
+	int err;
+
+	err = mlx5_core_alloc_pd(mdev, &aso->pdn);
+	if (err) {
+		mlx5_core_err(mdev,
+			      "MACsec offload: Failed to alloc pd for MACsec ASO, err=%d\n",
+			      err);
+		return err;
+	}
+
+	maso = mlx5_aso_create(mdev, aso->pdn);
+	if (IS_ERR(maso)) {
+		err = PTR_ERR(maso);
+		goto err_aso;
+	}
+
+	err = mlx5e_macsec_aso_reg_mr(mdev, aso);
+	if (err)
+		goto err_aso_reg;
+
+	mutex_init(&aso->aso_lock);
+
+	aso->maso = maso;
+
+	return 0;
+
+err_aso_reg:
+	mlx5_aso_destroy(maso);
+err_aso:
+	mlx5_core_dealloc_pd(mdev, aso->pdn);
+	return err;
+}
+
+static void mlx5e_macsec_aso_cleanup(struct mlx5e_macsec_aso *aso, struct mlx5_core_dev *mdev)
+{
+	if (!aso)
+		return;
+
+	mlx5e_macsec_aso_dereg_mr(mdev, aso);
+
+	mlx5_aso_destroy(aso->maso);
+
+	mlx5_core_dealloc_pd(mdev, aso->pdn);
+}
+
+bool mlx5e_is_macsec_device(const struct mlx5_core_dev *mdev)
+{
+	if (!(MLX5_CAP_GEN_64(mdev, general_obj_types) &
+	    MLX5_GENERAL_OBJ_TYPES_CAP_MACSEC_OFFLOAD))
+		return false;
+
+	if (!MLX5_CAP_GEN(mdev, log_max_dek))
+		return false;
+
+	if (!MLX5_CAP_MACSEC(mdev, log_max_macsec_offload))
+		return false;
+
+	if (!MLX5_CAP_FLOWTABLE_NIC_RX(mdev, macsec_decrypt) ||
+	    !MLX5_CAP_FLOWTABLE_NIC_RX(mdev, reformat_remove_macsec))
+		return false;
+
+	if (!MLX5_CAP_FLOWTABLE_NIC_TX(mdev, macsec_encrypt) ||
+	    !MLX5_CAP_FLOWTABLE_NIC_TX(mdev, reformat_add_macsec))
+		return false;
+
+	if (!MLX5_CAP_MACSEC(mdev, macsec_crypto_esp_aes_gcm_128_encrypt) &&
+	    !MLX5_CAP_MACSEC(mdev, macsec_crypto_esp_aes_gcm_256_encrypt))
+		return false;
+
+	if (!MLX5_CAP_MACSEC(mdev, macsec_crypto_esp_aes_gcm_128_decrypt) &&
+	    !MLX5_CAP_MACSEC(mdev, macsec_crypto_esp_aes_gcm_256_decrypt))
+		return false;
+
+	return true;
+}
+
+void mlx5e_macsec_get_stats_fill(struct mlx5e_macsec *macsec, void *macsec_stats)
+{
+	mlx5e_macsec_fs_get_stats_fill(macsec->macsec_fs, macsec_stats);
+}
+
+struct mlx5e_macsec_stats *mlx5e_macsec_get_stats(struct mlx5e_macsec *macsec)
+{
+	if (!macsec)
+		return NULL;
+
+	return &macsec->stats;
+}
+
+static const struct macsec_ops macsec_offload_ops = {
+	.mdo_add_txsa = mlx5e_macsec_add_txsa,
+	.mdo_upd_txsa = mlx5e_macsec_upd_txsa,
+	.mdo_del_txsa = mlx5e_macsec_del_txsa,
+	.mdo_add_rxsc = mlx5e_macsec_add_rxsc,
+	.mdo_upd_rxsc = mlx5e_macsec_upd_rxsc,
+	.mdo_del_rxsc = mlx5e_macsec_del_rxsc,
+	.mdo_add_rxsa = mlx5e_macsec_add_rxsa,
+	.mdo_upd_rxsa = mlx5e_macsec_upd_rxsa,
+	.mdo_del_rxsa = mlx5e_macsec_del_rxsa,
+	.mdo_add_secy = mlx5e_macsec_add_secy,
+	.mdo_upd_secy = mlx5e_macsec_upd_secy,
+	.mdo_del_secy = mlx5e_macsec_del_secy,
+};
+
+bool mlx5e_macsec_handle_tx_skb(struct mlx5e_macsec *macsec, struct sk_buff *skb)
+{
+	struct metadata_dst *md_dst = skb_metadata_dst(skb);
+	u32 fs_id;
+
+	fs_id = mlx5e_macsec_get_sa_from_hashtable(&macsec->sci_hash, &md_dst->u.macsec_info.sci);
+	if (!fs_id)
+		goto err_out;
+
+	return true;
+
+err_out:
+	dev_kfree_skb_any(skb);
+	return false;
+}
+
+void mlx5e_macsec_tx_build_eseg(struct mlx5e_macsec *macsec,
+				struct sk_buff *skb,
+				struct mlx5_wqe_eth_seg *eseg)
+{
+	struct metadata_dst *md_dst = skb_metadata_dst(skb);
+	u32 fs_id;
+
+	fs_id = mlx5e_macsec_get_sa_from_hashtable(&macsec->sci_hash, &md_dst->u.macsec_info.sci);
+	if (!fs_id)
+		return;
+
+	eseg->flow_table_metadata = cpu_to_be32(MLX5_ETH_WQE_FT_META_MACSEC | fs_id << 2);
+}
+
+void mlx5e_macsec_offload_handle_rx_skb(struct net_device *netdev,
+					struct sk_buff *skb,
+					struct mlx5_cqe64 *cqe)
+{
+	struct mlx5e_macsec_rx_sc_xarray_element *sc_xarray_element;
+	u32 macsec_meta_data = be32_to_cpu(cqe->ft_metadata);
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	struct mlx5e_macsec_rx_sc *rx_sc;
+	struct mlx5e_macsec *macsec;
+	u32  fs_id;
+
+	macsec = priv->macsec;
+	if (!macsec)
+		return;
+
+	fs_id = MLX5_MACSEC_RX_METADAT_HANDLE(macsec_meta_data);
+
+	rcu_read_lock();
+	sc_xarray_element = xa_load(&macsec->sc_xarray, fs_id);
+	rx_sc = sc_xarray_element->rx_sc;
+	if (rx_sc) {
+		dst_hold(&rx_sc->md_dst->dst);
+		skb_dst_set(skb, &rx_sc->md_dst->dst);
+	}
+
+	rcu_read_unlock();
+}
+
+void mlx5e_macsec_build_netdev(struct mlx5e_priv *priv)
+{
+	struct net_device *netdev = priv->netdev;
+
+	if (!mlx5e_is_macsec_device(priv->mdev))
+		return;
+
+	/* Enable MACsec */
+	mlx5_core_dbg(priv->mdev, "mlx5e: MACsec acceleration enabled\n");
+	netdev->macsec_ops = &macsec_offload_ops;
+	netdev->features |= NETIF_F_HW_MACSEC;
+	netif_keep_dst(netdev);
+}
+
+int mlx5e_macsec_init(struct mlx5e_priv *priv)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+	struct mlx5e_macsec *macsec = NULL;
+	struct mlx5e_macsec_fs *macsec_fs;
+	int err;
+
+	if (!mlx5e_is_macsec_device(priv->mdev)) {
+		mlx5_core_dbg(mdev, "Not a MACsec offload device\n");
+		return 0;
+	}
+
+	macsec = kzalloc(sizeof(*macsec), GFP_KERNEL);
+	if (!macsec)
+		return -ENOMEM;
+
+	INIT_LIST_HEAD(&macsec->macsec_device_list_head);
+	mutex_init(&macsec->lock);
+
+	err = rhashtable_init(&macsec->sci_hash, &rhash_sci);
+	if (err) {
+		mlx5_core_err(mdev, "MACsec offload: Failed to init SCI hash table, err=%d\n",
+			      err);
+		goto err_hash;
+	}
+
+	err = mlx5e_macsec_aso_init(&macsec->aso, priv->mdev);
+	if (err) {
+		mlx5_core_err(mdev, "MACsec offload: Failed to init aso, err=%d\n", err);
+		goto err_aso;
+	}
+
+	macsec->wq = alloc_ordered_workqueue("mlx5e_macsec_%s", 0, priv->netdev->name);
+	if (!macsec->wq) {
+		err = -ENOMEM;
+		goto err_wq;
+	}
+
+	xa_init_flags(&macsec->sc_xarray, XA_FLAGS_ALLOC1);
+
+	priv->macsec = macsec;
+
+	macsec->mdev = mdev;
+
+	macsec_fs = mlx5e_macsec_fs_init(mdev, priv->netdev);
+	if (!macsec_fs) {
+		err = -ENOMEM;
+		goto err_out;
+	}
+
+	macsec->macsec_fs = macsec_fs;
+
+	macsec->nb.notifier_call = macsec_obj_change_event;
+	mlx5_notifier_register(mdev, &macsec->nb);
+
+	mlx5_core_dbg(mdev, "MACsec attached to netdevice\n");
+
+	return 0;
+
+err_out:
+	destroy_workqueue(macsec->wq);
+err_wq:
+	mlx5e_macsec_aso_cleanup(&macsec->aso, priv->mdev);
+err_aso:
+	rhashtable_destroy(&macsec->sci_hash);
+err_hash:
+	kfree(macsec);
+	priv->macsec = NULL;
+	return err;
+}
+
+void mlx5e_macsec_cleanup(struct mlx5e_priv *priv)
+{
+	struct mlx5e_macsec *macsec = priv->macsec;
+	struct mlx5_core_dev *mdev = priv->mdev;
+
+	if (!macsec)
+		return;
+
+	mlx5_notifier_unregister(mdev, &macsec->nb);
+	mlx5e_macsec_fs_cleanup(macsec->macsec_fs);
+	destroy_workqueue(macsec->wq);
+	mlx5e_macsec_aso_cleanup(&macsec->aso, mdev);
+	rhashtable_destroy(&macsec->sci_hash);
+	mutex_destroy(&macsec->lock);
+	kfree(macsec);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec.h b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec.h
new file mode 100644
index 000000000..347380a2c
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec.h
@@ -0,0 +1,73 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2022, NVIDIA CORPORATION & AFFILIATES. All rights reserved. */
+
+#ifndef __MLX5_EN_ACCEL_MACSEC_H__
+#define __MLX5_EN_ACCEL_MACSEC_H__
+
+#ifdef CONFIG_MLX5_EN_MACSEC
+
+#include <linux/mlx5/driver.h>
+#include <net/macsec.h>
+#include <net/dst_metadata.h>
+
+/* Bit31 - 30: MACsec marker, Bit15-0: MACsec id */
+#define MLX5_MACEC_RX_FS_ID_MAX USHRT_MAX /* Must be power of two */
+#define MLX5_MACSEC_RX_FS_ID_MASK MLX5_MACEC_RX_FS_ID_MAX
+#define MLX5_MACSEC_METADATA_MARKER(metadata)  ((((metadata) >> 30) & 0x3)  == 0x1)
+#define MLX5_MACSEC_RX_METADAT_HANDLE(metadata)  ((metadata) & MLX5_MACSEC_RX_FS_ID_MASK)
+
+struct mlx5e_priv;
+struct mlx5e_macsec;
+
+struct mlx5e_macsec_stats {
+	u64 macsec_rx_pkts;
+	u64 macsec_rx_bytes;
+	u64 macsec_rx_pkts_drop;
+	u64 macsec_rx_bytes_drop;
+	u64 macsec_tx_pkts;
+	u64 macsec_tx_bytes;
+	u64 macsec_tx_pkts_drop;
+	u64 macsec_tx_bytes_drop;
+};
+
+void mlx5e_macsec_build_netdev(struct mlx5e_priv *priv);
+int mlx5e_macsec_init(struct mlx5e_priv *priv);
+void mlx5e_macsec_cleanup(struct mlx5e_priv *priv);
+bool mlx5e_macsec_handle_tx_skb(struct mlx5e_macsec *macsec, struct sk_buff *skb);
+void mlx5e_macsec_tx_build_eseg(struct mlx5e_macsec *macsec,
+				struct sk_buff *skb,
+				struct mlx5_wqe_eth_seg *eseg);
+
+static inline bool mlx5e_macsec_skb_is_offload(struct sk_buff *skb)
+{
+	struct metadata_dst *md_dst = skb_metadata_dst(skb);
+
+	return md_dst && (md_dst->type == METADATA_MACSEC);
+}
+
+static inline bool mlx5e_macsec_is_rx_flow(struct mlx5_cqe64 *cqe)
+{
+	return MLX5_MACSEC_METADATA_MARKER(be32_to_cpu(cqe->ft_metadata));
+}
+
+void mlx5e_macsec_offload_handle_rx_skb(struct net_device *netdev, struct sk_buff *skb,
+					struct mlx5_cqe64 *cqe);
+bool mlx5e_is_macsec_device(const struct mlx5_core_dev *mdev);
+void mlx5e_macsec_get_stats_fill(struct mlx5e_macsec *macsec, void *macsec_stats);
+struct mlx5e_macsec_stats *mlx5e_macsec_get_stats(struct mlx5e_macsec *macsec);
+
+#else
+
+static inline void mlx5e_macsec_build_netdev(struct mlx5e_priv *priv) {}
+static inline int mlx5e_macsec_init(struct mlx5e_priv *priv) { return 0; }
+static inline void mlx5e_macsec_cleanup(struct mlx5e_priv *priv) {}
+static inline bool mlx5e_macsec_skb_is_offload(struct sk_buff *skb) { return false; }
+static inline bool mlx5e_macsec_is_rx_flow(struct mlx5_cqe64 *cqe) { return false; }
+static inline void mlx5e_macsec_offload_handle_rx_skb(struct net_device *netdev,
+						      struct sk_buff *skb,
+						      struct mlx5_cqe64 *cqe)
+{}
+static inline bool mlx5e_is_macsec_device(const struct mlx5_core_dev *mdev) { return false; }
+#endif  /* CONFIG_MLX5_EN_MACSEC */
+
+#endif	/* __MLX5_ACCEL_EN_MACSEC_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec_fs.c b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec_fs.c
new file mode 100644
index 000000000..6ecf0bf23
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec_fs.c
@@ -0,0 +1,1390 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2022, NVIDIA CORPORATION & AFFILIATES. All rights reserved. */
+
+#include <net/macsec.h>
+#include <linux/netdevice.h>
+#include <linux/mlx5/qp.h>
+#include "fs_core.h"
+#include "en/fs.h"
+#include "en_accel/macsec_fs.h"
+#include "mlx5_core.h"
+
+/* MACsec TX flow steering */
+#define CRYPTO_NUM_MAXSEC_FTE BIT(15)
+#define CRYPTO_TABLE_DEFAULT_RULE_GROUP_SIZE 1
+
+#define TX_CRYPTO_TABLE_LEVEL 0
+#define TX_CRYPTO_TABLE_NUM_GROUPS 3
+#define TX_CRYPTO_TABLE_MKE_GROUP_SIZE 1
+#define TX_CRYPTO_TABLE_SA_GROUP_SIZE \
+	(CRYPTO_NUM_MAXSEC_FTE - (TX_CRYPTO_TABLE_MKE_GROUP_SIZE + \
+				  CRYPTO_TABLE_DEFAULT_RULE_GROUP_SIZE))
+#define TX_CHECK_TABLE_LEVEL 1
+#define TX_CHECK_TABLE_NUM_FTE 2
+#define RX_CRYPTO_TABLE_LEVEL 0
+#define RX_CHECK_TABLE_LEVEL 1
+#define RX_CHECK_TABLE_NUM_FTE 3
+#define RX_CRYPTO_TABLE_NUM_GROUPS 3
+#define RX_CRYPTO_TABLE_SA_RULE_WITH_SCI_GROUP_SIZE \
+	((CRYPTO_NUM_MAXSEC_FTE - CRYPTO_TABLE_DEFAULT_RULE_GROUP_SIZE) / 2)
+#define RX_CRYPTO_TABLE_SA_RULE_WITHOUT_SCI_GROUP_SIZE \
+	(CRYPTO_NUM_MAXSEC_FTE - RX_CRYPTO_TABLE_SA_RULE_WITH_SCI_GROUP_SIZE)
+#define RX_NUM_OF_RULES_PER_SA 2
+
+#define MLX5_MACSEC_TAG_LEN 8 /* SecTAG length with ethertype and without the optional SCI */
+#define MLX5_MACSEC_SECTAG_TCI_AN_FIELD_BITMASK 0x23
+#define MLX5_MACSEC_SECTAG_TCI_AN_FIELD_OFFSET 0x8
+#define MLX5_MACSEC_SECTAG_TCI_SC_FIELD_OFFSET 0x5
+#define MLX5_MACSEC_SECTAG_TCI_SC_FIELD_BIT (0x1 << MLX5_MACSEC_SECTAG_TCI_SC_FIELD_OFFSET)
+#define MLX5_SECTAG_HEADER_SIZE_WITHOUT_SCI 0x8
+#define MLX5_SECTAG_HEADER_SIZE_WITH_SCI (MLX5_SECTAG_HEADER_SIZE_WITHOUT_SCI + MACSEC_SCI_LEN)
+
+/* MACsec RX flow steering */
+#define MLX5_ETH_WQE_FT_META_MACSEC_MASK 0x3E
+
+struct mlx5_sectag_header {
+	__be16 ethertype;
+	u8 tci_an;
+	u8 sl;
+	u32 pn;
+	u8 sci[MACSEC_SCI_LEN]; /* optional */
+}  __packed;
+
+struct mlx5e_macsec_tx_rule {
+	struct mlx5_flow_handle *rule;
+	struct mlx5_pkt_reformat *pkt_reformat;
+	u32 fs_id;
+};
+
+struct mlx5e_macsec_tables {
+	struct mlx5e_flow_table ft_crypto;
+	struct mlx5_flow_handle *crypto_miss_rule;
+
+	struct mlx5_flow_table *ft_check;
+	struct mlx5_flow_group  *ft_check_group;
+	struct mlx5_fc *check_miss_rule_counter;
+	struct mlx5_flow_handle *check_miss_rule;
+	struct mlx5_fc *check_rule_counter;
+
+	u32 refcnt;
+};
+
+struct mlx5e_macsec_tx {
+	struct mlx5_flow_handle *crypto_mke_rule;
+	struct mlx5_flow_handle *check_rule;
+
+	struct ida tx_halloc;
+
+	struct mlx5e_macsec_tables tables;
+};
+
+struct mlx5e_macsec_rx_rule {
+	struct mlx5_flow_handle *rule[RX_NUM_OF_RULES_PER_SA];
+	struct mlx5_modify_hdr *meta_modhdr;
+};
+
+struct mlx5e_macsec_rx {
+	struct mlx5_flow_handle *check_rule[2];
+	struct mlx5_pkt_reformat *check_rule_pkt_reformat[2];
+
+	struct mlx5e_macsec_tables tables;
+};
+
+union mlx5e_macsec_rule {
+	struct mlx5e_macsec_tx_rule tx_rule;
+	struct mlx5e_macsec_rx_rule rx_rule;
+};
+
+struct mlx5e_macsec_fs {
+	struct mlx5_core_dev *mdev;
+	struct net_device *netdev;
+	struct mlx5e_macsec_tx *tx_fs;
+	struct mlx5e_macsec_rx *rx_fs;
+};
+
+static void macsec_fs_tx_destroy(struct mlx5e_macsec_fs *macsec_fs)
+{
+	struct mlx5e_macsec_tx *tx_fs = macsec_fs->tx_fs;
+	struct mlx5e_macsec_tables *tx_tables;
+
+	tx_tables = &tx_fs->tables;
+
+	/* Tx check table */
+	if (tx_fs->check_rule) {
+		mlx5_del_flow_rules(tx_fs->check_rule);
+		tx_fs->check_rule = NULL;
+	}
+
+	if (tx_tables->check_miss_rule) {
+		mlx5_del_flow_rules(tx_tables->check_miss_rule);
+		tx_tables->check_miss_rule = NULL;
+	}
+
+	if (tx_tables->ft_check_group) {
+		mlx5_destroy_flow_group(tx_tables->ft_check_group);
+		tx_tables->ft_check_group = NULL;
+	}
+
+	if (tx_tables->ft_check) {
+		mlx5_destroy_flow_table(tx_tables->ft_check);
+		tx_tables->ft_check = NULL;
+	}
+
+	/* Tx crypto table */
+	if (tx_fs->crypto_mke_rule) {
+		mlx5_del_flow_rules(tx_fs->crypto_mke_rule);
+		tx_fs->crypto_mke_rule = NULL;
+	}
+
+	if (tx_tables->crypto_miss_rule) {
+		mlx5_del_flow_rules(tx_tables->crypto_miss_rule);
+		tx_tables->crypto_miss_rule = NULL;
+	}
+
+	mlx5e_destroy_flow_table(&tx_tables->ft_crypto);
+}
+
+static int macsec_fs_tx_create_crypto_table_groups(struct mlx5e_flow_table *ft)
+{
+	int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
+	int mclen = MLX5_ST_SZ_BYTES(fte_match_param);
+	int ix = 0;
+	u32 *in;
+	int err;
+	u8 *mc;
+
+	ft->g = kcalloc(TX_CRYPTO_TABLE_NUM_GROUPS, sizeof(*ft->g), GFP_KERNEL);
+	if (!ft->g)
+		return -ENOMEM;
+	in = kvzalloc(inlen, GFP_KERNEL);
+
+	if (!in) {
+		kfree(ft->g);
+		ft->g = NULL;
+		return -ENOMEM;
+	}
+
+	mc = MLX5_ADDR_OF(create_flow_group_in, in, match_criteria);
+
+	/* Flow Group for MKE match */
+	MLX5_SET_CFG(in, match_criteria_enable, MLX5_MATCH_OUTER_HEADERS);
+	MLX5_SET_TO_ONES(fte_match_param, mc, outer_headers.ethertype);
+
+	MLX5_SET_CFG(in, start_flow_index, ix);
+	ix += TX_CRYPTO_TABLE_MKE_GROUP_SIZE;
+	MLX5_SET_CFG(in, end_flow_index, ix - 1);
+	ft->g[ft->num_groups] = mlx5_create_flow_group(ft->t, in);
+	if (IS_ERR(ft->g[ft->num_groups]))
+		goto err;
+	ft->num_groups++;
+
+	/* Flow Group for SA rules */
+	memset(in, 0, inlen);
+	memset(mc, 0, mclen);
+	MLX5_SET_CFG(in, match_criteria_enable, MLX5_MATCH_MISC_PARAMETERS_2);
+	MLX5_SET(fte_match_param, mc, misc_parameters_2.metadata_reg_a,
+		 MLX5_ETH_WQE_FT_META_MACSEC_MASK);
+
+	MLX5_SET_CFG(in, start_flow_index, ix);
+	ix += TX_CRYPTO_TABLE_SA_GROUP_SIZE;
+	MLX5_SET_CFG(in, end_flow_index, ix - 1);
+	ft->g[ft->num_groups] = mlx5_create_flow_group(ft->t, in);
+	if (IS_ERR(ft->g[ft->num_groups]))
+		goto err;
+	ft->num_groups++;
+
+	/* Flow Group for l2 traps */
+	memset(in, 0, inlen);
+	memset(mc, 0, mclen);
+	MLX5_SET_CFG(in, start_flow_index, ix);
+	ix += CRYPTO_TABLE_DEFAULT_RULE_GROUP_SIZE;
+	MLX5_SET_CFG(in, end_flow_index, ix - 1);
+	ft->g[ft->num_groups] = mlx5_create_flow_group(ft->t, in);
+	if (IS_ERR(ft->g[ft->num_groups]))
+		goto err;
+	ft->num_groups++;
+
+	kvfree(in);
+	return 0;
+
+err:
+	err = PTR_ERR(ft->g[ft->num_groups]);
+	ft->g[ft->num_groups] = NULL;
+	kvfree(in);
+
+	return err;
+}
+
+static struct mlx5_flow_table
+	*macsec_fs_auto_group_table_create(struct mlx5_flow_namespace *ns, int flags,
+					   int level, int max_fte)
+{
+	struct mlx5_flow_table_attr ft_attr = {};
+	struct mlx5_flow_table *fdb = NULL;
+
+	/* reserve entry for the match all miss group and rule */
+	ft_attr.autogroup.num_reserved_entries = 1;
+	ft_attr.autogroup.max_num_groups = 1;
+	ft_attr.prio = 0;
+	ft_attr.flags = flags;
+	ft_attr.level = level;
+	ft_attr.max_fte = max_fte;
+
+	fdb = mlx5_create_auto_grouped_flow_table(ns, &ft_attr);
+
+	return fdb;
+}
+
+static int macsec_fs_tx_create(struct mlx5e_macsec_fs *macsec_fs)
+{
+	int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
+	struct mlx5e_macsec_tx *tx_fs = macsec_fs->tx_fs;
+	struct net_device *netdev = macsec_fs->netdev;
+	struct mlx5_flow_table_attr ft_attr = {};
+	struct mlx5_flow_destination dest = {};
+	struct mlx5e_macsec_tables *tx_tables;
+	struct mlx5_flow_act flow_act = {};
+	struct mlx5e_flow_table *ft_crypto;
+	struct mlx5_flow_table *flow_table;
+	struct mlx5_flow_group *flow_group;
+	struct mlx5_flow_namespace *ns;
+	struct mlx5_flow_handle *rule;
+	struct mlx5_flow_spec *spec;
+	u32 *flow_group_in;
+	int err;
+
+	ns = mlx5_get_flow_namespace(macsec_fs->mdev, MLX5_FLOW_NAMESPACE_EGRESS_MACSEC);
+	if (!ns)
+		return -ENOMEM;
+
+	spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
+	if (!spec)
+		return -ENOMEM;
+
+	flow_group_in = kvzalloc(inlen, GFP_KERNEL);
+	if (!flow_group_in) {
+		err = -ENOMEM;
+		goto out_spec;
+	}
+
+	tx_tables = &tx_fs->tables;
+	ft_crypto = &tx_tables->ft_crypto;
+
+	/* Tx crypto table  */
+	ft_attr.flags = MLX5_FLOW_TABLE_TUNNEL_EN_REFORMAT;
+	ft_attr.level = TX_CRYPTO_TABLE_LEVEL;
+	ft_attr.max_fte = CRYPTO_NUM_MAXSEC_FTE;
+
+	flow_table = mlx5_create_flow_table(ns, &ft_attr);
+	if (IS_ERR(flow_table)) {
+		err = PTR_ERR(flow_table);
+		netdev_err(netdev, "Failed to create MACsec Tx crypto table err(%d)\n", err);
+		goto out_flow_group;
+	}
+	ft_crypto->t = flow_table;
+
+	/* Tx crypto table groups */
+	err = macsec_fs_tx_create_crypto_table_groups(ft_crypto);
+	if (err) {
+		netdev_err(netdev,
+			   "Failed to create default flow group for MACsec Tx crypto table err(%d)\n",
+			   err);
+		goto err;
+	}
+
+	/* Tx crypto table MKE rule - MKE packets shouldn't be offloaded */
+	memset(&flow_act, 0, sizeof(flow_act));
+	memset(spec, 0, sizeof(*spec));
+	spec->match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
+
+	MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria, outer_headers.ethertype);
+	MLX5_SET(fte_match_param, spec->match_value, outer_headers.ethertype, ETH_P_PAE);
+	flow_act.action = MLX5_FLOW_CONTEXT_ACTION_ALLOW;
+
+	rule = mlx5_add_flow_rules(ft_crypto->t, spec, &flow_act, NULL, 0);
+	if (IS_ERR(rule)) {
+		err = PTR_ERR(rule);
+		netdev_err(netdev, "Failed to add MACsec TX MKE rule, err=%d\n", err);
+		goto err;
+	}
+	tx_fs->crypto_mke_rule = rule;
+
+	/* Tx crypto table Default miss rule */
+	memset(&flow_act, 0, sizeof(flow_act));
+	flow_act.action = MLX5_FLOW_CONTEXT_ACTION_ALLOW;
+	rule = mlx5_add_flow_rules(ft_crypto->t, NULL, &flow_act, NULL, 0);
+	if (IS_ERR(rule)) {
+		err = PTR_ERR(rule);
+		netdev_err(netdev, "Failed to add MACsec Tx table default miss rule %d\n", err);
+		goto err;
+	}
+	tx_tables->crypto_miss_rule = rule;
+
+	/* Tx check table */
+	flow_table = macsec_fs_auto_group_table_create(ns, 0, TX_CHECK_TABLE_LEVEL,
+						       TX_CHECK_TABLE_NUM_FTE);
+	if (IS_ERR(flow_table)) {
+		err = PTR_ERR(flow_table);
+		netdev_err(netdev, "fail to create MACsec TX check table, err(%d)\n", err);
+		goto err;
+	}
+	tx_tables->ft_check = flow_table;
+
+	/* Tx check table Default miss group/rule */
+	memset(flow_group_in, 0, inlen);
+	MLX5_SET(create_flow_group_in, flow_group_in, start_flow_index, flow_table->max_fte - 1);
+	MLX5_SET(create_flow_group_in, flow_group_in, end_flow_index, flow_table->max_fte - 1);
+	flow_group = mlx5_create_flow_group(tx_tables->ft_check, flow_group_in);
+	if (IS_ERR(flow_group)) {
+		err = PTR_ERR(flow_group);
+		netdev_err(netdev,
+			   "Failed to create default flow group for MACsec Tx crypto table err(%d)\n",
+			   err);
+		goto err;
+	}
+	tx_tables->ft_check_group = flow_group;
+
+	/* Tx check table default drop rule */
+	memset(&dest, 0, sizeof(struct mlx5_flow_destination));
+	memset(&flow_act, 0, sizeof(flow_act));
+	dest.type = MLX5_FLOW_DESTINATION_TYPE_COUNTER;
+	dest.counter_id = mlx5_fc_id(tx_tables->check_miss_rule_counter);
+	flow_act.action = MLX5_FLOW_CONTEXT_ACTION_DROP | MLX5_FLOW_CONTEXT_ACTION_COUNT;
+	rule = mlx5_add_flow_rules(tx_tables->ft_check,  NULL, &flow_act, &dest, 1);
+	if (IS_ERR(rule)) {
+		err = PTR_ERR(rule);
+		netdev_err(netdev, "Failed to added MACsec tx check drop rule, err(%d)\n", err);
+		goto err;
+	}
+	tx_tables->check_miss_rule = rule;
+
+	/* Tx check table rule */
+	memset(spec, 0, sizeof(struct mlx5_flow_spec));
+	memset(&dest, 0, sizeof(struct mlx5_flow_destination));
+	memset(&flow_act, 0, sizeof(flow_act));
+
+	MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria, misc_parameters_2.metadata_reg_c_4);
+	MLX5_SET(fte_match_param, spec->match_value, misc_parameters_2.metadata_reg_c_4, 0);
+	spec->match_criteria_enable = MLX5_MATCH_MISC_PARAMETERS_2;
+
+	flow_act.flags = FLOW_ACT_NO_APPEND;
+	flow_act.action = MLX5_FLOW_CONTEXT_ACTION_ALLOW | MLX5_FLOW_CONTEXT_ACTION_COUNT;
+	dest.type = MLX5_FLOW_DESTINATION_TYPE_COUNTER;
+	dest.counter_id = mlx5_fc_id(tx_tables->check_rule_counter);
+	rule = mlx5_add_flow_rules(tx_tables->ft_check, spec, &flow_act, &dest, 1);
+	if (IS_ERR(rule)) {
+		err = PTR_ERR(rule);
+		netdev_err(netdev, "Failed to add MACsec check rule, err=%d\n", err);
+		goto err;
+	}
+	tx_fs->check_rule = rule;
+
+	goto out_flow_group;
+
+err:
+	macsec_fs_tx_destroy(macsec_fs);
+out_flow_group:
+	kvfree(flow_group_in);
+out_spec:
+	kvfree(spec);
+	return err;
+}
+
+static int macsec_fs_tx_ft_get(struct mlx5e_macsec_fs *macsec_fs)
+{
+	struct mlx5e_macsec_tx *tx_fs = macsec_fs->tx_fs;
+	struct mlx5e_macsec_tables *tx_tables;
+	int err = 0;
+
+	tx_tables = &tx_fs->tables;
+	if (tx_tables->refcnt)
+		goto out;
+
+	err = macsec_fs_tx_create(macsec_fs);
+	if (err)
+		return err;
+
+out:
+	tx_tables->refcnt++;
+	return err;
+}
+
+static void macsec_fs_tx_ft_put(struct mlx5e_macsec_fs *macsec_fs)
+{
+	struct mlx5e_macsec_tables *tx_tables = &macsec_fs->tx_fs->tables;
+
+	if (--tx_tables->refcnt)
+		return;
+
+	macsec_fs_tx_destroy(macsec_fs);
+}
+
+static int macsec_fs_tx_setup_fte(struct mlx5e_macsec_fs *macsec_fs,
+				  struct mlx5_flow_spec *spec,
+				  struct mlx5_flow_act *flow_act,
+				  u32 macsec_obj_id,
+				  u32 *fs_id)
+{
+	struct mlx5e_macsec_tx *tx_fs = macsec_fs->tx_fs;
+	int err = 0;
+	u32 id;
+
+	err = ida_alloc_range(&tx_fs->tx_halloc, 1,
+			      MLX5_MACSEC_NUM_OF_SUPPORTED_INTERFACES,
+			      GFP_KERNEL);
+	if (err < 0)
+		return err;
+
+	id = err;
+	spec->match_criteria_enable |= MLX5_MATCH_MISC_PARAMETERS_2;
+
+	/* Metadata match */
+	MLX5_SET(fte_match_param, spec->match_criteria, misc_parameters_2.metadata_reg_a,
+		 MLX5_ETH_WQE_FT_META_MACSEC_MASK);
+	MLX5_SET(fte_match_param, spec->match_value, misc_parameters_2.metadata_reg_a,
+		 MLX5_ETH_WQE_FT_META_MACSEC | id << 2);
+
+	*fs_id = id;
+	flow_act->crypto.type = MLX5_FLOW_CONTEXT_ENCRYPT_DECRYPT_TYPE_MACSEC;
+	flow_act->crypto.obj_id = macsec_obj_id;
+
+	mlx5_core_dbg(macsec_fs->mdev, "Tx fte: macsec obj_id %u, fs_id %u\n", macsec_obj_id, id);
+	return 0;
+}
+
+static void macsec_fs_tx_create_sectag_header(const struct macsec_context *ctx,
+					      char *reformatbf,
+					      size_t *reformat_size)
+{
+	const struct macsec_secy *secy = ctx->secy;
+	bool sci_present = macsec_send_sci(secy);
+	struct mlx5_sectag_header sectag = {};
+	const struct macsec_tx_sc *tx_sc;
+
+	tx_sc = &secy->tx_sc;
+	sectag.ethertype = htons(ETH_P_MACSEC);
+
+	if (sci_present) {
+		sectag.tci_an |= MACSEC_TCI_SC;
+		memcpy(&sectag.sci, &secy->sci,
+		       sizeof(sectag.sci));
+	} else {
+		if (tx_sc->end_station)
+			sectag.tci_an |= MACSEC_TCI_ES;
+		if (tx_sc->scb)
+			sectag.tci_an |= MACSEC_TCI_SCB;
+	}
+
+	/* With GCM, C/E clear for !encrypt, both set for encrypt */
+	if (tx_sc->encrypt)
+		sectag.tci_an |= MACSEC_TCI_CONFID;
+	else if (secy->icv_len != MACSEC_DEFAULT_ICV_LEN)
+		sectag.tci_an |= MACSEC_TCI_C;
+
+	sectag.tci_an |= tx_sc->encoding_sa;
+
+	*reformat_size = MLX5_MACSEC_TAG_LEN + (sci_present ? MACSEC_SCI_LEN : 0);
+
+	memcpy(reformatbf, &sectag, *reformat_size);
+}
+
+static void macsec_fs_tx_del_rule(struct mlx5e_macsec_fs *macsec_fs,
+				  struct mlx5e_macsec_tx_rule *tx_rule)
+{
+	if (tx_rule->rule) {
+		mlx5_del_flow_rules(tx_rule->rule);
+		tx_rule->rule = NULL;
+	}
+
+	if (tx_rule->pkt_reformat) {
+		mlx5_packet_reformat_dealloc(macsec_fs->mdev, tx_rule->pkt_reformat);
+		tx_rule->pkt_reformat = NULL;
+	}
+
+	if (tx_rule->fs_id) {
+		ida_free(&macsec_fs->tx_fs->tx_halloc, tx_rule->fs_id);
+		tx_rule->fs_id = 0;
+	}
+
+	kfree(tx_rule);
+
+	macsec_fs_tx_ft_put(macsec_fs);
+}
+
+static union mlx5e_macsec_rule *
+macsec_fs_tx_add_rule(struct mlx5e_macsec_fs *macsec_fs,
+		      const struct macsec_context *macsec_ctx,
+		      struct mlx5_macsec_rule_attrs *attrs,
+		      u32 *sa_fs_id)
+{
+	char reformatbf[MLX5_MACSEC_TAG_LEN + MACSEC_SCI_LEN];
+	struct mlx5_pkt_reformat_params reformat_params = {};
+	struct mlx5e_macsec_tx *tx_fs = macsec_fs->tx_fs;
+	struct net_device *netdev = macsec_fs->netdev;
+	union mlx5e_macsec_rule *macsec_rule = NULL;
+	struct mlx5_flow_destination dest = {};
+	struct mlx5e_macsec_tables *tx_tables;
+	struct mlx5e_macsec_tx_rule *tx_rule;
+	struct mlx5_flow_act flow_act = {};
+	struct mlx5_flow_handle *rule;
+	struct mlx5_flow_spec *spec;
+	size_t reformat_size;
+	int err = 0;
+	u32 fs_id;
+
+	tx_tables = &tx_fs->tables;
+
+	spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
+	if (!spec)
+		return NULL;
+
+	err = macsec_fs_tx_ft_get(macsec_fs);
+	if (err)
+		goto out_spec;
+
+	macsec_rule = kzalloc(sizeof(*macsec_rule), GFP_KERNEL);
+	if (!macsec_rule) {
+		macsec_fs_tx_ft_put(macsec_fs);
+		goto out_spec;
+	}
+
+	tx_rule = &macsec_rule->tx_rule;
+
+	/* Tx crypto table crypto rule */
+	macsec_fs_tx_create_sectag_header(macsec_ctx, reformatbf, &reformat_size);
+
+	reformat_params.type = MLX5_REFORMAT_TYPE_ADD_MACSEC;
+	reformat_params.size = reformat_size;
+	reformat_params.data = reformatbf;
+	flow_act.pkt_reformat = mlx5_packet_reformat_alloc(macsec_fs->mdev,
+							   &reformat_params,
+							   MLX5_FLOW_NAMESPACE_EGRESS_MACSEC);
+	if (IS_ERR(flow_act.pkt_reformat)) {
+		err = PTR_ERR(flow_act.pkt_reformat);
+		netdev_err(netdev, "Failed to allocate MACsec Tx reformat context err=%d\n",  err);
+		goto err;
+	}
+	tx_rule->pkt_reformat = flow_act.pkt_reformat;
+
+	err = macsec_fs_tx_setup_fte(macsec_fs, spec, &flow_act, attrs->macsec_obj_id, &fs_id);
+	if (err) {
+		netdev_err(netdev,
+			   "Failed to add packet reformat for MACsec TX crypto rule, err=%d\n",
+			   err);
+		goto err;
+	}
+
+	tx_rule->fs_id = fs_id;
+	*sa_fs_id = fs_id;
+
+	flow_act.action = MLX5_FLOW_CONTEXT_ACTION_FWD_DEST |
+			  MLX5_FLOW_CONTEXT_ACTION_CRYPTO_ENCRYPT |
+			  MLX5_FLOW_CONTEXT_ACTION_PACKET_REFORMAT;
+	dest.type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
+	dest.ft = tx_tables->ft_check;
+	rule = mlx5_add_flow_rules(tx_tables->ft_crypto.t, spec, &flow_act, &dest, 1);
+	if (IS_ERR(rule)) {
+		err = PTR_ERR(rule);
+		netdev_err(netdev, "Failed to add MACsec TX crypto rule, err=%d\n", err);
+		goto err;
+	}
+	tx_rule->rule = rule;
+
+	goto out_spec;
+
+err:
+	macsec_fs_tx_del_rule(macsec_fs, tx_rule);
+	macsec_rule = NULL;
+out_spec:
+	kvfree(spec);
+
+	return macsec_rule;
+}
+
+static void macsec_fs_tx_cleanup(struct mlx5e_macsec_fs *macsec_fs)
+{
+	struct mlx5e_macsec_tx *tx_fs = macsec_fs->tx_fs;
+	struct mlx5_core_dev *mdev = macsec_fs->mdev;
+	struct mlx5e_macsec_tables *tx_tables;
+
+	if (!tx_fs)
+		return;
+
+	tx_tables = &tx_fs->tables;
+	if (tx_tables->refcnt) {
+		netdev_err(macsec_fs->netdev,
+			   "Can't destroy MACsec offload tx_fs, refcnt(%u) isn't 0\n",
+			   tx_tables->refcnt);
+		return;
+	}
+
+	ida_destroy(&tx_fs->tx_halloc);
+
+	if (tx_tables->check_miss_rule_counter) {
+		mlx5_fc_destroy(mdev, tx_tables->check_miss_rule_counter);
+		tx_tables->check_miss_rule_counter = NULL;
+	}
+
+	if (tx_tables->check_rule_counter) {
+		mlx5_fc_destroy(mdev, tx_tables->check_rule_counter);
+		tx_tables->check_rule_counter = NULL;
+	}
+
+	kfree(tx_fs);
+	macsec_fs->tx_fs = NULL;
+}
+
+static int macsec_fs_tx_init(struct mlx5e_macsec_fs *macsec_fs)
+{
+	struct net_device *netdev = macsec_fs->netdev;
+	struct mlx5_core_dev *mdev = macsec_fs->mdev;
+	struct mlx5e_macsec_tables *tx_tables;
+	struct mlx5e_macsec_tx *tx_fs;
+	struct mlx5_fc *flow_counter;
+	int err;
+
+	tx_fs = kzalloc(sizeof(*tx_fs), GFP_KERNEL);
+	if (!tx_fs)
+		return -ENOMEM;
+
+	tx_tables = &tx_fs->tables;
+
+	flow_counter = mlx5_fc_create(mdev, false);
+	if (IS_ERR(flow_counter)) {
+		err = PTR_ERR(flow_counter);
+		netdev_err(netdev,
+			   "Failed to create MACsec Tx encrypt flow counter, err(%d)\n",
+			   err);
+		goto err_encrypt_counter;
+	}
+	tx_tables->check_rule_counter = flow_counter;
+
+	flow_counter = mlx5_fc_create(mdev, false);
+	if (IS_ERR(flow_counter)) {
+		err = PTR_ERR(flow_counter);
+		netdev_err(netdev,
+			   "Failed to create MACsec Tx drop flow counter, err(%d)\n",
+			   err);
+		goto err_drop_counter;
+	}
+	tx_tables->check_miss_rule_counter = flow_counter;
+
+	ida_init(&tx_fs->tx_halloc);
+
+	macsec_fs->tx_fs = tx_fs;
+
+	return 0;
+
+err_drop_counter:
+	mlx5_fc_destroy(mdev, tx_tables->check_rule_counter);
+	tx_tables->check_rule_counter = NULL;
+
+err_encrypt_counter:
+	kfree(tx_fs);
+	macsec_fs->tx_fs = NULL;
+
+	return err;
+}
+
+static void macsec_fs_rx_destroy(struct mlx5e_macsec_fs *macsec_fs)
+{
+	struct mlx5e_macsec_rx *rx_fs = macsec_fs->rx_fs;
+	struct mlx5e_macsec_tables *rx_tables;
+	int i;
+
+	/* Rx check table */
+	for (i = 1; i >= 0; --i) {
+		if (rx_fs->check_rule[i]) {
+			mlx5_del_flow_rules(rx_fs->check_rule[i]);
+			rx_fs->check_rule[i] = NULL;
+		}
+
+		if (rx_fs->check_rule_pkt_reformat[i]) {
+			mlx5_packet_reformat_dealloc(macsec_fs->mdev,
+						     rx_fs->check_rule_pkt_reformat[i]);
+			rx_fs->check_rule_pkt_reformat[i] = NULL;
+		}
+	}
+
+	rx_tables = &rx_fs->tables;
+
+	if (rx_tables->check_miss_rule) {
+		mlx5_del_flow_rules(rx_tables->check_miss_rule);
+		rx_tables->check_miss_rule = NULL;
+	}
+
+	if (rx_tables->ft_check_group) {
+		mlx5_destroy_flow_group(rx_tables->ft_check_group);
+		rx_tables->ft_check_group = NULL;
+	}
+
+	if (rx_tables->ft_check) {
+		mlx5_destroy_flow_table(rx_tables->ft_check);
+		rx_tables->ft_check = NULL;
+	}
+
+	/* Rx crypto table */
+	if (rx_tables->crypto_miss_rule) {
+		mlx5_del_flow_rules(rx_tables->crypto_miss_rule);
+		rx_tables->crypto_miss_rule = NULL;
+	}
+
+	mlx5e_destroy_flow_table(&rx_tables->ft_crypto);
+}
+
+static int macsec_fs_rx_create_crypto_table_groups(struct mlx5e_flow_table *ft)
+{
+	int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
+	int mclen = MLX5_ST_SZ_BYTES(fte_match_param);
+	int ix = 0;
+	u32 *in;
+	int err;
+	u8 *mc;
+
+	ft->g = kcalloc(RX_CRYPTO_TABLE_NUM_GROUPS, sizeof(*ft->g), GFP_KERNEL);
+	if (!ft->g)
+		return -ENOMEM;
+
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in) {
+		kfree(ft->g);
+		return -ENOMEM;
+	}
+
+	mc = MLX5_ADDR_OF(create_flow_group_in, in, match_criteria);
+
+	/* Flow group for SA rule with SCI */
+	MLX5_SET_CFG(in, match_criteria_enable, MLX5_MATCH_OUTER_HEADERS |
+						MLX5_MATCH_MISC_PARAMETERS_5);
+	MLX5_SET_TO_ONES(fte_match_param, mc, outer_headers.ethertype);
+
+	MLX5_SET(fte_match_param, mc, misc_parameters_5.macsec_tag_0,
+		 MLX5_MACSEC_SECTAG_TCI_AN_FIELD_BITMASK <<
+		 MLX5_MACSEC_SECTAG_TCI_AN_FIELD_OFFSET);
+	MLX5_SET_TO_ONES(fte_match_param, mc, misc_parameters_5.macsec_tag_2);
+	MLX5_SET_TO_ONES(fte_match_param, mc, misc_parameters_5.macsec_tag_3);
+
+	MLX5_SET_CFG(in, start_flow_index, ix);
+	ix += RX_CRYPTO_TABLE_SA_RULE_WITH_SCI_GROUP_SIZE;
+	MLX5_SET_CFG(in, end_flow_index, ix - 1);
+	ft->g[ft->num_groups] = mlx5_create_flow_group(ft->t, in);
+	if (IS_ERR(ft->g[ft->num_groups]))
+		goto err;
+	ft->num_groups++;
+
+	/* Flow group for SA rule without SCI */
+	memset(in, 0, inlen);
+	memset(mc, 0, mclen);
+	MLX5_SET_CFG(in, match_criteria_enable, MLX5_MATCH_OUTER_HEADERS |
+						MLX5_MATCH_MISC_PARAMETERS_5);
+	MLX5_SET_TO_ONES(fte_match_param, mc, outer_headers.smac_47_16);
+	MLX5_SET_TO_ONES(fte_match_param, mc, outer_headers.smac_15_0);
+	MLX5_SET_TO_ONES(fte_match_param, mc, outer_headers.ethertype);
+
+	MLX5_SET(fte_match_param, mc, misc_parameters_5.macsec_tag_0,
+		 MLX5_MACSEC_SECTAG_TCI_AN_FIELD_BITMASK << MLX5_MACSEC_SECTAG_TCI_AN_FIELD_OFFSET);
+
+	MLX5_SET_CFG(in, start_flow_index, ix);
+	ix += RX_CRYPTO_TABLE_SA_RULE_WITHOUT_SCI_GROUP_SIZE;
+	MLX5_SET_CFG(in, end_flow_index, ix - 1);
+	ft->g[ft->num_groups] = mlx5_create_flow_group(ft->t, in);
+	if (IS_ERR(ft->g[ft->num_groups]))
+		goto err;
+	ft->num_groups++;
+
+	/* Flow Group for l2 traps */
+	memset(in, 0, inlen);
+	memset(mc, 0, mclen);
+	MLX5_SET_CFG(in, start_flow_index, ix);
+	ix += CRYPTO_TABLE_DEFAULT_RULE_GROUP_SIZE;
+	MLX5_SET_CFG(in, end_flow_index, ix - 1);
+	ft->g[ft->num_groups] = mlx5_create_flow_group(ft->t, in);
+	if (IS_ERR(ft->g[ft->num_groups]))
+		goto err;
+	ft->num_groups++;
+
+	kvfree(in);
+	return 0;
+
+err:
+	err = PTR_ERR(ft->g[ft->num_groups]);
+	ft->g[ft->num_groups] = NULL;
+	kvfree(in);
+
+	return err;
+}
+
+static int macsec_fs_rx_create_check_decap_rule(struct mlx5e_macsec_fs *macsec_fs,
+						struct mlx5_flow_destination *dest,
+						struct mlx5_flow_act *flow_act,
+						struct mlx5_flow_spec *spec,
+						int reformat_param_size)
+{
+	int rule_index = (reformat_param_size == MLX5_SECTAG_HEADER_SIZE_WITH_SCI) ? 0 : 1;
+	u8 mlx5_reformat_buf[MLX5_SECTAG_HEADER_SIZE_WITH_SCI];
+	struct mlx5_pkt_reformat_params reformat_params = {};
+	struct mlx5e_macsec_rx *rx_fs = macsec_fs->rx_fs;
+	struct net_device *netdev = macsec_fs->netdev;
+	struct mlx5e_macsec_tables *rx_tables;
+	struct mlx5_flow_handle *rule;
+	int err = 0;
+
+	rx_tables = &rx_fs->tables;
+
+	/* Rx check table decap 16B rule */
+	memset(dest, 0, sizeof(*dest));
+	memset(flow_act, 0, sizeof(*flow_act));
+	memset(spec, 0, sizeof(*spec));
+
+	reformat_params.type = MLX5_REFORMAT_TYPE_DEL_MACSEC;
+	reformat_params.size = reformat_param_size;
+	reformat_params.data = mlx5_reformat_buf;
+	flow_act->pkt_reformat = mlx5_packet_reformat_alloc(macsec_fs->mdev,
+							    &reformat_params,
+							    MLX5_FLOW_NAMESPACE_KERNEL_RX_MACSEC);
+	if (IS_ERR(flow_act->pkt_reformat)) {
+		err = PTR_ERR(flow_act->pkt_reformat);
+		netdev_err(netdev, "Failed to allocate MACsec Rx reformat context err=%d\n", err);
+		return err;
+	}
+	rx_fs->check_rule_pkt_reformat[rule_index] = flow_act->pkt_reformat;
+
+	spec->match_criteria_enable = MLX5_MATCH_MISC_PARAMETERS_2;
+	/* MACsec syndrome match */
+	MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria, misc_parameters_2.macsec_syndrome);
+	MLX5_SET(fte_match_param, spec->match_value, misc_parameters_2.macsec_syndrome, 0);
+	/* ASO return reg syndrome match */
+	MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria, misc_parameters_2.metadata_reg_c_4);
+	MLX5_SET(fte_match_param, spec->match_value, misc_parameters_2.metadata_reg_c_4, 0);
+
+	spec->match_criteria_enable |= MLX5_MATCH_MISC_PARAMETERS_5;
+	/* Sectag TCI SC present bit*/
+	MLX5_SET(fte_match_param, spec->match_criteria, misc_parameters_5.macsec_tag_0,
+		 MLX5_MACSEC_SECTAG_TCI_SC_FIELD_BIT << MLX5_MACSEC_SECTAG_TCI_AN_FIELD_OFFSET);
+
+	if (reformat_param_size == MLX5_SECTAG_HEADER_SIZE_WITH_SCI)
+		MLX5_SET(fte_match_param, spec->match_value, misc_parameters_5.macsec_tag_0,
+			 MLX5_MACSEC_SECTAG_TCI_SC_FIELD_BIT <<
+			 MLX5_MACSEC_SECTAG_TCI_AN_FIELD_OFFSET);
+
+	flow_act->flags = FLOW_ACT_NO_APPEND;
+	flow_act->action = MLX5_FLOW_CONTEXT_ACTION_FWD_NEXT_PRIO |
+			   MLX5_FLOW_CONTEXT_ACTION_PACKET_REFORMAT |
+			   MLX5_FLOW_CONTEXT_ACTION_COUNT;
+	dest->type = MLX5_FLOW_DESTINATION_TYPE_COUNTER;
+	dest->counter_id = mlx5_fc_id(rx_tables->check_rule_counter);
+	rule = mlx5_add_flow_rules(rx_tables->ft_check, spec, flow_act, dest, 1);
+	if (IS_ERR(rule)) {
+		err = PTR_ERR(rule);
+		netdev_err(netdev, "Failed to add MACsec Rx check rule, err=%d\n", err);
+		return err;
+	}
+
+	rx_fs->check_rule[rule_index] = rule;
+
+	return 0;
+}
+
+static int macsec_fs_rx_create(struct mlx5e_macsec_fs *macsec_fs)
+{
+	int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
+	struct mlx5e_macsec_rx *rx_fs = macsec_fs->rx_fs;
+	struct net_device *netdev = macsec_fs->netdev;
+	struct mlx5_flow_table_attr ft_attr = {};
+	struct mlx5_flow_destination dest = {};
+	struct mlx5e_macsec_tables *rx_tables;
+	struct mlx5e_flow_table *ft_crypto;
+	struct mlx5_flow_table *flow_table;
+	struct mlx5_flow_group *flow_group;
+	struct mlx5_flow_act flow_act = {};
+	struct mlx5_flow_namespace *ns;
+	struct mlx5_flow_handle *rule;
+	struct mlx5_flow_spec *spec;
+	u32 *flow_group_in;
+	int err;
+
+	ns = mlx5_get_flow_namespace(macsec_fs->mdev, MLX5_FLOW_NAMESPACE_KERNEL_RX_MACSEC);
+	if (!ns)
+		return -ENOMEM;
+
+	spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
+	if (!spec)
+		return -ENOMEM;
+
+	flow_group_in = kvzalloc(inlen, GFP_KERNEL);
+	if (!flow_group_in) {
+		err = -ENOMEM;
+		goto free_spec;
+	}
+
+	rx_tables = &rx_fs->tables;
+	ft_crypto = &rx_tables->ft_crypto;
+
+	/* Rx crypto table */
+	ft_attr.level = RX_CRYPTO_TABLE_LEVEL;
+	ft_attr.max_fte = CRYPTO_NUM_MAXSEC_FTE;
+
+	flow_table = mlx5_create_flow_table(ns, &ft_attr);
+	if (IS_ERR(flow_table)) {
+		err = PTR_ERR(flow_table);
+		netdev_err(netdev, "Failed to create MACsec Rx crypto table err(%d)\n", err);
+		goto out_flow_group;
+	}
+	ft_crypto->t = flow_table;
+
+	/* Rx crypto table groups */
+	err = macsec_fs_rx_create_crypto_table_groups(ft_crypto);
+	if (err) {
+		netdev_err(netdev,
+			   "Failed to create default flow group for MACsec Tx crypto table err(%d)\n",
+			   err);
+		goto err;
+	}
+
+	flow_act.action = MLX5_FLOW_CONTEXT_ACTION_FWD_NEXT_PRIO;
+	rule = mlx5_add_flow_rules(ft_crypto->t, NULL, &flow_act, NULL, 0);
+	if (IS_ERR(rule)) {
+		err = PTR_ERR(rule);
+		netdev_err(netdev,
+			   "Failed to add MACsec Rx crypto table default miss rule %d\n",
+			   err);
+		goto err;
+	}
+	rx_tables->crypto_miss_rule = rule;
+
+	/* Rx check table */
+	flow_table = macsec_fs_auto_group_table_create(ns,
+						       MLX5_FLOW_TABLE_TUNNEL_EN_REFORMAT,
+						       RX_CHECK_TABLE_LEVEL,
+						       RX_CHECK_TABLE_NUM_FTE);
+	if (IS_ERR(flow_table)) {
+		err = PTR_ERR(flow_table);
+		netdev_err(netdev, "fail to create MACsec RX check table, err(%d)\n", err);
+		goto err;
+	}
+	rx_tables->ft_check = flow_table;
+
+	/* Rx check table Default miss group/rule */
+	MLX5_SET(create_flow_group_in, flow_group_in, start_flow_index, flow_table->max_fte - 1);
+	MLX5_SET(create_flow_group_in, flow_group_in, end_flow_index, flow_table->max_fte - 1);
+	flow_group = mlx5_create_flow_group(rx_tables->ft_check, flow_group_in);
+	if (IS_ERR(flow_group)) {
+		err = PTR_ERR(flow_group);
+		netdev_err(netdev,
+			   "Failed to create default flow group for MACsec Rx check table err(%d)\n",
+			   err);
+		goto err;
+	}
+	rx_tables->ft_check_group = flow_group;
+
+	/* Rx check table default drop rule */
+	memset(&flow_act, 0, sizeof(flow_act));
+
+	dest.type = MLX5_FLOW_DESTINATION_TYPE_COUNTER;
+	dest.counter_id = mlx5_fc_id(rx_tables->check_miss_rule_counter);
+	flow_act.action = MLX5_FLOW_CONTEXT_ACTION_DROP | MLX5_FLOW_CONTEXT_ACTION_COUNT;
+	rule = mlx5_add_flow_rules(rx_tables->ft_check,  NULL, &flow_act, &dest, 1);
+	if (IS_ERR(rule)) {
+		err = PTR_ERR(rule);
+		netdev_err(netdev, "Failed to added MACsec Rx check drop rule, err(%d)\n", err);
+		goto err;
+	}
+	rx_tables->check_miss_rule = rule;
+
+	/* Rx check table decap rules */
+	err = macsec_fs_rx_create_check_decap_rule(macsec_fs, &dest, &flow_act, spec,
+						   MLX5_SECTAG_HEADER_SIZE_WITH_SCI);
+	if (err)
+		goto err;
+
+	err = macsec_fs_rx_create_check_decap_rule(macsec_fs, &dest, &flow_act, spec,
+						   MLX5_SECTAG_HEADER_SIZE_WITHOUT_SCI);
+	if (err)
+		goto err;
+
+	goto out_flow_group;
+
+err:
+	macsec_fs_rx_destroy(macsec_fs);
+out_flow_group:
+	kvfree(flow_group_in);
+free_spec:
+	kvfree(spec);
+	return err;
+}
+
+static int macsec_fs_rx_ft_get(struct mlx5e_macsec_fs *macsec_fs)
+{
+	struct mlx5e_macsec_tables *rx_tables = &macsec_fs->rx_fs->tables;
+	int err = 0;
+
+	if (rx_tables->refcnt)
+		goto out;
+
+	err = macsec_fs_rx_create(macsec_fs);
+	if (err)
+		return err;
+
+out:
+	rx_tables->refcnt++;
+	return err;
+}
+
+static void macsec_fs_rx_ft_put(struct mlx5e_macsec_fs *macsec_fs)
+{
+	struct mlx5e_macsec_tables *rx_tables = &macsec_fs->rx_fs->tables;
+
+	if (--rx_tables->refcnt)
+		return;
+
+	macsec_fs_rx_destroy(macsec_fs);
+}
+
+static void macsec_fs_rx_del_rule(struct mlx5e_macsec_fs *macsec_fs,
+				  struct mlx5e_macsec_rx_rule *rx_rule)
+{
+	int i;
+
+	for (i = 0; i < RX_NUM_OF_RULES_PER_SA; ++i) {
+		if (rx_rule->rule[i]) {
+			mlx5_del_flow_rules(rx_rule->rule[i]);
+			rx_rule->rule[i] = NULL;
+		}
+	}
+
+	if (rx_rule->meta_modhdr) {
+		mlx5_modify_header_dealloc(macsec_fs->mdev, rx_rule->meta_modhdr);
+		rx_rule->meta_modhdr = NULL;
+	}
+
+	kfree(rx_rule);
+
+	macsec_fs_rx_ft_put(macsec_fs);
+}
+
+static void macsec_fs_rx_setup_fte(struct mlx5_flow_spec *spec,
+				   struct mlx5_flow_act *flow_act,
+				   struct mlx5_macsec_rule_attrs *attrs,
+				   bool sci_present)
+{
+	u8 tci_an = (sci_present << MLX5_MACSEC_SECTAG_TCI_SC_FIELD_OFFSET) | attrs->assoc_num;
+	struct mlx5_flow_act_crypto_params *crypto_params = &flow_act->crypto;
+	__be32 *sci_p = (__be32 *)(&attrs->sci);
+
+	spec->match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
+
+	/* MACsec ethertype */
+	MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria, outer_headers.ethertype);
+	MLX5_SET(fte_match_param, spec->match_value, outer_headers.ethertype, ETH_P_MACSEC);
+
+	spec->match_criteria_enable |= MLX5_MATCH_MISC_PARAMETERS_5;
+
+	/* Sectag AN + TCI SC present bit*/
+	MLX5_SET(fte_match_param, spec->match_criteria, misc_parameters_5.macsec_tag_0,
+		 MLX5_MACSEC_SECTAG_TCI_AN_FIELD_BITMASK << MLX5_MACSEC_SECTAG_TCI_AN_FIELD_OFFSET);
+	MLX5_SET(fte_match_param, spec->match_value, misc_parameters_5.macsec_tag_0,
+		 tci_an << MLX5_MACSEC_SECTAG_TCI_AN_FIELD_OFFSET);
+
+	if (sci_present) {
+		MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria,
+				 misc_parameters_5.macsec_tag_2);
+		MLX5_SET(fte_match_param, spec->match_value, misc_parameters_5.macsec_tag_2,
+			 be32_to_cpu(sci_p[0]));
+
+		MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria,
+				 misc_parameters_5.macsec_tag_3);
+		MLX5_SET(fte_match_param, spec->match_value, misc_parameters_5.macsec_tag_3,
+			 be32_to_cpu(sci_p[1]));
+	} else {
+		/* When SCI isn't present in the Sectag, need to match the source */
+		/* MAC address only if the SCI contains the default MACsec PORT	  */
+		MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria, outer_headers.smac_47_16);
+		MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria, outer_headers.smac_15_0);
+		memcpy(MLX5_ADDR_OF(fte_match_param, spec->match_value, outer_headers.smac_47_16),
+		       sci_p, ETH_ALEN);
+	}
+
+	crypto_params->type = MLX5_FLOW_CONTEXT_ENCRYPT_DECRYPT_TYPE_MACSEC;
+	crypto_params->obj_id = attrs->macsec_obj_id;
+}
+
+static union mlx5e_macsec_rule *
+macsec_fs_rx_add_rule(struct mlx5e_macsec_fs *macsec_fs,
+		      const struct macsec_context *macsec_ctx,
+		      struct mlx5_macsec_rule_attrs *attrs,
+		      u32 fs_id)
+{
+	u8 action[MLX5_UN_SZ_BYTES(set_add_copy_action_in_auto)] = {};
+	struct mlx5e_macsec_rx *rx_fs = macsec_fs->rx_fs;
+	struct net_device *netdev = macsec_fs->netdev;
+	union mlx5e_macsec_rule *macsec_rule = NULL;
+	struct mlx5_modify_hdr *modify_hdr = NULL;
+	struct mlx5_flow_destination dest = {};
+	struct mlx5e_macsec_tables *rx_tables;
+	struct mlx5e_macsec_rx_rule *rx_rule;
+	struct mlx5_flow_act flow_act = {};
+	struct mlx5e_flow_table *ft_crypto;
+	struct mlx5_flow_handle *rule;
+	struct mlx5_flow_spec *spec;
+	int err = 0;
+
+	spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
+	if (!spec)
+		return NULL;
+
+	err = macsec_fs_rx_ft_get(macsec_fs);
+	if (err)
+		goto out_spec;
+
+	macsec_rule = kzalloc(sizeof(*macsec_rule), GFP_KERNEL);
+	if (!macsec_rule) {
+		macsec_fs_rx_ft_put(macsec_fs);
+		goto out_spec;
+	}
+
+	rx_rule = &macsec_rule->rx_rule;
+	rx_tables = &rx_fs->tables;
+	ft_crypto = &rx_tables->ft_crypto;
+
+	/* Set bit[31 - 30] macsec marker - 0x01 */
+	/* Set bit[15-0] fs id */
+	MLX5_SET(set_action_in, action, action_type, MLX5_ACTION_TYPE_SET);
+	MLX5_SET(set_action_in, action, field, MLX5_ACTION_IN_FIELD_METADATA_REG_B);
+	MLX5_SET(set_action_in, action, data, MLX5_MACSEC_RX_METADAT_HANDLE(fs_id) | BIT(30));
+	MLX5_SET(set_action_in, action, offset, 0);
+	MLX5_SET(set_action_in, action, length, 32);
+
+	modify_hdr = mlx5_modify_header_alloc(macsec_fs->mdev, MLX5_FLOW_NAMESPACE_KERNEL_RX_MACSEC,
+					      1, action);
+	if (IS_ERR(modify_hdr)) {
+		err = PTR_ERR(modify_hdr);
+		netdev_err(netdev, "fail to alloc MACsec set modify_header_id err=%d\n", err);
+		modify_hdr = NULL;
+		goto err;
+	}
+	rx_rule->meta_modhdr = modify_hdr;
+
+	/* Rx crypto table with SCI rule */
+	macsec_fs_rx_setup_fte(spec, &flow_act, attrs, true);
+
+	flow_act.modify_hdr = modify_hdr;
+	flow_act.action = MLX5_FLOW_CONTEXT_ACTION_FWD_DEST |
+			  MLX5_FLOW_CONTEXT_ACTION_CRYPTO_DECRYPT |
+			  MLX5_FLOW_CONTEXT_ACTION_MOD_HDR;
+
+	dest.type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
+	dest.ft = rx_tables->ft_check;
+	rule = mlx5_add_flow_rules(ft_crypto->t, spec, &flow_act, &dest, 1);
+	if (IS_ERR(rule)) {
+		err = PTR_ERR(rule);
+		netdev_err(netdev,
+			   "Failed to add SA with SCI rule to Rx crypto rule, err=%d\n",
+			   err);
+		goto err;
+	}
+	rx_rule->rule[0] = rule;
+
+	/* Rx crypto table without SCI rule */
+	if ((cpu_to_be64((__force u64)attrs->sci) & 0xFFFF) == ntohs(MACSEC_PORT_ES)) {
+		memset(spec, 0, sizeof(struct mlx5_flow_spec));
+		memset(&dest, 0, sizeof(struct mlx5_flow_destination));
+		memset(&flow_act, 0, sizeof(flow_act));
+
+		macsec_fs_rx_setup_fte(spec, &flow_act, attrs, false);
+
+		flow_act.modify_hdr = modify_hdr;
+		flow_act.action = MLX5_FLOW_CONTEXT_ACTION_FWD_DEST |
+				  MLX5_FLOW_CONTEXT_ACTION_CRYPTO_DECRYPT |
+				  MLX5_FLOW_CONTEXT_ACTION_MOD_HDR;
+
+		dest.type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
+		dest.ft = rx_tables->ft_check;
+		rule = mlx5_add_flow_rules(ft_crypto->t, spec, &flow_act, &dest, 1);
+		if (IS_ERR(rule)) {
+			err = PTR_ERR(rule);
+			netdev_err(netdev,
+				   "Failed to add SA without SCI rule to Rx crypto rule, err=%d\n",
+				   err);
+			goto err;
+		}
+		rx_rule->rule[1] = rule;
+	}
+
+	kvfree(spec);
+	return macsec_rule;
+
+err:
+	macsec_fs_rx_del_rule(macsec_fs, rx_rule);
+	macsec_rule = NULL;
+out_spec:
+	kvfree(spec);
+	return macsec_rule;
+}
+
+static int macsec_fs_rx_init(struct mlx5e_macsec_fs *macsec_fs)
+{
+	struct net_device *netdev = macsec_fs->netdev;
+	struct mlx5_core_dev *mdev = macsec_fs->mdev;
+	struct mlx5e_macsec_tables *rx_tables;
+	struct mlx5e_macsec_rx *rx_fs;
+	struct mlx5_fc *flow_counter;
+	int err;
+
+	rx_fs =	kzalloc(sizeof(*rx_fs), GFP_KERNEL);
+	if (!rx_fs)
+		return -ENOMEM;
+
+	flow_counter = mlx5_fc_create(mdev, false);
+	if (IS_ERR(flow_counter)) {
+		err = PTR_ERR(flow_counter);
+		netdev_err(netdev,
+			   "Failed to create MACsec Rx encrypt flow counter, err(%d)\n",
+			   err);
+		goto err_encrypt_counter;
+	}
+
+	rx_tables = &rx_fs->tables;
+	rx_tables->check_rule_counter = flow_counter;
+
+	flow_counter = mlx5_fc_create(mdev, false);
+	if (IS_ERR(flow_counter)) {
+		err = PTR_ERR(flow_counter);
+		netdev_err(netdev,
+			   "Failed to create MACsec Rx drop flow counter, err(%d)\n",
+			   err);
+		goto err_drop_counter;
+	}
+	rx_tables->check_miss_rule_counter = flow_counter;
+
+	macsec_fs->rx_fs = rx_fs;
+
+	return 0;
+
+err_drop_counter:
+	mlx5_fc_destroy(mdev, rx_tables->check_rule_counter);
+	rx_tables->check_rule_counter = NULL;
+
+err_encrypt_counter:
+	kfree(rx_fs);
+	macsec_fs->rx_fs = NULL;
+
+	return err;
+}
+
+static void macsec_fs_rx_cleanup(struct mlx5e_macsec_fs *macsec_fs)
+{
+	struct mlx5e_macsec_rx *rx_fs = macsec_fs->rx_fs;
+	struct mlx5_core_dev *mdev = macsec_fs->mdev;
+	struct mlx5e_macsec_tables *rx_tables;
+
+	if (!rx_fs)
+		return;
+
+	rx_tables = &rx_fs->tables;
+
+	if (rx_tables->refcnt) {
+		netdev_err(macsec_fs->netdev,
+			   "Can't destroy MACsec offload rx_fs, refcnt(%u) isn't 0\n",
+			   rx_tables->refcnt);
+		return;
+	}
+
+	if (rx_tables->check_miss_rule_counter) {
+		mlx5_fc_destroy(mdev, rx_tables->check_miss_rule_counter);
+		rx_tables->check_miss_rule_counter = NULL;
+	}
+
+	if (rx_tables->check_rule_counter) {
+		mlx5_fc_destroy(mdev, rx_tables->check_rule_counter);
+		rx_tables->check_rule_counter = NULL;
+	}
+
+	kfree(rx_fs);
+	macsec_fs->rx_fs = NULL;
+}
+
+void mlx5e_macsec_fs_get_stats_fill(struct mlx5e_macsec_fs *macsec_fs, void *macsec_stats)
+{
+	struct mlx5e_macsec_stats *stats = (struct mlx5e_macsec_stats *)macsec_stats;
+	struct mlx5e_macsec_tables *tx_tables = &macsec_fs->tx_fs->tables;
+	struct mlx5e_macsec_tables *rx_tables = &macsec_fs->rx_fs->tables;
+	struct mlx5_core_dev *mdev = macsec_fs->mdev;
+
+	if (tx_tables->check_rule_counter)
+		mlx5_fc_query(mdev, tx_tables->check_rule_counter,
+			      &stats->macsec_tx_pkts, &stats->macsec_tx_bytes);
+
+	if (tx_tables->check_miss_rule_counter)
+		mlx5_fc_query(mdev, tx_tables->check_miss_rule_counter,
+			      &stats->macsec_tx_pkts_drop, &stats->macsec_tx_bytes_drop);
+
+	if (rx_tables->check_rule_counter)
+		mlx5_fc_query(mdev, rx_tables->check_rule_counter,
+			      &stats->macsec_rx_pkts, &stats->macsec_rx_bytes);
+
+	if (rx_tables->check_miss_rule_counter)
+		mlx5_fc_query(mdev, rx_tables->check_miss_rule_counter,
+			      &stats->macsec_rx_pkts_drop, &stats->macsec_rx_bytes_drop);
+}
+
+union mlx5e_macsec_rule *
+mlx5e_macsec_fs_add_rule(struct mlx5e_macsec_fs *macsec_fs,
+			 const struct macsec_context *macsec_ctx,
+			 struct mlx5_macsec_rule_attrs *attrs,
+			 u32 *sa_fs_id)
+{
+	return (attrs->action == MLX5_ACCEL_MACSEC_ACTION_ENCRYPT) ?
+		macsec_fs_tx_add_rule(macsec_fs, macsec_ctx, attrs, sa_fs_id) :
+		macsec_fs_rx_add_rule(macsec_fs, macsec_ctx, attrs, *sa_fs_id);
+}
+
+void mlx5e_macsec_fs_del_rule(struct mlx5e_macsec_fs *macsec_fs,
+			      union mlx5e_macsec_rule *macsec_rule,
+			      int action)
+{
+	(action == MLX5_ACCEL_MACSEC_ACTION_ENCRYPT) ?
+		macsec_fs_tx_del_rule(macsec_fs, &macsec_rule->tx_rule) :
+		macsec_fs_rx_del_rule(macsec_fs, &macsec_rule->rx_rule);
+}
+
+void mlx5e_macsec_fs_cleanup(struct mlx5e_macsec_fs *macsec_fs)
+{
+	macsec_fs_rx_cleanup(macsec_fs);
+	macsec_fs_tx_cleanup(macsec_fs);
+	kfree(macsec_fs);
+}
+
+struct mlx5e_macsec_fs *
+mlx5e_macsec_fs_init(struct mlx5_core_dev *mdev,
+		     struct net_device *netdev)
+{
+	struct mlx5e_macsec_fs *macsec_fs;
+	int err;
+
+	macsec_fs = kzalloc(sizeof(*macsec_fs), GFP_KERNEL);
+	if (!macsec_fs)
+		return NULL;
+
+	macsec_fs->mdev = mdev;
+	macsec_fs->netdev = netdev;
+
+	err = macsec_fs_tx_init(macsec_fs);
+	if (err) {
+		netdev_err(netdev, "MACsec offload: Failed to init tx_fs, err=%d\n", err);
+		goto err;
+	}
+
+	err = macsec_fs_rx_init(macsec_fs);
+	if (err) {
+		netdev_err(netdev, "MACsec offload: Failed to init tx_fs, err=%d\n", err);
+		goto tx_cleanup;
+	}
+
+	return macsec_fs;
+
+tx_cleanup:
+	macsec_fs_tx_cleanup(macsec_fs);
+err:
+	kfree(macsec_fs);
+	return NULL;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec_fs.h b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec_fs.h
new file mode 100644
index 000000000..b429648d4
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec_fs.h
@@ -0,0 +1,47 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2022, NVIDIA CORPORATION & AFFILIATES. All rights reserved. */
+
+#ifndef __MLX5_MACSEC_STEERING_H__
+#define __MLX5_MACSEC_STEERING_H__
+
+#ifdef CONFIG_MLX5_EN_MACSEC
+
+#include "en_accel/macsec.h"
+
+#define MLX5_MACSEC_NUM_OF_SUPPORTED_INTERFACES 16
+
+struct mlx5e_macsec_fs;
+union mlx5e_macsec_rule;
+
+struct mlx5_macsec_rule_attrs {
+	sci_t sci;
+	u32 macsec_obj_id;
+	u8 assoc_num;
+	int action;
+};
+
+enum mlx5_macsec_action {
+	MLX5_ACCEL_MACSEC_ACTION_ENCRYPT,
+	MLX5_ACCEL_MACSEC_ACTION_DECRYPT,
+};
+
+void mlx5e_macsec_fs_cleanup(struct mlx5e_macsec_fs *macsec_fs);
+
+struct mlx5e_macsec_fs *
+mlx5e_macsec_fs_init(struct mlx5_core_dev *mdev, struct net_device *netdev);
+
+union mlx5e_macsec_rule *
+mlx5e_macsec_fs_add_rule(struct mlx5e_macsec_fs *macsec_fs,
+			 const struct macsec_context *ctx,
+			 struct mlx5_macsec_rule_attrs *attrs,
+			 u32 *sa_fs_id);
+
+void mlx5e_macsec_fs_del_rule(struct mlx5e_macsec_fs *macsec_fs,
+			      union mlx5e_macsec_rule *macsec_rule,
+			      int action);
+
+void mlx5e_macsec_fs_get_stats_fill(struct mlx5e_macsec_fs *macsec_fs, void *macsec_stats);
+
+#endif
+
+#endif /* __MLX5_MACSEC_STEERING_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec_stats.c b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec_stats.c
new file mode 100644
index 000000000..e50a2e3f3
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec_stats.c
@@ -0,0 +1,72 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2022, NVIDIA CORPORATION & AFFILIATES. All rights reserved. */
+
+#include <linux/ethtool.h>
+#include <net/sock.h>
+
+#include "en.h"
+#include "en_accel/macsec.h"
+
+static const struct counter_desc mlx5e_macsec_hw_stats_desc[] = {
+	{ MLX5E_DECLARE_STAT(struct mlx5e_macsec_stats, macsec_rx_pkts) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_macsec_stats, macsec_rx_bytes) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_macsec_stats, macsec_rx_pkts_drop) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_macsec_stats, macsec_rx_bytes_drop) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_macsec_stats, macsec_tx_pkts) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_macsec_stats, macsec_tx_bytes) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_macsec_stats, macsec_tx_pkts_drop) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_macsec_stats, macsec_tx_bytes_drop) },
+};
+
+#define NUM_MACSEC_HW_COUNTERS ARRAY_SIZE(mlx5e_macsec_hw_stats_desc)
+
+static MLX5E_DECLARE_STATS_GRP_OP_NUM_STATS(macsec_hw)
+{
+	if (!priv->macsec)
+		return 0;
+
+	if (mlx5e_is_macsec_device(priv->mdev))
+		return NUM_MACSEC_HW_COUNTERS;
+
+	return 0;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_UPDATE_STATS(macsec_hw) {}
+
+static MLX5E_DECLARE_STATS_GRP_OP_FILL_STRS(macsec_hw)
+{
+	unsigned int i;
+
+	if (!priv->macsec)
+		return idx;
+
+	if (!mlx5e_is_macsec_device(priv->mdev))
+		return idx;
+
+	for (i = 0; i < NUM_MACSEC_HW_COUNTERS; i++)
+		strcpy(data + (idx++) * ETH_GSTRING_LEN,
+		       mlx5e_macsec_hw_stats_desc[i].format);
+
+	return idx;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_FILL_STATS(macsec_hw)
+{
+	int i;
+
+	if (!priv->macsec)
+		return idx;
+
+	if (!mlx5e_is_macsec_device(priv->mdev))
+		return idx;
+
+	mlx5e_macsec_get_stats_fill(priv->macsec, mlx5e_macsec_get_stats(priv->macsec));
+	for (i = 0; i < NUM_MACSEC_HW_COUNTERS; i++)
+		data[idx++] = MLX5E_READ_CTR64_CPU(mlx5e_macsec_get_stats(priv->macsec),
+						   mlx5e_macsec_hw_stats_desc,
+						   i);
+
+	return idx;
+}
+
+MLX5E_DEFINE_STATS_GRP(macsec_hw, 0);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_arfs.c b/drivers/net/ethernet/mellanox/mlx5/core/en_arfs.c
new file mode 100644
index 000000000..58eacba6d
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_arfs.c
@@ -0,0 +1,765 @@
+/*
+ * Copyright (c) 2016, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/hash.h>
+#include <linux/mlx5/fs.h>
+#include <linux/ip.h>
+#include <linux/ipv6.h>
+#include "en.h"
+
+#define ARFS_HASH_SHIFT BITS_PER_BYTE
+#define ARFS_HASH_SIZE BIT(BITS_PER_BYTE)
+
+struct arfs_table {
+	struct mlx5e_flow_table  ft;
+	struct mlx5_flow_handle	 *default_rule;
+	struct hlist_head	 rules_hash[ARFS_HASH_SIZE];
+};
+
+enum arfs_type {
+	ARFS_IPV4_TCP,
+	ARFS_IPV6_TCP,
+	ARFS_IPV4_UDP,
+	ARFS_IPV6_UDP,
+	ARFS_NUM_TYPES,
+};
+
+struct mlx5e_arfs_tables {
+	struct arfs_table arfs_tables[ARFS_NUM_TYPES];
+	/* Protect aRFS rules list */
+	spinlock_t                     arfs_lock;
+	struct list_head               rules;
+	int                            last_filter_id;
+	struct workqueue_struct        *wq;
+};
+
+struct arfs_tuple {
+	__be16 etype;
+	u8     ip_proto;
+	union {
+		__be32 src_ipv4;
+		struct in6_addr src_ipv6;
+	};
+	union {
+		__be32 dst_ipv4;
+		struct in6_addr dst_ipv6;
+	};
+	__be16 src_port;
+	__be16 dst_port;
+};
+
+struct arfs_rule {
+	struct mlx5e_priv	*priv;
+	struct work_struct      arfs_work;
+	struct mlx5_flow_handle *rule;
+	struct hlist_node	hlist;
+	int			rxq;
+	/* Flow ID passed to ndo_rx_flow_steer */
+	int			flow_id;
+	/* Filter ID returned by ndo_rx_flow_steer */
+	int			filter_id;
+	struct arfs_tuple	tuple;
+};
+
+#define mlx5e_for_each_arfs_rule(hn, tmp, arfs_tables, i, j) \
+	for (i = 0; i < ARFS_NUM_TYPES; i++) \
+		mlx5e_for_each_hash_arfs_rule(hn, tmp, arfs_tables[i].rules_hash, j)
+
+#define mlx5e_for_each_hash_arfs_rule(hn, tmp, hash, j) \
+	for (j = 0; j < ARFS_HASH_SIZE; j++) \
+		hlist_for_each_entry_safe(hn, tmp, &hash[j], hlist)
+
+static enum mlx5_traffic_types arfs_get_tt(enum arfs_type type)
+{
+	switch (type) {
+	case ARFS_IPV4_TCP:
+		return MLX5_TT_IPV4_TCP;
+	case ARFS_IPV4_UDP:
+		return MLX5_TT_IPV4_UDP;
+	case ARFS_IPV6_TCP:
+		return MLX5_TT_IPV6_TCP;
+	case ARFS_IPV6_UDP:
+		return MLX5_TT_IPV6_UDP;
+	default:
+		return -EINVAL;
+	}
+}
+
+static int arfs_disable(struct mlx5e_flow_steering *fs)
+{
+	struct mlx5_ttc_table *ttc = mlx5e_fs_get_ttc(fs, false);
+	int err, i;
+
+	for (i = 0; i < ARFS_NUM_TYPES; i++) {
+		/* Modify ttc rules destination back to their default */
+		err = mlx5_ttc_fwd_default_dest(ttc, arfs_get_tt(i));
+		if (err) {
+			fs_err(fs,
+			       "%s: modify ttc[%d] default destination failed, err(%d)\n",
+			       __func__, arfs_get_tt(i), err);
+			return err;
+		}
+	}
+	return 0;
+}
+
+static void arfs_del_rules(struct mlx5e_flow_steering *fs);
+
+int mlx5e_arfs_disable(struct mlx5e_flow_steering *fs)
+{
+	/* Moving to switchdev mode, fs->arfs is freed by mlx5e_nic_profile
+	 * cleanup_rx callback and it is not recreated when
+	 * mlx5e_uplink_rep_profile is loaded as mlx5e_create_flow_steering()
+	 * is not called by the uplink_rep profile init_rx callback. Thus, if
+	 * ntuple is set, moving to switchdev flow will enter this function
+	 * with fs->arfs nullified.
+	 */
+	if (!mlx5e_fs_get_arfs(fs))
+		return 0;
+
+	arfs_del_rules(fs);
+
+	return arfs_disable(fs);
+}
+
+int mlx5e_arfs_enable(struct mlx5e_flow_steering *fs)
+{
+	struct mlx5_ttc_table *ttc = mlx5e_fs_get_ttc(fs, false);
+	struct mlx5e_arfs_tables *arfs =  mlx5e_fs_get_arfs(fs);
+	struct mlx5_flow_destination dest = {};
+	int err, i;
+
+	dest.type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
+	for (i = 0; i < ARFS_NUM_TYPES; i++) {
+		dest.ft = arfs->arfs_tables[i].ft.t;
+		/* Modify ttc rules destination to point on the aRFS FTs */
+		err = mlx5_ttc_fwd_dest(ttc, arfs_get_tt(i), &dest);
+		if (err) {
+			fs_err(fs, "%s: modify ttc[%d] dest to arfs, failed err(%d)\n",
+			       __func__, arfs_get_tt(i), err);
+			arfs_disable(fs);
+			return err;
+		}
+	}
+	return 0;
+}
+
+static void arfs_destroy_table(struct arfs_table *arfs_t)
+{
+	mlx5_del_flow_rules(arfs_t->default_rule);
+	mlx5e_destroy_flow_table(&arfs_t->ft);
+}
+
+static void _mlx5e_cleanup_tables(struct mlx5e_flow_steering *fs)
+{
+	struct mlx5e_arfs_tables *arfs =  mlx5e_fs_get_arfs(fs);
+	int i;
+
+	arfs_del_rules(fs);
+	destroy_workqueue(arfs->wq);
+	for (i = 0; i < ARFS_NUM_TYPES; i++) {
+		if (!IS_ERR_OR_NULL(arfs->arfs_tables[i].ft.t))
+			arfs_destroy_table(&arfs->arfs_tables[i]);
+	}
+}
+
+void mlx5e_arfs_destroy_tables(struct mlx5e_flow_steering *fs, bool ntuple)
+{
+	struct mlx5e_arfs_tables *arfs =  mlx5e_fs_get_arfs(fs);
+
+	if (!ntuple)
+		return;
+
+	_mlx5e_cleanup_tables(fs);
+	mlx5e_fs_set_arfs(fs, NULL);
+	kvfree(arfs);
+}
+
+static int arfs_add_default_rule(struct mlx5e_flow_steering *fs,
+				 struct mlx5e_rx_res *rx_res,
+				 enum arfs_type type)
+{
+	struct mlx5e_arfs_tables *arfs =  mlx5e_fs_get_arfs(fs);
+	struct arfs_table *arfs_t = &arfs->arfs_tables[type];
+	struct mlx5_flow_destination dest = {};
+	MLX5_DECLARE_FLOW_ACT(flow_act);
+	enum mlx5_traffic_types tt;
+	int err = 0;
+
+	dest.type = MLX5_FLOW_DESTINATION_TYPE_TIR;
+	tt = arfs_get_tt(type);
+	if (tt == -EINVAL) {
+		fs_err(fs, "%s: bad arfs_type: %d\n", __func__, type);
+		return -EINVAL;
+	}
+
+	/* FIXME: Must use mlx5_ttc_get_default_dest(),
+	 * but can't since TTC default is not setup yet !
+	 */
+	dest.tir_num = mlx5e_rx_res_get_tirn_rss(rx_res, tt);
+	arfs_t->default_rule = mlx5_add_flow_rules(arfs_t->ft.t, NULL,
+						   &flow_act,
+						   &dest, 1);
+	if (IS_ERR(arfs_t->default_rule)) {
+		err = PTR_ERR(arfs_t->default_rule);
+		arfs_t->default_rule = NULL;
+		fs_err(fs, "%s: add rule failed, arfs type=%d\n", __func__, type);
+	}
+
+	return err;
+}
+
+#define MLX5E_ARFS_NUM_GROUPS	2
+#define MLX5E_ARFS_GROUP1_SIZE	(BIT(16) - 1)
+#define MLX5E_ARFS_GROUP2_SIZE	BIT(0)
+#define MLX5E_ARFS_TABLE_SIZE	(MLX5E_ARFS_GROUP1_SIZE +\
+				 MLX5E_ARFS_GROUP2_SIZE)
+static int arfs_create_groups(struct mlx5e_flow_table *ft,
+			      enum  arfs_type type)
+{
+	int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
+	void *outer_headers_c;
+	int ix = 0;
+	u32 *in;
+	int err;
+	u8 *mc;
+
+	ft->g = kcalloc(MLX5E_ARFS_NUM_GROUPS,
+			sizeof(*ft->g), GFP_KERNEL);
+	if (!ft->g)
+		return -ENOMEM;
+
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in) {
+		err = -ENOMEM;
+		goto err_free_g;
+	}
+
+	mc = MLX5_ADDR_OF(create_flow_group_in, in, match_criteria);
+	outer_headers_c = MLX5_ADDR_OF(fte_match_param, mc,
+				       outer_headers);
+	MLX5_SET_TO_ONES(fte_match_set_lyr_2_4, outer_headers_c, ethertype);
+	switch (type) {
+	case ARFS_IPV4_TCP:
+	case ARFS_IPV6_TCP:
+		MLX5_SET_TO_ONES(fte_match_set_lyr_2_4, outer_headers_c, tcp_dport);
+		MLX5_SET_TO_ONES(fte_match_set_lyr_2_4, outer_headers_c, tcp_sport);
+		break;
+	case ARFS_IPV4_UDP:
+	case ARFS_IPV6_UDP:
+		MLX5_SET_TO_ONES(fte_match_set_lyr_2_4, outer_headers_c, udp_dport);
+		MLX5_SET_TO_ONES(fte_match_set_lyr_2_4, outer_headers_c, udp_sport);
+		break;
+	default:
+		err = -EINVAL;
+		goto err_free_in;
+	}
+
+	switch (type) {
+	case ARFS_IPV4_TCP:
+	case ARFS_IPV4_UDP:
+		MLX5_SET_TO_ONES(fte_match_set_lyr_2_4, outer_headers_c,
+				 src_ipv4_src_ipv6.ipv4_layout.ipv4);
+		MLX5_SET_TO_ONES(fte_match_set_lyr_2_4, outer_headers_c,
+				 dst_ipv4_dst_ipv6.ipv4_layout.ipv4);
+		break;
+	case ARFS_IPV6_TCP:
+	case ARFS_IPV6_UDP:
+		memset(MLX5_ADDR_OF(fte_match_set_lyr_2_4, outer_headers_c,
+				    src_ipv4_src_ipv6.ipv6_layout.ipv6),
+		       0xff, 16);
+		memset(MLX5_ADDR_OF(fte_match_set_lyr_2_4, outer_headers_c,
+				    dst_ipv4_dst_ipv6.ipv6_layout.ipv6),
+		       0xff, 16);
+		break;
+	default:
+		err = -EINVAL;
+		goto err_free_in;
+	}
+
+	MLX5_SET_CFG(in, match_criteria_enable, MLX5_MATCH_OUTER_HEADERS);
+	MLX5_SET_CFG(in, start_flow_index, ix);
+	ix += MLX5E_ARFS_GROUP1_SIZE;
+	MLX5_SET_CFG(in, end_flow_index, ix - 1);
+	ft->g[ft->num_groups] = mlx5_create_flow_group(ft->t, in);
+	if (IS_ERR(ft->g[ft->num_groups]))
+		goto err_clean_group;
+	ft->num_groups++;
+
+	memset(in, 0, inlen);
+	MLX5_SET_CFG(in, start_flow_index, ix);
+	ix += MLX5E_ARFS_GROUP2_SIZE;
+	MLX5_SET_CFG(in, end_flow_index, ix - 1);
+	ft->g[ft->num_groups] = mlx5_create_flow_group(ft->t, in);
+	if (IS_ERR(ft->g[ft->num_groups]))
+		goto err_clean_group;
+	ft->num_groups++;
+
+	kvfree(in);
+	return 0;
+
+err_clean_group:
+	err = PTR_ERR(ft->g[ft->num_groups]);
+	ft->g[ft->num_groups] = NULL;
+err_free_in:
+	kvfree(in);
+err_free_g:
+	kfree(ft->g);
+	ft->g = NULL;
+	return err;
+}
+
+static int arfs_create_table(struct mlx5e_flow_steering *fs,
+			     struct mlx5e_rx_res *rx_res,
+			     enum arfs_type type)
+{
+	struct mlx5_flow_namespace *ns = mlx5e_fs_get_ns(fs, false);
+	struct mlx5e_arfs_tables *arfs = mlx5e_fs_get_arfs(fs);
+	struct mlx5e_flow_table *ft = &arfs->arfs_tables[type].ft;
+	struct mlx5_flow_table_attr ft_attr = {};
+	int err;
+
+	ft->num_groups = 0;
+
+	ft_attr.max_fte = MLX5E_ARFS_TABLE_SIZE;
+	ft_attr.level = MLX5E_ARFS_FT_LEVEL;
+	ft_attr.prio = MLX5E_NIC_PRIO;
+
+	ft->t = mlx5_create_flow_table(ns, &ft_attr);
+	if (IS_ERR(ft->t)) {
+		err = PTR_ERR(ft->t);
+		ft->t = NULL;
+		return err;
+	}
+
+	err = arfs_create_groups(ft, type);
+	if (err)
+		goto err;
+
+	err = arfs_add_default_rule(fs, rx_res,  type);
+	if (err)
+		goto err;
+
+	return 0;
+err:
+	mlx5e_destroy_flow_table(ft);
+	return err;
+}
+
+int mlx5e_arfs_create_tables(struct mlx5e_flow_steering *fs,
+			     struct mlx5e_rx_res *rx_res, bool ntuple)
+{
+	struct mlx5e_arfs_tables *arfs;
+	int err = -ENOMEM;
+	int i;
+
+	if (!ntuple)
+		return 0;
+
+	arfs = kvzalloc(sizeof(*arfs), GFP_KERNEL);
+	if (!arfs)
+		return -ENOMEM;
+
+	spin_lock_init(&arfs->arfs_lock);
+	INIT_LIST_HEAD(&arfs->rules);
+	arfs->wq = create_singlethread_workqueue("mlx5e_arfs");
+	if (!arfs->wq)
+		goto err;
+
+	mlx5e_fs_set_arfs(fs, arfs);
+
+	for (i = 0; i < ARFS_NUM_TYPES; i++) {
+		err = arfs_create_table(fs, rx_res, i);
+		if (err)
+			goto err_des;
+	}
+	return 0;
+
+err_des:
+	_mlx5e_cleanup_tables(fs);
+err:
+	mlx5e_fs_set_arfs(fs, NULL);
+	kvfree(arfs);
+	return err;
+}
+
+#define MLX5E_ARFS_EXPIRY_QUOTA 60
+
+static void arfs_may_expire_flow(struct mlx5e_priv *priv)
+{
+	struct mlx5e_arfs_tables *arfs = mlx5e_fs_get_arfs(priv->fs);
+	struct arfs_rule *arfs_rule;
+	struct hlist_node *htmp;
+	HLIST_HEAD(del_list);
+	int quota = 0;
+	int i;
+	int j;
+
+	spin_lock_bh(&arfs->arfs_lock);
+	mlx5e_for_each_arfs_rule(arfs_rule, htmp, arfs->arfs_tables, i, j) {
+		if (!work_pending(&arfs_rule->arfs_work) &&
+		    rps_may_expire_flow(priv->netdev,
+					arfs_rule->rxq, arfs_rule->flow_id,
+					arfs_rule->filter_id)) {
+			hlist_del_init(&arfs_rule->hlist);
+			hlist_add_head(&arfs_rule->hlist, &del_list);
+			if (quota++ > MLX5E_ARFS_EXPIRY_QUOTA)
+				break;
+		}
+	}
+	spin_unlock_bh(&arfs->arfs_lock);
+	hlist_for_each_entry_safe(arfs_rule, htmp, &del_list, hlist) {
+		if (arfs_rule->rule)
+			mlx5_del_flow_rules(arfs_rule->rule);
+		hlist_del(&arfs_rule->hlist);
+		kfree(arfs_rule);
+	}
+}
+
+static void arfs_del_rules(struct mlx5e_flow_steering *fs)
+{
+	struct mlx5e_arfs_tables *arfs = mlx5e_fs_get_arfs(fs);
+	struct hlist_node *htmp;
+	struct arfs_rule *rule;
+	HLIST_HEAD(del_list);
+	int i;
+	int j;
+
+	spin_lock_bh(&arfs->arfs_lock);
+	mlx5e_for_each_arfs_rule(rule, htmp, arfs->arfs_tables, i, j) {
+		hlist_del_init(&rule->hlist);
+		hlist_add_head(&rule->hlist, &del_list);
+	}
+	spin_unlock_bh(&arfs->arfs_lock);
+
+	hlist_for_each_entry_safe(rule, htmp, &del_list, hlist) {
+		cancel_work_sync(&rule->arfs_work);
+		if (rule->rule)
+			mlx5_del_flow_rules(rule->rule);
+		hlist_del(&rule->hlist);
+		kfree(rule);
+	}
+}
+
+static struct hlist_head *
+arfs_hash_bucket(struct arfs_table *arfs_t, __be16 src_port,
+		 __be16 dst_port)
+{
+	unsigned long l;
+	int bucket_idx;
+
+	l = (__force unsigned long)src_port |
+	    ((__force unsigned long)dst_port << 2);
+
+	bucket_idx = hash_long(l, ARFS_HASH_SHIFT);
+
+	return &arfs_t->rules_hash[bucket_idx];
+}
+
+static struct arfs_table *arfs_get_table(struct mlx5e_arfs_tables *arfs,
+					 u8 ip_proto, __be16 etype)
+{
+	if (etype == htons(ETH_P_IP) && ip_proto == IPPROTO_TCP)
+		return &arfs->arfs_tables[ARFS_IPV4_TCP];
+	if (etype == htons(ETH_P_IP) && ip_proto == IPPROTO_UDP)
+		return &arfs->arfs_tables[ARFS_IPV4_UDP];
+	if (etype == htons(ETH_P_IPV6) && ip_proto == IPPROTO_TCP)
+		return &arfs->arfs_tables[ARFS_IPV6_TCP];
+	if (etype == htons(ETH_P_IPV6) && ip_proto == IPPROTO_UDP)
+		return &arfs->arfs_tables[ARFS_IPV6_UDP];
+
+	return NULL;
+}
+
+static struct mlx5_flow_handle *arfs_add_rule(struct mlx5e_priv *priv,
+					      struct arfs_rule *arfs_rule)
+{
+	struct mlx5e_arfs_tables *arfs = mlx5e_fs_get_arfs(priv->fs);
+	struct arfs_tuple *tuple = &arfs_rule->tuple;
+	struct mlx5_flow_handle *rule = NULL;
+	struct mlx5_flow_destination dest = {};
+	MLX5_DECLARE_FLOW_ACT(flow_act);
+	struct arfs_table *arfs_table;
+	struct mlx5_flow_spec *spec;
+	struct mlx5_flow_table *ft;
+	int err = 0;
+
+	spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
+	if (!spec) {
+		err = -ENOMEM;
+		goto out;
+	}
+	spec->match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
+	MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria,
+			 outer_headers.ethertype);
+	MLX5_SET(fte_match_param, spec->match_value, outer_headers.ethertype,
+		 ntohs(tuple->etype));
+	arfs_table = arfs_get_table(arfs, tuple->ip_proto, tuple->etype);
+	if (!arfs_table) {
+		err = -EINVAL;
+		goto out;
+	}
+
+	ft = arfs_table->ft.t;
+	if (tuple->ip_proto == IPPROTO_TCP) {
+		MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria,
+				 outer_headers.tcp_dport);
+		MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria,
+				 outer_headers.tcp_sport);
+		MLX5_SET(fte_match_param, spec->match_value, outer_headers.tcp_dport,
+			 ntohs(tuple->dst_port));
+		MLX5_SET(fte_match_param, spec->match_value, outer_headers.tcp_sport,
+			 ntohs(tuple->src_port));
+	} else {
+		MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria,
+				 outer_headers.udp_dport);
+		MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria,
+				 outer_headers.udp_sport);
+		MLX5_SET(fte_match_param, spec->match_value, outer_headers.udp_dport,
+			 ntohs(tuple->dst_port));
+		MLX5_SET(fte_match_param, spec->match_value, outer_headers.udp_sport,
+			 ntohs(tuple->src_port));
+	}
+	if (tuple->etype == htons(ETH_P_IP)) {
+		memcpy(MLX5_ADDR_OF(fte_match_param, spec->match_value,
+				    outer_headers.src_ipv4_src_ipv6.ipv4_layout.ipv4),
+		       &tuple->src_ipv4,
+		       4);
+		memcpy(MLX5_ADDR_OF(fte_match_param, spec->match_value,
+				    outer_headers.dst_ipv4_dst_ipv6.ipv4_layout.ipv4),
+		       &tuple->dst_ipv4,
+		       4);
+		MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria,
+				 outer_headers.src_ipv4_src_ipv6.ipv4_layout.ipv4);
+		MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria,
+				 outer_headers.dst_ipv4_dst_ipv6.ipv4_layout.ipv4);
+	} else {
+		memcpy(MLX5_ADDR_OF(fte_match_param, spec->match_value,
+				    outer_headers.src_ipv4_src_ipv6.ipv6_layout.ipv6),
+		       &tuple->src_ipv6,
+		       16);
+		memcpy(MLX5_ADDR_OF(fte_match_param, spec->match_value,
+				    outer_headers.dst_ipv4_dst_ipv6.ipv6_layout.ipv6),
+		       &tuple->dst_ipv6,
+		       16);
+		memset(MLX5_ADDR_OF(fte_match_param, spec->match_criteria,
+				    outer_headers.src_ipv4_src_ipv6.ipv6_layout.ipv6),
+		       0xff,
+		       16);
+		memset(MLX5_ADDR_OF(fte_match_param, spec->match_criteria,
+				    outer_headers.dst_ipv4_dst_ipv6.ipv6_layout.ipv6),
+		       0xff,
+		       16);
+	}
+	dest.type = MLX5_FLOW_DESTINATION_TYPE_TIR;
+	dest.tir_num = mlx5e_rx_res_get_tirn_direct(priv->rx_res, arfs_rule->rxq);
+	rule = mlx5_add_flow_rules(ft, spec, &flow_act, &dest, 1);
+	if (IS_ERR(rule)) {
+		err = PTR_ERR(rule);
+		priv->channel_stats[arfs_rule->rxq]->rq.arfs_err++;
+		mlx5e_dbg(HW, priv,
+			  "%s: add rule(filter id=%d, rq idx=%d, ip proto=0x%x) failed,err=%d\n",
+			  __func__, arfs_rule->filter_id, arfs_rule->rxq,
+			  tuple->ip_proto, err);
+	}
+
+out:
+	kvfree(spec);
+	return err ? ERR_PTR(err) : rule;
+}
+
+static void arfs_modify_rule_rq(struct mlx5e_priv *priv,
+				struct mlx5_flow_handle *rule, u16 rxq)
+{
+	struct mlx5_flow_destination dst = {};
+	int err = 0;
+
+	dst.type = MLX5_FLOW_DESTINATION_TYPE_TIR;
+	dst.tir_num = mlx5e_rx_res_get_tirn_direct(priv->rx_res, rxq);
+	err =  mlx5_modify_rule_destination(rule, &dst, NULL);
+	if (err)
+		netdev_warn(priv->netdev,
+			    "Failed to modify aRFS rule destination to rq=%d\n", rxq);
+}
+
+static void arfs_handle_work(struct work_struct *work)
+{
+	struct arfs_rule *arfs_rule = container_of(work,
+						   struct arfs_rule,
+						   arfs_work);
+	struct mlx5e_priv *priv = arfs_rule->priv;
+	struct mlx5e_arfs_tables *arfs;
+	struct mlx5_flow_handle *rule;
+
+	arfs = mlx5e_fs_get_arfs(priv->fs);
+	mutex_lock(&priv->state_lock);
+	if (!test_bit(MLX5E_STATE_OPENED, &priv->state)) {
+		spin_lock_bh(&arfs->arfs_lock);
+		hlist_del(&arfs_rule->hlist);
+		spin_unlock_bh(&arfs->arfs_lock);
+
+		mutex_unlock(&priv->state_lock);
+		kfree(arfs_rule);
+		goto out;
+	}
+	mutex_unlock(&priv->state_lock);
+
+	if (!arfs_rule->rule) {
+		rule = arfs_add_rule(priv, arfs_rule);
+		if (IS_ERR(rule))
+			goto out;
+		arfs_rule->rule = rule;
+	} else {
+		arfs_modify_rule_rq(priv, arfs_rule->rule,
+				    arfs_rule->rxq);
+	}
+out:
+	arfs_may_expire_flow(priv);
+}
+
+static struct arfs_rule *arfs_alloc_rule(struct mlx5e_priv *priv,
+					 struct arfs_table *arfs_t,
+					 const struct flow_keys *fk,
+					 u16 rxq, u32 flow_id)
+{
+	struct mlx5e_arfs_tables *arfs =  mlx5e_fs_get_arfs(priv->fs);
+	struct arfs_rule *rule;
+	struct arfs_tuple *tuple;
+
+	rule = kzalloc(sizeof(*rule), GFP_ATOMIC);
+	if (!rule)
+		return NULL;
+
+	rule->priv = priv;
+	rule->rxq = rxq;
+	INIT_WORK(&rule->arfs_work, arfs_handle_work);
+
+	tuple = &rule->tuple;
+	tuple->etype = fk->basic.n_proto;
+	tuple->ip_proto = fk->basic.ip_proto;
+	if (tuple->etype == htons(ETH_P_IP)) {
+		tuple->src_ipv4 = fk->addrs.v4addrs.src;
+		tuple->dst_ipv4 = fk->addrs.v4addrs.dst;
+	} else {
+		memcpy(&tuple->src_ipv6, &fk->addrs.v6addrs.src,
+		       sizeof(struct in6_addr));
+		memcpy(&tuple->dst_ipv6, &fk->addrs.v6addrs.dst,
+		       sizeof(struct in6_addr));
+	}
+	tuple->src_port = fk->ports.src;
+	tuple->dst_port = fk->ports.dst;
+
+	rule->flow_id = flow_id;
+	rule->filter_id = arfs->last_filter_id++ % RPS_NO_FILTER;
+
+	hlist_add_head(&rule->hlist,
+		       arfs_hash_bucket(arfs_t, tuple->src_port,
+					tuple->dst_port));
+	return rule;
+}
+
+static bool arfs_cmp(const struct arfs_tuple *tuple, const struct flow_keys *fk)
+{
+	if (tuple->src_port != fk->ports.src || tuple->dst_port != fk->ports.dst)
+		return false;
+	if (tuple->etype != fk->basic.n_proto)
+		return false;
+	if (tuple->etype == htons(ETH_P_IP))
+		return tuple->src_ipv4 == fk->addrs.v4addrs.src &&
+		       tuple->dst_ipv4 == fk->addrs.v4addrs.dst;
+	if (tuple->etype == htons(ETH_P_IPV6))
+		return !memcmp(&tuple->src_ipv6, &fk->addrs.v6addrs.src,
+			       sizeof(struct in6_addr)) &&
+		       !memcmp(&tuple->dst_ipv6, &fk->addrs.v6addrs.dst,
+			       sizeof(struct in6_addr));
+	return false;
+}
+
+static struct arfs_rule *arfs_find_rule(struct arfs_table *arfs_t,
+					const struct flow_keys *fk)
+{
+	struct arfs_rule *arfs_rule;
+	struct hlist_head *head;
+
+	head = arfs_hash_bucket(arfs_t, fk->ports.src, fk->ports.dst);
+	hlist_for_each_entry(arfs_rule, head, hlist) {
+		if (arfs_cmp(&arfs_rule->tuple, fk))
+			return arfs_rule;
+	}
+
+	return NULL;
+}
+
+int mlx5e_rx_flow_steer(struct net_device *dev, const struct sk_buff *skb,
+			u16 rxq_index, u32 flow_id)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	struct mlx5e_arfs_tables *arfs;
+	struct arfs_rule *arfs_rule;
+	struct arfs_table *arfs_t;
+	struct flow_keys fk;
+
+	arfs =  mlx5e_fs_get_arfs(priv->fs);
+	if (!skb_flow_dissect_flow_keys(skb, &fk, 0))
+		return -EPROTONOSUPPORT;
+
+	if (fk.basic.n_proto != htons(ETH_P_IP) &&
+	    fk.basic.n_proto != htons(ETH_P_IPV6))
+		return -EPROTONOSUPPORT;
+
+	if (skb->encapsulation)
+		return -EPROTONOSUPPORT;
+
+	arfs_t = arfs_get_table(arfs, fk.basic.ip_proto, fk.basic.n_proto);
+	if (!arfs_t)
+		return -EPROTONOSUPPORT;
+
+	spin_lock_bh(&arfs->arfs_lock);
+	arfs_rule = arfs_find_rule(arfs_t, &fk);
+	if (arfs_rule) {
+		if (arfs_rule->rxq == rxq_index) {
+			spin_unlock_bh(&arfs->arfs_lock);
+			return arfs_rule->filter_id;
+		}
+		arfs_rule->rxq = rxq_index;
+	} else {
+		arfs_rule = arfs_alloc_rule(priv, arfs_t, &fk, rxq_index, flow_id);
+		if (!arfs_rule) {
+			spin_unlock_bh(&arfs->arfs_lock);
+			return -ENOMEM;
+		}
+	}
+	queue_work(arfs->wq, &arfs_rule->arfs_work);
+	spin_unlock_bh(&arfs->arfs_lock);
+	return arfs_rule->filter_id;
+}
+
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_common.c b/drivers/net/ethernet/mellanox/mlx5/core/en_common.c
new file mode 100644
index 000000000..03a99918a
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_common.c
@@ -0,0 +1,170 @@
+/*
+ * Copyright (c) 2016, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "en.h"
+
+/* mlx5e global resources should be placed in this file.
+ * Global resources are common to all the netdevices created on the same nic.
+ */
+
+void mlx5e_mkey_set_relaxed_ordering(struct mlx5_core_dev *mdev, void *mkc)
+{
+	bool ro_pci_enable = pcie_relaxed_ordering_enabled(mdev->pdev);
+	bool ro_write = MLX5_CAP_GEN(mdev, relaxed_ordering_write);
+	bool ro_read = MLX5_CAP_GEN(mdev, relaxed_ordering_read);
+
+	MLX5_SET(mkc, mkc, relaxed_ordering_read, ro_pci_enable && ro_read);
+	MLX5_SET(mkc, mkc, relaxed_ordering_write, ro_pci_enable && ro_write);
+}
+
+int mlx5e_create_mkey(struct mlx5_core_dev *mdev, u32 pdn, u32 *mkey)
+{
+	int inlen = MLX5_ST_SZ_BYTES(create_mkey_in);
+	void *mkc;
+	u32 *in;
+	int err;
+
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in)
+		return -ENOMEM;
+
+	mkc = MLX5_ADDR_OF(create_mkey_in, in, memory_key_mkey_entry);
+	MLX5_SET(mkc, mkc, access_mode_1_0, MLX5_MKC_ACCESS_MODE_PA);
+	MLX5_SET(mkc, mkc, lw, 1);
+	MLX5_SET(mkc, mkc, lr, 1);
+	mlx5e_mkey_set_relaxed_ordering(mdev, mkc);
+	MLX5_SET(mkc, mkc, pd, pdn);
+	MLX5_SET(mkc, mkc, length64, 1);
+	MLX5_SET(mkc, mkc, qpn, 0xffffff);
+
+	err = mlx5_core_create_mkey(mdev, mkey, in, inlen);
+
+	kvfree(in);
+	return err;
+}
+
+int mlx5e_create_mdev_resources(struct mlx5_core_dev *mdev)
+{
+	struct mlx5e_hw_objs *res = &mdev->mlx5e_res.hw_objs;
+	int err;
+
+	err = mlx5_core_alloc_pd(mdev, &res->pdn);
+	if (err) {
+		mlx5_core_err(mdev, "alloc pd failed, %d\n", err);
+		return err;
+	}
+
+	err = mlx5_core_alloc_transport_domain(mdev, &res->td.tdn);
+	if (err) {
+		mlx5_core_err(mdev, "alloc td failed, %d\n", err);
+		goto err_dealloc_pd;
+	}
+
+	err = mlx5e_create_mkey(mdev, res->pdn, &res->mkey);
+	if (err) {
+		mlx5_core_err(mdev, "create mkey failed, %d\n", err);
+		goto err_dealloc_transport_domain;
+	}
+
+	err = mlx5_alloc_bfreg(mdev, &res->bfreg, false, false);
+	if (err) {
+		mlx5_core_err(mdev, "alloc bfreg failed, %d\n", err);
+		goto err_destroy_mkey;
+	}
+
+	INIT_LIST_HEAD(&res->td.tirs_list);
+	mutex_init(&res->td.list_lock);
+
+	return 0;
+
+err_destroy_mkey:
+	mlx5_core_destroy_mkey(mdev, res->mkey);
+err_dealloc_transport_domain:
+	mlx5_core_dealloc_transport_domain(mdev, res->td.tdn);
+err_dealloc_pd:
+	mlx5_core_dealloc_pd(mdev, res->pdn);
+	return err;
+}
+
+void mlx5e_destroy_mdev_resources(struct mlx5_core_dev *mdev)
+{
+	struct mlx5e_hw_objs *res = &mdev->mlx5e_res.hw_objs;
+
+	mlx5_free_bfreg(mdev, &res->bfreg);
+	mlx5_core_destroy_mkey(mdev, res->mkey);
+	mlx5_core_dealloc_transport_domain(mdev, res->td.tdn);
+	mlx5_core_dealloc_pd(mdev, res->pdn);
+	memset(res, 0, sizeof(*res));
+}
+
+int mlx5e_refresh_tirs(struct mlx5e_priv *priv, bool enable_uc_lb,
+		       bool enable_mc_lb)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+	struct mlx5e_tir *tir;
+	u8 lb_flags = 0;
+	int err  = 0;
+	u32 tirn = 0;
+	int inlen;
+	void *in;
+
+	inlen = MLX5_ST_SZ_BYTES(modify_tir_in);
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in)
+		return -ENOMEM;
+
+	if (enable_uc_lb)
+		lb_flags = MLX5_TIRC_SELF_LB_BLOCK_BLOCK_UNICAST;
+
+	if (enable_mc_lb)
+		lb_flags |= MLX5_TIRC_SELF_LB_BLOCK_BLOCK_MULTICAST;
+
+	if (lb_flags)
+		MLX5_SET(modify_tir_in, in, ctx.self_lb_block, lb_flags);
+
+	MLX5_SET(modify_tir_in, in, bitmask.self_lb_en, 1);
+
+	mutex_lock(&mdev->mlx5e_res.hw_objs.td.list_lock);
+	list_for_each_entry(tir, &mdev->mlx5e_res.hw_objs.td.tirs_list, list) {
+		tirn = tir->tirn;
+		err = mlx5_core_modify_tir(mdev, tirn, in);
+		if (err)
+			break;
+	}
+	mutex_unlock(&mdev->mlx5e_res.hw_objs.td.list_lock);
+
+	kvfree(in);
+	if (err)
+		netdev_err(priv->netdev, "refresh tir(0x%x) failed, %d\n", tirn, err);
+
+	return err;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_dcbnl.c b/drivers/net/ethernet/mellanox/mlx5/core/en_dcbnl.c
new file mode 100644
index 000000000..89de92d06
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_dcbnl.c
@@ -0,0 +1,1256 @@
+/*
+ * Copyright (c) 2016, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+#include <linux/device.h>
+#include <linux/netdevice.h>
+#include "en.h"
+#include "en/port.h"
+#include "en/port_buffer.h"
+
+#define MLX5E_MAX_BW_ALLOC 100 /* Max percentage of BW allocation */
+
+#define MLX5E_100MB (100000)
+#define MLX5E_1GB   (1000000)
+
+#define MLX5E_CEE_STATE_UP    1
+#define MLX5E_CEE_STATE_DOWN  0
+
+/* Max supported cable length is 1000 meters */
+#define MLX5E_MAX_CABLE_LENGTH 1000
+
+enum {
+	MLX5E_VENDOR_TC_GROUP_NUM = 7,
+	MLX5E_LOWEST_PRIO_GROUP   = 0,
+};
+
+enum {
+	MLX5_DCB_CHG_RESET,
+	MLX5_DCB_NO_CHG,
+	MLX5_DCB_CHG_NO_RESET,
+};
+
+#define MLX5_DSCP_SUPPORTED(mdev) (MLX5_CAP_GEN(mdev, qcam_reg)  && \
+				   MLX5_CAP_QCAM_REG(mdev, qpts) && \
+				   MLX5_CAP_QCAM_REG(mdev, qpdpm))
+
+static int mlx5e_set_trust_state(struct mlx5e_priv *priv, u8 trust_state);
+static int mlx5e_set_dscp2prio(struct mlx5e_priv *priv, u8 dscp, u8 prio);
+
+/* If dcbx mode is non-host set the dcbx mode to host.
+ */
+static int mlx5e_dcbnl_set_dcbx_mode(struct mlx5e_priv *priv,
+				     enum mlx5_dcbx_oper_mode mode)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+	u32 param[MLX5_ST_SZ_DW(dcbx_param)];
+	int err;
+
+	err = mlx5_query_port_dcbx_param(mdev, param);
+	if (err)
+		return err;
+
+	MLX5_SET(dcbx_param, param, version_admin, mode);
+	if (mode != MLX5E_DCBX_PARAM_VER_OPER_HOST)
+		MLX5_SET(dcbx_param, param, willing_admin, 1);
+
+	return mlx5_set_port_dcbx_param(mdev, param);
+}
+
+static int mlx5e_dcbnl_switch_to_host_mode(struct mlx5e_priv *priv)
+{
+	struct mlx5e_dcbx *dcbx = &priv->dcbx;
+	int err;
+
+	if (!MLX5_CAP_GEN(priv->mdev, dcbx))
+		return 0;
+
+	if (dcbx->mode == MLX5E_DCBX_PARAM_VER_OPER_HOST)
+		return 0;
+
+	err = mlx5e_dcbnl_set_dcbx_mode(priv, MLX5E_DCBX_PARAM_VER_OPER_HOST);
+	if (err)
+		return err;
+
+	dcbx->mode = MLX5E_DCBX_PARAM_VER_OPER_HOST;
+	return 0;
+}
+
+static int mlx5e_dcbnl_ieee_getets(struct net_device *netdev,
+				   struct ieee_ets *ets)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+	u8 tc_group[IEEE_8021QAZ_MAX_TCS];
+	bool is_tc_group_6_exist = false;
+	bool is_zero_bw_ets_tc = false;
+	int err = 0;
+	int i;
+
+	if (!MLX5_CAP_GEN(priv->mdev, ets))
+		return -EOPNOTSUPP;
+
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) {
+		err = mlx5_query_port_prio_tc(mdev, i, &ets->prio_tc[i]);
+		if (err)
+			return err;
+	}
+
+	ets->ets_cap = mlx5_max_tc(priv->mdev) + 1;
+	for (i = 0; i < ets->ets_cap; i++) {
+		err = mlx5_query_port_tc_group(mdev, i, &tc_group[i]);
+		if (err)
+			return err;
+
+		err = mlx5_query_port_tc_bw_alloc(mdev, i, &ets->tc_tx_bw[i]);
+		if (err)
+			return err;
+
+		if (ets->tc_tx_bw[i] < MLX5E_MAX_BW_ALLOC &&
+		    tc_group[i] == (MLX5E_LOWEST_PRIO_GROUP + 1))
+			is_zero_bw_ets_tc = true;
+
+		if (tc_group[i] == (MLX5E_VENDOR_TC_GROUP_NUM - 1))
+			is_tc_group_6_exist = true;
+	}
+
+	/* Report 0% ets tc if exits*/
+	if (is_zero_bw_ets_tc) {
+		for (i = 0; i < ets->ets_cap; i++)
+			if (tc_group[i] == MLX5E_LOWEST_PRIO_GROUP)
+				ets->tc_tx_bw[i] = 0;
+	}
+
+	/* Update tc_tsa based on fw setting*/
+	for (i = 0; i < ets->ets_cap; i++) {
+		if (ets->tc_tx_bw[i] < MLX5E_MAX_BW_ALLOC)
+			priv->dcbx.tc_tsa[i] = IEEE_8021QAZ_TSA_ETS;
+		else if (tc_group[i] == MLX5E_VENDOR_TC_GROUP_NUM &&
+			 !is_tc_group_6_exist)
+			priv->dcbx.tc_tsa[i] = IEEE_8021QAZ_TSA_VENDOR;
+	}
+	memcpy(ets->tc_tsa, priv->dcbx.tc_tsa, sizeof(ets->tc_tsa));
+
+	return err;
+}
+
+static void mlx5e_build_tc_group(struct ieee_ets *ets, u8 *tc_group, int max_tc)
+{
+	bool any_tc_mapped_to_ets = false;
+	bool ets_zero_bw = false;
+	int strict_group;
+	int i;
+
+	for (i = 0; i <= max_tc; i++) {
+		if (ets->tc_tsa[i] == IEEE_8021QAZ_TSA_ETS) {
+			any_tc_mapped_to_ets = true;
+			if (!ets->tc_tx_bw[i])
+				ets_zero_bw = true;
+		}
+	}
+
+	/* strict group has higher priority than ets group */
+	strict_group = MLX5E_LOWEST_PRIO_GROUP;
+	if (any_tc_mapped_to_ets)
+		strict_group++;
+	if (ets_zero_bw)
+		strict_group++;
+
+	for (i = 0; i <= max_tc; i++) {
+		switch (ets->tc_tsa[i]) {
+		case IEEE_8021QAZ_TSA_VENDOR:
+			tc_group[i] = MLX5E_VENDOR_TC_GROUP_NUM;
+			break;
+		case IEEE_8021QAZ_TSA_STRICT:
+			tc_group[i] = strict_group++;
+			break;
+		case IEEE_8021QAZ_TSA_ETS:
+			tc_group[i] = MLX5E_LOWEST_PRIO_GROUP;
+			if (ets->tc_tx_bw[i] && ets_zero_bw)
+				tc_group[i] = MLX5E_LOWEST_PRIO_GROUP + 1;
+			break;
+		}
+	}
+}
+
+static void mlx5e_build_tc_tx_bw(struct ieee_ets *ets, u8 *tc_tx_bw,
+				 u8 *tc_group, int max_tc)
+{
+	int bw_for_ets_zero_bw_tc = 0;
+	int last_ets_zero_bw_tc = -1;
+	int num_ets_zero_bw = 0;
+	int i;
+
+	for (i = 0; i <= max_tc; i++) {
+		if (ets->tc_tsa[i] == IEEE_8021QAZ_TSA_ETS &&
+		    !ets->tc_tx_bw[i]) {
+			num_ets_zero_bw++;
+			last_ets_zero_bw_tc = i;
+		}
+	}
+
+	if (num_ets_zero_bw)
+		bw_for_ets_zero_bw_tc = MLX5E_MAX_BW_ALLOC / num_ets_zero_bw;
+
+	for (i = 0; i <= max_tc; i++) {
+		switch (ets->tc_tsa[i]) {
+		case IEEE_8021QAZ_TSA_VENDOR:
+			tc_tx_bw[i] = MLX5E_MAX_BW_ALLOC;
+			break;
+		case IEEE_8021QAZ_TSA_STRICT:
+			tc_tx_bw[i] = MLX5E_MAX_BW_ALLOC;
+			break;
+		case IEEE_8021QAZ_TSA_ETS:
+			tc_tx_bw[i] = ets->tc_tx_bw[i] ?
+				      ets->tc_tx_bw[i] :
+				      bw_for_ets_zero_bw_tc;
+			break;
+		}
+	}
+
+	/* Make sure the total bw for ets zero bw group is 100% */
+	if (last_ets_zero_bw_tc != -1)
+		tc_tx_bw[last_ets_zero_bw_tc] +=
+			MLX5E_MAX_BW_ALLOC % num_ets_zero_bw;
+}
+
+/* If there are ETS BW 0,
+ *   Set ETS group # to 1 for all ETS non zero BW tcs. Their sum must be 100%.
+ *   Set group #0 to all the ETS BW 0 tcs and
+ *     equally splits the 100% BW between them
+ *   Report both group #0 and #1 as ETS type.
+ *     All the tcs in group #0 will be reported with 0% BW.
+ */
+static int mlx5e_dcbnl_ieee_setets_core(struct mlx5e_priv *priv, struct ieee_ets *ets)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+	u8 tc_tx_bw[IEEE_8021QAZ_MAX_TCS];
+	u8 tc_group[IEEE_8021QAZ_MAX_TCS];
+	int max_tc = mlx5_max_tc(mdev);
+	int err, i;
+
+	mlx5e_build_tc_group(ets, tc_group, max_tc);
+	mlx5e_build_tc_tx_bw(ets, tc_tx_bw, tc_group, max_tc);
+
+	err = mlx5_set_port_prio_tc(mdev, ets->prio_tc);
+	if (err)
+		return err;
+
+	err = mlx5_set_port_tc_group(mdev, tc_group);
+	if (err)
+		return err;
+
+	err = mlx5_set_port_tc_bw_alloc(mdev, tc_tx_bw);
+
+	if (err)
+		return err;
+
+	memcpy(priv->dcbx.tc_tsa, ets->tc_tsa, sizeof(ets->tc_tsa));
+
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) {
+		mlx5e_dbg(HW, priv, "%s: prio_%d <=> tc_%d\n",
+			  __func__, i, ets->prio_tc[i]);
+		mlx5e_dbg(HW, priv, "%s: tc_%d <=> tx_bw_%d%%, group_%d\n",
+			  __func__, i, tc_tx_bw[i], tc_group[i]);
+	}
+
+	return err;
+}
+
+static int mlx5e_dbcnl_validate_ets(struct net_device *netdev,
+				    struct ieee_ets *ets,
+				    bool zero_sum_allowed)
+{
+	bool have_ets_tc = false;
+	int bw_sum = 0;
+	int i;
+
+	/* Validate Priority */
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) {
+		if (ets->prio_tc[i] >= MLX5E_MAX_PRIORITY) {
+			netdev_err(netdev,
+				   "Failed to validate ETS: priority value greater than max(%d)\n",
+				    MLX5E_MAX_PRIORITY);
+			return -EINVAL;
+		}
+	}
+
+	/* Validate Bandwidth Sum */
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) {
+		if (ets->tc_tsa[i] == IEEE_8021QAZ_TSA_ETS) {
+			have_ets_tc = true;
+			bw_sum += ets->tc_tx_bw[i];
+		}
+	}
+
+	if (have_ets_tc && bw_sum != 100) {
+		if (bw_sum || (!bw_sum && !zero_sum_allowed))
+			netdev_err(netdev,
+				   "Failed to validate ETS: BW sum is illegal\n");
+		return -EINVAL;
+	}
+	return 0;
+}
+
+static int mlx5e_dcbnl_ieee_setets(struct net_device *netdev,
+				   struct ieee_ets *ets)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	int err;
+
+	if (!MLX5_CAP_GEN(priv->mdev, ets))
+		return -EOPNOTSUPP;
+
+	err = mlx5e_dbcnl_validate_ets(netdev, ets, false);
+	if (err)
+		return err;
+
+	err = mlx5e_dcbnl_ieee_setets_core(priv, ets);
+	if (err)
+		return err;
+
+	return 0;
+}
+
+static int mlx5e_dcbnl_ieee_getpfc(struct net_device *dev,
+				   struct ieee_pfc *pfc)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+	struct mlx5e_pport_stats *pstats = &priv->stats.pport;
+	int i;
+
+	pfc->pfc_cap = mlx5_max_tc(mdev) + 1;
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) {
+		pfc->requests[i]    = PPORT_PER_PRIO_GET(pstats, i, tx_pause);
+		pfc->indications[i] = PPORT_PER_PRIO_GET(pstats, i, rx_pause);
+	}
+
+	if (MLX5_BUFFER_SUPPORTED(mdev))
+		pfc->delay = priv->dcbx.cable_len;
+
+	return mlx5_query_port_pfc(mdev, &pfc->pfc_en, NULL);
+}
+
+static int mlx5e_dcbnl_ieee_setpfc(struct net_device *dev,
+				   struct ieee_pfc *pfc)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+	u32 old_cable_len = priv->dcbx.cable_len;
+	struct ieee_pfc pfc_new;
+	u32 changed = 0;
+	u8 curr_pfc_en;
+	int ret = 0;
+
+	/* pfc_en */
+	mlx5_query_port_pfc(mdev, &curr_pfc_en, NULL);
+	if (pfc->pfc_en != curr_pfc_en) {
+		ret = mlx5_set_port_pfc(mdev, pfc->pfc_en, pfc->pfc_en);
+		if (ret)
+			return ret;
+		mlx5_toggle_port_link(mdev);
+		changed |= MLX5E_PORT_BUFFER_PFC;
+	}
+
+	if (pfc->delay &&
+	    pfc->delay < MLX5E_MAX_CABLE_LENGTH &&
+	    pfc->delay != priv->dcbx.cable_len) {
+		priv->dcbx.cable_len = pfc->delay;
+		changed |= MLX5E_PORT_BUFFER_CABLE_LEN;
+	}
+
+	if (MLX5_BUFFER_SUPPORTED(mdev)) {
+		pfc_new.pfc_en = (changed & MLX5E_PORT_BUFFER_PFC) ? pfc->pfc_en : curr_pfc_en;
+		if (priv->dcbx.manual_buffer)
+			ret = mlx5e_port_manual_buffer_config(priv, changed,
+							      dev->mtu, &pfc_new,
+							      NULL, NULL);
+
+		if (ret && (changed & MLX5E_PORT_BUFFER_CABLE_LEN))
+			priv->dcbx.cable_len = old_cable_len;
+	}
+
+	if (!ret) {
+		mlx5e_dbg(HW, priv,
+			  "%s: PFC per priority bit mask: 0x%x\n",
+			  __func__, pfc->pfc_en);
+	}
+	return ret;
+}
+
+static u8 mlx5e_dcbnl_getdcbx(struct net_device *dev)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+
+	return priv->dcbx.cap;
+}
+
+static u8 mlx5e_dcbnl_setdcbx(struct net_device *dev, u8 mode)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	struct mlx5e_dcbx *dcbx = &priv->dcbx;
+
+	if (mode & DCB_CAP_DCBX_LLD_MANAGED)
+		return 1;
+
+	if ((!mode) && MLX5_CAP_GEN(priv->mdev, dcbx)) {
+		if (dcbx->mode == MLX5E_DCBX_PARAM_VER_OPER_AUTO)
+			return 0;
+
+		/* set dcbx to fw controlled */
+		if (!mlx5e_dcbnl_set_dcbx_mode(priv, MLX5E_DCBX_PARAM_VER_OPER_AUTO)) {
+			dcbx->mode = MLX5E_DCBX_PARAM_VER_OPER_AUTO;
+			dcbx->cap &= ~DCB_CAP_DCBX_HOST;
+			return 0;
+		}
+
+		return 1;
+	}
+
+	if (!(mode & DCB_CAP_DCBX_HOST))
+		return 1;
+
+	if (mlx5e_dcbnl_switch_to_host_mode(netdev_priv(dev)))
+		return 1;
+
+	dcbx->cap = mode;
+
+	return 0;
+}
+
+static int mlx5e_dcbnl_ieee_setapp(struct net_device *dev, struct dcb_app *app)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	struct dcb_app temp;
+	bool is_new;
+	int err;
+
+	if (!MLX5_CAP_GEN(priv->mdev, vport_group_manager) ||
+	    !MLX5_DSCP_SUPPORTED(priv->mdev))
+		return -EOPNOTSUPP;
+
+	if ((app->selector != IEEE_8021QAZ_APP_SEL_DSCP) ||
+	    (app->protocol >= MLX5E_MAX_DSCP))
+		return -EINVAL;
+
+	/* Save the old entry info */
+	temp.selector = IEEE_8021QAZ_APP_SEL_DSCP;
+	temp.protocol = app->protocol;
+	temp.priority = priv->dcbx_dp.dscp2prio[app->protocol];
+
+	/* Check if need to switch to dscp trust state */
+	if (!priv->dcbx.dscp_app_cnt) {
+		err =  mlx5e_set_trust_state(priv, MLX5_QPTS_TRUST_DSCP);
+		if (err)
+			return err;
+	}
+
+	/* Skip the fw command if new and old mapping are the same */
+	if (app->priority != priv->dcbx_dp.dscp2prio[app->protocol]) {
+		err = mlx5e_set_dscp2prio(priv, app->protocol, app->priority);
+		if (err)
+			goto fw_err;
+	}
+
+	/* Delete the old entry if exists */
+	is_new = false;
+	err = dcb_ieee_delapp(dev, &temp);
+	if (err)
+		is_new = true;
+
+	/* Add new entry and update counter */
+	err = dcb_ieee_setapp(dev, app);
+	if (err)
+		return err;
+
+	if (is_new)
+		priv->dcbx.dscp_app_cnt++;
+
+	return err;
+
+fw_err:
+	mlx5e_set_trust_state(priv, MLX5_QPTS_TRUST_PCP);
+	return err;
+}
+
+static int mlx5e_dcbnl_ieee_delapp(struct net_device *dev, struct dcb_app *app)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	int err;
+
+	if  (!MLX5_CAP_GEN(priv->mdev, vport_group_manager) ||
+	     !MLX5_DSCP_SUPPORTED(priv->mdev))
+		return -EOPNOTSUPP;
+
+	if ((app->selector != IEEE_8021QAZ_APP_SEL_DSCP) ||
+	    (app->protocol >= MLX5E_MAX_DSCP))
+		return -EINVAL;
+
+	/* Skip if no dscp app entry */
+	if (!priv->dcbx.dscp_app_cnt)
+		return -ENOENT;
+
+	/* Check if the entry matches fw setting */
+	if (app->priority != priv->dcbx_dp.dscp2prio[app->protocol])
+		return -ENOENT;
+
+	/* Delete the app entry */
+	err = dcb_ieee_delapp(dev, app);
+	if (err)
+		return err;
+
+	/* Reset the priority mapping back to zero */
+	err = mlx5e_set_dscp2prio(priv, app->protocol, 0);
+	if (err)
+		goto fw_err;
+
+	priv->dcbx.dscp_app_cnt--;
+
+	/* Check if need to switch to pcp trust state */
+	if (!priv->dcbx.dscp_app_cnt)
+		err = mlx5e_set_trust_state(priv, MLX5_QPTS_TRUST_PCP);
+
+	return err;
+
+fw_err:
+	mlx5e_set_trust_state(priv, MLX5_QPTS_TRUST_PCP);
+	return err;
+}
+
+static int mlx5e_dcbnl_ieee_getmaxrate(struct net_device *netdev,
+				       struct ieee_maxrate *maxrate)
+{
+	struct mlx5e_priv *priv    = netdev_priv(netdev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+	u8 max_bw_value[IEEE_8021QAZ_MAX_TCS];
+	u8 max_bw_unit[IEEE_8021QAZ_MAX_TCS];
+	int err;
+	int i;
+
+	err = mlx5_query_port_ets_rate_limit(mdev, max_bw_value, max_bw_unit);
+	if (err)
+		return err;
+
+	memset(maxrate->tc_maxrate, 0, sizeof(maxrate->tc_maxrate));
+
+	for (i = 0; i <= mlx5_max_tc(mdev); i++) {
+		switch (max_bw_unit[i]) {
+		case MLX5_100_MBPS_UNIT:
+			maxrate->tc_maxrate[i] = max_bw_value[i] * MLX5E_100MB;
+			break;
+		case MLX5_GBPS_UNIT:
+			maxrate->tc_maxrate[i] = max_bw_value[i] * MLX5E_1GB;
+			break;
+		case MLX5_BW_NO_LIMIT:
+			break;
+		default:
+			WARN(true, "non-supported BW unit");
+			break;
+		}
+	}
+
+	return 0;
+}
+
+static int mlx5e_dcbnl_ieee_setmaxrate(struct net_device *netdev,
+				       struct ieee_maxrate *maxrate)
+{
+	struct mlx5e_priv *priv    = netdev_priv(netdev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+	u8 max_bw_value[IEEE_8021QAZ_MAX_TCS];
+	u8 max_bw_unit[IEEE_8021QAZ_MAX_TCS];
+	__u64 upper_limit_mbps = roundup(255 * MLX5E_100MB, MLX5E_1GB);
+	int i;
+
+	memset(max_bw_value, 0, sizeof(max_bw_value));
+	memset(max_bw_unit, 0, sizeof(max_bw_unit));
+
+	for (i = 0; i <= mlx5_max_tc(mdev); i++) {
+		if (!maxrate->tc_maxrate[i]) {
+			max_bw_unit[i]  = MLX5_BW_NO_LIMIT;
+			continue;
+		}
+		if (maxrate->tc_maxrate[i] < upper_limit_mbps) {
+			max_bw_value[i] = div_u64(maxrate->tc_maxrate[i],
+						  MLX5E_100MB);
+			max_bw_value[i] = max_bw_value[i] ? max_bw_value[i] : 1;
+			max_bw_unit[i]  = MLX5_100_MBPS_UNIT;
+		} else {
+			max_bw_value[i] = div_u64(maxrate->tc_maxrate[i],
+						  MLX5E_1GB);
+			max_bw_unit[i]  = MLX5_GBPS_UNIT;
+		}
+	}
+
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) {
+		mlx5e_dbg(HW, priv, "%s: tc_%d <=> max_bw %d Gbps\n",
+			  __func__, i, max_bw_value[i]);
+	}
+
+	return mlx5_modify_port_ets_rate_limit(mdev, max_bw_value, max_bw_unit);
+}
+
+static u8 mlx5e_dcbnl_setall(struct net_device *netdev)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	struct mlx5e_cee_config *cee_cfg = &priv->dcbx.cee_cfg;
+	struct mlx5_core_dev *mdev = priv->mdev;
+	struct ieee_ets ets;
+	struct ieee_pfc pfc;
+	int err = -EOPNOTSUPP;
+	int i;
+
+	if (!MLX5_CAP_GEN(mdev, ets))
+		goto out;
+
+	memset(&ets, 0, sizeof(ets));
+	memset(&pfc, 0, sizeof(pfc));
+
+	ets.ets_cap = IEEE_8021QAZ_MAX_TCS;
+	for (i = 0; i < CEE_DCBX_MAX_PGS; i++) {
+		ets.tc_tx_bw[i] = cee_cfg->pg_bw_pct[i];
+		ets.tc_rx_bw[i] = cee_cfg->pg_bw_pct[i];
+		ets.tc_tsa[i]   = IEEE_8021QAZ_TSA_ETS;
+		ets.prio_tc[i]  = cee_cfg->prio_to_pg_map[i];
+		mlx5e_dbg(HW, priv,
+			  "%s: Priority group %d: tx_bw %d, rx_bw %d, prio_tc %d\n",
+			  __func__, i, ets.tc_tx_bw[i], ets.tc_rx_bw[i],
+			  ets.prio_tc[i]);
+	}
+
+	err = mlx5e_dbcnl_validate_ets(netdev, &ets, true);
+	if (err)
+		goto out;
+
+	err = mlx5e_dcbnl_ieee_setets_core(priv, &ets);
+	if (err) {
+		netdev_err(netdev,
+			   "%s, Failed to set ETS: %d\n", __func__, err);
+		goto out;
+	}
+
+	/* Set PFC */
+	pfc.pfc_cap = mlx5_max_tc(mdev) + 1;
+	if (!cee_cfg->pfc_enable)
+		pfc.pfc_en = 0;
+	else
+		for (i = 0; i < CEE_DCBX_MAX_PRIO; i++)
+			pfc.pfc_en |= cee_cfg->pfc_setting[i] << i;
+
+	err = mlx5e_dcbnl_ieee_setpfc(netdev, &pfc);
+	if (err) {
+		netdev_err(netdev,
+			   "%s, Failed to set PFC: %d\n", __func__, err);
+		goto out;
+	}
+out:
+	return err ? MLX5_DCB_NO_CHG : MLX5_DCB_CHG_RESET;
+}
+
+static u8 mlx5e_dcbnl_getstate(struct net_device *netdev)
+{
+	return MLX5E_CEE_STATE_UP;
+}
+
+static void mlx5e_dcbnl_getpermhwaddr(struct net_device *netdev,
+				      u8 *perm_addr)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+
+	if (!perm_addr)
+		return;
+
+	memset(perm_addr, 0xff, MAX_ADDR_LEN);
+
+	mlx5_query_mac_address(priv->mdev, perm_addr);
+}
+
+static void mlx5e_dcbnl_setpgtccfgtx(struct net_device *netdev,
+				     int priority, u8 prio_type,
+				     u8 pgid, u8 bw_pct, u8 up_map)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	struct mlx5e_cee_config *cee_cfg = &priv->dcbx.cee_cfg;
+
+	if (priority >= CEE_DCBX_MAX_PRIO) {
+		netdev_err(netdev,
+			   "%s, priority is out of range\n", __func__);
+		return;
+	}
+
+	if (pgid >= CEE_DCBX_MAX_PGS) {
+		netdev_err(netdev,
+			   "%s, priority group is out of range\n", __func__);
+		return;
+	}
+
+	cee_cfg->prio_to_pg_map[priority] = pgid;
+}
+
+static void mlx5e_dcbnl_setpgbwgcfgtx(struct net_device *netdev,
+				      int pgid, u8 bw_pct)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	struct mlx5e_cee_config *cee_cfg = &priv->dcbx.cee_cfg;
+
+	if (pgid >= CEE_DCBX_MAX_PGS) {
+		netdev_err(netdev,
+			   "%s, priority group is out of range\n", __func__);
+		return;
+	}
+
+	cee_cfg->pg_bw_pct[pgid] = bw_pct;
+}
+
+static void mlx5e_dcbnl_getpgtccfgtx(struct net_device *netdev,
+				     int priority, u8 *prio_type,
+				     u8 *pgid, u8 *bw_pct, u8 *up_map)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+
+	if (!MLX5_CAP_GEN(priv->mdev, ets)) {
+		netdev_err(netdev, "%s, ets is not supported\n", __func__);
+		return;
+	}
+
+	if (priority >= CEE_DCBX_MAX_PRIO) {
+		netdev_err(netdev,
+			   "%s, priority is out of range\n", __func__);
+		return;
+	}
+
+	*prio_type = 0;
+	*bw_pct = 0;
+	*up_map = 0;
+
+	if (mlx5_query_port_prio_tc(mdev, priority, pgid))
+		*pgid = 0;
+}
+
+static void mlx5e_dcbnl_getpgbwgcfgtx(struct net_device *netdev,
+				      int pgid, u8 *bw_pct)
+{
+	struct ieee_ets ets;
+
+	if (pgid >= CEE_DCBX_MAX_PGS) {
+		netdev_err(netdev,
+			   "%s, priority group is out of range\n", __func__);
+		return;
+	}
+
+	mlx5e_dcbnl_ieee_getets(netdev, &ets);
+	*bw_pct = ets.tc_tx_bw[pgid];
+}
+
+static void mlx5e_dcbnl_setpfccfg(struct net_device *netdev,
+				  int priority, u8 setting)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	struct mlx5e_cee_config *cee_cfg = &priv->dcbx.cee_cfg;
+
+	if (priority >= CEE_DCBX_MAX_PRIO) {
+		netdev_err(netdev,
+			   "%s, priority is out of range\n", __func__);
+		return;
+	}
+
+	if (setting > 1)
+		return;
+
+	cee_cfg->pfc_setting[priority] = setting;
+}
+
+static int
+mlx5e_dcbnl_get_priority_pfc(struct net_device *netdev,
+			     int priority, u8 *setting)
+{
+	struct ieee_pfc pfc;
+	int err;
+
+	err = mlx5e_dcbnl_ieee_getpfc(netdev, &pfc);
+
+	if (err)
+		*setting = 0;
+	else
+		*setting = (pfc.pfc_en >> priority) & 0x01;
+
+	return err;
+}
+
+static void mlx5e_dcbnl_getpfccfg(struct net_device *netdev,
+				  int priority, u8 *setting)
+{
+	if (priority >= CEE_DCBX_MAX_PRIO) {
+		netdev_err(netdev,
+			   "%s, priority is out of range\n", __func__);
+		return;
+	}
+
+	if (!setting)
+		return;
+
+	mlx5e_dcbnl_get_priority_pfc(netdev, priority, setting);
+}
+
+static u8 mlx5e_dcbnl_getcap(struct net_device *netdev,
+			     int capid, u8 *cap)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+	u8 rval = 0;
+
+	switch (capid) {
+	case DCB_CAP_ATTR_PG:
+		*cap = true;
+		break;
+	case DCB_CAP_ATTR_PFC:
+		*cap = true;
+		break;
+	case DCB_CAP_ATTR_UP2TC:
+		*cap = false;
+		break;
+	case DCB_CAP_ATTR_PG_TCS:
+		*cap = 1 << mlx5_max_tc(mdev);
+		break;
+	case DCB_CAP_ATTR_PFC_TCS:
+		*cap = 1 << mlx5_max_tc(mdev);
+		break;
+	case DCB_CAP_ATTR_GSP:
+		*cap = false;
+		break;
+	case DCB_CAP_ATTR_BCN:
+		*cap = false;
+		break;
+	case DCB_CAP_ATTR_DCBX:
+		*cap = priv->dcbx.cap |
+		       DCB_CAP_DCBX_VER_CEE |
+		       DCB_CAP_DCBX_VER_IEEE;
+		break;
+	default:
+		*cap = 0;
+		rval = 1;
+		break;
+	}
+
+	return rval;
+}
+
+static int mlx5e_dcbnl_getnumtcs(struct net_device *netdev,
+				 int tcs_id, u8 *num)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+
+	switch (tcs_id) {
+	case DCB_NUMTCS_ATTR_PG:
+	case DCB_NUMTCS_ATTR_PFC:
+		*num = mlx5_max_tc(mdev) + 1;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static u8 mlx5e_dcbnl_getpfcstate(struct net_device *netdev)
+{
+	struct ieee_pfc pfc;
+
+	if (mlx5e_dcbnl_ieee_getpfc(netdev, &pfc))
+		return MLX5E_CEE_STATE_DOWN;
+
+	return pfc.pfc_en ? MLX5E_CEE_STATE_UP : MLX5E_CEE_STATE_DOWN;
+}
+
+static void mlx5e_dcbnl_setpfcstate(struct net_device *netdev, u8 state)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	struct mlx5e_cee_config *cee_cfg = &priv->dcbx.cee_cfg;
+
+	if ((state != MLX5E_CEE_STATE_UP) && (state != MLX5E_CEE_STATE_DOWN))
+		return;
+
+	cee_cfg->pfc_enable = state;
+}
+
+static int mlx5e_dcbnl_getbuffer(struct net_device *dev,
+				 struct dcbnl_buffer *dcb_buffer)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+	struct mlx5e_port_buffer port_buffer;
+	u8 buffer[MLX5E_MAX_PRIORITY];
+	int i, err;
+
+	if (!MLX5_BUFFER_SUPPORTED(mdev))
+		return -EOPNOTSUPP;
+
+	err = mlx5e_port_query_priority2buffer(mdev, buffer);
+	if (err)
+		return err;
+
+	for (i = 0; i < MLX5E_MAX_PRIORITY; i++)
+		dcb_buffer->prio2buffer[i] = buffer[i];
+
+	err = mlx5e_port_query_buffer(priv, &port_buffer);
+	if (err)
+		return err;
+
+	for (i = 0; i < MLX5E_MAX_BUFFER; i++)
+		dcb_buffer->buffer_size[i] = port_buffer.buffer[i].size;
+	dcb_buffer->total_size = port_buffer.port_buffer_size;
+
+	return 0;
+}
+
+static int mlx5e_dcbnl_setbuffer(struct net_device *dev,
+				 struct dcbnl_buffer *dcb_buffer)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+	struct mlx5e_port_buffer port_buffer;
+	u8 old_prio2buffer[MLX5E_MAX_PRIORITY];
+	u32 *buffer_size = NULL;
+	u8 *prio2buffer = NULL;
+	u32 changed = 0;
+	int i, err;
+
+	if (!MLX5_BUFFER_SUPPORTED(mdev))
+		return -EOPNOTSUPP;
+
+	for (i = 0; i < DCBX_MAX_BUFFERS; i++)
+		mlx5_core_dbg(mdev, "buffer[%d]=%d\n", i, dcb_buffer->buffer_size[i]);
+
+	for (i = 0; i < MLX5E_MAX_PRIORITY; i++)
+		mlx5_core_dbg(mdev, "priority %d buffer%d\n", i, dcb_buffer->prio2buffer[i]);
+
+	err = mlx5e_port_query_priority2buffer(mdev, old_prio2buffer);
+	if (err)
+		return err;
+
+	for (i = 0; i < MLX5E_MAX_PRIORITY; i++) {
+		if (dcb_buffer->prio2buffer[i] != old_prio2buffer[i]) {
+			changed |= MLX5E_PORT_BUFFER_PRIO2BUFFER;
+			prio2buffer = dcb_buffer->prio2buffer;
+			break;
+		}
+	}
+
+	err = mlx5e_port_query_buffer(priv, &port_buffer);
+	if (err)
+		return err;
+
+	for (i = 0; i < MLX5E_MAX_BUFFER; i++) {
+		if (port_buffer.buffer[i].size != dcb_buffer->buffer_size[i]) {
+			changed |= MLX5E_PORT_BUFFER_SIZE;
+			buffer_size = dcb_buffer->buffer_size;
+			break;
+		}
+	}
+
+	if (!changed)
+		return 0;
+
+	priv->dcbx.manual_buffer = true;
+	err = mlx5e_port_manual_buffer_config(priv, changed, dev->mtu, NULL,
+					      buffer_size, prio2buffer);
+	return err;
+}
+
+static const struct dcbnl_rtnl_ops mlx5e_dcbnl_ops = {
+	.ieee_getets	= mlx5e_dcbnl_ieee_getets,
+	.ieee_setets	= mlx5e_dcbnl_ieee_setets,
+	.ieee_getmaxrate = mlx5e_dcbnl_ieee_getmaxrate,
+	.ieee_setmaxrate = mlx5e_dcbnl_ieee_setmaxrate,
+	.ieee_getpfc	= mlx5e_dcbnl_ieee_getpfc,
+	.ieee_setpfc	= mlx5e_dcbnl_ieee_setpfc,
+	.ieee_setapp    = mlx5e_dcbnl_ieee_setapp,
+	.ieee_delapp    = mlx5e_dcbnl_ieee_delapp,
+	.getdcbx	= mlx5e_dcbnl_getdcbx,
+	.setdcbx	= mlx5e_dcbnl_setdcbx,
+	.dcbnl_getbuffer = mlx5e_dcbnl_getbuffer,
+	.dcbnl_setbuffer = mlx5e_dcbnl_setbuffer,
+
+/* CEE interfaces */
+	.setall         = mlx5e_dcbnl_setall,
+	.getstate       = mlx5e_dcbnl_getstate,
+	.getpermhwaddr  = mlx5e_dcbnl_getpermhwaddr,
+
+	.setpgtccfgtx   = mlx5e_dcbnl_setpgtccfgtx,
+	.setpgbwgcfgtx  = mlx5e_dcbnl_setpgbwgcfgtx,
+	.getpgtccfgtx   = mlx5e_dcbnl_getpgtccfgtx,
+	.getpgbwgcfgtx  = mlx5e_dcbnl_getpgbwgcfgtx,
+
+	.setpfccfg      = mlx5e_dcbnl_setpfccfg,
+	.getpfccfg      = mlx5e_dcbnl_getpfccfg,
+	.getcap         = mlx5e_dcbnl_getcap,
+	.getnumtcs      = mlx5e_dcbnl_getnumtcs,
+	.getpfcstate    = mlx5e_dcbnl_getpfcstate,
+	.setpfcstate    = mlx5e_dcbnl_setpfcstate,
+};
+
+void mlx5e_dcbnl_build_netdev(struct net_device *netdev)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+
+	if (MLX5_CAP_GEN(mdev, vport_group_manager) && MLX5_CAP_GEN(mdev, qos))
+		netdev->dcbnl_ops = &mlx5e_dcbnl_ops;
+}
+
+static void mlx5e_dcbnl_query_dcbx_mode(struct mlx5e_priv *priv,
+					enum mlx5_dcbx_oper_mode *mode)
+{
+	u32 out[MLX5_ST_SZ_DW(dcbx_param)];
+
+	*mode = MLX5E_DCBX_PARAM_VER_OPER_HOST;
+
+	if (!mlx5_query_port_dcbx_param(priv->mdev, out))
+		*mode = MLX5_GET(dcbx_param, out, version_oper);
+
+	/* From driver's point of view, we only care if the mode
+	 * is host (HOST) or non-host (AUTO)
+	 */
+	if (*mode != MLX5E_DCBX_PARAM_VER_OPER_HOST)
+		*mode = MLX5E_DCBX_PARAM_VER_OPER_AUTO;
+}
+
+static void mlx5e_ets_init(struct mlx5e_priv *priv)
+{
+	struct ieee_ets ets;
+	int err;
+	int i;
+
+	if (!MLX5_CAP_GEN(priv->mdev, ets))
+		return;
+
+	memset(&ets, 0, sizeof(ets));
+	ets.ets_cap = mlx5_max_tc(priv->mdev) + 1;
+	for (i = 0; i < ets.ets_cap; i++) {
+		ets.tc_tx_bw[i] = MLX5E_MAX_BW_ALLOC;
+		ets.tc_tsa[i] = IEEE_8021QAZ_TSA_VENDOR;
+		ets.prio_tc[i] = i;
+	}
+
+	if (ets.ets_cap > 1) {
+		/* tclass[prio=0]=1, tclass[prio=1]=0, tclass[prio=i]=i (for i>1) */
+		ets.prio_tc[0] = 1;
+		ets.prio_tc[1] = 0;
+	}
+
+	err = mlx5e_dcbnl_ieee_setets_core(priv, &ets);
+	if (err)
+		netdev_err(priv->netdev,
+			   "%s, Failed to init ETS: %d\n", __func__, err);
+}
+
+enum {
+	INIT,
+	DELETE,
+};
+
+static void mlx5e_dcbnl_dscp_app(struct mlx5e_priv *priv, int action)
+{
+	struct dcb_app temp;
+	int i;
+
+	if (!MLX5_CAP_GEN(priv->mdev, vport_group_manager))
+		return;
+
+	if (!MLX5_DSCP_SUPPORTED(priv->mdev))
+		return;
+
+	/* No SEL_DSCP entry in non DSCP state */
+	if (priv->dcbx_dp.trust_state != MLX5_QPTS_TRUST_DSCP)
+		return;
+
+	temp.selector = IEEE_8021QAZ_APP_SEL_DSCP;
+	for (i = 0; i < MLX5E_MAX_DSCP; i++) {
+		temp.protocol = i;
+		temp.priority = priv->dcbx_dp.dscp2prio[i];
+		if (action == INIT)
+			dcb_ieee_setapp(priv->netdev, &temp);
+		else
+			dcb_ieee_delapp(priv->netdev, &temp);
+	}
+
+	priv->dcbx.dscp_app_cnt = (action == INIT) ? MLX5E_MAX_DSCP : 0;
+}
+
+void mlx5e_dcbnl_init_app(struct mlx5e_priv *priv)
+{
+	mlx5e_dcbnl_dscp_app(priv, INIT);
+}
+
+void mlx5e_dcbnl_delete_app(struct mlx5e_priv *priv)
+{
+	mlx5e_dcbnl_dscp_app(priv, DELETE);
+}
+
+static void mlx5e_params_calc_trust_tx_min_inline_mode(struct mlx5_core_dev *mdev,
+						       struct mlx5e_params *params,
+						       u8 trust_state)
+{
+	mlx5_query_min_inline(mdev, &params->tx_min_inline_mode);
+	if (trust_state == MLX5_QPTS_TRUST_DSCP &&
+	    params->tx_min_inline_mode == MLX5_INLINE_MODE_L2)
+		params->tx_min_inline_mode = MLX5_INLINE_MODE_IP;
+}
+
+static int mlx5e_update_trust_state_hw(struct mlx5e_priv *priv, void *context)
+{
+	u8 *trust_state = context;
+	int err;
+
+	err = mlx5_set_trust_state(priv->mdev, *trust_state);
+	if (err)
+		return err;
+	WRITE_ONCE(priv->dcbx_dp.trust_state, *trust_state);
+
+	return 0;
+}
+
+static int mlx5e_set_trust_state(struct mlx5e_priv *priv, u8 trust_state)
+{
+	struct mlx5e_params new_params;
+	bool reset = true;
+	int err;
+
+	mutex_lock(&priv->state_lock);
+
+	new_params = priv->channels.params;
+	mlx5e_params_calc_trust_tx_min_inline_mode(priv->mdev, &new_params,
+						   trust_state);
+
+	/* Skip if tx_min_inline is the same */
+	if (new_params.tx_min_inline_mode == priv->channels.params.tx_min_inline_mode)
+		reset = false;
+
+	err = mlx5e_safe_switch_params(priv, &new_params,
+				       mlx5e_update_trust_state_hw,
+				       &trust_state, reset);
+
+	mutex_unlock(&priv->state_lock);
+
+	return err;
+}
+
+static int mlx5e_set_dscp2prio(struct mlx5e_priv *priv, u8 dscp, u8 prio)
+{
+	int err;
+
+	err = mlx5_set_dscp2prio(priv->mdev, dscp, prio);
+	if (err)
+		return err;
+
+	priv->dcbx_dp.dscp2prio[dscp] = prio;
+	return err;
+}
+
+static int mlx5e_trust_initialize(struct mlx5e_priv *priv)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+	u8 trust_state;
+	int err;
+
+	if (!MLX5_DSCP_SUPPORTED(mdev)) {
+		WRITE_ONCE(priv->dcbx_dp.trust_state, MLX5_QPTS_TRUST_PCP);
+		return 0;
+	}
+
+	err = mlx5_query_trust_state(priv->mdev, &trust_state);
+	if (err)
+		return err;
+	WRITE_ONCE(priv->dcbx_dp.trust_state, trust_state);
+
+	if (priv->dcbx_dp.trust_state == MLX5_QPTS_TRUST_PCP && priv->dcbx.dscp_app_cnt) {
+		/*
+		 * Align the driver state with the register state.
+		 * Temporary state change is required to enable the app list reset.
+		 */
+		priv->dcbx_dp.trust_state = MLX5_QPTS_TRUST_DSCP;
+		mlx5e_dcbnl_delete_app(priv);
+		priv->dcbx_dp.trust_state = MLX5_QPTS_TRUST_PCP;
+	}
+
+	mlx5e_params_calc_trust_tx_min_inline_mode(priv->mdev, &priv->channels.params,
+						   priv->dcbx_dp.trust_state);
+
+	err = mlx5_query_dscp2prio(priv->mdev, priv->dcbx_dp.dscp2prio);
+	if (err)
+		return err;
+
+	return 0;
+}
+
+#define MLX5E_BUFFER_CELL_SHIFT 7
+
+static u16 mlx5e_query_port_buffers_cell_size(struct mlx5e_priv *priv)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+	u32 out[MLX5_ST_SZ_DW(sbcam_reg)] = {};
+	u32 in[MLX5_ST_SZ_DW(sbcam_reg)] = {};
+
+	if (!MLX5_CAP_GEN(mdev, sbcam_reg))
+		return (1 << MLX5E_BUFFER_CELL_SHIFT);
+
+	if (mlx5_core_access_reg(mdev, in, sizeof(in), out, sizeof(out),
+				 MLX5_REG_SBCAM, 0, 0))
+		return (1 << MLX5E_BUFFER_CELL_SHIFT);
+
+	return MLX5_GET(sbcam_reg, out, cap_cell_size);
+}
+
+void mlx5e_dcbnl_initialize(struct mlx5e_priv *priv)
+{
+	struct mlx5e_dcbx *dcbx = &priv->dcbx;
+
+	mlx5e_trust_initialize(priv);
+
+	if (!MLX5_CAP_GEN(priv->mdev, qos))
+		return;
+
+	if (MLX5_CAP_GEN(priv->mdev, dcbx))
+		mlx5e_dcbnl_query_dcbx_mode(priv, &dcbx->mode);
+
+	priv->dcbx.cap = DCB_CAP_DCBX_VER_CEE |
+			 DCB_CAP_DCBX_VER_IEEE;
+	if (priv->dcbx.mode == MLX5E_DCBX_PARAM_VER_OPER_HOST)
+		priv->dcbx.cap |= DCB_CAP_DCBX_HOST;
+
+	priv->dcbx.port_buff_cell_sz = mlx5e_query_port_buffers_cell_size(priv);
+	priv->dcbx.manual_buffer = false;
+	priv->dcbx.cable_len = MLX5E_DEFAULT_CABLE_LEN;
+
+	mlx5e_ets_init(priv);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_dim.c b/drivers/net/ethernet/mellanox/mlx5/core/en_dim.c
new file mode 100644
index 000000000..ca9cfbf57
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_dim.c
@@ -0,0 +1,62 @@
+/*
+ * Copyright (c) 2016, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/dim.h>
+#include "en.h"
+
+static void
+mlx5e_complete_dim_work(struct dim *dim, struct dim_cq_moder moder,
+			struct mlx5_core_dev *mdev, struct mlx5_core_cq *mcq)
+{
+	mlx5_core_modify_cq_moderation(mdev, mcq, moder.usec, moder.pkts);
+	dim->state = DIM_START_MEASURE;
+}
+
+void mlx5e_rx_dim_work(struct work_struct *work)
+{
+	struct dim *dim = container_of(work, struct dim, work);
+	struct mlx5e_rq *rq = container_of(dim, struct mlx5e_rq, dim);
+	struct dim_cq_moder cur_moder =
+		net_dim_get_rx_moderation(dim->mode, dim->profile_ix);
+
+	mlx5e_complete_dim_work(dim, cur_moder, rq->mdev, &rq->cq.mcq);
+}
+
+void mlx5e_tx_dim_work(struct work_struct *work)
+{
+	struct dim *dim = container_of(work, struct dim, work);
+	struct mlx5e_txqsq *sq = container_of(dim, struct mlx5e_txqsq, dim);
+	struct dim_cq_moder cur_moder =
+		net_dim_get_tx_moderation(dim->mode, dim->profile_ix);
+
+	mlx5e_complete_dim_work(dim, cur_moder, sq->cq.mdev, &sq->cq.mcq);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c b/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c
new file mode 100644
index 000000000..ceeb23f47
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c
@@ -0,0 +1,2453 @@
+/*
+ * Copyright (c) 2015, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/ethtool_netlink.h>
+
+#include "en.h"
+#include "en/port.h"
+#include "en/params.h"
+#include "en/ptp.h"
+#include "lib/clock.h"
+#include "en/fs_ethtool.h"
+
+void mlx5e_ethtool_get_drvinfo(struct mlx5e_priv *priv,
+			       struct ethtool_drvinfo *drvinfo)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+	int count;
+
+	strscpy(drvinfo->driver, KBUILD_MODNAME, sizeof(drvinfo->driver));
+	count = snprintf(drvinfo->fw_version, sizeof(drvinfo->fw_version),
+			 "%d.%d.%04d (%.16s)", fw_rev_maj(mdev),
+			 fw_rev_min(mdev), fw_rev_sub(mdev), mdev->board_id);
+	if (count >= sizeof(drvinfo->fw_version))
+		snprintf(drvinfo->fw_version, sizeof(drvinfo->fw_version),
+			 "%d.%d.%04d", fw_rev_maj(mdev),
+			 fw_rev_min(mdev), fw_rev_sub(mdev));
+
+	strscpy(drvinfo->bus_info, dev_name(mdev->device),
+		sizeof(drvinfo->bus_info));
+}
+
+static void mlx5e_get_drvinfo(struct net_device *dev,
+			      struct ethtool_drvinfo *drvinfo)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+
+	mlx5e_ethtool_get_drvinfo(priv, drvinfo);
+}
+
+struct ptys2ethtool_config {
+	__ETHTOOL_DECLARE_LINK_MODE_MASK(supported);
+	__ETHTOOL_DECLARE_LINK_MODE_MASK(advertised);
+};
+
+static
+struct ptys2ethtool_config ptys2legacy_ethtool_table[MLX5E_LINK_MODES_NUMBER];
+static
+struct ptys2ethtool_config ptys2ext_ethtool_table[MLX5E_EXT_LINK_MODES_NUMBER];
+
+#define MLX5_BUILD_PTYS2ETHTOOL_CONFIG(reg_, table, ...)                  \
+	({                                                              \
+		struct ptys2ethtool_config *cfg;                        \
+		const unsigned int modes[] = { __VA_ARGS__ };           \
+		unsigned int i, bit, idx;                               \
+		cfg = &ptys2##table##_ethtool_table[reg_];		\
+		bitmap_zero(cfg->supported,                             \
+			    __ETHTOOL_LINK_MODE_MASK_NBITS);            \
+		bitmap_zero(cfg->advertised,                            \
+			    __ETHTOOL_LINK_MODE_MASK_NBITS);            \
+		for (i = 0 ; i < ARRAY_SIZE(modes) ; ++i) {             \
+			bit = modes[i] % 64;                            \
+			idx = modes[i] / 64;                            \
+			__set_bit(bit, &cfg->supported[idx]);           \
+			__set_bit(bit, &cfg->advertised[idx]);          \
+		}                                                       \
+	})
+
+void mlx5e_build_ptys2ethtool_map(void)
+{
+	memset(ptys2legacy_ethtool_table, 0, sizeof(ptys2legacy_ethtool_table));
+	memset(ptys2ext_ethtool_table, 0, sizeof(ptys2ext_ethtool_table));
+	MLX5_BUILD_PTYS2ETHTOOL_CONFIG(MLX5E_1000BASE_CX_SGMII, legacy,
+				       ETHTOOL_LINK_MODE_1000baseKX_Full_BIT);
+	MLX5_BUILD_PTYS2ETHTOOL_CONFIG(MLX5E_1000BASE_KX, legacy,
+				       ETHTOOL_LINK_MODE_1000baseKX_Full_BIT);
+	MLX5_BUILD_PTYS2ETHTOOL_CONFIG(MLX5E_10GBASE_CX4, legacy,
+				       ETHTOOL_LINK_MODE_10000baseKX4_Full_BIT);
+	MLX5_BUILD_PTYS2ETHTOOL_CONFIG(MLX5E_10GBASE_KX4, legacy,
+				       ETHTOOL_LINK_MODE_10000baseKX4_Full_BIT);
+	MLX5_BUILD_PTYS2ETHTOOL_CONFIG(MLX5E_10GBASE_KR, legacy,
+				       ETHTOOL_LINK_MODE_10000baseKR_Full_BIT);
+	MLX5_BUILD_PTYS2ETHTOOL_CONFIG(MLX5E_20GBASE_KR2, legacy,
+				       ETHTOOL_LINK_MODE_20000baseKR2_Full_BIT);
+	MLX5_BUILD_PTYS2ETHTOOL_CONFIG(MLX5E_40GBASE_CR4, legacy,
+				       ETHTOOL_LINK_MODE_40000baseCR4_Full_BIT);
+	MLX5_BUILD_PTYS2ETHTOOL_CONFIG(MLX5E_40GBASE_KR4, legacy,
+				       ETHTOOL_LINK_MODE_40000baseKR4_Full_BIT);
+	MLX5_BUILD_PTYS2ETHTOOL_CONFIG(MLX5E_56GBASE_R4, legacy,
+				       ETHTOOL_LINK_MODE_56000baseKR4_Full_BIT);
+	MLX5_BUILD_PTYS2ETHTOOL_CONFIG(MLX5E_10GBASE_CR, legacy,
+				       ETHTOOL_LINK_MODE_10000baseKR_Full_BIT);
+	MLX5_BUILD_PTYS2ETHTOOL_CONFIG(MLX5E_10GBASE_SR, legacy,
+				       ETHTOOL_LINK_MODE_10000baseKR_Full_BIT);
+	MLX5_BUILD_PTYS2ETHTOOL_CONFIG(MLX5E_10GBASE_ER, legacy,
+				       ETHTOOL_LINK_MODE_10000baseKR_Full_BIT);
+	MLX5_BUILD_PTYS2ETHTOOL_CONFIG(MLX5E_40GBASE_SR4, legacy,
+				       ETHTOOL_LINK_MODE_40000baseSR4_Full_BIT);
+	MLX5_BUILD_PTYS2ETHTOOL_CONFIG(MLX5E_40GBASE_LR4, legacy,
+				       ETHTOOL_LINK_MODE_40000baseLR4_Full_BIT);
+	MLX5_BUILD_PTYS2ETHTOOL_CONFIG(MLX5E_50GBASE_SR2, legacy,
+				       ETHTOOL_LINK_MODE_50000baseSR2_Full_BIT);
+	MLX5_BUILD_PTYS2ETHTOOL_CONFIG(MLX5E_100GBASE_CR4, legacy,
+				       ETHTOOL_LINK_MODE_100000baseCR4_Full_BIT);
+	MLX5_BUILD_PTYS2ETHTOOL_CONFIG(MLX5E_100GBASE_SR4, legacy,
+				       ETHTOOL_LINK_MODE_100000baseSR4_Full_BIT);
+	MLX5_BUILD_PTYS2ETHTOOL_CONFIG(MLX5E_100GBASE_KR4, legacy,
+				       ETHTOOL_LINK_MODE_100000baseKR4_Full_BIT);
+	MLX5_BUILD_PTYS2ETHTOOL_CONFIG(MLX5E_100GBASE_LR4, legacy,
+				       ETHTOOL_LINK_MODE_100000baseLR4_ER4_Full_BIT);
+	MLX5_BUILD_PTYS2ETHTOOL_CONFIG(MLX5E_10GBASE_T, legacy,
+				       ETHTOOL_LINK_MODE_10000baseT_Full_BIT);
+	MLX5_BUILD_PTYS2ETHTOOL_CONFIG(MLX5E_25GBASE_CR, legacy,
+				       ETHTOOL_LINK_MODE_25000baseCR_Full_BIT);
+	MLX5_BUILD_PTYS2ETHTOOL_CONFIG(MLX5E_25GBASE_KR, legacy,
+				       ETHTOOL_LINK_MODE_25000baseKR_Full_BIT);
+	MLX5_BUILD_PTYS2ETHTOOL_CONFIG(MLX5E_25GBASE_SR, legacy,
+				       ETHTOOL_LINK_MODE_25000baseSR_Full_BIT);
+	MLX5_BUILD_PTYS2ETHTOOL_CONFIG(MLX5E_50GBASE_CR2, legacy,
+				       ETHTOOL_LINK_MODE_50000baseCR2_Full_BIT);
+	MLX5_BUILD_PTYS2ETHTOOL_CONFIG(MLX5E_50GBASE_KR2, legacy,
+				       ETHTOOL_LINK_MODE_50000baseKR2_Full_BIT);
+	MLX5_BUILD_PTYS2ETHTOOL_CONFIG(MLX5E_SGMII_100M, ext,
+				       ETHTOOL_LINK_MODE_100baseT_Full_BIT);
+	MLX5_BUILD_PTYS2ETHTOOL_CONFIG(MLX5E_1000BASE_X_SGMII, ext,
+				       ETHTOOL_LINK_MODE_1000baseT_Full_BIT,
+				       ETHTOOL_LINK_MODE_1000baseKX_Full_BIT,
+				       ETHTOOL_LINK_MODE_1000baseX_Full_BIT);
+	MLX5_BUILD_PTYS2ETHTOOL_CONFIG(MLX5E_5GBASE_R, ext,
+				       ETHTOOL_LINK_MODE_5000baseT_Full_BIT);
+	MLX5_BUILD_PTYS2ETHTOOL_CONFIG(MLX5E_10GBASE_XFI_XAUI_1, ext,
+				       ETHTOOL_LINK_MODE_10000baseT_Full_BIT,
+				       ETHTOOL_LINK_MODE_10000baseKR_Full_BIT,
+				       ETHTOOL_LINK_MODE_10000baseR_FEC_BIT,
+				       ETHTOOL_LINK_MODE_10000baseCR_Full_BIT,
+				       ETHTOOL_LINK_MODE_10000baseSR_Full_BIT,
+				       ETHTOOL_LINK_MODE_10000baseLR_Full_BIT,
+				       ETHTOOL_LINK_MODE_10000baseER_Full_BIT);
+	MLX5_BUILD_PTYS2ETHTOOL_CONFIG(MLX5E_40GBASE_XLAUI_4_XLPPI_4, ext,
+				       ETHTOOL_LINK_MODE_40000baseKR4_Full_BIT,
+				       ETHTOOL_LINK_MODE_40000baseCR4_Full_BIT,
+				       ETHTOOL_LINK_MODE_40000baseSR4_Full_BIT,
+				       ETHTOOL_LINK_MODE_40000baseLR4_Full_BIT);
+	MLX5_BUILD_PTYS2ETHTOOL_CONFIG(MLX5E_25GAUI_1_25GBASE_CR_KR, ext,
+				       ETHTOOL_LINK_MODE_25000baseCR_Full_BIT,
+				       ETHTOOL_LINK_MODE_25000baseKR_Full_BIT,
+				       ETHTOOL_LINK_MODE_25000baseSR_Full_BIT);
+	MLX5_BUILD_PTYS2ETHTOOL_CONFIG(MLX5E_50GAUI_2_LAUI_2_50GBASE_CR2_KR2,
+				       ext,
+				       ETHTOOL_LINK_MODE_50000baseCR2_Full_BIT,
+				       ETHTOOL_LINK_MODE_50000baseKR2_Full_BIT,
+				       ETHTOOL_LINK_MODE_50000baseSR2_Full_BIT);
+	MLX5_BUILD_PTYS2ETHTOOL_CONFIG(MLX5E_50GAUI_1_LAUI_1_50GBASE_CR_KR, ext,
+				       ETHTOOL_LINK_MODE_50000baseKR_Full_BIT,
+				       ETHTOOL_LINK_MODE_50000baseSR_Full_BIT,
+				       ETHTOOL_LINK_MODE_50000baseCR_Full_BIT,
+				       ETHTOOL_LINK_MODE_50000baseLR_ER_FR_Full_BIT,
+				       ETHTOOL_LINK_MODE_50000baseDR_Full_BIT);
+	MLX5_BUILD_PTYS2ETHTOOL_CONFIG(MLX5E_CAUI_4_100GBASE_CR4_KR4, ext,
+				       ETHTOOL_LINK_MODE_100000baseKR4_Full_BIT,
+				       ETHTOOL_LINK_MODE_100000baseSR4_Full_BIT,
+				       ETHTOOL_LINK_MODE_100000baseCR4_Full_BIT,
+				       ETHTOOL_LINK_MODE_100000baseLR4_ER4_Full_BIT);
+	MLX5_BUILD_PTYS2ETHTOOL_CONFIG(MLX5E_100GAUI_2_100GBASE_CR2_KR2, ext,
+				       ETHTOOL_LINK_MODE_100000baseKR2_Full_BIT,
+				       ETHTOOL_LINK_MODE_100000baseSR2_Full_BIT,
+				       ETHTOOL_LINK_MODE_100000baseCR2_Full_BIT,
+				       ETHTOOL_LINK_MODE_100000baseLR2_ER2_FR2_Full_BIT,
+				       ETHTOOL_LINK_MODE_100000baseDR2_Full_BIT);
+	MLX5_BUILD_PTYS2ETHTOOL_CONFIG(MLX5E_200GAUI_4_200GBASE_CR4_KR4, ext,
+				       ETHTOOL_LINK_MODE_200000baseKR4_Full_BIT,
+				       ETHTOOL_LINK_MODE_200000baseSR4_Full_BIT,
+				       ETHTOOL_LINK_MODE_200000baseLR4_ER4_FR4_Full_BIT,
+				       ETHTOOL_LINK_MODE_200000baseDR4_Full_BIT,
+				       ETHTOOL_LINK_MODE_200000baseCR4_Full_BIT);
+	MLX5_BUILD_PTYS2ETHTOOL_CONFIG(MLX5E_100GAUI_1_100GBASE_CR_KR, ext,
+				       ETHTOOL_LINK_MODE_100000baseKR_Full_BIT,
+				       ETHTOOL_LINK_MODE_100000baseSR_Full_BIT,
+				       ETHTOOL_LINK_MODE_100000baseLR_ER_FR_Full_BIT,
+				       ETHTOOL_LINK_MODE_100000baseDR_Full_BIT,
+				       ETHTOOL_LINK_MODE_100000baseCR_Full_BIT);
+	MLX5_BUILD_PTYS2ETHTOOL_CONFIG(MLX5E_200GAUI_2_200GBASE_CR2_KR2, ext,
+				       ETHTOOL_LINK_MODE_200000baseKR2_Full_BIT,
+				       ETHTOOL_LINK_MODE_200000baseSR2_Full_BIT,
+				       ETHTOOL_LINK_MODE_200000baseLR2_ER2_FR2_Full_BIT,
+				       ETHTOOL_LINK_MODE_200000baseDR2_Full_BIT,
+				       ETHTOOL_LINK_MODE_200000baseCR2_Full_BIT);
+	MLX5_BUILD_PTYS2ETHTOOL_CONFIG(MLX5E_400GAUI_4_400GBASE_CR4_KR4, ext,
+				       ETHTOOL_LINK_MODE_400000baseKR4_Full_BIT,
+				       ETHTOOL_LINK_MODE_400000baseSR4_Full_BIT,
+				       ETHTOOL_LINK_MODE_400000baseLR4_ER4_FR4_Full_BIT,
+				       ETHTOOL_LINK_MODE_400000baseDR4_Full_BIT,
+				       ETHTOOL_LINK_MODE_400000baseCR4_Full_BIT);
+}
+
+static void mlx5e_ethtool_get_speed_arr(struct mlx5_core_dev *mdev,
+					struct ptys2ethtool_config **arr,
+					u32 *size)
+{
+	bool ext = mlx5e_ptys_ext_supported(mdev);
+
+	*arr = ext ? ptys2ext_ethtool_table : ptys2legacy_ethtool_table;
+	*size = ext ? ARRAY_SIZE(ptys2ext_ethtool_table) :
+		      ARRAY_SIZE(ptys2legacy_ethtool_table);
+}
+
+typedef int (*mlx5e_pflag_handler)(struct net_device *netdev, bool enable);
+
+struct pflag_desc {
+	char name[ETH_GSTRING_LEN];
+	mlx5e_pflag_handler handler;
+};
+
+static const struct pflag_desc mlx5e_priv_flags[MLX5E_NUM_PFLAGS];
+
+int mlx5e_ethtool_get_sset_count(struct mlx5e_priv *priv, int sset)
+{
+	switch (sset) {
+	case ETH_SS_STATS:
+		return mlx5e_stats_total_num(priv);
+	case ETH_SS_PRIV_FLAGS:
+		return MLX5E_NUM_PFLAGS;
+	case ETH_SS_TEST:
+		return mlx5e_self_test_num(priv);
+	default:
+		return -EOPNOTSUPP;
+	}
+}
+
+static int mlx5e_get_sset_count(struct net_device *dev, int sset)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+
+	return mlx5e_ethtool_get_sset_count(priv, sset);
+}
+
+void mlx5e_ethtool_get_strings(struct mlx5e_priv *priv, u32 stringset, u8 *data)
+{
+	int i;
+
+	switch (stringset) {
+	case ETH_SS_PRIV_FLAGS:
+		for (i = 0; i < MLX5E_NUM_PFLAGS; i++)
+			strcpy(data + i * ETH_GSTRING_LEN,
+			       mlx5e_priv_flags[i].name);
+		break;
+
+	case ETH_SS_TEST:
+		mlx5e_self_test_fill_strings(priv, data);
+		break;
+
+	case ETH_SS_STATS:
+		mlx5e_stats_fill_strings(priv, data);
+		break;
+	}
+}
+
+static void mlx5e_get_strings(struct net_device *dev, u32 stringset, u8 *data)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+
+	mlx5e_ethtool_get_strings(priv, stringset, data);
+}
+
+void mlx5e_ethtool_get_ethtool_stats(struct mlx5e_priv *priv,
+				     struct ethtool_stats *stats, u64 *data)
+{
+	int idx = 0;
+
+	mutex_lock(&priv->state_lock);
+	mlx5e_stats_update(priv);
+	mutex_unlock(&priv->state_lock);
+
+	mlx5e_stats_fill(priv, data, idx);
+}
+
+static void mlx5e_get_ethtool_stats(struct net_device *dev,
+				    struct ethtool_stats *stats,
+				    u64 *data)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+
+	mlx5e_ethtool_get_ethtool_stats(priv, stats, data);
+}
+
+void mlx5e_ethtool_get_ringparam(struct mlx5e_priv *priv,
+				 struct ethtool_ringparam *param,
+				 struct kernel_ethtool_ringparam *kernel_param)
+{
+	/* Limitation for regular RQ. XSK RQ may clamp the queue length in
+	 * mlx5e_mpwqe_get_log_rq_size.
+	 */
+	u8 max_log_mpwrq_pkts = mlx5e_mpwrq_max_log_rq_pkts(priv->mdev,
+							    PAGE_SHIFT,
+							    MLX5E_MPWRQ_UMR_MODE_ALIGNED);
+
+	param->rx_max_pending = 1 << min_t(u8, MLX5E_PARAMS_MAXIMUM_LOG_RQ_SIZE,
+					   max_log_mpwrq_pkts);
+	param->tx_max_pending = 1 << MLX5E_PARAMS_MAXIMUM_LOG_SQ_SIZE;
+	param->rx_pending     = 1 << priv->channels.params.log_rq_mtu_frames;
+	param->tx_pending     = 1 << priv->channels.params.log_sq_size;
+
+	kernel_param->tcp_data_split =
+		(priv->channels.params.packet_merge.type == MLX5E_PACKET_MERGE_SHAMPO) ?
+		ETHTOOL_TCP_DATA_SPLIT_ENABLED :
+		ETHTOOL_TCP_DATA_SPLIT_DISABLED;
+}
+
+static void mlx5e_get_ringparam(struct net_device *dev,
+				struct ethtool_ringparam *param,
+				struct kernel_ethtool_ringparam *kernel_param,
+				struct netlink_ext_ack *extack)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+
+	mlx5e_ethtool_get_ringparam(priv, param, kernel_param);
+}
+
+int mlx5e_ethtool_set_ringparam(struct mlx5e_priv *priv,
+				struct ethtool_ringparam *param)
+{
+	struct mlx5e_params new_params;
+	u8 log_rq_size;
+	u8 log_sq_size;
+	int err = 0;
+
+	if (param->rx_jumbo_pending) {
+		netdev_info(priv->netdev, "%s: rx_jumbo_pending not supported\n",
+			    __func__);
+		return -EINVAL;
+	}
+	if (param->rx_mini_pending) {
+		netdev_info(priv->netdev, "%s: rx_mini_pending not supported\n",
+			    __func__);
+		return -EINVAL;
+	}
+
+	if (param->rx_pending < (1 << MLX5E_PARAMS_MINIMUM_LOG_RQ_SIZE)) {
+		netdev_info(priv->netdev, "%s: rx_pending (%d) < min (%d)\n",
+			    __func__, param->rx_pending,
+			    1 << MLX5E_PARAMS_MINIMUM_LOG_RQ_SIZE);
+		return -EINVAL;
+	}
+
+	if (param->tx_pending < (1 << MLX5E_PARAMS_MINIMUM_LOG_SQ_SIZE)) {
+		netdev_info(priv->netdev, "%s: tx_pending (%d) < min (%d)\n",
+			    __func__, param->tx_pending,
+			    1 << MLX5E_PARAMS_MINIMUM_LOG_SQ_SIZE);
+		return -EINVAL;
+	}
+
+	log_rq_size = order_base_2(param->rx_pending);
+	log_sq_size = order_base_2(param->tx_pending);
+
+	if (log_rq_size == priv->channels.params.log_rq_mtu_frames &&
+	    log_sq_size == priv->channels.params.log_sq_size)
+		return 0;
+
+	mutex_lock(&priv->state_lock);
+
+	new_params = priv->channels.params;
+	new_params.log_rq_mtu_frames = log_rq_size;
+	new_params.log_sq_size = log_sq_size;
+
+	err = mlx5e_validate_params(priv->mdev, &new_params);
+	if (err)
+		goto unlock;
+
+	err = mlx5e_safe_switch_params(priv, &new_params, NULL, NULL, true);
+
+unlock:
+	mutex_unlock(&priv->state_lock);
+
+	return err;
+}
+
+static int mlx5e_set_ringparam(struct net_device *dev,
+			       struct ethtool_ringparam *param,
+			       struct kernel_ethtool_ringparam *kernel_param,
+			       struct netlink_ext_ack *extack)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+
+	return mlx5e_ethtool_set_ringparam(priv, param);
+}
+
+void mlx5e_ethtool_get_channels(struct mlx5e_priv *priv,
+				struct ethtool_channels *ch)
+{
+	mutex_lock(&priv->state_lock);
+	ch->max_combined   = priv->max_nch;
+	ch->combined_count = priv->channels.params.num_channels;
+	mutex_unlock(&priv->state_lock);
+}
+
+static void mlx5e_get_channels(struct net_device *dev,
+			       struct ethtool_channels *ch)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+
+	mlx5e_ethtool_get_channels(priv, ch);
+}
+
+int mlx5e_ethtool_set_channels(struct mlx5e_priv *priv,
+			       struct ethtool_channels *ch)
+{
+	struct mlx5e_params *cur_params = &priv->channels.params;
+	unsigned int count = ch->combined_count;
+	struct mlx5e_params new_params;
+	bool arfs_enabled;
+	int rss_cnt;
+	bool opened;
+	int err = 0;
+
+	if (!count) {
+		netdev_info(priv->netdev, "%s: combined_count=0 not supported\n",
+			    __func__);
+		return -EINVAL;
+	}
+
+	if (cur_params->num_channels == count)
+		return 0;
+
+	mutex_lock(&priv->state_lock);
+
+	/* Don't allow changing the number of channels if HTB offload is active,
+	 * because the numeration of the QoS SQs will change, while per-queue
+	 * qdiscs are attached.
+	 */
+	if (mlx5e_selq_is_htb_enabled(&priv->selq)) {
+		err = -EINVAL;
+		netdev_err(priv->netdev, "%s: HTB offload is active, cannot change the number of channels\n",
+			   __func__);
+		goto out;
+	}
+
+	/* Don't allow changing the number of channels if non-default RSS contexts exist,
+	 * the kernel doesn't protect against set_channels operations that break them.
+	 */
+	rss_cnt = mlx5e_rx_res_rss_cnt(priv->rx_res) - 1;
+	if (rss_cnt) {
+		err = -EINVAL;
+		netdev_err(priv->netdev, "%s: Non-default RSS contexts exist (%d), cannot change the number of channels\n",
+			   __func__, rss_cnt);
+		goto out;
+	}
+
+	/* Don't allow changing the number of channels if MQPRIO mode channel offload is active,
+	 * because it defines a partition over the channels queues.
+	 */
+	if (cur_params->mqprio.mode == TC_MQPRIO_MODE_CHANNEL) {
+		err = -EINVAL;
+		netdev_err(priv->netdev, "%s: MQPRIO mode channel offload is active, cannot change the number of channels\n",
+			   __func__);
+		goto out;
+	}
+
+	new_params = *cur_params;
+	new_params.num_channels = count;
+
+	opened = test_bit(MLX5E_STATE_OPENED, &priv->state);
+
+	arfs_enabled = opened && (priv->netdev->features & NETIF_F_NTUPLE);
+	if (arfs_enabled)
+		mlx5e_arfs_disable(priv->fs);
+
+	/* Switch to new channels, set new parameters and close old ones */
+	err = mlx5e_safe_switch_params(priv, &new_params,
+				       mlx5e_num_channels_changed_ctx, NULL, true);
+
+	if (arfs_enabled) {
+		int err2 = mlx5e_arfs_enable(priv->fs);
+
+		if (err2)
+			netdev_err(priv->netdev, "%s: mlx5e_arfs_enable failed: %d\n",
+				   __func__, err2);
+	}
+
+out:
+	mutex_unlock(&priv->state_lock);
+
+	return err;
+}
+
+static int mlx5e_set_channels(struct net_device *dev,
+			      struct ethtool_channels *ch)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+
+	return mlx5e_ethtool_set_channels(priv, ch);
+}
+
+int mlx5e_ethtool_get_coalesce(struct mlx5e_priv *priv,
+			       struct ethtool_coalesce *coal,
+			       struct kernel_ethtool_coalesce *kernel_coal)
+{
+	struct dim_cq_moder *rx_moder, *tx_moder;
+
+	if (!MLX5_CAP_GEN(priv->mdev, cq_moderation))
+		return -EOPNOTSUPP;
+
+	rx_moder = &priv->channels.params.rx_cq_moderation;
+	coal->rx_coalesce_usecs		= rx_moder->usec;
+	coal->rx_max_coalesced_frames	= rx_moder->pkts;
+	coal->use_adaptive_rx_coalesce	= priv->channels.params.rx_dim_enabled;
+
+	tx_moder = &priv->channels.params.tx_cq_moderation;
+	coal->tx_coalesce_usecs		= tx_moder->usec;
+	coal->tx_max_coalesced_frames	= tx_moder->pkts;
+	coal->use_adaptive_tx_coalesce	= priv->channels.params.tx_dim_enabled;
+
+	kernel_coal->use_cqe_mode_rx =
+		MLX5E_GET_PFLAG(&priv->channels.params, MLX5E_PFLAG_RX_CQE_BASED_MODER);
+	kernel_coal->use_cqe_mode_tx =
+		MLX5E_GET_PFLAG(&priv->channels.params, MLX5E_PFLAG_TX_CQE_BASED_MODER);
+
+	return 0;
+}
+
+static int mlx5e_get_coalesce(struct net_device *netdev,
+			      struct ethtool_coalesce *coal,
+			      struct kernel_ethtool_coalesce *kernel_coal,
+			      struct netlink_ext_ack *extack)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+
+	return mlx5e_ethtool_get_coalesce(priv, coal, kernel_coal);
+}
+
+#define MLX5E_MAX_COAL_TIME		MLX5_MAX_CQ_PERIOD
+#define MLX5E_MAX_COAL_FRAMES		MLX5_MAX_CQ_COUNT
+
+static void
+mlx5e_set_priv_channels_tx_coalesce(struct mlx5e_priv *priv, struct ethtool_coalesce *coal)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+	int tc;
+	int i;
+
+	for (i = 0; i < priv->channels.num; ++i) {
+		struct mlx5e_channel *c = priv->channels.c[i];
+
+		for (tc = 0; tc < c->num_tc; tc++) {
+			mlx5_core_modify_cq_moderation(mdev,
+						&c->sq[tc].cq.mcq,
+						coal->tx_coalesce_usecs,
+						coal->tx_max_coalesced_frames);
+		}
+	}
+}
+
+static void
+mlx5e_set_priv_channels_rx_coalesce(struct mlx5e_priv *priv, struct ethtool_coalesce *coal)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+	int i;
+
+	for (i = 0; i < priv->channels.num; ++i) {
+		struct mlx5e_channel *c = priv->channels.c[i];
+
+		mlx5_core_modify_cq_moderation(mdev, &c->rq.cq.mcq,
+					       coal->rx_coalesce_usecs,
+					       coal->rx_max_coalesced_frames);
+	}
+}
+
+/* convert a boolean value of cq_mode to mlx5 period mode
+ * true  : MLX5_CQ_PERIOD_MODE_START_FROM_CQE
+ * false : MLX5_CQ_PERIOD_MODE_START_FROM_EQE
+ */
+static int cqe_mode_to_period_mode(bool val)
+{
+	return val ? MLX5_CQ_PERIOD_MODE_START_FROM_CQE : MLX5_CQ_PERIOD_MODE_START_FROM_EQE;
+}
+
+int mlx5e_ethtool_set_coalesce(struct mlx5e_priv *priv,
+			       struct ethtool_coalesce *coal,
+			       struct kernel_ethtool_coalesce *kernel_coal,
+			       struct netlink_ext_ack *extack)
+{
+	struct dim_cq_moder *rx_moder, *tx_moder;
+	struct mlx5_core_dev *mdev = priv->mdev;
+	struct mlx5e_params new_params;
+	bool reset_rx, reset_tx;
+	bool reset = true;
+	u8 cq_period_mode;
+	int err = 0;
+
+	if (!MLX5_CAP_GEN(mdev, cq_moderation))
+		return -EOPNOTSUPP;
+
+	if (coal->tx_coalesce_usecs > MLX5E_MAX_COAL_TIME ||
+	    coal->rx_coalesce_usecs > MLX5E_MAX_COAL_TIME) {
+		netdev_info(priv->netdev, "%s: maximum coalesce time supported is %lu usecs\n",
+			    __func__, MLX5E_MAX_COAL_TIME);
+		return -ERANGE;
+	}
+
+	if (coal->tx_max_coalesced_frames > MLX5E_MAX_COAL_FRAMES ||
+	    coal->rx_max_coalesced_frames > MLX5E_MAX_COAL_FRAMES) {
+		netdev_info(priv->netdev, "%s: maximum coalesced frames supported is %lu\n",
+			    __func__, MLX5E_MAX_COAL_FRAMES);
+		return -ERANGE;
+	}
+
+	if ((kernel_coal->use_cqe_mode_rx || kernel_coal->use_cqe_mode_tx) &&
+	    !MLX5_CAP_GEN(priv->mdev, cq_period_start_from_cqe)) {
+		NL_SET_ERR_MSG_MOD(extack, "cqe_mode_rx/tx is not supported on this device");
+		return -EOPNOTSUPP;
+	}
+
+	mutex_lock(&priv->state_lock);
+	new_params = priv->channels.params;
+
+	rx_moder          = &new_params.rx_cq_moderation;
+	rx_moder->usec    = coal->rx_coalesce_usecs;
+	rx_moder->pkts    = coal->rx_max_coalesced_frames;
+	new_params.rx_dim_enabled = !!coal->use_adaptive_rx_coalesce;
+
+	tx_moder          = &new_params.tx_cq_moderation;
+	tx_moder->usec    = coal->tx_coalesce_usecs;
+	tx_moder->pkts    = coal->tx_max_coalesced_frames;
+	new_params.tx_dim_enabled = !!coal->use_adaptive_tx_coalesce;
+
+	reset_rx = !!coal->use_adaptive_rx_coalesce != priv->channels.params.rx_dim_enabled;
+	reset_tx = !!coal->use_adaptive_tx_coalesce != priv->channels.params.tx_dim_enabled;
+
+	cq_period_mode = cqe_mode_to_period_mode(kernel_coal->use_cqe_mode_rx);
+	if (cq_period_mode != rx_moder->cq_period_mode) {
+		mlx5e_set_rx_cq_mode_params(&new_params, cq_period_mode);
+		reset_rx = true;
+	}
+
+	cq_period_mode = cqe_mode_to_period_mode(kernel_coal->use_cqe_mode_tx);
+	if (cq_period_mode != tx_moder->cq_period_mode) {
+		mlx5e_set_tx_cq_mode_params(&new_params, cq_period_mode);
+		reset_tx = true;
+	}
+
+	if (reset_rx) {
+		u8 mode = MLX5E_GET_PFLAG(&new_params,
+					  MLX5E_PFLAG_RX_CQE_BASED_MODER);
+
+		mlx5e_reset_rx_moderation(&new_params, mode);
+	}
+	if (reset_tx) {
+		u8 mode = MLX5E_GET_PFLAG(&new_params,
+					  MLX5E_PFLAG_TX_CQE_BASED_MODER);
+
+		mlx5e_reset_tx_moderation(&new_params, mode);
+	}
+
+	/* If DIM state hasn't changed, it's possible to modify interrupt
+	 * moderation parameters on the fly, even if the channels are open.
+	 */
+	if (!reset_rx && !reset_tx && test_bit(MLX5E_STATE_OPENED, &priv->state)) {
+		if (!coal->use_adaptive_rx_coalesce)
+			mlx5e_set_priv_channels_rx_coalesce(priv, coal);
+		if (!coal->use_adaptive_tx_coalesce)
+			mlx5e_set_priv_channels_tx_coalesce(priv, coal);
+		reset = false;
+	}
+
+	err = mlx5e_safe_switch_params(priv, &new_params, NULL, NULL, reset);
+
+	mutex_unlock(&priv->state_lock);
+	return err;
+}
+
+static int mlx5e_set_coalesce(struct net_device *netdev,
+			      struct ethtool_coalesce *coal,
+			      struct kernel_ethtool_coalesce *kernel_coal,
+			      struct netlink_ext_ack *extack)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+
+	return mlx5e_ethtool_set_coalesce(priv, coal, kernel_coal, extack);
+}
+
+static void ptys2ethtool_supported_link(struct mlx5_core_dev *mdev,
+					unsigned long *supported_modes,
+					u32 eth_proto_cap)
+{
+	unsigned long proto_cap = eth_proto_cap;
+	struct ptys2ethtool_config *table;
+	u32 max_size;
+	int proto;
+
+	mlx5e_ethtool_get_speed_arr(mdev, &table, &max_size);
+	for_each_set_bit(proto, &proto_cap, max_size)
+		bitmap_or(supported_modes, supported_modes,
+			  table[proto].supported,
+			  __ETHTOOL_LINK_MODE_MASK_NBITS);
+}
+
+static void ptys2ethtool_adver_link(unsigned long *advertising_modes,
+				    u32 eth_proto_cap, bool ext)
+{
+	unsigned long proto_cap = eth_proto_cap;
+	struct ptys2ethtool_config *table;
+	u32 max_size;
+	int proto;
+
+	table = ext ? ptys2ext_ethtool_table : ptys2legacy_ethtool_table;
+	max_size = ext ? ARRAY_SIZE(ptys2ext_ethtool_table) :
+			 ARRAY_SIZE(ptys2legacy_ethtool_table);
+
+	for_each_set_bit(proto, &proto_cap, max_size)
+		bitmap_or(advertising_modes, advertising_modes,
+			  table[proto].advertised,
+			  __ETHTOOL_LINK_MODE_MASK_NBITS);
+}
+
+static const u32 pplm_fec_2_ethtool[] = {
+	[MLX5E_FEC_NOFEC] = ETHTOOL_FEC_OFF,
+	[MLX5E_FEC_FIRECODE] = ETHTOOL_FEC_BASER,
+	[MLX5E_FEC_RS_528_514] = ETHTOOL_FEC_RS,
+	[MLX5E_FEC_RS_544_514] = ETHTOOL_FEC_RS,
+	[MLX5E_FEC_LLRS_272_257_1] = ETHTOOL_FEC_LLRS,
+};
+
+static u32 pplm2ethtool_fec(u_long fec_mode, unsigned long size)
+{
+	int mode = 0;
+
+	if (!fec_mode)
+		return ETHTOOL_FEC_AUTO;
+
+	mode = find_first_bit(&fec_mode, size);
+
+	if (mode < ARRAY_SIZE(pplm_fec_2_ethtool))
+		return pplm_fec_2_ethtool[mode];
+
+	return 0;
+}
+
+#define MLX5E_ADVERTISE_SUPPORTED_FEC(mlx5_fec, ethtool_fec)		\
+	do {								\
+		if (mlx5e_fec_in_caps(dev, 1 << (mlx5_fec)))		\
+			__set_bit(ethtool_fec,				\
+				  link_ksettings->link_modes.supported);\
+	} while (0)
+
+static const u32 pplm_fec_2_ethtool_linkmodes[] = {
+	[MLX5E_FEC_NOFEC] = ETHTOOL_LINK_MODE_FEC_NONE_BIT,
+	[MLX5E_FEC_FIRECODE] = ETHTOOL_LINK_MODE_FEC_BASER_BIT,
+	[MLX5E_FEC_RS_528_514] = ETHTOOL_LINK_MODE_FEC_RS_BIT,
+	[MLX5E_FEC_RS_544_514] = ETHTOOL_LINK_MODE_FEC_RS_BIT,
+	[MLX5E_FEC_LLRS_272_257_1] = ETHTOOL_LINK_MODE_FEC_LLRS_BIT,
+};
+
+static int get_fec_supported_advertised(struct mlx5_core_dev *dev,
+					struct ethtool_link_ksettings *link_ksettings)
+{
+	unsigned long active_fec_long;
+	u32 active_fec;
+	u32 bitn;
+	int err;
+
+	err = mlx5e_get_fec_mode(dev, &active_fec, NULL);
+	if (err)
+		return (err == -EOPNOTSUPP) ? 0 : err;
+
+	MLX5E_ADVERTISE_SUPPORTED_FEC(MLX5E_FEC_NOFEC,
+				      ETHTOOL_LINK_MODE_FEC_NONE_BIT);
+	MLX5E_ADVERTISE_SUPPORTED_FEC(MLX5E_FEC_FIRECODE,
+				      ETHTOOL_LINK_MODE_FEC_BASER_BIT);
+	MLX5E_ADVERTISE_SUPPORTED_FEC(MLX5E_FEC_RS_528_514,
+				      ETHTOOL_LINK_MODE_FEC_RS_BIT);
+	MLX5E_ADVERTISE_SUPPORTED_FEC(MLX5E_FEC_LLRS_272_257_1,
+				      ETHTOOL_LINK_MODE_FEC_LLRS_BIT);
+
+	active_fec_long = active_fec;
+	/* active fec is a bit set, find out which bit is set and
+	 * advertise the corresponding ethtool bit
+	 */
+	bitn = find_first_bit(&active_fec_long, sizeof(active_fec_long) * BITS_PER_BYTE);
+	if (bitn < ARRAY_SIZE(pplm_fec_2_ethtool_linkmodes))
+		__set_bit(pplm_fec_2_ethtool_linkmodes[bitn],
+			  link_ksettings->link_modes.advertising);
+
+	return 0;
+}
+
+static void ptys2ethtool_supported_advertised_port(struct mlx5_core_dev *mdev,
+						   struct ethtool_link_ksettings *link_ksettings,
+						   u32 eth_proto_cap, u8 connector_type)
+{
+	if (!MLX5_CAP_PCAM_FEATURE(mdev, ptys_connector_type)) {
+		if (eth_proto_cap & (MLX5E_PROT_MASK(MLX5E_10GBASE_CR)
+				   | MLX5E_PROT_MASK(MLX5E_10GBASE_SR)
+				   | MLX5E_PROT_MASK(MLX5E_40GBASE_CR4)
+				   | MLX5E_PROT_MASK(MLX5E_40GBASE_SR4)
+				   | MLX5E_PROT_MASK(MLX5E_100GBASE_SR4)
+				   | MLX5E_PROT_MASK(MLX5E_1000BASE_CX_SGMII))) {
+			ethtool_link_ksettings_add_link_mode(link_ksettings,
+							     supported,
+							     FIBRE);
+			ethtool_link_ksettings_add_link_mode(link_ksettings,
+							     advertising,
+							     FIBRE);
+		}
+
+		if (eth_proto_cap & (MLX5E_PROT_MASK(MLX5E_100GBASE_KR4)
+				   | MLX5E_PROT_MASK(MLX5E_40GBASE_KR4)
+				   | MLX5E_PROT_MASK(MLX5E_10GBASE_KR)
+				   | MLX5E_PROT_MASK(MLX5E_10GBASE_KX4)
+				   | MLX5E_PROT_MASK(MLX5E_1000BASE_KX))) {
+			ethtool_link_ksettings_add_link_mode(link_ksettings,
+							     supported,
+							     Backplane);
+			ethtool_link_ksettings_add_link_mode(link_ksettings,
+							     advertising,
+							     Backplane);
+		}
+		return;
+	}
+
+	switch (connector_type) {
+	case MLX5E_PORT_TP:
+		ethtool_link_ksettings_add_link_mode(link_ksettings,
+						     supported, TP);
+		ethtool_link_ksettings_add_link_mode(link_ksettings,
+						     advertising, TP);
+		break;
+	case MLX5E_PORT_AUI:
+		ethtool_link_ksettings_add_link_mode(link_ksettings,
+						     supported, AUI);
+		ethtool_link_ksettings_add_link_mode(link_ksettings,
+						     advertising, AUI);
+		break;
+	case MLX5E_PORT_BNC:
+		ethtool_link_ksettings_add_link_mode(link_ksettings,
+						     supported, BNC);
+		ethtool_link_ksettings_add_link_mode(link_ksettings,
+						     advertising, BNC);
+		break;
+	case MLX5E_PORT_MII:
+		ethtool_link_ksettings_add_link_mode(link_ksettings,
+						     supported, MII);
+		ethtool_link_ksettings_add_link_mode(link_ksettings,
+						     advertising, MII);
+		break;
+	case MLX5E_PORT_FIBRE:
+		ethtool_link_ksettings_add_link_mode(link_ksettings,
+						     supported, FIBRE);
+		ethtool_link_ksettings_add_link_mode(link_ksettings,
+						     advertising, FIBRE);
+		break;
+	case MLX5E_PORT_DA:
+		ethtool_link_ksettings_add_link_mode(link_ksettings,
+						     supported, Backplane);
+		ethtool_link_ksettings_add_link_mode(link_ksettings,
+						     advertising, Backplane);
+		break;
+	case MLX5E_PORT_NONE:
+	case MLX5E_PORT_OTHER:
+	default:
+		break;
+	}
+}
+
+static void get_speed_duplex(struct net_device *netdev,
+			     u32 eth_proto_oper, bool force_legacy,
+			     u16 data_rate_oper,
+			     struct ethtool_link_ksettings *link_ksettings)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	u32 speed = SPEED_UNKNOWN;
+	u8 duplex = DUPLEX_UNKNOWN;
+
+	if (!netif_carrier_ok(netdev))
+		goto out;
+
+	speed = mlx5e_port_ptys2speed(priv->mdev, eth_proto_oper, force_legacy);
+	if (!speed) {
+		if (data_rate_oper)
+			speed = 100 * data_rate_oper;
+		else
+			speed = SPEED_UNKNOWN;
+		goto out;
+	}
+
+	duplex = DUPLEX_FULL;
+
+out:
+	link_ksettings->base.speed = speed;
+	link_ksettings->base.duplex = duplex;
+}
+
+static void get_supported(struct mlx5_core_dev *mdev, u32 eth_proto_cap,
+			  struct ethtool_link_ksettings *link_ksettings)
+{
+	unsigned long *supported = link_ksettings->link_modes.supported;
+	ptys2ethtool_supported_link(mdev, supported, eth_proto_cap);
+
+	ethtool_link_ksettings_add_link_mode(link_ksettings, supported, Pause);
+}
+
+static void get_advertising(u32 eth_proto_cap, u8 tx_pause, u8 rx_pause,
+			    struct ethtool_link_ksettings *link_ksettings,
+			    bool ext)
+{
+	unsigned long *advertising = link_ksettings->link_modes.advertising;
+	ptys2ethtool_adver_link(advertising, eth_proto_cap, ext);
+
+	if (rx_pause)
+		ethtool_link_ksettings_add_link_mode(link_ksettings, advertising, Pause);
+	if (tx_pause ^ rx_pause)
+		ethtool_link_ksettings_add_link_mode(link_ksettings, advertising, Asym_Pause);
+}
+
+static int ptys2connector_type[MLX5E_CONNECTOR_TYPE_NUMBER] = {
+		[MLX5E_PORT_UNKNOWN]            = PORT_OTHER,
+		[MLX5E_PORT_NONE]               = PORT_NONE,
+		[MLX5E_PORT_TP]                 = PORT_TP,
+		[MLX5E_PORT_AUI]                = PORT_AUI,
+		[MLX5E_PORT_BNC]                = PORT_BNC,
+		[MLX5E_PORT_MII]                = PORT_MII,
+		[MLX5E_PORT_FIBRE]              = PORT_FIBRE,
+		[MLX5E_PORT_DA]                 = PORT_DA,
+		[MLX5E_PORT_OTHER]              = PORT_OTHER,
+	};
+
+static u8 get_connector_port(struct mlx5_core_dev *mdev, u32 eth_proto, u8 connector_type)
+{
+	if (MLX5_CAP_PCAM_FEATURE(mdev, ptys_connector_type))
+		return ptys2connector_type[connector_type];
+
+	if (eth_proto &
+	    (MLX5E_PROT_MASK(MLX5E_10GBASE_SR)   |
+	     MLX5E_PROT_MASK(MLX5E_40GBASE_SR4)  |
+	     MLX5E_PROT_MASK(MLX5E_100GBASE_SR4) |
+	     MLX5E_PROT_MASK(MLX5E_1000BASE_CX_SGMII))) {
+		return PORT_FIBRE;
+	}
+
+	if (eth_proto &
+	    (MLX5E_PROT_MASK(MLX5E_40GBASE_CR4) |
+	     MLX5E_PROT_MASK(MLX5E_10GBASE_CR)  |
+	     MLX5E_PROT_MASK(MLX5E_100GBASE_CR4))) {
+		return PORT_DA;
+	}
+
+	if (eth_proto &
+	    (MLX5E_PROT_MASK(MLX5E_10GBASE_KX4) |
+	     MLX5E_PROT_MASK(MLX5E_10GBASE_KR)  |
+	     MLX5E_PROT_MASK(MLX5E_40GBASE_KR4) |
+	     MLX5E_PROT_MASK(MLX5E_100GBASE_KR4))) {
+		return PORT_NONE;
+	}
+
+	return PORT_OTHER;
+}
+
+static void get_lp_advertising(struct mlx5_core_dev *mdev, u32 eth_proto_lp,
+			       struct ethtool_link_ksettings *link_ksettings)
+{
+	unsigned long *lp_advertising = link_ksettings->link_modes.lp_advertising;
+	bool ext = mlx5e_ptys_ext_supported(mdev);
+
+	ptys2ethtool_adver_link(lp_advertising, eth_proto_lp, ext);
+}
+
+int mlx5e_ethtool_get_link_ksettings(struct mlx5e_priv *priv,
+				     struct ethtool_link_ksettings *link_ksettings)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+	u32 out[MLX5_ST_SZ_DW(ptys_reg)] = {};
+	u32 eth_proto_admin;
+	u8 an_disable_admin;
+	u16 data_rate_oper;
+	u32 eth_proto_oper;
+	u32 eth_proto_cap;
+	u8 connector_type;
+	u32 rx_pause = 0;
+	u32 tx_pause = 0;
+	u32 eth_proto_lp;
+	bool admin_ext;
+	u8 an_status;
+	bool ext;
+	int err;
+
+	err = mlx5_query_port_ptys(mdev, out, sizeof(out), MLX5_PTYS_EN, 1);
+	if (err) {
+		netdev_err(priv->netdev, "%s: query port ptys failed: %d\n",
+			   __func__, err);
+		goto err_query_regs;
+	}
+	ext = !!MLX5_GET_ETH_PROTO(ptys_reg, out, true, eth_proto_capability);
+	eth_proto_cap    = MLX5_GET_ETH_PROTO(ptys_reg, out, ext,
+					      eth_proto_capability);
+	eth_proto_admin  = MLX5_GET_ETH_PROTO(ptys_reg, out, ext,
+					      eth_proto_admin);
+	/* Fields: eth_proto_admin and ext_eth_proto_admin  are
+	 * mutually exclusive. Hence try reading legacy advertising
+	 * when extended advertising is zero.
+	 * admin_ext indicates which proto_admin (ext vs. legacy)
+	 * should be read and interpreted
+	 */
+	admin_ext = ext;
+	if (ext && !eth_proto_admin) {
+		eth_proto_admin  = MLX5_GET_ETH_PROTO(ptys_reg, out, false,
+						      eth_proto_admin);
+		admin_ext = false;
+	}
+
+	eth_proto_oper   = MLX5_GET_ETH_PROTO(ptys_reg, out, admin_ext,
+					      eth_proto_oper);
+	eth_proto_lp	    = MLX5_GET(ptys_reg, out, eth_proto_lp_advertise);
+	an_disable_admin    = MLX5_GET(ptys_reg, out, an_disable_admin);
+	an_status	    = MLX5_GET(ptys_reg, out, an_status);
+	connector_type	    = MLX5_GET(ptys_reg, out, connector_type);
+	data_rate_oper	    = MLX5_GET(ptys_reg, out, data_rate_oper);
+
+	mlx5_query_port_pause(mdev, &rx_pause, &tx_pause);
+
+	ethtool_link_ksettings_zero_link_mode(link_ksettings, supported);
+	ethtool_link_ksettings_zero_link_mode(link_ksettings, advertising);
+
+	get_supported(mdev, eth_proto_cap, link_ksettings);
+	get_advertising(eth_proto_admin, tx_pause, rx_pause, link_ksettings,
+			admin_ext);
+	get_speed_duplex(priv->netdev, eth_proto_oper, !admin_ext,
+			 data_rate_oper, link_ksettings);
+
+	eth_proto_oper = eth_proto_oper ? eth_proto_oper : eth_proto_cap;
+	connector_type = connector_type < MLX5E_CONNECTOR_TYPE_NUMBER ?
+			 connector_type : MLX5E_PORT_UNKNOWN;
+	link_ksettings->base.port = get_connector_port(mdev, eth_proto_oper, connector_type);
+	ptys2ethtool_supported_advertised_port(mdev, link_ksettings, eth_proto_admin,
+					       connector_type);
+	get_lp_advertising(mdev, eth_proto_lp, link_ksettings);
+
+	if (an_status == MLX5_AN_COMPLETE)
+		ethtool_link_ksettings_add_link_mode(link_ksettings,
+						     lp_advertising, Autoneg);
+
+	link_ksettings->base.autoneg = an_disable_admin ? AUTONEG_DISABLE :
+							  AUTONEG_ENABLE;
+	ethtool_link_ksettings_add_link_mode(link_ksettings, supported,
+					     Autoneg);
+
+	err = get_fec_supported_advertised(mdev, link_ksettings);
+	if (err) {
+		netdev_dbg(priv->netdev, "%s: FEC caps query failed: %d\n",
+			   __func__, err);
+		err = 0; /* don't fail caps query because of FEC error */
+	}
+
+	if (!an_disable_admin)
+		ethtool_link_ksettings_add_link_mode(link_ksettings,
+						     advertising, Autoneg);
+
+err_query_regs:
+	return err;
+}
+
+static int mlx5e_get_link_ksettings(struct net_device *netdev,
+				    struct ethtool_link_ksettings *link_ksettings)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+
+	return mlx5e_ethtool_get_link_ksettings(priv, link_ksettings);
+}
+
+static int mlx5e_speed_validate(struct net_device *netdev, bool ext,
+				const unsigned long link_modes, u8 autoneg)
+{
+	/* Extended link-mode has no speed limitations. */
+	if (ext)
+		return 0;
+
+	if ((link_modes & MLX5E_PROT_MASK(MLX5E_56GBASE_R4)) &&
+	    autoneg != AUTONEG_ENABLE) {
+		netdev_err(netdev, "%s: 56G link speed requires autoneg enabled\n",
+			   __func__);
+		return -EINVAL;
+	}
+	return 0;
+}
+
+static u32 mlx5e_ethtool2ptys_adver_link(const unsigned long *link_modes)
+{
+	u32 i, ptys_modes = 0;
+
+	for (i = 0; i < MLX5E_LINK_MODES_NUMBER; ++i) {
+		if (*ptys2legacy_ethtool_table[i].advertised == 0)
+			continue;
+		if (bitmap_intersects(ptys2legacy_ethtool_table[i].advertised,
+				      link_modes,
+				      __ETHTOOL_LINK_MODE_MASK_NBITS))
+			ptys_modes |= MLX5E_PROT_MASK(i);
+	}
+
+	return ptys_modes;
+}
+
+static u32 mlx5e_ethtool2ptys_ext_adver_link(const unsigned long *link_modes)
+{
+	u32 i, ptys_modes = 0;
+	unsigned long modes[2];
+
+	for (i = 0; i < MLX5E_EXT_LINK_MODES_NUMBER; ++i) {
+		if (ptys2ext_ethtool_table[i].advertised[0] == 0 &&
+		    ptys2ext_ethtool_table[i].advertised[1] == 0)
+			continue;
+		memset(modes, 0, sizeof(modes));
+		bitmap_and(modes, ptys2ext_ethtool_table[i].advertised,
+			   link_modes, __ETHTOOL_LINK_MODE_MASK_NBITS);
+
+		if (modes[0] == ptys2ext_ethtool_table[i].advertised[0] &&
+		    modes[1] == ptys2ext_ethtool_table[i].advertised[1])
+			ptys_modes |= MLX5E_PROT_MASK(i);
+	}
+	return ptys_modes;
+}
+
+static bool ext_link_mode_requested(const unsigned long *adver)
+{
+#define MLX5E_MIN_PTYS_EXT_LINK_MODE_BIT ETHTOOL_LINK_MODE_50000baseKR_Full_BIT
+	int size = __ETHTOOL_LINK_MODE_MASK_NBITS - MLX5E_MIN_PTYS_EXT_LINK_MODE_BIT;
+	__ETHTOOL_DECLARE_LINK_MODE_MASK(modes) = {0,};
+
+	bitmap_set(modes, MLX5E_MIN_PTYS_EXT_LINK_MODE_BIT, size);
+	return bitmap_intersects(modes, adver, __ETHTOOL_LINK_MODE_MASK_NBITS);
+}
+
+static bool ext_requested(u8 autoneg, const unsigned long *adver, bool ext_supported)
+{
+	bool ext_link_mode = ext_link_mode_requested(adver);
+
+	return  autoneg == AUTONEG_ENABLE ? ext_link_mode : ext_supported;
+}
+
+int mlx5e_ethtool_set_link_ksettings(struct mlx5e_priv *priv,
+				     const struct ethtool_link_ksettings *link_ksettings)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+	struct mlx5e_port_eth_proto eproto;
+	const unsigned long *adver;
+	bool an_changes = false;
+	u8 an_disable_admin;
+	bool ext_supported;
+	u8 an_disable_cap;
+	bool an_disable;
+	u32 link_modes;
+	u8 an_status;
+	u8 autoneg;
+	u32 speed;
+	bool ext;
+	int err;
+
+	u32 (*ethtool2ptys_adver_func)(const unsigned long *adver);
+
+	adver = link_ksettings->link_modes.advertising;
+	autoneg = link_ksettings->base.autoneg;
+	speed = link_ksettings->base.speed;
+
+	ext_supported = mlx5e_ptys_ext_supported(mdev);
+	ext = ext_requested(autoneg, adver, ext_supported);
+	if (!ext_supported && ext)
+		return -EOPNOTSUPP;
+
+	ethtool2ptys_adver_func = ext ? mlx5e_ethtool2ptys_ext_adver_link :
+				  mlx5e_ethtool2ptys_adver_link;
+	err = mlx5_port_query_eth_proto(mdev, 1, ext, &eproto);
+	if (err) {
+		netdev_err(priv->netdev, "%s: query port eth proto failed: %d\n",
+			   __func__, err);
+		goto out;
+	}
+	link_modes = autoneg == AUTONEG_ENABLE ? ethtool2ptys_adver_func(adver) :
+		mlx5e_port_speed2linkmodes(mdev, speed, !ext);
+
+	err = mlx5e_speed_validate(priv->netdev, ext, link_modes, autoneg);
+	if (err)
+		goto out;
+
+	link_modes = link_modes & eproto.cap;
+	if (!link_modes) {
+		netdev_err(priv->netdev, "%s: Not supported link mode(s) requested",
+			   __func__);
+		err = -EINVAL;
+		goto out;
+	}
+
+	mlx5_port_query_eth_autoneg(mdev, &an_status, &an_disable_cap,
+				    &an_disable_admin);
+
+	an_disable = autoneg == AUTONEG_DISABLE;
+	an_changes = ((!an_disable && an_disable_admin) ||
+		      (an_disable && !an_disable_admin));
+
+	if (!an_changes && link_modes == eproto.admin)
+		goto out;
+
+	mlx5_port_set_eth_ptys(mdev, an_disable, link_modes, ext);
+	mlx5_toggle_port_link(mdev);
+
+out:
+	return err;
+}
+
+static int mlx5e_set_link_ksettings(struct net_device *netdev,
+				    const struct ethtool_link_ksettings *link_ksettings)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+
+	return mlx5e_ethtool_set_link_ksettings(priv, link_ksettings);
+}
+
+u32 mlx5e_ethtool_get_rxfh_key_size(struct mlx5e_priv *priv)
+{
+	return sizeof_field(struct mlx5e_rss_params_hash, toeplitz_hash_key);
+}
+
+static u32 mlx5e_get_rxfh_key_size(struct net_device *netdev)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+
+	return mlx5e_ethtool_get_rxfh_key_size(priv);
+}
+
+u32 mlx5e_ethtool_get_rxfh_indir_size(struct mlx5e_priv *priv)
+{
+	return MLX5E_INDIR_RQT_SIZE;
+}
+
+static u32 mlx5e_get_rxfh_indir_size(struct net_device *netdev)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+
+	return mlx5e_ethtool_get_rxfh_indir_size(priv);
+}
+
+static int mlx5e_get_rxfh_context(struct net_device *dev, u32 *indir,
+				  u8 *key, u8 *hfunc, u32 rss_context)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	int err;
+
+	mutex_lock(&priv->state_lock);
+	err = mlx5e_rx_res_rss_get_rxfh(priv->rx_res, rss_context, indir, key, hfunc);
+	mutex_unlock(&priv->state_lock);
+	return err;
+}
+
+static int mlx5e_set_rxfh_context(struct net_device *dev, const u32 *indir,
+				  const u8 *key, const u8 hfunc,
+				  u32 *rss_context, bool delete)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	int err;
+
+	mutex_lock(&priv->state_lock);
+	if (delete) {
+		err = mlx5e_rx_res_rss_destroy(priv->rx_res, *rss_context);
+		goto unlock;
+	}
+
+	if (*rss_context == ETH_RXFH_CONTEXT_ALLOC) {
+		unsigned int count = priv->channels.params.num_channels;
+
+		err = mlx5e_rx_res_rss_init(priv->rx_res, rss_context, count);
+		if (err)
+			goto unlock;
+	}
+
+	err = mlx5e_rx_res_rss_set_rxfh(priv->rx_res, *rss_context, indir, key,
+					hfunc == ETH_RSS_HASH_NO_CHANGE ? NULL : &hfunc);
+
+unlock:
+	mutex_unlock(&priv->state_lock);
+	return err;
+}
+
+int mlx5e_get_rxfh(struct net_device *netdev, u32 *indir, u8 *key,
+		   u8 *hfunc)
+{
+	return mlx5e_get_rxfh_context(netdev, indir, key, hfunc, 0);
+}
+
+int mlx5e_set_rxfh(struct net_device *dev, const u32 *indir,
+		   const u8 *key, const u8 hfunc)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	int err;
+
+	mutex_lock(&priv->state_lock);
+	err = mlx5e_rx_res_rss_set_rxfh(priv->rx_res, 0, indir, key,
+					hfunc == ETH_RSS_HASH_NO_CHANGE ? NULL : &hfunc);
+	mutex_unlock(&priv->state_lock);
+	return err;
+}
+
+#define MLX5E_PFC_PREVEN_AUTO_TOUT_MSEC		100
+#define MLX5E_PFC_PREVEN_TOUT_MAX_MSEC		8000
+#define MLX5E_PFC_PREVEN_MINOR_PRECENT		85
+#define MLX5E_PFC_PREVEN_TOUT_MIN_MSEC		80
+#define MLX5E_DEVICE_STALL_MINOR_WATERMARK(critical_tout) \
+	max_t(u16, MLX5E_PFC_PREVEN_TOUT_MIN_MSEC, \
+	      (critical_tout * MLX5E_PFC_PREVEN_MINOR_PRECENT) / 100)
+
+static int mlx5e_get_pfc_prevention_tout(struct net_device *netdev,
+					 u16 *pfc_prevention_tout)
+{
+	struct mlx5e_priv *priv    = netdev_priv(netdev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+
+	if (!MLX5_CAP_PCAM_FEATURE((priv)->mdev, pfcc_mask) ||
+	    !MLX5_CAP_DEBUG((priv)->mdev, stall_detect))
+		return -EOPNOTSUPP;
+
+	return mlx5_query_port_stall_watermark(mdev, pfc_prevention_tout, NULL);
+}
+
+static int mlx5e_set_pfc_prevention_tout(struct net_device *netdev,
+					 u16 pfc_preven)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+	u16 critical_tout;
+	u16 minor;
+
+	if (!MLX5_CAP_PCAM_FEATURE((priv)->mdev, pfcc_mask) ||
+	    !MLX5_CAP_DEBUG((priv)->mdev, stall_detect))
+		return -EOPNOTSUPP;
+
+	critical_tout = (pfc_preven == PFC_STORM_PREVENTION_AUTO) ?
+			MLX5E_PFC_PREVEN_AUTO_TOUT_MSEC :
+			pfc_preven;
+
+	if (critical_tout != PFC_STORM_PREVENTION_DISABLE &&
+	    (critical_tout > MLX5E_PFC_PREVEN_TOUT_MAX_MSEC ||
+	     critical_tout < MLX5E_PFC_PREVEN_TOUT_MIN_MSEC)) {
+		netdev_info(netdev, "%s: pfc prevention tout not in range (%d-%d)\n",
+			    __func__, MLX5E_PFC_PREVEN_TOUT_MIN_MSEC,
+			    MLX5E_PFC_PREVEN_TOUT_MAX_MSEC);
+		return -EINVAL;
+	}
+
+	minor = MLX5E_DEVICE_STALL_MINOR_WATERMARK(critical_tout);
+	return mlx5_set_port_stall_watermark(mdev, critical_tout,
+					     minor);
+}
+
+static int mlx5e_get_tunable(struct net_device *dev,
+			     const struct ethtool_tunable *tuna,
+			     void *data)
+{
+	int err;
+
+	switch (tuna->id) {
+	case ETHTOOL_PFC_PREVENTION_TOUT:
+		err = mlx5e_get_pfc_prevention_tout(dev, data);
+		break;
+	default:
+		err = -EINVAL;
+		break;
+	}
+
+	return err;
+}
+
+static int mlx5e_set_tunable(struct net_device *dev,
+			     const struct ethtool_tunable *tuna,
+			     const void *data)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	int err;
+
+	mutex_lock(&priv->state_lock);
+
+	switch (tuna->id) {
+	case ETHTOOL_PFC_PREVENTION_TOUT:
+		err = mlx5e_set_pfc_prevention_tout(dev, *(u16 *)data);
+		break;
+	default:
+		err = -EINVAL;
+		break;
+	}
+
+	mutex_unlock(&priv->state_lock);
+	return err;
+}
+
+static void mlx5e_get_pause_stats(struct net_device *netdev,
+				  struct ethtool_pause_stats *pause_stats)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+
+	mlx5e_stats_pause_get(priv, pause_stats);
+}
+
+void mlx5e_ethtool_get_pauseparam(struct mlx5e_priv *priv,
+				  struct ethtool_pauseparam *pauseparam)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+	int err;
+
+	err = mlx5_query_port_pause(mdev, &pauseparam->rx_pause,
+				    &pauseparam->tx_pause);
+	if (err) {
+		netdev_err(priv->netdev, "%s: mlx5_query_port_pause failed:0x%x\n",
+			   __func__, err);
+	}
+}
+
+static void mlx5e_get_pauseparam(struct net_device *netdev,
+				 struct ethtool_pauseparam *pauseparam)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+
+	mlx5e_ethtool_get_pauseparam(priv, pauseparam);
+}
+
+int mlx5e_ethtool_set_pauseparam(struct mlx5e_priv *priv,
+				 struct ethtool_pauseparam *pauseparam)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+	int err;
+
+	if (!MLX5_CAP_GEN(mdev, vport_group_manager))
+		return -EOPNOTSUPP;
+
+	if (pauseparam->autoneg)
+		return -EINVAL;
+
+	err = mlx5_set_port_pause(mdev,
+				  pauseparam->rx_pause ? 1 : 0,
+				  pauseparam->tx_pause ? 1 : 0);
+	if (err) {
+		netdev_err(priv->netdev, "%s: mlx5_set_port_pause failed:0x%x\n",
+			   __func__, err);
+	}
+
+	return err;
+}
+
+static int mlx5e_set_pauseparam(struct net_device *netdev,
+				struct ethtool_pauseparam *pauseparam)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+
+	return mlx5e_ethtool_set_pauseparam(priv, pauseparam);
+}
+
+int mlx5e_ethtool_get_ts_info(struct mlx5e_priv *priv,
+			      struct ethtool_ts_info *info)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+
+	info->phc_index = mlx5_clock_get_ptp_index(mdev);
+
+	if (!MLX5_CAP_GEN(priv->mdev, device_frequency_khz) ||
+	    info->phc_index == -1)
+		return 0;
+
+	info->so_timestamping = SOF_TIMESTAMPING_TX_HARDWARE |
+				SOF_TIMESTAMPING_RX_HARDWARE |
+				SOF_TIMESTAMPING_RAW_HARDWARE;
+
+	info->tx_types = BIT(HWTSTAMP_TX_OFF) |
+			 BIT(HWTSTAMP_TX_ON);
+
+	info->rx_filters = BIT(HWTSTAMP_FILTER_NONE) |
+			   BIT(HWTSTAMP_FILTER_ALL);
+
+	return 0;
+}
+
+static int mlx5e_get_ts_info(struct net_device *dev,
+			     struct ethtool_ts_info *info)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+
+	return mlx5e_ethtool_get_ts_info(priv, info);
+}
+
+static __u32 mlx5e_get_wol_supported(struct mlx5_core_dev *mdev)
+{
+	__u32 ret = 0;
+
+	if (MLX5_CAP_GEN(mdev, wol_g))
+		ret |= WAKE_MAGIC;
+
+	if (MLX5_CAP_GEN(mdev, wol_s))
+		ret |= WAKE_MAGICSECURE;
+
+	if (MLX5_CAP_GEN(mdev, wol_a))
+		ret |= WAKE_ARP;
+
+	if (MLX5_CAP_GEN(mdev, wol_b))
+		ret |= WAKE_BCAST;
+
+	if (MLX5_CAP_GEN(mdev, wol_m))
+		ret |= WAKE_MCAST;
+
+	if (MLX5_CAP_GEN(mdev, wol_u))
+		ret |= WAKE_UCAST;
+
+	if (MLX5_CAP_GEN(mdev, wol_p))
+		ret |= WAKE_PHY;
+
+	return ret;
+}
+
+static __u32 mlx5e_reformat_wol_mode_mlx5_to_linux(u8 mode)
+{
+	__u32 ret = 0;
+
+	if (mode & MLX5_WOL_MAGIC)
+		ret |= WAKE_MAGIC;
+
+	if (mode & MLX5_WOL_SECURED_MAGIC)
+		ret |= WAKE_MAGICSECURE;
+
+	if (mode & MLX5_WOL_ARP)
+		ret |= WAKE_ARP;
+
+	if (mode & MLX5_WOL_BROADCAST)
+		ret |= WAKE_BCAST;
+
+	if (mode & MLX5_WOL_MULTICAST)
+		ret |= WAKE_MCAST;
+
+	if (mode & MLX5_WOL_UNICAST)
+		ret |= WAKE_UCAST;
+
+	if (mode & MLX5_WOL_PHY_ACTIVITY)
+		ret |= WAKE_PHY;
+
+	return ret;
+}
+
+static u8 mlx5e_reformat_wol_mode_linux_to_mlx5(__u32 mode)
+{
+	u8 ret = 0;
+
+	if (mode & WAKE_MAGIC)
+		ret |= MLX5_WOL_MAGIC;
+
+	if (mode & WAKE_MAGICSECURE)
+		ret |= MLX5_WOL_SECURED_MAGIC;
+
+	if (mode & WAKE_ARP)
+		ret |= MLX5_WOL_ARP;
+
+	if (mode & WAKE_BCAST)
+		ret |= MLX5_WOL_BROADCAST;
+
+	if (mode & WAKE_MCAST)
+		ret |= MLX5_WOL_MULTICAST;
+
+	if (mode & WAKE_UCAST)
+		ret |= MLX5_WOL_UNICAST;
+
+	if (mode & WAKE_PHY)
+		ret |= MLX5_WOL_PHY_ACTIVITY;
+
+	return ret;
+}
+
+static void mlx5e_get_wol(struct net_device *netdev,
+			  struct ethtool_wolinfo *wol)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+	u8 mlx5_wol_mode;
+	int err;
+
+	memset(wol, 0, sizeof(*wol));
+
+	wol->supported = mlx5e_get_wol_supported(mdev);
+	if (!wol->supported)
+		return;
+
+	err = mlx5_query_port_wol(mdev, &mlx5_wol_mode);
+	if (err)
+		return;
+
+	wol->wolopts = mlx5e_reformat_wol_mode_mlx5_to_linux(mlx5_wol_mode);
+}
+
+static int mlx5e_set_wol(struct net_device *netdev, struct ethtool_wolinfo *wol)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+	__u32 wol_supported = mlx5e_get_wol_supported(mdev);
+	u32 mlx5_wol_mode;
+
+	if (!wol_supported)
+		return -EOPNOTSUPP;
+
+	if (wol->wolopts & ~wol_supported)
+		return -EINVAL;
+
+	mlx5_wol_mode = mlx5e_reformat_wol_mode_linux_to_mlx5(wol->wolopts);
+
+	return mlx5_set_port_wol(mdev, mlx5_wol_mode);
+}
+
+static void mlx5e_get_fec_stats(struct net_device *netdev,
+				struct ethtool_fec_stats *fec_stats)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+
+	mlx5e_stats_fec_get(priv, fec_stats);
+}
+
+static int mlx5e_get_fecparam(struct net_device *netdev,
+			      struct ethtool_fecparam *fecparam)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+	u16 fec_configured;
+	u32 fec_active;
+	int err;
+
+	err = mlx5e_get_fec_mode(mdev, &fec_active, &fec_configured);
+
+	if (err)
+		return err;
+
+	fecparam->active_fec = pplm2ethtool_fec((unsigned long)fec_active,
+						sizeof(unsigned long) * BITS_PER_BYTE);
+
+	if (!fecparam->active_fec)
+		return -EOPNOTSUPP;
+
+	fecparam->fec = pplm2ethtool_fec((unsigned long)fec_configured,
+					 sizeof(unsigned long) * BITS_PER_BYTE);
+
+	return 0;
+}
+
+static int mlx5e_set_fecparam(struct net_device *netdev,
+			      struct ethtool_fecparam *fecparam)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+	unsigned long fec_bitmap;
+	u16 fec_policy = 0;
+	int mode;
+	int err;
+
+	bitmap_from_arr32(&fec_bitmap, &fecparam->fec, sizeof(fecparam->fec) * BITS_PER_BYTE);
+	if (bitmap_weight(&fec_bitmap, ETHTOOL_FEC_LLRS_BIT + 1) > 1)
+		return -EOPNOTSUPP;
+
+	for (mode = 0; mode < ARRAY_SIZE(pplm_fec_2_ethtool); mode++) {
+		if (!(pplm_fec_2_ethtool[mode] & fecparam->fec))
+			continue;
+		fec_policy |= (1 << mode);
+		break;
+	}
+
+	err = mlx5e_set_fec_mode(mdev, fec_policy);
+
+	if (err)
+		return err;
+
+	mlx5_toggle_port_link(mdev);
+
+	return 0;
+}
+
+static u32 mlx5e_get_msglevel(struct net_device *dev)
+{
+	return ((struct mlx5e_priv *)netdev_priv(dev))->msglevel;
+}
+
+static void mlx5e_set_msglevel(struct net_device *dev, u32 val)
+{
+	((struct mlx5e_priv *)netdev_priv(dev))->msglevel = val;
+}
+
+static int mlx5e_set_phys_id(struct net_device *dev,
+			     enum ethtool_phys_id_state state)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+	u16 beacon_duration;
+
+	if (!MLX5_CAP_GEN(mdev, beacon_led))
+		return -EOPNOTSUPP;
+
+	switch (state) {
+	case ETHTOOL_ID_ACTIVE:
+		beacon_duration = MLX5_BEACON_DURATION_INF;
+		break;
+	case ETHTOOL_ID_INACTIVE:
+		beacon_duration = MLX5_BEACON_DURATION_OFF;
+		break;
+	default:
+		return -EOPNOTSUPP;
+	}
+
+	return mlx5_set_port_beacon(mdev, beacon_duration);
+}
+
+static int mlx5e_get_module_info(struct net_device *netdev,
+				 struct ethtool_modinfo *modinfo)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	struct mlx5_core_dev *dev = priv->mdev;
+	int size_read = 0;
+	u8 data[4] = {0};
+
+	size_read = mlx5_query_module_eeprom(dev, 0, 2, data);
+	if (size_read < 2)
+		return -EIO;
+
+	/* data[0] = identifier byte */
+	switch (data[0]) {
+	case MLX5_MODULE_ID_QSFP:
+		modinfo->type       = ETH_MODULE_SFF_8436;
+		modinfo->eeprom_len = ETH_MODULE_SFF_8436_MAX_LEN;
+		break;
+	case MLX5_MODULE_ID_QSFP_PLUS:
+	case MLX5_MODULE_ID_QSFP28:
+		/* data[1] = revision id */
+		if (data[0] == MLX5_MODULE_ID_QSFP28 || data[1] >= 0x3) {
+			modinfo->type       = ETH_MODULE_SFF_8636;
+			modinfo->eeprom_len = ETH_MODULE_SFF_8636_MAX_LEN;
+		} else {
+			modinfo->type       = ETH_MODULE_SFF_8436;
+			modinfo->eeprom_len = ETH_MODULE_SFF_8436_MAX_LEN;
+		}
+		break;
+	case MLX5_MODULE_ID_SFP:
+		modinfo->type       = ETH_MODULE_SFF_8472;
+		modinfo->eeprom_len = ETH_MODULE_SFF_8472_LEN;
+		break;
+	default:
+		netdev_err(priv->netdev, "%s: cable type not recognized:0x%x\n",
+			   __func__, data[0]);
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int mlx5e_get_module_eeprom(struct net_device *netdev,
+				   struct ethtool_eeprom *ee,
+				   u8 *data)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+	int offset = ee->offset;
+	int size_read;
+	int i = 0;
+
+	if (!ee->len)
+		return -EINVAL;
+
+	memset(data, 0, ee->len);
+
+	while (i < ee->len) {
+		size_read = mlx5_query_module_eeprom(mdev, offset, ee->len - i,
+						     data + i);
+
+		if (!size_read)
+			/* Done reading */
+			return 0;
+
+		if (size_read < 0) {
+			netdev_err(priv->netdev, "%s: mlx5_query_eeprom failed:0x%x\n",
+				   __func__, size_read);
+			return size_read;
+		}
+
+		i += size_read;
+		offset += size_read;
+	}
+
+	return 0;
+}
+
+static int mlx5e_get_module_eeprom_by_page(struct net_device *netdev,
+					   const struct ethtool_module_eeprom *page_data,
+					   struct netlink_ext_ack *extack)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	struct mlx5_module_eeprom_query_params query;
+	struct mlx5_core_dev *mdev = priv->mdev;
+	u8 *data = page_data->data;
+	int size_read;
+	int i = 0;
+
+	if (!page_data->length)
+		return -EINVAL;
+
+	memset(data, 0, page_data->length);
+
+	query.offset = page_data->offset;
+	query.i2c_address = page_data->i2c_address;
+	query.bank = page_data->bank;
+	query.page = page_data->page;
+	while (i < page_data->length) {
+		query.size = page_data->length - i;
+		size_read = mlx5_query_module_eeprom_by_page(mdev, &query, data + i);
+
+		/* Done reading, return how many bytes was read */
+		if (!size_read)
+			return i;
+
+		if (size_read == -EINVAL)
+			return -EINVAL;
+		if (size_read < 0) {
+			netdev_err(priv->netdev, "%s: mlx5_query_module_eeprom_by_page failed:0x%x\n",
+				   __func__, size_read);
+			return i;
+		}
+
+		i += size_read;
+		query.offset += size_read;
+	}
+
+	return i;
+}
+
+int mlx5e_ethtool_flash_device(struct mlx5e_priv *priv,
+			       struct ethtool_flash *flash)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+	struct net_device *dev = priv->netdev;
+	const struct firmware *fw;
+	int err;
+
+	if (flash->region != ETHTOOL_FLASH_ALL_REGIONS)
+		return -EOPNOTSUPP;
+
+	err = request_firmware_direct(&fw, flash->data, &dev->dev);
+	if (err)
+		return err;
+
+	dev_hold(dev);
+	rtnl_unlock();
+
+	err = mlx5_firmware_flash(mdev, fw, NULL);
+	release_firmware(fw);
+
+	rtnl_lock();
+	dev_put(dev);
+	return err;
+}
+
+static int mlx5e_flash_device(struct net_device *dev,
+			      struct ethtool_flash *flash)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+
+	return mlx5e_ethtool_flash_device(priv, flash);
+}
+
+static int set_pflag_cqe_based_moder(struct net_device *netdev, bool enable,
+				     bool is_rx_cq)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	u8 cq_period_mode, current_cq_period_mode;
+	struct mlx5e_params new_params;
+
+	if (enable && !MLX5_CAP_GEN(priv->mdev, cq_period_start_from_cqe))
+		return -EOPNOTSUPP;
+
+	cq_period_mode = cqe_mode_to_period_mode(enable);
+
+	current_cq_period_mode = is_rx_cq ?
+		priv->channels.params.rx_cq_moderation.cq_period_mode :
+		priv->channels.params.tx_cq_moderation.cq_period_mode;
+
+	if (cq_period_mode == current_cq_period_mode)
+		return 0;
+
+	new_params = priv->channels.params;
+	if (is_rx_cq)
+		mlx5e_set_rx_cq_mode_params(&new_params, cq_period_mode);
+	else
+		mlx5e_set_tx_cq_mode_params(&new_params, cq_period_mode);
+
+	return mlx5e_safe_switch_params(priv, &new_params, NULL, NULL, true);
+}
+
+static int set_pflag_tx_cqe_based_moder(struct net_device *netdev, bool enable)
+{
+	return set_pflag_cqe_based_moder(netdev, enable, false);
+}
+
+static int set_pflag_rx_cqe_based_moder(struct net_device *netdev, bool enable)
+{
+	return set_pflag_cqe_based_moder(netdev, enable, true);
+}
+
+int mlx5e_modify_rx_cqe_compression_locked(struct mlx5e_priv *priv, bool new_val, bool rx_filter)
+{
+	bool curr_val = MLX5E_GET_PFLAG(&priv->channels.params, MLX5E_PFLAG_RX_CQE_COMPRESS);
+	struct mlx5e_params new_params;
+	int err = 0;
+
+	if (!MLX5_CAP_GEN(priv->mdev, cqe_compression))
+		return new_val ? -EOPNOTSUPP : 0;
+
+	if (curr_val == new_val)
+		return 0;
+
+	if (new_val && !mlx5e_profile_feature_cap(priv->profile, PTP_RX) && rx_filter) {
+		netdev_err(priv->netdev,
+			   "Profile doesn't support enabling of CQE compression while hardware time-stamping is enabled.\n");
+		return -EINVAL;
+	}
+
+	if (priv->channels.params.packet_merge.type == MLX5E_PACKET_MERGE_SHAMPO) {
+		netdev_warn(priv->netdev, "Can't set CQE compression with HW-GRO, disable it first.\n");
+		return -EINVAL;
+	}
+
+	new_params = priv->channels.params;
+	MLX5E_SET_PFLAG(&new_params, MLX5E_PFLAG_RX_CQE_COMPRESS, new_val);
+	if (rx_filter)
+		new_params.ptp_rx = new_val;
+
+	if (new_params.ptp_rx == priv->channels.params.ptp_rx)
+		err = mlx5e_safe_switch_params(priv, &new_params, NULL, NULL, true);
+	else
+		err = mlx5e_safe_switch_params(priv, &new_params, mlx5e_ptp_rx_manage_fs_ctx,
+					       &new_params.ptp_rx, true);
+	if (err)
+		return err;
+
+	mlx5e_dbg(DRV, priv, "MLX5E: RxCqeCmprss was turned %s\n",
+		  MLX5E_GET_PFLAG(&priv->channels.params,
+				  MLX5E_PFLAG_RX_CQE_COMPRESS) ? "ON" : "OFF");
+
+	return 0;
+}
+
+static int set_pflag_rx_cqe_compress(struct net_device *netdev,
+				     bool enable)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+	bool rx_filter;
+	int err;
+
+	if (!MLX5_CAP_GEN(mdev, cqe_compression))
+		return -EOPNOTSUPP;
+
+	rx_filter = priv->tstamp.rx_filter != HWTSTAMP_FILTER_NONE;
+	err = mlx5e_modify_rx_cqe_compression_locked(priv, enable, rx_filter);
+	if (err)
+		return err;
+
+	priv->channels.params.rx_cqe_compress_def = enable;
+
+	return 0;
+}
+
+static int set_pflag_rx_striding_rq(struct net_device *netdev, bool enable)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+	struct mlx5e_params new_params;
+
+	if (enable) {
+		/* Checking the regular RQ here; mlx5e_validate_xsk_param called
+		 * from mlx5e_open_xsk will check for each XSK queue, and
+		 * mlx5e_safe_switch_params will be reverted if any check fails.
+		 */
+		int err = mlx5e_mpwrq_validate_regular(mdev, &priv->channels.params);
+
+		if (err)
+			return err;
+	} else if (priv->channels.params.packet_merge.type != MLX5E_PACKET_MERGE_NONE) {
+		netdev_warn(netdev, "Can't set legacy RQ with HW-GRO/LRO, disable them first\n");
+		return -EINVAL;
+	}
+
+	new_params = priv->channels.params;
+
+	MLX5E_SET_PFLAG(&new_params, MLX5E_PFLAG_RX_STRIDING_RQ, enable);
+	mlx5e_set_rq_type(mdev, &new_params);
+
+	return mlx5e_safe_switch_params(priv, &new_params, NULL, NULL, true);
+}
+
+static int set_pflag_rx_no_csum_complete(struct net_device *netdev, bool enable)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	struct mlx5e_channels *channels = &priv->channels;
+	struct mlx5e_channel *c;
+	int i;
+
+	if (!test_bit(MLX5E_STATE_OPENED, &priv->state) ||
+	    priv->channels.params.xdp_prog)
+		return 0;
+
+	for (i = 0; i < channels->num; i++) {
+		c = channels->c[i];
+		if (enable)
+			__set_bit(MLX5E_RQ_STATE_NO_CSUM_COMPLETE, &c->rq.state);
+		else
+			__clear_bit(MLX5E_RQ_STATE_NO_CSUM_COMPLETE, &c->rq.state);
+	}
+
+	return 0;
+}
+
+static int set_pflag_tx_mpwqe_common(struct net_device *netdev, u32 flag, bool enable)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+	struct mlx5e_params new_params;
+
+	if (enable && !mlx5e_tx_mpwqe_supported(mdev))
+		return -EOPNOTSUPP;
+
+	new_params = priv->channels.params;
+
+	MLX5E_SET_PFLAG(&new_params, flag, enable);
+
+	return mlx5e_safe_switch_params(priv, &new_params, NULL, NULL, true);
+}
+
+static int set_pflag_xdp_tx_mpwqe(struct net_device *netdev, bool enable)
+{
+	return set_pflag_tx_mpwqe_common(netdev, MLX5E_PFLAG_XDP_TX_MPWQE, enable);
+}
+
+static int set_pflag_skb_tx_mpwqe(struct net_device *netdev, bool enable)
+{
+	return set_pflag_tx_mpwqe_common(netdev, MLX5E_PFLAG_SKB_TX_MPWQE, enable);
+}
+
+static int set_pflag_tx_port_ts(struct net_device *netdev, bool enable)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+	struct mlx5e_params new_params;
+	int err;
+
+	if (!MLX5_CAP_GEN(mdev, ts_cqe_to_dest_cqn))
+		return -EOPNOTSUPP;
+
+	/* Don't allow changing the PTP state if HTB offload is active, because
+	 * the numeration of the QoS SQs will change, while per-queue qdiscs are
+	 * attached.
+	 */
+	if (mlx5e_selq_is_htb_enabled(&priv->selq)) {
+		netdev_err(priv->netdev, "%s: HTB offload is active, cannot change the PTP state\n",
+			   __func__);
+		return -EINVAL;
+	}
+
+	new_params = priv->channels.params;
+	/* Don't allow enabling TX-port-TS if MQPRIO mode channel  offload is
+	 * active, since it defines explicitly which TC accepts the packet.
+	 * This conflicts with TX-port-TS hijacking the PTP traffic to a specific
+	 * HW TX-queue.
+	 */
+	if (enable && new_params.mqprio.mode == TC_MQPRIO_MODE_CHANNEL) {
+		netdev_err(priv->netdev,
+			   "%s: MQPRIO mode channel offload is active, cannot set the TX-port-TS\n",
+			   __func__);
+		return -EINVAL;
+	}
+	MLX5E_SET_PFLAG(&new_params, MLX5E_PFLAG_TX_PORT_TS, enable);
+	/* No need to verify SQ stop room as
+	 * ptpsq.txqsq.stop_room <= generic_sq->stop_room, and both
+	 * has the same log_sq_size.
+	 */
+
+	err = mlx5e_safe_switch_params(priv, &new_params,
+				       mlx5e_num_channels_changed_ctx, NULL, true);
+	if (!err)
+		priv->tx_ptp_opened = true;
+
+	return err;
+}
+
+static const struct pflag_desc mlx5e_priv_flags[MLX5E_NUM_PFLAGS] = {
+	{ "rx_cqe_moder",        set_pflag_rx_cqe_based_moder },
+	{ "tx_cqe_moder",        set_pflag_tx_cqe_based_moder },
+	{ "rx_cqe_compress",     set_pflag_rx_cqe_compress },
+	{ "rx_striding_rq",      set_pflag_rx_striding_rq },
+	{ "rx_no_csum_complete", set_pflag_rx_no_csum_complete },
+	{ "xdp_tx_mpwqe",        set_pflag_xdp_tx_mpwqe },
+	{ "skb_tx_mpwqe",        set_pflag_skb_tx_mpwqe },
+	{ "tx_port_ts",          set_pflag_tx_port_ts },
+};
+
+static int mlx5e_handle_pflag(struct net_device *netdev,
+			      u32 wanted_flags,
+			      enum mlx5e_priv_flag flag)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	bool enable = !!(wanted_flags & BIT(flag));
+	u32 changes = wanted_flags ^ priv->channels.params.pflags;
+	int err;
+
+	if (!(changes & BIT(flag)))
+		return 0;
+
+	err = mlx5e_priv_flags[flag].handler(netdev, enable);
+	if (err) {
+		netdev_err(netdev, "%s private flag '%s' failed err %d\n",
+			   enable ? "Enable" : "Disable", mlx5e_priv_flags[flag].name, err);
+		return err;
+	}
+
+	MLX5E_SET_PFLAG(&priv->channels.params, flag, enable);
+	return 0;
+}
+
+static int mlx5e_set_priv_flags(struct net_device *netdev, u32 pflags)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	enum mlx5e_priv_flag pflag;
+	int err;
+
+	mutex_lock(&priv->state_lock);
+
+	for (pflag = 0; pflag < MLX5E_NUM_PFLAGS; pflag++) {
+		err = mlx5e_handle_pflag(netdev, pflags, pflag);
+		if (err)
+			break;
+	}
+
+	mutex_unlock(&priv->state_lock);
+
+	/* Need to fix some features.. */
+	netdev_update_features(netdev);
+
+	return err;
+}
+
+static u32 mlx5e_get_priv_flags(struct net_device *netdev)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+
+	return priv->channels.params.pflags;
+}
+
+int mlx5e_get_rxnfc(struct net_device *dev, struct ethtool_rxnfc *info,
+		    u32 *rule_locs)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+
+	/* ETHTOOL_GRXRINGS is needed by ethtool -x which is not part
+	 * of rxnfc. We keep this logic out of mlx5e_ethtool_get_rxnfc,
+	 * to avoid breaking "ethtool -x" when mlx5e_ethtool_get_rxnfc
+	 * is compiled out via CONFIG_MLX5_EN_RXNFC=n.
+	 */
+	if (info->cmd == ETHTOOL_GRXRINGS) {
+		info->data = priv->channels.params.num_channels;
+		return 0;
+	}
+
+	return mlx5e_ethtool_get_rxnfc(priv, info, rule_locs);
+}
+
+int mlx5e_set_rxnfc(struct net_device *dev, struct ethtool_rxnfc *cmd)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+
+	return mlx5e_ethtool_set_rxnfc(priv, cmd);
+}
+
+static int query_port_status_opcode(struct mlx5_core_dev *mdev, u32 *status_opcode)
+{
+	struct mlx5_ifc_pddr_troubleshooting_page_bits *pddr_troubleshooting_page;
+	u32 in[MLX5_ST_SZ_DW(pddr_reg)] = {};
+	u32 out[MLX5_ST_SZ_DW(pddr_reg)];
+	int err;
+
+	MLX5_SET(pddr_reg, in, local_port, 1);
+	MLX5_SET(pddr_reg, in, page_select,
+		 MLX5_PDDR_REG_PAGE_SELECT_TROUBLESHOOTING_INFO_PAGE);
+
+	pddr_troubleshooting_page = MLX5_ADDR_OF(pddr_reg, in, page_data);
+	MLX5_SET(pddr_troubleshooting_page, pddr_troubleshooting_page,
+		 group_opcode, MLX5_PDDR_REG_TRBLSH_GROUP_OPCODE_MONITOR);
+	err = mlx5_core_access_reg(mdev, in, sizeof(in), out,
+				   sizeof(out), MLX5_REG_PDDR, 0, 0);
+	if (err)
+		return err;
+
+	pddr_troubleshooting_page = MLX5_ADDR_OF(pddr_reg, out, page_data);
+	*status_opcode = MLX5_GET(pddr_troubleshooting_page, pddr_troubleshooting_page,
+				  status_opcode);
+	return 0;
+}
+
+struct mlx5e_ethtool_link_ext_state_opcode_mapping {
+	u32 status_opcode;
+	enum ethtool_link_ext_state link_ext_state;
+	u8 link_ext_substate;
+};
+
+static const struct mlx5e_ethtool_link_ext_state_opcode_mapping
+mlx5e_link_ext_state_opcode_map[] = {
+	/* States relating to the autonegotiation or issues therein */
+	{2, ETHTOOL_LINK_EXT_STATE_AUTONEG,
+		ETHTOOL_LINK_EXT_SUBSTATE_AN_NO_PARTNER_DETECTED},
+	{3, ETHTOOL_LINK_EXT_STATE_AUTONEG,
+		ETHTOOL_LINK_EXT_SUBSTATE_AN_ACK_NOT_RECEIVED},
+	{4, ETHTOOL_LINK_EXT_STATE_AUTONEG,
+		ETHTOOL_LINK_EXT_SUBSTATE_AN_NEXT_PAGE_EXCHANGE_FAILED},
+	{36, ETHTOOL_LINK_EXT_STATE_AUTONEG,
+		ETHTOOL_LINK_EXT_SUBSTATE_AN_NO_PARTNER_DETECTED_FORCE_MODE},
+	{38, ETHTOOL_LINK_EXT_STATE_AUTONEG,
+		ETHTOOL_LINK_EXT_SUBSTATE_AN_FEC_MISMATCH_DURING_OVERRIDE},
+	{39, ETHTOOL_LINK_EXT_STATE_AUTONEG,
+		ETHTOOL_LINK_EXT_SUBSTATE_AN_NO_HCD},
+
+	/* Failure during link training */
+	{5, ETHTOOL_LINK_EXT_STATE_LINK_TRAINING_FAILURE,
+		ETHTOOL_LINK_EXT_SUBSTATE_LT_KR_FRAME_LOCK_NOT_ACQUIRED},
+	{6, ETHTOOL_LINK_EXT_STATE_LINK_TRAINING_FAILURE,
+		ETHTOOL_LINK_EXT_SUBSTATE_LT_KR_LINK_INHIBIT_TIMEOUT},
+	{7, ETHTOOL_LINK_EXT_STATE_LINK_TRAINING_FAILURE,
+		ETHTOOL_LINK_EXT_SUBSTATE_LT_KR_LINK_PARTNER_DID_NOT_SET_RECEIVER_READY},
+	{8, ETHTOOL_LINK_EXT_STATE_LINK_TRAINING_FAILURE, 0},
+	{14, ETHTOOL_LINK_EXT_STATE_LINK_TRAINING_FAILURE,
+		ETHTOOL_LINK_EXT_SUBSTATE_LT_REMOTE_FAULT},
+
+	/* Logical mismatch in physical coding sublayer or forward error correction sublayer */
+	{9, ETHTOOL_LINK_EXT_STATE_LINK_LOGICAL_MISMATCH,
+		ETHTOOL_LINK_EXT_SUBSTATE_LLM_PCS_DID_NOT_ACQUIRE_BLOCK_LOCK},
+	{10, ETHTOOL_LINK_EXT_STATE_LINK_LOGICAL_MISMATCH,
+		ETHTOOL_LINK_EXT_SUBSTATE_LLM_PCS_DID_NOT_ACQUIRE_AM_LOCK},
+	{11, ETHTOOL_LINK_EXT_STATE_LINK_LOGICAL_MISMATCH,
+		ETHTOOL_LINK_EXT_SUBSTATE_LLM_PCS_DID_NOT_GET_ALIGN_STATUS},
+	{12, ETHTOOL_LINK_EXT_STATE_LINK_LOGICAL_MISMATCH,
+		ETHTOOL_LINK_EXT_SUBSTATE_LLM_FC_FEC_IS_NOT_LOCKED},
+	{13, ETHTOOL_LINK_EXT_STATE_LINK_LOGICAL_MISMATCH,
+		ETHTOOL_LINK_EXT_SUBSTATE_LLM_RS_FEC_IS_NOT_LOCKED},
+
+	/* Signal integrity issues */
+	{15, ETHTOOL_LINK_EXT_STATE_BAD_SIGNAL_INTEGRITY, 0},
+	{17, ETHTOOL_LINK_EXT_STATE_BAD_SIGNAL_INTEGRITY,
+		ETHTOOL_LINK_EXT_SUBSTATE_BSI_LARGE_NUMBER_OF_PHYSICAL_ERRORS},
+	{42, ETHTOOL_LINK_EXT_STATE_BAD_SIGNAL_INTEGRITY,
+		ETHTOOL_LINK_EXT_SUBSTATE_BSI_UNSUPPORTED_RATE},
+
+	/* No cable connected */
+	{1024, ETHTOOL_LINK_EXT_STATE_NO_CABLE, 0},
+
+	/* Failure is related to cable, e.g., unsupported cable */
+	{16, ETHTOOL_LINK_EXT_STATE_CABLE_ISSUE,
+		ETHTOOL_LINK_EXT_SUBSTATE_CI_UNSUPPORTED_CABLE},
+	{20, ETHTOOL_LINK_EXT_STATE_CABLE_ISSUE,
+		ETHTOOL_LINK_EXT_SUBSTATE_CI_UNSUPPORTED_CABLE},
+	{29, ETHTOOL_LINK_EXT_STATE_CABLE_ISSUE,
+		ETHTOOL_LINK_EXT_SUBSTATE_CI_UNSUPPORTED_CABLE},
+	{1025, ETHTOOL_LINK_EXT_STATE_CABLE_ISSUE,
+		ETHTOOL_LINK_EXT_SUBSTATE_CI_UNSUPPORTED_CABLE},
+	{1029, ETHTOOL_LINK_EXT_STATE_CABLE_ISSUE,
+		ETHTOOL_LINK_EXT_SUBSTATE_CI_UNSUPPORTED_CABLE},
+	{1031, ETHTOOL_LINK_EXT_STATE_CABLE_ISSUE, 0},
+
+	/* Failure is related to EEPROM, e.g., failure during reading or parsing the data */
+	{1027, ETHTOOL_LINK_EXT_STATE_EEPROM_ISSUE, 0},
+
+	/* Failure during calibration algorithm */
+	{23, ETHTOOL_LINK_EXT_STATE_CALIBRATION_FAILURE, 0},
+
+	/* The hardware is not able to provide the power required from cable or module */
+	{1032, ETHTOOL_LINK_EXT_STATE_POWER_BUDGET_EXCEEDED, 0},
+
+	/* The module is overheated */
+	{1030, ETHTOOL_LINK_EXT_STATE_OVERHEAT, 0},
+};
+
+static void
+mlx5e_set_link_ext_state(struct mlx5e_ethtool_link_ext_state_opcode_mapping
+			 link_ext_state_mapping,
+			 struct ethtool_link_ext_state_info *link_ext_state_info)
+{
+	switch (link_ext_state_mapping.link_ext_state) {
+	case ETHTOOL_LINK_EXT_STATE_AUTONEG:
+		link_ext_state_info->autoneg =
+			link_ext_state_mapping.link_ext_substate;
+		break;
+	case ETHTOOL_LINK_EXT_STATE_LINK_TRAINING_FAILURE:
+		link_ext_state_info->link_training =
+			link_ext_state_mapping.link_ext_substate;
+		break;
+	case ETHTOOL_LINK_EXT_STATE_LINK_LOGICAL_MISMATCH:
+		link_ext_state_info->link_logical_mismatch =
+			link_ext_state_mapping.link_ext_substate;
+		break;
+	case ETHTOOL_LINK_EXT_STATE_BAD_SIGNAL_INTEGRITY:
+		link_ext_state_info->bad_signal_integrity =
+			link_ext_state_mapping.link_ext_substate;
+		break;
+	case ETHTOOL_LINK_EXT_STATE_CABLE_ISSUE:
+		link_ext_state_info->cable_issue =
+			link_ext_state_mapping.link_ext_substate;
+		break;
+	default:
+		break;
+	}
+
+	link_ext_state_info->link_ext_state = link_ext_state_mapping.link_ext_state;
+}
+
+static int
+mlx5e_get_link_ext_state(struct net_device *dev,
+			 struct ethtool_link_ext_state_info *link_ext_state_info)
+{
+	struct mlx5e_ethtool_link_ext_state_opcode_mapping link_ext_state_mapping;
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	u32 status_opcode = 0;
+	int i;
+
+	/* Exit without data if the interface state is OK, since no extended data is
+	 * available in such case
+	 */
+	if (netif_carrier_ok(dev))
+		return -ENODATA;
+
+	if (query_port_status_opcode(priv->mdev, &status_opcode) ||
+	    !status_opcode)
+		return -ENODATA;
+
+	for (i = 0; i < ARRAY_SIZE(mlx5e_link_ext_state_opcode_map); i++) {
+		link_ext_state_mapping = mlx5e_link_ext_state_opcode_map[i];
+		if (link_ext_state_mapping.status_opcode == status_opcode) {
+			mlx5e_set_link_ext_state(link_ext_state_mapping,
+						 link_ext_state_info);
+			return 0;
+		}
+	}
+
+	return -ENODATA;
+}
+
+static void mlx5e_get_eth_phy_stats(struct net_device *netdev,
+				    struct ethtool_eth_phy_stats *phy_stats)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+
+	mlx5e_stats_eth_phy_get(priv, phy_stats);
+}
+
+static void mlx5e_get_eth_mac_stats(struct net_device *netdev,
+				    struct ethtool_eth_mac_stats *mac_stats)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+
+	mlx5e_stats_eth_mac_get(priv, mac_stats);
+}
+
+static void mlx5e_get_eth_ctrl_stats(struct net_device *netdev,
+				     struct ethtool_eth_ctrl_stats *ctrl_stats)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+
+	mlx5e_stats_eth_ctrl_get(priv, ctrl_stats);
+}
+
+static void mlx5e_get_rmon_stats(struct net_device *netdev,
+				 struct ethtool_rmon_stats *rmon_stats,
+				 const struct ethtool_rmon_hist_range **ranges)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+
+	mlx5e_stats_rmon_get(priv, rmon_stats, ranges);
+}
+
+const struct ethtool_ops mlx5e_ethtool_ops = {
+	.supported_coalesce_params = ETHTOOL_COALESCE_USECS |
+				     ETHTOOL_COALESCE_MAX_FRAMES |
+				     ETHTOOL_COALESCE_USE_ADAPTIVE |
+				     ETHTOOL_COALESCE_USE_CQE,
+	.get_drvinfo       = mlx5e_get_drvinfo,
+	.get_link          = ethtool_op_get_link,
+	.get_link_ext_state  = mlx5e_get_link_ext_state,
+	.get_strings       = mlx5e_get_strings,
+	.get_sset_count    = mlx5e_get_sset_count,
+	.get_ethtool_stats = mlx5e_get_ethtool_stats,
+	.get_ringparam     = mlx5e_get_ringparam,
+	.set_ringparam     = mlx5e_set_ringparam,
+	.get_channels      = mlx5e_get_channels,
+	.set_channels      = mlx5e_set_channels,
+	.get_coalesce      = mlx5e_get_coalesce,
+	.set_coalesce      = mlx5e_set_coalesce,
+	.get_link_ksettings  = mlx5e_get_link_ksettings,
+	.set_link_ksettings  = mlx5e_set_link_ksettings,
+	.get_rxfh_key_size   = mlx5e_get_rxfh_key_size,
+	.get_rxfh_indir_size = mlx5e_get_rxfh_indir_size,
+	.get_rxfh          = mlx5e_get_rxfh,
+	.set_rxfh          = mlx5e_set_rxfh,
+	.get_rxfh_context  = mlx5e_get_rxfh_context,
+	.set_rxfh_context  = mlx5e_set_rxfh_context,
+	.get_rxnfc         = mlx5e_get_rxnfc,
+	.set_rxnfc         = mlx5e_set_rxnfc,
+	.get_tunable       = mlx5e_get_tunable,
+	.set_tunable       = mlx5e_set_tunable,
+	.get_pause_stats   = mlx5e_get_pause_stats,
+	.get_pauseparam    = mlx5e_get_pauseparam,
+	.set_pauseparam    = mlx5e_set_pauseparam,
+	.get_ts_info       = mlx5e_get_ts_info,
+	.set_phys_id       = mlx5e_set_phys_id,
+	.get_wol	   = mlx5e_get_wol,
+	.set_wol	   = mlx5e_set_wol,
+	.get_module_info   = mlx5e_get_module_info,
+	.get_module_eeprom = mlx5e_get_module_eeprom,
+	.get_module_eeprom_by_page = mlx5e_get_module_eeprom_by_page,
+	.flash_device      = mlx5e_flash_device,
+	.get_priv_flags    = mlx5e_get_priv_flags,
+	.set_priv_flags    = mlx5e_set_priv_flags,
+	.self_test         = mlx5e_self_test,
+	.get_msglevel      = mlx5e_get_msglevel,
+	.set_msglevel      = mlx5e_set_msglevel,
+	.get_fec_stats     = mlx5e_get_fec_stats,
+	.get_fecparam      = mlx5e_get_fecparam,
+	.set_fecparam      = mlx5e_set_fecparam,
+	.get_eth_phy_stats = mlx5e_get_eth_phy_stats,
+	.get_eth_mac_stats = mlx5e_get_eth_mac_stats,
+	.get_eth_ctrl_stats = mlx5e_get_eth_ctrl_stats,
+	.get_rmon_stats    = mlx5e_get_rmon_stats,
+};
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_fs.c b/drivers/net/ethernet/mellanox/mlx5/core/en_fs.c
new file mode 100644
index 000000000..eba601487
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_fs.c
@@ -0,0 +1,1582 @@
+/*
+ * Copyright (c) 2015, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/list.h>
+#include <linux/ip.h>
+#include <linux/ipv6.h>
+#include <linux/tcp.h>
+#include <linux/mlx5/fs.h>
+#include <linux/mlx5/mpfs.h>
+#include "en_tc.h"
+#include "lib/mpfs.h"
+#include "en/ptp.h"
+#include "en/fs_ethtool.h"
+
+struct mlx5e_flow_steering {
+	struct work_struct		set_rx_mode_work;
+	bool				state_destroy;
+	bool				vlan_strip_disable;
+	struct mlx5_core_dev		*mdev;
+	struct net_device		*netdev;
+	struct mlx5_flow_namespace      *ns;
+	struct mlx5_flow_namespace      *egress_ns;
+#ifdef CONFIG_MLX5_EN_RXNFC
+	struct mlx5e_ethtool_steering   *ethtool;
+#endif
+	struct mlx5e_tc_table           *tc;
+	struct mlx5e_promisc_table      promisc;
+	struct mlx5e_vlan_table         *vlan;
+	struct mlx5e_l2_table           l2;
+	struct mlx5_ttc_table           *ttc;
+	struct mlx5_ttc_table           *inner_ttc;
+#ifdef CONFIG_MLX5_EN_ARFS
+	struct mlx5e_arfs_tables       *arfs;
+#endif
+#ifdef CONFIG_MLX5_EN_TLS
+	struct mlx5e_accel_fs_tcp      *accel_tcp;
+#endif
+	struct mlx5e_fs_udp            *udp;
+	struct mlx5e_fs_any            *any;
+	struct mlx5e_ptp_fs            *ptp_fs;
+};
+
+static int mlx5e_add_l2_flow_rule(struct mlx5e_flow_steering *fs,
+				  struct mlx5e_l2_rule *ai, int type);
+static void mlx5e_del_l2_flow_rule(struct mlx5e_flow_steering *fs,
+				   struct mlx5e_l2_rule *ai);
+
+enum {
+	MLX5E_FULLMATCH = 0,
+	MLX5E_ALLMULTI  = 1,
+};
+
+enum {
+	MLX5E_UC        = 0,
+	MLX5E_MC_IPV4   = 1,
+	MLX5E_MC_IPV6   = 2,
+	MLX5E_MC_OTHER  = 3,
+};
+
+enum {
+	MLX5E_ACTION_NONE = 0,
+	MLX5E_ACTION_ADD  = 1,
+	MLX5E_ACTION_DEL  = 2,
+};
+
+struct mlx5e_l2_hash_node {
+	struct hlist_node          hlist;
+	u8                         action;
+	struct mlx5e_l2_rule ai;
+	bool   mpfs;
+};
+
+static inline int mlx5e_hash_l2(const u8 *addr)
+{
+	return addr[5];
+}
+
+static void mlx5e_add_l2_to_hash(struct hlist_head *hash, const u8 *addr)
+{
+	struct mlx5e_l2_hash_node *hn;
+	int ix = mlx5e_hash_l2(addr);
+	int found = 0;
+
+	hlist_for_each_entry(hn, &hash[ix], hlist)
+		if (ether_addr_equal_64bits(hn->ai.addr, addr)) {
+			found = 1;
+			break;
+		}
+
+	if (found) {
+		hn->action = MLX5E_ACTION_NONE;
+		return;
+	}
+
+	hn = kzalloc(sizeof(*hn), GFP_ATOMIC);
+	if (!hn)
+		return;
+
+	ether_addr_copy(hn->ai.addr, addr);
+	hn->action = MLX5E_ACTION_ADD;
+
+	hlist_add_head(&hn->hlist, &hash[ix]);
+}
+
+static void mlx5e_del_l2_from_hash(struct mlx5e_l2_hash_node *hn)
+{
+	hlist_del(&hn->hlist);
+	kfree(hn);
+}
+
+struct mlx5e_vlan_table {
+	struct mlx5e_flow_table		ft;
+	DECLARE_BITMAP(active_cvlans, VLAN_N_VID);
+	DECLARE_BITMAP(active_svlans, VLAN_N_VID);
+	struct mlx5_flow_handle	*active_cvlans_rule[VLAN_N_VID];
+	struct mlx5_flow_handle	*active_svlans_rule[VLAN_N_VID];
+	struct mlx5_flow_handle	*untagged_rule;
+	struct mlx5_flow_handle	*any_cvlan_rule;
+	struct mlx5_flow_handle	*any_svlan_rule;
+	struct mlx5_flow_handle	*trap_rule;
+	bool			cvlan_filter_disabled;
+};
+
+unsigned long *mlx5e_vlan_get_active_svlans(struct mlx5e_vlan_table *vlan)
+{
+	return vlan->active_svlans;
+}
+
+struct mlx5_flow_table *mlx5e_vlan_get_flowtable(struct mlx5e_vlan_table *vlan)
+{
+	return vlan->ft.t;
+}
+
+static int mlx5e_vport_context_update_vlans(struct mlx5e_flow_steering *fs)
+{
+	int max_list_size;
+	int list_size;
+	u16 *vlans;
+	int vlan;
+	int err;
+	int i;
+
+	list_size = 0;
+	for_each_set_bit(vlan, fs->vlan->active_cvlans, VLAN_N_VID)
+		list_size++;
+
+	max_list_size = 1 << MLX5_CAP_GEN(fs->mdev, log_max_vlan_list);
+
+	if (list_size > max_list_size) {
+		fs_warn(fs, "netdev vlans list size (%d) > (%d) max vport list size, some vlans will be dropped\n",
+			list_size, max_list_size);
+		list_size = max_list_size;
+	}
+
+	vlans = kvcalloc(list_size, sizeof(*vlans), GFP_KERNEL);
+	if (!vlans)
+		return -ENOMEM;
+
+	i = 0;
+	for_each_set_bit(vlan, fs->vlan->active_cvlans, VLAN_N_VID) {
+		if (i >= list_size)
+			break;
+		vlans[i++] = vlan;
+	}
+
+	err = mlx5_modify_nic_vport_vlans(fs->mdev, vlans, list_size);
+	if (err)
+		fs_err(fs, "Failed to modify vport vlans list err(%d)\n",
+		       err);
+
+	kvfree(vlans);
+	return err;
+}
+
+enum mlx5e_vlan_rule_type {
+	MLX5E_VLAN_RULE_TYPE_UNTAGGED,
+	MLX5E_VLAN_RULE_TYPE_ANY_CTAG_VID,
+	MLX5E_VLAN_RULE_TYPE_ANY_STAG_VID,
+	MLX5E_VLAN_RULE_TYPE_MATCH_CTAG_VID,
+	MLX5E_VLAN_RULE_TYPE_MATCH_STAG_VID,
+};
+
+static int __mlx5e_add_vlan_rule(struct mlx5e_flow_steering *fs,
+				 enum mlx5e_vlan_rule_type rule_type,
+				 u16 vid, struct mlx5_flow_spec *spec)
+{
+	struct mlx5_flow_table *ft = fs->vlan->ft.t;
+	struct mlx5_flow_destination dest = {};
+	struct mlx5_flow_handle **rule_p;
+	MLX5_DECLARE_FLOW_ACT(flow_act);
+	int err = 0;
+
+	dest.type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
+	dest.ft = fs->l2.ft.t;
+
+	spec->match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
+
+	switch (rule_type) {
+	case MLX5E_VLAN_RULE_TYPE_UNTAGGED:
+		/* cvlan_tag enabled in match criteria and
+		 * disabled in match value means both S & C tags
+		 * don't exist (untagged of both)
+		 */
+		rule_p = &fs->vlan->untagged_rule;
+		MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria,
+				 outer_headers.cvlan_tag);
+		break;
+	case MLX5E_VLAN_RULE_TYPE_ANY_CTAG_VID:
+		rule_p = &fs->vlan->any_cvlan_rule;
+		MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria,
+				 outer_headers.cvlan_tag);
+		MLX5_SET(fte_match_param, spec->match_value, outer_headers.cvlan_tag, 1);
+		break;
+	case MLX5E_VLAN_RULE_TYPE_ANY_STAG_VID:
+		rule_p = &fs->vlan->any_svlan_rule;
+		MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria,
+				 outer_headers.svlan_tag);
+		MLX5_SET(fte_match_param, spec->match_value, outer_headers.svlan_tag, 1);
+		break;
+	case MLX5E_VLAN_RULE_TYPE_MATCH_STAG_VID:
+		rule_p = &fs->vlan->active_svlans_rule[vid];
+		MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria,
+				 outer_headers.svlan_tag);
+		MLX5_SET(fte_match_param, spec->match_value, outer_headers.svlan_tag, 1);
+		MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria,
+				 outer_headers.first_vid);
+		MLX5_SET(fte_match_param, spec->match_value, outer_headers.first_vid,
+			 vid);
+		break;
+	default: /* MLX5E_VLAN_RULE_TYPE_MATCH_CTAG_VID */
+		rule_p = &fs->vlan->active_cvlans_rule[vid];
+		MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria,
+				 outer_headers.cvlan_tag);
+		MLX5_SET(fte_match_param, spec->match_value, outer_headers.cvlan_tag, 1);
+		MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria,
+				 outer_headers.first_vid);
+		MLX5_SET(fte_match_param, spec->match_value, outer_headers.first_vid,
+			 vid);
+		break;
+	}
+
+	if (WARN_ONCE(*rule_p, "VLAN rule already exists type %d", rule_type))
+		return 0;
+
+	*rule_p = mlx5_add_flow_rules(ft, spec, &flow_act, &dest, 1);
+
+	if (IS_ERR(*rule_p)) {
+		err = PTR_ERR(*rule_p);
+		*rule_p = NULL;
+		fs_err(fs, "%s: add rule failed\n", __func__);
+	}
+
+	return err;
+}
+
+static int mlx5e_add_vlan_rule(struct mlx5e_flow_steering *fs,
+			       enum mlx5e_vlan_rule_type rule_type, u16 vid)
+{
+	struct mlx5_flow_spec *spec;
+	int err = 0;
+
+	spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
+	if (!spec)
+		return -ENOMEM;
+
+	if (rule_type == MLX5E_VLAN_RULE_TYPE_MATCH_CTAG_VID)
+		mlx5e_vport_context_update_vlans(fs);
+
+	err = __mlx5e_add_vlan_rule(fs, rule_type, vid, spec);
+
+	kvfree(spec);
+
+	return err;
+}
+
+static void mlx5e_fs_del_vlan_rule(struct mlx5e_flow_steering *fs,
+				   enum mlx5e_vlan_rule_type rule_type, u16 vid)
+{
+	switch (rule_type) {
+	case MLX5E_VLAN_RULE_TYPE_UNTAGGED:
+		if (fs->vlan->untagged_rule) {
+			mlx5_del_flow_rules(fs->vlan->untagged_rule);
+			fs->vlan->untagged_rule = NULL;
+		}
+		break;
+	case MLX5E_VLAN_RULE_TYPE_ANY_CTAG_VID:
+		if (fs->vlan->any_cvlan_rule) {
+			mlx5_del_flow_rules(fs->vlan->any_cvlan_rule);
+			fs->vlan->any_cvlan_rule = NULL;
+		}
+		break;
+	case MLX5E_VLAN_RULE_TYPE_ANY_STAG_VID:
+		if (fs->vlan->any_svlan_rule) {
+			mlx5_del_flow_rules(fs->vlan->any_svlan_rule);
+			fs->vlan->any_svlan_rule = NULL;
+		}
+		break;
+	case MLX5E_VLAN_RULE_TYPE_MATCH_STAG_VID:
+		if (fs->vlan->active_svlans_rule[vid]) {
+			mlx5_del_flow_rules(fs->vlan->active_svlans_rule[vid]);
+			fs->vlan->active_svlans_rule[vid] = NULL;
+		}
+		break;
+	case MLX5E_VLAN_RULE_TYPE_MATCH_CTAG_VID:
+		if (fs->vlan->active_cvlans_rule[vid]) {
+			mlx5_del_flow_rules(fs->vlan->active_cvlans_rule[vid]);
+			fs->vlan->active_cvlans_rule[vid] = NULL;
+		}
+		mlx5e_vport_context_update_vlans(fs);
+		break;
+	}
+}
+
+static void mlx5e_fs_del_any_vid_rules(struct mlx5e_flow_steering *fs)
+{
+	mlx5e_fs_del_vlan_rule(fs, MLX5E_VLAN_RULE_TYPE_ANY_CTAG_VID, 0);
+	mlx5e_fs_del_vlan_rule(fs, MLX5E_VLAN_RULE_TYPE_ANY_STAG_VID, 0);
+}
+
+static int mlx5e_fs_add_any_vid_rules(struct mlx5e_flow_steering *fs)
+{
+	int err;
+
+	err = mlx5e_add_vlan_rule(fs, MLX5E_VLAN_RULE_TYPE_ANY_CTAG_VID, 0);
+	if (err)
+		return err;
+
+	return mlx5e_add_vlan_rule(fs, MLX5E_VLAN_RULE_TYPE_ANY_STAG_VID, 0);
+}
+
+static struct mlx5_flow_handle *
+mlx5e_add_trap_rule(struct mlx5_flow_table *ft, int trap_id, int tir_num)
+{
+	struct mlx5_flow_destination dest = {};
+	MLX5_DECLARE_FLOW_ACT(flow_act);
+	struct mlx5_flow_handle *rule;
+	struct mlx5_flow_spec *spec;
+
+	spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
+	if (!spec)
+		return ERR_PTR(-ENOMEM);
+	spec->flow_context.flags |= FLOW_CONTEXT_HAS_TAG;
+	spec->flow_context.flow_tag = trap_id;
+	dest.type = MLX5_FLOW_DESTINATION_TYPE_TIR;
+	dest.tir_num = tir_num;
+
+	rule = mlx5_add_flow_rules(ft, spec, &flow_act, &dest, 1);
+	kvfree(spec);
+	return rule;
+}
+
+int mlx5e_add_vlan_trap(struct mlx5e_flow_steering *fs, int trap_id, int tir_num)
+{
+	struct mlx5_flow_table *ft = fs->vlan->ft.t;
+	struct mlx5_flow_handle *rule;
+	int err;
+
+	rule = mlx5e_add_trap_rule(ft, trap_id, tir_num);
+	if (IS_ERR(rule)) {
+		err = PTR_ERR(rule);
+		fs->vlan->trap_rule = NULL;
+		fs_err(fs, "%s: add VLAN trap rule failed, err %d\n",
+		       __func__, err);
+		return err;
+	}
+	fs->vlan->trap_rule = rule;
+	return 0;
+}
+
+void mlx5e_remove_vlan_trap(struct mlx5e_flow_steering *fs)
+{
+	if (fs->vlan->trap_rule) {
+		mlx5_del_flow_rules(fs->vlan->trap_rule);
+		fs->vlan->trap_rule = NULL;
+	}
+}
+
+int mlx5e_add_mac_trap(struct mlx5e_flow_steering *fs, int trap_id, int tir_num)
+{
+	struct mlx5_flow_table *ft = fs->l2.ft.t;
+	struct mlx5_flow_handle *rule;
+	int err;
+
+	rule = mlx5e_add_trap_rule(ft, trap_id, tir_num);
+	if (IS_ERR(rule)) {
+		err = PTR_ERR(rule);
+		fs->l2.trap_rule = NULL;
+		fs_err(fs, "%s: add MAC trap rule failed, err %d\n",
+		       __func__, err);
+		return err;
+	}
+	fs->l2.trap_rule = rule;
+	return 0;
+}
+
+void mlx5e_remove_mac_trap(struct mlx5e_flow_steering *fs)
+{
+	if (fs->l2.trap_rule) {
+		mlx5_del_flow_rules(fs->l2.trap_rule);
+		fs->l2.trap_rule = NULL;
+	}
+}
+
+void mlx5e_enable_cvlan_filter(struct mlx5e_flow_steering *fs, bool promisc)
+{
+	if (!fs->vlan->cvlan_filter_disabled)
+		return;
+
+	fs->vlan->cvlan_filter_disabled = false;
+	if (promisc)
+		return;
+	mlx5e_fs_del_vlan_rule(fs, MLX5E_VLAN_RULE_TYPE_ANY_CTAG_VID, 0);
+}
+
+void mlx5e_disable_cvlan_filter(struct mlx5e_flow_steering *fs, bool promisc)
+{
+	if (!fs->vlan || fs->vlan->cvlan_filter_disabled)
+		return;
+
+	fs->vlan->cvlan_filter_disabled = true;
+	if (promisc)
+		return;
+	mlx5e_add_vlan_rule(fs, MLX5E_VLAN_RULE_TYPE_ANY_CTAG_VID, 0);
+}
+
+static int mlx5e_vlan_rx_add_cvid(struct mlx5e_flow_steering *fs, u16 vid)
+{
+	int err;
+
+	set_bit(vid, fs->vlan->active_cvlans);
+
+	err = mlx5e_add_vlan_rule(fs, MLX5E_VLAN_RULE_TYPE_MATCH_CTAG_VID, vid);
+	if (err)
+		clear_bit(vid, fs->vlan->active_cvlans);
+
+	return err;
+}
+
+static int mlx5e_vlan_rx_add_svid(struct mlx5e_flow_steering *fs,
+				  struct net_device *netdev, u16 vid)
+{
+	int err;
+
+	set_bit(vid, fs->vlan->active_svlans);
+
+	err = mlx5e_add_vlan_rule(fs, MLX5E_VLAN_RULE_TYPE_MATCH_STAG_VID, vid);
+	if (err) {
+		clear_bit(vid, fs->vlan->active_svlans);
+		return err;
+	}
+
+	/* Need to fix some features.. */
+	netdev_update_features(netdev);
+	return err;
+}
+
+int mlx5e_fs_vlan_rx_add_vid(struct mlx5e_flow_steering *fs,
+			     struct net_device *netdev,
+			     __be16 proto, u16 vid)
+{
+
+	if (!fs->vlan) {
+		fs_err(fs, "Vlan doesn't exist\n");
+		return -EINVAL;
+	}
+
+	if (be16_to_cpu(proto) == ETH_P_8021Q)
+		return mlx5e_vlan_rx_add_cvid(fs, vid);
+	else if (be16_to_cpu(proto) == ETH_P_8021AD)
+		return mlx5e_vlan_rx_add_svid(fs, netdev, vid);
+
+	return -EOPNOTSUPP;
+}
+
+int mlx5e_fs_vlan_rx_kill_vid(struct mlx5e_flow_steering *fs,
+			      struct net_device *netdev,
+			      __be16 proto, u16 vid)
+{
+	if (!fs->vlan) {
+		fs_err(fs, "Vlan doesn't exist\n");
+		return -EINVAL;
+	}
+
+	if (be16_to_cpu(proto) == ETH_P_8021Q) {
+		clear_bit(vid, fs->vlan->active_cvlans);
+		mlx5e_fs_del_vlan_rule(fs, MLX5E_VLAN_RULE_TYPE_MATCH_CTAG_VID, vid);
+	} else if (be16_to_cpu(proto) == ETH_P_8021AD) {
+		clear_bit(vid, fs->vlan->active_svlans);
+		mlx5e_fs_del_vlan_rule(fs, MLX5E_VLAN_RULE_TYPE_MATCH_STAG_VID, vid);
+		netdev_update_features(netdev);
+	}
+
+	return 0;
+}
+
+static void mlx5e_fs_add_vlan_rules(struct mlx5e_flow_steering *fs)
+{
+	int i;
+
+	mlx5e_add_vlan_rule(fs, MLX5E_VLAN_RULE_TYPE_UNTAGGED, 0);
+
+	for_each_set_bit(i, fs->vlan->active_cvlans, VLAN_N_VID) {
+		mlx5e_add_vlan_rule(fs, MLX5E_VLAN_RULE_TYPE_MATCH_CTAG_VID, i);
+	}
+
+	for_each_set_bit(i, fs->vlan->active_svlans, VLAN_N_VID)
+		mlx5e_add_vlan_rule(fs, MLX5E_VLAN_RULE_TYPE_MATCH_STAG_VID, i);
+
+	if (fs->vlan->cvlan_filter_disabled)
+		mlx5e_fs_add_any_vid_rules(fs);
+}
+
+static void mlx5e_del_vlan_rules(struct mlx5e_flow_steering *fs)
+{
+	int i;
+
+	mlx5e_fs_del_vlan_rule(fs, MLX5E_VLAN_RULE_TYPE_UNTAGGED, 0);
+
+	for_each_set_bit(i, fs->vlan->active_cvlans, VLAN_N_VID) {
+		mlx5e_fs_del_vlan_rule(fs, MLX5E_VLAN_RULE_TYPE_MATCH_CTAG_VID, i);
+	}
+
+	for_each_set_bit(i, fs->vlan->active_svlans, VLAN_N_VID)
+		mlx5e_fs_del_vlan_rule(fs, MLX5E_VLAN_RULE_TYPE_MATCH_STAG_VID, i);
+
+	WARN_ON_ONCE(fs->state_destroy);
+
+	mlx5e_remove_vlan_trap(fs);
+
+	/* must be called after DESTROY bit is set and
+	 * set_rx_mode is called and flushed
+	 */
+	if (fs->vlan->cvlan_filter_disabled)
+		mlx5e_fs_del_any_vid_rules(fs);
+}
+
+#define mlx5e_for_each_hash_node(hn, tmp, hash, i) \
+	for (i = 0; i < MLX5E_L2_ADDR_HASH_SIZE; i++) \
+		hlist_for_each_entry_safe(hn, tmp, &hash[i], hlist)
+
+static void mlx5e_execute_l2_action(struct mlx5e_flow_steering *fs,
+				    struct mlx5e_l2_hash_node *hn)
+{
+	u8 action = hn->action;
+	u8 mac_addr[ETH_ALEN];
+	int l2_err = 0;
+
+	ether_addr_copy(mac_addr, hn->ai.addr);
+
+	switch (action) {
+	case MLX5E_ACTION_ADD:
+		mlx5e_add_l2_flow_rule(fs, &hn->ai, MLX5E_FULLMATCH);
+		if (!is_multicast_ether_addr(mac_addr)) {
+			l2_err = mlx5_mpfs_add_mac(fs->mdev, mac_addr);
+			hn->mpfs = !l2_err;
+		}
+		hn->action = MLX5E_ACTION_NONE;
+		break;
+
+	case MLX5E_ACTION_DEL:
+		if (!is_multicast_ether_addr(mac_addr) && hn->mpfs)
+			l2_err = mlx5_mpfs_del_mac(fs->mdev, mac_addr);
+		mlx5e_del_l2_flow_rule(fs, &hn->ai);
+		mlx5e_del_l2_from_hash(hn);
+		break;
+	}
+
+	if (l2_err)
+		fs_warn(fs, "MPFS, failed to %s mac %pM, err(%d)\n",
+			action == MLX5E_ACTION_ADD ? "add" : "del",
+			mac_addr, l2_err);
+}
+
+static void mlx5e_sync_netdev_addr(struct mlx5e_flow_steering *fs,
+				   struct net_device *netdev)
+{
+	struct netdev_hw_addr *ha;
+
+	netif_addr_lock_bh(netdev);
+
+	mlx5e_add_l2_to_hash(fs->l2.netdev_uc, netdev->dev_addr);
+	netdev_for_each_uc_addr(ha, netdev)
+		mlx5e_add_l2_to_hash(fs->l2.netdev_uc, ha->addr);
+
+	netdev_for_each_mc_addr(ha, netdev)
+		mlx5e_add_l2_to_hash(fs->l2.netdev_mc, ha->addr);
+
+	netif_addr_unlock_bh(netdev);
+}
+
+static void mlx5e_fill_addr_array(struct mlx5e_flow_steering *fs, int list_type,
+				  struct net_device *ndev,
+				  u8 addr_array[][ETH_ALEN], int size)
+{
+	bool is_uc = (list_type == MLX5_NVPRT_LIST_TYPE_UC);
+	struct mlx5e_l2_hash_node *hn;
+	struct hlist_head *addr_list;
+	struct hlist_node *tmp;
+	int i = 0;
+	int hi;
+
+	addr_list = is_uc ? fs->l2.netdev_uc : fs->l2.netdev_mc;
+
+	if (is_uc) /* Make sure our own address is pushed first */
+		ether_addr_copy(addr_array[i++], ndev->dev_addr);
+	else if (fs->l2.broadcast_enabled)
+		ether_addr_copy(addr_array[i++], ndev->broadcast);
+
+	mlx5e_for_each_hash_node(hn, tmp, addr_list, hi) {
+		if (ether_addr_equal(ndev->dev_addr, hn->ai.addr))
+			continue;
+		if (i >= size)
+			break;
+		ether_addr_copy(addr_array[i++], hn->ai.addr);
+	}
+}
+
+static void mlx5e_vport_context_update_addr_list(struct mlx5e_flow_steering *fs,
+						 struct net_device *netdev,
+						 int list_type)
+{
+	bool is_uc = (list_type == MLX5_NVPRT_LIST_TYPE_UC);
+	struct mlx5e_l2_hash_node *hn;
+	u8 (*addr_array)[ETH_ALEN] = NULL;
+	struct hlist_head *addr_list;
+	struct hlist_node *tmp;
+	int max_size;
+	int size;
+	int err;
+	int hi;
+
+	size = is_uc ? 0 : (fs->l2.broadcast_enabled ? 1 : 0);
+	max_size = is_uc ?
+		1 << MLX5_CAP_GEN(fs->mdev, log_max_current_uc_list) :
+		1 << MLX5_CAP_GEN(fs->mdev, log_max_current_mc_list);
+
+	addr_list = is_uc ? fs->l2.netdev_uc : fs->l2.netdev_mc;
+	mlx5e_for_each_hash_node(hn, tmp, addr_list, hi)
+		size++;
+
+	if (size > max_size) {
+		fs_warn(fs, "mdev %s list size (%d) > (%d) max vport list size, some addresses will be dropped\n",
+			is_uc ? "UC" : "MC", size, max_size);
+		size = max_size;
+	}
+
+	if (size) {
+		addr_array = kcalloc(size, ETH_ALEN, GFP_KERNEL);
+		if (!addr_array) {
+			err = -ENOMEM;
+			goto out;
+		}
+		mlx5e_fill_addr_array(fs, list_type, netdev, addr_array, size);
+	}
+
+	err = mlx5_modify_nic_vport_mac_list(fs->mdev, list_type, addr_array, size);
+out:
+	if (err)
+		fs_err(fs, "Failed to modify vport %s list err(%d)\n",
+		       is_uc ? "UC" : "MC", err);
+	kfree(addr_array);
+}
+
+static void mlx5e_vport_context_update(struct mlx5e_flow_steering *fs,
+				       struct net_device *netdev)
+{
+	struct mlx5e_l2_table *ea = &fs->l2;
+
+	mlx5e_vport_context_update_addr_list(fs, netdev, MLX5_NVPRT_LIST_TYPE_UC);
+	mlx5e_vport_context_update_addr_list(fs, netdev, MLX5_NVPRT_LIST_TYPE_MC);
+	mlx5_modify_nic_vport_promisc(fs->mdev, 0,
+				      ea->allmulti_enabled,
+				      ea->promisc_enabled);
+}
+
+static void mlx5e_apply_netdev_addr(struct mlx5e_flow_steering *fs)
+{
+	struct mlx5e_l2_hash_node *hn;
+	struct hlist_node *tmp;
+	int i;
+
+	mlx5e_for_each_hash_node(hn, tmp, fs->l2.netdev_uc, i)
+		mlx5e_execute_l2_action(fs, hn);
+
+	mlx5e_for_each_hash_node(hn, tmp, fs->l2.netdev_mc, i)
+		mlx5e_execute_l2_action(fs, hn);
+}
+
+static void mlx5e_handle_netdev_addr(struct mlx5e_flow_steering *fs,
+				     struct net_device *netdev)
+{
+	struct mlx5e_l2_hash_node *hn;
+	struct hlist_node *tmp;
+	int i;
+
+	mlx5e_for_each_hash_node(hn, tmp, fs->l2.netdev_uc, i)
+		hn->action = MLX5E_ACTION_DEL;
+	mlx5e_for_each_hash_node(hn, tmp, fs->l2.netdev_mc, i)
+		hn->action = MLX5E_ACTION_DEL;
+
+	if (fs->state_destroy)
+		mlx5e_sync_netdev_addr(fs, netdev);
+
+	mlx5e_apply_netdev_addr(fs);
+}
+
+#define MLX5E_PROMISC_GROUP0_SIZE BIT(0)
+#define MLX5E_PROMISC_TABLE_SIZE MLX5E_PROMISC_GROUP0_SIZE
+
+static int mlx5e_add_promisc_rule(struct mlx5e_flow_steering *fs)
+{
+	struct mlx5_flow_table *ft = fs->promisc.ft.t;
+	struct mlx5_flow_destination dest = {};
+	struct mlx5_flow_handle **rule_p;
+	MLX5_DECLARE_FLOW_ACT(flow_act);
+	struct mlx5_flow_spec *spec;
+	int err = 0;
+
+	spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
+	if (!spec)
+		return -ENOMEM;
+	dest.type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
+	dest.ft = mlx5_get_ttc_flow_table(fs->ttc);
+
+	rule_p = &fs->promisc.rule;
+	*rule_p = mlx5_add_flow_rules(ft, spec, &flow_act, &dest, 1);
+	if (IS_ERR(*rule_p)) {
+		err = PTR_ERR(*rule_p);
+		*rule_p = NULL;
+		fs_err(fs, "%s: add promiscuous rule failed\n", __func__);
+	}
+	kvfree(spec);
+	return err;
+}
+
+static int mlx5e_create_promisc_table(struct mlx5e_flow_steering *fs)
+{
+	struct mlx5e_flow_table *ft = &fs->promisc.ft;
+	struct mlx5_flow_table_attr ft_attr = {};
+	int err;
+
+	ft_attr.max_fte = MLX5E_PROMISC_TABLE_SIZE;
+	ft_attr.autogroup.max_num_groups = 1;
+	ft_attr.level = MLX5E_PROMISC_FT_LEVEL;
+	ft_attr.prio = MLX5E_NIC_PRIO;
+
+	ft->t = mlx5_create_auto_grouped_flow_table(fs->ns, &ft_attr);
+	if (IS_ERR(ft->t)) {
+		err = PTR_ERR(ft->t);
+		ft->t = NULL;
+		fs_err(fs, "fail to create promisc table err=%d\n", err);
+		return err;
+	}
+
+	err = mlx5e_add_promisc_rule(fs);
+	if (err)
+		goto err_destroy_promisc_table;
+
+	return 0;
+
+err_destroy_promisc_table:
+	mlx5_destroy_flow_table(ft->t);
+	ft->t = NULL;
+
+	return err;
+}
+
+static void mlx5e_del_promisc_rule(struct mlx5e_flow_steering *fs)
+{
+	if (WARN(!fs->promisc.rule, "Trying to remove non-existing promiscuous rule"))
+		return;
+	mlx5_del_flow_rules(fs->promisc.rule);
+	fs->promisc.rule = NULL;
+}
+
+static void mlx5e_destroy_promisc_table(struct mlx5e_flow_steering *fs)
+{
+	if (!fs->promisc.ft.t)
+		return;
+	mlx5e_del_promisc_rule(fs);
+	mlx5_destroy_flow_table(fs->promisc.ft.t);
+	fs->promisc.ft.t = NULL;
+}
+
+void mlx5e_fs_set_rx_mode_work(struct mlx5e_flow_steering *fs,
+			       struct net_device *netdev)
+{
+	struct mlx5e_l2_table *ea = &fs->l2;
+
+	bool rx_mode_enable  = fs->state_destroy;
+	bool promisc_enabled   = rx_mode_enable && (netdev->flags & IFF_PROMISC);
+	bool allmulti_enabled  = rx_mode_enable && (netdev->flags & IFF_ALLMULTI);
+	bool broadcast_enabled = rx_mode_enable;
+
+	bool enable_promisc    = !ea->promisc_enabled   &&  promisc_enabled;
+	bool disable_promisc   =  ea->promisc_enabled   && !promisc_enabled;
+	bool enable_allmulti   = !ea->allmulti_enabled  &&  allmulti_enabled;
+	bool disable_allmulti  =  ea->allmulti_enabled  && !allmulti_enabled;
+	bool enable_broadcast  = !ea->broadcast_enabled &&  broadcast_enabled;
+	bool disable_broadcast =  ea->broadcast_enabled && !broadcast_enabled;
+	int err;
+
+	if (enable_promisc) {
+		err = mlx5e_create_promisc_table(fs);
+		if (err)
+			enable_promisc = false;
+		if (!fs->vlan_strip_disable && !err)
+			fs_warn_once(fs,
+				     "S-tagged traffic will be dropped while C-tag vlan stripping is enabled\n");
+	}
+	if (enable_allmulti)
+		mlx5e_add_l2_flow_rule(fs, &ea->allmulti, MLX5E_ALLMULTI);
+	if (enable_broadcast)
+		mlx5e_add_l2_flow_rule(fs, &ea->broadcast, MLX5E_FULLMATCH);
+
+	mlx5e_handle_netdev_addr(fs, netdev);
+
+	if (disable_broadcast)
+		mlx5e_del_l2_flow_rule(fs, &ea->broadcast);
+	if (disable_allmulti)
+		mlx5e_del_l2_flow_rule(fs, &ea->allmulti);
+	if (disable_promisc)
+		mlx5e_destroy_promisc_table(fs);
+
+	ea->promisc_enabled   = promisc_enabled;
+	ea->allmulti_enabled  = allmulti_enabled;
+	ea->broadcast_enabled = broadcast_enabled;
+
+	mlx5e_vport_context_update(fs, netdev);
+}
+
+static void mlx5e_destroy_groups(struct mlx5e_flow_table *ft)
+{
+	int i;
+
+	for (i = ft->num_groups - 1; i >= 0; i--) {
+		if (!IS_ERR_OR_NULL(ft->g[i]))
+			mlx5_destroy_flow_group(ft->g[i]);
+		ft->g[i] = NULL;
+	}
+	ft->num_groups = 0;
+}
+
+void mlx5e_fs_init_l2_addr(struct mlx5e_flow_steering *fs, struct net_device *netdev)
+{
+	ether_addr_copy(fs->l2.broadcast.addr, netdev->broadcast);
+}
+
+void mlx5e_destroy_flow_table(struct mlx5e_flow_table *ft)
+{
+	mlx5e_destroy_groups(ft);
+	kfree(ft->g);
+	mlx5_destroy_flow_table(ft->t);
+	ft->t = NULL;
+}
+
+static void mlx5e_set_inner_ttc_params(struct mlx5e_flow_steering *fs,
+				       struct mlx5e_rx_res *rx_res,
+				       struct ttc_params *ttc_params)
+{
+	struct mlx5_flow_table_attr *ft_attr = &ttc_params->ft_attr;
+	int tt;
+
+	memset(ttc_params, 0, sizeof(*ttc_params));
+	ttc_params->ns = mlx5_get_flow_namespace(fs->mdev,
+						 MLX5_FLOW_NAMESPACE_KERNEL);
+	ft_attr->level = MLX5E_INNER_TTC_FT_LEVEL;
+	ft_attr->prio = MLX5E_NIC_PRIO;
+
+	for (tt = 0; tt < MLX5_NUM_TT; tt++) {
+		ttc_params->dests[tt].type = MLX5_FLOW_DESTINATION_TYPE_TIR;
+		ttc_params->dests[tt].tir_num =
+			tt == MLX5_TT_ANY ?
+				mlx5e_rx_res_get_tirn_direct(rx_res, 0) :
+				mlx5e_rx_res_get_tirn_rss_inner(rx_res,
+								tt);
+	}
+}
+
+void mlx5e_set_ttc_params(struct mlx5e_flow_steering *fs,
+			  struct mlx5e_rx_res *rx_res,
+			  struct ttc_params *ttc_params, bool tunnel)
+
+{
+	struct mlx5_flow_table_attr *ft_attr = &ttc_params->ft_attr;
+	int tt;
+
+	memset(ttc_params, 0, sizeof(*ttc_params));
+	ttc_params->ns = mlx5_get_flow_namespace(fs->mdev,
+						 MLX5_FLOW_NAMESPACE_KERNEL);
+	ft_attr->level = MLX5E_TTC_FT_LEVEL;
+	ft_attr->prio = MLX5E_NIC_PRIO;
+
+	for (tt = 0; tt < MLX5_NUM_TT; tt++) {
+		ttc_params->dests[tt].type = MLX5_FLOW_DESTINATION_TYPE_TIR;
+		ttc_params->dests[tt].tir_num =
+			tt == MLX5_TT_ANY ?
+				mlx5e_rx_res_get_tirn_direct(rx_res, 0) :
+				mlx5e_rx_res_get_tirn_rss(rx_res, tt);
+	}
+
+	ttc_params->inner_ttc = tunnel;
+	if (!tunnel || !mlx5_tunnel_inner_ft_supported(fs->mdev))
+		return;
+
+	for (tt = 0; tt < MLX5_NUM_TUNNEL_TT; tt++) {
+		ttc_params->tunnel_dests[tt].type =
+			MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
+		ttc_params->tunnel_dests[tt].ft =
+			mlx5_get_ttc_flow_table(fs->inner_ttc);
+	}
+}
+
+static void mlx5e_del_l2_flow_rule(struct mlx5e_flow_steering *fs,
+				   struct mlx5e_l2_rule *ai)
+{
+	if (!IS_ERR_OR_NULL(ai->rule)) {
+		mlx5_del_flow_rules(ai->rule);
+		ai->rule = NULL;
+	}
+}
+
+static int mlx5e_add_l2_flow_rule(struct mlx5e_flow_steering *fs,
+				  struct mlx5e_l2_rule *ai, int type)
+{
+	struct mlx5_flow_table *ft = fs->l2.ft.t;
+	struct mlx5_flow_destination dest = {};
+	MLX5_DECLARE_FLOW_ACT(flow_act);
+	struct mlx5_flow_spec *spec;
+	int err = 0;
+	u8 *mc_dmac;
+	u8 *mv_dmac;
+
+	spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
+	if (!spec)
+		return -ENOMEM;
+
+	mc_dmac = MLX5_ADDR_OF(fte_match_param, spec->match_criteria,
+			       outer_headers.dmac_47_16);
+	mv_dmac = MLX5_ADDR_OF(fte_match_param, spec->match_value,
+			       outer_headers.dmac_47_16);
+
+	dest.type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
+	dest.ft = mlx5_get_ttc_flow_table(fs->ttc);
+
+	switch (type) {
+	case MLX5E_FULLMATCH:
+		spec->match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
+		eth_broadcast_addr(mc_dmac);
+		ether_addr_copy(mv_dmac, ai->addr);
+		break;
+
+	case MLX5E_ALLMULTI:
+		spec->match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
+		mc_dmac[0] = 0x01;
+		mv_dmac[0] = 0x01;
+		break;
+	}
+
+	ai->rule = mlx5_add_flow_rules(ft, spec, &flow_act, &dest, 1);
+	if (IS_ERR(ai->rule)) {
+		fs_err(fs, "%s: add l2 rule(mac:%pM) failed\n", __func__, mv_dmac);
+		err = PTR_ERR(ai->rule);
+		ai->rule = NULL;
+	}
+
+	kvfree(spec);
+
+	return err;
+}
+
+#define MLX5E_NUM_L2_GROUPS	   3
+#define MLX5E_L2_GROUP1_SIZE	   BIT(15)
+#define MLX5E_L2_GROUP2_SIZE	   BIT(0)
+#define MLX5E_L2_GROUP_TRAP_SIZE   BIT(0) /* must be last */
+#define MLX5E_L2_TABLE_SIZE	   (MLX5E_L2_GROUP1_SIZE +\
+				    MLX5E_L2_GROUP2_SIZE +\
+				    MLX5E_L2_GROUP_TRAP_SIZE)
+static int mlx5e_create_l2_table_groups(struct mlx5e_l2_table *l2_table)
+{
+	int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
+	struct mlx5e_flow_table *ft = &l2_table->ft;
+	int ix = 0;
+	u8 *mc_dmac;
+	u32 *in;
+	int err;
+	u8 *mc;
+
+	ft->g = kcalloc(MLX5E_NUM_L2_GROUPS, sizeof(*ft->g), GFP_KERNEL);
+	if (!ft->g)
+		return -ENOMEM;
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in) {
+		kfree(ft->g);
+		return -ENOMEM;
+	}
+
+	mc = MLX5_ADDR_OF(create_flow_group_in, in, match_criteria);
+	mc_dmac = MLX5_ADDR_OF(fte_match_param, mc,
+			       outer_headers.dmac_47_16);
+	/* Flow Group for full match */
+	eth_broadcast_addr(mc_dmac);
+	MLX5_SET_CFG(in, match_criteria_enable, MLX5_MATCH_OUTER_HEADERS);
+	MLX5_SET_CFG(in, start_flow_index, ix);
+	ix += MLX5E_L2_GROUP1_SIZE;
+	MLX5_SET_CFG(in, end_flow_index, ix - 1);
+	ft->g[ft->num_groups] = mlx5_create_flow_group(ft->t, in);
+	if (IS_ERR(ft->g[ft->num_groups]))
+		goto err_destroy_groups;
+	ft->num_groups++;
+
+	/* Flow Group for allmulti */
+	eth_zero_addr(mc_dmac);
+	mc_dmac[0] = 0x01;
+	MLX5_SET_CFG(in, start_flow_index, ix);
+	ix += MLX5E_L2_GROUP2_SIZE;
+	MLX5_SET_CFG(in, end_flow_index, ix - 1);
+	ft->g[ft->num_groups] = mlx5_create_flow_group(ft->t, in);
+	if (IS_ERR(ft->g[ft->num_groups]))
+		goto err_destroy_groups;
+	ft->num_groups++;
+
+	/* Flow Group for l2 traps */
+	memset(in, 0, inlen);
+	MLX5_SET_CFG(in, start_flow_index, ix);
+	ix += MLX5E_L2_GROUP_TRAP_SIZE;
+	MLX5_SET_CFG(in, end_flow_index, ix - 1);
+	ft->g[ft->num_groups] = mlx5_create_flow_group(ft->t, in);
+	if (IS_ERR(ft->g[ft->num_groups]))
+		goto err_destroy_groups;
+	ft->num_groups++;
+
+	kvfree(in);
+	return 0;
+
+err_destroy_groups:
+	err = PTR_ERR(ft->g[ft->num_groups]);
+	ft->g[ft->num_groups] = NULL;
+	mlx5e_destroy_groups(ft);
+	kvfree(in);
+	kfree(ft->g);
+
+	return err;
+}
+
+static void mlx5e_destroy_l2_table(struct mlx5e_flow_steering *fs)
+{
+	mlx5e_destroy_flow_table(&fs->l2.ft);
+}
+
+static int mlx5e_create_l2_table(struct mlx5e_flow_steering *fs)
+{
+	struct mlx5e_l2_table *l2_table = &fs->l2;
+	struct mlx5e_flow_table *ft = &l2_table->ft;
+	struct mlx5_flow_table_attr ft_attr = {};
+	int err;
+
+	ft->num_groups = 0;
+
+	ft_attr.max_fte = MLX5E_L2_TABLE_SIZE;
+	ft_attr.level = MLX5E_L2_FT_LEVEL;
+	ft_attr.prio = MLX5E_NIC_PRIO;
+
+	ft->t = mlx5_create_flow_table(fs->ns, &ft_attr);
+	if (IS_ERR(ft->t)) {
+		err = PTR_ERR(ft->t);
+		ft->t = NULL;
+		return err;
+	}
+
+	err = mlx5e_create_l2_table_groups(l2_table);
+	if (err)
+		goto err_destroy_flow_table;
+
+	return 0;
+
+err_destroy_flow_table:
+	mlx5_destroy_flow_table(ft->t);
+	ft->t = NULL;
+
+	return err;
+}
+
+#define MLX5E_NUM_VLAN_GROUPS	5
+#define MLX5E_VLAN_GROUP0_SIZE	BIT(12)
+#define MLX5E_VLAN_GROUP1_SIZE	BIT(12)
+#define MLX5E_VLAN_GROUP2_SIZE	BIT(1)
+#define MLX5E_VLAN_GROUP3_SIZE	BIT(0)
+#define MLX5E_VLAN_GROUP_TRAP_SIZE BIT(0) /* must be last */
+#define MLX5E_VLAN_TABLE_SIZE	(MLX5E_VLAN_GROUP0_SIZE +\
+				 MLX5E_VLAN_GROUP1_SIZE +\
+				 MLX5E_VLAN_GROUP2_SIZE +\
+				 MLX5E_VLAN_GROUP3_SIZE +\
+				 MLX5E_VLAN_GROUP_TRAP_SIZE)
+
+static int __mlx5e_create_vlan_table_groups(struct mlx5e_flow_table *ft, u32 *in,
+					    int inlen)
+{
+	int err;
+	int ix = 0;
+	u8 *mc = MLX5_ADDR_OF(create_flow_group_in, in, match_criteria);
+
+	memset(in, 0, inlen);
+	MLX5_SET_CFG(in, match_criteria_enable, MLX5_MATCH_OUTER_HEADERS);
+	MLX5_SET_TO_ONES(fte_match_param, mc, outer_headers.cvlan_tag);
+	MLX5_SET_TO_ONES(fte_match_param, mc, outer_headers.first_vid);
+	MLX5_SET_CFG(in, start_flow_index, ix);
+	ix += MLX5E_VLAN_GROUP0_SIZE;
+	MLX5_SET_CFG(in, end_flow_index, ix - 1);
+	ft->g[ft->num_groups] = mlx5_create_flow_group(ft->t, in);
+	if (IS_ERR(ft->g[ft->num_groups]))
+		goto err_destroy_groups;
+	ft->num_groups++;
+
+	memset(in, 0, inlen);
+	MLX5_SET_CFG(in, match_criteria_enable, MLX5_MATCH_OUTER_HEADERS);
+	MLX5_SET_TO_ONES(fte_match_param, mc, outer_headers.svlan_tag);
+	MLX5_SET_TO_ONES(fte_match_param, mc, outer_headers.first_vid);
+	MLX5_SET_CFG(in, start_flow_index, ix);
+	ix += MLX5E_VLAN_GROUP1_SIZE;
+	MLX5_SET_CFG(in, end_flow_index, ix - 1);
+	ft->g[ft->num_groups] = mlx5_create_flow_group(ft->t, in);
+	if (IS_ERR(ft->g[ft->num_groups]))
+		goto err_destroy_groups;
+	ft->num_groups++;
+
+	memset(in, 0, inlen);
+	MLX5_SET_CFG(in, match_criteria_enable, MLX5_MATCH_OUTER_HEADERS);
+	MLX5_SET_TO_ONES(fte_match_param, mc, outer_headers.cvlan_tag);
+	MLX5_SET_CFG(in, start_flow_index, ix);
+	ix += MLX5E_VLAN_GROUP2_SIZE;
+	MLX5_SET_CFG(in, end_flow_index, ix - 1);
+	ft->g[ft->num_groups] = mlx5_create_flow_group(ft->t, in);
+	if (IS_ERR(ft->g[ft->num_groups]))
+		goto err_destroy_groups;
+	ft->num_groups++;
+
+	memset(in, 0, inlen);
+	MLX5_SET_CFG(in, match_criteria_enable, MLX5_MATCH_OUTER_HEADERS);
+	MLX5_SET_TO_ONES(fte_match_param, mc, outer_headers.svlan_tag);
+	MLX5_SET_CFG(in, start_flow_index, ix);
+	ix += MLX5E_VLAN_GROUP3_SIZE;
+	MLX5_SET_CFG(in, end_flow_index, ix - 1);
+	ft->g[ft->num_groups] = mlx5_create_flow_group(ft->t, in);
+	if (IS_ERR(ft->g[ft->num_groups]))
+		goto err_destroy_groups;
+	ft->num_groups++;
+
+	memset(in, 0, inlen);
+	MLX5_SET_CFG(in, start_flow_index, ix);
+	ix += MLX5E_VLAN_GROUP_TRAP_SIZE;
+	MLX5_SET_CFG(in, end_flow_index, ix - 1);
+	ft->g[ft->num_groups] = mlx5_create_flow_group(ft->t, in);
+	if (IS_ERR(ft->g[ft->num_groups]))
+		goto err_destroy_groups;
+	ft->num_groups++;
+
+	return 0;
+
+err_destroy_groups:
+	err = PTR_ERR(ft->g[ft->num_groups]);
+	ft->g[ft->num_groups] = NULL;
+	mlx5e_destroy_groups(ft);
+
+	return err;
+}
+
+static int mlx5e_create_vlan_table_groups(struct mlx5e_flow_table *ft)
+{
+	u32 *in;
+	int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
+	int err;
+
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in)
+		return -ENOMEM;
+
+	err = __mlx5e_create_vlan_table_groups(ft, in, inlen);
+
+	kvfree(in);
+	return err;
+}
+
+static int mlx5e_fs_create_vlan_table(struct mlx5e_flow_steering *fs)
+{
+	struct mlx5_flow_table_attr ft_attr = {};
+	struct mlx5e_flow_table *ft;
+	int err;
+
+	ft = &fs->vlan->ft;
+	ft->num_groups = 0;
+
+	ft_attr.max_fte = MLX5E_VLAN_TABLE_SIZE;
+	ft_attr.level = MLX5E_VLAN_FT_LEVEL;
+	ft_attr.prio = MLX5E_NIC_PRIO;
+
+	ft->t = mlx5_create_flow_table(fs->ns, &ft_attr);
+	if (IS_ERR(ft->t))
+		return PTR_ERR(ft->t);
+
+	ft->g = kcalloc(MLX5E_NUM_VLAN_GROUPS, sizeof(*ft->g), GFP_KERNEL);
+	if (!ft->g) {
+		err = -ENOMEM;
+		goto err_destroy_vlan_table;
+	}
+
+	err = mlx5e_create_vlan_table_groups(ft);
+	if (err)
+		goto err_free_g;
+
+	mlx5e_fs_add_vlan_rules(fs);
+
+	return 0;
+
+err_free_g:
+	kfree(ft->g);
+err_destroy_vlan_table:
+	mlx5_destroy_flow_table(ft->t);
+
+	return err;
+}
+
+static void mlx5e_destroy_vlan_table(struct mlx5e_flow_steering *fs)
+{
+	mlx5e_del_vlan_rules(fs);
+	mlx5e_destroy_flow_table(&fs->vlan->ft);
+}
+
+static void mlx5e_destroy_inner_ttc_table(struct mlx5e_flow_steering *fs)
+{
+	if (!mlx5_tunnel_inner_ft_supported(fs->mdev))
+		return;
+	mlx5_destroy_ttc_table(fs->inner_ttc);
+}
+
+void mlx5e_destroy_ttc_table(struct mlx5e_flow_steering *fs)
+{
+	mlx5_destroy_ttc_table(fs->ttc);
+}
+
+static int mlx5e_create_inner_ttc_table(struct mlx5e_flow_steering *fs,
+					struct mlx5e_rx_res *rx_res)
+{
+	struct ttc_params ttc_params = {};
+
+	if (!mlx5_tunnel_inner_ft_supported(fs->mdev))
+		return 0;
+
+	mlx5e_set_inner_ttc_params(fs, rx_res, &ttc_params);
+	fs->inner_ttc = mlx5_create_inner_ttc_table(fs->mdev,
+						    &ttc_params);
+	if (IS_ERR(fs->inner_ttc))
+		return PTR_ERR(fs->inner_ttc);
+	return 0;
+}
+
+int mlx5e_create_ttc_table(struct mlx5e_flow_steering *fs,
+			   struct mlx5e_rx_res *rx_res)
+{
+	struct ttc_params ttc_params = {};
+
+	mlx5e_set_ttc_params(fs, rx_res, &ttc_params, true);
+	fs->ttc = mlx5_create_ttc_table(fs->mdev, &ttc_params);
+	if (IS_ERR(fs->ttc))
+		return PTR_ERR(fs->ttc);
+	return 0;
+}
+
+int mlx5e_create_flow_steering(struct mlx5e_flow_steering *fs,
+			       struct mlx5e_rx_res *rx_res,
+			       const struct mlx5e_profile *profile,
+			       struct net_device *netdev)
+{
+	struct mlx5_flow_namespace *ns = mlx5_get_flow_namespace(fs->mdev,
+								 MLX5_FLOW_NAMESPACE_KERNEL);
+	int err;
+
+	if (!ns)
+		return -EOPNOTSUPP;
+
+	mlx5e_fs_set_ns(fs, ns, false);
+	err = mlx5e_arfs_create_tables(fs, rx_res,
+				       !!(netdev->hw_features & NETIF_F_NTUPLE));
+	if (err) {
+		fs_err(fs, "Failed to create arfs tables, err=%d\n", err);
+		netdev->hw_features &= ~NETIF_F_NTUPLE;
+	}
+
+	err = mlx5e_create_inner_ttc_table(fs, rx_res);
+	if (err) {
+		fs_err(fs, "Failed to create inner ttc table, err=%d\n", err);
+		goto err_destroy_arfs_tables;
+	}
+
+	err = mlx5e_create_ttc_table(fs, rx_res);
+	if (err) {
+		fs_err(fs, "Failed to create ttc table, err=%d\n", err);
+		goto err_destroy_inner_ttc_table;
+	}
+
+	err = mlx5e_create_l2_table(fs);
+	if (err) {
+		fs_err(fs, "Failed to create l2 table, err=%d\n", err);
+		goto err_destroy_ttc_table;
+	}
+
+	err = mlx5e_fs_create_vlan_table(fs);
+	if (err) {
+		fs_err(fs, "Failed to create vlan table, err=%d\n", err);
+		goto err_destroy_l2_table;
+	}
+
+	err = mlx5e_ptp_alloc_rx_fs(fs, profile);
+	if (err)
+		goto err_destory_vlan_table;
+
+	mlx5e_ethtool_init_steering(fs);
+
+	return 0;
+
+err_destory_vlan_table:
+	mlx5e_destroy_vlan_table(fs);
+err_destroy_l2_table:
+	mlx5e_destroy_l2_table(fs);
+err_destroy_ttc_table:
+	mlx5e_destroy_ttc_table(fs);
+err_destroy_inner_ttc_table:
+	mlx5e_destroy_inner_ttc_table(fs);
+err_destroy_arfs_tables:
+	mlx5e_arfs_destroy_tables(fs, !!(netdev->hw_features & NETIF_F_NTUPLE));
+
+	return err;
+}
+
+void mlx5e_destroy_flow_steering(struct mlx5e_flow_steering *fs, bool ntuple,
+				 const struct mlx5e_profile *profile)
+{
+	mlx5e_ptp_free_rx_fs(fs, profile);
+	mlx5e_destroy_vlan_table(fs);
+	mlx5e_destroy_l2_table(fs);
+	mlx5e_destroy_ttc_table(fs);
+	mlx5e_destroy_inner_ttc_table(fs);
+	mlx5e_arfs_destroy_tables(fs, ntuple);
+	mlx5e_ethtool_cleanup_steering(fs);
+}
+
+static int mlx5e_fs_vlan_alloc(struct mlx5e_flow_steering *fs)
+{
+	fs->vlan = kvzalloc(sizeof(*fs->vlan), GFP_KERNEL);
+	if (!fs->vlan)
+		return -ENOMEM;
+	return 0;
+}
+
+static void mlx5e_fs_vlan_free(struct mlx5e_flow_steering *fs)
+{
+	kvfree(fs->vlan);
+}
+
+struct mlx5e_vlan_table *mlx5e_fs_get_vlan(struct mlx5e_flow_steering *fs)
+{
+	return fs->vlan;
+}
+
+static int mlx5e_fs_tc_alloc(struct mlx5e_flow_steering *fs)
+{
+	fs->tc = mlx5e_tc_table_alloc();
+	if (IS_ERR(fs->tc))
+		return -ENOMEM;
+	return 0;
+}
+
+static void mlx5e_fs_tc_free(struct mlx5e_flow_steering *fs)
+{
+	mlx5e_tc_table_free(fs->tc);
+}
+
+struct mlx5e_tc_table *mlx5e_fs_get_tc(struct mlx5e_flow_steering *fs)
+{
+	return fs->tc;
+}
+
+#ifdef CONFIG_MLX5_EN_RXNFC
+static int mlx5e_fs_ethtool_alloc(struct mlx5e_flow_steering *fs)
+{
+	return mlx5e_ethtool_alloc(&fs->ethtool);
+}
+
+static void mlx5e_fs_ethtool_free(struct mlx5e_flow_steering *fs)
+{
+	mlx5e_ethtool_free(fs->ethtool);
+}
+
+struct mlx5e_ethtool_steering *mlx5e_fs_get_ethtool(struct mlx5e_flow_steering *fs)
+{
+	return fs->ethtool;
+}
+#else
+static int mlx5e_fs_ethtool_alloc(struct mlx5e_flow_steering *fs)
+{ return 0; }
+static void mlx5e_fs_ethtool_free(struct mlx5e_flow_steering *fs) { }
+#endif
+
+struct mlx5e_flow_steering *mlx5e_fs_init(const struct mlx5e_profile *profile,
+					  struct mlx5_core_dev *mdev,
+					  bool state_destroy)
+{
+	struct mlx5e_flow_steering *fs;
+	int err;
+
+	fs = kvzalloc(sizeof(*fs), GFP_KERNEL);
+	if (!fs)
+		goto err;
+
+	fs->mdev = mdev;
+	fs->state_destroy = state_destroy;
+	if (mlx5e_profile_feature_cap(profile, FS_VLAN)) {
+		err = mlx5e_fs_vlan_alloc(fs);
+		if (err)
+			goto err_free_fs;
+	}
+
+	if (mlx5e_profile_feature_cap(profile, FS_TC)) {
+		err = mlx5e_fs_tc_alloc(fs);
+		if (err)
+			goto err_free_vlan;
+	}
+
+	err = mlx5e_fs_ethtool_alloc(fs);
+	if (err)
+		goto err_free_tc;
+
+	return fs;
+err_free_tc:
+	mlx5e_fs_tc_free(fs);
+err_free_vlan:
+	mlx5e_fs_vlan_free(fs);
+err_free_fs:
+	kvfree(fs);
+err:
+	return NULL;
+}
+
+void mlx5e_fs_cleanup(struct mlx5e_flow_steering *fs)
+{
+	if (!fs)
+		return;
+	mlx5e_fs_ethtool_free(fs);
+	mlx5e_fs_tc_free(fs);
+	mlx5e_fs_vlan_free(fs);
+	kvfree(fs);
+}
+
+struct mlx5e_l2_table *mlx5e_fs_get_l2(struct mlx5e_flow_steering *fs)
+{
+	return &fs->l2;
+}
+
+struct mlx5_flow_namespace *mlx5e_fs_get_ns(struct mlx5e_flow_steering *fs, bool egress)
+{
+	return  egress ? fs->egress_ns : fs->ns;
+}
+
+void mlx5e_fs_set_ns(struct mlx5e_flow_steering *fs, struct mlx5_flow_namespace *ns, bool egress)
+{
+	if (!egress)
+		fs->ns = ns;
+	else
+		fs->egress_ns = ns;
+}
+
+struct mlx5_ttc_table *mlx5e_fs_get_ttc(struct mlx5e_flow_steering *fs, bool inner)
+{
+	return inner ? fs->inner_ttc : fs->ttc;
+}
+
+void mlx5e_fs_set_ttc(struct mlx5e_flow_steering *fs, struct mlx5_ttc_table *ttc, bool inner)
+{
+	if (!inner)
+		fs->ttc = ttc;
+	else
+		fs->inner_ttc = ttc;
+}
+
+#ifdef CONFIG_MLX5_EN_ARFS
+struct mlx5e_arfs_tables *mlx5e_fs_get_arfs(struct mlx5e_flow_steering *fs)
+{
+	return fs->arfs;
+}
+
+void mlx5e_fs_set_arfs(struct mlx5e_flow_steering *fs, struct mlx5e_arfs_tables *arfs)
+{
+	fs->arfs = arfs;
+}
+#endif
+
+struct mlx5e_ptp_fs *mlx5e_fs_get_ptp(struct mlx5e_flow_steering *fs)
+{
+	return fs->ptp_fs;
+}
+
+void mlx5e_fs_set_ptp(struct mlx5e_flow_steering *fs, struct mlx5e_ptp_fs *ptp_fs)
+{
+	fs->ptp_fs = ptp_fs;
+}
+
+struct mlx5e_fs_any *mlx5e_fs_get_any(struct mlx5e_flow_steering *fs)
+{
+	return fs->any;
+}
+
+void mlx5e_fs_set_any(struct mlx5e_flow_steering *fs, struct mlx5e_fs_any *any)
+{
+	fs->any = any;
+}
+
+#ifdef CONFIG_MLX5_EN_TLS
+struct mlx5e_accel_fs_tcp *mlx5e_fs_get_accel_tcp(struct mlx5e_flow_steering *fs)
+{
+	return fs->accel_tcp;
+}
+
+void mlx5e_fs_set_accel_tcp(struct mlx5e_flow_steering *fs, struct mlx5e_accel_fs_tcp *accel_tcp)
+{
+	fs->accel_tcp = accel_tcp;
+}
+#endif
+
+void mlx5e_fs_set_state_destroy(struct mlx5e_flow_steering *fs, bool state_destroy)
+{
+	fs->state_destroy = state_destroy;
+}
+
+void mlx5e_fs_set_vlan_strip_disable(struct mlx5e_flow_steering *fs,
+				     bool vlan_strip_disable)
+{
+	fs->vlan_strip_disable = vlan_strip_disable;
+}
+
+struct mlx5e_fs_udp *mlx5e_fs_get_udp(struct mlx5e_flow_steering *fs)
+{
+	return fs->udp;
+}
+
+void mlx5e_fs_set_udp(struct mlx5e_flow_steering *fs, struct mlx5e_fs_udp *udp)
+{
+	fs->udp = udp;
+}
+
+struct mlx5_core_dev *mlx5e_fs_get_mdev(struct mlx5e_flow_steering *fs)
+{
+	return fs->mdev;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_fs_ethtool.c b/drivers/net/ethernet/mellanox/mlx5/core/en_fs_ethtool.c
new file mode 100644
index 000000000..aac32e505
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_fs_ethtool.c
@@ -0,0 +1,1013 @@
+/*
+ * Copyright (c) 2016, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/mlx5/fs.h>
+#include "en.h"
+#include "en/params.h"
+#include "en/xsk/pool.h"
+#include "en/fs_ethtool.h"
+
+struct mlx5e_ethtool_table {
+	struct mlx5_flow_table *ft;
+	int                    num_rules;
+};
+
+#define ETHTOOL_NUM_L3_L4_FTS 7
+#define ETHTOOL_NUM_L2_FTS 4
+
+struct mlx5e_ethtool_steering {
+	struct mlx5e_ethtool_table      l3_l4_ft[ETHTOOL_NUM_L3_L4_FTS];
+	struct mlx5e_ethtool_table      l2_ft[ETHTOOL_NUM_L2_FTS];
+	struct list_head                rules;
+	int                             tot_num_rules;
+};
+
+static int flow_type_to_traffic_type(u32 flow_type);
+
+static u32 flow_type_mask(u32 flow_type)
+{
+	return flow_type & ~(FLOW_EXT | FLOW_MAC_EXT | FLOW_RSS);
+}
+
+struct mlx5e_ethtool_rule {
+	struct list_head             list;
+	struct ethtool_rx_flow_spec  flow_spec;
+	struct mlx5_flow_handle	     *rule;
+	struct mlx5e_ethtool_table   *eth_ft;
+	struct mlx5e_rss             *rss;
+};
+
+static void put_flow_table(struct mlx5e_ethtool_table *eth_ft)
+{
+	if (!--eth_ft->num_rules) {
+		mlx5_destroy_flow_table(eth_ft->ft);
+		eth_ft->ft = NULL;
+	}
+}
+
+#define MLX5E_ETHTOOL_L3_L4_PRIO 0
+#define MLX5E_ETHTOOL_L2_PRIO (MLX5E_ETHTOOL_L3_L4_PRIO + ETHTOOL_NUM_L3_L4_FTS)
+#define MLX5E_ETHTOOL_NUM_ENTRIES 64000
+#define MLX5E_ETHTOOL_NUM_GROUPS  10
+static struct mlx5e_ethtool_table *get_flow_table(struct mlx5e_priv *priv,
+						  struct ethtool_rx_flow_spec *fs,
+						  int num_tuples)
+{
+	struct mlx5e_ethtool_steering *ethtool = mlx5e_fs_get_ethtool(priv->fs);
+	struct mlx5_flow_table_attr ft_attr = {};
+	struct mlx5e_ethtool_table *eth_ft;
+	struct mlx5_flow_namespace *ns;
+	struct mlx5_flow_table *ft;
+	int max_tuples;
+	int table_size;
+	int prio;
+
+	switch (flow_type_mask(fs->flow_type)) {
+	case TCP_V4_FLOW:
+	case UDP_V4_FLOW:
+	case TCP_V6_FLOW:
+	case UDP_V6_FLOW:
+		max_tuples = ETHTOOL_NUM_L3_L4_FTS;
+		prio = MLX5E_ETHTOOL_L3_L4_PRIO + (max_tuples - num_tuples);
+		eth_ft = &ethtool->l3_l4_ft[prio];
+		break;
+	case IP_USER_FLOW:
+	case IPV6_USER_FLOW:
+		max_tuples = ETHTOOL_NUM_L3_L4_FTS;
+		prio = MLX5E_ETHTOOL_L3_L4_PRIO + (max_tuples - num_tuples);
+		eth_ft = &ethtool->l3_l4_ft[prio];
+		break;
+	case ETHER_FLOW:
+		max_tuples = ETHTOOL_NUM_L2_FTS;
+		prio = max_tuples - num_tuples;
+		eth_ft = &ethtool->l2_ft[prio];
+		prio += MLX5E_ETHTOOL_L2_PRIO;
+		break;
+	default:
+		return ERR_PTR(-EINVAL);
+	}
+
+	eth_ft->num_rules++;
+	if (eth_ft->ft)
+		return eth_ft;
+
+	ns = mlx5_get_flow_namespace(priv->mdev,
+				     MLX5_FLOW_NAMESPACE_ETHTOOL);
+	if (!ns)
+		return ERR_PTR(-EOPNOTSUPP);
+
+	table_size = min_t(u32, BIT(MLX5_CAP_FLOWTABLE(priv->mdev,
+						       flow_table_properties_nic_receive.log_max_ft_size)),
+			   MLX5E_ETHTOOL_NUM_ENTRIES);
+
+	ft_attr.prio = prio;
+	ft_attr.max_fte = table_size;
+	ft_attr.autogroup.max_num_groups = MLX5E_ETHTOOL_NUM_GROUPS;
+	ft = mlx5_create_auto_grouped_flow_table(ns, &ft_attr);
+	if (IS_ERR(ft))
+		return (void *)ft;
+
+	eth_ft->ft = ft;
+	return eth_ft;
+}
+
+static void mask_spec(u8 *mask, u8 *val, size_t size)
+{
+	unsigned int i;
+
+	for (i = 0; i < size; i++, mask++, val++)
+		*((u8 *)val) = *((u8 *)mask) & *((u8 *)val);
+}
+
+#define MLX5E_FTE_SET(header_p, fld, v)  \
+	MLX5_SET(fte_match_set_lyr_2_4, header_p, fld, v)
+
+#define MLX5E_FTE_ADDR_OF(header_p, fld) \
+	MLX5_ADDR_OF(fte_match_set_lyr_2_4, header_p, fld)
+
+static void
+set_ip4(void *headers_c, void *headers_v, __be32 ip4src_m,
+	__be32 ip4src_v, __be32 ip4dst_m, __be32 ip4dst_v)
+{
+	if (ip4src_m) {
+		memcpy(MLX5E_FTE_ADDR_OF(headers_v, src_ipv4_src_ipv6.ipv4_layout.ipv4),
+		       &ip4src_v, sizeof(ip4src_v));
+		memcpy(MLX5E_FTE_ADDR_OF(headers_c, src_ipv4_src_ipv6.ipv4_layout.ipv4),
+		       &ip4src_m, sizeof(ip4src_m));
+	}
+	if (ip4dst_m) {
+		memcpy(MLX5E_FTE_ADDR_OF(headers_v, dst_ipv4_dst_ipv6.ipv4_layout.ipv4),
+		       &ip4dst_v, sizeof(ip4dst_v));
+		memcpy(MLX5E_FTE_ADDR_OF(headers_c, dst_ipv4_dst_ipv6.ipv4_layout.ipv4),
+		       &ip4dst_m, sizeof(ip4dst_m));
+	}
+
+	MLX5E_FTE_SET(headers_c, ethertype, 0xffff);
+	MLX5E_FTE_SET(headers_v, ethertype, ETH_P_IP);
+}
+
+static void
+set_ip6(void *headers_c, void *headers_v, __be32 ip6src_m[4],
+	__be32 ip6src_v[4], __be32 ip6dst_m[4], __be32 ip6dst_v[4])
+{
+	u8 ip6_sz = MLX5_FLD_SZ_BYTES(ipv6_layout, ipv6);
+
+	if (!ipv6_addr_any((struct in6_addr *)ip6src_m)) {
+		memcpy(MLX5E_FTE_ADDR_OF(headers_v, src_ipv4_src_ipv6.ipv6_layout.ipv6),
+		       ip6src_v, ip6_sz);
+		memcpy(MLX5E_FTE_ADDR_OF(headers_c, src_ipv4_src_ipv6.ipv6_layout.ipv6),
+		       ip6src_m, ip6_sz);
+	}
+	if (!ipv6_addr_any((struct in6_addr *)ip6dst_m)) {
+		memcpy(MLX5E_FTE_ADDR_OF(headers_v, dst_ipv4_dst_ipv6.ipv6_layout.ipv6),
+		       ip6dst_v, ip6_sz);
+		memcpy(MLX5E_FTE_ADDR_OF(headers_c, dst_ipv4_dst_ipv6.ipv6_layout.ipv6),
+		       ip6dst_m, ip6_sz);
+	}
+
+	MLX5E_FTE_SET(headers_c, ethertype, 0xffff);
+	MLX5E_FTE_SET(headers_v, ethertype, ETH_P_IPV6);
+}
+
+static void
+set_tcp(void *headers_c, void *headers_v, __be16 psrc_m, __be16 psrc_v,
+	__be16 pdst_m, __be16 pdst_v)
+{
+	if (psrc_m) {
+		MLX5E_FTE_SET(headers_c, tcp_sport, ntohs(psrc_m));
+		MLX5E_FTE_SET(headers_v, tcp_sport, ntohs(psrc_v));
+	}
+	if (pdst_m) {
+		MLX5E_FTE_SET(headers_c, tcp_dport, ntohs(pdst_m));
+		MLX5E_FTE_SET(headers_v, tcp_dport, ntohs(pdst_v));
+	}
+
+	MLX5E_FTE_SET(headers_c, ip_protocol, 0xffff);
+	MLX5E_FTE_SET(headers_v, ip_protocol, IPPROTO_TCP);
+}
+
+static void
+set_udp(void *headers_c, void *headers_v, __be16 psrc_m, __be16 psrc_v,
+	__be16 pdst_m, __be16 pdst_v)
+{
+	if (psrc_m) {
+		MLX5E_FTE_SET(headers_c, udp_sport, ntohs(psrc_m));
+		MLX5E_FTE_SET(headers_v, udp_sport, ntohs(psrc_v));
+	}
+
+	if (pdst_m) {
+		MLX5E_FTE_SET(headers_c, udp_dport, ntohs(pdst_m));
+		MLX5E_FTE_SET(headers_v, udp_dport, ntohs(pdst_v));
+	}
+
+	MLX5E_FTE_SET(headers_c, ip_protocol, 0xffff);
+	MLX5E_FTE_SET(headers_v, ip_protocol, IPPROTO_UDP);
+}
+
+static void
+parse_tcp4(void *headers_c, void *headers_v, struct ethtool_rx_flow_spec *fs)
+{
+	struct ethtool_tcpip4_spec *l4_mask = &fs->m_u.tcp_ip4_spec;
+	struct ethtool_tcpip4_spec *l4_val  = &fs->h_u.tcp_ip4_spec;
+
+	set_ip4(headers_c, headers_v, l4_mask->ip4src, l4_val->ip4src,
+		l4_mask->ip4dst, l4_val->ip4dst);
+
+	set_tcp(headers_c, headers_v, l4_mask->psrc, l4_val->psrc,
+		l4_mask->pdst, l4_val->pdst);
+}
+
+static void
+parse_udp4(void *headers_c, void *headers_v, struct ethtool_rx_flow_spec *fs)
+{
+	struct ethtool_tcpip4_spec *l4_mask = &fs->m_u.udp_ip4_spec;
+	struct ethtool_tcpip4_spec *l4_val  = &fs->h_u.udp_ip4_spec;
+
+	set_ip4(headers_c, headers_v, l4_mask->ip4src, l4_val->ip4src,
+		l4_mask->ip4dst, l4_val->ip4dst);
+
+	set_udp(headers_c, headers_v, l4_mask->psrc, l4_val->psrc,
+		l4_mask->pdst, l4_val->pdst);
+}
+
+static void
+parse_ip4(void *headers_c, void *headers_v, struct ethtool_rx_flow_spec *fs)
+{
+	struct ethtool_usrip4_spec *l3_mask = &fs->m_u.usr_ip4_spec;
+	struct ethtool_usrip4_spec *l3_val  = &fs->h_u.usr_ip4_spec;
+
+	set_ip4(headers_c, headers_v, l3_mask->ip4src, l3_val->ip4src,
+		l3_mask->ip4dst, l3_val->ip4dst);
+
+	if (l3_mask->proto) {
+		MLX5E_FTE_SET(headers_c, ip_protocol, l3_mask->proto);
+		MLX5E_FTE_SET(headers_v, ip_protocol, l3_val->proto);
+	}
+}
+
+static void
+parse_ip6(void *headers_c, void *headers_v, struct ethtool_rx_flow_spec *fs)
+{
+	struct ethtool_usrip6_spec *l3_mask = &fs->m_u.usr_ip6_spec;
+	struct ethtool_usrip6_spec *l3_val  = &fs->h_u.usr_ip6_spec;
+
+	set_ip6(headers_c, headers_v, l3_mask->ip6src,
+		l3_val->ip6src, l3_mask->ip6dst, l3_val->ip6dst);
+
+	if (l3_mask->l4_proto) {
+		MLX5E_FTE_SET(headers_c, ip_protocol, l3_mask->l4_proto);
+		MLX5E_FTE_SET(headers_v, ip_protocol, l3_val->l4_proto);
+	}
+}
+
+static void
+parse_tcp6(void *headers_c, void *headers_v, struct ethtool_rx_flow_spec *fs)
+{
+	struct ethtool_tcpip6_spec *l4_mask = &fs->m_u.tcp_ip6_spec;
+	struct ethtool_tcpip6_spec *l4_val  = &fs->h_u.tcp_ip6_spec;
+
+	set_ip6(headers_c, headers_v, l4_mask->ip6src,
+		l4_val->ip6src, l4_mask->ip6dst, l4_val->ip6dst);
+
+	set_tcp(headers_c, headers_v, l4_mask->psrc, l4_val->psrc,
+		l4_mask->pdst, l4_val->pdst);
+}
+
+static void
+parse_udp6(void *headers_c, void *headers_v, struct ethtool_rx_flow_spec *fs)
+{
+	struct ethtool_tcpip6_spec *l4_mask = &fs->m_u.udp_ip6_spec;
+	struct ethtool_tcpip6_spec *l4_val  = &fs->h_u.udp_ip6_spec;
+
+	set_ip6(headers_c, headers_v, l4_mask->ip6src,
+		l4_val->ip6src, l4_mask->ip6dst, l4_val->ip6dst);
+
+	set_udp(headers_c, headers_v, l4_mask->psrc, l4_val->psrc,
+		l4_mask->pdst, l4_val->pdst);
+}
+
+static void
+parse_ether(void *headers_c, void *headers_v, struct ethtool_rx_flow_spec *fs)
+{
+	struct ethhdr *eth_mask = &fs->m_u.ether_spec;
+	struct ethhdr *eth_val = &fs->h_u.ether_spec;
+
+	mask_spec((u8 *)eth_mask, (u8 *)eth_val, sizeof(*eth_mask));
+	ether_addr_copy(MLX5E_FTE_ADDR_OF(headers_c, smac_47_16), eth_mask->h_source);
+	ether_addr_copy(MLX5E_FTE_ADDR_OF(headers_v, smac_47_16), eth_val->h_source);
+	ether_addr_copy(MLX5E_FTE_ADDR_OF(headers_c, dmac_47_16), eth_mask->h_dest);
+	ether_addr_copy(MLX5E_FTE_ADDR_OF(headers_v, dmac_47_16), eth_val->h_dest);
+	MLX5E_FTE_SET(headers_c, ethertype, ntohs(eth_mask->h_proto));
+	MLX5E_FTE_SET(headers_v, ethertype, ntohs(eth_val->h_proto));
+}
+
+static void
+set_cvlan(void *headers_c, void *headers_v, __be16 vlan_tci)
+{
+	MLX5E_FTE_SET(headers_c, cvlan_tag, 1);
+	MLX5E_FTE_SET(headers_v, cvlan_tag, 1);
+	MLX5E_FTE_SET(headers_c, first_vid, 0xfff);
+	MLX5E_FTE_SET(headers_v, first_vid, ntohs(vlan_tci));
+}
+
+static void
+set_dmac(void *headers_c, void *headers_v,
+	 unsigned char m_dest[ETH_ALEN], unsigned char v_dest[ETH_ALEN])
+{
+	ether_addr_copy(MLX5E_FTE_ADDR_OF(headers_c, dmac_47_16), m_dest);
+	ether_addr_copy(MLX5E_FTE_ADDR_OF(headers_v, dmac_47_16), v_dest);
+}
+
+static int set_flow_attrs(u32 *match_c, u32 *match_v,
+			  struct ethtool_rx_flow_spec *fs)
+{
+	void *outer_headers_c = MLX5_ADDR_OF(fte_match_param, match_c,
+					     outer_headers);
+	void *outer_headers_v = MLX5_ADDR_OF(fte_match_param, match_v,
+					     outer_headers);
+	u32 flow_type = flow_type_mask(fs->flow_type);
+
+	switch (flow_type) {
+	case TCP_V4_FLOW:
+		parse_tcp4(outer_headers_c, outer_headers_v, fs);
+		break;
+	case UDP_V4_FLOW:
+		parse_udp4(outer_headers_c, outer_headers_v, fs);
+		break;
+	case IP_USER_FLOW:
+		parse_ip4(outer_headers_c, outer_headers_v, fs);
+		break;
+	case TCP_V6_FLOW:
+		parse_tcp6(outer_headers_c, outer_headers_v, fs);
+		break;
+	case UDP_V6_FLOW:
+		parse_udp6(outer_headers_c, outer_headers_v, fs);
+		break;
+	case IPV6_USER_FLOW:
+		parse_ip6(outer_headers_c, outer_headers_v, fs);
+		break;
+	case ETHER_FLOW:
+		parse_ether(outer_headers_c, outer_headers_v, fs);
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	if ((fs->flow_type & FLOW_EXT) &&
+	    (fs->m_ext.vlan_tci & cpu_to_be16(VLAN_VID_MASK)))
+		set_cvlan(outer_headers_c, outer_headers_v, fs->h_ext.vlan_tci);
+
+	if (fs->flow_type & FLOW_MAC_EXT &&
+	    !is_zero_ether_addr(fs->m_ext.h_dest)) {
+		mask_spec(fs->m_ext.h_dest, fs->h_ext.h_dest, ETH_ALEN);
+		set_dmac(outer_headers_c, outer_headers_v, fs->m_ext.h_dest,
+			 fs->h_ext.h_dest);
+	}
+
+	return 0;
+}
+
+static void add_rule_to_list(struct mlx5e_priv *priv,
+			     struct mlx5e_ethtool_rule *rule)
+{
+	struct mlx5e_ethtool_steering *ethtool = mlx5e_fs_get_ethtool(priv->fs);
+	struct list_head *head = &ethtool->rules;
+	struct mlx5e_ethtool_rule *iter;
+
+	list_for_each_entry(iter, &ethtool->rules, list) {
+		if (iter->flow_spec.location > rule->flow_spec.location)
+			break;
+		head = &iter->list;
+	}
+	ethtool->tot_num_rules++;
+	list_add(&rule->list, head);
+}
+
+static bool outer_header_zero(u32 *match_criteria)
+{
+	int size = MLX5_FLD_SZ_BYTES(fte_match_param, outer_headers);
+	char *outer_headers_c = MLX5_ADDR_OF(fte_match_param, match_criteria,
+					     outer_headers);
+
+	return outer_headers_c[0] == 0 && !memcmp(outer_headers_c,
+						  outer_headers_c + 1,
+						  size - 1);
+}
+
+static int flow_get_tirn(struct mlx5e_priv *priv,
+			 struct mlx5e_ethtool_rule *eth_rule,
+			 struct ethtool_rx_flow_spec *fs,
+			 u32 rss_context, u32 *tirn)
+{
+	if (fs->flow_type & FLOW_RSS) {
+		struct mlx5e_packet_merge_param pkt_merge_param;
+		struct mlx5e_rss *rss;
+		u32 flow_type;
+		int err;
+		int tt;
+
+		rss = mlx5e_rx_res_rss_get(priv->rx_res, rss_context);
+		if (!rss)
+			return -ENOENT;
+
+		flow_type = flow_type_mask(fs->flow_type);
+		tt = flow_type_to_traffic_type(flow_type);
+		if (tt < 0)
+			return -EINVAL;
+
+		pkt_merge_param = priv->channels.params.packet_merge;
+		err = mlx5e_rss_obtain_tirn(rss, tt, &pkt_merge_param, false, tirn);
+		if (err)
+			return err;
+		eth_rule->rss = rss;
+		mlx5e_rss_refcnt_inc(eth_rule->rss);
+	} else {
+		*tirn = mlx5e_rx_res_get_tirn_direct(priv->rx_res, fs->ring_cookie);
+	}
+
+	return 0;
+}
+
+static struct mlx5_flow_handle *
+add_ethtool_flow_rule(struct mlx5e_priv *priv,
+		      struct mlx5e_ethtool_rule *eth_rule,
+		      struct mlx5_flow_table *ft,
+		      struct ethtool_rx_flow_spec *fs, u32 rss_context)
+{
+	struct mlx5_flow_act flow_act = { .flags = FLOW_ACT_NO_APPEND };
+	struct mlx5_flow_destination *dst = NULL;
+	struct mlx5_flow_handle *rule;
+	struct mlx5_flow_spec *spec;
+	int err = 0;
+
+	spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
+	if (!spec)
+		return ERR_PTR(-ENOMEM);
+	err = set_flow_attrs(spec->match_criteria, spec->match_value,
+			     fs);
+	if (err)
+		goto free;
+
+	if (fs->ring_cookie == RX_CLS_FLOW_DISC) {
+		flow_act.action = MLX5_FLOW_CONTEXT_ACTION_DROP;
+	} else {
+		dst = kzalloc(sizeof(*dst), GFP_KERNEL);
+		if (!dst) {
+			err = -ENOMEM;
+			goto free;
+		}
+
+		err = flow_get_tirn(priv, eth_rule, fs, rss_context, &dst->tir_num);
+		if (err)
+			goto free;
+
+		dst->type = MLX5_FLOW_DESTINATION_TYPE_TIR;
+		flow_act.action = MLX5_FLOW_CONTEXT_ACTION_FWD_DEST;
+	}
+
+	spec->match_criteria_enable = (!outer_header_zero(spec->match_criteria));
+	spec->flow_context.flow_tag = MLX5_FS_DEFAULT_FLOW_TAG;
+	rule = mlx5_add_flow_rules(ft, spec, &flow_act, dst, dst ? 1 : 0);
+	if (IS_ERR(rule)) {
+		err = PTR_ERR(rule);
+		netdev_err(priv->netdev, "%s: failed to add ethtool steering rule: %d\n",
+			   __func__, err);
+		goto free;
+	}
+free:
+	kvfree(spec);
+	kfree(dst);
+	return err ? ERR_PTR(err) : rule;
+}
+
+static void del_ethtool_rule(struct mlx5e_flow_steering *fs,
+			     struct mlx5e_ethtool_rule *eth_rule)
+{
+	struct mlx5e_ethtool_steering *ethtool = mlx5e_fs_get_ethtool(fs);
+	if (eth_rule->rule)
+		mlx5_del_flow_rules(eth_rule->rule);
+	if (eth_rule->rss)
+		mlx5e_rss_refcnt_dec(eth_rule->rss);
+	list_del(&eth_rule->list);
+	ethtool->tot_num_rules--;
+	put_flow_table(eth_rule->eth_ft);
+	kfree(eth_rule);
+}
+
+static struct mlx5e_ethtool_rule *find_ethtool_rule(struct mlx5e_priv *priv,
+						    int location)
+{
+	struct mlx5e_ethtool_steering *ethtool = mlx5e_fs_get_ethtool(priv->fs);
+	struct mlx5e_ethtool_rule *iter;
+
+	list_for_each_entry(iter, &ethtool->rules, list) {
+		if (iter->flow_spec.location == location)
+			return iter;
+	}
+	return NULL;
+}
+
+static struct mlx5e_ethtool_rule *get_ethtool_rule(struct mlx5e_priv *priv,
+						   int location)
+{
+	struct mlx5e_ethtool_rule *eth_rule;
+
+	eth_rule = find_ethtool_rule(priv, location);
+	if (eth_rule)
+		del_ethtool_rule(priv->fs, eth_rule);
+
+	eth_rule = kzalloc(sizeof(*eth_rule), GFP_KERNEL);
+	if (!eth_rule)
+		return ERR_PTR(-ENOMEM);
+
+	add_rule_to_list(priv, eth_rule);
+	return eth_rule;
+}
+
+#define MAX_NUM_OF_ETHTOOL_RULES BIT(10)
+
+#define all_ones(field) (field == (__force typeof(field))-1)
+#define all_zeros_or_all_ones(field)		\
+	((field) == 0 || (field) == (__force typeof(field))-1)
+
+static int validate_ethter(struct ethtool_rx_flow_spec *fs)
+{
+	struct ethhdr *eth_mask = &fs->m_u.ether_spec;
+	int ntuples = 0;
+
+	if (!is_zero_ether_addr(eth_mask->h_dest))
+		ntuples++;
+	if (!is_zero_ether_addr(eth_mask->h_source))
+		ntuples++;
+	if (eth_mask->h_proto)
+		ntuples++;
+	return ntuples;
+}
+
+static int validate_tcpudp4(struct ethtool_rx_flow_spec *fs)
+{
+	struct ethtool_tcpip4_spec *l4_mask = &fs->m_u.tcp_ip4_spec;
+	int ntuples = 0;
+
+	if (l4_mask->tos)
+		return -EINVAL;
+
+	if (l4_mask->ip4src)
+		ntuples++;
+	if (l4_mask->ip4dst)
+		ntuples++;
+	if (l4_mask->psrc)
+		ntuples++;
+	if (l4_mask->pdst)
+		ntuples++;
+	/* Flow is TCP/UDP */
+	return ++ntuples;
+}
+
+static int validate_ip4(struct ethtool_rx_flow_spec *fs)
+{
+	struct ethtool_usrip4_spec *l3_mask = &fs->m_u.usr_ip4_spec;
+	int ntuples = 0;
+
+	if (l3_mask->l4_4_bytes || l3_mask->tos ||
+	    fs->h_u.usr_ip4_spec.ip_ver != ETH_RX_NFC_IP4)
+		return -EINVAL;
+	if (l3_mask->ip4src)
+		ntuples++;
+	if (l3_mask->ip4dst)
+		ntuples++;
+	if (l3_mask->proto)
+		ntuples++;
+	/* Flow is IPv4 */
+	return ++ntuples;
+}
+
+static int validate_ip6(struct ethtool_rx_flow_spec *fs)
+{
+	struct ethtool_usrip6_spec *l3_mask = &fs->m_u.usr_ip6_spec;
+	int ntuples = 0;
+
+	if (l3_mask->l4_4_bytes || l3_mask->tclass)
+		return -EINVAL;
+	if (!ipv6_addr_any((struct in6_addr *)l3_mask->ip6src))
+		ntuples++;
+
+	if (!ipv6_addr_any((struct in6_addr *)l3_mask->ip6dst))
+		ntuples++;
+	if (l3_mask->l4_proto)
+		ntuples++;
+	/* Flow is IPv6 */
+	return ++ntuples;
+}
+
+static int validate_tcpudp6(struct ethtool_rx_flow_spec *fs)
+{
+	struct ethtool_tcpip6_spec *l4_mask = &fs->m_u.tcp_ip6_spec;
+	int ntuples = 0;
+
+	if (l4_mask->tclass)
+		return -EINVAL;
+
+	if (!ipv6_addr_any((struct in6_addr *)l4_mask->ip6src))
+		ntuples++;
+
+	if (!ipv6_addr_any((struct in6_addr *)l4_mask->ip6dst))
+		ntuples++;
+
+	if (l4_mask->psrc)
+		ntuples++;
+	if (l4_mask->pdst)
+		ntuples++;
+	/* Flow is TCP/UDP */
+	return ++ntuples;
+}
+
+static int validate_vlan(struct ethtool_rx_flow_spec *fs)
+{
+	if (fs->m_ext.vlan_etype ||
+	    fs->m_ext.vlan_tci != cpu_to_be16(VLAN_VID_MASK))
+		return -EINVAL;
+
+	if (fs->m_ext.vlan_tci &&
+	    (be16_to_cpu(fs->h_ext.vlan_tci) >= VLAN_N_VID))
+		return -EINVAL;
+
+	return 1;
+}
+
+static int validate_flow(struct mlx5e_priv *priv,
+			 struct ethtool_rx_flow_spec *fs)
+{
+	int num_tuples = 0;
+	int ret = 0;
+
+	if (fs->location >= MAX_NUM_OF_ETHTOOL_RULES)
+		return -ENOSPC;
+
+	if (fs->ring_cookie != RX_CLS_FLOW_DISC)
+		if (fs->ring_cookie >= priv->channels.params.num_channels)
+			return -EINVAL;
+
+	switch (flow_type_mask(fs->flow_type)) {
+	case ETHER_FLOW:
+		num_tuples += validate_ethter(fs);
+		break;
+	case TCP_V4_FLOW:
+	case UDP_V4_FLOW:
+		ret = validate_tcpudp4(fs);
+		if (ret < 0)
+			return ret;
+		num_tuples += ret;
+		break;
+	case IP_USER_FLOW:
+		ret = validate_ip4(fs);
+		if (ret < 0)
+			return ret;
+		num_tuples += ret;
+		break;
+	case TCP_V6_FLOW:
+	case UDP_V6_FLOW:
+		ret = validate_tcpudp6(fs);
+		if (ret < 0)
+			return ret;
+		num_tuples += ret;
+		break;
+	case IPV6_USER_FLOW:
+		ret = validate_ip6(fs);
+		if (ret < 0)
+			return ret;
+		num_tuples += ret;
+		break;
+	default:
+		return -ENOTSUPP;
+	}
+	if ((fs->flow_type & FLOW_EXT)) {
+		ret = validate_vlan(fs);
+		if (ret < 0)
+			return ret;
+		num_tuples += ret;
+	}
+
+	if (fs->flow_type & FLOW_MAC_EXT &&
+	    !is_zero_ether_addr(fs->m_ext.h_dest))
+		num_tuples++;
+
+	return num_tuples;
+}
+
+static int
+mlx5e_ethtool_flow_replace(struct mlx5e_priv *priv,
+			   struct ethtool_rx_flow_spec *fs, u32 rss_context)
+{
+	struct mlx5e_ethtool_table *eth_ft;
+	struct mlx5e_ethtool_rule *eth_rule;
+	struct mlx5_flow_handle *rule;
+	int num_tuples;
+	int err;
+
+	num_tuples = validate_flow(priv, fs);
+	if (num_tuples <= 0) {
+		netdev_warn(priv->netdev, "%s: flow is not valid %d\n",
+			    __func__, num_tuples);
+		return num_tuples;
+	}
+
+	eth_ft = get_flow_table(priv, fs, num_tuples);
+	if (IS_ERR(eth_ft))
+		return PTR_ERR(eth_ft);
+
+	eth_rule = get_ethtool_rule(priv, fs->location);
+	if (IS_ERR(eth_rule)) {
+		put_flow_table(eth_ft);
+		return PTR_ERR(eth_rule);
+	}
+
+	eth_rule->flow_spec = *fs;
+	eth_rule->eth_ft = eth_ft;
+
+	rule = add_ethtool_flow_rule(priv, eth_rule, eth_ft->ft, fs, rss_context);
+	if (IS_ERR(rule)) {
+		err = PTR_ERR(rule);
+		goto del_ethtool_rule;
+	}
+
+	eth_rule->rule = rule;
+
+	return 0;
+
+del_ethtool_rule:
+	del_ethtool_rule(priv->fs, eth_rule);
+
+	return err;
+}
+
+static int
+mlx5e_ethtool_flow_remove(struct mlx5e_priv *priv, int location)
+{
+	struct mlx5e_ethtool_rule *eth_rule;
+	int err = 0;
+
+	if (location >= MAX_NUM_OF_ETHTOOL_RULES)
+		return -ENOSPC;
+
+	eth_rule = find_ethtool_rule(priv, location);
+	if (!eth_rule) {
+		err =  -ENOENT;
+		goto out;
+	}
+
+	del_ethtool_rule(priv->fs, eth_rule);
+out:
+	return err;
+}
+
+static int
+mlx5e_ethtool_get_flow(struct mlx5e_priv *priv,
+		       struct ethtool_rxnfc *info, int location)
+{
+	struct mlx5e_ethtool_steering *ethtool = mlx5e_fs_get_ethtool(priv->fs);
+	struct mlx5e_ethtool_rule *eth_rule;
+
+	if (location < 0 || location >= MAX_NUM_OF_ETHTOOL_RULES)
+		return -EINVAL;
+
+	list_for_each_entry(eth_rule, &ethtool->rules, list) {
+		int index;
+
+		if (eth_rule->flow_spec.location != location)
+			continue;
+		if (!info)
+			return 0;
+		info->fs = eth_rule->flow_spec;
+		if (!eth_rule->rss)
+			return 0;
+		index = mlx5e_rx_res_rss_index(priv->rx_res, eth_rule->rss);
+		if (index < 0)
+			return index;
+		info->rss_context = index;
+		return 0;
+	}
+
+	return -ENOENT;
+}
+
+static int
+mlx5e_ethtool_get_all_flows(struct mlx5e_priv *priv,
+			    struct ethtool_rxnfc *info, u32 *rule_locs)
+{
+	int location = 0;
+	int idx = 0;
+	int err = 0;
+
+	info->data = MAX_NUM_OF_ETHTOOL_RULES;
+	while ((!err || err == -ENOENT) && idx < info->rule_cnt) {
+		err = mlx5e_ethtool_get_flow(priv, NULL, location);
+		if (!err)
+			rule_locs[idx++] = location;
+		location++;
+	}
+	return err;
+}
+
+int mlx5e_ethtool_alloc(struct mlx5e_ethtool_steering **ethtool)
+{
+	*ethtool =  kvzalloc(sizeof(**ethtool), GFP_KERNEL);
+	if (!*ethtool)
+		return -ENOMEM;
+	return 0;
+}
+
+void mlx5e_ethtool_free(struct mlx5e_ethtool_steering *ethtool)
+{
+	kvfree(ethtool);
+}
+
+void mlx5e_ethtool_cleanup_steering(struct mlx5e_flow_steering *fs)
+{
+	struct mlx5e_ethtool_steering *ethtool = mlx5e_fs_get_ethtool(fs);
+	struct mlx5e_ethtool_rule *iter;
+	struct mlx5e_ethtool_rule *temp;
+
+	list_for_each_entry_safe(iter, temp, &ethtool->rules, list)
+		del_ethtool_rule(fs, iter);
+}
+
+void mlx5e_ethtool_init_steering(struct mlx5e_flow_steering *fs)
+{
+	struct mlx5e_ethtool_steering *ethtool = mlx5e_fs_get_ethtool(fs);
+
+	INIT_LIST_HEAD(&ethtool->rules);
+}
+
+static int flow_type_to_traffic_type(u32 flow_type)
+{
+	switch (flow_type) {
+	case TCP_V4_FLOW:
+		return MLX5_TT_IPV4_TCP;
+	case TCP_V6_FLOW:
+		return MLX5_TT_IPV6_TCP;
+	case UDP_V4_FLOW:
+		return MLX5_TT_IPV4_UDP;
+	case UDP_V6_FLOW:
+		return MLX5_TT_IPV6_UDP;
+	case AH_V4_FLOW:
+		return MLX5_TT_IPV4_IPSEC_AH;
+	case AH_V6_FLOW:
+		return MLX5_TT_IPV6_IPSEC_AH;
+	case ESP_V4_FLOW:
+		return MLX5_TT_IPV4_IPSEC_ESP;
+	case ESP_V6_FLOW:
+		return MLX5_TT_IPV6_IPSEC_ESP;
+	case IPV4_FLOW:
+		return MLX5_TT_IPV4;
+	case IPV6_FLOW:
+		return MLX5_TT_IPV6;
+	default:
+		return -EINVAL;
+	}
+}
+
+static int mlx5e_set_rss_hash_opt(struct mlx5e_priv *priv,
+				  struct ethtool_rxnfc *nfc)
+{
+	u8 rx_hash_field = 0;
+	int err;
+	int tt;
+
+	tt = flow_type_to_traffic_type(nfc->flow_type);
+	if (tt < 0)
+		return tt;
+
+	/*  RSS does not support anything other than hashing to queues
+	 *  on src IP, dest IP, TCP/UDP src port and TCP/UDP dest
+	 *  port.
+	 */
+	if (nfc->flow_type != TCP_V4_FLOW &&
+	    nfc->flow_type != TCP_V6_FLOW &&
+	    nfc->flow_type != UDP_V4_FLOW &&
+	    nfc->flow_type != UDP_V6_FLOW)
+		return -EOPNOTSUPP;
+
+	if (nfc->data & ~(RXH_IP_SRC | RXH_IP_DST |
+			  RXH_L4_B_0_1 | RXH_L4_B_2_3))
+		return -EOPNOTSUPP;
+
+	if (nfc->data & RXH_IP_SRC)
+		rx_hash_field |= MLX5_HASH_FIELD_SEL_SRC_IP;
+	if (nfc->data & RXH_IP_DST)
+		rx_hash_field |= MLX5_HASH_FIELD_SEL_DST_IP;
+	if (nfc->data & RXH_L4_B_0_1)
+		rx_hash_field |= MLX5_HASH_FIELD_SEL_L4_SPORT;
+	if (nfc->data & RXH_L4_B_2_3)
+		rx_hash_field |= MLX5_HASH_FIELD_SEL_L4_DPORT;
+
+	mutex_lock(&priv->state_lock);
+	err = mlx5e_rx_res_rss_set_hash_fields(priv->rx_res, tt, rx_hash_field);
+	mutex_unlock(&priv->state_lock);
+
+	return err;
+}
+
+static int mlx5e_get_rss_hash_opt(struct mlx5e_priv *priv,
+				  struct ethtool_rxnfc *nfc)
+{
+	u32 hash_field = 0;
+	int tt;
+
+	tt = flow_type_to_traffic_type(nfc->flow_type);
+	if (tt < 0)
+		return tt;
+
+	hash_field = mlx5e_rx_res_rss_get_hash_fields(priv->rx_res, tt);
+	nfc->data = 0;
+
+	if (hash_field & MLX5_HASH_FIELD_SEL_SRC_IP)
+		nfc->data |= RXH_IP_SRC;
+	if (hash_field & MLX5_HASH_FIELD_SEL_DST_IP)
+		nfc->data |= RXH_IP_DST;
+	if (hash_field & MLX5_HASH_FIELD_SEL_L4_SPORT)
+		nfc->data |= RXH_L4_B_0_1;
+	if (hash_field & MLX5_HASH_FIELD_SEL_L4_DPORT)
+		nfc->data |= RXH_L4_B_2_3;
+
+	return 0;
+}
+
+int mlx5e_ethtool_set_rxnfc(struct mlx5e_priv *priv, struct ethtool_rxnfc *cmd)
+{
+	int err = 0;
+
+	switch (cmd->cmd) {
+	case ETHTOOL_SRXCLSRLINS:
+		err = mlx5e_ethtool_flow_replace(priv, &cmd->fs, cmd->rss_context);
+		break;
+	case ETHTOOL_SRXCLSRLDEL:
+		err = mlx5e_ethtool_flow_remove(priv, cmd->fs.location);
+		break;
+	case ETHTOOL_SRXFH:
+		err = mlx5e_set_rss_hash_opt(priv, cmd);
+		break;
+	default:
+		err = -EOPNOTSUPP;
+		break;
+	}
+
+	return err;
+}
+
+int mlx5e_ethtool_get_rxnfc(struct mlx5e_priv *priv,
+			    struct ethtool_rxnfc *info, u32 *rule_locs)
+{
+	struct mlx5e_ethtool_steering *ethtool = mlx5e_fs_get_ethtool(priv->fs);
+	int err = 0;
+
+	switch (info->cmd) {
+	case ETHTOOL_GRXCLSRLCNT:
+		info->rule_cnt = ethtool->tot_num_rules;
+		break;
+	case ETHTOOL_GRXCLSRULE:
+		err = mlx5e_ethtool_get_flow(priv, info, info->fs.location);
+		break;
+	case ETHTOOL_GRXCLSRLALL:
+		err = mlx5e_ethtool_get_all_flows(priv, info, rule_locs);
+		break;
+	case ETHTOOL_GRXFH:
+		err =  mlx5e_get_rss_hash_opt(priv, info);
+		break;
+	default:
+		err = -EOPNOTSUPP;
+		break;
+	}
+
+	return err;
+}
+
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
new file mode 100644
index 000000000..9910a0480
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
@@ -0,0 +1,6021 @@
+/*
+ * Copyright (c) 2015-2016, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <net/tc_act/tc_gact.h>
+#include <linux/mlx5/fs.h>
+#include <net/vxlan.h>
+#include <net/geneve.h>
+#include <linux/bpf.h>
+#include <linux/if_bridge.h>
+#include <linux/filter.h>
+#include <net/page_pool.h>
+#include <net/xdp_sock_drv.h>
+#include "eswitch.h"
+#include "en.h"
+#include "en/txrx.h"
+#include "en_tc.h"
+#include "en_rep.h"
+#include "en_accel/ipsec.h"
+#include "en_accel/macsec.h"
+#include "en_accel/en_accel.h"
+#include "en_accel/ktls.h"
+#include "lib/vxlan.h"
+#include "lib/clock.h"
+#include "en/port.h"
+#include "en/xdp.h"
+#include "lib/eq.h"
+#include "en/monitor_stats.h"
+#include "en/health.h"
+#include "en/params.h"
+#include "en/xsk/pool.h"
+#include "en/xsk/setup.h"
+#include "en/xsk/rx.h"
+#include "en/xsk/tx.h"
+#include "en/hv_vhca_stats.h"
+#include "en/devlink.h"
+#include "lib/mlx5.h"
+#include "en/ptp.h"
+#include "en/htb.h"
+#include "qos.h"
+#include "en/trap.h"
+
+bool mlx5e_check_fragmented_striding_rq_cap(struct mlx5_core_dev *mdev, u8 page_shift,
+					    enum mlx5e_mpwrq_umr_mode umr_mode)
+{
+	u16 umr_wqebbs, max_wqebbs;
+	bool striding_rq_umr;
+
+	striding_rq_umr = MLX5_CAP_GEN(mdev, striding_rq) && MLX5_CAP_GEN(mdev, umr_ptr_rlky) &&
+			  MLX5_CAP_ETH(mdev, reg_umr_sq);
+	if (!striding_rq_umr)
+		return false;
+
+	umr_wqebbs = mlx5e_mpwrq_umr_wqebbs(mdev, page_shift, umr_mode);
+	max_wqebbs = mlx5e_get_max_sq_aligned_wqebbs(mdev);
+	/* Sanity check; should never happen, because mlx5e_mpwrq_umr_wqebbs is
+	 * calculated from mlx5e_get_max_sq_aligned_wqebbs.
+	 */
+	if (WARN_ON(umr_wqebbs > max_wqebbs))
+		return false;
+
+	return true;
+}
+
+void mlx5e_update_carrier(struct mlx5e_priv *priv)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+	u8 port_state;
+	bool up;
+
+	port_state = mlx5_query_vport_state(mdev,
+					    MLX5_VPORT_STATE_OP_MOD_VNIC_VPORT,
+					    0);
+
+	up = port_state == VPORT_STATE_UP;
+	if (up == netif_carrier_ok(priv->netdev))
+		netif_carrier_event(priv->netdev);
+	if (up) {
+		netdev_info(priv->netdev, "Link up\n");
+		netif_carrier_on(priv->netdev);
+	} else {
+		netdev_info(priv->netdev, "Link down\n");
+		netif_carrier_off(priv->netdev);
+	}
+}
+
+static void mlx5e_update_carrier_work(struct work_struct *work)
+{
+	struct mlx5e_priv *priv = container_of(work, struct mlx5e_priv,
+					       update_carrier_work);
+
+	mutex_lock(&priv->state_lock);
+	if (test_bit(MLX5E_STATE_OPENED, &priv->state))
+		if (priv->profile->update_carrier)
+			priv->profile->update_carrier(priv);
+	mutex_unlock(&priv->state_lock);
+}
+
+static void mlx5e_update_stats_work(struct work_struct *work)
+{
+	struct mlx5e_priv *priv = container_of(work, struct mlx5e_priv,
+					       update_stats_work);
+
+	mutex_lock(&priv->state_lock);
+	priv->profile->update_stats(priv);
+	mutex_unlock(&priv->state_lock);
+}
+
+void mlx5e_queue_update_stats(struct mlx5e_priv *priv)
+{
+	if (!priv->profile->update_stats)
+		return;
+
+	if (unlikely(test_bit(MLX5E_STATE_DESTROYING, &priv->state)))
+		return;
+
+	queue_work(priv->wq, &priv->update_stats_work);
+}
+
+static int async_event(struct notifier_block *nb, unsigned long event, void *data)
+{
+	struct mlx5e_priv *priv = container_of(nb, struct mlx5e_priv, events_nb);
+	struct mlx5_eqe   *eqe = data;
+
+	if (event != MLX5_EVENT_TYPE_PORT_CHANGE)
+		return NOTIFY_DONE;
+
+	switch (eqe->sub_type) {
+	case MLX5_PORT_CHANGE_SUBTYPE_DOWN:
+	case MLX5_PORT_CHANGE_SUBTYPE_ACTIVE:
+		queue_work(priv->wq, &priv->update_carrier_work);
+		break;
+	default:
+		return NOTIFY_DONE;
+	}
+
+	return NOTIFY_OK;
+}
+
+static void mlx5e_enable_async_events(struct mlx5e_priv *priv)
+{
+	priv->events_nb.notifier_call = async_event;
+	mlx5_notifier_register(priv->mdev, &priv->events_nb);
+}
+
+static void mlx5e_disable_async_events(struct mlx5e_priv *priv)
+{
+	mlx5_notifier_unregister(priv->mdev, &priv->events_nb);
+}
+
+static int blocking_event(struct notifier_block *nb, unsigned long event, void *data)
+{
+	struct mlx5e_priv *priv = container_of(nb, struct mlx5e_priv, blocking_events_nb);
+	int err;
+
+	switch (event) {
+	case MLX5_DRIVER_EVENT_TYPE_TRAP:
+		err = mlx5e_handle_trap_event(priv, data);
+		break;
+	default:
+		netdev_warn(priv->netdev, "Sync event: Unknown event %ld\n", event);
+		err = -EINVAL;
+	}
+	return err;
+}
+
+static void mlx5e_enable_blocking_events(struct mlx5e_priv *priv)
+{
+	priv->blocking_events_nb.notifier_call = blocking_event;
+	mlx5_blocking_notifier_register(priv->mdev, &priv->blocking_events_nb);
+}
+
+static void mlx5e_disable_blocking_events(struct mlx5e_priv *priv)
+{
+	mlx5_blocking_notifier_unregister(priv->mdev, &priv->blocking_events_nb);
+}
+
+static u16 mlx5e_mpwrq_umr_octowords(u32 entries, enum mlx5e_mpwrq_umr_mode umr_mode)
+{
+	u8 umr_entry_size = mlx5e_mpwrq_umr_entry_size(umr_mode);
+	u32 sz;
+
+	sz = ALIGN(entries * umr_entry_size, MLX5_UMR_MTT_ALIGNMENT);
+
+	return sz / MLX5_OCTWORD;
+}
+
+static inline void mlx5e_build_umr_wqe(struct mlx5e_rq *rq,
+				       struct mlx5e_icosq *sq,
+				       struct mlx5e_umr_wqe *wqe)
+{
+	struct mlx5_wqe_ctrl_seg      *cseg = &wqe->ctrl;
+	struct mlx5_wqe_umr_ctrl_seg *ucseg = &wqe->uctrl;
+	u16 octowords;
+	u8 ds_cnt;
+
+	ds_cnt = DIV_ROUND_UP(mlx5e_mpwrq_umr_wqe_sz(rq->mdev, rq->mpwqe.page_shift,
+						     rq->mpwqe.umr_mode),
+			      MLX5_SEND_WQE_DS);
+
+	cseg->qpn_ds    = cpu_to_be32((sq->sqn << MLX5_WQE_CTRL_QPN_SHIFT) |
+				      ds_cnt);
+	cseg->umr_mkey  = rq->mpwqe.umr_mkey_be;
+
+	ucseg->flags = MLX5_UMR_TRANSLATION_OFFSET_EN | MLX5_UMR_INLINE;
+	octowords = mlx5e_mpwrq_umr_octowords(rq->mpwqe.pages_per_wqe, rq->mpwqe.umr_mode);
+	ucseg->xlt_octowords = cpu_to_be16(octowords);
+	ucseg->mkey_mask     = cpu_to_be64(MLX5_MKEY_MASK_FREE);
+}
+
+static int mlx5e_rq_shampo_hd_alloc(struct mlx5e_rq *rq, int node)
+{
+	rq->mpwqe.shampo = kvzalloc_node(sizeof(*rq->mpwqe.shampo),
+					 GFP_KERNEL, node);
+	if (!rq->mpwqe.shampo)
+		return -ENOMEM;
+	return 0;
+}
+
+static void mlx5e_rq_shampo_hd_free(struct mlx5e_rq *rq)
+{
+	kvfree(rq->mpwqe.shampo);
+}
+
+static int mlx5e_rq_shampo_hd_info_alloc(struct mlx5e_rq *rq, int node)
+{
+	struct mlx5e_shampo_hd *shampo = rq->mpwqe.shampo;
+
+	shampo->bitmap = bitmap_zalloc_node(shampo->hd_per_wq, GFP_KERNEL,
+					    node);
+	if (!shampo->bitmap)
+		return -ENOMEM;
+
+	shampo->info = kvzalloc_node(array_size(shampo->hd_per_wq,
+						sizeof(*shampo->info)),
+				     GFP_KERNEL, node);
+	if (!shampo->info) {
+		kvfree(shampo->bitmap);
+		return -ENOMEM;
+	}
+	return 0;
+}
+
+static void mlx5e_rq_shampo_hd_info_free(struct mlx5e_rq *rq)
+{
+	kvfree(rq->mpwqe.shampo->bitmap);
+	kvfree(rq->mpwqe.shampo->info);
+}
+
+static int mlx5e_rq_alloc_mpwqe_info(struct mlx5e_rq *rq, int node)
+{
+	int wq_sz = mlx5_wq_ll_get_size(&rq->mpwqe.wq);
+	size_t alloc_size;
+
+	alloc_size = array_size(wq_sz, struct_size(rq->mpwqe.info, alloc_units,
+						   rq->mpwqe.pages_per_wqe));
+
+	rq->mpwqe.info = kvzalloc_node(alloc_size, GFP_KERNEL, node);
+	if (!rq->mpwqe.info)
+		return -ENOMEM;
+
+	mlx5e_build_umr_wqe(rq, rq->icosq, &rq->mpwqe.umr_wqe);
+
+	return 0;
+}
+
+
+static u8 mlx5e_mpwrq_access_mode(enum mlx5e_mpwrq_umr_mode umr_mode)
+{
+	switch (umr_mode) {
+	case MLX5E_MPWRQ_UMR_MODE_ALIGNED:
+		return MLX5_MKC_ACCESS_MODE_MTT;
+	case MLX5E_MPWRQ_UMR_MODE_UNALIGNED:
+		return MLX5_MKC_ACCESS_MODE_KSM;
+	case MLX5E_MPWRQ_UMR_MODE_OVERSIZED:
+		return MLX5_MKC_ACCESS_MODE_KLMS;
+	case MLX5E_MPWRQ_UMR_MODE_TRIPLE:
+		return MLX5_MKC_ACCESS_MODE_KSM;
+	}
+	WARN_ONCE(1, "MPWRQ UMR mode %d is not known\n", umr_mode);
+	return 0;
+}
+
+static int mlx5e_create_umr_mkey(struct mlx5_core_dev *mdev,
+				 u32 npages, u8 page_shift, u32 *umr_mkey,
+				 dma_addr_t filler_addr,
+				 enum mlx5e_mpwrq_umr_mode umr_mode,
+				 u32 xsk_chunk_size)
+{
+	struct mlx5_mtt *mtt;
+	struct mlx5_ksm *ksm;
+	struct mlx5_klm *klm;
+	u32 octwords;
+	int inlen;
+	void *mkc;
+	u32 *in;
+	int err;
+	int i;
+
+	if ((umr_mode == MLX5E_MPWRQ_UMR_MODE_UNALIGNED ||
+	     umr_mode == MLX5E_MPWRQ_UMR_MODE_TRIPLE) &&
+	    !MLX5_CAP_GEN(mdev, fixed_buffer_size)) {
+		mlx5_core_warn(mdev, "Unaligned AF_XDP requires fixed_buffer_size capability\n");
+		return -EINVAL;
+	}
+
+	octwords = mlx5e_mpwrq_umr_octowords(npages, umr_mode);
+
+	inlen = MLX5_FLEXIBLE_INLEN(mdev, MLX5_ST_SZ_BYTES(create_mkey_in),
+				    MLX5_OCTWORD, octwords);
+	if (inlen < 0)
+		return inlen;
+
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in)
+		return -ENOMEM;
+
+	mkc = MLX5_ADDR_OF(create_mkey_in, in, memory_key_mkey_entry);
+
+	MLX5_SET(mkc, mkc, free, 1);
+	MLX5_SET(mkc, mkc, umr_en, 1);
+	MLX5_SET(mkc, mkc, lw, 1);
+	MLX5_SET(mkc, mkc, lr, 1);
+	MLX5_SET(mkc, mkc, access_mode_1_0, mlx5e_mpwrq_access_mode(umr_mode));
+	mlx5e_mkey_set_relaxed_ordering(mdev, mkc);
+	MLX5_SET(mkc, mkc, qpn, 0xffffff);
+	MLX5_SET(mkc, mkc, pd, mdev->mlx5e_res.hw_objs.pdn);
+	MLX5_SET64(mkc, mkc, len, npages << page_shift);
+	MLX5_SET(mkc, mkc, translations_octword_size, octwords);
+	if (umr_mode == MLX5E_MPWRQ_UMR_MODE_TRIPLE)
+		MLX5_SET(mkc, mkc, log_page_size, page_shift - 2);
+	else if (umr_mode != MLX5E_MPWRQ_UMR_MODE_OVERSIZED)
+		MLX5_SET(mkc, mkc, log_page_size, page_shift);
+	MLX5_SET(create_mkey_in, in, translations_octword_actual_size, octwords);
+
+	/* Initialize the mkey with all MTTs pointing to a default
+	 * page (filler_addr). When the channels are activated, UMR
+	 * WQEs will redirect the RX WQEs to the actual memory from
+	 * the RQ's pool, while the gaps (wqe_overflow) remain mapped
+	 * to the default page.
+	 */
+	switch (umr_mode) {
+	case MLX5E_MPWRQ_UMR_MODE_OVERSIZED:
+		klm = MLX5_ADDR_OF(create_mkey_in, in, klm_pas_mtt);
+		for (i = 0; i < npages; i++) {
+			klm[i << 1] = (struct mlx5_klm) {
+				.va = cpu_to_be64(filler_addr),
+				.bcount = cpu_to_be32(xsk_chunk_size),
+				.key = cpu_to_be32(mdev->mlx5e_res.hw_objs.mkey),
+			};
+			klm[(i << 1) + 1] = (struct mlx5_klm) {
+				.va = cpu_to_be64(filler_addr),
+				.bcount = cpu_to_be32((1 << page_shift) - xsk_chunk_size),
+				.key = cpu_to_be32(mdev->mlx5e_res.hw_objs.mkey),
+			};
+		}
+		break;
+	case MLX5E_MPWRQ_UMR_MODE_UNALIGNED:
+		ksm = MLX5_ADDR_OF(create_mkey_in, in, klm_pas_mtt);
+		for (i = 0; i < npages; i++)
+			ksm[i] = (struct mlx5_ksm) {
+				.key = cpu_to_be32(mdev->mlx5e_res.hw_objs.mkey),
+				.va = cpu_to_be64(filler_addr),
+			};
+		break;
+	case MLX5E_MPWRQ_UMR_MODE_ALIGNED:
+		mtt = MLX5_ADDR_OF(create_mkey_in, in, klm_pas_mtt);
+		for (i = 0; i < npages; i++)
+			mtt[i] = (struct mlx5_mtt) {
+				.ptag = cpu_to_be64(filler_addr),
+			};
+		break;
+	case MLX5E_MPWRQ_UMR_MODE_TRIPLE:
+		ksm = MLX5_ADDR_OF(create_mkey_in, in, klm_pas_mtt);
+		for (i = 0; i < npages * 4; i++) {
+			ksm[i] = (struct mlx5_ksm) {
+				.key = cpu_to_be32(mdev->mlx5e_res.hw_objs.mkey),
+				.va = cpu_to_be64(filler_addr),
+			};
+		}
+		break;
+	}
+
+	err = mlx5_core_create_mkey(mdev, umr_mkey, in, inlen);
+
+	kvfree(in);
+	return err;
+}
+
+static int mlx5e_create_umr_klm_mkey(struct mlx5_core_dev *mdev,
+				     u64 nentries,
+				     u32 *umr_mkey)
+{
+	int inlen;
+	void *mkc;
+	u32 *in;
+	int err;
+
+	inlen = MLX5_ST_SZ_BYTES(create_mkey_in);
+
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in)
+		return -ENOMEM;
+
+	mkc = MLX5_ADDR_OF(create_mkey_in, in, memory_key_mkey_entry);
+
+	MLX5_SET(mkc, mkc, free, 1);
+	MLX5_SET(mkc, mkc, umr_en, 1);
+	MLX5_SET(mkc, mkc, lw, 1);
+	MLX5_SET(mkc, mkc, lr, 1);
+	MLX5_SET(mkc, mkc, access_mode_1_0, MLX5_MKC_ACCESS_MODE_KLMS);
+	mlx5e_mkey_set_relaxed_ordering(mdev, mkc);
+	MLX5_SET(mkc, mkc, qpn, 0xffffff);
+	MLX5_SET(mkc, mkc, pd, mdev->mlx5e_res.hw_objs.pdn);
+	MLX5_SET(mkc, mkc, translations_octword_size, nentries);
+	MLX5_SET(mkc, mkc, length64, 1);
+	err = mlx5_core_create_mkey(mdev, umr_mkey, in, inlen);
+
+	kvfree(in);
+	return err;
+}
+
+static int mlx5e_create_rq_umr_mkey(struct mlx5_core_dev *mdev, struct mlx5e_rq *rq)
+{
+	u32 xsk_chunk_size = rq->xsk_pool ? rq->xsk_pool->chunk_size : 0;
+	u32 wq_size = mlx5_wq_ll_get_size(&rq->mpwqe.wq);
+	u32 num_entries, max_num_entries;
+	u32 umr_mkey;
+	int err;
+
+	max_num_entries = mlx5e_mpwrq_max_num_entries(mdev, rq->mpwqe.umr_mode);
+
+	/* Shouldn't overflow, the result is at most MLX5E_MAX_RQ_NUM_MTTS. */
+	if (WARN_ON_ONCE(check_mul_overflow(wq_size, (u32)rq->mpwqe.mtts_per_wqe,
+					    &num_entries) ||
+			 num_entries > max_num_entries))
+		mlx5_core_err(mdev, "%s: multiplication overflow: %u * %u > %u\n",
+			      __func__, wq_size, rq->mpwqe.mtts_per_wqe,
+			      max_num_entries);
+
+	err = mlx5e_create_umr_mkey(mdev, num_entries, rq->mpwqe.page_shift,
+				    &umr_mkey, rq->wqe_overflow.addr,
+				    rq->mpwqe.umr_mode, xsk_chunk_size);
+	rq->mpwqe.umr_mkey_be = cpu_to_be32(umr_mkey);
+	return err;
+}
+
+static int mlx5e_create_rq_hd_umr_mkey(struct mlx5_core_dev *mdev,
+				       struct mlx5e_rq *rq)
+{
+	u32 max_klm_size = BIT(MLX5_CAP_GEN(mdev, log_max_klm_list_size));
+
+	if (max_klm_size < rq->mpwqe.shampo->hd_per_wq) {
+		mlx5_core_err(mdev, "max klm list size 0x%x is smaller than shampo header buffer list size 0x%x\n",
+			      max_klm_size, rq->mpwqe.shampo->hd_per_wq);
+		return -EINVAL;
+	}
+	return mlx5e_create_umr_klm_mkey(mdev, rq->mpwqe.shampo->hd_per_wq,
+					 &rq->mpwqe.shampo->mkey);
+}
+
+static void mlx5e_init_frags_partition(struct mlx5e_rq *rq)
+{
+	struct mlx5e_wqe_frag_info next_frag = {};
+	struct mlx5e_wqe_frag_info *prev = NULL;
+	int i;
+
+	if (rq->xsk_pool) {
+		/* Assumptions used by XSK batched allocator. */
+		WARN_ON(rq->wqe.info.num_frags != 1);
+		WARN_ON(rq->wqe.info.log_num_frags != 0);
+		WARN_ON(rq->wqe.info.arr[0].frag_stride != PAGE_SIZE);
+	}
+
+	next_frag.au = &rq->wqe.alloc_units[0];
+
+	for (i = 0; i < mlx5_wq_cyc_get_size(&rq->wqe.wq); i++) {
+		struct mlx5e_rq_frag_info *frag_info = &rq->wqe.info.arr[0];
+		struct mlx5e_wqe_frag_info *frag =
+			&rq->wqe.frags[i << rq->wqe.info.log_num_frags];
+		int f;
+
+		for (f = 0; f < rq->wqe.info.num_frags; f++, frag++) {
+			if (next_frag.offset + frag_info[f].frag_stride > PAGE_SIZE) {
+				next_frag.au++;
+				next_frag.offset = 0;
+				if (prev)
+					prev->last_in_page = true;
+			}
+			*frag = next_frag;
+
+			/* prepare next */
+			next_frag.offset += frag_info[f].frag_stride;
+			prev = frag;
+		}
+	}
+
+	if (prev)
+		prev->last_in_page = true;
+}
+
+static int mlx5e_init_au_list(struct mlx5e_rq *rq, int wq_sz, int node)
+{
+	int len = wq_sz << rq->wqe.info.log_num_frags;
+
+	rq->wqe.alloc_units = kvzalloc_node(array_size(len, sizeof(*rq->wqe.alloc_units)),
+					    GFP_KERNEL, node);
+	if (!rq->wqe.alloc_units)
+		return -ENOMEM;
+
+	mlx5e_init_frags_partition(rq);
+
+	return 0;
+}
+
+static void mlx5e_free_au_list(struct mlx5e_rq *rq)
+{
+	kvfree(rq->wqe.alloc_units);
+}
+
+static void mlx5e_rq_err_cqe_work(struct work_struct *recover_work)
+{
+	struct mlx5e_rq *rq = container_of(recover_work, struct mlx5e_rq, recover_work);
+
+	mlx5e_reporter_rq_cqe_err(rq);
+}
+
+static int mlx5e_alloc_mpwqe_rq_drop_page(struct mlx5e_rq *rq)
+{
+	rq->wqe_overflow.page = alloc_page(GFP_KERNEL);
+	if (!rq->wqe_overflow.page)
+		return -ENOMEM;
+
+	rq->wqe_overflow.addr = dma_map_page(rq->pdev, rq->wqe_overflow.page, 0,
+					     PAGE_SIZE, rq->buff.map_dir);
+	if (dma_mapping_error(rq->pdev, rq->wqe_overflow.addr)) {
+		__free_page(rq->wqe_overflow.page);
+		return -ENOMEM;
+	}
+	return 0;
+}
+
+static void mlx5e_free_mpwqe_rq_drop_page(struct mlx5e_rq *rq)
+{
+	 dma_unmap_page(rq->pdev, rq->wqe_overflow.addr, PAGE_SIZE,
+			rq->buff.map_dir);
+	 __free_page(rq->wqe_overflow.page);
+}
+
+static int mlx5e_init_rxq_rq(struct mlx5e_channel *c, struct mlx5e_params *params,
+			     struct mlx5e_rq *rq)
+{
+	struct mlx5_core_dev *mdev = c->mdev;
+	int err;
+
+	rq->wq_type      = params->rq_wq_type;
+	rq->pdev         = c->pdev;
+	rq->netdev       = c->netdev;
+	rq->priv         = c->priv;
+	rq->tstamp       = c->tstamp;
+	rq->clock        = &mdev->clock;
+	rq->icosq        = &c->icosq;
+	rq->ix           = c->ix;
+	rq->channel      = c;
+	rq->mdev         = mdev;
+	rq->hw_mtu =
+		MLX5E_SW2HW_MTU(params, params->sw_mtu) - ETH_FCS_LEN * !params->scatter_fcs_en;
+	rq->xdpsq        = &c->rq_xdpsq;
+	rq->stats        = &c->priv->channel_stats[c->ix]->rq;
+	rq->ptp_cyc2time = mlx5_rq_ts_translator(mdev);
+	err = mlx5e_rq_set_handlers(rq, params, NULL);
+	if (err)
+		return err;
+
+	return xdp_rxq_info_reg(&rq->xdp_rxq, rq->netdev, rq->ix, c->napi.napi_id);
+}
+
+static int mlx5_rq_shampo_alloc(struct mlx5_core_dev *mdev,
+				struct mlx5e_params *params,
+				struct mlx5e_rq_param *rqp,
+				struct mlx5e_rq *rq,
+				u32 *pool_size,
+				int node)
+{
+	void *wqc = MLX5_ADDR_OF(rqc, rqp->rqc, wq);
+	int wq_size;
+	int err;
+
+	if (!test_bit(MLX5E_RQ_STATE_SHAMPO, &rq->state))
+		return 0;
+	err = mlx5e_rq_shampo_hd_alloc(rq, node);
+	if (err)
+		goto out;
+	rq->mpwqe.shampo->hd_per_wq =
+		mlx5e_shampo_hd_per_wq(mdev, params, rqp);
+	err = mlx5e_create_rq_hd_umr_mkey(mdev, rq);
+	if (err)
+		goto err_shampo_hd;
+	err = mlx5e_rq_shampo_hd_info_alloc(rq, node);
+	if (err)
+		goto err_shampo_info;
+	rq->hw_gro_data = kvzalloc_node(sizeof(*rq->hw_gro_data), GFP_KERNEL, node);
+	if (!rq->hw_gro_data) {
+		err = -ENOMEM;
+		goto err_hw_gro_data;
+	}
+	rq->mpwqe.shampo->key =
+		cpu_to_be32(rq->mpwqe.shampo->mkey);
+	rq->mpwqe.shampo->hd_per_wqe =
+		mlx5e_shampo_hd_per_wqe(mdev, params, rqp);
+	wq_size = BIT(MLX5_GET(wq, wqc, log_wq_sz));
+	*pool_size += (rq->mpwqe.shampo->hd_per_wqe * wq_size) /
+		     MLX5E_SHAMPO_WQ_HEADER_PER_PAGE;
+	return 0;
+
+err_hw_gro_data:
+	mlx5e_rq_shampo_hd_info_free(rq);
+err_shampo_info:
+	mlx5_core_destroy_mkey(mdev, rq->mpwqe.shampo->mkey);
+err_shampo_hd:
+	mlx5e_rq_shampo_hd_free(rq);
+out:
+	return err;
+}
+
+static void mlx5e_rq_free_shampo(struct mlx5e_rq *rq)
+{
+	if (!test_bit(MLX5E_RQ_STATE_SHAMPO, &rq->state))
+		return;
+
+	kvfree(rq->hw_gro_data);
+	mlx5e_rq_shampo_hd_info_free(rq);
+	mlx5_core_destroy_mkey(rq->mdev, rq->mpwqe.shampo->mkey);
+	mlx5e_rq_shampo_hd_free(rq);
+}
+
+static int mlx5e_alloc_rq(struct mlx5e_params *params,
+			  struct mlx5e_xsk_param *xsk,
+			  struct mlx5e_rq_param *rqp,
+			  int node, struct mlx5e_rq *rq)
+{
+	struct page_pool_params pp_params = { 0 };
+	struct mlx5_core_dev *mdev = rq->mdev;
+	void *rqc = rqp->rqc;
+	void *rqc_wq = MLX5_ADDR_OF(rqc, rqc, wq);
+	u32 pool_size;
+	int wq_sz;
+	int err;
+	int i;
+
+	rqp->wq.db_numa_node = node;
+	INIT_WORK(&rq->recover_work, mlx5e_rq_err_cqe_work);
+
+	if (params->xdp_prog)
+		bpf_prog_inc(params->xdp_prog);
+	RCU_INIT_POINTER(rq->xdp_prog, params->xdp_prog);
+
+	rq->buff.map_dir = params->xdp_prog ? DMA_BIDIRECTIONAL : DMA_FROM_DEVICE;
+	rq->buff.headroom = mlx5e_get_rq_headroom(mdev, params, xsk);
+	pool_size = 1 << params->log_rq_mtu_frames;
+
+	rq->mkey_be = cpu_to_be32(mdev->mlx5e_res.hw_objs.mkey);
+
+	switch (rq->wq_type) {
+	case MLX5_WQ_TYPE_LINKED_LIST_STRIDING_RQ:
+		err = mlx5_wq_ll_create(mdev, &rqp->wq, rqc_wq, &rq->mpwqe.wq,
+					&rq->wq_ctrl);
+		if (err)
+			goto err_rq_xdp_prog;
+
+		err = mlx5e_alloc_mpwqe_rq_drop_page(rq);
+		if (err)
+			goto err_rq_wq_destroy;
+
+		rq->mpwqe.wq.db = &rq->mpwqe.wq.db[MLX5_RCV_DBR];
+
+		wq_sz = mlx5_wq_ll_get_size(&rq->mpwqe.wq);
+
+		rq->mpwqe.page_shift = mlx5e_mpwrq_page_shift(mdev, xsk);
+		rq->mpwqe.umr_mode = mlx5e_mpwrq_umr_mode(mdev, xsk);
+		rq->mpwqe.pages_per_wqe =
+			mlx5e_mpwrq_pages_per_wqe(mdev, rq->mpwqe.page_shift,
+						  rq->mpwqe.umr_mode);
+		rq->mpwqe.umr_wqebbs =
+			mlx5e_mpwrq_umr_wqebbs(mdev, rq->mpwqe.page_shift,
+					       rq->mpwqe.umr_mode);
+		rq->mpwqe.mtts_per_wqe =
+			mlx5e_mpwrq_mtts_per_wqe(mdev, rq->mpwqe.page_shift,
+						 rq->mpwqe.umr_mode);
+
+		pool_size = rq->mpwqe.pages_per_wqe <<
+			mlx5e_mpwqe_get_log_rq_size(mdev, params, xsk);
+
+		rq->mpwqe.log_stride_sz = mlx5e_mpwqe_get_log_stride_size(mdev, params, xsk);
+		rq->mpwqe.num_strides =
+			BIT(mlx5e_mpwqe_get_log_num_strides(mdev, params, xsk));
+		rq->mpwqe.min_wqe_bulk = mlx5e_mpwqe_get_min_wqe_bulk(wq_sz);
+
+		rq->buff.frame0_sz = (1 << rq->mpwqe.log_stride_sz);
+
+		err = mlx5e_create_rq_umr_mkey(mdev, rq);
+		if (err)
+			goto err_rq_drop_page;
+
+		err = mlx5e_rq_alloc_mpwqe_info(rq, node);
+		if (err)
+			goto err_rq_mkey;
+
+		err = mlx5_rq_shampo_alloc(mdev, params, rqp, rq, &pool_size, node);
+		if (err)
+			goto err_free_mpwqe_info;
+
+		break;
+	default: /* MLX5_WQ_TYPE_CYCLIC */
+		err = mlx5_wq_cyc_create(mdev, &rqp->wq, rqc_wq, &rq->wqe.wq,
+					 &rq->wq_ctrl);
+		if (err)
+			goto err_rq_xdp_prog;
+
+		rq->wqe.wq.db = &rq->wqe.wq.db[MLX5_RCV_DBR];
+
+		wq_sz = mlx5_wq_cyc_get_size(&rq->wqe.wq);
+
+		rq->wqe.info = rqp->frags_info;
+		rq->buff.frame0_sz = rq->wqe.info.arr[0].frag_stride;
+
+		rq->wqe.frags =
+			kvzalloc_node(array_size(sizeof(*rq->wqe.frags),
+					(wq_sz << rq->wqe.info.log_num_frags)),
+				      GFP_KERNEL, node);
+		if (!rq->wqe.frags) {
+			err = -ENOMEM;
+			goto err_rq_wq_destroy;
+		}
+
+		err = mlx5e_init_au_list(rq, wq_sz, node);
+		if (err)
+			goto err_rq_frags;
+	}
+
+	if (xsk) {
+		err = xdp_rxq_info_reg_mem_model(&rq->xdp_rxq,
+						 MEM_TYPE_XSK_BUFF_POOL, NULL);
+		xsk_pool_set_rxq_info(rq->xsk_pool, &rq->xdp_rxq);
+	} else {
+		/* Create a page_pool and register it with rxq */
+		pp_params.order     = 0;
+		pp_params.flags     = 0; /* No-internal DMA mapping in page_pool */
+		pp_params.pool_size = pool_size;
+		pp_params.nid       = node;
+		pp_params.dev       = rq->pdev;
+		pp_params.dma_dir   = rq->buff.map_dir;
+
+		/* page_pool can be used even when there is no rq->xdp_prog,
+		 * given page_pool does not handle DMA mapping there is no
+		 * required state to clear. And page_pool gracefully handle
+		 * elevated refcnt.
+		 */
+		rq->page_pool = page_pool_create(&pp_params);
+		if (IS_ERR(rq->page_pool)) {
+			err = PTR_ERR(rq->page_pool);
+			rq->page_pool = NULL;
+			goto err_free_by_rq_type;
+		}
+		if (xdp_rxq_info_is_reg(&rq->xdp_rxq))
+			err = xdp_rxq_info_reg_mem_model(&rq->xdp_rxq,
+							 MEM_TYPE_PAGE_POOL, rq->page_pool);
+	}
+	if (err)
+		goto err_destroy_page_pool;
+
+	for (i = 0; i < wq_sz; i++) {
+		if (rq->wq_type == MLX5_WQ_TYPE_LINKED_LIST_STRIDING_RQ) {
+			struct mlx5e_rx_wqe_ll *wqe =
+				mlx5_wq_ll_get_wqe(&rq->mpwqe.wq, i);
+			u32 byte_count =
+				rq->mpwqe.num_strides << rq->mpwqe.log_stride_sz;
+			u64 dma_offset = mul_u32_u32(i, rq->mpwqe.mtts_per_wqe) <<
+				rq->mpwqe.page_shift;
+			u16 headroom = test_bit(MLX5E_RQ_STATE_SHAMPO, &rq->state) ?
+				       0 : rq->buff.headroom;
+
+			wqe->data[0].addr = cpu_to_be64(dma_offset + headroom);
+			wqe->data[0].byte_count = cpu_to_be32(byte_count);
+			wqe->data[0].lkey = rq->mpwqe.umr_mkey_be;
+		} else {
+			struct mlx5e_rx_wqe_cyc *wqe =
+				mlx5_wq_cyc_get_wqe(&rq->wqe.wq, i);
+			int f;
+
+			for (f = 0; f < rq->wqe.info.num_frags; f++) {
+				u32 frag_size = rq->wqe.info.arr[f].frag_size |
+					MLX5_HW_START_PADDING;
+
+				wqe->data[f].byte_count = cpu_to_be32(frag_size);
+				wqe->data[f].lkey = rq->mkey_be;
+			}
+			/* check if num_frags is not a pow of two */
+			if (rq->wqe.info.num_frags < (1 << rq->wqe.info.log_num_frags)) {
+				wqe->data[f].byte_count = 0;
+				wqe->data[f].lkey = cpu_to_be32(MLX5_INVALID_LKEY);
+				wqe->data[f].addr = 0;
+			}
+		}
+	}
+
+	INIT_WORK(&rq->dim.work, mlx5e_rx_dim_work);
+
+	switch (params->rx_cq_moderation.cq_period_mode) {
+	case MLX5_CQ_PERIOD_MODE_START_FROM_CQE:
+		rq->dim.mode = DIM_CQ_PERIOD_MODE_START_FROM_CQE;
+		break;
+	case MLX5_CQ_PERIOD_MODE_START_FROM_EQE:
+	default:
+		rq->dim.mode = DIM_CQ_PERIOD_MODE_START_FROM_EQE;
+	}
+
+	rq->page_cache.head = 0;
+	rq->page_cache.tail = 0;
+
+	return 0;
+
+err_destroy_page_pool:
+	page_pool_destroy(rq->page_pool);
+err_free_by_rq_type:
+	switch (rq->wq_type) {
+	case MLX5_WQ_TYPE_LINKED_LIST_STRIDING_RQ:
+		mlx5e_rq_free_shampo(rq);
+err_free_mpwqe_info:
+		kvfree(rq->mpwqe.info);
+err_rq_mkey:
+		mlx5_core_destroy_mkey(mdev, be32_to_cpu(rq->mpwqe.umr_mkey_be));
+err_rq_drop_page:
+		mlx5e_free_mpwqe_rq_drop_page(rq);
+		break;
+	default: /* MLX5_WQ_TYPE_CYCLIC */
+		mlx5e_free_au_list(rq);
+err_rq_frags:
+		kvfree(rq->wqe.frags);
+	}
+err_rq_wq_destroy:
+	mlx5_wq_destroy(&rq->wq_ctrl);
+err_rq_xdp_prog:
+	if (params->xdp_prog)
+		bpf_prog_put(params->xdp_prog);
+
+	return err;
+}
+
+static void mlx5e_free_rq(struct mlx5e_rq *rq)
+{
+	struct bpf_prog *old_prog;
+	int i;
+
+	if (xdp_rxq_info_is_reg(&rq->xdp_rxq)) {
+		old_prog = rcu_dereference_protected(rq->xdp_prog,
+						     lockdep_is_held(&rq->priv->state_lock));
+		if (old_prog)
+			bpf_prog_put(old_prog);
+	}
+
+	switch (rq->wq_type) {
+	case MLX5_WQ_TYPE_LINKED_LIST_STRIDING_RQ:
+		kvfree(rq->mpwqe.info);
+		mlx5_core_destroy_mkey(rq->mdev, be32_to_cpu(rq->mpwqe.umr_mkey_be));
+		mlx5e_free_mpwqe_rq_drop_page(rq);
+		mlx5e_rq_free_shampo(rq);
+		break;
+	default: /* MLX5_WQ_TYPE_CYCLIC */
+		kvfree(rq->wqe.frags);
+		mlx5e_free_au_list(rq);
+	}
+
+	for (i = rq->page_cache.head; i != rq->page_cache.tail;
+	     i = (i + 1) & (MLX5E_CACHE_SIZE - 1)) {
+		/* With AF_XDP, page_cache is not used, so this loop is not
+		 * entered, and it's safe to call mlx5e_page_release_dynamic
+		 * directly.
+		 */
+		mlx5e_page_release_dynamic(rq, rq->page_cache.page_cache[i], false);
+	}
+
+	xdp_rxq_info_unreg(&rq->xdp_rxq);
+	page_pool_destroy(rq->page_pool);
+	mlx5_wq_destroy(&rq->wq_ctrl);
+}
+
+int mlx5e_create_rq(struct mlx5e_rq *rq, struct mlx5e_rq_param *param)
+{
+	struct mlx5_core_dev *mdev = rq->mdev;
+	u8 ts_format;
+	void *in;
+	void *rqc;
+	void *wq;
+	int inlen;
+	int err;
+
+	inlen = MLX5_ST_SZ_BYTES(create_rq_in) +
+		sizeof(u64) * rq->wq_ctrl.buf.npages;
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in)
+		return -ENOMEM;
+
+	ts_format = mlx5_is_real_time_rq(mdev) ?
+			    MLX5_TIMESTAMP_FORMAT_REAL_TIME :
+			    MLX5_TIMESTAMP_FORMAT_FREE_RUNNING;
+	rqc = MLX5_ADDR_OF(create_rq_in, in, ctx);
+	wq  = MLX5_ADDR_OF(rqc, rqc, wq);
+
+	memcpy(rqc, param->rqc, sizeof(param->rqc));
+
+	MLX5_SET(rqc,  rqc, cqn,		rq->cq.mcq.cqn);
+	MLX5_SET(rqc,  rqc, state,		MLX5_RQC_STATE_RST);
+	MLX5_SET(rqc,  rqc, ts_format,		ts_format);
+	MLX5_SET(wq,   wq,  log_wq_pg_sz,	rq->wq_ctrl.buf.page_shift -
+						MLX5_ADAPTER_PAGE_SHIFT);
+	MLX5_SET64(wq, wq,  dbr_addr,		rq->wq_ctrl.db.dma);
+
+	if (test_bit(MLX5E_RQ_STATE_SHAMPO, &rq->state)) {
+		MLX5_SET(wq, wq, log_headers_buffer_entry_num,
+			 order_base_2(rq->mpwqe.shampo->hd_per_wq));
+		MLX5_SET(wq, wq, headers_mkey, rq->mpwqe.shampo->mkey);
+	}
+
+	mlx5_fill_page_frag_array(&rq->wq_ctrl.buf,
+				  (__be64 *)MLX5_ADDR_OF(wq, wq, pas));
+
+	err = mlx5_core_create_rq(mdev, in, inlen, &rq->rqn);
+
+	kvfree(in);
+
+	return err;
+}
+
+static int mlx5e_modify_rq_state(struct mlx5e_rq *rq, int curr_state, int next_state)
+{
+	struct mlx5_core_dev *mdev = rq->mdev;
+
+	void *in;
+	void *rqc;
+	int inlen;
+	int err;
+
+	inlen = MLX5_ST_SZ_BYTES(modify_rq_in);
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in)
+		return -ENOMEM;
+
+	if (curr_state == MLX5_RQC_STATE_RST && next_state == MLX5_RQC_STATE_RDY)
+		mlx5e_rqwq_reset(rq);
+
+	rqc = MLX5_ADDR_OF(modify_rq_in, in, ctx);
+
+	MLX5_SET(modify_rq_in, in, rq_state, curr_state);
+	MLX5_SET(rqc, rqc, state, next_state);
+
+	err = mlx5_core_modify_rq(mdev, rq->rqn, in);
+
+	kvfree(in);
+
+	return err;
+}
+
+static int mlx5e_rq_to_ready(struct mlx5e_rq *rq, int curr_state)
+{
+	struct net_device *dev = rq->netdev;
+	int err;
+
+	err = mlx5e_modify_rq_state(rq, curr_state, MLX5_RQC_STATE_RST);
+	if (err) {
+		netdev_err(dev, "Failed to move rq 0x%x to reset\n", rq->rqn);
+		return err;
+	}
+	err = mlx5e_modify_rq_state(rq, MLX5_RQC_STATE_RST, MLX5_RQC_STATE_RDY);
+	if (err) {
+		netdev_err(dev, "Failed to move rq 0x%x to ready\n", rq->rqn);
+		return err;
+	}
+
+	return 0;
+}
+
+int mlx5e_flush_rq(struct mlx5e_rq *rq, int curr_state)
+{
+	mlx5e_free_rx_descs(rq);
+
+	return mlx5e_rq_to_ready(rq, curr_state);
+}
+
+static int mlx5e_modify_rq_vsd(struct mlx5e_rq *rq, bool vsd)
+{
+	struct mlx5_core_dev *mdev = rq->mdev;
+	void *in;
+	void *rqc;
+	int inlen;
+	int err;
+
+	inlen = MLX5_ST_SZ_BYTES(modify_rq_in);
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in)
+		return -ENOMEM;
+
+	rqc = MLX5_ADDR_OF(modify_rq_in, in, ctx);
+
+	MLX5_SET(modify_rq_in, in, rq_state, MLX5_RQC_STATE_RDY);
+	MLX5_SET64(modify_rq_in, in, modify_bitmask,
+		   MLX5_MODIFY_RQ_IN_MODIFY_BITMASK_VSD);
+	MLX5_SET(rqc, rqc, vsd, vsd);
+	MLX5_SET(rqc, rqc, state, MLX5_RQC_STATE_RDY);
+
+	err = mlx5_core_modify_rq(mdev, rq->rqn, in);
+
+	kvfree(in);
+
+	return err;
+}
+
+void mlx5e_destroy_rq(struct mlx5e_rq *rq)
+{
+	mlx5_core_destroy_rq(rq->mdev, rq->rqn);
+}
+
+int mlx5e_wait_for_min_rx_wqes(struct mlx5e_rq *rq, int wait_time)
+{
+	unsigned long exp_time = jiffies + msecs_to_jiffies(wait_time);
+
+	u16 min_wqes = mlx5_min_rx_wqes(rq->wq_type, mlx5e_rqwq_get_size(rq));
+
+	do {
+		if (mlx5e_rqwq_get_cur_sz(rq) >= min_wqes)
+			return 0;
+
+		msleep(20);
+	} while (time_before(jiffies, exp_time));
+
+	netdev_warn(rq->netdev, "Failed to get min RX wqes on Channel[%d] RQN[0x%x] wq cur_sz(%d) min_rx_wqes(%d)\n",
+		    rq->ix, rq->rqn, mlx5e_rqwq_get_cur_sz(rq), min_wqes);
+
+	mlx5e_reporter_rx_timeout(rq);
+	return -ETIMEDOUT;
+}
+
+void mlx5e_free_rx_in_progress_descs(struct mlx5e_rq *rq)
+{
+	struct mlx5_wq_ll *wq;
+	u16 head;
+	int i;
+
+	if (rq->wq_type != MLX5_WQ_TYPE_LINKED_LIST_STRIDING_RQ)
+		return;
+
+	wq = &rq->mpwqe.wq;
+	head = wq->head;
+
+	/* Outstanding UMR WQEs (in progress) start at wq->head */
+	for (i = 0; i < rq->mpwqe.umr_in_progress; i++) {
+		rq->dealloc_wqe(rq, head);
+		head = mlx5_wq_ll_get_wqe_next_ix(wq, head);
+	}
+
+	if (test_bit(MLX5E_RQ_STATE_SHAMPO, &rq->state)) {
+		u16 len;
+
+		len = (rq->mpwqe.shampo->pi - rq->mpwqe.shampo->ci) &
+		      (rq->mpwqe.shampo->hd_per_wq - 1);
+		mlx5e_shampo_dealloc_hd(rq, len, rq->mpwqe.shampo->ci, false);
+		rq->mpwqe.shampo->pi = rq->mpwqe.shampo->ci;
+	}
+
+	rq->mpwqe.actual_wq_head = wq->head;
+	rq->mpwqe.umr_in_progress = 0;
+	rq->mpwqe.umr_completed = 0;
+}
+
+void mlx5e_free_rx_descs(struct mlx5e_rq *rq)
+{
+	__be16 wqe_ix_be;
+	u16 wqe_ix;
+
+	if (rq->wq_type == MLX5_WQ_TYPE_LINKED_LIST_STRIDING_RQ) {
+		struct mlx5_wq_ll *wq = &rq->mpwqe.wq;
+
+		mlx5e_free_rx_in_progress_descs(rq);
+
+		while (!mlx5_wq_ll_is_empty(wq)) {
+			struct mlx5e_rx_wqe_ll *wqe;
+
+			wqe_ix_be = *wq->tail_next;
+			wqe_ix    = be16_to_cpu(wqe_ix_be);
+			wqe       = mlx5_wq_ll_get_wqe(wq, wqe_ix);
+			rq->dealloc_wqe(rq, wqe_ix);
+			mlx5_wq_ll_pop(wq, wqe_ix_be,
+				       &wqe->next.next_wqe_index);
+		}
+
+		if (test_bit(MLX5E_RQ_STATE_SHAMPO, &rq->state))
+			mlx5e_shampo_dealloc_hd(rq, rq->mpwqe.shampo->hd_per_wq,
+						0, true);
+	} else {
+		struct mlx5_wq_cyc *wq = &rq->wqe.wq;
+
+		while (!mlx5_wq_cyc_is_empty(wq)) {
+			wqe_ix = mlx5_wq_cyc_get_tail(wq);
+			rq->dealloc_wqe(rq, wqe_ix);
+			mlx5_wq_cyc_pop(wq);
+		}
+	}
+
+}
+
+int mlx5e_open_rq(struct mlx5e_params *params, struct mlx5e_rq_param *param,
+		  struct mlx5e_xsk_param *xsk, int node,
+		  struct mlx5e_rq *rq)
+{
+	struct mlx5_core_dev *mdev = rq->mdev;
+	int err;
+
+	if (params->packet_merge.type == MLX5E_PACKET_MERGE_SHAMPO)
+		__set_bit(MLX5E_RQ_STATE_SHAMPO, &rq->state);
+
+	err = mlx5e_alloc_rq(params, xsk, param, node, rq);
+	if (err)
+		return err;
+
+	err = mlx5e_create_rq(rq, param);
+	if (err)
+		goto err_free_rq;
+
+	err = mlx5e_modify_rq_state(rq, MLX5_RQC_STATE_RST, MLX5_RQC_STATE_RDY);
+	if (err)
+		goto err_destroy_rq;
+
+	if (MLX5_CAP_ETH(mdev, cqe_checksum_full))
+		__set_bit(MLX5E_RQ_STATE_CSUM_FULL, &rq->state);
+
+	if (params->rx_dim_enabled)
+		__set_bit(MLX5E_RQ_STATE_AM, &rq->state);
+
+	/* We disable csum_complete when XDP is enabled since
+	 * XDP programs might manipulate packets which will render
+	 * skb->checksum incorrect.
+	 */
+	if (MLX5E_GET_PFLAG(params, MLX5E_PFLAG_RX_NO_CSUM_COMPLETE) || params->xdp_prog)
+		__set_bit(MLX5E_RQ_STATE_NO_CSUM_COMPLETE, &rq->state);
+
+	/* For CQE compression on striding RQ, use stride index provided by
+	 * HW if capability is supported.
+	 */
+	if (MLX5E_GET_PFLAG(params, MLX5E_PFLAG_RX_STRIDING_RQ) &&
+	    MLX5_CAP_GEN(mdev, mini_cqe_resp_stride_index))
+		__set_bit(MLX5E_RQ_STATE_MINI_CQE_HW_STRIDX, &rq->state);
+
+	return 0;
+
+err_destroy_rq:
+	mlx5e_destroy_rq(rq);
+err_free_rq:
+	mlx5e_free_rq(rq);
+
+	return err;
+}
+
+void mlx5e_activate_rq(struct mlx5e_rq *rq)
+{
+	set_bit(MLX5E_RQ_STATE_ENABLED, &rq->state);
+}
+
+void mlx5e_deactivate_rq(struct mlx5e_rq *rq)
+{
+	clear_bit(MLX5E_RQ_STATE_ENABLED, &rq->state);
+	synchronize_net(); /* Sync with NAPI to prevent mlx5e_post_rx_wqes. */
+}
+
+void mlx5e_close_rq(struct mlx5e_rq *rq)
+{
+	cancel_work_sync(&rq->dim.work);
+	cancel_work_sync(&rq->recover_work);
+	mlx5e_destroy_rq(rq);
+	mlx5e_free_rx_descs(rq);
+	mlx5e_free_rq(rq);
+}
+
+static void mlx5e_free_xdpsq_db(struct mlx5e_xdpsq *sq)
+{
+	kvfree(sq->db.xdpi_fifo.xi);
+	kvfree(sq->db.wqe_info);
+}
+
+static int mlx5e_alloc_xdpsq_fifo(struct mlx5e_xdpsq *sq, int numa)
+{
+	struct mlx5e_xdp_info_fifo *xdpi_fifo = &sq->db.xdpi_fifo;
+	int wq_sz        = mlx5_wq_cyc_get_size(&sq->wq);
+	int dsegs_per_wq = wq_sz * MLX5_SEND_WQEBB_NUM_DS;
+	size_t size;
+
+	size = array_size(sizeof(*xdpi_fifo->xi), dsegs_per_wq);
+	xdpi_fifo->xi = kvzalloc_node(size, GFP_KERNEL, numa);
+	if (!xdpi_fifo->xi)
+		return -ENOMEM;
+
+	xdpi_fifo->pc   = &sq->xdpi_fifo_pc;
+	xdpi_fifo->cc   = &sq->xdpi_fifo_cc;
+	xdpi_fifo->mask = dsegs_per_wq - 1;
+
+	return 0;
+}
+
+static int mlx5e_alloc_xdpsq_db(struct mlx5e_xdpsq *sq, int numa)
+{
+	int wq_sz = mlx5_wq_cyc_get_size(&sq->wq);
+	size_t size;
+	int err;
+
+	size = array_size(sizeof(*sq->db.wqe_info), wq_sz);
+	sq->db.wqe_info = kvzalloc_node(size, GFP_KERNEL, numa);
+	if (!sq->db.wqe_info)
+		return -ENOMEM;
+
+	err = mlx5e_alloc_xdpsq_fifo(sq, numa);
+	if (err) {
+		mlx5e_free_xdpsq_db(sq);
+		return err;
+	}
+
+	return 0;
+}
+
+static int mlx5e_alloc_xdpsq(struct mlx5e_channel *c,
+			     struct mlx5e_params *params,
+			     struct xsk_buff_pool *xsk_pool,
+			     struct mlx5e_sq_param *param,
+			     struct mlx5e_xdpsq *sq,
+			     bool is_redirect)
+{
+	void *sqc_wq               = MLX5_ADDR_OF(sqc, param->sqc, wq);
+	struct mlx5_core_dev *mdev = c->mdev;
+	struct mlx5_wq_cyc *wq = &sq->wq;
+	int err;
+
+	sq->pdev      = c->pdev;
+	sq->mkey_be   = c->mkey_be;
+	sq->channel   = c;
+	sq->uar_map   = mdev->mlx5e_res.hw_objs.bfreg.map;
+	sq->min_inline_mode = params->tx_min_inline_mode;
+	sq->hw_mtu    = MLX5E_SW2HW_MTU(params, params->sw_mtu) - ETH_FCS_LEN;
+	sq->xsk_pool  = xsk_pool;
+
+	sq->stats = sq->xsk_pool ?
+		&c->priv->channel_stats[c->ix]->xsksq :
+		is_redirect ?
+			&c->priv->channel_stats[c->ix]->xdpsq :
+			&c->priv->channel_stats[c->ix]->rq_xdpsq;
+	sq->stop_room = param->is_mpw ? mlx5e_stop_room_for_mpwqe(mdev) :
+					mlx5e_stop_room_for_max_wqe(mdev);
+	sq->max_sq_mpw_wqebbs = mlx5e_get_max_sq_aligned_wqebbs(mdev);
+
+	param->wq.db_numa_node = cpu_to_node(c->cpu);
+	err = mlx5_wq_cyc_create(mdev, &param->wq, sqc_wq, wq, &sq->wq_ctrl);
+	if (err)
+		return err;
+	wq->db = &wq->db[MLX5_SND_DBR];
+
+	err = mlx5e_alloc_xdpsq_db(sq, cpu_to_node(c->cpu));
+	if (err)
+		goto err_sq_wq_destroy;
+
+	return 0;
+
+err_sq_wq_destroy:
+	mlx5_wq_destroy(&sq->wq_ctrl);
+
+	return err;
+}
+
+static void mlx5e_free_xdpsq(struct mlx5e_xdpsq *sq)
+{
+	mlx5e_free_xdpsq_db(sq);
+	mlx5_wq_destroy(&sq->wq_ctrl);
+}
+
+static void mlx5e_free_icosq_db(struct mlx5e_icosq *sq)
+{
+	kvfree(sq->db.wqe_info);
+}
+
+static int mlx5e_alloc_icosq_db(struct mlx5e_icosq *sq, int numa)
+{
+	int wq_sz = mlx5_wq_cyc_get_size(&sq->wq);
+	size_t size;
+
+	size = array_size(wq_sz, sizeof(*sq->db.wqe_info));
+	sq->db.wqe_info = kvzalloc_node(size, GFP_KERNEL, numa);
+	if (!sq->db.wqe_info)
+		return -ENOMEM;
+
+	return 0;
+}
+
+static void mlx5e_icosq_err_cqe_work(struct work_struct *recover_work)
+{
+	struct mlx5e_icosq *sq = container_of(recover_work, struct mlx5e_icosq,
+					      recover_work);
+
+	mlx5e_reporter_icosq_cqe_err(sq);
+}
+
+static void mlx5e_async_icosq_err_cqe_work(struct work_struct *recover_work)
+{
+	struct mlx5e_icosq *sq = container_of(recover_work, struct mlx5e_icosq,
+					      recover_work);
+
+	/* Not implemented yet. */
+
+	netdev_warn(sq->channel->netdev, "async_icosq recovery is not implemented\n");
+}
+
+static int mlx5e_alloc_icosq(struct mlx5e_channel *c,
+			     struct mlx5e_sq_param *param,
+			     struct mlx5e_icosq *sq,
+			     work_func_t recover_work_func)
+{
+	void *sqc_wq               = MLX5_ADDR_OF(sqc, param->sqc, wq);
+	struct mlx5_core_dev *mdev = c->mdev;
+	struct mlx5_wq_cyc *wq = &sq->wq;
+	int err;
+
+	sq->channel   = c;
+	sq->uar_map   = mdev->mlx5e_res.hw_objs.bfreg.map;
+	sq->reserved_room = param->stop_room;
+
+	param->wq.db_numa_node = cpu_to_node(c->cpu);
+	err = mlx5_wq_cyc_create(mdev, &param->wq, sqc_wq, wq, &sq->wq_ctrl);
+	if (err)
+		return err;
+	wq->db = &wq->db[MLX5_SND_DBR];
+
+	err = mlx5e_alloc_icosq_db(sq, cpu_to_node(c->cpu));
+	if (err)
+		goto err_sq_wq_destroy;
+
+	INIT_WORK(&sq->recover_work, recover_work_func);
+
+	return 0;
+
+err_sq_wq_destroy:
+	mlx5_wq_destroy(&sq->wq_ctrl);
+
+	return err;
+}
+
+static void mlx5e_free_icosq(struct mlx5e_icosq *sq)
+{
+	mlx5e_free_icosq_db(sq);
+	mlx5_wq_destroy(&sq->wq_ctrl);
+}
+
+void mlx5e_free_txqsq_db(struct mlx5e_txqsq *sq)
+{
+	kvfree(sq->db.wqe_info);
+	kvfree(sq->db.skb_fifo.fifo);
+	kvfree(sq->db.dma_fifo);
+}
+
+int mlx5e_alloc_txqsq_db(struct mlx5e_txqsq *sq, int numa)
+{
+	int wq_sz = mlx5_wq_cyc_get_size(&sq->wq);
+	int df_sz = wq_sz * MLX5_SEND_WQEBB_NUM_DS;
+
+	sq->db.dma_fifo = kvzalloc_node(array_size(df_sz,
+						   sizeof(*sq->db.dma_fifo)),
+					GFP_KERNEL, numa);
+	sq->db.skb_fifo.fifo = kvzalloc_node(array_size(df_sz,
+							sizeof(*sq->db.skb_fifo.fifo)),
+					GFP_KERNEL, numa);
+	sq->db.wqe_info = kvzalloc_node(array_size(wq_sz,
+						   sizeof(*sq->db.wqe_info)),
+					GFP_KERNEL, numa);
+	if (!sq->db.dma_fifo || !sq->db.skb_fifo.fifo || !sq->db.wqe_info) {
+		mlx5e_free_txqsq_db(sq);
+		return -ENOMEM;
+	}
+
+	sq->dma_fifo_mask = df_sz - 1;
+
+	sq->db.skb_fifo.pc   = &sq->skb_fifo_pc;
+	sq->db.skb_fifo.cc   = &sq->skb_fifo_cc;
+	sq->db.skb_fifo.mask = df_sz - 1;
+
+	return 0;
+}
+
+static int mlx5e_alloc_txqsq(struct mlx5e_channel *c,
+			     int txq_ix,
+			     struct mlx5e_params *params,
+			     struct mlx5e_sq_param *param,
+			     struct mlx5e_txqsq *sq,
+			     int tc)
+{
+	void *sqc_wq               = MLX5_ADDR_OF(sqc, param->sqc, wq);
+	struct mlx5_core_dev *mdev = c->mdev;
+	struct mlx5_wq_cyc *wq = &sq->wq;
+	int err;
+
+	sq->pdev      = c->pdev;
+	sq->clock     = &mdev->clock;
+	sq->mkey_be   = c->mkey_be;
+	sq->netdev    = c->netdev;
+	sq->mdev      = c->mdev;
+	sq->priv      = c->priv;
+	sq->ch_ix     = c->ix;
+	sq->txq_ix    = txq_ix;
+	sq->uar_map   = mdev->mlx5e_res.hw_objs.bfreg.map;
+	sq->min_inline_mode = params->tx_min_inline_mode;
+	sq->hw_mtu    = MLX5E_SW2HW_MTU(params, params->sw_mtu);
+	sq->max_sq_mpw_wqebbs = mlx5e_get_max_sq_aligned_wqebbs(mdev);
+	INIT_WORK(&sq->recover_work, mlx5e_tx_err_cqe_work);
+	if (!MLX5_CAP_ETH(mdev, wqe_vlan_insert))
+		set_bit(MLX5E_SQ_STATE_VLAN_NEED_L2_INLINE, &sq->state);
+	if (mlx5_ipsec_device_caps(c->priv->mdev))
+		set_bit(MLX5E_SQ_STATE_IPSEC, &sq->state);
+	if (param->is_mpw)
+		set_bit(MLX5E_SQ_STATE_MPWQE, &sq->state);
+	sq->stop_room = param->stop_room;
+	sq->ptp_cyc2time = mlx5_sq_ts_translator(mdev);
+
+	param->wq.db_numa_node = cpu_to_node(c->cpu);
+	err = mlx5_wq_cyc_create(mdev, &param->wq, sqc_wq, wq, &sq->wq_ctrl);
+	if (err)
+		return err;
+	wq->db    = &wq->db[MLX5_SND_DBR];
+
+	err = mlx5e_alloc_txqsq_db(sq, cpu_to_node(c->cpu));
+	if (err)
+		goto err_sq_wq_destroy;
+
+	INIT_WORK(&sq->dim.work, mlx5e_tx_dim_work);
+	sq->dim.mode = params->tx_cq_moderation.cq_period_mode;
+
+	return 0;
+
+err_sq_wq_destroy:
+	mlx5_wq_destroy(&sq->wq_ctrl);
+
+	return err;
+}
+
+void mlx5e_free_txqsq(struct mlx5e_txqsq *sq)
+{
+	mlx5e_free_txqsq_db(sq);
+	mlx5_wq_destroy(&sq->wq_ctrl);
+}
+
+static int mlx5e_create_sq(struct mlx5_core_dev *mdev,
+			   struct mlx5e_sq_param *param,
+			   struct mlx5e_create_sq_param *csp,
+			   u32 *sqn)
+{
+	u8 ts_format;
+	void *in;
+	void *sqc;
+	void *wq;
+	int inlen;
+	int err;
+
+	inlen = MLX5_ST_SZ_BYTES(create_sq_in) +
+		sizeof(u64) * csp->wq_ctrl->buf.npages;
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in)
+		return -ENOMEM;
+
+	ts_format = mlx5_is_real_time_sq(mdev) ?
+			    MLX5_TIMESTAMP_FORMAT_REAL_TIME :
+			    MLX5_TIMESTAMP_FORMAT_FREE_RUNNING;
+	sqc = MLX5_ADDR_OF(create_sq_in, in, ctx);
+	wq = MLX5_ADDR_OF(sqc, sqc, wq);
+
+	memcpy(sqc, param->sqc, sizeof(param->sqc));
+	MLX5_SET(sqc,  sqc, tis_lst_sz, csp->tis_lst_sz);
+	MLX5_SET(sqc,  sqc, tis_num_0, csp->tisn);
+	MLX5_SET(sqc,  sqc, cqn, csp->cqn);
+	MLX5_SET(sqc,  sqc, ts_cqe_to_dest_cqn, csp->ts_cqe_to_dest_cqn);
+	MLX5_SET(sqc,  sqc, ts_format, ts_format);
+
+
+	if (MLX5_CAP_ETH(mdev, wqe_inline_mode) == MLX5_CAP_INLINE_MODE_VPORT_CONTEXT)
+		MLX5_SET(sqc,  sqc, min_wqe_inline_mode, csp->min_inline_mode);
+
+	MLX5_SET(sqc,  sqc, state, MLX5_SQC_STATE_RST);
+	MLX5_SET(sqc,  sqc, flush_in_error_en, 1);
+
+	MLX5_SET(wq,   wq, wq_type,       MLX5_WQ_TYPE_CYCLIC);
+	MLX5_SET(wq,   wq, uar_page,      mdev->mlx5e_res.hw_objs.bfreg.index);
+	MLX5_SET(wq,   wq, log_wq_pg_sz,  csp->wq_ctrl->buf.page_shift -
+					  MLX5_ADAPTER_PAGE_SHIFT);
+	MLX5_SET64(wq, wq, dbr_addr,      csp->wq_ctrl->db.dma);
+
+	mlx5_fill_page_frag_array(&csp->wq_ctrl->buf,
+				  (__be64 *)MLX5_ADDR_OF(wq, wq, pas));
+
+	err = mlx5_core_create_sq(mdev, in, inlen, sqn);
+
+	kvfree(in);
+
+	return err;
+}
+
+int mlx5e_modify_sq(struct mlx5_core_dev *mdev, u32 sqn,
+		    struct mlx5e_modify_sq_param *p)
+{
+	u64 bitmask = 0;
+	void *in;
+	void *sqc;
+	int inlen;
+	int err;
+
+	inlen = MLX5_ST_SZ_BYTES(modify_sq_in);
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in)
+		return -ENOMEM;
+
+	sqc = MLX5_ADDR_OF(modify_sq_in, in, ctx);
+
+	MLX5_SET(modify_sq_in, in, sq_state, p->curr_state);
+	MLX5_SET(sqc, sqc, state, p->next_state);
+	if (p->rl_update && p->next_state == MLX5_SQC_STATE_RDY) {
+		bitmask |= 1;
+		MLX5_SET(sqc, sqc, packet_pacing_rate_limit_index, p->rl_index);
+	}
+	if (p->qos_update && p->next_state == MLX5_SQC_STATE_RDY) {
+		bitmask |= 1 << 2;
+		MLX5_SET(sqc, sqc, qos_queue_group_id, p->qos_queue_group_id);
+	}
+	MLX5_SET64(modify_sq_in, in, modify_bitmask, bitmask);
+
+	err = mlx5_core_modify_sq(mdev, sqn, in);
+
+	kvfree(in);
+
+	return err;
+}
+
+static void mlx5e_destroy_sq(struct mlx5_core_dev *mdev, u32 sqn)
+{
+	mlx5_core_destroy_sq(mdev, sqn);
+}
+
+int mlx5e_create_sq_rdy(struct mlx5_core_dev *mdev,
+			struct mlx5e_sq_param *param,
+			struct mlx5e_create_sq_param *csp,
+			u16 qos_queue_group_id,
+			u32 *sqn)
+{
+	struct mlx5e_modify_sq_param msp = {0};
+	int err;
+
+	err = mlx5e_create_sq(mdev, param, csp, sqn);
+	if (err)
+		return err;
+
+	msp.curr_state = MLX5_SQC_STATE_RST;
+	msp.next_state = MLX5_SQC_STATE_RDY;
+	if (qos_queue_group_id) {
+		msp.qos_update = true;
+		msp.qos_queue_group_id = qos_queue_group_id;
+	}
+	err = mlx5e_modify_sq(mdev, *sqn, &msp);
+	if (err)
+		mlx5e_destroy_sq(mdev, *sqn);
+
+	return err;
+}
+
+static int mlx5e_set_sq_maxrate(struct net_device *dev,
+				struct mlx5e_txqsq *sq, u32 rate);
+
+int mlx5e_open_txqsq(struct mlx5e_channel *c, u32 tisn, int txq_ix,
+		     struct mlx5e_params *params, struct mlx5e_sq_param *param,
+		     struct mlx5e_txqsq *sq, int tc, u16 qos_queue_group_id,
+		     struct mlx5e_sq_stats *sq_stats)
+{
+	struct mlx5e_create_sq_param csp = {};
+	u32 tx_rate;
+	int err;
+
+	err = mlx5e_alloc_txqsq(c, txq_ix, params, param, sq, tc);
+	if (err)
+		return err;
+
+	sq->stats = sq_stats;
+
+	csp.tisn            = tisn;
+	csp.tis_lst_sz      = 1;
+	csp.cqn             = sq->cq.mcq.cqn;
+	csp.wq_ctrl         = &sq->wq_ctrl;
+	csp.min_inline_mode = sq->min_inline_mode;
+	err = mlx5e_create_sq_rdy(c->mdev, param, &csp, qos_queue_group_id, &sq->sqn);
+	if (err)
+		goto err_free_txqsq;
+
+	tx_rate = c->priv->tx_rates[sq->txq_ix];
+	if (tx_rate)
+		mlx5e_set_sq_maxrate(c->netdev, sq, tx_rate);
+
+	if (params->tx_dim_enabled)
+		sq->state |= BIT(MLX5E_SQ_STATE_AM);
+
+	return 0;
+
+err_free_txqsq:
+	mlx5e_free_txqsq(sq);
+
+	return err;
+}
+
+void mlx5e_activate_txqsq(struct mlx5e_txqsq *sq)
+{
+	sq->txq = netdev_get_tx_queue(sq->netdev, sq->txq_ix);
+	set_bit(MLX5E_SQ_STATE_ENABLED, &sq->state);
+	netdev_tx_reset_queue(sq->txq);
+	netif_tx_start_queue(sq->txq);
+}
+
+void mlx5e_tx_disable_queue(struct netdev_queue *txq)
+{
+	__netif_tx_lock_bh(txq);
+	netif_tx_stop_queue(txq);
+	__netif_tx_unlock_bh(txq);
+}
+
+void mlx5e_deactivate_txqsq(struct mlx5e_txqsq *sq)
+{
+	struct mlx5_wq_cyc *wq = &sq->wq;
+
+	clear_bit(MLX5E_SQ_STATE_ENABLED, &sq->state);
+	synchronize_net(); /* Sync with NAPI to prevent netif_tx_wake_queue. */
+
+	mlx5e_tx_disable_queue(sq->txq);
+
+	/* last doorbell out, godspeed .. */
+	if (mlx5e_wqc_has_room_for(wq, sq->cc, sq->pc, 1)) {
+		u16 pi = mlx5_wq_cyc_ctr2ix(wq, sq->pc);
+		struct mlx5e_tx_wqe *nop;
+
+		sq->db.wqe_info[pi] = (struct mlx5e_tx_wqe_info) {
+			.num_wqebbs = 1,
+		};
+
+		nop = mlx5e_post_nop(wq, sq->sqn, &sq->pc);
+		mlx5e_notify_hw(wq, sq->pc, sq->uar_map, &nop->ctrl);
+	}
+}
+
+void mlx5e_close_txqsq(struct mlx5e_txqsq *sq)
+{
+	struct mlx5_core_dev *mdev = sq->mdev;
+	struct mlx5_rate_limit rl = {0};
+
+	cancel_work_sync(&sq->dim.work);
+	cancel_work_sync(&sq->recover_work);
+	mlx5e_destroy_sq(mdev, sq->sqn);
+	if (sq->rate_limit) {
+		rl.rate = sq->rate_limit;
+		mlx5_rl_remove_rate(mdev, &rl);
+	}
+	mlx5e_free_txqsq_descs(sq);
+	mlx5e_free_txqsq(sq);
+}
+
+void mlx5e_tx_err_cqe_work(struct work_struct *recover_work)
+{
+	struct mlx5e_txqsq *sq = container_of(recover_work, struct mlx5e_txqsq,
+					      recover_work);
+
+	mlx5e_reporter_tx_err_cqe(sq);
+}
+
+static int mlx5e_open_icosq(struct mlx5e_channel *c, struct mlx5e_params *params,
+			    struct mlx5e_sq_param *param, struct mlx5e_icosq *sq,
+			    work_func_t recover_work_func)
+{
+	struct mlx5e_create_sq_param csp = {};
+	int err;
+
+	err = mlx5e_alloc_icosq(c, param, sq, recover_work_func);
+	if (err)
+		return err;
+
+	csp.cqn             = sq->cq.mcq.cqn;
+	csp.wq_ctrl         = &sq->wq_ctrl;
+	csp.min_inline_mode = params->tx_min_inline_mode;
+	err = mlx5e_create_sq_rdy(c->mdev, param, &csp, 0, &sq->sqn);
+	if (err)
+		goto err_free_icosq;
+
+	if (param->is_tls) {
+		sq->ktls_resync = mlx5e_ktls_rx_resync_create_resp_list();
+		if (IS_ERR(sq->ktls_resync)) {
+			err = PTR_ERR(sq->ktls_resync);
+			goto err_destroy_icosq;
+		}
+	}
+	return 0;
+
+err_destroy_icosq:
+	mlx5e_destroy_sq(c->mdev, sq->sqn);
+err_free_icosq:
+	mlx5e_free_icosq(sq);
+
+	return err;
+}
+
+void mlx5e_activate_icosq(struct mlx5e_icosq *icosq)
+{
+	set_bit(MLX5E_SQ_STATE_ENABLED, &icosq->state);
+}
+
+void mlx5e_deactivate_icosq(struct mlx5e_icosq *icosq)
+{
+	clear_bit(MLX5E_SQ_STATE_ENABLED, &icosq->state);
+	synchronize_net(); /* Sync with NAPI. */
+}
+
+static void mlx5e_close_icosq(struct mlx5e_icosq *sq)
+{
+	struct mlx5e_channel *c = sq->channel;
+
+	if (sq->ktls_resync)
+		mlx5e_ktls_rx_resync_destroy_resp_list(sq->ktls_resync);
+	mlx5e_destroy_sq(c->mdev, sq->sqn);
+	mlx5e_free_icosq_descs(sq);
+	mlx5e_free_icosq(sq);
+}
+
+int mlx5e_open_xdpsq(struct mlx5e_channel *c, struct mlx5e_params *params,
+		     struct mlx5e_sq_param *param, struct xsk_buff_pool *xsk_pool,
+		     struct mlx5e_xdpsq *sq, bool is_redirect)
+{
+	struct mlx5e_create_sq_param csp = {};
+	int err;
+
+	err = mlx5e_alloc_xdpsq(c, params, xsk_pool, param, sq, is_redirect);
+	if (err)
+		return err;
+
+	csp.tis_lst_sz      = 1;
+	csp.tisn            = c->priv->tisn[c->lag_port][0]; /* tc = 0 */
+	csp.cqn             = sq->cq.mcq.cqn;
+	csp.wq_ctrl         = &sq->wq_ctrl;
+	csp.min_inline_mode = sq->min_inline_mode;
+	set_bit(MLX5E_SQ_STATE_ENABLED, &sq->state);
+
+	/* Don't enable multi buffer on XDP_REDIRECT SQ, as it's not yet
+	 * supported by upstream, and there is no defined trigger to allow
+	 * transmitting redirected multi-buffer frames.
+	 */
+	if (param->is_xdp_mb && !is_redirect)
+		set_bit(MLX5E_SQ_STATE_XDP_MULTIBUF, &sq->state);
+
+	err = mlx5e_create_sq_rdy(c->mdev, param, &csp, 0, &sq->sqn);
+	if (err)
+		goto err_free_xdpsq;
+
+	mlx5e_set_xmit_fp(sq, param->is_mpw);
+
+	if (!param->is_mpw && !test_bit(MLX5E_SQ_STATE_XDP_MULTIBUF, &sq->state)) {
+		unsigned int ds_cnt = MLX5E_TX_WQE_EMPTY_DS_COUNT + 1;
+		unsigned int inline_hdr_sz = 0;
+		int i;
+
+		if (sq->min_inline_mode != MLX5_INLINE_MODE_NONE) {
+			inline_hdr_sz = MLX5E_XDP_MIN_INLINE;
+			ds_cnt++;
+		}
+
+		/* Pre initialize fixed WQE fields */
+		for (i = 0; i < mlx5_wq_cyc_get_size(&sq->wq); i++) {
+			struct mlx5e_tx_wqe      *wqe  = mlx5_wq_cyc_get_wqe(&sq->wq, i);
+			struct mlx5_wqe_ctrl_seg *cseg = &wqe->ctrl;
+			struct mlx5_wqe_eth_seg  *eseg = &wqe->eth;
+			struct mlx5_wqe_data_seg *dseg;
+
+			sq->db.wqe_info[i] = (struct mlx5e_xdp_wqe_info) {
+				.num_wqebbs = 1,
+				.num_pkts   = 1,
+			};
+
+			cseg->qpn_ds = cpu_to_be32((sq->sqn << 8) | ds_cnt);
+			eseg->inline_hdr.sz = cpu_to_be16(inline_hdr_sz);
+
+			dseg = (struct mlx5_wqe_data_seg *)cseg + (ds_cnt - 1);
+			dseg->lkey = sq->mkey_be;
+		}
+	}
+
+	return 0;
+
+err_free_xdpsq:
+	clear_bit(MLX5E_SQ_STATE_ENABLED, &sq->state);
+	mlx5e_free_xdpsq(sq);
+
+	return err;
+}
+
+void mlx5e_close_xdpsq(struct mlx5e_xdpsq *sq)
+{
+	struct mlx5e_channel *c = sq->channel;
+
+	clear_bit(MLX5E_SQ_STATE_ENABLED, &sq->state);
+	synchronize_net(); /* Sync with NAPI. */
+
+	mlx5e_destroy_sq(c->mdev, sq->sqn);
+	mlx5e_free_xdpsq_descs(sq);
+	mlx5e_free_xdpsq(sq);
+}
+
+static int mlx5e_alloc_cq_common(struct mlx5e_priv *priv,
+				 struct mlx5e_cq_param *param,
+				 struct mlx5e_cq *cq)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+	struct mlx5_core_cq *mcq = &cq->mcq;
+	int err;
+	u32 i;
+
+	err = mlx5_cqwq_create(mdev, &param->wq, param->cqc, &cq->wq,
+			       &cq->wq_ctrl);
+	if (err)
+		return err;
+
+	mcq->cqe_sz     = 64;
+	mcq->set_ci_db  = cq->wq_ctrl.db.db;
+	mcq->arm_db     = cq->wq_ctrl.db.db + 1;
+	*mcq->set_ci_db = 0;
+	*mcq->arm_db    = 0;
+	mcq->vector     = param->eq_ix;
+	mcq->comp       = mlx5e_completion_event;
+	mcq->event      = mlx5e_cq_error_event;
+
+	for (i = 0; i < mlx5_cqwq_get_size(&cq->wq); i++) {
+		struct mlx5_cqe64 *cqe = mlx5_cqwq_get_wqe(&cq->wq, i);
+
+		cqe->op_own = 0xf1;
+	}
+
+	cq->mdev = mdev;
+	cq->netdev = priv->netdev;
+	cq->priv = priv;
+
+	return 0;
+}
+
+static int mlx5e_alloc_cq(struct mlx5e_priv *priv,
+			  struct mlx5e_cq_param *param,
+			  struct mlx5e_create_cq_param *ccp,
+			  struct mlx5e_cq *cq)
+{
+	int err;
+
+	param->wq.buf_numa_node = ccp->node;
+	param->wq.db_numa_node  = ccp->node;
+	param->eq_ix            = ccp->ix;
+
+	err = mlx5e_alloc_cq_common(priv, param, cq);
+
+	cq->napi     = ccp->napi;
+	cq->ch_stats = ccp->ch_stats;
+
+	return err;
+}
+
+static void mlx5e_free_cq(struct mlx5e_cq *cq)
+{
+	mlx5_wq_destroy(&cq->wq_ctrl);
+}
+
+static int mlx5e_create_cq(struct mlx5e_cq *cq, struct mlx5e_cq_param *param)
+{
+	u32 out[MLX5_ST_SZ_DW(create_cq_out)];
+	struct mlx5_core_dev *mdev = cq->mdev;
+	struct mlx5_core_cq *mcq = &cq->mcq;
+
+	void *in;
+	void *cqc;
+	int inlen;
+	int eqn;
+	int err;
+
+	err = mlx5_vector2eqn(mdev, param->eq_ix, &eqn);
+	if (err)
+		return err;
+
+	inlen = MLX5_ST_SZ_BYTES(create_cq_in) +
+		sizeof(u64) * cq->wq_ctrl.buf.npages;
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in)
+		return -ENOMEM;
+
+	cqc = MLX5_ADDR_OF(create_cq_in, in, cq_context);
+
+	memcpy(cqc, param->cqc, sizeof(param->cqc));
+
+	mlx5_fill_page_frag_array(&cq->wq_ctrl.buf,
+				  (__be64 *)MLX5_ADDR_OF(create_cq_in, in, pas));
+
+	MLX5_SET(cqc,   cqc, cq_period_mode, param->cq_period_mode);
+	MLX5_SET(cqc,   cqc, c_eqn_or_apu_element, eqn);
+	MLX5_SET(cqc,   cqc, uar_page,      mdev->priv.uar->index);
+	MLX5_SET(cqc,   cqc, log_page_size, cq->wq_ctrl.buf.page_shift -
+					    MLX5_ADAPTER_PAGE_SHIFT);
+	MLX5_SET64(cqc, cqc, dbr_addr,      cq->wq_ctrl.db.dma);
+
+	err = mlx5_core_create_cq(mdev, mcq, in, inlen, out, sizeof(out));
+
+	kvfree(in);
+
+	if (err)
+		return err;
+
+	mlx5e_cq_arm(cq);
+
+	return 0;
+}
+
+static void mlx5e_destroy_cq(struct mlx5e_cq *cq)
+{
+	mlx5_core_destroy_cq(cq->mdev, &cq->mcq);
+}
+
+int mlx5e_open_cq(struct mlx5e_priv *priv, struct dim_cq_moder moder,
+		  struct mlx5e_cq_param *param, struct mlx5e_create_cq_param *ccp,
+		  struct mlx5e_cq *cq)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+	int err;
+
+	err = mlx5e_alloc_cq(priv, param, ccp, cq);
+	if (err)
+		return err;
+
+	err = mlx5e_create_cq(cq, param);
+	if (err)
+		goto err_free_cq;
+
+	if (MLX5_CAP_GEN(mdev, cq_moderation))
+		mlx5_core_modify_cq_moderation(mdev, &cq->mcq, moder.usec, moder.pkts);
+	return 0;
+
+err_free_cq:
+	mlx5e_free_cq(cq);
+
+	return err;
+}
+
+void mlx5e_close_cq(struct mlx5e_cq *cq)
+{
+	mlx5e_destroy_cq(cq);
+	mlx5e_free_cq(cq);
+}
+
+static int mlx5e_open_tx_cqs(struct mlx5e_channel *c,
+			     struct mlx5e_params *params,
+			     struct mlx5e_create_cq_param *ccp,
+			     struct mlx5e_channel_param *cparam)
+{
+	int err;
+	int tc;
+
+	for (tc = 0; tc < c->num_tc; tc++) {
+		err = mlx5e_open_cq(c->priv, params->tx_cq_moderation, &cparam->txq_sq.cqp,
+				    ccp, &c->sq[tc].cq);
+		if (err)
+			goto err_close_tx_cqs;
+	}
+
+	return 0;
+
+err_close_tx_cqs:
+	for (tc--; tc >= 0; tc--)
+		mlx5e_close_cq(&c->sq[tc].cq);
+
+	return err;
+}
+
+static void mlx5e_close_tx_cqs(struct mlx5e_channel *c)
+{
+	int tc;
+
+	for (tc = 0; tc < c->num_tc; tc++)
+		mlx5e_close_cq(&c->sq[tc].cq);
+}
+
+static int mlx5e_mqprio_txq_to_tc(struct netdev_tc_txq *tc_to_txq, unsigned int txq)
+{
+	int tc;
+
+	for (tc = 0; tc < TC_MAX_QUEUE; tc++)
+		if (txq - tc_to_txq[tc].offset < tc_to_txq[tc].count)
+			return tc;
+
+	WARN(1, "Unexpected TCs configuration. No match found for txq %u", txq);
+	return -ENOENT;
+}
+
+static int mlx5e_txq_get_qos_node_hw_id(struct mlx5e_params *params, int txq_ix,
+					u32 *hw_id)
+{
+	int tc;
+
+	if (params->mqprio.mode != TC_MQPRIO_MODE_CHANNEL) {
+		*hw_id = 0;
+		return 0;
+	}
+
+	tc = mlx5e_mqprio_txq_to_tc(params->mqprio.tc_to_txq, txq_ix);
+	if (tc < 0)
+		return tc;
+
+	if (tc >= params->mqprio.num_tc) {
+		WARN(1, "Unexpected TCs configuration. tc %d is out of range of %u",
+		     tc, params->mqprio.num_tc);
+		return -EINVAL;
+	}
+
+	*hw_id = params->mqprio.channel.hw_id[tc];
+	return 0;
+}
+
+static int mlx5e_open_sqs(struct mlx5e_channel *c,
+			  struct mlx5e_params *params,
+			  struct mlx5e_channel_param *cparam)
+{
+	int err, tc;
+
+	for (tc = 0; tc < mlx5e_get_dcb_num_tc(params); tc++) {
+		int txq_ix = c->ix + tc * params->num_channels;
+		u32 qos_queue_group_id;
+
+		err = mlx5e_txq_get_qos_node_hw_id(params, txq_ix, &qos_queue_group_id);
+		if (err)
+			goto err_close_sqs;
+
+		err = mlx5e_open_txqsq(c, c->priv->tisn[c->lag_port][tc], txq_ix,
+				       params, &cparam->txq_sq, &c->sq[tc], tc,
+				       qos_queue_group_id,
+				       &c->priv->channel_stats[c->ix]->sq[tc]);
+		if (err)
+			goto err_close_sqs;
+	}
+
+	return 0;
+
+err_close_sqs:
+	for (tc--; tc >= 0; tc--)
+		mlx5e_close_txqsq(&c->sq[tc]);
+
+	return err;
+}
+
+static void mlx5e_close_sqs(struct mlx5e_channel *c)
+{
+	int tc;
+
+	for (tc = 0; tc < c->num_tc; tc++)
+		mlx5e_close_txqsq(&c->sq[tc]);
+}
+
+static int mlx5e_set_sq_maxrate(struct net_device *dev,
+				struct mlx5e_txqsq *sq, u32 rate)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+	struct mlx5e_modify_sq_param msp = {0};
+	struct mlx5_rate_limit rl = {0};
+	u16 rl_index = 0;
+	int err;
+
+	if (rate == sq->rate_limit)
+		/* nothing to do */
+		return 0;
+
+	if (sq->rate_limit) {
+		rl.rate = sq->rate_limit;
+		/* remove current rl index to free space to next ones */
+		mlx5_rl_remove_rate(mdev, &rl);
+	}
+
+	sq->rate_limit = 0;
+
+	if (rate) {
+		rl.rate = rate;
+		err = mlx5_rl_add_rate(mdev, &rl_index, &rl);
+		if (err) {
+			netdev_err(dev, "Failed configuring rate %u: %d\n",
+				   rate, err);
+			return err;
+		}
+	}
+
+	msp.curr_state = MLX5_SQC_STATE_RDY;
+	msp.next_state = MLX5_SQC_STATE_RDY;
+	msp.rl_index   = rl_index;
+	msp.rl_update  = true;
+	err = mlx5e_modify_sq(mdev, sq->sqn, &msp);
+	if (err) {
+		netdev_err(dev, "Failed configuring rate %u: %d\n",
+			   rate, err);
+		/* remove the rate from the table */
+		if (rate)
+			mlx5_rl_remove_rate(mdev, &rl);
+		return err;
+	}
+
+	sq->rate_limit = rate;
+	return 0;
+}
+
+static int mlx5e_set_tx_maxrate(struct net_device *dev, int index, u32 rate)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+	struct mlx5e_txqsq *sq = priv->txq2sq[index];
+	int err = 0;
+
+	if (!mlx5_rl_is_supported(mdev)) {
+		netdev_err(dev, "Rate limiting is not supported on this device\n");
+		return -EINVAL;
+	}
+
+	/* rate is given in Mb/sec, HW config is in Kb/sec */
+	rate = rate << 10;
+
+	/* Check whether rate in valid range, 0 is always valid */
+	if (rate && !mlx5_rl_is_in_range(mdev, rate)) {
+		netdev_err(dev, "TX rate %u, is not in range\n", rate);
+		return -ERANGE;
+	}
+
+	mutex_lock(&priv->state_lock);
+	if (test_bit(MLX5E_STATE_OPENED, &priv->state))
+		err = mlx5e_set_sq_maxrate(dev, sq, rate);
+	if (!err)
+		priv->tx_rates[index] = rate;
+	mutex_unlock(&priv->state_lock);
+
+	return err;
+}
+
+static int mlx5e_open_rxq_rq(struct mlx5e_channel *c, struct mlx5e_params *params,
+			     struct mlx5e_rq_param *rq_params)
+{
+	int err;
+
+	err = mlx5e_init_rxq_rq(c, params, &c->rq);
+	if (err)
+		return err;
+
+	return mlx5e_open_rq(params, rq_params, NULL, cpu_to_node(c->cpu), &c->rq);
+}
+
+static int mlx5e_open_queues(struct mlx5e_channel *c,
+			     struct mlx5e_params *params,
+			     struct mlx5e_channel_param *cparam)
+{
+	struct dim_cq_moder icocq_moder = {0, 0};
+	struct mlx5e_create_cq_param ccp;
+	int err;
+
+	mlx5e_build_create_cq_param(&ccp, c);
+
+	err = mlx5e_open_cq(c->priv, icocq_moder, &cparam->async_icosq.cqp, &ccp,
+			    &c->async_icosq.cq);
+	if (err)
+		return err;
+
+	err = mlx5e_open_cq(c->priv, icocq_moder, &cparam->icosq.cqp, &ccp,
+			    &c->icosq.cq);
+	if (err)
+		goto err_close_async_icosq_cq;
+
+	err = mlx5e_open_tx_cqs(c, params, &ccp, cparam);
+	if (err)
+		goto err_close_icosq_cq;
+
+	err = mlx5e_open_cq(c->priv, params->tx_cq_moderation, &cparam->xdp_sq.cqp, &ccp,
+			    &c->xdpsq.cq);
+	if (err)
+		goto err_close_tx_cqs;
+
+	err = mlx5e_open_cq(c->priv, params->rx_cq_moderation, &cparam->rq.cqp, &ccp,
+			    &c->rq.cq);
+	if (err)
+		goto err_close_xdp_tx_cqs;
+
+	err = c->xdp ? mlx5e_open_cq(c->priv, params->tx_cq_moderation, &cparam->xdp_sq.cqp,
+				     &ccp, &c->rq_xdpsq.cq) : 0;
+	if (err)
+		goto err_close_rx_cq;
+
+	spin_lock_init(&c->async_icosq_lock);
+
+	err = mlx5e_open_icosq(c, params, &cparam->async_icosq, &c->async_icosq,
+			       mlx5e_async_icosq_err_cqe_work);
+	if (err)
+		goto err_close_xdpsq_cq;
+
+	mutex_init(&c->icosq_recovery_lock);
+
+	err = mlx5e_open_icosq(c, params, &cparam->icosq, &c->icosq,
+			       mlx5e_icosq_err_cqe_work);
+	if (err)
+		goto err_close_async_icosq;
+
+	err = mlx5e_open_sqs(c, params, cparam);
+	if (err)
+		goto err_close_icosq;
+
+	err = mlx5e_open_rxq_rq(c, params, &cparam->rq);
+	if (err)
+		goto err_close_sqs;
+
+	if (c->xdp) {
+		err = mlx5e_open_xdpsq(c, params, &cparam->xdp_sq, NULL,
+				       &c->rq_xdpsq, false);
+		if (err)
+			goto err_close_rq;
+	}
+
+	err = mlx5e_open_xdpsq(c, params, &cparam->xdp_sq, NULL, &c->xdpsq, true);
+	if (err)
+		goto err_close_xdp_sq;
+
+	return 0;
+
+err_close_xdp_sq:
+	if (c->xdp)
+		mlx5e_close_xdpsq(&c->rq_xdpsq);
+
+err_close_rq:
+	mlx5e_close_rq(&c->rq);
+
+err_close_sqs:
+	mlx5e_close_sqs(c);
+
+err_close_icosq:
+	mlx5e_close_icosq(&c->icosq);
+
+err_close_async_icosq:
+	mlx5e_close_icosq(&c->async_icosq);
+
+err_close_xdpsq_cq:
+	if (c->xdp)
+		mlx5e_close_cq(&c->rq_xdpsq.cq);
+
+err_close_rx_cq:
+	mlx5e_close_cq(&c->rq.cq);
+
+err_close_xdp_tx_cqs:
+	mlx5e_close_cq(&c->xdpsq.cq);
+
+err_close_tx_cqs:
+	mlx5e_close_tx_cqs(c);
+
+err_close_icosq_cq:
+	mlx5e_close_cq(&c->icosq.cq);
+
+err_close_async_icosq_cq:
+	mlx5e_close_cq(&c->async_icosq.cq);
+
+	return err;
+}
+
+static void mlx5e_close_queues(struct mlx5e_channel *c)
+{
+	mlx5e_close_xdpsq(&c->xdpsq);
+	if (c->xdp)
+		mlx5e_close_xdpsq(&c->rq_xdpsq);
+	/* The same ICOSQ is used for UMRs for both RQ and XSKRQ. */
+	cancel_work_sync(&c->icosq.recover_work);
+	mlx5e_close_rq(&c->rq);
+	mlx5e_close_sqs(c);
+	mlx5e_close_icosq(&c->icosq);
+	mutex_destroy(&c->icosq_recovery_lock);
+	mlx5e_close_icosq(&c->async_icosq);
+	if (c->xdp)
+		mlx5e_close_cq(&c->rq_xdpsq.cq);
+	mlx5e_close_cq(&c->rq.cq);
+	mlx5e_close_cq(&c->xdpsq.cq);
+	mlx5e_close_tx_cqs(c);
+	mlx5e_close_cq(&c->icosq.cq);
+	mlx5e_close_cq(&c->async_icosq.cq);
+}
+
+static u8 mlx5e_enumerate_lag_port(struct mlx5_core_dev *mdev, int ix)
+{
+	u16 port_aff_bias = mlx5_core_is_pf(mdev) ? 0 : MLX5_CAP_GEN(mdev, vhca_id);
+
+	return (ix + port_aff_bias) % mlx5e_get_num_lag_ports(mdev);
+}
+
+static int mlx5e_channel_stats_alloc(struct mlx5e_priv *priv, int ix, int cpu)
+{
+	if (ix > priv->stats_nch)  {
+		netdev_warn(priv->netdev, "Unexpected channel stats index %d > %d\n", ix,
+			    priv->stats_nch);
+		return -EINVAL;
+	}
+
+	if (priv->channel_stats[ix])
+		return 0;
+
+	/* Asymmetric dynamic memory allocation.
+	 * Freed in mlx5e_priv_arrays_free, not on channel closure.
+	 */
+	mlx5e_dbg(DRV, priv, "Creating channel stats %d\n", ix);
+	priv->channel_stats[ix] = kvzalloc_node(sizeof(**priv->channel_stats),
+						GFP_KERNEL, cpu_to_node(cpu));
+	if (!priv->channel_stats[ix])
+		return -ENOMEM;
+	priv->stats_nch++;
+
+	return 0;
+}
+
+void mlx5e_trigger_napi_icosq(struct mlx5e_channel *c)
+{
+	spin_lock_bh(&c->async_icosq_lock);
+	mlx5e_trigger_irq(&c->async_icosq);
+	spin_unlock_bh(&c->async_icosq_lock);
+}
+
+void mlx5e_trigger_napi_sched(struct napi_struct *napi)
+{
+	local_bh_disable();
+	napi_schedule(napi);
+	local_bh_enable();
+}
+
+static int mlx5e_open_channel(struct mlx5e_priv *priv, int ix,
+			      struct mlx5e_params *params,
+			      struct mlx5e_channel_param *cparam,
+			      struct xsk_buff_pool *xsk_pool,
+			      struct mlx5e_channel **cp)
+{
+	int cpu = cpumask_first(mlx5_comp_irq_get_affinity_mask(priv->mdev, ix));
+	struct net_device *netdev = priv->netdev;
+	struct mlx5e_xsk_param xsk;
+	struct mlx5e_channel *c;
+	unsigned int irq;
+	int err;
+
+	err = mlx5_vector2irqn(priv->mdev, ix, &irq);
+	if (err)
+		return err;
+
+	err = mlx5e_channel_stats_alloc(priv, ix, cpu);
+	if (err)
+		return err;
+
+	c = kvzalloc_node(sizeof(*c), GFP_KERNEL, cpu_to_node(cpu));
+	if (!c)
+		return -ENOMEM;
+
+	c->priv     = priv;
+	c->mdev     = priv->mdev;
+	c->tstamp   = &priv->tstamp;
+	c->ix       = ix;
+	c->cpu      = cpu;
+	c->pdev     = mlx5_core_dma_dev(priv->mdev);
+	c->netdev   = priv->netdev;
+	c->mkey_be  = cpu_to_be32(priv->mdev->mlx5e_res.hw_objs.mkey);
+	c->num_tc   = mlx5e_get_dcb_num_tc(params);
+	c->xdp      = !!params->xdp_prog;
+	c->stats    = &priv->channel_stats[ix]->ch;
+	c->aff_mask = irq_get_effective_affinity_mask(irq);
+	c->lag_port = mlx5e_enumerate_lag_port(priv->mdev, ix);
+
+	netif_napi_add(netdev, &c->napi, mlx5e_napi_poll);
+
+	err = mlx5e_open_queues(c, params, cparam);
+	if (unlikely(err))
+		goto err_napi_del;
+
+	if (xsk_pool) {
+		mlx5e_build_xsk_param(xsk_pool, &xsk);
+		err = mlx5e_open_xsk(priv, params, &xsk, xsk_pool, c);
+		if (unlikely(err))
+			goto err_close_queues;
+	}
+
+	*cp = c;
+
+	return 0;
+
+err_close_queues:
+	mlx5e_close_queues(c);
+
+err_napi_del:
+	netif_napi_del(&c->napi);
+
+	kvfree(c);
+
+	return err;
+}
+
+static void mlx5e_activate_channel(struct mlx5e_channel *c)
+{
+	int tc;
+
+	napi_enable(&c->napi);
+
+	for (tc = 0; tc < c->num_tc; tc++)
+		mlx5e_activate_txqsq(&c->sq[tc]);
+	mlx5e_activate_icosq(&c->icosq);
+	mlx5e_activate_icosq(&c->async_icosq);
+
+	if (test_bit(MLX5E_CHANNEL_STATE_XSK, c->state))
+		mlx5e_activate_xsk(c);
+	else
+		mlx5e_activate_rq(&c->rq);
+
+	mlx5e_trigger_napi_icosq(c);
+}
+
+static void mlx5e_deactivate_channel(struct mlx5e_channel *c)
+{
+	int tc;
+
+	if (test_bit(MLX5E_CHANNEL_STATE_XSK, c->state))
+		mlx5e_deactivate_xsk(c);
+	else
+		mlx5e_deactivate_rq(&c->rq);
+
+	mlx5e_deactivate_icosq(&c->async_icosq);
+	mlx5e_deactivate_icosq(&c->icosq);
+	for (tc = 0; tc < c->num_tc; tc++)
+		mlx5e_deactivate_txqsq(&c->sq[tc]);
+	mlx5e_qos_deactivate_queues(c);
+
+	napi_disable(&c->napi);
+}
+
+static void mlx5e_close_channel(struct mlx5e_channel *c)
+{
+	if (test_bit(MLX5E_CHANNEL_STATE_XSK, c->state))
+		mlx5e_close_xsk(c);
+	mlx5e_close_queues(c);
+	mlx5e_qos_close_queues(c);
+	netif_napi_del(&c->napi);
+
+	kvfree(c);
+}
+
+int mlx5e_open_channels(struct mlx5e_priv *priv,
+			struct mlx5e_channels *chs)
+{
+	struct mlx5e_channel_param *cparam;
+	int err = -ENOMEM;
+	int i;
+
+	chs->num = chs->params.num_channels;
+
+	chs->c = kcalloc(chs->num, sizeof(struct mlx5e_channel *), GFP_KERNEL);
+	cparam = kvzalloc(sizeof(struct mlx5e_channel_param), GFP_KERNEL);
+	if (!chs->c || !cparam)
+		goto err_free;
+
+	err = mlx5e_build_channel_param(priv->mdev, &chs->params, priv->q_counter, cparam);
+	if (err)
+		goto err_free;
+
+	for (i = 0; i < chs->num; i++) {
+		struct xsk_buff_pool *xsk_pool = NULL;
+
+		if (chs->params.xdp_prog)
+			xsk_pool = mlx5e_xsk_get_pool(&chs->params, chs->params.xsk, i);
+
+		err = mlx5e_open_channel(priv, i, &chs->params, cparam, xsk_pool, &chs->c[i]);
+		if (err)
+			goto err_close_channels;
+	}
+
+	if (MLX5E_GET_PFLAG(&chs->params, MLX5E_PFLAG_TX_PORT_TS) || chs->params.ptp_rx) {
+		err = mlx5e_ptp_open(priv, &chs->params, chs->c[0]->lag_port, &chs->ptp);
+		if (err)
+			goto err_close_channels;
+	}
+
+	if (priv->htb) {
+		err = mlx5e_qos_open_queues(priv, chs);
+		if (err)
+			goto err_close_ptp;
+	}
+
+	mlx5e_health_channels_update(priv);
+	kvfree(cparam);
+	return 0;
+
+err_close_ptp:
+	if (chs->ptp)
+		mlx5e_ptp_close(chs->ptp);
+
+err_close_channels:
+	for (i--; i >= 0; i--)
+		mlx5e_close_channel(chs->c[i]);
+
+err_free:
+	kfree(chs->c);
+	kvfree(cparam);
+	chs->num = 0;
+	return err;
+}
+
+static void mlx5e_activate_channels(struct mlx5e_channels *chs)
+{
+	int i;
+
+	for (i = 0; i < chs->num; i++)
+		mlx5e_activate_channel(chs->c[i]);
+
+	if (chs->ptp)
+		mlx5e_ptp_activate_channel(chs->ptp);
+}
+
+static int mlx5e_wait_channels_min_rx_wqes(struct mlx5e_channels *chs)
+{
+	int err = 0;
+	int i;
+
+	for (i = 0; i < chs->num; i++) {
+		int timeout = err ? 0 : MLX5E_RQ_WQES_TIMEOUT;
+		struct mlx5e_channel *c = chs->c[i];
+
+		if (test_bit(MLX5E_CHANNEL_STATE_XSK, c->state))
+			continue;
+
+		err |= mlx5e_wait_for_min_rx_wqes(&c->rq, timeout);
+
+		/* Don't wait on the XSK RQ, because the newer xdpsock sample
+		 * doesn't provide any Fill Ring entries at the setup stage.
+		 */
+	}
+
+	return err ? -ETIMEDOUT : 0;
+}
+
+static void mlx5e_deactivate_channels(struct mlx5e_channels *chs)
+{
+	int i;
+
+	if (chs->ptp)
+		mlx5e_ptp_deactivate_channel(chs->ptp);
+
+	for (i = 0; i < chs->num; i++)
+		mlx5e_deactivate_channel(chs->c[i]);
+}
+
+void mlx5e_close_channels(struct mlx5e_channels *chs)
+{
+	int i;
+
+	ASSERT_RTNL();
+	if (chs->ptp) {
+		mlx5e_ptp_close(chs->ptp);
+		chs->ptp = NULL;
+	}
+	for (i = 0; i < chs->num; i++)
+		mlx5e_close_channel(chs->c[i]);
+
+	kfree(chs->c);
+	chs->num = 0;
+}
+
+static int mlx5e_modify_tirs_packet_merge(struct mlx5e_priv *priv)
+{
+	struct mlx5e_rx_res *res = priv->rx_res;
+
+	return mlx5e_rx_res_packet_merge_set_param(res, &priv->channels.params.packet_merge);
+}
+
+static MLX5E_DEFINE_PREACTIVATE_WRAPPER_CTX(mlx5e_modify_tirs_packet_merge);
+
+static int mlx5e_set_mtu(struct mlx5_core_dev *mdev,
+			 struct mlx5e_params *params, u16 mtu)
+{
+	u16 hw_mtu = MLX5E_SW2HW_MTU(params, mtu);
+	int err;
+
+	err = mlx5_set_port_mtu(mdev, hw_mtu, 1);
+	if (err)
+		return err;
+
+	/* Update vport context MTU */
+	mlx5_modify_nic_vport_mtu(mdev, hw_mtu);
+	return 0;
+}
+
+static void mlx5e_query_mtu(struct mlx5_core_dev *mdev,
+			    struct mlx5e_params *params, u16 *mtu)
+{
+	u16 hw_mtu = 0;
+	int err;
+
+	err = mlx5_query_nic_vport_mtu(mdev, &hw_mtu);
+	if (err || !hw_mtu) /* fallback to port oper mtu */
+		mlx5_query_port_oper_mtu(mdev, &hw_mtu, 1);
+
+	*mtu = MLX5E_HW2SW_MTU(params, hw_mtu);
+}
+
+int mlx5e_set_dev_port_mtu(struct mlx5e_priv *priv)
+{
+	struct mlx5e_params *params = &priv->channels.params;
+	struct net_device *netdev = priv->netdev;
+	struct mlx5_core_dev *mdev = priv->mdev;
+	u16 mtu;
+	int err;
+
+	err = mlx5e_set_mtu(mdev, params, params->sw_mtu);
+	if (err)
+		return err;
+
+	mlx5e_query_mtu(mdev, params, &mtu);
+	if (mtu != params->sw_mtu)
+		netdev_warn(netdev, "%s: VPort MTU %d is different than netdev mtu %d\n",
+			    __func__, mtu, params->sw_mtu);
+
+	params->sw_mtu = mtu;
+	return 0;
+}
+
+MLX5E_DEFINE_PREACTIVATE_WRAPPER_CTX(mlx5e_set_dev_port_mtu);
+
+void mlx5e_set_netdev_mtu_boundaries(struct mlx5e_priv *priv)
+{
+	struct mlx5e_params *params = &priv->channels.params;
+	struct net_device *netdev   = priv->netdev;
+	struct mlx5_core_dev *mdev  = priv->mdev;
+	u16 max_mtu;
+
+	/* MTU range: 68 - hw-specific max */
+	netdev->min_mtu = ETH_MIN_MTU;
+
+	mlx5_query_port_max_mtu(mdev, &max_mtu, 1);
+	netdev->max_mtu = min_t(unsigned int, MLX5E_HW2SW_MTU(params, max_mtu),
+				ETH_MAX_MTU);
+}
+
+static int mlx5e_netdev_set_tcs(struct net_device *netdev, u16 nch, u8 ntc,
+				struct netdev_tc_txq *tc_to_txq)
+{
+	int tc, err;
+
+	netdev_reset_tc(netdev);
+
+	if (ntc == 1)
+		return 0;
+
+	err = netdev_set_num_tc(netdev, ntc);
+	if (err) {
+		netdev_WARN(netdev, "netdev_set_num_tc failed (%d), ntc = %d\n", err, ntc);
+		return err;
+	}
+
+	for (tc = 0; tc < ntc; tc++) {
+		u16 count, offset;
+
+		count = tc_to_txq[tc].count;
+		offset = tc_to_txq[tc].offset;
+		netdev_set_tc_queue(netdev, tc, count, offset);
+	}
+
+	return 0;
+}
+
+int mlx5e_update_tx_netdev_queues(struct mlx5e_priv *priv)
+{
+	int nch, ntc, num_txqs, err;
+	int qos_queues = 0;
+
+	if (priv->htb)
+		qos_queues = mlx5e_htb_cur_leaf_nodes(priv->htb);
+
+	nch = priv->channels.params.num_channels;
+	ntc = mlx5e_get_dcb_num_tc(&priv->channels.params);
+	num_txqs = nch * ntc + qos_queues;
+	if (MLX5E_GET_PFLAG(&priv->channels.params, MLX5E_PFLAG_TX_PORT_TS))
+		num_txqs += ntc;
+
+	mlx5e_dbg(DRV, priv, "Setting num_txqs %d\n", num_txqs);
+	err = netif_set_real_num_tx_queues(priv->netdev, num_txqs);
+	if (err)
+		netdev_warn(priv->netdev, "netif_set_real_num_tx_queues failed, %d\n", err);
+
+	return err;
+}
+
+static int mlx5e_update_netdev_queues(struct mlx5e_priv *priv)
+{
+	struct netdev_tc_txq old_tc_to_txq[TC_MAX_QUEUE], *tc_to_txq;
+	struct net_device *netdev = priv->netdev;
+	int old_num_txqs, old_ntc;
+	int nch, ntc;
+	int err;
+	int i;
+
+	old_num_txqs = netdev->real_num_tx_queues;
+	old_ntc = netdev->num_tc ? : 1;
+	for (i = 0; i < ARRAY_SIZE(old_tc_to_txq); i++)
+		old_tc_to_txq[i] = netdev->tc_to_txq[i];
+
+	nch = priv->channels.params.num_channels;
+	ntc = priv->channels.params.mqprio.num_tc;
+	tc_to_txq = priv->channels.params.mqprio.tc_to_txq;
+
+	err = mlx5e_netdev_set_tcs(netdev, nch, ntc, tc_to_txq);
+	if (err)
+		goto err_out;
+	err = mlx5e_update_tx_netdev_queues(priv);
+	if (err)
+		goto err_tcs;
+	err = netif_set_real_num_rx_queues(netdev, nch);
+	if (err) {
+		netdev_warn(netdev, "netif_set_real_num_rx_queues failed, %d\n", err);
+		goto err_txqs;
+	}
+
+	return 0;
+
+err_txqs:
+	/* netif_set_real_num_rx_queues could fail only when nch increased. Only
+	 * one of nch and ntc is changed in this function. That means, the call
+	 * to netif_set_real_num_tx_queues below should not fail, because it
+	 * decreases the number of TX queues.
+	 */
+	WARN_ON_ONCE(netif_set_real_num_tx_queues(netdev, old_num_txqs));
+
+err_tcs:
+	WARN_ON_ONCE(mlx5e_netdev_set_tcs(netdev, old_num_txqs / old_ntc, old_ntc,
+					  old_tc_to_txq));
+err_out:
+	return err;
+}
+
+static MLX5E_DEFINE_PREACTIVATE_WRAPPER_CTX(mlx5e_update_netdev_queues);
+
+static void mlx5e_set_default_xps_cpumasks(struct mlx5e_priv *priv,
+					   struct mlx5e_params *params)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+	int num_comp_vectors, ix, irq;
+
+	num_comp_vectors = mlx5_comp_vectors_count(mdev);
+
+	for (ix = 0; ix < params->num_channels; ix++) {
+		cpumask_clear(priv->scratchpad.cpumask);
+
+		for (irq = ix; irq < num_comp_vectors; irq += params->num_channels) {
+			int cpu = cpumask_first(mlx5_comp_irq_get_affinity_mask(mdev, irq));
+
+			cpumask_set_cpu(cpu, priv->scratchpad.cpumask);
+		}
+
+		netif_set_xps_queue(priv->netdev, priv->scratchpad.cpumask, ix);
+	}
+}
+
+static int mlx5e_num_channels_changed(struct mlx5e_priv *priv)
+{
+	u16 count = priv->channels.params.num_channels;
+	int err;
+
+	err = mlx5e_update_netdev_queues(priv);
+	if (err)
+		return err;
+
+	mlx5e_set_default_xps_cpumasks(priv, &priv->channels.params);
+
+	/* This function may be called on attach, before priv->rx_res is created. */
+	if (!netif_is_rxfh_configured(priv->netdev) && priv->rx_res)
+		mlx5e_rx_res_rss_set_indir_uniform(priv->rx_res, count);
+
+	return 0;
+}
+
+MLX5E_DEFINE_PREACTIVATE_WRAPPER_CTX(mlx5e_num_channels_changed);
+
+static void mlx5e_build_txq_maps(struct mlx5e_priv *priv)
+{
+	int i, ch, tc, num_tc;
+
+	ch = priv->channels.num;
+	num_tc = mlx5e_get_dcb_num_tc(&priv->channels.params);
+
+	for (i = 0; i < ch; i++) {
+		for (tc = 0; tc < num_tc; tc++) {
+			struct mlx5e_channel *c = priv->channels.c[i];
+			struct mlx5e_txqsq *sq = &c->sq[tc];
+
+			priv->txq2sq[sq->txq_ix] = sq;
+		}
+	}
+
+	if (!priv->channels.ptp)
+		goto out;
+
+	if (!test_bit(MLX5E_PTP_STATE_TX, priv->channels.ptp->state))
+		goto out;
+
+	for (tc = 0; tc < num_tc; tc++) {
+		struct mlx5e_ptp *c = priv->channels.ptp;
+		struct mlx5e_txqsq *sq = &c->ptpsq[tc].txqsq;
+
+		priv->txq2sq[sq->txq_ix] = sq;
+	}
+
+out:
+	/* Make the change to txq2sq visible before the queue is started.
+	 * As mlx5e_xmit runs under a spinlock, there is an implicit ACQUIRE,
+	 * which pairs with this barrier.
+	 */
+	smp_wmb();
+}
+
+void mlx5e_activate_priv_channels(struct mlx5e_priv *priv)
+{
+	mlx5e_build_txq_maps(priv);
+	mlx5e_activate_channels(&priv->channels);
+	if (priv->htb)
+		mlx5e_qos_activate_queues(priv);
+	mlx5e_xdp_tx_enable(priv);
+
+	/* dev_watchdog() wants all TX queues to be started when the carrier is
+	 * OK, including the ones in range real_num_tx_queues..num_tx_queues-1.
+	 * Make it happy to avoid TX timeout false alarms.
+	 */
+	netif_tx_start_all_queues(priv->netdev);
+
+	if (mlx5e_is_vport_rep(priv))
+		mlx5e_rep_activate_channels(priv);
+
+	set_bit(MLX5E_STATE_CHANNELS_ACTIVE, &priv->state);
+
+	mlx5e_wait_channels_min_rx_wqes(&priv->channels);
+
+	if (priv->rx_res)
+		mlx5e_rx_res_channels_activate(priv->rx_res, &priv->channels);
+}
+
+static void mlx5e_cancel_tx_timeout_work(struct mlx5e_priv *priv)
+{
+	WARN_ON_ONCE(test_bit(MLX5E_STATE_CHANNELS_ACTIVE, &priv->state));
+	if (current_work() != &priv->tx_timeout_work)
+		cancel_work_sync(&priv->tx_timeout_work);
+}
+
+void mlx5e_deactivate_priv_channels(struct mlx5e_priv *priv)
+{
+	if (priv->rx_res)
+		mlx5e_rx_res_channels_deactivate(priv->rx_res);
+
+	clear_bit(MLX5E_STATE_CHANNELS_ACTIVE, &priv->state);
+	mlx5e_cancel_tx_timeout_work(priv);
+
+	if (mlx5e_is_vport_rep(priv))
+		mlx5e_rep_deactivate_channels(priv);
+
+	/* The results of ndo_select_queue are unreliable, while netdev config
+	 * is being changed (real_num_tx_queues, num_tc). Stop all queues to
+	 * prevent ndo_start_xmit from being called, so that it can assume that
+	 * the selected queue is always valid.
+	 */
+	netif_tx_disable(priv->netdev);
+
+	mlx5e_xdp_tx_disable(priv);
+	mlx5e_deactivate_channels(&priv->channels);
+}
+
+static int mlx5e_switch_priv_params(struct mlx5e_priv *priv,
+				    struct mlx5e_params *new_params,
+				    mlx5e_fp_preactivate preactivate,
+				    void *context)
+{
+	struct mlx5e_params old_params;
+
+	old_params = priv->channels.params;
+	priv->channels.params = *new_params;
+
+	if (preactivate) {
+		int err;
+
+		err = preactivate(priv, context);
+		if (err) {
+			priv->channels.params = old_params;
+			return err;
+		}
+	}
+
+	return 0;
+}
+
+static int mlx5e_switch_priv_channels(struct mlx5e_priv *priv,
+				      struct mlx5e_channels *new_chs,
+				      mlx5e_fp_preactivate preactivate,
+				      void *context)
+{
+	struct net_device *netdev = priv->netdev;
+	struct mlx5e_channels old_chs;
+	int carrier_ok;
+	int err = 0;
+
+	carrier_ok = netif_carrier_ok(netdev);
+	netif_carrier_off(netdev);
+
+	mlx5e_deactivate_priv_channels(priv);
+
+	old_chs = priv->channels;
+	priv->channels = *new_chs;
+
+	/* New channels are ready to roll, call the preactivate hook if needed
+	 * to modify HW settings or update kernel parameters.
+	 */
+	if (preactivate) {
+		err = preactivate(priv, context);
+		if (err) {
+			priv->channels = old_chs;
+			goto out;
+		}
+	}
+
+	mlx5e_close_channels(&old_chs);
+	priv->profile->update_rx(priv);
+
+	mlx5e_selq_apply(&priv->selq);
+out:
+	mlx5e_activate_priv_channels(priv);
+
+	/* return carrier back if needed */
+	if (carrier_ok)
+		netif_carrier_on(netdev);
+
+	return err;
+}
+
+int mlx5e_safe_switch_params(struct mlx5e_priv *priv,
+			     struct mlx5e_params *params,
+			     mlx5e_fp_preactivate preactivate,
+			     void *context, bool reset)
+{
+	struct mlx5e_channels new_chs = {};
+	int err;
+
+	reset &= test_bit(MLX5E_STATE_OPENED, &priv->state);
+	if (!reset)
+		return mlx5e_switch_priv_params(priv, params, preactivate, context);
+
+	new_chs.params = *params;
+
+	mlx5e_selq_prepare_params(&priv->selq, &new_chs.params);
+
+	err = mlx5e_open_channels(priv, &new_chs);
+	if (err)
+		goto err_cancel_selq;
+
+	err = mlx5e_switch_priv_channels(priv, &new_chs, preactivate, context);
+	if (err)
+		goto err_close;
+
+	return 0;
+
+err_close:
+	mlx5e_close_channels(&new_chs);
+
+err_cancel_selq:
+	mlx5e_selq_cancel(&priv->selq);
+	return err;
+}
+
+int mlx5e_safe_reopen_channels(struct mlx5e_priv *priv)
+{
+	return mlx5e_safe_switch_params(priv, &priv->channels.params, NULL, NULL, true);
+}
+
+void mlx5e_timestamp_init(struct mlx5e_priv *priv)
+{
+	priv->tstamp.tx_type   = HWTSTAMP_TX_OFF;
+	priv->tstamp.rx_filter = HWTSTAMP_FILTER_NONE;
+}
+
+static void mlx5e_modify_admin_state(struct mlx5_core_dev *mdev,
+				     enum mlx5_port_status state)
+{
+	struct mlx5_eswitch *esw = mdev->priv.eswitch;
+	int vport_admin_state;
+
+	mlx5_set_port_admin_status(mdev, state);
+
+	if (mlx5_eswitch_mode(mdev) == MLX5_ESWITCH_OFFLOADS ||
+	    !MLX5_CAP_GEN(mdev, uplink_follow))
+		return;
+
+	if (state == MLX5_PORT_UP)
+		vport_admin_state = MLX5_VPORT_ADMIN_STATE_AUTO;
+	else
+		vport_admin_state = MLX5_VPORT_ADMIN_STATE_DOWN;
+
+	mlx5_eswitch_set_vport_state(esw, MLX5_VPORT_UPLINK, vport_admin_state);
+}
+
+int mlx5e_open_locked(struct net_device *netdev)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	int err;
+
+	mlx5e_selq_prepare_params(&priv->selq, &priv->channels.params);
+
+	set_bit(MLX5E_STATE_OPENED, &priv->state);
+
+	err = mlx5e_open_channels(priv, &priv->channels);
+	if (err)
+		goto err_clear_state_opened_flag;
+
+	priv->profile->update_rx(priv);
+	mlx5e_selq_apply(&priv->selq);
+	mlx5e_activate_priv_channels(priv);
+	mlx5e_apply_traps(priv, true);
+	if (priv->profile->update_carrier)
+		priv->profile->update_carrier(priv);
+
+	mlx5e_queue_update_stats(priv);
+	return 0;
+
+err_clear_state_opened_flag:
+	clear_bit(MLX5E_STATE_OPENED, &priv->state);
+	mlx5e_selq_cancel(&priv->selq);
+	return err;
+}
+
+int mlx5e_open(struct net_device *netdev)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	int err;
+
+	mutex_lock(&priv->state_lock);
+	err = mlx5e_open_locked(netdev);
+	if (!err)
+		mlx5e_modify_admin_state(priv->mdev, MLX5_PORT_UP);
+	mutex_unlock(&priv->state_lock);
+
+	return err;
+}
+
+int mlx5e_close_locked(struct net_device *netdev)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+
+	/* May already be CLOSED in case a previous configuration operation
+	 * (e.g RX/TX queue size change) that involves close&open failed.
+	 */
+	if (!test_bit(MLX5E_STATE_OPENED, &priv->state))
+		return 0;
+
+	mlx5e_apply_traps(priv, false);
+	clear_bit(MLX5E_STATE_OPENED, &priv->state);
+
+	netif_carrier_off(priv->netdev);
+	mlx5e_deactivate_priv_channels(priv);
+	mlx5e_close_channels(&priv->channels);
+
+	return 0;
+}
+
+int mlx5e_close(struct net_device *netdev)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	int err;
+
+	if (!netif_device_present(netdev))
+		return -ENODEV;
+
+	mutex_lock(&priv->state_lock);
+	mlx5e_modify_admin_state(priv->mdev, MLX5_PORT_DOWN);
+	err = mlx5e_close_locked(netdev);
+	mutex_unlock(&priv->state_lock);
+
+	return err;
+}
+
+static void mlx5e_free_drop_rq(struct mlx5e_rq *rq)
+{
+	mlx5_wq_destroy(&rq->wq_ctrl);
+}
+
+static int mlx5e_alloc_drop_rq(struct mlx5_core_dev *mdev,
+			       struct mlx5e_rq *rq,
+			       struct mlx5e_rq_param *param)
+{
+	void *rqc = param->rqc;
+	void *rqc_wq = MLX5_ADDR_OF(rqc, rqc, wq);
+	int err;
+
+	param->wq.db_numa_node = param->wq.buf_numa_node;
+
+	err = mlx5_wq_cyc_create(mdev, &param->wq, rqc_wq, &rq->wqe.wq,
+				 &rq->wq_ctrl);
+	if (err)
+		return err;
+
+	/* Mark as unused given "Drop-RQ" packets never reach XDP */
+	xdp_rxq_info_unused(&rq->xdp_rxq);
+
+	rq->mdev = mdev;
+
+	return 0;
+}
+
+static int mlx5e_alloc_drop_cq(struct mlx5e_priv *priv,
+			       struct mlx5e_cq *cq,
+			       struct mlx5e_cq_param *param)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+
+	param->wq.buf_numa_node = dev_to_node(mlx5_core_dma_dev(mdev));
+	param->wq.db_numa_node  = dev_to_node(mlx5_core_dma_dev(mdev));
+
+	return mlx5e_alloc_cq_common(priv, param, cq);
+}
+
+int mlx5e_open_drop_rq(struct mlx5e_priv *priv,
+		       struct mlx5e_rq *drop_rq)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+	struct mlx5e_cq_param cq_param = {};
+	struct mlx5e_rq_param rq_param = {};
+	struct mlx5e_cq *cq = &drop_rq->cq;
+	int err;
+
+	mlx5e_build_drop_rq_param(mdev, priv->drop_rq_q_counter, &rq_param);
+
+	err = mlx5e_alloc_drop_cq(priv, cq, &cq_param);
+	if (err)
+		return err;
+
+	err = mlx5e_create_cq(cq, &cq_param);
+	if (err)
+		goto err_free_cq;
+
+	err = mlx5e_alloc_drop_rq(mdev, drop_rq, &rq_param);
+	if (err)
+		goto err_destroy_cq;
+
+	err = mlx5e_create_rq(drop_rq, &rq_param);
+	if (err)
+		goto err_free_rq;
+
+	err = mlx5e_modify_rq_state(drop_rq, MLX5_RQC_STATE_RST, MLX5_RQC_STATE_RDY);
+	if (err)
+		mlx5_core_warn(priv->mdev, "modify_rq_state failed, rx_if_down_packets won't be counted %d\n", err);
+
+	return 0;
+
+err_free_rq:
+	mlx5e_free_drop_rq(drop_rq);
+
+err_destroy_cq:
+	mlx5e_destroy_cq(cq);
+
+err_free_cq:
+	mlx5e_free_cq(cq);
+
+	return err;
+}
+
+void mlx5e_close_drop_rq(struct mlx5e_rq *drop_rq)
+{
+	mlx5e_destroy_rq(drop_rq);
+	mlx5e_free_drop_rq(drop_rq);
+	mlx5e_destroy_cq(&drop_rq->cq);
+	mlx5e_free_cq(&drop_rq->cq);
+}
+
+int mlx5e_create_tis(struct mlx5_core_dev *mdev, void *in, u32 *tisn)
+{
+	void *tisc = MLX5_ADDR_OF(create_tis_in, in, ctx);
+
+	MLX5_SET(tisc, tisc, transport_domain, mdev->mlx5e_res.hw_objs.td.tdn);
+
+	if (MLX5_GET(tisc, tisc, tls_en))
+		MLX5_SET(tisc, tisc, pd, mdev->mlx5e_res.hw_objs.pdn);
+
+	if (mlx5_lag_is_lacp_owner(mdev))
+		MLX5_SET(tisc, tisc, strict_lag_tx_port_affinity, 1);
+
+	return mlx5_core_create_tis(mdev, in, tisn);
+}
+
+void mlx5e_destroy_tis(struct mlx5_core_dev *mdev, u32 tisn)
+{
+	mlx5_core_destroy_tis(mdev, tisn);
+}
+
+void mlx5e_destroy_tises(struct mlx5e_priv *priv)
+{
+	int tc, i;
+
+	for (i = 0; i < mlx5e_get_num_lag_ports(priv->mdev); i++)
+		for (tc = 0; tc < priv->profile->max_tc; tc++)
+			mlx5e_destroy_tis(priv->mdev, priv->tisn[i][tc]);
+}
+
+static bool mlx5e_lag_should_assign_affinity(struct mlx5_core_dev *mdev)
+{
+	return MLX5_CAP_GEN(mdev, lag_tx_port_affinity) && mlx5e_get_num_lag_ports(mdev) > 1;
+}
+
+int mlx5e_create_tises(struct mlx5e_priv *priv)
+{
+	int tc, i;
+	int err;
+
+	for (i = 0; i < mlx5e_get_num_lag_ports(priv->mdev); i++) {
+		for (tc = 0; tc < priv->profile->max_tc; tc++) {
+			u32 in[MLX5_ST_SZ_DW(create_tis_in)] = {};
+			void *tisc;
+
+			tisc = MLX5_ADDR_OF(create_tis_in, in, ctx);
+
+			MLX5_SET(tisc, tisc, prio, tc << 1);
+
+			if (mlx5e_lag_should_assign_affinity(priv->mdev))
+				MLX5_SET(tisc, tisc, lag_tx_port_affinity, i + 1);
+
+			err = mlx5e_create_tis(priv->mdev, in, &priv->tisn[i][tc]);
+			if (err)
+				goto err_close_tises;
+		}
+	}
+
+	return 0;
+
+err_close_tises:
+	for (; i >= 0; i--) {
+		for (tc--; tc >= 0; tc--)
+			mlx5e_destroy_tis(priv->mdev, priv->tisn[i][tc]);
+		tc = priv->profile->max_tc;
+	}
+
+	return err;
+}
+
+static void mlx5e_cleanup_nic_tx(struct mlx5e_priv *priv)
+{
+	if (priv->mqprio_rl) {
+		mlx5e_mqprio_rl_cleanup(priv->mqprio_rl);
+		mlx5e_mqprio_rl_free(priv->mqprio_rl);
+		priv->mqprio_rl = NULL;
+	}
+	mlx5e_accel_cleanup_tx(priv);
+	mlx5e_destroy_tises(priv);
+}
+
+static int mlx5e_modify_channels_vsd(struct mlx5e_channels *chs, bool vsd)
+{
+	int err;
+	int i;
+
+	for (i = 0; i < chs->num; i++) {
+		err = mlx5e_modify_rq_vsd(&chs->c[i]->rq, vsd);
+		if (err)
+			return err;
+	}
+	if (chs->ptp && test_bit(MLX5E_PTP_STATE_RX, chs->ptp->state))
+		return mlx5e_modify_rq_vsd(&chs->ptp->rq, vsd);
+
+	return 0;
+}
+
+static void mlx5e_mqprio_build_default_tc_to_txq(struct netdev_tc_txq *tc_to_txq,
+						 int ntc, int nch)
+{
+	int tc;
+
+	memset(tc_to_txq, 0, sizeof(*tc_to_txq) * TC_MAX_QUEUE);
+
+	/* Map netdev TCs to offset 0.
+	 * We have our own UP to TXQ mapping for DCB mode of QoS
+	 */
+	for (tc = 0; tc < ntc; tc++) {
+		tc_to_txq[tc] = (struct netdev_tc_txq) {
+			.count = nch,
+			.offset = 0,
+		};
+	}
+}
+
+static void mlx5e_mqprio_build_tc_to_txq(struct netdev_tc_txq *tc_to_txq,
+					 struct tc_mqprio_qopt *qopt)
+{
+	int tc;
+
+	for (tc = 0; tc < TC_MAX_QUEUE; tc++) {
+		tc_to_txq[tc] = (struct netdev_tc_txq) {
+			.count = qopt->count[tc],
+			.offset = qopt->offset[tc],
+		};
+	}
+}
+
+static void mlx5e_params_mqprio_dcb_set(struct mlx5e_params *params, u8 num_tc)
+{
+	params->mqprio.mode = TC_MQPRIO_MODE_DCB;
+	params->mqprio.num_tc = num_tc;
+	mlx5e_mqprio_build_default_tc_to_txq(params->mqprio.tc_to_txq, num_tc,
+					     params->num_channels);
+}
+
+static void mlx5e_mqprio_rl_update_params(struct mlx5e_params *params,
+					  struct mlx5e_mqprio_rl *rl)
+{
+	int tc;
+
+	for (tc = 0; tc < TC_MAX_QUEUE; tc++) {
+		u32 hw_id = 0;
+
+		if (rl)
+			mlx5e_mqprio_rl_get_node_hw_id(rl, tc, &hw_id);
+		params->mqprio.channel.hw_id[tc] = hw_id;
+	}
+}
+
+static void mlx5e_params_mqprio_channel_set(struct mlx5e_params *params,
+					    struct tc_mqprio_qopt_offload *mqprio,
+					    struct mlx5e_mqprio_rl *rl)
+{
+	int tc;
+
+	params->mqprio.mode = TC_MQPRIO_MODE_CHANNEL;
+	params->mqprio.num_tc = mqprio->qopt.num_tc;
+
+	for (tc = 0; tc < TC_MAX_QUEUE; tc++)
+		params->mqprio.channel.max_rate[tc] = mqprio->max_rate[tc];
+
+	mlx5e_mqprio_rl_update_params(params, rl);
+	mlx5e_mqprio_build_tc_to_txq(params->mqprio.tc_to_txq, &mqprio->qopt);
+}
+
+static void mlx5e_params_mqprio_reset(struct mlx5e_params *params)
+{
+	mlx5e_params_mqprio_dcb_set(params, 1);
+}
+
+static int mlx5e_setup_tc_mqprio_dcb(struct mlx5e_priv *priv,
+				     struct tc_mqprio_qopt *mqprio)
+{
+	struct mlx5e_params new_params;
+	u8 tc = mqprio->num_tc;
+	int err;
+
+	mqprio->hw = TC_MQPRIO_HW_OFFLOAD_TCS;
+
+	if (tc && tc != MLX5E_MAX_NUM_TC)
+		return -EINVAL;
+
+	new_params = priv->channels.params;
+	mlx5e_params_mqprio_dcb_set(&new_params, tc ? tc : 1);
+
+	err = mlx5e_safe_switch_params(priv, &new_params,
+				       mlx5e_num_channels_changed_ctx, NULL, true);
+
+	if (!err && priv->mqprio_rl) {
+		mlx5e_mqprio_rl_cleanup(priv->mqprio_rl);
+		mlx5e_mqprio_rl_free(priv->mqprio_rl);
+		priv->mqprio_rl = NULL;
+	}
+
+	priv->max_opened_tc = max_t(u8, priv->max_opened_tc,
+				    mlx5e_get_dcb_num_tc(&priv->channels.params));
+	return err;
+}
+
+static int mlx5e_mqprio_channel_validate(struct mlx5e_priv *priv,
+					 struct tc_mqprio_qopt_offload *mqprio)
+{
+	struct net_device *netdev = priv->netdev;
+	struct mlx5e_ptp *ptp_channel;
+	int agg_count = 0;
+	int i;
+
+	ptp_channel = priv->channels.ptp;
+	if (ptp_channel && test_bit(MLX5E_PTP_STATE_TX, ptp_channel->state)) {
+		netdev_err(netdev,
+			   "Cannot activate MQPRIO mode channel since it conflicts with TX port TS\n");
+		return -EINVAL;
+	}
+
+	if (mqprio->qopt.offset[0] != 0 || mqprio->qopt.num_tc < 1 ||
+	    mqprio->qopt.num_tc > MLX5E_MAX_NUM_MQPRIO_CH_TC)
+		return -EINVAL;
+
+	for (i = 0; i < mqprio->qopt.num_tc; i++) {
+		if (!mqprio->qopt.count[i]) {
+			netdev_err(netdev, "Zero size for queue-group (%d) is not supported\n", i);
+			return -EINVAL;
+		}
+		if (mqprio->min_rate[i]) {
+			netdev_err(netdev, "Min tx rate is not supported\n");
+			return -EINVAL;
+		}
+
+		if (mqprio->max_rate[i]) {
+			int err;
+
+			err = mlx5e_qos_bytes_rate_check(priv->mdev, mqprio->max_rate[i]);
+			if (err)
+				return err;
+		}
+
+		if (mqprio->qopt.offset[i] != agg_count) {
+			netdev_err(netdev, "Discontinuous queues config is not supported\n");
+			return -EINVAL;
+		}
+		agg_count += mqprio->qopt.count[i];
+	}
+
+	if (priv->channels.params.num_channels != agg_count) {
+		netdev_err(netdev, "Num of queues (%d) does not match available (%d)\n",
+			   agg_count, priv->channels.params.num_channels);
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static bool mlx5e_mqprio_rate_limit(u8 num_tc, u64 max_rate[])
+{
+	int tc;
+
+	for (tc = 0; tc < num_tc; tc++)
+		if (max_rate[tc])
+			return true;
+	return false;
+}
+
+static struct mlx5e_mqprio_rl *mlx5e_mqprio_rl_create(struct mlx5_core_dev *mdev,
+						      u8 num_tc, u64 max_rate[])
+{
+	struct mlx5e_mqprio_rl *rl;
+	int err;
+
+	if (!mlx5e_mqprio_rate_limit(num_tc, max_rate))
+		return NULL;
+
+	rl = mlx5e_mqprio_rl_alloc();
+	if (!rl)
+		return ERR_PTR(-ENOMEM);
+
+	err = mlx5e_mqprio_rl_init(rl, mdev, num_tc, max_rate);
+	if (err) {
+		mlx5e_mqprio_rl_free(rl);
+		return ERR_PTR(err);
+	}
+
+	return rl;
+}
+
+static int mlx5e_setup_tc_mqprio_channel(struct mlx5e_priv *priv,
+					 struct tc_mqprio_qopt_offload *mqprio)
+{
+	mlx5e_fp_preactivate preactivate;
+	struct mlx5e_params new_params;
+	struct mlx5e_mqprio_rl *rl;
+	bool nch_changed;
+	int err;
+
+	err = mlx5e_mqprio_channel_validate(priv, mqprio);
+	if (err)
+		return err;
+
+	rl = mlx5e_mqprio_rl_create(priv->mdev, mqprio->qopt.num_tc, mqprio->max_rate);
+	if (IS_ERR(rl))
+		return PTR_ERR(rl);
+
+	new_params = priv->channels.params;
+	mlx5e_params_mqprio_channel_set(&new_params, mqprio, rl);
+
+	nch_changed = mlx5e_get_dcb_num_tc(&priv->channels.params) > 1;
+	preactivate = nch_changed ? mlx5e_num_channels_changed_ctx :
+		mlx5e_update_netdev_queues_ctx;
+	err = mlx5e_safe_switch_params(priv, &new_params, preactivate, NULL, true);
+	if (err) {
+		if (rl) {
+			mlx5e_mqprio_rl_cleanup(rl);
+			mlx5e_mqprio_rl_free(rl);
+		}
+		return err;
+	}
+
+	if (priv->mqprio_rl) {
+		mlx5e_mqprio_rl_cleanup(priv->mqprio_rl);
+		mlx5e_mqprio_rl_free(priv->mqprio_rl);
+	}
+	priv->mqprio_rl = rl;
+
+	return 0;
+}
+
+static int mlx5e_setup_tc_mqprio(struct mlx5e_priv *priv,
+				 struct tc_mqprio_qopt_offload *mqprio)
+{
+	/* MQPRIO is another toplevel qdisc that can't be attached
+	 * simultaneously with the offloaded HTB.
+	 */
+	if (WARN_ON(mlx5e_selq_is_htb_enabled(&priv->selq)))
+		return -EINVAL;
+
+	switch (mqprio->mode) {
+	case TC_MQPRIO_MODE_DCB:
+		return mlx5e_setup_tc_mqprio_dcb(priv, &mqprio->qopt);
+	case TC_MQPRIO_MODE_CHANNEL:
+		return mlx5e_setup_tc_mqprio_channel(priv, mqprio);
+	default:
+		return -EOPNOTSUPP;
+	}
+}
+
+static LIST_HEAD(mlx5e_block_cb_list);
+
+static int mlx5e_setup_tc(struct net_device *dev, enum tc_setup_type type,
+			  void *type_data)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	bool tc_unbind = false;
+	int err;
+
+	if (type == TC_SETUP_BLOCK &&
+	    ((struct flow_block_offload *)type_data)->command == FLOW_BLOCK_UNBIND)
+		tc_unbind = true;
+
+	if (!netif_device_present(dev) && !tc_unbind)
+		return -ENODEV;
+
+	switch (type) {
+	case TC_SETUP_BLOCK: {
+		struct flow_block_offload *f = type_data;
+
+		f->unlocked_driver_cb = true;
+		return flow_block_cb_setup_simple(type_data,
+						  &mlx5e_block_cb_list,
+						  mlx5e_setup_tc_block_cb,
+						  priv, priv, true);
+	}
+	case TC_SETUP_QDISC_MQPRIO:
+		mutex_lock(&priv->state_lock);
+		err = mlx5e_setup_tc_mqprio(priv, type_data);
+		mutex_unlock(&priv->state_lock);
+		return err;
+	case TC_SETUP_QDISC_HTB:
+		mutex_lock(&priv->state_lock);
+		err = mlx5e_htb_setup_tc(priv, type_data);
+		mutex_unlock(&priv->state_lock);
+		return err;
+	default:
+		return -EOPNOTSUPP;
+	}
+}
+
+void mlx5e_fold_sw_stats64(struct mlx5e_priv *priv, struct rtnl_link_stats64 *s)
+{
+	int i;
+
+	for (i = 0; i < priv->stats_nch; i++) {
+		struct mlx5e_channel_stats *channel_stats = priv->channel_stats[i];
+		struct mlx5e_rq_stats *xskrq_stats = &channel_stats->xskrq;
+		struct mlx5e_rq_stats *rq_stats = &channel_stats->rq;
+		int j;
+
+		s->rx_packets   += rq_stats->packets + xskrq_stats->packets;
+		s->rx_bytes     += rq_stats->bytes + xskrq_stats->bytes;
+		s->multicast    += rq_stats->mcast_packets + xskrq_stats->mcast_packets;
+
+		for (j = 0; j < priv->max_opened_tc; j++) {
+			struct mlx5e_sq_stats *sq_stats = &channel_stats->sq[j];
+
+			s->tx_packets    += sq_stats->packets;
+			s->tx_bytes      += sq_stats->bytes;
+			s->tx_dropped    += sq_stats->dropped;
+		}
+	}
+	if (priv->tx_ptp_opened) {
+		for (i = 0; i < priv->max_opened_tc; i++) {
+			struct mlx5e_sq_stats *sq_stats = &priv->ptp_stats.sq[i];
+
+			s->tx_packets    += sq_stats->packets;
+			s->tx_bytes      += sq_stats->bytes;
+			s->tx_dropped    += sq_stats->dropped;
+		}
+	}
+	if (priv->rx_ptp_opened) {
+		struct mlx5e_rq_stats *rq_stats = &priv->ptp_stats.rq;
+
+		s->rx_packets   += rq_stats->packets;
+		s->rx_bytes     += rq_stats->bytes;
+		s->multicast    += rq_stats->mcast_packets;
+	}
+}
+
+void
+mlx5e_get_stats(struct net_device *dev, struct rtnl_link_stats64 *stats)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	struct mlx5e_pport_stats *pstats = &priv->stats.pport;
+
+	if (!netif_device_present(dev))
+		return;
+
+	/* In switchdev mode, monitor counters doesn't monitor
+	 * rx/tx stats of 802_3. The update stats mechanism
+	 * should keep the 802_3 layout counters updated
+	 */
+	if (!mlx5e_monitor_counter_supported(priv) ||
+	    mlx5e_is_uplink_rep(priv)) {
+		/* update HW stats in background for next time */
+		mlx5e_queue_update_stats(priv);
+	}
+
+	if (mlx5e_is_uplink_rep(priv)) {
+		struct mlx5e_vport_stats *vstats = &priv->stats.vport;
+
+		stats->rx_packets = PPORT_802_3_GET(pstats, a_frames_received_ok);
+		stats->rx_bytes   = PPORT_802_3_GET(pstats, a_octets_received_ok);
+		stats->tx_packets = PPORT_802_3_GET(pstats, a_frames_transmitted_ok);
+		stats->tx_bytes   = PPORT_802_3_GET(pstats, a_octets_transmitted_ok);
+
+		/* vport multicast also counts packets that are dropped due to steering
+		 * or rx out of buffer
+		 */
+		stats->multicast = VPORT_COUNTER_GET(vstats, received_eth_multicast.packets);
+	} else {
+		mlx5e_fold_sw_stats64(priv, stats);
+	}
+
+	stats->rx_dropped = priv->stats.qcnt.rx_out_of_buffer;
+
+	stats->rx_length_errors =
+		PPORT_802_3_GET(pstats, a_in_range_length_errors) +
+		PPORT_802_3_GET(pstats, a_out_of_range_length_field) +
+		PPORT_802_3_GET(pstats, a_frame_too_long_errors) +
+		VNIC_ENV_GET(&priv->stats.vnic, eth_wqe_too_small);
+	stats->rx_crc_errors =
+		PPORT_802_3_GET(pstats, a_frame_check_sequence_errors);
+	stats->rx_frame_errors = PPORT_802_3_GET(pstats, a_alignment_errors);
+	stats->tx_aborted_errors = PPORT_2863_GET(pstats, if_out_discards);
+	stats->rx_errors = stats->rx_length_errors + stats->rx_crc_errors +
+			   stats->rx_frame_errors;
+	stats->tx_errors = stats->tx_aborted_errors + stats->tx_carrier_errors;
+}
+
+static void mlx5e_nic_set_rx_mode(struct mlx5e_priv *priv)
+{
+	if (mlx5e_is_uplink_rep(priv))
+		return; /* no rx mode for uplink rep */
+
+	queue_work(priv->wq, &priv->set_rx_mode_work);
+}
+
+static void mlx5e_set_rx_mode(struct net_device *dev)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+
+	mlx5e_nic_set_rx_mode(priv);
+}
+
+static int mlx5e_set_mac(struct net_device *netdev, void *addr)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	struct sockaddr *saddr = addr;
+
+	if (!is_valid_ether_addr(saddr->sa_data))
+		return -EADDRNOTAVAIL;
+
+	netif_addr_lock_bh(netdev);
+	eth_hw_addr_set(netdev, saddr->sa_data);
+	netif_addr_unlock_bh(netdev);
+
+	mlx5e_nic_set_rx_mode(priv);
+
+	return 0;
+}
+
+#define MLX5E_SET_FEATURE(features, feature, enable)	\
+	do {						\
+		if (enable)				\
+			*features |= feature;		\
+		else					\
+			*features &= ~feature;		\
+	} while (0)
+
+typedef int (*mlx5e_feature_handler)(struct net_device *netdev, bool enable);
+
+static int set_feature_lro(struct net_device *netdev, bool enable)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+	struct mlx5e_params *cur_params;
+	struct mlx5e_params new_params;
+	bool reset = true;
+	int err = 0;
+
+	mutex_lock(&priv->state_lock);
+
+	cur_params = &priv->channels.params;
+	new_params = *cur_params;
+
+	if (enable)
+		new_params.packet_merge.type = MLX5E_PACKET_MERGE_LRO;
+	else if (new_params.packet_merge.type == MLX5E_PACKET_MERGE_LRO)
+		new_params.packet_merge.type = MLX5E_PACKET_MERGE_NONE;
+	else
+		goto out;
+
+	if (!(cur_params->packet_merge.type == MLX5E_PACKET_MERGE_SHAMPO &&
+	      new_params.packet_merge.type == MLX5E_PACKET_MERGE_LRO)) {
+		if (cur_params->rq_wq_type == MLX5_WQ_TYPE_LINKED_LIST_STRIDING_RQ) {
+			if (mlx5e_rx_mpwqe_is_linear_skb(mdev, cur_params, NULL) ==
+			    mlx5e_rx_mpwqe_is_linear_skb(mdev, &new_params, NULL))
+				reset = false;
+		}
+	}
+
+	err = mlx5e_safe_switch_params(priv, &new_params,
+				       mlx5e_modify_tirs_packet_merge_ctx, NULL, reset);
+out:
+	mutex_unlock(&priv->state_lock);
+	return err;
+}
+
+static int set_feature_hw_gro(struct net_device *netdev, bool enable)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	struct mlx5e_params new_params;
+	bool reset = true;
+	int err = 0;
+
+	mutex_lock(&priv->state_lock);
+	new_params = priv->channels.params;
+
+	if (enable) {
+		new_params.packet_merge.type = MLX5E_PACKET_MERGE_SHAMPO;
+		new_params.packet_merge.shampo.match_criteria_type =
+			MLX5_RQC_SHAMPO_MATCH_CRITERIA_TYPE_EXTENDED;
+		new_params.packet_merge.shampo.alignment_granularity =
+			MLX5_RQC_SHAMPO_NO_MATCH_ALIGNMENT_GRANULARITY_STRIDE;
+	} else if (new_params.packet_merge.type == MLX5E_PACKET_MERGE_SHAMPO) {
+		new_params.packet_merge.type = MLX5E_PACKET_MERGE_NONE;
+	} else {
+		goto out;
+	}
+
+	err = mlx5e_safe_switch_params(priv, &new_params, NULL, NULL, reset);
+out:
+	mutex_unlock(&priv->state_lock);
+	return err;
+}
+
+static int set_feature_cvlan_filter(struct net_device *netdev, bool enable)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+
+	if (enable)
+		mlx5e_enable_cvlan_filter(priv->fs,
+					  !!(priv->netdev->flags & IFF_PROMISC));
+	else
+		mlx5e_disable_cvlan_filter(priv->fs,
+					   !!(priv->netdev->flags & IFF_PROMISC));
+
+	return 0;
+}
+
+static int set_feature_hw_tc(struct net_device *netdev, bool enable)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	int err = 0;
+
+#if IS_ENABLED(CONFIG_MLX5_CLS_ACT)
+	int tc_flag = mlx5e_is_uplink_rep(priv) ? MLX5_TC_FLAG(ESW_OFFLOAD) :
+						  MLX5_TC_FLAG(NIC_OFFLOAD);
+	if (!enable && mlx5e_tc_num_filters(priv, tc_flag)) {
+		netdev_err(netdev,
+			   "Active offloaded tc filters, can't turn hw_tc_offload off\n");
+		return -EINVAL;
+	}
+#endif
+
+	mutex_lock(&priv->state_lock);
+	if (!enable && mlx5e_selq_is_htb_enabled(&priv->selq)) {
+		netdev_err(netdev, "Active HTB offload, can't turn hw_tc_offload off\n");
+		err = -EINVAL;
+	}
+	mutex_unlock(&priv->state_lock);
+
+	return err;
+}
+
+static int set_feature_rx_all(struct net_device *netdev, bool enable)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+
+	return mlx5_set_port_fcs(mdev, !enable);
+}
+
+static int mlx5e_set_rx_port_ts(struct mlx5_core_dev *mdev, bool enable)
+{
+	u32 in[MLX5_ST_SZ_DW(pcmr_reg)] = {};
+	bool supported, curr_state;
+	int err;
+
+	if (!MLX5_CAP_GEN(mdev, ports_check))
+		return 0;
+
+	err = mlx5_query_ports_check(mdev, in, sizeof(in));
+	if (err)
+		return err;
+
+	supported = MLX5_GET(pcmr_reg, in, rx_ts_over_crc_cap);
+	curr_state = MLX5_GET(pcmr_reg, in, rx_ts_over_crc);
+
+	if (!supported || enable == curr_state)
+		return 0;
+
+	MLX5_SET(pcmr_reg, in, local_port, 1);
+	MLX5_SET(pcmr_reg, in, rx_ts_over_crc, enable);
+
+	return mlx5_set_ports_check(mdev, in, sizeof(in));
+}
+
+static int mlx5e_set_rx_port_ts_wrap(struct mlx5e_priv *priv, void *ctx)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+	bool enable = *(bool *)ctx;
+
+	return mlx5e_set_rx_port_ts(mdev, enable);
+}
+
+static int set_feature_rx_fcs(struct net_device *netdev, bool enable)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	struct mlx5e_channels *chs = &priv->channels;
+	struct mlx5e_params new_params;
+	int err;
+	bool rx_ts_over_crc = !enable;
+
+	mutex_lock(&priv->state_lock);
+
+	new_params = chs->params;
+	new_params.scatter_fcs_en = enable;
+	err = mlx5e_safe_switch_params(priv, &new_params, mlx5e_set_rx_port_ts_wrap,
+				       &rx_ts_over_crc, true);
+	mutex_unlock(&priv->state_lock);
+	return err;
+}
+
+static int set_feature_rx_vlan(struct net_device *netdev, bool enable)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	int err = 0;
+
+	mutex_lock(&priv->state_lock);
+
+	mlx5e_fs_set_vlan_strip_disable(priv->fs, !enable);
+	priv->channels.params.vlan_strip_disable = !enable;
+
+	if (!test_bit(MLX5E_STATE_OPENED, &priv->state))
+		goto unlock;
+
+	err = mlx5e_modify_channels_vsd(&priv->channels, !enable);
+	if (err) {
+		mlx5e_fs_set_vlan_strip_disable(priv->fs, enable);
+		priv->channels.params.vlan_strip_disable = enable;
+	}
+unlock:
+	mutex_unlock(&priv->state_lock);
+
+	return err;
+}
+
+int mlx5e_vlan_rx_add_vid(struct net_device *dev, __be16 proto, u16 vid)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	struct mlx5e_flow_steering *fs = priv->fs;
+
+	if (mlx5e_is_uplink_rep(priv))
+		return 0; /* no vlan table for uplink rep */
+
+	return mlx5e_fs_vlan_rx_add_vid(fs, dev, proto, vid);
+}
+
+int mlx5e_vlan_rx_kill_vid(struct net_device *dev, __be16 proto, u16 vid)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	struct mlx5e_flow_steering *fs = priv->fs;
+
+	if (mlx5e_is_uplink_rep(priv))
+		return 0; /* no vlan table for uplink rep */
+
+	return mlx5e_fs_vlan_rx_kill_vid(fs, dev, proto, vid);
+}
+
+#ifdef CONFIG_MLX5_EN_ARFS
+static int set_feature_arfs(struct net_device *netdev, bool enable)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	int err;
+
+	if (enable)
+		err = mlx5e_arfs_enable(priv->fs);
+	else
+		err = mlx5e_arfs_disable(priv->fs);
+
+	return err;
+}
+#endif
+
+static int mlx5e_handle_feature(struct net_device *netdev,
+				netdev_features_t *features,
+				netdev_features_t feature,
+				mlx5e_feature_handler feature_handler)
+{
+	netdev_features_t changes = *features ^ netdev->features;
+	bool enable = !!(*features & feature);
+	int err;
+
+	if (!(changes & feature))
+		return 0;
+
+	err = feature_handler(netdev, enable);
+	if (err) {
+		MLX5E_SET_FEATURE(features, feature, !enable);
+		netdev_err(netdev, "%s feature %pNF failed, err %d\n",
+			   enable ? "Enable" : "Disable", &feature, err);
+		return err;
+	}
+
+	return 0;
+}
+
+int mlx5e_set_features(struct net_device *netdev, netdev_features_t features)
+{
+	netdev_features_t oper_features = features;
+	int err = 0;
+
+#define MLX5E_HANDLE_FEATURE(feature, handler) \
+	mlx5e_handle_feature(netdev, &oper_features, feature, handler)
+
+	err |= MLX5E_HANDLE_FEATURE(NETIF_F_LRO, set_feature_lro);
+	err |= MLX5E_HANDLE_FEATURE(NETIF_F_GRO_HW, set_feature_hw_gro);
+	err |= MLX5E_HANDLE_FEATURE(NETIF_F_HW_VLAN_CTAG_FILTER,
+				    set_feature_cvlan_filter);
+	err |= MLX5E_HANDLE_FEATURE(NETIF_F_HW_TC, set_feature_hw_tc);
+	err |= MLX5E_HANDLE_FEATURE(NETIF_F_RXALL, set_feature_rx_all);
+	err |= MLX5E_HANDLE_FEATURE(NETIF_F_RXFCS, set_feature_rx_fcs);
+	err |= MLX5E_HANDLE_FEATURE(NETIF_F_HW_VLAN_CTAG_RX, set_feature_rx_vlan);
+#ifdef CONFIG_MLX5_EN_ARFS
+	err |= MLX5E_HANDLE_FEATURE(NETIF_F_NTUPLE, set_feature_arfs);
+#endif
+	err |= MLX5E_HANDLE_FEATURE(NETIF_F_HW_TLS_RX, mlx5e_ktls_set_feature_rx);
+
+	if (err) {
+		netdev->features = oper_features;
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static netdev_features_t mlx5e_fix_uplink_rep_features(struct net_device *netdev,
+						       netdev_features_t features)
+{
+	features &= ~NETIF_F_HW_TLS_RX;
+	if (netdev->features & NETIF_F_HW_TLS_RX)
+		netdev_warn(netdev, "Disabling hw_tls_rx, not supported in switchdev mode\n");
+
+	features &= ~NETIF_F_HW_TLS_TX;
+	if (netdev->features & NETIF_F_HW_TLS_TX)
+		netdev_warn(netdev, "Disabling hw_tls_tx, not supported in switchdev mode\n");
+
+	features &= ~NETIF_F_NTUPLE;
+	if (netdev->features & NETIF_F_NTUPLE)
+		netdev_warn(netdev, "Disabling ntuple, not supported in switchdev mode\n");
+
+	features &= ~NETIF_F_GRO_HW;
+	if (netdev->features & NETIF_F_GRO_HW)
+		netdev_warn(netdev, "Disabling HW_GRO, not supported in switchdev mode\n");
+
+	features &= ~NETIF_F_HW_VLAN_CTAG_FILTER;
+	if (netdev->features & NETIF_F_HW_VLAN_CTAG_FILTER)
+		netdev_warn(netdev, "Disabling HW_VLAN CTAG FILTERING, not supported in switchdev mode\n");
+
+	return features;
+}
+
+static netdev_features_t mlx5e_fix_features(struct net_device *netdev,
+					    netdev_features_t features)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	struct mlx5e_vlan_table *vlan;
+	struct mlx5e_params *params;
+
+	if (!netif_device_present(netdev))
+		return features;
+
+	vlan = mlx5e_fs_get_vlan(priv->fs);
+	mutex_lock(&priv->state_lock);
+	params = &priv->channels.params;
+	if (!vlan ||
+	    !bitmap_empty(mlx5e_vlan_get_active_svlans(vlan), VLAN_N_VID)) {
+		/* HW strips the outer C-tag header, this is a problem
+		 * for S-tag traffic.
+		 */
+		features &= ~NETIF_F_HW_VLAN_CTAG_RX;
+		if (!params->vlan_strip_disable)
+			netdev_warn(netdev, "Dropping C-tag vlan stripping offload due to S-tag vlan\n");
+	}
+
+	if (!MLX5E_GET_PFLAG(params, MLX5E_PFLAG_RX_STRIDING_RQ)) {
+		if (features & NETIF_F_LRO) {
+			netdev_warn(netdev, "Disabling LRO, not supported in legacy RQ\n");
+			features &= ~NETIF_F_LRO;
+		}
+		if (features & NETIF_F_GRO_HW) {
+			netdev_warn(netdev, "Disabling HW-GRO, not supported in legacy RQ\n");
+			features &= ~NETIF_F_GRO_HW;
+		}
+	}
+
+	if (params->xdp_prog) {
+		if (features & NETIF_F_LRO) {
+			netdev_warn(netdev, "LRO is incompatible with XDP\n");
+			features &= ~NETIF_F_LRO;
+		}
+		if (features & NETIF_F_GRO_HW) {
+			netdev_warn(netdev, "HW GRO is incompatible with XDP\n");
+			features &= ~NETIF_F_GRO_HW;
+		}
+	}
+
+	if (priv->xsk.refcnt) {
+		if (features & NETIF_F_LRO) {
+			netdev_warn(netdev, "LRO is incompatible with AF_XDP (%u XSKs are active)\n",
+				    priv->xsk.refcnt);
+			features &= ~NETIF_F_LRO;
+		}
+		if (features & NETIF_F_GRO_HW) {
+			netdev_warn(netdev, "HW GRO is incompatible with AF_XDP (%u XSKs are active)\n",
+				    priv->xsk.refcnt);
+			features &= ~NETIF_F_GRO_HW;
+		}
+	}
+
+	if (MLX5E_GET_PFLAG(params, MLX5E_PFLAG_RX_CQE_COMPRESS)) {
+		features &= ~NETIF_F_RXHASH;
+		if (netdev->features & NETIF_F_RXHASH)
+			netdev_warn(netdev, "Disabling rxhash, not supported when CQE compress is active\n");
+
+		if (features & NETIF_F_GRO_HW) {
+			netdev_warn(netdev, "Disabling HW-GRO, not supported when CQE compress is active\n");
+			features &= ~NETIF_F_GRO_HW;
+		}
+	}
+
+	if (mlx5e_is_uplink_rep(priv)) {
+		features = mlx5e_fix_uplink_rep_features(netdev, features);
+		features |= NETIF_F_NETNS_LOCAL;
+	} else {
+		features &= ~NETIF_F_NETNS_LOCAL;
+	}
+
+	mutex_unlock(&priv->state_lock);
+
+	return features;
+}
+
+static bool mlx5e_xsk_validate_mtu(struct net_device *netdev,
+				   struct mlx5e_channels *chs,
+				   struct mlx5e_params *new_params,
+				   struct mlx5_core_dev *mdev)
+{
+	u16 ix;
+
+	for (ix = 0; ix < chs->params.num_channels; ix++) {
+		struct xsk_buff_pool *xsk_pool =
+			mlx5e_xsk_get_pool(&chs->params, chs->params.xsk, ix);
+		struct mlx5e_xsk_param xsk;
+		int max_xdp_mtu;
+
+		if (!xsk_pool)
+			continue;
+
+		mlx5e_build_xsk_param(xsk_pool, &xsk);
+		max_xdp_mtu = mlx5e_xdp_max_mtu(new_params, &xsk);
+
+		/* Validate XSK params and XDP MTU in advance */
+		if (!mlx5e_validate_xsk_param(new_params, &xsk, mdev) ||
+		    new_params->sw_mtu > max_xdp_mtu) {
+			u32 hr = mlx5e_get_linear_rq_headroom(new_params, &xsk);
+			int max_mtu_frame, max_mtu_page, max_mtu;
+
+			/* Two criteria must be met:
+			 * 1. HW MTU + all headrooms <= XSK frame size.
+			 * 2. Size of SKBs allocated on XDP_PASS <= PAGE_SIZE.
+			 */
+			max_mtu_frame = MLX5E_HW2SW_MTU(new_params, xsk.chunk_size - hr);
+			max_mtu_page = MLX5E_HW2SW_MTU(new_params, SKB_MAX_HEAD(0));
+			max_mtu = min3(max_mtu_frame, max_mtu_page, max_xdp_mtu);
+
+			netdev_err(netdev, "MTU %d is too big for an XSK running on channel %u or its redirection XDP program. Try MTU <= %d\n",
+				   new_params->sw_mtu, ix, max_mtu);
+			return false;
+		}
+	}
+
+	return true;
+}
+
+static bool mlx5e_params_validate_xdp(struct net_device *netdev,
+				      struct mlx5_core_dev *mdev,
+				      struct mlx5e_params *params)
+{
+	bool is_linear;
+
+	/* No XSK params: AF_XDP can't be enabled yet at the point of setting
+	 * the XDP program.
+	 */
+	is_linear = mlx5e_rx_is_linear_skb(mdev, params, NULL);
+
+	if (!is_linear && params->rq_wq_type != MLX5_WQ_TYPE_CYCLIC) {
+		netdev_warn(netdev, "XDP is not allowed with striding RQ and MTU(%d) > %d\n",
+			    params->sw_mtu,
+			    mlx5e_xdp_max_mtu(params, NULL));
+		return false;
+	}
+	if (!is_linear && !params->xdp_prog->aux->xdp_has_frags) {
+		netdev_warn(netdev, "MTU(%d) > %d, too big for an XDP program not aware of multi buffer\n",
+			    params->sw_mtu,
+			    mlx5e_xdp_max_mtu(params, NULL));
+		return false;
+	}
+
+	return true;
+}
+
+int mlx5e_change_mtu(struct net_device *netdev, int new_mtu,
+		     mlx5e_fp_preactivate preactivate)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	struct mlx5e_params new_params;
+	struct mlx5e_params *params;
+	bool reset = true;
+	int err = 0;
+
+	mutex_lock(&priv->state_lock);
+
+	params = &priv->channels.params;
+
+	new_params = *params;
+	new_params.sw_mtu = new_mtu;
+	err = mlx5e_validate_params(priv->mdev, &new_params);
+	if (err)
+		goto out;
+
+	if (new_params.xdp_prog && !mlx5e_params_validate_xdp(netdev, priv->mdev,
+							      &new_params)) {
+		err = -EINVAL;
+		goto out;
+	}
+
+	if (priv->xsk.refcnt &&
+	    !mlx5e_xsk_validate_mtu(netdev, &priv->channels,
+				    &new_params, priv->mdev)) {
+		err = -EINVAL;
+		goto out;
+	}
+
+	if (params->packet_merge.type == MLX5E_PACKET_MERGE_LRO)
+		reset = false;
+
+	if (params->rq_wq_type == MLX5_WQ_TYPE_LINKED_LIST_STRIDING_RQ &&
+	    params->packet_merge.type != MLX5E_PACKET_MERGE_SHAMPO) {
+		bool is_linear_old = mlx5e_rx_mpwqe_is_linear_skb(priv->mdev, params, NULL);
+		bool is_linear_new = mlx5e_rx_mpwqe_is_linear_skb(priv->mdev,
+								  &new_params, NULL);
+		u8 sz_old = mlx5e_mpwqe_get_log_rq_size(priv->mdev, params, NULL);
+		u8 sz_new = mlx5e_mpwqe_get_log_rq_size(priv->mdev, &new_params, NULL);
+
+		/* Always reset in linear mode - hw_mtu is used in data path.
+		 * Check that the mode was non-linear and didn't change.
+		 * If XSK is active, XSK RQs are linear.
+		 * Reset if the RQ size changed, even if it's non-linear.
+		 */
+		if (!is_linear_old && !is_linear_new && !priv->xsk.refcnt &&
+		    sz_old == sz_new)
+			reset = false;
+	}
+
+	err = mlx5e_safe_switch_params(priv, &new_params, preactivate, NULL, reset);
+
+out:
+	netdev->mtu = params->sw_mtu;
+	mutex_unlock(&priv->state_lock);
+	return err;
+}
+
+static int mlx5e_change_nic_mtu(struct net_device *netdev, int new_mtu)
+{
+	return mlx5e_change_mtu(netdev, new_mtu, mlx5e_set_dev_port_mtu_ctx);
+}
+
+int mlx5e_ptp_rx_manage_fs_ctx(struct mlx5e_priv *priv, void *ctx)
+{
+	bool set  = *(bool *)ctx;
+
+	return mlx5e_ptp_rx_manage_fs(priv, set);
+}
+
+static int mlx5e_hwstamp_config_no_ptp_rx(struct mlx5e_priv *priv, bool rx_filter)
+{
+	bool rx_cqe_compress_def = priv->channels.params.rx_cqe_compress_def;
+	int err;
+
+	if (!rx_filter)
+		/* Reset CQE compression to Admin default */
+		return mlx5e_modify_rx_cqe_compression_locked(priv, rx_cqe_compress_def, false);
+
+	if (!MLX5E_GET_PFLAG(&priv->channels.params, MLX5E_PFLAG_RX_CQE_COMPRESS))
+		return 0;
+
+	/* Disable CQE compression */
+	netdev_warn(priv->netdev, "Disabling RX cqe compression\n");
+	err = mlx5e_modify_rx_cqe_compression_locked(priv, false, true);
+	if (err)
+		netdev_err(priv->netdev, "Failed disabling cqe compression err=%d\n", err);
+
+	return err;
+}
+
+static int mlx5e_hwstamp_config_ptp_rx(struct mlx5e_priv *priv, bool ptp_rx)
+{
+	struct mlx5e_params new_params;
+
+	if (ptp_rx == priv->channels.params.ptp_rx)
+		return 0;
+
+	new_params = priv->channels.params;
+	new_params.ptp_rx = ptp_rx;
+	return mlx5e_safe_switch_params(priv, &new_params, mlx5e_ptp_rx_manage_fs_ctx,
+					&new_params.ptp_rx, true);
+}
+
+int mlx5e_hwstamp_set(struct mlx5e_priv *priv, struct ifreq *ifr)
+{
+	struct hwtstamp_config config;
+	bool rx_cqe_compress_def;
+	bool ptp_rx;
+	int err;
+
+	if (!MLX5_CAP_GEN(priv->mdev, device_frequency_khz) ||
+	    (mlx5_clock_get_ptp_index(priv->mdev) == -1))
+		return -EOPNOTSUPP;
+
+	if (copy_from_user(&config, ifr->ifr_data, sizeof(config)))
+		return -EFAULT;
+
+	/* TX HW timestamp */
+	switch (config.tx_type) {
+	case HWTSTAMP_TX_OFF:
+	case HWTSTAMP_TX_ON:
+		break;
+	default:
+		return -ERANGE;
+	}
+
+	mutex_lock(&priv->state_lock);
+	rx_cqe_compress_def = priv->channels.params.rx_cqe_compress_def;
+
+	/* RX HW timestamp */
+	switch (config.rx_filter) {
+	case HWTSTAMP_FILTER_NONE:
+		ptp_rx = false;
+		break;
+	case HWTSTAMP_FILTER_ALL:
+	case HWTSTAMP_FILTER_SOME:
+	case HWTSTAMP_FILTER_PTP_V1_L4_EVENT:
+	case HWTSTAMP_FILTER_PTP_V1_L4_SYNC:
+	case HWTSTAMP_FILTER_PTP_V1_L4_DELAY_REQ:
+	case HWTSTAMP_FILTER_PTP_V2_L4_EVENT:
+	case HWTSTAMP_FILTER_PTP_V2_L4_SYNC:
+	case HWTSTAMP_FILTER_PTP_V2_L4_DELAY_REQ:
+	case HWTSTAMP_FILTER_PTP_V2_L2_EVENT:
+	case HWTSTAMP_FILTER_PTP_V2_L2_SYNC:
+	case HWTSTAMP_FILTER_PTP_V2_L2_DELAY_REQ:
+	case HWTSTAMP_FILTER_PTP_V2_EVENT:
+	case HWTSTAMP_FILTER_PTP_V2_SYNC:
+	case HWTSTAMP_FILTER_PTP_V2_DELAY_REQ:
+	case HWTSTAMP_FILTER_NTP_ALL:
+		config.rx_filter = HWTSTAMP_FILTER_ALL;
+		/* ptp_rx is set if both HW TS is set and CQE
+		 * compression is set
+		 */
+		ptp_rx = rx_cqe_compress_def;
+		break;
+	default:
+		err = -ERANGE;
+		goto err_unlock;
+	}
+
+	if (!mlx5e_profile_feature_cap(priv->profile, PTP_RX))
+		err = mlx5e_hwstamp_config_no_ptp_rx(priv,
+						     config.rx_filter != HWTSTAMP_FILTER_NONE);
+	else
+		err = mlx5e_hwstamp_config_ptp_rx(priv, ptp_rx);
+	if (err)
+		goto err_unlock;
+
+	memcpy(&priv->tstamp, &config, sizeof(config));
+	mutex_unlock(&priv->state_lock);
+
+	/* might need to fix some features */
+	netdev_update_features(priv->netdev);
+
+	return copy_to_user(ifr->ifr_data, &config,
+			    sizeof(config)) ? -EFAULT : 0;
+err_unlock:
+	mutex_unlock(&priv->state_lock);
+	return err;
+}
+
+int mlx5e_hwstamp_get(struct mlx5e_priv *priv, struct ifreq *ifr)
+{
+	struct hwtstamp_config *cfg = &priv->tstamp;
+
+	if (!MLX5_CAP_GEN(priv->mdev, device_frequency_khz))
+		return -EOPNOTSUPP;
+
+	return copy_to_user(ifr->ifr_data, cfg, sizeof(*cfg)) ? -EFAULT : 0;
+}
+
+static int mlx5e_ioctl(struct net_device *dev, struct ifreq *ifr, int cmd)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+
+	switch (cmd) {
+	case SIOCSHWTSTAMP:
+		return mlx5e_hwstamp_set(priv, ifr);
+	case SIOCGHWTSTAMP:
+		return mlx5e_hwstamp_get(priv, ifr);
+	default:
+		return -EOPNOTSUPP;
+	}
+}
+
+#ifdef CONFIG_MLX5_ESWITCH
+int mlx5e_set_vf_mac(struct net_device *dev, int vf, u8 *mac)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+
+	return mlx5_eswitch_set_vport_mac(mdev->priv.eswitch, vf + 1, mac);
+}
+
+static int mlx5e_set_vf_vlan(struct net_device *dev, int vf, u16 vlan, u8 qos,
+			     __be16 vlan_proto)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+
+	if (vlan_proto != htons(ETH_P_8021Q))
+		return -EPROTONOSUPPORT;
+
+	return mlx5_eswitch_set_vport_vlan(mdev->priv.eswitch, vf + 1,
+					   vlan, qos);
+}
+
+static int mlx5e_set_vf_spoofchk(struct net_device *dev, int vf, bool setting)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+
+	return mlx5_eswitch_set_vport_spoofchk(mdev->priv.eswitch, vf + 1, setting);
+}
+
+static int mlx5e_set_vf_trust(struct net_device *dev, int vf, bool setting)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+
+	return mlx5_eswitch_set_vport_trust(mdev->priv.eswitch, vf + 1, setting);
+}
+
+int mlx5e_set_vf_rate(struct net_device *dev, int vf, int min_tx_rate,
+		      int max_tx_rate)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+
+	return mlx5_eswitch_set_vport_rate(mdev->priv.eswitch, vf + 1,
+					   max_tx_rate, min_tx_rate);
+}
+
+static int mlx5_vport_link2ifla(u8 esw_link)
+{
+	switch (esw_link) {
+	case MLX5_VPORT_ADMIN_STATE_DOWN:
+		return IFLA_VF_LINK_STATE_DISABLE;
+	case MLX5_VPORT_ADMIN_STATE_UP:
+		return IFLA_VF_LINK_STATE_ENABLE;
+	}
+	return IFLA_VF_LINK_STATE_AUTO;
+}
+
+static int mlx5_ifla_link2vport(u8 ifla_link)
+{
+	switch (ifla_link) {
+	case IFLA_VF_LINK_STATE_DISABLE:
+		return MLX5_VPORT_ADMIN_STATE_DOWN;
+	case IFLA_VF_LINK_STATE_ENABLE:
+		return MLX5_VPORT_ADMIN_STATE_UP;
+	}
+	return MLX5_VPORT_ADMIN_STATE_AUTO;
+}
+
+static int mlx5e_set_vf_link_state(struct net_device *dev, int vf,
+				   int link_state)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+
+	if (mlx5e_is_uplink_rep(priv))
+		return -EOPNOTSUPP;
+
+	return mlx5_eswitch_set_vport_state(mdev->priv.eswitch, vf + 1,
+					    mlx5_ifla_link2vport(link_state));
+}
+
+int mlx5e_get_vf_config(struct net_device *dev,
+			int vf, struct ifla_vf_info *ivi)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+	int err;
+
+	if (!netif_device_present(dev))
+		return -EOPNOTSUPP;
+
+	err = mlx5_eswitch_get_vport_config(mdev->priv.eswitch, vf + 1, ivi);
+	if (err)
+		return err;
+	ivi->linkstate = mlx5_vport_link2ifla(ivi->linkstate);
+	return 0;
+}
+
+int mlx5e_get_vf_stats(struct net_device *dev,
+		       int vf, struct ifla_vf_stats *vf_stats)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+
+	return mlx5_eswitch_get_vport_stats(mdev->priv.eswitch, vf + 1,
+					    vf_stats);
+}
+
+static bool
+mlx5e_has_offload_stats(const struct net_device *dev, int attr_id)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+
+	if (!netif_device_present(dev))
+		return false;
+
+	if (!mlx5e_is_uplink_rep(priv))
+		return false;
+
+	return mlx5e_rep_has_offload_stats(dev, attr_id);
+}
+
+static int
+mlx5e_get_offload_stats(int attr_id, const struct net_device *dev,
+			void *sp)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+
+	if (!mlx5e_is_uplink_rep(priv))
+		return -EOPNOTSUPP;
+
+	return mlx5e_rep_get_offload_stats(attr_id, dev, sp);
+}
+#endif
+
+static bool mlx5e_tunnel_proto_supported_tx(struct mlx5_core_dev *mdev, u8 proto_type)
+{
+	switch (proto_type) {
+	case IPPROTO_GRE:
+		return MLX5_CAP_ETH(mdev, tunnel_stateless_gre);
+	case IPPROTO_IPIP:
+	case IPPROTO_IPV6:
+		return (MLX5_CAP_ETH(mdev, tunnel_stateless_ip_over_ip) ||
+			MLX5_CAP_ETH(mdev, tunnel_stateless_ip_over_ip_tx));
+	default:
+		return false;
+	}
+}
+
+static bool mlx5e_gre_tunnel_inner_proto_offload_supported(struct mlx5_core_dev *mdev,
+							   struct sk_buff *skb)
+{
+	switch (skb->inner_protocol) {
+	case htons(ETH_P_IP):
+	case htons(ETH_P_IPV6):
+	case htons(ETH_P_TEB):
+		return true;
+	case htons(ETH_P_MPLS_UC):
+	case htons(ETH_P_MPLS_MC):
+		return MLX5_CAP_ETH(mdev, tunnel_stateless_mpls_over_gre);
+	}
+	return false;
+}
+
+static netdev_features_t mlx5e_tunnel_features_check(struct mlx5e_priv *priv,
+						     struct sk_buff *skb,
+						     netdev_features_t features)
+{
+	unsigned int offset = 0;
+	struct udphdr *udph;
+	u8 proto;
+	u16 port;
+
+	switch (vlan_get_protocol(skb)) {
+	case htons(ETH_P_IP):
+		proto = ip_hdr(skb)->protocol;
+		break;
+	case htons(ETH_P_IPV6):
+		proto = ipv6_find_hdr(skb, &offset, -1, NULL, NULL);
+		break;
+	default:
+		goto out;
+	}
+
+	switch (proto) {
+	case IPPROTO_GRE:
+		if (mlx5e_gre_tunnel_inner_proto_offload_supported(priv->mdev, skb))
+			return features;
+		break;
+	case IPPROTO_IPIP:
+	case IPPROTO_IPV6:
+		if (mlx5e_tunnel_proto_supported_tx(priv->mdev, IPPROTO_IPIP))
+			return features;
+		break;
+	case IPPROTO_UDP:
+		udph = udp_hdr(skb);
+		port = be16_to_cpu(udph->dest);
+
+		/* Verify if UDP port is being offloaded by HW */
+		if (mlx5_vxlan_lookup_port(priv->mdev->vxlan, port))
+			return features;
+
+#if IS_ENABLED(CONFIG_GENEVE)
+		/* Support Geneve offload for default UDP port */
+		if (port == GENEVE_UDP_PORT && mlx5_geneve_tx_allowed(priv->mdev))
+			return features;
+#endif
+		break;
+#ifdef CONFIG_MLX5_EN_IPSEC
+	case IPPROTO_ESP:
+		return mlx5e_ipsec_feature_check(skb, features);
+#endif
+	}
+
+out:
+	/* Disable CSUM and GSO if the udp dport is not offloaded by HW */
+	return features & ~(NETIF_F_CSUM_MASK | NETIF_F_GSO_MASK);
+}
+
+netdev_features_t mlx5e_features_check(struct sk_buff *skb,
+				       struct net_device *netdev,
+				       netdev_features_t features)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+
+	features = vlan_features_check(skb, features);
+	features = vxlan_features_check(skb, features);
+
+	/* Validate if the tunneled packet is being offloaded by HW */
+	if (skb->encapsulation &&
+	    (features & NETIF_F_CSUM_MASK || features & NETIF_F_GSO_MASK))
+		return mlx5e_tunnel_features_check(priv, skb, features);
+
+	return features;
+}
+
+static void mlx5e_tx_timeout_work(struct work_struct *work)
+{
+	struct mlx5e_priv *priv = container_of(work, struct mlx5e_priv,
+					       tx_timeout_work);
+	struct net_device *netdev = priv->netdev;
+	int i;
+
+	/* Take rtnl_lock to ensure no change in netdev->real_num_tx_queues
+	 * through this flow. However, channel closing flows have to wait for
+	 * this work to finish while holding rtnl lock too. So either get the
+	 * lock or find that channels are being closed for other reason and
+	 * this work is not relevant anymore.
+	 */
+	while (!rtnl_trylock()) {
+		if (!test_bit(MLX5E_STATE_CHANNELS_ACTIVE, &priv->state))
+			return;
+		msleep(20);
+	}
+
+	if (!test_bit(MLX5E_STATE_OPENED, &priv->state))
+		goto unlock;
+
+	for (i = 0; i < netdev->real_num_tx_queues; i++) {
+		struct netdev_queue *dev_queue =
+			netdev_get_tx_queue(netdev, i);
+		struct mlx5e_txqsq *sq = priv->txq2sq[i];
+
+		if (!netif_xmit_stopped(dev_queue))
+			continue;
+
+		if (mlx5e_reporter_tx_timeout(sq))
+		/* break if tried to reopened channels */
+			break;
+	}
+
+unlock:
+	rtnl_unlock();
+}
+
+static void mlx5e_tx_timeout(struct net_device *dev, unsigned int txqueue)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+
+	netdev_err(dev, "TX timeout detected\n");
+	queue_work(priv->wq, &priv->tx_timeout_work);
+}
+
+static int mlx5e_xdp_allowed(struct mlx5e_priv *priv, struct bpf_prog *prog)
+{
+	struct net_device *netdev = priv->netdev;
+	struct mlx5e_params new_params;
+
+	if (priv->channels.params.packet_merge.type != MLX5E_PACKET_MERGE_NONE) {
+		netdev_warn(netdev, "can't set XDP while HW-GRO/LRO is on, disable them first\n");
+		return -EINVAL;
+	}
+
+	new_params = priv->channels.params;
+	new_params.xdp_prog = prog;
+
+	if (!mlx5e_params_validate_xdp(netdev, priv->mdev, &new_params))
+		return -EINVAL;
+
+	return 0;
+}
+
+static void mlx5e_rq_replace_xdp_prog(struct mlx5e_rq *rq, struct bpf_prog *prog)
+{
+	struct bpf_prog *old_prog;
+
+	old_prog = rcu_replace_pointer(rq->xdp_prog, prog,
+				       lockdep_is_held(&rq->priv->state_lock));
+	if (old_prog)
+		bpf_prog_put(old_prog);
+}
+
+static int mlx5e_xdp_set(struct net_device *netdev, struct bpf_prog *prog)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	struct mlx5e_params new_params;
+	struct bpf_prog *old_prog;
+	int err = 0;
+	bool reset;
+	int i;
+
+	mutex_lock(&priv->state_lock);
+
+	if (prog) {
+		err = mlx5e_xdp_allowed(priv, prog);
+		if (err)
+			goto unlock;
+	}
+
+	/* no need for full reset when exchanging programs */
+	reset = (!priv->channels.params.xdp_prog || !prog);
+
+	new_params = priv->channels.params;
+	new_params.xdp_prog = prog;
+
+	/* XDP affects striding RQ parameters. Block XDP if striding RQ won't be
+	 * supported with the new parameters: if PAGE_SIZE is bigger than
+	 * MLX5_MPWQE_LOG_STRIDE_SZ_MAX, striding RQ can't be used, even though
+	 * the MTU is small enough for the linear mode, because XDP uses strides
+	 * of PAGE_SIZE on regular RQs.
+	 */
+	if (reset && MLX5E_GET_PFLAG(&new_params, MLX5E_PFLAG_RX_STRIDING_RQ)) {
+		/* Checking for regular RQs here; XSK RQs were checked on XSK bind. */
+		err = mlx5e_mpwrq_validate_regular(priv->mdev, &new_params);
+		if (err)
+			goto unlock;
+	}
+
+	old_prog = priv->channels.params.xdp_prog;
+
+	err = mlx5e_safe_switch_params(priv, &new_params, NULL, NULL, reset);
+	if (err)
+		goto unlock;
+
+	if (old_prog)
+		bpf_prog_put(old_prog);
+
+	if (!test_bit(MLX5E_STATE_OPENED, &priv->state) || reset)
+		goto unlock;
+
+	/* exchanging programs w/o reset, we update ref counts on behalf
+	 * of the channels RQs here.
+	 */
+	bpf_prog_add(prog, priv->channels.num);
+	for (i = 0; i < priv->channels.num; i++) {
+		struct mlx5e_channel *c = priv->channels.c[i];
+
+		mlx5e_rq_replace_xdp_prog(&c->rq, prog);
+		if (test_bit(MLX5E_CHANNEL_STATE_XSK, c->state)) {
+			bpf_prog_inc(prog);
+			mlx5e_rq_replace_xdp_prog(&c->xskrq, prog);
+		}
+	}
+
+unlock:
+	mutex_unlock(&priv->state_lock);
+
+	/* Need to fix some features. */
+	if (!err)
+		netdev_update_features(netdev);
+
+	return err;
+}
+
+static int mlx5e_xdp(struct net_device *dev, struct netdev_bpf *xdp)
+{
+	switch (xdp->command) {
+	case XDP_SETUP_PROG:
+		return mlx5e_xdp_set(dev, xdp->prog);
+	case XDP_SETUP_XSK_POOL:
+		return mlx5e_xsk_setup_pool(dev, xdp->xsk.pool,
+					    xdp->xsk.queue_id);
+	default:
+		return -EINVAL;
+	}
+}
+
+#ifdef CONFIG_MLX5_ESWITCH
+static int mlx5e_bridge_getlink(struct sk_buff *skb, u32 pid, u32 seq,
+				struct net_device *dev, u32 filter_mask,
+				int nlflags)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+	u8 mode, setting;
+	int err;
+
+	err = mlx5_eswitch_get_vepa(mdev->priv.eswitch, &setting);
+	if (err)
+		return err;
+	mode = setting ? BRIDGE_MODE_VEPA : BRIDGE_MODE_VEB;
+	return ndo_dflt_bridge_getlink(skb, pid, seq, dev,
+				       mode,
+				       0, 0, nlflags, filter_mask, NULL);
+}
+
+static int mlx5e_bridge_setlink(struct net_device *dev, struct nlmsghdr *nlh,
+				u16 flags, struct netlink_ext_ack *extack)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+	struct nlattr *attr, *br_spec;
+	u16 mode = BRIDGE_MODE_UNDEF;
+	u8 setting;
+	int rem;
+
+	br_spec = nlmsg_find_attr(nlh, sizeof(struct ifinfomsg), IFLA_AF_SPEC);
+	if (!br_spec)
+		return -EINVAL;
+
+	nla_for_each_nested(attr, br_spec, rem) {
+		if (nla_type(attr) != IFLA_BRIDGE_MODE)
+			continue;
+
+		if (nla_len(attr) < sizeof(mode))
+			return -EINVAL;
+
+		mode = nla_get_u16(attr);
+		if (mode > BRIDGE_MODE_VEPA)
+			return -EINVAL;
+
+		break;
+	}
+
+	if (mode == BRIDGE_MODE_UNDEF)
+		return -EINVAL;
+
+	setting = (mode == BRIDGE_MODE_VEPA) ?  1 : 0;
+	return mlx5_eswitch_set_vepa(mdev->priv.eswitch, setting);
+}
+#endif
+
+const struct net_device_ops mlx5e_netdev_ops = {
+	.ndo_open                = mlx5e_open,
+	.ndo_stop                = mlx5e_close,
+	.ndo_start_xmit          = mlx5e_xmit,
+	.ndo_setup_tc            = mlx5e_setup_tc,
+	.ndo_select_queue        = mlx5e_select_queue,
+	.ndo_get_stats64         = mlx5e_get_stats,
+	.ndo_set_rx_mode         = mlx5e_set_rx_mode,
+	.ndo_set_mac_address     = mlx5e_set_mac,
+	.ndo_vlan_rx_add_vid     = mlx5e_vlan_rx_add_vid,
+	.ndo_vlan_rx_kill_vid    = mlx5e_vlan_rx_kill_vid,
+	.ndo_set_features        = mlx5e_set_features,
+	.ndo_fix_features        = mlx5e_fix_features,
+	.ndo_change_mtu          = mlx5e_change_nic_mtu,
+	.ndo_eth_ioctl            = mlx5e_ioctl,
+	.ndo_set_tx_maxrate      = mlx5e_set_tx_maxrate,
+	.ndo_features_check      = mlx5e_features_check,
+	.ndo_tx_timeout          = mlx5e_tx_timeout,
+	.ndo_bpf		 = mlx5e_xdp,
+	.ndo_xdp_xmit            = mlx5e_xdp_xmit,
+	.ndo_xsk_wakeup          = mlx5e_xsk_wakeup,
+#ifdef CONFIG_MLX5_EN_ARFS
+	.ndo_rx_flow_steer	 = mlx5e_rx_flow_steer,
+#endif
+#ifdef CONFIG_MLX5_ESWITCH
+	.ndo_bridge_setlink      = mlx5e_bridge_setlink,
+	.ndo_bridge_getlink      = mlx5e_bridge_getlink,
+
+	/* SRIOV E-Switch NDOs */
+	.ndo_set_vf_mac          = mlx5e_set_vf_mac,
+	.ndo_set_vf_vlan         = mlx5e_set_vf_vlan,
+	.ndo_set_vf_spoofchk     = mlx5e_set_vf_spoofchk,
+	.ndo_set_vf_trust        = mlx5e_set_vf_trust,
+	.ndo_set_vf_rate         = mlx5e_set_vf_rate,
+	.ndo_get_vf_config       = mlx5e_get_vf_config,
+	.ndo_set_vf_link_state   = mlx5e_set_vf_link_state,
+	.ndo_get_vf_stats        = mlx5e_get_vf_stats,
+	.ndo_has_offload_stats   = mlx5e_has_offload_stats,
+	.ndo_get_offload_stats   = mlx5e_get_offload_stats,
+#endif
+	.ndo_get_devlink_port    = mlx5e_get_devlink_port,
+};
+
+static u32 mlx5e_choose_lro_timeout(struct mlx5_core_dev *mdev, u32 wanted_timeout)
+{
+	int i;
+
+	/* The supported periods are organized in ascending order */
+	for (i = 0; i < MLX5E_LRO_TIMEOUT_ARR_SIZE - 1; i++)
+		if (MLX5_CAP_ETH(mdev, lro_timer_supported_periods[i]) >= wanted_timeout)
+			break;
+
+	return MLX5_CAP_ETH(mdev, lro_timer_supported_periods[i]);
+}
+
+void mlx5e_build_nic_params(struct mlx5e_priv *priv, struct mlx5e_xsk *xsk, u16 mtu)
+{
+	struct mlx5e_params *params = &priv->channels.params;
+	struct mlx5_core_dev *mdev = priv->mdev;
+	u8 rx_cq_period_mode;
+
+	params->sw_mtu = mtu;
+	params->hard_mtu = MLX5E_ETH_HARD_MTU;
+	params->num_channels = min_t(unsigned int, MLX5E_MAX_NUM_CHANNELS / 2,
+				     priv->max_nch);
+	mlx5e_params_mqprio_reset(params);
+
+	/* SQ */
+	params->log_sq_size = is_kdump_kernel() ?
+		MLX5E_PARAMS_MINIMUM_LOG_SQ_SIZE :
+		MLX5E_PARAMS_DEFAULT_LOG_SQ_SIZE;
+	MLX5E_SET_PFLAG(params, MLX5E_PFLAG_SKB_TX_MPWQE, mlx5e_tx_mpwqe_supported(mdev));
+
+	/* XDP SQ */
+	MLX5E_SET_PFLAG(params, MLX5E_PFLAG_XDP_TX_MPWQE, mlx5e_tx_mpwqe_supported(mdev));
+
+	/* set CQE compression */
+	params->rx_cqe_compress_def = false;
+	if (MLX5_CAP_GEN(mdev, cqe_compression) &&
+	    MLX5_CAP_GEN(mdev, vport_group_manager))
+		params->rx_cqe_compress_def = slow_pci_heuristic(mdev);
+
+	MLX5E_SET_PFLAG(params, MLX5E_PFLAG_RX_CQE_COMPRESS, params->rx_cqe_compress_def);
+	MLX5E_SET_PFLAG(params, MLX5E_PFLAG_RX_NO_CSUM_COMPLETE, false);
+
+	/* RQ */
+	mlx5e_build_rq_params(mdev, params);
+
+	params->packet_merge.timeout = mlx5e_choose_lro_timeout(mdev, MLX5E_DEFAULT_LRO_TIMEOUT);
+
+	/* CQ moderation params */
+	rx_cq_period_mode = MLX5_CAP_GEN(mdev, cq_period_start_from_cqe) ?
+			MLX5_CQ_PERIOD_MODE_START_FROM_CQE :
+			MLX5_CQ_PERIOD_MODE_START_FROM_EQE;
+	params->rx_dim_enabled = MLX5_CAP_GEN(mdev, cq_moderation);
+	params->tx_dim_enabled = MLX5_CAP_GEN(mdev, cq_moderation);
+	mlx5e_set_rx_cq_mode_params(params, rx_cq_period_mode);
+	mlx5e_set_tx_cq_mode_params(params, MLX5_CQ_PERIOD_MODE_START_FROM_EQE);
+
+	/* TX inline */
+	mlx5_query_min_inline(mdev, &params->tx_min_inline_mode);
+
+	/* AF_XDP */
+	params->xsk = xsk;
+
+	/* Do not update netdev->features directly in here
+	 * on mlx5e_attach_netdev() we will call mlx5e_update_features()
+	 * To update netdev->features please modify mlx5e_fix_features()
+	 */
+}
+
+static void mlx5e_set_netdev_dev_addr(struct net_device *netdev)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	u8 addr[ETH_ALEN];
+
+	mlx5_query_mac_address(priv->mdev, addr);
+	if (is_zero_ether_addr(addr) &&
+	    !MLX5_CAP_GEN(priv->mdev, vport_group_manager)) {
+		eth_hw_addr_random(netdev);
+		mlx5_core_info(priv->mdev, "Assigned random MAC address %pM\n", netdev->dev_addr);
+		return;
+	}
+
+	eth_hw_addr_set(netdev, addr);
+}
+
+static int mlx5e_vxlan_set_port(struct net_device *netdev, unsigned int table,
+				unsigned int entry, struct udp_tunnel_info *ti)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+
+	return mlx5_vxlan_add_port(priv->mdev->vxlan, ntohs(ti->port));
+}
+
+static int mlx5e_vxlan_unset_port(struct net_device *netdev, unsigned int table,
+				  unsigned int entry, struct udp_tunnel_info *ti)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+
+	return mlx5_vxlan_del_port(priv->mdev->vxlan, ntohs(ti->port));
+}
+
+void mlx5e_vxlan_set_netdev_info(struct mlx5e_priv *priv)
+{
+	if (!mlx5_vxlan_allowed(priv->mdev->vxlan))
+		return;
+
+	priv->nic_info.set_port = mlx5e_vxlan_set_port;
+	priv->nic_info.unset_port = mlx5e_vxlan_unset_port;
+	priv->nic_info.flags = UDP_TUNNEL_NIC_INFO_MAY_SLEEP |
+				UDP_TUNNEL_NIC_INFO_STATIC_IANA_VXLAN;
+	priv->nic_info.tables[0].tunnel_types = UDP_TUNNEL_TYPE_VXLAN;
+	/* Don't count the space hard-coded to the IANA port */
+	priv->nic_info.tables[0].n_entries =
+		mlx5_vxlan_max_udp_ports(priv->mdev) - 1;
+
+	priv->netdev->udp_tunnel_nic_info = &priv->nic_info;
+}
+
+static bool mlx5e_tunnel_any_tx_proto_supported(struct mlx5_core_dev *mdev)
+{
+	int tt;
+
+	for (tt = 0; tt < MLX5_NUM_TUNNEL_TT; tt++) {
+		if (mlx5e_tunnel_proto_supported_tx(mdev, mlx5_get_proto_by_tunnel_type(tt)))
+			return true;
+	}
+	return (mlx5_vxlan_allowed(mdev->vxlan) || mlx5_geneve_tx_allowed(mdev));
+}
+
+static void mlx5e_build_nic_netdev(struct net_device *netdev)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+	bool fcs_supported;
+	bool fcs_enabled;
+
+	SET_NETDEV_DEV(netdev, mdev->device);
+
+	netdev->netdev_ops = &mlx5e_netdev_ops;
+
+	mlx5e_dcbnl_build_netdev(netdev);
+
+	netdev->watchdog_timeo    = 15 * HZ;
+
+	netdev->ethtool_ops	  = &mlx5e_ethtool_ops;
+
+	netdev->vlan_features    |= NETIF_F_SG;
+	netdev->vlan_features    |= NETIF_F_HW_CSUM;
+	netdev->vlan_features    |= NETIF_F_GRO;
+	netdev->vlan_features    |= NETIF_F_TSO;
+	netdev->vlan_features    |= NETIF_F_TSO6;
+	netdev->vlan_features    |= NETIF_F_RXCSUM;
+	netdev->vlan_features    |= NETIF_F_RXHASH;
+	netdev->vlan_features    |= NETIF_F_GSO_PARTIAL;
+
+	netdev->mpls_features    |= NETIF_F_SG;
+	netdev->mpls_features    |= NETIF_F_HW_CSUM;
+	netdev->mpls_features    |= NETIF_F_TSO;
+	netdev->mpls_features    |= NETIF_F_TSO6;
+
+	netdev->hw_enc_features  |= NETIF_F_HW_VLAN_CTAG_TX;
+	netdev->hw_enc_features  |= NETIF_F_HW_VLAN_CTAG_RX;
+
+	/* Tunneled LRO is not supported in the driver, and the same RQs are
+	 * shared between inner and outer TIRs, so the driver can't disable LRO
+	 * for inner TIRs while having it enabled for outer TIRs. Due to this,
+	 * block LRO altogether if the firmware declares tunneled LRO support.
+	 */
+	if (!!MLX5_CAP_ETH(mdev, lro_cap) &&
+	    !MLX5_CAP_ETH(mdev, tunnel_lro_vxlan) &&
+	    !MLX5_CAP_ETH(mdev, tunnel_lro_gre) &&
+	    mlx5e_check_fragmented_striding_rq_cap(mdev, PAGE_SHIFT,
+						   MLX5E_MPWRQ_UMR_MODE_ALIGNED))
+		netdev->vlan_features    |= NETIF_F_LRO;
+
+	netdev->hw_features       = netdev->vlan_features;
+	netdev->hw_features      |= NETIF_F_HW_VLAN_CTAG_TX;
+	netdev->hw_features      |= NETIF_F_HW_VLAN_CTAG_RX;
+	netdev->hw_features      |= NETIF_F_HW_VLAN_CTAG_FILTER;
+	netdev->hw_features      |= NETIF_F_HW_VLAN_STAG_TX;
+
+	if (mlx5e_tunnel_any_tx_proto_supported(mdev)) {
+		netdev->hw_enc_features |= NETIF_F_HW_CSUM;
+		netdev->hw_enc_features |= NETIF_F_TSO;
+		netdev->hw_enc_features |= NETIF_F_TSO6;
+		netdev->hw_enc_features |= NETIF_F_GSO_PARTIAL;
+	}
+
+	if (mlx5_vxlan_allowed(mdev->vxlan) || mlx5_geneve_tx_allowed(mdev)) {
+		netdev->hw_features     |= NETIF_F_GSO_UDP_TUNNEL |
+					   NETIF_F_GSO_UDP_TUNNEL_CSUM;
+		netdev->hw_enc_features |= NETIF_F_GSO_UDP_TUNNEL |
+					   NETIF_F_GSO_UDP_TUNNEL_CSUM;
+		netdev->gso_partial_features = NETIF_F_GSO_UDP_TUNNEL_CSUM;
+		netdev->vlan_features |= NETIF_F_GSO_UDP_TUNNEL |
+					 NETIF_F_GSO_UDP_TUNNEL_CSUM;
+	}
+
+	if (mlx5e_tunnel_proto_supported_tx(mdev, IPPROTO_GRE)) {
+		netdev->hw_features     |= NETIF_F_GSO_GRE |
+					   NETIF_F_GSO_GRE_CSUM;
+		netdev->hw_enc_features |= NETIF_F_GSO_GRE |
+					   NETIF_F_GSO_GRE_CSUM;
+		netdev->gso_partial_features |= NETIF_F_GSO_GRE |
+						NETIF_F_GSO_GRE_CSUM;
+	}
+
+	if (mlx5e_tunnel_proto_supported_tx(mdev, IPPROTO_IPIP)) {
+		netdev->hw_features |= NETIF_F_GSO_IPXIP4 |
+				       NETIF_F_GSO_IPXIP6;
+		netdev->hw_enc_features |= NETIF_F_GSO_IPXIP4 |
+					   NETIF_F_GSO_IPXIP6;
+		netdev->gso_partial_features |= NETIF_F_GSO_IPXIP4 |
+						NETIF_F_GSO_IPXIP6;
+	}
+
+	netdev->gso_partial_features             |= NETIF_F_GSO_UDP_L4;
+	netdev->hw_features                      |= NETIF_F_GSO_UDP_L4;
+	netdev->features                         |= NETIF_F_GSO_UDP_L4;
+
+	mlx5_query_port_fcs(mdev, &fcs_supported, &fcs_enabled);
+
+	if (fcs_supported)
+		netdev->hw_features |= NETIF_F_RXALL;
+
+	if (MLX5_CAP_ETH(mdev, scatter_fcs))
+		netdev->hw_features |= NETIF_F_RXFCS;
+
+	if (mlx5_qos_is_supported(mdev))
+		netdev->hw_features |= NETIF_F_HW_TC;
+
+	netdev->features          = netdev->hw_features;
+
+	/* Defaults */
+	if (fcs_enabled)
+		netdev->features  &= ~NETIF_F_RXALL;
+	netdev->features  &= ~NETIF_F_LRO;
+	netdev->features  &= ~NETIF_F_GRO_HW;
+	netdev->features  &= ~NETIF_F_RXFCS;
+
+#define FT_CAP(f) MLX5_CAP_FLOWTABLE(mdev, flow_table_properties_nic_receive.f)
+	if (FT_CAP(flow_modify_en) &&
+	    FT_CAP(modify_root) &&
+	    FT_CAP(identified_miss_table_mode) &&
+	    FT_CAP(flow_table_modify)) {
+#if IS_ENABLED(CONFIG_MLX5_CLS_ACT)
+		netdev->hw_features      |= NETIF_F_HW_TC;
+#endif
+#ifdef CONFIG_MLX5_EN_ARFS
+		netdev->hw_features	 |= NETIF_F_NTUPLE;
+#endif
+	}
+
+	netdev->features         |= NETIF_F_HIGHDMA;
+	netdev->features         |= NETIF_F_HW_VLAN_STAG_FILTER;
+
+	netdev->priv_flags       |= IFF_UNICAST_FLT;
+
+	netif_set_tso_max_size(netdev, GSO_MAX_SIZE);
+	mlx5e_set_netdev_dev_addr(netdev);
+	mlx5e_macsec_build_netdev(priv);
+	mlx5e_ipsec_build_netdev(priv);
+	mlx5e_ktls_build_netdev(priv);
+}
+
+void mlx5e_create_q_counters(struct mlx5e_priv *priv)
+{
+	u32 out[MLX5_ST_SZ_DW(alloc_q_counter_out)] = {};
+	u32 in[MLX5_ST_SZ_DW(alloc_q_counter_in)] = {};
+	struct mlx5_core_dev *mdev = priv->mdev;
+	int err;
+
+	MLX5_SET(alloc_q_counter_in, in, opcode, MLX5_CMD_OP_ALLOC_Q_COUNTER);
+	err = mlx5_cmd_exec_inout(mdev, alloc_q_counter, in, out);
+	if (!err)
+		priv->q_counter =
+			MLX5_GET(alloc_q_counter_out, out, counter_set_id);
+
+	err = mlx5_cmd_exec_inout(mdev, alloc_q_counter, in, out);
+	if (!err)
+		priv->drop_rq_q_counter =
+			MLX5_GET(alloc_q_counter_out, out, counter_set_id);
+}
+
+void mlx5e_destroy_q_counters(struct mlx5e_priv *priv)
+{
+	u32 in[MLX5_ST_SZ_DW(dealloc_q_counter_in)] = {};
+
+	MLX5_SET(dealloc_q_counter_in, in, opcode,
+		 MLX5_CMD_OP_DEALLOC_Q_COUNTER);
+	if (priv->q_counter) {
+		MLX5_SET(dealloc_q_counter_in, in, counter_set_id,
+			 priv->q_counter);
+		mlx5_cmd_exec_in(priv->mdev, dealloc_q_counter, in);
+	}
+
+	if (priv->drop_rq_q_counter) {
+		MLX5_SET(dealloc_q_counter_in, in, counter_set_id,
+			 priv->drop_rq_q_counter);
+		mlx5_cmd_exec_in(priv->mdev, dealloc_q_counter, in);
+	}
+}
+
+static int mlx5e_nic_init(struct mlx5_core_dev *mdev,
+			  struct net_device *netdev)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	struct mlx5e_flow_steering *fs;
+	int err;
+
+	mlx5e_build_nic_params(priv, &priv->xsk, netdev->mtu);
+	mlx5e_vxlan_set_netdev_info(priv);
+
+	mlx5e_timestamp_init(priv);
+
+	fs = mlx5e_fs_init(priv->profile, mdev,
+			   !test_bit(MLX5E_STATE_DESTROYING, &priv->state));
+	if (!fs) {
+		err = -ENOMEM;
+		mlx5_core_err(mdev, "FS initialization failed, %d\n", err);
+		return err;
+	}
+	priv->fs = fs;
+
+	err = mlx5e_ipsec_init(priv);
+	if (err)
+		mlx5_core_err(mdev, "IPSec initialization failed, %d\n", err);
+
+	err = mlx5e_ktls_init(priv);
+	if (err)
+		mlx5_core_err(mdev, "TLS initialization failed, %d\n", err);
+
+	mlx5e_health_create_reporters(priv);
+	return 0;
+}
+
+static void mlx5e_nic_cleanup(struct mlx5e_priv *priv)
+{
+	mlx5e_health_destroy_reporters(priv);
+	mlx5e_ktls_cleanup(priv);
+	mlx5e_ipsec_cleanup(priv);
+	mlx5e_fs_cleanup(priv->fs);
+	priv->fs = NULL;
+}
+
+static int mlx5e_init_nic_rx(struct mlx5e_priv *priv)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+	enum mlx5e_rx_res_features features;
+	int err;
+
+	priv->rx_res = mlx5e_rx_res_alloc();
+	if (!priv->rx_res)
+		return -ENOMEM;
+
+	mlx5e_create_q_counters(priv);
+
+	err = mlx5e_open_drop_rq(priv, &priv->drop_rq);
+	if (err) {
+		mlx5_core_err(mdev, "open drop rq failed, %d\n", err);
+		goto err_destroy_q_counters;
+	}
+
+	features = MLX5E_RX_RES_FEATURE_PTP;
+	if (mlx5_tunnel_inner_ft_supported(mdev))
+		features |= MLX5E_RX_RES_FEATURE_INNER_FT;
+	err = mlx5e_rx_res_init(priv->rx_res, priv->mdev, features,
+				priv->max_nch, priv->drop_rq.rqn,
+				&priv->channels.params.packet_merge,
+				priv->channels.params.num_channels);
+	if (err)
+		goto err_close_drop_rq;
+
+	err = mlx5e_create_flow_steering(priv->fs, priv->rx_res, priv->profile,
+					 priv->netdev);
+	if (err) {
+		mlx5_core_warn(mdev, "create flow steering failed, %d\n", err);
+		goto err_destroy_rx_res;
+	}
+
+	err = mlx5e_tc_nic_init(priv);
+	if (err)
+		goto err_destroy_flow_steering;
+
+	err = mlx5e_accel_init_rx(priv);
+	if (err)
+		goto err_tc_nic_cleanup;
+
+#ifdef CONFIG_MLX5_EN_ARFS
+	priv->netdev->rx_cpu_rmap =  mlx5_eq_table_get_rmap(priv->mdev);
+#endif
+
+	return 0;
+
+err_tc_nic_cleanup:
+	mlx5e_tc_nic_cleanup(priv);
+err_destroy_flow_steering:
+	mlx5e_destroy_flow_steering(priv->fs, !!(priv->netdev->hw_features & NETIF_F_NTUPLE),
+				    priv->profile);
+err_destroy_rx_res:
+	mlx5e_rx_res_destroy(priv->rx_res);
+err_close_drop_rq:
+	mlx5e_close_drop_rq(&priv->drop_rq);
+err_destroy_q_counters:
+	mlx5e_destroy_q_counters(priv);
+	mlx5e_rx_res_free(priv->rx_res);
+	priv->rx_res = NULL;
+	return err;
+}
+
+static void mlx5e_cleanup_nic_rx(struct mlx5e_priv *priv)
+{
+	mlx5e_accel_cleanup_rx(priv);
+	mlx5e_tc_nic_cleanup(priv);
+	mlx5e_destroy_flow_steering(priv->fs, !!(priv->netdev->hw_features & NETIF_F_NTUPLE),
+				    priv->profile);
+	mlx5e_rx_res_destroy(priv->rx_res);
+	mlx5e_close_drop_rq(&priv->drop_rq);
+	mlx5e_destroy_q_counters(priv);
+	mlx5e_rx_res_free(priv->rx_res);
+	priv->rx_res = NULL;
+}
+
+static void mlx5e_set_mqprio_rl(struct mlx5e_priv *priv)
+{
+	struct mlx5e_params *params;
+	struct mlx5e_mqprio_rl *rl;
+
+	params = &priv->channels.params;
+	if (params->mqprio.mode != TC_MQPRIO_MODE_CHANNEL)
+		return;
+
+	rl = mlx5e_mqprio_rl_create(priv->mdev, params->mqprio.num_tc,
+				    params->mqprio.channel.max_rate);
+	if (IS_ERR(rl))
+		rl = NULL;
+	priv->mqprio_rl = rl;
+	mlx5e_mqprio_rl_update_params(params, rl);
+}
+
+static int mlx5e_init_nic_tx(struct mlx5e_priv *priv)
+{
+	int err;
+
+	err = mlx5e_create_tises(priv);
+	if (err) {
+		mlx5_core_warn(priv->mdev, "create tises failed, %d\n", err);
+		return err;
+	}
+
+	err = mlx5e_accel_init_tx(priv);
+	if (err)
+		goto err_destroy_tises;
+
+	mlx5e_set_mqprio_rl(priv);
+	mlx5e_dcbnl_initialize(priv);
+	return 0;
+
+err_destroy_tises:
+	mlx5e_destroy_tises(priv);
+	return err;
+}
+
+static void mlx5e_nic_enable(struct mlx5e_priv *priv)
+{
+	struct net_device *netdev = priv->netdev;
+	struct mlx5_core_dev *mdev = priv->mdev;
+	int err;
+
+	mlx5e_fs_init_l2_addr(priv->fs, netdev);
+
+	err = mlx5e_macsec_init(priv);
+	if (err)
+		mlx5_core_err(mdev, "MACsec initialization failed, %d\n", err);
+
+	/* Marking the link as currently not needed by the Driver */
+	if (!netif_running(netdev))
+		mlx5e_modify_admin_state(mdev, MLX5_PORT_DOWN);
+
+	mlx5e_set_netdev_mtu_boundaries(priv);
+	mlx5e_set_dev_port_mtu(priv);
+
+	mlx5_lag_add_netdev(mdev, netdev);
+
+	mlx5e_enable_async_events(priv);
+	mlx5e_enable_blocking_events(priv);
+	if (mlx5e_monitor_counter_supported(priv))
+		mlx5e_monitor_counter_init(priv);
+
+	mlx5e_hv_vhca_stats_create(priv);
+	if (netdev->reg_state != NETREG_REGISTERED)
+		return;
+	mlx5e_dcbnl_init_app(priv);
+
+	mlx5e_nic_set_rx_mode(priv);
+
+	rtnl_lock();
+	if (netif_running(netdev))
+		mlx5e_open(netdev);
+	udp_tunnel_nic_reset_ntf(priv->netdev);
+	netif_device_attach(netdev);
+	rtnl_unlock();
+}
+
+static void mlx5e_nic_disable(struct mlx5e_priv *priv)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+
+	if (priv->netdev->reg_state == NETREG_REGISTERED)
+		mlx5e_dcbnl_delete_app(priv);
+
+	rtnl_lock();
+	if (netif_running(priv->netdev))
+		mlx5e_close(priv->netdev);
+	netif_device_detach(priv->netdev);
+	rtnl_unlock();
+
+	mlx5e_nic_set_rx_mode(priv);
+
+	mlx5e_hv_vhca_stats_destroy(priv);
+	if (mlx5e_monitor_counter_supported(priv))
+		mlx5e_monitor_counter_cleanup(priv);
+
+	mlx5e_disable_blocking_events(priv);
+	if (priv->en_trap) {
+		mlx5e_deactivate_trap(priv);
+		mlx5e_close_trap(priv->en_trap);
+		priv->en_trap = NULL;
+	}
+	mlx5e_disable_async_events(priv);
+	mlx5_lag_remove_netdev(mdev, priv->netdev);
+	mlx5_vxlan_reset_to_default(mdev->vxlan);
+	mlx5e_macsec_cleanup(priv);
+}
+
+int mlx5e_update_nic_rx(struct mlx5e_priv *priv)
+{
+	return mlx5e_refresh_tirs(priv, false, false);
+}
+
+static const struct mlx5e_profile mlx5e_nic_profile = {
+	.init		   = mlx5e_nic_init,
+	.cleanup	   = mlx5e_nic_cleanup,
+	.init_rx	   = mlx5e_init_nic_rx,
+	.cleanup_rx	   = mlx5e_cleanup_nic_rx,
+	.init_tx	   = mlx5e_init_nic_tx,
+	.cleanup_tx	   = mlx5e_cleanup_nic_tx,
+	.enable		   = mlx5e_nic_enable,
+	.disable	   = mlx5e_nic_disable,
+	.update_rx	   = mlx5e_update_nic_rx,
+	.update_stats	   = mlx5e_stats_update_ndo_stats,
+	.update_carrier	   = mlx5e_update_carrier,
+	.rx_handlers       = &mlx5e_rx_handlers_nic,
+	.max_tc		   = MLX5E_MAX_NUM_TC,
+	.stats_grps	   = mlx5e_nic_stats_grps,
+	.stats_grps_num	   = mlx5e_nic_stats_grps_num,
+	.features          = BIT(MLX5E_PROFILE_FEATURE_PTP_RX) |
+		BIT(MLX5E_PROFILE_FEATURE_PTP_TX) |
+		BIT(MLX5E_PROFILE_FEATURE_QOS_HTB) |
+		BIT(MLX5E_PROFILE_FEATURE_FS_VLAN) |
+		BIT(MLX5E_PROFILE_FEATURE_FS_TC),
+};
+
+static int mlx5e_profile_max_num_channels(struct mlx5_core_dev *mdev,
+					  const struct mlx5e_profile *profile)
+{
+	int nch;
+
+	nch = mlx5e_get_max_num_channels(mdev);
+
+	if (profile->max_nch_limit)
+		nch = min_t(int, nch, profile->max_nch_limit(mdev));
+	return nch;
+}
+
+static unsigned int
+mlx5e_calc_max_nch(struct mlx5_core_dev *mdev, struct net_device *netdev,
+		   const struct mlx5e_profile *profile)
+
+{
+	unsigned int max_nch, tmp;
+
+	/* core resources */
+	max_nch = mlx5e_profile_max_num_channels(mdev, profile);
+
+	/* netdev rx queues */
+	max_nch = min_t(unsigned int, max_nch, netdev->num_rx_queues);
+
+	/* netdev tx queues */
+	tmp = netdev->num_tx_queues;
+	if (mlx5_qos_is_supported(mdev))
+		tmp -= mlx5e_qos_max_leaf_nodes(mdev);
+	if (MLX5_CAP_GEN(mdev, ts_cqe_to_dest_cqn))
+		tmp -= profile->max_tc;
+	tmp = tmp / profile->max_tc;
+	max_nch = min_t(unsigned int, max_nch, tmp);
+
+	return max_nch;
+}
+
+int mlx5e_get_pf_num_tirs(struct mlx5_core_dev *mdev)
+{
+	/* Indirect TIRS: 2 sets of TTCs (inner + outer steering)
+	 * and 1 set of direct TIRS
+	 */
+	return 2 * MLX5E_NUM_INDIR_TIRS
+		+ mlx5e_profile_max_num_channels(mdev, &mlx5e_nic_profile);
+}
+
+void mlx5e_set_rx_mode_work(struct work_struct *work)
+{
+	struct mlx5e_priv *priv = container_of(work, struct mlx5e_priv,
+					       set_rx_mode_work);
+
+	return mlx5e_fs_set_rx_mode_work(priv->fs, priv->netdev);
+}
+
+/* mlx5e generic netdev management API (move to en_common.c) */
+int mlx5e_priv_init(struct mlx5e_priv *priv,
+		    const struct mlx5e_profile *profile,
+		    struct net_device *netdev,
+		    struct mlx5_core_dev *mdev)
+{
+	int nch, num_txqs, node;
+	int err;
+
+	num_txqs = netdev->num_tx_queues;
+	nch = mlx5e_calc_max_nch(mdev, netdev, profile);
+	node = dev_to_node(mlx5_core_dma_dev(mdev));
+
+	/* priv init */
+	priv->mdev        = mdev;
+	priv->netdev      = netdev;
+	priv->msglevel    = MLX5E_MSG_LEVEL;
+	priv->max_nch     = nch;
+	priv->max_opened_tc = 1;
+
+	if (!alloc_cpumask_var(&priv->scratchpad.cpumask, GFP_KERNEL))
+		return -ENOMEM;
+
+	mutex_init(&priv->state_lock);
+
+	err = mlx5e_selq_init(&priv->selq, &priv->state_lock);
+	if (err)
+		goto err_free_cpumask;
+
+	INIT_WORK(&priv->update_carrier_work, mlx5e_update_carrier_work);
+	INIT_WORK(&priv->set_rx_mode_work, mlx5e_set_rx_mode_work);
+	INIT_WORK(&priv->tx_timeout_work, mlx5e_tx_timeout_work);
+	INIT_WORK(&priv->update_stats_work, mlx5e_update_stats_work);
+
+	priv->wq = create_singlethread_workqueue("mlx5e");
+	if (!priv->wq)
+		goto err_free_selq;
+
+	priv->txq2sq = kcalloc_node(num_txqs, sizeof(*priv->txq2sq), GFP_KERNEL, node);
+	if (!priv->txq2sq)
+		goto err_destroy_workqueue;
+
+	priv->tx_rates = kcalloc_node(num_txqs, sizeof(*priv->tx_rates), GFP_KERNEL, node);
+	if (!priv->tx_rates)
+		goto err_free_txq2sq;
+
+	priv->channel_stats =
+		kcalloc_node(nch, sizeof(*priv->channel_stats), GFP_KERNEL, node);
+	if (!priv->channel_stats)
+		goto err_free_tx_rates;
+
+	return 0;
+
+err_free_tx_rates:
+	kfree(priv->tx_rates);
+err_free_txq2sq:
+	kfree(priv->txq2sq);
+err_destroy_workqueue:
+	destroy_workqueue(priv->wq);
+err_free_selq:
+	mlx5e_selq_cleanup(&priv->selq);
+err_free_cpumask:
+	free_cpumask_var(priv->scratchpad.cpumask);
+	return -ENOMEM;
+}
+
+void mlx5e_priv_cleanup(struct mlx5e_priv *priv)
+{
+	int i;
+
+	/* bail if change profile failed and also rollback failed */
+	if (!priv->mdev)
+		return;
+
+	for (i = 0; i < priv->stats_nch; i++)
+		kvfree(priv->channel_stats[i]);
+	kfree(priv->channel_stats);
+	kfree(priv->tx_rates);
+	kfree(priv->txq2sq);
+	destroy_workqueue(priv->wq);
+	mutex_lock(&priv->state_lock);
+	mlx5e_selq_cleanup(&priv->selq);
+	mutex_unlock(&priv->state_lock);
+	free_cpumask_var(priv->scratchpad.cpumask);
+
+	for (i = 0; i < priv->htb_max_qos_sqs; i++)
+		kfree(priv->htb_qos_sq_stats[i]);
+	kvfree(priv->htb_qos_sq_stats);
+
+	memset(priv, 0, sizeof(*priv));
+}
+
+static unsigned int mlx5e_get_max_num_txqs(struct mlx5_core_dev *mdev,
+					   const struct mlx5e_profile *profile)
+{
+	unsigned int nch, ptp_txqs, qos_txqs;
+
+	nch = mlx5e_profile_max_num_channels(mdev, profile);
+
+	ptp_txqs = MLX5_CAP_GEN(mdev, ts_cqe_to_dest_cqn) &&
+		mlx5e_profile_feature_cap(profile, PTP_TX) ?
+		profile->max_tc : 0;
+
+	qos_txqs = mlx5_qos_is_supported(mdev) &&
+		mlx5e_profile_feature_cap(profile, QOS_HTB) ?
+		mlx5e_qos_max_leaf_nodes(mdev) : 0;
+
+	return nch * profile->max_tc + ptp_txqs + qos_txqs;
+}
+
+static unsigned int mlx5e_get_max_num_rxqs(struct mlx5_core_dev *mdev,
+					   const struct mlx5e_profile *profile)
+{
+	return mlx5e_profile_max_num_channels(mdev, profile);
+}
+
+struct net_device *
+mlx5e_create_netdev(struct mlx5_core_dev *mdev, const struct mlx5e_profile *profile)
+{
+	struct net_device *netdev;
+	unsigned int txqs, rxqs;
+	int err;
+
+	txqs = mlx5e_get_max_num_txqs(mdev, profile);
+	rxqs = mlx5e_get_max_num_rxqs(mdev, profile);
+
+	netdev = alloc_etherdev_mqs(sizeof(struct mlx5e_priv), txqs, rxqs);
+	if (!netdev) {
+		mlx5_core_err(mdev, "alloc_etherdev_mqs() failed\n");
+		return NULL;
+	}
+
+	err = mlx5e_priv_init(netdev_priv(netdev), profile, netdev, mdev);
+	if (err) {
+		mlx5_core_err(mdev, "mlx5e_priv_init failed, err=%d\n", err);
+		goto err_free_netdev;
+	}
+
+	netif_carrier_off(netdev);
+	netif_tx_disable(netdev);
+	dev_net_set(netdev, mlx5_core_net(mdev));
+
+	return netdev;
+
+err_free_netdev:
+	free_netdev(netdev);
+
+	return NULL;
+}
+
+static void mlx5e_update_features(struct net_device *netdev)
+{
+	if (netdev->reg_state != NETREG_REGISTERED)
+		return; /* features will be updated on netdev registration */
+
+	rtnl_lock();
+	netdev_update_features(netdev);
+	rtnl_unlock();
+}
+
+static void mlx5e_reset_channels(struct net_device *netdev)
+{
+	netdev_reset_tc(netdev);
+}
+
+int mlx5e_attach_netdev(struct mlx5e_priv *priv)
+{
+	const bool take_rtnl = priv->netdev->reg_state == NETREG_REGISTERED;
+	const struct mlx5e_profile *profile = priv->profile;
+	int max_nch;
+	int err;
+
+	clear_bit(MLX5E_STATE_DESTROYING, &priv->state);
+	if (priv->fs)
+		mlx5e_fs_set_state_destroy(priv->fs,
+					   !test_bit(MLX5E_STATE_DESTROYING, &priv->state));
+
+	/* Validate the max_wqe_size_sq capability. */
+	if (WARN_ON_ONCE(mlx5e_get_max_sq_wqebbs(priv->mdev) < MLX5E_MAX_TX_WQEBBS)) {
+		mlx5_core_warn(priv->mdev, "MLX5E: Max SQ WQEBBs firmware capability: %u, needed %lu\n",
+			       mlx5e_get_max_sq_wqebbs(priv->mdev), MLX5E_MAX_TX_WQEBBS);
+		return -EIO;
+	}
+
+	/* max number of channels may have changed */
+	max_nch = mlx5e_calc_max_nch(priv->mdev, priv->netdev, profile);
+	if (priv->channels.params.num_channels > max_nch) {
+		mlx5_core_warn(priv->mdev, "MLX5E: Reducing number of channels to %d\n", max_nch);
+		/* Reducing the number of channels - RXFH has to be reset, and
+		 * mlx5e_num_channels_changed below will build the RQT.
+		 */
+		priv->netdev->priv_flags &= ~IFF_RXFH_CONFIGURED;
+		priv->channels.params.num_channels = max_nch;
+		if (priv->channels.params.mqprio.mode == TC_MQPRIO_MODE_CHANNEL) {
+			mlx5_core_warn(priv->mdev, "MLX5E: Disabling MQPRIO channel mode\n");
+			mlx5e_params_mqprio_reset(&priv->channels.params);
+		}
+	}
+	if (max_nch != priv->max_nch) {
+		mlx5_core_warn(priv->mdev,
+			       "MLX5E: Updating max number of channels from %u to %u\n",
+			       priv->max_nch, max_nch);
+		priv->max_nch = max_nch;
+	}
+
+	/* 1. Set the real number of queues in the kernel the first time.
+	 * 2. Set our default XPS cpumask.
+	 * 3. Build the RQT.
+	 *
+	 * rtnl_lock is required by netif_set_real_num_*_queues in case the
+	 * netdev has been registered by this point (if this function was called
+	 * in the reload or resume flow).
+	 */
+	if (take_rtnl)
+		rtnl_lock();
+	err = mlx5e_num_channels_changed(priv);
+	if (take_rtnl)
+		rtnl_unlock();
+	if (err)
+		goto out;
+
+	err = profile->init_tx(priv);
+	if (err)
+		goto out;
+
+	err = profile->init_rx(priv);
+	if (err)
+		goto err_cleanup_tx;
+
+	if (profile->enable)
+		profile->enable(priv);
+
+	mlx5e_update_features(priv->netdev);
+
+	return 0;
+
+err_cleanup_tx:
+	profile->cleanup_tx(priv);
+
+out:
+	mlx5e_reset_channels(priv->netdev);
+	set_bit(MLX5E_STATE_DESTROYING, &priv->state);
+	if (priv->fs)
+		mlx5e_fs_set_state_destroy(priv->fs,
+					   !test_bit(MLX5E_STATE_DESTROYING, &priv->state));
+	cancel_work_sync(&priv->update_stats_work);
+	return err;
+}
+
+void mlx5e_detach_netdev(struct mlx5e_priv *priv)
+{
+	const struct mlx5e_profile *profile = priv->profile;
+
+	set_bit(MLX5E_STATE_DESTROYING, &priv->state);
+	if (priv->fs)
+		mlx5e_fs_set_state_destroy(priv->fs,
+					   !test_bit(MLX5E_STATE_DESTROYING, &priv->state));
+
+	if (profile->disable)
+		profile->disable(priv);
+	flush_workqueue(priv->wq);
+
+	profile->cleanup_rx(priv);
+	profile->cleanup_tx(priv);
+	mlx5e_reset_channels(priv->netdev);
+	cancel_work_sync(&priv->update_stats_work);
+}
+
+static int
+mlx5e_netdev_init_profile(struct net_device *netdev, struct mlx5_core_dev *mdev,
+			  const struct mlx5e_profile *new_profile, void *new_ppriv)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	int err;
+
+	err = mlx5e_priv_init(priv, new_profile, netdev, mdev);
+	if (err) {
+		mlx5_core_err(mdev, "mlx5e_priv_init failed, err=%d\n", err);
+		return err;
+	}
+	netif_carrier_off(netdev);
+	priv->profile = new_profile;
+	priv->ppriv = new_ppriv;
+	err = new_profile->init(priv->mdev, priv->netdev);
+	if (err)
+		goto priv_cleanup;
+
+	return 0;
+
+priv_cleanup:
+	mlx5e_priv_cleanup(priv);
+	return err;
+}
+
+static int
+mlx5e_netdev_attach_profile(struct net_device *netdev, struct mlx5_core_dev *mdev,
+			    const struct mlx5e_profile *new_profile, void *new_ppriv)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	int err;
+
+	err = mlx5e_netdev_init_profile(netdev, mdev, new_profile, new_ppriv);
+	if (err)
+		return err;
+
+	err = mlx5e_attach_netdev(priv);
+	if (err)
+		goto profile_cleanup;
+	return err;
+
+profile_cleanup:
+	new_profile->cleanup(priv);
+	mlx5e_priv_cleanup(priv);
+	return err;
+}
+
+int mlx5e_netdev_change_profile(struct mlx5e_priv *priv,
+				const struct mlx5e_profile *new_profile, void *new_ppriv)
+{
+	const struct mlx5e_profile *orig_profile = priv->profile;
+	struct net_device *netdev = priv->netdev;
+	struct mlx5_core_dev *mdev = priv->mdev;
+	void *orig_ppriv = priv->ppriv;
+	int err, rollback_err;
+
+	/* cleanup old profile */
+	mlx5e_detach_netdev(priv);
+	priv->profile->cleanup(priv);
+	mlx5e_priv_cleanup(priv);
+
+	if (mdev->state == MLX5_DEVICE_STATE_INTERNAL_ERROR) {
+		mlx5e_netdev_init_profile(netdev, mdev, new_profile, new_ppriv);
+		set_bit(MLX5E_STATE_DESTROYING, &priv->state);
+		return -EIO;
+	}
+
+	err = mlx5e_netdev_attach_profile(netdev, mdev, new_profile, new_ppriv);
+	if (err) { /* roll back to original profile */
+		netdev_warn(netdev, "%s: new profile init failed, %d\n", __func__, err);
+		goto rollback;
+	}
+
+	return 0;
+
+rollback:
+	rollback_err = mlx5e_netdev_attach_profile(netdev, mdev, orig_profile, orig_ppriv);
+	if (rollback_err)
+		netdev_err(netdev, "%s: failed to rollback to orig profile, %d\n",
+			   __func__, rollback_err);
+	return err;
+}
+
+void mlx5e_netdev_attach_nic_profile(struct mlx5e_priv *priv)
+{
+	mlx5e_netdev_change_profile(priv, &mlx5e_nic_profile, NULL);
+}
+
+void mlx5e_destroy_netdev(struct mlx5e_priv *priv)
+{
+	struct net_device *netdev = priv->netdev;
+
+	mlx5e_priv_cleanup(priv);
+	free_netdev(netdev);
+}
+
+static int mlx5e_resume(struct auxiliary_device *adev)
+{
+	struct mlx5_adev *edev = container_of(adev, struct mlx5_adev, adev);
+	struct mlx5e_priv *priv = auxiliary_get_drvdata(adev);
+	struct net_device *netdev = priv->netdev;
+	struct mlx5_core_dev *mdev = edev->mdev;
+	int err;
+
+	if (netif_device_present(netdev))
+		return 0;
+
+	err = mlx5e_create_mdev_resources(mdev);
+	if (err)
+		return err;
+
+	err = mlx5e_attach_netdev(priv);
+	if (err) {
+		mlx5e_destroy_mdev_resources(mdev);
+		return err;
+	}
+
+	return 0;
+}
+
+static int mlx5e_suspend(struct auxiliary_device *adev, pm_message_t state)
+{
+	struct mlx5e_priv *priv = auxiliary_get_drvdata(adev);
+	struct net_device *netdev = priv->netdev;
+	struct mlx5_core_dev *mdev = priv->mdev;
+
+	if (!netif_device_present(netdev)) {
+		if (test_bit(MLX5E_STATE_DESTROYING, &priv->state))
+			mlx5e_destroy_mdev_resources(mdev);
+		return -ENODEV;
+	}
+
+	mlx5e_detach_netdev(priv);
+	mlx5e_destroy_mdev_resources(mdev);
+	return 0;
+}
+
+static int mlx5e_probe(struct auxiliary_device *adev,
+		       const struct auxiliary_device_id *id)
+{
+	struct mlx5_adev *edev = container_of(adev, struct mlx5_adev, adev);
+	const struct mlx5e_profile *profile = &mlx5e_nic_profile;
+	struct mlx5_core_dev *mdev = edev->mdev;
+	struct net_device *netdev;
+	pm_message_t state = {};
+	struct mlx5e_priv *priv;
+	int err;
+
+	netdev = mlx5e_create_netdev(mdev, profile);
+	if (!netdev) {
+		mlx5_core_err(mdev, "mlx5e_create_netdev failed\n");
+		return -ENOMEM;
+	}
+
+	mlx5e_build_nic_netdev(netdev);
+
+	priv = netdev_priv(netdev);
+	auxiliary_set_drvdata(adev, priv);
+
+	priv->profile = profile;
+	priv->ppriv = NULL;
+
+	err = mlx5e_devlink_port_register(priv);
+	if (err) {
+		mlx5_core_err(mdev, "mlx5e_devlink_port_register failed, %d\n", err);
+		goto err_destroy_netdev;
+	}
+
+	err = profile->init(mdev, netdev);
+	if (err) {
+		mlx5_core_err(mdev, "mlx5e_nic_profile init failed, %d\n", err);
+		goto err_devlink_cleanup;
+	}
+
+	err = mlx5e_resume(adev);
+	if (err) {
+		mlx5_core_err(mdev, "mlx5e_resume failed, %d\n", err);
+		goto err_profile_cleanup;
+	}
+
+	err = register_netdev(netdev);
+	if (err) {
+		mlx5_core_err(mdev, "register_netdev failed, %d\n", err);
+		goto err_resume;
+	}
+
+	mlx5e_devlink_port_type_eth_set(priv);
+
+	mlx5e_dcbnl_init_app(priv);
+	mlx5_uplink_netdev_set(mdev, netdev);
+	return 0;
+
+err_resume:
+	mlx5e_suspend(adev, state);
+err_profile_cleanup:
+	profile->cleanup(priv);
+err_devlink_cleanup:
+	mlx5e_devlink_port_unregister(priv);
+err_destroy_netdev:
+	mlx5e_destroy_netdev(priv);
+	return err;
+}
+
+static void mlx5e_remove(struct auxiliary_device *adev)
+{
+	struct mlx5e_priv *priv = auxiliary_get_drvdata(adev);
+	pm_message_t state = {};
+
+	mlx5e_dcbnl_delete_app(priv);
+	unregister_netdev(priv->netdev);
+	mlx5e_suspend(adev, state);
+	priv->profile->cleanup(priv);
+	mlx5e_devlink_port_unregister(priv);
+	mlx5e_destroy_netdev(priv);
+}
+
+static const struct auxiliary_device_id mlx5e_id_table[] = {
+	{ .name = MLX5_ADEV_NAME ".eth", },
+	{},
+};
+
+MODULE_DEVICE_TABLE(auxiliary, mlx5e_id_table);
+
+static struct auxiliary_driver mlx5e_driver = {
+	.name = "eth",
+	.probe = mlx5e_probe,
+	.remove = mlx5e_remove,
+	.suspend = mlx5e_suspend,
+	.resume = mlx5e_resume,
+	.id_table = mlx5e_id_table,
+};
+
+int mlx5e_init(void)
+{
+	int ret;
+
+	mlx5e_build_ptys2ethtool_map();
+	ret = auxiliary_driver_register(&mlx5e_driver);
+	if (ret)
+		return ret;
+
+	ret = mlx5e_rep_init();
+	if (ret)
+		auxiliary_driver_unregister(&mlx5e_driver);
+	return ret;
+}
+
+void mlx5e_cleanup(void)
+{
+	mlx5e_rep_cleanup();
+	auxiliary_driver_unregister(&mlx5e_driver);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c b/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c
new file mode 100644
index 000000000..5aeca9534
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c
@@ -0,0 +1,1521 @@
+/*
+ * Copyright (c) 2016, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/mlx5/fs.h>
+#include <net/switchdev.h>
+#include <net/pkt_cls.h>
+#include <net/act_api.h>
+#include <net/devlink.h>
+#include <net/ipv6_stubs.h>
+
+#include "eswitch.h"
+#include "en.h"
+#include "en_rep.h"
+#include "en/params.h"
+#include "en/txrx.h"
+#include "en_tc.h"
+#include "en/rep/tc.h"
+#include "en/rep/neigh.h"
+#include "en/rep/bridge.h"
+#include "en/devlink.h"
+#include "fs_core.h"
+#include "lib/mlx5.h"
+#include "lib/devcom.h"
+#include "lib/vxlan.h"
+#define CREATE_TRACE_POINTS
+#include "diag/en_rep_tracepoint.h"
+#include "en_accel/ipsec.h"
+#include "en/tc/int_port.h"
+#include "en/ptp.h"
+#include "en/fs_ethtool.h"
+
+#define MLX5E_REP_PARAMS_DEF_LOG_SQ_SIZE \
+	max(0x7, MLX5E_PARAMS_MINIMUM_LOG_SQ_SIZE)
+#define MLX5E_REP_PARAMS_DEF_NUM_CHANNELS 1
+
+static const char mlx5e_rep_driver_name[] = "mlx5e_rep";
+
+static void mlx5e_rep_get_drvinfo(struct net_device *dev,
+				  struct ethtool_drvinfo *drvinfo)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+	int count;
+
+	strscpy(drvinfo->driver, mlx5e_rep_driver_name,
+		sizeof(drvinfo->driver));
+	count = snprintf(drvinfo->fw_version, sizeof(drvinfo->fw_version),
+			 "%d.%d.%04d (%.16s)", fw_rev_maj(mdev),
+			 fw_rev_min(mdev), fw_rev_sub(mdev), mdev->board_id);
+	if (count >= sizeof(drvinfo->fw_version))
+		snprintf(drvinfo->fw_version, sizeof(drvinfo->fw_version),
+			 "%d.%d.%04d", fw_rev_maj(mdev),
+			 fw_rev_min(mdev), fw_rev_sub(mdev));
+}
+
+static const struct counter_desc sw_rep_stats_desc[] = {
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_packets) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_bytes) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_packets) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_bytes) },
+};
+
+struct vport_stats {
+	u64 vport_rx_packets;
+	u64 vport_tx_packets;
+	u64 vport_rx_bytes;
+	u64 vport_tx_bytes;
+};
+
+static const struct counter_desc vport_rep_stats_desc[] = {
+	{ MLX5E_DECLARE_STAT(struct vport_stats, vport_rx_packets) },
+	{ MLX5E_DECLARE_STAT(struct vport_stats, vport_rx_bytes) },
+	{ MLX5E_DECLARE_STAT(struct vport_stats, vport_tx_packets) },
+	{ MLX5E_DECLARE_STAT(struct vport_stats, vport_tx_bytes) },
+};
+
+#define NUM_VPORT_REP_SW_COUNTERS ARRAY_SIZE(sw_rep_stats_desc)
+#define NUM_VPORT_REP_HW_COUNTERS ARRAY_SIZE(vport_rep_stats_desc)
+
+static MLX5E_DECLARE_STATS_GRP_OP_NUM_STATS(sw_rep)
+{
+	return NUM_VPORT_REP_SW_COUNTERS;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_FILL_STRS(sw_rep)
+{
+	int i;
+
+	for (i = 0; i < NUM_VPORT_REP_SW_COUNTERS; i++)
+		strcpy(data + (idx++) * ETH_GSTRING_LEN,
+		       sw_rep_stats_desc[i].format);
+	return idx;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_FILL_STATS(sw_rep)
+{
+	int i;
+
+	for (i = 0; i < NUM_VPORT_REP_SW_COUNTERS; i++)
+		data[idx++] = MLX5E_READ_CTR64_CPU(&priv->stats.sw,
+						   sw_rep_stats_desc, i);
+	return idx;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_UPDATE_STATS(sw_rep)
+{
+	struct mlx5e_sw_stats *s = &priv->stats.sw;
+	struct rtnl_link_stats64 stats64 = {};
+
+	memset(s, 0, sizeof(*s));
+	mlx5e_fold_sw_stats64(priv, &stats64);
+
+	s->rx_packets = stats64.rx_packets;
+	s->rx_bytes   = stats64.rx_bytes;
+	s->tx_packets = stats64.tx_packets;
+	s->tx_bytes   = stats64.tx_bytes;
+	s->tx_queue_dropped = stats64.tx_dropped;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_NUM_STATS(vport_rep)
+{
+	return NUM_VPORT_REP_HW_COUNTERS;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_FILL_STRS(vport_rep)
+{
+	int i;
+
+	for (i = 0; i < NUM_VPORT_REP_HW_COUNTERS; i++)
+		strcpy(data + (idx++) * ETH_GSTRING_LEN, vport_rep_stats_desc[i].format);
+	return idx;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_FILL_STATS(vport_rep)
+{
+	int i;
+
+	for (i = 0; i < NUM_VPORT_REP_HW_COUNTERS; i++)
+		data[idx++] = MLX5E_READ_CTR64_CPU(&priv->stats.vf_vport,
+						   vport_rep_stats_desc, i);
+	return idx;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_UPDATE_STATS(vport_rep)
+{
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	struct mlx5e_rep_priv *rpriv = priv->ppriv;
+	struct mlx5_eswitch_rep *rep = rpriv->rep;
+	struct rtnl_link_stats64 *vport_stats;
+	struct ifla_vf_stats vf_stats;
+	int err;
+
+	err = mlx5_eswitch_get_vport_stats(esw, rep->vport, &vf_stats);
+	if (err) {
+		netdev_warn(priv->netdev, "vport %d error %d reading stats\n",
+			    rep->vport, err);
+		return;
+	}
+
+	vport_stats = &priv->stats.vf_vport;
+	/* flip tx/rx as we are reporting the counters for the switch vport */
+	vport_stats->rx_packets = vf_stats.tx_packets;
+	vport_stats->rx_bytes   = vf_stats.tx_bytes;
+	vport_stats->tx_packets = vf_stats.rx_packets;
+	vport_stats->tx_bytes   = vf_stats.rx_bytes;
+}
+
+static void mlx5e_rep_get_strings(struct net_device *dev,
+				  u32 stringset, uint8_t *data)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+
+	switch (stringset) {
+	case ETH_SS_STATS:
+		mlx5e_stats_fill_strings(priv, data);
+		break;
+	}
+}
+
+static void mlx5e_rep_get_ethtool_stats(struct net_device *dev,
+					struct ethtool_stats *stats, u64 *data)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+
+	mlx5e_ethtool_get_ethtool_stats(priv, stats, data);
+}
+
+static int mlx5e_rep_get_sset_count(struct net_device *dev, int sset)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+
+	switch (sset) {
+	case ETH_SS_STATS:
+		return mlx5e_stats_total_num(priv);
+	default:
+		return -EOPNOTSUPP;
+	}
+}
+
+static void
+mlx5e_rep_get_ringparam(struct net_device *dev,
+			struct ethtool_ringparam *param,
+			struct kernel_ethtool_ringparam *kernel_param,
+			struct netlink_ext_ack *extack)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+
+	mlx5e_ethtool_get_ringparam(priv, param, kernel_param);
+}
+
+static int
+mlx5e_rep_set_ringparam(struct net_device *dev,
+			struct ethtool_ringparam *param,
+			struct kernel_ethtool_ringparam *kernel_param,
+			struct netlink_ext_ack *extack)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+
+	return mlx5e_ethtool_set_ringparam(priv, param);
+}
+
+static void mlx5e_rep_get_channels(struct net_device *dev,
+				   struct ethtool_channels *ch)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+
+	mlx5e_ethtool_get_channels(priv, ch);
+}
+
+static int mlx5e_rep_set_channels(struct net_device *dev,
+				  struct ethtool_channels *ch)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+
+	return mlx5e_ethtool_set_channels(priv, ch);
+}
+
+static int mlx5e_rep_get_coalesce(struct net_device *netdev,
+				  struct ethtool_coalesce *coal,
+				  struct kernel_ethtool_coalesce *kernel_coal,
+				  struct netlink_ext_ack *extack)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+
+	return mlx5e_ethtool_get_coalesce(priv, coal, kernel_coal);
+}
+
+static int mlx5e_rep_set_coalesce(struct net_device *netdev,
+				  struct ethtool_coalesce *coal,
+				  struct kernel_ethtool_coalesce *kernel_coal,
+				  struct netlink_ext_ack *extack)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+
+	return mlx5e_ethtool_set_coalesce(priv, coal, kernel_coal, extack);
+}
+
+static u32 mlx5e_rep_get_rxfh_key_size(struct net_device *netdev)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+
+	return mlx5e_ethtool_get_rxfh_key_size(priv);
+}
+
+static u32 mlx5e_rep_get_rxfh_indir_size(struct net_device *netdev)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+
+	return mlx5e_ethtool_get_rxfh_indir_size(priv);
+}
+
+static const struct ethtool_ops mlx5e_rep_ethtool_ops = {
+	.supported_coalesce_params = ETHTOOL_COALESCE_USECS |
+				     ETHTOOL_COALESCE_MAX_FRAMES |
+				     ETHTOOL_COALESCE_USE_ADAPTIVE,
+	.get_drvinfo	   = mlx5e_rep_get_drvinfo,
+	.get_link	   = ethtool_op_get_link,
+	.get_strings       = mlx5e_rep_get_strings,
+	.get_sset_count    = mlx5e_rep_get_sset_count,
+	.get_ethtool_stats = mlx5e_rep_get_ethtool_stats,
+	.get_ringparam     = mlx5e_rep_get_ringparam,
+	.set_ringparam     = mlx5e_rep_set_ringparam,
+	.get_channels      = mlx5e_rep_get_channels,
+	.set_channels      = mlx5e_rep_set_channels,
+	.get_coalesce      = mlx5e_rep_get_coalesce,
+	.set_coalesce      = mlx5e_rep_set_coalesce,
+	.get_rxfh_key_size   = mlx5e_rep_get_rxfh_key_size,
+	.get_rxfh_indir_size = mlx5e_rep_get_rxfh_indir_size,
+};
+
+static void mlx5e_sqs2vport_stop(struct mlx5_eswitch *esw,
+				 struct mlx5_eswitch_rep *rep)
+{
+	struct mlx5e_rep_sq *rep_sq, *tmp;
+	struct mlx5e_rep_priv *rpriv;
+
+	if (esw->mode != MLX5_ESWITCH_OFFLOADS)
+		return;
+
+	rpriv = mlx5e_rep_to_rep_priv(rep);
+	list_for_each_entry_safe(rep_sq, tmp, &rpriv->vport_sqs_list, list) {
+		mlx5_eswitch_del_send_to_vport_rule(rep_sq->send_to_vport_rule);
+		if (rep_sq->send_to_vport_rule_peer)
+			mlx5_eswitch_del_send_to_vport_rule(rep_sq->send_to_vport_rule_peer);
+		list_del(&rep_sq->list);
+		kfree(rep_sq);
+	}
+}
+
+static int mlx5e_sqs2vport_start(struct mlx5_eswitch *esw,
+				 struct mlx5_eswitch_rep *rep,
+				 u32 *sqns_array, int sqns_num)
+{
+	struct mlx5_eswitch *peer_esw = NULL;
+	struct mlx5_flow_handle *flow_rule;
+	struct mlx5e_rep_priv *rpriv;
+	struct mlx5e_rep_sq *rep_sq;
+	int err;
+	int i;
+
+	if (esw->mode != MLX5_ESWITCH_OFFLOADS)
+		return 0;
+
+	rpriv = mlx5e_rep_to_rep_priv(rep);
+	if (mlx5_devcom_is_paired(esw->dev->priv.devcom, MLX5_DEVCOM_ESW_OFFLOADS))
+		peer_esw = mlx5_devcom_get_peer_data(esw->dev->priv.devcom,
+						     MLX5_DEVCOM_ESW_OFFLOADS);
+
+	for (i = 0; i < sqns_num; i++) {
+		rep_sq = kzalloc(sizeof(*rep_sq), GFP_KERNEL);
+		if (!rep_sq) {
+			err = -ENOMEM;
+			goto out_err;
+		}
+
+		/* Add re-inject rule to the PF/representor sqs */
+		flow_rule = mlx5_eswitch_add_send_to_vport_rule(esw, esw, rep,
+								sqns_array[i]);
+		if (IS_ERR(flow_rule)) {
+			err = PTR_ERR(flow_rule);
+			kfree(rep_sq);
+			goto out_err;
+		}
+		rep_sq->send_to_vport_rule = flow_rule;
+		rep_sq->sqn = sqns_array[i];
+
+		if (peer_esw) {
+			flow_rule = mlx5_eswitch_add_send_to_vport_rule(peer_esw, esw,
+									rep, sqns_array[i]);
+			if (IS_ERR(flow_rule)) {
+				err = PTR_ERR(flow_rule);
+				mlx5_eswitch_del_send_to_vport_rule(rep_sq->send_to_vport_rule);
+				kfree(rep_sq);
+				goto out_err;
+			}
+			rep_sq->send_to_vport_rule_peer = flow_rule;
+		}
+
+		list_add(&rep_sq->list, &rpriv->vport_sqs_list);
+	}
+
+	if (peer_esw)
+		mlx5_devcom_release_peer_data(esw->dev->priv.devcom, MLX5_DEVCOM_ESW_OFFLOADS);
+
+	return 0;
+
+out_err:
+	mlx5e_sqs2vport_stop(esw, rep);
+
+	if (peer_esw)
+		mlx5_devcom_release_peer_data(esw->dev->priv.devcom, MLX5_DEVCOM_ESW_OFFLOADS);
+
+	return err;
+}
+
+static int
+mlx5e_add_sqs_fwd_rules(struct mlx5e_priv *priv)
+{
+	int sqs_per_channel = mlx5e_get_dcb_num_tc(&priv->channels.params);
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	bool is_uplink_rep = mlx5e_is_uplink_rep(priv);
+	struct mlx5e_rep_priv *rpriv = priv->ppriv;
+	struct mlx5_eswitch_rep *rep = rpriv->rep;
+	int n, tc, nch, num_sqs = 0;
+	struct mlx5e_channel *c;
+	int err = -ENOMEM;
+	bool ptp_sq;
+	u32 *sqs;
+
+	ptp_sq = !!(priv->channels.ptp &&
+		    MLX5E_GET_PFLAG(&priv->channels.params, MLX5E_PFLAG_TX_PORT_TS));
+	nch = priv->channels.num + ptp_sq;
+	/* +2 for xdpsqs, they don't exist on the ptp channel but will not be
+	 * counted for by num_sqs.
+	 */
+	if (is_uplink_rep)
+		sqs_per_channel += 2;
+
+	sqs = kvcalloc(nch * sqs_per_channel, sizeof(*sqs), GFP_KERNEL);
+	if (!sqs)
+		goto out;
+
+	for (n = 0; n < priv->channels.num; n++) {
+		c = priv->channels.c[n];
+		for (tc = 0; tc < c->num_tc; tc++)
+			sqs[num_sqs++] = c->sq[tc].sqn;
+
+		if (is_uplink_rep) {
+			if (c->xdp)
+				sqs[num_sqs++] = c->rq_xdpsq.sqn;
+
+			sqs[num_sqs++] = c->xdpsq.sqn;
+		}
+	}
+	if (ptp_sq) {
+		struct mlx5e_ptp *ptp_ch = priv->channels.ptp;
+
+		for (tc = 0; tc < ptp_ch->num_tc; tc++)
+			sqs[num_sqs++] = ptp_ch->ptpsq[tc].txqsq.sqn;
+	}
+
+	err = mlx5e_sqs2vport_start(esw, rep, sqs, num_sqs);
+	kvfree(sqs);
+
+out:
+	if (err)
+		netdev_warn(priv->netdev, "Failed to add SQs FWD rules %d\n", err);
+	return err;
+}
+
+static void
+mlx5e_remove_sqs_fwd_rules(struct mlx5e_priv *priv)
+{
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	struct mlx5e_rep_priv *rpriv = priv->ppriv;
+	struct mlx5_eswitch_rep *rep = rpriv->rep;
+
+	mlx5e_sqs2vport_stop(esw, rep);
+}
+
+static int
+mlx5e_rep_add_meta_tunnel_rule(struct mlx5e_priv *priv)
+{
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	struct mlx5e_rep_priv *rpriv = priv->ppriv;
+	struct mlx5_eswitch_rep *rep = rpriv->rep;
+	struct mlx5_flow_handle *flow_rule;
+	struct mlx5_flow_group *g;
+
+	g = esw->fdb_table.offloads.send_to_vport_meta_grp;
+	if (!g)
+		return 0;
+
+	flow_rule = mlx5_eswitch_add_send_to_vport_meta_rule(esw, rep->vport);
+	if (IS_ERR(flow_rule))
+		return PTR_ERR(flow_rule);
+
+	rpriv->send_to_vport_meta_rule = flow_rule;
+
+	return 0;
+}
+
+static void
+mlx5e_rep_del_meta_tunnel_rule(struct mlx5e_priv *priv)
+{
+	struct mlx5e_rep_priv *rpriv = priv->ppriv;
+
+	if (rpriv->send_to_vport_meta_rule)
+		mlx5_eswitch_del_send_to_vport_meta_rule(rpriv->send_to_vport_meta_rule);
+}
+
+void mlx5e_rep_activate_channels(struct mlx5e_priv *priv)
+{
+	mlx5e_add_sqs_fwd_rules(priv);
+	mlx5e_rep_add_meta_tunnel_rule(priv);
+}
+
+void mlx5e_rep_deactivate_channels(struct mlx5e_priv *priv)
+{
+	mlx5e_rep_del_meta_tunnel_rule(priv);
+	mlx5e_remove_sqs_fwd_rules(priv);
+}
+
+static int mlx5e_rep_open(struct net_device *dev)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	struct mlx5e_rep_priv *rpriv = priv->ppriv;
+	struct mlx5_eswitch_rep *rep = rpriv->rep;
+	int err;
+
+	mutex_lock(&priv->state_lock);
+	err = mlx5e_open_locked(dev);
+	if (err)
+		goto unlock;
+
+	if (!mlx5_modify_vport_admin_state(priv->mdev,
+					   MLX5_VPORT_STATE_OP_MOD_ESW_VPORT,
+					   rep->vport, 1,
+					   MLX5_VPORT_ADMIN_STATE_UP))
+		netif_carrier_on(dev);
+
+unlock:
+	mutex_unlock(&priv->state_lock);
+	return err;
+}
+
+static int mlx5e_rep_close(struct net_device *dev)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	struct mlx5e_rep_priv *rpriv = priv->ppriv;
+	struct mlx5_eswitch_rep *rep = rpriv->rep;
+	int ret;
+
+	mutex_lock(&priv->state_lock);
+	mlx5_modify_vport_admin_state(priv->mdev,
+				      MLX5_VPORT_STATE_OP_MOD_ESW_VPORT,
+				      rep->vport, 1,
+				      MLX5_VPORT_ADMIN_STATE_DOWN);
+	ret = mlx5e_close_locked(dev);
+	mutex_unlock(&priv->state_lock);
+	return ret;
+}
+
+bool mlx5e_is_uplink_rep(struct mlx5e_priv *priv)
+{
+	struct mlx5e_rep_priv *rpriv = priv->ppriv;
+	struct mlx5_eswitch_rep *rep;
+
+	if (!MLX5_ESWITCH_MANAGER(priv->mdev))
+		return false;
+
+	if (!rpriv) /* non vport rep mlx5e instances don't use this field */
+		return false;
+
+	rep = rpriv->rep;
+	return (rep->vport == MLX5_VPORT_UPLINK);
+}
+
+bool mlx5e_rep_has_offload_stats(const struct net_device *dev, int attr_id)
+{
+	switch (attr_id) {
+	case IFLA_OFFLOAD_XSTATS_CPU_HIT:
+			return true;
+	}
+
+	return false;
+}
+
+static int
+mlx5e_get_sw_stats64(const struct net_device *dev,
+		     struct rtnl_link_stats64 *stats)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+
+	mlx5e_fold_sw_stats64(priv, stats);
+	return 0;
+}
+
+int mlx5e_rep_get_offload_stats(int attr_id, const struct net_device *dev,
+				void *sp)
+{
+	switch (attr_id) {
+	case IFLA_OFFLOAD_XSTATS_CPU_HIT:
+		return mlx5e_get_sw_stats64(dev, sp);
+	}
+
+	return -EINVAL;
+}
+
+static void
+mlx5e_rep_get_stats(struct net_device *dev, struct rtnl_link_stats64 *stats)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+
+	/* update HW stats in background for next time */
+	mlx5e_queue_update_stats(priv);
+	memcpy(stats, &priv->stats.vf_vport, sizeof(*stats));
+}
+
+static int mlx5e_rep_change_mtu(struct net_device *netdev, int new_mtu)
+{
+	return mlx5e_change_mtu(netdev, new_mtu, NULL);
+}
+
+static struct devlink_port *mlx5e_rep_get_devlink_port(struct net_device *netdev)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	struct mlx5e_rep_priv *rpriv = priv->ppriv;
+	struct mlx5_core_dev *dev = priv->mdev;
+
+	return mlx5_esw_offloads_devlink_port(dev->priv.eswitch, rpriv->rep->vport);
+}
+
+static int mlx5e_rep_change_carrier(struct net_device *dev, bool new_carrier)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	struct mlx5e_rep_priv *rpriv = priv->ppriv;
+	struct mlx5_eswitch_rep *rep = rpriv->rep;
+	int err;
+
+	if (new_carrier) {
+		err = mlx5_modify_vport_admin_state(priv->mdev, MLX5_VPORT_STATE_OP_MOD_ESW_VPORT,
+						    rep->vport, 1, MLX5_VPORT_ADMIN_STATE_UP);
+		if (err)
+			return err;
+		netif_carrier_on(dev);
+	} else {
+		err = mlx5_modify_vport_admin_state(priv->mdev, MLX5_VPORT_STATE_OP_MOD_ESW_VPORT,
+						    rep->vport, 1, MLX5_VPORT_ADMIN_STATE_DOWN);
+		if (err)
+			return err;
+		netif_carrier_off(dev);
+	}
+	return 0;
+}
+
+static const struct net_device_ops mlx5e_netdev_ops_rep = {
+	.ndo_open                = mlx5e_rep_open,
+	.ndo_stop                = mlx5e_rep_close,
+	.ndo_start_xmit          = mlx5e_xmit,
+	.ndo_setup_tc            = mlx5e_rep_setup_tc,
+	.ndo_get_devlink_port    = mlx5e_rep_get_devlink_port,
+	.ndo_get_stats64         = mlx5e_rep_get_stats,
+	.ndo_has_offload_stats	 = mlx5e_rep_has_offload_stats,
+	.ndo_get_offload_stats	 = mlx5e_rep_get_offload_stats,
+	.ndo_change_mtu          = mlx5e_rep_change_mtu,
+	.ndo_change_carrier      = mlx5e_rep_change_carrier,
+};
+
+bool mlx5e_eswitch_uplink_rep(const struct net_device *netdev)
+{
+	return netdev->netdev_ops == &mlx5e_netdev_ops &&
+	       mlx5e_is_uplink_rep(netdev_priv(netdev));
+}
+
+bool mlx5e_eswitch_vf_rep(const struct net_device *netdev)
+{
+	return netdev->netdev_ops == &mlx5e_netdev_ops_rep;
+}
+
+/* One indirect TIR set for outer. Inner not supported in reps. */
+#define REP_NUM_INDIR_TIRS MLX5E_NUM_INDIR_TIRS
+
+static int mlx5e_rep_max_nch_limit(struct mlx5_core_dev *mdev)
+{
+	int max_tir_num = 1 << MLX5_CAP_GEN(mdev, log_max_tir);
+	int num_vports = mlx5_eswitch_get_total_vports(mdev);
+
+	return (max_tir_num - mlx5e_get_pf_num_tirs(mdev)
+		- (num_vports * REP_NUM_INDIR_TIRS)) / num_vports;
+}
+
+static void mlx5e_build_rep_params(struct net_device *netdev)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	struct mlx5e_rep_priv *rpriv = priv->ppriv;
+	struct mlx5_eswitch_rep *rep = rpriv->rep;
+	struct mlx5_core_dev *mdev = priv->mdev;
+	struct mlx5e_params *params;
+
+	u8 cq_period_mode = MLX5_CAP_GEN(mdev, cq_period_start_from_cqe) ?
+					 MLX5_CQ_PERIOD_MODE_START_FROM_CQE :
+					 MLX5_CQ_PERIOD_MODE_START_FROM_EQE;
+
+	params = &priv->channels.params;
+
+	params->num_channels = MLX5E_REP_PARAMS_DEF_NUM_CHANNELS;
+	params->hard_mtu    = MLX5E_ETH_HARD_MTU;
+	params->sw_mtu      = netdev->mtu;
+
+	/* SQ */
+	if (rep->vport == MLX5_VPORT_UPLINK)
+		params->log_sq_size = MLX5E_PARAMS_DEFAULT_LOG_SQ_SIZE;
+	else
+		params->log_sq_size = MLX5E_REP_PARAMS_DEF_LOG_SQ_SIZE;
+
+	/* RQ */
+	mlx5e_build_rq_params(mdev, params);
+
+	/* CQ moderation params */
+	params->rx_dim_enabled = MLX5_CAP_GEN(mdev, cq_moderation);
+	mlx5e_set_rx_cq_mode_params(params, cq_period_mode);
+
+	params->mqprio.num_tc       = 1;
+	if (rep->vport != MLX5_VPORT_UPLINK)
+		params->vlan_strip_disable = true;
+
+	mlx5_query_min_inline(mdev, &params->tx_min_inline_mode);
+}
+
+static void mlx5e_build_rep_netdev(struct net_device *netdev,
+				   struct mlx5_core_dev *mdev)
+{
+	SET_NETDEV_DEV(netdev, mdev->device);
+	netdev->netdev_ops = &mlx5e_netdev_ops_rep;
+	eth_hw_addr_random(netdev);
+	netdev->ethtool_ops = &mlx5e_rep_ethtool_ops;
+
+	netdev->watchdog_timeo    = 15 * HZ;
+
+#if IS_ENABLED(CONFIG_MLX5_CLS_ACT)
+	netdev->hw_features    |= NETIF_F_HW_TC;
+#endif
+	netdev->hw_features    |= NETIF_F_SG;
+	netdev->hw_features    |= NETIF_F_IP_CSUM;
+	netdev->hw_features    |= NETIF_F_IPV6_CSUM;
+	netdev->hw_features    |= NETIF_F_GRO;
+	netdev->hw_features    |= NETIF_F_TSO;
+	netdev->hw_features    |= NETIF_F_TSO6;
+	netdev->hw_features    |= NETIF_F_RXCSUM;
+
+	netdev->features |= netdev->hw_features;
+	netdev->features |= NETIF_F_NETNS_LOCAL;
+}
+
+static int mlx5e_init_rep(struct mlx5_core_dev *mdev,
+			  struct net_device *netdev)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+
+	priv->fs = mlx5e_fs_init(priv->profile, mdev,
+				 !test_bit(MLX5E_STATE_DESTROYING, &priv->state));
+	if (!priv->fs) {
+		netdev_err(priv->netdev, "FS allocation failed\n");
+		return -ENOMEM;
+	}
+
+	mlx5e_build_rep_params(netdev);
+	mlx5e_timestamp_init(priv);
+
+	return 0;
+}
+
+static int mlx5e_init_ul_rep(struct mlx5_core_dev *mdev,
+			     struct net_device *netdev)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	int err;
+
+	priv->fs = mlx5e_fs_init(priv->profile, mdev,
+				 !test_bit(MLX5E_STATE_DESTROYING, &priv->state));
+	if (!priv->fs) {
+		netdev_err(priv->netdev, "FS allocation failed\n");
+		return -ENOMEM;
+	}
+
+	err = mlx5e_ipsec_init(priv);
+	if (err)
+		mlx5_core_err(mdev, "Uplink rep IPsec initialization failed, %d\n", err);
+
+	mlx5e_vxlan_set_netdev_info(priv);
+	mlx5e_build_rep_params(netdev);
+	mlx5e_timestamp_init(priv);
+	return 0;
+}
+
+static void mlx5e_cleanup_rep(struct mlx5e_priv *priv)
+{
+	mlx5e_fs_cleanup(priv->fs);
+	mlx5e_ipsec_cleanup(priv);
+	priv->fs = NULL;
+}
+
+static int mlx5e_create_rep_ttc_table(struct mlx5e_priv *priv)
+{
+	struct mlx5e_rep_priv *rpriv = priv->ppriv;
+	struct mlx5_eswitch_rep *rep = rpriv->rep;
+	struct ttc_params ttc_params = {};
+	int err;
+
+	mlx5e_fs_set_ns(priv->fs,
+			mlx5_get_flow_namespace(priv->mdev,
+						MLX5_FLOW_NAMESPACE_KERNEL), false);
+
+	/* The inner_ttc in the ttc params is intentionally not set */
+	mlx5e_set_ttc_params(priv->fs, priv->rx_res, &ttc_params, false);
+
+	if (rep->vport != MLX5_VPORT_UPLINK)
+		/* To give uplik rep TTC a lower level for chaining from root ft */
+		ttc_params.ft_attr.level = MLX5E_TTC_FT_LEVEL + 1;
+
+	mlx5e_fs_set_ttc(priv->fs, mlx5_create_ttc_table(priv->mdev, &ttc_params), false);
+	if (IS_ERR(mlx5e_fs_get_ttc(priv->fs, false))) {
+		err = PTR_ERR(mlx5e_fs_get_ttc(priv->fs, false));
+		netdev_err(priv->netdev, "Failed to create rep ttc table, err=%d\n",
+			   err);
+		return err;
+	}
+	return 0;
+}
+
+static int mlx5e_create_rep_root_ft(struct mlx5e_priv *priv)
+{
+	struct mlx5e_rep_priv *rpriv = priv->ppriv;
+	struct mlx5_eswitch_rep *rep = rpriv->rep;
+	struct mlx5_flow_table_attr ft_attr = {};
+	struct mlx5_flow_namespace *ns;
+	int err = 0;
+
+	if (rep->vport != MLX5_VPORT_UPLINK) {
+		/* non uplik reps will skip any bypass tables and go directly to
+		 * their own ttc
+		 */
+		rpriv->root_ft = mlx5_get_ttc_flow_table(mlx5e_fs_get_ttc(priv->fs, false));
+		return 0;
+	}
+
+	/* uplink root ft will be used to auto chain, to ethtool or ttc tables */
+	ns = mlx5_get_flow_namespace(priv->mdev, MLX5_FLOW_NAMESPACE_OFFLOADS);
+	if (!ns) {
+		netdev_err(priv->netdev, "Failed to get reps offloads namespace\n");
+		return -EOPNOTSUPP;
+	}
+
+	ft_attr.max_fte = 0; /* Empty table, miss rule will always point to next table */
+	ft_attr.prio = 1;
+	ft_attr.level = 1;
+
+	rpriv->root_ft = mlx5_create_flow_table(ns, &ft_attr);
+	if (IS_ERR(rpriv->root_ft)) {
+		err = PTR_ERR(rpriv->root_ft);
+		rpriv->root_ft = NULL;
+	}
+
+	return err;
+}
+
+static void mlx5e_destroy_rep_root_ft(struct mlx5e_priv *priv)
+{
+	struct mlx5e_rep_priv *rpriv = priv->ppriv;
+	struct mlx5_eswitch_rep *rep = rpriv->rep;
+
+	if (rep->vport != MLX5_VPORT_UPLINK)
+		return;
+	mlx5_destroy_flow_table(rpriv->root_ft);
+}
+
+static int mlx5e_create_rep_vport_rx_rule(struct mlx5e_priv *priv)
+{
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	struct mlx5e_rep_priv *rpriv = priv->ppriv;
+	struct mlx5_eswitch_rep *rep = rpriv->rep;
+	struct mlx5_flow_handle *flow_rule;
+	struct mlx5_flow_destination dest;
+
+	dest.type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
+	dest.ft = rpriv->root_ft;
+
+	flow_rule = mlx5_eswitch_create_vport_rx_rule(esw, rep->vport, &dest);
+	if (IS_ERR(flow_rule))
+		return PTR_ERR(flow_rule);
+	rpriv->vport_rx_rule = flow_rule;
+	return 0;
+}
+
+static void rep_vport_rx_rule_destroy(struct mlx5e_priv *priv)
+{
+	struct mlx5e_rep_priv *rpriv = priv->ppriv;
+
+	if (!rpriv->vport_rx_rule)
+		return;
+
+	mlx5_del_flow_rules(rpriv->vport_rx_rule);
+	rpriv->vport_rx_rule = NULL;
+}
+
+int mlx5e_rep_bond_update(struct mlx5e_priv *priv, bool cleanup)
+{
+	rep_vport_rx_rule_destroy(priv);
+
+	return cleanup ? 0 : mlx5e_create_rep_vport_rx_rule(priv);
+}
+
+static int mlx5e_init_rep_rx(struct mlx5e_priv *priv)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+	int err;
+
+	priv->rx_res = mlx5e_rx_res_alloc();
+	if (!priv->rx_res) {
+		err = -ENOMEM;
+		goto err_free_fs;
+	}
+
+	mlx5e_fs_init_l2_addr(priv->fs, priv->netdev);
+
+	err = mlx5e_open_drop_rq(priv, &priv->drop_rq);
+	if (err) {
+		mlx5_core_err(mdev, "open drop rq failed, %d\n", err);
+		goto err_rx_res_free;
+	}
+
+	err = mlx5e_rx_res_init(priv->rx_res, priv->mdev, 0,
+				priv->max_nch, priv->drop_rq.rqn,
+				&priv->channels.params.packet_merge,
+				priv->channels.params.num_channels);
+	if (err)
+		goto err_close_drop_rq;
+
+	err = mlx5e_create_rep_ttc_table(priv);
+	if (err)
+		goto err_destroy_rx_res;
+
+	err = mlx5e_create_rep_root_ft(priv);
+	if (err)
+		goto err_destroy_ttc_table;
+
+	err = mlx5e_create_rep_vport_rx_rule(priv);
+	if (err)
+		goto err_destroy_root_ft;
+
+	mlx5e_ethtool_init_steering(priv->fs);
+
+	return 0;
+
+err_destroy_root_ft:
+	mlx5e_destroy_rep_root_ft(priv);
+err_destroy_ttc_table:
+	mlx5_destroy_ttc_table(mlx5e_fs_get_ttc(priv->fs, false));
+err_destroy_rx_res:
+	mlx5e_rx_res_destroy(priv->rx_res);
+err_close_drop_rq:
+	mlx5e_close_drop_rq(&priv->drop_rq);
+err_rx_res_free:
+	mlx5e_rx_res_free(priv->rx_res);
+	priv->rx_res = NULL;
+err_free_fs:
+	mlx5e_fs_cleanup(priv->fs);
+	priv->fs = NULL;
+	return err;
+}
+
+static void mlx5e_cleanup_rep_rx(struct mlx5e_priv *priv)
+{
+	mlx5e_ethtool_cleanup_steering(priv->fs);
+	rep_vport_rx_rule_destroy(priv);
+	mlx5e_destroy_rep_root_ft(priv);
+	mlx5_destroy_ttc_table(mlx5e_fs_get_ttc(priv->fs, false));
+	mlx5e_rx_res_destroy(priv->rx_res);
+	mlx5e_close_drop_rq(&priv->drop_rq);
+	mlx5e_rx_res_free(priv->rx_res);
+	priv->rx_res = NULL;
+}
+
+static int mlx5e_init_ul_rep_rx(struct mlx5e_priv *priv)
+{
+	int err;
+
+	mlx5e_create_q_counters(priv);
+	err = mlx5e_init_rep_rx(priv);
+	if (err)
+		goto out;
+
+	mlx5e_tc_int_port_init_rep_rx(priv);
+
+out:
+	return err;
+}
+
+static void mlx5e_cleanup_ul_rep_rx(struct mlx5e_priv *priv)
+{
+	mlx5e_tc_int_port_cleanup_rep_rx(priv);
+	mlx5e_cleanup_rep_rx(priv);
+	mlx5e_destroy_q_counters(priv);
+}
+
+static int mlx5e_init_uplink_rep_tx(struct mlx5e_rep_priv *rpriv)
+{
+	struct mlx5_rep_uplink_priv *uplink_priv;
+	struct net_device *netdev;
+	struct mlx5e_priv *priv;
+	int err;
+
+	netdev = rpriv->netdev;
+	priv = netdev_priv(netdev);
+	uplink_priv = &rpriv->uplink_priv;
+
+	err = mlx5e_rep_tc_init(rpriv);
+	if (err)
+		return err;
+
+	mlx5_init_port_tun_entropy(&uplink_priv->tun_entropy, priv->mdev);
+
+	mlx5e_rep_bond_init(rpriv);
+	err = mlx5e_rep_tc_netdevice_event_register(rpriv);
+	if (err) {
+		mlx5_core_err(priv->mdev, "Failed to register netdev notifier, err: %d\n",
+			      err);
+		goto err_event_reg;
+	}
+
+	return 0;
+
+err_event_reg:
+	mlx5e_rep_bond_cleanup(rpriv);
+	mlx5e_rep_tc_cleanup(rpriv);
+	return err;
+}
+
+static void mlx5e_cleanup_uplink_rep_tx(struct mlx5e_rep_priv *rpriv)
+{
+	mlx5e_rep_tc_netdevice_event_unregister(rpriv);
+	mlx5e_rep_bond_cleanup(rpriv);
+	mlx5e_rep_tc_cleanup(rpriv);
+}
+
+static int mlx5e_init_rep_tx(struct mlx5e_priv *priv)
+{
+	struct mlx5e_rep_priv *rpriv = priv->ppriv;
+	int err;
+
+	err = mlx5e_create_tises(priv);
+	if (err) {
+		mlx5_core_warn(priv->mdev, "create tises failed, %d\n", err);
+		return err;
+	}
+
+	err = mlx5e_rep_neigh_init(rpriv);
+	if (err)
+		goto err_neigh_init;
+
+	if (rpriv->rep->vport == MLX5_VPORT_UPLINK) {
+		err = mlx5e_init_uplink_rep_tx(rpriv);
+		if (err)
+			goto err_init_tx;
+	}
+
+	err = mlx5e_tc_ht_init(&rpriv->tc_ht);
+	if (err)
+		goto err_ht_init;
+
+	return 0;
+
+err_ht_init:
+	if (rpriv->rep->vport == MLX5_VPORT_UPLINK)
+		mlx5e_cleanup_uplink_rep_tx(rpriv);
+err_init_tx:
+	mlx5e_rep_neigh_cleanup(rpriv);
+err_neigh_init:
+	mlx5e_destroy_tises(priv);
+	return err;
+}
+
+static void mlx5e_cleanup_rep_tx(struct mlx5e_priv *priv)
+{
+	struct mlx5e_rep_priv *rpriv = priv->ppriv;
+
+	mlx5e_tc_ht_cleanup(&rpriv->tc_ht);
+
+	if (rpriv->rep->vport == MLX5_VPORT_UPLINK)
+		mlx5e_cleanup_uplink_rep_tx(rpriv);
+
+	mlx5e_rep_neigh_cleanup(rpriv);
+	mlx5e_destroy_tises(priv);
+}
+
+static void mlx5e_rep_enable(struct mlx5e_priv *priv)
+{
+	mlx5e_set_netdev_mtu_boundaries(priv);
+}
+
+static void mlx5e_rep_disable(struct mlx5e_priv *priv)
+{
+}
+
+static int mlx5e_update_rep_rx(struct mlx5e_priv *priv)
+{
+	return 0;
+}
+
+static int uplink_rep_async_event(struct notifier_block *nb, unsigned long event, void *data)
+{
+	struct mlx5e_priv *priv = container_of(nb, struct mlx5e_priv, events_nb);
+
+	if (event == MLX5_EVENT_TYPE_PORT_CHANGE) {
+		struct mlx5_eqe *eqe = data;
+
+		switch (eqe->sub_type) {
+		case MLX5_PORT_CHANGE_SUBTYPE_DOWN:
+		case MLX5_PORT_CHANGE_SUBTYPE_ACTIVE:
+			queue_work(priv->wq, &priv->update_carrier_work);
+			break;
+		default:
+			return NOTIFY_DONE;
+		}
+
+		return NOTIFY_OK;
+	}
+
+	if (event == MLX5_DEV_EVENT_PORT_AFFINITY)
+		return mlx5e_rep_tc_event_port_affinity(priv);
+
+	return NOTIFY_DONE;
+}
+
+static void mlx5e_uplink_rep_enable(struct mlx5e_priv *priv)
+{
+	struct net_device *netdev = priv->netdev;
+	struct mlx5_core_dev *mdev = priv->mdev;
+	u16 max_mtu;
+
+	netdev->min_mtu = ETH_MIN_MTU;
+	mlx5_query_port_max_mtu(priv->mdev, &max_mtu, 1);
+	netdev->max_mtu = MLX5E_HW2SW_MTU(&priv->channels.params, max_mtu);
+	mlx5e_set_dev_port_mtu(priv);
+
+	mlx5e_rep_tc_enable(priv);
+
+	if (MLX5_CAP_GEN(mdev, uplink_follow))
+		mlx5_modify_vport_admin_state(mdev, MLX5_VPORT_STATE_OP_MOD_UPLINK,
+					      0, 0, MLX5_VPORT_ADMIN_STATE_AUTO);
+	mlx5_lag_add_netdev(mdev, netdev);
+	priv->events_nb.notifier_call = uplink_rep_async_event;
+	mlx5_notifier_register(mdev, &priv->events_nb);
+	mlx5e_dcbnl_initialize(priv);
+	mlx5e_dcbnl_init_app(priv);
+	mlx5e_rep_bridge_init(priv);
+
+	netdev->wanted_features |= NETIF_F_HW_TC;
+
+	rtnl_lock();
+	if (netif_running(netdev))
+		mlx5e_open(netdev);
+	udp_tunnel_nic_reset_ntf(priv->netdev);
+	netif_device_attach(netdev);
+	rtnl_unlock();
+}
+
+static void mlx5e_uplink_rep_disable(struct mlx5e_priv *priv)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+
+	rtnl_lock();
+	if (netif_running(priv->netdev))
+		mlx5e_close(priv->netdev);
+	netif_device_detach(priv->netdev);
+	rtnl_unlock();
+
+	mlx5e_rep_bridge_cleanup(priv);
+	mlx5e_dcbnl_delete_app(priv);
+	mlx5_notifier_unregister(mdev, &priv->events_nb);
+	mlx5e_rep_tc_disable(priv);
+	mlx5_lag_remove_netdev(mdev, priv->netdev);
+	mlx5_vxlan_reset_to_default(mdev->vxlan);
+}
+
+static MLX5E_DEFINE_STATS_GRP(sw_rep, 0);
+static MLX5E_DEFINE_STATS_GRP(vport_rep, MLX5E_NDO_UPDATE_STATS);
+
+/* The stats groups order is opposite to the update_stats() order calls */
+static mlx5e_stats_grp_t mlx5e_rep_stats_grps[] = {
+	&MLX5E_STATS_GRP(sw_rep),
+	&MLX5E_STATS_GRP(vport_rep),
+};
+
+static unsigned int mlx5e_rep_stats_grps_num(struct mlx5e_priv *priv)
+{
+	return ARRAY_SIZE(mlx5e_rep_stats_grps);
+}
+
+/* The stats groups order is opposite to the update_stats() order calls */
+static mlx5e_stats_grp_t mlx5e_ul_rep_stats_grps[] = {
+	&MLX5E_STATS_GRP(sw),
+	&MLX5E_STATS_GRP(qcnt),
+	&MLX5E_STATS_GRP(vnic_env),
+	&MLX5E_STATS_GRP(vport),
+	&MLX5E_STATS_GRP(802_3),
+	&MLX5E_STATS_GRP(2863),
+	&MLX5E_STATS_GRP(2819),
+	&MLX5E_STATS_GRP(phy),
+	&MLX5E_STATS_GRP(eth_ext),
+	&MLX5E_STATS_GRP(pcie),
+	&MLX5E_STATS_GRP(per_prio),
+	&MLX5E_STATS_GRP(pme),
+	&MLX5E_STATS_GRP(channels),
+	&MLX5E_STATS_GRP(per_port_buff_congest),
+#ifdef CONFIG_MLX5_EN_IPSEC
+	&MLX5E_STATS_GRP(ipsec_sw),
+#endif
+	&MLX5E_STATS_GRP(ptp),
+};
+
+static unsigned int mlx5e_ul_rep_stats_grps_num(struct mlx5e_priv *priv)
+{
+	return ARRAY_SIZE(mlx5e_ul_rep_stats_grps);
+}
+
+static const struct mlx5e_profile mlx5e_rep_profile = {
+	.init			= mlx5e_init_rep,
+	.cleanup		= mlx5e_cleanup_rep,
+	.init_rx		= mlx5e_init_rep_rx,
+	.cleanup_rx		= mlx5e_cleanup_rep_rx,
+	.init_tx		= mlx5e_init_rep_tx,
+	.cleanup_tx		= mlx5e_cleanup_rep_tx,
+	.enable		        = mlx5e_rep_enable,
+	.disable	        = mlx5e_rep_disable,
+	.update_rx		= mlx5e_update_rep_rx,
+	.update_stats           = mlx5e_stats_update_ndo_stats,
+	.rx_handlers            = &mlx5e_rx_handlers_rep,
+	.max_tc			= 1,
+	.stats_grps		= mlx5e_rep_stats_grps,
+	.stats_grps_num		= mlx5e_rep_stats_grps_num,
+	.max_nch_limit		= mlx5e_rep_max_nch_limit,
+};
+
+static const struct mlx5e_profile mlx5e_uplink_rep_profile = {
+	.init			= mlx5e_init_ul_rep,
+	.cleanup		= mlx5e_cleanup_rep,
+	.init_rx		= mlx5e_init_ul_rep_rx,
+	.cleanup_rx		= mlx5e_cleanup_ul_rep_rx,
+	.init_tx		= mlx5e_init_rep_tx,
+	.cleanup_tx		= mlx5e_cleanup_rep_tx,
+	.enable		        = mlx5e_uplink_rep_enable,
+	.disable	        = mlx5e_uplink_rep_disable,
+	.update_rx		= mlx5e_update_rep_rx,
+	.update_stats           = mlx5e_stats_update_ndo_stats,
+	.update_carrier	        = mlx5e_update_carrier,
+	.rx_handlers            = &mlx5e_rx_handlers_rep,
+	.max_tc			= MLX5E_MAX_NUM_TC,
+	.stats_grps		= mlx5e_ul_rep_stats_grps,
+	.stats_grps_num		= mlx5e_ul_rep_stats_grps_num,
+};
+
+/* e-Switch vport representors */
+static int
+mlx5e_vport_uplink_rep_load(struct mlx5_core_dev *dev, struct mlx5_eswitch_rep *rep)
+{
+	struct mlx5e_priv *priv = netdev_priv(mlx5_uplink_netdev_get(dev));
+	struct mlx5e_rep_priv *rpriv = mlx5e_rep_to_rep_priv(rep);
+	struct devlink_port *dl_port;
+	int err;
+
+	rpriv->netdev = priv->netdev;
+
+	err = mlx5e_netdev_change_profile(priv, &mlx5e_uplink_rep_profile,
+					  rpriv);
+	if (err)
+		return err;
+
+	dl_port = mlx5_esw_offloads_devlink_port(dev->priv.eswitch, rpriv->rep->vport);
+	if (dl_port)
+		devlink_port_type_eth_set(dl_port, rpriv->netdev);
+
+	return 0;
+}
+
+static void
+mlx5e_vport_uplink_rep_unload(struct mlx5e_rep_priv *rpriv)
+{
+	struct net_device *netdev = rpriv->netdev;
+	struct devlink_port *dl_port;
+	struct mlx5_core_dev *dev;
+	struct mlx5e_priv *priv;
+
+	priv = netdev_priv(netdev);
+	dev = priv->mdev;
+
+	dl_port = mlx5_esw_offloads_devlink_port(dev->priv.eswitch, rpriv->rep->vport);
+	if (dl_port)
+		devlink_port_type_clear(dl_port);
+	mlx5e_netdev_attach_nic_profile(priv);
+}
+
+static int
+mlx5e_vport_vf_rep_load(struct mlx5_core_dev *dev, struct mlx5_eswitch_rep *rep)
+{
+	struct mlx5e_rep_priv *rpriv = mlx5e_rep_to_rep_priv(rep);
+	const struct mlx5e_profile *profile;
+	struct devlink_port *dl_port;
+	struct net_device *netdev;
+	struct mlx5e_priv *priv;
+	int err;
+
+	profile = &mlx5e_rep_profile;
+	netdev = mlx5e_create_netdev(dev, profile);
+	if (!netdev) {
+		mlx5_core_warn(dev,
+			       "Failed to create representor netdev for vport %d\n",
+			       rep->vport);
+		return -EINVAL;
+	}
+
+	mlx5e_build_rep_netdev(netdev, dev);
+	rpriv->netdev = netdev;
+
+	priv = netdev_priv(netdev);
+	priv->profile = profile;
+	priv->ppriv = rpriv;
+	err = profile->init(dev, netdev);
+	if (err) {
+		netdev_warn(netdev, "rep profile init failed, %d\n", err);
+		goto err_destroy_netdev;
+	}
+
+	err = mlx5e_attach_netdev(netdev_priv(netdev));
+	if (err) {
+		netdev_warn(netdev,
+			    "Failed to attach representor netdev for vport %d\n",
+			    rep->vport);
+		goto err_cleanup_profile;
+	}
+
+	err = register_netdev(netdev);
+	if (err) {
+		netdev_warn(netdev,
+			    "Failed to register representor netdev for vport %d\n",
+			    rep->vport);
+		goto err_detach_netdev;
+	}
+
+	dl_port = mlx5_esw_offloads_devlink_port(dev->priv.eswitch, rpriv->rep->vport);
+	if (dl_port)
+		devlink_port_type_eth_set(dl_port, netdev);
+	return 0;
+
+err_detach_netdev:
+	mlx5e_detach_netdev(netdev_priv(netdev));
+
+err_cleanup_profile:
+	priv->profile->cleanup(priv);
+
+err_destroy_netdev:
+	mlx5e_destroy_netdev(netdev_priv(netdev));
+	return err;
+}
+
+static int
+mlx5e_vport_rep_load(struct mlx5_core_dev *dev, struct mlx5_eswitch_rep *rep)
+{
+	struct mlx5e_rep_priv *rpriv;
+	int err;
+
+	rpriv = kvzalloc(sizeof(*rpriv), GFP_KERNEL);
+	if (!rpriv)
+		return -ENOMEM;
+
+	/* rpriv->rep to be looked up when profile->init() is called */
+	rpriv->rep = rep;
+	rep->rep_data[REP_ETH].priv = rpriv;
+	INIT_LIST_HEAD(&rpriv->vport_sqs_list);
+
+	if (rep->vport == MLX5_VPORT_UPLINK)
+		err = mlx5e_vport_uplink_rep_load(dev, rep);
+	else
+		err = mlx5e_vport_vf_rep_load(dev, rep);
+
+	if (err)
+		kvfree(rpriv);
+
+	return err;
+}
+
+static void
+mlx5e_vport_rep_unload(struct mlx5_eswitch_rep *rep)
+{
+	struct mlx5e_rep_priv *rpriv = mlx5e_rep_to_rep_priv(rep);
+	struct net_device *netdev = rpriv->netdev;
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	struct mlx5_core_dev *dev = priv->mdev;
+	struct devlink_port *dl_port;
+	void *ppriv = priv->ppriv;
+
+	if (rep->vport == MLX5_VPORT_UPLINK) {
+		mlx5e_vport_uplink_rep_unload(rpriv);
+		goto free_ppriv;
+	}
+
+	dl_port = mlx5_esw_offloads_devlink_port(dev->priv.eswitch, rpriv->rep->vport);
+	if (dl_port)
+		devlink_port_type_clear(dl_port);
+	unregister_netdev(netdev);
+	mlx5e_detach_netdev(priv);
+	priv->profile->cleanup(priv);
+	mlx5e_destroy_netdev(priv);
+free_ppriv:
+	kvfree(ppriv); /* mlx5e_rep_priv */
+}
+
+static void *mlx5e_vport_rep_get_proto_dev(struct mlx5_eswitch_rep *rep)
+{
+	struct mlx5e_rep_priv *rpriv;
+
+	rpriv = mlx5e_rep_to_rep_priv(rep);
+
+	return rpriv->netdev;
+}
+
+static void mlx5e_vport_rep_event_unpair(struct mlx5_eswitch_rep *rep)
+{
+	struct mlx5e_rep_priv *rpriv;
+	struct mlx5e_rep_sq *rep_sq;
+
+	rpriv = mlx5e_rep_to_rep_priv(rep);
+	list_for_each_entry(rep_sq, &rpriv->vport_sqs_list, list) {
+		if (!rep_sq->send_to_vport_rule_peer)
+			continue;
+		mlx5_eswitch_del_send_to_vport_rule(rep_sq->send_to_vport_rule_peer);
+		rep_sq->send_to_vport_rule_peer = NULL;
+	}
+}
+
+static int mlx5e_vport_rep_event_pair(struct mlx5_eswitch *esw,
+				      struct mlx5_eswitch_rep *rep,
+				      struct mlx5_eswitch *peer_esw)
+{
+	struct mlx5_flow_handle *flow_rule;
+	struct mlx5e_rep_priv *rpriv;
+	struct mlx5e_rep_sq *rep_sq;
+
+	rpriv = mlx5e_rep_to_rep_priv(rep);
+	list_for_each_entry(rep_sq, &rpriv->vport_sqs_list, list) {
+		if (rep_sq->send_to_vport_rule_peer)
+			continue;
+		flow_rule = mlx5_eswitch_add_send_to_vport_rule(peer_esw, esw, rep, rep_sq->sqn);
+		if (IS_ERR(flow_rule))
+			goto err_out;
+		rep_sq->send_to_vport_rule_peer = flow_rule;
+	}
+
+	return 0;
+err_out:
+	mlx5e_vport_rep_event_unpair(rep);
+	return PTR_ERR(flow_rule);
+}
+
+static int mlx5e_vport_rep_event(struct mlx5_eswitch *esw,
+				 struct mlx5_eswitch_rep *rep,
+				 enum mlx5_switchdev_event event,
+				 void *data)
+{
+	int err = 0;
+
+	if (event == MLX5_SWITCHDEV_EVENT_PAIR)
+		err = mlx5e_vport_rep_event_pair(esw, rep, data);
+	else if (event == MLX5_SWITCHDEV_EVENT_UNPAIR)
+		mlx5e_vport_rep_event_unpair(rep);
+
+	return err;
+}
+
+static const struct mlx5_eswitch_rep_ops rep_ops = {
+	.load = mlx5e_vport_rep_load,
+	.unload = mlx5e_vport_rep_unload,
+	.get_proto_dev = mlx5e_vport_rep_get_proto_dev,
+	.event = mlx5e_vport_rep_event,
+};
+
+static int mlx5e_rep_probe(struct auxiliary_device *adev,
+			   const struct auxiliary_device_id *id)
+{
+	struct mlx5_adev *edev = container_of(adev, struct mlx5_adev, adev);
+	struct mlx5_core_dev *mdev = edev->mdev;
+	struct mlx5_eswitch *esw;
+
+	esw = mdev->priv.eswitch;
+	mlx5_eswitch_register_vport_reps(esw, &rep_ops, REP_ETH);
+	return 0;
+}
+
+static void mlx5e_rep_remove(struct auxiliary_device *adev)
+{
+	struct mlx5_adev *vdev = container_of(adev, struct mlx5_adev, adev);
+	struct mlx5_core_dev *mdev = vdev->mdev;
+	struct mlx5_eswitch *esw;
+
+	esw = mdev->priv.eswitch;
+	mlx5_eswitch_unregister_vport_reps(esw, REP_ETH);
+}
+
+static const struct auxiliary_device_id mlx5e_rep_id_table[] = {
+	{ .name = MLX5_ADEV_NAME ".eth-rep", },
+	{},
+};
+
+MODULE_DEVICE_TABLE(auxiliary, mlx5e_rep_id_table);
+
+static struct auxiliary_driver mlx5e_rep_driver = {
+	.name = "eth-rep",
+	.probe = mlx5e_rep_probe,
+	.remove = mlx5e_rep_remove,
+	.id_table = mlx5e_rep_id_table,
+};
+
+int mlx5e_rep_init(void)
+{
+	return auxiliary_driver_register(&mlx5e_rep_driver);
+}
+
+void mlx5e_rep_cleanup(void)
+{
+	auxiliary_driver_unregister(&mlx5e_rep_driver);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_rep.h b/drivers/net/ethernet/mellanox/mlx5/core/en_rep.h
new file mode 100644
index 000000000..b4e691760
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_rep.h
@@ -0,0 +1,275 @@
+/*
+ * Copyright (c) 2017, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef __MLX5E_REP_H__
+#define __MLX5E_REP_H__
+
+#include <net/ip_tunnels.h>
+#include <linux/rhashtable.h>
+#include <linux/mutex.h>
+#include "eswitch.h"
+#include "en.h"
+#include "lib/port_tun.h"
+
+#ifdef CONFIG_MLX5_ESWITCH
+extern const struct mlx5e_rx_handlers mlx5e_rx_handlers_rep;
+
+struct mlx5e_neigh_update_table {
+	struct rhashtable       neigh_ht;
+	/* Save the neigh hash entries in a list in addition to the hash table
+	 * (neigh_ht). In order to iterate easily over the neigh entries.
+	 * Used for stats query.
+	 */
+	struct list_head	neigh_list;
+	/* protect lookup/remove operations */
+	struct mutex		encap_lock;
+	struct notifier_block   netevent_nb;
+	struct delayed_work     neigh_stats_work;
+	unsigned long           min_interval; /* jiffies */
+};
+
+struct mlx5_tc_ct_priv;
+struct mlx5_tc_int_port_priv;
+struct mlx5e_rep_bond;
+struct mlx5e_tc_tun_encap;
+struct mlx5e_post_act;
+struct mlx5e_flow_meters;
+
+struct mlx5_rep_uplink_priv {
+	/* indirect block callbacks are invoked on bind/unbind events
+	 * on registered higher level devices (e.g. tunnel devices)
+	 *
+	 * tc_indr_block_cb_priv_list is used to lookup indirect callback
+	 * private data
+	 *
+	 */
+	struct list_head	    tc_indr_block_priv_list;
+
+	struct mlx5_tun_entropy tun_entropy;
+
+	/* protects unready_flows */
+	struct mutex                unready_flows_lock;
+	struct list_head            unready_flows;
+	struct work_struct          reoffload_flows_work;
+
+	/* maps tun_info to a unique id*/
+	struct mapping_ctx *tunnel_mapping;
+	/* maps tun_enc_opts to a unique id*/
+	struct mapping_ctx *tunnel_enc_opts_mapping;
+
+	struct mlx5e_post_act *post_act;
+	struct mlx5_tc_ct_priv *ct_priv;
+	struct mlx5e_tc_psample *tc_psample;
+
+	/* support eswitch vports bonding */
+	struct mlx5e_rep_bond *bond;
+
+	/* tc tunneling encapsulation private data */
+	struct mlx5e_tc_tun_encap *encap;
+
+	/* OVS internal port support */
+	struct mlx5e_tc_int_port_priv *int_port_priv;
+
+	struct mlx5e_flow_meters *flow_meters;
+};
+
+struct mlx5e_rep_priv {
+	struct mlx5_eswitch_rep *rep;
+	struct mlx5e_neigh_update_table neigh_update;
+	struct net_device      *netdev;
+	struct mlx5_flow_table *root_ft;
+	struct mlx5_flow_handle *vport_rx_rule;
+	struct list_head       vport_sqs_list;
+	struct mlx5_rep_uplink_priv uplink_priv; /* valid for uplink rep */
+	struct rtnl_link_stats64 prev_vf_vport_stats;
+	struct mlx5_flow_handle *send_to_vport_meta_rule;
+	struct rhashtable tc_ht;
+};
+
+static inline
+struct mlx5e_rep_priv *mlx5e_rep_to_rep_priv(struct mlx5_eswitch_rep *rep)
+{
+	return rep->rep_data[REP_ETH].priv;
+}
+
+struct mlx5e_neigh {
+	union {
+		__be32	v4;
+		struct in6_addr v6;
+	} dst_ip;
+	int family;
+};
+
+struct mlx5e_neigh_hash_entry {
+	struct rhash_head rhash_node;
+	struct mlx5e_neigh m_neigh;
+	struct mlx5e_priv *priv;
+	struct net_device *neigh_dev;
+
+	/* Save the neigh hash entry in a list on the representor in
+	 * addition to the hash table. In order to iterate easily over the
+	 * neighbour entries. Used for stats query.
+	 */
+	struct list_head neigh_list;
+
+	/* protects encap list */
+	spinlock_t encap_list_lock;
+	/* encap list sharing the same neigh */
+	struct list_head encap_list;
+
+	/* neigh hash entry can be deleted only when the refcount is zero.
+	 * refcount is needed to avoid neigh hash entry removal by TC, while
+	 * it's used by the neigh notification call.
+	 */
+	refcount_t refcnt;
+
+	/* Save the last reported time offloaded traffic pass over one of the
+	 * neigh hash entry flows. Use it to periodically update the neigh
+	 * 'used' value and avoid neigh deleting by the kernel.
+	 */
+	unsigned long reported_lastuse;
+
+	struct rcu_head rcu;
+};
+
+enum {
+	/* set when the encap entry is successfully offloaded into HW */
+	MLX5_ENCAP_ENTRY_VALID     = BIT(0),
+	MLX5_REFORMAT_DECAP        = BIT(1),
+	MLX5_ENCAP_ENTRY_NO_ROUTE  = BIT(2),
+};
+
+struct mlx5e_decap_key {
+	struct ethhdr key;
+};
+
+struct mlx5e_decap_entry {
+	struct mlx5e_decap_key key;
+	struct list_head flows;
+	struct hlist_node hlist;
+	refcount_t refcnt;
+	struct completion res_ready;
+	int compl_result;
+	struct mlx5_pkt_reformat *pkt_reformat;
+	struct rcu_head rcu;
+};
+
+struct mlx5e_mpls_info {
+	u32             label;
+	u8              tc;
+	u8              bos;
+	u8              ttl;
+};
+
+struct mlx5e_encap_entry {
+	/* attached neigh hash entry */
+	struct mlx5e_neigh_hash_entry *nhe;
+	/* neigh hash entry list of encaps sharing the same neigh */
+	struct list_head encap_list;
+	/* a node of the eswitch encap hash table which keeping all the encap
+	 * entries
+	 */
+	struct hlist_node encap_hlist;
+	struct list_head flows;
+	struct list_head route_list;
+	struct mlx5_pkt_reformat *pkt_reformat;
+	const struct ip_tunnel_info *tun_info;
+	struct mlx5e_mpls_info mpls_info;
+	unsigned char h_dest[ETH_ALEN];	/* destination eth addr	*/
+
+	struct net_device *out_dev;
+	int route_dev_ifindex;
+	struct mlx5e_tc_tunnel *tunnel;
+	int reformat_type;
+	u8 flags;
+	char *encap_header;
+	int encap_size;
+	refcount_t refcnt;
+	struct completion res_ready;
+	int compl_result;
+	struct rcu_head rcu;
+};
+
+struct mlx5e_rep_sq {
+	struct mlx5_flow_handle	*send_to_vport_rule;
+	struct mlx5_flow_handle *send_to_vport_rule_peer;
+	u32 sqn;
+	struct list_head	 list;
+};
+
+int mlx5e_rep_init(void);
+void mlx5e_rep_cleanup(void);
+int mlx5e_rep_bond_init(struct mlx5e_rep_priv *rpriv);
+void mlx5e_rep_bond_cleanup(struct mlx5e_rep_priv *rpriv);
+int mlx5e_rep_bond_enslave(struct mlx5_eswitch *esw, struct net_device *netdev,
+			   struct net_device *lag_dev);
+void mlx5e_rep_bond_unslave(struct mlx5_eswitch *esw,
+			    const struct net_device *netdev,
+			    const struct net_device *lag_dev);
+int mlx5e_rep_bond_update(struct mlx5e_priv *priv, bool cleanup);
+
+bool mlx5e_rep_has_offload_stats(const struct net_device *dev, int attr_id);
+int mlx5e_rep_get_offload_stats(int attr_id, const struct net_device *dev,
+				void *sp);
+
+bool mlx5e_is_uplink_rep(struct mlx5e_priv *priv);
+void mlx5e_rep_activate_channels(struct mlx5e_priv *priv);
+void mlx5e_rep_deactivate_channels(struct mlx5e_priv *priv);
+
+void mlx5e_rep_queue_neigh_stats_work(struct mlx5e_priv *priv);
+
+bool mlx5e_eswitch_vf_rep(const struct net_device *netdev);
+bool mlx5e_eswitch_uplink_rep(const struct net_device *netdev);
+static inline bool mlx5e_eswitch_rep(const struct net_device *netdev)
+{
+	return mlx5e_eswitch_vf_rep(netdev) ||
+	       mlx5e_eswitch_uplink_rep(netdev);
+}
+
+#else /* CONFIG_MLX5_ESWITCH */
+static inline bool mlx5e_is_uplink_rep(struct mlx5e_priv *priv) { return false; }
+static inline void mlx5e_rep_activate_channels(struct mlx5e_priv *priv) {}
+static inline void mlx5e_rep_deactivate_channels(struct mlx5e_priv *priv) {}
+static inline int mlx5e_rep_init(void) { return 0; };
+static inline void mlx5e_rep_cleanup(void) {};
+static inline bool mlx5e_rep_has_offload_stats(const struct net_device *dev,
+					       int attr_id) { return false; }
+static inline int mlx5e_rep_get_offload_stats(int attr_id,
+					      const struct net_device *dev,
+					      void *sp) { return -EOPNOTSUPP; }
+#endif
+
+static inline bool mlx5e_is_vport_rep(struct mlx5e_priv *priv)
+{
+	return (MLX5_ESWITCH_MANAGER(priv->mdev) && priv->ppriv);
+}
+#endif /* __MLX5E_REP_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c b/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c
new file mode 100644
index 000000000..56d1bd22c
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c
@@ -0,0 +1,2494 @@
+/*
+ * Copyright (c) 2015, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/ip.h>
+#include <linux/ipv6.h>
+#include <linux/tcp.h>
+#include <linux/bitmap.h>
+#include <linux/filter.h>
+#include <net/ip6_checksum.h>
+#include <net/page_pool.h>
+#include <net/inet_ecn.h>
+#include <net/gro.h>
+#include <net/udp.h>
+#include <net/tcp.h>
+#include <net/xdp_sock_drv.h>
+#include "en.h"
+#include "en/txrx.h"
+#include "en_tc.h"
+#include "eswitch.h"
+#include "en_rep.h"
+#include "en/rep/tc.h"
+#include "ipoib/ipoib.h"
+#include "en_accel/ipsec.h"
+#include "en_accel/macsec.h"
+#include "en_accel/ipsec_rxtx.h"
+#include "en_accel/ktls_txrx.h"
+#include "en/xdp.h"
+#include "en/xsk/rx.h"
+#include "en/health.h"
+#include "en/params.h"
+#include "devlink.h"
+#include "en/devlink.h"
+
+static struct sk_buff *
+mlx5e_skb_from_cqe_mpwrq_linear(struct mlx5e_rq *rq, struct mlx5e_mpw_info *wi,
+				u16 cqe_bcnt, u32 head_offset, u32 page_idx);
+static struct sk_buff *
+mlx5e_skb_from_cqe_mpwrq_nonlinear(struct mlx5e_rq *rq, struct mlx5e_mpw_info *wi,
+				   u16 cqe_bcnt, u32 head_offset, u32 page_idx);
+static void mlx5e_handle_rx_cqe(struct mlx5e_rq *rq, struct mlx5_cqe64 *cqe);
+static void mlx5e_handle_rx_cqe_mpwrq(struct mlx5e_rq *rq, struct mlx5_cqe64 *cqe);
+static void mlx5e_handle_rx_cqe_mpwrq_shampo(struct mlx5e_rq *rq, struct mlx5_cqe64 *cqe);
+
+const struct mlx5e_rx_handlers mlx5e_rx_handlers_nic = {
+	.handle_rx_cqe       = mlx5e_handle_rx_cqe,
+	.handle_rx_cqe_mpwqe = mlx5e_handle_rx_cqe_mpwrq,
+	.handle_rx_cqe_mpwqe_shampo = mlx5e_handle_rx_cqe_mpwrq_shampo,
+};
+
+static inline bool mlx5e_rx_hw_stamp(struct hwtstamp_config *config)
+{
+	return config->rx_filter == HWTSTAMP_FILTER_ALL;
+}
+
+static inline void mlx5e_read_cqe_slot(struct mlx5_cqwq *wq,
+				       u32 cqcc, void *data)
+{
+	u32 ci = mlx5_cqwq_ctr2ix(wq, cqcc);
+
+	memcpy(data, mlx5_cqwq_get_wqe(wq, ci), sizeof(struct mlx5_cqe64));
+}
+
+static inline void mlx5e_read_title_slot(struct mlx5e_rq *rq,
+					 struct mlx5_cqwq *wq,
+					 u32 cqcc)
+{
+	struct mlx5e_cq_decomp *cqd = &rq->cqd;
+	struct mlx5_cqe64 *title = &cqd->title;
+
+	mlx5e_read_cqe_slot(wq, cqcc, title);
+	cqd->left        = be32_to_cpu(title->byte_cnt);
+	cqd->wqe_counter = be16_to_cpu(title->wqe_counter);
+	rq->stats->cqe_compress_blks++;
+}
+
+static inline void mlx5e_read_mini_arr_slot(struct mlx5_cqwq *wq,
+					    struct mlx5e_cq_decomp *cqd,
+					    u32 cqcc)
+{
+	mlx5e_read_cqe_slot(wq, cqcc, cqd->mini_arr);
+	cqd->mini_arr_idx = 0;
+}
+
+static inline void mlx5e_cqes_update_owner(struct mlx5_cqwq *wq, int n)
+{
+	u32 cqcc   = wq->cc;
+	u8  op_own = mlx5_cqwq_get_ctr_wrap_cnt(wq, cqcc) & 1;
+	u32 ci     = mlx5_cqwq_ctr2ix(wq, cqcc);
+	u32 wq_sz  = mlx5_cqwq_get_size(wq);
+	u32 ci_top = min_t(u32, wq_sz, ci + n);
+
+	for (; ci < ci_top; ci++, n--) {
+		struct mlx5_cqe64 *cqe = mlx5_cqwq_get_wqe(wq, ci);
+
+		cqe->op_own = op_own;
+	}
+
+	if (unlikely(ci == wq_sz)) {
+		op_own = !op_own;
+		for (ci = 0; ci < n; ci++) {
+			struct mlx5_cqe64 *cqe = mlx5_cqwq_get_wqe(wq, ci);
+
+			cqe->op_own = op_own;
+		}
+	}
+}
+
+static inline void mlx5e_decompress_cqe(struct mlx5e_rq *rq,
+					struct mlx5_cqwq *wq,
+					u32 cqcc)
+{
+	struct mlx5e_cq_decomp *cqd = &rq->cqd;
+	struct mlx5_mini_cqe8 *mini_cqe = &cqd->mini_arr[cqd->mini_arr_idx];
+	struct mlx5_cqe64 *title = &cqd->title;
+
+	title->byte_cnt     = mini_cqe->byte_cnt;
+	title->check_sum    = mini_cqe->checksum;
+	title->op_own      &= 0xf0;
+	title->op_own      |= 0x01 & (cqcc >> wq->fbc.log_sz);
+
+	/* state bit set implies linked-list striding RQ wq type and
+	 * HW stride index capability supported
+	 */
+	if (test_bit(MLX5E_RQ_STATE_MINI_CQE_HW_STRIDX, &rq->state)) {
+		title->wqe_counter = mini_cqe->stridx;
+		return;
+	}
+
+	/* HW stride index capability not supported */
+	title->wqe_counter = cpu_to_be16(cqd->wqe_counter);
+	if (rq->wq_type == MLX5_WQ_TYPE_LINKED_LIST_STRIDING_RQ)
+		cqd->wqe_counter += mpwrq_get_cqe_consumed_strides(title);
+	else
+		cqd->wqe_counter =
+			mlx5_wq_cyc_ctr2ix(&rq->wqe.wq, cqd->wqe_counter + 1);
+}
+
+static inline void mlx5e_decompress_cqe_no_hash(struct mlx5e_rq *rq,
+						struct mlx5_cqwq *wq,
+						u32 cqcc)
+{
+	struct mlx5e_cq_decomp *cqd = &rq->cqd;
+
+	mlx5e_decompress_cqe(rq, wq, cqcc);
+	cqd->title.rss_hash_type   = 0;
+	cqd->title.rss_hash_result = 0;
+}
+
+static inline u32 mlx5e_decompress_cqes_cont(struct mlx5e_rq *rq,
+					     struct mlx5_cqwq *wq,
+					     int update_owner_only,
+					     int budget_rem)
+{
+	struct mlx5e_cq_decomp *cqd = &rq->cqd;
+	u32 cqcc = wq->cc + update_owner_only;
+	u32 cqe_count;
+	u32 i;
+
+	cqe_count = min_t(u32, cqd->left, budget_rem);
+
+	for (i = update_owner_only; i < cqe_count;
+	     i++, cqd->mini_arr_idx++, cqcc++) {
+		if (cqd->mini_arr_idx == MLX5_MINI_CQE_ARRAY_SIZE)
+			mlx5e_read_mini_arr_slot(wq, cqd, cqcc);
+
+		mlx5e_decompress_cqe_no_hash(rq, wq, cqcc);
+		INDIRECT_CALL_3(rq->handle_rx_cqe, mlx5e_handle_rx_cqe_mpwrq,
+				mlx5e_handle_rx_cqe_mpwrq_shampo, mlx5e_handle_rx_cqe,
+				rq, &cqd->title);
+	}
+	mlx5e_cqes_update_owner(wq, cqcc - wq->cc);
+	wq->cc = cqcc;
+	cqd->left -= cqe_count;
+	rq->stats->cqe_compress_pkts += cqe_count;
+
+	return cqe_count;
+}
+
+static inline u32 mlx5e_decompress_cqes_start(struct mlx5e_rq *rq,
+					      struct mlx5_cqwq *wq,
+					      int budget_rem)
+{
+	struct mlx5e_cq_decomp *cqd = &rq->cqd;
+	u32 cc = wq->cc;
+
+	mlx5e_read_title_slot(rq, wq, cc);
+	mlx5e_read_mini_arr_slot(wq, cqd, cc + 1);
+	mlx5e_decompress_cqe(rq, wq, cc);
+	INDIRECT_CALL_3(rq->handle_rx_cqe, mlx5e_handle_rx_cqe_mpwrq,
+			mlx5e_handle_rx_cqe_mpwrq_shampo, mlx5e_handle_rx_cqe,
+			rq, &cqd->title);
+	cqd->mini_arr_idx++;
+
+	return mlx5e_decompress_cqes_cont(rq, wq, 1, budget_rem) - 1;
+}
+
+static inline bool mlx5e_rx_cache_put(struct mlx5e_rq *rq, struct page *page)
+{
+	struct mlx5e_page_cache *cache = &rq->page_cache;
+	u32 tail_next = (cache->tail + 1) & (MLX5E_CACHE_SIZE - 1);
+	struct mlx5e_rq_stats *stats = rq->stats;
+
+	if (tail_next == cache->head) {
+		stats->cache_full++;
+		return false;
+	}
+
+	if (!dev_page_is_reusable(page)) {
+		stats->cache_waive++;
+		return false;
+	}
+
+	cache->page_cache[cache->tail] = page;
+	cache->tail = tail_next;
+	return true;
+}
+
+static inline bool mlx5e_rx_cache_get(struct mlx5e_rq *rq, union mlx5e_alloc_unit *au)
+{
+	struct mlx5e_page_cache *cache = &rq->page_cache;
+	struct mlx5e_rq_stats *stats = rq->stats;
+	dma_addr_t addr;
+
+	if (unlikely(cache->head == cache->tail)) {
+		stats->cache_empty++;
+		return false;
+	}
+
+	if (page_ref_count(cache->page_cache[cache->head]) != 1) {
+		stats->cache_busy++;
+		return false;
+	}
+
+	au->page = cache->page_cache[cache->head];
+	cache->head = (cache->head + 1) & (MLX5E_CACHE_SIZE - 1);
+	stats->cache_reuse++;
+
+	addr = page_pool_get_dma_addr(au->page);
+	/* Non-XSK always uses PAGE_SIZE. */
+	dma_sync_single_for_device(rq->pdev, addr, PAGE_SIZE, rq->buff.map_dir);
+	return true;
+}
+
+static inline int mlx5e_page_alloc_pool(struct mlx5e_rq *rq, union mlx5e_alloc_unit *au)
+{
+	dma_addr_t addr;
+
+	if (mlx5e_rx_cache_get(rq, au))
+		return 0;
+
+	au->page = page_pool_dev_alloc_pages(rq->page_pool);
+	if (unlikely(!au->page))
+		return -ENOMEM;
+
+	/* Non-XSK always uses PAGE_SIZE. */
+	addr = dma_map_page(rq->pdev, au->page, 0, PAGE_SIZE, rq->buff.map_dir);
+	if (unlikely(dma_mapping_error(rq->pdev, addr))) {
+		page_pool_recycle_direct(rq->page_pool, au->page);
+		au->page = NULL;
+		return -ENOMEM;
+	}
+	page_pool_set_dma_addr(au->page, addr);
+
+	return 0;
+}
+
+void mlx5e_page_dma_unmap(struct mlx5e_rq *rq, struct page *page)
+{
+	dma_addr_t dma_addr = page_pool_get_dma_addr(page);
+
+	dma_unmap_page_attrs(rq->pdev, dma_addr, PAGE_SIZE, rq->buff.map_dir,
+			     DMA_ATTR_SKIP_CPU_SYNC);
+	page_pool_set_dma_addr(page, 0);
+}
+
+void mlx5e_page_release_dynamic(struct mlx5e_rq *rq, struct page *page, bool recycle)
+{
+	if (likely(recycle)) {
+		if (mlx5e_rx_cache_put(rq, page))
+			return;
+
+		mlx5e_page_dma_unmap(rq, page);
+		page_pool_recycle_direct(rq->page_pool, page);
+	} else {
+		mlx5e_page_dma_unmap(rq, page);
+		page_pool_release_page(rq->page_pool, page);
+		put_page(page);
+	}
+}
+
+static inline int mlx5e_get_rx_frag(struct mlx5e_rq *rq,
+				    struct mlx5e_wqe_frag_info *frag)
+{
+	int err = 0;
+
+	if (!frag->offset)
+		/* On first frag (offset == 0), replenish page (alloc_unit actually).
+		 * Other frags that point to the same alloc_unit (with a different
+		 * offset) should just use the new one without replenishing again
+		 * by themselves.
+		 */
+		err = mlx5e_page_alloc_pool(rq, frag->au);
+
+	return err;
+}
+
+static inline void mlx5e_put_rx_frag(struct mlx5e_rq *rq,
+				     struct mlx5e_wqe_frag_info *frag,
+				     bool recycle)
+{
+	if (frag->last_in_page)
+		mlx5e_page_release_dynamic(rq, frag->au->page, recycle);
+}
+
+static inline struct mlx5e_wqe_frag_info *get_frag(struct mlx5e_rq *rq, u16 ix)
+{
+	return &rq->wqe.frags[ix << rq->wqe.info.log_num_frags];
+}
+
+static int mlx5e_alloc_rx_wqe(struct mlx5e_rq *rq, struct mlx5e_rx_wqe_cyc *wqe,
+			      u16 ix)
+{
+	struct mlx5e_wqe_frag_info *frag = get_frag(rq, ix);
+	int err;
+	int i;
+
+	for (i = 0; i < rq->wqe.info.num_frags; i++, frag++) {
+		dma_addr_t addr;
+		u16 headroom;
+
+		err = mlx5e_get_rx_frag(rq, frag);
+		if (unlikely(err))
+			goto free_frags;
+
+		headroom = i == 0 ? rq->buff.headroom : 0;
+		addr = page_pool_get_dma_addr(frag->au->page);
+		wqe->data[i].addr = cpu_to_be64(addr + frag->offset + headroom);
+	}
+
+	return 0;
+
+free_frags:
+	while (--i >= 0)
+		mlx5e_put_rx_frag(rq, --frag, true);
+
+	return err;
+}
+
+static inline void mlx5e_free_rx_wqe(struct mlx5e_rq *rq,
+				     struct mlx5e_wqe_frag_info *wi,
+				     bool recycle)
+{
+	int i;
+
+	if (rq->xsk_pool) {
+		/* The `recycle` parameter is ignored, and the page is always
+		 * put into the Reuse Ring, because there is no way to return
+		 * the page to the userspace when the interface goes down.
+		 */
+		xsk_buff_free(wi->au->xsk);
+		return;
+	}
+
+	for (i = 0; i < rq->wqe.info.num_frags; i++, wi++)
+		mlx5e_put_rx_frag(rq, wi, recycle);
+}
+
+static void mlx5e_dealloc_rx_wqe(struct mlx5e_rq *rq, u16 ix)
+{
+	struct mlx5e_wqe_frag_info *wi = get_frag(rq, ix);
+
+	mlx5e_free_rx_wqe(rq, wi, false);
+}
+
+static int mlx5e_alloc_rx_wqes(struct mlx5e_rq *rq, u16 ix, int wqe_bulk)
+{
+	struct mlx5_wq_cyc *wq = &rq->wqe.wq;
+	int i;
+
+	for (i = 0; i < wqe_bulk; i++) {
+		int j = mlx5_wq_cyc_ctr2ix(wq, ix + i);
+		struct mlx5e_rx_wqe_cyc *wqe;
+
+		wqe = mlx5_wq_cyc_get_wqe(wq, j);
+
+		if (unlikely(mlx5e_alloc_rx_wqe(rq, wqe, j)))
+			break;
+	}
+
+	return i;
+}
+
+static inline void
+mlx5e_add_skb_frag(struct mlx5e_rq *rq, struct sk_buff *skb,
+		   union mlx5e_alloc_unit *au, u32 frag_offset, u32 len,
+		   unsigned int truesize)
+{
+	dma_addr_t addr = page_pool_get_dma_addr(au->page);
+
+	dma_sync_single_for_cpu(rq->pdev, addr + frag_offset, len,
+				rq->buff.map_dir);
+	page_ref_inc(au->page);
+	skb_add_rx_frag(skb, skb_shinfo(skb)->nr_frags,
+			au->page, frag_offset, len, truesize);
+}
+
+static inline void
+mlx5e_copy_skb_header(struct mlx5e_rq *rq, struct sk_buff *skb,
+		      struct page *page, dma_addr_t addr,
+		      int offset_from, int dma_offset, u32 headlen)
+{
+	const void *from = page_address(page) + offset_from;
+	/* Aligning len to sizeof(long) optimizes memcpy performance */
+	unsigned int len = ALIGN(headlen, sizeof(long));
+
+	dma_sync_single_for_cpu(rq->pdev, addr + dma_offset, len,
+				rq->buff.map_dir);
+	skb_copy_to_linear_data(skb, from, len);
+}
+
+static void
+mlx5e_free_rx_mpwqe(struct mlx5e_rq *rq, struct mlx5e_mpw_info *wi, bool recycle)
+{
+	union mlx5e_alloc_unit *alloc_units = wi->alloc_units;
+	bool no_xdp_xmit;
+	int i;
+
+	/* A common case for AF_XDP. */
+	if (bitmap_full(wi->xdp_xmit_bitmap, rq->mpwqe.pages_per_wqe))
+		return;
+
+	no_xdp_xmit = bitmap_empty(wi->xdp_xmit_bitmap, rq->mpwqe.pages_per_wqe);
+
+	if (rq->xsk_pool) {
+		/* The `recycle` parameter is ignored, and the page is always
+		 * put into the Reuse Ring, because there is no way to return
+		 * the page to the userspace when the interface goes down.
+		 */
+		for (i = 0; i < rq->mpwqe.pages_per_wqe; i++)
+			if (no_xdp_xmit || !test_bit(i, wi->xdp_xmit_bitmap))
+				xsk_buff_free(alloc_units[i].xsk);
+	} else {
+		for (i = 0; i < rq->mpwqe.pages_per_wqe; i++)
+			if (no_xdp_xmit || !test_bit(i, wi->xdp_xmit_bitmap))
+				mlx5e_page_release_dynamic(rq, alloc_units[i].page, recycle);
+	}
+}
+
+static void mlx5e_post_rx_mpwqe(struct mlx5e_rq *rq, u8 n)
+{
+	struct mlx5_wq_ll *wq = &rq->mpwqe.wq;
+
+	do {
+		u16 next_wqe_index = mlx5_wq_ll_get_wqe_next_ix(wq, wq->head);
+
+		mlx5_wq_ll_push(wq, next_wqe_index);
+	} while (--n);
+
+	/* ensure wqes are visible to device before updating doorbell record */
+	dma_wmb();
+
+	mlx5_wq_ll_update_db_record(wq);
+}
+
+/* This function returns the size of the continuous free space inside a bitmap
+ * that starts from first and no longer than len including circular ones.
+ */
+static int bitmap_find_window(unsigned long *bitmap, int len,
+			      int bitmap_size, int first)
+{
+	int next_one, count;
+
+	next_one = find_next_bit(bitmap, bitmap_size, first);
+	if (next_one == bitmap_size) {
+		if (bitmap_size - first >= len)
+			return len;
+		next_one = find_next_bit(bitmap, bitmap_size, 0);
+		count = next_one + bitmap_size - first;
+	} else {
+		count = next_one - first;
+	}
+
+	return min(len, count);
+}
+
+static void build_klm_umr(struct mlx5e_icosq *sq, struct mlx5e_umr_wqe *umr_wqe,
+			  __be32 key, u16 offset, u16 klm_len, u16 wqe_bbs)
+{
+	memset(umr_wqe, 0, offsetof(struct mlx5e_umr_wqe, inline_klms));
+	umr_wqe->ctrl.opmod_idx_opcode =
+		cpu_to_be32((sq->pc << MLX5_WQE_CTRL_WQE_INDEX_SHIFT) |
+			     MLX5_OPCODE_UMR);
+	umr_wqe->ctrl.umr_mkey = key;
+	umr_wqe->ctrl.qpn_ds = cpu_to_be32((sq->sqn << MLX5_WQE_CTRL_QPN_SHIFT)
+					    | MLX5E_KLM_UMR_DS_CNT(klm_len));
+	umr_wqe->uctrl.flags = MLX5_UMR_TRANSLATION_OFFSET_EN | MLX5_UMR_INLINE;
+	umr_wqe->uctrl.xlt_offset = cpu_to_be16(offset);
+	umr_wqe->uctrl.xlt_octowords = cpu_to_be16(klm_len);
+	umr_wqe->uctrl.mkey_mask     = cpu_to_be64(MLX5_MKEY_MASK_FREE);
+}
+
+static int mlx5e_build_shampo_hd_umr(struct mlx5e_rq *rq,
+				     struct mlx5e_icosq *sq,
+				     u16 klm_entries, u16 index)
+{
+	struct mlx5e_shampo_hd *shampo = rq->mpwqe.shampo;
+	u16 entries, pi, header_offset, err, wqe_bbs, new_entries;
+	u32 lkey = rq->mdev->mlx5e_res.hw_objs.mkey;
+	struct page *page = shampo->last_page;
+	u64 addr = shampo->last_addr;
+	struct mlx5e_dma_info *dma_info;
+	struct mlx5e_umr_wqe *umr_wqe;
+	int headroom, i;
+
+	headroom = rq->buff.headroom;
+	new_entries = klm_entries - (shampo->pi & (MLX5_UMR_KLM_ALIGNMENT - 1));
+	entries = ALIGN(klm_entries, MLX5_UMR_KLM_ALIGNMENT);
+	wqe_bbs = MLX5E_KLM_UMR_WQEBBS(entries);
+	pi = mlx5e_icosq_get_next_pi(sq, wqe_bbs);
+	umr_wqe = mlx5_wq_cyc_get_wqe(&sq->wq, pi);
+	build_klm_umr(sq, umr_wqe, shampo->key, index, entries, wqe_bbs);
+
+	for (i = 0; i < entries; i++, index++) {
+		dma_info = &shampo->info[index];
+		if (i >= klm_entries || (index < shampo->pi && shampo->pi - index <
+					 MLX5_UMR_KLM_ALIGNMENT))
+			goto update_klm;
+		header_offset = (index & (MLX5E_SHAMPO_WQ_HEADER_PER_PAGE - 1)) <<
+			MLX5E_SHAMPO_LOG_MAX_HEADER_ENTRY_SIZE;
+		if (!(header_offset & (PAGE_SIZE - 1))) {
+			union mlx5e_alloc_unit au;
+
+			err = mlx5e_page_alloc_pool(rq, &au);
+			if (unlikely(err))
+				goto err_unmap;
+			page = dma_info->page = au.page;
+			addr = dma_info->addr = page_pool_get_dma_addr(au.page);
+		} else {
+			dma_info->addr = addr + header_offset;
+			dma_info->page = page;
+		}
+
+update_klm:
+		umr_wqe->inline_klms[i].bcount =
+			cpu_to_be32(MLX5E_RX_MAX_HEAD);
+		umr_wqe->inline_klms[i].key    = cpu_to_be32(lkey);
+		umr_wqe->inline_klms[i].va     =
+			cpu_to_be64(dma_info->addr + headroom);
+	}
+
+	sq->db.wqe_info[pi] = (struct mlx5e_icosq_wqe_info) {
+		.wqe_type	= MLX5E_ICOSQ_WQE_SHAMPO_HD_UMR,
+		.num_wqebbs	= wqe_bbs,
+		.shampo.len	= new_entries,
+	};
+
+	shampo->pi = (shampo->pi + new_entries) & (shampo->hd_per_wq - 1);
+	shampo->last_page = page;
+	shampo->last_addr = addr;
+	sq->pc += wqe_bbs;
+	sq->doorbell_cseg = &umr_wqe->ctrl;
+
+	return 0;
+
+err_unmap:
+	while (--i >= 0) {
+		dma_info = &shampo->info[--index];
+		if (!(i & (MLX5E_SHAMPO_WQ_HEADER_PER_PAGE - 1))) {
+			dma_info->addr = ALIGN_DOWN(dma_info->addr, PAGE_SIZE);
+			mlx5e_page_release_dynamic(rq, dma_info->page, true);
+		}
+	}
+	rq->stats->buff_alloc_err++;
+	return err;
+}
+
+static int mlx5e_alloc_rx_hd_mpwqe(struct mlx5e_rq *rq)
+{
+	struct mlx5e_shampo_hd *shampo = rq->mpwqe.shampo;
+	u16 klm_entries, num_wqe, index, entries_before;
+	struct mlx5e_icosq *sq = rq->icosq;
+	int i, err, max_klm_entries, len;
+
+	max_klm_entries = MLX5E_MAX_KLM_PER_WQE(rq->mdev);
+	klm_entries = bitmap_find_window(shampo->bitmap,
+					 shampo->hd_per_wqe,
+					 shampo->hd_per_wq, shampo->pi);
+	if (!klm_entries)
+		return 0;
+
+	klm_entries += (shampo->pi & (MLX5_UMR_KLM_ALIGNMENT - 1));
+	index = ALIGN_DOWN(shampo->pi, MLX5_UMR_KLM_ALIGNMENT);
+	entries_before = shampo->hd_per_wq - index;
+
+	if (unlikely(entries_before < klm_entries))
+		num_wqe = DIV_ROUND_UP(entries_before, max_klm_entries) +
+			  DIV_ROUND_UP(klm_entries - entries_before, max_klm_entries);
+	else
+		num_wqe = DIV_ROUND_UP(klm_entries, max_klm_entries);
+
+	for (i = 0; i < num_wqe; i++) {
+		len = (klm_entries > max_klm_entries) ? max_klm_entries :
+							klm_entries;
+		if (unlikely(index + len > shampo->hd_per_wq))
+			len = shampo->hd_per_wq - index;
+		err = mlx5e_build_shampo_hd_umr(rq, sq, len, index);
+		if (unlikely(err))
+			return err;
+		index = (index + len) & (rq->mpwqe.shampo->hd_per_wq - 1);
+		klm_entries -= len;
+	}
+
+	return 0;
+}
+
+static int mlx5e_alloc_rx_mpwqe(struct mlx5e_rq *rq, u16 ix)
+{
+	struct mlx5e_mpw_info *wi = mlx5e_get_mpw_info(rq, ix);
+	union mlx5e_alloc_unit *au = &wi->alloc_units[0];
+	struct mlx5e_icosq *sq = rq->icosq;
+	struct mlx5_wq_cyc *wq = &sq->wq;
+	struct mlx5e_umr_wqe *umr_wqe;
+	u32 offset; /* 17-bit value with MTT. */
+	u16 pi;
+	int err;
+	int i;
+
+	if (test_bit(MLX5E_RQ_STATE_SHAMPO, &rq->state)) {
+		err = mlx5e_alloc_rx_hd_mpwqe(rq);
+		if (unlikely(err))
+			goto err;
+	}
+
+	pi = mlx5e_icosq_get_next_pi(sq, rq->mpwqe.umr_wqebbs);
+	umr_wqe = mlx5_wq_cyc_get_wqe(wq, pi);
+	memcpy(umr_wqe, &rq->mpwqe.umr_wqe, sizeof(struct mlx5e_umr_wqe));
+
+	for (i = 0; i < rq->mpwqe.pages_per_wqe; i++, au++) {
+		dma_addr_t addr;
+
+		err = mlx5e_page_alloc_pool(rq, au);
+		if (unlikely(err))
+			goto err_unmap;
+		addr = page_pool_get_dma_addr(au->page);
+		umr_wqe->inline_mtts[i] = (struct mlx5_mtt) {
+			.ptag = cpu_to_be64(addr | MLX5_EN_WR),
+		};
+	}
+
+	bitmap_zero(wi->xdp_xmit_bitmap, rq->mpwqe.pages_per_wqe);
+	wi->consumed_strides = 0;
+
+	umr_wqe->ctrl.opmod_idx_opcode =
+		cpu_to_be32((sq->pc << MLX5_WQE_CTRL_WQE_INDEX_SHIFT) |
+			    MLX5_OPCODE_UMR);
+
+	offset = (ix * rq->mpwqe.mtts_per_wqe) * sizeof(struct mlx5_mtt) / MLX5_OCTWORD;
+	umr_wqe->uctrl.xlt_offset = cpu_to_be16(offset);
+
+	sq->db.wqe_info[pi] = (struct mlx5e_icosq_wqe_info) {
+		.wqe_type   = MLX5E_ICOSQ_WQE_UMR_RX,
+		.num_wqebbs = rq->mpwqe.umr_wqebbs,
+		.umr.rq     = rq,
+	};
+
+	sq->pc += rq->mpwqe.umr_wqebbs;
+
+	sq->doorbell_cseg = &umr_wqe->ctrl;
+
+	return 0;
+
+err_unmap:
+	while (--i >= 0) {
+		au--;
+		mlx5e_page_release_dynamic(rq, au->page, true);
+	}
+
+err:
+	rq->stats->buff_alloc_err++;
+
+	return err;
+}
+
+/* This function is responsible to dealloc SHAMPO header buffer.
+ * close == true specifies that we are in the middle of closing RQ operation so
+ * we go over all the entries and if they are not in use we free them,
+ * otherwise we only go over a specific range inside the header buffer that are
+ * not in use.
+ */
+void mlx5e_shampo_dealloc_hd(struct mlx5e_rq *rq, u16 len, u16 start, bool close)
+{
+	struct mlx5e_shampo_hd *shampo = rq->mpwqe.shampo;
+	int hd_per_wq = shampo->hd_per_wq;
+	struct page *deleted_page = NULL;
+	struct mlx5e_dma_info *hd_info;
+	int i, index = start;
+
+	for (i = 0; i < len; i++, index++) {
+		if (index == hd_per_wq)
+			index = 0;
+
+		if (close && !test_bit(index, shampo->bitmap))
+			continue;
+
+		hd_info = &shampo->info[index];
+		hd_info->addr = ALIGN_DOWN(hd_info->addr, PAGE_SIZE);
+		if (hd_info->page != deleted_page) {
+			deleted_page = hd_info->page;
+			mlx5e_page_release_dynamic(rq, hd_info->page, false);
+		}
+	}
+
+	if (start + len > hd_per_wq) {
+		len -= hd_per_wq - start;
+		bitmap_clear(shampo->bitmap, start, hd_per_wq - start);
+		start = 0;
+	}
+
+	bitmap_clear(shampo->bitmap, start, len);
+}
+
+static void mlx5e_dealloc_rx_mpwqe(struct mlx5e_rq *rq, u16 ix)
+{
+	struct mlx5e_mpw_info *wi = mlx5e_get_mpw_info(rq, ix);
+	/* Don't recycle, this function is called on rq/netdev close */
+	mlx5e_free_rx_mpwqe(rq, wi, false);
+}
+
+INDIRECT_CALLABLE_SCOPE bool mlx5e_post_rx_wqes(struct mlx5e_rq *rq)
+{
+	struct mlx5_wq_cyc *wq = &rq->wqe.wq;
+	int wqe_bulk, count;
+	bool busy = false;
+	u16 head;
+
+	if (unlikely(!test_bit(MLX5E_RQ_STATE_ENABLED, &rq->state)))
+		return false;
+
+	if (mlx5_wq_cyc_missing(wq) < rq->wqe.info.wqe_bulk)
+		return false;
+
+	if (rq->page_pool)
+		page_pool_nid_changed(rq->page_pool, numa_mem_id());
+
+	wqe_bulk = mlx5_wq_cyc_missing(wq);
+	head = mlx5_wq_cyc_get_head(wq);
+
+	/* Don't allow any newly allocated WQEs to share the same page with old
+	 * WQEs that aren't completed yet. Stop earlier.
+	 */
+	wqe_bulk -= (head + wqe_bulk) & rq->wqe.info.wqe_index_mask;
+
+	if (!rq->xsk_pool)
+		count = mlx5e_alloc_rx_wqes(rq, head, wqe_bulk);
+	else if (likely(!rq->xsk_pool->dma_need_sync))
+		count = mlx5e_xsk_alloc_rx_wqes_batched(rq, head, wqe_bulk);
+	else
+		/* If dma_need_sync is true, it's more efficient to call
+		 * xsk_buff_alloc in a loop, rather than xsk_buff_alloc_batch,
+		 * because the latter does the same check and returns only one
+		 * frame.
+		 */
+		count = mlx5e_xsk_alloc_rx_wqes(rq, head, wqe_bulk);
+
+	mlx5_wq_cyc_push_n(wq, count);
+	if (unlikely(count != wqe_bulk)) {
+		rq->stats->buff_alloc_err++;
+		busy = true;
+	}
+
+	/* ensure wqes are visible to device before updating doorbell record */
+	dma_wmb();
+
+	mlx5_wq_cyc_update_db_record(wq);
+
+	return busy;
+}
+
+void mlx5e_free_icosq_descs(struct mlx5e_icosq *sq)
+{
+	u16 sqcc;
+
+	sqcc = sq->cc;
+
+	while (sqcc != sq->pc) {
+		struct mlx5e_icosq_wqe_info *wi;
+		u16 ci;
+
+		ci = mlx5_wq_cyc_ctr2ix(&sq->wq, sqcc);
+		wi = &sq->db.wqe_info[ci];
+		sqcc += wi->num_wqebbs;
+#ifdef CONFIG_MLX5_EN_TLS
+		switch (wi->wqe_type) {
+		case MLX5E_ICOSQ_WQE_SET_PSV_TLS:
+			mlx5e_ktls_handle_ctx_completion(wi);
+			break;
+		case MLX5E_ICOSQ_WQE_GET_PSV_TLS:
+			mlx5e_ktls_handle_get_psv_completion(wi, sq);
+			break;
+		}
+#endif
+	}
+	sq->cc = sqcc;
+}
+
+static void mlx5e_handle_shampo_hd_umr(struct mlx5e_shampo_umr umr,
+				       struct mlx5e_icosq *sq)
+{
+	struct mlx5e_channel *c = container_of(sq, struct mlx5e_channel, icosq);
+	struct mlx5e_shampo_hd *shampo;
+	/* assume 1:1 relationship between RQ and icosq */
+	struct mlx5e_rq *rq = &c->rq;
+	int end, from, len = umr.len;
+
+	shampo = rq->mpwqe.shampo;
+	end = shampo->hd_per_wq;
+	from = shampo->ci;
+	if (from + len > shampo->hd_per_wq) {
+		len -= end - from;
+		bitmap_set(shampo->bitmap, from, end - from);
+		from = 0;
+	}
+
+	bitmap_set(shampo->bitmap, from, len);
+	shampo->ci = (shampo->ci + umr.len) & (shampo->hd_per_wq - 1);
+}
+
+int mlx5e_poll_ico_cq(struct mlx5e_cq *cq)
+{
+	struct mlx5e_icosq *sq = container_of(cq, struct mlx5e_icosq, cq);
+	struct mlx5_cqe64 *cqe;
+	u16 sqcc;
+	int i;
+
+	if (unlikely(!test_bit(MLX5E_SQ_STATE_ENABLED, &sq->state)))
+		return 0;
+
+	cqe = mlx5_cqwq_get_cqe(&cq->wq);
+	if (likely(!cqe))
+		return 0;
+
+	/* sq->cc must be updated only after mlx5_cqwq_update_db_record(),
+	 * otherwise a cq overrun may occur
+	 */
+	sqcc = sq->cc;
+
+	i = 0;
+	do {
+		u16 wqe_counter;
+		bool last_wqe;
+
+		mlx5_cqwq_pop(&cq->wq);
+
+		wqe_counter = be16_to_cpu(cqe->wqe_counter);
+
+		do {
+			struct mlx5e_icosq_wqe_info *wi;
+			u16 ci;
+
+			last_wqe = (sqcc == wqe_counter);
+
+			ci = mlx5_wq_cyc_ctr2ix(&sq->wq, sqcc);
+			wi = &sq->db.wqe_info[ci];
+			sqcc += wi->num_wqebbs;
+
+			if (last_wqe && unlikely(get_cqe_opcode(cqe) != MLX5_CQE_REQ)) {
+				netdev_WARN_ONCE(cq->netdev,
+						 "Bad OP in ICOSQ CQE: 0x%x\n",
+						 get_cqe_opcode(cqe));
+				mlx5e_dump_error_cqe(&sq->cq, sq->sqn,
+						     (struct mlx5_err_cqe *)cqe);
+				mlx5_wq_cyc_wqe_dump(&sq->wq, ci, wi->num_wqebbs);
+				if (!test_and_set_bit(MLX5E_SQ_STATE_RECOVERING, &sq->state))
+					queue_work(cq->priv->wq, &sq->recover_work);
+				break;
+			}
+
+			switch (wi->wqe_type) {
+			case MLX5E_ICOSQ_WQE_UMR_RX:
+				wi->umr.rq->mpwqe.umr_completed++;
+				break;
+			case MLX5E_ICOSQ_WQE_NOP:
+				break;
+			case MLX5E_ICOSQ_WQE_SHAMPO_HD_UMR:
+				mlx5e_handle_shampo_hd_umr(wi->shampo, sq);
+				break;
+#ifdef CONFIG_MLX5_EN_TLS
+			case MLX5E_ICOSQ_WQE_UMR_TLS:
+				break;
+			case MLX5E_ICOSQ_WQE_SET_PSV_TLS:
+				mlx5e_ktls_handle_ctx_completion(wi);
+				break;
+			case MLX5E_ICOSQ_WQE_GET_PSV_TLS:
+				mlx5e_ktls_handle_get_psv_completion(wi, sq);
+				break;
+#endif
+			default:
+				netdev_WARN_ONCE(cq->netdev,
+						 "Bad WQE type in ICOSQ WQE info: 0x%x\n",
+						 wi->wqe_type);
+			}
+		} while (!last_wqe);
+	} while ((++i < MLX5E_TX_CQ_POLL_BUDGET) && (cqe = mlx5_cqwq_get_cqe(&cq->wq)));
+
+	sq->cc = sqcc;
+
+	mlx5_cqwq_update_db_record(&cq->wq);
+
+	return i;
+}
+
+INDIRECT_CALLABLE_SCOPE bool mlx5e_post_rx_mpwqes(struct mlx5e_rq *rq)
+{
+	struct mlx5_wq_ll *wq = &rq->mpwqe.wq;
+	u8  umr_completed = rq->mpwqe.umr_completed;
+	struct mlx5e_icosq *sq = rq->icosq;
+	int alloc_err = 0;
+	u8  missing, i;
+	u16 head;
+
+	if (unlikely(!test_bit(MLX5E_RQ_STATE_ENABLED, &rq->state)))
+		return false;
+
+	if (umr_completed) {
+		mlx5e_post_rx_mpwqe(rq, umr_completed);
+		rq->mpwqe.umr_in_progress -= umr_completed;
+		rq->mpwqe.umr_completed = 0;
+	}
+
+	missing = mlx5_wq_ll_missing(wq) - rq->mpwqe.umr_in_progress;
+
+	if (unlikely(rq->mpwqe.umr_in_progress > rq->mpwqe.umr_last_bulk))
+		rq->stats->congst_umr++;
+
+	if (likely(missing < rq->mpwqe.min_wqe_bulk))
+		return false;
+
+	if (rq->page_pool)
+		page_pool_nid_changed(rq->page_pool, numa_mem_id());
+
+	head = rq->mpwqe.actual_wq_head;
+	i = missing;
+	do {
+		alloc_err = rq->xsk_pool ? mlx5e_xsk_alloc_rx_mpwqe(rq, head) :
+					   mlx5e_alloc_rx_mpwqe(rq, head);
+
+		if (unlikely(alloc_err))
+			break;
+		head = mlx5_wq_ll_get_wqe_next_ix(wq, head);
+	} while (--i);
+
+	rq->mpwqe.umr_last_bulk    = missing - i;
+	if (sq->doorbell_cseg) {
+		mlx5e_notify_hw(&sq->wq, sq->pc, sq->uar_map, sq->doorbell_cseg);
+		sq->doorbell_cseg = NULL;
+	}
+
+	rq->mpwqe.umr_in_progress += rq->mpwqe.umr_last_bulk;
+	rq->mpwqe.actual_wq_head   = head;
+
+	/* If XSK Fill Ring doesn't have enough frames, report the error, so
+	 * that one of the actions can be performed:
+	 * 1. If need_wakeup is used, signal that the application has to kick
+	 * the driver when it refills the Fill Ring.
+	 * 2. Otherwise, busy poll by rescheduling the NAPI poll.
+	 */
+	if (unlikely(alloc_err == -ENOMEM && rq->xsk_pool))
+		return true;
+
+	return false;
+}
+
+static void mlx5e_lro_update_tcp_hdr(struct mlx5_cqe64 *cqe, struct tcphdr *tcp)
+{
+	u8 l4_hdr_type = get_cqe_l4_hdr_type(cqe);
+	u8 tcp_ack     = (l4_hdr_type == CQE_L4_HDR_TYPE_TCP_ACK_NO_DATA) ||
+			 (l4_hdr_type == CQE_L4_HDR_TYPE_TCP_ACK_AND_DATA);
+
+	tcp->check                      = 0;
+	tcp->psh                        = get_cqe_lro_tcppsh(cqe);
+
+	if (tcp_ack) {
+		tcp->ack                = 1;
+		tcp->ack_seq            = cqe->lro.ack_seq_num;
+		tcp->window             = cqe->lro.tcp_win;
+	}
+}
+
+static void mlx5e_lro_update_hdr(struct sk_buff *skb, struct mlx5_cqe64 *cqe,
+				 u32 cqe_bcnt)
+{
+	struct ethhdr	*eth = (struct ethhdr *)(skb->data);
+	struct tcphdr	*tcp;
+	int network_depth = 0;
+	__wsum check;
+	__be16 proto;
+	u16 tot_len;
+	void *ip_p;
+
+	proto = __vlan_get_protocol(skb, eth->h_proto, &network_depth);
+
+	tot_len = cqe_bcnt - network_depth;
+	ip_p = skb->data + network_depth;
+
+	if (proto == htons(ETH_P_IP)) {
+		struct iphdr *ipv4 = ip_p;
+
+		tcp = ip_p + sizeof(struct iphdr);
+		skb_shinfo(skb)->gso_type = SKB_GSO_TCPV4;
+
+		ipv4->ttl               = cqe->lro.min_ttl;
+		ipv4->tot_len           = cpu_to_be16(tot_len);
+		ipv4->check             = 0;
+		ipv4->check             = ip_fast_csum((unsigned char *)ipv4,
+						       ipv4->ihl);
+
+		mlx5e_lro_update_tcp_hdr(cqe, tcp);
+		check = csum_partial(tcp, tcp->doff * 4,
+				     csum_unfold((__force __sum16)cqe->check_sum));
+		/* Almost done, don't forget the pseudo header */
+		tcp->check = csum_tcpudp_magic(ipv4->saddr, ipv4->daddr,
+					       tot_len - sizeof(struct iphdr),
+					       IPPROTO_TCP, check);
+	} else {
+		u16 payload_len = tot_len - sizeof(struct ipv6hdr);
+		struct ipv6hdr *ipv6 = ip_p;
+
+		tcp = ip_p + sizeof(struct ipv6hdr);
+		skb_shinfo(skb)->gso_type = SKB_GSO_TCPV6;
+
+		ipv6->hop_limit         = cqe->lro.min_ttl;
+		ipv6->payload_len       = cpu_to_be16(payload_len);
+
+		mlx5e_lro_update_tcp_hdr(cqe, tcp);
+		check = csum_partial(tcp, tcp->doff * 4,
+				     csum_unfold((__force __sum16)cqe->check_sum));
+		/* Almost done, don't forget the pseudo header */
+		tcp->check = csum_ipv6_magic(&ipv6->saddr, &ipv6->daddr, payload_len,
+					     IPPROTO_TCP, check);
+	}
+}
+
+static void *mlx5e_shampo_get_packet_hd(struct mlx5e_rq *rq, u16 header_index)
+{
+	struct mlx5e_dma_info *last_head = &rq->mpwqe.shampo->info[header_index];
+	u16 head_offset = (last_head->addr & (PAGE_SIZE - 1)) + rq->buff.headroom;
+
+	return page_address(last_head->page) + head_offset;
+}
+
+static void mlx5e_shampo_update_ipv4_udp_hdr(struct mlx5e_rq *rq, struct iphdr *ipv4)
+{
+	int udp_off = rq->hw_gro_data->fk.control.thoff;
+	struct sk_buff *skb = rq->hw_gro_data->skb;
+	struct udphdr *uh;
+
+	uh = (struct udphdr *)(skb->data + udp_off);
+	uh->len = htons(skb->len - udp_off);
+
+	if (uh->check)
+		uh->check = ~udp_v4_check(skb->len - udp_off, ipv4->saddr,
+					  ipv4->daddr, 0);
+
+	skb->csum_start = (unsigned char *)uh - skb->head;
+	skb->csum_offset = offsetof(struct udphdr, check);
+
+	skb_shinfo(skb)->gso_type |= SKB_GSO_UDP_L4;
+}
+
+static void mlx5e_shampo_update_ipv6_udp_hdr(struct mlx5e_rq *rq, struct ipv6hdr *ipv6)
+{
+	int udp_off = rq->hw_gro_data->fk.control.thoff;
+	struct sk_buff *skb = rq->hw_gro_data->skb;
+	struct udphdr *uh;
+
+	uh = (struct udphdr *)(skb->data + udp_off);
+	uh->len = htons(skb->len - udp_off);
+
+	if (uh->check)
+		uh->check = ~udp_v6_check(skb->len - udp_off, &ipv6->saddr,
+					  &ipv6->daddr, 0);
+
+	skb->csum_start = (unsigned char *)uh - skb->head;
+	skb->csum_offset = offsetof(struct udphdr, check);
+
+	skb_shinfo(skb)->gso_type |= SKB_GSO_UDP_L4;
+}
+
+static void mlx5e_shampo_update_fin_psh_flags(struct mlx5e_rq *rq, struct mlx5_cqe64 *cqe,
+					      struct tcphdr *skb_tcp_hd)
+{
+	u16 header_index = mlx5e_shampo_get_cqe_header_index(rq, cqe);
+	struct tcphdr *last_tcp_hd;
+	void *last_hd_addr;
+
+	last_hd_addr = mlx5e_shampo_get_packet_hd(rq, header_index);
+	last_tcp_hd =  last_hd_addr + ETH_HLEN + rq->hw_gro_data->fk.control.thoff;
+	tcp_flag_word(skb_tcp_hd) |= tcp_flag_word(last_tcp_hd) & (TCP_FLAG_FIN | TCP_FLAG_PSH);
+}
+
+static void mlx5e_shampo_update_ipv4_tcp_hdr(struct mlx5e_rq *rq, struct iphdr *ipv4,
+					     struct mlx5_cqe64 *cqe, bool match)
+{
+	int tcp_off = rq->hw_gro_data->fk.control.thoff;
+	struct sk_buff *skb = rq->hw_gro_data->skb;
+	struct tcphdr *tcp;
+
+	tcp = (struct tcphdr *)(skb->data + tcp_off);
+	if (match)
+		mlx5e_shampo_update_fin_psh_flags(rq, cqe, tcp);
+
+	tcp->check = ~tcp_v4_check(skb->len - tcp_off, ipv4->saddr,
+				   ipv4->daddr, 0);
+	skb_shinfo(skb)->gso_type |= SKB_GSO_TCPV4;
+	if (ntohs(ipv4->id) == rq->hw_gro_data->second_ip_id)
+		skb_shinfo(skb)->gso_type |= SKB_GSO_TCP_FIXEDID;
+
+	skb->csum_start = (unsigned char *)tcp - skb->head;
+	skb->csum_offset = offsetof(struct tcphdr, check);
+
+	if (tcp->cwr)
+		skb_shinfo(skb)->gso_type |= SKB_GSO_TCP_ECN;
+}
+
+static void mlx5e_shampo_update_ipv6_tcp_hdr(struct mlx5e_rq *rq, struct ipv6hdr *ipv6,
+					     struct mlx5_cqe64 *cqe, bool match)
+{
+	int tcp_off = rq->hw_gro_data->fk.control.thoff;
+	struct sk_buff *skb = rq->hw_gro_data->skb;
+	struct tcphdr *tcp;
+
+	tcp = (struct tcphdr *)(skb->data + tcp_off);
+	if (match)
+		mlx5e_shampo_update_fin_psh_flags(rq, cqe, tcp);
+
+	tcp->check = ~tcp_v6_check(skb->len - tcp_off, &ipv6->saddr,
+				   &ipv6->daddr, 0);
+	skb_shinfo(skb)->gso_type |= SKB_GSO_TCPV6;
+	skb->csum_start = (unsigned char *)tcp - skb->head;
+	skb->csum_offset = offsetof(struct tcphdr, check);
+
+	if (tcp->cwr)
+		skb_shinfo(skb)->gso_type |= SKB_GSO_TCP_ECN;
+}
+
+static void mlx5e_shampo_update_hdr(struct mlx5e_rq *rq, struct mlx5_cqe64 *cqe, bool match)
+{
+	bool is_ipv4 = (rq->hw_gro_data->fk.basic.n_proto == htons(ETH_P_IP));
+	struct sk_buff *skb = rq->hw_gro_data->skb;
+
+	skb_shinfo(skb)->gso_segs = NAPI_GRO_CB(skb)->count;
+	skb->ip_summed = CHECKSUM_PARTIAL;
+
+	if (is_ipv4) {
+		int nhoff = rq->hw_gro_data->fk.control.thoff - sizeof(struct iphdr);
+		struct iphdr *ipv4 = (struct iphdr *)(skb->data + nhoff);
+		__be16 newlen = htons(skb->len - nhoff);
+
+		csum_replace2(&ipv4->check, ipv4->tot_len, newlen);
+		ipv4->tot_len = newlen;
+
+		if (ipv4->protocol == IPPROTO_TCP)
+			mlx5e_shampo_update_ipv4_tcp_hdr(rq, ipv4, cqe, match);
+		else
+			mlx5e_shampo_update_ipv4_udp_hdr(rq, ipv4);
+	} else {
+		int nhoff = rq->hw_gro_data->fk.control.thoff - sizeof(struct ipv6hdr);
+		struct ipv6hdr *ipv6 = (struct ipv6hdr *)(skb->data + nhoff);
+
+		ipv6->payload_len = htons(skb->len - nhoff - sizeof(*ipv6));
+
+		if (ipv6->nexthdr == IPPROTO_TCP)
+			mlx5e_shampo_update_ipv6_tcp_hdr(rq, ipv6, cqe, match);
+		else
+			mlx5e_shampo_update_ipv6_udp_hdr(rq, ipv6);
+	}
+}
+
+static inline void mlx5e_skb_set_hash(struct mlx5_cqe64 *cqe,
+				      struct sk_buff *skb)
+{
+	u8 cht = cqe->rss_hash_type;
+	int ht = (cht & CQE_RSS_HTYPE_L4) ? PKT_HASH_TYPE_L4 :
+		 (cht & CQE_RSS_HTYPE_IP) ? PKT_HASH_TYPE_L3 :
+					    PKT_HASH_TYPE_NONE;
+	skb_set_hash(skb, be32_to_cpu(cqe->rss_hash_result), ht);
+}
+
+static inline bool is_last_ethertype_ip(struct sk_buff *skb, int *network_depth,
+					__be16 *proto)
+{
+	*proto = ((struct ethhdr *)skb->data)->h_proto;
+	*proto = __vlan_get_protocol(skb, *proto, network_depth);
+
+	if (*proto == htons(ETH_P_IP))
+		return pskb_may_pull(skb, *network_depth + sizeof(struct iphdr));
+
+	if (*proto == htons(ETH_P_IPV6))
+		return pskb_may_pull(skb, *network_depth + sizeof(struct ipv6hdr));
+
+	return false;
+}
+
+static inline void mlx5e_enable_ecn(struct mlx5e_rq *rq, struct sk_buff *skb)
+{
+	int network_depth = 0;
+	__be16 proto;
+	void *ip;
+	int rc;
+
+	if (unlikely(!is_last_ethertype_ip(skb, &network_depth, &proto)))
+		return;
+
+	ip = skb->data + network_depth;
+	rc = ((proto == htons(ETH_P_IP)) ? IP_ECN_set_ce((struct iphdr *)ip) :
+					 IP6_ECN_set_ce(skb, (struct ipv6hdr *)ip));
+
+	rq->stats->ecn_mark += !!rc;
+}
+
+static u8 get_ip_proto(struct sk_buff *skb, int network_depth, __be16 proto)
+{
+	void *ip_p = skb->data + network_depth;
+
+	return (proto == htons(ETH_P_IP)) ? ((struct iphdr *)ip_p)->protocol :
+					    ((struct ipv6hdr *)ip_p)->nexthdr;
+}
+
+#define short_frame(size) ((size) <= ETH_ZLEN + ETH_FCS_LEN)
+
+#define MAX_PADDING 8
+
+static void
+tail_padding_csum_slow(struct sk_buff *skb, int offset, int len,
+		       struct mlx5e_rq_stats *stats)
+{
+	stats->csum_complete_tail_slow++;
+	skb->csum = csum_block_add(skb->csum,
+				   skb_checksum(skb, offset, len, 0),
+				   offset);
+}
+
+static void
+tail_padding_csum(struct sk_buff *skb, int offset,
+		  struct mlx5e_rq_stats *stats)
+{
+	u8 tail_padding[MAX_PADDING];
+	int len = skb->len - offset;
+	void *tail;
+
+	if (unlikely(len > MAX_PADDING)) {
+		tail_padding_csum_slow(skb, offset, len, stats);
+		return;
+	}
+
+	tail = skb_header_pointer(skb, offset, len, tail_padding);
+	if (unlikely(!tail)) {
+		tail_padding_csum_slow(skb, offset, len, stats);
+		return;
+	}
+
+	stats->csum_complete_tail++;
+	skb->csum = csum_block_add(skb->csum, csum_partial(tail, len, 0), offset);
+}
+
+static void
+mlx5e_skb_csum_fixup(struct sk_buff *skb, int network_depth, __be16 proto,
+		     struct mlx5e_rq_stats *stats)
+{
+	struct ipv6hdr *ip6;
+	struct iphdr   *ip4;
+	int pkt_len;
+
+	/* Fixup vlan headers, if any */
+	if (network_depth > ETH_HLEN)
+		/* CQE csum is calculated from the IP header and does
+		 * not cover VLAN headers (if present). This will add
+		 * the checksum manually.
+		 */
+		skb->csum = csum_partial(skb->data + ETH_HLEN,
+					 network_depth - ETH_HLEN,
+					 skb->csum);
+
+	/* Fixup tail padding, if any */
+	switch (proto) {
+	case htons(ETH_P_IP):
+		ip4 = (struct iphdr *)(skb->data + network_depth);
+		pkt_len = network_depth + ntohs(ip4->tot_len);
+		break;
+	case htons(ETH_P_IPV6):
+		ip6 = (struct ipv6hdr *)(skb->data + network_depth);
+		pkt_len = network_depth + sizeof(*ip6) + ntohs(ip6->payload_len);
+		break;
+	default:
+		return;
+	}
+
+	if (likely(pkt_len >= skb->len))
+		return;
+
+	tail_padding_csum(skb, pkt_len, stats);
+}
+
+static inline void mlx5e_handle_csum(struct net_device *netdev,
+				     struct mlx5_cqe64 *cqe,
+				     struct mlx5e_rq *rq,
+				     struct sk_buff *skb,
+				     bool   lro)
+{
+	struct mlx5e_rq_stats *stats = rq->stats;
+	int network_depth = 0;
+	__be16 proto;
+
+	if (unlikely(!(netdev->features & NETIF_F_RXCSUM)))
+		goto csum_none;
+
+	if (lro) {
+		skb->ip_summed = CHECKSUM_UNNECESSARY;
+		stats->csum_unnecessary++;
+		return;
+	}
+
+	/* True when explicitly set via priv flag, or XDP prog is loaded */
+	if (test_bit(MLX5E_RQ_STATE_NO_CSUM_COMPLETE, &rq->state) ||
+	    get_cqe_tls_offload(cqe))
+		goto csum_unnecessary;
+
+	/* CQE csum doesn't cover padding octets in short ethernet
+	 * frames. And the pad field is appended prior to calculating
+	 * and appending the FCS field.
+	 *
+	 * Detecting these padded frames requires to verify and parse
+	 * IP headers, so we simply force all those small frames to be
+	 * CHECKSUM_UNNECESSARY even if they are not padded.
+	 */
+	if (short_frame(skb->len))
+		goto csum_unnecessary;
+
+	if (likely(is_last_ethertype_ip(skb, &network_depth, &proto))) {
+		if (unlikely(get_ip_proto(skb, network_depth, proto) == IPPROTO_SCTP))
+			goto csum_unnecessary;
+
+		stats->csum_complete++;
+		skb->ip_summed = CHECKSUM_COMPLETE;
+		skb->csum = csum_unfold((__force __sum16)cqe->check_sum);
+
+		if (test_bit(MLX5E_RQ_STATE_CSUM_FULL, &rq->state))
+			return; /* CQE csum covers all received bytes */
+
+		/* csum might need some fixups ...*/
+		mlx5e_skb_csum_fixup(skb, network_depth, proto, stats);
+		return;
+	}
+
+csum_unnecessary:
+	if (likely((cqe->hds_ip_ext & CQE_L3_OK) &&
+		   (cqe->hds_ip_ext & CQE_L4_OK))) {
+		skb->ip_summed = CHECKSUM_UNNECESSARY;
+		if (cqe_is_tunneled(cqe)) {
+			skb->csum_level = 1;
+			skb->encapsulation = 1;
+			stats->csum_unnecessary_inner++;
+			return;
+		}
+		stats->csum_unnecessary++;
+		return;
+	}
+csum_none:
+	skb->ip_summed = CHECKSUM_NONE;
+	stats->csum_none++;
+}
+
+#define MLX5E_CE_BIT_MASK 0x80
+
+static inline void mlx5e_build_rx_skb(struct mlx5_cqe64 *cqe,
+				      u32 cqe_bcnt,
+				      struct mlx5e_rq *rq,
+				      struct sk_buff *skb)
+{
+	u8 lro_num_seg = be32_to_cpu(cqe->srqn) >> 24;
+	struct mlx5e_rq_stats *stats = rq->stats;
+	struct net_device *netdev = rq->netdev;
+
+	skb->mac_len = ETH_HLEN;
+
+	if (unlikely(get_cqe_tls_offload(cqe)))
+		mlx5e_ktls_handle_rx_skb(rq, skb, cqe, &cqe_bcnt);
+
+	if (unlikely(mlx5_ipsec_is_rx_flow(cqe)))
+		mlx5e_ipsec_offload_handle_rx_skb(netdev, skb, cqe);
+
+	if (unlikely(mlx5e_macsec_is_rx_flow(cqe)))
+		mlx5e_macsec_offload_handle_rx_skb(netdev, skb, cqe);
+
+	if (lro_num_seg > 1) {
+		mlx5e_lro_update_hdr(skb, cqe, cqe_bcnt);
+		skb_shinfo(skb)->gso_size = DIV_ROUND_UP(cqe_bcnt, lro_num_seg);
+		/* Subtract one since we already counted this as one
+		 * "regular" packet in mlx5e_complete_rx_cqe()
+		 */
+		stats->packets += lro_num_seg - 1;
+		stats->lro_packets++;
+		stats->lro_bytes += cqe_bcnt;
+	}
+
+	if (unlikely(mlx5e_rx_hw_stamp(rq->tstamp)))
+		skb_hwtstamps(skb)->hwtstamp = mlx5e_cqe_ts_to_ns(rq->ptp_cyc2time,
+								  rq->clock, get_cqe_ts(cqe));
+	skb_record_rx_queue(skb, rq->ix);
+
+	if (likely(netdev->features & NETIF_F_RXHASH))
+		mlx5e_skb_set_hash(cqe, skb);
+
+	if (cqe_has_vlan(cqe)) {
+		__vlan_hwaccel_put_tag(skb, htons(ETH_P_8021Q),
+				       be16_to_cpu(cqe->vlan_info));
+		stats->removed_vlan_packets++;
+	}
+
+	skb->mark = be32_to_cpu(cqe->sop_drop_qpn) & MLX5E_TC_FLOW_ID_MASK;
+
+	mlx5e_handle_csum(netdev, cqe, rq, skb, !!lro_num_seg);
+	/* checking CE bit in cqe - MSB in ml_path field */
+	if (unlikely(cqe->ml_path & MLX5E_CE_BIT_MASK))
+		mlx5e_enable_ecn(rq, skb);
+
+	skb->protocol = eth_type_trans(skb, netdev);
+
+	if (unlikely(mlx5e_skb_is_multicast(skb)))
+		stats->mcast_packets++;
+}
+
+static void mlx5e_shampo_complete_rx_cqe(struct mlx5e_rq *rq,
+					 struct mlx5_cqe64 *cqe,
+					 u32 cqe_bcnt,
+					 struct sk_buff *skb)
+{
+	struct mlx5e_rq_stats *stats = rq->stats;
+
+	stats->packets++;
+	stats->gro_packets++;
+	stats->bytes += cqe_bcnt;
+	stats->gro_bytes += cqe_bcnt;
+	if (NAPI_GRO_CB(skb)->count != 1)
+		return;
+	mlx5e_build_rx_skb(cqe, cqe_bcnt, rq, skb);
+	skb_reset_network_header(skb);
+	if (!skb_flow_dissect_flow_keys(skb, &rq->hw_gro_data->fk, 0)) {
+		napi_gro_receive(rq->cq.napi, skb);
+		rq->hw_gro_data->skb = NULL;
+	}
+}
+
+static inline void mlx5e_complete_rx_cqe(struct mlx5e_rq *rq,
+					 struct mlx5_cqe64 *cqe,
+					 u32 cqe_bcnt,
+					 struct sk_buff *skb)
+{
+	struct mlx5e_rq_stats *stats = rq->stats;
+
+	stats->packets++;
+	stats->bytes += cqe_bcnt;
+	mlx5e_build_rx_skb(cqe, cqe_bcnt, rq, skb);
+}
+
+static inline
+struct sk_buff *mlx5e_build_linear_skb(struct mlx5e_rq *rq, void *va,
+				       u32 frag_size, u16 headroom,
+				       u32 cqe_bcnt, u32 metasize)
+{
+	struct sk_buff *skb = build_skb(va, frag_size);
+
+	if (unlikely(!skb)) {
+		rq->stats->buff_alloc_err++;
+		return NULL;
+	}
+
+	skb_reserve(skb, headroom);
+	skb_put(skb, cqe_bcnt);
+
+	if (metasize)
+		skb_metadata_set(skb, metasize);
+
+	return skb;
+}
+
+static void mlx5e_fill_xdp_buff(struct mlx5e_rq *rq, void *va, u16 headroom,
+				u32 len, struct xdp_buff *xdp)
+{
+	xdp_init_buff(xdp, rq->buff.frame0_sz, &rq->xdp_rxq);
+	xdp_prepare_buff(xdp, va, headroom, len, true);
+}
+
+static struct sk_buff *
+mlx5e_skb_from_cqe_linear(struct mlx5e_rq *rq, struct mlx5e_wqe_frag_info *wi,
+			  u32 cqe_bcnt)
+{
+	union mlx5e_alloc_unit *au = wi->au;
+	u16 rx_headroom = rq->buff.headroom;
+	struct bpf_prog *prog;
+	struct sk_buff *skb;
+	u32 metasize = 0;
+	void *va, *data;
+	dma_addr_t addr;
+	u32 frag_size;
+
+	va             = page_address(au->page) + wi->offset;
+	data           = va + rx_headroom;
+	frag_size      = MLX5_SKB_FRAG_SZ(rx_headroom + cqe_bcnt);
+
+	addr = page_pool_get_dma_addr(au->page);
+	dma_sync_single_range_for_cpu(rq->pdev, addr, wi->offset,
+				      frag_size, rq->buff.map_dir);
+	net_prefetch(data);
+
+	prog = rcu_dereference(rq->xdp_prog);
+	if (prog) {
+		struct xdp_buff xdp;
+
+		net_prefetchw(va); /* xdp_frame data area */
+		mlx5e_fill_xdp_buff(rq, va, rx_headroom, cqe_bcnt, &xdp);
+		if (mlx5e_xdp_handle(rq, au->page, prog, &xdp))
+			return NULL; /* page/packet was consumed by XDP */
+
+		rx_headroom = xdp.data - xdp.data_hard_start;
+		metasize = xdp.data - xdp.data_meta;
+		cqe_bcnt = xdp.data_end - xdp.data;
+	}
+	frag_size = MLX5_SKB_FRAG_SZ(rx_headroom + cqe_bcnt);
+	skb = mlx5e_build_linear_skb(rq, va, frag_size, rx_headroom, cqe_bcnt, metasize);
+	if (unlikely(!skb))
+		return NULL;
+
+	/* queue up for recycling/reuse */
+	page_ref_inc(au->page);
+
+	return skb;
+}
+
+static struct sk_buff *
+mlx5e_skb_from_cqe_nonlinear(struct mlx5e_rq *rq, struct mlx5e_wqe_frag_info *wi,
+			     u32 cqe_bcnt)
+{
+	struct mlx5e_rq_frag_info *frag_info = &rq->wqe.info.arr[0];
+	struct mlx5e_wqe_frag_info *head_wi = wi;
+	union mlx5e_alloc_unit *au = wi->au;
+	u16 rx_headroom = rq->buff.headroom;
+	struct skb_shared_info *sinfo;
+	u32 frag_consumed_bytes;
+	struct bpf_prog *prog;
+	struct xdp_buff xdp;
+	struct sk_buff *skb;
+	dma_addr_t addr;
+	u32 truesize;
+	void *va;
+
+	va = page_address(au->page) + wi->offset;
+	frag_consumed_bytes = min_t(u32, frag_info->frag_size, cqe_bcnt);
+
+	addr = page_pool_get_dma_addr(au->page);
+	dma_sync_single_range_for_cpu(rq->pdev, addr, wi->offset,
+				      rq->buff.frame0_sz, rq->buff.map_dir);
+	net_prefetchw(va); /* xdp_frame data area */
+	net_prefetch(va + rx_headroom);
+
+	mlx5e_fill_xdp_buff(rq, va, rx_headroom, frag_consumed_bytes, &xdp);
+	sinfo = xdp_get_shared_info_from_buff(&xdp);
+	truesize = 0;
+
+	cqe_bcnt -= frag_consumed_bytes;
+	frag_info++;
+	wi++;
+
+	while (cqe_bcnt) {
+		skb_frag_t *frag;
+
+		au = wi->au;
+
+		frag_consumed_bytes = min_t(u32, frag_info->frag_size, cqe_bcnt);
+
+		addr = page_pool_get_dma_addr(au->page);
+		dma_sync_single_for_cpu(rq->pdev, addr + wi->offset,
+					frag_consumed_bytes, rq->buff.map_dir);
+
+		if (!xdp_buff_has_frags(&xdp)) {
+			/* Init on the first fragment to avoid cold cache access
+			 * when possible.
+			 */
+			sinfo->nr_frags = 0;
+			sinfo->xdp_frags_size = 0;
+			xdp_buff_set_frags_flag(&xdp);
+		}
+
+		frag = &sinfo->frags[sinfo->nr_frags++];
+		__skb_frag_set_page(frag, au->page);
+		skb_frag_off_set(frag, wi->offset);
+		skb_frag_size_set(frag, frag_consumed_bytes);
+
+		if (page_is_pfmemalloc(au->page))
+			xdp_buff_set_frag_pfmemalloc(&xdp);
+
+		sinfo->xdp_frags_size += frag_consumed_bytes;
+		truesize += frag_info->frag_stride;
+
+		cqe_bcnt -= frag_consumed_bytes;
+		frag_info++;
+		wi++;
+	}
+
+	au = head_wi->au;
+
+	prog = rcu_dereference(rq->xdp_prog);
+	if (prog && mlx5e_xdp_handle(rq, au->page, prog, &xdp)) {
+		if (test_bit(MLX5E_RQ_FLAG_XDP_XMIT, rq->flags)) {
+			int i;
+
+			for (i = wi - head_wi; i < rq->wqe.info.num_frags; i++)
+				mlx5e_put_rx_frag(rq, &head_wi[i], true);
+		}
+		return NULL; /* page/packet was consumed by XDP */
+	}
+
+	skb = mlx5e_build_linear_skb(rq, xdp.data_hard_start, rq->buff.frame0_sz,
+				     xdp.data - xdp.data_hard_start,
+				     xdp.data_end - xdp.data,
+				     xdp.data - xdp.data_meta);
+	if (unlikely(!skb))
+		return NULL;
+
+	page_ref_inc(au->page);
+
+	if (unlikely(xdp_buff_has_frags(&xdp))) {
+		int i;
+
+		/* sinfo->nr_frags is reset by build_skb, calculate again. */
+		xdp_update_skb_shared_info(skb, wi - head_wi - 1,
+					   sinfo->xdp_frags_size, truesize,
+					   xdp_buff_is_frag_pfmemalloc(&xdp));
+
+		for (i = 0; i < sinfo->nr_frags; i++) {
+			skb_frag_t *frag = &sinfo->frags[i];
+
+			page_ref_inc(skb_frag_page(frag));
+		}
+	}
+
+	return skb;
+}
+
+static void trigger_report(struct mlx5e_rq *rq, struct mlx5_cqe64 *cqe)
+{
+	struct mlx5_err_cqe *err_cqe = (struct mlx5_err_cqe *)cqe;
+	struct mlx5e_priv *priv = rq->priv;
+
+	if (cqe_syndrome_needs_recover(err_cqe->syndrome) &&
+	    !test_and_set_bit(MLX5E_RQ_STATE_RECOVERING, &rq->state)) {
+		mlx5e_dump_error_cqe(&rq->cq, rq->rqn, err_cqe);
+		queue_work(priv->wq, &rq->recover_work);
+	}
+}
+
+static void mlx5e_handle_rx_err_cqe(struct mlx5e_rq *rq, struct mlx5_cqe64 *cqe)
+{
+	trigger_report(rq, cqe);
+	rq->stats->wqe_err++;
+}
+
+static void mlx5e_handle_rx_cqe(struct mlx5e_rq *rq, struct mlx5_cqe64 *cqe)
+{
+	struct mlx5_wq_cyc *wq = &rq->wqe.wq;
+	struct mlx5e_wqe_frag_info *wi;
+	struct sk_buff *skb;
+	u32 cqe_bcnt;
+	u16 ci;
+
+	ci       = mlx5_wq_cyc_ctr2ix(wq, be16_to_cpu(cqe->wqe_counter));
+	wi       = get_frag(rq, ci);
+	cqe_bcnt = be32_to_cpu(cqe->byte_cnt);
+
+	if (unlikely(MLX5E_RX_ERR_CQE(cqe))) {
+		mlx5e_handle_rx_err_cqe(rq, cqe);
+		goto free_wqe;
+	}
+
+	skb = INDIRECT_CALL_3(rq->wqe.skb_from_cqe,
+			      mlx5e_skb_from_cqe_linear,
+			      mlx5e_skb_from_cqe_nonlinear,
+			      mlx5e_xsk_skb_from_cqe_linear,
+			      rq, wi, cqe_bcnt);
+	if (!skb) {
+		/* probably for XDP */
+		if (__test_and_clear_bit(MLX5E_RQ_FLAG_XDP_XMIT, rq->flags)) {
+			/* do not return page to cache,
+			 * it will be returned on XDP_TX completion.
+			 */
+			goto wq_cyc_pop;
+		}
+		goto free_wqe;
+	}
+
+	mlx5e_complete_rx_cqe(rq, cqe, cqe_bcnt, skb);
+
+	if (mlx5e_cqe_regb_chain(cqe))
+		if (!mlx5e_tc_update_skb(cqe, skb)) {
+			dev_kfree_skb_any(skb);
+			goto free_wqe;
+		}
+
+	napi_gro_receive(rq->cq.napi, skb);
+
+free_wqe:
+	mlx5e_free_rx_wqe(rq, wi, true);
+wq_cyc_pop:
+	mlx5_wq_cyc_pop(wq);
+}
+
+#ifdef CONFIG_MLX5_ESWITCH
+static void mlx5e_handle_rx_cqe_rep(struct mlx5e_rq *rq, struct mlx5_cqe64 *cqe)
+{
+	struct net_device *netdev = rq->netdev;
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	struct mlx5e_rep_priv *rpriv  = priv->ppriv;
+	struct mlx5_eswitch_rep *rep = rpriv->rep;
+	struct mlx5_wq_cyc *wq = &rq->wqe.wq;
+	struct mlx5e_wqe_frag_info *wi;
+	struct sk_buff *skb;
+	u32 cqe_bcnt;
+	u16 ci;
+
+	ci       = mlx5_wq_cyc_ctr2ix(wq, be16_to_cpu(cqe->wqe_counter));
+	wi       = get_frag(rq, ci);
+	cqe_bcnt = be32_to_cpu(cqe->byte_cnt);
+
+	if (unlikely(MLX5E_RX_ERR_CQE(cqe))) {
+		mlx5e_handle_rx_err_cqe(rq, cqe);
+		goto free_wqe;
+	}
+
+	skb = INDIRECT_CALL_2(rq->wqe.skb_from_cqe,
+			      mlx5e_skb_from_cqe_linear,
+			      mlx5e_skb_from_cqe_nonlinear,
+			      rq, wi, cqe_bcnt);
+	if (!skb) {
+		/* probably for XDP */
+		if (__test_and_clear_bit(MLX5E_RQ_FLAG_XDP_XMIT, rq->flags)) {
+			/* do not return page to cache,
+			 * it will be returned on XDP_TX completion.
+			 */
+			goto wq_cyc_pop;
+		}
+		goto free_wqe;
+	}
+
+	mlx5e_complete_rx_cqe(rq, cqe, cqe_bcnt, skb);
+
+	if (rep->vlan && skb_vlan_tag_present(skb))
+		skb_vlan_pop(skb);
+
+	mlx5e_rep_tc_receive(cqe, rq, skb);
+
+free_wqe:
+	mlx5e_free_rx_wqe(rq, wi, true);
+wq_cyc_pop:
+	mlx5_wq_cyc_pop(wq);
+}
+
+static void mlx5e_handle_rx_cqe_mpwrq_rep(struct mlx5e_rq *rq, struct mlx5_cqe64 *cqe)
+{
+	u16 cstrides       = mpwrq_get_cqe_consumed_strides(cqe);
+	u16 wqe_id         = be16_to_cpu(cqe->wqe_id);
+	struct mlx5e_mpw_info *wi = mlx5e_get_mpw_info(rq, wqe_id);
+	u16 stride_ix      = mpwrq_get_cqe_stride_index(cqe);
+	u32 wqe_offset     = stride_ix << rq->mpwqe.log_stride_sz;
+	u32 head_offset    = wqe_offset & ((1 << rq->mpwqe.page_shift) - 1);
+	u32 page_idx       = wqe_offset >> rq->mpwqe.page_shift;
+	struct mlx5e_rx_wqe_ll *wqe;
+	struct mlx5_wq_ll *wq;
+	struct sk_buff *skb;
+	u16 cqe_bcnt;
+
+	wi->consumed_strides += cstrides;
+
+	if (unlikely(MLX5E_RX_ERR_CQE(cqe))) {
+		mlx5e_handle_rx_err_cqe(rq, cqe);
+		goto mpwrq_cqe_out;
+	}
+
+	if (unlikely(mpwrq_is_filler_cqe(cqe))) {
+		struct mlx5e_rq_stats *stats = rq->stats;
+
+		stats->mpwqe_filler_cqes++;
+		stats->mpwqe_filler_strides += cstrides;
+		goto mpwrq_cqe_out;
+	}
+
+	cqe_bcnt = mpwrq_get_cqe_byte_cnt(cqe);
+
+	skb = INDIRECT_CALL_2(rq->mpwqe.skb_from_cqe_mpwrq,
+			      mlx5e_skb_from_cqe_mpwrq_linear,
+			      mlx5e_skb_from_cqe_mpwrq_nonlinear,
+			      rq, wi, cqe_bcnt, head_offset, page_idx);
+	if (!skb)
+		goto mpwrq_cqe_out;
+
+	mlx5e_complete_rx_cqe(rq, cqe, cqe_bcnt, skb);
+
+	mlx5e_rep_tc_receive(cqe, rq, skb);
+
+mpwrq_cqe_out:
+	if (likely(wi->consumed_strides < rq->mpwqe.num_strides))
+		return;
+
+	wq  = &rq->mpwqe.wq;
+	wqe = mlx5_wq_ll_get_wqe(wq, wqe_id);
+	mlx5e_free_rx_mpwqe(rq, wi, true);
+	mlx5_wq_ll_pop(wq, cqe->wqe_id, &wqe->next.next_wqe_index);
+}
+
+const struct mlx5e_rx_handlers mlx5e_rx_handlers_rep = {
+	.handle_rx_cqe       = mlx5e_handle_rx_cqe_rep,
+	.handle_rx_cqe_mpwqe = mlx5e_handle_rx_cqe_mpwrq_rep,
+};
+#endif
+
+static void
+mlx5e_fill_skb_data(struct sk_buff *skb, struct mlx5e_rq *rq,
+		    union mlx5e_alloc_unit *au, u32 data_bcnt, u32 data_offset)
+{
+	net_prefetchw(skb->data);
+
+	while (data_bcnt) {
+		/* Non-linear mode, hence non-XSK, which always uses PAGE_SIZE. */
+		u32 pg_consumed_bytes = min_t(u32, PAGE_SIZE - data_offset, data_bcnt);
+		unsigned int truesize;
+
+		if (test_bit(MLX5E_RQ_STATE_SHAMPO, &rq->state))
+			truesize = pg_consumed_bytes;
+		else
+			truesize = ALIGN(pg_consumed_bytes, BIT(rq->mpwqe.log_stride_sz));
+
+		mlx5e_add_skb_frag(rq, skb, au, data_offset,
+				   pg_consumed_bytes, truesize);
+
+		data_bcnt -= pg_consumed_bytes;
+		data_offset = 0;
+		au++;
+	}
+}
+
+static struct sk_buff *
+mlx5e_skb_from_cqe_mpwrq_nonlinear(struct mlx5e_rq *rq, struct mlx5e_mpw_info *wi,
+				   u16 cqe_bcnt, u32 head_offset, u32 page_idx)
+{
+	union mlx5e_alloc_unit *au = &wi->alloc_units[page_idx];
+	u16 headlen = min_t(u16, MLX5E_RX_MAX_HEAD, cqe_bcnt);
+	u32 frag_offset    = head_offset + headlen;
+	u32 byte_cnt       = cqe_bcnt - headlen;
+	union mlx5e_alloc_unit *head_au = au;
+	struct sk_buff *skb;
+	dma_addr_t addr;
+
+	skb = napi_alloc_skb(rq->cq.napi,
+			     ALIGN(MLX5E_RX_MAX_HEAD, sizeof(long)));
+	if (unlikely(!skb)) {
+		rq->stats->buff_alloc_err++;
+		return NULL;
+	}
+
+	net_prefetchw(skb->data);
+
+	/* Non-linear mode, hence non-XSK, which always uses PAGE_SIZE. */
+	if (unlikely(frag_offset >= PAGE_SIZE)) {
+		au++;
+		frag_offset -= PAGE_SIZE;
+	}
+
+	mlx5e_fill_skb_data(skb, rq, au, byte_cnt, frag_offset);
+	/* copy header */
+	addr = page_pool_get_dma_addr(head_au->page);
+	mlx5e_copy_skb_header(rq, skb, head_au->page, addr,
+			      head_offset, head_offset, headlen);
+	/* skb linear part was allocated with headlen and aligned to long */
+	skb->tail += headlen;
+	skb->len  += headlen;
+
+	return skb;
+}
+
+static struct sk_buff *
+mlx5e_skb_from_cqe_mpwrq_linear(struct mlx5e_rq *rq, struct mlx5e_mpw_info *wi,
+				u16 cqe_bcnt, u32 head_offset, u32 page_idx)
+{
+	union mlx5e_alloc_unit *au = &wi->alloc_units[page_idx];
+	u16 rx_headroom = rq->buff.headroom;
+	struct bpf_prog *prog;
+	struct sk_buff *skb;
+	u32 metasize = 0;
+	void *va, *data;
+	dma_addr_t addr;
+	u32 frag_size;
+
+	/* Check packet size. Note LRO doesn't use linear SKB */
+	if (unlikely(cqe_bcnt > rq->hw_mtu)) {
+		rq->stats->oversize_pkts_sw_drop++;
+		return NULL;
+	}
+
+	va             = page_address(au->page) + head_offset;
+	data           = va + rx_headroom;
+	frag_size      = MLX5_SKB_FRAG_SZ(rx_headroom + cqe_bcnt);
+
+	addr = page_pool_get_dma_addr(au->page);
+	dma_sync_single_range_for_cpu(rq->pdev, addr, head_offset,
+				      frag_size, rq->buff.map_dir);
+	net_prefetch(data);
+
+	prog = rcu_dereference(rq->xdp_prog);
+	if (prog) {
+		struct xdp_buff xdp;
+
+		net_prefetchw(va); /* xdp_frame data area */
+		mlx5e_fill_xdp_buff(rq, va, rx_headroom, cqe_bcnt, &xdp);
+		if (mlx5e_xdp_handle(rq, au->page, prog, &xdp)) {
+			if (__test_and_clear_bit(MLX5E_RQ_FLAG_XDP_XMIT, rq->flags))
+				__set_bit(page_idx, wi->xdp_xmit_bitmap); /* non-atomic */
+			return NULL; /* page/packet was consumed by XDP */
+		}
+
+		rx_headroom = xdp.data - xdp.data_hard_start;
+		metasize = xdp.data - xdp.data_meta;
+		cqe_bcnt = xdp.data_end - xdp.data;
+	}
+	frag_size = MLX5_SKB_FRAG_SZ(rx_headroom + cqe_bcnt);
+	skb = mlx5e_build_linear_skb(rq, va, frag_size, rx_headroom, cqe_bcnt, metasize);
+	if (unlikely(!skb))
+		return NULL;
+
+	/* queue up for recycling/reuse */
+	page_ref_inc(au->page);
+
+	return skb;
+}
+
+static struct sk_buff *
+mlx5e_skb_from_cqe_shampo(struct mlx5e_rq *rq, struct mlx5e_mpw_info *wi,
+			  struct mlx5_cqe64 *cqe, u16 header_index)
+{
+	struct mlx5e_dma_info *head = &rq->mpwqe.shampo->info[header_index];
+	u16 head_offset = head->addr & (PAGE_SIZE - 1);
+	u16 head_size = cqe->shampo.header_size;
+	u16 rx_headroom = rq->buff.headroom;
+	struct sk_buff *skb = NULL;
+	void *hdr, *data;
+	u32 frag_size;
+
+	hdr		= page_address(head->page) + head_offset;
+	data		= hdr + rx_headroom;
+	frag_size	= MLX5_SKB_FRAG_SZ(rx_headroom + head_size);
+
+	if (likely(frag_size <= BIT(MLX5E_SHAMPO_LOG_MAX_HEADER_ENTRY_SIZE))) {
+		/* build SKB around header */
+		dma_sync_single_range_for_cpu(rq->pdev, head->addr, 0, frag_size, rq->buff.map_dir);
+		prefetchw(hdr);
+		prefetch(data);
+		skb = mlx5e_build_linear_skb(rq, hdr, frag_size, rx_headroom, head_size, 0);
+
+		if (unlikely(!skb))
+			return NULL;
+
+		/* queue up for recycling/reuse */
+		page_ref_inc(head->page);
+
+	} else {
+		/* allocate SKB and copy header for large header */
+		rq->stats->gro_large_hds++;
+		skb = napi_alloc_skb(rq->cq.napi,
+				     ALIGN(head_size, sizeof(long)));
+		if (unlikely(!skb)) {
+			rq->stats->buff_alloc_err++;
+			return NULL;
+		}
+
+		prefetchw(skb->data);
+		mlx5e_copy_skb_header(rq, skb, head->page, head->addr,
+				      head_offset + rx_headroom,
+				      rx_headroom, head_size);
+		/* skb linear part was allocated with headlen and aligned to long */
+		skb->tail += head_size;
+		skb->len  += head_size;
+	}
+	return skb;
+}
+
+static void
+mlx5e_shampo_align_fragment(struct sk_buff *skb, u8 log_stride_sz)
+{
+	skb_frag_t *last_frag = &skb_shinfo(skb)->frags[skb_shinfo(skb)->nr_frags - 1];
+	unsigned int frag_size = skb_frag_size(last_frag);
+	unsigned int frag_truesize;
+
+	frag_truesize = ALIGN(frag_size, BIT(log_stride_sz));
+	skb->truesize += frag_truesize - frag_size;
+}
+
+static void
+mlx5e_shampo_flush_skb(struct mlx5e_rq *rq, struct mlx5_cqe64 *cqe, bool match)
+{
+	struct sk_buff *skb = rq->hw_gro_data->skb;
+	struct mlx5e_rq_stats *stats = rq->stats;
+
+	stats->gro_skbs++;
+	if (likely(skb_shinfo(skb)->nr_frags))
+		mlx5e_shampo_align_fragment(skb, rq->mpwqe.log_stride_sz);
+	if (NAPI_GRO_CB(skb)->count > 1)
+		mlx5e_shampo_update_hdr(rq, cqe, match);
+	napi_gro_receive(rq->cq.napi, skb);
+	rq->hw_gro_data->skb = NULL;
+}
+
+static bool
+mlx5e_hw_gro_skb_has_enough_space(struct sk_buff *skb, u16 data_bcnt)
+{
+	int nr_frags = skb_shinfo(skb)->nr_frags;
+
+	return PAGE_SIZE * nr_frags + data_bcnt <= GRO_LEGACY_MAX_SIZE;
+}
+
+static void
+mlx5e_free_rx_shampo_hd_entry(struct mlx5e_rq *rq, u16 header_index)
+{
+	struct mlx5e_shampo_hd *shampo = rq->mpwqe.shampo;
+	u64 addr = shampo->info[header_index].addr;
+
+	if (((header_index + 1) & (MLX5E_SHAMPO_WQ_HEADER_PER_PAGE - 1)) == 0) {
+		shampo->info[header_index].addr = ALIGN_DOWN(addr, PAGE_SIZE);
+		mlx5e_page_release_dynamic(rq, shampo->info[header_index].page, true);
+	}
+	bitmap_clear(shampo->bitmap, header_index, 1);
+}
+
+static void mlx5e_handle_rx_cqe_mpwrq_shampo(struct mlx5e_rq *rq, struct mlx5_cqe64 *cqe)
+{
+	u16 data_bcnt		= mpwrq_get_cqe_byte_cnt(cqe) - cqe->shampo.header_size;
+	u16 header_index	= mlx5e_shampo_get_cqe_header_index(rq, cqe);
+	u32 wqe_offset		= be32_to_cpu(cqe->shampo.data_offset);
+	u16 cstrides		= mpwrq_get_cqe_consumed_strides(cqe);
+	u32 data_offset		= wqe_offset & (PAGE_SIZE - 1);
+	u32 cqe_bcnt		= mpwrq_get_cqe_byte_cnt(cqe);
+	u16 wqe_id		= be16_to_cpu(cqe->wqe_id);
+	u32 page_idx		= wqe_offset >> PAGE_SHIFT;
+	u16 head_size		= cqe->shampo.header_size;
+	struct sk_buff **skb	= &rq->hw_gro_data->skb;
+	bool flush		= cqe->shampo.flush;
+	bool match		= cqe->shampo.match;
+	struct mlx5e_rq_stats *stats = rq->stats;
+	struct mlx5e_rx_wqe_ll *wqe;
+	union mlx5e_alloc_unit *au;
+	struct mlx5e_mpw_info *wi;
+	struct mlx5_wq_ll *wq;
+
+	wi = mlx5e_get_mpw_info(rq, wqe_id);
+	wi->consumed_strides += cstrides;
+
+	if (unlikely(MLX5E_RX_ERR_CQE(cqe))) {
+		mlx5e_handle_rx_err_cqe(rq, cqe);
+		goto mpwrq_cqe_out;
+	}
+
+	if (unlikely(mpwrq_is_filler_cqe(cqe))) {
+		stats->mpwqe_filler_cqes++;
+		stats->mpwqe_filler_strides += cstrides;
+		goto mpwrq_cqe_out;
+	}
+
+	stats->gro_match_packets += match;
+
+	if (*skb && (!match || !(mlx5e_hw_gro_skb_has_enough_space(*skb, data_bcnt)))) {
+		match = false;
+		mlx5e_shampo_flush_skb(rq, cqe, match);
+	}
+
+	if (!*skb) {
+		if (likely(head_size))
+			*skb = mlx5e_skb_from_cqe_shampo(rq, wi, cqe, header_index);
+		else
+			*skb = mlx5e_skb_from_cqe_mpwrq_nonlinear(rq, wi, cqe_bcnt, data_offset,
+								  page_idx);
+		if (unlikely(!*skb))
+			goto free_hd_entry;
+
+		NAPI_GRO_CB(*skb)->count = 1;
+		skb_shinfo(*skb)->gso_size = cqe_bcnt - head_size;
+	} else {
+		NAPI_GRO_CB(*skb)->count++;
+		if (NAPI_GRO_CB(*skb)->count == 2 &&
+		    rq->hw_gro_data->fk.basic.n_proto == htons(ETH_P_IP)) {
+			void *hd_addr = mlx5e_shampo_get_packet_hd(rq, header_index);
+			int nhoff = ETH_HLEN + rq->hw_gro_data->fk.control.thoff -
+				    sizeof(struct iphdr);
+			struct iphdr *iph = (struct iphdr *)(hd_addr + nhoff);
+
+			rq->hw_gro_data->second_ip_id = ntohs(iph->id);
+		}
+	}
+
+	if (likely(head_size)) {
+		au = &wi->alloc_units[page_idx];
+		mlx5e_fill_skb_data(*skb, rq, au, data_bcnt, data_offset);
+	}
+
+	mlx5e_shampo_complete_rx_cqe(rq, cqe, cqe_bcnt, *skb);
+	if (flush)
+		mlx5e_shampo_flush_skb(rq, cqe, match);
+free_hd_entry:
+	mlx5e_free_rx_shampo_hd_entry(rq, header_index);
+mpwrq_cqe_out:
+	if (likely(wi->consumed_strides < rq->mpwqe.num_strides))
+		return;
+
+	wq  = &rq->mpwqe.wq;
+	wqe = mlx5_wq_ll_get_wqe(wq, wqe_id);
+	mlx5e_free_rx_mpwqe(rq, wi, true);
+	mlx5_wq_ll_pop(wq, cqe->wqe_id, &wqe->next.next_wqe_index);
+}
+
+static void mlx5e_handle_rx_cqe_mpwrq(struct mlx5e_rq *rq, struct mlx5_cqe64 *cqe)
+{
+	u16 cstrides       = mpwrq_get_cqe_consumed_strides(cqe);
+	u16 wqe_id         = be16_to_cpu(cqe->wqe_id);
+	struct mlx5e_mpw_info *wi = mlx5e_get_mpw_info(rq, wqe_id);
+	u16 stride_ix      = mpwrq_get_cqe_stride_index(cqe);
+	u32 wqe_offset     = stride_ix << rq->mpwqe.log_stride_sz;
+	u32 head_offset    = wqe_offset & ((1 << rq->mpwqe.page_shift) - 1);
+	u32 page_idx       = wqe_offset >> rq->mpwqe.page_shift;
+	struct mlx5e_rx_wqe_ll *wqe;
+	struct mlx5_wq_ll *wq;
+	struct sk_buff *skb;
+	u16 cqe_bcnt;
+
+	wi->consumed_strides += cstrides;
+
+	if (unlikely(MLX5E_RX_ERR_CQE(cqe))) {
+		mlx5e_handle_rx_err_cqe(rq, cqe);
+		goto mpwrq_cqe_out;
+	}
+
+	if (unlikely(mpwrq_is_filler_cqe(cqe))) {
+		struct mlx5e_rq_stats *stats = rq->stats;
+
+		stats->mpwqe_filler_cqes++;
+		stats->mpwqe_filler_strides += cstrides;
+		goto mpwrq_cqe_out;
+	}
+
+	cqe_bcnt = mpwrq_get_cqe_byte_cnt(cqe);
+
+	skb = INDIRECT_CALL_3(rq->mpwqe.skb_from_cqe_mpwrq,
+			      mlx5e_skb_from_cqe_mpwrq_linear,
+			      mlx5e_skb_from_cqe_mpwrq_nonlinear,
+			      mlx5e_xsk_skb_from_cqe_mpwrq_linear,
+			      rq, wi, cqe_bcnt, head_offset, page_idx);
+	if (!skb)
+		goto mpwrq_cqe_out;
+
+	mlx5e_complete_rx_cqe(rq, cqe, cqe_bcnt, skb);
+
+	if (mlx5e_cqe_regb_chain(cqe))
+		if (!mlx5e_tc_update_skb(cqe, skb)) {
+			dev_kfree_skb_any(skb);
+			goto mpwrq_cqe_out;
+		}
+
+	napi_gro_receive(rq->cq.napi, skb);
+
+mpwrq_cqe_out:
+	if (likely(wi->consumed_strides < rq->mpwqe.num_strides))
+		return;
+
+	wq  = &rq->mpwqe.wq;
+	wqe = mlx5_wq_ll_get_wqe(wq, wqe_id);
+	mlx5e_free_rx_mpwqe(rq, wi, true);
+	mlx5_wq_ll_pop(wq, cqe->wqe_id, &wqe->next.next_wqe_index);
+}
+
+int mlx5e_poll_rx_cq(struct mlx5e_cq *cq, int budget)
+{
+	struct mlx5e_rq *rq = container_of(cq, struct mlx5e_rq, cq);
+	struct mlx5_cqwq *cqwq = &cq->wq;
+	struct mlx5_cqe64 *cqe;
+	int work_done = 0;
+
+	if (unlikely(!test_bit(MLX5E_RQ_STATE_ENABLED, &rq->state)))
+		return 0;
+
+	if (rq->cqd.left) {
+		work_done += mlx5e_decompress_cqes_cont(rq, cqwq, 0, budget);
+		if (work_done >= budget)
+			goto out;
+	}
+
+	cqe = mlx5_cqwq_get_cqe(cqwq);
+	if (!cqe) {
+		if (unlikely(work_done))
+			goto out;
+		return 0;
+	}
+
+	do {
+		if (mlx5_get_cqe_format(cqe) == MLX5_COMPRESSED) {
+			work_done +=
+				mlx5e_decompress_cqes_start(rq, cqwq,
+							    budget - work_done);
+			continue;
+		}
+
+		mlx5_cqwq_pop(cqwq);
+
+		INDIRECT_CALL_3(rq->handle_rx_cqe, mlx5e_handle_rx_cqe_mpwrq,
+				mlx5e_handle_rx_cqe, mlx5e_handle_rx_cqe_mpwrq_shampo,
+				rq, cqe);
+	} while ((++work_done < budget) && (cqe = mlx5_cqwq_get_cqe(cqwq)));
+
+out:
+	if (test_bit(MLX5E_RQ_STATE_SHAMPO, &rq->state) && rq->hw_gro_data->skb)
+		mlx5e_shampo_flush_skb(rq, NULL, false);
+
+	if (rcu_access_pointer(rq->xdp_prog))
+		mlx5e_xdp_rx_poll_complete(rq);
+
+	mlx5_cqwq_update_db_record(cqwq);
+
+	/* ensure cq space is freed before enabling more cqes */
+	wmb();
+
+	return work_done;
+}
+
+#ifdef CONFIG_MLX5_CORE_IPOIB
+
+#define MLX5_IB_GRH_SGID_OFFSET 8
+#define MLX5_IB_GRH_DGID_OFFSET 24
+#define MLX5_GID_SIZE           16
+
+static inline void mlx5i_complete_rx_cqe(struct mlx5e_rq *rq,
+					 struct mlx5_cqe64 *cqe,
+					 u32 cqe_bcnt,
+					 struct sk_buff *skb)
+{
+	struct hwtstamp_config *tstamp;
+	struct mlx5e_rq_stats *stats;
+	struct net_device *netdev;
+	struct mlx5e_priv *priv;
+	char *pseudo_header;
+	u32 flags_rqpn;
+	u32 qpn;
+	u8 *dgid;
+	u8 g;
+
+	qpn = be32_to_cpu(cqe->sop_drop_qpn) & 0xffffff;
+	netdev = mlx5i_pkey_get_netdev(rq->netdev, qpn);
+
+	/* No mapping present, cannot process SKB. This might happen if a child
+	 * interface is going down while having unprocessed CQEs on parent RQ
+	 */
+	if (unlikely(!netdev)) {
+		/* TODO: add drop counters support */
+		skb->dev = NULL;
+		pr_warn_once("Unable to map QPN %u to dev - dropping skb\n", qpn);
+		return;
+	}
+
+	priv = mlx5i_epriv(netdev);
+	tstamp = &priv->tstamp;
+	stats = &priv->channel_stats[rq->ix]->rq;
+
+	flags_rqpn = be32_to_cpu(cqe->flags_rqpn);
+	g = (flags_rqpn >> 28) & 3;
+	dgid = skb->data + MLX5_IB_GRH_DGID_OFFSET;
+	if ((!g) || dgid[0] != 0xff)
+		skb->pkt_type = PACKET_HOST;
+	else if (memcmp(dgid, netdev->broadcast + 4, MLX5_GID_SIZE) == 0)
+		skb->pkt_type = PACKET_BROADCAST;
+	else
+		skb->pkt_type = PACKET_MULTICAST;
+
+	/* Drop packets that this interface sent, ie multicast packets
+	 * that the HCA has replicated.
+	 */
+	if (g && (qpn == (flags_rqpn & 0xffffff)) &&
+	    (memcmp(netdev->dev_addr + 4, skb->data + MLX5_IB_GRH_SGID_OFFSET,
+		    MLX5_GID_SIZE) == 0)) {
+		skb->dev = NULL;
+		return;
+	}
+
+	skb_pull(skb, MLX5_IB_GRH_BYTES);
+
+	skb->protocol = *((__be16 *)(skb->data));
+
+	if (netdev->features & NETIF_F_RXCSUM) {
+		skb->ip_summed = CHECKSUM_COMPLETE;
+		skb->csum = csum_unfold((__force __sum16)cqe->check_sum);
+		stats->csum_complete++;
+	} else {
+		skb->ip_summed = CHECKSUM_NONE;
+		stats->csum_none++;
+	}
+
+	if (unlikely(mlx5e_rx_hw_stamp(tstamp)))
+		skb_hwtstamps(skb)->hwtstamp = mlx5e_cqe_ts_to_ns(rq->ptp_cyc2time,
+								  rq->clock, get_cqe_ts(cqe));
+	skb_record_rx_queue(skb, rq->ix);
+
+	if (likely(netdev->features & NETIF_F_RXHASH))
+		mlx5e_skb_set_hash(cqe, skb);
+
+	/* 20 bytes of ipoib header and 4 for encap existing */
+	pseudo_header = skb_push(skb, MLX5_IPOIB_PSEUDO_LEN);
+	memset(pseudo_header, 0, MLX5_IPOIB_PSEUDO_LEN);
+	skb_reset_mac_header(skb);
+	skb_pull(skb, MLX5_IPOIB_HARD_LEN);
+
+	skb->dev = netdev;
+
+	stats->packets++;
+	stats->bytes += cqe_bcnt;
+}
+
+static void mlx5i_handle_rx_cqe(struct mlx5e_rq *rq, struct mlx5_cqe64 *cqe)
+{
+	struct mlx5_wq_cyc *wq = &rq->wqe.wq;
+	struct mlx5e_wqe_frag_info *wi;
+	struct sk_buff *skb;
+	u32 cqe_bcnt;
+	u16 ci;
+
+	ci       = mlx5_wq_cyc_ctr2ix(wq, be16_to_cpu(cqe->wqe_counter));
+	wi       = get_frag(rq, ci);
+	cqe_bcnt = be32_to_cpu(cqe->byte_cnt);
+
+	if (unlikely(MLX5E_RX_ERR_CQE(cqe))) {
+		rq->stats->wqe_err++;
+		goto wq_free_wqe;
+	}
+
+	skb = INDIRECT_CALL_2(rq->wqe.skb_from_cqe,
+			      mlx5e_skb_from_cqe_linear,
+			      mlx5e_skb_from_cqe_nonlinear,
+			      rq, wi, cqe_bcnt);
+	if (!skb)
+		goto wq_free_wqe;
+
+	mlx5i_complete_rx_cqe(rq, cqe, cqe_bcnt, skb);
+	if (unlikely(!skb->dev)) {
+		dev_kfree_skb_any(skb);
+		goto wq_free_wqe;
+	}
+	napi_gro_receive(rq->cq.napi, skb);
+
+wq_free_wqe:
+	mlx5e_free_rx_wqe(rq, wi, true);
+	mlx5_wq_cyc_pop(wq);
+}
+
+const struct mlx5e_rx_handlers mlx5i_rx_handlers = {
+	.handle_rx_cqe       = mlx5i_handle_rx_cqe,
+	.handle_rx_cqe_mpwqe = NULL, /* Not supported */
+};
+#endif /* CONFIG_MLX5_CORE_IPOIB */
+
+int mlx5e_rq_set_handlers(struct mlx5e_rq *rq, struct mlx5e_params *params, bool xsk)
+{
+	struct net_device *netdev = rq->netdev;
+	struct mlx5_core_dev *mdev = rq->mdev;
+	struct mlx5e_priv *priv = rq->priv;
+
+	switch (rq->wq_type) {
+	case MLX5_WQ_TYPE_LINKED_LIST_STRIDING_RQ:
+		rq->mpwqe.skb_from_cqe_mpwrq = xsk ?
+			mlx5e_xsk_skb_from_cqe_mpwrq_linear :
+			mlx5e_rx_mpwqe_is_linear_skb(mdev, params, NULL) ?
+				mlx5e_skb_from_cqe_mpwrq_linear :
+				mlx5e_skb_from_cqe_mpwrq_nonlinear;
+		rq->post_wqes = mlx5e_post_rx_mpwqes;
+		rq->dealloc_wqe = mlx5e_dealloc_rx_mpwqe;
+
+		if (params->packet_merge.type == MLX5E_PACKET_MERGE_SHAMPO) {
+			rq->handle_rx_cqe = priv->profile->rx_handlers->handle_rx_cqe_mpwqe_shampo;
+			if (!rq->handle_rx_cqe) {
+				netdev_err(netdev, "RX handler of SHAMPO MPWQE RQ is not set\n");
+				return -EINVAL;
+			}
+		} else {
+			rq->handle_rx_cqe = priv->profile->rx_handlers->handle_rx_cqe_mpwqe;
+			if (!rq->handle_rx_cqe) {
+				netdev_err(netdev, "RX handler of MPWQE RQ is not set\n");
+				return -EINVAL;
+			}
+		}
+
+		break;
+	default: /* MLX5_WQ_TYPE_CYCLIC */
+		rq->wqe.skb_from_cqe = xsk ?
+			mlx5e_xsk_skb_from_cqe_linear :
+			mlx5e_rx_is_linear_skb(mdev, params, NULL) ?
+				mlx5e_skb_from_cqe_linear :
+				mlx5e_skb_from_cqe_nonlinear;
+		rq->post_wqes = mlx5e_post_rx_wqes;
+		rq->dealloc_wqe = mlx5e_dealloc_rx_wqe;
+		rq->handle_rx_cqe = priv->profile->rx_handlers->handle_rx_cqe;
+		if (!rq->handle_rx_cqe) {
+			netdev_err(netdev, "RX handler of RQ is not set\n");
+			return -EINVAL;
+		}
+	}
+
+	return 0;
+}
+
+static void mlx5e_trap_handle_rx_cqe(struct mlx5e_rq *rq, struct mlx5_cqe64 *cqe)
+{
+	struct mlx5e_priv *priv = netdev_priv(rq->netdev);
+	struct mlx5_wq_cyc *wq = &rq->wqe.wq;
+	struct mlx5e_wqe_frag_info *wi;
+	struct devlink_port *dl_port;
+	struct sk_buff *skb;
+	u32 cqe_bcnt;
+	u16 trap_id;
+	u16 ci;
+
+	trap_id  = get_cqe_flow_tag(cqe);
+	ci       = mlx5_wq_cyc_ctr2ix(wq, be16_to_cpu(cqe->wqe_counter));
+	wi       = get_frag(rq, ci);
+	cqe_bcnt = be32_to_cpu(cqe->byte_cnt);
+
+	if (unlikely(MLX5E_RX_ERR_CQE(cqe))) {
+		rq->stats->wqe_err++;
+		goto free_wqe;
+	}
+
+	skb = mlx5e_skb_from_cqe_nonlinear(rq, wi, cqe_bcnt);
+	if (!skb)
+		goto free_wqe;
+
+	mlx5e_complete_rx_cqe(rq, cqe, cqe_bcnt, skb);
+	skb_push(skb, ETH_HLEN);
+
+	dl_port = mlx5e_devlink_get_dl_port(priv);
+	mlx5_devlink_trap_report(rq->mdev, trap_id, skb, dl_port);
+	dev_kfree_skb_any(skb);
+
+free_wqe:
+	mlx5e_free_rx_wqe(rq, wi, false);
+	mlx5_wq_cyc_pop(wq);
+}
+
+void mlx5e_rq_set_trap_handlers(struct mlx5e_rq *rq, struct mlx5e_params *params)
+{
+	rq->wqe.skb_from_cqe = mlx5e_rx_is_linear_skb(rq->mdev, params, NULL) ?
+			       mlx5e_skb_from_cqe_linear :
+			       mlx5e_skb_from_cqe_nonlinear;
+	rq->post_wqes = mlx5e_post_rx_wqes;
+	rq->dealloc_wqe = mlx5e_dealloc_rx_wqe;
+	rq->handle_rx_cqe = mlx5e_trap_handle_rx_cqe;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_selftest.c b/drivers/net/ethernet/mellanox/mlx5/core/en_selftest.c
new file mode 100644
index 000000000..08a75654f
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_selftest.c
@@ -0,0 +1,371 @@
+/*
+ * Copyright (c) 2016, Mellanox Technologies, Ltd.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/ip.h>
+#include <linux/udp.h>
+#include <net/udp.h>
+#include "en.h"
+#include "en/port.h"
+#include "eswitch.h"
+
+static int mlx5e_test_health_info(struct mlx5e_priv *priv)
+{
+	struct mlx5_core_health *health = &priv->mdev->priv.health;
+
+	return health->fatal_error ? 1 : 0;
+}
+
+static int mlx5e_test_link_state(struct mlx5e_priv *priv)
+{
+	u8 port_state;
+
+	if (!netif_carrier_ok(priv->netdev))
+		return 1;
+
+	port_state = mlx5_query_vport_state(priv->mdev, MLX5_VPORT_STATE_OP_MOD_VNIC_VPORT, 0);
+	return port_state == VPORT_STATE_UP ? 0 : 1;
+}
+
+static int mlx5e_test_link_speed(struct mlx5e_priv *priv)
+{
+	u32 speed;
+
+	if (!netif_carrier_ok(priv->netdev))
+		return 1;
+
+	return mlx5e_port_linkspeed(priv->mdev, &speed);
+}
+
+struct mlx5ehdr {
+	__be32 version;
+	__be64 magic;
+};
+
+#ifdef CONFIG_INET
+/* loopback test */
+#define MLX5E_TEST_PKT_SIZE (sizeof(struct ethhdr) + sizeof(struct iphdr) +\
+			     sizeof(struct udphdr) + sizeof(struct mlx5ehdr))
+#define MLX5E_TEST_MAGIC 0x5AEED15C001ULL
+
+static struct sk_buff *mlx5e_test_get_udp_skb(struct mlx5e_priv *priv)
+{
+	struct sk_buff *skb = NULL;
+	struct mlx5ehdr *mlxh;
+	struct ethhdr *ethh;
+	struct udphdr *udph;
+	struct iphdr *iph;
+	int    iplen;
+
+	skb = netdev_alloc_skb(priv->netdev, MLX5E_TEST_PKT_SIZE);
+	if (!skb) {
+		netdev_err(priv->netdev, "\tFailed to alloc loopback skb\n");
+		return NULL;
+	}
+
+	net_prefetchw(skb->data);
+	skb_reserve(skb, NET_IP_ALIGN);
+
+	/*  Reserve for ethernet and IP header  */
+	ethh = skb_push(skb, ETH_HLEN);
+	skb_reset_mac_header(skb);
+
+	skb_set_network_header(skb, skb->len);
+	iph = skb_put(skb, sizeof(struct iphdr));
+
+	skb_set_transport_header(skb, skb->len);
+	udph = skb_put(skb, sizeof(struct udphdr));
+
+	/* Fill ETH header */
+	ether_addr_copy(ethh->h_dest, priv->netdev->dev_addr);
+	eth_zero_addr(ethh->h_source);
+	ethh->h_proto = htons(ETH_P_IP);
+
+	/* Fill UDP header */
+	udph->source = htons(9);
+	udph->dest = htons(9); /* Discard Protocol */
+	udph->len = htons(sizeof(struct mlx5ehdr) + sizeof(struct udphdr));
+	udph->check = 0;
+
+	/* Fill IP header */
+	iph->ihl = 5;
+	iph->ttl = 32;
+	iph->version = 4;
+	iph->protocol = IPPROTO_UDP;
+	iplen = sizeof(struct iphdr) + sizeof(struct udphdr) +
+		sizeof(struct mlx5ehdr);
+	iph->tot_len = htons(iplen);
+	iph->frag_off = 0;
+	iph->saddr = 0;
+	iph->daddr = 0;
+	iph->tos = 0;
+	iph->id = 0;
+	ip_send_check(iph);
+
+	/* Fill test header and data */
+	mlxh = skb_put(skb, sizeof(*mlxh));
+	mlxh->version = 0;
+	mlxh->magic = cpu_to_be64(MLX5E_TEST_MAGIC);
+
+	skb->csum = 0;
+	skb->ip_summed = CHECKSUM_PARTIAL;
+	udp4_hwcsum(skb, iph->saddr, iph->daddr);
+
+	skb->protocol = htons(ETH_P_IP);
+	skb->pkt_type = PACKET_HOST;
+	skb->dev = priv->netdev;
+
+	return skb;
+}
+
+struct mlx5e_lbt_priv {
+	struct packet_type pt;
+	struct completion comp;
+	bool loopback_ok;
+	bool local_lb;
+};
+
+static int
+mlx5e_test_loopback_validate(struct sk_buff *skb,
+			     struct net_device *ndev,
+			     struct packet_type *pt,
+			     struct net_device *orig_ndev)
+{
+	struct mlx5e_lbt_priv *lbtp = pt->af_packet_priv;
+	struct mlx5ehdr *mlxh;
+	struct ethhdr *ethh;
+	struct udphdr *udph;
+	struct iphdr *iph;
+
+	/* We are only going to peek, no need to clone the SKB */
+	if (MLX5E_TEST_PKT_SIZE - ETH_HLEN > skb_headlen(skb))
+		goto out;
+
+	ethh = (struct ethhdr *)skb_mac_header(skb);
+	if (!ether_addr_equal(ethh->h_dest, orig_ndev->dev_addr))
+		goto out;
+
+	iph = ip_hdr(skb);
+	if (iph->protocol != IPPROTO_UDP)
+		goto out;
+
+	/* Don't assume skb_transport_header() was set */
+	udph = (struct udphdr *)((u8 *)iph + 4 * iph->ihl);
+	if (udph->dest != htons(9))
+		goto out;
+
+	mlxh = (struct mlx5ehdr *)((char *)udph + sizeof(*udph));
+	if (mlxh->magic != cpu_to_be64(MLX5E_TEST_MAGIC))
+		goto out; /* so close ! */
+
+	/* bingo */
+	lbtp->loopback_ok = true;
+	complete(&lbtp->comp);
+out:
+	kfree_skb(skb);
+	return 0;
+}
+
+static int mlx5e_test_loopback_setup(struct mlx5e_priv *priv,
+				     struct mlx5e_lbt_priv *lbtp)
+{
+	int err = 0;
+
+	/* Temporarily enable local_lb */
+	err = mlx5_nic_vport_query_local_lb(priv->mdev, &lbtp->local_lb);
+	if (err)
+		return err;
+
+	if (!lbtp->local_lb) {
+		err = mlx5_nic_vport_update_local_lb(priv->mdev, true);
+		if (err)
+			return err;
+	}
+
+	err = mlx5e_refresh_tirs(priv, true, false);
+	if (err)
+		goto out;
+
+	lbtp->loopback_ok = false;
+	init_completion(&lbtp->comp);
+
+	lbtp->pt.type = htons(ETH_P_IP);
+	lbtp->pt.func = mlx5e_test_loopback_validate;
+	lbtp->pt.dev = priv->netdev;
+	lbtp->pt.af_packet_priv = lbtp;
+	dev_add_pack(&lbtp->pt);
+
+	return 0;
+
+out:
+	if (!lbtp->local_lb)
+		mlx5_nic_vport_update_local_lb(priv->mdev, false);
+
+	return err;
+}
+
+static void mlx5e_test_loopback_cleanup(struct mlx5e_priv *priv,
+					struct mlx5e_lbt_priv *lbtp)
+{
+	if (!lbtp->local_lb)
+		mlx5_nic_vport_update_local_lb(priv->mdev, false);
+
+	dev_remove_pack(&lbtp->pt);
+	mlx5e_refresh_tirs(priv, false, false);
+}
+
+static int mlx5e_cond_loopback(struct mlx5e_priv *priv)
+{
+	if (is_mdev_switchdev_mode(priv->mdev))
+		return -EOPNOTSUPP;
+
+	return 0;
+}
+
+#define MLX5E_LB_VERIFY_TIMEOUT (msecs_to_jiffies(200))
+static int mlx5e_test_loopback(struct mlx5e_priv *priv)
+{
+	struct mlx5e_lbt_priv *lbtp;
+	struct sk_buff *skb = NULL;
+	int err;
+
+	if (!test_bit(MLX5E_STATE_OPENED, &priv->state)) {
+		netdev_err(priv->netdev,
+			   "\tCan't perform loopback test while device is down\n");
+		return -ENODEV;
+	}
+
+	lbtp = kzalloc(sizeof(*lbtp), GFP_KERNEL);
+	if (!lbtp)
+		return -ENOMEM;
+	lbtp->loopback_ok = false;
+
+	err = mlx5e_test_loopback_setup(priv, lbtp);
+	if (err)
+		goto out;
+
+	skb = mlx5e_test_get_udp_skb(priv);
+	if (!skb) {
+		err = -ENOMEM;
+		goto cleanup;
+	}
+
+	skb_set_queue_mapping(skb, 0);
+	err = dev_queue_xmit(skb);
+	if (err) {
+		netdev_err(priv->netdev,
+			   "\tFailed to xmit loopback packet err(%d)\n",
+			   err);
+		goto cleanup;
+	}
+
+	wait_for_completion_timeout(&lbtp->comp, MLX5E_LB_VERIFY_TIMEOUT);
+	err = !lbtp->loopback_ok;
+
+cleanup:
+	mlx5e_test_loopback_cleanup(priv, lbtp);
+out:
+	kfree(lbtp);
+	return err;
+}
+#endif
+
+typedef int (*mlx5e_st_func)(struct mlx5e_priv *);
+
+struct mlx5e_st {
+	char name[ETH_GSTRING_LEN];
+	mlx5e_st_func st_func;
+	mlx5e_st_func cond_func;
+};
+
+static struct mlx5e_st mlx5e_sts[] = {
+	{ "Link Test", mlx5e_test_link_state },
+	{ "Speed Test", mlx5e_test_link_speed },
+	{ "Health Test", mlx5e_test_health_info },
+#ifdef CONFIG_INET
+	{ "Loopback Test", mlx5e_test_loopback, mlx5e_cond_loopback },
+#endif
+};
+
+#define MLX5E_ST_NUM ARRAY_SIZE(mlx5e_sts)
+
+void mlx5e_self_test(struct net_device *ndev, struct ethtool_test *etest,
+		     u64 *buf)
+{
+	struct mlx5e_priv *priv = netdev_priv(ndev);
+	int i, count = 0;
+
+	mutex_lock(&priv->state_lock);
+	netdev_info(ndev, "Self test begin..\n");
+
+	for (i = 0; i < MLX5E_ST_NUM; i++) {
+		struct mlx5e_st st = mlx5e_sts[i];
+
+		if (st.cond_func && st.cond_func(priv))
+			continue;
+		netdev_info(ndev, "\t[%d] %s start..\n", i, st.name);
+		buf[count] = st.st_func(priv);
+		netdev_info(ndev, "\t[%d] %s end: result(%lld)\n", i, st.name, buf[count]);
+		count++;
+	}
+
+	mutex_unlock(&priv->state_lock);
+
+	for (i = 0; i < count; i++) {
+		if (buf[i]) {
+			etest->flags |= ETH_TEST_FL_FAILED;
+			break;
+		}
+	}
+	netdev_info(ndev, "Self test out: status flags(0x%x)\n",
+		    etest->flags);
+}
+
+int mlx5e_self_test_fill_strings(struct mlx5e_priv *priv, u8 *data)
+{
+	int i, count = 0;
+
+	for (i = 0; i < MLX5E_ST_NUM; i++) {
+		struct mlx5e_st st = mlx5e_sts[i];
+
+		if (st.cond_func && st.cond_func(priv))
+			continue;
+		if (data)
+			strcpy(data + count * ETH_GSTRING_LEN, st.name);
+		count++;
+	}
+	return count;
+}
+
+int mlx5e_self_test_num(struct mlx5e_priv *priv)
+{
+	return mlx5e_self_test_fill_strings(priv, NULL);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_stats.c b/drivers/net/ethernet/mellanox/mlx5/core/en_stats.c
new file mode 100644
index 000000000..f7f54550a
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_stats.c
@@ -0,0 +1,2482 @@
+/*
+ * Copyright (c) 2017, Mellanox Technologies, Ltd.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "lib/mlx5.h"
+#include "en.h"
+#include "en_accel/ktls.h"
+#include "en_accel/en_accel.h"
+#include "en/ptp.h"
+#include "en/port.h"
+
+#ifdef CONFIG_PAGE_POOL_STATS
+#include <net/page_pool.h>
+#endif
+
+static unsigned int stats_grps_num(struct mlx5e_priv *priv)
+{
+	return !priv->profile->stats_grps_num ? 0 :
+		priv->profile->stats_grps_num(priv);
+}
+
+unsigned int mlx5e_stats_total_num(struct mlx5e_priv *priv)
+{
+	mlx5e_stats_grp_t *stats_grps = priv->profile->stats_grps;
+	const unsigned int num_stats_grps = stats_grps_num(priv);
+	unsigned int total = 0;
+	int i;
+
+	for (i = 0; i < num_stats_grps; i++)
+		total += stats_grps[i]->get_num_stats(priv);
+
+	return total;
+}
+
+void mlx5e_stats_update_ndo_stats(struct mlx5e_priv *priv)
+{
+	mlx5e_stats_grp_t *stats_grps = priv->profile->stats_grps;
+	const unsigned int num_stats_grps = stats_grps_num(priv);
+	int i;
+
+	for (i = num_stats_grps - 1; i >= 0; i--)
+		if (stats_grps[i]->update_stats &&
+		    stats_grps[i]->update_stats_mask & MLX5E_NDO_UPDATE_STATS)
+			stats_grps[i]->update_stats(priv);
+}
+
+void mlx5e_stats_update(struct mlx5e_priv *priv)
+{
+	mlx5e_stats_grp_t *stats_grps = priv->profile->stats_grps;
+	const unsigned int num_stats_grps = stats_grps_num(priv);
+	int i;
+
+	for (i = num_stats_grps - 1; i >= 0; i--)
+		if (stats_grps[i]->update_stats)
+			stats_grps[i]->update_stats(priv);
+}
+
+void mlx5e_stats_fill(struct mlx5e_priv *priv, u64 *data, int idx)
+{
+	mlx5e_stats_grp_t *stats_grps = priv->profile->stats_grps;
+	const unsigned int num_stats_grps = stats_grps_num(priv);
+	int i;
+
+	for (i = 0; i < num_stats_grps; i++)
+		idx = stats_grps[i]->fill_stats(priv, data, idx);
+}
+
+void mlx5e_stats_fill_strings(struct mlx5e_priv *priv, u8 *data)
+{
+	mlx5e_stats_grp_t *stats_grps = priv->profile->stats_grps;
+	const unsigned int num_stats_grps = stats_grps_num(priv);
+	int i, idx = 0;
+
+	for (i = 0; i < num_stats_grps; i++)
+		idx = stats_grps[i]->fill_strings(priv, data, idx);
+}
+
+/* Concrete NIC Stats */
+
+static const struct counter_desc sw_stats_desc[] = {
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_packets) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_bytes) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_packets) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_bytes) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_tso_packets) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_tso_bytes) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_tso_inner_packets) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_tso_inner_bytes) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_added_vlan_packets) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_nop) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_mpwqe_blks) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_mpwqe_pkts) },
+
+#ifdef CONFIG_MLX5_EN_TLS
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_tls_encrypted_packets) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_tls_encrypted_bytes) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_tls_ooo) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_tls_dump_packets) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_tls_dump_bytes) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_tls_resync_bytes) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_tls_skip_no_sync_data) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_tls_drop_no_sync_data) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_tls_drop_bypass_req) },
+#endif
+
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_lro_packets) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_lro_bytes) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_gro_packets) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_gro_bytes) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_gro_skbs) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_gro_match_packets) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_gro_large_hds) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_ecn_mark) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_removed_vlan_packets) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_csum_unnecessary) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_csum_none) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_csum_complete) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_csum_complete_tail) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_csum_complete_tail_slow) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_csum_unnecessary_inner) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_xdp_drop) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_xdp_redirect) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_xdp_tx_xmit) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_xdp_tx_mpwqe) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_xdp_tx_inlnw) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_xdp_tx_nops) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_xdp_tx_full) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_xdp_tx_err) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_xdp_tx_cqe) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_csum_none) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_csum_partial) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_csum_partial_inner) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_queue_stopped) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_queue_dropped) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_xmit_more) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_recover) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_cqes) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_queue_wake) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_cqe_err) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_xdp_xmit) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_xdp_mpwqe) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_xdp_inlnw) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_xdp_nops) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_xdp_full) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_xdp_err) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_xdp_cqes) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_wqe_err) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_mpwqe_filler_cqes) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_mpwqe_filler_strides) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_oversize_pkts_sw_drop) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_buff_alloc_err) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_cqe_compress_blks) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_cqe_compress_pkts) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_cache_reuse) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_cache_full) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_cache_empty) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_cache_busy) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_cache_waive) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_congst_umr) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_arfs_err) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_recover) },
+#ifdef CONFIG_PAGE_POOL_STATS
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_pp_alloc_fast) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_pp_alloc_slow) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_pp_alloc_slow_high_order) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_pp_alloc_empty) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_pp_alloc_refill) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_pp_alloc_waive) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_pp_recycle_cached) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_pp_recycle_cache_full) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_pp_recycle_ring) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_pp_recycle_ring_full) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_pp_recycle_released_ref) },
+#endif
+#ifdef CONFIG_MLX5_EN_TLS
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_tls_decrypted_packets) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_tls_decrypted_bytes) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_tls_resync_req_pkt) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_tls_resync_req_start) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_tls_resync_req_end) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_tls_resync_req_skip) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_tls_resync_res_ok) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_tls_resync_res_retry) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_tls_resync_res_skip) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_tls_err) },
+#endif
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, ch_events) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, ch_poll) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, ch_arm) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, ch_aff_change) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, ch_force_irq) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, ch_eq_rearm) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_xsk_packets) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_xsk_bytes) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_xsk_csum_complete) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_xsk_csum_unnecessary) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_xsk_csum_unnecessary_inner) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_xsk_csum_none) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_xsk_ecn_mark) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_xsk_removed_vlan_packets) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_xsk_xdp_drop) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_xsk_xdp_redirect) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_xsk_wqe_err) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_xsk_mpwqe_filler_cqes) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_xsk_mpwqe_filler_strides) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_xsk_oversize_pkts_sw_drop) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_xsk_buff_alloc_err) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_xsk_cqe_compress_blks) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_xsk_cqe_compress_pkts) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_xsk_congst_umr) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_xsk_arfs_err) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_xsk_xmit) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_xsk_mpwqe) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_xsk_inlnw) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_xsk_full) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_xsk_err) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_xsk_cqes) },
+};
+
+#define NUM_SW_COUNTERS			ARRAY_SIZE(sw_stats_desc)
+
+static MLX5E_DECLARE_STATS_GRP_OP_NUM_STATS(sw)
+{
+	return NUM_SW_COUNTERS;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_FILL_STRS(sw)
+{
+	int i;
+
+	for (i = 0; i < NUM_SW_COUNTERS; i++)
+		strcpy(data + (idx++) * ETH_GSTRING_LEN, sw_stats_desc[i].format);
+	return idx;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_FILL_STATS(sw)
+{
+	int i;
+
+	for (i = 0; i < NUM_SW_COUNTERS; i++)
+		data[idx++] = MLX5E_READ_CTR64_CPU(&priv->stats.sw, sw_stats_desc, i);
+	return idx;
+}
+
+static void mlx5e_stats_grp_sw_update_stats_xdp_red(struct mlx5e_sw_stats *s,
+						    struct mlx5e_xdpsq_stats *xdpsq_red_stats)
+{
+	s->tx_xdp_xmit  += xdpsq_red_stats->xmit;
+	s->tx_xdp_mpwqe += xdpsq_red_stats->mpwqe;
+	s->tx_xdp_inlnw += xdpsq_red_stats->inlnw;
+	s->tx_xdp_nops  += xdpsq_red_stats->nops;
+	s->tx_xdp_full  += xdpsq_red_stats->full;
+	s->tx_xdp_err   += xdpsq_red_stats->err;
+	s->tx_xdp_cqes  += xdpsq_red_stats->cqes;
+}
+
+static void mlx5e_stats_grp_sw_update_stats_xdpsq(struct mlx5e_sw_stats *s,
+						  struct mlx5e_xdpsq_stats *xdpsq_stats)
+{
+	s->rx_xdp_tx_xmit  += xdpsq_stats->xmit;
+	s->rx_xdp_tx_mpwqe += xdpsq_stats->mpwqe;
+	s->rx_xdp_tx_inlnw += xdpsq_stats->inlnw;
+	s->rx_xdp_tx_nops  += xdpsq_stats->nops;
+	s->rx_xdp_tx_full  += xdpsq_stats->full;
+	s->rx_xdp_tx_err   += xdpsq_stats->err;
+	s->rx_xdp_tx_cqe   += xdpsq_stats->cqes;
+}
+
+static void mlx5e_stats_grp_sw_update_stats_xsksq(struct mlx5e_sw_stats *s,
+						  struct mlx5e_xdpsq_stats *xsksq_stats)
+{
+	s->tx_xsk_xmit  += xsksq_stats->xmit;
+	s->tx_xsk_mpwqe += xsksq_stats->mpwqe;
+	s->tx_xsk_inlnw += xsksq_stats->inlnw;
+	s->tx_xsk_full  += xsksq_stats->full;
+	s->tx_xsk_err   += xsksq_stats->err;
+	s->tx_xsk_cqes  += xsksq_stats->cqes;
+}
+
+static void mlx5e_stats_grp_sw_update_stats_xskrq(struct mlx5e_sw_stats *s,
+						  struct mlx5e_rq_stats *xskrq_stats)
+{
+	s->rx_xsk_packets                += xskrq_stats->packets;
+	s->rx_xsk_bytes                  += xskrq_stats->bytes;
+	s->rx_xsk_csum_complete          += xskrq_stats->csum_complete;
+	s->rx_xsk_csum_unnecessary       += xskrq_stats->csum_unnecessary;
+	s->rx_xsk_csum_unnecessary_inner += xskrq_stats->csum_unnecessary_inner;
+	s->rx_xsk_csum_none              += xskrq_stats->csum_none;
+	s->rx_xsk_ecn_mark               += xskrq_stats->ecn_mark;
+	s->rx_xsk_removed_vlan_packets   += xskrq_stats->removed_vlan_packets;
+	s->rx_xsk_xdp_drop               += xskrq_stats->xdp_drop;
+	s->rx_xsk_xdp_redirect           += xskrq_stats->xdp_redirect;
+	s->rx_xsk_wqe_err                += xskrq_stats->wqe_err;
+	s->rx_xsk_mpwqe_filler_cqes      += xskrq_stats->mpwqe_filler_cqes;
+	s->rx_xsk_mpwqe_filler_strides   += xskrq_stats->mpwqe_filler_strides;
+	s->rx_xsk_oversize_pkts_sw_drop  += xskrq_stats->oversize_pkts_sw_drop;
+	s->rx_xsk_buff_alloc_err         += xskrq_stats->buff_alloc_err;
+	s->rx_xsk_cqe_compress_blks      += xskrq_stats->cqe_compress_blks;
+	s->rx_xsk_cqe_compress_pkts      += xskrq_stats->cqe_compress_pkts;
+	s->rx_xsk_congst_umr             += xskrq_stats->congst_umr;
+	s->rx_xsk_arfs_err               += xskrq_stats->arfs_err;
+}
+
+static void mlx5e_stats_grp_sw_update_stats_rq_stats(struct mlx5e_sw_stats *s,
+						     struct mlx5e_rq_stats *rq_stats)
+{
+	s->rx_packets                 += rq_stats->packets;
+	s->rx_bytes                   += rq_stats->bytes;
+	s->rx_lro_packets             += rq_stats->lro_packets;
+	s->rx_lro_bytes               += rq_stats->lro_bytes;
+	s->rx_gro_packets             += rq_stats->gro_packets;
+	s->rx_gro_bytes               += rq_stats->gro_bytes;
+	s->rx_gro_skbs                += rq_stats->gro_skbs;
+	s->rx_gro_match_packets       += rq_stats->gro_match_packets;
+	s->rx_gro_large_hds           += rq_stats->gro_large_hds;
+	s->rx_ecn_mark                += rq_stats->ecn_mark;
+	s->rx_removed_vlan_packets    += rq_stats->removed_vlan_packets;
+	s->rx_csum_none               += rq_stats->csum_none;
+	s->rx_csum_complete           += rq_stats->csum_complete;
+	s->rx_csum_complete_tail      += rq_stats->csum_complete_tail;
+	s->rx_csum_complete_tail_slow += rq_stats->csum_complete_tail_slow;
+	s->rx_csum_unnecessary        += rq_stats->csum_unnecessary;
+	s->rx_csum_unnecessary_inner  += rq_stats->csum_unnecessary_inner;
+	s->rx_xdp_drop                += rq_stats->xdp_drop;
+	s->rx_xdp_redirect            += rq_stats->xdp_redirect;
+	s->rx_wqe_err                 += rq_stats->wqe_err;
+	s->rx_mpwqe_filler_cqes       += rq_stats->mpwqe_filler_cqes;
+	s->rx_mpwqe_filler_strides    += rq_stats->mpwqe_filler_strides;
+	s->rx_oversize_pkts_sw_drop   += rq_stats->oversize_pkts_sw_drop;
+	s->rx_buff_alloc_err          += rq_stats->buff_alloc_err;
+	s->rx_cqe_compress_blks       += rq_stats->cqe_compress_blks;
+	s->rx_cqe_compress_pkts       += rq_stats->cqe_compress_pkts;
+	s->rx_cache_reuse             += rq_stats->cache_reuse;
+	s->rx_cache_full              += rq_stats->cache_full;
+	s->rx_cache_empty             += rq_stats->cache_empty;
+	s->rx_cache_busy              += rq_stats->cache_busy;
+	s->rx_cache_waive             += rq_stats->cache_waive;
+	s->rx_congst_umr              += rq_stats->congst_umr;
+	s->rx_arfs_err                += rq_stats->arfs_err;
+	s->rx_recover                 += rq_stats->recover;
+#ifdef CONFIG_PAGE_POOL_STATS
+	s->rx_pp_alloc_fast          += rq_stats->pp_alloc_fast;
+	s->rx_pp_alloc_slow          += rq_stats->pp_alloc_slow;
+	s->rx_pp_alloc_empty         += rq_stats->pp_alloc_empty;
+	s->rx_pp_alloc_refill        += rq_stats->pp_alloc_refill;
+	s->rx_pp_alloc_waive         += rq_stats->pp_alloc_waive;
+	s->rx_pp_alloc_slow_high_order		+= rq_stats->pp_alloc_slow_high_order;
+	s->rx_pp_recycle_cached			+= rq_stats->pp_recycle_cached;
+	s->rx_pp_recycle_cache_full		+= rq_stats->pp_recycle_cache_full;
+	s->rx_pp_recycle_ring			+= rq_stats->pp_recycle_ring;
+	s->rx_pp_recycle_ring_full		+= rq_stats->pp_recycle_ring_full;
+	s->rx_pp_recycle_released_ref		+= rq_stats->pp_recycle_released_ref;
+#endif
+#ifdef CONFIG_MLX5_EN_TLS
+	s->rx_tls_decrypted_packets   += rq_stats->tls_decrypted_packets;
+	s->rx_tls_decrypted_bytes     += rq_stats->tls_decrypted_bytes;
+	s->rx_tls_resync_req_pkt      += rq_stats->tls_resync_req_pkt;
+	s->rx_tls_resync_req_start    += rq_stats->tls_resync_req_start;
+	s->rx_tls_resync_req_end      += rq_stats->tls_resync_req_end;
+	s->rx_tls_resync_req_skip     += rq_stats->tls_resync_req_skip;
+	s->rx_tls_resync_res_ok       += rq_stats->tls_resync_res_ok;
+	s->rx_tls_resync_res_retry    += rq_stats->tls_resync_res_retry;
+	s->rx_tls_resync_res_skip     += rq_stats->tls_resync_res_skip;
+	s->rx_tls_err                 += rq_stats->tls_err;
+#endif
+}
+
+static void mlx5e_stats_grp_sw_update_stats_ch_stats(struct mlx5e_sw_stats *s,
+						     struct mlx5e_ch_stats *ch_stats)
+{
+	s->ch_events      += ch_stats->events;
+	s->ch_poll        += ch_stats->poll;
+	s->ch_arm         += ch_stats->arm;
+	s->ch_aff_change  += ch_stats->aff_change;
+	s->ch_force_irq   += ch_stats->force_irq;
+	s->ch_eq_rearm    += ch_stats->eq_rearm;
+}
+
+static void mlx5e_stats_grp_sw_update_stats_sq(struct mlx5e_sw_stats *s,
+					       struct mlx5e_sq_stats *sq_stats)
+{
+	s->tx_packets               += sq_stats->packets;
+	s->tx_bytes                 += sq_stats->bytes;
+	s->tx_tso_packets           += sq_stats->tso_packets;
+	s->tx_tso_bytes             += sq_stats->tso_bytes;
+	s->tx_tso_inner_packets     += sq_stats->tso_inner_packets;
+	s->tx_tso_inner_bytes       += sq_stats->tso_inner_bytes;
+	s->tx_added_vlan_packets    += sq_stats->added_vlan_packets;
+	s->tx_nop                   += sq_stats->nop;
+	s->tx_mpwqe_blks            += sq_stats->mpwqe_blks;
+	s->tx_mpwqe_pkts            += sq_stats->mpwqe_pkts;
+	s->tx_queue_stopped         += sq_stats->stopped;
+	s->tx_queue_wake            += sq_stats->wake;
+	s->tx_queue_dropped         += sq_stats->dropped;
+	s->tx_cqe_err               += sq_stats->cqe_err;
+	s->tx_recover               += sq_stats->recover;
+	s->tx_xmit_more             += sq_stats->xmit_more;
+	s->tx_csum_partial_inner    += sq_stats->csum_partial_inner;
+	s->tx_csum_none             += sq_stats->csum_none;
+	s->tx_csum_partial          += sq_stats->csum_partial;
+#ifdef CONFIG_MLX5_EN_TLS
+	s->tx_tls_encrypted_packets += sq_stats->tls_encrypted_packets;
+	s->tx_tls_encrypted_bytes   += sq_stats->tls_encrypted_bytes;
+	s->tx_tls_ooo               += sq_stats->tls_ooo;
+	s->tx_tls_dump_bytes        += sq_stats->tls_dump_bytes;
+	s->tx_tls_dump_packets      += sq_stats->tls_dump_packets;
+	s->tx_tls_resync_bytes      += sq_stats->tls_resync_bytes;
+	s->tx_tls_skip_no_sync_data += sq_stats->tls_skip_no_sync_data;
+	s->tx_tls_drop_no_sync_data += sq_stats->tls_drop_no_sync_data;
+	s->tx_tls_drop_bypass_req   += sq_stats->tls_drop_bypass_req;
+#endif
+	s->tx_cqes                  += sq_stats->cqes;
+}
+
+static void mlx5e_stats_grp_sw_update_stats_ptp(struct mlx5e_priv *priv,
+						struct mlx5e_sw_stats *s)
+{
+	int i;
+
+	if (!priv->tx_ptp_opened && !priv->rx_ptp_opened)
+		return;
+
+	mlx5e_stats_grp_sw_update_stats_ch_stats(s, &priv->ptp_stats.ch);
+
+	if (priv->tx_ptp_opened) {
+		for (i = 0; i < priv->max_opened_tc; i++) {
+			mlx5e_stats_grp_sw_update_stats_sq(s, &priv->ptp_stats.sq[i]);
+
+			/* https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92657 */
+			barrier();
+		}
+	}
+	if (priv->rx_ptp_opened) {
+		mlx5e_stats_grp_sw_update_stats_rq_stats(s, &priv->ptp_stats.rq);
+
+		/* https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92657 */
+		barrier();
+	}
+}
+
+static void mlx5e_stats_grp_sw_update_stats_qos(struct mlx5e_priv *priv,
+						struct mlx5e_sw_stats *s)
+{
+	struct mlx5e_sq_stats **stats;
+	u16 max_qos_sqs;
+	int i;
+
+	/* Pairs with smp_store_release in mlx5e_open_qos_sq. */
+	max_qos_sqs = smp_load_acquire(&priv->htb_max_qos_sqs);
+	stats = READ_ONCE(priv->htb_qos_sq_stats);
+
+	for (i = 0; i < max_qos_sqs; i++) {
+		mlx5e_stats_grp_sw_update_stats_sq(s, READ_ONCE(stats[i]));
+
+		/* https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92657 */
+		barrier();
+	}
+}
+
+#ifdef CONFIG_PAGE_POOL_STATS
+static void mlx5e_stats_update_stats_rq_page_pool(struct mlx5e_channel *c)
+{
+	struct mlx5e_rq_stats *rq_stats = c->rq.stats;
+	struct page_pool *pool = c->rq.page_pool;
+	struct page_pool_stats stats = { 0 };
+
+	if (!page_pool_get_stats(pool, &stats))
+		return;
+
+	rq_stats->pp_alloc_fast = stats.alloc_stats.fast;
+	rq_stats->pp_alloc_slow = stats.alloc_stats.slow;
+	rq_stats->pp_alloc_slow_high_order = stats.alloc_stats.slow_high_order;
+	rq_stats->pp_alloc_empty = stats.alloc_stats.empty;
+	rq_stats->pp_alloc_waive = stats.alloc_stats.waive;
+	rq_stats->pp_alloc_refill = stats.alloc_stats.refill;
+
+	rq_stats->pp_recycle_cached = stats.recycle_stats.cached;
+	rq_stats->pp_recycle_cache_full = stats.recycle_stats.cache_full;
+	rq_stats->pp_recycle_ring = stats.recycle_stats.ring;
+	rq_stats->pp_recycle_ring_full = stats.recycle_stats.ring_full;
+	rq_stats->pp_recycle_released_ref = stats.recycle_stats.released_refcnt;
+}
+#else
+static void mlx5e_stats_update_stats_rq_page_pool(struct mlx5e_channel *c)
+{
+}
+#endif
+
+static MLX5E_DECLARE_STATS_GRP_OP_UPDATE_STATS(sw)
+{
+	struct mlx5e_sw_stats *s = &priv->stats.sw;
+	int i;
+
+	memset(s, 0, sizeof(*s));
+
+	for (i = 0; i < priv->channels.num; i++) /* for active channels only */
+		mlx5e_stats_update_stats_rq_page_pool(priv->channels.c[i]);
+
+	for (i = 0; i < priv->stats_nch; i++) {
+		struct mlx5e_channel_stats *channel_stats =
+			priv->channel_stats[i];
+
+		int j;
+
+		mlx5e_stats_grp_sw_update_stats_rq_stats(s, &channel_stats->rq);
+		mlx5e_stats_grp_sw_update_stats_xdpsq(s, &channel_stats->rq_xdpsq);
+		mlx5e_stats_grp_sw_update_stats_ch_stats(s, &channel_stats->ch);
+		/* xdp redirect */
+		mlx5e_stats_grp_sw_update_stats_xdp_red(s, &channel_stats->xdpsq);
+		/* AF_XDP zero-copy */
+		mlx5e_stats_grp_sw_update_stats_xskrq(s, &channel_stats->xskrq);
+		mlx5e_stats_grp_sw_update_stats_xsksq(s, &channel_stats->xsksq);
+
+		for (j = 0; j < priv->max_opened_tc; j++) {
+			mlx5e_stats_grp_sw_update_stats_sq(s, &channel_stats->sq[j]);
+
+			/* https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92657 */
+			barrier();
+		}
+	}
+	mlx5e_stats_grp_sw_update_stats_ptp(priv, s);
+	mlx5e_stats_grp_sw_update_stats_qos(priv, s);
+}
+
+static const struct counter_desc q_stats_desc[] = {
+	{ MLX5E_DECLARE_STAT(struct mlx5e_qcounter_stats, rx_out_of_buffer) },
+};
+
+static const struct counter_desc drop_rq_stats_desc[] = {
+	{ MLX5E_DECLARE_STAT(struct mlx5e_qcounter_stats, rx_if_down_packets) },
+};
+
+#define NUM_Q_COUNTERS			ARRAY_SIZE(q_stats_desc)
+#define NUM_DROP_RQ_COUNTERS		ARRAY_SIZE(drop_rq_stats_desc)
+
+static MLX5E_DECLARE_STATS_GRP_OP_NUM_STATS(qcnt)
+{
+	int num_stats = 0;
+
+	if (priv->q_counter)
+		num_stats += NUM_Q_COUNTERS;
+
+	if (priv->drop_rq_q_counter)
+		num_stats += NUM_DROP_RQ_COUNTERS;
+
+	return num_stats;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_FILL_STRS(qcnt)
+{
+	int i;
+
+	for (i = 0; i < NUM_Q_COUNTERS && priv->q_counter; i++)
+		strcpy(data + (idx++) * ETH_GSTRING_LEN,
+		       q_stats_desc[i].format);
+
+	for (i = 0; i < NUM_DROP_RQ_COUNTERS && priv->drop_rq_q_counter; i++)
+		strcpy(data + (idx++) * ETH_GSTRING_LEN,
+		       drop_rq_stats_desc[i].format);
+
+	return idx;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_FILL_STATS(qcnt)
+{
+	int i;
+
+	for (i = 0; i < NUM_Q_COUNTERS && priv->q_counter; i++)
+		data[idx++] = MLX5E_READ_CTR32_CPU(&priv->stats.qcnt,
+						   q_stats_desc, i);
+	for (i = 0; i < NUM_DROP_RQ_COUNTERS && priv->drop_rq_q_counter; i++)
+		data[idx++] = MLX5E_READ_CTR32_CPU(&priv->stats.qcnt,
+						   drop_rq_stats_desc, i);
+	return idx;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_UPDATE_STATS(qcnt)
+{
+	struct mlx5e_qcounter_stats *qcnt = &priv->stats.qcnt;
+	u32 out[MLX5_ST_SZ_DW(query_q_counter_out)] = {};
+	u32 in[MLX5_ST_SZ_DW(query_q_counter_in)] = {};
+	int ret;
+
+	MLX5_SET(query_q_counter_in, in, opcode, MLX5_CMD_OP_QUERY_Q_COUNTER);
+
+	if (priv->q_counter) {
+		MLX5_SET(query_q_counter_in, in, counter_set_id,
+			 priv->q_counter);
+		ret = mlx5_cmd_exec_inout(priv->mdev, query_q_counter, in, out);
+		if (!ret)
+			qcnt->rx_out_of_buffer = MLX5_GET(query_q_counter_out,
+							  out, out_of_buffer);
+	}
+
+	if (priv->drop_rq_q_counter) {
+		MLX5_SET(query_q_counter_in, in, counter_set_id,
+			 priv->drop_rq_q_counter);
+		ret = mlx5_cmd_exec_inout(priv->mdev, query_q_counter, in, out);
+		if (!ret)
+			qcnt->rx_if_down_packets = MLX5_GET(query_q_counter_out,
+							    out, out_of_buffer);
+	}
+}
+
+#define VNIC_ENV_OFF(c) MLX5_BYTE_OFF(query_vnic_env_out, c)
+static const struct counter_desc vnic_env_stats_steer_desc[] = {
+	{ "rx_steer_missed_packets",
+		VNIC_ENV_OFF(vport_env.nic_receive_steering_discard) },
+};
+
+static const struct counter_desc vnic_env_stats_dev_oob_desc[] = {
+	{ "dev_internal_queue_oob",
+		VNIC_ENV_OFF(vport_env.internal_rq_out_of_buffer) },
+};
+
+static const struct counter_desc vnic_env_stats_drop_desc[] = {
+	{ "rx_oversize_pkts_buffer",
+		VNIC_ENV_OFF(vport_env.eth_wqe_too_small) },
+};
+
+#define NUM_VNIC_ENV_STEER_COUNTERS(dev) \
+	(MLX5_CAP_GEN(dev, nic_receive_steering_discard) ? \
+	 ARRAY_SIZE(vnic_env_stats_steer_desc) : 0)
+#define NUM_VNIC_ENV_DEV_OOB_COUNTERS(dev) \
+	(MLX5_CAP_GEN(dev, vnic_env_int_rq_oob) ? \
+	 ARRAY_SIZE(vnic_env_stats_dev_oob_desc) : 0)
+#define NUM_VNIC_ENV_DROP_COUNTERS(dev) \
+	(MLX5_CAP_GEN(dev, eth_wqe_too_small) ? \
+	 ARRAY_SIZE(vnic_env_stats_drop_desc) : 0)
+
+static MLX5E_DECLARE_STATS_GRP_OP_NUM_STATS(vnic_env)
+{
+	return NUM_VNIC_ENV_STEER_COUNTERS(priv->mdev) +
+	       NUM_VNIC_ENV_DEV_OOB_COUNTERS(priv->mdev) +
+	       NUM_VNIC_ENV_DROP_COUNTERS(priv->mdev);
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_FILL_STRS(vnic_env)
+{
+	int i;
+
+	for (i = 0; i < NUM_VNIC_ENV_STEER_COUNTERS(priv->mdev); i++)
+		strcpy(data + (idx++) * ETH_GSTRING_LEN,
+		       vnic_env_stats_steer_desc[i].format);
+
+	for (i = 0; i < NUM_VNIC_ENV_DEV_OOB_COUNTERS(priv->mdev); i++)
+		strcpy(data + (idx++) * ETH_GSTRING_LEN,
+		       vnic_env_stats_dev_oob_desc[i].format);
+
+	for (i = 0; i < NUM_VNIC_ENV_DROP_COUNTERS(priv->mdev); i++)
+		strcpy(data + (idx++) * ETH_GSTRING_LEN,
+		       vnic_env_stats_drop_desc[i].format);
+
+	return idx;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_FILL_STATS(vnic_env)
+{
+	int i;
+
+	for (i = 0; i < NUM_VNIC_ENV_STEER_COUNTERS(priv->mdev); i++)
+		data[idx++] = MLX5E_READ_CTR64_BE(priv->stats.vnic.query_vnic_env_out,
+						  vnic_env_stats_steer_desc, i);
+
+	for (i = 0; i < NUM_VNIC_ENV_DEV_OOB_COUNTERS(priv->mdev); i++)
+		data[idx++] = MLX5E_READ_CTR32_BE(priv->stats.vnic.query_vnic_env_out,
+						  vnic_env_stats_dev_oob_desc, i);
+
+	for (i = 0; i < NUM_VNIC_ENV_DROP_COUNTERS(priv->mdev); i++)
+		data[idx++] = MLX5E_READ_CTR32_BE(priv->stats.vnic.query_vnic_env_out,
+						  vnic_env_stats_drop_desc, i);
+
+	return idx;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_UPDATE_STATS(vnic_env)
+{
+	u32 *out = (u32 *)priv->stats.vnic.query_vnic_env_out;
+	u32 in[MLX5_ST_SZ_DW(query_vnic_env_in)] = {};
+	struct mlx5_core_dev *mdev = priv->mdev;
+
+	if (!mlx5e_stats_grp_vnic_env_num_stats(priv))
+		return;
+
+	MLX5_SET(query_vnic_env_in, in, opcode, MLX5_CMD_OP_QUERY_VNIC_ENV);
+	mlx5_cmd_exec_inout(mdev, query_vnic_env, in, out);
+}
+
+#define VPORT_COUNTER_OFF(c) MLX5_BYTE_OFF(query_vport_counter_out, c)
+static const struct counter_desc vport_stats_desc[] = {
+	{ "rx_vport_unicast_packets",
+		VPORT_COUNTER_OFF(received_eth_unicast.packets) },
+	{ "rx_vport_unicast_bytes",
+		VPORT_COUNTER_OFF(received_eth_unicast.octets) },
+	{ "tx_vport_unicast_packets",
+		VPORT_COUNTER_OFF(transmitted_eth_unicast.packets) },
+	{ "tx_vport_unicast_bytes",
+		VPORT_COUNTER_OFF(transmitted_eth_unicast.octets) },
+	{ "rx_vport_multicast_packets",
+		VPORT_COUNTER_OFF(received_eth_multicast.packets) },
+	{ "rx_vport_multicast_bytes",
+		VPORT_COUNTER_OFF(received_eth_multicast.octets) },
+	{ "tx_vport_multicast_packets",
+		VPORT_COUNTER_OFF(transmitted_eth_multicast.packets) },
+	{ "tx_vport_multicast_bytes",
+		VPORT_COUNTER_OFF(transmitted_eth_multicast.octets) },
+	{ "rx_vport_broadcast_packets",
+		VPORT_COUNTER_OFF(received_eth_broadcast.packets) },
+	{ "rx_vport_broadcast_bytes",
+		VPORT_COUNTER_OFF(received_eth_broadcast.octets) },
+	{ "tx_vport_broadcast_packets",
+		VPORT_COUNTER_OFF(transmitted_eth_broadcast.packets) },
+	{ "tx_vport_broadcast_bytes",
+		VPORT_COUNTER_OFF(transmitted_eth_broadcast.octets) },
+	{ "rx_vport_rdma_unicast_packets",
+		VPORT_COUNTER_OFF(received_ib_unicast.packets) },
+	{ "rx_vport_rdma_unicast_bytes",
+		VPORT_COUNTER_OFF(received_ib_unicast.octets) },
+	{ "tx_vport_rdma_unicast_packets",
+		VPORT_COUNTER_OFF(transmitted_ib_unicast.packets) },
+	{ "tx_vport_rdma_unicast_bytes",
+		VPORT_COUNTER_OFF(transmitted_ib_unicast.octets) },
+	{ "rx_vport_rdma_multicast_packets",
+		VPORT_COUNTER_OFF(received_ib_multicast.packets) },
+	{ "rx_vport_rdma_multicast_bytes",
+		VPORT_COUNTER_OFF(received_ib_multicast.octets) },
+	{ "tx_vport_rdma_multicast_packets",
+		VPORT_COUNTER_OFF(transmitted_ib_multicast.packets) },
+	{ "tx_vport_rdma_multicast_bytes",
+		VPORT_COUNTER_OFF(transmitted_ib_multicast.octets) },
+};
+
+#define NUM_VPORT_COUNTERS		ARRAY_SIZE(vport_stats_desc)
+
+static MLX5E_DECLARE_STATS_GRP_OP_NUM_STATS(vport)
+{
+	return NUM_VPORT_COUNTERS;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_FILL_STRS(vport)
+{
+	int i;
+
+	for (i = 0; i < NUM_VPORT_COUNTERS; i++)
+		strcpy(data + (idx++) * ETH_GSTRING_LEN, vport_stats_desc[i].format);
+	return idx;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_FILL_STATS(vport)
+{
+	int i;
+
+	for (i = 0; i < NUM_VPORT_COUNTERS; i++)
+		data[idx++] = MLX5E_READ_CTR64_BE(priv->stats.vport.query_vport_out,
+						  vport_stats_desc, i);
+	return idx;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_UPDATE_STATS(vport)
+{
+	u32 *out = (u32 *)priv->stats.vport.query_vport_out;
+	u32 in[MLX5_ST_SZ_DW(query_vport_counter_in)] = {};
+	struct mlx5_core_dev *mdev = priv->mdev;
+
+	MLX5_SET(query_vport_counter_in, in, opcode, MLX5_CMD_OP_QUERY_VPORT_COUNTER);
+	mlx5_cmd_exec_inout(mdev, query_vport_counter, in, out);
+}
+
+#define PPORT_802_3_OFF(c) \
+	MLX5_BYTE_OFF(ppcnt_reg, \
+		      counter_set.eth_802_3_cntrs_grp_data_layout.c##_high)
+static const struct counter_desc pport_802_3_stats_desc[] = {
+	{ "tx_packets_phy", PPORT_802_3_OFF(a_frames_transmitted_ok) },
+	{ "rx_packets_phy", PPORT_802_3_OFF(a_frames_received_ok) },
+	{ "rx_crc_errors_phy", PPORT_802_3_OFF(a_frame_check_sequence_errors) },
+	{ "tx_bytes_phy", PPORT_802_3_OFF(a_octets_transmitted_ok) },
+	{ "rx_bytes_phy", PPORT_802_3_OFF(a_octets_received_ok) },
+	{ "tx_multicast_phy", PPORT_802_3_OFF(a_multicast_frames_xmitted_ok) },
+	{ "tx_broadcast_phy", PPORT_802_3_OFF(a_broadcast_frames_xmitted_ok) },
+	{ "rx_multicast_phy", PPORT_802_3_OFF(a_multicast_frames_received_ok) },
+	{ "rx_broadcast_phy", PPORT_802_3_OFF(a_broadcast_frames_received_ok) },
+	{ "rx_in_range_len_errors_phy", PPORT_802_3_OFF(a_in_range_length_errors) },
+	{ "rx_out_of_range_len_phy", PPORT_802_3_OFF(a_out_of_range_length_field) },
+	{ "rx_oversize_pkts_phy", PPORT_802_3_OFF(a_frame_too_long_errors) },
+	{ "rx_symbol_err_phy", PPORT_802_3_OFF(a_symbol_error_during_carrier) },
+	{ "tx_mac_control_phy", PPORT_802_3_OFF(a_mac_control_frames_transmitted) },
+	{ "rx_mac_control_phy", PPORT_802_3_OFF(a_mac_control_frames_received) },
+	{ "rx_unsupported_op_phy", PPORT_802_3_OFF(a_unsupported_opcodes_received) },
+	{ "rx_pause_ctrl_phy", PPORT_802_3_OFF(a_pause_mac_ctrl_frames_received) },
+	{ "tx_pause_ctrl_phy", PPORT_802_3_OFF(a_pause_mac_ctrl_frames_transmitted) },
+};
+
+#define NUM_PPORT_802_3_COUNTERS	ARRAY_SIZE(pport_802_3_stats_desc)
+
+static MLX5E_DECLARE_STATS_GRP_OP_NUM_STATS(802_3)
+{
+	return NUM_PPORT_802_3_COUNTERS;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_FILL_STRS(802_3)
+{
+	int i;
+
+	for (i = 0; i < NUM_PPORT_802_3_COUNTERS; i++)
+		strcpy(data + (idx++) * ETH_GSTRING_LEN, pport_802_3_stats_desc[i].format);
+	return idx;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_FILL_STATS(802_3)
+{
+	int i;
+
+	for (i = 0; i < NUM_PPORT_802_3_COUNTERS; i++)
+		data[idx++] = MLX5E_READ_CTR64_BE(&priv->stats.pport.IEEE_802_3_counters,
+						  pport_802_3_stats_desc, i);
+	return idx;
+}
+
+#define MLX5_BASIC_PPCNT_SUPPORTED(mdev) \
+	(MLX5_CAP_GEN(mdev, pcam_reg) ? MLX5_CAP_PCAM_REG(mdev, ppcnt) : 1)
+
+static MLX5E_DECLARE_STATS_GRP_OP_UPDATE_STATS(802_3)
+{
+	struct mlx5e_pport_stats *pstats = &priv->stats.pport;
+	struct mlx5_core_dev *mdev = priv->mdev;
+	u32 in[MLX5_ST_SZ_DW(ppcnt_reg)] = {0};
+	int sz = MLX5_ST_SZ_BYTES(ppcnt_reg);
+	void *out;
+
+	if (!MLX5_BASIC_PPCNT_SUPPORTED(mdev))
+		return;
+
+	MLX5_SET(ppcnt_reg, in, local_port, 1);
+	out = pstats->IEEE_802_3_counters;
+	MLX5_SET(ppcnt_reg, in, grp, MLX5_IEEE_802_3_COUNTERS_GROUP);
+	mlx5_core_access_reg(mdev, in, sz, out, sz, MLX5_REG_PPCNT, 0, 0);
+}
+
+#define MLX5E_READ_CTR64_BE_F(ptr, set, c)		\
+	be64_to_cpu(*(__be64 *)((char *)ptr +		\
+		MLX5_BYTE_OFF(ppcnt_reg,		\
+			      counter_set.set.c##_high)))
+
+static int mlx5e_stats_get_ieee(struct mlx5_core_dev *mdev,
+				u32 *ppcnt_ieee_802_3)
+{
+	u32 in[MLX5_ST_SZ_DW(ppcnt_reg)] = {};
+	int sz = MLX5_ST_SZ_BYTES(ppcnt_reg);
+
+	if (!MLX5_BASIC_PPCNT_SUPPORTED(mdev))
+		return -EOPNOTSUPP;
+
+	MLX5_SET(ppcnt_reg, in, local_port, 1);
+	MLX5_SET(ppcnt_reg, in, grp, MLX5_IEEE_802_3_COUNTERS_GROUP);
+	return mlx5_core_access_reg(mdev, in, sz, ppcnt_ieee_802_3,
+				    sz, MLX5_REG_PPCNT, 0, 0);
+}
+
+void mlx5e_stats_pause_get(struct mlx5e_priv *priv,
+			   struct ethtool_pause_stats *pause_stats)
+{
+	u32 ppcnt_ieee_802_3[MLX5_ST_SZ_DW(ppcnt_reg)];
+	struct mlx5_core_dev *mdev = priv->mdev;
+
+	if (mlx5e_stats_get_ieee(mdev, ppcnt_ieee_802_3))
+		return;
+
+	pause_stats->tx_pause_frames =
+		MLX5E_READ_CTR64_BE_F(ppcnt_ieee_802_3,
+				      eth_802_3_cntrs_grp_data_layout,
+				      a_pause_mac_ctrl_frames_transmitted);
+	pause_stats->rx_pause_frames =
+		MLX5E_READ_CTR64_BE_F(ppcnt_ieee_802_3,
+				      eth_802_3_cntrs_grp_data_layout,
+				      a_pause_mac_ctrl_frames_received);
+}
+
+void mlx5e_stats_eth_phy_get(struct mlx5e_priv *priv,
+			     struct ethtool_eth_phy_stats *phy_stats)
+{
+	u32 ppcnt_ieee_802_3[MLX5_ST_SZ_DW(ppcnt_reg)];
+	struct mlx5_core_dev *mdev = priv->mdev;
+
+	if (mlx5e_stats_get_ieee(mdev, ppcnt_ieee_802_3))
+		return;
+
+	phy_stats->SymbolErrorDuringCarrier =
+		MLX5E_READ_CTR64_BE_F(ppcnt_ieee_802_3,
+				      eth_802_3_cntrs_grp_data_layout,
+				      a_symbol_error_during_carrier);
+}
+
+void mlx5e_stats_eth_mac_get(struct mlx5e_priv *priv,
+			     struct ethtool_eth_mac_stats *mac_stats)
+{
+	u32 ppcnt_ieee_802_3[MLX5_ST_SZ_DW(ppcnt_reg)];
+	struct mlx5_core_dev *mdev = priv->mdev;
+
+	if (mlx5e_stats_get_ieee(mdev, ppcnt_ieee_802_3))
+		return;
+
+#define RD(name)							\
+	MLX5E_READ_CTR64_BE_F(ppcnt_ieee_802_3,				\
+			      eth_802_3_cntrs_grp_data_layout,		\
+			      name)
+
+	mac_stats->FramesTransmittedOK	= RD(a_frames_transmitted_ok);
+	mac_stats->FramesReceivedOK	= RD(a_frames_received_ok);
+	mac_stats->FrameCheckSequenceErrors = RD(a_frame_check_sequence_errors);
+	mac_stats->OctetsTransmittedOK	= RD(a_octets_transmitted_ok);
+	mac_stats->OctetsReceivedOK	= RD(a_octets_received_ok);
+	mac_stats->MulticastFramesXmittedOK = RD(a_multicast_frames_xmitted_ok);
+	mac_stats->BroadcastFramesXmittedOK = RD(a_broadcast_frames_xmitted_ok);
+	mac_stats->MulticastFramesReceivedOK = RD(a_multicast_frames_received_ok);
+	mac_stats->BroadcastFramesReceivedOK = RD(a_broadcast_frames_received_ok);
+	mac_stats->InRangeLengthErrors	= RD(a_in_range_length_errors);
+	mac_stats->OutOfRangeLengthField = RD(a_out_of_range_length_field);
+	mac_stats->FrameTooLongErrors	= RD(a_frame_too_long_errors);
+#undef RD
+}
+
+void mlx5e_stats_eth_ctrl_get(struct mlx5e_priv *priv,
+			      struct ethtool_eth_ctrl_stats *ctrl_stats)
+{
+	u32 ppcnt_ieee_802_3[MLX5_ST_SZ_DW(ppcnt_reg)];
+	struct mlx5_core_dev *mdev = priv->mdev;
+
+	if (mlx5e_stats_get_ieee(mdev, ppcnt_ieee_802_3))
+		return;
+
+	ctrl_stats->MACControlFramesTransmitted =
+		MLX5E_READ_CTR64_BE_F(ppcnt_ieee_802_3,
+				      eth_802_3_cntrs_grp_data_layout,
+				      a_mac_control_frames_transmitted);
+	ctrl_stats->MACControlFramesReceived =
+		MLX5E_READ_CTR64_BE_F(ppcnt_ieee_802_3,
+				      eth_802_3_cntrs_grp_data_layout,
+				      a_mac_control_frames_received);
+	ctrl_stats->UnsupportedOpcodesReceived =
+		MLX5E_READ_CTR64_BE_F(ppcnt_ieee_802_3,
+				      eth_802_3_cntrs_grp_data_layout,
+				      a_unsupported_opcodes_received);
+}
+
+#define PPORT_2863_OFF(c) \
+	MLX5_BYTE_OFF(ppcnt_reg, \
+		      counter_set.eth_2863_cntrs_grp_data_layout.c##_high)
+static const struct counter_desc pport_2863_stats_desc[] = {
+	{ "rx_discards_phy", PPORT_2863_OFF(if_in_discards) },
+	{ "tx_discards_phy", PPORT_2863_OFF(if_out_discards) },
+	{ "tx_errors_phy", PPORT_2863_OFF(if_out_errors) },
+};
+
+#define NUM_PPORT_2863_COUNTERS		ARRAY_SIZE(pport_2863_stats_desc)
+
+static MLX5E_DECLARE_STATS_GRP_OP_NUM_STATS(2863)
+{
+	return NUM_PPORT_2863_COUNTERS;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_FILL_STRS(2863)
+{
+	int i;
+
+	for (i = 0; i < NUM_PPORT_2863_COUNTERS; i++)
+		strcpy(data + (idx++) * ETH_GSTRING_LEN, pport_2863_stats_desc[i].format);
+	return idx;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_FILL_STATS(2863)
+{
+	int i;
+
+	for (i = 0; i < NUM_PPORT_2863_COUNTERS; i++)
+		data[idx++] = MLX5E_READ_CTR64_BE(&priv->stats.pport.RFC_2863_counters,
+						  pport_2863_stats_desc, i);
+	return idx;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_UPDATE_STATS(2863)
+{
+	struct mlx5e_pport_stats *pstats = &priv->stats.pport;
+	struct mlx5_core_dev *mdev = priv->mdev;
+	u32 in[MLX5_ST_SZ_DW(ppcnt_reg)] = {0};
+	int sz = MLX5_ST_SZ_BYTES(ppcnt_reg);
+	void *out;
+
+	MLX5_SET(ppcnt_reg, in, local_port, 1);
+	out = pstats->RFC_2863_counters;
+	MLX5_SET(ppcnt_reg, in, grp, MLX5_RFC_2863_COUNTERS_GROUP);
+	mlx5_core_access_reg(mdev, in, sz, out, sz, MLX5_REG_PPCNT, 0, 0);
+}
+
+#define PPORT_2819_OFF(c) \
+	MLX5_BYTE_OFF(ppcnt_reg, \
+		      counter_set.eth_2819_cntrs_grp_data_layout.c##_high)
+static const struct counter_desc pport_2819_stats_desc[] = {
+	{ "rx_undersize_pkts_phy", PPORT_2819_OFF(ether_stats_undersize_pkts) },
+	{ "rx_fragments_phy", PPORT_2819_OFF(ether_stats_fragments) },
+	{ "rx_jabbers_phy", PPORT_2819_OFF(ether_stats_jabbers) },
+	{ "rx_64_bytes_phy", PPORT_2819_OFF(ether_stats_pkts64octets) },
+	{ "rx_65_to_127_bytes_phy", PPORT_2819_OFF(ether_stats_pkts65to127octets) },
+	{ "rx_128_to_255_bytes_phy", PPORT_2819_OFF(ether_stats_pkts128to255octets) },
+	{ "rx_256_to_511_bytes_phy", PPORT_2819_OFF(ether_stats_pkts256to511octets) },
+	{ "rx_512_to_1023_bytes_phy", PPORT_2819_OFF(ether_stats_pkts512to1023octets) },
+	{ "rx_1024_to_1518_bytes_phy", PPORT_2819_OFF(ether_stats_pkts1024to1518octets) },
+	{ "rx_1519_to_2047_bytes_phy", PPORT_2819_OFF(ether_stats_pkts1519to2047octets) },
+	{ "rx_2048_to_4095_bytes_phy", PPORT_2819_OFF(ether_stats_pkts2048to4095octets) },
+	{ "rx_4096_to_8191_bytes_phy", PPORT_2819_OFF(ether_stats_pkts4096to8191octets) },
+	{ "rx_8192_to_10239_bytes_phy", PPORT_2819_OFF(ether_stats_pkts8192to10239octets) },
+};
+
+#define NUM_PPORT_2819_COUNTERS		ARRAY_SIZE(pport_2819_stats_desc)
+
+static MLX5E_DECLARE_STATS_GRP_OP_NUM_STATS(2819)
+{
+	return NUM_PPORT_2819_COUNTERS;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_FILL_STRS(2819)
+{
+	int i;
+
+	for (i = 0; i < NUM_PPORT_2819_COUNTERS; i++)
+		strcpy(data + (idx++) * ETH_GSTRING_LEN, pport_2819_stats_desc[i].format);
+	return idx;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_FILL_STATS(2819)
+{
+	int i;
+
+	for (i = 0; i < NUM_PPORT_2819_COUNTERS; i++)
+		data[idx++] = MLX5E_READ_CTR64_BE(&priv->stats.pport.RFC_2819_counters,
+						  pport_2819_stats_desc, i);
+	return idx;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_UPDATE_STATS(2819)
+{
+	struct mlx5e_pport_stats *pstats = &priv->stats.pport;
+	struct mlx5_core_dev *mdev = priv->mdev;
+	u32 in[MLX5_ST_SZ_DW(ppcnt_reg)] = {0};
+	int sz = MLX5_ST_SZ_BYTES(ppcnt_reg);
+	void *out;
+
+	if (!MLX5_BASIC_PPCNT_SUPPORTED(mdev))
+		return;
+
+	MLX5_SET(ppcnt_reg, in, local_port, 1);
+	out = pstats->RFC_2819_counters;
+	MLX5_SET(ppcnt_reg, in, grp, MLX5_RFC_2819_COUNTERS_GROUP);
+	mlx5_core_access_reg(mdev, in, sz, out, sz, MLX5_REG_PPCNT, 0, 0);
+}
+
+static const struct ethtool_rmon_hist_range mlx5e_rmon_ranges[] = {
+	{    0,    64 },
+	{   65,   127 },
+	{  128,   255 },
+	{  256,   511 },
+	{  512,  1023 },
+	{ 1024,  1518 },
+	{ 1519,  2047 },
+	{ 2048,  4095 },
+	{ 4096,  8191 },
+	{ 8192, 10239 },
+	{}
+};
+
+void mlx5e_stats_rmon_get(struct mlx5e_priv *priv,
+			  struct ethtool_rmon_stats *rmon,
+			  const struct ethtool_rmon_hist_range **ranges)
+{
+	u32 ppcnt_RFC_2819_counters[MLX5_ST_SZ_DW(ppcnt_reg)];
+	struct mlx5_core_dev *mdev = priv->mdev;
+	u32 in[MLX5_ST_SZ_DW(ppcnt_reg)] = {0};
+	int sz = MLX5_ST_SZ_BYTES(ppcnt_reg);
+
+	MLX5_SET(ppcnt_reg, in, local_port, 1);
+	MLX5_SET(ppcnt_reg, in, grp, MLX5_RFC_2819_COUNTERS_GROUP);
+	if (mlx5_core_access_reg(mdev, in, sz, ppcnt_RFC_2819_counters,
+				 sz, MLX5_REG_PPCNT, 0, 0))
+		return;
+
+#define RD(name)						\
+	MLX5E_READ_CTR64_BE_F(ppcnt_RFC_2819_counters,		\
+			      eth_2819_cntrs_grp_data_layout,	\
+			      name)
+
+	rmon->undersize_pkts	= RD(ether_stats_undersize_pkts);
+	rmon->fragments		= RD(ether_stats_fragments);
+	rmon->jabbers		= RD(ether_stats_jabbers);
+
+	rmon->hist[0]		= RD(ether_stats_pkts64octets);
+	rmon->hist[1]		= RD(ether_stats_pkts65to127octets);
+	rmon->hist[2]		= RD(ether_stats_pkts128to255octets);
+	rmon->hist[3]		= RD(ether_stats_pkts256to511octets);
+	rmon->hist[4]		= RD(ether_stats_pkts512to1023octets);
+	rmon->hist[5]		= RD(ether_stats_pkts1024to1518octets);
+	rmon->hist[6]		= RD(ether_stats_pkts1519to2047octets);
+	rmon->hist[7]		= RD(ether_stats_pkts2048to4095octets);
+	rmon->hist[8]		= RD(ether_stats_pkts4096to8191octets);
+	rmon->hist[9]		= RD(ether_stats_pkts8192to10239octets);
+#undef RD
+
+	*ranges = mlx5e_rmon_ranges;
+}
+
+#define PPORT_PHY_STATISTICAL_OFF(c) \
+	MLX5_BYTE_OFF(ppcnt_reg, \
+		      counter_set.phys_layer_statistical_cntrs.c##_high)
+static const struct counter_desc pport_phy_statistical_stats_desc[] = {
+	{ "rx_pcs_symbol_err_phy", PPORT_PHY_STATISTICAL_OFF(phy_symbol_errors) },
+	{ "rx_corrected_bits_phy", PPORT_PHY_STATISTICAL_OFF(phy_corrected_bits) },
+};
+
+static const struct counter_desc
+pport_phy_statistical_err_lanes_stats_desc[] = {
+	{ "rx_err_lane_0_phy", PPORT_PHY_STATISTICAL_OFF(phy_corrected_bits_lane0) },
+	{ "rx_err_lane_1_phy", PPORT_PHY_STATISTICAL_OFF(phy_corrected_bits_lane1) },
+	{ "rx_err_lane_2_phy", PPORT_PHY_STATISTICAL_OFF(phy_corrected_bits_lane2) },
+	{ "rx_err_lane_3_phy", PPORT_PHY_STATISTICAL_OFF(phy_corrected_bits_lane3) },
+};
+
+#define NUM_PPORT_PHY_STATISTICAL_COUNTERS \
+	ARRAY_SIZE(pport_phy_statistical_stats_desc)
+#define NUM_PPORT_PHY_STATISTICAL_PER_LANE_COUNTERS \
+	ARRAY_SIZE(pport_phy_statistical_err_lanes_stats_desc)
+
+static MLX5E_DECLARE_STATS_GRP_OP_NUM_STATS(phy)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+	int num_stats;
+
+	/* "1" for link_down_events special counter */
+	num_stats = 1;
+
+	num_stats += MLX5_CAP_PCAM_FEATURE(mdev, ppcnt_statistical_group) ?
+		     NUM_PPORT_PHY_STATISTICAL_COUNTERS : 0;
+
+	num_stats += MLX5_CAP_PCAM_FEATURE(mdev, per_lane_error_counters) ?
+		     NUM_PPORT_PHY_STATISTICAL_PER_LANE_COUNTERS : 0;
+
+	return num_stats;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_FILL_STRS(phy)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+	int i;
+
+	strcpy(data + (idx++) * ETH_GSTRING_LEN, "link_down_events_phy");
+
+	if (!MLX5_CAP_PCAM_FEATURE(mdev, ppcnt_statistical_group))
+		return idx;
+
+	for (i = 0; i < NUM_PPORT_PHY_STATISTICAL_COUNTERS; i++)
+		strcpy(data + (idx++) * ETH_GSTRING_LEN,
+		       pport_phy_statistical_stats_desc[i].format);
+
+	if (MLX5_CAP_PCAM_FEATURE(mdev, per_lane_error_counters))
+		for (i = 0; i < NUM_PPORT_PHY_STATISTICAL_PER_LANE_COUNTERS; i++)
+			strcpy(data + (idx++) * ETH_GSTRING_LEN,
+			       pport_phy_statistical_err_lanes_stats_desc[i].format);
+
+	return idx;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_FILL_STATS(phy)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+	int i;
+
+	/* link_down_events_phy has special handling since it is not stored in __be64 format */
+	data[idx++] = MLX5_GET(ppcnt_reg, priv->stats.pport.phy_counters,
+			       counter_set.phys_layer_cntrs.link_down_events);
+
+	if (!MLX5_CAP_PCAM_FEATURE(mdev, ppcnt_statistical_group))
+		return idx;
+
+	for (i = 0; i < NUM_PPORT_PHY_STATISTICAL_COUNTERS; i++)
+		data[idx++] =
+			MLX5E_READ_CTR64_BE(&priv->stats.pport.phy_statistical_counters,
+					    pport_phy_statistical_stats_desc, i);
+
+	if (MLX5_CAP_PCAM_FEATURE(mdev, per_lane_error_counters))
+		for (i = 0; i < NUM_PPORT_PHY_STATISTICAL_PER_LANE_COUNTERS; i++)
+			data[idx++] =
+				MLX5E_READ_CTR64_BE(&priv->stats.pport.phy_statistical_counters,
+						    pport_phy_statistical_err_lanes_stats_desc,
+						    i);
+	return idx;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_UPDATE_STATS(phy)
+{
+	struct mlx5e_pport_stats *pstats = &priv->stats.pport;
+	struct mlx5_core_dev *mdev = priv->mdev;
+	u32 in[MLX5_ST_SZ_DW(ppcnt_reg)] = {0};
+	int sz = MLX5_ST_SZ_BYTES(ppcnt_reg);
+	void *out;
+
+	MLX5_SET(ppcnt_reg, in, local_port, 1);
+	out = pstats->phy_counters;
+	MLX5_SET(ppcnt_reg, in, grp, MLX5_PHYSICAL_LAYER_COUNTERS_GROUP);
+	mlx5_core_access_reg(mdev, in, sz, out, sz, MLX5_REG_PPCNT, 0, 0);
+
+	if (!MLX5_CAP_PCAM_FEATURE(mdev, ppcnt_statistical_group))
+		return;
+
+	out = pstats->phy_statistical_counters;
+	MLX5_SET(ppcnt_reg, in, grp, MLX5_PHYSICAL_LAYER_STATISTICAL_GROUP);
+	mlx5_core_access_reg(mdev, in, sz, out, sz, MLX5_REG_PPCNT, 0, 0);
+}
+
+static int fec_num_lanes(struct mlx5_core_dev *dev)
+{
+	u32 out[MLX5_ST_SZ_DW(pmlp_reg)] = {};
+	u32 in[MLX5_ST_SZ_DW(pmlp_reg)] = {};
+	int err;
+
+	MLX5_SET(pmlp_reg, in, local_port, 1);
+	err = mlx5_core_access_reg(dev, in, sizeof(in), out, sizeof(out),
+				   MLX5_REG_PMLP, 0, 0);
+	if (err)
+		return 0;
+
+	return MLX5_GET(pmlp_reg, out, width);
+}
+
+static int fec_active_mode(struct mlx5_core_dev *mdev)
+{
+	unsigned long fec_active_long;
+	u32 fec_active;
+
+	if (mlx5e_get_fec_mode(mdev, &fec_active, NULL))
+		return MLX5E_FEC_NOFEC;
+
+	fec_active_long = fec_active;
+	return find_first_bit(&fec_active_long, sizeof(unsigned long) * BITS_PER_BYTE);
+}
+
+#define MLX5E_STATS_SET_FEC_BLOCK(idx) ({ \
+	fec_stats->corrected_blocks.lanes[(idx)] = \
+		MLX5E_READ_CTR64_BE_F(ppcnt, phys_layer_cntrs, \
+				      fc_fec_corrected_blocks_lane##idx); \
+	fec_stats->uncorrectable_blocks.lanes[(idx)] = \
+		MLX5E_READ_CTR64_BE_F(ppcnt, phys_layer_cntrs, \
+				      fc_fec_uncorrectable_blocks_lane##idx); \
+})
+
+static void fec_set_fc_stats(struct ethtool_fec_stats *fec_stats,
+			     u32 *ppcnt, u8 lanes)
+{
+	if (lanes > 3) { /* 4 lanes */
+		MLX5E_STATS_SET_FEC_BLOCK(3);
+		MLX5E_STATS_SET_FEC_BLOCK(2);
+	}
+	if (lanes > 1) /* 2 lanes */
+		MLX5E_STATS_SET_FEC_BLOCK(1);
+	if (lanes > 0) /* 1 lane */
+		MLX5E_STATS_SET_FEC_BLOCK(0);
+}
+
+static void fec_set_rs_stats(struct ethtool_fec_stats *fec_stats, u32 *ppcnt)
+{
+	fec_stats->corrected_blocks.total =
+		MLX5E_READ_CTR64_BE_F(ppcnt, phys_layer_cntrs,
+				      rs_fec_corrected_blocks);
+	fec_stats->uncorrectable_blocks.total =
+		MLX5E_READ_CTR64_BE_F(ppcnt, phys_layer_cntrs,
+				      rs_fec_uncorrectable_blocks);
+}
+
+static void fec_set_block_stats(struct mlx5e_priv *priv,
+				struct ethtool_fec_stats *fec_stats)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+	u32 out[MLX5_ST_SZ_DW(ppcnt_reg)] = {};
+	u32 in[MLX5_ST_SZ_DW(ppcnt_reg)] = {};
+	int sz = MLX5_ST_SZ_BYTES(ppcnt_reg);
+	int mode = fec_active_mode(mdev);
+
+	if (mode == MLX5E_FEC_NOFEC)
+		return;
+
+	MLX5_SET(ppcnt_reg, in, local_port, 1);
+	MLX5_SET(ppcnt_reg, in, grp, MLX5_PHYSICAL_LAYER_COUNTERS_GROUP);
+	if (mlx5_core_access_reg(mdev, in, sz, out, sz, MLX5_REG_PPCNT, 0, 0))
+		return;
+
+	switch (mode) {
+	case MLX5E_FEC_RS_528_514:
+	case MLX5E_FEC_RS_544_514:
+	case MLX5E_FEC_LLRS_272_257_1:
+		fec_set_rs_stats(fec_stats, out);
+		return;
+	case MLX5E_FEC_FIRECODE:
+		fec_set_fc_stats(fec_stats, out, fec_num_lanes(mdev));
+	}
+}
+
+static void fec_set_corrected_bits_total(struct mlx5e_priv *priv,
+					 struct ethtool_fec_stats *fec_stats)
+{
+	u32 ppcnt_phy_statistical[MLX5_ST_SZ_DW(ppcnt_reg)];
+	struct mlx5_core_dev *mdev = priv->mdev;
+	u32 in[MLX5_ST_SZ_DW(ppcnt_reg)] = {};
+	int sz = MLX5_ST_SZ_BYTES(ppcnt_reg);
+
+	MLX5_SET(ppcnt_reg, in, local_port, 1);
+	MLX5_SET(ppcnt_reg, in, grp, MLX5_PHYSICAL_LAYER_STATISTICAL_GROUP);
+	if (mlx5_core_access_reg(mdev, in, sz, ppcnt_phy_statistical,
+				 sz, MLX5_REG_PPCNT, 0, 0))
+		return;
+
+	fec_stats->corrected_bits.total =
+		MLX5E_READ_CTR64_BE_F(ppcnt_phy_statistical,
+				      phys_layer_statistical_cntrs,
+				      phy_corrected_bits);
+}
+
+void mlx5e_stats_fec_get(struct mlx5e_priv *priv,
+			 struct ethtool_fec_stats *fec_stats)
+{
+	if (!MLX5_CAP_PCAM_FEATURE(priv->mdev, ppcnt_statistical_group))
+		return;
+
+	fec_set_corrected_bits_total(priv, fec_stats);
+	fec_set_block_stats(priv, fec_stats);
+}
+
+#define PPORT_ETH_EXT_OFF(c) \
+	MLX5_BYTE_OFF(ppcnt_reg, \
+		      counter_set.eth_extended_cntrs_grp_data_layout.c##_high)
+static const struct counter_desc pport_eth_ext_stats_desc[] = {
+	{ "rx_buffer_passed_thres_phy", PPORT_ETH_EXT_OFF(rx_buffer_almost_full) },
+};
+
+#define NUM_PPORT_ETH_EXT_COUNTERS	ARRAY_SIZE(pport_eth_ext_stats_desc)
+
+static MLX5E_DECLARE_STATS_GRP_OP_NUM_STATS(eth_ext)
+{
+	if (MLX5_CAP_PCAM_FEATURE((priv)->mdev, rx_buffer_fullness_counters))
+		return NUM_PPORT_ETH_EXT_COUNTERS;
+
+	return 0;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_FILL_STRS(eth_ext)
+{
+	int i;
+
+	if (MLX5_CAP_PCAM_FEATURE((priv)->mdev, rx_buffer_fullness_counters))
+		for (i = 0; i < NUM_PPORT_ETH_EXT_COUNTERS; i++)
+			strcpy(data + (idx++) * ETH_GSTRING_LEN,
+			       pport_eth_ext_stats_desc[i].format);
+	return idx;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_FILL_STATS(eth_ext)
+{
+	int i;
+
+	if (MLX5_CAP_PCAM_FEATURE((priv)->mdev, rx_buffer_fullness_counters))
+		for (i = 0; i < NUM_PPORT_ETH_EXT_COUNTERS; i++)
+			data[idx++] =
+				MLX5E_READ_CTR64_BE(&priv->stats.pport.eth_ext_counters,
+						    pport_eth_ext_stats_desc, i);
+	return idx;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_UPDATE_STATS(eth_ext)
+{
+	struct mlx5e_pport_stats *pstats = &priv->stats.pport;
+	struct mlx5_core_dev *mdev = priv->mdev;
+	u32 in[MLX5_ST_SZ_DW(ppcnt_reg)] = {0};
+	int sz = MLX5_ST_SZ_BYTES(ppcnt_reg);
+	void *out;
+
+	if (!MLX5_CAP_PCAM_FEATURE(mdev, rx_buffer_fullness_counters))
+		return;
+
+	MLX5_SET(ppcnt_reg, in, local_port, 1);
+	out = pstats->eth_ext_counters;
+	MLX5_SET(ppcnt_reg, in, grp, MLX5_ETHERNET_EXTENDED_COUNTERS_GROUP);
+	mlx5_core_access_reg(mdev, in, sz, out, sz, MLX5_REG_PPCNT, 0, 0);
+}
+
+#define PCIE_PERF_OFF(c) \
+	MLX5_BYTE_OFF(mpcnt_reg, counter_set.pcie_perf_cntrs_grp_data_layout.c)
+static const struct counter_desc pcie_perf_stats_desc[] = {
+	{ "rx_pci_signal_integrity", PCIE_PERF_OFF(rx_errors) },
+	{ "tx_pci_signal_integrity", PCIE_PERF_OFF(tx_errors) },
+};
+
+#define PCIE_PERF_OFF64(c) \
+	MLX5_BYTE_OFF(mpcnt_reg, counter_set.pcie_perf_cntrs_grp_data_layout.c##_high)
+static const struct counter_desc pcie_perf_stats_desc64[] = {
+	{ "outbound_pci_buffer_overflow", PCIE_PERF_OFF64(tx_overflow_buffer_pkt) },
+};
+
+static const struct counter_desc pcie_perf_stall_stats_desc[] = {
+	{ "outbound_pci_stalled_rd", PCIE_PERF_OFF(outbound_stalled_reads) },
+	{ "outbound_pci_stalled_wr", PCIE_PERF_OFF(outbound_stalled_writes) },
+	{ "outbound_pci_stalled_rd_events", PCIE_PERF_OFF(outbound_stalled_reads_events) },
+	{ "outbound_pci_stalled_wr_events", PCIE_PERF_OFF(outbound_stalled_writes_events) },
+};
+
+#define NUM_PCIE_PERF_COUNTERS		ARRAY_SIZE(pcie_perf_stats_desc)
+#define NUM_PCIE_PERF_COUNTERS64	ARRAY_SIZE(pcie_perf_stats_desc64)
+#define NUM_PCIE_PERF_STALL_COUNTERS	ARRAY_SIZE(pcie_perf_stall_stats_desc)
+
+static MLX5E_DECLARE_STATS_GRP_OP_NUM_STATS(pcie)
+{
+	int num_stats = 0;
+
+	if (MLX5_CAP_MCAM_FEATURE((priv)->mdev, pcie_performance_group))
+		num_stats += NUM_PCIE_PERF_COUNTERS;
+
+	if (MLX5_CAP_MCAM_FEATURE((priv)->mdev, tx_overflow_buffer_pkt))
+		num_stats += NUM_PCIE_PERF_COUNTERS64;
+
+	if (MLX5_CAP_MCAM_FEATURE((priv)->mdev, pcie_outbound_stalled))
+		num_stats += NUM_PCIE_PERF_STALL_COUNTERS;
+
+	return num_stats;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_FILL_STRS(pcie)
+{
+	int i;
+
+	if (MLX5_CAP_MCAM_FEATURE((priv)->mdev, pcie_performance_group))
+		for (i = 0; i < NUM_PCIE_PERF_COUNTERS; i++)
+			strcpy(data + (idx++) * ETH_GSTRING_LEN,
+			       pcie_perf_stats_desc[i].format);
+
+	if (MLX5_CAP_MCAM_FEATURE((priv)->mdev, tx_overflow_buffer_pkt))
+		for (i = 0; i < NUM_PCIE_PERF_COUNTERS64; i++)
+			strcpy(data + (idx++) * ETH_GSTRING_LEN,
+			       pcie_perf_stats_desc64[i].format);
+
+	if (MLX5_CAP_MCAM_FEATURE((priv)->mdev, pcie_outbound_stalled))
+		for (i = 0; i < NUM_PCIE_PERF_STALL_COUNTERS; i++)
+			strcpy(data + (idx++) * ETH_GSTRING_LEN,
+			       pcie_perf_stall_stats_desc[i].format);
+	return idx;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_FILL_STATS(pcie)
+{
+	int i;
+
+	if (MLX5_CAP_MCAM_FEATURE((priv)->mdev, pcie_performance_group))
+		for (i = 0; i < NUM_PCIE_PERF_COUNTERS; i++)
+			data[idx++] =
+				MLX5E_READ_CTR32_BE(&priv->stats.pcie.pcie_perf_counters,
+						    pcie_perf_stats_desc, i);
+
+	if (MLX5_CAP_MCAM_FEATURE((priv)->mdev, tx_overflow_buffer_pkt))
+		for (i = 0; i < NUM_PCIE_PERF_COUNTERS64; i++)
+			data[idx++] =
+				MLX5E_READ_CTR64_BE(&priv->stats.pcie.pcie_perf_counters,
+						    pcie_perf_stats_desc64, i);
+
+	if (MLX5_CAP_MCAM_FEATURE((priv)->mdev, pcie_outbound_stalled))
+		for (i = 0; i < NUM_PCIE_PERF_STALL_COUNTERS; i++)
+			data[idx++] =
+				MLX5E_READ_CTR32_BE(&priv->stats.pcie.pcie_perf_counters,
+						    pcie_perf_stall_stats_desc, i);
+	return idx;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_UPDATE_STATS(pcie)
+{
+	struct mlx5e_pcie_stats *pcie_stats = &priv->stats.pcie;
+	struct mlx5_core_dev *mdev = priv->mdev;
+	u32 in[MLX5_ST_SZ_DW(mpcnt_reg)] = {0};
+	int sz = MLX5_ST_SZ_BYTES(mpcnt_reg);
+	void *out;
+
+	if (!MLX5_CAP_MCAM_FEATURE(mdev, pcie_performance_group))
+		return;
+
+	out = pcie_stats->pcie_perf_counters;
+	MLX5_SET(mpcnt_reg, in, grp, MLX5_PCIE_PERFORMANCE_COUNTERS_GROUP);
+	mlx5_core_access_reg(mdev, in, sz, out, sz, MLX5_REG_MPCNT, 0, 0);
+}
+
+#define PPORT_PER_TC_PRIO_OFF(c) \
+	MLX5_BYTE_OFF(ppcnt_reg, \
+		      counter_set.eth_per_tc_prio_grp_data_layout.c##_high)
+
+static const struct counter_desc pport_per_tc_prio_stats_desc[] = {
+	{ "rx_prio%d_buf_discard", PPORT_PER_TC_PRIO_OFF(no_buffer_discard_uc) },
+};
+
+#define NUM_PPORT_PER_TC_PRIO_COUNTERS	ARRAY_SIZE(pport_per_tc_prio_stats_desc)
+
+#define PPORT_PER_TC_CONGEST_PRIO_OFF(c) \
+	MLX5_BYTE_OFF(ppcnt_reg, \
+		      counter_set.eth_per_tc_congest_prio_grp_data_layout.c##_high)
+
+static const struct counter_desc pport_per_tc_congest_prio_stats_desc[] = {
+	{ "rx_prio%d_cong_discard", PPORT_PER_TC_CONGEST_PRIO_OFF(wred_discard) },
+	{ "rx_prio%d_marked", PPORT_PER_TC_CONGEST_PRIO_OFF(ecn_marked_tc) },
+};
+
+#define NUM_PPORT_PER_TC_CONGEST_PRIO_COUNTERS \
+	ARRAY_SIZE(pport_per_tc_congest_prio_stats_desc)
+
+static int mlx5e_grp_per_tc_prio_get_num_stats(struct mlx5e_priv *priv)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+
+	if (!MLX5_CAP_GEN(mdev, sbcam_reg))
+		return 0;
+
+	return NUM_PPORT_PER_TC_PRIO_COUNTERS * NUM_PPORT_PRIO;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_FILL_STRS(per_port_buff_congest)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+	int i, prio;
+
+	if (!MLX5_CAP_GEN(mdev, sbcam_reg))
+		return idx;
+
+	for (prio = 0; prio < NUM_PPORT_PRIO; prio++) {
+		for (i = 0; i < NUM_PPORT_PER_TC_PRIO_COUNTERS; i++)
+			sprintf(data + (idx++) * ETH_GSTRING_LEN,
+				pport_per_tc_prio_stats_desc[i].format, prio);
+		for (i = 0; i < NUM_PPORT_PER_TC_CONGEST_PRIO_COUNTERS; i++)
+			sprintf(data + (idx++) * ETH_GSTRING_LEN,
+				pport_per_tc_congest_prio_stats_desc[i].format, prio);
+	}
+
+	return idx;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_FILL_STATS(per_port_buff_congest)
+{
+	struct mlx5e_pport_stats *pport = &priv->stats.pport;
+	struct mlx5_core_dev *mdev = priv->mdev;
+	int i, prio;
+
+	if (!MLX5_CAP_GEN(mdev, sbcam_reg))
+		return idx;
+
+	for (prio = 0; prio < NUM_PPORT_PRIO; prio++) {
+		for (i = 0; i < NUM_PPORT_PER_TC_PRIO_COUNTERS; i++)
+			data[idx++] =
+				MLX5E_READ_CTR64_BE(&pport->per_tc_prio_counters[prio],
+						    pport_per_tc_prio_stats_desc, i);
+		for (i = 0; i < NUM_PPORT_PER_TC_CONGEST_PRIO_COUNTERS ; i++)
+			data[idx++] =
+				MLX5E_READ_CTR64_BE(&pport->per_tc_congest_prio_counters[prio],
+						    pport_per_tc_congest_prio_stats_desc, i);
+	}
+
+	return idx;
+}
+
+static void mlx5e_grp_per_tc_prio_update_stats(struct mlx5e_priv *priv)
+{
+	struct mlx5e_pport_stats *pstats = &priv->stats.pport;
+	struct mlx5_core_dev *mdev = priv->mdev;
+	u32 in[MLX5_ST_SZ_DW(ppcnt_reg)] = {};
+	int sz = MLX5_ST_SZ_BYTES(ppcnt_reg);
+	void *out;
+	int prio;
+
+	if (!MLX5_CAP_GEN(mdev, sbcam_reg))
+		return;
+
+	MLX5_SET(ppcnt_reg, in, pnat, 2);
+	MLX5_SET(ppcnt_reg, in, grp, MLX5_PER_TRAFFIC_CLASS_COUNTERS_GROUP);
+	for (prio = 0; prio < NUM_PPORT_PRIO; prio++) {
+		out = pstats->per_tc_prio_counters[prio];
+		MLX5_SET(ppcnt_reg, in, prio_tc, prio);
+		mlx5_core_access_reg(mdev, in, sz, out, sz, MLX5_REG_PPCNT, 0, 0);
+	}
+}
+
+static int mlx5e_grp_per_tc_congest_prio_get_num_stats(struct mlx5e_priv *priv)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+
+	if (!MLX5_CAP_GEN(mdev, sbcam_reg))
+		return 0;
+
+	return NUM_PPORT_PER_TC_CONGEST_PRIO_COUNTERS * NUM_PPORT_PRIO;
+}
+
+static void mlx5e_grp_per_tc_congest_prio_update_stats(struct mlx5e_priv *priv)
+{
+	struct mlx5e_pport_stats *pstats = &priv->stats.pport;
+	struct mlx5_core_dev *mdev = priv->mdev;
+	u32 in[MLX5_ST_SZ_DW(ppcnt_reg)] = {};
+	int sz = MLX5_ST_SZ_BYTES(ppcnt_reg);
+	void *out;
+	int prio;
+
+	if (!MLX5_CAP_GEN(mdev, sbcam_reg))
+		return;
+
+	MLX5_SET(ppcnt_reg, in, pnat, 2);
+	MLX5_SET(ppcnt_reg, in, grp, MLX5_PER_TRAFFIC_CLASS_CONGESTION_GROUP);
+	for (prio = 0; prio < NUM_PPORT_PRIO; prio++) {
+		out = pstats->per_tc_congest_prio_counters[prio];
+		MLX5_SET(ppcnt_reg, in, prio_tc, prio);
+		mlx5_core_access_reg(mdev, in, sz, out, sz, MLX5_REG_PPCNT, 0, 0);
+	}
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_NUM_STATS(per_port_buff_congest)
+{
+	return mlx5e_grp_per_tc_prio_get_num_stats(priv) +
+		mlx5e_grp_per_tc_congest_prio_get_num_stats(priv);
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_UPDATE_STATS(per_port_buff_congest)
+{
+	mlx5e_grp_per_tc_prio_update_stats(priv);
+	mlx5e_grp_per_tc_congest_prio_update_stats(priv);
+}
+
+#define PPORT_PER_PRIO_OFF(c) \
+	MLX5_BYTE_OFF(ppcnt_reg, \
+		      counter_set.eth_per_prio_grp_data_layout.c##_high)
+static const struct counter_desc pport_per_prio_traffic_stats_desc[] = {
+	{ "rx_prio%d_bytes", PPORT_PER_PRIO_OFF(rx_octets) },
+	{ "rx_prio%d_packets", PPORT_PER_PRIO_OFF(rx_frames) },
+	{ "rx_prio%d_discards", PPORT_PER_PRIO_OFF(rx_discards) },
+	{ "tx_prio%d_bytes", PPORT_PER_PRIO_OFF(tx_octets) },
+	{ "tx_prio%d_packets", PPORT_PER_PRIO_OFF(tx_frames) },
+};
+
+#define NUM_PPORT_PER_PRIO_TRAFFIC_COUNTERS	ARRAY_SIZE(pport_per_prio_traffic_stats_desc)
+
+static int mlx5e_grp_per_prio_traffic_get_num_stats(void)
+{
+	return NUM_PPORT_PER_PRIO_TRAFFIC_COUNTERS * NUM_PPORT_PRIO;
+}
+
+static int mlx5e_grp_per_prio_traffic_fill_strings(struct mlx5e_priv *priv,
+						   u8 *data,
+						   int idx)
+{
+	int i, prio;
+
+	for (prio = 0; prio < NUM_PPORT_PRIO; prio++) {
+		for (i = 0; i < NUM_PPORT_PER_PRIO_TRAFFIC_COUNTERS; i++)
+			sprintf(data + (idx++) * ETH_GSTRING_LEN,
+				pport_per_prio_traffic_stats_desc[i].format, prio);
+	}
+
+	return idx;
+}
+
+static int mlx5e_grp_per_prio_traffic_fill_stats(struct mlx5e_priv *priv,
+						 u64 *data,
+						 int idx)
+{
+	int i, prio;
+
+	for (prio = 0; prio < NUM_PPORT_PRIO; prio++) {
+		for (i = 0; i < NUM_PPORT_PER_PRIO_TRAFFIC_COUNTERS; i++)
+			data[idx++] =
+				MLX5E_READ_CTR64_BE(&priv->stats.pport.per_prio_counters[prio],
+						    pport_per_prio_traffic_stats_desc, i);
+	}
+
+	return idx;
+}
+
+static const struct counter_desc pport_per_prio_pfc_stats_desc[] = {
+	/* %s is "global" or "prio{i}" */
+	{ "rx_%s_pause", PPORT_PER_PRIO_OFF(rx_pause) },
+	{ "rx_%s_pause_duration", PPORT_PER_PRIO_OFF(rx_pause_duration) },
+	{ "tx_%s_pause", PPORT_PER_PRIO_OFF(tx_pause) },
+	{ "tx_%s_pause_duration", PPORT_PER_PRIO_OFF(tx_pause_duration) },
+	{ "rx_%s_pause_transition", PPORT_PER_PRIO_OFF(rx_pause_transition) },
+};
+
+static const struct counter_desc pport_pfc_stall_stats_desc[] = {
+	{ "tx_pause_storm_warning_events", PPORT_PER_PRIO_OFF(device_stall_minor_watermark_cnt) },
+	{ "tx_pause_storm_error_events", PPORT_PER_PRIO_OFF(device_stall_critical_watermark_cnt) },
+};
+
+#define NUM_PPORT_PER_PRIO_PFC_COUNTERS		ARRAY_SIZE(pport_per_prio_pfc_stats_desc)
+#define NUM_PPORT_PFC_STALL_COUNTERS(priv)	(ARRAY_SIZE(pport_pfc_stall_stats_desc) * \
+						 MLX5_CAP_PCAM_FEATURE((priv)->mdev, pfcc_mask) * \
+						 MLX5_CAP_DEBUG((priv)->mdev, stall_detect))
+
+static unsigned long mlx5e_query_pfc_combined(struct mlx5e_priv *priv)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+	u8 pfc_en_tx;
+	u8 pfc_en_rx;
+	int err;
+
+	if (MLX5_CAP_GEN(mdev, port_type) != MLX5_CAP_PORT_TYPE_ETH)
+		return 0;
+
+	err = mlx5_query_port_pfc(mdev, &pfc_en_tx, &pfc_en_rx);
+
+	return err ? 0 : pfc_en_tx | pfc_en_rx;
+}
+
+static bool mlx5e_query_global_pause_combined(struct mlx5e_priv *priv)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+	u32 rx_pause;
+	u32 tx_pause;
+	int err;
+
+	if (MLX5_CAP_GEN(mdev, port_type) != MLX5_CAP_PORT_TYPE_ETH)
+		return false;
+
+	err = mlx5_query_port_pause(mdev, &rx_pause, &tx_pause);
+
+	return err ? false : rx_pause | tx_pause;
+}
+
+static int mlx5e_grp_per_prio_pfc_get_num_stats(struct mlx5e_priv *priv)
+{
+	return (mlx5e_query_global_pause_combined(priv) +
+		hweight8(mlx5e_query_pfc_combined(priv))) *
+		NUM_PPORT_PER_PRIO_PFC_COUNTERS +
+		NUM_PPORT_PFC_STALL_COUNTERS(priv);
+}
+
+static int mlx5e_grp_per_prio_pfc_fill_strings(struct mlx5e_priv *priv,
+					       u8 *data,
+					       int idx)
+{
+	unsigned long pfc_combined;
+	int i, prio;
+
+	pfc_combined = mlx5e_query_pfc_combined(priv);
+	for_each_set_bit(prio, &pfc_combined, NUM_PPORT_PRIO) {
+		for (i = 0; i < NUM_PPORT_PER_PRIO_PFC_COUNTERS; i++) {
+			char pfc_string[ETH_GSTRING_LEN];
+
+			snprintf(pfc_string, sizeof(pfc_string), "prio%d", prio);
+			sprintf(data + (idx++) * ETH_GSTRING_LEN,
+				pport_per_prio_pfc_stats_desc[i].format, pfc_string);
+		}
+	}
+
+	if (mlx5e_query_global_pause_combined(priv)) {
+		for (i = 0; i < NUM_PPORT_PER_PRIO_PFC_COUNTERS; i++) {
+			sprintf(data + (idx++) * ETH_GSTRING_LEN,
+				pport_per_prio_pfc_stats_desc[i].format, "global");
+		}
+	}
+
+	for (i = 0; i < NUM_PPORT_PFC_STALL_COUNTERS(priv); i++)
+		strcpy(data + (idx++) * ETH_GSTRING_LEN,
+		       pport_pfc_stall_stats_desc[i].format);
+
+	return idx;
+}
+
+static int mlx5e_grp_per_prio_pfc_fill_stats(struct mlx5e_priv *priv,
+					     u64 *data,
+					     int idx)
+{
+	unsigned long pfc_combined;
+	int i, prio;
+
+	pfc_combined = mlx5e_query_pfc_combined(priv);
+	for_each_set_bit(prio, &pfc_combined, NUM_PPORT_PRIO) {
+		for (i = 0; i < NUM_PPORT_PER_PRIO_PFC_COUNTERS; i++) {
+			data[idx++] =
+				MLX5E_READ_CTR64_BE(&priv->stats.pport.per_prio_counters[prio],
+						    pport_per_prio_pfc_stats_desc, i);
+		}
+	}
+
+	if (mlx5e_query_global_pause_combined(priv)) {
+		for (i = 0; i < NUM_PPORT_PER_PRIO_PFC_COUNTERS; i++) {
+			data[idx++] =
+				MLX5E_READ_CTR64_BE(&priv->stats.pport.per_prio_counters[0],
+						    pport_per_prio_pfc_stats_desc, i);
+		}
+	}
+
+	for (i = 0; i < NUM_PPORT_PFC_STALL_COUNTERS(priv); i++)
+		data[idx++] = MLX5E_READ_CTR64_BE(&priv->stats.pport.per_prio_counters[0],
+						  pport_pfc_stall_stats_desc, i);
+
+	return idx;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_NUM_STATS(per_prio)
+{
+	return mlx5e_grp_per_prio_traffic_get_num_stats() +
+		mlx5e_grp_per_prio_pfc_get_num_stats(priv);
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_FILL_STRS(per_prio)
+{
+	idx = mlx5e_grp_per_prio_traffic_fill_strings(priv, data, idx);
+	idx = mlx5e_grp_per_prio_pfc_fill_strings(priv, data, idx);
+	return idx;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_FILL_STATS(per_prio)
+{
+	idx = mlx5e_grp_per_prio_traffic_fill_stats(priv, data, idx);
+	idx = mlx5e_grp_per_prio_pfc_fill_stats(priv, data, idx);
+	return idx;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_UPDATE_STATS(per_prio)
+{
+	struct mlx5e_pport_stats *pstats = &priv->stats.pport;
+	struct mlx5_core_dev *mdev = priv->mdev;
+	u32 in[MLX5_ST_SZ_DW(ppcnt_reg)] = {0};
+	int sz = MLX5_ST_SZ_BYTES(ppcnt_reg);
+	int prio;
+	void *out;
+
+	if (!MLX5_BASIC_PPCNT_SUPPORTED(mdev))
+		return;
+
+	MLX5_SET(ppcnt_reg, in, local_port, 1);
+	MLX5_SET(ppcnt_reg, in, grp, MLX5_PER_PRIORITY_COUNTERS_GROUP);
+	for (prio = 0; prio < NUM_PPORT_PRIO; prio++) {
+		out = pstats->per_prio_counters[prio];
+		MLX5_SET(ppcnt_reg, in, prio_tc, prio);
+		mlx5_core_access_reg(mdev, in, sz, out, sz,
+				     MLX5_REG_PPCNT, 0, 0);
+	}
+}
+
+static const struct counter_desc mlx5e_pme_status_desc[] = {
+	{ "module_unplug",       sizeof(u64) * MLX5_MODULE_STATUS_UNPLUGGED },
+};
+
+static const struct counter_desc mlx5e_pme_error_desc[] = {
+	{ "module_bus_stuck",    sizeof(u64) * MLX5_MODULE_EVENT_ERROR_BUS_STUCK },
+	{ "module_high_temp",    sizeof(u64) * MLX5_MODULE_EVENT_ERROR_HIGH_TEMPERATURE },
+	{ "module_bad_shorted",  sizeof(u64) * MLX5_MODULE_EVENT_ERROR_BAD_CABLE },
+};
+
+#define NUM_PME_STATUS_STATS		ARRAY_SIZE(mlx5e_pme_status_desc)
+#define NUM_PME_ERR_STATS		ARRAY_SIZE(mlx5e_pme_error_desc)
+
+static MLX5E_DECLARE_STATS_GRP_OP_NUM_STATS(pme)
+{
+	return NUM_PME_STATUS_STATS + NUM_PME_ERR_STATS;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_FILL_STRS(pme)
+{
+	int i;
+
+	for (i = 0; i < NUM_PME_STATUS_STATS; i++)
+		strcpy(data + (idx++) * ETH_GSTRING_LEN, mlx5e_pme_status_desc[i].format);
+
+	for (i = 0; i < NUM_PME_ERR_STATS; i++)
+		strcpy(data + (idx++) * ETH_GSTRING_LEN, mlx5e_pme_error_desc[i].format);
+
+	return idx;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_FILL_STATS(pme)
+{
+	struct mlx5_pme_stats pme_stats;
+	int i;
+
+	mlx5_get_pme_stats(priv->mdev, &pme_stats);
+
+	for (i = 0; i < NUM_PME_STATUS_STATS; i++)
+		data[idx++] = MLX5E_READ_CTR64_CPU(pme_stats.status_counters,
+						   mlx5e_pme_status_desc, i);
+
+	for (i = 0; i < NUM_PME_ERR_STATS; i++)
+		data[idx++] = MLX5E_READ_CTR64_CPU(pme_stats.error_counters,
+						   mlx5e_pme_error_desc, i);
+
+	return idx;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_UPDATE_STATS(pme) { return; }
+
+static MLX5E_DECLARE_STATS_GRP_OP_NUM_STATS(tls)
+{
+	return mlx5e_ktls_get_count(priv);
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_FILL_STRS(tls)
+{
+	return idx + mlx5e_ktls_get_strings(priv, data + idx * ETH_GSTRING_LEN);
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_FILL_STATS(tls)
+{
+	return idx + mlx5e_ktls_get_stats(priv, data + idx);
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_UPDATE_STATS(tls) { return; }
+
+static const struct counter_desc rq_stats_desc[] = {
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, packets) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, bytes) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, csum_complete) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, csum_complete_tail) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, csum_complete_tail_slow) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, csum_unnecessary) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, csum_unnecessary_inner) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, csum_none) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, xdp_drop) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, xdp_redirect) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, lro_packets) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, lro_bytes) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, gro_packets) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, gro_bytes) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, gro_skbs) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, gro_match_packets) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, gro_large_hds) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, ecn_mark) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, removed_vlan_packets) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, wqe_err) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, mpwqe_filler_cqes) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, mpwqe_filler_strides) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, oversize_pkts_sw_drop) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, buff_alloc_err) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, cqe_compress_blks) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, cqe_compress_pkts) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, cache_reuse) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, cache_full) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, cache_empty) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, cache_busy) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, cache_waive) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, congst_umr) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, arfs_err) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, recover) },
+#ifdef CONFIG_PAGE_POOL_STATS
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, pp_alloc_fast) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, pp_alloc_slow) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, pp_alloc_slow_high_order) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, pp_alloc_empty) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, pp_alloc_refill) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, pp_alloc_waive) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, pp_recycle_cached) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, pp_recycle_cache_full) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, pp_recycle_ring) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, pp_recycle_ring_full) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, pp_recycle_released_ref) },
+#endif
+#ifdef CONFIG_MLX5_EN_TLS
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, tls_decrypted_packets) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, tls_decrypted_bytes) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, tls_resync_req_pkt) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, tls_resync_req_start) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, tls_resync_req_end) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, tls_resync_req_skip) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, tls_resync_res_ok) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, tls_resync_res_retry) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, tls_resync_res_skip) },
+	{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats, tls_err) },
+#endif
+};
+
+static const struct counter_desc sq_stats_desc[] = {
+	{ MLX5E_DECLARE_TX_STAT(struct mlx5e_sq_stats, packets) },
+	{ MLX5E_DECLARE_TX_STAT(struct mlx5e_sq_stats, bytes) },
+	{ MLX5E_DECLARE_TX_STAT(struct mlx5e_sq_stats, tso_packets) },
+	{ MLX5E_DECLARE_TX_STAT(struct mlx5e_sq_stats, tso_bytes) },
+	{ MLX5E_DECLARE_TX_STAT(struct mlx5e_sq_stats, tso_inner_packets) },
+	{ MLX5E_DECLARE_TX_STAT(struct mlx5e_sq_stats, tso_inner_bytes) },
+	{ MLX5E_DECLARE_TX_STAT(struct mlx5e_sq_stats, csum_partial) },
+	{ MLX5E_DECLARE_TX_STAT(struct mlx5e_sq_stats, csum_partial_inner) },
+	{ MLX5E_DECLARE_TX_STAT(struct mlx5e_sq_stats, added_vlan_packets) },
+	{ MLX5E_DECLARE_TX_STAT(struct mlx5e_sq_stats, nop) },
+	{ MLX5E_DECLARE_TX_STAT(struct mlx5e_sq_stats, mpwqe_blks) },
+	{ MLX5E_DECLARE_TX_STAT(struct mlx5e_sq_stats, mpwqe_pkts) },
+#ifdef CONFIG_MLX5_EN_TLS
+	{ MLX5E_DECLARE_TX_STAT(struct mlx5e_sq_stats, tls_encrypted_packets) },
+	{ MLX5E_DECLARE_TX_STAT(struct mlx5e_sq_stats, tls_encrypted_bytes) },
+	{ MLX5E_DECLARE_TX_STAT(struct mlx5e_sq_stats, tls_ooo) },
+	{ MLX5E_DECLARE_TX_STAT(struct mlx5e_sq_stats, tls_dump_packets) },
+	{ MLX5E_DECLARE_TX_STAT(struct mlx5e_sq_stats, tls_dump_bytes) },
+	{ MLX5E_DECLARE_TX_STAT(struct mlx5e_sq_stats, tls_resync_bytes) },
+	{ MLX5E_DECLARE_TX_STAT(struct mlx5e_sq_stats, tls_skip_no_sync_data) },
+	{ MLX5E_DECLARE_TX_STAT(struct mlx5e_sq_stats, tls_drop_no_sync_data) },
+	{ MLX5E_DECLARE_TX_STAT(struct mlx5e_sq_stats, tls_drop_bypass_req) },
+#endif
+	{ MLX5E_DECLARE_TX_STAT(struct mlx5e_sq_stats, csum_none) },
+	{ MLX5E_DECLARE_TX_STAT(struct mlx5e_sq_stats, stopped) },
+	{ MLX5E_DECLARE_TX_STAT(struct mlx5e_sq_stats, dropped) },
+	{ MLX5E_DECLARE_TX_STAT(struct mlx5e_sq_stats, xmit_more) },
+	{ MLX5E_DECLARE_TX_STAT(struct mlx5e_sq_stats, recover) },
+	{ MLX5E_DECLARE_TX_STAT(struct mlx5e_sq_stats, cqes) },
+	{ MLX5E_DECLARE_TX_STAT(struct mlx5e_sq_stats, wake) },
+	{ MLX5E_DECLARE_TX_STAT(struct mlx5e_sq_stats, cqe_err) },
+};
+
+static const struct counter_desc rq_xdpsq_stats_desc[] = {
+	{ MLX5E_DECLARE_RQ_XDPSQ_STAT(struct mlx5e_xdpsq_stats, xmit) },
+	{ MLX5E_DECLARE_RQ_XDPSQ_STAT(struct mlx5e_xdpsq_stats, mpwqe) },
+	{ MLX5E_DECLARE_RQ_XDPSQ_STAT(struct mlx5e_xdpsq_stats, inlnw) },
+	{ MLX5E_DECLARE_RQ_XDPSQ_STAT(struct mlx5e_xdpsq_stats, nops) },
+	{ MLX5E_DECLARE_RQ_XDPSQ_STAT(struct mlx5e_xdpsq_stats, full) },
+	{ MLX5E_DECLARE_RQ_XDPSQ_STAT(struct mlx5e_xdpsq_stats, err) },
+	{ MLX5E_DECLARE_RQ_XDPSQ_STAT(struct mlx5e_xdpsq_stats, cqes) },
+};
+
+static const struct counter_desc xdpsq_stats_desc[] = {
+	{ MLX5E_DECLARE_XDPSQ_STAT(struct mlx5e_xdpsq_stats, xmit) },
+	{ MLX5E_DECLARE_XDPSQ_STAT(struct mlx5e_xdpsq_stats, mpwqe) },
+	{ MLX5E_DECLARE_XDPSQ_STAT(struct mlx5e_xdpsq_stats, inlnw) },
+	{ MLX5E_DECLARE_XDPSQ_STAT(struct mlx5e_xdpsq_stats, nops) },
+	{ MLX5E_DECLARE_XDPSQ_STAT(struct mlx5e_xdpsq_stats, full) },
+	{ MLX5E_DECLARE_XDPSQ_STAT(struct mlx5e_xdpsq_stats, err) },
+	{ MLX5E_DECLARE_XDPSQ_STAT(struct mlx5e_xdpsq_stats, cqes) },
+};
+
+static const struct counter_desc xskrq_stats_desc[] = {
+	{ MLX5E_DECLARE_XSKRQ_STAT(struct mlx5e_rq_stats, packets) },
+	{ MLX5E_DECLARE_XSKRQ_STAT(struct mlx5e_rq_stats, bytes) },
+	{ MLX5E_DECLARE_XSKRQ_STAT(struct mlx5e_rq_stats, csum_complete) },
+	{ MLX5E_DECLARE_XSKRQ_STAT(struct mlx5e_rq_stats, csum_unnecessary) },
+	{ MLX5E_DECLARE_XSKRQ_STAT(struct mlx5e_rq_stats, csum_unnecessary_inner) },
+	{ MLX5E_DECLARE_XSKRQ_STAT(struct mlx5e_rq_stats, csum_none) },
+	{ MLX5E_DECLARE_XSKRQ_STAT(struct mlx5e_rq_stats, ecn_mark) },
+	{ MLX5E_DECLARE_XSKRQ_STAT(struct mlx5e_rq_stats, removed_vlan_packets) },
+	{ MLX5E_DECLARE_XSKRQ_STAT(struct mlx5e_rq_stats, xdp_drop) },
+	{ MLX5E_DECLARE_XSKRQ_STAT(struct mlx5e_rq_stats, xdp_redirect) },
+	{ MLX5E_DECLARE_XSKRQ_STAT(struct mlx5e_rq_stats, wqe_err) },
+	{ MLX5E_DECLARE_XSKRQ_STAT(struct mlx5e_rq_stats, mpwqe_filler_cqes) },
+	{ MLX5E_DECLARE_XSKRQ_STAT(struct mlx5e_rq_stats, mpwqe_filler_strides) },
+	{ MLX5E_DECLARE_XSKRQ_STAT(struct mlx5e_rq_stats, oversize_pkts_sw_drop) },
+	{ MLX5E_DECLARE_XSKRQ_STAT(struct mlx5e_rq_stats, buff_alloc_err) },
+	{ MLX5E_DECLARE_XSKRQ_STAT(struct mlx5e_rq_stats, cqe_compress_blks) },
+	{ MLX5E_DECLARE_XSKRQ_STAT(struct mlx5e_rq_stats, cqe_compress_pkts) },
+	{ MLX5E_DECLARE_XSKRQ_STAT(struct mlx5e_rq_stats, congst_umr) },
+	{ MLX5E_DECLARE_XSKRQ_STAT(struct mlx5e_rq_stats, arfs_err) },
+};
+
+static const struct counter_desc xsksq_stats_desc[] = {
+	{ MLX5E_DECLARE_XSKSQ_STAT(struct mlx5e_xdpsq_stats, xmit) },
+	{ MLX5E_DECLARE_XSKSQ_STAT(struct mlx5e_xdpsq_stats, mpwqe) },
+	{ MLX5E_DECLARE_XSKSQ_STAT(struct mlx5e_xdpsq_stats, inlnw) },
+	{ MLX5E_DECLARE_XSKSQ_STAT(struct mlx5e_xdpsq_stats, full) },
+	{ MLX5E_DECLARE_XSKSQ_STAT(struct mlx5e_xdpsq_stats, err) },
+	{ MLX5E_DECLARE_XSKSQ_STAT(struct mlx5e_xdpsq_stats, cqes) },
+};
+
+static const struct counter_desc ch_stats_desc[] = {
+	{ MLX5E_DECLARE_CH_STAT(struct mlx5e_ch_stats, events) },
+	{ MLX5E_DECLARE_CH_STAT(struct mlx5e_ch_stats, poll) },
+	{ MLX5E_DECLARE_CH_STAT(struct mlx5e_ch_stats, arm) },
+	{ MLX5E_DECLARE_CH_STAT(struct mlx5e_ch_stats, aff_change) },
+	{ MLX5E_DECLARE_CH_STAT(struct mlx5e_ch_stats, force_irq) },
+	{ MLX5E_DECLARE_CH_STAT(struct mlx5e_ch_stats, eq_rearm) },
+};
+
+static const struct counter_desc ptp_sq_stats_desc[] = {
+	{ MLX5E_DECLARE_PTP_TX_STAT(struct mlx5e_sq_stats, packets) },
+	{ MLX5E_DECLARE_PTP_TX_STAT(struct mlx5e_sq_stats, bytes) },
+	{ MLX5E_DECLARE_PTP_TX_STAT(struct mlx5e_sq_stats, csum_partial) },
+	{ MLX5E_DECLARE_PTP_TX_STAT(struct mlx5e_sq_stats, csum_partial_inner) },
+	{ MLX5E_DECLARE_PTP_TX_STAT(struct mlx5e_sq_stats, added_vlan_packets) },
+	{ MLX5E_DECLARE_PTP_TX_STAT(struct mlx5e_sq_stats, nop) },
+	{ MLX5E_DECLARE_PTP_TX_STAT(struct mlx5e_sq_stats, csum_none) },
+	{ MLX5E_DECLARE_PTP_TX_STAT(struct mlx5e_sq_stats, stopped) },
+	{ MLX5E_DECLARE_PTP_TX_STAT(struct mlx5e_sq_stats, dropped) },
+	{ MLX5E_DECLARE_PTP_TX_STAT(struct mlx5e_sq_stats, xmit_more) },
+	{ MLX5E_DECLARE_PTP_TX_STAT(struct mlx5e_sq_stats, recover) },
+	{ MLX5E_DECLARE_PTP_TX_STAT(struct mlx5e_sq_stats, cqes) },
+	{ MLX5E_DECLARE_PTP_TX_STAT(struct mlx5e_sq_stats, wake) },
+	{ MLX5E_DECLARE_PTP_TX_STAT(struct mlx5e_sq_stats, cqe_err) },
+};
+
+static const struct counter_desc ptp_ch_stats_desc[] = {
+	{ MLX5E_DECLARE_PTP_CH_STAT(struct mlx5e_ch_stats, events) },
+	{ MLX5E_DECLARE_PTP_CH_STAT(struct mlx5e_ch_stats, poll) },
+	{ MLX5E_DECLARE_PTP_CH_STAT(struct mlx5e_ch_stats, arm) },
+	{ MLX5E_DECLARE_PTP_CH_STAT(struct mlx5e_ch_stats, eq_rearm) },
+};
+
+static const struct counter_desc ptp_cq_stats_desc[] = {
+	{ MLX5E_DECLARE_PTP_CQ_STAT(struct mlx5e_ptp_cq_stats, cqe) },
+	{ MLX5E_DECLARE_PTP_CQ_STAT(struct mlx5e_ptp_cq_stats, err_cqe) },
+	{ MLX5E_DECLARE_PTP_CQ_STAT(struct mlx5e_ptp_cq_stats, abort) },
+	{ MLX5E_DECLARE_PTP_CQ_STAT(struct mlx5e_ptp_cq_stats, abort_abs_diff_ns) },
+	{ MLX5E_DECLARE_PTP_CQ_STAT(struct mlx5e_ptp_cq_stats, resync_cqe) },
+	{ MLX5E_DECLARE_PTP_CQ_STAT(struct mlx5e_ptp_cq_stats, resync_event) },
+	{ MLX5E_DECLARE_PTP_CQ_STAT(struct mlx5e_ptp_cq_stats, ooo_cqe_drop) },
+};
+
+static const struct counter_desc ptp_rq_stats_desc[] = {
+	{ MLX5E_DECLARE_PTP_RQ_STAT(struct mlx5e_rq_stats, packets) },
+	{ MLX5E_DECLARE_PTP_RQ_STAT(struct mlx5e_rq_stats, bytes) },
+	{ MLX5E_DECLARE_PTP_RQ_STAT(struct mlx5e_rq_stats, csum_complete) },
+	{ MLX5E_DECLARE_PTP_RQ_STAT(struct mlx5e_rq_stats, csum_complete_tail) },
+	{ MLX5E_DECLARE_PTP_RQ_STAT(struct mlx5e_rq_stats, csum_complete_tail_slow) },
+	{ MLX5E_DECLARE_PTP_RQ_STAT(struct mlx5e_rq_stats, csum_unnecessary) },
+	{ MLX5E_DECLARE_PTP_RQ_STAT(struct mlx5e_rq_stats, csum_unnecessary_inner) },
+	{ MLX5E_DECLARE_PTP_RQ_STAT(struct mlx5e_rq_stats, csum_none) },
+	{ MLX5E_DECLARE_PTP_RQ_STAT(struct mlx5e_rq_stats, xdp_drop) },
+	{ MLX5E_DECLARE_PTP_RQ_STAT(struct mlx5e_rq_stats, xdp_redirect) },
+	{ MLX5E_DECLARE_PTP_RQ_STAT(struct mlx5e_rq_stats, lro_packets) },
+	{ MLX5E_DECLARE_PTP_RQ_STAT(struct mlx5e_rq_stats, lro_bytes) },
+	{ MLX5E_DECLARE_PTP_RQ_STAT(struct mlx5e_rq_stats, ecn_mark) },
+	{ MLX5E_DECLARE_PTP_RQ_STAT(struct mlx5e_rq_stats, removed_vlan_packets) },
+	{ MLX5E_DECLARE_PTP_RQ_STAT(struct mlx5e_rq_stats, wqe_err) },
+	{ MLX5E_DECLARE_PTP_RQ_STAT(struct mlx5e_rq_stats, mpwqe_filler_cqes) },
+	{ MLX5E_DECLARE_PTP_RQ_STAT(struct mlx5e_rq_stats, mpwqe_filler_strides) },
+	{ MLX5E_DECLARE_PTP_RQ_STAT(struct mlx5e_rq_stats, oversize_pkts_sw_drop) },
+	{ MLX5E_DECLARE_PTP_RQ_STAT(struct mlx5e_rq_stats, buff_alloc_err) },
+	{ MLX5E_DECLARE_PTP_RQ_STAT(struct mlx5e_rq_stats, cqe_compress_blks) },
+	{ MLX5E_DECLARE_PTP_RQ_STAT(struct mlx5e_rq_stats, cqe_compress_pkts) },
+	{ MLX5E_DECLARE_PTP_RQ_STAT(struct mlx5e_rq_stats, cache_reuse) },
+	{ MLX5E_DECLARE_PTP_RQ_STAT(struct mlx5e_rq_stats, cache_full) },
+	{ MLX5E_DECLARE_PTP_RQ_STAT(struct mlx5e_rq_stats, cache_empty) },
+	{ MLX5E_DECLARE_PTP_RQ_STAT(struct mlx5e_rq_stats, cache_busy) },
+	{ MLX5E_DECLARE_PTP_RQ_STAT(struct mlx5e_rq_stats, cache_waive) },
+	{ MLX5E_DECLARE_PTP_RQ_STAT(struct mlx5e_rq_stats, congst_umr) },
+	{ MLX5E_DECLARE_PTP_RQ_STAT(struct mlx5e_rq_stats, arfs_err) },
+	{ MLX5E_DECLARE_PTP_RQ_STAT(struct mlx5e_rq_stats, recover) },
+};
+
+static const struct counter_desc qos_sq_stats_desc[] = {
+	{ MLX5E_DECLARE_QOS_TX_STAT(struct mlx5e_sq_stats, packets) },
+	{ MLX5E_DECLARE_QOS_TX_STAT(struct mlx5e_sq_stats, bytes) },
+	{ MLX5E_DECLARE_QOS_TX_STAT(struct mlx5e_sq_stats, tso_packets) },
+	{ MLX5E_DECLARE_QOS_TX_STAT(struct mlx5e_sq_stats, tso_bytes) },
+	{ MLX5E_DECLARE_QOS_TX_STAT(struct mlx5e_sq_stats, tso_inner_packets) },
+	{ MLX5E_DECLARE_QOS_TX_STAT(struct mlx5e_sq_stats, tso_inner_bytes) },
+	{ MLX5E_DECLARE_QOS_TX_STAT(struct mlx5e_sq_stats, csum_partial) },
+	{ MLX5E_DECLARE_QOS_TX_STAT(struct mlx5e_sq_stats, csum_partial_inner) },
+	{ MLX5E_DECLARE_QOS_TX_STAT(struct mlx5e_sq_stats, added_vlan_packets) },
+	{ MLX5E_DECLARE_QOS_TX_STAT(struct mlx5e_sq_stats, nop) },
+	{ MLX5E_DECLARE_QOS_TX_STAT(struct mlx5e_sq_stats, mpwqe_blks) },
+	{ MLX5E_DECLARE_QOS_TX_STAT(struct mlx5e_sq_stats, mpwqe_pkts) },
+#ifdef CONFIG_MLX5_EN_TLS
+	{ MLX5E_DECLARE_QOS_TX_STAT(struct mlx5e_sq_stats, tls_encrypted_packets) },
+	{ MLX5E_DECLARE_QOS_TX_STAT(struct mlx5e_sq_stats, tls_encrypted_bytes) },
+	{ MLX5E_DECLARE_QOS_TX_STAT(struct mlx5e_sq_stats, tls_ooo) },
+	{ MLX5E_DECLARE_QOS_TX_STAT(struct mlx5e_sq_stats, tls_dump_packets) },
+	{ MLX5E_DECLARE_QOS_TX_STAT(struct mlx5e_sq_stats, tls_dump_bytes) },
+	{ MLX5E_DECLARE_QOS_TX_STAT(struct mlx5e_sq_stats, tls_resync_bytes) },
+	{ MLX5E_DECLARE_QOS_TX_STAT(struct mlx5e_sq_stats, tls_skip_no_sync_data) },
+	{ MLX5E_DECLARE_QOS_TX_STAT(struct mlx5e_sq_stats, tls_drop_no_sync_data) },
+	{ MLX5E_DECLARE_QOS_TX_STAT(struct mlx5e_sq_stats, tls_drop_bypass_req) },
+#endif
+	{ MLX5E_DECLARE_QOS_TX_STAT(struct mlx5e_sq_stats, csum_none) },
+	{ MLX5E_DECLARE_QOS_TX_STAT(struct mlx5e_sq_stats, stopped) },
+	{ MLX5E_DECLARE_QOS_TX_STAT(struct mlx5e_sq_stats, dropped) },
+	{ MLX5E_DECLARE_QOS_TX_STAT(struct mlx5e_sq_stats, xmit_more) },
+	{ MLX5E_DECLARE_QOS_TX_STAT(struct mlx5e_sq_stats, recover) },
+	{ MLX5E_DECLARE_QOS_TX_STAT(struct mlx5e_sq_stats, cqes) },
+	{ MLX5E_DECLARE_QOS_TX_STAT(struct mlx5e_sq_stats, wake) },
+	{ MLX5E_DECLARE_QOS_TX_STAT(struct mlx5e_sq_stats, cqe_err) },
+};
+
+#define NUM_RQ_STATS			ARRAY_SIZE(rq_stats_desc)
+#define NUM_SQ_STATS			ARRAY_SIZE(sq_stats_desc)
+#define NUM_XDPSQ_STATS			ARRAY_SIZE(xdpsq_stats_desc)
+#define NUM_RQ_XDPSQ_STATS		ARRAY_SIZE(rq_xdpsq_stats_desc)
+#define NUM_XSKRQ_STATS			ARRAY_SIZE(xskrq_stats_desc)
+#define NUM_XSKSQ_STATS			ARRAY_SIZE(xsksq_stats_desc)
+#define NUM_CH_STATS			ARRAY_SIZE(ch_stats_desc)
+#define NUM_PTP_SQ_STATS		ARRAY_SIZE(ptp_sq_stats_desc)
+#define NUM_PTP_CH_STATS		ARRAY_SIZE(ptp_ch_stats_desc)
+#define NUM_PTP_CQ_STATS		ARRAY_SIZE(ptp_cq_stats_desc)
+#define NUM_PTP_RQ_STATS                ARRAY_SIZE(ptp_rq_stats_desc)
+#define NUM_QOS_SQ_STATS		ARRAY_SIZE(qos_sq_stats_desc)
+
+static MLX5E_DECLARE_STATS_GRP_OP_NUM_STATS(qos)
+{
+	/* Pairs with smp_store_release in mlx5e_open_qos_sq. */
+	return NUM_QOS_SQ_STATS * smp_load_acquire(&priv->htb_max_qos_sqs);
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_FILL_STRS(qos)
+{
+	/* Pairs with smp_store_release in mlx5e_open_qos_sq. */
+	u16 max_qos_sqs = smp_load_acquire(&priv->htb_max_qos_sqs);
+	int i, qid;
+
+	for (qid = 0; qid < max_qos_sqs; qid++)
+		for (i = 0; i < NUM_QOS_SQ_STATS; i++)
+			sprintf(data + (idx++) * ETH_GSTRING_LEN,
+				qos_sq_stats_desc[i].format, qid);
+
+	return idx;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_FILL_STATS(qos)
+{
+	struct mlx5e_sq_stats **stats;
+	u16 max_qos_sqs;
+	int i, qid;
+
+	/* Pairs with smp_store_release in mlx5e_open_qos_sq. */
+	max_qos_sqs = smp_load_acquire(&priv->htb_max_qos_sqs);
+	stats = READ_ONCE(priv->htb_qos_sq_stats);
+
+	for (qid = 0; qid < max_qos_sqs; qid++) {
+		struct mlx5e_sq_stats *s = READ_ONCE(stats[qid]);
+
+		for (i = 0; i < NUM_QOS_SQ_STATS; i++)
+			data[idx++] = MLX5E_READ_CTR64_CPU(s, qos_sq_stats_desc, i);
+	}
+
+	return idx;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_UPDATE_STATS(qos) { return; }
+
+static MLX5E_DECLARE_STATS_GRP_OP_NUM_STATS(ptp)
+{
+	int num = NUM_PTP_CH_STATS;
+
+	if (!priv->tx_ptp_opened && !priv->rx_ptp_opened)
+		return 0;
+
+	if (priv->tx_ptp_opened)
+		num += (NUM_PTP_SQ_STATS + NUM_PTP_CQ_STATS) * priv->max_opened_tc;
+	if (priv->rx_ptp_opened)
+		num += NUM_PTP_RQ_STATS;
+
+	return num;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_FILL_STRS(ptp)
+{
+	int i, tc;
+
+	if (!priv->tx_ptp_opened && !priv->rx_ptp_opened)
+		return idx;
+
+	for (i = 0; i < NUM_PTP_CH_STATS; i++)
+		sprintf(data + (idx++) * ETH_GSTRING_LEN,
+			"%s", ptp_ch_stats_desc[i].format);
+
+	if (priv->tx_ptp_opened) {
+		for (tc = 0; tc < priv->max_opened_tc; tc++)
+			for (i = 0; i < NUM_PTP_SQ_STATS; i++)
+				sprintf(data + (idx++) * ETH_GSTRING_LEN,
+					ptp_sq_stats_desc[i].format, tc);
+
+		for (tc = 0; tc < priv->max_opened_tc; tc++)
+			for (i = 0; i < NUM_PTP_CQ_STATS; i++)
+				sprintf(data + (idx++) * ETH_GSTRING_LEN,
+					ptp_cq_stats_desc[i].format, tc);
+	}
+	if (priv->rx_ptp_opened) {
+		for (i = 0; i < NUM_PTP_RQ_STATS; i++)
+			sprintf(data + (idx++) * ETH_GSTRING_LEN,
+				ptp_rq_stats_desc[i].format, MLX5E_PTP_CHANNEL_IX);
+	}
+	return idx;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_FILL_STATS(ptp)
+{
+	int i, tc;
+
+	if (!priv->tx_ptp_opened && !priv->rx_ptp_opened)
+		return idx;
+
+	for (i = 0; i < NUM_PTP_CH_STATS; i++)
+		data[idx++] =
+			MLX5E_READ_CTR64_CPU(&priv->ptp_stats.ch,
+					     ptp_ch_stats_desc, i);
+
+	if (priv->tx_ptp_opened) {
+		for (tc = 0; tc < priv->max_opened_tc; tc++)
+			for (i = 0; i < NUM_PTP_SQ_STATS; i++)
+				data[idx++] =
+					MLX5E_READ_CTR64_CPU(&priv->ptp_stats.sq[tc],
+							     ptp_sq_stats_desc, i);
+
+		for (tc = 0; tc < priv->max_opened_tc; tc++)
+			for (i = 0; i < NUM_PTP_CQ_STATS; i++)
+				data[idx++] =
+					MLX5E_READ_CTR64_CPU(&priv->ptp_stats.cq[tc],
+							     ptp_cq_stats_desc, i);
+	}
+	if (priv->rx_ptp_opened) {
+		for (i = 0; i < NUM_PTP_RQ_STATS; i++)
+			data[idx++] =
+				MLX5E_READ_CTR64_CPU(&priv->ptp_stats.rq,
+						     ptp_rq_stats_desc, i);
+	}
+	return idx;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_UPDATE_STATS(ptp) { return; }
+
+static MLX5E_DECLARE_STATS_GRP_OP_NUM_STATS(channels)
+{
+	int max_nch = priv->stats_nch;
+
+	return (NUM_RQ_STATS * max_nch) +
+	       (NUM_CH_STATS * max_nch) +
+	       (NUM_SQ_STATS * max_nch * priv->max_opened_tc) +
+	       (NUM_RQ_XDPSQ_STATS * max_nch) +
+	       (NUM_XDPSQ_STATS * max_nch) +
+	       (NUM_XSKRQ_STATS * max_nch * priv->xsk.ever_used) +
+	       (NUM_XSKSQ_STATS * max_nch * priv->xsk.ever_used);
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_FILL_STRS(channels)
+{
+	bool is_xsk = priv->xsk.ever_used;
+	int max_nch = priv->stats_nch;
+	int i, j, tc;
+
+	for (i = 0; i < max_nch; i++)
+		for (j = 0; j < NUM_CH_STATS; j++)
+			sprintf(data + (idx++) * ETH_GSTRING_LEN,
+				ch_stats_desc[j].format, i);
+
+	for (i = 0; i < max_nch; i++) {
+		for (j = 0; j < NUM_RQ_STATS; j++)
+			sprintf(data + (idx++) * ETH_GSTRING_LEN,
+				rq_stats_desc[j].format, i);
+		for (j = 0; j < NUM_XSKRQ_STATS * is_xsk; j++)
+			sprintf(data + (idx++) * ETH_GSTRING_LEN,
+				xskrq_stats_desc[j].format, i);
+		for (j = 0; j < NUM_RQ_XDPSQ_STATS; j++)
+			sprintf(data + (idx++) * ETH_GSTRING_LEN,
+				rq_xdpsq_stats_desc[j].format, i);
+	}
+
+	for (tc = 0; tc < priv->max_opened_tc; tc++)
+		for (i = 0; i < max_nch; i++)
+			for (j = 0; j < NUM_SQ_STATS; j++)
+				sprintf(data + (idx++) * ETH_GSTRING_LEN,
+					sq_stats_desc[j].format,
+					i + tc * max_nch);
+
+	for (i = 0; i < max_nch; i++) {
+		for (j = 0; j < NUM_XSKSQ_STATS * is_xsk; j++)
+			sprintf(data + (idx++) * ETH_GSTRING_LEN,
+				xsksq_stats_desc[j].format, i);
+		for (j = 0; j < NUM_XDPSQ_STATS; j++)
+			sprintf(data + (idx++) * ETH_GSTRING_LEN,
+				xdpsq_stats_desc[j].format, i);
+	}
+
+	return idx;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_FILL_STATS(channels)
+{
+	bool is_xsk = priv->xsk.ever_used;
+	int max_nch = priv->stats_nch;
+	int i, j, tc;
+
+	for (i = 0; i < max_nch; i++)
+		for (j = 0; j < NUM_CH_STATS; j++)
+			data[idx++] =
+				MLX5E_READ_CTR64_CPU(&priv->channel_stats[i]->ch,
+						     ch_stats_desc, j);
+
+	for (i = 0; i < max_nch; i++) {
+		for (j = 0; j < NUM_RQ_STATS; j++)
+			data[idx++] =
+				MLX5E_READ_CTR64_CPU(&priv->channel_stats[i]->rq,
+						     rq_stats_desc, j);
+		for (j = 0; j < NUM_XSKRQ_STATS * is_xsk; j++)
+			data[idx++] =
+				MLX5E_READ_CTR64_CPU(&priv->channel_stats[i]->xskrq,
+						     xskrq_stats_desc, j);
+		for (j = 0; j < NUM_RQ_XDPSQ_STATS; j++)
+			data[idx++] =
+				MLX5E_READ_CTR64_CPU(&priv->channel_stats[i]->rq_xdpsq,
+						     rq_xdpsq_stats_desc, j);
+	}
+
+	for (tc = 0; tc < priv->max_opened_tc; tc++)
+		for (i = 0; i < max_nch; i++)
+			for (j = 0; j < NUM_SQ_STATS; j++)
+				data[idx++] =
+					MLX5E_READ_CTR64_CPU(&priv->channel_stats[i]->sq[tc],
+							     sq_stats_desc, j);
+
+	for (i = 0; i < max_nch; i++) {
+		for (j = 0; j < NUM_XSKSQ_STATS * is_xsk; j++)
+			data[idx++] =
+				MLX5E_READ_CTR64_CPU(&priv->channel_stats[i]->xsksq,
+						     xsksq_stats_desc, j);
+		for (j = 0; j < NUM_XDPSQ_STATS; j++)
+			data[idx++] =
+				MLX5E_READ_CTR64_CPU(&priv->channel_stats[i]->xdpsq,
+						     xdpsq_stats_desc, j);
+	}
+
+	return idx;
+}
+
+static MLX5E_DECLARE_STATS_GRP_OP_UPDATE_STATS(channels) { return; }
+
+MLX5E_DEFINE_STATS_GRP(sw, 0);
+MLX5E_DEFINE_STATS_GRP(qcnt, MLX5E_NDO_UPDATE_STATS);
+MLX5E_DEFINE_STATS_GRP(vnic_env, 0);
+MLX5E_DEFINE_STATS_GRP(vport, MLX5E_NDO_UPDATE_STATS);
+MLX5E_DEFINE_STATS_GRP(802_3, MLX5E_NDO_UPDATE_STATS);
+MLX5E_DEFINE_STATS_GRP(2863, 0);
+MLX5E_DEFINE_STATS_GRP(2819, 0);
+MLX5E_DEFINE_STATS_GRP(phy, 0);
+MLX5E_DEFINE_STATS_GRP(pcie, 0);
+MLX5E_DEFINE_STATS_GRP(per_prio, 0);
+MLX5E_DEFINE_STATS_GRP(pme, 0);
+MLX5E_DEFINE_STATS_GRP(channels, 0);
+MLX5E_DEFINE_STATS_GRP(per_port_buff_congest, 0);
+MLX5E_DEFINE_STATS_GRP(eth_ext, 0);
+static MLX5E_DEFINE_STATS_GRP(tls, 0);
+MLX5E_DEFINE_STATS_GRP(ptp, 0);
+static MLX5E_DEFINE_STATS_GRP(qos, 0);
+
+/* The stats groups order is opposite to the update_stats() order calls */
+mlx5e_stats_grp_t mlx5e_nic_stats_grps[] = {
+	&MLX5E_STATS_GRP(sw),
+	&MLX5E_STATS_GRP(qcnt),
+	&MLX5E_STATS_GRP(vnic_env),
+	&MLX5E_STATS_GRP(vport),
+	&MLX5E_STATS_GRP(802_3),
+	&MLX5E_STATS_GRP(2863),
+	&MLX5E_STATS_GRP(2819),
+	&MLX5E_STATS_GRP(phy),
+	&MLX5E_STATS_GRP(eth_ext),
+	&MLX5E_STATS_GRP(pcie),
+	&MLX5E_STATS_GRP(per_prio),
+	&MLX5E_STATS_GRP(pme),
+#ifdef CONFIG_MLX5_EN_IPSEC
+	&MLX5E_STATS_GRP(ipsec_sw),
+#endif
+	&MLX5E_STATS_GRP(tls),
+	&MLX5E_STATS_GRP(channels),
+	&MLX5E_STATS_GRP(per_port_buff_congest),
+	&MLX5E_STATS_GRP(ptp),
+	&MLX5E_STATS_GRP(qos),
+#ifdef CONFIG_MLX5_EN_MACSEC
+	&MLX5E_STATS_GRP(macsec_hw),
+#endif
+};
+
+unsigned int mlx5e_nic_stats_grps_num(struct mlx5e_priv *priv)
+{
+	return ARRAY_SIZE(mlx5e_nic_stats_grps);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_stats.h b/drivers/net/ethernet/mellanox/mlx5/core/en_stats.h
new file mode 100644
index 000000000..52a67efaf
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_stats.h
@@ -0,0 +1,496 @@
+/*
+ * Copyright (c) 2015-2016, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef __MLX5_EN_STATS_H__
+#define __MLX5_EN_STATS_H__
+
+#define MLX5E_READ_CTR64_CPU(ptr, dsc, i) \
+	(*(u64 *)((char *)ptr + dsc[i].offset))
+#define MLX5E_READ_CTR64_BE(ptr, dsc, i) \
+	be64_to_cpu(*(__be64 *)((char *)ptr + dsc[i].offset))
+#define MLX5E_READ_CTR32_CPU(ptr, dsc, i) \
+	(*(u32 *)((char *)ptr + dsc[i].offset))
+#define MLX5E_READ_CTR32_BE(ptr, dsc, i) \
+	be32_to_cpu(*(__be32 *)((char *)ptr + dsc[i].offset))
+
+#define MLX5E_DECLARE_STAT(type, fld) #fld, offsetof(type, fld)
+#define MLX5E_DECLARE_RX_STAT(type, fld) "rx%d_"#fld, offsetof(type, fld)
+#define MLX5E_DECLARE_TX_STAT(type, fld) "tx%d_"#fld, offsetof(type, fld)
+#define MLX5E_DECLARE_XDPSQ_STAT(type, fld) "tx%d_xdp_"#fld, offsetof(type, fld)
+#define MLX5E_DECLARE_RQ_XDPSQ_STAT(type, fld) "rx%d_xdp_tx_"#fld, offsetof(type, fld)
+#define MLX5E_DECLARE_XSKRQ_STAT(type, fld) "rx%d_xsk_"#fld, offsetof(type, fld)
+#define MLX5E_DECLARE_XSKSQ_STAT(type, fld) "tx%d_xsk_"#fld, offsetof(type, fld)
+#define MLX5E_DECLARE_CH_STAT(type, fld) "ch%d_"#fld, offsetof(type, fld)
+
+#define MLX5E_DECLARE_PTP_TX_STAT(type, fld) "ptp_tx%d_"#fld, offsetof(type, fld)
+#define MLX5E_DECLARE_PTP_CH_STAT(type, fld) "ptp_ch_"#fld, offsetof(type, fld)
+#define MLX5E_DECLARE_PTP_CQ_STAT(type, fld) "ptp_cq%d_"#fld, offsetof(type, fld)
+#define MLX5E_DECLARE_PTP_RQ_STAT(type, fld) "ptp_rq%d_"#fld, offsetof(type, fld)
+
+#define MLX5E_DECLARE_QOS_TX_STAT(type, fld) "qos_tx%d_"#fld, offsetof(type, fld)
+
+struct counter_desc {
+	char		format[ETH_GSTRING_LEN];
+	size_t		offset; /* Byte offset */
+};
+
+enum {
+	MLX5E_NDO_UPDATE_STATS = BIT(0x1),
+};
+
+struct mlx5e_priv;
+struct mlx5e_stats_grp {
+	u16 update_stats_mask;
+	int (*get_num_stats)(struct mlx5e_priv *priv);
+	int (*fill_strings)(struct mlx5e_priv *priv, u8 *data, int idx);
+	int (*fill_stats)(struct mlx5e_priv *priv, u64 *data, int idx);
+	void (*update_stats)(struct mlx5e_priv *priv);
+};
+
+typedef const struct mlx5e_stats_grp *const mlx5e_stats_grp_t;
+
+#define MLX5E_STATS_GRP_OP(grp, name) mlx5e_stats_grp_ ## grp ## _ ## name
+
+#define MLX5E_DECLARE_STATS_GRP_OP_NUM_STATS(grp) \
+	int MLX5E_STATS_GRP_OP(grp, num_stats)(struct mlx5e_priv *priv)
+
+#define MLX5E_DECLARE_STATS_GRP_OP_UPDATE_STATS(grp) \
+	void MLX5E_STATS_GRP_OP(grp, update_stats)(struct mlx5e_priv *priv)
+
+#define MLX5E_DECLARE_STATS_GRP_OP_FILL_STRS(grp) \
+	int MLX5E_STATS_GRP_OP(grp, fill_strings)(struct mlx5e_priv *priv, u8 *data, int idx)
+
+#define MLX5E_DECLARE_STATS_GRP_OP_FILL_STATS(grp) \
+	int MLX5E_STATS_GRP_OP(grp, fill_stats)(struct mlx5e_priv *priv, u64 *data, int idx)
+
+#define MLX5E_STATS_GRP(grp) mlx5e_stats_grp_ ## grp
+
+#define MLX5E_DECLARE_STATS_GRP(grp) \
+	const struct mlx5e_stats_grp MLX5E_STATS_GRP(grp)
+
+#define MLX5E_DEFINE_STATS_GRP(grp, mask) \
+MLX5E_DECLARE_STATS_GRP(grp) = { \
+	.get_num_stats = MLX5E_STATS_GRP_OP(grp, num_stats), \
+	.fill_stats    = MLX5E_STATS_GRP_OP(grp, fill_stats), \
+	.fill_strings  = MLX5E_STATS_GRP_OP(grp, fill_strings), \
+	.update_stats  = MLX5E_STATS_GRP_OP(grp, update_stats), \
+	.update_stats_mask = mask, \
+}
+
+unsigned int mlx5e_stats_total_num(struct mlx5e_priv *priv);
+void mlx5e_stats_update(struct mlx5e_priv *priv);
+void mlx5e_stats_fill(struct mlx5e_priv *priv, u64 *data, int idx);
+void mlx5e_stats_fill_strings(struct mlx5e_priv *priv, u8 *data);
+void mlx5e_stats_update_ndo_stats(struct mlx5e_priv *priv);
+
+void mlx5e_stats_pause_get(struct mlx5e_priv *priv,
+			   struct ethtool_pause_stats *pause_stats);
+void mlx5e_stats_fec_get(struct mlx5e_priv *priv,
+			 struct ethtool_fec_stats *fec_stats);
+
+void mlx5e_stats_eth_phy_get(struct mlx5e_priv *priv,
+			     struct ethtool_eth_phy_stats *phy_stats);
+void mlx5e_stats_eth_mac_get(struct mlx5e_priv *priv,
+			     struct ethtool_eth_mac_stats *mac_stats);
+void mlx5e_stats_eth_ctrl_get(struct mlx5e_priv *priv,
+			      struct ethtool_eth_ctrl_stats *ctrl_stats);
+void mlx5e_stats_rmon_get(struct mlx5e_priv *priv,
+			  struct ethtool_rmon_stats *rmon,
+			  const struct ethtool_rmon_hist_range **ranges);
+
+/* Concrete NIC Stats */
+
+struct mlx5e_sw_stats {
+	u64 rx_packets;
+	u64 rx_bytes;
+	u64 tx_packets;
+	u64 tx_bytes;
+	u64 tx_tso_packets;
+	u64 tx_tso_bytes;
+	u64 tx_tso_inner_packets;
+	u64 tx_tso_inner_bytes;
+	u64 tx_added_vlan_packets;
+	u64 tx_nop;
+	u64 tx_mpwqe_blks;
+	u64 tx_mpwqe_pkts;
+	u64 rx_lro_packets;
+	u64 rx_lro_bytes;
+	u64 rx_gro_packets;
+	u64 rx_gro_bytes;
+	u64 rx_gro_skbs;
+	u64 rx_gro_match_packets;
+	u64 rx_gro_large_hds;
+	u64 rx_mcast_packets;
+	u64 rx_ecn_mark;
+	u64 rx_removed_vlan_packets;
+	u64 rx_csum_unnecessary;
+	u64 rx_csum_none;
+	u64 rx_csum_complete;
+	u64 rx_csum_complete_tail;
+	u64 rx_csum_complete_tail_slow;
+	u64 rx_csum_unnecessary_inner;
+	u64 rx_xdp_drop;
+	u64 rx_xdp_redirect;
+	u64 rx_xdp_tx_xmit;
+	u64 rx_xdp_tx_mpwqe;
+	u64 rx_xdp_tx_inlnw;
+	u64 rx_xdp_tx_nops;
+	u64 rx_xdp_tx_full;
+	u64 rx_xdp_tx_err;
+	u64 rx_xdp_tx_cqe;
+	u64 tx_csum_none;
+	u64 tx_csum_partial;
+	u64 tx_csum_partial_inner;
+	u64 tx_queue_stopped;
+	u64 tx_queue_dropped;
+	u64 tx_xmit_more;
+	u64 tx_recover;
+	u64 tx_cqes;
+	u64 tx_queue_wake;
+	u64 tx_cqe_err;
+	u64 tx_xdp_xmit;
+	u64 tx_xdp_mpwqe;
+	u64 tx_xdp_inlnw;
+	u64 tx_xdp_nops;
+	u64 tx_xdp_full;
+	u64 tx_xdp_err;
+	u64 tx_xdp_cqes;
+	u64 rx_wqe_err;
+	u64 rx_mpwqe_filler_cqes;
+	u64 rx_mpwqe_filler_strides;
+	u64 rx_oversize_pkts_sw_drop;
+	u64 rx_buff_alloc_err;
+	u64 rx_cqe_compress_blks;
+	u64 rx_cqe_compress_pkts;
+	u64 rx_cache_reuse;
+	u64 rx_cache_full;
+	u64 rx_cache_empty;
+	u64 rx_cache_busy;
+	u64 rx_cache_waive;
+	u64 rx_congst_umr;
+	u64 rx_arfs_err;
+	u64 rx_recover;
+	u64 ch_events;
+	u64 ch_poll;
+	u64 ch_arm;
+	u64 ch_aff_change;
+	u64 ch_force_irq;
+	u64 ch_eq_rearm;
+#ifdef CONFIG_PAGE_POOL_STATS
+	u64 rx_pp_alloc_fast;
+	u64 rx_pp_alloc_slow;
+	u64 rx_pp_alloc_slow_high_order;
+	u64 rx_pp_alloc_empty;
+	u64 rx_pp_alloc_refill;
+	u64 rx_pp_alloc_waive;
+	u64 rx_pp_recycle_cached;
+	u64 rx_pp_recycle_cache_full;
+	u64 rx_pp_recycle_ring;
+	u64 rx_pp_recycle_ring_full;
+	u64 rx_pp_recycle_released_ref;
+#endif
+#ifdef CONFIG_MLX5_EN_TLS
+	u64 tx_tls_encrypted_packets;
+	u64 tx_tls_encrypted_bytes;
+	u64 tx_tls_ooo;
+	u64 tx_tls_dump_packets;
+	u64 tx_tls_dump_bytes;
+	u64 tx_tls_resync_bytes;
+	u64 tx_tls_skip_no_sync_data;
+	u64 tx_tls_drop_no_sync_data;
+	u64 tx_tls_drop_bypass_req;
+
+	u64 rx_tls_decrypted_packets;
+	u64 rx_tls_decrypted_bytes;
+	u64 rx_tls_resync_req_pkt;
+	u64 rx_tls_resync_req_start;
+	u64 rx_tls_resync_req_end;
+	u64 rx_tls_resync_req_skip;
+	u64 rx_tls_resync_res_ok;
+	u64 rx_tls_resync_res_retry;
+	u64 rx_tls_resync_res_skip;
+	u64 rx_tls_err;
+#endif
+
+	u64 rx_xsk_packets;
+	u64 rx_xsk_bytes;
+	u64 rx_xsk_csum_complete;
+	u64 rx_xsk_csum_unnecessary;
+	u64 rx_xsk_csum_unnecessary_inner;
+	u64 rx_xsk_csum_none;
+	u64 rx_xsk_ecn_mark;
+	u64 rx_xsk_removed_vlan_packets;
+	u64 rx_xsk_xdp_drop;
+	u64 rx_xsk_xdp_redirect;
+	u64 rx_xsk_wqe_err;
+	u64 rx_xsk_mpwqe_filler_cqes;
+	u64 rx_xsk_mpwqe_filler_strides;
+	u64 rx_xsk_oversize_pkts_sw_drop;
+	u64 rx_xsk_buff_alloc_err;
+	u64 rx_xsk_cqe_compress_blks;
+	u64 rx_xsk_cqe_compress_pkts;
+	u64 rx_xsk_congst_umr;
+	u64 rx_xsk_arfs_err;
+	u64 tx_xsk_xmit;
+	u64 tx_xsk_mpwqe;
+	u64 tx_xsk_inlnw;
+	u64 tx_xsk_full;
+	u64 tx_xsk_err;
+	u64 tx_xsk_cqes;
+};
+
+struct mlx5e_qcounter_stats {
+	u32 rx_out_of_buffer;
+	u32 rx_if_down_packets;
+};
+
+#define VNIC_ENV_GET(vnic_env_stats, c) \
+	MLX5_GET(query_vnic_env_out, (vnic_env_stats)->query_vnic_env_out, \
+		 vport_env.c)
+
+struct mlx5e_vnic_env_stats {
+	__be64 query_vnic_env_out[MLX5_ST_SZ_QW(query_vnic_env_out)];
+};
+
+#define VPORT_COUNTER_GET(vstats, c) MLX5_GET64(query_vport_counter_out, \
+						vstats->query_vport_out, c)
+
+struct mlx5e_vport_stats {
+	__be64 query_vport_out[MLX5_ST_SZ_QW(query_vport_counter_out)];
+};
+
+#define PPORT_802_3_GET(pstats, c) \
+	MLX5_GET64(ppcnt_reg, pstats->IEEE_802_3_counters, \
+		   counter_set.eth_802_3_cntrs_grp_data_layout.c##_high)
+#define PPORT_2863_GET(pstats, c) \
+	MLX5_GET64(ppcnt_reg, pstats->RFC_2863_counters, \
+		   counter_set.eth_2863_cntrs_grp_data_layout.c##_high)
+#define PPORT_2819_GET(pstats, c) \
+	MLX5_GET64(ppcnt_reg, pstats->RFC_2819_counters, \
+		   counter_set.eth_2819_cntrs_grp_data_layout.c##_high)
+#define PPORT_PHY_STATISTICAL_GET(pstats, c) \
+	MLX5_GET64(ppcnt_reg, (pstats)->phy_statistical_counters, \
+		   counter_set.phys_layer_statistical_cntrs.c##_high)
+#define PPORT_PER_PRIO_GET(pstats, prio, c) \
+	MLX5_GET64(ppcnt_reg, pstats->per_prio_counters[prio], \
+		   counter_set.eth_per_prio_grp_data_layout.c##_high)
+#define NUM_PPORT_PRIO				8
+#define PPORT_ETH_EXT_GET(pstats, c) \
+	MLX5_GET64(ppcnt_reg, (pstats)->eth_ext_counters, \
+		   counter_set.eth_extended_cntrs_grp_data_layout.c##_high)
+
+struct mlx5e_pport_stats {
+	__be64 IEEE_802_3_counters[MLX5_ST_SZ_QW(ppcnt_reg)];
+	__be64 RFC_2863_counters[MLX5_ST_SZ_QW(ppcnt_reg)];
+	__be64 RFC_2819_counters[MLX5_ST_SZ_QW(ppcnt_reg)];
+	__be64 per_prio_counters[NUM_PPORT_PRIO][MLX5_ST_SZ_QW(ppcnt_reg)];
+	__be64 phy_counters[MLX5_ST_SZ_QW(ppcnt_reg)];
+	__be64 phy_statistical_counters[MLX5_ST_SZ_QW(ppcnt_reg)];
+	__be64 eth_ext_counters[MLX5_ST_SZ_QW(ppcnt_reg)];
+	__be64 per_tc_prio_counters[NUM_PPORT_PRIO][MLX5_ST_SZ_QW(ppcnt_reg)];
+	__be64 per_tc_congest_prio_counters[NUM_PPORT_PRIO][MLX5_ST_SZ_QW(ppcnt_reg)];
+};
+
+#define PCIE_PERF_GET(pcie_stats, c) \
+	MLX5_GET(mpcnt_reg, (pcie_stats)->pcie_perf_counters, \
+		 counter_set.pcie_perf_cntrs_grp_data_layout.c)
+
+#define PCIE_PERF_GET64(pcie_stats, c) \
+	MLX5_GET64(mpcnt_reg, (pcie_stats)->pcie_perf_counters, \
+		   counter_set.pcie_perf_cntrs_grp_data_layout.c##_high)
+
+struct mlx5e_pcie_stats {
+	__be64 pcie_perf_counters[MLX5_ST_SZ_QW(mpcnt_reg)];
+};
+
+struct mlx5e_rq_stats {
+	u64 packets;
+	u64 bytes;
+	u64 csum_complete;
+	u64 csum_complete_tail;
+	u64 csum_complete_tail_slow;
+	u64 csum_unnecessary;
+	u64 csum_unnecessary_inner;
+	u64 csum_none;
+	u64 lro_packets;
+	u64 lro_bytes;
+	u64 gro_packets;
+	u64 gro_bytes;
+	u64 gro_skbs;
+	u64 gro_match_packets;
+	u64 gro_large_hds;
+	u64 mcast_packets;
+	u64 ecn_mark;
+	u64 removed_vlan_packets;
+	u64 xdp_drop;
+	u64 xdp_redirect;
+	u64 wqe_err;
+	u64 mpwqe_filler_cqes;
+	u64 mpwqe_filler_strides;
+	u64 oversize_pkts_sw_drop;
+	u64 buff_alloc_err;
+	u64 cqe_compress_blks;
+	u64 cqe_compress_pkts;
+	u64 cache_reuse;
+	u64 cache_full;
+	u64 cache_empty;
+	u64 cache_busy;
+	u64 cache_waive;
+	u64 congst_umr;
+	u64 arfs_err;
+	u64 recover;
+#ifdef CONFIG_PAGE_POOL_STATS
+	u64 pp_alloc_fast;
+	u64 pp_alloc_slow;
+	u64 pp_alloc_slow_high_order;
+	u64 pp_alloc_empty;
+	u64 pp_alloc_refill;
+	u64 pp_alloc_waive;
+	u64 pp_recycle_cached;
+	u64 pp_recycle_cache_full;
+	u64 pp_recycle_ring;
+	u64 pp_recycle_ring_full;
+	u64 pp_recycle_released_ref;
+#endif
+#ifdef CONFIG_MLX5_EN_TLS
+	u64 tls_decrypted_packets;
+	u64 tls_decrypted_bytes;
+	u64 tls_resync_req_pkt;
+	u64 tls_resync_req_start;
+	u64 tls_resync_req_end;
+	u64 tls_resync_req_skip;
+	u64 tls_resync_res_ok;
+	u64 tls_resync_res_retry;
+	u64 tls_resync_res_skip;
+	u64 tls_err;
+#endif
+};
+
+struct mlx5e_sq_stats {
+	/* commonly accessed in data path */
+	u64 packets;
+	u64 bytes;
+	u64 xmit_more;
+	u64 tso_packets;
+	u64 tso_bytes;
+	u64 tso_inner_packets;
+	u64 tso_inner_bytes;
+	u64 csum_partial;
+	u64 csum_partial_inner;
+	u64 added_vlan_packets;
+	u64 nop;
+	u64 mpwqe_blks;
+	u64 mpwqe_pkts;
+#ifdef CONFIG_MLX5_EN_TLS
+	u64 tls_encrypted_packets;
+	u64 tls_encrypted_bytes;
+	u64 tls_ooo;
+	u64 tls_dump_packets;
+	u64 tls_dump_bytes;
+	u64 tls_resync_bytes;
+	u64 tls_skip_no_sync_data;
+	u64 tls_drop_no_sync_data;
+	u64 tls_drop_bypass_req;
+#endif
+	/* less likely accessed in data path */
+	u64 csum_none;
+	u64 stopped;
+	u64 dropped;
+	u64 recover;
+	/* dirtied @completion */
+	u64 cqes ____cacheline_aligned_in_smp;
+	u64 wake;
+	u64 cqe_err;
+};
+
+struct mlx5e_xdpsq_stats {
+	u64 xmit;
+	u64 mpwqe;
+	u64 inlnw;
+	u64 nops;
+	u64 full;
+	u64 err;
+	/* dirtied @completion */
+	u64 cqes ____cacheline_aligned_in_smp;
+};
+
+struct mlx5e_ch_stats {
+	u64 events;
+	u64 poll;
+	u64 arm;
+	u64 aff_change;
+	u64 force_irq;
+	u64 eq_rearm;
+};
+
+struct mlx5e_ptp_cq_stats {
+	u64 cqe;
+	u64 err_cqe;
+	u64 abort;
+	u64 abort_abs_diff_ns;
+	u64 resync_cqe;
+	u64 resync_event;
+	u64 ooo_cqe_drop;
+};
+
+struct mlx5e_stats {
+	struct mlx5e_sw_stats sw;
+	struct mlx5e_qcounter_stats qcnt;
+	struct mlx5e_vnic_env_stats vnic;
+	struct mlx5e_vport_stats vport;
+	struct mlx5e_pport_stats pport;
+	struct rtnl_link_stats64 vf_vport;
+	struct mlx5e_pcie_stats pcie;
+};
+
+extern mlx5e_stats_grp_t mlx5e_nic_stats_grps[];
+unsigned int mlx5e_nic_stats_grps_num(struct mlx5e_priv *priv);
+
+extern MLX5E_DECLARE_STATS_GRP(sw);
+extern MLX5E_DECLARE_STATS_GRP(qcnt);
+extern MLX5E_DECLARE_STATS_GRP(vnic_env);
+extern MLX5E_DECLARE_STATS_GRP(vport);
+extern MLX5E_DECLARE_STATS_GRP(802_3);
+extern MLX5E_DECLARE_STATS_GRP(2863);
+extern MLX5E_DECLARE_STATS_GRP(2819);
+extern MLX5E_DECLARE_STATS_GRP(phy);
+extern MLX5E_DECLARE_STATS_GRP(eth_ext);
+extern MLX5E_DECLARE_STATS_GRP(pcie);
+extern MLX5E_DECLARE_STATS_GRP(per_prio);
+extern MLX5E_DECLARE_STATS_GRP(pme);
+extern MLX5E_DECLARE_STATS_GRP(channels);
+extern MLX5E_DECLARE_STATS_GRP(per_port_buff_congest);
+extern MLX5E_DECLARE_STATS_GRP(ipsec_sw);
+extern MLX5E_DECLARE_STATS_GRP(ptp);
+extern MLX5E_DECLARE_STATS_GRP(macsec_hw);
+
+#endif /* __MLX5_EN_STATS_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
new file mode 100644
index 000000000..43239555f
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
@@ -0,0 +1,5307 @@
+/*
+ * Copyright (c) 2016, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <net/flow_dissector.h>
+#include <net/flow_offload.h>
+#include <net/sch_generic.h>
+#include <net/pkt_cls.h>
+#include <linux/mlx5/fs.h>
+#include <linux/mlx5/device.h>
+#include <linux/rhashtable.h>
+#include <linux/refcount.h>
+#include <linux/completion.h>
+#include <net/arp.h>
+#include <net/ipv6_stubs.h>
+#include <net/bareudp.h>
+#include <net/bonding.h>
+#include "en.h"
+#include "en/tc/post_act.h"
+#include "en_rep.h"
+#include "en/rep/tc.h"
+#include "en/rep/neigh.h"
+#include "en_tc.h"
+#include "eswitch.h"
+#include "fs_core.h"
+#include "en/port.h"
+#include "en/tc_tun.h"
+#include "en/mapping.h"
+#include "en/tc_ct.h"
+#include "en/mod_hdr.h"
+#include "en/tc_tun_encap.h"
+#include "en/tc/sample.h"
+#include "en/tc/act/act.h"
+#include "en/tc/post_meter.h"
+#include "lib/devcom.h"
+#include "lib/geneve.h"
+#include "lib/fs_chains.h"
+#include "diag/en_tc_tracepoint.h"
+#include <asm/div64.h>
+#include "lag/lag.h"
+#include "lag/mp.h"
+
+#define MLX5E_TC_TABLE_NUM_GROUPS 4
+#define MLX5E_TC_TABLE_MAX_GROUP_SIZE BIT(18)
+
+struct mlx5e_tc_table {
+	/* Protects the dynamic assignment of the t parameter
+	 * which is the nic tc root table.
+	 */
+	struct mutex			t_lock;
+	struct mlx5e_priv		*priv;
+	struct mlx5_flow_table		*t;
+	struct mlx5_flow_table		*miss_t;
+	struct mlx5_fs_chains           *chains;
+	struct mlx5e_post_act		*post_act;
+
+	struct rhashtable               ht;
+
+	struct mod_hdr_tbl mod_hdr;
+	struct mutex hairpin_tbl_lock; /* protects hairpin_tbl */
+	DECLARE_HASHTABLE(hairpin_tbl, 8);
+
+	struct notifier_block     netdevice_nb;
+	struct netdev_net_notifier	netdevice_nn;
+
+	struct mlx5_tc_ct_priv         *ct;
+	struct mapping_ctx             *mapping;
+};
+
+struct mlx5e_tc_attr_to_reg_mapping mlx5e_tc_attr_to_reg_mappings[] = {
+	[CHAIN_TO_REG] = {
+		.mfield = MLX5_ACTION_IN_FIELD_METADATA_REG_C_0,
+		.moffset = 0,
+		.mlen = 16,
+	},
+	[VPORT_TO_REG] = {
+		.mfield = MLX5_ACTION_IN_FIELD_METADATA_REG_C_0,
+		.moffset = 16,
+		.mlen = 16,
+	},
+	[TUNNEL_TO_REG] = {
+		.mfield = MLX5_ACTION_IN_FIELD_METADATA_REG_C_1,
+		.moffset = 8,
+		.mlen = ESW_TUN_OPTS_BITS + ESW_TUN_ID_BITS,
+		.soffset = MLX5_BYTE_OFF(fte_match_param,
+					 misc_parameters_2.metadata_reg_c_1),
+	},
+	[ZONE_TO_REG] = zone_to_reg_ct,
+	[ZONE_RESTORE_TO_REG] = zone_restore_to_reg_ct,
+	[CTSTATE_TO_REG] = ctstate_to_reg_ct,
+	[MARK_TO_REG] = mark_to_reg_ct,
+	[LABELS_TO_REG] = labels_to_reg_ct,
+	[FTEID_TO_REG] = fteid_to_reg_ct,
+	/* For NIC rules we store the restore metadata directly
+	 * into reg_b that is passed to SW since we don't
+	 * jump between steering domains.
+	 */
+	[NIC_CHAIN_TO_REG] = {
+		.mfield = MLX5_ACTION_IN_FIELD_METADATA_REG_B,
+		.moffset = 0,
+		.mlen = 16,
+	},
+	[NIC_ZONE_RESTORE_TO_REG] = nic_zone_restore_to_reg_ct,
+	[PACKET_COLOR_TO_REG] = packet_color_to_reg,
+};
+
+struct mlx5e_tc_table *mlx5e_tc_table_alloc(void)
+{
+	struct mlx5e_tc_table *tc;
+
+	tc = kvzalloc(sizeof(*tc), GFP_KERNEL);
+	return tc ? tc : ERR_PTR(-ENOMEM);
+}
+
+void mlx5e_tc_table_free(struct mlx5e_tc_table *tc)
+{
+	kvfree(tc);
+}
+
+struct mlx5_fs_chains *mlx5e_nic_chains(struct mlx5e_tc_table *tc)
+{
+	return tc->chains;
+}
+
+/* To avoid false lock dependency warning set the tc_ht lock
+ * class different than the lock class of the ht being used when deleting
+ * last flow from a group and then deleting a group, we get into del_sw_flow_group()
+ * which call rhashtable_destroy on fg->ftes_hash which will take ht->mutex but
+ * it's different than the ht->mutex here.
+ */
+static struct lock_class_key tc_ht_lock_key;
+static struct lock_class_key tc_ht_wq_key;
+
+static void mlx5e_put_flow_tunnel_id(struct mlx5e_tc_flow *flow);
+static void free_flow_post_acts(struct mlx5e_tc_flow *flow);
+
+void
+mlx5e_tc_match_to_reg_match(struct mlx5_flow_spec *spec,
+			    enum mlx5e_tc_attr_to_reg type,
+			    u32 val,
+			    u32 mask)
+{
+	void *headers_c = spec->match_criteria, *headers_v = spec->match_value, *fmask, *fval;
+	int soffset = mlx5e_tc_attr_to_reg_mappings[type].soffset;
+	int moffset = mlx5e_tc_attr_to_reg_mappings[type].moffset;
+	int match_len = mlx5e_tc_attr_to_reg_mappings[type].mlen;
+	u32 max_mask = GENMASK(match_len - 1, 0);
+	__be32 curr_mask_be, curr_val_be;
+	u32 curr_mask, curr_val;
+
+	fmask = headers_c + soffset;
+	fval = headers_v + soffset;
+
+	memcpy(&curr_mask_be, fmask, 4);
+	memcpy(&curr_val_be, fval, 4);
+
+	curr_mask = be32_to_cpu(curr_mask_be);
+	curr_val = be32_to_cpu(curr_val_be);
+
+	//move to correct offset
+	WARN_ON(mask > max_mask);
+	mask <<= moffset;
+	val <<= moffset;
+	max_mask <<= moffset;
+
+	//zero val and mask
+	curr_mask &= ~max_mask;
+	curr_val &= ~max_mask;
+
+	//add current to mask
+	curr_mask |= mask;
+	curr_val |= val;
+
+	//back to be32 and write
+	curr_mask_be = cpu_to_be32(curr_mask);
+	curr_val_be = cpu_to_be32(curr_val);
+
+	memcpy(fmask, &curr_mask_be, 4);
+	memcpy(fval, &curr_val_be, 4);
+
+	spec->match_criteria_enable |= MLX5_MATCH_MISC_PARAMETERS_2;
+}
+
+void
+mlx5e_tc_match_to_reg_get_match(struct mlx5_flow_spec *spec,
+				enum mlx5e_tc_attr_to_reg type,
+				u32 *val,
+				u32 *mask)
+{
+	void *headers_c = spec->match_criteria, *headers_v = spec->match_value, *fmask, *fval;
+	int soffset = mlx5e_tc_attr_to_reg_mappings[type].soffset;
+	int moffset = mlx5e_tc_attr_to_reg_mappings[type].moffset;
+	int match_len = mlx5e_tc_attr_to_reg_mappings[type].mlen;
+	u32 max_mask = GENMASK(match_len - 1, 0);
+	__be32 curr_mask_be, curr_val_be;
+	u32 curr_mask, curr_val;
+
+	fmask = headers_c + soffset;
+	fval = headers_v + soffset;
+
+	memcpy(&curr_mask_be, fmask, 4);
+	memcpy(&curr_val_be, fval, 4);
+
+	curr_mask = be32_to_cpu(curr_mask_be);
+	curr_val = be32_to_cpu(curr_val_be);
+
+	*mask = (curr_mask >> moffset) & max_mask;
+	*val = (curr_val >> moffset) & max_mask;
+}
+
+int
+mlx5e_tc_match_to_reg_set_and_get_id(struct mlx5_core_dev *mdev,
+				     struct mlx5e_tc_mod_hdr_acts *mod_hdr_acts,
+				     enum mlx5_flow_namespace_type ns,
+				     enum mlx5e_tc_attr_to_reg type,
+				     u32 data)
+{
+	int moffset = mlx5e_tc_attr_to_reg_mappings[type].moffset;
+	int mfield = mlx5e_tc_attr_to_reg_mappings[type].mfield;
+	int mlen = mlx5e_tc_attr_to_reg_mappings[type].mlen;
+	char *modact;
+	int err;
+
+	modact = mlx5e_mod_hdr_alloc(mdev, ns, mod_hdr_acts);
+	if (IS_ERR(modact))
+		return PTR_ERR(modact);
+
+	/* Firmware has 5bit length field and 0 means 32bits */
+	if (mlen == 32)
+		mlen = 0;
+
+	MLX5_SET(set_action_in, modact, action_type, MLX5_ACTION_TYPE_SET);
+	MLX5_SET(set_action_in, modact, field, mfield);
+	MLX5_SET(set_action_in, modact, offset, moffset);
+	MLX5_SET(set_action_in, modact, length, mlen);
+	MLX5_SET(set_action_in, modact, data, data);
+	err = mod_hdr_acts->num_actions;
+	mod_hdr_acts->num_actions++;
+
+	return err;
+}
+
+struct mlx5e_tc_int_port_priv *
+mlx5e_get_int_port_priv(struct mlx5e_priv *priv)
+{
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	struct mlx5_rep_uplink_priv *uplink_priv;
+	struct mlx5e_rep_priv *uplink_rpriv;
+
+	if (is_mdev_switchdev_mode(priv->mdev)) {
+		uplink_rpriv = mlx5_eswitch_get_uplink_priv(esw, REP_ETH);
+		uplink_priv = &uplink_rpriv->uplink_priv;
+
+		return uplink_priv->int_port_priv;
+	}
+
+	return NULL;
+}
+
+struct mlx5e_flow_meters *
+mlx5e_get_flow_meters(struct mlx5_core_dev *dev)
+{
+	struct mlx5_eswitch *esw = dev->priv.eswitch;
+	struct mlx5_rep_uplink_priv *uplink_priv;
+	struct mlx5e_rep_priv *uplink_rpriv;
+	struct mlx5e_priv *priv;
+
+	if (is_mdev_switchdev_mode(dev)) {
+		uplink_rpriv = mlx5_eswitch_get_uplink_priv(esw, REP_ETH);
+		uplink_priv = &uplink_rpriv->uplink_priv;
+		priv = netdev_priv(uplink_rpriv->netdev);
+		if (!uplink_priv->flow_meters)
+			uplink_priv->flow_meters =
+				mlx5e_flow_meters_init(priv,
+						       MLX5_FLOW_NAMESPACE_FDB,
+						       uplink_priv->post_act);
+		if (!IS_ERR(uplink_priv->flow_meters))
+			return uplink_priv->flow_meters;
+	}
+
+	return NULL;
+}
+
+static struct mlx5_tc_ct_priv *
+get_ct_priv(struct mlx5e_priv *priv)
+{
+	struct mlx5e_tc_table *tc = mlx5e_fs_get_tc(priv->fs);
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	struct mlx5_rep_uplink_priv *uplink_priv;
+	struct mlx5e_rep_priv *uplink_rpriv;
+
+	if (is_mdev_switchdev_mode(priv->mdev)) {
+		uplink_rpriv = mlx5_eswitch_get_uplink_priv(esw, REP_ETH);
+		uplink_priv = &uplink_rpriv->uplink_priv;
+
+		return uplink_priv->ct_priv;
+	}
+
+	return tc->ct;
+}
+
+static struct mlx5e_tc_psample *
+get_sample_priv(struct mlx5e_priv *priv)
+{
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	struct mlx5_rep_uplink_priv *uplink_priv;
+	struct mlx5e_rep_priv *uplink_rpriv;
+
+	if (is_mdev_switchdev_mode(priv->mdev)) {
+		uplink_rpriv = mlx5_eswitch_get_uplink_priv(esw, REP_ETH);
+		uplink_priv = &uplink_rpriv->uplink_priv;
+
+		return uplink_priv->tc_psample;
+	}
+
+	return NULL;
+}
+
+static struct mlx5e_post_act *
+get_post_action(struct mlx5e_priv *priv)
+{
+	struct mlx5e_tc_table *tc = mlx5e_fs_get_tc(priv->fs);
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	struct mlx5_rep_uplink_priv *uplink_priv;
+	struct mlx5e_rep_priv *uplink_rpriv;
+
+	if (is_mdev_switchdev_mode(priv->mdev)) {
+		uplink_rpriv = mlx5_eswitch_get_uplink_priv(esw, REP_ETH);
+		uplink_priv = &uplink_rpriv->uplink_priv;
+
+		return uplink_priv->post_act;
+	}
+
+	return tc->post_act;
+}
+
+struct mlx5_flow_handle *
+mlx5_tc_rule_insert(struct mlx5e_priv *priv,
+		    struct mlx5_flow_spec *spec,
+		    struct mlx5_flow_attr *attr)
+{
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+
+	if (is_mdev_switchdev_mode(priv->mdev))
+		return mlx5_eswitch_add_offloaded_rule(esw, spec, attr);
+
+	return	mlx5e_add_offloaded_nic_rule(priv, spec, attr);
+}
+
+void
+mlx5_tc_rule_delete(struct mlx5e_priv *priv,
+		    struct mlx5_flow_handle *rule,
+		    struct mlx5_flow_attr *attr)
+{
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+
+	if (is_mdev_switchdev_mode(priv->mdev)) {
+		mlx5_eswitch_del_offloaded_rule(esw, rule, attr);
+		return;
+	}
+
+	mlx5e_del_offloaded_nic_rule(priv, rule, attr);
+}
+
+static bool
+is_flow_meter_action(struct mlx5_flow_attr *attr)
+{
+	return ((attr->action & MLX5_FLOW_CONTEXT_ACTION_EXECUTE_ASO) &&
+		(attr->exe_aso_type == MLX5_EXE_ASO_FLOW_METER));
+}
+
+static int
+mlx5e_tc_add_flow_meter(struct mlx5e_priv *priv,
+			struct mlx5_flow_attr *attr)
+{
+	struct mlx5e_post_act *post_act = get_post_action(priv);
+	struct mlx5e_post_meter_priv *post_meter;
+	enum mlx5_flow_namespace_type ns_type;
+	struct mlx5e_flow_meter_handle *meter;
+
+	meter = mlx5e_tc_meter_replace(priv->mdev, &attr->meter_attr.params);
+	if (IS_ERR(meter)) {
+		mlx5_core_err(priv->mdev, "Failed to get flow meter\n");
+		return PTR_ERR(meter);
+	}
+
+	ns_type = mlx5e_tc_meter_get_namespace(meter->flow_meters);
+	post_meter = mlx5e_post_meter_init(priv, ns_type, post_act, meter->green_counter,
+					   meter->red_counter);
+	if (IS_ERR(post_meter)) {
+		mlx5_core_err(priv->mdev, "Failed to init post meter\n");
+		goto err_meter_init;
+	}
+
+	attr->meter_attr.meter = meter;
+	attr->meter_attr.post_meter = post_meter;
+	attr->dest_ft = mlx5e_post_meter_get_ft(post_meter);
+	attr->action |= MLX5_FLOW_CONTEXT_ACTION_FWD_DEST;
+
+	return 0;
+
+err_meter_init:
+	mlx5e_tc_meter_put(meter);
+	return PTR_ERR(post_meter);
+}
+
+static void
+mlx5e_tc_del_flow_meter(struct mlx5_flow_attr *attr)
+{
+	mlx5e_post_meter_cleanup(attr->meter_attr.post_meter);
+	mlx5e_tc_meter_put(attr->meter_attr.meter);
+}
+
+struct mlx5_flow_handle *
+mlx5e_tc_rule_offload(struct mlx5e_priv *priv,
+		      struct mlx5_flow_spec *spec,
+		      struct mlx5_flow_attr *attr)
+{
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	int err;
+
+	if (attr->flags & MLX5_ATTR_FLAG_CT) {
+		struct mlx5e_tc_mod_hdr_acts *mod_hdr_acts =
+			&attr->parse_attr->mod_hdr_acts;
+
+		return mlx5_tc_ct_flow_offload(get_ct_priv(priv),
+					       spec, attr,
+					       mod_hdr_acts);
+	}
+
+	if (!is_mdev_switchdev_mode(priv->mdev))
+		return mlx5e_add_offloaded_nic_rule(priv, spec, attr);
+
+	if (attr->flags & MLX5_ATTR_FLAG_SAMPLE)
+		return mlx5e_tc_sample_offload(get_sample_priv(priv), spec, attr);
+
+	if (is_flow_meter_action(attr)) {
+		err = mlx5e_tc_add_flow_meter(priv, attr);
+		if (err)
+			return ERR_PTR(err);
+	}
+
+	return mlx5_eswitch_add_offloaded_rule(esw, spec, attr);
+}
+
+void
+mlx5e_tc_rule_unoffload(struct mlx5e_priv *priv,
+			struct mlx5_flow_handle *rule,
+			struct mlx5_flow_attr *attr)
+{
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+
+	if (attr->flags & MLX5_ATTR_FLAG_CT) {
+		mlx5_tc_ct_delete_flow(get_ct_priv(priv), attr);
+		return;
+	}
+
+	if (!is_mdev_switchdev_mode(priv->mdev)) {
+		mlx5e_del_offloaded_nic_rule(priv, rule, attr);
+		return;
+	}
+
+	if (attr->flags & MLX5_ATTR_FLAG_SAMPLE) {
+		mlx5e_tc_sample_unoffload(get_sample_priv(priv), rule, attr);
+		return;
+	}
+
+	mlx5_eswitch_del_offloaded_rule(esw, rule, attr);
+
+	if (attr->meter_attr.meter)
+		mlx5e_tc_del_flow_meter(attr);
+}
+
+int
+mlx5e_tc_match_to_reg_set(struct mlx5_core_dev *mdev,
+			  struct mlx5e_tc_mod_hdr_acts *mod_hdr_acts,
+			  enum mlx5_flow_namespace_type ns,
+			  enum mlx5e_tc_attr_to_reg type,
+			  u32 data)
+{
+	int ret = mlx5e_tc_match_to_reg_set_and_get_id(mdev, mod_hdr_acts, ns, type, data);
+
+	return ret < 0 ? ret : 0;
+}
+
+void mlx5e_tc_match_to_reg_mod_hdr_change(struct mlx5_core_dev *mdev,
+					  struct mlx5e_tc_mod_hdr_acts *mod_hdr_acts,
+					  enum mlx5e_tc_attr_to_reg type,
+					  int act_id, u32 data)
+{
+	int moffset = mlx5e_tc_attr_to_reg_mappings[type].moffset;
+	int mfield = mlx5e_tc_attr_to_reg_mappings[type].mfield;
+	int mlen = mlx5e_tc_attr_to_reg_mappings[type].mlen;
+	char *modact;
+
+	modact = mlx5e_mod_hdr_get_item(mod_hdr_acts, act_id);
+
+	/* Firmware has 5bit length field and 0 means 32bits */
+	if (mlen == 32)
+		mlen = 0;
+
+	MLX5_SET(set_action_in, modact, action_type, MLX5_ACTION_TYPE_SET);
+	MLX5_SET(set_action_in, modact, field, mfield);
+	MLX5_SET(set_action_in, modact, offset, moffset);
+	MLX5_SET(set_action_in, modact, length, mlen);
+	MLX5_SET(set_action_in, modact, data, data);
+}
+
+struct mlx5e_hairpin {
+	struct mlx5_hairpin *pair;
+
+	struct mlx5_core_dev *func_mdev;
+	struct mlx5e_priv *func_priv;
+	u32 tdn;
+	struct mlx5e_tir direct_tir;
+
+	int num_channels;
+	struct mlx5e_rqt indir_rqt;
+	struct mlx5e_tir indir_tir[MLX5E_NUM_INDIR_TIRS];
+	struct mlx5_ttc_table *ttc;
+};
+
+struct mlx5e_hairpin_entry {
+	/* a node of a hash table which keeps all the  hairpin entries */
+	struct hlist_node hairpin_hlist;
+
+	/* protects flows list */
+	spinlock_t flows_lock;
+	/* flows sharing the same hairpin */
+	struct list_head flows;
+	/* hpe's that were not fully initialized when dead peer update event
+	 * function traversed them.
+	 */
+	struct list_head dead_peer_wait_list;
+
+	u16 peer_vhca_id;
+	u8 prio;
+	struct mlx5e_hairpin *hp;
+	refcount_t refcnt;
+	struct completion res_ready;
+};
+
+static void mlx5e_tc_del_flow(struct mlx5e_priv *priv,
+			      struct mlx5e_tc_flow *flow);
+
+struct mlx5e_tc_flow *mlx5e_flow_get(struct mlx5e_tc_flow *flow)
+{
+	if (!flow || !refcount_inc_not_zero(&flow->refcnt))
+		return ERR_PTR(-EINVAL);
+	return flow;
+}
+
+void mlx5e_flow_put(struct mlx5e_priv *priv, struct mlx5e_tc_flow *flow)
+{
+	if (refcount_dec_and_test(&flow->refcnt)) {
+		mlx5e_tc_del_flow(priv, flow);
+		kfree_rcu(flow, rcu_head);
+	}
+}
+
+bool mlx5e_is_eswitch_flow(struct mlx5e_tc_flow *flow)
+{
+	return flow_flag_test(flow, ESWITCH);
+}
+
+bool mlx5e_is_ft_flow(struct mlx5e_tc_flow *flow)
+{
+	return flow_flag_test(flow, FT);
+}
+
+bool mlx5e_is_offloaded_flow(struct mlx5e_tc_flow *flow)
+{
+	return flow_flag_test(flow, OFFLOADED);
+}
+
+int mlx5e_get_flow_namespace(struct mlx5e_tc_flow *flow)
+{
+	return mlx5e_is_eswitch_flow(flow) ?
+		MLX5_FLOW_NAMESPACE_FDB : MLX5_FLOW_NAMESPACE_KERNEL;
+}
+
+static struct mod_hdr_tbl *
+get_mod_hdr_table(struct mlx5e_priv *priv, struct mlx5e_tc_flow *flow)
+{
+	struct mlx5e_tc_table *tc = mlx5e_fs_get_tc(priv->fs);
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+
+	return mlx5e_get_flow_namespace(flow) == MLX5_FLOW_NAMESPACE_FDB ?
+		&esw->offloads.mod_hdr :
+		&tc->mod_hdr;
+}
+
+static int mlx5e_attach_mod_hdr(struct mlx5e_priv *priv,
+				struct mlx5e_tc_flow *flow,
+				struct mlx5e_tc_flow_parse_attr *parse_attr)
+{
+	struct mlx5_modify_hdr *modify_hdr;
+	struct mlx5e_mod_hdr_handle *mh;
+
+	mh = mlx5e_mod_hdr_attach(priv->mdev, get_mod_hdr_table(priv, flow),
+				  mlx5e_get_flow_namespace(flow),
+				  &parse_attr->mod_hdr_acts);
+	if (IS_ERR(mh))
+		return PTR_ERR(mh);
+
+	modify_hdr = mlx5e_mod_hdr_get(mh);
+	flow->attr->modify_hdr = modify_hdr;
+	flow->mh = mh;
+
+	return 0;
+}
+
+static void mlx5e_detach_mod_hdr(struct mlx5e_priv *priv,
+				 struct mlx5e_tc_flow *flow)
+{
+	/* flow wasn't fully initialized */
+	if (!flow->mh)
+		return;
+
+	mlx5e_mod_hdr_detach(priv->mdev, get_mod_hdr_table(priv, flow),
+			     flow->mh);
+	flow->mh = NULL;
+}
+
+static
+struct mlx5_core_dev *mlx5e_hairpin_get_mdev(struct net *net, int ifindex)
+{
+	struct mlx5_core_dev *mdev;
+	struct net_device *netdev;
+	struct mlx5e_priv *priv;
+
+	netdev = dev_get_by_index(net, ifindex);
+	if (!netdev)
+		return ERR_PTR(-ENODEV);
+
+	priv = netdev_priv(netdev);
+	mdev = priv->mdev;
+	dev_put(netdev);
+
+	/* Mirred tc action holds a refcount on the ifindex net_device (see
+	 * net/sched/act_mirred.c:tcf_mirred_get_dev). So, it's okay to continue using mdev
+	 * after dev_put(netdev), while we're in the context of adding a tc flow.
+	 *
+	 * The mdev pointer corresponds to the peer/out net_device of a hairpin. It is then
+	 * stored in a hairpin object, which exists until all flows, that refer to it, get
+	 * removed.
+	 *
+	 * On the other hand, after a hairpin object has been created, the peer net_device may
+	 * be removed/unbound while there are still some hairpin flows that are using it. This
+	 * case is handled by mlx5e_tc_hairpin_update_dead_peer, which is hooked to
+	 * NETDEV_UNREGISTER event of the peer net_device.
+	 */
+	return mdev;
+}
+
+static int mlx5e_hairpin_create_transport(struct mlx5e_hairpin *hp)
+{
+	struct mlx5e_tir_builder *builder;
+	int err;
+
+	builder = mlx5e_tir_builder_alloc(false);
+	if (!builder)
+		return -ENOMEM;
+
+	err = mlx5_core_alloc_transport_domain(hp->func_mdev, &hp->tdn);
+	if (err)
+		goto out;
+
+	mlx5e_tir_builder_build_inline(builder, hp->tdn, hp->pair->rqn[0]);
+	err = mlx5e_tir_init(&hp->direct_tir, builder, hp->func_mdev, false);
+	if (err)
+		goto create_tir_err;
+
+out:
+	mlx5e_tir_builder_free(builder);
+	return err;
+
+create_tir_err:
+	mlx5_core_dealloc_transport_domain(hp->func_mdev, hp->tdn);
+
+	goto out;
+}
+
+static void mlx5e_hairpin_destroy_transport(struct mlx5e_hairpin *hp)
+{
+	mlx5e_tir_destroy(&hp->direct_tir);
+	mlx5_core_dealloc_transport_domain(hp->func_mdev, hp->tdn);
+}
+
+static int mlx5e_hairpin_create_indirect_rqt(struct mlx5e_hairpin *hp)
+{
+	struct mlx5e_priv *priv = hp->func_priv;
+	struct mlx5_core_dev *mdev = priv->mdev;
+	struct mlx5e_rss_params_indir *indir;
+	int err;
+
+	indir = kvmalloc(sizeof(*indir), GFP_KERNEL);
+	if (!indir)
+		return -ENOMEM;
+
+	mlx5e_rss_params_indir_init_uniform(indir, hp->num_channels);
+	err = mlx5e_rqt_init_indir(&hp->indir_rqt, mdev, hp->pair->rqn, hp->num_channels,
+				   mlx5e_rx_res_get_current_hash(priv->rx_res).hfunc,
+				   indir);
+
+	kvfree(indir);
+	return err;
+}
+
+static int mlx5e_hairpin_create_indirect_tirs(struct mlx5e_hairpin *hp)
+{
+	struct mlx5e_priv *priv = hp->func_priv;
+	struct mlx5e_rss_params_hash rss_hash;
+	enum mlx5_traffic_types tt, max_tt;
+	struct mlx5e_tir_builder *builder;
+	int err = 0;
+
+	builder = mlx5e_tir_builder_alloc(false);
+	if (!builder)
+		return -ENOMEM;
+
+	rss_hash = mlx5e_rx_res_get_current_hash(priv->rx_res);
+
+	for (tt = 0; tt < MLX5E_NUM_INDIR_TIRS; tt++) {
+		struct mlx5e_rss_params_traffic_type rss_tt;
+
+		rss_tt = mlx5e_rss_get_default_tt_config(tt);
+
+		mlx5e_tir_builder_build_rqt(builder, hp->tdn,
+					    mlx5e_rqt_get_rqtn(&hp->indir_rqt),
+					    false);
+		mlx5e_tir_builder_build_rss(builder, &rss_hash, &rss_tt, false);
+
+		err = mlx5e_tir_init(&hp->indir_tir[tt], builder, hp->func_mdev, false);
+		if (err) {
+			mlx5_core_warn(hp->func_mdev, "create indirect tirs failed, %d\n", err);
+			goto err_destroy_tirs;
+		}
+
+		mlx5e_tir_builder_clear(builder);
+	}
+
+out:
+	mlx5e_tir_builder_free(builder);
+	return err;
+
+err_destroy_tirs:
+	max_tt = tt;
+	for (tt = 0; tt < max_tt; tt++)
+		mlx5e_tir_destroy(&hp->indir_tir[tt]);
+
+	goto out;
+}
+
+static void mlx5e_hairpin_destroy_indirect_tirs(struct mlx5e_hairpin *hp)
+{
+	int tt;
+
+	for (tt = 0; tt < MLX5E_NUM_INDIR_TIRS; tt++)
+		mlx5e_tir_destroy(&hp->indir_tir[tt]);
+}
+
+static void mlx5e_hairpin_set_ttc_params(struct mlx5e_hairpin *hp,
+					 struct ttc_params *ttc_params)
+{
+	struct mlx5_flow_table_attr *ft_attr = &ttc_params->ft_attr;
+	int tt;
+
+	memset(ttc_params, 0, sizeof(*ttc_params));
+
+	ttc_params->ns = mlx5_get_flow_namespace(hp->func_mdev,
+						 MLX5_FLOW_NAMESPACE_KERNEL);
+	for (tt = 0; tt < MLX5_NUM_TT; tt++) {
+		ttc_params->dests[tt].type = MLX5_FLOW_DESTINATION_TYPE_TIR;
+		ttc_params->dests[tt].tir_num =
+			tt == MLX5_TT_ANY ?
+				mlx5e_tir_get_tirn(&hp->direct_tir) :
+				mlx5e_tir_get_tirn(&hp->indir_tir[tt]);
+	}
+
+	ft_attr->level = MLX5E_TC_TTC_FT_LEVEL;
+	ft_attr->prio = MLX5E_TC_PRIO;
+}
+
+static int mlx5e_hairpin_rss_init(struct mlx5e_hairpin *hp)
+{
+	struct mlx5e_priv *priv = hp->func_priv;
+	struct ttc_params ttc_params;
+	struct mlx5_ttc_table *ttc;
+	int err;
+
+	err = mlx5e_hairpin_create_indirect_rqt(hp);
+	if (err)
+		return err;
+
+	err = mlx5e_hairpin_create_indirect_tirs(hp);
+	if (err)
+		goto err_create_indirect_tirs;
+
+	mlx5e_hairpin_set_ttc_params(hp, &ttc_params);
+	hp->ttc = mlx5_create_ttc_table(priv->mdev, &ttc_params);
+	if (IS_ERR(hp->ttc)) {
+		err = PTR_ERR(hp->ttc);
+		goto err_create_ttc_table;
+	}
+
+	ttc = mlx5e_fs_get_ttc(priv->fs, false);
+	netdev_dbg(priv->netdev, "add hairpin: using %d channels rss ttc table id %x\n",
+		   hp->num_channels,
+		   mlx5_get_ttc_flow_table(ttc)->id);
+
+	return 0;
+
+err_create_ttc_table:
+	mlx5e_hairpin_destroy_indirect_tirs(hp);
+err_create_indirect_tirs:
+	mlx5e_rqt_destroy(&hp->indir_rqt);
+
+	return err;
+}
+
+static void mlx5e_hairpin_rss_cleanup(struct mlx5e_hairpin *hp)
+{
+	mlx5_destroy_ttc_table(hp->ttc);
+	mlx5e_hairpin_destroy_indirect_tirs(hp);
+	mlx5e_rqt_destroy(&hp->indir_rqt);
+}
+
+static struct mlx5e_hairpin *
+mlx5e_hairpin_create(struct mlx5e_priv *priv, struct mlx5_hairpin_params *params,
+		     int peer_ifindex)
+{
+	struct mlx5_core_dev *func_mdev, *peer_mdev;
+	struct mlx5e_hairpin *hp;
+	struct mlx5_hairpin *pair;
+	int err;
+
+	hp = kzalloc(sizeof(*hp), GFP_KERNEL);
+	if (!hp)
+		return ERR_PTR(-ENOMEM);
+
+	func_mdev = priv->mdev;
+	peer_mdev = mlx5e_hairpin_get_mdev(dev_net(priv->netdev), peer_ifindex);
+	if (IS_ERR(peer_mdev)) {
+		err = PTR_ERR(peer_mdev);
+		goto create_pair_err;
+	}
+
+	pair = mlx5_core_hairpin_create(func_mdev, peer_mdev, params);
+	if (IS_ERR(pair)) {
+		err = PTR_ERR(pair);
+		goto create_pair_err;
+	}
+	hp->pair = pair;
+	hp->func_mdev = func_mdev;
+	hp->func_priv = priv;
+	hp->num_channels = params->num_channels;
+
+	err = mlx5e_hairpin_create_transport(hp);
+	if (err)
+		goto create_transport_err;
+
+	if (hp->num_channels > 1) {
+		err = mlx5e_hairpin_rss_init(hp);
+		if (err)
+			goto rss_init_err;
+	}
+
+	return hp;
+
+rss_init_err:
+	mlx5e_hairpin_destroy_transport(hp);
+create_transport_err:
+	mlx5_core_hairpin_destroy(hp->pair);
+create_pair_err:
+	kfree(hp);
+	return ERR_PTR(err);
+}
+
+static void mlx5e_hairpin_destroy(struct mlx5e_hairpin *hp)
+{
+	if (hp->num_channels > 1)
+		mlx5e_hairpin_rss_cleanup(hp);
+	mlx5e_hairpin_destroy_transport(hp);
+	mlx5_core_hairpin_destroy(hp->pair);
+	kvfree(hp);
+}
+
+static inline u32 hash_hairpin_info(u16 peer_vhca_id, u8 prio)
+{
+	return (peer_vhca_id << 16 | prio);
+}
+
+static struct mlx5e_hairpin_entry *mlx5e_hairpin_get(struct mlx5e_priv *priv,
+						     u16 peer_vhca_id, u8 prio)
+{
+	struct mlx5e_tc_table *tc = mlx5e_fs_get_tc(priv->fs);
+	struct mlx5e_hairpin_entry *hpe;
+	u32 hash_key = hash_hairpin_info(peer_vhca_id, prio);
+
+	hash_for_each_possible(tc->hairpin_tbl, hpe,
+			       hairpin_hlist, hash_key) {
+		if (hpe->peer_vhca_id == peer_vhca_id && hpe->prio == prio) {
+			refcount_inc(&hpe->refcnt);
+			return hpe;
+		}
+	}
+
+	return NULL;
+}
+
+static void mlx5e_hairpin_put(struct mlx5e_priv *priv,
+			      struct mlx5e_hairpin_entry *hpe)
+{
+	struct mlx5e_tc_table *tc = mlx5e_fs_get_tc(priv->fs);
+	/* no more hairpin flows for us, release the hairpin pair */
+	if (!refcount_dec_and_mutex_lock(&hpe->refcnt, &tc->hairpin_tbl_lock))
+		return;
+	hash_del(&hpe->hairpin_hlist);
+	mutex_unlock(&tc->hairpin_tbl_lock);
+
+	if (!IS_ERR_OR_NULL(hpe->hp)) {
+		netdev_dbg(priv->netdev, "del hairpin: peer %s\n",
+			   dev_name(hpe->hp->pair->peer_mdev->device));
+
+		mlx5e_hairpin_destroy(hpe->hp);
+	}
+
+	WARN_ON(!list_empty(&hpe->flows));
+	kfree(hpe);
+}
+
+#define UNKNOWN_MATCH_PRIO 8
+
+static int mlx5e_hairpin_get_prio(struct mlx5e_priv *priv,
+				  struct mlx5_flow_spec *spec, u8 *match_prio,
+				  struct netlink_ext_ack *extack)
+{
+	void *headers_c, *headers_v;
+	u8 prio_val, prio_mask = 0;
+	bool vlan_present;
+
+#ifdef CONFIG_MLX5_CORE_EN_DCB
+	if (priv->dcbx_dp.trust_state != MLX5_QPTS_TRUST_PCP) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "only PCP trust state supported for hairpin");
+		return -EOPNOTSUPP;
+	}
+#endif
+	headers_c = MLX5_ADDR_OF(fte_match_param, spec->match_criteria, outer_headers);
+	headers_v = MLX5_ADDR_OF(fte_match_param, spec->match_value, outer_headers);
+
+	vlan_present = MLX5_GET(fte_match_set_lyr_2_4, headers_v, cvlan_tag);
+	if (vlan_present) {
+		prio_mask = MLX5_GET(fte_match_set_lyr_2_4, headers_c, first_prio);
+		prio_val = MLX5_GET(fte_match_set_lyr_2_4, headers_v, first_prio);
+	}
+
+	if (!vlan_present || !prio_mask) {
+		prio_val = UNKNOWN_MATCH_PRIO;
+	} else if (prio_mask != 0x7) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "masked priority match not supported for hairpin");
+		return -EOPNOTSUPP;
+	}
+
+	*match_prio = prio_val;
+	return 0;
+}
+
+static int mlx5e_hairpin_flow_add(struct mlx5e_priv *priv,
+				  struct mlx5e_tc_flow *flow,
+				  struct mlx5e_tc_flow_parse_attr *parse_attr,
+				  struct netlink_ext_ack *extack)
+{
+	struct mlx5e_tc_table *tc = mlx5e_fs_get_tc(priv->fs);
+	int peer_ifindex = parse_attr->mirred_ifindex[0];
+	struct mlx5_hairpin_params params;
+	struct mlx5_core_dev *peer_mdev;
+	struct mlx5e_hairpin_entry *hpe;
+	struct mlx5e_hairpin *hp;
+	u64 link_speed64;
+	u32 link_speed;
+	u8 match_prio;
+	u16 peer_id;
+	int err;
+
+	peer_mdev = mlx5e_hairpin_get_mdev(dev_net(priv->netdev), peer_ifindex);
+	if (IS_ERR(peer_mdev)) {
+		NL_SET_ERR_MSG_MOD(extack, "invalid ifindex of mirred device");
+		return PTR_ERR(peer_mdev);
+	}
+
+	if (!MLX5_CAP_GEN(priv->mdev, hairpin) || !MLX5_CAP_GEN(peer_mdev, hairpin)) {
+		NL_SET_ERR_MSG_MOD(extack, "hairpin is not supported");
+		return -EOPNOTSUPP;
+	}
+
+	peer_id = MLX5_CAP_GEN(peer_mdev, vhca_id);
+	err = mlx5e_hairpin_get_prio(priv, &parse_attr->spec, &match_prio,
+				     extack);
+	if (err)
+		return err;
+
+	mutex_lock(&tc->hairpin_tbl_lock);
+	hpe = mlx5e_hairpin_get(priv, peer_id, match_prio);
+	if (hpe) {
+		mutex_unlock(&tc->hairpin_tbl_lock);
+		wait_for_completion(&hpe->res_ready);
+
+		if (IS_ERR(hpe->hp)) {
+			err = -EREMOTEIO;
+			goto out_err;
+		}
+		goto attach_flow;
+	}
+
+	hpe = kzalloc(sizeof(*hpe), GFP_KERNEL);
+	if (!hpe) {
+		mutex_unlock(&tc->hairpin_tbl_lock);
+		return -ENOMEM;
+	}
+
+	spin_lock_init(&hpe->flows_lock);
+	INIT_LIST_HEAD(&hpe->flows);
+	INIT_LIST_HEAD(&hpe->dead_peer_wait_list);
+	hpe->peer_vhca_id = peer_id;
+	hpe->prio = match_prio;
+	refcount_set(&hpe->refcnt, 1);
+	init_completion(&hpe->res_ready);
+
+	hash_add(tc->hairpin_tbl, &hpe->hairpin_hlist,
+		 hash_hairpin_info(peer_id, match_prio));
+	mutex_unlock(&tc->hairpin_tbl_lock);
+
+	params.log_data_size = 16;
+	params.log_data_size = min_t(u8, params.log_data_size,
+				     MLX5_CAP_GEN(priv->mdev, log_max_hairpin_wq_data_sz));
+	params.log_data_size = max_t(u8, params.log_data_size,
+				     MLX5_CAP_GEN(priv->mdev, log_min_hairpin_wq_data_sz));
+
+	params.log_num_packets = params.log_data_size -
+				 MLX5_MPWRQ_MIN_LOG_STRIDE_SZ(priv->mdev);
+	params.log_num_packets = min_t(u8, params.log_num_packets,
+				       MLX5_CAP_GEN(priv->mdev, log_max_hairpin_num_packets));
+
+	params.q_counter = priv->q_counter;
+	/* set hairpin pair per each 50Gbs share of the link */
+	mlx5e_port_max_linkspeed(priv->mdev, &link_speed);
+	link_speed = max_t(u32, link_speed, 50000);
+	link_speed64 = link_speed;
+	do_div(link_speed64, 50000);
+	params.num_channels = link_speed64;
+
+	hp = mlx5e_hairpin_create(priv, &params, peer_ifindex);
+	hpe->hp = hp;
+	complete_all(&hpe->res_ready);
+	if (IS_ERR(hp)) {
+		err = PTR_ERR(hp);
+		goto out_err;
+	}
+
+	netdev_dbg(priv->netdev, "add hairpin: tirn %x rqn %x peer %s sqn %x prio %d (log) data %d packets %d\n",
+		   mlx5e_tir_get_tirn(&hp->direct_tir), hp->pair->rqn[0],
+		   dev_name(hp->pair->peer_mdev->device),
+		   hp->pair->sqn[0], match_prio, params.log_data_size, params.log_num_packets);
+
+attach_flow:
+	if (hpe->hp->num_channels > 1) {
+		flow_flag_set(flow, HAIRPIN_RSS);
+		flow->attr->nic_attr->hairpin_ft =
+			mlx5_get_ttc_flow_table(hpe->hp->ttc);
+	} else {
+		flow->attr->nic_attr->hairpin_tirn = mlx5e_tir_get_tirn(&hpe->hp->direct_tir);
+	}
+
+	flow->hpe = hpe;
+	spin_lock(&hpe->flows_lock);
+	list_add(&flow->hairpin, &hpe->flows);
+	spin_unlock(&hpe->flows_lock);
+
+	return 0;
+
+out_err:
+	mlx5e_hairpin_put(priv, hpe);
+	return err;
+}
+
+static void mlx5e_hairpin_flow_del(struct mlx5e_priv *priv,
+				   struct mlx5e_tc_flow *flow)
+{
+	/* flow wasn't fully initialized */
+	if (!flow->hpe)
+		return;
+
+	spin_lock(&flow->hpe->flows_lock);
+	list_del(&flow->hairpin);
+	spin_unlock(&flow->hpe->flows_lock);
+
+	mlx5e_hairpin_put(priv, flow->hpe);
+	flow->hpe = NULL;
+}
+
+struct mlx5_flow_handle *
+mlx5e_add_offloaded_nic_rule(struct mlx5e_priv *priv,
+			     struct mlx5_flow_spec *spec,
+			     struct mlx5_flow_attr *attr)
+{
+	struct mlx5_flow_context *flow_context = &spec->flow_context;
+	struct mlx5e_vlan_table *vlan = mlx5e_fs_get_vlan(priv->fs);
+	struct mlx5e_tc_table *tc = mlx5e_fs_get_tc(priv->fs);
+	struct mlx5_nic_flow_attr *nic_attr = attr->nic_attr;
+	struct mlx5_flow_destination dest[2] = {};
+	struct mlx5_fs_chains *nic_chains;
+	struct mlx5_flow_act flow_act = {
+		.action = attr->action,
+		.flags    = FLOW_ACT_NO_APPEND,
+	};
+	struct mlx5_flow_handle *rule;
+	struct mlx5_flow_table *ft;
+	int dest_ix = 0;
+
+	nic_chains = mlx5e_nic_chains(tc);
+	flow_context->flags |= FLOW_CONTEXT_HAS_TAG;
+	flow_context->flow_tag = nic_attr->flow_tag;
+
+	if (attr->dest_ft) {
+		dest[dest_ix].type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
+		dest[dest_ix].ft = attr->dest_ft;
+		dest_ix++;
+	} else if (nic_attr->hairpin_ft) {
+		dest[dest_ix].type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
+		dest[dest_ix].ft = nic_attr->hairpin_ft;
+		dest_ix++;
+	} else if (nic_attr->hairpin_tirn) {
+		dest[dest_ix].type = MLX5_FLOW_DESTINATION_TYPE_TIR;
+		dest[dest_ix].tir_num = nic_attr->hairpin_tirn;
+		dest_ix++;
+	} else if (attr->action & MLX5_FLOW_CONTEXT_ACTION_FWD_DEST) {
+		dest[dest_ix].type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
+		if (attr->dest_chain) {
+			dest[dest_ix].ft = mlx5_chains_get_table(nic_chains,
+								 attr->dest_chain, 1,
+								 MLX5E_TC_FT_LEVEL);
+			if (IS_ERR(dest[dest_ix].ft))
+				return ERR_CAST(dest[dest_ix].ft);
+		} else {
+			dest[dest_ix].ft = mlx5e_vlan_get_flowtable(vlan);
+		}
+		dest_ix++;
+	}
+
+	if (dest[0].type == MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE &&
+	    MLX5_CAP_FLOWTABLE_NIC_RX(priv->mdev, ignore_flow_level))
+		flow_act.flags |= FLOW_ACT_IGNORE_FLOW_LEVEL;
+
+	if (flow_act.action & MLX5_FLOW_CONTEXT_ACTION_COUNT) {
+		dest[dest_ix].type = MLX5_FLOW_DESTINATION_TYPE_COUNTER;
+		dest[dest_ix].counter_id = mlx5_fc_id(attr->counter);
+		dest_ix++;
+	}
+
+	if (attr->action & MLX5_FLOW_CONTEXT_ACTION_MOD_HDR)
+		flow_act.modify_hdr = attr->modify_hdr;
+
+	mutex_lock(&tc->t_lock);
+	if (IS_ERR_OR_NULL(tc->t)) {
+		/* Create the root table here if doesn't exist yet */
+		tc->t =
+			mlx5_chains_get_table(nic_chains, 0, 1, MLX5E_TC_FT_LEVEL);
+
+		if (IS_ERR(tc->t)) {
+			mutex_unlock(&tc->t_lock);
+			netdev_err(priv->netdev,
+				   "Failed to create tc offload table\n");
+			rule = ERR_CAST(tc->t);
+			goto err_ft_get;
+		}
+	}
+	mutex_unlock(&tc->t_lock);
+
+	if (attr->chain || attr->prio)
+		ft = mlx5_chains_get_table(nic_chains,
+					   attr->chain, attr->prio,
+					   MLX5E_TC_FT_LEVEL);
+	else
+		ft = attr->ft;
+
+	if (IS_ERR(ft)) {
+		rule = ERR_CAST(ft);
+		goto err_ft_get;
+	}
+
+	if (attr->outer_match_level != MLX5_MATCH_NONE)
+		spec->match_criteria_enable |= MLX5_MATCH_OUTER_HEADERS;
+
+	rule = mlx5_add_flow_rules(ft, spec,
+				   &flow_act, dest, dest_ix);
+	if (IS_ERR(rule))
+		goto err_rule;
+
+	return rule;
+
+err_rule:
+	if (attr->chain || attr->prio)
+		mlx5_chains_put_table(nic_chains,
+				      attr->chain, attr->prio,
+				      MLX5E_TC_FT_LEVEL);
+err_ft_get:
+	if (attr->dest_chain)
+		mlx5_chains_put_table(nic_chains,
+				      attr->dest_chain, 1,
+				      MLX5E_TC_FT_LEVEL);
+
+	return ERR_CAST(rule);
+}
+
+static int
+alloc_flow_attr_counter(struct mlx5_core_dev *counter_dev,
+			struct mlx5_flow_attr *attr)
+
+{
+	struct mlx5_fc *counter;
+
+	counter = mlx5_fc_create(counter_dev, true);
+	if (IS_ERR(counter))
+		return PTR_ERR(counter);
+
+	attr->counter = counter;
+	return 0;
+}
+
+static int
+mlx5e_tc_add_nic_flow(struct mlx5e_priv *priv,
+		      struct mlx5e_tc_flow *flow,
+		      struct netlink_ext_ack *extack)
+{
+	struct mlx5e_tc_flow_parse_attr *parse_attr;
+	struct mlx5_flow_attr *attr = flow->attr;
+	struct mlx5_core_dev *dev = priv->mdev;
+	int err;
+
+	parse_attr = attr->parse_attr;
+
+	if (flow_flag_test(flow, HAIRPIN)) {
+		err = mlx5e_hairpin_flow_add(priv, flow, parse_attr, extack);
+		if (err)
+			return err;
+	}
+
+	if (attr->action & MLX5_FLOW_CONTEXT_ACTION_COUNT) {
+		err = alloc_flow_attr_counter(dev, attr);
+		if (err)
+			return err;
+	}
+
+	if (attr->action & MLX5_FLOW_CONTEXT_ACTION_MOD_HDR) {
+		err = mlx5e_attach_mod_hdr(priv, flow, parse_attr);
+		if (err)
+			return err;
+	}
+
+	if (attr->flags & MLX5_ATTR_FLAG_CT)
+		flow->rule[0] = mlx5_tc_ct_flow_offload(get_ct_priv(priv), &parse_attr->spec,
+							attr, &parse_attr->mod_hdr_acts);
+	else
+		flow->rule[0] = mlx5e_add_offloaded_nic_rule(priv, &parse_attr->spec,
+							     attr);
+
+	return PTR_ERR_OR_ZERO(flow->rule[0]);
+}
+
+void mlx5e_del_offloaded_nic_rule(struct mlx5e_priv *priv,
+				  struct mlx5_flow_handle *rule,
+				  struct mlx5_flow_attr *attr)
+{
+	struct mlx5e_tc_table *tc = mlx5e_fs_get_tc(priv->fs);
+	struct mlx5_fs_chains *nic_chains;
+
+	nic_chains = mlx5e_nic_chains(tc);
+	mlx5_del_flow_rules(rule);
+
+	if (attr->chain || attr->prio)
+		mlx5_chains_put_table(nic_chains, attr->chain, attr->prio,
+				      MLX5E_TC_FT_LEVEL);
+
+	if (attr->dest_chain)
+		mlx5_chains_put_table(nic_chains, attr->dest_chain, 1,
+				      MLX5E_TC_FT_LEVEL);
+}
+
+static void mlx5e_tc_del_nic_flow(struct mlx5e_priv *priv,
+				  struct mlx5e_tc_flow *flow)
+{
+	struct mlx5e_tc_table *tc = mlx5e_fs_get_tc(priv->fs);
+	struct mlx5_flow_attr *attr = flow->attr;
+
+	flow_flag_clear(flow, OFFLOADED);
+
+	if (attr->flags & MLX5_ATTR_FLAG_CT)
+		mlx5_tc_ct_delete_flow(get_ct_priv(flow->priv), attr);
+	else if (!IS_ERR_OR_NULL(flow->rule[0]))
+		mlx5e_del_offloaded_nic_rule(priv, flow->rule[0], attr);
+
+	/* Remove root table if no rules are left to avoid
+	 * extra steering hops.
+	 */
+	mutex_lock(&tc->t_lock);
+	if (!mlx5e_tc_num_filters(priv, MLX5_TC_FLAG(NIC_OFFLOAD)) &&
+	    !IS_ERR_OR_NULL(tc->t)) {
+		mlx5_chains_put_table(mlx5e_nic_chains(tc), 0, 1, MLX5E_TC_FT_LEVEL);
+		tc->t = NULL;
+	}
+	mutex_unlock(&tc->t_lock);
+
+	if (attr->action & MLX5_FLOW_CONTEXT_ACTION_MOD_HDR) {
+		mlx5e_mod_hdr_dealloc(&attr->parse_attr->mod_hdr_acts);
+		mlx5e_detach_mod_hdr(priv, flow);
+	}
+
+	if (attr->action & MLX5_FLOW_CONTEXT_ACTION_COUNT)
+		mlx5_fc_destroy(priv->mdev, attr->counter);
+
+	if (flow_flag_test(flow, HAIRPIN))
+		mlx5e_hairpin_flow_del(priv, flow);
+
+	free_flow_post_acts(flow);
+
+	kvfree(attr->parse_attr);
+	kfree(flow->attr);
+}
+
+struct mlx5_flow_handle *
+mlx5e_tc_offload_fdb_rules(struct mlx5_eswitch *esw,
+			   struct mlx5e_tc_flow *flow,
+			   struct mlx5_flow_spec *spec,
+			   struct mlx5_flow_attr *attr)
+{
+	struct mlx5_flow_handle *rule;
+
+	if (attr->flags & MLX5_ATTR_FLAG_SLOW_PATH)
+		return mlx5_eswitch_add_offloaded_rule(esw, spec, attr);
+
+	rule = mlx5e_tc_rule_offload(flow->priv, spec, attr);
+
+	if (IS_ERR(rule))
+		return rule;
+
+	if (attr->esw_attr->split_count) {
+		flow->rule[1] = mlx5_eswitch_add_fwd_rule(esw, spec, attr);
+		if (IS_ERR(flow->rule[1]))
+			goto err_rule1;
+	}
+
+	return rule;
+
+err_rule1:
+	mlx5e_tc_rule_unoffload(flow->priv, rule, attr);
+	return flow->rule[1];
+}
+
+void mlx5e_tc_unoffload_fdb_rules(struct mlx5_eswitch *esw,
+				  struct mlx5e_tc_flow *flow,
+				  struct mlx5_flow_attr *attr)
+{
+	flow_flag_clear(flow, OFFLOADED);
+
+	if (attr->flags & MLX5_ATTR_FLAG_SLOW_PATH)
+		return mlx5_eswitch_del_offloaded_rule(esw, flow->rule[0], attr);
+
+	if (attr->esw_attr->split_count)
+		mlx5_eswitch_del_fwd_rule(esw, flow->rule[1], attr);
+
+	mlx5e_tc_rule_unoffload(flow->priv, flow->rule[0], attr);
+}
+
+struct mlx5_flow_handle *
+mlx5e_tc_offload_to_slow_path(struct mlx5_eswitch *esw,
+			      struct mlx5e_tc_flow *flow,
+			      struct mlx5_flow_spec *spec)
+{
+	struct mlx5e_tc_mod_hdr_acts mod_acts = {};
+	struct mlx5e_mod_hdr_handle *mh = NULL;
+	struct mlx5_flow_attr *slow_attr;
+	struct mlx5_flow_handle *rule;
+	bool fwd_and_modify_cap;
+	u32 chain_mapping = 0;
+	int err;
+
+	slow_attr = mlx5_alloc_flow_attr(MLX5_FLOW_NAMESPACE_FDB);
+	if (!slow_attr)
+		return ERR_PTR(-ENOMEM);
+
+	memcpy(slow_attr, flow->attr, ESW_FLOW_ATTR_SZ);
+	slow_attr->action = MLX5_FLOW_CONTEXT_ACTION_FWD_DEST;
+	slow_attr->esw_attr->split_count = 0;
+	slow_attr->flags |= MLX5_ATTR_FLAG_SLOW_PATH;
+
+	fwd_and_modify_cap = MLX5_CAP_ESW_FLOWTABLE((esw)->dev, fdb_modify_header_fwd_to_table);
+	if (!fwd_and_modify_cap)
+		goto skip_restore;
+
+	err = mlx5_chains_get_chain_mapping(esw_chains(esw), flow->attr->chain, &chain_mapping);
+	if (err)
+		goto err_get_chain;
+
+	err = mlx5e_tc_match_to_reg_set(esw->dev, &mod_acts, MLX5_FLOW_NAMESPACE_FDB,
+					CHAIN_TO_REG, chain_mapping);
+	if (err)
+		goto err_reg_set;
+
+	mh = mlx5e_mod_hdr_attach(esw->dev, get_mod_hdr_table(flow->priv, flow),
+				  MLX5_FLOW_NAMESPACE_FDB, &mod_acts);
+	if (IS_ERR(mh)) {
+		err = PTR_ERR(mh);
+		goto err_attach;
+	}
+
+	slow_attr->action |= MLX5_FLOW_CONTEXT_ACTION_MOD_HDR;
+	slow_attr->modify_hdr = mlx5e_mod_hdr_get(mh);
+
+skip_restore:
+	rule = mlx5e_tc_offload_fdb_rules(esw, flow, spec, slow_attr);
+	if (IS_ERR(rule)) {
+		err = PTR_ERR(rule);
+		goto err_offload;
+	}
+
+	flow->slow_mh = mh;
+	flow->chain_mapping = chain_mapping;
+	flow_flag_set(flow, SLOW);
+
+	mlx5e_mod_hdr_dealloc(&mod_acts);
+	kfree(slow_attr);
+
+	return rule;
+
+err_offload:
+	if (fwd_and_modify_cap)
+		mlx5e_mod_hdr_detach(esw->dev, get_mod_hdr_table(flow->priv, flow), mh);
+err_attach:
+err_reg_set:
+	if (fwd_and_modify_cap)
+		mlx5_chains_put_chain_mapping(esw_chains(esw), chain_mapping);
+err_get_chain:
+	mlx5e_mod_hdr_dealloc(&mod_acts);
+	kfree(slow_attr);
+	return ERR_PTR(err);
+}
+
+void mlx5e_tc_unoffload_from_slow_path(struct mlx5_eswitch *esw,
+				       struct mlx5e_tc_flow *flow)
+{
+	struct mlx5_flow_attr *slow_attr;
+
+	slow_attr = mlx5_alloc_flow_attr(MLX5_FLOW_NAMESPACE_FDB);
+	if (!slow_attr) {
+		mlx5_core_warn(flow->priv->mdev, "Unable to alloc attr to unoffload slow path rule\n");
+		return;
+	}
+
+	memcpy(slow_attr, flow->attr, ESW_FLOW_ATTR_SZ);
+	slow_attr->action = MLX5_FLOW_CONTEXT_ACTION_FWD_DEST;
+	slow_attr->esw_attr->split_count = 0;
+	slow_attr->flags |= MLX5_ATTR_FLAG_SLOW_PATH;
+	if (flow->slow_mh) {
+		slow_attr->action |= MLX5_FLOW_CONTEXT_ACTION_MOD_HDR;
+		slow_attr->modify_hdr = mlx5e_mod_hdr_get(flow->slow_mh);
+	}
+	mlx5e_tc_unoffload_fdb_rules(esw, flow, slow_attr);
+	if (flow->slow_mh) {
+		mlx5e_mod_hdr_detach(esw->dev, get_mod_hdr_table(flow->priv, flow), flow->slow_mh);
+		mlx5_chains_put_chain_mapping(esw_chains(esw), flow->chain_mapping);
+		flow->chain_mapping = 0;
+		flow->slow_mh = NULL;
+	}
+	flow_flag_clear(flow, SLOW);
+	kfree(slow_attr);
+}
+
+/* Caller must obtain uplink_priv->unready_flows_lock mutex before calling this
+ * function.
+ */
+static void unready_flow_add(struct mlx5e_tc_flow *flow,
+			     struct list_head *unready_flows)
+{
+	flow_flag_set(flow, NOT_READY);
+	list_add_tail(&flow->unready, unready_flows);
+}
+
+/* Caller must obtain uplink_priv->unready_flows_lock mutex before calling this
+ * function.
+ */
+static void unready_flow_del(struct mlx5e_tc_flow *flow)
+{
+	list_del(&flow->unready);
+	flow_flag_clear(flow, NOT_READY);
+}
+
+static void add_unready_flow(struct mlx5e_tc_flow *flow)
+{
+	struct mlx5_rep_uplink_priv *uplink_priv;
+	struct mlx5e_rep_priv *rpriv;
+	struct mlx5_eswitch *esw;
+
+	esw = flow->priv->mdev->priv.eswitch;
+	rpriv = mlx5_eswitch_get_uplink_priv(esw, REP_ETH);
+	uplink_priv = &rpriv->uplink_priv;
+
+	mutex_lock(&uplink_priv->unready_flows_lock);
+	unready_flow_add(flow, &uplink_priv->unready_flows);
+	mutex_unlock(&uplink_priv->unready_flows_lock);
+}
+
+static void remove_unready_flow(struct mlx5e_tc_flow *flow)
+{
+	struct mlx5_rep_uplink_priv *uplink_priv;
+	struct mlx5e_rep_priv *rpriv;
+	struct mlx5_eswitch *esw;
+
+	esw = flow->priv->mdev->priv.eswitch;
+	rpriv = mlx5_eswitch_get_uplink_priv(esw, REP_ETH);
+	uplink_priv = &rpriv->uplink_priv;
+
+	mutex_lock(&uplink_priv->unready_flows_lock);
+	if (flow_flag_test(flow, NOT_READY))
+		unready_flow_del(flow);
+	mutex_unlock(&uplink_priv->unready_flows_lock);
+}
+
+bool mlx5e_tc_is_vf_tunnel(struct net_device *out_dev, struct net_device *route_dev)
+{
+	struct mlx5_core_dev *out_mdev, *route_mdev;
+	struct mlx5e_priv *out_priv, *route_priv;
+
+	out_priv = netdev_priv(out_dev);
+	out_mdev = out_priv->mdev;
+	route_priv = netdev_priv(route_dev);
+	route_mdev = route_priv->mdev;
+
+	if (out_mdev->coredev_type != MLX5_COREDEV_PF)
+		return false;
+
+	if (route_mdev->coredev_type != MLX5_COREDEV_VF &&
+	    route_mdev->coredev_type != MLX5_COREDEV_SF)
+		return false;
+
+	return mlx5e_same_hw_devs(out_priv, route_priv);
+}
+
+int mlx5e_tc_query_route_vport(struct net_device *out_dev, struct net_device *route_dev, u16 *vport)
+{
+	struct mlx5e_priv *out_priv, *route_priv;
+	struct mlx5_core_dev *route_mdev;
+	struct mlx5_eswitch *esw;
+	u16 vhca_id;
+
+	out_priv = netdev_priv(out_dev);
+	esw = out_priv->mdev->priv.eswitch;
+	route_priv = netdev_priv(route_dev);
+	route_mdev = route_priv->mdev;
+
+	vhca_id = MLX5_CAP_GEN(route_mdev, vhca_id);
+	if (mlx5_lag_is_active(out_priv->mdev)) {
+		struct mlx5_devcom *devcom;
+		int err;
+
+		/* In lag case we may get devices from different eswitch instances.
+		 * If we failed to get vport num, it means, mostly, that we on the wrong
+		 * eswitch.
+		 */
+		err = mlx5_eswitch_vhca_id_to_vport(esw, vhca_id, vport);
+		if (err != -ENOENT)
+			return err;
+
+		rcu_read_lock();
+		devcom = out_priv->mdev->priv.devcom;
+		esw = mlx5_devcom_get_peer_data_rcu(devcom, MLX5_DEVCOM_ESW_OFFLOADS);
+		err = esw ? mlx5_eswitch_vhca_id_to_vport(esw, vhca_id, vport) : -ENODEV;
+		rcu_read_unlock();
+
+		return err;
+	}
+
+	return mlx5_eswitch_vhca_id_to_vport(esw, vhca_id, vport);
+}
+
+int mlx5e_tc_add_flow_mod_hdr(struct mlx5e_priv *priv,
+			      struct mlx5e_tc_flow *flow,
+			      struct mlx5_flow_attr *attr)
+{
+	struct mlx5e_tc_mod_hdr_acts *mod_hdr_acts = &attr->parse_attr->mod_hdr_acts;
+	struct mlx5_modify_hdr *mod_hdr;
+
+	mod_hdr = mlx5_modify_header_alloc(priv->mdev,
+					   mlx5e_get_flow_namespace(flow),
+					   mod_hdr_acts->num_actions,
+					   mod_hdr_acts->actions);
+	if (IS_ERR(mod_hdr))
+		return PTR_ERR(mod_hdr);
+
+	WARN_ON(attr->modify_hdr);
+	attr->modify_hdr = mod_hdr;
+
+	return 0;
+}
+
+static int
+set_encap_dests(struct mlx5e_priv *priv,
+		struct mlx5e_tc_flow *flow,
+		struct mlx5_flow_attr *attr,
+		struct netlink_ext_ack *extack,
+		bool *vf_tun)
+{
+	struct mlx5e_tc_flow_parse_attr *parse_attr;
+	struct mlx5_esw_flow_attr *esw_attr;
+	struct net_device *encap_dev = NULL;
+	struct mlx5e_rep_priv *rpriv;
+	struct mlx5e_priv *out_priv;
+	int out_index;
+	int err = 0;
+
+	if (!mlx5e_is_eswitch_flow(flow))
+		return 0;
+
+	parse_attr = attr->parse_attr;
+	esw_attr = attr->esw_attr;
+	*vf_tun = false;
+
+	for (out_index = 0; out_index < MLX5_MAX_FLOW_FWD_VPORTS; out_index++) {
+		struct net_device *out_dev;
+		int mirred_ifindex;
+
+		if (!(esw_attr->dests[out_index].flags & MLX5_ESW_DEST_ENCAP))
+			continue;
+
+		mirred_ifindex = parse_attr->mirred_ifindex[out_index];
+		out_dev = dev_get_by_index(dev_net(priv->netdev), mirred_ifindex);
+		if (!out_dev) {
+			NL_SET_ERR_MSG_MOD(extack, "Requested mirred device not found");
+			err = -ENODEV;
+			goto out;
+		}
+		err = mlx5e_attach_encap(priv, flow, attr, out_dev, out_index,
+					 extack, &encap_dev);
+		dev_put(out_dev);
+		if (err)
+			goto out;
+
+		if (esw_attr->dests[out_index].flags &
+		    MLX5_ESW_DEST_CHAIN_WITH_SRC_PORT_CHANGE &&
+		    !esw_attr->dest_int_port)
+			*vf_tun = true;
+
+		out_priv = netdev_priv(encap_dev);
+		rpriv = out_priv->ppriv;
+		esw_attr->dests[out_index].rep = rpriv->rep;
+		esw_attr->dests[out_index].mdev = out_priv->mdev;
+	}
+
+	if (*vf_tun && esw_attr->out_count > 1) {
+		NL_SET_ERR_MSG_MOD(extack, "VF tunnel encap with mirroring is not supported");
+		err = -EOPNOTSUPP;
+		goto out;
+	}
+
+out:
+	return err;
+}
+
+static void
+clean_encap_dests(struct mlx5e_priv *priv,
+		  struct mlx5e_tc_flow *flow,
+		  struct mlx5_flow_attr *attr,
+		  bool *vf_tun)
+{
+	struct mlx5_esw_flow_attr *esw_attr;
+	int out_index;
+
+	if (!mlx5e_is_eswitch_flow(flow))
+		return;
+
+	esw_attr = attr->esw_attr;
+	*vf_tun = false;
+
+	for (out_index = 0; out_index < MLX5_MAX_FLOW_FWD_VPORTS; out_index++) {
+		if (!(esw_attr->dests[out_index].flags & MLX5_ESW_DEST_ENCAP))
+			continue;
+
+		if (esw_attr->dests[out_index].flags &
+		    MLX5_ESW_DEST_CHAIN_WITH_SRC_PORT_CHANGE &&
+		    !esw_attr->dest_int_port)
+			*vf_tun = true;
+
+		mlx5e_detach_encap(priv, flow, attr, out_index);
+		kfree(attr->parse_attr->tun_info[out_index]);
+	}
+}
+
+static int
+mlx5e_tc_add_fdb_flow(struct mlx5e_priv *priv,
+		      struct mlx5e_tc_flow *flow,
+		      struct netlink_ext_ack *extack)
+{
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	struct mlx5e_tc_flow_parse_attr *parse_attr;
+	struct mlx5_flow_attr *attr = flow->attr;
+	struct mlx5_esw_flow_attr *esw_attr;
+	u32 max_prio, max_chain;
+	bool vf_tun;
+	int err = 0;
+
+	parse_attr = attr->parse_attr;
+	esw_attr = attr->esw_attr;
+
+	/* We check chain range only for tc flows.
+	 * For ft flows, we checked attr->chain was originally 0 and set it to
+	 * FDB_FT_CHAIN which is outside tc range.
+	 * See mlx5e_rep_setup_ft_cb().
+	 */
+	max_chain = mlx5_chains_get_chain_range(esw_chains(esw));
+	if (!mlx5e_is_ft_flow(flow) && attr->chain > max_chain) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "Requested chain is out of supported range");
+		err = -EOPNOTSUPP;
+		goto err_out;
+	}
+
+	max_prio = mlx5_chains_get_prio_range(esw_chains(esw));
+	if (attr->prio > max_prio) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "Requested priority is out of supported range");
+		err = -EOPNOTSUPP;
+		goto err_out;
+	}
+
+	if (flow_flag_test(flow, TUN_RX)) {
+		err = mlx5e_attach_decap_route(priv, flow);
+		if (err)
+			goto err_out;
+
+		if (!attr->chain && esw_attr->int_port &&
+		    attr->action & MLX5_FLOW_CONTEXT_ACTION_FWD_DEST) {
+			/* If decap route device is internal port, change the
+			 * source vport value in reg_c0 back to uplink just in
+			 * case the rule performs goto chain > 0. If we have a miss
+			 * on chain > 0 we want the metadata regs to hold the
+			 * chain id so SW will resume handling of this packet
+			 * from the proper chain.
+			 */
+			u32 metadata = mlx5_eswitch_get_vport_metadata_for_set(esw,
+									esw_attr->in_rep->vport);
+
+			err = mlx5e_tc_match_to_reg_set(priv->mdev, &parse_attr->mod_hdr_acts,
+							MLX5_FLOW_NAMESPACE_FDB, VPORT_TO_REG,
+							metadata);
+			if (err)
+				goto err_out;
+
+			attr->action |= MLX5_FLOW_CONTEXT_ACTION_MOD_HDR;
+		}
+	}
+
+	if (flow_flag_test(flow, L3_TO_L2_DECAP)) {
+		err = mlx5e_attach_decap(priv, flow, extack);
+		if (err)
+			goto err_out;
+	}
+
+	if (netif_is_ovs_master(parse_attr->filter_dev)) {
+		struct mlx5e_tc_int_port *int_port;
+
+		if (attr->chain) {
+			NL_SET_ERR_MSG_MOD(extack,
+					   "Internal port rule is only supported on chain 0");
+			err = -EOPNOTSUPP;
+			goto err_out;
+		}
+
+		if (attr->dest_chain) {
+			NL_SET_ERR_MSG_MOD(extack,
+					   "Internal port rule offload doesn't support goto action");
+			err = -EOPNOTSUPP;
+			goto err_out;
+		}
+
+		int_port = mlx5e_tc_int_port_get(mlx5e_get_int_port_priv(priv),
+						 parse_attr->filter_dev->ifindex,
+						 flow_flag_test(flow, EGRESS) ?
+						 MLX5E_TC_INT_PORT_EGRESS :
+						 MLX5E_TC_INT_PORT_INGRESS);
+		if (IS_ERR(int_port)) {
+			err = PTR_ERR(int_port);
+			goto err_out;
+		}
+
+		esw_attr->int_port = int_port;
+	}
+
+	err = set_encap_dests(priv, flow, attr, extack, &vf_tun);
+	if (err)
+		goto err_out;
+
+	err = mlx5_eswitch_add_vlan_action(esw, attr);
+	if (err)
+		goto err_out;
+
+	if (attr->action & MLX5_FLOW_CONTEXT_ACTION_MOD_HDR) {
+		if (vf_tun) {
+			err = mlx5e_tc_add_flow_mod_hdr(priv, flow, attr);
+			if (err)
+				goto err_out;
+		} else {
+			err = mlx5e_attach_mod_hdr(priv, flow, parse_attr);
+			if (err)
+				goto err_out;
+		}
+	}
+
+	if (attr->action & MLX5_FLOW_CONTEXT_ACTION_COUNT) {
+		err = alloc_flow_attr_counter(esw_attr->counter_dev, attr);
+		if (err)
+			goto err_out;
+	}
+
+	/* we get here if one of the following takes place:
+	 * (1) there's no error
+	 * (2) there's an encap action and we don't have valid neigh
+	 */
+	if (flow_flag_test(flow, SLOW))
+		flow->rule[0] = mlx5e_tc_offload_to_slow_path(esw, flow, &parse_attr->spec);
+	else
+		flow->rule[0] = mlx5e_tc_offload_fdb_rules(esw, flow, &parse_attr->spec, attr);
+
+	if (IS_ERR(flow->rule[0])) {
+		err = PTR_ERR(flow->rule[0]);
+		goto err_out;
+	}
+	flow_flag_set(flow, OFFLOADED);
+
+	return 0;
+
+err_out:
+	flow_flag_set(flow, FAILED);
+	return err;
+}
+
+static bool mlx5_flow_has_geneve_opt(struct mlx5e_tc_flow *flow)
+{
+	struct mlx5_flow_spec *spec = &flow->attr->parse_attr->spec;
+	void *headers_v = MLX5_ADDR_OF(fte_match_param,
+				       spec->match_value,
+				       misc_parameters_3);
+	u32 geneve_tlv_opt_0_data = MLX5_GET(fte_match_set_misc3,
+					     headers_v,
+					     geneve_tlv_option_0_data);
+
+	return !!geneve_tlv_opt_0_data;
+}
+
+static void mlx5e_tc_del_fdb_flow(struct mlx5e_priv *priv,
+				  struct mlx5e_tc_flow *flow)
+{
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	struct mlx5_flow_attr *attr = flow->attr;
+	struct mlx5_esw_flow_attr *esw_attr;
+	bool vf_tun;
+
+	esw_attr = attr->esw_attr;
+	mlx5e_put_flow_tunnel_id(flow);
+
+	remove_unready_flow(flow);
+
+	if (mlx5e_is_offloaded_flow(flow)) {
+		if (flow_flag_test(flow, SLOW))
+			mlx5e_tc_unoffload_from_slow_path(esw, flow);
+		else
+			mlx5e_tc_unoffload_fdb_rules(esw, flow, attr);
+	}
+	complete_all(&flow->del_hw_done);
+
+	if (mlx5_flow_has_geneve_opt(flow))
+		mlx5_geneve_tlv_option_del(priv->mdev->geneve);
+
+	mlx5_eswitch_del_vlan_action(esw, attr);
+
+	if (flow->decap_route)
+		mlx5e_detach_decap_route(priv, flow);
+
+	clean_encap_dests(priv, flow, attr, &vf_tun);
+
+	mlx5_tc_ct_match_del(get_ct_priv(priv), &flow->attr->ct_attr);
+
+	if (attr->action & MLX5_FLOW_CONTEXT_ACTION_MOD_HDR) {
+		mlx5e_mod_hdr_dealloc(&attr->parse_attr->mod_hdr_acts);
+		if (vf_tun && attr->modify_hdr)
+			mlx5_modify_header_dealloc(priv->mdev, attr->modify_hdr);
+		else
+			mlx5e_detach_mod_hdr(priv, flow);
+	}
+
+	if (attr->action & MLX5_FLOW_CONTEXT_ACTION_COUNT)
+		mlx5_fc_destroy(esw_attr->counter_dev, attr->counter);
+
+	if (esw_attr->int_port)
+		mlx5e_tc_int_port_put(mlx5e_get_int_port_priv(priv), esw_attr->int_port);
+
+	if (esw_attr->dest_int_port)
+		mlx5e_tc_int_port_put(mlx5e_get_int_port_priv(priv), esw_attr->dest_int_port);
+
+	if (flow_flag_test(flow, L3_TO_L2_DECAP))
+		mlx5e_detach_decap(priv, flow);
+
+	free_flow_post_acts(flow);
+
+	if (flow->attr->lag.count)
+		mlx5_lag_del_mpesw_rule(esw->dev);
+
+	kvfree(attr->esw_attr->rx_tun_attr);
+	kvfree(attr->parse_attr);
+	kfree(flow->attr);
+}
+
+struct mlx5_fc *mlx5e_tc_get_counter(struct mlx5e_tc_flow *flow)
+{
+	struct mlx5_flow_attr *attr;
+
+	attr = list_first_entry(&flow->attrs, struct mlx5_flow_attr, list);
+	return attr->counter;
+}
+
+/* Iterate over tmp_list of flows attached to flow_list head. */
+void mlx5e_put_flow_list(struct mlx5e_priv *priv, struct list_head *flow_list)
+{
+	struct mlx5e_tc_flow *flow, *tmp;
+
+	list_for_each_entry_safe(flow, tmp, flow_list, tmp_list)
+		mlx5e_flow_put(priv, flow);
+}
+
+static void __mlx5e_tc_del_fdb_peer_flow(struct mlx5e_tc_flow *flow)
+{
+	struct mlx5_eswitch *esw = flow->priv->mdev->priv.eswitch;
+
+	if (!flow_flag_test(flow, ESWITCH) ||
+	    !flow_flag_test(flow, DUP))
+		return;
+
+	mutex_lock(&esw->offloads.peer_mutex);
+	list_del(&flow->peer);
+	mutex_unlock(&esw->offloads.peer_mutex);
+
+	flow_flag_clear(flow, DUP);
+
+	if (refcount_dec_and_test(&flow->peer_flow->refcnt)) {
+		mlx5e_tc_del_fdb_flow(flow->peer_flow->priv, flow->peer_flow);
+		kfree(flow->peer_flow);
+	}
+
+	flow->peer_flow = NULL;
+}
+
+static void mlx5e_tc_del_fdb_peer_flow(struct mlx5e_tc_flow *flow)
+{
+	struct mlx5_core_dev *dev = flow->priv->mdev;
+	struct mlx5_devcom *devcom = dev->priv.devcom;
+	struct mlx5_eswitch *peer_esw;
+
+	peer_esw = mlx5_devcom_get_peer_data(devcom, MLX5_DEVCOM_ESW_OFFLOADS);
+	if (!peer_esw)
+		return;
+
+	__mlx5e_tc_del_fdb_peer_flow(flow);
+	mlx5_devcom_release_peer_data(devcom, MLX5_DEVCOM_ESW_OFFLOADS);
+}
+
+static void mlx5e_tc_del_flow(struct mlx5e_priv *priv,
+			      struct mlx5e_tc_flow *flow)
+{
+	if (mlx5e_is_eswitch_flow(flow)) {
+		mlx5e_tc_del_fdb_peer_flow(flow);
+		mlx5e_tc_del_fdb_flow(priv, flow);
+	} else {
+		mlx5e_tc_del_nic_flow(priv, flow);
+	}
+}
+
+static bool flow_requires_tunnel_mapping(u32 chain, struct flow_cls_offload *f)
+{
+	struct flow_rule *rule = flow_cls_offload_flow_rule(f);
+	struct flow_action *flow_action = &rule->action;
+	const struct flow_action_entry *act;
+	int i;
+
+	if (chain)
+		return false;
+
+	flow_action_for_each(i, act, flow_action) {
+		switch (act->id) {
+		case FLOW_ACTION_GOTO:
+			return true;
+		case FLOW_ACTION_SAMPLE:
+			return true;
+		default:
+			continue;
+		}
+	}
+
+	return false;
+}
+
+static int
+enc_opts_is_dont_care_or_full_match(struct mlx5e_priv *priv,
+				    struct flow_dissector_key_enc_opts *opts,
+				    struct netlink_ext_ack *extack,
+				    bool *dont_care)
+{
+	struct geneve_opt *opt;
+	int off = 0;
+
+	*dont_care = true;
+
+	while (opts->len > off) {
+		opt = (struct geneve_opt *)&opts->data[off];
+
+		if (!(*dont_care) || opt->opt_class || opt->type ||
+		    memchr_inv(opt->opt_data, 0, opt->length * 4)) {
+			*dont_care = false;
+
+			if (opt->opt_class != htons(U16_MAX) ||
+			    opt->type != U8_MAX) {
+				NL_SET_ERR_MSG_MOD(extack,
+						   "Partial match of tunnel options in chain > 0 isn't supported");
+				netdev_warn(priv->netdev,
+					    "Partial match of tunnel options in chain > 0 isn't supported");
+				return -EOPNOTSUPP;
+			}
+		}
+
+		off += sizeof(struct geneve_opt) + opt->length * 4;
+	}
+
+	return 0;
+}
+
+#define COPY_DISSECTOR(rule, diss_key, dst)\
+({ \
+	struct flow_rule *__rule = (rule);\
+	typeof(dst) __dst = dst;\
+\
+	memcpy(__dst,\
+	       skb_flow_dissector_target(__rule->match.dissector,\
+					 diss_key,\
+					 __rule->match.key),\
+	       sizeof(*__dst));\
+})
+
+static int mlx5e_get_flow_tunnel_id(struct mlx5e_priv *priv,
+				    struct mlx5e_tc_flow *flow,
+				    struct flow_cls_offload *f,
+				    struct net_device *filter_dev)
+{
+	struct flow_rule *rule = flow_cls_offload_flow_rule(f);
+	struct netlink_ext_ack *extack = f->common.extack;
+	struct mlx5e_tc_mod_hdr_acts *mod_hdr_acts;
+	struct flow_match_enc_opts enc_opts_match;
+	struct tunnel_match_enc_opts tun_enc_opts;
+	struct mlx5_rep_uplink_priv *uplink_priv;
+	struct mlx5_flow_attr *attr = flow->attr;
+	struct mlx5e_rep_priv *uplink_rpriv;
+	struct tunnel_match_key tunnel_key;
+	bool enc_opts_is_dont_care = true;
+	u32 tun_id, enc_opts_id = 0;
+	struct mlx5_eswitch *esw;
+	u32 value, mask;
+	int err;
+
+	esw = priv->mdev->priv.eswitch;
+	uplink_rpriv = mlx5_eswitch_get_uplink_priv(esw, REP_ETH);
+	uplink_priv = &uplink_rpriv->uplink_priv;
+
+	memset(&tunnel_key, 0, sizeof(tunnel_key));
+	COPY_DISSECTOR(rule, FLOW_DISSECTOR_KEY_ENC_CONTROL,
+		       &tunnel_key.enc_control);
+	if (tunnel_key.enc_control.addr_type == FLOW_DISSECTOR_KEY_IPV4_ADDRS)
+		COPY_DISSECTOR(rule, FLOW_DISSECTOR_KEY_ENC_IPV4_ADDRS,
+			       &tunnel_key.enc_ipv4);
+	else
+		COPY_DISSECTOR(rule, FLOW_DISSECTOR_KEY_ENC_IPV6_ADDRS,
+			       &tunnel_key.enc_ipv6);
+	COPY_DISSECTOR(rule, FLOW_DISSECTOR_KEY_ENC_IP, &tunnel_key.enc_ip);
+	COPY_DISSECTOR(rule, FLOW_DISSECTOR_KEY_ENC_PORTS,
+		       &tunnel_key.enc_tp);
+	COPY_DISSECTOR(rule, FLOW_DISSECTOR_KEY_ENC_KEYID,
+		       &tunnel_key.enc_key_id);
+	tunnel_key.filter_ifindex = filter_dev->ifindex;
+
+	err = mapping_add(uplink_priv->tunnel_mapping, &tunnel_key, &tun_id);
+	if (err)
+		return err;
+
+	flow_rule_match_enc_opts(rule, &enc_opts_match);
+	err = enc_opts_is_dont_care_or_full_match(priv,
+						  enc_opts_match.mask,
+						  extack,
+						  &enc_opts_is_dont_care);
+	if (err)
+		goto err_enc_opts;
+
+	if (!enc_opts_is_dont_care) {
+		memset(&tun_enc_opts, 0, sizeof(tun_enc_opts));
+		memcpy(&tun_enc_opts.key, enc_opts_match.key,
+		       sizeof(*enc_opts_match.key));
+		memcpy(&tun_enc_opts.mask, enc_opts_match.mask,
+		       sizeof(*enc_opts_match.mask));
+
+		err = mapping_add(uplink_priv->tunnel_enc_opts_mapping,
+				  &tun_enc_opts, &enc_opts_id);
+		if (err)
+			goto err_enc_opts;
+	}
+
+	value = tun_id << ENC_OPTS_BITS | enc_opts_id;
+	mask = enc_opts_id ? TUNNEL_ID_MASK :
+			     (TUNNEL_ID_MASK & ~ENC_OPTS_BITS_MASK);
+
+	if (attr->chain) {
+		mlx5e_tc_match_to_reg_match(&attr->parse_attr->spec,
+					    TUNNEL_TO_REG, value, mask);
+	} else {
+		mod_hdr_acts = &attr->parse_attr->mod_hdr_acts;
+		err = mlx5e_tc_match_to_reg_set(priv->mdev,
+						mod_hdr_acts, MLX5_FLOW_NAMESPACE_FDB,
+						TUNNEL_TO_REG, value);
+		if (err)
+			goto err_set;
+
+		attr->action |= MLX5_FLOW_CONTEXT_ACTION_MOD_HDR;
+	}
+
+	flow->attr->tunnel_id = value;
+	return 0;
+
+err_set:
+	if (enc_opts_id)
+		mapping_remove(uplink_priv->tunnel_enc_opts_mapping,
+			       enc_opts_id);
+err_enc_opts:
+	mapping_remove(uplink_priv->tunnel_mapping, tun_id);
+	return err;
+}
+
+static void mlx5e_put_flow_tunnel_id(struct mlx5e_tc_flow *flow)
+{
+	u32 enc_opts_id = flow->attr->tunnel_id & ENC_OPTS_BITS_MASK;
+	u32 tun_id = flow->attr->tunnel_id >> ENC_OPTS_BITS;
+	struct mlx5_rep_uplink_priv *uplink_priv;
+	struct mlx5e_rep_priv *uplink_rpriv;
+	struct mlx5_eswitch *esw;
+
+	esw = flow->priv->mdev->priv.eswitch;
+	uplink_rpriv = mlx5_eswitch_get_uplink_priv(esw, REP_ETH);
+	uplink_priv = &uplink_rpriv->uplink_priv;
+
+	if (tun_id)
+		mapping_remove(uplink_priv->tunnel_mapping, tun_id);
+	if (enc_opts_id)
+		mapping_remove(uplink_priv->tunnel_enc_opts_mapping,
+			       enc_opts_id);
+}
+
+void mlx5e_tc_set_ethertype(struct mlx5_core_dev *mdev,
+			    struct flow_match_basic *match, bool outer,
+			    void *headers_c, void *headers_v)
+{
+	bool ip_version_cap;
+
+	ip_version_cap = outer ?
+		MLX5_CAP_FLOWTABLE_NIC_RX(mdev,
+					  ft_field_support.outer_ip_version) :
+		MLX5_CAP_FLOWTABLE_NIC_RX(mdev,
+					  ft_field_support.inner_ip_version);
+
+	if (ip_version_cap && match->mask->n_proto == htons(0xFFFF) &&
+	    (match->key->n_proto == htons(ETH_P_IP) ||
+	     match->key->n_proto == htons(ETH_P_IPV6))) {
+		MLX5_SET_TO_ONES(fte_match_set_lyr_2_4, headers_c, ip_version);
+		MLX5_SET(fte_match_set_lyr_2_4, headers_v, ip_version,
+			 match->key->n_proto == htons(ETH_P_IP) ? 4 : 6);
+	} else {
+		MLX5_SET(fte_match_set_lyr_2_4, headers_c, ethertype,
+			 ntohs(match->mask->n_proto));
+		MLX5_SET(fte_match_set_lyr_2_4, headers_v, ethertype,
+			 ntohs(match->key->n_proto));
+	}
+}
+
+u8 mlx5e_tc_get_ip_version(struct mlx5_flow_spec *spec, bool outer)
+{
+	void *headers_v;
+	u16 ethertype;
+	u8 ip_version;
+
+	if (outer)
+		headers_v = MLX5_ADDR_OF(fte_match_param, spec->match_value, outer_headers);
+	else
+		headers_v = MLX5_ADDR_OF(fte_match_param, spec->match_value, inner_headers);
+
+	ip_version = MLX5_GET(fte_match_set_lyr_2_4, headers_v, ip_version);
+	/* Return ip_version converted from ethertype anyway */
+	if (!ip_version) {
+		ethertype = MLX5_GET(fte_match_set_lyr_2_4, headers_v, ethertype);
+		if (ethertype == ETH_P_IP || ethertype == ETH_P_ARP)
+			ip_version = 4;
+		else if (ethertype == ETH_P_IPV6)
+			ip_version = 6;
+	}
+	return ip_version;
+}
+
+/* Tunnel device follows RFC 6040, see include/net/inet_ecn.h.
+ * And changes inner ip_ecn depending on inner and outer ip_ecn as follows:
+ *      +---------+----------------------------------------+
+ *      |Arriving |         Arriving Outer Header          |
+ *      |   Inner +---------+---------+---------+----------+
+ *      |  Header | Not-ECT | ECT(0)  | ECT(1)  |   CE     |
+ *      +---------+---------+---------+---------+----------+
+ *      | Not-ECT | Not-ECT | Not-ECT | Not-ECT | <drop>   |
+ *      |  ECT(0) |  ECT(0) | ECT(0)  | ECT(1)  |   CE*    |
+ *      |  ECT(1) |  ECT(1) | ECT(1)  | ECT(1)* |   CE*    |
+ *      |    CE   |   CE    |  CE     | CE      |   CE     |
+ *      +---------+---------+---------+---------+----------+
+ *
+ * Tc matches on inner after decapsulation on tunnel device, but hw offload matches
+ * the inner ip_ecn value before hardware decap action.
+ *
+ * Cells marked are changed from original inner packet ip_ecn value during decap, and
+ * so matching those values on inner ip_ecn before decap will fail.
+ *
+ * The following helper allows offload when inner ip_ecn won't be changed by outer ip_ecn,
+ * except for the outer ip_ecn = CE, where in all cases inner ip_ecn will be changed to CE,
+ * and such we can drop the inner ip_ecn=CE match.
+ */
+
+static int mlx5e_tc_verify_tunnel_ecn(struct mlx5e_priv *priv,
+				      struct flow_cls_offload *f,
+				      bool *match_inner_ecn)
+{
+	u8 outer_ecn_mask = 0, outer_ecn_key = 0, inner_ecn_mask = 0, inner_ecn_key = 0;
+	struct flow_rule *rule = flow_cls_offload_flow_rule(f);
+	struct netlink_ext_ack *extack = f->common.extack;
+	struct flow_match_ip match;
+
+	*match_inner_ecn = true;
+
+	if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_ENC_IP)) {
+		flow_rule_match_enc_ip(rule, &match);
+		outer_ecn_key = match.key->tos & INET_ECN_MASK;
+		outer_ecn_mask = match.mask->tos & INET_ECN_MASK;
+	}
+
+	if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_IP)) {
+		flow_rule_match_ip(rule, &match);
+		inner_ecn_key = match.key->tos & INET_ECN_MASK;
+		inner_ecn_mask = match.mask->tos & INET_ECN_MASK;
+	}
+
+	if (outer_ecn_mask != 0 && outer_ecn_mask != INET_ECN_MASK) {
+		NL_SET_ERR_MSG_MOD(extack, "Partial match on enc_tos ecn bits isn't supported");
+		netdev_warn(priv->netdev, "Partial match on enc_tos ecn bits isn't supported");
+		return -EOPNOTSUPP;
+	}
+
+	if (!outer_ecn_mask) {
+		if (!inner_ecn_mask)
+			return 0;
+
+		NL_SET_ERR_MSG_MOD(extack,
+				   "Matching on tos ecn bits without also matching enc_tos ecn bits isn't supported");
+		netdev_warn(priv->netdev,
+			    "Matching on tos ecn bits without also matching enc_tos ecn bits isn't supported");
+		return -EOPNOTSUPP;
+	}
+
+	if (inner_ecn_mask && inner_ecn_mask != INET_ECN_MASK) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "Partial match on tos ecn bits with match on enc_tos ecn bits isn't supported");
+		netdev_warn(priv->netdev,
+			    "Partial match on tos ecn bits with match on enc_tos ecn bits isn't supported");
+		return -EOPNOTSUPP;
+	}
+
+	if (!inner_ecn_mask)
+		return 0;
+
+	/* Both inner and outer have full mask on ecn */
+
+	if (outer_ecn_key == INET_ECN_ECT_1) {
+		/* inner ecn might change by DECAP action */
+
+		NL_SET_ERR_MSG_MOD(extack, "Match on enc_tos ecn = ECT(1) isn't supported");
+		netdev_warn(priv->netdev, "Match on enc_tos ecn = ECT(1) isn't supported");
+		return -EOPNOTSUPP;
+	}
+
+	if (outer_ecn_key != INET_ECN_CE)
+		return 0;
+
+	if (inner_ecn_key != INET_ECN_CE) {
+		/* Can't happen in software, as packet ecn will be changed to CE after decap */
+		NL_SET_ERR_MSG_MOD(extack,
+				   "Match on tos enc_tos ecn = CE while match on tos ecn != CE isn't supported");
+		netdev_warn(priv->netdev,
+			    "Match on tos enc_tos ecn = CE while match on tos ecn != CE isn't supported");
+		return -EOPNOTSUPP;
+	}
+
+	/* outer ecn = CE, inner ecn = CE, as decap will change inner ecn to CE in anycase,
+	 * drop match on inner ecn
+	 */
+	*match_inner_ecn = false;
+
+	return 0;
+}
+
+static int parse_tunnel_attr(struct mlx5e_priv *priv,
+			     struct mlx5e_tc_flow *flow,
+			     struct mlx5_flow_spec *spec,
+			     struct flow_cls_offload *f,
+			     struct net_device *filter_dev,
+			     u8 *match_level,
+			     bool *match_inner)
+{
+	struct mlx5e_tc_tunnel *tunnel = mlx5e_get_tc_tun(filter_dev);
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	struct netlink_ext_ack *extack = f->common.extack;
+	bool needs_mapping, sets_mapping;
+	int err;
+
+	if (!mlx5e_is_eswitch_flow(flow)) {
+		NL_SET_ERR_MSG_MOD(extack, "Match on tunnel is not supported");
+		return -EOPNOTSUPP;
+	}
+
+	needs_mapping = !!flow->attr->chain;
+	sets_mapping = flow_requires_tunnel_mapping(flow->attr->chain, f);
+	*match_inner = !needs_mapping;
+
+	if ((needs_mapping || sets_mapping) &&
+	    !mlx5_eswitch_reg_c1_loopback_enabled(esw)) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "Chains on tunnel devices isn't supported without register loopback support");
+		netdev_warn(priv->netdev,
+			    "Chains on tunnel devices isn't supported without register loopback support");
+		return -EOPNOTSUPP;
+	}
+
+	if (!flow->attr->chain) {
+		err = mlx5e_tc_tun_parse(filter_dev, priv, spec, f,
+					 match_level);
+		if (err) {
+			NL_SET_ERR_MSG_MOD(extack,
+					   "Failed to parse tunnel attributes");
+			netdev_warn(priv->netdev,
+				    "Failed to parse tunnel attributes");
+			return err;
+		}
+
+		/* With mpls over udp we decapsulate using packet reformat
+		 * object
+		 */
+		if (!netif_is_bareudp(filter_dev))
+			flow->attr->action |= MLX5_FLOW_CONTEXT_ACTION_DECAP;
+		err = mlx5e_tc_set_attr_rx_tun(flow, spec);
+		if (err)
+			return err;
+	} else if (tunnel) {
+		struct mlx5_flow_spec *tmp_spec;
+
+		tmp_spec = kvzalloc(sizeof(*tmp_spec), GFP_KERNEL);
+		if (!tmp_spec) {
+			NL_SET_ERR_MSG_MOD(extack, "Failed to allocate memory for tunnel tmp spec");
+			netdev_warn(priv->netdev, "Failed to allocate memory for tunnel tmp spec");
+			return -ENOMEM;
+		}
+		memcpy(tmp_spec, spec, sizeof(*tmp_spec));
+
+		err = mlx5e_tc_tun_parse(filter_dev, priv, tmp_spec, f, match_level);
+		if (err) {
+			kvfree(tmp_spec);
+			NL_SET_ERR_MSG_MOD(extack, "Failed to parse tunnel attributes");
+			netdev_warn(priv->netdev, "Failed to parse tunnel attributes");
+			return err;
+		}
+		err = mlx5e_tc_set_attr_rx_tun(flow, tmp_spec);
+		kvfree(tmp_spec);
+		if (err)
+			return err;
+	}
+
+	if (!needs_mapping && !sets_mapping)
+		return 0;
+
+	return mlx5e_get_flow_tunnel_id(priv, flow, f, filter_dev);
+}
+
+static void *get_match_inner_headers_criteria(struct mlx5_flow_spec *spec)
+{
+	return MLX5_ADDR_OF(fte_match_param, spec->match_criteria,
+			    inner_headers);
+}
+
+static void *get_match_inner_headers_value(struct mlx5_flow_spec *spec)
+{
+	return MLX5_ADDR_OF(fte_match_param, spec->match_value,
+			    inner_headers);
+}
+
+static void *get_match_outer_headers_criteria(struct mlx5_flow_spec *spec)
+{
+	return MLX5_ADDR_OF(fte_match_param, spec->match_criteria,
+			    outer_headers);
+}
+
+static void *get_match_outer_headers_value(struct mlx5_flow_spec *spec)
+{
+	return MLX5_ADDR_OF(fte_match_param, spec->match_value,
+			    outer_headers);
+}
+
+void *mlx5e_get_match_headers_value(u32 flags, struct mlx5_flow_spec *spec)
+{
+	return (flags & MLX5_FLOW_CONTEXT_ACTION_DECAP) ?
+		get_match_inner_headers_value(spec) :
+		get_match_outer_headers_value(spec);
+}
+
+void *mlx5e_get_match_headers_criteria(u32 flags, struct mlx5_flow_spec *spec)
+{
+	return (flags & MLX5_FLOW_CONTEXT_ACTION_DECAP) ?
+		get_match_inner_headers_criteria(spec) :
+		get_match_outer_headers_criteria(spec);
+}
+
+static int mlx5e_flower_parse_meta(struct net_device *filter_dev,
+				   struct flow_cls_offload *f)
+{
+	struct flow_rule *rule = flow_cls_offload_flow_rule(f);
+	struct netlink_ext_ack *extack = f->common.extack;
+	struct net_device *ingress_dev;
+	struct flow_match_meta match;
+
+	if (!flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_META))
+		return 0;
+
+	flow_rule_match_meta(rule, &match);
+	if (!match.mask->ingress_ifindex)
+		return 0;
+
+	if (match.mask->ingress_ifindex != 0xFFFFFFFF) {
+		NL_SET_ERR_MSG_MOD(extack, "Unsupported ingress ifindex mask");
+		return -EOPNOTSUPP;
+	}
+
+	ingress_dev = __dev_get_by_index(dev_net(filter_dev),
+					 match.key->ingress_ifindex);
+	if (!ingress_dev) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "Can't find the ingress port to match on");
+		return -ENOENT;
+	}
+
+	if (ingress_dev != filter_dev) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "Can't match on the ingress filter port");
+		return -EOPNOTSUPP;
+	}
+
+	return 0;
+}
+
+static bool skip_key_basic(struct net_device *filter_dev,
+			   struct flow_cls_offload *f)
+{
+	/* When doing mpls over udp decap, the user needs to provide
+	 * MPLS_UC as the protocol in order to be able to match on mpls
+	 * label fields.  However, the actual ethertype is IP so we want to
+	 * avoid matching on this, otherwise we'll fail the match.
+	 */
+	if (netif_is_bareudp(filter_dev) && f->common.chain_index == 0)
+		return true;
+
+	return false;
+}
+
+static int __parse_cls_flower(struct mlx5e_priv *priv,
+			      struct mlx5e_tc_flow *flow,
+			      struct mlx5_flow_spec *spec,
+			      struct flow_cls_offload *f,
+			      struct net_device *filter_dev,
+			      u8 *inner_match_level, u8 *outer_match_level)
+{
+	struct netlink_ext_ack *extack = f->common.extack;
+	void *headers_c = MLX5_ADDR_OF(fte_match_param, spec->match_criteria,
+				       outer_headers);
+	void *headers_v = MLX5_ADDR_OF(fte_match_param, spec->match_value,
+				       outer_headers);
+	void *misc_c = MLX5_ADDR_OF(fte_match_param, spec->match_criteria,
+				    misc_parameters);
+	void *misc_v = MLX5_ADDR_OF(fte_match_param, spec->match_value,
+				    misc_parameters);
+	void *misc_c_3 = MLX5_ADDR_OF(fte_match_param, spec->match_criteria,
+				    misc_parameters_3);
+	void *misc_v_3 = MLX5_ADDR_OF(fte_match_param, spec->match_value,
+				    misc_parameters_3);
+	struct flow_rule *rule = flow_cls_offload_flow_rule(f);
+	struct flow_dissector *dissector = rule->match.dissector;
+	enum fs_flow_table_type fs_type;
+	bool match_inner_ecn = true;
+	u16 addr_type = 0;
+	u8 ip_proto = 0;
+	u8 *match_level;
+	int err;
+
+	fs_type = mlx5e_is_eswitch_flow(flow) ? FS_FT_FDB : FS_FT_NIC_RX;
+	match_level = outer_match_level;
+
+	if (dissector->used_keys &
+	    ~(BIT(FLOW_DISSECTOR_KEY_META) |
+	      BIT(FLOW_DISSECTOR_KEY_CONTROL) |
+	      BIT(FLOW_DISSECTOR_KEY_BASIC) |
+	      BIT(FLOW_DISSECTOR_KEY_ETH_ADDRS) |
+	      BIT(FLOW_DISSECTOR_KEY_VLAN) |
+	      BIT(FLOW_DISSECTOR_KEY_CVLAN) |
+	      BIT(FLOW_DISSECTOR_KEY_IPV4_ADDRS) |
+	      BIT(FLOW_DISSECTOR_KEY_IPV6_ADDRS) |
+	      BIT(FLOW_DISSECTOR_KEY_PORTS) |
+	      BIT(FLOW_DISSECTOR_KEY_ENC_KEYID) |
+	      BIT(FLOW_DISSECTOR_KEY_ENC_IPV4_ADDRS) |
+	      BIT(FLOW_DISSECTOR_KEY_ENC_IPV6_ADDRS) |
+	      BIT(FLOW_DISSECTOR_KEY_ENC_PORTS)	|
+	      BIT(FLOW_DISSECTOR_KEY_ENC_CONTROL) |
+	      BIT(FLOW_DISSECTOR_KEY_TCP) |
+	      BIT(FLOW_DISSECTOR_KEY_IP)  |
+	      BIT(FLOW_DISSECTOR_KEY_CT) |
+	      BIT(FLOW_DISSECTOR_KEY_ENC_IP) |
+	      BIT(FLOW_DISSECTOR_KEY_ENC_OPTS) |
+	      BIT(FLOW_DISSECTOR_KEY_ICMP) |
+	      BIT(FLOW_DISSECTOR_KEY_MPLS))) {
+		NL_SET_ERR_MSG_MOD(extack, "Unsupported key");
+		netdev_dbg(priv->netdev, "Unsupported key used: 0x%x\n",
+			   dissector->used_keys);
+		return -EOPNOTSUPP;
+	}
+
+	if (mlx5e_get_tc_tun(filter_dev)) {
+		bool match_inner = false;
+
+		err = parse_tunnel_attr(priv, flow, spec, f, filter_dev,
+					outer_match_level, &match_inner);
+		if (err)
+			return err;
+
+		if (match_inner) {
+			/* header pointers should point to the inner headers
+			 * if the packet was decapsulated already.
+			 * outer headers are set by parse_tunnel_attr.
+			 */
+			match_level = inner_match_level;
+			headers_c = get_match_inner_headers_criteria(spec);
+			headers_v = get_match_inner_headers_value(spec);
+		}
+
+		err = mlx5e_tc_verify_tunnel_ecn(priv, f, &match_inner_ecn);
+		if (err)
+			return err;
+	}
+
+	err = mlx5e_flower_parse_meta(filter_dev, f);
+	if (err)
+		return err;
+
+	if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_BASIC) &&
+	    !skip_key_basic(filter_dev, f)) {
+		struct flow_match_basic match;
+
+		flow_rule_match_basic(rule, &match);
+		mlx5e_tc_set_ethertype(priv->mdev, &match,
+				       match_level == outer_match_level,
+				       headers_c, headers_v);
+
+		if (match.mask->n_proto)
+			*match_level = MLX5_MATCH_L2;
+	}
+	if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_VLAN) ||
+	    is_vlan_dev(filter_dev)) {
+		struct flow_dissector_key_vlan filter_dev_mask;
+		struct flow_dissector_key_vlan filter_dev_key;
+		struct flow_match_vlan match;
+
+		if (is_vlan_dev(filter_dev)) {
+			match.key = &filter_dev_key;
+			match.key->vlan_id = vlan_dev_vlan_id(filter_dev);
+			match.key->vlan_tpid = vlan_dev_vlan_proto(filter_dev);
+			match.key->vlan_priority = 0;
+			match.mask = &filter_dev_mask;
+			memset(match.mask, 0xff, sizeof(*match.mask));
+			match.mask->vlan_priority = 0;
+		} else {
+			flow_rule_match_vlan(rule, &match);
+		}
+		if (match.mask->vlan_id ||
+		    match.mask->vlan_priority ||
+		    match.mask->vlan_tpid) {
+			if (match.key->vlan_tpid == htons(ETH_P_8021AD)) {
+				MLX5_SET(fte_match_set_lyr_2_4, headers_c,
+					 svlan_tag, 1);
+				MLX5_SET(fte_match_set_lyr_2_4, headers_v,
+					 svlan_tag, 1);
+			} else {
+				MLX5_SET(fte_match_set_lyr_2_4, headers_c,
+					 cvlan_tag, 1);
+				MLX5_SET(fte_match_set_lyr_2_4, headers_v,
+					 cvlan_tag, 1);
+			}
+
+			MLX5_SET(fte_match_set_lyr_2_4, headers_c, first_vid,
+				 match.mask->vlan_id);
+			MLX5_SET(fte_match_set_lyr_2_4, headers_v, first_vid,
+				 match.key->vlan_id);
+
+			MLX5_SET(fte_match_set_lyr_2_4, headers_c, first_prio,
+				 match.mask->vlan_priority);
+			MLX5_SET(fte_match_set_lyr_2_4, headers_v, first_prio,
+				 match.key->vlan_priority);
+
+			*match_level = MLX5_MATCH_L2;
+
+			if (!flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_CVLAN) &&
+			    match.mask->vlan_eth_type &&
+			    MLX5_CAP_FLOWTABLE_TYPE(priv->mdev,
+						    ft_field_support.outer_second_vid,
+						    fs_type)) {
+				MLX5_SET(fte_match_set_misc, misc_c,
+					 outer_second_cvlan_tag, 1);
+				spec->match_criteria_enable |=
+					MLX5_MATCH_MISC_PARAMETERS;
+			}
+		}
+	} else if (*match_level != MLX5_MATCH_NONE) {
+		/* cvlan_tag enabled in match criteria and
+		 * disabled in match value means both S & C tags
+		 * don't exist (untagged of both)
+		 */
+		MLX5_SET(fte_match_set_lyr_2_4, headers_c, cvlan_tag, 1);
+		*match_level = MLX5_MATCH_L2;
+	}
+
+	if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_CVLAN)) {
+		struct flow_match_vlan match;
+
+		flow_rule_match_cvlan(rule, &match);
+		if (match.mask->vlan_id ||
+		    match.mask->vlan_priority ||
+		    match.mask->vlan_tpid) {
+			if (!MLX5_CAP_FLOWTABLE_TYPE(priv->mdev, ft_field_support.outer_second_vid,
+						     fs_type)) {
+				NL_SET_ERR_MSG_MOD(extack,
+						   "Matching on CVLAN is not supported");
+				return -EOPNOTSUPP;
+			}
+
+			if (match.key->vlan_tpid == htons(ETH_P_8021AD)) {
+				MLX5_SET(fte_match_set_misc, misc_c,
+					 outer_second_svlan_tag, 1);
+				MLX5_SET(fte_match_set_misc, misc_v,
+					 outer_second_svlan_tag, 1);
+			} else {
+				MLX5_SET(fte_match_set_misc, misc_c,
+					 outer_second_cvlan_tag, 1);
+				MLX5_SET(fte_match_set_misc, misc_v,
+					 outer_second_cvlan_tag, 1);
+			}
+
+			MLX5_SET(fte_match_set_misc, misc_c, outer_second_vid,
+				 match.mask->vlan_id);
+			MLX5_SET(fte_match_set_misc, misc_v, outer_second_vid,
+				 match.key->vlan_id);
+			MLX5_SET(fte_match_set_misc, misc_c, outer_second_prio,
+				 match.mask->vlan_priority);
+			MLX5_SET(fte_match_set_misc, misc_v, outer_second_prio,
+				 match.key->vlan_priority);
+
+			*match_level = MLX5_MATCH_L2;
+			spec->match_criteria_enable |= MLX5_MATCH_MISC_PARAMETERS;
+		}
+	}
+
+	if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_ETH_ADDRS)) {
+		struct flow_match_eth_addrs match;
+
+		flow_rule_match_eth_addrs(rule, &match);
+		ether_addr_copy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_c,
+					     dmac_47_16),
+				match.mask->dst);
+		ether_addr_copy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_v,
+					     dmac_47_16),
+				match.key->dst);
+
+		ether_addr_copy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_c,
+					     smac_47_16),
+				match.mask->src);
+		ether_addr_copy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_v,
+					     smac_47_16),
+				match.key->src);
+
+		if (!is_zero_ether_addr(match.mask->src) ||
+		    !is_zero_ether_addr(match.mask->dst))
+			*match_level = MLX5_MATCH_L2;
+	}
+
+	if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_CONTROL)) {
+		struct flow_match_control match;
+
+		flow_rule_match_control(rule, &match);
+		addr_type = match.key->addr_type;
+
+		/* the HW doesn't support frag first/later */
+		if (match.mask->flags & FLOW_DIS_FIRST_FRAG) {
+			NL_SET_ERR_MSG_MOD(extack, "Match on frag first/later is not supported");
+			return -EOPNOTSUPP;
+		}
+
+		if (match.mask->flags & FLOW_DIS_IS_FRAGMENT) {
+			MLX5_SET(fte_match_set_lyr_2_4, headers_c, frag, 1);
+			MLX5_SET(fte_match_set_lyr_2_4, headers_v, frag,
+				 match.key->flags & FLOW_DIS_IS_FRAGMENT);
+
+			/* the HW doesn't need L3 inline to match on frag=no */
+			if (!(match.key->flags & FLOW_DIS_IS_FRAGMENT))
+				*match_level = MLX5_MATCH_L2;
+	/* ***  L2 attributes parsing up to here *** */
+			else
+				*match_level = MLX5_MATCH_L3;
+		}
+	}
+
+	if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_BASIC)) {
+		struct flow_match_basic match;
+
+		flow_rule_match_basic(rule, &match);
+		ip_proto = match.key->ip_proto;
+
+		MLX5_SET(fte_match_set_lyr_2_4, headers_c, ip_protocol,
+			 match.mask->ip_proto);
+		MLX5_SET(fte_match_set_lyr_2_4, headers_v, ip_protocol,
+			 match.key->ip_proto);
+
+		if (match.mask->ip_proto)
+			*match_level = MLX5_MATCH_L3;
+	}
+
+	if (addr_type == FLOW_DISSECTOR_KEY_IPV4_ADDRS) {
+		struct flow_match_ipv4_addrs match;
+
+		flow_rule_match_ipv4_addrs(rule, &match);
+		memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_c,
+				    src_ipv4_src_ipv6.ipv4_layout.ipv4),
+		       &match.mask->src, sizeof(match.mask->src));
+		memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_v,
+				    src_ipv4_src_ipv6.ipv4_layout.ipv4),
+		       &match.key->src, sizeof(match.key->src));
+		memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_c,
+				    dst_ipv4_dst_ipv6.ipv4_layout.ipv4),
+		       &match.mask->dst, sizeof(match.mask->dst));
+		memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_v,
+				    dst_ipv4_dst_ipv6.ipv4_layout.ipv4),
+		       &match.key->dst, sizeof(match.key->dst));
+
+		if (match.mask->src || match.mask->dst)
+			*match_level = MLX5_MATCH_L3;
+	}
+
+	if (addr_type == FLOW_DISSECTOR_KEY_IPV6_ADDRS) {
+		struct flow_match_ipv6_addrs match;
+
+		flow_rule_match_ipv6_addrs(rule, &match);
+		memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_c,
+				    src_ipv4_src_ipv6.ipv6_layout.ipv6),
+		       &match.mask->src, sizeof(match.mask->src));
+		memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_v,
+				    src_ipv4_src_ipv6.ipv6_layout.ipv6),
+		       &match.key->src, sizeof(match.key->src));
+
+		memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_c,
+				    dst_ipv4_dst_ipv6.ipv6_layout.ipv6),
+		       &match.mask->dst, sizeof(match.mask->dst));
+		memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_v,
+				    dst_ipv4_dst_ipv6.ipv6_layout.ipv6),
+		       &match.key->dst, sizeof(match.key->dst));
+
+		if (ipv6_addr_type(&match.mask->src) != IPV6_ADDR_ANY ||
+		    ipv6_addr_type(&match.mask->dst) != IPV6_ADDR_ANY)
+			*match_level = MLX5_MATCH_L3;
+	}
+
+	if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_IP)) {
+		struct flow_match_ip match;
+
+		flow_rule_match_ip(rule, &match);
+		if (match_inner_ecn) {
+			MLX5_SET(fte_match_set_lyr_2_4, headers_c, ip_ecn,
+				 match.mask->tos & 0x3);
+			MLX5_SET(fte_match_set_lyr_2_4, headers_v, ip_ecn,
+				 match.key->tos & 0x3);
+		}
+
+		MLX5_SET(fte_match_set_lyr_2_4, headers_c, ip_dscp,
+			 match.mask->tos >> 2);
+		MLX5_SET(fte_match_set_lyr_2_4, headers_v, ip_dscp,
+			 match.key->tos  >> 2);
+
+		MLX5_SET(fte_match_set_lyr_2_4, headers_c, ttl_hoplimit,
+			 match.mask->ttl);
+		MLX5_SET(fte_match_set_lyr_2_4, headers_v, ttl_hoplimit,
+			 match.key->ttl);
+
+		if (match.mask->ttl &&
+		    !MLX5_CAP_ESW_FLOWTABLE_FDB(priv->mdev,
+						ft_field_support.outer_ipv4_ttl)) {
+			NL_SET_ERR_MSG_MOD(extack,
+					   "Matching on TTL is not supported");
+			return -EOPNOTSUPP;
+		}
+
+		if (match.mask->tos || match.mask->ttl)
+			*match_level = MLX5_MATCH_L3;
+	}
+
+	/* ***  L3 attributes parsing up to here *** */
+
+	if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_PORTS)) {
+		struct flow_match_ports match;
+
+		flow_rule_match_ports(rule, &match);
+		switch (ip_proto) {
+		case IPPROTO_TCP:
+			MLX5_SET(fte_match_set_lyr_2_4, headers_c,
+				 tcp_sport, ntohs(match.mask->src));
+			MLX5_SET(fte_match_set_lyr_2_4, headers_v,
+				 tcp_sport, ntohs(match.key->src));
+
+			MLX5_SET(fte_match_set_lyr_2_4, headers_c,
+				 tcp_dport, ntohs(match.mask->dst));
+			MLX5_SET(fte_match_set_lyr_2_4, headers_v,
+				 tcp_dport, ntohs(match.key->dst));
+			break;
+
+		case IPPROTO_UDP:
+			MLX5_SET(fte_match_set_lyr_2_4, headers_c,
+				 udp_sport, ntohs(match.mask->src));
+			MLX5_SET(fte_match_set_lyr_2_4, headers_v,
+				 udp_sport, ntohs(match.key->src));
+
+			MLX5_SET(fte_match_set_lyr_2_4, headers_c,
+				 udp_dport, ntohs(match.mask->dst));
+			MLX5_SET(fte_match_set_lyr_2_4, headers_v,
+				 udp_dport, ntohs(match.key->dst));
+			break;
+		default:
+			NL_SET_ERR_MSG_MOD(extack,
+					   "Only UDP and TCP transports are supported for L4 matching");
+			netdev_err(priv->netdev,
+				   "Only UDP and TCP transport are supported\n");
+			return -EINVAL;
+		}
+
+		if (match.mask->src || match.mask->dst)
+			*match_level = MLX5_MATCH_L4;
+	}
+
+	if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_TCP)) {
+		struct flow_match_tcp match;
+
+		flow_rule_match_tcp(rule, &match);
+		MLX5_SET(fte_match_set_lyr_2_4, headers_c, tcp_flags,
+			 ntohs(match.mask->flags));
+		MLX5_SET(fte_match_set_lyr_2_4, headers_v, tcp_flags,
+			 ntohs(match.key->flags));
+
+		if (match.mask->flags)
+			*match_level = MLX5_MATCH_L4;
+	}
+	if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_ICMP)) {
+		struct flow_match_icmp match;
+
+		flow_rule_match_icmp(rule, &match);
+		switch (ip_proto) {
+		case IPPROTO_ICMP:
+			if (!(MLX5_CAP_GEN(priv->mdev, flex_parser_protocols) &
+			      MLX5_FLEX_PROTO_ICMP)) {
+				NL_SET_ERR_MSG_MOD(extack,
+						   "Match on Flex protocols for ICMP is not supported");
+				return -EOPNOTSUPP;
+			}
+			MLX5_SET(fte_match_set_misc3, misc_c_3, icmp_type,
+				 match.mask->type);
+			MLX5_SET(fte_match_set_misc3, misc_v_3, icmp_type,
+				 match.key->type);
+			MLX5_SET(fte_match_set_misc3, misc_c_3, icmp_code,
+				 match.mask->code);
+			MLX5_SET(fte_match_set_misc3, misc_v_3, icmp_code,
+				 match.key->code);
+			break;
+		case IPPROTO_ICMPV6:
+			if (!(MLX5_CAP_GEN(priv->mdev, flex_parser_protocols) &
+			      MLX5_FLEX_PROTO_ICMPV6)) {
+				NL_SET_ERR_MSG_MOD(extack,
+						   "Match on Flex protocols for ICMPV6 is not supported");
+				return -EOPNOTSUPP;
+			}
+			MLX5_SET(fte_match_set_misc3, misc_c_3, icmpv6_type,
+				 match.mask->type);
+			MLX5_SET(fte_match_set_misc3, misc_v_3, icmpv6_type,
+				 match.key->type);
+			MLX5_SET(fte_match_set_misc3, misc_c_3, icmpv6_code,
+				 match.mask->code);
+			MLX5_SET(fte_match_set_misc3, misc_v_3, icmpv6_code,
+				 match.key->code);
+			break;
+		default:
+			NL_SET_ERR_MSG_MOD(extack,
+					   "Code and type matching only with ICMP and ICMPv6");
+			netdev_err(priv->netdev,
+				   "Code and type matching only with ICMP and ICMPv6\n");
+			return -EINVAL;
+		}
+		if (match.mask->code || match.mask->type) {
+			*match_level = MLX5_MATCH_L4;
+			spec->match_criteria_enable |= MLX5_MATCH_MISC_PARAMETERS_3;
+		}
+	}
+	/* Currently supported only for MPLS over UDP */
+	if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_MPLS) &&
+	    !netif_is_bareudp(filter_dev)) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "Matching on MPLS is supported only for MPLS over UDP");
+		netdev_err(priv->netdev,
+			   "Matching on MPLS is supported only for MPLS over UDP\n");
+		return -EOPNOTSUPP;
+	}
+
+	return 0;
+}
+
+static int parse_cls_flower(struct mlx5e_priv *priv,
+			    struct mlx5e_tc_flow *flow,
+			    struct mlx5_flow_spec *spec,
+			    struct flow_cls_offload *f,
+			    struct net_device *filter_dev)
+{
+	u8 inner_match_level, outer_match_level, non_tunnel_match_level;
+	struct netlink_ext_ack *extack = f->common.extack;
+	struct mlx5_core_dev *dev = priv->mdev;
+	struct mlx5_eswitch *esw = dev->priv.eswitch;
+	struct mlx5e_rep_priv *rpriv = priv->ppriv;
+	struct mlx5_eswitch_rep *rep;
+	bool is_eswitch_flow;
+	int err;
+
+	inner_match_level = MLX5_MATCH_NONE;
+	outer_match_level = MLX5_MATCH_NONE;
+
+	err = __parse_cls_flower(priv, flow, spec, f, filter_dev,
+				 &inner_match_level, &outer_match_level);
+	non_tunnel_match_level = (inner_match_level == MLX5_MATCH_NONE) ?
+				 outer_match_level : inner_match_level;
+
+	is_eswitch_flow = mlx5e_is_eswitch_flow(flow);
+	if (!err && is_eswitch_flow) {
+		rep = rpriv->rep;
+		if (rep->vport != MLX5_VPORT_UPLINK &&
+		    (esw->offloads.inline_mode != MLX5_INLINE_MODE_NONE &&
+		    esw->offloads.inline_mode < non_tunnel_match_level)) {
+			NL_SET_ERR_MSG_MOD(extack,
+					   "Flow is not offloaded due to min inline setting");
+			netdev_warn(priv->netdev,
+				    "Flow is not offloaded due to min inline setting, required %d actual %d\n",
+				    non_tunnel_match_level, esw->offloads.inline_mode);
+			return -EOPNOTSUPP;
+		}
+	}
+
+	flow->attr->inner_match_level = inner_match_level;
+	flow->attr->outer_match_level = outer_match_level;
+
+
+	return err;
+}
+
+struct mlx5_fields {
+	u8  field;
+	u8  field_bsize;
+	u32 field_mask;
+	u32 offset;
+	u32 match_offset;
+};
+
+#define OFFLOAD(fw_field, field_bsize, field_mask, field, off, match_field) \
+		{MLX5_ACTION_IN_FIELD_OUT_ ## fw_field, field_bsize, field_mask, \
+		 offsetof(struct pedit_headers, field) + (off), \
+		 MLX5_BYTE_OFF(fte_match_set_lyr_2_4, match_field)}
+
+/* masked values are the same and there are no rewrites that do not have a
+ * match.
+ */
+#define SAME_VAL_MASK(type, valp, maskp, matchvalp, matchmaskp) ({ \
+	type matchmaskx = *(type *)(matchmaskp); \
+	type matchvalx = *(type *)(matchvalp); \
+	type maskx = *(type *)(maskp); \
+	type valx = *(type *)(valp); \
+	\
+	(valx & maskx) == (matchvalx & matchmaskx) && !(maskx & (maskx ^ \
+								 matchmaskx)); \
+})
+
+static bool cmp_val_mask(void *valp, void *maskp, void *matchvalp,
+			 void *matchmaskp, u8 bsize)
+{
+	bool same = false;
+
+	switch (bsize) {
+	case 8:
+		same = SAME_VAL_MASK(u8, valp, maskp, matchvalp, matchmaskp);
+		break;
+	case 16:
+		same = SAME_VAL_MASK(u16, valp, maskp, matchvalp, matchmaskp);
+		break;
+	case 32:
+		same = SAME_VAL_MASK(u32, valp, maskp, matchvalp, matchmaskp);
+		break;
+	}
+
+	return same;
+}
+
+static struct mlx5_fields fields[] = {
+	OFFLOAD(DMAC_47_16, 32, U32_MAX, eth.h_dest[0], 0, dmac_47_16),
+	OFFLOAD(DMAC_15_0,  16, U16_MAX, eth.h_dest[4], 0, dmac_15_0),
+	OFFLOAD(SMAC_47_16, 32, U32_MAX, eth.h_source[0], 0, smac_47_16),
+	OFFLOAD(SMAC_15_0,  16, U16_MAX, eth.h_source[4], 0, smac_15_0),
+	OFFLOAD(ETHERTYPE,  16, U16_MAX, eth.h_proto, 0, ethertype),
+	OFFLOAD(FIRST_VID,  16, U16_MAX, vlan.h_vlan_TCI, 0, first_vid),
+
+	OFFLOAD(IP_DSCP, 8,    0xfc, ip4.tos,   0, ip_dscp),
+	OFFLOAD(IP_TTL,  8,  U8_MAX, ip4.ttl,   0, ttl_hoplimit),
+	OFFLOAD(SIPV4,  32, U32_MAX, ip4.saddr, 0, src_ipv4_src_ipv6.ipv4_layout.ipv4),
+	OFFLOAD(DIPV4,  32, U32_MAX, ip4.daddr, 0, dst_ipv4_dst_ipv6.ipv4_layout.ipv4),
+
+	OFFLOAD(SIPV6_127_96, 32, U32_MAX, ip6.saddr.s6_addr32[0], 0,
+		src_ipv4_src_ipv6.ipv6_layout.ipv6[0]),
+	OFFLOAD(SIPV6_95_64,  32, U32_MAX, ip6.saddr.s6_addr32[1], 0,
+		src_ipv4_src_ipv6.ipv6_layout.ipv6[4]),
+	OFFLOAD(SIPV6_63_32,  32, U32_MAX, ip6.saddr.s6_addr32[2], 0,
+		src_ipv4_src_ipv6.ipv6_layout.ipv6[8]),
+	OFFLOAD(SIPV6_31_0,   32, U32_MAX, ip6.saddr.s6_addr32[3], 0,
+		src_ipv4_src_ipv6.ipv6_layout.ipv6[12]),
+	OFFLOAD(DIPV6_127_96, 32, U32_MAX, ip6.daddr.s6_addr32[0], 0,
+		dst_ipv4_dst_ipv6.ipv6_layout.ipv6[0]),
+	OFFLOAD(DIPV6_95_64,  32, U32_MAX, ip6.daddr.s6_addr32[1], 0,
+		dst_ipv4_dst_ipv6.ipv6_layout.ipv6[4]),
+	OFFLOAD(DIPV6_63_32,  32, U32_MAX, ip6.daddr.s6_addr32[2], 0,
+		dst_ipv4_dst_ipv6.ipv6_layout.ipv6[8]),
+	OFFLOAD(DIPV6_31_0,   32, U32_MAX, ip6.daddr.s6_addr32[3], 0,
+		dst_ipv4_dst_ipv6.ipv6_layout.ipv6[12]),
+	OFFLOAD(IPV6_HOPLIMIT, 8,  U8_MAX, ip6.hop_limit, 0, ttl_hoplimit),
+	OFFLOAD(IP_DSCP, 16,  0x0fc0, ip6, 0, ip_dscp),
+
+	OFFLOAD(TCP_SPORT, 16, U16_MAX, tcp.source,  0, tcp_sport),
+	OFFLOAD(TCP_DPORT, 16, U16_MAX, tcp.dest,    0, tcp_dport),
+	/* in linux iphdr tcp_flags is 8 bits long */
+	OFFLOAD(TCP_FLAGS,  8,  U8_MAX, tcp.ack_seq, 5, tcp_flags),
+
+	OFFLOAD(UDP_SPORT, 16, U16_MAX, udp.source, 0, udp_sport),
+	OFFLOAD(UDP_DPORT, 16, U16_MAX, udp.dest,   0, udp_dport),
+};
+
+static u32 mask_field_get(void *mask, struct mlx5_fields *f)
+{
+	switch (f->field_bsize) {
+	case 32:
+		return be32_to_cpu(*(__be32 *)mask) & f->field_mask;
+	case 16:
+		return be16_to_cpu(*(__be16 *)mask) & (u16)f->field_mask;
+	default:
+		return *(u8 *)mask & (u8)f->field_mask;
+	}
+}
+
+static void mask_field_clear(void *mask, struct mlx5_fields *f)
+{
+	switch (f->field_bsize) {
+	case 32:
+		*(__be32 *)mask &= ~cpu_to_be32(f->field_mask);
+		break;
+	case 16:
+		*(__be16 *)mask &= ~cpu_to_be16((u16)f->field_mask);
+		break;
+	default:
+		*(u8 *)mask &= ~(u8)f->field_mask;
+		break;
+	}
+}
+
+static int offload_pedit_fields(struct mlx5e_priv *priv,
+				int namespace,
+				struct mlx5e_tc_flow_parse_attr *parse_attr,
+				u32 *action_flags,
+				struct netlink_ext_ack *extack)
+{
+	struct pedit_headers *set_masks, *add_masks, *set_vals, *add_vals;
+	struct pedit_headers_action *hdrs = parse_attr->hdrs;
+	void *headers_c, *headers_v, *action, *vals_p;
+	struct mlx5e_tc_mod_hdr_acts *mod_acts;
+	void *s_masks_p, *a_masks_p;
+	int i, first, last, next_z;
+	struct mlx5_fields *f;
+	unsigned long mask;
+	u32 s_mask, a_mask;
+	u8 cmd;
+
+	mod_acts = &parse_attr->mod_hdr_acts;
+	headers_c = mlx5e_get_match_headers_criteria(*action_flags, &parse_attr->spec);
+	headers_v = mlx5e_get_match_headers_value(*action_flags, &parse_attr->spec);
+
+	set_masks = &hdrs[0].masks;
+	add_masks = &hdrs[1].masks;
+	set_vals = &hdrs[0].vals;
+	add_vals = &hdrs[1].vals;
+
+	for (i = 0; i < ARRAY_SIZE(fields); i++) {
+		bool skip;
+
+		f = &fields[i];
+		s_masks_p = (void *)set_masks + f->offset;
+		a_masks_p = (void *)add_masks + f->offset;
+
+		s_mask = mask_field_get(s_masks_p, f);
+		a_mask = mask_field_get(a_masks_p, f);
+
+		if (!s_mask && !a_mask) /* nothing to offload here */
+			continue;
+
+		if (s_mask && a_mask) {
+			NL_SET_ERR_MSG_MOD(extack,
+					   "can't set and add to the same HW field");
+			netdev_warn(priv->netdev,
+				    "mlx5: can't set and add to the same HW field (%x)\n",
+				    f->field);
+			return -EOPNOTSUPP;
+		}
+
+		skip = false;
+		if (s_mask) {
+			void *match_mask = headers_c + f->match_offset;
+			void *match_val = headers_v + f->match_offset;
+
+			cmd  = MLX5_ACTION_TYPE_SET;
+			mask = s_mask;
+			vals_p = (void *)set_vals + f->offset;
+			/* don't rewrite if we have a match on the same value */
+			if (cmp_val_mask(vals_p, s_masks_p, match_val,
+					 match_mask, f->field_bsize))
+				skip = true;
+			/* clear to denote we consumed this field */
+			mask_field_clear(s_masks_p, f);
+		} else {
+			cmd  = MLX5_ACTION_TYPE_ADD;
+			mask = a_mask;
+			vals_p = (void *)add_vals + f->offset;
+			/* add 0 is no change */
+			if (!mask_field_get(vals_p, f))
+				skip = true;
+			/* clear to denote we consumed this field */
+			mask_field_clear(a_masks_p, f);
+		}
+		if (skip)
+			continue;
+
+		first = find_first_bit(&mask, f->field_bsize);
+		next_z = find_next_zero_bit(&mask, f->field_bsize, first);
+		last  = find_last_bit(&mask, f->field_bsize);
+		if (first < next_z && next_z < last) {
+			NL_SET_ERR_MSG_MOD(extack,
+					   "rewrite of few sub-fields isn't supported");
+			netdev_warn(priv->netdev,
+				    "mlx5: rewrite of few sub-fields (mask %lx) isn't offloaded\n",
+				    mask);
+			return -EOPNOTSUPP;
+		}
+
+		action = mlx5e_mod_hdr_alloc(priv->mdev, namespace, mod_acts);
+		if (IS_ERR(action)) {
+			NL_SET_ERR_MSG_MOD(extack,
+					   "too many pedit actions, can't offload");
+			mlx5_core_warn(priv->mdev,
+				       "mlx5: parsed %d pedit actions, can't do more\n",
+				       mod_acts->num_actions);
+			return PTR_ERR(action);
+		}
+
+		MLX5_SET(set_action_in, action, action_type, cmd);
+		MLX5_SET(set_action_in, action, field, f->field);
+
+		if (cmd == MLX5_ACTION_TYPE_SET) {
+			unsigned long field_mask = f->field_mask;
+			int start;
+
+			/* if field is bit sized it can start not from first bit */
+			start = find_first_bit(&field_mask, f->field_bsize);
+
+			MLX5_SET(set_action_in, action, offset, first - start);
+			/* length is num of bits to be written, zero means length of 32 */
+			MLX5_SET(set_action_in, action, length, (last - first + 1));
+		}
+
+		if (f->field_bsize == 32)
+			MLX5_SET(set_action_in, action, data, ntohl(*(__be32 *)vals_p) >> first);
+		else if (f->field_bsize == 16)
+			MLX5_SET(set_action_in, action, data, ntohs(*(__be16 *)vals_p) >> first);
+		else if (f->field_bsize == 8)
+			MLX5_SET(set_action_in, action, data, *(u8 *)vals_p >> first);
+
+		++mod_acts->num_actions;
+	}
+
+	return 0;
+}
+
+static const struct pedit_headers zero_masks = {};
+
+static int verify_offload_pedit_fields(struct mlx5e_priv *priv,
+				       struct mlx5e_tc_flow_parse_attr *parse_attr,
+				       struct netlink_ext_ack *extack)
+{
+	struct pedit_headers *cmd_masks;
+	u8 cmd;
+
+	for (cmd = 0; cmd < __PEDIT_CMD_MAX; cmd++) {
+		cmd_masks = &parse_attr->hdrs[cmd].masks;
+		if (memcmp(cmd_masks, &zero_masks, sizeof(zero_masks))) {
+			NL_SET_ERR_MSG_MOD(extack, "attempt to offload an unsupported field");
+			netdev_warn(priv->netdev, "attempt to offload an unsupported field (cmd %d)\n", cmd);
+			print_hex_dump(KERN_WARNING, "mask: ", DUMP_PREFIX_ADDRESS,
+				       16, 1, cmd_masks, sizeof(zero_masks), true);
+			return -EOPNOTSUPP;
+		}
+	}
+
+	return 0;
+}
+
+static int alloc_tc_pedit_action(struct mlx5e_priv *priv, int namespace,
+				 struct mlx5e_tc_flow_parse_attr *parse_attr,
+				 u32 *action_flags,
+				 struct netlink_ext_ack *extack)
+{
+	int err;
+
+	err = offload_pedit_fields(priv, namespace, parse_attr, action_flags, extack);
+	if (err)
+		goto out_dealloc_parsed_actions;
+
+	err = verify_offload_pedit_fields(priv, parse_attr, extack);
+	if (err)
+		goto out_dealloc_parsed_actions;
+
+	return 0;
+
+out_dealloc_parsed_actions:
+	mlx5e_mod_hdr_dealloc(&parse_attr->mod_hdr_acts);
+	return err;
+}
+
+struct ip_ttl_word {
+	__u8	ttl;
+	__u8	protocol;
+	__sum16	check;
+};
+
+struct ipv6_hoplimit_word {
+	__be16	payload_len;
+	__u8	nexthdr;
+	__u8	hop_limit;
+};
+
+static bool
+is_action_keys_supported(const struct flow_action_entry *act, bool ct_flow,
+			 bool *modify_ip_header, bool *modify_tuple,
+			 struct netlink_ext_ack *extack)
+{
+	u32 mask, offset;
+	u8 htype;
+
+	htype = act->mangle.htype;
+	offset = act->mangle.offset;
+	mask = ~act->mangle.mask;
+	/* For IPv4 & IPv6 header check 4 byte word,
+	 * to determine that modified fields
+	 * are NOT ttl & hop_limit only.
+	 */
+	if (htype == FLOW_ACT_MANGLE_HDR_TYPE_IP4) {
+		struct ip_ttl_word *ttl_word =
+			(struct ip_ttl_word *)&mask;
+
+		if (offset != offsetof(struct iphdr, ttl) ||
+		    ttl_word->protocol ||
+		    ttl_word->check) {
+			*modify_ip_header = true;
+		}
+
+		if (offset >= offsetof(struct iphdr, saddr))
+			*modify_tuple = true;
+
+		if (ct_flow && *modify_tuple) {
+			NL_SET_ERR_MSG_MOD(extack,
+					   "can't offload re-write of ipv4 address with action ct");
+			return false;
+		}
+	} else if (htype == FLOW_ACT_MANGLE_HDR_TYPE_IP6) {
+		struct ipv6_hoplimit_word *hoplimit_word =
+			(struct ipv6_hoplimit_word *)&mask;
+
+		if (offset != offsetof(struct ipv6hdr, payload_len) ||
+		    hoplimit_word->payload_len ||
+		    hoplimit_word->nexthdr) {
+			*modify_ip_header = true;
+		}
+
+		if (ct_flow && offset >= offsetof(struct ipv6hdr, saddr))
+			*modify_tuple = true;
+
+		if (ct_flow && *modify_tuple) {
+			NL_SET_ERR_MSG_MOD(extack,
+					   "can't offload re-write of ipv6 address with action ct");
+			return false;
+		}
+	} else if (htype == FLOW_ACT_MANGLE_HDR_TYPE_TCP ||
+		   htype == FLOW_ACT_MANGLE_HDR_TYPE_UDP) {
+		*modify_tuple = true;
+		if (ct_flow) {
+			NL_SET_ERR_MSG_MOD(extack,
+					   "can't offload re-write of transport header ports with action ct");
+			return false;
+		}
+	}
+
+	return true;
+}
+
+static bool modify_tuple_supported(bool modify_tuple, bool ct_clear,
+				   bool ct_flow, struct netlink_ext_ack *extack,
+				   struct mlx5e_priv *priv,
+				   struct mlx5_flow_spec *spec)
+{
+	if (!modify_tuple || ct_clear)
+		return true;
+
+	if (ct_flow) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "can't offload tuple modification with non-clear ct()");
+		netdev_info(priv->netdev,
+			    "can't offload tuple modification with non-clear ct()");
+		return false;
+	}
+
+	/* Add ct_state=-trk match so it will be offloaded for non ct flows
+	 * (or after clear action), as otherwise, since the tuple is changed,
+	 * we can't restore ct state
+	 */
+	if (mlx5_tc_ct_add_no_trk_match(spec)) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "can't offload tuple modification with ct matches and no ct(clear) action");
+		netdev_info(priv->netdev,
+			    "can't offload tuple modification with ct matches and no ct(clear) action");
+		return false;
+	}
+
+	return true;
+}
+
+static bool modify_header_match_supported(struct mlx5e_priv *priv,
+					  struct mlx5_flow_spec *spec,
+					  struct flow_action *flow_action,
+					  u32 actions, bool ct_flow,
+					  bool ct_clear,
+					  struct netlink_ext_ack *extack)
+{
+	const struct flow_action_entry *act;
+	bool modify_ip_header, modify_tuple;
+	void *headers_c;
+	void *headers_v;
+	u16 ethertype;
+	u8 ip_proto;
+	int i;
+
+	headers_c = mlx5e_get_match_headers_criteria(actions, spec);
+	headers_v = mlx5e_get_match_headers_value(actions, spec);
+	ethertype = MLX5_GET(fte_match_set_lyr_2_4, headers_v, ethertype);
+
+	/* for non-IP we only re-write MACs, so we're okay */
+	if (MLX5_GET(fte_match_set_lyr_2_4, headers_c, ip_version) == 0 &&
+	    ethertype != ETH_P_IP && ethertype != ETH_P_IPV6)
+		goto out_ok;
+
+	modify_ip_header = false;
+	modify_tuple = false;
+	flow_action_for_each(i, act, flow_action) {
+		if (act->id != FLOW_ACTION_MANGLE &&
+		    act->id != FLOW_ACTION_ADD)
+			continue;
+
+		if (!is_action_keys_supported(act, ct_flow,
+					      &modify_ip_header,
+					      &modify_tuple, extack))
+			return false;
+	}
+
+	if (!modify_tuple_supported(modify_tuple, ct_clear, ct_flow, extack,
+				    priv, spec))
+		return false;
+
+	ip_proto = MLX5_GET(fte_match_set_lyr_2_4, headers_v, ip_protocol);
+	if (modify_ip_header && ip_proto != IPPROTO_TCP &&
+	    ip_proto != IPPROTO_UDP && ip_proto != IPPROTO_ICMP) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "can't offload re-write of non TCP/UDP");
+		netdev_info(priv->netdev, "can't offload re-write of ip proto %d\n",
+			    ip_proto);
+		return false;
+	}
+
+out_ok:
+	return true;
+}
+
+static bool
+actions_match_supported_fdb(struct mlx5e_priv *priv,
+			    struct mlx5e_tc_flow_parse_attr *parse_attr,
+			    struct mlx5e_tc_flow *flow,
+			    struct netlink_ext_ack *extack)
+{
+	struct mlx5_esw_flow_attr *esw_attr = flow->attr->esw_attr;
+	bool ct_flow, ct_clear;
+
+	ct_clear = flow->attr->ct_attr.ct_action & TCA_CT_ACT_CLEAR;
+	ct_flow = flow_flag_test(flow, CT) && !ct_clear;
+
+	if (esw_attr->split_count && ct_flow &&
+	    !MLX5_CAP_GEN(esw_attr->in_mdev, reg_c_preserve)) {
+		/* All registers used by ct are cleared when using
+		 * split rules.
+		 */
+		NL_SET_ERR_MSG_MOD(extack, "Can't offload mirroring with action ct");
+		return false;
+	}
+
+	if (esw_attr->split_count > 0 && !mlx5_esw_has_fwd_fdb(priv->mdev)) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "current firmware doesn't support split rule for port mirroring");
+		netdev_warn_once(priv->netdev,
+				 "current firmware doesn't support split rule for port mirroring\n");
+		return false;
+	}
+
+	return true;
+}
+
+static bool
+actions_match_supported(struct mlx5e_priv *priv,
+			struct flow_action *flow_action,
+			u32 actions,
+			struct mlx5e_tc_flow_parse_attr *parse_attr,
+			struct mlx5e_tc_flow *flow,
+			struct netlink_ext_ack *extack)
+{
+	bool ct_flow, ct_clear;
+
+	ct_clear = flow->attr->ct_attr.ct_action & TCA_CT_ACT_CLEAR;
+	ct_flow = flow_flag_test(flow, CT) && !ct_clear;
+
+	if (!(actions &
+	      (MLX5_FLOW_CONTEXT_ACTION_FWD_DEST | MLX5_FLOW_CONTEXT_ACTION_DROP))) {
+		NL_SET_ERR_MSG_MOD(extack, "Rule must have at least one forward/drop action");
+		return false;
+	}
+
+	if (!(~actions &
+	      (MLX5_FLOW_CONTEXT_ACTION_FWD_DEST | MLX5_FLOW_CONTEXT_ACTION_DROP))) {
+		NL_SET_ERR_MSG_MOD(extack, "Rule cannot support forward+drop action");
+		return false;
+	}
+
+	if (actions & MLX5_FLOW_CONTEXT_ACTION_MOD_HDR &&
+	    actions & MLX5_FLOW_CONTEXT_ACTION_DROP) {
+		NL_SET_ERR_MSG_MOD(extack, "Drop with modify header action is not supported");
+		return false;
+	}
+
+	if (!(~actions &
+	      (MLX5_FLOW_CONTEXT_ACTION_FWD_DEST | MLX5_FLOW_CONTEXT_ACTION_DROP))) {
+		NL_SET_ERR_MSG_MOD(extack, "Rule cannot support forward+drop action");
+		return false;
+	}
+
+	if (actions & MLX5_FLOW_CONTEXT_ACTION_MOD_HDR &&
+	    actions & MLX5_FLOW_CONTEXT_ACTION_DROP) {
+		NL_SET_ERR_MSG_MOD(extack, "Drop with modify header action is not supported");
+		return false;
+	}
+
+	if (actions & MLX5_FLOW_CONTEXT_ACTION_MOD_HDR &&
+	    !modify_header_match_supported(priv, &parse_attr->spec, flow_action,
+					   actions, ct_flow, ct_clear, extack))
+		return false;
+
+	if (mlx5e_is_eswitch_flow(flow) &&
+	    !actions_match_supported_fdb(priv, parse_attr, flow, extack))
+		return false;
+
+	return true;
+}
+
+static bool same_port_devs(struct mlx5e_priv *priv, struct mlx5e_priv *peer_priv)
+{
+	return priv->mdev == peer_priv->mdev;
+}
+
+bool mlx5e_same_hw_devs(struct mlx5e_priv *priv, struct mlx5e_priv *peer_priv)
+{
+	struct mlx5_core_dev *fmdev, *pmdev;
+	u64 fsystem_guid, psystem_guid;
+
+	fmdev = priv->mdev;
+	pmdev = peer_priv->mdev;
+
+	fsystem_guid = mlx5_query_nic_system_image_guid(fmdev);
+	psystem_guid = mlx5_query_nic_system_image_guid(pmdev);
+
+	return (fsystem_guid == psystem_guid);
+}
+
+static int
+actions_prepare_mod_hdr_actions(struct mlx5e_priv *priv,
+				struct mlx5e_tc_flow *flow,
+				struct mlx5_flow_attr *attr,
+				struct netlink_ext_ack *extack)
+{
+	struct mlx5e_tc_flow_parse_attr *parse_attr = attr->parse_attr;
+	struct pedit_headers_action *hdrs = parse_attr->hdrs;
+	enum mlx5_flow_namespace_type ns_type;
+	int err;
+
+	if (!hdrs[TCA_PEDIT_KEY_EX_CMD_SET].pedits &&
+	    !hdrs[TCA_PEDIT_KEY_EX_CMD_ADD].pedits)
+		return 0;
+
+	ns_type = mlx5e_get_flow_namespace(flow);
+
+	err = alloc_tc_pedit_action(priv, ns_type, parse_attr, &attr->action, extack);
+	if (err)
+		return err;
+
+	if (parse_attr->mod_hdr_acts.num_actions > 0)
+		return 0;
+
+	/* In case all pedit actions are skipped, remove the MOD_HDR flag. */
+	attr->action &= ~MLX5_FLOW_CONTEXT_ACTION_MOD_HDR;
+	mlx5e_mod_hdr_dealloc(&parse_attr->mod_hdr_acts);
+
+	if (ns_type != MLX5_FLOW_NAMESPACE_FDB)
+		return 0;
+
+	if (!((attr->action & MLX5_FLOW_CONTEXT_ACTION_VLAN_POP) ||
+	      (attr->action & MLX5_FLOW_CONTEXT_ACTION_VLAN_PUSH)))
+		attr->esw_attr->split_count = 0;
+
+	return 0;
+}
+
+static struct mlx5_flow_attr*
+mlx5e_clone_flow_attr_for_post_act(struct mlx5_flow_attr *attr,
+				   enum mlx5_flow_namespace_type ns_type)
+{
+	struct mlx5e_tc_flow_parse_attr *parse_attr;
+	u32 attr_sz = ns_to_attr_sz(ns_type);
+	struct mlx5_flow_attr *attr2;
+
+	attr2 = mlx5_alloc_flow_attr(ns_type);
+	parse_attr = kvzalloc(sizeof(*parse_attr), GFP_KERNEL);
+	if (!attr2 || !parse_attr) {
+		kvfree(parse_attr);
+		kfree(attr2);
+		return NULL;
+	}
+
+	memcpy(attr2, attr, attr_sz);
+	INIT_LIST_HEAD(&attr2->list);
+	parse_attr->filter_dev = attr->parse_attr->filter_dev;
+	attr2->action = 0;
+	attr2->flags = 0;
+	attr2->parse_attr = parse_attr;
+	attr2->dest_chain = 0;
+	attr2->dest_ft = NULL;
+
+	if (ns_type == MLX5_FLOW_NAMESPACE_FDB) {
+		attr2->esw_attr->out_count = 0;
+		attr2->esw_attr->split_count = 0;
+	}
+
+	return attr2;
+}
+
+static struct mlx5_core_dev *
+get_flow_counter_dev(struct mlx5e_tc_flow *flow)
+{
+	return mlx5e_is_eswitch_flow(flow) ? flow->attr->esw_attr->counter_dev : flow->priv->mdev;
+}
+
+struct mlx5_flow_attr *
+mlx5e_tc_get_encap_attr(struct mlx5e_tc_flow *flow)
+{
+	struct mlx5_esw_flow_attr *esw_attr;
+	struct mlx5_flow_attr *attr;
+	int i;
+
+	list_for_each_entry(attr, &flow->attrs, list) {
+		esw_attr = attr->esw_attr;
+		for (i = 0; i < MLX5_MAX_FLOW_FWD_VPORTS; i++) {
+			if (esw_attr->dests[i].flags & MLX5_ESW_DEST_ENCAP)
+				return attr;
+		}
+	}
+
+	return NULL;
+}
+
+void
+mlx5e_tc_unoffload_flow_post_acts(struct mlx5e_tc_flow *flow)
+{
+	struct mlx5e_post_act *post_act = get_post_action(flow->priv);
+	struct mlx5_flow_attr *attr;
+
+	list_for_each_entry(attr, &flow->attrs, list) {
+		if (list_is_last(&attr->list, &flow->attrs))
+			break;
+
+		mlx5e_tc_post_act_unoffload(post_act, attr->post_act_handle);
+	}
+}
+
+static void
+free_flow_post_acts(struct mlx5e_tc_flow *flow)
+{
+	struct mlx5_core_dev *counter_dev = get_flow_counter_dev(flow);
+	struct mlx5e_post_act *post_act = get_post_action(flow->priv);
+	struct mlx5_flow_attr *attr, *tmp;
+	bool vf_tun;
+
+	list_for_each_entry_safe(attr, tmp, &flow->attrs, list) {
+		if (list_is_last(&attr->list, &flow->attrs))
+			break;
+
+		if (attr->post_act_handle)
+			mlx5e_tc_post_act_del(post_act, attr->post_act_handle);
+
+		clean_encap_dests(flow->priv, flow, attr, &vf_tun);
+
+		if (attr->action & MLX5_FLOW_CONTEXT_ACTION_COUNT)
+			mlx5_fc_destroy(counter_dev, attr->counter);
+
+		if (attr->action & MLX5_FLOW_CONTEXT_ACTION_MOD_HDR) {
+			mlx5e_mod_hdr_dealloc(&attr->parse_attr->mod_hdr_acts);
+			if (attr->modify_hdr)
+				mlx5_modify_header_dealloc(flow->priv->mdev, attr->modify_hdr);
+		}
+
+		list_del(&attr->list);
+		kvfree(attr->parse_attr);
+		kfree(attr);
+	}
+}
+
+int
+mlx5e_tc_offload_flow_post_acts(struct mlx5e_tc_flow *flow)
+{
+	struct mlx5e_post_act *post_act = get_post_action(flow->priv);
+	struct mlx5_flow_attr *attr;
+	int err = 0;
+
+	list_for_each_entry(attr, &flow->attrs, list) {
+		if (list_is_last(&attr->list, &flow->attrs))
+			break;
+
+		err = mlx5e_tc_post_act_offload(post_act, attr->post_act_handle);
+		if (err)
+			break;
+	}
+
+	return err;
+}
+
+/* TC filter rule HW translation:
+ *
+ * +---------------------+
+ * + ft prio (tc chain)  +
+ * + original match      +
+ * +---------------------+
+ *           |
+ *           | if multi table action
+ *           |
+ *           v
+ * +---------------------+
+ * + post act ft         |<----.
+ * + match fte id        |     | split on multi table action
+ * + do actions          |-----'
+ * +---------------------+
+ *           |
+ *           |
+ *           v
+ * Do rest of the actions after last multi table action.
+ */
+static int
+alloc_flow_post_acts(struct mlx5e_tc_flow *flow, struct netlink_ext_ack *extack)
+{
+	struct mlx5e_post_act *post_act = get_post_action(flow->priv);
+	struct mlx5_flow_attr *attr, *next_attr = NULL;
+	struct mlx5e_post_act_handle *handle;
+	bool vf_tun;
+	int err;
+
+	/* This is going in reverse order as needed.
+	 * The first entry is the last attribute.
+	 */
+	list_for_each_entry(attr, &flow->attrs, list) {
+		if (!next_attr) {
+			/* Set counter action on last post act rule. */
+			attr->action |= MLX5_FLOW_CONTEXT_ACTION_COUNT;
+		} else {
+			err = mlx5e_tc_act_set_next_post_act(flow, attr, next_attr);
+			if (err)
+				goto out_free;
+		}
+
+		/* Don't add post_act rule for first attr (last in the list).
+		 * It's being handled by the caller.
+		 */
+		if (list_is_last(&attr->list, &flow->attrs))
+			break;
+
+		err = set_encap_dests(flow->priv, flow, attr, extack, &vf_tun);
+		if (err)
+			goto out_free;
+
+		err = actions_prepare_mod_hdr_actions(flow->priv, flow, attr, extack);
+		if (err)
+			goto out_free;
+
+		if (attr->action & MLX5_FLOW_CONTEXT_ACTION_MOD_HDR) {
+			err = mlx5e_tc_add_flow_mod_hdr(flow->priv, flow, attr);
+			if (err)
+				goto out_free;
+		}
+
+		if (attr->action & MLX5_FLOW_CONTEXT_ACTION_COUNT) {
+			err = alloc_flow_attr_counter(get_flow_counter_dev(flow), attr);
+			if (err)
+				goto out_free;
+		}
+
+		handle = mlx5e_tc_post_act_add(post_act, attr);
+		if (IS_ERR(handle)) {
+			err = PTR_ERR(handle);
+			goto out_free;
+		}
+
+		attr->post_act_handle = handle;
+		next_attr = attr;
+	}
+
+	if (flow_flag_test(flow, SLOW))
+		goto out;
+
+	err = mlx5e_tc_offload_flow_post_acts(flow);
+	if (err)
+		goto out_free;
+
+out:
+	return 0;
+
+out_free:
+	free_flow_post_acts(flow);
+	return err;
+}
+
+static int
+parse_tc_actions(struct mlx5e_tc_act_parse_state *parse_state,
+		 struct flow_action *flow_action)
+{
+	struct netlink_ext_ack *extack = parse_state->extack;
+	struct mlx5e_tc_flow_action flow_action_reorder;
+	struct mlx5e_tc_flow *flow = parse_state->flow;
+	struct mlx5_flow_attr *attr = flow->attr;
+	enum mlx5_flow_namespace_type ns_type;
+	struct mlx5e_priv *priv = flow->priv;
+	struct flow_action_entry *act, **_act;
+	struct mlx5e_tc_act *tc_act;
+	int err, i;
+
+	flow_action_reorder.num_entries = flow_action->num_entries;
+	flow_action_reorder.entries = kcalloc(flow_action->num_entries,
+					      sizeof(flow_action), GFP_KERNEL);
+	if (!flow_action_reorder.entries)
+		return -ENOMEM;
+
+	mlx5e_tc_act_reorder_flow_actions(flow_action, &flow_action_reorder);
+
+	ns_type = mlx5e_get_flow_namespace(flow);
+	list_add(&attr->list, &flow->attrs);
+
+	flow_action_for_each(i, _act, &flow_action_reorder) {
+		act = *_act;
+		tc_act = mlx5e_tc_act_get(act->id, ns_type);
+		if (!tc_act) {
+			NL_SET_ERR_MSG_MOD(extack, "Not implemented offload action");
+			err = -EOPNOTSUPP;
+			goto out_free;
+		}
+
+		if (!tc_act->can_offload(parse_state, act, i, attr)) {
+			err = -EOPNOTSUPP;
+			goto out_free;
+		}
+
+		err = tc_act->parse_action(parse_state, act, priv, attr);
+		if (err)
+			goto out_free;
+
+		parse_state->actions |= attr->action;
+
+		/* Split attr for multi table act if not the last act. */
+		if (tc_act->is_multi_table_act &&
+		    tc_act->is_multi_table_act(priv, act, attr) &&
+		    i < flow_action_reorder.num_entries - 1) {
+			err = mlx5e_tc_act_post_parse(parse_state, flow_action, attr, ns_type);
+			if (err)
+				goto out_free;
+
+			attr = mlx5e_clone_flow_attr_for_post_act(flow->attr, ns_type);
+			if (!attr) {
+				err = -ENOMEM;
+				goto out_free;
+			}
+
+			list_add(&attr->list, &flow->attrs);
+		}
+	}
+
+	kfree(flow_action_reorder.entries);
+
+	err = mlx5e_tc_act_post_parse(parse_state, flow_action, attr, ns_type);
+	if (err)
+		goto out_free_post_acts;
+
+	err = alloc_flow_post_acts(flow, extack);
+	if (err)
+		goto out_free_post_acts;
+
+	return 0;
+
+out_free:
+	kfree(flow_action_reorder.entries);
+out_free_post_acts:
+	free_flow_post_acts(flow);
+
+	return err;
+}
+
+static int
+flow_action_supported(struct flow_action *flow_action,
+		      struct netlink_ext_ack *extack)
+{
+	if (!flow_action_has_entries(flow_action)) {
+		NL_SET_ERR_MSG_MOD(extack, "Flow action doesn't have any entries");
+		return -EINVAL;
+	}
+
+	if (!flow_action_hw_stats_check(flow_action, extack,
+					FLOW_ACTION_HW_STATS_DELAYED_BIT)) {
+		NL_SET_ERR_MSG_MOD(extack, "Flow action HW stats type is not supported");
+		return -EOPNOTSUPP;
+	}
+
+	return 0;
+}
+
+static int
+parse_tc_nic_actions(struct mlx5e_priv *priv,
+		     struct flow_action *flow_action,
+		     struct mlx5e_tc_flow *flow,
+		     struct netlink_ext_ack *extack)
+{
+	struct mlx5e_tc_act_parse_state *parse_state;
+	struct mlx5e_tc_flow_parse_attr *parse_attr;
+	struct mlx5_flow_attr *attr = flow->attr;
+	int err;
+
+	err = flow_action_supported(flow_action, extack);
+	if (err)
+		return err;
+
+	attr->nic_attr->flow_tag = MLX5_FS_DEFAULT_FLOW_TAG;
+	parse_attr = attr->parse_attr;
+	parse_state = &parse_attr->parse_state;
+	mlx5e_tc_act_init_parse_state(parse_state, flow, flow_action, extack);
+	parse_state->ct_priv = get_ct_priv(priv);
+
+	err = parse_tc_actions(parse_state, flow_action);
+	if (err)
+		return err;
+
+	err = actions_prepare_mod_hdr_actions(priv, flow, attr, extack);
+	if (err)
+		return err;
+
+	if (!actions_match_supported(priv, flow_action, parse_state->actions,
+				     parse_attr, flow, extack))
+		return -EOPNOTSUPP;
+
+	return 0;
+}
+
+static bool is_merged_eswitch_vfs(struct mlx5e_priv *priv,
+				  struct net_device *peer_netdev)
+{
+	struct mlx5e_priv *peer_priv;
+
+	peer_priv = netdev_priv(peer_netdev);
+
+	return (MLX5_CAP_ESW(priv->mdev, merged_eswitch) &&
+		mlx5e_eswitch_vf_rep(priv->netdev) &&
+		mlx5e_eswitch_vf_rep(peer_netdev) &&
+		mlx5e_same_hw_devs(priv, peer_priv));
+}
+
+static bool same_hw_reps(struct mlx5e_priv *priv,
+			 struct net_device *peer_netdev)
+{
+	struct mlx5e_priv *peer_priv;
+
+	peer_priv = netdev_priv(peer_netdev);
+
+	return mlx5e_eswitch_rep(priv->netdev) &&
+	       mlx5e_eswitch_rep(peer_netdev) &&
+	       mlx5e_same_hw_devs(priv, peer_priv);
+}
+
+static bool is_lag_dev(struct mlx5e_priv *priv,
+		       struct net_device *peer_netdev)
+{
+	return ((mlx5_lag_is_sriov(priv->mdev) ||
+		 mlx5_lag_is_multipath(priv->mdev)) &&
+		 same_hw_reps(priv, peer_netdev));
+}
+
+static bool is_multiport_eligible(struct mlx5e_priv *priv, struct net_device *out_dev)
+{
+	if (same_hw_reps(priv, out_dev) &&
+	    MLX5_CAP_PORT_SELECTION(priv->mdev, port_select_flow_table) &&
+	    MLX5_CAP_GEN(priv->mdev, create_lag_when_not_master_up))
+		return true;
+
+	return false;
+}
+
+bool mlx5e_is_valid_eswitch_fwd_dev(struct mlx5e_priv *priv,
+				    struct net_device *out_dev)
+{
+	if (is_merged_eswitch_vfs(priv, out_dev))
+		return true;
+
+	if (is_multiport_eligible(priv, out_dev))
+		return true;
+
+	if (is_lag_dev(priv, out_dev))
+		return true;
+
+	return mlx5e_eswitch_rep(out_dev) &&
+	       same_port_devs(priv, netdev_priv(out_dev));
+}
+
+int mlx5e_set_fwd_to_int_port_actions(struct mlx5e_priv *priv,
+				      struct mlx5_flow_attr *attr,
+				      int ifindex,
+				      enum mlx5e_tc_int_port_type type,
+				      u32 *action,
+				      int out_index)
+{
+	struct mlx5_esw_flow_attr *esw_attr = attr->esw_attr;
+	struct mlx5e_tc_int_port_priv *int_port_priv;
+	struct mlx5e_tc_flow_parse_attr *parse_attr;
+	struct mlx5e_tc_int_port *dest_int_port;
+	int err;
+
+	parse_attr = attr->parse_attr;
+	int_port_priv = mlx5e_get_int_port_priv(priv);
+
+	dest_int_port = mlx5e_tc_int_port_get(int_port_priv, ifindex, type);
+	if (IS_ERR(dest_int_port))
+		return PTR_ERR(dest_int_port);
+
+	err = mlx5e_tc_match_to_reg_set(priv->mdev, &parse_attr->mod_hdr_acts,
+					MLX5_FLOW_NAMESPACE_FDB, VPORT_TO_REG,
+					mlx5e_tc_int_port_get_metadata(dest_int_port));
+	if (err) {
+		mlx5e_tc_int_port_put(int_port_priv, dest_int_port);
+		return err;
+	}
+
+	*action |= MLX5_FLOW_CONTEXT_ACTION_MOD_HDR;
+
+	esw_attr->dest_int_port = dest_int_port;
+	esw_attr->dests[out_index].flags |= MLX5_ESW_DEST_CHAIN_WITH_SRC_PORT_CHANGE;
+	esw_attr->split_count = out_index;
+
+	/* Forward to root fdb for matching against the new source vport */
+	attr->dest_chain = 0;
+
+	return 0;
+}
+
+static int
+parse_tc_fdb_actions(struct mlx5e_priv *priv,
+		     struct flow_action *flow_action,
+		     struct mlx5e_tc_flow *flow,
+		     struct netlink_ext_ack *extack)
+{
+	struct mlx5e_tc_act_parse_state *parse_state;
+	struct mlx5e_tc_flow_parse_attr *parse_attr;
+	struct mlx5_flow_attr *attr = flow->attr;
+	struct mlx5_esw_flow_attr *esw_attr;
+	struct net_device *filter_dev;
+	int err;
+
+	err = flow_action_supported(flow_action, extack);
+	if (err)
+		return err;
+
+	esw_attr = attr->esw_attr;
+	parse_attr = attr->parse_attr;
+	filter_dev = parse_attr->filter_dev;
+	parse_state = &parse_attr->parse_state;
+	mlx5e_tc_act_init_parse_state(parse_state, flow, flow_action, extack);
+	parse_state->ct_priv = get_ct_priv(priv);
+
+	err = parse_tc_actions(parse_state, flow_action);
+	if (err)
+		return err;
+
+	/* Forward to/from internal port can only have 1 dest */
+	if ((netif_is_ovs_master(filter_dev) || esw_attr->dest_int_port) &&
+	    esw_attr->out_count > 1) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "Rules with internal port can have only one destination");
+		return -EOPNOTSUPP;
+	}
+
+	/* Forward from tunnel/internal port to internal port is not supported */
+	if ((mlx5e_get_tc_tun(filter_dev) || netif_is_ovs_master(filter_dev)) &&
+	    esw_attr->dest_int_port) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "Forwarding from tunnel/internal port to internal port is not supported");
+		return -EOPNOTSUPP;
+	}
+
+	err = actions_prepare_mod_hdr_actions(priv, flow, attr, extack);
+	if (err)
+		return err;
+
+	if (!actions_match_supported(priv, flow_action, parse_state->actions,
+				     parse_attr, flow, extack))
+		return -EOPNOTSUPP;
+
+	return 0;
+}
+
+static void get_flags(int flags, unsigned long *flow_flags)
+{
+	unsigned long __flow_flags = 0;
+
+	if (flags & MLX5_TC_FLAG(INGRESS))
+		__flow_flags |= BIT(MLX5E_TC_FLOW_FLAG_INGRESS);
+	if (flags & MLX5_TC_FLAG(EGRESS))
+		__flow_flags |= BIT(MLX5E_TC_FLOW_FLAG_EGRESS);
+
+	if (flags & MLX5_TC_FLAG(ESW_OFFLOAD))
+		__flow_flags |= BIT(MLX5E_TC_FLOW_FLAG_ESWITCH);
+	if (flags & MLX5_TC_FLAG(NIC_OFFLOAD))
+		__flow_flags |= BIT(MLX5E_TC_FLOW_FLAG_NIC);
+	if (flags & MLX5_TC_FLAG(FT_OFFLOAD))
+		__flow_flags |= BIT(MLX5E_TC_FLOW_FLAG_FT);
+
+	*flow_flags = __flow_flags;
+}
+
+static const struct rhashtable_params tc_ht_params = {
+	.head_offset = offsetof(struct mlx5e_tc_flow, node),
+	.key_offset = offsetof(struct mlx5e_tc_flow, cookie),
+	.key_len = sizeof(((struct mlx5e_tc_flow *)0)->cookie),
+	.automatic_shrinking = true,
+};
+
+static struct rhashtable *get_tc_ht(struct mlx5e_priv *priv,
+				    unsigned long flags)
+{
+	struct mlx5e_tc_table *tc = mlx5e_fs_get_tc(priv->fs);
+	struct mlx5e_rep_priv *rpriv;
+
+	if (flags & MLX5_TC_FLAG(ESW_OFFLOAD)) {
+		rpriv = priv->ppriv;
+		return &rpriv->tc_ht;
+	} else /* NIC offload */
+		return &tc->ht;
+}
+
+static bool is_peer_flow_needed(struct mlx5e_tc_flow *flow)
+{
+	struct mlx5_esw_flow_attr *esw_attr = flow->attr->esw_attr;
+	struct mlx5_flow_attr *attr = flow->attr;
+	bool is_rep_ingress = esw_attr->in_rep->vport != MLX5_VPORT_UPLINK &&
+		flow_flag_test(flow, INGRESS);
+	bool act_is_encap = !!(attr->action &
+			       MLX5_FLOW_CONTEXT_ACTION_PACKET_REFORMAT);
+	bool esw_paired = mlx5_devcom_is_paired(esw_attr->in_mdev->priv.devcom,
+						MLX5_DEVCOM_ESW_OFFLOADS);
+
+	if (!esw_paired)
+		return false;
+
+	if ((mlx5_lag_is_sriov(esw_attr->in_mdev) ||
+	     mlx5_lag_is_multipath(esw_attr->in_mdev)) &&
+	    (is_rep_ingress || act_is_encap))
+		return true;
+
+	return false;
+}
+
+struct mlx5_flow_attr *
+mlx5_alloc_flow_attr(enum mlx5_flow_namespace_type type)
+{
+	u32 ex_attr_size = (type == MLX5_FLOW_NAMESPACE_FDB)  ?
+				sizeof(struct mlx5_esw_flow_attr) :
+				sizeof(struct mlx5_nic_flow_attr);
+	struct mlx5_flow_attr *attr;
+
+	attr = kzalloc(sizeof(*attr) + ex_attr_size, GFP_KERNEL);
+	if (!attr)
+		return attr;
+
+	INIT_LIST_HEAD(&attr->list);
+	return attr;
+}
+
+static int
+mlx5e_alloc_flow(struct mlx5e_priv *priv, int attr_size,
+		 struct flow_cls_offload *f, unsigned long flow_flags,
+		 struct mlx5e_tc_flow_parse_attr **__parse_attr,
+		 struct mlx5e_tc_flow **__flow)
+{
+	struct mlx5e_tc_flow_parse_attr *parse_attr;
+	struct mlx5_flow_attr *attr;
+	struct mlx5e_tc_flow *flow;
+	int err = -ENOMEM;
+	int out_index;
+
+	flow = kzalloc(sizeof(*flow), GFP_KERNEL);
+	parse_attr = kvzalloc(sizeof(*parse_attr), GFP_KERNEL);
+	if (!parse_attr || !flow)
+		goto err_free;
+
+	flow->flags = flow_flags;
+	flow->cookie = f->cookie;
+	flow->priv = priv;
+
+	attr = mlx5_alloc_flow_attr(mlx5e_get_flow_namespace(flow));
+	if (!attr)
+		goto err_free;
+
+	flow->attr = attr;
+
+	for (out_index = 0; out_index < MLX5_MAX_FLOW_FWD_VPORTS; out_index++)
+		INIT_LIST_HEAD(&flow->encaps[out_index].list);
+	INIT_LIST_HEAD(&flow->hairpin);
+	INIT_LIST_HEAD(&flow->l3_to_l2_reformat);
+	INIT_LIST_HEAD(&flow->attrs);
+	refcount_set(&flow->refcnt, 1);
+	init_completion(&flow->init_done);
+	init_completion(&flow->del_hw_done);
+
+	*__flow = flow;
+	*__parse_attr = parse_attr;
+
+	return 0;
+
+err_free:
+	kfree(flow);
+	kvfree(parse_attr);
+	return err;
+}
+
+static void
+mlx5e_flow_attr_init(struct mlx5_flow_attr *attr,
+		     struct mlx5e_tc_flow_parse_attr *parse_attr,
+		     struct flow_cls_offload *f)
+{
+	attr->parse_attr = parse_attr;
+	attr->chain = f->common.chain_index;
+	attr->prio = f->common.prio;
+}
+
+static void
+mlx5e_flow_esw_attr_init(struct mlx5_flow_attr *attr,
+			 struct mlx5e_priv *priv,
+			 struct mlx5e_tc_flow_parse_attr *parse_attr,
+			 struct flow_cls_offload *f,
+			 struct mlx5_eswitch_rep *in_rep,
+			 struct mlx5_core_dev *in_mdev)
+{
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	struct mlx5_esw_flow_attr *esw_attr = attr->esw_attr;
+
+	mlx5e_flow_attr_init(attr, parse_attr, f);
+
+	esw_attr->in_rep = in_rep;
+	esw_attr->in_mdev = in_mdev;
+
+	if (MLX5_CAP_ESW(esw->dev, counter_eswitch_affinity) ==
+	    MLX5_COUNTER_SOURCE_ESWITCH)
+		esw_attr->counter_dev = in_mdev;
+	else
+		esw_attr->counter_dev = priv->mdev;
+}
+
+static struct mlx5e_tc_flow *
+__mlx5e_add_fdb_flow(struct mlx5e_priv *priv,
+		     struct flow_cls_offload *f,
+		     unsigned long flow_flags,
+		     struct net_device *filter_dev,
+		     struct mlx5_eswitch_rep *in_rep,
+		     struct mlx5_core_dev *in_mdev)
+{
+	struct flow_rule *rule = flow_cls_offload_flow_rule(f);
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	struct netlink_ext_ack *extack = f->common.extack;
+	struct mlx5e_tc_flow_parse_attr *parse_attr;
+	struct mlx5e_tc_flow *flow;
+	int attr_size, err;
+
+	flow_flags |= BIT(MLX5E_TC_FLOW_FLAG_ESWITCH);
+	attr_size  = sizeof(struct mlx5_esw_flow_attr);
+	err = mlx5e_alloc_flow(priv, attr_size, f, flow_flags,
+			       &parse_attr, &flow);
+	if (err)
+		goto out;
+
+	parse_attr->filter_dev = filter_dev;
+	mlx5e_flow_esw_attr_init(flow->attr,
+				 priv, parse_attr,
+				 f, in_rep, in_mdev);
+
+	err = parse_cls_flower(flow->priv, flow, &parse_attr->spec,
+			       f, filter_dev);
+	if (err)
+		goto err_free;
+
+	/* actions validation depends on parsing the ct matches first */
+	err = mlx5_tc_ct_match_add(get_ct_priv(priv), &parse_attr->spec, f,
+				   &flow->attr->ct_attr, extack);
+	if (err)
+		goto err_free;
+
+	err = parse_tc_fdb_actions(priv, &rule->action, flow, extack);
+	if (err)
+		goto err_free;
+
+	if (flow->attr->lag.count) {
+		err = mlx5_lag_add_mpesw_rule(esw->dev);
+		if (err)
+			goto err_free;
+	}
+
+	err = mlx5e_tc_add_fdb_flow(priv, flow, extack);
+	complete_all(&flow->init_done);
+	if (err) {
+		if (!(err == -ENETUNREACH && mlx5_lag_is_multipath(in_mdev)))
+			goto err_lag;
+
+		add_unready_flow(flow);
+	}
+
+	return flow;
+
+err_lag:
+	if (flow->attr->lag.count)
+		mlx5_lag_del_mpesw_rule(esw->dev);
+err_free:
+	mlx5e_flow_put(priv, flow);
+out:
+	return ERR_PTR(err);
+}
+
+static int mlx5e_tc_add_fdb_peer_flow(struct flow_cls_offload *f,
+				      struct mlx5e_tc_flow *flow,
+				      unsigned long flow_flags)
+{
+	struct mlx5e_priv *priv = flow->priv, *peer_priv;
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch, *peer_esw;
+	struct mlx5_esw_flow_attr *attr = flow->attr->esw_attr;
+	struct mlx5_devcom *devcom = priv->mdev->priv.devcom;
+	struct mlx5e_tc_flow_parse_attr *parse_attr;
+	struct mlx5e_rep_priv *peer_urpriv;
+	struct mlx5e_tc_flow *peer_flow;
+	struct mlx5_core_dev *in_mdev;
+	int err = 0;
+
+	peer_esw = mlx5_devcom_get_peer_data(devcom, MLX5_DEVCOM_ESW_OFFLOADS);
+	if (!peer_esw)
+		return -ENODEV;
+
+	peer_urpriv = mlx5_eswitch_get_uplink_priv(peer_esw, REP_ETH);
+	peer_priv = netdev_priv(peer_urpriv->netdev);
+
+	/* in_mdev is assigned of which the packet originated from.
+	 * So packets redirected to uplink use the same mdev of the
+	 * original flow and packets redirected from uplink use the
+	 * peer mdev.
+	 */
+	if (attr->in_rep->vport == MLX5_VPORT_UPLINK)
+		in_mdev = peer_priv->mdev;
+	else
+		in_mdev = priv->mdev;
+
+	parse_attr = flow->attr->parse_attr;
+	peer_flow = __mlx5e_add_fdb_flow(peer_priv, f, flow_flags,
+					 parse_attr->filter_dev,
+					 attr->in_rep, in_mdev);
+	if (IS_ERR(peer_flow)) {
+		err = PTR_ERR(peer_flow);
+		goto out;
+	}
+
+	flow->peer_flow = peer_flow;
+	flow_flag_set(flow, DUP);
+	mutex_lock(&esw->offloads.peer_mutex);
+	list_add_tail(&flow->peer, &esw->offloads.peer_flows);
+	mutex_unlock(&esw->offloads.peer_mutex);
+
+out:
+	mlx5_devcom_release_peer_data(devcom, MLX5_DEVCOM_ESW_OFFLOADS);
+	return err;
+}
+
+static int
+mlx5e_add_fdb_flow(struct mlx5e_priv *priv,
+		   struct flow_cls_offload *f,
+		   unsigned long flow_flags,
+		   struct net_device *filter_dev,
+		   struct mlx5e_tc_flow **__flow)
+{
+	struct mlx5e_rep_priv *rpriv = priv->ppriv;
+	struct mlx5_eswitch_rep *in_rep = rpriv->rep;
+	struct mlx5_core_dev *in_mdev = priv->mdev;
+	struct mlx5e_tc_flow *flow;
+	int err;
+
+	flow = __mlx5e_add_fdb_flow(priv, f, flow_flags, filter_dev, in_rep,
+				    in_mdev);
+	if (IS_ERR(flow))
+		return PTR_ERR(flow);
+
+	if (is_peer_flow_needed(flow)) {
+		err = mlx5e_tc_add_fdb_peer_flow(f, flow, flow_flags);
+		if (err) {
+			mlx5e_tc_del_fdb_flow(priv, flow);
+			goto out;
+		}
+	}
+
+	*__flow = flow;
+
+	return 0;
+
+out:
+	return err;
+}
+
+static int
+mlx5e_add_nic_flow(struct mlx5e_priv *priv,
+		   struct flow_cls_offload *f,
+		   unsigned long flow_flags,
+		   struct net_device *filter_dev,
+		   struct mlx5e_tc_flow **__flow)
+{
+	struct flow_rule *rule = flow_cls_offload_flow_rule(f);
+	struct netlink_ext_ack *extack = f->common.extack;
+	struct mlx5e_tc_flow_parse_attr *parse_attr;
+	struct mlx5e_tc_flow *flow;
+	int attr_size, err;
+
+	if (!MLX5_CAP_FLOWTABLE_NIC_RX(priv->mdev, ignore_flow_level)) {
+		if (!tc_cls_can_offload_and_chain0(priv->netdev, &f->common))
+			return -EOPNOTSUPP;
+	} else if (!tc_can_offload_extack(priv->netdev, f->common.extack)) {
+		return -EOPNOTSUPP;
+	}
+
+	flow_flags |= BIT(MLX5E_TC_FLOW_FLAG_NIC);
+	attr_size  = sizeof(struct mlx5_nic_flow_attr);
+	err = mlx5e_alloc_flow(priv, attr_size, f, flow_flags,
+			       &parse_attr, &flow);
+	if (err)
+		goto out;
+
+	parse_attr->filter_dev = filter_dev;
+	mlx5e_flow_attr_init(flow->attr, parse_attr, f);
+
+	err = parse_cls_flower(flow->priv, flow, &parse_attr->spec,
+			       f, filter_dev);
+	if (err)
+		goto err_free;
+
+	err = mlx5_tc_ct_match_add(get_ct_priv(priv), &parse_attr->spec, f,
+				   &flow->attr->ct_attr, extack);
+	if (err)
+		goto err_free;
+
+	err = parse_tc_nic_actions(priv, &rule->action, flow, extack);
+	if (err)
+		goto err_free;
+
+	err = mlx5e_tc_add_nic_flow(priv, flow, extack);
+	if (err)
+		goto err_free;
+
+	flow_flag_set(flow, OFFLOADED);
+	*__flow = flow;
+
+	return 0;
+
+err_free:
+	flow_flag_set(flow, FAILED);
+	mlx5e_mod_hdr_dealloc(&parse_attr->mod_hdr_acts);
+	mlx5e_flow_put(priv, flow);
+out:
+	return err;
+}
+
+static int
+mlx5e_tc_add_flow(struct mlx5e_priv *priv,
+		  struct flow_cls_offload *f,
+		  unsigned long flags,
+		  struct net_device *filter_dev,
+		  struct mlx5e_tc_flow **flow)
+{
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	unsigned long flow_flags;
+	int err;
+
+	get_flags(flags, &flow_flags);
+
+	if (!tc_can_offload_extack(priv->netdev, f->common.extack))
+		return -EOPNOTSUPP;
+
+	if (esw && esw->mode == MLX5_ESWITCH_OFFLOADS)
+		err = mlx5e_add_fdb_flow(priv, f, flow_flags,
+					 filter_dev, flow);
+	else
+		err = mlx5e_add_nic_flow(priv, f, flow_flags,
+					 filter_dev, flow);
+
+	return err;
+}
+
+static bool is_flow_rule_duplicate_allowed(struct net_device *dev,
+					   struct mlx5e_rep_priv *rpriv)
+{
+	/* Offloaded flow rule is allowed to duplicate on non-uplink representor
+	 * sharing tc block with other slaves of a lag device. Rpriv can be NULL if this
+	 * function is called from NIC mode.
+	 */
+	return netif_is_lag_port(dev) && rpriv && rpriv->rep->vport != MLX5_VPORT_UPLINK;
+}
+
+int mlx5e_configure_flower(struct net_device *dev, struct mlx5e_priv *priv,
+			   struct flow_cls_offload *f, unsigned long flags)
+{
+	struct netlink_ext_ack *extack = f->common.extack;
+	struct rhashtable *tc_ht = get_tc_ht(priv, flags);
+	struct mlx5e_rep_priv *rpriv = priv->ppriv;
+	struct mlx5e_tc_flow *flow;
+	int err = 0;
+
+	if (!mlx5_esw_hold(priv->mdev))
+		return -EBUSY;
+
+	mlx5_esw_get(priv->mdev);
+
+	rcu_read_lock();
+	flow = rhashtable_lookup(tc_ht, &f->cookie, tc_ht_params);
+	if (flow) {
+		/* Same flow rule offloaded to non-uplink representor sharing tc block,
+		 * just return 0.
+		 */
+		if (is_flow_rule_duplicate_allowed(dev, rpriv) && flow->orig_dev != dev)
+			goto rcu_unlock;
+
+		NL_SET_ERR_MSG_MOD(extack,
+				   "flow cookie already exists, ignoring");
+		netdev_warn_once(priv->netdev,
+				 "flow cookie %lx already exists, ignoring\n",
+				 f->cookie);
+		err = -EEXIST;
+		goto rcu_unlock;
+	}
+rcu_unlock:
+	rcu_read_unlock();
+	if (flow)
+		goto out;
+
+	trace_mlx5e_configure_flower(f);
+	err = mlx5e_tc_add_flow(priv, f, flags, dev, &flow);
+	if (err)
+		goto out;
+
+	/* Flow rule offloaded to non-uplink representor sharing tc block,
+	 * set the flow's owner dev.
+	 */
+	if (is_flow_rule_duplicate_allowed(dev, rpriv))
+		flow->orig_dev = dev;
+
+	err = rhashtable_lookup_insert_fast(tc_ht, &flow->node, tc_ht_params);
+	if (err)
+		goto err_free;
+
+	mlx5_esw_release(priv->mdev);
+	return 0;
+
+err_free:
+	mlx5e_flow_put(priv, flow);
+out:
+	mlx5_esw_put(priv->mdev);
+	mlx5_esw_release(priv->mdev);
+	return err;
+}
+
+static bool same_flow_direction(struct mlx5e_tc_flow *flow, int flags)
+{
+	bool dir_ingress = !!(flags & MLX5_TC_FLAG(INGRESS));
+	bool dir_egress = !!(flags & MLX5_TC_FLAG(EGRESS));
+
+	return flow_flag_test(flow, INGRESS) == dir_ingress &&
+		flow_flag_test(flow, EGRESS) == dir_egress;
+}
+
+int mlx5e_delete_flower(struct net_device *dev, struct mlx5e_priv *priv,
+			struct flow_cls_offload *f, unsigned long flags)
+{
+	struct rhashtable *tc_ht = get_tc_ht(priv, flags);
+	struct mlx5e_tc_flow *flow;
+	int err;
+
+	rcu_read_lock();
+	flow = rhashtable_lookup(tc_ht, &f->cookie, tc_ht_params);
+	if (!flow || !same_flow_direction(flow, flags)) {
+		err = -EINVAL;
+		goto errout;
+	}
+
+	/* Only delete the flow if it doesn't have MLX5E_TC_FLOW_DELETED flag
+	 * set.
+	 */
+	if (flow_flag_test_and_set(flow, DELETED)) {
+		err = -EINVAL;
+		goto errout;
+	}
+	rhashtable_remove_fast(tc_ht, &flow->node, tc_ht_params);
+	rcu_read_unlock();
+
+	trace_mlx5e_delete_flower(f);
+	mlx5e_flow_put(priv, flow);
+
+	mlx5_esw_put(priv->mdev);
+	return 0;
+
+errout:
+	rcu_read_unlock();
+	return err;
+}
+
+int mlx5e_stats_flower(struct net_device *dev, struct mlx5e_priv *priv,
+		       struct flow_cls_offload *f, unsigned long flags)
+{
+	struct mlx5_devcom *devcom = priv->mdev->priv.devcom;
+	struct rhashtable *tc_ht = get_tc_ht(priv, flags);
+	struct mlx5_eswitch *peer_esw;
+	struct mlx5e_tc_flow *flow;
+	struct mlx5_fc *counter;
+	u64 lastuse = 0;
+	u64 packets = 0;
+	u64 bytes = 0;
+	int err = 0;
+
+	rcu_read_lock();
+	flow = mlx5e_flow_get(rhashtable_lookup(tc_ht, &f->cookie,
+						tc_ht_params));
+	rcu_read_unlock();
+	if (IS_ERR(flow))
+		return PTR_ERR(flow);
+
+	if (!same_flow_direction(flow, flags)) {
+		err = -EINVAL;
+		goto errout;
+	}
+
+	if (mlx5e_is_offloaded_flow(flow) || flow_flag_test(flow, CT)) {
+		counter = mlx5e_tc_get_counter(flow);
+		if (!counter)
+			goto errout;
+
+		mlx5_fc_query_cached(counter, &bytes, &packets, &lastuse);
+	}
+
+	/* Under multipath it's possible for one rule to be currently
+	 * un-offloaded while the other rule is offloaded.
+	 */
+	peer_esw = mlx5_devcom_get_peer_data(devcom, MLX5_DEVCOM_ESW_OFFLOADS);
+	if (!peer_esw)
+		goto out;
+
+	if (flow_flag_test(flow, DUP) &&
+	    flow_flag_test(flow->peer_flow, OFFLOADED)) {
+		u64 bytes2;
+		u64 packets2;
+		u64 lastuse2;
+
+		counter = mlx5e_tc_get_counter(flow->peer_flow);
+		if (!counter)
+			goto no_peer_counter;
+		mlx5_fc_query_cached(counter, &bytes2, &packets2, &lastuse2);
+
+		bytes += bytes2;
+		packets += packets2;
+		lastuse = max_t(u64, lastuse, lastuse2);
+	}
+
+no_peer_counter:
+	mlx5_devcom_release_peer_data(devcom, MLX5_DEVCOM_ESW_OFFLOADS);
+out:
+	flow_stats_update(&f->stats, bytes, packets, 0, lastuse,
+			  FLOW_ACTION_HW_STATS_DELAYED);
+	trace_mlx5e_stats_flower(f);
+errout:
+	mlx5e_flow_put(priv, flow);
+	return err;
+}
+
+static int apply_police_params(struct mlx5e_priv *priv, u64 rate,
+			       struct netlink_ext_ack *extack)
+{
+	struct mlx5e_rep_priv *rpriv = priv->ppriv;
+	struct mlx5_eswitch *esw;
+	u32 rate_mbps = 0;
+	u16 vport_num;
+	int err;
+
+	vport_num = rpriv->rep->vport;
+	if (vport_num >= MLX5_VPORT_ECPF) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "Ingress rate limit is supported only for Eswitch ports connected to VFs");
+		return -EOPNOTSUPP;
+	}
+
+	esw = priv->mdev->priv.eswitch;
+	/* rate is given in bytes/sec.
+	 * First convert to bits/sec and then round to the nearest mbit/secs.
+	 * mbit means million bits.
+	 * Moreover, if rate is non zero we choose to configure to a minimum of
+	 * 1 mbit/sec.
+	 */
+	if (rate) {
+		rate = (rate * BITS_PER_BYTE) + 500000;
+		do_div(rate, 1000000);
+		rate_mbps = max_t(u32, rate, 1);
+	}
+
+	err = mlx5_esw_qos_modify_vport_rate(esw, vport_num, rate_mbps);
+	if (err)
+		NL_SET_ERR_MSG_MOD(extack, "failed applying action to hardware");
+
+	return err;
+}
+
+int mlx5e_policer_validate(const struct flow_action *action,
+			   const struct flow_action_entry *act,
+			   struct netlink_ext_ack *extack)
+{
+	if (act->police.exceed.act_id != FLOW_ACTION_DROP) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "Offload not supported when exceed action is not drop");
+		return -EOPNOTSUPP;
+	}
+
+	if (act->police.notexceed.act_id == FLOW_ACTION_ACCEPT &&
+	    !flow_action_is_last_entry(action, act)) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "Offload not supported when conform action is ok, but action is not last");
+		return -EOPNOTSUPP;
+	}
+
+	if (act->police.peakrate_bytes_ps ||
+	    act->police.avrate || act->police.overhead) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "Offload not supported when peakrate/avrate/overhead is configured");
+		return -EOPNOTSUPP;
+	}
+
+	return 0;
+}
+
+static int scan_tc_matchall_fdb_actions(struct mlx5e_priv *priv,
+					struct flow_action *flow_action,
+					struct netlink_ext_ack *extack)
+{
+	struct mlx5e_rep_priv *rpriv = priv->ppriv;
+	const struct flow_action_entry *act;
+	int err;
+	int i;
+
+	if (!flow_action_has_entries(flow_action)) {
+		NL_SET_ERR_MSG_MOD(extack, "matchall called with no action");
+		return -EINVAL;
+	}
+
+	if (!flow_offload_has_one_action(flow_action)) {
+		NL_SET_ERR_MSG_MOD(extack, "matchall policing support only a single action");
+		return -EOPNOTSUPP;
+	}
+
+	if (!flow_action_basic_hw_stats_check(flow_action, extack)) {
+		NL_SET_ERR_MSG_MOD(extack, "Flow action HW stats type is not supported");
+		return -EOPNOTSUPP;
+	}
+
+	flow_action_for_each(i, act, flow_action) {
+		switch (act->id) {
+		case FLOW_ACTION_POLICE:
+			if (act->police.notexceed.act_id != FLOW_ACTION_CONTINUE) {
+				NL_SET_ERR_MSG_MOD(extack,
+						   "Offload not supported when conform action is not continue");
+				return -EOPNOTSUPP;
+			}
+
+			err = mlx5e_policer_validate(flow_action, act, extack);
+			if (err)
+				return err;
+
+			err = apply_police_params(priv, act->police.rate_bytes_ps, extack);
+			if (err)
+				return err;
+
+			rpriv->prev_vf_vport_stats = priv->stats.vf_vport;
+			break;
+		default:
+			NL_SET_ERR_MSG_MOD(extack, "mlx5 supports only police action for matchall");
+			return -EOPNOTSUPP;
+		}
+	}
+
+	return 0;
+}
+
+int mlx5e_tc_configure_matchall(struct mlx5e_priv *priv,
+				struct tc_cls_matchall_offload *ma)
+{
+	struct netlink_ext_ack *extack = ma->common.extack;
+
+	if (ma->common.prio != 1) {
+		NL_SET_ERR_MSG_MOD(extack, "only priority 1 is supported");
+		return -EINVAL;
+	}
+
+	return scan_tc_matchall_fdb_actions(priv, &ma->rule->action, extack);
+}
+
+int mlx5e_tc_delete_matchall(struct mlx5e_priv *priv,
+			     struct tc_cls_matchall_offload *ma)
+{
+	struct netlink_ext_ack *extack = ma->common.extack;
+
+	return apply_police_params(priv, 0, extack);
+}
+
+void mlx5e_tc_stats_matchall(struct mlx5e_priv *priv,
+			     struct tc_cls_matchall_offload *ma)
+{
+	struct mlx5e_rep_priv *rpriv = priv->ppriv;
+	struct rtnl_link_stats64 cur_stats;
+	u64 dbytes;
+	u64 dpkts;
+
+	cur_stats = priv->stats.vf_vport;
+	dpkts = cur_stats.rx_packets - rpriv->prev_vf_vport_stats.rx_packets;
+	dbytes = cur_stats.rx_bytes - rpriv->prev_vf_vport_stats.rx_bytes;
+	rpriv->prev_vf_vport_stats = cur_stats;
+	flow_stats_update(&ma->stats, dbytes, dpkts, 0, jiffies,
+			  FLOW_ACTION_HW_STATS_DELAYED);
+}
+
+static void mlx5e_tc_hairpin_update_dead_peer(struct mlx5e_priv *priv,
+					      struct mlx5e_priv *peer_priv)
+{
+	struct mlx5e_tc_table *tc = mlx5e_fs_get_tc(priv->fs);
+	struct mlx5_core_dev *peer_mdev = peer_priv->mdev;
+	struct mlx5e_hairpin_entry *hpe, *tmp;
+	LIST_HEAD(init_wait_list);
+	u16 peer_vhca_id;
+	int bkt;
+
+	if (!mlx5e_same_hw_devs(priv, peer_priv))
+		return;
+
+	peer_vhca_id = MLX5_CAP_GEN(peer_mdev, vhca_id);
+
+	mutex_lock(&tc->hairpin_tbl_lock);
+	hash_for_each(tc->hairpin_tbl, bkt, hpe, hairpin_hlist)
+		if (refcount_inc_not_zero(&hpe->refcnt))
+			list_add(&hpe->dead_peer_wait_list, &init_wait_list);
+	mutex_unlock(&tc->hairpin_tbl_lock);
+
+	list_for_each_entry_safe(hpe, tmp, &init_wait_list, dead_peer_wait_list) {
+		wait_for_completion(&hpe->res_ready);
+		if (!IS_ERR_OR_NULL(hpe->hp) && hpe->peer_vhca_id == peer_vhca_id)
+			mlx5_core_hairpin_clear_dead_peer(hpe->hp->pair);
+
+		mlx5e_hairpin_put(priv, hpe);
+	}
+}
+
+static int mlx5e_tc_netdev_event(struct notifier_block *this,
+				 unsigned long event, void *ptr)
+{
+	struct net_device *ndev = netdev_notifier_info_to_dev(ptr);
+	struct mlx5e_priv *peer_priv;
+	struct mlx5e_tc_table *tc;
+	struct mlx5e_priv *priv;
+
+	if (ndev->netdev_ops != &mlx5e_netdev_ops ||
+	    event != NETDEV_UNREGISTER ||
+	    ndev->reg_state == NETREG_REGISTERED)
+		return NOTIFY_DONE;
+
+	tc = container_of(this, struct mlx5e_tc_table, netdevice_nb);
+	priv = tc->priv;
+	peer_priv = netdev_priv(ndev);
+	if (priv == peer_priv ||
+	    !(priv->netdev->features & NETIF_F_HW_TC))
+		return NOTIFY_DONE;
+
+	mlx5e_tc_hairpin_update_dead_peer(priv, peer_priv);
+
+	return NOTIFY_DONE;
+}
+
+static int mlx5e_tc_nic_get_ft_size(struct mlx5_core_dev *dev)
+{
+	int tc_grp_size, tc_tbl_size;
+	u32 max_flow_counter;
+
+	max_flow_counter = (MLX5_CAP_GEN(dev, max_flow_counter_31_16) << 16) |
+			    MLX5_CAP_GEN(dev, max_flow_counter_15_0);
+
+	tc_grp_size = min_t(int, max_flow_counter, MLX5E_TC_TABLE_MAX_GROUP_SIZE);
+
+	tc_tbl_size = min_t(int, tc_grp_size * MLX5E_TC_TABLE_NUM_GROUPS,
+			    BIT(MLX5_CAP_FLOWTABLE_NIC_RX(dev, log_max_ft_size)));
+
+	return tc_tbl_size;
+}
+
+static int mlx5e_tc_nic_create_miss_table(struct mlx5e_priv *priv)
+{
+	struct mlx5e_tc_table *tc = mlx5e_fs_get_tc(priv->fs);
+	struct mlx5_flow_table **ft = &tc->miss_t;
+	struct mlx5_flow_table_attr ft_attr = {};
+	struct mlx5_flow_namespace *ns;
+	int err = 0;
+
+	ft_attr.max_fte = 1;
+	ft_attr.autogroup.max_num_groups = 1;
+	ft_attr.level = MLX5E_TC_MISS_LEVEL;
+	ft_attr.prio = 0;
+	ns = mlx5_get_flow_namespace(priv->mdev, MLX5_FLOW_NAMESPACE_KERNEL);
+
+	*ft = mlx5_create_auto_grouped_flow_table(ns, &ft_attr);
+	if (IS_ERR(*ft)) {
+		err = PTR_ERR(*ft);
+		netdev_err(priv->netdev, "failed to create tc nic miss table err=%d\n", err);
+	}
+
+	return err;
+}
+
+static void mlx5e_tc_nic_destroy_miss_table(struct mlx5e_priv *priv)
+{
+	struct mlx5e_tc_table *tc = mlx5e_fs_get_tc(priv->fs);
+
+	mlx5_destroy_flow_table(tc->miss_t);
+}
+
+int mlx5e_tc_nic_init(struct mlx5e_priv *priv)
+{
+	struct mlx5e_tc_table *tc = mlx5e_fs_get_tc(priv->fs);
+	struct mlx5_core_dev *dev = priv->mdev;
+	struct mapping_ctx *chains_mapping;
+	struct mlx5_chains_attr attr = {};
+	u64 mapping_id;
+	int err;
+
+	mlx5e_mod_hdr_tbl_init(&tc->mod_hdr);
+	mutex_init(&tc->t_lock);
+	mutex_init(&tc->hairpin_tbl_lock);
+	hash_init(tc->hairpin_tbl);
+	tc->priv = priv;
+
+	err = rhashtable_init(&tc->ht, &tc_ht_params);
+	if (err)
+		return err;
+
+	lockdep_set_class(&tc->ht.mutex, &tc_ht_lock_key);
+	lockdep_init_map(&tc->ht.run_work.lockdep_map, "tc_ht_wq_key", &tc_ht_wq_key, 0);
+
+	mapping_id = mlx5_query_nic_system_image_guid(dev);
+
+	chains_mapping = mapping_create_for_id(mapping_id, MAPPING_TYPE_CHAIN,
+					       sizeof(struct mlx5_mapped_obj),
+					       MLX5E_TC_TABLE_CHAIN_TAG_MASK, true);
+
+	if (IS_ERR(chains_mapping)) {
+		err = PTR_ERR(chains_mapping);
+		goto err_mapping;
+	}
+	tc->mapping = chains_mapping;
+
+	err = mlx5e_tc_nic_create_miss_table(priv);
+	if (err)
+		goto err_chains;
+
+	if (MLX5_CAP_FLOWTABLE_NIC_RX(priv->mdev, ignore_flow_level))
+		attr.flags = MLX5_CHAINS_AND_PRIOS_SUPPORTED |
+			MLX5_CHAINS_IGNORE_FLOW_LEVEL_SUPPORTED;
+	attr.ns = MLX5_FLOW_NAMESPACE_KERNEL;
+	attr.max_ft_sz = mlx5e_tc_nic_get_ft_size(dev);
+	attr.max_grp_num = MLX5E_TC_TABLE_NUM_GROUPS;
+	attr.default_ft = tc->miss_t;
+	attr.mapping = chains_mapping;
+
+	tc->chains = mlx5_chains_create(dev, &attr);
+	if (IS_ERR(tc->chains)) {
+		err = PTR_ERR(tc->chains);
+		goto err_miss;
+	}
+
+	tc->post_act = mlx5e_tc_post_act_init(priv, tc->chains, MLX5_FLOW_NAMESPACE_KERNEL);
+	tc->ct = mlx5_tc_ct_init(priv, tc->chains, &tc->mod_hdr,
+				 MLX5_FLOW_NAMESPACE_KERNEL, tc->post_act);
+
+	tc->netdevice_nb.notifier_call = mlx5e_tc_netdev_event;
+	err = register_netdevice_notifier_dev_net(priv->netdev,
+						  &tc->netdevice_nb,
+						  &tc->netdevice_nn);
+	if (err) {
+		tc->netdevice_nb.notifier_call = NULL;
+		mlx5_core_warn(priv->mdev, "Failed to register netdev notifier\n");
+		goto err_reg;
+	}
+
+	return 0;
+
+err_reg:
+	mlx5_tc_ct_clean(tc->ct);
+	mlx5e_tc_post_act_destroy(tc->post_act);
+	mlx5_chains_destroy(tc->chains);
+err_miss:
+	mlx5e_tc_nic_destroy_miss_table(priv);
+err_chains:
+	mapping_destroy(chains_mapping);
+err_mapping:
+	rhashtable_destroy(&tc->ht);
+	return err;
+}
+
+static void _mlx5e_tc_del_flow(void *ptr, void *arg)
+{
+	struct mlx5e_tc_flow *flow = ptr;
+	struct mlx5e_priv *priv = flow->priv;
+
+	mlx5e_tc_del_flow(priv, flow);
+	kfree(flow);
+}
+
+void mlx5e_tc_nic_cleanup(struct mlx5e_priv *priv)
+{
+	struct mlx5e_tc_table *tc = mlx5e_fs_get_tc(priv->fs);
+
+	if (tc->netdevice_nb.notifier_call)
+		unregister_netdevice_notifier_dev_net(priv->netdev,
+						      &tc->netdevice_nb,
+						      &tc->netdevice_nn);
+
+	mlx5e_mod_hdr_tbl_destroy(&tc->mod_hdr);
+	mutex_destroy(&tc->hairpin_tbl_lock);
+
+	rhashtable_free_and_destroy(&tc->ht, _mlx5e_tc_del_flow, NULL);
+
+	if (!IS_ERR_OR_NULL(tc->t)) {
+		mlx5_chains_put_table(tc->chains, 0, 1, MLX5E_TC_FT_LEVEL);
+		tc->t = NULL;
+	}
+	mutex_destroy(&tc->t_lock);
+
+	mlx5_tc_ct_clean(tc->ct);
+	mlx5e_tc_post_act_destroy(tc->post_act);
+	mapping_destroy(tc->mapping);
+	mlx5_chains_destroy(tc->chains);
+	mlx5e_tc_nic_destroy_miss_table(priv);
+}
+
+int mlx5e_tc_ht_init(struct rhashtable *tc_ht)
+{
+	int err;
+
+	err = rhashtable_init(tc_ht, &tc_ht_params);
+	if (err)
+		return err;
+
+	lockdep_set_class(&tc_ht->mutex, &tc_ht_lock_key);
+	lockdep_init_map(&tc_ht->run_work.lockdep_map, "tc_ht_wq_key", &tc_ht_wq_key, 0);
+
+	return 0;
+}
+
+void mlx5e_tc_ht_cleanup(struct rhashtable *tc_ht)
+{
+	rhashtable_free_and_destroy(tc_ht, _mlx5e_tc_del_flow, NULL);
+}
+
+int mlx5e_tc_esw_init(struct mlx5_rep_uplink_priv *uplink_priv)
+{
+	const size_t sz_enc_opts = sizeof(struct tunnel_match_enc_opts);
+	struct mlx5e_rep_priv *rpriv;
+	struct mapping_ctx *mapping;
+	struct mlx5_eswitch *esw;
+	struct mlx5e_priv *priv;
+	u64 mapping_id;
+	int err = 0;
+
+	rpriv = container_of(uplink_priv, struct mlx5e_rep_priv, uplink_priv);
+	priv = netdev_priv(rpriv->netdev);
+	esw = priv->mdev->priv.eswitch;
+
+	uplink_priv->post_act = mlx5e_tc_post_act_init(priv, esw_chains(esw),
+						       MLX5_FLOW_NAMESPACE_FDB);
+	uplink_priv->ct_priv = mlx5_tc_ct_init(netdev_priv(priv->netdev),
+					       esw_chains(esw),
+					       &esw->offloads.mod_hdr,
+					       MLX5_FLOW_NAMESPACE_FDB,
+					       uplink_priv->post_act);
+
+	uplink_priv->int_port_priv = mlx5e_tc_int_port_init(netdev_priv(priv->netdev));
+
+	uplink_priv->tc_psample = mlx5e_tc_sample_init(esw, uplink_priv->post_act);
+
+	mapping_id = mlx5_query_nic_system_image_guid(esw->dev);
+
+	mapping = mapping_create_for_id(mapping_id, MAPPING_TYPE_TUNNEL,
+					sizeof(struct tunnel_match_key),
+					TUNNEL_INFO_BITS_MASK, true);
+
+	if (IS_ERR(mapping)) {
+		err = PTR_ERR(mapping);
+		goto err_tun_mapping;
+	}
+	uplink_priv->tunnel_mapping = mapping;
+
+	/* Two last values are reserved for stack devices slow path table mark
+	 * and bridge ingress push mark.
+	 */
+	mapping = mapping_create_for_id(mapping_id, MAPPING_TYPE_TUNNEL_ENC_OPTS,
+					sz_enc_opts, ENC_OPTS_BITS_MASK - 2, true);
+	if (IS_ERR(mapping)) {
+		err = PTR_ERR(mapping);
+		goto err_enc_opts_mapping;
+	}
+	uplink_priv->tunnel_enc_opts_mapping = mapping;
+
+	uplink_priv->encap = mlx5e_tc_tun_init(priv);
+	if (IS_ERR(uplink_priv->encap)) {
+		err = PTR_ERR(uplink_priv->encap);
+		goto err_register_fib_notifier;
+	}
+
+	mlx5_esw_offloads_devcom_init(esw);
+
+	return 0;
+
+err_register_fib_notifier:
+	mapping_destroy(uplink_priv->tunnel_enc_opts_mapping);
+err_enc_opts_mapping:
+	mapping_destroy(uplink_priv->tunnel_mapping);
+err_tun_mapping:
+	mlx5e_tc_sample_cleanup(uplink_priv->tc_psample);
+	mlx5e_tc_int_port_cleanup(uplink_priv->int_port_priv);
+	mlx5_tc_ct_clean(uplink_priv->ct_priv);
+	netdev_warn(priv->netdev,
+		    "Failed to initialize tc (eswitch), err: %d", err);
+	mlx5e_tc_post_act_destroy(uplink_priv->post_act);
+	return err;
+}
+
+void mlx5e_tc_esw_cleanup(struct mlx5_rep_uplink_priv *uplink_priv)
+{
+	struct mlx5e_rep_priv *rpriv;
+	struct mlx5_eswitch *esw;
+	struct mlx5e_priv *priv;
+
+	rpriv = container_of(uplink_priv, struct mlx5e_rep_priv, uplink_priv);
+	priv = netdev_priv(rpriv->netdev);
+	esw = priv->mdev->priv.eswitch;
+
+	mlx5_esw_offloads_devcom_cleanup(esw);
+
+	mlx5e_tc_tun_cleanup(uplink_priv->encap);
+
+	mapping_destroy(uplink_priv->tunnel_enc_opts_mapping);
+	mapping_destroy(uplink_priv->tunnel_mapping);
+
+	mlx5e_tc_sample_cleanup(uplink_priv->tc_psample);
+	mlx5e_tc_int_port_cleanup(uplink_priv->int_port_priv);
+	mlx5_tc_ct_clean(uplink_priv->ct_priv);
+	mlx5e_flow_meters_cleanup(uplink_priv->flow_meters);
+	mlx5e_tc_post_act_destroy(uplink_priv->post_act);
+}
+
+int mlx5e_tc_num_filters(struct mlx5e_priv *priv, unsigned long flags)
+{
+	struct rhashtable *tc_ht = get_tc_ht(priv, flags);
+
+	return atomic_read(&tc_ht->nelems);
+}
+
+void mlx5e_tc_clean_fdb_peer_flows(struct mlx5_eswitch *esw)
+{
+	struct mlx5e_tc_flow *flow, *tmp;
+
+	list_for_each_entry_safe(flow, tmp, &esw->offloads.peer_flows, peer)
+		__mlx5e_tc_del_fdb_peer_flow(flow);
+}
+
+void mlx5e_tc_reoffload_flows_work(struct work_struct *work)
+{
+	struct mlx5_rep_uplink_priv *rpriv =
+		container_of(work, struct mlx5_rep_uplink_priv,
+			     reoffload_flows_work);
+	struct mlx5e_tc_flow *flow, *tmp;
+
+	mutex_lock(&rpriv->unready_flows_lock);
+	list_for_each_entry_safe(flow, tmp, &rpriv->unready_flows, unready) {
+		if (!mlx5e_tc_add_fdb_flow(flow->priv, flow, NULL))
+			unready_flow_del(flow);
+	}
+	mutex_unlock(&rpriv->unready_flows_lock);
+}
+
+static int mlx5e_setup_tc_cls_flower(struct mlx5e_priv *priv,
+				     struct flow_cls_offload *cls_flower,
+				     unsigned long flags)
+{
+	switch (cls_flower->command) {
+	case FLOW_CLS_REPLACE:
+		return mlx5e_configure_flower(priv->netdev, priv, cls_flower,
+					      flags);
+	case FLOW_CLS_DESTROY:
+		return mlx5e_delete_flower(priv->netdev, priv, cls_flower,
+					   flags);
+	case FLOW_CLS_STATS:
+		return mlx5e_stats_flower(priv->netdev, priv, cls_flower,
+					  flags);
+	default:
+		return -EOPNOTSUPP;
+	}
+}
+
+int mlx5e_setup_tc_block_cb(enum tc_setup_type type, void *type_data,
+			    void *cb_priv)
+{
+	unsigned long flags = MLX5_TC_FLAG(INGRESS);
+	struct mlx5e_priv *priv = cb_priv;
+
+	if (!priv->netdev || !netif_device_present(priv->netdev))
+		return -EOPNOTSUPP;
+
+	if (mlx5e_is_uplink_rep(priv))
+		flags |= MLX5_TC_FLAG(ESW_OFFLOAD);
+	else
+		flags |= MLX5_TC_FLAG(NIC_OFFLOAD);
+
+	switch (type) {
+	case TC_SETUP_CLSFLOWER:
+		return mlx5e_setup_tc_cls_flower(priv, type_data, flags);
+	default:
+		return -EOPNOTSUPP;
+	}
+}
+
+bool mlx5e_tc_update_skb(struct mlx5_cqe64 *cqe,
+			 struct sk_buff *skb)
+{
+#if IS_ENABLED(CONFIG_NET_TC_SKB_EXT)
+	u32 chain = 0, chain_tag, reg_b, zone_restore_id;
+	struct mlx5e_priv *priv = netdev_priv(skb->dev);
+	struct mlx5_mapped_obj mapped_obj;
+	struct tc_skb_ext *tc_skb_ext;
+	struct mlx5e_tc_table *tc;
+	int err;
+
+	reg_b = be32_to_cpu(cqe->ft_metadata);
+	tc = mlx5e_fs_get_tc(priv->fs);
+	chain_tag = reg_b & MLX5E_TC_TABLE_CHAIN_TAG_MASK;
+
+	err = mapping_find(tc->mapping, chain_tag, &mapped_obj);
+	if (err) {
+		netdev_dbg(priv->netdev,
+			   "Couldn't find chain for chain tag: %d, err: %d\n",
+			   chain_tag, err);
+		return false;
+	}
+
+	if (mapped_obj.type == MLX5_MAPPED_OBJ_CHAIN) {
+		chain = mapped_obj.chain;
+		tc_skb_ext = tc_skb_ext_alloc(skb);
+		if (WARN_ON(!tc_skb_ext))
+			return false;
+
+		tc_skb_ext->chain = chain;
+
+		zone_restore_id = (reg_b >> MLX5_REG_MAPPING_MOFFSET(NIC_ZONE_RESTORE_TO_REG)) &
+			ESW_ZONE_ID_MASK;
+
+		if (!mlx5e_tc_ct_restore_flow(tc->ct, skb,
+					      zone_restore_id))
+			return false;
+	} else {
+		netdev_dbg(priv->netdev, "Invalid mapped object type: %d\n", mapped_obj.type);
+		return false;
+	}
+#endif /* CONFIG_NET_TC_SKB_EXT */
+
+	return true;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.h b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.h
new file mode 100644
index 000000000..edd5f0944
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.h
@@ -0,0 +1,389 @@
+/*
+ * Copyright (c) 2016, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef __MLX5_EN_TC_H__
+#define __MLX5_EN_TC_H__
+
+#include <net/pkt_cls.h>
+#include "en.h"
+#include "eswitch.h"
+#include "en/tc_ct.h"
+#include "en/tc_tun.h"
+#include "en/tc/int_port.h"
+#include "en/tc/meter.h"
+#include "en_rep.h"
+
+#define MLX5E_TC_FLOW_ID_MASK 0x0000ffff
+
+#ifdef CONFIG_MLX5_ESWITCH
+
+#define NIC_FLOW_ATTR_SZ (sizeof(struct mlx5_flow_attr) +\
+			  sizeof(struct mlx5_nic_flow_attr))
+#define ESW_FLOW_ATTR_SZ (sizeof(struct mlx5_flow_attr) +\
+			  sizeof(struct mlx5_esw_flow_attr))
+#define ns_to_attr_sz(ns) (((ns) == MLX5_FLOW_NAMESPACE_FDB) ?\
+			    ESW_FLOW_ATTR_SZ :\
+			    NIC_FLOW_ATTR_SZ)
+
+struct mlx5_fs_chains *mlx5e_nic_chains(struct mlx5e_tc_table *tc);
+int mlx5e_tc_num_filters(struct mlx5e_priv *priv, unsigned long flags);
+
+struct mlx5e_tc_update_priv {
+	struct net_device *fwd_dev;
+};
+
+struct mlx5_nic_flow_attr {
+	u32 flow_tag;
+	u32 hairpin_tirn;
+	struct mlx5_flow_table *hairpin_ft;
+};
+
+struct mlx5_flow_attr {
+	u32 action;
+	struct mlx5_fc *counter;
+	struct mlx5_modify_hdr *modify_hdr;
+	struct mlx5_ct_attr ct_attr;
+	struct mlx5e_sample_attr sample_attr;
+	struct mlx5e_meter_attr meter_attr;
+	struct mlx5e_tc_flow_parse_attr *parse_attr;
+	u32 chain;
+	u16 prio;
+	u32 dest_chain;
+	struct mlx5_flow_table *ft;
+	struct mlx5_flow_table *dest_ft;
+	u8 inner_match_level;
+	u8 outer_match_level;
+	u8 tun_ip_version;
+	int tunnel_id; /* mapped tunnel id */
+	u32 flags;
+	u32 exe_aso_type;
+	struct list_head list;
+	struct mlx5e_post_act_handle *post_act_handle;
+	struct {
+		/* Indicate whether the parsed flow should be counted for lag mode decision
+		 * making
+		 */
+		bool count;
+	} lag;
+	/* keep this union last */
+	union {
+		struct mlx5_esw_flow_attr esw_attr[0];
+		struct mlx5_nic_flow_attr nic_attr[0];
+	};
+};
+
+enum {
+	MLX5_ATTR_FLAG_VLAN_HANDLED  = BIT(0),
+	MLX5_ATTR_FLAG_SLOW_PATH     = BIT(1),
+	MLX5_ATTR_FLAG_NO_IN_PORT    = BIT(2),
+	MLX5_ATTR_FLAG_SRC_REWRITE   = BIT(3),
+	MLX5_ATTR_FLAG_SAMPLE        = BIT(4),
+	MLX5_ATTR_FLAG_ACCEPT        = BIT(5),
+	MLX5_ATTR_FLAG_CT            = BIT(6),
+};
+
+/* Returns true if any of the flags that require skipping further TC/NF processing are set. */
+static inline bool
+mlx5e_tc_attr_flags_skip(u32 attr_flags)
+{
+	return attr_flags & (MLX5_ATTR_FLAG_SLOW_PATH | MLX5_ATTR_FLAG_ACCEPT);
+}
+
+struct mlx5_rx_tun_attr {
+	u16 decap_vport;
+	union {
+		__be32 v4;
+		struct in6_addr v6;
+	} src_ip; /* Valid if decap_vport is not zero */
+	union {
+		__be32 v4;
+		struct in6_addr v6;
+	} dst_ip; /* Valid if decap_vport is not zero */
+};
+
+#define MLX5E_TC_TABLE_CHAIN_TAG_BITS 16
+#define MLX5E_TC_TABLE_CHAIN_TAG_MASK GENMASK(MLX5E_TC_TABLE_CHAIN_TAG_BITS - 1, 0)
+
+#define MLX5E_TC_MAX_INT_PORT_NUM (8)
+
+#if IS_ENABLED(CONFIG_MLX5_CLS_ACT)
+
+struct tunnel_match_key {
+	struct flow_dissector_key_control enc_control;
+	struct flow_dissector_key_keyid enc_key_id;
+	struct flow_dissector_key_ports enc_tp;
+	struct flow_dissector_key_ip enc_ip;
+	union {
+		struct flow_dissector_key_ipv4_addrs enc_ipv4;
+		struct flow_dissector_key_ipv6_addrs enc_ipv6;
+	};
+
+	int filter_ifindex;
+};
+
+struct tunnel_match_enc_opts {
+	struct flow_dissector_key_enc_opts key;
+	struct flow_dissector_key_enc_opts mask;
+};
+
+/* Tunnel_id mapping is TUNNEL_INFO_BITS + ENC_OPTS_BITS.
+ * Upper TUNNEL_INFO_BITS for general tunnel info.
+ * Lower ENC_OPTS_BITS bits for enc_opts.
+ */
+#define TUNNEL_INFO_BITS 12
+#define TUNNEL_INFO_BITS_MASK GENMASK(TUNNEL_INFO_BITS - 1, 0)
+#define ENC_OPTS_BITS 11
+#define ENC_OPTS_BITS_MASK GENMASK(ENC_OPTS_BITS - 1, 0)
+#define TUNNEL_ID_BITS (TUNNEL_INFO_BITS + ENC_OPTS_BITS)
+#define TUNNEL_ID_MASK GENMASK(TUNNEL_ID_BITS - 1, 0)
+
+enum {
+	MLX5E_TC_FLAG_INGRESS_BIT,
+	MLX5E_TC_FLAG_EGRESS_BIT,
+	MLX5E_TC_FLAG_NIC_OFFLOAD_BIT,
+	MLX5E_TC_FLAG_ESW_OFFLOAD_BIT,
+	MLX5E_TC_FLAG_FT_OFFLOAD_BIT,
+	MLX5E_TC_FLAG_LAST_EXPORTED_BIT = MLX5E_TC_FLAG_FT_OFFLOAD_BIT,
+};
+
+#define MLX5_TC_FLAG(flag) BIT(MLX5E_TC_FLAG_##flag##_BIT)
+
+int mlx5e_tc_esw_init(struct mlx5_rep_uplink_priv *uplink_priv);
+void mlx5e_tc_esw_cleanup(struct mlx5_rep_uplink_priv *uplink_priv);
+
+int mlx5e_tc_ht_init(struct rhashtable *tc_ht);
+void mlx5e_tc_ht_cleanup(struct rhashtable *tc_ht);
+
+int mlx5e_configure_flower(struct net_device *dev, struct mlx5e_priv *priv,
+			   struct flow_cls_offload *f, unsigned long flags);
+int mlx5e_delete_flower(struct net_device *dev, struct mlx5e_priv *priv,
+			struct flow_cls_offload *f, unsigned long flags);
+
+int mlx5e_stats_flower(struct net_device *dev, struct mlx5e_priv *priv,
+		       struct flow_cls_offload *f, unsigned long flags);
+
+int mlx5e_tc_configure_matchall(struct mlx5e_priv *priv,
+				struct tc_cls_matchall_offload *f);
+int mlx5e_tc_delete_matchall(struct mlx5e_priv *priv,
+			     struct tc_cls_matchall_offload *f);
+void mlx5e_tc_stats_matchall(struct mlx5e_priv *priv,
+			     struct tc_cls_matchall_offload *ma);
+
+struct mlx5e_encap_entry;
+void mlx5e_tc_encap_flows_add(struct mlx5e_priv *priv,
+			      struct mlx5e_encap_entry *e,
+			      struct list_head *flow_list);
+void mlx5e_tc_encap_flows_del(struct mlx5e_priv *priv,
+			      struct mlx5e_encap_entry *e,
+			      struct list_head *flow_list);
+bool mlx5e_encap_take(struct mlx5e_encap_entry *e);
+void mlx5e_encap_put(struct mlx5e_priv *priv, struct mlx5e_encap_entry *e);
+
+void mlx5e_take_all_encap_flows(struct mlx5e_encap_entry *e, struct list_head *flow_list);
+void mlx5e_put_flow_list(struct mlx5e_priv *priv, struct list_head *flow_list);
+
+struct mlx5e_neigh_hash_entry;
+struct mlx5e_encap_entry *
+mlx5e_get_next_init_encap(struct mlx5e_neigh_hash_entry *nhe,
+			  struct mlx5e_encap_entry *e);
+void mlx5e_tc_update_neigh_used_value(struct mlx5e_neigh_hash_entry *nhe);
+
+void mlx5e_tc_reoffload_flows_work(struct work_struct *work);
+
+enum mlx5e_tc_attr_to_reg {
+	CHAIN_TO_REG,
+	VPORT_TO_REG,
+	TUNNEL_TO_REG,
+	CTSTATE_TO_REG,
+	ZONE_TO_REG,
+	ZONE_RESTORE_TO_REG,
+	MARK_TO_REG,
+	LABELS_TO_REG,
+	FTEID_TO_REG,
+	NIC_CHAIN_TO_REG,
+	NIC_ZONE_RESTORE_TO_REG,
+	PACKET_COLOR_TO_REG,
+};
+
+struct mlx5e_tc_attr_to_reg_mapping {
+	int mfield; /* rewrite field */
+	int moffset; /* bit offset of mfield */
+	int mlen; /* bits to rewrite/match */
+
+	int soffset; /* byte offset of spec for match */
+};
+
+extern struct mlx5e_tc_attr_to_reg_mapping mlx5e_tc_attr_to_reg_mappings[];
+
+#define MLX5_REG_MAPPING_MOFFSET(reg_id) (mlx5e_tc_attr_to_reg_mappings[reg_id].moffset)
+#define MLX5_REG_MAPPING_MBITS(reg_id) (mlx5e_tc_attr_to_reg_mappings[reg_id].mlen)
+#define MLX5_REG_MAPPING_MASK(reg_id) (GENMASK(mlx5e_tc_attr_to_reg_mappings[reg_id].mlen - 1, 0))
+
+bool mlx5e_is_valid_eswitch_fwd_dev(struct mlx5e_priv *priv,
+				    struct net_device *out_dev);
+
+int mlx5e_tc_match_to_reg_set(struct mlx5_core_dev *mdev,
+			      struct mlx5e_tc_mod_hdr_acts *mod_hdr_acts,
+			      enum mlx5_flow_namespace_type ns,
+			      enum mlx5e_tc_attr_to_reg type,
+			      u32 data);
+
+void mlx5e_tc_match_to_reg_mod_hdr_change(struct mlx5_core_dev *mdev,
+					  struct mlx5e_tc_mod_hdr_acts *mod_hdr_acts,
+					  enum mlx5e_tc_attr_to_reg type,
+					  int act_id, u32 data);
+
+void mlx5e_tc_match_to_reg_match(struct mlx5_flow_spec *spec,
+				 enum mlx5e_tc_attr_to_reg type,
+				 u32 data,
+				 u32 mask);
+
+void mlx5e_tc_match_to_reg_get_match(struct mlx5_flow_spec *spec,
+				     enum mlx5e_tc_attr_to_reg type,
+				     u32 *data,
+				     u32 *mask);
+
+int mlx5e_tc_match_to_reg_set_and_get_id(struct mlx5_core_dev *mdev,
+					 struct mlx5e_tc_mod_hdr_acts *mod_hdr_acts,
+					 enum mlx5_flow_namespace_type ns,
+					 enum mlx5e_tc_attr_to_reg type,
+					 u32 data);
+
+int mlx5e_tc_add_flow_mod_hdr(struct mlx5e_priv *priv,
+			      struct mlx5e_tc_flow *flow,
+			      struct mlx5_flow_attr *attr);
+
+void mlx5e_tc_set_ethertype(struct mlx5_core_dev *mdev,
+			    struct flow_match_basic *match, bool outer,
+			    void *headers_c, void *headers_v);
+
+int mlx5e_tc_nic_init(struct mlx5e_priv *priv);
+void mlx5e_tc_nic_cleanup(struct mlx5e_priv *priv);
+
+int mlx5e_setup_tc_block_cb(enum tc_setup_type type, void *type_data,
+			    void *cb_priv);
+
+struct mlx5_flow_handle *
+mlx5e_add_offloaded_nic_rule(struct mlx5e_priv *priv,
+			     struct mlx5_flow_spec *spec,
+			     struct mlx5_flow_attr *attr);
+void mlx5e_del_offloaded_nic_rule(struct mlx5e_priv *priv,
+				  struct mlx5_flow_handle *rule,
+				  struct mlx5_flow_attr *attr);
+
+struct mlx5_flow_handle *
+mlx5_tc_rule_insert(struct mlx5e_priv *priv,
+		    struct mlx5_flow_spec *spec,
+		    struct mlx5_flow_attr *attr);
+void
+mlx5_tc_rule_delete(struct mlx5e_priv *priv,
+		    struct mlx5_flow_handle *rule,
+		    struct mlx5_flow_attr *attr);
+
+bool mlx5e_tc_is_vf_tunnel(struct net_device *out_dev, struct net_device *route_dev);
+int mlx5e_tc_query_route_vport(struct net_device *out_dev, struct net_device *route_dev,
+			       u16 *vport);
+
+int mlx5e_set_fwd_to_int_port_actions(struct mlx5e_priv *priv,
+				      struct mlx5_flow_attr *attr,
+				      int ifindex,
+				      enum mlx5e_tc_int_port_type type,
+				      u32 *action,
+				      int out_index);
+#else /* CONFIG_MLX5_CLS_ACT */
+static inline int  mlx5e_tc_nic_init(struct mlx5e_priv *priv) { return 0; }
+static inline void mlx5e_tc_nic_cleanup(struct mlx5e_priv *priv) {}
+static inline int mlx5e_tc_ht_init(struct rhashtable *tc_ht) { return 0; }
+static inline void mlx5e_tc_ht_cleanup(struct rhashtable *tc_ht) {}
+static inline int
+mlx5e_setup_tc_block_cb(enum tc_setup_type type, void *type_data, void *cb_priv)
+{ return -EOPNOTSUPP; }
+
+#endif /* CONFIG_MLX5_CLS_ACT */
+
+struct mlx5_flow_attr *mlx5_alloc_flow_attr(enum mlx5_flow_namespace_type type);
+
+struct mlx5_flow_handle *
+mlx5e_add_offloaded_nic_rule(struct mlx5e_priv *priv,
+			     struct mlx5_flow_spec *spec,
+			     struct mlx5_flow_attr *attr);
+void mlx5e_del_offloaded_nic_rule(struct mlx5e_priv *priv,
+				  struct mlx5_flow_handle *rule,
+				  struct mlx5_flow_attr *attr);
+
+#else /* CONFIG_MLX5_ESWITCH */
+static inline int  mlx5e_tc_nic_init(struct mlx5e_priv *priv) { return 0; }
+static inline void mlx5e_tc_nic_cleanup(struct mlx5e_priv *priv) {}
+static inline int  mlx5e_tc_num_filters(struct mlx5e_priv *priv,
+					unsigned long flags)
+{
+	return 0;
+}
+
+static inline int
+mlx5e_setup_tc_block_cb(enum tc_setup_type type, void *type_data, void *cb_priv)
+{ return -EOPNOTSUPP; }
+#endif
+
+#if IS_ENABLED(CONFIG_MLX5_CLS_ACT)
+struct mlx5e_tc_table *mlx5e_tc_table_alloc(void);
+void mlx5e_tc_table_free(struct mlx5e_tc_table *tc);
+static inline bool mlx5e_cqe_regb_chain(struct mlx5_cqe64 *cqe)
+{
+#if IS_ENABLED(CONFIG_NET_TC_SKB_EXT)
+	u32 chain, reg_b;
+
+	reg_b = be32_to_cpu(cqe->ft_metadata);
+
+	if (reg_b >> (MLX5E_TC_TABLE_CHAIN_TAG_BITS + ESW_ZONE_ID_BITS))
+		return false;
+
+	chain = reg_b & MLX5E_TC_TABLE_CHAIN_TAG_MASK;
+	if (chain)
+		return true;
+#endif
+
+	return false;
+}
+
+bool mlx5e_tc_update_skb(struct mlx5_cqe64 *cqe, struct sk_buff *skb);
+#else /* CONFIG_MLX5_CLS_ACT */
+static inline struct mlx5e_tc_table *mlx5e_tc_table_alloc(void) { return NULL; }
+static inline void mlx5e_tc_table_free(struct mlx5e_tc_table *tc) {}
+static inline bool mlx5e_cqe_regb_chain(struct mlx5_cqe64 *cqe)
+{ return false; }
+static inline bool
+mlx5e_tc_update_skb(struct mlx5_cqe64 *cqe, struct sk_buff *skb)
+{ return true; }
+#endif
+
+#endif /* __MLX5_EN_TC_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_tx.c b/drivers/net/ethernet/mellanox/mlx5/core/en_tx.c
new file mode 100644
index 000000000..a6d7e2cfc
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_tx.c
@@ -0,0 +1,1058 @@
+/*
+ * Copyright (c) 2015-2016, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/tcp.h>
+#include <linux/if_vlan.h>
+#include <net/geneve.h>
+#include <net/dsfield.h>
+#include "en.h"
+#include "en/txrx.h"
+#include "ipoib/ipoib.h"
+#include "en_accel/en_accel.h"
+#include "en_accel/ipsec_rxtx.h"
+#include "en_accel/macsec.h"
+#include "en/ptp.h"
+#include <net/ipv6.h>
+
+static void mlx5e_dma_unmap_wqe_err(struct mlx5e_txqsq *sq, u8 num_dma)
+{
+	int i;
+
+	for (i = 0; i < num_dma; i++) {
+		struct mlx5e_sq_dma *last_pushed_dma =
+			mlx5e_dma_get(sq, --sq->dma_fifo_pc);
+
+		mlx5e_tx_dma_unmap(sq->pdev, last_pushed_dma);
+	}
+}
+
+static inline int mlx5e_skb_l2_header_offset(struct sk_buff *skb)
+{
+#define MLX5E_MIN_INLINE (ETH_HLEN + VLAN_HLEN)
+
+	return max(skb_network_offset(skb), MLX5E_MIN_INLINE);
+}
+
+static inline int mlx5e_skb_l3_header_offset(struct sk_buff *skb)
+{
+	if (skb_transport_header_was_set(skb))
+		return skb_transport_offset(skb);
+	else
+		return mlx5e_skb_l2_header_offset(skb);
+}
+
+static inline u16 mlx5e_calc_min_inline(enum mlx5_inline_modes mode,
+					struct sk_buff *skb)
+{
+	u16 hlen;
+
+	switch (mode) {
+	case MLX5_INLINE_MODE_NONE:
+		return 0;
+	case MLX5_INLINE_MODE_TCP_UDP:
+		hlen = eth_get_headlen(skb->dev, skb->data, skb_headlen(skb));
+		if (hlen == ETH_HLEN && !skb_vlan_tag_present(skb))
+			hlen += VLAN_HLEN;
+		break;
+	case MLX5_INLINE_MODE_IP:
+		hlen = mlx5e_skb_l3_header_offset(skb);
+		break;
+	case MLX5_INLINE_MODE_L2:
+	default:
+		hlen = mlx5e_skb_l2_header_offset(skb);
+	}
+	return min_t(u16, hlen, skb_headlen(skb));
+}
+
+#define MLX5_UNSAFE_MEMCPY_DISCLAIMER				\
+	"This copy has been bounds-checked earlier in "		\
+	"mlx5i_sq_calc_wqe_attr() and intentionally "		\
+	"crosses a flex array boundary. Since it is "		\
+	"performance sensitive, splitting the copy is "		\
+	"undesirable."
+
+static inline void mlx5e_insert_vlan(void *start, struct sk_buff *skb, u16 ihs)
+{
+	struct vlan_ethhdr *vhdr = (struct vlan_ethhdr *)start;
+	int cpy1_sz = 2 * ETH_ALEN;
+	int cpy2_sz = ihs - cpy1_sz;
+
+	memcpy(&vhdr->addrs, skb->data, cpy1_sz);
+	vhdr->h_vlan_proto = skb->vlan_proto;
+	vhdr->h_vlan_TCI = cpu_to_be16(skb_vlan_tag_get(skb));
+	unsafe_memcpy(&vhdr->h_vlan_encapsulated_proto,
+		      skb->data + cpy1_sz,
+		      cpy2_sz,
+		      MLX5_UNSAFE_MEMCPY_DISCLAIMER);
+}
+
+static inline void
+mlx5e_txwqe_build_eseg_csum(struct mlx5e_txqsq *sq, struct sk_buff *skb,
+			    struct mlx5e_accel_tx_state *accel,
+			    struct mlx5_wqe_eth_seg *eseg)
+{
+	if (unlikely(mlx5e_ipsec_txwqe_build_eseg_csum(sq, skb, eseg)))
+		return;
+
+	if (likely(skb->ip_summed == CHECKSUM_PARTIAL)) {
+		eseg->cs_flags = MLX5_ETH_WQE_L3_CSUM;
+		if (skb->encapsulation) {
+			eseg->cs_flags |= MLX5_ETH_WQE_L3_INNER_CSUM |
+					  MLX5_ETH_WQE_L4_INNER_CSUM;
+			sq->stats->csum_partial_inner++;
+		} else {
+			eseg->cs_flags |= MLX5_ETH_WQE_L4_CSUM;
+			sq->stats->csum_partial++;
+		}
+#ifdef CONFIG_MLX5_EN_TLS
+	} else if (unlikely(accel && accel->tls.tls_tisn)) {
+		eseg->cs_flags = MLX5_ETH_WQE_L3_CSUM | MLX5_ETH_WQE_L4_CSUM;
+		sq->stats->csum_partial++;
+#endif
+	} else
+		sq->stats->csum_none++;
+}
+
+/* Returns the number of header bytes that we plan
+ * to inline later in the transmit descriptor
+ */
+static inline u16
+mlx5e_tx_get_gso_ihs(struct mlx5e_txqsq *sq, struct sk_buff *skb, int *hopbyhop)
+{
+	struct mlx5e_sq_stats *stats = sq->stats;
+	u16 ihs;
+
+	*hopbyhop = 0;
+	if (skb->encapsulation) {
+		ihs = skb_inner_tcp_all_headers(skb);
+		stats->tso_inner_packets++;
+		stats->tso_inner_bytes += skb->len - ihs;
+	} else {
+		if (skb_shinfo(skb)->gso_type & SKB_GSO_UDP_L4) {
+			ihs = skb_transport_offset(skb) + sizeof(struct udphdr);
+		} else {
+			ihs = skb_tcp_all_headers(skb);
+			if (ipv6_has_hopopt_jumbo(skb)) {
+				*hopbyhop = sizeof(struct hop_jumbo_hdr);
+				ihs -= sizeof(struct hop_jumbo_hdr);
+			}
+		}
+		stats->tso_packets++;
+		stats->tso_bytes += skb->len - ihs - *hopbyhop;
+	}
+
+	return ihs;
+}
+
+static inline int
+mlx5e_txwqe_build_dsegs(struct mlx5e_txqsq *sq, struct sk_buff *skb,
+			unsigned char *skb_data, u16 headlen,
+			struct mlx5_wqe_data_seg *dseg)
+{
+	dma_addr_t dma_addr = 0;
+	u8 num_dma          = 0;
+	int i;
+
+	if (headlen) {
+		dma_addr = dma_map_single(sq->pdev, skb_data, headlen,
+					  DMA_TO_DEVICE);
+		if (unlikely(dma_mapping_error(sq->pdev, dma_addr)))
+			goto dma_unmap_wqe_err;
+
+		dseg->addr       = cpu_to_be64(dma_addr);
+		dseg->lkey       = sq->mkey_be;
+		dseg->byte_count = cpu_to_be32(headlen);
+
+		mlx5e_dma_push(sq, dma_addr, headlen, MLX5E_DMA_MAP_SINGLE);
+		num_dma++;
+		dseg++;
+	}
+
+	for (i = 0; i < skb_shinfo(skb)->nr_frags; i++) {
+		skb_frag_t *frag = &skb_shinfo(skb)->frags[i];
+		int fsz = skb_frag_size(frag);
+
+		dma_addr = skb_frag_dma_map(sq->pdev, frag, 0, fsz,
+					    DMA_TO_DEVICE);
+		if (unlikely(dma_mapping_error(sq->pdev, dma_addr)))
+			goto dma_unmap_wqe_err;
+
+		dseg->addr       = cpu_to_be64(dma_addr);
+		dseg->lkey       = sq->mkey_be;
+		dseg->byte_count = cpu_to_be32(fsz);
+
+		mlx5e_dma_push(sq, dma_addr, fsz, MLX5E_DMA_MAP_PAGE);
+		num_dma++;
+		dseg++;
+	}
+
+	return num_dma;
+
+dma_unmap_wqe_err:
+	mlx5e_dma_unmap_wqe_err(sq, num_dma);
+	return -ENOMEM;
+}
+
+struct mlx5e_tx_attr {
+	u32 num_bytes;
+	u16 headlen;
+	u16 ihs;
+	__be16 mss;
+	u16 insz;
+	u8 opcode;
+	u8 hopbyhop;
+};
+
+struct mlx5e_tx_wqe_attr {
+	u16 ds_cnt;
+	u16 ds_cnt_inl;
+	u16 ds_cnt_ids;
+	u8 num_wqebbs;
+};
+
+static u8
+mlx5e_tx_wqe_inline_mode(struct mlx5e_txqsq *sq, struct sk_buff *skb,
+			 struct mlx5e_accel_tx_state *accel)
+{
+	u8 mode;
+
+#ifdef CONFIG_MLX5_EN_TLS
+	if (accel && accel->tls.tls_tisn)
+		return MLX5_INLINE_MODE_TCP_UDP;
+#endif
+
+	mode = sq->min_inline_mode;
+
+	if (skb_vlan_tag_present(skb) &&
+	    test_bit(MLX5E_SQ_STATE_VLAN_NEED_L2_INLINE, &sq->state))
+		mode = max_t(u8, MLX5_INLINE_MODE_L2, mode);
+
+	return mode;
+}
+
+static void mlx5e_sq_xmit_prepare(struct mlx5e_txqsq *sq, struct sk_buff *skb,
+				  struct mlx5e_accel_tx_state *accel,
+				  struct mlx5e_tx_attr *attr)
+{
+	struct mlx5e_sq_stats *stats = sq->stats;
+
+	if (skb_is_gso(skb)) {
+		int hopbyhop;
+		u16 ihs = mlx5e_tx_get_gso_ihs(sq, skb, &hopbyhop);
+
+		*attr = (struct mlx5e_tx_attr) {
+			.opcode    = MLX5_OPCODE_LSO,
+			.mss       = cpu_to_be16(skb_shinfo(skb)->gso_size),
+			.ihs       = ihs,
+			.num_bytes = skb->len + (skb_shinfo(skb)->gso_segs - 1) * ihs,
+			.headlen   = skb_headlen(skb) - ihs - hopbyhop,
+			.hopbyhop  = hopbyhop,
+		};
+
+		stats->packets += skb_shinfo(skb)->gso_segs;
+	} else {
+		u8 mode = mlx5e_tx_wqe_inline_mode(sq, skb, accel);
+		u16 ihs = mlx5e_calc_min_inline(mode, skb);
+
+		*attr = (struct mlx5e_tx_attr) {
+			.opcode    = MLX5_OPCODE_SEND,
+			.mss       = cpu_to_be16(0),
+			.ihs       = ihs,
+			.num_bytes = max_t(unsigned int, skb->len, ETH_ZLEN),
+			.headlen   = skb_headlen(skb) - ihs,
+		};
+
+		stats->packets++;
+	}
+
+	attr->insz = mlx5e_accel_tx_ids_len(sq, accel);
+	stats->bytes += attr->num_bytes;
+}
+
+static void mlx5e_sq_calc_wqe_attr(struct sk_buff *skb, const struct mlx5e_tx_attr *attr,
+				   struct mlx5e_tx_wqe_attr *wqe_attr)
+{
+	u16 ds_cnt = MLX5E_TX_WQE_EMPTY_DS_COUNT;
+	u16 ds_cnt_inl = 0;
+	u16 ds_cnt_ids = 0;
+
+	/* Sync the calculation with MLX5E_MAX_TX_WQEBBS. */
+
+	if (attr->insz)
+		ds_cnt_ids = DIV_ROUND_UP(sizeof(struct mlx5_wqe_inline_seg) + attr->insz,
+					  MLX5_SEND_WQE_DS);
+
+	ds_cnt += !!attr->headlen + skb_shinfo(skb)->nr_frags + ds_cnt_ids;
+	if (attr->ihs) {
+		u16 inl = attr->ihs - INL_HDR_START_SZ;
+
+		if (skb_vlan_tag_present(skb))
+			inl += VLAN_HLEN;
+
+		ds_cnt_inl = DIV_ROUND_UP(inl, MLX5_SEND_WQE_DS);
+		if (WARN_ON_ONCE(ds_cnt_inl > MLX5E_MAX_TX_INLINE_DS))
+			netdev_warn(skb->dev, "ds_cnt_inl = %u > max %u\n", ds_cnt_inl,
+				    (u16)MLX5E_MAX_TX_INLINE_DS);
+		ds_cnt += ds_cnt_inl;
+	}
+
+	*wqe_attr = (struct mlx5e_tx_wqe_attr) {
+		.ds_cnt     = ds_cnt,
+		.ds_cnt_inl = ds_cnt_inl,
+		.ds_cnt_ids = ds_cnt_ids,
+		.num_wqebbs = DIV_ROUND_UP(ds_cnt, MLX5_SEND_WQEBB_NUM_DS),
+	};
+}
+
+static void mlx5e_tx_skb_update_hwts_flags(struct sk_buff *skb)
+{
+	if (unlikely(skb_shinfo(skb)->tx_flags & SKBTX_HW_TSTAMP))
+		skb_shinfo(skb)->tx_flags |= SKBTX_IN_PROGRESS;
+}
+
+static void mlx5e_tx_check_stop(struct mlx5e_txqsq *sq)
+{
+	if (unlikely(!mlx5e_wqc_has_room_for(&sq->wq, sq->cc, sq->pc, sq->stop_room))) {
+		netif_tx_stop_queue(sq->txq);
+		sq->stats->stopped++;
+	}
+}
+
+static void mlx5e_tx_flush(struct mlx5e_txqsq *sq)
+{
+	struct mlx5e_tx_wqe_info *wi;
+	struct mlx5e_tx_wqe *wqe;
+	u16 pi;
+
+	/* Must not be called when a MPWQE session is active but empty. */
+	mlx5e_tx_mpwqe_ensure_complete(sq);
+
+	pi = mlx5_wq_cyc_ctr2ix(&sq->wq, sq->pc);
+	wi = &sq->db.wqe_info[pi];
+
+	*wi = (struct mlx5e_tx_wqe_info) {
+		.num_wqebbs = 1,
+	};
+
+	wqe = mlx5e_post_nop(&sq->wq, sq->sqn, &sq->pc);
+	mlx5e_notify_hw(&sq->wq, sq->pc, sq->uar_map, &wqe->ctrl);
+}
+
+static inline void
+mlx5e_txwqe_complete(struct mlx5e_txqsq *sq, struct sk_buff *skb,
+		     const struct mlx5e_tx_attr *attr,
+		     const struct mlx5e_tx_wqe_attr *wqe_attr, u8 num_dma,
+		     struct mlx5e_tx_wqe_info *wi, struct mlx5_wqe_ctrl_seg *cseg,
+		     bool xmit_more)
+{
+	struct mlx5_wq_cyc *wq = &sq->wq;
+	bool send_doorbell;
+
+	*wi = (struct mlx5e_tx_wqe_info) {
+		.skb = skb,
+		.num_bytes = attr->num_bytes,
+		.num_dma = num_dma,
+		.num_wqebbs = wqe_attr->num_wqebbs,
+		.num_fifo_pkts = 0,
+	};
+
+	cseg->opmod_idx_opcode = cpu_to_be32((sq->pc << 8) | attr->opcode);
+	cseg->qpn_ds           = cpu_to_be32((sq->sqn << 8) | wqe_attr->ds_cnt);
+
+	mlx5e_tx_skb_update_hwts_flags(skb);
+
+	sq->pc += wi->num_wqebbs;
+
+	mlx5e_tx_check_stop(sq);
+
+	if (unlikely(sq->ptpsq)) {
+		mlx5e_skb_cb_hwtstamp_init(skb);
+		mlx5e_skb_fifo_push(&sq->ptpsq->skb_fifo, skb);
+		if (!netif_tx_queue_stopped(sq->txq) &&
+		    !mlx5e_skb_fifo_has_room(&sq->ptpsq->skb_fifo)) {
+			netif_tx_stop_queue(sq->txq);
+			sq->stats->stopped++;
+		}
+		skb_get(skb);
+	}
+
+	send_doorbell = __netdev_tx_sent_queue(sq->txq, attr->num_bytes, xmit_more);
+	if (send_doorbell)
+		mlx5e_notify_hw(wq, sq->pc, sq->uar_map, cseg);
+}
+
+static void
+mlx5e_sq_xmit_wqe(struct mlx5e_txqsq *sq, struct sk_buff *skb,
+		  const struct mlx5e_tx_attr *attr, const struct mlx5e_tx_wqe_attr *wqe_attr,
+		  struct mlx5e_tx_wqe *wqe, u16 pi, bool xmit_more)
+{
+	struct mlx5_wqe_ctrl_seg *cseg;
+	struct mlx5_wqe_eth_seg  *eseg;
+	struct mlx5_wqe_data_seg *dseg;
+	struct mlx5e_tx_wqe_info *wi;
+	u16 ihs = attr->ihs;
+	struct ipv6hdr *h6;
+	struct mlx5e_sq_stats *stats = sq->stats;
+	int num_dma;
+
+	stats->xmit_more += xmit_more;
+
+	/* fill wqe */
+	wi   = &sq->db.wqe_info[pi];
+	cseg = &wqe->ctrl;
+	eseg = &wqe->eth;
+	dseg =  wqe->data;
+
+	eseg->mss = attr->mss;
+
+	if (ihs) {
+		u8 *start = eseg->inline_hdr.start;
+
+		if (unlikely(attr->hopbyhop)) {
+			/* remove the HBH header.
+			 * Layout: [Ethernet header][IPv6 header][HBH][TCP header]
+			 */
+			if (skb_vlan_tag_present(skb)) {
+				mlx5e_insert_vlan(start, skb, ETH_HLEN + sizeof(*h6));
+				ihs += VLAN_HLEN;
+				h6 = (struct ipv6hdr *)(start + sizeof(struct vlan_ethhdr));
+			} else {
+				unsafe_memcpy(start, skb->data,
+					      ETH_HLEN + sizeof(*h6),
+					      MLX5_UNSAFE_MEMCPY_DISCLAIMER);
+				h6 = (struct ipv6hdr *)(start + ETH_HLEN);
+			}
+			h6->nexthdr = IPPROTO_TCP;
+			/* Copy the TCP header after the IPv6 one */
+			memcpy(h6 + 1,
+			       skb->data + ETH_HLEN + sizeof(*h6) +
+					sizeof(struct hop_jumbo_hdr),
+			       tcp_hdrlen(skb));
+			/* Leave ipv6 payload_len set to 0, as LSO v2 specs request. */
+		} else if (skb_vlan_tag_present(skb)) {
+			mlx5e_insert_vlan(start, skb, ihs);
+			ihs += VLAN_HLEN;
+			stats->added_vlan_packets++;
+		} else {
+			unsafe_memcpy(eseg->inline_hdr.start, skb->data,
+				      attr->ihs,
+				      MLX5_UNSAFE_MEMCPY_DISCLAIMER);
+		}
+		eseg->inline_hdr.sz |= cpu_to_be16(ihs);
+		dseg += wqe_attr->ds_cnt_inl;
+	} else if (skb_vlan_tag_present(skb)) {
+		eseg->insert.type = cpu_to_be16(MLX5_ETH_WQE_INSERT_VLAN);
+		if (skb->vlan_proto == cpu_to_be16(ETH_P_8021AD))
+			eseg->insert.type |= cpu_to_be16(MLX5_ETH_WQE_SVLAN);
+		eseg->insert.vlan_tci = cpu_to_be16(skb_vlan_tag_get(skb));
+		stats->added_vlan_packets++;
+	}
+
+	dseg += wqe_attr->ds_cnt_ids;
+	num_dma = mlx5e_txwqe_build_dsegs(sq, skb, skb->data + attr->ihs + attr->hopbyhop,
+					  attr->headlen, dseg);
+	if (unlikely(num_dma < 0))
+		goto err_drop;
+
+	mlx5e_txwqe_complete(sq, skb, attr, wqe_attr, num_dma, wi, cseg, xmit_more);
+
+	return;
+
+err_drop:
+	stats->dropped++;
+	dev_kfree_skb_any(skb);
+	mlx5e_tx_flush(sq);
+}
+
+static bool mlx5e_tx_skb_supports_mpwqe(struct sk_buff *skb, struct mlx5e_tx_attr *attr)
+{
+	return !skb_is_nonlinear(skb) && !skb_vlan_tag_present(skb) && !attr->ihs &&
+	       !attr->insz && !mlx5e_macsec_skb_is_offload(skb);
+}
+
+static bool mlx5e_tx_mpwqe_same_eseg(struct mlx5e_txqsq *sq, struct mlx5_wqe_eth_seg *eseg)
+{
+	struct mlx5e_tx_mpwqe *session = &sq->mpwqe;
+
+	/* Assumes the session is already running and has at least one packet. */
+	return !memcmp(&session->wqe->eth, eseg, MLX5E_ACCEL_ESEG_LEN);
+}
+
+static void mlx5e_tx_mpwqe_session_start(struct mlx5e_txqsq *sq,
+					 struct mlx5_wqe_eth_seg *eseg)
+{
+	struct mlx5e_tx_mpwqe *session = &sq->mpwqe;
+	struct mlx5e_tx_wqe *wqe;
+	u16 pi;
+
+	pi = mlx5e_txqsq_get_next_pi(sq, sq->max_sq_mpw_wqebbs);
+	wqe = MLX5E_TX_FETCH_WQE(sq, pi);
+	net_prefetchw(wqe->data);
+
+	*session = (struct mlx5e_tx_mpwqe) {
+		.wqe = wqe,
+		.bytes_count = 0,
+		.ds_count = MLX5E_TX_WQE_EMPTY_DS_COUNT,
+		.pkt_count = 0,
+		.inline_on = 0,
+	};
+
+	memcpy(&session->wqe->eth, eseg, MLX5E_ACCEL_ESEG_LEN);
+
+	sq->stats->mpwqe_blks++;
+}
+
+static bool mlx5e_tx_mpwqe_session_is_active(struct mlx5e_txqsq *sq)
+{
+	return sq->mpwqe.wqe;
+}
+
+static void mlx5e_tx_mpwqe_add_dseg(struct mlx5e_txqsq *sq, struct mlx5e_xmit_data *txd)
+{
+	struct mlx5e_tx_mpwqe *session = &sq->mpwqe;
+	struct mlx5_wqe_data_seg *dseg;
+
+	dseg = (struct mlx5_wqe_data_seg *)session->wqe + session->ds_count;
+
+	session->pkt_count++;
+	session->bytes_count += txd->len;
+
+	dseg->addr = cpu_to_be64(txd->dma_addr);
+	dseg->byte_count = cpu_to_be32(txd->len);
+	dseg->lkey = sq->mkey_be;
+	session->ds_count++;
+
+	sq->stats->mpwqe_pkts++;
+}
+
+static struct mlx5_wqe_ctrl_seg *mlx5e_tx_mpwqe_session_complete(struct mlx5e_txqsq *sq)
+{
+	struct mlx5e_tx_mpwqe *session = &sq->mpwqe;
+	u8 ds_count = session->ds_count;
+	struct mlx5_wqe_ctrl_seg *cseg;
+	struct mlx5e_tx_wqe_info *wi;
+	u16 pi;
+
+	cseg = &session->wqe->ctrl;
+	cseg->opmod_idx_opcode = cpu_to_be32((sq->pc << 8) | MLX5_OPCODE_ENHANCED_MPSW);
+	cseg->qpn_ds = cpu_to_be32((sq->sqn << 8) | ds_count);
+
+	pi = mlx5_wq_cyc_ctr2ix(&sq->wq, sq->pc);
+	wi = &sq->db.wqe_info[pi];
+	*wi = (struct mlx5e_tx_wqe_info) {
+		.skb = NULL,
+		.num_bytes = session->bytes_count,
+		.num_wqebbs = DIV_ROUND_UP(ds_count, MLX5_SEND_WQEBB_NUM_DS),
+		.num_dma = session->pkt_count,
+		.num_fifo_pkts = session->pkt_count,
+	};
+
+	sq->pc += wi->num_wqebbs;
+
+	session->wqe = NULL;
+
+	mlx5e_tx_check_stop(sq);
+
+	return cseg;
+}
+
+static void
+mlx5e_sq_xmit_mpwqe(struct mlx5e_txqsq *sq, struct sk_buff *skb,
+		    struct mlx5_wqe_eth_seg *eseg, bool xmit_more)
+{
+	struct mlx5_wqe_ctrl_seg *cseg;
+	struct mlx5e_xmit_data txd;
+
+	txd.data = skb->data;
+	txd.len = skb->len;
+
+	txd.dma_addr = dma_map_single(sq->pdev, txd.data, txd.len, DMA_TO_DEVICE);
+	if (unlikely(dma_mapping_error(sq->pdev, txd.dma_addr)))
+		goto err_unmap;
+
+	if (!mlx5e_tx_mpwqe_session_is_active(sq)) {
+		mlx5e_tx_mpwqe_session_start(sq, eseg);
+	} else if (!mlx5e_tx_mpwqe_same_eseg(sq, eseg)) {
+		mlx5e_tx_mpwqe_session_complete(sq);
+		mlx5e_tx_mpwqe_session_start(sq, eseg);
+	}
+
+	sq->stats->xmit_more += xmit_more;
+
+	mlx5e_dma_push(sq, txd.dma_addr, txd.len, MLX5E_DMA_MAP_SINGLE);
+	mlx5e_skb_fifo_push(&sq->db.skb_fifo, skb);
+	mlx5e_tx_mpwqe_add_dseg(sq, &txd);
+	mlx5e_tx_skb_update_hwts_flags(skb);
+
+	if (unlikely(mlx5e_tx_mpwqe_is_full(&sq->mpwqe, sq->max_sq_mpw_wqebbs))) {
+		/* Might stop the queue and affect the retval of __netdev_tx_sent_queue. */
+		cseg = mlx5e_tx_mpwqe_session_complete(sq);
+
+		if (__netdev_tx_sent_queue(sq->txq, txd.len, xmit_more))
+			mlx5e_notify_hw(&sq->wq, sq->pc, sq->uar_map, cseg);
+	} else if (__netdev_tx_sent_queue(sq->txq, txd.len, xmit_more)) {
+		/* Might stop the queue, but we were asked to ring the doorbell anyway. */
+		cseg = mlx5e_tx_mpwqe_session_complete(sq);
+
+		mlx5e_notify_hw(&sq->wq, sq->pc, sq->uar_map, cseg);
+	}
+
+	return;
+
+err_unmap:
+	mlx5e_dma_unmap_wqe_err(sq, 1);
+	sq->stats->dropped++;
+	dev_kfree_skb_any(skb);
+	mlx5e_tx_flush(sq);
+}
+
+void mlx5e_tx_mpwqe_ensure_complete(struct mlx5e_txqsq *sq)
+{
+	/* Unlikely in non-MPWQE workloads; not important in MPWQE workloads. */
+	if (unlikely(mlx5e_tx_mpwqe_session_is_active(sq)))
+		mlx5e_tx_mpwqe_session_complete(sq);
+}
+
+static void mlx5e_cqe_ts_id_eseg(struct mlx5e_ptpsq *ptpsq, struct sk_buff *skb,
+				 struct mlx5_wqe_eth_seg *eseg)
+{
+	if (ptpsq->ts_cqe_ctr_mask && unlikely(skb_shinfo(skb)->tx_flags & SKBTX_HW_TSTAMP))
+		eseg->flow_table_metadata = cpu_to_be32(ptpsq->skb_fifo_pc &
+							ptpsq->ts_cqe_ctr_mask);
+}
+
+static void mlx5e_txwqe_build_eseg(struct mlx5e_priv *priv, struct mlx5e_txqsq *sq,
+				   struct sk_buff *skb, struct mlx5e_accel_tx_state *accel,
+				   struct mlx5_wqe_eth_seg *eseg, u16 ihs)
+{
+	mlx5e_accel_tx_eseg(priv, skb, eseg, ihs);
+	mlx5e_txwqe_build_eseg_csum(sq, skb, accel, eseg);
+	if (unlikely(sq->ptpsq))
+		mlx5e_cqe_ts_id_eseg(sq->ptpsq, skb, eseg);
+}
+
+netdev_tx_t mlx5e_xmit(struct sk_buff *skb, struct net_device *dev)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	struct mlx5e_accel_tx_state accel = {};
+	struct mlx5e_tx_wqe_attr wqe_attr;
+	struct mlx5e_tx_attr attr;
+	struct mlx5e_tx_wqe *wqe;
+	struct mlx5e_txqsq *sq;
+	u16 pi;
+
+	/* All changes to txq2sq are performed in sync with mlx5e_xmit, when the
+	 * queue being changed is disabled, and smp_wmb guarantees that the
+	 * changes are visible before mlx5e_xmit tries to read from txq2sq. It
+	 * guarantees that the value of txq2sq[qid] doesn't change while
+	 * mlx5e_xmit is running on queue number qid. smb_wmb is paired with
+	 * HARD_TX_LOCK around ndo_start_xmit, which serves as an ACQUIRE.
+	 */
+	sq = priv->txq2sq[skb_get_queue_mapping(skb)];
+	if (unlikely(!sq)) {
+		/* Two cases when sq can be NULL:
+		 * 1. The HTB node is registered, and mlx5e_select_queue
+		 * selected its queue ID, but the SQ itself is not yet created.
+		 * 2. HTB SQ creation failed. Similar to the previous case, but
+		 * the SQ won't be created.
+		 */
+		dev_kfree_skb_any(skb);
+		return NETDEV_TX_OK;
+	}
+
+	/* May send SKBs and WQEs. */
+	if (unlikely(!mlx5e_accel_tx_begin(dev, sq, skb, &accel)))
+		return NETDEV_TX_OK;
+
+	mlx5e_sq_xmit_prepare(sq, skb, &accel, &attr);
+
+	if (test_bit(MLX5E_SQ_STATE_MPWQE, &sq->state)) {
+		if (mlx5e_tx_skb_supports_mpwqe(skb, &attr)) {
+			struct mlx5_wqe_eth_seg eseg = {};
+
+			mlx5e_txwqe_build_eseg(priv, sq, skb, &accel, &eseg, attr.ihs);
+			mlx5e_sq_xmit_mpwqe(sq, skb, &eseg, netdev_xmit_more());
+			return NETDEV_TX_OK;
+		}
+
+		mlx5e_tx_mpwqe_ensure_complete(sq);
+	}
+
+	mlx5e_sq_calc_wqe_attr(skb, &attr, &wqe_attr);
+	pi = mlx5e_txqsq_get_next_pi(sq, wqe_attr.num_wqebbs);
+	wqe = MLX5E_TX_FETCH_WQE(sq, pi);
+
+	/* May update the WQE, but may not post other WQEs. */
+	mlx5e_accel_tx_finish(sq, wqe, &accel,
+			      (struct mlx5_wqe_inline_seg *)(wqe->data + wqe_attr.ds_cnt_inl));
+	mlx5e_txwqe_build_eseg(priv, sq, skb, &accel, &wqe->eth, attr.ihs);
+	mlx5e_sq_xmit_wqe(sq, skb, &attr, &wqe_attr, wqe, pi, netdev_xmit_more());
+
+	return NETDEV_TX_OK;
+}
+
+void mlx5e_sq_xmit_simple(struct mlx5e_txqsq *sq, struct sk_buff *skb, bool xmit_more)
+{
+	struct mlx5e_tx_wqe_attr wqe_attr;
+	struct mlx5e_tx_attr attr;
+	struct mlx5e_tx_wqe *wqe;
+	u16 pi;
+
+	mlx5e_sq_xmit_prepare(sq, skb, NULL, &attr);
+	mlx5e_sq_calc_wqe_attr(skb, &attr, &wqe_attr);
+	pi = mlx5e_txqsq_get_next_pi(sq, wqe_attr.num_wqebbs);
+	wqe = MLX5E_TX_FETCH_WQE(sq, pi);
+	mlx5e_txwqe_build_eseg_csum(sq, skb, NULL, &wqe->eth);
+	mlx5e_sq_xmit_wqe(sq, skb, &attr, &wqe_attr, wqe, pi, xmit_more);
+}
+
+static void mlx5e_tx_wi_dma_unmap(struct mlx5e_txqsq *sq, struct mlx5e_tx_wqe_info *wi,
+				  u32 *dma_fifo_cc)
+{
+	int i;
+
+	for (i = 0; i < wi->num_dma; i++) {
+		struct mlx5e_sq_dma *dma = mlx5e_dma_get(sq, (*dma_fifo_cc)++);
+
+		mlx5e_tx_dma_unmap(sq->pdev, dma);
+	}
+}
+
+static void mlx5e_consume_skb(struct mlx5e_txqsq *sq, struct sk_buff *skb,
+			      struct mlx5_cqe64 *cqe, int napi_budget)
+{
+	if (unlikely(skb_shinfo(skb)->tx_flags & SKBTX_HW_TSTAMP)) {
+		struct skb_shared_hwtstamps hwts = {};
+		u64 ts = get_cqe_ts(cqe);
+
+		hwts.hwtstamp = mlx5e_cqe_ts_to_ns(sq->ptp_cyc2time, sq->clock, ts);
+		if (sq->ptpsq)
+			mlx5e_skb_cb_hwtstamp_handler(skb, MLX5E_SKB_CB_CQE_HWTSTAMP,
+						      hwts.hwtstamp, sq->ptpsq->cq_stats);
+		else
+			skb_tstamp_tx(skb, &hwts);
+	}
+
+	napi_consume_skb(skb, napi_budget);
+}
+
+static void mlx5e_tx_wi_consume_fifo_skbs(struct mlx5e_txqsq *sq, struct mlx5e_tx_wqe_info *wi,
+					  struct mlx5_cqe64 *cqe, int napi_budget)
+{
+	int i;
+
+	for (i = 0; i < wi->num_fifo_pkts; i++) {
+		struct sk_buff *skb = mlx5e_skb_fifo_pop(&sq->db.skb_fifo);
+
+		mlx5e_consume_skb(sq, skb, cqe, napi_budget);
+	}
+}
+
+void mlx5e_txqsq_wake(struct mlx5e_txqsq *sq)
+{
+	if (netif_tx_queue_stopped(sq->txq) &&
+	    mlx5e_wqc_has_room_for(&sq->wq, sq->cc, sq->pc, sq->stop_room) &&
+	    mlx5e_ptpsq_fifo_has_room(sq) &&
+	    !test_bit(MLX5E_SQ_STATE_RECOVERING, &sq->state)) {
+		netif_tx_wake_queue(sq->txq);
+		sq->stats->wake++;
+	}
+}
+
+bool mlx5e_poll_tx_cq(struct mlx5e_cq *cq, int napi_budget)
+{
+	struct mlx5e_sq_stats *stats;
+	struct mlx5e_txqsq *sq;
+	struct mlx5_cqe64 *cqe;
+	u32 dma_fifo_cc;
+	u32 nbytes;
+	u16 npkts;
+	u16 sqcc;
+	int i;
+
+	sq = container_of(cq, struct mlx5e_txqsq, cq);
+
+	if (unlikely(!test_bit(MLX5E_SQ_STATE_ENABLED, &sq->state)))
+		return false;
+
+	cqe = mlx5_cqwq_get_cqe(&cq->wq);
+	if (!cqe)
+		return false;
+
+	stats = sq->stats;
+
+	npkts = 0;
+	nbytes = 0;
+
+	/* sq->cc must be updated only after mlx5_cqwq_update_db_record(),
+	 * otherwise a cq overrun may occur
+	 */
+	sqcc = sq->cc;
+
+	/* avoid dirtying sq cache line every cqe */
+	dma_fifo_cc = sq->dma_fifo_cc;
+
+	i = 0;
+	do {
+		struct mlx5e_tx_wqe_info *wi;
+		u16 wqe_counter;
+		bool last_wqe;
+		u16 ci;
+
+		mlx5_cqwq_pop(&cq->wq);
+
+		wqe_counter = be16_to_cpu(cqe->wqe_counter);
+
+		do {
+			last_wqe = (sqcc == wqe_counter);
+
+			ci = mlx5_wq_cyc_ctr2ix(&sq->wq, sqcc);
+			wi = &sq->db.wqe_info[ci];
+
+			sqcc += wi->num_wqebbs;
+
+			if (likely(wi->skb)) {
+				mlx5e_tx_wi_dma_unmap(sq, wi, &dma_fifo_cc);
+				mlx5e_consume_skb(sq, wi->skb, cqe, napi_budget);
+
+				npkts++;
+				nbytes += wi->num_bytes;
+				continue;
+			}
+
+			if (unlikely(mlx5e_ktls_tx_try_handle_resync_dump_comp(sq, wi,
+									       &dma_fifo_cc)))
+				continue;
+
+			if (wi->num_fifo_pkts) {
+				mlx5e_tx_wi_dma_unmap(sq, wi, &dma_fifo_cc);
+				mlx5e_tx_wi_consume_fifo_skbs(sq, wi, cqe, napi_budget);
+
+				npkts += wi->num_fifo_pkts;
+				nbytes += wi->num_bytes;
+			}
+		} while (!last_wqe);
+
+		if (unlikely(get_cqe_opcode(cqe) == MLX5_CQE_REQ_ERR)) {
+			if (!test_and_set_bit(MLX5E_SQ_STATE_RECOVERING,
+					      &sq->state)) {
+				mlx5e_dump_error_cqe(&sq->cq, sq->sqn,
+						     (struct mlx5_err_cqe *)cqe);
+				mlx5_wq_cyc_wqe_dump(&sq->wq, ci, wi->num_wqebbs);
+				queue_work(cq->priv->wq, &sq->recover_work);
+			}
+			stats->cqe_err++;
+		}
+
+	} while ((++i < MLX5E_TX_CQ_POLL_BUDGET) && (cqe = mlx5_cqwq_get_cqe(&cq->wq)));
+
+	stats->cqes += i;
+
+	mlx5_cqwq_update_db_record(&cq->wq);
+
+	/* ensure cq space is freed before enabling more cqes */
+	wmb();
+
+	sq->dma_fifo_cc = dma_fifo_cc;
+	sq->cc = sqcc;
+
+	netdev_tx_completed_queue(sq->txq, npkts, nbytes);
+
+	mlx5e_txqsq_wake(sq);
+
+	return (i == MLX5E_TX_CQ_POLL_BUDGET);
+}
+
+static void mlx5e_tx_wi_kfree_fifo_skbs(struct mlx5e_txqsq *sq, struct mlx5e_tx_wqe_info *wi)
+{
+	int i;
+
+	for (i = 0; i < wi->num_fifo_pkts; i++)
+		dev_kfree_skb_any(mlx5e_skb_fifo_pop(&sq->db.skb_fifo));
+}
+
+void mlx5e_free_txqsq_descs(struct mlx5e_txqsq *sq)
+{
+	struct mlx5e_tx_wqe_info *wi;
+	u32 dma_fifo_cc, nbytes = 0;
+	u16 ci, sqcc, npkts = 0;
+
+	sqcc = sq->cc;
+	dma_fifo_cc = sq->dma_fifo_cc;
+
+	while (sqcc != sq->pc) {
+		ci = mlx5_wq_cyc_ctr2ix(&sq->wq, sqcc);
+		wi = &sq->db.wqe_info[ci];
+
+		sqcc += wi->num_wqebbs;
+
+		if (likely(wi->skb)) {
+			mlx5e_tx_wi_dma_unmap(sq, wi, &dma_fifo_cc);
+			dev_kfree_skb_any(wi->skb);
+
+			npkts++;
+			nbytes += wi->num_bytes;
+			continue;
+		}
+
+		if (unlikely(mlx5e_ktls_tx_try_handle_resync_dump_comp(sq, wi, &dma_fifo_cc)))
+			continue;
+
+		if (wi->num_fifo_pkts) {
+			mlx5e_tx_wi_dma_unmap(sq, wi, &dma_fifo_cc);
+			mlx5e_tx_wi_kfree_fifo_skbs(sq, wi);
+
+			npkts += wi->num_fifo_pkts;
+			nbytes += wi->num_bytes;
+		}
+	}
+
+	sq->dma_fifo_cc = dma_fifo_cc;
+	sq->cc = sqcc;
+
+	netdev_tx_completed_queue(sq->txq, npkts, nbytes);
+}
+
+#ifdef CONFIG_MLX5_CORE_IPOIB
+static inline void
+mlx5i_txwqe_build_datagram(struct mlx5_av *av, u32 dqpn, u32 dqkey,
+			   struct mlx5_wqe_datagram_seg *dseg)
+{
+	memcpy(&dseg->av, av, sizeof(struct mlx5_av));
+	dseg->av.dqp_dct = cpu_to_be32(dqpn | MLX5_EXTENDED_UD_AV);
+	dseg->av.key.qkey.qkey = cpu_to_be32(dqkey);
+}
+
+static void mlx5i_sq_calc_wqe_attr(struct sk_buff *skb,
+				   const struct mlx5e_tx_attr *attr,
+				   struct mlx5e_tx_wqe_attr *wqe_attr)
+{
+	u16 ds_cnt = sizeof(struct mlx5i_tx_wqe) / MLX5_SEND_WQE_DS;
+	u16 ds_cnt_inl = 0;
+
+	ds_cnt += !!attr->headlen + skb_shinfo(skb)->nr_frags;
+
+	if (attr->ihs) {
+		u16 inl = attr->ihs - INL_HDR_START_SZ;
+
+		ds_cnt_inl = DIV_ROUND_UP(inl, MLX5_SEND_WQE_DS);
+		ds_cnt += ds_cnt_inl;
+	}
+
+	*wqe_attr = (struct mlx5e_tx_wqe_attr) {
+		.ds_cnt     = ds_cnt,
+		.ds_cnt_inl = ds_cnt_inl,
+		.num_wqebbs = DIV_ROUND_UP(ds_cnt, MLX5_SEND_WQEBB_NUM_DS),
+	};
+}
+
+void mlx5i_sq_xmit(struct mlx5e_txqsq *sq, struct sk_buff *skb,
+		   struct mlx5_av *av, u32 dqpn, u32 dqkey, bool xmit_more)
+{
+	struct mlx5e_tx_wqe_attr wqe_attr;
+	struct mlx5e_tx_attr attr;
+	struct mlx5i_tx_wqe *wqe;
+
+	struct mlx5_wqe_datagram_seg *datagram;
+	struct mlx5_wqe_ctrl_seg *cseg;
+	struct mlx5_wqe_eth_seg  *eseg;
+	struct mlx5_wqe_data_seg *dseg;
+	struct mlx5e_tx_wqe_info *wi;
+
+	struct mlx5e_sq_stats *stats = sq->stats;
+	int num_dma;
+	u16 pi;
+
+	mlx5e_sq_xmit_prepare(sq, skb, NULL, &attr);
+	mlx5i_sq_calc_wqe_attr(skb, &attr, &wqe_attr);
+
+	pi = mlx5e_txqsq_get_next_pi(sq, wqe_attr.num_wqebbs);
+	wqe = MLX5I_SQ_FETCH_WQE(sq, pi);
+
+	stats->xmit_more += xmit_more;
+
+	/* fill wqe */
+	wi       = &sq->db.wqe_info[pi];
+	cseg     = &wqe->ctrl;
+	datagram = &wqe->datagram;
+	eseg     = &wqe->eth;
+	dseg     =  wqe->data;
+
+	mlx5i_txwqe_build_datagram(av, dqpn, dqkey, datagram);
+
+	mlx5e_txwqe_build_eseg_csum(sq, skb, NULL, eseg);
+
+	eseg->mss = attr.mss;
+
+	if (attr.ihs) {
+		if (unlikely(attr.hopbyhop)) {
+			struct ipv6hdr *h6;
+
+			/* remove the HBH header.
+			 * Layout: [Ethernet header][IPv6 header][HBH][TCP header]
+			 */
+			unsafe_memcpy(eseg->inline_hdr.start, skb->data,
+				      ETH_HLEN + sizeof(*h6),
+				      MLX5_UNSAFE_MEMCPY_DISCLAIMER);
+			h6 = (struct ipv6hdr *)((char *)eseg->inline_hdr.start + ETH_HLEN);
+			h6->nexthdr = IPPROTO_TCP;
+			/* Copy the TCP header after the IPv6 one */
+			unsafe_memcpy(h6 + 1,
+				      skb->data + ETH_HLEN + sizeof(*h6) +
+						  sizeof(struct hop_jumbo_hdr),
+				      tcp_hdrlen(skb),
+				      MLX5_UNSAFE_MEMCPY_DISCLAIMER);
+			/* Leave ipv6 payload_len set to 0, as LSO v2 specs request. */
+		} else {
+			unsafe_memcpy(eseg->inline_hdr.start, skb->data,
+				      attr.ihs,
+				      MLX5_UNSAFE_MEMCPY_DISCLAIMER);
+		}
+		eseg->inline_hdr.sz = cpu_to_be16(attr.ihs);
+		dseg += wqe_attr.ds_cnt_inl;
+	}
+
+	num_dma = mlx5e_txwqe_build_dsegs(sq, skb, skb->data + attr.ihs + attr.hopbyhop,
+					  attr.headlen, dseg);
+	if (unlikely(num_dma < 0))
+		goto err_drop;
+
+	mlx5e_txwqe_complete(sq, skb, &attr, &wqe_attr, num_dma, wi, cseg, xmit_more);
+
+	return;
+
+err_drop:
+	stats->dropped++;
+	dev_kfree_skb_any(skb);
+	mlx5e_tx_flush(sq);
+}
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_txrx.c b/drivers/net/ethernet/mellanox/mlx5/core/en_txrx.c
new file mode 100644
index 000000000..44547b22a
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_txrx.c
@@ -0,0 +1,274 @@
+/*
+ * Copyright (c) 2015, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/irq.h>
+#include <net/xdp_sock_drv.h>
+#include "en.h"
+#include "en/txrx.h"
+#include "en/xdp.h"
+#include "en/xsk/rx.h"
+#include "en/xsk/tx.h"
+#include "en_accel/ktls_txrx.h"
+
+static inline bool mlx5e_channel_no_affinity_change(struct mlx5e_channel *c)
+{
+	int current_cpu = smp_processor_id();
+
+	return cpumask_test_cpu(current_cpu, c->aff_mask);
+}
+
+static void mlx5e_handle_tx_dim(struct mlx5e_txqsq *sq)
+{
+	struct mlx5e_sq_stats *stats = sq->stats;
+	struct dim_sample dim_sample = {};
+
+	if (unlikely(!test_bit(MLX5E_SQ_STATE_AM, &sq->state)))
+		return;
+
+	dim_update_sample(sq->cq.event_ctr, stats->packets, stats->bytes, &dim_sample);
+	net_dim(&sq->dim, dim_sample);
+}
+
+static void mlx5e_handle_rx_dim(struct mlx5e_rq *rq)
+{
+	struct mlx5e_rq_stats *stats = rq->stats;
+	struct dim_sample dim_sample = {};
+
+	if (unlikely(!test_bit(MLX5E_RQ_STATE_AM, &rq->state)))
+		return;
+
+	dim_update_sample(rq->cq.event_ctr, stats->packets, stats->bytes, &dim_sample);
+	net_dim(&rq->dim, dim_sample);
+}
+
+void mlx5e_trigger_irq(struct mlx5e_icosq *sq)
+{
+	struct mlx5_wq_cyc *wq = &sq->wq;
+	struct mlx5e_tx_wqe *nopwqe;
+	u16 pi = mlx5_wq_cyc_ctr2ix(wq, sq->pc);
+
+	sq->db.wqe_info[pi] = (struct mlx5e_icosq_wqe_info) {
+		.wqe_type   = MLX5E_ICOSQ_WQE_NOP,
+		.num_wqebbs = 1,
+	};
+
+	nopwqe = mlx5e_post_nop(wq, sq->sqn, &sq->pc);
+	mlx5e_notify_hw(wq, sq->pc, sq->uar_map, &nopwqe->ctrl);
+}
+
+static bool mlx5e_napi_xsk_post(struct mlx5e_xdpsq *xsksq, struct mlx5e_rq *xskrq)
+{
+	bool need_wakeup = xsk_uses_need_wakeup(xskrq->xsk_pool);
+	bool busy_xsk = false, xsk_rx_alloc_err;
+
+	/* If SQ is empty, there are no TX completions to trigger NAPI, so set
+	 * need_wakeup. Do it before queuing packets for TX to avoid race
+	 * condition with userspace.
+	 */
+	if (need_wakeup && xsksq->pc == xsksq->cc)
+		xsk_set_tx_need_wakeup(xsksq->xsk_pool);
+	busy_xsk |= mlx5e_xsk_tx(xsksq, MLX5E_TX_XSK_POLL_BUDGET);
+	/* If we queued some packets for TX, no need for wakeup anymore. */
+	if (need_wakeup && xsksq->pc != xsksq->cc)
+		xsk_clear_tx_need_wakeup(xsksq->xsk_pool);
+
+	/* If WQ is empty, RX won't trigger NAPI, so set need_wakeup. Do it
+	 * before refilling to avoid race condition with userspace.
+	 */
+	if (need_wakeup && !mlx5e_rqwq_get_cur_sz(xskrq))
+		xsk_set_rx_need_wakeup(xskrq->xsk_pool);
+	xsk_rx_alloc_err = INDIRECT_CALL_2(xskrq->post_wqes,
+					   mlx5e_post_rx_mpwqes,
+					   mlx5e_post_rx_wqes,
+					   xskrq);
+	/* Ask for wakeup if WQ is not full after refill. */
+	if (!need_wakeup)
+		busy_xsk |= xsk_rx_alloc_err;
+	else if (xsk_rx_alloc_err)
+		xsk_set_rx_need_wakeup(xskrq->xsk_pool);
+	else
+		xsk_clear_rx_need_wakeup(xskrq->xsk_pool);
+
+	return busy_xsk;
+}
+
+int mlx5e_napi_poll(struct napi_struct *napi, int budget)
+{
+	struct mlx5e_channel *c = container_of(napi, struct mlx5e_channel,
+					       napi);
+	struct mlx5e_ch_stats *ch_stats = c->stats;
+	struct mlx5e_xdpsq *xsksq = &c->xsksq;
+	struct mlx5e_txqsq __rcu **qos_sqs;
+	struct mlx5e_rq *xskrq = &c->xskrq;
+	struct mlx5e_rq *rq = &c->rq;
+	bool aff_change = false;
+	bool busy_xsk = false;
+	bool busy = false;
+	int work_done = 0;
+	u16 qos_sqs_size;
+	bool xsk_open;
+	int i;
+
+	rcu_read_lock();
+
+	qos_sqs = rcu_dereference(c->qos_sqs);
+
+	xsk_open = test_bit(MLX5E_CHANNEL_STATE_XSK, c->state);
+
+	ch_stats->poll++;
+
+	for (i = 0; i < c->num_tc; i++)
+		busy |= mlx5e_poll_tx_cq(&c->sq[i].cq, budget);
+
+	if (unlikely(qos_sqs)) {
+		smp_rmb(); /* Pairs with mlx5e_qos_alloc_queues. */
+		qos_sqs_size = READ_ONCE(c->qos_sqs_size);
+
+		for (i = 0; i < qos_sqs_size; i++) {
+			struct mlx5e_txqsq *sq = rcu_dereference(qos_sqs[i]);
+
+			if (sq)
+				busy |= mlx5e_poll_tx_cq(&sq->cq, budget);
+		}
+	}
+
+	/* budget=0 means we may be in IRQ context, do as little as possible */
+	if (unlikely(!budget))
+		goto out;
+
+	busy |= mlx5e_poll_xdpsq_cq(&c->xdpsq.cq);
+
+	if (c->xdp)
+		busy |= mlx5e_poll_xdpsq_cq(&c->rq_xdpsq.cq);
+
+	if (xsk_open)
+		work_done = mlx5e_poll_rx_cq(&xskrq->cq, budget);
+
+	if (likely(budget - work_done))
+		work_done += mlx5e_poll_rx_cq(&rq->cq, budget - work_done);
+
+	busy |= work_done == budget;
+
+	mlx5e_poll_ico_cq(&c->icosq.cq);
+	if (mlx5e_poll_ico_cq(&c->async_icosq.cq))
+		/* Don't clear the flag if nothing was polled to prevent
+		 * queueing more WQEs and overflowing the async ICOSQ.
+		 */
+		clear_bit(MLX5E_SQ_STATE_PENDING_XSK_TX, &c->async_icosq.state);
+
+	/* Keep after async ICOSQ CQ poll */
+	if (unlikely(mlx5e_ktls_rx_pending_resync_list(c, budget)))
+		busy |= mlx5e_ktls_rx_handle_resync_list(c, budget);
+
+	busy |= INDIRECT_CALL_2(rq->post_wqes,
+				mlx5e_post_rx_mpwqes,
+				mlx5e_post_rx_wqes,
+				rq);
+	if (xsk_open) {
+		busy |= mlx5e_poll_xdpsq_cq(&xsksq->cq);
+		busy_xsk |= mlx5e_napi_xsk_post(xsksq, xskrq);
+	}
+
+	busy |= busy_xsk;
+
+	if (busy) {
+		if (likely(mlx5e_channel_no_affinity_change(c))) {
+			work_done = budget;
+			goto out;
+		}
+		ch_stats->aff_change++;
+		aff_change = true;
+		if (budget && work_done == budget)
+			work_done--;
+	}
+
+	if (unlikely(!napi_complete_done(napi, work_done)))
+		goto out;
+
+	ch_stats->arm++;
+
+	for (i = 0; i < c->num_tc; i++) {
+		mlx5e_handle_tx_dim(&c->sq[i]);
+		mlx5e_cq_arm(&c->sq[i].cq);
+	}
+	if (unlikely(qos_sqs)) {
+		for (i = 0; i < qos_sqs_size; i++) {
+			struct mlx5e_txqsq *sq = rcu_dereference(qos_sqs[i]);
+
+			if (sq) {
+				mlx5e_handle_tx_dim(sq);
+				mlx5e_cq_arm(&sq->cq);
+			}
+		}
+	}
+
+	mlx5e_handle_rx_dim(rq);
+
+	mlx5e_cq_arm(&rq->cq);
+	mlx5e_cq_arm(&c->icosq.cq);
+	mlx5e_cq_arm(&c->async_icosq.cq);
+	mlx5e_cq_arm(&c->xdpsq.cq);
+
+	if (xsk_open) {
+		mlx5e_handle_rx_dim(xskrq);
+		mlx5e_cq_arm(&xsksq->cq);
+		mlx5e_cq_arm(&xskrq->cq);
+	}
+
+	if (unlikely(aff_change && busy_xsk)) {
+		mlx5e_trigger_irq(&c->icosq);
+		ch_stats->force_irq++;
+	}
+
+out:
+	rcu_read_unlock();
+
+	return work_done;
+}
+
+void mlx5e_completion_event(struct mlx5_core_cq *mcq, struct mlx5_eqe *eqe)
+{
+	struct mlx5e_cq *cq = container_of(mcq, struct mlx5e_cq, mcq);
+
+	napi_schedule(cq->napi);
+	cq->event_ctr++;
+	cq->ch_stats->events++;
+}
+
+void mlx5e_cq_error_event(struct mlx5_core_cq *mcq, enum mlx5_event event)
+{
+	struct mlx5e_cq *cq = container_of(mcq, struct mlx5e_cq, mcq);
+	struct net_device *netdev = cq->netdev;
+
+	netdev_err(netdev, "%s: cqn=0x%.6x event=0x%.2x\n",
+		   __func__, mcq->cqn, event);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eq.c b/drivers/net/ethernet/mellanox/mlx5/core/eq.c
new file mode 100644
index 000000000..e112b5685
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eq.c
@@ -0,0 +1,1150 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/*
+ * Copyright (c) 2013-2021, Mellanox Technologies inc.  All rights reserved.
+ */
+
+#include <linux/interrupt.h>
+#include <linux/notifier.h>
+#include <linux/mlx5/driver.h>
+#include <linux/mlx5/vport.h>
+#include <linux/mlx5/eq.h>
+#ifdef CONFIG_RFS_ACCEL
+#include <linux/cpu_rmap.h>
+#endif
+#include "mlx5_core.h"
+#include "lib/eq.h"
+#include "fpga/core.h"
+#include "eswitch.h"
+#include "lib/clock.h"
+#include "diag/fw_tracer.h"
+#include "mlx5_irq.h"
+#include "devlink.h"
+
+enum {
+	MLX5_EQE_OWNER_INIT_VAL	= 0x1,
+};
+
+enum {
+	MLX5_EQ_STATE_ARMED		= 0x9,
+	MLX5_EQ_STATE_FIRED		= 0xa,
+	MLX5_EQ_STATE_ALWAYS_ARMED	= 0xb,
+};
+
+enum {
+	MLX5_EQ_DOORBEL_OFFSET	= 0x40,
+};
+
+/* budget must be smaller than MLX5_NUM_SPARE_EQE to guarantee that we update
+ * the ci before we polled all the entries in the EQ. MLX5_NUM_SPARE_EQE is
+ * used to set the EQ size, budget must be smaller than the EQ size.
+ */
+enum {
+	MLX5_EQ_POLLING_BUDGET	= 128,
+};
+
+static_assert(MLX5_EQ_POLLING_BUDGET <= MLX5_NUM_SPARE_EQE);
+
+struct mlx5_eq_table {
+	struct list_head        comp_eqs_list;
+	struct mlx5_eq_async    pages_eq;
+	struct mlx5_eq_async    cmd_eq;
+	struct mlx5_eq_async    async_eq;
+
+	struct atomic_notifier_head nh[MLX5_EVENT_TYPE_MAX];
+
+	/* Since CQ DB is stored in async_eq */
+	struct mlx5_nb          cq_err_nb;
+
+	struct mutex            lock; /* sync async eqs creations */
+	int			num_comp_eqs;
+	struct mlx5_irq_table	*irq_table;
+	struct mlx5_irq         **comp_irqs;
+	struct mlx5_irq         *ctrl_irq;
+#ifdef CONFIG_RFS_ACCEL
+	struct cpu_rmap		*rmap;
+#endif
+};
+
+#define MLX5_ASYNC_EVENT_MASK ((1ull << MLX5_EVENT_TYPE_PATH_MIG)	    | \
+			       (1ull << MLX5_EVENT_TYPE_COMM_EST)	    | \
+			       (1ull << MLX5_EVENT_TYPE_SQ_DRAINED)	    | \
+			       (1ull << MLX5_EVENT_TYPE_CQ_ERROR)	    | \
+			       (1ull << MLX5_EVENT_TYPE_WQ_CATAS_ERROR)	    | \
+			       (1ull << MLX5_EVENT_TYPE_PATH_MIG_FAILED)    | \
+			       (1ull << MLX5_EVENT_TYPE_WQ_INVAL_REQ_ERROR) | \
+			       (1ull << MLX5_EVENT_TYPE_WQ_ACCESS_ERROR)    | \
+			       (1ull << MLX5_EVENT_TYPE_PORT_CHANGE)	    | \
+			       (1ull << MLX5_EVENT_TYPE_SRQ_CATAS_ERROR)    | \
+			       (1ull << MLX5_EVENT_TYPE_SRQ_LAST_WQE)	    | \
+			       (1ull << MLX5_EVENT_TYPE_SRQ_RQ_LIMIT))
+
+static int mlx5_cmd_destroy_eq(struct mlx5_core_dev *dev, u8 eqn)
+{
+	u32 in[MLX5_ST_SZ_DW(destroy_eq_in)] = {};
+
+	MLX5_SET(destroy_eq_in, in, opcode, MLX5_CMD_OP_DESTROY_EQ);
+	MLX5_SET(destroy_eq_in, in, eq_number, eqn);
+	return mlx5_cmd_exec_in(dev, destroy_eq, in);
+}
+
+/* caller must eventually call mlx5_cq_put on the returned cq */
+static struct mlx5_core_cq *mlx5_eq_cq_get(struct mlx5_eq *eq, u32 cqn)
+{
+	struct mlx5_cq_table *table = &eq->cq_table;
+	struct mlx5_core_cq *cq = NULL;
+
+	rcu_read_lock();
+	cq = radix_tree_lookup(&table->tree, cqn);
+	if (likely(cq))
+		mlx5_cq_hold(cq);
+	rcu_read_unlock();
+
+	return cq;
+}
+
+static int mlx5_eq_comp_int(struct notifier_block *nb,
+			    __always_unused unsigned long action,
+			    __always_unused void *data)
+{
+	struct mlx5_eq_comp *eq_comp =
+		container_of(nb, struct mlx5_eq_comp, irq_nb);
+	struct mlx5_eq *eq = &eq_comp->core;
+	struct mlx5_eqe *eqe;
+	int num_eqes = 0;
+	u32 cqn = -1;
+
+	eqe = next_eqe_sw(eq);
+	if (!eqe)
+		goto out;
+
+	do {
+		struct mlx5_core_cq *cq;
+
+		/* Make sure we read EQ entry contents after we've
+		 * checked the ownership bit.
+		 */
+		dma_rmb();
+		/* Assume (eqe->type) is always MLX5_EVENT_TYPE_COMP */
+		cqn = be32_to_cpu(eqe->data.comp.cqn) & 0xffffff;
+
+		cq = mlx5_eq_cq_get(eq, cqn);
+		if (likely(cq)) {
+			++cq->arm_sn;
+			cq->comp(cq, eqe);
+			mlx5_cq_put(cq);
+		} else {
+			dev_dbg_ratelimited(eq->dev->device,
+					    "Completion event for bogus CQ 0x%x\n", cqn);
+		}
+
+		++eq->cons_index;
+
+	} while ((++num_eqes < MLX5_EQ_POLLING_BUDGET) && (eqe = next_eqe_sw(eq)));
+
+out:
+	eq_update_ci(eq, 1);
+
+	if (cqn != -1)
+		tasklet_schedule(&eq_comp->tasklet_ctx.task);
+
+	return 0;
+}
+
+/* Some architectures don't latch interrupts when they are disabled, so using
+ * mlx5_eq_poll_irq_disabled could end up losing interrupts while trying to
+ * avoid losing them.  It is not recommended to use it, unless this is the last
+ * resort.
+ */
+u32 mlx5_eq_poll_irq_disabled(struct mlx5_eq_comp *eq)
+{
+	u32 count_eqe;
+
+	disable_irq(eq->core.irqn);
+	count_eqe = eq->core.cons_index;
+	mlx5_eq_comp_int(&eq->irq_nb, 0, NULL);
+	count_eqe = eq->core.cons_index - count_eqe;
+	enable_irq(eq->core.irqn);
+
+	return count_eqe;
+}
+
+static void mlx5_eq_async_int_lock(struct mlx5_eq_async *eq, bool recovery,
+				   unsigned long *flags)
+	__acquires(&eq->lock)
+{
+	if (!recovery)
+		spin_lock(&eq->lock);
+	else
+		spin_lock_irqsave(&eq->lock, *flags);
+}
+
+static void mlx5_eq_async_int_unlock(struct mlx5_eq_async *eq, bool recovery,
+				     unsigned long *flags)
+	__releases(&eq->lock)
+{
+	if (!recovery)
+		spin_unlock(&eq->lock);
+	else
+		spin_unlock_irqrestore(&eq->lock, *flags);
+}
+
+enum async_eq_nb_action {
+	ASYNC_EQ_IRQ_HANDLER = 0,
+	ASYNC_EQ_RECOVER = 1,
+};
+
+static int mlx5_eq_async_int(struct notifier_block *nb,
+			     unsigned long action, void *data)
+{
+	struct mlx5_eq_async *eq_async =
+		container_of(nb, struct mlx5_eq_async, irq_nb);
+	struct mlx5_eq *eq = &eq_async->core;
+	struct mlx5_eq_table *eqt;
+	struct mlx5_core_dev *dev;
+	struct mlx5_eqe *eqe;
+	unsigned long flags;
+	int num_eqes = 0;
+	bool recovery;
+
+	dev = eq->dev;
+	eqt = dev->priv.eq_table;
+
+	recovery = action == ASYNC_EQ_RECOVER;
+	mlx5_eq_async_int_lock(eq_async, recovery, &flags);
+
+	eqe = next_eqe_sw(eq);
+	if (!eqe)
+		goto out;
+
+	do {
+		/*
+		 * Make sure we read EQ entry contents after we've
+		 * checked the ownership bit.
+		 */
+		dma_rmb();
+
+		atomic_notifier_call_chain(&eqt->nh[eqe->type], eqe->type, eqe);
+		atomic_notifier_call_chain(&eqt->nh[MLX5_EVENT_TYPE_NOTIFY_ANY], eqe->type, eqe);
+
+		++eq->cons_index;
+
+	} while ((++num_eqes < MLX5_EQ_POLLING_BUDGET) && (eqe = next_eqe_sw(eq)));
+
+out:
+	eq_update_ci(eq, 1);
+	mlx5_eq_async_int_unlock(eq_async, recovery, &flags);
+
+	return unlikely(recovery) ? num_eqes : 0;
+}
+
+void mlx5_cmd_eq_recover(struct mlx5_core_dev *dev)
+{
+	struct mlx5_eq_async *eq = &dev->priv.eq_table->cmd_eq;
+	int eqes;
+
+	eqes = mlx5_eq_async_int(&eq->irq_nb, ASYNC_EQ_RECOVER, NULL);
+	if (eqes)
+		mlx5_core_warn(dev, "Recovered %d EQEs on cmd_eq\n", eqes);
+}
+
+static void init_eq_buf(struct mlx5_eq *eq)
+{
+	struct mlx5_eqe *eqe;
+	int i;
+
+	for (i = 0; i < eq_get_size(eq); i++) {
+		eqe = get_eqe(eq, i);
+		eqe->owner = MLX5_EQE_OWNER_INIT_VAL;
+	}
+}
+
+static int
+create_map_eq(struct mlx5_core_dev *dev, struct mlx5_eq *eq,
+	      struct mlx5_eq_param *param)
+{
+	u8 log_eq_size = order_base_2(param->nent + MLX5_NUM_SPARE_EQE);
+	struct mlx5_cq_table *cq_table = &eq->cq_table;
+	u32 out[MLX5_ST_SZ_DW(create_eq_out)] = {0};
+	u8 log_eq_stride = ilog2(MLX5_EQE_SIZE);
+	struct mlx5_priv *priv = &dev->priv;
+	__be64 *pas;
+	u16 vecidx;
+	void *eqc;
+	int inlen;
+	u32 *in;
+	int err;
+	int i;
+
+	/* Init CQ table */
+	memset(cq_table, 0, sizeof(*cq_table));
+	spin_lock_init(&cq_table->lock);
+	INIT_RADIX_TREE(&cq_table->tree, GFP_ATOMIC);
+
+	eq->cons_index = 0;
+
+	err = mlx5_frag_buf_alloc_node(dev, wq_get_byte_sz(log_eq_size, log_eq_stride),
+				       &eq->frag_buf, dev->priv.numa_node);
+	if (err)
+		return err;
+
+	mlx5_init_fbc(eq->frag_buf.frags, log_eq_stride, log_eq_size, &eq->fbc);
+	init_eq_buf(eq);
+
+	eq->irq = param->irq;
+	vecidx = mlx5_irq_get_index(eq->irq);
+
+	inlen = MLX5_ST_SZ_BYTES(create_eq_in) +
+		MLX5_FLD_SZ_BYTES(create_eq_in, pas[0]) * eq->frag_buf.npages;
+
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in) {
+		err = -ENOMEM;
+		goto err_buf;
+	}
+
+	pas = (__be64 *)MLX5_ADDR_OF(create_eq_in, in, pas);
+	mlx5_fill_page_frag_array(&eq->frag_buf, pas);
+
+	MLX5_SET(create_eq_in, in, opcode, MLX5_CMD_OP_CREATE_EQ);
+	if (!param->mask[0] && MLX5_CAP_GEN(dev, log_max_uctx))
+		MLX5_SET(create_eq_in, in, uid, MLX5_SHARED_RESOURCE_UID);
+
+	for (i = 0; i < 4; i++)
+		MLX5_ARRAY_SET64(create_eq_in, in, event_bitmask, i,
+				 param->mask[i]);
+
+	eqc = MLX5_ADDR_OF(create_eq_in, in, eq_context_entry);
+	MLX5_SET(eqc, eqc, log_eq_size, eq->fbc.log_sz);
+	MLX5_SET(eqc, eqc, uar_page, priv->uar->index);
+	MLX5_SET(eqc, eqc, intr, vecidx);
+	MLX5_SET(eqc, eqc, log_page_size,
+		 eq->frag_buf.page_shift - MLX5_ADAPTER_PAGE_SHIFT);
+
+	err = mlx5_cmd_exec(dev, in, inlen, out, sizeof(out));
+	if (err)
+		goto err_in;
+
+	eq->vecidx = vecidx;
+	eq->eqn = MLX5_GET(create_eq_out, out, eq_number);
+	eq->irqn = pci_irq_vector(dev->pdev, vecidx);
+	eq->dev = dev;
+	eq->doorbell = priv->uar->map + MLX5_EQ_DOORBEL_OFFSET;
+
+	err = mlx5_debug_eq_add(dev, eq);
+	if (err)
+		goto err_eq;
+
+	kvfree(in);
+	return 0;
+
+err_eq:
+	mlx5_cmd_destroy_eq(dev, eq->eqn);
+
+err_in:
+	kvfree(in);
+
+err_buf:
+	mlx5_frag_buf_free(dev, &eq->frag_buf);
+	return err;
+}
+
+/**
+ * mlx5_eq_enable - Enable EQ for receiving EQEs
+ * @dev : Device which owns the eq
+ * @eq  : EQ to enable
+ * @nb  : Notifier call block
+ *
+ * Must be called after EQ is created in device.
+ *
+ * @return: 0 if no error
+ */
+int mlx5_eq_enable(struct mlx5_core_dev *dev, struct mlx5_eq *eq,
+		   struct notifier_block *nb)
+{
+	int err;
+
+	err = mlx5_irq_attach_nb(eq->irq, nb);
+	if (!err)
+		eq_update_ci(eq, 1);
+
+	return err;
+}
+EXPORT_SYMBOL(mlx5_eq_enable);
+
+/**
+ * mlx5_eq_disable - Disable EQ for receiving EQEs
+ * @dev : Device which owns the eq
+ * @eq  : EQ to disable
+ * @nb  : Notifier call block
+ *
+ * Must be called before EQ is destroyed.
+ */
+void mlx5_eq_disable(struct mlx5_core_dev *dev, struct mlx5_eq *eq,
+		     struct notifier_block *nb)
+{
+	mlx5_irq_detach_nb(eq->irq, nb);
+}
+EXPORT_SYMBOL(mlx5_eq_disable);
+
+static int destroy_unmap_eq(struct mlx5_core_dev *dev, struct mlx5_eq *eq)
+{
+	int err;
+
+	mlx5_debug_eq_remove(dev, eq);
+
+	err = mlx5_cmd_destroy_eq(dev, eq->eqn);
+	if (err)
+		mlx5_core_warn(dev, "failed to destroy a previously created eq: eqn %d\n",
+			       eq->eqn);
+
+	mlx5_frag_buf_free(dev, &eq->frag_buf);
+	return err;
+}
+
+int mlx5_eq_add_cq(struct mlx5_eq *eq, struct mlx5_core_cq *cq)
+{
+	struct mlx5_cq_table *table = &eq->cq_table;
+	int err;
+
+	spin_lock(&table->lock);
+	err = radix_tree_insert(&table->tree, cq->cqn, cq);
+	spin_unlock(&table->lock);
+
+	return err;
+}
+
+void mlx5_eq_del_cq(struct mlx5_eq *eq, struct mlx5_core_cq *cq)
+{
+	struct mlx5_cq_table *table = &eq->cq_table;
+	struct mlx5_core_cq *tmp;
+
+	spin_lock(&table->lock);
+	tmp = radix_tree_delete(&table->tree, cq->cqn);
+	spin_unlock(&table->lock);
+
+	if (!tmp) {
+		mlx5_core_dbg(eq->dev, "cq 0x%x not found in eq 0x%x tree\n",
+			      eq->eqn, cq->cqn);
+		return;
+	}
+
+	if (tmp != cq)
+		mlx5_core_dbg(eq->dev, "corruption on cqn 0x%x in eq 0x%x\n",
+			      eq->eqn, cq->cqn);
+}
+
+int mlx5_eq_table_init(struct mlx5_core_dev *dev)
+{
+	struct mlx5_eq_table *eq_table;
+	int i;
+
+	eq_table = kvzalloc_node(sizeof(*eq_table), GFP_KERNEL,
+				 dev->priv.numa_node);
+	if (!eq_table)
+		return -ENOMEM;
+
+	dev->priv.eq_table = eq_table;
+
+	mlx5_eq_debugfs_init(dev);
+
+	mutex_init(&eq_table->lock);
+	for (i = 0; i < MLX5_EVENT_TYPE_MAX; i++)
+		ATOMIC_INIT_NOTIFIER_HEAD(&eq_table->nh[i]);
+
+	eq_table->irq_table = mlx5_irq_table_get(dev);
+	return 0;
+}
+
+void mlx5_eq_table_cleanup(struct mlx5_core_dev *dev)
+{
+	mlx5_eq_debugfs_cleanup(dev);
+	kvfree(dev->priv.eq_table);
+}
+
+/* Async EQs */
+
+static int create_async_eq(struct mlx5_core_dev *dev,
+			   struct mlx5_eq *eq, struct mlx5_eq_param *param)
+{
+	struct mlx5_eq_table *eq_table = dev->priv.eq_table;
+	int err;
+
+	mutex_lock(&eq_table->lock);
+	err = create_map_eq(dev, eq, param);
+	mutex_unlock(&eq_table->lock);
+	return err;
+}
+
+static int destroy_async_eq(struct mlx5_core_dev *dev, struct mlx5_eq *eq)
+{
+	struct mlx5_eq_table *eq_table = dev->priv.eq_table;
+	int err;
+
+	mutex_lock(&eq_table->lock);
+	err = destroy_unmap_eq(dev, eq);
+	mutex_unlock(&eq_table->lock);
+	return err;
+}
+
+static int cq_err_event_notifier(struct notifier_block *nb,
+				 unsigned long type, void *data)
+{
+	struct mlx5_eq_table *eqt;
+	struct mlx5_core_cq *cq;
+	struct mlx5_eqe *eqe;
+	struct mlx5_eq *eq;
+	u32 cqn;
+
+	/* type == MLX5_EVENT_TYPE_CQ_ERROR */
+
+	eqt = mlx5_nb_cof(nb, struct mlx5_eq_table, cq_err_nb);
+	eq  = &eqt->async_eq.core;
+	eqe = data;
+
+	cqn = be32_to_cpu(eqe->data.cq_err.cqn) & 0xffffff;
+	mlx5_core_warn(eq->dev, "CQ error on CQN 0x%x, syndrome 0x%x\n",
+		       cqn, eqe->data.cq_err.syndrome);
+
+	cq = mlx5_eq_cq_get(eq, cqn);
+	if (unlikely(!cq)) {
+		mlx5_core_warn(eq->dev, "Async event for bogus CQ 0x%x\n", cqn);
+		return NOTIFY_OK;
+	}
+
+	if (cq->event)
+		cq->event(cq, type);
+
+	mlx5_cq_put(cq);
+
+	return NOTIFY_OK;
+}
+
+static void gather_user_async_events(struct mlx5_core_dev *dev, u64 mask[4])
+{
+	__be64 *user_unaffiliated_events;
+	__be64 *user_affiliated_events;
+	int i;
+
+	user_affiliated_events =
+		MLX5_CAP_DEV_EVENT(dev, user_affiliated_events);
+	user_unaffiliated_events =
+		MLX5_CAP_DEV_EVENT(dev, user_unaffiliated_events);
+
+	for (i = 0; i < 4; i++)
+		mask[i] |= be64_to_cpu(user_affiliated_events[i] |
+				       user_unaffiliated_events[i]);
+}
+
+static void gather_async_events_mask(struct mlx5_core_dev *dev, u64 mask[4])
+{
+	u64 async_event_mask = MLX5_ASYNC_EVENT_MASK;
+
+	if (MLX5_VPORT_MANAGER(dev))
+		async_event_mask |= (1ull << MLX5_EVENT_TYPE_NIC_VPORT_CHANGE);
+
+	if (MLX5_CAP_GEN(dev, general_notification_event))
+		async_event_mask |= (1ull << MLX5_EVENT_TYPE_GENERAL_EVENT);
+
+	if (MLX5_CAP_GEN(dev, port_module_event))
+		async_event_mask |= (1ull << MLX5_EVENT_TYPE_PORT_MODULE_EVENT);
+	else
+		mlx5_core_dbg(dev, "port_module_event is not set\n");
+
+	if (MLX5_PPS_CAP(dev))
+		async_event_mask |= (1ull << MLX5_EVENT_TYPE_PPS_EVENT);
+
+	if (MLX5_CAP_GEN(dev, fpga))
+		async_event_mask |= (1ull << MLX5_EVENT_TYPE_FPGA_ERROR) |
+				    (1ull << MLX5_EVENT_TYPE_FPGA_QP_ERROR);
+	if (MLX5_CAP_GEN_MAX(dev, dct))
+		async_event_mask |= (1ull << MLX5_EVENT_TYPE_DCT_DRAINED);
+
+	if (MLX5_CAP_GEN(dev, temp_warn_event))
+		async_event_mask |= (1ull << MLX5_EVENT_TYPE_TEMP_WARN_EVENT);
+
+	if (MLX5_CAP_MCAM_REG(dev, tracer_registers))
+		async_event_mask |= (1ull << MLX5_EVENT_TYPE_DEVICE_TRACER);
+
+	if (MLX5_CAP_GEN(dev, max_num_of_monitor_counters))
+		async_event_mask |= (1ull << MLX5_EVENT_TYPE_MONITOR_COUNTER);
+
+	if (mlx5_eswitch_is_funcs_handler(dev))
+		async_event_mask |=
+			(1ull << MLX5_EVENT_TYPE_ESW_FUNCTIONS_CHANGED);
+
+	if (MLX5_CAP_GEN_MAX(dev, vhca_state))
+		async_event_mask |= (1ull << MLX5_EVENT_TYPE_VHCA_STATE_CHANGE);
+
+	if (MLX5_CAP_MACSEC(dev, log_max_macsec_offload))
+		async_event_mask |= (1ull << MLX5_EVENT_TYPE_OBJECT_CHANGE);
+
+	mask[0] = async_event_mask;
+
+	if (MLX5_CAP_GEN(dev, event_cap))
+		gather_user_async_events(dev, mask);
+}
+
+static int
+setup_async_eq(struct mlx5_core_dev *dev, struct mlx5_eq_async *eq,
+	       struct mlx5_eq_param *param, const char *name)
+{
+	int err;
+
+	eq->irq_nb.notifier_call = mlx5_eq_async_int;
+	spin_lock_init(&eq->lock);
+
+	err = create_async_eq(dev, &eq->core, param);
+	if (err) {
+		mlx5_core_warn(dev, "failed to create %s EQ %d\n", name, err);
+		return err;
+	}
+	err = mlx5_eq_enable(dev, &eq->core, &eq->irq_nb);
+	if (err) {
+		mlx5_core_warn(dev, "failed to enable %s EQ %d\n", name, err);
+		destroy_async_eq(dev, &eq->core);
+	}
+	return err;
+}
+
+static void cleanup_async_eq(struct mlx5_core_dev *dev,
+			     struct mlx5_eq_async *eq, const char *name)
+{
+	int err;
+
+	mlx5_eq_disable(dev, &eq->core, &eq->irq_nb);
+	err = destroy_async_eq(dev, &eq->core);
+	if (err)
+		mlx5_core_err(dev, "failed to destroy %s eq, err(%d)\n",
+			      name, err);
+}
+
+static u16 async_eq_depth_devlink_param_get(struct mlx5_core_dev *dev)
+{
+	struct devlink *devlink = priv_to_devlink(dev);
+	union devlink_param_value val;
+	int err;
+
+	err = devlink_param_driverinit_value_get(devlink,
+						 DEVLINK_PARAM_GENERIC_ID_EVENT_EQ_SIZE,
+						 &val);
+	if (!err)
+		return val.vu32;
+	mlx5_core_dbg(dev, "Failed to get param. using default. err = %d\n", err);
+	return MLX5_NUM_ASYNC_EQE;
+}
+static int create_async_eqs(struct mlx5_core_dev *dev)
+{
+	struct mlx5_eq_table *table = dev->priv.eq_table;
+	struct mlx5_eq_param param = {};
+	int err;
+
+	/* All the async_eqs are using single IRQ, request one IRQ and share its
+	 * index among all the async_eqs of this device.
+	 */
+	table->ctrl_irq = mlx5_ctrl_irq_request(dev);
+	if (IS_ERR(table->ctrl_irq))
+		return PTR_ERR(table->ctrl_irq);
+
+	MLX5_NB_INIT(&table->cq_err_nb, cq_err_event_notifier, CQ_ERROR);
+	mlx5_eq_notifier_register(dev, &table->cq_err_nb);
+
+	param = (struct mlx5_eq_param) {
+		.irq = table->ctrl_irq,
+		.nent = MLX5_NUM_CMD_EQE,
+		.mask[0] = 1ull << MLX5_EVENT_TYPE_CMD,
+	};
+	mlx5_cmd_allowed_opcode(dev, MLX5_CMD_OP_CREATE_EQ);
+	err = setup_async_eq(dev, &table->cmd_eq, &param, "cmd");
+	if (err)
+		goto err1;
+
+	mlx5_cmd_use_events(dev);
+	mlx5_cmd_allowed_opcode(dev, CMD_ALLOWED_OPCODE_ALL);
+
+	param = (struct mlx5_eq_param) {
+		.irq = table->ctrl_irq,
+		.nent = async_eq_depth_devlink_param_get(dev),
+	};
+
+	gather_async_events_mask(dev, param.mask);
+	err = setup_async_eq(dev, &table->async_eq, &param, "async");
+	if (err)
+		goto err2;
+
+	param = (struct mlx5_eq_param) {
+		.irq = table->ctrl_irq,
+		.nent = /* TODO: sriov max_vf + */ 1,
+		.mask[0] = 1ull << MLX5_EVENT_TYPE_PAGE_REQUEST,
+	};
+
+	err = setup_async_eq(dev, &table->pages_eq, &param, "pages");
+	if (err)
+		goto err3;
+
+	return 0;
+
+err3:
+	cleanup_async_eq(dev, &table->async_eq, "async");
+err2:
+	mlx5_cmd_use_polling(dev);
+	cleanup_async_eq(dev, &table->cmd_eq, "cmd");
+err1:
+	mlx5_cmd_allowed_opcode(dev, CMD_ALLOWED_OPCODE_ALL);
+	mlx5_eq_notifier_unregister(dev, &table->cq_err_nb);
+	mlx5_ctrl_irq_release(table->ctrl_irq);
+	return err;
+}
+
+static void destroy_async_eqs(struct mlx5_core_dev *dev)
+{
+	struct mlx5_eq_table *table = dev->priv.eq_table;
+
+	cleanup_async_eq(dev, &table->pages_eq, "pages");
+	cleanup_async_eq(dev, &table->async_eq, "async");
+	mlx5_cmd_allowed_opcode(dev, MLX5_CMD_OP_DESTROY_EQ);
+	mlx5_cmd_use_polling(dev);
+	cleanup_async_eq(dev, &table->cmd_eq, "cmd");
+	mlx5_cmd_allowed_opcode(dev, CMD_ALLOWED_OPCODE_ALL);
+	mlx5_eq_notifier_unregister(dev, &table->cq_err_nb);
+	mlx5_ctrl_irq_release(table->ctrl_irq);
+}
+
+struct mlx5_eq *mlx5_get_async_eq(struct mlx5_core_dev *dev)
+{
+	return &dev->priv.eq_table->async_eq.core;
+}
+
+void mlx5_eq_synchronize_async_irq(struct mlx5_core_dev *dev)
+{
+	synchronize_irq(dev->priv.eq_table->async_eq.core.irqn);
+}
+
+void mlx5_eq_synchronize_cmd_irq(struct mlx5_core_dev *dev)
+{
+	synchronize_irq(dev->priv.eq_table->cmd_eq.core.irqn);
+}
+
+/* Generic EQ API for mlx5_core consumers
+ * Needed For RDMA ODP EQ for now
+ */
+struct mlx5_eq *
+mlx5_eq_create_generic(struct mlx5_core_dev *dev,
+		       struct mlx5_eq_param *param)
+{
+	struct mlx5_eq *eq = kvzalloc_node(sizeof(*eq), GFP_KERNEL,
+					   dev->priv.numa_node);
+	int err;
+
+	if (!eq)
+		return ERR_PTR(-ENOMEM);
+
+	param->irq = dev->priv.eq_table->ctrl_irq;
+	err = create_async_eq(dev, eq, param);
+	if (err) {
+		kvfree(eq);
+		eq = ERR_PTR(err);
+	}
+
+	return eq;
+}
+EXPORT_SYMBOL(mlx5_eq_create_generic);
+
+int mlx5_eq_destroy_generic(struct mlx5_core_dev *dev, struct mlx5_eq *eq)
+{
+	int err;
+
+	if (IS_ERR(eq))
+		return -EINVAL;
+
+	err = destroy_async_eq(dev, eq);
+	if (err)
+		goto out;
+
+	kvfree(eq);
+out:
+	return err;
+}
+EXPORT_SYMBOL(mlx5_eq_destroy_generic);
+
+struct mlx5_eqe *mlx5_eq_get_eqe(struct mlx5_eq *eq, u32 cc)
+{
+	u32 ci = eq->cons_index + cc;
+	u32 nent = eq_get_size(eq);
+	struct mlx5_eqe *eqe;
+
+	eqe = get_eqe(eq, ci & (nent - 1));
+	eqe = ((eqe->owner & 1) ^ !!(ci & nent)) ? NULL : eqe;
+	/* Make sure we read EQ entry contents after we've
+	 * checked the ownership bit.
+	 */
+	if (eqe)
+		dma_rmb();
+
+	return eqe;
+}
+EXPORT_SYMBOL(mlx5_eq_get_eqe);
+
+void mlx5_eq_update_ci(struct mlx5_eq *eq, u32 cc, bool arm)
+{
+	__be32 __iomem *addr = eq->doorbell + (arm ? 0 : 2);
+	u32 val;
+
+	eq->cons_index += cc;
+	val = (eq->cons_index & 0xffffff) | (eq->eqn << 24);
+
+	__raw_writel((__force u32)cpu_to_be32(val), addr);
+	/* We still want ordering, just not swabbing, so add a barrier */
+	wmb();
+}
+EXPORT_SYMBOL(mlx5_eq_update_ci);
+
+static void comp_irqs_release(struct mlx5_core_dev *dev)
+{
+	struct mlx5_eq_table *table = dev->priv.eq_table;
+
+	if (mlx5_core_is_sf(dev))
+		mlx5_irq_affinity_irqs_release(dev, table->comp_irqs, table->num_comp_eqs);
+	else
+		mlx5_irqs_release_vectors(table->comp_irqs, table->num_comp_eqs);
+	kfree(table->comp_irqs);
+}
+
+static int comp_irqs_request(struct mlx5_core_dev *dev)
+{
+	struct mlx5_eq_table *table = dev->priv.eq_table;
+	int ncomp_eqs = table->num_comp_eqs;
+	u16 *cpus;
+	int ret;
+	int i;
+
+	ncomp_eqs = table->num_comp_eqs;
+	table->comp_irqs = kcalloc(ncomp_eqs, sizeof(*table->comp_irqs), GFP_KERNEL);
+	if (!table->comp_irqs)
+		return -ENOMEM;
+	if (mlx5_core_is_sf(dev)) {
+		ret = mlx5_irq_affinity_irqs_request_auto(dev, ncomp_eqs, table->comp_irqs);
+		if (ret < 0)
+			goto free_irqs;
+		return ret;
+	}
+
+	cpus = kcalloc(ncomp_eqs, sizeof(*cpus), GFP_KERNEL);
+	if (!cpus) {
+		ret = -ENOMEM;
+		goto free_irqs;
+	}
+	for (i = 0; i < ncomp_eqs; i++)
+		cpus[i] = cpumask_local_spread(i, dev->priv.numa_node);
+	ret = mlx5_irqs_request_vectors(dev, cpus, ncomp_eqs, table->comp_irqs);
+	kfree(cpus);
+	if (ret < 0)
+		goto free_irqs;
+	return ret;
+
+free_irqs:
+	kfree(table->comp_irqs);
+	return ret;
+}
+
+static void destroy_comp_eqs(struct mlx5_core_dev *dev)
+{
+	struct mlx5_eq_table *table = dev->priv.eq_table;
+	struct mlx5_eq_comp *eq, *n;
+
+	list_for_each_entry_safe(eq, n, &table->comp_eqs_list, list) {
+		list_del(&eq->list);
+		mlx5_eq_disable(dev, &eq->core, &eq->irq_nb);
+		if (destroy_unmap_eq(dev, &eq->core))
+			mlx5_core_warn(dev, "failed to destroy comp EQ 0x%x\n",
+				       eq->core.eqn);
+		tasklet_disable(&eq->tasklet_ctx.task);
+		kfree(eq);
+	}
+	comp_irqs_release(dev);
+}
+
+static u16 comp_eq_depth_devlink_param_get(struct mlx5_core_dev *dev)
+{
+	struct devlink *devlink = priv_to_devlink(dev);
+	union devlink_param_value val;
+	int err;
+
+	err = devlink_param_driverinit_value_get(devlink,
+						 DEVLINK_PARAM_GENERIC_ID_IO_EQ_SIZE,
+						 &val);
+	if (!err)
+		return val.vu32;
+	mlx5_core_dbg(dev, "Failed to get param. using default. err = %d\n", err);
+	return MLX5_COMP_EQ_SIZE;
+}
+
+static int create_comp_eqs(struct mlx5_core_dev *dev)
+{
+	struct mlx5_eq_table *table = dev->priv.eq_table;
+	struct mlx5_eq_comp *eq;
+	int ncomp_eqs;
+	int nent;
+	int err;
+	int i;
+
+	ncomp_eqs = comp_irqs_request(dev);
+	if (ncomp_eqs < 0)
+		return ncomp_eqs;
+	INIT_LIST_HEAD(&table->comp_eqs_list);
+	nent = comp_eq_depth_devlink_param_get(dev);
+
+	for (i = 0; i < ncomp_eqs; i++) {
+		struct mlx5_eq_param param = {};
+
+		eq = kzalloc_node(sizeof(*eq), GFP_KERNEL, dev->priv.numa_node);
+		if (!eq) {
+			err = -ENOMEM;
+			goto clean;
+		}
+
+		INIT_LIST_HEAD(&eq->tasklet_ctx.list);
+		INIT_LIST_HEAD(&eq->tasklet_ctx.process_list);
+		spin_lock_init(&eq->tasklet_ctx.lock);
+		tasklet_setup(&eq->tasklet_ctx.task, mlx5_cq_tasklet_cb);
+
+		eq->irq_nb.notifier_call = mlx5_eq_comp_int;
+		param = (struct mlx5_eq_param) {
+			.irq = table->comp_irqs[i],
+			.nent = nent,
+		};
+
+		err = create_map_eq(dev, &eq->core, &param);
+		if (err)
+			goto clean_eq;
+		err = mlx5_eq_enable(dev, &eq->core, &eq->irq_nb);
+		if (err) {
+			destroy_unmap_eq(dev, &eq->core);
+			goto clean_eq;
+		}
+
+		mlx5_core_dbg(dev, "allocated completion EQN %d\n", eq->core.eqn);
+		/* add tail, to keep the list ordered, for mlx5_vector2eqn to work */
+		list_add_tail(&eq->list, &table->comp_eqs_list);
+	}
+
+	table->num_comp_eqs = ncomp_eqs;
+	return 0;
+
+clean_eq:
+	kfree(eq);
+clean:
+	destroy_comp_eqs(dev);
+	return err;
+}
+
+static int vector2eqnirqn(struct mlx5_core_dev *dev, int vector, int *eqn,
+			  unsigned int *irqn)
+{
+	struct mlx5_eq_table *table = dev->priv.eq_table;
+	struct mlx5_eq_comp *eq, *n;
+	int err = -ENOENT;
+	int i = 0;
+
+	list_for_each_entry_safe(eq, n, &table->comp_eqs_list, list) {
+		if (i++ == vector) {
+			if (irqn)
+				*irqn = eq->core.irqn;
+			if (eqn)
+				*eqn = eq->core.eqn;
+			err = 0;
+			break;
+		}
+	}
+
+	return err;
+}
+
+int mlx5_vector2eqn(struct mlx5_core_dev *dev, int vector, int *eqn)
+{
+	return vector2eqnirqn(dev, vector, eqn, NULL);
+}
+EXPORT_SYMBOL(mlx5_vector2eqn);
+
+int mlx5_vector2irqn(struct mlx5_core_dev *dev, int vector, unsigned int *irqn)
+{
+	return vector2eqnirqn(dev, vector, NULL, irqn);
+}
+
+unsigned int mlx5_comp_vectors_count(struct mlx5_core_dev *dev)
+{
+	return dev->priv.eq_table->num_comp_eqs;
+}
+EXPORT_SYMBOL(mlx5_comp_vectors_count);
+
+struct cpumask *
+mlx5_comp_irq_get_affinity_mask(struct mlx5_core_dev *dev, int vector)
+{
+	struct mlx5_eq_table *table = dev->priv.eq_table;
+	struct mlx5_eq_comp *eq, *n;
+	int i = 0;
+
+	list_for_each_entry_safe(eq, n, &table->comp_eqs_list, list) {
+		if (i++ == vector)
+			break;
+	}
+
+	return mlx5_irq_get_affinity_mask(eq->core.irq);
+}
+EXPORT_SYMBOL(mlx5_comp_irq_get_affinity_mask);
+
+#ifdef CONFIG_RFS_ACCEL
+struct cpu_rmap *mlx5_eq_table_get_rmap(struct mlx5_core_dev *dev)
+{
+	return dev->priv.eq_table->rmap;
+}
+#endif
+
+struct mlx5_eq_comp *mlx5_eqn2comp_eq(struct mlx5_core_dev *dev, int eqn)
+{
+	struct mlx5_eq_table *table = dev->priv.eq_table;
+	struct mlx5_eq_comp *eq;
+
+	list_for_each_entry(eq, &table->comp_eqs_list, list) {
+		if (eq->core.eqn == eqn)
+			return eq;
+	}
+
+	return ERR_PTR(-ENOENT);
+}
+
+static void clear_rmap(struct mlx5_core_dev *dev)
+{
+#ifdef CONFIG_RFS_ACCEL
+	struct mlx5_eq_table *eq_table = dev->priv.eq_table;
+
+	free_irq_cpu_rmap(eq_table->rmap);
+#endif
+}
+
+static int set_rmap(struct mlx5_core_dev *mdev)
+{
+	int err = 0;
+#ifdef CONFIG_RFS_ACCEL
+	struct mlx5_eq_table *eq_table = mdev->priv.eq_table;
+	int vecidx;
+
+	eq_table->rmap = alloc_irq_cpu_rmap(eq_table->num_comp_eqs);
+	if (!eq_table->rmap) {
+		err = -ENOMEM;
+		mlx5_core_err(mdev, "Failed to allocate cpu_rmap. err %d", err);
+		goto err_out;
+	}
+
+	for (vecidx = 0; vecidx < eq_table->num_comp_eqs; vecidx++) {
+		err = irq_cpu_rmap_add(eq_table->rmap,
+				       pci_irq_vector(mdev->pdev, vecidx));
+		if (err) {
+			mlx5_core_err(mdev, "irq_cpu_rmap_add failed. err %d",
+				      err);
+			goto err_irq_cpu_rmap_add;
+		}
+	}
+	return 0;
+
+err_irq_cpu_rmap_add:
+	clear_rmap(mdev);
+err_out:
+#endif
+	return err;
+}
+
+/* This function should only be called after mlx5_cmd_force_teardown_hca */
+void mlx5_core_eq_free_irqs(struct mlx5_core_dev *dev)
+{
+	struct mlx5_eq_table *table = dev->priv.eq_table;
+
+	mutex_lock(&table->lock); /* sync with create/destroy_async_eq */
+	if (!mlx5_core_is_sf(dev))
+		clear_rmap(dev);
+	mlx5_irq_table_free_irqs(dev);
+	mutex_unlock(&table->lock);
+}
+
+#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
+#define MLX5_MAX_ASYNC_EQS 4
+#else
+#define MLX5_MAX_ASYNC_EQS 3
+#endif
+
+int mlx5_eq_table_create(struct mlx5_core_dev *dev)
+{
+	struct mlx5_eq_table *eq_table = dev->priv.eq_table;
+	int num_eqs = MLX5_CAP_GEN(dev, max_num_eqs) ?
+		      MLX5_CAP_GEN(dev, max_num_eqs) :
+		      1 << MLX5_CAP_GEN(dev, log_max_eq);
+	int max_eqs_sf;
+	int err;
+
+	eq_table->num_comp_eqs =
+		min_t(int,
+		      mlx5_irq_table_get_num_comp(eq_table->irq_table),
+		      num_eqs - MLX5_MAX_ASYNC_EQS);
+	if (mlx5_core_is_sf(dev)) {
+		max_eqs_sf = min_t(int, MLX5_COMP_EQS_PER_SF,
+				   mlx5_irq_table_get_sfs_vec(eq_table->irq_table));
+		eq_table->num_comp_eqs = min_t(int, eq_table->num_comp_eqs,
+					       max_eqs_sf);
+	}
+
+	err = create_async_eqs(dev);
+	if (err) {
+		mlx5_core_err(dev, "Failed to create async EQs\n");
+		goto err_async_eqs;
+	}
+
+	if (!mlx5_core_is_sf(dev)) {
+		/* rmap is a mapping between irq number and queue number.
+		 * each irq can be assign only to a single rmap.
+		 * since SFs share IRQs, rmap mapping cannot function correctly
+		 * for irqs that are shared for different core/netdev RX rings.
+		 * Hence we don't allow netdev rmap for SFs
+		 */
+		err = set_rmap(dev);
+		if (err)
+			goto err_rmap;
+	}
+
+	err = create_comp_eqs(dev);
+	if (err) {
+		mlx5_core_err(dev, "Failed to create completion EQs\n");
+		goto err_comp_eqs;
+	}
+
+	return 0;
+err_comp_eqs:
+	if (!mlx5_core_is_sf(dev))
+		clear_rmap(dev);
+err_rmap:
+	destroy_async_eqs(dev);
+err_async_eqs:
+	return err;
+}
+
+void mlx5_eq_table_destroy(struct mlx5_core_dev *dev)
+{
+	if (!mlx5_core_is_sf(dev))
+		clear_rmap(dev);
+	destroy_comp_eqs(dev);
+	destroy_async_eqs(dev);
+}
+
+int mlx5_eq_notifier_register(struct mlx5_core_dev *dev, struct mlx5_nb *nb)
+{
+	struct mlx5_eq_table *eqt = dev->priv.eq_table;
+
+	return atomic_notifier_chain_register(&eqt->nh[nb->event_type], &nb->nb);
+}
+EXPORT_SYMBOL(mlx5_eq_notifier_register);
+
+int mlx5_eq_notifier_unregister(struct mlx5_core_dev *dev, struct mlx5_nb *nb)
+{
+	struct mlx5_eq_table *eqt = dev->priv.eq_table;
+
+	return atomic_notifier_chain_unregister(&eqt->nh[nb->event_type], &nb->nb);
+}
+EXPORT_SYMBOL(mlx5_eq_notifier_unregister);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/esw/Makefile b/drivers/net/ethernet/mellanox/mlx5/core/esw/Makefile
new file mode 100644
index 000000000..c78512eed
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/esw/Makefile
@@ -0,0 +1,2 @@
+# SPDX-License-Identifier: GPL-2.0-only
+subdir-ccflags-y += -I$(src)/..
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/esw/acl/egress_lgcy.c b/drivers/net/ethernet/mellanox/mlx5/core/esw/acl/egress_lgcy.c
new file mode 100644
index 000000000..6b4c9ffad
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/esw/acl/egress_lgcy.c
@@ -0,0 +1,178 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2020 Mellanox Technologies Inc. All rights reserved. */
+
+#include "mlx5_core.h"
+#include "eswitch.h"
+#include "helper.h"
+#include "lgcy.h"
+
+static void esw_acl_egress_lgcy_rules_destroy(struct mlx5_vport *vport)
+{
+	esw_acl_egress_vlan_destroy(vport);
+	if (!IS_ERR_OR_NULL(vport->egress.legacy.drop_rule)) {
+		mlx5_del_flow_rules(vport->egress.legacy.drop_rule);
+		vport->egress.legacy.drop_rule = NULL;
+	}
+}
+
+static int esw_acl_egress_lgcy_groups_create(struct mlx5_eswitch *esw,
+					     struct mlx5_vport *vport)
+{
+	int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
+	struct mlx5_core_dev *dev = esw->dev;
+	struct mlx5_flow_group *drop_grp;
+	u32 *flow_group_in;
+	int err = 0;
+
+	err = esw_acl_egress_vlan_grp_create(esw, vport);
+	if (err)
+		return err;
+
+	flow_group_in = kvzalloc(inlen, GFP_KERNEL);
+	if (!flow_group_in) {
+		err = -ENOMEM;
+		goto alloc_err;
+	}
+
+	MLX5_SET(create_flow_group_in, flow_group_in, start_flow_index, 1);
+	MLX5_SET(create_flow_group_in, flow_group_in, end_flow_index, 1);
+	drop_grp = mlx5_create_flow_group(vport->egress.acl, flow_group_in);
+	if (IS_ERR(drop_grp)) {
+		err = PTR_ERR(drop_grp);
+		esw_warn(dev, "Failed to create E-Switch vport[%d] egress drop flow group, err(%d)\n",
+			 vport->vport, err);
+		goto drop_grp_err;
+	}
+
+	vport->egress.legacy.drop_grp = drop_grp;
+	kvfree(flow_group_in);
+	return 0;
+
+drop_grp_err:
+	kvfree(flow_group_in);
+alloc_err:
+	esw_acl_egress_vlan_grp_destroy(vport);
+	return err;
+}
+
+static void esw_acl_egress_lgcy_groups_destroy(struct mlx5_vport *vport)
+{
+	if (!IS_ERR_OR_NULL(vport->egress.legacy.drop_grp)) {
+		mlx5_destroy_flow_group(vport->egress.legacy.drop_grp);
+		vport->egress.legacy.drop_grp = NULL;
+	}
+	esw_acl_egress_vlan_grp_destroy(vport);
+}
+
+int esw_acl_egress_lgcy_setup(struct mlx5_eswitch *esw,
+			      struct mlx5_vport *vport)
+{
+	bool vst_mode_steering = esw_vst_mode_is_steering(esw);
+	struct mlx5_flow_destination drop_ctr_dst = {};
+	struct mlx5_flow_destination *dst = NULL;
+	struct mlx5_fc *drop_counter = NULL;
+	struct mlx5_flow_act flow_act = {};
+	/* The egress acl table contains 2 rules:
+	 * 1)Allow traffic with vlan_tag=vst_vlan_id
+	 * 2)Drop all other traffic.
+	 */
+	int table_size = 2;
+	int dest_num = 0;
+	int actions_flag;
+	int err = 0;
+
+	if (vport->egress.legacy.drop_counter) {
+		drop_counter = vport->egress.legacy.drop_counter;
+	} else if (MLX5_CAP_ESW_EGRESS_ACL(esw->dev, flow_counter)) {
+		drop_counter = mlx5_fc_create(esw->dev, false);
+		if (IS_ERR(drop_counter)) {
+			esw_warn(esw->dev,
+				 "vport[%d] configure egress drop rule counter err(%ld)\n",
+				 vport->vport, PTR_ERR(drop_counter));
+			drop_counter = NULL;
+		}
+		vport->egress.legacy.drop_counter = drop_counter;
+	}
+
+	esw_acl_egress_lgcy_rules_destroy(vport);
+
+	if (!vport->info.vlan && !vport->info.qos) {
+		esw_acl_egress_lgcy_cleanup(esw, vport);
+		return 0;
+	}
+
+	if (!vport->egress.acl) {
+		vport->egress.acl = esw_acl_table_create(esw, vport,
+							 MLX5_FLOW_NAMESPACE_ESW_EGRESS,
+							 table_size);
+		if (IS_ERR(vport->egress.acl)) {
+			err = PTR_ERR(vport->egress.acl);
+			vport->egress.acl = NULL;
+			goto out;
+		}
+
+		err = esw_acl_egress_lgcy_groups_create(esw, vport);
+		if (err)
+			goto out;
+	}
+
+	esw_debug(esw->dev,
+		  "vport[%d] configure egress rules, vlan(%d) qos(%d)\n",
+		  vport->vport, vport->info.vlan, vport->info.qos);
+
+	/* Allowed vlan rule */
+	actions_flag = MLX5_FLOW_CONTEXT_ACTION_ALLOW;
+	if (vst_mode_steering)
+		actions_flag |= MLX5_FLOW_CONTEXT_ACTION_VLAN_POP;
+	err = esw_egress_acl_vlan_create(esw, vport, NULL, vport->info.vlan,
+					 actions_flag);
+	if (err)
+		goto out;
+
+	flow_act.action = MLX5_FLOW_CONTEXT_ACTION_DROP;
+
+	/* Attach egress drop flow counter */
+	if (drop_counter) {
+		flow_act.action |= MLX5_FLOW_CONTEXT_ACTION_COUNT;
+		drop_ctr_dst.type = MLX5_FLOW_DESTINATION_TYPE_COUNTER;
+		drop_ctr_dst.counter_id = mlx5_fc_id(drop_counter);
+		dst = &drop_ctr_dst;
+		dest_num++;
+	}
+	vport->egress.legacy.drop_rule =
+		mlx5_add_flow_rules(vport->egress.acl, NULL,
+				    &flow_act, dst, dest_num);
+	if (IS_ERR(vport->egress.legacy.drop_rule)) {
+		err = PTR_ERR(vport->egress.legacy.drop_rule);
+		esw_warn(esw->dev,
+			 "vport[%d] configure egress drop rule failed, err(%d)\n",
+			 vport->vport, err);
+		vport->egress.legacy.drop_rule = NULL;
+		goto out;
+	}
+
+	return err;
+
+out:
+	esw_acl_egress_lgcy_cleanup(esw, vport);
+	return err;
+}
+
+void esw_acl_egress_lgcy_cleanup(struct mlx5_eswitch *esw,
+				 struct mlx5_vport *vport)
+{
+	if (IS_ERR_OR_NULL(vport->egress.acl))
+		goto clean_drop_counter;
+
+	esw_debug(esw->dev, "Destroy vport[%d] E-Switch egress ACL\n", vport->vport);
+
+	esw_acl_egress_lgcy_rules_destroy(vport);
+	esw_acl_egress_lgcy_groups_destroy(vport);
+	esw_acl_egress_table_destroy(vport);
+
+clean_drop_counter:
+	if (vport->egress.legacy.drop_counter) {
+		mlx5_fc_destroy(esw->dev, vport->egress.legacy.drop_counter);
+		vport->egress.legacy.drop_counter = NULL;
+	}
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/esw/acl/egress_ofld.c b/drivers/net/ethernet/mellanox/mlx5/core/esw/acl/egress_ofld.c
new file mode 100644
index 000000000..2e504c746
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/esw/acl/egress_ofld.c
@@ -0,0 +1,259 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2020 Mellanox Technologies Inc. All rights reserved. */
+
+#include "mlx5_core.h"
+#include "eswitch.h"
+#include "helper.h"
+#include "ofld.h"
+
+static void esw_acl_egress_ofld_fwd2vport_destroy(struct mlx5_vport *vport)
+{
+	if (!vport->egress.offloads.fwd_rule)
+		return;
+
+	mlx5_del_flow_rules(vport->egress.offloads.fwd_rule);
+	vport->egress.offloads.fwd_rule = NULL;
+}
+
+static void esw_acl_egress_ofld_bounce_rule_destroy(struct mlx5_vport *vport)
+{
+	if (!vport->egress.offloads.bounce_rule)
+		return;
+
+	mlx5_del_flow_rules(vport->egress.offloads.bounce_rule);
+	vport->egress.offloads.bounce_rule = NULL;
+}
+
+static int esw_acl_egress_ofld_fwd2vport_create(struct mlx5_eswitch *esw,
+						struct mlx5_vport *vport,
+						struct mlx5_flow_destination *fwd_dest)
+{
+	struct mlx5_flow_act flow_act = {};
+	int err = 0;
+
+	esw_debug(esw->dev, "vport(%d) configure egress acl rule fwd2vport(%d)\n",
+		  vport->vport, fwd_dest->vport.num);
+
+	/* Delete the old egress forward-to-vport rule if any */
+	esw_acl_egress_ofld_fwd2vport_destroy(vport);
+
+	flow_act.action = MLX5_FLOW_CONTEXT_ACTION_FWD_DEST;
+
+	vport->egress.offloads.fwd_rule =
+		mlx5_add_flow_rules(vport->egress.acl, NULL,
+				    &flow_act, fwd_dest, 1);
+	if (IS_ERR(vport->egress.offloads.fwd_rule)) {
+		err = PTR_ERR(vport->egress.offloads.fwd_rule);
+		esw_warn(esw->dev,
+			 "vport(%d) failed to add fwd2vport acl rule err(%d)\n",
+			 vport->vport, err);
+		vport->egress.offloads.fwd_rule = NULL;
+	}
+
+	return err;
+}
+
+static int esw_acl_egress_ofld_rules_create(struct mlx5_eswitch *esw,
+					    struct mlx5_vport *vport,
+					    struct mlx5_flow_destination *fwd_dest)
+{
+	int err = 0;
+	int action;
+
+	if (MLX5_CAP_GEN(esw->dev, prio_tag_required)) {
+		/* For prio tag mode, there is only 1 FTEs:
+		 * 1) prio tag packets - pop the prio tag VLAN, allow
+		 * Unmatched traffic is allowed by default
+		 */
+		esw_debug(esw->dev,
+			  "vport[%d] configure prio tag egress rules\n", vport->vport);
+
+		action = MLX5_FLOW_CONTEXT_ACTION_VLAN_POP;
+		action |= fwd_dest ? MLX5_FLOW_CONTEXT_ACTION_FWD_DEST :
+			  MLX5_FLOW_CONTEXT_ACTION_ALLOW;
+
+		/* prio tag vlan rule - pop it so vport receives untagged packets */
+		err = esw_egress_acl_vlan_create(esw, vport, fwd_dest, 0, action);
+		if (err)
+			goto prio_err;
+	}
+
+	if (fwd_dest) {
+		err = esw_acl_egress_ofld_fwd2vport_create(esw, vport, fwd_dest);
+		if (err)
+			goto fwd_err;
+	}
+
+	return 0;
+
+fwd_err:
+	esw_acl_egress_vlan_destroy(vport);
+prio_err:
+	return err;
+}
+
+static void esw_acl_egress_ofld_rules_destroy(struct mlx5_vport *vport)
+{
+	esw_acl_egress_vlan_destroy(vport);
+	esw_acl_egress_ofld_fwd2vport_destroy(vport);
+	esw_acl_egress_ofld_bounce_rule_destroy(vport);
+}
+
+static int esw_acl_egress_ofld_groups_create(struct mlx5_eswitch *esw,
+					     struct mlx5_vport *vport)
+{
+	int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
+	struct mlx5_flow_group *fwd_grp;
+	u32 *flow_group_in;
+	u32 flow_index = 0;
+	int ret = 0;
+
+	if (MLX5_CAP_GEN(esw->dev, prio_tag_required)) {
+		ret = esw_acl_egress_vlan_grp_create(esw, vport);
+		if (ret)
+			return ret;
+
+		flow_index++;
+	}
+
+	if (!mlx5_esw_acl_egress_fwd2vport_supported(esw))
+		goto out;
+
+	flow_group_in = kvzalloc(inlen, GFP_KERNEL);
+	if (!flow_group_in) {
+		ret = -ENOMEM;
+		goto fwd_grp_err;
+	}
+
+	/* This group holds 1 FTE to forward all packets to other vport
+	 * when bond vports is supported.
+	 */
+	MLX5_SET(create_flow_group_in, flow_group_in, start_flow_index, flow_index);
+	MLX5_SET(create_flow_group_in, flow_group_in, end_flow_index, flow_index);
+	fwd_grp = mlx5_create_flow_group(vport->egress.acl, flow_group_in);
+	if (IS_ERR(fwd_grp)) {
+		ret = PTR_ERR(fwd_grp);
+		esw_warn(esw->dev,
+			 "Failed to create vport[%d] egress fwd2vport flow group, err(%d)\n",
+			 vport->vport, ret);
+		kvfree(flow_group_in);
+		goto fwd_grp_err;
+	}
+	vport->egress.offloads.fwd_grp = fwd_grp;
+	kvfree(flow_group_in);
+	return 0;
+
+fwd_grp_err:
+	esw_acl_egress_vlan_grp_destroy(vport);
+out:
+	return ret;
+}
+
+static void esw_acl_egress_ofld_groups_destroy(struct mlx5_vport *vport)
+{
+	if (!IS_ERR_OR_NULL(vport->egress.offloads.fwd_grp)) {
+		mlx5_destroy_flow_group(vport->egress.offloads.fwd_grp);
+		vport->egress.offloads.fwd_grp = NULL;
+	}
+
+	if (!IS_ERR_OR_NULL(vport->egress.offloads.bounce_grp)) {
+		mlx5_destroy_flow_group(vport->egress.offloads.bounce_grp);
+		vport->egress.offloads.bounce_grp = NULL;
+	}
+
+	esw_acl_egress_vlan_grp_destroy(vport);
+}
+
+static bool esw_acl_egress_needed(struct mlx5_eswitch *esw, u16 vport_num)
+{
+	return mlx5_eswitch_is_vf_vport(esw, vport_num) || mlx5_esw_is_sf_vport(esw, vport_num);
+}
+
+int esw_acl_egress_ofld_setup(struct mlx5_eswitch *esw, struct mlx5_vport *vport)
+{
+	int table_size = 0;
+	int err;
+
+	if (!mlx5_esw_acl_egress_fwd2vport_supported(esw) &&
+	    !MLX5_CAP_GEN(esw->dev, prio_tag_required))
+		return 0;
+
+	if (!esw_acl_egress_needed(esw, vport->vport))
+		return 0;
+
+	esw_acl_egress_ofld_rules_destroy(vport);
+
+	if (mlx5_esw_acl_egress_fwd2vport_supported(esw))
+		table_size++;
+	if (MLX5_CAP_GEN(esw->dev, prio_tag_required))
+		table_size++;
+	vport->egress.acl = esw_acl_table_create(esw, vport,
+						 MLX5_FLOW_NAMESPACE_ESW_EGRESS, table_size);
+	if (IS_ERR(vport->egress.acl)) {
+		err = PTR_ERR(vport->egress.acl);
+		vport->egress.acl = NULL;
+		return err;
+	}
+
+	err = esw_acl_egress_ofld_groups_create(esw, vport);
+	if (err)
+		goto group_err;
+
+	esw_debug(esw->dev, "vport[%d] configure egress rules\n", vport->vport);
+
+	err = esw_acl_egress_ofld_rules_create(esw, vport, NULL);
+	if (err)
+		goto rules_err;
+
+	return 0;
+
+rules_err:
+	esw_acl_egress_ofld_groups_destroy(vport);
+group_err:
+	esw_acl_egress_table_destroy(vport);
+	return err;
+}
+
+void esw_acl_egress_ofld_cleanup(struct mlx5_vport *vport)
+{
+	esw_acl_egress_ofld_rules_destroy(vport);
+	esw_acl_egress_ofld_groups_destroy(vport);
+	esw_acl_egress_table_destroy(vport);
+}
+
+int mlx5_esw_acl_egress_vport_bond(struct mlx5_eswitch *esw, u16 active_vport_num,
+				   u16 passive_vport_num)
+{
+	struct mlx5_vport *passive_vport = mlx5_eswitch_get_vport(esw, passive_vport_num);
+	struct mlx5_vport *active_vport = mlx5_eswitch_get_vport(esw, active_vport_num);
+	struct mlx5_flow_destination fwd_dest = {};
+
+	if (IS_ERR(active_vport))
+		return PTR_ERR(active_vport);
+	if (IS_ERR(passive_vport))
+		return PTR_ERR(passive_vport);
+
+	/* Cleanup and recreate rules WITHOUT fwd2vport of active vport */
+	esw_acl_egress_ofld_rules_destroy(active_vport);
+	esw_acl_egress_ofld_rules_create(esw, active_vport, NULL);
+
+	/* Cleanup and recreate all rules + fwd2vport rule of passive vport to forward */
+	esw_acl_egress_ofld_rules_destroy(passive_vport);
+	fwd_dest.type = MLX5_FLOW_DESTINATION_TYPE_VPORT;
+	fwd_dest.vport.num = active_vport_num;
+	fwd_dest.vport.vhca_id = MLX5_CAP_GEN(esw->dev, vhca_id);
+	fwd_dest.vport.flags = MLX5_FLOW_DEST_VPORT_VHCA_ID;
+
+	return esw_acl_egress_ofld_rules_create(esw, passive_vport, &fwd_dest);
+}
+
+int mlx5_esw_acl_egress_vport_unbond(struct mlx5_eswitch *esw, u16 vport_num)
+{
+	struct mlx5_vport *vport = mlx5_eswitch_get_vport(esw, vport_num);
+
+	if (IS_ERR(vport))
+		return PTR_ERR(vport);
+
+	esw_acl_egress_ofld_rules_destroy(vport);
+	return esw_acl_egress_ofld_rules_create(esw, vport, NULL);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/esw/acl/helper.c b/drivers/net/ethernet/mellanox/mlx5/core/esw/acl/helper.c
new file mode 100644
index 000000000..45b839116
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/esw/acl/helper.c
@@ -0,0 +1,163 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2020 Mellanox Technologies Inc. All rights reserved. */
+
+#include "mlx5_core.h"
+#include "eswitch.h"
+#include "helper.h"
+
+struct mlx5_flow_table *
+esw_acl_table_create(struct mlx5_eswitch *esw, struct mlx5_vport *vport, int ns, int size)
+{
+	struct mlx5_flow_table_attr ft_attr = {};
+	struct mlx5_core_dev *dev = esw->dev;
+	struct mlx5_flow_namespace *root_ns;
+	struct mlx5_flow_table *acl;
+	int acl_supported;
+	u16 vport_num;
+	int err;
+
+	acl_supported = (ns == MLX5_FLOW_NAMESPACE_ESW_INGRESS) ?
+			MLX5_CAP_ESW_INGRESS_ACL(dev, ft_support) :
+			MLX5_CAP_ESW_EGRESS_ACL(dev, ft_support);
+
+	if (!acl_supported)
+		return ERR_PTR(-EOPNOTSUPP);
+
+	vport_num = vport->vport;
+	esw_debug(dev, "Create vport[%d] %s ACL table\n", vport_num,
+		  ns == MLX5_FLOW_NAMESPACE_ESW_INGRESS ? "ingress" : "egress");
+
+	root_ns = mlx5_get_flow_vport_acl_namespace(dev, ns, vport->index);
+	if (!root_ns) {
+		esw_warn(dev, "Failed to get E-Switch root namespace for vport (%d)\n",
+			 vport_num);
+		return ERR_PTR(-EOPNOTSUPP);
+	}
+
+	ft_attr.max_fte = size;
+	ft_attr.flags = MLX5_FLOW_TABLE_OTHER_VPORT;
+	acl = mlx5_create_vport_flow_table(root_ns, &ft_attr, vport_num);
+	if (IS_ERR(acl)) {
+		err = PTR_ERR(acl);
+		esw_warn(dev, "vport[%d] create %s ACL table, err(%d)\n", vport_num,
+			 ns == MLX5_FLOW_NAMESPACE_ESW_INGRESS ? "ingress" : "egress", err);
+	}
+	return acl;
+}
+
+int esw_egress_acl_vlan_create(struct mlx5_eswitch *esw,
+			       struct mlx5_vport *vport,
+			       struct mlx5_flow_destination *fwd_dest,
+			       u16 vlan_id, u32 flow_action)
+{
+	struct mlx5_flow_act flow_act = {};
+	struct mlx5_flow_spec *spec;
+	int err = 0;
+
+	if (vport->egress.allowed_vlan)
+		return -EEXIST;
+
+	spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
+	if (!spec)
+		return -ENOMEM;
+
+	MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria, outer_headers.cvlan_tag);
+	MLX5_SET_TO_ONES(fte_match_param, spec->match_value, outer_headers.cvlan_tag);
+	MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria, outer_headers.first_vid);
+	MLX5_SET(fte_match_param, spec->match_value, outer_headers.first_vid, vlan_id);
+
+	spec->match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
+	flow_act.action = flow_action;
+	vport->egress.allowed_vlan =
+		mlx5_add_flow_rules(vport->egress.acl, spec,
+				    &flow_act, fwd_dest, 0);
+	if (IS_ERR(vport->egress.allowed_vlan)) {
+		err = PTR_ERR(vport->egress.allowed_vlan);
+		esw_warn(esw->dev,
+			 "vport[%d] configure egress vlan rule failed, err(%d)\n",
+			 vport->vport, err);
+		vport->egress.allowed_vlan = NULL;
+	}
+
+	kvfree(spec);
+	return err;
+}
+
+void esw_acl_egress_vlan_destroy(struct mlx5_vport *vport)
+{
+	if (!IS_ERR_OR_NULL(vport->egress.allowed_vlan)) {
+		mlx5_del_flow_rules(vport->egress.allowed_vlan);
+		vport->egress.allowed_vlan = NULL;
+	}
+}
+
+int esw_acl_egress_vlan_grp_create(struct mlx5_eswitch *esw, struct mlx5_vport *vport)
+{
+	int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
+	struct mlx5_flow_group *vlan_grp;
+	void *match_criteria;
+	u32 *flow_group_in;
+	int ret = 0;
+
+	flow_group_in = kvzalloc(inlen, GFP_KERNEL);
+	if (!flow_group_in)
+		return -ENOMEM;
+
+	MLX5_SET(create_flow_group_in, flow_group_in,
+		 match_criteria_enable, MLX5_MATCH_OUTER_HEADERS);
+	match_criteria = MLX5_ADDR_OF(create_flow_group_in,
+				      flow_group_in, match_criteria);
+	MLX5_SET_TO_ONES(fte_match_param, match_criteria, outer_headers.cvlan_tag);
+	MLX5_SET_TO_ONES(fte_match_param, match_criteria, outer_headers.first_vid);
+	MLX5_SET(create_flow_group_in, flow_group_in, start_flow_index, 0);
+	MLX5_SET(create_flow_group_in, flow_group_in, end_flow_index, 0);
+
+	vlan_grp = mlx5_create_flow_group(vport->egress.acl, flow_group_in);
+	if (IS_ERR(vlan_grp)) {
+		ret = PTR_ERR(vlan_grp);
+		esw_warn(esw->dev,
+			 "Failed to create E-Switch vport[%d] egress pop vlans flow group, err(%d)\n",
+			 vport->vport, ret);
+		goto out;
+	}
+	vport->egress.vlan_grp = vlan_grp;
+
+out:
+	kvfree(flow_group_in);
+	return ret;
+}
+
+void esw_acl_egress_vlan_grp_destroy(struct mlx5_vport *vport)
+{
+	if (!IS_ERR_OR_NULL(vport->egress.vlan_grp)) {
+		mlx5_destroy_flow_group(vport->egress.vlan_grp);
+		vport->egress.vlan_grp = NULL;
+	}
+}
+
+void esw_acl_egress_table_destroy(struct mlx5_vport *vport)
+{
+	if (IS_ERR_OR_NULL(vport->egress.acl))
+		return;
+
+	mlx5_destroy_flow_table(vport->egress.acl);
+	vport->egress.acl = NULL;
+}
+
+void esw_acl_ingress_table_destroy(struct mlx5_vport *vport)
+{
+	if (!vport->ingress.acl)
+		return;
+
+	mlx5_destroy_flow_table(vport->ingress.acl);
+	vport->ingress.acl = NULL;
+}
+
+void esw_acl_ingress_allow_rule_destroy(struct mlx5_vport *vport)
+{
+	if (!vport->ingress.allow_rule)
+		return;
+
+	mlx5_del_flow_rules(vport->ingress.allow_rule);
+	vport->ingress.allow_rule = NULL;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/esw/acl/helper.h b/drivers/net/ethernet/mellanox/mlx5/core/esw/acl/helper.h
new file mode 100644
index 000000000..a47063fab
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/esw/acl/helper.h
@@ -0,0 +1,26 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2020 Mellanox Technologies Inc. All rights reserved. */
+
+#ifndef __MLX5_ESWITCH_ACL_HELPER_H__
+#define __MLX5_ESWITCH_ACL_HELPER_H__
+
+#include "eswitch.h"
+
+/* General acl helper functions */
+struct mlx5_flow_table *
+esw_acl_table_create(struct mlx5_eswitch *esw, struct mlx5_vport *vport, int ns, int size);
+
+/* Egress acl helper functions */
+void esw_acl_egress_table_destroy(struct mlx5_vport *vport);
+int esw_egress_acl_vlan_create(struct mlx5_eswitch *esw, struct mlx5_vport *vport,
+			       struct mlx5_flow_destination *fwd_dest,
+			       u16 vlan_id, u32 flow_action);
+void esw_acl_egress_vlan_destroy(struct mlx5_vport *vport);
+int esw_acl_egress_vlan_grp_create(struct mlx5_eswitch *esw, struct mlx5_vport *vport);
+void esw_acl_egress_vlan_grp_destroy(struct mlx5_vport *vport);
+
+/* Ingress acl helper functions */
+void esw_acl_ingress_table_destroy(struct mlx5_vport *vport);
+void esw_acl_ingress_allow_rule_destroy(struct mlx5_vport *vport);
+
+#endif /* __MLX5_ESWITCH_ACL_HELPER_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/esw/acl/ingress_lgcy.c b/drivers/net/ethernet/mellanox/mlx5/core/esw/acl/ingress_lgcy.c
new file mode 100644
index 000000000..093ed86a0
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/esw/acl/ingress_lgcy.c
@@ -0,0 +1,305 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2020 Mellanox Technologies Inc. All rights reserved. */
+
+#include "mlx5_core.h"
+#include "eswitch.h"
+#include "helper.h"
+#include "lgcy.h"
+
+static void esw_acl_ingress_lgcy_rules_destroy(struct mlx5_vport *vport)
+{
+	if (vport->ingress.legacy.drop_rule) {
+		mlx5_del_flow_rules(vport->ingress.legacy.drop_rule);
+		vport->ingress.legacy.drop_rule = NULL;
+	}
+	esw_acl_ingress_allow_rule_destroy(vport);
+}
+
+static int esw_acl_ingress_lgcy_groups_create(struct mlx5_eswitch *esw,
+					      struct mlx5_vport *vport)
+{
+	int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
+	struct mlx5_core_dev *dev = esw->dev;
+	struct mlx5_flow_group *g;
+	void *match_criteria;
+	u32 *flow_group_in;
+	int err;
+
+	flow_group_in = kvzalloc(inlen, GFP_KERNEL);
+	if (!flow_group_in)
+		return -ENOMEM;
+
+	match_criteria = MLX5_ADDR_OF(create_flow_group_in, flow_group_in, match_criteria);
+
+	MLX5_SET(create_flow_group_in, flow_group_in, match_criteria_enable,
+		 MLX5_MATCH_OUTER_HEADERS);
+	MLX5_SET_TO_ONES(fte_match_param, match_criteria, outer_headers.cvlan_tag);
+	MLX5_SET_TO_ONES(fte_match_param, match_criteria, outer_headers.smac_47_16);
+	MLX5_SET_TO_ONES(fte_match_param, match_criteria, outer_headers.smac_15_0);
+	MLX5_SET(create_flow_group_in, flow_group_in, start_flow_index, 0);
+	MLX5_SET(create_flow_group_in, flow_group_in, end_flow_index, 0);
+
+	g = mlx5_create_flow_group(vport->ingress.acl, flow_group_in);
+	if (IS_ERR(g)) {
+		err = PTR_ERR(g);
+		esw_warn(dev, "vport[%d] ingress create untagged spoofchk flow group, err(%d)\n",
+			 vport->vport, err);
+		goto spoof_err;
+	}
+	vport->ingress.legacy.allow_untagged_spoofchk_grp = g;
+
+	memset(flow_group_in, 0, inlen);
+	MLX5_SET(create_flow_group_in, flow_group_in, match_criteria_enable,
+		 MLX5_MATCH_OUTER_HEADERS);
+	MLX5_SET_TO_ONES(fte_match_param, match_criteria, outer_headers.cvlan_tag);
+	MLX5_SET(create_flow_group_in, flow_group_in, start_flow_index, 1);
+	MLX5_SET(create_flow_group_in, flow_group_in, end_flow_index, 1);
+
+	g = mlx5_create_flow_group(vport->ingress.acl, flow_group_in);
+	if (IS_ERR(g)) {
+		err = PTR_ERR(g);
+		esw_warn(dev, "vport[%d] ingress create untagged flow group, err(%d)\n",
+			 vport->vport, err);
+		goto untagged_err;
+	}
+	vport->ingress.legacy.allow_untagged_only_grp = g;
+
+	memset(flow_group_in, 0, inlen);
+	MLX5_SET(create_flow_group_in, flow_group_in, match_criteria_enable,
+		 MLX5_MATCH_OUTER_HEADERS);
+	MLX5_SET_TO_ONES(fte_match_param, match_criteria, outer_headers.smac_47_16);
+	MLX5_SET_TO_ONES(fte_match_param, match_criteria, outer_headers.smac_15_0);
+	MLX5_SET(create_flow_group_in, flow_group_in, start_flow_index, 2);
+	MLX5_SET(create_flow_group_in, flow_group_in, end_flow_index, 2);
+
+	g = mlx5_create_flow_group(vport->ingress.acl, flow_group_in);
+	if (IS_ERR(g)) {
+		err = PTR_ERR(g);
+		esw_warn(dev, "vport[%d] ingress create spoofchk flow group, err(%d)\n",
+			 vport->vport, err);
+		goto allow_spoof_err;
+	}
+	vport->ingress.legacy.allow_spoofchk_only_grp = g;
+
+	memset(flow_group_in, 0, inlen);
+	MLX5_SET(create_flow_group_in, flow_group_in, start_flow_index, 3);
+	MLX5_SET(create_flow_group_in, flow_group_in, end_flow_index, 3);
+
+	g = mlx5_create_flow_group(vport->ingress.acl, flow_group_in);
+	if (IS_ERR(g)) {
+		err = PTR_ERR(g);
+		esw_warn(dev, "vport[%d] ingress create drop flow group, err(%d)\n",
+			 vport->vport, err);
+		goto drop_err;
+	}
+	vport->ingress.legacy.drop_grp = g;
+	kvfree(flow_group_in);
+	return 0;
+
+drop_err:
+	if (!IS_ERR_OR_NULL(vport->ingress.legacy.allow_spoofchk_only_grp)) {
+		mlx5_destroy_flow_group(vport->ingress.legacy.allow_spoofchk_only_grp);
+		vport->ingress.legacy.allow_spoofchk_only_grp = NULL;
+	}
+allow_spoof_err:
+	if (!IS_ERR_OR_NULL(vport->ingress.legacy.allow_untagged_only_grp)) {
+		mlx5_destroy_flow_group(vport->ingress.legacy.allow_untagged_only_grp);
+		vport->ingress.legacy.allow_untagged_only_grp = NULL;
+	}
+untagged_err:
+	if (!IS_ERR_OR_NULL(vport->ingress.legacy.allow_untagged_spoofchk_grp)) {
+		mlx5_destroy_flow_group(vport->ingress.legacy.allow_untagged_spoofchk_grp);
+		vport->ingress.legacy.allow_untagged_spoofchk_grp = NULL;
+	}
+spoof_err:
+	kvfree(flow_group_in);
+	return err;
+}
+
+static void esw_acl_ingress_lgcy_groups_destroy(struct mlx5_vport *vport)
+{
+	if (vport->ingress.legacy.allow_spoofchk_only_grp) {
+		mlx5_destroy_flow_group(vport->ingress.legacy.allow_spoofchk_only_grp);
+		vport->ingress.legacy.allow_spoofchk_only_grp = NULL;
+	}
+	if (vport->ingress.legacy.allow_untagged_only_grp) {
+		mlx5_destroy_flow_group(vport->ingress.legacy.allow_untagged_only_grp);
+		vport->ingress.legacy.allow_untagged_only_grp = NULL;
+	}
+	if (vport->ingress.legacy.allow_untagged_spoofchk_grp) {
+		mlx5_destroy_flow_group(vport->ingress.legacy.allow_untagged_spoofchk_grp);
+		vport->ingress.legacy.allow_untagged_spoofchk_grp = NULL;
+	}
+	if (vport->ingress.legacy.drop_grp) {
+		mlx5_destroy_flow_group(vport->ingress.legacy.drop_grp);
+		vport->ingress.legacy.drop_grp = NULL;
+	}
+}
+
+int esw_acl_ingress_lgcy_setup(struct mlx5_eswitch *esw,
+			       struct mlx5_vport *vport)
+{
+	bool vst_mode_steering = esw_vst_mode_is_steering(esw);
+	struct mlx5_flow_destination drop_ctr_dst = {};
+	struct mlx5_flow_destination *dst = NULL;
+	struct mlx5_flow_act flow_act = {};
+	struct mlx5_flow_spec *spec = NULL;
+	struct mlx5_fc *counter = NULL;
+	bool vst_check_cvlan = false;
+	bool vst_push_cvlan = false;
+	/* The ingress acl table contains 4 groups
+	 * (2 active rules at the same time -
+	 *      1 allow rule from one of the first 3 groups.
+	 *      1 drop rule from the last group):
+	 * 1)Allow untagged traffic with smac=original mac.
+	 * 2)Allow untagged traffic.
+	 * 3)Allow traffic with smac=original mac.
+	 * 4)Drop all other traffic.
+	 */
+	int table_size = 4;
+	int dest_num = 0;
+	int err = 0;
+	u8 *smac_v;
+
+	esw_acl_ingress_lgcy_rules_destroy(vport);
+
+	if (vport->ingress.legacy.drop_counter) {
+		counter = vport->ingress.legacy.drop_counter;
+	} else if (MLX5_CAP_ESW_INGRESS_ACL(esw->dev, flow_counter)) {
+		counter = mlx5_fc_create(esw->dev, false);
+		if (IS_ERR(counter)) {
+			esw_warn(esw->dev,
+				 "vport[%d] configure ingress drop rule counter failed\n",
+				 vport->vport);
+			counter = NULL;
+		}
+		vport->ingress.legacy.drop_counter = counter;
+	}
+
+	if (!vport->info.vlan && !vport->info.qos && !vport->info.spoofchk) {
+		esw_acl_ingress_lgcy_cleanup(esw, vport);
+		return 0;
+	}
+
+	if (!vport->ingress.acl) {
+		vport->ingress.acl = esw_acl_table_create(esw, vport,
+							  MLX5_FLOW_NAMESPACE_ESW_INGRESS,
+							  table_size);
+		if (IS_ERR(vport->ingress.acl)) {
+			err = PTR_ERR(vport->ingress.acl);
+			vport->ingress.acl = NULL;
+			return err;
+		}
+
+		err = esw_acl_ingress_lgcy_groups_create(esw, vport);
+		if (err)
+			goto out;
+	}
+
+	esw_debug(esw->dev,
+		  "vport[%d] configure ingress rules, vlan(%d) qos(%d)\n",
+		  vport->vport, vport->info.vlan, vport->info.qos);
+
+	spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
+	if (!spec) {
+		err = -ENOMEM;
+		goto out;
+	}
+
+	if ((vport->info.vlan || vport->info.qos)) {
+		if (vst_mode_steering)
+			vst_push_cvlan = true;
+		else if (!MLX5_CAP_ESW(esw->dev, vport_cvlan_insert_always))
+			vst_check_cvlan = true;
+	}
+
+	if (vst_check_cvlan || vport->info.spoofchk)
+		spec->match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
+
+	/* Create ingress allow rule */
+	flow_act.action = MLX5_FLOW_CONTEXT_ACTION_ALLOW;
+	if (vst_push_cvlan) {
+		flow_act.action |= MLX5_FLOW_CONTEXT_ACTION_VLAN_PUSH;
+		flow_act.vlan[0].prio = vport->info.qos;
+		flow_act.vlan[0].vid = vport->info.vlan;
+		flow_act.vlan[0].ethtype = ETH_P_8021Q;
+	}
+
+	if (vst_check_cvlan)
+		MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria,
+				 outer_headers.cvlan_tag);
+
+	if (vport->info.spoofchk) {
+		MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria,
+				 outer_headers.smac_47_16);
+		MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria,
+				 outer_headers.smac_15_0);
+		smac_v = MLX5_ADDR_OF(fte_match_param,
+				      spec->match_value,
+				      outer_headers.smac_47_16);
+		ether_addr_copy(smac_v, vport->info.mac);
+	}
+
+	vport->ingress.allow_rule = mlx5_add_flow_rules(vport->ingress.acl, spec,
+							&flow_act, NULL, 0);
+	if (IS_ERR(vport->ingress.allow_rule)) {
+		err = PTR_ERR(vport->ingress.allow_rule);
+		esw_warn(esw->dev,
+			 "vport[%d] configure ingress allow rule, err(%d)\n",
+			 vport->vport, err);
+		vport->ingress.allow_rule = NULL;
+		goto out;
+	}
+
+	if (!vst_check_cvlan && !vport->info.spoofchk)
+		goto out;
+
+	memset(&flow_act, 0, sizeof(flow_act));
+	flow_act.action = MLX5_FLOW_CONTEXT_ACTION_DROP;
+	/* Attach drop flow counter */
+	if (counter) {
+		flow_act.action |= MLX5_FLOW_CONTEXT_ACTION_COUNT;
+		drop_ctr_dst.type = MLX5_FLOW_DESTINATION_TYPE_COUNTER;
+		drop_ctr_dst.counter_id = mlx5_fc_id(counter);
+		dst = &drop_ctr_dst;
+		dest_num++;
+	}
+	vport->ingress.legacy.drop_rule =
+		mlx5_add_flow_rules(vport->ingress.acl, NULL,
+				    &flow_act, dst, dest_num);
+	if (IS_ERR(vport->ingress.legacy.drop_rule)) {
+		err = PTR_ERR(vport->ingress.legacy.drop_rule);
+		esw_warn(esw->dev,
+			 "vport[%d] configure ingress drop rule, err(%d)\n",
+			 vport->vport, err);
+		vport->ingress.legacy.drop_rule = NULL;
+		goto out;
+	}
+	kvfree(spec);
+	return 0;
+
+out:
+	if (err)
+		esw_acl_ingress_lgcy_cleanup(esw, vport);
+	kvfree(spec);
+	return err;
+}
+
+void esw_acl_ingress_lgcy_cleanup(struct mlx5_eswitch *esw,
+				  struct mlx5_vport *vport)
+{
+	if (IS_ERR_OR_NULL(vport->ingress.acl))
+		goto clean_drop_counter;
+
+	esw_debug(esw->dev, "Destroy vport[%d] E-Switch ingress ACL\n", vport->vport);
+
+	esw_acl_ingress_lgcy_rules_destroy(vport);
+	esw_acl_ingress_lgcy_groups_destroy(vport);
+	esw_acl_ingress_table_destroy(vport);
+
+clean_drop_counter:
+	if (vport->ingress.legacy.drop_counter) {
+		mlx5_fc_destroy(esw->dev, vport->ingress.legacy.drop_counter);
+		vport->ingress.legacy.drop_counter = NULL;
+	}
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/esw/acl/ingress_ofld.c b/drivers/net/ethernet/mellanox/mlx5/core/esw/acl/ingress_ofld.c
new file mode 100644
index 000000000..db578a7e7
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/esw/acl/ingress_ofld.c
@@ -0,0 +1,408 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2020 Mellanox Technologies Inc. All rights reserved. */
+
+#include "mlx5_core.h"
+#include "eswitch.h"
+#include "helper.h"
+#include "ofld.h"
+
+static bool
+esw_acl_ingress_prio_tag_enabled(struct mlx5_eswitch *esw,
+				 const struct mlx5_vport *vport)
+{
+	return (MLX5_CAP_GEN(esw->dev, prio_tag_required) &&
+		mlx5_eswitch_is_vf_vport(esw, vport->vport));
+}
+
+static int esw_acl_ingress_prio_tag_create(struct mlx5_eswitch *esw,
+					   struct mlx5_vport *vport)
+{
+	struct mlx5_flow_act flow_act = {};
+	struct mlx5_flow_spec *spec;
+	int err = 0;
+
+	/* For prio tag mode, there is only 1 FTEs:
+	 * 1) Untagged packets - push prio tag VLAN and modify metadata if
+	 * required, allow
+	 * Unmatched traffic is allowed by default
+	 */
+	spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
+	if (!spec)
+		return -ENOMEM;
+
+	/* Untagged packets - push prio tag VLAN, allow */
+	MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria, outer_headers.cvlan_tag);
+	MLX5_SET(fte_match_param, spec->match_value, outer_headers.cvlan_tag, 0);
+	spec->match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
+	flow_act.action = MLX5_FLOW_CONTEXT_ACTION_VLAN_PUSH |
+			  MLX5_FLOW_CONTEXT_ACTION_ALLOW;
+	flow_act.vlan[0].ethtype = ETH_P_8021Q;
+	flow_act.vlan[0].vid = 0;
+	flow_act.vlan[0].prio = 0;
+
+	if (vport->ingress.offloads.modify_metadata_rule) {
+		flow_act.action |= MLX5_FLOW_CONTEXT_ACTION_MOD_HDR;
+		flow_act.modify_hdr = vport->ingress.offloads.modify_metadata;
+	}
+
+	vport->ingress.allow_rule = mlx5_add_flow_rules(vport->ingress.acl, spec,
+							&flow_act, NULL, 0);
+	if (IS_ERR(vport->ingress.allow_rule)) {
+		err = PTR_ERR(vport->ingress.allow_rule);
+		esw_warn(esw->dev,
+			 "vport[%d] configure ingress untagged allow rule, err(%d)\n",
+			 vport->vport, err);
+		vport->ingress.allow_rule = NULL;
+	}
+
+	kvfree(spec);
+	return err;
+}
+
+static int esw_acl_ingress_mod_metadata_create(struct mlx5_eswitch *esw,
+					       struct mlx5_vport *vport)
+{
+	u8 action[MLX5_UN_SZ_BYTES(set_add_copy_action_in_auto)] = {};
+	struct mlx5_flow_act flow_act = {};
+	int err = 0;
+	u32 key;
+
+	key = mlx5_eswitch_get_vport_metadata_for_match(esw, vport->vport);
+	key >>= ESW_SOURCE_PORT_METADATA_OFFSET;
+
+	MLX5_SET(set_action_in, action, action_type, MLX5_ACTION_TYPE_SET);
+	MLX5_SET(set_action_in, action, field,
+		 MLX5_ACTION_IN_FIELD_METADATA_REG_C_0);
+	MLX5_SET(set_action_in, action, data, key);
+	MLX5_SET(set_action_in, action, offset,
+		 ESW_SOURCE_PORT_METADATA_OFFSET);
+	MLX5_SET(set_action_in, action, length,
+		 ESW_SOURCE_PORT_METADATA_BITS);
+
+	vport->ingress.offloads.modify_metadata =
+		mlx5_modify_header_alloc(esw->dev, MLX5_FLOW_NAMESPACE_ESW_INGRESS,
+					 1, action);
+	if (IS_ERR(vport->ingress.offloads.modify_metadata)) {
+		err = PTR_ERR(vport->ingress.offloads.modify_metadata);
+		esw_warn(esw->dev,
+			 "failed to alloc modify header for vport %d ingress acl (%d)\n",
+			 vport->vport, err);
+		return err;
+	}
+
+	flow_act.action = MLX5_FLOW_CONTEXT_ACTION_MOD_HDR | MLX5_FLOW_CONTEXT_ACTION_ALLOW;
+	flow_act.modify_hdr = vport->ingress.offloads.modify_metadata;
+	flow_act.fg = vport->ingress.offloads.metadata_allmatch_grp;
+	vport->ingress.offloads.modify_metadata_rule =
+				mlx5_add_flow_rules(vport->ingress.acl,
+						    NULL, &flow_act, NULL, 0);
+	if (IS_ERR(vport->ingress.offloads.modify_metadata_rule)) {
+		err = PTR_ERR(vport->ingress.offloads.modify_metadata_rule);
+		esw_warn(esw->dev,
+			 "failed to add setting metadata rule for vport %d ingress acl, err(%d)\n",
+			 vport->vport, err);
+		mlx5_modify_header_dealloc(esw->dev, vport->ingress.offloads.modify_metadata);
+		vport->ingress.offloads.modify_metadata_rule = NULL;
+	}
+	return err;
+}
+
+static void esw_acl_ingress_mod_metadata_destroy(struct mlx5_eswitch *esw,
+						 struct mlx5_vport *vport)
+{
+	if (!vport->ingress.offloads.modify_metadata_rule)
+		return;
+
+	mlx5_del_flow_rules(vport->ingress.offloads.modify_metadata_rule);
+	mlx5_modify_header_dealloc(esw->dev, vport->ingress.offloads.modify_metadata);
+	vport->ingress.offloads.modify_metadata_rule = NULL;
+}
+
+static int esw_acl_ingress_src_port_drop_create(struct mlx5_eswitch *esw,
+						struct mlx5_vport *vport)
+{
+	struct mlx5_flow_act flow_act = {};
+	struct mlx5_flow_handle *flow_rule;
+	int err = 0;
+
+	flow_act.action = MLX5_FLOW_CONTEXT_ACTION_DROP;
+	flow_act.fg = vport->ingress.offloads.drop_grp;
+	flow_rule = mlx5_add_flow_rules(vport->ingress.acl, NULL, &flow_act, NULL, 0);
+	if (IS_ERR(flow_rule)) {
+		err = PTR_ERR(flow_rule);
+		goto out;
+	}
+
+	vport->ingress.offloads.drop_rule = flow_rule;
+out:
+	return err;
+}
+
+static void esw_acl_ingress_src_port_drop_destroy(struct mlx5_eswitch *esw,
+						  struct mlx5_vport *vport)
+{
+	if (!vport->ingress.offloads.drop_rule)
+		return;
+
+	mlx5_del_flow_rules(vport->ingress.offloads.drop_rule);
+	vport->ingress.offloads.drop_rule = NULL;
+}
+
+static int esw_acl_ingress_ofld_rules_create(struct mlx5_eswitch *esw,
+					     struct mlx5_vport *vport)
+{
+	int err;
+
+	if (mlx5_eswitch_vport_match_metadata_enabled(esw)) {
+		err = esw_acl_ingress_mod_metadata_create(esw, vport);
+		if (err) {
+			esw_warn(esw->dev,
+				 "vport(%d) create ingress modify metadata, err(%d)\n",
+				 vport->vport, err);
+			return err;
+		}
+	}
+
+	if (esw_acl_ingress_prio_tag_enabled(esw, vport)) {
+		err = esw_acl_ingress_prio_tag_create(esw, vport);
+		if (err) {
+			esw_warn(esw->dev,
+				 "vport(%d) create ingress prio tag rule, err(%d)\n",
+				 vport->vport, err);
+			goto prio_tag_err;
+		}
+	}
+
+	return 0;
+
+prio_tag_err:
+	esw_acl_ingress_mod_metadata_destroy(esw, vport);
+	return err;
+}
+
+static void esw_acl_ingress_ofld_rules_destroy(struct mlx5_eswitch *esw,
+					       struct mlx5_vport *vport)
+{
+	esw_acl_ingress_allow_rule_destroy(vport);
+	esw_acl_ingress_mod_metadata_destroy(esw, vport);
+	esw_acl_ingress_src_port_drop_destroy(esw, vport);
+}
+
+static int esw_acl_ingress_ofld_groups_create(struct mlx5_eswitch *esw,
+					      struct mlx5_vport *vport)
+{
+	int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
+	struct mlx5_flow_group *g;
+	void *match_criteria;
+	u32 *flow_group_in;
+	u32 flow_index = 0;
+	int ret = 0;
+
+	flow_group_in = kvzalloc(inlen, GFP_KERNEL);
+	if (!flow_group_in)
+		return -ENOMEM;
+
+	if (vport->vport == MLX5_VPORT_UPLINK) {
+		/* This group can hold an FTE to drop all traffic.
+		 * Need in case LAG is enabled.
+		 */
+		MLX5_SET(create_flow_group_in, flow_group_in, start_flow_index, flow_index);
+		MLX5_SET(create_flow_group_in, flow_group_in, end_flow_index, flow_index);
+
+		g = mlx5_create_flow_group(vport->ingress.acl, flow_group_in);
+		if (IS_ERR(g)) {
+			ret = PTR_ERR(g);
+			esw_warn(esw->dev, "vport[%d] ingress create drop flow group, err(%d)\n",
+				 vport->vport, ret);
+			goto drop_err;
+		}
+		vport->ingress.offloads.drop_grp = g;
+		flow_index++;
+	}
+
+	if (esw_acl_ingress_prio_tag_enabled(esw, vport)) {
+		/* This group is to hold FTE to match untagged packets when prio_tag
+		 * is enabled.
+		 */
+		memset(flow_group_in, 0, inlen);
+		match_criteria = MLX5_ADDR_OF(create_flow_group_in,
+					      flow_group_in, match_criteria);
+		MLX5_SET(create_flow_group_in, flow_group_in,
+			 match_criteria_enable, MLX5_MATCH_OUTER_HEADERS);
+		MLX5_SET_TO_ONES(fte_match_param, match_criteria, outer_headers.cvlan_tag);
+		MLX5_SET(create_flow_group_in, flow_group_in, start_flow_index, flow_index);
+		MLX5_SET(create_flow_group_in, flow_group_in, end_flow_index, flow_index);
+
+		g = mlx5_create_flow_group(vport->ingress.acl, flow_group_in);
+		if (IS_ERR(g)) {
+			ret = PTR_ERR(g);
+			esw_warn(esw->dev, "vport[%d] ingress create untagged flow group, err(%d)\n",
+				 vport->vport, ret);
+			goto prio_tag_err;
+		}
+		vport->ingress.offloads.metadata_prio_tag_grp = g;
+		flow_index++;
+	}
+
+	if (mlx5_eswitch_vport_match_metadata_enabled(esw)) {
+		/* This group holds an FTE with no match to add metadata for
+		 * tagged packets if prio-tag is enabled, or for all untagged
+		 * traffic in case prio-tag is disabled.
+		 */
+		memset(flow_group_in, 0, inlen);
+		MLX5_SET(create_flow_group_in, flow_group_in, start_flow_index, flow_index);
+		MLX5_SET(create_flow_group_in, flow_group_in, end_flow_index, flow_index);
+
+		g = mlx5_create_flow_group(vport->ingress.acl, flow_group_in);
+		if (IS_ERR(g)) {
+			ret = PTR_ERR(g);
+			esw_warn(esw->dev, "vport[%d] ingress create drop flow group, err(%d)\n",
+				 vport->vport, ret);
+			goto metadata_err;
+		}
+		vport->ingress.offloads.metadata_allmatch_grp = g;
+	}
+
+	kvfree(flow_group_in);
+	return 0;
+
+metadata_err:
+	if (!IS_ERR_OR_NULL(vport->ingress.offloads.metadata_prio_tag_grp)) {
+		mlx5_destroy_flow_group(vport->ingress.offloads.metadata_prio_tag_grp);
+		vport->ingress.offloads.metadata_prio_tag_grp = NULL;
+	}
+prio_tag_err:
+	if (!IS_ERR_OR_NULL(vport->ingress.offloads.drop_grp)) {
+		mlx5_destroy_flow_group(vport->ingress.offloads.drop_grp);
+		vport->ingress.offloads.drop_grp = NULL;
+	}
+drop_err:
+	kvfree(flow_group_in);
+	return ret;
+}
+
+static void esw_acl_ingress_ofld_groups_destroy(struct mlx5_vport *vport)
+{
+	if (vport->ingress.offloads.metadata_allmatch_grp) {
+		mlx5_destroy_flow_group(vport->ingress.offloads.metadata_allmatch_grp);
+		vport->ingress.offloads.metadata_allmatch_grp = NULL;
+	}
+
+	if (vport->ingress.offloads.metadata_prio_tag_grp) {
+		mlx5_destroy_flow_group(vport->ingress.offloads.metadata_prio_tag_grp);
+		vport->ingress.offloads.metadata_prio_tag_grp = NULL;
+	}
+
+	if (vport->ingress.offloads.drop_grp) {
+		mlx5_destroy_flow_group(vport->ingress.offloads.drop_grp);
+		vport->ingress.offloads.drop_grp = NULL;
+	}
+}
+
+int esw_acl_ingress_ofld_setup(struct mlx5_eswitch *esw,
+			       struct mlx5_vport *vport)
+{
+	int num_ftes = 0;
+	int err;
+
+	if (!mlx5_eswitch_vport_match_metadata_enabled(esw) &&
+	    !esw_acl_ingress_prio_tag_enabled(esw, vport))
+		return 0;
+
+	esw_acl_ingress_allow_rule_destroy(vport);
+
+	if (mlx5_eswitch_vport_match_metadata_enabled(esw))
+		num_ftes++;
+	if (vport->vport == MLX5_VPORT_UPLINK)
+		num_ftes++;
+	if (esw_acl_ingress_prio_tag_enabled(esw, vport))
+		num_ftes++;
+
+	vport->ingress.acl = esw_acl_table_create(esw, vport,
+						  MLX5_FLOW_NAMESPACE_ESW_INGRESS,
+						  num_ftes);
+	if (IS_ERR(vport->ingress.acl)) {
+		err = PTR_ERR(vport->ingress.acl);
+		vport->ingress.acl = NULL;
+		return err;
+	}
+
+	err = esw_acl_ingress_ofld_groups_create(esw, vport);
+	if (err)
+		goto group_err;
+
+	esw_debug(esw->dev,
+		  "vport[%d] configure ingress rules\n", vport->vport);
+
+	err = esw_acl_ingress_ofld_rules_create(esw, vport);
+	if (err)
+		goto rules_err;
+
+	return 0;
+
+rules_err:
+	esw_acl_ingress_ofld_groups_destroy(vport);
+group_err:
+	esw_acl_ingress_table_destroy(vport);
+	return err;
+}
+
+void esw_acl_ingress_ofld_cleanup(struct mlx5_eswitch *esw,
+				  struct mlx5_vport *vport)
+{
+	esw_acl_ingress_ofld_rules_destroy(esw, vport);
+	esw_acl_ingress_ofld_groups_destroy(vport);
+	esw_acl_ingress_table_destroy(vport);
+}
+
+/* Caller must hold rtnl_lock */
+int mlx5_esw_acl_ingress_vport_bond_update(struct mlx5_eswitch *esw, u16 vport_num,
+					   u32 metadata)
+{
+	struct mlx5_vport *vport = mlx5_eswitch_get_vport(esw, vport_num);
+	int err;
+
+	if (WARN_ON_ONCE(IS_ERR(vport))) {
+		esw_warn(esw->dev, "vport(%d) invalid!\n", vport_num);
+		return PTR_ERR(vport);
+	}
+
+	esw_acl_ingress_ofld_rules_destroy(esw, vport);
+
+	vport->metadata = metadata ? metadata : vport->default_metadata;
+
+	/* Recreate ingress acl rules with vport->metadata */
+	err = esw_acl_ingress_ofld_rules_create(esw, vport);
+	if (err)
+		goto out;
+
+	return 0;
+
+out:
+	vport->metadata = vport->default_metadata;
+	return err;
+}
+
+int mlx5_esw_acl_ingress_vport_drop_rule_create(struct mlx5_eswitch *esw, u16 vport_num)
+{
+	struct mlx5_vport *vport = mlx5_eswitch_get_vport(esw, vport_num);
+
+	if (IS_ERR(vport)) {
+		esw_warn(esw->dev, "vport(%d) invalid!\n", vport_num);
+		return PTR_ERR(vport);
+	}
+
+	return esw_acl_ingress_src_port_drop_create(esw, vport);
+}
+
+void mlx5_esw_acl_ingress_vport_drop_rule_destroy(struct mlx5_eswitch *esw, u16 vport_num)
+{
+	struct mlx5_vport *vport = mlx5_eswitch_get_vport(esw, vport_num);
+
+	if (WARN_ON_ONCE(IS_ERR(vport))) {
+		esw_warn(esw->dev, "vport(%d) invalid!\n", vport_num);
+		return;
+	}
+
+	esw_acl_ingress_src_port_drop_destroy(esw, vport);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/esw/acl/lgcy.h b/drivers/net/ethernet/mellanox/mlx5/core/esw/acl/lgcy.h
new file mode 100644
index 000000000..44c152da3
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/esw/acl/lgcy.h
@@ -0,0 +1,17 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2020 Mellanox Technologies Inc. All rights reserved. */
+
+#ifndef __MLX5_ESWITCH_ACL_LGCY_H__
+#define __MLX5_ESWITCH_ACL_LGCY_H__
+
+#include "eswitch.h"
+
+/* Eswitch acl egress external APIs */
+int esw_acl_egress_lgcy_setup(struct mlx5_eswitch *esw, struct mlx5_vport *vport);
+void esw_acl_egress_lgcy_cleanup(struct mlx5_eswitch *esw, struct mlx5_vport *vport);
+
+/* Eswitch acl ingress external APIs */
+int esw_acl_ingress_lgcy_setup(struct mlx5_eswitch *esw, struct mlx5_vport *vport);
+void esw_acl_ingress_lgcy_cleanup(struct mlx5_eswitch *esw, struct mlx5_vport *vport);
+
+#endif /* __MLX5_ESWITCH_ACL_LGCY_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/esw/acl/ofld.h b/drivers/net/ethernet/mellanox/mlx5/core/esw/acl/ofld.h
new file mode 100644
index 000000000..11d3d3978
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/esw/acl/ofld.h
@@ -0,0 +1,44 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2020 Mellanox Technologies Inc. All rights reserved. */
+
+#ifndef __MLX5_ESWITCH_ACL_OFLD_H__
+#define __MLX5_ESWITCH_ACL_OFLD_H__
+
+#include "eswitch.h"
+
+#ifdef CONFIG_MLX5_ESWITCH
+/* Eswitch acl egress external APIs */
+int esw_acl_egress_ofld_setup(struct mlx5_eswitch *esw, struct mlx5_vport *vport);
+void esw_acl_egress_ofld_cleanup(struct mlx5_vport *vport);
+int mlx5_esw_acl_egress_vport_bond(struct mlx5_eswitch *esw, u16 active_vport_num,
+				   u16 passive_vport_num);
+int mlx5_esw_acl_egress_vport_unbond(struct mlx5_eswitch *esw, u16 vport_num);
+
+static inline bool mlx5_esw_acl_egress_fwd2vport_supported(struct mlx5_eswitch *esw)
+{
+	return esw && esw->mode == MLX5_ESWITCH_OFFLOADS &&
+		mlx5_eswitch_vport_match_metadata_enabled(esw) &&
+		MLX5_CAP_ESW_FLOWTABLE(esw->dev, egress_acl_forward_to_vport);
+}
+
+/* Eswitch acl ingress external APIs */
+int esw_acl_ingress_ofld_setup(struct mlx5_eswitch *esw, struct mlx5_vport *vport);
+void esw_acl_ingress_ofld_cleanup(struct mlx5_eswitch *esw, struct mlx5_vport *vport);
+int mlx5_esw_acl_ingress_vport_bond_update(struct mlx5_eswitch *esw, u16 vport_num,
+					   u32 metadata);
+void mlx5_esw_acl_ingress_vport_drop_rule_destroy(struct mlx5_eswitch *esw, u16 vport_num);
+int mlx5_esw_acl_ingress_vport_drop_rule_create(struct mlx5_eswitch *esw, u16 vport_num);
+
+#else /* CONFIG_MLX5_ESWITCH */
+static void
+mlx5_esw_acl_ingress_vport_drop_rule_destroy(struct mlx5_eswitch *esw,
+					     u16 vport_num)
+{}
+
+static int mlx5_esw_acl_ingress_vport_drop_rule_create(struct mlx5_eswitch *esw,
+						       u16 vport_num)
+{
+	return 0;
+}
+#endif /* CONFIG_MLX5_ESWITCH */
+#endif /* __MLX5_ESWITCH_ACL_OFLD_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/esw/bridge.c b/drivers/net/ethernet/mellanox/mlx5/core/esw/bridge.c
new file mode 100644
index 000000000..d0b2676c3
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/esw/bridge.c
@@ -0,0 +1,1853 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2021 Mellanox Technologies. */
+
+#include <linux/build_bug.h>
+#include <linux/list.h>
+#include <linux/notifier.h>
+#include <net/netevent.h>
+#include <net/switchdev.h>
+#include "lib/devcom.h"
+#include "bridge.h"
+#include "eswitch.h"
+#include "bridge_priv.h"
+#define CREATE_TRACE_POINTS
+#include "diag/bridge_tracepoint.h"
+
+#define MLX5_ESW_BRIDGE_INGRESS_TABLE_VLAN_GRP_SIZE 12000
+#define MLX5_ESW_BRIDGE_INGRESS_TABLE_UNTAGGED_GRP_SIZE 16000
+#define MLX5_ESW_BRIDGE_INGRESS_TABLE_VLAN_GRP_IDX_FROM 0
+#define MLX5_ESW_BRIDGE_INGRESS_TABLE_VLAN_GRP_IDX_TO		\
+	(MLX5_ESW_BRIDGE_INGRESS_TABLE_VLAN_GRP_SIZE - 1)
+#define MLX5_ESW_BRIDGE_INGRESS_TABLE_VLAN_FILTER_GRP_IDX_FROM	\
+	(MLX5_ESW_BRIDGE_INGRESS_TABLE_VLAN_GRP_IDX_TO + 1)
+#define MLX5_ESW_BRIDGE_INGRESS_TABLE_VLAN_FILTER_GRP_IDX_TO		\
+	(MLX5_ESW_BRIDGE_INGRESS_TABLE_VLAN_FILTER_GRP_IDX_FROM +	\
+	 MLX5_ESW_BRIDGE_INGRESS_TABLE_VLAN_GRP_SIZE - 1)
+#define MLX5_ESW_BRIDGE_INGRESS_TABLE_QINQ_GRP_IDX_FROM			\
+	(MLX5_ESW_BRIDGE_INGRESS_TABLE_VLAN_FILTER_GRP_IDX_TO + 1)
+#define MLX5_ESW_BRIDGE_INGRESS_TABLE_QINQ_GRP_IDX_TO			\
+	(MLX5_ESW_BRIDGE_INGRESS_TABLE_QINQ_GRP_IDX_FROM +		\
+	 MLX5_ESW_BRIDGE_INGRESS_TABLE_VLAN_GRP_SIZE - 1)
+#define MLX5_ESW_BRIDGE_INGRESS_TABLE_QINQ_FILTER_GRP_IDX_FROM	\
+	(MLX5_ESW_BRIDGE_INGRESS_TABLE_QINQ_GRP_IDX_TO + 1)
+#define MLX5_ESW_BRIDGE_INGRESS_TABLE_QINQ_FILTER_GRP_IDX_TO		\
+	(MLX5_ESW_BRIDGE_INGRESS_TABLE_QINQ_FILTER_GRP_IDX_FROM +	\
+	 MLX5_ESW_BRIDGE_INGRESS_TABLE_VLAN_GRP_SIZE - 1)
+#define MLX5_ESW_BRIDGE_INGRESS_TABLE_MAC_GRP_IDX_FROM			\
+	(MLX5_ESW_BRIDGE_INGRESS_TABLE_QINQ_FILTER_GRP_IDX_TO + 1)
+#define MLX5_ESW_BRIDGE_INGRESS_TABLE_MAC_GRP_IDX_TO			\
+	(MLX5_ESW_BRIDGE_INGRESS_TABLE_MAC_GRP_IDX_FROM +		\
+	 MLX5_ESW_BRIDGE_INGRESS_TABLE_UNTAGGED_GRP_SIZE - 1)
+#define MLX5_ESW_BRIDGE_INGRESS_TABLE_SIZE			\
+	(MLX5_ESW_BRIDGE_INGRESS_TABLE_MAC_GRP_IDX_TO + 1)
+static_assert(MLX5_ESW_BRIDGE_INGRESS_TABLE_SIZE == 64000);
+
+#define MLX5_ESW_BRIDGE_EGRESS_TABLE_VLAN_GRP_SIZE 16000
+#define MLX5_ESW_BRIDGE_EGRESS_TABLE_MAC_GRP_SIZE (32000 - 1)
+#define MLX5_ESW_BRIDGE_EGRESS_TABLE_VLAN_GRP_IDX_FROM 0
+#define MLX5_ESW_BRIDGE_EGRESS_TABLE_VLAN_GRP_IDX_TO		\
+	(MLX5_ESW_BRIDGE_EGRESS_TABLE_VLAN_GRP_SIZE - 1)
+#define MLX5_ESW_BRIDGE_EGRESS_TABLE_QINQ_GRP_IDX_FROM		\
+	(MLX5_ESW_BRIDGE_EGRESS_TABLE_VLAN_GRP_IDX_TO + 1)
+#define MLX5_ESW_BRIDGE_EGRESS_TABLE_QINQ_GRP_IDX_TO			\
+	(MLX5_ESW_BRIDGE_EGRESS_TABLE_QINQ_GRP_IDX_FROM +		\
+	 MLX5_ESW_BRIDGE_EGRESS_TABLE_VLAN_GRP_SIZE - 1)
+#define MLX5_ESW_BRIDGE_EGRESS_TABLE_MAC_GRP_IDX_FROM \
+	(MLX5_ESW_BRIDGE_EGRESS_TABLE_QINQ_GRP_IDX_TO + 1)
+#define MLX5_ESW_BRIDGE_EGRESS_TABLE_MAC_GRP_IDX_TO			\
+	(MLX5_ESW_BRIDGE_EGRESS_TABLE_MAC_GRP_IDX_FROM +		\
+	 MLX5_ESW_BRIDGE_EGRESS_TABLE_MAC_GRP_SIZE - 1)
+#define MLX5_ESW_BRIDGE_EGRESS_TABLE_MISS_GRP_IDX_FROM \
+	(MLX5_ESW_BRIDGE_EGRESS_TABLE_MAC_GRP_IDX_TO + 1)
+#define MLX5_ESW_BRIDGE_EGRESS_TABLE_MISS_GRP_IDX_TO	\
+	MLX5_ESW_BRIDGE_EGRESS_TABLE_MISS_GRP_IDX_FROM
+#define MLX5_ESW_BRIDGE_EGRESS_TABLE_SIZE			\
+	(MLX5_ESW_BRIDGE_EGRESS_TABLE_MISS_GRP_IDX_TO + 1)
+static_assert(MLX5_ESW_BRIDGE_EGRESS_TABLE_SIZE == 64000);
+
+#define MLX5_ESW_BRIDGE_SKIP_TABLE_SIZE 0
+
+enum {
+	MLX5_ESW_BRIDGE_LEVEL_INGRESS_TABLE,
+	MLX5_ESW_BRIDGE_LEVEL_EGRESS_TABLE,
+	MLX5_ESW_BRIDGE_LEVEL_SKIP_TABLE,
+};
+
+static const struct rhashtable_params fdb_ht_params = {
+	.key_offset = offsetof(struct mlx5_esw_bridge_fdb_entry, key),
+	.key_len = sizeof(struct mlx5_esw_bridge_fdb_key),
+	.head_offset = offsetof(struct mlx5_esw_bridge_fdb_entry, ht_node),
+	.automatic_shrinking = true,
+};
+
+enum {
+	MLX5_ESW_BRIDGE_VLAN_FILTERING_FLAG = BIT(0),
+};
+
+struct mlx5_esw_bridge {
+	int ifindex;
+	int refcnt;
+	struct list_head list;
+	struct mlx5_esw_bridge_offloads *br_offloads;
+
+	struct list_head fdb_list;
+	struct rhashtable fdb_ht;
+
+	struct mlx5_flow_table *egress_ft;
+	struct mlx5_flow_group *egress_vlan_fg;
+	struct mlx5_flow_group *egress_qinq_fg;
+	struct mlx5_flow_group *egress_mac_fg;
+	struct mlx5_flow_group *egress_miss_fg;
+	struct mlx5_pkt_reformat *egress_miss_pkt_reformat;
+	struct mlx5_flow_handle *egress_miss_handle;
+	unsigned long ageing_time;
+	u32 flags;
+	u16 vlan_proto;
+};
+
+static void
+mlx5_esw_bridge_fdb_offload_notify(struct net_device *dev, const unsigned char *addr, u16 vid,
+				   unsigned long val)
+{
+	struct switchdev_notifier_fdb_info send_info = {};
+
+	send_info.addr = addr;
+	send_info.vid = vid;
+	send_info.offloaded = true;
+	call_switchdev_notifiers(val, dev, &send_info.info, NULL);
+}
+
+static void
+mlx5_esw_bridge_fdb_del_notify(struct mlx5_esw_bridge_fdb_entry *entry)
+{
+	if (!(entry->flags & (MLX5_ESW_BRIDGE_FLAG_ADDED_BY_USER | MLX5_ESW_BRIDGE_FLAG_PEER)))
+		mlx5_esw_bridge_fdb_offload_notify(entry->dev, entry->key.addr,
+						   entry->key.vid,
+						   SWITCHDEV_FDB_DEL_TO_BRIDGE);
+}
+
+static bool mlx5_esw_bridge_pkt_reformat_vlan_pop_supported(struct mlx5_eswitch *esw)
+{
+	return BIT(MLX5_CAP_ESW_FLOWTABLE_FDB(esw->dev, reformat_remove)) &&
+		MLX5_CAP_GEN_2(esw->dev, max_reformat_remove_size) >= sizeof(struct vlan_hdr) &&
+		MLX5_CAP_GEN_2(esw->dev, max_reformat_remove_offset) >=
+		offsetof(struct vlan_ethhdr, h_vlan_proto);
+}
+
+static struct mlx5_pkt_reformat *
+mlx5_esw_bridge_pkt_reformat_vlan_pop_create(struct mlx5_eswitch *esw)
+{
+	struct mlx5_pkt_reformat_params reformat_params = {};
+
+	reformat_params.type = MLX5_REFORMAT_TYPE_REMOVE_HDR;
+	reformat_params.param_0 = MLX5_REFORMAT_CONTEXT_ANCHOR_MAC_START;
+	reformat_params.param_1 = offsetof(struct vlan_ethhdr, h_vlan_proto);
+	reformat_params.size = sizeof(struct vlan_hdr);
+	return mlx5_packet_reformat_alloc(esw->dev, &reformat_params, MLX5_FLOW_NAMESPACE_FDB);
+}
+
+static struct mlx5_flow_table *
+mlx5_esw_bridge_table_create(int max_fte, u32 level, struct mlx5_eswitch *esw)
+{
+	struct mlx5_flow_table_attr ft_attr = {};
+	struct mlx5_core_dev *dev = esw->dev;
+	struct mlx5_flow_namespace *ns;
+	struct mlx5_flow_table *fdb;
+
+	ns = mlx5_get_flow_namespace(dev, MLX5_FLOW_NAMESPACE_FDB);
+	if (!ns) {
+		esw_warn(dev, "Failed to get FDB namespace\n");
+		return ERR_PTR(-ENOENT);
+	}
+
+	ft_attr.flags = MLX5_FLOW_TABLE_TUNNEL_EN_REFORMAT;
+	ft_attr.max_fte = max_fte;
+	ft_attr.level = level;
+	ft_attr.prio = FDB_BR_OFFLOAD;
+	fdb = mlx5_create_flow_table(ns, &ft_attr);
+	if (IS_ERR(fdb))
+		esw_warn(dev, "Failed to create bridge FDB Table (err=%ld)\n", PTR_ERR(fdb));
+
+	return fdb;
+}
+
+static struct mlx5_flow_group *
+mlx5_esw_bridge_ingress_vlan_proto_fg_create(unsigned int from, unsigned int to, u16 vlan_proto,
+					     struct mlx5_eswitch *esw,
+					     struct mlx5_flow_table *ingress_ft)
+{
+	int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
+	struct mlx5_flow_group *fg;
+	u32 *in, *match;
+
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in)
+		return ERR_PTR(-ENOMEM);
+
+	MLX5_SET(create_flow_group_in, in, match_criteria_enable,
+		 MLX5_MATCH_OUTER_HEADERS | MLX5_MATCH_MISC_PARAMETERS_2);
+	match = MLX5_ADDR_OF(create_flow_group_in, in, match_criteria);
+
+	MLX5_SET_TO_ONES(fte_match_param, match, outer_headers.smac_47_16);
+	MLX5_SET_TO_ONES(fte_match_param, match, outer_headers.smac_15_0);
+	if (vlan_proto == ETH_P_8021Q)
+		MLX5_SET_TO_ONES(fte_match_param, match, outer_headers.cvlan_tag);
+	else if (vlan_proto == ETH_P_8021AD)
+		MLX5_SET_TO_ONES(fte_match_param, match, outer_headers.svlan_tag);
+	MLX5_SET_TO_ONES(fte_match_param, match, outer_headers.first_vid);
+
+	MLX5_SET(fte_match_param, match, misc_parameters_2.metadata_reg_c_0,
+		 mlx5_eswitch_get_vport_metadata_mask());
+
+	MLX5_SET(create_flow_group_in, in, start_flow_index, from);
+	MLX5_SET(create_flow_group_in, in, end_flow_index, to);
+
+	fg = mlx5_create_flow_group(ingress_ft, in);
+	kvfree(in);
+	if (IS_ERR(fg))
+		esw_warn(esw->dev,
+			 "Failed to create VLAN(proto=%x) flow group for bridge ingress table (err=%ld)\n",
+			 vlan_proto, PTR_ERR(fg));
+
+	return fg;
+}
+
+static struct mlx5_flow_group *
+mlx5_esw_bridge_ingress_vlan_fg_create(struct mlx5_eswitch *esw,
+				       struct mlx5_flow_table *ingress_ft)
+{
+	unsigned int from = MLX5_ESW_BRIDGE_INGRESS_TABLE_VLAN_GRP_IDX_FROM;
+	unsigned int to = MLX5_ESW_BRIDGE_INGRESS_TABLE_VLAN_GRP_IDX_TO;
+
+	return mlx5_esw_bridge_ingress_vlan_proto_fg_create(from, to, ETH_P_8021Q, esw, ingress_ft);
+}
+
+static struct mlx5_flow_group *
+mlx5_esw_bridge_ingress_qinq_fg_create(struct mlx5_eswitch *esw,
+				       struct mlx5_flow_table *ingress_ft)
+{
+	unsigned int from = MLX5_ESW_BRIDGE_INGRESS_TABLE_QINQ_GRP_IDX_FROM;
+	unsigned int to = MLX5_ESW_BRIDGE_INGRESS_TABLE_QINQ_GRP_IDX_TO;
+
+	return mlx5_esw_bridge_ingress_vlan_proto_fg_create(from, to, ETH_P_8021AD, esw,
+							    ingress_ft);
+}
+
+static struct mlx5_flow_group *
+mlx5_esw_bridge_ingress_vlan_proto_filter_fg_create(unsigned int from, unsigned int to,
+						    u16 vlan_proto, struct mlx5_eswitch *esw,
+						    struct mlx5_flow_table *ingress_ft)
+{
+	int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
+	struct mlx5_flow_group *fg;
+	u32 *in, *match;
+
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in)
+		return ERR_PTR(-ENOMEM);
+
+	MLX5_SET(create_flow_group_in, in, match_criteria_enable,
+		 MLX5_MATCH_OUTER_HEADERS | MLX5_MATCH_MISC_PARAMETERS_2);
+	match = MLX5_ADDR_OF(create_flow_group_in, in, match_criteria);
+
+	MLX5_SET_TO_ONES(fte_match_param, match, outer_headers.smac_47_16);
+	MLX5_SET_TO_ONES(fte_match_param, match, outer_headers.smac_15_0);
+	if (vlan_proto == ETH_P_8021Q)
+		MLX5_SET_TO_ONES(fte_match_param, match, outer_headers.cvlan_tag);
+	else if (vlan_proto == ETH_P_8021AD)
+		MLX5_SET_TO_ONES(fte_match_param, match, outer_headers.svlan_tag);
+	MLX5_SET(fte_match_param, match, misc_parameters_2.metadata_reg_c_0,
+		 mlx5_eswitch_get_vport_metadata_mask());
+
+	MLX5_SET(create_flow_group_in, in, start_flow_index, from);
+	MLX5_SET(create_flow_group_in, in, end_flow_index, to);
+
+	fg = mlx5_create_flow_group(ingress_ft, in);
+	if (IS_ERR(fg))
+		esw_warn(esw->dev,
+			 "Failed to create bridge ingress table VLAN filter flow group (err=%ld)\n",
+			 PTR_ERR(fg));
+	kvfree(in);
+	return fg;
+}
+
+static struct mlx5_flow_group *
+mlx5_esw_bridge_ingress_vlan_filter_fg_create(struct mlx5_eswitch *esw,
+					      struct mlx5_flow_table *ingress_ft)
+{
+	unsigned int from = MLX5_ESW_BRIDGE_INGRESS_TABLE_VLAN_FILTER_GRP_IDX_FROM;
+	unsigned int to = MLX5_ESW_BRIDGE_INGRESS_TABLE_VLAN_FILTER_GRP_IDX_TO;
+
+	return mlx5_esw_bridge_ingress_vlan_proto_filter_fg_create(from, to, ETH_P_8021Q, esw,
+								   ingress_ft);
+}
+
+static struct mlx5_flow_group *
+mlx5_esw_bridge_ingress_qinq_filter_fg_create(struct mlx5_eswitch *esw,
+					      struct mlx5_flow_table *ingress_ft)
+{
+	unsigned int from = MLX5_ESW_BRIDGE_INGRESS_TABLE_QINQ_FILTER_GRP_IDX_FROM;
+	unsigned int to = MLX5_ESW_BRIDGE_INGRESS_TABLE_QINQ_FILTER_GRP_IDX_TO;
+
+	return mlx5_esw_bridge_ingress_vlan_proto_filter_fg_create(from, to, ETH_P_8021AD, esw,
+								   ingress_ft);
+}
+
+static struct mlx5_flow_group *
+mlx5_esw_bridge_ingress_mac_fg_create(struct mlx5_eswitch *esw, struct mlx5_flow_table *ingress_ft)
+{
+	int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
+	struct mlx5_flow_group *fg;
+	u32 *in, *match;
+
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in)
+		return ERR_PTR(-ENOMEM);
+
+	MLX5_SET(create_flow_group_in, in, match_criteria_enable,
+		 MLX5_MATCH_OUTER_HEADERS | MLX5_MATCH_MISC_PARAMETERS_2);
+	match = MLX5_ADDR_OF(create_flow_group_in, in, match_criteria);
+
+	MLX5_SET_TO_ONES(fte_match_param, match, outer_headers.smac_47_16);
+	MLX5_SET_TO_ONES(fte_match_param, match, outer_headers.smac_15_0);
+
+	MLX5_SET(fte_match_param, match, misc_parameters_2.metadata_reg_c_0,
+		 mlx5_eswitch_get_vport_metadata_mask());
+
+	MLX5_SET(create_flow_group_in, in, start_flow_index,
+		 MLX5_ESW_BRIDGE_INGRESS_TABLE_MAC_GRP_IDX_FROM);
+	MLX5_SET(create_flow_group_in, in, end_flow_index,
+		 MLX5_ESW_BRIDGE_INGRESS_TABLE_MAC_GRP_IDX_TO);
+
+	fg = mlx5_create_flow_group(ingress_ft, in);
+	if (IS_ERR(fg))
+		esw_warn(esw->dev,
+			 "Failed to create MAC flow group for bridge ingress table (err=%ld)\n",
+			 PTR_ERR(fg));
+
+	kvfree(in);
+	return fg;
+}
+
+static struct mlx5_flow_group *
+mlx5_esw_bridge_egress_vlan_proto_fg_create(unsigned int from, unsigned int to, u16 vlan_proto,
+					    struct mlx5_eswitch *esw,
+					    struct mlx5_flow_table *egress_ft)
+{
+	int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
+	struct mlx5_flow_group *fg;
+	u32 *in, *match;
+
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in)
+		return ERR_PTR(-ENOMEM);
+
+	MLX5_SET(create_flow_group_in, in, match_criteria_enable, MLX5_MATCH_OUTER_HEADERS);
+	match = MLX5_ADDR_OF(create_flow_group_in, in, match_criteria);
+
+	MLX5_SET_TO_ONES(fte_match_param, match, outer_headers.dmac_47_16);
+	MLX5_SET_TO_ONES(fte_match_param, match, outer_headers.dmac_15_0);
+	if (vlan_proto == ETH_P_8021Q)
+		MLX5_SET_TO_ONES(fte_match_param, match, outer_headers.cvlan_tag);
+	else if (vlan_proto == ETH_P_8021AD)
+		MLX5_SET_TO_ONES(fte_match_param, match, outer_headers.svlan_tag);
+	MLX5_SET_TO_ONES(fte_match_param, match, outer_headers.first_vid);
+
+	MLX5_SET(create_flow_group_in, in, start_flow_index, from);
+	MLX5_SET(create_flow_group_in, in, end_flow_index, to);
+
+	fg = mlx5_create_flow_group(egress_ft, in);
+	if (IS_ERR(fg))
+		esw_warn(esw->dev,
+			 "Failed to create VLAN flow group for bridge egress table (err=%ld)\n",
+			 PTR_ERR(fg));
+	kvfree(in);
+	return fg;
+}
+
+static struct mlx5_flow_group *
+mlx5_esw_bridge_egress_vlan_fg_create(struct mlx5_eswitch *esw, struct mlx5_flow_table *egress_ft)
+{
+	unsigned int from = MLX5_ESW_BRIDGE_EGRESS_TABLE_VLAN_GRP_IDX_FROM;
+	unsigned int to = MLX5_ESW_BRIDGE_EGRESS_TABLE_VLAN_GRP_IDX_TO;
+
+	return mlx5_esw_bridge_egress_vlan_proto_fg_create(from, to, ETH_P_8021Q, esw, egress_ft);
+}
+
+static struct mlx5_flow_group *
+mlx5_esw_bridge_egress_qinq_fg_create(struct mlx5_eswitch *esw,
+				      struct mlx5_flow_table *egress_ft)
+{
+	unsigned int from = MLX5_ESW_BRIDGE_EGRESS_TABLE_QINQ_GRP_IDX_FROM;
+	unsigned int to = MLX5_ESW_BRIDGE_EGRESS_TABLE_QINQ_GRP_IDX_TO;
+
+	return mlx5_esw_bridge_egress_vlan_proto_fg_create(from, to, ETH_P_8021AD, esw, egress_ft);
+}
+
+static struct mlx5_flow_group *
+mlx5_esw_bridge_egress_mac_fg_create(struct mlx5_eswitch *esw, struct mlx5_flow_table *egress_ft)
+{
+	int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
+	struct mlx5_flow_group *fg;
+	u32 *in, *match;
+
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in)
+		return ERR_PTR(-ENOMEM);
+
+	MLX5_SET(create_flow_group_in, in, match_criteria_enable, MLX5_MATCH_OUTER_HEADERS);
+	match = MLX5_ADDR_OF(create_flow_group_in, in, match_criteria);
+
+	MLX5_SET_TO_ONES(fte_match_param, match, outer_headers.dmac_47_16);
+	MLX5_SET_TO_ONES(fte_match_param, match, outer_headers.dmac_15_0);
+
+	MLX5_SET(create_flow_group_in, in, start_flow_index,
+		 MLX5_ESW_BRIDGE_EGRESS_TABLE_MAC_GRP_IDX_FROM);
+	MLX5_SET(create_flow_group_in, in, end_flow_index,
+		 MLX5_ESW_BRIDGE_EGRESS_TABLE_MAC_GRP_IDX_TO);
+
+	fg = mlx5_create_flow_group(egress_ft, in);
+	if (IS_ERR(fg))
+		esw_warn(esw->dev,
+			 "Failed to create bridge egress table MAC flow group (err=%ld)\n",
+			 PTR_ERR(fg));
+	kvfree(in);
+	return fg;
+}
+
+static struct mlx5_flow_group *
+mlx5_esw_bridge_egress_miss_fg_create(struct mlx5_eswitch *esw, struct mlx5_flow_table *egress_ft)
+{
+	int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
+	struct mlx5_flow_group *fg;
+	u32 *in, *match;
+
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in)
+		return ERR_PTR(-ENOMEM);
+
+	MLX5_SET(create_flow_group_in, in, match_criteria_enable, MLX5_MATCH_MISC_PARAMETERS_2);
+	match = MLX5_ADDR_OF(create_flow_group_in, in, match_criteria);
+
+	MLX5_SET(fte_match_param, match, misc_parameters_2.metadata_reg_c_1, ESW_TUN_MASK);
+
+	MLX5_SET(create_flow_group_in, in, start_flow_index,
+		 MLX5_ESW_BRIDGE_EGRESS_TABLE_MISS_GRP_IDX_FROM);
+	MLX5_SET(create_flow_group_in, in, end_flow_index,
+		 MLX5_ESW_BRIDGE_EGRESS_TABLE_MISS_GRP_IDX_TO);
+
+	fg = mlx5_create_flow_group(egress_ft, in);
+	if (IS_ERR(fg))
+		esw_warn(esw->dev,
+			 "Failed to create bridge egress table miss flow group (err=%ld)\n",
+			 PTR_ERR(fg));
+	kvfree(in);
+	return fg;
+}
+
+static int
+mlx5_esw_bridge_ingress_table_init(struct mlx5_esw_bridge_offloads *br_offloads)
+{
+	struct mlx5_flow_group *mac_fg, *qinq_filter_fg, *qinq_fg, *vlan_filter_fg, *vlan_fg;
+	struct mlx5_flow_table *ingress_ft, *skip_ft;
+	struct mlx5_eswitch *esw = br_offloads->esw;
+	int err;
+
+	if (!mlx5_eswitch_vport_match_metadata_enabled(esw))
+		return -EOPNOTSUPP;
+
+	ingress_ft = mlx5_esw_bridge_table_create(MLX5_ESW_BRIDGE_INGRESS_TABLE_SIZE,
+						  MLX5_ESW_BRIDGE_LEVEL_INGRESS_TABLE,
+						  esw);
+	if (IS_ERR(ingress_ft))
+		return PTR_ERR(ingress_ft);
+
+	skip_ft = mlx5_esw_bridge_table_create(MLX5_ESW_BRIDGE_SKIP_TABLE_SIZE,
+					       MLX5_ESW_BRIDGE_LEVEL_SKIP_TABLE,
+					       esw);
+	if (IS_ERR(skip_ft)) {
+		err = PTR_ERR(skip_ft);
+		goto err_skip_tbl;
+	}
+
+	vlan_fg = mlx5_esw_bridge_ingress_vlan_fg_create(esw, ingress_ft);
+	if (IS_ERR(vlan_fg)) {
+		err = PTR_ERR(vlan_fg);
+		goto err_vlan_fg;
+	}
+
+	vlan_filter_fg = mlx5_esw_bridge_ingress_vlan_filter_fg_create(esw, ingress_ft);
+	if (IS_ERR(vlan_filter_fg)) {
+		err = PTR_ERR(vlan_filter_fg);
+		goto err_vlan_filter_fg;
+	}
+
+	qinq_fg = mlx5_esw_bridge_ingress_qinq_fg_create(esw, ingress_ft);
+	if (IS_ERR(qinq_fg)) {
+		err = PTR_ERR(qinq_fg);
+		goto err_qinq_fg;
+	}
+
+	qinq_filter_fg = mlx5_esw_bridge_ingress_qinq_filter_fg_create(esw, ingress_ft);
+	if (IS_ERR(qinq_filter_fg)) {
+		err = PTR_ERR(qinq_filter_fg);
+		goto err_qinq_filter_fg;
+	}
+
+	mac_fg = mlx5_esw_bridge_ingress_mac_fg_create(esw, ingress_ft);
+	if (IS_ERR(mac_fg)) {
+		err = PTR_ERR(mac_fg);
+		goto err_mac_fg;
+	}
+
+	br_offloads->ingress_ft = ingress_ft;
+	br_offloads->skip_ft = skip_ft;
+	br_offloads->ingress_vlan_fg = vlan_fg;
+	br_offloads->ingress_vlan_filter_fg = vlan_filter_fg;
+	br_offloads->ingress_qinq_fg = qinq_fg;
+	br_offloads->ingress_qinq_filter_fg = qinq_filter_fg;
+	br_offloads->ingress_mac_fg = mac_fg;
+	return 0;
+
+err_mac_fg:
+	mlx5_destroy_flow_group(qinq_filter_fg);
+err_qinq_filter_fg:
+	mlx5_destroy_flow_group(qinq_fg);
+err_qinq_fg:
+	mlx5_destroy_flow_group(vlan_filter_fg);
+err_vlan_filter_fg:
+	mlx5_destroy_flow_group(vlan_fg);
+err_vlan_fg:
+	mlx5_destroy_flow_table(skip_ft);
+err_skip_tbl:
+	mlx5_destroy_flow_table(ingress_ft);
+	return err;
+}
+
+static void
+mlx5_esw_bridge_ingress_table_cleanup(struct mlx5_esw_bridge_offloads *br_offloads)
+{
+	mlx5_destroy_flow_group(br_offloads->ingress_mac_fg);
+	br_offloads->ingress_mac_fg = NULL;
+	mlx5_destroy_flow_group(br_offloads->ingress_qinq_filter_fg);
+	br_offloads->ingress_qinq_filter_fg = NULL;
+	mlx5_destroy_flow_group(br_offloads->ingress_qinq_fg);
+	br_offloads->ingress_qinq_fg = NULL;
+	mlx5_destroy_flow_group(br_offloads->ingress_vlan_filter_fg);
+	br_offloads->ingress_vlan_filter_fg = NULL;
+	mlx5_destroy_flow_group(br_offloads->ingress_vlan_fg);
+	br_offloads->ingress_vlan_fg = NULL;
+	mlx5_destroy_flow_table(br_offloads->skip_ft);
+	br_offloads->skip_ft = NULL;
+	mlx5_destroy_flow_table(br_offloads->ingress_ft);
+	br_offloads->ingress_ft = NULL;
+}
+
+static struct mlx5_flow_handle *
+mlx5_esw_bridge_egress_miss_flow_create(struct mlx5_flow_table *egress_ft,
+					struct mlx5_flow_table *skip_ft,
+					struct mlx5_pkt_reformat *pkt_reformat);
+
+static int
+mlx5_esw_bridge_egress_table_init(struct mlx5_esw_bridge_offloads *br_offloads,
+				  struct mlx5_esw_bridge *bridge)
+{
+	struct mlx5_flow_group *miss_fg = NULL, *mac_fg, *vlan_fg, *qinq_fg;
+	struct mlx5_pkt_reformat *miss_pkt_reformat = NULL;
+	struct mlx5_flow_handle *miss_handle = NULL;
+	struct mlx5_eswitch *esw = br_offloads->esw;
+	struct mlx5_flow_table *egress_ft;
+	int err;
+
+	egress_ft = mlx5_esw_bridge_table_create(MLX5_ESW_BRIDGE_EGRESS_TABLE_SIZE,
+						 MLX5_ESW_BRIDGE_LEVEL_EGRESS_TABLE,
+						 esw);
+	if (IS_ERR(egress_ft))
+		return PTR_ERR(egress_ft);
+
+	vlan_fg = mlx5_esw_bridge_egress_vlan_fg_create(esw, egress_ft);
+	if (IS_ERR(vlan_fg)) {
+		err = PTR_ERR(vlan_fg);
+		goto err_vlan_fg;
+	}
+
+	qinq_fg = mlx5_esw_bridge_egress_qinq_fg_create(esw, egress_ft);
+	if (IS_ERR(qinq_fg)) {
+		err = PTR_ERR(qinq_fg);
+		goto err_qinq_fg;
+	}
+
+	mac_fg = mlx5_esw_bridge_egress_mac_fg_create(esw, egress_ft);
+	if (IS_ERR(mac_fg)) {
+		err = PTR_ERR(mac_fg);
+		goto err_mac_fg;
+	}
+
+	if (mlx5_esw_bridge_pkt_reformat_vlan_pop_supported(esw)) {
+		miss_fg = mlx5_esw_bridge_egress_miss_fg_create(esw, egress_ft);
+		if (IS_ERR(miss_fg)) {
+			esw_warn(esw->dev, "Failed to create miss flow group (err=%ld)\n",
+				 PTR_ERR(miss_fg));
+			miss_fg = NULL;
+			goto skip_miss_flow;
+		}
+
+		miss_pkt_reformat = mlx5_esw_bridge_pkt_reformat_vlan_pop_create(esw);
+		if (IS_ERR(miss_pkt_reformat)) {
+			esw_warn(esw->dev,
+				 "Failed to alloc packet reformat REMOVE_HEADER (err=%ld)\n",
+				 PTR_ERR(miss_pkt_reformat));
+			miss_pkt_reformat = NULL;
+			mlx5_destroy_flow_group(miss_fg);
+			miss_fg = NULL;
+			goto skip_miss_flow;
+		}
+
+		miss_handle = mlx5_esw_bridge_egress_miss_flow_create(egress_ft,
+								      br_offloads->skip_ft,
+								      miss_pkt_reformat);
+		if (IS_ERR(miss_handle)) {
+			esw_warn(esw->dev, "Failed to create miss flow (err=%ld)\n",
+				 PTR_ERR(miss_handle));
+			miss_handle = NULL;
+			mlx5_packet_reformat_dealloc(esw->dev, miss_pkt_reformat);
+			miss_pkt_reformat = NULL;
+			mlx5_destroy_flow_group(miss_fg);
+			miss_fg = NULL;
+			goto skip_miss_flow;
+		}
+	}
+skip_miss_flow:
+
+	bridge->egress_ft = egress_ft;
+	bridge->egress_vlan_fg = vlan_fg;
+	bridge->egress_qinq_fg = qinq_fg;
+	bridge->egress_mac_fg = mac_fg;
+	bridge->egress_miss_fg = miss_fg;
+	bridge->egress_miss_pkt_reformat = miss_pkt_reformat;
+	bridge->egress_miss_handle = miss_handle;
+	return 0;
+
+err_mac_fg:
+	mlx5_destroy_flow_group(qinq_fg);
+err_qinq_fg:
+	mlx5_destroy_flow_group(vlan_fg);
+err_vlan_fg:
+	mlx5_destroy_flow_table(egress_ft);
+	return err;
+}
+
+static void
+mlx5_esw_bridge_egress_table_cleanup(struct mlx5_esw_bridge *bridge)
+{
+	if (bridge->egress_miss_handle)
+		mlx5_del_flow_rules(bridge->egress_miss_handle);
+	if (bridge->egress_miss_pkt_reformat)
+		mlx5_packet_reformat_dealloc(bridge->br_offloads->esw->dev,
+					     bridge->egress_miss_pkt_reformat);
+	if (bridge->egress_miss_fg)
+		mlx5_destroy_flow_group(bridge->egress_miss_fg);
+	mlx5_destroy_flow_group(bridge->egress_mac_fg);
+	mlx5_destroy_flow_group(bridge->egress_qinq_fg);
+	mlx5_destroy_flow_group(bridge->egress_vlan_fg);
+	mlx5_destroy_flow_table(bridge->egress_ft);
+}
+
+static struct mlx5_flow_handle *
+mlx5_esw_bridge_ingress_flow_with_esw_create(u16 vport_num, const unsigned char *addr,
+					     struct mlx5_esw_bridge_vlan *vlan, u32 counter_id,
+					     struct mlx5_esw_bridge *bridge,
+					     struct mlx5_eswitch *esw)
+{
+	struct mlx5_esw_bridge_offloads *br_offloads = bridge->br_offloads;
+	struct mlx5_flow_act flow_act = {
+		.action = MLX5_FLOW_CONTEXT_ACTION_FWD_DEST | MLX5_FLOW_CONTEXT_ACTION_COUNT,
+		.flags = FLOW_ACT_NO_APPEND,
+	};
+	struct mlx5_flow_destination dests[2] = {};
+	struct mlx5_flow_spec *rule_spec;
+	struct mlx5_flow_handle *handle;
+	u8 *smac_v, *smac_c;
+
+	rule_spec = kvzalloc(sizeof(*rule_spec), GFP_KERNEL);
+	if (!rule_spec)
+		return ERR_PTR(-ENOMEM);
+
+	rule_spec->match_criteria_enable = MLX5_MATCH_OUTER_HEADERS | MLX5_MATCH_MISC_PARAMETERS_2;
+
+	smac_v = MLX5_ADDR_OF(fte_match_param, rule_spec->match_value,
+			      outer_headers.smac_47_16);
+	ether_addr_copy(smac_v, addr);
+	smac_c = MLX5_ADDR_OF(fte_match_param, rule_spec->match_criteria,
+			      outer_headers.smac_47_16);
+	eth_broadcast_addr(smac_c);
+
+	MLX5_SET(fte_match_param, rule_spec->match_criteria,
+		 misc_parameters_2.metadata_reg_c_0, mlx5_eswitch_get_vport_metadata_mask());
+	MLX5_SET(fte_match_param, rule_spec->match_value, misc_parameters_2.metadata_reg_c_0,
+		 mlx5_eswitch_get_vport_metadata_for_match(esw, vport_num));
+
+	if (vlan && vlan->pkt_reformat_push) {
+		flow_act.action |= MLX5_FLOW_CONTEXT_ACTION_PACKET_REFORMAT |
+			MLX5_FLOW_CONTEXT_ACTION_MOD_HDR;
+		flow_act.pkt_reformat = vlan->pkt_reformat_push;
+		flow_act.modify_hdr = vlan->pkt_mod_hdr_push_mark;
+	} else if (vlan) {
+		if (bridge->vlan_proto == ETH_P_8021Q) {
+			MLX5_SET_TO_ONES(fte_match_param, rule_spec->match_criteria,
+					 outer_headers.cvlan_tag);
+			MLX5_SET_TO_ONES(fte_match_param, rule_spec->match_value,
+					 outer_headers.cvlan_tag);
+		} else if (bridge->vlan_proto == ETH_P_8021AD) {
+			MLX5_SET_TO_ONES(fte_match_param, rule_spec->match_criteria,
+					 outer_headers.svlan_tag);
+			MLX5_SET_TO_ONES(fte_match_param, rule_spec->match_value,
+					 outer_headers.svlan_tag);
+		}
+		MLX5_SET_TO_ONES(fte_match_param, rule_spec->match_criteria,
+				 outer_headers.first_vid);
+		MLX5_SET(fte_match_param, rule_spec->match_value, outer_headers.first_vid,
+			 vlan->vid);
+	}
+
+	dests[0].type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
+	dests[0].ft = bridge->egress_ft;
+	dests[1].type = MLX5_FLOW_DESTINATION_TYPE_COUNTER;
+	dests[1].counter_id = counter_id;
+
+	handle = mlx5_add_flow_rules(br_offloads->ingress_ft, rule_spec, &flow_act, dests,
+				     ARRAY_SIZE(dests));
+
+	kvfree(rule_spec);
+	return handle;
+}
+
+static struct mlx5_flow_handle *
+mlx5_esw_bridge_ingress_flow_create(u16 vport_num, const unsigned char *addr,
+				    struct mlx5_esw_bridge_vlan *vlan, u32 counter_id,
+				    struct mlx5_esw_bridge *bridge)
+{
+	return mlx5_esw_bridge_ingress_flow_with_esw_create(vport_num, addr, vlan, counter_id,
+							    bridge, bridge->br_offloads->esw);
+}
+
+static struct mlx5_flow_handle *
+mlx5_esw_bridge_ingress_flow_peer_create(u16 vport_num, const unsigned char *addr,
+					 struct mlx5_esw_bridge_vlan *vlan, u32 counter_id,
+					 struct mlx5_esw_bridge *bridge)
+{
+	struct mlx5_devcom *devcom = bridge->br_offloads->esw->dev->priv.devcom;
+	static struct mlx5_flow_handle *handle;
+	struct mlx5_eswitch *peer_esw;
+
+	peer_esw = mlx5_devcom_get_peer_data(devcom, MLX5_DEVCOM_ESW_OFFLOADS);
+	if (!peer_esw)
+		return ERR_PTR(-ENODEV);
+
+	handle = mlx5_esw_bridge_ingress_flow_with_esw_create(vport_num, addr, vlan, counter_id,
+							      bridge, peer_esw);
+
+	mlx5_devcom_release_peer_data(devcom, MLX5_DEVCOM_ESW_OFFLOADS);
+	return handle;
+}
+
+static struct mlx5_flow_handle *
+mlx5_esw_bridge_ingress_filter_flow_create(u16 vport_num, const unsigned char *addr,
+					   struct mlx5_esw_bridge *bridge)
+{
+	struct mlx5_esw_bridge_offloads *br_offloads = bridge->br_offloads;
+	struct mlx5_flow_destination dest = {
+		.type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE,
+		.ft = br_offloads->skip_ft,
+	};
+	struct mlx5_flow_act flow_act = {
+		.action = MLX5_FLOW_CONTEXT_ACTION_FWD_DEST,
+		.flags = FLOW_ACT_NO_APPEND,
+	};
+	struct mlx5_flow_spec *rule_spec;
+	struct mlx5_flow_handle *handle;
+	u8 *smac_v, *smac_c;
+
+	rule_spec = kvzalloc(sizeof(*rule_spec), GFP_KERNEL);
+	if (!rule_spec)
+		return ERR_PTR(-ENOMEM);
+
+	rule_spec->match_criteria_enable = MLX5_MATCH_OUTER_HEADERS | MLX5_MATCH_MISC_PARAMETERS_2;
+
+	smac_v = MLX5_ADDR_OF(fte_match_param, rule_spec->match_value,
+			      outer_headers.smac_47_16);
+	ether_addr_copy(smac_v, addr);
+	smac_c = MLX5_ADDR_OF(fte_match_param, rule_spec->match_criteria,
+			      outer_headers.smac_47_16);
+	eth_broadcast_addr(smac_c);
+
+	MLX5_SET(fte_match_param, rule_spec->match_criteria,
+		 misc_parameters_2.metadata_reg_c_0, mlx5_eswitch_get_vport_metadata_mask());
+	MLX5_SET(fte_match_param, rule_spec->match_value, misc_parameters_2.metadata_reg_c_0,
+		 mlx5_eswitch_get_vport_metadata_for_match(br_offloads->esw, vport_num));
+
+	if (bridge->vlan_proto == ETH_P_8021Q) {
+		MLX5_SET_TO_ONES(fte_match_param, rule_spec->match_criteria,
+				 outer_headers.cvlan_tag);
+		MLX5_SET_TO_ONES(fte_match_param, rule_spec->match_value,
+				 outer_headers.cvlan_tag);
+	} else if (bridge->vlan_proto == ETH_P_8021AD) {
+		MLX5_SET_TO_ONES(fte_match_param, rule_spec->match_criteria,
+				 outer_headers.svlan_tag);
+		MLX5_SET_TO_ONES(fte_match_param, rule_spec->match_value,
+				 outer_headers.svlan_tag);
+	}
+
+	handle = mlx5_add_flow_rules(br_offloads->ingress_ft, rule_spec, &flow_act, &dest, 1);
+
+	kvfree(rule_spec);
+	return handle;
+}
+
+static struct mlx5_flow_handle *
+mlx5_esw_bridge_egress_flow_create(u16 vport_num, u16 esw_owner_vhca_id, const unsigned char *addr,
+				   struct mlx5_esw_bridge_vlan *vlan,
+				   struct mlx5_esw_bridge *bridge)
+{
+	struct mlx5_flow_destination dest = {
+		.type = MLX5_FLOW_DESTINATION_TYPE_VPORT,
+		.vport.num = vport_num,
+	};
+	struct mlx5_flow_act flow_act = {
+		.action = MLX5_FLOW_CONTEXT_ACTION_FWD_DEST,
+		.flags = FLOW_ACT_NO_APPEND,
+	};
+	struct mlx5_flow_spec *rule_spec;
+	struct mlx5_flow_handle *handle;
+	u8 *dmac_v, *dmac_c;
+
+	rule_spec = kvzalloc(sizeof(*rule_spec), GFP_KERNEL);
+	if (!rule_spec)
+		return ERR_PTR(-ENOMEM);
+
+	if (MLX5_CAP_ESW_FLOWTABLE(bridge->br_offloads->esw->dev, flow_source) &&
+	    vport_num == MLX5_VPORT_UPLINK)
+		rule_spec->flow_context.flow_source =
+			MLX5_FLOW_CONTEXT_FLOW_SOURCE_LOCAL_VPORT;
+	rule_spec->match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
+
+	dmac_v = MLX5_ADDR_OF(fte_match_param, rule_spec->match_value,
+			      outer_headers.dmac_47_16);
+	ether_addr_copy(dmac_v, addr);
+	dmac_c = MLX5_ADDR_OF(fte_match_param, rule_spec->match_criteria,
+			      outer_headers.dmac_47_16);
+	eth_broadcast_addr(dmac_c);
+
+	if (vlan) {
+		if (vlan->pkt_reformat_pop) {
+			flow_act.action |= MLX5_FLOW_CONTEXT_ACTION_PACKET_REFORMAT;
+			flow_act.pkt_reformat = vlan->pkt_reformat_pop;
+		}
+
+		if (bridge->vlan_proto == ETH_P_8021Q) {
+			MLX5_SET_TO_ONES(fte_match_param, rule_spec->match_criteria,
+					 outer_headers.cvlan_tag);
+			MLX5_SET_TO_ONES(fte_match_param, rule_spec->match_value,
+					 outer_headers.cvlan_tag);
+		} else if (bridge->vlan_proto == ETH_P_8021AD) {
+			MLX5_SET_TO_ONES(fte_match_param, rule_spec->match_criteria,
+					 outer_headers.svlan_tag);
+			MLX5_SET_TO_ONES(fte_match_param, rule_spec->match_value,
+					 outer_headers.svlan_tag);
+		}
+		MLX5_SET_TO_ONES(fte_match_param, rule_spec->match_criteria,
+				 outer_headers.first_vid);
+		MLX5_SET(fte_match_param, rule_spec->match_value, outer_headers.first_vid,
+			 vlan->vid);
+	}
+
+	if (MLX5_CAP_ESW(bridge->br_offloads->esw->dev, merged_eswitch)) {
+		dest.vport.flags = MLX5_FLOW_DEST_VPORT_VHCA_ID;
+		dest.vport.vhca_id = esw_owner_vhca_id;
+	}
+	handle = mlx5_add_flow_rules(bridge->egress_ft, rule_spec, &flow_act, &dest, 1);
+
+	kvfree(rule_spec);
+	return handle;
+}
+
+static struct mlx5_flow_handle *
+mlx5_esw_bridge_egress_miss_flow_create(struct mlx5_flow_table *egress_ft,
+					struct mlx5_flow_table *skip_ft,
+					struct mlx5_pkt_reformat *pkt_reformat)
+{
+	struct mlx5_flow_destination dest = {
+		.type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE,
+		.ft = skip_ft,
+	};
+	struct mlx5_flow_act flow_act = {
+		.action = MLX5_FLOW_CONTEXT_ACTION_FWD_DEST |
+		MLX5_FLOW_CONTEXT_ACTION_PACKET_REFORMAT,
+		.flags = FLOW_ACT_NO_APPEND,
+		.pkt_reformat = pkt_reformat,
+	};
+	struct mlx5_flow_spec *rule_spec;
+	struct mlx5_flow_handle *handle;
+
+	rule_spec = kvzalloc(sizeof(*rule_spec), GFP_KERNEL);
+	if (!rule_spec)
+		return ERR_PTR(-ENOMEM);
+
+	rule_spec->match_criteria_enable = MLX5_MATCH_MISC_PARAMETERS_2;
+
+	MLX5_SET(fte_match_param, rule_spec->match_criteria,
+		 misc_parameters_2.metadata_reg_c_1, ESW_TUN_MASK);
+	MLX5_SET(fte_match_param, rule_spec->match_value, misc_parameters_2.metadata_reg_c_1,
+		 ESW_TUN_BRIDGE_INGRESS_PUSH_VLAN_MARK);
+
+	handle = mlx5_add_flow_rules(egress_ft, rule_spec, &flow_act, &dest, 1);
+
+	kvfree(rule_spec);
+	return handle;
+}
+
+static struct mlx5_esw_bridge *mlx5_esw_bridge_create(int ifindex,
+						      struct mlx5_esw_bridge_offloads *br_offloads)
+{
+	struct mlx5_esw_bridge *bridge;
+	int err;
+
+	bridge = kvzalloc(sizeof(*bridge), GFP_KERNEL);
+	if (!bridge)
+		return ERR_PTR(-ENOMEM);
+
+	bridge->br_offloads = br_offloads;
+	err = mlx5_esw_bridge_egress_table_init(br_offloads, bridge);
+	if (err)
+		goto err_egress_tbl;
+
+	err = rhashtable_init(&bridge->fdb_ht, &fdb_ht_params);
+	if (err)
+		goto err_fdb_ht;
+
+	INIT_LIST_HEAD(&bridge->fdb_list);
+	bridge->ifindex = ifindex;
+	bridge->refcnt = 1;
+	bridge->ageing_time = clock_t_to_jiffies(BR_DEFAULT_AGEING_TIME);
+	bridge->vlan_proto = ETH_P_8021Q;
+	list_add(&bridge->list, &br_offloads->bridges);
+
+	return bridge;
+
+err_fdb_ht:
+	mlx5_esw_bridge_egress_table_cleanup(bridge);
+err_egress_tbl:
+	kvfree(bridge);
+	return ERR_PTR(err);
+}
+
+static void mlx5_esw_bridge_get(struct mlx5_esw_bridge *bridge)
+{
+	bridge->refcnt++;
+}
+
+static void mlx5_esw_bridge_put(struct mlx5_esw_bridge_offloads *br_offloads,
+				struct mlx5_esw_bridge *bridge)
+{
+	if (--bridge->refcnt)
+		return;
+
+	mlx5_esw_bridge_egress_table_cleanup(bridge);
+	list_del(&bridge->list);
+	rhashtable_destroy(&bridge->fdb_ht);
+	kvfree(bridge);
+
+	if (list_empty(&br_offloads->bridges))
+		mlx5_esw_bridge_ingress_table_cleanup(br_offloads);
+}
+
+static struct mlx5_esw_bridge *
+mlx5_esw_bridge_lookup(int ifindex, struct mlx5_esw_bridge_offloads *br_offloads)
+{
+	struct mlx5_esw_bridge *bridge;
+
+	ASSERT_RTNL();
+
+	list_for_each_entry(bridge, &br_offloads->bridges, list) {
+		if (bridge->ifindex == ifindex) {
+			mlx5_esw_bridge_get(bridge);
+			return bridge;
+		}
+	}
+
+	if (!br_offloads->ingress_ft) {
+		int err = mlx5_esw_bridge_ingress_table_init(br_offloads);
+
+		if (err)
+			return ERR_PTR(err);
+	}
+
+	bridge = mlx5_esw_bridge_create(ifindex, br_offloads);
+	if (IS_ERR(bridge) && list_empty(&br_offloads->bridges))
+		mlx5_esw_bridge_ingress_table_cleanup(br_offloads);
+	return bridge;
+}
+
+static unsigned long mlx5_esw_bridge_port_key_from_data(u16 vport_num, u16 esw_owner_vhca_id)
+{
+	return vport_num | (unsigned long)esw_owner_vhca_id << sizeof(vport_num) * BITS_PER_BYTE;
+}
+
+static unsigned long mlx5_esw_bridge_port_key(struct mlx5_esw_bridge_port *port)
+{
+	return mlx5_esw_bridge_port_key_from_data(port->vport_num, port->esw_owner_vhca_id);
+}
+
+static int mlx5_esw_bridge_port_insert(struct mlx5_esw_bridge_port *port,
+				       struct mlx5_esw_bridge_offloads *br_offloads)
+{
+	return xa_insert(&br_offloads->ports, mlx5_esw_bridge_port_key(port), port, GFP_KERNEL);
+}
+
+static struct mlx5_esw_bridge_port *
+mlx5_esw_bridge_port_lookup(u16 vport_num, u16 esw_owner_vhca_id,
+			    struct mlx5_esw_bridge_offloads *br_offloads)
+{
+	return xa_load(&br_offloads->ports, mlx5_esw_bridge_port_key_from_data(vport_num,
+									       esw_owner_vhca_id));
+}
+
+static void mlx5_esw_bridge_port_erase(struct mlx5_esw_bridge_port *port,
+				       struct mlx5_esw_bridge_offloads *br_offloads)
+{
+	xa_erase(&br_offloads->ports, mlx5_esw_bridge_port_key(port));
+}
+
+static void mlx5_esw_bridge_fdb_entry_refresh(struct mlx5_esw_bridge_fdb_entry *entry)
+{
+	trace_mlx5_esw_bridge_fdb_entry_refresh(entry);
+
+	mlx5_esw_bridge_fdb_offload_notify(entry->dev, entry->key.addr,
+					   entry->key.vid,
+					   SWITCHDEV_FDB_ADD_TO_BRIDGE);
+}
+
+static void
+mlx5_esw_bridge_fdb_entry_cleanup(struct mlx5_esw_bridge_fdb_entry *entry,
+				  struct mlx5_esw_bridge *bridge)
+{
+	trace_mlx5_esw_bridge_fdb_entry_cleanup(entry);
+
+	rhashtable_remove_fast(&bridge->fdb_ht, &entry->ht_node, fdb_ht_params);
+	mlx5_del_flow_rules(entry->egress_handle);
+	if (entry->filter_handle)
+		mlx5_del_flow_rules(entry->filter_handle);
+	mlx5_del_flow_rules(entry->ingress_handle);
+	mlx5_fc_destroy(bridge->br_offloads->esw->dev, entry->ingress_counter);
+	list_del(&entry->vlan_list);
+	list_del(&entry->list);
+	kvfree(entry);
+}
+
+static void
+mlx5_esw_bridge_fdb_entry_notify_and_cleanup(struct mlx5_esw_bridge_fdb_entry *entry,
+					     struct mlx5_esw_bridge *bridge)
+{
+	mlx5_esw_bridge_fdb_del_notify(entry);
+	mlx5_esw_bridge_fdb_entry_cleanup(entry, bridge);
+}
+
+static void mlx5_esw_bridge_fdb_flush(struct mlx5_esw_bridge *bridge)
+{
+	struct mlx5_esw_bridge_fdb_entry *entry, *tmp;
+
+	list_for_each_entry_safe(entry, tmp, &bridge->fdb_list, list)
+		mlx5_esw_bridge_fdb_entry_notify_and_cleanup(entry, bridge);
+}
+
+static struct mlx5_esw_bridge_vlan *
+mlx5_esw_bridge_vlan_lookup(u16 vid, struct mlx5_esw_bridge_port *port)
+{
+	return xa_load(&port->vlans, vid);
+}
+
+static int
+mlx5_esw_bridge_vlan_push_create(u16 vlan_proto, struct mlx5_esw_bridge_vlan *vlan,
+				 struct mlx5_eswitch *esw)
+{
+	struct {
+		__be16	h_vlan_proto;
+		__be16	h_vlan_TCI;
+	} vlan_hdr = { htons(vlan_proto), htons(vlan->vid) };
+	struct mlx5_pkt_reformat_params reformat_params = {};
+	struct mlx5_pkt_reformat *pkt_reformat;
+
+	if (!BIT(MLX5_CAP_ESW_FLOWTABLE_FDB(esw->dev, reformat_insert)) ||
+	    MLX5_CAP_GEN_2(esw->dev, max_reformat_insert_size) < sizeof(vlan_hdr) ||
+	    MLX5_CAP_GEN_2(esw->dev, max_reformat_insert_offset) <
+	    offsetof(struct vlan_ethhdr, h_vlan_proto)) {
+		esw_warn(esw->dev, "Packet reformat INSERT_HEADER is not supported\n");
+		return -EOPNOTSUPP;
+	}
+
+	reformat_params.type = MLX5_REFORMAT_TYPE_INSERT_HDR;
+	reformat_params.param_0 = MLX5_REFORMAT_CONTEXT_ANCHOR_MAC_START;
+	reformat_params.param_1 = offsetof(struct vlan_ethhdr, h_vlan_proto);
+	reformat_params.size = sizeof(vlan_hdr);
+	reformat_params.data = &vlan_hdr;
+	pkt_reformat = mlx5_packet_reformat_alloc(esw->dev,
+						  &reformat_params,
+						  MLX5_FLOW_NAMESPACE_FDB);
+	if (IS_ERR(pkt_reformat)) {
+		esw_warn(esw->dev, "Failed to alloc packet reformat INSERT_HEADER (err=%ld)\n",
+			 PTR_ERR(pkt_reformat));
+		return PTR_ERR(pkt_reformat);
+	}
+
+	vlan->pkt_reformat_push = pkt_reformat;
+	return 0;
+}
+
+static void
+mlx5_esw_bridge_vlan_push_cleanup(struct mlx5_esw_bridge_vlan *vlan, struct mlx5_eswitch *esw)
+{
+	mlx5_packet_reformat_dealloc(esw->dev, vlan->pkt_reformat_push);
+	vlan->pkt_reformat_push = NULL;
+}
+
+static int
+mlx5_esw_bridge_vlan_pop_create(struct mlx5_esw_bridge_vlan *vlan, struct mlx5_eswitch *esw)
+{
+	struct mlx5_pkt_reformat *pkt_reformat;
+
+	if (!mlx5_esw_bridge_pkt_reformat_vlan_pop_supported(esw)) {
+		esw_warn(esw->dev, "Packet reformat REMOVE_HEADER is not supported\n");
+		return -EOPNOTSUPP;
+	}
+
+	pkt_reformat = mlx5_esw_bridge_pkt_reformat_vlan_pop_create(esw);
+	if (IS_ERR(pkt_reformat)) {
+		esw_warn(esw->dev, "Failed to alloc packet reformat REMOVE_HEADER (err=%ld)\n",
+			 PTR_ERR(pkt_reformat));
+		return PTR_ERR(pkt_reformat);
+	}
+
+	vlan->pkt_reformat_pop = pkt_reformat;
+	return 0;
+}
+
+static void
+mlx5_esw_bridge_vlan_pop_cleanup(struct mlx5_esw_bridge_vlan *vlan, struct mlx5_eswitch *esw)
+{
+	mlx5_packet_reformat_dealloc(esw->dev, vlan->pkt_reformat_pop);
+	vlan->pkt_reformat_pop = NULL;
+}
+
+static int
+mlx5_esw_bridge_vlan_push_mark_create(struct mlx5_esw_bridge_vlan *vlan, struct mlx5_eswitch *esw)
+{
+	u8 action[MLX5_UN_SZ_BYTES(set_add_copy_action_in_auto)] = {};
+	struct mlx5_modify_hdr *pkt_mod_hdr;
+
+	MLX5_SET(set_action_in, action, action_type, MLX5_ACTION_TYPE_SET);
+	MLX5_SET(set_action_in, action, field, MLX5_ACTION_IN_FIELD_METADATA_REG_C_1);
+	MLX5_SET(set_action_in, action, offset, 8);
+	MLX5_SET(set_action_in, action, length, ESW_TUN_OPTS_BITS + ESW_TUN_ID_BITS);
+	MLX5_SET(set_action_in, action, data, ESW_TUN_BRIDGE_INGRESS_PUSH_VLAN);
+
+	pkt_mod_hdr = mlx5_modify_header_alloc(esw->dev, MLX5_FLOW_NAMESPACE_FDB, 1, action);
+	if (IS_ERR(pkt_mod_hdr))
+		return PTR_ERR(pkt_mod_hdr);
+
+	vlan->pkt_mod_hdr_push_mark = pkt_mod_hdr;
+	return 0;
+}
+
+static void
+mlx5_esw_bridge_vlan_push_mark_cleanup(struct mlx5_esw_bridge_vlan *vlan, struct mlx5_eswitch *esw)
+{
+	mlx5_modify_header_dealloc(esw->dev, vlan->pkt_mod_hdr_push_mark);
+	vlan->pkt_mod_hdr_push_mark = NULL;
+}
+
+static int
+mlx5_esw_bridge_vlan_push_pop_create(u16 vlan_proto, u16 flags, struct mlx5_esw_bridge_vlan *vlan,
+				     struct mlx5_eswitch *esw)
+{
+	int err;
+
+	if (flags & BRIDGE_VLAN_INFO_PVID) {
+		err = mlx5_esw_bridge_vlan_push_create(vlan_proto, vlan, esw);
+		if (err)
+			return err;
+
+		err = mlx5_esw_bridge_vlan_push_mark_create(vlan, esw);
+		if (err)
+			goto err_vlan_push_mark;
+	}
+
+	if (flags & BRIDGE_VLAN_INFO_UNTAGGED) {
+		err = mlx5_esw_bridge_vlan_pop_create(vlan, esw);
+		if (err)
+			goto err_vlan_pop;
+	}
+
+	return 0;
+
+err_vlan_pop:
+	if (vlan->pkt_mod_hdr_push_mark)
+		mlx5_esw_bridge_vlan_push_mark_cleanup(vlan, esw);
+err_vlan_push_mark:
+	if (vlan->pkt_reformat_push)
+		mlx5_esw_bridge_vlan_push_cleanup(vlan, esw);
+	return err;
+}
+
+static struct mlx5_esw_bridge_vlan *
+mlx5_esw_bridge_vlan_create(u16 vlan_proto, u16 vid, u16 flags, struct mlx5_esw_bridge_port *port,
+			    struct mlx5_eswitch *esw)
+{
+	struct mlx5_esw_bridge_vlan *vlan;
+	int err;
+
+	vlan = kvzalloc(sizeof(*vlan), GFP_KERNEL);
+	if (!vlan)
+		return ERR_PTR(-ENOMEM);
+
+	vlan->vid = vid;
+	vlan->flags = flags;
+	INIT_LIST_HEAD(&vlan->fdb_list);
+
+	err = mlx5_esw_bridge_vlan_push_pop_create(vlan_proto, flags, vlan, esw);
+	if (err)
+		goto err_vlan_push_pop;
+
+	err = xa_insert(&port->vlans, vid, vlan, GFP_KERNEL);
+	if (err)
+		goto err_xa_insert;
+
+	trace_mlx5_esw_bridge_vlan_create(vlan);
+	return vlan;
+
+err_xa_insert:
+	if (vlan->pkt_reformat_pop)
+		mlx5_esw_bridge_vlan_pop_cleanup(vlan, esw);
+	if (vlan->pkt_mod_hdr_push_mark)
+		mlx5_esw_bridge_vlan_push_mark_cleanup(vlan, esw);
+	if (vlan->pkt_reformat_push)
+		mlx5_esw_bridge_vlan_push_cleanup(vlan, esw);
+err_vlan_push_pop:
+	kvfree(vlan);
+	return ERR_PTR(err);
+}
+
+static void mlx5_esw_bridge_vlan_erase(struct mlx5_esw_bridge_port *port,
+				       struct mlx5_esw_bridge_vlan *vlan)
+{
+	xa_erase(&port->vlans, vlan->vid);
+}
+
+static void mlx5_esw_bridge_vlan_flush(struct mlx5_esw_bridge_vlan *vlan,
+				       struct mlx5_esw_bridge *bridge)
+{
+	struct mlx5_eswitch *esw = bridge->br_offloads->esw;
+	struct mlx5_esw_bridge_fdb_entry *entry, *tmp;
+
+	list_for_each_entry_safe(entry, tmp, &vlan->fdb_list, vlan_list)
+		mlx5_esw_bridge_fdb_entry_notify_and_cleanup(entry, bridge);
+
+	if (vlan->pkt_reformat_pop)
+		mlx5_esw_bridge_vlan_pop_cleanup(vlan, esw);
+	if (vlan->pkt_mod_hdr_push_mark)
+		mlx5_esw_bridge_vlan_push_mark_cleanup(vlan, esw);
+	if (vlan->pkt_reformat_push)
+		mlx5_esw_bridge_vlan_push_cleanup(vlan, esw);
+}
+
+static void mlx5_esw_bridge_vlan_cleanup(struct mlx5_esw_bridge_port *port,
+					 struct mlx5_esw_bridge_vlan *vlan,
+					 struct mlx5_esw_bridge *bridge)
+{
+	trace_mlx5_esw_bridge_vlan_cleanup(vlan);
+	mlx5_esw_bridge_vlan_flush(vlan, bridge);
+	mlx5_esw_bridge_vlan_erase(port, vlan);
+	kvfree(vlan);
+}
+
+static void mlx5_esw_bridge_port_vlans_flush(struct mlx5_esw_bridge_port *port,
+					     struct mlx5_esw_bridge *bridge)
+{
+	struct mlx5_esw_bridge_vlan *vlan;
+	unsigned long index;
+
+	xa_for_each(&port->vlans, index, vlan)
+		mlx5_esw_bridge_vlan_cleanup(port, vlan, bridge);
+}
+
+static int mlx5_esw_bridge_port_vlans_recreate(struct mlx5_esw_bridge_port *port,
+					       struct mlx5_esw_bridge *bridge)
+{
+	struct mlx5_esw_bridge_offloads *br_offloads = bridge->br_offloads;
+	struct mlx5_esw_bridge_vlan *vlan;
+	unsigned long i;
+	int err;
+
+	xa_for_each(&port->vlans, i, vlan) {
+		mlx5_esw_bridge_vlan_flush(vlan, bridge);
+		err = mlx5_esw_bridge_vlan_push_pop_create(bridge->vlan_proto, vlan->flags, vlan,
+							   br_offloads->esw);
+		if (err) {
+			esw_warn(br_offloads->esw->dev,
+				 "Failed to create VLAN=%u(proto=%x) push/pop actions (vport=%u,err=%d)\n",
+				 vlan->vid, bridge->vlan_proto, port->vport_num,
+				 err);
+			return err;
+		}
+	}
+
+	return 0;
+}
+
+static int
+mlx5_esw_bridge_vlans_recreate(struct mlx5_esw_bridge *bridge)
+{
+	struct mlx5_esw_bridge_offloads *br_offloads = bridge->br_offloads;
+	struct mlx5_esw_bridge_port *port;
+	unsigned long i;
+	int err;
+
+	xa_for_each(&br_offloads->ports, i, port) {
+		if (port->bridge != bridge)
+			continue;
+
+		err = mlx5_esw_bridge_port_vlans_recreate(port, bridge);
+		if (err)
+			return err;
+	}
+
+	return 0;
+}
+
+static struct mlx5_esw_bridge_vlan *
+mlx5_esw_bridge_port_vlan_lookup(u16 vid, u16 vport_num, u16 esw_owner_vhca_id,
+				 struct mlx5_esw_bridge *bridge, struct mlx5_eswitch *esw)
+{
+	struct mlx5_esw_bridge_port *port;
+	struct mlx5_esw_bridge_vlan *vlan;
+
+	port = mlx5_esw_bridge_port_lookup(vport_num, esw_owner_vhca_id, bridge->br_offloads);
+	if (!port) {
+		/* FDB is added asynchronously on wq while port might have been deleted
+		 * concurrently. Report on 'info' logging level and skip the FDB offload.
+		 */
+		esw_info(esw->dev, "Failed to lookup bridge port (vport=%u)\n", vport_num);
+		return ERR_PTR(-EINVAL);
+	}
+
+	vlan = mlx5_esw_bridge_vlan_lookup(vid, port);
+	if (!vlan) {
+		/* FDB is added asynchronously on wq while vlan might have been deleted
+		 * concurrently. Report on 'info' logging level and skip the FDB offload.
+		 */
+		esw_info(esw->dev, "Failed to lookup bridge port vlan metadata (vport=%u)\n",
+			 vport_num);
+		return ERR_PTR(-EINVAL);
+	}
+
+	return vlan;
+}
+
+static struct mlx5_esw_bridge_fdb_entry *
+mlx5_esw_bridge_fdb_lookup(struct mlx5_esw_bridge *bridge,
+			   const unsigned char *addr, u16 vid)
+{
+	struct mlx5_esw_bridge_fdb_key key = {};
+
+	ether_addr_copy(key.addr, addr);
+	key.vid = vid;
+	return rhashtable_lookup_fast(&bridge->fdb_ht, &key, fdb_ht_params);
+}
+
+static struct mlx5_esw_bridge_fdb_entry *
+mlx5_esw_bridge_fdb_entry_init(struct net_device *dev, u16 vport_num, u16 esw_owner_vhca_id,
+			       const unsigned char *addr, u16 vid, bool added_by_user, bool peer,
+			       struct mlx5_eswitch *esw, struct mlx5_esw_bridge *bridge)
+{
+	struct mlx5_esw_bridge_vlan *vlan = NULL;
+	struct mlx5_esw_bridge_fdb_entry *entry;
+	struct mlx5_flow_handle *handle;
+	struct mlx5_fc *counter;
+	int err;
+
+	if (bridge->flags & MLX5_ESW_BRIDGE_VLAN_FILTERING_FLAG && vid) {
+		vlan = mlx5_esw_bridge_port_vlan_lookup(vid, vport_num, esw_owner_vhca_id, bridge,
+							esw);
+		if (IS_ERR(vlan))
+			return ERR_CAST(vlan);
+	}
+
+	entry = mlx5_esw_bridge_fdb_lookup(bridge, addr, vid);
+	if (entry)
+		mlx5_esw_bridge_fdb_entry_notify_and_cleanup(entry, bridge);
+
+	entry = kvzalloc(sizeof(*entry), GFP_KERNEL);
+	if (!entry)
+		return ERR_PTR(-ENOMEM);
+
+	ether_addr_copy(entry->key.addr, addr);
+	entry->key.vid = vid;
+	entry->dev = dev;
+	entry->vport_num = vport_num;
+	entry->esw_owner_vhca_id = esw_owner_vhca_id;
+	entry->lastuse = jiffies;
+	if (added_by_user)
+		entry->flags |= MLX5_ESW_BRIDGE_FLAG_ADDED_BY_USER;
+	if (peer)
+		entry->flags |= MLX5_ESW_BRIDGE_FLAG_PEER;
+
+	counter = mlx5_fc_create(esw->dev, true);
+	if (IS_ERR(counter)) {
+		err = PTR_ERR(counter);
+		goto err_ingress_fc_create;
+	}
+	entry->ingress_counter = counter;
+
+	handle = peer ?
+		mlx5_esw_bridge_ingress_flow_peer_create(vport_num, addr, vlan,
+							 mlx5_fc_id(counter), bridge) :
+		mlx5_esw_bridge_ingress_flow_create(vport_num, addr, vlan,
+						    mlx5_fc_id(counter), bridge);
+	if (IS_ERR(handle)) {
+		err = PTR_ERR(handle);
+		esw_warn(esw->dev, "Failed to create ingress flow(vport=%u,err=%d)\n",
+			 vport_num, err);
+		goto err_ingress_flow_create;
+	}
+	entry->ingress_handle = handle;
+
+	if (bridge->flags & MLX5_ESW_BRIDGE_VLAN_FILTERING_FLAG) {
+		handle = mlx5_esw_bridge_ingress_filter_flow_create(vport_num, addr, bridge);
+		if (IS_ERR(handle)) {
+			err = PTR_ERR(handle);
+			esw_warn(esw->dev, "Failed to create ingress filter(vport=%u,err=%d)\n",
+				 vport_num, err);
+			goto err_ingress_filter_flow_create;
+		}
+		entry->filter_handle = handle;
+	}
+
+	handle = mlx5_esw_bridge_egress_flow_create(vport_num, esw_owner_vhca_id, addr, vlan,
+						    bridge);
+	if (IS_ERR(handle)) {
+		err = PTR_ERR(handle);
+		esw_warn(esw->dev, "Failed to create egress flow(vport=%u,err=%d)\n",
+			 vport_num, err);
+		goto err_egress_flow_create;
+	}
+	entry->egress_handle = handle;
+
+	err = rhashtable_insert_fast(&bridge->fdb_ht, &entry->ht_node, fdb_ht_params);
+	if (err) {
+		esw_warn(esw->dev, "Failed to insert FDB flow(vport=%u,err=%d)\n", vport_num, err);
+		goto err_ht_init;
+	}
+
+	if (vlan)
+		list_add(&entry->vlan_list, &vlan->fdb_list);
+	else
+		INIT_LIST_HEAD(&entry->vlan_list);
+	list_add(&entry->list, &bridge->fdb_list);
+
+	trace_mlx5_esw_bridge_fdb_entry_init(entry);
+	return entry;
+
+err_ht_init:
+	mlx5_del_flow_rules(entry->egress_handle);
+err_egress_flow_create:
+	if (entry->filter_handle)
+		mlx5_del_flow_rules(entry->filter_handle);
+err_ingress_filter_flow_create:
+	mlx5_del_flow_rules(entry->ingress_handle);
+err_ingress_flow_create:
+	mlx5_fc_destroy(esw->dev, entry->ingress_counter);
+err_ingress_fc_create:
+	kvfree(entry);
+	return ERR_PTR(err);
+}
+
+int mlx5_esw_bridge_ageing_time_set(u16 vport_num, u16 esw_owner_vhca_id, unsigned long ageing_time,
+				    struct mlx5_esw_bridge_offloads *br_offloads)
+{
+	struct mlx5_esw_bridge_port *port;
+
+	port = mlx5_esw_bridge_port_lookup(vport_num, esw_owner_vhca_id, br_offloads);
+	if (!port)
+		return -EINVAL;
+
+	port->bridge->ageing_time = clock_t_to_jiffies(ageing_time);
+	return 0;
+}
+
+int mlx5_esw_bridge_vlan_filtering_set(u16 vport_num, u16 esw_owner_vhca_id, bool enable,
+				       struct mlx5_esw_bridge_offloads *br_offloads)
+{
+	struct mlx5_esw_bridge_port *port;
+	struct mlx5_esw_bridge *bridge;
+	bool filtering;
+
+	port = mlx5_esw_bridge_port_lookup(vport_num, esw_owner_vhca_id, br_offloads);
+	if (!port)
+		return -EINVAL;
+
+	bridge = port->bridge;
+	filtering = bridge->flags & MLX5_ESW_BRIDGE_VLAN_FILTERING_FLAG;
+	if (filtering == enable)
+		return 0;
+
+	mlx5_esw_bridge_fdb_flush(bridge);
+	if (enable)
+		bridge->flags |= MLX5_ESW_BRIDGE_VLAN_FILTERING_FLAG;
+	else
+		bridge->flags &= ~MLX5_ESW_BRIDGE_VLAN_FILTERING_FLAG;
+
+	return 0;
+}
+
+int mlx5_esw_bridge_vlan_proto_set(u16 vport_num, u16 esw_owner_vhca_id, u16 proto,
+				   struct mlx5_esw_bridge_offloads *br_offloads)
+{
+	struct mlx5_esw_bridge_port *port;
+	struct mlx5_esw_bridge *bridge;
+
+	port = mlx5_esw_bridge_port_lookup(vport_num, esw_owner_vhca_id,
+					   br_offloads);
+	if (!port)
+		return -EINVAL;
+
+	bridge = port->bridge;
+	if (bridge->vlan_proto == proto)
+		return 0;
+	if (proto != ETH_P_8021Q && proto != ETH_P_8021AD) {
+		esw_warn(br_offloads->esw->dev, "Can't set unsupported VLAN protocol %x", proto);
+		return -EOPNOTSUPP;
+	}
+
+	mlx5_esw_bridge_fdb_flush(bridge);
+	bridge->vlan_proto = proto;
+	mlx5_esw_bridge_vlans_recreate(bridge);
+
+	return 0;
+}
+
+static int mlx5_esw_bridge_vport_init(u16 vport_num, u16 esw_owner_vhca_id, u16 flags,
+				      struct mlx5_esw_bridge_offloads *br_offloads,
+				      struct mlx5_esw_bridge *bridge)
+{
+	struct mlx5_eswitch *esw = br_offloads->esw;
+	struct mlx5_esw_bridge_port *port;
+	int err;
+
+	port = kvzalloc(sizeof(*port), GFP_KERNEL);
+	if (!port)
+		return -ENOMEM;
+
+	port->vport_num = vport_num;
+	port->esw_owner_vhca_id = esw_owner_vhca_id;
+	port->bridge = bridge;
+	port->flags |= flags;
+	xa_init(&port->vlans);
+	err = mlx5_esw_bridge_port_insert(port, br_offloads);
+	if (err) {
+		esw_warn(esw->dev,
+			 "Failed to insert port metadata (vport=%u,esw_owner_vhca_id=%u,err=%d)\n",
+			 port->vport_num, port->esw_owner_vhca_id, err);
+		goto err_port_insert;
+	}
+	trace_mlx5_esw_bridge_vport_init(port);
+
+	return 0;
+
+err_port_insert:
+	kvfree(port);
+	return err;
+}
+
+static int mlx5_esw_bridge_vport_cleanup(struct mlx5_esw_bridge_offloads *br_offloads,
+					 struct mlx5_esw_bridge_port *port)
+{
+	u16 vport_num = port->vport_num, esw_owner_vhca_id = port->esw_owner_vhca_id;
+	struct mlx5_esw_bridge *bridge = port->bridge;
+	struct mlx5_esw_bridge_fdb_entry *entry, *tmp;
+
+	list_for_each_entry_safe(entry, tmp, &bridge->fdb_list, list)
+		if (entry->vport_num == vport_num && entry->esw_owner_vhca_id == esw_owner_vhca_id)
+			mlx5_esw_bridge_fdb_entry_cleanup(entry, bridge);
+
+	trace_mlx5_esw_bridge_vport_cleanup(port);
+	mlx5_esw_bridge_port_vlans_flush(port, bridge);
+	mlx5_esw_bridge_port_erase(port, br_offloads);
+	kvfree(port);
+	mlx5_esw_bridge_put(br_offloads, bridge);
+	return 0;
+}
+
+static int mlx5_esw_bridge_vport_link_with_flags(int ifindex, u16 vport_num, u16 esw_owner_vhca_id,
+						 u16 flags,
+						 struct mlx5_esw_bridge_offloads *br_offloads,
+						 struct netlink_ext_ack *extack)
+{
+	struct mlx5_esw_bridge *bridge;
+	int err;
+
+	bridge = mlx5_esw_bridge_lookup(ifindex, br_offloads);
+	if (IS_ERR(bridge)) {
+		NL_SET_ERR_MSG_MOD(extack, "Error checking for existing bridge with same ifindex");
+		return PTR_ERR(bridge);
+	}
+
+	err = mlx5_esw_bridge_vport_init(vport_num, esw_owner_vhca_id, flags, br_offloads, bridge);
+	if (err) {
+		NL_SET_ERR_MSG_MOD(extack, "Error initializing port");
+		goto err_vport;
+	}
+	return 0;
+
+err_vport:
+	mlx5_esw_bridge_put(br_offloads, bridge);
+	return err;
+}
+
+int mlx5_esw_bridge_vport_link(int ifindex, u16 vport_num, u16 esw_owner_vhca_id,
+			       struct mlx5_esw_bridge_offloads *br_offloads,
+			       struct netlink_ext_ack *extack)
+{
+	return mlx5_esw_bridge_vport_link_with_flags(ifindex, vport_num, esw_owner_vhca_id, 0,
+						     br_offloads, extack);
+}
+
+int mlx5_esw_bridge_vport_unlink(int ifindex, u16 vport_num, u16 esw_owner_vhca_id,
+				 struct mlx5_esw_bridge_offloads *br_offloads,
+				 struct netlink_ext_ack *extack)
+{
+	struct mlx5_esw_bridge_port *port;
+	int err;
+
+	port = mlx5_esw_bridge_port_lookup(vport_num, esw_owner_vhca_id, br_offloads);
+	if (!port) {
+		NL_SET_ERR_MSG_MOD(extack, "Port is not attached to any bridge");
+		return -EINVAL;
+	}
+	if (port->bridge->ifindex != ifindex) {
+		NL_SET_ERR_MSG_MOD(extack, "Port is attached to another bridge");
+		return -EINVAL;
+	}
+
+	err = mlx5_esw_bridge_vport_cleanup(br_offloads, port);
+	if (err)
+		NL_SET_ERR_MSG_MOD(extack, "Port cleanup failed");
+	return err;
+}
+
+int mlx5_esw_bridge_vport_peer_link(int ifindex, u16 vport_num, u16 esw_owner_vhca_id,
+				    struct mlx5_esw_bridge_offloads *br_offloads,
+				    struct netlink_ext_ack *extack)
+{
+	if (!MLX5_CAP_ESW(br_offloads->esw->dev, merged_eswitch))
+		return 0;
+
+	return mlx5_esw_bridge_vport_link_with_flags(ifindex, vport_num, esw_owner_vhca_id,
+						     MLX5_ESW_BRIDGE_PORT_FLAG_PEER,
+						     br_offloads, extack);
+}
+
+int mlx5_esw_bridge_vport_peer_unlink(int ifindex, u16 vport_num, u16 esw_owner_vhca_id,
+				      struct mlx5_esw_bridge_offloads *br_offloads,
+				      struct netlink_ext_ack *extack)
+{
+	return mlx5_esw_bridge_vport_unlink(ifindex, vport_num, esw_owner_vhca_id, br_offloads,
+					    extack);
+}
+
+int mlx5_esw_bridge_port_vlan_add(u16 vport_num, u16 esw_owner_vhca_id, u16 vid, u16 flags,
+				  struct mlx5_esw_bridge_offloads *br_offloads,
+				  struct netlink_ext_ack *extack)
+{
+	struct mlx5_esw_bridge_port *port;
+	struct mlx5_esw_bridge_vlan *vlan;
+
+	port = mlx5_esw_bridge_port_lookup(vport_num, esw_owner_vhca_id, br_offloads);
+	if (!port)
+		return -EINVAL;
+
+	vlan = mlx5_esw_bridge_vlan_lookup(vid, port);
+	if (vlan) {
+		if (vlan->flags == flags)
+			return 0;
+		mlx5_esw_bridge_vlan_cleanup(port, vlan, port->bridge);
+	}
+
+	vlan = mlx5_esw_bridge_vlan_create(port->bridge->vlan_proto, vid, flags, port,
+					   br_offloads->esw);
+	if (IS_ERR(vlan)) {
+		NL_SET_ERR_MSG_MOD(extack, "Failed to create VLAN entry");
+		return PTR_ERR(vlan);
+	}
+	return 0;
+}
+
+void mlx5_esw_bridge_port_vlan_del(u16 vport_num, u16 esw_owner_vhca_id, u16 vid,
+				   struct mlx5_esw_bridge_offloads *br_offloads)
+{
+	struct mlx5_esw_bridge_port *port;
+	struct mlx5_esw_bridge_vlan *vlan;
+
+	port = mlx5_esw_bridge_port_lookup(vport_num, esw_owner_vhca_id, br_offloads);
+	if (!port)
+		return;
+
+	vlan = mlx5_esw_bridge_vlan_lookup(vid, port);
+	if (!vlan)
+		return;
+	mlx5_esw_bridge_vlan_cleanup(port, vlan, port->bridge);
+}
+
+void mlx5_esw_bridge_fdb_update_used(struct net_device *dev, u16 vport_num, u16 esw_owner_vhca_id,
+				     struct mlx5_esw_bridge_offloads *br_offloads,
+				     struct switchdev_notifier_fdb_info *fdb_info)
+{
+	struct mlx5_esw_bridge_fdb_entry *entry;
+	struct mlx5_esw_bridge_port *port;
+	struct mlx5_esw_bridge *bridge;
+
+	port = mlx5_esw_bridge_port_lookup(vport_num, esw_owner_vhca_id, br_offloads);
+	if (!port)
+		return;
+
+	bridge = port->bridge;
+	entry = mlx5_esw_bridge_fdb_lookup(bridge, fdb_info->addr, fdb_info->vid);
+	if (!entry) {
+		esw_debug(br_offloads->esw->dev,
+			  "FDB entry with specified key not found (MAC=%pM,vid=%u,vport=%u)\n",
+			  fdb_info->addr, fdb_info->vid, vport_num);
+		return;
+	}
+
+	entry->lastuse = jiffies;
+}
+
+void mlx5_esw_bridge_fdb_create(struct net_device *dev, u16 vport_num, u16 esw_owner_vhca_id,
+				struct mlx5_esw_bridge_offloads *br_offloads,
+				struct switchdev_notifier_fdb_info *fdb_info)
+{
+	struct mlx5_esw_bridge_fdb_entry *entry;
+	struct mlx5_esw_bridge_port *port;
+	struct mlx5_esw_bridge *bridge;
+
+	port = mlx5_esw_bridge_port_lookup(vport_num, esw_owner_vhca_id, br_offloads);
+	if (!port)
+		return;
+
+	bridge = port->bridge;
+	entry = mlx5_esw_bridge_fdb_entry_init(dev, vport_num, esw_owner_vhca_id, fdb_info->addr,
+					       fdb_info->vid, fdb_info->added_by_user,
+					       port->flags & MLX5_ESW_BRIDGE_PORT_FLAG_PEER,
+					       br_offloads->esw, bridge);
+	if (IS_ERR(entry))
+		return;
+
+	if (entry->flags & MLX5_ESW_BRIDGE_FLAG_ADDED_BY_USER)
+		mlx5_esw_bridge_fdb_offload_notify(dev, entry->key.addr, entry->key.vid,
+						   SWITCHDEV_FDB_OFFLOADED);
+	else if (!(entry->flags & MLX5_ESW_BRIDGE_FLAG_PEER))
+		/* Take over dynamic entries to prevent kernel bridge from aging them out. */
+		mlx5_esw_bridge_fdb_offload_notify(dev, entry->key.addr, entry->key.vid,
+						   SWITCHDEV_FDB_ADD_TO_BRIDGE);
+}
+
+void mlx5_esw_bridge_fdb_remove(struct net_device *dev, u16 vport_num, u16 esw_owner_vhca_id,
+				struct mlx5_esw_bridge_offloads *br_offloads,
+				struct switchdev_notifier_fdb_info *fdb_info)
+{
+	struct mlx5_eswitch *esw = br_offloads->esw;
+	struct mlx5_esw_bridge_fdb_entry *entry;
+	struct mlx5_esw_bridge_port *port;
+	struct mlx5_esw_bridge *bridge;
+
+	port = mlx5_esw_bridge_port_lookup(vport_num, esw_owner_vhca_id, br_offloads);
+	if (!port)
+		return;
+
+	bridge = port->bridge;
+	entry = mlx5_esw_bridge_fdb_lookup(bridge, fdb_info->addr, fdb_info->vid);
+	if (!entry) {
+		esw_warn(esw->dev,
+			 "FDB entry with specified key not found (MAC=%pM,vid=%u,vport=%u)\n",
+			 fdb_info->addr, fdb_info->vid, vport_num);
+		return;
+	}
+
+	mlx5_esw_bridge_fdb_entry_notify_and_cleanup(entry, bridge);
+}
+
+void mlx5_esw_bridge_update(struct mlx5_esw_bridge_offloads *br_offloads)
+{
+	struct mlx5_esw_bridge_fdb_entry *entry, *tmp;
+	struct mlx5_esw_bridge *bridge;
+
+	list_for_each_entry(bridge, &br_offloads->bridges, list) {
+		list_for_each_entry_safe(entry, tmp, &bridge->fdb_list, list) {
+			unsigned long lastuse =
+				(unsigned long)mlx5_fc_query_lastuse(entry->ingress_counter);
+
+			if (entry->flags & MLX5_ESW_BRIDGE_FLAG_ADDED_BY_USER)
+				continue;
+
+			if (time_after(lastuse, entry->lastuse))
+				mlx5_esw_bridge_fdb_entry_refresh(entry);
+			else if (!(entry->flags & MLX5_ESW_BRIDGE_FLAG_PEER) &&
+				 time_is_before_jiffies(entry->lastuse + bridge->ageing_time))
+				mlx5_esw_bridge_fdb_entry_notify_and_cleanup(entry, bridge);
+		}
+	}
+}
+
+static void mlx5_esw_bridge_flush(struct mlx5_esw_bridge_offloads *br_offloads)
+{
+	struct mlx5_esw_bridge_port *port;
+	unsigned long i;
+
+	xa_for_each(&br_offloads->ports, i, port)
+		mlx5_esw_bridge_vport_cleanup(br_offloads, port);
+
+	WARN_ONCE(!list_empty(&br_offloads->bridges),
+		  "Cleaning up bridge offloads while still having bridges attached\n");
+}
+
+struct mlx5_esw_bridge_offloads *mlx5_esw_bridge_init(struct mlx5_eswitch *esw)
+{
+	struct mlx5_esw_bridge_offloads *br_offloads;
+
+	ASSERT_RTNL();
+
+	br_offloads = kvzalloc(sizeof(*br_offloads), GFP_KERNEL);
+	if (!br_offloads)
+		return ERR_PTR(-ENOMEM);
+
+	INIT_LIST_HEAD(&br_offloads->bridges);
+	xa_init(&br_offloads->ports);
+	br_offloads->esw = esw;
+	esw->br_offloads = br_offloads;
+
+	return br_offloads;
+}
+
+void mlx5_esw_bridge_cleanup(struct mlx5_eswitch *esw)
+{
+	struct mlx5_esw_bridge_offloads *br_offloads = esw->br_offloads;
+
+	ASSERT_RTNL();
+
+	if (!br_offloads)
+		return;
+
+	mlx5_esw_bridge_flush(br_offloads);
+	WARN_ON(!xa_empty(&br_offloads->ports));
+
+	esw->br_offloads = NULL;
+	kvfree(br_offloads);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/esw/bridge.h b/drivers/net/ethernet/mellanox/mlx5/core/esw/bridge.h
new file mode 100644
index 000000000..10851a515
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/esw/bridge.h
@@ -0,0 +1,73 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2021 Mellanox Technologies. */
+
+#ifndef __MLX5_ESW_BRIDGE_H__
+#define __MLX5_ESW_BRIDGE_H__
+
+#include <linux/notifier.h>
+#include <linux/list.h>
+#include <linux/workqueue.h>
+#include <linux/xarray.h>
+#include "eswitch.h"
+
+struct mlx5_flow_table;
+struct mlx5_flow_group;
+
+struct mlx5_esw_bridge_offloads {
+	struct mlx5_eswitch *esw;
+	struct list_head bridges;
+	struct xarray ports;
+
+	struct notifier_block netdev_nb;
+	struct notifier_block nb_blk;
+	struct notifier_block nb;
+	struct workqueue_struct *wq;
+	struct delayed_work update_work;
+
+	struct mlx5_flow_table *ingress_ft;
+	struct mlx5_flow_group *ingress_vlan_fg;
+	struct mlx5_flow_group *ingress_vlan_filter_fg;
+	struct mlx5_flow_group *ingress_qinq_fg;
+	struct mlx5_flow_group *ingress_qinq_filter_fg;
+	struct mlx5_flow_group *ingress_mac_fg;
+
+	struct mlx5_flow_table *skip_ft;
+};
+
+struct mlx5_esw_bridge_offloads *mlx5_esw_bridge_init(struct mlx5_eswitch *esw);
+void mlx5_esw_bridge_cleanup(struct mlx5_eswitch *esw);
+int mlx5_esw_bridge_vport_link(int ifindex, u16 vport_num, u16 esw_owner_vhca_id,
+			       struct mlx5_esw_bridge_offloads *br_offloads,
+			       struct netlink_ext_ack *extack);
+int mlx5_esw_bridge_vport_unlink(int ifindex, u16 vport_num, u16 esw_owner_vhca_id,
+				 struct mlx5_esw_bridge_offloads *br_offloads,
+				 struct netlink_ext_ack *extack);
+int mlx5_esw_bridge_vport_peer_link(int ifindex, u16 vport_num, u16 esw_owner_vhca_id,
+				    struct mlx5_esw_bridge_offloads *br_offloads,
+				    struct netlink_ext_ack *extack);
+int mlx5_esw_bridge_vport_peer_unlink(int ifindex, u16 vport_num, u16 esw_owner_vhca_id,
+				      struct mlx5_esw_bridge_offloads *br_offloads,
+				      struct netlink_ext_ack *extack);
+void mlx5_esw_bridge_fdb_update_used(struct net_device *dev, u16 vport_num, u16 esw_owner_vhca_id,
+				     struct mlx5_esw_bridge_offloads *br_offloads,
+				     struct switchdev_notifier_fdb_info *fdb_info);
+void mlx5_esw_bridge_fdb_create(struct net_device *dev, u16 vport_num, u16 esw_owner_vhca_id,
+				struct mlx5_esw_bridge_offloads *br_offloads,
+				struct switchdev_notifier_fdb_info *fdb_info);
+void mlx5_esw_bridge_fdb_remove(struct net_device *dev, u16 vport_num, u16 esw_owner_vhca_id,
+				struct mlx5_esw_bridge_offloads *br_offloads,
+				struct switchdev_notifier_fdb_info *fdb_info);
+void mlx5_esw_bridge_update(struct mlx5_esw_bridge_offloads *br_offloads);
+int mlx5_esw_bridge_ageing_time_set(u16 vport_num, u16 esw_owner_vhca_id, unsigned long ageing_time,
+				    struct mlx5_esw_bridge_offloads *br_offloads);
+int mlx5_esw_bridge_vlan_filtering_set(u16 vport_num, u16 esw_owner_vhca_id, bool enable,
+				       struct mlx5_esw_bridge_offloads *br_offloads);
+int mlx5_esw_bridge_vlan_proto_set(u16 vport_num, u16 esw_owner_vhca_id, u16 proto,
+				   struct mlx5_esw_bridge_offloads *br_offloads);
+int mlx5_esw_bridge_port_vlan_add(u16 vport_num, u16 esw_owner_vhca_id, u16 vid, u16 flags,
+				  struct mlx5_esw_bridge_offloads *br_offloads,
+				  struct netlink_ext_ack *extack);
+void mlx5_esw_bridge_port_vlan_del(u16 vport_num, u16 esw_owner_vhca_id, u16 vid,
+				   struct mlx5_esw_bridge_offloads *br_offloads);
+
+#endif /* __MLX5_ESW_BRIDGE_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/esw/bridge_priv.h b/drivers/net/ethernet/mellanox/mlx5/core/esw/bridge_priv.h
new file mode 100644
index 000000000..878311fe9
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/esw/bridge_priv.h
@@ -0,0 +1,63 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2021 Mellanox Technologies. */
+
+#ifndef _MLX5_ESW_BRIDGE_PRIVATE_
+#define _MLX5_ESW_BRIDGE_PRIVATE_
+
+#include <linux/netdevice.h>
+#include <linux/if_bridge.h>
+#include <linux/if_vlan.h>
+#include <linux/if_ether.h>
+#include <linux/rhashtable.h>
+#include <linux/xarray.h>
+#include "fs_core.h"
+
+struct mlx5_esw_bridge_fdb_key {
+	unsigned char addr[ETH_ALEN];
+	u16 vid;
+};
+
+enum {
+	MLX5_ESW_BRIDGE_FLAG_ADDED_BY_USER = BIT(0),
+	MLX5_ESW_BRIDGE_FLAG_PEER = BIT(1),
+};
+
+enum {
+	MLX5_ESW_BRIDGE_PORT_FLAG_PEER = BIT(0),
+};
+
+struct mlx5_esw_bridge_fdb_entry {
+	struct mlx5_esw_bridge_fdb_key key;
+	struct rhash_head ht_node;
+	struct net_device *dev;
+	struct list_head list;
+	struct list_head vlan_list;
+	u16 vport_num;
+	u16 esw_owner_vhca_id;
+	u16 flags;
+
+	struct mlx5_flow_handle *ingress_handle;
+	struct mlx5_fc *ingress_counter;
+	unsigned long lastuse;
+	struct mlx5_flow_handle *egress_handle;
+	struct mlx5_flow_handle *filter_handle;
+};
+
+struct mlx5_esw_bridge_vlan {
+	u16 vid;
+	u16 flags;
+	struct list_head fdb_list;
+	struct mlx5_pkt_reformat *pkt_reformat_push;
+	struct mlx5_pkt_reformat *pkt_reformat_pop;
+	struct mlx5_modify_hdr *pkt_mod_hdr_push_mark;
+};
+
+struct mlx5_esw_bridge_port {
+	u16 vport_num;
+	u16 esw_owner_vhca_id;
+	u16 flags;
+	struct mlx5_esw_bridge *bridge;
+	struct xarray vlans;
+};
+
+#endif /* _MLX5_ESW_BRIDGE_PRIVATE_ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/esw/debugfs.c b/drivers/net/ethernet/mellanox/mlx5/core/esw/debugfs.c
new file mode 100644
index 000000000..2db13c71e
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/esw/debugfs.c
@@ -0,0 +1,182 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2022, NVIDIA CORPORATION & AFFILIATES. All rights reserved. */
+
+#include <linux/debugfs.h>
+#include "eswitch.h"
+
+enum vnic_diag_counter {
+	MLX5_VNIC_DIAG_TOTAL_Q_UNDER_PROCESSOR_HANDLE,
+	MLX5_VNIC_DIAG_SEND_QUEUE_PRIORITY_UPDATE_FLOW,
+	MLX5_VNIC_DIAG_COMP_EQ_OVERRUN,
+	MLX5_VNIC_DIAG_ASYNC_EQ_OVERRUN,
+	MLX5_VNIC_DIAG_CQ_OVERRUN,
+	MLX5_VNIC_DIAG_INVALID_COMMAND,
+	MLX5_VNIC_DIAG_QOUTA_EXCEEDED_COMMAND,
+};
+
+static int mlx5_esw_query_vnic_diag(struct mlx5_vport *vport, enum vnic_diag_counter counter,
+				    u32 *val)
+{
+	u32 out[MLX5_ST_SZ_DW(query_vnic_env_out)] = {};
+	u32 in[MLX5_ST_SZ_DW(query_vnic_env_in)] = {};
+	struct mlx5_core_dev *dev = vport->dev;
+	u16 vport_num = vport->vport;
+	void *vnic_diag_out;
+	int err;
+
+	MLX5_SET(query_vnic_env_in, in, opcode, MLX5_CMD_OP_QUERY_VNIC_ENV);
+	MLX5_SET(query_vnic_env_in, in, vport_number, vport_num);
+	if (!mlx5_esw_is_manager_vport(dev->priv.eswitch, vport_num))
+		MLX5_SET(query_vnic_env_in, in, other_vport, 1);
+
+	err = mlx5_cmd_exec(dev, in, sizeof(in), out, sizeof(out));
+	if (err)
+		return err;
+
+	vnic_diag_out = MLX5_ADDR_OF(query_vnic_env_out, out, vport_env);
+	switch (counter) {
+	case MLX5_VNIC_DIAG_TOTAL_Q_UNDER_PROCESSOR_HANDLE:
+		*val = MLX5_GET(vnic_diagnostic_statistics, vnic_diag_out, total_error_queues);
+		break;
+	case MLX5_VNIC_DIAG_SEND_QUEUE_PRIORITY_UPDATE_FLOW:
+		*val = MLX5_GET(vnic_diagnostic_statistics, vnic_diag_out,
+				send_queue_priority_update_flow);
+		break;
+	case MLX5_VNIC_DIAG_COMP_EQ_OVERRUN:
+		*val = MLX5_GET(vnic_diagnostic_statistics, vnic_diag_out, comp_eq_overrun);
+		break;
+	case MLX5_VNIC_DIAG_ASYNC_EQ_OVERRUN:
+		*val = MLX5_GET(vnic_diagnostic_statistics, vnic_diag_out, async_eq_overrun);
+		break;
+	case MLX5_VNIC_DIAG_CQ_OVERRUN:
+		*val = MLX5_GET(vnic_diagnostic_statistics, vnic_diag_out, cq_overrun);
+		break;
+	case MLX5_VNIC_DIAG_INVALID_COMMAND:
+		*val = MLX5_GET(vnic_diagnostic_statistics, vnic_diag_out, invalid_command);
+		break;
+	case MLX5_VNIC_DIAG_QOUTA_EXCEEDED_COMMAND:
+		*val = MLX5_GET(vnic_diagnostic_statistics, vnic_diag_out, quota_exceeded_command);
+		break;
+	}
+
+	return 0;
+}
+
+static int __show_vnic_diag(struct seq_file *file, struct mlx5_vport *vport,
+			    enum vnic_diag_counter type)
+{
+	u32 val = 0;
+	int ret;
+
+	ret = mlx5_esw_query_vnic_diag(vport, type, &val);
+	if (ret)
+		return ret;
+
+	seq_printf(file, "%d\n", val);
+	return 0;
+}
+
+static int total_q_under_processor_handle_show(struct seq_file *file, void *priv)
+{
+	return __show_vnic_diag(file, file->private, MLX5_VNIC_DIAG_TOTAL_Q_UNDER_PROCESSOR_HANDLE);
+}
+
+static int send_queue_priority_update_flow_show(struct seq_file *file, void *priv)
+{
+	return __show_vnic_diag(file, file->private,
+				MLX5_VNIC_DIAG_SEND_QUEUE_PRIORITY_UPDATE_FLOW);
+}
+
+static int comp_eq_overrun_show(struct seq_file *file, void *priv)
+{
+	return __show_vnic_diag(file, file->private, MLX5_VNIC_DIAG_COMP_EQ_OVERRUN);
+}
+
+static int async_eq_overrun_show(struct seq_file *file, void *priv)
+{
+	return __show_vnic_diag(file, file->private, MLX5_VNIC_DIAG_ASYNC_EQ_OVERRUN);
+}
+
+static int cq_overrun_show(struct seq_file *file, void *priv)
+{
+	return __show_vnic_diag(file, file->private, MLX5_VNIC_DIAG_CQ_OVERRUN);
+}
+
+static int invalid_command_show(struct seq_file *file, void *priv)
+{
+	return __show_vnic_diag(file, file->private, MLX5_VNIC_DIAG_INVALID_COMMAND);
+}
+
+static int quota_exceeded_command_show(struct seq_file *file, void *priv)
+{
+	return __show_vnic_diag(file, file->private, MLX5_VNIC_DIAG_QOUTA_EXCEEDED_COMMAND);
+}
+
+DEFINE_SHOW_ATTRIBUTE(total_q_under_processor_handle);
+DEFINE_SHOW_ATTRIBUTE(send_queue_priority_update_flow);
+DEFINE_SHOW_ATTRIBUTE(comp_eq_overrun);
+DEFINE_SHOW_ATTRIBUTE(async_eq_overrun);
+DEFINE_SHOW_ATTRIBUTE(cq_overrun);
+DEFINE_SHOW_ATTRIBUTE(invalid_command);
+DEFINE_SHOW_ATTRIBUTE(quota_exceeded_command);
+
+void mlx5_esw_vport_debugfs_destroy(struct mlx5_eswitch *esw, u16 vport_num)
+{
+	struct mlx5_vport *vport = mlx5_eswitch_get_vport(esw, vport_num);
+
+	debugfs_remove_recursive(vport->dbgfs);
+	vport->dbgfs = NULL;
+}
+
+/* vnic diag dir name is "pf", "ecpf" or "{vf/sf}_xxxx" */
+#define VNIC_DIAG_DIR_NAME_MAX_LEN 8
+
+void mlx5_esw_vport_debugfs_create(struct mlx5_eswitch *esw, u16 vport_num, bool is_sf, u16 sf_num)
+{
+	struct mlx5_vport *vport = mlx5_eswitch_get_vport(esw, vport_num);
+	struct dentry *vnic_diag;
+	char dir_name[VNIC_DIAG_DIR_NAME_MAX_LEN];
+	int err;
+
+	if (!MLX5_CAP_GEN(esw->dev, vport_group_manager))
+		return;
+
+	if (vport_num == MLX5_VPORT_PF) {
+		strcpy(dir_name, "pf");
+	} else if (vport_num == MLX5_VPORT_ECPF) {
+		strcpy(dir_name, "ecpf");
+	} else {
+		err = snprintf(dir_name, VNIC_DIAG_DIR_NAME_MAX_LEN, "%s_%d", is_sf ? "sf" : "vf",
+			       is_sf ? sf_num : vport_num - MLX5_VPORT_FIRST_VF);
+		if (WARN_ON(err < 0))
+			return;
+	}
+
+	vport->dbgfs = debugfs_create_dir(dir_name, esw->dbgfs);
+	vnic_diag = debugfs_create_dir("vnic_diag", vport->dbgfs);
+
+	if (MLX5_CAP_GEN(esw->dev, vnic_env_queue_counters)) {
+		debugfs_create_file("total_q_under_processor_handle", 0444, vnic_diag, vport,
+				    &total_q_under_processor_handle_fops);
+		debugfs_create_file("send_queue_priority_update_flow", 0444, vnic_diag, vport,
+				    &send_queue_priority_update_flow_fops);
+	}
+
+	if (MLX5_CAP_GEN(esw->dev, eq_overrun_count)) {
+		debugfs_create_file("comp_eq_overrun", 0444, vnic_diag, vport,
+				    &comp_eq_overrun_fops);
+		debugfs_create_file("async_eq_overrun", 0444, vnic_diag, vport,
+				    &async_eq_overrun_fops);
+	}
+
+	if (MLX5_CAP_GEN(esw->dev, vnic_env_cq_overrun))
+		debugfs_create_file("cq_overrun", 0444, vnic_diag, vport, &cq_overrun_fops);
+
+	if (MLX5_CAP_GEN(esw->dev, invalid_command_count))
+		debugfs_create_file("invalid_command", 0444, vnic_diag, vport,
+				    &invalid_command_fops);
+
+	if (MLX5_CAP_GEN(esw->dev, quota_exceeded_count))
+		debugfs_create_file("quota_exceeded_command", 0444, vnic_diag, vport,
+				    &quota_exceeded_command_fops);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/esw/devlink_port.c b/drivers/net/ethernet/mellanox/mlx5/core/esw/devlink_port.c
new file mode 100644
index 000000000..9bc7be95d
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/esw/devlink_port.c
@@ -0,0 +1,190 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2020 Mellanox Technologies Ltd. */
+
+#include <linux/mlx5/driver.h>
+#include "eswitch.h"
+
+static void
+mlx5_esw_get_port_parent_id(struct mlx5_core_dev *dev, struct netdev_phys_item_id *ppid)
+{
+	u64 parent_id;
+
+	parent_id = mlx5_query_nic_system_image_guid(dev);
+	ppid->id_len = sizeof(parent_id);
+	memcpy(ppid->id, &parent_id, sizeof(parent_id));
+}
+
+static bool mlx5_esw_devlink_port_supported(struct mlx5_eswitch *esw, u16 vport_num)
+{
+	return vport_num == MLX5_VPORT_UPLINK ||
+	       (mlx5_core_is_ecpf(esw->dev) && vport_num == MLX5_VPORT_PF) ||
+	       mlx5_eswitch_is_vf_vport(esw, vport_num);
+}
+
+static struct devlink_port *mlx5_esw_dl_port_alloc(struct mlx5_eswitch *esw, u16 vport_num)
+{
+	struct mlx5_core_dev *dev = esw->dev;
+	struct devlink_port_attrs attrs = {};
+	struct netdev_phys_item_id ppid = {};
+	struct devlink_port *dl_port;
+	u32 controller_num = 0;
+	bool external;
+	u16 pfnum;
+
+	dl_port = kzalloc(sizeof(*dl_port), GFP_KERNEL);
+	if (!dl_port)
+		return NULL;
+
+	mlx5_esw_get_port_parent_id(dev, &ppid);
+	pfnum = mlx5_get_dev_index(dev);
+	external = mlx5_core_is_ecpf_esw_manager(dev);
+	if (external)
+		controller_num = dev->priv.eswitch->offloads.host_number + 1;
+
+	if (vport_num == MLX5_VPORT_UPLINK) {
+		attrs.flavour = DEVLINK_PORT_FLAVOUR_PHYSICAL;
+		attrs.phys.port_number = pfnum;
+		memcpy(attrs.switch_id.id, ppid.id, ppid.id_len);
+		attrs.switch_id.id_len = ppid.id_len;
+		devlink_port_attrs_set(dl_port, &attrs);
+	} else if (vport_num == MLX5_VPORT_PF) {
+		memcpy(dl_port->attrs.switch_id.id, ppid.id, ppid.id_len);
+		dl_port->attrs.switch_id.id_len = ppid.id_len;
+		devlink_port_attrs_pci_pf_set(dl_port, controller_num, pfnum, external);
+	} else if (mlx5_eswitch_is_vf_vport(esw, vport_num)) {
+		memcpy(dl_port->attrs.switch_id.id, ppid.id, ppid.id_len);
+		dl_port->attrs.switch_id.id_len = ppid.id_len;
+		devlink_port_attrs_pci_vf_set(dl_port, controller_num, pfnum,
+					      vport_num - 1, external);
+	}
+	return dl_port;
+}
+
+static void mlx5_esw_dl_port_free(struct devlink_port *dl_port)
+{
+	kfree(dl_port);
+}
+
+int mlx5_esw_offloads_devlink_port_register(struct mlx5_eswitch *esw, u16 vport_num)
+{
+	struct mlx5_core_dev *dev = esw->dev;
+	struct devlink_port *dl_port;
+	unsigned int dl_port_index;
+	struct mlx5_vport *vport;
+	struct devlink *devlink;
+	int err;
+
+	if (!mlx5_esw_devlink_port_supported(esw, vport_num))
+		return 0;
+
+	vport = mlx5_eswitch_get_vport(esw, vport_num);
+	if (IS_ERR(vport))
+		return PTR_ERR(vport);
+
+	dl_port = mlx5_esw_dl_port_alloc(esw, vport_num);
+	if (!dl_port)
+		return -ENOMEM;
+
+	devlink = priv_to_devlink(dev);
+	dl_port_index = mlx5_esw_vport_to_devlink_port_index(dev, vport_num);
+	err = devl_port_register(devlink, dl_port, dl_port_index);
+	if (err)
+		goto reg_err;
+
+	err = devl_rate_leaf_create(dl_port, vport);
+	if (err)
+		goto rate_err;
+
+	vport->dl_port = dl_port;
+	return 0;
+
+rate_err:
+	devl_port_unregister(dl_port);
+reg_err:
+	mlx5_esw_dl_port_free(dl_port);
+	return err;
+}
+
+void mlx5_esw_offloads_devlink_port_unregister(struct mlx5_eswitch *esw, u16 vport_num)
+{
+	struct mlx5_vport *vport;
+
+	if (!mlx5_esw_devlink_port_supported(esw, vport_num))
+		return;
+
+	vport = mlx5_eswitch_get_vport(esw, vport_num);
+	if (IS_ERR(vport))
+		return;
+
+	if (vport->dl_port->devlink_rate) {
+		mlx5_esw_qos_vport_update_group(esw, vport, NULL, NULL);
+		devl_rate_leaf_destroy(vport->dl_port);
+	}
+
+	devl_port_unregister(vport->dl_port);
+	mlx5_esw_dl_port_free(vport->dl_port);
+	vport->dl_port = NULL;
+}
+
+struct devlink_port *mlx5_esw_offloads_devlink_port(struct mlx5_eswitch *esw, u16 vport_num)
+{
+	struct mlx5_vport *vport;
+
+	vport = mlx5_eswitch_get_vport(esw, vport_num);
+	return IS_ERR(vport) ? ERR_CAST(vport) : vport->dl_port;
+}
+
+int mlx5_esw_devlink_sf_port_register(struct mlx5_eswitch *esw, struct devlink_port *dl_port,
+				      u16 vport_num, u32 controller, u32 sfnum)
+{
+	struct mlx5_core_dev *dev = esw->dev;
+	struct netdev_phys_item_id ppid = {};
+	unsigned int dl_port_index;
+	struct mlx5_vport *vport;
+	struct devlink *devlink;
+	u16 pfnum;
+	int err;
+
+	vport = mlx5_eswitch_get_vport(esw, vport_num);
+	if (IS_ERR(vport))
+		return PTR_ERR(vport);
+
+	pfnum = mlx5_get_dev_index(dev);
+	mlx5_esw_get_port_parent_id(dev, &ppid);
+	memcpy(dl_port->attrs.switch_id.id, &ppid.id[0], ppid.id_len);
+	dl_port->attrs.switch_id.id_len = ppid.id_len;
+	devlink_port_attrs_pci_sf_set(dl_port, controller, pfnum, sfnum, !!controller);
+	devlink = priv_to_devlink(dev);
+	dl_port_index = mlx5_esw_vport_to_devlink_port_index(dev, vport_num);
+	err = devl_port_register(devlink, dl_port, dl_port_index);
+	if (err)
+		return err;
+
+	err = devl_rate_leaf_create(dl_port, vport);
+	if (err)
+		goto rate_err;
+
+	vport->dl_port = dl_port;
+	return 0;
+
+rate_err:
+	devl_port_unregister(dl_port);
+	return err;
+}
+
+void mlx5_esw_devlink_sf_port_unregister(struct mlx5_eswitch *esw, u16 vport_num)
+{
+	struct mlx5_vport *vport;
+
+	vport = mlx5_eswitch_get_vport(esw, vport_num);
+	if (IS_ERR(vport))
+		return;
+
+	if (vport->dl_port->devlink_rate) {
+		mlx5_esw_qos_vport_update_group(esw, vport, NULL, NULL);
+		devl_rate_leaf_destroy(vport->dl_port);
+	}
+
+	devl_port_unregister(vport->dl_port);
+	vport->dl_port = NULL;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/esw/diag/bridge_tracepoint.h b/drivers/net/ethernet/mellanox/mlx5/core/esw/diag/bridge_tracepoint.h
new file mode 100644
index 000000000..51ac24e6e
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/esw/diag/bridge_tracepoint.h
@@ -0,0 +1,120 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2021 Mellanox Technologies. */
+
+#undef TRACE_SYSTEM
+#define TRACE_SYSTEM mlx5
+
+#if !defined(_MLX5_ESW_BRIDGE_TRACEPOINT_) || defined(TRACE_HEADER_MULTI_READ)
+#define _MLX5_ESW_BRIDGE_TRACEPOINT_
+
+#include <linux/tracepoint.h>
+#include "../bridge_priv.h"
+
+DECLARE_EVENT_CLASS(mlx5_esw_bridge_fdb_template,
+		    TP_PROTO(const struct mlx5_esw_bridge_fdb_entry *fdb),
+		    TP_ARGS(fdb),
+		    TP_STRUCT__entry(
+			    __array(char, dev_name, IFNAMSIZ)
+			    __array(unsigned char, addr, ETH_ALEN)
+			    __field(u16, vid)
+			    __field(u16, flags)
+			    __field(unsigned int, used)
+			    ),
+		    TP_fast_assign(
+			    strscpy(__entry->dev_name,
+				    netdev_name(fdb->dev),
+				    IFNAMSIZ);
+			    memcpy(__entry->addr, fdb->key.addr, ETH_ALEN);
+			    __entry->vid = fdb->key.vid;
+			    __entry->flags = fdb->flags;
+			    __entry->used = jiffies_to_msecs(jiffies - fdb->lastuse)
+			    ),
+		    TP_printk("net_device=%s addr=%pM vid=%hu flags=%hx used=%u",
+			      __entry->dev_name,
+			      __entry->addr,
+			      __entry->vid,
+			      __entry->flags,
+			      __entry->used / 1000)
+	);
+
+DEFINE_EVENT(mlx5_esw_bridge_fdb_template,
+	     mlx5_esw_bridge_fdb_entry_init,
+	     TP_PROTO(const struct mlx5_esw_bridge_fdb_entry *fdb),
+	     TP_ARGS(fdb)
+	);
+DEFINE_EVENT(mlx5_esw_bridge_fdb_template,
+	     mlx5_esw_bridge_fdb_entry_refresh,
+	     TP_PROTO(const struct mlx5_esw_bridge_fdb_entry *fdb),
+	     TP_ARGS(fdb)
+	);
+DEFINE_EVENT(mlx5_esw_bridge_fdb_template,
+	     mlx5_esw_bridge_fdb_entry_cleanup,
+	     TP_PROTO(const struct mlx5_esw_bridge_fdb_entry *fdb),
+	     TP_ARGS(fdb)
+	);
+
+DECLARE_EVENT_CLASS(mlx5_esw_bridge_vlan_template,
+		    TP_PROTO(const struct mlx5_esw_bridge_vlan *vlan),
+		    TP_ARGS(vlan),
+		    TP_STRUCT__entry(
+			    __field(u16, vid)
+			    __field(u16, flags)
+			    ),
+		    TP_fast_assign(
+			    __entry->vid = vlan->vid;
+			    __entry->flags = vlan->flags;
+			    ),
+		    TP_printk("vid=%hu flags=%hx",
+			      __entry->vid,
+			      __entry->flags)
+	);
+
+DEFINE_EVENT(mlx5_esw_bridge_vlan_template,
+	     mlx5_esw_bridge_vlan_create,
+	     TP_PROTO(const struct mlx5_esw_bridge_vlan *vlan),
+	     TP_ARGS(vlan)
+	);
+DEFINE_EVENT(mlx5_esw_bridge_vlan_template,
+	     mlx5_esw_bridge_vlan_cleanup,
+	     TP_PROTO(const struct mlx5_esw_bridge_vlan *vlan),
+	     TP_ARGS(vlan)
+	);
+
+DECLARE_EVENT_CLASS(mlx5_esw_bridge_port_template,
+		    TP_PROTO(const struct mlx5_esw_bridge_port *port),
+		    TP_ARGS(port),
+		    TP_STRUCT__entry(
+			    __field(u16, vport_num)
+			    __field(u16, esw_owner_vhca_id)
+			    __field(u16, flags)
+			    ),
+		    TP_fast_assign(
+			    __entry->vport_num = port->vport_num;
+			    __entry->esw_owner_vhca_id = port->esw_owner_vhca_id;
+			    __entry->flags = port->flags;
+			    ),
+		    TP_printk("vport_num=%hu esw_owner_vhca_id=%hu flags=%hx",
+			      __entry->vport_num,
+			      __entry->esw_owner_vhca_id,
+			      __entry->flags)
+	);
+
+DEFINE_EVENT(mlx5_esw_bridge_port_template,
+	     mlx5_esw_bridge_vport_init,
+	     TP_PROTO(const struct mlx5_esw_bridge_port *port),
+	     TP_ARGS(port)
+	);
+DEFINE_EVENT(mlx5_esw_bridge_port_template,
+	     mlx5_esw_bridge_vport_cleanup,
+	     TP_PROTO(const struct mlx5_esw_bridge_port *port),
+	     TP_ARGS(port)
+	);
+
+#endif
+
+/* This part must be outside protection */
+#undef TRACE_INCLUDE_PATH
+#define TRACE_INCLUDE_PATH esw/diag
+#undef TRACE_INCLUDE_FILE
+#define TRACE_INCLUDE_FILE bridge_tracepoint
+#include <trace/define_trace.h>
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/esw/diag/qos_tracepoint.h b/drivers/net/ethernet/mellanox/mlx5/core/esw/diag/qos_tracepoint.h
new file mode 100644
index 000000000..458baf0c6
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/esw/diag/qos_tracepoint.h
@@ -0,0 +1,123 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. All rights reserved. */
+
+#undef TRACE_SYSTEM
+#define TRACE_SYSTEM mlx5
+
+#if !defined(_MLX5_ESW_TP_) || defined(TRACE_HEADER_MULTI_READ)
+#define _MLX5_ESW_TP_
+
+#include <linux/tracepoint.h>
+#include "eswitch.h"
+
+TRACE_EVENT(mlx5_esw_vport_qos_destroy,
+	    TP_PROTO(const struct mlx5_vport *vport),
+	    TP_ARGS(vport),
+	    TP_STRUCT__entry(__string(devname, dev_name(vport->dev->device))
+			     __field(unsigned short, vport_id)
+			     __field(unsigned int,   tsar_ix)
+			     ),
+	    TP_fast_assign(__assign_str(devname, dev_name(vport->dev->device));
+		    __entry->vport_id = vport->vport;
+		    __entry->tsar_ix = vport->qos.esw_tsar_ix;
+	    ),
+	    TP_printk("(%s) vport=%hu tsar_ix=%u\n",
+		      __get_str(devname), __entry->vport_id, __entry->tsar_ix
+		      )
+);
+
+DECLARE_EVENT_CLASS(mlx5_esw_vport_qos_template,
+		    TP_PROTO(const struct mlx5_vport *vport, u32 bw_share, u32 max_rate),
+		    TP_ARGS(vport, bw_share, max_rate),
+		    TP_STRUCT__entry(__string(devname, dev_name(vport->dev->device))
+				     __field(unsigned short, vport_id)
+				     __field(unsigned int, tsar_ix)
+				     __field(unsigned int, bw_share)
+				     __field(unsigned int, max_rate)
+				     __field(void *, group)
+				     ),
+		    TP_fast_assign(__assign_str(devname, dev_name(vport->dev->device));
+			    __entry->vport_id = vport->vport;
+			    __entry->tsar_ix = vport->qos.esw_tsar_ix;
+			    __entry->bw_share = bw_share;
+			    __entry->max_rate = max_rate;
+			    __entry->group = vport->qos.group;
+		    ),
+		    TP_printk("(%s) vport=%hu tsar_ix=%u bw_share=%u, max_rate=%u group=%p\n",
+			      __get_str(devname), __entry->vport_id, __entry->tsar_ix,
+			      __entry->bw_share, __entry->max_rate, __entry->group
+			      )
+);
+
+DEFINE_EVENT(mlx5_esw_vport_qos_template, mlx5_esw_vport_qos_create,
+	     TP_PROTO(const struct mlx5_vport *vport, u32 bw_share, u32 max_rate),
+	     TP_ARGS(vport, bw_share, max_rate)
+	     );
+
+DEFINE_EVENT(mlx5_esw_vport_qos_template, mlx5_esw_vport_qos_config,
+	     TP_PROTO(const struct mlx5_vport *vport, u32 bw_share, u32 max_rate),
+	     TP_ARGS(vport, bw_share, max_rate)
+	     );
+
+DECLARE_EVENT_CLASS(mlx5_esw_group_qos_template,
+		    TP_PROTO(const struct mlx5_core_dev *dev,
+			     const struct mlx5_esw_rate_group *group,
+			     unsigned int tsar_ix),
+		    TP_ARGS(dev, group, tsar_ix),
+		    TP_STRUCT__entry(__string(devname, dev_name(dev->device))
+				     __field(const void *, group)
+				     __field(unsigned int, tsar_ix)
+				     ),
+		    TP_fast_assign(__assign_str(devname, dev_name(dev->device));
+			    __entry->group = group;
+			    __entry->tsar_ix = tsar_ix;
+		    ),
+		    TP_printk("(%s) group=%p tsar_ix=%u\n",
+			      __get_str(devname), __entry->group, __entry->tsar_ix
+			      )
+);
+
+DEFINE_EVENT(mlx5_esw_group_qos_template, mlx5_esw_group_qos_create,
+	     TP_PROTO(const struct mlx5_core_dev *dev,
+		      const struct mlx5_esw_rate_group *group,
+		      unsigned int tsar_ix),
+	     TP_ARGS(dev, group, tsar_ix)
+	     );
+
+DEFINE_EVENT(mlx5_esw_group_qos_template, mlx5_esw_group_qos_destroy,
+	     TP_PROTO(const struct mlx5_core_dev *dev,
+		      const struct mlx5_esw_rate_group *group,
+		      unsigned int tsar_ix),
+	     TP_ARGS(dev, group, tsar_ix)
+	     );
+
+TRACE_EVENT(mlx5_esw_group_qos_config,
+	    TP_PROTO(const struct mlx5_core_dev *dev,
+		     const struct mlx5_esw_rate_group *group,
+		     unsigned int tsar_ix, u32 bw_share, u32 max_rate),
+	    TP_ARGS(dev, group, tsar_ix, bw_share, max_rate),
+	    TP_STRUCT__entry(__string(devname, dev_name(dev->device))
+			     __field(const void *, group)
+			     __field(unsigned int, tsar_ix)
+			     __field(unsigned int, bw_share)
+			     __field(unsigned int, max_rate)
+			     ),
+	    TP_fast_assign(__assign_str(devname, dev_name(dev->device));
+		    __entry->group = group;
+		    __entry->tsar_ix = tsar_ix;
+		    __entry->bw_share = bw_share;
+		    __entry->max_rate = max_rate;
+	    ),
+	    TP_printk("(%s) group=%p tsar_ix=%u bw_share=%u max_rate=%u\n",
+		      __get_str(devname), __entry->group, __entry->tsar_ix,
+		      __entry->bw_share, __entry->max_rate
+		      )
+);
+#endif /* _MLX5_ESW_TP_ */
+
+/* This part must be outside protection */
+#undef TRACE_INCLUDE_PATH
+#define TRACE_INCLUDE_PATH esw/diag
+#undef TRACE_INCLUDE_FILE
+#define TRACE_INCLUDE_FILE qos_tracepoint
+#include <trace/define_trace.h>
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/esw/indir_table.c b/drivers/net/ethernet/mellanox/mlx5/core/esw/indir_table.c
new file mode 100644
index 000000000..8a94870c5
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/esw/indir_table.c
@@ -0,0 +1,390 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2021 Mellanox Technologies. */
+
+#include <linux/etherdevice.h>
+#include <linux/idr.h>
+#include <linux/mlx5/driver.h>
+#include <linux/mlx5/mlx5_ifc.h>
+#include <linux/mlx5/vport.h>
+#include <linux/mlx5/fs.h>
+#include "mlx5_core.h"
+#include "eswitch.h"
+#include "en.h"
+#include "en_tc.h"
+#include "fs_core.h"
+#include "esw/indir_table.h"
+#include "lib/fs_chains.h"
+#include "en/mod_hdr.h"
+
+#define MLX5_ESW_INDIR_TABLE_SIZE 2
+#define MLX5_ESW_INDIR_TABLE_RECIRC_IDX (MLX5_ESW_INDIR_TABLE_SIZE - 2)
+#define MLX5_ESW_INDIR_TABLE_FWD_IDX (MLX5_ESW_INDIR_TABLE_SIZE - 1)
+
+struct mlx5_esw_indir_table_rule {
+	struct mlx5_flow_handle *handle;
+	struct mlx5_modify_hdr *mh;
+	refcount_t refcnt;
+};
+
+struct mlx5_esw_indir_table_entry {
+	struct hlist_node hlist;
+	struct mlx5_flow_table *ft;
+	struct mlx5_flow_group *recirc_grp;
+	struct mlx5_flow_group *fwd_grp;
+	struct mlx5_flow_handle *fwd_rule;
+	struct mlx5_esw_indir_table_rule *recirc_rule;
+	int fwd_ref;
+
+	u16 vport;
+};
+
+struct mlx5_esw_indir_table {
+	struct mutex lock; /* protects table */
+	DECLARE_HASHTABLE(table, 8);
+};
+
+struct mlx5_esw_indir_table *
+mlx5_esw_indir_table_init(void)
+{
+	struct mlx5_esw_indir_table *indir = kvzalloc(sizeof(*indir), GFP_KERNEL);
+
+	if (!indir)
+		return ERR_PTR(-ENOMEM);
+
+	mutex_init(&indir->lock);
+	hash_init(indir->table);
+	return indir;
+}
+
+void
+mlx5_esw_indir_table_destroy(struct mlx5_esw_indir_table *indir)
+{
+	mutex_destroy(&indir->lock);
+	kvfree(indir);
+}
+
+bool
+mlx5_esw_indir_table_needed(struct mlx5_eswitch *esw,
+			    struct mlx5_flow_attr *attr,
+			    u16 vport_num,
+			    struct mlx5_core_dev *dest_mdev)
+{
+	struct mlx5_esw_flow_attr *esw_attr = attr->esw_attr;
+	bool vf_sf_vport;
+
+	vf_sf_vport = mlx5_eswitch_is_vf_vport(esw, vport_num) ||
+		      mlx5_esw_is_sf_vport(esw, vport_num);
+
+	/* Use indirect table for all IP traffic from UL to VF with vport
+	 * destination when source rewrite flag is set.
+	 */
+	return esw_attr->in_rep->vport == MLX5_VPORT_UPLINK &&
+		vf_sf_vport &&
+		esw->dev == dest_mdev &&
+		attr->flags & MLX5_ATTR_FLAG_SRC_REWRITE;
+}
+
+u16
+mlx5_esw_indir_table_decap_vport(struct mlx5_flow_attr *attr)
+{
+	struct mlx5_esw_flow_attr *esw_attr = attr->esw_attr;
+
+	return esw_attr->rx_tun_attr ? esw_attr->rx_tun_attr->decap_vport : 0;
+}
+
+static int mlx5_esw_indir_table_rule_get(struct mlx5_eswitch *esw,
+					 struct mlx5_flow_attr *attr,
+					 struct mlx5_esw_indir_table_entry *e)
+{
+	struct mlx5_esw_flow_attr *esw_attr = attr->esw_attr;
+	struct mlx5_fs_chains *chains = esw_chains(esw);
+	struct mlx5e_tc_mod_hdr_acts mod_acts = {};
+	struct mlx5_flow_destination dest = {};
+	struct mlx5_esw_indir_table_rule *rule;
+	struct mlx5_flow_act flow_act = {};
+	struct mlx5_flow_handle *handle;
+	int err = 0;
+	u32 data;
+
+	if (e->recirc_rule) {
+		refcount_inc(&e->recirc_rule->refcnt);
+		return 0;
+	}
+
+	rule = kzalloc(sizeof(*rule), GFP_KERNEL);
+	if (!rule)
+		return -ENOMEM;
+
+	/* Modify flow source to recirculate packet */
+	data = mlx5_eswitch_get_vport_metadata_for_set(esw, esw_attr->rx_tun_attr->decap_vport);
+	err = mlx5e_tc_match_to_reg_set(esw->dev, &mod_acts, MLX5_FLOW_NAMESPACE_FDB,
+					VPORT_TO_REG, data);
+	if (err)
+		goto err_mod_hdr_regc0;
+
+	err = mlx5e_tc_match_to_reg_set(esw->dev, &mod_acts, MLX5_FLOW_NAMESPACE_FDB,
+					TUNNEL_TO_REG, ESW_TUN_SLOW_TABLE_GOTO_VPORT);
+	if (err)
+		goto err_mod_hdr_regc1;
+
+	flow_act.modify_hdr = mlx5_modify_header_alloc(esw->dev, MLX5_FLOW_NAMESPACE_FDB,
+						       mod_acts.num_actions, mod_acts.actions);
+	if (IS_ERR(flow_act.modify_hdr)) {
+		err = PTR_ERR(flow_act.modify_hdr);
+		goto err_mod_hdr_alloc;
+	}
+
+	flow_act.action = MLX5_FLOW_CONTEXT_ACTION_FWD_DEST | MLX5_FLOW_CONTEXT_ACTION_MOD_HDR;
+	flow_act.flags = FLOW_ACT_IGNORE_FLOW_LEVEL | FLOW_ACT_NO_APPEND;
+	flow_act.fg = e->recirc_grp;
+	dest.type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
+	dest.ft = mlx5_chains_get_table(chains, 0, 1, 0);
+	if (IS_ERR(dest.ft)) {
+		err = PTR_ERR(dest.ft);
+		goto err_table;
+	}
+	handle = mlx5_add_flow_rules(e->ft, NULL, &flow_act, &dest, 1);
+	if (IS_ERR(handle)) {
+		err = PTR_ERR(handle);
+		goto err_handle;
+	}
+
+	mlx5e_mod_hdr_dealloc(&mod_acts);
+	rule->handle = handle;
+	rule->mh = flow_act.modify_hdr;
+	refcount_set(&rule->refcnt, 1);
+	e->recirc_rule = rule;
+	return 0;
+
+err_handle:
+	mlx5_chains_put_table(chains, 0, 1, 0);
+err_table:
+	mlx5_modify_header_dealloc(esw->dev, flow_act.modify_hdr);
+err_mod_hdr_alloc:
+err_mod_hdr_regc1:
+	mlx5e_mod_hdr_dealloc(&mod_acts);
+err_mod_hdr_regc0:
+	kfree(rule);
+	return err;
+}
+
+static void mlx5_esw_indir_table_rule_put(struct mlx5_eswitch *esw,
+					  struct mlx5_esw_indir_table_entry *e)
+{
+	struct mlx5_esw_indir_table_rule *rule = e->recirc_rule;
+	struct mlx5_fs_chains *chains = esw_chains(esw);
+
+	if (!rule)
+		return;
+
+	if (!refcount_dec_and_test(&rule->refcnt))
+		return;
+
+	mlx5_del_flow_rules(rule->handle);
+	mlx5_chains_put_table(chains, 0, 1, 0);
+	mlx5_modify_header_dealloc(esw->dev, rule->mh);
+	kfree(rule);
+	e->recirc_rule = NULL;
+}
+
+static int mlx5_create_indir_recirc_group(struct mlx5_esw_indir_table_entry *e)
+{
+	int err = 0, inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
+	u32 *in;
+
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in)
+		return -ENOMEM;
+
+	MLX5_SET(create_flow_group_in, in, start_flow_index, 0);
+	MLX5_SET(create_flow_group_in, in, end_flow_index, MLX5_ESW_INDIR_TABLE_RECIRC_IDX);
+	e->recirc_grp = mlx5_create_flow_group(e->ft, in);
+	if (IS_ERR(e->recirc_grp))
+		err = PTR_ERR(e->recirc_grp);
+
+	kvfree(in);
+	return err;
+}
+
+static int mlx5_create_indir_fwd_group(struct mlx5_eswitch *esw,
+				       struct mlx5_esw_indir_table_entry *e)
+{
+	int err = 0, inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
+	struct mlx5_flow_destination dest = {};
+	struct mlx5_flow_act flow_act = {};
+	struct mlx5_flow_spec *spec;
+	u32 *in;
+
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in)
+		return -ENOMEM;
+
+	spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
+	if (!spec) {
+		kvfree(in);
+		return -ENOMEM;
+	}
+
+	/* Hold one entry */
+	MLX5_SET(create_flow_group_in, in, start_flow_index, MLX5_ESW_INDIR_TABLE_FWD_IDX);
+	MLX5_SET(create_flow_group_in, in, end_flow_index, MLX5_ESW_INDIR_TABLE_FWD_IDX);
+	e->fwd_grp = mlx5_create_flow_group(e->ft, in);
+	if (IS_ERR(e->fwd_grp)) {
+		err = PTR_ERR(e->fwd_grp);
+		goto err_out;
+	}
+
+	flow_act.action = MLX5_FLOW_CONTEXT_ACTION_FWD_DEST;
+	flow_act.fg = e->fwd_grp;
+	dest.type = MLX5_FLOW_DESTINATION_TYPE_VPORT;
+	dest.vport.num = e->vport;
+	dest.vport.vhca_id = MLX5_CAP_GEN(esw->dev, vhca_id);
+	dest.vport.flags = MLX5_FLOW_DEST_VPORT_VHCA_ID;
+	e->fwd_rule = mlx5_add_flow_rules(e->ft, spec, &flow_act, &dest, 1);
+	if (IS_ERR(e->fwd_rule)) {
+		mlx5_destroy_flow_group(e->fwd_grp);
+		err = PTR_ERR(e->fwd_rule);
+	}
+
+err_out:
+	kvfree(spec);
+	kvfree(in);
+	return err;
+}
+
+static struct mlx5_esw_indir_table_entry *
+mlx5_esw_indir_table_entry_create(struct mlx5_eswitch *esw, struct mlx5_flow_attr *attr,
+				  u16 vport, bool decap)
+{
+	struct mlx5_flow_table_attr ft_attr = {};
+	struct mlx5_flow_namespace *root_ns;
+	struct mlx5_esw_indir_table_entry *e;
+	struct mlx5_flow_table *ft;
+	int err = 0;
+
+	root_ns = mlx5_get_flow_namespace(esw->dev, MLX5_FLOW_NAMESPACE_FDB);
+	if (!root_ns)
+		return ERR_PTR(-ENOENT);
+
+	e = kzalloc(sizeof(*e), GFP_KERNEL);
+	if (!e)
+		return ERR_PTR(-ENOMEM);
+
+	ft_attr.prio = FDB_TC_OFFLOAD;
+	ft_attr.max_fte = MLX5_ESW_INDIR_TABLE_SIZE;
+	ft_attr.flags = MLX5_FLOW_TABLE_UNMANAGED;
+	ft_attr.level = 1;
+
+	ft = mlx5_create_flow_table(root_ns, &ft_attr);
+	if (IS_ERR(ft)) {
+		err = PTR_ERR(ft);
+		goto tbl_err;
+	}
+	e->ft = ft;
+	e->vport = vport;
+	e->fwd_ref = !decap;
+
+	err = mlx5_create_indir_recirc_group(e);
+	if (err)
+		goto recirc_grp_err;
+
+	if (decap) {
+		err = mlx5_esw_indir_table_rule_get(esw, attr, e);
+		if (err)
+			goto recirc_rule_err;
+	}
+
+	err = mlx5_create_indir_fwd_group(esw, e);
+	if (err)
+		goto fwd_grp_err;
+
+	hash_add(esw->fdb_table.offloads.indir->table, &e->hlist,
+		 vport << 16);
+
+	return e;
+
+fwd_grp_err:
+	if (decap)
+		mlx5_esw_indir_table_rule_put(esw, e);
+recirc_rule_err:
+	mlx5_destroy_flow_group(e->recirc_grp);
+recirc_grp_err:
+	mlx5_destroy_flow_table(e->ft);
+tbl_err:
+	kfree(e);
+	return ERR_PTR(err);
+}
+
+static struct mlx5_esw_indir_table_entry *
+mlx5_esw_indir_table_entry_lookup(struct mlx5_eswitch *esw, u16 vport)
+{
+	struct mlx5_esw_indir_table_entry *e;
+	u32 key = vport << 16;
+
+	hash_for_each_possible(esw->fdb_table.offloads.indir->table, e, hlist, key)
+		if (e->vport == vport)
+			return e;
+
+	return NULL;
+}
+
+struct mlx5_flow_table *mlx5_esw_indir_table_get(struct mlx5_eswitch *esw,
+						 struct mlx5_flow_attr *attr,
+						 u16 vport, bool decap)
+{
+	struct mlx5_esw_indir_table_entry *e;
+	int err;
+
+	mutex_lock(&esw->fdb_table.offloads.indir->lock);
+	e = mlx5_esw_indir_table_entry_lookup(esw, vport);
+	if (e) {
+		if (!decap) {
+			e->fwd_ref++;
+		} else {
+			err = mlx5_esw_indir_table_rule_get(esw, attr, e);
+			if (err)
+				goto out_err;
+		}
+	} else {
+		e = mlx5_esw_indir_table_entry_create(esw, attr, vport, decap);
+		if (IS_ERR(e)) {
+			err = PTR_ERR(e);
+			esw_warn(esw->dev, "Failed to create indirection table, err %d.\n", err);
+			goto out_err;
+		}
+	}
+	mutex_unlock(&esw->fdb_table.offloads.indir->lock);
+	return e->ft;
+
+out_err:
+	mutex_unlock(&esw->fdb_table.offloads.indir->lock);
+	return ERR_PTR(err);
+}
+
+void mlx5_esw_indir_table_put(struct mlx5_eswitch *esw,
+			      u16 vport, bool decap)
+{
+	struct mlx5_esw_indir_table_entry *e;
+
+	mutex_lock(&esw->fdb_table.offloads.indir->lock);
+	e = mlx5_esw_indir_table_entry_lookup(esw, vport);
+	if (!e)
+		goto out;
+
+	if (!decap)
+		e->fwd_ref--;
+	else
+		mlx5_esw_indir_table_rule_put(esw, e);
+
+	if (e->fwd_ref || e->recirc_rule)
+		goto out;
+
+	hash_del(&e->hlist);
+	mlx5_destroy_flow_group(e->recirc_grp);
+	mlx5_del_flow_rules(e->fwd_rule);
+	mlx5_destroy_flow_group(e->fwd_grp);
+	mlx5_destroy_flow_table(e->ft);
+	kfree(e);
+out:
+	mutex_unlock(&esw->fdb_table.offloads.indir->lock);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/esw/indir_table.h b/drivers/net/ethernet/mellanox/mlx5/core/esw/indir_table.h
new file mode 100644
index 000000000..036f5b3a3
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/esw/indir_table.h
@@ -0,0 +1,72 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2021 Mellanox Technologies. */
+
+#ifndef __MLX5_ESW_FT_H__
+#define __MLX5_ESW_FT_H__
+
+#ifdef CONFIG_MLX5_CLS_ACT
+
+struct mlx5_esw_indir_table *
+mlx5_esw_indir_table_init(void);
+void
+mlx5_esw_indir_table_destroy(struct mlx5_esw_indir_table *indir);
+
+struct mlx5_flow_table *mlx5_esw_indir_table_get(struct mlx5_eswitch *esw,
+						 struct mlx5_flow_attr *attr,
+						 u16 vport, bool decap);
+void mlx5_esw_indir_table_put(struct mlx5_eswitch *esw,
+			      u16 vport, bool decap);
+
+bool
+mlx5_esw_indir_table_needed(struct mlx5_eswitch *esw,
+			    struct mlx5_flow_attr *attr,
+			    u16 vport_num,
+			    struct mlx5_core_dev *dest_mdev);
+
+u16
+mlx5_esw_indir_table_decap_vport(struct mlx5_flow_attr *attr);
+
+#else
+/* indir API stubs */
+static inline struct mlx5_esw_indir_table *
+mlx5_esw_indir_table_init(void)
+{
+	return NULL;
+}
+
+static inline void
+mlx5_esw_indir_table_destroy(struct mlx5_esw_indir_table *indir)
+{
+}
+
+static inline struct mlx5_flow_table *
+mlx5_esw_indir_table_get(struct mlx5_eswitch *esw,
+			 struct mlx5_flow_attr *attr,
+			 u16 vport, bool decap)
+{
+	return ERR_PTR(-EOPNOTSUPP);
+}
+
+static inline void
+mlx5_esw_indir_table_put(struct mlx5_eswitch *esw,
+			 u16 vport, bool decap)
+{
+}
+
+static inline bool
+mlx5_esw_indir_table_needed(struct mlx5_eswitch *esw,
+			    struct mlx5_flow_attr *attr,
+			    u16 vport_num,
+			    struct mlx5_core_dev *dest_mdev)
+{
+	return false;
+}
+
+static inline u16
+mlx5_esw_indir_table_decap_vport(struct mlx5_flow_attr *attr)
+{
+	return 0;
+}
+#endif
+
+#endif /* __MLX5_ESW_FT_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/esw/legacy.c b/drivers/net/ethernet/mellanox/mlx5/core/esw/legacy.c
new file mode 100644
index 000000000..fabe49a35
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/esw/legacy.c
@@ -0,0 +1,529 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2021 Mellanox Technologies Ltd */
+
+#include <linux/etherdevice.h>
+#include <linux/mlx5/driver.h>
+#include <linux/mlx5/mlx5_ifc.h>
+#include <linux/mlx5/vport.h>
+#include <linux/mlx5/fs.h>
+#include "esw/acl/lgcy.h"
+#include "esw/legacy.h"
+#include "mlx5_core.h"
+#include "eswitch.h"
+#include "fs_core.h"
+#include "fs_ft_pool.h"
+#include "esw/qos.h"
+
+enum {
+	LEGACY_VEPA_PRIO = 0,
+	LEGACY_FDB_PRIO,
+};
+
+static int esw_create_legacy_vepa_table(struct mlx5_eswitch *esw)
+{
+	struct mlx5_flow_table_attr ft_attr = {};
+	struct mlx5_core_dev *dev = esw->dev;
+	struct mlx5_flow_namespace *root_ns;
+	struct mlx5_flow_table *fdb;
+	int err;
+
+	root_ns = mlx5_get_fdb_sub_ns(dev, 0);
+	if (!root_ns) {
+		esw_warn(dev, "Failed to get FDB flow namespace\n");
+		return -EOPNOTSUPP;
+	}
+
+	/* num FTE 2, num FG 2 */
+	ft_attr.prio = LEGACY_VEPA_PRIO;
+	ft_attr.max_fte = 2;
+	ft_attr.autogroup.max_num_groups = 2;
+	fdb = mlx5_create_auto_grouped_flow_table(root_ns, &ft_attr);
+	if (IS_ERR(fdb)) {
+		err = PTR_ERR(fdb);
+		esw_warn(dev, "Failed to create VEPA FDB err %d\n", err);
+		return err;
+	}
+	esw->fdb_table.legacy.vepa_fdb = fdb;
+
+	return 0;
+}
+
+static void esw_destroy_legacy_fdb_table(struct mlx5_eswitch *esw)
+{
+	esw_debug(esw->dev, "Destroy FDB Table\n");
+	if (!esw->fdb_table.legacy.fdb)
+		return;
+
+	if (esw->fdb_table.legacy.promisc_grp)
+		mlx5_destroy_flow_group(esw->fdb_table.legacy.promisc_grp);
+	if (esw->fdb_table.legacy.allmulti_grp)
+		mlx5_destroy_flow_group(esw->fdb_table.legacy.allmulti_grp);
+	if (esw->fdb_table.legacy.addr_grp)
+		mlx5_destroy_flow_group(esw->fdb_table.legacy.addr_grp);
+	mlx5_destroy_flow_table(esw->fdb_table.legacy.fdb);
+
+	esw->fdb_table.legacy.fdb = NULL;
+	esw->fdb_table.legacy.addr_grp = NULL;
+	esw->fdb_table.legacy.allmulti_grp = NULL;
+	esw->fdb_table.legacy.promisc_grp = NULL;
+	atomic64_set(&esw->user_count, 0);
+}
+
+static int esw_create_legacy_fdb_table(struct mlx5_eswitch *esw)
+{
+	int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
+	struct mlx5_flow_table_attr ft_attr = {};
+	struct mlx5_core_dev *dev = esw->dev;
+	struct mlx5_flow_namespace *root_ns;
+	struct mlx5_flow_table *fdb;
+	struct mlx5_flow_group *g;
+	void *match_criteria;
+	int table_size;
+	u32 *flow_group_in;
+	u8 *dmac;
+	int err = 0;
+
+	esw_debug(dev, "Create FDB log_max_size(%d)\n",
+		  MLX5_CAP_ESW_FLOWTABLE_FDB(dev, log_max_ft_size));
+
+	root_ns = mlx5_get_fdb_sub_ns(dev, 0);
+	if (!root_ns) {
+		esw_warn(dev, "Failed to get FDB flow namespace\n");
+		return -EOPNOTSUPP;
+	}
+
+	flow_group_in = kvzalloc(inlen, GFP_KERNEL);
+	if (!flow_group_in)
+		return -ENOMEM;
+
+	ft_attr.max_fte = POOL_NEXT_SIZE;
+	ft_attr.prio = LEGACY_FDB_PRIO;
+	fdb = mlx5_create_flow_table(root_ns, &ft_attr);
+	if (IS_ERR(fdb)) {
+		err = PTR_ERR(fdb);
+		esw_warn(dev, "Failed to create FDB Table err %d\n", err);
+		goto out;
+	}
+	esw->fdb_table.legacy.fdb = fdb;
+	table_size = fdb->max_fte;
+
+	/* Addresses group : Full match unicast/multicast addresses */
+	MLX5_SET(create_flow_group_in, flow_group_in, match_criteria_enable,
+		 MLX5_MATCH_OUTER_HEADERS);
+	match_criteria = MLX5_ADDR_OF(create_flow_group_in, flow_group_in, match_criteria);
+	dmac = MLX5_ADDR_OF(fte_match_param, match_criteria, outer_headers.dmac_47_16);
+	MLX5_SET(create_flow_group_in, flow_group_in, start_flow_index, 0);
+	/* Preserve 2 entries for allmulti and promisc rules*/
+	MLX5_SET(create_flow_group_in, flow_group_in, end_flow_index, table_size - 3);
+	eth_broadcast_addr(dmac);
+	g = mlx5_create_flow_group(fdb, flow_group_in);
+	if (IS_ERR(g)) {
+		err = PTR_ERR(g);
+		esw_warn(dev, "Failed to create flow group err(%d)\n", err);
+		goto out;
+	}
+	esw->fdb_table.legacy.addr_grp = g;
+
+	/* Allmulti group : One rule that forwards any mcast traffic */
+	MLX5_SET(create_flow_group_in, flow_group_in, match_criteria_enable,
+		 MLX5_MATCH_OUTER_HEADERS);
+	MLX5_SET(create_flow_group_in, flow_group_in, start_flow_index, table_size - 2);
+	MLX5_SET(create_flow_group_in, flow_group_in, end_flow_index, table_size - 2);
+	eth_zero_addr(dmac);
+	dmac[0] = 0x01;
+	g = mlx5_create_flow_group(fdb, flow_group_in);
+	if (IS_ERR(g)) {
+		err = PTR_ERR(g);
+		esw_warn(dev, "Failed to create allmulti flow group err(%d)\n", err);
+		goto out;
+	}
+	esw->fdb_table.legacy.allmulti_grp = g;
+
+	/* Promiscuous group :
+	 * One rule that forward all unmatched traffic from previous groups
+	 */
+	eth_zero_addr(dmac);
+	MLX5_SET(create_flow_group_in, flow_group_in, match_criteria_enable,
+		 MLX5_MATCH_MISC_PARAMETERS);
+	MLX5_SET_TO_ONES(fte_match_param, match_criteria, misc_parameters.source_port);
+	MLX5_SET(create_flow_group_in, flow_group_in, start_flow_index, table_size - 1);
+	MLX5_SET(create_flow_group_in, flow_group_in, end_flow_index, table_size - 1);
+	g = mlx5_create_flow_group(fdb, flow_group_in);
+	if (IS_ERR(g)) {
+		err = PTR_ERR(g);
+		esw_warn(dev, "Failed to create promisc flow group err(%d)\n", err);
+		goto out;
+	}
+	esw->fdb_table.legacy.promisc_grp = g;
+
+out:
+	if (err)
+		esw_destroy_legacy_fdb_table(esw);
+
+	kvfree(flow_group_in);
+	return err;
+}
+
+static void esw_destroy_legacy_vepa_table(struct mlx5_eswitch *esw)
+{
+	esw_debug(esw->dev, "Destroy VEPA Table\n");
+	if (!esw->fdb_table.legacy.vepa_fdb)
+		return;
+
+	mlx5_destroy_flow_table(esw->fdb_table.legacy.vepa_fdb);
+	esw->fdb_table.legacy.vepa_fdb = NULL;
+}
+
+static int esw_create_legacy_table(struct mlx5_eswitch *esw)
+{
+	int err;
+
+	memset(&esw->fdb_table.legacy, 0, sizeof(struct legacy_fdb));
+	atomic64_set(&esw->user_count, 0);
+
+	err = esw_create_legacy_vepa_table(esw);
+	if (err)
+		return err;
+
+	err = esw_create_legacy_fdb_table(esw);
+	if (err)
+		esw_destroy_legacy_vepa_table(esw);
+
+	return err;
+}
+
+static void esw_cleanup_vepa_rules(struct mlx5_eswitch *esw)
+{
+	if (esw->fdb_table.legacy.vepa_uplink_rule)
+		mlx5_del_flow_rules(esw->fdb_table.legacy.vepa_uplink_rule);
+
+	if (esw->fdb_table.legacy.vepa_star_rule)
+		mlx5_del_flow_rules(esw->fdb_table.legacy.vepa_star_rule);
+
+	esw->fdb_table.legacy.vepa_uplink_rule = NULL;
+	esw->fdb_table.legacy.vepa_star_rule = NULL;
+}
+
+static void esw_destroy_legacy_table(struct mlx5_eswitch *esw)
+{
+	esw_cleanup_vepa_rules(esw);
+	esw_destroy_legacy_fdb_table(esw);
+	esw_destroy_legacy_vepa_table(esw);
+}
+
+#define MLX5_LEGACY_SRIOV_VPORT_EVENTS (MLX5_VPORT_UC_ADDR_CHANGE | \
+					MLX5_VPORT_MC_ADDR_CHANGE | \
+					MLX5_VPORT_PROMISC_CHANGE)
+
+int esw_legacy_enable(struct mlx5_eswitch *esw)
+{
+	struct mlx5_vport *vport;
+	unsigned long i;
+	int ret;
+
+	ret = esw_create_legacy_table(esw);
+	if (ret)
+		return ret;
+
+	mlx5_esw_for_each_vf_vport(esw, i, vport, esw->esw_funcs.num_vfs)
+		vport->info.link_state = MLX5_VPORT_ADMIN_STATE_AUTO;
+
+	ret = mlx5_eswitch_enable_pf_vf_vports(esw, MLX5_LEGACY_SRIOV_VPORT_EVENTS);
+	if (ret)
+		esw_destroy_legacy_table(esw);
+	return ret;
+}
+
+void esw_legacy_disable(struct mlx5_eswitch *esw)
+{
+	struct esw_mc_addr *mc_promisc;
+
+	mlx5_eswitch_disable_pf_vf_vports(esw);
+
+	mc_promisc = &esw->mc_promisc;
+	if (mc_promisc->uplink_rule)
+		mlx5_del_flow_rules(mc_promisc->uplink_rule);
+
+	esw_destroy_legacy_table(esw);
+}
+
+static int _mlx5_eswitch_set_vepa_locked(struct mlx5_eswitch *esw,
+					 u8 setting)
+{
+	struct mlx5_flow_destination dest = {};
+	struct mlx5_flow_act flow_act = {};
+	struct mlx5_flow_handle *flow_rule;
+	struct mlx5_flow_spec *spec;
+	int err = 0;
+	void *misc;
+
+	if (!setting) {
+		esw_cleanup_vepa_rules(esw);
+		return 0;
+	}
+
+	if (esw->fdb_table.legacy.vepa_uplink_rule)
+		return 0;
+
+	spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
+	if (!spec)
+		return -ENOMEM;
+
+	/* Uplink rule forward uplink traffic to FDB */
+	misc = MLX5_ADDR_OF(fte_match_param, spec->match_value, misc_parameters);
+	MLX5_SET(fte_match_set_misc, misc, source_port, MLX5_VPORT_UPLINK);
+
+	misc = MLX5_ADDR_OF(fte_match_param, spec->match_criteria, misc_parameters);
+	MLX5_SET_TO_ONES(fte_match_set_misc, misc, source_port);
+
+	spec->match_criteria_enable = MLX5_MATCH_MISC_PARAMETERS;
+	dest.type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
+	dest.ft = esw->fdb_table.legacy.fdb;
+	flow_act.action = MLX5_FLOW_CONTEXT_ACTION_FWD_DEST;
+	flow_rule = mlx5_add_flow_rules(esw->fdb_table.legacy.vepa_fdb, spec,
+					&flow_act, &dest, 1);
+	if (IS_ERR(flow_rule)) {
+		err = PTR_ERR(flow_rule);
+		goto out;
+	} else {
+		esw->fdb_table.legacy.vepa_uplink_rule = flow_rule;
+	}
+
+	/* Star rule to forward all traffic to uplink vport */
+	memset(&dest, 0, sizeof(dest));
+	dest.type = MLX5_FLOW_DESTINATION_TYPE_VPORT;
+	dest.vport.num = MLX5_VPORT_UPLINK;
+	flow_act.action = MLX5_FLOW_CONTEXT_ACTION_FWD_DEST;
+	flow_rule = mlx5_add_flow_rules(esw->fdb_table.legacy.vepa_fdb, NULL,
+					&flow_act, &dest, 1);
+	if (IS_ERR(flow_rule)) {
+		err = PTR_ERR(flow_rule);
+		goto out;
+	} else {
+		esw->fdb_table.legacy.vepa_star_rule = flow_rule;
+	}
+
+out:
+	kvfree(spec);
+	if (err)
+		esw_cleanup_vepa_rules(esw);
+	return err;
+}
+
+int mlx5_eswitch_set_vepa(struct mlx5_eswitch *esw, u8 setting)
+{
+	int err = 0;
+
+	if (!esw)
+		return -EOPNOTSUPP;
+
+	if (!mlx5_esw_allowed(esw))
+		return -EPERM;
+
+	mutex_lock(&esw->state_lock);
+	if (esw->mode != MLX5_ESWITCH_LEGACY) {
+		err = -EOPNOTSUPP;
+		goto out;
+	}
+
+	err = _mlx5_eswitch_set_vepa_locked(esw, setting);
+
+out:
+	mutex_unlock(&esw->state_lock);
+	return err;
+}
+
+int mlx5_eswitch_get_vepa(struct mlx5_eswitch *esw, u8 *setting)
+{
+	if (!esw)
+		return -EOPNOTSUPP;
+
+	if (!mlx5_esw_allowed(esw))
+		return -EPERM;
+
+	if (esw->mode != MLX5_ESWITCH_LEGACY)
+		return -EOPNOTSUPP;
+
+	*setting = esw->fdb_table.legacy.vepa_uplink_rule ? 1 : 0;
+	return 0;
+}
+
+int esw_legacy_vport_acl_setup(struct mlx5_eswitch *esw, struct mlx5_vport *vport)
+{
+	int ret;
+
+	/* Only non manager vports need ACL in legacy mode */
+	if (mlx5_esw_is_manager_vport(esw, vport->vport))
+		return 0;
+
+	ret = esw_acl_ingress_lgcy_setup(esw, vport);
+	if (ret)
+		goto ingress_err;
+
+	ret = esw_acl_egress_lgcy_setup(esw, vport);
+	if (ret)
+		goto egress_err;
+
+	return 0;
+
+egress_err:
+	esw_acl_ingress_lgcy_cleanup(esw, vport);
+ingress_err:
+	return ret;
+}
+
+void esw_legacy_vport_acl_cleanup(struct mlx5_eswitch *esw, struct mlx5_vport *vport)
+{
+	if (mlx5_esw_is_manager_vport(esw, vport->vport))
+		return;
+
+	esw_acl_egress_lgcy_cleanup(esw, vport);
+	esw_acl_ingress_lgcy_cleanup(esw, vport);
+}
+
+int mlx5_esw_query_vport_drop_stats(struct mlx5_core_dev *dev,
+				    struct mlx5_vport *vport,
+				    struct mlx5_vport_drop_stats *stats)
+{
+	u64 rx_discard_vport_down, tx_discard_vport_down;
+	struct mlx5_eswitch *esw = dev->priv.eswitch;
+	u64 bytes = 0;
+	int err = 0;
+
+	if (esw->mode != MLX5_ESWITCH_LEGACY)
+		return 0;
+
+	mutex_lock(&esw->state_lock);
+	if (!vport->enabled)
+		goto unlock;
+
+	if (!IS_ERR_OR_NULL(vport->egress.legacy.drop_counter))
+		mlx5_fc_query(dev, vport->egress.legacy.drop_counter,
+			      &stats->rx_dropped, &bytes);
+
+	if (vport->ingress.legacy.drop_counter)
+		mlx5_fc_query(dev, vport->ingress.legacy.drop_counter,
+			      &stats->tx_dropped, &bytes);
+
+	if (!MLX5_CAP_GEN(dev, receive_discard_vport_down) &&
+	    !MLX5_CAP_GEN(dev, transmit_discard_vport_down))
+		goto unlock;
+
+	err = mlx5_query_vport_down_stats(dev, vport->vport, 1,
+					  &rx_discard_vport_down,
+					  &tx_discard_vport_down);
+	if (err)
+		goto unlock;
+
+	if (MLX5_CAP_GEN(dev, receive_discard_vport_down))
+		stats->rx_dropped += rx_discard_vport_down;
+	if (MLX5_CAP_GEN(dev, transmit_discard_vport_down))
+		stats->tx_dropped += tx_discard_vport_down;
+
+unlock:
+	mutex_unlock(&esw->state_lock);
+	return err;
+}
+
+int mlx5_eswitch_set_vport_vlan(struct mlx5_eswitch *esw,
+				u16 vport, u16 vlan, u8 qos)
+{
+	u8 set_flags = 0;
+	int err = 0;
+
+	if (!mlx5_esw_allowed(esw))
+		return vlan ? -EPERM : 0;
+
+	if (vlan || qos)
+		set_flags = SET_VLAN_STRIP | SET_VLAN_INSERT;
+
+	mutex_lock(&esw->state_lock);
+	if (esw->mode != MLX5_ESWITCH_LEGACY) {
+		if (!vlan)
+			goto unlock; /* compatibility with libvirt */
+
+		err = -EOPNOTSUPP;
+		goto unlock;
+	}
+
+	err = __mlx5_eswitch_set_vport_vlan(esw, vport, vlan, qos, set_flags);
+
+unlock:
+	mutex_unlock(&esw->state_lock);
+	return err;
+}
+
+int mlx5_eswitch_set_vport_spoofchk(struct mlx5_eswitch *esw,
+				    u16 vport, bool spoofchk)
+{
+	struct mlx5_vport *evport = mlx5_eswitch_get_vport(esw, vport);
+	bool pschk;
+	int err = 0;
+
+	if (!mlx5_esw_allowed(esw))
+		return -EPERM;
+	if (IS_ERR(evport))
+		return PTR_ERR(evport);
+
+	mutex_lock(&esw->state_lock);
+	if (esw->mode != MLX5_ESWITCH_LEGACY) {
+		err = -EOPNOTSUPP;
+		goto unlock;
+	}
+	pschk = evport->info.spoofchk;
+	evport->info.spoofchk = spoofchk;
+	if (pschk && !is_valid_ether_addr(evport->info.mac))
+		mlx5_core_warn(esw->dev,
+			       "Spoofchk in set while MAC is invalid, vport(%d)\n",
+			       evport->vport);
+	if (evport->enabled && esw->mode == MLX5_ESWITCH_LEGACY)
+		err = esw_acl_ingress_lgcy_setup(esw, evport);
+	if (err)
+		evport->info.spoofchk = pschk;
+
+unlock:
+	mutex_unlock(&esw->state_lock);
+	return err;
+}
+
+int mlx5_eswitch_set_vport_trust(struct mlx5_eswitch *esw,
+				 u16 vport, bool setting)
+{
+	struct mlx5_vport *evport = mlx5_eswitch_get_vport(esw, vport);
+	int err = 0;
+
+	if (!mlx5_esw_allowed(esw))
+		return -EPERM;
+	if (IS_ERR(evport))
+		return PTR_ERR(evport);
+
+	mutex_lock(&esw->state_lock);
+	if (esw->mode != MLX5_ESWITCH_LEGACY) {
+		err = -EOPNOTSUPP;
+		goto unlock;
+	}
+	evport->info.trusted = setting;
+	if (evport->enabled)
+		esw_vport_change_handle_locked(evport);
+
+unlock:
+	mutex_unlock(&esw->state_lock);
+	return err;
+}
+
+int mlx5_eswitch_set_vport_rate(struct mlx5_eswitch *esw, u16 vport,
+				u32 max_rate, u32 min_rate)
+{
+	struct mlx5_vport *evport = mlx5_eswitch_get_vport(esw, vport);
+	int err;
+
+	if (!mlx5_esw_allowed(esw))
+		return -EPERM;
+	if (IS_ERR(evport))
+		return PTR_ERR(evport);
+
+	mutex_lock(&esw->state_lock);
+	err = mlx5_esw_qos_set_vport_rate(esw, evport, max_rate, min_rate);
+	mutex_unlock(&esw->state_lock);
+	return err;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/esw/legacy.h b/drivers/net/ethernet/mellanox/mlx5/core/esw/legacy.h
new file mode 100644
index 000000000..e0820bb72
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/esw/legacy.h
@@ -0,0 +1,22 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2021 Mellanox Technologies Ltd */
+
+#ifndef __MLX5_ESW_LEGACY_H__
+#define __MLX5_ESW_LEGACY_H__
+
+#define MLX5_LEGACY_SRIOV_VPORT_EVENTS (MLX5_VPORT_UC_ADDR_CHANGE | \
+					MLX5_VPORT_MC_ADDR_CHANGE | \
+					MLX5_VPORT_PROMISC_CHANGE)
+
+struct mlx5_eswitch;
+
+int esw_legacy_enable(struct mlx5_eswitch *esw);
+void esw_legacy_disable(struct mlx5_eswitch *esw);
+
+int esw_legacy_vport_acl_setup(struct mlx5_eswitch *esw, struct mlx5_vport *vport);
+void esw_legacy_vport_acl_cleanup(struct mlx5_eswitch *esw, struct mlx5_vport *vport);
+
+int mlx5_esw_query_vport_drop_stats(struct mlx5_core_dev *dev,
+				    struct mlx5_vport *vport,
+				    struct mlx5_vport_drop_stats *stats);
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c b/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c
new file mode 100644
index 000000000..75015d370
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c
@@ -0,0 +1,943 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. All rights reserved. */
+
+#include "eswitch.h"
+#include "esw/qos.h"
+#include "en/port.h"
+#define CREATE_TRACE_POINTS
+#include "diag/qos_tracepoint.h"
+
+/* Minimum supported BW share value by the HW is 1 Mbit/sec */
+#define MLX5_MIN_BW_SHARE 1
+
+#define MLX5_RATE_TO_BW_SHARE(rate, divider, limit) \
+	min_t(u32, max_t(u32, DIV_ROUND_UP(rate, divider), MLX5_MIN_BW_SHARE), limit)
+
+struct mlx5_esw_rate_group {
+	u32 tsar_ix;
+	u32 max_rate;
+	u32 min_rate;
+	u32 bw_share;
+	struct list_head list;
+};
+
+static int esw_qos_tsar_config(struct mlx5_core_dev *dev, u32 *sched_ctx,
+			       u32 tsar_ix, u32 max_rate, u32 bw_share)
+{
+	u32 bitmask = 0;
+
+	if (!MLX5_CAP_GEN(dev, qos) || !MLX5_CAP_QOS(dev, esw_scheduling))
+		return -EOPNOTSUPP;
+
+	MLX5_SET(scheduling_context, sched_ctx, max_average_bw, max_rate);
+	MLX5_SET(scheduling_context, sched_ctx, bw_share, bw_share);
+	bitmask |= MODIFY_SCHEDULING_ELEMENT_IN_MODIFY_BITMASK_MAX_AVERAGE_BW;
+	bitmask |= MODIFY_SCHEDULING_ELEMENT_IN_MODIFY_BITMASK_BW_SHARE;
+
+	return mlx5_modify_scheduling_element_cmd(dev,
+						  SCHEDULING_HIERARCHY_E_SWITCH,
+						  sched_ctx,
+						  tsar_ix,
+						  bitmask);
+}
+
+static int esw_qos_group_config(struct mlx5_eswitch *esw, struct mlx5_esw_rate_group *group,
+				u32 max_rate, u32 bw_share, struct netlink_ext_ack *extack)
+{
+	u32 sched_ctx[MLX5_ST_SZ_DW(scheduling_context)] = {};
+	struct mlx5_core_dev *dev = esw->dev;
+	int err;
+
+	err = esw_qos_tsar_config(dev, sched_ctx,
+				  group->tsar_ix,
+				  max_rate, bw_share);
+	if (err)
+		NL_SET_ERR_MSG_MOD(extack, "E-Switch modify group TSAR element failed");
+
+	trace_mlx5_esw_group_qos_config(dev, group, group->tsar_ix, bw_share, max_rate);
+
+	return err;
+}
+
+static int esw_qos_vport_config(struct mlx5_eswitch *esw,
+				struct mlx5_vport *vport,
+				u32 max_rate, u32 bw_share,
+				struct netlink_ext_ack *extack)
+{
+	u32 sched_ctx[MLX5_ST_SZ_DW(scheduling_context)] = {};
+	struct mlx5_core_dev *dev = esw->dev;
+	int err;
+
+	if (!vport->qos.enabled)
+		return -EIO;
+
+	err = esw_qos_tsar_config(dev, sched_ctx, vport->qos.esw_tsar_ix,
+				  max_rate, bw_share);
+	if (err) {
+		esw_warn(esw->dev,
+			 "E-Switch modify TSAR vport element failed (vport=%d,err=%d)\n",
+			 vport->vport, err);
+		NL_SET_ERR_MSG_MOD(extack, "E-Switch modify TSAR vport element failed");
+		return err;
+	}
+
+	trace_mlx5_esw_vport_qos_config(vport, bw_share, max_rate);
+
+	return 0;
+}
+
+static u32 esw_qos_calculate_min_rate_divider(struct mlx5_eswitch *esw,
+					      struct mlx5_esw_rate_group *group,
+					      bool group_level)
+{
+	u32 fw_max_bw_share = MLX5_CAP_QOS(esw->dev, max_tsar_bw_share);
+	struct mlx5_vport *evport;
+	u32 max_guarantee = 0;
+	unsigned long i;
+
+	if (group_level) {
+		struct mlx5_esw_rate_group *group;
+
+		list_for_each_entry(group, &esw->qos.groups, list) {
+			if (group->min_rate < max_guarantee)
+				continue;
+			max_guarantee = group->min_rate;
+		}
+	} else {
+		mlx5_esw_for_each_vport(esw, i, evport) {
+			if (!evport->enabled || !evport->qos.enabled ||
+			    evport->qos.group != group || evport->qos.min_rate < max_guarantee)
+				continue;
+			max_guarantee = evport->qos.min_rate;
+		}
+	}
+
+	if (max_guarantee)
+		return max_t(u32, max_guarantee / fw_max_bw_share, 1);
+
+	/* If vports min rate divider is 0 but their group has bw_share configured, then
+	 * need to set bw_share for vports to minimal value.
+	 */
+	if (!group_level && !max_guarantee && group && group->bw_share)
+		return 1;
+	return 0;
+}
+
+static u32 esw_qos_calc_bw_share(u32 min_rate, u32 divider, u32 fw_max)
+{
+	if (divider)
+		return MLX5_RATE_TO_BW_SHARE(min_rate, divider, fw_max);
+
+	return 0;
+}
+
+static int esw_qos_normalize_vports_min_rate(struct mlx5_eswitch *esw,
+					     struct mlx5_esw_rate_group *group,
+					     struct netlink_ext_ack *extack)
+{
+	u32 fw_max_bw_share = MLX5_CAP_QOS(esw->dev, max_tsar_bw_share);
+	u32 divider = esw_qos_calculate_min_rate_divider(esw, group, false);
+	struct mlx5_vport *evport;
+	unsigned long i;
+	u32 bw_share;
+	int err;
+
+	mlx5_esw_for_each_vport(esw, i, evport) {
+		if (!evport->enabled || !evport->qos.enabled || evport->qos.group != group)
+			continue;
+		bw_share = esw_qos_calc_bw_share(evport->qos.min_rate, divider, fw_max_bw_share);
+
+		if (bw_share == evport->qos.bw_share)
+			continue;
+
+		err = esw_qos_vport_config(esw, evport, evport->qos.max_rate, bw_share, extack);
+		if (err)
+			return err;
+
+		evport->qos.bw_share = bw_share;
+	}
+
+	return 0;
+}
+
+static int esw_qos_normalize_groups_min_rate(struct mlx5_eswitch *esw, u32 divider,
+					     struct netlink_ext_ack *extack)
+{
+	u32 fw_max_bw_share = MLX5_CAP_QOS(esw->dev, max_tsar_bw_share);
+	struct mlx5_esw_rate_group *group;
+	u32 bw_share;
+	int err;
+
+	list_for_each_entry(group, &esw->qos.groups, list) {
+		bw_share = esw_qos_calc_bw_share(group->min_rate, divider, fw_max_bw_share);
+
+		if (bw_share == group->bw_share)
+			continue;
+
+		err = esw_qos_group_config(esw, group, group->max_rate, bw_share, extack);
+		if (err)
+			return err;
+
+		group->bw_share = bw_share;
+
+		/* All the group's vports need to be set with default bw_share
+		 * to enable them with QOS
+		 */
+		err = esw_qos_normalize_vports_min_rate(esw, group, extack);
+
+		if (err)
+			return err;
+	}
+
+	return 0;
+}
+
+static int esw_qos_set_vport_min_rate(struct mlx5_eswitch *esw, struct mlx5_vport *evport,
+				      u32 min_rate, struct netlink_ext_ack *extack)
+{
+	u32 fw_max_bw_share, previous_min_rate;
+	bool min_rate_supported;
+	int err;
+
+	lockdep_assert_held(&esw->state_lock);
+	fw_max_bw_share = MLX5_CAP_QOS(esw->dev, max_tsar_bw_share);
+	min_rate_supported = MLX5_CAP_QOS(esw->dev, esw_bw_share) &&
+				fw_max_bw_share >= MLX5_MIN_BW_SHARE;
+	if (min_rate && !min_rate_supported)
+		return -EOPNOTSUPP;
+	if (min_rate == evport->qos.min_rate)
+		return 0;
+
+	previous_min_rate = evport->qos.min_rate;
+	evport->qos.min_rate = min_rate;
+	err = esw_qos_normalize_vports_min_rate(esw, evport->qos.group, extack);
+	if (err)
+		evport->qos.min_rate = previous_min_rate;
+
+	return err;
+}
+
+static int esw_qos_set_vport_max_rate(struct mlx5_eswitch *esw, struct mlx5_vport *evport,
+				      u32 max_rate, struct netlink_ext_ack *extack)
+{
+	u32 act_max_rate = max_rate;
+	bool max_rate_supported;
+	int err;
+
+	lockdep_assert_held(&esw->state_lock);
+	max_rate_supported = MLX5_CAP_QOS(esw->dev, esw_rate_limit);
+
+	if (max_rate && !max_rate_supported)
+		return -EOPNOTSUPP;
+	if (max_rate == evport->qos.max_rate)
+		return 0;
+
+	/* If parent group has rate limit need to set to group
+	 * value when new max rate is 0.
+	 */
+	if (evport->qos.group && !max_rate)
+		act_max_rate = evport->qos.group->max_rate;
+
+	err = esw_qos_vport_config(esw, evport, act_max_rate, evport->qos.bw_share, extack);
+
+	if (!err)
+		evport->qos.max_rate = max_rate;
+
+	return err;
+}
+
+static int esw_qos_set_group_min_rate(struct mlx5_eswitch *esw, struct mlx5_esw_rate_group *group,
+				      u32 min_rate, struct netlink_ext_ack *extack)
+{
+	u32 fw_max_bw_share = MLX5_CAP_QOS(esw->dev, max_tsar_bw_share);
+	struct mlx5_core_dev *dev = esw->dev;
+	u32 previous_min_rate, divider;
+	int err;
+
+	if (!(MLX5_CAP_QOS(dev, esw_bw_share) && fw_max_bw_share >= MLX5_MIN_BW_SHARE))
+		return -EOPNOTSUPP;
+
+	if (min_rate == group->min_rate)
+		return 0;
+
+	previous_min_rate = group->min_rate;
+	group->min_rate = min_rate;
+	divider = esw_qos_calculate_min_rate_divider(esw, group, true);
+	err = esw_qos_normalize_groups_min_rate(esw, divider, extack);
+	if (err) {
+		group->min_rate = previous_min_rate;
+		NL_SET_ERR_MSG_MOD(extack, "E-Switch group min rate setting failed");
+
+		/* Attempt restoring previous configuration */
+		divider = esw_qos_calculate_min_rate_divider(esw, group, true);
+		if (esw_qos_normalize_groups_min_rate(esw, divider, extack))
+			NL_SET_ERR_MSG_MOD(extack, "E-Switch BW share restore failed");
+	}
+
+	return err;
+}
+
+static int esw_qos_set_group_max_rate(struct mlx5_eswitch *esw,
+				      struct mlx5_esw_rate_group *group,
+				      u32 max_rate, struct netlink_ext_ack *extack)
+{
+	struct mlx5_vport *vport;
+	unsigned long i;
+	int err;
+
+	if (group->max_rate == max_rate)
+		return 0;
+
+	err = esw_qos_group_config(esw, group, max_rate, group->bw_share, extack);
+	if (err)
+		return err;
+
+	group->max_rate = max_rate;
+
+	/* Any unlimited vports in the group should be set
+	 * with the value of the group.
+	 */
+	mlx5_esw_for_each_vport(esw, i, vport) {
+		if (!vport->enabled || !vport->qos.enabled ||
+		    vport->qos.group != group || vport->qos.max_rate)
+			continue;
+
+		err = esw_qos_vport_config(esw, vport, max_rate, vport->qos.bw_share, extack);
+		if (err)
+			NL_SET_ERR_MSG_MOD(extack,
+					   "E-Switch vport implicit rate limit setting failed");
+	}
+
+	return err;
+}
+
+static int esw_qos_vport_create_sched_element(struct mlx5_eswitch *esw,
+					      struct mlx5_vport *vport,
+					      u32 max_rate, u32 bw_share)
+{
+	u32 sched_ctx[MLX5_ST_SZ_DW(scheduling_context)] = {};
+	struct mlx5_esw_rate_group *group = vport->qos.group;
+	struct mlx5_core_dev *dev = esw->dev;
+	u32 parent_tsar_ix;
+	void *vport_elem;
+	int err;
+
+	parent_tsar_ix = group ? group->tsar_ix : esw->qos.root_tsar_ix;
+	MLX5_SET(scheduling_context, sched_ctx, element_type,
+		 SCHEDULING_CONTEXT_ELEMENT_TYPE_VPORT);
+	vport_elem = MLX5_ADDR_OF(scheduling_context, sched_ctx, element_attributes);
+	MLX5_SET(vport_element, vport_elem, vport_number, vport->vport);
+	MLX5_SET(scheduling_context, sched_ctx, parent_element_id, parent_tsar_ix);
+	MLX5_SET(scheduling_context, sched_ctx, max_average_bw, max_rate);
+	MLX5_SET(scheduling_context, sched_ctx, bw_share, bw_share);
+
+	err = mlx5_create_scheduling_element_cmd(dev,
+						 SCHEDULING_HIERARCHY_E_SWITCH,
+						 sched_ctx,
+						 &vport->qos.esw_tsar_ix);
+	if (err) {
+		esw_warn(esw->dev, "E-Switch create TSAR vport element failed (vport=%d,err=%d)\n",
+			 vport->vport, err);
+		return err;
+	}
+
+	return 0;
+}
+
+static int esw_qos_update_group_scheduling_element(struct mlx5_eswitch *esw,
+						   struct mlx5_vport *vport,
+						   struct mlx5_esw_rate_group *curr_group,
+						   struct mlx5_esw_rate_group *new_group,
+						   struct netlink_ext_ack *extack)
+{
+	u32 max_rate;
+	int err;
+
+	err = mlx5_destroy_scheduling_element_cmd(esw->dev,
+						  SCHEDULING_HIERARCHY_E_SWITCH,
+						  vport->qos.esw_tsar_ix);
+	if (err) {
+		NL_SET_ERR_MSG_MOD(extack, "E-Switch destroy TSAR vport element failed");
+		return err;
+	}
+
+	vport->qos.group = new_group;
+	max_rate = vport->qos.max_rate ? vport->qos.max_rate : new_group->max_rate;
+
+	/* If vport is unlimited, we set the group's value.
+	 * Therefore, if the group is limited it will apply to
+	 * the vport as well and if not, vport will remain unlimited.
+	 */
+	err = esw_qos_vport_create_sched_element(esw, vport, max_rate, vport->qos.bw_share);
+	if (err) {
+		NL_SET_ERR_MSG_MOD(extack, "E-Switch vport group set failed.");
+		goto err_sched;
+	}
+
+	return 0;
+
+err_sched:
+	vport->qos.group = curr_group;
+	max_rate = vport->qos.max_rate ? vport->qos.max_rate : curr_group->max_rate;
+	if (esw_qos_vport_create_sched_element(esw, vport, max_rate, vport->qos.bw_share))
+		esw_warn(esw->dev, "E-Switch vport group restore failed (vport=%d)\n",
+			 vport->vport);
+
+	return err;
+}
+
+static int esw_qos_vport_update_group(struct mlx5_eswitch *esw,
+				      struct mlx5_vport *vport,
+				      struct mlx5_esw_rate_group *group,
+				      struct netlink_ext_ack *extack)
+{
+	struct mlx5_esw_rate_group *new_group, *curr_group;
+	int err;
+
+	if (!vport->enabled)
+		return -EINVAL;
+
+	curr_group = vport->qos.group;
+	new_group = group ?: esw->qos.group0;
+	if (curr_group == new_group)
+		return 0;
+
+	err = esw_qos_update_group_scheduling_element(esw, vport, curr_group, new_group, extack);
+	if (err)
+		return err;
+
+	/* Recalculate bw share weights of old and new groups */
+	if (vport->qos.bw_share || new_group->bw_share) {
+		esw_qos_normalize_vports_min_rate(esw, curr_group, extack);
+		esw_qos_normalize_vports_min_rate(esw, new_group, extack);
+	}
+
+	return 0;
+}
+
+static struct mlx5_esw_rate_group *
+__esw_qos_create_rate_group(struct mlx5_eswitch *esw, struct netlink_ext_ack *extack)
+{
+	u32 tsar_ctx[MLX5_ST_SZ_DW(scheduling_context)] = {};
+	struct mlx5_esw_rate_group *group;
+	u32 divider;
+	int err;
+
+	group = kzalloc(sizeof(*group), GFP_KERNEL);
+	if (!group)
+		return ERR_PTR(-ENOMEM);
+
+	MLX5_SET(scheduling_context, tsar_ctx, parent_element_id,
+		 esw->qos.root_tsar_ix);
+	err = mlx5_create_scheduling_element_cmd(esw->dev,
+						 SCHEDULING_HIERARCHY_E_SWITCH,
+						 tsar_ctx,
+						 &group->tsar_ix);
+	if (err) {
+		NL_SET_ERR_MSG_MOD(extack, "E-Switch create TSAR for group failed");
+		goto err_sched_elem;
+	}
+
+	list_add_tail(&group->list, &esw->qos.groups);
+
+	divider = esw_qos_calculate_min_rate_divider(esw, group, true);
+	if (divider) {
+		err = esw_qos_normalize_groups_min_rate(esw, divider, extack);
+		if (err) {
+			NL_SET_ERR_MSG_MOD(extack, "E-Switch groups normalization failed");
+			goto err_min_rate;
+		}
+	}
+	trace_mlx5_esw_group_qos_create(esw->dev, group, group->tsar_ix);
+
+	return group;
+
+err_min_rate:
+	list_del(&group->list);
+	if (mlx5_destroy_scheduling_element_cmd(esw->dev,
+						SCHEDULING_HIERARCHY_E_SWITCH,
+						group->tsar_ix))
+		NL_SET_ERR_MSG_MOD(extack, "E-Switch destroy TSAR for group failed");
+err_sched_elem:
+	kfree(group);
+	return ERR_PTR(err);
+}
+
+static int esw_qos_get(struct mlx5_eswitch *esw, struct netlink_ext_ack *extack);
+static void esw_qos_put(struct mlx5_eswitch *esw);
+
+static struct mlx5_esw_rate_group *
+esw_qos_create_rate_group(struct mlx5_eswitch *esw, struct netlink_ext_ack *extack)
+{
+	struct mlx5_esw_rate_group *group;
+	int err;
+
+	if (!MLX5_CAP_QOS(esw->dev, log_esw_max_sched_depth))
+		return ERR_PTR(-EOPNOTSUPP);
+
+	err = esw_qos_get(esw, extack);
+	if (err)
+		return ERR_PTR(err);
+
+	group = __esw_qos_create_rate_group(esw, extack);
+	if (IS_ERR(group))
+		esw_qos_put(esw);
+
+	return group;
+}
+
+static int __esw_qos_destroy_rate_group(struct mlx5_eswitch *esw,
+					struct mlx5_esw_rate_group *group,
+					struct netlink_ext_ack *extack)
+{
+	u32 divider;
+	int err;
+
+	list_del(&group->list);
+
+	divider = esw_qos_calculate_min_rate_divider(esw, NULL, true);
+	err = esw_qos_normalize_groups_min_rate(esw, divider, extack);
+	if (err)
+		NL_SET_ERR_MSG_MOD(extack, "E-Switch groups' normalization failed");
+
+	err = mlx5_destroy_scheduling_element_cmd(esw->dev,
+						  SCHEDULING_HIERARCHY_E_SWITCH,
+						  group->tsar_ix);
+	if (err)
+		NL_SET_ERR_MSG_MOD(extack, "E-Switch destroy TSAR_ID failed");
+
+	trace_mlx5_esw_group_qos_destroy(esw->dev, group, group->tsar_ix);
+
+	kfree(group);
+
+	return err;
+}
+
+static int esw_qos_destroy_rate_group(struct mlx5_eswitch *esw,
+				      struct mlx5_esw_rate_group *group,
+				      struct netlink_ext_ack *extack)
+{
+	int err;
+
+	err = __esw_qos_destroy_rate_group(esw, group, extack);
+	esw_qos_put(esw);
+
+	return err;
+}
+
+static bool esw_qos_element_type_supported(struct mlx5_core_dev *dev, int type)
+{
+	switch (type) {
+	case SCHEDULING_CONTEXT_ELEMENT_TYPE_TSAR:
+		return MLX5_CAP_QOS(dev, esw_element_type) &
+		       ELEMENT_TYPE_CAP_MASK_TASR;
+	case SCHEDULING_CONTEXT_ELEMENT_TYPE_VPORT:
+		return MLX5_CAP_QOS(dev, esw_element_type) &
+		       ELEMENT_TYPE_CAP_MASK_VPORT;
+	case SCHEDULING_CONTEXT_ELEMENT_TYPE_VPORT_TC:
+		return MLX5_CAP_QOS(dev, esw_element_type) &
+		       ELEMENT_TYPE_CAP_MASK_VPORT_TC;
+	case SCHEDULING_CONTEXT_ELEMENT_TYPE_PARA_VPORT_TC:
+		return MLX5_CAP_QOS(dev, esw_element_type) &
+		       ELEMENT_TYPE_CAP_MASK_PARA_VPORT_TC;
+	}
+	return false;
+}
+
+static int esw_qos_create(struct mlx5_eswitch *esw, struct netlink_ext_ack *extack)
+{
+	u32 tsar_ctx[MLX5_ST_SZ_DW(scheduling_context)] = {};
+	struct mlx5_core_dev *dev = esw->dev;
+	__be32 *attr;
+	int err;
+
+	if (!MLX5_CAP_GEN(dev, qos) || !MLX5_CAP_QOS(dev, esw_scheduling))
+		return -EOPNOTSUPP;
+
+	if (!esw_qos_element_type_supported(dev, SCHEDULING_CONTEXT_ELEMENT_TYPE_TSAR))
+		return -EOPNOTSUPP;
+
+	MLX5_SET(scheduling_context, tsar_ctx, element_type,
+		 SCHEDULING_CONTEXT_ELEMENT_TYPE_TSAR);
+
+	attr = MLX5_ADDR_OF(scheduling_context, tsar_ctx, element_attributes);
+	*attr = cpu_to_be32(TSAR_ELEMENT_TSAR_TYPE_DWRR << 16);
+
+	err = mlx5_create_scheduling_element_cmd(dev,
+						 SCHEDULING_HIERARCHY_E_SWITCH,
+						 tsar_ctx,
+						 &esw->qos.root_tsar_ix);
+	if (err) {
+		esw_warn(dev, "E-Switch create root TSAR failed (%d)\n", err);
+		return err;
+	}
+
+	INIT_LIST_HEAD(&esw->qos.groups);
+	if (MLX5_CAP_QOS(dev, log_esw_max_sched_depth)) {
+		esw->qos.group0 = __esw_qos_create_rate_group(esw, extack);
+		if (IS_ERR(esw->qos.group0)) {
+			esw_warn(dev, "E-Switch create rate group 0 failed (%ld)\n",
+				 PTR_ERR(esw->qos.group0));
+			err = PTR_ERR(esw->qos.group0);
+			goto err_group0;
+		}
+	}
+	refcount_set(&esw->qos.refcnt, 1);
+
+	return 0;
+
+err_group0:
+	if (mlx5_destroy_scheduling_element_cmd(esw->dev, SCHEDULING_HIERARCHY_E_SWITCH,
+						esw->qos.root_tsar_ix))
+		esw_warn(esw->dev, "E-Switch destroy root TSAR failed.\n");
+
+	return err;
+}
+
+static void esw_qos_destroy(struct mlx5_eswitch *esw)
+{
+	int err;
+
+	if (esw->qos.group0)
+		__esw_qos_destroy_rate_group(esw, esw->qos.group0, NULL);
+
+	err = mlx5_destroy_scheduling_element_cmd(esw->dev,
+						  SCHEDULING_HIERARCHY_E_SWITCH,
+						  esw->qos.root_tsar_ix);
+	if (err)
+		esw_warn(esw->dev, "E-Switch destroy root TSAR failed (%d)\n", err);
+}
+
+static int esw_qos_get(struct mlx5_eswitch *esw, struct netlink_ext_ack *extack)
+{
+	int err = 0;
+
+	lockdep_assert_held(&esw->state_lock);
+
+	if (!refcount_inc_not_zero(&esw->qos.refcnt)) {
+		/* esw_qos_create() set refcount to 1 only on success.
+		 * No need to decrement on failure.
+		 */
+		err = esw_qos_create(esw, extack);
+	}
+
+	return err;
+}
+
+static void esw_qos_put(struct mlx5_eswitch *esw)
+{
+	lockdep_assert_held(&esw->state_lock);
+	if (refcount_dec_and_test(&esw->qos.refcnt))
+		esw_qos_destroy(esw);
+}
+
+static int esw_qos_vport_enable(struct mlx5_eswitch *esw, struct mlx5_vport *vport,
+				u32 max_rate, u32 bw_share, struct netlink_ext_ack *extack)
+{
+	int err;
+
+	lockdep_assert_held(&esw->state_lock);
+	if (vport->qos.enabled)
+		return 0;
+
+	err = esw_qos_get(esw, extack);
+	if (err)
+		return err;
+
+	vport->qos.group = esw->qos.group0;
+
+	err = esw_qos_vport_create_sched_element(esw, vport, max_rate, bw_share);
+	if (err)
+		goto err_out;
+
+	vport->qos.enabled = true;
+	trace_mlx5_esw_vport_qos_create(vport, bw_share, max_rate);
+
+	return 0;
+
+err_out:
+	esw_qos_put(esw);
+
+	return err;
+}
+
+void mlx5_esw_qos_vport_disable(struct mlx5_eswitch *esw, struct mlx5_vport *vport)
+{
+	int err;
+
+	lockdep_assert_held(&esw->state_lock);
+	if (!vport->qos.enabled)
+		return;
+	WARN(vport->qos.group && vport->qos.group != esw->qos.group0,
+	     "Disabling QoS on port before detaching it from group");
+
+	err = mlx5_destroy_scheduling_element_cmd(esw->dev,
+						  SCHEDULING_HIERARCHY_E_SWITCH,
+						  vport->qos.esw_tsar_ix);
+	if (err)
+		esw_warn(esw->dev, "E-Switch destroy TSAR vport element failed (vport=%d,err=%d)\n",
+			 vport->vport, err);
+
+	memset(&vport->qos, 0, sizeof(vport->qos));
+	trace_mlx5_esw_vport_qos_destroy(vport);
+
+	esw_qos_put(esw);
+}
+
+int mlx5_esw_qos_set_vport_rate(struct mlx5_eswitch *esw, struct mlx5_vport *vport,
+				u32 max_rate, u32 min_rate)
+{
+	int err;
+
+	lockdep_assert_held(&esw->state_lock);
+	err = esw_qos_vport_enable(esw, vport, 0, 0, NULL);
+	if (err)
+		return err;
+
+	err = esw_qos_set_vport_min_rate(esw, vport, min_rate, NULL);
+	if (!err)
+		err = esw_qos_set_vport_max_rate(esw, vport, max_rate, NULL);
+
+	return err;
+}
+
+int mlx5_esw_qos_modify_vport_rate(struct mlx5_eswitch *esw, u16 vport_num, u32 rate_mbps)
+{
+	u32 ctx[MLX5_ST_SZ_DW(scheduling_context)] = {};
+	struct mlx5_vport *vport;
+	u32 bitmask;
+	int err;
+
+	vport = mlx5_eswitch_get_vport(esw, vport_num);
+	if (IS_ERR(vport))
+		return PTR_ERR(vport);
+
+	mutex_lock(&esw->state_lock);
+	if (!vport->qos.enabled) {
+		/* Eswitch QoS wasn't enabled yet. Enable it and vport QoS. */
+		err = esw_qos_vport_enable(esw, vport, rate_mbps, vport->qos.bw_share, NULL);
+	} else {
+		MLX5_SET(scheduling_context, ctx, max_average_bw, rate_mbps);
+
+		bitmask = MODIFY_SCHEDULING_ELEMENT_IN_MODIFY_BITMASK_MAX_AVERAGE_BW;
+		err = mlx5_modify_scheduling_element_cmd(esw->dev,
+							 SCHEDULING_HIERARCHY_E_SWITCH,
+							 ctx,
+							 vport->qos.esw_tsar_ix,
+							 bitmask);
+	}
+	mutex_unlock(&esw->state_lock);
+
+	return err;
+}
+
+#define MLX5_LINKSPEED_UNIT 125000 /* 1Mbps in Bps */
+
+/* Converts bytes per second value passed in a pointer into megabits per
+ * second, rewriting last. If converted rate exceed link speed or is not a
+ * fraction of Mbps - returns error.
+ */
+static int esw_qos_devlink_rate_to_mbps(struct mlx5_core_dev *mdev, const char *name,
+					u64 *rate, struct netlink_ext_ack *extack)
+{
+	u32 link_speed_max, reminder;
+	u64 value;
+	int err;
+
+	err = mlx5e_port_max_linkspeed(mdev, &link_speed_max);
+	if (err) {
+		NL_SET_ERR_MSG_MOD(extack, "Failed to get link maximum speed");
+		return err;
+	}
+
+	value = div_u64_rem(*rate, MLX5_LINKSPEED_UNIT, &reminder);
+	if (reminder) {
+		pr_err("%s rate value %lluBps not in link speed units of 1Mbps.\n",
+		       name, *rate);
+		NL_SET_ERR_MSG_MOD(extack, "TX rate value not in link speed units of 1Mbps");
+		return -EINVAL;
+	}
+
+	if (value > link_speed_max) {
+		pr_err("%s rate value %lluMbps exceed link maximum speed %u.\n",
+		       name, value, link_speed_max);
+		NL_SET_ERR_MSG_MOD(extack, "TX rate value exceed link maximum speed");
+		return -EINVAL;
+	}
+
+	*rate = value;
+	return 0;
+}
+
+/* Eswitch devlink rate API */
+
+int mlx5_esw_devlink_rate_leaf_tx_share_set(struct devlink_rate *rate_leaf, void *priv,
+					    u64 tx_share, struct netlink_ext_ack *extack)
+{
+	struct mlx5_vport *vport = priv;
+	struct mlx5_eswitch *esw;
+	int err;
+
+	esw = vport->dev->priv.eswitch;
+	if (!mlx5_esw_allowed(esw))
+		return -EPERM;
+
+	err = esw_qos_devlink_rate_to_mbps(vport->dev, "tx_share", &tx_share, extack);
+	if (err)
+		return err;
+
+	mutex_lock(&esw->state_lock);
+	err = esw_qos_vport_enable(esw, vport, 0, 0, extack);
+	if (err)
+		goto unlock;
+
+	err = esw_qos_set_vport_min_rate(esw, vport, tx_share, extack);
+unlock:
+	mutex_unlock(&esw->state_lock);
+	return err;
+}
+
+int mlx5_esw_devlink_rate_leaf_tx_max_set(struct devlink_rate *rate_leaf, void *priv,
+					  u64 tx_max, struct netlink_ext_ack *extack)
+{
+	struct mlx5_vport *vport = priv;
+	struct mlx5_eswitch *esw;
+	int err;
+
+	esw = vport->dev->priv.eswitch;
+	if (!mlx5_esw_allowed(esw))
+		return -EPERM;
+
+	err = esw_qos_devlink_rate_to_mbps(vport->dev, "tx_max", &tx_max, extack);
+	if (err)
+		return err;
+
+	mutex_lock(&esw->state_lock);
+	err = esw_qos_vport_enable(esw, vport, 0, 0, extack);
+	if (err)
+		goto unlock;
+
+	err = esw_qos_set_vport_max_rate(esw, vport, tx_max, extack);
+unlock:
+	mutex_unlock(&esw->state_lock);
+	return err;
+}
+
+int mlx5_esw_devlink_rate_node_tx_share_set(struct devlink_rate *rate_node, void *priv,
+					    u64 tx_share, struct netlink_ext_ack *extack)
+{
+	struct mlx5_core_dev *dev = devlink_priv(rate_node->devlink);
+	struct mlx5_eswitch *esw = dev->priv.eswitch;
+	struct mlx5_esw_rate_group *group = priv;
+	int err;
+
+	err = esw_qos_devlink_rate_to_mbps(dev, "tx_share", &tx_share, extack);
+	if (err)
+		return err;
+
+	mutex_lock(&esw->state_lock);
+	err = esw_qos_set_group_min_rate(esw, group, tx_share, extack);
+	mutex_unlock(&esw->state_lock);
+	return err;
+}
+
+int mlx5_esw_devlink_rate_node_tx_max_set(struct devlink_rate *rate_node, void *priv,
+					  u64 tx_max, struct netlink_ext_ack *extack)
+{
+	struct mlx5_core_dev *dev = devlink_priv(rate_node->devlink);
+	struct mlx5_eswitch *esw = dev->priv.eswitch;
+	struct mlx5_esw_rate_group *group = priv;
+	int err;
+
+	err = esw_qos_devlink_rate_to_mbps(dev, "tx_max", &tx_max, extack);
+	if (err)
+		return err;
+
+	mutex_lock(&esw->state_lock);
+	err = esw_qos_set_group_max_rate(esw, group, tx_max, extack);
+	mutex_unlock(&esw->state_lock);
+	return err;
+}
+
+int mlx5_esw_devlink_rate_node_new(struct devlink_rate *rate_node, void **priv,
+				   struct netlink_ext_ack *extack)
+{
+	struct mlx5_esw_rate_group *group;
+	struct mlx5_eswitch *esw;
+	int err = 0;
+
+	esw = mlx5_devlink_eswitch_get(rate_node->devlink);
+	if (IS_ERR(esw))
+		return PTR_ERR(esw);
+
+	mutex_lock(&esw->state_lock);
+	if (esw->mode != MLX5_ESWITCH_OFFLOADS) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "Rate node creation supported only in switchdev mode");
+		err = -EOPNOTSUPP;
+		goto unlock;
+	}
+
+	group = esw_qos_create_rate_group(esw, extack);
+	if (IS_ERR(group)) {
+		err = PTR_ERR(group);
+		goto unlock;
+	}
+
+	*priv = group;
+unlock:
+	mutex_unlock(&esw->state_lock);
+	return err;
+}
+
+int mlx5_esw_devlink_rate_node_del(struct devlink_rate *rate_node, void *priv,
+				   struct netlink_ext_ack *extack)
+{
+	struct mlx5_esw_rate_group *group = priv;
+	struct mlx5_eswitch *esw;
+	int err;
+
+	esw = mlx5_devlink_eswitch_get(rate_node->devlink);
+	if (IS_ERR(esw))
+		return PTR_ERR(esw);
+
+	mutex_lock(&esw->state_lock);
+	err = esw_qos_destroy_rate_group(esw, group, extack);
+	mutex_unlock(&esw->state_lock);
+	return err;
+}
+
+int mlx5_esw_qos_vport_update_group(struct mlx5_eswitch *esw,
+				    struct mlx5_vport *vport,
+				    struct mlx5_esw_rate_group *group,
+				    struct netlink_ext_ack *extack)
+{
+	int err = 0;
+
+	mutex_lock(&esw->state_lock);
+	if (!vport->qos.enabled && !group)
+		goto unlock;
+
+	err = esw_qos_vport_enable(esw, vport, 0, 0, extack);
+	if (!err)
+		err = esw_qos_vport_update_group(esw, vport, group, extack);
+unlock:
+	mutex_unlock(&esw->state_lock);
+	return err;
+}
+
+int mlx5_esw_devlink_rate_parent_set(struct devlink_rate *devlink_rate,
+				     struct devlink_rate *parent,
+				     void *priv, void *parent_priv,
+				     struct netlink_ext_ack *extack)
+{
+	struct mlx5_esw_rate_group *group;
+	struct mlx5_vport *vport = priv;
+
+	if (!parent)
+		return mlx5_esw_qos_vport_update_group(vport->dev->priv.eswitch,
+						       vport, NULL, extack);
+
+	group = parent_priv;
+	return mlx5_esw_qos_vport_update_group(vport->dev->priv.eswitch, vport, group, extack);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.h b/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.h
new file mode 100644
index 000000000..0141e9d52
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.h
@@ -0,0 +1,31 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. All rights reserved. */
+
+#ifndef __MLX5_ESW_QOS_H__
+#define __MLX5_ESW_QOS_H__
+
+#ifdef CONFIG_MLX5_ESWITCH
+
+int mlx5_esw_qos_set_vport_rate(struct mlx5_eswitch *esw, struct mlx5_vport *evport,
+				u32 max_rate, u32 min_rate);
+void mlx5_esw_qos_vport_disable(struct mlx5_eswitch *esw, struct mlx5_vport *vport);
+
+int mlx5_esw_devlink_rate_leaf_tx_share_set(struct devlink_rate *rate_leaf, void *priv,
+					    u64 tx_share, struct netlink_ext_ack *extack);
+int mlx5_esw_devlink_rate_leaf_tx_max_set(struct devlink_rate *rate_leaf, void *priv,
+					  u64 tx_max, struct netlink_ext_ack *extack);
+int mlx5_esw_devlink_rate_node_tx_share_set(struct devlink_rate *rate_node, void *priv,
+					    u64 tx_share, struct netlink_ext_ack *extack);
+int mlx5_esw_devlink_rate_node_tx_max_set(struct devlink_rate *rate_node, void *priv,
+					  u64 tx_max, struct netlink_ext_ack *extack);
+int mlx5_esw_devlink_rate_node_new(struct devlink_rate *rate_node, void **priv,
+				   struct netlink_ext_ack *extack);
+int mlx5_esw_devlink_rate_node_del(struct devlink_rate *rate_node, void *priv,
+				   struct netlink_ext_ack *extack);
+int mlx5_esw_devlink_rate_parent_set(struct devlink_rate *devlink_rate,
+				     struct devlink_rate *parent,
+				     void *priv, void *parent_priv,
+				     struct netlink_ext_ack *extack);
+#endif
+
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/esw/vporttbl.c b/drivers/net/ethernet/mellanox/mlx5/core/esw/vporttbl.c
new file mode 100644
index 000000000..749c3957a
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/esw/vporttbl.c
@@ -0,0 +1,150 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+// Copyright (c) 2021 Mellanox Technologies.
+
+#include "eswitch.h"
+
+/* This struct is used as a key to the hash table and we need it to be packed
+ * so hash result is consistent
+ */
+struct mlx5_vport_key {
+	u32 chain;
+	u16 prio;
+	u16 vport;
+	u16 vhca_id;
+	struct esw_vport_tbl_namespace *vport_ns;
+} __packed;
+
+struct mlx5_vport_table {
+	struct hlist_node hlist;
+	struct mlx5_flow_table *fdb;
+	u32 num_rules;
+	struct mlx5_vport_key key;
+};
+
+static void
+esw_vport_tbl_init(struct mlx5_eswitch *esw, struct esw_vport_tbl_namespace *ns)
+{
+	if (esw->offloads.encap != DEVLINK_ESWITCH_ENCAP_MODE_NONE)
+		ns->flags |= (MLX5_FLOW_TABLE_TUNNEL_EN_REFORMAT |
+			      MLX5_FLOW_TABLE_TUNNEL_EN_DECAP);
+}
+
+static struct mlx5_flow_table *
+esw_vport_tbl_create(struct mlx5_eswitch *esw, struct mlx5_flow_namespace *ns,
+		     const struct esw_vport_tbl_namespace *vport_ns)
+{
+	struct mlx5_flow_table_attr ft_attr = {};
+	struct mlx5_flow_table *fdb;
+
+	if (vport_ns->max_num_groups)
+		ft_attr.autogroup.max_num_groups = vport_ns->max_num_groups;
+	else
+		ft_attr.autogroup.max_num_groups = esw->params.large_group_num;
+	ft_attr.max_fte = vport_ns->max_fte;
+	ft_attr.prio = FDB_PER_VPORT;
+	ft_attr.flags = vport_ns->flags;
+	fdb = mlx5_create_auto_grouped_flow_table(ns, &ft_attr);
+	if (IS_ERR(fdb)) {
+		esw_warn(esw->dev, "Failed to create per vport FDB Table err %ld\n",
+			 PTR_ERR(fdb));
+	}
+
+	return fdb;
+}
+
+static u32 flow_attr_to_vport_key(struct mlx5_eswitch *esw,
+				  struct mlx5_vport_tbl_attr *attr,
+				  struct mlx5_vport_key *key)
+{
+	key->vport = attr->vport;
+	key->chain = attr->chain;
+	key->prio = attr->prio;
+	key->vhca_id = MLX5_CAP_GEN(esw->dev, vhca_id);
+	key->vport_ns  = attr->vport_ns;
+	return jhash(key, sizeof(*key), 0);
+}
+
+/* caller must hold vports.lock */
+static struct mlx5_vport_table *
+esw_vport_tbl_lookup(struct mlx5_eswitch *esw, struct mlx5_vport_key *skey, u32 key)
+{
+	struct mlx5_vport_table *e;
+
+	hash_for_each_possible(esw->fdb_table.offloads.vports.table, e, hlist, key)
+		if (!memcmp(&e->key, skey, sizeof(*skey)))
+			return e;
+
+	return NULL;
+}
+
+struct mlx5_flow_table *
+mlx5_esw_vporttbl_get(struct mlx5_eswitch *esw, struct mlx5_vport_tbl_attr *attr)
+{
+	struct mlx5_core_dev *dev = esw->dev;
+	struct mlx5_flow_namespace *ns;
+	struct mlx5_flow_table *fdb;
+	struct mlx5_vport_table *e;
+	struct mlx5_vport_key skey;
+	u32 hkey;
+
+	mutex_lock(&esw->fdb_table.offloads.vports.lock);
+	esw_vport_tbl_init(esw, attr->vport_ns);
+	hkey = flow_attr_to_vport_key(esw, attr, &skey);
+	e = esw_vport_tbl_lookup(esw, &skey, hkey);
+	if (e) {
+		e->num_rules++;
+		goto out;
+	}
+
+	e = kzalloc(sizeof(*e), GFP_KERNEL);
+	if (!e) {
+		fdb = ERR_PTR(-ENOMEM);
+		goto err_alloc;
+	}
+
+	ns = mlx5_get_flow_namespace(dev, MLX5_FLOW_NAMESPACE_FDB);
+	if (!ns) {
+		esw_warn(dev, "Failed to get FDB namespace\n");
+		fdb = ERR_PTR(-ENOENT);
+		goto err_ns;
+	}
+
+	fdb = esw_vport_tbl_create(esw, ns, attr->vport_ns);
+	if (IS_ERR(fdb))
+		goto err_ns;
+
+	e->fdb = fdb;
+	e->num_rules = 1;
+	e->key = skey;
+	hash_add(esw->fdb_table.offloads.vports.table, &e->hlist, hkey);
+out:
+	mutex_unlock(&esw->fdb_table.offloads.vports.lock);
+	return e->fdb;
+
+err_ns:
+	kfree(e);
+err_alloc:
+	mutex_unlock(&esw->fdb_table.offloads.vports.lock);
+	return fdb;
+}
+
+void
+mlx5_esw_vporttbl_put(struct mlx5_eswitch *esw, struct mlx5_vport_tbl_attr *attr)
+{
+	struct mlx5_vport_table *e;
+	struct mlx5_vport_key key;
+	u32 hkey;
+
+	mutex_lock(&esw->fdb_table.offloads.vports.lock);
+	esw_vport_tbl_init(esw, attr->vport_ns);
+	hkey = flow_attr_to_vport_key(esw, attr, &key);
+	e = esw_vport_tbl_lookup(esw, &key, hkey);
+	if (!e || --e->num_rules)
+		goto out;
+
+	hash_del(&e->hlist);
+	mlx5_destroy_flow_table(e->fdb);
+	kfree(e);
+out:
+	mutex_unlock(&esw->fdb_table.offloads.vports.lock);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.c b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.c
new file mode 100644
index 000000000..48939c72b
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.c
@@ -0,0 +1,2075 @@
+/*
+ * Copyright (c) 2015, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/etherdevice.h>
+#include <linux/mlx5/driver.h>
+#include <linux/mlx5/mlx5_ifc.h>
+#include <linux/mlx5/vport.h>
+#include <linux/mlx5/fs.h>
+#include <linux/mlx5/mpfs.h>
+#include <linux/debugfs.h>
+#include "esw/acl/lgcy.h"
+#include "esw/legacy.h"
+#include "esw/qos.h"
+#include "mlx5_core.h"
+#include "lib/eq.h"
+#include "eswitch.h"
+#include "fs_core.h"
+#include "devlink.h"
+#include "ecpf.h"
+#include "en/mod_hdr.h"
+
+enum {
+	MLX5_ACTION_NONE = 0,
+	MLX5_ACTION_ADD  = 1,
+	MLX5_ACTION_DEL  = 2,
+};
+
+/* Vport UC/MC hash node */
+struct vport_addr {
+	struct l2addr_node     node;
+	u8                     action;
+	u16                    vport;
+	struct mlx5_flow_handle *flow_rule;
+	bool mpfs; /* UC MAC was added to MPFs */
+	/* A flag indicating that mac was added due to mc promiscuous vport */
+	bool mc_promisc;
+};
+
+static int mlx5_eswitch_check(const struct mlx5_core_dev *dev)
+{
+	if (MLX5_CAP_GEN(dev, port_type) != MLX5_CAP_PORT_TYPE_ETH)
+		return -EOPNOTSUPP;
+
+	if (!MLX5_ESWITCH_MANAGER(dev))
+		return -EOPNOTSUPP;
+
+	return 0;
+}
+
+struct mlx5_eswitch *mlx5_devlink_eswitch_get(struct devlink *devlink)
+{
+	struct mlx5_core_dev *dev = devlink_priv(devlink);
+	int err;
+
+	err = mlx5_eswitch_check(dev);
+	if (err)
+		return ERR_PTR(err);
+
+	return dev->priv.eswitch;
+}
+
+struct mlx5_vport *__must_check
+mlx5_eswitch_get_vport(struct mlx5_eswitch *esw, u16 vport_num)
+{
+	struct mlx5_vport *vport;
+
+	if (!esw || !MLX5_CAP_GEN(esw->dev, vport_group_manager))
+		return ERR_PTR(-EPERM);
+
+	vport = xa_load(&esw->vports, vport_num);
+	if (!vport) {
+		esw_debug(esw->dev, "vport out of range: num(0x%x)\n", vport_num);
+		return ERR_PTR(-EINVAL);
+	}
+	return vport;
+}
+
+static int arm_vport_context_events_cmd(struct mlx5_core_dev *dev, u16 vport,
+					u32 events_mask)
+{
+	u32 in[MLX5_ST_SZ_DW(modify_nic_vport_context_in)] = {};
+	void *nic_vport_ctx;
+
+	MLX5_SET(modify_nic_vport_context_in, in,
+		 opcode, MLX5_CMD_OP_MODIFY_NIC_VPORT_CONTEXT);
+	MLX5_SET(modify_nic_vport_context_in, in, field_select.change_event, 1);
+	MLX5_SET(modify_nic_vport_context_in, in, vport_number, vport);
+	MLX5_SET(modify_nic_vport_context_in, in, other_vport, 1);
+	nic_vport_ctx = MLX5_ADDR_OF(modify_nic_vport_context_in,
+				     in, nic_vport_context);
+
+	MLX5_SET(nic_vport_context, nic_vport_ctx, arm_change_event, 1);
+
+	if (events_mask & MLX5_VPORT_UC_ADDR_CHANGE)
+		MLX5_SET(nic_vport_context, nic_vport_ctx,
+			 event_on_uc_address_change, 1);
+	if (events_mask & MLX5_VPORT_MC_ADDR_CHANGE)
+		MLX5_SET(nic_vport_context, nic_vport_ctx,
+			 event_on_mc_address_change, 1);
+	if (events_mask & MLX5_VPORT_PROMISC_CHANGE)
+		MLX5_SET(nic_vport_context, nic_vport_ctx,
+			 event_on_promisc_change, 1);
+
+	return mlx5_cmd_exec_in(dev, modify_nic_vport_context, in);
+}
+
+/* E-Switch vport context HW commands */
+int mlx5_eswitch_modify_esw_vport_context(struct mlx5_core_dev *dev, u16 vport,
+					  bool other_vport, void *in)
+{
+	MLX5_SET(modify_esw_vport_context_in, in, opcode,
+		 MLX5_CMD_OP_MODIFY_ESW_VPORT_CONTEXT);
+	MLX5_SET(modify_esw_vport_context_in, in, vport_number, vport);
+	MLX5_SET(modify_esw_vport_context_in, in, other_vport, other_vport);
+	return mlx5_cmd_exec_in(dev, modify_esw_vport_context, in);
+}
+
+static int modify_esw_vport_cvlan(struct mlx5_core_dev *dev, u16 vport,
+				  u16 vlan, u8 qos, u8 set_flags)
+{
+	u32 in[MLX5_ST_SZ_DW(modify_esw_vport_context_in)] = {};
+
+	if (!MLX5_CAP_ESW(dev, vport_cvlan_strip) ||
+	    !MLX5_CAP_ESW(dev, vport_cvlan_insert_if_not_exist))
+		return -EOPNOTSUPP;
+
+	esw_debug(dev, "Set Vport[%d] VLAN %d qos %d set=%x\n",
+		  vport, vlan, qos, set_flags);
+
+	if (set_flags & SET_VLAN_STRIP)
+		MLX5_SET(modify_esw_vport_context_in, in,
+			 esw_vport_context.vport_cvlan_strip, 1);
+
+	if (set_flags & SET_VLAN_INSERT) {
+		if (MLX5_CAP_ESW(dev, vport_cvlan_insert_always)) {
+			/* insert either if vlan exist in packet or not */
+			MLX5_SET(modify_esw_vport_context_in, in,
+				 esw_vport_context.vport_cvlan_insert,
+				 MLX5_VPORT_CVLAN_INSERT_ALWAYS);
+		} else {
+			/* insert only if no vlan in packet */
+			MLX5_SET(modify_esw_vport_context_in, in,
+				 esw_vport_context.vport_cvlan_insert,
+				 MLX5_VPORT_CVLAN_INSERT_WHEN_NO_CVLAN);
+		}
+		MLX5_SET(modify_esw_vport_context_in, in,
+			 esw_vport_context.cvlan_pcp, qos);
+		MLX5_SET(modify_esw_vport_context_in, in,
+			 esw_vport_context.cvlan_id, vlan);
+	}
+
+	MLX5_SET(modify_esw_vport_context_in, in,
+		 field_select.vport_cvlan_strip, 1);
+	MLX5_SET(modify_esw_vport_context_in, in,
+		 field_select.vport_cvlan_insert, 1);
+
+	return mlx5_eswitch_modify_esw_vport_context(dev, vport, true, in);
+}
+
+/* E-Switch FDB */
+static struct mlx5_flow_handle *
+__esw_fdb_set_vport_rule(struct mlx5_eswitch *esw, u16 vport, bool rx_rule,
+			 u8 mac_c[ETH_ALEN], u8 mac_v[ETH_ALEN])
+{
+	int match_header = (is_zero_ether_addr(mac_c) ? 0 :
+			    MLX5_MATCH_OUTER_HEADERS);
+	struct mlx5_flow_handle *flow_rule = NULL;
+	struct mlx5_flow_act flow_act = {0};
+	struct mlx5_flow_destination dest = {};
+	struct mlx5_flow_spec *spec;
+	void *mv_misc = NULL;
+	void *mc_misc = NULL;
+	u8 *dmac_v = NULL;
+	u8 *dmac_c = NULL;
+
+	if (rx_rule)
+		match_header |= MLX5_MATCH_MISC_PARAMETERS;
+
+	spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
+	if (!spec)
+		return NULL;
+
+	dmac_v = MLX5_ADDR_OF(fte_match_param, spec->match_value,
+			      outer_headers.dmac_47_16);
+	dmac_c = MLX5_ADDR_OF(fte_match_param, spec->match_criteria,
+			      outer_headers.dmac_47_16);
+
+	if (match_header & MLX5_MATCH_OUTER_HEADERS) {
+		ether_addr_copy(dmac_v, mac_v);
+		ether_addr_copy(dmac_c, mac_c);
+	}
+
+	if (match_header & MLX5_MATCH_MISC_PARAMETERS) {
+		mv_misc  = MLX5_ADDR_OF(fte_match_param, spec->match_value,
+					misc_parameters);
+		mc_misc  = MLX5_ADDR_OF(fte_match_param, spec->match_criteria,
+					misc_parameters);
+		MLX5_SET(fte_match_set_misc, mv_misc, source_port, MLX5_VPORT_UPLINK);
+		MLX5_SET_TO_ONES(fte_match_set_misc, mc_misc, source_port);
+	}
+
+	dest.type = MLX5_FLOW_DESTINATION_TYPE_VPORT;
+	dest.vport.num = vport;
+
+	esw_debug(esw->dev,
+		  "\tFDB add rule dmac_v(%pM) dmac_c(%pM) -> vport(%d)\n",
+		  dmac_v, dmac_c, vport);
+	spec->match_criteria_enable = match_header;
+	flow_act.action =  MLX5_FLOW_CONTEXT_ACTION_FWD_DEST;
+	flow_rule =
+		mlx5_add_flow_rules(esw->fdb_table.legacy.fdb, spec,
+				    &flow_act, &dest, 1);
+	if (IS_ERR(flow_rule)) {
+		esw_warn(esw->dev,
+			 "FDB: Failed to add flow rule: dmac_v(%pM) dmac_c(%pM) -> vport(%d), err(%ld)\n",
+			 dmac_v, dmac_c, vport, PTR_ERR(flow_rule));
+		flow_rule = NULL;
+	}
+
+	kvfree(spec);
+	return flow_rule;
+}
+
+static struct mlx5_flow_handle *
+esw_fdb_set_vport_rule(struct mlx5_eswitch *esw, u8 mac[ETH_ALEN], u16 vport)
+{
+	u8 mac_c[ETH_ALEN];
+
+	eth_broadcast_addr(mac_c);
+	return __esw_fdb_set_vport_rule(esw, vport, false, mac_c, mac);
+}
+
+static struct mlx5_flow_handle *
+esw_fdb_set_vport_allmulti_rule(struct mlx5_eswitch *esw, u16 vport)
+{
+	u8 mac_c[ETH_ALEN];
+	u8 mac_v[ETH_ALEN];
+
+	eth_zero_addr(mac_c);
+	eth_zero_addr(mac_v);
+	mac_c[0] = 0x01;
+	mac_v[0] = 0x01;
+	return __esw_fdb_set_vport_rule(esw, vport, false, mac_c, mac_v);
+}
+
+static struct mlx5_flow_handle *
+esw_fdb_set_vport_promisc_rule(struct mlx5_eswitch *esw, u16 vport)
+{
+	u8 mac_c[ETH_ALEN];
+	u8 mac_v[ETH_ALEN];
+
+	eth_zero_addr(mac_c);
+	eth_zero_addr(mac_v);
+	return __esw_fdb_set_vport_rule(esw, vport, true, mac_c, mac_v);
+}
+
+/* E-Switch vport UC/MC lists management */
+typedef int (*vport_addr_action)(struct mlx5_eswitch *esw,
+				 struct vport_addr *vaddr);
+
+static int esw_add_uc_addr(struct mlx5_eswitch *esw, struct vport_addr *vaddr)
+{
+	u8 *mac = vaddr->node.addr;
+	u16 vport = vaddr->vport;
+	int err;
+
+	/* Skip mlx5_mpfs_add_mac for eswitch_managers,
+	 * it is already done by its netdev in mlx5e_execute_l2_action
+	 */
+	if (mlx5_esw_is_manager_vport(esw, vport))
+		goto fdb_add;
+
+	err = mlx5_mpfs_add_mac(esw->dev, mac);
+	if (err) {
+		esw_warn(esw->dev,
+			 "Failed to add L2 table mac(%pM) for vport(0x%x), err(%d)\n",
+			 mac, vport, err);
+		return err;
+	}
+	vaddr->mpfs = true;
+
+fdb_add:
+	/* SRIOV is enabled: Forward UC MAC to vport */
+	if (esw->fdb_table.legacy.fdb && esw->mode == MLX5_ESWITCH_LEGACY)
+		vaddr->flow_rule = esw_fdb_set_vport_rule(esw, mac, vport);
+
+	esw_debug(esw->dev, "\tADDED UC MAC: vport[%d] %pM fr(%p)\n",
+		  vport, mac, vaddr->flow_rule);
+
+	return 0;
+}
+
+static int esw_del_uc_addr(struct mlx5_eswitch *esw, struct vport_addr *vaddr)
+{
+	u8 *mac = vaddr->node.addr;
+	u16 vport = vaddr->vport;
+	int err = 0;
+
+	/* Skip mlx5_mpfs_del_mac for eswitch managers,
+	 * it is already done by its netdev in mlx5e_execute_l2_action
+	 */
+	if (!vaddr->mpfs || mlx5_esw_is_manager_vport(esw, vport))
+		goto fdb_del;
+
+	err = mlx5_mpfs_del_mac(esw->dev, mac);
+	if (err)
+		esw_warn(esw->dev,
+			 "Failed to del L2 table mac(%pM) for vport(%d), err(%d)\n",
+			 mac, vport, err);
+	vaddr->mpfs = false;
+
+fdb_del:
+	if (vaddr->flow_rule)
+		mlx5_del_flow_rules(vaddr->flow_rule);
+	vaddr->flow_rule = NULL;
+
+	return 0;
+}
+
+static void update_allmulti_vports(struct mlx5_eswitch *esw,
+				   struct vport_addr *vaddr,
+				   struct esw_mc_addr *esw_mc)
+{
+	u8 *mac = vaddr->node.addr;
+	struct mlx5_vport *vport;
+	unsigned long i;
+	u16 vport_num;
+
+	mlx5_esw_for_each_vport(esw, i, vport) {
+		struct hlist_head *vport_hash = vport->mc_list;
+		struct vport_addr *iter_vaddr =
+					l2addr_hash_find(vport_hash,
+							 mac,
+							 struct vport_addr);
+		vport_num = vport->vport;
+		if (IS_ERR_OR_NULL(vport->allmulti_rule) ||
+		    vaddr->vport == vport_num)
+			continue;
+		switch (vaddr->action) {
+		case MLX5_ACTION_ADD:
+			if (iter_vaddr)
+				continue;
+			iter_vaddr = l2addr_hash_add(vport_hash, mac,
+						     struct vport_addr,
+						     GFP_KERNEL);
+			if (!iter_vaddr) {
+				esw_warn(esw->dev,
+					 "ALL-MULTI: Failed to add MAC(%pM) to vport[%d] DB\n",
+					 mac, vport_num);
+				continue;
+			}
+			iter_vaddr->vport = vport_num;
+			iter_vaddr->flow_rule =
+					esw_fdb_set_vport_rule(esw,
+							       mac,
+							       vport_num);
+			iter_vaddr->mc_promisc = true;
+			break;
+		case MLX5_ACTION_DEL:
+			if (!iter_vaddr)
+				continue;
+			mlx5_del_flow_rules(iter_vaddr->flow_rule);
+			l2addr_hash_del(iter_vaddr);
+			break;
+		}
+	}
+}
+
+static int esw_add_mc_addr(struct mlx5_eswitch *esw, struct vport_addr *vaddr)
+{
+	struct hlist_head *hash = esw->mc_table;
+	struct esw_mc_addr *esw_mc;
+	u8 *mac = vaddr->node.addr;
+	u16 vport = vaddr->vport;
+
+	if (!esw->fdb_table.legacy.fdb)
+		return 0;
+
+	esw_mc = l2addr_hash_find(hash, mac, struct esw_mc_addr);
+	if (esw_mc)
+		goto add;
+
+	esw_mc = l2addr_hash_add(hash, mac, struct esw_mc_addr, GFP_KERNEL);
+	if (!esw_mc)
+		return -ENOMEM;
+
+	esw_mc->uplink_rule = /* Forward MC MAC to Uplink */
+		esw_fdb_set_vport_rule(esw, mac, MLX5_VPORT_UPLINK);
+
+	/* Add this multicast mac to all the mc promiscuous vports */
+	update_allmulti_vports(esw, vaddr, esw_mc);
+
+add:
+	/* If the multicast mac is added as a result of mc promiscuous vport,
+	 * don't increment the multicast ref count
+	 */
+	if (!vaddr->mc_promisc)
+		esw_mc->refcnt++;
+
+	/* Forward MC MAC to vport */
+	vaddr->flow_rule = esw_fdb_set_vport_rule(esw, mac, vport);
+	esw_debug(esw->dev,
+		  "\tADDED MC MAC: vport[%d] %pM fr(%p) refcnt(%d) uplinkfr(%p)\n",
+		  vport, mac, vaddr->flow_rule,
+		  esw_mc->refcnt, esw_mc->uplink_rule);
+	return 0;
+}
+
+static int esw_del_mc_addr(struct mlx5_eswitch *esw, struct vport_addr *vaddr)
+{
+	struct hlist_head *hash = esw->mc_table;
+	struct esw_mc_addr *esw_mc;
+	u8 *mac = vaddr->node.addr;
+	u16 vport = vaddr->vport;
+
+	if (!esw->fdb_table.legacy.fdb)
+		return 0;
+
+	esw_mc = l2addr_hash_find(hash, mac, struct esw_mc_addr);
+	if (!esw_mc) {
+		esw_warn(esw->dev,
+			 "Failed to find eswitch MC addr for MAC(%pM) vport(%d)",
+			 mac, vport);
+		return -EINVAL;
+	}
+	esw_debug(esw->dev,
+		  "\tDELETE MC MAC: vport[%d] %pM fr(%p) refcnt(%d) uplinkfr(%p)\n",
+		  vport, mac, vaddr->flow_rule, esw_mc->refcnt,
+		  esw_mc->uplink_rule);
+
+	if (vaddr->flow_rule)
+		mlx5_del_flow_rules(vaddr->flow_rule);
+	vaddr->flow_rule = NULL;
+
+	/* If the multicast mac is added as a result of mc promiscuous vport,
+	 * don't decrement the multicast ref count.
+	 */
+	if (vaddr->mc_promisc || (--esw_mc->refcnt > 0))
+		return 0;
+
+	/* Remove this multicast mac from all the mc promiscuous vports */
+	update_allmulti_vports(esw, vaddr, esw_mc);
+
+	if (esw_mc->uplink_rule)
+		mlx5_del_flow_rules(esw_mc->uplink_rule);
+
+	l2addr_hash_del(esw_mc);
+	return 0;
+}
+
+/* Apply vport UC/MC list to HW l2 table and FDB table */
+static void esw_apply_vport_addr_list(struct mlx5_eswitch *esw,
+				      struct mlx5_vport *vport, int list_type)
+{
+	bool is_uc = list_type == MLX5_NVPRT_LIST_TYPE_UC;
+	vport_addr_action vport_addr_add;
+	vport_addr_action vport_addr_del;
+	struct vport_addr *addr;
+	struct l2addr_node *node;
+	struct hlist_head *hash;
+	struct hlist_node *tmp;
+	int hi;
+
+	vport_addr_add = is_uc ? esw_add_uc_addr :
+				 esw_add_mc_addr;
+	vport_addr_del = is_uc ? esw_del_uc_addr :
+				 esw_del_mc_addr;
+
+	hash = is_uc ? vport->uc_list : vport->mc_list;
+	for_each_l2hash_node(node, tmp, hash, hi) {
+		addr = container_of(node, struct vport_addr, node);
+		switch (addr->action) {
+		case MLX5_ACTION_ADD:
+			vport_addr_add(esw, addr);
+			addr->action = MLX5_ACTION_NONE;
+			break;
+		case MLX5_ACTION_DEL:
+			vport_addr_del(esw, addr);
+			l2addr_hash_del(addr);
+			break;
+		}
+	}
+}
+
+/* Sync vport UC/MC list from vport context */
+static void esw_update_vport_addr_list(struct mlx5_eswitch *esw,
+				       struct mlx5_vport *vport, int list_type)
+{
+	bool is_uc = list_type == MLX5_NVPRT_LIST_TYPE_UC;
+	u8 (*mac_list)[ETH_ALEN];
+	struct l2addr_node *node;
+	struct vport_addr *addr;
+	struct hlist_head *hash;
+	struct hlist_node *tmp;
+	int size;
+	int err;
+	int hi;
+	int i;
+
+	size = is_uc ? MLX5_MAX_UC_PER_VPORT(esw->dev) :
+		       MLX5_MAX_MC_PER_VPORT(esw->dev);
+
+	mac_list = kcalloc(size, ETH_ALEN, GFP_KERNEL);
+	if (!mac_list)
+		return;
+
+	hash = is_uc ? vport->uc_list : vport->mc_list;
+
+	for_each_l2hash_node(node, tmp, hash, hi) {
+		addr = container_of(node, struct vport_addr, node);
+		addr->action = MLX5_ACTION_DEL;
+	}
+
+	if (!vport->enabled)
+		goto out;
+
+	err = mlx5_query_nic_vport_mac_list(esw->dev, vport->vport, list_type,
+					    mac_list, &size);
+	if (err)
+		goto out;
+	esw_debug(esw->dev, "vport[%d] context update %s list size (%d)\n",
+		  vport->vport, is_uc ? "UC" : "MC", size);
+
+	for (i = 0; i < size; i++) {
+		if (is_uc && !is_valid_ether_addr(mac_list[i]))
+			continue;
+
+		if (!is_uc && !is_multicast_ether_addr(mac_list[i]))
+			continue;
+
+		addr = l2addr_hash_find(hash, mac_list[i], struct vport_addr);
+		if (addr) {
+			addr->action = MLX5_ACTION_NONE;
+			/* If this mac was previously added because of allmulti
+			 * promiscuous rx mode, its now converted to be original
+			 * vport mac.
+			 */
+			if (addr->mc_promisc) {
+				struct esw_mc_addr *esw_mc =
+					l2addr_hash_find(esw->mc_table,
+							 mac_list[i],
+							 struct esw_mc_addr);
+				if (!esw_mc) {
+					esw_warn(esw->dev,
+						 "Failed to MAC(%pM) in mcast DB\n",
+						 mac_list[i]);
+					continue;
+				}
+				esw_mc->refcnt++;
+				addr->mc_promisc = false;
+			}
+			continue;
+		}
+
+		addr = l2addr_hash_add(hash, mac_list[i], struct vport_addr,
+				       GFP_KERNEL);
+		if (!addr) {
+			esw_warn(esw->dev,
+				 "Failed to add MAC(%pM) to vport[%d] DB\n",
+				 mac_list[i], vport->vport);
+			continue;
+		}
+		addr->vport = vport->vport;
+		addr->action = MLX5_ACTION_ADD;
+	}
+out:
+	kfree(mac_list);
+}
+
+/* Sync vport UC/MC list from vport context
+ * Must be called after esw_update_vport_addr_list
+ */
+static void esw_update_vport_mc_promisc(struct mlx5_eswitch *esw,
+					struct mlx5_vport *vport)
+{
+	struct l2addr_node *node;
+	struct vport_addr *addr;
+	struct hlist_head *hash;
+	struct hlist_node *tmp;
+	int hi;
+
+	hash = vport->mc_list;
+
+	for_each_l2hash_node(node, tmp, esw->mc_table, hi) {
+		u8 *mac = node->addr;
+
+		addr = l2addr_hash_find(hash, mac, struct vport_addr);
+		if (addr) {
+			if (addr->action == MLX5_ACTION_DEL)
+				addr->action = MLX5_ACTION_NONE;
+			continue;
+		}
+		addr = l2addr_hash_add(hash, mac, struct vport_addr,
+				       GFP_KERNEL);
+		if (!addr) {
+			esw_warn(esw->dev,
+				 "Failed to add allmulti MAC(%pM) to vport[%d] DB\n",
+				 mac, vport->vport);
+			continue;
+		}
+		addr->vport = vport->vport;
+		addr->action = MLX5_ACTION_ADD;
+		addr->mc_promisc = true;
+	}
+}
+
+/* Apply vport rx mode to HW FDB table */
+static void esw_apply_vport_rx_mode(struct mlx5_eswitch *esw,
+				    struct mlx5_vport *vport,
+				    bool promisc, bool mc_promisc)
+{
+	struct esw_mc_addr *allmulti_addr = &esw->mc_promisc;
+
+	if (IS_ERR_OR_NULL(vport->allmulti_rule) != mc_promisc)
+		goto promisc;
+
+	if (mc_promisc) {
+		vport->allmulti_rule =
+			esw_fdb_set_vport_allmulti_rule(esw, vport->vport);
+		if (!allmulti_addr->uplink_rule)
+			allmulti_addr->uplink_rule =
+				esw_fdb_set_vport_allmulti_rule(esw,
+								MLX5_VPORT_UPLINK);
+		allmulti_addr->refcnt++;
+	} else if (vport->allmulti_rule) {
+		mlx5_del_flow_rules(vport->allmulti_rule);
+		vport->allmulti_rule = NULL;
+
+		if (--allmulti_addr->refcnt > 0)
+			goto promisc;
+
+		if (allmulti_addr->uplink_rule)
+			mlx5_del_flow_rules(allmulti_addr->uplink_rule);
+		allmulti_addr->uplink_rule = NULL;
+	}
+
+promisc:
+	if (IS_ERR_OR_NULL(vport->promisc_rule) != promisc)
+		return;
+
+	if (promisc) {
+		vport->promisc_rule =
+			esw_fdb_set_vport_promisc_rule(esw, vport->vport);
+	} else if (vport->promisc_rule) {
+		mlx5_del_flow_rules(vport->promisc_rule);
+		vport->promisc_rule = NULL;
+	}
+}
+
+/* Sync vport rx mode from vport context */
+static void esw_update_vport_rx_mode(struct mlx5_eswitch *esw,
+				     struct mlx5_vport *vport)
+{
+	int promisc_all = 0;
+	int promisc_uc = 0;
+	int promisc_mc = 0;
+	int err;
+
+	err = mlx5_query_nic_vport_promisc(esw->dev,
+					   vport->vport,
+					   &promisc_uc,
+					   &promisc_mc,
+					   &promisc_all);
+	if (err)
+		return;
+	esw_debug(esw->dev, "vport[%d] context update rx mode promisc_all=%d, all_multi=%d\n",
+		  vport->vport, promisc_all, promisc_mc);
+
+	if (!vport->info.trusted || !vport->enabled) {
+		promisc_uc = 0;
+		promisc_mc = 0;
+		promisc_all = 0;
+	}
+
+	esw_apply_vport_rx_mode(esw, vport, promisc_all,
+				(promisc_all || promisc_mc));
+}
+
+void esw_vport_change_handle_locked(struct mlx5_vport *vport)
+{
+	struct mlx5_core_dev *dev = vport->dev;
+	struct mlx5_eswitch *esw = dev->priv.eswitch;
+	u8 mac[ETH_ALEN];
+
+	mlx5_query_nic_vport_mac_address(dev, vport->vport, true, mac);
+	esw_debug(dev, "vport[%d] Context Changed: perm mac: %pM\n",
+		  vport->vport, mac);
+
+	if (vport->enabled_events & MLX5_VPORT_UC_ADDR_CHANGE) {
+		esw_update_vport_addr_list(esw, vport, MLX5_NVPRT_LIST_TYPE_UC);
+		esw_apply_vport_addr_list(esw, vport, MLX5_NVPRT_LIST_TYPE_UC);
+	}
+
+	if (vport->enabled_events & MLX5_VPORT_MC_ADDR_CHANGE)
+		esw_update_vport_addr_list(esw, vport, MLX5_NVPRT_LIST_TYPE_MC);
+
+	if (vport->enabled_events & MLX5_VPORT_PROMISC_CHANGE) {
+		esw_update_vport_rx_mode(esw, vport);
+		if (!IS_ERR_OR_NULL(vport->allmulti_rule))
+			esw_update_vport_mc_promisc(esw, vport);
+	}
+
+	if (vport->enabled_events & (MLX5_VPORT_PROMISC_CHANGE | MLX5_VPORT_MC_ADDR_CHANGE))
+		esw_apply_vport_addr_list(esw, vport, MLX5_NVPRT_LIST_TYPE_MC);
+
+	esw_debug(esw->dev, "vport[%d] Context Changed: Done\n", vport->vport);
+	if (vport->enabled)
+		arm_vport_context_events_cmd(dev, vport->vport,
+					     vport->enabled_events);
+}
+
+static void esw_vport_change_handler(struct work_struct *work)
+{
+	struct mlx5_vport *vport =
+		container_of(work, struct mlx5_vport, vport_change_handler);
+	struct mlx5_eswitch *esw = vport->dev->priv.eswitch;
+
+	mutex_lock(&esw->state_lock);
+	esw_vport_change_handle_locked(vport);
+	mutex_unlock(&esw->state_lock);
+}
+
+static void node_guid_gen_from_mac(u64 *node_guid, const u8 *mac)
+{
+	((u8 *)node_guid)[7] = mac[0];
+	((u8 *)node_guid)[6] = mac[1];
+	((u8 *)node_guid)[5] = mac[2];
+	((u8 *)node_guid)[4] = 0xff;
+	((u8 *)node_guid)[3] = 0xfe;
+	((u8 *)node_guid)[2] = mac[3];
+	((u8 *)node_guid)[1] = mac[4];
+	((u8 *)node_guid)[0] = mac[5];
+}
+
+static int esw_vport_setup_acl(struct mlx5_eswitch *esw,
+			       struct mlx5_vport *vport)
+{
+	if (esw->mode == MLX5_ESWITCH_LEGACY)
+		return esw_legacy_vport_acl_setup(esw, vport);
+	else
+		return esw_vport_create_offloads_acl_tables(esw, vport);
+}
+
+static void esw_vport_cleanup_acl(struct mlx5_eswitch *esw,
+				  struct mlx5_vport *vport)
+{
+	if (esw->mode == MLX5_ESWITCH_LEGACY)
+		esw_legacy_vport_acl_cleanup(esw, vport);
+	else
+		esw_vport_destroy_offloads_acl_tables(esw, vport);
+}
+
+static int esw_vport_setup(struct mlx5_eswitch *esw, struct mlx5_vport *vport)
+{
+	bool vst_mode_steering = esw_vst_mode_is_steering(esw);
+	u16 vport_num = vport->vport;
+	int flags;
+	int err;
+
+	err = esw_vport_setup_acl(esw, vport);
+	if (err)
+		return err;
+
+	if (mlx5_esw_is_manager_vport(esw, vport_num))
+		return 0;
+
+	mlx5_modify_vport_admin_state(esw->dev,
+				      MLX5_VPORT_STATE_OP_MOD_ESW_VPORT,
+				      vport_num, 1,
+				      vport->info.link_state);
+
+	/* Host PF has its own mac/guid. */
+	if (vport_num) {
+		mlx5_modify_nic_vport_mac_address(esw->dev, vport_num,
+						  vport->info.mac);
+		mlx5_modify_nic_vport_node_guid(esw->dev, vport_num,
+						vport->info.node_guid);
+	}
+
+	flags = (vport->info.vlan || vport->info.qos) ?
+		SET_VLAN_STRIP | SET_VLAN_INSERT : 0;
+	if (esw->mode == MLX5_ESWITCH_OFFLOADS || !vst_mode_steering)
+		modify_esw_vport_cvlan(esw->dev, vport_num, vport->info.vlan,
+				       vport->info.qos, flags);
+
+	return 0;
+}
+
+/* Don't cleanup vport->info, it's needed to restore vport configuration */
+static void esw_vport_cleanup(struct mlx5_eswitch *esw, struct mlx5_vport *vport)
+{
+	u16 vport_num = vport->vport;
+
+	if (!mlx5_esw_is_manager_vport(esw, vport_num))
+		mlx5_modify_vport_admin_state(esw->dev,
+					      MLX5_VPORT_STATE_OP_MOD_ESW_VPORT,
+					      vport_num, 1,
+					      MLX5_VPORT_ADMIN_STATE_DOWN);
+
+	mlx5_esw_qos_vport_disable(esw, vport);
+	esw_vport_cleanup_acl(esw, vport);
+}
+
+int mlx5_esw_vport_enable(struct mlx5_eswitch *esw, u16 vport_num,
+			  enum mlx5_eswitch_vport_event enabled_events)
+{
+	struct mlx5_vport *vport;
+	int ret;
+
+	vport = mlx5_eswitch_get_vport(esw, vport_num);
+	if (IS_ERR(vport))
+		return PTR_ERR(vport);
+
+	mutex_lock(&esw->state_lock);
+	WARN_ON(vport->enabled);
+
+	esw_debug(esw->dev, "Enabling VPORT(%d)\n", vport_num);
+
+	ret = esw_vport_setup(esw, vport);
+	if (ret)
+		goto done;
+
+	/* Sync with current vport context */
+	vport->enabled_events = enabled_events;
+	vport->enabled = true;
+
+	/* Esw manager is trusted by default. Host PF (vport 0) is trusted as well
+	 * in smartNIC as it's a vport group manager.
+	 */
+	if (mlx5_esw_is_manager_vport(esw, vport_num) ||
+	    (!vport_num && mlx5_core_is_ecpf(esw->dev)))
+		vport->info.trusted = true;
+
+	if (!mlx5_esw_is_manager_vport(esw, vport->vport) &&
+	    MLX5_CAP_GEN(esw->dev, vhca_resource_manager)) {
+		ret = mlx5_esw_vport_vhca_id_set(esw, vport_num);
+		if (ret)
+			goto err_vhca_mapping;
+	}
+
+	/* External controller host PF has factory programmed MAC.
+	 * Read it from the device.
+	 */
+	if (mlx5_core_is_ecpf(esw->dev) && vport_num == MLX5_VPORT_PF)
+		mlx5_query_nic_vport_mac_address(esw->dev, vport_num, true, vport->info.mac);
+
+	esw_vport_change_handle_locked(vport);
+
+	esw->enabled_vports++;
+	esw_debug(esw->dev, "Enabled VPORT(%d)\n", vport_num);
+done:
+	mutex_unlock(&esw->state_lock);
+	return ret;
+
+err_vhca_mapping:
+	esw_vport_cleanup(esw, vport);
+	mutex_unlock(&esw->state_lock);
+	return ret;
+}
+
+void mlx5_esw_vport_disable(struct mlx5_eswitch *esw, u16 vport_num)
+{
+	struct mlx5_vport *vport;
+
+	vport = mlx5_eswitch_get_vport(esw, vport_num);
+	if (IS_ERR(vport))
+		return;
+
+	mutex_lock(&esw->state_lock);
+	if (!vport->enabled)
+		goto done;
+
+	esw_debug(esw->dev, "Disabling vport(%d)\n", vport_num);
+	/* Mark this vport as disabled to discard new events */
+	vport->enabled = false;
+
+	/* Disable events from this vport */
+	arm_vport_context_events_cmd(esw->dev, vport->vport, 0);
+
+	if (!mlx5_esw_is_manager_vport(esw, vport->vport) &&
+	    MLX5_CAP_GEN(esw->dev, vhca_resource_manager))
+		mlx5_esw_vport_vhca_id_clear(esw, vport_num);
+
+	/* We don't assume VFs will cleanup after themselves.
+	 * Calling vport change handler while vport is disabled will cleanup
+	 * the vport resources.
+	 */
+	esw_vport_change_handle_locked(vport);
+	vport->enabled_events = 0;
+	esw_apply_vport_rx_mode(esw, vport, false, false);
+	esw_vport_cleanup(esw, vport);
+	esw->enabled_vports--;
+
+done:
+	mutex_unlock(&esw->state_lock);
+}
+
+static int eswitch_vport_event(struct notifier_block *nb,
+			       unsigned long type, void *data)
+{
+	struct mlx5_eswitch *esw = mlx5_nb_cof(nb, struct mlx5_eswitch, nb);
+	struct mlx5_eqe *eqe = data;
+	struct mlx5_vport *vport;
+	u16 vport_num;
+
+	vport_num = be16_to_cpu(eqe->data.vport_change.vport_num);
+	vport = mlx5_eswitch_get_vport(esw, vport_num);
+	if (!IS_ERR(vport))
+		queue_work(esw->work_queue, &vport->vport_change_handler);
+	return NOTIFY_OK;
+}
+
+/**
+ * mlx5_esw_query_functions - Returns raw output about functions state
+ * @dev:	Pointer to device to query
+ *
+ * mlx5_esw_query_functions() allocates and returns functions changed
+ * raw output memory pointer from device on success. Otherwise returns ERR_PTR.
+ * Caller must free the memory using kvfree() when valid pointer is returned.
+ */
+const u32 *mlx5_esw_query_functions(struct mlx5_core_dev *dev)
+{
+	int outlen = MLX5_ST_SZ_BYTES(query_esw_functions_out);
+	u32 in[MLX5_ST_SZ_DW(query_esw_functions_in)] = {};
+	u32 *out;
+	int err;
+
+	out = kvzalloc(outlen, GFP_KERNEL);
+	if (!out)
+		return ERR_PTR(-ENOMEM);
+
+	MLX5_SET(query_esw_functions_in, in, opcode,
+		 MLX5_CMD_OP_QUERY_ESW_FUNCTIONS);
+
+	err = mlx5_cmd_exec(dev, in, sizeof(in), out, outlen);
+	if (!err)
+		return out;
+
+	kvfree(out);
+	return ERR_PTR(err);
+}
+
+static void mlx5_eswitch_event_handler_register(struct mlx5_eswitch *esw)
+{
+	if (esw->mode == MLX5_ESWITCH_OFFLOADS && mlx5_eswitch_is_funcs_handler(esw->dev)) {
+		MLX5_NB_INIT(&esw->esw_funcs.nb, mlx5_esw_funcs_changed_handler,
+			     ESW_FUNCTIONS_CHANGED);
+		mlx5_eq_notifier_register(esw->dev, &esw->esw_funcs.nb);
+	}
+}
+
+static void mlx5_eswitch_event_handler_unregister(struct mlx5_eswitch *esw)
+{
+	if (esw->mode == MLX5_ESWITCH_OFFLOADS && mlx5_eswitch_is_funcs_handler(esw->dev))
+		mlx5_eq_notifier_unregister(esw->dev, &esw->esw_funcs.nb);
+
+	flush_workqueue(esw->work_queue);
+}
+
+static void mlx5_eswitch_clear_vf_vports_info(struct mlx5_eswitch *esw)
+{
+	struct mlx5_vport *vport;
+	unsigned long i;
+
+	mlx5_esw_for_each_vf_vport(esw, i, vport, esw->esw_funcs.num_vfs) {
+		memset(&vport->qos, 0, sizeof(vport->qos));
+		memset(&vport->info, 0, sizeof(vport->info));
+		vport->info.link_state = MLX5_VPORT_ADMIN_STATE_AUTO;
+	}
+}
+
+/* Public E-Switch API */
+int mlx5_eswitch_load_vport(struct mlx5_eswitch *esw, u16 vport_num,
+			    enum mlx5_eswitch_vport_event enabled_events)
+{
+	int err;
+
+	err = mlx5_esw_vport_enable(esw, vport_num, enabled_events);
+	if (err)
+		return err;
+
+	mlx5_esw_vport_debugfs_create(esw, vport_num, false, 0);
+	err = esw_offloads_load_rep(esw, vport_num);
+	if (err)
+		goto err_rep;
+
+	return err;
+
+err_rep:
+	mlx5_esw_vport_debugfs_destroy(esw, vport_num);
+	mlx5_esw_vport_disable(esw, vport_num);
+	return err;
+}
+
+void mlx5_eswitch_unload_vport(struct mlx5_eswitch *esw, u16 vport_num)
+{
+	esw_offloads_unload_rep(esw, vport_num);
+	mlx5_esw_vport_debugfs_destroy(esw, vport_num);
+	mlx5_esw_vport_disable(esw, vport_num);
+}
+
+void mlx5_eswitch_unload_vf_vports(struct mlx5_eswitch *esw, u16 num_vfs)
+{
+	struct mlx5_vport *vport;
+	unsigned long i;
+
+	mlx5_esw_for_each_vf_vport(esw, i, vport, num_vfs) {
+		if (!vport->enabled)
+			continue;
+		mlx5_eswitch_unload_vport(esw, vport->vport);
+	}
+}
+
+int mlx5_eswitch_load_vf_vports(struct mlx5_eswitch *esw, u16 num_vfs,
+				enum mlx5_eswitch_vport_event enabled_events)
+{
+	struct mlx5_vport *vport;
+	unsigned long i;
+	int err;
+
+	mlx5_esw_for_each_vf_vport(esw, i, vport, num_vfs) {
+		err = mlx5_eswitch_load_vport(esw, vport->vport, enabled_events);
+		if (err)
+			goto vf_err;
+	}
+
+	return 0;
+
+vf_err:
+	mlx5_eswitch_unload_vf_vports(esw, num_vfs);
+	return err;
+}
+
+static int host_pf_enable_hca(struct mlx5_core_dev *dev)
+{
+	if (!mlx5_core_is_ecpf(dev))
+		return 0;
+
+	/* Once vport and representor are ready, take out the external host PF
+	 * out of initializing state. Enabling HCA clears the iser->initializing
+	 * bit and host PF driver loading can progress.
+	 */
+	return mlx5_cmd_host_pf_enable_hca(dev);
+}
+
+static void host_pf_disable_hca(struct mlx5_core_dev *dev)
+{
+	if (!mlx5_core_is_ecpf(dev))
+		return;
+
+	mlx5_cmd_host_pf_disable_hca(dev);
+}
+
+/* mlx5_eswitch_enable_pf_vf_vports() enables vports of PF, ECPF and VFs
+ * whichever are present on the eswitch.
+ */
+int
+mlx5_eswitch_enable_pf_vf_vports(struct mlx5_eswitch *esw,
+				 enum mlx5_eswitch_vport_event enabled_events)
+{
+	int ret;
+
+	/* Enable PF vport */
+	ret = mlx5_eswitch_load_vport(esw, MLX5_VPORT_PF, enabled_events);
+	if (ret)
+		return ret;
+
+	/* Enable external host PF HCA */
+	ret = host_pf_enable_hca(esw->dev);
+	if (ret)
+		goto pf_hca_err;
+
+	/* Enable ECPF vport */
+	if (mlx5_ecpf_vport_exists(esw->dev)) {
+		ret = mlx5_eswitch_load_vport(esw, MLX5_VPORT_ECPF, enabled_events);
+		if (ret)
+			goto ecpf_err;
+	}
+
+	/* Enable VF vports */
+	ret = mlx5_eswitch_load_vf_vports(esw, esw->esw_funcs.num_vfs,
+					  enabled_events);
+	if (ret)
+		goto vf_err;
+	return 0;
+
+vf_err:
+	if (mlx5_ecpf_vport_exists(esw->dev))
+		mlx5_eswitch_unload_vport(esw, MLX5_VPORT_ECPF);
+ecpf_err:
+	host_pf_disable_hca(esw->dev);
+pf_hca_err:
+	mlx5_eswitch_unload_vport(esw, MLX5_VPORT_PF);
+	return ret;
+}
+
+/* mlx5_eswitch_disable_pf_vf_vports() disables vports of PF, ECPF and VFs
+ * whichever are previously enabled on the eswitch.
+ */
+void mlx5_eswitch_disable_pf_vf_vports(struct mlx5_eswitch *esw)
+{
+	mlx5_eswitch_unload_vf_vports(esw, esw->esw_funcs.num_vfs);
+
+	if (mlx5_ecpf_vport_exists(esw->dev))
+		mlx5_eswitch_unload_vport(esw, MLX5_VPORT_ECPF);
+
+	host_pf_disable_hca(esw->dev);
+	mlx5_eswitch_unload_vport(esw, MLX5_VPORT_PF);
+}
+
+static void mlx5_eswitch_get_devlink_param(struct mlx5_eswitch *esw)
+{
+	struct devlink *devlink = priv_to_devlink(esw->dev);
+	union devlink_param_value val;
+	int err;
+
+	err = devlink_param_driverinit_value_get(devlink,
+						 MLX5_DEVLINK_PARAM_ID_ESW_LARGE_GROUP_NUM,
+						 &val);
+	if (!err) {
+		esw->params.large_group_num = val.vu32;
+	} else {
+		esw_warn(esw->dev,
+			 "Devlink can't get param fdb_large_groups, uses default (%d).\n",
+			 ESW_OFFLOADS_DEFAULT_NUM_GROUPS);
+		esw->params.large_group_num = ESW_OFFLOADS_DEFAULT_NUM_GROUPS;
+	}
+}
+
+static void
+mlx5_eswitch_update_num_of_vfs(struct mlx5_eswitch *esw, int num_vfs)
+{
+	const u32 *out;
+
+	if (num_vfs < 0)
+		return;
+
+	if (!mlx5_core_is_ecpf_esw_manager(esw->dev)) {
+		esw->esw_funcs.num_vfs = num_vfs;
+		return;
+	}
+
+	out = mlx5_esw_query_functions(esw->dev);
+	if (IS_ERR(out))
+		return;
+
+	esw->esw_funcs.num_vfs = MLX5_GET(query_esw_functions_out, out,
+					  host_params_context.host_num_of_vfs);
+	kvfree(out);
+}
+
+static void mlx5_esw_mode_change_notify(struct mlx5_eswitch *esw, u16 mode)
+{
+	struct mlx5_esw_event_info info = {};
+
+	info.new_mode = mode;
+
+	blocking_notifier_call_chain(&esw->n_head, 0, &info);
+}
+
+static int mlx5_esw_acls_ns_init(struct mlx5_eswitch *esw)
+{
+	struct mlx5_core_dev *dev = esw->dev;
+	int total_vports;
+	int err;
+
+	if (esw->flags & MLX5_ESWITCH_VPORT_ACL_NS_CREATED)
+		return 0;
+
+	total_vports = mlx5_eswitch_get_total_vports(dev);
+
+	if (MLX5_CAP_ESW_EGRESS_ACL(dev, ft_support)) {
+		err = mlx5_fs_egress_acls_init(dev, total_vports);
+		if (err)
+			return err;
+	} else {
+		esw_warn(dev, "engress ACL is not supported by FW\n");
+	}
+
+	if (MLX5_CAP_ESW_INGRESS_ACL(dev, ft_support)) {
+		err = mlx5_fs_ingress_acls_init(dev, total_vports);
+		if (err)
+			goto err;
+	} else {
+		esw_warn(dev, "ingress ACL is not supported by FW\n");
+	}
+	esw->flags |= MLX5_ESWITCH_VPORT_ACL_NS_CREATED;
+	return 0;
+
+err:
+	if (MLX5_CAP_ESW_EGRESS_ACL(dev, ft_support))
+		mlx5_fs_egress_acls_cleanup(dev);
+	return err;
+}
+
+static void mlx5_esw_acls_ns_cleanup(struct mlx5_eswitch *esw)
+{
+	struct mlx5_core_dev *dev = esw->dev;
+
+	esw->flags &= ~MLX5_ESWITCH_VPORT_ACL_NS_CREATED;
+	if (MLX5_CAP_ESW_INGRESS_ACL(dev, ft_support))
+		mlx5_fs_ingress_acls_cleanup(dev);
+	if (MLX5_CAP_ESW_EGRESS_ACL(dev, ft_support))
+		mlx5_fs_egress_acls_cleanup(dev);
+}
+
+/**
+ * mlx5_eswitch_enable_locked - Enable eswitch
+ * @esw:	Pointer to eswitch
+ * @num_vfs:	Enable eswitch for given number of VFs. This is optional.
+ *		Valid value are 0, > 0 and MLX5_ESWITCH_IGNORE_NUM_VFS.
+ *		Caller should pass num_vfs > 0 when enabling eswitch for
+ *		vf vports. Caller should pass num_vfs = 0, when eswitch
+ *		is enabled without sriov VFs or when caller
+ *		is unaware of the sriov state of the host PF on ECPF based
+ *		eswitch. Caller should pass < 0 when num_vfs should be
+ *		completely ignored. This is typically the case when eswitch
+ *		is enabled without sriov regardless of PF/ECPF system.
+ * mlx5_eswitch_enable_locked() Enables eswitch in either legacy or offloads
+ * mode. If num_vfs >=0 is provided, it setup VF related eswitch vports.
+ * It returns 0 on success or error code on failure.
+ */
+int mlx5_eswitch_enable_locked(struct mlx5_eswitch *esw, int num_vfs)
+{
+	int err;
+
+	lockdep_assert_held(&esw->mode_lock);
+
+	if (!MLX5_CAP_ESW_FLOWTABLE_FDB(esw->dev, ft_support)) {
+		esw_warn(esw->dev, "FDB is not supported, aborting ...\n");
+		return -EOPNOTSUPP;
+	}
+
+	mlx5_eswitch_get_devlink_param(esw);
+
+	err = mlx5_esw_acls_ns_init(esw);
+	if (err)
+		return err;
+
+	mlx5_eswitch_update_num_of_vfs(esw, num_vfs);
+
+	MLX5_NB_INIT(&esw->nb, eswitch_vport_event, NIC_VPORT_CHANGE);
+	mlx5_eq_notifier_register(esw->dev, &esw->nb);
+
+	if (esw->mode == MLX5_ESWITCH_LEGACY) {
+		err = esw_legacy_enable(esw);
+	} else {
+		mlx5_rescan_drivers(esw->dev);
+		err = esw_offloads_enable(esw);
+	}
+
+	if (err)
+		goto abort;
+
+	esw->fdb_table.flags |= MLX5_ESW_FDB_CREATED;
+
+	mlx5_eswitch_event_handler_register(esw);
+
+	esw_info(esw->dev, "Enable: mode(%s), nvfs(%d), active vports(%d)\n",
+		 esw->mode == MLX5_ESWITCH_LEGACY ? "LEGACY" : "OFFLOADS",
+		 esw->esw_funcs.num_vfs, esw->enabled_vports);
+
+	mlx5_esw_mode_change_notify(esw, esw->mode);
+
+	return 0;
+
+abort:
+	mlx5_esw_acls_ns_cleanup(esw);
+	return err;
+}
+
+/**
+ * mlx5_eswitch_enable - Enable eswitch
+ * @esw:	Pointer to eswitch
+ * @num_vfs:	Enable eswitch switch for given number of VFs.
+ *		Caller must pass num_vfs > 0 when enabling eswitch for
+ *		vf vports.
+ * mlx5_eswitch_enable() returns 0 on success or error code on failure.
+ */
+int mlx5_eswitch_enable(struct mlx5_eswitch *esw, int num_vfs)
+{
+	bool toggle_lag;
+	int ret;
+
+	if (!mlx5_esw_allowed(esw))
+		return 0;
+
+	devl_assert_locked(priv_to_devlink(esw->dev));
+
+	toggle_lag = !mlx5_esw_is_fdb_created(esw);
+
+	if (toggle_lag)
+		mlx5_lag_disable_change(esw->dev);
+
+	down_write(&esw->mode_lock);
+	if (!mlx5_esw_is_fdb_created(esw)) {
+		ret = mlx5_eswitch_enable_locked(esw, num_vfs);
+	} else {
+		enum mlx5_eswitch_vport_event vport_events;
+
+		vport_events = (esw->mode == MLX5_ESWITCH_LEGACY) ?
+					MLX5_LEGACY_SRIOV_VPORT_EVENTS : MLX5_VPORT_UC_ADDR_CHANGE;
+		ret = mlx5_eswitch_load_vf_vports(esw, num_vfs, vport_events);
+		if (!ret)
+			esw->esw_funcs.num_vfs = num_vfs;
+	}
+	up_write(&esw->mode_lock);
+
+	if (toggle_lag)
+		mlx5_lag_enable_change(esw->dev);
+
+	return ret;
+}
+
+/* When disabling sriov, free driver level resources. */
+void mlx5_eswitch_disable_sriov(struct mlx5_eswitch *esw, bool clear_vf)
+{
+	if (!mlx5_esw_allowed(esw))
+		return;
+
+	devl_assert_locked(priv_to_devlink(esw->dev));
+	down_write(&esw->mode_lock);
+	/* If driver is unloaded, this function is called twice by remove_one()
+	 * and mlx5_unload(). Prevent the second call.
+	 */
+	if (!esw->esw_funcs.num_vfs && !clear_vf)
+		goto unlock;
+
+	esw_info(esw->dev, "Unload vfs: mode(%s), nvfs(%d), active vports(%d)\n",
+		 esw->mode == MLX5_ESWITCH_LEGACY ? "LEGACY" : "OFFLOADS",
+		 esw->esw_funcs.num_vfs, esw->enabled_vports);
+
+	mlx5_eswitch_unload_vf_vports(esw, esw->esw_funcs.num_vfs);
+	if (clear_vf)
+		mlx5_eswitch_clear_vf_vports_info(esw);
+	/* If disabling sriov in switchdev mode, free meta rules here
+	 * because it depends on num_vfs.
+	 */
+	if (esw->mode == MLX5_ESWITCH_OFFLOADS) {
+		struct devlink *devlink = priv_to_devlink(esw->dev);
+
+		devl_rate_nodes_destroy(devlink);
+	}
+	/* Destroy legacy fdb when disabling sriov in legacy mode. */
+	if (esw->mode == MLX5_ESWITCH_LEGACY)
+		mlx5_eswitch_disable_locked(esw);
+
+	esw->esw_funcs.num_vfs = 0;
+
+unlock:
+	up_write(&esw->mode_lock);
+}
+
+/* Free resources for corresponding eswitch mode. It is called by devlink
+ * when changing eswitch mode or modprobe when unloading driver.
+ */
+void mlx5_eswitch_disable_locked(struct mlx5_eswitch *esw)
+{
+	struct devlink *devlink = priv_to_devlink(esw->dev);
+
+	/* Notify eswitch users that it is exiting from current mode.
+	 * So that it can do necessary cleanup before the eswitch is disabled.
+	 */
+	mlx5_esw_mode_change_notify(esw, MLX5_ESWITCH_LEGACY);
+
+	mlx5_eq_notifier_unregister(esw->dev, &esw->nb);
+	mlx5_eswitch_event_handler_unregister(esw);
+
+	esw_info(esw->dev, "Disable: mode(%s), nvfs(%d), active vports(%d)\n",
+		 esw->mode == MLX5_ESWITCH_LEGACY ? "LEGACY" : "OFFLOADS",
+		 esw->esw_funcs.num_vfs, esw->enabled_vports);
+
+	if (esw->fdb_table.flags & MLX5_ESW_FDB_CREATED) {
+		esw->fdb_table.flags &= ~MLX5_ESW_FDB_CREATED;
+		if (esw->mode == MLX5_ESWITCH_OFFLOADS)
+			esw_offloads_disable(esw);
+		else if (esw->mode == MLX5_ESWITCH_LEGACY)
+			esw_legacy_disable(esw);
+		mlx5_esw_acls_ns_cleanup(esw);
+	}
+
+	if (esw->mode == MLX5_ESWITCH_OFFLOADS)
+		devl_rate_nodes_destroy(devlink);
+}
+
+void mlx5_eswitch_disable(struct mlx5_eswitch *esw)
+{
+	if (!mlx5_esw_allowed(esw))
+		return;
+
+	devl_assert_locked(priv_to_devlink(esw->dev));
+	mlx5_lag_disable_change(esw->dev);
+	down_write(&esw->mode_lock);
+	mlx5_eswitch_disable_locked(esw);
+	esw->mode = MLX5_ESWITCH_LEGACY;
+	up_write(&esw->mode_lock);
+	mlx5_lag_enable_change(esw->dev);
+}
+
+static int mlx5_query_hca_cap_host_pf(struct mlx5_core_dev *dev, void *out)
+{
+	u16 opmod = (MLX5_CAP_GENERAL << 1) | (HCA_CAP_OPMOD_GET_MAX & 0x01);
+	u8 in[MLX5_ST_SZ_BYTES(query_hca_cap_in)] = {};
+
+	MLX5_SET(query_hca_cap_in, in, opcode, MLX5_CMD_OP_QUERY_HCA_CAP);
+	MLX5_SET(query_hca_cap_in, in, op_mod, opmod);
+	MLX5_SET(query_hca_cap_in, in, function_id, MLX5_VPORT_PF);
+	MLX5_SET(query_hca_cap_in, in, other_function, true);
+	return mlx5_cmd_exec_inout(dev, query_hca_cap, in, out);
+}
+
+int mlx5_esw_sf_max_hpf_functions(struct mlx5_core_dev *dev, u16 *max_sfs, u16 *sf_base_id)
+
+{
+	int query_out_sz = MLX5_ST_SZ_BYTES(query_hca_cap_out);
+	void *query_ctx;
+	void *hca_caps;
+	int err;
+
+	if (!mlx5_core_is_ecpf(dev)) {
+		*max_sfs = 0;
+		return 0;
+	}
+
+	query_ctx = kzalloc(query_out_sz, GFP_KERNEL);
+	if (!query_ctx)
+		return -ENOMEM;
+
+	err = mlx5_query_hca_cap_host_pf(dev, query_ctx);
+	if (err)
+		goto out_free;
+
+	hca_caps = MLX5_ADDR_OF(query_hca_cap_out, query_ctx, capability);
+	*max_sfs = MLX5_GET(cmd_hca_cap, hca_caps, max_num_sf);
+	*sf_base_id = MLX5_GET(cmd_hca_cap, hca_caps, sf_base_id);
+
+out_free:
+	kfree(query_ctx);
+	return err;
+}
+
+static int mlx5_esw_vport_alloc(struct mlx5_eswitch *esw, struct mlx5_core_dev *dev,
+				int index, u16 vport_num)
+{
+	struct mlx5_vport *vport;
+	int err;
+
+	vport = kzalloc(sizeof(*vport), GFP_KERNEL);
+	if (!vport)
+		return -ENOMEM;
+
+	vport->dev = esw->dev;
+	vport->vport = vport_num;
+	vport->index = index;
+	vport->info.link_state = MLX5_VPORT_ADMIN_STATE_AUTO;
+	INIT_WORK(&vport->vport_change_handler, esw_vport_change_handler);
+	err = xa_insert(&esw->vports, vport_num, vport, GFP_KERNEL);
+	if (err)
+		goto insert_err;
+
+	esw->total_vports++;
+	return 0;
+
+insert_err:
+	kfree(vport);
+	return err;
+}
+
+static void mlx5_esw_vport_free(struct mlx5_eswitch *esw, struct mlx5_vport *vport)
+{
+	xa_erase(&esw->vports, vport->vport);
+	kfree(vport);
+}
+
+static void mlx5_esw_vports_cleanup(struct mlx5_eswitch *esw)
+{
+	struct mlx5_vport *vport;
+	unsigned long i;
+
+	mlx5_esw_for_each_vport(esw, i, vport)
+		mlx5_esw_vport_free(esw, vport);
+	xa_destroy(&esw->vports);
+}
+
+static int mlx5_esw_vports_init(struct mlx5_eswitch *esw)
+{
+	struct mlx5_core_dev *dev = esw->dev;
+	u16 max_host_pf_sfs;
+	u16 base_sf_num;
+	int idx = 0;
+	int err;
+	int i;
+
+	xa_init(&esw->vports);
+
+	err = mlx5_esw_vport_alloc(esw, dev, idx, MLX5_VPORT_PF);
+	if (err)
+		goto err;
+	if (esw->first_host_vport == MLX5_VPORT_PF)
+		xa_set_mark(&esw->vports, idx, MLX5_ESW_VPT_HOST_FN);
+	idx++;
+
+	for (i = 0; i < mlx5_core_max_vfs(dev); i++) {
+		err = mlx5_esw_vport_alloc(esw, dev, idx, idx);
+		if (err)
+			goto err;
+		xa_set_mark(&esw->vports, idx, MLX5_ESW_VPT_VF);
+		xa_set_mark(&esw->vports, idx, MLX5_ESW_VPT_HOST_FN);
+		idx++;
+	}
+	base_sf_num = mlx5_sf_start_function_id(dev);
+	for (i = 0; i < mlx5_sf_max_functions(dev); i++) {
+		err = mlx5_esw_vport_alloc(esw, dev, idx, base_sf_num + i);
+		if (err)
+			goto err;
+		xa_set_mark(&esw->vports, base_sf_num + i, MLX5_ESW_VPT_SF);
+		idx++;
+	}
+
+	err = mlx5_esw_sf_max_hpf_functions(dev, &max_host_pf_sfs, &base_sf_num);
+	if (err)
+		goto err;
+	for (i = 0; i < max_host_pf_sfs; i++) {
+		err = mlx5_esw_vport_alloc(esw, dev, idx, base_sf_num + i);
+		if (err)
+			goto err;
+		xa_set_mark(&esw->vports, base_sf_num + i, MLX5_ESW_VPT_SF);
+		idx++;
+	}
+
+	if (mlx5_ecpf_vport_exists(dev)) {
+		err = mlx5_esw_vport_alloc(esw, dev, idx, MLX5_VPORT_ECPF);
+		if (err)
+			goto err;
+		idx++;
+	}
+	err = mlx5_esw_vport_alloc(esw, dev, idx, MLX5_VPORT_UPLINK);
+	if (err)
+		goto err;
+	return 0;
+
+err:
+	mlx5_esw_vports_cleanup(esw);
+	return err;
+}
+
+int mlx5_eswitch_init(struct mlx5_core_dev *dev)
+{
+	struct mlx5_eswitch *esw;
+	int err;
+
+	if (!MLX5_VPORT_MANAGER(dev))
+		return 0;
+
+	esw = kzalloc(sizeof(*esw), GFP_KERNEL);
+	if (!esw)
+		return -ENOMEM;
+
+	esw->dev = dev;
+	esw->manager_vport = mlx5_eswitch_manager_vport(dev);
+	esw->first_host_vport = mlx5_eswitch_first_host_vport_num(dev);
+
+	esw->work_queue = create_singlethread_workqueue("mlx5_esw_wq");
+	if (!esw->work_queue) {
+		err = -ENOMEM;
+		goto abort;
+	}
+
+	err = mlx5_esw_vports_init(esw);
+	if (err)
+		goto abort;
+
+	err = esw_offloads_init_reps(esw);
+	if (err)
+		goto reps_err;
+
+	mutex_init(&esw->offloads.encap_tbl_lock);
+	hash_init(esw->offloads.encap_tbl);
+	mutex_init(&esw->offloads.decap_tbl_lock);
+	hash_init(esw->offloads.decap_tbl);
+	mlx5e_mod_hdr_tbl_init(&esw->offloads.mod_hdr);
+	atomic64_set(&esw->offloads.num_flows, 0);
+	ida_init(&esw->offloads.vport_metadata_ida);
+	xa_init_flags(&esw->offloads.vhca_map, XA_FLAGS_ALLOC);
+	mutex_init(&esw->state_lock);
+	init_rwsem(&esw->mode_lock);
+	refcount_set(&esw->qos.refcnt, 0);
+
+	esw->enabled_vports = 0;
+	esw->mode = MLX5_ESWITCH_LEGACY;
+	esw->offloads.inline_mode = MLX5_INLINE_MODE_NONE;
+	if (MLX5_CAP_ESW_FLOWTABLE_FDB(dev, reformat) &&
+	    MLX5_CAP_ESW_FLOWTABLE_FDB(dev, decap))
+		esw->offloads.encap = DEVLINK_ESWITCH_ENCAP_MODE_BASIC;
+	else
+		esw->offloads.encap = DEVLINK_ESWITCH_ENCAP_MODE_NONE;
+	if (MLX5_ESWITCH_MANAGER(dev) &&
+	    mlx5_esw_vport_match_metadata_supported(esw))
+		esw->flags |= MLX5_ESWITCH_VPORT_MATCH_METADATA;
+
+	dev->priv.eswitch = esw;
+	BLOCKING_INIT_NOTIFIER_HEAD(&esw->n_head);
+
+	esw->dbgfs = debugfs_create_dir("esw", mlx5_debugfs_get_dev_root(esw->dev));
+	esw_info(dev,
+		 "Total vports %d, per vport: max uc(%d) max mc(%d)\n",
+		 esw->total_vports,
+		 MLX5_MAX_UC_PER_VPORT(dev),
+		 MLX5_MAX_MC_PER_VPORT(dev));
+	return 0;
+
+reps_err:
+	mlx5_esw_vports_cleanup(esw);
+abort:
+	if (esw->work_queue)
+		destroy_workqueue(esw->work_queue);
+	kfree(esw);
+	return err;
+}
+
+void mlx5_eswitch_cleanup(struct mlx5_eswitch *esw)
+{
+	if (!esw || !MLX5_VPORT_MANAGER(esw->dev))
+		return;
+
+	esw_info(esw->dev, "cleanup\n");
+
+	debugfs_remove_recursive(esw->dbgfs);
+	esw->dev->priv.eswitch = NULL;
+	destroy_workqueue(esw->work_queue);
+	WARN_ON(refcount_read(&esw->qos.refcnt));
+	mutex_destroy(&esw->state_lock);
+	WARN_ON(!xa_empty(&esw->offloads.vhca_map));
+	xa_destroy(&esw->offloads.vhca_map);
+	ida_destroy(&esw->offloads.vport_metadata_ida);
+	mlx5e_mod_hdr_tbl_destroy(&esw->offloads.mod_hdr);
+	mutex_destroy(&esw->offloads.encap_tbl_lock);
+	mutex_destroy(&esw->offloads.decap_tbl_lock);
+	esw_offloads_cleanup_reps(esw);
+	mlx5_esw_vports_cleanup(esw);
+	kfree(esw);
+}
+
+/* Vport Administration */
+static int
+mlx5_esw_set_vport_mac_locked(struct mlx5_eswitch *esw,
+			      struct mlx5_vport *evport, const u8 *mac)
+{
+	u16 vport_num = evport->vport;
+	u64 node_guid;
+	int err = 0;
+
+	if (is_multicast_ether_addr(mac))
+		return -EINVAL;
+
+	if (evport->info.spoofchk && !is_valid_ether_addr(mac))
+		mlx5_core_warn(esw->dev,
+			       "Set invalid MAC while spoofchk is on, vport(%d)\n",
+			       vport_num);
+
+	err = mlx5_modify_nic_vport_mac_address(esw->dev, vport_num, mac);
+	if (err) {
+		mlx5_core_warn(esw->dev,
+			       "Failed to mlx5_modify_nic_vport_mac vport(%d) err=(%d)\n",
+			       vport_num, err);
+		return err;
+	}
+
+	node_guid_gen_from_mac(&node_guid, mac);
+	err = mlx5_modify_nic_vport_node_guid(esw->dev, vport_num, node_guid);
+	if (err)
+		mlx5_core_warn(esw->dev,
+			       "Failed to set vport %d node guid, err = %d. RDMA_CM will not function properly for this VF.\n",
+			       vport_num, err);
+
+	ether_addr_copy(evport->info.mac, mac);
+	evport->info.node_guid = node_guid;
+	if (evport->enabled && esw->mode == MLX5_ESWITCH_LEGACY)
+		err = esw_acl_ingress_lgcy_setup(esw, evport);
+
+	return err;
+}
+
+int mlx5_eswitch_set_vport_mac(struct mlx5_eswitch *esw,
+			       u16 vport, const u8 *mac)
+{
+	struct mlx5_vport *evport = mlx5_eswitch_get_vport(esw, vport);
+	int err = 0;
+
+	if (IS_ERR(evport))
+		return PTR_ERR(evport);
+
+	mutex_lock(&esw->state_lock);
+	err = mlx5_esw_set_vport_mac_locked(esw, evport, mac);
+	mutex_unlock(&esw->state_lock);
+	return err;
+}
+
+static bool mlx5_esw_check_port_type(struct mlx5_eswitch *esw, u16 vport_num, xa_mark_t mark)
+{
+	struct mlx5_vport *vport;
+
+	vport = mlx5_eswitch_get_vport(esw, vport_num);
+	if (IS_ERR(vport))
+		return false;
+
+	return xa_get_mark(&esw->vports, vport_num, mark);
+}
+
+bool mlx5_eswitch_is_vf_vport(struct mlx5_eswitch *esw, u16 vport_num)
+{
+	return mlx5_esw_check_port_type(esw, vport_num, MLX5_ESW_VPT_VF);
+}
+
+bool mlx5_esw_is_sf_vport(struct mlx5_eswitch *esw, u16 vport_num)
+{
+	return mlx5_esw_check_port_type(esw, vport_num, MLX5_ESW_VPT_SF);
+}
+
+int mlx5_eswitch_set_vport_state(struct mlx5_eswitch *esw,
+				 u16 vport, int link_state)
+{
+	struct mlx5_vport *evport = mlx5_eswitch_get_vport(esw, vport);
+	int opmod = MLX5_VPORT_STATE_OP_MOD_ESW_VPORT;
+	int other_vport = 1;
+	int err = 0;
+
+	if (!mlx5_esw_allowed(esw))
+		return -EPERM;
+	if (IS_ERR(evport))
+		return PTR_ERR(evport);
+
+	if (vport == MLX5_VPORT_UPLINK) {
+		opmod = MLX5_VPORT_STATE_OP_MOD_UPLINK;
+		other_vport = 0;
+		vport = 0;
+	}
+	mutex_lock(&esw->state_lock);
+	if (esw->mode != MLX5_ESWITCH_LEGACY) {
+		err = -EOPNOTSUPP;
+		goto unlock;
+	}
+
+	err = mlx5_modify_vport_admin_state(esw->dev, opmod, vport, other_vport, link_state);
+	if (err) {
+		mlx5_core_warn(esw->dev, "Failed to set vport %d link state, opmod = %d, err = %d",
+			       vport, opmod, err);
+		goto unlock;
+	}
+
+	evport->info.link_state = link_state;
+
+unlock:
+	mutex_unlock(&esw->state_lock);
+	return err;
+}
+
+int mlx5_eswitch_get_vport_config(struct mlx5_eswitch *esw,
+				  u16 vport, struct ifla_vf_info *ivi)
+{
+	struct mlx5_vport *evport = mlx5_eswitch_get_vport(esw, vport);
+
+	if (IS_ERR(evport))
+		return PTR_ERR(evport);
+
+	memset(ivi, 0, sizeof(*ivi));
+	ivi->vf = vport - 1;
+
+	mutex_lock(&esw->state_lock);
+	ether_addr_copy(ivi->mac, evport->info.mac);
+	ivi->linkstate = evport->info.link_state;
+	ivi->vlan = evport->info.vlan;
+	ivi->qos = evport->info.qos;
+	ivi->spoofchk = evport->info.spoofchk;
+	ivi->trusted = evport->info.trusted;
+	if (evport->qos.enabled) {
+		ivi->min_tx_rate = evport->qos.min_rate;
+		ivi->max_tx_rate = evport->qos.max_rate;
+	}
+	mutex_unlock(&esw->state_lock);
+
+	return 0;
+}
+
+int __mlx5_eswitch_set_vport_vlan(struct mlx5_eswitch *esw,
+				  u16 vport, u16 vlan, u8 qos, u8 set_flags)
+{
+	struct mlx5_vport *evport = mlx5_eswitch_get_vport(esw, vport);
+	bool vst_mode_steering = esw_vst_mode_is_steering(esw);
+	int err = 0;
+
+	if (IS_ERR(evport))
+		return PTR_ERR(evport);
+	if (vlan > 4095 || qos > 7)
+		return -EINVAL;
+
+	if (esw->mode == MLX5_ESWITCH_OFFLOADS || !vst_mode_steering) {
+		err = modify_esw_vport_cvlan(esw->dev, vport, vlan, qos, set_flags);
+		if (err)
+			return err;
+	}
+
+	evport->info.vlan = vlan;
+	evport->info.qos = qos;
+	if (evport->enabled && esw->mode == MLX5_ESWITCH_LEGACY) {
+		err = esw_acl_ingress_lgcy_setup(esw, evport);
+		if (err)
+			return err;
+		err = esw_acl_egress_lgcy_setup(esw, evport);
+	}
+
+	return err;
+}
+
+int mlx5_eswitch_get_vport_stats(struct mlx5_eswitch *esw,
+				 u16 vport_num,
+				 struct ifla_vf_stats *vf_stats)
+{
+	struct mlx5_vport *vport = mlx5_eswitch_get_vport(esw, vport_num);
+	int outlen = MLX5_ST_SZ_BYTES(query_vport_counter_out);
+	u32 in[MLX5_ST_SZ_DW(query_vport_counter_in)] = {};
+	struct mlx5_vport_drop_stats stats = {};
+	int err = 0;
+	u32 *out;
+
+	if (IS_ERR(vport))
+		return PTR_ERR(vport);
+
+	out = kvzalloc(outlen, GFP_KERNEL);
+	if (!out)
+		return -ENOMEM;
+
+	MLX5_SET(query_vport_counter_in, in, opcode,
+		 MLX5_CMD_OP_QUERY_VPORT_COUNTER);
+	MLX5_SET(query_vport_counter_in, in, op_mod, 0);
+	MLX5_SET(query_vport_counter_in, in, vport_number, vport->vport);
+	MLX5_SET(query_vport_counter_in, in, other_vport, 1);
+
+	err = mlx5_cmd_exec_inout(esw->dev, query_vport_counter, in, out);
+	if (err)
+		goto free_out;
+
+	#define MLX5_GET_CTR(p, x) \
+		MLX5_GET64(query_vport_counter_out, p, x)
+
+	memset(vf_stats, 0, sizeof(*vf_stats));
+	vf_stats->rx_packets =
+		MLX5_GET_CTR(out, received_eth_unicast.packets) +
+		MLX5_GET_CTR(out, received_ib_unicast.packets) +
+		MLX5_GET_CTR(out, received_eth_multicast.packets) +
+		MLX5_GET_CTR(out, received_ib_multicast.packets) +
+		MLX5_GET_CTR(out, received_eth_broadcast.packets);
+
+	vf_stats->rx_bytes =
+		MLX5_GET_CTR(out, received_eth_unicast.octets) +
+		MLX5_GET_CTR(out, received_ib_unicast.octets) +
+		MLX5_GET_CTR(out, received_eth_multicast.octets) +
+		MLX5_GET_CTR(out, received_ib_multicast.octets) +
+		MLX5_GET_CTR(out, received_eth_broadcast.octets);
+
+	vf_stats->tx_packets =
+		MLX5_GET_CTR(out, transmitted_eth_unicast.packets) +
+		MLX5_GET_CTR(out, transmitted_ib_unicast.packets) +
+		MLX5_GET_CTR(out, transmitted_eth_multicast.packets) +
+		MLX5_GET_CTR(out, transmitted_ib_multicast.packets) +
+		MLX5_GET_CTR(out, transmitted_eth_broadcast.packets);
+
+	vf_stats->tx_bytes =
+		MLX5_GET_CTR(out, transmitted_eth_unicast.octets) +
+		MLX5_GET_CTR(out, transmitted_ib_unicast.octets) +
+		MLX5_GET_CTR(out, transmitted_eth_multicast.octets) +
+		MLX5_GET_CTR(out, transmitted_ib_multicast.octets) +
+		MLX5_GET_CTR(out, transmitted_eth_broadcast.octets);
+
+	vf_stats->multicast =
+		MLX5_GET_CTR(out, received_eth_multicast.packets) +
+		MLX5_GET_CTR(out, received_ib_multicast.packets);
+
+	vf_stats->broadcast =
+		MLX5_GET_CTR(out, received_eth_broadcast.packets);
+
+	err = mlx5_esw_query_vport_drop_stats(esw->dev, vport, &stats);
+	if (err)
+		goto free_out;
+	vf_stats->rx_dropped = stats.rx_dropped;
+	vf_stats->tx_dropped = stats.tx_dropped;
+
+free_out:
+	kvfree(out);
+	return err;
+}
+
+u8 mlx5_eswitch_mode(const struct mlx5_core_dev *dev)
+{
+	struct mlx5_eswitch *esw = dev->priv.eswitch;
+
+	return mlx5_esw_allowed(esw) ? esw->mode : MLX5_ESWITCH_LEGACY;
+}
+EXPORT_SYMBOL_GPL(mlx5_eswitch_mode);
+
+enum devlink_eswitch_encap_mode
+mlx5_eswitch_get_encap_mode(const struct mlx5_core_dev *dev)
+{
+	struct mlx5_eswitch *esw;
+
+	esw = dev->priv.eswitch;
+	return (mlx5_eswitch_mode(dev) == MLX5_ESWITCH_OFFLOADS)  ? esw->offloads.encap :
+		DEVLINK_ESWITCH_ENCAP_MODE_NONE;
+}
+EXPORT_SYMBOL(mlx5_eswitch_get_encap_mode);
+
+bool mlx5_esw_multipath_prereq(struct mlx5_core_dev *dev0,
+			       struct mlx5_core_dev *dev1)
+{
+	return (dev0->priv.eswitch->mode == MLX5_ESWITCH_OFFLOADS &&
+		dev1->priv.eswitch->mode == MLX5_ESWITCH_OFFLOADS);
+}
+
+int mlx5_esw_event_notifier_register(struct mlx5_eswitch *esw, struct notifier_block *nb)
+{
+	return blocking_notifier_chain_register(&esw->n_head, nb);
+}
+
+void mlx5_esw_event_notifier_unregister(struct mlx5_eswitch *esw, struct notifier_block *nb)
+{
+	blocking_notifier_chain_unregister(&esw->n_head, nb);
+}
+
+/**
+ * mlx5_esw_hold() - Try to take a read lock on esw mode lock.
+ * @mdev: mlx5 core device.
+ *
+ * Should be called by esw resources callers.
+ *
+ * Return: true on success or false.
+ */
+bool mlx5_esw_hold(struct mlx5_core_dev *mdev)
+{
+	struct mlx5_eswitch *esw = mdev->priv.eswitch;
+
+	/* e.g. VF doesn't have eswitch so nothing to do */
+	if (!mlx5_esw_allowed(esw))
+		return true;
+
+	if (down_read_trylock(&esw->mode_lock) != 0)
+		return true;
+
+	return false;
+}
+
+/**
+ * mlx5_esw_release() - Release a read lock on esw mode lock.
+ * @mdev: mlx5 core device.
+ */
+void mlx5_esw_release(struct mlx5_core_dev *mdev)
+{
+	struct mlx5_eswitch *esw = mdev->priv.eswitch;
+
+	if (mlx5_esw_allowed(esw))
+		up_read(&esw->mode_lock);
+}
+
+/**
+ * mlx5_esw_get() - Increase esw user count.
+ * @mdev: mlx5 core device.
+ */
+void mlx5_esw_get(struct mlx5_core_dev *mdev)
+{
+	struct mlx5_eswitch *esw = mdev->priv.eswitch;
+
+	if (mlx5_esw_allowed(esw))
+		atomic64_inc(&esw->user_count);
+}
+
+/**
+ * mlx5_esw_put() - Decrease esw user count.
+ * @mdev: mlx5 core device.
+ */
+void mlx5_esw_put(struct mlx5_core_dev *mdev)
+{
+	struct mlx5_eswitch *esw = mdev->priv.eswitch;
+
+	if (mlx5_esw_allowed(esw))
+		atomic64_dec_if_positive(&esw->user_count);
+}
+
+/**
+ * mlx5_esw_try_lock() - Take a write lock on esw mode lock.
+ * @esw: eswitch device.
+ *
+ * Should be called by esw mode change routine.
+ *
+ * Return:
+ * * 0       - esw mode if successfully locked and refcount is 0.
+ * * -EBUSY  - refcount is not 0.
+ * * -EINVAL - In the middle of switching mode or lock is already held.
+ */
+int mlx5_esw_try_lock(struct mlx5_eswitch *esw)
+{
+	if (down_write_trylock(&esw->mode_lock) == 0)
+		return -EINVAL;
+
+	if (atomic64_read(&esw->user_count) > 0) {
+		up_write(&esw->mode_lock);
+		return -EBUSY;
+	}
+
+	return esw->mode;
+}
+
+/**
+ * mlx5_esw_unlock() - Release write lock on esw mode lock
+ * @esw: eswitch device.
+ */
+void mlx5_esw_unlock(struct mlx5_eswitch *esw)
+{
+	up_write(&esw->mode_lock);
+}
+
+/**
+ * mlx5_eswitch_get_total_vports - Get total vports of the eswitch
+ *
+ * @dev: Pointer to core device
+ *
+ * mlx5_eswitch_get_total_vports returns total number of eswitch vports.
+ */
+u16 mlx5_eswitch_get_total_vports(const struct mlx5_core_dev *dev)
+{
+	struct mlx5_eswitch *esw;
+
+	esw = dev->priv.eswitch;
+	return mlx5_esw_allowed(esw) ? esw->total_vports : 0;
+}
+EXPORT_SYMBOL_GPL(mlx5_eswitch_get_total_vports);
+
+/**
+ * mlx5_eswitch_get_core_dev - Get the mdev device
+ * @esw : eswitch device.
+ *
+ * Return the mellanox core device which manages the eswitch.
+ */
+struct mlx5_core_dev *mlx5_eswitch_get_core_dev(struct mlx5_eswitch *esw)
+{
+	return mlx5_esw_allowed(esw) ? esw->dev : NULL;
+}
+EXPORT_SYMBOL(mlx5_eswitch_get_core_dev);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h
new file mode 100644
index 000000000..a3daca44f
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h
@@ -0,0 +1,804 @@
+/*
+ * Copyright (c) 2015, Mellanox Technologies, Ltd.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef __MLX5_ESWITCH_H__
+#define __MLX5_ESWITCH_H__
+
+#include <linux/if_ether.h>
+#include <linux/if_link.h>
+#include <linux/atomic.h>
+#include <linux/xarray.h>
+#include <net/devlink.h>
+#include <linux/mlx5/device.h>
+#include <linux/mlx5/eswitch.h>
+#include <linux/mlx5/vport.h>
+#include <linux/mlx5/fs.h>
+#include "lib/mpfs.h"
+#include "lib/fs_chains.h"
+#include "sf/sf.h"
+#include "en/tc_ct.h"
+#include "en/tc/sample.h"
+
+enum mlx5_mapped_obj_type {
+	MLX5_MAPPED_OBJ_CHAIN,
+	MLX5_MAPPED_OBJ_SAMPLE,
+	MLX5_MAPPED_OBJ_INT_PORT_METADATA,
+};
+
+struct mlx5_mapped_obj {
+	enum mlx5_mapped_obj_type type;
+	union {
+		u32 chain;
+		struct {
+			u32 group_id;
+			u32 rate;
+			u32 trunc_size;
+			u32 tunnel_id;
+		} sample;
+		u32 int_port_metadata;
+	};
+};
+
+#ifdef CONFIG_MLX5_ESWITCH
+
+#define ESW_OFFLOADS_DEFAULT_NUM_GROUPS 15
+
+#define MLX5_MAX_UC_PER_VPORT(dev) \
+	(1 << MLX5_CAP_GEN(dev, log_max_current_uc_list))
+
+#define MLX5_MAX_MC_PER_VPORT(dev) \
+	(1 << MLX5_CAP_GEN(dev, log_max_current_mc_list))
+
+#define mlx5_esw_has_fwd_fdb(dev) \
+	MLX5_CAP_ESW_FLOWTABLE(dev, fdb_multi_path_to_table)
+
+#define esw_chains(esw) \
+	((esw)->fdb_table.offloads.esw_chains_priv)
+
+enum {
+	MAPPING_TYPE_CHAIN,
+	MAPPING_TYPE_TUNNEL,
+	MAPPING_TYPE_TUNNEL_ENC_OPTS,
+	MAPPING_TYPE_LABELS,
+	MAPPING_TYPE_ZONE,
+	MAPPING_TYPE_INT_PORT,
+};
+
+struct vport_ingress {
+	struct mlx5_flow_table *acl;
+	struct mlx5_flow_handle *allow_rule;
+	struct {
+		struct mlx5_flow_group *allow_spoofchk_only_grp;
+		struct mlx5_flow_group *allow_untagged_spoofchk_grp;
+		struct mlx5_flow_group *allow_untagged_only_grp;
+		struct mlx5_flow_group *drop_grp;
+		struct mlx5_flow_handle *drop_rule;
+		struct mlx5_fc *drop_counter;
+	} legacy;
+	struct {
+		/* Optional group to add an FTE to do internal priority
+		 * tagging on ingress packets.
+		 */
+		struct mlx5_flow_group *metadata_prio_tag_grp;
+		/* Group to add default match-all FTE entry to tag ingress
+		 * packet with metadata.
+		 */
+		struct mlx5_flow_group *metadata_allmatch_grp;
+		/* Optional group to add a drop all rule */
+		struct mlx5_flow_group *drop_grp;
+		struct mlx5_modify_hdr *modify_metadata;
+		struct mlx5_flow_handle *modify_metadata_rule;
+		struct mlx5_flow_handle *drop_rule;
+	} offloads;
+};
+
+struct vport_egress {
+	struct mlx5_flow_table *acl;
+	struct mlx5_flow_handle  *allowed_vlan;
+	struct mlx5_flow_group *vlan_grp;
+	union {
+		struct {
+			struct mlx5_flow_group *drop_grp;
+			struct mlx5_flow_handle *drop_rule;
+			struct mlx5_fc *drop_counter;
+		} legacy;
+		struct {
+			struct mlx5_flow_group *fwd_grp;
+			struct mlx5_flow_handle *fwd_rule;
+			struct mlx5_flow_handle *bounce_rule;
+			struct mlx5_flow_group *bounce_grp;
+		} offloads;
+	};
+};
+
+struct mlx5_vport_drop_stats {
+	u64 rx_dropped;
+	u64 tx_dropped;
+};
+
+struct mlx5_vport_info {
+	u8                      mac[ETH_ALEN];
+	u16                     vlan;
+	u64                     node_guid;
+	int                     link_state;
+	u8                      qos;
+	u8                      spoofchk: 1;
+	u8                      trusted: 1;
+};
+
+/* Vport context events */
+enum mlx5_eswitch_vport_event {
+	MLX5_VPORT_UC_ADDR_CHANGE = BIT(0),
+	MLX5_VPORT_MC_ADDR_CHANGE = BIT(1),
+	MLX5_VPORT_PROMISC_CHANGE = BIT(3),
+};
+
+struct mlx5_vport {
+	struct mlx5_core_dev    *dev;
+	struct hlist_head       uc_list[MLX5_L2_ADDR_HASH_SIZE];
+	struct hlist_head       mc_list[MLX5_L2_ADDR_HASH_SIZE];
+	struct mlx5_flow_handle *promisc_rule;
+	struct mlx5_flow_handle *allmulti_rule;
+	struct work_struct      vport_change_handler;
+
+	struct vport_ingress    ingress;
+	struct vport_egress     egress;
+	u32                     default_metadata;
+	u32                     metadata;
+
+	struct mlx5_vport_info  info;
+
+	struct {
+		bool            enabled;
+		u32             esw_tsar_ix;
+		u32             bw_share;
+		u32 min_rate;
+		u32 max_rate;
+		struct mlx5_esw_rate_group *group;
+	} qos;
+
+	u16 vport;
+	bool                    enabled;
+	enum mlx5_eswitch_vport_event enabled_events;
+	int index;
+	struct devlink_port *dl_port;
+	struct dentry *dbgfs;
+};
+
+struct mlx5_esw_indir_table;
+
+struct mlx5_eswitch_fdb {
+	union {
+		struct legacy_fdb {
+			struct mlx5_flow_table *fdb;
+			struct mlx5_flow_group *addr_grp;
+			struct mlx5_flow_group *allmulti_grp;
+			struct mlx5_flow_group *promisc_grp;
+			struct mlx5_flow_table *vepa_fdb;
+			struct mlx5_flow_handle *vepa_uplink_rule;
+			struct mlx5_flow_handle *vepa_star_rule;
+		} legacy;
+
+		struct offloads_fdb {
+			struct mlx5_flow_namespace *ns;
+			struct mlx5_flow_table *tc_miss_table;
+			struct mlx5_flow_table *slow_fdb;
+			struct mlx5_flow_group *send_to_vport_grp;
+			struct mlx5_flow_group *send_to_vport_meta_grp;
+			struct mlx5_flow_group *peer_miss_grp;
+			struct mlx5_flow_handle **peer_miss_rules;
+			struct mlx5_flow_group *miss_grp;
+			struct mlx5_flow_handle **send_to_vport_meta_rules;
+			struct mlx5_flow_handle *miss_rule_uni;
+			struct mlx5_flow_handle *miss_rule_multi;
+			int vlan_push_pop_refcount;
+
+			struct mlx5_fs_chains *esw_chains_priv;
+			struct {
+				DECLARE_HASHTABLE(table, 8);
+				/* Protects vports.table */
+				struct mutex lock;
+			} vports;
+
+			struct mlx5_esw_indir_table *indir;
+
+		} offloads;
+	};
+	u32 flags;
+};
+
+struct mlx5_esw_offload {
+	struct mlx5_flow_table *ft_offloads_restore;
+	struct mlx5_flow_group *restore_group;
+	struct mlx5_modify_hdr *restore_copy_hdr_id;
+	struct mapping_ctx *reg_c0_obj_pool;
+
+	struct mlx5_flow_table *ft_offloads;
+	struct mlx5_flow_group *vport_rx_group;
+	struct mlx5_flow_group *vport_rx_drop_group;
+	struct mlx5_flow_handle *vport_rx_drop_rule;
+	struct xarray vport_reps;
+	struct list_head peer_flows;
+	struct mutex peer_mutex;
+	struct mutex encap_tbl_lock; /* protects encap_tbl */
+	DECLARE_HASHTABLE(encap_tbl, 8);
+	struct mutex decap_tbl_lock; /* protects decap_tbl */
+	DECLARE_HASHTABLE(decap_tbl, 8);
+	struct mod_hdr_tbl mod_hdr;
+	DECLARE_HASHTABLE(termtbl_tbl, 8);
+	struct mutex termtbl_mutex; /* protects termtbl hash */
+	struct xarray vhca_map;
+	const struct mlx5_eswitch_rep_ops *rep_ops[NUM_REP_TYPES];
+	u8 inline_mode;
+	atomic64_t num_flows;
+	enum devlink_eswitch_encap_mode encap;
+	struct ida vport_metadata_ida;
+	unsigned int host_number; /* ECPF supports one external host */
+};
+
+/* E-Switch MC FDB table hash node */
+struct esw_mc_addr { /* SRIOV only */
+	struct l2addr_node     node;
+	struct mlx5_flow_handle *uplink_rule; /* Forward to uplink rule */
+	u32                    refcnt;
+};
+
+struct mlx5_host_work {
+	struct work_struct	work;
+	struct mlx5_eswitch	*esw;
+};
+
+struct mlx5_esw_functions {
+	struct mlx5_nb		nb;
+	u16			num_vfs;
+};
+
+enum {
+	MLX5_ESWITCH_VPORT_MATCH_METADATA = BIT(0),
+	MLX5_ESWITCH_REG_C1_LOOPBACK_ENABLED = BIT(1),
+	MLX5_ESWITCH_VPORT_ACL_NS_CREATED = BIT(2),
+};
+
+struct mlx5_esw_bridge_offloads;
+
+enum {
+	MLX5_ESW_FDB_CREATED = BIT(0),
+};
+
+struct mlx5_eswitch {
+	struct mlx5_core_dev    *dev;
+	struct mlx5_nb          nb;
+	struct mlx5_eswitch_fdb fdb_table;
+	/* legacy data structures */
+	struct hlist_head       mc_table[MLX5_L2_ADDR_HASH_SIZE];
+	struct esw_mc_addr mc_promisc;
+	/* end of legacy */
+	struct workqueue_struct *work_queue;
+	struct xarray vports;
+	u32 flags;
+	int                     total_vports;
+	int                     enabled_vports;
+	/* Synchronize between vport change events
+	 * and async SRIOV admin state changes
+	 */
+	struct mutex            state_lock;
+
+	/* Protects eswitch mode change that occurs via one or more
+	 * user commands, i.e. sriov state change, devlink commands.
+	 */
+	struct rw_semaphore mode_lock;
+	atomic64_t user_count;
+
+	struct {
+		u32             root_tsar_ix;
+		struct mlx5_esw_rate_group *group0;
+		struct list_head groups; /* Protected by esw->state_lock */
+
+		/* Protected by esw->state_lock.
+		 * Initially 0, meaning no QoS users and QoS is disabled.
+		 */
+		refcount_t refcnt;
+	} qos;
+
+	struct mlx5_esw_bridge_offloads *br_offloads;
+	struct mlx5_esw_offload offloads;
+	int                     mode;
+	u16                     manager_vport;
+	u16                     first_host_vport;
+	struct mlx5_esw_functions esw_funcs;
+	struct {
+		u32             large_group_num;
+	}  params;
+	struct blocking_notifier_head n_head;
+	struct dentry *dbgfs;
+	bool paired[MLX5_MAX_PORTS];
+};
+
+void esw_offloads_disable(struct mlx5_eswitch *esw);
+int esw_offloads_enable(struct mlx5_eswitch *esw);
+void esw_offloads_cleanup_reps(struct mlx5_eswitch *esw);
+int esw_offloads_init_reps(struct mlx5_eswitch *esw);
+
+struct mlx5_flow_handle *
+mlx5_eswitch_add_send_to_vport_meta_rule(struct mlx5_eswitch *esw, u16 vport_num);
+void mlx5_eswitch_del_send_to_vport_meta_rule(struct mlx5_flow_handle *rule);
+
+bool mlx5_esw_vport_match_metadata_supported(const struct mlx5_eswitch *esw);
+int mlx5_esw_offloads_vport_metadata_set(struct mlx5_eswitch *esw, bool enable);
+u32 mlx5_esw_match_metadata_alloc(struct mlx5_eswitch *esw);
+void mlx5_esw_match_metadata_free(struct mlx5_eswitch *esw, u32 metadata);
+
+int mlx5_esw_qos_modify_vport_rate(struct mlx5_eswitch *esw, u16 vport_num, u32 rate_mbps);
+
+/* E-Switch API */
+int mlx5_eswitch_init(struct mlx5_core_dev *dev);
+void mlx5_eswitch_cleanup(struct mlx5_eswitch *esw);
+
+#define MLX5_ESWITCH_IGNORE_NUM_VFS (-1)
+int mlx5_eswitch_enable_locked(struct mlx5_eswitch *esw, int num_vfs);
+int mlx5_eswitch_enable(struct mlx5_eswitch *esw, int num_vfs);
+void mlx5_eswitch_disable_sriov(struct mlx5_eswitch *esw, bool clear_vf);
+void mlx5_eswitch_disable_locked(struct mlx5_eswitch *esw);
+void mlx5_eswitch_disable(struct mlx5_eswitch *esw);
+void mlx5_esw_offloads_devcom_init(struct mlx5_eswitch *esw);
+void mlx5_esw_offloads_devcom_cleanup(struct mlx5_eswitch *esw);
+int mlx5_eswitch_set_vport_mac(struct mlx5_eswitch *esw,
+			       u16 vport, const u8 *mac);
+int mlx5_eswitch_set_vport_state(struct mlx5_eswitch *esw,
+				 u16 vport, int link_state);
+int mlx5_eswitch_set_vport_vlan(struct mlx5_eswitch *esw,
+				u16 vport, u16 vlan, u8 qos);
+int mlx5_eswitch_set_vport_spoofchk(struct mlx5_eswitch *esw,
+				    u16 vport, bool spoofchk);
+int mlx5_eswitch_set_vport_trust(struct mlx5_eswitch *esw,
+				 u16 vport_num, bool setting);
+int mlx5_eswitch_set_vport_rate(struct mlx5_eswitch *esw, u16 vport,
+				u32 max_rate, u32 min_rate);
+int mlx5_esw_qos_vport_update_group(struct mlx5_eswitch *esw,
+				    struct mlx5_vport *vport,
+				    struct mlx5_esw_rate_group *group,
+				    struct netlink_ext_ack *extack);
+int mlx5_eswitch_set_vepa(struct mlx5_eswitch *esw, u8 setting);
+int mlx5_eswitch_get_vepa(struct mlx5_eswitch *esw, u8 *setting);
+int mlx5_eswitch_get_vport_config(struct mlx5_eswitch *esw,
+				  u16 vport, struct ifla_vf_info *ivi);
+int mlx5_eswitch_get_vport_stats(struct mlx5_eswitch *esw,
+				 u16 vport,
+				 struct ifla_vf_stats *vf_stats);
+void mlx5_eswitch_del_send_to_vport_rule(struct mlx5_flow_handle *rule);
+
+int mlx5_eswitch_modify_esw_vport_context(struct mlx5_core_dev *dev, u16 vport,
+					  bool other_vport, void *in);
+
+struct mlx5_flow_spec;
+struct mlx5_esw_flow_attr;
+struct mlx5_termtbl_handle;
+
+bool
+mlx5_eswitch_termtbl_required(struct mlx5_eswitch *esw,
+			      struct mlx5_flow_attr *attr,
+			      struct mlx5_flow_act *flow_act,
+			      struct mlx5_flow_spec *spec);
+
+struct mlx5_flow_handle *
+mlx5_eswitch_add_termtbl_rule(struct mlx5_eswitch *esw,
+			      struct mlx5_flow_table *ft,
+			      struct mlx5_flow_spec *spec,
+			      struct mlx5_esw_flow_attr *attr,
+			      struct mlx5_flow_act *flow_act,
+			      struct mlx5_flow_destination *dest,
+			      int num_dest);
+
+void
+mlx5_eswitch_termtbl_put(struct mlx5_eswitch *esw,
+			 struct mlx5_termtbl_handle *tt);
+
+void
+mlx5_eswitch_clear_rule_source_port(struct mlx5_eswitch *esw, struct mlx5_flow_spec *spec);
+
+struct mlx5_flow_handle *
+mlx5_eswitch_add_offloaded_rule(struct mlx5_eswitch *esw,
+				struct mlx5_flow_spec *spec,
+				struct mlx5_flow_attr *attr);
+struct mlx5_flow_handle *
+mlx5_eswitch_add_fwd_rule(struct mlx5_eswitch *esw,
+			  struct mlx5_flow_spec *spec,
+			  struct mlx5_flow_attr *attr);
+void
+mlx5_eswitch_del_offloaded_rule(struct mlx5_eswitch *esw,
+				struct mlx5_flow_handle *rule,
+				struct mlx5_flow_attr *attr);
+void
+mlx5_eswitch_del_fwd_rule(struct mlx5_eswitch *esw,
+			  struct mlx5_flow_handle *rule,
+			  struct mlx5_flow_attr *attr);
+
+struct mlx5_flow_handle *
+mlx5_eswitch_create_vport_rx_rule(struct mlx5_eswitch *esw, u16 vport,
+				  struct mlx5_flow_destination *dest);
+
+enum {
+	SET_VLAN_STRIP	= BIT(0),
+	SET_VLAN_INSERT	= BIT(1)
+};
+
+enum mlx5_flow_match_level {
+	MLX5_MATCH_NONE	= MLX5_INLINE_MODE_NONE,
+	MLX5_MATCH_L2	= MLX5_INLINE_MODE_L2,
+	MLX5_MATCH_L3	= MLX5_INLINE_MODE_IP,
+	MLX5_MATCH_L4	= MLX5_INLINE_MODE_TCP_UDP,
+};
+
+/* current maximum for flow based vport multicasting */
+#define MLX5_MAX_FLOW_FWD_VPORTS 32
+
+enum {
+	MLX5_ESW_DEST_ENCAP         = BIT(0),
+	MLX5_ESW_DEST_ENCAP_VALID   = BIT(1),
+	MLX5_ESW_DEST_CHAIN_WITH_SRC_PORT_CHANGE  = BIT(2),
+};
+
+struct mlx5_esw_flow_attr {
+	struct mlx5_eswitch_rep *in_rep;
+	struct mlx5_core_dev	*in_mdev;
+	struct mlx5_core_dev    *counter_dev;
+	struct mlx5e_tc_int_port *dest_int_port;
+	struct mlx5e_tc_int_port *int_port;
+
+	int split_count;
+	int out_count;
+
+	__be16	vlan_proto[MLX5_FS_VLAN_DEPTH];
+	u16	vlan_vid[MLX5_FS_VLAN_DEPTH];
+	u8	vlan_prio[MLX5_FS_VLAN_DEPTH];
+	u8	total_vlan;
+	struct {
+		u32 flags;
+		struct mlx5_eswitch_rep *rep;
+		struct mlx5_pkt_reformat *pkt_reformat;
+		struct mlx5_core_dev *mdev;
+		struct mlx5_termtbl_handle *termtbl;
+		int src_port_rewrite_act_id;
+	} dests[MLX5_MAX_FLOW_FWD_VPORTS];
+	struct mlx5_rx_tun_attr *rx_tun_attr;
+	struct ethhdr eth;
+	struct mlx5_pkt_reformat *decap_pkt_reformat;
+};
+
+int mlx5_devlink_eswitch_mode_set(struct devlink *devlink, u16 mode,
+				  struct netlink_ext_ack *extack);
+int mlx5_devlink_eswitch_mode_get(struct devlink *devlink, u16 *mode);
+int mlx5_devlink_eswitch_inline_mode_set(struct devlink *devlink, u8 mode,
+					 struct netlink_ext_ack *extack);
+int mlx5_devlink_eswitch_inline_mode_get(struct devlink *devlink, u8 *mode);
+int mlx5_devlink_eswitch_encap_mode_set(struct devlink *devlink,
+					enum devlink_eswitch_encap_mode encap,
+					struct netlink_ext_ack *extack);
+int mlx5_devlink_eswitch_encap_mode_get(struct devlink *devlink,
+					enum devlink_eswitch_encap_mode *encap);
+int mlx5_devlink_port_function_hw_addr_get(struct devlink_port *port,
+					   u8 *hw_addr, int *hw_addr_len,
+					   struct netlink_ext_ack *extack);
+int mlx5_devlink_port_function_hw_addr_set(struct devlink_port *port,
+					   const u8 *hw_addr, int hw_addr_len,
+					   struct netlink_ext_ack *extack);
+
+void *mlx5_eswitch_get_uplink_priv(struct mlx5_eswitch *esw, u8 rep_type);
+
+int mlx5_eswitch_add_vlan_action(struct mlx5_eswitch *esw,
+				 struct mlx5_flow_attr *attr);
+int mlx5_eswitch_del_vlan_action(struct mlx5_eswitch *esw,
+				 struct mlx5_flow_attr *attr);
+int __mlx5_eswitch_set_vport_vlan(struct mlx5_eswitch *esw,
+				  u16 vport, u16 vlan, u8 qos, u8 set_flags);
+
+static inline bool esw_vst_mode_is_steering(struct mlx5_eswitch *esw)
+{
+	return (MLX5_CAP_ESW_EGRESS_ACL(esw->dev, pop_vlan) &&
+		MLX5_CAP_ESW_INGRESS_ACL(esw->dev, push_vlan));
+}
+
+static inline bool mlx5_eswitch_vlan_actions_supported(struct mlx5_core_dev *dev,
+						       u8 vlan_depth)
+{
+	bool ret = MLX5_CAP_ESW_FLOWTABLE_FDB(dev, pop_vlan) &&
+		   MLX5_CAP_ESW_FLOWTABLE_FDB(dev, push_vlan);
+
+	if (vlan_depth == 1)
+		return ret;
+
+	return  ret && MLX5_CAP_ESW_FLOWTABLE_FDB(dev, pop_vlan_2) &&
+		MLX5_CAP_ESW_FLOWTABLE_FDB(dev, push_vlan_2);
+}
+
+bool mlx5_esw_multipath_prereq(struct mlx5_core_dev *dev0,
+			       struct mlx5_core_dev *dev1);
+
+const u32 *mlx5_esw_query_functions(struct mlx5_core_dev *dev);
+
+#define MLX5_DEBUG_ESWITCH_MASK BIT(3)
+
+#define esw_info(__dev, format, ...)			\
+	dev_info((__dev)->device, "E-Switch: " format, ##__VA_ARGS__)
+
+#define esw_warn(__dev, format, ...)			\
+	dev_warn((__dev)->device, "E-Switch: " format, ##__VA_ARGS__)
+
+#define esw_debug(dev, format, ...)				\
+	mlx5_core_dbg_mask(dev, MLX5_DEBUG_ESWITCH_MASK, format, ##__VA_ARGS__)
+
+static inline bool mlx5_esw_allowed(const struct mlx5_eswitch *esw)
+{
+	return esw && MLX5_ESWITCH_MANAGER(esw->dev);
+}
+
+/* The returned number is valid only when the dev is eswitch manager. */
+static inline u16 mlx5_eswitch_manager_vport(struct mlx5_core_dev *dev)
+{
+	return mlx5_core_is_ecpf_esw_manager(dev) ?
+		MLX5_VPORT_ECPF : MLX5_VPORT_PF;
+}
+
+static inline bool
+mlx5_esw_is_manager_vport(const struct mlx5_eswitch *esw, u16 vport_num)
+{
+	return esw->manager_vport == vport_num;
+}
+
+static inline u16 mlx5_eswitch_first_host_vport_num(struct mlx5_core_dev *dev)
+{
+	return mlx5_core_is_ecpf_esw_manager(dev) ?
+		MLX5_VPORT_PF : MLX5_VPORT_FIRST_VF;
+}
+
+static inline bool mlx5_eswitch_is_funcs_handler(const struct mlx5_core_dev *dev)
+{
+	return mlx5_core_is_ecpf_esw_manager(dev);
+}
+
+static inline unsigned int
+mlx5_esw_vport_to_devlink_port_index(const struct mlx5_core_dev *dev,
+				     u16 vport_num)
+{
+	return (MLX5_CAP_GEN(dev, vhca_id) << 16) | vport_num;
+}
+
+static inline u16
+mlx5_esw_devlink_port_index_to_vport_num(unsigned int dl_port_index)
+{
+	return dl_port_index & 0xffff;
+}
+
+static inline bool mlx5_esw_is_fdb_created(struct mlx5_eswitch *esw)
+{
+	return esw->fdb_table.flags & MLX5_ESW_FDB_CREATED;
+}
+
+/* TODO: This mlx5e_tc function shouldn't be called by eswitch */
+void mlx5e_tc_clean_fdb_peer_flows(struct mlx5_eswitch *esw);
+
+/* Each mark identifies eswitch vport type.
+ * MLX5_ESW_VPT_HOST_FN is used to identify both PF and VF ports using
+ * a single mark.
+ * MLX5_ESW_VPT_VF identifies a SRIOV VF vport.
+ * MLX5_ESW_VPT_SF identifies SF vport.
+ */
+#define MLX5_ESW_VPT_HOST_FN XA_MARK_0
+#define MLX5_ESW_VPT_VF XA_MARK_1
+#define MLX5_ESW_VPT_SF XA_MARK_2
+
+/* The vport iterator is valid only after vport are initialized in mlx5_eswitch_init.
+ * Borrowed the idea from xa_for_each_marked() but with support for desired last element.
+ */
+
+#define mlx5_esw_for_each_vport(esw, index, vport) \
+	xa_for_each(&((esw)->vports), index, vport)
+
+#define mlx5_esw_for_each_entry_marked(xa, index, entry, last, filter)	\
+	for (index = 0, entry = xa_find(xa, &index, last, filter); \
+	     entry; entry = xa_find_after(xa, &index, last, filter))
+
+#define mlx5_esw_for_each_vport_marked(esw, index, vport, last, filter)	\
+	mlx5_esw_for_each_entry_marked(&((esw)->vports), index, vport, last, filter)
+
+#define mlx5_esw_for_each_vf_vport(esw, index, vport, last)	\
+	mlx5_esw_for_each_vport_marked(esw, index, vport, last, MLX5_ESW_VPT_VF)
+
+#define mlx5_esw_for_each_host_func_vport(esw, index, vport, last)	\
+	mlx5_esw_for_each_vport_marked(esw, index, vport, last, MLX5_ESW_VPT_HOST_FN)
+
+struct mlx5_eswitch *mlx5_devlink_eswitch_get(struct devlink *devlink);
+struct mlx5_vport *__must_check
+mlx5_eswitch_get_vport(struct mlx5_eswitch *esw, u16 vport_num);
+
+bool mlx5_eswitch_is_vf_vport(struct mlx5_eswitch *esw, u16 vport_num);
+bool mlx5_esw_is_sf_vport(struct mlx5_eswitch *esw, u16 vport_num);
+
+int mlx5_esw_funcs_changed_handler(struct notifier_block *nb, unsigned long type, void *data);
+
+int
+mlx5_eswitch_enable_pf_vf_vports(struct mlx5_eswitch *esw,
+				 enum mlx5_eswitch_vport_event enabled_events);
+void mlx5_eswitch_disable_pf_vf_vports(struct mlx5_eswitch *esw);
+
+int mlx5_esw_vport_enable(struct mlx5_eswitch *esw, u16 vport_num,
+			  enum mlx5_eswitch_vport_event enabled_events);
+void mlx5_esw_vport_disable(struct mlx5_eswitch *esw, u16 vport_num);
+
+int
+esw_vport_create_offloads_acl_tables(struct mlx5_eswitch *esw,
+				     struct mlx5_vport *vport);
+void
+esw_vport_destroy_offloads_acl_tables(struct mlx5_eswitch *esw,
+				      struct mlx5_vport *vport);
+
+struct esw_vport_tbl_namespace {
+	int max_fte;
+	int max_num_groups;
+	u32 flags;
+};
+
+struct mlx5_vport_tbl_attr {
+	u32 chain;
+	u16 prio;
+	u16 vport;
+	struct esw_vport_tbl_namespace *vport_ns;
+};
+
+struct mlx5_flow_table *
+mlx5_esw_vporttbl_get(struct mlx5_eswitch *esw, struct mlx5_vport_tbl_attr *attr);
+void
+mlx5_esw_vporttbl_put(struct mlx5_eswitch *esw, struct mlx5_vport_tbl_attr *attr);
+
+struct mlx5_flow_handle *
+esw_add_restore_rule(struct mlx5_eswitch *esw, u32 tag);
+
+int esw_offloads_load_rep(struct mlx5_eswitch *esw, u16 vport_num);
+void esw_offloads_unload_rep(struct mlx5_eswitch *esw, u16 vport_num);
+
+int mlx5_esw_offloads_rep_load(struct mlx5_eswitch *esw, u16 vport_num);
+void mlx5_esw_offloads_rep_unload(struct mlx5_eswitch *esw, u16 vport_num);
+
+int mlx5_eswitch_load_vport(struct mlx5_eswitch *esw, u16 vport_num,
+			    enum mlx5_eswitch_vport_event enabled_events);
+void mlx5_eswitch_unload_vport(struct mlx5_eswitch *esw, u16 vport_num);
+
+int mlx5_eswitch_load_vf_vports(struct mlx5_eswitch *esw, u16 num_vfs,
+				enum mlx5_eswitch_vport_event enabled_events);
+void mlx5_eswitch_unload_vf_vports(struct mlx5_eswitch *esw, u16 num_vfs);
+
+int mlx5_esw_offloads_devlink_port_register(struct mlx5_eswitch *esw, u16 vport_num);
+void mlx5_esw_offloads_devlink_port_unregister(struct mlx5_eswitch *esw, u16 vport_num);
+struct devlink_port *mlx5_esw_offloads_devlink_port(struct mlx5_eswitch *esw, u16 vport_num);
+
+void mlx5_esw_vport_debugfs_create(struct mlx5_eswitch *esw, u16 vport_num, bool is_sf, u16 sf_num);
+void mlx5_esw_vport_debugfs_destroy(struct mlx5_eswitch *esw, u16 vport_num);
+
+int mlx5_esw_devlink_sf_port_register(struct mlx5_eswitch *esw, struct devlink_port *dl_port,
+				      u16 vport_num, u32 controller, u32 sfnum);
+void mlx5_esw_devlink_sf_port_unregister(struct mlx5_eswitch *esw, u16 vport_num);
+
+int mlx5_esw_offloads_sf_vport_enable(struct mlx5_eswitch *esw, struct devlink_port *dl_port,
+				      u16 vport_num, u32 controller, u32 sfnum);
+void mlx5_esw_offloads_sf_vport_disable(struct mlx5_eswitch *esw, u16 vport_num);
+int mlx5_esw_sf_max_hpf_functions(struct mlx5_core_dev *dev, u16 *max_sfs, u16 *sf_base_id);
+
+int mlx5_esw_vport_vhca_id_set(struct mlx5_eswitch *esw, u16 vport_num);
+void mlx5_esw_vport_vhca_id_clear(struct mlx5_eswitch *esw, u16 vport_num);
+int mlx5_eswitch_vhca_id_to_vport(struct mlx5_eswitch *esw, u16 vhca_id, u16 *vport_num);
+
+/**
+ * mlx5_esw_event_info - Indicates eswitch mode changed/changing.
+ *
+ * @new_mode: New mode of eswitch.
+ */
+struct mlx5_esw_event_info {
+	u16 new_mode;
+};
+
+int mlx5_esw_event_notifier_register(struct mlx5_eswitch *esw, struct notifier_block *n);
+void mlx5_esw_event_notifier_unregister(struct mlx5_eswitch *esw, struct notifier_block *n);
+
+bool mlx5_esw_hold(struct mlx5_core_dev *dev);
+void mlx5_esw_release(struct mlx5_core_dev *dev);
+void mlx5_esw_get(struct mlx5_core_dev *dev);
+void mlx5_esw_put(struct mlx5_core_dev *dev);
+int mlx5_esw_try_lock(struct mlx5_eswitch *esw);
+void mlx5_esw_unlock(struct mlx5_eswitch *esw);
+
+void esw_vport_change_handle_locked(struct mlx5_vport *vport);
+
+bool mlx5_esw_offloads_controller_valid(const struct mlx5_eswitch *esw, u32 controller);
+
+int mlx5_eswitch_offloads_config_single_fdb(struct mlx5_eswitch *master_esw,
+					    struct mlx5_eswitch *slave_esw);
+void mlx5_eswitch_offloads_destroy_single_fdb(struct mlx5_eswitch *master_esw,
+					      struct mlx5_eswitch *slave_esw);
+int mlx5_eswitch_reload_reps(struct mlx5_eswitch *esw);
+
+static inline int mlx5_eswitch_num_vfs(struct mlx5_eswitch *esw)
+{
+	if (mlx5_esw_allowed(esw))
+		return esw->esw_funcs.num_vfs;
+
+	return 0;
+}
+
+#else  /* CONFIG_MLX5_ESWITCH */
+/* eswitch API stubs */
+static inline int  mlx5_eswitch_init(struct mlx5_core_dev *dev) { return 0; }
+static inline void mlx5_eswitch_cleanup(struct mlx5_eswitch *esw) {}
+static inline int mlx5_eswitch_enable(struct mlx5_eswitch *esw, int num_vfs) { return 0; }
+static inline void mlx5_eswitch_disable_sriov(struct mlx5_eswitch *esw, bool clear_vf) {}
+static inline void mlx5_eswitch_disable(struct mlx5_eswitch *esw) {}
+static inline void mlx5_esw_offloads_devcom_init(struct mlx5_eswitch *esw) {}
+static inline void mlx5_esw_offloads_devcom_cleanup(struct mlx5_eswitch *esw) {}
+static inline bool mlx5_eswitch_is_funcs_handler(struct mlx5_core_dev *dev) { return false; }
+static inline
+int mlx5_eswitch_set_vport_state(struct mlx5_eswitch *esw, u16 vport, int link_state) { return 0; }
+static inline const u32 *mlx5_esw_query_functions(struct mlx5_core_dev *dev)
+{
+	return ERR_PTR(-EOPNOTSUPP);
+}
+
+static inline struct mlx5_flow_handle *
+esw_add_restore_rule(struct mlx5_eswitch *esw, u32 tag)
+{
+	return ERR_PTR(-EOPNOTSUPP);
+}
+
+static inline unsigned int
+mlx5_esw_vport_to_devlink_port_index(const struct mlx5_core_dev *dev,
+				     u16 vport_num)
+{
+	return vport_num;
+}
+
+static inline int
+mlx5_eswitch_offloads_config_single_fdb(struct mlx5_eswitch *master_esw,
+					struct mlx5_eswitch *slave_esw)
+{
+	return 0;
+}
+
+static inline void
+mlx5_eswitch_offloads_destroy_single_fdb(struct mlx5_eswitch *master_esw,
+					 struct mlx5_eswitch *slave_esw) {}
+
+static inline int
+mlx5_eswitch_reload_reps(struct mlx5_eswitch *esw)
+{
+	return 0;
+}
+#endif /* CONFIG_MLX5_ESWITCH */
+
+#endif /* __MLX5_ESWITCH_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c b/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
new file mode 100644
index 000000000..433cdd0a2
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
@@ -0,0 +1,4040 @@
+/*
+ * Copyright (c) 2016, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/etherdevice.h>
+#include <linux/idr.h>
+#include <linux/mlx5/driver.h>
+#include <linux/mlx5/mlx5_ifc.h>
+#include <linux/mlx5/vport.h>
+#include <linux/mlx5/fs.h>
+#include "mlx5_core.h"
+#include "eswitch.h"
+#include "esw/indir_table.h"
+#include "esw/acl/ofld.h"
+#include "rdma.h"
+#include "en.h"
+#include "fs_core.h"
+#include "lib/devcom.h"
+#include "lib/eq.h"
+#include "lib/fs_chains.h"
+#include "en_tc.h"
+#include "en/mapping.h"
+#include "devlink.h"
+#include "lag/lag.h"
+
+#define mlx5_esw_for_each_rep(esw, i, rep) \
+	xa_for_each(&((esw)->offloads.vport_reps), i, rep)
+
+#define mlx5_esw_for_each_sf_rep(esw, i, rep) \
+	xa_for_each_marked(&((esw)->offloads.vport_reps), i, rep, MLX5_ESW_VPT_SF)
+
+#define mlx5_esw_for_each_vf_rep(esw, index, rep)	\
+	mlx5_esw_for_each_entry_marked(&((esw)->offloads.vport_reps), index, \
+				       rep, (esw)->esw_funcs.num_vfs, MLX5_ESW_VPT_VF)
+
+/* There are two match-all miss flows, one for unicast dst mac and
+ * one for multicast.
+ */
+#define MLX5_ESW_MISS_FLOWS (2)
+#define UPLINK_REP_INDEX 0
+
+#define MLX5_ESW_VPORT_TBL_SIZE 128
+#define MLX5_ESW_VPORT_TBL_NUM_GROUPS  4
+
+#define MLX5_ESW_FT_OFFLOADS_DROP_RULE (1)
+
+static struct esw_vport_tbl_namespace mlx5_esw_vport_tbl_mirror_ns = {
+	.max_fte = MLX5_ESW_VPORT_TBL_SIZE,
+	.max_num_groups = MLX5_ESW_VPORT_TBL_NUM_GROUPS,
+	.flags = 0,
+};
+
+static struct mlx5_eswitch_rep *mlx5_eswitch_get_rep(struct mlx5_eswitch *esw,
+						     u16 vport_num)
+{
+	return xa_load(&esw->offloads.vport_reps, vport_num);
+}
+
+static void
+mlx5_eswitch_set_rule_flow_source(struct mlx5_eswitch *esw,
+				  struct mlx5_flow_spec *spec,
+				  struct mlx5_esw_flow_attr *attr)
+{
+	if (!MLX5_CAP_ESW_FLOWTABLE(esw->dev, flow_source) || !attr || !attr->in_rep)
+		return;
+
+	if (attr->int_port) {
+		spec->flow_context.flow_source = mlx5e_tc_int_port_get_flow_source(attr->int_port);
+
+		return;
+	}
+
+	spec->flow_context.flow_source = (attr->in_rep->vport == MLX5_VPORT_UPLINK) ?
+					 MLX5_FLOW_CONTEXT_FLOW_SOURCE_UPLINK :
+					 MLX5_FLOW_CONTEXT_FLOW_SOURCE_LOCAL_VPORT;
+}
+
+/* Actually only the upper 16 bits of reg c0 need to be cleared, but the lower 16 bits
+ * are not needed as well in the following process. So clear them all for simplicity.
+ */
+void
+mlx5_eswitch_clear_rule_source_port(struct mlx5_eswitch *esw, struct mlx5_flow_spec *spec)
+{
+	if (mlx5_eswitch_vport_match_metadata_enabled(esw)) {
+		void *misc2;
+
+		misc2 = MLX5_ADDR_OF(fte_match_param, spec->match_value, misc_parameters_2);
+		MLX5_SET(fte_match_set_misc2, misc2, metadata_reg_c_0, 0);
+
+		misc2 = MLX5_ADDR_OF(fte_match_param, spec->match_criteria, misc_parameters_2);
+		MLX5_SET(fte_match_set_misc2, misc2, metadata_reg_c_0, 0);
+
+		if (!memchr_inv(misc2, 0, MLX5_ST_SZ_BYTES(fte_match_set_misc2)))
+			spec->match_criteria_enable &= ~MLX5_MATCH_MISC_PARAMETERS_2;
+	}
+}
+
+static void
+mlx5_eswitch_set_rule_source_port(struct mlx5_eswitch *esw,
+				  struct mlx5_flow_spec *spec,
+				  struct mlx5_flow_attr *attr,
+				  struct mlx5_eswitch *src_esw,
+				  u16 vport)
+{
+	struct mlx5_esw_flow_attr *esw_attr = attr->esw_attr;
+	u32 metadata;
+	void *misc2;
+	void *misc;
+
+	/* Use metadata matching because vport is not represented by single
+	 * VHCA in dual-port RoCE mode, and matching on source vport may fail.
+	 */
+	if (mlx5_eswitch_vport_match_metadata_enabled(esw)) {
+		if (mlx5_esw_indir_table_decap_vport(attr))
+			vport = mlx5_esw_indir_table_decap_vport(attr);
+
+		if (!attr->chain && esw_attr && esw_attr->int_port)
+			metadata =
+				mlx5e_tc_int_port_get_metadata_for_match(esw_attr->int_port);
+		else
+			metadata =
+				mlx5_eswitch_get_vport_metadata_for_match(src_esw, vport);
+
+		misc2 = MLX5_ADDR_OF(fte_match_param, spec->match_value, misc_parameters_2);
+		MLX5_SET(fte_match_set_misc2, misc2, metadata_reg_c_0, metadata);
+
+		misc2 = MLX5_ADDR_OF(fte_match_param, spec->match_criteria, misc_parameters_2);
+		MLX5_SET(fte_match_set_misc2, misc2, metadata_reg_c_0,
+			 mlx5_eswitch_get_vport_metadata_mask());
+
+		spec->match_criteria_enable |= MLX5_MATCH_MISC_PARAMETERS_2;
+	} else {
+		misc = MLX5_ADDR_OF(fte_match_param, spec->match_value, misc_parameters);
+		MLX5_SET(fte_match_set_misc, misc, source_port, vport);
+
+		if (MLX5_CAP_ESW(esw->dev, merged_eswitch))
+			MLX5_SET(fte_match_set_misc, misc,
+				 source_eswitch_owner_vhca_id,
+				 MLX5_CAP_GEN(src_esw->dev, vhca_id));
+
+		misc = MLX5_ADDR_OF(fte_match_param, spec->match_criteria, misc_parameters);
+		MLX5_SET_TO_ONES(fte_match_set_misc, misc, source_port);
+		if (MLX5_CAP_ESW(esw->dev, merged_eswitch))
+			MLX5_SET_TO_ONES(fte_match_set_misc, misc,
+					 source_eswitch_owner_vhca_id);
+
+		spec->match_criteria_enable |= MLX5_MATCH_MISC_PARAMETERS;
+	}
+}
+
+static int
+esw_setup_decap_indir(struct mlx5_eswitch *esw,
+		      struct mlx5_flow_attr *attr)
+{
+	struct mlx5_flow_table *ft;
+
+	if (!(attr->flags & MLX5_ATTR_FLAG_SRC_REWRITE))
+		return -EOPNOTSUPP;
+
+	ft = mlx5_esw_indir_table_get(esw, attr,
+				      mlx5_esw_indir_table_decap_vport(attr), true);
+	return PTR_ERR_OR_ZERO(ft);
+}
+
+static void
+esw_cleanup_decap_indir(struct mlx5_eswitch *esw,
+			struct mlx5_flow_attr *attr)
+{
+	if (mlx5_esw_indir_table_decap_vport(attr))
+		mlx5_esw_indir_table_put(esw,
+					 mlx5_esw_indir_table_decap_vport(attr),
+					 true);
+}
+
+static int
+esw_setup_sampler_dest(struct mlx5_flow_destination *dest,
+		       struct mlx5_flow_act *flow_act,
+		       u32 sampler_id,
+		       int i)
+{
+	flow_act->flags |= FLOW_ACT_IGNORE_FLOW_LEVEL;
+	dest[i].type = MLX5_FLOW_DESTINATION_TYPE_FLOW_SAMPLER;
+	dest[i].sampler_id = sampler_id;
+
+	return 0;
+}
+
+static int
+esw_setup_ft_dest(struct mlx5_flow_destination *dest,
+		  struct mlx5_flow_act *flow_act,
+		  struct mlx5_eswitch *esw,
+		  struct mlx5_flow_attr *attr,
+		  int i)
+{
+	flow_act->flags |= FLOW_ACT_IGNORE_FLOW_LEVEL;
+	dest[i].type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
+	dest[i].ft = attr->dest_ft;
+
+	if (mlx5_esw_indir_table_decap_vport(attr))
+		return esw_setup_decap_indir(esw, attr);
+	return 0;
+}
+
+static void
+esw_setup_accept_dest(struct mlx5_flow_destination *dest, struct mlx5_flow_act *flow_act,
+		      struct mlx5_fs_chains *chains, int i)
+{
+	if (mlx5_chains_ignore_flow_level_supported(chains))
+		flow_act->flags |= FLOW_ACT_IGNORE_FLOW_LEVEL;
+	dest[i].type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
+	dest[i].ft = mlx5_chains_get_tc_end_ft(chains);
+}
+
+static void
+esw_setup_slow_path_dest(struct mlx5_flow_destination *dest, struct mlx5_flow_act *flow_act,
+			 struct mlx5_eswitch *esw, int i)
+{
+	if (MLX5_CAP_ESW_FLOWTABLE_FDB(esw->dev, ignore_flow_level))
+		flow_act->flags |= FLOW_ACT_IGNORE_FLOW_LEVEL;
+	dest[i].type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
+	dest[i].ft = esw->fdb_table.offloads.slow_fdb;
+}
+
+static int
+esw_setup_chain_dest(struct mlx5_flow_destination *dest,
+		     struct mlx5_flow_act *flow_act,
+		     struct mlx5_fs_chains *chains,
+		     u32 chain, u32 prio, u32 level,
+		     int i)
+{
+	struct mlx5_flow_table *ft;
+
+	flow_act->flags |= FLOW_ACT_IGNORE_FLOW_LEVEL;
+	ft = mlx5_chains_get_table(chains, chain, prio, level);
+	if (IS_ERR(ft))
+		return PTR_ERR(ft);
+
+	dest[i].type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
+	dest[i].ft = ft;
+	return  0;
+}
+
+static void esw_put_dest_tables_loop(struct mlx5_eswitch *esw, struct mlx5_flow_attr *attr,
+				     int from, int to)
+{
+	struct mlx5_esw_flow_attr *esw_attr = attr->esw_attr;
+	struct mlx5_fs_chains *chains = esw_chains(esw);
+	int i;
+
+	for (i = from; i < to; i++)
+		if (esw_attr->dests[i].flags & MLX5_ESW_DEST_CHAIN_WITH_SRC_PORT_CHANGE)
+			mlx5_chains_put_table(chains, 0, 1, 0);
+		else if (mlx5_esw_indir_table_needed(esw, attr, esw_attr->dests[i].rep->vport,
+						     esw_attr->dests[i].mdev))
+			mlx5_esw_indir_table_put(esw, esw_attr->dests[i].rep->vport,
+						 false);
+}
+
+static bool
+esw_is_chain_src_port_rewrite(struct mlx5_eswitch *esw, struct mlx5_esw_flow_attr *esw_attr)
+{
+	int i;
+
+	for (i = esw_attr->split_count; i < esw_attr->out_count; i++)
+		if (esw_attr->dests[i].flags & MLX5_ESW_DEST_CHAIN_WITH_SRC_PORT_CHANGE)
+			return true;
+	return false;
+}
+
+static int
+esw_setup_chain_src_port_rewrite(struct mlx5_flow_destination *dest,
+				 struct mlx5_flow_act *flow_act,
+				 struct mlx5_eswitch *esw,
+				 struct mlx5_fs_chains *chains,
+				 struct mlx5_flow_attr *attr,
+				 int *i)
+{
+	struct mlx5_esw_flow_attr *esw_attr = attr->esw_attr;
+	int err;
+
+	if (!(attr->flags & MLX5_ATTR_FLAG_SRC_REWRITE))
+		return -EOPNOTSUPP;
+
+	/* flow steering cannot handle more than one dest with the same ft
+	 * in a single flow
+	 */
+	if (esw_attr->out_count - esw_attr->split_count > 1)
+		return -EOPNOTSUPP;
+
+	err = esw_setup_chain_dest(dest, flow_act, chains, attr->dest_chain, 1, 0, *i);
+	if (err)
+		return err;
+
+	if (esw_attr->dests[esw_attr->split_count].pkt_reformat) {
+		flow_act->action |= MLX5_FLOW_CONTEXT_ACTION_PACKET_REFORMAT;
+		flow_act->pkt_reformat = esw_attr->dests[esw_attr->split_count].pkt_reformat;
+	}
+	(*i)++;
+
+	return 0;
+}
+
+static void esw_cleanup_chain_src_port_rewrite(struct mlx5_eswitch *esw,
+					       struct mlx5_flow_attr *attr)
+{
+	struct mlx5_esw_flow_attr *esw_attr = attr->esw_attr;
+
+	esw_put_dest_tables_loop(esw, attr, esw_attr->split_count, esw_attr->out_count);
+}
+
+static bool
+esw_is_indir_table(struct mlx5_eswitch *esw, struct mlx5_flow_attr *attr)
+{
+	struct mlx5_esw_flow_attr *esw_attr = attr->esw_attr;
+	bool result = false;
+	int i;
+
+	/* Indirect table is supported only for flows with in_port uplink
+	 * and the destination is vport on the same eswitch as the uplink,
+	 * return false in case at least one of destinations doesn't meet
+	 * this criteria.
+	 */
+	for (i = esw_attr->split_count; i < esw_attr->out_count; i++) {
+		if (esw_attr->dests[i].rep &&
+		    mlx5_esw_indir_table_needed(esw, attr, esw_attr->dests[i].rep->vport,
+						esw_attr->dests[i].mdev)) {
+			result = true;
+		} else {
+			result = false;
+			break;
+		}
+	}
+	return result;
+}
+
+static int
+esw_setup_indir_table(struct mlx5_flow_destination *dest,
+		      struct mlx5_flow_act *flow_act,
+		      struct mlx5_eswitch *esw,
+		      struct mlx5_flow_attr *attr,
+		      bool ignore_flow_lvl,
+		      int *i)
+{
+	struct mlx5_esw_flow_attr *esw_attr = attr->esw_attr;
+	int j, err;
+
+	if (!(attr->flags & MLX5_ATTR_FLAG_SRC_REWRITE))
+		return -EOPNOTSUPP;
+
+	for (j = esw_attr->split_count; j < esw_attr->out_count; j++, (*i)++) {
+		if (ignore_flow_lvl)
+			flow_act->flags |= FLOW_ACT_IGNORE_FLOW_LEVEL;
+		dest[*i].type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
+
+		dest[*i].ft = mlx5_esw_indir_table_get(esw, attr,
+						       esw_attr->dests[j].rep->vport, false);
+		if (IS_ERR(dest[*i].ft)) {
+			err = PTR_ERR(dest[*i].ft);
+			goto err_indir_tbl_get;
+		}
+	}
+
+	if (mlx5_esw_indir_table_decap_vport(attr)) {
+		err = esw_setup_decap_indir(esw, attr);
+		if (err)
+			goto err_indir_tbl_get;
+	}
+
+	return 0;
+
+err_indir_tbl_get:
+	esw_put_dest_tables_loop(esw, attr, esw_attr->split_count, j);
+	return err;
+}
+
+static void esw_cleanup_indir_table(struct mlx5_eswitch *esw, struct mlx5_flow_attr *attr)
+{
+	struct mlx5_esw_flow_attr *esw_attr = attr->esw_attr;
+
+	esw_put_dest_tables_loop(esw, attr, esw_attr->split_count, esw_attr->out_count);
+	esw_cleanup_decap_indir(esw, attr);
+}
+
+static void
+esw_cleanup_chain_dest(struct mlx5_fs_chains *chains, u32 chain, u32 prio, u32 level)
+{
+	mlx5_chains_put_table(chains, chain, prio, level);
+}
+
+static void
+esw_setup_vport_dest(struct mlx5_flow_destination *dest, struct mlx5_flow_act *flow_act,
+		     struct mlx5_eswitch *esw, struct mlx5_esw_flow_attr *esw_attr,
+		     int attr_idx, int dest_idx, bool pkt_reformat)
+{
+	dest[dest_idx].type = MLX5_FLOW_DESTINATION_TYPE_VPORT;
+	dest[dest_idx].vport.num = esw_attr->dests[attr_idx].rep->vport;
+	if (MLX5_CAP_ESW(esw->dev, merged_eswitch)) {
+		dest[dest_idx].vport.vhca_id =
+			MLX5_CAP_GEN(esw_attr->dests[attr_idx].mdev, vhca_id);
+		dest[dest_idx].vport.flags |= MLX5_FLOW_DEST_VPORT_VHCA_ID;
+		if (dest[dest_idx].vport.num == MLX5_VPORT_UPLINK &&
+		    mlx5_lag_mpesw_is_activated(esw->dev))
+			dest[dest_idx].type = MLX5_FLOW_DESTINATION_TYPE_UPLINK;
+	}
+	if (esw_attr->dests[attr_idx].flags & MLX5_ESW_DEST_ENCAP_VALID) {
+		if (pkt_reformat) {
+			flow_act->action |= MLX5_FLOW_CONTEXT_ACTION_PACKET_REFORMAT;
+			flow_act->pkt_reformat = esw_attr->dests[attr_idx].pkt_reformat;
+		}
+		dest[dest_idx].vport.flags |= MLX5_FLOW_DEST_VPORT_REFORMAT_ID;
+		dest[dest_idx].vport.pkt_reformat = esw_attr->dests[attr_idx].pkt_reformat;
+	}
+}
+
+static int
+esw_setup_vport_dests(struct mlx5_flow_destination *dest, struct mlx5_flow_act *flow_act,
+		      struct mlx5_eswitch *esw, struct mlx5_esw_flow_attr *esw_attr,
+		      int i)
+{
+	int j;
+
+	for (j = esw_attr->split_count; j < esw_attr->out_count; j++, i++)
+		esw_setup_vport_dest(dest, flow_act, esw, esw_attr, j, i, true);
+	return i;
+}
+
+static bool
+esw_src_port_rewrite_supported(struct mlx5_eswitch *esw)
+{
+	return MLX5_CAP_GEN(esw->dev, reg_c_preserve) &&
+	       mlx5_eswitch_vport_match_metadata_enabled(esw) &&
+	       MLX5_CAP_ESW_FLOWTABLE_FDB(esw->dev, ignore_flow_level);
+}
+
+static int
+esw_setup_dests(struct mlx5_flow_destination *dest,
+		struct mlx5_flow_act *flow_act,
+		struct mlx5_eswitch *esw,
+		struct mlx5_flow_attr *attr,
+		struct mlx5_flow_spec *spec,
+		int *i)
+{
+	struct mlx5_esw_flow_attr *esw_attr = attr->esw_attr;
+	struct mlx5_fs_chains *chains = esw_chains(esw);
+	int err = 0;
+
+	if (!mlx5_eswitch_termtbl_required(esw, attr, flow_act, spec) &&
+	    esw_src_port_rewrite_supported(esw))
+		attr->flags |= MLX5_ATTR_FLAG_SRC_REWRITE;
+
+	if (attr->flags & MLX5_ATTR_FLAG_SAMPLE &&
+	    !(attr->flags & MLX5_ATTR_FLAG_SLOW_PATH)) {
+		esw_setup_sampler_dest(dest, flow_act, attr->sample_attr.sampler_id, *i);
+		(*i)++;
+	} else if (attr->flags & MLX5_ATTR_FLAG_SLOW_PATH) {
+		esw_setup_slow_path_dest(dest, flow_act, esw, *i);
+		(*i)++;
+	} else if (attr->flags & MLX5_ATTR_FLAG_ACCEPT) {
+		esw_setup_accept_dest(dest, flow_act, chains, *i);
+		(*i)++;
+	} else if (esw_is_indir_table(esw, attr)) {
+		err = esw_setup_indir_table(dest, flow_act, esw, attr, true, i);
+	} else if (esw_is_chain_src_port_rewrite(esw, esw_attr)) {
+		err = esw_setup_chain_src_port_rewrite(dest, flow_act, esw, chains, attr, i);
+	} else {
+		*i = esw_setup_vport_dests(dest, flow_act, esw, esw_attr, *i);
+
+		if (attr->dest_ft) {
+			err = esw_setup_ft_dest(dest, flow_act, esw, attr, *i);
+			(*i)++;
+		} else if (attr->dest_chain) {
+			err = esw_setup_chain_dest(dest, flow_act, chains, attr->dest_chain,
+						   1, 0, *i);
+			(*i)++;
+		}
+	}
+
+	return err;
+}
+
+static void
+esw_cleanup_dests(struct mlx5_eswitch *esw,
+		  struct mlx5_flow_attr *attr)
+{
+	struct mlx5_esw_flow_attr *esw_attr = attr->esw_attr;
+	struct mlx5_fs_chains *chains = esw_chains(esw);
+
+	if (attr->dest_ft) {
+		esw_cleanup_decap_indir(esw, attr);
+	} else if (!mlx5e_tc_attr_flags_skip(attr->flags)) {
+		if (attr->dest_chain)
+			esw_cleanup_chain_dest(chains, attr->dest_chain, 1, 0);
+		else if (esw_is_indir_table(esw, attr))
+			esw_cleanup_indir_table(esw, attr);
+		else if (esw_is_chain_src_port_rewrite(esw, esw_attr))
+			esw_cleanup_chain_src_port_rewrite(esw, attr);
+	}
+}
+
+static void
+esw_setup_meter(struct mlx5_flow_attr *attr, struct mlx5_flow_act *flow_act)
+{
+	struct mlx5e_flow_meter_handle *meter;
+
+	meter = attr->meter_attr.meter;
+	flow_act->exe_aso.type = attr->exe_aso_type;
+	flow_act->exe_aso.object_id = meter->obj_id;
+	flow_act->exe_aso.flow_meter.meter_idx = meter->idx;
+	flow_act->exe_aso.flow_meter.init_color = MLX5_FLOW_METER_COLOR_GREEN;
+	/* use metadata reg 5 for packet color */
+	flow_act->exe_aso.return_reg_id = 5;
+}
+
+struct mlx5_flow_handle *
+mlx5_eswitch_add_offloaded_rule(struct mlx5_eswitch *esw,
+				struct mlx5_flow_spec *spec,
+				struct mlx5_flow_attr *attr)
+{
+	struct mlx5_flow_act flow_act = { .flags = FLOW_ACT_NO_APPEND, };
+	struct mlx5_esw_flow_attr *esw_attr = attr->esw_attr;
+	struct mlx5_fs_chains *chains = esw_chains(esw);
+	bool split = !!(esw_attr->split_count);
+	struct mlx5_vport_tbl_attr fwd_attr;
+	struct mlx5_flow_destination *dest;
+	struct mlx5_flow_handle *rule;
+	struct mlx5_flow_table *fdb;
+	int i = 0;
+
+	if (esw->mode != MLX5_ESWITCH_OFFLOADS)
+		return ERR_PTR(-EOPNOTSUPP);
+
+	dest = kcalloc(MLX5_MAX_FLOW_FWD_VPORTS + 1, sizeof(*dest), GFP_KERNEL);
+	if (!dest)
+		return ERR_PTR(-ENOMEM);
+
+	flow_act.action = attr->action;
+	/* if per flow vlan pop/push is emulated, don't set that into the firmware */
+	if (!mlx5_eswitch_vlan_actions_supported(esw->dev, 1))
+		flow_act.action &= ~(MLX5_FLOW_CONTEXT_ACTION_VLAN_PUSH |
+				     MLX5_FLOW_CONTEXT_ACTION_VLAN_POP);
+	else if (flow_act.action & MLX5_FLOW_CONTEXT_ACTION_VLAN_PUSH) {
+		flow_act.vlan[0].ethtype = ntohs(esw_attr->vlan_proto[0]);
+		flow_act.vlan[0].vid = esw_attr->vlan_vid[0];
+		flow_act.vlan[0].prio = esw_attr->vlan_prio[0];
+		if (flow_act.action & MLX5_FLOW_CONTEXT_ACTION_VLAN_PUSH_2) {
+			flow_act.vlan[1].ethtype = ntohs(esw_attr->vlan_proto[1]);
+			flow_act.vlan[1].vid = esw_attr->vlan_vid[1];
+			flow_act.vlan[1].prio = esw_attr->vlan_prio[1];
+		}
+	}
+
+	mlx5_eswitch_set_rule_flow_source(esw, spec, esw_attr);
+
+	if (flow_act.action & MLX5_FLOW_CONTEXT_ACTION_FWD_DEST) {
+		int err;
+
+		err = esw_setup_dests(dest, &flow_act, esw, attr, spec, &i);
+		if (err) {
+			rule = ERR_PTR(err);
+			goto err_create_goto_table;
+		}
+	}
+
+	if (esw_attr->decap_pkt_reformat)
+		flow_act.pkt_reformat = esw_attr->decap_pkt_reformat;
+
+	if (flow_act.action & MLX5_FLOW_CONTEXT_ACTION_COUNT) {
+		dest[i].type = MLX5_FLOW_DESTINATION_TYPE_COUNTER;
+		dest[i].counter_id = mlx5_fc_id(attr->counter);
+		i++;
+	}
+
+	if (attr->outer_match_level != MLX5_MATCH_NONE)
+		spec->match_criteria_enable |= MLX5_MATCH_OUTER_HEADERS;
+	if (attr->inner_match_level != MLX5_MATCH_NONE)
+		spec->match_criteria_enable |= MLX5_MATCH_INNER_HEADERS;
+
+	if (flow_act.action & MLX5_FLOW_CONTEXT_ACTION_MOD_HDR)
+		flow_act.modify_hdr = attr->modify_hdr;
+
+	if ((flow_act.action & MLX5_FLOW_CONTEXT_ACTION_EXECUTE_ASO) &&
+	    attr->exe_aso_type == MLX5_EXE_ASO_FLOW_METER)
+		esw_setup_meter(attr, &flow_act);
+
+	if (split) {
+		fwd_attr.chain = attr->chain;
+		fwd_attr.prio = attr->prio;
+		fwd_attr.vport = esw_attr->in_rep->vport;
+		fwd_attr.vport_ns = &mlx5_esw_vport_tbl_mirror_ns;
+
+		fdb = mlx5_esw_vporttbl_get(esw, &fwd_attr);
+	} else {
+		if (attr->chain || attr->prio)
+			fdb = mlx5_chains_get_table(chains, attr->chain,
+						    attr->prio, 0);
+		else
+			fdb = attr->ft;
+
+		if (!(attr->flags & MLX5_ATTR_FLAG_NO_IN_PORT))
+			mlx5_eswitch_set_rule_source_port(esw, spec, attr,
+							  esw_attr->in_mdev->priv.eswitch,
+							  esw_attr->in_rep->vport);
+	}
+	if (IS_ERR(fdb)) {
+		rule = ERR_CAST(fdb);
+		goto err_esw_get;
+	}
+
+	if (mlx5_eswitch_termtbl_required(esw, attr, &flow_act, spec))
+		rule = mlx5_eswitch_add_termtbl_rule(esw, fdb, spec, esw_attr,
+						     &flow_act, dest, i);
+	else
+		rule = mlx5_add_flow_rules(fdb, spec, &flow_act, dest, i);
+	if (IS_ERR(rule))
+		goto err_add_rule;
+	else
+		atomic64_inc(&esw->offloads.num_flows);
+
+	kfree(dest);
+	return rule;
+
+err_add_rule:
+	if (split)
+		mlx5_esw_vporttbl_put(esw, &fwd_attr);
+	else if (attr->chain || attr->prio)
+		mlx5_chains_put_table(chains, attr->chain, attr->prio, 0);
+err_esw_get:
+	esw_cleanup_dests(esw, attr);
+err_create_goto_table:
+	kfree(dest);
+	return rule;
+}
+
+struct mlx5_flow_handle *
+mlx5_eswitch_add_fwd_rule(struct mlx5_eswitch *esw,
+			  struct mlx5_flow_spec *spec,
+			  struct mlx5_flow_attr *attr)
+{
+	struct mlx5_flow_act flow_act = { .flags = FLOW_ACT_NO_APPEND, };
+	struct mlx5_esw_flow_attr *esw_attr = attr->esw_attr;
+	struct mlx5_fs_chains *chains = esw_chains(esw);
+	struct mlx5_vport_tbl_attr fwd_attr;
+	struct mlx5_flow_destination *dest;
+	struct mlx5_flow_table *fast_fdb;
+	struct mlx5_flow_table *fwd_fdb;
+	struct mlx5_flow_handle *rule;
+	int i, err = 0;
+
+	dest = kcalloc(MLX5_MAX_FLOW_FWD_VPORTS + 1, sizeof(*dest), GFP_KERNEL);
+	if (!dest)
+		return ERR_PTR(-ENOMEM);
+
+	fast_fdb = mlx5_chains_get_table(chains, attr->chain, attr->prio, 0);
+	if (IS_ERR(fast_fdb)) {
+		rule = ERR_CAST(fast_fdb);
+		goto err_get_fast;
+	}
+
+	fwd_attr.chain = attr->chain;
+	fwd_attr.prio = attr->prio;
+	fwd_attr.vport = esw_attr->in_rep->vport;
+	fwd_attr.vport_ns = &mlx5_esw_vport_tbl_mirror_ns;
+	fwd_fdb = mlx5_esw_vporttbl_get(esw, &fwd_attr);
+	if (IS_ERR(fwd_fdb)) {
+		rule = ERR_CAST(fwd_fdb);
+		goto err_get_fwd;
+	}
+
+	flow_act.action = MLX5_FLOW_CONTEXT_ACTION_FWD_DEST;
+	for (i = 0; i < esw_attr->split_count; i++) {
+		if (esw_attr->dests[i].flags & MLX5_ESW_DEST_CHAIN_WITH_SRC_PORT_CHANGE)
+			/* Source port rewrite (forward to ovs internal port or statck device) isn't
+			 * supported in the rule of split action.
+			 */
+			err = -EOPNOTSUPP;
+		else
+			esw_setup_vport_dest(dest, &flow_act, esw, esw_attr, i, i, false);
+
+		if (err) {
+			rule = ERR_PTR(err);
+			goto err_chain_src_rewrite;
+		}
+	}
+	dest[i].type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
+	dest[i].ft = fwd_fdb;
+	i++;
+
+	mlx5_eswitch_set_rule_source_port(esw, spec, attr,
+					  esw_attr->in_mdev->priv.eswitch,
+					  esw_attr->in_rep->vport);
+
+	if (attr->outer_match_level != MLX5_MATCH_NONE)
+		spec->match_criteria_enable |= MLX5_MATCH_OUTER_HEADERS;
+
+	flow_act.flags |= FLOW_ACT_IGNORE_FLOW_LEVEL;
+	rule = mlx5_add_flow_rules(fast_fdb, spec, &flow_act, dest, i);
+
+	if (IS_ERR(rule)) {
+		i = esw_attr->split_count;
+		goto err_chain_src_rewrite;
+	}
+
+	atomic64_inc(&esw->offloads.num_flows);
+
+	kfree(dest);
+	return rule;
+err_chain_src_rewrite:
+	mlx5_esw_vporttbl_put(esw, &fwd_attr);
+err_get_fwd:
+	mlx5_chains_put_table(chains, attr->chain, attr->prio, 0);
+err_get_fast:
+	kfree(dest);
+	return rule;
+}
+
+static void
+__mlx5_eswitch_del_rule(struct mlx5_eswitch *esw,
+			struct mlx5_flow_handle *rule,
+			struct mlx5_flow_attr *attr,
+			bool fwd_rule)
+{
+	struct mlx5_esw_flow_attr *esw_attr = attr->esw_attr;
+	struct mlx5_fs_chains *chains = esw_chains(esw);
+	bool split = (esw_attr->split_count > 0);
+	struct mlx5_vport_tbl_attr fwd_attr;
+	int i;
+
+	mlx5_del_flow_rules(rule);
+
+	if (!mlx5e_tc_attr_flags_skip(attr->flags)) {
+		/* unref the term table */
+		for (i = 0; i < MLX5_MAX_FLOW_FWD_VPORTS; i++) {
+			if (esw_attr->dests[i].termtbl)
+				mlx5_eswitch_termtbl_put(esw, esw_attr->dests[i].termtbl);
+		}
+	}
+
+	atomic64_dec(&esw->offloads.num_flows);
+
+	if (fwd_rule || split) {
+		fwd_attr.chain = attr->chain;
+		fwd_attr.prio = attr->prio;
+		fwd_attr.vport = esw_attr->in_rep->vport;
+		fwd_attr.vport_ns = &mlx5_esw_vport_tbl_mirror_ns;
+	}
+
+	if (fwd_rule)  {
+		mlx5_esw_vporttbl_put(esw, &fwd_attr);
+		mlx5_chains_put_table(chains, attr->chain, attr->prio, 0);
+	} else {
+		if (split)
+			mlx5_esw_vporttbl_put(esw, &fwd_attr);
+		else if (attr->chain || attr->prio)
+			mlx5_chains_put_table(chains, attr->chain, attr->prio, 0);
+		esw_cleanup_dests(esw, attr);
+	}
+}
+
+void
+mlx5_eswitch_del_offloaded_rule(struct mlx5_eswitch *esw,
+				struct mlx5_flow_handle *rule,
+				struct mlx5_flow_attr *attr)
+{
+	__mlx5_eswitch_del_rule(esw, rule, attr, false);
+}
+
+void
+mlx5_eswitch_del_fwd_rule(struct mlx5_eswitch *esw,
+			  struct mlx5_flow_handle *rule,
+			  struct mlx5_flow_attr *attr)
+{
+	__mlx5_eswitch_del_rule(esw, rule, attr, true);
+}
+
+static int esw_set_global_vlan_pop(struct mlx5_eswitch *esw, u8 val)
+{
+	struct mlx5_eswitch_rep *rep;
+	unsigned long i;
+	int err = 0;
+
+	esw_debug(esw->dev, "%s applying global %s policy\n", __func__, val ? "pop" : "none");
+	mlx5_esw_for_each_host_func_vport(esw, i, rep, esw->esw_funcs.num_vfs) {
+		if (atomic_read(&rep->rep_data[REP_ETH].state) != REP_LOADED)
+			continue;
+
+		err = __mlx5_eswitch_set_vport_vlan(esw, rep->vport, 0, 0, val);
+		if (err)
+			goto out;
+	}
+
+out:
+	return err;
+}
+
+static struct mlx5_eswitch_rep *
+esw_vlan_action_get_vport(struct mlx5_esw_flow_attr *attr, bool push, bool pop)
+{
+	struct mlx5_eswitch_rep *in_rep, *out_rep, *vport = NULL;
+
+	in_rep  = attr->in_rep;
+	out_rep = attr->dests[0].rep;
+
+	if (push)
+		vport = in_rep;
+	else if (pop)
+		vport = out_rep;
+	else
+		vport = in_rep;
+
+	return vport;
+}
+
+static int esw_add_vlan_action_check(struct mlx5_esw_flow_attr *attr,
+				     bool push, bool pop, bool fwd)
+{
+	struct mlx5_eswitch_rep *in_rep, *out_rep;
+
+	if ((push || pop) && !fwd)
+		goto out_notsupp;
+
+	in_rep  = attr->in_rep;
+	out_rep = attr->dests[0].rep;
+
+	if (push && in_rep->vport == MLX5_VPORT_UPLINK)
+		goto out_notsupp;
+
+	if (pop && out_rep->vport == MLX5_VPORT_UPLINK)
+		goto out_notsupp;
+
+	/* vport has vlan push configured, can't offload VF --> wire rules w.o it */
+	if (!push && !pop && fwd)
+		if (in_rep->vlan && out_rep->vport == MLX5_VPORT_UPLINK)
+			goto out_notsupp;
+
+	/* protects against (1) setting rules with different vlans to push and
+	 * (2) setting rules w.o vlans (attr->vlan = 0) && w. vlans to push (!= 0)
+	 */
+	if (push && in_rep->vlan_refcount && (in_rep->vlan != attr->vlan_vid[0]))
+		goto out_notsupp;
+
+	return 0;
+
+out_notsupp:
+	return -EOPNOTSUPP;
+}
+
+int mlx5_eswitch_add_vlan_action(struct mlx5_eswitch *esw,
+				 struct mlx5_flow_attr *attr)
+{
+	struct offloads_fdb *offloads = &esw->fdb_table.offloads;
+	struct mlx5_esw_flow_attr *esw_attr = attr->esw_attr;
+	struct mlx5_eswitch_rep *vport = NULL;
+	bool push, pop, fwd;
+	int err = 0;
+
+	/* nop if we're on the vlan push/pop non emulation mode */
+	if (mlx5_eswitch_vlan_actions_supported(esw->dev, 1))
+		return 0;
+
+	push = !!(attr->action & MLX5_FLOW_CONTEXT_ACTION_VLAN_PUSH);
+	pop  = !!(attr->action & MLX5_FLOW_CONTEXT_ACTION_VLAN_POP);
+	fwd  = !!((attr->action & MLX5_FLOW_CONTEXT_ACTION_FWD_DEST) &&
+		   !attr->dest_chain);
+
+	mutex_lock(&esw->state_lock);
+
+	err = esw_add_vlan_action_check(esw_attr, push, pop, fwd);
+	if (err)
+		goto unlock;
+
+	attr->flags &= ~MLX5_ATTR_FLAG_VLAN_HANDLED;
+
+	vport = esw_vlan_action_get_vport(esw_attr, push, pop);
+
+	if (!push && !pop && fwd) {
+		/* tracks VF --> wire rules without vlan push action */
+		if (esw_attr->dests[0].rep->vport == MLX5_VPORT_UPLINK) {
+			vport->vlan_refcount++;
+			attr->flags |= MLX5_ATTR_FLAG_VLAN_HANDLED;
+		}
+
+		goto unlock;
+	}
+
+	if (!push && !pop)
+		goto unlock;
+
+	if (!(offloads->vlan_push_pop_refcount)) {
+		/* it's the 1st vlan rule, apply global vlan pop policy */
+		err = esw_set_global_vlan_pop(esw, SET_VLAN_STRIP);
+		if (err)
+			goto out;
+	}
+	offloads->vlan_push_pop_refcount++;
+
+	if (push) {
+		if (vport->vlan_refcount)
+			goto skip_set_push;
+
+		err = __mlx5_eswitch_set_vport_vlan(esw, vport->vport, esw_attr->vlan_vid[0],
+						    0, SET_VLAN_INSERT | SET_VLAN_STRIP);
+		if (err)
+			goto out;
+		vport->vlan = esw_attr->vlan_vid[0];
+skip_set_push:
+		vport->vlan_refcount++;
+	}
+out:
+	if (!err)
+		attr->flags |= MLX5_ATTR_FLAG_VLAN_HANDLED;
+unlock:
+	mutex_unlock(&esw->state_lock);
+	return err;
+}
+
+int mlx5_eswitch_del_vlan_action(struct mlx5_eswitch *esw,
+				 struct mlx5_flow_attr *attr)
+{
+	struct offloads_fdb *offloads = &esw->fdb_table.offloads;
+	struct mlx5_esw_flow_attr *esw_attr = attr->esw_attr;
+	struct mlx5_eswitch_rep *vport = NULL;
+	bool push, pop, fwd;
+	int err = 0;
+
+	/* nop if we're on the vlan push/pop non emulation mode */
+	if (mlx5_eswitch_vlan_actions_supported(esw->dev, 1))
+		return 0;
+
+	if (!(attr->flags & MLX5_ATTR_FLAG_VLAN_HANDLED))
+		return 0;
+
+	push = !!(attr->action & MLX5_FLOW_CONTEXT_ACTION_VLAN_PUSH);
+	pop  = !!(attr->action & MLX5_FLOW_CONTEXT_ACTION_VLAN_POP);
+	fwd  = !!(attr->action & MLX5_FLOW_CONTEXT_ACTION_FWD_DEST);
+
+	mutex_lock(&esw->state_lock);
+
+	vport = esw_vlan_action_get_vport(esw_attr, push, pop);
+
+	if (!push && !pop && fwd) {
+		/* tracks VF --> wire rules without vlan push action */
+		if (esw_attr->dests[0].rep->vport == MLX5_VPORT_UPLINK)
+			vport->vlan_refcount--;
+
+		goto out;
+	}
+
+	if (push) {
+		vport->vlan_refcount--;
+		if (vport->vlan_refcount)
+			goto skip_unset_push;
+
+		vport->vlan = 0;
+		err = __mlx5_eswitch_set_vport_vlan(esw, vport->vport,
+						    0, 0, SET_VLAN_STRIP);
+		if (err)
+			goto out;
+	}
+
+skip_unset_push:
+	offloads->vlan_push_pop_refcount--;
+	if (offloads->vlan_push_pop_refcount)
+		goto out;
+
+	/* no more vlan rules, stop global vlan pop policy */
+	err = esw_set_global_vlan_pop(esw, 0);
+
+out:
+	mutex_unlock(&esw->state_lock);
+	return err;
+}
+
+struct mlx5_flow_handle *
+mlx5_eswitch_add_send_to_vport_rule(struct mlx5_eswitch *on_esw,
+				    struct mlx5_eswitch *from_esw,
+				    struct mlx5_eswitch_rep *rep,
+				    u32 sqn)
+{
+	struct mlx5_flow_act flow_act = {0};
+	struct mlx5_flow_destination dest = {};
+	struct mlx5_flow_handle *flow_rule;
+	struct mlx5_flow_spec *spec;
+	void *misc;
+
+	spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
+	if (!spec) {
+		flow_rule = ERR_PTR(-ENOMEM);
+		goto out;
+	}
+
+	misc = MLX5_ADDR_OF(fte_match_param, spec->match_value, misc_parameters);
+	MLX5_SET(fte_match_set_misc, misc, source_sqn, sqn);
+	/* source vport is the esw manager */
+	MLX5_SET(fte_match_set_misc, misc, source_port, from_esw->manager_vport);
+	if (MLX5_CAP_ESW(on_esw->dev, merged_eswitch))
+		MLX5_SET(fte_match_set_misc, misc, source_eswitch_owner_vhca_id,
+			 MLX5_CAP_GEN(from_esw->dev, vhca_id));
+
+	misc = MLX5_ADDR_OF(fte_match_param, spec->match_criteria, misc_parameters);
+	MLX5_SET_TO_ONES(fte_match_set_misc, misc, source_sqn);
+	MLX5_SET_TO_ONES(fte_match_set_misc, misc, source_port);
+	if (MLX5_CAP_ESW(on_esw->dev, merged_eswitch))
+		MLX5_SET_TO_ONES(fte_match_set_misc, misc,
+				 source_eswitch_owner_vhca_id);
+
+	spec->match_criteria_enable = MLX5_MATCH_MISC_PARAMETERS;
+	dest.type = MLX5_FLOW_DESTINATION_TYPE_VPORT;
+	dest.vport.num = rep->vport;
+	dest.vport.vhca_id = MLX5_CAP_GEN(rep->esw->dev, vhca_id);
+	dest.vport.flags |= MLX5_FLOW_DEST_VPORT_VHCA_ID;
+	flow_act.action = MLX5_FLOW_CONTEXT_ACTION_FWD_DEST;
+
+	if (MLX5_CAP_ESW_FLOWTABLE(on_esw->dev, flow_source) &&
+	    rep->vport == MLX5_VPORT_UPLINK)
+		spec->flow_context.flow_source = MLX5_FLOW_CONTEXT_FLOW_SOURCE_LOCAL_VPORT;
+
+	flow_rule = mlx5_add_flow_rules(on_esw->fdb_table.offloads.slow_fdb,
+					spec, &flow_act, &dest, 1);
+	if (IS_ERR(flow_rule))
+		esw_warn(on_esw->dev, "FDB: Failed to add send to vport rule err %ld\n",
+			 PTR_ERR(flow_rule));
+out:
+	kvfree(spec);
+	return flow_rule;
+}
+EXPORT_SYMBOL(mlx5_eswitch_add_send_to_vport_rule);
+
+void mlx5_eswitch_del_send_to_vport_rule(struct mlx5_flow_handle *rule)
+{
+	mlx5_del_flow_rules(rule);
+}
+
+void mlx5_eswitch_del_send_to_vport_meta_rule(struct mlx5_flow_handle *rule)
+{
+	if (rule)
+		mlx5_del_flow_rules(rule);
+}
+
+struct mlx5_flow_handle *
+mlx5_eswitch_add_send_to_vport_meta_rule(struct mlx5_eswitch *esw, u16 vport_num)
+{
+	struct mlx5_flow_destination dest = {};
+	struct mlx5_flow_act flow_act = {0};
+	struct mlx5_flow_handle *flow_rule;
+	struct mlx5_flow_spec *spec;
+
+	spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
+	if (!spec)
+		return ERR_PTR(-ENOMEM);
+
+	MLX5_SET(fte_match_param, spec->match_criteria,
+		 misc_parameters_2.metadata_reg_c_0, mlx5_eswitch_get_vport_metadata_mask());
+	MLX5_SET(fte_match_param, spec->match_criteria,
+		 misc_parameters_2.metadata_reg_c_1, ESW_TUN_MASK);
+	MLX5_SET(fte_match_param, spec->match_value, misc_parameters_2.metadata_reg_c_1,
+		 ESW_TUN_SLOW_TABLE_GOTO_VPORT_MARK);
+
+	spec->match_criteria_enable = MLX5_MATCH_MISC_PARAMETERS_2;
+	dest.type = MLX5_FLOW_DESTINATION_TYPE_VPORT;
+	flow_act.action = MLX5_FLOW_CONTEXT_ACTION_FWD_DEST;
+
+	MLX5_SET(fte_match_param, spec->match_value, misc_parameters_2.metadata_reg_c_0,
+		 mlx5_eswitch_get_vport_metadata_for_match(esw, vport_num));
+	dest.vport.num = vport_num;
+
+	flow_rule = mlx5_add_flow_rules(esw->fdb_table.offloads.slow_fdb,
+					spec, &flow_act, &dest, 1);
+	if (IS_ERR(flow_rule))
+		esw_warn(esw->dev, "FDB: Failed to add send to vport meta rule vport %d, err %ld\n",
+			 vport_num, PTR_ERR(flow_rule));
+
+	kvfree(spec);
+	return flow_rule;
+}
+
+static bool mlx5_eswitch_reg_c1_loopback_supported(struct mlx5_eswitch *esw)
+{
+	return MLX5_CAP_ESW_FLOWTABLE(esw->dev, fdb_to_vport_reg_c_id) &
+	       MLX5_FDB_TO_VPORT_REG_C_1;
+}
+
+static int esw_set_passing_vport_metadata(struct mlx5_eswitch *esw, bool enable)
+{
+	u32 out[MLX5_ST_SZ_DW(query_esw_vport_context_out)] = {};
+	u32 min[MLX5_ST_SZ_DW(modify_esw_vport_context_in)] = {};
+	u32 in[MLX5_ST_SZ_DW(query_esw_vport_context_in)] = {};
+	u8 curr, wanted;
+	int err;
+
+	if (!mlx5_eswitch_reg_c1_loopback_supported(esw) &&
+	    !mlx5_eswitch_vport_match_metadata_enabled(esw))
+		return 0;
+
+	MLX5_SET(query_esw_vport_context_in, in, opcode,
+		 MLX5_CMD_OP_QUERY_ESW_VPORT_CONTEXT);
+	err = mlx5_cmd_exec_inout(esw->dev, query_esw_vport_context, in, out);
+	if (err)
+		return err;
+
+	curr = MLX5_GET(query_esw_vport_context_out, out,
+			esw_vport_context.fdb_to_vport_reg_c_id);
+	wanted = MLX5_FDB_TO_VPORT_REG_C_0;
+	if (mlx5_eswitch_reg_c1_loopback_supported(esw))
+		wanted |= MLX5_FDB_TO_VPORT_REG_C_1;
+
+	if (enable)
+		curr |= wanted;
+	else
+		curr &= ~wanted;
+
+	MLX5_SET(modify_esw_vport_context_in, min,
+		 esw_vport_context.fdb_to_vport_reg_c_id, curr);
+	MLX5_SET(modify_esw_vport_context_in, min,
+		 field_select.fdb_to_vport_reg_c_id, 1);
+
+	err = mlx5_eswitch_modify_esw_vport_context(esw->dev, 0, false, min);
+	if (!err) {
+		if (enable && (curr & MLX5_FDB_TO_VPORT_REG_C_1))
+			esw->flags |= MLX5_ESWITCH_REG_C1_LOOPBACK_ENABLED;
+		else
+			esw->flags &= ~MLX5_ESWITCH_REG_C1_LOOPBACK_ENABLED;
+	}
+
+	return err;
+}
+
+static void peer_miss_rules_setup(struct mlx5_eswitch *esw,
+				  struct mlx5_core_dev *peer_dev,
+				  struct mlx5_flow_spec *spec,
+				  struct mlx5_flow_destination *dest)
+{
+	void *misc;
+
+	if (mlx5_eswitch_vport_match_metadata_enabled(esw)) {
+		misc = MLX5_ADDR_OF(fte_match_param, spec->match_criteria,
+				    misc_parameters_2);
+		MLX5_SET(fte_match_set_misc2, misc, metadata_reg_c_0,
+			 mlx5_eswitch_get_vport_metadata_mask());
+
+		spec->match_criteria_enable = MLX5_MATCH_MISC_PARAMETERS_2;
+	} else {
+		misc = MLX5_ADDR_OF(fte_match_param, spec->match_value,
+				    misc_parameters);
+
+		MLX5_SET(fte_match_set_misc, misc, source_eswitch_owner_vhca_id,
+			 MLX5_CAP_GEN(peer_dev, vhca_id));
+
+		spec->match_criteria_enable = MLX5_MATCH_MISC_PARAMETERS;
+
+		misc = MLX5_ADDR_OF(fte_match_param, spec->match_criteria,
+				    misc_parameters);
+		MLX5_SET_TO_ONES(fte_match_set_misc, misc, source_port);
+		MLX5_SET_TO_ONES(fte_match_set_misc, misc,
+				 source_eswitch_owner_vhca_id);
+	}
+
+	dest->type = MLX5_FLOW_DESTINATION_TYPE_VPORT;
+	dest->vport.num = peer_dev->priv.eswitch->manager_vport;
+	dest->vport.vhca_id = MLX5_CAP_GEN(peer_dev, vhca_id);
+	dest->vport.flags |= MLX5_FLOW_DEST_VPORT_VHCA_ID;
+}
+
+static void esw_set_peer_miss_rule_source_port(struct mlx5_eswitch *esw,
+					       struct mlx5_eswitch *peer_esw,
+					       struct mlx5_flow_spec *spec,
+					       u16 vport)
+{
+	void *misc;
+
+	if (mlx5_eswitch_vport_match_metadata_enabled(esw)) {
+		misc = MLX5_ADDR_OF(fte_match_param, spec->match_value,
+				    misc_parameters_2);
+		MLX5_SET(fte_match_set_misc2, misc, metadata_reg_c_0,
+			 mlx5_eswitch_get_vport_metadata_for_match(peer_esw,
+								   vport));
+	} else {
+		misc = MLX5_ADDR_OF(fte_match_param, spec->match_value,
+				    misc_parameters);
+		MLX5_SET(fte_match_set_misc, misc, source_port, vport);
+	}
+}
+
+static int esw_add_fdb_peer_miss_rules(struct mlx5_eswitch *esw,
+				       struct mlx5_core_dev *peer_dev)
+{
+	struct mlx5_flow_destination dest = {};
+	struct mlx5_flow_act flow_act = {0};
+	struct mlx5_flow_handle **flows;
+	/* total vports is the same for both e-switches */
+	int nvports = esw->total_vports;
+	struct mlx5_flow_handle *flow;
+	struct mlx5_flow_spec *spec;
+	struct mlx5_vport *vport;
+	unsigned long i;
+	void *misc;
+	int err;
+
+	spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
+	if (!spec)
+		return -ENOMEM;
+
+	peer_miss_rules_setup(esw, peer_dev, spec, &dest);
+
+	flows = kvcalloc(nvports, sizeof(*flows), GFP_KERNEL);
+	if (!flows) {
+		err = -ENOMEM;
+		goto alloc_flows_err;
+	}
+
+	flow_act.action = MLX5_FLOW_CONTEXT_ACTION_FWD_DEST;
+	misc = MLX5_ADDR_OF(fte_match_param, spec->match_value,
+			    misc_parameters);
+
+	if (mlx5_core_is_ecpf_esw_manager(esw->dev)) {
+		vport = mlx5_eswitch_get_vport(esw, MLX5_VPORT_PF);
+		esw_set_peer_miss_rule_source_port(esw, peer_dev->priv.eswitch,
+						   spec, MLX5_VPORT_PF);
+
+		flow = mlx5_add_flow_rules(esw->fdb_table.offloads.slow_fdb,
+					   spec, &flow_act, &dest, 1);
+		if (IS_ERR(flow)) {
+			err = PTR_ERR(flow);
+			goto add_pf_flow_err;
+		}
+		flows[vport->index] = flow;
+	}
+
+	if (mlx5_ecpf_vport_exists(esw->dev)) {
+		vport = mlx5_eswitch_get_vport(esw, MLX5_VPORT_ECPF);
+		MLX5_SET(fte_match_set_misc, misc, source_port, MLX5_VPORT_ECPF);
+		flow = mlx5_add_flow_rules(esw->fdb_table.offloads.slow_fdb,
+					   spec, &flow_act, &dest, 1);
+		if (IS_ERR(flow)) {
+			err = PTR_ERR(flow);
+			goto add_ecpf_flow_err;
+		}
+		flows[vport->index] = flow;
+	}
+
+	mlx5_esw_for_each_vf_vport(esw, i, vport, mlx5_core_max_vfs(esw->dev)) {
+		esw_set_peer_miss_rule_source_port(esw,
+						   peer_dev->priv.eswitch,
+						   spec, vport->vport);
+
+		flow = mlx5_add_flow_rules(esw->fdb_table.offloads.slow_fdb,
+					   spec, &flow_act, &dest, 1);
+		if (IS_ERR(flow)) {
+			err = PTR_ERR(flow);
+			goto add_vf_flow_err;
+		}
+		flows[vport->index] = flow;
+	}
+
+	esw->fdb_table.offloads.peer_miss_rules = flows;
+
+	kvfree(spec);
+	return 0;
+
+add_vf_flow_err:
+	mlx5_esw_for_each_vf_vport(esw, i, vport, mlx5_core_max_vfs(esw->dev)) {
+		if (!flows[vport->index])
+			continue;
+		mlx5_del_flow_rules(flows[vport->index]);
+	}
+	if (mlx5_ecpf_vport_exists(esw->dev)) {
+		vport = mlx5_eswitch_get_vport(esw, MLX5_VPORT_ECPF);
+		mlx5_del_flow_rules(flows[vport->index]);
+	}
+add_ecpf_flow_err:
+	if (mlx5_core_is_ecpf_esw_manager(esw->dev)) {
+		vport = mlx5_eswitch_get_vport(esw, MLX5_VPORT_PF);
+		mlx5_del_flow_rules(flows[vport->index]);
+	}
+add_pf_flow_err:
+	esw_warn(esw->dev, "FDB: Failed to add peer miss flow rule err %d\n", err);
+	kvfree(flows);
+alloc_flows_err:
+	kvfree(spec);
+	return err;
+}
+
+static void esw_del_fdb_peer_miss_rules(struct mlx5_eswitch *esw)
+{
+	struct mlx5_flow_handle **flows;
+	struct mlx5_vport *vport;
+	unsigned long i;
+
+	flows = esw->fdb_table.offloads.peer_miss_rules;
+
+	mlx5_esw_for_each_vf_vport(esw, i, vport, mlx5_core_max_vfs(esw->dev))
+		mlx5_del_flow_rules(flows[vport->index]);
+
+	if (mlx5_ecpf_vport_exists(esw->dev)) {
+		vport = mlx5_eswitch_get_vport(esw, MLX5_VPORT_ECPF);
+		mlx5_del_flow_rules(flows[vport->index]);
+	}
+
+	if (mlx5_core_is_ecpf_esw_manager(esw->dev)) {
+		vport = mlx5_eswitch_get_vport(esw, MLX5_VPORT_PF);
+		mlx5_del_flow_rules(flows[vport->index]);
+	}
+	kvfree(flows);
+}
+
+static int esw_add_fdb_miss_rule(struct mlx5_eswitch *esw)
+{
+	struct mlx5_flow_act flow_act = {0};
+	struct mlx5_flow_destination dest = {};
+	struct mlx5_flow_handle *flow_rule = NULL;
+	struct mlx5_flow_spec *spec;
+	void *headers_c;
+	void *headers_v;
+	int err = 0;
+	u8 *dmac_c;
+	u8 *dmac_v;
+
+	spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
+	if (!spec) {
+		err = -ENOMEM;
+		goto out;
+	}
+
+	spec->match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
+	headers_c = MLX5_ADDR_OF(fte_match_param, spec->match_criteria,
+				 outer_headers);
+	dmac_c = MLX5_ADDR_OF(fte_match_param, headers_c,
+			      outer_headers.dmac_47_16);
+	dmac_c[0] = 0x01;
+
+	dest.type = MLX5_FLOW_DESTINATION_TYPE_VPORT;
+	dest.vport.num = esw->manager_vport;
+	flow_act.action = MLX5_FLOW_CONTEXT_ACTION_FWD_DEST;
+
+	flow_rule = mlx5_add_flow_rules(esw->fdb_table.offloads.slow_fdb,
+					spec, &flow_act, &dest, 1);
+	if (IS_ERR(flow_rule)) {
+		err = PTR_ERR(flow_rule);
+		esw_warn(esw->dev,  "FDB: Failed to add unicast miss flow rule err %d\n", err);
+		goto out;
+	}
+
+	esw->fdb_table.offloads.miss_rule_uni = flow_rule;
+
+	headers_v = MLX5_ADDR_OF(fte_match_param, spec->match_value,
+				 outer_headers);
+	dmac_v = MLX5_ADDR_OF(fte_match_param, headers_v,
+			      outer_headers.dmac_47_16);
+	dmac_v[0] = 0x01;
+	flow_rule = mlx5_add_flow_rules(esw->fdb_table.offloads.slow_fdb,
+					spec, &flow_act, &dest, 1);
+	if (IS_ERR(flow_rule)) {
+		err = PTR_ERR(flow_rule);
+		esw_warn(esw->dev, "FDB: Failed to add multicast miss flow rule err %d\n", err);
+		mlx5_del_flow_rules(esw->fdb_table.offloads.miss_rule_uni);
+		goto out;
+	}
+
+	esw->fdb_table.offloads.miss_rule_multi = flow_rule;
+
+out:
+	kvfree(spec);
+	return err;
+}
+
+struct mlx5_flow_handle *
+esw_add_restore_rule(struct mlx5_eswitch *esw, u32 tag)
+{
+	struct mlx5_flow_act flow_act = { .flags = FLOW_ACT_NO_APPEND, };
+	struct mlx5_flow_table *ft = esw->offloads.ft_offloads_restore;
+	struct mlx5_flow_context *flow_context;
+	struct mlx5_flow_handle *flow_rule;
+	struct mlx5_flow_destination dest;
+	struct mlx5_flow_spec *spec;
+	void *misc;
+
+	if (!mlx5_eswitch_reg_c1_loopback_supported(esw))
+		return ERR_PTR(-EOPNOTSUPP);
+
+	spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
+	if (!spec)
+		return ERR_PTR(-ENOMEM);
+
+	misc = MLX5_ADDR_OF(fte_match_param, spec->match_criteria,
+			    misc_parameters_2);
+	MLX5_SET(fte_match_set_misc2, misc, metadata_reg_c_0,
+		 ESW_REG_C0_USER_DATA_METADATA_MASK);
+	misc = MLX5_ADDR_OF(fte_match_param, spec->match_value,
+			    misc_parameters_2);
+	MLX5_SET(fte_match_set_misc2, misc, metadata_reg_c_0, tag);
+	spec->match_criteria_enable = MLX5_MATCH_MISC_PARAMETERS_2;
+	flow_act.action = MLX5_FLOW_CONTEXT_ACTION_FWD_DEST |
+			  MLX5_FLOW_CONTEXT_ACTION_MOD_HDR;
+	flow_act.modify_hdr = esw->offloads.restore_copy_hdr_id;
+
+	flow_context = &spec->flow_context;
+	flow_context->flags |= FLOW_CONTEXT_HAS_TAG;
+	flow_context->flow_tag = tag;
+	dest.type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
+	dest.ft = esw->offloads.ft_offloads;
+
+	flow_rule = mlx5_add_flow_rules(ft, spec, &flow_act, &dest, 1);
+	kvfree(spec);
+
+	if (IS_ERR(flow_rule))
+		esw_warn(esw->dev,
+			 "Failed to create restore rule for tag: %d, err(%d)\n",
+			 tag, (int)PTR_ERR(flow_rule));
+
+	return flow_rule;
+}
+
+#define MAX_PF_SQ 256
+#define MAX_SQ_NVPORTS 32
+
+static void esw_set_flow_group_source_port(struct mlx5_eswitch *esw,
+					   u32 *flow_group_in)
+{
+	void *match_criteria = MLX5_ADDR_OF(create_flow_group_in,
+					    flow_group_in,
+					    match_criteria);
+
+	if (mlx5_eswitch_vport_match_metadata_enabled(esw)) {
+		MLX5_SET(create_flow_group_in, flow_group_in,
+			 match_criteria_enable,
+			 MLX5_MATCH_MISC_PARAMETERS_2);
+
+		MLX5_SET(fte_match_param, match_criteria,
+			 misc_parameters_2.metadata_reg_c_0,
+			 mlx5_eswitch_get_vport_metadata_mask());
+	} else {
+		MLX5_SET(create_flow_group_in, flow_group_in,
+			 match_criteria_enable,
+			 MLX5_MATCH_MISC_PARAMETERS);
+
+		MLX5_SET_TO_ONES(fte_match_param, match_criteria,
+				 misc_parameters.source_port);
+	}
+}
+
+#if IS_ENABLED(CONFIG_MLX5_CLS_ACT)
+static void esw_vport_tbl_put(struct mlx5_eswitch *esw)
+{
+	struct mlx5_vport_tbl_attr attr;
+	struct mlx5_vport *vport;
+	unsigned long i;
+
+	attr.chain = 0;
+	attr.prio = 1;
+	mlx5_esw_for_each_vport(esw, i, vport) {
+		attr.vport = vport->vport;
+		attr.vport_ns = &mlx5_esw_vport_tbl_mirror_ns;
+		mlx5_esw_vporttbl_put(esw, &attr);
+	}
+}
+
+static int esw_vport_tbl_get(struct mlx5_eswitch *esw)
+{
+	struct mlx5_vport_tbl_attr attr;
+	struct mlx5_flow_table *fdb;
+	struct mlx5_vport *vport;
+	unsigned long i;
+
+	attr.chain = 0;
+	attr.prio = 1;
+	mlx5_esw_for_each_vport(esw, i, vport) {
+		attr.vport = vport->vport;
+		attr.vport_ns = &mlx5_esw_vport_tbl_mirror_ns;
+		fdb = mlx5_esw_vporttbl_get(esw, &attr);
+		if (IS_ERR(fdb))
+			goto out;
+	}
+	return 0;
+
+out:
+	esw_vport_tbl_put(esw);
+	return PTR_ERR(fdb);
+}
+
+#define fdb_modify_header_fwd_to_table_supported(esw) \
+	(MLX5_CAP_ESW_FLOWTABLE((esw)->dev, fdb_modify_header_fwd_to_table))
+static void esw_init_chains_offload_flags(struct mlx5_eswitch *esw, u32 *flags)
+{
+	struct mlx5_core_dev *dev = esw->dev;
+
+	if (MLX5_CAP_ESW_FLOWTABLE_FDB(dev, ignore_flow_level))
+		*flags |= MLX5_CHAINS_IGNORE_FLOW_LEVEL_SUPPORTED;
+
+	if (!MLX5_CAP_ESW_FLOWTABLE(dev, multi_fdb_encap) &&
+	    esw->offloads.encap != DEVLINK_ESWITCH_ENCAP_MODE_NONE) {
+		*flags &= ~MLX5_CHAINS_AND_PRIOS_SUPPORTED;
+		esw_warn(dev, "Tc chains and priorities offload aren't supported, update firmware if needed\n");
+	} else if (!mlx5_eswitch_reg_c1_loopback_enabled(esw)) {
+		*flags &= ~MLX5_CHAINS_AND_PRIOS_SUPPORTED;
+		esw_warn(dev, "Tc chains and priorities offload aren't supported\n");
+	} else if (!fdb_modify_header_fwd_to_table_supported(esw)) {
+		/* Disabled when ttl workaround is needed, e.g
+		 * when ESWITCH_IPV4_TTL_MODIFY_ENABLE = true in mlxconfig
+		 */
+		esw_warn(dev,
+			 "Tc chains and priorities offload aren't supported, check firmware version, or mlxconfig settings\n");
+		*flags &= ~MLX5_CHAINS_AND_PRIOS_SUPPORTED;
+	} else {
+		*flags |= MLX5_CHAINS_AND_PRIOS_SUPPORTED;
+		esw_info(dev, "Supported tc chains and prios offload\n");
+	}
+
+	if (esw->offloads.encap != DEVLINK_ESWITCH_ENCAP_MODE_NONE)
+		*flags |= MLX5_CHAINS_FT_TUNNEL_SUPPORTED;
+}
+
+static int
+esw_chains_create(struct mlx5_eswitch *esw, struct mlx5_flow_table *miss_fdb)
+{
+	struct mlx5_core_dev *dev = esw->dev;
+	struct mlx5_flow_table *nf_ft, *ft;
+	struct mlx5_chains_attr attr = {};
+	struct mlx5_fs_chains *chains;
+	u32 fdb_max;
+	int err;
+
+	fdb_max = 1 << MLX5_CAP_ESW_FLOWTABLE_FDB(dev, log_max_ft_size);
+
+	esw_init_chains_offload_flags(esw, &attr.flags);
+	attr.ns = MLX5_FLOW_NAMESPACE_FDB;
+	attr.max_ft_sz = fdb_max;
+	attr.max_grp_num = esw->params.large_group_num;
+	attr.default_ft = miss_fdb;
+	attr.mapping = esw->offloads.reg_c0_obj_pool;
+
+	chains = mlx5_chains_create(dev, &attr);
+	if (IS_ERR(chains)) {
+		err = PTR_ERR(chains);
+		esw_warn(dev, "Failed to create fdb chains err(%d)\n", err);
+		return err;
+	}
+
+	esw->fdb_table.offloads.esw_chains_priv = chains;
+
+	/* Create tc_end_ft which is the always created ft chain */
+	nf_ft = mlx5_chains_get_table(chains, mlx5_chains_get_nf_ft_chain(chains),
+				      1, 0);
+	if (IS_ERR(nf_ft)) {
+		err = PTR_ERR(nf_ft);
+		goto nf_ft_err;
+	}
+
+	/* Always open the root for fast path */
+	ft = mlx5_chains_get_table(chains, 0, 1, 0);
+	if (IS_ERR(ft)) {
+		err = PTR_ERR(ft);
+		goto level_0_err;
+	}
+
+	/* Open level 1 for split fdb rules now if prios isn't supported  */
+	if (!mlx5_chains_prios_supported(chains)) {
+		err = esw_vport_tbl_get(esw);
+		if (err)
+			goto level_1_err;
+	}
+
+	mlx5_chains_set_end_ft(chains, nf_ft);
+
+	return 0;
+
+level_1_err:
+	mlx5_chains_put_table(chains, 0, 1, 0);
+level_0_err:
+	mlx5_chains_put_table(chains, mlx5_chains_get_nf_ft_chain(chains), 1, 0);
+nf_ft_err:
+	mlx5_chains_destroy(chains);
+	esw->fdb_table.offloads.esw_chains_priv = NULL;
+
+	return err;
+}
+
+static void
+esw_chains_destroy(struct mlx5_eswitch *esw, struct mlx5_fs_chains *chains)
+{
+	if (!mlx5_chains_prios_supported(chains))
+		esw_vport_tbl_put(esw);
+	mlx5_chains_put_table(chains, 0, 1, 0);
+	mlx5_chains_put_table(chains, mlx5_chains_get_nf_ft_chain(chains), 1, 0);
+	mlx5_chains_destroy(chains);
+}
+
+#else /* CONFIG_MLX5_CLS_ACT */
+
+static int
+esw_chains_create(struct mlx5_eswitch *esw, struct mlx5_flow_table *miss_fdb)
+{ return 0; }
+
+static void
+esw_chains_destroy(struct mlx5_eswitch *esw, struct mlx5_fs_chains *chains)
+{}
+
+#endif
+
+static int
+esw_create_send_to_vport_group(struct mlx5_eswitch *esw,
+			       struct mlx5_flow_table *fdb,
+			       u32 *flow_group_in,
+			       int *ix)
+{
+	int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
+	struct mlx5_flow_group *g;
+	void *match_criteria;
+	int count, err = 0;
+
+	memset(flow_group_in, 0, inlen);
+
+	MLX5_SET(create_flow_group_in, flow_group_in, match_criteria_enable,
+		 MLX5_MATCH_MISC_PARAMETERS);
+
+	match_criteria = MLX5_ADDR_OF(create_flow_group_in, flow_group_in, match_criteria);
+
+	MLX5_SET_TO_ONES(fte_match_param, match_criteria, misc_parameters.source_sqn);
+	MLX5_SET_TO_ONES(fte_match_param, match_criteria, misc_parameters.source_port);
+	if (MLX5_CAP_ESW(esw->dev, merged_eswitch)) {
+		MLX5_SET_TO_ONES(fte_match_param, match_criteria,
+				 misc_parameters.source_eswitch_owner_vhca_id);
+		MLX5_SET(create_flow_group_in, flow_group_in,
+			 source_eswitch_owner_vhca_id_valid, 1);
+	}
+
+	/* See comment at table_size calculation */
+	count = MLX5_MAX_PORTS * (esw->total_vports * MAX_SQ_NVPORTS + MAX_PF_SQ);
+	MLX5_SET(create_flow_group_in, flow_group_in, start_flow_index, 0);
+	MLX5_SET(create_flow_group_in, flow_group_in, end_flow_index, *ix + count - 1);
+	*ix += count;
+
+	g = mlx5_create_flow_group(fdb, flow_group_in);
+	if (IS_ERR(g)) {
+		err = PTR_ERR(g);
+		esw_warn(esw->dev, "Failed to create send-to-vport flow group err(%d)\n", err);
+		goto out;
+	}
+	esw->fdb_table.offloads.send_to_vport_grp = g;
+
+out:
+	return err;
+}
+
+static int
+esw_create_meta_send_to_vport_group(struct mlx5_eswitch *esw,
+				    struct mlx5_flow_table *fdb,
+				    u32 *flow_group_in,
+				    int *ix)
+{
+	int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
+	struct mlx5_flow_group *g;
+	void *match_criteria;
+	int err = 0;
+
+	if (!esw_src_port_rewrite_supported(esw))
+		return 0;
+
+	memset(flow_group_in, 0, inlen);
+
+	MLX5_SET(create_flow_group_in, flow_group_in, match_criteria_enable,
+		 MLX5_MATCH_MISC_PARAMETERS_2);
+
+	match_criteria = MLX5_ADDR_OF(create_flow_group_in, flow_group_in, match_criteria);
+
+	MLX5_SET(fte_match_param, match_criteria,
+		 misc_parameters_2.metadata_reg_c_0,
+		 mlx5_eswitch_get_vport_metadata_mask());
+	MLX5_SET(fte_match_param, match_criteria,
+		 misc_parameters_2.metadata_reg_c_1, ESW_TUN_MASK);
+
+	MLX5_SET(create_flow_group_in, flow_group_in, start_flow_index, *ix);
+	MLX5_SET(create_flow_group_in, flow_group_in,
+		 end_flow_index, *ix + esw->total_vports - 1);
+	*ix += esw->total_vports;
+
+	g = mlx5_create_flow_group(fdb, flow_group_in);
+	if (IS_ERR(g)) {
+		err = PTR_ERR(g);
+		esw_warn(esw->dev,
+			 "Failed to create send-to-vport meta flow group err(%d)\n", err);
+		goto send_vport_meta_err;
+	}
+	esw->fdb_table.offloads.send_to_vport_meta_grp = g;
+
+	return 0;
+
+send_vport_meta_err:
+	return err;
+}
+
+static int
+esw_create_peer_esw_miss_group(struct mlx5_eswitch *esw,
+			       struct mlx5_flow_table *fdb,
+			       u32 *flow_group_in,
+			       int *ix)
+{
+	int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
+	struct mlx5_flow_group *g;
+	void *match_criteria;
+	int err = 0;
+
+	if (!MLX5_CAP_ESW(esw->dev, merged_eswitch))
+		return 0;
+
+	memset(flow_group_in, 0, inlen);
+
+	esw_set_flow_group_source_port(esw, flow_group_in);
+
+	if (!mlx5_eswitch_vport_match_metadata_enabled(esw)) {
+		match_criteria = MLX5_ADDR_OF(create_flow_group_in,
+					      flow_group_in,
+					      match_criteria);
+
+		MLX5_SET_TO_ONES(fte_match_param, match_criteria,
+				 misc_parameters.source_eswitch_owner_vhca_id);
+
+		MLX5_SET(create_flow_group_in, flow_group_in,
+			 source_eswitch_owner_vhca_id_valid, 1);
+	}
+
+	MLX5_SET(create_flow_group_in, flow_group_in, start_flow_index, *ix);
+	MLX5_SET(create_flow_group_in, flow_group_in, end_flow_index,
+		 *ix + esw->total_vports - 1);
+	*ix += esw->total_vports;
+
+	g = mlx5_create_flow_group(fdb, flow_group_in);
+	if (IS_ERR(g)) {
+		err = PTR_ERR(g);
+		esw_warn(esw->dev, "Failed to create peer miss flow group err(%d)\n", err);
+		goto out;
+	}
+	esw->fdb_table.offloads.peer_miss_grp = g;
+
+out:
+	return err;
+}
+
+static int
+esw_create_miss_group(struct mlx5_eswitch *esw,
+		      struct mlx5_flow_table *fdb,
+		      u32 *flow_group_in,
+		      int *ix)
+{
+	int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
+	struct mlx5_flow_group *g;
+	void *match_criteria;
+	int err = 0;
+	u8 *dmac;
+
+	memset(flow_group_in, 0, inlen);
+
+	MLX5_SET(create_flow_group_in, flow_group_in, match_criteria_enable,
+		 MLX5_MATCH_OUTER_HEADERS);
+	match_criteria = MLX5_ADDR_OF(create_flow_group_in, flow_group_in,
+				      match_criteria);
+	dmac = MLX5_ADDR_OF(fte_match_param, match_criteria,
+			    outer_headers.dmac_47_16);
+	dmac[0] = 0x01;
+
+	MLX5_SET(create_flow_group_in, flow_group_in, start_flow_index, *ix);
+	MLX5_SET(create_flow_group_in, flow_group_in, end_flow_index,
+		 *ix + MLX5_ESW_MISS_FLOWS);
+
+	g = mlx5_create_flow_group(fdb, flow_group_in);
+	if (IS_ERR(g)) {
+		err = PTR_ERR(g);
+		esw_warn(esw->dev, "Failed to create miss flow group err(%d)\n", err);
+		goto miss_err;
+	}
+	esw->fdb_table.offloads.miss_grp = g;
+
+	err = esw_add_fdb_miss_rule(esw);
+	if (err)
+		goto miss_rule_err;
+
+	return 0;
+
+miss_rule_err:
+	mlx5_destroy_flow_group(esw->fdb_table.offloads.miss_grp);
+miss_err:
+	return err;
+}
+
+static int esw_create_offloads_fdb_tables(struct mlx5_eswitch *esw)
+{
+	int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
+	struct mlx5_flow_table_attr ft_attr = {};
+	struct mlx5_core_dev *dev = esw->dev;
+	struct mlx5_flow_namespace *root_ns;
+	struct mlx5_flow_table *fdb = NULL;
+	int table_size, ix = 0, err = 0;
+	u32 flags = 0, *flow_group_in;
+
+	esw_debug(esw->dev, "Create offloads FDB Tables\n");
+
+	flow_group_in = kvzalloc(inlen, GFP_KERNEL);
+	if (!flow_group_in)
+		return -ENOMEM;
+
+	root_ns = mlx5_get_flow_namespace(dev, MLX5_FLOW_NAMESPACE_FDB);
+	if (!root_ns) {
+		esw_warn(dev, "Failed to get FDB flow namespace\n");
+		err = -EOPNOTSUPP;
+		goto ns_err;
+	}
+	esw->fdb_table.offloads.ns = root_ns;
+	err = mlx5_flow_namespace_set_mode(root_ns,
+					   esw->dev->priv.steering->mode);
+	if (err) {
+		esw_warn(dev, "Failed to set FDB namespace steering mode\n");
+		goto ns_err;
+	}
+
+	/* To be strictly correct:
+	 *	MLX5_MAX_PORTS * (esw->total_vports * MAX_SQ_NVPORTS + MAX_PF_SQ)
+	 * should be:
+	 *	esw->total_vports * MAX_SQ_NVPORTS + MAX_PF_SQ +
+	 *	peer_esw->total_vports * MAX_SQ_NVPORTS + MAX_PF_SQ
+	 * but as the peer device might not be in switchdev mode it's not
+	 * possible. We use the fact that by default FW sets max vfs and max sfs
+	 * to the same value on both devices. If it needs to be changed in the future note
+	 * the peer miss group should also be created based on the number of
+	 * total vports of the peer (currently is also uses esw->total_vports).
+	 */
+	table_size = MLX5_MAX_PORTS * (esw->total_vports * MAX_SQ_NVPORTS + MAX_PF_SQ) +
+		     esw->total_vports * 2 + MLX5_ESW_MISS_FLOWS;
+
+	/* create the slow path fdb with encap set, so further table instances
+	 * can be created at run time while VFs are probed if the FW allows that.
+	 */
+	if (esw->offloads.encap != DEVLINK_ESWITCH_ENCAP_MODE_NONE)
+		flags |= (MLX5_FLOW_TABLE_TUNNEL_EN_REFORMAT |
+			  MLX5_FLOW_TABLE_TUNNEL_EN_DECAP);
+
+	ft_attr.flags = flags;
+	ft_attr.max_fte = table_size;
+	ft_attr.prio = FDB_SLOW_PATH;
+
+	fdb = mlx5_create_flow_table(root_ns, &ft_attr);
+	if (IS_ERR(fdb)) {
+		err = PTR_ERR(fdb);
+		esw_warn(dev, "Failed to create slow path FDB Table err %d\n", err);
+		goto slow_fdb_err;
+	}
+	esw->fdb_table.offloads.slow_fdb = fdb;
+
+	/* Create empty TC-miss managed table. This allows plugging in following
+	 * priorities without directly exposing their level 0 table to
+	 * eswitch_offloads and passing it as miss_fdb to following call to
+	 * esw_chains_create().
+	 */
+	memset(&ft_attr, 0, sizeof(ft_attr));
+	ft_attr.prio = FDB_TC_MISS;
+	esw->fdb_table.offloads.tc_miss_table = mlx5_create_flow_table(root_ns, &ft_attr);
+	if (IS_ERR(esw->fdb_table.offloads.tc_miss_table)) {
+		err = PTR_ERR(esw->fdb_table.offloads.tc_miss_table);
+		esw_warn(dev, "Failed to create TC miss FDB Table err %d\n", err);
+		goto tc_miss_table_err;
+	}
+
+	err = esw_chains_create(esw, esw->fdb_table.offloads.tc_miss_table);
+	if (err) {
+		esw_warn(dev, "Failed to open fdb chains err(%d)\n", err);
+		goto fdb_chains_err;
+	}
+
+	err = esw_create_send_to_vport_group(esw, fdb, flow_group_in, &ix);
+	if (err)
+		goto send_vport_err;
+
+	err = esw_create_meta_send_to_vport_group(esw, fdb, flow_group_in, &ix);
+	if (err)
+		goto send_vport_meta_err;
+
+	err = esw_create_peer_esw_miss_group(esw, fdb, flow_group_in, &ix);
+	if (err)
+		goto peer_miss_err;
+
+	err = esw_create_miss_group(esw, fdb, flow_group_in, &ix);
+	if (err)
+		goto miss_err;
+
+	kvfree(flow_group_in);
+	return 0;
+
+miss_err:
+	if (MLX5_CAP_ESW(esw->dev, merged_eswitch))
+		mlx5_destroy_flow_group(esw->fdb_table.offloads.peer_miss_grp);
+peer_miss_err:
+	if (esw->fdb_table.offloads.send_to_vport_meta_grp)
+		mlx5_destroy_flow_group(esw->fdb_table.offloads.send_to_vport_meta_grp);
+send_vport_meta_err:
+	mlx5_destroy_flow_group(esw->fdb_table.offloads.send_to_vport_grp);
+send_vport_err:
+	esw_chains_destroy(esw, esw_chains(esw));
+fdb_chains_err:
+	mlx5_destroy_flow_table(esw->fdb_table.offloads.tc_miss_table);
+tc_miss_table_err:
+	mlx5_destroy_flow_table(esw->fdb_table.offloads.slow_fdb);
+slow_fdb_err:
+	/* Holds true only as long as DMFS is the default */
+	mlx5_flow_namespace_set_mode(root_ns, MLX5_FLOW_STEERING_MODE_DMFS);
+ns_err:
+	kvfree(flow_group_in);
+	return err;
+}
+
+static void esw_destroy_offloads_fdb_tables(struct mlx5_eswitch *esw)
+{
+	if (!esw->fdb_table.offloads.slow_fdb)
+		return;
+
+	esw_debug(esw->dev, "Destroy offloads FDB Tables\n");
+	mlx5_del_flow_rules(esw->fdb_table.offloads.miss_rule_multi);
+	mlx5_del_flow_rules(esw->fdb_table.offloads.miss_rule_uni);
+	mlx5_destroy_flow_group(esw->fdb_table.offloads.send_to_vport_grp);
+	if (esw->fdb_table.offloads.send_to_vport_meta_grp)
+		mlx5_destroy_flow_group(esw->fdb_table.offloads.send_to_vport_meta_grp);
+	if (MLX5_CAP_ESW(esw->dev, merged_eswitch))
+		mlx5_destroy_flow_group(esw->fdb_table.offloads.peer_miss_grp);
+	mlx5_destroy_flow_group(esw->fdb_table.offloads.miss_grp);
+
+	esw_chains_destroy(esw, esw_chains(esw));
+
+	mlx5_destroy_flow_table(esw->fdb_table.offloads.tc_miss_table);
+	mlx5_destroy_flow_table(esw->fdb_table.offloads.slow_fdb);
+	/* Holds true only as long as DMFS is the default */
+	mlx5_flow_namespace_set_mode(esw->fdb_table.offloads.ns,
+				     MLX5_FLOW_STEERING_MODE_DMFS);
+	atomic64_set(&esw->user_count, 0);
+}
+
+static int esw_get_nr_ft_offloads_steering_src_ports(struct mlx5_eswitch *esw)
+{
+	int nvports;
+
+	nvports = esw->total_vports + MLX5_ESW_MISS_FLOWS;
+	if (mlx5e_tc_int_port_supported(esw))
+		nvports += MLX5E_TC_MAX_INT_PORT_NUM;
+
+	return nvports;
+}
+
+static int esw_create_offloads_table(struct mlx5_eswitch *esw)
+{
+	struct mlx5_flow_table_attr ft_attr = {};
+	struct mlx5_core_dev *dev = esw->dev;
+	struct mlx5_flow_table *ft_offloads;
+	struct mlx5_flow_namespace *ns;
+	int err = 0;
+
+	ns = mlx5_get_flow_namespace(dev, MLX5_FLOW_NAMESPACE_OFFLOADS);
+	if (!ns) {
+		esw_warn(esw->dev, "Failed to get offloads flow namespace\n");
+		return -EOPNOTSUPP;
+	}
+
+	ft_attr.max_fte = esw_get_nr_ft_offloads_steering_src_ports(esw) +
+			  MLX5_ESW_FT_OFFLOADS_DROP_RULE;
+	ft_attr.prio = 1;
+
+	ft_offloads = mlx5_create_flow_table(ns, &ft_attr);
+	if (IS_ERR(ft_offloads)) {
+		err = PTR_ERR(ft_offloads);
+		esw_warn(esw->dev, "Failed to create offloads table, err %d\n", err);
+		return err;
+	}
+
+	esw->offloads.ft_offloads = ft_offloads;
+	return 0;
+}
+
+static void esw_destroy_offloads_table(struct mlx5_eswitch *esw)
+{
+	struct mlx5_esw_offload *offloads = &esw->offloads;
+
+	mlx5_destroy_flow_table(offloads->ft_offloads);
+}
+
+static int esw_create_vport_rx_group(struct mlx5_eswitch *esw)
+{
+	int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
+	struct mlx5_flow_group *g;
+	u32 *flow_group_in;
+	int nvports;
+	int err = 0;
+
+	nvports = esw_get_nr_ft_offloads_steering_src_ports(esw);
+	flow_group_in = kvzalloc(inlen, GFP_KERNEL);
+	if (!flow_group_in)
+		return -ENOMEM;
+
+	/* create vport rx group */
+	esw_set_flow_group_source_port(esw, flow_group_in);
+
+	MLX5_SET(create_flow_group_in, flow_group_in, start_flow_index, 0);
+	MLX5_SET(create_flow_group_in, flow_group_in, end_flow_index, nvports - 1);
+
+	g = mlx5_create_flow_group(esw->offloads.ft_offloads, flow_group_in);
+
+	if (IS_ERR(g)) {
+		err = PTR_ERR(g);
+		mlx5_core_warn(esw->dev, "Failed to create vport rx group err %d\n", err);
+		goto out;
+	}
+
+	esw->offloads.vport_rx_group = g;
+out:
+	kvfree(flow_group_in);
+	return err;
+}
+
+static void esw_destroy_vport_rx_group(struct mlx5_eswitch *esw)
+{
+	mlx5_destroy_flow_group(esw->offloads.vport_rx_group);
+}
+
+static int esw_create_vport_rx_drop_rule_index(struct mlx5_eswitch *esw)
+{
+	/* ft_offloads table is enlarged by MLX5_ESW_FT_OFFLOADS_DROP_RULE (1)
+	 * for the drop rule, which is placed at the end of the table.
+	 * So return the total of vport and int_port as rule index.
+	 */
+	return esw_get_nr_ft_offloads_steering_src_ports(esw);
+}
+
+static int esw_create_vport_rx_drop_group(struct mlx5_eswitch *esw)
+{
+	int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
+	struct mlx5_flow_group *g;
+	u32 *flow_group_in;
+	int flow_index;
+	int err = 0;
+
+	flow_index = esw_create_vport_rx_drop_rule_index(esw);
+
+	flow_group_in = kvzalloc(inlen, GFP_KERNEL);
+	if (!flow_group_in)
+		return -ENOMEM;
+
+	MLX5_SET(create_flow_group_in, flow_group_in, start_flow_index, flow_index);
+	MLX5_SET(create_flow_group_in, flow_group_in, end_flow_index, flow_index);
+
+	g = mlx5_create_flow_group(esw->offloads.ft_offloads, flow_group_in);
+
+	if (IS_ERR(g)) {
+		err = PTR_ERR(g);
+		mlx5_core_warn(esw->dev, "Failed to create vport rx drop group err %d\n", err);
+		goto out;
+	}
+
+	esw->offloads.vport_rx_drop_group = g;
+out:
+	kvfree(flow_group_in);
+	return err;
+}
+
+static void esw_destroy_vport_rx_drop_group(struct mlx5_eswitch *esw)
+{
+	if (esw->offloads.vport_rx_drop_group)
+		mlx5_destroy_flow_group(esw->offloads.vport_rx_drop_group);
+}
+
+struct mlx5_flow_handle *
+mlx5_eswitch_create_vport_rx_rule(struct mlx5_eswitch *esw, u16 vport,
+				  struct mlx5_flow_destination *dest)
+{
+	struct mlx5_flow_act flow_act = {0};
+	struct mlx5_flow_handle *flow_rule;
+	struct mlx5_flow_spec *spec;
+	void *misc;
+
+	spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
+	if (!spec) {
+		flow_rule = ERR_PTR(-ENOMEM);
+		goto out;
+	}
+
+	if (mlx5_eswitch_vport_match_metadata_enabled(esw)) {
+		misc = MLX5_ADDR_OF(fte_match_param, spec->match_value, misc_parameters_2);
+		MLX5_SET(fte_match_set_misc2, misc, metadata_reg_c_0,
+			 mlx5_eswitch_get_vport_metadata_for_match(esw, vport));
+
+		misc = MLX5_ADDR_OF(fte_match_param, spec->match_criteria, misc_parameters_2);
+		MLX5_SET(fte_match_set_misc2, misc, metadata_reg_c_0,
+			 mlx5_eswitch_get_vport_metadata_mask());
+
+		spec->match_criteria_enable = MLX5_MATCH_MISC_PARAMETERS_2;
+	} else {
+		misc = MLX5_ADDR_OF(fte_match_param, spec->match_value, misc_parameters);
+		MLX5_SET(fte_match_set_misc, misc, source_port, vport);
+
+		misc = MLX5_ADDR_OF(fte_match_param, spec->match_criteria, misc_parameters);
+		MLX5_SET_TO_ONES(fte_match_set_misc, misc, source_port);
+
+		spec->match_criteria_enable = MLX5_MATCH_MISC_PARAMETERS;
+	}
+
+	flow_act.action = MLX5_FLOW_CONTEXT_ACTION_FWD_DEST;
+	flow_rule = mlx5_add_flow_rules(esw->offloads.ft_offloads, spec,
+					&flow_act, dest, 1);
+	if (IS_ERR(flow_rule)) {
+		esw_warn(esw->dev, "fs offloads: Failed to add vport rx rule err %ld\n", PTR_ERR(flow_rule));
+		goto out;
+	}
+
+out:
+	kvfree(spec);
+	return flow_rule;
+}
+
+static int esw_create_vport_rx_drop_rule(struct mlx5_eswitch *esw)
+{
+	struct mlx5_flow_act flow_act = {};
+	struct mlx5_flow_handle *flow_rule;
+
+	flow_act.action = MLX5_FLOW_CONTEXT_ACTION_DROP;
+	flow_rule = mlx5_add_flow_rules(esw->offloads.ft_offloads, NULL,
+					&flow_act, NULL, 0);
+	if (IS_ERR(flow_rule)) {
+		esw_warn(esw->dev,
+			 "fs offloads: Failed to add vport rx drop rule err %ld\n",
+			 PTR_ERR(flow_rule));
+		return PTR_ERR(flow_rule);
+	}
+
+	esw->offloads.vport_rx_drop_rule = flow_rule;
+
+	return 0;
+}
+
+static void esw_destroy_vport_rx_drop_rule(struct mlx5_eswitch *esw)
+{
+	if (esw->offloads.vport_rx_drop_rule)
+		mlx5_del_flow_rules(esw->offloads.vport_rx_drop_rule);
+}
+
+static int mlx5_eswitch_inline_mode_get(struct mlx5_eswitch *esw, u8 *mode)
+{
+	u8 prev_mlx5_mode, mlx5_mode = MLX5_INLINE_MODE_L2;
+	struct mlx5_core_dev *dev = esw->dev;
+	struct mlx5_vport *vport;
+	unsigned long i;
+
+	if (!MLX5_CAP_GEN(dev, vport_group_manager))
+		return -EOPNOTSUPP;
+
+	if (!mlx5_esw_is_fdb_created(esw))
+		return -EOPNOTSUPP;
+
+	switch (MLX5_CAP_ETH(dev, wqe_inline_mode)) {
+	case MLX5_CAP_INLINE_MODE_NOT_REQUIRED:
+		mlx5_mode = MLX5_INLINE_MODE_NONE;
+		goto out;
+	case MLX5_CAP_INLINE_MODE_L2:
+		mlx5_mode = MLX5_INLINE_MODE_L2;
+		goto out;
+	case MLX5_CAP_INLINE_MODE_VPORT_CONTEXT:
+		goto query_vports;
+	}
+
+query_vports:
+	mlx5_query_nic_vport_min_inline(dev, esw->first_host_vport, &prev_mlx5_mode);
+	mlx5_esw_for_each_host_func_vport(esw, i, vport, esw->esw_funcs.num_vfs) {
+		mlx5_query_nic_vport_min_inline(dev, vport->vport, &mlx5_mode);
+		if (prev_mlx5_mode != mlx5_mode)
+			return -EINVAL;
+		prev_mlx5_mode = mlx5_mode;
+	}
+
+out:
+	*mode = mlx5_mode;
+	return 0;
+}
+
+static void esw_destroy_restore_table(struct mlx5_eswitch *esw)
+{
+	struct mlx5_esw_offload *offloads = &esw->offloads;
+
+	if (!mlx5_eswitch_reg_c1_loopback_supported(esw))
+		return;
+
+	mlx5_modify_header_dealloc(esw->dev, offloads->restore_copy_hdr_id);
+	mlx5_destroy_flow_group(offloads->restore_group);
+	mlx5_destroy_flow_table(offloads->ft_offloads_restore);
+}
+
+static int esw_create_restore_table(struct mlx5_eswitch *esw)
+{
+	u8 modact[MLX5_UN_SZ_BYTES(set_add_copy_action_in_auto)] = {};
+	int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
+	struct mlx5_flow_table_attr ft_attr = {};
+	struct mlx5_core_dev *dev = esw->dev;
+	struct mlx5_flow_namespace *ns;
+	struct mlx5_modify_hdr *mod_hdr;
+	void *match_criteria, *misc;
+	struct mlx5_flow_table *ft;
+	struct mlx5_flow_group *g;
+	u32 *flow_group_in;
+	int err = 0;
+
+	if (!mlx5_eswitch_reg_c1_loopback_supported(esw))
+		return 0;
+
+	ns = mlx5_get_flow_namespace(dev, MLX5_FLOW_NAMESPACE_OFFLOADS);
+	if (!ns) {
+		esw_warn(esw->dev, "Failed to get offloads flow namespace\n");
+		return -EOPNOTSUPP;
+	}
+
+	flow_group_in = kvzalloc(inlen, GFP_KERNEL);
+	if (!flow_group_in) {
+		err = -ENOMEM;
+		goto out_free;
+	}
+
+	ft_attr.max_fte = 1 << ESW_REG_C0_USER_DATA_METADATA_BITS;
+	ft = mlx5_create_flow_table(ns, &ft_attr);
+	if (IS_ERR(ft)) {
+		err = PTR_ERR(ft);
+		esw_warn(esw->dev, "Failed to create restore table, err %d\n",
+			 err);
+		goto out_free;
+	}
+
+	match_criteria = MLX5_ADDR_OF(create_flow_group_in, flow_group_in,
+				      match_criteria);
+	misc = MLX5_ADDR_OF(fte_match_param, match_criteria,
+			    misc_parameters_2);
+
+	MLX5_SET(fte_match_set_misc2, misc, metadata_reg_c_0,
+		 ESW_REG_C0_USER_DATA_METADATA_MASK);
+	MLX5_SET(create_flow_group_in, flow_group_in, start_flow_index, 0);
+	MLX5_SET(create_flow_group_in, flow_group_in, end_flow_index,
+		 ft_attr.max_fte - 1);
+	MLX5_SET(create_flow_group_in, flow_group_in, match_criteria_enable,
+		 MLX5_MATCH_MISC_PARAMETERS_2);
+	g = mlx5_create_flow_group(ft, flow_group_in);
+	if (IS_ERR(g)) {
+		err = PTR_ERR(g);
+		esw_warn(dev, "Failed to create restore flow group, err: %d\n",
+			 err);
+		goto err_group;
+	}
+
+	MLX5_SET(copy_action_in, modact, action_type, MLX5_ACTION_TYPE_COPY);
+	MLX5_SET(copy_action_in, modact, src_field,
+		 MLX5_ACTION_IN_FIELD_METADATA_REG_C_1);
+	MLX5_SET(copy_action_in, modact, dst_field,
+		 MLX5_ACTION_IN_FIELD_METADATA_REG_B);
+	mod_hdr = mlx5_modify_header_alloc(esw->dev,
+					   MLX5_FLOW_NAMESPACE_KERNEL, 1,
+					   modact);
+	if (IS_ERR(mod_hdr)) {
+		err = PTR_ERR(mod_hdr);
+		esw_warn(dev, "Failed to create restore mod header, err: %d\n",
+			 err);
+		goto err_mod_hdr;
+	}
+
+	esw->offloads.ft_offloads_restore = ft;
+	esw->offloads.restore_group = g;
+	esw->offloads.restore_copy_hdr_id = mod_hdr;
+
+	kvfree(flow_group_in);
+
+	return 0;
+
+err_mod_hdr:
+	mlx5_destroy_flow_group(g);
+err_group:
+	mlx5_destroy_flow_table(ft);
+out_free:
+	kvfree(flow_group_in);
+
+	return err;
+}
+
+static int esw_offloads_start(struct mlx5_eswitch *esw,
+			      struct netlink_ext_ack *extack)
+{
+	int err;
+
+	esw->mode = MLX5_ESWITCH_OFFLOADS;
+	err = mlx5_eswitch_enable_locked(esw, esw->dev->priv.sriov.num_vfs);
+	if (err) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "Failed setting eswitch to offloads");
+		esw->mode = MLX5_ESWITCH_LEGACY;
+		mlx5_rescan_drivers(esw->dev);
+	}
+	if (esw->offloads.inline_mode == MLX5_INLINE_MODE_NONE) {
+		if (mlx5_eswitch_inline_mode_get(esw,
+						 &esw->offloads.inline_mode)) {
+			esw->offloads.inline_mode = MLX5_INLINE_MODE_L2;
+			NL_SET_ERR_MSG_MOD(extack,
+					   "Inline mode is different between vports");
+		}
+	}
+	return err;
+}
+
+static void mlx5_esw_offloads_rep_mark_set(struct mlx5_eswitch *esw,
+					   struct mlx5_eswitch_rep *rep,
+					   xa_mark_t mark)
+{
+	bool mark_set;
+
+	/* Copy the mark from vport to its rep */
+	mark_set = xa_get_mark(&esw->vports, rep->vport, mark);
+	if (mark_set)
+		xa_set_mark(&esw->offloads.vport_reps, rep->vport, mark);
+}
+
+static int mlx5_esw_offloads_rep_init(struct mlx5_eswitch *esw, const struct mlx5_vport *vport)
+{
+	struct mlx5_eswitch_rep *rep;
+	int rep_type;
+	int err;
+
+	rep = kzalloc(sizeof(*rep), GFP_KERNEL);
+	if (!rep)
+		return -ENOMEM;
+
+	rep->vport = vport->vport;
+	rep->vport_index = vport->index;
+	for (rep_type = 0; rep_type < NUM_REP_TYPES; rep_type++)
+		atomic_set(&rep->rep_data[rep_type].state, REP_UNREGISTERED);
+
+	err = xa_insert(&esw->offloads.vport_reps, rep->vport, rep, GFP_KERNEL);
+	if (err)
+		goto insert_err;
+
+	mlx5_esw_offloads_rep_mark_set(esw, rep, MLX5_ESW_VPT_HOST_FN);
+	mlx5_esw_offloads_rep_mark_set(esw, rep, MLX5_ESW_VPT_VF);
+	mlx5_esw_offloads_rep_mark_set(esw, rep, MLX5_ESW_VPT_SF);
+	return 0;
+
+insert_err:
+	kfree(rep);
+	return err;
+}
+
+static void mlx5_esw_offloads_rep_cleanup(struct mlx5_eswitch *esw,
+					  struct mlx5_eswitch_rep *rep)
+{
+	xa_erase(&esw->offloads.vport_reps, rep->vport);
+	kfree(rep);
+}
+
+void esw_offloads_cleanup_reps(struct mlx5_eswitch *esw)
+{
+	struct mlx5_eswitch_rep *rep;
+	unsigned long i;
+
+	mlx5_esw_for_each_rep(esw, i, rep)
+		mlx5_esw_offloads_rep_cleanup(esw, rep);
+	xa_destroy(&esw->offloads.vport_reps);
+}
+
+int esw_offloads_init_reps(struct mlx5_eswitch *esw)
+{
+	struct mlx5_vport *vport;
+	unsigned long i;
+	int err;
+
+	xa_init(&esw->offloads.vport_reps);
+
+	mlx5_esw_for_each_vport(esw, i, vport) {
+		err = mlx5_esw_offloads_rep_init(esw, vport);
+		if (err)
+			goto err;
+	}
+	return 0;
+
+err:
+	esw_offloads_cleanup_reps(esw);
+	return err;
+}
+
+static void __esw_offloads_unload_rep(struct mlx5_eswitch *esw,
+				      struct mlx5_eswitch_rep *rep, u8 rep_type)
+{
+	if (atomic_cmpxchg(&rep->rep_data[rep_type].state,
+			   REP_LOADED, REP_REGISTERED) == REP_LOADED)
+		esw->offloads.rep_ops[rep_type]->unload(rep);
+}
+
+static void __unload_reps_sf_vport(struct mlx5_eswitch *esw, u8 rep_type)
+{
+	struct mlx5_eswitch_rep *rep;
+	unsigned long i;
+
+	mlx5_esw_for_each_sf_rep(esw, i, rep)
+		__esw_offloads_unload_rep(esw, rep, rep_type);
+}
+
+static void __unload_reps_all_vport(struct mlx5_eswitch *esw, u8 rep_type)
+{
+	struct mlx5_eswitch_rep *rep;
+	unsigned long i;
+
+	__unload_reps_sf_vport(esw, rep_type);
+
+	mlx5_esw_for_each_vf_rep(esw, i, rep)
+		__esw_offloads_unload_rep(esw, rep, rep_type);
+
+	if (mlx5_ecpf_vport_exists(esw->dev)) {
+		rep = mlx5_eswitch_get_rep(esw, MLX5_VPORT_ECPF);
+		__esw_offloads_unload_rep(esw, rep, rep_type);
+	}
+
+	if (mlx5_core_is_ecpf_esw_manager(esw->dev)) {
+		rep = mlx5_eswitch_get_rep(esw, MLX5_VPORT_PF);
+		__esw_offloads_unload_rep(esw, rep, rep_type);
+	}
+
+	rep = mlx5_eswitch_get_rep(esw, MLX5_VPORT_UPLINK);
+	__esw_offloads_unload_rep(esw, rep, rep_type);
+}
+
+int mlx5_esw_offloads_rep_load(struct mlx5_eswitch *esw, u16 vport_num)
+{
+	struct mlx5_eswitch_rep *rep;
+	int rep_type;
+	int err;
+
+	rep = mlx5_eswitch_get_rep(esw, vport_num);
+	for (rep_type = 0; rep_type < NUM_REP_TYPES; rep_type++)
+		if (atomic_cmpxchg(&rep->rep_data[rep_type].state,
+				   REP_REGISTERED, REP_LOADED) == REP_REGISTERED) {
+			err = esw->offloads.rep_ops[rep_type]->load(esw->dev, rep);
+			if (err)
+				goto err_reps;
+		}
+
+	return 0;
+
+err_reps:
+	atomic_set(&rep->rep_data[rep_type].state, REP_REGISTERED);
+	for (--rep_type; rep_type >= 0; rep_type--)
+		__esw_offloads_unload_rep(esw, rep, rep_type);
+	return err;
+}
+
+void mlx5_esw_offloads_rep_unload(struct mlx5_eswitch *esw, u16 vport_num)
+{
+	struct mlx5_eswitch_rep *rep;
+	int rep_type;
+
+	rep = mlx5_eswitch_get_rep(esw, vport_num);
+	for (rep_type = NUM_REP_TYPES - 1; rep_type >= 0; rep_type--)
+		__esw_offloads_unload_rep(esw, rep, rep_type);
+}
+
+int esw_offloads_load_rep(struct mlx5_eswitch *esw, u16 vport_num)
+{
+	int err;
+
+	if (esw->mode != MLX5_ESWITCH_OFFLOADS)
+		return 0;
+
+	if (vport_num != MLX5_VPORT_UPLINK) {
+		err = mlx5_esw_offloads_devlink_port_register(esw, vport_num);
+		if (err)
+			return err;
+	}
+
+	err = mlx5_esw_offloads_rep_load(esw, vport_num);
+	if (err)
+		goto load_err;
+	return err;
+
+load_err:
+	if (vport_num != MLX5_VPORT_UPLINK)
+		mlx5_esw_offloads_devlink_port_unregister(esw, vport_num);
+	return err;
+}
+
+void esw_offloads_unload_rep(struct mlx5_eswitch *esw, u16 vport_num)
+{
+	if (esw->mode != MLX5_ESWITCH_OFFLOADS)
+		return;
+
+	mlx5_esw_offloads_rep_unload(esw, vport_num);
+
+	if (vport_num != MLX5_VPORT_UPLINK)
+		mlx5_esw_offloads_devlink_port_unregister(esw, vport_num);
+}
+
+static int esw_set_slave_root_fdb(struct mlx5_core_dev *master,
+				  struct mlx5_core_dev *slave)
+{
+	u32 in[MLX5_ST_SZ_DW(set_flow_table_root_in)]   = {};
+	u32 out[MLX5_ST_SZ_DW(set_flow_table_root_out)] = {};
+	struct mlx5_flow_root_namespace *root;
+	struct mlx5_flow_namespace *ns;
+	int err;
+
+	MLX5_SET(set_flow_table_root_in, in, opcode,
+		 MLX5_CMD_OP_SET_FLOW_TABLE_ROOT);
+	MLX5_SET(set_flow_table_root_in, in, table_type,
+		 FS_FT_FDB);
+
+	if (master) {
+		ns = mlx5_get_flow_namespace(master,
+					     MLX5_FLOW_NAMESPACE_FDB);
+		root = find_root(&ns->node);
+		mutex_lock(&root->chain_lock);
+		MLX5_SET(set_flow_table_root_in, in,
+			 table_eswitch_owner_vhca_id_valid, 1);
+		MLX5_SET(set_flow_table_root_in, in,
+			 table_eswitch_owner_vhca_id,
+			 MLX5_CAP_GEN(master, vhca_id));
+		MLX5_SET(set_flow_table_root_in, in, table_id,
+			 root->root_ft->id);
+	} else {
+		ns = mlx5_get_flow_namespace(slave,
+					     MLX5_FLOW_NAMESPACE_FDB);
+		root = find_root(&ns->node);
+		mutex_lock(&root->chain_lock);
+		MLX5_SET(set_flow_table_root_in, in, table_id,
+			 root->root_ft->id);
+	}
+
+	err = mlx5_cmd_exec(slave, in, sizeof(in), out, sizeof(out));
+	mutex_unlock(&root->chain_lock);
+
+	return err;
+}
+
+static int __esw_set_master_egress_rule(struct mlx5_core_dev *master,
+					struct mlx5_core_dev *slave,
+					struct mlx5_vport *vport,
+					struct mlx5_flow_table *acl)
+{
+	struct mlx5_flow_handle *flow_rule = NULL;
+	struct mlx5_flow_destination dest = {};
+	struct mlx5_flow_act flow_act = {};
+	struct mlx5_flow_spec *spec;
+	int err = 0;
+	void *misc;
+
+	spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
+	if (!spec)
+		return -ENOMEM;
+
+	spec->match_criteria_enable = MLX5_MATCH_MISC_PARAMETERS;
+	misc = MLX5_ADDR_OF(fte_match_param, spec->match_value,
+			    misc_parameters);
+	MLX5_SET(fte_match_set_misc, misc, source_port, MLX5_VPORT_UPLINK);
+	MLX5_SET(fte_match_set_misc, misc, source_eswitch_owner_vhca_id,
+		 MLX5_CAP_GEN(slave, vhca_id));
+
+	misc = MLX5_ADDR_OF(fte_match_param, spec->match_criteria, misc_parameters);
+	MLX5_SET_TO_ONES(fte_match_set_misc, misc, source_port);
+	MLX5_SET_TO_ONES(fte_match_set_misc, misc,
+			 source_eswitch_owner_vhca_id);
+
+	flow_act.action = MLX5_FLOW_CONTEXT_ACTION_FWD_DEST;
+	dest.type = MLX5_FLOW_DESTINATION_TYPE_VPORT;
+	dest.vport.num = slave->priv.eswitch->manager_vport;
+	dest.vport.vhca_id = MLX5_CAP_GEN(slave, vhca_id);
+	dest.vport.flags |= MLX5_FLOW_DEST_VPORT_VHCA_ID;
+
+	flow_rule = mlx5_add_flow_rules(acl, spec, &flow_act,
+					&dest, 1);
+	if (IS_ERR(flow_rule))
+		err = PTR_ERR(flow_rule);
+	else
+		vport->egress.offloads.bounce_rule = flow_rule;
+
+	kvfree(spec);
+	return err;
+}
+
+static int esw_set_master_egress_rule(struct mlx5_core_dev *master,
+				      struct mlx5_core_dev *slave)
+{
+	int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
+	struct mlx5_eswitch *esw = master->priv.eswitch;
+	struct mlx5_flow_table_attr ft_attr = {
+		.max_fte = 1, .prio = 0, .level = 0,
+		.flags = MLX5_FLOW_TABLE_OTHER_VPORT,
+	};
+	struct mlx5_flow_namespace *egress_ns;
+	struct mlx5_flow_table *acl;
+	struct mlx5_flow_group *g;
+	struct mlx5_vport *vport;
+	void *match_criteria;
+	u32 *flow_group_in;
+	int err;
+
+	vport = mlx5_eswitch_get_vport(esw, esw->manager_vport);
+	if (IS_ERR(vport))
+		return PTR_ERR(vport);
+
+	egress_ns = mlx5_get_flow_vport_acl_namespace(master,
+						      MLX5_FLOW_NAMESPACE_ESW_EGRESS,
+						      vport->index);
+	if (!egress_ns)
+		return -EINVAL;
+
+	if (vport->egress.acl)
+		return -EINVAL;
+
+	flow_group_in = kvzalloc(inlen, GFP_KERNEL);
+	if (!flow_group_in)
+		return -ENOMEM;
+
+	acl = mlx5_create_vport_flow_table(egress_ns, &ft_attr, vport->vport);
+	if (IS_ERR(acl)) {
+		err = PTR_ERR(acl);
+		goto out;
+	}
+
+	match_criteria = MLX5_ADDR_OF(create_flow_group_in, flow_group_in,
+				      match_criteria);
+	MLX5_SET_TO_ONES(fte_match_param, match_criteria,
+			 misc_parameters.source_port);
+	MLX5_SET_TO_ONES(fte_match_param, match_criteria,
+			 misc_parameters.source_eswitch_owner_vhca_id);
+	MLX5_SET(create_flow_group_in, flow_group_in, match_criteria_enable,
+		 MLX5_MATCH_MISC_PARAMETERS);
+
+	MLX5_SET(create_flow_group_in, flow_group_in,
+		 source_eswitch_owner_vhca_id_valid, 1);
+	MLX5_SET(create_flow_group_in, flow_group_in, start_flow_index, 0);
+	MLX5_SET(create_flow_group_in, flow_group_in, end_flow_index, 0);
+
+	g = mlx5_create_flow_group(acl, flow_group_in);
+	if (IS_ERR(g)) {
+		err = PTR_ERR(g);
+		goto err_group;
+	}
+
+	err = __esw_set_master_egress_rule(master, slave, vport, acl);
+	if (err)
+		goto err_rule;
+
+	vport->egress.acl = acl;
+	vport->egress.offloads.bounce_grp = g;
+
+	kvfree(flow_group_in);
+
+	return 0;
+
+err_rule:
+	mlx5_destroy_flow_group(g);
+err_group:
+	mlx5_destroy_flow_table(acl);
+out:
+	kvfree(flow_group_in);
+	return err;
+}
+
+static void esw_unset_master_egress_rule(struct mlx5_core_dev *dev)
+{
+	struct mlx5_vport *vport;
+
+	vport = mlx5_eswitch_get_vport(dev->priv.eswitch,
+				       dev->priv.eswitch->manager_vport);
+
+	esw_acl_egress_ofld_cleanup(vport);
+}
+
+int mlx5_eswitch_offloads_config_single_fdb(struct mlx5_eswitch *master_esw,
+					    struct mlx5_eswitch *slave_esw)
+{
+	int err;
+
+	err = esw_set_slave_root_fdb(master_esw->dev,
+				     slave_esw->dev);
+	if (err)
+		return err;
+
+	err = esw_set_master_egress_rule(master_esw->dev,
+					 slave_esw->dev);
+	if (err)
+		goto err_acl;
+
+	return err;
+
+err_acl:
+	esw_set_slave_root_fdb(NULL, slave_esw->dev);
+
+	return err;
+}
+
+void mlx5_eswitch_offloads_destroy_single_fdb(struct mlx5_eswitch *master_esw,
+					      struct mlx5_eswitch *slave_esw)
+{
+	esw_unset_master_egress_rule(master_esw->dev);
+	esw_set_slave_root_fdb(NULL, slave_esw->dev);
+}
+
+#define ESW_OFFLOADS_DEVCOM_PAIR	(0)
+#define ESW_OFFLOADS_DEVCOM_UNPAIR	(1)
+
+static void mlx5_esw_offloads_rep_event_unpair(struct mlx5_eswitch *esw)
+{
+	const struct mlx5_eswitch_rep_ops *ops;
+	struct mlx5_eswitch_rep *rep;
+	unsigned long i;
+	u8 rep_type;
+
+	mlx5_esw_for_each_rep(esw, i, rep) {
+		rep_type = NUM_REP_TYPES;
+		while (rep_type--) {
+			ops = esw->offloads.rep_ops[rep_type];
+			if (atomic_read(&rep->rep_data[rep_type].state) == REP_LOADED &&
+			    ops->event)
+				ops->event(esw, rep, MLX5_SWITCHDEV_EVENT_UNPAIR, NULL);
+		}
+	}
+}
+
+static void mlx5_esw_offloads_unpair(struct mlx5_eswitch *esw)
+{
+#if IS_ENABLED(CONFIG_MLX5_CLS_ACT)
+	mlx5e_tc_clean_fdb_peer_flows(esw);
+#endif
+	mlx5_esw_offloads_rep_event_unpair(esw);
+	esw_del_fdb_peer_miss_rules(esw);
+}
+
+static int mlx5_esw_offloads_pair(struct mlx5_eswitch *esw,
+				  struct mlx5_eswitch *peer_esw)
+{
+	const struct mlx5_eswitch_rep_ops *ops;
+	struct mlx5_eswitch_rep *rep;
+	unsigned long i;
+	u8 rep_type;
+	int err;
+
+	err = esw_add_fdb_peer_miss_rules(esw, peer_esw->dev);
+	if (err)
+		return err;
+
+	mlx5_esw_for_each_rep(esw, i, rep) {
+		for (rep_type = 0; rep_type < NUM_REP_TYPES; rep_type++) {
+			ops = esw->offloads.rep_ops[rep_type];
+			if (atomic_read(&rep->rep_data[rep_type].state) == REP_LOADED &&
+			    ops->event) {
+				err = ops->event(esw, rep, MLX5_SWITCHDEV_EVENT_PAIR, peer_esw);
+				if (err)
+					goto err_out;
+			}
+		}
+	}
+
+	return 0;
+
+err_out:
+	mlx5_esw_offloads_unpair(esw);
+	return err;
+}
+
+static int mlx5_esw_offloads_set_ns_peer(struct mlx5_eswitch *esw,
+					 struct mlx5_eswitch *peer_esw,
+					 bool pair)
+{
+	struct mlx5_flow_root_namespace *peer_ns;
+	struct mlx5_flow_root_namespace *ns;
+	int err;
+
+	peer_ns = peer_esw->dev->priv.steering->fdb_root_ns;
+	ns = esw->dev->priv.steering->fdb_root_ns;
+
+	if (pair) {
+		err = mlx5_flow_namespace_set_peer(ns, peer_ns);
+		if (err)
+			return err;
+
+		err = mlx5_flow_namespace_set_peer(peer_ns, ns);
+		if (err) {
+			mlx5_flow_namespace_set_peer(ns, NULL);
+			return err;
+		}
+	} else {
+		mlx5_flow_namespace_set_peer(ns, NULL);
+		mlx5_flow_namespace_set_peer(peer_ns, NULL);
+	}
+
+	return 0;
+}
+
+static int mlx5_esw_offloads_devcom_event(int event,
+					  void *my_data,
+					  void *event_data)
+{
+	struct mlx5_eswitch *esw = my_data;
+	struct mlx5_devcom *devcom = esw->dev->priv.devcom;
+	struct mlx5_eswitch *peer_esw = event_data;
+	int err;
+
+	switch (event) {
+	case ESW_OFFLOADS_DEVCOM_PAIR:
+		if (mlx5_eswitch_vport_match_metadata_enabled(esw) !=
+		    mlx5_eswitch_vport_match_metadata_enabled(peer_esw))
+			break;
+
+		if (esw->paired[mlx5_get_dev_index(peer_esw->dev)])
+			break;
+
+		err = mlx5_esw_offloads_set_ns_peer(esw, peer_esw, true);
+		if (err)
+			goto err_out;
+		err = mlx5_esw_offloads_pair(esw, peer_esw);
+		if (err)
+			goto err_peer;
+
+		err = mlx5_esw_offloads_pair(peer_esw, esw);
+		if (err)
+			goto err_pair;
+
+		esw->paired[mlx5_get_dev_index(peer_esw->dev)] = true;
+		peer_esw->paired[mlx5_get_dev_index(esw->dev)] = true;
+		mlx5_devcom_set_paired(devcom, MLX5_DEVCOM_ESW_OFFLOADS, true);
+		break;
+
+	case ESW_OFFLOADS_DEVCOM_UNPAIR:
+		if (!esw->paired[mlx5_get_dev_index(peer_esw->dev)])
+			break;
+
+		mlx5_devcom_set_paired(devcom, MLX5_DEVCOM_ESW_OFFLOADS, false);
+		esw->paired[mlx5_get_dev_index(peer_esw->dev)] = false;
+		peer_esw->paired[mlx5_get_dev_index(esw->dev)] = false;
+		mlx5_esw_offloads_unpair(peer_esw);
+		mlx5_esw_offloads_unpair(esw);
+		mlx5_esw_offloads_set_ns_peer(esw, peer_esw, false);
+		break;
+	}
+
+	return 0;
+
+err_pair:
+	mlx5_esw_offloads_unpair(esw);
+err_peer:
+	mlx5_esw_offloads_set_ns_peer(esw, peer_esw, false);
+err_out:
+	mlx5_core_err(esw->dev, "esw offloads devcom event failure, event %u err %d",
+		      event, err);
+	return err;
+}
+
+void mlx5_esw_offloads_devcom_init(struct mlx5_eswitch *esw)
+{
+	struct mlx5_devcom *devcom = esw->dev->priv.devcom;
+
+	INIT_LIST_HEAD(&esw->offloads.peer_flows);
+	mutex_init(&esw->offloads.peer_mutex);
+
+	if (!MLX5_CAP_ESW(esw->dev, merged_eswitch))
+		return;
+
+	if (!mlx5_is_lag_supported(esw->dev))
+		return;
+
+	mlx5_devcom_register_component(devcom,
+				       MLX5_DEVCOM_ESW_OFFLOADS,
+				       mlx5_esw_offloads_devcom_event,
+				       esw);
+
+	mlx5_devcom_send_event(devcom,
+			       MLX5_DEVCOM_ESW_OFFLOADS,
+			       ESW_OFFLOADS_DEVCOM_PAIR, esw);
+}
+
+void mlx5_esw_offloads_devcom_cleanup(struct mlx5_eswitch *esw)
+{
+	struct mlx5_devcom *devcom = esw->dev->priv.devcom;
+
+	if (!MLX5_CAP_ESW(esw->dev, merged_eswitch))
+		return;
+
+	if (!mlx5_is_lag_supported(esw->dev))
+		return;
+
+	mlx5_devcom_send_event(devcom, MLX5_DEVCOM_ESW_OFFLOADS,
+			       ESW_OFFLOADS_DEVCOM_UNPAIR, esw);
+
+	mlx5_devcom_unregister_component(devcom, MLX5_DEVCOM_ESW_OFFLOADS);
+}
+
+bool mlx5_esw_vport_match_metadata_supported(const struct mlx5_eswitch *esw)
+{
+	if (!MLX5_CAP_ESW(esw->dev, esw_uplink_ingress_acl))
+		return false;
+
+	if (!(MLX5_CAP_ESW_FLOWTABLE(esw->dev, fdb_to_vport_reg_c_id) &
+	      MLX5_FDB_TO_VPORT_REG_C_0))
+		return false;
+
+	if (!MLX5_CAP_ESW_FLOWTABLE(esw->dev, flow_source))
+		return false;
+
+	return true;
+}
+
+#define MLX5_ESW_METADATA_RSVD_UPLINK 1
+
+/* Share the same metadata for uplink's. This is fine because:
+ * (a) In shared FDB mode (LAG) both uplink's are treated the
+ *     same and tagged with the same metadata.
+ * (b) In non shared FDB mode, packets from physical port0
+ *     cannot hit eswitch of PF1 and vice versa.
+ */
+static u32 mlx5_esw_match_metadata_reserved(struct mlx5_eswitch *esw)
+{
+	return MLX5_ESW_METADATA_RSVD_UPLINK;
+}
+
+u32 mlx5_esw_match_metadata_alloc(struct mlx5_eswitch *esw)
+{
+	u32 vport_end_ida = (1 << ESW_VPORT_BITS) - 1;
+	/* Reserve 0xf for internal port offload */
+	u32 max_pf_num = (1 << ESW_PFNUM_BITS) - 2;
+	u32 pf_num;
+	int id;
+
+	/* Only 4 bits of pf_num */
+	pf_num = mlx5_get_dev_index(esw->dev);
+	if (pf_num > max_pf_num)
+		return 0;
+
+	/* Metadata is 4 bits of PFNUM and 12 bits of unique id */
+	/* Use only non-zero vport_id (2-4095) for all PF's */
+	id = ida_alloc_range(&esw->offloads.vport_metadata_ida,
+			     MLX5_ESW_METADATA_RSVD_UPLINK + 1,
+			     vport_end_ida, GFP_KERNEL);
+	if (id < 0)
+		return 0;
+	id = (pf_num << ESW_VPORT_BITS) | id;
+	return id;
+}
+
+void mlx5_esw_match_metadata_free(struct mlx5_eswitch *esw, u32 metadata)
+{
+	u32 vport_bit_mask = (1 << ESW_VPORT_BITS) - 1;
+
+	/* Metadata contains only 12 bits of actual ida id */
+	ida_free(&esw->offloads.vport_metadata_ida, metadata & vport_bit_mask);
+}
+
+static int esw_offloads_vport_metadata_setup(struct mlx5_eswitch *esw,
+					     struct mlx5_vport *vport)
+{
+	if (vport->vport == MLX5_VPORT_UPLINK)
+		vport->default_metadata = mlx5_esw_match_metadata_reserved(esw);
+	else
+		vport->default_metadata = mlx5_esw_match_metadata_alloc(esw);
+
+	vport->metadata = vport->default_metadata;
+	return vport->metadata ? 0 : -ENOSPC;
+}
+
+static void esw_offloads_vport_metadata_cleanup(struct mlx5_eswitch *esw,
+						struct mlx5_vport *vport)
+{
+	if (!vport->default_metadata)
+		return;
+
+	if (vport->vport == MLX5_VPORT_UPLINK)
+		return;
+
+	WARN_ON(vport->metadata != vport->default_metadata);
+	mlx5_esw_match_metadata_free(esw, vport->default_metadata);
+}
+
+static void esw_offloads_metadata_uninit(struct mlx5_eswitch *esw)
+{
+	struct mlx5_vport *vport;
+	unsigned long i;
+
+	if (!mlx5_eswitch_vport_match_metadata_enabled(esw))
+		return;
+
+	mlx5_esw_for_each_vport(esw, i, vport)
+		esw_offloads_vport_metadata_cleanup(esw, vport);
+}
+
+static int esw_offloads_metadata_init(struct mlx5_eswitch *esw)
+{
+	struct mlx5_vport *vport;
+	unsigned long i;
+	int err;
+
+	if (!mlx5_eswitch_vport_match_metadata_enabled(esw))
+		return 0;
+
+	mlx5_esw_for_each_vport(esw, i, vport) {
+		err = esw_offloads_vport_metadata_setup(esw, vport);
+		if (err)
+			goto metadata_err;
+	}
+
+	return 0;
+
+metadata_err:
+	esw_offloads_metadata_uninit(esw);
+	return err;
+}
+
+int mlx5_esw_offloads_vport_metadata_set(struct mlx5_eswitch *esw, bool enable)
+{
+	int err = 0;
+
+	down_write(&esw->mode_lock);
+	if (mlx5_esw_is_fdb_created(esw)) {
+		err = -EBUSY;
+		goto done;
+	}
+	if (!mlx5_esw_vport_match_metadata_supported(esw)) {
+		err = -EOPNOTSUPP;
+		goto done;
+	}
+	if (enable)
+		esw->flags |= MLX5_ESWITCH_VPORT_MATCH_METADATA;
+	else
+		esw->flags &= ~MLX5_ESWITCH_VPORT_MATCH_METADATA;
+done:
+	up_write(&esw->mode_lock);
+	return err;
+}
+
+int
+esw_vport_create_offloads_acl_tables(struct mlx5_eswitch *esw,
+				     struct mlx5_vport *vport)
+{
+	int err;
+
+	err = esw_acl_ingress_ofld_setup(esw, vport);
+	if (err)
+		return err;
+
+	err = esw_acl_egress_ofld_setup(esw, vport);
+	if (err)
+		goto egress_err;
+
+	return 0;
+
+egress_err:
+	esw_acl_ingress_ofld_cleanup(esw, vport);
+	return err;
+}
+
+void
+esw_vport_destroy_offloads_acl_tables(struct mlx5_eswitch *esw,
+				      struct mlx5_vport *vport)
+{
+	esw_acl_egress_ofld_cleanup(vport);
+	esw_acl_ingress_ofld_cleanup(esw, vport);
+}
+
+static int esw_create_uplink_offloads_acl_tables(struct mlx5_eswitch *esw)
+{
+	struct mlx5_vport *vport;
+
+	vport = mlx5_eswitch_get_vport(esw, MLX5_VPORT_UPLINK);
+	if (IS_ERR(vport))
+		return PTR_ERR(vport);
+
+	return esw_vport_create_offloads_acl_tables(esw, vport);
+}
+
+static void esw_destroy_uplink_offloads_acl_tables(struct mlx5_eswitch *esw)
+{
+	struct mlx5_vport *vport;
+
+	vport = mlx5_eswitch_get_vport(esw, MLX5_VPORT_UPLINK);
+	if (IS_ERR(vport))
+		return;
+
+	esw_vport_destroy_offloads_acl_tables(esw, vport);
+}
+
+int mlx5_eswitch_reload_reps(struct mlx5_eswitch *esw)
+{
+	struct mlx5_eswitch_rep *rep;
+	unsigned long i;
+	int ret;
+
+	if (!esw || esw->mode != MLX5_ESWITCH_OFFLOADS)
+		return 0;
+
+	rep = mlx5_eswitch_get_rep(esw, MLX5_VPORT_UPLINK);
+	if (atomic_read(&rep->rep_data[REP_ETH].state) != REP_LOADED)
+		return 0;
+
+	ret = mlx5_esw_offloads_rep_load(esw, MLX5_VPORT_UPLINK);
+	if (ret)
+		return ret;
+
+	mlx5_esw_for_each_rep(esw, i, rep) {
+		if (atomic_read(&rep->rep_data[REP_ETH].state) == REP_LOADED)
+			mlx5_esw_offloads_rep_load(esw, rep->vport);
+	}
+
+	return 0;
+}
+
+static int esw_offloads_steering_init(struct mlx5_eswitch *esw)
+{
+	struct mlx5_esw_indir_table *indir;
+	int err;
+
+	memset(&esw->fdb_table.offloads, 0, sizeof(struct offloads_fdb));
+	mutex_init(&esw->fdb_table.offloads.vports.lock);
+	hash_init(esw->fdb_table.offloads.vports.table);
+	atomic64_set(&esw->user_count, 0);
+
+	indir = mlx5_esw_indir_table_init();
+	if (IS_ERR(indir)) {
+		err = PTR_ERR(indir);
+		goto create_indir_err;
+	}
+	esw->fdb_table.offloads.indir = indir;
+
+	err = esw_create_uplink_offloads_acl_tables(esw);
+	if (err)
+		goto create_acl_err;
+
+	err = esw_create_offloads_table(esw);
+	if (err)
+		goto create_offloads_err;
+
+	err = esw_create_restore_table(esw);
+	if (err)
+		goto create_restore_err;
+
+	err = esw_create_offloads_fdb_tables(esw);
+	if (err)
+		goto create_fdb_err;
+
+	err = esw_create_vport_rx_group(esw);
+	if (err)
+		goto create_fg_err;
+
+	err = esw_create_vport_rx_drop_group(esw);
+	if (err)
+		goto create_rx_drop_fg_err;
+
+	err = esw_create_vport_rx_drop_rule(esw);
+	if (err)
+		goto create_rx_drop_rule_err;
+
+	return 0;
+
+create_rx_drop_rule_err:
+	esw_destroy_vport_rx_drop_group(esw);
+create_rx_drop_fg_err:
+	esw_destroy_vport_rx_group(esw);
+create_fg_err:
+	esw_destroy_offloads_fdb_tables(esw);
+create_fdb_err:
+	esw_destroy_restore_table(esw);
+create_restore_err:
+	esw_destroy_offloads_table(esw);
+create_offloads_err:
+	esw_destroy_uplink_offloads_acl_tables(esw);
+create_acl_err:
+	mlx5_esw_indir_table_destroy(esw->fdb_table.offloads.indir);
+create_indir_err:
+	mutex_destroy(&esw->fdb_table.offloads.vports.lock);
+	return err;
+}
+
+static void esw_offloads_steering_cleanup(struct mlx5_eswitch *esw)
+{
+	esw_destroy_vport_rx_drop_rule(esw);
+	esw_destroy_vport_rx_drop_group(esw);
+	esw_destroy_vport_rx_group(esw);
+	esw_destroy_offloads_fdb_tables(esw);
+	esw_destroy_restore_table(esw);
+	esw_destroy_offloads_table(esw);
+	esw_destroy_uplink_offloads_acl_tables(esw);
+	mlx5_esw_indir_table_destroy(esw->fdb_table.offloads.indir);
+	mutex_destroy(&esw->fdb_table.offloads.vports.lock);
+}
+
+static void
+esw_vfs_changed_event_handler(struct mlx5_eswitch *esw, const u32 *out)
+{
+	struct devlink *devlink;
+	bool host_pf_disabled;
+	u16 new_num_vfs;
+
+	new_num_vfs = MLX5_GET(query_esw_functions_out, out,
+			       host_params_context.host_num_of_vfs);
+	host_pf_disabled = MLX5_GET(query_esw_functions_out, out,
+				    host_params_context.host_pf_disabled);
+
+	if (new_num_vfs == esw->esw_funcs.num_vfs || host_pf_disabled)
+		return;
+
+	devlink = priv_to_devlink(esw->dev);
+	devl_lock(devlink);
+	/* Number of VFs can only change from "0 to x" or "x to 0". */
+	if (esw->esw_funcs.num_vfs > 0) {
+		mlx5_eswitch_unload_vf_vports(esw, esw->esw_funcs.num_vfs);
+	} else {
+		int err;
+
+		err = mlx5_eswitch_load_vf_vports(esw, new_num_vfs,
+						  MLX5_VPORT_UC_ADDR_CHANGE);
+		if (err) {
+			devl_unlock(devlink);
+			return;
+		}
+	}
+	esw->esw_funcs.num_vfs = new_num_vfs;
+	devl_unlock(devlink);
+}
+
+static void esw_functions_changed_event_handler(struct work_struct *work)
+{
+	struct mlx5_host_work *host_work;
+	struct mlx5_eswitch *esw;
+	const u32 *out;
+
+	host_work = container_of(work, struct mlx5_host_work, work);
+	esw = host_work->esw;
+
+	out = mlx5_esw_query_functions(esw->dev);
+	if (IS_ERR(out))
+		goto out;
+
+	esw_vfs_changed_event_handler(esw, out);
+	kvfree(out);
+out:
+	kfree(host_work);
+}
+
+int mlx5_esw_funcs_changed_handler(struct notifier_block *nb, unsigned long type, void *data)
+{
+	struct mlx5_esw_functions *esw_funcs;
+	struct mlx5_host_work *host_work;
+	struct mlx5_eswitch *esw;
+
+	host_work = kzalloc(sizeof(*host_work), GFP_ATOMIC);
+	if (!host_work)
+		return NOTIFY_DONE;
+
+	esw_funcs = mlx5_nb_cof(nb, struct mlx5_esw_functions, nb);
+	esw = container_of(esw_funcs, struct mlx5_eswitch, esw_funcs);
+
+	host_work->esw = esw;
+
+	INIT_WORK(&host_work->work, esw_functions_changed_event_handler);
+	queue_work(esw->work_queue, &host_work->work);
+
+	return NOTIFY_OK;
+}
+
+static int mlx5_esw_host_number_init(struct mlx5_eswitch *esw)
+{
+	const u32 *query_host_out;
+
+	if (!mlx5_core_is_ecpf_esw_manager(esw->dev))
+		return 0;
+
+	query_host_out = mlx5_esw_query_functions(esw->dev);
+	if (IS_ERR(query_host_out))
+		return PTR_ERR(query_host_out);
+
+	/* Mark non local controller with non zero controller number. */
+	esw->offloads.host_number = MLX5_GET(query_esw_functions_out, query_host_out,
+					     host_params_context.host_number);
+	kvfree(query_host_out);
+	return 0;
+}
+
+bool mlx5_esw_offloads_controller_valid(const struct mlx5_eswitch *esw, u32 controller)
+{
+	/* Local controller is always valid */
+	if (controller == 0)
+		return true;
+
+	if (!mlx5_core_is_ecpf_esw_manager(esw->dev))
+		return false;
+
+	/* External host number starts with zero in device */
+	return (controller == esw->offloads.host_number + 1);
+}
+
+int esw_offloads_enable(struct mlx5_eswitch *esw)
+{
+	struct mapping_ctx *reg_c0_obj_pool;
+	struct mlx5_vport *vport;
+	unsigned long i;
+	u64 mapping_id;
+	int err;
+
+	mutex_init(&esw->offloads.termtbl_mutex);
+	mlx5_rdma_enable_roce(esw->dev);
+
+	err = mlx5_esw_host_number_init(esw);
+	if (err)
+		goto err_metadata;
+
+	err = esw_offloads_metadata_init(esw);
+	if (err)
+		goto err_metadata;
+
+	err = esw_set_passing_vport_metadata(esw, true);
+	if (err)
+		goto err_vport_metadata;
+
+	mapping_id = mlx5_query_nic_system_image_guid(esw->dev);
+
+	reg_c0_obj_pool = mapping_create_for_id(mapping_id, MAPPING_TYPE_CHAIN,
+						sizeof(struct mlx5_mapped_obj),
+						ESW_REG_C0_USER_DATA_METADATA_MASK,
+						true);
+
+	if (IS_ERR(reg_c0_obj_pool)) {
+		err = PTR_ERR(reg_c0_obj_pool);
+		goto err_pool;
+	}
+	esw->offloads.reg_c0_obj_pool = reg_c0_obj_pool;
+
+	err = esw_offloads_steering_init(esw);
+	if (err)
+		goto err_steering_init;
+
+	/* Representor will control the vport link state */
+	mlx5_esw_for_each_vf_vport(esw, i, vport, esw->esw_funcs.num_vfs)
+		vport->info.link_state = MLX5_VPORT_ADMIN_STATE_DOWN;
+
+	/* Uplink vport rep must load first. */
+	err = esw_offloads_load_rep(esw, MLX5_VPORT_UPLINK);
+	if (err)
+		goto err_uplink;
+
+	err = mlx5_eswitch_enable_pf_vf_vports(esw, MLX5_VPORT_UC_ADDR_CHANGE);
+	if (err)
+		goto err_vports;
+
+	return 0;
+
+err_vports:
+	esw_offloads_unload_rep(esw, MLX5_VPORT_UPLINK);
+err_uplink:
+	esw_offloads_steering_cleanup(esw);
+err_steering_init:
+	mapping_destroy(reg_c0_obj_pool);
+err_pool:
+	esw_set_passing_vport_metadata(esw, false);
+err_vport_metadata:
+	esw_offloads_metadata_uninit(esw);
+err_metadata:
+	mlx5_rdma_disable_roce(esw->dev);
+	mutex_destroy(&esw->offloads.termtbl_mutex);
+	return err;
+}
+
+static int esw_offloads_stop(struct mlx5_eswitch *esw,
+			     struct netlink_ext_ack *extack)
+{
+	int err;
+
+	esw->mode = MLX5_ESWITCH_LEGACY;
+
+	/* If changing from switchdev to legacy mode without sriov enabled,
+	 * no need to create legacy fdb.
+	 */
+	if (!mlx5_sriov_is_enabled(esw->dev))
+		return 0;
+
+	err = mlx5_eswitch_enable_locked(esw, MLX5_ESWITCH_IGNORE_NUM_VFS);
+	if (err)
+		NL_SET_ERR_MSG_MOD(extack, "Failed setting eswitch to legacy");
+
+	return err;
+}
+
+void esw_offloads_disable(struct mlx5_eswitch *esw)
+{
+	mlx5_eswitch_disable_pf_vf_vports(esw);
+	esw_offloads_unload_rep(esw, MLX5_VPORT_UPLINK);
+	esw_set_passing_vport_metadata(esw, false);
+	esw_offloads_steering_cleanup(esw);
+	mapping_destroy(esw->offloads.reg_c0_obj_pool);
+	esw_offloads_metadata_uninit(esw);
+	mlx5_rdma_disable_roce(esw->dev);
+	mutex_destroy(&esw->offloads.termtbl_mutex);
+}
+
+static int esw_mode_from_devlink(u16 mode, u16 *mlx5_mode)
+{
+	switch (mode) {
+	case DEVLINK_ESWITCH_MODE_LEGACY:
+		*mlx5_mode = MLX5_ESWITCH_LEGACY;
+		break;
+	case DEVLINK_ESWITCH_MODE_SWITCHDEV:
+		*mlx5_mode = MLX5_ESWITCH_OFFLOADS;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int esw_mode_to_devlink(u16 mlx5_mode, u16 *mode)
+{
+	switch (mlx5_mode) {
+	case MLX5_ESWITCH_LEGACY:
+		*mode = DEVLINK_ESWITCH_MODE_LEGACY;
+		break;
+	case MLX5_ESWITCH_OFFLOADS:
+		*mode = DEVLINK_ESWITCH_MODE_SWITCHDEV;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int esw_inline_mode_from_devlink(u8 mode, u8 *mlx5_mode)
+{
+	switch (mode) {
+	case DEVLINK_ESWITCH_INLINE_MODE_NONE:
+		*mlx5_mode = MLX5_INLINE_MODE_NONE;
+		break;
+	case DEVLINK_ESWITCH_INLINE_MODE_LINK:
+		*mlx5_mode = MLX5_INLINE_MODE_L2;
+		break;
+	case DEVLINK_ESWITCH_INLINE_MODE_NETWORK:
+		*mlx5_mode = MLX5_INLINE_MODE_IP;
+		break;
+	case DEVLINK_ESWITCH_INLINE_MODE_TRANSPORT:
+		*mlx5_mode = MLX5_INLINE_MODE_TCP_UDP;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int esw_inline_mode_to_devlink(u8 mlx5_mode, u8 *mode)
+{
+	switch (mlx5_mode) {
+	case MLX5_INLINE_MODE_NONE:
+		*mode = DEVLINK_ESWITCH_INLINE_MODE_NONE;
+		break;
+	case MLX5_INLINE_MODE_L2:
+		*mode = DEVLINK_ESWITCH_INLINE_MODE_LINK;
+		break;
+	case MLX5_INLINE_MODE_IP:
+		*mode = DEVLINK_ESWITCH_INLINE_MODE_NETWORK;
+		break;
+	case MLX5_INLINE_MODE_TCP_UDP:
+		*mode = DEVLINK_ESWITCH_INLINE_MODE_TRANSPORT;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static bool esw_offloads_devlink_ns_eq_netdev_ns(struct devlink *devlink)
+{
+	struct net *devl_net, *netdev_net;
+	struct mlx5_eswitch *esw;
+
+	esw = mlx5_devlink_eswitch_get(devlink);
+	netdev_net = dev_net(esw->dev->mlx5e_res.uplink_netdev);
+	devl_net = devlink_net(devlink);
+
+	return net_eq(devl_net, netdev_net);
+}
+
+int mlx5_devlink_eswitch_mode_set(struct devlink *devlink, u16 mode,
+				  struct netlink_ext_ack *extack)
+{
+	u16 cur_mlx5_mode, mlx5_mode = 0;
+	struct mlx5_eswitch *esw;
+	int err = 0;
+
+	esw = mlx5_devlink_eswitch_get(devlink);
+	if (IS_ERR(esw))
+		return PTR_ERR(esw);
+
+	if (esw_mode_from_devlink(mode, &mlx5_mode))
+		return -EINVAL;
+
+	if (mode == DEVLINK_ESWITCH_MODE_SWITCHDEV &&
+	    !esw_offloads_devlink_ns_eq_netdev_ns(devlink)) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "Can't change E-Switch mode to switchdev when netdev net namespace has diverged from the devlink's.");
+		return -EPERM;
+	}
+
+	mlx5_lag_disable_change(esw->dev);
+	err = mlx5_esw_try_lock(esw);
+	if (err < 0) {
+		NL_SET_ERR_MSG_MOD(extack, "Can't change mode, E-Switch is busy");
+		goto enable_lag;
+	}
+	cur_mlx5_mode = err;
+	err = 0;
+
+	if (cur_mlx5_mode == mlx5_mode)
+		goto unlock;
+
+	mlx5_eswitch_disable_locked(esw);
+	if (mode == DEVLINK_ESWITCH_MODE_SWITCHDEV) {
+		if (mlx5_devlink_trap_get_num_active(esw->dev)) {
+			NL_SET_ERR_MSG_MOD(extack,
+					   "Can't change mode while devlink traps are active");
+			err = -EOPNOTSUPP;
+			goto unlock;
+		}
+		err = esw_offloads_start(esw, extack);
+	} else if (mode == DEVLINK_ESWITCH_MODE_LEGACY) {
+		err = esw_offloads_stop(esw, extack);
+		mlx5_rescan_drivers(esw->dev);
+	} else {
+		err = -EINVAL;
+	}
+
+unlock:
+	mlx5_esw_unlock(esw);
+enable_lag:
+	mlx5_lag_enable_change(esw->dev);
+	return err;
+}
+
+int mlx5_devlink_eswitch_mode_get(struct devlink *devlink, u16 *mode)
+{
+	struct mlx5_eswitch *esw;
+	int err;
+
+	esw = mlx5_devlink_eswitch_get(devlink);
+	if (IS_ERR(esw))
+		return PTR_ERR(esw);
+
+	down_write(&esw->mode_lock);
+	err = esw_mode_to_devlink(esw->mode, mode);
+	up_write(&esw->mode_lock);
+	return err;
+}
+
+static int mlx5_esw_vports_inline_set(struct mlx5_eswitch *esw, u8 mlx5_mode,
+				      struct netlink_ext_ack *extack)
+{
+	struct mlx5_core_dev *dev = esw->dev;
+	struct mlx5_vport *vport;
+	u16 err_vport_num = 0;
+	unsigned long i;
+	int err = 0;
+
+	mlx5_esw_for_each_host_func_vport(esw, i, vport, esw->esw_funcs.num_vfs) {
+		err = mlx5_modify_nic_vport_min_inline(dev, vport->vport, mlx5_mode);
+		if (err) {
+			err_vport_num = vport->vport;
+			NL_SET_ERR_MSG_MOD(extack,
+					   "Failed to set min inline on vport");
+			goto revert_inline_mode;
+		}
+	}
+	return 0;
+
+revert_inline_mode:
+	mlx5_esw_for_each_host_func_vport(esw, i, vport, esw->esw_funcs.num_vfs) {
+		if (vport->vport == err_vport_num)
+			break;
+		mlx5_modify_nic_vport_min_inline(dev,
+						 vport->vport,
+						 esw->offloads.inline_mode);
+	}
+	return err;
+}
+
+int mlx5_devlink_eswitch_inline_mode_set(struct devlink *devlink, u8 mode,
+					 struct netlink_ext_ack *extack)
+{
+	struct mlx5_core_dev *dev = devlink_priv(devlink);
+	struct mlx5_eswitch *esw;
+	u8 mlx5_mode;
+	int err;
+
+	esw = mlx5_devlink_eswitch_get(devlink);
+	if (IS_ERR(esw))
+		return PTR_ERR(esw);
+
+	down_write(&esw->mode_lock);
+
+	switch (MLX5_CAP_ETH(dev, wqe_inline_mode)) {
+	case MLX5_CAP_INLINE_MODE_NOT_REQUIRED:
+		if (mode == DEVLINK_ESWITCH_INLINE_MODE_NONE) {
+			err = 0;
+			goto out;
+		}
+
+		fallthrough;
+	case MLX5_CAP_INLINE_MODE_L2:
+		NL_SET_ERR_MSG_MOD(extack, "Inline mode can't be set");
+		err = -EOPNOTSUPP;
+		goto out;
+	case MLX5_CAP_INLINE_MODE_VPORT_CONTEXT:
+		break;
+	}
+
+	if (atomic64_read(&esw->offloads.num_flows) > 0) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "Can't set inline mode when flows are configured");
+		err = -EOPNOTSUPP;
+		goto out;
+	}
+
+	err = esw_inline_mode_from_devlink(mode, &mlx5_mode);
+	if (err)
+		goto out;
+
+	err = mlx5_esw_vports_inline_set(esw, mlx5_mode, extack);
+	if (err)
+		goto out;
+
+	esw->offloads.inline_mode = mlx5_mode;
+	up_write(&esw->mode_lock);
+	return 0;
+
+out:
+	up_write(&esw->mode_lock);
+	return err;
+}
+
+int mlx5_devlink_eswitch_inline_mode_get(struct devlink *devlink, u8 *mode)
+{
+	struct mlx5_eswitch *esw;
+	int err;
+
+	esw = mlx5_devlink_eswitch_get(devlink);
+	if (IS_ERR(esw))
+		return PTR_ERR(esw);
+
+	down_write(&esw->mode_lock);
+	err = esw_inline_mode_to_devlink(esw->offloads.inline_mode, mode);
+	up_write(&esw->mode_lock);
+	return err;
+}
+
+int mlx5_devlink_eswitch_encap_mode_set(struct devlink *devlink,
+					enum devlink_eswitch_encap_mode encap,
+					struct netlink_ext_ack *extack)
+{
+	struct mlx5_core_dev *dev = devlink_priv(devlink);
+	struct mlx5_eswitch *esw;
+	int err = 0;
+
+	esw = mlx5_devlink_eswitch_get(devlink);
+	if (IS_ERR(esw))
+		return PTR_ERR(esw);
+
+	down_write(&esw->mode_lock);
+
+	if (encap != DEVLINK_ESWITCH_ENCAP_MODE_NONE &&
+	    (!MLX5_CAP_ESW_FLOWTABLE_FDB(dev, reformat) ||
+	     !MLX5_CAP_ESW_FLOWTABLE_FDB(dev, decap))) {
+		err = -EOPNOTSUPP;
+		goto unlock;
+	}
+
+	if (encap && encap != DEVLINK_ESWITCH_ENCAP_MODE_BASIC) {
+		err = -EOPNOTSUPP;
+		goto unlock;
+	}
+
+	if (esw->mode == MLX5_ESWITCH_LEGACY) {
+		esw->offloads.encap = encap;
+		goto unlock;
+	}
+
+	if (esw->offloads.encap == encap)
+		goto unlock;
+
+	if (atomic64_read(&esw->offloads.num_flows) > 0) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "Can't set encapsulation when flows are configured");
+		err = -EOPNOTSUPP;
+		goto unlock;
+	}
+
+	esw_destroy_offloads_fdb_tables(esw);
+
+	esw->offloads.encap = encap;
+
+	err = esw_create_offloads_fdb_tables(esw);
+
+	if (err) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "Failed re-creating fast FDB table");
+		esw->offloads.encap = !encap;
+		(void)esw_create_offloads_fdb_tables(esw);
+	}
+
+unlock:
+	up_write(&esw->mode_lock);
+	return err;
+}
+
+int mlx5_devlink_eswitch_encap_mode_get(struct devlink *devlink,
+					enum devlink_eswitch_encap_mode *encap)
+{
+	struct mlx5_eswitch *esw;
+
+	esw = mlx5_devlink_eswitch_get(devlink);
+	if (IS_ERR(esw))
+		return PTR_ERR(esw);
+
+	down_write(&esw->mode_lock);
+	*encap = esw->offloads.encap;
+	up_write(&esw->mode_lock);
+	return 0;
+}
+
+static bool
+mlx5_eswitch_vport_has_rep(const struct mlx5_eswitch *esw, u16 vport_num)
+{
+	/* Currently, only ECPF based device has representor for host PF. */
+	if (vport_num == MLX5_VPORT_PF &&
+	    !mlx5_core_is_ecpf_esw_manager(esw->dev))
+		return false;
+
+	if (vport_num == MLX5_VPORT_ECPF &&
+	    !mlx5_ecpf_vport_exists(esw->dev))
+		return false;
+
+	return true;
+}
+
+void mlx5_eswitch_register_vport_reps(struct mlx5_eswitch *esw,
+				      const struct mlx5_eswitch_rep_ops *ops,
+				      u8 rep_type)
+{
+	struct mlx5_eswitch_rep_data *rep_data;
+	struct mlx5_eswitch_rep *rep;
+	unsigned long i;
+
+	esw->offloads.rep_ops[rep_type] = ops;
+	mlx5_esw_for_each_rep(esw, i, rep) {
+		if (likely(mlx5_eswitch_vport_has_rep(esw, rep->vport))) {
+			rep->esw = esw;
+			rep_data = &rep->rep_data[rep_type];
+			atomic_set(&rep_data->state, REP_REGISTERED);
+		}
+	}
+}
+EXPORT_SYMBOL(mlx5_eswitch_register_vport_reps);
+
+void mlx5_eswitch_unregister_vport_reps(struct mlx5_eswitch *esw, u8 rep_type)
+{
+	struct mlx5_eswitch_rep *rep;
+	unsigned long i;
+
+	if (esw->mode == MLX5_ESWITCH_OFFLOADS)
+		__unload_reps_all_vport(esw, rep_type);
+
+	mlx5_esw_for_each_rep(esw, i, rep)
+		atomic_set(&rep->rep_data[rep_type].state, REP_UNREGISTERED);
+}
+EXPORT_SYMBOL(mlx5_eswitch_unregister_vport_reps);
+
+void *mlx5_eswitch_get_uplink_priv(struct mlx5_eswitch *esw, u8 rep_type)
+{
+	struct mlx5_eswitch_rep *rep;
+
+	rep = mlx5_eswitch_get_rep(esw, MLX5_VPORT_UPLINK);
+	return rep->rep_data[rep_type].priv;
+}
+
+void *mlx5_eswitch_get_proto_dev(struct mlx5_eswitch *esw,
+				 u16 vport,
+				 u8 rep_type)
+{
+	struct mlx5_eswitch_rep *rep;
+
+	rep = mlx5_eswitch_get_rep(esw, vport);
+
+	if (atomic_read(&rep->rep_data[rep_type].state) == REP_LOADED &&
+	    esw->offloads.rep_ops[rep_type]->get_proto_dev)
+		return esw->offloads.rep_ops[rep_type]->get_proto_dev(rep);
+	return NULL;
+}
+EXPORT_SYMBOL(mlx5_eswitch_get_proto_dev);
+
+void *mlx5_eswitch_uplink_get_proto_dev(struct mlx5_eswitch *esw, u8 rep_type)
+{
+	return mlx5_eswitch_get_proto_dev(esw, MLX5_VPORT_UPLINK, rep_type);
+}
+EXPORT_SYMBOL(mlx5_eswitch_uplink_get_proto_dev);
+
+struct mlx5_eswitch_rep *mlx5_eswitch_vport_rep(struct mlx5_eswitch *esw,
+						u16 vport)
+{
+	return mlx5_eswitch_get_rep(esw, vport);
+}
+EXPORT_SYMBOL(mlx5_eswitch_vport_rep);
+
+bool mlx5_eswitch_reg_c1_loopback_enabled(const struct mlx5_eswitch *esw)
+{
+	return !!(esw->flags & MLX5_ESWITCH_REG_C1_LOOPBACK_ENABLED);
+}
+EXPORT_SYMBOL(mlx5_eswitch_reg_c1_loopback_enabled);
+
+bool mlx5_eswitch_vport_match_metadata_enabled(const struct mlx5_eswitch *esw)
+{
+	return !!(esw->flags & MLX5_ESWITCH_VPORT_MATCH_METADATA);
+}
+EXPORT_SYMBOL(mlx5_eswitch_vport_match_metadata_enabled);
+
+u32 mlx5_eswitch_get_vport_metadata_for_match(struct mlx5_eswitch *esw,
+					      u16 vport_num)
+{
+	struct mlx5_vport *vport = mlx5_eswitch_get_vport(esw, vport_num);
+
+	if (WARN_ON_ONCE(IS_ERR(vport)))
+		return 0;
+
+	return vport->metadata << (32 - ESW_SOURCE_PORT_METADATA_BITS);
+}
+EXPORT_SYMBOL(mlx5_eswitch_get_vport_metadata_for_match);
+
+int mlx5_esw_offloads_sf_vport_enable(struct mlx5_eswitch *esw, struct devlink_port *dl_port,
+				      u16 vport_num, u32 controller, u32 sfnum)
+{
+	int err;
+
+	err = mlx5_esw_vport_enable(esw, vport_num, MLX5_VPORT_UC_ADDR_CHANGE);
+	if (err)
+		return err;
+
+	err = mlx5_esw_devlink_sf_port_register(esw, dl_port, vport_num, controller, sfnum);
+	if (err)
+		goto devlink_err;
+
+	mlx5_esw_vport_debugfs_create(esw, vport_num, true, sfnum);
+	err = mlx5_esw_offloads_rep_load(esw, vport_num);
+	if (err)
+		goto rep_err;
+	return 0;
+
+rep_err:
+	mlx5_esw_vport_debugfs_destroy(esw, vport_num);
+	mlx5_esw_devlink_sf_port_unregister(esw, vport_num);
+devlink_err:
+	mlx5_esw_vport_disable(esw, vport_num);
+	return err;
+}
+
+void mlx5_esw_offloads_sf_vport_disable(struct mlx5_eswitch *esw, u16 vport_num)
+{
+	mlx5_esw_offloads_rep_unload(esw, vport_num);
+	mlx5_esw_vport_debugfs_destroy(esw, vport_num);
+	mlx5_esw_devlink_sf_port_unregister(esw, vport_num);
+	mlx5_esw_vport_disable(esw, vport_num);
+}
+
+static int mlx5_esw_query_vport_vhca_id(struct mlx5_eswitch *esw, u16 vport_num, u16 *vhca_id)
+{
+	int query_out_sz = MLX5_ST_SZ_BYTES(query_hca_cap_out);
+	void *query_ctx;
+	void *hca_caps;
+	int err;
+
+	*vhca_id = 0;
+	if (mlx5_esw_is_manager_vport(esw, vport_num) ||
+	    !MLX5_CAP_GEN(esw->dev, vhca_resource_manager))
+		return -EPERM;
+
+	query_ctx = kzalloc(query_out_sz, GFP_KERNEL);
+	if (!query_ctx)
+		return -ENOMEM;
+
+	err = mlx5_vport_get_other_func_cap(esw->dev, vport_num, query_ctx);
+	if (err)
+		goto out_free;
+
+	hca_caps = MLX5_ADDR_OF(query_hca_cap_out, query_ctx, capability);
+	*vhca_id = MLX5_GET(cmd_hca_cap, hca_caps, vhca_id);
+
+out_free:
+	kfree(query_ctx);
+	return err;
+}
+
+int mlx5_esw_vport_vhca_id_set(struct mlx5_eswitch *esw, u16 vport_num)
+{
+	u16 *old_entry, *vhca_map_entry, vhca_id;
+	int err;
+
+	err = mlx5_esw_query_vport_vhca_id(esw, vport_num, &vhca_id);
+	if (err) {
+		esw_warn(esw->dev, "Getting vhca_id for vport failed (vport=%u,err=%d)\n",
+			 vport_num, err);
+		return err;
+	}
+
+	vhca_map_entry = kmalloc(sizeof(*vhca_map_entry), GFP_KERNEL);
+	if (!vhca_map_entry)
+		return -ENOMEM;
+
+	*vhca_map_entry = vport_num;
+	old_entry = xa_store(&esw->offloads.vhca_map, vhca_id, vhca_map_entry, GFP_KERNEL);
+	if (xa_is_err(old_entry)) {
+		kfree(vhca_map_entry);
+		return xa_err(old_entry);
+	}
+	kfree(old_entry);
+	return 0;
+}
+
+void mlx5_esw_vport_vhca_id_clear(struct mlx5_eswitch *esw, u16 vport_num)
+{
+	u16 *vhca_map_entry, vhca_id;
+	int err;
+
+	err = mlx5_esw_query_vport_vhca_id(esw, vport_num, &vhca_id);
+	if (err)
+		esw_warn(esw->dev, "Getting vhca_id for vport failed (vport=%hu,err=%d)\n",
+			 vport_num, err);
+
+	vhca_map_entry = xa_erase(&esw->offloads.vhca_map, vhca_id);
+	kfree(vhca_map_entry);
+}
+
+int mlx5_eswitch_vhca_id_to_vport(struct mlx5_eswitch *esw, u16 vhca_id, u16 *vport_num)
+{
+	u16 *res = xa_load(&esw->offloads.vhca_map, vhca_id);
+
+	if (!res)
+		return -ENOENT;
+
+	*vport_num = *res;
+	return 0;
+}
+
+u32 mlx5_eswitch_get_vport_metadata_for_set(struct mlx5_eswitch *esw,
+					    u16 vport_num)
+{
+	struct mlx5_vport *vport = mlx5_eswitch_get_vport(esw, vport_num);
+
+	if (WARN_ON_ONCE(IS_ERR(vport)))
+		return 0;
+
+	return vport->metadata;
+}
+EXPORT_SYMBOL(mlx5_eswitch_get_vport_metadata_for_set);
+
+static bool
+is_port_function_supported(struct mlx5_eswitch *esw, u16 vport_num)
+{
+	return vport_num == MLX5_VPORT_PF ||
+	       mlx5_eswitch_is_vf_vport(esw, vport_num) ||
+	       mlx5_esw_is_sf_vport(esw, vport_num);
+}
+
+int mlx5_devlink_port_function_hw_addr_get(struct devlink_port *port,
+					   u8 *hw_addr, int *hw_addr_len,
+					   struct netlink_ext_ack *extack)
+{
+	struct mlx5_eswitch *esw;
+	struct mlx5_vport *vport;
+	u16 vport_num;
+
+	esw = mlx5_devlink_eswitch_get(port->devlink);
+	if (IS_ERR(esw))
+		return PTR_ERR(esw);
+
+	vport_num = mlx5_esw_devlink_port_index_to_vport_num(port->index);
+	if (!is_port_function_supported(esw, vport_num))
+		return -EOPNOTSUPP;
+
+	vport = mlx5_eswitch_get_vport(esw, vport_num);
+	if (IS_ERR(vport)) {
+		NL_SET_ERR_MSG_MOD(extack, "Invalid port");
+		return PTR_ERR(vport);
+	}
+
+	mutex_lock(&esw->state_lock);
+	ether_addr_copy(hw_addr, vport->info.mac);
+	*hw_addr_len = ETH_ALEN;
+	mutex_unlock(&esw->state_lock);
+	return 0;
+}
+
+int mlx5_devlink_port_function_hw_addr_set(struct devlink_port *port,
+					   const u8 *hw_addr, int hw_addr_len,
+					   struct netlink_ext_ack *extack)
+{
+	struct mlx5_eswitch *esw;
+	u16 vport_num;
+
+	esw = mlx5_devlink_eswitch_get(port->devlink);
+	if (IS_ERR(esw)) {
+		NL_SET_ERR_MSG_MOD(extack, "Eswitch doesn't support set hw_addr");
+		return PTR_ERR(esw);
+	}
+
+	vport_num = mlx5_esw_devlink_port_index_to_vport_num(port->index);
+	if (!is_port_function_supported(esw, vport_num)) {
+		NL_SET_ERR_MSG_MOD(extack, "Port doesn't support set hw_addr");
+		return -EINVAL;
+	}
+
+	return mlx5_eswitch_set_vport_mac(esw, vport_num, hw_addr);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads_termtbl.c b/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads_termtbl.c
new file mode 100644
index 000000000..edd910258
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads_termtbl.c
@@ -0,0 +1,335 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+// Copyright (c) 2019 Mellanox Technologies.
+
+#include <linux/mlx5/fs.h>
+#include "eswitch.h"
+#include "en_tc.h"
+#include "fs_core.h"
+
+struct mlx5_termtbl_handle {
+	struct hlist_node termtbl_hlist;
+
+	struct mlx5_flow_table *termtbl;
+	struct mlx5_flow_act flow_act;
+	struct mlx5_flow_destination dest;
+
+	struct mlx5_flow_handle *rule;
+	int ref_count;
+};
+
+static u32
+mlx5_eswitch_termtbl_hash(struct mlx5_flow_act *flow_act,
+			  struct mlx5_flow_destination *dest)
+{
+	u32 hash;
+
+	hash = jhash_1word(flow_act->action, 0);
+	hash = jhash((const void *)&flow_act->vlan,
+		     sizeof(flow_act->vlan), hash);
+	hash = jhash((const void *)&dest->vport.num,
+		     sizeof(dest->vport.num), hash);
+	hash = jhash((const void *)&dest->vport.vhca_id,
+		     sizeof(dest->vport.num), hash);
+	if (flow_act->pkt_reformat)
+		hash = jhash(flow_act->pkt_reformat,
+			     sizeof(*flow_act->pkt_reformat),
+			     hash);
+	return hash;
+}
+
+static int
+mlx5_eswitch_termtbl_cmp(struct mlx5_flow_act *flow_act1,
+			 struct mlx5_flow_destination *dest1,
+			 struct mlx5_flow_act *flow_act2,
+			 struct mlx5_flow_destination *dest2)
+{
+	int ret;
+
+	ret = flow_act1->action != flow_act2->action ||
+	      dest1->vport.num != dest2->vport.num ||
+	      dest1->vport.vhca_id != dest2->vport.vhca_id ||
+	      memcmp(&flow_act1->vlan, &flow_act2->vlan,
+		     sizeof(flow_act1->vlan));
+	if (ret)
+		return ret;
+
+	if (flow_act1->pkt_reformat && flow_act2->pkt_reformat)
+		return memcmp(flow_act1->pkt_reformat, flow_act2->pkt_reformat,
+			      sizeof(*flow_act1->pkt_reformat));
+
+	return !(flow_act1->pkt_reformat == flow_act2->pkt_reformat);
+}
+
+static int
+mlx5_eswitch_termtbl_create(struct mlx5_core_dev *dev,
+			    struct mlx5_termtbl_handle *tt,
+			    struct mlx5_flow_act *flow_act)
+{
+	struct mlx5_flow_table_attr ft_attr = {};
+	struct mlx5_flow_namespace *root_ns;
+	int err, err2;
+
+	root_ns = mlx5_get_flow_namespace(dev, MLX5_FLOW_NAMESPACE_FDB);
+	if (!root_ns) {
+		esw_warn(dev, "Failed to get FDB flow namespace\n");
+		return -EOPNOTSUPP;
+	}
+
+	/* As this is the terminating action then the termination table is the
+	 * same prio as the slow path
+	 */
+	ft_attr.flags = MLX5_FLOW_TABLE_TERMINATION | MLX5_FLOW_TABLE_UNMANAGED |
+			MLX5_FLOW_TABLE_TUNNEL_EN_REFORMAT;
+	ft_attr.prio = FDB_TC_OFFLOAD;
+	ft_attr.max_fte = 1;
+	ft_attr.level = 1;
+	ft_attr.autogroup.max_num_groups = 1;
+	tt->termtbl = mlx5_create_auto_grouped_flow_table(root_ns, &ft_attr);
+	if (IS_ERR(tt->termtbl)) {
+		err = PTR_ERR(tt->termtbl);
+		esw_warn(dev, "Failed to create termination table, err %pe\n", tt->termtbl);
+		return err;
+	}
+
+	tt->rule = mlx5_add_flow_rules(tt->termtbl, NULL, flow_act,
+				       &tt->dest, 1);
+	if (IS_ERR(tt->rule)) {
+		err = PTR_ERR(tt->rule);
+		esw_warn(dev, "Failed to create termination table rule, err %pe\n", tt->rule);
+		goto add_flow_err;
+	}
+	return 0;
+
+add_flow_err:
+	err2 = mlx5_destroy_flow_table(tt->termtbl);
+	if (err2)
+		esw_warn(dev, "Failed to destroy termination table, err %d\n", err2);
+
+	return err;
+}
+
+static struct mlx5_termtbl_handle *
+mlx5_eswitch_termtbl_get_create(struct mlx5_eswitch *esw,
+				struct mlx5_flow_act *flow_act,
+				struct mlx5_flow_destination *dest,
+				struct mlx5_esw_flow_attr *attr)
+{
+	struct mlx5_termtbl_handle *tt;
+	bool found = false;
+	u32 hash_key;
+	int err;
+
+	mutex_lock(&esw->offloads.termtbl_mutex);
+	hash_key = mlx5_eswitch_termtbl_hash(flow_act, dest);
+	hash_for_each_possible(esw->offloads.termtbl_tbl, tt,
+			       termtbl_hlist, hash_key) {
+		if (!mlx5_eswitch_termtbl_cmp(&tt->flow_act, &tt->dest,
+					      flow_act, dest)) {
+			found = true;
+			break;
+		}
+	}
+	if (found)
+		goto tt_add_ref;
+
+	tt = kzalloc(sizeof(*tt), GFP_KERNEL);
+	if (!tt) {
+		err = -ENOMEM;
+		goto tt_create_err;
+	}
+
+	tt->dest.type = MLX5_FLOW_DESTINATION_TYPE_VPORT;
+	tt->dest.vport.num = dest->vport.num;
+	tt->dest.vport.vhca_id = dest->vport.vhca_id;
+	tt->dest.vport.flags = dest->vport.flags;
+	memcpy(&tt->flow_act, flow_act, sizeof(*flow_act));
+
+	err = mlx5_eswitch_termtbl_create(esw->dev, tt, flow_act);
+	if (err)
+		goto tt_create_err;
+
+	hash_add(esw->offloads.termtbl_tbl, &tt->termtbl_hlist, hash_key);
+tt_add_ref:
+	tt->ref_count++;
+	mutex_unlock(&esw->offloads.termtbl_mutex);
+	return tt;
+tt_create_err:
+	kfree(tt);
+	mutex_unlock(&esw->offloads.termtbl_mutex);
+	return ERR_PTR(err);
+}
+
+void
+mlx5_eswitch_termtbl_put(struct mlx5_eswitch *esw,
+			 struct mlx5_termtbl_handle *tt)
+{
+	mutex_lock(&esw->offloads.termtbl_mutex);
+	if (--tt->ref_count == 0)
+		hash_del(&tt->termtbl_hlist);
+	mutex_unlock(&esw->offloads.termtbl_mutex);
+
+	if (!tt->ref_count) {
+		mlx5_del_flow_rules(tt->rule);
+		mlx5_destroy_flow_table(tt->termtbl);
+		kfree(tt);
+	}
+}
+
+static void
+mlx5_eswitch_termtbl_actions_move(struct mlx5_flow_act *src,
+				  struct mlx5_flow_act *dst)
+{
+	if (src->action & MLX5_FLOW_CONTEXT_ACTION_VLAN_PUSH) {
+		src->action &= ~MLX5_FLOW_CONTEXT_ACTION_VLAN_PUSH;
+		dst->action |= MLX5_FLOW_CONTEXT_ACTION_VLAN_PUSH;
+		memcpy(&dst->vlan[0], &src->vlan[0], sizeof(src->vlan[0]));
+		memset(&src->vlan[0], 0, sizeof(src->vlan[0]));
+
+		if (src->action & MLX5_FLOW_CONTEXT_ACTION_VLAN_PUSH_2) {
+			src->action &= ~MLX5_FLOW_CONTEXT_ACTION_VLAN_PUSH_2;
+			dst->action |= MLX5_FLOW_CONTEXT_ACTION_VLAN_PUSH_2;
+			memcpy(&dst->vlan[1], &src->vlan[1], sizeof(src->vlan[1]));
+			memset(&src->vlan[1], 0, sizeof(src->vlan[1]));
+		}
+	}
+}
+
+static bool mlx5_eswitch_offload_is_uplink_port(const struct mlx5_eswitch *esw,
+						const struct mlx5_flow_spec *spec)
+{
+	u16 port_mask, port_value;
+
+	if (MLX5_CAP_ESW_FLOWTABLE(esw->dev, flow_source))
+		return spec->flow_context.flow_source ==
+					MLX5_FLOW_CONTEXT_FLOW_SOURCE_UPLINK;
+
+	port_mask = MLX5_GET(fte_match_param, spec->match_criteria,
+			     misc_parameters.source_port);
+	port_value = MLX5_GET(fte_match_param, spec->match_value,
+			      misc_parameters.source_port);
+	return (port_mask & port_value) == MLX5_VPORT_UPLINK;
+}
+
+bool
+mlx5_eswitch_termtbl_required(struct mlx5_eswitch *esw,
+			      struct mlx5_flow_attr *attr,
+			      struct mlx5_flow_act *flow_act,
+			      struct mlx5_flow_spec *spec)
+{
+	struct mlx5_esw_flow_attr *esw_attr = attr->esw_attr;
+	int i;
+
+	if (!MLX5_CAP_ESW_FLOWTABLE_FDB(esw->dev, termination_table) ||
+	    !MLX5_CAP_ESW_FLOWTABLE_FDB(esw->dev, ignore_flow_level) ||
+	    mlx5e_tc_attr_flags_skip(attr->flags) ||
+	    (!mlx5_eswitch_offload_is_uplink_port(esw, spec) && !esw_attr->int_port))
+		return false;
+
+	/* push vlan on RX */
+	if (flow_act->action & MLX5_FLOW_CONTEXT_ACTION_VLAN_PUSH &&
+	    !(mlx5_fs_get_capabilities(esw->dev, MLX5_FLOW_NAMESPACE_FDB) &
+	      MLX5_FLOW_STEERING_CAP_VLAN_PUSH_ON_RX))
+		return true;
+
+	/* hairpin */
+	for (i = esw_attr->split_count; i < esw_attr->out_count; i++)
+		if (!esw_attr->dest_int_port && esw_attr->dests[i].rep &&
+		    esw_attr->dests[i].rep->vport == MLX5_VPORT_UPLINK)
+			return true;
+
+	return false;
+}
+
+struct mlx5_flow_handle *
+mlx5_eswitch_add_termtbl_rule(struct mlx5_eswitch *esw,
+			      struct mlx5_flow_table *fdb,
+			      struct mlx5_flow_spec *spec,
+			      struct mlx5_esw_flow_attr *attr,
+			      struct mlx5_flow_act *flow_act,
+			      struct mlx5_flow_destination *dest,
+			      int num_dest)
+{
+	struct mlx5_flow_act term_tbl_act = {};
+	struct mlx5_flow_handle *rule = NULL;
+	bool term_table_created = false;
+	int num_vport_dests = 0;
+	int i, curr_dest;
+
+	mlx5_eswitch_termtbl_actions_move(flow_act, &term_tbl_act);
+	term_tbl_act.action |= MLX5_FLOW_CONTEXT_ACTION_FWD_DEST;
+
+	for (i = 0; i < num_dest; i++) {
+		struct mlx5_termtbl_handle *tt;
+
+		/* only vport destinations can be terminated */
+		if (dest[i].type != MLX5_FLOW_DESTINATION_TYPE_VPORT)
+			continue;
+
+		if (attr->dests[num_vport_dests].flags & MLX5_ESW_DEST_ENCAP) {
+			term_tbl_act.action |= MLX5_FLOW_CONTEXT_ACTION_PACKET_REFORMAT;
+			term_tbl_act.pkt_reformat = attr->dests[num_vport_dests].pkt_reformat;
+		} else {
+			term_tbl_act.action &= ~MLX5_FLOW_CONTEXT_ACTION_PACKET_REFORMAT;
+			term_tbl_act.pkt_reformat = NULL;
+		}
+
+		/* get the terminating table for the action list */
+		tt = mlx5_eswitch_termtbl_get_create(esw, &term_tbl_act,
+						     &dest[i], attr);
+		if (IS_ERR(tt)) {
+			esw_warn(esw->dev, "Failed to get termination table, err %pe\n", tt);
+			goto revert_changes;
+		}
+		attr->dests[num_vport_dests].termtbl = tt;
+		num_vport_dests++;
+
+		/* link the destination with the termination table */
+		dest[i].type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
+		dest[i].ft = tt->termtbl;
+		term_table_created = true;
+	}
+
+	/* at least one destination should reference a termination table */
+	if (!term_table_created)
+		goto revert_changes;
+
+	/* create the FTE */
+	flow_act->action &= ~MLX5_FLOW_CONTEXT_ACTION_PACKET_REFORMAT;
+	flow_act->pkt_reformat = NULL;
+	flow_act->flags |= FLOW_ACT_IGNORE_FLOW_LEVEL;
+	rule = mlx5_add_flow_rules(fdb, spec, flow_act, dest, num_dest);
+	if (IS_ERR(rule))
+		goto revert_changes;
+
+	goto out;
+
+revert_changes:
+	/* revert the changes that were made to the original flow_act
+	 * and fall-back to the original rule actions
+	 */
+	mlx5_eswitch_termtbl_actions_move(&term_tbl_act, flow_act);
+
+	for (curr_dest = 0; curr_dest < num_vport_dests; curr_dest++) {
+		struct mlx5_termtbl_handle *tt = attr->dests[curr_dest].termtbl;
+
+		attr->dests[curr_dest].termtbl = NULL;
+
+		/* search for the destination associated with the
+		 * current term table
+		 */
+		for (i = 0; i < num_dest; i++) {
+			if (dest[i].ft != tt->termtbl)
+				continue;
+
+			memset(&dest[i], 0, sizeof(dest[i]));
+			dest[i].type = MLX5_FLOW_DESTINATION_TYPE_VPORT;
+			dest[i].vport.num = tt->dest.vport.num;
+			dest[i].vport.vhca_id = tt->dest.vport.vhca_id;
+			mlx5_eswitch_termtbl_put(esw, tt);
+			break;
+		}
+	}
+	rule = mlx5_add_flow_rules(fdb, spec, flow_act, dest, num_dest);
+out:
+	return rule;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/events.c b/drivers/net/ethernet/mellanox/mlx5/core/events.c
new file mode 100644
index 000000000..9459e56ee
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/events.c
@@ -0,0 +1,446 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+// Copyright (c) 2018 Mellanox Technologies
+
+#include <linux/mlx5/driver.h>
+
+#include "mlx5_core.h"
+#include "lib/eq.h"
+#include "lib/mlx5.h"
+
+struct mlx5_event_nb {
+	struct mlx5_nb  nb;
+	void           *ctx;
+};
+
+/* General events handlers for the low level mlx5_core driver
+ *
+ * Other Major feature specific events such as
+ * clock/eswitch/fpga/FW trace and many others, are handled elsewhere, with
+ * separate notifiers callbacks, specifically by those mlx5 components.
+ */
+static int any_notifier(struct notifier_block *, unsigned long, void *);
+static int temp_warn(struct notifier_block *, unsigned long, void *);
+static int port_module(struct notifier_block *, unsigned long, void *);
+static int pcie_core(struct notifier_block *, unsigned long, void *);
+
+/* handler which forwards the event to events->fw_nh, driver notifiers */
+static int forward_event(struct notifier_block *, unsigned long, void *);
+
+static struct mlx5_nb events_nbs_ref[] = {
+	/* Events to be processed by mlx5_core */
+	{.nb.notifier_call = any_notifier,  .event_type = MLX5_EVENT_TYPE_NOTIFY_ANY },
+	{.nb.notifier_call = temp_warn,     .event_type = MLX5_EVENT_TYPE_TEMP_WARN_EVENT },
+	{.nb.notifier_call = port_module,   .event_type = MLX5_EVENT_TYPE_PORT_MODULE_EVENT },
+	{.nb.notifier_call = pcie_core,     .event_type = MLX5_EVENT_TYPE_GENERAL_EVENT },
+
+	/* Events to be forwarded (as is) to mlx5 core interfaces (mlx5e/mlx5_ib) */
+	{.nb.notifier_call = forward_event,   .event_type = MLX5_EVENT_TYPE_PORT_CHANGE },
+	{.nb.notifier_call = forward_event,   .event_type = MLX5_EVENT_TYPE_GENERAL_EVENT },
+	{.nb.notifier_call = forward_event,   .event_type = MLX5_EVENT_TYPE_OBJECT_CHANGE },
+	/* QP/WQ resource events to forward */
+	{.nb.notifier_call = forward_event,   .event_type = MLX5_EVENT_TYPE_DCT_DRAINED },
+	{.nb.notifier_call = forward_event,   .event_type = MLX5_EVENT_TYPE_PATH_MIG },
+	{.nb.notifier_call = forward_event,   .event_type = MLX5_EVENT_TYPE_COMM_EST },
+	{.nb.notifier_call = forward_event,   .event_type = MLX5_EVENT_TYPE_SQ_DRAINED },
+	{.nb.notifier_call = forward_event,   .event_type = MLX5_EVENT_TYPE_SRQ_LAST_WQE },
+	{.nb.notifier_call = forward_event,   .event_type = MLX5_EVENT_TYPE_WQ_CATAS_ERROR },
+	{.nb.notifier_call = forward_event,   .event_type = MLX5_EVENT_TYPE_PATH_MIG_FAILED },
+	{.nb.notifier_call = forward_event,   .event_type = MLX5_EVENT_TYPE_WQ_INVAL_REQ_ERROR },
+	{.nb.notifier_call = forward_event,   .event_type = MLX5_EVENT_TYPE_WQ_ACCESS_ERROR },
+	/* SRQ events */
+	{.nb.notifier_call = forward_event,   .event_type = MLX5_EVENT_TYPE_SRQ_CATAS_ERROR },
+	{.nb.notifier_call = forward_event,   .event_type = MLX5_EVENT_TYPE_SRQ_RQ_LIMIT },
+};
+
+struct mlx5_events {
+	struct mlx5_core_dev *dev;
+	struct workqueue_struct *wq;
+	struct mlx5_event_nb  notifiers[ARRAY_SIZE(events_nbs_ref)];
+	/* driver notifier chain for fw events */
+	struct atomic_notifier_head fw_nh;
+	/* port module events stats */
+	struct mlx5_pme_stats pme_stats;
+	/*pcie_core*/
+	struct work_struct pcie_core_work;
+	/* driver notifier chain for sw events */
+	struct blocking_notifier_head sw_nh;
+};
+
+static const char *eqe_type_str(u8 type)
+{
+	switch (type) {
+	case MLX5_EVENT_TYPE_COMP:
+		return "MLX5_EVENT_TYPE_COMP";
+	case MLX5_EVENT_TYPE_PATH_MIG:
+		return "MLX5_EVENT_TYPE_PATH_MIG";
+	case MLX5_EVENT_TYPE_COMM_EST:
+		return "MLX5_EVENT_TYPE_COMM_EST";
+	case MLX5_EVENT_TYPE_SQ_DRAINED:
+		return "MLX5_EVENT_TYPE_SQ_DRAINED";
+	case MLX5_EVENT_TYPE_SRQ_LAST_WQE:
+		return "MLX5_EVENT_TYPE_SRQ_LAST_WQE";
+	case MLX5_EVENT_TYPE_SRQ_RQ_LIMIT:
+		return "MLX5_EVENT_TYPE_SRQ_RQ_LIMIT";
+	case MLX5_EVENT_TYPE_CQ_ERROR:
+		return "MLX5_EVENT_TYPE_CQ_ERROR";
+	case MLX5_EVENT_TYPE_WQ_CATAS_ERROR:
+		return "MLX5_EVENT_TYPE_WQ_CATAS_ERROR";
+	case MLX5_EVENT_TYPE_PATH_MIG_FAILED:
+		return "MLX5_EVENT_TYPE_PATH_MIG_FAILED";
+	case MLX5_EVENT_TYPE_WQ_INVAL_REQ_ERROR:
+		return "MLX5_EVENT_TYPE_WQ_INVAL_REQ_ERROR";
+	case MLX5_EVENT_TYPE_WQ_ACCESS_ERROR:
+		return "MLX5_EVENT_TYPE_WQ_ACCESS_ERROR";
+	case MLX5_EVENT_TYPE_SRQ_CATAS_ERROR:
+		return "MLX5_EVENT_TYPE_SRQ_CATAS_ERROR";
+	case MLX5_EVENT_TYPE_INTERNAL_ERROR:
+		return "MLX5_EVENT_TYPE_INTERNAL_ERROR";
+	case MLX5_EVENT_TYPE_PORT_CHANGE:
+		return "MLX5_EVENT_TYPE_PORT_CHANGE";
+	case MLX5_EVENT_TYPE_GPIO_EVENT:
+		return "MLX5_EVENT_TYPE_GPIO_EVENT";
+	case MLX5_EVENT_TYPE_PORT_MODULE_EVENT:
+		return "MLX5_EVENT_TYPE_PORT_MODULE_EVENT";
+	case MLX5_EVENT_TYPE_TEMP_WARN_EVENT:
+		return "MLX5_EVENT_TYPE_TEMP_WARN_EVENT";
+	case MLX5_EVENT_TYPE_REMOTE_CONFIG:
+		return "MLX5_EVENT_TYPE_REMOTE_CONFIG";
+	case MLX5_EVENT_TYPE_DB_BF_CONGESTION:
+		return "MLX5_EVENT_TYPE_DB_BF_CONGESTION";
+	case MLX5_EVENT_TYPE_STALL_EVENT:
+		return "MLX5_EVENT_TYPE_STALL_EVENT";
+	case MLX5_EVENT_TYPE_CMD:
+		return "MLX5_EVENT_TYPE_CMD";
+	case MLX5_EVENT_TYPE_ESW_FUNCTIONS_CHANGED:
+		return "MLX5_EVENT_TYPE_ESW_FUNCTIONS_CHANGED";
+	case MLX5_EVENT_TYPE_VHCA_STATE_CHANGE:
+		return "MLX5_EVENT_TYPE_VHCA_STATE_CHANGE";
+	case MLX5_EVENT_TYPE_PAGE_REQUEST:
+		return "MLX5_EVENT_TYPE_PAGE_REQUEST";
+	case MLX5_EVENT_TYPE_PAGE_FAULT:
+		return "MLX5_EVENT_TYPE_PAGE_FAULT";
+	case MLX5_EVENT_TYPE_PPS_EVENT:
+		return "MLX5_EVENT_TYPE_PPS_EVENT";
+	case MLX5_EVENT_TYPE_NIC_VPORT_CHANGE:
+		return "MLX5_EVENT_TYPE_NIC_VPORT_CHANGE";
+	case MLX5_EVENT_TYPE_FPGA_ERROR:
+		return "MLX5_EVENT_TYPE_FPGA_ERROR";
+	case MLX5_EVENT_TYPE_FPGA_QP_ERROR:
+		return "MLX5_EVENT_TYPE_FPGA_QP_ERROR";
+	case MLX5_EVENT_TYPE_GENERAL_EVENT:
+		return "MLX5_EVENT_TYPE_GENERAL_EVENT";
+	case MLX5_EVENT_TYPE_MONITOR_COUNTER:
+		return "MLX5_EVENT_TYPE_MONITOR_COUNTER";
+	case MLX5_EVENT_TYPE_DEVICE_TRACER:
+		return "MLX5_EVENT_TYPE_DEVICE_TRACER";
+	case MLX5_EVENT_TYPE_OBJECT_CHANGE:
+		return "MLX5_EVENT_TYPE_OBJECT_CHANGE";
+	default:
+		return "Unrecognized event";
+	}
+}
+
+/* handles all FW events, type == eqe->type */
+static int any_notifier(struct notifier_block *nb,
+			unsigned long type, void *data)
+{
+	struct mlx5_event_nb *event_nb = mlx5_nb_cof(nb, struct mlx5_event_nb, nb);
+	struct mlx5_events   *events   = event_nb->ctx;
+	struct mlx5_eqe      *eqe      = data;
+
+	mlx5_core_dbg(events->dev, "Async eqe type %s, subtype (%d)\n",
+		      eqe_type_str(eqe->type), eqe->sub_type);
+	return NOTIFY_OK;
+}
+
+/* type == MLX5_EVENT_TYPE_TEMP_WARN_EVENT */
+static int temp_warn(struct notifier_block *nb, unsigned long type, void *data)
+{
+	struct mlx5_event_nb *event_nb = mlx5_nb_cof(nb, struct mlx5_event_nb, nb);
+	struct mlx5_events   *events   = event_nb->ctx;
+	struct mlx5_eqe      *eqe      = data;
+	u64 value_lsb;
+	u64 value_msb;
+
+	value_lsb = be64_to_cpu(eqe->data.temp_warning.sensor_warning_lsb);
+	value_msb = be64_to_cpu(eqe->data.temp_warning.sensor_warning_msb);
+
+	mlx5_core_warn(events->dev,
+		       "High temperature on sensors with bit set %llx %llx",
+		       value_msb, value_lsb);
+
+	return NOTIFY_OK;
+}
+
+/* MLX5_EVENT_TYPE_PORT_MODULE_EVENT */
+static const char *mlx5_pme_status_to_string(enum port_module_event_status_type status)
+{
+	switch (status) {
+	case MLX5_MODULE_STATUS_PLUGGED:
+		return "Cable plugged";
+	case MLX5_MODULE_STATUS_UNPLUGGED:
+		return "Cable unplugged";
+	case MLX5_MODULE_STATUS_ERROR:
+		return "Cable error";
+	case MLX5_MODULE_STATUS_DISABLED:
+		return "Cable disabled";
+	default:
+		return "Unknown status";
+	}
+}
+
+static const char *mlx5_pme_error_to_string(enum port_module_event_error_type error)
+{
+	switch (error) {
+	case MLX5_MODULE_EVENT_ERROR_POWER_BUDGET_EXCEEDED:
+		return "Power budget exceeded";
+	case MLX5_MODULE_EVENT_ERROR_LONG_RANGE_FOR_NON_MLNX:
+		return "Long Range for non MLNX cable";
+	case MLX5_MODULE_EVENT_ERROR_BUS_STUCK:
+		return "Bus stuck (I2C or data shorted)";
+	case MLX5_MODULE_EVENT_ERROR_NO_EEPROM_RETRY_TIMEOUT:
+		return "No EEPROM/retry timeout";
+	case MLX5_MODULE_EVENT_ERROR_ENFORCE_PART_NUMBER_LIST:
+		return "Enforce part number list";
+	case MLX5_MODULE_EVENT_ERROR_UNKNOWN_IDENTIFIER:
+		return "Unknown identifier";
+	case MLX5_MODULE_EVENT_ERROR_HIGH_TEMPERATURE:
+		return "High Temperature";
+	case MLX5_MODULE_EVENT_ERROR_BAD_CABLE:
+		return "Bad or shorted cable/module";
+	case MLX5_MODULE_EVENT_ERROR_PCIE_POWER_SLOT_EXCEEDED:
+		return "One or more network ports have been powered down due to insufficient/unadvertised power on the PCIe slot";
+	default:
+		return "Unknown error";
+	}
+}
+
+/* type == MLX5_EVENT_TYPE_PORT_MODULE_EVENT */
+static int port_module(struct notifier_block *nb, unsigned long type, void *data)
+{
+	struct mlx5_event_nb *event_nb = mlx5_nb_cof(nb, struct mlx5_event_nb, nb);
+	struct mlx5_events   *events   = event_nb->ctx;
+	struct mlx5_eqe      *eqe      = data;
+
+	enum port_module_event_status_type module_status;
+	enum port_module_event_error_type error_type;
+	struct mlx5_eqe_port_module *module_event_eqe;
+	const char *status_str;
+	u8 module_num;
+
+	module_event_eqe = &eqe->data.port_module;
+	module_status = module_event_eqe->module_status &
+			PORT_MODULE_EVENT_MODULE_STATUS_MASK;
+	error_type = module_event_eqe->error_type &
+		     PORT_MODULE_EVENT_ERROR_TYPE_MASK;
+
+	if (module_status < MLX5_MODULE_STATUS_NUM)
+		events->pme_stats.status_counters[module_status]++;
+
+	if (module_status == MLX5_MODULE_STATUS_ERROR)
+		if (error_type < MLX5_MODULE_EVENT_ERROR_NUM)
+			events->pme_stats.error_counters[error_type]++;
+
+	if (!printk_ratelimit())
+		return NOTIFY_OK;
+
+	module_num = module_event_eqe->module;
+	status_str = mlx5_pme_status_to_string(module_status);
+	if (module_status == MLX5_MODULE_STATUS_ERROR) {
+		const char *error_str = mlx5_pme_error_to_string(error_type);
+
+		mlx5_core_err(events->dev,
+			      "Port module event[error]: module %u, %s, %s\n",
+			      module_num, status_str, error_str);
+	} else {
+		mlx5_core_info(events->dev,
+			       "Port module event: module %u, %s\n",
+			       module_num, status_str);
+	}
+
+	return NOTIFY_OK;
+}
+
+enum {
+	MLX5_PCI_POWER_COULD_NOT_BE_READ = 0x0,
+	MLX5_PCI_POWER_SUFFICIENT_REPORTED = 0x1,
+	MLX5_PCI_POWER_INSUFFICIENT_REPORTED = 0x2,
+};
+
+static void mlx5_pcie_event(struct work_struct *work)
+{
+	u32 out[MLX5_ST_SZ_DW(mpein_reg)] = {0};
+	u32 in[MLX5_ST_SZ_DW(mpein_reg)] = {0};
+	struct mlx5_events *events;
+	struct mlx5_core_dev *dev;
+	u8 power_status;
+	u16 pci_power;
+
+	events = container_of(work, struct mlx5_events, pcie_core_work);
+	dev  = events->dev;
+
+	if (!MLX5_CAP_MCAM_FEATURE(dev, pci_status_and_power))
+		return;
+
+	mlx5_core_access_reg(dev, in, sizeof(in), out, sizeof(out),
+			     MLX5_REG_MPEIN, 0, 0);
+	power_status = MLX5_GET(mpein_reg, out, pwr_status);
+	pci_power = MLX5_GET(mpein_reg, out, pci_power);
+
+	switch (power_status) {
+	case MLX5_PCI_POWER_COULD_NOT_BE_READ:
+		mlx5_core_info_rl(dev,
+				  "PCIe slot power capability was not advertised.\n");
+		break;
+	case MLX5_PCI_POWER_INSUFFICIENT_REPORTED:
+		mlx5_core_warn_rl(dev,
+				  "Detected insufficient power on the PCIe slot (%uW).\n",
+				  pci_power);
+		break;
+	case MLX5_PCI_POWER_SUFFICIENT_REPORTED:
+		mlx5_core_info_rl(dev,
+				  "PCIe slot advertised sufficient power (%uW).\n",
+				  pci_power);
+		break;
+	}
+}
+
+static int pcie_core(struct notifier_block *nb, unsigned long type, void *data)
+{
+	struct mlx5_event_nb    *event_nb = mlx5_nb_cof(nb,
+							struct mlx5_event_nb,
+							nb);
+	struct mlx5_events      *events   = event_nb->ctx;
+	struct mlx5_eqe         *eqe      = data;
+
+	switch (eqe->sub_type) {
+	case MLX5_GENERAL_SUBTYPE_PCI_POWER_CHANGE_EVENT:
+			queue_work(events->wq, &events->pcie_core_work);
+		break;
+	default:
+		return NOTIFY_DONE;
+	}
+
+	return NOTIFY_OK;
+}
+
+void mlx5_get_pme_stats(struct mlx5_core_dev *dev, struct mlx5_pme_stats *stats)
+{
+	*stats = dev->priv.events->pme_stats;
+}
+
+/* forward event as is to registered interfaces (mlx5e/mlx5_ib) */
+static int forward_event(struct notifier_block *nb, unsigned long event, void *data)
+{
+	struct mlx5_event_nb *event_nb = mlx5_nb_cof(nb, struct mlx5_event_nb, nb);
+	struct mlx5_events   *events   = event_nb->ctx;
+	struct mlx5_eqe      *eqe      = data;
+
+	mlx5_core_dbg(events->dev, "Async eqe type %s, subtype (%d) forward to interfaces\n",
+		      eqe_type_str(eqe->type), eqe->sub_type);
+	atomic_notifier_call_chain(&events->fw_nh, event, data);
+	return NOTIFY_OK;
+}
+
+int mlx5_events_init(struct mlx5_core_dev *dev)
+{
+	struct mlx5_events *events = kzalloc(sizeof(*events), GFP_KERNEL);
+
+	if (!events)
+		return -ENOMEM;
+
+	ATOMIC_INIT_NOTIFIER_HEAD(&events->fw_nh);
+	events->dev = dev;
+	dev->priv.events = events;
+	events->wq = create_singlethread_workqueue("mlx5_events");
+	if (!events->wq) {
+		kfree(events);
+		return -ENOMEM;
+	}
+	INIT_WORK(&events->pcie_core_work, mlx5_pcie_event);
+	BLOCKING_INIT_NOTIFIER_HEAD(&events->sw_nh);
+
+	return 0;
+}
+
+void mlx5_events_cleanup(struct mlx5_core_dev *dev)
+{
+	destroy_workqueue(dev->priv.events->wq);
+	kvfree(dev->priv.events);
+}
+
+void mlx5_events_start(struct mlx5_core_dev *dev)
+{
+	struct mlx5_events *events = dev->priv.events;
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(events_nbs_ref); i++) {
+		events->notifiers[i].nb  = events_nbs_ref[i];
+		events->notifiers[i].ctx = events;
+		mlx5_eq_notifier_register(dev, &events->notifiers[i].nb);
+	}
+}
+
+void mlx5_events_stop(struct mlx5_core_dev *dev)
+{
+	struct mlx5_events *events = dev->priv.events;
+	int i;
+
+	for (i = ARRAY_SIZE(events_nbs_ref) - 1; i >= 0 ; i--)
+		mlx5_eq_notifier_unregister(dev, &events->notifiers[i].nb);
+	flush_workqueue(events->wq);
+}
+
+/* This API is used only for processing and forwarding firmware
+ * events to mlx5 consumer.
+ */
+int mlx5_notifier_register(struct mlx5_core_dev *dev, struct notifier_block *nb)
+{
+	struct mlx5_events *events = dev->priv.events;
+
+	return atomic_notifier_chain_register(&events->fw_nh, nb);
+}
+EXPORT_SYMBOL(mlx5_notifier_register);
+
+int mlx5_notifier_unregister(struct mlx5_core_dev *dev, struct notifier_block *nb)
+{
+	struct mlx5_events *events = dev->priv.events;
+
+	return atomic_notifier_chain_unregister(&events->fw_nh, nb);
+}
+EXPORT_SYMBOL(mlx5_notifier_unregister);
+
+int mlx5_notifier_call_chain(struct mlx5_events *events, unsigned int event, void *data)
+{
+	return atomic_notifier_call_chain(&events->fw_nh, event, data);
+}
+
+/* This API is used only for processing and forwarding driver-specific
+ * events to mlx5 consumers.
+ */
+int mlx5_blocking_notifier_register(struct mlx5_core_dev *dev, struct notifier_block *nb)
+{
+	struct mlx5_events *events = dev->priv.events;
+
+	return blocking_notifier_chain_register(&events->sw_nh, nb);
+}
+
+int mlx5_blocking_notifier_unregister(struct mlx5_core_dev *dev, struct notifier_block *nb)
+{
+	struct mlx5_events *events = dev->priv.events;
+
+	return blocking_notifier_chain_unregister(&events->sw_nh, nb);
+}
+
+int mlx5_blocking_notifier_call_chain(struct mlx5_core_dev *dev, unsigned int event,
+				      void *data)
+{
+	struct mlx5_events *events = dev->priv.events;
+
+	return blocking_notifier_call_chain(&events->sw_nh, event, data);
+}
+
+void mlx5_events_work_enqueue(struct mlx5_core_dev *dev, struct work_struct *work)
+{
+	queue_work(dev->priv.events->wq, work);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fpga/cmd.c b/drivers/net/ethernet/mellanox/mlx5/core/fpga/cmd.c
new file mode 100644
index 000000000..9a3707715
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fpga/cmd.c
@@ -0,0 +1,235 @@
+/*
+ * Copyright (c) 2017, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/etherdevice.h>
+#include <linux/mlx5/driver.h>
+#include <linux/mlx5/device.h>
+
+#include "mlx5_core.h"
+#include "fpga/cmd.h"
+
+#define MLX5_FPGA_ACCESS_REG_SZ (MLX5_ST_SZ_DW(fpga_access_reg) + \
+				 MLX5_FPGA_ACCESS_REG_SIZE_MAX)
+
+int mlx5_fpga_access_reg(struct mlx5_core_dev *dev, u8 size, u64 addr,
+			 void *buf, bool write)
+{
+	u32 in[MLX5_FPGA_ACCESS_REG_SZ] = {0};
+	u32 out[MLX5_FPGA_ACCESS_REG_SZ];
+	int err;
+
+	if (size & 3)
+		return -EINVAL;
+	if (addr & 3)
+		return -EINVAL;
+	if (size > MLX5_FPGA_ACCESS_REG_SIZE_MAX)
+		return -EINVAL;
+
+	MLX5_SET(fpga_access_reg, in, size, size);
+	MLX5_SET64(fpga_access_reg, in, address, addr);
+	if (write)
+		memcpy(MLX5_ADDR_OF(fpga_access_reg, in, data), buf, size);
+
+	err = mlx5_core_access_reg(dev, in, sizeof(in), out, sizeof(out),
+				   MLX5_REG_FPGA_ACCESS_REG, 0, write);
+	if (err)
+		return err;
+
+	if (!write)
+		memcpy(buf, MLX5_ADDR_OF(fpga_access_reg, out, data), size);
+
+	return 0;
+}
+
+int mlx5_fpga_caps(struct mlx5_core_dev *dev)
+{
+	u32 in[MLX5_ST_SZ_DW(fpga_cap)] = {0};
+
+	return mlx5_core_access_reg(dev, in, sizeof(in), dev->caps.fpga,
+				    MLX5_ST_SZ_BYTES(fpga_cap),
+				    MLX5_REG_FPGA_CAP, 0, 0);
+}
+
+int mlx5_fpga_ctrl_op(struct mlx5_core_dev *dev, u8 op)
+{
+	u32 in[MLX5_ST_SZ_DW(fpga_ctrl)] = {0};
+	u32 out[MLX5_ST_SZ_DW(fpga_ctrl)];
+
+	MLX5_SET(fpga_ctrl, in, operation, op);
+
+	return mlx5_core_access_reg(dev, in, sizeof(in), out, sizeof(out),
+				    MLX5_REG_FPGA_CTRL, 0, true);
+}
+
+int mlx5_fpga_sbu_caps(struct mlx5_core_dev *dev, void *caps, int size)
+{
+	unsigned int cap_size = MLX5_CAP_FPGA(dev, sandbox_extended_caps_len);
+	u64 addr = MLX5_CAP64_FPGA(dev, sandbox_extended_caps_addr);
+	unsigned int read;
+	int ret = 0;
+
+	if (cap_size > size) {
+		mlx5_core_warn(dev, "Not enough buffer %u for FPGA SBU caps %u",
+			       size, cap_size);
+		return -EINVAL;
+	}
+
+	while (cap_size > 0) {
+		read = min_t(unsigned int, cap_size,
+			     MLX5_FPGA_ACCESS_REG_SIZE_MAX);
+
+		ret = mlx5_fpga_access_reg(dev, read, addr, caps, false);
+		if (ret) {
+			mlx5_core_warn(dev, "Error reading FPGA SBU caps %u bytes at address 0x%llx: %d",
+				       read, addr, ret);
+			return ret;
+		}
+
+		cap_size -= read;
+		addr += read;
+		caps += read;
+	}
+
+	return ret;
+}
+
+int mlx5_fpga_query(struct mlx5_core_dev *dev, struct mlx5_fpga_query *query)
+{
+	u32 in[MLX5_ST_SZ_DW(fpga_ctrl)] = {0};
+	u32 out[MLX5_ST_SZ_DW(fpga_ctrl)];
+	int err;
+
+	err = mlx5_core_access_reg(dev, in, sizeof(in), out, sizeof(out),
+				   MLX5_REG_FPGA_CTRL, 0, false);
+	if (err)
+		return err;
+
+	query->status = MLX5_GET(fpga_ctrl, out, status);
+	query->admin_image = MLX5_GET(fpga_ctrl, out, flash_select_admin);
+	query->oper_image = MLX5_GET(fpga_ctrl, out, flash_select_oper);
+	return 0;
+}
+
+int mlx5_fpga_create_qp(struct mlx5_core_dev *dev, void *fpga_qpc,
+			u32 *fpga_qpn)
+{
+	u32 out[MLX5_ST_SZ_DW(fpga_create_qp_out)] = {};
+	u32 in[MLX5_ST_SZ_DW(fpga_create_qp_in)] = {};
+	int ret;
+
+	MLX5_SET(fpga_create_qp_in, in, opcode, MLX5_CMD_OP_FPGA_CREATE_QP);
+	memcpy(MLX5_ADDR_OF(fpga_create_qp_in, in, fpga_qpc), fpga_qpc,
+	       MLX5_FLD_SZ_BYTES(fpga_create_qp_in, fpga_qpc));
+
+	ret = mlx5_cmd_exec_inout(dev, fpga_create_qp, in, out);
+	if (ret)
+		return ret;
+
+	memcpy(fpga_qpc, MLX5_ADDR_OF(fpga_create_qp_out, out, fpga_qpc),
+	       MLX5_FLD_SZ_BYTES(fpga_create_qp_out, fpga_qpc));
+	*fpga_qpn = MLX5_GET(fpga_create_qp_out, out, fpga_qpn);
+	return ret;
+}
+
+int mlx5_fpga_modify_qp(struct mlx5_core_dev *dev, u32 fpga_qpn,
+			enum mlx5_fpga_qpc_field_select fields,
+			void *fpga_qpc)
+{
+	u32 in[MLX5_ST_SZ_DW(fpga_modify_qp_in)] = {};
+
+	MLX5_SET(fpga_modify_qp_in, in, opcode, MLX5_CMD_OP_FPGA_MODIFY_QP);
+	MLX5_SET(fpga_modify_qp_in, in, field_select, fields);
+	MLX5_SET(fpga_modify_qp_in, in, fpga_qpn, fpga_qpn);
+	memcpy(MLX5_ADDR_OF(fpga_modify_qp_in, in, fpga_qpc), fpga_qpc,
+	       MLX5_FLD_SZ_BYTES(fpga_modify_qp_in, fpga_qpc));
+
+	return mlx5_cmd_exec_in(dev, fpga_modify_qp, in);
+}
+
+int mlx5_fpga_query_qp(struct mlx5_core_dev *dev,
+		       u32 fpga_qpn, void *fpga_qpc)
+{
+	u32 out[MLX5_ST_SZ_DW(fpga_query_qp_out)] = {};
+	u32 in[MLX5_ST_SZ_DW(fpga_query_qp_in)] = {};
+	int ret;
+
+	MLX5_SET(fpga_query_qp_in, in, opcode, MLX5_CMD_OP_FPGA_QUERY_QP);
+	MLX5_SET(fpga_query_qp_in, in, fpga_qpn, fpga_qpn);
+
+	ret = mlx5_cmd_exec_inout(dev, fpga_query_qp, in, out);
+	if (ret)
+		return ret;
+
+	memcpy(fpga_qpc, MLX5_ADDR_OF(fpga_query_qp_out, out, fpga_qpc),
+	       MLX5_FLD_SZ_BYTES(fpga_query_qp_out, fpga_qpc));
+	return ret;
+}
+
+int mlx5_fpga_destroy_qp(struct mlx5_core_dev *dev, u32 fpga_qpn)
+{
+	u32 in[MLX5_ST_SZ_DW(fpga_destroy_qp_in)] = {};
+
+	MLX5_SET(fpga_destroy_qp_in, in, opcode, MLX5_CMD_OP_FPGA_DESTROY_QP);
+	MLX5_SET(fpga_destroy_qp_in, in, fpga_qpn, fpga_qpn);
+
+	return mlx5_cmd_exec_in(dev, fpga_destroy_qp, in);
+}
+
+int mlx5_fpga_query_qp_counters(struct mlx5_core_dev *dev, u32 fpga_qpn,
+				bool clear, struct mlx5_fpga_qp_counters *data)
+{
+	u32 out[MLX5_ST_SZ_DW(fpga_query_qp_counters_out)] = {};
+	u32 in[MLX5_ST_SZ_DW(fpga_query_qp_counters_in)] = {};
+	int ret;
+
+	MLX5_SET(fpga_query_qp_counters_in, in, opcode,
+		 MLX5_CMD_OP_FPGA_QUERY_QP_COUNTERS);
+	MLX5_SET(fpga_query_qp_counters_in, in, clear, clear);
+	MLX5_SET(fpga_query_qp_counters_in, in, fpga_qpn, fpga_qpn);
+
+	ret = mlx5_cmd_exec_inout(dev, fpga_query_qp_counters, in, out);
+	if (ret)
+		return ret;
+
+	data->rx_ack_packets = MLX5_GET64(fpga_query_qp_counters_out, out,
+					  rx_ack_packets);
+	data->rx_send_packets = MLX5_GET64(fpga_query_qp_counters_out, out,
+					   rx_send_packets);
+	data->tx_ack_packets = MLX5_GET64(fpga_query_qp_counters_out, out,
+					  tx_ack_packets);
+	data->tx_send_packets = MLX5_GET64(fpga_query_qp_counters_out, out,
+					   tx_send_packets);
+	data->rx_total_drop = MLX5_GET64(fpga_query_qp_counters_out, out,
+					 rx_total_drop);
+
+	return ret;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fpga/cmd.h b/drivers/net/ethernet/mellanox/mlx5/core/fpga/cmd.h
new file mode 100644
index 000000000..11621d265
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fpga/cmd.h
@@ -0,0 +1,91 @@
+/*
+ * Copyright (c) 2017, Mellanox Technologies, Ltd.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef __MLX5_FPGA_H__
+#define __MLX5_FPGA_H__
+
+#include <linux/mlx5/driver.h>
+
+enum mlx5_fpga_id {
+	MLX5_FPGA_NEWTON = 0,
+	MLX5_FPGA_EDISON = 1,
+	MLX5_FPGA_MORSE = 2,
+	MLX5_FPGA_MORSEQ = 3,
+};
+
+enum mlx5_fpga_image {
+	MLX5_FPGA_IMAGE_USER = 0,
+	MLX5_FPGA_IMAGE_FACTORY,
+};
+
+enum mlx5_fpga_status {
+	MLX5_FPGA_STATUS_SUCCESS = 0,
+	MLX5_FPGA_STATUS_FAILURE = 1,
+	MLX5_FPGA_STATUS_IN_PROGRESS = 2,
+	MLX5_FPGA_STATUS_NONE = 0xFFFF,
+};
+
+struct mlx5_fpga_query {
+	enum mlx5_fpga_image admin_image;
+	enum mlx5_fpga_image oper_image;
+	enum mlx5_fpga_status status;
+};
+
+enum mlx5_fpga_qpc_field_select {
+	MLX5_FPGA_QPC_STATE = BIT(0),
+};
+
+struct mlx5_fpga_qp_counters {
+	u64 rx_ack_packets;
+	u64 rx_send_packets;
+	u64 tx_ack_packets;
+	u64 tx_send_packets;
+	u64 rx_total_drop;
+};
+
+int mlx5_fpga_caps(struct mlx5_core_dev *dev);
+int mlx5_fpga_query(struct mlx5_core_dev *dev, struct mlx5_fpga_query *query);
+int mlx5_fpga_ctrl_op(struct mlx5_core_dev *dev, u8 op);
+int mlx5_fpga_access_reg(struct mlx5_core_dev *dev, u8 size, u64 addr,
+			 void *buf, bool write);
+int mlx5_fpga_sbu_caps(struct mlx5_core_dev *dev, void *caps, int size);
+
+int mlx5_fpga_create_qp(struct mlx5_core_dev *dev, void *fpga_qpc,
+			u32 *fpga_qpn);
+int mlx5_fpga_modify_qp(struct mlx5_core_dev *dev, u32 fpga_qpn,
+			enum mlx5_fpga_qpc_field_select fields, void *fpga_qpc);
+int mlx5_fpga_query_qp(struct mlx5_core_dev *dev, u32 fpga_qpn, void *fpga_qpc);
+int mlx5_fpga_query_qp_counters(struct mlx5_core_dev *dev, u32 fpga_qpn,
+				bool clear, struct mlx5_fpga_qp_counters *data);
+int mlx5_fpga_destroy_qp(struct mlx5_core_dev *dev, u32 fpga_qpn);
+
+#endif /* __MLX5_FPGA_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fpga/conn.c b/drivers/net/ethernet/mellanox/mlx5/core/fpga/conn.c
new file mode 100644
index 000000000..12abe9915
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fpga/conn.c
@@ -0,0 +1,1001 @@
+/*
+ * Copyright (c) 2017 Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+
+#include <net/addrconf.h>
+#include <linux/etherdevice.h>
+#include <linux/mlx5/vport.h>
+
+#include "mlx5_core.h"
+#include "lib/mlx5.h"
+#include "fpga/conn.h"
+
+#define MLX5_FPGA_PKEY 0xFFFF
+#define MLX5_FPGA_PKEY_INDEX 0 /* RoCE PKEY 0xFFFF is always at index 0 */
+#define MLX5_FPGA_RECV_SIZE 2048
+#define MLX5_FPGA_PORT_NUM 1
+#define MLX5_FPGA_CQ_BUDGET 64
+
+static int mlx5_fpga_conn_map_buf(struct mlx5_fpga_conn *conn,
+				  struct mlx5_fpga_dma_buf *buf)
+{
+	struct device *dma_device;
+	int err = 0;
+
+	if (unlikely(!buf->sg[0].data))
+		goto out;
+
+	dma_device = mlx5_core_dma_dev(conn->fdev->mdev);
+	buf->sg[0].dma_addr = dma_map_single(dma_device, buf->sg[0].data,
+					     buf->sg[0].size, buf->dma_dir);
+	err = dma_mapping_error(dma_device, buf->sg[0].dma_addr);
+	if (unlikely(err)) {
+		mlx5_fpga_warn(conn->fdev, "DMA error on sg 0: %d\n", err);
+		err = -ENOMEM;
+		goto out;
+	}
+
+	if (!buf->sg[1].data)
+		goto out;
+
+	buf->sg[1].dma_addr = dma_map_single(dma_device, buf->sg[1].data,
+					     buf->sg[1].size, buf->dma_dir);
+	err = dma_mapping_error(dma_device, buf->sg[1].dma_addr);
+	if (unlikely(err)) {
+		mlx5_fpga_warn(conn->fdev, "DMA error on sg 1: %d\n", err);
+		dma_unmap_single(dma_device, buf->sg[0].dma_addr,
+				 buf->sg[0].size, buf->dma_dir);
+		err = -ENOMEM;
+	}
+
+out:
+	return err;
+}
+
+static void mlx5_fpga_conn_unmap_buf(struct mlx5_fpga_conn *conn,
+				     struct mlx5_fpga_dma_buf *buf)
+{
+	struct device *dma_device;
+
+	dma_device = mlx5_core_dma_dev(conn->fdev->mdev);
+	if (buf->sg[1].data)
+		dma_unmap_single(dma_device, buf->sg[1].dma_addr,
+				 buf->sg[1].size, buf->dma_dir);
+
+	if (likely(buf->sg[0].data))
+		dma_unmap_single(dma_device, buf->sg[0].dma_addr,
+				 buf->sg[0].size, buf->dma_dir);
+}
+
+static int mlx5_fpga_conn_post_recv(struct mlx5_fpga_conn *conn,
+				    struct mlx5_fpga_dma_buf *buf)
+{
+	struct mlx5_wqe_data_seg *data;
+	unsigned int ix;
+	int err = 0;
+
+	err = mlx5_fpga_conn_map_buf(conn, buf);
+	if (unlikely(err))
+		goto out;
+
+	if (unlikely(conn->qp.rq.pc - conn->qp.rq.cc >= conn->qp.rq.size)) {
+		mlx5_fpga_conn_unmap_buf(conn, buf);
+		return -EBUSY;
+	}
+
+	ix = conn->qp.rq.pc & (conn->qp.rq.size - 1);
+	data = mlx5_wq_cyc_get_wqe(&conn->qp.wq.rq, ix);
+	data->byte_count = cpu_to_be32(buf->sg[0].size);
+	data->lkey = cpu_to_be32(conn->fdev->conn_res.mkey);
+	data->addr = cpu_to_be64(buf->sg[0].dma_addr);
+
+	conn->qp.rq.pc++;
+	conn->qp.rq.bufs[ix] = buf;
+
+	/* Make sure that descriptors are written before doorbell record. */
+	dma_wmb();
+	*conn->qp.wq.rq.db = cpu_to_be32(conn->qp.rq.pc & 0xffff);
+out:
+	return err;
+}
+
+static void mlx5_fpga_conn_notify_hw(struct mlx5_fpga_conn *conn, void *wqe)
+{
+	/* ensure wqe is visible to device before updating doorbell record */
+	dma_wmb();
+	*conn->qp.wq.sq.db = cpu_to_be32(conn->qp.sq.pc);
+	/* Make sure that doorbell record is visible before ringing */
+	wmb();
+	mlx5_write64(wqe, conn->fdev->conn_res.uar->map + MLX5_BF_OFFSET);
+}
+
+static void mlx5_fpga_conn_post_send(struct mlx5_fpga_conn *conn,
+				     struct mlx5_fpga_dma_buf *buf)
+{
+	struct mlx5_wqe_ctrl_seg *ctrl;
+	struct mlx5_wqe_data_seg *data;
+	unsigned int ix, sgi;
+	int size = 1;
+
+	ix = conn->qp.sq.pc & (conn->qp.sq.size - 1);
+
+	ctrl = mlx5_wq_cyc_get_wqe(&conn->qp.wq.sq, ix);
+	data = (void *)(ctrl + 1);
+
+	for (sgi = 0; sgi < ARRAY_SIZE(buf->sg); sgi++) {
+		if (!buf->sg[sgi].data)
+			break;
+		data->byte_count = cpu_to_be32(buf->sg[sgi].size);
+		data->lkey = cpu_to_be32(conn->fdev->conn_res.mkey);
+		data->addr = cpu_to_be64(buf->sg[sgi].dma_addr);
+		data++;
+		size++;
+	}
+
+	ctrl->imm = 0;
+	ctrl->fm_ce_se = MLX5_WQE_CTRL_CQ_UPDATE;
+	ctrl->opmod_idx_opcode = cpu_to_be32(((conn->qp.sq.pc & 0xffff) << 8) |
+					     MLX5_OPCODE_SEND);
+	ctrl->qpn_ds = cpu_to_be32(size | (conn->qp.qpn << 8));
+
+	conn->qp.sq.pc++;
+	conn->qp.sq.bufs[ix] = buf;
+	mlx5_fpga_conn_notify_hw(conn, ctrl);
+}
+
+int mlx5_fpga_conn_send(struct mlx5_fpga_conn *conn,
+			struct mlx5_fpga_dma_buf *buf)
+{
+	unsigned long flags;
+	int err;
+
+	if (!conn->qp.active)
+		return -ENOTCONN;
+
+	buf->dma_dir = DMA_TO_DEVICE;
+	err = mlx5_fpga_conn_map_buf(conn, buf);
+	if (err)
+		return err;
+
+	spin_lock_irqsave(&conn->qp.sq.lock, flags);
+
+	if (conn->qp.sq.pc - conn->qp.sq.cc >= conn->qp.sq.size) {
+		list_add_tail(&buf->list, &conn->qp.sq.backlog);
+		goto out_unlock;
+	}
+
+	mlx5_fpga_conn_post_send(conn, buf);
+
+out_unlock:
+	spin_unlock_irqrestore(&conn->qp.sq.lock, flags);
+	return err;
+}
+
+static int mlx5_fpga_conn_post_recv_buf(struct mlx5_fpga_conn *conn)
+{
+	struct mlx5_fpga_dma_buf *buf;
+	int err;
+
+	buf = kzalloc(sizeof(*buf) + MLX5_FPGA_RECV_SIZE, 0);
+	if (!buf)
+		return -ENOMEM;
+
+	buf->sg[0].data = (void *)(buf + 1);
+	buf->sg[0].size = MLX5_FPGA_RECV_SIZE;
+	buf->dma_dir = DMA_FROM_DEVICE;
+
+	err = mlx5_fpga_conn_post_recv(conn, buf);
+	if (err)
+		kfree(buf);
+
+	return err;
+}
+
+static int mlx5_fpga_conn_create_mkey(struct mlx5_core_dev *mdev, u32 pdn,
+				      u32 *mkey)
+{
+	int inlen = MLX5_ST_SZ_BYTES(create_mkey_in);
+	void *mkc;
+	u32 *in;
+	int err;
+
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in)
+		return -ENOMEM;
+
+	mkc = MLX5_ADDR_OF(create_mkey_in, in, memory_key_mkey_entry);
+	MLX5_SET(mkc, mkc, access_mode_1_0, MLX5_MKC_ACCESS_MODE_PA);
+	MLX5_SET(mkc, mkc, lw, 1);
+	MLX5_SET(mkc, mkc, lr, 1);
+
+	MLX5_SET(mkc, mkc, pd, pdn);
+	MLX5_SET(mkc, mkc, length64, 1);
+	MLX5_SET(mkc, mkc, qpn, 0xffffff);
+
+	err = mlx5_core_create_mkey(mdev, mkey, in, inlen);
+
+	kvfree(in);
+	return err;
+}
+
+static void mlx5_fpga_conn_rq_cqe(struct mlx5_fpga_conn *conn,
+				  struct mlx5_cqe64 *cqe, u8 status)
+{
+	struct mlx5_fpga_dma_buf *buf;
+	int ix, err;
+
+	ix = be16_to_cpu(cqe->wqe_counter) & (conn->qp.rq.size - 1);
+	buf = conn->qp.rq.bufs[ix];
+	conn->qp.rq.bufs[ix] = NULL;
+	conn->qp.rq.cc++;
+
+	if (unlikely(status && (status != MLX5_CQE_SYNDROME_WR_FLUSH_ERR)))
+		mlx5_fpga_warn(conn->fdev, "RQ buf %p on FPGA QP %u completion status %d\n",
+			       buf, conn->fpga_qpn, status);
+	else
+		mlx5_fpga_dbg(conn->fdev, "RQ buf %p on FPGA QP %u completion status %d\n",
+			      buf, conn->fpga_qpn, status);
+
+	mlx5_fpga_conn_unmap_buf(conn, buf);
+
+	if (unlikely(status || !conn->qp.active)) {
+		conn->qp.active = false;
+		kfree(buf);
+		return;
+	}
+
+	buf->sg[0].size = be32_to_cpu(cqe->byte_cnt);
+	mlx5_fpga_dbg(conn->fdev, "Message with %u bytes received successfully\n",
+		      buf->sg[0].size);
+	conn->recv_cb(conn->cb_arg, buf);
+
+	buf->sg[0].size = MLX5_FPGA_RECV_SIZE;
+	err = mlx5_fpga_conn_post_recv(conn, buf);
+	if (unlikely(err)) {
+		mlx5_fpga_warn(conn->fdev,
+			       "Failed to re-post recv buf: %d\n", err);
+		kfree(buf);
+	}
+}
+
+static void mlx5_fpga_conn_sq_cqe(struct mlx5_fpga_conn *conn,
+				  struct mlx5_cqe64 *cqe, u8 status)
+{
+	struct mlx5_fpga_dma_buf *buf, *nextbuf;
+	unsigned long flags;
+	int ix;
+
+	spin_lock_irqsave(&conn->qp.sq.lock, flags);
+
+	ix = be16_to_cpu(cqe->wqe_counter) & (conn->qp.sq.size - 1);
+	buf = conn->qp.sq.bufs[ix];
+	conn->qp.sq.bufs[ix] = NULL;
+	conn->qp.sq.cc++;
+
+	/* Handle backlog still under the spinlock to ensure message post order */
+	if (unlikely(!list_empty(&conn->qp.sq.backlog))) {
+		if (likely(conn->qp.active)) {
+			nextbuf = list_first_entry(&conn->qp.sq.backlog,
+						   struct mlx5_fpga_dma_buf, list);
+			list_del(&nextbuf->list);
+			mlx5_fpga_conn_post_send(conn, nextbuf);
+		}
+	}
+
+	spin_unlock_irqrestore(&conn->qp.sq.lock, flags);
+
+	if (unlikely(status && (status != MLX5_CQE_SYNDROME_WR_FLUSH_ERR)))
+		mlx5_fpga_warn(conn->fdev, "SQ buf %p on FPGA QP %u completion status %d\n",
+			       buf, conn->fpga_qpn, status);
+	else
+		mlx5_fpga_dbg(conn->fdev, "SQ buf %p on FPGA QP %u completion status %d\n",
+			      buf, conn->fpga_qpn, status);
+
+	mlx5_fpga_conn_unmap_buf(conn, buf);
+
+	if (likely(buf->complete))
+		buf->complete(conn, conn->fdev, buf, status);
+
+	if (unlikely(status))
+		conn->qp.active = false;
+}
+
+static void mlx5_fpga_conn_handle_cqe(struct mlx5_fpga_conn *conn,
+				      struct mlx5_cqe64 *cqe)
+{
+	u8 opcode, status = 0;
+
+	opcode = get_cqe_opcode(cqe);
+
+	switch (opcode) {
+	case MLX5_CQE_REQ_ERR:
+		status = ((struct mlx5_err_cqe *)cqe)->syndrome;
+		fallthrough;
+	case MLX5_CQE_REQ:
+		mlx5_fpga_conn_sq_cqe(conn, cqe, status);
+		break;
+
+	case MLX5_CQE_RESP_ERR:
+		status = ((struct mlx5_err_cqe *)cqe)->syndrome;
+		fallthrough;
+	case MLX5_CQE_RESP_SEND:
+		mlx5_fpga_conn_rq_cqe(conn, cqe, status);
+		break;
+	default:
+		mlx5_fpga_warn(conn->fdev, "Unexpected cqe opcode %u\n",
+			       opcode);
+	}
+}
+
+static void mlx5_fpga_conn_arm_cq(struct mlx5_fpga_conn *conn)
+{
+	mlx5_cq_arm(&conn->cq.mcq, MLX5_CQ_DB_REQ_NOT,
+		    conn->fdev->conn_res.uar->map, conn->cq.wq.cc);
+}
+
+static inline void mlx5_fpga_conn_cqes(struct mlx5_fpga_conn *conn,
+				       unsigned int budget)
+{
+	struct mlx5_cqe64 *cqe;
+
+	while (budget) {
+		cqe = mlx5_cqwq_get_cqe(&conn->cq.wq);
+		if (!cqe)
+			break;
+
+		budget--;
+		mlx5_cqwq_pop(&conn->cq.wq);
+		mlx5_fpga_conn_handle_cqe(conn, cqe);
+		mlx5_cqwq_update_db_record(&conn->cq.wq);
+	}
+	if (!budget) {
+		tasklet_schedule(&conn->cq.tasklet);
+		return;
+	}
+
+	mlx5_fpga_dbg(conn->fdev, "Re-arming CQ with cc# %u\n", conn->cq.wq.cc);
+	/* ensure cq space is freed before enabling more cqes */
+	wmb();
+	mlx5_fpga_conn_arm_cq(conn);
+}
+
+static void mlx5_fpga_conn_cq_tasklet(struct tasklet_struct *t)
+{
+	struct mlx5_fpga_conn *conn = from_tasklet(conn, t, cq.tasklet);
+
+	if (unlikely(!conn->qp.active))
+		return;
+	mlx5_fpga_conn_cqes(conn, MLX5_FPGA_CQ_BUDGET);
+}
+
+static void mlx5_fpga_conn_cq_complete(struct mlx5_core_cq *mcq,
+				       struct mlx5_eqe *eqe)
+{
+	struct mlx5_fpga_conn *conn;
+
+	conn = container_of(mcq, struct mlx5_fpga_conn, cq.mcq);
+	if (unlikely(!conn->qp.active))
+		return;
+	mlx5_fpga_conn_cqes(conn, MLX5_FPGA_CQ_BUDGET);
+}
+
+static int mlx5_fpga_conn_create_cq(struct mlx5_fpga_conn *conn, int cq_size)
+{
+	struct mlx5_fpga_device *fdev = conn->fdev;
+	struct mlx5_core_dev *mdev = fdev->mdev;
+	u32 temp_cqc[MLX5_ST_SZ_DW(cqc)] = {0};
+	u32 out[MLX5_ST_SZ_DW(create_cq_out)];
+	struct mlx5_wq_param wqp;
+	struct mlx5_cqe64 *cqe;
+	int inlen, err, eqn;
+	void *cqc, *in;
+	__be64 *pas;
+	u32 i;
+
+	cq_size = roundup_pow_of_two(cq_size);
+	MLX5_SET(cqc, temp_cqc, log_cq_size, ilog2(cq_size));
+
+	wqp.buf_numa_node = mdev->priv.numa_node;
+	wqp.db_numa_node  = mdev->priv.numa_node;
+
+	err = mlx5_cqwq_create(mdev, &wqp, temp_cqc, &conn->cq.wq,
+			       &conn->cq.wq_ctrl);
+	if (err)
+		return err;
+
+	for (i = 0; i < mlx5_cqwq_get_size(&conn->cq.wq); i++) {
+		cqe = mlx5_cqwq_get_wqe(&conn->cq.wq, i);
+		cqe->op_own = MLX5_CQE_INVALID << 4 | MLX5_CQE_OWNER_MASK;
+	}
+
+	inlen = MLX5_ST_SZ_BYTES(create_cq_in) +
+		sizeof(u64) * conn->cq.wq_ctrl.buf.npages;
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in) {
+		err = -ENOMEM;
+		goto err_cqwq;
+	}
+
+	err = mlx5_vector2eqn(mdev, smp_processor_id(), &eqn);
+	if (err) {
+		kvfree(in);
+		goto err_cqwq;
+	}
+
+	cqc = MLX5_ADDR_OF(create_cq_in, in, cq_context);
+	MLX5_SET(cqc, cqc, log_cq_size, ilog2(cq_size));
+	MLX5_SET(cqc, cqc, c_eqn_or_apu_element, eqn);
+	MLX5_SET(cqc, cqc, uar_page, fdev->conn_res.uar->index);
+	MLX5_SET(cqc, cqc, log_page_size, conn->cq.wq_ctrl.buf.page_shift -
+			   MLX5_ADAPTER_PAGE_SHIFT);
+	MLX5_SET64(cqc, cqc, dbr_addr, conn->cq.wq_ctrl.db.dma);
+
+	pas = (__be64 *)MLX5_ADDR_OF(create_cq_in, in, pas);
+	mlx5_fill_page_frag_array(&conn->cq.wq_ctrl.buf, pas);
+
+	err = mlx5_core_create_cq(mdev, &conn->cq.mcq, in, inlen, out, sizeof(out));
+	kvfree(in);
+
+	if (err)
+		goto err_cqwq;
+
+	conn->cq.mcq.cqe_sz     = 64;
+	conn->cq.mcq.set_ci_db  = conn->cq.wq_ctrl.db.db;
+	conn->cq.mcq.arm_db     = conn->cq.wq_ctrl.db.db + 1;
+	*conn->cq.mcq.set_ci_db = 0;
+	*conn->cq.mcq.arm_db    = 0;
+	conn->cq.mcq.vector     = 0;
+	conn->cq.mcq.comp       = mlx5_fpga_conn_cq_complete;
+	conn->cq.mcq.uar        = fdev->conn_res.uar;
+	tasklet_setup(&conn->cq.tasklet, mlx5_fpga_conn_cq_tasklet);
+
+	mlx5_fpga_dbg(fdev, "Created CQ #0x%x\n", conn->cq.mcq.cqn);
+
+	goto out;
+
+err_cqwq:
+	mlx5_wq_destroy(&conn->cq.wq_ctrl);
+out:
+	return err;
+}
+
+static void mlx5_fpga_conn_destroy_cq(struct mlx5_fpga_conn *conn)
+{
+	tasklet_disable(&conn->cq.tasklet);
+	tasklet_kill(&conn->cq.tasklet);
+	mlx5_core_destroy_cq(conn->fdev->mdev, &conn->cq.mcq);
+	mlx5_wq_destroy(&conn->cq.wq_ctrl);
+}
+
+static int mlx5_fpga_conn_create_wq(struct mlx5_fpga_conn *conn, void *qpc)
+{
+	struct mlx5_fpga_device *fdev = conn->fdev;
+	struct mlx5_core_dev *mdev = fdev->mdev;
+	struct mlx5_wq_param wqp;
+
+	wqp.buf_numa_node = mdev->priv.numa_node;
+	wqp.db_numa_node  = mdev->priv.numa_node;
+
+	return mlx5_wq_qp_create(mdev, &wqp, qpc, &conn->qp.wq,
+				 &conn->qp.wq_ctrl);
+}
+
+static int mlx5_fpga_conn_create_qp(struct mlx5_fpga_conn *conn,
+				    unsigned int tx_size, unsigned int rx_size)
+{
+	struct mlx5_fpga_device *fdev = conn->fdev;
+	u32 out[MLX5_ST_SZ_DW(create_qp_out)] = {};
+	struct mlx5_core_dev *mdev = fdev->mdev;
+	u32 temp_qpc[MLX5_ST_SZ_DW(qpc)] = {};
+	void *in = NULL, *qpc;
+	int err, inlen;
+
+	conn->qp.rq.pc = 0;
+	conn->qp.rq.cc = 0;
+	conn->qp.rq.size = roundup_pow_of_two(rx_size);
+	conn->qp.sq.pc = 0;
+	conn->qp.sq.cc = 0;
+	conn->qp.sq.size = roundup_pow_of_two(tx_size);
+
+	MLX5_SET(qpc, temp_qpc, log_rq_stride, ilog2(MLX5_SEND_WQE_DS) - 4);
+	MLX5_SET(qpc, temp_qpc, log_rq_size, ilog2(conn->qp.rq.size));
+	MLX5_SET(qpc, temp_qpc, log_sq_size, ilog2(conn->qp.sq.size));
+	err = mlx5_fpga_conn_create_wq(conn, temp_qpc);
+	if (err)
+		goto out;
+
+	conn->qp.rq.bufs = kvcalloc(conn->qp.rq.size,
+				    sizeof(conn->qp.rq.bufs[0]),
+				    GFP_KERNEL);
+	if (!conn->qp.rq.bufs) {
+		err = -ENOMEM;
+		goto err_wq;
+	}
+
+	conn->qp.sq.bufs = kvcalloc(conn->qp.sq.size,
+				    sizeof(conn->qp.sq.bufs[0]),
+				    GFP_KERNEL);
+	if (!conn->qp.sq.bufs) {
+		err = -ENOMEM;
+		goto err_rq_bufs;
+	}
+
+	inlen = MLX5_ST_SZ_BYTES(create_qp_in) +
+		MLX5_FLD_SZ_BYTES(create_qp_in, pas[0]) *
+		conn->qp.wq_ctrl.buf.npages;
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in) {
+		err = -ENOMEM;
+		goto err_sq_bufs;
+	}
+
+	qpc = MLX5_ADDR_OF(create_qp_in, in, qpc);
+	MLX5_SET(qpc, qpc, uar_page, fdev->conn_res.uar->index);
+	MLX5_SET(qpc, qpc, log_page_size,
+		 conn->qp.wq_ctrl.buf.page_shift - MLX5_ADAPTER_PAGE_SHIFT);
+	MLX5_SET(qpc, qpc, fre, 1);
+	MLX5_SET(qpc, qpc, rlky, 1);
+	MLX5_SET(qpc, qpc, st, MLX5_QP_ST_RC);
+	MLX5_SET(qpc, qpc, pm_state, MLX5_QP_PM_MIGRATED);
+	MLX5_SET(qpc, qpc, pd, fdev->conn_res.pdn);
+	MLX5_SET(qpc, qpc, log_rq_stride, ilog2(MLX5_SEND_WQE_DS) - 4);
+	MLX5_SET(qpc, qpc, log_rq_size, ilog2(conn->qp.rq.size));
+	MLX5_SET(qpc, qpc, rq_type, MLX5_NON_ZERO_RQ);
+	MLX5_SET(qpc, qpc, log_sq_size, ilog2(conn->qp.sq.size));
+	MLX5_SET(qpc, qpc, cqn_snd, conn->cq.mcq.cqn);
+	MLX5_SET(qpc, qpc, cqn_rcv, conn->cq.mcq.cqn);
+	MLX5_SET(qpc, qpc, ts_format, mlx5_get_qp_default_ts(mdev));
+	MLX5_SET64(qpc, qpc, dbr_addr, conn->qp.wq_ctrl.db.dma);
+	if (MLX5_CAP_GEN(mdev, cqe_version) == 1)
+		MLX5_SET(qpc, qpc, user_index, 0xFFFFFF);
+
+	mlx5_fill_page_frag_array(&conn->qp.wq_ctrl.buf,
+				  (__be64 *)MLX5_ADDR_OF(create_qp_in, in, pas));
+
+	MLX5_SET(create_qp_in, in, opcode, MLX5_CMD_OP_CREATE_QP);
+	err = mlx5_cmd_exec(mdev, in, inlen, out, sizeof(out));
+	if (err)
+		goto err_sq_bufs;
+
+	conn->qp.qpn = MLX5_GET(create_qp_out, out, qpn);
+	mlx5_fpga_dbg(fdev, "Created QP #0x%x\n", conn->qp.qpn);
+
+	goto out;
+
+err_sq_bufs:
+	kvfree(conn->qp.sq.bufs);
+err_rq_bufs:
+	kvfree(conn->qp.rq.bufs);
+err_wq:
+	mlx5_wq_destroy(&conn->qp.wq_ctrl);
+out:
+	kvfree(in);
+	return err;
+}
+
+static void mlx5_fpga_conn_free_recv_bufs(struct mlx5_fpga_conn *conn)
+{
+	int ix;
+
+	for (ix = 0; ix < conn->qp.rq.size; ix++) {
+		if (!conn->qp.rq.bufs[ix])
+			continue;
+		mlx5_fpga_conn_unmap_buf(conn, conn->qp.rq.bufs[ix]);
+		kfree(conn->qp.rq.bufs[ix]);
+		conn->qp.rq.bufs[ix] = NULL;
+	}
+}
+
+static void mlx5_fpga_conn_flush_send_bufs(struct mlx5_fpga_conn *conn)
+{
+	struct mlx5_fpga_dma_buf *buf, *temp;
+	int ix;
+
+	for (ix = 0; ix < conn->qp.sq.size; ix++) {
+		buf = conn->qp.sq.bufs[ix];
+		if (!buf)
+			continue;
+		conn->qp.sq.bufs[ix] = NULL;
+		mlx5_fpga_conn_unmap_buf(conn, buf);
+		if (!buf->complete)
+			continue;
+		buf->complete(conn, conn->fdev, buf, MLX5_CQE_SYNDROME_WR_FLUSH_ERR);
+	}
+	list_for_each_entry_safe(buf, temp, &conn->qp.sq.backlog, list) {
+		mlx5_fpga_conn_unmap_buf(conn, buf);
+		if (!buf->complete)
+			continue;
+		buf->complete(conn, conn->fdev, buf, MLX5_CQE_SYNDROME_WR_FLUSH_ERR);
+	}
+}
+
+static void mlx5_fpga_conn_destroy_qp(struct mlx5_fpga_conn *conn)
+{
+	struct mlx5_core_dev *dev = conn->fdev->mdev;
+	u32 in[MLX5_ST_SZ_DW(destroy_qp_in)] = {};
+
+	MLX5_SET(destroy_qp_in, in, opcode, MLX5_CMD_OP_DESTROY_QP);
+	MLX5_SET(destroy_qp_in, in, qpn, conn->qp.qpn);
+	mlx5_cmd_exec_in(dev, destroy_qp, in);
+
+	mlx5_fpga_conn_free_recv_bufs(conn);
+	mlx5_fpga_conn_flush_send_bufs(conn);
+	kvfree(conn->qp.sq.bufs);
+	kvfree(conn->qp.rq.bufs);
+	mlx5_wq_destroy(&conn->qp.wq_ctrl);
+}
+
+static int mlx5_fpga_conn_reset_qp(struct mlx5_fpga_conn *conn)
+{
+	struct mlx5_core_dev *mdev = conn->fdev->mdev;
+	u32 in[MLX5_ST_SZ_DW(qp_2rst_in)] = {};
+
+	mlx5_fpga_dbg(conn->fdev, "Modifying QP %u to RST\n", conn->qp.qpn);
+
+	MLX5_SET(qp_2rst_in, in, opcode, MLX5_CMD_OP_2RST_QP);
+	MLX5_SET(qp_2rst_in, in, qpn, conn->qp.qpn);
+
+	return mlx5_cmd_exec_in(mdev, qp_2rst, in);
+}
+
+static int mlx5_fpga_conn_init_qp(struct mlx5_fpga_conn *conn)
+{
+	u32 in[MLX5_ST_SZ_DW(rst2init_qp_in)] = {};
+	struct mlx5_fpga_device *fdev = conn->fdev;
+	struct mlx5_core_dev *mdev = fdev->mdev;
+	u32 *qpc;
+
+	mlx5_fpga_dbg(conn->fdev, "Modifying QP %u to INIT\n", conn->qp.qpn);
+
+	qpc = MLX5_ADDR_OF(rst2init_qp_in, in, qpc);
+
+	MLX5_SET(qpc, qpc, st, MLX5_QP_ST_RC);
+	MLX5_SET(qpc, qpc, pm_state, MLX5_QP_PM_MIGRATED);
+	MLX5_SET(qpc, qpc, primary_address_path.pkey_index, MLX5_FPGA_PKEY_INDEX);
+	MLX5_SET(qpc, qpc, primary_address_path.vhca_port_num, MLX5_FPGA_PORT_NUM);
+	MLX5_SET(qpc, qpc, pd, conn->fdev->conn_res.pdn);
+	MLX5_SET(qpc, qpc, cqn_snd, conn->cq.mcq.cqn);
+	MLX5_SET(qpc, qpc, cqn_rcv, conn->cq.mcq.cqn);
+	MLX5_SET64(qpc, qpc, dbr_addr, conn->qp.wq_ctrl.db.dma);
+
+	MLX5_SET(rst2init_qp_in, in, opcode, MLX5_CMD_OP_RST2INIT_QP);
+	MLX5_SET(rst2init_qp_in, in, qpn, conn->qp.qpn);
+
+	return mlx5_cmd_exec_in(mdev, rst2init_qp, in);
+}
+
+static int mlx5_fpga_conn_rtr_qp(struct mlx5_fpga_conn *conn)
+{
+	u32 in[MLX5_ST_SZ_DW(init2rtr_qp_in)] = {};
+	struct mlx5_fpga_device *fdev = conn->fdev;
+	struct mlx5_core_dev *mdev = fdev->mdev;
+	u32 *qpc;
+
+	mlx5_fpga_dbg(conn->fdev, "QP RTR\n");
+
+	qpc = MLX5_ADDR_OF(init2rtr_qp_in, in, qpc);
+
+	MLX5_SET(qpc, qpc, mtu, MLX5_QPC_MTU_1K_BYTES);
+	MLX5_SET(qpc, qpc, log_msg_max, (u8)MLX5_CAP_GEN(mdev, log_max_msg));
+	MLX5_SET(qpc, qpc, remote_qpn, conn->fpga_qpn);
+	MLX5_SET(qpc, qpc, next_rcv_psn,
+		 MLX5_GET(fpga_qpc, conn->fpga_qpc, next_send_psn));
+	MLX5_SET(qpc, qpc, primary_address_path.pkey_index, MLX5_FPGA_PKEY_INDEX);
+	MLX5_SET(qpc, qpc, primary_address_path.vhca_port_num, MLX5_FPGA_PORT_NUM);
+	ether_addr_copy(MLX5_ADDR_OF(qpc, qpc, primary_address_path.rmac_47_32),
+			MLX5_ADDR_OF(fpga_qpc, conn->fpga_qpc, fpga_mac_47_32));
+	MLX5_SET(qpc, qpc, primary_address_path.udp_sport,
+		 MLX5_CAP_ROCE(mdev, r_roce_min_src_udp_port));
+	MLX5_SET(qpc, qpc, primary_address_path.src_addr_index,
+		 conn->qp.sgid_index);
+	MLX5_SET(qpc, qpc, primary_address_path.hop_limit, 0);
+	memcpy(MLX5_ADDR_OF(qpc, qpc, primary_address_path.rgid_rip),
+	       MLX5_ADDR_OF(fpga_qpc, conn->fpga_qpc, fpga_ip),
+	       MLX5_FLD_SZ_BYTES(qpc, primary_address_path.rgid_rip));
+
+	MLX5_SET(init2rtr_qp_in, in, opcode, MLX5_CMD_OP_INIT2RTR_QP);
+	MLX5_SET(init2rtr_qp_in, in, qpn, conn->qp.qpn);
+
+	return mlx5_cmd_exec_in(mdev, init2rtr_qp, in);
+}
+
+static int mlx5_fpga_conn_rts_qp(struct mlx5_fpga_conn *conn)
+{
+	struct mlx5_fpga_device *fdev = conn->fdev;
+	u32 in[MLX5_ST_SZ_DW(rtr2rts_qp_in)] = {};
+	struct mlx5_core_dev *mdev = fdev->mdev;
+	u32 *qpc;
+
+	mlx5_fpga_dbg(conn->fdev, "QP RTS\n");
+
+	qpc = MLX5_ADDR_OF(rtr2rts_qp_in, in, qpc);
+
+	MLX5_SET(qpc, qpc, log_ack_req_freq, 8);
+	MLX5_SET(qpc, qpc, min_rnr_nak, 0x12);
+	MLX5_SET(qpc, qpc, primary_address_path.ack_timeout, 0x12); /* ~1.07s */
+	MLX5_SET(qpc, qpc, next_send_psn,
+		 MLX5_GET(fpga_qpc, conn->fpga_qpc, next_rcv_psn));
+	MLX5_SET(qpc, qpc, retry_count, 7);
+	MLX5_SET(qpc, qpc, rnr_retry, 7); /* Infinite retry if RNR NACK */
+
+	MLX5_SET(rtr2rts_qp_in, in, opcode, MLX5_CMD_OP_RTR2RTS_QP);
+	MLX5_SET(rtr2rts_qp_in, in, qpn, conn->qp.qpn);
+	MLX5_SET(rtr2rts_qp_in, in, opt_param_mask, MLX5_QP_OPTPAR_RNR_TIMEOUT);
+
+	return mlx5_cmd_exec_in(mdev, rtr2rts_qp, in);
+}
+
+static int mlx5_fpga_conn_connect(struct mlx5_fpga_conn *conn)
+{
+	struct mlx5_fpga_device *fdev = conn->fdev;
+	int err;
+
+	MLX5_SET(fpga_qpc, conn->fpga_qpc, state, MLX5_FPGA_QPC_STATE_ACTIVE);
+	err = mlx5_fpga_modify_qp(conn->fdev->mdev, conn->fpga_qpn,
+				  MLX5_FPGA_QPC_STATE, &conn->fpga_qpc);
+	if (err) {
+		mlx5_fpga_err(fdev, "Failed to activate FPGA RC QP: %d\n", err);
+		goto out;
+	}
+
+	err = mlx5_fpga_conn_reset_qp(conn);
+	if (err) {
+		mlx5_fpga_err(fdev, "Failed to change QP state to reset\n");
+		goto err_fpga_qp;
+	}
+
+	err = mlx5_fpga_conn_init_qp(conn);
+	if (err) {
+		mlx5_fpga_err(fdev, "Failed to modify QP from RESET to INIT\n");
+		goto err_fpga_qp;
+	}
+	conn->qp.active = true;
+
+	while (!mlx5_fpga_conn_post_recv_buf(conn))
+		;
+
+	err = mlx5_fpga_conn_rtr_qp(conn);
+	if (err) {
+		mlx5_fpga_err(fdev, "Failed to change QP state from INIT to RTR\n");
+		goto err_recv_bufs;
+	}
+
+	err = mlx5_fpga_conn_rts_qp(conn);
+	if (err) {
+		mlx5_fpga_err(fdev, "Failed to change QP state from RTR to RTS\n");
+		goto err_recv_bufs;
+	}
+	goto out;
+
+err_recv_bufs:
+	mlx5_fpga_conn_free_recv_bufs(conn);
+err_fpga_qp:
+	MLX5_SET(fpga_qpc, conn->fpga_qpc, state, MLX5_FPGA_QPC_STATE_INIT);
+	if (mlx5_fpga_modify_qp(conn->fdev->mdev, conn->fpga_qpn,
+				MLX5_FPGA_QPC_STATE, &conn->fpga_qpc))
+		mlx5_fpga_err(fdev, "Failed to revert FPGA QP to INIT\n");
+out:
+	return err;
+}
+
+struct mlx5_fpga_conn *mlx5_fpga_conn_create(struct mlx5_fpga_device *fdev,
+					     struct mlx5_fpga_conn_attr *attr,
+					     enum mlx5_ifc_fpga_qp_type qp_type)
+{
+	struct mlx5_fpga_conn *ret, *conn;
+	u8 *remote_mac, *remote_ip;
+	int err;
+
+	if (!attr->recv_cb)
+		return ERR_PTR(-EINVAL);
+
+	conn = kzalloc(sizeof(*conn), GFP_KERNEL);
+	if (!conn)
+		return ERR_PTR(-ENOMEM);
+
+	conn->fdev = fdev;
+	INIT_LIST_HEAD(&conn->qp.sq.backlog);
+
+	spin_lock_init(&conn->qp.sq.lock);
+
+	conn->recv_cb = attr->recv_cb;
+	conn->cb_arg = attr->cb_arg;
+
+	remote_mac = MLX5_ADDR_OF(fpga_qpc, conn->fpga_qpc, remote_mac_47_32);
+	err = mlx5_query_mac_address(fdev->mdev, remote_mac);
+	if (err) {
+		mlx5_fpga_err(fdev, "Failed to query local MAC: %d\n", err);
+		ret = ERR_PTR(err);
+		goto err;
+	}
+
+	/* Build Modified EUI-64 IPv6 address from the MAC address */
+	remote_ip = MLX5_ADDR_OF(fpga_qpc, conn->fpga_qpc, remote_ip);
+	remote_ip[0] = 0xfe;
+	remote_ip[1] = 0x80;
+	addrconf_addr_eui48(&remote_ip[8], remote_mac);
+
+	err = mlx5_core_reserved_gid_alloc(fdev->mdev, &conn->qp.sgid_index);
+	if (err) {
+		mlx5_fpga_err(fdev, "Failed to allocate SGID: %d\n", err);
+		ret = ERR_PTR(err);
+		goto err;
+	}
+
+	err = mlx5_core_roce_gid_set(fdev->mdev, conn->qp.sgid_index,
+				     MLX5_ROCE_VERSION_2,
+				     MLX5_ROCE_L3_TYPE_IPV6,
+				     remote_ip, remote_mac, true, 0,
+				     MLX5_FPGA_PORT_NUM);
+	if (err) {
+		mlx5_fpga_err(fdev, "Failed to set SGID: %d\n", err);
+		ret = ERR_PTR(err);
+		goto err_rsvd_gid;
+	}
+	mlx5_fpga_dbg(fdev, "Reserved SGID index %u\n", conn->qp.sgid_index);
+
+	/* Allow for one cqe per rx/tx wqe, plus one cqe for the next wqe,
+	 * created during processing of the cqe
+	 */
+	err = mlx5_fpga_conn_create_cq(conn,
+				       (attr->tx_size + attr->rx_size) * 2);
+	if (err) {
+		mlx5_fpga_err(fdev, "Failed to create CQ: %d\n", err);
+		ret = ERR_PTR(err);
+		goto err_gid;
+	}
+
+	mlx5_fpga_conn_arm_cq(conn);
+
+	err = mlx5_fpga_conn_create_qp(conn, attr->tx_size, attr->rx_size);
+	if (err) {
+		mlx5_fpga_err(fdev, "Failed to create QP: %d\n", err);
+		ret = ERR_PTR(err);
+		goto err_cq;
+	}
+
+	MLX5_SET(fpga_qpc, conn->fpga_qpc, state, MLX5_FPGA_QPC_STATE_INIT);
+	MLX5_SET(fpga_qpc, conn->fpga_qpc, qp_type, qp_type);
+	MLX5_SET(fpga_qpc, conn->fpga_qpc, st, MLX5_FPGA_QPC_ST_RC);
+	MLX5_SET(fpga_qpc, conn->fpga_qpc, ether_type, ETH_P_8021Q);
+	MLX5_SET(fpga_qpc, conn->fpga_qpc, vid, 0);
+	MLX5_SET(fpga_qpc, conn->fpga_qpc, next_rcv_psn, 1);
+	MLX5_SET(fpga_qpc, conn->fpga_qpc, next_send_psn, 0);
+	MLX5_SET(fpga_qpc, conn->fpga_qpc, pkey, MLX5_FPGA_PKEY);
+	MLX5_SET(fpga_qpc, conn->fpga_qpc, remote_qpn, conn->qp.qpn);
+	MLX5_SET(fpga_qpc, conn->fpga_qpc, rnr_retry, 7);
+	MLX5_SET(fpga_qpc, conn->fpga_qpc, retry_count, 7);
+
+	err = mlx5_fpga_create_qp(fdev->mdev, &conn->fpga_qpc,
+				  &conn->fpga_qpn);
+	if (err) {
+		mlx5_fpga_err(fdev, "Failed to create FPGA RC QP: %d\n", err);
+		ret = ERR_PTR(err);
+		goto err_qp;
+	}
+
+	err = mlx5_fpga_conn_connect(conn);
+	if (err) {
+		ret = ERR_PTR(err);
+		goto err_conn;
+	}
+
+	mlx5_fpga_dbg(fdev, "FPGA QPN is %u\n", conn->fpga_qpn);
+	ret = conn;
+	goto out;
+
+err_conn:
+	mlx5_fpga_destroy_qp(conn->fdev->mdev, conn->fpga_qpn);
+err_qp:
+	mlx5_fpga_conn_destroy_qp(conn);
+err_cq:
+	mlx5_fpga_conn_destroy_cq(conn);
+err_gid:
+	mlx5_core_roce_gid_set(fdev->mdev, conn->qp.sgid_index, 0, 0, NULL,
+			       NULL, false, 0, MLX5_FPGA_PORT_NUM);
+err_rsvd_gid:
+	mlx5_core_reserved_gid_free(fdev->mdev, conn->qp.sgid_index);
+err:
+	kfree(conn);
+out:
+	return ret;
+}
+
+void mlx5_fpga_conn_destroy(struct mlx5_fpga_conn *conn)
+{
+	conn->qp.active = false;
+	tasklet_disable(&conn->cq.tasklet);
+	synchronize_irq(conn->cq.mcq.irqn);
+
+	mlx5_fpga_destroy_qp(conn->fdev->mdev, conn->fpga_qpn);
+	mlx5_fpga_conn_destroy_qp(conn);
+	mlx5_fpga_conn_destroy_cq(conn);
+
+	mlx5_core_roce_gid_set(conn->fdev->mdev, conn->qp.sgid_index, 0, 0,
+			       NULL, NULL, false, 0, MLX5_FPGA_PORT_NUM);
+	mlx5_core_reserved_gid_free(conn->fdev->mdev, conn->qp.sgid_index);
+	kfree(conn);
+}
+
+int mlx5_fpga_conn_device_init(struct mlx5_fpga_device *fdev)
+{
+	int err;
+
+	err = mlx5_nic_vport_enable_roce(fdev->mdev);
+	if (err) {
+		mlx5_fpga_err(fdev, "Failed to enable RoCE: %d\n", err);
+		goto out;
+	}
+
+	fdev->conn_res.uar = mlx5_get_uars_page(fdev->mdev);
+	if (IS_ERR(fdev->conn_res.uar)) {
+		err = PTR_ERR(fdev->conn_res.uar);
+		mlx5_fpga_err(fdev, "get_uars_page failed, %d\n", err);
+		goto err_roce;
+	}
+	mlx5_fpga_dbg(fdev, "Allocated UAR index %u\n",
+		      fdev->conn_res.uar->index);
+
+	err = mlx5_core_alloc_pd(fdev->mdev, &fdev->conn_res.pdn);
+	if (err) {
+		mlx5_fpga_err(fdev, "alloc pd failed, %d\n", err);
+		goto err_uar;
+	}
+	mlx5_fpga_dbg(fdev, "Allocated PD %u\n", fdev->conn_res.pdn);
+
+	err = mlx5_fpga_conn_create_mkey(fdev->mdev, fdev->conn_res.pdn,
+					 &fdev->conn_res.mkey);
+	if (err) {
+		mlx5_fpga_err(fdev, "create mkey failed, %d\n", err);
+		goto err_dealloc_pd;
+	}
+	mlx5_fpga_dbg(fdev, "Created mkey 0x%x\n", fdev->conn_res.mkey);
+
+	return 0;
+
+err_dealloc_pd:
+	mlx5_core_dealloc_pd(fdev->mdev, fdev->conn_res.pdn);
+err_uar:
+	mlx5_put_uars_page(fdev->mdev, fdev->conn_res.uar);
+err_roce:
+	mlx5_nic_vport_disable_roce(fdev->mdev);
+out:
+	return err;
+}
+
+void mlx5_fpga_conn_device_cleanup(struct mlx5_fpga_device *fdev)
+{
+	mlx5_core_destroy_mkey(fdev->mdev, fdev->conn_res.mkey);
+	mlx5_core_dealloc_pd(fdev->mdev, fdev->conn_res.pdn);
+	mlx5_put_uars_page(fdev->mdev, fdev->conn_res.uar);
+	mlx5_nic_vport_disable_roce(fdev->mdev);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fpga/conn.h b/drivers/net/ethernet/mellanox/mlx5/core/fpga/conn.h
new file mode 100644
index 000000000..5116e869a
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fpga/conn.h
@@ -0,0 +1,96 @@
+/*
+ * Copyright (c) 2017 Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+
+#ifndef __MLX5_FPGA_CONN_H__
+#define __MLX5_FPGA_CONN_H__
+
+#include <linux/mlx5/cq.h>
+#include <linux/mlx5/qp.h>
+
+#include "fpga/core.h"
+#include "fpga/sdk.h"
+#include "wq.h"
+
+struct mlx5_fpga_conn {
+	struct mlx5_fpga_device *fdev;
+
+	void (*recv_cb)(void *cb_arg, struct mlx5_fpga_dma_buf *buf);
+	void *cb_arg;
+
+	/* FPGA QP */
+	u32 fpga_qpc[MLX5_ST_SZ_DW(fpga_qpc)];
+	u32 fpga_qpn;
+
+	/* CQ */
+	struct {
+		struct mlx5_cqwq wq;
+		struct mlx5_wq_ctrl wq_ctrl;
+		struct mlx5_core_cq mcq;
+		struct tasklet_struct tasklet;
+	} cq;
+
+	/* QP */
+	struct {
+		bool active;
+		int sgid_index;
+		struct mlx5_wq_qp wq;
+		struct mlx5_wq_ctrl wq_ctrl;
+		u32 qpn;
+		struct {
+			spinlock_t lock; /* Protects all SQ state */
+			unsigned int pc;
+			unsigned int cc;
+			unsigned int size;
+			struct mlx5_fpga_dma_buf **bufs;
+			struct list_head backlog;
+		} sq;
+		struct {
+			unsigned int pc;
+			unsigned int cc;
+			unsigned int size;
+			struct mlx5_fpga_dma_buf **bufs;
+		} rq;
+	} qp;
+};
+
+int mlx5_fpga_conn_device_init(struct mlx5_fpga_device *fdev);
+void mlx5_fpga_conn_device_cleanup(struct mlx5_fpga_device *fdev);
+struct mlx5_fpga_conn *
+mlx5_fpga_conn_create(struct mlx5_fpga_device *fdev,
+		      struct mlx5_fpga_conn_attr *attr,
+		      enum mlx5_ifc_fpga_qp_type qp_type);
+void mlx5_fpga_conn_destroy(struct mlx5_fpga_conn *conn);
+int mlx5_fpga_conn_send(struct mlx5_fpga_conn *conn,
+			struct mlx5_fpga_dma_buf *buf);
+
+#endif /* __MLX5_FPGA_CONN_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fpga/core.c b/drivers/net/ethernet/mellanox/mlx5/core/fpga/core.c
new file mode 100644
index 000000000..39c03dcbd
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fpga/core.c
@@ -0,0 +1,375 @@
+/*
+ * Copyright (c) 2017, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/etherdevice.h>
+#include <linux/mlx5/driver.h>
+
+#include "mlx5_core.h"
+#include "lib/mlx5.h"
+#include "lib/eq.h"
+#include "fpga/core.h"
+#include "fpga/conn.h"
+
+static const char *const mlx5_fpga_error_strings[] = {
+	"Null Syndrome",
+	"Corrupted DDR",
+	"Flash Timeout",
+	"Internal Link Error",
+	"Watchdog HW Failure",
+	"I2C Failure",
+	"Image Changed",
+	"Temperature Critical",
+};
+
+static const char * const mlx5_fpga_qp_error_strings[] = {
+	"Null Syndrome",
+	"Retry Counter Expired",
+	"RNR Expired",
+};
+static struct mlx5_fpga_device *mlx5_fpga_device_alloc(void)
+{
+	struct mlx5_fpga_device *fdev = NULL;
+
+	fdev = kzalloc(sizeof(*fdev), GFP_KERNEL);
+	if (!fdev)
+		return NULL;
+
+	spin_lock_init(&fdev->state_lock);
+	fdev->state = MLX5_FPGA_STATUS_NONE;
+	return fdev;
+}
+
+static const char *mlx5_fpga_image_name(enum mlx5_fpga_image image)
+{
+	switch (image) {
+	case MLX5_FPGA_IMAGE_USER:
+		return "user";
+	case MLX5_FPGA_IMAGE_FACTORY:
+		return "factory";
+	default:
+		return "unknown";
+	}
+}
+
+static const char *mlx5_fpga_name(u32 fpga_id)
+{
+	static char ret[32];
+
+	switch (fpga_id) {
+	case MLX5_FPGA_NEWTON:
+		return "Newton";
+	case MLX5_FPGA_EDISON:
+		return "Edison";
+	case MLX5_FPGA_MORSE:
+		return "Morse";
+	case MLX5_FPGA_MORSEQ:
+		return "MorseQ";
+	}
+
+	snprintf(ret, sizeof(ret), "Unknown %d", fpga_id);
+	return ret;
+}
+
+static int mlx5_is_fpga_lookaside(u32 fpga_id)
+{
+	return fpga_id != MLX5_FPGA_NEWTON && fpga_id != MLX5_FPGA_EDISON;
+}
+
+static int mlx5_fpga_device_load_check(struct mlx5_fpga_device *fdev)
+{
+	struct mlx5_fpga_query query;
+	int err;
+
+	err = mlx5_fpga_query(fdev->mdev, &query);
+	if (err) {
+		mlx5_fpga_err(fdev, "Failed to query status: %d\n", err);
+		return err;
+	}
+
+	fdev->last_admin_image = query.admin_image;
+	fdev->last_oper_image = query.oper_image;
+
+	mlx5_fpga_info(fdev, "Status %u; Admin image %u; Oper image %u\n",
+		       query.status, query.admin_image, query.oper_image);
+
+	/* for FPGA lookaside projects FPGA load status is not important */
+	if (mlx5_is_fpga_lookaside(MLX5_CAP_FPGA(fdev->mdev, fpga_id)))
+		return 0;
+
+	if (query.status != MLX5_FPGA_STATUS_SUCCESS) {
+		mlx5_fpga_err(fdev, "%s image failed to load; status %u\n",
+			      mlx5_fpga_image_name(fdev->last_oper_image),
+			      query.status);
+		return -EIO;
+	}
+
+	return 0;
+}
+
+static int mlx5_fpga_device_brb(struct mlx5_fpga_device *fdev)
+{
+	int err;
+	struct mlx5_core_dev *mdev = fdev->mdev;
+
+	err = mlx5_fpga_ctrl_op(mdev, MLX5_FPGA_CTRL_OPERATION_SANDBOX_BYPASS_ON);
+	if (err) {
+		mlx5_fpga_err(fdev, "Failed to set bypass on: %d\n", err);
+		return err;
+	}
+	err = mlx5_fpga_ctrl_op(mdev, MLX5_FPGA_CTRL_OPERATION_RESET_SANDBOX);
+	if (err) {
+		mlx5_fpga_err(fdev, "Failed to reset SBU: %d\n", err);
+		return err;
+	}
+	err = mlx5_fpga_ctrl_op(mdev, MLX5_FPGA_CTRL_OPERATION_SANDBOX_BYPASS_OFF);
+	if (err) {
+		mlx5_fpga_err(fdev, "Failed to set bypass off: %d\n", err);
+		return err;
+	}
+	return 0;
+}
+
+static int mlx5_fpga_event(struct mlx5_fpga_device *, unsigned long, void *);
+
+static int fpga_err_event(struct notifier_block *nb, unsigned long event, void *eqe)
+{
+	struct mlx5_fpga_device *fdev = mlx5_nb_cof(nb, struct mlx5_fpga_device, fpga_err_nb);
+
+	return mlx5_fpga_event(fdev, event, eqe);
+}
+
+static int fpga_qp_err_event(struct notifier_block *nb, unsigned long event, void *eqe)
+{
+	struct mlx5_fpga_device *fdev = mlx5_nb_cof(nb, struct mlx5_fpga_device, fpga_qp_err_nb);
+
+	return mlx5_fpga_event(fdev, event, eqe);
+}
+
+int mlx5_fpga_device_start(struct mlx5_core_dev *mdev)
+{
+	struct mlx5_fpga_device *fdev = mdev->fpga;
+	unsigned int max_num_qps;
+	unsigned long flags;
+	u32 fpga_id;
+	int err;
+
+	if (!fdev)
+		return 0;
+
+	err = mlx5_fpga_caps(fdev->mdev);
+	if (err)
+		goto out;
+
+	err = mlx5_fpga_device_load_check(fdev);
+	if (err)
+		goto out;
+
+	fpga_id = MLX5_CAP_FPGA(fdev->mdev, fpga_id);
+	mlx5_fpga_info(fdev, "FPGA card %s:%u\n", mlx5_fpga_name(fpga_id), fpga_id);
+
+	/* No QPs if FPGA does not participate in net processing */
+	if (mlx5_is_fpga_lookaside(fpga_id))
+		goto out;
+
+	mlx5_fpga_info(fdev, "%s(%d): image, version %u; SBU %06x:%04x version %d\n",
+		       mlx5_fpga_image_name(fdev->last_oper_image),
+		       fdev->last_oper_image,
+		       MLX5_CAP_FPGA(fdev->mdev, image_version),
+		       MLX5_CAP_FPGA(fdev->mdev, ieee_vendor_id),
+		       MLX5_CAP_FPGA(fdev->mdev, sandbox_product_id),
+		       MLX5_CAP_FPGA(fdev->mdev, sandbox_product_version));
+
+	max_num_qps = MLX5_CAP_FPGA(mdev, shell_caps.max_num_qps);
+	if (!max_num_qps) {
+		mlx5_fpga_err(fdev, "FPGA reports 0 QPs in SHELL_CAPS\n");
+		err = -ENOTSUPP;
+		goto out;
+	}
+
+	err = mlx5_core_reserve_gids(mdev, max_num_qps);
+	if (err)
+		goto out;
+
+	MLX5_NB_INIT(&fdev->fpga_err_nb, fpga_err_event, FPGA_ERROR);
+	MLX5_NB_INIT(&fdev->fpga_qp_err_nb, fpga_qp_err_event, FPGA_QP_ERROR);
+	mlx5_eq_notifier_register(fdev->mdev, &fdev->fpga_err_nb);
+	mlx5_eq_notifier_register(fdev->mdev, &fdev->fpga_qp_err_nb);
+
+	err = mlx5_fpga_conn_device_init(fdev);
+	if (err)
+		goto err_rsvd_gid;
+
+	if (fdev->last_oper_image == MLX5_FPGA_IMAGE_USER) {
+		err = mlx5_fpga_device_brb(fdev);
+		if (err)
+			goto err_conn_init;
+	}
+
+	goto out;
+
+err_conn_init:
+	mlx5_fpga_conn_device_cleanup(fdev);
+
+err_rsvd_gid:
+	mlx5_eq_notifier_unregister(fdev->mdev, &fdev->fpga_err_nb);
+	mlx5_eq_notifier_unregister(fdev->mdev, &fdev->fpga_qp_err_nb);
+	mlx5_core_unreserve_gids(mdev, max_num_qps);
+out:
+	spin_lock_irqsave(&fdev->state_lock, flags);
+	fdev->state = err ? MLX5_FPGA_STATUS_FAILURE : MLX5_FPGA_STATUS_SUCCESS;
+	spin_unlock_irqrestore(&fdev->state_lock, flags);
+	return err;
+}
+
+int mlx5_fpga_init(struct mlx5_core_dev *mdev)
+{
+	struct mlx5_fpga_device *fdev = NULL;
+
+	if (!MLX5_CAP_GEN(mdev, fpga)) {
+		mlx5_core_dbg(mdev, "FPGA capability not present\n");
+		return 0;
+	}
+
+	mlx5_core_dbg(mdev, "Initializing FPGA\n");
+
+	fdev = mlx5_fpga_device_alloc();
+	if (!fdev)
+		return -ENOMEM;
+
+	fdev->mdev = mdev;
+	mdev->fpga = fdev;
+
+	return 0;
+}
+
+void mlx5_fpga_device_stop(struct mlx5_core_dev *mdev)
+{
+	struct mlx5_fpga_device *fdev = mdev->fpga;
+	unsigned int max_num_qps;
+	unsigned long flags;
+	int err;
+
+	if (!fdev)
+		return;
+
+	if (mlx5_is_fpga_lookaside(MLX5_CAP_FPGA(fdev->mdev, fpga_id)))
+		return;
+
+	spin_lock_irqsave(&fdev->state_lock, flags);
+	if (fdev->state != MLX5_FPGA_STATUS_SUCCESS) {
+		spin_unlock_irqrestore(&fdev->state_lock, flags);
+		return;
+	}
+	fdev->state = MLX5_FPGA_STATUS_NONE;
+	spin_unlock_irqrestore(&fdev->state_lock, flags);
+
+	if (fdev->last_oper_image == MLX5_FPGA_IMAGE_USER) {
+		err = mlx5_fpga_ctrl_op(mdev, MLX5_FPGA_CTRL_OPERATION_SANDBOX_BYPASS_ON);
+		if (err)
+			mlx5_fpga_err(fdev, "Failed to re-set SBU bypass on: %d\n",
+				      err);
+	}
+
+	mlx5_fpga_conn_device_cleanup(fdev);
+	mlx5_eq_notifier_unregister(fdev->mdev, &fdev->fpga_err_nb);
+	mlx5_eq_notifier_unregister(fdev->mdev, &fdev->fpga_qp_err_nb);
+
+	max_num_qps = MLX5_CAP_FPGA(mdev, shell_caps.max_num_qps);
+	mlx5_core_unreserve_gids(mdev, max_num_qps);
+}
+
+void mlx5_fpga_cleanup(struct mlx5_core_dev *mdev)
+{
+	struct mlx5_fpga_device *fdev = mdev->fpga;
+
+	mlx5_fpga_device_stop(mdev);
+	kfree(fdev);
+	mdev->fpga = NULL;
+}
+
+static const char *mlx5_fpga_syndrome_to_string(u8 syndrome)
+{
+	if (syndrome < ARRAY_SIZE(mlx5_fpga_error_strings))
+		return mlx5_fpga_error_strings[syndrome];
+	return "Unknown";
+}
+
+static const char *mlx5_fpga_qp_syndrome_to_string(u8 syndrome)
+{
+	if (syndrome < ARRAY_SIZE(mlx5_fpga_qp_error_strings))
+		return mlx5_fpga_qp_error_strings[syndrome];
+	return "Unknown";
+}
+
+static int mlx5_fpga_event(struct mlx5_fpga_device *fdev,
+			   unsigned long event, void *eqe)
+{
+	void *data = ((struct mlx5_eqe *)eqe)->data.raw;
+	const char *event_name;
+	bool teardown = false;
+	unsigned long flags;
+	u8 syndrome;
+
+	switch (event) {
+	case MLX5_EVENT_TYPE_FPGA_ERROR:
+		syndrome = MLX5_GET(fpga_error_event, data, syndrome);
+		event_name = mlx5_fpga_syndrome_to_string(syndrome);
+		break;
+	case MLX5_EVENT_TYPE_FPGA_QP_ERROR:
+		syndrome = MLX5_GET(fpga_qp_error_event, data, syndrome);
+		event_name = mlx5_fpga_qp_syndrome_to_string(syndrome);
+		break;
+	default:
+		return NOTIFY_DONE;
+	}
+
+	spin_lock_irqsave(&fdev->state_lock, flags);
+	switch (fdev->state) {
+	case MLX5_FPGA_STATUS_SUCCESS:
+		mlx5_fpga_warn(fdev, "Error %u: %s\n", syndrome, event_name);
+		teardown = true;
+		break;
+	default:
+		mlx5_fpga_warn_ratelimited(fdev, "Unexpected error event %u: %s\n",
+					   syndrome, event_name);
+	}
+	spin_unlock_irqrestore(&fdev->state_lock, flags);
+	/* We tear-down the card's interfaces and functionality because
+	 * the FPGA bump-on-the-wire is misbehaving and we lose ability
+	 * to communicate with the network. User may still be able to
+	 * recover by re-programming or debugging the FPGA
+	 */
+	if (teardown)
+		mlx5_trigger_health_work(fdev->mdev);
+
+	return NOTIFY_OK;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fpga/core.h b/drivers/net/ethernet/mellanox/mlx5/core/fpga/core.h
new file mode 100644
index 000000000..750c32050
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fpga/core.h
@@ -0,0 +1,111 @@
+/*
+ * Copyright (c) 2017, Mellanox Technologies, Ltd.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef __MLX5_FPGA_CORE_H__
+#define __MLX5_FPGA_CORE_H__
+
+#ifdef CONFIG_MLX5_FPGA
+
+#include <linux/mlx5/eq.h>
+
+#include "mlx5_core.h"
+#include "lib/eq.h"
+#include "fpga/cmd.h"
+
+/* Represents an Innova device */
+struct mlx5_fpga_device {
+	struct mlx5_core_dev *mdev;
+	struct mlx5_nb fpga_err_nb;
+	struct mlx5_nb fpga_qp_err_nb;
+	spinlock_t state_lock; /* Protects state transitions */
+	enum mlx5_fpga_status state;
+	enum mlx5_fpga_image last_admin_image;
+	enum mlx5_fpga_image last_oper_image;
+
+	/* QP Connection resources */
+	struct {
+		u32 pdn;
+		u32 mkey;
+		struct mlx5_uars_page *uar;
+	} conn_res;
+};
+
+#define mlx5_fpga_dbg(__adev, format, ...) \
+	mlx5_core_dbg((__adev)->mdev, "FPGA: %s:%d:(pid %d): " format, \
+		       __func__, __LINE__, current->pid, ##__VA_ARGS__)
+
+#define mlx5_fpga_err(__adev, format, ...) \
+	mlx5_core_err((__adev)->mdev, "FPGA: %s:%d:(pid %d): " format, \
+		      __func__, __LINE__, current->pid, ##__VA_ARGS__)
+
+#define mlx5_fpga_warn(__adev, format, ...) \
+	mlx5_core_warn((__adev)->mdev, "FPGA: %s:%d:(pid %d): " format, \
+		       __func__, __LINE__, current->pid, ##__VA_ARGS__)
+
+#define mlx5_fpga_warn_ratelimited(__adev, format, ...) \
+	mlx5_core_err_rl((__adev)->mdev, "FPGA: %s:%d: " \
+			 format, __func__, __LINE__, ##__VA_ARGS__)
+
+#define mlx5_fpga_notice(__adev, format, ...) \
+	mlx5_core_info((__adev)->mdev, "FPGA: " format, ##__VA_ARGS__)
+
+#define mlx5_fpga_info(__adev, format, ...) \
+	mlx5_core_info((__adev)->mdev, "FPGA: " format, ##__VA_ARGS__)
+
+int mlx5_fpga_init(struct mlx5_core_dev *mdev);
+void mlx5_fpga_cleanup(struct mlx5_core_dev *mdev);
+int mlx5_fpga_device_start(struct mlx5_core_dev *mdev);
+void mlx5_fpga_device_stop(struct mlx5_core_dev *mdev);
+
+#else
+
+static inline int mlx5_fpga_init(struct mlx5_core_dev *mdev)
+{
+	return 0;
+}
+
+static inline void mlx5_fpga_cleanup(struct mlx5_core_dev *mdev)
+{
+}
+
+static inline int mlx5_fpga_device_start(struct mlx5_core_dev *mdev)
+{
+	return 0;
+}
+
+static inline void mlx5_fpga_device_stop(struct mlx5_core_dev *mdev)
+{
+}
+
+#endif
+
+#endif /* __MLX5_FPGA_CORE_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fpga/sdk.c b/drivers/net/ethernet/mellanox/mlx5/core/fpga/sdk.c
new file mode 100644
index 000000000..14962969c
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fpga/sdk.c
@@ -0,0 +1,170 @@
+/*
+ * Copyright (c) 2017 Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+
+#include <linux/mlx5/device.h>
+
+#include "fpga/core.h"
+#include "fpga/conn.h"
+#include "fpga/sdk.h"
+
+struct mlx5_fpga_conn *
+mlx5_fpga_sbu_conn_create(struct mlx5_fpga_device *fdev,
+			  struct mlx5_fpga_conn_attr *attr)
+{
+	return mlx5_fpga_conn_create(fdev, attr, MLX5_FPGA_QPC_QP_TYPE_SANDBOX_QP);
+}
+EXPORT_SYMBOL(mlx5_fpga_sbu_conn_create);
+
+void mlx5_fpga_sbu_conn_destroy(struct mlx5_fpga_conn *conn)
+{
+	mlx5_fpga_conn_destroy(conn);
+}
+EXPORT_SYMBOL(mlx5_fpga_sbu_conn_destroy);
+
+int mlx5_fpga_sbu_conn_sendmsg(struct mlx5_fpga_conn *conn,
+			       struct mlx5_fpga_dma_buf *buf)
+{
+	return mlx5_fpga_conn_send(conn, buf);
+}
+EXPORT_SYMBOL(mlx5_fpga_sbu_conn_sendmsg);
+
+static int mlx5_fpga_mem_read_i2c(struct mlx5_fpga_device *fdev, size_t size,
+				  u64 addr, u8 *buf)
+{
+	size_t max_size = MLX5_FPGA_ACCESS_REG_SIZE_MAX;
+	size_t bytes_done = 0;
+	u8 actual_size;
+	int err;
+
+	if (!size)
+		return -EINVAL;
+
+	if (!fdev->mdev)
+		return -ENOTCONN;
+
+	while (bytes_done < size) {
+		actual_size = min(max_size, (size - bytes_done));
+
+		err = mlx5_fpga_access_reg(fdev->mdev, actual_size,
+					   addr + bytes_done,
+					   buf + bytes_done, false);
+		if (err) {
+			mlx5_fpga_err(fdev, "Failed to read over I2C: %d\n",
+				      err);
+			break;
+		}
+
+		bytes_done += actual_size;
+	}
+
+	return err;
+}
+
+static int mlx5_fpga_mem_write_i2c(struct mlx5_fpga_device *fdev, size_t size,
+				   u64 addr, u8 *buf)
+{
+	size_t max_size = MLX5_FPGA_ACCESS_REG_SIZE_MAX;
+	size_t bytes_done = 0;
+	u8 actual_size;
+	int err;
+
+	if (!size)
+		return -EINVAL;
+
+	if (!fdev->mdev)
+		return -ENOTCONN;
+
+	while (bytes_done < size) {
+		actual_size = min(max_size, (size - bytes_done));
+
+		err = mlx5_fpga_access_reg(fdev->mdev, actual_size,
+					   addr + bytes_done,
+					   buf + bytes_done, true);
+		if (err) {
+			mlx5_fpga_err(fdev, "Failed to write FPGA crspace\n");
+			break;
+		}
+
+		bytes_done += actual_size;
+	}
+
+	return err;
+}
+
+int mlx5_fpga_mem_read(struct mlx5_fpga_device *fdev, size_t size, u64 addr,
+		       void *buf, enum mlx5_fpga_access_type access_type)
+{
+	int ret;
+
+	switch (access_type) {
+	case MLX5_FPGA_ACCESS_TYPE_I2C:
+		ret = mlx5_fpga_mem_read_i2c(fdev, size, addr, buf);
+		if (ret)
+			return ret;
+		break;
+	default:
+		mlx5_fpga_warn(fdev, "Unexpected read access_type %u\n",
+			       access_type);
+		return -EACCES;
+	}
+
+	return size;
+}
+EXPORT_SYMBOL(mlx5_fpga_mem_read);
+
+int mlx5_fpga_mem_write(struct mlx5_fpga_device *fdev, size_t size, u64 addr,
+			void *buf, enum mlx5_fpga_access_type access_type)
+{
+	int ret;
+
+	switch (access_type) {
+	case MLX5_FPGA_ACCESS_TYPE_I2C:
+		ret = mlx5_fpga_mem_write_i2c(fdev, size, addr, buf);
+		if (ret)
+			return ret;
+		break;
+	default:
+		mlx5_fpga_warn(fdev, "Unexpected write access_type %u\n",
+			       access_type);
+		return -EACCES;
+	}
+
+	return size;
+}
+EXPORT_SYMBOL(mlx5_fpga_mem_write);
+
+int mlx5_fpga_get_sbu_caps(struct mlx5_fpga_device *fdev, int size, void *buf)
+{
+	return mlx5_fpga_sbu_caps(fdev->mdev, buf, size);
+}
+EXPORT_SYMBOL(mlx5_fpga_get_sbu_caps);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fpga/sdk.h b/drivers/net/ethernet/mellanox/mlx5/core/fpga/sdk.h
new file mode 100644
index 000000000..89ef59265
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fpga/sdk.h
@@ -0,0 +1,214 @@
+/*
+ * Copyright (c) 2017 Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+
+#ifndef MLX5_FPGA_SDK_H
+#define MLX5_FPGA_SDK_H
+
+#include <linux/types.h>
+#include <linux/dma-direction.h>
+
+/**
+ * DOC: Innova SDK
+ * This header defines the in-kernel API for Innova FPGA client drivers.
+ */
+#define SBU_QP_QUEUE_SIZE 8
+#define MLX5_FPGA_CMD_TIMEOUT_MSEC (60 * 1000)
+
+/**
+ * enum mlx5_fpga_access_type - Enumerated the different methods possible for
+ * accessing the device memory address space
+ *
+ * @MLX5_FPGA_ACCESS_TYPE_I2C: Use the slow CX-FPGA I2C bus
+ * @MLX5_FPGA_ACCESS_TYPE_DONTCARE: Use the fastest available method
+ */
+enum mlx5_fpga_access_type {
+	MLX5_FPGA_ACCESS_TYPE_I2C = 0x0,
+	MLX5_FPGA_ACCESS_TYPE_DONTCARE = 0x0,
+};
+
+struct mlx5_fpga_conn;
+struct mlx5_fpga_device;
+
+/**
+ * struct mlx5_fpga_dma_entry - A scatter-gather DMA entry
+ */
+struct mlx5_fpga_dma_entry {
+	/** @data: Virtual address pointer to the data */
+	void *data;
+	/** @size: Size in bytes of the data */
+	unsigned int size;
+	/** @dma_addr: Private member. Physical DMA-mapped address of the data */
+	dma_addr_t dma_addr;
+};
+
+/**
+ * struct mlx5_fpga_dma_buf - A packet buffer
+ * May contain up to 2 scatter-gather data entries
+ */
+struct mlx5_fpga_dma_buf {
+	/** @dma_dir: DMA direction */
+	enum dma_data_direction dma_dir;
+	/** @sg: Scatter-gather entries pointing to the data in memory */
+	struct mlx5_fpga_dma_entry sg[2];
+	/** @list: Item in SQ backlog, for TX packets */
+	struct list_head list;
+	/**
+	 * @complete: Completion routine, for TX packets
+	 * @conn: FPGA Connection this packet was sent to
+	 * @fdev: FPGA device this packet was sent to
+	 * @buf: The packet buffer
+	 * @status: 0 if successful, or an error code otherwise
+	 */
+	void (*complete)(struct mlx5_fpga_conn *conn,
+			 struct mlx5_fpga_device *fdev,
+			 struct mlx5_fpga_dma_buf *buf, u8 status);
+};
+
+/**
+ * struct mlx5_fpga_conn_attr - FPGA connection attributes
+ * Describes the attributes of a connection
+ */
+struct mlx5_fpga_conn_attr {
+	/** @tx_size: Size of connection TX queue, in packets */
+	unsigned int tx_size;
+	/** @rx_size: Size of connection RX queue, in packets */
+	unsigned int rx_size;
+	/**
+	 * @recv_cb: Callback function which is called for received packets
+	 * @cb_arg: The value provided in mlx5_fpga_conn_attr.cb_arg
+	 * @buf: A buffer containing a received packet
+	 *
+	 * buf is guaranteed to only contain a single scatter-gather entry.
+	 * The size of the actual packet received is specified in buf.sg[0].size
+	 * When this callback returns, the packet buffer may be re-used for
+	 * subsequent receives.
+	 */
+	void (*recv_cb)(void *cb_arg, struct mlx5_fpga_dma_buf *buf);
+	/** @cb_arg: A context to be passed to recv_cb callback */
+	void *cb_arg;
+};
+
+/**
+ * mlx5_fpga_sbu_conn_create() - Initialize a new FPGA SBU connection
+ * @fdev: The FPGA device
+ * @attr: Attributes of the new connection
+ *
+ * Sets up a new FPGA SBU connection with the specified attributes.
+ * The receive callback function may be called for incoming messages even
+ * before this function returns.
+ *
+ * The caller must eventually destroy the connection by calling
+ * mlx5_fpga_sbu_conn_destroy.
+ *
+ * Return: A new connection, or ERR_PTR() error value otherwise.
+ */
+struct mlx5_fpga_conn *
+mlx5_fpga_sbu_conn_create(struct mlx5_fpga_device *fdev,
+			  struct mlx5_fpga_conn_attr *attr);
+
+/**
+ * mlx5_fpga_sbu_conn_destroy() - Destroy an FPGA SBU connection
+ * @conn: The FPGA SBU connection to destroy
+ *
+ * Cleans up an FPGA SBU connection which was previously created with
+ * mlx5_fpga_sbu_conn_create.
+ */
+void mlx5_fpga_sbu_conn_destroy(struct mlx5_fpga_conn *conn);
+
+/**
+ * mlx5_fpga_sbu_conn_sendmsg() - Queue the transmission of a packet
+ * @conn: An FPGA SBU connection
+ * @buf: The packet buffer
+ *
+ * Queues a packet for transmission over an FPGA SBU connection.
+ * The buffer should not be modified or freed until completion.
+ * Upon completion, the buf's complete() callback is invoked, indicating the
+ * success or error status of the transmission.
+ *
+ * Return: 0 if successful, or an error value otherwise.
+ */
+int mlx5_fpga_sbu_conn_sendmsg(struct mlx5_fpga_conn *conn,
+			       struct mlx5_fpga_dma_buf *buf);
+
+/**
+ * mlx5_fpga_mem_read() - Read from FPGA memory address space
+ * @fdev: The FPGA device
+ * @size: Size of chunk to read, in bytes
+ * @addr: Starting address to read from, in FPGA address space
+ * @buf: Buffer to read into
+ * @access_type: Method for reading
+ *
+ * Reads from the specified address into the specified buffer.
+ * The address may point to configuration space or to DDR.
+ * Large reads may be performed internally as several non-atomic operations.
+ * This function may sleep, so should not be called from atomic contexts.
+ *
+ * Return: 0 if successful, or an error value otherwise.
+ */
+int mlx5_fpga_mem_read(struct mlx5_fpga_device *fdev, size_t size, u64 addr,
+		       void *buf, enum mlx5_fpga_access_type access_type);
+
+/**
+ * mlx5_fpga_mem_write() - Write to FPGA memory address space
+ * @fdev: The FPGA device
+ * @size: Size of chunk to write, in bytes
+ * @addr: Starting address to write to, in FPGA address space
+ * @buf: Buffer which contains data to write
+ * @access_type: Method for writing
+ *
+ * Writes the specified buffer data to FPGA memory at the specified address.
+ * The address may point to configuration space or to DDR.
+ * Large writes may be performed internally as several non-atomic operations.
+ * This function may sleep, so should not be called from atomic contexts.
+ *
+ * Return: 0 if successful, or an error value otherwise.
+ */
+int mlx5_fpga_mem_write(struct mlx5_fpga_device *fdev, size_t size, u64 addr,
+			void *buf, enum mlx5_fpga_access_type access_type);
+
+/**
+ * mlx5_fpga_get_sbu_caps() - Read the SBU capabilities
+ * @fdev: The FPGA device
+ * @size: Size of the buffer to read into
+ * @buf: Buffer to read the capabilities into
+ *
+ * Reads the FPGA SBU capabilities into the specified buffer.
+ * The format of the capabilities buffer is SBU-dependent.
+ *
+ * Return: 0 if successful
+ *         -EINVAL if the buffer is not large enough to contain SBU caps
+ *         or any other error value otherwise.
+ */
+int mlx5_fpga_get_sbu_caps(struct mlx5_fpga_device *fdev, int size, void *buf);
+
+#endif /* MLX5_FPGA_SDK_H */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fs_cmd.c b/drivers/net/ethernet/mellanox/mlx5/core/fs_cmd.c
new file mode 100644
index 000000000..32d4c9674
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fs_cmd.c
@@ -0,0 +1,1107 @@
+/*
+ * Copyright (c) 2015, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/mlx5/driver.h>
+#include <linux/mlx5/device.h>
+#include <linux/mlx5/mlx5_ifc.h>
+
+#include "fs_core.h"
+#include "fs_cmd.h"
+#include "fs_ft_pool.h"
+#include "mlx5_core.h"
+#include "eswitch.h"
+
+static int mlx5_cmd_stub_update_root_ft(struct mlx5_flow_root_namespace *ns,
+					struct mlx5_flow_table *ft,
+					u32 underlay_qpn,
+					bool disconnect)
+{
+	return 0;
+}
+
+static int mlx5_cmd_stub_create_flow_table(struct mlx5_flow_root_namespace *ns,
+					   struct mlx5_flow_table *ft,
+					   struct mlx5_flow_table_attr *ft_attr,
+					   struct mlx5_flow_table *next_ft)
+{
+	int max_fte = ft_attr->max_fte;
+
+	ft->max_fte = max_fte ? roundup_pow_of_two(max_fte) : 1;
+
+	return 0;
+}
+
+static int mlx5_cmd_stub_destroy_flow_table(struct mlx5_flow_root_namespace *ns,
+					    struct mlx5_flow_table *ft)
+{
+	return 0;
+}
+
+static int mlx5_cmd_stub_modify_flow_table(struct mlx5_flow_root_namespace *ns,
+					   struct mlx5_flow_table *ft,
+					   struct mlx5_flow_table *next_ft)
+{
+	return 0;
+}
+
+static int mlx5_cmd_stub_create_flow_group(struct mlx5_flow_root_namespace *ns,
+					   struct mlx5_flow_table *ft,
+					   u32 *in,
+					   struct mlx5_flow_group *fg)
+{
+	return 0;
+}
+
+static int mlx5_cmd_stub_destroy_flow_group(struct mlx5_flow_root_namespace *ns,
+					    struct mlx5_flow_table *ft,
+					    struct mlx5_flow_group *fg)
+{
+	return 0;
+}
+
+static int mlx5_cmd_stub_create_fte(struct mlx5_flow_root_namespace *ns,
+				    struct mlx5_flow_table *ft,
+				    struct mlx5_flow_group *group,
+				    struct fs_fte *fte)
+{
+	return 0;
+}
+
+static int mlx5_cmd_stub_update_fte(struct mlx5_flow_root_namespace *ns,
+				    struct mlx5_flow_table *ft,
+				    struct mlx5_flow_group *group,
+				    int modify_mask,
+				    struct fs_fte *fte)
+{
+	return -EOPNOTSUPP;
+}
+
+static int mlx5_cmd_stub_delete_fte(struct mlx5_flow_root_namespace *ns,
+				    struct mlx5_flow_table *ft,
+				    struct fs_fte *fte)
+{
+	return 0;
+}
+
+static int mlx5_cmd_stub_packet_reformat_alloc(struct mlx5_flow_root_namespace *ns,
+					       struct mlx5_pkt_reformat_params *params,
+					       enum mlx5_flow_namespace_type namespace,
+					       struct mlx5_pkt_reformat *pkt_reformat)
+{
+	return 0;
+}
+
+static void mlx5_cmd_stub_packet_reformat_dealloc(struct mlx5_flow_root_namespace *ns,
+						  struct mlx5_pkt_reformat *pkt_reformat)
+{
+}
+
+static int mlx5_cmd_stub_modify_header_alloc(struct mlx5_flow_root_namespace *ns,
+					     u8 namespace, u8 num_actions,
+					     void *modify_actions,
+					     struct mlx5_modify_hdr *modify_hdr)
+{
+	return 0;
+}
+
+static void mlx5_cmd_stub_modify_header_dealloc(struct mlx5_flow_root_namespace *ns,
+						struct mlx5_modify_hdr *modify_hdr)
+{
+}
+
+static int mlx5_cmd_stub_set_peer(struct mlx5_flow_root_namespace *ns,
+				  struct mlx5_flow_root_namespace *peer_ns)
+{
+	return 0;
+}
+
+static int mlx5_cmd_stub_create_ns(struct mlx5_flow_root_namespace *ns)
+{
+	return 0;
+}
+
+static int mlx5_cmd_stub_destroy_ns(struct mlx5_flow_root_namespace *ns)
+{
+	return 0;
+}
+
+static u32 mlx5_cmd_stub_get_capabilities(struct mlx5_flow_root_namespace *ns,
+					  enum fs_flow_table_type ft_type)
+{
+	return 0;
+}
+
+static int mlx5_cmd_set_slave_root_fdb(struct mlx5_core_dev *master,
+				       struct mlx5_core_dev *slave,
+				       bool ft_id_valid,
+				       u32 ft_id)
+{
+	u32 out[MLX5_ST_SZ_DW(set_flow_table_root_out)] = {};
+	u32 in[MLX5_ST_SZ_DW(set_flow_table_root_in)] = {};
+	struct mlx5_flow_root_namespace *root;
+	struct mlx5_flow_namespace *ns;
+
+	MLX5_SET(set_flow_table_root_in, in, opcode,
+		 MLX5_CMD_OP_SET_FLOW_TABLE_ROOT);
+	MLX5_SET(set_flow_table_root_in, in, table_type,
+		 FS_FT_FDB);
+	if (ft_id_valid) {
+		MLX5_SET(set_flow_table_root_in, in,
+			 table_eswitch_owner_vhca_id_valid, 1);
+		MLX5_SET(set_flow_table_root_in, in,
+			 table_eswitch_owner_vhca_id,
+			 MLX5_CAP_GEN(master, vhca_id));
+		MLX5_SET(set_flow_table_root_in, in, table_id,
+			 ft_id);
+	} else {
+		ns = mlx5_get_flow_namespace(slave,
+					     MLX5_FLOW_NAMESPACE_FDB);
+		root = find_root(&ns->node);
+		MLX5_SET(set_flow_table_root_in, in, table_id,
+			 root->root_ft->id);
+	}
+
+	return mlx5_cmd_exec(slave, in, sizeof(in), out, sizeof(out));
+}
+
+static int
+mlx5_cmd_stub_destroy_match_definer(struct mlx5_flow_root_namespace *ns,
+				    int definer_id)
+{
+	return 0;
+}
+
+static int
+mlx5_cmd_stub_create_match_definer(struct mlx5_flow_root_namespace *ns,
+				   u16 format_id, u32 *match_mask)
+{
+	return 0;
+}
+
+static int mlx5_cmd_update_root_ft(struct mlx5_flow_root_namespace *ns,
+				   struct mlx5_flow_table *ft, u32 underlay_qpn,
+				   bool disconnect)
+{
+	u32 in[MLX5_ST_SZ_DW(set_flow_table_root_in)] = {};
+	struct mlx5_core_dev *dev = ns->dev;
+	int err;
+
+	if ((MLX5_CAP_GEN(dev, port_type) == MLX5_CAP_PORT_TYPE_IB) &&
+	    underlay_qpn == 0)
+		return 0;
+
+	if (ft->type == FS_FT_FDB &&
+	    mlx5_lag_is_shared_fdb(dev) &&
+	    !mlx5_lag_is_master(dev))
+		return 0;
+
+	MLX5_SET(set_flow_table_root_in, in, opcode,
+		 MLX5_CMD_OP_SET_FLOW_TABLE_ROOT);
+	MLX5_SET(set_flow_table_root_in, in, table_type, ft->type);
+
+	if (disconnect)
+		MLX5_SET(set_flow_table_root_in, in, op_mod, 1);
+	else
+		MLX5_SET(set_flow_table_root_in, in, table_id, ft->id);
+
+	MLX5_SET(set_flow_table_root_in, in, underlay_qpn, underlay_qpn);
+	MLX5_SET(set_flow_table_root_in, in, vport_number, ft->vport);
+	MLX5_SET(set_flow_table_root_in, in, other_vport,
+		 !!(ft->flags & MLX5_FLOW_TABLE_OTHER_VPORT));
+
+	err = mlx5_cmd_exec_in(dev, set_flow_table_root, in);
+	if (!err &&
+	    ft->type == FS_FT_FDB &&
+	    mlx5_lag_is_shared_fdb(dev) &&
+	    mlx5_lag_is_master(dev)) {
+		err = mlx5_cmd_set_slave_root_fdb(dev,
+						  mlx5_lag_get_peer_mdev(dev),
+						  !disconnect, (!disconnect) ?
+						  ft->id : 0);
+		if (err && !disconnect) {
+			MLX5_SET(set_flow_table_root_in, in, op_mod, 0);
+			MLX5_SET(set_flow_table_root_in, in, table_id,
+				 ns->root_ft->id);
+			mlx5_cmd_exec_in(dev, set_flow_table_root, in);
+		}
+	}
+
+	return err;
+}
+
+static int mlx5_cmd_create_flow_table(struct mlx5_flow_root_namespace *ns,
+				      struct mlx5_flow_table *ft,
+				      struct mlx5_flow_table_attr *ft_attr,
+				      struct mlx5_flow_table *next_ft)
+{
+	int en_encap = !!(ft->flags & MLX5_FLOW_TABLE_TUNNEL_EN_REFORMAT);
+	int en_decap = !!(ft->flags & MLX5_FLOW_TABLE_TUNNEL_EN_DECAP);
+	int term = !!(ft->flags & MLX5_FLOW_TABLE_TERMINATION);
+	u32 out[MLX5_ST_SZ_DW(create_flow_table_out)] = {};
+	u32 in[MLX5_ST_SZ_DW(create_flow_table_in)] = {};
+	struct mlx5_core_dev *dev = ns->dev;
+	unsigned int size;
+	int err;
+
+	if (ft_attr->max_fte != POOL_NEXT_SIZE)
+		size = roundup_pow_of_two(ft_attr->max_fte);
+	size = mlx5_ft_pool_get_avail_sz(dev, ft->type, ft_attr->max_fte);
+	if (!size)
+		return -ENOSPC;
+
+	MLX5_SET(create_flow_table_in, in, opcode,
+		 MLX5_CMD_OP_CREATE_FLOW_TABLE);
+
+	MLX5_SET(create_flow_table_in, in, uid, ft_attr->uid);
+	MLX5_SET(create_flow_table_in, in, table_type, ft->type);
+	MLX5_SET(create_flow_table_in, in, flow_table_context.level, ft->level);
+	MLX5_SET(create_flow_table_in, in, flow_table_context.log_size, size ? ilog2(size) : 0);
+	MLX5_SET(create_flow_table_in, in, vport_number, ft->vport);
+	MLX5_SET(create_flow_table_in, in, other_vport,
+		 !!(ft->flags & MLX5_FLOW_TABLE_OTHER_VPORT));
+
+	MLX5_SET(create_flow_table_in, in, flow_table_context.decap_en,
+		 en_decap);
+	MLX5_SET(create_flow_table_in, in, flow_table_context.reformat_en,
+		 en_encap);
+	MLX5_SET(create_flow_table_in, in, flow_table_context.termination_table,
+		 term);
+
+	switch (ft->op_mod) {
+	case FS_FT_OP_MOD_NORMAL:
+		if (next_ft) {
+			MLX5_SET(create_flow_table_in, in,
+				 flow_table_context.table_miss_action,
+				 MLX5_FLOW_TABLE_MISS_ACTION_FWD);
+			MLX5_SET(create_flow_table_in, in,
+				 flow_table_context.table_miss_id, next_ft->id);
+		} else {
+			MLX5_SET(create_flow_table_in, in,
+				 flow_table_context.table_miss_action,
+				 ft->def_miss_action);
+		}
+		break;
+
+	case FS_FT_OP_MOD_LAG_DEMUX:
+		MLX5_SET(create_flow_table_in, in, op_mod, 0x1);
+		if (next_ft)
+			MLX5_SET(create_flow_table_in, in,
+				 flow_table_context.lag_master_next_table_id,
+				 next_ft->id);
+		break;
+	}
+
+	err = mlx5_cmd_exec_inout(dev, create_flow_table, in, out);
+	if (!err) {
+		ft->id = MLX5_GET(create_flow_table_out, out,
+				  table_id);
+		ft->max_fte = size;
+	} else {
+		mlx5_ft_pool_put_sz(ns->dev, size);
+	}
+
+	return err;
+}
+
+static int mlx5_cmd_destroy_flow_table(struct mlx5_flow_root_namespace *ns,
+				       struct mlx5_flow_table *ft)
+{
+	u32 in[MLX5_ST_SZ_DW(destroy_flow_table_in)] = {};
+	struct mlx5_core_dev *dev = ns->dev;
+	int err;
+
+	MLX5_SET(destroy_flow_table_in, in, opcode,
+		 MLX5_CMD_OP_DESTROY_FLOW_TABLE);
+	MLX5_SET(destroy_flow_table_in, in, table_type, ft->type);
+	MLX5_SET(destroy_flow_table_in, in, table_id, ft->id);
+	MLX5_SET(destroy_flow_table_in, in, vport_number, ft->vport);
+	MLX5_SET(destroy_flow_table_in, in, other_vport,
+		 !!(ft->flags & MLX5_FLOW_TABLE_OTHER_VPORT));
+
+	err = mlx5_cmd_exec_in(dev, destroy_flow_table, in);
+	if (!err)
+		mlx5_ft_pool_put_sz(ns->dev, ft->max_fte);
+
+	return err;
+}
+
+static int mlx5_cmd_modify_flow_table(struct mlx5_flow_root_namespace *ns,
+				      struct mlx5_flow_table *ft,
+				      struct mlx5_flow_table *next_ft)
+{
+	u32 in[MLX5_ST_SZ_DW(modify_flow_table_in)] = {};
+	struct mlx5_core_dev *dev = ns->dev;
+
+	MLX5_SET(modify_flow_table_in, in, opcode,
+		 MLX5_CMD_OP_MODIFY_FLOW_TABLE);
+	MLX5_SET(modify_flow_table_in, in, table_type, ft->type);
+	MLX5_SET(modify_flow_table_in, in, table_id, ft->id);
+
+	if (ft->op_mod == FS_FT_OP_MOD_LAG_DEMUX) {
+		MLX5_SET(modify_flow_table_in, in, modify_field_select,
+			 MLX5_MODIFY_FLOW_TABLE_LAG_NEXT_TABLE_ID);
+		if (next_ft) {
+			MLX5_SET(modify_flow_table_in, in,
+				 flow_table_context.lag_master_next_table_id, next_ft->id);
+		} else {
+			MLX5_SET(modify_flow_table_in, in,
+				 flow_table_context.lag_master_next_table_id, 0);
+		}
+	} else {
+		MLX5_SET(modify_flow_table_in, in, vport_number, ft->vport);
+		MLX5_SET(modify_flow_table_in, in, other_vport,
+			 !!(ft->flags & MLX5_FLOW_TABLE_OTHER_VPORT));
+		MLX5_SET(modify_flow_table_in, in, modify_field_select,
+			 MLX5_MODIFY_FLOW_TABLE_MISS_TABLE_ID);
+		if (next_ft) {
+			MLX5_SET(modify_flow_table_in, in,
+				 flow_table_context.table_miss_action,
+				 MLX5_FLOW_TABLE_MISS_ACTION_FWD);
+			MLX5_SET(modify_flow_table_in, in,
+				 flow_table_context.table_miss_id,
+				 next_ft->id);
+		} else {
+			MLX5_SET(modify_flow_table_in, in,
+				 flow_table_context.table_miss_action,
+				 ft->def_miss_action);
+		}
+	}
+
+	return mlx5_cmd_exec_in(dev, modify_flow_table, in);
+}
+
+static int mlx5_cmd_create_flow_group(struct mlx5_flow_root_namespace *ns,
+				      struct mlx5_flow_table *ft,
+				      u32 *in,
+				      struct mlx5_flow_group *fg)
+{
+	u32 out[MLX5_ST_SZ_DW(create_flow_group_out)] = {};
+	struct mlx5_core_dev *dev = ns->dev;
+	int err;
+
+	MLX5_SET(create_flow_group_in, in, opcode,
+		 MLX5_CMD_OP_CREATE_FLOW_GROUP);
+	MLX5_SET(create_flow_group_in, in, table_type, ft->type);
+	MLX5_SET(create_flow_group_in, in, table_id, ft->id);
+	if (ft->vport) {
+		MLX5_SET(create_flow_group_in, in, vport_number, ft->vport);
+		MLX5_SET(create_flow_group_in, in, other_vport, 1);
+	}
+
+	MLX5_SET(create_flow_group_in, in, vport_number, ft->vport);
+	MLX5_SET(create_flow_group_in, in, other_vport,
+		 !!(ft->flags & MLX5_FLOW_TABLE_OTHER_VPORT));
+	err = mlx5_cmd_exec_inout(dev, create_flow_group, in, out);
+	if (!err)
+		fg->id = MLX5_GET(create_flow_group_out, out,
+				  group_id);
+	return err;
+}
+
+static int mlx5_cmd_destroy_flow_group(struct mlx5_flow_root_namespace *ns,
+				       struct mlx5_flow_table *ft,
+				       struct mlx5_flow_group *fg)
+{
+	u32 in[MLX5_ST_SZ_DW(destroy_flow_group_in)] = {};
+	struct mlx5_core_dev *dev = ns->dev;
+
+	MLX5_SET(destroy_flow_group_in, in, opcode,
+		 MLX5_CMD_OP_DESTROY_FLOW_GROUP);
+	MLX5_SET(destroy_flow_group_in, in, table_type, ft->type);
+	MLX5_SET(destroy_flow_group_in, in, table_id, ft->id);
+	MLX5_SET(destroy_flow_group_in, in, group_id, fg->id);
+	MLX5_SET(destroy_flow_group_in, in, vport_number, ft->vport);
+	MLX5_SET(destroy_flow_group_in, in, other_vport,
+		 !!(ft->flags & MLX5_FLOW_TABLE_OTHER_VPORT));
+	return mlx5_cmd_exec_in(dev, destroy_flow_group, in);
+}
+
+static int mlx5_set_extended_dest(struct mlx5_core_dev *dev,
+				  struct fs_fte *fte, bool *extended_dest)
+{
+	int fw_log_max_fdb_encap_uplink =
+		MLX5_CAP_ESW(dev, log_max_fdb_encap_uplink);
+	int num_fwd_destinations = 0;
+	struct mlx5_flow_rule *dst;
+	int num_encap = 0;
+
+	*extended_dest = false;
+	if (!(fte->action.action & MLX5_FLOW_CONTEXT_ACTION_FWD_DEST))
+		return 0;
+
+	list_for_each_entry(dst, &fte->node.children, node.list) {
+		if (dst->dest_attr.type == MLX5_FLOW_DESTINATION_TYPE_COUNTER ||
+		    dst->dest_attr.type == MLX5_FLOW_DESTINATION_TYPE_NONE)
+			continue;
+		if ((dst->dest_attr.type == MLX5_FLOW_DESTINATION_TYPE_VPORT ||
+		     dst->dest_attr.type == MLX5_FLOW_DESTINATION_TYPE_UPLINK) &&
+		    dst->dest_attr.vport.flags & MLX5_FLOW_DEST_VPORT_REFORMAT_ID)
+			num_encap++;
+		num_fwd_destinations++;
+	}
+	if (num_fwd_destinations > 1 && num_encap > 0)
+		*extended_dest = true;
+
+	if (*extended_dest && !fw_log_max_fdb_encap_uplink) {
+		mlx5_core_warn(dev, "FW does not support extended destination");
+		return -EOPNOTSUPP;
+	}
+	if (num_encap > (1 << fw_log_max_fdb_encap_uplink)) {
+		mlx5_core_warn(dev, "FW does not support more than %d encaps",
+			       1 << fw_log_max_fdb_encap_uplink);
+		return -EOPNOTSUPP;
+	}
+
+	return 0;
+}
+
+static void
+mlx5_cmd_set_fte_flow_meter(struct fs_fte *fte, void *in_flow_context)
+{
+	void *exe_aso_ctrl;
+	void *execute_aso;
+
+	execute_aso = MLX5_ADDR_OF(flow_context, in_flow_context,
+				   execute_aso[0]);
+	MLX5_SET(execute_aso, execute_aso, valid, 1);
+	MLX5_SET(execute_aso, execute_aso, aso_object_id,
+		 fte->action.exe_aso.object_id);
+
+	exe_aso_ctrl = MLX5_ADDR_OF(execute_aso, execute_aso, exe_aso_ctrl);
+	MLX5_SET(exe_aso_ctrl_flow_meter, exe_aso_ctrl, return_reg_id,
+		 fte->action.exe_aso.return_reg_id);
+	MLX5_SET(exe_aso_ctrl_flow_meter, exe_aso_ctrl, aso_type,
+		 fte->action.exe_aso.type);
+	MLX5_SET(exe_aso_ctrl_flow_meter, exe_aso_ctrl, init_color,
+		 fte->action.exe_aso.flow_meter.init_color);
+	MLX5_SET(exe_aso_ctrl_flow_meter, exe_aso_ctrl, meter_id,
+		 fte->action.exe_aso.flow_meter.meter_idx);
+}
+
+static int mlx5_cmd_set_fte(struct mlx5_core_dev *dev,
+			    int opmod, int modify_mask,
+			    struct mlx5_flow_table *ft,
+			    unsigned group_id,
+			    struct fs_fte *fte)
+{
+	u32 out[MLX5_ST_SZ_DW(set_fte_out)] = {0};
+	bool extended_dest = false;
+	struct mlx5_flow_rule *dst;
+	void *in_flow_context, *vlan;
+	void *in_match_value;
+	unsigned int inlen;
+	int dst_cnt_size;
+	void *in_dests;
+	u32 *in;
+	int err;
+
+	if (mlx5_set_extended_dest(dev, fte, &extended_dest))
+		return -EOPNOTSUPP;
+
+	if (!extended_dest)
+		dst_cnt_size = MLX5_ST_SZ_BYTES(dest_format_struct);
+	else
+		dst_cnt_size = MLX5_ST_SZ_BYTES(extended_dest_format);
+
+	inlen = MLX5_ST_SZ_BYTES(set_fte_in) + fte->dests_size * dst_cnt_size;
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in)
+		return -ENOMEM;
+
+	MLX5_SET(set_fte_in, in, opcode, MLX5_CMD_OP_SET_FLOW_TABLE_ENTRY);
+	MLX5_SET(set_fte_in, in, op_mod, opmod);
+	MLX5_SET(set_fte_in, in, modify_enable_mask, modify_mask);
+	MLX5_SET(set_fte_in, in, table_type, ft->type);
+	MLX5_SET(set_fte_in, in, table_id,   ft->id);
+	MLX5_SET(set_fte_in, in, flow_index, fte->index);
+	MLX5_SET(set_fte_in, in, ignore_flow_level,
+		 !!(fte->action.flags & FLOW_ACT_IGNORE_FLOW_LEVEL));
+
+	MLX5_SET(set_fte_in, in, vport_number, ft->vport);
+	MLX5_SET(set_fte_in, in, other_vport,
+		 !!(ft->flags & MLX5_FLOW_TABLE_OTHER_VPORT));
+
+	in_flow_context = MLX5_ADDR_OF(set_fte_in, in, flow_context);
+	MLX5_SET(flow_context, in_flow_context, group_id, group_id);
+
+	MLX5_SET(flow_context, in_flow_context, flow_tag,
+		 fte->flow_context.flow_tag);
+	MLX5_SET(flow_context, in_flow_context, flow_source,
+		 fte->flow_context.flow_source);
+
+	MLX5_SET(flow_context, in_flow_context, extended_destination,
+		 extended_dest);
+	if (extended_dest) {
+		u32 action;
+
+		action = fte->action.action &
+			~MLX5_FLOW_CONTEXT_ACTION_PACKET_REFORMAT;
+		MLX5_SET(flow_context, in_flow_context, action, action);
+	} else {
+		MLX5_SET(flow_context, in_flow_context, action,
+			 fte->action.action);
+		if (fte->action.pkt_reformat)
+			MLX5_SET(flow_context, in_flow_context, packet_reformat_id,
+				 fte->action.pkt_reformat->id);
+	}
+	if (fte->action.modify_hdr)
+		MLX5_SET(flow_context, in_flow_context, modify_header_id,
+			 fte->action.modify_hdr->id);
+
+	MLX5_SET(flow_context, in_flow_context, encrypt_decrypt_type,
+		 fte->action.crypto.type);
+	MLX5_SET(flow_context, in_flow_context, encrypt_decrypt_obj_id,
+		 fte->action.crypto.obj_id);
+
+	vlan = MLX5_ADDR_OF(flow_context, in_flow_context, push_vlan);
+
+	MLX5_SET(vlan, vlan, ethtype, fte->action.vlan[0].ethtype);
+	MLX5_SET(vlan, vlan, vid, fte->action.vlan[0].vid);
+	MLX5_SET(vlan, vlan, prio, fte->action.vlan[0].prio);
+
+	vlan = MLX5_ADDR_OF(flow_context, in_flow_context, push_vlan_2);
+
+	MLX5_SET(vlan, vlan, ethtype, fte->action.vlan[1].ethtype);
+	MLX5_SET(vlan, vlan, vid, fte->action.vlan[1].vid);
+	MLX5_SET(vlan, vlan, prio, fte->action.vlan[1].prio);
+
+	in_match_value = MLX5_ADDR_OF(flow_context, in_flow_context,
+				      match_value);
+	memcpy(in_match_value, &fte->val, sizeof(fte->val));
+
+	in_dests = MLX5_ADDR_OF(flow_context, in_flow_context, destination);
+	if (fte->action.action & MLX5_FLOW_CONTEXT_ACTION_FWD_DEST) {
+		int list_size = 0;
+
+		list_for_each_entry(dst, &fte->node.children, node.list) {
+			enum mlx5_flow_destination_type type = dst->dest_attr.type;
+			enum mlx5_ifc_flow_destination_type ifc_type;
+			unsigned int id;
+
+			if (type == MLX5_FLOW_DESTINATION_TYPE_COUNTER)
+				continue;
+
+			switch (type) {
+			case MLX5_FLOW_DESTINATION_TYPE_NONE:
+				continue;
+			case MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE_NUM:
+				id = dst->dest_attr.ft_num;
+				ifc_type = MLX5_IFC_FLOW_DESTINATION_TYPE_FLOW_TABLE;
+				break;
+			case MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE:
+				id = dst->dest_attr.ft->id;
+				ifc_type = MLX5_IFC_FLOW_DESTINATION_TYPE_FLOW_TABLE;
+				break;
+			case MLX5_FLOW_DESTINATION_TYPE_UPLINK:
+			case MLX5_FLOW_DESTINATION_TYPE_VPORT:
+				MLX5_SET(dest_format_struct, in_dests,
+					 destination_eswitch_owner_vhca_id_valid,
+					 !!(dst->dest_attr.vport.flags &
+					    MLX5_FLOW_DEST_VPORT_VHCA_ID));
+				MLX5_SET(dest_format_struct, in_dests,
+					 destination_eswitch_owner_vhca_id,
+					 dst->dest_attr.vport.vhca_id);
+				if (type == MLX5_FLOW_DESTINATION_TYPE_UPLINK) {
+					/* destination_id is reserved */
+					id = 0;
+					ifc_type = MLX5_IFC_FLOW_DESTINATION_TYPE_UPLINK;
+					break;
+				}
+				ifc_type = MLX5_IFC_FLOW_DESTINATION_TYPE_VPORT;
+				id = dst->dest_attr.vport.num;
+				if (extended_dest &&
+				    dst->dest_attr.vport.pkt_reformat) {
+					MLX5_SET(dest_format_struct, in_dests,
+						 packet_reformat,
+						 !!(dst->dest_attr.vport.flags &
+						    MLX5_FLOW_DEST_VPORT_REFORMAT_ID));
+					MLX5_SET(extended_dest_format, in_dests,
+						 packet_reformat_id,
+						 dst->dest_attr.vport.pkt_reformat->id);
+				}
+				break;
+			case MLX5_FLOW_DESTINATION_TYPE_FLOW_SAMPLER:
+				id = dst->dest_attr.sampler_id;
+				ifc_type = MLX5_IFC_FLOW_DESTINATION_TYPE_FLOW_SAMPLER;
+				break;
+			default:
+				id = dst->dest_attr.tir_num;
+				ifc_type = MLX5_IFC_FLOW_DESTINATION_TYPE_TIR;
+			}
+
+			MLX5_SET(dest_format_struct, in_dests, destination_type,
+				 ifc_type);
+			MLX5_SET(dest_format_struct, in_dests, destination_id, id);
+			in_dests += dst_cnt_size;
+			list_size++;
+		}
+
+		MLX5_SET(flow_context, in_flow_context, destination_list_size,
+			 list_size);
+	}
+
+	if (fte->action.action & MLX5_FLOW_CONTEXT_ACTION_COUNT) {
+		int max_list_size = BIT(MLX5_CAP_FLOWTABLE_TYPE(dev,
+					log_max_flow_counter,
+					ft->type));
+		int list_size = 0;
+
+		list_for_each_entry(dst, &fte->node.children, node.list) {
+			if (dst->dest_attr.type !=
+			    MLX5_FLOW_DESTINATION_TYPE_COUNTER)
+				continue;
+
+			MLX5_SET(flow_counter_list, in_dests, flow_counter_id,
+				 dst->dest_attr.counter_id);
+			in_dests += dst_cnt_size;
+			list_size++;
+		}
+		if (list_size > max_list_size) {
+			err = -EINVAL;
+			goto err_out;
+		}
+
+		MLX5_SET(flow_context, in_flow_context, flow_counter_list_size,
+			 list_size);
+	}
+
+	if (fte->action.action & MLX5_FLOW_CONTEXT_ACTION_EXECUTE_ASO) {
+		if (fte->action.exe_aso.type == MLX5_EXE_ASO_FLOW_METER) {
+			mlx5_cmd_set_fte_flow_meter(fte, in_flow_context);
+		} else {
+			err = -EOPNOTSUPP;
+			goto err_out;
+		}
+	}
+
+	err = mlx5_cmd_exec(dev, in, inlen, out, sizeof(out));
+err_out:
+	kvfree(in);
+	return err;
+}
+
+static int mlx5_cmd_create_fte(struct mlx5_flow_root_namespace *ns,
+			       struct mlx5_flow_table *ft,
+			       struct mlx5_flow_group *group,
+			       struct fs_fte *fte)
+{
+	struct mlx5_core_dev *dev = ns->dev;
+	unsigned int group_id = group->id;
+
+	return mlx5_cmd_set_fte(dev, 0, 0, ft, group_id, fte);
+}
+
+static int mlx5_cmd_update_fte(struct mlx5_flow_root_namespace *ns,
+			       struct mlx5_flow_table *ft,
+			       struct mlx5_flow_group *fg,
+			       int modify_mask,
+			       struct fs_fte *fte)
+{
+	int opmod;
+	struct mlx5_core_dev *dev = ns->dev;
+	int atomic_mod_cap = MLX5_CAP_FLOWTABLE(dev,
+						flow_table_properties_nic_receive.
+						flow_modify_en);
+	if (!atomic_mod_cap)
+		return -EOPNOTSUPP;
+	opmod = 1;
+
+	return	mlx5_cmd_set_fte(dev, opmod, modify_mask, ft, fg->id, fte);
+}
+
+static int mlx5_cmd_delete_fte(struct mlx5_flow_root_namespace *ns,
+			       struct mlx5_flow_table *ft,
+			       struct fs_fte *fte)
+{
+	u32 in[MLX5_ST_SZ_DW(delete_fte_in)] = {};
+	struct mlx5_core_dev *dev = ns->dev;
+
+	MLX5_SET(delete_fte_in, in, opcode, MLX5_CMD_OP_DELETE_FLOW_TABLE_ENTRY);
+	MLX5_SET(delete_fte_in, in, table_type, ft->type);
+	MLX5_SET(delete_fte_in, in, table_id, ft->id);
+	MLX5_SET(delete_fte_in, in, flow_index, fte->index);
+	MLX5_SET(delete_fte_in, in, vport_number, ft->vport);
+	MLX5_SET(delete_fte_in, in, other_vport,
+		 !!(ft->flags & MLX5_FLOW_TABLE_OTHER_VPORT));
+
+	return mlx5_cmd_exec_in(dev, delete_fte, in);
+}
+
+int mlx5_cmd_fc_bulk_alloc(struct mlx5_core_dev *dev,
+			   enum mlx5_fc_bulk_alloc_bitmask alloc_bitmask,
+			   u32 *id)
+{
+	u32 out[MLX5_ST_SZ_DW(alloc_flow_counter_out)] = {};
+	u32 in[MLX5_ST_SZ_DW(alloc_flow_counter_in)] = {};
+	int err;
+
+	MLX5_SET(alloc_flow_counter_in, in, opcode,
+		 MLX5_CMD_OP_ALLOC_FLOW_COUNTER);
+	MLX5_SET(alloc_flow_counter_in, in, flow_counter_bulk, alloc_bitmask);
+
+	err = mlx5_cmd_exec_inout(dev, alloc_flow_counter, in, out);
+	if (!err)
+		*id = MLX5_GET(alloc_flow_counter_out, out, flow_counter_id);
+	return err;
+}
+
+int mlx5_cmd_fc_alloc(struct mlx5_core_dev *dev, u32 *id)
+{
+	return mlx5_cmd_fc_bulk_alloc(dev, 0, id);
+}
+
+int mlx5_cmd_fc_free(struct mlx5_core_dev *dev, u32 id)
+{
+	u32 in[MLX5_ST_SZ_DW(dealloc_flow_counter_in)] = {};
+
+	MLX5_SET(dealloc_flow_counter_in, in, opcode,
+		 MLX5_CMD_OP_DEALLOC_FLOW_COUNTER);
+	MLX5_SET(dealloc_flow_counter_in, in, flow_counter_id, id);
+	return mlx5_cmd_exec_in(dev, dealloc_flow_counter, in);
+}
+
+int mlx5_cmd_fc_query(struct mlx5_core_dev *dev, u32 id,
+		      u64 *packets, u64 *bytes)
+{
+	u32 out[MLX5_ST_SZ_BYTES(query_flow_counter_out) +
+		MLX5_ST_SZ_BYTES(traffic_counter)] = {};
+	u32 in[MLX5_ST_SZ_DW(query_flow_counter_in)] = {};
+	void *stats;
+	int err = 0;
+
+	MLX5_SET(query_flow_counter_in, in, opcode,
+		 MLX5_CMD_OP_QUERY_FLOW_COUNTER);
+	MLX5_SET(query_flow_counter_in, in, op_mod, 0);
+	MLX5_SET(query_flow_counter_in, in, flow_counter_id, id);
+	err = mlx5_cmd_exec(dev, in, sizeof(in), out, sizeof(out));
+	if (err)
+		return err;
+
+	stats = MLX5_ADDR_OF(query_flow_counter_out, out, flow_statistics);
+	*packets = MLX5_GET64(traffic_counter, stats, packets);
+	*bytes = MLX5_GET64(traffic_counter, stats, octets);
+	return 0;
+}
+
+int mlx5_cmd_fc_get_bulk_query_out_len(int bulk_len)
+{
+	return MLX5_ST_SZ_BYTES(query_flow_counter_out) +
+		MLX5_ST_SZ_BYTES(traffic_counter) * bulk_len;
+}
+
+int mlx5_cmd_fc_bulk_query(struct mlx5_core_dev *dev, u32 base_id, int bulk_len,
+			   u32 *out)
+{
+	int outlen = mlx5_cmd_fc_get_bulk_query_out_len(bulk_len);
+	u32 in[MLX5_ST_SZ_DW(query_flow_counter_in)] = {};
+
+	MLX5_SET(query_flow_counter_in, in, opcode,
+		 MLX5_CMD_OP_QUERY_FLOW_COUNTER);
+	MLX5_SET(query_flow_counter_in, in, flow_counter_id, base_id);
+	MLX5_SET(query_flow_counter_in, in, num_of_counters, bulk_len);
+	return mlx5_cmd_exec(dev, in, sizeof(in), out, outlen);
+}
+
+static int mlx5_cmd_packet_reformat_alloc(struct mlx5_flow_root_namespace *ns,
+					  struct mlx5_pkt_reformat_params *params,
+					  enum mlx5_flow_namespace_type namespace,
+					  struct mlx5_pkt_reformat *pkt_reformat)
+{
+	u32 out[MLX5_ST_SZ_DW(alloc_packet_reformat_context_out)] = {};
+	struct mlx5_core_dev *dev = ns->dev;
+	void *packet_reformat_context_in;
+	int max_encap_size;
+	void *reformat;
+	int inlen;
+	int err;
+	u32 *in;
+
+	if (namespace == MLX5_FLOW_NAMESPACE_FDB ||
+	    namespace == MLX5_FLOW_NAMESPACE_FDB_BYPASS)
+		max_encap_size = MLX5_CAP_ESW(dev, max_encap_header_size);
+	else
+		max_encap_size = MLX5_CAP_FLOWTABLE(dev, max_encap_header_size);
+
+	if (params->size > max_encap_size) {
+		mlx5_core_warn(dev, "encap size %zd too big, max supported is %d\n",
+			       params->size, max_encap_size);
+		return -EINVAL;
+	}
+
+	in = kzalloc(MLX5_ST_SZ_BYTES(alloc_packet_reformat_context_in) +
+		     params->size, GFP_KERNEL);
+	if (!in)
+		return -ENOMEM;
+
+	packet_reformat_context_in = MLX5_ADDR_OF(alloc_packet_reformat_context_in,
+						  in, packet_reformat_context);
+	reformat = MLX5_ADDR_OF(packet_reformat_context_in,
+				packet_reformat_context_in,
+				reformat_data);
+	inlen = reformat - (void *)in + params->size;
+
+	MLX5_SET(alloc_packet_reformat_context_in, in, opcode,
+		 MLX5_CMD_OP_ALLOC_PACKET_REFORMAT_CONTEXT);
+	MLX5_SET(packet_reformat_context_in, packet_reformat_context_in,
+		 reformat_data_size, params->size);
+	MLX5_SET(packet_reformat_context_in, packet_reformat_context_in,
+		 reformat_type, params->type);
+	MLX5_SET(packet_reformat_context_in, packet_reformat_context_in,
+		 reformat_param_0, params->param_0);
+	MLX5_SET(packet_reformat_context_in, packet_reformat_context_in,
+		 reformat_param_1, params->param_1);
+	if (params->data && params->size)
+		memcpy(reformat, params->data, params->size);
+
+	err = mlx5_cmd_exec(dev, in, inlen, out, sizeof(out));
+
+	pkt_reformat->id = MLX5_GET(alloc_packet_reformat_context_out,
+				    out, packet_reformat_id);
+	kfree(in);
+	return err;
+}
+
+static void mlx5_cmd_packet_reformat_dealloc(struct mlx5_flow_root_namespace *ns,
+					     struct mlx5_pkt_reformat *pkt_reformat)
+{
+	u32 in[MLX5_ST_SZ_DW(dealloc_packet_reformat_context_in)] = {};
+	struct mlx5_core_dev *dev = ns->dev;
+
+	MLX5_SET(dealloc_packet_reformat_context_in, in, opcode,
+		 MLX5_CMD_OP_DEALLOC_PACKET_REFORMAT_CONTEXT);
+	MLX5_SET(dealloc_packet_reformat_context_in, in, packet_reformat_id,
+		 pkt_reformat->id);
+
+	mlx5_cmd_exec_in(dev, dealloc_packet_reformat_context, in);
+}
+
+static int mlx5_cmd_modify_header_alloc(struct mlx5_flow_root_namespace *ns,
+					u8 namespace, u8 num_actions,
+					void *modify_actions,
+					struct mlx5_modify_hdr *modify_hdr)
+{
+	u32 out[MLX5_ST_SZ_DW(alloc_modify_header_context_out)] = {};
+	int max_actions, actions_size, inlen, err;
+	struct mlx5_core_dev *dev = ns->dev;
+	void *actions_in;
+	u8 table_type;
+	u32 *in;
+
+	switch (namespace) {
+	case MLX5_FLOW_NAMESPACE_FDB:
+	case MLX5_FLOW_NAMESPACE_FDB_BYPASS:
+		max_actions = MLX5_CAP_ESW_FLOWTABLE_FDB(dev, max_modify_header_actions);
+		table_type = FS_FT_FDB;
+		break;
+	case MLX5_FLOW_NAMESPACE_KERNEL_RX_MACSEC:
+	case MLX5_FLOW_NAMESPACE_KERNEL:
+	case MLX5_FLOW_NAMESPACE_BYPASS:
+		max_actions = MLX5_CAP_FLOWTABLE_NIC_RX(dev, max_modify_header_actions);
+		table_type = FS_FT_NIC_RX;
+		break;
+	case MLX5_FLOW_NAMESPACE_EGRESS:
+	case MLX5_FLOW_NAMESPACE_EGRESS_IPSEC:
+	case MLX5_FLOW_NAMESPACE_EGRESS_MACSEC:
+		max_actions = MLX5_CAP_FLOWTABLE_NIC_TX(dev, max_modify_header_actions);
+		table_type = FS_FT_NIC_TX;
+		break;
+	case MLX5_FLOW_NAMESPACE_ESW_INGRESS:
+		max_actions = MLX5_CAP_ESW_INGRESS_ACL(dev, max_modify_header_actions);
+		table_type = FS_FT_ESW_INGRESS_ACL;
+		break;
+	case MLX5_FLOW_NAMESPACE_RDMA_TX:
+		max_actions = MLX5_CAP_FLOWTABLE_RDMA_TX(dev, max_modify_header_actions);
+		table_type = FS_FT_RDMA_TX;
+		break;
+	default:
+		return -EOPNOTSUPP;
+	}
+
+	if (num_actions > max_actions) {
+		mlx5_core_warn(dev, "too many modify header actions %d, max supported %d\n",
+			       num_actions, max_actions);
+		return -EOPNOTSUPP;
+	}
+
+	actions_size = MLX5_UN_SZ_BYTES(set_add_copy_action_in_auto) * num_actions;
+	inlen = MLX5_ST_SZ_BYTES(alloc_modify_header_context_in) + actions_size;
+
+	in = kzalloc(inlen, GFP_KERNEL);
+	if (!in)
+		return -ENOMEM;
+
+	MLX5_SET(alloc_modify_header_context_in, in, opcode,
+		 MLX5_CMD_OP_ALLOC_MODIFY_HEADER_CONTEXT);
+	MLX5_SET(alloc_modify_header_context_in, in, table_type, table_type);
+	MLX5_SET(alloc_modify_header_context_in, in, num_of_actions, num_actions);
+
+	actions_in = MLX5_ADDR_OF(alloc_modify_header_context_in, in, actions);
+	memcpy(actions_in, modify_actions, actions_size);
+
+	err = mlx5_cmd_exec(dev, in, inlen, out, sizeof(out));
+
+	modify_hdr->id = MLX5_GET(alloc_modify_header_context_out, out, modify_header_id);
+	kfree(in);
+	return err;
+}
+
+static void mlx5_cmd_modify_header_dealloc(struct mlx5_flow_root_namespace *ns,
+					   struct mlx5_modify_hdr *modify_hdr)
+{
+	u32 in[MLX5_ST_SZ_DW(dealloc_modify_header_context_in)] = {};
+	struct mlx5_core_dev *dev = ns->dev;
+
+	MLX5_SET(dealloc_modify_header_context_in, in, opcode,
+		 MLX5_CMD_OP_DEALLOC_MODIFY_HEADER_CONTEXT);
+	MLX5_SET(dealloc_modify_header_context_in, in, modify_header_id,
+		 modify_hdr->id);
+
+	mlx5_cmd_exec_in(dev, dealloc_modify_header_context, in);
+}
+
+static int mlx5_cmd_destroy_match_definer(struct mlx5_flow_root_namespace *ns,
+					  int definer_id)
+{
+	u32 in[MLX5_ST_SZ_DW(general_obj_in_cmd_hdr)] = {};
+	u32 out[MLX5_ST_SZ_DW(general_obj_out_cmd_hdr)];
+
+	MLX5_SET(general_obj_in_cmd_hdr, in, opcode,
+		 MLX5_CMD_OP_DESTROY_GENERAL_OBJECT);
+	MLX5_SET(general_obj_in_cmd_hdr, in, obj_type,
+		 MLX5_OBJ_TYPE_MATCH_DEFINER);
+	MLX5_SET(general_obj_in_cmd_hdr, in, obj_id, definer_id);
+
+	return mlx5_cmd_exec(ns->dev, in, sizeof(in), out, sizeof(out));
+}
+
+static int mlx5_cmd_create_match_definer(struct mlx5_flow_root_namespace *ns,
+					 u16 format_id, u32 *match_mask)
+{
+	u32 out[MLX5_ST_SZ_DW(create_match_definer_out)] = {};
+	u32 in[MLX5_ST_SZ_DW(create_match_definer_in)] = {};
+	struct mlx5_core_dev *dev = ns->dev;
+	void *ptr;
+	int err;
+
+	MLX5_SET(create_match_definer_in, in, general_obj_in_cmd_hdr.opcode,
+		 MLX5_CMD_OP_CREATE_GENERAL_OBJECT);
+	MLX5_SET(create_match_definer_in, in, general_obj_in_cmd_hdr.obj_type,
+		 MLX5_OBJ_TYPE_MATCH_DEFINER);
+
+	ptr = MLX5_ADDR_OF(create_match_definer_in, in, obj_context);
+	MLX5_SET(match_definer, ptr, format_id, format_id);
+
+	ptr = MLX5_ADDR_OF(match_definer, ptr, match_mask);
+	memcpy(ptr, match_mask, MLX5_FLD_SZ_BYTES(match_definer, match_mask));
+
+	err = mlx5_cmd_exec_inout(dev, create_match_definer, in, out);
+	return err ? err : MLX5_GET(general_obj_out_cmd_hdr, out, obj_id);
+}
+
+static u32 mlx5_cmd_get_capabilities(struct mlx5_flow_root_namespace *ns,
+				     enum fs_flow_table_type ft_type)
+{
+	return 0;
+}
+
+static const struct mlx5_flow_cmds mlx5_flow_cmds = {
+	.create_flow_table = mlx5_cmd_create_flow_table,
+	.destroy_flow_table = mlx5_cmd_destroy_flow_table,
+	.modify_flow_table = mlx5_cmd_modify_flow_table,
+	.create_flow_group = mlx5_cmd_create_flow_group,
+	.destroy_flow_group = mlx5_cmd_destroy_flow_group,
+	.create_fte = mlx5_cmd_create_fte,
+	.update_fte = mlx5_cmd_update_fte,
+	.delete_fte = mlx5_cmd_delete_fte,
+	.update_root_ft = mlx5_cmd_update_root_ft,
+	.packet_reformat_alloc = mlx5_cmd_packet_reformat_alloc,
+	.packet_reformat_dealloc = mlx5_cmd_packet_reformat_dealloc,
+	.modify_header_alloc = mlx5_cmd_modify_header_alloc,
+	.modify_header_dealloc = mlx5_cmd_modify_header_dealloc,
+	.create_match_definer = mlx5_cmd_create_match_definer,
+	.destroy_match_definer = mlx5_cmd_destroy_match_definer,
+	.set_peer = mlx5_cmd_stub_set_peer,
+	.create_ns = mlx5_cmd_stub_create_ns,
+	.destroy_ns = mlx5_cmd_stub_destroy_ns,
+	.get_capabilities = mlx5_cmd_get_capabilities,
+};
+
+static const struct mlx5_flow_cmds mlx5_flow_cmd_stubs = {
+	.create_flow_table = mlx5_cmd_stub_create_flow_table,
+	.destroy_flow_table = mlx5_cmd_stub_destroy_flow_table,
+	.modify_flow_table = mlx5_cmd_stub_modify_flow_table,
+	.create_flow_group = mlx5_cmd_stub_create_flow_group,
+	.destroy_flow_group = mlx5_cmd_stub_destroy_flow_group,
+	.create_fte = mlx5_cmd_stub_create_fte,
+	.update_fte = mlx5_cmd_stub_update_fte,
+	.delete_fte = mlx5_cmd_stub_delete_fte,
+	.update_root_ft = mlx5_cmd_stub_update_root_ft,
+	.packet_reformat_alloc = mlx5_cmd_stub_packet_reformat_alloc,
+	.packet_reformat_dealloc = mlx5_cmd_stub_packet_reformat_dealloc,
+	.modify_header_alloc = mlx5_cmd_stub_modify_header_alloc,
+	.modify_header_dealloc = mlx5_cmd_stub_modify_header_dealloc,
+	.create_match_definer = mlx5_cmd_stub_create_match_definer,
+	.destroy_match_definer = mlx5_cmd_stub_destroy_match_definer,
+	.set_peer = mlx5_cmd_stub_set_peer,
+	.create_ns = mlx5_cmd_stub_create_ns,
+	.destroy_ns = mlx5_cmd_stub_destroy_ns,
+	.get_capabilities = mlx5_cmd_stub_get_capabilities,
+};
+
+const struct mlx5_flow_cmds *mlx5_fs_cmd_get_fw_cmds(void)
+{
+	return &mlx5_flow_cmds;
+}
+
+static const struct mlx5_flow_cmds *mlx5_fs_cmd_get_stub_cmds(void)
+{
+	return &mlx5_flow_cmd_stubs;
+}
+
+const struct mlx5_flow_cmds *mlx5_fs_cmd_get_default(enum fs_flow_table_type type)
+{
+	switch (type) {
+	case FS_FT_NIC_RX:
+	case FS_FT_ESW_EGRESS_ACL:
+	case FS_FT_ESW_INGRESS_ACL:
+	case FS_FT_FDB:
+	case FS_FT_SNIFFER_RX:
+	case FS_FT_SNIFFER_TX:
+	case FS_FT_NIC_TX:
+	case FS_FT_RDMA_RX:
+	case FS_FT_RDMA_TX:
+	case FS_FT_PORT_SEL:
+		return mlx5_fs_cmd_get_fw_cmds();
+	default:
+		return mlx5_fs_cmd_get_stub_cmds();
+	}
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fs_cmd.h b/drivers/net/ethernet/mellanox/mlx5/core/fs_cmd.h
new file mode 100644
index 000000000..8ef4254b9
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fs_cmd.h
@@ -0,0 +1,124 @@
+/*
+ * Copyright (c) 2015, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef _MLX5_FS_CMD_
+#define _MLX5_FS_CMD_
+
+#include "fs_core.h"
+
+struct mlx5_flow_cmds {
+	int (*create_flow_table)(struct mlx5_flow_root_namespace *ns,
+				 struct mlx5_flow_table *ft,
+				 struct mlx5_flow_table_attr *ft_attr,
+				 struct mlx5_flow_table *next_ft);
+	int (*destroy_flow_table)(struct mlx5_flow_root_namespace *ns,
+				  struct mlx5_flow_table *ft);
+
+	int (*modify_flow_table)(struct mlx5_flow_root_namespace *ns,
+				 struct mlx5_flow_table *ft,
+				 struct mlx5_flow_table *next_ft);
+
+	int (*create_flow_group)(struct mlx5_flow_root_namespace *ns,
+				 struct mlx5_flow_table *ft,
+				 u32 *in,
+				 struct mlx5_flow_group *fg);
+
+	int (*destroy_flow_group)(struct mlx5_flow_root_namespace *ns,
+				  struct mlx5_flow_table *ft,
+				  struct mlx5_flow_group *fg);
+
+	int (*create_fte)(struct mlx5_flow_root_namespace *ns,
+			  struct mlx5_flow_table *ft,
+			  struct mlx5_flow_group *fg,
+			  struct fs_fte *fte);
+
+	int (*update_fte)(struct mlx5_flow_root_namespace *ns,
+			  struct mlx5_flow_table *ft,
+			  struct mlx5_flow_group *fg,
+			  int modify_mask,
+			  struct fs_fte *fte);
+
+	int (*delete_fte)(struct mlx5_flow_root_namespace *ns,
+			  struct mlx5_flow_table *ft,
+			  struct fs_fte *fte);
+
+	int (*update_root_ft)(struct mlx5_flow_root_namespace *ns,
+			      struct mlx5_flow_table *ft,
+			      u32 underlay_qpn,
+			      bool disconnect);
+
+	int (*packet_reformat_alloc)(struct mlx5_flow_root_namespace *ns,
+				     struct mlx5_pkt_reformat_params *params,
+				     enum mlx5_flow_namespace_type namespace,
+				     struct mlx5_pkt_reformat *pkt_reformat);
+
+	void (*packet_reformat_dealloc)(struct mlx5_flow_root_namespace *ns,
+					struct mlx5_pkt_reformat *pkt_reformat);
+
+	int (*modify_header_alloc)(struct mlx5_flow_root_namespace *ns,
+				   u8 namespace, u8 num_actions,
+				   void *modify_actions,
+				   struct mlx5_modify_hdr *modify_hdr);
+
+	void (*modify_header_dealloc)(struct mlx5_flow_root_namespace *ns,
+				      struct mlx5_modify_hdr *modify_hdr);
+
+	int (*set_peer)(struct mlx5_flow_root_namespace *ns,
+			struct mlx5_flow_root_namespace *peer_ns);
+
+	int (*create_ns)(struct mlx5_flow_root_namespace *ns);
+	int (*destroy_ns)(struct mlx5_flow_root_namespace *ns);
+	int (*create_match_definer)(struct mlx5_flow_root_namespace *ns,
+				    u16 format_id, u32 *match_mask);
+	int (*destroy_match_definer)(struct mlx5_flow_root_namespace *ns,
+				     int definer_id);
+
+	u32 (*get_capabilities)(struct mlx5_flow_root_namespace *ns,
+				enum fs_flow_table_type ft_type);
+};
+
+int mlx5_cmd_fc_alloc(struct mlx5_core_dev *dev, u32 *id);
+int mlx5_cmd_fc_bulk_alloc(struct mlx5_core_dev *dev,
+			   enum mlx5_fc_bulk_alloc_bitmask alloc_bitmask,
+			   u32 *id);
+int mlx5_cmd_fc_free(struct mlx5_core_dev *dev, u32 id);
+int mlx5_cmd_fc_query(struct mlx5_core_dev *dev, u32 id,
+		      u64 *packets, u64 *bytes);
+
+int mlx5_cmd_fc_get_bulk_query_out_len(int bulk_len);
+int mlx5_cmd_fc_bulk_query(struct mlx5_core_dev *dev, u32 base_id, int bulk_len,
+			   u32 *out);
+
+const struct mlx5_flow_cmds *mlx5_fs_cmd_get_default(enum fs_flow_table_type type);
+const struct mlx5_flow_cmds *mlx5_fs_cmd_get_fw_cmds(void);
+
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c b/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c
new file mode 100644
index 000000000..e6674118b
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c
@@ -0,0 +1,3612 @@
+/*
+ * Copyright (c) 2015, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/mutex.h>
+#include <linux/mlx5/driver.h>
+#include <linux/mlx5/vport.h>
+#include <linux/mlx5/eswitch.h>
+
+#include "mlx5_core.h"
+#include "fs_core.h"
+#include "fs_cmd.h"
+#include "fs_ft_pool.h"
+#include "diag/fs_tracepoint.h"
+
+#define INIT_TREE_NODE_ARRAY_SIZE(...)	(sizeof((struct init_tree_node[]){__VA_ARGS__}) /\
+					 sizeof(struct init_tree_node))
+
+#define ADD_PRIO(num_prios_val, min_level_val, num_levels_val, caps_val,\
+		 ...) {.type = FS_TYPE_PRIO,\
+	.min_ft_level = min_level_val,\
+	.num_levels = num_levels_val,\
+	.num_leaf_prios = num_prios_val,\
+	.caps = caps_val,\
+	.children = (struct init_tree_node[]) {__VA_ARGS__},\
+	.ar_size = INIT_TREE_NODE_ARRAY_SIZE(__VA_ARGS__) \
+}
+
+#define ADD_MULTIPLE_PRIO(num_prios_val, num_levels_val, ...)\
+	ADD_PRIO(num_prios_val, 0, num_levels_val, {},\
+		 __VA_ARGS__)\
+
+#define ADD_NS(def_miss_act, ...) {.type = FS_TYPE_NAMESPACE,	\
+	.def_miss_action = def_miss_act,\
+	.children = (struct init_tree_node[]) {__VA_ARGS__},\
+	.ar_size = INIT_TREE_NODE_ARRAY_SIZE(__VA_ARGS__) \
+}
+
+#define INIT_CAPS_ARRAY_SIZE(...) (sizeof((long[]){__VA_ARGS__}) /\
+				   sizeof(long))
+
+#define FS_CAP(cap) (__mlx5_bit_off(flow_table_nic_cap, cap))
+
+#define FS_REQUIRED_CAPS(...) {.arr_sz = INIT_CAPS_ARRAY_SIZE(__VA_ARGS__), \
+			       .caps = (long[]) {__VA_ARGS__} }
+
+#define FS_CHAINING_CAPS  FS_REQUIRED_CAPS(FS_CAP(flow_table_properties_nic_receive.flow_modify_en), \
+					   FS_CAP(flow_table_properties_nic_receive.modify_root), \
+					   FS_CAP(flow_table_properties_nic_receive.identified_miss_table_mode), \
+					   FS_CAP(flow_table_properties_nic_receive.flow_table_modify))
+
+#define FS_CHAINING_CAPS_EGRESS                                                \
+	FS_REQUIRED_CAPS(                                                      \
+		FS_CAP(flow_table_properties_nic_transmit.flow_modify_en),     \
+		FS_CAP(flow_table_properties_nic_transmit.modify_root),        \
+		FS_CAP(flow_table_properties_nic_transmit                      \
+			       .identified_miss_table_mode),                   \
+		FS_CAP(flow_table_properties_nic_transmit.flow_table_modify))
+
+#define FS_CHAINING_CAPS_RDMA_TX                                                \
+	FS_REQUIRED_CAPS(                                                       \
+		FS_CAP(flow_table_properties_nic_transmit_rdma.flow_modify_en), \
+		FS_CAP(flow_table_properties_nic_transmit_rdma.modify_root),    \
+		FS_CAP(flow_table_properties_nic_transmit_rdma                  \
+			       .identified_miss_table_mode),                    \
+		FS_CAP(flow_table_properties_nic_transmit_rdma                  \
+			       .flow_table_modify))
+
+#define LEFTOVERS_NUM_LEVELS 1
+#define LEFTOVERS_NUM_PRIOS 1
+
+#define RDMA_RX_COUNTERS_PRIO_NUM_LEVELS 1
+#define RDMA_TX_COUNTERS_PRIO_NUM_LEVELS 1
+
+#define BY_PASS_PRIO_NUM_LEVELS 1
+#define BY_PASS_MIN_LEVEL (ETHTOOL_MIN_LEVEL + MLX5_BY_PASS_NUM_PRIOS +\
+			   LEFTOVERS_NUM_PRIOS)
+
+#define KERNEL_RX_MACSEC_NUM_PRIOS  1
+#define KERNEL_RX_MACSEC_NUM_LEVELS 2
+#define KERNEL_RX_MACSEC_MIN_LEVEL (BY_PASS_MIN_LEVEL + KERNEL_RX_MACSEC_NUM_PRIOS)
+
+#define ETHTOOL_PRIO_NUM_LEVELS 1
+#define ETHTOOL_NUM_PRIOS 11
+#define ETHTOOL_MIN_LEVEL (KERNEL_MIN_LEVEL + ETHTOOL_NUM_PRIOS)
+/* Promiscuous, Vlan, mac, ttc, inner ttc, {UDP/ANY/aRFS/accel/{esp, esp_err}} */
+#define KERNEL_NIC_PRIO_NUM_LEVELS 7
+#define KERNEL_NIC_NUM_PRIOS 1
+/* One more level for tc */
+#define KERNEL_MIN_LEVEL (KERNEL_NIC_PRIO_NUM_LEVELS + 1)
+
+#define KERNEL_NIC_TC_NUM_PRIOS  1
+#define KERNEL_NIC_TC_NUM_LEVELS 3
+
+#define ANCHOR_NUM_LEVELS 1
+#define ANCHOR_NUM_PRIOS 1
+#define ANCHOR_MIN_LEVEL (BY_PASS_MIN_LEVEL + 1)
+
+#define OFFLOADS_MAX_FT 2
+#define OFFLOADS_NUM_PRIOS 2
+#define OFFLOADS_MIN_LEVEL (ANCHOR_MIN_LEVEL + OFFLOADS_NUM_PRIOS)
+
+#define LAG_PRIO_NUM_LEVELS 1
+#define LAG_NUM_PRIOS 1
+#define LAG_MIN_LEVEL (OFFLOADS_MIN_LEVEL + KERNEL_RX_MACSEC_MIN_LEVEL + 1)
+
+#define KERNEL_TX_IPSEC_NUM_PRIOS  1
+#define KERNEL_TX_IPSEC_NUM_LEVELS 1
+#define KERNEL_TX_IPSEC_MIN_LEVEL        (KERNEL_TX_IPSEC_NUM_LEVELS)
+
+#define KERNEL_TX_MACSEC_NUM_PRIOS  1
+#define KERNEL_TX_MACSEC_NUM_LEVELS 2
+#define KERNEL_TX_MACSEC_MIN_LEVEL       (KERNEL_TX_IPSEC_MIN_LEVEL + KERNEL_TX_MACSEC_NUM_PRIOS)
+
+struct node_caps {
+	size_t	arr_sz;
+	long	*caps;
+};
+
+static struct init_tree_node {
+	enum fs_node_type	type;
+	struct init_tree_node *children;
+	int ar_size;
+	struct node_caps caps;
+	int min_ft_level;
+	int num_leaf_prios;
+	int prio;
+	int num_levels;
+	enum mlx5_flow_table_miss_action def_miss_action;
+} root_fs = {
+	.type = FS_TYPE_NAMESPACE,
+	.ar_size = 8,
+	  .children = (struct init_tree_node[]){
+		  ADD_PRIO(0, BY_PASS_MIN_LEVEL, 0, FS_CHAINING_CAPS,
+			   ADD_NS(MLX5_FLOW_TABLE_MISS_ACTION_DEF,
+				  ADD_MULTIPLE_PRIO(MLX5_BY_PASS_NUM_PRIOS,
+						    BY_PASS_PRIO_NUM_LEVELS))),
+		  ADD_PRIO(0, KERNEL_RX_MACSEC_MIN_LEVEL, 0, FS_CHAINING_CAPS,
+			   ADD_NS(MLX5_FLOW_TABLE_MISS_ACTION_DEF,
+				  ADD_MULTIPLE_PRIO(KERNEL_RX_MACSEC_NUM_PRIOS,
+						    KERNEL_RX_MACSEC_NUM_LEVELS))),
+		  ADD_PRIO(0, LAG_MIN_LEVEL, 0, FS_CHAINING_CAPS,
+			   ADD_NS(MLX5_FLOW_TABLE_MISS_ACTION_DEF,
+				  ADD_MULTIPLE_PRIO(LAG_NUM_PRIOS,
+						    LAG_PRIO_NUM_LEVELS))),
+		  ADD_PRIO(0, OFFLOADS_MIN_LEVEL, 0, FS_CHAINING_CAPS,
+			   ADD_NS(MLX5_FLOW_TABLE_MISS_ACTION_DEF,
+				  ADD_MULTIPLE_PRIO(OFFLOADS_NUM_PRIOS,
+						    OFFLOADS_MAX_FT))),
+		  ADD_PRIO(0, ETHTOOL_MIN_LEVEL, 0, FS_CHAINING_CAPS,
+			   ADD_NS(MLX5_FLOW_TABLE_MISS_ACTION_DEF,
+				  ADD_MULTIPLE_PRIO(ETHTOOL_NUM_PRIOS,
+						    ETHTOOL_PRIO_NUM_LEVELS))),
+		  ADD_PRIO(0, KERNEL_MIN_LEVEL, 0, {},
+			   ADD_NS(MLX5_FLOW_TABLE_MISS_ACTION_DEF,
+				  ADD_MULTIPLE_PRIO(KERNEL_NIC_TC_NUM_PRIOS,
+						    KERNEL_NIC_TC_NUM_LEVELS),
+				  ADD_MULTIPLE_PRIO(KERNEL_NIC_NUM_PRIOS,
+						    KERNEL_NIC_PRIO_NUM_LEVELS))),
+		  ADD_PRIO(0, BY_PASS_MIN_LEVEL, 0, FS_CHAINING_CAPS,
+			   ADD_NS(MLX5_FLOW_TABLE_MISS_ACTION_DEF,
+				  ADD_MULTIPLE_PRIO(LEFTOVERS_NUM_PRIOS,
+						    LEFTOVERS_NUM_LEVELS))),
+		  ADD_PRIO(0, ANCHOR_MIN_LEVEL, 0, {},
+			   ADD_NS(MLX5_FLOW_TABLE_MISS_ACTION_DEF,
+				  ADD_MULTIPLE_PRIO(ANCHOR_NUM_PRIOS,
+						    ANCHOR_NUM_LEVELS))),
+	}
+};
+
+static struct init_tree_node egress_root_fs = {
+	.type = FS_TYPE_NAMESPACE,
+	.ar_size = 3,
+	.children = (struct init_tree_node[]) {
+		ADD_PRIO(0, MLX5_BY_PASS_NUM_PRIOS, 0,
+			 FS_CHAINING_CAPS_EGRESS,
+			 ADD_NS(MLX5_FLOW_TABLE_MISS_ACTION_DEF,
+				ADD_MULTIPLE_PRIO(MLX5_BY_PASS_NUM_PRIOS,
+						  BY_PASS_PRIO_NUM_LEVELS))),
+		ADD_PRIO(0, KERNEL_TX_IPSEC_MIN_LEVEL, 0,
+			 FS_CHAINING_CAPS_EGRESS,
+			 ADD_NS(MLX5_FLOW_TABLE_MISS_ACTION_DEF,
+				ADD_MULTIPLE_PRIO(KERNEL_TX_IPSEC_NUM_PRIOS,
+						  KERNEL_TX_IPSEC_NUM_LEVELS))),
+		ADD_PRIO(0, KERNEL_TX_MACSEC_MIN_LEVEL, 0,
+			 FS_CHAINING_CAPS_EGRESS,
+			 ADD_NS(MLX5_FLOW_TABLE_MISS_ACTION_DEF,
+				ADD_MULTIPLE_PRIO(KERNEL_TX_MACSEC_NUM_PRIOS,
+						  KERNEL_TX_MACSEC_NUM_LEVELS))),
+	}
+};
+
+enum {
+	RDMA_RX_COUNTERS_PRIO,
+	RDMA_RX_BYPASS_PRIO,
+	RDMA_RX_KERNEL_PRIO,
+};
+
+#define RDMA_RX_BYPASS_MIN_LEVEL MLX5_BY_PASS_NUM_REGULAR_PRIOS
+#define RDMA_RX_KERNEL_MIN_LEVEL (RDMA_RX_BYPASS_MIN_LEVEL + 1)
+#define RDMA_RX_COUNTERS_MIN_LEVEL (RDMA_RX_KERNEL_MIN_LEVEL + 2)
+
+static struct init_tree_node rdma_rx_root_fs = {
+	.type = FS_TYPE_NAMESPACE,
+	.ar_size = 3,
+	.children = (struct init_tree_node[]) {
+		[RDMA_RX_COUNTERS_PRIO] =
+		ADD_PRIO(0, RDMA_RX_COUNTERS_MIN_LEVEL, 0,
+			 FS_CHAINING_CAPS,
+			 ADD_NS(MLX5_FLOW_TABLE_MISS_ACTION_DEF,
+				ADD_MULTIPLE_PRIO(MLX5_RDMA_RX_NUM_COUNTERS_PRIOS,
+						  RDMA_RX_COUNTERS_PRIO_NUM_LEVELS))),
+		[RDMA_RX_BYPASS_PRIO] =
+		ADD_PRIO(0, RDMA_RX_BYPASS_MIN_LEVEL, 0,
+			 FS_CHAINING_CAPS,
+			 ADD_NS(MLX5_FLOW_TABLE_MISS_ACTION_DEF,
+				ADD_MULTIPLE_PRIO(MLX5_BY_PASS_NUM_REGULAR_PRIOS,
+						  BY_PASS_PRIO_NUM_LEVELS))),
+		[RDMA_RX_KERNEL_PRIO] =
+		ADD_PRIO(0, RDMA_RX_KERNEL_MIN_LEVEL, 0,
+			 FS_CHAINING_CAPS,
+			 ADD_NS(MLX5_FLOW_TABLE_MISS_ACTION_SWITCH_DOMAIN,
+				ADD_MULTIPLE_PRIO(1, 1))),
+	}
+};
+
+enum {
+	RDMA_TX_COUNTERS_PRIO,
+	RDMA_TX_BYPASS_PRIO,
+};
+
+#define RDMA_TX_BYPASS_MIN_LEVEL MLX5_BY_PASS_NUM_PRIOS
+#define RDMA_TX_COUNTERS_MIN_LEVEL (RDMA_TX_BYPASS_MIN_LEVEL + 1)
+
+static struct init_tree_node rdma_tx_root_fs = {
+	.type = FS_TYPE_NAMESPACE,
+	.ar_size = 2,
+	.children = (struct init_tree_node[]) {
+		[RDMA_TX_COUNTERS_PRIO] =
+		ADD_PRIO(0, RDMA_TX_COUNTERS_MIN_LEVEL, 0,
+			 FS_CHAINING_CAPS,
+			 ADD_NS(MLX5_FLOW_TABLE_MISS_ACTION_DEF,
+				ADD_MULTIPLE_PRIO(MLX5_RDMA_TX_NUM_COUNTERS_PRIOS,
+						  RDMA_TX_COUNTERS_PRIO_NUM_LEVELS))),
+		[RDMA_TX_BYPASS_PRIO] =
+		ADD_PRIO(0, RDMA_TX_BYPASS_MIN_LEVEL, 0,
+			 FS_CHAINING_CAPS_RDMA_TX,
+			 ADD_NS(MLX5_FLOW_TABLE_MISS_ACTION_DEF,
+				ADD_MULTIPLE_PRIO(RDMA_TX_BYPASS_MIN_LEVEL,
+						  BY_PASS_PRIO_NUM_LEVELS))),
+	}
+};
+
+enum fs_i_lock_class {
+	FS_LOCK_GRANDPARENT,
+	FS_LOCK_PARENT,
+	FS_LOCK_CHILD
+};
+
+static const struct rhashtable_params rhash_fte = {
+	.key_len = sizeof_field(struct fs_fte, val),
+	.key_offset = offsetof(struct fs_fte, val),
+	.head_offset = offsetof(struct fs_fte, hash),
+	.automatic_shrinking = true,
+	.min_size = 1,
+};
+
+static const struct rhashtable_params rhash_fg = {
+	.key_len = sizeof_field(struct mlx5_flow_group, mask),
+	.key_offset = offsetof(struct mlx5_flow_group, mask),
+	.head_offset = offsetof(struct mlx5_flow_group, hash),
+	.automatic_shrinking = true,
+	.min_size = 1,
+
+};
+
+static void del_hw_flow_table(struct fs_node *node);
+static void del_hw_flow_group(struct fs_node *node);
+static void del_hw_fte(struct fs_node *node);
+static void del_sw_flow_table(struct fs_node *node);
+static void del_sw_flow_group(struct fs_node *node);
+static void del_sw_fte(struct fs_node *node);
+static void del_sw_prio(struct fs_node *node);
+static void del_sw_ns(struct fs_node *node);
+/* Delete rule (destination) is special case that
+ * requires to lock the FTE for all the deletion process.
+ */
+static void del_sw_hw_rule(struct fs_node *node);
+static bool mlx5_flow_dests_cmp(struct mlx5_flow_destination *d1,
+				struct mlx5_flow_destination *d2);
+static void cleanup_root_ns(struct mlx5_flow_root_namespace *root_ns);
+static struct mlx5_flow_rule *
+find_flow_rule(struct fs_fte *fte,
+	       struct mlx5_flow_destination *dest);
+
+static void tree_init_node(struct fs_node *node,
+			   void (*del_hw_func)(struct fs_node *),
+			   void (*del_sw_func)(struct fs_node *))
+{
+	refcount_set(&node->refcount, 1);
+	INIT_LIST_HEAD(&node->list);
+	INIT_LIST_HEAD(&node->children);
+	init_rwsem(&node->lock);
+	node->del_hw_func = del_hw_func;
+	node->del_sw_func = del_sw_func;
+	node->active = false;
+}
+
+static void tree_add_node(struct fs_node *node, struct fs_node *parent)
+{
+	if (parent)
+		refcount_inc(&parent->refcount);
+	node->parent = parent;
+
+	/* Parent is the root */
+	if (!parent)
+		node->root = node;
+	else
+		node->root = parent->root;
+}
+
+static int tree_get_node(struct fs_node *node)
+{
+	return refcount_inc_not_zero(&node->refcount);
+}
+
+static void nested_down_read_ref_node(struct fs_node *node,
+				      enum fs_i_lock_class class)
+{
+	if (node) {
+		down_read_nested(&node->lock, class);
+		refcount_inc(&node->refcount);
+	}
+}
+
+static void nested_down_write_ref_node(struct fs_node *node,
+				       enum fs_i_lock_class class)
+{
+	if (node) {
+		down_write_nested(&node->lock, class);
+		refcount_inc(&node->refcount);
+	}
+}
+
+static void down_write_ref_node(struct fs_node *node, bool locked)
+{
+	if (node) {
+		if (!locked)
+			down_write(&node->lock);
+		refcount_inc(&node->refcount);
+	}
+}
+
+static void up_read_ref_node(struct fs_node *node)
+{
+	refcount_dec(&node->refcount);
+	up_read(&node->lock);
+}
+
+static void up_write_ref_node(struct fs_node *node, bool locked)
+{
+	refcount_dec(&node->refcount);
+	if (!locked)
+		up_write(&node->lock);
+}
+
+static void tree_put_node(struct fs_node *node, bool locked)
+{
+	struct fs_node *parent_node = node->parent;
+
+	if (refcount_dec_and_test(&node->refcount)) {
+		if (node->del_hw_func)
+			node->del_hw_func(node);
+		if (parent_node) {
+			down_write_ref_node(parent_node, locked);
+			list_del_init(&node->list);
+		}
+		node->del_sw_func(node);
+		if (parent_node)
+			up_write_ref_node(parent_node, locked);
+		node = NULL;
+	}
+	if (!node && parent_node)
+		tree_put_node(parent_node, locked);
+}
+
+static int tree_remove_node(struct fs_node *node, bool locked)
+{
+	if (refcount_read(&node->refcount) > 1) {
+		refcount_dec(&node->refcount);
+		return -EEXIST;
+	}
+	tree_put_node(node, locked);
+	return 0;
+}
+
+static struct fs_prio *find_prio(struct mlx5_flow_namespace *ns,
+				 unsigned int prio)
+{
+	struct fs_prio *iter_prio;
+
+	fs_for_each_prio(iter_prio, ns) {
+		if (iter_prio->prio == prio)
+			return iter_prio;
+	}
+
+	return NULL;
+}
+
+static bool is_fwd_next_action(u32 action)
+{
+	return action & (MLX5_FLOW_CONTEXT_ACTION_FWD_NEXT_PRIO |
+			 MLX5_FLOW_CONTEXT_ACTION_FWD_NEXT_NS);
+}
+
+static bool is_fwd_dest_type(enum mlx5_flow_destination_type type)
+{
+	return type == MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE_NUM ||
+		type == MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE ||
+		type == MLX5_FLOW_DESTINATION_TYPE_UPLINK ||
+		type == MLX5_FLOW_DESTINATION_TYPE_VPORT ||
+		type == MLX5_FLOW_DESTINATION_TYPE_FLOW_SAMPLER ||
+		type == MLX5_FLOW_DESTINATION_TYPE_TIR;
+}
+
+static bool check_valid_spec(const struct mlx5_flow_spec *spec)
+{
+	int i;
+
+	for (i = 0; i < MLX5_ST_SZ_DW_MATCH_PARAM; i++)
+		if (spec->match_value[i] & ~spec->match_criteria[i]) {
+			pr_warn("mlx5_core: match_value differs from match_criteria\n");
+			return false;
+		}
+
+	return true;
+}
+
+struct mlx5_flow_root_namespace *find_root(struct fs_node *node)
+{
+	struct fs_node *root;
+	struct mlx5_flow_namespace *ns;
+
+	root = node->root;
+
+	if (WARN_ON(root->type != FS_TYPE_NAMESPACE)) {
+		pr_warn("mlx5: flow steering node is not in tree or garbaged\n");
+		return NULL;
+	}
+
+	ns = container_of(root, struct mlx5_flow_namespace, node);
+	return container_of(ns, struct mlx5_flow_root_namespace, ns);
+}
+
+static inline struct mlx5_flow_steering *get_steering(struct fs_node *node)
+{
+	struct mlx5_flow_root_namespace *root = find_root(node);
+
+	if (root)
+		return root->dev->priv.steering;
+	return NULL;
+}
+
+static inline struct mlx5_core_dev *get_dev(struct fs_node *node)
+{
+	struct mlx5_flow_root_namespace *root = find_root(node);
+
+	if (root)
+		return root->dev;
+	return NULL;
+}
+
+static void del_sw_ns(struct fs_node *node)
+{
+	kfree(node);
+}
+
+static void del_sw_prio(struct fs_node *node)
+{
+	kfree(node);
+}
+
+static void del_hw_flow_table(struct fs_node *node)
+{
+	struct mlx5_flow_root_namespace *root;
+	struct mlx5_flow_table *ft;
+	struct mlx5_core_dev *dev;
+	int err;
+
+	fs_get_obj(ft, node);
+	dev = get_dev(&ft->node);
+	root = find_root(&ft->node);
+	trace_mlx5_fs_del_ft(ft);
+
+	if (node->active) {
+		err = root->cmds->destroy_flow_table(root, ft);
+		if (err)
+			mlx5_core_warn(dev, "flow steering can't destroy ft\n");
+	}
+}
+
+static void del_sw_flow_table(struct fs_node *node)
+{
+	struct mlx5_flow_table *ft;
+	struct fs_prio *prio;
+
+	fs_get_obj(ft, node);
+
+	rhltable_destroy(&ft->fgs_hash);
+	if (ft->node.parent) {
+		fs_get_obj(prio, ft->node.parent);
+		prio->num_ft--;
+	}
+	kfree(ft);
+}
+
+static void modify_fte(struct fs_fte *fte)
+{
+	struct mlx5_flow_root_namespace *root;
+	struct mlx5_flow_table *ft;
+	struct mlx5_flow_group *fg;
+	struct mlx5_core_dev *dev;
+	int err;
+
+	fs_get_obj(fg, fte->node.parent);
+	fs_get_obj(ft, fg->node.parent);
+	dev = get_dev(&fte->node);
+
+	root = find_root(&ft->node);
+	err = root->cmds->update_fte(root, ft, fg, fte->modify_mask, fte);
+	if (err)
+		mlx5_core_warn(dev,
+			       "%s can't del rule fg id=%d fte_index=%d\n",
+			       __func__, fg->id, fte->index);
+	fte->modify_mask = 0;
+}
+
+static void del_sw_hw_rule(struct fs_node *node)
+{
+	struct mlx5_flow_rule *rule;
+	struct fs_fte *fte;
+
+	fs_get_obj(rule, node);
+	fs_get_obj(fte, rule->node.parent);
+	trace_mlx5_fs_del_rule(rule);
+	if (is_fwd_next_action(rule->sw_action)) {
+		mutex_lock(&rule->dest_attr.ft->lock);
+		list_del(&rule->next_ft);
+		mutex_unlock(&rule->dest_attr.ft->lock);
+	}
+
+	if (rule->dest_attr.type == MLX5_FLOW_DESTINATION_TYPE_COUNTER) {
+		--fte->dests_size;
+		fte->modify_mask |=
+			BIT(MLX5_SET_FTE_MODIFY_ENABLE_MASK_ACTION) |
+			BIT(MLX5_SET_FTE_MODIFY_ENABLE_MASK_FLOW_COUNTERS);
+		fte->action.action &= ~MLX5_FLOW_CONTEXT_ACTION_COUNT;
+		goto out;
+	}
+
+	if (rule->dest_attr.type == MLX5_FLOW_DESTINATION_TYPE_PORT) {
+		--fte->dests_size;
+		fte->modify_mask |= BIT(MLX5_SET_FTE_MODIFY_ENABLE_MASK_ACTION);
+		fte->action.action &= ~MLX5_FLOW_CONTEXT_ACTION_ALLOW;
+		goto out;
+	}
+
+	if (is_fwd_dest_type(rule->dest_attr.type)) {
+		--fte->dests_size;
+		--fte->fwd_dests;
+
+		if (!fte->fwd_dests)
+			fte->action.action &=
+				~MLX5_FLOW_CONTEXT_ACTION_FWD_DEST;
+		fte->modify_mask |=
+			BIT(MLX5_SET_FTE_MODIFY_ENABLE_MASK_DESTINATION_LIST);
+		goto out;
+	}
+out:
+	kfree(rule);
+}
+
+static void del_hw_fte(struct fs_node *node)
+{
+	struct mlx5_flow_root_namespace *root;
+	struct mlx5_flow_table *ft;
+	struct mlx5_flow_group *fg;
+	struct mlx5_core_dev *dev;
+	struct fs_fte *fte;
+	int err;
+
+	fs_get_obj(fte, node);
+	fs_get_obj(fg, fte->node.parent);
+	fs_get_obj(ft, fg->node.parent);
+
+	trace_mlx5_fs_del_fte(fte);
+	WARN_ON(fte->dests_size);
+	dev = get_dev(&ft->node);
+	root = find_root(&ft->node);
+	if (node->active) {
+		err = root->cmds->delete_fte(root, ft, fte);
+		if (err)
+			mlx5_core_warn(dev,
+				       "flow steering can't delete fte in index %d of flow group id %d\n",
+				       fte->index, fg->id);
+		node->active = false;
+	}
+}
+
+static void del_sw_fte(struct fs_node *node)
+{
+	struct mlx5_flow_steering *steering = get_steering(node);
+	struct mlx5_flow_group *fg;
+	struct fs_fte *fte;
+	int err;
+
+	fs_get_obj(fte, node);
+	fs_get_obj(fg, fte->node.parent);
+
+	err = rhashtable_remove_fast(&fg->ftes_hash,
+				     &fte->hash,
+				     rhash_fte);
+	WARN_ON(err);
+	ida_free(&fg->fte_allocator, fte->index - fg->start_index);
+	kmem_cache_free(steering->ftes_cache, fte);
+}
+
+static void del_hw_flow_group(struct fs_node *node)
+{
+	struct mlx5_flow_root_namespace *root;
+	struct mlx5_flow_group *fg;
+	struct mlx5_flow_table *ft;
+	struct mlx5_core_dev *dev;
+
+	fs_get_obj(fg, node);
+	fs_get_obj(ft, fg->node.parent);
+	dev = get_dev(&ft->node);
+	trace_mlx5_fs_del_fg(fg);
+
+	root = find_root(&ft->node);
+	if (fg->node.active && root->cmds->destroy_flow_group(root, ft, fg))
+		mlx5_core_warn(dev, "flow steering can't destroy fg %d of ft %d\n",
+			       fg->id, ft->id);
+}
+
+static void del_sw_flow_group(struct fs_node *node)
+{
+	struct mlx5_flow_steering *steering = get_steering(node);
+	struct mlx5_flow_group *fg;
+	struct mlx5_flow_table *ft;
+	int err;
+
+	fs_get_obj(fg, node);
+	fs_get_obj(ft, fg->node.parent);
+
+	rhashtable_destroy(&fg->ftes_hash);
+	ida_destroy(&fg->fte_allocator);
+	if (ft->autogroup.active &&
+	    fg->max_ftes == ft->autogroup.group_size &&
+	    fg->start_index < ft->autogroup.max_fte)
+		ft->autogroup.num_groups--;
+	err = rhltable_remove(&ft->fgs_hash,
+			      &fg->hash,
+			      rhash_fg);
+	WARN_ON(err);
+	kmem_cache_free(steering->fgs_cache, fg);
+}
+
+static int insert_fte(struct mlx5_flow_group *fg, struct fs_fte *fte)
+{
+	int index;
+	int ret;
+
+	index = ida_alloc_max(&fg->fte_allocator, fg->max_ftes - 1, GFP_KERNEL);
+	if (index < 0)
+		return index;
+
+	fte->index = index + fg->start_index;
+	ret = rhashtable_insert_fast(&fg->ftes_hash,
+				     &fte->hash,
+				     rhash_fte);
+	if (ret)
+		goto err_ida_remove;
+
+	tree_add_node(&fte->node, &fg->node);
+	list_add_tail(&fte->node.list, &fg->node.children);
+	return 0;
+
+err_ida_remove:
+	ida_free(&fg->fte_allocator, index);
+	return ret;
+}
+
+static struct fs_fte *alloc_fte(struct mlx5_flow_table *ft,
+				const struct mlx5_flow_spec *spec,
+				struct mlx5_flow_act *flow_act)
+{
+	struct mlx5_flow_steering *steering = get_steering(&ft->node);
+	struct fs_fte *fte;
+
+	fte = kmem_cache_zalloc(steering->ftes_cache, GFP_KERNEL);
+	if (!fte)
+		return ERR_PTR(-ENOMEM);
+
+	memcpy(fte->val, &spec->match_value, sizeof(fte->val));
+	fte->node.type =  FS_TYPE_FLOW_ENTRY;
+	fte->action = *flow_act;
+	fte->flow_context = spec->flow_context;
+
+	tree_init_node(&fte->node, del_hw_fte, del_sw_fte);
+
+	return fte;
+}
+
+static void dealloc_flow_group(struct mlx5_flow_steering *steering,
+			       struct mlx5_flow_group *fg)
+{
+	rhashtable_destroy(&fg->ftes_hash);
+	kmem_cache_free(steering->fgs_cache, fg);
+}
+
+static struct mlx5_flow_group *alloc_flow_group(struct mlx5_flow_steering *steering,
+						u8 match_criteria_enable,
+						const void *match_criteria,
+						int start_index,
+						int end_index)
+{
+	struct mlx5_flow_group *fg;
+	int ret;
+
+	fg = kmem_cache_zalloc(steering->fgs_cache, GFP_KERNEL);
+	if (!fg)
+		return ERR_PTR(-ENOMEM);
+
+	ret = rhashtable_init(&fg->ftes_hash, &rhash_fte);
+	if (ret) {
+		kmem_cache_free(steering->fgs_cache, fg);
+		return ERR_PTR(ret);
+	}
+
+	ida_init(&fg->fte_allocator);
+	fg->mask.match_criteria_enable = match_criteria_enable;
+	memcpy(&fg->mask.match_criteria, match_criteria,
+	       sizeof(fg->mask.match_criteria));
+	fg->node.type =  FS_TYPE_FLOW_GROUP;
+	fg->start_index = start_index;
+	fg->max_ftes = end_index - start_index + 1;
+
+	return fg;
+}
+
+static struct mlx5_flow_group *alloc_insert_flow_group(struct mlx5_flow_table *ft,
+						       u8 match_criteria_enable,
+						       const void *match_criteria,
+						       int start_index,
+						       int end_index,
+						       struct list_head *prev)
+{
+	struct mlx5_flow_steering *steering = get_steering(&ft->node);
+	struct mlx5_flow_group *fg;
+	int ret;
+
+	fg = alloc_flow_group(steering, match_criteria_enable, match_criteria,
+			      start_index, end_index);
+	if (IS_ERR(fg))
+		return fg;
+
+	/* initialize refcnt, add to parent list */
+	ret = rhltable_insert(&ft->fgs_hash,
+			      &fg->hash,
+			      rhash_fg);
+	if (ret) {
+		dealloc_flow_group(steering, fg);
+		return ERR_PTR(ret);
+	}
+
+	tree_init_node(&fg->node, del_hw_flow_group, del_sw_flow_group);
+	tree_add_node(&fg->node, &ft->node);
+	/* Add node to group list */
+	list_add(&fg->node.list, prev);
+	atomic_inc(&ft->node.version);
+
+	return fg;
+}
+
+static struct mlx5_flow_table *alloc_flow_table(int level, u16 vport,
+						enum fs_flow_table_type table_type,
+						enum fs_flow_table_op_mod op_mod,
+						u32 flags)
+{
+	struct mlx5_flow_table *ft;
+	int ret;
+
+	ft  = kzalloc(sizeof(*ft), GFP_KERNEL);
+	if (!ft)
+		return ERR_PTR(-ENOMEM);
+
+	ret = rhltable_init(&ft->fgs_hash, &rhash_fg);
+	if (ret) {
+		kfree(ft);
+		return ERR_PTR(ret);
+	}
+
+	ft->level = level;
+	ft->node.type = FS_TYPE_FLOW_TABLE;
+	ft->op_mod = op_mod;
+	ft->type = table_type;
+	ft->vport = vport;
+	ft->flags = flags;
+	INIT_LIST_HEAD(&ft->fwd_rules);
+	mutex_init(&ft->lock);
+
+	return ft;
+}
+
+/* If reverse is false, then we search for the first flow table in the
+ * root sub-tree from start(closest from right), else we search for the
+ * last flow table in the root sub-tree till start(closest from left).
+ */
+static struct mlx5_flow_table *find_closest_ft_recursive(struct fs_node  *root,
+							 struct list_head *start,
+							 bool reverse)
+{
+#define list_advance_entry(pos, reverse)		\
+	((reverse) ? list_prev_entry(pos, list) : list_next_entry(pos, list))
+
+#define list_for_each_advance_continue(pos, head, reverse)	\
+	for (pos = list_advance_entry(pos, reverse);		\
+	     &pos->list != (head);				\
+	     pos = list_advance_entry(pos, reverse))
+
+	struct fs_node *iter = list_entry(start, struct fs_node, list);
+	struct mlx5_flow_table *ft = NULL;
+
+	if (!root)
+		return NULL;
+
+	list_for_each_advance_continue(iter, &root->children, reverse) {
+		if (iter->type == FS_TYPE_FLOW_TABLE) {
+			fs_get_obj(ft, iter);
+			return ft;
+		}
+		ft = find_closest_ft_recursive(iter, &iter->children, reverse);
+		if (ft)
+			return ft;
+	}
+
+	return ft;
+}
+
+static struct fs_node *find_prio_chains_parent(struct fs_node *parent,
+					       struct fs_node **child)
+{
+	struct fs_node *node = NULL;
+
+	while (parent && parent->type != FS_TYPE_PRIO_CHAINS) {
+		node = parent;
+		parent = parent->parent;
+	}
+
+	if (child)
+		*child = node;
+
+	return parent;
+}
+
+/* If reverse is false then return the first flow table next to the passed node
+ * in the tree, else return the last flow table before the node in the tree.
+ * If skip is true, skip the flow tables in the same prio_chains prio.
+ */
+static struct mlx5_flow_table *find_closest_ft(struct fs_node *node, bool reverse,
+					       bool skip)
+{
+	struct fs_node *prio_chains_parent = NULL;
+	struct mlx5_flow_table *ft = NULL;
+	struct fs_node *curr_node;
+	struct fs_node *parent;
+
+	if (skip)
+		prio_chains_parent = find_prio_chains_parent(node, NULL);
+	parent = node->parent;
+	curr_node = node;
+	while (!ft && parent) {
+		if (parent != prio_chains_parent)
+			ft = find_closest_ft_recursive(parent, &curr_node->list,
+						       reverse);
+		curr_node = parent;
+		parent = curr_node->parent;
+	}
+	return ft;
+}
+
+/* Assuming all the tree is locked by mutex chain lock */
+static struct mlx5_flow_table *find_next_chained_ft(struct fs_node *node)
+{
+	return find_closest_ft(node, false, true);
+}
+
+/* Assuming all the tree is locked by mutex chain lock */
+static struct mlx5_flow_table *find_prev_chained_ft(struct fs_node *node)
+{
+	return find_closest_ft(node, true, true);
+}
+
+static struct mlx5_flow_table *find_next_fwd_ft(struct mlx5_flow_table *ft,
+						struct mlx5_flow_act *flow_act)
+{
+	struct fs_prio *prio;
+	bool next_ns;
+
+	next_ns = flow_act->action & MLX5_FLOW_CONTEXT_ACTION_FWD_NEXT_NS;
+	fs_get_obj(prio, next_ns ? ft->ns->node.parent : ft->node.parent);
+
+	return find_next_chained_ft(&prio->node);
+}
+
+static int connect_fts_in_prio(struct mlx5_core_dev *dev,
+			       struct fs_prio *prio,
+			       struct mlx5_flow_table *ft)
+{
+	struct mlx5_flow_root_namespace *root = find_root(&prio->node);
+	struct mlx5_flow_table *iter;
+	int err;
+
+	fs_for_each_ft(iter, prio) {
+		err = root->cmds->modify_flow_table(root, iter, ft);
+		if (err) {
+			mlx5_core_err(dev,
+				      "Failed to modify flow table id %d, type %d, err %d\n",
+				      iter->id, iter->type, err);
+			/* The driver is out of sync with the FW */
+			return err;
+		}
+	}
+	return 0;
+}
+
+static struct mlx5_flow_table *find_closet_ft_prio_chains(struct fs_node *node,
+							  struct fs_node *parent,
+							  struct fs_node **child,
+							  bool reverse)
+{
+	struct mlx5_flow_table *ft;
+
+	ft = find_closest_ft(node, reverse, false);
+
+	if (ft && parent == find_prio_chains_parent(&ft->node, child))
+		return ft;
+
+	return NULL;
+}
+
+/* Connect flow tables from previous priority of prio to ft */
+static int connect_prev_fts(struct mlx5_core_dev *dev,
+			    struct mlx5_flow_table *ft,
+			    struct fs_prio *prio)
+{
+	struct fs_node *prio_parent, *parent = NULL, *child, *node;
+	struct mlx5_flow_table *prev_ft;
+	int err = 0;
+
+	prio_parent = find_prio_chains_parent(&prio->node, &child);
+
+	/* return directly if not under the first sub ns of prio_chains prio */
+	if (prio_parent && !list_is_first(&child->list, &prio_parent->children))
+		return 0;
+
+	prev_ft = find_prev_chained_ft(&prio->node);
+	while (prev_ft) {
+		struct fs_prio *prev_prio;
+
+		fs_get_obj(prev_prio, prev_ft->node.parent);
+		err = connect_fts_in_prio(dev, prev_prio, ft);
+		if (err)
+			break;
+
+		if (!parent) {
+			parent = find_prio_chains_parent(&prev_prio->node, &child);
+			if (!parent)
+				break;
+		}
+
+		node = child;
+		prev_ft = find_closet_ft_prio_chains(node, parent, &child, true);
+	}
+	return err;
+}
+
+static int update_root_ft_create(struct mlx5_flow_table *ft, struct fs_prio
+				 *prio)
+{
+	struct mlx5_flow_root_namespace *root = find_root(&prio->node);
+	struct mlx5_ft_underlay_qp *uqp;
+	int min_level = INT_MAX;
+	int err = 0;
+	u32 qpn;
+
+	if (root->root_ft)
+		min_level = root->root_ft->level;
+
+	if (ft->level >= min_level)
+		return 0;
+
+	if (list_empty(&root->underlay_qpns)) {
+		/* Don't set any QPN (zero) in case QPN list is empty */
+		qpn = 0;
+		err = root->cmds->update_root_ft(root, ft, qpn, false);
+	} else {
+		list_for_each_entry(uqp, &root->underlay_qpns, list) {
+			qpn = uqp->qpn;
+			err = root->cmds->update_root_ft(root, ft,
+							 qpn, false);
+			if (err)
+				break;
+		}
+	}
+
+	if (err)
+		mlx5_core_warn(root->dev,
+			       "Update root flow table of id(%u) qpn(%d) failed\n",
+			       ft->id, qpn);
+	else
+		root->root_ft = ft;
+
+	return err;
+}
+
+static int _mlx5_modify_rule_destination(struct mlx5_flow_rule *rule,
+					 struct mlx5_flow_destination *dest)
+{
+	struct mlx5_flow_root_namespace *root;
+	struct mlx5_flow_table *ft;
+	struct mlx5_flow_group *fg;
+	struct fs_fte *fte;
+	int modify_mask = BIT(MLX5_SET_FTE_MODIFY_ENABLE_MASK_DESTINATION_LIST);
+	int err = 0;
+
+	fs_get_obj(fte, rule->node.parent);
+	if (!(fte->action.action & MLX5_FLOW_CONTEXT_ACTION_FWD_DEST))
+		return -EINVAL;
+	down_write_ref_node(&fte->node, false);
+	fs_get_obj(fg, fte->node.parent);
+	fs_get_obj(ft, fg->node.parent);
+
+	memcpy(&rule->dest_attr, dest, sizeof(*dest));
+	root = find_root(&ft->node);
+	err = root->cmds->update_fte(root, ft, fg,
+				     modify_mask, fte);
+	up_write_ref_node(&fte->node, false);
+
+	return err;
+}
+
+int mlx5_modify_rule_destination(struct mlx5_flow_handle *handle,
+				 struct mlx5_flow_destination *new_dest,
+				 struct mlx5_flow_destination *old_dest)
+{
+	int i;
+
+	if (!old_dest) {
+		if (handle->num_rules != 1)
+			return -EINVAL;
+		return _mlx5_modify_rule_destination(handle->rule[0],
+						     new_dest);
+	}
+
+	for (i = 0; i < handle->num_rules; i++) {
+		if (mlx5_flow_dests_cmp(new_dest, &handle->rule[i]->dest_attr))
+			return _mlx5_modify_rule_destination(handle->rule[i],
+							     new_dest);
+	}
+
+	return -EINVAL;
+}
+
+/* Modify/set FWD rules that point on old_next_ft to point on new_next_ft  */
+static int connect_fwd_rules(struct mlx5_core_dev *dev,
+			     struct mlx5_flow_table *new_next_ft,
+			     struct mlx5_flow_table *old_next_ft)
+{
+	struct mlx5_flow_destination dest = {};
+	struct mlx5_flow_rule *iter;
+	int err = 0;
+
+	/* new_next_ft and old_next_ft could be NULL only
+	 * when we create/destroy the anchor flow table.
+	 */
+	if (!new_next_ft || !old_next_ft)
+		return 0;
+
+	dest.type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
+	dest.ft = new_next_ft;
+
+	mutex_lock(&old_next_ft->lock);
+	list_splice_init(&old_next_ft->fwd_rules, &new_next_ft->fwd_rules);
+	mutex_unlock(&old_next_ft->lock);
+	list_for_each_entry(iter, &new_next_ft->fwd_rules, next_ft) {
+		if ((iter->sw_action & MLX5_FLOW_CONTEXT_ACTION_FWD_NEXT_NS) &&
+		    iter->ft->ns == new_next_ft->ns)
+			continue;
+
+		err = _mlx5_modify_rule_destination(iter, &dest);
+		if (err)
+			pr_err("mlx5_core: failed to modify rule to point on flow table %d\n",
+			       new_next_ft->id);
+	}
+	return 0;
+}
+
+static int connect_flow_table(struct mlx5_core_dev *dev, struct mlx5_flow_table *ft,
+			      struct fs_prio *prio)
+{
+	struct mlx5_flow_table *next_ft, *first_ft;
+	int err = 0;
+
+	/* Connect_prev_fts and update_root_ft_create are mutually exclusive */
+
+	first_ft = list_first_entry_or_null(&prio->node.children,
+					    struct mlx5_flow_table, node.list);
+	if (!first_ft || first_ft->level > ft->level) {
+		err = connect_prev_fts(dev, ft, prio);
+		if (err)
+			return err;
+
+		next_ft = first_ft ? first_ft : find_next_chained_ft(&prio->node);
+		err = connect_fwd_rules(dev, ft, next_ft);
+		if (err)
+			return err;
+	}
+
+	if (MLX5_CAP_FLOWTABLE(dev,
+			       flow_table_properties_nic_receive.modify_root))
+		err = update_root_ft_create(ft, prio);
+	return err;
+}
+
+static void list_add_flow_table(struct mlx5_flow_table *ft,
+				struct fs_prio *prio)
+{
+	struct list_head *prev = &prio->node.children;
+	struct mlx5_flow_table *iter;
+
+	fs_for_each_ft(iter, prio) {
+		if (iter->level > ft->level)
+			break;
+		prev = &iter->node.list;
+	}
+	list_add(&ft->node.list, prev);
+}
+
+static struct mlx5_flow_table *__mlx5_create_flow_table(struct mlx5_flow_namespace *ns,
+							struct mlx5_flow_table_attr *ft_attr,
+							enum fs_flow_table_op_mod op_mod,
+							u16 vport)
+{
+	struct mlx5_flow_root_namespace *root = find_root(&ns->node);
+	bool unmanaged = ft_attr->flags & MLX5_FLOW_TABLE_UNMANAGED;
+	struct mlx5_flow_table *next_ft;
+	struct fs_prio *fs_prio = NULL;
+	struct mlx5_flow_table *ft;
+	int err;
+
+	if (!root) {
+		pr_err("mlx5: flow steering failed to find root of namespace\n");
+		return ERR_PTR(-ENODEV);
+	}
+
+	mutex_lock(&root->chain_lock);
+	fs_prio = find_prio(ns, ft_attr->prio);
+	if (!fs_prio) {
+		err = -EINVAL;
+		goto unlock_root;
+	}
+	if (!unmanaged) {
+		/* The level is related to the
+		 * priority level range.
+		 */
+		if (ft_attr->level >= fs_prio->num_levels) {
+			err = -ENOSPC;
+			goto unlock_root;
+		}
+
+		ft_attr->level += fs_prio->start_level;
+	}
+
+	/* The level is related to the
+	 * priority level range.
+	 */
+	ft = alloc_flow_table(ft_attr->level,
+			      vport,
+			      root->table_type,
+			      op_mod, ft_attr->flags);
+	if (IS_ERR(ft)) {
+		err = PTR_ERR(ft);
+		goto unlock_root;
+	}
+
+	tree_init_node(&ft->node, del_hw_flow_table, del_sw_flow_table);
+	next_ft = unmanaged ? ft_attr->next_ft :
+			      find_next_chained_ft(&fs_prio->node);
+	ft->def_miss_action = ns->def_miss_action;
+	ft->ns = ns;
+	err = root->cmds->create_flow_table(root, ft, ft_attr, next_ft);
+	if (err)
+		goto free_ft;
+
+	if (!unmanaged) {
+		err = connect_flow_table(root->dev, ft, fs_prio);
+		if (err)
+			goto destroy_ft;
+	}
+
+	ft->node.active = true;
+	down_write_ref_node(&fs_prio->node, false);
+	if (!unmanaged) {
+		tree_add_node(&ft->node, &fs_prio->node);
+		list_add_flow_table(ft, fs_prio);
+	} else {
+		ft->node.root = fs_prio->node.root;
+	}
+	fs_prio->num_ft++;
+	up_write_ref_node(&fs_prio->node, false);
+	mutex_unlock(&root->chain_lock);
+	trace_mlx5_fs_add_ft(ft);
+	return ft;
+destroy_ft:
+	root->cmds->destroy_flow_table(root, ft);
+free_ft:
+	rhltable_destroy(&ft->fgs_hash);
+	kfree(ft);
+unlock_root:
+	mutex_unlock(&root->chain_lock);
+	return ERR_PTR(err);
+}
+
+struct mlx5_flow_table *mlx5_create_flow_table(struct mlx5_flow_namespace *ns,
+					       struct mlx5_flow_table_attr *ft_attr)
+{
+	return __mlx5_create_flow_table(ns, ft_attr, FS_FT_OP_MOD_NORMAL, 0);
+}
+EXPORT_SYMBOL(mlx5_create_flow_table);
+
+u32 mlx5_flow_table_id(struct mlx5_flow_table *ft)
+{
+	return ft->id;
+}
+EXPORT_SYMBOL(mlx5_flow_table_id);
+
+struct mlx5_flow_table *
+mlx5_create_vport_flow_table(struct mlx5_flow_namespace *ns,
+			     struct mlx5_flow_table_attr *ft_attr, u16 vport)
+{
+	return __mlx5_create_flow_table(ns, ft_attr, FS_FT_OP_MOD_NORMAL, vport);
+}
+
+struct mlx5_flow_table*
+mlx5_create_lag_demux_flow_table(struct mlx5_flow_namespace *ns,
+				 int prio, u32 level)
+{
+	struct mlx5_flow_table_attr ft_attr = {};
+
+	ft_attr.level = level;
+	ft_attr.prio  = prio;
+	ft_attr.max_fte = 1;
+
+	return __mlx5_create_flow_table(ns, &ft_attr, FS_FT_OP_MOD_LAG_DEMUX, 0);
+}
+EXPORT_SYMBOL(mlx5_create_lag_demux_flow_table);
+
+#define MAX_FLOW_GROUP_SIZE BIT(24)
+struct mlx5_flow_table*
+mlx5_create_auto_grouped_flow_table(struct mlx5_flow_namespace *ns,
+				    struct mlx5_flow_table_attr *ft_attr)
+{
+	int num_reserved_entries = ft_attr->autogroup.num_reserved_entries;
+	int max_num_groups = ft_attr->autogroup.max_num_groups;
+	struct mlx5_flow_table *ft;
+	int autogroups_max_fte;
+
+	ft = mlx5_create_flow_table(ns, ft_attr);
+	if (IS_ERR(ft))
+		return ft;
+
+	autogroups_max_fte = ft->max_fte - num_reserved_entries;
+	if (max_num_groups > autogroups_max_fte)
+		goto err_validate;
+	if (num_reserved_entries > ft->max_fte)
+		goto err_validate;
+
+	/* Align the number of groups according to the largest group size */
+	if (autogroups_max_fte / (max_num_groups + 1) > MAX_FLOW_GROUP_SIZE)
+		max_num_groups = (autogroups_max_fte / MAX_FLOW_GROUP_SIZE) - 1;
+
+	ft->autogroup.active = true;
+	ft->autogroup.required_groups = max_num_groups;
+	ft->autogroup.max_fte = autogroups_max_fte;
+	/* We save place for flow groups in addition to max types */
+	ft->autogroup.group_size = autogroups_max_fte / (max_num_groups + 1);
+
+	return ft;
+
+err_validate:
+	mlx5_destroy_flow_table(ft);
+	return ERR_PTR(-ENOSPC);
+}
+EXPORT_SYMBOL(mlx5_create_auto_grouped_flow_table);
+
+struct mlx5_flow_group *mlx5_create_flow_group(struct mlx5_flow_table *ft,
+					       u32 *fg_in)
+{
+	struct mlx5_flow_root_namespace *root = find_root(&ft->node);
+	void *match_criteria = MLX5_ADDR_OF(create_flow_group_in,
+					    fg_in, match_criteria);
+	u8 match_criteria_enable = MLX5_GET(create_flow_group_in,
+					    fg_in,
+					    match_criteria_enable);
+	int start_index = MLX5_GET(create_flow_group_in, fg_in,
+				   start_flow_index);
+	int end_index = MLX5_GET(create_flow_group_in, fg_in,
+				 end_flow_index);
+	struct mlx5_flow_group *fg;
+	int err;
+
+	if (ft->autogroup.active && start_index < ft->autogroup.max_fte)
+		return ERR_PTR(-EPERM);
+
+	down_write_ref_node(&ft->node, false);
+	fg = alloc_insert_flow_group(ft, match_criteria_enable, match_criteria,
+				     start_index, end_index,
+				     ft->node.children.prev);
+	up_write_ref_node(&ft->node, false);
+	if (IS_ERR(fg))
+		return fg;
+
+	err = root->cmds->create_flow_group(root, ft, fg_in, fg);
+	if (err) {
+		tree_put_node(&fg->node, false);
+		return ERR_PTR(err);
+	}
+	trace_mlx5_fs_add_fg(fg);
+	fg->node.active = true;
+
+	return fg;
+}
+EXPORT_SYMBOL(mlx5_create_flow_group);
+
+static struct mlx5_flow_rule *alloc_rule(struct mlx5_flow_destination *dest)
+{
+	struct mlx5_flow_rule *rule;
+
+	rule = kzalloc(sizeof(*rule), GFP_KERNEL);
+	if (!rule)
+		return NULL;
+
+	INIT_LIST_HEAD(&rule->next_ft);
+	rule->node.type = FS_TYPE_FLOW_DEST;
+	if (dest)
+		memcpy(&rule->dest_attr, dest, sizeof(*dest));
+	else
+		rule->dest_attr.type = MLX5_FLOW_DESTINATION_TYPE_NONE;
+
+	return rule;
+}
+
+static struct mlx5_flow_handle *alloc_handle(int num_rules)
+{
+	struct mlx5_flow_handle *handle;
+
+	handle = kzalloc(struct_size(handle, rule, num_rules), GFP_KERNEL);
+	if (!handle)
+		return NULL;
+
+	handle->num_rules = num_rules;
+
+	return handle;
+}
+
+static void destroy_flow_handle(struct fs_fte *fte,
+				struct mlx5_flow_handle *handle,
+				struct mlx5_flow_destination *dest,
+				int i)
+{
+	for (; --i >= 0;) {
+		if (refcount_dec_and_test(&handle->rule[i]->node.refcount)) {
+			fte->dests_size--;
+			list_del(&handle->rule[i]->node.list);
+			kfree(handle->rule[i]);
+		}
+	}
+	kfree(handle);
+}
+
+static struct mlx5_flow_handle *
+create_flow_handle(struct fs_fte *fte,
+		   struct mlx5_flow_destination *dest,
+		   int dest_num,
+		   int *modify_mask,
+		   bool *new_rule)
+{
+	struct mlx5_flow_handle *handle;
+	struct mlx5_flow_rule *rule = NULL;
+	static int count = BIT(MLX5_SET_FTE_MODIFY_ENABLE_MASK_FLOW_COUNTERS);
+	static int dst = BIT(MLX5_SET_FTE_MODIFY_ENABLE_MASK_DESTINATION_LIST);
+	int type;
+	int i = 0;
+
+	handle = alloc_handle((dest_num) ? dest_num : 1);
+	if (!handle)
+		return ERR_PTR(-ENOMEM);
+
+	do {
+		if (dest) {
+			rule = find_flow_rule(fte, dest + i);
+			if (rule) {
+				refcount_inc(&rule->node.refcount);
+				goto rule_found;
+			}
+		}
+
+		*new_rule = true;
+		rule = alloc_rule(dest + i);
+		if (!rule)
+			goto free_rules;
+
+		/* Add dest to dests list- we need flow tables to be in the
+		 * end of the list for forward to next prio rules.
+		 */
+		tree_init_node(&rule->node, NULL, del_sw_hw_rule);
+		if (dest &&
+		    dest[i].type != MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE)
+			list_add(&rule->node.list, &fte->node.children);
+		else
+			list_add_tail(&rule->node.list, &fte->node.children);
+		if (dest) {
+			fte->dests_size++;
+
+			if (is_fwd_dest_type(dest[i].type))
+				fte->fwd_dests++;
+
+			type = dest[i].type ==
+				MLX5_FLOW_DESTINATION_TYPE_COUNTER;
+			*modify_mask |= type ? count : dst;
+		}
+rule_found:
+		handle->rule[i] = rule;
+	} while (++i < dest_num);
+
+	return handle;
+
+free_rules:
+	destroy_flow_handle(fte, handle, dest, i);
+	return ERR_PTR(-ENOMEM);
+}
+
+/* fte should not be deleted while calling this function */
+static struct mlx5_flow_handle *
+add_rule_fte(struct fs_fte *fte,
+	     struct mlx5_flow_group *fg,
+	     struct mlx5_flow_destination *dest,
+	     int dest_num,
+	     bool update_action)
+{
+	struct mlx5_flow_root_namespace *root;
+	struct mlx5_flow_handle *handle;
+	struct mlx5_flow_table *ft;
+	int modify_mask = 0;
+	int err;
+	bool new_rule = false;
+
+	handle = create_flow_handle(fte, dest, dest_num, &modify_mask,
+				    &new_rule);
+	if (IS_ERR(handle) || !new_rule)
+		goto out;
+
+	if (update_action)
+		modify_mask |= BIT(MLX5_SET_FTE_MODIFY_ENABLE_MASK_ACTION);
+
+	fs_get_obj(ft, fg->node.parent);
+	root = find_root(&fg->node);
+	if (!(fte->status & FS_FTE_STATUS_EXISTING))
+		err = root->cmds->create_fte(root, ft, fg, fte);
+	else
+		err = root->cmds->update_fte(root, ft, fg, modify_mask, fte);
+	if (err)
+		goto free_handle;
+
+	fte->node.active = true;
+	fte->status |= FS_FTE_STATUS_EXISTING;
+	atomic_inc(&fg->node.version);
+
+out:
+	return handle;
+
+free_handle:
+	destroy_flow_handle(fte, handle, dest, handle->num_rules);
+	return ERR_PTR(err);
+}
+
+static struct mlx5_flow_group *alloc_auto_flow_group(struct mlx5_flow_table  *ft,
+						     const struct mlx5_flow_spec *spec)
+{
+	struct list_head *prev = &ft->node.children;
+	u32 max_fte = ft->autogroup.max_fte;
+	unsigned int candidate_index = 0;
+	unsigned int group_size = 0;
+	struct mlx5_flow_group *fg;
+
+	if (!ft->autogroup.active)
+		return ERR_PTR(-ENOENT);
+
+	if (ft->autogroup.num_groups < ft->autogroup.required_groups)
+		group_size = ft->autogroup.group_size;
+
+	/*  max_fte == ft->autogroup.max_types */
+	if (group_size == 0)
+		group_size = 1;
+
+	/* sorted by start_index */
+	fs_for_each_fg(fg, ft) {
+		if (candidate_index + group_size > fg->start_index)
+			candidate_index = fg->start_index + fg->max_ftes;
+		else
+			break;
+		prev = &fg->node.list;
+	}
+
+	if (candidate_index + group_size > max_fte)
+		return ERR_PTR(-ENOSPC);
+
+	fg = alloc_insert_flow_group(ft,
+				     spec->match_criteria_enable,
+				     spec->match_criteria,
+				     candidate_index,
+				     candidate_index + group_size - 1,
+				     prev);
+	if (IS_ERR(fg))
+		goto out;
+
+	if (group_size == ft->autogroup.group_size)
+		ft->autogroup.num_groups++;
+
+out:
+	return fg;
+}
+
+static int create_auto_flow_group(struct mlx5_flow_table *ft,
+				  struct mlx5_flow_group *fg)
+{
+	struct mlx5_flow_root_namespace *root = find_root(&ft->node);
+	int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
+	void *match_criteria_addr;
+	u8 src_esw_owner_mask_on;
+	void *misc;
+	int err;
+	u32 *in;
+
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in)
+		return -ENOMEM;
+
+	MLX5_SET(create_flow_group_in, in, match_criteria_enable,
+		 fg->mask.match_criteria_enable);
+	MLX5_SET(create_flow_group_in, in, start_flow_index, fg->start_index);
+	MLX5_SET(create_flow_group_in, in, end_flow_index,   fg->start_index +
+		 fg->max_ftes - 1);
+
+	misc = MLX5_ADDR_OF(fte_match_param, fg->mask.match_criteria,
+			    misc_parameters);
+	src_esw_owner_mask_on = !!MLX5_GET(fte_match_set_misc, misc,
+					 source_eswitch_owner_vhca_id);
+	MLX5_SET(create_flow_group_in, in,
+		 source_eswitch_owner_vhca_id_valid, src_esw_owner_mask_on);
+
+	match_criteria_addr = MLX5_ADDR_OF(create_flow_group_in,
+					   in, match_criteria);
+	memcpy(match_criteria_addr, fg->mask.match_criteria,
+	       sizeof(fg->mask.match_criteria));
+
+	err = root->cmds->create_flow_group(root, ft, in, fg);
+	if (!err) {
+		fg->node.active = true;
+		trace_mlx5_fs_add_fg(fg);
+	}
+
+	kvfree(in);
+	return err;
+}
+
+static bool mlx5_flow_dests_cmp(struct mlx5_flow_destination *d1,
+				struct mlx5_flow_destination *d2)
+{
+	if (d1->type == d2->type) {
+		if (((d1->type == MLX5_FLOW_DESTINATION_TYPE_VPORT ||
+		      d1->type == MLX5_FLOW_DESTINATION_TYPE_UPLINK) &&
+		     d1->vport.num == d2->vport.num &&
+		     d1->vport.flags == d2->vport.flags &&
+		     ((d1->vport.flags & MLX5_FLOW_DEST_VPORT_VHCA_ID) ?
+		      (d1->vport.vhca_id == d2->vport.vhca_id) : true) &&
+		     ((d1->vport.flags & MLX5_FLOW_DEST_VPORT_REFORMAT_ID) ?
+		      (d1->vport.pkt_reformat->id ==
+		       d2->vport.pkt_reformat->id) : true)) ||
+		    (d1->type == MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE &&
+		     d1->ft == d2->ft) ||
+		    (d1->type == MLX5_FLOW_DESTINATION_TYPE_TIR &&
+		     d1->tir_num == d2->tir_num) ||
+		    (d1->type == MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE_NUM &&
+		     d1->ft_num == d2->ft_num) ||
+		    (d1->type == MLX5_FLOW_DESTINATION_TYPE_FLOW_SAMPLER &&
+		     d1->sampler_id == d2->sampler_id))
+			return true;
+	}
+
+	return false;
+}
+
+static struct mlx5_flow_rule *find_flow_rule(struct fs_fte *fte,
+					     struct mlx5_flow_destination *dest)
+{
+	struct mlx5_flow_rule *rule;
+
+	list_for_each_entry(rule, &fte->node.children, node.list) {
+		if (mlx5_flow_dests_cmp(&rule->dest_attr, dest))
+			return rule;
+	}
+	return NULL;
+}
+
+static bool check_conflicting_actions_vlan(const struct mlx5_fs_vlan *vlan0,
+					   const struct mlx5_fs_vlan *vlan1)
+{
+	return vlan0->ethtype != vlan1->ethtype ||
+	       vlan0->vid != vlan1->vid ||
+	       vlan0->prio != vlan1->prio;
+}
+
+static bool check_conflicting_actions(const struct mlx5_flow_act *act1,
+				      const struct mlx5_flow_act *act2)
+{
+	u32 action1 = act1->action;
+	u32 action2 = act2->action;
+	u32 xored_actions;
+
+	xored_actions = action1 ^ action2;
+
+	/* if one rule only wants to count, it's ok */
+	if (action1 == MLX5_FLOW_CONTEXT_ACTION_COUNT ||
+	    action2 == MLX5_FLOW_CONTEXT_ACTION_COUNT)
+		return false;
+
+	if (xored_actions & (MLX5_FLOW_CONTEXT_ACTION_DROP  |
+			     MLX5_FLOW_CONTEXT_ACTION_PACKET_REFORMAT |
+			     MLX5_FLOW_CONTEXT_ACTION_DECAP |
+			     MLX5_FLOW_CONTEXT_ACTION_MOD_HDR  |
+			     MLX5_FLOW_CONTEXT_ACTION_VLAN_POP |
+			     MLX5_FLOW_CONTEXT_ACTION_VLAN_PUSH |
+			     MLX5_FLOW_CONTEXT_ACTION_VLAN_POP_2 |
+			     MLX5_FLOW_CONTEXT_ACTION_VLAN_PUSH_2))
+		return true;
+
+	if (action1 & MLX5_FLOW_CONTEXT_ACTION_PACKET_REFORMAT &&
+	    act1->pkt_reformat != act2->pkt_reformat)
+		return true;
+
+	if (action1 & MLX5_FLOW_CONTEXT_ACTION_MOD_HDR &&
+	    act1->modify_hdr != act2->modify_hdr)
+		return true;
+
+	if (action1 & MLX5_FLOW_CONTEXT_ACTION_VLAN_PUSH &&
+	    check_conflicting_actions_vlan(&act1->vlan[0], &act2->vlan[0]))
+		return true;
+
+	if (action1 & MLX5_FLOW_CONTEXT_ACTION_VLAN_PUSH_2 &&
+	    check_conflicting_actions_vlan(&act1->vlan[1], &act2->vlan[1]))
+		return true;
+
+	return false;
+}
+
+static int check_conflicting_ftes(struct fs_fte *fte,
+				  const struct mlx5_flow_context *flow_context,
+				  const struct mlx5_flow_act *flow_act)
+{
+	if (check_conflicting_actions(flow_act, &fte->action)) {
+		mlx5_core_warn(get_dev(&fte->node),
+			       "Found two FTEs with conflicting actions\n");
+		return -EEXIST;
+	}
+
+	if ((flow_context->flags & FLOW_CONTEXT_HAS_TAG) &&
+	    fte->flow_context.flow_tag != flow_context->flow_tag) {
+		mlx5_core_warn(get_dev(&fte->node),
+			       "FTE flow tag %u already exists with different flow tag %u\n",
+			       fte->flow_context.flow_tag,
+			       flow_context->flow_tag);
+		return -EEXIST;
+	}
+
+	return 0;
+}
+
+static struct mlx5_flow_handle *add_rule_fg(struct mlx5_flow_group *fg,
+					    const struct mlx5_flow_spec *spec,
+					    struct mlx5_flow_act *flow_act,
+					    struct mlx5_flow_destination *dest,
+					    int dest_num,
+					    struct fs_fte *fte)
+{
+	struct mlx5_flow_handle *handle;
+	int old_action;
+	int i;
+	int ret;
+
+	ret = check_conflicting_ftes(fte, &spec->flow_context, flow_act);
+	if (ret)
+		return ERR_PTR(ret);
+
+	old_action = fte->action.action;
+	fte->action.action |= flow_act->action;
+	handle = add_rule_fte(fte, fg, dest, dest_num,
+			      old_action != flow_act->action);
+	if (IS_ERR(handle)) {
+		fte->action.action = old_action;
+		return handle;
+	}
+	trace_mlx5_fs_set_fte(fte, false);
+
+	for (i = 0; i < handle->num_rules; i++) {
+		if (refcount_read(&handle->rule[i]->node.refcount) == 1) {
+			tree_add_node(&handle->rule[i]->node, &fte->node);
+			trace_mlx5_fs_add_rule(handle->rule[i]);
+		}
+	}
+	return handle;
+}
+
+static bool counter_is_valid(u32 action)
+{
+	return (action & (MLX5_FLOW_CONTEXT_ACTION_DROP |
+			  MLX5_FLOW_CONTEXT_ACTION_ALLOW |
+			  MLX5_FLOW_CONTEXT_ACTION_FWD_DEST));
+}
+
+static bool dest_is_valid(struct mlx5_flow_destination *dest,
+			  struct mlx5_flow_act *flow_act,
+			  struct mlx5_flow_table *ft)
+{
+	bool ignore_level = flow_act->flags & FLOW_ACT_IGNORE_FLOW_LEVEL;
+	u32 action = flow_act->action;
+
+	if (dest && (dest->type == MLX5_FLOW_DESTINATION_TYPE_COUNTER))
+		return counter_is_valid(action);
+
+	if (!(action & MLX5_FLOW_CONTEXT_ACTION_FWD_DEST))
+		return true;
+
+	if (ignore_level) {
+		if (ft->type != FS_FT_FDB &&
+		    ft->type != FS_FT_NIC_RX)
+			return false;
+
+		if (dest->type == MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE &&
+		    ft->type != dest->ft->type)
+			return false;
+	}
+
+	if (!dest || ((dest->type ==
+	    MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE) &&
+	    (dest->ft->level <= ft->level && !ignore_level)))
+		return false;
+	return true;
+}
+
+struct match_list {
+	struct list_head	list;
+	struct mlx5_flow_group *g;
+};
+
+static void free_match_list(struct match_list *head, bool ft_locked)
+{
+	struct match_list *iter, *match_tmp;
+
+	list_for_each_entry_safe(iter, match_tmp, &head->list,
+				 list) {
+		tree_put_node(&iter->g->node, ft_locked);
+		list_del(&iter->list);
+		kfree(iter);
+	}
+}
+
+static int build_match_list(struct match_list *match_head,
+			    struct mlx5_flow_table *ft,
+			    const struct mlx5_flow_spec *spec,
+			    struct mlx5_flow_group *fg,
+			    bool ft_locked)
+{
+	struct rhlist_head *tmp, *list;
+	struct mlx5_flow_group *g;
+	int err = 0;
+
+	rcu_read_lock();
+	INIT_LIST_HEAD(&match_head->list);
+	/* Collect all fgs which has a matching match_criteria */
+	list = rhltable_lookup(&ft->fgs_hash, spec, rhash_fg);
+	/* RCU is atomic, we can't execute FW commands here */
+	rhl_for_each_entry_rcu(g, tmp, list, hash) {
+		struct match_list *curr_match;
+
+		if (fg && fg != g)
+			continue;
+
+		if (unlikely(!tree_get_node(&g->node)))
+			continue;
+
+		curr_match = kmalloc(sizeof(*curr_match), GFP_ATOMIC);
+		if (!curr_match) {
+			rcu_read_unlock();
+			free_match_list(match_head, ft_locked);
+			return -ENOMEM;
+		}
+		curr_match->g = g;
+		list_add_tail(&curr_match->list, &match_head->list);
+	}
+	rcu_read_unlock();
+	return err;
+}
+
+static u64 matched_fgs_get_version(struct list_head *match_head)
+{
+	struct match_list *iter;
+	u64 version = 0;
+
+	list_for_each_entry(iter, match_head, list)
+		version += (u64)atomic_read(&iter->g->node.version);
+	return version;
+}
+
+static struct fs_fte *
+lookup_fte_locked(struct mlx5_flow_group *g,
+		  const u32 *match_value,
+		  bool take_write)
+{
+	struct fs_fte *fte_tmp;
+
+	if (take_write)
+		nested_down_write_ref_node(&g->node, FS_LOCK_PARENT);
+	else
+		nested_down_read_ref_node(&g->node, FS_LOCK_PARENT);
+	fte_tmp = rhashtable_lookup_fast(&g->ftes_hash, match_value,
+					 rhash_fte);
+	if (!fte_tmp || !tree_get_node(&fte_tmp->node)) {
+		fte_tmp = NULL;
+		goto out;
+	}
+	if (!fte_tmp->node.active) {
+		tree_put_node(&fte_tmp->node, false);
+		fte_tmp = NULL;
+		goto out;
+	}
+
+	nested_down_write_ref_node(&fte_tmp->node, FS_LOCK_CHILD);
+out:
+	if (take_write)
+		up_write_ref_node(&g->node, false);
+	else
+		up_read_ref_node(&g->node);
+	return fte_tmp;
+}
+
+static struct mlx5_flow_handle *
+try_add_to_existing_fg(struct mlx5_flow_table *ft,
+		       struct list_head *match_head,
+		       const struct mlx5_flow_spec *spec,
+		       struct mlx5_flow_act *flow_act,
+		       struct mlx5_flow_destination *dest,
+		       int dest_num,
+		       int ft_version)
+{
+	struct mlx5_flow_steering *steering = get_steering(&ft->node);
+	struct mlx5_flow_group *g;
+	struct mlx5_flow_handle *rule;
+	struct match_list *iter;
+	bool take_write = false;
+	struct fs_fte *fte;
+	u64  version = 0;
+	int err;
+
+	fte = alloc_fte(ft, spec, flow_act);
+	if (IS_ERR(fte))
+		return  ERR_PTR(-ENOMEM);
+
+search_again_locked:
+	if (flow_act->flags & FLOW_ACT_NO_APPEND)
+		goto skip_search;
+	version = matched_fgs_get_version(match_head);
+	/* Try to find an fte with identical match value and attempt update its
+	 * action.
+	 */
+	list_for_each_entry(iter, match_head, list) {
+		struct fs_fte *fte_tmp;
+
+		g = iter->g;
+		fte_tmp = lookup_fte_locked(g, spec->match_value, take_write);
+		if (!fte_tmp)
+			continue;
+		rule = add_rule_fg(g, spec, flow_act, dest, dest_num, fte_tmp);
+		/* No error check needed here, because insert_fte() is not called */
+		up_write_ref_node(&fte_tmp->node, false);
+		tree_put_node(&fte_tmp->node, false);
+		kmem_cache_free(steering->ftes_cache, fte);
+		return rule;
+	}
+
+skip_search:
+	/* No group with matching fte found, or we skipped the search.
+	 * Try to add a new fte to any matching fg.
+	 */
+
+	/* Check the ft version, for case that new flow group
+	 * was added while the fgs weren't locked
+	 */
+	if (atomic_read(&ft->node.version) != ft_version) {
+		rule = ERR_PTR(-EAGAIN);
+		goto out;
+	}
+
+	/* Check the fgs version. If version have changed it could be that an
+	 * FTE with the same match value was added while the fgs weren't
+	 * locked.
+	 */
+	if (!(flow_act->flags & FLOW_ACT_NO_APPEND) &&
+	    version != matched_fgs_get_version(match_head)) {
+		take_write = true;
+		goto search_again_locked;
+	}
+
+	list_for_each_entry(iter, match_head, list) {
+		g = iter->g;
+
+		nested_down_write_ref_node(&g->node, FS_LOCK_PARENT);
+
+		if (!g->node.active) {
+			up_write_ref_node(&g->node, false);
+			continue;
+		}
+
+		err = insert_fte(g, fte);
+		if (err) {
+			up_write_ref_node(&g->node, false);
+			if (err == -ENOSPC)
+				continue;
+			kmem_cache_free(steering->ftes_cache, fte);
+			return ERR_PTR(err);
+		}
+
+		nested_down_write_ref_node(&fte->node, FS_LOCK_CHILD);
+		up_write_ref_node(&g->node, false);
+		rule = add_rule_fg(g, spec, flow_act, dest, dest_num, fte);
+		up_write_ref_node(&fte->node, false);
+		if (IS_ERR(rule))
+			tree_put_node(&fte->node, false);
+		return rule;
+	}
+	rule = ERR_PTR(-ENOENT);
+out:
+	kmem_cache_free(steering->ftes_cache, fte);
+	return rule;
+}
+
+static struct mlx5_flow_handle *
+_mlx5_add_flow_rules(struct mlx5_flow_table *ft,
+		     const struct mlx5_flow_spec *spec,
+		     struct mlx5_flow_act *flow_act,
+		     struct mlx5_flow_destination *dest,
+		     int dest_num)
+
+{
+	struct mlx5_flow_steering *steering = get_steering(&ft->node);
+	struct mlx5_flow_handle *rule;
+	struct match_list match_head;
+	struct mlx5_flow_group *g;
+	bool take_write = false;
+	struct fs_fte *fte;
+	int version;
+	int err;
+	int i;
+
+	if (!check_valid_spec(spec))
+		return ERR_PTR(-EINVAL);
+
+	if (flow_act->fg && ft->autogroup.active)
+		return ERR_PTR(-EINVAL);
+
+	for (i = 0; i < dest_num; i++) {
+		if (!dest_is_valid(&dest[i], flow_act, ft))
+			return ERR_PTR(-EINVAL);
+	}
+	nested_down_read_ref_node(&ft->node, FS_LOCK_GRANDPARENT);
+search_again_locked:
+	version = atomic_read(&ft->node.version);
+
+	/* Collect all fgs which has a matching match_criteria */
+	err = build_match_list(&match_head, ft, spec, flow_act->fg, take_write);
+	if (err) {
+		if (take_write)
+			up_write_ref_node(&ft->node, false);
+		else
+			up_read_ref_node(&ft->node);
+		return ERR_PTR(err);
+	}
+
+	if (!take_write)
+		up_read_ref_node(&ft->node);
+
+	rule = try_add_to_existing_fg(ft, &match_head.list, spec, flow_act, dest,
+				      dest_num, version);
+	free_match_list(&match_head, take_write);
+	if (!IS_ERR(rule) ||
+	    (PTR_ERR(rule) != -ENOENT && PTR_ERR(rule) != -EAGAIN)) {
+		if (take_write)
+			up_write_ref_node(&ft->node, false);
+		return rule;
+	}
+
+	if (!take_write) {
+		nested_down_write_ref_node(&ft->node, FS_LOCK_GRANDPARENT);
+		take_write = true;
+	}
+
+	if (PTR_ERR(rule) == -EAGAIN ||
+	    version != atomic_read(&ft->node.version))
+		goto search_again_locked;
+
+	g = alloc_auto_flow_group(ft, spec);
+	if (IS_ERR(g)) {
+		rule = ERR_CAST(g);
+		up_write_ref_node(&ft->node, false);
+		return rule;
+	}
+
+	fte = alloc_fte(ft, spec, flow_act);
+	if (IS_ERR(fte)) {
+		up_write_ref_node(&ft->node, false);
+		err = PTR_ERR(fte);
+		goto err_alloc_fte;
+	}
+
+	nested_down_write_ref_node(&g->node, FS_LOCK_PARENT);
+	up_write_ref_node(&ft->node, false);
+
+	err = create_auto_flow_group(ft, g);
+	if (err)
+		goto err_release_fg;
+
+	err = insert_fte(g, fte);
+	if (err)
+		goto err_release_fg;
+
+	nested_down_write_ref_node(&fte->node, FS_LOCK_CHILD);
+	up_write_ref_node(&g->node, false);
+	rule = add_rule_fg(g, spec, flow_act, dest, dest_num, fte);
+	up_write_ref_node(&fte->node, false);
+	if (IS_ERR(rule))
+		tree_put_node(&fte->node, false);
+	tree_put_node(&g->node, false);
+	return rule;
+
+err_release_fg:
+	up_write_ref_node(&g->node, false);
+	kmem_cache_free(steering->ftes_cache, fte);
+err_alloc_fte:
+	tree_put_node(&g->node, false);
+	return ERR_PTR(err);
+}
+
+static bool fwd_next_prio_supported(struct mlx5_flow_table *ft)
+{
+	return ((ft->type == FS_FT_NIC_RX) &&
+		(MLX5_CAP_FLOWTABLE(get_dev(&ft->node), nic_rx_multi_path_tirs)));
+}
+
+struct mlx5_flow_handle *
+mlx5_add_flow_rules(struct mlx5_flow_table *ft,
+		    const struct mlx5_flow_spec *spec,
+		    struct mlx5_flow_act *flow_act,
+		    struct mlx5_flow_destination *dest,
+		    int num_dest)
+{
+	struct mlx5_flow_root_namespace *root = find_root(&ft->node);
+	static const struct mlx5_flow_spec zero_spec = {};
+	struct mlx5_flow_destination *gen_dest = NULL;
+	struct mlx5_flow_table *next_ft = NULL;
+	struct mlx5_flow_handle *handle = NULL;
+	u32 sw_action = flow_act->action;
+	int i;
+
+	if (!spec)
+		spec = &zero_spec;
+
+	if (!is_fwd_next_action(sw_action))
+		return _mlx5_add_flow_rules(ft, spec, flow_act, dest, num_dest);
+
+	if (!fwd_next_prio_supported(ft))
+		return ERR_PTR(-EOPNOTSUPP);
+
+	mutex_lock(&root->chain_lock);
+	next_ft = find_next_fwd_ft(ft, flow_act);
+	if (!next_ft) {
+		handle = ERR_PTR(-EOPNOTSUPP);
+		goto unlock;
+	}
+
+	gen_dest = kcalloc(num_dest + 1, sizeof(*dest),
+			   GFP_KERNEL);
+	if (!gen_dest) {
+		handle = ERR_PTR(-ENOMEM);
+		goto unlock;
+	}
+	for (i = 0; i < num_dest; i++)
+		gen_dest[i] = dest[i];
+	gen_dest[i].type =
+		MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
+	gen_dest[i].ft = next_ft;
+	dest = gen_dest;
+	num_dest++;
+	flow_act->action &= ~(MLX5_FLOW_CONTEXT_ACTION_FWD_NEXT_PRIO |
+			      MLX5_FLOW_CONTEXT_ACTION_FWD_NEXT_NS);
+	flow_act->action |= MLX5_FLOW_CONTEXT_ACTION_FWD_DEST;
+	handle = _mlx5_add_flow_rules(ft, spec, flow_act, dest, num_dest);
+	if (IS_ERR(handle))
+		goto unlock;
+
+	if (list_empty(&handle->rule[num_dest - 1]->next_ft)) {
+		mutex_lock(&next_ft->lock);
+		list_add(&handle->rule[num_dest - 1]->next_ft,
+			 &next_ft->fwd_rules);
+		mutex_unlock(&next_ft->lock);
+		handle->rule[num_dest - 1]->sw_action = sw_action;
+		handle->rule[num_dest - 1]->ft = ft;
+	}
+unlock:
+	mutex_unlock(&root->chain_lock);
+	kfree(gen_dest);
+	return handle;
+}
+EXPORT_SYMBOL(mlx5_add_flow_rules);
+
+void mlx5_del_flow_rules(struct mlx5_flow_handle *handle)
+{
+	struct fs_fte *fte;
+	int i;
+
+	/* In order to consolidate the HW changes we lock the FTE for other
+	 * changes, and increase its refcount, in order not to perform the
+	 * "del" functions of the FTE. Will handle them here.
+	 * The removal of the rules is done under locked FTE.
+	 * After removing all the handle's rules, if there are remaining
+	 * rules, it means we just need to modify the FTE in FW, and
+	 * unlock/decrease the refcount we increased before.
+	 * Otherwise, it means the FTE should be deleted. First delete the
+	 * FTE in FW. Then, unlock the FTE, and proceed the tree_put_node of
+	 * the FTE, which will handle the last decrease of the refcount, as
+	 * well as required handling of its parent.
+	 */
+	fs_get_obj(fte, handle->rule[0]->node.parent);
+	down_write_ref_node(&fte->node, false);
+	for (i = handle->num_rules - 1; i >= 0; i--)
+		tree_remove_node(&handle->rule[i]->node, true);
+	if (list_empty(&fte->node.children)) {
+		fte->node.del_hw_func(&fte->node);
+		/* Avoid double call to del_hw_fte */
+		fte->node.del_hw_func = NULL;
+		up_write_ref_node(&fte->node, false);
+		tree_put_node(&fte->node, false);
+	} else if (fte->dests_size) {
+		if (fte->modify_mask)
+			modify_fte(fte);
+		up_write_ref_node(&fte->node, false);
+	} else {
+		up_write_ref_node(&fte->node, false);
+	}
+	kfree(handle);
+}
+EXPORT_SYMBOL(mlx5_del_flow_rules);
+
+/* Assuming prio->node.children(flow tables) is sorted by level */
+static struct mlx5_flow_table *find_next_ft(struct mlx5_flow_table *ft)
+{
+	struct fs_node *prio_parent, *child;
+	struct fs_prio *prio;
+
+	fs_get_obj(prio, ft->node.parent);
+
+	if (!list_is_last(&ft->node.list, &prio->node.children))
+		return list_next_entry(ft, node.list);
+
+	prio_parent = find_prio_chains_parent(&prio->node, &child);
+
+	if (prio_parent && list_is_first(&child->list, &prio_parent->children))
+		return find_closest_ft(&prio->node, false, false);
+
+	return find_next_chained_ft(&prio->node);
+}
+
+static int update_root_ft_destroy(struct mlx5_flow_table *ft)
+{
+	struct mlx5_flow_root_namespace *root = find_root(&ft->node);
+	struct mlx5_ft_underlay_qp *uqp;
+	struct mlx5_flow_table *new_root_ft = NULL;
+	int err = 0;
+	u32 qpn;
+
+	if (root->root_ft != ft)
+		return 0;
+
+	new_root_ft = find_next_ft(ft);
+	if (!new_root_ft) {
+		root->root_ft = NULL;
+		return 0;
+	}
+
+	if (list_empty(&root->underlay_qpns)) {
+		/* Don't set any QPN (zero) in case QPN list is empty */
+		qpn = 0;
+		err = root->cmds->update_root_ft(root, new_root_ft,
+						 qpn, false);
+	} else {
+		list_for_each_entry(uqp, &root->underlay_qpns, list) {
+			qpn = uqp->qpn;
+			err = root->cmds->update_root_ft(root,
+							 new_root_ft, qpn,
+							 false);
+			if (err)
+				break;
+		}
+	}
+
+	if (err)
+		mlx5_core_warn(root->dev,
+			       "Update root flow table of id(%u) qpn(%d) failed\n",
+			       ft->id, qpn);
+	else
+		root->root_ft = new_root_ft;
+
+	return 0;
+}
+
+/* Connect flow table from previous priority to
+ * the next flow table.
+ */
+static int disconnect_flow_table(struct mlx5_flow_table *ft)
+{
+	struct mlx5_core_dev *dev = get_dev(&ft->node);
+	struct mlx5_flow_table *next_ft;
+	struct fs_prio *prio;
+	int err = 0;
+
+	err = update_root_ft_destroy(ft);
+	if (err)
+		return err;
+
+	fs_get_obj(prio, ft->node.parent);
+	if  (!(list_first_entry(&prio->node.children,
+				struct mlx5_flow_table,
+				node.list) == ft))
+		return 0;
+
+	next_ft = find_next_ft(ft);
+	err = connect_fwd_rules(dev, next_ft, ft);
+	if (err)
+		return err;
+
+	err = connect_prev_fts(dev, next_ft, prio);
+	if (err)
+		mlx5_core_warn(dev, "Failed to disconnect flow table %d\n",
+			       ft->id);
+	return err;
+}
+
+int mlx5_destroy_flow_table(struct mlx5_flow_table *ft)
+{
+	struct mlx5_flow_root_namespace *root = find_root(&ft->node);
+	int err = 0;
+
+	mutex_lock(&root->chain_lock);
+	if (!(ft->flags & MLX5_FLOW_TABLE_UNMANAGED))
+		err = disconnect_flow_table(ft);
+	if (err) {
+		mutex_unlock(&root->chain_lock);
+		return err;
+	}
+	if (tree_remove_node(&ft->node, false))
+		mlx5_core_warn(get_dev(&ft->node), "Flow table %d wasn't destroyed, refcount > 1\n",
+			       ft->id);
+	mutex_unlock(&root->chain_lock);
+
+	return err;
+}
+EXPORT_SYMBOL(mlx5_destroy_flow_table);
+
+void mlx5_destroy_flow_group(struct mlx5_flow_group *fg)
+{
+	if (tree_remove_node(&fg->node, false))
+		mlx5_core_warn(get_dev(&fg->node), "Flow group %d wasn't destroyed, refcount > 1\n",
+			       fg->id);
+}
+EXPORT_SYMBOL(mlx5_destroy_flow_group);
+
+struct mlx5_flow_namespace *mlx5_get_fdb_sub_ns(struct mlx5_core_dev *dev,
+						int n)
+{
+	struct mlx5_flow_steering *steering = dev->priv.steering;
+
+	if (!steering || !steering->fdb_sub_ns)
+		return NULL;
+
+	return steering->fdb_sub_ns[n];
+}
+EXPORT_SYMBOL(mlx5_get_fdb_sub_ns);
+
+static bool is_nic_rx_ns(enum mlx5_flow_namespace_type type)
+{
+	switch (type) {
+	case MLX5_FLOW_NAMESPACE_BYPASS:
+	case MLX5_FLOW_NAMESPACE_KERNEL_RX_MACSEC:
+	case MLX5_FLOW_NAMESPACE_LAG:
+	case MLX5_FLOW_NAMESPACE_OFFLOADS:
+	case MLX5_FLOW_NAMESPACE_ETHTOOL:
+	case MLX5_FLOW_NAMESPACE_KERNEL:
+	case MLX5_FLOW_NAMESPACE_LEFTOVERS:
+	case MLX5_FLOW_NAMESPACE_ANCHOR:
+		return true;
+	default:
+		return false;
+	}
+}
+
+struct mlx5_flow_namespace *mlx5_get_flow_namespace(struct mlx5_core_dev *dev,
+						    enum mlx5_flow_namespace_type type)
+{
+	struct mlx5_flow_steering *steering = dev->priv.steering;
+	struct mlx5_flow_root_namespace *root_ns;
+	int prio = 0;
+	struct fs_prio *fs_prio;
+	struct mlx5_flow_namespace *ns;
+
+	if (!steering)
+		return NULL;
+
+	switch (type) {
+	case MLX5_FLOW_NAMESPACE_FDB:
+		if (steering->fdb_root_ns)
+			return &steering->fdb_root_ns->ns;
+		return NULL;
+	case MLX5_FLOW_NAMESPACE_PORT_SEL:
+		if (steering->port_sel_root_ns)
+			return &steering->port_sel_root_ns->ns;
+		return NULL;
+	case MLX5_FLOW_NAMESPACE_SNIFFER_RX:
+		if (steering->sniffer_rx_root_ns)
+			return &steering->sniffer_rx_root_ns->ns;
+		return NULL;
+	case MLX5_FLOW_NAMESPACE_SNIFFER_TX:
+		if (steering->sniffer_tx_root_ns)
+			return &steering->sniffer_tx_root_ns->ns;
+		return NULL;
+	case MLX5_FLOW_NAMESPACE_FDB_BYPASS:
+		root_ns = steering->fdb_root_ns;
+		prio =  FDB_BYPASS_PATH;
+		break;
+	case MLX5_FLOW_NAMESPACE_EGRESS:
+	case MLX5_FLOW_NAMESPACE_EGRESS_IPSEC:
+	case MLX5_FLOW_NAMESPACE_EGRESS_MACSEC:
+		root_ns = steering->egress_root_ns;
+		prio = type - MLX5_FLOW_NAMESPACE_EGRESS;
+		break;
+	case MLX5_FLOW_NAMESPACE_RDMA_RX:
+		root_ns = steering->rdma_rx_root_ns;
+		prio = RDMA_RX_BYPASS_PRIO;
+		break;
+	case MLX5_FLOW_NAMESPACE_RDMA_RX_KERNEL:
+		root_ns = steering->rdma_rx_root_ns;
+		prio = RDMA_RX_KERNEL_PRIO;
+		break;
+	case MLX5_FLOW_NAMESPACE_RDMA_TX:
+		root_ns = steering->rdma_tx_root_ns;
+		break;
+	case MLX5_FLOW_NAMESPACE_RDMA_RX_COUNTERS:
+		root_ns = steering->rdma_rx_root_ns;
+		prio = RDMA_RX_COUNTERS_PRIO;
+		break;
+	case MLX5_FLOW_NAMESPACE_RDMA_TX_COUNTERS:
+		root_ns = steering->rdma_tx_root_ns;
+		prio = RDMA_TX_COUNTERS_PRIO;
+		break;
+	default: /* Must be NIC RX */
+		WARN_ON(!is_nic_rx_ns(type));
+		root_ns = steering->root_ns;
+		prio = type;
+		break;
+	}
+
+	if (!root_ns)
+		return NULL;
+
+	fs_prio = find_prio(&root_ns->ns, prio);
+	if (!fs_prio)
+		return NULL;
+
+	ns = list_first_entry(&fs_prio->node.children,
+			      typeof(*ns),
+			      node.list);
+
+	return ns;
+}
+EXPORT_SYMBOL(mlx5_get_flow_namespace);
+
+struct mlx5_flow_namespace *mlx5_get_flow_vport_acl_namespace(struct mlx5_core_dev *dev,
+							      enum mlx5_flow_namespace_type type,
+							      int vport)
+{
+	struct mlx5_flow_steering *steering = dev->priv.steering;
+
+	if (!steering)
+		return NULL;
+
+	switch (type) {
+	case MLX5_FLOW_NAMESPACE_ESW_EGRESS:
+		if (vport >= steering->esw_egress_acl_vports)
+			return NULL;
+		if (steering->esw_egress_root_ns &&
+		    steering->esw_egress_root_ns[vport])
+			return &steering->esw_egress_root_ns[vport]->ns;
+		else
+			return NULL;
+	case MLX5_FLOW_NAMESPACE_ESW_INGRESS:
+		if (vport >= steering->esw_ingress_acl_vports)
+			return NULL;
+		if (steering->esw_ingress_root_ns &&
+		    steering->esw_ingress_root_ns[vport])
+			return &steering->esw_ingress_root_ns[vport]->ns;
+		else
+			return NULL;
+	default:
+		return NULL;
+	}
+}
+
+static struct fs_prio *_fs_create_prio(struct mlx5_flow_namespace *ns,
+				       unsigned int prio,
+				       int num_levels,
+				       enum fs_node_type type)
+{
+	struct fs_prio *fs_prio;
+
+	fs_prio = kzalloc(sizeof(*fs_prio), GFP_KERNEL);
+	if (!fs_prio)
+		return ERR_PTR(-ENOMEM);
+
+	fs_prio->node.type = type;
+	tree_init_node(&fs_prio->node, NULL, del_sw_prio);
+	tree_add_node(&fs_prio->node, &ns->node);
+	fs_prio->num_levels = num_levels;
+	fs_prio->prio = prio;
+	list_add_tail(&fs_prio->node.list, &ns->node.children);
+
+	return fs_prio;
+}
+
+static struct fs_prio *fs_create_prio_chained(struct mlx5_flow_namespace *ns,
+					      unsigned int prio,
+					      int num_levels)
+{
+	return _fs_create_prio(ns, prio, num_levels, FS_TYPE_PRIO_CHAINS);
+}
+
+static struct fs_prio *fs_create_prio(struct mlx5_flow_namespace *ns,
+				      unsigned int prio, int num_levels)
+{
+	return _fs_create_prio(ns, prio, num_levels, FS_TYPE_PRIO);
+}
+
+static struct mlx5_flow_namespace *fs_init_namespace(struct mlx5_flow_namespace
+						     *ns)
+{
+	ns->node.type = FS_TYPE_NAMESPACE;
+
+	return ns;
+}
+
+static struct mlx5_flow_namespace *fs_create_namespace(struct fs_prio *prio,
+						       int def_miss_act)
+{
+	struct mlx5_flow_namespace	*ns;
+
+	ns = kzalloc(sizeof(*ns), GFP_KERNEL);
+	if (!ns)
+		return ERR_PTR(-ENOMEM);
+
+	fs_init_namespace(ns);
+	ns->def_miss_action = def_miss_act;
+	tree_init_node(&ns->node, NULL, del_sw_ns);
+	tree_add_node(&ns->node, &prio->node);
+	list_add_tail(&ns->node.list, &prio->node.children);
+
+	return ns;
+}
+
+static int create_leaf_prios(struct mlx5_flow_namespace *ns, int prio,
+			     struct init_tree_node *prio_metadata)
+{
+	struct fs_prio *fs_prio;
+	int i;
+
+	for (i = 0; i < prio_metadata->num_leaf_prios; i++) {
+		fs_prio = fs_create_prio(ns, prio++, prio_metadata->num_levels);
+		if (IS_ERR(fs_prio))
+			return PTR_ERR(fs_prio);
+	}
+	return 0;
+}
+
+#define FLOW_TABLE_BIT_SZ 1
+#define GET_FLOW_TABLE_CAP(dev, offset) \
+	((be32_to_cpu(*((__be32 *)(dev->caps.hca[MLX5_CAP_FLOW_TABLE]->cur) +	\
+			offset / 32)) >>					\
+	  (32 - FLOW_TABLE_BIT_SZ - (offset & 0x1f))) & FLOW_TABLE_BIT_SZ)
+static bool has_required_caps(struct mlx5_core_dev *dev, struct node_caps *caps)
+{
+	int i;
+
+	for (i = 0; i < caps->arr_sz; i++) {
+		if (!GET_FLOW_TABLE_CAP(dev, caps->caps[i]))
+			return false;
+	}
+	return true;
+}
+
+static int init_root_tree_recursive(struct mlx5_flow_steering *steering,
+				    struct init_tree_node *init_node,
+				    struct fs_node *fs_parent_node,
+				    struct init_tree_node *init_parent_node,
+				    int prio)
+{
+	int max_ft_level = MLX5_CAP_FLOWTABLE(steering->dev,
+					      flow_table_properties_nic_receive.
+					      max_ft_level);
+	struct mlx5_flow_namespace *fs_ns;
+	struct fs_prio *fs_prio;
+	struct fs_node *base;
+	int i;
+	int err;
+
+	if (init_node->type == FS_TYPE_PRIO) {
+		if ((init_node->min_ft_level > max_ft_level) ||
+		    !has_required_caps(steering->dev, &init_node->caps))
+			return 0;
+
+		fs_get_obj(fs_ns, fs_parent_node);
+		if (init_node->num_leaf_prios)
+			return create_leaf_prios(fs_ns, prio, init_node);
+		fs_prio = fs_create_prio(fs_ns, prio, init_node->num_levels);
+		if (IS_ERR(fs_prio))
+			return PTR_ERR(fs_prio);
+		base = &fs_prio->node;
+	} else if (init_node->type == FS_TYPE_NAMESPACE) {
+		fs_get_obj(fs_prio, fs_parent_node);
+		fs_ns = fs_create_namespace(fs_prio, init_node->def_miss_action);
+		if (IS_ERR(fs_ns))
+			return PTR_ERR(fs_ns);
+		base = &fs_ns->node;
+	} else {
+		return -EINVAL;
+	}
+	prio = 0;
+	for (i = 0; i < init_node->ar_size; i++) {
+		err = init_root_tree_recursive(steering, &init_node->children[i],
+					       base, init_node, prio);
+		if (err)
+			return err;
+		if (init_node->children[i].type == FS_TYPE_PRIO &&
+		    init_node->children[i].num_leaf_prios) {
+			prio += init_node->children[i].num_leaf_prios;
+		}
+	}
+
+	return 0;
+}
+
+static int init_root_tree(struct mlx5_flow_steering *steering,
+			  struct init_tree_node *init_node,
+			  struct fs_node *fs_parent_node)
+{
+	int err;
+	int i;
+
+	for (i = 0; i < init_node->ar_size; i++) {
+		err = init_root_tree_recursive(steering, &init_node->children[i],
+					       fs_parent_node,
+					       init_node, i);
+		if (err)
+			return err;
+	}
+	return 0;
+}
+
+static void del_sw_root_ns(struct fs_node *node)
+{
+	struct mlx5_flow_root_namespace *root_ns;
+	struct mlx5_flow_namespace *ns;
+
+	fs_get_obj(ns, node);
+	root_ns = container_of(ns, struct mlx5_flow_root_namespace, ns);
+	mutex_destroy(&root_ns->chain_lock);
+	kfree(node);
+}
+
+static struct mlx5_flow_root_namespace
+*create_root_ns(struct mlx5_flow_steering *steering,
+		enum fs_flow_table_type table_type)
+{
+	const struct mlx5_flow_cmds *cmds = mlx5_fs_cmd_get_default(table_type);
+	struct mlx5_flow_root_namespace *root_ns;
+	struct mlx5_flow_namespace *ns;
+
+	/* Create the root namespace */
+	root_ns = kzalloc(sizeof(*root_ns), GFP_KERNEL);
+	if (!root_ns)
+		return NULL;
+
+	root_ns->dev = steering->dev;
+	root_ns->table_type = table_type;
+	root_ns->cmds = cmds;
+
+	INIT_LIST_HEAD(&root_ns->underlay_qpns);
+
+	ns = &root_ns->ns;
+	fs_init_namespace(ns);
+	mutex_init(&root_ns->chain_lock);
+	tree_init_node(&ns->node, NULL, del_sw_root_ns);
+	tree_add_node(&ns->node, NULL);
+
+	return root_ns;
+}
+
+static void set_prio_attrs_in_prio(struct fs_prio *prio, int acc_level);
+
+static int set_prio_attrs_in_ns(struct mlx5_flow_namespace *ns, int acc_level)
+{
+	struct fs_prio *prio;
+
+	fs_for_each_prio(prio, ns) {
+		 /* This updates prio start_level and num_levels */
+		set_prio_attrs_in_prio(prio, acc_level);
+		acc_level += prio->num_levels;
+	}
+	return acc_level;
+}
+
+static void set_prio_attrs_in_prio(struct fs_prio *prio, int acc_level)
+{
+	struct mlx5_flow_namespace *ns;
+	int acc_level_ns = acc_level;
+
+	prio->start_level = acc_level;
+	fs_for_each_ns(ns, prio) {
+		/* This updates start_level and num_levels of ns's priority descendants */
+		acc_level_ns = set_prio_attrs_in_ns(ns, acc_level);
+
+		/* If this a prio with chains, and we can jump from one chain
+		 * (namespace) to another, so we accumulate the levels
+		 */
+		if (prio->node.type == FS_TYPE_PRIO_CHAINS)
+			acc_level = acc_level_ns;
+	}
+
+	if (!prio->num_levels)
+		prio->num_levels = acc_level_ns - prio->start_level;
+	WARN_ON(prio->num_levels < acc_level_ns - prio->start_level);
+}
+
+static void set_prio_attrs(struct mlx5_flow_root_namespace *root_ns)
+{
+	struct mlx5_flow_namespace *ns = &root_ns->ns;
+	struct fs_prio *prio;
+	int start_level = 0;
+
+	fs_for_each_prio(prio, ns) {
+		set_prio_attrs_in_prio(prio, start_level);
+		start_level += prio->num_levels;
+	}
+}
+
+#define ANCHOR_PRIO 0
+#define ANCHOR_SIZE 1
+#define ANCHOR_LEVEL 0
+static int create_anchor_flow_table(struct mlx5_flow_steering *steering)
+{
+	struct mlx5_flow_namespace *ns = NULL;
+	struct mlx5_flow_table_attr ft_attr = {};
+	struct mlx5_flow_table *ft;
+
+	ns = mlx5_get_flow_namespace(steering->dev, MLX5_FLOW_NAMESPACE_ANCHOR);
+	if (WARN_ON(!ns))
+		return -EINVAL;
+
+	ft_attr.max_fte = ANCHOR_SIZE;
+	ft_attr.level   = ANCHOR_LEVEL;
+	ft_attr.prio    = ANCHOR_PRIO;
+
+	ft = mlx5_create_flow_table(ns, &ft_attr);
+	if (IS_ERR(ft)) {
+		mlx5_core_err(steering->dev, "Failed to create last anchor flow table");
+		return PTR_ERR(ft);
+	}
+	return 0;
+}
+
+static int init_root_ns(struct mlx5_flow_steering *steering)
+{
+	int err;
+
+	steering->root_ns = create_root_ns(steering, FS_FT_NIC_RX);
+	if (!steering->root_ns)
+		return -ENOMEM;
+
+	err = init_root_tree(steering, &root_fs, &steering->root_ns->ns.node);
+	if (err)
+		goto out_err;
+
+	set_prio_attrs(steering->root_ns);
+	err = create_anchor_flow_table(steering);
+	if (err)
+		goto out_err;
+
+	return 0;
+
+out_err:
+	cleanup_root_ns(steering->root_ns);
+	steering->root_ns = NULL;
+	return err;
+}
+
+static void clean_tree(struct fs_node *node)
+{
+	if (node) {
+		struct fs_node *iter;
+		struct fs_node *temp;
+
+		tree_get_node(node);
+		list_for_each_entry_safe(iter, temp, &node->children, list)
+			clean_tree(iter);
+		tree_put_node(node, false);
+		tree_remove_node(node, false);
+	}
+}
+
+static void cleanup_root_ns(struct mlx5_flow_root_namespace *root_ns)
+{
+	if (!root_ns)
+		return;
+
+	clean_tree(&root_ns->ns.node);
+}
+
+static int init_sniffer_tx_root_ns(struct mlx5_flow_steering *steering)
+{
+	struct fs_prio *prio;
+
+	steering->sniffer_tx_root_ns = create_root_ns(steering, FS_FT_SNIFFER_TX);
+	if (!steering->sniffer_tx_root_ns)
+		return -ENOMEM;
+
+	/* Create single prio */
+	prio = fs_create_prio(&steering->sniffer_tx_root_ns->ns, 0, 1);
+	return PTR_ERR_OR_ZERO(prio);
+}
+
+static int init_sniffer_rx_root_ns(struct mlx5_flow_steering *steering)
+{
+	struct fs_prio *prio;
+
+	steering->sniffer_rx_root_ns = create_root_ns(steering, FS_FT_SNIFFER_RX);
+	if (!steering->sniffer_rx_root_ns)
+		return -ENOMEM;
+
+	/* Create single prio */
+	prio = fs_create_prio(&steering->sniffer_rx_root_ns->ns, 0, 1);
+	return PTR_ERR_OR_ZERO(prio);
+}
+
+#define PORT_SEL_NUM_LEVELS 3
+static int init_port_sel_root_ns(struct mlx5_flow_steering *steering)
+{
+	struct fs_prio *prio;
+
+	steering->port_sel_root_ns = create_root_ns(steering, FS_FT_PORT_SEL);
+	if (!steering->port_sel_root_ns)
+		return -ENOMEM;
+
+	/* Create single prio */
+	prio = fs_create_prio(&steering->port_sel_root_ns->ns, 0,
+			      PORT_SEL_NUM_LEVELS);
+	return PTR_ERR_OR_ZERO(prio);
+}
+
+static int init_rdma_rx_root_ns(struct mlx5_flow_steering *steering)
+{
+	int err;
+
+	steering->rdma_rx_root_ns = create_root_ns(steering, FS_FT_RDMA_RX);
+	if (!steering->rdma_rx_root_ns)
+		return -ENOMEM;
+
+	err = init_root_tree(steering, &rdma_rx_root_fs,
+			     &steering->rdma_rx_root_ns->ns.node);
+	if (err)
+		goto out_err;
+
+	set_prio_attrs(steering->rdma_rx_root_ns);
+
+	return 0;
+
+out_err:
+	cleanup_root_ns(steering->rdma_rx_root_ns);
+	steering->rdma_rx_root_ns = NULL;
+	return err;
+}
+
+static int init_rdma_tx_root_ns(struct mlx5_flow_steering *steering)
+{
+	int err;
+
+	steering->rdma_tx_root_ns = create_root_ns(steering, FS_FT_RDMA_TX);
+	if (!steering->rdma_tx_root_ns)
+		return -ENOMEM;
+
+	err = init_root_tree(steering, &rdma_tx_root_fs,
+			     &steering->rdma_tx_root_ns->ns.node);
+	if (err)
+		goto out_err;
+
+	set_prio_attrs(steering->rdma_tx_root_ns);
+
+	return 0;
+
+out_err:
+	cleanup_root_ns(steering->rdma_tx_root_ns);
+	steering->rdma_tx_root_ns = NULL;
+	return err;
+}
+
+/* FT and tc chains are stored in the same array so we can re-use the
+ * mlx5_get_fdb_sub_ns() and tc api for FT chains.
+ * When creating a new ns for each chain store it in the first available slot.
+ * Assume tc chains are created and stored first and only then the FT chain.
+ */
+static void store_fdb_sub_ns_prio_chain(struct mlx5_flow_steering *steering,
+					struct mlx5_flow_namespace *ns)
+{
+	int chain = 0;
+
+	while (steering->fdb_sub_ns[chain])
+		++chain;
+
+	steering->fdb_sub_ns[chain] = ns;
+}
+
+static int create_fdb_sub_ns_prio_chain(struct mlx5_flow_steering *steering,
+					struct fs_prio *maj_prio)
+{
+	struct mlx5_flow_namespace *ns;
+	struct fs_prio *min_prio;
+	int prio;
+
+	ns = fs_create_namespace(maj_prio, MLX5_FLOW_TABLE_MISS_ACTION_DEF);
+	if (IS_ERR(ns))
+		return PTR_ERR(ns);
+
+	for (prio = 0; prio < FDB_TC_MAX_PRIO; prio++) {
+		min_prio = fs_create_prio(ns, prio, FDB_TC_LEVELS_PER_PRIO);
+		if (IS_ERR(min_prio))
+			return PTR_ERR(min_prio);
+	}
+
+	store_fdb_sub_ns_prio_chain(steering, ns);
+
+	return 0;
+}
+
+static int create_fdb_chains(struct mlx5_flow_steering *steering,
+			     int fs_prio,
+			     int chains)
+{
+	struct fs_prio *maj_prio;
+	int levels;
+	int chain;
+	int err;
+
+	levels = FDB_TC_LEVELS_PER_PRIO * FDB_TC_MAX_PRIO * chains;
+	maj_prio = fs_create_prio_chained(&steering->fdb_root_ns->ns,
+					  fs_prio,
+					  levels);
+	if (IS_ERR(maj_prio))
+		return PTR_ERR(maj_prio);
+
+	for (chain = 0; chain < chains; chain++) {
+		err = create_fdb_sub_ns_prio_chain(steering, maj_prio);
+		if (err)
+			return err;
+	}
+
+	return 0;
+}
+
+static int create_fdb_fast_path(struct mlx5_flow_steering *steering)
+{
+	int err;
+
+	steering->fdb_sub_ns = kcalloc(FDB_NUM_CHAINS,
+				       sizeof(*steering->fdb_sub_ns),
+				       GFP_KERNEL);
+	if (!steering->fdb_sub_ns)
+		return -ENOMEM;
+
+	err = create_fdb_chains(steering, FDB_TC_OFFLOAD, FDB_TC_MAX_CHAIN + 1);
+	if (err)
+		return err;
+
+	err = create_fdb_chains(steering, FDB_FT_OFFLOAD, 1);
+	if (err)
+		return err;
+
+	return 0;
+}
+
+static int create_fdb_bypass(struct mlx5_flow_steering *steering)
+{
+	struct mlx5_flow_namespace *ns;
+	struct fs_prio *prio;
+	int i;
+
+	prio = fs_create_prio(&steering->fdb_root_ns->ns, FDB_BYPASS_PATH, 0);
+	if (IS_ERR(prio))
+		return PTR_ERR(prio);
+
+	ns = fs_create_namespace(prio, MLX5_FLOW_TABLE_MISS_ACTION_DEF);
+	if (IS_ERR(ns))
+		return PTR_ERR(ns);
+
+	for (i = 0; i < MLX5_BY_PASS_NUM_REGULAR_PRIOS; i++) {
+		prio = fs_create_prio(ns, i, 1);
+		if (IS_ERR(prio))
+			return PTR_ERR(prio);
+	}
+	return 0;
+}
+
+static void cleanup_fdb_root_ns(struct mlx5_flow_steering *steering)
+{
+	cleanup_root_ns(steering->fdb_root_ns);
+	steering->fdb_root_ns = NULL;
+	kfree(steering->fdb_sub_ns);
+	steering->fdb_sub_ns = NULL;
+}
+
+static int init_fdb_root_ns(struct mlx5_flow_steering *steering)
+{
+	struct fs_prio *maj_prio;
+	int err;
+
+	steering->fdb_root_ns = create_root_ns(steering, FS_FT_FDB);
+	if (!steering->fdb_root_ns)
+		return -ENOMEM;
+
+	err = create_fdb_bypass(steering);
+	if (err)
+		goto out_err;
+
+	err = create_fdb_fast_path(steering);
+	if (err)
+		goto out_err;
+
+	maj_prio = fs_create_prio(&steering->fdb_root_ns->ns, FDB_TC_MISS, 1);
+	if (IS_ERR(maj_prio)) {
+		err = PTR_ERR(maj_prio);
+		goto out_err;
+	}
+
+	maj_prio = fs_create_prio(&steering->fdb_root_ns->ns, FDB_BR_OFFLOAD, 3);
+	if (IS_ERR(maj_prio)) {
+		err = PTR_ERR(maj_prio);
+		goto out_err;
+	}
+
+	maj_prio = fs_create_prio(&steering->fdb_root_ns->ns, FDB_SLOW_PATH, 1);
+	if (IS_ERR(maj_prio)) {
+		err = PTR_ERR(maj_prio);
+		goto out_err;
+	}
+
+	/* We put this priority last, knowing that nothing will get here
+	 * unless explicitly forwarded to. This is possible because the
+	 * slow path tables have catch all rules and nothing gets passed
+	 * those tables.
+	 */
+	maj_prio = fs_create_prio(&steering->fdb_root_ns->ns, FDB_PER_VPORT, 1);
+	if (IS_ERR(maj_prio)) {
+		err = PTR_ERR(maj_prio);
+		goto out_err;
+	}
+
+	set_prio_attrs(steering->fdb_root_ns);
+	return 0;
+
+out_err:
+	cleanup_fdb_root_ns(steering);
+	return err;
+}
+
+static int init_egress_acl_root_ns(struct mlx5_flow_steering *steering, int vport)
+{
+	struct fs_prio *prio;
+
+	steering->esw_egress_root_ns[vport] = create_root_ns(steering, FS_FT_ESW_EGRESS_ACL);
+	if (!steering->esw_egress_root_ns[vport])
+		return -ENOMEM;
+
+	/* create 1 prio*/
+	prio = fs_create_prio(&steering->esw_egress_root_ns[vport]->ns, 0, 1);
+	return PTR_ERR_OR_ZERO(prio);
+}
+
+static int init_ingress_acl_root_ns(struct mlx5_flow_steering *steering, int vport)
+{
+	struct fs_prio *prio;
+
+	steering->esw_ingress_root_ns[vport] = create_root_ns(steering, FS_FT_ESW_INGRESS_ACL);
+	if (!steering->esw_ingress_root_ns[vport])
+		return -ENOMEM;
+
+	/* create 1 prio*/
+	prio = fs_create_prio(&steering->esw_ingress_root_ns[vport]->ns, 0, 1);
+	return PTR_ERR_OR_ZERO(prio);
+}
+
+int mlx5_fs_egress_acls_init(struct mlx5_core_dev *dev, int total_vports)
+{
+	struct mlx5_flow_steering *steering = dev->priv.steering;
+	int err;
+	int i;
+
+	steering->esw_egress_root_ns =
+			kcalloc(total_vports,
+				sizeof(*steering->esw_egress_root_ns),
+				GFP_KERNEL);
+	if (!steering->esw_egress_root_ns)
+		return -ENOMEM;
+
+	for (i = 0; i < total_vports; i++) {
+		err = init_egress_acl_root_ns(steering, i);
+		if (err)
+			goto cleanup_root_ns;
+	}
+	steering->esw_egress_acl_vports = total_vports;
+	return 0;
+
+cleanup_root_ns:
+	for (i--; i >= 0; i--)
+		cleanup_root_ns(steering->esw_egress_root_ns[i]);
+	kfree(steering->esw_egress_root_ns);
+	steering->esw_egress_root_ns = NULL;
+	return err;
+}
+
+void mlx5_fs_egress_acls_cleanup(struct mlx5_core_dev *dev)
+{
+	struct mlx5_flow_steering *steering = dev->priv.steering;
+	int i;
+
+	if (!steering->esw_egress_root_ns)
+		return;
+
+	for (i = 0; i < steering->esw_egress_acl_vports; i++)
+		cleanup_root_ns(steering->esw_egress_root_ns[i]);
+
+	kfree(steering->esw_egress_root_ns);
+	steering->esw_egress_root_ns = NULL;
+}
+
+int mlx5_fs_ingress_acls_init(struct mlx5_core_dev *dev, int total_vports)
+{
+	struct mlx5_flow_steering *steering = dev->priv.steering;
+	int err;
+	int i;
+
+	steering->esw_ingress_root_ns =
+			kcalloc(total_vports,
+				sizeof(*steering->esw_ingress_root_ns),
+				GFP_KERNEL);
+	if (!steering->esw_ingress_root_ns)
+		return -ENOMEM;
+
+	for (i = 0; i < total_vports; i++) {
+		err = init_ingress_acl_root_ns(steering, i);
+		if (err)
+			goto cleanup_root_ns;
+	}
+	steering->esw_ingress_acl_vports = total_vports;
+	return 0;
+
+cleanup_root_ns:
+	for (i--; i >= 0; i--)
+		cleanup_root_ns(steering->esw_ingress_root_ns[i]);
+	kfree(steering->esw_ingress_root_ns);
+	steering->esw_ingress_root_ns = NULL;
+	return err;
+}
+
+void mlx5_fs_ingress_acls_cleanup(struct mlx5_core_dev *dev)
+{
+	struct mlx5_flow_steering *steering = dev->priv.steering;
+	int i;
+
+	if (!steering->esw_ingress_root_ns)
+		return;
+
+	for (i = 0; i < steering->esw_ingress_acl_vports; i++)
+		cleanup_root_ns(steering->esw_ingress_root_ns[i]);
+
+	kfree(steering->esw_ingress_root_ns);
+	steering->esw_ingress_root_ns = NULL;
+}
+
+u32 mlx5_fs_get_capabilities(struct mlx5_core_dev *dev, enum mlx5_flow_namespace_type type)
+{
+	struct mlx5_flow_root_namespace *root;
+	struct mlx5_flow_namespace *ns;
+
+	ns = mlx5_get_flow_namespace(dev, type);
+	if (!ns)
+		return 0;
+
+	root = find_root(&ns->node);
+	if (!root)
+		return 0;
+
+	return root->cmds->get_capabilities(root, root->table_type);
+}
+
+static int init_egress_root_ns(struct mlx5_flow_steering *steering)
+{
+	int err;
+
+	steering->egress_root_ns = create_root_ns(steering,
+						  FS_FT_NIC_TX);
+	if (!steering->egress_root_ns)
+		return -ENOMEM;
+
+	err = init_root_tree(steering, &egress_root_fs,
+			     &steering->egress_root_ns->ns.node);
+	if (err)
+		goto cleanup;
+	set_prio_attrs(steering->egress_root_ns);
+	return 0;
+cleanup:
+	cleanup_root_ns(steering->egress_root_ns);
+	steering->egress_root_ns = NULL;
+	return err;
+}
+
+void mlx5_fs_core_cleanup(struct mlx5_core_dev *dev)
+{
+	struct mlx5_flow_steering *steering = dev->priv.steering;
+
+	cleanup_root_ns(steering->root_ns);
+	cleanup_fdb_root_ns(steering);
+	cleanup_root_ns(steering->port_sel_root_ns);
+	cleanup_root_ns(steering->sniffer_rx_root_ns);
+	cleanup_root_ns(steering->sniffer_tx_root_ns);
+	cleanup_root_ns(steering->rdma_rx_root_ns);
+	cleanup_root_ns(steering->rdma_tx_root_ns);
+	cleanup_root_ns(steering->egress_root_ns);
+}
+
+int mlx5_fs_core_init(struct mlx5_core_dev *dev)
+{
+	struct mlx5_flow_steering *steering = dev->priv.steering;
+	int err = 0;
+
+	if ((((MLX5_CAP_GEN(dev, port_type) == MLX5_CAP_PORT_TYPE_ETH) &&
+	      (MLX5_CAP_GEN(dev, nic_flow_table))) ||
+	     ((MLX5_CAP_GEN(dev, port_type) == MLX5_CAP_PORT_TYPE_IB) &&
+	      MLX5_CAP_GEN(dev, ipoib_enhanced_offloads))) &&
+	    MLX5_CAP_FLOWTABLE_NIC_RX(dev, ft_support)) {
+		err = init_root_ns(steering);
+		if (err)
+			goto err;
+	}
+
+	if (MLX5_ESWITCH_MANAGER(dev)) {
+		if (MLX5_CAP_ESW_FLOWTABLE_FDB(dev, ft_support)) {
+			err = init_fdb_root_ns(steering);
+			if (err)
+				goto err;
+		}
+	}
+
+	if (MLX5_CAP_FLOWTABLE_SNIFFER_RX(dev, ft_support)) {
+		err = init_sniffer_rx_root_ns(steering);
+		if (err)
+			goto err;
+	}
+
+	if (MLX5_CAP_FLOWTABLE_SNIFFER_TX(dev, ft_support)) {
+		err = init_sniffer_tx_root_ns(steering);
+		if (err)
+			goto err;
+	}
+
+	if (MLX5_CAP_FLOWTABLE_PORT_SELECTION(dev, ft_support)) {
+		err = init_port_sel_root_ns(steering);
+		if (err)
+			goto err;
+	}
+
+	if (MLX5_CAP_FLOWTABLE_RDMA_RX(dev, ft_support) &&
+	    MLX5_CAP_FLOWTABLE_RDMA_RX(dev, table_miss_action_domain)) {
+		err = init_rdma_rx_root_ns(steering);
+		if (err)
+			goto err;
+	}
+
+	if (MLX5_CAP_FLOWTABLE_RDMA_TX(dev, ft_support)) {
+		err = init_rdma_tx_root_ns(steering);
+		if (err)
+			goto err;
+	}
+
+	if (MLX5_CAP_FLOWTABLE_NIC_TX(dev, ft_support)) {
+		err = init_egress_root_ns(steering);
+		if (err)
+			goto err;
+	}
+
+	return 0;
+
+err:
+	mlx5_fs_core_cleanup(dev);
+	return err;
+}
+
+void mlx5_fs_core_free(struct mlx5_core_dev *dev)
+{
+	struct mlx5_flow_steering *steering = dev->priv.steering;
+
+	kmem_cache_destroy(steering->ftes_cache);
+	kmem_cache_destroy(steering->fgs_cache);
+	kfree(steering);
+	mlx5_ft_pool_destroy(dev);
+	mlx5_cleanup_fc_stats(dev);
+}
+
+int mlx5_fs_core_alloc(struct mlx5_core_dev *dev)
+{
+	struct mlx5_flow_steering *steering;
+	int err = 0;
+
+	err = mlx5_init_fc_stats(dev);
+	if (err)
+		return err;
+
+	err = mlx5_ft_pool_init(dev);
+	if (err)
+		goto err;
+
+	steering = kzalloc(sizeof(*steering), GFP_KERNEL);
+	if (!steering) {
+		err = -ENOMEM;
+		goto err;
+	}
+
+	steering->dev = dev;
+	dev->priv.steering = steering;
+
+	if (mlx5_fs_dr_is_supported(dev))
+		steering->mode = MLX5_FLOW_STEERING_MODE_SMFS;
+	else
+		steering->mode = MLX5_FLOW_STEERING_MODE_DMFS;
+
+	steering->fgs_cache = kmem_cache_create("mlx5_fs_fgs",
+						sizeof(struct mlx5_flow_group), 0,
+						0, NULL);
+	steering->ftes_cache = kmem_cache_create("mlx5_fs_ftes", sizeof(struct fs_fte), 0,
+						 0, NULL);
+	if (!steering->ftes_cache || !steering->fgs_cache) {
+		err = -ENOMEM;
+		goto err;
+	}
+
+	return 0;
+
+err:
+	mlx5_fs_core_free(dev);
+	return err;
+}
+
+int mlx5_fs_add_rx_underlay_qpn(struct mlx5_core_dev *dev, u32 underlay_qpn)
+{
+	struct mlx5_flow_root_namespace *root = dev->priv.steering->root_ns;
+	struct mlx5_ft_underlay_qp *new_uqp;
+	int err = 0;
+
+	new_uqp = kzalloc(sizeof(*new_uqp), GFP_KERNEL);
+	if (!new_uqp)
+		return -ENOMEM;
+
+	mutex_lock(&root->chain_lock);
+
+	if (!root->root_ft) {
+		err = -EINVAL;
+		goto update_ft_fail;
+	}
+
+	err = root->cmds->update_root_ft(root, root->root_ft, underlay_qpn,
+					 false);
+	if (err) {
+		mlx5_core_warn(dev, "Failed adding underlay QPN (%u) to root FT err(%d)\n",
+			       underlay_qpn, err);
+		goto update_ft_fail;
+	}
+
+	new_uqp->qpn = underlay_qpn;
+	list_add_tail(&new_uqp->list, &root->underlay_qpns);
+
+	mutex_unlock(&root->chain_lock);
+
+	return 0;
+
+update_ft_fail:
+	mutex_unlock(&root->chain_lock);
+	kfree(new_uqp);
+	return err;
+}
+EXPORT_SYMBOL(mlx5_fs_add_rx_underlay_qpn);
+
+int mlx5_fs_remove_rx_underlay_qpn(struct mlx5_core_dev *dev, u32 underlay_qpn)
+{
+	struct mlx5_flow_root_namespace *root = dev->priv.steering->root_ns;
+	struct mlx5_ft_underlay_qp *uqp;
+	bool found = false;
+	int err = 0;
+
+	mutex_lock(&root->chain_lock);
+	list_for_each_entry(uqp, &root->underlay_qpns, list) {
+		if (uqp->qpn == underlay_qpn) {
+			found = true;
+			break;
+		}
+	}
+
+	if (!found) {
+		mlx5_core_warn(dev, "Failed finding underlay qp (%u) in qpn list\n",
+			       underlay_qpn);
+		err = -EINVAL;
+		goto out;
+	}
+
+	err = root->cmds->update_root_ft(root, root->root_ft, underlay_qpn,
+					 true);
+	if (err)
+		mlx5_core_warn(dev, "Failed removing underlay QPN (%u) from root FT err(%d)\n",
+			       underlay_qpn, err);
+
+	list_del(&uqp->list);
+	mutex_unlock(&root->chain_lock);
+	kfree(uqp);
+
+	return 0;
+
+out:
+	mutex_unlock(&root->chain_lock);
+	return err;
+}
+EXPORT_SYMBOL(mlx5_fs_remove_rx_underlay_qpn);
+
+static struct mlx5_flow_root_namespace
+*get_root_namespace(struct mlx5_core_dev *dev, enum mlx5_flow_namespace_type ns_type)
+{
+	struct mlx5_flow_namespace *ns;
+
+	if (ns_type == MLX5_FLOW_NAMESPACE_ESW_EGRESS ||
+	    ns_type == MLX5_FLOW_NAMESPACE_ESW_INGRESS)
+		ns = mlx5_get_flow_vport_acl_namespace(dev, ns_type, 0);
+	else
+		ns = mlx5_get_flow_namespace(dev, ns_type);
+	if (!ns)
+		return NULL;
+
+	return find_root(&ns->node);
+}
+
+struct mlx5_modify_hdr *mlx5_modify_header_alloc(struct mlx5_core_dev *dev,
+						 u8 ns_type, u8 num_actions,
+						 void *modify_actions)
+{
+	struct mlx5_flow_root_namespace *root;
+	struct mlx5_modify_hdr *modify_hdr;
+	int err;
+
+	root = get_root_namespace(dev, ns_type);
+	if (!root)
+		return ERR_PTR(-EOPNOTSUPP);
+
+	modify_hdr = kzalloc(sizeof(*modify_hdr), GFP_KERNEL);
+	if (!modify_hdr)
+		return ERR_PTR(-ENOMEM);
+
+	modify_hdr->ns_type = ns_type;
+	err = root->cmds->modify_header_alloc(root, ns_type, num_actions,
+					      modify_actions, modify_hdr);
+	if (err) {
+		kfree(modify_hdr);
+		return ERR_PTR(err);
+	}
+
+	return modify_hdr;
+}
+EXPORT_SYMBOL(mlx5_modify_header_alloc);
+
+void mlx5_modify_header_dealloc(struct mlx5_core_dev *dev,
+				struct mlx5_modify_hdr *modify_hdr)
+{
+	struct mlx5_flow_root_namespace *root;
+
+	root = get_root_namespace(dev, modify_hdr->ns_type);
+	if (WARN_ON(!root))
+		return;
+	root->cmds->modify_header_dealloc(root, modify_hdr);
+	kfree(modify_hdr);
+}
+EXPORT_SYMBOL(mlx5_modify_header_dealloc);
+
+struct mlx5_pkt_reformat *mlx5_packet_reformat_alloc(struct mlx5_core_dev *dev,
+						     struct mlx5_pkt_reformat_params *params,
+						     enum mlx5_flow_namespace_type ns_type)
+{
+	struct mlx5_pkt_reformat *pkt_reformat;
+	struct mlx5_flow_root_namespace *root;
+	int err;
+
+	root = get_root_namespace(dev, ns_type);
+	if (!root)
+		return ERR_PTR(-EOPNOTSUPP);
+
+	pkt_reformat = kzalloc(sizeof(*pkt_reformat), GFP_KERNEL);
+	if (!pkt_reformat)
+		return ERR_PTR(-ENOMEM);
+
+	pkt_reformat->ns_type = ns_type;
+	pkt_reformat->reformat_type = params->type;
+	err = root->cmds->packet_reformat_alloc(root, params, ns_type,
+						pkt_reformat);
+	if (err) {
+		kfree(pkt_reformat);
+		return ERR_PTR(err);
+	}
+
+	return pkt_reformat;
+}
+EXPORT_SYMBOL(mlx5_packet_reformat_alloc);
+
+void mlx5_packet_reformat_dealloc(struct mlx5_core_dev *dev,
+				  struct mlx5_pkt_reformat *pkt_reformat)
+{
+	struct mlx5_flow_root_namespace *root;
+
+	root = get_root_namespace(dev, pkt_reformat->ns_type);
+	if (WARN_ON(!root))
+		return;
+	root->cmds->packet_reformat_dealloc(root, pkt_reformat);
+	kfree(pkt_reformat);
+}
+EXPORT_SYMBOL(mlx5_packet_reformat_dealloc);
+
+int mlx5_get_match_definer_id(struct mlx5_flow_definer *definer)
+{
+	return definer->id;
+}
+
+struct mlx5_flow_definer *
+mlx5_create_match_definer(struct mlx5_core_dev *dev,
+			  enum mlx5_flow_namespace_type ns_type, u16 format_id,
+			  u32 *match_mask)
+{
+	struct mlx5_flow_root_namespace *root;
+	struct mlx5_flow_definer *definer;
+	int id;
+
+	root = get_root_namespace(dev, ns_type);
+	if (!root)
+		return ERR_PTR(-EOPNOTSUPP);
+
+	definer = kzalloc(sizeof(*definer), GFP_KERNEL);
+	if (!definer)
+		return ERR_PTR(-ENOMEM);
+
+	definer->ns_type = ns_type;
+	id = root->cmds->create_match_definer(root, format_id, match_mask);
+	if (id < 0) {
+		mlx5_core_warn(root->dev, "Failed to create match definer (%d)\n", id);
+		kfree(definer);
+		return ERR_PTR(id);
+	}
+	definer->id = id;
+	return definer;
+}
+
+void mlx5_destroy_match_definer(struct mlx5_core_dev *dev,
+				struct mlx5_flow_definer *definer)
+{
+	struct mlx5_flow_root_namespace *root;
+
+	root = get_root_namespace(dev, definer->ns_type);
+	if (WARN_ON(!root))
+		return;
+
+	root->cmds->destroy_match_definer(root, definer->id);
+	kfree(definer);
+}
+
+int mlx5_flow_namespace_set_peer(struct mlx5_flow_root_namespace *ns,
+				 struct mlx5_flow_root_namespace *peer_ns)
+{
+	if (peer_ns && ns->mode != peer_ns->mode) {
+		mlx5_core_err(ns->dev,
+			      "Can't peer namespace of different steering mode\n");
+		return -EINVAL;
+	}
+
+	return ns->cmds->set_peer(ns, peer_ns);
+}
+
+/* This function should be called only at init stage of the namespace.
+ * It is not safe to call this function while steering operations
+ * are executed in the namespace.
+ */
+int mlx5_flow_namespace_set_mode(struct mlx5_flow_namespace *ns,
+				 enum mlx5_flow_steering_mode mode)
+{
+	struct mlx5_flow_root_namespace *root;
+	const struct mlx5_flow_cmds *cmds;
+	int err;
+
+	root = find_root(&ns->node);
+	if (&root->ns != ns)
+	/* Can't set cmds to non root namespace */
+		return -EINVAL;
+
+	if (root->table_type != FS_FT_FDB)
+		return -EOPNOTSUPP;
+
+	if (root->mode == mode)
+		return 0;
+
+	if (mode == MLX5_FLOW_STEERING_MODE_SMFS)
+		cmds = mlx5_fs_cmd_get_dr_cmds();
+	else
+		cmds = mlx5_fs_cmd_get_fw_cmds();
+	if (!cmds)
+		return -EOPNOTSUPP;
+
+	err = cmds->create_ns(root);
+	if (err) {
+		mlx5_core_err(root->dev, "Failed to create flow namespace (%d)\n",
+			      err);
+		return err;
+	}
+
+	root->cmds->destroy_ns(root);
+	root->cmds = cmds;
+	root->mode = mode;
+
+	return 0;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fs_core.h b/drivers/net/ethernet/mellanox/mlx5/core/fs_core.h
new file mode 100644
index 000000000..3af50fd04
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fs_core.h
@@ -0,0 +1,365 @@
+/*
+ * Copyright (c) 2015, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef _MLX5_FS_CORE_
+#define _MLX5_FS_CORE_
+
+#include <linux/refcount.h>
+#include <linux/mlx5/fs.h>
+#include <linux/rhashtable.h>
+#include <linux/llist.h>
+#include <steering/fs_dr.h>
+
+#define FDB_TC_MAX_CHAIN 3
+#define FDB_FT_CHAIN (FDB_TC_MAX_CHAIN + 1)
+#define FDB_TC_SLOW_PATH_CHAIN (FDB_FT_CHAIN + 1)
+
+/* The index of the last real chain (FT) + 1 as chain zero is valid as well */
+#define FDB_NUM_CHAINS (FDB_FT_CHAIN + 1)
+
+#define FDB_TC_MAX_PRIO 16
+#define FDB_TC_LEVELS_PER_PRIO 2
+
+struct mlx5_flow_definer {
+	enum mlx5_flow_namespace_type ns_type;
+	u32 id;
+};
+
+struct mlx5_modify_hdr {
+	enum mlx5_flow_namespace_type ns_type;
+	union {
+		struct mlx5_fs_dr_action action;
+		u32 id;
+	};
+};
+
+struct mlx5_pkt_reformat {
+	enum mlx5_flow_namespace_type ns_type;
+	int reformat_type; /* from mlx5_ifc */
+	union {
+		struct mlx5_fs_dr_action action;
+		u32 id;
+	};
+};
+
+/* FS_TYPE_PRIO_CHAINS is a PRIO that will have namespaces only,
+ * and those are in parallel to one another when going over them to connect
+ * a new flow table. Meaning the last flow table in a TYPE_PRIO prio in one
+ * parallel namespace will not automatically connect to the first flow table
+ * found in any prio in any next namespace, but skip the entire containing
+ * TYPE_PRIO_CHAINS prio.
+ *
+ * This is used to implement tc chains, each chain of prios is a different
+ * namespace inside a containing TYPE_PRIO_CHAINS prio.
+ */
+
+enum fs_node_type {
+	FS_TYPE_NAMESPACE,
+	FS_TYPE_PRIO,
+	FS_TYPE_PRIO_CHAINS,
+	FS_TYPE_FLOW_TABLE,
+	FS_TYPE_FLOW_GROUP,
+	FS_TYPE_FLOW_ENTRY,
+	FS_TYPE_FLOW_DEST
+};
+
+enum fs_flow_table_type {
+	FS_FT_NIC_RX          = 0x0,
+	FS_FT_NIC_TX          = 0x1,
+	FS_FT_ESW_EGRESS_ACL  = 0x2,
+	FS_FT_ESW_INGRESS_ACL = 0x3,
+	FS_FT_FDB             = 0X4,
+	FS_FT_SNIFFER_RX	= 0X5,
+	FS_FT_SNIFFER_TX	= 0X6,
+	FS_FT_RDMA_RX		= 0X7,
+	FS_FT_RDMA_TX		= 0X8,
+	FS_FT_PORT_SEL		= 0X9,
+	FS_FT_MAX_TYPE = FS_FT_PORT_SEL,
+};
+
+enum fs_flow_table_op_mod {
+	FS_FT_OP_MOD_NORMAL,
+	FS_FT_OP_MOD_LAG_DEMUX,
+};
+
+enum fs_fte_status {
+	FS_FTE_STATUS_EXISTING = 1UL << 0,
+};
+
+enum mlx5_flow_steering_mode {
+	MLX5_FLOW_STEERING_MODE_DMFS,
+	MLX5_FLOW_STEERING_MODE_SMFS
+};
+
+enum mlx5_flow_steering_capabilty {
+	MLX5_FLOW_STEERING_CAP_VLAN_PUSH_ON_RX = 1UL << 0,
+	MLX5_FLOW_STEERING_CAP_VLAN_POP_ON_TX = 1UL << 1,
+};
+
+struct mlx5_flow_steering {
+	struct mlx5_core_dev *dev;
+	enum   mlx5_flow_steering_mode	mode;
+	struct kmem_cache		*fgs_cache;
+	struct kmem_cache               *ftes_cache;
+	struct mlx5_flow_root_namespace *root_ns;
+	struct mlx5_flow_root_namespace *fdb_root_ns;
+	struct mlx5_flow_namespace	**fdb_sub_ns;
+	struct mlx5_flow_root_namespace **esw_egress_root_ns;
+	struct mlx5_flow_root_namespace **esw_ingress_root_ns;
+	struct mlx5_flow_root_namespace	*sniffer_tx_root_ns;
+	struct mlx5_flow_root_namespace	*sniffer_rx_root_ns;
+	struct mlx5_flow_root_namespace	*rdma_rx_root_ns;
+	struct mlx5_flow_root_namespace	*rdma_tx_root_ns;
+	struct mlx5_flow_root_namespace	*egress_root_ns;
+	struct mlx5_flow_root_namespace	*port_sel_root_ns;
+	int esw_egress_acl_vports;
+	int esw_ingress_acl_vports;
+};
+
+struct fs_node {
+	struct list_head	list;
+	struct list_head	children;
+	enum fs_node_type	type;
+	struct fs_node		*parent;
+	struct fs_node		*root;
+	/* lock the node for writing and traversing */
+	struct rw_semaphore	lock;
+	refcount_t		refcount;
+	bool			active;
+	void			(*del_hw_func)(struct fs_node *);
+	void			(*del_sw_func)(struct fs_node *);
+	atomic_t		version;
+};
+
+struct mlx5_flow_rule {
+	struct fs_node				node;
+	struct mlx5_flow_table			*ft;
+	struct mlx5_flow_destination		dest_attr;
+	/* next_ft should be accessed under chain_lock and only of
+	 * destination type is FWD_NEXT_fT.
+	 */
+	struct list_head			next_ft;
+	u32					sw_action;
+};
+
+struct mlx5_flow_handle {
+	int num_rules;
+	struct mlx5_flow_rule *rule[];
+};
+
+/* Type of children is mlx5_flow_group */
+struct mlx5_flow_table {
+	struct fs_node			node;
+	struct mlx5_fs_dr_table		fs_dr_table;
+	u32				id;
+	u16				vport;
+	unsigned int			max_fte;
+	unsigned int			level;
+	enum fs_flow_table_type		type;
+	enum fs_flow_table_op_mod	op_mod;
+	struct {
+		bool			active;
+		unsigned int		required_groups;
+		unsigned int		group_size;
+		unsigned int		num_groups;
+		unsigned int		max_fte;
+	} autogroup;
+	/* Protect fwd_rules */
+	struct mutex			lock;
+	/* FWD rules that point on this flow table */
+	struct list_head		fwd_rules;
+	u32				flags;
+	struct rhltable			fgs_hash;
+	enum mlx5_flow_table_miss_action def_miss_action;
+	struct mlx5_flow_namespace	*ns;
+};
+
+struct mlx5_ft_underlay_qp {
+	struct list_head list;
+	u32 qpn;
+};
+
+#define MLX5_FTE_MATCH_PARAM_RESERVED	reserved_at_e00
+/* Calculate the fte_match_param length and without the reserved length.
+ * Make sure the reserved field is the last.
+ */
+#define MLX5_ST_SZ_DW_MATCH_PARAM					    \
+	((MLX5_BYTE_OFF(fte_match_param, MLX5_FTE_MATCH_PARAM_RESERVED) / sizeof(u32)) + \
+	 BUILD_BUG_ON_ZERO(MLX5_ST_SZ_BYTES(fte_match_param) !=		     \
+			   MLX5_FLD_SZ_BYTES(fte_match_param,		     \
+					     MLX5_FTE_MATCH_PARAM_RESERVED) +\
+			   MLX5_BYTE_OFF(fte_match_param,		     \
+					 MLX5_FTE_MATCH_PARAM_RESERVED)))
+
+/* Type of children is mlx5_flow_rule */
+struct fs_fte {
+	struct fs_node			node;
+	struct mlx5_fs_dr_rule		fs_dr_rule;
+	u32				val[MLX5_ST_SZ_DW_MATCH_PARAM];
+	u32				dests_size;
+	u32				fwd_dests;
+	u32				index;
+	struct mlx5_flow_context	flow_context;
+	struct mlx5_flow_act		action;
+	enum fs_fte_status		status;
+	struct mlx5_fc			*counter;
+	struct rhash_head		hash;
+	int				modify_mask;
+};
+
+/* Type of children is mlx5_flow_table/namespace */
+struct fs_prio {
+	struct fs_node			node;
+	unsigned int			num_levels;
+	unsigned int			start_level;
+	unsigned int			prio;
+	unsigned int			num_ft;
+};
+
+/* Type of children is fs_prio */
+struct mlx5_flow_namespace {
+	/* parent == NULL => root ns */
+	struct	fs_node			node;
+	enum mlx5_flow_table_miss_action def_miss_action;
+};
+
+struct mlx5_flow_group_mask {
+	u8	match_criteria_enable;
+	u32	match_criteria[MLX5_ST_SZ_DW_MATCH_PARAM];
+};
+
+/* Type of children is fs_fte */
+struct mlx5_flow_group {
+	struct fs_node			node;
+	struct mlx5_fs_dr_matcher	fs_dr_matcher;
+	struct mlx5_flow_group_mask	mask;
+	u32				start_index;
+	u32				max_ftes;
+	struct ida			fte_allocator;
+	u32				id;
+	struct rhashtable		ftes_hash;
+	struct rhlist_head		hash;
+};
+
+struct mlx5_flow_root_namespace {
+	struct mlx5_flow_namespace	ns;
+	enum   mlx5_flow_steering_mode	mode;
+	struct mlx5_fs_dr_domain	fs_dr_domain;
+	enum   fs_flow_table_type	table_type;
+	struct mlx5_core_dev		*dev;
+	struct mlx5_flow_table		*root_ft;
+	/* Should be held when chaining flow tables */
+	struct mutex			chain_lock;
+	struct list_head		underlay_qpns;
+	const struct mlx5_flow_cmds	*cmds;
+};
+
+int mlx5_init_fc_stats(struct mlx5_core_dev *dev);
+void mlx5_cleanup_fc_stats(struct mlx5_core_dev *dev);
+void mlx5_fc_queue_stats_work(struct mlx5_core_dev *dev,
+			      struct delayed_work *dwork,
+			      unsigned long delay);
+void mlx5_fc_update_sampling_interval(struct mlx5_core_dev *dev,
+				      unsigned long interval);
+
+const struct mlx5_flow_cmds *mlx5_fs_cmd_get_fw_cmds(void);
+
+int mlx5_flow_namespace_set_peer(struct mlx5_flow_root_namespace *ns,
+				 struct mlx5_flow_root_namespace *peer_ns);
+
+int mlx5_flow_namespace_set_mode(struct mlx5_flow_namespace *ns,
+				 enum mlx5_flow_steering_mode mode);
+
+int mlx5_fs_core_alloc(struct mlx5_core_dev *dev);
+void mlx5_fs_core_free(struct mlx5_core_dev *dev);
+int mlx5_fs_core_init(struct mlx5_core_dev *dev);
+void mlx5_fs_core_cleanup(struct mlx5_core_dev *dev);
+
+int mlx5_fs_egress_acls_init(struct mlx5_core_dev *dev, int total_vports);
+void mlx5_fs_egress_acls_cleanup(struct mlx5_core_dev *dev);
+int mlx5_fs_ingress_acls_init(struct mlx5_core_dev *dev, int total_vports);
+void mlx5_fs_ingress_acls_cleanup(struct mlx5_core_dev *dev);
+
+u32 mlx5_fs_get_capabilities(struct mlx5_core_dev *dev, enum mlx5_flow_namespace_type type);
+
+struct mlx5_flow_root_namespace *find_root(struct fs_node *node);
+
+#define fs_get_obj(v, _node)  {v = container_of((_node), typeof(*v), node); }
+
+#define fs_list_for_each_entry(pos, root)		\
+	list_for_each_entry(pos, root, node.list)
+
+#define fs_list_for_each_entry_safe(pos, tmp, root)		\
+	list_for_each_entry_safe(pos, tmp, root, node.list)
+
+#define fs_for_each_ns_or_ft_reverse(pos, prio)				\
+	list_for_each_entry_reverse(pos, &(prio)->node.children, list)
+
+#define fs_for_each_ns_or_ft(pos, prio)					\
+	list_for_each_entry(pos, (&(prio)->node.children), list)
+
+#define fs_for_each_prio(pos, ns)			\
+	fs_list_for_each_entry(pos, &(ns)->node.children)
+
+#define fs_for_each_ns(pos, prio)			\
+	fs_list_for_each_entry(pos, &(prio)->node.children)
+
+#define fs_for_each_ft(pos, prio)			\
+	fs_list_for_each_entry(pos, &(prio)->node.children)
+
+#define fs_for_each_ft_safe(pos, tmp, prio)			\
+	fs_list_for_each_entry_safe(pos, tmp, &(prio)->node.children)
+
+#define fs_for_each_fg(pos, ft)			\
+	fs_list_for_each_entry(pos, &(ft)->node.children)
+
+#define fs_for_each_fte(pos, fg)			\
+	fs_list_for_each_entry(pos, &(fg)->node.children)
+
+#define fs_for_each_dst(pos, fte)			\
+	fs_list_for_each_entry(pos, &(fte)->node.children)
+
+#define MLX5_CAP_FLOWTABLE_TYPE(mdev, cap, type) (		\
+	(type == FS_FT_NIC_RX) ? MLX5_CAP_FLOWTABLE_NIC_RX(mdev, cap) :		\
+	(type == FS_FT_NIC_TX) ? MLX5_CAP_FLOWTABLE_NIC_TX(mdev, cap) :		\
+	(type == FS_FT_ESW_EGRESS_ACL) ? MLX5_CAP_ESW_EGRESS_ACL(mdev, cap) :		\
+	(type == FS_FT_ESW_INGRESS_ACL) ? MLX5_CAP_ESW_INGRESS_ACL(mdev, cap) :		\
+	(type == FS_FT_FDB) ? MLX5_CAP_ESW_FLOWTABLE_FDB(mdev, cap) :		\
+	(type == FS_FT_SNIFFER_RX) ? MLX5_CAP_FLOWTABLE_SNIFFER_RX(mdev, cap) :		\
+	(type == FS_FT_SNIFFER_TX) ? MLX5_CAP_FLOWTABLE_SNIFFER_TX(mdev, cap) :		\
+	(type == FS_FT_RDMA_RX) ? MLX5_CAP_FLOWTABLE_RDMA_RX(mdev, cap) :		\
+	(type == FS_FT_RDMA_TX) ? MLX5_CAP_FLOWTABLE_RDMA_TX(mdev, cap) :      \
+	(type == FS_FT_PORT_SEL) ? MLX5_CAP_FLOWTABLE_PORT_SELECTION(mdev, cap) :      \
+	(BUILD_BUG_ON_ZERO(FS_FT_PORT_SEL != FS_FT_MAX_TYPE))\
+	)
+
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fs_counters.c b/drivers/net/ethernet/mellanox/mlx5/core/fs_counters.c
new file mode 100644
index 000000000..b406e0367
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fs_counters.c
@@ -0,0 +1,762 @@
+/*
+ * Copyright (c) 2016, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/mlx5/driver.h>
+#include <linux/mlx5/fs.h>
+#include <linux/rbtree.h>
+#include "mlx5_core.h"
+#include "fs_core.h"
+#include "fs_cmd.h"
+
+#define MLX5_FC_STATS_PERIOD msecs_to_jiffies(1000)
+#define MLX5_FC_BULK_QUERY_ALLOC_PERIOD msecs_to_jiffies(180 * 1000)
+/* Max number of counters to query in bulk read is 32K */
+#define MLX5_SW_MAX_COUNTERS_BULK BIT(15)
+#define MLX5_INIT_COUNTERS_BULK 8
+#define MLX5_FC_POOL_MAX_THRESHOLD BIT(18)
+#define MLX5_FC_POOL_USED_BUFF_RATIO 10
+
+struct mlx5_fc_cache {
+	u64 packets;
+	u64 bytes;
+	u64 lastuse;
+};
+
+struct mlx5_fc {
+	struct list_head list;
+	struct llist_node addlist;
+	struct llist_node dellist;
+
+	/* last{packets,bytes} members are used when calculating the delta since
+	 * last reading
+	 */
+	u64 lastpackets;
+	u64 lastbytes;
+
+	struct mlx5_fc_bulk *bulk;
+	u32 id;
+	bool aging;
+
+	struct mlx5_fc_cache cache ____cacheline_aligned_in_smp;
+};
+
+static void mlx5_fc_pool_init(struct mlx5_fc_pool *fc_pool, struct mlx5_core_dev *dev);
+static void mlx5_fc_pool_cleanup(struct mlx5_fc_pool *fc_pool);
+static struct mlx5_fc *mlx5_fc_pool_acquire_counter(struct mlx5_fc_pool *fc_pool);
+static void mlx5_fc_pool_release_counter(struct mlx5_fc_pool *fc_pool, struct mlx5_fc *fc);
+
+/* locking scheme:
+ *
+ * It is the responsibility of the user to prevent concurrent calls or bad
+ * ordering to mlx5_fc_create(), mlx5_fc_destroy() and accessing a reference
+ * to struct mlx5_fc.
+ * e.g en_tc.c is protected by RTNL lock of its caller, and will never call a
+ * dump (access to struct mlx5_fc) after a counter is destroyed.
+ *
+ * access to counter list:
+ * - create (user context)
+ *   - mlx5_fc_create() only adds to an addlist to be used by
+ *     mlx5_fc_stats_work(). addlist is a lockless single linked list
+ *     that doesn't require any additional synchronization when adding single
+ *     node.
+ *   - spawn thread to do the actual destroy
+ *
+ * - destroy (user context)
+ *   - add a counter to lockless dellist
+ *   - spawn thread to do the actual del
+ *
+ * - dump (user context)
+ *   user should not call dump after destroy
+ *
+ * - query (single thread workqueue context)
+ *   destroy/dump - no conflict (see destroy)
+ *   query/dump - packets and bytes might be inconsistent (since update is not
+ *                atomic)
+ *   query/create - no conflict (see create)
+ *   since every create/destroy spawn the work, only after necessary time has
+ *   elapsed, the thread will actually query the hardware.
+ */
+
+static struct list_head *mlx5_fc_counters_lookup_next(struct mlx5_core_dev *dev,
+						      u32 id)
+{
+	struct mlx5_fc_stats *fc_stats = &dev->priv.fc_stats;
+	unsigned long next_id = (unsigned long)id + 1;
+	struct mlx5_fc *counter;
+	unsigned long tmp;
+
+	rcu_read_lock();
+	/* skip counters that are in idr, but not yet in counters list */
+	idr_for_each_entry_continue_ul(&fc_stats->counters_idr,
+				       counter, tmp, next_id) {
+		if (!list_empty(&counter->list))
+			break;
+	}
+	rcu_read_unlock();
+
+	return counter ? &counter->list : &fc_stats->counters;
+}
+
+static void mlx5_fc_stats_insert(struct mlx5_core_dev *dev,
+				 struct mlx5_fc *counter)
+{
+	struct list_head *next = mlx5_fc_counters_lookup_next(dev, counter->id);
+
+	list_add_tail(&counter->list, next);
+}
+
+static void mlx5_fc_stats_remove(struct mlx5_core_dev *dev,
+				 struct mlx5_fc *counter)
+{
+	struct mlx5_fc_stats *fc_stats = &dev->priv.fc_stats;
+
+	list_del(&counter->list);
+
+	spin_lock(&fc_stats->counters_idr_lock);
+	WARN_ON(!idr_remove(&fc_stats->counters_idr, counter->id));
+	spin_unlock(&fc_stats->counters_idr_lock);
+}
+
+static int get_init_bulk_query_len(struct mlx5_core_dev *dev)
+{
+	return min_t(int, MLX5_INIT_COUNTERS_BULK,
+		     (1 << MLX5_CAP_GEN(dev, log_max_flow_counter_bulk)));
+}
+
+static int get_max_bulk_query_len(struct mlx5_core_dev *dev)
+{
+	return min_t(int, MLX5_SW_MAX_COUNTERS_BULK,
+		     (1 << MLX5_CAP_GEN(dev, log_max_flow_counter_bulk)));
+}
+
+static void update_counter_cache(int index, u32 *bulk_raw_data,
+				 struct mlx5_fc_cache *cache)
+{
+	void *stats = MLX5_ADDR_OF(query_flow_counter_out, bulk_raw_data,
+			     flow_statistics[index]);
+	u64 packets = MLX5_GET64(traffic_counter, stats, packets);
+	u64 bytes = MLX5_GET64(traffic_counter, stats, octets);
+
+	if (cache->packets == packets)
+		return;
+
+	cache->packets = packets;
+	cache->bytes = bytes;
+	cache->lastuse = jiffies;
+}
+
+static void mlx5_fc_stats_query_counter_range(struct mlx5_core_dev *dev,
+					      struct mlx5_fc *first,
+					      u32 last_id)
+{
+	struct mlx5_fc_stats *fc_stats = &dev->priv.fc_stats;
+	bool query_more_counters = (first->id <= last_id);
+	int cur_bulk_len = fc_stats->bulk_query_len;
+	u32 *data = fc_stats->bulk_query_out;
+	struct mlx5_fc *counter = first;
+	u32 bulk_base_id;
+	int bulk_len;
+	int err;
+
+	while (query_more_counters) {
+		/* first id must be aligned to 4 when using bulk query */
+		bulk_base_id = counter->id & ~0x3;
+
+		/* number of counters to query inc. the last counter */
+		bulk_len = min_t(int, cur_bulk_len,
+				 ALIGN(last_id - bulk_base_id + 1, 4));
+
+		err = mlx5_cmd_fc_bulk_query(dev, bulk_base_id, bulk_len,
+					     data);
+		if (err) {
+			mlx5_core_err(dev, "Error doing bulk query: %d\n", err);
+			return;
+		}
+		query_more_counters = false;
+
+		list_for_each_entry_from(counter, &fc_stats->counters, list) {
+			int counter_index = counter->id - bulk_base_id;
+			struct mlx5_fc_cache *cache = &counter->cache;
+
+			if (counter->id >= bulk_base_id + bulk_len) {
+				query_more_counters = true;
+				break;
+			}
+
+			update_counter_cache(counter_index, data, cache);
+		}
+	}
+}
+
+static void mlx5_fc_free(struct mlx5_core_dev *dev, struct mlx5_fc *counter)
+{
+	mlx5_cmd_fc_free(dev, counter->id);
+	kfree(counter);
+}
+
+static void mlx5_fc_release(struct mlx5_core_dev *dev, struct mlx5_fc *counter)
+{
+	struct mlx5_fc_stats *fc_stats = &dev->priv.fc_stats;
+
+	if (counter->bulk)
+		mlx5_fc_pool_release_counter(&fc_stats->fc_pool, counter);
+	else
+		mlx5_fc_free(dev, counter);
+}
+
+static void mlx5_fc_stats_bulk_query_size_increase(struct mlx5_core_dev *dev)
+{
+	struct mlx5_fc_stats *fc_stats = &dev->priv.fc_stats;
+	int max_bulk_len = get_max_bulk_query_len(dev);
+	unsigned long now = jiffies;
+	u32 *bulk_query_out_tmp;
+	int max_out_len;
+
+	if (fc_stats->bulk_query_alloc_failed &&
+	    time_before(now, fc_stats->next_bulk_query_alloc))
+		return;
+
+	max_out_len = mlx5_cmd_fc_get_bulk_query_out_len(max_bulk_len);
+	bulk_query_out_tmp = kzalloc(max_out_len, GFP_KERNEL);
+	if (!bulk_query_out_tmp) {
+		mlx5_core_warn_once(dev,
+				    "Can't increase flow counters bulk query buffer size, insufficient memory, bulk_size(%d)\n",
+				    max_bulk_len);
+		fc_stats->bulk_query_alloc_failed = true;
+		fc_stats->next_bulk_query_alloc =
+			now + MLX5_FC_BULK_QUERY_ALLOC_PERIOD;
+		return;
+	}
+
+	kfree(fc_stats->bulk_query_out);
+	fc_stats->bulk_query_out = bulk_query_out_tmp;
+	fc_stats->bulk_query_len = max_bulk_len;
+	if (fc_stats->bulk_query_alloc_failed) {
+		mlx5_core_info(dev,
+			       "Flow counters bulk query buffer size increased, bulk_size(%d)\n",
+			       max_bulk_len);
+		fc_stats->bulk_query_alloc_failed = false;
+	}
+}
+
+static void mlx5_fc_stats_work(struct work_struct *work)
+{
+	struct mlx5_core_dev *dev = container_of(work, struct mlx5_core_dev,
+						 priv.fc_stats.work.work);
+	struct mlx5_fc_stats *fc_stats = &dev->priv.fc_stats;
+	/* Take dellist first to ensure that counters cannot be deleted before
+	 * they are inserted.
+	 */
+	struct llist_node *dellist = llist_del_all(&fc_stats->dellist);
+	struct llist_node *addlist = llist_del_all(&fc_stats->addlist);
+	struct mlx5_fc *counter = NULL, *last = NULL, *tmp;
+	unsigned long now = jiffies;
+
+	if (addlist || !list_empty(&fc_stats->counters))
+		queue_delayed_work(fc_stats->wq, &fc_stats->work,
+				   fc_stats->sampling_interval);
+
+	llist_for_each_entry(counter, addlist, addlist) {
+		mlx5_fc_stats_insert(dev, counter);
+		fc_stats->num_counters++;
+	}
+
+	llist_for_each_entry_safe(counter, tmp, dellist, dellist) {
+		mlx5_fc_stats_remove(dev, counter);
+
+		mlx5_fc_release(dev, counter);
+		fc_stats->num_counters--;
+	}
+
+	if (fc_stats->bulk_query_len < get_max_bulk_query_len(dev) &&
+	    fc_stats->num_counters > get_init_bulk_query_len(dev))
+		mlx5_fc_stats_bulk_query_size_increase(dev);
+
+	if (time_before(now, fc_stats->next_query) ||
+	    list_empty(&fc_stats->counters))
+		return;
+	last = list_last_entry(&fc_stats->counters, struct mlx5_fc, list);
+
+	counter = list_first_entry(&fc_stats->counters, struct mlx5_fc,
+				   list);
+	if (counter)
+		mlx5_fc_stats_query_counter_range(dev, counter, last->id);
+
+	fc_stats->next_query = now + fc_stats->sampling_interval;
+}
+
+static struct mlx5_fc *mlx5_fc_single_alloc(struct mlx5_core_dev *dev)
+{
+	struct mlx5_fc *counter;
+	int err;
+
+	counter = kzalloc(sizeof(*counter), GFP_KERNEL);
+	if (!counter)
+		return ERR_PTR(-ENOMEM);
+
+	err = mlx5_cmd_fc_alloc(dev, &counter->id);
+	if (err) {
+		kfree(counter);
+		return ERR_PTR(err);
+	}
+
+	return counter;
+}
+
+static struct mlx5_fc *mlx5_fc_acquire(struct mlx5_core_dev *dev, bool aging)
+{
+	struct mlx5_fc_stats *fc_stats = &dev->priv.fc_stats;
+	struct mlx5_fc *counter;
+
+	if (aging && MLX5_CAP_GEN(dev, flow_counter_bulk_alloc) != 0) {
+		counter = mlx5_fc_pool_acquire_counter(&fc_stats->fc_pool);
+		if (!IS_ERR(counter))
+			return counter;
+	}
+
+	return mlx5_fc_single_alloc(dev);
+}
+
+struct mlx5_fc *mlx5_fc_create_ex(struct mlx5_core_dev *dev, bool aging)
+{
+	struct mlx5_fc *counter = mlx5_fc_acquire(dev, aging);
+	struct mlx5_fc_stats *fc_stats = &dev->priv.fc_stats;
+	int err;
+
+	if (IS_ERR(counter))
+		return counter;
+
+	INIT_LIST_HEAD(&counter->list);
+	counter->aging = aging;
+
+	if (aging) {
+		u32 id = counter->id;
+
+		counter->cache.lastuse = jiffies;
+		counter->lastbytes = counter->cache.bytes;
+		counter->lastpackets = counter->cache.packets;
+
+		idr_preload(GFP_KERNEL);
+		spin_lock(&fc_stats->counters_idr_lock);
+
+		err = idr_alloc_u32(&fc_stats->counters_idr, counter, &id, id,
+				    GFP_NOWAIT);
+
+		spin_unlock(&fc_stats->counters_idr_lock);
+		idr_preload_end();
+		if (err)
+			goto err_out_alloc;
+
+		llist_add(&counter->addlist, &fc_stats->addlist);
+	}
+
+	return counter;
+
+err_out_alloc:
+	mlx5_fc_release(dev, counter);
+	return ERR_PTR(err);
+}
+
+struct mlx5_fc *mlx5_fc_create(struct mlx5_core_dev *dev, bool aging)
+{
+	struct mlx5_fc *counter = mlx5_fc_create_ex(dev, aging);
+	struct mlx5_fc_stats *fc_stats = &dev->priv.fc_stats;
+
+	if (aging)
+		mod_delayed_work(fc_stats->wq, &fc_stats->work, 0);
+	return counter;
+}
+EXPORT_SYMBOL(mlx5_fc_create);
+
+u32 mlx5_fc_id(struct mlx5_fc *counter)
+{
+	return counter->id;
+}
+EXPORT_SYMBOL(mlx5_fc_id);
+
+void mlx5_fc_destroy(struct mlx5_core_dev *dev, struct mlx5_fc *counter)
+{
+	struct mlx5_fc_stats *fc_stats = &dev->priv.fc_stats;
+
+	if (!counter)
+		return;
+
+	if (counter->aging) {
+		llist_add(&counter->dellist, &fc_stats->dellist);
+		mod_delayed_work(fc_stats->wq, &fc_stats->work, 0);
+		return;
+	}
+
+	mlx5_fc_release(dev, counter);
+}
+EXPORT_SYMBOL(mlx5_fc_destroy);
+
+int mlx5_init_fc_stats(struct mlx5_core_dev *dev)
+{
+	struct mlx5_fc_stats *fc_stats = &dev->priv.fc_stats;
+	int init_bulk_len;
+	int init_out_len;
+
+	spin_lock_init(&fc_stats->counters_idr_lock);
+	idr_init(&fc_stats->counters_idr);
+	INIT_LIST_HEAD(&fc_stats->counters);
+	init_llist_head(&fc_stats->addlist);
+	init_llist_head(&fc_stats->dellist);
+
+	init_bulk_len = get_init_bulk_query_len(dev);
+	init_out_len = mlx5_cmd_fc_get_bulk_query_out_len(init_bulk_len);
+	fc_stats->bulk_query_out = kzalloc(init_out_len, GFP_KERNEL);
+	if (!fc_stats->bulk_query_out)
+		return -ENOMEM;
+	fc_stats->bulk_query_len = init_bulk_len;
+
+	fc_stats->wq = create_singlethread_workqueue("mlx5_fc");
+	if (!fc_stats->wq)
+		goto err_wq_create;
+
+	fc_stats->sampling_interval = MLX5_FC_STATS_PERIOD;
+	INIT_DELAYED_WORK(&fc_stats->work, mlx5_fc_stats_work);
+
+	mlx5_fc_pool_init(&fc_stats->fc_pool, dev);
+	return 0;
+
+err_wq_create:
+	kfree(fc_stats->bulk_query_out);
+	return -ENOMEM;
+}
+
+void mlx5_cleanup_fc_stats(struct mlx5_core_dev *dev)
+{
+	struct mlx5_fc_stats *fc_stats = &dev->priv.fc_stats;
+	struct llist_node *tmplist;
+	struct mlx5_fc *counter;
+	struct mlx5_fc *tmp;
+
+	cancel_delayed_work_sync(&dev->priv.fc_stats.work);
+	destroy_workqueue(dev->priv.fc_stats.wq);
+	dev->priv.fc_stats.wq = NULL;
+
+	tmplist = llist_del_all(&fc_stats->addlist);
+	llist_for_each_entry_safe(counter, tmp, tmplist, addlist)
+		mlx5_fc_release(dev, counter);
+
+	list_for_each_entry_safe(counter, tmp, &fc_stats->counters, list)
+		mlx5_fc_release(dev, counter);
+
+	mlx5_fc_pool_cleanup(&fc_stats->fc_pool);
+	idr_destroy(&fc_stats->counters_idr);
+	kfree(fc_stats->bulk_query_out);
+}
+
+int mlx5_fc_query(struct mlx5_core_dev *dev, struct mlx5_fc *counter,
+		  u64 *packets, u64 *bytes)
+{
+	return mlx5_cmd_fc_query(dev, counter->id, packets, bytes);
+}
+EXPORT_SYMBOL(mlx5_fc_query);
+
+u64 mlx5_fc_query_lastuse(struct mlx5_fc *counter)
+{
+	return counter->cache.lastuse;
+}
+
+void mlx5_fc_query_cached(struct mlx5_fc *counter,
+			  u64 *bytes, u64 *packets, u64 *lastuse)
+{
+	struct mlx5_fc_cache c;
+
+	c = counter->cache;
+
+	*bytes = c.bytes - counter->lastbytes;
+	*packets = c.packets - counter->lastpackets;
+	*lastuse = c.lastuse;
+
+	counter->lastbytes = c.bytes;
+	counter->lastpackets = c.packets;
+}
+
+void mlx5_fc_queue_stats_work(struct mlx5_core_dev *dev,
+			      struct delayed_work *dwork,
+			      unsigned long delay)
+{
+	struct mlx5_fc_stats *fc_stats = &dev->priv.fc_stats;
+
+	queue_delayed_work(fc_stats->wq, dwork, delay);
+}
+
+void mlx5_fc_update_sampling_interval(struct mlx5_core_dev *dev,
+				      unsigned long interval)
+{
+	struct mlx5_fc_stats *fc_stats = &dev->priv.fc_stats;
+
+	fc_stats->sampling_interval = min_t(unsigned long, interval,
+					    fc_stats->sampling_interval);
+}
+
+/* Flow counter bluks */
+
+struct mlx5_fc_bulk {
+	struct list_head pool_list;
+	u32 base_id;
+	int bulk_len;
+	unsigned long *bitmask;
+	struct mlx5_fc fcs[];
+};
+
+static void mlx5_fc_init(struct mlx5_fc *counter, struct mlx5_fc_bulk *bulk,
+			 u32 id)
+{
+	counter->bulk = bulk;
+	counter->id = id;
+}
+
+static int mlx5_fc_bulk_get_free_fcs_amount(struct mlx5_fc_bulk *bulk)
+{
+	return bitmap_weight(bulk->bitmask, bulk->bulk_len);
+}
+
+static struct mlx5_fc_bulk *mlx5_fc_bulk_create(struct mlx5_core_dev *dev)
+{
+	enum mlx5_fc_bulk_alloc_bitmask alloc_bitmask;
+	struct mlx5_fc_bulk *bulk;
+	int err = -ENOMEM;
+	int bulk_len;
+	u32 base_id;
+	int i;
+
+	alloc_bitmask = MLX5_CAP_GEN(dev, flow_counter_bulk_alloc);
+	bulk_len = alloc_bitmask > 0 ? MLX5_FC_BULK_NUM_FCS(alloc_bitmask) : 1;
+
+	bulk = kvzalloc(struct_size(bulk, fcs, bulk_len), GFP_KERNEL);
+	if (!bulk)
+		goto err_alloc_bulk;
+
+	bulk->bitmask = kvcalloc(BITS_TO_LONGS(bulk_len), sizeof(unsigned long),
+				 GFP_KERNEL);
+	if (!bulk->bitmask)
+		goto err_alloc_bitmask;
+
+	err = mlx5_cmd_fc_bulk_alloc(dev, alloc_bitmask, &base_id);
+	if (err)
+		goto err_mlx5_cmd_bulk_alloc;
+
+	bulk->base_id = base_id;
+	bulk->bulk_len = bulk_len;
+	for (i = 0; i < bulk_len; i++) {
+		mlx5_fc_init(&bulk->fcs[i], bulk, base_id + i);
+		set_bit(i, bulk->bitmask);
+	}
+
+	return bulk;
+
+err_mlx5_cmd_bulk_alloc:
+	kvfree(bulk->bitmask);
+err_alloc_bitmask:
+	kvfree(bulk);
+err_alloc_bulk:
+	return ERR_PTR(err);
+}
+
+static int
+mlx5_fc_bulk_destroy(struct mlx5_core_dev *dev, struct mlx5_fc_bulk *bulk)
+{
+	if (mlx5_fc_bulk_get_free_fcs_amount(bulk) < bulk->bulk_len) {
+		mlx5_core_err(dev, "Freeing bulk before all counters were released\n");
+		return -EBUSY;
+	}
+
+	mlx5_cmd_fc_free(dev, bulk->base_id);
+	kvfree(bulk->bitmask);
+	kvfree(bulk);
+
+	return 0;
+}
+
+static struct mlx5_fc *mlx5_fc_bulk_acquire_fc(struct mlx5_fc_bulk *bulk)
+{
+	int free_fc_index = find_first_bit(bulk->bitmask, bulk->bulk_len);
+
+	if (free_fc_index >= bulk->bulk_len)
+		return ERR_PTR(-ENOSPC);
+
+	clear_bit(free_fc_index, bulk->bitmask);
+	return &bulk->fcs[free_fc_index];
+}
+
+static int mlx5_fc_bulk_release_fc(struct mlx5_fc_bulk *bulk, struct mlx5_fc *fc)
+{
+	int fc_index = fc->id - bulk->base_id;
+
+	if (test_bit(fc_index, bulk->bitmask))
+		return -EINVAL;
+
+	set_bit(fc_index, bulk->bitmask);
+	return 0;
+}
+
+/* Flow counters pool API */
+
+static void mlx5_fc_pool_init(struct mlx5_fc_pool *fc_pool, struct mlx5_core_dev *dev)
+{
+	fc_pool->dev = dev;
+	mutex_init(&fc_pool->pool_lock);
+	INIT_LIST_HEAD(&fc_pool->fully_used);
+	INIT_LIST_HEAD(&fc_pool->partially_used);
+	INIT_LIST_HEAD(&fc_pool->unused);
+	fc_pool->available_fcs = 0;
+	fc_pool->used_fcs = 0;
+	fc_pool->threshold = 0;
+}
+
+static void mlx5_fc_pool_cleanup(struct mlx5_fc_pool *fc_pool)
+{
+	struct mlx5_core_dev *dev = fc_pool->dev;
+	struct mlx5_fc_bulk *bulk;
+	struct mlx5_fc_bulk *tmp;
+
+	list_for_each_entry_safe(bulk, tmp, &fc_pool->fully_used, pool_list)
+		mlx5_fc_bulk_destroy(dev, bulk);
+	list_for_each_entry_safe(bulk, tmp, &fc_pool->partially_used, pool_list)
+		mlx5_fc_bulk_destroy(dev, bulk);
+	list_for_each_entry_safe(bulk, tmp, &fc_pool->unused, pool_list)
+		mlx5_fc_bulk_destroy(dev, bulk);
+}
+
+static void mlx5_fc_pool_update_threshold(struct mlx5_fc_pool *fc_pool)
+{
+	fc_pool->threshold = min_t(int, MLX5_FC_POOL_MAX_THRESHOLD,
+				   fc_pool->used_fcs / MLX5_FC_POOL_USED_BUFF_RATIO);
+}
+
+static struct mlx5_fc_bulk *
+mlx5_fc_pool_alloc_new_bulk(struct mlx5_fc_pool *fc_pool)
+{
+	struct mlx5_core_dev *dev = fc_pool->dev;
+	struct mlx5_fc_bulk *new_bulk;
+
+	new_bulk = mlx5_fc_bulk_create(dev);
+	if (!IS_ERR(new_bulk))
+		fc_pool->available_fcs += new_bulk->bulk_len;
+	mlx5_fc_pool_update_threshold(fc_pool);
+	return new_bulk;
+}
+
+static void
+mlx5_fc_pool_free_bulk(struct mlx5_fc_pool *fc_pool, struct mlx5_fc_bulk *bulk)
+{
+	struct mlx5_core_dev *dev = fc_pool->dev;
+
+	fc_pool->available_fcs -= bulk->bulk_len;
+	mlx5_fc_bulk_destroy(dev, bulk);
+	mlx5_fc_pool_update_threshold(fc_pool);
+}
+
+static struct mlx5_fc *
+mlx5_fc_pool_acquire_from_list(struct list_head *src_list,
+			       struct list_head *next_list,
+			       bool move_non_full_bulk)
+{
+	struct mlx5_fc_bulk *bulk;
+	struct mlx5_fc *fc;
+
+	if (list_empty(src_list))
+		return ERR_PTR(-ENODATA);
+
+	bulk = list_first_entry(src_list, struct mlx5_fc_bulk, pool_list);
+	fc = mlx5_fc_bulk_acquire_fc(bulk);
+	if (move_non_full_bulk || mlx5_fc_bulk_get_free_fcs_amount(bulk) == 0)
+		list_move(&bulk->pool_list, next_list);
+	return fc;
+}
+
+static struct mlx5_fc *
+mlx5_fc_pool_acquire_counter(struct mlx5_fc_pool *fc_pool)
+{
+	struct mlx5_fc_bulk *new_bulk;
+	struct mlx5_fc *fc;
+
+	mutex_lock(&fc_pool->pool_lock);
+
+	fc = mlx5_fc_pool_acquire_from_list(&fc_pool->partially_used,
+					    &fc_pool->fully_used, false);
+	if (IS_ERR(fc))
+		fc = mlx5_fc_pool_acquire_from_list(&fc_pool->unused,
+						    &fc_pool->partially_used,
+						    true);
+	if (IS_ERR(fc)) {
+		new_bulk = mlx5_fc_pool_alloc_new_bulk(fc_pool);
+		if (IS_ERR(new_bulk)) {
+			fc = ERR_CAST(new_bulk);
+			goto out;
+		}
+		fc = mlx5_fc_bulk_acquire_fc(new_bulk);
+		list_add(&new_bulk->pool_list, &fc_pool->partially_used);
+	}
+	fc_pool->available_fcs--;
+	fc_pool->used_fcs++;
+
+out:
+	mutex_unlock(&fc_pool->pool_lock);
+	return fc;
+}
+
+static void
+mlx5_fc_pool_release_counter(struct mlx5_fc_pool *fc_pool, struct mlx5_fc *fc)
+{
+	struct mlx5_core_dev *dev = fc_pool->dev;
+	struct mlx5_fc_bulk *bulk = fc->bulk;
+	int bulk_free_fcs_amount;
+
+	mutex_lock(&fc_pool->pool_lock);
+
+	if (mlx5_fc_bulk_release_fc(bulk, fc)) {
+		mlx5_core_warn(dev, "Attempted to release a counter which is not acquired\n");
+		goto unlock;
+	}
+
+	fc_pool->available_fcs++;
+	fc_pool->used_fcs--;
+
+	bulk_free_fcs_amount = mlx5_fc_bulk_get_free_fcs_amount(bulk);
+	if (bulk_free_fcs_amount == 1)
+		list_move_tail(&bulk->pool_list, &fc_pool->partially_used);
+	if (bulk_free_fcs_amount == bulk->bulk_len) {
+		list_del(&bulk->pool_list);
+		if (fc_pool->available_fcs > fc_pool->threshold)
+			mlx5_fc_pool_free_bulk(fc_pool, bulk);
+		else
+			list_add(&bulk->pool_list, &fc_pool->unused);
+	}
+
+unlock:
+	mutex_unlock(&fc_pool->pool_lock);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fs_ft_pool.c b/drivers/net/ethernet/mellanox/mlx5/core/fs_ft_pool.c
new file mode 100644
index 000000000..c14590acc
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fs_ft_pool.c
@@ -0,0 +1,85 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2021 Mellanox Technologies. */
+
+#include "fs_ft_pool.h"
+
+/* Firmware currently has 4 pool of 4 sizes that it supports (FT_POOLS),
+ * and a virtual memory region of 16M (MLX5_FT_SIZE), this region is duplicated
+ * for each flow table pool. We can allocate up to 16M of each pool,
+ * and we keep track of how much we used via mlx5_ft_pool_get_avail_sz.
+ * Firmware doesn't report any of this for now.
+ * ESW_POOL is expected to be sorted from large to small and match firmware
+ * pools.
+ */
+#define FT_SIZE (16 * 1024 * 1024)
+static const unsigned int FT_POOLS[] = { 4 * 1024 * 1024,
+					 1 * 1024 * 1024,
+					 64 * 1024,
+					 128,
+					 1 /* size for termination tables */ };
+struct mlx5_ft_pool {
+	int ft_left[ARRAY_SIZE(FT_POOLS)];
+};
+
+int mlx5_ft_pool_init(struct mlx5_core_dev *dev)
+{
+	struct mlx5_ft_pool *ft_pool;
+	int i;
+
+	ft_pool = kzalloc(sizeof(*ft_pool), GFP_KERNEL);
+	if (!ft_pool)
+		return -ENOMEM;
+
+	for (i = ARRAY_SIZE(FT_POOLS) - 1; i >= 0; i--)
+		ft_pool->ft_left[i] = FT_SIZE / FT_POOLS[i];
+
+	dev->priv.ft_pool = ft_pool;
+	return 0;
+}
+
+void mlx5_ft_pool_destroy(struct mlx5_core_dev *dev)
+{
+	kfree(dev->priv.ft_pool);
+}
+
+int
+mlx5_ft_pool_get_avail_sz(struct mlx5_core_dev *dev, enum fs_flow_table_type table_type,
+			  int desired_size)
+{
+	u32 max_ft_size = 1 << MLX5_CAP_FLOWTABLE_TYPE(dev, log_max_ft_size, table_type);
+	int i, found_i = -1;
+
+	for (i = ARRAY_SIZE(FT_POOLS) - 1; i >= 0; i--) {
+		if (dev->priv.ft_pool->ft_left[i] && FT_POOLS[i] >= desired_size &&
+		    FT_POOLS[i] <= max_ft_size) {
+			found_i = i;
+			if (desired_size != POOL_NEXT_SIZE)
+				break;
+		}
+	}
+
+	if (found_i != -1) {
+		--dev->priv.ft_pool->ft_left[found_i];
+		return FT_POOLS[found_i];
+	}
+
+	return 0;
+}
+
+void
+mlx5_ft_pool_put_sz(struct mlx5_core_dev *dev, int sz)
+{
+	int i;
+
+	if (!sz)
+		return;
+
+	for (i = ARRAY_SIZE(FT_POOLS) - 1; i >= 0; i--) {
+		if (sz == FT_POOLS[i]) {
+			++dev->priv.ft_pool->ft_left[i];
+			return;
+		}
+	}
+
+	WARN_ONCE(1, "Couldn't find size %d in flow table size pool", sz);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fs_ft_pool.h b/drivers/net/ethernet/mellanox/mlx5/core/fs_ft_pool.h
new file mode 100644
index 000000000..25f4274b3
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fs_ft_pool.h
@@ -0,0 +1,21 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2021 Mellanox Technologies. */
+
+#ifndef __MLX5_FS_FT_POOL_H__
+#define __MLX5_FS_FT_POOL_H__
+
+#include <linux/mlx5/driver.h>
+#include "fs_core.h"
+
+#define POOL_NEXT_SIZE 0
+
+int mlx5_ft_pool_init(struct mlx5_core_dev *dev);
+void mlx5_ft_pool_destroy(struct mlx5_core_dev *dev);
+
+int
+mlx5_ft_pool_get_avail_sz(struct mlx5_core_dev *dev, enum fs_flow_table_type table_type,
+			  int desired_size);
+void
+mlx5_ft_pool_put_sz(struct mlx5_core_dev *dev, int sz);
+
+#endif /* __MLX5_FS_FT_POOL_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fw.c b/drivers/net/ethernet/mellanox/mlx5/core/fw.c
new file mode 100644
index 000000000..f34e758a2
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fw.c
@@ -0,0 +1,851 @@
+/*
+ * Copyright (c) 2013-2015, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/mlx5/driver.h>
+#include <linux/mlx5/eswitch.h>
+#include "mlx5_core.h"
+#include "../../mlxfw/mlxfw.h"
+#include "lib/tout.h"
+
+enum {
+	MCQS_IDENTIFIER_BOOT_IMG	= 0x1,
+	MCQS_IDENTIFIER_OEM_NVCONFIG	= 0x4,
+	MCQS_IDENTIFIER_MLNX_NVCONFIG	= 0x5,
+	MCQS_IDENTIFIER_CS_TOKEN	= 0x6,
+	MCQS_IDENTIFIER_DBG_TOKEN	= 0x7,
+	MCQS_IDENTIFIER_GEARBOX		= 0xA,
+};
+
+enum {
+	MCQS_UPDATE_STATE_IDLE,
+	MCQS_UPDATE_STATE_IN_PROGRESS,
+	MCQS_UPDATE_STATE_APPLIED,
+	MCQS_UPDATE_STATE_ACTIVE,
+	MCQS_UPDATE_STATE_ACTIVE_PENDING_RESET,
+	MCQS_UPDATE_STATE_FAILED,
+	MCQS_UPDATE_STATE_CANCELED,
+	MCQS_UPDATE_STATE_BUSY,
+};
+
+enum {
+	MCQI_INFO_TYPE_CAPABILITIES	  = 0x0,
+	MCQI_INFO_TYPE_VERSION		  = 0x1,
+	MCQI_INFO_TYPE_ACTIVATION_METHOD  = 0x5,
+};
+
+enum {
+	MCQI_FW_RUNNING_VERSION = 0,
+	MCQI_FW_STORED_VERSION  = 1,
+};
+
+int mlx5_query_board_id(struct mlx5_core_dev *dev)
+{
+	u32 *out;
+	int outlen = MLX5_ST_SZ_BYTES(query_adapter_out);
+	u32 in[MLX5_ST_SZ_DW(query_adapter_in)] = {};
+	int err;
+
+	out = kzalloc(outlen, GFP_KERNEL);
+	if (!out)
+		return -ENOMEM;
+
+	MLX5_SET(query_adapter_in, in, opcode, MLX5_CMD_OP_QUERY_ADAPTER);
+	err = mlx5_cmd_exec_inout(dev, query_adapter, in, out);
+	if (err)
+		goto out;
+
+	memcpy(dev->board_id,
+	       MLX5_ADDR_OF(query_adapter_out, out,
+			    query_adapter_struct.vsd_contd_psid),
+	       MLX5_FLD_SZ_BYTES(query_adapter_out,
+				 query_adapter_struct.vsd_contd_psid));
+
+out:
+	kfree(out);
+	return err;
+}
+
+int mlx5_core_query_vendor_id(struct mlx5_core_dev *mdev, u32 *vendor_id)
+{
+	u32 *out;
+	int outlen = MLX5_ST_SZ_BYTES(query_adapter_out);
+	u32 in[MLX5_ST_SZ_DW(query_adapter_in)] = {};
+	int err;
+
+	out = kzalloc(outlen, GFP_KERNEL);
+	if (!out)
+		return -ENOMEM;
+
+	MLX5_SET(query_adapter_in, in, opcode, MLX5_CMD_OP_QUERY_ADAPTER);
+	err = mlx5_cmd_exec_inout(mdev, query_adapter, in, out);
+	if (err)
+		goto out;
+
+	*vendor_id = MLX5_GET(query_adapter_out, out,
+			      query_adapter_struct.ieee_vendor_id);
+out:
+	kfree(out);
+	return err;
+}
+EXPORT_SYMBOL(mlx5_core_query_vendor_id);
+
+static int mlx5_get_pcam_reg(struct mlx5_core_dev *dev)
+{
+	return mlx5_query_pcam_reg(dev, dev->caps.pcam,
+				   MLX5_PCAM_FEATURE_ENHANCED_FEATURES,
+				   MLX5_PCAM_REGS_5000_TO_507F);
+}
+
+static int mlx5_get_mcam_access_reg_group(struct mlx5_core_dev *dev,
+					  enum mlx5_mcam_reg_groups group)
+{
+	return mlx5_query_mcam_reg(dev, dev->caps.mcam[group],
+				   MLX5_MCAM_FEATURE_ENHANCED_FEATURES, group);
+}
+
+static int mlx5_get_qcam_reg(struct mlx5_core_dev *dev)
+{
+	return mlx5_query_qcam_reg(dev, dev->caps.qcam,
+				   MLX5_QCAM_FEATURE_ENHANCED_FEATURES,
+				   MLX5_QCAM_REGS_FIRST_128);
+}
+
+int mlx5_query_hca_caps(struct mlx5_core_dev *dev)
+{
+	int err;
+
+	err = mlx5_core_get_caps(dev, MLX5_CAP_GENERAL);
+	if (err)
+		return err;
+
+	if (MLX5_CAP_GEN(dev, port_selection_cap)) {
+		err = mlx5_core_get_caps(dev, MLX5_CAP_PORT_SELECTION);
+		if (err)
+			return err;
+	}
+
+	if (MLX5_CAP_GEN(dev, hca_cap_2)) {
+		err = mlx5_core_get_caps(dev, MLX5_CAP_GENERAL_2);
+		if (err)
+			return err;
+	}
+
+	if (MLX5_CAP_GEN(dev, eth_net_offloads)) {
+		err = mlx5_core_get_caps(dev, MLX5_CAP_ETHERNET_OFFLOADS);
+		if (err)
+			return err;
+	}
+
+	if (MLX5_CAP_GEN(dev, ipoib_enhanced_offloads)) {
+		err = mlx5_core_get_caps(dev, MLX5_CAP_IPOIB_ENHANCED_OFFLOADS);
+		if (err)
+			return err;
+	}
+
+	if (MLX5_CAP_GEN(dev, pg)) {
+		err = mlx5_core_get_caps(dev, MLX5_CAP_ODP);
+		if (err)
+			return err;
+	}
+
+	if (MLX5_CAP_GEN(dev, atomic)) {
+		err = mlx5_core_get_caps(dev, MLX5_CAP_ATOMIC);
+		if (err)
+			return err;
+	}
+
+	if (MLX5_CAP_GEN(dev, roce)) {
+		err = mlx5_core_get_caps(dev, MLX5_CAP_ROCE);
+		if (err)
+			return err;
+	}
+
+	if (MLX5_CAP_GEN(dev, nic_flow_table) ||
+	    MLX5_CAP_GEN(dev, ipoib_enhanced_offloads)) {
+		err = mlx5_core_get_caps(dev, MLX5_CAP_FLOW_TABLE);
+		if (err)
+			return err;
+	}
+
+	if (MLX5_CAP_GEN(dev, vport_group_manager) &&
+	    MLX5_ESWITCH_MANAGER(dev)) {
+		err = mlx5_core_get_caps(dev, MLX5_CAP_ESWITCH_FLOW_TABLE);
+		if (err)
+			return err;
+	}
+
+	if (MLX5_ESWITCH_MANAGER(dev)) {
+		err = mlx5_core_get_caps(dev, MLX5_CAP_ESWITCH);
+		if (err)
+			return err;
+	}
+
+	if (MLX5_CAP_GEN(dev, vector_calc)) {
+		err = mlx5_core_get_caps(dev, MLX5_CAP_VECTOR_CALC);
+		if (err)
+			return err;
+	}
+
+	if (MLX5_CAP_GEN(dev, qos)) {
+		err = mlx5_core_get_caps(dev, MLX5_CAP_QOS);
+		if (err)
+			return err;
+	}
+
+	if (MLX5_CAP_GEN(dev, debug))
+		mlx5_core_get_caps(dev, MLX5_CAP_DEBUG);
+
+	if (MLX5_CAP_GEN(dev, pcam_reg))
+		mlx5_get_pcam_reg(dev);
+
+	if (MLX5_CAP_GEN(dev, mcam_reg)) {
+		mlx5_get_mcam_access_reg_group(dev, MLX5_MCAM_REGS_FIRST_128);
+		mlx5_get_mcam_access_reg_group(dev, MLX5_MCAM_REGS_0x9080_0x90FF);
+		mlx5_get_mcam_access_reg_group(dev, MLX5_MCAM_REGS_0x9100_0x917F);
+	}
+
+	if (MLX5_CAP_GEN(dev, qcam_reg))
+		mlx5_get_qcam_reg(dev);
+
+	if (MLX5_CAP_GEN(dev, device_memory)) {
+		err = mlx5_core_get_caps(dev, MLX5_CAP_DEV_MEM);
+		if (err)
+			return err;
+	}
+
+	if (MLX5_CAP_GEN(dev, event_cap)) {
+		err = mlx5_core_get_caps(dev, MLX5_CAP_DEV_EVENT);
+		if (err)
+			return err;
+	}
+
+	if (MLX5_CAP_GEN(dev, tls_tx) || MLX5_CAP_GEN(dev, tls_rx)) {
+		err = mlx5_core_get_caps(dev, MLX5_CAP_TLS);
+		if (err)
+			return err;
+	}
+
+	if (MLX5_CAP_GEN_64(dev, general_obj_types) &
+		MLX5_GENERAL_OBJ_TYPES_CAP_VIRTIO_NET_Q) {
+		err = mlx5_core_get_caps(dev, MLX5_CAP_VDPA_EMULATION);
+		if (err)
+			return err;
+	}
+
+	if (MLX5_CAP_GEN(dev, ipsec_offload)) {
+		err = mlx5_core_get_caps(dev, MLX5_CAP_IPSEC);
+		if (err)
+			return err;
+	}
+
+	if (MLX5_CAP_GEN(dev, shampo)) {
+		err = mlx5_core_get_caps(dev, MLX5_CAP_DEV_SHAMPO);
+		if (err)
+			return err;
+	}
+
+	if (MLX5_CAP_GEN_64(dev, general_obj_types) &
+	    MLX5_GENERAL_OBJ_TYPES_CAP_MACSEC_OFFLOAD) {
+		err = mlx5_core_get_caps(dev, MLX5_CAP_MACSEC);
+		if (err)
+			return err;
+	}
+
+	if (MLX5_CAP_GEN(dev, adv_virtualization)) {
+		err = mlx5_core_get_caps(dev, MLX5_CAP_ADV_VIRTUALIZATION);
+		if (err)
+			return err;
+	}
+
+	return 0;
+}
+
+int mlx5_cmd_init_hca(struct mlx5_core_dev *dev, uint32_t *sw_owner_id)
+{
+	u32 in[MLX5_ST_SZ_DW(init_hca_in)] = {};
+	int i;
+
+	MLX5_SET(init_hca_in, in, opcode, MLX5_CMD_OP_INIT_HCA);
+
+	if (MLX5_CAP_GEN(dev, sw_owner_id)) {
+		for (i = 0; i < 4; i++)
+			MLX5_ARRAY_SET(init_hca_in, in, sw_owner_id, i,
+				       sw_owner_id[i]);
+	}
+
+	if (MLX5_CAP_GEN_2_MAX(dev, sw_vhca_id_valid) &&
+	    dev->priv.sw_vhca_id > 0)
+		MLX5_SET(init_hca_in, in, sw_vhca_id, dev->priv.sw_vhca_id);
+
+	return mlx5_cmd_exec_in(dev, init_hca, in);
+}
+
+int mlx5_cmd_teardown_hca(struct mlx5_core_dev *dev)
+{
+	u32 in[MLX5_ST_SZ_DW(teardown_hca_in)] = {};
+
+	MLX5_SET(teardown_hca_in, in, opcode, MLX5_CMD_OP_TEARDOWN_HCA);
+	return mlx5_cmd_exec_in(dev, teardown_hca, in);
+}
+
+int mlx5_cmd_force_teardown_hca(struct mlx5_core_dev *dev)
+{
+	u32 out[MLX5_ST_SZ_DW(teardown_hca_out)] = {0};
+	u32 in[MLX5_ST_SZ_DW(teardown_hca_in)] = {0};
+	int force_state;
+	int ret;
+
+	if (!MLX5_CAP_GEN(dev, force_teardown)) {
+		mlx5_core_dbg(dev, "force teardown is not supported in the firmware\n");
+		return -EOPNOTSUPP;
+	}
+
+	MLX5_SET(teardown_hca_in, in, opcode, MLX5_CMD_OP_TEARDOWN_HCA);
+	MLX5_SET(teardown_hca_in, in, profile, MLX5_TEARDOWN_HCA_IN_PROFILE_FORCE_CLOSE);
+
+	ret = mlx5_cmd_exec_polling(dev, in, sizeof(in), out, sizeof(out));
+	if (ret)
+		return ret;
+
+	force_state = MLX5_GET(teardown_hca_out, out, state);
+	if (force_state == MLX5_TEARDOWN_HCA_OUT_FORCE_STATE_FAIL) {
+		mlx5_core_warn(dev, "teardown with force mode failed, doing normal teardown\n");
+		return -EIO;
+	}
+
+	return 0;
+}
+
+int mlx5_cmd_fast_teardown_hca(struct mlx5_core_dev *dev)
+{
+	unsigned long end, delay_ms = mlx5_tout_ms(dev, TEARDOWN);
+	u32 out[MLX5_ST_SZ_DW(teardown_hca_out)] = {};
+	u32 in[MLX5_ST_SZ_DW(teardown_hca_in)] = {};
+	int state;
+	int ret;
+
+	if (!MLX5_CAP_GEN(dev, fast_teardown)) {
+		mlx5_core_dbg(dev, "fast teardown is not supported in the firmware\n");
+		return -EOPNOTSUPP;
+	}
+
+	MLX5_SET(teardown_hca_in, in, opcode, MLX5_CMD_OP_TEARDOWN_HCA);
+	MLX5_SET(teardown_hca_in, in, profile,
+		 MLX5_TEARDOWN_HCA_IN_PROFILE_PREPARE_FAST_TEARDOWN);
+
+	ret = mlx5_cmd_exec_inout(dev, teardown_hca, in, out);
+	if (ret)
+		return ret;
+
+	state = MLX5_GET(teardown_hca_out, out, state);
+	if (state == MLX5_TEARDOWN_HCA_OUT_FORCE_STATE_FAIL) {
+		mlx5_core_warn(dev, "teardown with fast mode failed\n");
+		return -EIO;
+	}
+
+	mlx5_set_nic_state(dev, MLX5_NIC_IFC_DISABLED);
+
+	/* Loop until device state turns to disable */
+	end = jiffies + msecs_to_jiffies(delay_ms);
+	do {
+		if (mlx5_get_nic_state(dev) == MLX5_NIC_IFC_DISABLED)
+			break;
+
+		cond_resched();
+	} while (!time_after(jiffies, end));
+
+	if (mlx5_get_nic_state(dev) != MLX5_NIC_IFC_DISABLED) {
+		dev_err(&dev->pdev->dev, "NIC IFC still %d after %lums.\n",
+			mlx5_get_nic_state(dev), delay_ms);
+		return -EIO;
+	}
+
+	return 0;
+}
+
+enum mlxsw_reg_mcc_instruction {
+	MLX5_REG_MCC_INSTRUCTION_LOCK_UPDATE_HANDLE = 0x01,
+	MLX5_REG_MCC_INSTRUCTION_RELEASE_UPDATE_HANDLE = 0x02,
+	MLX5_REG_MCC_INSTRUCTION_UPDATE_COMPONENT = 0x03,
+	MLX5_REG_MCC_INSTRUCTION_VERIFY_COMPONENT = 0x04,
+	MLX5_REG_MCC_INSTRUCTION_ACTIVATE = 0x06,
+	MLX5_REG_MCC_INSTRUCTION_CANCEL = 0x08,
+};
+
+static int mlx5_reg_mcc_set(struct mlx5_core_dev *dev,
+			    enum mlxsw_reg_mcc_instruction instr,
+			    u16 component_index, u32 update_handle,
+			    u32 component_size)
+{
+	u32 out[MLX5_ST_SZ_DW(mcc_reg)];
+	u32 in[MLX5_ST_SZ_DW(mcc_reg)];
+
+	memset(in, 0, sizeof(in));
+
+	MLX5_SET(mcc_reg, in, instruction, instr);
+	MLX5_SET(mcc_reg, in, component_index, component_index);
+	MLX5_SET(mcc_reg, in, update_handle, update_handle);
+	MLX5_SET(mcc_reg, in, component_size, component_size);
+
+	return mlx5_core_access_reg(dev, in, sizeof(in), out,
+				    sizeof(out), MLX5_REG_MCC, 0, 1);
+}
+
+static int mlx5_reg_mcc_query(struct mlx5_core_dev *dev,
+			      u32 *update_handle, u8 *error_code,
+			      u8 *control_state)
+{
+	u32 out[MLX5_ST_SZ_DW(mcc_reg)];
+	u32 in[MLX5_ST_SZ_DW(mcc_reg)];
+	int err;
+
+	memset(in, 0, sizeof(in));
+	memset(out, 0, sizeof(out));
+	MLX5_SET(mcc_reg, in, update_handle, *update_handle);
+
+	err = mlx5_core_access_reg(dev, in, sizeof(in), out,
+				   sizeof(out), MLX5_REG_MCC, 0, 0);
+	if (err)
+		goto out;
+
+	*update_handle = MLX5_GET(mcc_reg, out, update_handle);
+	*error_code = MLX5_GET(mcc_reg, out, error_code);
+	*control_state = MLX5_GET(mcc_reg, out, control_state);
+
+out:
+	return err;
+}
+
+static int mlx5_reg_mcda_set(struct mlx5_core_dev *dev,
+			     u32 update_handle,
+			     u32 offset, u16 size,
+			     u8 *data)
+{
+	int err, in_size = MLX5_ST_SZ_BYTES(mcda_reg) + size;
+	u32 out[MLX5_ST_SZ_DW(mcda_reg)];
+	int i, j, dw_size = size >> 2;
+	__be32 data_element;
+	u32 *in;
+
+	in = kzalloc(in_size, GFP_KERNEL);
+	if (!in)
+		return -ENOMEM;
+
+	MLX5_SET(mcda_reg, in, update_handle, update_handle);
+	MLX5_SET(mcda_reg, in, offset, offset);
+	MLX5_SET(mcda_reg, in, size, size);
+
+	for (i = 0; i < dw_size; i++) {
+		j = i * 4;
+		data_element = htonl(*(u32 *)&data[j]);
+		memcpy(MLX5_ADDR_OF(mcda_reg, in, data) + j, &data_element, 4);
+	}
+
+	err = mlx5_core_access_reg(dev, in, in_size, out,
+				   sizeof(out), MLX5_REG_MCDA, 0, 1);
+	kfree(in);
+	return err;
+}
+
+static int mlx5_reg_mcqi_query(struct mlx5_core_dev *dev,
+			       u16 component_index, bool read_pending,
+			       u8 info_type, u16 data_size, void *mcqi_data)
+{
+	u32 out[MLX5_ST_SZ_DW(mcqi_reg) + MLX5_UN_SZ_DW(mcqi_reg_data)] = {};
+	u32 in[MLX5_ST_SZ_DW(mcqi_reg)] = {};
+	void *data;
+	int err;
+
+	MLX5_SET(mcqi_reg, in, component_index, component_index);
+	MLX5_SET(mcqi_reg, in, read_pending_component, read_pending);
+	MLX5_SET(mcqi_reg, in, info_type, info_type);
+	MLX5_SET(mcqi_reg, in, data_size, data_size);
+
+	err = mlx5_core_access_reg(dev, in, sizeof(in), out,
+				   MLX5_ST_SZ_BYTES(mcqi_reg) + data_size,
+				   MLX5_REG_MCQI, 0, 0);
+	if (err)
+		return err;
+
+	data = MLX5_ADDR_OF(mcqi_reg, out, data);
+	memcpy(mcqi_data, data, data_size);
+
+	return 0;
+}
+
+static int mlx5_reg_mcqi_caps_query(struct mlx5_core_dev *dev, u16 component_index,
+				    u32 *max_component_size, u8 *log_mcda_word_size,
+				    u16 *mcda_max_write_size)
+{
+	u32 mcqi_reg[MLX5_ST_SZ_DW(mcqi_cap)] = {};
+	int err;
+
+	err = mlx5_reg_mcqi_query(dev, component_index, 0,
+				  MCQI_INFO_TYPE_CAPABILITIES,
+				  MLX5_ST_SZ_BYTES(mcqi_cap), mcqi_reg);
+	if (err)
+		return err;
+
+	*max_component_size = MLX5_GET(mcqi_cap, mcqi_reg, max_component_size);
+	*log_mcda_word_size = MLX5_GET(mcqi_cap, mcqi_reg, log_mcda_word_size);
+	*mcda_max_write_size = MLX5_GET(mcqi_cap, mcqi_reg, mcda_max_write_size);
+
+	return 0;
+}
+
+struct mlx5_mlxfw_dev {
+	struct mlxfw_dev mlxfw_dev;
+	struct mlx5_core_dev *mlx5_core_dev;
+};
+
+static int mlx5_component_query(struct mlxfw_dev *mlxfw_dev,
+				u16 component_index, u32 *p_max_size,
+				u8 *p_align_bits, u16 *p_max_write_size)
+{
+	struct mlx5_mlxfw_dev *mlx5_mlxfw_dev =
+		container_of(mlxfw_dev, struct mlx5_mlxfw_dev, mlxfw_dev);
+	struct mlx5_core_dev *dev = mlx5_mlxfw_dev->mlx5_core_dev;
+
+	if (!MLX5_CAP_GEN(dev, mcam_reg) || !MLX5_CAP_MCAM_REG(dev, mcqi)) {
+		mlx5_core_warn(dev, "caps query isn't supported by running FW\n");
+		return -EOPNOTSUPP;
+	}
+
+	return mlx5_reg_mcqi_caps_query(dev, component_index, p_max_size,
+					p_align_bits, p_max_write_size);
+}
+
+static int mlx5_fsm_lock(struct mlxfw_dev *mlxfw_dev, u32 *fwhandle)
+{
+	struct mlx5_mlxfw_dev *mlx5_mlxfw_dev =
+		container_of(mlxfw_dev, struct mlx5_mlxfw_dev, mlxfw_dev);
+	struct mlx5_core_dev *dev = mlx5_mlxfw_dev->mlx5_core_dev;
+	u8 control_state, error_code;
+	int err;
+
+	*fwhandle = 0;
+	err = mlx5_reg_mcc_query(dev, fwhandle, &error_code, &control_state);
+	if (err)
+		return err;
+
+	if (control_state != MLXFW_FSM_STATE_IDLE)
+		return -EBUSY;
+
+	return mlx5_reg_mcc_set(dev, MLX5_REG_MCC_INSTRUCTION_LOCK_UPDATE_HANDLE,
+				0, *fwhandle, 0);
+}
+
+static int mlx5_fsm_component_update(struct mlxfw_dev *mlxfw_dev, u32 fwhandle,
+				     u16 component_index, u32 component_size)
+{
+	struct mlx5_mlxfw_dev *mlx5_mlxfw_dev =
+		container_of(mlxfw_dev, struct mlx5_mlxfw_dev, mlxfw_dev);
+	struct mlx5_core_dev *dev = mlx5_mlxfw_dev->mlx5_core_dev;
+
+	return mlx5_reg_mcc_set(dev, MLX5_REG_MCC_INSTRUCTION_UPDATE_COMPONENT,
+				component_index, fwhandle, component_size);
+}
+
+static int mlx5_fsm_block_download(struct mlxfw_dev *mlxfw_dev, u32 fwhandle,
+				   u8 *data, u16 size, u32 offset)
+{
+	struct mlx5_mlxfw_dev *mlx5_mlxfw_dev =
+		container_of(mlxfw_dev, struct mlx5_mlxfw_dev, mlxfw_dev);
+	struct mlx5_core_dev *dev = mlx5_mlxfw_dev->mlx5_core_dev;
+
+	return mlx5_reg_mcda_set(dev, fwhandle, offset, size, data);
+}
+
+static int mlx5_fsm_component_verify(struct mlxfw_dev *mlxfw_dev, u32 fwhandle,
+				     u16 component_index)
+{
+	struct mlx5_mlxfw_dev *mlx5_mlxfw_dev =
+		container_of(mlxfw_dev, struct mlx5_mlxfw_dev, mlxfw_dev);
+	struct mlx5_core_dev *dev = mlx5_mlxfw_dev->mlx5_core_dev;
+
+	return mlx5_reg_mcc_set(dev, MLX5_REG_MCC_INSTRUCTION_VERIFY_COMPONENT,
+				component_index, fwhandle, 0);
+}
+
+static int mlx5_fsm_activate(struct mlxfw_dev *mlxfw_dev, u32 fwhandle)
+{
+	struct mlx5_mlxfw_dev *mlx5_mlxfw_dev =
+		container_of(mlxfw_dev, struct mlx5_mlxfw_dev, mlxfw_dev);
+	struct mlx5_core_dev *dev = mlx5_mlxfw_dev->mlx5_core_dev;
+
+	return mlx5_reg_mcc_set(dev, MLX5_REG_MCC_INSTRUCTION_ACTIVATE,	0,
+				fwhandle, 0);
+}
+
+static int mlx5_fsm_query_state(struct mlxfw_dev *mlxfw_dev, u32 fwhandle,
+				enum mlxfw_fsm_state *fsm_state,
+				enum mlxfw_fsm_state_err *fsm_state_err)
+{
+	struct mlx5_mlxfw_dev *mlx5_mlxfw_dev =
+		container_of(mlxfw_dev, struct mlx5_mlxfw_dev, mlxfw_dev);
+	struct mlx5_core_dev *dev = mlx5_mlxfw_dev->mlx5_core_dev;
+	u8 control_state, error_code;
+	int err;
+
+	err = mlx5_reg_mcc_query(dev, &fwhandle, &error_code, &control_state);
+	if (err)
+		return err;
+
+	*fsm_state = control_state;
+	*fsm_state_err = min_t(enum mlxfw_fsm_state_err, error_code,
+			       MLXFW_FSM_STATE_ERR_MAX);
+	return 0;
+}
+
+static void mlx5_fsm_cancel(struct mlxfw_dev *mlxfw_dev, u32 fwhandle)
+{
+	struct mlx5_mlxfw_dev *mlx5_mlxfw_dev =
+		container_of(mlxfw_dev, struct mlx5_mlxfw_dev, mlxfw_dev);
+	struct mlx5_core_dev *dev = mlx5_mlxfw_dev->mlx5_core_dev;
+
+	mlx5_reg_mcc_set(dev, MLX5_REG_MCC_INSTRUCTION_CANCEL, 0, fwhandle, 0);
+}
+
+static void mlx5_fsm_release(struct mlxfw_dev *mlxfw_dev, u32 fwhandle)
+{
+	struct mlx5_mlxfw_dev *mlx5_mlxfw_dev =
+		container_of(mlxfw_dev, struct mlx5_mlxfw_dev, mlxfw_dev);
+	struct mlx5_core_dev *dev = mlx5_mlxfw_dev->mlx5_core_dev;
+
+	mlx5_reg_mcc_set(dev, MLX5_REG_MCC_INSTRUCTION_RELEASE_UPDATE_HANDLE, 0,
+			 fwhandle, 0);
+}
+
+static int mlx5_fsm_reactivate(struct mlxfw_dev *mlxfw_dev, u8 *status)
+{
+	struct mlx5_mlxfw_dev *mlx5_mlxfw_dev =
+		container_of(mlxfw_dev, struct mlx5_mlxfw_dev, mlxfw_dev);
+	struct mlx5_core_dev *dev = mlx5_mlxfw_dev->mlx5_core_dev;
+	u32 out[MLX5_ST_SZ_DW(mirc_reg)];
+	u32 in[MLX5_ST_SZ_DW(mirc_reg)];
+	unsigned long exp_time;
+	int err;
+
+	exp_time = jiffies + msecs_to_jiffies(mlx5_tout_ms(dev, FSM_REACTIVATE));
+
+	if (!MLX5_CAP_MCAM_REG2(dev, mirc))
+		return -EOPNOTSUPP;
+
+	memset(in, 0, sizeof(in));
+
+	err = mlx5_core_access_reg(dev, in, sizeof(in), out,
+				   sizeof(out), MLX5_REG_MIRC, 0, 1);
+	if (err)
+		return err;
+
+	do {
+		memset(out, 0, sizeof(out));
+		err = mlx5_core_access_reg(dev, in, sizeof(in), out,
+					   sizeof(out), MLX5_REG_MIRC, 0, 0);
+		if (err)
+			return err;
+
+		*status = MLX5_GET(mirc_reg, out, status_code);
+		if (*status != MLXFW_FSM_REACTIVATE_STATUS_BUSY)
+			return 0;
+
+		msleep(20);
+	} while (time_before(jiffies, exp_time));
+
+	return 0;
+}
+
+static const struct mlxfw_dev_ops mlx5_mlxfw_dev_ops = {
+	.component_query	= mlx5_component_query,
+	.fsm_lock		= mlx5_fsm_lock,
+	.fsm_component_update	= mlx5_fsm_component_update,
+	.fsm_block_download	= mlx5_fsm_block_download,
+	.fsm_component_verify	= mlx5_fsm_component_verify,
+	.fsm_activate		= mlx5_fsm_activate,
+	.fsm_reactivate		= mlx5_fsm_reactivate,
+	.fsm_query_state	= mlx5_fsm_query_state,
+	.fsm_cancel		= mlx5_fsm_cancel,
+	.fsm_release		= mlx5_fsm_release
+};
+
+int mlx5_firmware_flash(struct mlx5_core_dev *dev,
+			const struct firmware *firmware,
+			struct netlink_ext_ack *extack)
+{
+	struct mlx5_mlxfw_dev mlx5_mlxfw_dev = {
+		.mlxfw_dev = {
+			.ops = &mlx5_mlxfw_dev_ops,
+			.psid = dev->board_id,
+			.psid_size = strlen(dev->board_id),
+			.devlink = priv_to_devlink(dev),
+		},
+		.mlx5_core_dev = dev
+	};
+
+	if (!MLX5_CAP_GEN(dev, mcam_reg)  ||
+	    !MLX5_CAP_MCAM_REG(dev, mcqi) ||
+	    !MLX5_CAP_MCAM_REG(dev, mcc)  ||
+	    !MLX5_CAP_MCAM_REG(dev, mcda)) {
+		pr_info("%s flashing isn't supported by the running FW\n", __func__);
+		return -EOPNOTSUPP;
+	}
+
+	return mlxfw_firmware_flash(&mlx5_mlxfw_dev.mlxfw_dev,
+				    firmware, extack);
+}
+
+static int mlx5_reg_mcqi_version_query(struct mlx5_core_dev *dev,
+				       u16 component_index, bool read_pending,
+				       u32 *mcqi_version_out)
+{
+	return mlx5_reg_mcqi_query(dev, component_index, read_pending,
+				   MCQI_INFO_TYPE_VERSION,
+				   MLX5_ST_SZ_BYTES(mcqi_version),
+				   mcqi_version_out);
+}
+
+static int mlx5_reg_mcqs_query(struct mlx5_core_dev *dev, u32 *out,
+			       u16 component_index)
+{
+	u8 out_sz = MLX5_ST_SZ_BYTES(mcqs_reg);
+	u32 in[MLX5_ST_SZ_DW(mcqs_reg)] = {};
+	int err;
+
+	memset(out, 0, out_sz);
+
+	MLX5_SET(mcqs_reg, in, component_index, component_index);
+
+	err = mlx5_core_access_reg(dev, in, sizeof(in), out,
+				   out_sz, MLX5_REG_MCQS, 0, 0);
+	return err;
+}
+
+/* scans component index sequentially, to find the boot img index */
+static int mlx5_get_boot_img_component_index(struct mlx5_core_dev *dev)
+{
+	u32 out[MLX5_ST_SZ_DW(mcqs_reg)] = {};
+	u16 identifier, component_idx = 0;
+	bool quit;
+	int err;
+
+	do {
+		err = mlx5_reg_mcqs_query(dev, out, component_idx);
+		if (err)
+			return err;
+
+		identifier = MLX5_GET(mcqs_reg, out, identifier);
+		quit = !!MLX5_GET(mcqs_reg, out, last_index_flag);
+		quit |= identifier == MCQS_IDENTIFIER_BOOT_IMG;
+	} while (!quit && ++component_idx);
+
+	if (identifier != MCQS_IDENTIFIER_BOOT_IMG) {
+		mlx5_core_warn(dev, "mcqs: can't find boot_img component ix, last scanned idx %d\n",
+			       component_idx);
+		return -EOPNOTSUPP;
+	}
+
+	return component_idx;
+}
+
+static int
+mlx5_fw_image_pending(struct mlx5_core_dev *dev,
+		      int component_index,
+		      bool *pending_version_exists)
+{
+	u32 out[MLX5_ST_SZ_DW(mcqs_reg)];
+	u8 component_update_state;
+	int err;
+
+	err = mlx5_reg_mcqs_query(dev, out, component_index);
+	if (err)
+		return err;
+
+	component_update_state = MLX5_GET(mcqs_reg, out, component_update_state);
+
+	if (component_update_state == MCQS_UPDATE_STATE_IDLE) {
+		*pending_version_exists = false;
+	} else if (component_update_state == MCQS_UPDATE_STATE_ACTIVE_PENDING_RESET) {
+		*pending_version_exists = true;
+	} else {
+		mlx5_core_warn(dev,
+			       "mcqs: can't read pending fw version while fw state is %d\n",
+			       component_update_state);
+		return -ENODATA;
+	}
+	return 0;
+}
+
+int mlx5_fw_version_query(struct mlx5_core_dev *dev,
+			  u32 *running_ver, u32 *pending_ver)
+{
+	u32 reg_mcqi_version[MLX5_ST_SZ_DW(mcqi_version)] = {};
+	bool pending_version_exists;
+	int component_index;
+	int err;
+
+	if (!MLX5_CAP_GEN(dev, mcam_reg) || !MLX5_CAP_MCAM_REG(dev, mcqi) ||
+	    !MLX5_CAP_MCAM_REG(dev, mcqs)) {
+		mlx5_core_warn(dev, "fw query isn't supported by the FW\n");
+		return -EOPNOTSUPP;
+	}
+
+	component_index = mlx5_get_boot_img_component_index(dev);
+	if (component_index < 0)
+		return component_index;
+
+	err = mlx5_reg_mcqi_version_query(dev, component_index,
+					  MCQI_FW_RUNNING_VERSION,
+					  reg_mcqi_version);
+	if (err)
+		return err;
+
+	*running_ver = MLX5_GET(mcqi_version, reg_mcqi_version, version);
+
+	err = mlx5_fw_image_pending(dev, component_index, &pending_version_exists);
+	if (err)
+		return err;
+
+	if (!pending_version_exists) {
+		*pending_ver = 0;
+		return 0;
+	}
+
+	err = mlx5_reg_mcqi_version_query(dev, component_index,
+					  MCQI_FW_STORED_VERSION,
+					  reg_mcqi_version);
+	if (err)
+		return err;
+
+	*pending_ver = MLX5_GET(mcqi_version, reg_mcqi_version, version);
+
+	return 0;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fw_reset.c b/drivers/net/ethernet/mellanox/mlx5/core/fw_reset.c
new file mode 100644
index 000000000..dec1492da
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fw_reset.c
@@ -0,0 +1,546 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2020, Mellanox Technologies inc.  All rights reserved. */
+
+#include "fw_reset.h"
+#include "diag/fw_tracer.h"
+#include "lib/tout.h"
+
+enum {
+	MLX5_FW_RESET_FLAGS_RESET_REQUESTED,
+	MLX5_FW_RESET_FLAGS_NACK_RESET_REQUEST,
+	MLX5_FW_RESET_FLAGS_PENDING_COMP,
+	MLX5_FW_RESET_FLAGS_DROP_NEW_REQUESTS,
+	MLX5_FW_RESET_FLAGS_RELOAD_REQUIRED
+};
+
+struct mlx5_fw_reset {
+	struct mlx5_core_dev *dev;
+	struct mlx5_nb nb;
+	struct workqueue_struct *wq;
+	struct work_struct fw_live_patch_work;
+	struct work_struct reset_request_work;
+	struct work_struct reset_reload_work;
+	struct work_struct reset_now_work;
+	struct work_struct reset_abort_work;
+	unsigned long reset_flags;
+	struct timer_list timer;
+	struct completion done;
+	int ret;
+};
+
+void mlx5_fw_reset_enable_remote_dev_reset_set(struct mlx5_core_dev *dev, bool enable)
+{
+	struct mlx5_fw_reset *fw_reset = dev->priv.fw_reset;
+
+	if (enable)
+		clear_bit(MLX5_FW_RESET_FLAGS_NACK_RESET_REQUEST, &fw_reset->reset_flags);
+	else
+		set_bit(MLX5_FW_RESET_FLAGS_NACK_RESET_REQUEST, &fw_reset->reset_flags);
+}
+
+bool mlx5_fw_reset_enable_remote_dev_reset_get(struct mlx5_core_dev *dev)
+{
+	struct mlx5_fw_reset *fw_reset = dev->priv.fw_reset;
+
+	return !test_bit(MLX5_FW_RESET_FLAGS_NACK_RESET_REQUEST, &fw_reset->reset_flags);
+}
+
+static int mlx5_reg_mfrl_set(struct mlx5_core_dev *dev, u8 reset_level,
+			     u8 reset_type_sel, u8 sync_resp, bool sync_start)
+{
+	u32 out[MLX5_ST_SZ_DW(mfrl_reg)] = {};
+	u32 in[MLX5_ST_SZ_DW(mfrl_reg)] = {};
+
+	MLX5_SET(mfrl_reg, in, reset_level, reset_level);
+	MLX5_SET(mfrl_reg, in, rst_type_sel, reset_type_sel);
+	MLX5_SET(mfrl_reg, in, pci_sync_for_fw_update_resp, sync_resp);
+	MLX5_SET(mfrl_reg, in, pci_sync_for_fw_update_start, sync_start);
+
+	return mlx5_core_access_reg(dev, in, sizeof(in), out, sizeof(out), MLX5_REG_MFRL, 0, 1);
+}
+
+static int mlx5_reg_mfrl_query(struct mlx5_core_dev *dev, u8 *reset_level,
+			       u8 *reset_type, u8 *reset_state)
+{
+	u32 out[MLX5_ST_SZ_DW(mfrl_reg)] = {};
+	u32 in[MLX5_ST_SZ_DW(mfrl_reg)] = {};
+	int err;
+
+	err = mlx5_core_access_reg(dev, in, sizeof(in), out, sizeof(out), MLX5_REG_MFRL, 0, 0);
+	if (err)
+		return err;
+
+	if (reset_level)
+		*reset_level = MLX5_GET(mfrl_reg, out, reset_level);
+	if (reset_type)
+		*reset_type = MLX5_GET(mfrl_reg, out, reset_type);
+	if (reset_state)
+		*reset_state = MLX5_GET(mfrl_reg, out, reset_state);
+
+	return 0;
+}
+
+int mlx5_fw_reset_query(struct mlx5_core_dev *dev, u8 *reset_level, u8 *reset_type)
+{
+	return mlx5_reg_mfrl_query(dev, reset_level, reset_type, NULL);
+}
+
+static int mlx5_fw_reset_get_reset_state_err(struct mlx5_core_dev *dev,
+					     struct netlink_ext_ack *extack)
+{
+	u8 reset_state;
+
+	if (mlx5_reg_mfrl_query(dev, NULL, NULL, &reset_state))
+		goto out;
+
+	switch (reset_state) {
+	case MLX5_MFRL_REG_RESET_STATE_IN_NEGOTIATION:
+	case MLX5_MFRL_REG_RESET_STATE_RESET_IN_PROGRESS:
+		NL_SET_ERR_MSG_MOD(extack, "Sync reset was already triggered");
+		return -EBUSY;
+	case MLX5_MFRL_REG_RESET_STATE_TIMEOUT:
+		NL_SET_ERR_MSG_MOD(extack, "Sync reset got timeout");
+		return -ETIMEDOUT;
+	case MLX5_MFRL_REG_RESET_STATE_NACK:
+		NL_SET_ERR_MSG_MOD(extack, "One of the hosts disabled reset");
+		return -EPERM;
+	}
+
+out:
+	NL_SET_ERR_MSG_MOD(extack, "Sync reset failed");
+	return -EIO;
+}
+
+int mlx5_fw_reset_set_reset_sync(struct mlx5_core_dev *dev, u8 reset_type_sel,
+				 struct netlink_ext_ack *extack)
+{
+	struct mlx5_fw_reset *fw_reset = dev->priv.fw_reset;
+	u32 out[MLX5_ST_SZ_DW(mfrl_reg)] = {};
+	u32 in[MLX5_ST_SZ_DW(mfrl_reg)] = {};
+	int err;
+
+	set_bit(MLX5_FW_RESET_FLAGS_PENDING_COMP, &fw_reset->reset_flags);
+
+	MLX5_SET(mfrl_reg, in, reset_level, MLX5_MFRL_REG_RESET_LEVEL3);
+	MLX5_SET(mfrl_reg, in, rst_type_sel, reset_type_sel);
+	MLX5_SET(mfrl_reg, in, pci_sync_for_fw_update_start, 1);
+	err = mlx5_access_reg(dev, in, sizeof(in), out, sizeof(out),
+			      MLX5_REG_MFRL, 0, 1, false);
+	if (!err)
+		return 0;
+
+	clear_bit(MLX5_FW_RESET_FLAGS_PENDING_COMP, &fw_reset->reset_flags);
+	if (err == -EREMOTEIO && MLX5_CAP_MCAM_FEATURE(dev, reset_state))
+		return mlx5_fw_reset_get_reset_state_err(dev, extack);
+
+	NL_SET_ERR_MSG_MOD(extack, "Sync reset command failed");
+	return mlx5_cmd_check(dev, err, in, out);
+}
+
+int mlx5_fw_reset_set_live_patch(struct mlx5_core_dev *dev)
+{
+	return mlx5_reg_mfrl_set(dev, MLX5_MFRL_REG_RESET_LEVEL0, 0, 0, false);
+}
+
+static void mlx5_fw_reset_complete_reload(struct mlx5_core_dev *dev)
+{
+	struct mlx5_fw_reset *fw_reset = dev->priv.fw_reset;
+
+	/* if this is the driver that initiated the fw reset, devlink completed the reload */
+	if (test_bit(MLX5_FW_RESET_FLAGS_PENDING_COMP, &fw_reset->reset_flags)) {
+		complete(&fw_reset->done);
+	} else {
+		mlx5_unload_one(dev, false);
+		if (mlx5_health_wait_pci_up(dev))
+			mlx5_core_err(dev, "reset reload flow aborted, PCI reads still not working\n");
+		else
+			mlx5_load_one(dev, true);
+		devlink_remote_reload_actions_performed(priv_to_devlink(dev), 0,
+							BIT(DEVLINK_RELOAD_ACTION_DRIVER_REINIT) |
+							BIT(DEVLINK_RELOAD_ACTION_FW_ACTIVATE));
+	}
+}
+
+static void mlx5_stop_sync_reset_poll(struct mlx5_core_dev *dev)
+{
+	struct mlx5_fw_reset *fw_reset = dev->priv.fw_reset;
+
+	del_timer_sync(&fw_reset->timer);
+}
+
+static int mlx5_sync_reset_clear_reset_requested(struct mlx5_core_dev *dev, bool poll_health)
+{
+	struct mlx5_fw_reset *fw_reset = dev->priv.fw_reset;
+
+	if (!test_and_clear_bit(MLX5_FW_RESET_FLAGS_RESET_REQUESTED, &fw_reset->reset_flags)) {
+		mlx5_core_warn(dev, "Reset request was already cleared\n");
+		return -EALREADY;
+	}
+
+	mlx5_stop_sync_reset_poll(dev);
+	if (poll_health)
+		mlx5_start_health_poll(dev);
+	return 0;
+}
+
+static void mlx5_sync_reset_reload_work(struct work_struct *work)
+{
+	struct mlx5_fw_reset *fw_reset = container_of(work, struct mlx5_fw_reset,
+						      reset_reload_work);
+	struct mlx5_core_dev *dev = fw_reset->dev;
+
+	mlx5_sync_reset_clear_reset_requested(dev, false);
+	mlx5_enter_error_state(dev, true);
+	mlx5_fw_reset_complete_reload(dev);
+}
+
+#define MLX5_RESET_POLL_INTERVAL	(HZ / 10)
+static void poll_sync_reset(struct timer_list *t)
+{
+	struct mlx5_fw_reset *fw_reset = from_timer(fw_reset, t, timer);
+	struct mlx5_core_dev *dev = fw_reset->dev;
+	u32 fatal_error;
+
+	if (!test_bit(MLX5_FW_RESET_FLAGS_RESET_REQUESTED, &fw_reset->reset_flags))
+		return;
+
+	fatal_error = mlx5_health_check_fatal_sensors(dev);
+
+	if (fatal_error) {
+		mlx5_core_warn(dev, "Got Device Reset\n");
+		if (!test_bit(MLX5_FW_RESET_FLAGS_DROP_NEW_REQUESTS, &fw_reset->reset_flags))
+			queue_work(fw_reset->wq, &fw_reset->reset_reload_work);
+		else
+			mlx5_core_err(dev, "Device is being removed, Drop new reset work\n");
+		return;
+	}
+
+	mod_timer(&fw_reset->timer, round_jiffies(jiffies + MLX5_RESET_POLL_INTERVAL));
+}
+
+static void mlx5_start_sync_reset_poll(struct mlx5_core_dev *dev)
+{
+	struct mlx5_fw_reset *fw_reset = dev->priv.fw_reset;
+
+	timer_setup(&fw_reset->timer, poll_sync_reset, 0);
+	fw_reset->timer.expires = round_jiffies(jiffies + MLX5_RESET_POLL_INTERVAL);
+	add_timer(&fw_reset->timer);
+}
+
+static int mlx5_fw_reset_set_reset_sync_ack(struct mlx5_core_dev *dev)
+{
+	return mlx5_reg_mfrl_set(dev, MLX5_MFRL_REG_RESET_LEVEL3, 0, 1, false);
+}
+
+static int mlx5_fw_reset_set_reset_sync_nack(struct mlx5_core_dev *dev)
+{
+	return mlx5_reg_mfrl_set(dev, MLX5_MFRL_REG_RESET_LEVEL3, 0, 2, false);
+}
+
+static int mlx5_sync_reset_set_reset_requested(struct mlx5_core_dev *dev)
+{
+	struct mlx5_fw_reset *fw_reset = dev->priv.fw_reset;
+
+	if (test_and_set_bit(MLX5_FW_RESET_FLAGS_RESET_REQUESTED, &fw_reset->reset_flags)) {
+		mlx5_core_warn(dev, "Reset request was already set\n");
+		return -EALREADY;
+	}
+	mlx5_stop_health_poll(dev, true);
+	mlx5_start_sync_reset_poll(dev);
+	return 0;
+}
+
+static void mlx5_fw_live_patch_event(struct work_struct *work)
+{
+	struct mlx5_fw_reset *fw_reset = container_of(work, struct mlx5_fw_reset,
+						      fw_live_patch_work);
+	struct mlx5_core_dev *dev = fw_reset->dev;
+
+	mlx5_core_info(dev, "Live patch updated firmware version: %d.%d.%d\n", fw_rev_maj(dev),
+		       fw_rev_min(dev), fw_rev_sub(dev));
+
+	if (mlx5_fw_tracer_reload(dev->tracer))
+		mlx5_core_err(dev, "Failed to reload FW tracer\n");
+}
+
+static void mlx5_sync_reset_request_event(struct work_struct *work)
+{
+	struct mlx5_fw_reset *fw_reset = container_of(work, struct mlx5_fw_reset,
+						      reset_request_work);
+	struct mlx5_core_dev *dev = fw_reset->dev;
+	int err;
+
+	if (test_bit(MLX5_FW_RESET_FLAGS_NACK_RESET_REQUEST, &fw_reset->reset_flags)) {
+		err = mlx5_fw_reset_set_reset_sync_nack(dev);
+		mlx5_core_warn(dev, "PCI Sync FW Update Reset Nack %s",
+			       err ? "Failed" : "Sent");
+		return;
+	}
+	if (mlx5_sync_reset_set_reset_requested(dev))
+		return;
+
+	err = mlx5_fw_reset_set_reset_sync_ack(dev);
+	if (err)
+		mlx5_core_warn(dev, "PCI Sync FW Update Reset Ack Failed. Error code: %d\n", err);
+	else
+		mlx5_core_warn(dev, "PCI Sync FW Update Reset Ack. Device reset is expected.\n");
+}
+
+static int mlx5_pci_link_toggle(struct mlx5_core_dev *dev)
+{
+	struct pci_bus *bridge_bus = dev->pdev->bus;
+	struct pci_dev *bridge = bridge_bus->self;
+	u16 reg16, dev_id, sdev_id;
+	unsigned long timeout;
+	struct pci_dev *sdev;
+	int cap, err;
+	u32 reg32;
+
+	/* Check that all functions under the pci bridge are PFs of
+	 * this device otherwise fail this function.
+	 */
+	err = pci_read_config_word(dev->pdev, PCI_DEVICE_ID, &dev_id);
+	if (err)
+		return err;
+	list_for_each_entry(sdev, &bridge_bus->devices, bus_list) {
+		err = pci_read_config_word(sdev, PCI_DEVICE_ID, &sdev_id);
+		if (err)
+			return err;
+		if (sdev_id != dev_id)
+			return -EPERM;
+	}
+
+	cap = pci_find_capability(bridge, PCI_CAP_ID_EXP);
+	if (!cap)
+		return -EOPNOTSUPP;
+
+	list_for_each_entry(sdev, &bridge_bus->devices, bus_list) {
+		pci_save_state(sdev);
+		pci_cfg_access_lock(sdev);
+	}
+	/* PCI link toggle */
+	err = pcie_capability_set_word(bridge, PCI_EXP_LNKCTL, PCI_EXP_LNKCTL_LD);
+	if (err)
+		return err;
+	msleep(500);
+	err = pcie_capability_clear_word(bridge, PCI_EXP_LNKCTL, PCI_EXP_LNKCTL_LD);
+	if (err)
+		return err;
+
+	/* Check link */
+	err = pci_read_config_dword(bridge, cap + PCI_EXP_LNKCAP, &reg32);
+	if (err)
+		return err;
+	if (!(reg32 & PCI_EXP_LNKCAP_DLLLARC)) {
+		mlx5_core_warn(dev, "No PCI link reporting capability (0x%08x)\n", reg32);
+		msleep(1000);
+		goto restore;
+	}
+
+	timeout = jiffies + msecs_to_jiffies(mlx5_tout_ms(dev, PCI_TOGGLE));
+	do {
+		err = pci_read_config_word(bridge, cap + PCI_EXP_LNKSTA, &reg16);
+		if (err)
+			return err;
+		if (reg16 & PCI_EXP_LNKSTA_DLLLA)
+			break;
+		msleep(20);
+	} while (!time_after(jiffies, timeout));
+
+	if (reg16 & PCI_EXP_LNKSTA_DLLLA) {
+		mlx5_core_info(dev, "PCI Link up\n");
+	} else {
+		mlx5_core_err(dev, "PCI link not ready (0x%04x) after %llu ms\n",
+			      reg16, mlx5_tout_ms(dev, PCI_TOGGLE));
+		err = -ETIMEDOUT;
+	}
+
+	do {
+		err = pci_read_config_word(dev->pdev, PCI_DEVICE_ID, &reg16);
+		if (err)
+			return err;
+		if (reg16 == dev_id)
+			break;
+		msleep(20);
+	} while (!time_after(jiffies, timeout));
+
+	if (reg16 == dev_id) {
+		mlx5_core_info(dev, "Firmware responds to PCI config cycles again\n");
+	} else {
+		mlx5_core_err(dev, "Firmware is not responsive (0x%04x) after %llu ms\n",
+			      reg16, mlx5_tout_ms(dev, PCI_TOGGLE));
+		err = -ETIMEDOUT;
+	}
+
+restore:
+	list_for_each_entry(sdev, &bridge_bus->devices, bus_list) {
+		pci_cfg_access_unlock(sdev);
+		pci_restore_state(sdev);
+	}
+
+	return err;
+}
+
+static void mlx5_sync_reset_now_event(struct work_struct *work)
+{
+	struct mlx5_fw_reset *fw_reset = container_of(work, struct mlx5_fw_reset,
+						      reset_now_work);
+	struct mlx5_core_dev *dev = fw_reset->dev;
+	int err;
+
+	if (mlx5_sync_reset_clear_reset_requested(dev, false))
+		return;
+
+	mlx5_core_warn(dev, "Sync Reset now. Device is going to reset.\n");
+
+	err = mlx5_cmd_fast_teardown_hca(dev);
+	if (err) {
+		mlx5_core_warn(dev, "Fast teardown failed, no reset done, err %d\n", err);
+		goto done;
+	}
+
+	err = mlx5_pci_link_toggle(dev);
+	if (err) {
+		mlx5_core_warn(dev, "mlx5_pci_link_toggle failed, no reset done, err %d\n", err);
+		set_bit(MLX5_FW_RESET_FLAGS_RELOAD_REQUIRED, &fw_reset->reset_flags);
+	}
+
+	mlx5_enter_error_state(dev, true);
+done:
+	fw_reset->ret = err;
+	mlx5_fw_reset_complete_reload(dev);
+}
+
+static void mlx5_sync_reset_abort_event(struct work_struct *work)
+{
+	struct mlx5_fw_reset *fw_reset = container_of(work, struct mlx5_fw_reset,
+						      reset_abort_work);
+	struct mlx5_core_dev *dev = fw_reset->dev;
+
+	if (mlx5_sync_reset_clear_reset_requested(dev, true))
+		return;
+	mlx5_core_warn(dev, "PCI Sync FW Update Reset Aborted.\n");
+}
+
+static void mlx5_sync_reset_events_handle(struct mlx5_fw_reset *fw_reset, struct mlx5_eqe *eqe)
+{
+	struct mlx5_eqe_sync_fw_update *sync_fw_update_eqe;
+	u8 sync_event_rst_type;
+
+	sync_fw_update_eqe = &eqe->data.sync_fw_update;
+	sync_event_rst_type = sync_fw_update_eqe->sync_rst_state & SYNC_RST_STATE_MASK;
+	switch (sync_event_rst_type) {
+	case MLX5_SYNC_RST_STATE_RESET_REQUEST:
+		queue_work(fw_reset->wq, &fw_reset->reset_request_work);
+		break;
+	case MLX5_SYNC_RST_STATE_RESET_NOW:
+		queue_work(fw_reset->wq, &fw_reset->reset_now_work);
+		break;
+	case MLX5_SYNC_RST_STATE_RESET_ABORT:
+		queue_work(fw_reset->wq, &fw_reset->reset_abort_work);
+		break;
+	}
+}
+
+static int fw_reset_event_notifier(struct notifier_block *nb, unsigned long action, void *data)
+{
+	struct mlx5_fw_reset *fw_reset = mlx5_nb_cof(nb, struct mlx5_fw_reset, nb);
+	struct mlx5_eqe *eqe = data;
+
+	if (test_bit(MLX5_FW_RESET_FLAGS_DROP_NEW_REQUESTS, &fw_reset->reset_flags))
+		return NOTIFY_DONE;
+
+	switch (eqe->sub_type) {
+	case MLX5_GENERAL_SUBTYPE_FW_LIVE_PATCH_EVENT:
+		queue_work(fw_reset->wq, &fw_reset->fw_live_patch_work);
+		break;
+	case MLX5_GENERAL_SUBTYPE_PCI_SYNC_FOR_FW_UPDATE_EVENT:
+		mlx5_sync_reset_events_handle(fw_reset, eqe);
+		break;
+	default:
+		return NOTIFY_DONE;
+	}
+
+	return NOTIFY_OK;
+}
+
+int mlx5_fw_reset_wait_reset_done(struct mlx5_core_dev *dev)
+{
+	unsigned long pci_sync_update_timeout = mlx5_tout_ms(dev, PCI_SYNC_UPDATE);
+	unsigned long timeout = msecs_to_jiffies(pci_sync_update_timeout);
+	struct mlx5_fw_reset *fw_reset = dev->priv.fw_reset;
+	int err;
+
+	if (!wait_for_completion_timeout(&fw_reset->done, timeout)) {
+		mlx5_core_warn(dev, "FW sync reset timeout after %lu seconds\n",
+			       pci_sync_update_timeout / 1000);
+		err = -ETIMEDOUT;
+		goto out;
+	}
+	err = fw_reset->ret;
+	if (test_and_clear_bit(MLX5_FW_RESET_FLAGS_RELOAD_REQUIRED, &fw_reset->reset_flags)) {
+		mlx5_unload_one_devl_locked(dev, false);
+		mlx5_load_one_devl_locked(dev, true);
+	}
+out:
+	clear_bit(MLX5_FW_RESET_FLAGS_PENDING_COMP, &fw_reset->reset_flags);
+	return err;
+}
+
+void mlx5_fw_reset_events_start(struct mlx5_core_dev *dev)
+{
+	struct mlx5_fw_reset *fw_reset = dev->priv.fw_reset;
+
+	MLX5_NB_INIT(&fw_reset->nb, fw_reset_event_notifier, GENERAL_EVENT);
+	mlx5_eq_notifier_register(dev, &fw_reset->nb);
+}
+
+void mlx5_fw_reset_events_stop(struct mlx5_core_dev *dev)
+{
+	mlx5_eq_notifier_unregister(dev, &dev->priv.fw_reset->nb);
+}
+
+void mlx5_drain_fw_reset(struct mlx5_core_dev *dev)
+{
+	struct mlx5_fw_reset *fw_reset = dev->priv.fw_reset;
+
+	set_bit(MLX5_FW_RESET_FLAGS_DROP_NEW_REQUESTS, &fw_reset->reset_flags);
+	cancel_work_sync(&fw_reset->fw_live_patch_work);
+	cancel_work_sync(&fw_reset->reset_request_work);
+	cancel_work_sync(&fw_reset->reset_reload_work);
+	cancel_work_sync(&fw_reset->reset_now_work);
+	cancel_work_sync(&fw_reset->reset_abort_work);
+}
+
+int mlx5_fw_reset_init(struct mlx5_core_dev *dev)
+{
+	struct mlx5_fw_reset *fw_reset = kzalloc(sizeof(*fw_reset), GFP_KERNEL);
+
+	if (!fw_reset)
+		return -ENOMEM;
+	fw_reset->wq = create_singlethread_workqueue("mlx5_fw_reset_events");
+	if (!fw_reset->wq) {
+		kfree(fw_reset);
+		return -ENOMEM;
+	}
+
+	fw_reset->dev = dev;
+	dev->priv.fw_reset = fw_reset;
+
+	INIT_WORK(&fw_reset->fw_live_patch_work, mlx5_fw_live_patch_event);
+	INIT_WORK(&fw_reset->reset_request_work, mlx5_sync_reset_request_event);
+	INIT_WORK(&fw_reset->reset_reload_work, mlx5_sync_reset_reload_work);
+	INIT_WORK(&fw_reset->reset_now_work, mlx5_sync_reset_now_event);
+	INIT_WORK(&fw_reset->reset_abort_work, mlx5_sync_reset_abort_event);
+
+	init_completion(&fw_reset->done);
+	return 0;
+}
+
+void mlx5_fw_reset_cleanup(struct mlx5_core_dev *dev)
+{
+	struct mlx5_fw_reset *fw_reset = dev->priv.fw_reset;
+
+	destroy_workqueue(fw_reset->wq);
+	kfree(dev->priv.fw_reset);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fw_reset.h b/drivers/net/ethernet/mellanox/mlx5/core/fw_reset.h
new file mode 100644
index 000000000..dc141c7e6
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fw_reset.h
@@ -0,0 +1,23 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2020, Mellanox Technologies inc.  All rights reserved. */
+
+#ifndef __MLX5_FW_RESET_H
+#define __MLX5_FW_RESET_H
+
+#include "mlx5_core.h"
+
+void mlx5_fw_reset_enable_remote_dev_reset_set(struct mlx5_core_dev *dev, bool enable);
+bool mlx5_fw_reset_enable_remote_dev_reset_get(struct mlx5_core_dev *dev);
+int mlx5_fw_reset_query(struct mlx5_core_dev *dev, u8 *reset_level, u8 *reset_type);
+int mlx5_fw_reset_set_reset_sync(struct mlx5_core_dev *dev, u8 reset_type_sel,
+				 struct netlink_ext_ack *extack);
+int mlx5_fw_reset_set_live_patch(struct mlx5_core_dev *dev);
+
+int mlx5_fw_reset_wait_reset_done(struct mlx5_core_dev *dev);
+void mlx5_fw_reset_events_start(struct mlx5_core_dev *dev);
+void mlx5_fw_reset_events_stop(struct mlx5_core_dev *dev);
+void mlx5_drain_fw_reset(struct mlx5_core_dev *dev);
+int mlx5_fw_reset_init(struct mlx5_core_dev *dev);
+void mlx5_fw_reset_cleanup(struct mlx5_core_dev *dev);
+
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/health.c b/drivers/net/ethernet/mellanox/mlx5/core/health.c
new file mode 100644
index 000000000..e42e4ac23
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/health.c
@@ -0,0 +1,941 @@
+/*
+ * Copyright (c) 2013-2015, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/kernel.h>
+#include <linux/random.h>
+#include <linux/vmalloc.h>
+#include <linux/hardirq.h>
+#include <linux/mlx5/driver.h>
+#include <linux/kern_levels.h>
+#include "mlx5_core.h"
+#include "lib/eq.h"
+#include "lib/mlx5.h"
+#include "lib/pci_vsc.h"
+#include "lib/tout.h"
+#include "diag/fw_tracer.h"
+
+enum {
+	MAX_MISSES			= 3,
+};
+
+enum {
+	MLX5_HEALTH_SYNDR_FW_ERR		= 0x1,
+	MLX5_HEALTH_SYNDR_IRISC_ERR		= 0x7,
+	MLX5_HEALTH_SYNDR_HW_UNRECOVERABLE_ERR	= 0x8,
+	MLX5_HEALTH_SYNDR_CRC_ERR		= 0x9,
+	MLX5_HEALTH_SYNDR_FETCH_PCI_ERR		= 0xa,
+	MLX5_HEALTH_SYNDR_HW_FTL_ERR		= 0xb,
+	MLX5_HEALTH_SYNDR_ASYNC_EQ_OVERRUN_ERR	= 0xc,
+	MLX5_HEALTH_SYNDR_EQ_ERR		= 0xd,
+	MLX5_HEALTH_SYNDR_EQ_INV		= 0xe,
+	MLX5_HEALTH_SYNDR_FFSER_ERR		= 0xf,
+	MLX5_HEALTH_SYNDR_HIGH_TEMP		= 0x10
+};
+
+enum {
+	MLX5_DROP_NEW_HEALTH_WORK,
+};
+
+enum  {
+	MLX5_SENSOR_NO_ERR		= 0,
+	MLX5_SENSOR_PCI_COMM_ERR	= 1,
+	MLX5_SENSOR_PCI_ERR		= 2,
+	MLX5_SENSOR_NIC_DISABLED	= 3,
+	MLX5_SENSOR_NIC_SW_RESET	= 4,
+	MLX5_SENSOR_FW_SYND_RFR		= 5,
+};
+
+enum {
+	MLX5_SEVERITY_MASK		= 0x7,
+	MLX5_SEVERITY_VALID_MASK	= 0x8,
+};
+
+u8 mlx5_get_nic_state(struct mlx5_core_dev *dev)
+{
+	return (ioread32be(&dev->iseg->cmdq_addr_l_sz) >> 8) & 7;
+}
+
+void mlx5_set_nic_state(struct mlx5_core_dev *dev, u8 state)
+{
+	u32 cur_cmdq_addr_l_sz;
+
+	cur_cmdq_addr_l_sz = ioread32be(&dev->iseg->cmdq_addr_l_sz);
+	iowrite32be((cur_cmdq_addr_l_sz & 0xFFFFF000) |
+		    state << MLX5_NIC_IFC_OFFSET,
+		    &dev->iseg->cmdq_addr_l_sz);
+}
+
+static bool sensor_pci_not_working(struct mlx5_core_dev *dev)
+{
+	struct mlx5_core_health *health = &dev->priv.health;
+	struct health_buffer __iomem *h = health->health;
+
+	/* Offline PCI reads return 0xffffffff */
+	return (ioread32be(&h->fw_ver) == 0xffffffff);
+}
+
+static int mlx5_health_get_rfr(u8 rfr_severity)
+{
+	return rfr_severity >> MLX5_RFR_BIT_OFFSET;
+}
+
+static bool sensor_fw_synd_rfr(struct mlx5_core_dev *dev)
+{
+	struct mlx5_core_health *health = &dev->priv.health;
+	struct health_buffer __iomem *h = health->health;
+	u8 synd = ioread8(&h->synd);
+	u8 rfr;
+
+	rfr = mlx5_health_get_rfr(ioread8(&h->rfr_severity));
+
+	if (rfr && synd)
+		mlx5_core_dbg(dev, "FW requests reset, synd: %d\n", synd);
+	return rfr && synd;
+}
+
+u32 mlx5_health_check_fatal_sensors(struct mlx5_core_dev *dev)
+{
+	if (sensor_pci_not_working(dev))
+		return MLX5_SENSOR_PCI_COMM_ERR;
+	if (pci_channel_offline(dev->pdev))
+		return MLX5_SENSOR_PCI_ERR;
+	if (mlx5_get_nic_state(dev) == MLX5_NIC_IFC_DISABLED)
+		return MLX5_SENSOR_NIC_DISABLED;
+	if (mlx5_get_nic_state(dev) == MLX5_NIC_IFC_SW_RESET)
+		return MLX5_SENSOR_NIC_SW_RESET;
+	if (sensor_fw_synd_rfr(dev))
+		return MLX5_SENSOR_FW_SYND_RFR;
+
+	return MLX5_SENSOR_NO_ERR;
+}
+
+static int lock_sem_sw_reset(struct mlx5_core_dev *dev, bool lock)
+{
+	enum mlx5_vsc_state state;
+	int ret;
+
+	if (!mlx5_core_is_pf(dev))
+		return -EBUSY;
+
+	/* Try to lock GW access, this stage doesn't return
+	 * EBUSY because locked GW does not mean that other PF
+	 * already started the reset.
+	 */
+	ret = mlx5_vsc_gw_lock(dev);
+	if (ret == -EBUSY)
+		return -EINVAL;
+	if (ret)
+		return ret;
+
+	state = lock ? MLX5_VSC_LOCK : MLX5_VSC_UNLOCK;
+	/* At this stage, if the return status == EBUSY, then we know
+	 * for sure that another PF started the reset, so don't allow
+	 * another reset.
+	 */
+	ret = mlx5_vsc_sem_set_space(dev, MLX5_SEMAPHORE_SW_RESET, state);
+	if (ret)
+		mlx5_core_warn(dev, "Failed to lock SW reset semaphore\n");
+
+	/* Unlock GW access */
+	mlx5_vsc_gw_unlock(dev);
+
+	return ret;
+}
+
+static bool reset_fw_if_needed(struct mlx5_core_dev *dev)
+{
+	bool supported = (ioread32be(&dev->iseg->initializing) >>
+			  MLX5_FW_RESET_SUPPORTED_OFFSET) & 1;
+	u32 fatal_error;
+
+	if (!supported)
+		return false;
+
+	/* The reset only needs to be issued by one PF. The health buffer is
+	 * shared between all functions, and will be cleared during a reset.
+	 * Check again to avoid a redundant 2nd reset. If the fatal errors was
+	 * PCI related a reset won't help.
+	 */
+	fatal_error = mlx5_health_check_fatal_sensors(dev);
+	if (fatal_error == MLX5_SENSOR_PCI_COMM_ERR ||
+	    fatal_error == MLX5_SENSOR_NIC_DISABLED ||
+	    fatal_error == MLX5_SENSOR_NIC_SW_RESET) {
+		mlx5_core_warn(dev, "Not issuing FW reset. Either it's already done or won't help.");
+		return false;
+	}
+
+	mlx5_core_warn(dev, "Issuing FW Reset\n");
+	/* Write the NIC interface field to initiate the reset, the command
+	 * interface address also resides here, don't overwrite it.
+	 */
+	mlx5_set_nic_state(dev, MLX5_NIC_IFC_SW_RESET);
+
+	return true;
+}
+
+static void enter_error_state(struct mlx5_core_dev *dev, bool force)
+{
+	if (mlx5_health_check_fatal_sensors(dev) || force) { /* protected state setting */
+		dev->state = MLX5_DEVICE_STATE_INTERNAL_ERROR;
+		mlx5_cmd_flush(dev);
+	}
+
+	mlx5_notifier_call_chain(dev->priv.events, MLX5_DEV_EVENT_SYS_ERROR, (void *)1);
+}
+
+void mlx5_enter_error_state(struct mlx5_core_dev *dev, bool force)
+{
+	bool err_detected = false;
+
+	/* Mark the device as fatal in order to abort FW commands */
+	if ((mlx5_health_check_fatal_sensors(dev) || force) &&
+	    dev->state == MLX5_DEVICE_STATE_UP) {
+		dev->state = MLX5_DEVICE_STATE_INTERNAL_ERROR;
+		err_detected = true;
+	}
+	mutex_lock(&dev->intf_state_mutex);
+	if (!err_detected && dev->state == MLX5_DEVICE_STATE_INTERNAL_ERROR)
+		goto unlock;/* a previous error is still being handled */
+
+	enter_error_state(dev, force);
+unlock:
+	mutex_unlock(&dev->intf_state_mutex);
+}
+
+void mlx5_error_sw_reset(struct mlx5_core_dev *dev)
+{
+	unsigned long end, delay_ms = mlx5_tout_ms(dev, PCI_TOGGLE);
+	int lock = -EBUSY;
+
+	mutex_lock(&dev->intf_state_mutex);
+	if (dev->state != MLX5_DEVICE_STATE_INTERNAL_ERROR)
+		goto unlock;
+
+	mlx5_core_err(dev, "start\n");
+
+	if (mlx5_health_check_fatal_sensors(dev) == MLX5_SENSOR_FW_SYND_RFR) {
+		/* Get cr-dump and reset FW semaphore */
+		lock = lock_sem_sw_reset(dev, true);
+
+		if (lock == -EBUSY) {
+			delay_ms = mlx5_tout_ms(dev, FULL_CRDUMP);
+			goto recover_from_sw_reset;
+		}
+		/* Execute SW reset */
+		reset_fw_if_needed(dev);
+	}
+
+recover_from_sw_reset:
+	/* Recover from SW reset */
+	end = jiffies + msecs_to_jiffies(delay_ms);
+	do {
+		if (mlx5_get_nic_state(dev) == MLX5_NIC_IFC_DISABLED)
+			break;
+
+		msleep(20);
+	} while (!time_after(jiffies, end));
+
+	if (mlx5_get_nic_state(dev) != MLX5_NIC_IFC_DISABLED) {
+		dev_err(&dev->pdev->dev, "NIC IFC still %d after %lums.\n",
+			mlx5_get_nic_state(dev), delay_ms);
+	}
+
+	/* Release FW semaphore if you are the lock owner */
+	if (!lock)
+		lock_sem_sw_reset(dev, false);
+
+	mlx5_core_err(dev, "end\n");
+
+unlock:
+	mutex_unlock(&dev->intf_state_mutex);
+}
+
+static void mlx5_handle_bad_state(struct mlx5_core_dev *dev)
+{
+	u8 nic_interface = mlx5_get_nic_state(dev);
+
+	switch (nic_interface) {
+	case MLX5_NIC_IFC_FULL:
+		mlx5_core_warn(dev, "Expected to see disabled NIC but it is full driver\n");
+		break;
+
+	case MLX5_NIC_IFC_DISABLED:
+		mlx5_core_warn(dev, "starting teardown\n");
+		break;
+
+	case MLX5_NIC_IFC_NO_DRAM_NIC:
+		mlx5_core_warn(dev, "Expected to see disabled NIC but it is no dram nic\n");
+		break;
+
+	case MLX5_NIC_IFC_SW_RESET:
+		/* The IFC mode field is 3 bits, so it will read 0x7 in 2 cases:
+		 * 1. PCI has been disabled (ie. PCI-AER, PF driver unloaded
+		 *    and this is a VF), this is not recoverable by SW reset.
+		 *    Logging of this is handled elsewhere.
+		 * 2. FW reset has been issued by another function, driver can
+		 *    be reloaded to recover after the mode switches to
+		 *    MLX5_NIC_IFC_DISABLED.
+		 */
+		if (dev->priv.health.fatal_error != MLX5_SENSOR_PCI_COMM_ERR)
+			mlx5_core_warn(dev, "NIC SW reset in progress\n");
+		break;
+
+	default:
+		mlx5_core_warn(dev, "Expected to see disabled NIC but it is has invalid value %d\n",
+			       nic_interface);
+	}
+
+	mlx5_disable_device(dev);
+}
+
+int mlx5_health_wait_pci_up(struct mlx5_core_dev *dev)
+{
+	unsigned long end;
+
+	end = jiffies + msecs_to_jiffies(mlx5_tout_ms(dev, FW_RESET));
+	while (sensor_pci_not_working(dev)) {
+		if (time_after(jiffies, end))
+			return -ETIMEDOUT;
+		msleep(100);
+	}
+	return 0;
+}
+
+static int mlx5_health_try_recover(struct mlx5_core_dev *dev)
+{
+	mlx5_core_warn(dev, "handling bad device here\n");
+	mlx5_handle_bad_state(dev);
+	if (mlx5_health_wait_pci_up(dev)) {
+		mlx5_core_err(dev, "health recovery flow aborted, PCI reads still not working\n");
+		return -EIO;
+	}
+	mlx5_core_err(dev, "starting health recovery flow\n");
+	if (mlx5_recover_device(dev) || mlx5_health_check_fatal_sensors(dev)) {
+		mlx5_core_err(dev, "health recovery failed\n");
+		return -EIO;
+	}
+
+	mlx5_core_info(dev, "health recovery succeeded\n");
+	return 0;
+}
+
+static const char *hsynd_str(u8 synd)
+{
+	switch (synd) {
+	case MLX5_HEALTH_SYNDR_FW_ERR:
+		return "firmware internal error";
+	case MLX5_HEALTH_SYNDR_IRISC_ERR:
+		return "irisc not responding";
+	case MLX5_HEALTH_SYNDR_HW_UNRECOVERABLE_ERR:
+		return "unrecoverable hardware error";
+	case MLX5_HEALTH_SYNDR_CRC_ERR:
+		return "firmware CRC error";
+	case MLX5_HEALTH_SYNDR_FETCH_PCI_ERR:
+		return "ICM fetch PCI error";
+	case MLX5_HEALTH_SYNDR_HW_FTL_ERR:
+		return "HW fatal error\n";
+	case MLX5_HEALTH_SYNDR_ASYNC_EQ_OVERRUN_ERR:
+		return "async EQ buffer overrun";
+	case MLX5_HEALTH_SYNDR_EQ_ERR:
+		return "EQ error";
+	case MLX5_HEALTH_SYNDR_EQ_INV:
+		return "Invalid EQ referenced";
+	case MLX5_HEALTH_SYNDR_FFSER_ERR:
+		return "FFSER error";
+	case MLX5_HEALTH_SYNDR_HIGH_TEMP:
+		return "High temperature";
+	default:
+		return "unrecognized error";
+	}
+}
+
+static const char *mlx5_loglevel_str(int level)
+{
+	switch (level) {
+	case LOGLEVEL_EMERG:
+		return "EMERGENCY";
+	case LOGLEVEL_ALERT:
+		return "ALERT";
+	case LOGLEVEL_CRIT:
+		return "CRITICAL";
+	case LOGLEVEL_ERR:
+		return "ERROR";
+	case LOGLEVEL_WARNING:
+		return "WARNING";
+	case LOGLEVEL_NOTICE:
+		return "NOTICE";
+	case LOGLEVEL_INFO:
+		return "INFO";
+	case LOGLEVEL_DEBUG:
+		return "DEBUG";
+	}
+	return "Unknown log level";
+}
+
+static int mlx5_health_get_severity(u8 rfr_severity)
+{
+	return rfr_severity & MLX5_SEVERITY_VALID_MASK ?
+	       rfr_severity & MLX5_SEVERITY_MASK : LOGLEVEL_ERR;
+}
+
+static void print_health_info(struct mlx5_core_dev *dev)
+{
+	struct mlx5_core_health *health = &dev->priv.health;
+	struct health_buffer __iomem *h = health->health;
+	u8 rfr_severity;
+	int severity;
+	int i;
+
+	/* If the syndrome is 0, the device is OK and no need to print buffer */
+	if (!ioread8(&h->synd))
+		return;
+
+	if (ioread32be(&h->fw_ver) == 0xFFFFFFFF) {
+		mlx5_log(dev, LOGLEVEL_ERR, "PCI slot is unavailable\n");
+		return;
+	}
+
+	rfr_severity = ioread8(&h->rfr_severity);
+	severity  = mlx5_health_get_severity(rfr_severity);
+	mlx5_log(dev, severity, "Health issue observed, %s, severity(%d) %s:\n",
+		 hsynd_str(ioread8(&h->synd)), severity, mlx5_loglevel_str(severity));
+
+	for (i = 0; i < ARRAY_SIZE(h->assert_var); i++)
+		mlx5_log(dev, severity, "assert_var[%d] 0x%08x\n", i,
+			 ioread32be(h->assert_var + i));
+
+	mlx5_log(dev, severity, "assert_exit_ptr 0x%08x\n", ioread32be(&h->assert_exit_ptr));
+	mlx5_log(dev, severity, "assert_callra 0x%08x\n", ioread32be(&h->assert_callra));
+	mlx5_log(dev, severity, "fw_ver %d.%d.%d", fw_rev_maj(dev), fw_rev_min(dev),
+		 fw_rev_sub(dev));
+	mlx5_log(dev, severity, "time %u\n", ioread32be(&h->time));
+	mlx5_log(dev, severity, "hw_id 0x%08x\n", ioread32be(&h->hw_id));
+	mlx5_log(dev, severity, "rfr %d\n", mlx5_health_get_rfr(rfr_severity));
+	mlx5_log(dev, severity, "severity %d (%s)\n", severity, mlx5_loglevel_str(severity));
+	mlx5_log(dev, severity, "irisc_index %d\n", ioread8(&h->irisc_index));
+	mlx5_log(dev, severity, "synd 0x%x: %s\n", ioread8(&h->synd),
+		 hsynd_str(ioread8(&h->synd)));
+	mlx5_log(dev, severity, "ext_synd 0x%04x\n", ioread16be(&h->ext_synd));
+	mlx5_log(dev, severity, "raw fw_ver 0x%08x\n", ioread32be(&h->fw_ver));
+}
+
+static int
+mlx5_fw_reporter_diagnose(struct devlink_health_reporter *reporter,
+			  struct devlink_fmsg *fmsg,
+			  struct netlink_ext_ack *extack)
+{
+	struct mlx5_core_dev *dev = devlink_health_reporter_priv(reporter);
+	struct mlx5_core_health *health = &dev->priv.health;
+	struct health_buffer __iomem *h = health->health;
+	u8 synd;
+	int err;
+
+	synd = ioread8(&h->synd);
+	err = devlink_fmsg_u8_pair_put(fmsg, "Syndrome", synd);
+	if (err || !synd)
+		return err;
+	return devlink_fmsg_string_pair_put(fmsg, "Description", hsynd_str(synd));
+}
+
+struct mlx5_fw_reporter_ctx {
+	u8 err_synd;
+	int miss_counter;
+};
+
+static int
+mlx5_fw_reporter_ctx_pairs_put(struct devlink_fmsg *fmsg,
+			       struct mlx5_fw_reporter_ctx *fw_reporter_ctx)
+{
+	int err;
+
+	err = devlink_fmsg_u8_pair_put(fmsg, "syndrome",
+				       fw_reporter_ctx->err_synd);
+	if (err)
+		return err;
+	err = devlink_fmsg_u32_pair_put(fmsg, "fw_miss_counter",
+					fw_reporter_ctx->miss_counter);
+	if (err)
+		return err;
+	return 0;
+}
+
+static int
+mlx5_fw_reporter_heath_buffer_data_put(struct mlx5_core_dev *dev,
+				       struct devlink_fmsg *fmsg)
+{
+	struct mlx5_core_health *health = &dev->priv.health;
+	struct health_buffer __iomem *h = health->health;
+	u8 rfr_severity;
+	int err;
+	int i;
+
+	if (!ioread8(&h->synd))
+		return 0;
+
+	err = devlink_fmsg_pair_nest_start(fmsg, "health buffer");
+	if (err)
+		return err;
+	err = devlink_fmsg_obj_nest_start(fmsg);
+	if (err)
+		return err;
+	err = devlink_fmsg_arr_pair_nest_start(fmsg, "assert_var");
+	if (err)
+		return err;
+
+	for (i = 0; i < ARRAY_SIZE(h->assert_var); i++) {
+		err = devlink_fmsg_u32_put(fmsg, ioread32be(h->assert_var + i));
+		if (err)
+			return err;
+	}
+	err = devlink_fmsg_arr_pair_nest_end(fmsg);
+	if (err)
+		return err;
+	err = devlink_fmsg_u32_pair_put(fmsg, "assert_exit_ptr",
+					ioread32be(&h->assert_exit_ptr));
+	if (err)
+		return err;
+	err = devlink_fmsg_u32_pair_put(fmsg, "assert_callra",
+					ioread32be(&h->assert_callra));
+	if (err)
+		return err;
+	err = devlink_fmsg_u32_pair_put(fmsg, "time", ioread32be(&h->time));
+	if (err)
+		return err;
+	err = devlink_fmsg_u32_pair_put(fmsg, "hw_id", ioread32be(&h->hw_id));
+	if (err)
+		return err;
+	rfr_severity = ioread8(&h->rfr_severity);
+	err = devlink_fmsg_u8_pair_put(fmsg, "rfr", mlx5_health_get_rfr(rfr_severity));
+	if (err)
+		return err;
+	err = devlink_fmsg_u8_pair_put(fmsg, "severity", mlx5_health_get_severity(rfr_severity));
+	if (err)
+		return err;
+	err = devlink_fmsg_u8_pair_put(fmsg, "irisc_index",
+				       ioread8(&h->irisc_index));
+	if (err)
+		return err;
+	err = devlink_fmsg_u8_pair_put(fmsg, "synd", ioread8(&h->synd));
+	if (err)
+		return err;
+	err = devlink_fmsg_u32_pair_put(fmsg, "ext_synd",
+					ioread16be(&h->ext_synd));
+	if (err)
+		return err;
+	err = devlink_fmsg_u32_pair_put(fmsg, "raw_fw_ver",
+					ioread32be(&h->fw_ver));
+	if (err)
+		return err;
+	err = devlink_fmsg_obj_nest_end(fmsg);
+	if (err)
+		return err;
+	return devlink_fmsg_pair_nest_end(fmsg);
+}
+
+static int
+mlx5_fw_reporter_dump(struct devlink_health_reporter *reporter,
+		      struct devlink_fmsg *fmsg, void *priv_ctx,
+		      struct netlink_ext_ack *extack)
+{
+	struct mlx5_core_dev *dev = devlink_health_reporter_priv(reporter);
+	int err;
+
+	err = mlx5_fw_tracer_trigger_core_dump_general(dev);
+	if (err)
+		return err;
+
+	if (priv_ctx) {
+		struct mlx5_fw_reporter_ctx *fw_reporter_ctx = priv_ctx;
+
+		err = mlx5_fw_reporter_ctx_pairs_put(fmsg, fw_reporter_ctx);
+		if (err)
+			return err;
+	}
+
+	err = mlx5_fw_reporter_heath_buffer_data_put(dev, fmsg);
+	if (err)
+		return err;
+	return mlx5_fw_tracer_get_saved_traces_objects(dev->tracer, fmsg);
+}
+
+static void mlx5_fw_reporter_err_work(struct work_struct *work)
+{
+	struct mlx5_fw_reporter_ctx fw_reporter_ctx;
+	struct mlx5_core_health *health;
+
+	health = container_of(work, struct mlx5_core_health, report_work);
+
+	if (IS_ERR_OR_NULL(health->fw_reporter))
+		return;
+
+	fw_reporter_ctx.err_synd = health->synd;
+	fw_reporter_ctx.miss_counter = health->miss_counter;
+	if (fw_reporter_ctx.err_synd) {
+		devlink_health_report(health->fw_reporter,
+				      "FW syndrome reported", &fw_reporter_ctx);
+		return;
+	}
+	if (fw_reporter_ctx.miss_counter)
+		devlink_health_report(health->fw_reporter,
+				      "FW miss counter reported",
+				      &fw_reporter_ctx);
+}
+
+static const struct devlink_health_reporter_ops mlx5_fw_reporter_ops = {
+		.name = "fw",
+		.diagnose = mlx5_fw_reporter_diagnose,
+		.dump = mlx5_fw_reporter_dump,
+};
+
+static int
+mlx5_fw_fatal_reporter_recover(struct devlink_health_reporter *reporter,
+			       void *priv_ctx,
+			       struct netlink_ext_ack *extack)
+{
+	struct mlx5_core_dev *dev = devlink_health_reporter_priv(reporter);
+
+	return mlx5_health_try_recover(dev);
+}
+
+static int
+mlx5_fw_fatal_reporter_dump(struct devlink_health_reporter *reporter,
+			    struct devlink_fmsg *fmsg, void *priv_ctx,
+			    struct netlink_ext_ack *extack)
+{
+	struct mlx5_core_dev *dev = devlink_health_reporter_priv(reporter);
+	u32 crdump_size = dev->priv.health.crdump_size;
+	u32 *cr_data;
+	int err;
+
+	if (!mlx5_core_is_pf(dev))
+		return -EPERM;
+
+	cr_data = kvmalloc(crdump_size, GFP_KERNEL);
+	if (!cr_data)
+		return -ENOMEM;
+	err = mlx5_crdump_collect(dev, cr_data);
+	if (err)
+		goto free_data;
+
+	if (priv_ctx) {
+		struct mlx5_fw_reporter_ctx *fw_reporter_ctx = priv_ctx;
+
+		err = mlx5_fw_reporter_ctx_pairs_put(fmsg, fw_reporter_ctx);
+		if (err)
+			goto free_data;
+	}
+
+	err = devlink_fmsg_binary_pair_put(fmsg, "crdump_data", cr_data, crdump_size);
+
+free_data:
+	kvfree(cr_data);
+	return err;
+}
+
+static void mlx5_fw_fatal_reporter_err_work(struct work_struct *work)
+{
+	struct mlx5_fw_reporter_ctx fw_reporter_ctx;
+	struct mlx5_core_health *health;
+	struct mlx5_core_dev *dev;
+	struct devlink *devlink;
+	struct mlx5_priv *priv;
+
+	health = container_of(work, struct mlx5_core_health, fatal_report_work);
+	priv = container_of(health, struct mlx5_priv, health);
+	dev = container_of(priv, struct mlx5_core_dev, priv);
+	devlink = priv_to_devlink(dev);
+
+	mutex_lock(&dev->intf_state_mutex);
+	if (test_bit(MLX5_DROP_NEW_HEALTH_WORK, &health->flags)) {
+		mlx5_core_err(dev, "health works are not permitted at this stage\n");
+		mutex_unlock(&dev->intf_state_mutex);
+		return;
+	}
+	mutex_unlock(&dev->intf_state_mutex);
+	enter_error_state(dev, false);
+	if (IS_ERR_OR_NULL(health->fw_fatal_reporter)) {
+		devl_lock(devlink);
+		if (mlx5_health_try_recover(dev))
+			mlx5_core_err(dev, "health recovery failed\n");
+		devl_unlock(devlink);
+		return;
+	}
+	fw_reporter_ctx.err_synd = health->synd;
+	fw_reporter_ctx.miss_counter = health->miss_counter;
+	if (devlink_health_report(health->fw_fatal_reporter,
+				  "FW fatal error reported", &fw_reporter_ctx) == -ECANCELED) {
+		/* If recovery wasn't performed, due to grace period,
+		 * unload the driver. This ensures that the driver
+		 * closes all its resources and it is not subjected to
+		 * requests from the kernel.
+		 */
+		mlx5_core_err(dev, "Driver is in error state. Unloading\n");
+		mlx5_unload_one(dev, false);
+	}
+}
+
+static const struct devlink_health_reporter_ops mlx5_fw_fatal_reporter_ops = {
+		.name = "fw_fatal",
+		.recover = mlx5_fw_fatal_reporter_recover,
+		.dump = mlx5_fw_fatal_reporter_dump,
+};
+
+#define MLX5_FW_REPORTER_ECPF_GRACEFUL_PERIOD 180000
+#define MLX5_FW_REPORTER_PF_GRACEFUL_PERIOD 60000
+#define MLX5_FW_REPORTER_VF_GRACEFUL_PERIOD 30000
+#define MLX5_FW_REPORTER_DEFAULT_GRACEFUL_PERIOD MLX5_FW_REPORTER_VF_GRACEFUL_PERIOD
+
+static void mlx5_fw_reporters_create(struct mlx5_core_dev *dev)
+{
+	struct mlx5_core_health *health = &dev->priv.health;
+	struct devlink *devlink = priv_to_devlink(dev);
+	u64 grace_period;
+
+	if (mlx5_core_is_ecpf(dev)) {
+		grace_period = MLX5_FW_REPORTER_ECPF_GRACEFUL_PERIOD;
+	} else if (mlx5_core_is_pf(dev)) {
+		grace_period = MLX5_FW_REPORTER_PF_GRACEFUL_PERIOD;
+	} else {
+		/* VF or SF */
+		grace_period = MLX5_FW_REPORTER_DEFAULT_GRACEFUL_PERIOD;
+	}
+
+	health->fw_reporter =
+		devlink_health_reporter_create(devlink, &mlx5_fw_reporter_ops,
+					       0, dev);
+	if (IS_ERR(health->fw_reporter))
+		mlx5_core_warn(dev, "Failed to create fw reporter, err = %ld\n",
+			       PTR_ERR(health->fw_reporter));
+
+	health->fw_fatal_reporter =
+		devlink_health_reporter_create(devlink,
+					       &mlx5_fw_fatal_reporter_ops,
+					       grace_period,
+					       dev);
+	if (IS_ERR(health->fw_fatal_reporter))
+		mlx5_core_warn(dev, "Failed to create fw fatal reporter, err = %ld\n",
+			       PTR_ERR(health->fw_fatal_reporter));
+}
+
+static void mlx5_fw_reporters_destroy(struct mlx5_core_dev *dev)
+{
+	struct mlx5_core_health *health = &dev->priv.health;
+
+	if (!IS_ERR_OR_NULL(health->fw_reporter))
+		devlink_health_reporter_destroy(health->fw_reporter);
+
+	if (!IS_ERR_OR_NULL(health->fw_fatal_reporter))
+		devlink_health_reporter_destroy(health->fw_fatal_reporter);
+}
+
+static unsigned long get_next_poll_jiffies(struct mlx5_core_dev *dev)
+{
+	unsigned long next;
+
+	get_random_bytes(&next, sizeof(next));
+	next %= HZ;
+	next += jiffies + msecs_to_jiffies(mlx5_tout_ms(dev, HEALTH_POLL_INTERVAL));
+
+	return next;
+}
+
+void mlx5_trigger_health_work(struct mlx5_core_dev *dev)
+{
+	struct mlx5_core_health *health = &dev->priv.health;
+	unsigned long flags;
+
+	spin_lock_irqsave(&health->wq_lock, flags);
+	if (!test_bit(MLX5_DROP_NEW_HEALTH_WORK, &health->flags))
+		queue_work(health->wq, &health->fatal_report_work);
+	else
+		mlx5_core_err(dev, "new health works are not permitted at this stage\n");
+	spin_unlock_irqrestore(&health->wq_lock, flags);
+}
+
+#define MLX5_MSEC_PER_HOUR (MSEC_PER_SEC * 60 * 60)
+static void mlx5_health_log_ts_update(struct work_struct *work)
+{
+	struct delayed_work *dwork = to_delayed_work(work);
+	u32 out[MLX5_ST_SZ_DW(mrtc_reg)] = {};
+	u32 in[MLX5_ST_SZ_DW(mrtc_reg)] = {};
+	struct mlx5_core_health *health;
+	struct mlx5_core_dev *dev;
+	struct mlx5_priv *priv;
+	u64 now_us;
+
+	health = container_of(dwork, struct mlx5_core_health, update_fw_log_ts_work);
+	priv = container_of(health, struct mlx5_priv, health);
+	dev = container_of(priv, struct mlx5_core_dev, priv);
+
+	now_us =  ktime_to_us(ktime_get_real());
+
+	MLX5_SET(mrtc_reg, in, time_h, now_us >> 32);
+	MLX5_SET(mrtc_reg, in, time_l, now_us & 0xFFFFFFFF);
+	mlx5_core_access_reg(dev, in, sizeof(in), out, sizeof(out), MLX5_REG_MRTC, 0, 1);
+
+	queue_delayed_work(health->wq, &health->update_fw_log_ts_work,
+			   msecs_to_jiffies(MLX5_MSEC_PER_HOUR));
+}
+
+static void poll_health(struct timer_list *t)
+{
+	struct mlx5_core_dev *dev = from_timer(dev, t, priv.health.timer);
+	struct mlx5_core_health *health = &dev->priv.health;
+	struct health_buffer __iomem *h = health->health;
+	u32 fatal_error;
+	u8 prev_synd;
+	u32 count;
+
+	if (dev->state == MLX5_DEVICE_STATE_INTERNAL_ERROR)
+		goto out;
+
+	fatal_error = mlx5_health_check_fatal_sensors(dev);
+
+	if (fatal_error && !health->fatal_error) {
+		mlx5_core_err(dev, "Fatal error %u detected\n", fatal_error);
+		dev->priv.health.fatal_error = fatal_error;
+		print_health_info(dev);
+		dev->state = MLX5_DEVICE_STATE_INTERNAL_ERROR;
+		mlx5_trigger_health_work(dev);
+		return;
+	}
+
+	count = ioread32be(health->health_counter);
+	if (count == health->prev)
+		++health->miss_counter;
+	else
+		health->miss_counter = 0;
+
+	health->prev = count;
+	if (health->miss_counter == MAX_MISSES) {
+		mlx5_core_err(dev, "device's health compromised - reached miss count\n");
+		print_health_info(dev);
+		queue_work(health->wq, &health->report_work);
+	}
+
+	prev_synd = health->synd;
+	health->synd = ioread8(&h->synd);
+	if (health->synd && health->synd != prev_synd)
+		queue_work(health->wq, &health->report_work);
+
+out:
+	mod_timer(&health->timer, get_next_poll_jiffies(dev));
+}
+
+void mlx5_start_health_poll(struct mlx5_core_dev *dev)
+{
+	u64 poll_interval_ms =  mlx5_tout_ms(dev, HEALTH_POLL_INTERVAL);
+	struct mlx5_core_health *health = &dev->priv.health;
+
+	timer_setup(&health->timer, poll_health, 0);
+	health->fatal_error = MLX5_SENSOR_NO_ERR;
+	clear_bit(MLX5_DROP_NEW_HEALTH_WORK, &health->flags);
+	health->health = &dev->iseg->health;
+	health->health_counter = &dev->iseg->health_counter;
+
+	health->timer.expires = jiffies + msecs_to_jiffies(poll_interval_ms);
+	add_timer(&health->timer);
+}
+
+void mlx5_stop_health_poll(struct mlx5_core_dev *dev, bool disable_health)
+{
+	struct mlx5_core_health *health = &dev->priv.health;
+	unsigned long flags;
+
+	if (disable_health) {
+		spin_lock_irqsave(&health->wq_lock, flags);
+		set_bit(MLX5_DROP_NEW_HEALTH_WORK, &health->flags);
+		spin_unlock_irqrestore(&health->wq_lock, flags);
+	}
+
+	del_timer_sync(&health->timer);
+}
+
+void mlx5_start_health_fw_log_up(struct mlx5_core_dev *dev)
+{
+	struct mlx5_core_health *health = &dev->priv.health;
+
+	if (mlx5_core_is_pf(dev) && MLX5_CAP_MCAM_REG(dev, mrtc))
+		queue_delayed_work(health->wq, &health->update_fw_log_ts_work, 0);
+}
+
+void mlx5_drain_health_wq(struct mlx5_core_dev *dev)
+{
+	struct mlx5_core_health *health = &dev->priv.health;
+	unsigned long flags;
+
+	spin_lock_irqsave(&health->wq_lock, flags);
+	set_bit(MLX5_DROP_NEW_HEALTH_WORK, &health->flags);
+	spin_unlock_irqrestore(&health->wq_lock, flags);
+	cancel_delayed_work_sync(&health->update_fw_log_ts_work);
+	cancel_work_sync(&health->report_work);
+	cancel_work_sync(&health->fatal_report_work);
+}
+
+void mlx5_health_cleanup(struct mlx5_core_dev *dev)
+{
+	struct mlx5_core_health *health = &dev->priv.health;
+
+	cancel_delayed_work_sync(&health->update_fw_log_ts_work);
+	destroy_workqueue(health->wq);
+	mlx5_fw_reporters_destroy(dev);
+}
+
+int mlx5_health_init(struct mlx5_core_dev *dev)
+{
+	struct mlx5_core_health *health;
+	char *name;
+
+	mlx5_fw_reporters_create(dev);
+
+	health = &dev->priv.health;
+	name = kmalloc(64, GFP_KERNEL);
+	if (!name)
+		goto out_err;
+
+	strcpy(name, "mlx5_health");
+	strcat(name, dev_name(dev->device));
+	health->wq = create_singlethread_workqueue(name);
+	kfree(name);
+	if (!health->wq)
+		goto out_err;
+	spin_lock_init(&health->wq_lock);
+	INIT_WORK(&health->fatal_report_work, mlx5_fw_fatal_reporter_err_work);
+	INIT_WORK(&health->report_work, mlx5_fw_reporter_err_work);
+	INIT_DELAYED_WORK(&health->update_fw_log_ts_work, mlx5_health_log_ts_update);
+
+	return 0;
+
+out_err:
+	mlx5_fw_reporters_destroy(dev);
+	return -ENOMEM;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/ipoib/ethtool.c b/drivers/net/ethernet/mellanox/mlx5/core/ipoib/ethtool.c
new file mode 100644
index 000000000..e09518f88
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/ipoib/ethtool.c
@@ -0,0 +1,294 @@
+/*
+ * Copyright (c) 2017, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "en.h"
+#include "ipoib.h"
+#include "en/fs_ethtool.h"
+
+static void mlx5i_get_drvinfo(struct net_device *dev,
+			      struct ethtool_drvinfo *drvinfo)
+{
+	struct mlx5e_priv *priv = mlx5i_epriv(dev);
+
+	mlx5e_ethtool_get_drvinfo(priv, drvinfo);
+	strscpy(drvinfo->driver, KBUILD_MODNAME "[ib_ipoib]",
+		sizeof(drvinfo->driver));
+}
+
+static void mlx5i_get_strings(struct net_device *dev, u32 stringset, u8 *data)
+{
+	struct mlx5e_priv *priv  = mlx5i_epriv(dev);
+
+	mlx5e_ethtool_get_strings(priv, stringset, data);
+}
+
+static int mlx5i_get_sset_count(struct net_device *dev, int sset)
+{
+	struct mlx5e_priv *priv = mlx5i_epriv(dev);
+
+	return mlx5e_ethtool_get_sset_count(priv, sset);
+}
+
+static void mlx5i_get_ethtool_stats(struct net_device *dev,
+				    struct ethtool_stats *stats,
+				    u64 *data)
+{
+	struct mlx5e_priv *priv = mlx5i_epriv(dev);
+
+	mlx5e_ethtool_get_ethtool_stats(priv, stats, data);
+}
+
+static int mlx5i_set_ringparam(struct net_device *dev,
+			       struct ethtool_ringparam *param,
+			       struct kernel_ethtool_ringparam *kernel_param,
+			       struct netlink_ext_ack *extack)
+{
+	struct mlx5e_priv *priv = mlx5i_epriv(dev);
+
+	return mlx5e_ethtool_set_ringparam(priv, param);
+}
+
+static void mlx5i_get_ringparam(struct net_device *dev,
+				struct ethtool_ringparam *param,
+				struct kernel_ethtool_ringparam *kernel_param,
+				struct netlink_ext_ack *extack)
+{
+	struct mlx5e_priv *priv = mlx5i_epriv(dev);
+
+	mlx5e_ethtool_get_ringparam(priv, param, kernel_param);
+}
+
+static int mlx5i_set_channels(struct net_device *dev,
+			      struct ethtool_channels *ch)
+{
+	struct mlx5i_priv *ipriv = netdev_priv(dev);
+	struct mlx5e_priv *epriv = mlx5i_epriv(dev);
+
+	/* rtnl lock protects from race between this ethtool op and sub
+	 * interface ndo_init/uninit.
+	 */
+	ASSERT_RTNL();
+	if (ipriv->num_sub_interfaces > 0) {
+		mlx5_core_warn(epriv->mdev,
+			       "can't change number of channels for interfaces with sub interfaces (%u)\n",
+			       ipriv->num_sub_interfaces);
+		return -EINVAL;
+	}
+
+	return mlx5e_ethtool_set_channels(epriv, ch);
+}
+
+static void mlx5i_get_channels(struct net_device *dev,
+			       struct ethtool_channels *ch)
+{
+	struct mlx5e_priv *priv = mlx5i_epriv(dev);
+
+	mlx5e_ethtool_get_channels(priv, ch);
+}
+
+static int mlx5i_set_coalesce(struct net_device *netdev,
+			      struct ethtool_coalesce *coal,
+			      struct kernel_ethtool_coalesce *kernel_coal,
+			      struct netlink_ext_ack *extack)
+{
+	struct mlx5e_priv *priv = mlx5i_epriv(netdev);
+
+	return mlx5e_ethtool_set_coalesce(priv, coal, kernel_coal, extack);
+}
+
+static int mlx5i_get_coalesce(struct net_device *netdev,
+			      struct ethtool_coalesce *coal,
+			      struct kernel_ethtool_coalesce *kernel_coal,
+			      struct netlink_ext_ack *extack)
+{
+	struct mlx5e_priv *priv = mlx5i_epriv(netdev);
+
+	return mlx5e_ethtool_get_coalesce(priv, coal, kernel_coal);
+}
+
+static int mlx5i_get_ts_info(struct net_device *netdev,
+			     struct ethtool_ts_info *info)
+{
+	struct mlx5e_priv *priv = mlx5i_epriv(netdev);
+
+	return mlx5e_ethtool_get_ts_info(priv, info);
+}
+
+static int mlx5i_flash_device(struct net_device *netdev,
+			      struct ethtool_flash *flash)
+{
+	struct mlx5e_priv *priv = mlx5i_epriv(netdev);
+
+	return mlx5e_ethtool_flash_device(priv, flash);
+}
+
+static inline int mlx5_ptys_width_enum_to_int(enum mlx5_ptys_width width)
+{
+	switch (width) {
+	case MLX5_PTYS_WIDTH_1X:  return  1;
+	case MLX5_PTYS_WIDTH_2X:  return  2;
+	case MLX5_PTYS_WIDTH_4X:  return  4;
+	case MLX5_PTYS_WIDTH_8X:  return  8;
+	case MLX5_PTYS_WIDTH_12X: return 12;
+	default:		  return -1;
+	}
+}
+
+enum mlx5_ptys_rate {
+	MLX5_PTYS_RATE_SDR	= 1 << 0,
+	MLX5_PTYS_RATE_DDR	= 1 << 1,
+	MLX5_PTYS_RATE_QDR	= 1 << 2,
+	MLX5_PTYS_RATE_FDR10	= 1 << 3,
+	MLX5_PTYS_RATE_FDR	= 1 << 4,
+	MLX5_PTYS_RATE_EDR	= 1 << 5,
+	MLX5_PTYS_RATE_HDR	= 1 << 6,
+	MLX5_PTYS_RATE_NDR	= 1 << 7,
+};
+
+static inline int mlx5_ptys_rate_enum_to_int(enum mlx5_ptys_rate rate)
+{
+	switch (rate) {
+	case MLX5_PTYS_RATE_SDR:   return 2500;
+	case MLX5_PTYS_RATE_DDR:   return 5000;
+	case MLX5_PTYS_RATE_QDR:
+	case MLX5_PTYS_RATE_FDR10: return 10000;
+	case MLX5_PTYS_RATE_FDR:   return 14000;
+	case MLX5_PTYS_RATE_EDR:   return 25000;
+	case MLX5_PTYS_RATE_HDR:   return 50000;
+	case MLX5_PTYS_RATE_NDR:   return 100000;
+	default:		   return -1;
+	}
+}
+
+static u32 mlx5i_get_speed_settings(u16 ib_link_width_oper, u16 ib_proto_oper)
+{
+	int rate, width;
+
+	rate = mlx5_ptys_rate_enum_to_int(ib_proto_oper);
+	if (rate < 0)
+		return SPEED_UNKNOWN;
+	width = mlx5_ptys_width_enum_to_int(ib_link_width_oper);
+	if (width < 0)
+		return SPEED_UNKNOWN;
+
+	return rate * width;
+}
+
+static int mlx5i_get_link_ksettings(struct net_device *netdev,
+				    struct ethtool_link_ksettings *link_ksettings)
+{
+	struct mlx5e_priv *priv = mlx5i_epriv(netdev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+	u16 ib_link_width_oper;
+	u16 ib_proto_oper;
+	int speed, ret;
+
+	ret = mlx5_query_ib_port_oper(mdev, &ib_link_width_oper, &ib_proto_oper,
+				      1);
+	if (ret)
+		return ret;
+
+	ethtool_link_ksettings_zero_link_mode(link_ksettings, supported);
+	ethtool_link_ksettings_zero_link_mode(link_ksettings, advertising);
+
+	speed = mlx5i_get_speed_settings(ib_link_width_oper, ib_proto_oper);
+	link_ksettings->base.speed = speed;
+	link_ksettings->base.duplex = speed == SPEED_UNKNOWN ? DUPLEX_UNKNOWN : DUPLEX_FULL;
+
+	link_ksettings->base.port = PORT_OTHER;
+
+	link_ksettings->base.autoneg = AUTONEG_DISABLE;
+
+	return 0;
+}
+
+static u32 mlx5i_flow_type_mask(u32 flow_type)
+{
+	return flow_type & ~(FLOW_EXT | FLOW_MAC_EXT | FLOW_RSS);
+}
+
+static int mlx5i_set_rxnfc(struct net_device *dev, struct ethtool_rxnfc *cmd)
+{
+	struct mlx5e_priv *priv = mlx5i_epriv(dev);
+	struct ethtool_rx_flow_spec *fs = &cmd->fs;
+
+	if (mlx5i_flow_type_mask(fs->flow_type) == ETHER_FLOW)
+		return -EINVAL;
+
+	return mlx5e_ethtool_set_rxnfc(priv, cmd);
+}
+
+static int mlx5i_get_rxnfc(struct net_device *dev, struct ethtool_rxnfc *info,
+			   u32 *rule_locs)
+{
+	struct mlx5e_priv *priv = mlx5i_epriv(dev);
+
+	/* ETHTOOL_GRXRINGS is needed by ethtool -x which is not part
+	 * of rxnfc. We keep this logic out of mlx5e_ethtool_get_rxnfc,
+	 * to avoid breaking "ethtool -x" when mlx5e_ethtool_get_rxnfc
+	 * is compiled out via CONFIG_MLX5_EN_RXNFC=n.
+	 */
+	if (info->cmd == ETHTOOL_GRXRINGS) {
+		info->data = priv->channels.params.num_channels;
+		return 0;
+	}
+
+	return mlx5e_ethtool_get_rxnfc(priv, info, rule_locs);
+}
+
+const struct ethtool_ops mlx5i_ethtool_ops = {
+	.supported_coalesce_params = ETHTOOL_COALESCE_USECS |
+				     ETHTOOL_COALESCE_MAX_FRAMES |
+				     ETHTOOL_COALESCE_USE_ADAPTIVE,
+	.get_drvinfo        = mlx5i_get_drvinfo,
+	.get_strings        = mlx5i_get_strings,
+	.get_sset_count     = mlx5i_get_sset_count,
+	.get_ethtool_stats  = mlx5i_get_ethtool_stats,
+	.get_ringparam      = mlx5i_get_ringparam,
+	.set_ringparam      = mlx5i_set_ringparam,
+	.flash_device       = mlx5i_flash_device,
+	.get_channels       = mlx5i_get_channels,
+	.set_channels       = mlx5i_set_channels,
+	.get_coalesce       = mlx5i_get_coalesce,
+	.set_coalesce       = mlx5i_set_coalesce,
+	.get_ts_info        = mlx5i_get_ts_info,
+	.get_rxnfc          = mlx5i_get_rxnfc,
+	.set_rxnfc          = mlx5i_set_rxnfc,
+	.get_link_ksettings = mlx5i_get_link_ksettings,
+	.get_link           = ethtool_op_get_link,
+};
+
+const struct ethtool_ops mlx5i_pkey_ethtool_ops = {
+	.get_drvinfo        = mlx5i_get_drvinfo,
+	.get_link           = ethtool_op_get_link,
+	.get_ts_info        = mlx5i_get_ts_info,
+};
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/ipoib/ipoib.c b/drivers/net/ethernet/mellanox/mlx5/core/ipoib/ipoib.c
new file mode 100644
index 000000000..aed4e8961
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/ipoib/ipoib.c
@@ -0,0 +1,842 @@
+/*
+ * Copyright (c) 2017, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <rdma/ib_verbs.h>
+#include <linux/mlx5/fs.h>
+#include "en.h"
+#include "en/params.h"
+#include "ipoib.h"
+#include "en/fs_ethtool.h"
+
+#define IB_DEFAULT_Q_KEY   0xb1b
+#define MLX5I_PARAMS_DEFAULT_LOG_RQ_SIZE 9
+
+static int mlx5i_open(struct net_device *netdev);
+static int mlx5i_close(struct net_device *netdev);
+static int mlx5i_change_mtu(struct net_device *netdev, int new_mtu);
+
+static const struct net_device_ops mlx5i_netdev_ops = {
+	.ndo_open                = mlx5i_open,
+	.ndo_stop                = mlx5i_close,
+	.ndo_get_stats64         = mlx5i_get_stats,
+	.ndo_init                = mlx5i_dev_init,
+	.ndo_uninit              = mlx5i_dev_cleanup,
+	.ndo_change_mtu          = mlx5i_change_mtu,
+	.ndo_eth_ioctl            = mlx5i_ioctl,
+};
+
+/* IPoIB mlx5 netdev profile */
+static void mlx5i_build_nic_params(struct mlx5_core_dev *mdev,
+				   struct mlx5e_params *params)
+{
+	/* Override RQ params as IPoIB supports only LINKED LIST RQ for now */
+	MLX5E_SET_PFLAG(params, MLX5E_PFLAG_RX_STRIDING_RQ, false);
+	mlx5e_set_rq_type(mdev, params);
+	mlx5e_init_rq_type_params(mdev, params);
+
+	/* RQ size in ipoib by default is 512 */
+	params->log_rq_mtu_frames = is_kdump_kernel() ?
+		MLX5E_PARAMS_MINIMUM_LOG_RQ_SIZE :
+		MLX5I_PARAMS_DEFAULT_LOG_RQ_SIZE;
+
+	params->packet_merge.type = MLX5E_PACKET_MERGE_NONE;
+	params->hard_mtu = MLX5_IB_GRH_BYTES + MLX5_IPOIB_HARD_LEN;
+
+	/* CQE compression is not supported for IPoIB */
+	params->rx_cqe_compress_def = false;
+	MLX5E_SET_PFLAG(params, MLX5E_PFLAG_RX_CQE_COMPRESS, params->rx_cqe_compress_def);
+}
+
+/* Called directly after IPoIB netdevice was created to initialize SW structs */
+int mlx5i_init(struct mlx5_core_dev *mdev, struct net_device *netdev)
+{
+	struct mlx5e_priv *priv  = mlx5i_epriv(netdev);
+
+	netif_carrier_off(netdev);
+	mlx5e_set_netdev_mtu_boundaries(priv);
+	netdev->mtu = netdev->max_mtu;
+
+	mlx5e_build_nic_params(priv, NULL, netdev->mtu);
+	mlx5i_build_nic_params(mdev, &priv->channels.params);
+
+	mlx5e_timestamp_init(priv);
+
+	/* netdev init */
+	netdev->hw_features    |= NETIF_F_SG;
+	netdev->hw_features    |= NETIF_F_IP_CSUM;
+	netdev->hw_features    |= NETIF_F_IPV6_CSUM;
+	netdev->hw_features    |= NETIF_F_GRO;
+	netdev->hw_features    |= NETIF_F_TSO;
+	netdev->hw_features    |= NETIF_F_TSO6;
+	netdev->hw_features    |= NETIF_F_RXCSUM;
+	netdev->hw_features    |= NETIF_F_RXHASH;
+
+	netdev->netdev_ops = &mlx5i_netdev_ops;
+	netdev->ethtool_ops = &mlx5i_ethtool_ops;
+
+	return 0;
+}
+
+/* Called directly before IPoIB netdevice is destroyed to cleanup SW structs */
+void mlx5i_cleanup(struct mlx5e_priv *priv)
+{
+	mlx5e_priv_cleanup(priv);
+}
+
+static void mlx5i_grp_sw_update_stats(struct mlx5e_priv *priv)
+{
+	struct rtnl_link_stats64 s = {};
+	int i, j;
+
+	for (i = 0; i < priv->stats_nch; i++) {
+		struct mlx5e_channel_stats *channel_stats;
+		struct mlx5e_rq_stats *rq_stats;
+
+		channel_stats = priv->channel_stats[i];
+		rq_stats = &channel_stats->rq;
+
+		s.rx_packets += rq_stats->packets;
+		s.rx_bytes   += rq_stats->bytes;
+
+		for (j = 0; j < priv->max_opened_tc; j++) {
+			struct mlx5e_sq_stats *sq_stats = &channel_stats->sq[j];
+
+			s.tx_packets           += sq_stats->packets;
+			s.tx_bytes             += sq_stats->bytes;
+			s.tx_dropped           += sq_stats->dropped;
+		}
+	}
+
+	memset(&priv->stats.sw, 0, sizeof(s));
+
+	priv->stats.sw.rx_packets = s.rx_packets;
+	priv->stats.sw.rx_bytes = s.rx_bytes;
+	priv->stats.sw.tx_packets = s.tx_packets;
+	priv->stats.sw.tx_bytes = s.tx_bytes;
+	priv->stats.sw.tx_queue_dropped = s.tx_dropped;
+}
+
+void mlx5i_get_stats(struct net_device *dev, struct rtnl_link_stats64 *stats)
+{
+	struct mlx5e_priv     *priv   = mlx5i_epriv(dev);
+	struct mlx5e_sw_stats *sstats = &priv->stats.sw;
+
+	mlx5i_grp_sw_update_stats(priv);
+
+	stats->rx_packets = sstats->rx_packets;
+	stats->rx_bytes   = sstats->rx_bytes;
+	stats->tx_packets = sstats->tx_packets;
+	stats->tx_bytes   = sstats->tx_bytes;
+	stats->tx_dropped = sstats->tx_queue_dropped;
+}
+
+struct net_device *mlx5i_parent_get(struct net_device *netdev)
+{
+	struct mlx5e_priv *priv = mlx5i_epriv(netdev);
+	struct mlx5i_priv *ipriv, *parent_ipriv;
+	struct net_device *parent_dev;
+	int parent_ifindex;
+
+	ipriv = priv->ppriv;
+
+	parent_ifindex = netdev->netdev_ops->ndo_get_iflink(netdev);
+	parent_dev = dev_get_by_index(dev_net(netdev), parent_ifindex);
+	if (!parent_dev)
+		return NULL;
+
+	parent_ipriv = netdev_priv(parent_dev);
+
+	ASSERT_RTNL();
+	parent_ipriv->num_sub_interfaces++;
+
+	ipriv->parent_dev = parent_dev;
+
+	return parent_dev;
+}
+
+void mlx5i_parent_put(struct net_device *netdev)
+{
+	struct mlx5e_priv *priv = mlx5i_epriv(netdev);
+	struct mlx5i_priv *ipriv, *parent_ipriv;
+
+	ipriv = priv->ppriv;
+	parent_ipriv = netdev_priv(ipriv->parent_dev);
+
+	ASSERT_RTNL();
+	parent_ipriv->num_sub_interfaces--;
+
+	dev_put(ipriv->parent_dev);
+}
+
+int mlx5i_init_underlay_qp(struct mlx5e_priv *priv)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+	struct mlx5i_priv *ipriv = priv->ppriv;
+	int ret;
+
+	{
+		u32 in[MLX5_ST_SZ_DW(rst2init_qp_in)] = {};
+		u32 *qpc;
+
+		qpc = MLX5_ADDR_OF(rst2init_qp_in, in, qpc);
+
+		MLX5_SET(qpc, qpc, pm_state, MLX5_QP_PM_MIGRATED);
+		MLX5_SET(qpc, qpc, primary_address_path.pkey_index,
+			 ipriv->pkey_index);
+		MLX5_SET(qpc, qpc, primary_address_path.vhca_port_num, 1);
+		MLX5_SET(qpc, qpc, q_key, IB_DEFAULT_Q_KEY);
+
+		MLX5_SET(rst2init_qp_in, in, opcode, MLX5_CMD_OP_RST2INIT_QP);
+		MLX5_SET(rst2init_qp_in, in, qpn, ipriv->qpn);
+		ret = mlx5_cmd_exec_in(mdev, rst2init_qp, in);
+		if (ret)
+			goto err_qp_modify_to_err;
+	}
+	{
+		u32 in[MLX5_ST_SZ_DW(init2rtr_qp_in)] = {};
+
+		MLX5_SET(init2rtr_qp_in, in, opcode, MLX5_CMD_OP_INIT2RTR_QP);
+		MLX5_SET(init2rtr_qp_in, in, qpn, ipriv->qpn);
+		ret = mlx5_cmd_exec_in(mdev, init2rtr_qp, in);
+		if (ret)
+			goto err_qp_modify_to_err;
+	}
+	{
+		u32 in[MLX5_ST_SZ_DW(rtr2rts_qp_in)] = {};
+
+		MLX5_SET(rtr2rts_qp_in, in, opcode, MLX5_CMD_OP_RTR2RTS_QP);
+		MLX5_SET(rtr2rts_qp_in, in, qpn, ipriv->qpn);
+		ret = mlx5_cmd_exec_in(mdev, rtr2rts_qp, in);
+		if (ret)
+			goto err_qp_modify_to_err;
+	}
+	return 0;
+
+err_qp_modify_to_err:
+	{
+		u32 in[MLX5_ST_SZ_DW(qp_2err_in)] = {};
+
+		MLX5_SET(qp_2err_in, in, opcode, MLX5_CMD_OP_2ERR_QP);
+		MLX5_SET(qp_2err_in, in, qpn, ipriv->qpn);
+		mlx5_cmd_exec_in(mdev, qp_2err, in);
+	}
+	return ret;
+}
+
+void mlx5i_uninit_underlay_qp(struct mlx5e_priv *priv)
+{
+	struct mlx5i_priv *ipriv = priv->ppriv;
+	struct mlx5_core_dev *mdev = priv->mdev;
+	u32 in[MLX5_ST_SZ_DW(qp_2rst_in)] = {};
+
+	MLX5_SET(qp_2rst_in, in, opcode, MLX5_CMD_OP_2RST_QP);
+	MLX5_SET(qp_2rst_in, in, qpn, ipriv->qpn);
+	mlx5_cmd_exec_in(mdev, qp_2rst, in);
+}
+
+#define MLX5_QP_ENHANCED_ULP_STATELESS_MODE 2
+
+int mlx5i_create_underlay_qp(struct mlx5e_priv *priv)
+{
+	const unsigned char *dev_addr = priv->netdev->dev_addr;
+	u32 out[MLX5_ST_SZ_DW(create_qp_out)] = {};
+	u32 in[MLX5_ST_SZ_DW(create_qp_in)] = {};
+	struct mlx5i_priv *ipriv = priv->ppriv;
+	void *addr_path;
+	int qpn = 0;
+	int ret = 0;
+	void *qpc;
+
+	if (MLX5_CAP_GEN(priv->mdev, mkey_by_name)) {
+		qpn = (dev_addr[1] << 16) + (dev_addr[2] << 8) + dev_addr[3];
+		MLX5_SET(create_qp_in, in, input_qpn, qpn);
+	}
+
+	qpc = MLX5_ADDR_OF(create_qp_in, in, qpc);
+	MLX5_SET(qpc, qpc, ts_format, mlx5_get_qp_default_ts(priv->mdev));
+	MLX5_SET(qpc, qpc, st, MLX5_QP_ST_UD);
+	MLX5_SET(qpc, qpc, pm_state, MLX5_QP_PM_MIGRATED);
+	MLX5_SET(qpc, qpc, ulp_stateless_offload_mode,
+		 MLX5_QP_ENHANCED_ULP_STATELESS_MODE);
+
+	addr_path = MLX5_ADDR_OF(qpc, qpc, primary_address_path);
+	MLX5_SET(ads, addr_path, vhca_port_num, 1);
+	MLX5_SET(ads, addr_path, grh, 1);
+
+	MLX5_SET(create_qp_in, in, opcode, MLX5_CMD_OP_CREATE_QP);
+	ret = mlx5_cmd_exec_inout(priv->mdev, create_qp, in, out);
+	if (ret)
+		return ret;
+
+	ipriv->qpn = MLX5_GET(create_qp_out, out, qpn);
+
+	return 0;
+}
+
+void mlx5i_destroy_underlay_qp(struct mlx5_core_dev *mdev, u32 qpn)
+{
+	u32 in[MLX5_ST_SZ_DW(destroy_qp_in)] = {};
+
+	MLX5_SET(destroy_qp_in, in, opcode, MLX5_CMD_OP_DESTROY_QP);
+	MLX5_SET(destroy_qp_in, in, qpn, qpn);
+	mlx5_cmd_exec_in(mdev, destroy_qp, in);
+}
+
+int mlx5i_update_nic_rx(struct mlx5e_priv *priv)
+{
+	return mlx5e_refresh_tirs(priv, true, true);
+}
+
+int mlx5i_create_tis(struct mlx5_core_dev *mdev, u32 underlay_qpn, u32 *tisn)
+{
+	u32 in[MLX5_ST_SZ_DW(create_tis_in)] = {};
+	void *tisc;
+
+	tisc = MLX5_ADDR_OF(create_tis_in, in, ctx);
+
+	MLX5_SET(tisc, tisc, underlay_qpn, underlay_qpn);
+
+	return mlx5e_create_tis(mdev, in, tisn);
+}
+
+static int mlx5i_init_tx(struct mlx5e_priv *priv)
+{
+	struct mlx5i_priv *ipriv = priv->ppriv;
+	int err;
+
+	err = mlx5i_create_underlay_qp(priv);
+	if (err) {
+		mlx5_core_warn(priv->mdev, "create underlay QP failed, %d\n", err);
+		return err;
+	}
+
+	err = mlx5i_create_tis(priv->mdev, ipriv->qpn, &priv->tisn[0][0]);
+	if (err) {
+		mlx5_core_warn(priv->mdev, "create tis failed, %d\n", err);
+		goto err_destroy_underlay_qp;
+	}
+
+	return 0;
+
+err_destroy_underlay_qp:
+	mlx5i_destroy_underlay_qp(priv->mdev, ipriv->qpn);
+	return err;
+}
+
+static void mlx5i_cleanup_tx(struct mlx5e_priv *priv)
+{
+	struct mlx5i_priv *ipriv = priv->ppriv;
+
+	mlx5e_destroy_tis(priv->mdev, priv->tisn[0][0]);
+	mlx5i_destroy_underlay_qp(priv->mdev, ipriv->qpn);
+}
+
+static int mlx5i_create_flow_steering(struct mlx5e_priv *priv)
+{
+	struct mlx5_flow_namespace *ns =
+		mlx5_get_flow_namespace(priv->mdev, MLX5_FLOW_NAMESPACE_KERNEL);
+	int err;
+
+
+	if (!ns)
+		return -EINVAL;
+
+	mlx5e_fs_set_ns(priv->fs, ns, false);
+	err = mlx5e_arfs_create_tables(priv->fs, priv->rx_res,
+				       !!(priv->netdev->hw_features & NETIF_F_NTUPLE));
+	if (err) {
+		netdev_err(priv->netdev, "Failed to create arfs tables, err=%d\n",
+			   err);
+		priv->netdev->hw_features &= ~NETIF_F_NTUPLE;
+	}
+
+	err = mlx5e_create_ttc_table(priv->fs, priv->rx_res);
+	if (err) {
+		netdev_err(priv->netdev, "Failed to create ttc table, err=%d\n",
+			   err);
+		goto err_destroy_arfs_tables;
+	}
+
+	mlx5e_ethtool_init_steering(priv->fs);
+
+	return 0;
+
+err_destroy_arfs_tables:
+	mlx5e_arfs_destroy_tables(priv->fs,
+				  !!(priv->netdev->hw_features & NETIF_F_NTUPLE));
+
+	return err;
+}
+
+static void mlx5i_destroy_flow_steering(struct mlx5e_priv *priv)
+{
+	mlx5e_destroy_ttc_table(priv->fs);
+	mlx5e_arfs_destroy_tables(priv->fs,
+				  !!(priv->netdev->hw_features & NETIF_F_NTUPLE));
+	mlx5e_ethtool_cleanup_steering(priv->fs);
+}
+
+static int mlx5i_init_rx(struct mlx5e_priv *priv)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+	int err;
+
+	priv->fs = mlx5e_fs_init(priv->profile, mdev,
+				 !test_bit(MLX5E_STATE_DESTROYING, &priv->state));
+	if (!priv->fs) {
+		netdev_err(priv->netdev, "FS allocation failed\n");
+		return -ENOMEM;
+	}
+
+	priv->rx_res = mlx5e_rx_res_alloc();
+	if (!priv->rx_res) {
+		err = -ENOMEM;
+		goto err_free_fs;
+	}
+
+	mlx5e_create_q_counters(priv);
+
+	err = mlx5e_open_drop_rq(priv, &priv->drop_rq);
+	if (err) {
+		mlx5_core_err(mdev, "open drop rq failed, %d\n", err);
+		goto err_destroy_q_counters;
+	}
+
+	err = mlx5e_rx_res_init(priv->rx_res, priv->mdev, 0,
+				priv->max_nch, priv->drop_rq.rqn,
+				&priv->channels.params.packet_merge,
+				priv->channels.params.num_channels);
+	if (err)
+		goto err_close_drop_rq;
+
+	err = mlx5i_create_flow_steering(priv);
+	if (err)
+		goto err_destroy_rx_res;
+
+	return 0;
+
+err_destroy_rx_res:
+	mlx5e_rx_res_destroy(priv->rx_res);
+err_close_drop_rq:
+	mlx5e_close_drop_rq(&priv->drop_rq);
+err_destroy_q_counters:
+	mlx5e_destroy_q_counters(priv);
+	mlx5e_rx_res_free(priv->rx_res);
+	priv->rx_res = NULL;
+err_free_fs:
+	mlx5e_fs_cleanup(priv->fs);
+	return err;
+}
+
+static void mlx5i_cleanup_rx(struct mlx5e_priv *priv)
+{
+	mlx5i_destroy_flow_steering(priv);
+	mlx5e_rx_res_destroy(priv->rx_res);
+	mlx5e_close_drop_rq(&priv->drop_rq);
+	mlx5e_destroy_q_counters(priv);
+	mlx5e_rx_res_free(priv->rx_res);
+	priv->rx_res = NULL;
+	mlx5e_fs_cleanup(priv->fs);
+}
+
+/* The stats groups order is opposite to the update_stats() order calls */
+static mlx5e_stats_grp_t mlx5i_stats_grps[] = {
+	&MLX5E_STATS_GRP(sw),
+	&MLX5E_STATS_GRP(qcnt),
+	&MLX5E_STATS_GRP(vnic_env),
+	&MLX5E_STATS_GRP(vport),
+	&MLX5E_STATS_GRP(802_3),
+	&MLX5E_STATS_GRP(2863),
+	&MLX5E_STATS_GRP(2819),
+	&MLX5E_STATS_GRP(phy),
+	&MLX5E_STATS_GRP(pcie),
+	&MLX5E_STATS_GRP(per_prio),
+	&MLX5E_STATS_GRP(pme),
+	&MLX5E_STATS_GRP(channels),
+	&MLX5E_STATS_GRP(per_port_buff_congest),
+};
+
+static unsigned int mlx5i_stats_grps_num(struct mlx5e_priv *priv)
+{
+	return ARRAY_SIZE(mlx5i_stats_grps);
+}
+
+static const struct mlx5e_profile mlx5i_nic_profile = {
+	.init		   = mlx5i_init,
+	.cleanup	   = mlx5i_cleanup,
+	.init_tx	   = mlx5i_init_tx,
+	.cleanup_tx	   = mlx5i_cleanup_tx,
+	.init_rx	   = mlx5i_init_rx,
+	.cleanup_rx	   = mlx5i_cleanup_rx,
+	.enable		   = NULL, /* mlx5i_enable */
+	.disable	   = NULL, /* mlx5i_disable */
+	.update_rx	   = mlx5i_update_nic_rx,
+	.update_stats	   = NULL, /* mlx5i_update_stats */
+	.update_carrier    = NULL, /* no HW update in IB link */
+	.rx_handlers       = &mlx5i_rx_handlers,
+	.max_tc		   = MLX5I_MAX_NUM_TC,
+	.stats_grps        = mlx5i_stats_grps,
+	.stats_grps_num    = mlx5i_stats_grps_num,
+};
+
+/* mlx5i netdev NDos */
+
+static int mlx5i_change_mtu(struct net_device *netdev, int new_mtu)
+{
+	struct mlx5e_priv *priv = mlx5i_epriv(netdev);
+	struct mlx5e_params new_params;
+	int err = 0;
+
+	mutex_lock(&priv->state_lock);
+
+	new_params = priv->channels.params;
+	new_params.sw_mtu = new_mtu;
+
+	err = mlx5e_safe_switch_params(priv, &new_params, NULL, NULL, true);
+	if (err)
+		goto out;
+
+	netdev->mtu = new_params.sw_mtu;
+
+out:
+	mutex_unlock(&priv->state_lock);
+	return err;
+}
+
+int mlx5i_dev_init(struct net_device *dev)
+{
+	struct mlx5e_priv    *priv   = mlx5i_epriv(dev);
+	struct mlx5i_priv    *ipriv  = priv->ppriv;
+	u8 addr_mod[3];
+
+	/* Set dev address using underlay QP */
+	addr_mod[0] = (ipriv->qpn >> 16) & 0xff;
+	addr_mod[1] = (ipriv->qpn >>  8) & 0xff;
+	addr_mod[2] = (ipriv->qpn) & 0xff;
+	dev_addr_mod(dev, 1, addr_mod, sizeof(addr_mod));
+
+	/* Add QPN to net-device mapping to HT */
+	mlx5i_pkey_add_qpn(dev, ipriv->qpn);
+
+	return 0;
+}
+
+int mlx5i_ioctl(struct net_device *dev, struct ifreq *ifr, int cmd)
+{
+	struct mlx5e_priv *priv = mlx5i_epriv(dev);
+
+	switch (cmd) {
+	case SIOCSHWTSTAMP:
+		return mlx5e_hwstamp_set(priv, ifr);
+	case SIOCGHWTSTAMP:
+		return mlx5e_hwstamp_get(priv, ifr);
+	default:
+		return -EOPNOTSUPP;
+	}
+}
+
+void mlx5i_dev_cleanup(struct net_device *dev)
+{
+	struct mlx5e_priv    *priv   = mlx5i_epriv(dev);
+	struct mlx5i_priv    *ipriv = priv->ppriv;
+
+	mlx5i_uninit_underlay_qp(priv);
+
+	/* Delete QPN to net-device mapping from HT */
+	mlx5i_pkey_del_qpn(dev, ipriv->qpn);
+}
+
+static int mlx5i_open(struct net_device *netdev)
+{
+	struct mlx5e_priv *epriv = mlx5i_epriv(netdev);
+	struct mlx5i_priv *ipriv = epriv->ppriv;
+	struct mlx5_core_dev *mdev = epriv->mdev;
+	int err;
+
+	mutex_lock(&epriv->state_lock);
+
+	set_bit(MLX5E_STATE_OPENED, &epriv->state);
+
+	err = mlx5i_init_underlay_qp(epriv);
+	if (err) {
+		mlx5_core_warn(mdev, "prepare underlay qp state failed, %d\n", err);
+		goto err_clear_state_opened_flag;
+	}
+
+	err = mlx5_fs_add_rx_underlay_qpn(mdev, ipriv->qpn);
+	if (err) {
+		mlx5_core_warn(mdev, "attach underlay qp to ft failed, %d\n", err);
+		goto err_reset_qp;
+	}
+
+	err = mlx5e_open_channels(epriv, &epriv->channels);
+	if (err)
+		goto err_remove_fs_underlay_qp;
+
+	epriv->profile->update_rx(epriv);
+	mlx5e_activate_priv_channels(epriv);
+
+	mutex_unlock(&epriv->state_lock);
+	return 0;
+
+err_remove_fs_underlay_qp:
+	mlx5_fs_remove_rx_underlay_qpn(mdev, ipriv->qpn);
+err_reset_qp:
+	mlx5i_uninit_underlay_qp(epriv);
+err_clear_state_opened_flag:
+	clear_bit(MLX5E_STATE_OPENED, &epriv->state);
+	mutex_unlock(&epriv->state_lock);
+	return err;
+}
+
+static int mlx5i_close(struct net_device *netdev)
+{
+	struct mlx5e_priv *epriv = mlx5i_epriv(netdev);
+	struct mlx5i_priv *ipriv = epriv->ppriv;
+	struct mlx5_core_dev *mdev = epriv->mdev;
+
+	/* May already be CLOSED in case a previous configuration operation
+	 * (e.g RX/TX queue size change) that involves close&open failed.
+	 */
+	mutex_lock(&epriv->state_lock);
+
+	if (!test_bit(MLX5E_STATE_OPENED, &epriv->state))
+		goto unlock;
+
+	clear_bit(MLX5E_STATE_OPENED, &epriv->state);
+
+	netif_carrier_off(epriv->netdev);
+	mlx5_fs_remove_rx_underlay_qpn(mdev, ipriv->qpn);
+	mlx5e_deactivate_priv_channels(epriv);
+	mlx5e_close_channels(&epriv->channels);
+	mlx5i_uninit_underlay_qp(epriv);
+unlock:
+	mutex_unlock(&epriv->state_lock);
+	return 0;
+}
+
+/* IPoIB RDMA netdev callbacks */
+static int mlx5i_attach_mcast(struct net_device *netdev, struct ib_device *hca,
+			      union ib_gid *gid, u16 lid, int set_qkey,
+			      u32 qkey)
+{
+	struct mlx5e_priv    *epriv = mlx5i_epriv(netdev);
+	struct mlx5_core_dev *mdev  = epriv->mdev;
+	struct mlx5i_priv    *ipriv = epriv->ppriv;
+	int err;
+
+	mlx5_core_dbg(mdev, "attaching QPN 0x%x, MGID %pI6\n", ipriv->qpn,
+		      gid->raw);
+	err = mlx5_core_attach_mcg(mdev, gid, ipriv->qpn);
+	if (err)
+		mlx5_core_warn(mdev, "failed attaching QPN 0x%x, MGID %pI6\n",
+			       ipriv->qpn, gid->raw);
+
+	if (set_qkey) {
+		mlx5_core_dbg(mdev, "%s setting qkey 0x%x\n",
+			      netdev->name, qkey);
+		ipriv->qkey = qkey;
+	}
+
+	return err;
+}
+
+static int mlx5i_detach_mcast(struct net_device *netdev, struct ib_device *hca,
+			      union ib_gid *gid, u16 lid)
+{
+	struct mlx5e_priv    *epriv = mlx5i_epriv(netdev);
+	struct mlx5_core_dev *mdev  = epriv->mdev;
+	struct mlx5i_priv    *ipriv = epriv->ppriv;
+	int err;
+
+	mlx5_core_dbg(mdev, "detaching QPN 0x%x, MGID %pI6\n", ipriv->qpn,
+		      gid->raw);
+
+	err = mlx5_core_detach_mcg(mdev, gid, ipriv->qpn);
+	if (err)
+		mlx5_core_dbg(mdev, "failed detaching QPN 0x%x, MGID %pI6\n",
+			      ipriv->qpn, gid->raw);
+
+	return err;
+}
+
+static int mlx5i_xmit(struct net_device *dev, struct sk_buff *skb,
+		      struct ib_ah *address, u32 dqpn)
+{
+	struct mlx5e_priv *epriv = mlx5i_epriv(dev);
+	struct mlx5e_txqsq *sq   = epriv->txq2sq[skb_get_queue_mapping(skb)];
+	struct mlx5_ib_ah *mah   = to_mah(address);
+	struct mlx5i_priv *ipriv = epriv->ppriv;
+
+	mlx5i_sq_xmit(sq, skb, &mah->av, dqpn, ipriv->qkey, netdev_xmit_more());
+
+	return NETDEV_TX_OK;
+}
+
+static void mlx5i_set_pkey_index(struct net_device *netdev, int id)
+{
+	struct mlx5i_priv *ipriv = netdev_priv(netdev);
+
+	ipriv->pkey_index = (u16)id;
+}
+
+static int mlx5i_check_required_hca_cap(struct mlx5_core_dev *mdev)
+{
+	if (MLX5_CAP_GEN(mdev, port_type) != MLX5_CAP_PORT_TYPE_IB)
+		return -EOPNOTSUPP;
+
+	if (!MLX5_CAP_GEN(mdev, ipoib_enhanced_offloads)) {
+		mlx5_core_warn(mdev, "IPoIB enhanced offloads are not supported\n");
+		return -EOPNOTSUPP;
+	}
+
+	return 0;
+}
+
+static void mlx5_rdma_netdev_free(struct net_device *netdev)
+{
+	struct mlx5e_priv *priv = mlx5i_epriv(netdev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+	struct mlx5i_priv *ipriv = priv->ppriv;
+	const struct mlx5e_profile *profile = priv->profile;
+
+	mlx5e_detach_netdev(priv);
+	profile->cleanup(priv);
+
+	if (!ipriv->sub_interface) {
+		mlx5i_pkey_qpn_ht_cleanup(netdev);
+		mlx5e_destroy_mdev_resources(mdev);
+	}
+}
+
+static bool mlx5_is_sub_interface(struct mlx5_core_dev *mdev)
+{
+	return mdev->mlx5e_res.hw_objs.pdn != 0;
+}
+
+static const struct mlx5e_profile *mlx5_get_profile(struct mlx5_core_dev *mdev)
+{
+	if (mlx5_is_sub_interface(mdev))
+		return mlx5i_pkey_get_profile();
+	return &mlx5i_nic_profile;
+}
+
+static int mlx5_rdma_setup_rn(struct ib_device *ibdev, u32 port_num,
+			      struct net_device *netdev, void *param)
+{
+	struct mlx5_core_dev *mdev = (struct mlx5_core_dev *)param;
+	const struct mlx5e_profile *prof = mlx5_get_profile(mdev);
+	struct mlx5i_priv *ipriv;
+	struct mlx5e_priv *epriv;
+	struct rdma_netdev *rn;
+	int err;
+
+	ipriv = netdev_priv(netdev);
+	epriv = mlx5i_epriv(netdev);
+
+	ipriv->sub_interface = mlx5_is_sub_interface(mdev);
+	if (!ipriv->sub_interface) {
+		err = mlx5i_pkey_qpn_ht_init(netdev);
+		if (err) {
+			mlx5_core_warn(mdev, "allocate qpn_to_netdev ht failed\n");
+			return err;
+		}
+
+		/* This should only be called once per mdev */
+		err = mlx5e_create_mdev_resources(mdev);
+		if (err)
+			goto destroy_ht;
+	}
+
+	err = mlx5e_priv_init(epriv, prof, netdev, mdev);
+	if (err)
+		goto destroy_mdev_resources;
+
+	epriv->profile = prof;
+	epriv->ppriv = ipriv;
+
+	prof->init(mdev, netdev);
+
+	err = mlx5e_attach_netdev(epriv);
+	if (err)
+		goto detach;
+	netif_carrier_off(netdev);
+
+	/* set rdma_netdev func pointers */
+	rn = &ipriv->rn;
+	rn->hca  = ibdev;
+	rn->send = mlx5i_xmit;
+	rn->attach_mcast = mlx5i_attach_mcast;
+	rn->detach_mcast = mlx5i_detach_mcast;
+	rn->set_id = mlx5i_set_pkey_index;
+
+	netdev->priv_destructor = mlx5_rdma_netdev_free;
+	netdev->needs_free_netdev = 1;
+
+	return 0;
+
+detach:
+	prof->cleanup(epriv);
+	if (ipriv->sub_interface)
+		return err;
+destroy_mdev_resources:
+	mlx5e_destroy_mdev_resources(mdev);
+destroy_ht:
+	mlx5i_pkey_qpn_ht_cleanup(netdev);
+	return err;
+}
+
+int mlx5_rdma_rn_get_params(struct mlx5_core_dev *mdev,
+			    struct ib_device *device,
+			    struct rdma_netdev_alloc_params *params)
+{
+	int nch;
+	int rc;
+
+	rc = mlx5i_check_required_hca_cap(mdev);
+	if (rc)
+		return rc;
+
+	nch = mlx5e_get_max_num_channels(mdev);
+
+	*params = (struct rdma_netdev_alloc_params){
+		.sizeof_priv = sizeof(struct mlx5i_priv) +
+			       sizeof(struct mlx5e_priv),
+		.txqs = nch * MLX5E_MAX_NUM_TC,
+		.rxqs = nch,
+		.param = mdev,
+		.initialize_rdma_netdev = mlx5_rdma_setup_rn,
+	};
+
+	return 0;
+}
+EXPORT_SYMBOL(mlx5_rdma_rn_get_params);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/ipoib/ipoib.h b/drivers/net/ethernet/mellanox/mlx5/core/ipoib/ipoib.h
new file mode 100644
index 000000000..f3f2af972
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/ipoib/ipoib.h
@@ -0,0 +1,127 @@
+/*
+ * Copyright (c) 2017, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef __MLX5E_IPOB_H__
+#define __MLX5E_IPOB_H__
+
+#ifdef CONFIG_MLX5_CORE_IPOIB
+
+#include <linux/mlx5/fs.h>
+#include "en.h"
+
+#define MLX5I_MAX_NUM_TC 1
+
+extern const struct ethtool_ops mlx5i_ethtool_ops;
+extern const struct ethtool_ops mlx5i_pkey_ethtool_ops;
+extern const struct mlx5e_rx_handlers mlx5i_rx_handlers;
+
+#define MLX5_IB_GRH_BYTES       40
+#define MLX5_IPOIB_ENCAP_LEN    4
+#define MLX5_IPOIB_PSEUDO_LEN   20
+#define MLX5_IPOIB_HARD_LEN     (MLX5_IPOIB_PSEUDO_LEN + MLX5_IPOIB_ENCAP_LEN)
+
+/* ipoib rdma netdev's private data structure */
+struct mlx5i_priv {
+	struct rdma_netdev rn; /* keep this first */
+	u32 qpn;
+	bool   sub_interface;
+	u32    num_sub_interfaces;
+	u32    qkey;
+	u16    pkey_index;
+	struct mlx5i_pkey_qpn_ht *qpn_htbl;
+	struct net_device *parent_dev;
+	char  *mlx5e_priv[];
+};
+
+int mlx5i_create_tis(struct mlx5_core_dev *mdev, u32 underlay_qpn, u32 *tisn);
+
+/* Underlay QP create/destroy functions */
+int mlx5i_create_underlay_qp(struct mlx5e_priv *priv);
+void mlx5i_destroy_underlay_qp(struct mlx5_core_dev *mdev, u32 qpn);
+
+/* Underlay QP state modification init/uninit functions */
+int mlx5i_init_underlay_qp(struct mlx5e_priv *priv);
+void mlx5i_uninit_underlay_qp(struct mlx5e_priv *priv);
+
+/* Allocate/Free underlay QPN to net-device hash table */
+int mlx5i_pkey_qpn_ht_init(struct net_device *netdev);
+void mlx5i_pkey_qpn_ht_cleanup(struct net_device *netdev);
+
+/* Add/Remove an underlay QPN to net-device mapping to/from the hash table */
+int mlx5i_pkey_add_qpn(struct net_device *netdev, u32 qpn);
+int mlx5i_pkey_del_qpn(struct net_device *netdev, u32 qpn);
+
+/* Get the net-device corresponding to the given underlay QPN */
+struct net_device *mlx5i_pkey_get_netdev(struct net_device *netdev, u32 qpn);
+
+/* Shared ndo functions */
+int mlx5i_dev_init(struct net_device *dev);
+void mlx5i_dev_cleanup(struct net_device *dev);
+int mlx5i_ioctl(struct net_device *dev, struct ifreq *ifr, int cmd);
+
+/* Parent profile functions */
+int mlx5i_init(struct mlx5_core_dev *mdev, struct net_device *netdev);
+void mlx5i_cleanup(struct mlx5e_priv *priv);
+
+int mlx5i_update_nic_rx(struct mlx5e_priv *priv);
+
+/* Get child interface nic profile */
+const struct mlx5e_profile *mlx5i_pkey_get_profile(void);
+
+/* Extract mlx5e_priv from IPoIB netdev */
+#define mlx5i_epriv(netdev) ((void *)(((struct mlx5i_priv *)netdev_priv(netdev))->mlx5e_priv))
+
+struct mlx5_wqe_eth_pad {
+	u8 rsvd0[16];
+};
+
+struct mlx5i_tx_wqe {
+	struct mlx5_wqe_ctrl_seg     ctrl;
+	struct mlx5_wqe_datagram_seg datagram;
+	struct mlx5_wqe_eth_pad      pad;
+	struct mlx5_wqe_eth_seg      eth;
+	struct mlx5_wqe_data_seg     data[];
+};
+
+#define MLX5I_SQ_FETCH_WQE(sq, pi) \
+	((struct mlx5i_tx_wqe *)mlx5e_fetch_wqe(&(sq)->wq, pi, sizeof(struct mlx5i_tx_wqe)))
+
+void mlx5i_sq_xmit(struct mlx5e_txqsq *sq, struct sk_buff *skb,
+		   struct mlx5_av *av, u32 dqpn, u32 dqkey, bool xmit_more);
+void mlx5i_get_stats(struct net_device *dev, struct rtnl_link_stats64 *stats);
+
+/* Reference management for child to parent interfaces. */
+struct net_device *mlx5i_parent_get(struct net_device *netdev);
+void mlx5i_parent_put(struct net_device *netdev);
+
+#endif /* CONFIG_MLX5_CORE_IPOIB */
+#endif /* __MLX5E_IPOB_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/ipoib/ipoib_vlan.c b/drivers/net/ethernet/mellanox/mlx5/core/ipoib/ipoib_vlan.c
new file mode 100644
index 000000000..0cf4eaf85
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/ipoib/ipoib_vlan.c
@@ -0,0 +1,365 @@
+/*
+ * Copyright (c) 2017, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/hash.h>
+#include "ipoib.h"
+
+#define MLX5I_MAX_LOG_PKEY_SUP 7
+
+struct qpn_to_netdev {
+	struct net_device *netdev;
+	struct hlist_node hlist;
+	u32 underlay_qpn;
+};
+
+struct mlx5i_pkey_qpn_ht {
+	struct hlist_head buckets[1 << MLX5I_MAX_LOG_PKEY_SUP];
+	spinlock_t ht_lock; /* Synchronise with NAPI */
+};
+
+int mlx5i_pkey_qpn_ht_init(struct net_device *netdev)
+{
+	struct mlx5i_priv *ipriv = netdev_priv(netdev);
+	struct mlx5i_pkey_qpn_ht *qpn_htbl;
+
+	qpn_htbl = kzalloc(sizeof(*qpn_htbl), GFP_KERNEL);
+	if (!qpn_htbl)
+		return -ENOMEM;
+
+	ipriv->qpn_htbl = qpn_htbl;
+	spin_lock_init(&qpn_htbl->ht_lock);
+
+	return 0;
+}
+
+void mlx5i_pkey_qpn_ht_cleanup(struct net_device *netdev)
+{
+	struct mlx5i_priv *ipriv = netdev_priv(netdev);
+
+	kfree(ipriv->qpn_htbl);
+}
+
+static struct qpn_to_netdev *mlx5i_find_qpn_to_netdev_node(struct hlist_head *buckets,
+							   u32 qpn)
+{
+	struct hlist_head *h = &buckets[hash_32(qpn, MLX5I_MAX_LOG_PKEY_SUP)];
+	struct qpn_to_netdev *node;
+
+	hlist_for_each_entry(node, h, hlist) {
+		if (node->underlay_qpn == qpn)
+			return node;
+	}
+
+	return NULL;
+}
+
+int mlx5i_pkey_add_qpn(struct net_device *netdev, u32 qpn)
+{
+	struct mlx5i_priv *ipriv = netdev_priv(netdev);
+	struct mlx5i_pkey_qpn_ht *ht = ipriv->qpn_htbl;
+	u8 key = hash_32(qpn, MLX5I_MAX_LOG_PKEY_SUP);
+	struct qpn_to_netdev *new_node;
+
+	new_node = kzalloc(sizeof(*new_node), GFP_KERNEL);
+	if (!new_node)
+		return -ENOMEM;
+
+	new_node->netdev = netdev;
+	new_node->underlay_qpn = qpn;
+	spin_lock_bh(&ht->ht_lock);
+	hlist_add_head(&new_node->hlist, &ht->buckets[key]);
+	spin_unlock_bh(&ht->ht_lock);
+
+	return 0;
+}
+
+int mlx5i_pkey_del_qpn(struct net_device *netdev, u32 qpn)
+{
+	struct mlx5e_priv *epriv = mlx5i_epriv(netdev);
+	struct mlx5i_priv *ipriv = epriv->ppriv;
+	struct mlx5i_pkey_qpn_ht *ht = ipriv->qpn_htbl;
+	struct qpn_to_netdev *node;
+
+	node = mlx5i_find_qpn_to_netdev_node(ht->buckets, qpn);
+	if (!node) {
+		mlx5_core_warn(epriv->mdev, "QPN to netdev delete from HT failed\n");
+		return -EINVAL;
+	}
+
+	spin_lock_bh(&ht->ht_lock);
+	hlist_del_init(&node->hlist);
+	spin_unlock_bh(&ht->ht_lock);
+	kfree(node);
+
+	return 0;
+}
+
+struct net_device *mlx5i_pkey_get_netdev(struct net_device *netdev, u32 qpn)
+{
+	struct mlx5i_priv *ipriv = netdev_priv(netdev);
+	struct qpn_to_netdev *node;
+
+	node = mlx5i_find_qpn_to_netdev_node(ipriv->qpn_htbl->buckets, qpn);
+	if (!node)
+		return NULL;
+
+	return node->netdev;
+}
+
+static int mlx5i_pkey_open(struct net_device *netdev);
+static int mlx5i_pkey_close(struct net_device *netdev);
+static int mlx5i_pkey_dev_init(struct net_device *dev);
+static void mlx5i_pkey_dev_cleanup(struct net_device *netdev);
+static int mlx5i_pkey_change_mtu(struct net_device *netdev, int new_mtu);
+static int mlx5i_pkey_ioctl(struct net_device *dev, struct ifreq *ifr, int cmd);
+
+static const struct net_device_ops mlx5i_pkey_netdev_ops = {
+	.ndo_open                = mlx5i_pkey_open,
+	.ndo_stop                = mlx5i_pkey_close,
+	.ndo_init                = mlx5i_pkey_dev_init,
+	.ndo_get_stats64         = mlx5i_get_stats,
+	.ndo_uninit              = mlx5i_pkey_dev_cleanup,
+	.ndo_change_mtu          = mlx5i_pkey_change_mtu,
+	.ndo_eth_ioctl            = mlx5i_pkey_ioctl,
+};
+
+/* Child NDOs */
+static int mlx5i_pkey_dev_init(struct net_device *dev)
+{
+	struct mlx5e_priv *priv = mlx5i_epriv(dev);
+	struct mlx5i_priv *ipriv, *parent_ipriv;
+	struct net_device *parent_dev;
+
+	ipriv = priv->ppriv;
+
+	/* Link to parent */
+	parent_dev = mlx5i_parent_get(dev);
+	if (!parent_dev) {
+		mlx5_core_warn(priv->mdev, "failed to get parent device\n");
+		return -EINVAL;
+	}
+
+	if (dev->num_rx_queues < parent_dev->real_num_rx_queues) {
+		mlx5_core_warn(priv->mdev,
+			       "failed to create child device with rx queues [%d] less than parent's [%d]\n",
+			       dev->num_rx_queues,
+			       parent_dev->real_num_rx_queues);
+		mlx5i_parent_put(dev);
+		return -EINVAL;
+	}
+
+	/* Get QPN to netdevice hash table from parent */
+	parent_ipriv = netdev_priv(parent_dev);
+	ipriv->qpn_htbl = parent_ipriv->qpn_htbl;
+
+	return mlx5i_dev_init(dev);
+}
+
+static int mlx5i_pkey_ioctl(struct net_device *dev, struct ifreq *ifr, int cmd)
+{
+	return mlx5i_ioctl(dev, ifr, cmd);
+}
+
+static void mlx5i_pkey_dev_cleanup(struct net_device *netdev)
+{
+	mlx5i_parent_put(netdev);
+	return mlx5i_dev_cleanup(netdev);
+}
+
+static int mlx5i_pkey_open(struct net_device *netdev)
+{
+	struct mlx5e_priv *epriv = mlx5i_epriv(netdev);
+	struct mlx5i_priv *ipriv = epriv->ppriv;
+	struct mlx5_core_dev *mdev = epriv->mdev;
+	int err;
+
+	mutex_lock(&epriv->state_lock);
+
+	set_bit(MLX5E_STATE_OPENED, &epriv->state);
+
+	err = mlx5i_init_underlay_qp(epriv);
+	if (err) {
+		mlx5_core_warn(mdev, "prepare child underlay qp state failed, %d\n", err);
+		goto err_release_lock;
+	}
+
+	err = mlx5_fs_add_rx_underlay_qpn(mdev, ipriv->qpn);
+	if (err) {
+		mlx5_core_warn(mdev, "attach child underlay qp to ft failed, %d\n", err);
+		goto err_unint_underlay_qp;
+	}
+
+	err = mlx5i_create_tis(mdev, ipriv->qpn, &epriv->tisn[0][0]);
+	if (err) {
+		mlx5_core_warn(mdev, "create child tis failed, %d\n", err);
+		goto err_remove_rx_uderlay_qp;
+	}
+
+	err = mlx5e_open_channels(epriv, &epriv->channels);
+	if (err) {
+		mlx5_core_warn(mdev, "opening child channels failed, %d\n", err);
+		goto err_clear_state_opened_flag;
+	}
+	epriv->profile->update_rx(epriv);
+	mlx5e_activate_priv_channels(epriv);
+	mutex_unlock(&epriv->state_lock);
+
+	return 0;
+
+err_clear_state_opened_flag:
+	mlx5e_destroy_tis(mdev, epriv->tisn[0][0]);
+err_remove_rx_uderlay_qp:
+	mlx5_fs_remove_rx_underlay_qpn(mdev, ipriv->qpn);
+err_unint_underlay_qp:
+	mlx5i_uninit_underlay_qp(epriv);
+err_release_lock:
+	clear_bit(MLX5E_STATE_OPENED, &epriv->state);
+	mutex_unlock(&epriv->state_lock);
+	return err;
+}
+
+static int mlx5i_pkey_close(struct net_device *netdev)
+{
+	struct mlx5e_priv *priv = mlx5i_epriv(netdev);
+	struct mlx5i_priv *ipriv = priv->ppriv;
+	struct mlx5_core_dev *mdev = priv->mdev;
+
+	mutex_lock(&priv->state_lock);
+
+	if (!test_bit(MLX5E_STATE_OPENED, &priv->state))
+		goto unlock;
+
+	clear_bit(MLX5E_STATE_OPENED, &priv->state);
+
+	netif_carrier_off(priv->netdev);
+	mlx5_fs_remove_rx_underlay_qpn(mdev, ipriv->qpn);
+	mlx5i_uninit_underlay_qp(priv);
+	mlx5e_deactivate_priv_channels(priv);
+	mlx5e_close_channels(&priv->channels);
+	mlx5e_destroy_tis(mdev, priv->tisn[0][0]);
+unlock:
+	mutex_unlock(&priv->state_lock);
+	return 0;
+}
+
+static int mlx5i_pkey_change_mtu(struct net_device *netdev, int new_mtu)
+{
+	struct mlx5e_priv *priv = mlx5i_epriv(netdev);
+
+	mutex_lock(&priv->state_lock);
+	netdev->mtu = new_mtu;
+	mutex_unlock(&priv->state_lock);
+
+	return 0;
+}
+
+/* Called directly after IPoIB netdevice was created to initialize SW structs */
+static int mlx5i_pkey_init(struct mlx5_core_dev *mdev,
+			   struct net_device *netdev)
+{
+	struct mlx5e_priv *priv  = mlx5i_epriv(netdev);
+	int err;
+
+	err = mlx5i_init(mdev, netdev);
+	if (err)
+		return err;
+
+	/* Override parent ndo */
+	netdev->netdev_ops = &mlx5i_pkey_netdev_ops;
+
+	/* Set child limited ethtool support */
+	netdev->ethtool_ops = &mlx5i_pkey_ethtool_ops;
+
+	/* Use dummy rqs */
+	priv->channels.params.log_rq_mtu_frames = MLX5E_PARAMS_MINIMUM_LOG_RQ_SIZE;
+
+	return 0;
+}
+
+/* Called directly before IPoIB netdevice is destroyed to cleanup SW structs */
+static void mlx5i_pkey_cleanup(struct mlx5e_priv *priv)
+{
+	mlx5i_cleanup(priv);
+}
+
+static int mlx5i_pkey_init_tx(struct mlx5e_priv *priv)
+{
+	int err;
+
+	err = mlx5i_create_underlay_qp(priv);
+	if (err)
+		mlx5_core_warn(priv->mdev, "create child underlay QP failed, %d\n", err);
+
+	return err;
+}
+
+static void mlx5i_pkey_cleanup_tx(struct mlx5e_priv *priv)
+{
+	struct mlx5i_priv *ipriv = priv->ppriv;
+
+	mlx5i_destroy_underlay_qp(priv->mdev, ipriv->qpn);
+}
+
+static int mlx5i_pkey_init_rx(struct mlx5e_priv *priv)
+{
+	/* Since the rx resources are shared between child and parent, the
+	 * parent interface is taking care of rx resource allocation and init
+	 */
+	return 0;
+}
+
+static void mlx5i_pkey_cleanup_rx(struct mlx5e_priv *priv)
+{
+	/* Since the rx resources are shared between child and parent, the
+	 * parent interface is taking care of rx resource free and de-init
+	 */
+}
+
+static const struct mlx5e_profile mlx5i_pkey_nic_profile = {
+	.init		   = mlx5i_pkey_init,
+	.cleanup	   = mlx5i_pkey_cleanup,
+	.init_tx	   = mlx5i_pkey_init_tx,
+	.cleanup_tx	   = mlx5i_pkey_cleanup_tx,
+	.init_rx	   = mlx5i_pkey_init_rx,
+	.cleanup_rx	   = mlx5i_pkey_cleanup_rx,
+	.enable		   = NULL,
+	.disable	   = NULL,
+	.update_rx	   = mlx5i_update_nic_rx,
+	.update_stats	   = NULL,
+	.rx_handlers       = &mlx5i_rx_handlers,
+	.max_tc		   = MLX5I_MAX_NUM_TC,
+};
+
+const struct mlx5e_profile *mlx5i_pkey_get_profile(void)
+{
+	return &mlx5i_pkey_nic_profile;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/irq_affinity.c b/drivers/net/ethernet/mellanox/mlx5/core/irq_affinity.c
new file mode 100644
index 000000000..380a208ab
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/irq_affinity.c
@@ -0,0 +1,226 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. All rights reserved. */
+
+#include "mlx5_core.h"
+#include "mlx5_irq.h"
+#include "pci_irq.h"
+
+static void cpu_put(struct mlx5_irq_pool *pool, int cpu)
+{
+	pool->irqs_per_cpu[cpu]--;
+}
+
+static void cpu_get(struct mlx5_irq_pool *pool, int cpu)
+{
+	pool->irqs_per_cpu[cpu]++;
+}
+
+/* Gets the least loaded CPU. e.g.: the CPU with least IRQs bound to it */
+static int cpu_get_least_loaded(struct mlx5_irq_pool *pool,
+				const struct cpumask *req_mask)
+{
+	int best_cpu = -1;
+	int cpu;
+
+	for_each_cpu_and(cpu, req_mask, cpu_online_mask) {
+		/* CPU has zero IRQs on it. No need to search any more CPUs. */
+		if (!pool->irqs_per_cpu[cpu]) {
+			best_cpu = cpu;
+			break;
+		}
+		if (best_cpu < 0)
+			best_cpu = cpu;
+		if (pool->irqs_per_cpu[cpu] < pool->irqs_per_cpu[best_cpu])
+			best_cpu = cpu;
+	}
+	if (best_cpu == -1) {
+		/* There isn't online CPUs in req_mask */
+		mlx5_core_err(pool->dev, "NO online CPUs in req_mask (%*pbl)\n",
+			      cpumask_pr_args(req_mask));
+		best_cpu = cpumask_first(cpu_online_mask);
+	}
+	pool->irqs_per_cpu[best_cpu]++;
+	return best_cpu;
+}
+
+/* Creating an IRQ from irq_pool */
+static struct mlx5_irq *
+irq_pool_request_irq(struct mlx5_irq_pool *pool, const struct cpumask *req_mask)
+{
+	cpumask_var_t auto_mask;
+	struct mlx5_irq *irq;
+	u32 irq_index;
+	int err;
+
+	if (!zalloc_cpumask_var(&auto_mask, GFP_KERNEL))
+		return ERR_PTR(-ENOMEM);
+	err = xa_alloc(&pool->irqs, &irq_index, NULL, pool->xa_num_irqs, GFP_KERNEL);
+	if (err)
+		return ERR_PTR(err);
+	if (pool->irqs_per_cpu) {
+		if (cpumask_weight(req_mask) > 1)
+			/* if req_mask contain more then one CPU, set the least loadad CPU
+			 * of req_mask
+			 */
+			cpumask_set_cpu(cpu_get_least_loaded(pool, req_mask), auto_mask);
+		else
+			cpu_get(pool, cpumask_first(req_mask));
+	}
+	irq = mlx5_irq_alloc(pool, irq_index, cpumask_empty(auto_mask) ? req_mask : auto_mask);
+	free_cpumask_var(auto_mask);
+	return irq;
+}
+
+/* Looking for the IRQ with the smallest refcount that fits req_mask.
+ * If pool is sf_comp_pool, then we are looking for an IRQ with any of the
+ * requested CPUs in req_mask.
+ * for example: req_mask = 0xf, irq0_mask = 0x10, irq1_mask = 0x1. irq0_mask
+ * isn't subset of req_mask, so we will skip it. irq1_mask is subset of req_mask,
+ * we don't skip it.
+ * If pool is sf_ctrl_pool, then all IRQs have the same mask, so any IRQ will
+ * fit. And since mask is subset of itself, we will pass the first if bellow.
+ */
+static struct mlx5_irq *
+irq_pool_find_least_loaded(struct mlx5_irq_pool *pool, const struct cpumask *req_mask)
+{
+	int start = pool->xa_num_irqs.min;
+	int end = pool->xa_num_irqs.max;
+	struct mlx5_irq *irq = NULL;
+	struct mlx5_irq *iter;
+	int irq_refcount = 0;
+	unsigned long index;
+
+	lockdep_assert_held(&pool->lock);
+	xa_for_each_range(&pool->irqs, index, iter, start, end) {
+		struct cpumask *iter_mask = mlx5_irq_get_affinity_mask(iter);
+		int iter_refcount = mlx5_irq_read_locked(iter);
+
+		if (!cpumask_subset(iter_mask, req_mask))
+			/* skip IRQs with a mask which is not subset of req_mask */
+			continue;
+		if (iter_refcount < pool->min_threshold)
+			/* If we found an IRQ with less than min_thres, return it */
+			return iter;
+		if (!irq || iter_refcount < irq_refcount) {
+			/* In case we won't find an IRQ with less than min_thres,
+			 * keep a pointer to the least used IRQ
+			 */
+			irq_refcount = iter_refcount;
+			irq = iter;
+		}
+	}
+	return irq;
+}
+
+/**
+ * mlx5_irq_affinity_request - request an IRQ according to the given mask.
+ * @pool: IRQ pool to request from.
+ * @req_mask: cpumask requested for this IRQ.
+ *
+ * This function returns a pointer to IRQ, or ERR_PTR in case of error.
+ */
+struct mlx5_irq *
+mlx5_irq_affinity_request(struct mlx5_irq_pool *pool, const struct cpumask *req_mask)
+{
+	struct mlx5_irq *least_loaded_irq, *new_irq;
+
+	mutex_lock(&pool->lock);
+	least_loaded_irq = irq_pool_find_least_loaded(pool, req_mask);
+	if (least_loaded_irq &&
+	    mlx5_irq_read_locked(least_loaded_irq) < pool->min_threshold)
+		goto out;
+	/* We didn't find an IRQ with less than min_thres, try to allocate a new IRQ */
+	new_irq = irq_pool_request_irq(pool, req_mask);
+	if (IS_ERR(new_irq)) {
+		if (!least_loaded_irq) {
+			/* We failed to create an IRQ and we didn't find an IRQ */
+			mlx5_core_err(pool->dev, "Didn't find a matching IRQ. err = %ld\n",
+				      PTR_ERR(new_irq));
+			mutex_unlock(&pool->lock);
+			return new_irq;
+		}
+		/* We failed to create a new IRQ for the requested affinity,
+		 * sharing existing IRQ.
+		 */
+		goto out;
+	}
+	least_loaded_irq = new_irq;
+	goto unlock;
+out:
+	mlx5_irq_get_locked(least_loaded_irq);
+	if (mlx5_irq_read_locked(least_loaded_irq) > pool->max_threshold)
+		mlx5_core_dbg(pool->dev, "IRQ %u overloaded, pool_name: %s, %u EQs on this irq\n",
+			      pci_irq_vector(pool->dev->pdev,
+					     mlx5_irq_get_index(least_loaded_irq)), pool->name,
+			      mlx5_irq_read_locked(least_loaded_irq) / MLX5_EQ_REFS_PER_IRQ);
+unlock:
+	mutex_unlock(&pool->lock);
+	return least_loaded_irq;
+}
+
+void mlx5_irq_affinity_irqs_release(struct mlx5_core_dev *dev, struct mlx5_irq **irqs,
+				    int num_irqs)
+{
+	struct mlx5_irq_pool *pool = mlx5_irq_pool_get(dev);
+	int i;
+
+	for (i = 0; i < num_irqs; i++) {
+		int cpu = cpumask_first(mlx5_irq_get_affinity_mask(irqs[i]));
+
+		synchronize_irq(pci_irq_vector(pool->dev->pdev,
+					       mlx5_irq_get_index(irqs[i])));
+		if (mlx5_irq_put(irqs[i]))
+			if (pool->irqs_per_cpu)
+				cpu_put(pool, cpu);
+	}
+}
+
+/**
+ * mlx5_irq_affinity_irqs_request_auto - request one or more IRQs for mlx5 device.
+ * @dev: mlx5 device that is requesting the IRQs.
+ * @nirqs: number of IRQs to request.
+ * @irqs: an output array of IRQs pointers.
+ *
+ * Each IRQ is bounded to at most 1 CPU.
+ * This function is requesting IRQs according to the default assignment.
+ * The default assignment policy is:
+ * - in each iteration, request the least loaded IRQ which is not bound to any
+ *   CPU of the previous IRQs requested.
+ *
+ * This function returns the number of IRQs requested, (which might be smaller than
+ * @nirqs), if successful, or a negative error code in case of an error.
+ */
+int mlx5_irq_affinity_irqs_request_auto(struct mlx5_core_dev *dev, int nirqs,
+					struct mlx5_irq **irqs)
+{
+	struct mlx5_irq_pool *pool = mlx5_irq_pool_get(dev);
+	cpumask_var_t req_mask;
+	struct mlx5_irq *irq;
+	int i = 0;
+
+	if (!zalloc_cpumask_var(&req_mask, GFP_KERNEL))
+		return -ENOMEM;
+	cpumask_copy(req_mask, cpu_online_mask);
+	for (i = 0; i < nirqs; i++) {
+		if (mlx5_irq_pool_is_sf_pool(pool))
+			irq = mlx5_irq_affinity_request(pool, req_mask);
+		else
+			/* In case SF pool doesn't exists, fallback to the PF IRQs.
+			 * The PF IRQs are already allocated and binded to CPU
+			 * at this point. Hence, only an index is needed.
+			 */
+			irq = mlx5_irq_request(dev, i, NULL);
+		if (IS_ERR(irq))
+			break;
+		irqs[i] = irq;
+		cpumask_clear_cpu(cpumask_first(mlx5_irq_get_affinity_mask(irq)), req_mask);
+		mlx5_core_dbg(pool->dev, "IRQ %u mapped to cpu %*pbl, %u EQs on this irq\n",
+			      pci_irq_vector(dev->pdev, mlx5_irq_get_index(irq)),
+			      cpumask_pr_args(mlx5_irq_get_affinity_mask(irq)),
+			      mlx5_irq_read_locked(irq) / MLX5_EQ_REFS_PER_IRQ);
+	}
+	free_cpumask_var(req_mask);
+	if (!i)
+		return PTR_ERR(irq);
+	return i;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lag/debugfs.c b/drivers/net/ethernet/mellanox/mlx5/core/lag/debugfs.c
new file mode 100644
index 000000000..b8feaf0f5
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lag/debugfs.c
@@ -0,0 +1,182 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2022, NVIDIA CORPORATION & AFFILIATES. All rights reserved. */
+
+#include "lag.h"
+
+static char *get_str_mode_type(struct mlx5_lag *ldev)
+{
+	switch (ldev->mode) {
+	case MLX5_LAG_MODE_ROCE: return "roce";
+	case MLX5_LAG_MODE_SRIOV: return "switchdev";
+	case MLX5_LAG_MODE_MULTIPATH: return "multipath";
+	case MLX5_LAG_MODE_MPESW: return "multiport_eswitch";
+	default: return "invalid";
+	}
+
+	return NULL;
+}
+
+static int type_show(struct seq_file *file, void *priv)
+{
+	struct mlx5_core_dev *dev = file->private;
+	struct mlx5_lag *ldev;
+	char *mode = NULL;
+
+	ldev = dev->priv.lag;
+	mutex_lock(&ldev->lock);
+	if (__mlx5_lag_is_active(ldev))
+		mode = get_str_mode_type(ldev);
+	mutex_unlock(&ldev->lock);
+	if (!mode)
+		return -EINVAL;
+	seq_printf(file, "%s\n", mode);
+
+	return 0;
+}
+
+static int port_sel_mode_show(struct seq_file *file, void *priv)
+{
+	struct mlx5_core_dev *dev = file->private;
+	struct mlx5_lag *ldev;
+	int ret = 0;
+	char *mode;
+
+	ldev = dev->priv.lag;
+	mutex_lock(&ldev->lock);
+	if (__mlx5_lag_is_active(ldev))
+		mode = mlx5_get_str_port_sel_mode(ldev->mode, ldev->mode_flags);
+	else
+		ret = -EINVAL;
+	mutex_unlock(&ldev->lock);
+	if (ret)
+		return ret;
+
+	seq_printf(file, "%s\n", mode);
+	return 0;
+}
+
+static int state_show(struct seq_file *file, void *priv)
+{
+	struct mlx5_core_dev *dev = file->private;
+	struct mlx5_lag *ldev;
+	bool active;
+
+	ldev = dev->priv.lag;
+	mutex_lock(&ldev->lock);
+	active = __mlx5_lag_is_active(ldev);
+	mutex_unlock(&ldev->lock);
+	seq_printf(file, "%s\n", active ? "active" : "disabled");
+	return 0;
+}
+
+static int flags_show(struct seq_file *file, void *priv)
+{
+	struct mlx5_core_dev *dev = file->private;
+	bool fdb_sel_mode_native;
+	struct mlx5_lag *ldev;
+	bool shared_fdb;
+	bool lag_active;
+
+	ldev = dev->priv.lag;
+	mutex_lock(&ldev->lock);
+	lag_active = __mlx5_lag_is_active(ldev);
+	if (!lag_active)
+		goto unlock;
+
+	shared_fdb = test_bit(MLX5_LAG_MODE_FLAG_SHARED_FDB, &ldev->mode_flags);
+	fdb_sel_mode_native = test_bit(MLX5_LAG_MODE_FLAG_FDB_SEL_MODE_NATIVE,
+				       &ldev->mode_flags);
+
+unlock:
+	mutex_unlock(&ldev->lock);
+	if (!lag_active)
+		return -EINVAL;
+
+	seq_printf(file, "%s:%s\n", "shared_fdb", shared_fdb ? "on" : "off");
+	seq_printf(file, "%s:%s\n", "fdb_selection_mode",
+		   fdb_sel_mode_native ? "native" : "affinity");
+	return 0;
+}
+
+static int mapping_show(struct seq_file *file, void *priv)
+{
+	struct mlx5_core_dev *dev = file->private;
+	u8 ports[MLX5_MAX_PORTS] = {};
+	struct mlx5_lag *ldev;
+	bool hash = false;
+	bool lag_active;
+	int num_ports;
+	int i;
+
+	ldev = dev->priv.lag;
+	mutex_lock(&ldev->lock);
+	lag_active = __mlx5_lag_is_active(ldev);
+	if (lag_active) {
+		if (test_bit(MLX5_LAG_MODE_FLAG_HASH_BASED, &ldev->mode_flags)) {
+			mlx5_infer_tx_enabled(&ldev->tracker, ldev->ports, ports,
+					      &num_ports);
+			hash = true;
+		} else {
+			for (i = 0; i < ldev->ports; i++)
+				ports[i] = ldev->v2p_map[i];
+			num_ports = ldev->ports;
+		}
+	}
+	mutex_unlock(&ldev->lock);
+	if (!lag_active)
+		return -EINVAL;
+
+	for (i = 0; i < num_ports; i++) {
+		if (hash)
+			seq_printf(file, "%d\n", ports[i] + 1);
+		else
+			seq_printf(file, "%d:%d\n", i + 1, ports[i]);
+	}
+
+	return 0;
+}
+
+static int members_show(struct seq_file *file, void *priv)
+{
+	struct mlx5_core_dev *dev = file->private;
+	struct mlx5_lag *ldev;
+	int i;
+
+	ldev = dev->priv.lag;
+	mutex_lock(&ldev->lock);
+	for (i = 0; i < ldev->ports; i++) {
+		if (!ldev->pf[i].dev)
+			continue;
+		seq_printf(file, "%s\n", dev_name(ldev->pf[i].dev->device));
+	}
+	mutex_unlock(&ldev->lock);
+
+	return 0;
+}
+
+DEFINE_SHOW_ATTRIBUTE(type);
+DEFINE_SHOW_ATTRIBUTE(port_sel_mode);
+DEFINE_SHOW_ATTRIBUTE(state);
+DEFINE_SHOW_ATTRIBUTE(flags);
+DEFINE_SHOW_ATTRIBUTE(mapping);
+DEFINE_SHOW_ATTRIBUTE(members);
+
+void mlx5_ldev_add_debugfs(struct mlx5_core_dev *dev)
+{
+	struct dentry *dbg;
+
+	dbg = debugfs_create_dir("lag", mlx5_debugfs_get_dev_root(dev));
+	dev->priv.dbg.lag_debugfs = dbg;
+
+	debugfs_create_file("type", 0444, dbg, dev, &type_fops);
+	debugfs_create_file("port_sel_mode", 0444, dbg, dev, &port_sel_mode_fops);
+	debugfs_create_file("state", 0444, dbg, dev, &state_fops);
+	debugfs_create_file("flags", 0444, dbg, dev, &flags_fops);
+	debugfs_create_file("mapping", 0444, dbg, dev, &mapping_fops);
+	debugfs_create_file("members", 0444, dbg, dev, &members_fops);
+}
+
+void mlx5_ldev_remove_debugfs(struct dentry *dbg)
+{
+	debugfs_remove_recursive(dbg);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.c b/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.c
new file mode 100644
index 000000000..ad32b80e8
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.c
@@ -0,0 +1,1580 @@
+/*
+ * Copyright (c) 2016, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/netdevice.h>
+#include <net/bonding.h>
+#include <linux/mlx5/driver.h>
+#include <linux/mlx5/eswitch.h>
+#include <linux/mlx5/vport.h>
+#include "lib/devcom.h"
+#include "mlx5_core.h"
+#include "eswitch.h"
+#include "esw/acl/ofld.h"
+#include "lag.h"
+#include "mp.h"
+#include "mpesw.h"
+
+enum {
+	MLX5_LAG_EGRESS_PORT_1 = 1,
+	MLX5_LAG_EGRESS_PORT_2,
+};
+
+/* General purpose, use for short periods of time.
+ * Beware of lock dependencies (preferably, no locks should be acquired
+ * under it).
+ */
+static DEFINE_SPINLOCK(lag_lock);
+
+static int get_port_sel_mode(enum mlx5_lag_mode mode, unsigned long flags)
+{
+	if (test_bit(MLX5_LAG_MODE_FLAG_HASH_BASED, &flags))
+		return MLX5_LAG_PORT_SELECT_MODE_PORT_SELECT_FT;
+
+	if (mode == MLX5_LAG_MODE_MPESW)
+		return MLX5_LAG_PORT_SELECT_MODE_PORT_SELECT_MPESW;
+
+	return MLX5_LAG_PORT_SELECT_MODE_QUEUE_AFFINITY;
+}
+
+static u8 lag_active_port_bits(struct mlx5_lag *ldev)
+{
+	u8 enabled_ports[MLX5_MAX_PORTS] = {};
+	u8 active_port = 0;
+	int num_enabled;
+	int idx;
+
+	mlx5_infer_tx_enabled(&ldev->tracker, ldev->ports, enabled_ports,
+			      &num_enabled);
+	for (idx = 0; idx < num_enabled; idx++)
+		active_port |= BIT_MASK(enabled_ports[idx]);
+
+	return active_port;
+}
+
+static int mlx5_cmd_create_lag(struct mlx5_core_dev *dev, u8 *ports, int mode,
+			       unsigned long flags)
+{
+	bool fdb_sel_mode = test_bit(MLX5_LAG_MODE_FLAG_FDB_SEL_MODE_NATIVE,
+				     &flags);
+	int port_sel_mode = get_port_sel_mode(mode, flags);
+	u32 in[MLX5_ST_SZ_DW(create_lag_in)] = {};
+	void *lag_ctx;
+
+	lag_ctx = MLX5_ADDR_OF(create_lag_in, in, ctx);
+	MLX5_SET(create_lag_in, in, opcode, MLX5_CMD_OP_CREATE_LAG);
+	MLX5_SET(lagc, lag_ctx, fdb_selection_mode, fdb_sel_mode);
+
+	switch (port_sel_mode) {
+	case MLX5_LAG_PORT_SELECT_MODE_QUEUE_AFFINITY:
+		MLX5_SET(lagc, lag_ctx, tx_remap_affinity_1, ports[0]);
+		MLX5_SET(lagc, lag_ctx, tx_remap_affinity_2, ports[1]);
+		break;
+	case MLX5_LAG_PORT_SELECT_MODE_PORT_SELECT_FT:
+		if (!MLX5_CAP_PORT_SELECTION(dev, port_select_flow_table_bypass))
+			break;
+
+		MLX5_SET(lagc, lag_ctx, active_port,
+			 lag_active_port_bits(mlx5_lag_dev(dev)));
+		break;
+	default:
+		break;
+	}
+	MLX5_SET(lagc, lag_ctx, port_select_mode, port_sel_mode);
+
+	return mlx5_cmd_exec_in(dev, create_lag, in);
+}
+
+static int mlx5_cmd_modify_lag(struct mlx5_core_dev *dev, u8 num_ports,
+			       u8 *ports)
+{
+	u32 in[MLX5_ST_SZ_DW(modify_lag_in)] = {};
+	void *lag_ctx = MLX5_ADDR_OF(modify_lag_in, in, ctx);
+
+	MLX5_SET(modify_lag_in, in, opcode, MLX5_CMD_OP_MODIFY_LAG);
+	MLX5_SET(modify_lag_in, in, field_select, 0x1);
+
+	MLX5_SET(lagc, lag_ctx, tx_remap_affinity_1, ports[0]);
+	MLX5_SET(lagc, lag_ctx, tx_remap_affinity_2, ports[1]);
+
+	return mlx5_cmd_exec_in(dev, modify_lag, in);
+}
+
+int mlx5_cmd_create_vport_lag(struct mlx5_core_dev *dev)
+{
+	u32 in[MLX5_ST_SZ_DW(create_vport_lag_in)] = {};
+
+	MLX5_SET(create_vport_lag_in, in, opcode, MLX5_CMD_OP_CREATE_VPORT_LAG);
+
+	return mlx5_cmd_exec_in(dev, create_vport_lag, in);
+}
+EXPORT_SYMBOL(mlx5_cmd_create_vport_lag);
+
+int mlx5_cmd_destroy_vport_lag(struct mlx5_core_dev *dev)
+{
+	u32 in[MLX5_ST_SZ_DW(destroy_vport_lag_in)] = {};
+
+	MLX5_SET(destroy_vport_lag_in, in, opcode, MLX5_CMD_OP_DESTROY_VPORT_LAG);
+
+	return mlx5_cmd_exec_in(dev, destroy_vport_lag, in);
+}
+EXPORT_SYMBOL(mlx5_cmd_destroy_vport_lag);
+
+static void mlx5_infer_tx_disabled(struct lag_tracker *tracker, u8 num_ports,
+				   u8 *ports, int *num_disabled)
+{
+	int i;
+
+	*num_disabled = 0;
+	for (i = 0; i < num_ports; i++) {
+		if (!tracker->netdev_state[i].tx_enabled ||
+		    !tracker->netdev_state[i].link_up)
+			ports[(*num_disabled)++] = i;
+	}
+}
+
+void mlx5_infer_tx_enabled(struct lag_tracker *tracker, u8 num_ports,
+			   u8 *ports, int *num_enabled)
+{
+	int i;
+
+	*num_enabled = 0;
+	for (i = 0; i < num_ports; i++) {
+		if (tracker->netdev_state[i].tx_enabled &&
+		    tracker->netdev_state[i].link_up)
+			ports[(*num_enabled)++] = i;
+	}
+
+	if (*num_enabled == 0)
+		mlx5_infer_tx_disabled(tracker, num_ports, ports, num_enabled);
+}
+
+static void mlx5_lag_print_mapping(struct mlx5_core_dev *dev,
+				   struct mlx5_lag *ldev,
+				   struct lag_tracker *tracker,
+				   unsigned long flags)
+{
+	char buf[MLX5_MAX_PORTS * 10 + 1] = {};
+	u8 enabled_ports[MLX5_MAX_PORTS] = {};
+	int written = 0;
+	int num_enabled;
+	int idx;
+	int err;
+	int i;
+	int j;
+
+	if (test_bit(MLX5_LAG_MODE_FLAG_HASH_BASED, &flags)) {
+		mlx5_infer_tx_enabled(tracker, ldev->ports, enabled_ports,
+				      &num_enabled);
+		for (i = 0; i < num_enabled; i++) {
+			err = scnprintf(buf + written, 4, "%d, ", enabled_ports[i] + 1);
+			if (err != 3)
+				return;
+			written += err;
+		}
+		buf[written - 2] = 0;
+		mlx5_core_info(dev, "lag map active ports: %s\n", buf);
+	} else {
+		for (i = 0; i < ldev->ports; i++) {
+			for (j  = 0; j < ldev->buckets; j++) {
+				idx = i * ldev->buckets + j;
+				err = scnprintf(buf + written, 10,
+						" port %d:%d", i + 1, ldev->v2p_map[idx]);
+				if (err != 9)
+					return;
+				written += err;
+			}
+		}
+		mlx5_core_info(dev, "lag map:%s\n", buf);
+	}
+}
+
+static int mlx5_lag_netdev_event(struct notifier_block *this,
+				 unsigned long event, void *ptr);
+static void mlx5_do_bond_work(struct work_struct *work);
+
+static void mlx5_ldev_free(struct kref *ref)
+{
+	struct mlx5_lag *ldev = container_of(ref, struct mlx5_lag, ref);
+
+	if (ldev->nb.notifier_call)
+		unregister_netdevice_notifier_net(&init_net, &ldev->nb);
+	mlx5_lag_mp_cleanup(ldev);
+	cancel_delayed_work_sync(&ldev->bond_work);
+	destroy_workqueue(ldev->wq);
+	mlx5_lag_mpesw_cleanup(ldev);
+	mutex_destroy(&ldev->lock);
+	kfree(ldev);
+}
+
+static void mlx5_ldev_put(struct mlx5_lag *ldev)
+{
+	kref_put(&ldev->ref, mlx5_ldev_free);
+}
+
+static void mlx5_ldev_get(struct mlx5_lag *ldev)
+{
+	kref_get(&ldev->ref);
+}
+
+static struct mlx5_lag *mlx5_lag_dev_alloc(struct mlx5_core_dev *dev)
+{
+	struct mlx5_lag *ldev;
+	int err;
+
+	ldev = kzalloc(sizeof(*ldev), GFP_KERNEL);
+	if (!ldev)
+		return NULL;
+
+	ldev->wq = create_singlethread_workqueue("mlx5_lag");
+	if (!ldev->wq) {
+		kfree(ldev);
+		return NULL;
+	}
+
+	kref_init(&ldev->ref);
+	mutex_init(&ldev->lock);
+	INIT_DELAYED_WORK(&ldev->bond_work, mlx5_do_bond_work);
+
+	ldev->nb.notifier_call = mlx5_lag_netdev_event;
+	if (register_netdevice_notifier_net(&init_net, &ldev->nb)) {
+		ldev->nb.notifier_call = NULL;
+		mlx5_core_err(dev, "Failed to register LAG netdev notifier\n");
+	}
+	ldev->mode = MLX5_LAG_MODE_NONE;
+
+	err = mlx5_lag_mp_init(ldev);
+	if (err)
+		mlx5_core_err(dev, "Failed to init multipath lag err=%d\n",
+			      err);
+
+	mlx5_lag_mpesw_init(ldev);
+	ldev->ports = MLX5_CAP_GEN(dev, num_lag_ports);
+	ldev->buckets = 1;
+
+	return ldev;
+}
+
+int mlx5_lag_dev_get_netdev_idx(struct mlx5_lag *ldev,
+				struct net_device *ndev)
+{
+	int i;
+
+	for (i = 0; i < ldev->ports; i++)
+		if (ldev->pf[i].netdev == ndev)
+			return i;
+
+	return -ENOENT;
+}
+
+static bool __mlx5_lag_is_roce(struct mlx5_lag *ldev)
+{
+	return ldev->mode == MLX5_LAG_MODE_ROCE;
+}
+
+static bool __mlx5_lag_is_sriov(struct mlx5_lag *ldev)
+{
+	return ldev->mode == MLX5_LAG_MODE_SRIOV;
+}
+
+/* Create a mapping between steering slots and active ports.
+ * As we have ldev->buckets slots per port first assume the native
+ * mapping should be used.
+ * If there are ports that are disabled fill the relevant slots
+ * with mapping that points to active ports.
+ */
+static void mlx5_infer_tx_affinity_mapping(struct lag_tracker *tracker,
+					   u8 num_ports,
+					   u8 buckets,
+					   u8 *ports)
+{
+	int disabled[MLX5_MAX_PORTS] = {};
+	int enabled[MLX5_MAX_PORTS] = {};
+	int disabled_ports_num = 0;
+	int enabled_ports_num = 0;
+	int idx;
+	u32 rand;
+	int i;
+	int j;
+
+	for (i = 0; i < num_ports; i++) {
+		if (tracker->netdev_state[i].tx_enabled &&
+		    tracker->netdev_state[i].link_up)
+			enabled[enabled_ports_num++] = i;
+		else
+			disabled[disabled_ports_num++] = i;
+	}
+
+	/* Use native mapping by default where each port's buckets
+	 * point the native port: 1 1 1 .. 1 2 2 2 ... 2 3 3 3 ... 3 etc
+	 */
+	for (i = 0; i < num_ports; i++)
+		for (j = 0; j < buckets; j++) {
+			idx = i * buckets + j;
+			ports[idx] = MLX5_LAG_EGRESS_PORT_1 + i;
+		}
+
+	/* If all ports are disabled/enabled keep native mapping */
+	if (enabled_ports_num == num_ports ||
+	    disabled_ports_num == num_ports)
+		return;
+
+	/* Go over the disabled ports and for each assign a random active port */
+	for (i = 0; i < disabled_ports_num; i++) {
+		for (j = 0; j < buckets; j++) {
+			get_random_bytes(&rand, 4);
+			ports[disabled[i] * buckets + j] = enabled[rand % enabled_ports_num] + 1;
+		}
+	}
+}
+
+static bool mlx5_lag_has_drop_rule(struct mlx5_lag *ldev)
+{
+	int i;
+
+	for (i = 0; i < ldev->ports; i++)
+		if (ldev->pf[i].has_drop)
+			return true;
+	return false;
+}
+
+static void mlx5_lag_drop_rule_cleanup(struct mlx5_lag *ldev)
+{
+	int i;
+
+	for (i = 0; i < ldev->ports; i++) {
+		if (!ldev->pf[i].has_drop)
+			continue;
+
+		mlx5_esw_acl_ingress_vport_drop_rule_destroy(ldev->pf[i].dev->priv.eswitch,
+							     MLX5_VPORT_UPLINK);
+		ldev->pf[i].has_drop = false;
+	}
+}
+
+static void mlx5_lag_drop_rule_setup(struct mlx5_lag *ldev,
+				     struct lag_tracker *tracker)
+{
+	u8 disabled_ports[MLX5_MAX_PORTS] = {};
+	struct mlx5_core_dev *dev;
+	int disabled_index;
+	int num_disabled;
+	int err;
+	int i;
+
+	/* First delete the current drop rule so there won't be any dropped
+	 * packets
+	 */
+	mlx5_lag_drop_rule_cleanup(ldev);
+
+	if (!ldev->tracker.has_inactive)
+		return;
+
+	mlx5_infer_tx_disabled(tracker, ldev->ports, disabled_ports, &num_disabled);
+
+	for (i = 0; i < num_disabled; i++) {
+		disabled_index = disabled_ports[i];
+		dev = ldev->pf[disabled_index].dev;
+		err = mlx5_esw_acl_ingress_vport_drop_rule_create(dev->priv.eswitch,
+								  MLX5_VPORT_UPLINK);
+		if (!err)
+			ldev->pf[disabled_index].has_drop = true;
+		else
+			mlx5_core_err(dev,
+				      "Failed to create lag drop rule, error: %d", err);
+	}
+}
+
+static int mlx5_cmd_modify_active_port(struct mlx5_core_dev *dev, u8 ports)
+{
+	u32 in[MLX5_ST_SZ_DW(modify_lag_in)] = {};
+	void *lag_ctx;
+
+	lag_ctx = MLX5_ADDR_OF(modify_lag_in, in, ctx);
+
+	MLX5_SET(modify_lag_in, in, opcode, MLX5_CMD_OP_MODIFY_LAG);
+	MLX5_SET(modify_lag_in, in, field_select, 0x2);
+
+	MLX5_SET(lagc, lag_ctx, active_port, ports);
+
+	return mlx5_cmd_exec_in(dev, modify_lag, in);
+}
+
+static int _mlx5_modify_lag(struct mlx5_lag *ldev, u8 *ports)
+{
+	struct mlx5_core_dev *dev0 = ldev->pf[MLX5_LAG_P1].dev;
+	u8 active_ports;
+	int ret;
+
+	if (test_bit(MLX5_LAG_MODE_FLAG_HASH_BASED, &ldev->mode_flags)) {
+		ret = mlx5_lag_port_sel_modify(ldev, ports);
+		if (ret ||
+		    !MLX5_CAP_PORT_SELECTION(dev0, port_select_flow_table_bypass))
+			return ret;
+
+		active_ports = lag_active_port_bits(ldev);
+
+		return mlx5_cmd_modify_active_port(dev0, active_ports);
+	}
+	return mlx5_cmd_modify_lag(dev0, ldev->ports, ports);
+}
+
+void mlx5_modify_lag(struct mlx5_lag *ldev,
+		     struct lag_tracker *tracker)
+{
+	u8 ports[MLX5_MAX_PORTS * MLX5_LAG_MAX_HASH_BUCKETS] = {};
+	struct mlx5_core_dev *dev0 = ldev->pf[MLX5_LAG_P1].dev;
+	int idx;
+	int err;
+	int i;
+	int j;
+
+	mlx5_infer_tx_affinity_mapping(tracker, ldev->ports, ldev->buckets, ports);
+
+	for (i = 0; i < ldev->ports; i++) {
+		for (j = 0; j < ldev->buckets; j++) {
+			idx = i * ldev->buckets + j;
+			if (ports[idx] == ldev->v2p_map[idx])
+				continue;
+			err = _mlx5_modify_lag(ldev, ports);
+			if (err) {
+				mlx5_core_err(dev0,
+					      "Failed to modify LAG (%d)\n",
+					      err);
+				return;
+			}
+			memcpy(ldev->v2p_map, ports, sizeof(ports));
+
+			mlx5_lag_print_mapping(dev0, ldev, tracker,
+					       ldev->mode_flags);
+			break;
+		}
+	}
+
+	if (tracker->tx_type == NETDEV_LAG_TX_TYPE_ACTIVEBACKUP &&
+	    !(ldev->mode == MLX5_LAG_MODE_ROCE))
+		mlx5_lag_drop_rule_setup(ldev, tracker);
+}
+
+static int mlx5_lag_set_port_sel_mode_roce(struct mlx5_lag *ldev,
+					   unsigned long *flags)
+{
+	struct mlx5_core_dev *dev0 = ldev->pf[MLX5_LAG_P1].dev;
+
+	if (!MLX5_CAP_PORT_SELECTION(dev0, port_select_flow_table)) {
+		if (ldev->ports > 2)
+			return -EINVAL;
+		return 0;
+	}
+
+	if (ldev->ports > 2)
+		ldev->buckets = MLX5_LAG_MAX_HASH_BUCKETS;
+
+	set_bit(MLX5_LAG_MODE_FLAG_HASH_BASED, flags);
+
+	return 0;
+}
+
+static void mlx5_lag_set_port_sel_mode_offloads(struct mlx5_lag *ldev,
+						struct lag_tracker *tracker,
+						enum mlx5_lag_mode mode,
+						unsigned long *flags)
+{
+	struct lag_func *dev0 = &ldev->pf[MLX5_LAG_P1];
+
+	if (mode == MLX5_LAG_MODE_MPESW)
+		return;
+
+	if (MLX5_CAP_PORT_SELECTION(dev0->dev, port_select_flow_table) &&
+	    tracker->tx_type == NETDEV_LAG_TX_TYPE_HASH)
+		set_bit(MLX5_LAG_MODE_FLAG_HASH_BASED, flags);
+}
+
+static int mlx5_lag_set_flags(struct mlx5_lag *ldev, enum mlx5_lag_mode mode,
+			      struct lag_tracker *tracker, bool shared_fdb,
+			      unsigned long *flags)
+{
+	bool roce_lag = mode == MLX5_LAG_MODE_ROCE;
+
+	*flags = 0;
+	if (shared_fdb) {
+		set_bit(MLX5_LAG_MODE_FLAG_SHARED_FDB, flags);
+		set_bit(MLX5_LAG_MODE_FLAG_FDB_SEL_MODE_NATIVE, flags);
+	}
+
+	if (mode == MLX5_LAG_MODE_MPESW)
+		set_bit(MLX5_LAG_MODE_FLAG_FDB_SEL_MODE_NATIVE, flags);
+
+	if (roce_lag)
+		return mlx5_lag_set_port_sel_mode_roce(ldev, flags);
+
+	mlx5_lag_set_port_sel_mode_offloads(ldev, tracker, mode, flags);
+	return 0;
+}
+
+char *mlx5_get_str_port_sel_mode(enum mlx5_lag_mode mode, unsigned long flags)
+{
+	int port_sel_mode = get_port_sel_mode(mode, flags);
+
+	switch (port_sel_mode) {
+	case MLX5_LAG_PORT_SELECT_MODE_QUEUE_AFFINITY: return "queue_affinity";
+	case MLX5_LAG_PORT_SELECT_MODE_PORT_SELECT_FT: return "hash";
+	case MLX5_LAG_PORT_SELECT_MODE_PORT_SELECT_MPESW: return "mpesw";
+	default: return "invalid";
+	}
+}
+
+static int mlx5_create_lag(struct mlx5_lag *ldev,
+			   struct lag_tracker *tracker,
+			   enum mlx5_lag_mode mode,
+			   unsigned long flags)
+{
+	bool shared_fdb = test_bit(MLX5_LAG_MODE_FLAG_SHARED_FDB, &flags);
+	struct mlx5_core_dev *dev0 = ldev->pf[MLX5_LAG_P1].dev;
+	struct mlx5_core_dev *dev1 = ldev->pf[MLX5_LAG_P2].dev;
+	u32 in[MLX5_ST_SZ_DW(destroy_lag_in)] = {};
+	int err;
+
+	if (tracker)
+		mlx5_lag_print_mapping(dev0, ldev, tracker, flags);
+	mlx5_core_info(dev0, "shared_fdb:%d mode:%s\n",
+		       shared_fdb, mlx5_get_str_port_sel_mode(mode, flags));
+
+	err = mlx5_cmd_create_lag(dev0, ldev->v2p_map, mode, flags);
+	if (err) {
+		mlx5_core_err(dev0,
+			      "Failed to create LAG (%d)\n",
+			      err);
+		return err;
+	}
+
+	if (shared_fdb) {
+		err = mlx5_eswitch_offloads_config_single_fdb(dev0->priv.eswitch,
+							      dev1->priv.eswitch);
+		if (err)
+			mlx5_core_err(dev0, "Can't enable single FDB mode\n");
+		else
+			mlx5_core_info(dev0, "Operation mode is single FDB\n");
+	}
+
+	if (err) {
+		MLX5_SET(destroy_lag_in, in, opcode, MLX5_CMD_OP_DESTROY_LAG);
+		if (mlx5_cmd_exec_in(dev0, destroy_lag, in))
+			mlx5_core_err(dev0,
+				      "Failed to deactivate RoCE LAG; driver restart required\n");
+	}
+
+	return err;
+}
+
+int mlx5_activate_lag(struct mlx5_lag *ldev,
+		      struct lag_tracker *tracker,
+		      enum mlx5_lag_mode mode,
+		      bool shared_fdb)
+{
+	bool roce_lag = mode == MLX5_LAG_MODE_ROCE;
+	struct mlx5_core_dev *dev0 = ldev->pf[MLX5_LAG_P1].dev;
+	unsigned long flags = 0;
+	int err;
+
+	err = mlx5_lag_set_flags(ldev, mode, tracker, shared_fdb, &flags);
+	if (err)
+		return err;
+
+	if (mode != MLX5_LAG_MODE_MPESW) {
+		mlx5_infer_tx_affinity_mapping(tracker, ldev->ports, ldev->buckets, ldev->v2p_map);
+		if (test_bit(MLX5_LAG_MODE_FLAG_HASH_BASED, &flags)) {
+			err = mlx5_lag_port_sel_create(ldev, tracker->hash_type,
+						       ldev->v2p_map);
+			if (err) {
+				mlx5_core_err(dev0,
+					      "Failed to create LAG port selection(%d)\n",
+					      err);
+				return err;
+			}
+		}
+	}
+
+	err = mlx5_create_lag(ldev, tracker, mode, flags);
+	if (err) {
+		if (test_bit(MLX5_LAG_MODE_FLAG_HASH_BASED, &flags))
+			mlx5_lag_port_sel_destroy(ldev);
+		if (roce_lag)
+			mlx5_core_err(dev0,
+				      "Failed to activate RoCE LAG\n");
+		else
+			mlx5_core_err(dev0,
+				      "Failed to activate VF LAG\n"
+				      "Make sure all VFs are unbound prior to VF LAG activation or deactivation\n");
+		return err;
+	}
+
+	if (tracker && tracker->tx_type == NETDEV_LAG_TX_TYPE_ACTIVEBACKUP &&
+	    !roce_lag)
+		mlx5_lag_drop_rule_setup(ldev, tracker);
+
+	ldev->mode = mode;
+	ldev->mode_flags = flags;
+	return 0;
+}
+
+static int mlx5_deactivate_lag(struct mlx5_lag *ldev)
+{
+	struct mlx5_core_dev *dev0 = ldev->pf[MLX5_LAG_P1].dev;
+	struct mlx5_core_dev *dev1 = ldev->pf[MLX5_LAG_P2].dev;
+	u32 in[MLX5_ST_SZ_DW(destroy_lag_in)] = {};
+	bool roce_lag = __mlx5_lag_is_roce(ldev);
+	unsigned long flags = ldev->mode_flags;
+	int err;
+
+	ldev->mode = MLX5_LAG_MODE_NONE;
+	ldev->mode_flags = 0;
+	mlx5_lag_mp_reset(ldev);
+
+	if (test_bit(MLX5_LAG_MODE_FLAG_SHARED_FDB, &flags)) {
+		mlx5_eswitch_offloads_destroy_single_fdb(dev0->priv.eswitch,
+							 dev1->priv.eswitch);
+		clear_bit(MLX5_LAG_MODE_FLAG_SHARED_FDB, &flags);
+	}
+
+	MLX5_SET(destroy_lag_in, in, opcode, MLX5_CMD_OP_DESTROY_LAG);
+	err = mlx5_cmd_exec_in(dev0, destroy_lag, in);
+	if (err) {
+		if (roce_lag) {
+			mlx5_core_err(dev0,
+				      "Failed to deactivate RoCE LAG; driver restart required\n");
+		} else {
+			mlx5_core_err(dev0,
+				      "Failed to deactivate VF LAG; driver restart required\n"
+				      "Make sure all VFs are unbound prior to VF LAG activation or deactivation\n");
+		}
+		return err;
+	}
+
+	if (test_bit(MLX5_LAG_MODE_FLAG_HASH_BASED, &flags))
+		mlx5_lag_port_sel_destroy(ldev);
+	if (mlx5_lag_has_drop_rule(ldev))
+		mlx5_lag_drop_rule_cleanup(ldev);
+
+	return 0;
+}
+
+#define MLX5_LAG_OFFLOADS_SUPPORTED_PORTS 2
+static bool mlx5_lag_check_prereq(struct mlx5_lag *ldev)
+{
+#ifdef CONFIG_MLX5_ESWITCH
+	struct mlx5_core_dev *dev;
+	u8 mode;
+#endif
+	int i;
+
+	for (i = 0; i < ldev->ports; i++)
+		if (!ldev->pf[i].dev)
+			return false;
+
+#ifdef CONFIG_MLX5_ESWITCH
+	for (i = 0; i < ldev->ports; i++) {
+		dev = ldev->pf[i].dev;
+		if (mlx5_eswitch_num_vfs(dev->priv.eswitch) && !is_mdev_switchdev_mode(dev))
+			return false;
+	}
+
+	dev = ldev->pf[MLX5_LAG_P1].dev;
+	mode = mlx5_eswitch_mode(dev);
+	for (i = 0; i < ldev->ports; i++)
+		if (mlx5_eswitch_mode(ldev->pf[i].dev) != mode)
+			return false;
+
+	if (mode == MLX5_ESWITCH_OFFLOADS && ldev->ports != MLX5_LAG_OFFLOADS_SUPPORTED_PORTS)
+		return false;
+#else
+	for (i = 0; i < ldev->ports; i++)
+		if (mlx5_sriov_is_enabled(ldev->pf[i].dev))
+			return false;
+#endif
+	return true;
+}
+
+static void mlx5_lag_add_devices(struct mlx5_lag *ldev)
+{
+	int i;
+
+	for (i = 0; i < ldev->ports; i++) {
+		if (!ldev->pf[i].dev)
+			continue;
+
+		if (ldev->pf[i].dev->priv.flags &
+		    MLX5_PRIV_FLAGS_DISABLE_ALL_ADEV)
+			continue;
+
+		ldev->pf[i].dev->priv.flags &= ~MLX5_PRIV_FLAGS_DISABLE_IB_ADEV;
+		mlx5_rescan_drivers_locked(ldev->pf[i].dev);
+	}
+}
+
+static void mlx5_lag_remove_devices(struct mlx5_lag *ldev)
+{
+	int i;
+
+	for (i = 0; i < ldev->ports; i++) {
+		if (!ldev->pf[i].dev)
+			continue;
+
+		if (ldev->pf[i].dev->priv.flags &
+		    MLX5_PRIV_FLAGS_DISABLE_ALL_ADEV)
+			continue;
+
+		ldev->pf[i].dev->priv.flags |= MLX5_PRIV_FLAGS_DISABLE_IB_ADEV;
+		mlx5_rescan_drivers_locked(ldev->pf[i].dev);
+	}
+}
+
+void mlx5_disable_lag(struct mlx5_lag *ldev)
+{
+	bool shared_fdb = test_bit(MLX5_LAG_MODE_FLAG_SHARED_FDB, &ldev->mode_flags);
+	struct mlx5_core_dev *dev0 = ldev->pf[MLX5_LAG_P1].dev;
+	struct mlx5_core_dev *dev1 = ldev->pf[MLX5_LAG_P2].dev;
+	bool roce_lag;
+	int err;
+	int i;
+
+	roce_lag = __mlx5_lag_is_roce(ldev);
+
+	if (shared_fdb) {
+		mlx5_lag_remove_devices(ldev);
+	} else if (roce_lag) {
+		if (!(dev0->priv.flags & MLX5_PRIV_FLAGS_DISABLE_ALL_ADEV)) {
+			dev0->priv.flags |= MLX5_PRIV_FLAGS_DISABLE_IB_ADEV;
+			mlx5_rescan_drivers_locked(dev0);
+		}
+		for (i = 1; i < ldev->ports; i++)
+			mlx5_nic_vport_disable_roce(ldev->pf[i].dev);
+	}
+
+	err = mlx5_deactivate_lag(ldev);
+	if (err)
+		return;
+
+	if (shared_fdb || roce_lag)
+		mlx5_lag_add_devices(ldev);
+
+	if (shared_fdb) {
+		if (!(dev0->priv.flags & MLX5_PRIV_FLAGS_DISABLE_ALL_ADEV))
+			mlx5_eswitch_reload_reps(dev0->priv.eswitch);
+		if (!(dev1->priv.flags & MLX5_PRIV_FLAGS_DISABLE_ALL_ADEV))
+			mlx5_eswitch_reload_reps(dev1->priv.eswitch);
+	}
+}
+
+bool mlx5_shared_fdb_supported(struct mlx5_lag *ldev)
+{
+	struct mlx5_core_dev *dev0 = ldev->pf[MLX5_LAG_P1].dev;
+	struct mlx5_core_dev *dev1 = ldev->pf[MLX5_LAG_P2].dev;
+
+	if (is_mdev_switchdev_mode(dev0) &&
+	    is_mdev_switchdev_mode(dev1) &&
+	    mlx5_eswitch_vport_match_metadata_enabled(dev0->priv.eswitch) &&
+	    mlx5_eswitch_vport_match_metadata_enabled(dev1->priv.eswitch) &&
+	    mlx5_devcom_is_paired(dev0->priv.devcom,
+				  MLX5_DEVCOM_ESW_OFFLOADS) &&
+	    MLX5_CAP_GEN(dev1, lag_native_fdb_selection) &&
+	    MLX5_CAP_ESW(dev1, root_ft_on_other_esw) &&
+	    MLX5_CAP_ESW(dev0, esw_shared_ingress_acl))
+		return true;
+
+	return false;
+}
+
+static bool mlx5_lag_is_roce_lag(struct mlx5_lag *ldev)
+{
+	bool roce_lag = true;
+	int i;
+
+	for (i = 0; i < ldev->ports; i++)
+		roce_lag = roce_lag && !mlx5_sriov_is_enabled(ldev->pf[i].dev);
+
+#ifdef CONFIG_MLX5_ESWITCH
+	for (i = 0; i < ldev->ports; i++)
+		roce_lag = roce_lag && is_mdev_legacy_mode(ldev->pf[i].dev);
+#endif
+
+	return roce_lag;
+}
+
+static bool mlx5_lag_should_modify_lag(struct mlx5_lag *ldev, bool do_bond)
+{
+	return do_bond && __mlx5_lag_is_active(ldev) &&
+	       ldev->mode != MLX5_LAG_MODE_MPESW;
+}
+
+static bool mlx5_lag_should_disable_lag(struct mlx5_lag *ldev, bool do_bond)
+{
+	return !do_bond && __mlx5_lag_is_active(ldev) &&
+	       ldev->mode != MLX5_LAG_MODE_MPESW;
+}
+
+static void mlx5_do_bond(struct mlx5_lag *ldev)
+{
+	struct mlx5_core_dev *dev0 = ldev->pf[MLX5_LAG_P1].dev;
+	struct mlx5_core_dev *dev1 = ldev->pf[MLX5_LAG_P2].dev;
+	struct lag_tracker tracker = { };
+	bool do_bond, roce_lag;
+	int err;
+	int i;
+
+	if (!mlx5_lag_is_ready(ldev)) {
+		do_bond = false;
+	} else {
+		/* VF LAG is in multipath mode, ignore bond change requests */
+		if (mlx5_lag_is_multipath(dev0))
+			return;
+
+		tracker = ldev->tracker;
+
+		do_bond = tracker.is_bonded && mlx5_lag_check_prereq(ldev);
+	}
+
+	if (do_bond && !__mlx5_lag_is_active(ldev)) {
+		bool shared_fdb = mlx5_shared_fdb_supported(ldev);
+
+		roce_lag = mlx5_lag_is_roce_lag(ldev);
+
+		if (shared_fdb || roce_lag)
+			mlx5_lag_remove_devices(ldev);
+
+		err = mlx5_activate_lag(ldev, &tracker,
+					roce_lag ? MLX5_LAG_MODE_ROCE :
+						   MLX5_LAG_MODE_SRIOV,
+					shared_fdb);
+		if (err) {
+			if (shared_fdb || roce_lag)
+				mlx5_lag_add_devices(ldev);
+
+			return;
+		} else if (roce_lag) {
+			dev0->priv.flags &= ~MLX5_PRIV_FLAGS_DISABLE_IB_ADEV;
+			mlx5_rescan_drivers_locked(dev0);
+			for (i = 1; i < ldev->ports; i++)
+				mlx5_nic_vport_enable_roce(ldev->pf[i].dev);
+		} else if (shared_fdb) {
+			dev0->priv.flags &= ~MLX5_PRIV_FLAGS_DISABLE_IB_ADEV;
+			mlx5_rescan_drivers_locked(dev0);
+
+			err = mlx5_eswitch_reload_reps(dev0->priv.eswitch);
+			if (!err)
+				err = mlx5_eswitch_reload_reps(dev1->priv.eswitch);
+
+			if (err) {
+				dev0->priv.flags |= MLX5_PRIV_FLAGS_DISABLE_IB_ADEV;
+				mlx5_rescan_drivers_locked(dev0);
+				mlx5_deactivate_lag(ldev);
+				mlx5_lag_add_devices(ldev);
+				mlx5_eswitch_reload_reps(dev0->priv.eswitch);
+				mlx5_eswitch_reload_reps(dev1->priv.eswitch);
+				mlx5_core_err(dev0, "Failed to enable lag\n");
+				return;
+			}
+		}
+	} else if (mlx5_lag_should_modify_lag(ldev, do_bond)) {
+		mlx5_modify_lag(ldev, &tracker);
+	} else if (mlx5_lag_should_disable_lag(ldev, do_bond)) {
+		mlx5_disable_lag(ldev);
+	}
+}
+
+static void mlx5_queue_bond_work(struct mlx5_lag *ldev, unsigned long delay)
+{
+	queue_delayed_work(ldev->wq, &ldev->bond_work, delay);
+}
+
+static void mlx5_do_bond_work(struct work_struct *work)
+{
+	struct delayed_work *delayed_work = to_delayed_work(work);
+	struct mlx5_lag *ldev = container_of(delayed_work, struct mlx5_lag,
+					     bond_work);
+	int status;
+
+	status = mlx5_dev_list_trylock();
+	if (!status) {
+		mlx5_queue_bond_work(ldev, HZ);
+		return;
+	}
+
+	mutex_lock(&ldev->lock);
+	if (ldev->mode_changes_in_progress) {
+		mutex_unlock(&ldev->lock);
+		mlx5_dev_list_unlock();
+		mlx5_queue_bond_work(ldev, HZ);
+		return;
+	}
+
+	mlx5_do_bond(ldev);
+	mutex_unlock(&ldev->lock);
+	mlx5_dev_list_unlock();
+}
+
+static int mlx5_handle_changeupper_event(struct mlx5_lag *ldev,
+					 struct lag_tracker *tracker,
+					 struct netdev_notifier_changeupper_info *info)
+{
+	struct net_device *upper = info->upper_dev, *ndev_tmp;
+	struct netdev_lag_upper_info *lag_upper_info = NULL;
+	bool is_bonded, is_in_lag, mode_supported;
+	bool has_inactive = 0;
+	struct slave *slave;
+	u8 bond_status = 0;
+	int num_slaves = 0;
+	int changed = 0;
+	int idx;
+
+	if (!netif_is_lag_master(upper))
+		return 0;
+
+	if (info->linking)
+		lag_upper_info = info->upper_info;
+
+	/* The event may still be of interest if the slave does not belong to
+	 * us, but is enslaved to a master which has one or more of our netdevs
+	 * as slaves (e.g., if a new slave is added to a master that bonds two
+	 * of our netdevs, we should unbond).
+	 */
+	rcu_read_lock();
+	for_each_netdev_in_bond_rcu(upper, ndev_tmp) {
+		idx = mlx5_lag_dev_get_netdev_idx(ldev, ndev_tmp);
+		if (idx >= 0) {
+			slave = bond_slave_get_rcu(ndev_tmp);
+			if (slave)
+				has_inactive |= bond_is_slave_inactive(slave);
+			bond_status |= (1 << idx);
+		}
+
+		num_slaves++;
+	}
+	rcu_read_unlock();
+
+	/* None of this lagdev's netdevs are slaves of this master. */
+	if (!(bond_status & GENMASK(ldev->ports - 1, 0)))
+		return 0;
+
+	if (lag_upper_info) {
+		tracker->tx_type = lag_upper_info->tx_type;
+		tracker->hash_type = lag_upper_info->hash_type;
+	}
+
+	tracker->has_inactive = has_inactive;
+	/* Determine bonding status:
+	 * A device is considered bonded if both its physical ports are slaves
+	 * of the same lag master, and only them.
+	 */
+	is_in_lag = num_slaves == ldev->ports &&
+		bond_status == GENMASK(ldev->ports - 1, 0);
+
+	/* Lag mode must be activebackup or hash. */
+	mode_supported = tracker->tx_type == NETDEV_LAG_TX_TYPE_ACTIVEBACKUP ||
+			 tracker->tx_type == NETDEV_LAG_TX_TYPE_HASH;
+
+	is_bonded = is_in_lag && mode_supported;
+	if (tracker->is_bonded != is_bonded) {
+		tracker->is_bonded = is_bonded;
+		changed = 1;
+	}
+
+	if (!is_in_lag)
+		return changed;
+
+	if (!mlx5_lag_is_ready(ldev))
+		NL_SET_ERR_MSG_MOD(info->info.extack,
+				   "Can't activate LAG offload, PF is configured with more than 64 VFs");
+	else if (!mode_supported)
+		NL_SET_ERR_MSG_MOD(info->info.extack,
+				   "Can't activate LAG offload, TX type isn't supported");
+
+	return changed;
+}
+
+static int mlx5_handle_changelowerstate_event(struct mlx5_lag *ldev,
+					      struct lag_tracker *tracker,
+					      struct net_device *ndev,
+					      struct netdev_notifier_changelowerstate_info *info)
+{
+	struct netdev_lag_lower_state_info *lag_lower_info;
+	int idx;
+
+	if (!netif_is_lag_port(ndev))
+		return 0;
+
+	idx = mlx5_lag_dev_get_netdev_idx(ldev, ndev);
+	if (idx < 0)
+		return 0;
+
+	/* This information is used to determine virtual to physical
+	 * port mapping.
+	 */
+	lag_lower_info = info->lower_state_info;
+	if (!lag_lower_info)
+		return 0;
+
+	tracker->netdev_state[idx] = *lag_lower_info;
+
+	return 1;
+}
+
+static int mlx5_handle_changeinfodata_event(struct mlx5_lag *ldev,
+					    struct lag_tracker *tracker,
+					    struct net_device *ndev)
+{
+	struct net_device *ndev_tmp;
+	struct slave *slave;
+	bool has_inactive = 0;
+	int idx;
+
+	if (!netif_is_lag_master(ndev))
+		return 0;
+
+	rcu_read_lock();
+	for_each_netdev_in_bond_rcu(ndev, ndev_tmp) {
+		idx = mlx5_lag_dev_get_netdev_idx(ldev, ndev_tmp);
+		if (idx < 0)
+			continue;
+
+		slave = bond_slave_get_rcu(ndev_tmp);
+		if (slave)
+			has_inactive |= bond_is_slave_inactive(slave);
+	}
+	rcu_read_unlock();
+
+	if (tracker->has_inactive == has_inactive)
+		return 0;
+
+	tracker->has_inactive = has_inactive;
+
+	return 1;
+}
+
+/* this handler is always registered to netdev events */
+static int mlx5_lag_netdev_event(struct notifier_block *this,
+				 unsigned long event, void *ptr)
+{
+	struct net_device *ndev = netdev_notifier_info_to_dev(ptr);
+	struct lag_tracker tracker;
+	struct mlx5_lag *ldev;
+	int changed = 0;
+
+	if (event != NETDEV_CHANGEUPPER &&
+	    event != NETDEV_CHANGELOWERSTATE &&
+	    event != NETDEV_CHANGEINFODATA)
+		return NOTIFY_DONE;
+
+	ldev    = container_of(this, struct mlx5_lag, nb);
+
+	tracker = ldev->tracker;
+
+	switch (event) {
+	case NETDEV_CHANGEUPPER:
+		changed = mlx5_handle_changeupper_event(ldev, &tracker, ptr);
+		break;
+	case NETDEV_CHANGELOWERSTATE:
+		changed = mlx5_handle_changelowerstate_event(ldev, &tracker,
+							     ndev, ptr);
+		break;
+	case NETDEV_CHANGEINFODATA:
+		changed = mlx5_handle_changeinfodata_event(ldev, &tracker, ndev);
+		break;
+	}
+
+	ldev->tracker = tracker;
+
+	if (changed)
+		mlx5_queue_bond_work(ldev, 0);
+
+	return NOTIFY_DONE;
+}
+
+static void mlx5_ldev_add_netdev(struct mlx5_lag *ldev,
+				 struct mlx5_core_dev *dev,
+				 struct net_device *netdev)
+{
+	unsigned int fn = mlx5_get_dev_index(dev);
+	unsigned long flags;
+
+	if (fn >= ldev->ports)
+		return;
+
+	spin_lock_irqsave(&lag_lock, flags);
+	ldev->pf[fn].netdev = netdev;
+	ldev->tracker.netdev_state[fn].link_up = 0;
+	ldev->tracker.netdev_state[fn].tx_enabled = 0;
+	spin_unlock_irqrestore(&lag_lock, flags);
+}
+
+static void mlx5_ldev_remove_netdev(struct mlx5_lag *ldev,
+				    struct net_device *netdev)
+{
+	unsigned long flags;
+	int i;
+
+	spin_lock_irqsave(&lag_lock, flags);
+	for (i = 0; i < ldev->ports; i++) {
+		if (ldev->pf[i].netdev == netdev) {
+			ldev->pf[i].netdev = NULL;
+			break;
+		}
+	}
+	spin_unlock_irqrestore(&lag_lock, flags);
+}
+
+static void mlx5_ldev_add_mdev(struct mlx5_lag *ldev,
+			       struct mlx5_core_dev *dev)
+{
+	unsigned int fn = mlx5_get_dev_index(dev);
+
+	if (fn >= ldev->ports)
+		return;
+
+	ldev->pf[fn].dev = dev;
+	dev->priv.lag = ldev;
+}
+
+static void mlx5_ldev_remove_mdev(struct mlx5_lag *ldev,
+				  struct mlx5_core_dev *dev)
+{
+	int i;
+
+	for (i = 0; i < ldev->ports; i++)
+		if (ldev->pf[i].dev == dev)
+			break;
+
+	if (i == ldev->ports)
+		return;
+
+	ldev->pf[i].dev = NULL;
+	dev->priv.lag = NULL;
+}
+
+/* Must be called with intf_mutex held */
+static int __mlx5_lag_dev_add_mdev(struct mlx5_core_dev *dev)
+{
+	struct mlx5_lag *ldev = NULL;
+	struct mlx5_core_dev *tmp_dev;
+
+	tmp_dev = mlx5_get_next_phys_dev_lag(dev);
+	if (tmp_dev)
+		ldev = tmp_dev->priv.lag;
+
+	if (!ldev) {
+		ldev = mlx5_lag_dev_alloc(dev);
+		if (!ldev) {
+			mlx5_core_err(dev, "Failed to alloc lag dev\n");
+			return 0;
+		}
+		mlx5_ldev_add_mdev(ldev, dev);
+		return 0;
+	}
+
+	mutex_lock(&ldev->lock);
+	if (ldev->mode_changes_in_progress) {
+		mutex_unlock(&ldev->lock);
+		return -EAGAIN;
+	}
+	mlx5_ldev_get(ldev);
+	mlx5_ldev_add_mdev(ldev, dev);
+	mutex_unlock(&ldev->lock);
+
+	return 0;
+}
+
+void mlx5_lag_remove_mdev(struct mlx5_core_dev *dev)
+{
+	struct mlx5_lag *ldev;
+
+	ldev = mlx5_lag_dev(dev);
+	if (!ldev)
+		return;
+
+	/* mdev is being removed, might as well remove debugfs
+	 * as early as possible.
+	 */
+	mlx5_ldev_remove_debugfs(dev->priv.dbg.lag_debugfs);
+recheck:
+	mutex_lock(&ldev->lock);
+	if (ldev->mode_changes_in_progress) {
+		mutex_unlock(&ldev->lock);
+		msleep(100);
+		goto recheck;
+	}
+	mlx5_ldev_remove_mdev(ldev, dev);
+	mutex_unlock(&ldev->lock);
+	mlx5_ldev_put(ldev);
+}
+
+void mlx5_lag_add_mdev(struct mlx5_core_dev *dev)
+{
+	int err;
+
+	if (!MLX5_CAP_GEN(dev, vport_group_manager) ||
+	    !MLX5_CAP_GEN(dev, lag_master) ||
+	    (MLX5_CAP_GEN(dev, num_lag_ports) > MLX5_MAX_PORTS ||
+	     MLX5_CAP_GEN(dev, num_lag_ports) <= 1))
+		return;
+
+recheck:
+	mlx5_dev_list_lock();
+	err = __mlx5_lag_dev_add_mdev(dev);
+	mlx5_dev_list_unlock();
+
+	if (err) {
+		msleep(100);
+		goto recheck;
+	}
+	mlx5_ldev_add_debugfs(dev);
+}
+
+void mlx5_lag_remove_netdev(struct mlx5_core_dev *dev,
+			    struct net_device *netdev)
+{
+	struct mlx5_lag *ldev;
+	bool lag_is_active;
+
+	ldev = mlx5_lag_dev(dev);
+	if (!ldev)
+		return;
+
+	mutex_lock(&ldev->lock);
+	mlx5_ldev_remove_netdev(ldev, netdev);
+	clear_bit(MLX5_LAG_FLAG_NDEVS_READY, &ldev->state_flags);
+
+	lag_is_active = __mlx5_lag_is_active(ldev);
+	mutex_unlock(&ldev->lock);
+
+	if (lag_is_active)
+		mlx5_queue_bond_work(ldev, 0);
+}
+
+void mlx5_lag_add_netdev(struct mlx5_core_dev *dev,
+			 struct net_device *netdev)
+{
+	struct mlx5_lag *ldev;
+	int i;
+
+	ldev = mlx5_lag_dev(dev);
+	if (!ldev)
+		return;
+
+	mutex_lock(&ldev->lock);
+	mlx5_ldev_add_netdev(ldev, dev, netdev);
+
+	for (i = 0; i < ldev->ports; i++)
+		if (!ldev->pf[i].netdev)
+			break;
+
+	if (i >= ldev->ports)
+		set_bit(MLX5_LAG_FLAG_NDEVS_READY, &ldev->state_flags);
+	mutex_unlock(&ldev->lock);
+	mlx5_queue_bond_work(ldev, 0);
+}
+
+bool mlx5_lag_is_roce(struct mlx5_core_dev *dev)
+{
+	struct mlx5_lag *ldev;
+	unsigned long flags;
+	bool res;
+
+	spin_lock_irqsave(&lag_lock, flags);
+	ldev = mlx5_lag_dev(dev);
+	res  = ldev && __mlx5_lag_is_roce(ldev);
+	spin_unlock_irqrestore(&lag_lock, flags);
+
+	return res;
+}
+EXPORT_SYMBOL(mlx5_lag_is_roce);
+
+bool mlx5_lag_is_active(struct mlx5_core_dev *dev)
+{
+	struct mlx5_lag *ldev;
+	unsigned long flags;
+	bool res;
+
+	spin_lock_irqsave(&lag_lock, flags);
+	ldev = mlx5_lag_dev(dev);
+	res  = ldev && __mlx5_lag_is_active(ldev);
+	spin_unlock_irqrestore(&lag_lock, flags);
+
+	return res;
+}
+EXPORT_SYMBOL(mlx5_lag_is_active);
+
+bool mlx5_lag_mode_is_hash(struct mlx5_core_dev *dev)
+{
+	struct mlx5_lag *ldev;
+	unsigned long flags;
+	bool res = 0;
+
+	spin_lock_irqsave(&lag_lock, flags);
+	ldev = mlx5_lag_dev(dev);
+	if (ldev)
+		res = test_bit(MLX5_LAG_MODE_FLAG_HASH_BASED, &ldev->mode_flags);
+	spin_unlock_irqrestore(&lag_lock, flags);
+
+	return res;
+}
+EXPORT_SYMBOL(mlx5_lag_mode_is_hash);
+
+bool mlx5_lag_is_master(struct mlx5_core_dev *dev)
+{
+	struct mlx5_lag *ldev;
+	unsigned long flags;
+	bool res;
+
+	spin_lock_irqsave(&lag_lock, flags);
+	ldev = mlx5_lag_dev(dev);
+	res = ldev && __mlx5_lag_is_active(ldev) &&
+		dev == ldev->pf[MLX5_LAG_P1].dev;
+	spin_unlock_irqrestore(&lag_lock, flags);
+
+	return res;
+}
+EXPORT_SYMBOL(mlx5_lag_is_master);
+
+bool mlx5_lag_is_sriov(struct mlx5_core_dev *dev)
+{
+	struct mlx5_lag *ldev;
+	unsigned long flags;
+	bool res;
+
+	spin_lock_irqsave(&lag_lock, flags);
+	ldev = mlx5_lag_dev(dev);
+	res  = ldev && __mlx5_lag_is_sriov(ldev);
+	spin_unlock_irqrestore(&lag_lock, flags);
+
+	return res;
+}
+EXPORT_SYMBOL(mlx5_lag_is_sriov);
+
+bool mlx5_lag_is_shared_fdb(struct mlx5_core_dev *dev)
+{
+	struct mlx5_lag *ldev;
+	unsigned long flags;
+	bool res;
+
+	spin_lock_irqsave(&lag_lock, flags);
+	ldev = mlx5_lag_dev(dev);
+	res = ldev && __mlx5_lag_is_sriov(ldev) &&
+	      test_bit(MLX5_LAG_MODE_FLAG_SHARED_FDB, &ldev->mode_flags);
+	spin_unlock_irqrestore(&lag_lock, flags);
+
+	return res;
+}
+EXPORT_SYMBOL(mlx5_lag_is_shared_fdb);
+
+void mlx5_lag_disable_change(struct mlx5_core_dev *dev)
+{
+	struct mlx5_lag *ldev;
+
+	ldev = mlx5_lag_dev(dev);
+	if (!ldev)
+		return;
+
+	mlx5_dev_list_lock();
+	mutex_lock(&ldev->lock);
+
+	ldev->mode_changes_in_progress++;
+	if (__mlx5_lag_is_active(ldev))
+		mlx5_disable_lag(ldev);
+
+	mutex_unlock(&ldev->lock);
+	mlx5_dev_list_unlock();
+}
+
+void mlx5_lag_enable_change(struct mlx5_core_dev *dev)
+{
+	struct mlx5_lag *ldev;
+
+	ldev = mlx5_lag_dev(dev);
+	if (!ldev)
+		return;
+
+	mutex_lock(&ldev->lock);
+	ldev->mode_changes_in_progress--;
+	mutex_unlock(&ldev->lock);
+	mlx5_queue_bond_work(ldev, 0);
+}
+
+struct net_device *mlx5_lag_get_roce_netdev(struct mlx5_core_dev *dev)
+{
+	struct net_device *ndev = NULL;
+	struct mlx5_lag *ldev;
+	unsigned long flags;
+	int i;
+
+	spin_lock_irqsave(&lag_lock, flags);
+	ldev = mlx5_lag_dev(dev);
+
+	if (!(ldev && __mlx5_lag_is_roce(ldev)))
+		goto unlock;
+
+	if (ldev->tracker.tx_type == NETDEV_LAG_TX_TYPE_ACTIVEBACKUP) {
+		for (i = 0; i < ldev->ports; i++)
+			if (ldev->tracker.netdev_state[i].tx_enabled)
+				ndev = ldev->pf[i].netdev;
+		if (!ndev)
+			ndev = ldev->pf[ldev->ports - 1].netdev;
+	} else {
+		ndev = ldev->pf[MLX5_LAG_P1].netdev;
+	}
+	if (ndev)
+		dev_hold(ndev);
+
+unlock:
+	spin_unlock_irqrestore(&lag_lock, flags);
+
+	return ndev;
+}
+EXPORT_SYMBOL(mlx5_lag_get_roce_netdev);
+
+u8 mlx5_lag_get_slave_port(struct mlx5_core_dev *dev,
+			   struct net_device *slave)
+{
+	struct mlx5_lag *ldev;
+	unsigned long flags;
+	u8 port = 0;
+	int i;
+
+	spin_lock_irqsave(&lag_lock, flags);
+	ldev = mlx5_lag_dev(dev);
+	if (!(ldev && __mlx5_lag_is_roce(ldev)))
+		goto unlock;
+
+	for (i = 0; i < ldev->ports; i++) {
+		if (ldev->pf[MLX5_LAG_P1].netdev == slave) {
+			port = i;
+			break;
+		}
+	}
+
+	port = ldev->v2p_map[port * ldev->buckets];
+
+unlock:
+	spin_unlock_irqrestore(&lag_lock, flags);
+	return port;
+}
+EXPORT_SYMBOL(mlx5_lag_get_slave_port);
+
+u8 mlx5_lag_get_num_ports(struct mlx5_core_dev *dev)
+{
+	struct mlx5_lag *ldev;
+
+	ldev = mlx5_lag_dev(dev);
+	if (!ldev)
+		return 0;
+
+	return ldev->ports;
+}
+EXPORT_SYMBOL(mlx5_lag_get_num_ports);
+
+struct mlx5_core_dev *mlx5_lag_get_peer_mdev(struct mlx5_core_dev *dev)
+{
+	struct mlx5_core_dev *peer_dev = NULL;
+	struct mlx5_lag *ldev;
+	unsigned long flags;
+
+	spin_lock_irqsave(&lag_lock, flags);
+	ldev = mlx5_lag_dev(dev);
+	if (!ldev)
+		goto unlock;
+
+	peer_dev = ldev->pf[MLX5_LAG_P1].dev == dev ?
+			   ldev->pf[MLX5_LAG_P2].dev :
+			   ldev->pf[MLX5_LAG_P1].dev;
+
+unlock:
+	spin_unlock_irqrestore(&lag_lock, flags);
+	return peer_dev;
+}
+EXPORT_SYMBOL(mlx5_lag_get_peer_mdev);
+
+int mlx5_lag_query_cong_counters(struct mlx5_core_dev *dev,
+				 u64 *values,
+				 int num_counters,
+				 size_t *offsets)
+{
+	int outlen = MLX5_ST_SZ_BYTES(query_cong_statistics_out);
+	struct mlx5_core_dev **mdev;
+	struct mlx5_lag *ldev;
+	unsigned long flags;
+	int num_ports;
+	int ret, i, j;
+	void *out;
+
+	out = kvzalloc(outlen, GFP_KERNEL);
+	if (!out)
+		return -ENOMEM;
+
+	mdev = kvzalloc(sizeof(mdev[0]) * MLX5_MAX_PORTS, GFP_KERNEL);
+	if (!mdev) {
+		ret = -ENOMEM;
+		goto free_out;
+	}
+
+	memset(values, 0, sizeof(*values) * num_counters);
+
+	spin_lock_irqsave(&lag_lock, flags);
+	ldev = mlx5_lag_dev(dev);
+	if (ldev && __mlx5_lag_is_active(ldev)) {
+		num_ports = ldev->ports;
+		for (i = 0; i < ldev->ports; i++)
+			mdev[i] = ldev->pf[i].dev;
+	} else {
+		num_ports = 1;
+		mdev[MLX5_LAG_P1] = dev;
+	}
+	spin_unlock_irqrestore(&lag_lock, flags);
+
+	for (i = 0; i < num_ports; ++i) {
+		u32 in[MLX5_ST_SZ_DW(query_cong_statistics_in)] = {};
+
+		MLX5_SET(query_cong_statistics_in, in, opcode,
+			 MLX5_CMD_OP_QUERY_CONG_STATISTICS);
+		ret = mlx5_cmd_exec_inout(mdev[i], query_cong_statistics, in,
+					  out);
+		if (ret)
+			goto free_mdev;
+
+		for (j = 0; j < num_counters; ++j)
+			values[j] += be64_to_cpup((__be64 *)(out + offsets[j]));
+	}
+
+free_mdev:
+	kvfree(mdev);
+free_out:
+	kvfree(out);
+	return ret;
+}
+EXPORT_SYMBOL(mlx5_lag_query_cong_counters);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.h b/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.h
new file mode 100644
index 000000000..f30ac2de6
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.h
@@ -0,0 +1,138 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2019 Mellanox Technologies. */
+
+#ifndef __MLX5_LAG_H__
+#define __MLX5_LAG_H__
+
+#include <linux/debugfs.h>
+
+#define MLX5_LAG_MAX_HASH_BUCKETS 16
+#include "mlx5_core.h"
+#include "mp.h"
+#include "port_sel.h"
+#include "mpesw.h"
+
+enum {
+	MLX5_LAG_P1,
+	MLX5_LAG_P2,
+};
+
+enum {
+	MLX5_LAG_FLAG_NDEVS_READY,
+};
+
+enum {
+	MLX5_LAG_MODE_FLAG_HASH_BASED,
+	MLX5_LAG_MODE_FLAG_SHARED_FDB,
+	MLX5_LAG_MODE_FLAG_FDB_SEL_MODE_NATIVE,
+};
+
+enum mlx5_lag_mode {
+	MLX5_LAG_MODE_NONE,
+	MLX5_LAG_MODE_ROCE,
+	MLX5_LAG_MODE_SRIOV,
+	MLX5_LAG_MODE_MULTIPATH,
+	MLX5_LAG_MODE_MPESW,
+};
+
+struct lag_func {
+	struct mlx5_core_dev *dev;
+	struct net_device    *netdev;
+	bool has_drop;
+};
+
+/* Used for collection of netdev event info. */
+struct lag_tracker {
+	enum   netdev_lag_tx_type           tx_type;
+	struct netdev_lag_lower_state_info  netdev_state[MLX5_MAX_PORTS];
+	unsigned int is_bonded:1;
+	unsigned int has_inactive:1;
+	enum netdev_lag_hash hash_type;
+};
+
+enum mpesw_op {
+	MLX5_MPESW_OP_ENABLE,
+	MLX5_MPESW_OP_DISABLE,
+};
+
+struct mlx5_mpesw_work_st {
+	struct work_struct work;
+	struct mlx5_lag    *lag;
+	enum mpesw_op	   op;
+	struct completion  comp;
+	int result;
+};
+
+/* LAG data of a ConnectX card.
+ * It serves both its phys functions.
+ */
+struct mlx5_lag {
+	enum mlx5_lag_mode        mode;
+	unsigned long		  mode_flags;
+	unsigned long		  state_flags;
+	u8			  ports;
+	u8			  buckets;
+	int			  mode_changes_in_progress;
+	u8			  v2p_map[MLX5_MAX_PORTS * MLX5_LAG_MAX_HASH_BUCKETS];
+	struct kref               ref;
+	struct lag_func           pf[MLX5_MAX_PORTS];
+	struct lag_tracker        tracker;
+	struct workqueue_struct   *wq;
+	struct delayed_work       bond_work;
+	struct notifier_block     nb;
+	struct lag_mp             lag_mp;
+	struct mlx5_lag_port_sel  port_sel;
+	/* Protect lag fields/state changes */
+	struct mutex		  lock;
+	struct lag_mpesw	  lag_mpesw;
+};
+
+static inline bool mlx5_is_lag_supported(struct mlx5_core_dev *dev)
+{
+	if (!MLX5_CAP_GEN(dev, vport_group_manager) ||
+	    !MLX5_CAP_GEN(dev, lag_master) ||
+	    MLX5_CAP_GEN(dev, num_lag_ports) < 2 ||
+	    MLX5_CAP_GEN(dev, num_lag_ports) > MLX5_MAX_PORTS)
+		return false;
+	return true;
+}
+
+static inline struct mlx5_lag *
+mlx5_lag_dev(struct mlx5_core_dev *dev)
+{
+	return dev->priv.lag;
+}
+
+static inline bool
+__mlx5_lag_is_active(struct mlx5_lag *ldev)
+{
+	return ldev->mode != MLX5_LAG_MODE_NONE;
+}
+
+static inline bool
+mlx5_lag_is_ready(struct mlx5_lag *ldev)
+{
+	return test_bit(MLX5_LAG_FLAG_NDEVS_READY, &ldev->state_flags);
+}
+
+void mlx5_modify_lag(struct mlx5_lag *ldev,
+		     struct lag_tracker *tracker);
+int mlx5_activate_lag(struct mlx5_lag *ldev,
+		      struct lag_tracker *tracker,
+		      enum mlx5_lag_mode mode,
+		      bool shared_fdb);
+int mlx5_lag_dev_get_netdev_idx(struct mlx5_lag *ldev,
+				struct net_device *ndev);
+bool mlx5_shared_fdb_supported(struct mlx5_lag *ldev);
+void mlx5_lag_del_mpesw_rule(struct mlx5_core_dev *dev);
+int mlx5_lag_add_mpesw_rule(struct mlx5_core_dev *dev);
+
+char *mlx5_get_str_port_sel_mode(enum mlx5_lag_mode mode, unsigned long flags);
+void mlx5_infer_tx_enabled(struct lag_tracker *tracker, u8 num_ports,
+			   u8 *ports, int *num_enabled);
+
+void mlx5_ldev_add_debugfs(struct mlx5_core_dev *dev);
+void mlx5_ldev_remove_debugfs(struct dentry *dbg);
+void mlx5_disable_lag(struct mlx5_lag *ldev);
+
+#endif /* __MLX5_LAG_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lag/mp.c b/drivers/net/ethernet/mellanox/mlx5/core/lag/mp.c
new file mode 100644
index 000000000..0259a149a
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lag/mp.c
@@ -0,0 +1,368 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2019 Mellanox Technologies. */
+
+#include <linux/netdevice.h>
+#include <net/nexthop.h>
+#include "lag/lag.h"
+#include "lag/mp.h"
+#include "mlx5_core.h"
+#include "eswitch.h"
+#include "lib/mlx5.h"
+
+static bool __mlx5_lag_is_multipath(struct mlx5_lag *ldev)
+{
+	return ldev->mode == MLX5_LAG_MODE_MULTIPATH;
+}
+
+static bool mlx5_lag_multipath_check_prereq(struct mlx5_lag *ldev)
+{
+	if (!mlx5_lag_is_ready(ldev))
+		return false;
+
+	if (__mlx5_lag_is_active(ldev) && !__mlx5_lag_is_multipath(ldev))
+		return false;
+
+	return mlx5_esw_multipath_prereq(ldev->pf[MLX5_LAG_P1].dev,
+					 ldev->pf[MLX5_LAG_P2].dev);
+}
+
+bool mlx5_lag_is_multipath(struct mlx5_core_dev *dev)
+{
+	struct mlx5_lag *ldev;
+	bool res;
+
+	ldev = mlx5_lag_dev(dev);
+	res  = ldev && __mlx5_lag_is_multipath(ldev);
+
+	return res;
+}
+
+/**
+ * mlx5_lag_set_port_affinity
+ *
+ * @ldev: lag device
+ * @port:
+ *     0 - set normal affinity.
+ *     1 - set affinity to port 1.
+ *     2 - set affinity to port 2.
+ *
+ **/
+static void mlx5_lag_set_port_affinity(struct mlx5_lag *ldev,
+				       enum mlx5_lag_port_affinity port)
+{
+	struct lag_tracker tracker = {};
+
+	if (!__mlx5_lag_is_multipath(ldev))
+		return;
+
+	switch (port) {
+	case MLX5_LAG_NORMAL_AFFINITY:
+		tracker.netdev_state[MLX5_LAG_P1].tx_enabled = true;
+		tracker.netdev_state[MLX5_LAG_P2].tx_enabled = true;
+		tracker.netdev_state[MLX5_LAG_P1].link_up = true;
+		tracker.netdev_state[MLX5_LAG_P2].link_up = true;
+		break;
+	case MLX5_LAG_P1_AFFINITY:
+		tracker.netdev_state[MLX5_LAG_P1].tx_enabled = true;
+		tracker.netdev_state[MLX5_LAG_P1].link_up = true;
+		tracker.netdev_state[MLX5_LAG_P2].tx_enabled = false;
+		tracker.netdev_state[MLX5_LAG_P2].link_up = false;
+		break;
+	case MLX5_LAG_P2_AFFINITY:
+		tracker.netdev_state[MLX5_LAG_P1].tx_enabled = false;
+		tracker.netdev_state[MLX5_LAG_P1].link_up = false;
+		tracker.netdev_state[MLX5_LAG_P2].tx_enabled = true;
+		tracker.netdev_state[MLX5_LAG_P2].link_up = true;
+		break;
+	default:
+		mlx5_core_warn(ldev->pf[MLX5_LAG_P1].dev,
+			       "Invalid affinity port %d", port);
+		return;
+	}
+
+	if (tracker.netdev_state[MLX5_LAG_P1].tx_enabled)
+		mlx5_notifier_call_chain(ldev->pf[MLX5_LAG_P1].dev->priv.events,
+					 MLX5_DEV_EVENT_PORT_AFFINITY,
+					 (void *)0);
+
+	if (tracker.netdev_state[MLX5_LAG_P2].tx_enabled)
+		mlx5_notifier_call_chain(ldev->pf[MLX5_LAG_P2].dev->priv.events,
+					 MLX5_DEV_EVENT_PORT_AFFINITY,
+					 (void *)0);
+
+	mlx5_modify_lag(ldev, &tracker);
+}
+
+static void mlx5_lag_fib_event_flush(struct notifier_block *nb)
+{
+	struct lag_mp *mp = container_of(nb, struct lag_mp, fib_nb);
+
+	flush_workqueue(mp->wq);
+}
+
+static void mlx5_lag_fib_set(struct lag_mp *mp, struct fib_info *fi, u32 dst, int dst_len)
+{
+	mp->fib.mfi = fi;
+	mp->fib.priority = fi->fib_priority;
+	mp->fib.dst = dst;
+	mp->fib.dst_len = dst_len;
+}
+
+struct mlx5_fib_event_work {
+	struct work_struct work;
+	struct mlx5_lag *ldev;
+	unsigned long event;
+	union {
+		struct fib_entry_notifier_info fen_info;
+		struct fib_nh_notifier_info fnh_info;
+	};
+};
+
+static void mlx5_lag_fib_route_event(struct mlx5_lag *ldev, unsigned long event,
+				     struct fib_entry_notifier_info *fen_info)
+{
+	struct fib_info *fi = fen_info->fi;
+	struct lag_mp *mp = &ldev->lag_mp;
+	struct fib_nh *fib_nh0, *fib_nh1;
+	unsigned int nhs;
+
+	/* Handle delete event */
+	if (event == FIB_EVENT_ENTRY_DEL) {
+		/* stop track */
+		if (mp->fib.mfi == fi)
+			mp->fib.mfi = NULL;
+		return;
+	}
+
+	/* Handle multipath entry with lower priority value */
+	if (mp->fib.mfi && mp->fib.mfi != fi &&
+	    (mp->fib.dst != fen_info->dst || mp->fib.dst_len != fen_info->dst_len) &&
+	    fi->fib_priority >= mp->fib.priority)
+		return;
+
+	/* Handle add/replace event */
+	nhs = fib_info_num_path(fi);
+	if (nhs == 1) {
+		if (__mlx5_lag_is_active(ldev)) {
+			struct fib_nh *nh = fib_info_nh(fi, 0);
+			struct net_device *nh_dev = nh->fib_nh_dev;
+			int i = mlx5_lag_dev_get_netdev_idx(ldev, nh_dev);
+
+			if (i < 0)
+				return;
+
+			i++;
+			mlx5_lag_set_port_affinity(ldev, i);
+			mlx5_lag_fib_set(mp, fi, fen_info->dst, fen_info->dst_len);
+		}
+
+		return;
+	}
+
+	if (nhs != 2)
+		return;
+
+	/* Verify next hops are ports of the same hca */
+	fib_nh0 = fib_info_nh(fi, 0);
+	fib_nh1 = fib_info_nh(fi, 1);
+	if (!(fib_nh0->fib_nh_dev == ldev->pf[MLX5_LAG_P1].netdev &&
+	      fib_nh1->fib_nh_dev == ldev->pf[MLX5_LAG_P2].netdev) &&
+	    !(fib_nh0->fib_nh_dev == ldev->pf[MLX5_LAG_P2].netdev &&
+	      fib_nh1->fib_nh_dev == ldev->pf[MLX5_LAG_P1].netdev)) {
+		mlx5_core_warn(ldev->pf[MLX5_LAG_P1].dev,
+			       "Multipath offload require two ports of the same HCA\n");
+		return;
+	}
+
+	/* First time we see multipath route */
+	if (!mp->fib.mfi && !__mlx5_lag_is_active(ldev)) {
+		struct lag_tracker tracker;
+
+		tracker = ldev->tracker;
+		mlx5_activate_lag(ldev, &tracker, MLX5_LAG_MODE_MULTIPATH, false);
+	}
+
+	mlx5_lag_set_port_affinity(ldev, MLX5_LAG_NORMAL_AFFINITY);
+	mlx5_lag_fib_set(mp, fi, fen_info->dst, fen_info->dst_len);
+}
+
+static void mlx5_lag_fib_nexthop_event(struct mlx5_lag *ldev,
+				       unsigned long event,
+				       struct fib_nh *fib_nh,
+				       struct fib_info *fi)
+{
+	struct lag_mp *mp = &ldev->lag_mp;
+
+	/* Check the nh event is related to the route */
+	if (!mp->fib.mfi || mp->fib.mfi != fi)
+		return;
+
+	/* nh added/removed */
+	if (event == FIB_EVENT_NH_DEL) {
+		int i = mlx5_lag_dev_get_netdev_idx(ldev, fib_nh->fib_nh_dev);
+
+		if (i >= 0) {
+			i = (i + 1) % 2 + 1; /* peer port */
+			mlx5_lag_set_port_affinity(ldev, i);
+		}
+	} else if (event == FIB_EVENT_NH_ADD &&
+		   fib_info_num_path(fi) == 2) {
+		mlx5_lag_set_port_affinity(ldev, MLX5_LAG_NORMAL_AFFINITY);
+	}
+}
+
+static void mlx5_lag_fib_update(struct work_struct *work)
+{
+	struct mlx5_fib_event_work *fib_work =
+		container_of(work, struct mlx5_fib_event_work, work);
+	struct mlx5_lag *ldev = fib_work->ldev;
+	struct fib_nh *fib_nh;
+
+	/* Protect internal structures from changes */
+	rtnl_lock();
+	switch (fib_work->event) {
+	case FIB_EVENT_ENTRY_REPLACE:
+	case FIB_EVENT_ENTRY_DEL:
+		mlx5_lag_fib_route_event(ldev, fib_work->event,
+					 &fib_work->fen_info);
+		fib_info_put(fib_work->fen_info.fi);
+		break;
+	case FIB_EVENT_NH_ADD:
+	case FIB_EVENT_NH_DEL:
+		fib_nh = fib_work->fnh_info.fib_nh;
+		mlx5_lag_fib_nexthop_event(ldev,
+					   fib_work->event,
+					   fib_work->fnh_info.fib_nh,
+					   fib_nh->nh_parent);
+		fib_info_put(fib_work->fnh_info.fib_nh->nh_parent);
+		break;
+	}
+
+	rtnl_unlock();
+	kfree(fib_work);
+}
+
+static struct mlx5_fib_event_work *
+mlx5_lag_init_fib_work(struct mlx5_lag *ldev, unsigned long event)
+{
+	struct mlx5_fib_event_work *fib_work;
+
+	fib_work = kzalloc(sizeof(*fib_work), GFP_ATOMIC);
+	if (WARN_ON(!fib_work))
+		return NULL;
+
+	INIT_WORK(&fib_work->work, mlx5_lag_fib_update);
+	fib_work->ldev = ldev;
+	fib_work->event = event;
+
+	return fib_work;
+}
+
+static int mlx5_lag_fib_event(struct notifier_block *nb,
+			      unsigned long event,
+			      void *ptr)
+{
+	struct lag_mp *mp = container_of(nb, struct lag_mp, fib_nb);
+	struct mlx5_lag *ldev = container_of(mp, struct mlx5_lag, lag_mp);
+	struct fib_notifier_info *info = ptr;
+	struct mlx5_fib_event_work *fib_work;
+	struct fib_entry_notifier_info *fen_info;
+	struct fib_nh_notifier_info *fnh_info;
+	struct net_device *fib_dev;
+	struct fib_info *fi;
+
+	if (info->family != AF_INET)
+		return NOTIFY_DONE;
+
+	if (!mlx5_lag_multipath_check_prereq(ldev))
+		return NOTIFY_DONE;
+
+	switch (event) {
+	case FIB_EVENT_ENTRY_REPLACE:
+	case FIB_EVENT_ENTRY_DEL:
+		fen_info = container_of(info, struct fib_entry_notifier_info,
+					info);
+		fi = fen_info->fi;
+		if (fi->nh)
+			return NOTIFY_DONE;
+		fib_dev = fib_info_nh(fen_info->fi, 0)->fib_nh_dev;
+		if (fib_dev != ldev->pf[MLX5_LAG_P1].netdev &&
+		    fib_dev != ldev->pf[MLX5_LAG_P2].netdev) {
+			return NOTIFY_DONE;
+		}
+		fib_work = mlx5_lag_init_fib_work(ldev, event);
+		if (!fib_work)
+			return NOTIFY_DONE;
+		fib_work->fen_info = *fen_info;
+		/* Take reference on fib_info to prevent it from being
+		 * freed while work is queued. Release it afterwards.
+		 */
+		fib_info_hold(fib_work->fen_info.fi);
+		break;
+	case FIB_EVENT_NH_ADD:
+	case FIB_EVENT_NH_DEL:
+		fnh_info = container_of(info, struct fib_nh_notifier_info,
+					info);
+		fib_work = mlx5_lag_init_fib_work(ldev, event);
+		if (!fib_work)
+			return NOTIFY_DONE;
+		fib_work->fnh_info = *fnh_info;
+		fib_info_hold(fib_work->fnh_info.fib_nh->nh_parent);
+		break;
+	default:
+		return NOTIFY_DONE;
+	}
+
+	queue_work(mp->wq, &fib_work->work);
+
+	return NOTIFY_DONE;
+}
+
+void mlx5_lag_mp_reset(struct mlx5_lag *ldev)
+{
+	/* Clear mfi, as it might become stale when a route delete event
+	 * has been missed, see mlx5_lag_fib_route_event().
+	 */
+	ldev->lag_mp.fib.mfi = NULL;
+}
+
+int mlx5_lag_mp_init(struct mlx5_lag *ldev)
+{
+	struct lag_mp *mp = &ldev->lag_mp;
+	int err;
+
+	/* always clear mfi, as it might become stale when a route delete event
+	 * has been missed
+	 */
+	mp->fib.mfi = NULL;
+
+	if (mp->fib_nb.notifier_call)
+		return 0;
+
+	mp->wq = create_singlethread_workqueue("mlx5_lag_mp");
+	if (!mp->wq)
+		return -ENOMEM;
+
+	mp->fib_nb.notifier_call = mlx5_lag_fib_event;
+	err = register_fib_notifier(&init_net, &mp->fib_nb,
+				    mlx5_lag_fib_event_flush, NULL);
+	if (err) {
+		destroy_workqueue(mp->wq);
+		mp->fib_nb.notifier_call = NULL;
+	}
+
+	return err;
+}
+
+void mlx5_lag_mp_cleanup(struct mlx5_lag *ldev)
+{
+	struct lag_mp *mp = &ldev->lag_mp;
+
+	if (!mp->fib_nb.notifier_call)
+		return;
+
+	unregister_fib_notifier(&init_net, &mp->fib_nb);
+	destroy_workqueue(mp->wq);
+	mp->fib_nb.notifier_call = NULL;
+	mp->fib.mfi = NULL;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lag/mp.h b/drivers/net/ethernet/mellanox/mlx5/core/lag/mp.h
new file mode 100644
index 000000000..056a066da
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lag/mp.h
@@ -0,0 +1,42 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2019 Mellanox Technologies. */
+
+#ifndef __MLX5_LAG_MP_H__
+#define __MLX5_LAG_MP_H__
+
+#include "lag.h"
+#include "mlx5_core.h"
+
+enum mlx5_lag_port_affinity {
+	MLX5_LAG_NORMAL_AFFINITY,
+	MLX5_LAG_P1_AFFINITY,
+	MLX5_LAG_P2_AFFINITY,
+};
+
+struct lag_mp {
+	struct notifier_block     fib_nb;
+	struct {
+		const void        *mfi; /* used in tracking fib events */
+		u32               priority;
+		u32               dst;
+		int               dst_len;
+	} fib;
+	struct workqueue_struct   *wq;
+};
+
+#ifdef CONFIG_MLX5_ESWITCH
+
+void mlx5_lag_mp_reset(struct mlx5_lag *ldev);
+int mlx5_lag_mp_init(struct mlx5_lag *ldev);
+void mlx5_lag_mp_cleanup(struct mlx5_lag *ldev);
+bool mlx5_lag_is_multipath(struct mlx5_core_dev *dev);
+
+#else /* CONFIG_MLX5_ESWITCH */
+
+static inline void mlx5_lag_mp_reset(struct mlx5_lag *ldev) {};
+static inline int mlx5_lag_mp_init(struct mlx5_lag *ldev) { return 0; }
+static inline void mlx5_lag_mp_cleanup(struct mlx5_lag *ldev) {}
+static inline bool mlx5_lag_is_multipath(struct mlx5_core_dev *dev) { return false; }
+
+#endif /* CONFIG_MLX5_ESWITCH */
+#endif /* __MLX5_LAG_MP_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.c b/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.c
new file mode 100644
index 000000000..c17e8f1ec
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.c
@@ -0,0 +1,128 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2022, NVIDIA CORPORATION & AFFILIATES. All rights reserved. */
+
+#include <linux/netdevice.h>
+#include <net/nexthop.h>
+#include "lag/lag.h"
+#include "eswitch.h"
+#include "lib/mlx5.h"
+
+static int add_mpesw_rule(struct mlx5_lag *ldev)
+{
+	struct mlx5_core_dev *dev = ldev->pf[MLX5_LAG_P1].dev;
+	int err;
+
+	if (atomic_add_return(1, &ldev->lag_mpesw.mpesw_rule_count) != 1)
+		return 0;
+
+	if (ldev->mode != MLX5_LAG_MODE_NONE) {
+		err = -EINVAL;
+		goto out_err;
+	}
+
+	err = mlx5_activate_lag(ldev, NULL, MLX5_LAG_MODE_MPESW, false);
+	if (err) {
+		mlx5_core_warn(dev, "Failed to create LAG in MPESW mode (%d)\n", err);
+		goto out_err;
+	}
+
+	return 0;
+
+out_err:
+	atomic_dec(&ldev->lag_mpesw.mpesw_rule_count);
+	return err;
+}
+
+static void del_mpesw_rule(struct mlx5_lag *ldev)
+{
+	if (!atomic_dec_return(&ldev->lag_mpesw.mpesw_rule_count) &&
+	    ldev->mode == MLX5_LAG_MODE_MPESW)
+		mlx5_disable_lag(ldev);
+}
+
+static void mlx5_mpesw_work(struct work_struct *work)
+{
+	struct mlx5_mpesw_work_st *mpesww = container_of(work, struct mlx5_mpesw_work_st, work);
+	struct mlx5_lag *ldev = mpesww->lag;
+
+	mutex_lock(&ldev->lock);
+	if (mpesww->op == MLX5_MPESW_OP_ENABLE)
+		mpesww->result = add_mpesw_rule(ldev);
+	else if (mpesww->op == MLX5_MPESW_OP_DISABLE)
+		del_mpesw_rule(ldev);
+	mutex_unlock(&ldev->lock);
+
+	complete(&mpesww->comp);
+}
+
+static int mlx5_lag_mpesw_queue_work(struct mlx5_core_dev *dev,
+				     enum mpesw_op op)
+{
+	struct mlx5_lag *ldev = dev->priv.lag;
+	struct mlx5_mpesw_work_st *work;
+	int err = 0;
+
+	if (!ldev)
+		return 0;
+
+	work = kzalloc(sizeof(*work), GFP_KERNEL);
+	if (!work)
+		return -ENOMEM;
+
+	INIT_WORK(&work->work, mlx5_mpesw_work);
+	init_completion(&work->comp);
+	work->op = op;
+	work->lag = ldev;
+
+	if (!queue_work(ldev->wq, &work->work)) {
+		mlx5_core_warn(dev, "failed to queue mpesw work\n");
+		err = -EINVAL;
+		goto out;
+	}
+	wait_for_completion(&work->comp);
+	err = work->result;
+out:
+	kfree(work);
+	return err;
+}
+
+void mlx5_lag_del_mpesw_rule(struct mlx5_core_dev *dev)
+{
+	mlx5_lag_mpesw_queue_work(dev, MLX5_MPESW_OP_DISABLE);
+}
+
+int mlx5_lag_add_mpesw_rule(struct mlx5_core_dev *dev)
+{
+	return mlx5_lag_mpesw_queue_work(dev, MLX5_MPESW_OP_ENABLE);
+}
+
+int mlx5_lag_do_mirred(struct mlx5_core_dev *mdev, struct net_device *out_dev)
+{
+	struct mlx5_lag *ldev = mdev->priv.lag;
+
+	if (!netif_is_bond_master(out_dev) || !ldev)
+		return 0;
+
+	if (ldev->mode == MLX5_LAG_MODE_MPESW)
+		return -EOPNOTSUPP;
+
+	return 0;
+}
+
+bool mlx5_lag_mpesw_is_activated(struct mlx5_core_dev *dev)
+{
+	bool ret;
+
+	ret = dev->priv.lag && dev->priv.lag->mode == MLX5_LAG_MODE_MPESW;
+	return ret;
+}
+
+void mlx5_lag_mpesw_init(struct mlx5_lag *ldev)
+{
+	atomic_set(&ldev->lag_mpesw.mpesw_rule_count, 0);
+}
+
+void mlx5_lag_mpesw_cleanup(struct mlx5_lag *ldev)
+{
+	WARN_ON(atomic_read(&ldev->lag_mpesw.mpesw_rule_count));
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.h b/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.h
new file mode 100644
index 000000000..88e8daffc
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.h
@@ -0,0 +1,25 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2022, NVIDIA CORPORATION & AFFILIATES. All rights reserved. */
+
+#ifndef __MLX5_LAG_MPESW_H__
+#define __MLX5_LAG_MPESW_H__
+
+#include "lag.h"
+#include "mlx5_core.h"
+
+struct lag_mpesw {
+	struct work_struct mpesw_work;
+	atomic_t mpesw_rule_count;
+};
+
+int mlx5_lag_do_mirred(struct mlx5_core_dev *mdev, struct net_device *out_dev);
+bool mlx5_lag_mpesw_is_activated(struct mlx5_core_dev *dev);
+#if IS_ENABLED(CONFIG_MLX5_ESWITCH)
+void mlx5_lag_mpesw_init(struct mlx5_lag *ldev);
+void mlx5_lag_mpesw_cleanup(struct mlx5_lag *ldev);
+#else
+static inline void mlx5_lag_mpesw_init(struct mlx5_lag *ldev) {}
+static inline void mlx5_lag_mpesw_cleanup(struct mlx5_lag *ldev) {}
+#endif
+
+#endif /* __MLX5_LAG_MPESW_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lag/port_sel.c b/drivers/net/ethernet/mellanox/mlx5/core/lag/port_sel.c
new file mode 100644
index 000000000..7d9bbb494
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lag/port_sel.c
@@ -0,0 +1,637 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. */
+
+#include <linux/netdevice.h>
+#include "lag.h"
+
+enum {
+	MLX5_LAG_FT_LEVEL_TTC,
+	MLX5_LAG_FT_LEVEL_INNER_TTC,
+	MLX5_LAG_FT_LEVEL_DEFINER,
+};
+
+static struct mlx5_flow_group *
+mlx5_create_hash_flow_group(struct mlx5_flow_table *ft,
+			    struct mlx5_flow_definer *definer,
+			    u8 rules)
+{
+	int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
+	struct mlx5_flow_group *fg;
+	u32 *in;
+
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in)
+		return ERR_PTR(-ENOMEM);
+
+	MLX5_SET(create_flow_group_in, in, match_definer_id,
+		 mlx5_get_match_definer_id(definer));
+	MLX5_SET(create_flow_group_in, in, start_flow_index, 0);
+	MLX5_SET(create_flow_group_in, in, end_flow_index, rules - 1);
+	MLX5_SET(create_flow_group_in, in, group_type,
+		 MLX5_CREATE_FLOW_GROUP_IN_GROUP_TYPE_HASH_SPLIT);
+
+	fg = mlx5_create_flow_group(ft, in);
+	kvfree(in);
+	return fg;
+}
+
+static int mlx5_lag_create_port_sel_table(struct mlx5_lag *ldev,
+					  struct mlx5_lag_definer *lag_definer,
+					  u8 *ports)
+{
+	struct mlx5_core_dev *dev = ldev->pf[MLX5_LAG_P1].dev;
+	struct mlx5_flow_table_attr ft_attr = {};
+	struct mlx5_flow_destination dest = {};
+	MLX5_DECLARE_FLOW_ACT(flow_act);
+	struct mlx5_flow_namespace *ns;
+	int err, i;
+	int idx;
+	int j;
+
+	ft_attr.max_fte = ldev->ports * ldev->buckets;
+	ft_attr.level = MLX5_LAG_FT_LEVEL_DEFINER;
+
+	ns = mlx5_get_flow_namespace(dev, MLX5_FLOW_NAMESPACE_PORT_SEL);
+	if (!ns) {
+		mlx5_core_warn(dev, "Failed to get port selection namespace\n");
+		return -EOPNOTSUPP;
+	}
+
+	lag_definer->ft = mlx5_create_flow_table(ns, &ft_attr);
+	if (IS_ERR(lag_definer->ft)) {
+		mlx5_core_warn(dev, "Failed to create port selection table\n");
+		return PTR_ERR(lag_definer->ft);
+	}
+
+	lag_definer->fg = mlx5_create_hash_flow_group(lag_definer->ft,
+						      lag_definer->definer,
+						      ft_attr.max_fte);
+	if (IS_ERR(lag_definer->fg)) {
+		err = PTR_ERR(lag_definer->fg);
+		goto destroy_ft;
+	}
+
+	dest.type = MLX5_FLOW_DESTINATION_TYPE_UPLINK;
+	dest.vport.flags |= MLX5_FLOW_DEST_VPORT_VHCA_ID;
+	flow_act.flags |= FLOW_ACT_NO_APPEND;
+	for (i = 0; i < ldev->ports; i++) {
+		for (j = 0; j < ldev->buckets; j++) {
+			u8 affinity;
+
+			idx = i * ldev->buckets + j;
+			affinity = ports[idx];
+
+			dest.vport.vhca_id = MLX5_CAP_GEN(ldev->pf[affinity - 1].dev,
+							  vhca_id);
+			lag_definer->rules[idx] = mlx5_add_flow_rules(lag_definer->ft,
+								      NULL, &flow_act,
+								      &dest, 1);
+			if (IS_ERR(lag_definer->rules[idx])) {
+				err = PTR_ERR(lag_definer->rules[idx]);
+				while (i--)
+					while (j--)
+						mlx5_del_flow_rules(lag_definer->rules[idx]);
+				goto destroy_fg;
+			}
+		}
+	}
+
+	return 0;
+
+destroy_fg:
+	mlx5_destroy_flow_group(lag_definer->fg);
+destroy_ft:
+	mlx5_destroy_flow_table(lag_definer->ft);
+	return err;
+}
+
+static int mlx5_lag_set_definer_inner(u32 *match_definer_mask,
+				      enum mlx5_traffic_types tt)
+{
+	int format_id;
+	u8 *ipv6;
+
+	switch (tt) {
+	case MLX5_TT_IPV4_UDP:
+	case MLX5_TT_IPV4_TCP:
+		format_id = 23;
+		MLX5_SET_TO_ONES(match_definer_format_23, match_definer_mask,
+				 inner_l4_sport);
+		MLX5_SET_TO_ONES(match_definer_format_23, match_definer_mask,
+				 inner_l4_dport);
+		MLX5_SET_TO_ONES(match_definer_format_23, match_definer_mask,
+				 inner_ip_src_addr);
+		MLX5_SET_TO_ONES(match_definer_format_23, match_definer_mask,
+				 inner_ip_dest_addr);
+		break;
+	case MLX5_TT_IPV4:
+		format_id = 23;
+		MLX5_SET_TO_ONES(match_definer_format_23, match_definer_mask,
+				 inner_l3_type);
+		MLX5_SET_TO_ONES(match_definer_format_23, match_definer_mask,
+				 inner_dmac_47_16);
+		MLX5_SET_TO_ONES(match_definer_format_23, match_definer_mask,
+				 inner_dmac_15_0);
+		MLX5_SET_TO_ONES(match_definer_format_23, match_definer_mask,
+				 inner_smac_47_16);
+		MLX5_SET_TO_ONES(match_definer_format_23, match_definer_mask,
+				 inner_smac_15_0);
+		MLX5_SET_TO_ONES(match_definer_format_23, match_definer_mask,
+				 inner_ip_src_addr);
+		MLX5_SET_TO_ONES(match_definer_format_23, match_definer_mask,
+				 inner_ip_dest_addr);
+		break;
+	case MLX5_TT_IPV6_TCP:
+	case MLX5_TT_IPV6_UDP:
+		format_id = 31;
+		MLX5_SET_TO_ONES(match_definer_format_31, match_definer_mask,
+				 inner_l4_sport);
+		MLX5_SET_TO_ONES(match_definer_format_31, match_definer_mask,
+				 inner_l4_dport);
+		ipv6 = MLX5_ADDR_OF(match_definer_format_31, match_definer_mask,
+				    inner_ip_dest_addr);
+		memset(ipv6, 0xff, 16);
+		ipv6 = MLX5_ADDR_OF(match_definer_format_31, match_definer_mask,
+				    inner_ip_src_addr);
+		memset(ipv6, 0xff, 16);
+		break;
+	case MLX5_TT_IPV6:
+		format_id = 32;
+		ipv6 = MLX5_ADDR_OF(match_definer_format_32, match_definer_mask,
+				    inner_ip_dest_addr);
+		memset(ipv6, 0xff, 16);
+		ipv6 = MLX5_ADDR_OF(match_definer_format_32, match_definer_mask,
+				    inner_ip_src_addr);
+		memset(ipv6, 0xff, 16);
+		MLX5_SET_TO_ONES(match_definer_format_32, match_definer_mask,
+				 inner_dmac_47_16);
+		MLX5_SET_TO_ONES(match_definer_format_32, match_definer_mask,
+				 inner_dmac_15_0);
+		MLX5_SET_TO_ONES(match_definer_format_32, match_definer_mask,
+				 inner_smac_47_16);
+		MLX5_SET_TO_ONES(match_definer_format_32, match_definer_mask,
+				 inner_smac_15_0);
+		break;
+	default:
+		format_id = 23;
+		MLX5_SET_TO_ONES(match_definer_format_23, match_definer_mask,
+				 inner_l3_type);
+		MLX5_SET_TO_ONES(match_definer_format_23, match_definer_mask,
+				 inner_dmac_47_16);
+		MLX5_SET_TO_ONES(match_definer_format_23, match_definer_mask,
+				 inner_dmac_15_0);
+		MLX5_SET_TO_ONES(match_definer_format_23, match_definer_mask,
+				 inner_smac_47_16);
+		MLX5_SET_TO_ONES(match_definer_format_23, match_definer_mask,
+				 inner_smac_15_0);
+		break;
+	}
+
+	return format_id;
+}
+
+static int mlx5_lag_set_definer(u32 *match_definer_mask,
+				enum mlx5_traffic_types tt, bool tunnel,
+				enum netdev_lag_hash hash)
+{
+	int format_id;
+	u8 *ipv6;
+
+	if (tunnel)
+		return mlx5_lag_set_definer_inner(match_definer_mask, tt);
+
+	switch (tt) {
+	case MLX5_TT_IPV4_UDP:
+	case MLX5_TT_IPV4_TCP:
+		format_id = 22;
+		MLX5_SET_TO_ONES(match_definer_format_22, match_definer_mask,
+				 outer_l4_sport);
+		MLX5_SET_TO_ONES(match_definer_format_22, match_definer_mask,
+				 outer_l4_dport);
+		MLX5_SET_TO_ONES(match_definer_format_22, match_definer_mask,
+				 outer_ip_src_addr);
+		MLX5_SET_TO_ONES(match_definer_format_22, match_definer_mask,
+				 outer_ip_dest_addr);
+		break;
+	case MLX5_TT_IPV4:
+		format_id = 22;
+		MLX5_SET_TO_ONES(match_definer_format_22, match_definer_mask,
+				 outer_l3_type);
+		MLX5_SET_TO_ONES(match_definer_format_22, match_definer_mask,
+				 outer_dmac_47_16);
+		MLX5_SET_TO_ONES(match_definer_format_22, match_definer_mask,
+				 outer_dmac_15_0);
+		MLX5_SET_TO_ONES(match_definer_format_22, match_definer_mask,
+				 outer_smac_47_16);
+		MLX5_SET_TO_ONES(match_definer_format_22, match_definer_mask,
+				 outer_smac_15_0);
+		MLX5_SET_TO_ONES(match_definer_format_22, match_definer_mask,
+				 outer_ip_src_addr);
+		MLX5_SET_TO_ONES(match_definer_format_22, match_definer_mask,
+				 outer_ip_dest_addr);
+		break;
+	case MLX5_TT_IPV6_TCP:
+	case MLX5_TT_IPV6_UDP:
+		format_id = 29;
+		MLX5_SET_TO_ONES(match_definer_format_29, match_definer_mask,
+				 outer_l4_sport);
+		MLX5_SET_TO_ONES(match_definer_format_29, match_definer_mask,
+				 outer_l4_dport);
+		ipv6 = MLX5_ADDR_OF(match_definer_format_29, match_definer_mask,
+				    outer_ip_dest_addr);
+		memset(ipv6, 0xff, 16);
+		ipv6 = MLX5_ADDR_OF(match_definer_format_29, match_definer_mask,
+				    outer_ip_src_addr);
+		memset(ipv6, 0xff, 16);
+		break;
+	case MLX5_TT_IPV6:
+		format_id = 30;
+		ipv6 = MLX5_ADDR_OF(match_definer_format_30, match_definer_mask,
+				    outer_ip_dest_addr);
+		memset(ipv6, 0xff, 16);
+		ipv6 = MLX5_ADDR_OF(match_definer_format_30, match_definer_mask,
+				    outer_ip_src_addr);
+		memset(ipv6, 0xff, 16);
+		MLX5_SET_TO_ONES(match_definer_format_30, match_definer_mask,
+				 outer_dmac_47_16);
+		MLX5_SET_TO_ONES(match_definer_format_30, match_definer_mask,
+				 outer_dmac_15_0);
+		MLX5_SET_TO_ONES(match_definer_format_30, match_definer_mask,
+				 outer_smac_47_16);
+		MLX5_SET_TO_ONES(match_definer_format_30, match_definer_mask,
+				 outer_smac_15_0);
+		break;
+	default:
+		format_id = 0;
+		MLX5_SET_TO_ONES(match_definer_format_0, match_definer_mask,
+				 outer_smac_47_16);
+		MLX5_SET_TO_ONES(match_definer_format_0, match_definer_mask,
+				 outer_smac_15_0);
+
+		if (hash == NETDEV_LAG_HASH_VLAN_SRCMAC) {
+			MLX5_SET_TO_ONES(match_definer_format_0,
+					 match_definer_mask,
+					 outer_first_vlan_vid);
+			break;
+		}
+
+		MLX5_SET_TO_ONES(match_definer_format_0, match_definer_mask,
+				 outer_ethertype);
+		MLX5_SET_TO_ONES(match_definer_format_0, match_definer_mask,
+				 outer_dmac_47_16);
+		MLX5_SET_TO_ONES(match_definer_format_0, match_definer_mask,
+				 outer_dmac_15_0);
+		break;
+	}
+
+	return format_id;
+}
+
+static struct mlx5_lag_definer *
+mlx5_lag_create_definer(struct mlx5_lag *ldev, enum netdev_lag_hash hash,
+			enum mlx5_traffic_types tt, bool tunnel, u8 *ports)
+{
+	struct mlx5_core_dev *dev = ldev->pf[MLX5_LAG_P1].dev;
+	struct mlx5_lag_definer *lag_definer;
+	u32 *match_definer_mask;
+	int format_id, err;
+
+	lag_definer = kzalloc(sizeof(*lag_definer), GFP_KERNEL);
+	if (!lag_definer)
+		return ERR_PTR(-ENOMEM);
+
+	match_definer_mask = kvzalloc(MLX5_FLD_SZ_BYTES(match_definer,
+							match_mask),
+				      GFP_KERNEL);
+	if (!match_definer_mask) {
+		err = -ENOMEM;
+		goto free_lag_definer;
+	}
+
+	format_id = mlx5_lag_set_definer(match_definer_mask, tt, tunnel, hash);
+	lag_definer->definer =
+		mlx5_create_match_definer(dev, MLX5_FLOW_NAMESPACE_PORT_SEL,
+					  format_id, match_definer_mask);
+	if (IS_ERR(lag_definer->definer)) {
+		err = PTR_ERR(lag_definer->definer);
+		goto free_mask;
+	}
+
+	err = mlx5_lag_create_port_sel_table(ldev, lag_definer, ports);
+	if (err)
+		goto destroy_match_definer;
+
+	kvfree(match_definer_mask);
+
+	return lag_definer;
+
+destroy_match_definer:
+	mlx5_destroy_match_definer(dev, lag_definer->definer);
+free_mask:
+	kvfree(match_definer_mask);
+free_lag_definer:
+	kfree(lag_definer);
+	return ERR_PTR(err);
+}
+
+static void mlx5_lag_destroy_definer(struct mlx5_lag *ldev,
+				     struct mlx5_lag_definer *lag_definer)
+{
+	struct mlx5_core_dev *dev = ldev->pf[MLX5_LAG_P1].dev;
+	int idx;
+	int i;
+	int j;
+
+	for (i = 0; i < ldev->ports; i++) {
+		for (j = 0; j < ldev->buckets; j++) {
+			idx = i * ldev->buckets + j;
+			mlx5_del_flow_rules(lag_definer->rules[idx]);
+		}
+	}
+	mlx5_destroy_flow_group(lag_definer->fg);
+	mlx5_destroy_flow_table(lag_definer->ft);
+	mlx5_destroy_match_definer(dev, lag_definer->definer);
+	kfree(lag_definer);
+}
+
+static void mlx5_lag_destroy_definers(struct mlx5_lag *ldev)
+{
+	struct mlx5_lag_port_sel *port_sel = &ldev->port_sel;
+	int tt;
+
+	for_each_set_bit(tt, port_sel->tt_map, MLX5_NUM_TT) {
+		if (port_sel->outer.definers[tt])
+			mlx5_lag_destroy_definer(ldev,
+						 port_sel->outer.definers[tt]);
+		if (port_sel->inner.definers[tt])
+			mlx5_lag_destroy_definer(ldev,
+						 port_sel->inner.definers[tt]);
+	}
+}
+
+static int mlx5_lag_create_definers(struct mlx5_lag *ldev,
+				    enum netdev_lag_hash hash_type,
+				    u8 *ports)
+{
+	struct mlx5_lag_port_sel *port_sel = &ldev->port_sel;
+	struct mlx5_lag_definer *lag_definer;
+	int tt, err;
+
+	for_each_set_bit(tt, port_sel->tt_map, MLX5_NUM_TT) {
+		lag_definer = mlx5_lag_create_definer(ldev, hash_type, tt,
+						      false, ports);
+		if (IS_ERR(lag_definer)) {
+			err = PTR_ERR(lag_definer);
+			goto destroy_definers;
+		}
+		port_sel->outer.definers[tt] = lag_definer;
+
+		if (!port_sel->tunnel)
+			continue;
+
+		lag_definer =
+			mlx5_lag_create_definer(ldev, hash_type, tt,
+						true, ports);
+		if (IS_ERR(lag_definer)) {
+			err = PTR_ERR(lag_definer);
+			goto destroy_definers;
+		}
+		port_sel->inner.definers[tt] = lag_definer;
+	}
+
+	return 0;
+
+destroy_definers:
+	mlx5_lag_destroy_definers(ldev);
+	return err;
+}
+
+static void set_tt_map(struct mlx5_lag_port_sel *port_sel,
+		       enum netdev_lag_hash hash)
+{
+	port_sel->tunnel = false;
+
+	switch (hash) {
+	case NETDEV_LAG_HASH_E34:
+		port_sel->tunnel = true;
+		fallthrough;
+	case NETDEV_LAG_HASH_L34:
+		set_bit(MLX5_TT_IPV4_TCP, port_sel->tt_map);
+		set_bit(MLX5_TT_IPV4_UDP, port_sel->tt_map);
+		set_bit(MLX5_TT_IPV6_TCP, port_sel->tt_map);
+		set_bit(MLX5_TT_IPV6_UDP, port_sel->tt_map);
+		set_bit(MLX5_TT_IPV4, port_sel->tt_map);
+		set_bit(MLX5_TT_IPV6, port_sel->tt_map);
+		set_bit(MLX5_TT_ANY, port_sel->tt_map);
+		break;
+	case NETDEV_LAG_HASH_E23:
+		port_sel->tunnel = true;
+		fallthrough;
+	case NETDEV_LAG_HASH_L23:
+		set_bit(MLX5_TT_IPV4, port_sel->tt_map);
+		set_bit(MLX5_TT_IPV6, port_sel->tt_map);
+		set_bit(MLX5_TT_ANY, port_sel->tt_map);
+		break;
+	default:
+		set_bit(MLX5_TT_ANY, port_sel->tt_map);
+		break;
+	}
+}
+
+#define SET_IGNORE_DESTS_BITS(tt_map, dests)				\
+	do {								\
+		int idx;						\
+									\
+		for_each_clear_bit(idx, tt_map, MLX5_NUM_TT)		\
+			set_bit(idx, dests);				\
+	} while (0)
+
+static void mlx5_lag_set_inner_ttc_params(struct mlx5_lag *ldev,
+					  struct ttc_params *ttc_params)
+{
+	struct mlx5_core_dev *dev = ldev->pf[MLX5_LAG_P1].dev;
+	struct mlx5_lag_port_sel *port_sel = &ldev->port_sel;
+	struct mlx5_flow_table_attr *ft_attr;
+	int tt;
+
+	ttc_params->ns = mlx5_get_flow_namespace(dev,
+						 MLX5_FLOW_NAMESPACE_PORT_SEL);
+	ft_attr = &ttc_params->ft_attr;
+	ft_attr->level = MLX5_LAG_FT_LEVEL_INNER_TTC;
+
+	for_each_set_bit(tt, port_sel->tt_map, MLX5_NUM_TT) {
+		ttc_params->dests[tt].type =
+			MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
+		ttc_params->dests[tt].ft = port_sel->inner.definers[tt]->ft;
+	}
+	SET_IGNORE_DESTS_BITS(port_sel->tt_map, ttc_params->ignore_dests);
+}
+
+static void mlx5_lag_set_outer_ttc_params(struct mlx5_lag *ldev,
+					  struct ttc_params *ttc_params)
+{
+	struct mlx5_core_dev *dev = ldev->pf[MLX5_LAG_P1].dev;
+	struct mlx5_lag_port_sel *port_sel = &ldev->port_sel;
+	struct mlx5_flow_table_attr *ft_attr;
+	int tt;
+
+	ttc_params->ns = mlx5_get_flow_namespace(dev,
+						 MLX5_FLOW_NAMESPACE_PORT_SEL);
+	ft_attr = &ttc_params->ft_attr;
+	ft_attr->level = MLX5_LAG_FT_LEVEL_TTC;
+
+	for_each_set_bit(tt, port_sel->tt_map, MLX5_NUM_TT) {
+		ttc_params->dests[tt].type =
+			MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
+		ttc_params->dests[tt].ft = port_sel->outer.definers[tt]->ft;
+	}
+	SET_IGNORE_DESTS_BITS(port_sel->tt_map, ttc_params->ignore_dests);
+
+	ttc_params->inner_ttc = port_sel->tunnel;
+	if (!port_sel->tunnel)
+		return;
+
+	for (tt = 0; tt < MLX5_NUM_TUNNEL_TT; tt++) {
+		ttc_params->tunnel_dests[tt].type =
+			MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
+		ttc_params->tunnel_dests[tt].ft =
+			mlx5_get_ttc_flow_table(port_sel->inner.ttc);
+	}
+}
+
+static int mlx5_lag_create_ttc_table(struct mlx5_lag *ldev)
+{
+	struct mlx5_core_dev *dev = ldev->pf[MLX5_LAG_P1].dev;
+	struct mlx5_lag_port_sel *port_sel = &ldev->port_sel;
+	struct ttc_params ttc_params = {};
+
+	mlx5_lag_set_outer_ttc_params(ldev, &ttc_params);
+	port_sel->outer.ttc = mlx5_create_ttc_table(dev, &ttc_params);
+	if (IS_ERR(port_sel->outer.ttc))
+		return PTR_ERR(port_sel->outer.ttc);
+
+	return 0;
+}
+
+static int mlx5_lag_create_inner_ttc_table(struct mlx5_lag *ldev)
+{
+	struct mlx5_core_dev *dev = ldev->pf[MLX5_LAG_P1].dev;
+	struct mlx5_lag_port_sel *port_sel = &ldev->port_sel;
+	struct ttc_params ttc_params = {};
+
+	mlx5_lag_set_inner_ttc_params(ldev, &ttc_params);
+	port_sel->inner.ttc = mlx5_create_inner_ttc_table(dev, &ttc_params);
+	if (IS_ERR(port_sel->inner.ttc))
+		return PTR_ERR(port_sel->inner.ttc);
+
+	return 0;
+}
+
+int mlx5_lag_port_sel_create(struct mlx5_lag *ldev,
+			     enum netdev_lag_hash hash_type, u8 *ports)
+{
+	struct mlx5_lag_port_sel *port_sel = &ldev->port_sel;
+	int err;
+
+	set_tt_map(port_sel, hash_type);
+	err = mlx5_lag_create_definers(ldev, hash_type, ports);
+	if (err)
+		return err;
+
+	if (port_sel->tunnel) {
+		err = mlx5_lag_create_inner_ttc_table(ldev);
+		if (err)
+			goto destroy_definers;
+	}
+
+	err = mlx5_lag_create_ttc_table(ldev);
+	if (err)
+		goto destroy_inner;
+
+	return 0;
+
+destroy_inner:
+	if (port_sel->tunnel)
+		mlx5_destroy_ttc_table(port_sel->inner.ttc);
+destroy_definers:
+	mlx5_lag_destroy_definers(ldev);
+	return err;
+}
+
+static int __mlx5_lag_modify_definers_destinations(struct mlx5_lag *ldev,
+						   struct mlx5_lag_definer *def,
+						   u8 *ports)
+{
+	struct mlx5_flow_destination dest = {};
+	int idx;
+	int err;
+	int i;
+	int j;
+
+	dest.type = MLX5_FLOW_DESTINATION_TYPE_UPLINK;
+	dest.vport.flags |= MLX5_FLOW_DEST_VPORT_VHCA_ID;
+
+	for (i = 0; i < ldev->ports; i++) {
+		for (j = 0; j < ldev->buckets; j++) {
+			idx = i * ldev->buckets + j;
+			if (ldev->v2p_map[idx] == ports[idx])
+				continue;
+
+			dest.vport.vhca_id = MLX5_CAP_GEN(ldev->pf[ports[idx] - 1].dev,
+							  vhca_id);
+			err = mlx5_modify_rule_destination(def->rules[idx], &dest, NULL);
+			if (err)
+				return err;
+		}
+	}
+
+	return 0;
+}
+
+static int
+mlx5_lag_modify_definers_destinations(struct mlx5_lag *ldev,
+				      struct mlx5_lag_definer **definers,
+				      u8 *ports)
+{
+	struct mlx5_lag_port_sel *port_sel = &ldev->port_sel;
+	int err;
+	int tt;
+
+	for_each_set_bit(tt, port_sel->tt_map, MLX5_NUM_TT) {
+		err = __mlx5_lag_modify_definers_destinations(ldev, definers[tt], ports);
+		if (err)
+			return err;
+	}
+
+	return 0;
+}
+
+int mlx5_lag_port_sel_modify(struct mlx5_lag *ldev, u8 *ports)
+{
+	struct mlx5_lag_port_sel *port_sel = &ldev->port_sel;
+	int err;
+
+	err = mlx5_lag_modify_definers_destinations(ldev,
+						    port_sel->outer.definers,
+						    ports);
+	if (err)
+		return err;
+
+	if (!port_sel->tunnel)
+		return 0;
+
+	return mlx5_lag_modify_definers_destinations(ldev,
+						     port_sel->inner.definers,
+						     ports);
+}
+
+void mlx5_lag_port_sel_destroy(struct mlx5_lag *ldev)
+{
+	struct mlx5_lag_port_sel *port_sel = &ldev->port_sel;
+
+	mlx5_destroy_ttc_table(port_sel->outer.ttc);
+	if (port_sel->tunnel)
+		mlx5_destroy_ttc_table(port_sel->inner.ttc);
+	mlx5_lag_destroy_definers(ldev);
+	memset(port_sel, 0, sizeof(*port_sel));
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lag/port_sel.h b/drivers/net/ethernet/mellanox/mlx5/core/lag/port_sel.h
new file mode 100644
index 000000000..5ec3af2a3
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lag/port_sel.h
@@ -0,0 +1,53 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. */
+
+#ifndef __MLX5_LAG_FS_H__
+#define __MLX5_LAG_FS_H__
+
+#include "lib/fs_ttc.h"
+
+struct mlx5_lag_definer {
+	struct mlx5_flow_definer *definer;
+	struct mlx5_flow_table *ft;
+	struct mlx5_flow_group *fg;
+	/* Each port has ldev->buckets number of rules and they are arrange in
+	 * [port * buckets .. port * buckets + buckets) locations
+	 */
+	struct mlx5_flow_handle *rules[MLX5_MAX_PORTS * MLX5_LAG_MAX_HASH_BUCKETS];
+};
+
+struct mlx5_lag_ttc {
+	struct mlx5_ttc_table *ttc;
+	struct mlx5_lag_definer *definers[MLX5_NUM_TT];
+};
+
+struct mlx5_lag_port_sel {
+	DECLARE_BITMAP(tt_map, MLX5_NUM_TT);
+	bool   tunnel;
+	struct mlx5_lag_ttc outer;
+	struct mlx5_lag_ttc inner;
+};
+
+#ifdef CONFIG_MLX5_ESWITCH
+
+int mlx5_lag_port_sel_modify(struct mlx5_lag *ldev, u8 *ports);
+void mlx5_lag_port_sel_destroy(struct mlx5_lag *ldev);
+int mlx5_lag_port_sel_create(struct mlx5_lag *ldev,
+			     enum netdev_lag_hash hash_type, u8 *ports);
+
+#else /* CONFIG_MLX5_ESWITCH */
+static inline int mlx5_lag_port_sel_create(struct mlx5_lag *ldev,
+					   enum netdev_lag_hash hash_type,
+					   u8 *ports)
+{
+	return 0;
+}
+
+static inline int mlx5_lag_port_sel_modify(struct mlx5_lag *ldev, u8 *ports)
+{
+	return 0;
+}
+
+static inline void mlx5_lag_port_sel_destroy(struct mlx5_lag *ldev) {}
+#endif /* CONFIG_MLX5_ESWITCH */
+#endif /* __MLX5_LAG_FS_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/aso.c b/drivers/net/ethernet/mellanox/mlx5/core/lib/aso.c
new file mode 100644
index 000000000..c215252f2
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/aso.c
@@ -0,0 +1,432 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+// Copyright (c) 2022, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+
+#include <linux/mlx5/device.h>
+#include <linux/mlx5/transobj.h>
+#include "clock.h"
+#include "aso.h"
+#include "wq.h"
+
+struct mlx5_aso_cq {
+	/* data path - accessed per cqe */
+	struct mlx5_cqwq           wq;
+
+	/* data path - accessed per napi poll */
+	struct mlx5_core_cq        mcq;
+
+	/* control */
+	struct mlx5_core_dev      *mdev;
+	struct mlx5_wq_ctrl        wq_ctrl;
+} ____cacheline_aligned_in_smp;
+
+struct mlx5_aso {
+	/* data path */
+	u16                        cc;
+	u16                        pc;
+
+	struct mlx5_wqe_ctrl_seg  *doorbell_cseg;
+	struct mlx5_aso_cq         cq;
+
+	/* read only */
+	struct mlx5_wq_cyc         wq;
+	void __iomem              *uar_map;
+	u32                        sqn;
+
+	/* control path */
+	struct mlx5_wq_ctrl        wq_ctrl;
+
+} ____cacheline_aligned_in_smp;
+
+static void mlx5_aso_free_cq(struct mlx5_aso_cq *cq)
+{
+	mlx5_wq_destroy(&cq->wq_ctrl);
+}
+
+static int mlx5_aso_alloc_cq(struct mlx5_core_dev *mdev, int numa_node,
+			     void *cqc_data, struct mlx5_aso_cq *cq)
+{
+	struct mlx5_core_cq *mcq = &cq->mcq;
+	struct mlx5_wq_param param;
+	int err;
+	u32 i;
+
+	param.buf_numa_node = numa_node;
+	param.db_numa_node = numa_node;
+
+	err = mlx5_cqwq_create(mdev, &param, cqc_data, &cq->wq, &cq->wq_ctrl);
+	if (err)
+		return err;
+
+	mcq->cqe_sz     = 64;
+	mcq->set_ci_db  = cq->wq_ctrl.db.db;
+	mcq->arm_db     = cq->wq_ctrl.db.db + 1;
+
+	for (i = 0; i < mlx5_cqwq_get_size(&cq->wq); i++) {
+		struct mlx5_cqe64 *cqe = mlx5_cqwq_get_wqe(&cq->wq, i);
+
+		cqe->op_own = 0xf1;
+	}
+
+	cq->mdev = mdev;
+
+	return 0;
+}
+
+static int create_aso_cq(struct mlx5_aso_cq *cq, void *cqc_data)
+{
+	u32 out[MLX5_ST_SZ_DW(create_cq_out)];
+	struct mlx5_core_dev *mdev = cq->mdev;
+	struct mlx5_core_cq *mcq = &cq->mcq;
+	void *in, *cqc;
+	int inlen, eqn;
+	int err;
+
+	err = mlx5_vector2eqn(mdev, 0, &eqn);
+	if (err)
+		return err;
+
+	inlen = MLX5_ST_SZ_BYTES(create_cq_in) +
+		sizeof(u64) * cq->wq_ctrl.buf.npages;
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in)
+		return -ENOMEM;
+
+	cqc = MLX5_ADDR_OF(create_cq_in, in, cq_context);
+
+	memcpy(cqc, cqc_data, MLX5_ST_SZ_BYTES(cqc));
+
+	mlx5_fill_page_frag_array(&cq->wq_ctrl.buf,
+				  (__be64 *)MLX5_ADDR_OF(create_cq_in, in, pas));
+
+	MLX5_SET(cqc,   cqc, cq_period_mode, MLX5_CQ_PERIOD_MODE_START_FROM_EQE);
+	MLX5_SET(cqc,   cqc, c_eqn_or_apu_element, eqn);
+	MLX5_SET(cqc,   cqc, uar_page,      mdev->priv.uar->index);
+	MLX5_SET(cqc,   cqc, log_page_size, cq->wq_ctrl.buf.page_shift -
+					    MLX5_ADAPTER_PAGE_SHIFT);
+	MLX5_SET64(cqc, cqc, dbr_addr,      cq->wq_ctrl.db.dma);
+
+	err = mlx5_core_create_cq(mdev, mcq, in, inlen, out, sizeof(out));
+
+	kvfree(in);
+
+	return err;
+}
+
+static void mlx5_aso_destroy_cq(struct mlx5_aso_cq *cq)
+{
+	mlx5_core_destroy_cq(cq->mdev, &cq->mcq);
+	mlx5_wq_destroy(&cq->wq_ctrl);
+}
+
+static int mlx5_aso_create_cq(struct mlx5_core_dev *mdev, int numa_node,
+			      struct mlx5_aso_cq *cq)
+{
+	void *cqc_data;
+	int err;
+
+	cqc_data = kvzalloc(MLX5_ST_SZ_BYTES(cqc), GFP_KERNEL);
+	if (!cqc_data)
+		return -ENOMEM;
+
+	MLX5_SET(cqc, cqc_data, log_cq_size, 1);
+	MLX5_SET(cqc, cqc_data, uar_page, mdev->priv.uar->index);
+	if (MLX5_CAP_GEN(mdev, cqe_128_always) && cache_line_size() >= 128)
+		MLX5_SET(cqc, cqc_data, cqe_sz, CQE_STRIDE_128_PAD);
+
+	err = mlx5_aso_alloc_cq(mdev, numa_node, cqc_data, cq);
+	if (err) {
+		mlx5_core_err(mdev, "Failed to alloc aso wq cq, err=%d\n", err);
+		goto err_out;
+	}
+
+	err = create_aso_cq(cq, cqc_data);
+	if (err) {
+		mlx5_core_err(mdev, "Failed to create aso wq cq, err=%d\n", err);
+		goto err_free_cq;
+	}
+
+	kvfree(cqc_data);
+	return 0;
+
+err_free_cq:
+	mlx5_aso_free_cq(cq);
+err_out:
+	kvfree(cqc_data);
+	return err;
+}
+
+static int mlx5_aso_alloc_sq(struct mlx5_core_dev *mdev, int numa_node,
+			     void *sqc_data, struct mlx5_aso *sq)
+{
+	void *sqc_wq = MLX5_ADDR_OF(sqc, sqc_data, wq);
+	struct mlx5_wq_cyc *wq = &sq->wq;
+	struct mlx5_wq_param param;
+	int err;
+
+	sq->uar_map = mdev->mlx5e_res.hw_objs.bfreg.map;
+
+	param.db_numa_node = numa_node;
+	param.buf_numa_node = numa_node;
+	err = mlx5_wq_cyc_create(mdev, &param, sqc_wq, wq, &sq->wq_ctrl);
+	if (err)
+		return err;
+	wq->db = &wq->db[MLX5_SND_DBR];
+
+	return 0;
+}
+
+static int create_aso_sq(struct mlx5_core_dev *mdev, int pdn,
+			 void *sqc_data, struct mlx5_aso *sq)
+{
+	void *in, *sqc, *wq;
+	int inlen, err;
+	u8 ts_format;
+
+	inlen = MLX5_ST_SZ_BYTES(create_sq_in) +
+		sizeof(u64) * sq->wq_ctrl.buf.npages;
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in)
+		return -ENOMEM;
+
+	sqc = MLX5_ADDR_OF(create_sq_in, in, ctx);
+	wq = MLX5_ADDR_OF(sqc, sqc, wq);
+
+	memcpy(sqc, sqc_data, MLX5_ST_SZ_BYTES(sqc));
+	MLX5_SET(sqc,  sqc, cqn, sq->cq.mcq.cqn);
+
+	MLX5_SET(sqc,  sqc, state, MLX5_SQC_STATE_RST);
+	MLX5_SET(sqc,  sqc, flush_in_error_en, 1);
+
+	ts_format = mlx5_is_real_time_sq(mdev) ?
+			MLX5_TIMESTAMP_FORMAT_REAL_TIME :
+			MLX5_TIMESTAMP_FORMAT_FREE_RUNNING;
+	MLX5_SET(sqc, sqc, ts_format, ts_format);
+
+	MLX5_SET(wq,   wq, wq_type,       MLX5_WQ_TYPE_CYCLIC);
+	MLX5_SET(wq,   wq, uar_page,      mdev->mlx5e_res.hw_objs.bfreg.index);
+	MLX5_SET(wq,   wq, log_wq_pg_sz,  sq->wq_ctrl.buf.page_shift -
+					  MLX5_ADAPTER_PAGE_SHIFT);
+	MLX5_SET64(wq, wq, dbr_addr,      sq->wq_ctrl.db.dma);
+
+	mlx5_fill_page_frag_array(&sq->wq_ctrl.buf,
+				  (__be64 *)MLX5_ADDR_OF(wq, wq, pas));
+
+	err = mlx5_core_create_sq(mdev, in, inlen, &sq->sqn);
+
+	kvfree(in);
+
+	return err;
+}
+
+static int mlx5_aso_set_sq_rdy(struct mlx5_core_dev *mdev, u32 sqn)
+{
+	void *in, *sqc;
+	int inlen, err;
+
+	inlen = MLX5_ST_SZ_BYTES(modify_sq_in);
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in)
+		return -ENOMEM;
+
+	MLX5_SET(modify_sq_in, in, sq_state, MLX5_SQC_STATE_RST);
+	sqc = MLX5_ADDR_OF(modify_sq_in, in, ctx);
+	MLX5_SET(sqc, sqc, state, MLX5_SQC_STATE_RDY);
+
+	err = mlx5_core_modify_sq(mdev, sqn, in);
+
+	kvfree(in);
+
+	return err;
+}
+
+static int mlx5_aso_create_sq_rdy(struct mlx5_core_dev *mdev, u32 pdn,
+				  void *sqc_data, struct mlx5_aso *sq)
+{
+	int err;
+
+	err = create_aso_sq(mdev, pdn, sqc_data, sq);
+	if (err)
+		return err;
+
+	err = mlx5_aso_set_sq_rdy(mdev, sq->sqn);
+	if (err)
+		mlx5_core_destroy_sq(mdev, sq->sqn);
+
+	return err;
+}
+
+static void mlx5_aso_free_sq(struct mlx5_aso *sq)
+{
+	mlx5_wq_destroy(&sq->wq_ctrl);
+}
+
+static void mlx5_aso_destroy_sq(struct mlx5_aso *sq)
+{
+	mlx5_core_destroy_sq(sq->cq.mdev, sq->sqn);
+	mlx5_aso_free_sq(sq);
+}
+
+static int mlx5_aso_create_sq(struct mlx5_core_dev *mdev, int numa_node,
+			      u32 pdn, struct mlx5_aso *sq)
+{
+	void *sqc_data, *wq;
+	int err;
+
+	sqc_data = kvzalloc(MLX5_ST_SZ_BYTES(sqc), GFP_KERNEL);
+	if (!sqc_data)
+		return -ENOMEM;
+
+	wq = MLX5_ADDR_OF(sqc, sqc_data, wq);
+	MLX5_SET(wq, wq, log_wq_stride, ilog2(MLX5_SEND_WQE_BB));
+	MLX5_SET(wq, wq, pd, pdn);
+	MLX5_SET(wq, wq, log_wq_sz, 1);
+
+	err = mlx5_aso_alloc_sq(mdev, numa_node, sqc_data, sq);
+	if (err) {
+		mlx5_core_err(mdev, "Failed to alloc aso wq sq, err=%d\n", err);
+		goto err_out;
+	}
+
+	err = mlx5_aso_create_sq_rdy(mdev, pdn, sqc_data, sq);
+	if (err) {
+		mlx5_core_err(mdev, "Failed to open aso wq sq, err=%d\n", err);
+		goto err_free_asosq;
+	}
+
+	mlx5_core_dbg(mdev, "aso sq->sqn = 0x%x\n", sq->sqn);
+
+	kvfree(sqc_data);
+	return 0;
+
+err_free_asosq:
+	mlx5_aso_free_sq(sq);
+err_out:
+	kvfree(sqc_data);
+	return err;
+}
+
+struct mlx5_aso *mlx5_aso_create(struct mlx5_core_dev *mdev, u32 pdn)
+{
+	int numa_node = dev_to_node(mlx5_core_dma_dev(mdev));
+	struct mlx5_aso *aso;
+	int err;
+
+	aso = kzalloc(sizeof(*aso), GFP_KERNEL);
+	if (!aso)
+		return ERR_PTR(-ENOMEM);
+
+	err = mlx5_aso_create_cq(mdev, numa_node, &aso->cq);
+	if (err)
+		goto err_cq;
+
+	err = mlx5_aso_create_sq(mdev, numa_node, pdn, aso);
+	if (err)
+		goto err_sq;
+
+	return aso;
+
+err_sq:
+	mlx5_aso_destroy_cq(&aso->cq);
+err_cq:
+	kfree(aso);
+	return ERR_PTR(err);
+}
+
+void mlx5_aso_destroy(struct mlx5_aso *aso)
+{
+	if (IS_ERR_OR_NULL(aso))
+		return;
+
+	mlx5_aso_destroy_sq(aso);
+	mlx5_aso_destroy_cq(&aso->cq);
+	kfree(aso);
+}
+
+void mlx5_aso_build_wqe(struct mlx5_aso *aso, u8 ds_cnt,
+			struct mlx5_aso_wqe *aso_wqe,
+			u32 obj_id, u32 opc_mode)
+{
+	struct mlx5_wqe_ctrl_seg *cseg = &aso_wqe->ctrl;
+
+	cseg->opmod_idx_opcode = cpu_to_be32((opc_mode << MLX5_WQE_CTRL_WQE_OPC_MOD_SHIFT) |
+					     (aso->pc << MLX5_WQE_CTRL_WQE_INDEX_SHIFT) |
+					     MLX5_OPCODE_ACCESS_ASO);
+	cseg->qpn_ds     = cpu_to_be32((aso->sqn << MLX5_WQE_CTRL_QPN_SHIFT) | ds_cnt);
+	cseg->fm_ce_se   = MLX5_WQE_CTRL_CQ_UPDATE;
+	cseg->general_id = cpu_to_be32(obj_id);
+}
+
+void *mlx5_aso_get_wqe(struct mlx5_aso *aso)
+{
+	u16 pi;
+
+	pi = mlx5_wq_cyc_ctr2ix(&aso->wq, aso->pc);
+	return mlx5_wq_cyc_get_wqe(&aso->wq, pi);
+}
+
+void mlx5_aso_post_wqe(struct mlx5_aso *aso, bool with_data,
+		       struct mlx5_wqe_ctrl_seg *doorbell_cseg)
+{
+	doorbell_cseg->fm_ce_se |= MLX5_WQE_CTRL_CQ_UPDATE;
+	/* ensure wqe is visible to device before updating doorbell record */
+	dma_wmb();
+
+	if (with_data)
+		aso->pc += MLX5_ASO_WQEBBS_DATA;
+	else
+		aso->pc += MLX5_ASO_WQEBBS;
+	*aso->wq.db = cpu_to_be32(aso->pc);
+
+	/* ensure doorbell record is visible to device before ringing the
+	 * doorbell
+	 */
+	wmb();
+
+	mlx5_write64((__be32 *)doorbell_cseg, aso->uar_map);
+
+	/* Ensure doorbell is written on uar_page before poll_cq */
+	WRITE_ONCE(doorbell_cseg, NULL);
+}
+
+int mlx5_aso_poll_cq(struct mlx5_aso *aso, bool with_data)
+{
+	struct mlx5_aso_cq *cq = &aso->cq;
+	struct mlx5_cqe64 *cqe;
+
+	cqe = mlx5_cqwq_get_cqe(&cq->wq);
+	if (!cqe)
+		return -ETIMEDOUT;
+
+	/* sq->cc must be updated only after mlx5_cqwq_update_db_record(),
+	 * otherwise a cq overrun may occur
+	 */
+	mlx5_cqwq_pop(&cq->wq);
+
+	if (unlikely(get_cqe_opcode(cqe) != MLX5_CQE_REQ)) {
+		struct mlx5_err_cqe *err_cqe;
+
+		mlx5_core_err(cq->mdev, "Bad OP in ASOSQ CQE: 0x%x\n",
+			      get_cqe_opcode(cqe));
+
+		err_cqe = (struct mlx5_err_cqe *)cqe;
+		mlx5_core_err(cq->mdev, "vendor_err_synd=%x\n",
+			      err_cqe->vendor_err_synd);
+		mlx5_core_err(cq->mdev, "syndrome=%x\n",
+			      err_cqe->syndrome);
+		print_hex_dump(KERN_WARNING, "", DUMP_PREFIX_OFFSET,
+			       16, 1, err_cqe,
+			       sizeof(*err_cqe), false);
+	}
+
+	mlx5_cqwq_update_db_record(&cq->wq);
+
+	/* ensure cq space is freed before enabling more cqes */
+	wmb();
+
+	if (with_data)
+		aso->cc += MLX5_ASO_WQEBBS_DATA;
+	else
+		aso->cc += MLX5_ASO_WQEBBS;
+
+	return 0;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/aso.h b/drivers/net/ethernet/mellanox/mlx5/core/lib/aso.h
new file mode 100644
index 000000000..2d40dcf9d
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/aso.h
@@ -0,0 +1,90 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2022, NVIDIA CORPORATION & AFFILIATES. All rights reserved. */
+
+#ifndef __MLX5_LIB_ASO_H__
+#define __MLX5_LIB_ASO_H__
+
+#include <linux/mlx5/qp.h>
+#include "mlx5_core.h"
+
+#define MLX5_ASO_WQEBBS \
+	(DIV_ROUND_UP(sizeof(struct mlx5_aso_wqe), MLX5_SEND_WQE_BB))
+#define MLX5_ASO_WQEBBS_DATA \
+	(DIV_ROUND_UP(sizeof(struct mlx5_aso_wqe_data), MLX5_SEND_WQE_BB))
+#define ASO_CTRL_READ_EN BIT(0)
+#define MLX5_WQE_CTRL_WQE_OPC_MOD_SHIFT 24
+#define MLX5_MACSEC_ASO_DS_CNT (DIV_ROUND_UP(sizeof(struct mlx5_aso_wqe), MLX5_SEND_WQE_DS))
+
+struct mlx5_wqe_aso_ctrl_seg {
+	__be32  va_h;
+	__be32  va_l; /* include read_enable */
+	__be32  l_key;
+	u8      data_mask_mode;
+	u8      condition_1_0_operand;
+	u8      condition_1_0_offset;
+	u8      data_offset_condition_operand;
+	__be32  condition_0_data;
+	__be32  condition_0_mask;
+	__be32  condition_1_data;
+	__be32  condition_1_mask;
+	__be64  bitwise_data;
+	__be64  data_mask;
+};
+
+struct mlx5_wqe_aso_data_seg {
+	__be32  bytewise_data[16];
+};
+
+struct mlx5_aso_wqe {
+	struct mlx5_wqe_ctrl_seg      ctrl;
+	struct mlx5_wqe_aso_ctrl_seg  aso_ctrl;
+};
+
+struct mlx5_aso_wqe_data {
+	struct mlx5_wqe_ctrl_seg      ctrl;
+	struct mlx5_wqe_aso_ctrl_seg  aso_ctrl;
+	struct mlx5_wqe_aso_data_seg  aso_data;
+};
+
+enum {
+	MLX5_ASO_LOGICAL_AND,
+	MLX5_ASO_LOGICAL_OR,
+};
+
+enum {
+	MLX5_ASO_ALWAYS_FALSE,
+	MLX5_ASO_ALWAYS_TRUE,
+	MLX5_ASO_EQUAL,
+	MLX5_ASO_NOT_EQUAL,
+	MLX5_ASO_GREATER_OR_EQUAL,
+	MLX5_ASO_LESSER_OR_EQUAL,
+	MLX5_ASO_LESSER,
+	MLX5_ASO_GREATER,
+	MLX5_ASO_CYCLIC_GREATER,
+	MLX5_ASO_CYCLIC_LESSER,
+};
+
+enum {
+	MLX5_ASO_DATA_MASK_MODE_BITWISE_64BIT,
+	MLX5_ASO_DATA_MASK_MODE_BYTEWISE_64BYTE,
+	MLX5_ASO_DATA_MASK_MODE_CALCULATED_64BYTE,
+};
+
+enum {
+	MLX5_ACCESS_ASO_OPC_MOD_FLOW_METER = 0x2,
+	MLX5_ACCESS_ASO_OPC_MOD_MACSEC = 0x5,
+};
+
+struct mlx5_aso;
+
+void *mlx5_aso_get_wqe(struct mlx5_aso *aso);
+void mlx5_aso_build_wqe(struct mlx5_aso *aso, u8 ds_cnt,
+			struct mlx5_aso_wqe *aso_wqe,
+			u32 obj_id, u32 opc_mode);
+void mlx5_aso_post_wqe(struct mlx5_aso *aso, bool with_data,
+		       struct mlx5_wqe_ctrl_seg *doorbell_cseg);
+int mlx5_aso_poll_cq(struct mlx5_aso *aso, bool with_data);
+
+struct mlx5_aso *mlx5_aso_create(struct mlx5_core_dev *mdev, u32 pdn);
+void mlx5_aso_destroy(struct mlx5_aso *aso);
+#endif /* __MLX5_LIB_ASO_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/clock.c b/drivers/net/ethernet/mellanox/mlx5/core/lib/clock.c
new file mode 100644
index 000000000..2ac255bb9
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/clock.c
@@ -0,0 +1,1026 @@
+/*
+ * Copyright (c) 2015, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/clocksource.h>
+#include <linux/highmem.h>
+#include <linux/ptp_clock_kernel.h>
+#include <rdma/mlx5-abi.h>
+#include "lib/eq.h"
+#include "en.h"
+#include "clock.h"
+
+enum {
+	MLX5_CYCLES_SHIFT	= 23
+};
+
+enum {
+	MLX5_PIN_MODE_IN		= 0x0,
+	MLX5_PIN_MODE_OUT		= 0x1,
+};
+
+enum {
+	MLX5_OUT_PATTERN_PULSE		= 0x0,
+	MLX5_OUT_PATTERN_PERIODIC	= 0x1,
+};
+
+enum {
+	MLX5_EVENT_MODE_DISABLE	= 0x0,
+	MLX5_EVENT_MODE_REPETETIVE	= 0x1,
+	MLX5_EVENT_MODE_ONCE_TILL_ARM	= 0x2,
+};
+
+enum {
+	MLX5_MTPPS_FS_ENABLE			= BIT(0x0),
+	MLX5_MTPPS_FS_PATTERN			= BIT(0x2),
+	MLX5_MTPPS_FS_PIN_MODE			= BIT(0x3),
+	MLX5_MTPPS_FS_TIME_STAMP		= BIT(0x4),
+	MLX5_MTPPS_FS_OUT_PULSE_DURATION	= BIT(0x5),
+	MLX5_MTPPS_FS_ENH_OUT_PER_ADJ		= BIT(0x7),
+	MLX5_MTPPS_FS_NPPS_PERIOD               = BIT(0x9),
+	MLX5_MTPPS_FS_OUT_PULSE_DURATION_NS     = BIT(0xa),
+};
+
+static bool mlx5_real_time_mode(struct mlx5_core_dev *mdev)
+{
+	return (mlx5_is_real_time_rq(mdev) || mlx5_is_real_time_sq(mdev));
+}
+
+static bool mlx5_npps_real_time_supported(struct mlx5_core_dev *mdev)
+{
+	return (mlx5_real_time_mode(mdev) &&
+		MLX5_CAP_MCAM_FEATURE(mdev, npps_period) &&
+		MLX5_CAP_MCAM_FEATURE(mdev, out_pulse_duration_ns));
+}
+
+static bool mlx5_modify_mtutc_allowed(struct mlx5_core_dev *mdev)
+{
+	return MLX5_CAP_MCAM_FEATURE(mdev, ptpcyc2realtime_modify);
+}
+
+static int mlx5_set_mtutc(struct mlx5_core_dev *dev, u32 *mtutc, u32 size)
+{
+	u32 out[MLX5_ST_SZ_DW(mtutc_reg)] = {};
+
+	if (!MLX5_CAP_MCAM_REG(dev, mtutc))
+		return -EOPNOTSUPP;
+
+	return mlx5_core_access_reg(dev, mtutc, size, out, sizeof(out),
+				    MLX5_REG_MTUTC, 0, 1);
+}
+
+static u64 mlx5_read_time(struct mlx5_core_dev *dev,
+			  struct ptp_system_timestamp *sts,
+			  bool real_time)
+{
+	u32 timer_h, timer_h1, timer_l;
+
+	timer_h = ioread32be(real_time ? &dev->iseg->real_time_h :
+			     &dev->iseg->internal_timer_h);
+	ptp_read_system_prets(sts);
+	timer_l = ioread32be(real_time ? &dev->iseg->real_time_l :
+			     &dev->iseg->internal_timer_l);
+	ptp_read_system_postts(sts);
+	timer_h1 = ioread32be(real_time ? &dev->iseg->real_time_h :
+			      &dev->iseg->internal_timer_h);
+	if (timer_h != timer_h1) {
+		/* wrap around */
+		ptp_read_system_prets(sts);
+		timer_l = ioread32be(real_time ? &dev->iseg->real_time_l :
+				     &dev->iseg->internal_timer_l);
+		ptp_read_system_postts(sts);
+	}
+
+	return real_time ? REAL_TIME_TO_NS(timer_h1, timer_l) :
+			   (u64)timer_l | (u64)timer_h1 << 32;
+}
+
+static u64 read_internal_timer(const struct cyclecounter *cc)
+{
+	struct mlx5_timer *timer = container_of(cc, struct mlx5_timer, cycles);
+	struct mlx5_clock *clock = container_of(timer, struct mlx5_clock, timer);
+	struct mlx5_core_dev *mdev = container_of(clock, struct mlx5_core_dev,
+						  clock);
+
+	return mlx5_read_time(mdev, NULL, false) & cc->mask;
+}
+
+static void mlx5_update_clock_info_page(struct mlx5_core_dev *mdev)
+{
+	struct mlx5_ib_clock_info *clock_info = mdev->clock_info;
+	struct mlx5_clock *clock = &mdev->clock;
+	struct mlx5_timer *timer;
+	u32 sign;
+
+	if (!clock_info)
+		return;
+
+	sign = smp_load_acquire(&clock_info->sign);
+	smp_store_mb(clock_info->sign,
+		     sign | MLX5_IB_CLOCK_INFO_KERNEL_UPDATING);
+
+	timer = &clock->timer;
+	clock_info->cycles = timer->tc.cycle_last;
+	clock_info->mult   = timer->cycles.mult;
+	clock_info->nsec   = timer->tc.nsec;
+	clock_info->frac   = timer->tc.frac;
+
+	smp_store_release(&clock_info->sign,
+			  sign + MLX5_IB_CLOCK_INFO_KERNEL_UPDATING * 2);
+}
+
+static void mlx5_pps_out(struct work_struct *work)
+{
+	struct mlx5_pps *pps_info = container_of(work, struct mlx5_pps,
+						 out_work);
+	struct mlx5_clock *clock = container_of(pps_info, struct mlx5_clock,
+						pps_info);
+	struct mlx5_core_dev *mdev = container_of(clock, struct mlx5_core_dev,
+						  clock);
+	u32 in[MLX5_ST_SZ_DW(mtpps_reg)] = {0};
+	unsigned long flags;
+	int i;
+
+	for (i = 0; i < clock->ptp_info.n_pins; i++) {
+		u64 tstart;
+
+		write_seqlock_irqsave(&clock->lock, flags);
+		tstart = clock->pps_info.start[i];
+		clock->pps_info.start[i] = 0;
+		write_sequnlock_irqrestore(&clock->lock, flags);
+		if (!tstart)
+			continue;
+
+		MLX5_SET(mtpps_reg, in, pin, i);
+		MLX5_SET64(mtpps_reg, in, time_stamp, tstart);
+		MLX5_SET(mtpps_reg, in, field_select, MLX5_MTPPS_FS_TIME_STAMP);
+		mlx5_set_mtpps(mdev, in, sizeof(in));
+	}
+}
+
+static void mlx5_timestamp_overflow(struct work_struct *work)
+{
+	struct delayed_work *dwork = to_delayed_work(work);
+	struct mlx5_core_dev *mdev;
+	struct mlx5_timer *timer;
+	struct mlx5_clock *clock;
+	unsigned long flags;
+
+	timer = container_of(dwork, struct mlx5_timer, overflow_work);
+	clock = container_of(timer, struct mlx5_clock, timer);
+	mdev = container_of(clock, struct mlx5_core_dev, clock);
+
+	if (mdev->state == MLX5_DEVICE_STATE_INTERNAL_ERROR)
+		goto out;
+
+	write_seqlock_irqsave(&clock->lock, flags);
+	timecounter_read(&timer->tc);
+	mlx5_update_clock_info_page(mdev);
+	write_sequnlock_irqrestore(&clock->lock, flags);
+
+out:
+	schedule_delayed_work(&timer->overflow_work, timer->overflow_period);
+}
+
+static int mlx5_ptp_settime_real_time(struct mlx5_core_dev *mdev,
+				      const struct timespec64 *ts)
+{
+	u32 in[MLX5_ST_SZ_DW(mtutc_reg)] = {};
+
+	if (!mlx5_modify_mtutc_allowed(mdev))
+		return 0;
+
+	if (ts->tv_sec < 0 || ts->tv_sec > U32_MAX ||
+	    ts->tv_nsec < 0 || ts->tv_nsec > NSEC_PER_SEC)
+		return -EINVAL;
+
+	MLX5_SET(mtutc_reg, in, operation, MLX5_MTUTC_OPERATION_SET_TIME_IMMEDIATE);
+	MLX5_SET(mtutc_reg, in, utc_sec, ts->tv_sec);
+	MLX5_SET(mtutc_reg, in, utc_nsec, ts->tv_nsec);
+
+	return mlx5_set_mtutc(mdev, in, sizeof(in));
+}
+
+static int mlx5_ptp_settime(struct ptp_clock_info *ptp, const struct timespec64 *ts)
+{
+	struct mlx5_clock *clock = container_of(ptp, struct mlx5_clock, ptp_info);
+	struct mlx5_timer *timer = &clock->timer;
+	struct mlx5_core_dev *mdev;
+	unsigned long flags;
+	int err;
+
+	mdev = container_of(clock, struct mlx5_core_dev, clock);
+	err = mlx5_ptp_settime_real_time(mdev, ts);
+	if (err)
+		return err;
+
+	write_seqlock_irqsave(&clock->lock, flags);
+	timecounter_init(&timer->tc, &timer->cycles, timespec64_to_ns(ts));
+	mlx5_update_clock_info_page(mdev);
+	write_sequnlock_irqrestore(&clock->lock, flags);
+
+	return 0;
+}
+
+static
+struct timespec64 mlx5_ptp_gettimex_real_time(struct mlx5_core_dev *mdev,
+					      struct ptp_system_timestamp *sts)
+{
+	struct timespec64 ts;
+	u64 time;
+
+	time = mlx5_read_time(mdev, sts, true);
+	ts = ns_to_timespec64(time);
+	return ts;
+}
+
+static int mlx5_ptp_gettimex(struct ptp_clock_info *ptp, struct timespec64 *ts,
+			     struct ptp_system_timestamp *sts)
+{
+	struct mlx5_clock *clock = container_of(ptp, struct mlx5_clock, ptp_info);
+	struct mlx5_timer *timer = &clock->timer;
+	struct mlx5_core_dev *mdev;
+	unsigned long flags;
+	u64 cycles, ns;
+
+	mdev = container_of(clock, struct mlx5_core_dev, clock);
+	if (mlx5_real_time_mode(mdev)) {
+		*ts = mlx5_ptp_gettimex_real_time(mdev, sts);
+		goto out;
+	}
+
+	write_seqlock_irqsave(&clock->lock, flags);
+	cycles = mlx5_read_time(mdev, sts, false);
+	ns = timecounter_cyc2time(&timer->tc, cycles);
+	write_sequnlock_irqrestore(&clock->lock, flags);
+	*ts = ns_to_timespec64(ns);
+out:
+	return 0;
+}
+
+static int mlx5_ptp_adjtime_real_time(struct mlx5_core_dev *mdev, s64 delta)
+{
+	u32 in[MLX5_ST_SZ_DW(mtutc_reg)] = {};
+
+	if (!mlx5_modify_mtutc_allowed(mdev))
+		return 0;
+
+	/* HW time adjustment range is s16. If out of range, settime instead */
+	if (delta < S16_MIN || delta > S16_MAX) {
+		struct timespec64 ts;
+		s64 ns;
+
+		ts = mlx5_ptp_gettimex_real_time(mdev, NULL);
+		ns = timespec64_to_ns(&ts) + delta;
+		ts = ns_to_timespec64(ns);
+		return mlx5_ptp_settime_real_time(mdev, &ts);
+	}
+
+	MLX5_SET(mtutc_reg, in, operation, MLX5_MTUTC_OPERATION_ADJUST_TIME);
+	MLX5_SET(mtutc_reg, in, time_adjustment, delta);
+
+	return mlx5_set_mtutc(mdev, in, sizeof(in));
+}
+
+static int mlx5_ptp_adjtime(struct ptp_clock_info *ptp, s64 delta)
+{
+	struct mlx5_clock *clock = container_of(ptp, struct mlx5_clock, ptp_info);
+	struct mlx5_timer *timer = &clock->timer;
+	struct mlx5_core_dev *mdev;
+	unsigned long flags;
+	int err;
+
+	mdev = container_of(clock, struct mlx5_core_dev, clock);
+
+	err = mlx5_ptp_adjtime_real_time(mdev, delta);
+	if (err)
+		return err;
+	write_seqlock_irqsave(&clock->lock, flags);
+	timecounter_adjtime(&timer->tc, delta);
+	mlx5_update_clock_info_page(mdev);
+	write_sequnlock_irqrestore(&clock->lock, flags);
+
+	return 0;
+}
+
+static int mlx5_ptp_adjfreq_real_time(struct mlx5_core_dev *mdev, s32 freq)
+{
+	u32 in[MLX5_ST_SZ_DW(mtutc_reg)] = {};
+
+	if (!mlx5_modify_mtutc_allowed(mdev))
+		return 0;
+
+	MLX5_SET(mtutc_reg, in, operation, MLX5_MTUTC_OPERATION_ADJUST_FREQ_UTC);
+	MLX5_SET(mtutc_reg, in, freq_adjustment, freq);
+
+	return mlx5_set_mtutc(mdev, in, sizeof(in));
+}
+
+static int mlx5_ptp_adjfreq(struct ptp_clock_info *ptp, s32 delta)
+{
+	struct mlx5_clock *clock = container_of(ptp, struct mlx5_clock, ptp_info);
+	struct mlx5_timer *timer = &clock->timer;
+	struct mlx5_core_dev *mdev;
+	unsigned long flags;
+	int neg_adj = 0;
+	u32 diff;
+	u64 adj;
+	int err;
+
+	mdev = container_of(clock, struct mlx5_core_dev, clock);
+	err = mlx5_ptp_adjfreq_real_time(mdev, delta);
+	if (err)
+		return err;
+
+	if (delta < 0) {
+		neg_adj = 1;
+		delta = -delta;
+	}
+
+	adj = timer->nominal_c_mult;
+	adj *= delta;
+	diff = div_u64(adj, 1000000000ULL);
+
+	write_seqlock_irqsave(&clock->lock, flags);
+	timecounter_read(&timer->tc);
+	timer->cycles.mult = neg_adj ? timer->nominal_c_mult - diff :
+				       timer->nominal_c_mult + diff;
+	mlx5_update_clock_info_page(mdev);
+	write_sequnlock_irqrestore(&clock->lock, flags);
+
+	return 0;
+}
+
+static int mlx5_extts_configure(struct ptp_clock_info *ptp,
+				struct ptp_clock_request *rq,
+				int on)
+{
+	struct mlx5_clock *clock =
+			container_of(ptp, struct mlx5_clock, ptp_info);
+	struct mlx5_core_dev *mdev =
+			container_of(clock, struct mlx5_core_dev, clock);
+	u32 in[MLX5_ST_SZ_DW(mtpps_reg)] = {0};
+	u32 field_select = 0;
+	u8 pin_mode = 0;
+	u8 pattern = 0;
+	int pin = -1;
+	int err = 0;
+
+	if (!MLX5_PPS_CAP(mdev))
+		return -EOPNOTSUPP;
+
+	/* Reject requests with unsupported flags */
+	if (rq->extts.flags & ~(PTP_ENABLE_FEATURE |
+				PTP_RISING_EDGE |
+				PTP_FALLING_EDGE |
+				PTP_STRICT_FLAGS))
+		return -EOPNOTSUPP;
+
+	/* Reject requests to enable time stamping on both edges. */
+	if ((rq->extts.flags & PTP_STRICT_FLAGS) &&
+	    (rq->extts.flags & PTP_ENABLE_FEATURE) &&
+	    (rq->extts.flags & PTP_EXTTS_EDGES) == PTP_EXTTS_EDGES)
+		return -EOPNOTSUPP;
+
+	if (rq->extts.index >= clock->ptp_info.n_pins)
+		return -EINVAL;
+
+	pin = ptp_find_pin(clock->ptp, PTP_PF_EXTTS, rq->extts.index);
+	if (pin < 0)
+		return -EBUSY;
+
+	if (on) {
+		pin_mode = MLX5_PIN_MODE_IN;
+		pattern = !!(rq->extts.flags & PTP_FALLING_EDGE);
+		field_select = MLX5_MTPPS_FS_PIN_MODE |
+			       MLX5_MTPPS_FS_PATTERN |
+			       MLX5_MTPPS_FS_ENABLE;
+	} else {
+		field_select = MLX5_MTPPS_FS_ENABLE;
+	}
+
+	MLX5_SET(mtpps_reg, in, pin, pin);
+	MLX5_SET(mtpps_reg, in, pin_mode, pin_mode);
+	MLX5_SET(mtpps_reg, in, pattern, pattern);
+	MLX5_SET(mtpps_reg, in, enable, on);
+	MLX5_SET(mtpps_reg, in, field_select, field_select);
+
+	err = mlx5_set_mtpps(mdev, in, sizeof(in));
+	if (err)
+		return err;
+
+	return mlx5_set_mtppse(mdev, pin, 0,
+			       MLX5_EVENT_MODE_REPETETIVE & on);
+}
+
+static u64 find_target_cycles(struct mlx5_core_dev *mdev, s64 target_ns)
+{
+	struct mlx5_clock *clock = &mdev->clock;
+	u64 cycles_now, cycles_delta;
+	u64 nsec_now, nsec_delta;
+	struct mlx5_timer *timer;
+	unsigned long flags;
+
+	timer = &clock->timer;
+
+	cycles_now = mlx5_read_time(mdev, NULL, false);
+	write_seqlock_irqsave(&clock->lock, flags);
+	nsec_now = timecounter_cyc2time(&timer->tc, cycles_now);
+	nsec_delta = target_ns - nsec_now;
+	cycles_delta = div64_u64(nsec_delta << timer->cycles.shift,
+				 timer->cycles.mult);
+	write_sequnlock_irqrestore(&clock->lock, flags);
+
+	return cycles_now + cycles_delta;
+}
+
+static u64 perout_conf_internal_timer(struct mlx5_core_dev *mdev, s64 sec)
+{
+	struct timespec64 ts = {};
+	s64 target_ns;
+
+	ts.tv_sec = sec;
+	target_ns = timespec64_to_ns(&ts);
+
+	return find_target_cycles(mdev, target_ns);
+}
+
+static u64 perout_conf_real_time(s64 sec, u32 nsec)
+{
+	return (u64)nsec | (u64)sec << 32;
+}
+
+static int perout_conf_1pps(struct mlx5_core_dev *mdev, struct ptp_clock_request *rq,
+			    u64 *time_stamp, bool real_time)
+{
+	struct timespec64 ts;
+	s64 ns;
+
+	ts.tv_nsec = rq->perout.period.nsec;
+	ts.tv_sec = rq->perout.period.sec;
+	ns = timespec64_to_ns(&ts);
+
+	if ((ns >> 1) != 500000000LL)
+		return -EINVAL;
+
+	*time_stamp = real_time ? perout_conf_real_time(rq->perout.start.sec, 0) :
+		      perout_conf_internal_timer(mdev, rq->perout.start.sec);
+
+	return 0;
+}
+
+#define MLX5_MAX_PULSE_DURATION (BIT(__mlx5_bit_sz(mtpps_reg, out_pulse_duration_ns)) - 1)
+static int mlx5_perout_conf_out_pulse_duration(struct mlx5_core_dev *mdev,
+					       struct ptp_clock_request *rq,
+					       u32 *out_pulse_duration_ns)
+{
+	struct mlx5_pps *pps_info = &mdev->clock.pps_info;
+	u32 out_pulse_duration;
+	struct timespec64 ts;
+
+	if (rq->perout.flags & PTP_PEROUT_DUTY_CYCLE) {
+		ts.tv_sec = rq->perout.on.sec;
+		ts.tv_nsec = rq->perout.on.nsec;
+		out_pulse_duration = (u32)timespec64_to_ns(&ts);
+	} else {
+		/* out_pulse_duration_ns should be up to 50% of the
+		 * pulse period as default
+		 */
+		ts.tv_sec = rq->perout.period.sec;
+		ts.tv_nsec = rq->perout.period.nsec;
+		out_pulse_duration = (u32)timespec64_to_ns(&ts) >> 1;
+	}
+
+	if (out_pulse_duration < pps_info->min_out_pulse_duration_ns ||
+	    out_pulse_duration > MLX5_MAX_PULSE_DURATION) {
+		mlx5_core_err(mdev, "NPPS pulse duration %u is not in [%llu, %lu]\n",
+			      out_pulse_duration, pps_info->min_out_pulse_duration_ns,
+			      MLX5_MAX_PULSE_DURATION);
+		return -EINVAL;
+	}
+	*out_pulse_duration_ns = out_pulse_duration;
+
+	return 0;
+}
+
+static int perout_conf_npps_real_time(struct mlx5_core_dev *mdev, struct ptp_clock_request *rq,
+				      u32 *field_select, u32 *out_pulse_duration_ns,
+				      u64 *period, u64 *time_stamp)
+{
+	struct mlx5_pps *pps_info = &mdev->clock.pps_info;
+	struct ptp_clock_time *time = &rq->perout.start;
+	struct timespec64 ts;
+
+	ts.tv_sec = rq->perout.period.sec;
+	ts.tv_nsec = rq->perout.period.nsec;
+	if (timespec64_to_ns(&ts) < pps_info->min_npps_period) {
+		mlx5_core_err(mdev, "NPPS period is lower than minimal npps period %llu\n",
+			      pps_info->min_npps_period);
+		return -EINVAL;
+	}
+	*period = perout_conf_real_time(rq->perout.period.sec, rq->perout.period.nsec);
+
+	if (mlx5_perout_conf_out_pulse_duration(mdev, rq, out_pulse_duration_ns))
+		return -EINVAL;
+
+	*time_stamp = perout_conf_real_time(time->sec, time->nsec);
+	*field_select |= MLX5_MTPPS_FS_NPPS_PERIOD |
+			 MLX5_MTPPS_FS_OUT_PULSE_DURATION_NS;
+
+	return 0;
+}
+
+static bool mlx5_perout_verify_flags(struct mlx5_core_dev *mdev, unsigned int flags)
+{
+	return ((!mlx5_npps_real_time_supported(mdev) && flags) ||
+		(mlx5_npps_real_time_supported(mdev) && flags & ~PTP_PEROUT_DUTY_CYCLE));
+}
+
+static int mlx5_perout_configure(struct ptp_clock_info *ptp,
+				 struct ptp_clock_request *rq,
+				 int on)
+{
+	struct mlx5_clock *clock =
+			container_of(ptp, struct mlx5_clock, ptp_info);
+	struct mlx5_core_dev *mdev =
+			container_of(clock, struct mlx5_core_dev, clock);
+	bool rt_mode = mlx5_real_time_mode(mdev);
+	u32 in[MLX5_ST_SZ_DW(mtpps_reg)] = {0};
+	u32 out_pulse_duration_ns = 0;
+	u32 field_select = 0;
+	u64 npps_period = 0;
+	u64 time_stamp = 0;
+	u8 pin_mode = 0;
+	u8 pattern = 0;
+	int pin = -1;
+	int err = 0;
+
+	if (!MLX5_PPS_CAP(mdev))
+		return -EOPNOTSUPP;
+
+	/* Reject requests with unsupported flags */
+	if (mlx5_perout_verify_flags(mdev, rq->perout.flags))
+		return -EOPNOTSUPP;
+
+	if (rq->perout.index >= clock->ptp_info.n_pins)
+		return -EINVAL;
+
+	field_select = MLX5_MTPPS_FS_ENABLE;
+	pin = ptp_find_pin(clock->ptp, PTP_PF_PEROUT, rq->perout.index);
+	if (pin < 0)
+		return -EBUSY;
+
+	if (on) {
+		bool rt_mode = mlx5_real_time_mode(mdev);
+
+		pin_mode = MLX5_PIN_MODE_OUT;
+		pattern = MLX5_OUT_PATTERN_PERIODIC;
+
+		if (rt_mode &&  rq->perout.start.sec > U32_MAX)
+			return -EINVAL;
+
+		field_select |= MLX5_MTPPS_FS_PIN_MODE |
+				MLX5_MTPPS_FS_PATTERN |
+				MLX5_MTPPS_FS_TIME_STAMP;
+
+		if (mlx5_npps_real_time_supported(mdev))
+			err = perout_conf_npps_real_time(mdev, rq, &field_select,
+							 &out_pulse_duration_ns, &npps_period,
+							 &time_stamp);
+		else
+			err = perout_conf_1pps(mdev, rq, &time_stamp, rt_mode);
+		if (err)
+			return err;
+	}
+
+	MLX5_SET(mtpps_reg, in, pin, pin);
+	MLX5_SET(mtpps_reg, in, pin_mode, pin_mode);
+	MLX5_SET(mtpps_reg, in, pattern, pattern);
+	MLX5_SET(mtpps_reg, in, enable, on);
+	MLX5_SET64(mtpps_reg, in, time_stamp, time_stamp);
+	MLX5_SET(mtpps_reg, in, field_select, field_select);
+	MLX5_SET64(mtpps_reg, in, npps_period, npps_period);
+	MLX5_SET(mtpps_reg, in, out_pulse_duration_ns, out_pulse_duration_ns);
+	err = mlx5_set_mtpps(mdev, in, sizeof(in));
+	if (err)
+		return err;
+
+	if (rt_mode)
+		return 0;
+
+	return mlx5_set_mtppse(mdev, pin, 0,
+			       MLX5_EVENT_MODE_REPETETIVE & on);
+}
+
+static int mlx5_pps_configure(struct ptp_clock_info *ptp,
+			      struct ptp_clock_request *rq,
+			      int on)
+{
+	struct mlx5_clock *clock =
+			container_of(ptp, struct mlx5_clock, ptp_info);
+
+	clock->pps_info.enabled = !!on;
+	return 0;
+}
+
+static int mlx5_ptp_enable(struct ptp_clock_info *ptp,
+			   struct ptp_clock_request *rq,
+			   int on)
+{
+	switch (rq->type) {
+	case PTP_CLK_REQ_EXTTS:
+		return mlx5_extts_configure(ptp, rq, on);
+	case PTP_CLK_REQ_PEROUT:
+		return mlx5_perout_configure(ptp, rq, on);
+	case PTP_CLK_REQ_PPS:
+		return mlx5_pps_configure(ptp, rq, on);
+	default:
+		return -EOPNOTSUPP;
+	}
+	return 0;
+}
+
+enum {
+	MLX5_MTPPS_REG_CAP_PIN_X_MODE_SUPPORT_PPS_IN = BIT(0),
+	MLX5_MTPPS_REG_CAP_PIN_X_MODE_SUPPORT_PPS_OUT = BIT(1),
+};
+
+static int mlx5_ptp_verify(struct ptp_clock_info *ptp, unsigned int pin,
+			   enum ptp_pin_function func, unsigned int chan)
+{
+	struct mlx5_clock *clock = container_of(ptp, struct mlx5_clock,
+						ptp_info);
+
+	switch (func) {
+	case PTP_PF_NONE:
+		return 0;
+	case PTP_PF_EXTTS:
+		return !(clock->pps_info.pin_caps[pin] &
+			 MLX5_MTPPS_REG_CAP_PIN_X_MODE_SUPPORT_PPS_IN);
+	case PTP_PF_PEROUT:
+		return !(clock->pps_info.pin_caps[pin] &
+			 MLX5_MTPPS_REG_CAP_PIN_X_MODE_SUPPORT_PPS_OUT);
+	default:
+		return -EOPNOTSUPP;
+	}
+}
+
+static const struct ptp_clock_info mlx5_ptp_clock_info = {
+	.owner		= THIS_MODULE,
+	.name		= "mlx5_ptp",
+	.max_adj	= 50000000,
+	.n_alarm	= 0,
+	.n_ext_ts	= 0,
+	.n_per_out	= 0,
+	.n_pins		= 0,
+	.pps		= 0,
+	.adjfreq	= mlx5_ptp_adjfreq,
+	.adjtime	= mlx5_ptp_adjtime,
+	.gettimex64	= mlx5_ptp_gettimex,
+	.settime64	= mlx5_ptp_settime,
+	.enable		= NULL,
+	.verify		= NULL,
+};
+
+static int mlx5_query_mtpps_pin_mode(struct mlx5_core_dev *mdev, u8 pin,
+				     u32 *mtpps, u32 mtpps_size)
+{
+	u32 in[MLX5_ST_SZ_DW(mtpps_reg)] = {};
+
+	MLX5_SET(mtpps_reg, in, pin, pin);
+
+	return mlx5_core_access_reg(mdev, in, sizeof(in), mtpps,
+				    mtpps_size, MLX5_REG_MTPPS, 0, 0);
+}
+
+static int mlx5_get_pps_pin_mode(struct mlx5_clock *clock, u8 pin)
+{
+	struct mlx5_core_dev *mdev = container_of(clock, struct mlx5_core_dev, clock);
+
+	u32 out[MLX5_ST_SZ_DW(mtpps_reg)] = {};
+	u8 mode;
+	int err;
+
+	err = mlx5_query_mtpps_pin_mode(mdev, pin, out, sizeof(out));
+	if (err || !MLX5_GET(mtpps_reg, out, enable))
+		return PTP_PF_NONE;
+
+	mode = MLX5_GET(mtpps_reg, out, pin_mode);
+
+	if (mode == MLX5_PIN_MODE_IN)
+		return PTP_PF_EXTTS;
+	else if (mode == MLX5_PIN_MODE_OUT)
+		return PTP_PF_PEROUT;
+
+	return PTP_PF_NONE;
+}
+
+static void mlx5_init_pin_config(struct mlx5_clock *clock)
+{
+	int i;
+
+	if (!clock->ptp_info.n_pins)
+		return;
+
+	clock->ptp_info.pin_config =
+			kcalloc(clock->ptp_info.n_pins,
+				sizeof(*clock->ptp_info.pin_config),
+				GFP_KERNEL);
+	if (!clock->ptp_info.pin_config)
+		return;
+	clock->ptp_info.enable = mlx5_ptp_enable;
+	clock->ptp_info.verify = mlx5_ptp_verify;
+	clock->ptp_info.pps = 1;
+
+	for (i = 0; i < clock->ptp_info.n_pins; i++) {
+		snprintf(clock->ptp_info.pin_config[i].name,
+			 sizeof(clock->ptp_info.pin_config[i].name),
+			 "mlx5_pps%d", i);
+		clock->ptp_info.pin_config[i].index = i;
+		clock->ptp_info.pin_config[i].func = mlx5_get_pps_pin_mode(clock, i);
+		clock->ptp_info.pin_config[i].chan = 0;
+	}
+}
+
+static void mlx5_get_pps_caps(struct mlx5_core_dev *mdev)
+{
+	struct mlx5_clock *clock = &mdev->clock;
+	u32 out[MLX5_ST_SZ_DW(mtpps_reg)] = {0};
+
+	mlx5_query_mtpps(mdev, out, sizeof(out));
+
+	clock->ptp_info.n_pins = MLX5_GET(mtpps_reg, out,
+					  cap_number_of_pps_pins);
+	clock->ptp_info.n_ext_ts = MLX5_GET(mtpps_reg, out,
+					    cap_max_num_of_pps_in_pins);
+	clock->ptp_info.n_per_out = MLX5_GET(mtpps_reg, out,
+					     cap_max_num_of_pps_out_pins);
+
+	if (MLX5_CAP_MCAM_FEATURE(mdev, npps_period))
+		clock->pps_info.min_npps_period = 1 << MLX5_GET(mtpps_reg, out,
+								cap_log_min_npps_period);
+	if (MLX5_CAP_MCAM_FEATURE(mdev, out_pulse_duration_ns))
+		clock->pps_info.min_out_pulse_duration_ns = 1 << MLX5_GET(mtpps_reg, out,
+								cap_log_min_out_pulse_duration_ns);
+
+	clock->pps_info.pin_caps[0] = MLX5_GET(mtpps_reg, out, cap_pin_0_mode);
+	clock->pps_info.pin_caps[1] = MLX5_GET(mtpps_reg, out, cap_pin_1_mode);
+	clock->pps_info.pin_caps[2] = MLX5_GET(mtpps_reg, out, cap_pin_2_mode);
+	clock->pps_info.pin_caps[3] = MLX5_GET(mtpps_reg, out, cap_pin_3_mode);
+	clock->pps_info.pin_caps[4] = MLX5_GET(mtpps_reg, out, cap_pin_4_mode);
+	clock->pps_info.pin_caps[5] = MLX5_GET(mtpps_reg, out, cap_pin_5_mode);
+	clock->pps_info.pin_caps[6] = MLX5_GET(mtpps_reg, out, cap_pin_6_mode);
+	clock->pps_info.pin_caps[7] = MLX5_GET(mtpps_reg, out, cap_pin_7_mode);
+}
+
+static void ts_next_sec(struct timespec64 *ts)
+{
+	ts->tv_sec += 1;
+	ts->tv_nsec = 0;
+}
+
+static u64 perout_conf_next_event_timer(struct mlx5_core_dev *mdev,
+					struct mlx5_clock *clock)
+{
+	struct timespec64 ts;
+	s64 target_ns;
+
+	mlx5_ptp_gettimex(&clock->ptp_info, &ts, NULL);
+	ts_next_sec(&ts);
+	target_ns = timespec64_to_ns(&ts);
+
+	return find_target_cycles(mdev, target_ns);
+}
+
+static int mlx5_pps_event(struct notifier_block *nb,
+			  unsigned long type, void *data)
+{
+	struct mlx5_clock *clock = mlx5_nb_cof(nb, struct mlx5_clock, pps_nb);
+	struct ptp_clock_event ptp_event;
+	struct mlx5_eqe *eqe = data;
+	int pin = eqe->data.pps.pin;
+	struct mlx5_core_dev *mdev;
+	unsigned long flags;
+	u64 ns;
+
+	mdev = container_of(clock, struct mlx5_core_dev, clock);
+
+	switch (clock->ptp_info.pin_config[pin].func) {
+	case PTP_PF_EXTTS:
+		ptp_event.index = pin;
+		ptp_event.timestamp = mlx5_real_time_mode(mdev) ?
+			mlx5_real_time_cyc2time(clock,
+						be64_to_cpu(eqe->data.pps.time_stamp)) :
+			mlx5_timecounter_cyc2time(clock,
+						  be64_to_cpu(eqe->data.pps.time_stamp));
+		if (clock->pps_info.enabled) {
+			ptp_event.type = PTP_CLOCK_PPSUSR;
+			ptp_event.pps_times.ts_real =
+					ns_to_timespec64(ptp_event.timestamp);
+		} else {
+			ptp_event.type = PTP_CLOCK_EXTTS;
+		}
+		/* TODOL clock->ptp can be NULL if ptp_clock_register fails */
+		ptp_clock_event(clock->ptp, &ptp_event);
+		break;
+	case PTP_PF_PEROUT:
+		ns = perout_conf_next_event_timer(mdev, clock);
+		write_seqlock_irqsave(&clock->lock, flags);
+		clock->pps_info.start[pin] = ns;
+		write_sequnlock_irqrestore(&clock->lock, flags);
+		schedule_work(&clock->pps_info.out_work);
+		break;
+	default:
+		mlx5_core_err(mdev, " Unhandled clock PPS event, func %d\n",
+			      clock->ptp_info.pin_config[pin].func);
+	}
+
+	return NOTIFY_OK;
+}
+
+static void mlx5_timecounter_init(struct mlx5_core_dev *mdev)
+{
+	struct mlx5_clock *clock = &mdev->clock;
+	struct mlx5_timer *timer = &clock->timer;
+	u32 dev_freq;
+
+	dev_freq = MLX5_CAP_GEN(mdev, device_frequency_khz);
+	timer->cycles.read = read_internal_timer;
+	timer->cycles.shift = MLX5_CYCLES_SHIFT;
+	timer->cycles.mult = clocksource_khz2mult(dev_freq,
+						  timer->cycles.shift);
+	timer->nominal_c_mult = timer->cycles.mult;
+	timer->cycles.mask = CLOCKSOURCE_MASK(41);
+
+	timecounter_init(&timer->tc, &timer->cycles,
+			 ktime_to_ns(ktime_get_real()));
+}
+
+static void mlx5_init_overflow_period(struct mlx5_clock *clock)
+{
+	struct mlx5_core_dev *mdev = container_of(clock, struct mlx5_core_dev, clock);
+	struct mlx5_ib_clock_info *clock_info = mdev->clock_info;
+	struct mlx5_timer *timer = &clock->timer;
+	u64 overflow_cycles;
+	u64 frac = 0;
+	u64 ns;
+
+	/* Calculate period in seconds to call the overflow watchdog - to make
+	 * sure counter is checked at least twice every wrap around.
+	 * The period is calculated as the minimum between max HW cycles count
+	 * (The clock source mask) and max amount of cycles that can be
+	 * multiplied by clock multiplier where the result doesn't exceed
+	 * 64bits.
+	 */
+	overflow_cycles = div64_u64(~0ULL >> 1, timer->cycles.mult);
+	overflow_cycles = min(overflow_cycles, div_u64(timer->cycles.mask, 3));
+
+	ns = cyclecounter_cyc2ns(&timer->cycles, overflow_cycles,
+				 frac, &frac);
+	do_div(ns, NSEC_PER_SEC / HZ);
+	timer->overflow_period = ns;
+
+	INIT_DELAYED_WORK(&timer->overflow_work, mlx5_timestamp_overflow);
+	if (timer->overflow_period)
+		schedule_delayed_work(&timer->overflow_work, 0);
+	else
+		mlx5_core_warn(mdev,
+			       "invalid overflow period, overflow_work is not scheduled\n");
+
+	if (clock_info)
+		clock_info->overflow_period = timer->overflow_period;
+}
+
+static void mlx5_init_clock_info(struct mlx5_core_dev *mdev)
+{
+	struct mlx5_clock *clock = &mdev->clock;
+	struct mlx5_ib_clock_info *info;
+	struct mlx5_timer *timer;
+
+	mdev->clock_info = (struct mlx5_ib_clock_info *)get_zeroed_page(GFP_KERNEL);
+	if (!mdev->clock_info) {
+		mlx5_core_warn(mdev, "Failed to allocate IB clock info page\n");
+		return;
+	}
+
+	info = mdev->clock_info;
+	timer = &clock->timer;
+
+	info->nsec = timer->tc.nsec;
+	info->cycles = timer->tc.cycle_last;
+	info->mask = timer->cycles.mask;
+	info->mult = timer->nominal_c_mult;
+	info->shift = timer->cycles.shift;
+	info->frac = timer->tc.frac;
+}
+
+static void mlx5_init_timer_clock(struct mlx5_core_dev *mdev)
+{
+	struct mlx5_clock *clock = &mdev->clock;
+
+	mlx5_timecounter_init(mdev);
+	mlx5_init_clock_info(mdev);
+	mlx5_init_overflow_period(clock);
+	clock->ptp_info = mlx5_ptp_clock_info;
+
+	if (mlx5_real_time_mode(mdev)) {
+		struct timespec64 ts;
+
+		ktime_get_real_ts64(&ts);
+		mlx5_ptp_settime(&clock->ptp_info, &ts);
+	}
+}
+
+static void mlx5_init_pps(struct mlx5_core_dev *mdev)
+{
+	struct mlx5_clock *clock = &mdev->clock;
+
+	if (!MLX5_PPS_CAP(mdev))
+		return;
+
+	mlx5_get_pps_caps(mdev);
+	mlx5_init_pin_config(clock);
+}
+
+void mlx5_init_clock(struct mlx5_core_dev *mdev)
+{
+	struct mlx5_clock *clock = &mdev->clock;
+
+	if (!MLX5_CAP_GEN(mdev, device_frequency_khz)) {
+		mlx5_core_warn(mdev, "invalid device_frequency_khz, aborting HW clock init\n");
+		return;
+	}
+
+	seqlock_init(&clock->lock);
+	mlx5_init_timer_clock(mdev);
+	INIT_WORK(&clock->pps_info.out_work, mlx5_pps_out);
+
+	/* Configure the PHC */
+	clock->ptp_info = mlx5_ptp_clock_info;
+
+	/* Initialize 1PPS data structures */
+	mlx5_init_pps(mdev);
+
+	clock->ptp = ptp_clock_register(&clock->ptp_info,
+					&mdev->pdev->dev);
+	if (IS_ERR(clock->ptp)) {
+		mlx5_core_warn(mdev, "ptp_clock_register failed %ld\n",
+			       PTR_ERR(clock->ptp));
+		clock->ptp = NULL;
+	}
+
+	MLX5_NB_INIT(&clock->pps_nb, mlx5_pps_event, PPS_EVENT);
+	mlx5_eq_notifier_register(mdev, &clock->pps_nb);
+}
+
+void mlx5_cleanup_clock(struct mlx5_core_dev *mdev)
+{
+	struct mlx5_clock *clock = &mdev->clock;
+
+	if (!MLX5_CAP_GEN(mdev, device_frequency_khz))
+		return;
+
+	mlx5_eq_notifier_unregister(mdev, &clock->pps_nb);
+	if (clock->ptp) {
+		ptp_clock_unregister(clock->ptp);
+		clock->ptp = NULL;
+	}
+
+	cancel_work_sync(&clock->pps_info.out_work);
+	cancel_delayed_work_sync(&clock->timer.overflow_work);
+
+	if (mdev->clock_info) {
+		free_page((unsigned long)mdev->clock_info);
+		mdev->clock_info = NULL;
+	}
+
+	kfree(clock->ptp_info.pin_config);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/clock.h b/drivers/net/ethernet/mellanox/mlx5/core/lib/clock.h
new file mode 100644
index 000000000..bd95b9f8d
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/clock.h
@@ -0,0 +1,121 @@
+/*
+ * Copyright (c) 2017, Mellanox Technologies, Ltd.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef __LIB_CLOCK_H__
+#define __LIB_CLOCK_H__
+
+static inline bool mlx5_is_real_time_rq(struct mlx5_core_dev *mdev)
+{
+	u8 rq_ts_format_cap = MLX5_CAP_GEN(mdev, rq_ts_format);
+
+	return (rq_ts_format_cap == MLX5_TIMESTAMP_FORMAT_CAP_REAL_TIME ||
+		rq_ts_format_cap ==
+			MLX5_TIMESTAMP_FORMAT_CAP_FREE_RUNNING_AND_REAL_TIME);
+}
+
+static inline bool mlx5_is_real_time_sq(struct mlx5_core_dev *mdev)
+{
+	u8 sq_ts_format_cap = MLX5_CAP_GEN(mdev, sq_ts_format);
+
+	return (sq_ts_format_cap == MLX5_TIMESTAMP_FORMAT_CAP_REAL_TIME ||
+		sq_ts_format_cap ==
+			MLX5_TIMESTAMP_FORMAT_CAP_FREE_RUNNING_AND_REAL_TIME);
+}
+
+typedef ktime_t (*cqe_ts_to_ns)(struct mlx5_clock *, u64);
+
+#if IS_ENABLED(CONFIG_PTP_1588_CLOCK)
+void mlx5_init_clock(struct mlx5_core_dev *mdev);
+void mlx5_cleanup_clock(struct mlx5_core_dev *mdev);
+
+static inline int mlx5_clock_get_ptp_index(struct mlx5_core_dev *mdev)
+{
+	return mdev->clock.ptp ? ptp_clock_index(mdev->clock.ptp) : -1;
+}
+
+static inline ktime_t mlx5_timecounter_cyc2time(struct mlx5_clock *clock,
+						u64 timestamp)
+{
+	struct mlx5_timer *timer = &clock->timer;
+	unsigned int seq;
+	u64 nsec;
+
+	do {
+		seq = read_seqbegin(&clock->lock);
+		nsec = timecounter_cyc2time(&timer->tc, timestamp);
+	} while (read_seqretry(&clock->lock, seq));
+
+	return ns_to_ktime(nsec);
+}
+
+#define REAL_TIME_TO_NS(hi, low) (((u64)hi) * NSEC_PER_SEC + ((u64)low))
+
+static inline ktime_t mlx5_real_time_cyc2time(struct mlx5_clock *clock,
+					      u64 timestamp)
+{
+	u64 time = REAL_TIME_TO_NS(timestamp >> 32, timestamp & 0xFFFFFFFF);
+
+	return ns_to_ktime(time);
+}
+#else
+static inline void mlx5_init_clock(struct mlx5_core_dev *mdev) {}
+static inline void mlx5_cleanup_clock(struct mlx5_core_dev *mdev) {}
+static inline int mlx5_clock_get_ptp_index(struct mlx5_core_dev *mdev)
+{
+	return -1;
+}
+
+static inline ktime_t mlx5_timecounter_cyc2time(struct mlx5_clock *clock,
+						u64 timestamp)
+{
+	return 0;
+}
+
+static inline ktime_t mlx5_real_time_cyc2time(struct mlx5_clock *clock,
+					      u64 timestamp)
+{
+	return 0;
+}
+#endif
+
+static inline cqe_ts_to_ns mlx5_rq_ts_translator(struct mlx5_core_dev *mdev)
+{
+	return mlx5_is_real_time_rq(mdev) ? mlx5_real_time_cyc2time :
+					    mlx5_timecounter_cyc2time;
+}
+
+static inline cqe_ts_to_ns mlx5_sq_ts_translator(struct mlx5_core_dev *mdev)
+{
+	return mlx5_is_real_time_sq(mdev) ? mlx5_real_time_cyc2time :
+					    mlx5_timecounter_cyc2time;
+}
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/crypto.c b/drivers/net/ethernet/mellanox/mlx5/core/lib/crypto.c
new file mode 100644
index 000000000..e995f8378
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/crypto.c
@@ -0,0 +1,73 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+// Copyright (c) 2019 Mellanox Technologies.
+
+#include "mlx5_core.h"
+#include "lib/mlx5.h"
+
+int mlx5_create_encryption_key(struct mlx5_core_dev *mdev,
+			       void *key, u32 sz_bytes,
+			       u32 key_type, u32 *p_key_id)
+{
+	u32 in[MLX5_ST_SZ_DW(create_encryption_key_in)] = {};
+	u32 out[MLX5_ST_SZ_DW(general_obj_out_cmd_hdr)];
+	u32 sz_bits = sz_bytes * BITS_PER_BYTE;
+	u8  general_obj_key_size;
+	u64 general_obj_types;
+	void *obj, *key_p;
+	int err;
+
+	obj = MLX5_ADDR_OF(create_encryption_key_in, in, encryption_key_object);
+	key_p = MLX5_ADDR_OF(encryption_key_obj, obj, key);
+
+	general_obj_types = MLX5_CAP_GEN_64(mdev, general_obj_types);
+	if (!(general_obj_types &
+	      MLX5_HCA_CAP_GENERAL_OBJECT_TYPES_ENCRYPTION_KEY))
+		return -EINVAL;
+
+	switch (sz_bits) {
+	case 128:
+		general_obj_key_size =
+			MLX5_GENERAL_OBJECT_TYPE_ENCRYPTION_KEY_KEY_SIZE_128;
+		key_p += sz_bytes;
+		break;
+	case 256:
+		general_obj_key_size =
+			MLX5_GENERAL_OBJECT_TYPE_ENCRYPTION_KEY_KEY_SIZE_256;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	memcpy(key_p, key, sz_bytes);
+
+	MLX5_SET(encryption_key_obj, obj, key_size, general_obj_key_size);
+	MLX5_SET(encryption_key_obj, obj, key_type, key_type);
+	MLX5_SET(general_obj_in_cmd_hdr, in, opcode,
+		 MLX5_CMD_OP_CREATE_GENERAL_OBJECT);
+	MLX5_SET(general_obj_in_cmd_hdr, in, obj_type,
+		 MLX5_GENERAL_OBJECT_TYPES_ENCRYPTION_KEY);
+	MLX5_SET(encryption_key_obj, obj, pd, mdev->mlx5e_res.hw_objs.pdn);
+
+	err = mlx5_cmd_exec(mdev, in, sizeof(in), out, sizeof(out));
+	if (!err)
+		*p_key_id = MLX5_GET(general_obj_out_cmd_hdr, out, obj_id);
+
+	/* avoid leaking key on the stack */
+	memzero_explicit(in, sizeof(in));
+
+	return err;
+}
+
+void mlx5_destroy_encryption_key(struct mlx5_core_dev *mdev, u32 key_id)
+{
+	u32 in[MLX5_ST_SZ_DW(general_obj_in_cmd_hdr)] = {};
+	u32 out[MLX5_ST_SZ_DW(general_obj_out_cmd_hdr)];
+
+	MLX5_SET(general_obj_in_cmd_hdr, in, opcode,
+		 MLX5_CMD_OP_DESTROY_GENERAL_OBJECT);
+	MLX5_SET(general_obj_in_cmd_hdr, in, obj_type,
+		 MLX5_GENERAL_OBJECT_TYPES_ENCRYPTION_KEY);
+	MLX5_SET(general_obj_in_cmd_hdr, in, obj_id, key_id);
+
+	mlx5_cmd_exec(mdev, in, sizeof(in), out, sizeof(out));
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/devcom.c b/drivers/net/ethernet/mellanox/mlx5/core/lib/devcom.c
new file mode 100644
index 000000000..b7d779d08
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/devcom.c
@@ -0,0 +1,293 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2018 Mellanox Technologies */
+
+#include <linux/mlx5/vport.h>
+#include "lib/devcom.h"
+#include "mlx5_core.h"
+
+static LIST_HEAD(devcom_list);
+
+#define devcom_for_each_component(priv, comp, iter) \
+	for (iter = 0; \
+	     comp = &(priv)->components[iter], iter < MLX5_DEVCOM_NUM_COMPONENTS; \
+	     iter++)
+
+struct mlx5_devcom_component {
+	struct {
+		void __rcu *data;
+	} device[MLX5_DEVCOM_PORTS_SUPPORTED];
+
+	mlx5_devcom_event_handler_t handler;
+	struct rw_semaphore sem;
+	bool paired;
+};
+
+struct mlx5_devcom_list {
+	struct list_head list;
+
+	struct mlx5_devcom_component components[MLX5_DEVCOM_NUM_COMPONENTS];
+	struct mlx5_core_dev *devs[MLX5_DEVCOM_PORTS_SUPPORTED];
+};
+
+struct mlx5_devcom {
+	struct mlx5_devcom_list *priv;
+	int idx;
+};
+
+static struct mlx5_devcom_list *mlx5_devcom_list_alloc(void)
+{
+	struct mlx5_devcom_component *comp;
+	struct mlx5_devcom_list *priv;
+	int i;
+
+	priv = kzalloc(sizeof(*priv), GFP_KERNEL);
+	if (!priv)
+		return NULL;
+
+	devcom_for_each_component(priv, comp, i)
+		init_rwsem(&comp->sem);
+
+	return priv;
+}
+
+static struct mlx5_devcom *mlx5_devcom_alloc(struct mlx5_devcom_list *priv,
+					     u8 idx)
+{
+	struct mlx5_devcom *devcom;
+
+	devcom = kzalloc(sizeof(*devcom), GFP_KERNEL);
+	if (!devcom)
+		return NULL;
+
+	devcom->priv = priv;
+	devcom->idx = idx;
+	return devcom;
+}
+
+/* Must be called with intf_mutex held */
+struct mlx5_devcom *mlx5_devcom_register_device(struct mlx5_core_dev *dev)
+{
+	struct mlx5_devcom_list *priv = NULL, *iter;
+	struct mlx5_devcom *devcom = NULL;
+	bool new_priv = false;
+	u64 sguid0, sguid1;
+	int idx, i;
+
+	if (!mlx5_core_is_pf(dev))
+		return NULL;
+	if (MLX5_CAP_GEN(dev, num_lag_ports) != MLX5_DEVCOM_PORTS_SUPPORTED)
+		return NULL;
+
+	mlx5_dev_list_lock();
+	sguid0 = mlx5_query_nic_system_image_guid(dev);
+	list_for_each_entry(iter, &devcom_list, list) {
+		struct mlx5_core_dev *tmp_dev = NULL;
+
+		idx = -1;
+		for (i = 0; i < MLX5_DEVCOM_PORTS_SUPPORTED; i++) {
+			if (iter->devs[i])
+				tmp_dev = iter->devs[i];
+			else
+				idx = i;
+		}
+
+		if (idx == -1)
+			continue;
+
+		sguid1 = mlx5_query_nic_system_image_guid(tmp_dev);
+		if (sguid0 != sguid1)
+			continue;
+
+		priv = iter;
+		break;
+	}
+
+	if (!priv) {
+		priv = mlx5_devcom_list_alloc();
+		if (!priv) {
+			devcom = ERR_PTR(-ENOMEM);
+			goto out;
+		}
+
+		idx = 0;
+		new_priv = true;
+	}
+
+	priv->devs[idx] = dev;
+	devcom = mlx5_devcom_alloc(priv, idx);
+	if (!devcom) {
+		if (new_priv)
+			kfree(priv);
+		devcom = ERR_PTR(-ENOMEM);
+		goto out;
+	}
+
+	if (new_priv)
+		list_add(&priv->list, &devcom_list);
+out:
+	mlx5_dev_list_unlock();
+	return devcom;
+}
+
+/* Must be called with intf_mutex held */
+void mlx5_devcom_unregister_device(struct mlx5_devcom *devcom)
+{
+	struct mlx5_devcom_list *priv;
+	int i;
+
+	if (IS_ERR_OR_NULL(devcom))
+		return;
+
+	mlx5_dev_list_lock();
+	priv = devcom->priv;
+	priv->devs[devcom->idx] = NULL;
+
+	kfree(devcom);
+
+	for (i = 0; i < MLX5_DEVCOM_PORTS_SUPPORTED; i++)
+		if (priv->devs[i])
+			break;
+
+	if (i != MLX5_DEVCOM_PORTS_SUPPORTED)
+		goto out;
+
+	list_del(&priv->list);
+	kfree(priv);
+out:
+	mlx5_dev_list_unlock();
+}
+
+void mlx5_devcom_register_component(struct mlx5_devcom *devcom,
+				    enum mlx5_devcom_components id,
+				    mlx5_devcom_event_handler_t handler,
+				    void *data)
+{
+	struct mlx5_devcom_component *comp;
+
+	if (IS_ERR_OR_NULL(devcom))
+		return;
+
+	WARN_ON(!data);
+
+	comp = &devcom->priv->components[id];
+	down_write(&comp->sem);
+	comp->handler = handler;
+	rcu_assign_pointer(comp->device[devcom->idx].data, data);
+	up_write(&comp->sem);
+}
+
+void mlx5_devcom_unregister_component(struct mlx5_devcom *devcom,
+				      enum mlx5_devcom_components id)
+{
+	struct mlx5_devcom_component *comp;
+
+	if (IS_ERR_OR_NULL(devcom))
+		return;
+
+	comp = &devcom->priv->components[id];
+	down_write(&comp->sem);
+	RCU_INIT_POINTER(comp->device[devcom->idx].data, NULL);
+	up_write(&comp->sem);
+	synchronize_rcu();
+}
+
+int mlx5_devcom_send_event(struct mlx5_devcom *devcom,
+			   enum mlx5_devcom_components id,
+			   int event,
+			   void *event_data)
+{
+	struct mlx5_devcom_component *comp;
+	int err = -ENODEV, i;
+
+	if (IS_ERR_OR_NULL(devcom))
+		return err;
+
+	comp = &devcom->priv->components[id];
+	down_write(&comp->sem);
+	for (i = 0; i < MLX5_DEVCOM_PORTS_SUPPORTED; i++) {
+		void *data = rcu_dereference_protected(comp->device[i].data,
+						       lockdep_is_held(&comp->sem));
+
+		if (i != devcom->idx && data) {
+			err = comp->handler(event, data, event_data);
+			break;
+		}
+	}
+
+	up_write(&comp->sem);
+	return err;
+}
+
+void mlx5_devcom_set_paired(struct mlx5_devcom *devcom,
+			    enum mlx5_devcom_components id,
+			    bool paired)
+{
+	struct mlx5_devcom_component *comp;
+
+	comp = &devcom->priv->components[id];
+	WARN_ON(!rwsem_is_locked(&comp->sem));
+
+	WRITE_ONCE(comp->paired, paired);
+}
+
+bool mlx5_devcom_is_paired(struct mlx5_devcom *devcom,
+			   enum mlx5_devcom_components id)
+{
+	if (IS_ERR_OR_NULL(devcom))
+		return false;
+
+	return READ_ONCE(devcom->priv->components[id].paired);
+}
+
+void *mlx5_devcom_get_peer_data(struct mlx5_devcom *devcom,
+				enum mlx5_devcom_components id)
+{
+	struct mlx5_devcom_component *comp;
+	int i;
+
+	if (IS_ERR_OR_NULL(devcom))
+		return NULL;
+
+	comp = &devcom->priv->components[id];
+	down_read(&comp->sem);
+	if (!READ_ONCE(comp->paired)) {
+		up_read(&comp->sem);
+		return NULL;
+	}
+
+	for (i = 0; i < MLX5_DEVCOM_PORTS_SUPPORTED; i++)
+		if (i != devcom->idx)
+			break;
+
+	return rcu_dereference_protected(comp->device[i].data, lockdep_is_held(&comp->sem));
+}
+
+void *mlx5_devcom_get_peer_data_rcu(struct mlx5_devcom *devcom, enum mlx5_devcom_components id)
+{
+	struct mlx5_devcom_component *comp;
+	int i;
+
+	if (IS_ERR_OR_NULL(devcom))
+		return NULL;
+
+	for (i = 0; i < MLX5_DEVCOM_PORTS_SUPPORTED; i++)
+		if (i != devcom->idx)
+			break;
+
+	comp = &devcom->priv->components[id];
+	/* This can change concurrently, however 'data' pointer will remain
+	 * valid for the duration of RCU read section.
+	 */
+	if (!READ_ONCE(comp->paired))
+		return NULL;
+
+	return rcu_dereference(comp->device[i].data);
+}
+
+void mlx5_devcom_release_peer_data(struct mlx5_devcom *devcom,
+				   enum mlx5_devcom_components id)
+{
+	struct mlx5_devcom_component *comp = &devcom->priv->components[id];
+
+	up_read(&comp->sem);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/devcom.h b/drivers/net/ethernet/mellanox/mlx5/core/lib/devcom.h
new file mode 100644
index 000000000..9a496f472
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/devcom.h
@@ -0,0 +1,49 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2018 Mellanox Technologies */
+
+#ifndef __LIB_MLX5_DEVCOM_H__
+#define __LIB_MLX5_DEVCOM_H__
+
+#include <linux/mlx5/driver.h>
+
+#define MLX5_DEVCOM_PORTS_SUPPORTED 2
+
+enum mlx5_devcom_components {
+	MLX5_DEVCOM_ESW_OFFLOADS,
+
+	MLX5_DEVCOM_NUM_COMPONENTS,
+};
+
+typedef int (*mlx5_devcom_event_handler_t)(int event,
+					   void *my_data,
+					   void *event_data);
+
+struct mlx5_devcom *mlx5_devcom_register_device(struct mlx5_core_dev *dev);
+void mlx5_devcom_unregister_device(struct mlx5_devcom *devcom);
+
+void mlx5_devcom_register_component(struct mlx5_devcom *devcom,
+				    enum mlx5_devcom_components id,
+				    mlx5_devcom_event_handler_t handler,
+				    void *data);
+void mlx5_devcom_unregister_component(struct mlx5_devcom *devcom,
+				      enum mlx5_devcom_components id);
+
+int mlx5_devcom_send_event(struct mlx5_devcom *devcom,
+			   enum mlx5_devcom_components id,
+			   int event,
+			   void *event_data);
+
+void mlx5_devcom_set_paired(struct mlx5_devcom *devcom,
+			    enum mlx5_devcom_components id,
+			    bool paired);
+bool mlx5_devcom_is_paired(struct mlx5_devcom *devcom,
+			   enum mlx5_devcom_components id);
+
+void *mlx5_devcom_get_peer_data(struct mlx5_devcom *devcom,
+				enum mlx5_devcom_components id);
+void *mlx5_devcom_get_peer_data_rcu(struct mlx5_devcom *devcom, enum mlx5_devcom_components id);
+void mlx5_devcom_release_peer_data(struct mlx5_devcom *devcom,
+				   enum mlx5_devcom_components id);
+
+#endif
+
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/dm.c b/drivers/net/ethernet/mellanox/mlx5/core/lib/dm.c
new file mode 100644
index 000000000..9482e51ac
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/dm.c
@@ -0,0 +1,267 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+// Copyright (c) 2019 Mellanox Technologies
+
+#include <linux/mlx5/driver.h>
+#include <linux/mlx5/device.h>
+
+#include "mlx5_core.h"
+#include "lib/mlx5.h"
+
+struct mlx5_dm {
+	/* protect access to icm bitmask */
+	spinlock_t lock;
+	unsigned long *steering_sw_icm_alloc_blocks;
+	unsigned long *header_modify_sw_icm_alloc_blocks;
+	unsigned long *header_modify_pattern_sw_icm_alloc_blocks;
+};
+
+struct mlx5_dm *mlx5_dm_create(struct mlx5_core_dev *dev)
+{
+	u64 header_modify_pattern_icm_blocks = 0;
+	u64 header_modify_icm_blocks = 0;
+	u64 steering_icm_blocks = 0;
+	struct mlx5_dm *dm;
+	bool support_v2;
+
+	if (!(MLX5_CAP_GEN_64(dev, general_obj_types) & MLX5_GENERAL_OBJ_TYPES_CAP_SW_ICM))
+		return NULL;
+
+	dm = kzalloc(sizeof(*dm), GFP_KERNEL);
+	if (!dm)
+		return ERR_PTR(-ENOMEM);
+
+	spin_lock_init(&dm->lock);
+
+	if (MLX5_CAP64_DEV_MEM(dev, steering_sw_icm_start_address)) {
+		steering_icm_blocks =
+			BIT(MLX5_CAP_DEV_MEM(dev, log_steering_sw_icm_size) -
+			    MLX5_LOG_SW_ICM_BLOCK_SIZE(dev));
+
+		dm->steering_sw_icm_alloc_blocks =
+			bitmap_zalloc(steering_icm_blocks, GFP_KERNEL);
+		if (!dm->steering_sw_icm_alloc_blocks)
+			goto err_steering;
+	}
+
+	if (MLX5_CAP64_DEV_MEM(dev, header_modify_sw_icm_start_address)) {
+		header_modify_icm_blocks =
+			BIT(MLX5_CAP_DEV_MEM(dev, log_header_modify_sw_icm_size) -
+			    MLX5_LOG_SW_ICM_BLOCK_SIZE(dev));
+
+		dm->header_modify_sw_icm_alloc_blocks =
+			bitmap_zalloc(header_modify_icm_blocks, GFP_KERNEL);
+		if (!dm->header_modify_sw_icm_alloc_blocks)
+			goto err_modify_hdr;
+	}
+
+	support_v2 = MLX5_CAP_FLOWTABLE_NIC_RX(dev, sw_owner_v2) &&
+		     MLX5_CAP_FLOWTABLE_NIC_TX(dev, sw_owner_v2) &&
+		     MLX5_CAP64_DEV_MEM(dev, header_modify_pattern_sw_icm_start_address);
+
+	if (support_v2) {
+		header_modify_pattern_icm_blocks =
+			BIT(MLX5_CAP_DEV_MEM(dev, log_header_modify_pattern_sw_icm_size) -
+			    MLX5_LOG_SW_ICM_BLOCK_SIZE(dev));
+
+		dm->header_modify_pattern_sw_icm_alloc_blocks =
+			bitmap_zalloc(header_modify_pattern_icm_blocks, GFP_KERNEL);
+		if (!dm->header_modify_pattern_sw_icm_alloc_blocks)
+			goto err_pattern;
+	}
+
+	return dm;
+
+err_pattern:
+	bitmap_free(dm->header_modify_sw_icm_alloc_blocks);
+
+err_modify_hdr:
+	bitmap_free(dm->steering_sw_icm_alloc_blocks);
+
+err_steering:
+	kfree(dm);
+
+	return ERR_PTR(-ENOMEM);
+}
+
+void mlx5_dm_cleanup(struct mlx5_core_dev *dev)
+{
+	struct mlx5_dm *dm = dev->dm;
+
+	if (!dev->dm)
+		return;
+
+	if (dm->steering_sw_icm_alloc_blocks) {
+		WARN_ON(!bitmap_empty(dm->steering_sw_icm_alloc_blocks,
+				      BIT(MLX5_CAP_DEV_MEM(dev, log_steering_sw_icm_size) -
+					  MLX5_LOG_SW_ICM_BLOCK_SIZE(dev))));
+		bitmap_free(dm->steering_sw_icm_alloc_blocks);
+	}
+
+	if (dm->header_modify_sw_icm_alloc_blocks) {
+		WARN_ON(!bitmap_empty(dm->header_modify_sw_icm_alloc_blocks,
+				      BIT(MLX5_CAP_DEV_MEM(dev,
+							   log_header_modify_sw_icm_size) -
+				      MLX5_LOG_SW_ICM_BLOCK_SIZE(dev))));
+		bitmap_free(dm->header_modify_sw_icm_alloc_blocks);
+	}
+
+	if (dm->header_modify_pattern_sw_icm_alloc_blocks) {
+		WARN_ON(!bitmap_empty(dm->header_modify_pattern_sw_icm_alloc_blocks,
+				      BIT(MLX5_CAP_DEV_MEM(dev,
+							   log_header_modify_pattern_sw_icm_size) -
+					  MLX5_LOG_SW_ICM_BLOCK_SIZE(dev))));
+		bitmap_free(dm->header_modify_pattern_sw_icm_alloc_blocks);
+	}
+
+	kfree(dm);
+}
+
+int mlx5_dm_sw_icm_alloc(struct mlx5_core_dev *dev, enum mlx5_sw_icm_type type,
+			 u64 length, u32 log_alignment, u16 uid,
+			 phys_addr_t *addr, u32 *obj_id)
+{
+	u32 num_blocks = DIV_ROUND_UP_ULL(length, MLX5_SW_ICM_BLOCK_SIZE(dev));
+	u32 out[MLX5_ST_SZ_DW(general_obj_out_cmd_hdr)] = {};
+	u32 in[MLX5_ST_SZ_DW(create_sw_icm_in)] = {};
+	struct mlx5_dm *dm = dev->dm;
+	unsigned long *block_map;
+	u64 icm_start_addr;
+	u32 log_icm_size;
+	u64 align_mask;
+	u32 max_blocks;
+	u64 block_idx;
+	void *sw_icm;
+	int ret;
+
+	if (!dev->dm)
+		return -EOPNOTSUPP;
+
+	if (!length || (length & (length - 1)) ||
+	    length & (MLX5_SW_ICM_BLOCK_SIZE(dev) - 1))
+		return -EINVAL;
+
+	MLX5_SET(general_obj_in_cmd_hdr, in, opcode,
+		 MLX5_CMD_OP_CREATE_GENERAL_OBJECT);
+	MLX5_SET(general_obj_in_cmd_hdr, in, obj_type, MLX5_OBJ_TYPE_SW_ICM);
+	MLX5_SET(general_obj_in_cmd_hdr, in, uid, uid);
+
+	switch (type) {
+	case MLX5_SW_ICM_TYPE_STEERING:
+		icm_start_addr = MLX5_CAP64_DEV_MEM(dev, steering_sw_icm_start_address);
+		log_icm_size = MLX5_CAP_DEV_MEM(dev, log_steering_sw_icm_size);
+		block_map = dm->steering_sw_icm_alloc_blocks;
+		break;
+	case MLX5_SW_ICM_TYPE_HEADER_MODIFY:
+		icm_start_addr = MLX5_CAP64_DEV_MEM(dev, header_modify_sw_icm_start_address);
+		log_icm_size = MLX5_CAP_DEV_MEM(dev,
+						log_header_modify_sw_icm_size);
+		block_map = dm->header_modify_sw_icm_alloc_blocks;
+		break;
+	case MLX5_SW_ICM_TYPE_HEADER_MODIFY_PATTERN:
+		icm_start_addr = MLX5_CAP64_DEV_MEM(dev,
+						    header_modify_pattern_sw_icm_start_address);
+		log_icm_size = MLX5_CAP_DEV_MEM(dev,
+						log_header_modify_pattern_sw_icm_size);
+		block_map = dm->header_modify_pattern_sw_icm_alloc_blocks;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	if (!block_map)
+		return -EOPNOTSUPP;
+
+	max_blocks = BIT(log_icm_size - MLX5_LOG_SW_ICM_BLOCK_SIZE(dev));
+
+	if (log_alignment < MLX5_LOG_SW_ICM_BLOCK_SIZE(dev))
+		log_alignment = MLX5_LOG_SW_ICM_BLOCK_SIZE(dev);
+	align_mask = BIT(log_alignment - MLX5_LOG_SW_ICM_BLOCK_SIZE(dev)) - 1;
+
+	spin_lock(&dm->lock);
+	block_idx = bitmap_find_next_zero_area(block_map, max_blocks, 0,
+					       num_blocks, align_mask);
+
+	if (block_idx < max_blocks)
+		bitmap_set(block_map,
+			   block_idx, num_blocks);
+
+	spin_unlock(&dm->lock);
+
+	if (block_idx >= max_blocks)
+		return -ENOMEM;
+
+	sw_icm = MLX5_ADDR_OF(create_sw_icm_in, in, sw_icm);
+	icm_start_addr += block_idx << MLX5_LOG_SW_ICM_BLOCK_SIZE(dev);
+	MLX5_SET64(sw_icm, sw_icm, sw_icm_start_addr,
+		   icm_start_addr);
+	MLX5_SET(sw_icm, sw_icm, log_sw_icm_size, ilog2(length));
+
+	ret = mlx5_cmd_exec(dev, in, sizeof(in), out, sizeof(out));
+	if (ret) {
+		spin_lock(&dm->lock);
+		bitmap_clear(block_map,
+			     block_idx, num_blocks);
+		spin_unlock(&dm->lock);
+
+		return ret;
+	}
+
+	*addr = icm_start_addr;
+	*obj_id = MLX5_GET(general_obj_out_cmd_hdr, out, obj_id);
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(mlx5_dm_sw_icm_alloc);
+
+int mlx5_dm_sw_icm_dealloc(struct mlx5_core_dev *dev, enum mlx5_sw_icm_type type,
+			   u64 length, u16 uid, phys_addr_t addr, u32 obj_id)
+{
+	u32 num_blocks = DIV_ROUND_UP_ULL(length, MLX5_SW_ICM_BLOCK_SIZE(dev));
+	u32 out[MLX5_ST_SZ_DW(general_obj_out_cmd_hdr)] = {};
+	u32 in[MLX5_ST_SZ_DW(general_obj_in_cmd_hdr)] = {};
+	struct mlx5_dm *dm = dev->dm;
+	unsigned long *block_map;
+	u64 icm_start_addr;
+	u64 start_idx;
+	int err;
+
+	if (!dev->dm)
+		return -EOPNOTSUPP;
+
+	switch (type) {
+	case MLX5_SW_ICM_TYPE_STEERING:
+		icm_start_addr = MLX5_CAP64_DEV_MEM(dev, steering_sw_icm_start_address);
+		block_map = dm->steering_sw_icm_alloc_blocks;
+		break;
+	case MLX5_SW_ICM_TYPE_HEADER_MODIFY:
+		icm_start_addr = MLX5_CAP64_DEV_MEM(dev, header_modify_sw_icm_start_address);
+		block_map = dm->header_modify_sw_icm_alloc_blocks;
+		break;
+	case MLX5_SW_ICM_TYPE_HEADER_MODIFY_PATTERN:
+		icm_start_addr = MLX5_CAP64_DEV_MEM(dev,
+						    header_modify_pattern_sw_icm_start_address);
+		block_map = dm->header_modify_pattern_sw_icm_alloc_blocks;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	MLX5_SET(general_obj_in_cmd_hdr, in, opcode,
+		 MLX5_CMD_OP_DESTROY_GENERAL_OBJECT);
+	MLX5_SET(general_obj_in_cmd_hdr, in, obj_type, MLX5_OBJ_TYPE_SW_ICM);
+	MLX5_SET(general_obj_in_cmd_hdr, in, obj_id, obj_id);
+	MLX5_SET(general_obj_in_cmd_hdr, in, uid, uid);
+
+	err =  mlx5_cmd_exec(dev, in, sizeof(in), out, sizeof(out));
+	if (err)
+		return err;
+
+	start_idx = (addr - icm_start_addr) >> MLX5_LOG_SW_ICM_BLOCK_SIZE(dev);
+	spin_lock(&dm->lock);
+	bitmap_clear(block_map,
+		     start_idx, num_blocks);
+	spin_unlock(&dm->lock);
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(mlx5_dm_sw_icm_dealloc);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/eq.h b/drivers/net/ethernet/mellanox/mlx5/core/lib/eq.h
new file mode 100644
index 000000000..d3d628b86
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/eq.h
@@ -0,0 +1,109 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2018-2021, Mellanox Technologies inc.  All rights reserved. */
+
+#ifndef __LIB_MLX5_EQ_H__
+#define __LIB_MLX5_EQ_H__
+#include <linux/mlx5/driver.h>
+#include <linux/mlx5/eq.h>
+#include <linux/mlx5/cq.h>
+
+#define MLX5_EQE_SIZE       (sizeof(struct mlx5_eqe))
+
+struct mlx5_eq_tasklet {
+	struct list_head      list;
+	struct list_head      process_list;
+	struct tasklet_struct task;
+	spinlock_t            lock; /* lock completion tasklet list */
+};
+
+struct mlx5_cq_table {
+	spinlock_t              lock;	/* protect radix tree */
+	struct radix_tree_root  tree;
+};
+
+struct mlx5_eq {
+	struct mlx5_frag_buf_ctrl fbc;
+	struct mlx5_frag_buf    frag_buf;
+	struct mlx5_core_dev    *dev;
+	struct mlx5_cq_table    cq_table;
+	__be32 __iomem	        *doorbell;
+	u32                     cons_index;
+	unsigned int            vecidx;
+	unsigned int            irqn;
+	u8                      eqn;
+	struct mlx5_rsc_debug   *dbg;
+	struct mlx5_irq         *irq;
+};
+
+struct mlx5_eq_async {
+	struct mlx5_eq          core;
+	struct notifier_block   irq_nb;
+	spinlock_t              lock; /* To avoid irq EQ handle races with resiliency flows */
+};
+
+struct mlx5_eq_comp {
+	struct mlx5_eq          core;
+	struct notifier_block   irq_nb;
+	struct mlx5_eq_tasklet  tasklet_ctx;
+	struct list_head        list;
+};
+
+static inline u32 eq_get_size(struct mlx5_eq *eq)
+{
+	return eq->fbc.sz_m1 + 1;
+}
+
+static inline struct mlx5_eqe *get_eqe(struct mlx5_eq *eq, u32 entry)
+{
+	return mlx5_frag_buf_get_wqe(&eq->fbc, entry);
+}
+
+static inline struct mlx5_eqe *next_eqe_sw(struct mlx5_eq *eq)
+{
+	struct mlx5_eqe *eqe = get_eqe(eq, eq->cons_index & eq->fbc.sz_m1);
+
+	return (eqe->owner ^ (eq->cons_index >> eq->fbc.log_sz)) & 1 ? NULL : eqe;
+}
+
+static inline void eq_update_ci(struct mlx5_eq *eq, int arm)
+{
+	__be32 __iomem *addr = eq->doorbell + (arm ? 0 : 2);
+	u32 val = (eq->cons_index & 0xffffff) | (eq->eqn << 24);
+
+	__raw_writel((__force u32)cpu_to_be32(val), addr);
+	/* We still want ordering, just not swabbing, so add a barrier */
+	mb();
+}
+
+int mlx5_eq_table_init(struct mlx5_core_dev *dev);
+void mlx5_eq_table_cleanup(struct mlx5_core_dev *dev);
+int mlx5_eq_table_create(struct mlx5_core_dev *dev);
+void mlx5_eq_table_destroy(struct mlx5_core_dev *dev);
+
+int mlx5_eq_add_cq(struct mlx5_eq *eq, struct mlx5_core_cq *cq);
+void mlx5_eq_del_cq(struct mlx5_eq *eq, struct mlx5_core_cq *cq);
+struct mlx5_eq_comp *mlx5_eqn2comp_eq(struct mlx5_core_dev *dev, int eqn);
+struct mlx5_eq *mlx5_get_async_eq(struct mlx5_core_dev *dev);
+void mlx5_cq_tasklet_cb(struct tasklet_struct *t);
+struct cpumask *mlx5_eq_comp_cpumask(struct mlx5_core_dev *dev, int ix);
+
+u32 mlx5_eq_poll_irq_disabled(struct mlx5_eq_comp *eq);
+void mlx5_cmd_eq_recover(struct mlx5_core_dev *dev);
+void mlx5_eq_synchronize_async_irq(struct mlx5_core_dev *dev);
+void mlx5_eq_synchronize_cmd_irq(struct mlx5_core_dev *dev);
+
+int mlx5_debug_eq_add(struct mlx5_core_dev *dev, struct mlx5_eq *eq);
+void mlx5_debug_eq_remove(struct mlx5_core_dev *dev, struct mlx5_eq *eq);
+void mlx5_eq_debugfs_init(struct mlx5_core_dev *dev);
+void mlx5_eq_debugfs_cleanup(struct mlx5_core_dev *dev);
+
+/* This function should only be called after mlx5_cmd_force_teardown_hca */
+void mlx5_core_eq_free_irqs(struct mlx5_core_dev *dev);
+
+#ifdef CONFIG_RFS_ACCEL
+struct cpu_rmap *mlx5_eq_table_get_rmap(struct mlx5_core_dev *dev);
+#endif
+
+int mlx5_vector2irqn(struct mlx5_core_dev *dev, int vector, unsigned int *irqn);
+
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/fs_chains.c b/drivers/net/ethernet/mellanox/mlx5/core/lib/fs_chains.c
new file mode 100644
index 000000000..df58cba37
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/fs_chains.c
@@ -0,0 +1,810 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+// Copyright (c) 2020 Mellanox Technologies.
+
+#include <linux/mlx5/driver.h>
+#include <linux/mlx5/mlx5_ifc.h>
+#include <linux/mlx5/fs.h>
+
+#include "lib/fs_chains.h"
+#include "fs_ft_pool.h"
+#include "en/mapping.h"
+#include "fs_core.h"
+#include "en_tc.h"
+
+#define chains_lock(chains) ((chains)->lock)
+#define chains_ht(chains) ((chains)->chains_ht)
+#define prios_ht(chains) ((chains)->prios_ht)
+#define tc_default_ft(chains) ((chains)->tc_default_ft)
+#define tc_end_ft(chains) ((chains)->tc_end_ft)
+#define ns_to_chains_fs_prio(ns) ((ns) == MLX5_FLOW_NAMESPACE_FDB ? \
+				  FDB_TC_OFFLOAD : MLX5E_TC_PRIO)
+#define FT_TBL_SZ (64 * 1024)
+
+struct mlx5_fs_chains {
+	struct mlx5_core_dev *dev;
+
+	struct rhashtable chains_ht;
+	struct rhashtable prios_ht;
+	/* Protects above chains_ht and prios_ht */
+	struct mutex lock;
+
+	struct mlx5_flow_table *tc_default_ft;
+	struct mlx5_flow_table *tc_end_ft;
+	struct mapping_ctx *chains_mapping;
+
+	enum mlx5_flow_namespace_type ns;
+	u32 group_num;
+	u32 flags;
+};
+
+struct fs_chain {
+	struct rhash_head node;
+
+	u32 chain;
+
+	int ref;
+	int id;
+
+	struct mlx5_fs_chains *chains;
+	struct list_head prios_list;
+	struct mlx5_flow_handle *restore_rule;
+	struct mlx5_modify_hdr *miss_modify_hdr;
+};
+
+struct prio_key {
+	u32 chain;
+	u32 prio;
+	u32 level;
+};
+
+struct prio {
+	struct rhash_head node;
+	struct list_head list;
+
+	struct prio_key key;
+
+	int ref;
+
+	struct fs_chain *chain;
+	struct mlx5_flow_table *ft;
+	struct mlx5_flow_table *next_ft;
+	struct mlx5_flow_group *miss_group;
+	struct mlx5_flow_handle *miss_rule;
+};
+
+static const struct rhashtable_params chain_params = {
+	.head_offset = offsetof(struct fs_chain, node),
+	.key_offset = offsetof(struct fs_chain, chain),
+	.key_len = sizeof_field(struct fs_chain, chain),
+	.automatic_shrinking = true,
+};
+
+static const struct rhashtable_params prio_params = {
+	.head_offset = offsetof(struct prio, node),
+	.key_offset = offsetof(struct prio, key),
+	.key_len = sizeof_field(struct prio, key),
+	.automatic_shrinking = true,
+};
+
+bool mlx5_chains_prios_supported(struct mlx5_fs_chains *chains)
+{
+	return chains->flags & MLX5_CHAINS_AND_PRIOS_SUPPORTED;
+}
+
+bool mlx5_chains_ignore_flow_level_supported(struct mlx5_fs_chains *chains)
+{
+	return chains->flags & MLX5_CHAINS_IGNORE_FLOW_LEVEL_SUPPORTED;
+}
+
+bool mlx5_chains_backwards_supported(struct mlx5_fs_chains *chains)
+{
+	return mlx5_chains_prios_supported(chains) &&
+	       mlx5_chains_ignore_flow_level_supported(chains);
+}
+
+u32 mlx5_chains_get_chain_range(struct mlx5_fs_chains *chains)
+{
+	if (!mlx5_chains_prios_supported(chains))
+		return 1;
+
+	if (mlx5_chains_ignore_flow_level_supported(chains))
+		return UINT_MAX - 1;
+
+	/* We should get here only for eswitch case */
+	return FDB_TC_MAX_CHAIN;
+}
+
+u32 mlx5_chains_get_nf_ft_chain(struct mlx5_fs_chains *chains)
+{
+	return mlx5_chains_get_chain_range(chains) + 1;
+}
+
+u32 mlx5_chains_get_prio_range(struct mlx5_fs_chains *chains)
+{
+	if (mlx5_chains_ignore_flow_level_supported(chains))
+		return UINT_MAX;
+
+	if (!chains->dev->priv.eswitch ||
+	    chains->dev->priv.eswitch->mode != MLX5_ESWITCH_OFFLOADS)
+		return 1;
+
+	/* We should get here only for eswitch case */
+	return FDB_TC_MAX_PRIO;
+}
+
+static unsigned int mlx5_chains_get_level_range(struct mlx5_fs_chains *chains)
+{
+	if (mlx5_chains_ignore_flow_level_supported(chains))
+		return UINT_MAX;
+
+	/* Same value for FDB and NIC RX tables */
+	return FDB_TC_LEVELS_PER_PRIO;
+}
+
+void
+mlx5_chains_set_end_ft(struct mlx5_fs_chains *chains,
+		       struct mlx5_flow_table *ft)
+{
+	tc_end_ft(chains) = ft;
+}
+
+static struct mlx5_flow_table *
+mlx5_chains_create_table(struct mlx5_fs_chains *chains,
+			 u32 chain, u32 prio, u32 level)
+{
+	struct mlx5_flow_table_attr ft_attr = {};
+	struct mlx5_flow_namespace *ns;
+	struct mlx5_flow_table *ft;
+	int sz;
+
+	if (chains->flags & MLX5_CHAINS_FT_TUNNEL_SUPPORTED)
+		ft_attr.flags |= (MLX5_FLOW_TABLE_TUNNEL_EN_REFORMAT |
+				  MLX5_FLOW_TABLE_TUNNEL_EN_DECAP);
+
+	sz = (chain == mlx5_chains_get_nf_ft_chain(chains)) ? FT_TBL_SZ : POOL_NEXT_SIZE;
+	ft_attr.max_fte = sz;
+
+	/* We use tc_default_ft(chains) as the table's next_ft till
+	 * ignore_flow_level is allowed on FT creation and not just for FTEs.
+	 * Instead caller should add an explicit miss rule if needed.
+	 */
+	ft_attr.next_ft = tc_default_ft(chains);
+
+	/* The root table(chain 0, prio 1, level 0) is required to be
+	 * connected to the previous fs_core managed prio.
+	 * We always create it, as a managed table, in order to align with
+	 * fs_core logic.
+	 */
+	if (!mlx5_chains_ignore_flow_level_supported(chains) ||
+	    (chain == 0 && prio == 1 && level == 0)) {
+		ft_attr.level = level;
+		ft_attr.prio = prio - 1;
+		ns = (chains->ns == MLX5_FLOW_NAMESPACE_FDB) ?
+			mlx5_get_fdb_sub_ns(chains->dev, chain) :
+			mlx5_get_flow_namespace(chains->dev, chains->ns);
+	} else {
+		ft_attr.flags |= MLX5_FLOW_TABLE_UNMANAGED;
+		ft_attr.prio = ns_to_chains_fs_prio(chains->ns);
+		/* Firmware doesn't allow us to create another level 0 table,
+		 * so we create all unmanaged tables as level 1.
+		 *
+		 * To connect them, we use explicit miss rules with
+		 * ignore_flow_level. Caller is responsible to create
+		 * these rules (if needed).
+		 */
+		ft_attr.level = 1;
+		ns = mlx5_get_flow_namespace(chains->dev, chains->ns);
+	}
+
+	ft_attr.autogroup.num_reserved_entries = 2;
+	ft_attr.autogroup.max_num_groups = chains->group_num;
+	ft = mlx5_create_auto_grouped_flow_table(ns, &ft_attr);
+	if (IS_ERR(ft)) {
+		mlx5_core_warn(chains->dev, "Failed to create chains table err %d (chain: %d, prio: %d, level: %d, size: %d)\n",
+			       (int)PTR_ERR(ft), chain, prio, level, sz);
+		return ft;
+	}
+
+	return ft;
+}
+
+static int
+create_chain_restore(struct fs_chain *chain)
+{
+	struct mlx5_eswitch *esw = chain->chains->dev->priv.eswitch;
+	u8 modact[MLX5_UN_SZ_BYTES(set_add_copy_action_in_auto)] = {};
+	struct mlx5_fs_chains *chains = chain->chains;
+	enum mlx5e_tc_attr_to_reg chain_to_reg;
+	struct mlx5_modify_hdr *mod_hdr;
+	u32 index;
+	int err;
+
+	if (chain->chain == mlx5_chains_get_nf_ft_chain(chains) ||
+	    !mlx5_chains_prios_supported(chains))
+		return 0;
+
+	err = mlx5_chains_get_chain_mapping(chains, chain->chain, &index);
+	if (err)
+		return err;
+	if (index == MLX5_FS_DEFAULT_FLOW_TAG) {
+		/* we got the special default flow tag id, so we won't know
+		 * if we actually marked the packet with the restore rule
+		 * we create.
+		 *
+		 * This case isn't possible with MLX5_FS_DEFAULT_FLOW_TAG = 0.
+		 */
+		err = mlx5_chains_get_chain_mapping(chains, chain->chain, &index);
+		mapping_remove(chains->chains_mapping, MLX5_FS_DEFAULT_FLOW_TAG);
+		if (err)
+			return err;
+	}
+
+	chain->id = index;
+
+	if (chains->ns == MLX5_FLOW_NAMESPACE_FDB) {
+		chain_to_reg = CHAIN_TO_REG;
+		chain->restore_rule = esw_add_restore_rule(esw, chain->id);
+		if (IS_ERR(chain->restore_rule)) {
+			err = PTR_ERR(chain->restore_rule);
+			goto err_rule;
+		}
+	} else if (chains->ns == MLX5_FLOW_NAMESPACE_KERNEL) {
+		/* For NIC RX we don't need a restore rule
+		 * since we write the metadata to reg_b
+		 * that is passed to SW directly.
+		 */
+		chain_to_reg = NIC_CHAIN_TO_REG;
+	} else {
+		err = -EINVAL;
+		goto err_rule;
+	}
+
+	MLX5_SET(set_action_in, modact, action_type, MLX5_ACTION_TYPE_SET);
+	MLX5_SET(set_action_in, modact, field,
+		 mlx5e_tc_attr_to_reg_mappings[chain_to_reg].mfield);
+	MLX5_SET(set_action_in, modact, offset,
+		 mlx5e_tc_attr_to_reg_mappings[chain_to_reg].moffset);
+	MLX5_SET(set_action_in, modact, length,
+		 mlx5e_tc_attr_to_reg_mappings[chain_to_reg].mlen == 32 ?
+		 0 : mlx5e_tc_attr_to_reg_mappings[chain_to_reg].mlen);
+	MLX5_SET(set_action_in, modact, data, chain->id);
+	mod_hdr = mlx5_modify_header_alloc(chains->dev, chains->ns,
+					   1, modact);
+	if (IS_ERR(mod_hdr)) {
+		err = PTR_ERR(mod_hdr);
+		goto err_mod_hdr;
+	}
+	chain->miss_modify_hdr = mod_hdr;
+
+	return 0;
+
+err_mod_hdr:
+	if (!IS_ERR_OR_NULL(chain->restore_rule))
+		mlx5_del_flow_rules(chain->restore_rule);
+err_rule:
+	/* Datapath can't find this mapping, so we can safely remove it */
+	mapping_remove(chains->chains_mapping, chain->id);
+	return err;
+}
+
+static void destroy_chain_restore(struct fs_chain *chain)
+{
+	struct mlx5_fs_chains *chains = chain->chains;
+
+	if (!chain->miss_modify_hdr)
+		return;
+
+	if (chain->restore_rule)
+		mlx5_del_flow_rules(chain->restore_rule);
+
+	mlx5_modify_header_dealloc(chains->dev, chain->miss_modify_hdr);
+	mapping_remove(chains->chains_mapping, chain->id);
+}
+
+static struct fs_chain *
+mlx5_chains_create_chain(struct mlx5_fs_chains *chains, u32 chain)
+{
+	struct fs_chain *chain_s = NULL;
+	int err;
+
+	chain_s = kvzalloc(sizeof(*chain_s), GFP_KERNEL);
+	if (!chain_s)
+		return ERR_PTR(-ENOMEM);
+
+	chain_s->chains = chains;
+	chain_s->chain = chain;
+	INIT_LIST_HEAD(&chain_s->prios_list);
+
+	err = create_chain_restore(chain_s);
+	if (err)
+		goto err_restore;
+
+	err = rhashtable_insert_fast(&chains_ht(chains), &chain_s->node,
+				     chain_params);
+	if (err)
+		goto err_insert;
+
+	return chain_s;
+
+err_insert:
+	destroy_chain_restore(chain_s);
+err_restore:
+	kvfree(chain_s);
+	return ERR_PTR(err);
+}
+
+static void
+mlx5_chains_destroy_chain(struct fs_chain *chain)
+{
+	struct mlx5_fs_chains *chains = chain->chains;
+
+	rhashtable_remove_fast(&chains_ht(chains), &chain->node,
+			       chain_params);
+
+	destroy_chain_restore(chain);
+	kvfree(chain);
+}
+
+static struct fs_chain *
+mlx5_chains_get_chain(struct mlx5_fs_chains *chains, u32 chain)
+{
+	struct fs_chain *chain_s;
+
+	chain_s = rhashtable_lookup_fast(&chains_ht(chains), &chain,
+					 chain_params);
+	if (!chain_s) {
+		chain_s = mlx5_chains_create_chain(chains, chain);
+		if (IS_ERR(chain_s))
+			return chain_s;
+	}
+
+	chain_s->ref++;
+
+	return chain_s;
+}
+
+static struct mlx5_flow_handle *
+mlx5_chains_add_miss_rule(struct fs_chain *chain,
+			  struct mlx5_flow_table *ft,
+			  struct mlx5_flow_table *next_ft)
+{
+	struct mlx5_fs_chains *chains = chain->chains;
+	struct mlx5_flow_destination dest = {};
+	struct mlx5_flow_act act = {};
+
+	act.flags  = FLOW_ACT_NO_APPEND;
+	if (mlx5_chains_ignore_flow_level_supported(chain->chains))
+		act.flags |= FLOW_ACT_IGNORE_FLOW_LEVEL;
+
+	act.action = MLX5_FLOW_CONTEXT_ACTION_FWD_DEST;
+	dest.type  = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
+	dest.ft = next_ft;
+
+	if (next_ft == tc_end_ft(chains) &&
+	    chain->chain != mlx5_chains_get_nf_ft_chain(chains) &&
+	    mlx5_chains_prios_supported(chains)) {
+		act.modify_hdr = chain->miss_modify_hdr;
+		act.action |= MLX5_FLOW_CONTEXT_ACTION_MOD_HDR;
+	}
+
+	return mlx5_add_flow_rules(ft, NULL, &act, &dest, 1);
+}
+
+static int
+mlx5_chains_update_prio_prevs(struct prio *prio,
+			      struct mlx5_flow_table *next_ft)
+{
+	struct mlx5_flow_handle *miss_rules[FDB_TC_LEVELS_PER_PRIO + 1] = {};
+	struct fs_chain *chain = prio->chain;
+	struct prio *pos;
+	int n = 0, err;
+
+	if (prio->key.level)
+		return 0;
+
+	/* Iterate in reverse order until reaching the level 0 rule of
+	 * the previous priority, adding all the miss rules first, so we can
+	 * revert them if any of them fails.
+	 */
+	pos = prio;
+	list_for_each_entry_continue_reverse(pos,
+					     &chain->prios_list,
+					     list) {
+		miss_rules[n] = mlx5_chains_add_miss_rule(chain,
+							  pos->ft,
+							  next_ft);
+		if (IS_ERR(miss_rules[n])) {
+			err = PTR_ERR(miss_rules[n]);
+			goto err_prev_rule;
+		}
+
+		n++;
+		if (!pos->key.level)
+			break;
+	}
+
+	/* Success, delete old miss rules, and update the pointers. */
+	n = 0;
+	pos = prio;
+	list_for_each_entry_continue_reverse(pos,
+					     &chain->prios_list,
+					     list) {
+		mlx5_del_flow_rules(pos->miss_rule);
+
+		pos->miss_rule = miss_rules[n];
+		pos->next_ft = next_ft;
+
+		n++;
+		if (!pos->key.level)
+			break;
+	}
+
+	return 0;
+
+err_prev_rule:
+	while (--n >= 0)
+		mlx5_del_flow_rules(miss_rules[n]);
+
+	return err;
+}
+
+static void
+mlx5_chains_put_chain(struct fs_chain *chain)
+{
+	if (--chain->ref == 0)
+		mlx5_chains_destroy_chain(chain);
+}
+
+static struct prio *
+mlx5_chains_create_prio(struct mlx5_fs_chains *chains,
+			u32 chain, u32 prio, u32 level)
+{
+	int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
+	struct mlx5_flow_handle *miss_rule;
+	struct mlx5_flow_group *miss_group;
+	struct mlx5_flow_table *next_ft;
+	struct mlx5_flow_table *ft;
+	struct fs_chain *chain_s;
+	struct list_head *pos;
+	struct prio *prio_s;
+	u32 *flow_group_in;
+	int err;
+
+	chain_s = mlx5_chains_get_chain(chains, chain);
+	if (IS_ERR(chain_s))
+		return ERR_CAST(chain_s);
+
+	prio_s = kvzalloc(sizeof(*prio_s), GFP_KERNEL);
+	flow_group_in = kvzalloc(inlen, GFP_KERNEL);
+	if (!prio_s || !flow_group_in) {
+		err = -ENOMEM;
+		goto err_alloc;
+	}
+
+	/* Chain's prio list is sorted by prio and level.
+	 * And all levels of some prio point to the next prio's level 0.
+	 * Example list (prio, level):
+	 * (3,0)->(3,1)->(5,0)->(5,1)->(6,1)->(7,0)
+	 * In hardware, we will we have the following pointers:
+	 * (3,0) -> (5,0) -> (7,0) -> Slow path
+	 * (3,1) -> (5,0)
+	 * (5,1) -> (7,0)
+	 * (6,1) -> (7,0)
+	 */
+
+	/* Default miss for each chain: */
+	next_ft = (chain == mlx5_chains_get_nf_ft_chain(chains)) ?
+		  tc_default_ft(chains) :
+		  tc_end_ft(chains);
+	list_for_each(pos, &chain_s->prios_list) {
+		struct prio *p = list_entry(pos, struct prio, list);
+
+		/* exit on first pos that is larger */
+		if (prio < p->key.prio || (prio == p->key.prio &&
+					   level < p->key.level)) {
+			/* Get next level 0 table */
+			next_ft = p->key.level == 0 ? p->ft : p->next_ft;
+			break;
+		}
+	}
+
+	ft = mlx5_chains_create_table(chains, chain, prio, level);
+	if (IS_ERR(ft)) {
+		err = PTR_ERR(ft);
+		goto err_create;
+	}
+
+	MLX5_SET(create_flow_group_in, flow_group_in, start_flow_index,
+		 ft->max_fte - 2);
+	MLX5_SET(create_flow_group_in, flow_group_in, end_flow_index,
+		 ft->max_fte - 1);
+	miss_group = mlx5_create_flow_group(ft, flow_group_in);
+	if (IS_ERR(miss_group)) {
+		err = PTR_ERR(miss_group);
+		goto err_group;
+	}
+
+	/* Add miss rule to next_ft */
+	miss_rule = mlx5_chains_add_miss_rule(chain_s, ft, next_ft);
+	if (IS_ERR(miss_rule)) {
+		err = PTR_ERR(miss_rule);
+		goto err_miss_rule;
+	}
+
+	prio_s->miss_group = miss_group;
+	prio_s->miss_rule = miss_rule;
+	prio_s->next_ft = next_ft;
+	prio_s->chain = chain_s;
+	prio_s->key.chain = chain;
+	prio_s->key.prio = prio;
+	prio_s->key.level = level;
+	prio_s->ft = ft;
+
+	err = rhashtable_insert_fast(&prios_ht(chains), &prio_s->node,
+				     prio_params);
+	if (err)
+		goto err_insert;
+
+	list_add(&prio_s->list, pos->prev);
+
+	/* Table is ready, connect it */
+	err = mlx5_chains_update_prio_prevs(prio_s, ft);
+	if (err)
+		goto err_update;
+
+	kvfree(flow_group_in);
+	return prio_s;
+
+err_update:
+	list_del(&prio_s->list);
+	rhashtable_remove_fast(&prios_ht(chains), &prio_s->node,
+			       prio_params);
+err_insert:
+	mlx5_del_flow_rules(miss_rule);
+err_miss_rule:
+	mlx5_destroy_flow_group(miss_group);
+err_group:
+	mlx5_destroy_flow_table(ft);
+err_create:
+err_alloc:
+	kvfree(prio_s);
+	kvfree(flow_group_in);
+	mlx5_chains_put_chain(chain_s);
+	return ERR_PTR(err);
+}
+
+static void
+mlx5_chains_destroy_prio(struct mlx5_fs_chains *chains,
+			 struct prio *prio)
+{
+	struct fs_chain *chain = prio->chain;
+
+	WARN_ON(mlx5_chains_update_prio_prevs(prio,
+					      prio->next_ft));
+
+	list_del(&prio->list);
+	rhashtable_remove_fast(&prios_ht(chains), &prio->node,
+			       prio_params);
+	mlx5_del_flow_rules(prio->miss_rule);
+	mlx5_destroy_flow_group(prio->miss_group);
+	mlx5_destroy_flow_table(prio->ft);
+	mlx5_chains_put_chain(chain);
+	kvfree(prio);
+}
+
+struct mlx5_flow_table *
+mlx5_chains_get_table(struct mlx5_fs_chains *chains, u32 chain, u32 prio,
+		      u32 level)
+{
+	struct mlx5_flow_table *prev_fts;
+	struct prio *prio_s;
+	struct prio_key key;
+	int l = 0;
+
+	if ((chain > mlx5_chains_get_chain_range(chains) &&
+	     chain != mlx5_chains_get_nf_ft_chain(chains)) ||
+	    prio > mlx5_chains_get_prio_range(chains) ||
+	    level > mlx5_chains_get_level_range(chains))
+		return ERR_PTR(-EOPNOTSUPP);
+
+	/* create earlier levels for correct fs_core lookup when
+	 * connecting tables.
+	 */
+	for (l = 0; l < level; l++) {
+		prev_fts = mlx5_chains_get_table(chains, chain, prio, l);
+		if (IS_ERR(prev_fts)) {
+			prio_s = ERR_CAST(prev_fts);
+			goto err_get_prevs;
+		}
+	}
+
+	key.chain = chain;
+	key.prio = prio;
+	key.level = level;
+
+	mutex_lock(&chains_lock(chains));
+	prio_s = rhashtable_lookup_fast(&prios_ht(chains), &key,
+					prio_params);
+	if (!prio_s) {
+		prio_s = mlx5_chains_create_prio(chains, chain,
+						 prio, level);
+		if (IS_ERR(prio_s))
+			goto err_create_prio;
+	}
+
+	++prio_s->ref;
+	mutex_unlock(&chains_lock(chains));
+
+	return prio_s->ft;
+
+err_create_prio:
+	mutex_unlock(&chains_lock(chains));
+err_get_prevs:
+	while (--l >= 0)
+		mlx5_chains_put_table(chains, chain, prio, l);
+	return ERR_CAST(prio_s);
+}
+
+void
+mlx5_chains_put_table(struct mlx5_fs_chains *chains, u32 chain, u32 prio,
+		      u32 level)
+{
+	struct prio *prio_s;
+	struct prio_key key;
+
+	key.chain = chain;
+	key.prio = prio;
+	key.level = level;
+
+	mutex_lock(&chains_lock(chains));
+	prio_s = rhashtable_lookup_fast(&prios_ht(chains), &key,
+					prio_params);
+	if (!prio_s)
+		goto err_get_prio;
+
+	if (--prio_s->ref == 0)
+		mlx5_chains_destroy_prio(chains, prio_s);
+	mutex_unlock(&chains_lock(chains));
+
+	while (level-- > 0)
+		mlx5_chains_put_table(chains, chain, prio, level);
+
+	return;
+
+err_get_prio:
+	mutex_unlock(&chains_lock(chains));
+	WARN_ONCE(1,
+		  "Couldn't find table: (chain: %d prio: %d level: %d)",
+		  chain, prio, level);
+}
+
+struct mlx5_flow_table *
+mlx5_chains_get_tc_end_ft(struct mlx5_fs_chains *chains)
+{
+	return tc_end_ft(chains);
+}
+
+struct mlx5_flow_table *
+mlx5_chains_create_global_table(struct mlx5_fs_chains *chains)
+{
+	u32 chain, prio, level;
+	int err;
+
+	if (!mlx5_chains_ignore_flow_level_supported(chains)) {
+		err = -EOPNOTSUPP;
+
+		mlx5_core_warn(chains->dev,
+			       "Couldn't create global flow table, ignore_flow_level not supported.");
+		goto err_ignore;
+	}
+
+	chain = mlx5_chains_get_chain_range(chains),
+	prio = mlx5_chains_get_prio_range(chains);
+	level = mlx5_chains_get_level_range(chains);
+
+	return mlx5_chains_create_table(chains, chain, prio, level);
+
+err_ignore:
+	return ERR_PTR(err);
+}
+
+void
+mlx5_chains_destroy_global_table(struct mlx5_fs_chains *chains,
+				 struct mlx5_flow_table *ft)
+{
+	mlx5_destroy_flow_table(ft);
+}
+
+static struct mlx5_fs_chains *
+mlx5_chains_init(struct mlx5_core_dev *dev, struct mlx5_chains_attr *attr)
+{
+	struct mlx5_fs_chains *chains_priv;
+	u32 max_flow_counter;
+	int err;
+
+	chains_priv = kzalloc(sizeof(*chains_priv), GFP_KERNEL);
+	if (!chains_priv)
+		return ERR_PTR(-ENOMEM);
+
+	max_flow_counter = (MLX5_CAP_GEN(dev, max_flow_counter_31_16) << 16) |
+			    MLX5_CAP_GEN(dev, max_flow_counter_15_0);
+
+	mlx5_core_dbg(dev,
+		      "Init flow table chains, max counters(%d), groups(%d), max flow table size(%d)\n",
+		      max_flow_counter, attr->max_grp_num, attr->max_ft_sz);
+
+	chains_priv->dev = dev;
+	chains_priv->flags = attr->flags;
+	chains_priv->ns = attr->ns;
+	chains_priv->group_num = attr->max_grp_num;
+	chains_priv->chains_mapping = attr->mapping;
+	tc_default_ft(chains_priv) = tc_end_ft(chains_priv) = attr->default_ft;
+
+	mlx5_core_info(dev, "Supported tc offload range - chains: %u, prios: %u\n",
+		       mlx5_chains_get_chain_range(chains_priv),
+		       mlx5_chains_get_prio_range(chains_priv));
+
+	err = rhashtable_init(&chains_ht(chains_priv), &chain_params);
+	if (err)
+		goto init_chains_ht_err;
+
+	err = rhashtable_init(&prios_ht(chains_priv), &prio_params);
+	if (err)
+		goto init_prios_ht_err;
+
+	mutex_init(&chains_lock(chains_priv));
+
+	return chains_priv;
+
+init_prios_ht_err:
+	rhashtable_destroy(&chains_ht(chains_priv));
+init_chains_ht_err:
+	kfree(chains_priv);
+	return ERR_PTR(err);
+}
+
+static void
+mlx5_chains_cleanup(struct mlx5_fs_chains *chains)
+{
+	mutex_destroy(&chains_lock(chains));
+	rhashtable_destroy(&prios_ht(chains));
+	rhashtable_destroy(&chains_ht(chains));
+
+	kfree(chains);
+}
+
+struct mlx5_fs_chains *
+mlx5_chains_create(struct mlx5_core_dev *dev, struct mlx5_chains_attr *attr)
+{
+	struct mlx5_fs_chains *chains;
+
+	chains = mlx5_chains_init(dev, attr);
+
+	return chains;
+}
+
+void
+mlx5_chains_destroy(struct mlx5_fs_chains *chains)
+{
+	mlx5_chains_cleanup(chains);
+}
+
+int
+mlx5_chains_get_chain_mapping(struct mlx5_fs_chains *chains, u32 chain,
+			      u32 *chain_mapping)
+{
+	struct mapping_ctx *ctx = chains->chains_mapping;
+	struct mlx5_mapped_obj mapped_obj = {};
+
+	mapped_obj.type = MLX5_MAPPED_OBJ_CHAIN;
+	mapped_obj.chain = chain;
+	return mapping_add(ctx, &mapped_obj, chain_mapping);
+}
+
+int
+mlx5_chains_put_chain_mapping(struct mlx5_fs_chains *chains, u32 chain_mapping)
+{
+	struct mapping_ctx *ctx = chains->chains_mapping;
+
+	return mapping_remove(ctx, chain_mapping);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/fs_chains.h b/drivers/net/ethernet/mellanox/mlx5/core/lib/fs_chains.h
new file mode 100644
index 000000000..d50bdb226
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/fs_chains.h
@@ -0,0 +1,96 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2020 Mellanox Technologies. */
+
+#ifndef __ML5_ESW_CHAINS_H__
+#define __ML5_ESW_CHAINS_H__
+
+#include <linux/mlx5/fs.h>
+
+struct mlx5_fs_chains;
+struct mlx5_mapped_obj;
+
+enum mlx5_chains_flags {
+	MLX5_CHAINS_AND_PRIOS_SUPPORTED = BIT(0),
+	MLX5_CHAINS_IGNORE_FLOW_LEVEL_SUPPORTED = BIT(1),
+	MLX5_CHAINS_FT_TUNNEL_SUPPORTED = BIT(2),
+};
+
+struct mlx5_chains_attr {
+	enum mlx5_flow_namespace_type ns;
+	u32 flags;
+	u32 max_ft_sz;
+	u32 max_grp_num;
+	struct mlx5_flow_table *default_ft;
+	struct mapping_ctx *mapping;
+};
+
+#if IS_ENABLED(CONFIG_MLX5_CLS_ACT)
+
+bool
+mlx5_chains_prios_supported(struct mlx5_fs_chains *chains);
+bool mlx5_chains_ignore_flow_level_supported(struct mlx5_fs_chains *chains);
+bool
+mlx5_chains_backwards_supported(struct mlx5_fs_chains *chains);
+u32
+mlx5_chains_get_prio_range(struct mlx5_fs_chains *chains);
+u32
+mlx5_chains_get_chain_range(struct mlx5_fs_chains *chains);
+u32
+mlx5_chains_get_nf_ft_chain(struct mlx5_fs_chains *chains);
+
+struct mlx5_flow_table *
+mlx5_chains_get_table(struct mlx5_fs_chains *chains, u32 chain, u32 prio,
+		      u32 level);
+void
+mlx5_chains_put_table(struct mlx5_fs_chains *chains, u32 chain, u32 prio,
+		      u32 level);
+
+struct mlx5_flow_table *
+mlx5_chains_get_tc_end_ft(struct mlx5_fs_chains *chains);
+
+struct mlx5_flow_table *
+mlx5_chains_create_global_table(struct mlx5_fs_chains *chains);
+void
+mlx5_chains_destroy_global_table(struct mlx5_fs_chains *chains,
+				 struct mlx5_flow_table *ft);
+
+int
+mlx5_chains_get_chain_mapping(struct mlx5_fs_chains *chains, u32 chain,
+			      u32 *chain_mapping);
+int
+mlx5_chains_put_chain_mapping(struct mlx5_fs_chains *chains,
+			      u32 chain_mapping);
+
+struct mlx5_fs_chains *
+mlx5_chains_create(struct mlx5_core_dev *dev, struct mlx5_chains_attr *attr);
+void mlx5_chains_destroy(struct mlx5_fs_chains *chains);
+
+void
+mlx5_chains_set_end_ft(struct mlx5_fs_chains *chains,
+		       struct mlx5_flow_table *ft);
+
+#else /* CONFIG_MLX5_CLS_ACT */
+
+static inline bool
+mlx5_chains_ignore_flow_level_supported(struct mlx5_fs_chains *chains)
+{ return false; }
+
+static inline struct mlx5_flow_table *
+mlx5_chains_get_table(struct mlx5_fs_chains *chains, u32 chain, u32 prio,
+		      u32 level) { return ERR_PTR(-EOPNOTSUPP); }
+static inline void
+mlx5_chains_put_table(struct mlx5_fs_chains *chains, u32 chain, u32 prio,
+		      u32 level) {};
+
+static inline struct mlx5_flow_table *
+mlx5_chains_get_tc_end_ft(struct mlx5_fs_chains *chains) { return ERR_PTR(-EOPNOTSUPP); }
+
+static inline struct mlx5_fs_chains *
+mlx5_chains_create(struct mlx5_core_dev *dev, struct mlx5_chains_attr *attr)
+{ return NULL; }
+static inline void
+mlx5_chains_destroy(struct mlx5_fs_chains *chains) {};
+
+#endif /* CONFIG_MLX5_CLS_ACT */
+
+#endif /* __ML5_ESW_CHAINS_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/fs_ttc.c b/drivers/net/ethernet/mellanox/mlx5/core/lib/fs_ttc.c
new file mode 100644
index 000000000..b78f2ba25
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/fs_ttc.c
@@ -0,0 +1,608 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+// Copyright (c) 2021 NVIDIA CORPORATION & AFFILIATES.
+
+#include <linux/ip.h>
+#include <linux/ipv6.h>
+#include <linux/tcp.h>
+#include <linux/mlx5/fs.h>
+#include <linux/mlx5/driver.h>
+#include "mlx5_core.h"
+#include "lib/fs_ttc.h"
+
+#define MLX5_TTC_NUM_GROUPS	3
+#define MLX5_TTC_GROUP1_SIZE	(BIT(3) + MLX5_NUM_TUNNEL_TT)
+#define MLX5_TTC_GROUP2_SIZE	 BIT(1)
+#define MLX5_TTC_GROUP3_SIZE	 BIT(0)
+#define MLX5_TTC_TABLE_SIZE	(MLX5_TTC_GROUP1_SIZE +\
+				 MLX5_TTC_GROUP2_SIZE +\
+				 MLX5_TTC_GROUP3_SIZE)
+
+#define MLX5_INNER_TTC_NUM_GROUPS	3
+#define MLX5_INNER_TTC_GROUP1_SIZE	BIT(3)
+#define MLX5_INNER_TTC_GROUP2_SIZE	BIT(1)
+#define MLX5_INNER_TTC_GROUP3_SIZE	BIT(0)
+#define MLX5_INNER_TTC_TABLE_SIZE	(MLX5_INNER_TTC_GROUP1_SIZE +\
+					 MLX5_INNER_TTC_GROUP2_SIZE +\
+					 MLX5_INNER_TTC_GROUP3_SIZE)
+
+/* L3/L4 traffic type classifier */
+struct mlx5_ttc_table {
+	int num_groups;
+	struct mlx5_flow_table *t;
+	struct mlx5_flow_group **g;
+	struct mlx5_ttc_rule rules[MLX5_NUM_TT];
+	struct mlx5_flow_handle *tunnel_rules[MLX5_NUM_TUNNEL_TT];
+};
+
+struct mlx5_flow_table *mlx5_get_ttc_flow_table(struct mlx5_ttc_table *ttc)
+{
+	return ttc->t;
+}
+
+static void mlx5_cleanup_ttc_rules(struct mlx5_ttc_table *ttc)
+{
+	int i;
+
+	for (i = 0; i < MLX5_NUM_TT; i++) {
+		if (!IS_ERR_OR_NULL(ttc->rules[i].rule)) {
+			mlx5_del_flow_rules(ttc->rules[i].rule);
+			ttc->rules[i].rule = NULL;
+		}
+	}
+
+	for (i = 0; i < MLX5_NUM_TUNNEL_TT; i++) {
+		if (!IS_ERR_OR_NULL(ttc->tunnel_rules[i])) {
+			mlx5_del_flow_rules(ttc->tunnel_rules[i]);
+			ttc->tunnel_rules[i] = NULL;
+		}
+	}
+}
+
+struct mlx5_etype_proto {
+	u16 etype;
+	u8 proto;
+};
+
+static struct mlx5_etype_proto ttc_rules[] = {
+	[MLX5_TT_IPV4_TCP] = {
+		.etype = ETH_P_IP,
+		.proto = IPPROTO_TCP,
+	},
+	[MLX5_TT_IPV6_TCP] = {
+		.etype = ETH_P_IPV6,
+		.proto = IPPROTO_TCP,
+	},
+	[MLX5_TT_IPV4_UDP] = {
+		.etype = ETH_P_IP,
+		.proto = IPPROTO_UDP,
+	},
+	[MLX5_TT_IPV6_UDP] = {
+		.etype = ETH_P_IPV6,
+		.proto = IPPROTO_UDP,
+	},
+	[MLX5_TT_IPV4_IPSEC_AH] = {
+		.etype = ETH_P_IP,
+		.proto = IPPROTO_AH,
+	},
+	[MLX5_TT_IPV6_IPSEC_AH] = {
+		.etype = ETH_P_IPV6,
+		.proto = IPPROTO_AH,
+	},
+	[MLX5_TT_IPV4_IPSEC_ESP] = {
+		.etype = ETH_P_IP,
+		.proto = IPPROTO_ESP,
+	},
+	[MLX5_TT_IPV6_IPSEC_ESP] = {
+		.etype = ETH_P_IPV6,
+		.proto = IPPROTO_ESP,
+	},
+	[MLX5_TT_IPV4] = {
+		.etype = ETH_P_IP,
+		.proto = 0,
+	},
+	[MLX5_TT_IPV6] = {
+		.etype = ETH_P_IPV6,
+		.proto = 0,
+	},
+	[MLX5_TT_ANY] = {
+		.etype = 0,
+		.proto = 0,
+	},
+};
+
+static struct mlx5_etype_proto ttc_tunnel_rules[] = {
+	[MLX5_TT_IPV4_GRE] = {
+		.etype = ETH_P_IP,
+		.proto = IPPROTO_GRE,
+	},
+	[MLX5_TT_IPV6_GRE] = {
+		.etype = ETH_P_IPV6,
+		.proto = IPPROTO_GRE,
+	},
+	[MLX5_TT_IPV4_IPIP] = {
+		.etype = ETH_P_IP,
+		.proto = IPPROTO_IPIP,
+	},
+	[MLX5_TT_IPV6_IPIP] = {
+		.etype = ETH_P_IPV6,
+		.proto = IPPROTO_IPIP,
+	},
+	[MLX5_TT_IPV4_IPV6] = {
+		.etype = ETH_P_IP,
+		.proto = IPPROTO_IPV6,
+	},
+	[MLX5_TT_IPV6_IPV6] = {
+		.etype = ETH_P_IPV6,
+		.proto = IPPROTO_IPV6,
+	},
+
+};
+
+u8 mlx5_get_proto_by_tunnel_type(enum mlx5_tunnel_types tt)
+{
+	return ttc_tunnel_rules[tt].proto;
+}
+
+static bool mlx5_tunnel_proto_supported_rx(struct mlx5_core_dev *mdev,
+					   u8 proto_type)
+{
+	switch (proto_type) {
+	case IPPROTO_GRE:
+		return MLX5_CAP_ETH(mdev, tunnel_stateless_gre);
+	case IPPROTO_IPIP:
+	case IPPROTO_IPV6:
+		return (MLX5_CAP_ETH(mdev, tunnel_stateless_ip_over_ip) ||
+			MLX5_CAP_ETH(mdev, tunnel_stateless_ip_over_ip_rx));
+	default:
+		return false;
+	}
+}
+
+static bool mlx5_tunnel_any_rx_proto_supported(struct mlx5_core_dev *mdev)
+{
+	int tt;
+
+	for (tt = 0; tt < MLX5_NUM_TUNNEL_TT; tt++) {
+		if (mlx5_tunnel_proto_supported_rx(mdev,
+						   ttc_tunnel_rules[tt].proto))
+			return true;
+	}
+	return false;
+}
+
+bool mlx5_tunnel_inner_ft_supported(struct mlx5_core_dev *mdev)
+{
+	return (mlx5_tunnel_any_rx_proto_supported(mdev) &&
+		MLX5_CAP_FLOWTABLE_NIC_RX(mdev,
+					  ft_field_support.inner_ip_version));
+}
+
+static u8 mlx5_etype_to_ipv(u16 ethertype)
+{
+	if (ethertype == ETH_P_IP)
+		return 4;
+
+	if (ethertype == ETH_P_IPV6)
+		return 6;
+
+	return 0;
+}
+
+static struct mlx5_flow_handle *
+mlx5_generate_ttc_rule(struct mlx5_core_dev *dev, struct mlx5_flow_table *ft,
+		       struct mlx5_flow_destination *dest, u16 etype, u8 proto)
+{
+	int match_ipv_outer =
+		MLX5_CAP_FLOWTABLE_NIC_RX(dev,
+					  ft_field_support.outer_ip_version);
+	MLX5_DECLARE_FLOW_ACT(flow_act);
+	struct mlx5_flow_handle *rule;
+	struct mlx5_flow_spec *spec;
+	int err = 0;
+	u8 ipv;
+
+	spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
+	if (!spec)
+		return ERR_PTR(-ENOMEM);
+
+	if (proto) {
+		spec->match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
+		MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria, outer_headers.ip_protocol);
+		MLX5_SET(fte_match_param, spec->match_value, outer_headers.ip_protocol, proto);
+	}
+
+	ipv = mlx5_etype_to_ipv(etype);
+	if (match_ipv_outer && ipv) {
+		spec->match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
+		MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria, outer_headers.ip_version);
+		MLX5_SET(fte_match_param, spec->match_value, outer_headers.ip_version, ipv);
+	} else if (etype) {
+		spec->match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
+		MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria, outer_headers.ethertype);
+		MLX5_SET(fte_match_param, spec->match_value, outer_headers.ethertype, etype);
+	}
+
+	rule = mlx5_add_flow_rules(ft, spec, &flow_act, dest, 1);
+	if (IS_ERR(rule)) {
+		err = PTR_ERR(rule);
+		mlx5_core_err(dev, "%s: add rule failed\n", __func__);
+	}
+
+	kvfree(spec);
+	return err ? ERR_PTR(err) : rule;
+}
+
+static int mlx5_generate_ttc_table_rules(struct mlx5_core_dev *dev,
+					 struct ttc_params *params,
+					 struct mlx5_ttc_table *ttc)
+{
+	struct mlx5_flow_handle **trules;
+	struct mlx5_ttc_rule *rules;
+	struct mlx5_flow_table *ft;
+	int tt;
+	int err;
+
+	ft = ttc->t;
+	rules = ttc->rules;
+	for (tt = 0; tt < MLX5_NUM_TT; tt++) {
+		struct mlx5_ttc_rule *rule = &rules[tt];
+
+		if (test_bit(tt, params->ignore_dests))
+			continue;
+		rule->rule = mlx5_generate_ttc_rule(dev, ft, &params->dests[tt],
+						    ttc_rules[tt].etype,
+						    ttc_rules[tt].proto);
+		if (IS_ERR(rule->rule)) {
+			err = PTR_ERR(rule->rule);
+			rule->rule = NULL;
+			goto del_rules;
+		}
+		rule->default_dest = params->dests[tt];
+	}
+
+	if (!params->inner_ttc || !mlx5_tunnel_inner_ft_supported(dev))
+		return 0;
+
+	trules    = ttc->tunnel_rules;
+	for (tt = 0; tt < MLX5_NUM_TUNNEL_TT; tt++) {
+		if (!mlx5_tunnel_proto_supported_rx(dev,
+						    ttc_tunnel_rules[tt].proto))
+			continue;
+		if (test_bit(tt, params->ignore_tunnel_dests))
+			continue;
+		trules[tt] = mlx5_generate_ttc_rule(dev, ft,
+						    &params->tunnel_dests[tt],
+						    ttc_tunnel_rules[tt].etype,
+						    ttc_tunnel_rules[tt].proto);
+		if (IS_ERR(trules[tt])) {
+			err = PTR_ERR(trules[tt]);
+			trules[tt] = NULL;
+			goto del_rules;
+		}
+	}
+
+	return 0;
+
+del_rules:
+	mlx5_cleanup_ttc_rules(ttc);
+	return err;
+}
+
+static int mlx5_create_ttc_table_groups(struct mlx5_ttc_table *ttc,
+					bool use_ipv)
+{
+	int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
+	int ix = 0;
+	u32 *in;
+	int err;
+	u8 *mc;
+
+	ttc->g = kcalloc(MLX5_TTC_NUM_GROUPS, sizeof(*ttc->g), GFP_KERNEL);
+	if (!ttc->g)
+		return -ENOMEM;
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in) {
+		kfree(ttc->g);
+		ttc->g = NULL;
+		return -ENOMEM;
+	}
+
+	/* L4 Group */
+	mc = MLX5_ADDR_OF(create_flow_group_in, in, match_criteria);
+	MLX5_SET_TO_ONES(fte_match_param, mc, outer_headers.ip_protocol);
+	if (use_ipv)
+		MLX5_SET_TO_ONES(fte_match_param, mc, outer_headers.ip_version);
+	else
+		MLX5_SET_TO_ONES(fte_match_param, mc, outer_headers.ethertype);
+	MLX5_SET_CFG(in, match_criteria_enable, MLX5_MATCH_OUTER_HEADERS);
+	MLX5_SET_CFG(in, start_flow_index, ix);
+	ix += MLX5_TTC_GROUP1_SIZE;
+	MLX5_SET_CFG(in, end_flow_index, ix - 1);
+	ttc->g[ttc->num_groups] = mlx5_create_flow_group(ttc->t, in);
+	if (IS_ERR(ttc->g[ttc->num_groups]))
+		goto err;
+	ttc->num_groups++;
+
+	/* L3 Group */
+	MLX5_SET(fte_match_param, mc, outer_headers.ip_protocol, 0);
+	MLX5_SET_CFG(in, start_flow_index, ix);
+	ix += MLX5_TTC_GROUP2_SIZE;
+	MLX5_SET_CFG(in, end_flow_index, ix - 1);
+	ttc->g[ttc->num_groups] = mlx5_create_flow_group(ttc->t, in);
+	if (IS_ERR(ttc->g[ttc->num_groups]))
+		goto err;
+	ttc->num_groups++;
+
+	/* Any Group */
+	memset(in, 0, inlen);
+	MLX5_SET_CFG(in, start_flow_index, ix);
+	ix += MLX5_TTC_GROUP3_SIZE;
+	MLX5_SET_CFG(in, end_flow_index, ix - 1);
+	ttc->g[ttc->num_groups] = mlx5_create_flow_group(ttc->t, in);
+	if (IS_ERR(ttc->g[ttc->num_groups]))
+		goto err;
+	ttc->num_groups++;
+
+	kvfree(in);
+	return 0;
+
+err:
+	err = PTR_ERR(ttc->g[ttc->num_groups]);
+	ttc->g[ttc->num_groups] = NULL;
+	kvfree(in);
+
+	return err;
+}
+
+static struct mlx5_flow_handle *
+mlx5_generate_inner_ttc_rule(struct mlx5_core_dev *dev,
+			     struct mlx5_flow_table *ft,
+			     struct mlx5_flow_destination *dest,
+			     u16 etype, u8 proto)
+{
+	MLX5_DECLARE_FLOW_ACT(flow_act);
+	struct mlx5_flow_handle *rule;
+	struct mlx5_flow_spec *spec;
+	int err = 0;
+	u8 ipv;
+
+	spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
+	if (!spec)
+		return ERR_PTR(-ENOMEM);
+
+	ipv = mlx5_etype_to_ipv(etype);
+	if (etype && ipv) {
+		spec->match_criteria_enable = MLX5_MATCH_INNER_HEADERS;
+		MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria, inner_headers.ip_version);
+		MLX5_SET(fte_match_param, spec->match_value, inner_headers.ip_version, ipv);
+	}
+
+	if (proto) {
+		spec->match_criteria_enable = MLX5_MATCH_INNER_HEADERS;
+		MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria, inner_headers.ip_protocol);
+		MLX5_SET(fte_match_param, spec->match_value, inner_headers.ip_protocol, proto);
+	}
+
+	rule = mlx5_add_flow_rules(ft, spec, &flow_act, dest, 1);
+	if (IS_ERR(rule)) {
+		err = PTR_ERR(rule);
+		mlx5_core_err(dev, "%s: add inner TTC rule failed\n", __func__);
+	}
+
+	kvfree(spec);
+	return err ? ERR_PTR(err) : rule;
+}
+
+static int mlx5_generate_inner_ttc_table_rules(struct mlx5_core_dev *dev,
+					       struct ttc_params *params,
+					       struct mlx5_ttc_table *ttc)
+{
+	struct mlx5_ttc_rule *rules;
+	struct mlx5_flow_table *ft;
+	int err;
+	int tt;
+
+	ft = ttc->t;
+	rules = ttc->rules;
+
+	for (tt = 0; tt < MLX5_NUM_TT; tt++) {
+		struct mlx5_ttc_rule *rule = &rules[tt];
+
+		if (test_bit(tt, params->ignore_dests))
+			continue;
+		rule->rule = mlx5_generate_inner_ttc_rule(dev, ft,
+							  &params->dests[tt],
+							  ttc_rules[tt].etype,
+							  ttc_rules[tt].proto);
+		if (IS_ERR(rule->rule)) {
+			err = PTR_ERR(rule->rule);
+			rule->rule = NULL;
+			goto del_rules;
+		}
+		rule->default_dest = params->dests[tt];
+	}
+
+	return 0;
+
+del_rules:
+
+	mlx5_cleanup_ttc_rules(ttc);
+	return err;
+}
+
+static int mlx5_create_inner_ttc_table_groups(struct mlx5_ttc_table *ttc)
+{
+	int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
+	int ix = 0;
+	u32 *in;
+	int err;
+	u8 *mc;
+
+	ttc->g = kcalloc(MLX5_INNER_TTC_NUM_GROUPS, sizeof(*ttc->g),
+			 GFP_KERNEL);
+	if (!ttc->g)
+		return -ENOMEM;
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in) {
+		kfree(ttc->g);
+		ttc->g = NULL;
+		return -ENOMEM;
+	}
+
+	/* L4 Group */
+	mc = MLX5_ADDR_OF(create_flow_group_in, in, match_criteria);
+	MLX5_SET_TO_ONES(fte_match_param, mc, inner_headers.ip_protocol);
+	MLX5_SET_TO_ONES(fte_match_param, mc, inner_headers.ip_version);
+	MLX5_SET_CFG(in, match_criteria_enable, MLX5_MATCH_INNER_HEADERS);
+	MLX5_SET_CFG(in, start_flow_index, ix);
+	ix += MLX5_INNER_TTC_GROUP1_SIZE;
+	MLX5_SET_CFG(in, end_flow_index, ix - 1);
+	ttc->g[ttc->num_groups] = mlx5_create_flow_group(ttc->t, in);
+	if (IS_ERR(ttc->g[ttc->num_groups]))
+		goto err;
+	ttc->num_groups++;
+
+	/* L3 Group */
+	MLX5_SET(fte_match_param, mc, inner_headers.ip_protocol, 0);
+	MLX5_SET_CFG(in, start_flow_index, ix);
+	ix += MLX5_INNER_TTC_GROUP2_SIZE;
+	MLX5_SET_CFG(in, end_flow_index, ix - 1);
+	ttc->g[ttc->num_groups] = mlx5_create_flow_group(ttc->t, in);
+	if (IS_ERR(ttc->g[ttc->num_groups]))
+		goto err;
+	ttc->num_groups++;
+
+	/* Any Group */
+	memset(in, 0, inlen);
+	MLX5_SET_CFG(in, start_flow_index, ix);
+	ix += MLX5_INNER_TTC_GROUP3_SIZE;
+	MLX5_SET_CFG(in, end_flow_index, ix - 1);
+	ttc->g[ttc->num_groups] = mlx5_create_flow_group(ttc->t, in);
+	if (IS_ERR(ttc->g[ttc->num_groups]))
+		goto err;
+	ttc->num_groups++;
+
+	kvfree(in);
+	return 0;
+
+err:
+	err = PTR_ERR(ttc->g[ttc->num_groups]);
+	ttc->g[ttc->num_groups] = NULL;
+	kvfree(in);
+
+	return err;
+}
+
+struct mlx5_ttc_table *mlx5_create_inner_ttc_table(struct mlx5_core_dev *dev,
+						   struct ttc_params *params)
+{
+	struct mlx5_ttc_table *ttc;
+	int err;
+
+	ttc = kvzalloc(sizeof(*ttc), GFP_KERNEL);
+	if (!ttc)
+		return ERR_PTR(-ENOMEM);
+
+	WARN_ON_ONCE(params->ft_attr.max_fte);
+	params->ft_attr.max_fte = MLX5_INNER_TTC_TABLE_SIZE;
+	ttc->t = mlx5_create_flow_table(params->ns, &params->ft_attr);
+	if (IS_ERR(ttc->t)) {
+		err = PTR_ERR(ttc->t);
+		kvfree(ttc);
+		return ERR_PTR(err);
+	}
+
+	err = mlx5_create_inner_ttc_table_groups(ttc);
+	if (err)
+		goto destroy_ft;
+
+	err = mlx5_generate_inner_ttc_table_rules(dev, params, ttc);
+	if (err)
+		goto destroy_ft;
+
+	return ttc;
+
+destroy_ft:
+	mlx5_destroy_ttc_table(ttc);
+	return ERR_PTR(err);
+}
+
+void mlx5_destroy_ttc_table(struct mlx5_ttc_table *ttc)
+{
+	int i;
+
+	mlx5_cleanup_ttc_rules(ttc);
+	for (i = ttc->num_groups - 1; i >= 0; i--) {
+		if (!IS_ERR_OR_NULL(ttc->g[i]))
+			mlx5_destroy_flow_group(ttc->g[i]);
+		ttc->g[i] = NULL;
+	}
+
+	kfree(ttc->g);
+	mlx5_destroy_flow_table(ttc->t);
+	kvfree(ttc);
+}
+
+struct mlx5_ttc_table *mlx5_create_ttc_table(struct mlx5_core_dev *dev,
+					     struct ttc_params *params)
+{
+	bool match_ipv_outer =
+		MLX5_CAP_FLOWTABLE_NIC_RX(dev,
+					  ft_field_support.outer_ip_version);
+	struct mlx5_ttc_table *ttc;
+	int err;
+
+	ttc = kvzalloc(sizeof(*ttc), GFP_KERNEL);
+	if (!ttc)
+		return ERR_PTR(-ENOMEM);
+
+	WARN_ON_ONCE(params->ft_attr.max_fte);
+	params->ft_attr.max_fte = MLX5_TTC_TABLE_SIZE;
+	ttc->t = mlx5_create_flow_table(params->ns, &params->ft_attr);
+	if (IS_ERR(ttc->t)) {
+		err = PTR_ERR(ttc->t);
+		kvfree(ttc);
+		return ERR_PTR(err);
+	}
+
+	err = mlx5_create_ttc_table_groups(ttc, match_ipv_outer);
+	if (err)
+		goto destroy_ft;
+
+	err = mlx5_generate_ttc_table_rules(dev, params, ttc);
+	if (err)
+		goto destroy_ft;
+
+	return ttc;
+
+destroy_ft:
+	mlx5_destroy_ttc_table(ttc);
+	return ERR_PTR(err);
+}
+
+int mlx5_ttc_fwd_dest(struct mlx5_ttc_table *ttc, enum mlx5_traffic_types type,
+		      struct mlx5_flow_destination *new_dest)
+{
+	return mlx5_modify_rule_destination(ttc->rules[type].rule, new_dest,
+					    NULL);
+}
+
+struct mlx5_flow_destination
+mlx5_ttc_get_default_dest(struct mlx5_ttc_table *ttc,
+			  enum mlx5_traffic_types type)
+{
+	struct mlx5_flow_destination *dest = &ttc->rules[type].default_dest;
+
+	WARN_ONCE(dest->type != MLX5_FLOW_DESTINATION_TYPE_TIR,
+		  "TTC[%d] default dest is not setup yet", type);
+
+	return *dest;
+}
+
+int mlx5_ttc_fwd_default_dest(struct mlx5_ttc_table *ttc,
+			      enum mlx5_traffic_types type)
+{
+	struct mlx5_flow_destination dest = mlx5_ttc_get_default_dest(ttc, type);
+
+	return mlx5_ttc_fwd_dest(ttc, type, &dest);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/fs_ttc.h b/drivers/net/ethernet/mellanox/mlx5/core/lib/fs_ttc.h
new file mode 100644
index 000000000..85fef0cd1
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/fs_ttc.h
@@ -0,0 +1,72 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2020 Mellanox Technologies. */
+
+#ifndef __ML5_FS_TTC_H__
+#define __ML5_FS_TTC_H__
+
+#include <linux/mlx5/fs.h>
+
+enum mlx5_traffic_types {
+	MLX5_TT_IPV4_TCP,
+	MLX5_TT_IPV6_TCP,
+	MLX5_TT_IPV4_UDP,
+	MLX5_TT_IPV6_UDP,
+	MLX5_TT_IPV4_IPSEC_AH,
+	MLX5_TT_IPV6_IPSEC_AH,
+	MLX5_TT_IPV4_IPSEC_ESP,
+	MLX5_TT_IPV6_IPSEC_ESP,
+	MLX5_TT_IPV4,
+	MLX5_TT_IPV6,
+	MLX5_TT_ANY,
+	MLX5_NUM_TT,
+	MLX5_NUM_INDIR_TIRS = MLX5_TT_ANY,
+};
+
+enum mlx5_tunnel_types {
+	MLX5_TT_IPV4_GRE,
+	MLX5_TT_IPV6_GRE,
+	MLX5_TT_IPV4_IPIP,
+	MLX5_TT_IPV6_IPIP,
+	MLX5_TT_IPV4_IPV6,
+	MLX5_TT_IPV6_IPV6,
+	MLX5_NUM_TUNNEL_TT,
+};
+
+struct mlx5_ttc_rule {
+	struct mlx5_flow_handle *rule;
+	struct mlx5_flow_destination default_dest;
+};
+
+struct mlx5_ttc_table;
+
+struct ttc_params {
+	struct mlx5_flow_namespace *ns;
+	struct mlx5_flow_table_attr ft_attr;
+	struct mlx5_flow_destination dests[MLX5_NUM_TT];
+	DECLARE_BITMAP(ignore_dests, MLX5_NUM_TT);
+	bool   inner_ttc;
+	DECLARE_BITMAP(ignore_tunnel_dests, MLX5_NUM_TUNNEL_TT);
+	struct mlx5_flow_destination tunnel_dests[MLX5_NUM_TUNNEL_TT];
+};
+
+struct mlx5_flow_table *mlx5_get_ttc_flow_table(struct mlx5_ttc_table *ttc);
+
+struct mlx5_ttc_table *mlx5_create_ttc_table(struct mlx5_core_dev *dev,
+					     struct ttc_params *params);
+void mlx5_destroy_ttc_table(struct mlx5_ttc_table *ttc);
+
+struct mlx5_ttc_table *mlx5_create_inner_ttc_table(struct mlx5_core_dev *dev,
+						   struct ttc_params *params);
+
+int mlx5_ttc_fwd_dest(struct mlx5_ttc_table *ttc, enum mlx5_traffic_types type,
+		      struct mlx5_flow_destination *new_dest);
+struct mlx5_flow_destination
+mlx5_ttc_get_default_dest(struct mlx5_ttc_table *ttc,
+			  enum mlx5_traffic_types type);
+int mlx5_ttc_fwd_default_dest(struct mlx5_ttc_table *ttc,
+			      enum mlx5_traffic_types type);
+
+bool mlx5_tunnel_inner_ft_supported(struct mlx5_core_dev *mdev);
+u8 mlx5_get_proto_by_tunnel_type(enum mlx5_tunnel_types tt);
+
+#endif /* __MLX5_FS_TTC_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/geneve.c b/drivers/net/ethernet/mellanox/mlx5/core/lib/geneve.c
new file mode 100644
index 000000000..6dc83e871
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/geneve.c
@@ -0,0 +1,158 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2019 Mellanox Technologies. */
+
+#include <linux/kernel.h>
+#include "mlx5_core.h"
+#include "geneve.h"
+
+struct mlx5_geneve {
+	struct mlx5_core_dev *mdev;
+	__be16 opt_class;
+	u8 opt_type;
+	u32 obj_id;
+	struct mutex sync_lock; /* protect GENEVE obj operations */
+	u32 refcount;
+};
+
+static int mlx5_geneve_tlv_option_create(struct mlx5_core_dev *mdev,
+					 __be16 class,
+					 u8 type,
+					 u8 len)
+{
+	u32 in[MLX5_ST_SZ_DW(create_geneve_tlv_option_in)] = {};
+	u32 out[MLX5_ST_SZ_DW(general_obj_out_cmd_hdr)] = {};
+	u64 general_obj_types;
+	void *hdr, *opt;
+	u16 obj_id;
+	int err;
+
+	general_obj_types = MLX5_CAP_GEN_64(mdev, general_obj_types);
+	if (!(general_obj_types & MLX5_GENERAL_OBJ_TYPES_CAP_GENEVE_TLV_OPT))
+		return -EINVAL;
+
+	hdr = MLX5_ADDR_OF(create_geneve_tlv_option_in, in, hdr);
+	opt = MLX5_ADDR_OF(create_geneve_tlv_option_in, in, geneve_tlv_opt);
+
+	MLX5_SET(general_obj_in_cmd_hdr, hdr, opcode, MLX5_CMD_OP_CREATE_GENERAL_OBJECT);
+	MLX5_SET(general_obj_in_cmd_hdr, hdr, obj_type, MLX5_OBJ_TYPE_GENEVE_TLV_OPT);
+
+	MLX5_SET(geneve_tlv_option, opt, option_class, be16_to_cpu(class));
+	MLX5_SET(geneve_tlv_option, opt, option_type, type);
+	MLX5_SET(geneve_tlv_option, opt, option_data_length, len);
+
+	err = mlx5_cmd_exec(mdev, in, sizeof(in), out, sizeof(out));
+	if (err)
+		return err;
+
+	obj_id = MLX5_GET(general_obj_out_cmd_hdr, out, obj_id);
+	return obj_id;
+}
+
+static void mlx5_geneve_tlv_option_destroy(struct mlx5_core_dev *mdev, u16 obj_id)
+{
+	u32 out[MLX5_ST_SZ_DW(general_obj_out_cmd_hdr)] = {};
+	u32 in[MLX5_ST_SZ_DW(general_obj_in_cmd_hdr)] = {};
+
+	MLX5_SET(general_obj_in_cmd_hdr, in, opcode, MLX5_CMD_OP_DESTROY_GENERAL_OBJECT);
+	MLX5_SET(general_obj_in_cmd_hdr, in, obj_type, MLX5_OBJ_TYPE_GENEVE_TLV_OPT);
+	MLX5_SET(general_obj_in_cmd_hdr, in, obj_id, obj_id);
+
+	mlx5_cmd_exec(mdev, in, sizeof(in), out, sizeof(out));
+}
+
+int mlx5_geneve_tlv_option_add(struct mlx5_geneve *geneve, struct geneve_opt *opt)
+{
+	int res = 0;
+
+	if (IS_ERR_OR_NULL(geneve))
+		return -EOPNOTSUPP;
+
+	mutex_lock(&geneve->sync_lock);
+
+	if (geneve->refcount) {
+		if (geneve->opt_class == opt->opt_class &&
+		    geneve->opt_type == opt->type) {
+			/* We already have TLV options obj allocated */
+			geneve->refcount++;
+		} else {
+			/* TLV options obj allocated, but its params
+			 * do not match the new request.
+			 * We support only one such object.
+			 */
+			mlx5_core_warn(geneve->mdev,
+				       "Won't create Geneve TLV opt object with class:type:len = 0x%x:0x%x:%d (another class:type already exists)\n",
+				       be16_to_cpu(opt->opt_class),
+				       opt->type,
+				       opt->length);
+			res = -EOPNOTSUPP;
+			goto unlock;
+		}
+	} else {
+		/* We don't have any TLV options obj allocated */
+
+		res = mlx5_geneve_tlv_option_create(geneve->mdev,
+						    opt->opt_class,
+						    opt->type,
+						    opt->length);
+		if (res < 0) {
+			mlx5_core_warn(geneve->mdev,
+				       "Failed creating Geneve TLV opt object class:type:len = 0x%x:0x%x:%d (err=%d)\n",
+				       be16_to_cpu(opt->opt_class),
+				       opt->type, opt->length, res);
+			goto unlock;
+		}
+		geneve->opt_class = opt->opt_class;
+		geneve->opt_type = opt->type;
+		geneve->obj_id = res;
+		geneve->refcount++;
+		res = 0;
+	}
+
+unlock:
+	mutex_unlock(&geneve->sync_lock);
+	return res;
+}
+
+void mlx5_geneve_tlv_option_del(struct mlx5_geneve *geneve)
+{
+	if (IS_ERR_OR_NULL(geneve))
+		return;
+
+	mutex_lock(&geneve->sync_lock);
+	if (--geneve->refcount == 0) {
+		/* We've just removed the last user of Geneve option.
+		 * Now delete the object in FW.
+		 */
+		mlx5_geneve_tlv_option_destroy(geneve->mdev, geneve->obj_id);
+
+		geneve->opt_class = 0;
+		geneve->opt_type = 0;
+		geneve->obj_id = 0;
+	}
+	mutex_unlock(&geneve->sync_lock);
+}
+
+struct mlx5_geneve *mlx5_geneve_create(struct mlx5_core_dev *mdev)
+{
+	struct mlx5_geneve *geneve =
+		kzalloc(sizeof(*geneve), GFP_KERNEL);
+
+	if (!geneve)
+		return ERR_PTR(-ENOMEM);
+	geneve->mdev = mdev;
+	mutex_init(&geneve->sync_lock);
+
+	return geneve;
+}
+
+void mlx5_geneve_destroy(struct mlx5_geneve *geneve)
+{
+	if (IS_ERR_OR_NULL(geneve))
+		return;
+
+	/* Lockless since we are unloading */
+	if (geneve->refcount)
+		mlx5_geneve_tlv_option_destroy(geneve->mdev, geneve->obj_id);
+
+	kfree(geneve);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/geneve.h b/drivers/net/ethernet/mellanox/mlx5/core/lib/geneve.h
new file mode 100644
index 000000000..adee0cbba
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/geneve.h
@@ -0,0 +1,33 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2019 Mellanox Technologies. */
+
+#ifndef __MLX5_GENEVE_H__
+#define __MLX5_GENEVE_H__
+
+#include <net/geneve.h>
+#include <linux/mlx5/driver.h>
+
+struct mlx5_geneve;
+
+#ifdef CONFIG_MLX5_ESWITCH
+
+struct mlx5_geneve *mlx5_geneve_create(struct mlx5_core_dev *mdev);
+void mlx5_geneve_destroy(struct mlx5_geneve *geneve);
+
+int mlx5_geneve_tlv_option_add(struct mlx5_geneve *geneve, struct geneve_opt *opt);
+void mlx5_geneve_tlv_option_del(struct mlx5_geneve *geneve);
+
+#else /* CONFIG_MLX5_ESWITCH */
+
+static inline struct mlx5_geneve
+*mlx5_geneve_create(struct mlx5_core_dev *mdev) { return NULL; }
+static inline void
+mlx5_geneve_destroy(struct mlx5_geneve *geneve) {}
+static inline int
+mlx5_geneve_tlv_option_add(struct mlx5_geneve *geneve, struct geneve_opt *opt) { return 0; }
+static inline void
+mlx5_geneve_tlv_option_del(struct mlx5_geneve *geneve) {}
+
+#endif /* CONFIG_MLX5_ESWITCH */
+
+#endif /* __MLX5_GENEVE_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/gid.c b/drivers/net/ethernet/mellanox/mlx5/core/lib/gid.c
new file mode 100644
index 000000000..96ffc0a0e
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/gid.c
@@ -0,0 +1,152 @@
+/*
+ * Copyright (c) 2017, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/mlx5/driver.h>
+#include <linux/etherdevice.h>
+#include <linux/idr.h>
+#include "mlx5_core.h"
+#include "lib/mlx5.h"
+
+void mlx5_init_reserved_gids(struct mlx5_core_dev *dev)
+{
+	unsigned int tblsz = MLX5_CAP_ROCE(dev, roce_address_table_size);
+
+	ida_init(&dev->roce.reserved_gids.ida);
+	dev->roce.reserved_gids.start = tblsz;
+	dev->roce.reserved_gids.count = 0;
+}
+
+void mlx5_cleanup_reserved_gids(struct mlx5_core_dev *dev)
+{
+	WARN_ON(!ida_is_empty(&dev->roce.reserved_gids.ida));
+	dev->roce.reserved_gids.start = 0;
+	dev->roce.reserved_gids.count = 0;
+	ida_destroy(&dev->roce.reserved_gids.ida);
+}
+
+int mlx5_core_reserve_gids(struct mlx5_core_dev *dev, unsigned int count)
+{
+	if (dev->roce.reserved_gids.start < count) {
+		mlx5_core_warn(dev, "GID table exhausted attempting to reserve %d more GIDs\n",
+			       count);
+		return -ENOMEM;
+	}
+	if (dev->roce.reserved_gids.count + count > MLX5_MAX_RESERVED_GIDS) {
+		mlx5_core_warn(dev, "Unable to reserve %d more GIDs\n", count);
+		return -ENOMEM;
+	}
+
+	dev->roce.reserved_gids.start -= count;
+	dev->roce.reserved_gids.count += count;
+	mlx5_core_dbg(dev, "Reserved %u GIDs starting at %u\n",
+		      dev->roce.reserved_gids.count,
+		      dev->roce.reserved_gids.start);
+	return 0;
+}
+
+void mlx5_core_unreserve_gids(struct mlx5_core_dev *dev, unsigned int count)
+{
+	WARN(count > dev->roce.reserved_gids.count, "Unreserving %u GIDs when only %u reserved",
+	     count, dev->roce.reserved_gids.count);
+
+	dev->roce.reserved_gids.start += count;
+	dev->roce.reserved_gids.count -= count;
+	mlx5_core_dbg(dev, "%u GIDs starting at %u left reserved\n",
+		      dev->roce.reserved_gids.count,
+		      dev->roce.reserved_gids.start);
+}
+
+int mlx5_core_reserved_gid_alloc(struct mlx5_core_dev *dev, int *gid_index)
+{
+	int end = dev->roce.reserved_gids.start +
+		  dev->roce.reserved_gids.count - 1;
+	int index = 0;
+
+	index = ida_alloc_range(&dev->roce.reserved_gids.ida,
+				dev->roce.reserved_gids.start, end,
+				GFP_KERNEL);
+	if (index < 0)
+		return index;
+
+	mlx5_core_dbg(dev, "Allocating reserved GID %u\n", index);
+	*gid_index = index;
+	return 0;
+}
+
+void mlx5_core_reserved_gid_free(struct mlx5_core_dev *dev, int gid_index)
+{
+	mlx5_core_dbg(dev, "Freeing reserved GID %u\n", gid_index);
+	ida_free(&dev->roce.reserved_gids.ida, gid_index);
+}
+
+unsigned int mlx5_core_reserved_gids_count(struct mlx5_core_dev *dev)
+{
+	return dev->roce.reserved_gids.count;
+}
+EXPORT_SYMBOL_GPL(mlx5_core_reserved_gids_count);
+
+int mlx5_core_roce_gid_set(struct mlx5_core_dev *dev, unsigned int index,
+			   u8 roce_version, u8 roce_l3_type, const u8 *gid,
+			   const u8 *mac, bool vlan, u16 vlan_id, u8 port_num)
+{
+#define MLX5_SET_RA(p, f, v) MLX5_SET(roce_addr_layout, p, f, v)
+	u32 in[MLX5_ST_SZ_DW(set_roce_address_in)] = {};
+	void *in_addr = MLX5_ADDR_OF(set_roce_address_in, in, roce_address);
+	char *addr_l3_addr = MLX5_ADDR_OF(roce_addr_layout, in_addr,
+					  source_l3_address);
+	void *addr_mac = MLX5_ADDR_OF(roce_addr_layout, in_addr,
+				      source_mac_47_32);
+	int gidsz = MLX5_FLD_SZ_BYTES(roce_addr_layout, source_l3_address);
+
+	if (MLX5_CAP_GEN(dev, port_type) != MLX5_CAP_PORT_TYPE_ETH)
+		return -EINVAL;
+
+	if (gid) {
+		if (vlan) {
+			MLX5_SET_RA(in_addr, vlan_valid, 1);
+			MLX5_SET_RA(in_addr, vlan_id, vlan_id);
+		}
+
+		ether_addr_copy(addr_mac, mac);
+		memcpy(addr_l3_addr, gid, gidsz);
+	}
+	MLX5_SET_RA(in_addr, roce_version, roce_version);
+	MLX5_SET_RA(in_addr, roce_l3_type, roce_l3_type);
+
+	if (MLX5_CAP_GEN(dev, num_vhca_ports) > 0)
+		MLX5_SET(set_roce_address_in, in, vhca_port_num, port_num);
+
+	MLX5_SET(set_roce_address_in, in, roce_address_index, index);
+	MLX5_SET(set_roce_address_in, in, opcode, MLX5_CMD_OP_SET_ROCE_ADDRESS);
+	return mlx5_cmd_exec_in(dev, set_roce_address, in);
+}
+EXPORT_SYMBOL(mlx5_core_roce_gid_set);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/hv.c b/drivers/net/ethernet/mellanox/mlx5/core/lib/hv.c
new file mode 100644
index 000000000..583dc7e2a
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/hv.c
@@ -0,0 +1,64 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+// Copyright (c) 2018 Mellanox Technologies
+
+#include <linux/hyperv.h>
+#include "mlx5_core.h"
+#include "lib/hv.h"
+
+static int mlx5_hv_config_common(struct mlx5_core_dev *dev, void *buf, int len,
+				 int offset, bool read)
+{
+	int rc = -EOPNOTSUPP;
+	int bytes_returned;
+	int block_id;
+
+	if (offset % HV_CONFIG_BLOCK_SIZE_MAX || len != HV_CONFIG_BLOCK_SIZE_MAX)
+		return -EINVAL;
+
+	block_id = offset / HV_CONFIG_BLOCK_SIZE_MAX;
+
+	rc = read ?
+	     hyperv_read_cfg_blk(dev->pdev, buf,
+				 HV_CONFIG_BLOCK_SIZE_MAX, block_id,
+				 &bytes_returned) :
+	     hyperv_write_cfg_blk(dev->pdev, buf,
+				  HV_CONFIG_BLOCK_SIZE_MAX, block_id);
+
+	/* Make sure len bytes were read successfully  */
+	if (read && !rc && len != bytes_returned)
+		rc = -EIO;
+
+	if (rc) {
+		mlx5_core_err(dev, "Failed to %s hv config, err = %d, len = %d, offset = %d\n",
+			      read ? "read" : "write", rc, len,
+			      offset);
+		return rc;
+	}
+
+	return 0;
+}
+
+int mlx5_hv_read_config(struct mlx5_core_dev *dev, void *buf, int len,
+			int offset)
+{
+	return mlx5_hv_config_common(dev, buf, len, offset, true);
+}
+
+int mlx5_hv_write_config(struct mlx5_core_dev *dev, void *buf, int len,
+			 int offset)
+{
+	return mlx5_hv_config_common(dev, buf, len, offset, false);
+}
+
+int mlx5_hv_register_invalidate(struct mlx5_core_dev *dev, void *context,
+				void (*block_invalidate)(void *context,
+							 u64 block_mask))
+{
+	return hyperv_reg_block_invalidate(dev->pdev, context,
+					   block_invalidate);
+}
+
+void mlx5_hv_unregister_invalidate(struct mlx5_core_dev *dev)
+{
+	hyperv_reg_block_invalidate(dev->pdev, NULL, NULL);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/hv.h b/drivers/net/ethernet/mellanox/mlx5/core/lib/hv.h
new file mode 100644
index 000000000..f9a45573f
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/hv.h
@@ -0,0 +1,22 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2019 Mellanox Technologies. */
+
+#ifndef __LIB_HV_H__
+#define __LIB_HV_H__
+
+#if IS_ENABLED(CONFIG_PCI_HYPERV_INTERFACE)
+
+#include <linux/hyperv.h>
+#include <linux/mlx5/driver.h>
+
+int mlx5_hv_read_config(struct mlx5_core_dev *dev, void *buf, int len,
+			int offset);
+int mlx5_hv_write_config(struct mlx5_core_dev *dev, void *buf, int len,
+			 int offset);
+int mlx5_hv_register_invalidate(struct mlx5_core_dev *dev, void *context,
+				void (*block_invalidate)(void *context,
+							 u64 block_mask));
+void mlx5_hv_unregister_invalidate(struct mlx5_core_dev *dev);
+#endif
+
+#endif /* __LIB_HV_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/hv_vhca.c b/drivers/net/ethernet/mellanox/mlx5/core/lib/hv_vhca.c
new file mode 100644
index 000000000..4047629a8
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/hv_vhca.c
@@ -0,0 +1,371 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+// Copyright (c) 2018 Mellanox Technologies
+
+#include <linux/hyperv.h>
+#include "mlx5_core.h"
+#include "lib/hv.h"
+#include "lib/hv_vhca.h"
+
+struct mlx5_hv_vhca {
+	struct mlx5_core_dev       *dev;
+	struct workqueue_struct    *work_queue;
+	struct mlx5_hv_vhca_agent  *agents[MLX5_HV_VHCA_AGENT_MAX];
+	struct mutex                agents_lock; /* Protect agents array */
+};
+
+struct mlx5_hv_vhca_work {
+	struct work_struct     invalidate_work;
+	struct mlx5_hv_vhca   *hv_vhca;
+	u64                    block_mask;
+};
+
+struct mlx5_hv_vhca_data_block {
+	u16     sequence;
+	u16     offset;
+	u8      reserved[4];
+	u64     data[15];
+};
+
+struct mlx5_hv_vhca_agent {
+	enum mlx5_hv_vhca_agent_type	 type;
+	struct mlx5_hv_vhca		*hv_vhca;
+	void				*priv;
+	u16                              seq;
+	void (*control)(struct mlx5_hv_vhca_agent *agent,
+			struct mlx5_hv_vhca_control_block *block);
+	void (*invalidate)(struct mlx5_hv_vhca_agent *agent,
+			   u64 block_mask);
+	void (*cleanup)(struct mlx5_hv_vhca_agent *agent);
+};
+
+struct mlx5_hv_vhca *mlx5_hv_vhca_create(struct mlx5_core_dev *dev)
+{
+	struct mlx5_hv_vhca *hv_vhca = NULL;
+
+	hv_vhca = kzalloc(sizeof(*hv_vhca), GFP_KERNEL);
+	if (!hv_vhca)
+		return ERR_PTR(-ENOMEM);
+
+	hv_vhca->work_queue = create_singlethread_workqueue("mlx5_hv_vhca");
+	if (!hv_vhca->work_queue) {
+		kfree(hv_vhca);
+		return ERR_PTR(-ENOMEM);
+	}
+
+	hv_vhca->dev = dev;
+	mutex_init(&hv_vhca->agents_lock);
+
+	return hv_vhca;
+}
+
+void mlx5_hv_vhca_destroy(struct mlx5_hv_vhca *hv_vhca)
+{
+	if (IS_ERR_OR_NULL(hv_vhca))
+		return;
+
+	destroy_workqueue(hv_vhca->work_queue);
+	kfree(hv_vhca);
+}
+
+static void mlx5_hv_vhca_invalidate_work(struct work_struct *work)
+{
+	struct mlx5_hv_vhca_work *hwork;
+	struct mlx5_hv_vhca *hv_vhca;
+	int i;
+
+	hwork = container_of(work, struct mlx5_hv_vhca_work, invalidate_work);
+	hv_vhca = hwork->hv_vhca;
+
+	mutex_lock(&hv_vhca->agents_lock);
+	for (i = 0; i < MLX5_HV_VHCA_AGENT_MAX; i++) {
+		struct mlx5_hv_vhca_agent *agent = hv_vhca->agents[i];
+
+		if (!agent || !agent->invalidate)
+			continue;
+
+		if (!(BIT(agent->type) & hwork->block_mask))
+			continue;
+
+		agent->invalidate(agent, hwork->block_mask);
+	}
+	mutex_unlock(&hv_vhca->agents_lock);
+
+	kfree(hwork);
+}
+
+void mlx5_hv_vhca_invalidate(void *context, u64 block_mask)
+{
+	struct mlx5_hv_vhca *hv_vhca = (struct mlx5_hv_vhca *)context;
+	struct mlx5_hv_vhca_work *work;
+
+	work = kzalloc(sizeof(*work), GFP_ATOMIC);
+	if (!work)
+		return;
+
+	INIT_WORK(&work->invalidate_work, mlx5_hv_vhca_invalidate_work);
+	work->hv_vhca    = hv_vhca;
+	work->block_mask = block_mask;
+
+	queue_work(hv_vhca->work_queue, &work->invalidate_work);
+}
+
+#define AGENT_MASK(type) (type ? BIT(type - 1) : 0 /* control */)
+
+static void mlx5_hv_vhca_agents_control(struct mlx5_hv_vhca *hv_vhca,
+					struct mlx5_hv_vhca_control_block *block)
+{
+	int i;
+
+	for (i = 0; i < MLX5_HV_VHCA_AGENT_MAX; i++) {
+		struct mlx5_hv_vhca_agent *agent = hv_vhca->agents[i];
+
+		if (!agent || !agent->control)
+			continue;
+
+		if (!(AGENT_MASK(agent->type) & block->control))
+			continue;
+
+		agent->control(agent, block);
+	}
+}
+
+static void mlx5_hv_vhca_capabilities(struct mlx5_hv_vhca *hv_vhca,
+				      u32 *capabilities)
+{
+	int i;
+
+	for (i = 0; i < MLX5_HV_VHCA_AGENT_MAX; i++) {
+		struct mlx5_hv_vhca_agent *agent = hv_vhca->agents[i];
+
+		if (agent)
+			*capabilities |= AGENT_MASK(agent->type);
+	}
+}
+
+static void
+mlx5_hv_vhca_control_agent_invalidate(struct mlx5_hv_vhca_agent *agent,
+				      u64 block_mask)
+{
+	struct mlx5_hv_vhca *hv_vhca = agent->hv_vhca;
+	struct mlx5_core_dev *dev = hv_vhca->dev;
+	struct mlx5_hv_vhca_control_block *block;
+	u32 capabilities = 0;
+	int err;
+
+	block = kzalloc(sizeof(*block), GFP_KERNEL);
+	if (!block)
+		return;
+
+	err = mlx5_hv_read_config(dev, block, sizeof(*block), 0);
+	if (err)
+		goto free_block;
+
+	mlx5_hv_vhca_capabilities(hv_vhca, &capabilities);
+
+	/* In case no capabilities, send empty block in return */
+	if (!capabilities) {
+		memset(block, 0, sizeof(*block));
+		goto write;
+	}
+
+	if (block->capabilities != capabilities)
+		block->capabilities = capabilities;
+
+	if (block->control & ~capabilities)
+		goto free_block;
+
+	mlx5_hv_vhca_agents_control(hv_vhca, block);
+	block->command_ack = block->command;
+
+write:
+	mlx5_hv_write_config(dev, block, sizeof(*block), 0);
+
+free_block:
+	kfree(block);
+}
+
+static struct mlx5_hv_vhca_agent *
+mlx5_hv_vhca_control_agent_create(struct mlx5_hv_vhca *hv_vhca)
+{
+	return mlx5_hv_vhca_agent_create(hv_vhca, MLX5_HV_VHCA_AGENT_CONTROL,
+					 NULL,
+					 mlx5_hv_vhca_control_agent_invalidate,
+					 NULL, NULL);
+}
+
+static void mlx5_hv_vhca_control_agent_destroy(struct mlx5_hv_vhca_agent *agent)
+{
+	mlx5_hv_vhca_agent_destroy(agent);
+}
+
+int mlx5_hv_vhca_init(struct mlx5_hv_vhca *hv_vhca)
+{
+	struct mlx5_hv_vhca_agent *agent;
+	int err;
+
+	if (IS_ERR_OR_NULL(hv_vhca))
+		return IS_ERR_OR_NULL(hv_vhca);
+
+	err = mlx5_hv_register_invalidate(hv_vhca->dev, hv_vhca,
+					  mlx5_hv_vhca_invalidate);
+	if (err)
+		return err;
+
+	agent = mlx5_hv_vhca_control_agent_create(hv_vhca);
+	if (IS_ERR_OR_NULL(agent)) {
+		mlx5_hv_unregister_invalidate(hv_vhca->dev);
+		return IS_ERR_OR_NULL(agent);
+	}
+
+	hv_vhca->agents[MLX5_HV_VHCA_AGENT_CONTROL] = agent;
+
+	return 0;
+}
+
+void mlx5_hv_vhca_cleanup(struct mlx5_hv_vhca *hv_vhca)
+{
+	struct mlx5_hv_vhca_agent *agent;
+	int i;
+
+	if (IS_ERR_OR_NULL(hv_vhca))
+		return;
+
+	agent = hv_vhca->agents[MLX5_HV_VHCA_AGENT_CONTROL];
+	if (agent)
+		mlx5_hv_vhca_control_agent_destroy(agent);
+
+	mutex_lock(&hv_vhca->agents_lock);
+	for (i = 0; i < MLX5_HV_VHCA_AGENT_MAX; i++)
+		WARN_ON(hv_vhca->agents[i]);
+
+	mutex_unlock(&hv_vhca->agents_lock);
+
+	mlx5_hv_unregister_invalidate(hv_vhca->dev);
+}
+
+static void mlx5_hv_vhca_agents_update(struct mlx5_hv_vhca *hv_vhca)
+{
+	mlx5_hv_vhca_invalidate(hv_vhca, BIT(MLX5_HV_VHCA_AGENT_CONTROL));
+}
+
+struct mlx5_hv_vhca_agent *
+mlx5_hv_vhca_agent_create(struct mlx5_hv_vhca *hv_vhca,
+			  enum mlx5_hv_vhca_agent_type type,
+			  void (*control)(struct mlx5_hv_vhca_agent*,
+					  struct mlx5_hv_vhca_control_block *block),
+			  void (*invalidate)(struct mlx5_hv_vhca_agent*,
+					     u64 block_mask),
+			  void (*cleaup)(struct mlx5_hv_vhca_agent *agent),
+			  void *priv)
+{
+	struct mlx5_hv_vhca_agent *agent;
+
+	if (IS_ERR_OR_NULL(hv_vhca))
+		return ERR_PTR(-ENOMEM);
+
+	if (type >= MLX5_HV_VHCA_AGENT_MAX)
+		return ERR_PTR(-EINVAL);
+
+	mutex_lock(&hv_vhca->agents_lock);
+	if (hv_vhca->agents[type]) {
+		mutex_unlock(&hv_vhca->agents_lock);
+		return ERR_PTR(-EINVAL);
+	}
+	mutex_unlock(&hv_vhca->agents_lock);
+
+	agent = kzalloc(sizeof(*agent), GFP_KERNEL);
+	if (!agent)
+		return ERR_PTR(-ENOMEM);
+
+	agent->type      = type;
+	agent->hv_vhca   = hv_vhca;
+	agent->priv      = priv;
+	agent->control   = control;
+	agent->invalidate = invalidate;
+	agent->cleanup   = cleaup;
+
+	mutex_lock(&hv_vhca->agents_lock);
+	hv_vhca->agents[type] = agent;
+	mutex_unlock(&hv_vhca->agents_lock);
+
+	mlx5_hv_vhca_agents_update(hv_vhca);
+
+	return agent;
+}
+
+void mlx5_hv_vhca_agent_destroy(struct mlx5_hv_vhca_agent *agent)
+{
+	struct mlx5_hv_vhca *hv_vhca = agent->hv_vhca;
+
+	mutex_lock(&hv_vhca->agents_lock);
+
+	if (WARN_ON(agent != hv_vhca->agents[agent->type])) {
+		mutex_unlock(&hv_vhca->agents_lock);
+		return;
+	}
+
+	hv_vhca->agents[agent->type] = NULL;
+	mutex_unlock(&hv_vhca->agents_lock);
+
+	if (agent->cleanup)
+		agent->cleanup(agent);
+
+	kfree(agent);
+
+	mlx5_hv_vhca_agents_update(hv_vhca);
+}
+
+static int mlx5_hv_vhca_data_block_prepare(struct mlx5_hv_vhca_agent *agent,
+					   struct mlx5_hv_vhca_data_block *data_block,
+					   void *src, int len, int *offset)
+{
+	int bytes = min_t(int, (int)sizeof(data_block->data), len);
+
+	data_block->sequence = agent->seq;
+	data_block->offset   = (*offset)++;
+	memcpy(data_block->data, src, bytes);
+
+	return bytes;
+}
+
+static void mlx5_hv_vhca_agent_seq_update(struct mlx5_hv_vhca_agent *agent)
+{
+	agent->seq++;
+}
+
+int mlx5_hv_vhca_agent_write(struct mlx5_hv_vhca_agent *agent,
+			     void *buf, int len)
+{
+	int offset = agent->type * HV_CONFIG_BLOCK_SIZE_MAX;
+	int block_offset = 0;
+	int total = 0;
+	int err;
+
+	while (len) {
+		struct mlx5_hv_vhca_data_block data_block = {0};
+		int bytes;
+
+		bytes = mlx5_hv_vhca_data_block_prepare(agent, &data_block,
+							buf + total,
+							len, &block_offset);
+		if (!bytes)
+			return -ENOMEM;
+
+		err = mlx5_hv_write_config(agent->hv_vhca->dev, &data_block,
+					   sizeof(data_block), offset);
+		if (err)
+			return err;
+
+		total += bytes;
+		len   -= bytes;
+	}
+
+	mlx5_hv_vhca_agent_seq_update(agent);
+
+	return 0;
+}
+
+void *mlx5_hv_vhca_agent_priv(struct mlx5_hv_vhca_agent *agent)
+{
+	return agent->priv;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/hv_vhca.h b/drivers/net/ethernet/mellanox/mlx5/core/lib/hv_vhca.h
new file mode 100644
index 000000000..f240ffe51
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/hv_vhca.h
@@ -0,0 +1,97 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2019 Mellanox Technologies. */
+
+#ifndef __LIB_HV_VHCA_H__
+#define __LIB_HV_VHCA_H__
+
+#include "en.h"
+#include "lib/hv.h"
+
+struct mlx5_hv_vhca_agent;
+struct mlx5_hv_vhca;
+struct mlx5_hv_vhca_control_block;
+
+enum mlx5_hv_vhca_agent_type {
+	MLX5_HV_VHCA_AGENT_CONTROL = 0,
+	MLX5_HV_VHCA_AGENT_STATS   = 1,
+	MLX5_HV_VHCA_AGENT_MAX = 32,
+};
+
+#if IS_ENABLED(CONFIG_PCI_HYPERV_INTERFACE)
+
+struct mlx5_hv_vhca_control_block {
+	u32     capabilities;
+	u32     control;
+	u16     command;
+	u16     command_ack;
+	u16     version;
+	u16     rings;
+	u32     reserved1[28];
+};
+
+struct mlx5_hv_vhca *mlx5_hv_vhca_create(struct mlx5_core_dev *dev);
+void mlx5_hv_vhca_destroy(struct mlx5_hv_vhca *hv_vhca);
+int mlx5_hv_vhca_init(struct mlx5_hv_vhca *hv_vhca);
+void mlx5_hv_vhca_cleanup(struct mlx5_hv_vhca *hv_vhca);
+void mlx5_hv_vhca_invalidate(void *context, u64 block_mask);
+
+struct mlx5_hv_vhca_agent *
+mlx5_hv_vhca_agent_create(struct mlx5_hv_vhca *hv_vhca,
+			  enum mlx5_hv_vhca_agent_type type,
+			  void (*control)(struct mlx5_hv_vhca_agent*,
+					  struct mlx5_hv_vhca_control_block *block),
+			  void (*invalidate)(struct mlx5_hv_vhca_agent*,
+					     u64 block_mask),
+			  void (*cleanup)(struct mlx5_hv_vhca_agent *agent),
+			  void *context);
+
+void mlx5_hv_vhca_agent_destroy(struct mlx5_hv_vhca_agent *agent);
+int mlx5_hv_vhca_agent_write(struct mlx5_hv_vhca_agent *agent,
+			     void *buf, int len);
+void *mlx5_hv_vhca_agent_priv(struct mlx5_hv_vhca_agent *agent);
+
+#else
+
+static inline struct mlx5_hv_vhca *
+mlx5_hv_vhca_create(struct mlx5_core_dev *dev)
+{
+	return NULL;
+}
+
+static inline void mlx5_hv_vhca_destroy(struct mlx5_hv_vhca *hv_vhca)
+{
+}
+
+static inline int mlx5_hv_vhca_init(struct mlx5_hv_vhca *hv_vhca)
+{
+	return 0;
+}
+
+static inline void mlx5_hv_vhca_cleanup(struct mlx5_hv_vhca *hv_vhca)
+{
+}
+
+static inline void mlx5_hv_vhca_invalidate(void *context,
+					   u64 block_mask)
+{
+}
+
+static inline struct mlx5_hv_vhca_agent *
+mlx5_hv_vhca_agent_create(struct mlx5_hv_vhca *hv_vhca,
+			  enum mlx5_hv_vhca_agent_type type,
+			  void (*control)(struct mlx5_hv_vhca_agent*,
+					  struct mlx5_hv_vhca_control_block *block),
+			  void (*invalidate)(struct mlx5_hv_vhca_agent*,
+					     u64 block_mask),
+			  void (*cleanup)(struct mlx5_hv_vhca_agent *agent),
+			  void *context)
+{
+	return NULL;
+}
+
+static inline void mlx5_hv_vhca_agent_destroy(struct mlx5_hv_vhca_agent *agent)
+{
+}
+#endif
+
+#endif /* __LIB_HV_VHCA_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/mlx5.h b/drivers/net/ethernet/mellanox/mlx5/core/lib/mlx5.h
new file mode 100644
index 000000000..032adb21a
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/mlx5.h
@@ -0,0 +1,108 @@
+/*
+ * Copyright (c) 2017, Mellanox Technologies, Ltd.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef __LIB_MLX5_H__
+#define __LIB_MLX5_H__
+
+#include "mlx5_core.h"
+
+void mlx5_init_reserved_gids(struct mlx5_core_dev *dev);
+void mlx5_cleanup_reserved_gids(struct mlx5_core_dev *dev);
+int  mlx5_core_reserve_gids(struct mlx5_core_dev *dev, unsigned int count);
+void mlx5_core_unreserve_gids(struct mlx5_core_dev *dev, unsigned int count);
+int  mlx5_core_reserved_gid_alloc(struct mlx5_core_dev *dev, int *gid_index);
+void mlx5_core_reserved_gid_free(struct mlx5_core_dev *dev, int gid_index);
+int mlx5_crdump_enable(struct mlx5_core_dev *dev);
+void mlx5_crdump_disable(struct mlx5_core_dev *dev);
+int mlx5_crdump_collect(struct mlx5_core_dev *dev, u32 *cr_data);
+
+/* TODO move to lib/events.h */
+
+#define PORT_MODULE_EVENT_MODULE_STATUS_MASK 0xF
+#define PORT_MODULE_EVENT_ERROR_TYPE_MASK    0xF
+
+enum port_module_event_status_type {
+	MLX5_MODULE_STATUS_PLUGGED   = 0x1,
+	MLX5_MODULE_STATUS_UNPLUGGED = 0x2,
+	MLX5_MODULE_STATUS_ERROR     = 0x3,
+	MLX5_MODULE_STATUS_DISABLED  = 0x4,
+	MLX5_MODULE_STATUS_NUM,
+};
+
+enum  port_module_event_error_type {
+	MLX5_MODULE_EVENT_ERROR_POWER_BUDGET_EXCEEDED    = 0x0,
+	MLX5_MODULE_EVENT_ERROR_LONG_RANGE_FOR_NON_MLNX  = 0x1,
+	MLX5_MODULE_EVENT_ERROR_BUS_STUCK                = 0x2,
+	MLX5_MODULE_EVENT_ERROR_NO_EEPROM_RETRY_TIMEOUT  = 0x3,
+	MLX5_MODULE_EVENT_ERROR_ENFORCE_PART_NUMBER_LIST = 0x4,
+	MLX5_MODULE_EVENT_ERROR_UNKNOWN_IDENTIFIER       = 0x5,
+	MLX5_MODULE_EVENT_ERROR_HIGH_TEMPERATURE         = 0x6,
+	MLX5_MODULE_EVENT_ERROR_BAD_CABLE                = 0x7,
+	MLX5_MODULE_EVENT_ERROR_PCIE_POWER_SLOT_EXCEEDED = 0xc,
+	MLX5_MODULE_EVENT_ERROR_NUM,
+};
+
+struct mlx5_pme_stats {
+	u64 status_counters[MLX5_MODULE_STATUS_NUM];
+	u64 error_counters[MLX5_MODULE_EVENT_ERROR_NUM];
+};
+
+void mlx5_get_pme_stats(struct mlx5_core_dev *dev, struct mlx5_pme_stats *stats);
+int mlx5_notifier_call_chain(struct mlx5_events *events, unsigned int event, void *data);
+
+/* Crypto */
+enum {
+	MLX5_ACCEL_OBJ_TLS_KEY = MLX5_GENERAL_OBJECT_TYPE_ENCRYPTION_KEY_TYPE_TLS,
+	MLX5_ACCEL_OBJ_IPSEC_KEY = MLX5_GENERAL_OBJECT_TYPE_ENCRYPTION_KEY_TYPE_IPSEC,
+	MLX5_ACCEL_OBJ_MACSEC_KEY = MLX5_GENERAL_OBJECT_TYPE_ENCRYPTION_KEY_TYPE_MACSEC,
+};
+
+int mlx5_create_encryption_key(struct mlx5_core_dev *mdev,
+			       void *key, u32 sz_bytes,
+			       u32 key_type, u32 *p_key_id);
+void mlx5_destroy_encryption_key(struct mlx5_core_dev *mdev, u32 key_id);
+
+static inline struct net *mlx5_core_net(struct mlx5_core_dev *dev)
+{
+	return devlink_net(priv_to_devlink(dev));
+}
+
+static inline void mlx5_uplink_netdev_set(struct mlx5_core_dev *mdev, struct net_device *netdev)
+{
+	mdev->mlx5e_res.uplink_netdev = netdev;
+}
+
+static inline struct net_device *mlx5_uplink_netdev_get(struct mlx5_core_dev *mdev)
+{
+	return mdev->mlx5e_res.uplink_netdev;
+}
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/mpfs.c b/drivers/net/ethernet/mellanox/mlx5/core/lib/mpfs.c
new file mode 100644
index 000000000..8ff16318e
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/mpfs.c
@@ -0,0 +1,211 @@
+/*
+ * Copyright (c) 2017, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/etherdevice.h>
+#include <linux/mlx5/driver.h>
+#include <linux/mlx5/mlx5_ifc.h>
+#include <linux/mlx5/mpfs.h>
+#include <linux/mlx5/eswitch.h>
+#include "mlx5_core.h"
+#include "lib/mpfs.h"
+
+/* HW L2 Table (MPFS) management */
+static int set_l2table_entry_cmd(struct mlx5_core_dev *dev, u32 index, u8 *mac)
+{
+	u32 in[MLX5_ST_SZ_DW(set_l2_table_entry_in)] = {};
+	u8 *in_mac_addr;
+
+	MLX5_SET(set_l2_table_entry_in, in, opcode, MLX5_CMD_OP_SET_L2_TABLE_ENTRY);
+	MLX5_SET(set_l2_table_entry_in, in, table_index, index);
+
+	in_mac_addr = MLX5_ADDR_OF(set_l2_table_entry_in, in, mac_address);
+	ether_addr_copy(&in_mac_addr[2], mac);
+
+	return mlx5_cmd_exec_in(dev, set_l2_table_entry, in);
+}
+
+static int del_l2table_entry_cmd(struct mlx5_core_dev *dev, u32 index)
+{
+	u32 in[MLX5_ST_SZ_DW(delete_l2_table_entry_in)] = {};
+
+	MLX5_SET(delete_l2_table_entry_in, in, opcode, MLX5_CMD_OP_DELETE_L2_TABLE_ENTRY);
+	MLX5_SET(delete_l2_table_entry_in, in, table_index, index);
+	return mlx5_cmd_exec_in(dev, delete_l2_table_entry, in);
+}
+
+/* UC L2 table hash node */
+struct l2table_node {
+	struct l2addr_node node;
+	u32                index; /* index in HW l2 table */
+	int                ref_count;
+};
+
+struct mlx5_mpfs {
+	struct hlist_head    hash[MLX5_L2_ADDR_HASH_SIZE];
+	struct mutex         lock; /* Synchronize l2 table access */
+	u32                  size;
+	unsigned long        *bitmap;
+};
+
+static int alloc_l2table_index(struct mlx5_mpfs *l2table, u32 *ix)
+{
+	int err = 0;
+
+	*ix = find_first_zero_bit(l2table->bitmap, l2table->size);
+	if (*ix >= l2table->size)
+		err = -ENOSPC;
+	else
+		__set_bit(*ix, l2table->bitmap);
+
+	return err;
+}
+
+static void free_l2table_index(struct mlx5_mpfs *l2table, u32 ix)
+{
+	__clear_bit(ix, l2table->bitmap);
+}
+
+int mlx5_mpfs_init(struct mlx5_core_dev *dev)
+{
+	int l2table_size = 1 << MLX5_CAP_GEN(dev, log_max_l2_table);
+	struct mlx5_mpfs *mpfs;
+
+	if (!MLX5_ESWITCH_MANAGER(dev))
+		return 0;
+
+	mpfs = kzalloc(sizeof(*mpfs), GFP_KERNEL);
+	if (!mpfs)
+		return -ENOMEM;
+
+	mutex_init(&mpfs->lock);
+	mpfs->size   = l2table_size;
+	mpfs->bitmap = bitmap_zalloc(l2table_size, GFP_KERNEL);
+	if (!mpfs->bitmap) {
+		kfree(mpfs);
+		return -ENOMEM;
+	}
+
+	dev->priv.mpfs = mpfs;
+	return 0;
+}
+
+void mlx5_mpfs_cleanup(struct mlx5_core_dev *dev)
+{
+	struct mlx5_mpfs *mpfs = dev->priv.mpfs;
+
+	if (!mpfs)
+		return;
+
+	WARN_ON(!hlist_empty(mpfs->hash));
+	bitmap_free(mpfs->bitmap);
+	kfree(mpfs);
+}
+
+int mlx5_mpfs_add_mac(struct mlx5_core_dev *dev, u8 *mac)
+{
+	struct mlx5_mpfs *mpfs = dev->priv.mpfs;
+	struct l2table_node *l2addr;
+	int err = 0;
+	u32 index;
+
+	if (!mpfs)
+		return 0;
+
+	mutex_lock(&mpfs->lock);
+
+	l2addr = l2addr_hash_find(mpfs->hash, mac, struct l2table_node);
+	if (l2addr) {
+		l2addr->ref_count++;
+		goto out;
+	}
+
+	err = alloc_l2table_index(mpfs, &index);
+	if (err)
+		goto out;
+
+	l2addr = l2addr_hash_add(mpfs->hash, mac, struct l2table_node, GFP_KERNEL);
+	if (!l2addr) {
+		err = -ENOMEM;
+		goto hash_add_err;
+	}
+
+	err = set_l2table_entry_cmd(dev, index, mac);
+	if (err)
+		goto set_table_entry_err;
+
+	l2addr->index = index;
+	l2addr->ref_count = 1;
+
+	mlx5_core_dbg(dev, "MPFS mac added %pM, index (%d)\n", mac, index);
+	goto out;
+
+set_table_entry_err:
+	l2addr_hash_del(l2addr);
+hash_add_err:
+	free_l2table_index(mpfs, index);
+out:
+	mutex_unlock(&mpfs->lock);
+	return err;
+}
+EXPORT_SYMBOL(mlx5_mpfs_add_mac);
+
+int mlx5_mpfs_del_mac(struct mlx5_core_dev *dev, u8 *mac)
+{
+	struct mlx5_mpfs *mpfs = dev->priv.mpfs;
+	struct l2table_node *l2addr;
+	int err = 0;
+	u32 index;
+
+	if (!mpfs)
+		return 0;
+
+	mutex_lock(&mpfs->lock);
+
+	l2addr = l2addr_hash_find(mpfs->hash, mac, struct l2table_node);
+	if (!l2addr) {
+		err = -ENOENT;
+		goto unlock;
+	}
+
+	if (--l2addr->ref_count > 0)
+		goto unlock;
+
+	index = l2addr->index;
+	del_l2table_entry_cmd(dev, index);
+	l2addr_hash_del(l2addr);
+	free_l2table_index(mpfs, index);
+	mlx5_core_dbg(dev, "MPFS mac deleted %pM, index (%d)\n", mac, index);
+unlock:
+	mutex_unlock(&mpfs->lock);
+	return err;
+}
+EXPORT_SYMBOL(mlx5_mpfs_del_mac);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/mpfs.h b/drivers/net/ethernet/mellanox/mlx5/core/lib/mpfs.h
new file mode 100644
index 000000000..4a293542a
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/mpfs.h
@@ -0,0 +1,92 @@
+/*
+ * Copyright (c) 2017, Mellanox Technologies, Ltd.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef __MLX5_MPFS_H__
+#define __MLX5_MPFS_H__
+
+#include <linux/if_ether.h>
+#include <linux/mlx5/device.h>
+
+/* L2 -mac address based- hash helpers */
+#define MLX5_L2_ADDR_HASH_SIZE (BIT(BITS_PER_BYTE))
+#define MLX5_L2_ADDR_HASH(addr) (addr[5])
+
+struct l2addr_node {
+	struct hlist_node hlist;
+	u8                addr[ETH_ALEN];
+};
+
+#define for_each_l2hash_node(hn, tmp, hash, i) \
+	for (i = 0; i < MLX5_L2_ADDR_HASH_SIZE; i++) \
+		hlist_for_each_entry_safe(hn, tmp, &(hash)[i], hlist)
+
+#define l2addr_hash_find(hash, mac, type) ({                \
+	int ix = MLX5_L2_ADDR_HASH(mac);                    \
+	bool found = false;                                 \
+	type *ptr = NULL;                                   \
+							    \
+	hlist_for_each_entry(ptr, &(hash)[ix], node.hlist)  \
+		if (ether_addr_equal(ptr->node.addr, mac)) {\
+			found = true;                       \
+			break;                              \
+		}                                           \
+	if (!found)                                         \
+		ptr = NULL;                                 \
+	ptr;                                                \
+})
+
+#define l2addr_hash_add(hash, mac, type, gfp) ({            \
+	int ix = MLX5_L2_ADDR_HASH(mac);                    \
+	type *ptr = NULL;                                   \
+							    \
+	ptr = kzalloc(sizeof(type), gfp);                   \
+	if (ptr) {                                          \
+		ether_addr_copy(ptr->node.addr, mac);       \
+		hlist_add_head(&ptr->node.hlist, &(hash)[ix]);\
+	}                                                   \
+	ptr;                                                \
+})
+
+#define l2addr_hash_del(ptr) ({                             \
+	hlist_del(&(ptr)->node.hlist);                      \
+	kfree(ptr);                                         \
+})
+
+#ifdef CONFIG_MLX5_MPFS
+int  mlx5_mpfs_init(struct mlx5_core_dev *dev);
+void mlx5_mpfs_cleanup(struct mlx5_core_dev *dev);
+#else /* #ifndef CONFIG_MLX5_MPFS */
+static inline int  mlx5_mpfs_init(struct mlx5_core_dev *dev) { return 0; }
+static inline void mlx5_mpfs_cleanup(struct mlx5_core_dev *dev) {}
+#endif
+
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/pci_vsc.c b/drivers/net/ethernet/mellanox/mlx5/core/lib/pci_vsc.c
new file mode 100644
index 000000000..6b774e0c2
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/pci_vsc.c
@@ -0,0 +1,316 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2019 Mellanox Technologies */
+
+#include <linux/pci.h>
+#include "mlx5_core.h"
+#include "pci_vsc.h"
+
+#define MLX5_EXTRACT_C(source, offset, size)	\
+	((((u32)(source)) >> (offset)) & MLX5_ONES32(size))
+#define MLX5_EXTRACT(src, start, len)		\
+	(((len) == 32) ? (src) : MLX5_EXTRACT_C(src, start, len))
+#define MLX5_ONES32(size)			\
+	((size) ? (0xffffffff >> (32 - (size))) : 0)
+#define MLX5_MASK32(offset, size)		\
+	(MLX5_ONES32(size) << (offset))
+#define MLX5_MERGE_C(rsrc1, rsrc2, start, len)  \
+	((((rsrc2) << (start)) & (MLX5_MASK32((start), (len)))) | \
+	((rsrc1) & (~MLX5_MASK32((start), (len)))))
+#define MLX5_MERGE(rsrc1, rsrc2, start, len)	\
+	(((len) == 32) ? (rsrc2) : MLX5_MERGE_C(rsrc1, rsrc2, start, len))
+#define vsc_read(dev, offset, val) \
+	pci_read_config_dword((dev)->pdev, (dev)->vsc_addr + (offset), (val))
+#define vsc_write(dev, offset, val) \
+	pci_write_config_dword((dev)->pdev, (dev)->vsc_addr + (offset), (val))
+#define VSC_MAX_RETRIES 2048
+
+enum {
+	VSC_CTRL_OFFSET = 0x4,
+	VSC_COUNTER_OFFSET = 0x8,
+	VSC_SEMAPHORE_OFFSET = 0xc,
+	VSC_ADDR_OFFSET = 0x10,
+	VSC_DATA_OFFSET = 0x14,
+
+	VSC_FLAG_BIT_OFFS = 31,
+	VSC_FLAG_BIT_LEN = 1,
+
+	VSC_SYND_BIT_OFFS = 30,
+	VSC_SYND_BIT_LEN = 1,
+
+	VSC_ADDR_BIT_OFFS = 0,
+	VSC_ADDR_BIT_LEN = 30,
+
+	VSC_SPACE_BIT_OFFS = 0,
+	VSC_SPACE_BIT_LEN = 16,
+
+	VSC_SIZE_VLD_BIT_OFFS = 28,
+	VSC_SIZE_VLD_BIT_LEN = 1,
+
+	VSC_STATUS_BIT_OFFS = 29,
+	VSC_STATUS_BIT_LEN = 3,
+};
+
+void mlx5_pci_vsc_init(struct mlx5_core_dev *dev)
+{
+	if (!mlx5_core_is_pf(dev))
+		return;
+
+	dev->vsc_addr = pci_find_capability(dev->pdev,
+					    PCI_CAP_ID_VNDR);
+	if (!dev->vsc_addr)
+		mlx5_core_warn(dev, "Failed to get valid vendor specific ID\n");
+}
+
+int mlx5_vsc_gw_lock(struct mlx5_core_dev *dev)
+{
+	u32 counter = 0;
+	int retries = 0;
+	u32 lock_val;
+	int ret;
+
+	pci_cfg_access_lock(dev->pdev);
+	do {
+		if (retries > VSC_MAX_RETRIES) {
+			ret = -EBUSY;
+			goto pci_unlock;
+		}
+
+		/* Check if semaphore is already locked */
+		ret = vsc_read(dev, VSC_SEMAPHORE_OFFSET, &lock_val);
+		if (ret)
+			goto pci_unlock;
+
+		if (lock_val) {
+			retries++;
+			usleep_range(1000, 2000);
+			continue;
+		}
+
+		/* Read and write counter value, if written value is
+		 * the same, semaphore was acquired successfully.
+		 */
+		ret = vsc_read(dev, VSC_COUNTER_OFFSET, &counter);
+		if (ret)
+			goto pci_unlock;
+
+		ret = vsc_write(dev, VSC_SEMAPHORE_OFFSET, counter);
+		if (ret)
+			goto pci_unlock;
+
+		ret = vsc_read(dev, VSC_SEMAPHORE_OFFSET, &lock_val);
+		if (ret)
+			goto pci_unlock;
+
+		retries++;
+	} while (counter != lock_val);
+
+	return 0;
+
+pci_unlock:
+	pci_cfg_access_unlock(dev->pdev);
+	return ret;
+}
+
+int mlx5_vsc_gw_unlock(struct mlx5_core_dev *dev)
+{
+	int ret;
+
+	ret = vsc_write(dev, VSC_SEMAPHORE_OFFSET, MLX5_VSC_UNLOCK);
+	pci_cfg_access_unlock(dev->pdev);
+	return ret;
+}
+
+int mlx5_vsc_gw_set_space(struct mlx5_core_dev *dev, u16 space,
+			  u32 *ret_space_size)
+{
+	int ret;
+	u32 val = 0;
+
+	if (!mlx5_vsc_accessible(dev))
+		return -EINVAL;
+
+	if (ret_space_size)
+		*ret_space_size = 0;
+
+	/* Get a unique val */
+	ret = vsc_read(dev, VSC_CTRL_OFFSET, &val);
+	if (ret)
+		goto out;
+
+	/* Try to modify the lock */
+	val = MLX5_MERGE(val, space, VSC_SPACE_BIT_OFFS, VSC_SPACE_BIT_LEN);
+	ret = vsc_write(dev, VSC_CTRL_OFFSET, val);
+	if (ret)
+		goto out;
+
+	/* Verify lock was modified */
+	ret = vsc_read(dev, VSC_CTRL_OFFSET, &val);
+	if (ret)
+		goto out;
+
+	if (MLX5_EXTRACT(val, VSC_STATUS_BIT_OFFS, VSC_STATUS_BIT_LEN) == 0)
+		return -EINVAL;
+
+	/* Get space max address if indicated by size valid bit */
+	if (ret_space_size &&
+	    MLX5_EXTRACT(val, VSC_SIZE_VLD_BIT_OFFS, VSC_SIZE_VLD_BIT_LEN)) {
+		ret = vsc_read(dev, VSC_ADDR_OFFSET, &val);
+		if (ret) {
+			mlx5_core_warn(dev, "Failed to get max space size\n");
+			goto out;
+		}
+		*ret_space_size = MLX5_EXTRACT(val, VSC_ADDR_BIT_OFFS,
+					       VSC_ADDR_BIT_LEN);
+	}
+	return 0;
+
+out:
+	return ret;
+}
+
+static int mlx5_vsc_wait_on_flag(struct mlx5_core_dev *dev, u8 expected_val)
+{
+	int retries = 0;
+	u32 flag;
+	int ret;
+
+	do {
+		if (retries > VSC_MAX_RETRIES)
+			return -EBUSY;
+
+		ret = vsc_read(dev, VSC_ADDR_OFFSET, &flag);
+		if (ret)
+			return ret;
+		flag = MLX5_EXTRACT(flag, VSC_FLAG_BIT_OFFS, VSC_FLAG_BIT_LEN);
+		retries++;
+
+		if ((retries & 0xf) == 0)
+			usleep_range(1000, 2000);
+
+	} while (flag != expected_val);
+
+	return 0;
+}
+
+static int mlx5_vsc_gw_write(struct mlx5_core_dev *dev, unsigned int address,
+			     u32 data)
+{
+	int ret;
+
+	if (MLX5_EXTRACT(address, VSC_SYND_BIT_OFFS,
+			 VSC_FLAG_BIT_LEN + VSC_SYND_BIT_LEN))
+		return -EINVAL;
+
+	/* Set flag to 0x1 */
+	address = MLX5_MERGE(address, 1, VSC_FLAG_BIT_OFFS, 1);
+	ret = vsc_write(dev, VSC_DATA_OFFSET, data);
+	if (ret)
+		goto out;
+
+	ret = vsc_write(dev, VSC_ADDR_OFFSET, address);
+	if (ret)
+		goto out;
+
+	/* Wait for the flag to be cleared */
+	ret = mlx5_vsc_wait_on_flag(dev, 0);
+
+out:
+	return ret;
+}
+
+static int mlx5_vsc_gw_read(struct mlx5_core_dev *dev, unsigned int address,
+			    u32 *data)
+{
+	int ret;
+
+	if (MLX5_EXTRACT(address, VSC_SYND_BIT_OFFS,
+			 VSC_FLAG_BIT_LEN + VSC_SYND_BIT_LEN))
+		return -EINVAL;
+
+	ret = vsc_write(dev, VSC_ADDR_OFFSET, address);
+	if (ret)
+		goto out;
+
+	ret = mlx5_vsc_wait_on_flag(dev, 1);
+	if (ret)
+		goto out;
+
+	ret = vsc_read(dev, VSC_DATA_OFFSET, data);
+out:
+	return ret;
+}
+
+static int mlx5_vsc_gw_read_fast(struct mlx5_core_dev *dev,
+				 unsigned int read_addr,
+				 unsigned int *next_read_addr,
+				 u32 *data)
+{
+	int ret;
+
+	ret = mlx5_vsc_gw_read(dev, read_addr, data);
+	if (ret)
+		goto out;
+
+	ret = vsc_read(dev, VSC_ADDR_OFFSET, next_read_addr);
+	if (ret)
+		goto out;
+
+	*next_read_addr = MLX5_EXTRACT(*next_read_addr, VSC_ADDR_BIT_OFFS,
+				       VSC_ADDR_BIT_LEN);
+
+	if (*next_read_addr <= read_addr)
+		ret = -EINVAL;
+out:
+	return ret;
+}
+
+int mlx5_vsc_gw_read_block_fast(struct mlx5_core_dev *dev, u32 *data,
+				int length)
+{
+	unsigned int next_read_addr = 0;
+	unsigned int read_addr = 0;
+
+	while (read_addr < length) {
+		if (mlx5_vsc_gw_read_fast(dev, read_addr, &next_read_addr,
+					  &data[(read_addr >> 2)]))
+			return read_addr;
+
+		read_addr = next_read_addr;
+	}
+	return length;
+}
+
+int mlx5_vsc_sem_set_space(struct mlx5_core_dev *dev, u16 space,
+			   enum mlx5_vsc_state state)
+{
+	u32 data, id = 0;
+	int ret;
+
+	ret = mlx5_vsc_gw_set_space(dev, MLX5_SEMAPHORE_SPACE_DOMAIN, NULL);
+	if (ret) {
+		mlx5_core_warn(dev, "Failed to set gw space %d\n", ret);
+		return ret;
+	}
+
+	if (state == MLX5_VSC_LOCK) {
+		/* Get a unique ID based on the counter */
+		ret = vsc_read(dev, VSC_COUNTER_OFFSET, &id);
+		if (ret)
+			return ret;
+	}
+
+	/* Try to modify lock */
+	ret = mlx5_vsc_gw_write(dev, space, id);
+	if (ret)
+		return ret;
+
+	/* Verify lock was modified */
+	ret = mlx5_vsc_gw_read(dev, space, &data);
+	if (ret)
+		return -EINVAL;
+
+	if (data != id)
+		return -EBUSY;
+
+	return 0;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/pci_vsc.h b/drivers/net/ethernet/mellanox/mlx5/core/lib/pci_vsc.h
new file mode 100644
index 000000000..64272a6d7
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/pci_vsc.h
@@ -0,0 +1,32 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2019 Mellanox Technologies */
+
+#ifndef __MLX5_PCI_VSC_H__
+#define __MLX5_PCI_VSC_H__
+
+enum mlx5_vsc_state {
+	MLX5_VSC_UNLOCK,
+	MLX5_VSC_LOCK,
+};
+
+enum {
+	MLX5_VSC_SPACE_SCAN_CRSPACE = 0x7,
+};
+
+void mlx5_pci_vsc_init(struct mlx5_core_dev *dev);
+int mlx5_vsc_gw_lock(struct mlx5_core_dev *dev);
+int mlx5_vsc_gw_unlock(struct mlx5_core_dev *dev);
+int mlx5_vsc_gw_set_space(struct mlx5_core_dev *dev, u16 space,
+			  u32 *ret_space_size);
+int mlx5_vsc_gw_read_block_fast(struct mlx5_core_dev *dev, u32 *data,
+				int length);
+
+static inline bool mlx5_vsc_accessible(struct mlx5_core_dev *dev)
+{
+	return !!dev->vsc_addr;
+}
+
+int mlx5_vsc_sem_set_space(struct mlx5_core_dev *dev, u16 space,
+			   enum mlx5_vsc_state state);
+
+#endif /* __MLX5_PCI_VSC_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/port_tun.c b/drivers/net/ethernet/mellanox/mlx5/core/lib/port_tun.c
new file mode 100644
index 000000000..4571c56ec
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/port_tun.c
@@ -0,0 +1,186 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2019 Mellanox Technologies. */
+
+#include <linux/mlx5/driver.h>
+#include <linux/mlx5/port.h>
+#include "mlx5_core.h"
+#include "lib/port_tun.h"
+
+struct mlx5_port_tun_entropy_flags {
+	bool force_supported, force_enabled;
+	bool calc_supported, calc_enabled;
+	bool gre_calc_supported, gre_calc_enabled;
+};
+
+static void mlx5_query_port_tun_entropy(struct mlx5_core_dev *mdev,
+					struct mlx5_port_tun_entropy_flags *entropy_flags)
+{
+	u32 out[MLX5_ST_SZ_DW(pcmr_reg)];
+	/* Default values for FW which do not support MLX5_REG_PCMR */
+	entropy_flags->force_supported = false;
+	entropy_flags->calc_supported = false;
+	entropy_flags->gre_calc_supported = false;
+	entropy_flags->force_enabled = false;
+	entropy_flags->calc_enabled = true;
+	entropy_flags->gre_calc_enabled = true;
+
+	if (!MLX5_CAP_GEN(mdev, ports_check))
+		return;
+
+	if (mlx5_query_ports_check(mdev, out, sizeof(out)))
+		return;
+
+	entropy_flags->force_supported = !!(MLX5_GET(pcmr_reg, out, entropy_force_cap));
+	entropy_flags->calc_supported = !!(MLX5_GET(pcmr_reg, out, entropy_calc_cap));
+	entropy_flags->gre_calc_supported = !!(MLX5_GET(pcmr_reg, out, entropy_gre_calc_cap));
+	entropy_flags->force_enabled = !!(MLX5_GET(pcmr_reg, out, entropy_force));
+	entropy_flags->calc_enabled = !!(MLX5_GET(pcmr_reg, out, entropy_calc));
+	entropy_flags->gre_calc_enabled = !!(MLX5_GET(pcmr_reg, out, entropy_gre_calc));
+}
+
+static int mlx5_set_port_tun_entropy_calc(struct mlx5_core_dev *mdev, u8 enable,
+					  u8 force)
+{
+	u32 in[MLX5_ST_SZ_DW(pcmr_reg)] = {0};
+	int err;
+
+	err = mlx5_query_ports_check(mdev, in, sizeof(in));
+	if (err)
+		return err;
+	MLX5_SET(pcmr_reg, in, local_port, 1);
+	MLX5_SET(pcmr_reg, in, entropy_force, force);
+	MLX5_SET(pcmr_reg, in, entropy_calc, enable);
+	return mlx5_set_ports_check(mdev, in, sizeof(in));
+}
+
+static int mlx5_set_port_gre_tun_entropy_calc(struct mlx5_core_dev *mdev,
+					      u8 enable, u8 force)
+{
+	u32 in[MLX5_ST_SZ_DW(pcmr_reg)] = {0};
+	int err;
+
+	err = mlx5_query_ports_check(mdev, in, sizeof(in));
+	if (err)
+		return err;
+	MLX5_SET(pcmr_reg, in, local_port, 1);
+	MLX5_SET(pcmr_reg, in, entropy_force, force);
+	MLX5_SET(pcmr_reg, in, entropy_gre_calc, enable);
+	return mlx5_set_ports_check(mdev, in, sizeof(in));
+}
+
+void mlx5_init_port_tun_entropy(struct mlx5_tun_entropy *tun_entropy,
+				struct mlx5_core_dev *mdev)
+{
+	struct mlx5_port_tun_entropy_flags entropy_flags;
+
+	tun_entropy->mdev = mdev;
+	mutex_init(&tun_entropy->lock);
+	mlx5_query_port_tun_entropy(mdev, &entropy_flags);
+	tun_entropy->num_enabling_entries = 0;
+	tun_entropy->num_disabling_entries = 0;
+	tun_entropy->enabled = entropy_flags.calc_supported ?
+			       entropy_flags.calc_enabled : true;
+}
+
+static int mlx5_set_entropy(struct mlx5_tun_entropy *tun_entropy,
+			    int reformat_type, bool enable)
+{
+	struct mlx5_port_tun_entropy_flags entropy_flags;
+	int err;
+
+	mlx5_query_port_tun_entropy(tun_entropy->mdev, &entropy_flags);
+	/* Tunnel entropy calculation may be controlled either on port basis
+	 * for all tunneling protocols or specifically for GRE protocol.
+	 * Prioritize GRE protocol control (if capable) over global port
+	 * configuration.
+	 */
+	if (entropy_flags.gre_calc_supported &&
+	    reformat_type == MLX5_REFORMAT_TYPE_L2_TO_NVGRE) {
+		if (!entropy_flags.force_supported)
+			return 0;
+		err = mlx5_set_port_gre_tun_entropy_calc(tun_entropy->mdev,
+							 enable, !enable);
+		if (err)
+			return err;
+	} else if (entropy_flags.calc_supported) {
+		/* Other applications may change the global FW entropy
+		 * calculations settings. Check that the current entropy value
+		 * is the negative of the updated value.
+		 */
+		if (entropy_flags.force_enabled &&
+		    enable == entropy_flags.calc_enabled) {
+			mlx5_core_warn(tun_entropy->mdev,
+				       "Unexpected entropy calc setting - expected %d",
+				       !entropy_flags.calc_enabled);
+			return -EOPNOTSUPP;
+		}
+		/* GRE requires disabling entropy calculation. if there are
+		 * enabling entries (i.e VXLAN) we cannot turn it off for them,
+		 * thus fail.
+		 */
+		if (tun_entropy->num_enabling_entries)
+			return -EOPNOTSUPP;
+		err = mlx5_set_port_tun_entropy_calc(tun_entropy->mdev, enable,
+						     entropy_flags.force_supported);
+		if (err)
+			return err;
+		tun_entropy->enabled = enable;
+		/* if we turn on the entropy we don't need to force it anymore */
+		if (entropy_flags.force_supported && enable) {
+			err = mlx5_set_port_tun_entropy_calc(tun_entropy->mdev, 1, 0);
+			if (err)
+				return err;
+		}
+	}
+
+	return 0;
+}
+
+/* the function manages the refcount for enabling/disabling tunnel types.
+ * the return value indicates if the inc is successful or not, depending on
+ * entropy capabilities and configuration.
+ */
+int mlx5_tun_entropy_refcount_inc(struct mlx5_tun_entropy *tun_entropy,
+				  int reformat_type)
+{
+	int err = -EOPNOTSUPP;
+
+	mutex_lock(&tun_entropy->lock);
+	if ((reformat_type == MLX5_REFORMAT_TYPE_L2_TO_VXLAN ||
+	     reformat_type == MLX5_REFORMAT_TYPE_L2_TO_L3_TUNNEL) &&
+	    tun_entropy->enabled) {
+		/* in case entropy calculation is enabled for all tunneling
+		 * types, it is ok for VXLAN, so approve.
+		 * otherwise keep the error default.
+		 */
+		tun_entropy->num_enabling_entries++;
+		err = 0;
+	} else if (reformat_type == MLX5_REFORMAT_TYPE_L2_TO_NVGRE) {
+		/* turn off the entropy only for the first GRE rule.
+		 * for the next rules the entropy was already disabled
+		 * successfully.
+		 */
+		if (tun_entropy->num_disabling_entries == 0)
+			err = mlx5_set_entropy(tun_entropy, reformat_type, 0);
+		else
+			err = 0;
+		if (!err)
+			tun_entropy->num_disabling_entries++;
+	}
+	mutex_unlock(&tun_entropy->lock);
+
+	return err;
+}
+
+void mlx5_tun_entropy_refcount_dec(struct mlx5_tun_entropy *tun_entropy,
+				   int reformat_type)
+{
+	mutex_lock(&tun_entropy->lock);
+	if (reformat_type == MLX5_REFORMAT_TYPE_L2_TO_VXLAN)
+		tun_entropy->num_enabling_entries--;
+	else if (reformat_type == MLX5_REFORMAT_TYPE_L2_TO_NVGRE &&
+		 --tun_entropy->num_disabling_entries == 0)
+		mlx5_set_entropy(tun_entropy, reformat_type, 1);
+	mutex_unlock(&tun_entropy->lock);
+}
+
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/port_tun.h b/drivers/net/ethernet/mellanox/mlx5/core/lib/port_tun.h
new file mode 100644
index 000000000..54c42a887
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/port_tun.h
@@ -0,0 +1,24 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2019 Mellanox Technologies. */
+
+#ifndef __MLX5_PORT_TUN_H__
+#define __MLX5_PORT_TUN_H__
+
+#include <linux/mlx5/driver.h>
+
+struct mlx5_tun_entropy {
+	struct mlx5_core_dev *mdev;
+	u32 num_enabling_entries;
+	u32 num_disabling_entries;
+	u8  enabled;
+	struct mutex lock;	/* lock the entropy fields */
+};
+
+void mlx5_init_port_tun_entropy(struct mlx5_tun_entropy *tun_entropy,
+				struct mlx5_core_dev *mdev);
+int mlx5_tun_entropy_refcount_inc(struct mlx5_tun_entropy *tun_entropy,
+				  int reformat_type);
+void mlx5_tun_entropy_refcount_dec(struct mlx5_tun_entropy *tun_entropy,
+				   int reformat_type);
+
+#endif /* __MLX5_PORT_TUN_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/sf.h b/drivers/net/ethernet/mellanox/mlx5/core/lib/sf.h
new file mode 100644
index 000000000..84e568386
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/sf.h
@@ -0,0 +1,45 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2021 Mellanox Technologies Ltd */
+
+#ifndef __LIB_MLX5_SF_H__
+#define __LIB_MLX5_SF_H__
+
+#include <linux/mlx5/driver.h>
+
+static inline u16 mlx5_sf_start_function_id(const struct mlx5_core_dev *dev)
+{
+	return MLX5_CAP_GEN(dev, sf_base_id);
+}
+
+#ifdef CONFIG_MLX5_SF
+
+static inline bool mlx5_sf_supported(const struct mlx5_core_dev *dev)
+{
+	return MLX5_CAP_GEN(dev, sf);
+}
+
+static inline u16 mlx5_sf_max_functions(const struct mlx5_core_dev *dev)
+{
+	if (!mlx5_sf_supported(dev))
+		return 0;
+	if (MLX5_CAP_GEN(dev, max_num_sf))
+		return MLX5_CAP_GEN(dev, max_num_sf);
+	else
+		return 1 << MLX5_CAP_GEN(dev, log_max_sf);
+}
+
+#else
+
+static inline bool mlx5_sf_supported(const struct mlx5_core_dev *dev)
+{
+	return false;
+}
+
+static inline u16 mlx5_sf_max_functions(const struct mlx5_core_dev *dev)
+{
+	return 0;
+}
+
+#endif
+
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/smfs.c b/drivers/net/ethernet/mellanox/mlx5/core/lib/smfs.c
new file mode 100644
index 000000000..9b8c051cc
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/smfs.c
@@ -0,0 +1,68 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. */
+
+#include <linux/kernel.h>
+#include <linux/mlx5/driver.h>
+
+#include "smfs.h"
+
+struct mlx5dr_matcher *
+mlx5_smfs_matcher_create(struct mlx5dr_table *table, u32 priority, struct mlx5_flow_spec *spec)
+{
+	struct mlx5dr_match_parameters matcher_mask = {};
+
+	matcher_mask.match_buf = (u64 *)&spec->match_criteria;
+	matcher_mask.match_sz = DR_SZ_MATCH_PARAM;
+
+	return mlx5dr_matcher_create(table, priority, spec->match_criteria_enable, &matcher_mask);
+}
+
+void
+mlx5_smfs_matcher_destroy(struct mlx5dr_matcher *matcher)
+{
+	mlx5dr_matcher_destroy(matcher);
+}
+
+struct mlx5dr_table *
+mlx5_smfs_table_get_from_fs_ft(struct mlx5_flow_table *ft)
+{
+	return mlx5dr_table_get_from_fs_ft(ft);
+}
+
+struct mlx5dr_action *
+mlx5_smfs_action_create_dest_table(struct mlx5dr_table *table)
+{
+	return mlx5dr_action_create_dest_table(table);
+}
+
+struct mlx5dr_action *
+mlx5_smfs_action_create_flow_counter(u32 counter_id)
+{
+	return mlx5dr_action_create_flow_counter(counter_id);
+}
+
+void
+mlx5_smfs_action_destroy(struct mlx5dr_action *action)
+{
+	mlx5dr_action_destroy(action);
+}
+
+struct mlx5dr_rule *
+mlx5_smfs_rule_create(struct mlx5dr_matcher *matcher, struct mlx5_flow_spec *spec,
+		      size_t num_actions, struct mlx5dr_action *actions[],
+		      u32 flow_source)
+{
+	struct mlx5dr_match_parameters value = {};
+
+	value.match_buf = (u64 *)spec->match_value;
+	value.match_sz = DR_SZ_MATCH_PARAM;
+
+	return mlx5dr_rule_create(matcher, &value, num_actions, actions, flow_source);
+}
+
+void
+mlx5_smfs_rule_destroy(struct mlx5dr_rule *rule)
+{
+	mlx5dr_rule_destroy(rule);
+}
+
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/smfs.h b/drivers/net/ethernet/mellanox/mlx5/core/lib/smfs.h
new file mode 100644
index 000000000..452d0df33
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/smfs.h
@@ -0,0 +1,36 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. */
+
+#ifndef __MLX5_LIB_SMFS_H__
+#define __MLX5_LIB_SMFS_H__
+
+#include "steering/mlx5dr.h"
+#include "steering/dr_types.h"
+
+struct mlx5dr_matcher *
+mlx5_smfs_matcher_create(struct mlx5dr_table *table, u32 priority, struct mlx5_flow_spec *spec);
+
+void
+mlx5_smfs_matcher_destroy(struct mlx5dr_matcher *matcher);
+
+struct mlx5dr_table *
+mlx5_smfs_table_get_from_fs_ft(struct mlx5_flow_table *ft);
+
+struct mlx5dr_action *
+mlx5_smfs_action_create_dest_table(struct mlx5dr_table *table);
+
+struct mlx5dr_action *
+mlx5_smfs_action_create_flow_counter(u32 counter_id);
+
+void
+mlx5_smfs_action_destroy(struct mlx5dr_action *action);
+
+struct mlx5dr_rule *
+mlx5_smfs_rule_create(struct mlx5dr_matcher *matcher, struct mlx5_flow_spec *spec,
+		      size_t num_actions, struct mlx5dr_action *actions[],
+		      u32 flow_source);
+
+void
+mlx5_smfs_rule_destroy(struct mlx5dr_rule *rule);
+
+#endif /* __MLX5_LIB_SMFS_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/tout.c b/drivers/net/ethernet/mellanox/mlx5/core/lib/tout.c
new file mode 100644
index 000000000..696e45e2b
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/tout.c
@@ -0,0 +1,159 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. All rights reserved. */
+
+#include <linux/mlx5/driver.h>
+#include "lib/tout.h"
+
+struct mlx5_timeouts {
+	u64 to[MAX_TIMEOUT_TYPES];
+};
+
+static const u32 tout_def_sw_val[MAX_TIMEOUT_TYPES] = {
+	[MLX5_TO_FW_PRE_INIT_TIMEOUT_MS] = 120000,
+	[MLX5_TO_FW_PRE_INIT_ON_RECOVERY_TIMEOUT_MS] = 7200000,
+	[MLX5_TO_FW_PRE_INIT_WARN_MESSAGE_INTERVAL_MS] = 20000,
+	[MLX5_TO_FW_PRE_INIT_WAIT_MS] = 2,
+	[MLX5_TO_FW_INIT_MS] = 2000,
+	[MLX5_TO_CMD_MS] = 60000,
+	[MLX5_TO_PCI_TOGGLE_MS] =  2000,
+	[MLX5_TO_HEALTH_POLL_INTERVAL_MS] =  2000,
+	[MLX5_TO_FULL_CRDUMP_MS] = 60000,
+	[MLX5_TO_FW_RESET_MS] = 60000,
+	[MLX5_TO_FLUSH_ON_ERROR_MS] = 2000,
+	[MLX5_TO_PCI_SYNC_UPDATE_MS] = 5000,
+	[MLX5_TO_TEARDOWN_MS] = 3000,
+	[MLX5_TO_FSM_REACTIVATE_MS] = 5000,
+	[MLX5_TO_RECLAIM_PAGES_MS] = 5000,
+	[MLX5_TO_RECLAIM_VFS_PAGES_MS] = 120000
+};
+
+static void tout_set(struct mlx5_core_dev *dev, u64 val, enum mlx5_timeouts_types type)
+{
+	dev->timeouts->to[type] = val;
+}
+
+int mlx5_tout_init(struct mlx5_core_dev *dev)
+{
+	int i;
+
+	dev->timeouts = kmalloc(sizeof(*dev->timeouts), GFP_KERNEL);
+	if (!dev->timeouts)
+		return -ENOMEM;
+
+	for (i = 0; i < MAX_TIMEOUT_TYPES; i++)
+		tout_set(dev, tout_def_sw_val[i], i);
+
+	return 0;
+}
+
+void mlx5_tout_cleanup(struct mlx5_core_dev *dev)
+{
+	kfree(dev->timeouts);
+}
+
+/* Time register consists of two fields to_multiplier(time out multiplier)
+ * and to_value(time out value). to_value is the quantity of the time units and
+ * to_multiplier is the type and should be one off these four values.
+ * 0x0: millisecond
+ * 0x1: seconds
+ * 0x2: minutes
+ * 0x3: hours
+ * this function converts the time stored in the two register fields into
+ * millisecond.
+ */
+static u64 tout_convert_reg_field_to_ms(u32 to_mul, u32 to_val)
+{
+	u64 msec = to_val;
+
+	to_mul &= 0x3;
+	/* convert hours/minutes/seconds to miliseconds */
+	if (to_mul)
+		msec *= 1000 * int_pow(60, to_mul - 1);
+
+	return msec;
+}
+
+static u64 tout_convert_iseg_to_ms(u32 iseg_to)
+{
+	return tout_convert_reg_field_to_ms(iseg_to >> 29, iseg_to & 0xfffff);
+}
+
+static bool tout_is_supported(struct mlx5_core_dev *dev)
+{
+	return !!ioread32be(&dev->iseg->cmd_q_init_to);
+}
+
+void mlx5_tout_query_iseg(struct mlx5_core_dev *dev)
+{
+	u32 to;
+
+	if (!tout_is_supported(dev))
+		return;
+
+	to = ioread32be(&dev->iseg->cmd_q_init_to);
+	tout_set(dev, tout_convert_iseg_to_ms(to), MLX5_TO_FW_INIT_MS);
+
+	to = ioread32be(&dev->iseg->cmd_exec_to);
+	tout_set(dev, tout_convert_iseg_to_ms(to), MLX5_TO_CMD_MS);
+}
+
+u64 _mlx5_tout_ms(struct mlx5_core_dev *dev, enum mlx5_timeouts_types type)
+{
+	return dev->timeouts->to[type];
+}
+
+#define MLX5_TIMEOUT_QUERY(fld, reg_out) \
+	({ \
+	struct mlx5_ifc_default_timeout_bits *time_field; \
+	u32 to_multi, to_value; \
+	u64 to_val_ms; \
+	\
+	time_field = MLX5_ADDR_OF(dtor_reg, reg_out, fld); \
+	to_multi = MLX5_GET(default_timeout, time_field, to_multiplier); \
+	to_value = MLX5_GET(default_timeout, time_field, to_value); \
+	to_val_ms = tout_convert_reg_field_to_ms(to_multi, to_value); \
+	to_val_ms; \
+	})
+
+#define MLX5_TIMEOUT_FILL(fld, reg_out, dev, to_type, to_extra) \
+	({ \
+	u64 fw_to = MLX5_TIMEOUT_QUERY(fld, reg_out); \
+	tout_set(dev, fw_to + (to_extra), to_type); \
+	fw_to; \
+	})
+
+static int tout_query_dtor(struct mlx5_core_dev *dev)
+{
+	u64 pcie_toggle_to_val, tear_down_to_val;
+	u32 out[MLX5_ST_SZ_DW(dtor_reg)] = {};
+	u32 in[MLX5_ST_SZ_DW(dtor_reg)] = {};
+	int err;
+
+	err = mlx5_core_access_reg(dev, in, sizeof(in), out, sizeof(out), MLX5_REG_DTOR, 0, 0);
+	if (err)
+		return err;
+
+	pcie_toggle_to_val = MLX5_TIMEOUT_FILL(pcie_toggle_to, out, dev, MLX5_TO_PCI_TOGGLE_MS, 0);
+	MLX5_TIMEOUT_FILL(fw_reset_to, out, dev, MLX5_TO_FW_RESET_MS, pcie_toggle_to_val);
+
+	tear_down_to_val = MLX5_TIMEOUT_FILL(tear_down_to, out, dev, MLX5_TO_TEARDOWN_MS, 0);
+	MLX5_TIMEOUT_FILL(pci_sync_update_to, out, dev, MLX5_TO_PCI_SYNC_UPDATE_MS,
+			  tear_down_to_val);
+
+	MLX5_TIMEOUT_FILL(health_poll_to, out, dev, MLX5_TO_HEALTH_POLL_INTERVAL_MS, 0);
+	MLX5_TIMEOUT_FILL(full_crdump_to, out, dev, MLX5_TO_FULL_CRDUMP_MS, 0);
+	MLX5_TIMEOUT_FILL(flush_on_err_to, out, dev, MLX5_TO_FLUSH_ON_ERROR_MS, 0);
+	MLX5_TIMEOUT_FILL(fsm_reactivate_to, out, dev, MLX5_TO_FSM_REACTIVATE_MS, 0);
+	MLX5_TIMEOUT_FILL(reclaim_pages_to, out, dev, MLX5_TO_RECLAIM_PAGES_MS, 0);
+	MLX5_TIMEOUT_FILL(reclaim_vfs_pages_to, out, dev, MLX5_TO_RECLAIM_VFS_PAGES_MS, 0);
+
+	return 0;
+}
+
+int mlx5_tout_query_dtor(struct mlx5_core_dev *dev)
+{
+	if (tout_is_supported(dev))
+		return tout_query_dtor(dev);
+
+	return 0;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/tout.h b/drivers/net/ethernet/mellanox/mlx5/core/lib/tout.h
new file mode 100644
index 000000000..bc9e9aeda
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/tout.h
@@ -0,0 +1,42 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. All rights reserved. */
+
+#ifndef MLX5_TIMEOUTS_H
+#define MLX5_TIMEOUTS_H
+
+enum mlx5_timeouts_types {
+	/* pre init timeouts (not read from FW) */
+	MLX5_TO_FW_PRE_INIT_TIMEOUT_MS,
+	MLX5_TO_FW_PRE_INIT_ON_RECOVERY_TIMEOUT_MS,
+	MLX5_TO_FW_PRE_INIT_WARN_MESSAGE_INTERVAL_MS,
+	MLX5_TO_FW_PRE_INIT_WAIT_MS,
+
+	/* init segment timeouts */
+	MLX5_TO_FW_INIT_MS,
+	MLX5_TO_CMD_MS,
+
+	/* DTOR timeouts */
+	MLX5_TO_PCI_TOGGLE_MS,
+	MLX5_TO_HEALTH_POLL_INTERVAL_MS,
+	MLX5_TO_FULL_CRDUMP_MS,
+	MLX5_TO_FW_RESET_MS,
+	MLX5_TO_FLUSH_ON_ERROR_MS,
+	MLX5_TO_PCI_SYNC_UPDATE_MS,
+	MLX5_TO_TEARDOWN_MS,
+	MLX5_TO_FSM_REACTIVATE_MS,
+	MLX5_TO_RECLAIM_PAGES_MS,
+	MLX5_TO_RECLAIM_VFS_PAGES_MS,
+
+	MAX_TIMEOUT_TYPES
+};
+
+struct mlx5_core_dev;
+int mlx5_tout_init(struct mlx5_core_dev *dev);
+void mlx5_tout_cleanup(struct mlx5_core_dev *dev);
+void mlx5_tout_query_iseg(struct mlx5_core_dev *dev);
+int mlx5_tout_query_dtor(struct mlx5_core_dev *dev);
+u64 _mlx5_tout_ms(struct mlx5_core_dev *dev, enum mlx5_timeouts_types type);
+
+#define mlx5_tout_ms(dev, type) _mlx5_tout_ms(dev, MLX5_TO_##type##_MS)
+
+# endif /* MLX5_TIMEOUTS_H */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/vxlan.c b/drivers/net/ethernet/mellanox/mlx5/core/lib/vxlan.c
new file mode 100644
index 000000000..d55e15c1f
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/vxlan.c
@@ -0,0 +1,196 @@
+/*
+ * Copyright (c) 2016, Mellanox Technologies, Ltd.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/kernel.h>
+#include <linux/refcount.h>
+#include <linux/mlx5/driver.h>
+#include <net/vxlan.h>
+#include "mlx5_core.h"
+#include "vxlan.h"
+
+struct mlx5_vxlan {
+	struct mlx5_core_dev		*mdev;
+	/* max_num_ports is usually 4, 16 buckets is more than enough */
+	DECLARE_HASHTABLE(htable, 4);
+	struct mutex                    sync_lock; /* sync add/del port HW operations */
+};
+
+struct mlx5_vxlan_port {
+	struct hlist_node hlist;
+	u16 udp_port;
+};
+
+static int mlx5_vxlan_core_add_port_cmd(struct mlx5_core_dev *mdev, u16 port)
+{
+	u32 in[MLX5_ST_SZ_DW(add_vxlan_udp_dport_in)] = {};
+
+	MLX5_SET(add_vxlan_udp_dport_in, in, opcode,
+		 MLX5_CMD_OP_ADD_VXLAN_UDP_DPORT);
+	MLX5_SET(add_vxlan_udp_dport_in, in, vxlan_udp_port, port);
+	return mlx5_cmd_exec_in(mdev, add_vxlan_udp_dport, in);
+}
+
+static int mlx5_vxlan_core_del_port_cmd(struct mlx5_core_dev *mdev, u16 port)
+{
+	u32 in[MLX5_ST_SZ_DW(delete_vxlan_udp_dport_in)] = {};
+
+	MLX5_SET(delete_vxlan_udp_dport_in, in, opcode,
+		 MLX5_CMD_OP_DELETE_VXLAN_UDP_DPORT);
+	MLX5_SET(delete_vxlan_udp_dport_in, in, vxlan_udp_port, port);
+	return mlx5_cmd_exec_in(mdev, delete_vxlan_udp_dport, in);
+}
+
+bool mlx5_vxlan_lookup_port(struct mlx5_vxlan *vxlan, u16 port)
+{
+	struct mlx5_vxlan_port *vxlanp;
+	bool found = false;
+
+	if (!mlx5_vxlan_allowed(vxlan))
+		return NULL;
+
+	rcu_read_lock();
+	hash_for_each_possible_rcu(vxlan->htable, vxlanp, hlist, port)
+		if (vxlanp->udp_port == port) {
+			found = true;
+			break;
+		}
+	rcu_read_unlock();
+
+	return found;
+}
+
+static struct mlx5_vxlan_port *vxlan_lookup_port(struct mlx5_vxlan *vxlan, u16 port)
+{
+	struct mlx5_vxlan_port *vxlanp;
+
+	hash_for_each_possible(vxlan->htable, vxlanp, hlist, port)
+		if (vxlanp->udp_port == port)
+			return vxlanp;
+	return NULL;
+}
+
+int mlx5_vxlan_add_port(struct mlx5_vxlan *vxlan, u16 port)
+{
+	struct mlx5_vxlan_port *vxlanp;
+	int ret;
+
+	vxlanp = kzalloc(sizeof(*vxlanp), GFP_KERNEL);
+	if (!vxlanp)
+		return -ENOMEM;
+	vxlanp->udp_port = port;
+
+	ret = mlx5_vxlan_core_add_port_cmd(vxlan->mdev, port);
+	if (ret) {
+		kfree(vxlanp);
+		return ret;
+	}
+
+	mutex_lock(&vxlan->sync_lock);
+	hash_add_rcu(vxlan->htable, &vxlanp->hlist, port);
+	mutex_unlock(&vxlan->sync_lock);
+
+	return 0;
+}
+
+int mlx5_vxlan_del_port(struct mlx5_vxlan *vxlan, u16 port)
+{
+	struct mlx5_vxlan_port *vxlanp;
+	int ret = 0;
+
+	mutex_lock(&vxlan->sync_lock);
+
+	vxlanp = vxlan_lookup_port(vxlan, port);
+	if (WARN_ON(!vxlanp)) {
+		ret = -ENOENT;
+		goto out_unlock;
+	}
+
+	hash_del_rcu(&vxlanp->hlist);
+	synchronize_rcu();
+	mlx5_vxlan_core_del_port_cmd(vxlan->mdev, port);
+	kfree(vxlanp);
+
+out_unlock:
+	mutex_unlock(&vxlan->sync_lock);
+	return ret;
+}
+
+struct mlx5_vxlan *mlx5_vxlan_create(struct mlx5_core_dev *mdev)
+{
+	struct mlx5_vxlan *vxlan;
+
+	if (!MLX5_CAP_ETH(mdev, tunnel_stateless_vxlan) || !mlx5_core_is_pf(mdev))
+		return ERR_PTR(-ENOTSUPP);
+
+	vxlan = kzalloc(sizeof(*vxlan), GFP_KERNEL);
+	if (!vxlan)
+		return ERR_PTR(-ENOMEM);
+
+	vxlan->mdev = mdev;
+	mutex_init(&vxlan->sync_lock);
+	hash_init(vxlan->htable);
+
+	/* Hardware adds 4789 (IANA_VXLAN_UDP_PORT) by default */
+	mlx5_vxlan_add_port(vxlan, IANA_VXLAN_UDP_PORT);
+
+	return vxlan;
+}
+
+void mlx5_vxlan_destroy(struct mlx5_vxlan *vxlan)
+{
+	if (!mlx5_vxlan_allowed(vxlan))
+		return;
+
+	mlx5_vxlan_del_port(vxlan, IANA_VXLAN_UDP_PORT);
+	WARN_ON(!hash_empty(vxlan->htable));
+
+	kfree(vxlan);
+}
+
+void mlx5_vxlan_reset_to_default(struct mlx5_vxlan *vxlan)
+{
+	struct mlx5_vxlan_port *vxlanp;
+	struct hlist_node *tmp;
+	int bkt;
+
+	if (!mlx5_vxlan_allowed(vxlan))
+		return;
+
+	hash_for_each_safe(vxlan->htable, bkt, tmp, vxlanp, hlist) {
+		/* Don't delete default UDP port added by the HW.
+		 * Remove only user configured ports
+		 */
+		if (vxlanp->udp_port == IANA_VXLAN_UDP_PORT)
+			continue;
+		mlx5_vxlan_del_port(vxlan, vxlanp->udp_port);
+	}
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/vxlan.h b/drivers/net/ethernet/mellanox/mlx5/core/lib/vxlan.h
new file mode 100644
index 000000000..34ef662da
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/vxlan.h
@@ -0,0 +1,70 @@
+/*
+ * Copyright (c) 2016, Mellanox Technologies, Ltd.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+#ifndef __MLX5_VXLAN_H__
+#define __MLX5_VXLAN_H__
+
+#include <linux/mlx5/driver.h>
+
+struct mlx5_vxlan;
+struct mlx5_vxlan_port;
+
+static inline u8 mlx5_vxlan_max_udp_ports(struct mlx5_core_dev *mdev)
+{
+	return MLX5_CAP_ETH(mdev, max_vxlan_udp_ports) ?: 4;
+}
+
+static inline bool mlx5_vxlan_allowed(struct mlx5_vxlan *vxlan)
+{
+	/* not allowed reason is encoded in vxlan pointer as error,
+	 * on mlx5_vxlan_create
+	 */
+	return !IS_ERR_OR_NULL(vxlan);
+}
+
+#if IS_ENABLED(CONFIG_VXLAN)
+struct mlx5_vxlan *mlx5_vxlan_create(struct mlx5_core_dev *mdev);
+void mlx5_vxlan_destroy(struct mlx5_vxlan *vxlan);
+int mlx5_vxlan_add_port(struct mlx5_vxlan *vxlan, u16 port);
+int mlx5_vxlan_del_port(struct mlx5_vxlan *vxlan, u16 port);
+bool mlx5_vxlan_lookup_port(struct mlx5_vxlan *vxlan, u16 port);
+void mlx5_vxlan_reset_to_default(struct mlx5_vxlan *vxlan);
+#else
+static inline struct mlx5_vxlan*
+mlx5_vxlan_create(struct mlx5_core_dev *mdev) { return ERR_PTR(-EOPNOTSUPP); }
+static inline void mlx5_vxlan_destroy(struct mlx5_vxlan *vxlan) { return; }
+static inline int mlx5_vxlan_add_port(struct mlx5_vxlan *vxlan, u16 port) { return -EOPNOTSUPP; }
+static inline int mlx5_vxlan_del_port(struct mlx5_vxlan *vxlan, u16 port) { return -EOPNOTSUPP; }
+static inline bool mlx5_vxlan_lookup_port(struct mlx5_vxlan *vxlan, u16 port) { return false; }
+static inline void mlx5_vxlan_reset_to_default(struct mlx5_vxlan *vxlan) { return; }
+#endif
+
+#endif /* __MLX5_VXLAN_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/main.c b/drivers/net/ethernet/mellanox/mlx5/core/main.c
new file mode 100644
index 000000000..6ab0642e9
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/main.c
@@ -0,0 +1,2130 @@
+/*
+ * Copyright (c) 2013-2015, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/highmem.h>
+#include <linux/module.h>
+#include <linux/init.h>
+#include <linux/errno.h>
+#include <linux/pci.h>
+#include <linux/dma-mapping.h>
+#include <linux/slab.h>
+#include <linux/io-mapping.h>
+#include <linux/interrupt.h>
+#include <linux/delay.h>
+#include <linux/mlx5/driver.h>
+#include <linux/mlx5/cq.h>
+#include <linux/mlx5/qp.h>
+#include <linux/debugfs.h>
+#include <linux/kmod.h>
+#include <linux/mlx5/mlx5_ifc.h>
+#include <linux/mlx5/vport.h>
+#ifdef CONFIG_RFS_ACCEL
+#include <linux/cpu_rmap.h>
+#endif
+#include <linux/version.h>
+#include <net/devlink.h>
+#include "mlx5_core.h"
+#include "lib/eq.h"
+#include "fs_core.h"
+#include "lib/mpfs.h"
+#include "eswitch.h"
+#include "devlink.h"
+#include "fw_reset.h"
+#include "lib/mlx5.h"
+#include "lib/tout.h"
+#include "fpga/core.h"
+#include "en_accel/ipsec.h"
+#include "lib/clock.h"
+#include "lib/vxlan.h"
+#include "lib/geneve.h"
+#include "lib/devcom.h"
+#include "lib/pci_vsc.h"
+#include "diag/fw_tracer.h"
+#include "ecpf.h"
+#include "lib/hv_vhca.h"
+#include "diag/rsc_dump.h"
+#include "sf/vhca_event.h"
+#include "sf/dev/dev.h"
+#include "sf/sf.h"
+#include "mlx5_irq.h"
+
+MODULE_AUTHOR("Eli Cohen <eli@mellanox.com>");
+MODULE_DESCRIPTION("Mellanox 5th generation network adapters (ConnectX series) core driver");
+MODULE_LICENSE("Dual BSD/GPL");
+
+unsigned int mlx5_core_debug_mask;
+module_param_named(debug_mask, mlx5_core_debug_mask, uint, 0644);
+MODULE_PARM_DESC(debug_mask, "debug mask: 1 = dump cmd data, 2 = dump cmd exec time, 3 = both. Default=0");
+
+static unsigned int prof_sel = MLX5_DEFAULT_PROF;
+module_param_named(prof_sel, prof_sel, uint, 0444);
+MODULE_PARM_DESC(prof_sel, "profile selector. Valid range 0 - 2");
+
+static u32 sw_owner_id[4];
+#define MAX_SW_VHCA_ID (BIT(__mlx5_bit_sz(cmd_hca_cap_2, sw_vhca_id)) - 1)
+static DEFINE_IDA(sw_vhca_ida);
+
+enum {
+	MLX5_ATOMIC_REQ_MODE_BE = 0x0,
+	MLX5_ATOMIC_REQ_MODE_HOST_ENDIANNESS = 0x1,
+};
+
+#define LOG_MAX_SUPPORTED_QPS 0xff
+
+static struct mlx5_profile profile[] = {
+	[0] = {
+		.mask           = 0,
+	},
+	[1] = {
+		.mask		= MLX5_PROF_MASK_QP_SIZE,
+		.log_max_qp	= 12,
+	},
+	[2] = {
+		.mask		= MLX5_PROF_MASK_QP_SIZE |
+				  MLX5_PROF_MASK_MR_CACHE,
+		.log_max_qp	= LOG_MAX_SUPPORTED_QPS,
+		.mr_cache[0]	= {
+			.size	= 500,
+			.limit	= 250
+		},
+		.mr_cache[1]	= {
+			.size	= 500,
+			.limit	= 250
+		},
+		.mr_cache[2]	= {
+			.size	= 500,
+			.limit	= 250
+		},
+		.mr_cache[3]	= {
+			.size	= 500,
+			.limit	= 250
+		},
+		.mr_cache[4]	= {
+			.size	= 500,
+			.limit	= 250
+		},
+		.mr_cache[5]	= {
+			.size	= 500,
+			.limit	= 250
+		},
+		.mr_cache[6]	= {
+			.size	= 500,
+			.limit	= 250
+		},
+		.mr_cache[7]	= {
+			.size	= 500,
+			.limit	= 250
+		},
+		.mr_cache[8]	= {
+			.size	= 500,
+			.limit	= 250
+		},
+		.mr_cache[9]	= {
+			.size	= 500,
+			.limit	= 250
+		},
+		.mr_cache[10]	= {
+			.size	= 500,
+			.limit	= 250
+		},
+		.mr_cache[11]	= {
+			.size	= 500,
+			.limit	= 250
+		},
+		.mr_cache[12]	= {
+			.size	= 64,
+			.limit	= 32
+		},
+		.mr_cache[13]	= {
+			.size	= 32,
+			.limit	= 16
+		},
+		.mr_cache[14]	= {
+			.size	= 16,
+			.limit	= 8
+		},
+		.mr_cache[15]	= {
+			.size	= 8,
+			.limit	= 4
+		},
+	},
+};
+
+static int wait_fw_init(struct mlx5_core_dev *dev, u32 max_wait_mili,
+			u32 warn_time_mili)
+{
+	unsigned long warn = jiffies + msecs_to_jiffies(warn_time_mili);
+	unsigned long end = jiffies + msecs_to_jiffies(max_wait_mili);
+	u32 fw_initializing;
+	int err = 0;
+
+	do {
+		fw_initializing = ioread32be(&dev->iseg->initializing);
+		if (!(fw_initializing >> 31))
+			break;
+		if (time_after(jiffies, end) ||
+		    test_and_clear_bit(MLX5_BREAK_FW_WAIT, &dev->intf_state)) {
+			err = -EBUSY;
+			break;
+		}
+		if (warn_time_mili && time_after(jiffies, warn)) {
+			mlx5_core_warn(dev, "Waiting for FW initialization, timeout abort in %ds (0x%x)\n",
+				       jiffies_to_msecs(end - warn) / 1000, fw_initializing);
+			warn = jiffies + msecs_to_jiffies(warn_time_mili);
+		}
+		msleep(mlx5_tout_ms(dev, FW_PRE_INIT_WAIT));
+	} while (true);
+
+	return err;
+}
+
+static void mlx5_set_driver_version(struct mlx5_core_dev *dev)
+{
+	int driver_ver_sz = MLX5_FLD_SZ_BYTES(set_driver_version_in,
+					      driver_version);
+	u8 in[MLX5_ST_SZ_BYTES(set_driver_version_in)] = {};
+	int remaining_size = driver_ver_sz;
+	char *string;
+
+	if (!MLX5_CAP_GEN(dev, driver_version))
+		return;
+
+	string = MLX5_ADDR_OF(set_driver_version_in, in, driver_version);
+
+	strncpy(string, "Linux", remaining_size);
+
+	remaining_size = max_t(int, 0, driver_ver_sz - strlen(string));
+	strncat(string, ",", remaining_size);
+
+	remaining_size = max_t(int, 0, driver_ver_sz - strlen(string));
+	strncat(string, KBUILD_MODNAME, remaining_size);
+
+	remaining_size = max_t(int, 0, driver_ver_sz - strlen(string));
+	strncat(string, ",", remaining_size);
+
+	remaining_size = max_t(int, 0, driver_ver_sz - strlen(string));
+
+	snprintf(string + strlen(string), remaining_size, "%u.%u.%u",
+		LINUX_VERSION_MAJOR, LINUX_VERSION_PATCHLEVEL,
+		LINUX_VERSION_SUBLEVEL);
+
+	/*Send the command*/
+	MLX5_SET(set_driver_version_in, in, opcode,
+		 MLX5_CMD_OP_SET_DRIVER_VERSION);
+
+	mlx5_cmd_exec_in(dev, set_driver_version, in);
+}
+
+static int set_dma_caps(struct pci_dev *pdev)
+{
+	int err;
+
+	err = dma_set_mask_and_coherent(&pdev->dev, DMA_BIT_MASK(64));
+	if (err) {
+		dev_warn(&pdev->dev, "Warning: couldn't set 64-bit PCI DMA mask\n");
+		err = dma_set_mask_and_coherent(&pdev->dev, DMA_BIT_MASK(32));
+		if (err) {
+			dev_err(&pdev->dev, "Can't set PCI DMA mask, aborting\n");
+			return err;
+		}
+	}
+
+	dma_set_max_seg_size(&pdev->dev, 2u * 1024 * 1024 * 1024);
+	return err;
+}
+
+static int mlx5_pci_enable_device(struct mlx5_core_dev *dev)
+{
+	struct pci_dev *pdev = dev->pdev;
+	int err = 0;
+
+	mutex_lock(&dev->pci_status_mutex);
+	if (dev->pci_status == MLX5_PCI_STATUS_DISABLED) {
+		err = pci_enable_device(pdev);
+		if (!err)
+			dev->pci_status = MLX5_PCI_STATUS_ENABLED;
+	}
+	mutex_unlock(&dev->pci_status_mutex);
+
+	return err;
+}
+
+static void mlx5_pci_disable_device(struct mlx5_core_dev *dev)
+{
+	struct pci_dev *pdev = dev->pdev;
+
+	mutex_lock(&dev->pci_status_mutex);
+	if (dev->pci_status == MLX5_PCI_STATUS_ENABLED) {
+		pci_disable_device(pdev);
+		dev->pci_status = MLX5_PCI_STATUS_DISABLED;
+	}
+	mutex_unlock(&dev->pci_status_mutex);
+}
+
+static int request_bar(struct pci_dev *pdev)
+{
+	int err = 0;
+
+	if (!(pci_resource_flags(pdev, 0) & IORESOURCE_MEM)) {
+		dev_err(&pdev->dev, "Missing registers BAR, aborting\n");
+		return -ENODEV;
+	}
+
+	err = pci_request_regions(pdev, KBUILD_MODNAME);
+	if (err)
+		dev_err(&pdev->dev, "Couldn't get PCI resources, aborting\n");
+
+	return err;
+}
+
+static void release_bar(struct pci_dev *pdev)
+{
+	pci_release_regions(pdev);
+}
+
+struct mlx5_reg_host_endianness {
+	u8	he;
+	u8      rsvd[15];
+};
+
+static u16 to_fw_pkey_sz(struct mlx5_core_dev *dev, u32 size)
+{
+	switch (size) {
+	case 128:
+		return 0;
+	case 256:
+		return 1;
+	case 512:
+		return 2;
+	case 1024:
+		return 3;
+	case 2048:
+		return 4;
+	case 4096:
+		return 5;
+	default:
+		mlx5_core_warn(dev, "invalid pkey table size %d\n", size);
+		return 0;
+	}
+}
+
+static int mlx5_core_get_caps_mode(struct mlx5_core_dev *dev,
+				   enum mlx5_cap_type cap_type,
+				   enum mlx5_cap_mode cap_mode)
+{
+	u8 in[MLX5_ST_SZ_BYTES(query_hca_cap_in)];
+	int out_sz = MLX5_ST_SZ_BYTES(query_hca_cap_out);
+	void *out, *hca_caps;
+	u16 opmod = (cap_type << 1) | (cap_mode & 0x01);
+	int err;
+
+	memset(in, 0, sizeof(in));
+	out = kzalloc(out_sz, GFP_KERNEL);
+	if (!out)
+		return -ENOMEM;
+
+	MLX5_SET(query_hca_cap_in, in, opcode, MLX5_CMD_OP_QUERY_HCA_CAP);
+	MLX5_SET(query_hca_cap_in, in, op_mod, opmod);
+	err = mlx5_cmd_exec_inout(dev, query_hca_cap, in, out);
+	if (err) {
+		mlx5_core_warn(dev,
+			       "QUERY_HCA_CAP : type(%x) opmode(%x) Failed(%d)\n",
+			       cap_type, cap_mode, err);
+		goto query_ex;
+	}
+
+	hca_caps =  MLX5_ADDR_OF(query_hca_cap_out, out, capability);
+
+	switch (cap_mode) {
+	case HCA_CAP_OPMOD_GET_MAX:
+		memcpy(dev->caps.hca[cap_type]->max, hca_caps,
+		       MLX5_UN_SZ_BYTES(hca_cap_union));
+		break;
+	case HCA_CAP_OPMOD_GET_CUR:
+		memcpy(dev->caps.hca[cap_type]->cur, hca_caps,
+		       MLX5_UN_SZ_BYTES(hca_cap_union));
+		break;
+	default:
+		mlx5_core_warn(dev,
+			       "Tried to query dev cap type(%x) with wrong opmode(%x)\n",
+			       cap_type, cap_mode);
+		err = -EINVAL;
+		break;
+	}
+query_ex:
+	kfree(out);
+	return err;
+}
+
+int mlx5_core_get_caps(struct mlx5_core_dev *dev, enum mlx5_cap_type cap_type)
+{
+	int ret;
+
+	ret = mlx5_core_get_caps_mode(dev, cap_type, HCA_CAP_OPMOD_GET_CUR);
+	if (ret)
+		return ret;
+	return mlx5_core_get_caps_mode(dev, cap_type, HCA_CAP_OPMOD_GET_MAX);
+}
+
+static int set_caps(struct mlx5_core_dev *dev, void *in, int opmod)
+{
+	MLX5_SET(set_hca_cap_in, in, opcode, MLX5_CMD_OP_SET_HCA_CAP);
+	MLX5_SET(set_hca_cap_in, in, op_mod, opmod << 1);
+	return mlx5_cmd_exec_in(dev, set_hca_cap, in);
+}
+
+static int handle_hca_cap_atomic(struct mlx5_core_dev *dev, void *set_ctx)
+{
+	void *set_hca_cap;
+	int req_endianness;
+	int err;
+
+	if (!MLX5_CAP_GEN(dev, atomic))
+		return 0;
+
+	err = mlx5_core_get_caps(dev, MLX5_CAP_ATOMIC);
+	if (err)
+		return err;
+
+	req_endianness =
+		MLX5_CAP_ATOMIC(dev,
+				supported_atomic_req_8B_endianness_mode_1);
+
+	if (req_endianness != MLX5_ATOMIC_REQ_MODE_HOST_ENDIANNESS)
+		return 0;
+
+	set_hca_cap = MLX5_ADDR_OF(set_hca_cap_in, set_ctx, capability);
+
+	/* Set requestor to host endianness */
+	MLX5_SET(atomic_caps, set_hca_cap, atomic_req_8B_endianness_mode,
+		 MLX5_ATOMIC_REQ_MODE_HOST_ENDIANNESS);
+
+	return set_caps(dev, set_ctx, MLX5_SET_HCA_CAP_OP_MOD_ATOMIC);
+}
+
+static int handle_hca_cap_odp(struct mlx5_core_dev *dev, void *set_ctx)
+{
+	void *set_hca_cap;
+	bool do_set = false;
+	int err;
+
+	if (!IS_ENABLED(CONFIG_INFINIBAND_ON_DEMAND_PAGING) ||
+	    !MLX5_CAP_GEN(dev, pg))
+		return 0;
+
+	err = mlx5_core_get_caps(dev, MLX5_CAP_ODP);
+	if (err)
+		return err;
+
+	set_hca_cap = MLX5_ADDR_OF(set_hca_cap_in, set_ctx, capability);
+	memcpy(set_hca_cap, dev->caps.hca[MLX5_CAP_ODP]->cur,
+	       MLX5_ST_SZ_BYTES(odp_cap));
+
+#define ODP_CAP_SET_MAX(dev, field)                                            \
+	do {                                                                   \
+		u32 _res = MLX5_CAP_ODP_MAX(dev, field);                       \
+		if (_res) {                                                    \
+			do_set = true;                                         \
+			MLX5_SET(odp_cap, set_hca_cap, field, _res);           \
+		}                                                              \
+	} while (0)
+
+	ODP_CAP_SET_MAX(dev, ud_odp_caps.srq_receive);
+	ODP_CAP_SET_MAX(dev, rc_odp_caps.srq_receive);
+	ODP_CAP_SET_MAX(dev, xrc_odp_caps.srq_receive);
+	ODP_CAP_SET_MAX(dev, xrc_odp_caps.send);
+	ODP_CAP_SET_MAX(dev, xrc_odp_caps.receive);
+	ODP_CAP_SET_MAX(dev, xrc_odp_caps.write);
+	ODP_CAP_SET_MAX(dev, xrc_odp_caps.read);
+	ODP_CAP_SET_MAX(dev, xrc_odp_caps.atomic);
+	ODP_CAP_SET_MAX(dev, dc_odp_caps.srq_receive);
+	ODP_CAP_SET_MAX(dev, dc_odp_caps.send);
+	ODP_CAP_SET_MAX(dev, dc_odp_caps.receive);
+	ODP_CAP_SET_MAX(dev, dc_odp_caps.write);
+	ODP_CAP_SET_MAX(dev, dc_odp_caps.read);
+	ODP_CAP_SET_MAX(dev, dc_odp_caps.atomic);
+
+	if (!do_set)
+		return 0;
+
+	return set_caps(dev, set_ctx, MLX5_SET_HCA_CAP_OP_MOD_ODP);
+}
+
+static int max_uc_list_get_devlink_param(struct mlx5_core_dev *dev)
+{
+	struct devlink *devlink = priv_to_devlink(dev);
+	union devlink_param_value val;
+	int err;
+
+	err = devlink_param_driverinit_value_get(devlink,
+						 DEVLINK_PARAM_GENERIC_ID_MAX_MACS,
+						 &val);
+	if (!err)
+		return val.vu32;
+	mlx5_core_dbg(dev, "Failed to get param. err = %d\n", err);
+	return err;
+}
+
+bool mlx5_is_roce_on(struct mlx5_core_dev *dev)
+{
+	struct devlink *devlink = priv_to_devlink(dev);
+	union devlink_param_value val;
+	int err;
+
+	err = devlink_param_driverinit_value_get(devlink,
+						 DEVLINK_PARAM_GENERIC_ID_ENABLE_ROCE,
+						 &val);
+
+	if (!err)
+		return val.vbool;
+
+	mlx5_core_dbg(dev, "Failed to get param. err = %d\n", err);
+	return MLX5_CAP_GEN(dev, roce);
+}
+EXPORT_SYMBOL(mlx5_is_roce_on);
+
+static int handle_hca_cap_2(struct mlx5_core_dev *dev, void *set_ctx)
+{
+	void *set_hca_cap;
+	int err;
+
+	if (!MLX5_CAP_GEN_MAX(dev, hca_cap_2))
+		return 0;
+
+	err = mlx5_core_get_caps(dev, MLX5_CAP_GENERAL_2);
+	if (err)
+		return err;
+
+	if (!MLX5_CAP_GEN_2_MAX(dev, sw_vhca_id_valid) ||
+	    !(dev->priv.sw_vhca_id > 0))
+		return 0;
+
+	set_hca_cap = MLX5_ADDR_OF(set_hca_cap_in, set_ctx,
+				   capability);
+	memcpy(set_hca_cap, dev->caps.hca[MLX5_CAP_GENERAL_2]->cur,
+	       MLX5_ST_SZ_BYTES(cmd_hca_cap_2));
+	MLX5_SET(cmd_hca_cap_2, set_hca_cap, sw_vhca_id_valid, 1);
+
+	return set_caps(dev, set_ctx, MLX5_CAP_GENERAL_2);
+}
+
+static int handle_hca_cap(struct mlx5_core_dev *dev, void *set_ctx)
+{
+	struct mlx5_profile *prof = &dev->profile;
+	void *set_hca_cap;
+	int max_uc_list;
+	int err;
+
+	err = mlx5_core_get_caps(dev, MLX5_CAP_GENERAL);
+	if (err)
+		return err;
+
+	set_hca_cap = MLX5_ADDR_OF(set_hca_cap_in, set_ctx,
+				   capability);
+	memcpy(set_hca_cap, dev->caps.hca[MLX5_CAP_GENERAL]->cur,
+	       MLX5_ST_SZ_BYTES(cmd_hca_cap));
+
+	mlx5_core_dbg(dev, "Current Pkey table size %d Setting new size %d\n",
+		      mlx5_to_sw_pkey_sz(MLX5_CAP_GEN(dev, pkey_table_size)),
+		      128);
+	/* we limit the size of the pkey table to 128 entries for now */
+	MLX5_SET(cmd_hca_cap, set_hca_cap, pkey_table_size,
+		 to_fw_pkey_sz(dev, 128));
+
+	/* Check log_max_qp from HCA caps to set in current profile */
+	if (prof->log_max_qp == LOG_MAX_SUPPORTED_QPS) {
+		prof->log_max_qp = min_t(u8, 18, MLX5_CAP_GEN_MAX(dev, log_max_qp));
+	} else if (MLX5_CAP_GEN_MAX(dev, log_max_qp) < prof->log_max_qp) {
+		mlx5_core_warn(dev, "log_max_qp value in current profile is %d, changing it to HCA capability limit (%d)\n",
+			       prof->log_max_qp,
+			       MLX5_CAP_GEN_MAX(dev, log_max_qp));
+		prof->log_max_qp = MLX5_CAP_GEN_MAX(dev, log_max_qp);
+	}
+	if (prof->mask & MLX5_PROF_MASK_QP_SIZE)
+		MLX5_SET(cmd_hca_cap, set_hca_cap, log_max_qp,
+			 prof->log_max_qp);
+
+	/* disable cmdif checksum */
+	MLX5_SET(cmd_hca_cap, set_hca_cap, cmdif_checksum, 0);
+
+	/* Enable 4K UAR only when HCA supports it and page size is bigger
+	 * than 4K.
+	 */
+	if (MLX5_CAP_GEN_MAX(dev, uar_4k) && PAGE_SIZE > 4096)
+		MLX5_SET(cmd_hca_cap, set_hca_cap, uar_4k, 1);
+
+	MLX5_SET(cmd_hca_cap, set_hca_cap, log_uar_page_sz, PAGE_SHIFT - 12);
+
+	if (MLX5_CAP_GEN_MAX(dev, cache_line_128byte))
+		MLX5_SET(cmd_hca_cap,
+			 set_hca_cap,
+			 cache_line_128byte,
+			 cache_line_size() >= 128 ? 1 : 0);
+
+	if (MLX5_CAP_GEN_MAX(dev, dct))
+		MLX5_SET(cmd_hca_cap, set_hca_cap, dct, 1);
+
+	if (MLX5_CAP_GEN_MAX(dev, pci_sync_for_fw_update_event))
+		MLX5_SET(cmd_hca_cap, set_hca_cap, pci_sync_for_fw_update_event, 1);
+
+	if (MLX5_CAP_GEN_MAX(dev, num_vhca_ports))
+		MLX5_SET(cmd_hca_cap,
+			 set_hca_cap,
+			 num_vhca_ports,
+			 MLX5_CAP_GEN_MAX(dev, num_vhca_ports));
+
+	if (MLX5_CAP_GEN_MAX(dev, release_all_pages))
+		MLX5_SET(cmd_hca_cap, set_hca_cap, release_all_pages, 1);
+
+	if (MLX5_CAP_GEN_MAX(dev, mkey_by_name))
+		MLX5_SET(cmd_hca_cap, set_hca_cap, mkey_by_name, 1);
+
+	mlx5_vhca_state_cap_handle(dev, set_hca_cap);
+
+	if (MLX5_CAP_GEN_MAX(dev, num_total_dynamic_vf_msix))
+		MLX5_SET(cmd_hca_cap, set_hca_cap, num_total_dynamic_vf_msix,
+			 MLX5_CAP_GEN_MAX(dev, num_total_dynamic_vf_msix));
+
+	if (MLX5_CAP_GEN(dev, roce_rw_supported) && MLX5_CAP_GEN_MAX(dev, roce))
+		MLX5_SET(cmd_hca_cap, set_hca_cap, roce,
+			 mlx5_is_roce_on(dev));
+
+	max_uc_list = max_uc_list_get_devlink_param(dev);
+	if (max_uc_list > 0)
+		MLX5_SET(cmd_hca_cap, set_hca_cap, log_max_current_uc_list,
+			 ilog2(max_uc_list));
+
+	return set_caps(dev, set_ctx, MLX5_SET_HCA_CAP_OP_MOD_GENERAL_DEVICE);
+}
+
+/* Cached MLX5_CAP_GEN(dev, roce) can be out of sync this early in the
+ * boot process.
+ * In case RoCE cap is writable in FW and user/devlink requested to change the
+ * cap, we are yet to query the final state of the above cap.
+ * Hence, the need for this function.
+ *
+ * Returns
+ * True:
+ * 1) RoCE cap is read only in FW and already disabled
+ * OR:
+ * 2) RoCE cap is writable in FW and user/devlink requested it off.
+ *
+ * In any other case, return False.
+ */
+static bool is_roce_fw_disabled(struct mlx5_core_dev *dev)
+{
+	return (MLX5_CAP_GEN(dev, roce_rw_supported) && !mlx5_is_roce_on(dev)) ||
+		(!MLX5_CAP_GEN(dev, roce_rw_supported) && !MLX5_CAP_GEN(dev, roce));
+}
+
+static int handle_hca_cap_roce(struct mlx5_core_dev *dev, void *set_ctx)
+{
+	void *set_hca_cap;
+	int err;
+
+	if (is_roce_fw_disabled(dev))
+		return 0;
+
+	err = mlx5_core_get_caps(dev, MLX5_CAP_ROCE);
+	if (err)
+		return err;
+
+	if (MLX5_CAP_ROCE(dev, sw_r_roce_src_udp_port) ||
+	    !MLX5_CAP_ROCE_MAX(dev, sw_r_roce_src_udp_port))
+		return 0;
+
+	set_hca_cap = MLX5_ADDR_OF(set_hca_cap_in, set_ctx, capability);
+	memcpy(set_hca_cap, dev->caps.hca[MLX5_CAP_ROCE]->cur,
+	       MLX5_ST_SZ_BYTES(roce_cap));
+	MLX5_SET(roce_cap, set_hca_cap, sw_r_roce_src_udp_port, 1);
+
+	err = set_caps(dev, set_ctx, MLX5_SET_HCA_CAP_OP_MOD_ROCE);
+	return err;
+}
+
+static int handle_hca_cap_port_selection(struct mlx5_core_dev *dev,
+					 void *set_ctx)
+{
+	void *set_hca_cap;
+	int err;
+
+	if (!MLX5_CAP_GEN(dev, port_selection_cap))
+		return 0;
+
+	err = mlx5_core_get_caps(dev, MLX5_CAP_PORT_SELECTION);
+	if (err)
+		return err;
+
+	if (MLX5_CAP_PORT_SELECTION(dev, port_select_flow_table_bypass) ||
+	    !MLX5_CAP_PORT_SELECTION_MAX(dev, port_select_flow_table_bypass))
+		return 0;
+
+	set_hca_cap = MLX5_ADDR_OF(set_hca_cap_in, set_ctx, capability);
+	memcpy(set_hca_cap, dev->caps.hca[MLX5_CAP_PORT_SELECTION]->cur,
+	       MLX5_ST_SZ_BYTES(port_selection_cap));
+	MLX5_SET(port_selection_cap, set_hca_cap, port_select_flow_table_bypass, 1);
+
+	err = set_caps(dev, set_ctx, MLX5_SET_HCA_CAP_OP_MODE_PORT_SELECTION);
+
+	return err;
+}
+
+static int set_hca_cap(struct mlx5_core_dev *dev)
+{
+	int set_sz = MLX5_ST_SZ_BYTES(set_hca_cap_in);
+	void *set_ctx;
+	int err;
+
+	set_ctx = kzalloc(set_sz, GFP_KERNEL);
+	if (!set_ctx)
+		return -ENOMEM;
+
+	err = handle_hca_cap(dev, set_ctx);
+	if (err) {
+		mlx5_core_err(dev, "handle_hca_cap failed\n");
+		goto out;
+	}
+
+	memset(set_ctx, 0, set_sz);
+	err = handle_hca_cap_atomic(dev, set_ctx);
+	if (err) {
+		mlx5_core_err(dev, "handle_hca_cap_atomic failed\n");
+		goto out;
+	}
+
+	memset(set_ctx, 0, set_sz);
+	err = handle_hca_cap_odp(dev, set_ctx);
+	if (err) {
+		mlx5_core_err(dev, "handle_hca_cap_odp failed\n");
+		goto out;
+	}
+
+	memset(set_ctx, 0, set_sz);
+	err = handle_hca_cap_roce(dev, set_ctx);
+	if (err) {
+		mlx5_core_err(dev, "handle_hca_cap_roce failed\n");
+		goto out;
+	}
+
+	memset(set_ctx, 0, set_sz);
+	err = handle_hca_cap_2(dev, set_ctx);
+	if (err) {
+		mlx5_core_err(dev, "handle_hca_cap_2 failed\n");
+		goto out;
+	}
+
+	memset(set_ctx, 0, set_sz);
+	err = handle_hca_cap_port_selection(dev, set_ctx);
+	if (err) {
+		mlx5_core_err(dev, "handle_hca_cap_port_selection failed\n");
+		goto out;
+	}
+
+out:
+	kfree(set_ctx);
+	return err;
+}
+
+static int set_hca_ctrl(struct mlx5_core_dev *dev)
+{
+	struct mlx5_reg_host_endianness he_in;
+	struct mlx5_reg_host_endianness he_out;
+	int err;
+
+	if (!mlx5_core_is_pf(dev))
+		return 0;
+
+	memset(&he_in, 0, sizeof(he_in));
+	he_in.he = MLX5_SET_HOST_ENDIANNESS;
+	err = mlx5_core_access_reg(dev, &he_in,  sizeof(he_in),
+					&he_out, sizeof(he_out),
+					MLX5_REG_HOST_ENDIANNESS, 0, 1);
+	return err;
+}
+
+static int mlx5_core_set_hca_defaults(struct mlx5_core_dev *dev)
+{
+	int ret = 0;
+
+	/* Disable local_lb by default */
+	if (MLX5_CAP_GEN(dev, port_type) == MLX5_CAP_PORT_TYPE_ETH)
+		ret = mlx5_nic_vport_update_local_lb(dev, false);
+
+	return ret;
+}
+
+int mlx5_core_enable_hca(struct mlx5_core_dev *dev, u16 func_id)
+{
+	u32 in[MLX5_ST_SZ_DW(enable_hca_in)] = {};
+
+	MLX5_SET(enable_hca_in, in, opcode, MLX5_CMD_OP_ENABLE_HCA);
+	MLX5_SET(enable_hca_in, in, function_id, func_id);
+	MLX5_SET(enable_hca_in, in, embedded_cpu_function,
+		 dev->caps.embedded_cpu);
+	return mlx5_cmd_exec_in(dev, enable_hca, in);
+}
+
+int mlx5_core_disable_hca(struct mlx5_core_dev *dev, u16 func_id)
+{
+	u32 in[MLX5_ST_SZ_DW(disable_hca_in)] = {};
+
+	MLX5_SET(disable_hca_in, in, opcode, MLX5_CMD_OP_DISABLE_HCA);
+	MLX5_SET(disable_hca_in, in, function_id, func_id);
+	MLX5_SET(enable_hca_in, in, embedded_cpu_function,
+		 dev->caps.embedded_cpu);
+	return mlx5_cmd_exec_in(dev, disable_hca, in);
+}
+
+static int mlx5_core_set_issi(struct mlx5_core_dev *dev)
+{
+	u32 query_out[MLX5_ST_SZ_DW(query_issi_out)] = {};
+	u32 query_in[MLX5_ST_SZ_DW(query_issi_in)] = {};
+	u32 sup_issi;
+	int err;
+
+	MLX5_SET(query_issi_in, query_in, opcode, MLX5_CMD_OP_QUERY_ISSI);
+	err = mlx5_cmd_exec_inout(dev, query_issi, query_in, query_out);
+	if (err) {
+		u32 syndrome = MLX5_GET(query_issi_out, query_out, syndrome);
+		u8 status = MLX5_GET(query_issi_out, query_out, status);
+
+		if (!status || syndrome == MLX5_DRIVER_SYND) {
+			mlx5_core_err(dev, "Failed to query ISSI err(%d) status(%d) synd(%d)\n",
+				      err, status, syndrome);
+			return err;
+		}
+
+		mlx5_core_warn(dev, "Query ISSI is not supported by FW, ISSI is 0\n");
+		dev->issi = 0;
+		return 0;
+	}
+
+	sup_issi = MLX5_GET(query_issi_out, query_out, supported_issi_dw0);
+
+	if (sup_issi & (1 << 1)) {
+		u32 set_in[MLX5_ST_SZ_DW(set_issi_in)] = {};
+
+		MLX5_SET(set_issi_in, set_in, opcode, MLX5_CMD_OP_SET_ISSI);
+		MLX5_SET(set_issi_in, set_in, current_issi, 1);
+		err = mlx5_cmd_exec_in(dev, set_issi, set_in);
+		if (err) {
+			mlx5_core_err(dev, "Failed to set ISSI to 1 err(%d)\n",
+				      err);
+			return err;
+		}
+
+		dev->issi = 1;
+
+		return 0;
+	} else if (sup_issi & (1 << 0) || !sup_issi) {
+		return 0;
+	}
+
+	return -EOPNOTSUPP;
+}
+
+static int mlx5_pci_init(struct mlx5_core_dev *dev, struct pci_dev *pdev,
+			 const struct pci_device_id *id)
+{
+	int err = 0;
+
+	mutex_init(&dev->pci_status_mutex);
+	pci_set_drvdata(dev->pdev, dev);
+
+	dev->bar_addr = pci_resource_start(pdev, 0);
+
+	err = mlx5_pci_enable_device(dev);
+	if (err) {
+		mlx5_core_err(dev, "Cannot enable PCI device, aborting\n");
+		return err;
+	}
+
+	err = request_bar(pdev);
+	if (err) {
+		mlx5_core_err(dev, "error requesting BARs, aborting\n");
+		goto err_disable;
+	}
+
+	pci_set_master(pdev);
+
+	err = set_dma_caps(pdev);
+	if (err) {
+		mlx5_core_err(dev, "Failed setting DMA capabilities mask, aborting\n");
+		goto err_clr_master;
+	}
+
+	if (pci_enable_atomic_ops_to_root(pdev, PCI_EXP_DEVCAP2_ATOMIC_COMP32) &&
+	    pci_enable_atomic_ops_to_root(pdev, PCI_EXP_DEVCAP2_ATOMIC_COMP64) &&
+	    pci_enable_atomic_ops_to_root(pdev, PCI_EXP_DEVCAP2_ATOMIC_COMP128))
+		mlx5_core_dbg(dev, "Enabling pci atomics failed\n");
+
+	dev->iseg_base = dev->bar_addr;
+	dev->iseg = ioremap(dev->iseg_base, sizeof(*dev->iseg));
+	if (!dev->iseg) {
+		err = -ENOMEM;
+		mlx5_core_err(dev, "Failed mapping initialization segment, aborting\n");
+		goto err_clr_master;
+	}
+
+	mlx5_pci_vsc_init(dev);
+	return 0;
+
+err_clr_master:
+	pci_clear_master(dev->pdev);
+	release_bar(dev->pdev);
+err_disable:
+	mlx5_pci_disable_device(dev);
+	return err;
+}
+
+static void mlx5_pci_close(struct mlx5_core_dev *dev)
+{
+	/* health work might still be active, and it needs pci bar in
+	 * order to know the NIC state. Therefore, drain the health WQ
+	 * before removing the pci bars
+	 */
+	mlx5_drain_health_wq(dev);
+	iounmap(dev->iseg);
+	pci_clear_master(dev->pdev);
+	release_bar(dev->pdev);
+	mlx5_pci_disable_device(dev);
+}
+
+static int mlx5_init_once(struct mlx5_core_dev *dev)
+{
+	int err;
+
+	dev->priv.devcom = mlx5_devcom_register_device(dev);
+	if (IS_ERR(dev->priv.devcom))
+		mlx5_core_err(dev, "failed to register with devcom (0x%p)\n",
+			      dev->priv.devcom);
+
+	err = mlx5_query_board_id(dev);
+	if (err) {
+		mlx5_core_err(dev, "query board id failed\n");
+		goto err_devcom;
+	}
+
+	err = mlx5_irq_table_init(dev);
+	if (err) {
+		mlx5_core_err(dev, "failed to initialize irq table\n");
+		goto err_devcom;
+	}
+
+	err = mlx5_eq_table_init(dev);
+	if (err) {
+		mlx5_core_err(dev, "failed to initialize eq\n");
+		goto err_irq_cleanup;
+	}
+
+	err = mlx5_events_init(dev);
+	if (err) {
+		mlx5_core_err(dev, "failed to initialize events\n");
+		goto err_eq_cleanup;
+	}
+
+	err = mlx5_fw_reset_init(dev);
+	if (err) {
+		mlx5_core_err(dev, "failed to initialize fw reset events\n");
+		goto err_events_cleanup;
+	}
+
+	mlx5_cq_debugfs_init(dev);
+
+	mlx5_init_reserved_gids(dev);
+
+	mlx5_init_clock(dev);
+
+	dev->vxlan = mlx5_vxlan_create(dev);
+	dev->geneve = mlx5_geneve_create(dev);
+
+	err = mlx5_init_rl_table(dev);
+	if (err) {
+		mlx5_core_err(dev, "Failed to init rate limiting\n");
+		goto err_tables_cleanup;
+	}
+
+	err = mlx5_mpfs_init(dev);
+	if (err) {
+		mlx5_core_err(dev, "Failed to init l2 table %d\n", err);
+		goto err_rl_cleanup;
+	}
+
+	err = mlx5_sriov_init(dev);
+	if (err) {
+		mlx5_core_err(dev, "Failed to init sriov %d\n", err);
+		goto err_mpfs_cleanup;
+	}
+
+	err = mlx5_eswitch_init(dev);
+	if (err) {
+		mlx5_core_err(dev, "Failed to init eswitch %d\n", err);
+		goto err_sriov_cleanup;
+	}
+
+	err = mlx5_fpga_init(dev);
+	if (err) {
+		mlx5_core_err(dev, "Failed to init fpga device %d\n", err);
+		goto err_eswitch_cleanup;
+	}
+
+	err = mlx5_vhca_event_init(dev);
+	if (err) {
+		mlx5_core_err(dev, "Failed to init vhca event notifier %d\n", err);
+		goto err_fpga_cleanup;
+	}
+
+	err = mlx5_sf_hw_table_init(dev);
+	if (err) {
+		mlx5_core_err(dev, "Failed to init SF HW table %d\n", err);
+		goto err_sf_hw_table_cleanup;
+	}
+
+	err = mlx5_sf_table_init(dev);
+	if (err) {
+		mlx5_core_err(dev, "Failed to init SF table %d\n", err);
+		goto err_sf_table_cleanup;
+	}
+
+	err = mlx5_fs_core_alloc(dev);
+	if (err) {
+		mlx5_core_err(dev, "Failed to alloc flow steering\n");
+		goto err_fs;
+	}
+
+	dev->dm = mlx5_dm_create(dev);
+	if (IS_ERR(dev->dm))
+		mlx5_core_warn(dev, "Failed to init device memory %ld\n", PTR_ERR(dev->dm));
+
+	dev->tracer = mlx5_fw_tracer_create(dev);
+	dev->hv_vhca = mlx5_hv_vhca_create(dev);
+	dev->rsc_dump = mlx5_rsc_dump_create(dev);
+
+	return 0;
+
+err_fs:
+	mlx5_sf_table_cleanup(dev);
+err_sf_table_cleanup:
+	mlx5_sf_hw_table_cleanup(dev);
+err_sf_hw_table_cleanup:
+	mlx5_vhca_event_cleanup(dev);
+err_fpga_cleanup:
+	mlx5_fpga_cleanup(dev);
+err_eswitch_cleanup:
+	mlx5_eswitch_cleanup(dev->priv.eswitch);
+err_sriov_cleanup:
+	mlx5_sriov_cleanup(dev);
+err_mpfs_cleanup:
+	mlx5_mpfs_cleanup(dev);
+err_rl_cleanup:
+	mlx5_cleanup_rl_table(dev);
+err_tables_cleanup:
+	mlx5_geneve_destroy(dev->geneve);
+	mlx5_vxlan_destroy(dev->vxlan);
+	mlx5_cleanup_clock(dev);
+	mlx5_cleanup_reserved_gids(dev);
+	mlx5_cq_debugfs_cleanup(dev);
+	mlx5_fw_reset_cleanup(dev);
+err_events_cleanup:
+	mlx5_events_cleanup(dev);
+err_eq_cleanup:
+	mlx5_eq_table_cleanup(dev);
+err_irq_cleanup:
+	mlx5_irq_table_cleanup(dev);
+err_devcom:
+	mlx5_devcom_unregister_device(dev->priv.devcom);
+
+	return err;
+}
+
+static void mlx5_cleanup_once(struct mlx5_core_dev *dev)
+{
+	mlx5_rsc_dump_destroy(dev);
+	mlx5_hv_vhca_destroy(dev->hv_vhca);
+	mlx5_fw_tracer_destroy(dev->tracer);
+	mlx5_dm_cleanup(dev);
+	mlx5_fs_core_free(dev);
+	mlx5_sf_table_cleanup(dev);
+	mlx5_sf_hw_table_cleanup(dev);
+	mlx5_vhca_event_cleanup(dev);
+	mlx5_fpga_cleanup(dev);
+	mlx5_eswitch_cleanup(dev->priv.eswitch);
+	mlx5_sriov_cleanup(dev);
+	mlx5_mpfs_cleanup(dev);
+	mlx5_cleanup_rl_table(dev);
+	mlx5_geneve_destroy(dev->geneve);
+	mlx5_vxlan_destroy(dev->vxlan);
+	mlx5_cleanup_clock(dev);
+	mlx5_cleanup_reserved_gids(dev);
+	mlx5_cq_debugfs_cleanup(dev);
+	mlx5_fw_reset_cleanup(dev);
+	mlx5_events_cleanup(dev);
+	mlx5_eq_table_cleanup(dev);
+	mlx5_irq_table_cleanup(dev);
+	mlx5_devcom_unregister_device(dev->priv.devcom);
+}
+
+static int mlx5_function_setup(struct mlx5_core_dev *dev, bool boot, u64 timeout)
+{
+	int err;
+
+	mlx5_core_info(dev, "firmware version: %d.%d.%d\n", fw_rev_maj(dev),
+		       fw_rev_min(dev), fw_rev_sub(dev));
+
+	/* Only PFs hold the relevant PCIe information for this query */
+	if (mlx5_core_is_pf(dev))
+		pcie_print_link_status(dev->pdev);
+
+	/* wait for firmware to accept initialization segments configurations
+	 */
+	err = wait_fw_init(dev, timeout,
+			   mlx5_tout_ms(dev, FW_PRE_INIT_WARN_MESSAGE_INTERVAL));
+	if (err) {
+		mlx5_core_err(dev, "Firmware over %llu MS in pre-initializing state, aborting\n",
+			      timeout);
+		return err;
+	}
+
+	err = mlx5_cmd_init(dev);
+	if (err) {
+		mlx5_core_err(dev, "Failed initializing command interface, aborting\n");
+		return err;
+	}
+
+	mlx5_tout_query_iseg(dev);
+
+	err = wait_fw_init(dev, mlx5_tout_ms(dev, FW_INIT), 0);
+	if (err) {
+		mlx5_core_err(dev, "Firmware over %llu MS in initializing state, aborting\n",
+			      mlx5_tout_ms(dev, FW_INIT));
+		goto err_cmd_cleanup;
+	}
+
+	dev->caps.embedded_cpu = mlx5_read_embedded_cpu(dev);
+	mlx5_cmd_set_state(dev, MLX5_CMDIF_STATE_UP);
+
+	mlx5_start_health_poll(dev);
+
+	err = mlx5_core_enable_hca(dev, 0);
+	if (err) {
+		mlx5_core_err(dev, "enable hca failed\n");
+		goto stop_health_poll;
+	}
+
+	err = mlx5_core_set_issi(dev);
+	if (err) {
+		mlx5_core_err(dev, "failed to set issi\n");
+		goto err_disable_hca;
+	}
+
+	err = mlx5_satisfy_startup_pages(dev, 1);
+	if (err) {
+		mlx5_core_err(dev, "failed to allocate boot pages\n");
+		goto err_disable_hca;
+	}
+
+	err = mlx5_tout_query_dtor(dev);
+	if (err) {
+		mlx5_core_err(dev, "failed to read dtor\n");
+		goto reclaim_boot_pages;
+	}
+
+	err = set_hca_ctrl(dev);
+	if (err) {
+		mlx5_core_err(dev, "set_hca_ctrl failed\n");
+		goto reclaim_boot_pages;
+	}
+
+	err = set_hca_cap(dev);
+	if (err) {
+		mlx5_core_err(dev, "set_hca_cap failed\n");
+		goto reclaim_boot_pages;
+	}
+
+	err = mlx5_satisfy_startup_pages(dev, 0);
+	if (err) {
+		mlx5_core_err(dev, "failed to allocate init pages\n");
+		goto reclaim_boot_pages;
+	}
+
+	err = mlx5_cmd_init_hca(dev, sw_owner_id);
+	if (err) {
+		mlx5_core_err(dev, "init hca failed\n");
+		goto reclaim_boot_pages;
+	}
+
+	mlx5_set_driver_version(dev);
+
+	err = mlx5_query_hca_caps(dev);
+	if (err) {
+		mlx5_core_err(dev, "query hca failed\n");
+		goto reclaim_boot_pages;
+	}
+	mlx5_start_health_fw_log_up(dev);
+
+	return 0;
+
+reclaim_boot_pages:
+	mlx5_reclaim_startup_pages(dev);
+err_disable_hca:
+	mlx5_core_disable_hca(dev, 0);
+stop_health_poll:
+	mlx5_stop_health_poll(dev, boot);
+err_cmd_cleanup:
+	mlx5_cmd_set_state(dev, MLX5_CMDIF_STATE_DOWN);
+	mlx5_cmd_cleanup(dev);
+
+	return err;
+}
+
+static int mlx5_function_teardown(struct mlx5_core_dev *dev, bool boot)
+{
+	int err;
+
+	err = mlx5_cmd_teardown_hca(dev);
+	if (err) {
+		mlx5_core_err(dev, "tear_down_hca failed, skip cleanup\n");
+		return err;
+	}
+	mlx5_reclaim_startup_pages(dev);
+	mlx5_core_disable_hca(dev, 0);
+	mlx5_stop_health_poll(dev, boot);
+	mlx5_cmd_set_state(dev, MLX5_CMDIF_STATE_DOWN);
+	mlx5_cmd_cleanup(dev);
+
+	return 0;
+}
+
+static int mlx5_load(struct mlx5_core_dev *dev)
+{
+	int err;
+
+	dev->priv.uar = mlx5_get_uars_page(dev);
+	if (IS_ERR(dev->priv.uar)) {
+		mlx5_core_err(dev, "Failed allocating uar, aborting\n");
+		err = PTR_ERR(dev->priv.uar);
+		return err;
+	}
+
+	mlx5_events_start(dev);
+	mlx5_pagealloc_start(dev);
+
+	err = mlx5_irq_table_create(dev);
+	if (err) {
+		mlx5_core_err(dev, "Failed to alloc IRQs\n");
+		goto err_irq_table;
+	}
+
+	err = mlx5_eq_table_create(dev);
+	if (err) {
+		mlx5_core_err(dev, "Failed to create EQs\n");
+		goto err_eq_table;
+	}
+
+	err = mlx5_fw_tracer_init(dev->tracer);
+	if (err) {
+		mlx5_core_err(dev, "Failed to init FW tracer %d\n", err);
+		mlx5_fw_tracer_destroy(dev->tracer);
+		dev->tracer = NULL;
+	}
+
+	mlx5_fw_reset_events_start(dev);
+	mlx5_hv_vhca_init(dev->hv_vhca);
+
+	err = mlx5_rsc_dump_init(dev);
+	if (err) {
+		mlx5_core_err(dev, "Failed to init Resource dump %d\n", err);
+		mlx5_rsc_dump_destroy(dev);
+		dev->rsc_dump = NULL;
+	}
+
+	err = mlx5_fpga_device_start(dev);
+	if (err) {
+		mlx5_core_err(dev, "fpga device start failed %d\n", err);
+		goto err_fpga_start;
+	}
+
+	err = mlx5_fs_core_init(dev);
+	if (err) {
+		mlx5_core_err(dev, "Failed to init flow steering\n");
+		goto err_fs;
+	}
+
+	err = mlx5_core_set_hca_defaults(dev);
+	if (err) {
+		mlx5_core_err(dev, "Failed to set hca defaults\n");
+		goto err_set_hca;
+	}
+
+	mlx5_vhca_event_start(dev);
+
+	err = mlx5_sf_hw_table_create(dev);
+	if (err) {
+		mlx5_core_err(dev, "sf table create failed %d\n", err);
+		goto err_vhca;
+	}
+
+	err = mlx5_ec_init(dev);
+	if (err) {
+		mlx5_core_err(dev, "Failed to init embedded CPU\n");
+		goto err_ec;
+	}
+
+	mlx5_lag_add_mdev(dev);
+	err = mlx5_sriov_attach(dev);
+	if (err) {
+		mlx5_core_err(dev, "sriov init failed %d\n", err);
+		goto err_sriov;
+	}
+
+	mlx5_sf_dev_table_create(dev);
+
+	return 0;
+
+err_sriov:
+	mlx5_lag_remove_mdev(dev);
+	mlx5_ec_cleanup(dev);
+err_ec:
+	mlx5_sf_hw_table_destroy(dev);
+err_vhca:
+	mlx5_vhca_event_stop(dev);
+err_set_hca:
+	mlx5_fs_core_cleanup(dev);
+err_fs:
+	mlx5_fpga_device_stop(dev);
+err_fpga_start:
+	mlx5_rsc_dump_cleanup(dev);
+	mlx5_hv_vhca_cleanup(dev->hv_vhca);
+	mlx5_fw_reset_events_stop(dev);
+	mlx5_fw_tracer_cleanup(dev->tracer);
+	mlx5_eq_table_destroy(dev);
+err_eq_table:
+	mlx5_irq_table_destroy(dev);
+err_irq_table:
+	mlx5_pagealloc_stop(dev);
+	mlx5_events_stop(dev);
+	mlx5_put_uars_page(dev, dev->priv.uar);
+	return err;
+}
+
+static void mlx5_unload(struct mlx5_core_dev *dev)
+{
+	mlx5_sf_dev_table_destroy(dev);
+	mlx5_eswitch_disable(dev->priv.eswitch);
+	mlx5_sriov_detach(dev);
+	mlx5_lag_remove_mdev(dev);
+	mlx5_ec_cleanup(dev);
+	mlx5_sf_hw_table_destroy(dev);
+	mlx5_vhca_event_stop(dev);
+	mlx5_fs_core_cleanup(dev);
+	mlx5_fpga_device_stop(dev);
+	mlx5_rsc_dump_cleanup(dev);
+	mlx5_hv_vhca_cleanup(dev->hv_vhca);
+	mlx5_fw_reset_events_stop(dev);
+	mlx5_fw_tracer_cleanup(dev->tracer);
+	mlx5_eq_table_destroy(dev);
+	mlx5_irq_table_destroy(dev);
+	mlx5_pagealloc_stop(dev);
+	mlx5_events_stop(dev);
+	mlx5_put_uars_page(dev, dev->priv.uar);
+}
+
+int mlx5_init_one(struct mlx5_core_dev *dev)
+{
+	struct devlink *devlink = priv_to_devlink(dev);
+	int err = 0;
+
+	devl_lock(devlink);
+	mutex_lock(&dev->intf_state_mutex);
+	dev->state = MLX5_DEVICE_STATE_UP;
+
+	err = mlx5_function_setup(dev, true, mlx5_tout_ms(dev, FW_PRE_INIT_TIMEOUT));
+	if (err)
+		goto err_function;
+
+	err = mlx5_init_once(dev);
+	if (err) {
+		mlx5_core_err(dev, "sw objs init failed\n");
+		goto function_teardown;
+	}
+
+	err = mlx5_load(dev);
+	if (err)
+		goto err_load;
+
+	set_bit(MLX5_INTERFACE_STATE_UP, &dev->intf_state);
+
+	err = mlx5_devlink_register(priv_to_devlink(dev));
+	if (err)
+		goto err_devlink_reg;
+
+	err = mlx5_register_device(dev);
+	if (err)
+		goto err_register;
+
+	mutex_unlock(&dev->intf_state_mutex);
+	devl_unlock(devlink);
+	return 0;
+
+err_register:
+	mlx5_devlink_unregister(priv_to_devlink(dev));
+err_devlink_reg:
+	clear_bit(MLX5_INTERFACE_STATE_UP, &dev->intf_state);
+	mlx5_unload(dev);
+err_load:
+	mlx5_cleanup_once(dev);
+function_teardown:
+	mlx5_function_teardown(dev, true);
+err_function:
+	dev->state = MLX5_DEVICE_STATE_INTERNAL_ERROR;
+	mutex_unlock(&dev->intf_state_mutex);
+	devl_unlock(devlink);
+	return err;
+}
+
+void mlx5_uninit_one(struct mlx5_core_dev *dev)
+{
+	struct devlink *devlink = priv_to_devlink(dev);
+
+	devl_lock(devlink);
+	mutex_lock(&dev->intf_state_mutex);
+
+	mlx5_unregister_device(dev);
+	mlx5_devlink_unregister(priv_to_devlink(dev));
+
+	if (!test_bit(MLX5_INTERFACE_STATE_UP, &dev->intf_state)) {
+		mlx5_core_warn(dev, "%s: interface is down, NOP\n",
+			       __func__);
+		mlx5_cleanup_once(dev);
+		goto out;
+	}
+
+	clear_bit(MLX5_INTERFACE_STATE_UP, &dev->intf_state);
+	mlx5_unload(dev);
+	mlx5_cleanup_once(dev);
+	mlx5_function_teardown(dev, true);
+out:
+	mutex_unlock(&dev->intf_state_mutex);
+	devl_unlock(devlink);
+}
+
+int mlx5_load_one_devl_locked(struct mlx5_core_dev *dev, bool recovery)
+{
+	int err = 0;
+	u64 timeout;
+
+	devl_assert_locked(priv_to_devlink(dev));
+	mutex_lock(&dev->intf_state_mutex);
+	if (test_bit(MLX5_INTERFACE_STATE_UP, &dev->intf_state)) {
+		mlx5_core_warn(dev, "interface is up, NOP\n");
+		goto out;
+	}
+	/* remove any previous indication of internal error */
+	dev->state = MLX5_DEVICE_STATE_UP;
+
+	if (recovery)
+		timeout = mlx5_tout_ms(dev, FW_PRE_INIT_ON_RECOVERY_TIMEOUT);
+	else
+		timeout = mlx5_tout_ms(dev, FW_PRE_INIT_TIMEOUT);
+	err = mlx5_function_setup(dev, false, timeout);
+	if (err)
+		goto err_function;
+
+	err = mlx5_load(dev);
+	if (err)
+		goto err_load;
+
+	set_bit(MLX5_INTERFACE_STATE_UP, &dev->intf_state);
+
+	err = mlx5_attach_device(dev);
+	if (err)
+		goto err_attach;
+
+	mutex_unlock(&dev->intf_state_mutex);
+	return 0;
+
+err_attach:
+	clear_bit(MLX5_INTERFACE_STATE_UP, &dev->intf_state);
+	mlx5_unload(dev);
+err_load:
+	mlx5_function_teardown(dev, false);
+err_function:
+	dev->state = MLX5_DEVICE_STATE_INTERNAL_ERROR;
+out:
+	mutex_unlock(&dev->intf_state_mutex);
+	return err;
+}
+
+int mlx5_load_one(struct mlx5_core_dev *dev, bool recovery)
+{
+	struct devlink *devlink = priv_to_devlink(dev);
+	int ret;
+
+	devl_lock(devlink);
+	ret = mlx5_load_one_devl_locked(dev, recovery);
+	devl_unlock(devlink);
+	return ret;
+}
+
+void mlx5_unload_one_devl_locked(struct mlx5_core_dev *dev, bool suspend)
+{
+	devl_assert_locked(priv_to_devlink(dev));
+	mutex_lock(&dev->intf_state_mutex);
+
+	mlx5_detach_device(dev, suspend);
+
+	if (!test_bit(MLX5_INTERFACE_STATE_UP, &dev->intf_state)) {
+		mlx5_core_warn(dev, "%s: interface is down, NOP\n",
+			       __func__);
+		goto out;
+	}
+
+	clear_bit(MLX5_INTERFACE_STATE_UP, &dev->intf_state);
+	mlx5_unload(dev);
+	mlx5_function_teardown(dev, false);
+out:
+	mutex_unlock(&dev->intf_state_mutex);
+}
+
+void mlx5_unload_one(struct mlx5_core_dev *dev, bool suspend)
+{
+	struct devlink *devlink = priv_to_devlink(dev);
+
+	devl_lock(devlink);
+	mlx5_unload_one_devl_locked(dev, suspend);
+	devl_unlock(devlink);
+}
+
+static const int types[] = {
+	MLX5_CAP_GENERAL,
+	MLX5_CAP_GENERAL_2,
+	MLX5_CAP_ETHERNET_OFFLOADS,
+	MLX5_CAP_IPOIB_ENHANCED_OFFLOADS,
+	MLX5_CAP_ODP,
+	MLX5_CAP_ATOMIC,
+	MLX5_CAP_ROCE,
+	MLX5_CAP_IPOIB_OFFLOADS,
+	MLX5_CAP_FLOW_TABLE,
+	MLX5_CAP_ESWITCH_FLOW_TABLE,
+	MLX5_CAP_ESWITCH,
+	MLX5_CAP_VECTOR_CALC,
+	MLX5_CAP_QOS,
+	MLX5_CAP_DEBUG,
+	MLX5_CAP_DEV_MEM,
+	MLX5_CAP_DEV_EVENT,
+	MLX5_CAP_TLS,
+	MLX5_CAP_VDPA_EMULATION,
+	MLX5_CAP_IPSEC,
+	MLX5_CAP_PORT_SELECTION,
+	MLX5_CAP_DEV_SHAMPO,
+	MLX5_CAP_MACSEC,
+	MLX5_CAP_ADV_VIRTUALIZATION,
+};
+
+static void mlx5_hca_caps_free(struct mlx5_core_dev *dev)
+{
+	int type;
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(types); i++) {
+		type = types[i];
+		kfree(dev->caps.hca[type]);
+	}
+}
+
+static int mlx5_hca_caps_alloc(struct mlx5_core_dev *dev)
+{
+	struct mlx5_hca_cap *cap;
+	int type;
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(types); i++) {
+		cap = kzalloc(sizeof(*cap), GFP_KERNEL);
+		if (!cap)
+			goto err;
+		type = types[i];
+		dev->caps.hca[type] = cap;
+	}
+
+	return 0;
+
+err:
+	mlx5_hca_caps_free(dev);
+	return -ENOMEM;
+}
+
+int mlx5_mdev_init(struct mlx5_core_dev *dev, int profile_idx)
+{
+	struct mlx5_priv *priv = &dev->priv;
+	int err;
+
+	memcpy(&dev->profile, &profile[profile_idx], sizeof(dev->profile));
+	INIT_LIST_HEAD(&priv->ctx_list);
+	spin_lock_init(&priv->ctx_lock);
+	lockdep_register_key(&dev->lock_key);
+	mutex_init(&dev->intf_state_mutex);
+	lockdep_set_class(&dev->intf_state_mutex, &dev->lock_key);
+
+	mutex_init(&priv->bfregs.reg_head.lock);
+	mutex_init(&priv->bfregs.wc_head.lock);
+	INIT_LIST_HEAD(&priv->bfregs.reg_head.list);
+	INIT_LIST_HEAD(&priv->bfregs.wc_head.list);
+
+	mutex_init(&priv->alloc_mutex);
+	mutex_init(&priv->pgdir_mutex);
+	INIT_LIST_HEAD(&priv->pgdir_list);
+
+	priv->numa_node = dev_to_node(mlx5_core_dma_dev(dev));
+	priv->dbg.dbg_root = debugfs_create_dir(dev_name(dev->device),
+						mlx5_debugfs_root);
+	INIT_LIST_HEAD(&priv->traps);
+
+	err = mlx5_tout_init(dev);
+	if (err) {
+		mlx5_core_err(dev, "Failed initializing timeouts, aborting\n");
+		goto err_timeout_init;
+	}
+
+	err = mlx5_health_init(dev);
+	if (err)
+		goto err_health_init;
+
+	err = mlx5_pagealloc_init(dev);
+	if (err)
+		goto err_pagealloc_init;
+
+	err = mlx5_adev_init(dev);
+	if (err)
+		goto err_adev_init;
+
+	err = mlx5_hca_caps_alloc(dev);
+	if (err)
+		goto err_hca_caps;
+
+	/* The conjunction of sw_vhca_id with sw_owner_id will be a global
+	 * unique id per function which uses mlx5_core.
+	 * Those values are supplied to FW as part of the init HCA command to
+	 * be used by both driver and FW when it's applicable.
+	 */
+	dev->priv.sw_vhca_id = ida_alloc_range(&sw_vhca_ida, 1,
+					       MAX_SW_VHCA_ID,
+					       GFP_KERNEL);
+	if (dev->priv.sw_vhca_id < 0)
+		mlx5_core_err(dev, "failed to allocate sw_vhca_id, err=%d\n",
+			      dev->priv.sw_vhca_id);
+
+	return 0;
+
+err_hca_caps:
+	mlx5_adev_cleanup(dev);
+err_adev_init:
+	mlx5_pagealloc_cleanup(dev);
+err_pagealloc_init:
+	mlx5_health_cleanup(dev);
+err_health_init:
+	mlx5_tout_cleanup(dev);
+err_timeout_init:
+	debugfs_remove(dev->priv.dbg.dbg_root);
+	mutex_destroy(&priv->pgdir_mutex);
+	mutex_destroy(&priv->alloc_mutex);
+	mutex_destroy(&priv->bfregs.wc_head.lock);
+	mutex_destroy(&priv->bfregs.reg_head.lock);
+	mutex_destroy(&dev->intf_state_mutex);
+	lockdep_unregister_key(&dev->lock_key);
+	return err;
+}
+
+void mlx5_mdev_uninit(struct mlx5_core_dev *dev)
+{
+	struct mlx5_priv *priv = &dev->priv;
+
+	if (priv->sw_vhca_id > 0)
+		ida_free(&sw_vhca_ida, dev->priv.sw_vhca_id);
+
+	mlx5_hca_caps_free(dev);
+	mlx5_adev_cleanup(dev);
+	mlx5_pagealloc_cleanup(dev);
+	mlx5_health_cleanup(dev);
+	mlx5_tout_cleanup(dev);
+	debugfs_remove_recursive(dev->priv.dbg.dbg_root);
+	mutex_destroy(&priv->pgdir_mutex);
+	mutex_destroy(&priv->alloc_mutex);
+	mutex_destroy(&priv->bfregs.wc_head.lock);
+	mutex_destroy(&priv->bfregs.reg_head.lock);
+	mutex_destroy(&dev->intf_state_mutex);
+	lockdep_unregister_key(&dev->lock_key);
+}
+
+static int probe_one(struct pci_dev *pdev, const struct pci_device_id *id)
+{
+	struct mlx5_core_dev *dev;
+	struct devlink *devlink;
+	int err;
+
+	devlink = mlx5_devlink_alloc(&pdev->dev);
+	if (!devlink) {
+		dev_err(&pdev->dev, "devlink alloc failed\n");
+		return -ENOMEM;
+	}
+
+	dev = devlink_priv(devlink);
+	dev->device = &pdev->dev;
+	dev->pdev = pdev;
+
+	dev->coredev_type = id->driver_data & MLX5_PCI_DEV_IS_VF ?
+			 MLX5_COREDEV_VF : MLX5_COREDEV_PF;
+
+	dev->priv.adev_idx = mlx5_adev_idx_alloc();
+	if (dev->priv.adev_idx < 0) {
+		err = dev->priv.adev_idx;
+		goto adev_init_err;
+	}
+
+	err = mlx5_mdev_init(dev, prof_sel);
+	if (err)
+		goto mdev_init_err;
+
+	err = mlx5_pci_init(dev, pdev, id);
+	if (err) {
+		mlx5_core_err(dev, "mlx5_pci_init failed with error code %d\n",
+			      err);
+		goto pci_init_err;
+	}
+
+	err = mlx5_init_one(dev);
+	if (err) {
+		mlx5_core_err(dev, "mlx5_init_one failed with error code %d\n",
+			      err);
+		goto err_init_one;
+	}
+
+	err = mlx5_crdump_enable(dev);
+	if (err)
+		dev_err(&pdev->dev, "mlx5_crdump_enable failed with error code %d\n", err);
+
+	pci_save_state(pdev);
+	devlink_register(devlink);
+	return 0;
+
+err_init_one:
+	mlx5_pci_close(dev);
+pci_init_err:
+	mlx5_mdev_uninit(dev);
+mdev_init_err:
+	mlx5_adev_idx_free(dev->priv.adev_idx);
+adev_init_err:
+	mlx5_devlink_free(devlink);
+
+	return err;
+}
+
+static void remove_one(struct pci_dev *pdev)
+{
+	struct mlx5_core_dev *dev  = pci_get_drvdata(pdev);
+	struct devlink *devlink = priv_to_devlink(dev);
+
+	set_bit(MLX5_BREAK_FW_WAIT, &dev->intf_state);
+	/* mlx5_drain_fw_reset() and mlx5_drain_health_wq() are using
+	 * devlink notify APIs.
+	 * Hence, we must drain them before unregistering the devlink.
+	 */
+	mlx5_drain_fw_reset(dev);
+	mlx5_drain_health_wq(dev);
+	devlink_unregister(devlink);
+	mlx5_sriov_disable(pdev);
+	mlx5_crdump_disable(dev);
+	mlx5_uninit_one(dev);
+	mlx5_pci_close(dev);
+	mlx5_mdev_uninit(dev);
+	mlx5_adev_idx_free(dev->priv.adev_idx);
+	mlx5_devlink_free(devlink);
+}
+
+#define mlx5_pci_trace(dev, fmt, ...) ({ \
+	struct mlx5_core_dev *__dev = (dev); \
+	mlx5_core_info(__dev, "%s Device state = %d health sensors: %d pci_status: %d. " fmt, \
+		       __func__, __dev->state, mlx5_health_check_fatal_sensors(__dev), \
+		       __dev->pci_status, ##__VA_ARGS__); \
+})
+
+static const char *result2str(enum pci_ers_result result)
+{
+	return  result == PCI_ERS_RESULT_NEED_RESET ? "need reset" :
+		result == PCI_ERS_RESULT_DISCONNECT ? "disconnect" :
+		result == PCI_ERS_RESULT_RECOVERED  ? "recovered" :
+		"unknown";
+}
+
+static pci_ers_result_t mlx5_pci_err_detected(struct pci_dev *pdev,
+					      pci_channel_state_t state)
+{
+	struct mlx5_core_dev *dev = pci_get_drvdata(pdev);
+	enum pci_ers_result res;
+
+	mlx5_pci_trace(dev, "Enter, pci channel state = %d\n", state);
+
+	mlx5_enter_error_state(dev, false);
+	mlx5_error_sw_reset(dev);
+	mlx5_unload_one(dev, false);
+	mlx5_drain_health_wq(dev);
+	mlx5_pci_disable_device(dev);
+
+	res = state == pci_channel_io_perm_failure ?
+		PCI_ERS_RESULT_DISCONNECT : PCI_ERS_RESULT_NEED_RESET;
+
+	mlx5_core_info(dev, "%s Device state = %d pci_status: %d. Exit, result = %d, %s\n",
+		       __func__, dev->state, dev->pci_status, res, result2str(res));
+	return res;
+}
+
+/* wait for the device to show vital signs by waiting
+ * for the health counter to start counting.
+ */
+static int wait_vital(struct pci_dev *pdev)
+{
+	struct mlx5_core_dev *dev = pci_get_drvdata(pdev);
+	struct mlx5_core_health *health = &dev->priv.health;
+	const int niter = 100;
+	u32 last_count = 0;
+	u32 count;
+	int i;
+
+	for (i = 0; i < niter; i++) {
+		count = ioread32be(health->health_counter);
+		if (count && count != 0xffffffff) {
+			if (last_count && last_count != count) {
+				mlx5_core_info(dev,
+					       "wait vital counter value 0x%x after %d iterations\n",
+					       count, i);
+				return 0;
+			}
+			last_count = count;
+		}
+		msleep(50);
+	}
+
+	return -ETIMEDOUT;
+}
+
+static pci_ers_result_t mlx5_pci_slot_reset(struct pci_dev *pdev)
+{
+	enum pci_ers_result res = PCI_ERS_RESULT_DISCONNECT;
+	struct mlx5_core_dev *dev = pci_get_drvdata(pdev);
+	int err;
+
+	mlx5_core_info(dev, "%s Device state = %d pci_status: %d. Enter\n",
+		       __func__, dev->state, dev->pci_status);
+
+	err = mlx5_pci_enable_device(dev);
+	if (err) {
+		mlx5_core_err(dev, "%s: mlx5_pci_enable_device failed with error code: %d\n",
+			      __func__, err);
+		goto out;
+	}
+
+	pci_set_master(pdev);
+	pci_restore_state(pdev);
+	pci_save_state(pdev);
+
+	err = wait_vital(pdev);
+	if (err) {
+		mlx5_core_err(dev, "%s: wait vital failed with error code: %d\n",
+			      __func__, err);
+		goto out;
+	}
+
+	res = PCI_ERS_RESULT_RECOVERED;
+out:
+	mlx5_core_info(dev, "%s Device state = %d pci_status: %d. Exit, err = %d, result = %d, %s\n",
+		       __func__, dev->state, dev->pci_status, err, res, result2str(res));
+	return res;
+}
+
+static void mlx5_pci_resume(struct pci_dev *pdev)
+{
+	struct mlx5_core_dev *dev = pci_get_drvdata(pdev);
+	int err;
+
+	mlx5_pci_trace(dev, "Enter, loading driver..\n");
+
+	err = mlx5_load_one(dev, false);
+
+	if (!err)
+		devlink_health_reporter_state_update(dev->priv.health.fw_fatal_reporter,
+						     DEVLINK_HEALTH_REPORTER_STATE_HEALTHY);
+
+	mlx5_pci_trace(dev, "Done, err = %d, device %s\n", err,
+		       !err ? "recovered" : "Failed");
+}
+
+static const struct pci_error_handlers mlx5_err_handler = {
+	.error_detected = mlx5_pci_err_detected,
+	.slot_reset	= mlx5_pci_slot_reset,
+	.resume		= mlx5_pci_resume
+};
+
+static int mlx5_try_fast_unload(struct mlx5_core_dev *dev)
+{
+	bool fast_teardown = false, force_teardown = false;
+	int ret = 1;
+
+	fast_teardown = MLX5_CAP_GEN(dev, fast_teardown);
+	force_teardown = MLX5_CAP_GEN(dev, force_teardown);
+
+	mlx5_core_dbg(dev, "force teardown firmware support=%d\n", force_teardown);
+	mlx5_core_dbg(dev, "fast teardown firmware support=%d\n", fast_teardown);
+
+	if (!fast_teardown && !force_teardown)
+		return -EOPNOTSUPP;
+
+	if (dev->state == MLX5_DEVICE_STATE_INTERNAL_ERROR) {
+		mlx5_core_dbg(dev, "Device in internal error state, giving up\n");
+		return -EAGAIN;
+	}
+
+	/* Panic tear down fw command will stop the PCI bus communication
+	 * with the HCA, so the health poll is no longer needed.
+	 */
+	mlx5_drain_health_wq(dev);
+	mlx5_stop_health_poll(dev, false);
+
+	ret = mlx5_cmd_fast_teardown_hca(dev);
+	if (!ret)
+		goto succeed;
+
+	ret = mlx5_cmd_force_teardown_hca(dev);
+	if (!ret)
+		goto succeed;
+
+	mlx5_core_dbg(dev, "Firmware couldn't do fast unload error: %d\n", ret);
+	mlx5_start_health_poll(dev);
+	return ret;
+
+succeed:
+	mlx5_enter_error_state(dev, true);
+
+	/* Some platforms requiring freeing the IRQ's in the shutdown
+	 * flow. If they aren't freed they can't be allocated after
+	 * kexec. There is no need to cleanup the mlx5_core software
+	 * contexts.
+	 */
+	mlx5_core_eq_free_irqs(dev);
+
+	return 0;
+}
+
+static void shutdown(struct pci_dev *pdev)
+{
+	struct mlx5_core_dev *dev  = pci_get_drvdata(pdev);
+	int err;
+
+	mlx5_core_info(dev, "Shutdown was called\n");
+	set_bit(MLX5_BREAK_FW_WAIT, &dev->intf_state);
+	err = mlx5_try_fast_unload(dev);
+	if (err)
+		mlx5_unload_one(dev, false);
+	mlx5_pci_disable_device(dev);
+}
+
+static int mlx5_suspend(struct pci_dev *pdev, pm_message_t state)
+{
+	struct mlx5_core_dev *dev = pci_get_drvdata(pdev);
+
+	mlx5_unload_one(dev, true);
+
+	return 0;
+}
+
+static int mlx5_resume(struct pci_dev *pdev)
+{
+	struct mlx5_core_dev *dev = pci_get_drvdata(pdev);
+
+	return mlx5_load_one(dev, false);
+}
+
+static const struct pci_device_id mlx5_core_pci_table[] = {
+	{ PCI_VDEVICE(MELLANOX, PCI_DEVICE_ID_MELLANOX_CONNECTIB) },
+	{ PCI_VDEVICE(MELLANOX, 0x1012), MLX5_PCI_DEV_IS_VF},	/* Connect-IB VF */
+	{ PCI_VDEVICE(MELLANOX, PCI_DEVICE_ID_MELLANOX_CONNECTX4) },
+	{ PCI_VDEVICE(MELLANOX, 0x1014), MLX5_PCI_DEV_IS_VF},	/* ConnectX-4 VF */
+	{ PCI_VDEVICE(MELLANOX, PCI_DEVICE_ID_MELLANOX_CONNECTX4_LX) },
+	{ PCI_VDEVICE(MELLANOX, 0x1016), MLX5_PCI_DEV_IS_VF},	/* ConnectX-4LX VF */
+	{ PCI_VDEVICE(MELLANOX, 0x1017) },			/* ConnectX-5, PCIe 3.0 */
+	{ PCI_VDEVICE(MELLANOX, 0x1018), MLX5_PCI_DEV_IS_VF},	/* ConnectX-5 VF */
+	{ PCI_VDEVICE(MELLANOX, 0x1019) },			/* ConnectX-5 Ex */
+	{ PCI_VDEVICE(MELLANOX, 0x101a), MLX5_PCI_DEV_IS_VF},	/* ConnectX-5 Ex VF */
+	{ PCI_VDEVICE(MELLANOX, 0x101b) },			/* ConnectX-6 */
+	{ PCI_VDEVICE(MELLANOX, 0x101c), MLX5_PCI_DEV_IS_VF},	/* ConnectX-6 VF */
+	{ PCI_VDEVICE(MELLANOX, 0x101d) },			/* ConnectX-6 Dx */
+	{ PCI_VDEVICE(MELLANOX, 0x101e), MLX5_PCI_DEV_IS_VF},	/* ConnectX Family mlx5Gen Virtual Function */
+	{ PCI_VDEVICE(MELLANOX, 0x101f) },			/* ConnectX-6 LX */
+	{ PCI_VDEVICE(MELLANOX, 0x1021) },			/* ConnectX-7 */
+	{ PCI_VDEVICE(MELLANOX, 0x1023) },			/* ConnectX-8 */
+	{ PCI_VDEVICE(MELLANOX, 0xa2d2) },			/* BlueField integrated ConnectX-5 network controller */
+	{ PCI_VDEVICE(MELLANOX, 0xa2d3), MLX5_PCI_DEV_IS_VF},	/* BlueField integrated ConnectX-5 network controller VF */
+	{ PCI_VDEVICE(MELLANOX, 0xa2d6) },			/* BlueField-2 integrated ConnectX-6 Dx network controller */
+	{ PCI_VDEVICE(MELLANOX, 0xa2dc) },			/* BlueField-3 integrated ConnectX-7 network controller */
+	{ PCI_VDEVICE(MELLANOX, 0xa2df) },			/* BlueField-4 integrated ConnectX-8 network controller */
+	{ 0, }
+};
+
+MODULE_DEVICE_TABLE(pci, mlx5_core_pci_table);
+
+void mlx5_disable_device(struct mlx5_core_dev *dev)
+{
+	mlx5_error_sw_reset(dev);
+	mlx5_unload_one_devl_locked(dev, false);
+}
+
+int mlx5_recover_device(struct mlx5_core_dev *dev)
+{
+	if (!mlx5_core_is_sf(dev)) {
+		mlx5_pci_disable_device(dev);
+		if (mlx5_pci_slot_reset(dev->pdev) != PCI_ERS_RESULT_RECOVERED)
+			return -EIO;
+	}
+
+	return mlx5_load_one_devl_locked(dev, true);
+}
+
+static struct pci_driver mlx5_core_driver = {
+	.name           = KBUILD_MODNAME,
+	.id_table       = mlx5_core_pci_table,
+	.probe          = probe_one,
+	.remove         = remove_one,
+	.suspend        = mlx5_suspend,
+	.resume         = mlx5_resume,
+	.shutdown	= shutdown,
+	.err_handler	= &mlx5_err_handler,
+	.sriov_configure   = mlx5_core_sriov_configure,
+	.sriov_get_vf_total_msix = mlx5_sriov_get_vf_total_msix,
+	.sriov_set_msix_vec_count = mlx5_core_sriov_set_msix_vec_count,
+};
+
+/**
+ * mlx5_vf_get_core_dev - Get the mlx5 core device from a given VF PCI device if
+ *                     mlx5_core is its driver.
+ * @pdev: The associated PCI device.
+ *
+ * Upon return the interface state lock stay held to let caller uses it safely.
+ * Caller must ensure to use the returned mlx5 device for a narrow window
+ * and put it back with mlx5_vf_put_core_dev() immediately once usage was over.
+ *
+ * Return: Pointer to the associated mlx5_core_dev or NULL.
+ */
+struct mlx5_core_dev *mlx5_vf_get_core_dev(struct pci_dev *pdev)
+{
+	struct mlx5_core_dev *mdev;
+
+	mdev = pci_iov_get_pf_drvdata(pdev, &mlx5_core_driver);
+	if (IS_ERR(mdev))
+		return NULL;
+
+	mutex_lock(&mdev->intf_state_mutex);
+	if (!test_bit(MLX5_INTERFACE_STATE_UP, &mdev->intf_state)) {
+		mutex_unlock(&mdev->intf_state_mutex);
+		return NULL;
+	}
+
+	return mdev;
+}
+EXPORT_SYMBOL(mlx5_vf_get_core_dev);
+
+/**
+ * mlx5_vf_put_core_dev - Put the mlx5 core device back.
+ * @mdev: The mlx5 core device.
+ *
+ * Upon return the interface state lock is unlocked and caller should not
+ * access the mdev any more.
+ */
+void mlx5_vf_put_core_dev(struct mlx5_core_dev *mdev)
+{
+	mutex_unlock(&mdev->intf_state_mutex);
+}
+EXPORT_SYMBOL(mlx5_vf_put_core_dev);
+
+static void mlx5_core_verify_params(void)
+{
+	if (prof_sel >= ARRAY_SIZE(profile)) {
+		pr_warn("mlx5_core: WARNING: Invalid module parameter prof_sel %d, valid range 0-%zu, changing back to default(%d)\n",
+			prof_sel,
+			ARRAY_SIZE(profile) - 1,
+			MLX5_DEFAULT_PROF);
+		prof_sel = MLX5_DEFAULT_PROF;
+	}
+}
+
+static int __init mlx5_init(void)
+{
+	int err;
+
+	WARN_ONCE(strcmp(MLX5_ADEV_NAME, KBUILD_MODNAME),
+		  "mlx5_core name not in sync with kernel module name");
+
+	get_random_bytes(&sw_owner_id, sizeof(sw_owner_id));
+
+	mlx5_core_verify_params();
+	mlx5_register_debugfs();
+
+	err = mlx5e_init();
+	if (err)
+		goto err_debug;
+
+	err = mlx5_sf_driver_register();
+	if (err)
+		goto err_sf;
+
+	err = pci_register_driver(&mlx5_core_driver);
+	if (err)
+		goto err_pci;
+
+	return 0;
+
+err_pci:
+	mlx5_sf_driver_unregister();
+err_sf:
+	mlx5e_cleanup();
+err_debug:
+	mlx5_unregister_debugfs();
+	return err;
+}
+
+static void __exit mlx5_cleanup(void)
+{
+	pci_unregister_driver(&mlx5_core_driver);
+	mlx5_sf_driver_unregister();
+	mlx5e_cleanup();
+	mlx5_unregister_debugfs();
+}
+
+module_init(mlx5_init);
+module_exit(mlx5_cleanup);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/mcg.c b/drivers/net/ethernet/mellanox/mlx5/core/mcg.c
new file mode 100644
index 000000000..495cca58d
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/mcg.c
@@ -0,0 +1,62 @@
+/*
+ * Copyright (c) 2013-2015, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/kernel.h>
+#include <linux/mlx5/driver.h>
+#include <rdma/ib_verbs.h>
+#include "mlx5_core.h"
+
+int mlx5_core_attach_mcg(struct mlx5_core_dev *dev, union ib_gid *mgid, u32 qpn)
+{
+	u32 in[MLX5_ST_SZ_DW(attach_to_mcg_in)] = {};
+	void *gid;
+
+	MLX5_SET(attach_to_mcg_in, in, opcode, MLX5_CMD_OP_ATTACH_TO_MCG);
+	MLX5_SET(attach_to_mcg_in, in, qpn, qpn);
+	gid = MLX5_ADDR_OF(attach_to_mcg_in, in, multicast_gid);
+	memcpy(gid, mgid, sizeof(*mgid));
+	return mlx5_cmd_exec_in(dev, attach_to_mcg, in);
+}
+EXPORT_SYMBOL(mlx5_core_attach_mcg);
+
+int mlx5_core_detach_mcg(struct mlx5_core_dev *dev, union ib_gid *mgid, u32 qpn)
+{
+	u32 in[MLX5_ST_SZ_DW(detach_from_mcg_in)] = {};
+	void *gid;
+
+	MLX5_SET(detach_from_mcg_in, in, opcode, MLX5_CMD_OP_DETACH_FROM_MCG);
+	MLX5_SET(detach_from_mcg_in, in, qpn, qpn);
+	gid = MLX5_ADDR_OF(detach_from_mcg_in, in, multicast_gid);
+	memcpy(gid, mgid, sizeof(*mgid));
+	return mlx5_cmd_exec_in(dev, detach_from_mcg, in);
+}
+EXPORT_SYMBOL(mlx5_core_detach_mcg);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/mlx5_core.h b/drivers/net/ethernet/mellanox/mlx5/core/mlx5_core.h
new file mode 100644
index 000000000..0b560e97a
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/mlx5_core.h
@@ -0,0 +1,330 @@
+/*
+ * Copyright (c) 2013-2015, Mellanox Technologies, Ltd.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef __MLX5_CORE_H__
+#define __MLX5_CORE_H__
+
+#include <linux/types.h>
+#include <linux/kernel.h>
+#include <linux/sched.h>
+#include <linux/if_link.h>
+#include <linux/firmware.h>
+#include <linux/mlx5/cq.h>
+#include <linux/mlx5/fs.h>
+#include <linux/mlx5/driver.h>
+
+extern uint mlx5_core_debug_mask;
+
+#define mlx5_core_dbg(__dev, format, ...)				\
+	dev_dbg((__dev)->device, "%s:%d:(pid %d): " format,		\
+		 __func__, __LINE__, current->pid,			\
+		 ##__VA_ARGS__)
+
+#define mlx5_core_dbg_once(__dev, format, ...)		\
+	dev_dbg_once((__dev)->device,		\
+		     "%s:%d:(pid %d): " format,		\
+		     __func__, __LINE__, current->pid,	\
+		     ##__VA_ARGS__)
+
+#define mlx5_core_dbg_mask(__dev, mask, format, ...)		\
+do {								\
+	if ((mask) & mlx5_core_debug_mask)			\
+		mlx5_core_dbg(__dev, format, ##__VA_ARGS__);	\
+} while (0)
+
+#define mlx5_core_err(__dev, format, ...)			\
+	dev_err((__dev)->device, "%s:%d:(pid %d): " format,	\
+		__func__, __LINE__, current->pid,		\
+	       ##__VA_ARGS__)
+
+#define mlx5_core_err_rl(__dev, format, ...)			\
+	dev_err_ratelimited((__dev)->device,			\
+			    "%s:%d:(pid %d): " format,		\
+			    __func__, __LINE__, current->pid,	\
+			    ##__VA_ARGS__)
+
+#define mlx5_core_warn(__dev, format, ...)			\
+	dev_warn((__dev)->device, "%s:%d:(pid %d): " format,	\
+		 __func__, __LINE__, current->pid,		\
+		 ##__VA_ARGS__)
+
+#define mlx5_core_warn_once(__dev, format, ...)				\
+	dev_warn_once((__dev)->device, "%s:%d:(pid %d): " format,	\
+		      __func__, __LINE__, current->pid,			\
+		      ##__VA_ARGS__)
+
+#define mlx5_core_warn_rl(__dev, format, ...)			\
+	dev_warn_ratelimited((__dev)->device,			\
+			     "%s:%d:(pid %d): " format,		\
+			     __func__, __LINE__, current->pid,	\
+			     ##__VA_ARGS__)
+
+#define mlx5_core_info(__dev, format, ...)		\
+	dev_info((__dev)->device, format, ##__VA_ARGS__)
+
+#define mlx5_core_info_rl(__dev, format, ...)			\
+	dev_info_ratelimited((__dev)->device,			\
+			     "%s:%d:(pid %d): " format,		\
+			     __func__, __LINE__, current->pid,	\
+			     ##__VA_ARGS__)
+
+static inline void mlx5_printk(struct mlx5_core_dev *dev, int level, const char *format, ...)
+{
+	struct device *device = dev->device;
+	struct va_format vaf;
+	va_list args;
+
+	if (WARN_ONCE(level < LOGLEVEL_EMERG || level > LOGLEVEL_DEBUG,
+		      "Level %d is out of range, set to default level\n", level))
+		level = LOGLEVEL_DEFAULT;
+
+	va_start(args, format);
+	vaf.fmt = format;
+	vaf.va = &args;
+
+	dev_printk_emit(level, device, "%s %s: %pV", dev_driver_string(device), dev_name(device),
+			&vaf);
+	va_end(args);
+}
+
+#define mlx5_log(__dev, level, format, ...)			\
+	mlx5_printk(__dev, level, "%s:%d:(pid %d): " format,	\
+		    __func__, __LINE__, current->pid,		\
+		    ##__VA_ARGS__)
+
+static inline struct device *mlx5_core_dma_dev(struct mlx5_core_dev *dev)
+{
+	return &dev->pdev->dev;
+}
+
+enum {
+	MLX5_CMD_DATA, /* print command payload only */
+	MLX5_CMD_TIME, /* print command execution time */
+};
+
+enum {
+	MLX5_DRIVER_STATUS_ABORTED = 0xfe,
+	MLX5_DRIVER_SYND = 0xbadd00de,
+};
+
+enum mlx5_semaphore_space_address {
+	MLX5_SEMAPHORE_SPACE_DOMAIN     = 0xA,
+	MLX5_SEMAPHORE_SW_RESET         = 0x20,
+};
+
+#define MLX5_DEFAULT_PROF       2
+
+static inline int mlx5_flexible_inlen(struct mlx5_core_dev *dev, size_t fixed,
+				      size_t item_size, size_t num_items,
+				      const char *func, int line)
+{
+	int inlen;
+
+	if (fixed > INT_MAX || item_size > INT_MAX || num_items > INT_MAX) {
+		mlx5_core_err(dev, "%s: %s:%d: input values too big: %zu + %zu * %zu\n",
+			      __func__, func, line, fixed, item_size, num_items);
+		return -ENOMEM;
+	}
+
+	if (check_mul_overflow((int)item_size, (int)num_items, &inlen)) {
+		mlx5_core_err(dev, "%s: %s:%d: multiplication overflow: %zu + %zu * %zu\n",
+			      __func__, func, line, fixed, item_size, num_items);
+		return -ENOMEM;
+	}
+
+	if (check_add_overflow((int)fixed, inlen, &inlen)) {
+		mlx5_core_err(dev, "%s: %s:%d: addition overflow: %zu + %zu * %zu\n",
+			      __func__, func, line, fixed, item_size, num_items);
+		return -ENOMEM;
+	}
+
+	return inlen;
+}
+
+#define MLX5_FLEXIBLE_INLEN(dev, fixed, item_size, num_items) \
+	mlx5_flexible_inlen(dev, fixed, item_size, num_items, __func__, __LINE__)
+
+int mlx5_query_hca_caps(struct mlx5_core_dev *dev);
+int mlx5_query_board_id(struct mlx5_core_dev *dev);
+int mlx5_cmd_init(struct mlx5_core_dev *dev);
+void mlx5_cmd_cleanup(struct mlx5_core_dev *dev);
+void mlx5_cmd_set_state(struct mlx5_core_dev *dev,
+			enum mlx5_cmdif_state cmdif_state);
+int mlx5_cmd_init_hca(struct mlx5_core_dev *dev, uint32_t *sw_owner_id);
+int mlx5_cmd_teardown_hca(struct mlx5_core_dev *dev);
+int mlx5_cmd_force_teardown_hca(struct mlx5_core_dev *dev);
+int mlx5_cmd_fast_teardown_hca(struct mlx5_core_dev *dev);
+void mlx5_enter_error_state(struct mlx5_core_dev *dev, bool force);
+void mlx5_error_sw_reset(struct mlx5_core_dev *dev);
+u32 mlx5_health_check_fatal_sensors(struct mlx5_core_dev *dev);
+int mlx5_health_wait_pci_up(struct mlx5_core_dev *dev);
+void mlx5_disable_device(struct mlx5_core_dev *dev);
+int mlx5_recover_device(struct mlx5_core_dev *dev);
+int mlx5_sriov_init(struct mlx5_core_dev *dev);
+void mlx5_sriov_cleanup(struct mlx5_core_dev *dev);
+int mlx5_sriov_attach(struct mlx5_core_dev *dev);
+void mlx5_sriov_detach(struct mlx5_core_dev *dev);
+int mlx5_core_sriov_configure(struct pci_dev *dev, int num_vfs);
+void mlx5_sriov_disable(struct pci_dev *pdev);
+int mlx5_core_sriov_set_msix_vec_count(struct pci_dev *vf, int msix_vec_count);
+int mlx5_core_enable_hca(struct mlx5_core_dev *dev, u16 func_id);
+int mlx5_core_disable_hca(struct mlx5_core_dev *dev, u16 func_id);
+int mlx5_create_scheduling_element_cmd(struct mlx5_core_dev *dev, u8 hierarchy,
+				       void *context, u32 *element_id);
+int mlx5_modify_scheduling_element_cmd(struct mlx5_core_dev *dev, u8 hierarchy,
+				       void *context, u32 element_id,
+				       u32 modify_bitmask);
+int mlx5_destroy_scheduling_element_cmd(struct mlx5_core_dev *dev, u8 hierarchy,
+					u32 element_id);
+int mlx5_wait_for_pages(struct mlx5_core_dev *dev, int *pages);
+
+void mlx5_cmd_flush(struct mlx5_core_dev *dev);
+void mlx5_cq_debugfs_init(struct mlx5_core_dev *dev);
+void mlx5_cq_debugfs_cleanup(struct mlx5_core_dev *dev);
+
+int mlx5_query_pcam_reg(struct mlx5_core_dev *dev, u32 *pcam, u8 feature_group,
+			u8 access_reg_group);
+int mlx5_query_mcam_reg(struct mlx5_core_dev *dev, u32 *mcap, u8 feature_group,
+			u8 access_reg_group);
+int mlx5_query_qcam_reg(struct mlx5_core_dev *mdev, u32 *qcam,
+			u8 feature_group, u8 access_reg_group);
+
+void mlx5_lag_add_netdev(struct mlx5_core_dev *dev, struct net_device *netdev);
+void mlx5_lag_remove_netdev(struct mlx5_core_dev *dev, struct net_device *netdev);
+void mlx5_lag_add_mdev(struct mlx5_core_dev *dev);
+void mlx5_lag_remove_mdev(struct mlx5_core_dev *dev);
+void mlx5_lag_disable_change(struct mlx5_core_dev *dev);
+void mlx5_lag_enable_change(struct mlx5_core_dev *dev);
+
+int mlx5_events_init(struct mlx5_core_dev *dev);
+void mlx5_events_cleanup(struct mlx5_core_dev *dev);
+void mlx5_events_start(struct mlx5_core_dev *dev);
+void mlx5_events_stop(struct mlx5_core_dev *dev);
+
+int mlx5_adev_idx_alloc(void);
+void mlx5_adev_idx_free(int idx);
+void mlx5_adev_cleanup(struct mlx5_core_dev *dev);
+int mlx5_adev_init(struct mlx5_core_dev *dev);
+
+int mlx5_attach_device(struct mlx5_core_dev *dev);
+void mlx5_detach_device(struct mlx5_core_dev *dev, bool suspend);
+int mlx5_register_device(struct mlx5_core_dev *dev);
+void mlx5_unregister_device(struct mlx5_core_dev *dev);
+struct mlx5_core_dev *mlx5_get_next_phys_dev_lag(struct mlx5_core_dev *dev);
+void mlx5_dev_list_lock(void);
+void mlx5_dev_list_unlock(void);
+int mlx5_dev_list_trylock(void);
+
+int mlx5_query_mtpps(struct mlx5_core_dev *dev, u32 *mtpps, u32 mtpps_size);
+int mlx5_set_mtpps(struct mlx5_core_dev *mdev, u32 *mtpps, u32 mtpps_size);
+int mlx5_query_mtppse(struct mlx5_core_dev *mdev, u8 pin, u8 *arm, u8 *mode);
+int mlx5_set_mtppse(struct mlx5_core_dev *mdev, u8 pin, u8 arm, u8 mode);
+
+struct mlx5_dm *mlx5_dm_create(struct mlx5_core_dev *dev);
+void mlx5_dm_cleanup(struct mlx5_core_dev *dev);
+
+#define MLX5_PPS_CAP(mdev) (MLX5_CAP_GEN((mdev), pps) &&		\
+			    MLX5_CAP_GEN((mdev), pps_modify) &&		\
+			    MLX5_CAP_MCAM_FEATURE((mdev), mtpps_fs) &&	\
+			    MLX5_CAP_MCAM_FEATURE((mdev), mtpps_enh_out_per_adj))
+
+int mlx5_firmware_flash(struct mlx5_core_dev *dev, const struct firmware *fw,
+			struct netlink_ext_ack *extack);
+int mlx5_fw_version_query(struct mlx5_core_dev *dev,
+			  u32 *running_ver, u32 *stored_ver);
+
+#ifdef CONFIG_MLX5_CORE_EN
+int mlx5e_init(void);
+void mlx5e_cleanup(void);
+#else
+static inline int mlx5e_init(void){ return 0; }
+static inline void mlx5e_cleanup(void){}
+#endif
+
+static inline bool mlx5_sriov_is_enabled(struct mlx5_core_dev *dev)
+{
+	return pci_num_vf(dev->pdev) ? true : false;
+}
+
+int mlx5_rescan_drivers_locked(struct mlx5_core_dev *dev);
+static inline int mlx5_rescan_drivers(struct mlx5_core_dev *dev)
+{
+	int ret;
+
+	mlx5_dev_list_lock();
+	ret = mlx5_rescan_drivers_locked(dev);
+	mlx5_dev_list_unlock();
+	return ret;
+}
+
+void mlx5_lag_update(struct mlx5_core_dev *dev);
+
+enum {
+	MLX5_NIC_IFC_FULL		= 0,
+	MLX5_NIC_IFC_DISABLED		= 1,
+	MLX5_NIC_IFC_NO_DRAM_NIC	= 2,
+	MLX5_NIC_IFC_SW_RESET		= 7
+};
+
+u8 mlx5_get_nic_state(struct mlx5_core_dev *dev);
+void mlx5_set_nic_state(struct mlx5_core_dev *dev, u8 state);
+
+static inline bool mlx5_core_is_sf(const struct mlx5_core_dev *dev)
+{
+	return dev->coredev_type == MLX5_COREDEV_SF;
+}
+
+int mlx5_mdev_init(struct mlx5_core_dev *dev, int profile_idx);
+void mlx5_mdev_uninit(struct mlx5_core_dev *dev);
+int mlx5_init_one(struct mlx5_core_dev *dev);
+void mlx5_uninit_one(struct mlx5_core_dev *dev);
+void mlx5_unload_one(struct mlx5_core_dev *dev, bool suspend);
+void mlx5_unload_one_devl_locked(struct mlx5_core_dev *dev, bool suspend);
+int mlx5_load_one(struct mlx5_core_dev *dev, bool recovery);
+int mlx5_load_one_devl_locked(struct mlx5_core_dev *dev, bool recovery);
+
+int mlx5_vport_get_other_func_cap(struct mlx5_core_dev *dev, u16 function_id, void *out);
+
+void mlx5_events_work_enqueue(struct mlx5_core_dev *dev, struct work_struct *work);
+static inline u32 mlx5_sriov_get_vf_total_msix(struct pci_dev *pdev)
+{
+	struct mlx5_core_dev *dev = pci_get_drvdata(pdev);
+
+	return MLX5_CAP_GEN_MAX(dev, num_total_dynamic_vf_msix);
+}
+
+bool mlx5_eth_supported(struct mlx5_core_dev *dev);
+bool mlx5_rdma_supported(struct mlx5_core_dev *dev);
+bool mlx5_vnet_supported(struct mlx5_core_dev *dev);
+bool mlx5_same_hw_devs(struct mlx5_core_dev *dev, struct mlx5_core_dev *peer_dev);
+
+#endif /* __MLX5_CORE_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/mlx5_irq.h b/drivers/net/ethernet/mellanox/mlx5/core/mlx5_irq.h
new file mode 100644
index 000000000..2e728e4e8
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/mlx5_irq.h
@@ -0,0 +1,62 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2021 Mellanox Technologies. */
+
+#ifndef __MLX5_IRQ_H__
+#define __MLX5_IRQ_H__
+
+#include <linux/mlx5/driver.h>
+
+#define MLX5_COMP_EQS_PER_SF 8
+
+struct mlx5_irq;
+
+int mlx5_irq_table_init(struct mlx5_core_dev *dev);
+void mlx5_irq_table_cleanup(struct mlx5_core_dev *dev);
+int mlx5_irq_table_create(struct mlx5_core_dev *dev);
+void mlx5_irq_table_destroy(struct mlx5_core_dev *dev);
+void mlx5_irq_table_free_irqs(struct mlx5_core_dev *dev);
+int mlx5_irq_table_get_num_comp(struct mlx5_irq_table *table);
+int mlx5_irq_table_get_sfs_vec(struct mlx5_irq_table *table);
+struct mlx5_irq_table *mlx5_irq_table_get(struct mlx5_core_dev *dev);
+
+int mlx5_set_msix_vec_count(struct mlx5_core_dev *dev, int devfn,
+			    int msix_vec_count);
+int mlx5_get_default_msix_vec_count(struct mlx5_core_dev *dev, int num_vfs);
+
+struct mlx5_irq *mlx5_ctrl_irq_request(struct mlx5_core_dev *dev);
+void mlx5_ctrl_irq_release(struct mlx5_irq *ctrl_irq);
+struct mlx5_irq *mlx5_irq_request(struct mlx5_core_dev *dev, u16 vecidx,
+				  struct cpumask *affinity);
+int mlx5_irqs_request_vectors(struct mlx5_core_dev *dev, u16 *cpus, int nirqs,
+			      struct mlx5_irq **irqs);
+void mlx5_irqs_release_vectors(struct mlx5_irq **irqs, int nirqs);
+int mlx5_irq_attach_nb(struct mlx5_irq *irq, struct notifier_block *nb);
+int mlx5_irq_detach_nb(struct mlx5_irq *irq, struct notifier_block *nb);
+struct cpumask *mlx5_irq_get_affinity_mask(struct mlx5_irq *irq);
+int mlx5_irq_get_index(struct mlx5_irq *irq);
+
+struct mlx5_irq_pool;
+#ifdef CONFIG_MLX5_SF
+int mlx5_irq_affinity_irqs_request_auto(struct mlx5_core_dev *dev, int nirqs,
+					struct mlx5_irq **irqs);
+struct mlx5_irq *mlx5_irq_affinity_request(struct mlx5_irq_pool *pool,
+					   const struct cpumask *req_mask);
+void mlx5_irq_affinity_irqs_release(struct mlx5_core_dev *dev, struct mlx5_irq **irqs,
+				    int num_irqs);
+#else
+static inline int mlx5_irq_affinity_irqs_request_auto(struct mlx5_core_dev *dev, int nirqs,
+						      struct mlx5_irq **irqs)
+{
+	return -EOPNOTSUPP;
+}
+
+static inline struct mlx5_irq *
+mlx5_irq_affinity_request(struct mlx5_irq_pool *pool, const struct cpumask *req_mask)
+{
+	return ERR_PTR(-EOPNOTSUPP);
+}
+
+static inline void mlx5_irq_affinity_irqs_release(struct mlx5_core_dev *dev,
+						  struct mlx5_irq **irqs, int num_irqs) {}
+#endif
+#endif /* __MLX5_IRQ_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/mr.c b/drivers/net/ethernet/mellanox/mlx5/core/mr.c
new file mode 100644
index 000000000..9d735c343
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/mr.c
@@ -0,0 +1,124 @@
+/*
+ * Copyright (c) 2013-2015, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/kernel.h>
+#include <linux/mlx5/driver.h>
+#include "mlx5_core.h"
+
+int mlx5_core_create_mkey(struct mlx5_core_dev *dev, u32 *mkey, u32 *in,
+			  int inlen)
+{
+	u32 lout[MLX5_ST_SZ_DW(create_mkey_out)] = {};
+	u32 mkey_index;
+	int err;
+
+	MLX5_SET(create_mkey_in, in, opcode, MLX5_CMD_OP_CREATE_MKEY);
+
+	err = mlx5_cmd_exec(dev, in, inlen, lout, sizeof(lout));
+	if (err)
+		return err;
+
+	mkey_index = MLX5_GET(create_mkey_out, lout, mkey_index);
+	*mkey = MLX5_GET(create_mkey_in, in, memory_key_mkey_entry.mkey_7_0) |
+		mlx5_idx_to_mkey(mkey_index);
+
+	mlx5_core_dbg(dev, "out 0x%x, mkey 0x%x\n", mkey_index, *mkey);
+	return 0;
+}
+EXPORT_SYMBOL(mlx5_core_create_mkey);
+
+int mlx5_core_destroy_mkey(struct mlx5_core_dev *dev, u32 mkey)
+{
+	u32 in[MLX5_ST_SZ_DW(destroy_mkey_in)] = {};
+
+	MLX5_SET(destroy_mkey_in, in, opcode, MLX5_CMD_OP_DESTROY_MKEY);
+	MLX5_SET(destroy_mkey_in, in, mkey_index, mlx5_mkey_to_idx(mkey));
+	return mlx5_cmd_exec_in(dev, destroy_mkey, in);
+}
+EXPORT_SYMBOL(mlx5_core_destroy_mkey);
+
+int mlx5_core_query_mkey(struct mlx5_core_dev *dev, u32 mkey, u32 *out,
+			 int outlen)
+{
+	u32 in[MLX5_ST_SZ_DW(query_mkey_in)] = {};
+
+	memset(out, 0, outlen);
+	MLX5_SET(query_mkey_in, in, opcode, MLX5_CMD_OP_QUERY_MKEY);
+	MLX5_SET(query_mkey_in, in, mkey_index, mlx5_mkey_to_idx(mkey));
+	return mlx5_cmd_exec(dev, in, sizeof(in), out, outlen);
+}
+EXPORT_SYMBOL(mlx5_core_query_mkey);
+
+static inline u32 mlx5_get_psv(u32 *out, int psv_index)
+{
+	switch (psv_index) {
+	case 1: return MLX5_GET(create_psv_out, out, psv1_index);
+	case 2: return MLX5_GET(create_psv_out, out, psv2_index);
+	case 3: return MLX5_GET(create_psv_out, out, psv3_index);
+	default: return MLX5_GET(create_psv_out, out, psv0_index);
+	}
+}
+
+int mlx5_core_create_psv(struct mlx5_core_dev *dev, u32 pdn,
+			 int npsvs, u32 *sig_index)
+{
+	u32 out[MLX5_ST_SZ_DW(create_psv_out)] = {};
+	u32 in[MLX5_ST_SZ_DW(create_psv_in)] = {};
+	int i, err;
+
+	if (npsvs > MLX5_MAX_PSVS)
+		return -EINVAL;
+
+	MLX5_SET(create_psv_in, in, opcode, MLX5_CMD_OP_CREATE_PSV);
+	MLX5_SET(create_psv_in, in, pd, pdn);
+	MLX5_SET(create_psv_in, in, num_psv, npsvs);
+
+	err = mlx5_cmd_exec_inout(dev, create_psv, in, out);
+	if (err)
+		return err;
+
+	for (i = 0; i < npsvs; i++)
+		sig_index[i] = mlx5_get_psv(out, i);
+
+	return err;
+}
+EXPORT_SYMBOL(mlx5_core_create_psv);
+
+int mlx5_core_destroy_psv(struct mlx5_core_dev *dev, int psv_num)
+{
+	u32 in[MLX5_ST_SZ_DW(destroy_psv_in)] = {};
+
+	MLX5_SET(destroy_psv_in, in, opcode, MLX5_CMD_OP_DESTROY_PSV);
+	MLX5_SET(destroy_psv_in, in, psvn, psv_num);
+	return mlx5_cmd_exec_in(dev, destroy_psv, in);
+}
+EXPORT_SYMBOL(mlx5_core_destroy_psv);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/pagealloc.c b/drivers/net/ethernet/mellanox/mlx5/core/pagealloc.c
new file mode 100644
index 000000000..95dc67fb3
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/pagealloc.c
@@ -0,0 +1,796 @@
+/*
+ * Copyright (c) 2013-2015, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/highmem.h>
+#include <linux/kernel.h>
+#include <linux/delay.h>
+#include <linux/mlx5/driver.h>
+#include <linux/xarray.h>
+#include "mlx5_core.h"
+#include "lib/eq.h"
+#include "lib/tout.h"
+
+enum {
+	MLX5_PAGES_CANT_GIVE	= 0,
+	MLX5_PAGES_GIVE		= 1,
+	MLX5_PAGES_TAKE		= 2
+};
+
+struct mlx5_pages_req {
+	struct mlx5_core_dev *dev;
+	u16	func_id;
+	u8	ec_function;
+	s32	npages;
+	struct work_struct work;
+	u8	release_all;
+};
+
+struct fw_page {
+	struct rb_node		rb_node;
+	u64			addr;
+	struct page	       *page;
+	u32			function;
+	unsigned long		bitmask;
+	struct list_head	list;
+	unsigned int free_count;
+};
+
+enum {
+	MLX5_MAX_RECLAIM_TIME_MILI	= 5000,
+	MLX5_NUM_4K_IN_PAGE		= PAGE_SIZE / MLX5_ADAPTER_PAGE_SIZE,
+};
+
+static u32 get_function(u16 func_id, bool ec_function)
+{
+	return (u32)func_id | (ec_function << 16);
+}
+
+static u16 func_id_to_type(struct mlx5_core_dev *dev, u16 func_id, bool ec_function)
+{
+	if (!func_id)
+		return mlx5_core_is_ecpf(dev) && !ec_function ? MLX5_HOST_PF : MLX5_PF;
+
+	return func_id <= mlx5_core_max_vfs(dev) ?  MLX5_VF : MLX5_SF;
+}
+
+static u32 mlx5_get_ec_function(u32 function)
+{
+	return function >> 16;
+}
+
+static u32 mlx5_get_func_id(u32 function)
+{
+	return function & 0xffff;
+}
+
+static struct rb_root *page_root_per_function(struct mlx5_core_dev *dev, u32 function)
+{
+	struct rb_root *root;
+	int err;
+
+	root = xa_load(&dev->priv.page_root_xa, function);
+	if (root)
+		return root;
+
+	root = kzalloc(sizeof(*root), GFP_KERNEL);
+	if (!root)
+		return ERR_PTR(-ENOMEM);
+
+	err = xa_insert(&dev->priv.page_root_xa, function, root, GFP_KERNEL);
+	if (err) {
+		kfree(root);
+		return ERR_PTR(err);
+	}
+
+	*root = RB_ROOT;
+
+	return root;
+}
+
+static int insert_page(struct mlx5_core_dev *dev, u64 addr, struct page *page, u32 function)
+{
+	struct rb_node *parent = NULL;
+	struct rb_root *root;
+	struct rb_node **new;
+	struct fw_page *nfp;
+	struct fw_page *tfp;
+	int i;
+
+	root = page_root_per_function(dev, function);
+	if (IS_ERR(root))
+		return PTR_ERR(root);
+
+	new = &root->rb_node;
+
+	while (*new) {
+		parent = *new;
+		tfp = rb_entry(parent, struct fw_page, rb_node);
+		if (tfp->addr < addr)
+			new = &parent->rb_left;
+		else if (tfp->addr > addr)
+			new = &parent->rb_right;
+		else
+			return -EEXIST;
+	}
+
+	nfp = kzalloc(sizeof(*nfp), GFP_KERNEL);
+	if (!nfp)
+		return -ENOMEM;
+
+	nfp->addr = addr;
+	nfp->page = page;
+	nfp->function = function;
+	nfp->free_count = MLX5_NUM_4K_IN_PAGE;
+	for (i = 0; i < MLX5_NUM_4K_IN_PAGE; i++)
+		set_bit(i, &nfp->bitmask);
+
+	rb_link_node(&nfp->rb_node, parent, new);
+	rb_insert_color(&nfp->rb_node, root);
+	list_add(&nfp->list, &dev->priv.free_list);
+
+	return 0;
+}
+
+static struct fw_page *find_fw_page(struct mlx5_core_dev *dev, u64 addr,
+				    u32 function)
+{
+	struct fw_page *result = NULL;
+	struct rb_root *root;
+	struct rb_node *tmp;
+	struct fw_page *tfp;
+
+	root = xa_load(&dev->priv.page_root_xa, function);
+	if (WARN_ON_ONCE(!root))
+		return NULL;
+
+	tmp = root->rb_node;
+
+	while (tmp) {
+		tfp = rb_entry(tmp, struct fw_page, rb_node);
+		if (tfp->addr < addr) {
+			tmp = tmp->rb_left;
+		} else if (tfp->addr > addr) {
+			tmp = tmp->rb_right;
+		} else {
+			result = tfp;
+			break;
+		}
+	}
+
+	return result;
+}
+
+static int mlx5_cmd_query_pages(struct mlx5_core_dev *dev, u16 *func_id,
+				s32 *npages, int boot)
+{
+	u32 out[MLX5_ST_SZ_DW(query_pages_out)] = {};
+	u32 in[MLX5_ST_SZ_DW(query_pages_in)] = {};
+	int err;
+
+	MLX5_SET(query_pages_in, in, opcode, MLX5_CMD_OP_QUERY_PAGES);
+	MLX5_SET(query_pages_in, in, op_mod, boot ?
+		 MLX5_QUERY_PAGES_IN_OP_MOD_BOOT_PAGES :
+		 MLX5_QUERY_PAGES_IN_OP_MOD_INIT_PAGES);
+	MLX5_SET(query_pages_in, in, embedded_cpu_function, mlx5_core_is_ecpf(dev));
+
+	err = mlx5_cmd_exec_inout(dev, query_pages, in, out);
+	if (err)
+		return err;
+
+	*npages = MLX5_GET(query_pages_out, out, num_pages);
+	*func_id = MLX5_GET(query_pages_out, out, function_id);
+
+	return err;
+}
+
+static int alloc_4k(struct mlx5_core_dev *dev, u64 *addr, u32 function)
+{
+	struct fw_page *fp = NULL;
+	struct fw_page *iter;
+	unsigned n;
+
+	list_for_each_entry(iter, &dev->priv.free_list, list) {
+		if (iter->function != function)
+			continue;
+		fp = iter;
+	}
+
+	if (list_empty(&dev->priv.free_list) || !fp)
+		return -ENOMEM;
+
+	n = find_first_bit(&fp->bitmask, 8 * sizeof(fp->bitmask));
+	if (n >= MLX5_NUM_4K_IN_PAGE) {
+		mlx5_core_warn(dev, "alloc 4k bug: fw page = 0x%llx, n = %u, bitmask: %lu, max num of 4K pages: %d\n",
+			       fp->addr, n, fp->bitmask,  MLX5_NUM_4K_IN_PAGE);
+		return -ENOENT;
+	}
+	clear_bit(n, &fp->bitmask);
+	fp->free_count--;
+	if (!fp->free_count)
+		list_del(&fp->list);
+
+	*addr = fp->addr + n * MLX5_ADAPTER_PAGE_SIZE;
+
+	return 0;
+}
+
+#define MLX5_U64_4K_PAGE_MASK ((~(u64)0U) << PAGE_SHIFT)
+
+static void free_fwp(struct mlx5_core_dev *dev, struct fw_page *fwp,
+		     bool in_free_list)
+{
+	struct rb_root *root;
+
+	root = xa_load(&dev->priv.page_root_xa, fwp->function);
+	if (WARN_ON_ONCE(!root))
+		return;
+
+	rb_erase(&fwp->rb_node, root);
+	if (in_free_list)
+		list_del(&fwp->list);
+	dma_unmap_page(mlx5_core_dma_dev(dev), fwp->addr & MLX5_U64_4K_PAGE_MASK,
+		       PAGE_SIZE, DMA_BIDIRECTIONAL);
+	__free_page(fwp->page);
+	kfree(fwp);
+}
+
+static void free_4k(struct mlx5_core_dev *dev, u64 addr, u32 function)
+{
+	struct fw_page *fwp;
+	int n;
+
+	fwp = find_fw_page(dev, addr & MLX5_U64_4K_PAGE_MASK, function);
+	if (!fwp) {
+		mlx5_core_warn_rl(dev, "page not found\n");
+		return;
+	}
+	n = (addr & ~MLX5_U64_4K_PAGE_MASK) >> MLX5_ADAPTER_PAGE_SHIFT;
+	fwp->free_count++;
+	set_bit(n, &fwp->bitmask);
+	if (fwp->free_count == MLX5_NUM_4K_IN_PAGE)
+		free_fwp(dev, fwp, fwp->free_count != 1);
+	else if (fwp->free_count == 1)
+		list_add(&fwp->list, &dev->priv.free_list);
+}
+
+static int alloc_system_page(struct mlx5_core_dev *dev, u32 function)
+{
+	struct device *device = mlx5_core_dma_dev(dev);
+	int nid = dev_to_node(device);
+	struct page *page;
+	u64 zero_addr = 1;
+	u64 addr;
+	int err;
+
+	page = alloc_pages_node(nid, GFP_HIGHUSER, 0);
+	if (!page) {
+		mlx5_core_warn(dev, "failed to allocate page\n");
+		return -ENOMEM;
+	}
+map:
+	addr = dma_map_page(device, page, 0, PAGE_SIZE, DMA_BIDIRECTIONAL);
+	if (dma_mapping_error(device, addr)) {
+		mlx5_core_warn(dev, "failed dma mapping page\n");
+		err = -ENOMEM;
+		goto err_mapping;
+	}
+
+	/* Firmware doesn't support page with physical address 0 */
+	if (addr == 0) {
+		zero_addr = addr;
+		goto map;
+	}
+
+	err = insert_page(dev, addr, page, function);
+	if (err) {
+		mlx5_core_err(dev, "failed to track allocated page\n");
+		dma_unmap_page(device, addr, PAGE_SIZE, DMA_BIDIRECTIONAL);
+	}
+
+err_mapping:
+	if (err)
+		__free_page(page);
+
+	if (zero_addr == 0)
+		dma_unmap_page(device, zero_addr, PAGE_SIZE,
+			       DMA_BIDIRECTIONAL);
+
+	return err;
+}
+
+static void page_notify_fail(struct mlx5_core_dev *dev, u16 func_id,
+			     bool ec_function)
+{
+	u32 in[MLX5_ST_SZ_DW(manage_pages_in)] = {};
+	int err;
+
+	MLX5_SET(manage_pages_in, in, opcode, MLX5_CMD_OP_MANAGE_PAGES);
+	MLX5_SET(manage_pages_in, in, op_mod, MLX5_PAGES_CANT_GIVE);
+	MLX5_SET(manage_pages_in, in, function_id, func_id);
+	MLX5_SET(manage_pages_in, in, embedded_cpu_function, ec_function);
+
+	err = mlx5_cmd_exec_in(dev, manage_pages, in);
+	if (err)
+		mlx5_core_warn(dev, "page notify failed func_id(%d) err(%d)\n",
+			       func_id, err);
+}
+
+static int give_pages(struct mlx5_core_dev *dev, u16 func_id, int npages,
+		      int event, bool ec_function)
+{
+	u32 function = get_function(func_id, ec_function);
+	u32 out[MLX5_ST_SZ_DW(manage_pages_out)] = {0};
+	int inlen = MLX5_ST_SZ_BYTES(manage_pages_in);
+	int notify_fail = event;
+	u16 func_type;
+	u64 addr;
+	int err;
+	u32 *in;
+	int i;
+
+	inlen += npages * MLX5_FLD_SZ_BYTES(manage_pages_in, pas[0]);
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in) {
+		err = -ENOMEM;
+		mlx5_core_warn(dev, "vzalloc failed %d\n", inlen);
+		goto out_free;
+	}
+
+	for (i = 0; i < npages; i++) {
+retry:
+		err = alloc_4k(dev, &addr, function);
+		if (err) {
+			if (err == -ENOMEM)
+				err = alloc_system_page(dev, function);
+			if (err) {
+				dev->priv.fw_pages_alloc_failed += (npages - i);
+				goto out_4k;
+			}
+
+			goto retry;
+		}
+		MLX5_ARRAY_SET64(manage_pages_in, in, pas, i, addr);
+	}
+
+	MLX5_SET(manage_pages_in, in, opcode, MLX5_CMD_OP_MANAGE_PAGES);
+	MLX5_SET(manage_pages_in, in, op_mod, MLX5_PAGES_GIVE);
+	MLX5_SET(manage_pages_in, in, function_id, func_id);
+	MLX5_SET(manage_pages_in, in, input_num_entries, npages);
+	MLX5_SET(manage_pages_in, in, embedded_cpu_function, ec_function);
+
+	err = mlx5_cmd_do(dev, in, inlen, out, sizeof(out));
+	if (err == -EREMOTEIO) {
+		notify_fail = 0;
+		/* if triggered by FW and failed by FW ignore */
+		if (event) {
+			err = 0;
+			goto out_dropped;
+		}
+	}
+	err = mlx5_cmd_check(dev, err, in, out);
+	if (err) {
+		mlx5_core_warn(dev, "func_id 0x%x, npages %d, err %d\n",
+			       func_id, npages, err);
+		goto out_dropped;
+	}
+
+	func_type = func_id_to_type(dev, func_id, ec_function);
+	dev->priv.page_counters[func_type] += npages;
+	dev->priv.fw_pages += npages;
+
+	mlx5_core_dbg(dev, "npages %d, ec_function %d, func_id 0x%x, err %d\n",
+		      npages, ec_function, func_id, err);
+
+	kvfree(in);
+	return 0;
+
+out_dropped:
+	dev->priv.give_pages_dropped += npages;
+out_4k:
+	for (i--; i >= 0; i--)
+		free_4k(dev, MLX5_GET64(manage_pages_in, in, pas[i]), function);
+out_free:
+	kvfree(in);
+	if (notify_fail)
+		page_notify_fail(dev, func_id, ec_function);
+	return err;
+}
+
+static void release_all_pages(struct mlx5_core_dev *dev, u16 func_id,
+			      bool ec_function)
+{
+	u32 function = get_function(func_id, ec_function);
+	struct rb_root *root;
+	struct rb_node *p;
+	int npages = 0;
+	u16 func_type;
+
+	root = xa_load(&dev->priv.page_root_xa, function);
+	if (WARN_ON_ONCE(!root))
+		return;
+
+	p = rb_first(root);
+	while (p) {
+		struct fw_page *fwp = rb_entry(p, struct fw_page, rb_node);
+
+		p = rb_next(p);
+		npages += (MLX5_NUM_4K_IN_PAGE - fwp->free_count);
+		free_fwp(dev, fwp, fwp->free_count);
+	}
+
+	func_type = func_id_to_type(dev, func_id, ec_function);
+	dev->priv.page_counters[func_type] -= npages;
+	dev->priv.fw_pages -= npages;
+
+	mlx5_core_dbg(dev, "npages %d, ec_function %d, func_id 0x%x\n",
+		      npages, ec_function, func_id);
+}
+
+static u32 fwp_fill_manage_pages_out(struct fw_page *fwp, u32 *out, u32 index,
+				     u32 npages)
+{
+	u32 pages_set = 0;
+	unsigned int n;
+
+	for_each_clear_bit(n, &fwp->bitmask, MLX5_NUM_4K_IN_PAGE) {
+		MLX5_ARRAY_SET64(manage_pages_out, out, pas, index + pages_set,
+				 fwp->addr + (n * MLX5_ADAPTER_PAGE_SIZE));
+		pages_set++;
+
+		if (!--npages)
+			break;
+	}
+
+	return pages_set;
+}
+
+static int reclaim_pages_cmd(struct mlx5_core_dev *dev,
+			     u32 *in, int in_size, u32 *out, int out_size)
+{
+	struct rb_root *root;
+	struct fw_page *fwp;
+	struct rb_node *p;
+	bool ec_function;
+	u32 func_id;
+	u32 npages;
+	u32 i = 0;
+
+	if (!mlx5_cmd_is_down(dev))
+		return mlx5_cmd_do(dev, in, in_size, out, out_size);
+
+	/* No hard feelings, we want our pages back! */
+	npages = MLX5_GET(manage_pages_in, in, input_num_entries);
+	func_id = MLX5_GET(manage_pages_in, in, function_id);
+	ec_function = MLX5_GET(manage_pages_in, in, embedded_cpu_function);
+
+	root = xa_load(&dev->priv.page_root_xa, get_function(func_id, ec_function));
+	if (WARN_ON_ONCE(!root))
+		return -EEXIST;
+
+	p = rb_first(root);
+	while (p && i < npages) {
+		fwp = rb_entry(p, struct fw_page, rb_node);
+		p = rb_next(p);
+
+		i += fwp_fill_manage_pages_out(fwp, out, i, npages - i);
+	}
+
+	MLX5_SET(manage_pages_out, out, output_num_entries, i);
+	return 0;
+}
+
+static int reclaim_pages(struct mlx5_core_dev *dev, u16 func_id, int npages,
+			 int *nclaimed, bool event, bool ec_function)
+{
+	u32 function = get_function(func_id, ec_function);
+	int outlen = MLX5_ST_SZ_BYTES(manage_pages_out);
+	u32 in[MLX5_ST_SZ_DW(manage_pages_in)] = {};
+	int num_claimed;
+	u16 func_type;
+	u32 *out;
+	int err;
+	int i;
+
+	if (nclaimed)
+		*nclaimed = 0;
+
+	outlen += npages * MLX5_FLD_SZ_BYTES(manage_pages_out, pas[0]);
+	out = kvzalloc(outlen, GFP_KERNEL);
+	if (!out)
+		return -ENOMEM;
+
+	MLX5_SET(manage_pages_in, in, opcode, MLX5_CMD_OP_MANAGE_PAGES);
+	MLX5_SET(manage_pages_in, in, op_mod, MLX5_PAGES_TAKE);
+	MLX5_SET(manage_pages_in, in, function_id, func_id);
+	MLX5_SET(manage_pages_in, in, input_num_entries, npages);
+	MLX5_SET(manage_pages_in, in, embedded_cpu_function, ec_function);
+
+	mlx5_core_dbg(dev, "func 0x%x, npages %d, outlen %d\n",
+		      func_id, npages, outlen);
+	err = reclaim_pages_cmd(dev, in, sizeof(in), out, outlen);
+	if (err) {
+		npages = MLX5_GET(manage_pages_in, in, input_num_entries);
+		dev->priv.reclaim_pages_discard += npages;
+	}
+	/* if triggered by FW event and failed by FW then ignore */
+	if (event && err == -EREMOTEIO) {
+		err = 0;
+		goto out_free;
+	}
+
+	err = mlx5_cmd_check(dev, err, in, out);
+	if (err) {
+		mlx5_core_err(dev, "failed reclaiming pages: err %d\n", err);
+		goto out_free;
+	}
+
+	num_claimed = MLX5_GET(manage_pages_out, out, output_num_entries);
+	if (num_claimed > npages) {
+		mlx5_core_warn(dev, "fw returned %d, driver asked %d => corruption\n",
+			       num_claimed, npages);
+		err = -EINVAL;
+		goto out_free;
+	}
+
+	for (i = 0; i < num_claimed; i++)
+		free_4k(dev, MLX5_GET64(manage_pages_out, out, pas[i]), function);
+
+	if (nclaimed)
+		*nclaimed = num_claimed;
+
+	func_type = func_id_to_type(dev, func_id, ec_function);
+	dev->priv.page_counters[func_type] -= num_claimed;
+	dev->priv.fw_pages -= num_claimed;
+
+out_free:
+	kvfree(out);
+	return err;
+}
+
+static void pages_work_handler(struct work_struct *work)
+{
+	struct mlx5_pages_req *req = container_of(work, struct mlx5_pages_req, work);
+	struct mlx5_core_dev *dev = req->dev;
+	int err = 0;
+
+	if (req->release_all)
+		release_all_pages(dev, req->func_id, req->ec_function);
+	else if (req->npages < 0)
+		err = reclaim_pages(dev, req->func_id, -1 * req->npages, NULL,
+				    true, req->ec_function);
+	else if (req->npages > 0)
+		err = give_pages(dev, req->func_id, req->npages, 1, req->ec_function);
+
+	if (err)
+		mlx5_core_warn(dev, "%s fail %d\n",
+			       req->npages < 0 ? "reclaim" : "give", err);
+
+	kfree(req);
+}
+
+enum {
+	EC_FUNCTION_MASK = 0x8000,
+	RELEASE_ALL_PAGES_MASK = 0x4000,
+};
+
+static int req_pages_handler(struct notifier_block *nb,
+			     unsigned long type, void *data)
+{
+	struct mlx5_pages_req *req;
+	struct mlx5_core_dev *dev;
+	struct mlx5_priv *priv;
+	struct mlx5_eqe *eqe;
+	bool ec_function;
+	bool release_all;
+	u16 func_id;
+	s32 npages;
+
+	priv = mlx5_nb_cof(nb, struct mlx5_priv, pg_nb);
+	dev  = container_of(priv, struct mlx5_core_dev, priv);
+	eqe  = data;
+
+	func_id = be16_to_cpu(eqe->data.req_pages.func_id);
+	npages  = be32_to_cpu(eqe->data.req_pages.num_pages);
+	ec_function = be16_to_cpu(eqe->data.req_pages.ec_function) & EC_FUNCTION_MASK;
+	release_all = be16_to_cpu(eqe->data.req_pages.ec_function) &
+		      RELEASE_ALL_PAGES_MASK;
+	mlx5_core_dbg(dev, "page request for func 0x%x, npages %d, release_all %d\n",
+		      func_id, npages, release_all);
+	req = kzalloc(sizeof(*req), GFP_ATOMIC);
+	if (!req) {
+		mlx5_core_warn(dev, "failed to allocate pages request\n");
+		return NOTIFY_DONE;
+	}
+
+	req->dev = dev;
+	req->func_id = func_id;
+	req->npages = npages;
+	req->ec_function = ec_function;
+	req->release_all = release_all;
+	INIT_WORK(&req->work, pages_work_handler);
+	queue_work(dev->priv.pg_wq, &req->work);
+	return NOTIFY_OK;
+}
+
+int mlx5_satisfy_startup_pages(struct mlx5_core_dev *dev, int boot)
+{
+	u16 func_id;
+	s32 npages;
+	int err;
+
+	err = mlx5_cmd_query_pages(dev, &func_id, &npages, boot);
+	if (err)
+		return err;
+
+	mlx5_core_dbg(dev, "requested %d %s pages for func_id 0x%x\n",
+		      npages, boot ? "boot" : "init", func_id);
+
+	return give_pages(dev, func_id, npages, 0, mlx5_core_is_ecpf(dev));
+}
+
+enum {
+	MLX5_BLKS_FOR_RECLAIM_PAGES = 12
+};
+
+static int optimal_reclaimed_pages(void)
+{
+	struct mlx5_cmd_prot_block *block;
+	struct mlx5_cmd_layout *lay;
+	int ret;
+
+	ret = (sizeof(lay->out) + MLX5_BLKS_FOR_RECLAIM_PAGES * sizeof(block->data) -
+	       MLX5_ST_SZ_BYTES(manage_pages_out)) /
+	       MLX5_FLD_SZ_BYTES(manage_pages_out, pas[0]);
+
+	return ret;
+}
+
+static int mlx5_reclaim_root_pages(struct mlx5_core_dev *dev,
+				   struct rb_root *root, u32 function)
+{
+	u64 recl_pages_to_jiffies = msecs_to_jiffies(mlx5_tout_ms(dev, RECLAIM_PAGES));
+	unsigned long end = jiffies + recl_pages_to_jiffies;
+
+	while (!RB_EMPTY_ROOT(root)) {
+		u32 ec_function = mlx5_get_ec_function(function);
+		u32 function_id = mlx5_get_func_id(function);
+		int nclaimed;
+		int err;
+
+		err = reclaim_pages(dev, function_id, optimal_reclaimed_pages(),
+				    &nclaimed, false, ec_function);
+		if (err) {
+			mlx5_core_warn(dev, "reclaim_pages err (%d) func_id=0x%x ec_func=0x%x\n",
+				       err, function_id, ec_function);
+			return err;
+		}
+
+		if (nclaimed)
+			end = jiffies + recl_pages_to_jiffies;
+
+		if (time_after(jiffies, end)) {
+			mlx5_core_warn(dev, "FW did not return all pages. giving up...\n");
+			break;
+		}
+	}
+
+	return 0;
+}
+
+int mlx5_reclaim_startup_pages(struct mlx5_core_dev *dev)
+{
+	struct rb_root *root;
+	unsigned long id;
+	void *entry;
+
+	xa_for_each(&dev->priv.page_root_xa, id, entry) {
+		root = entry;
+		mlx5_reclaim_root_pages(dev, root, id);
+		xa_erase(&dev->priv.page_root_xa, id);
+		kfree(root);
+	}
+
+	WARN_ON(!xa_empty(&dev->priv.page_root_xa));
+
+	WARN(dev->priv.fw_pages,
+	     "FW pages counter is %d after reclaiming all pages\n",
+	     dev->priv.fw_pages);
+	WARN(dev->priv.page_counters[MLX5_VF],
+	     "VFs FW pages counter is %d after reclaiming all pages\n",
+	     dev->priv.page_counters[MLX5_VF]);
+	WARN(dev->priv.page_counters[MLX5_HOST_PF],
+	     "External host PF FW pages counter is %d after reclaiming all pages\n",
+	     dev->priv.page_counters[MLX5_HOST_PF]);
+
+	return 0;
+}
+
+int mlx5_pagealloc_init(struct mlx5_core_dev *dev)
+{
+	INIT_LIST_HEAD(&dev->priv.free_list);
+	dev->priv.pg_wq = create_singlethread_workqueue("mlx5_page_allocator");
+	if (!dev->priv.pg_wq)
+		return -ENOMEM;
+
+	xa_init(&dev->priv.page_root_xa);
+	mlx5_pages_debugfs_init(dev);
+
+	return 0;
+}
+
+void mlx5_pagealloc_cleanup(struct mlx5_core_dev *dev)
+{
+	mlx5_pages_debugfs_cleanup(dev);
+	xa_destroy(&dev->priv.page_root_xa);
+	destroy_workqueue(dev->priv.pg_wq);
+}
+
+void mlx5_pagealloc_start(struct mlx5_core_dev *dev)
+{
+	MLX5_NB_INIT(&dev->priv.pg_nb, req_pages_handler, PAGE_REQUEST);
+	mlx5_eq_notifier_register(dev, &dev->priv.pg_nb);
+}
+
+void mlx5_pagealloc_stop(struct mlx5_core_dev *dev)
+{
+	mlx5_eq_notifier_unregister(dev, &dev->priv.pg_nb);
+	flush_workqueue(dev->priv.pg_wq);
+}
+
+int mlx5_wait_for_pages(struct mlx5_core_dev *dev, int *pages)
+{
+	u64 recl_vf_pages_to_jiffies = msecs_to_jiffies(mlx5_tout_ms(dev, RECLAIM_VFS_PAGES));
+	unsigned long end = jiffies + recl_vf_pages_to_jiffies;
+	int prev_pages = *pages;
+
+	/* In case of internal error we will free the pages manually later */
+	if (dev->state == MLX5_DEVICE_STATE_INTERNAL_ERROR) {
+		mlx5_core_warn(dev, "Skipping wait for vf pages stage");
+		return 0;
+	}
+
+	mlx5_core_dbg(dev, "Waiting for %d pages\n", prev_pages);
+	while (*pages) {
+		if (time_after(jiffies, end)) {
+			mlx5_core_warn(dev, "aborting while there are %d pending pages\n", *pages);
+			return -ETIMEDOUT;
+		}
+		if (*pages < prev_pages) {
+			end = jiffies + recl_vf_pages_to_jiffies;
+			prev_pages = *pages;
+		}
+		msleep(50);
+	}
+
+	mlx5_core_dbg(dev, "All pages received\n");
+	return 0;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/pci_irq.c b/drivers/net/ethernet/mellanox/mlx5/core/pci_irq.c
new file mode 100644
index 000000000..a6d3fc96e
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/pci_irq.c
@@ -0,0 +1,733 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2019 Mellanox Technologies. */
+
+#include <linux/interrupt.h>
+#include <linux/notifier.h>
+#include <linux/mlx5/driver.h>
+#include "mlx5_core.h"
+#include "mlx5_irq.h"
+#include "pci_irq.h"
+#include "lib/sf.h"
+#ifdef CONFIG_RFS_ACCEL
+#include <linux/cpu_rmap.h>
+#endif
+
+#define MLX5_SFS_PER_CTRL_IRQ 64
+#define MLX5_IRQ_CTRL_SF_MAX 8
+/* min num of vectors for SFs to be enabled */
+#define MLX5_IRQ_VEC_COMP_BASE_SF 2
+
+#define MLX5_EQ_SHARE_IRQ_MAX_COMP (8)
+#define MLX5_EQ_SHARE_IRQ_MAX_CTRL (UINT_MAX)
+#define MLX5_EQ_SHARE_IRQ_MIN_COMP (1)
+#define MLX5_EQ_SHARE_IRQ_MIN_CTRL (4)
+
+struct mlx5_irq {
+	struct atomic_notifier_head nh;
+	cpumask_var_t mask;
+	char name[MLX5_MAX_IRQ_FORMATTED_NAME];
+	struct mlx5_irq_pool *pool;
+	int refcount;
+	u32 index;
+	int irqn;
+};
+
+struct mlx5_irq_table {
+	struct mlx5_irq_pool *pf_pool;
+	struct mlx5_irq_pool *sf_ctrl_pool;
+	struct mlx5_irq_pool *sf_comp_pool;
+};
+
+/**
+ * mlx5_get_default_msix_vec_count - Get the default number of MSI-X vectors
+ *                                   to be ssigned to each VF.
+ * @dev: PF to work on
+ * @num_vfs: Number of enabled VFs
+ */
+int mlx5_get_default_msix_vec_count(struct mlx5_core_dev *dev, int num_vfs)
+{
+	int num_vf_msix, min_msix, max_msix;
+
+	num_vf_msix = MLX5_CAP_GEN_MAX(dev, num_total_dynamic_vf_msix);
+	if (!num_vf_msix)
+		return 0;
+
+	min_msix = MLX5_CAP_GEN(dev, min_dynamic_vf_msix_table_size);
+	max_msix = MLX5_CAP_GEN(dev, max_dynamic_vf_msix_table_size);
+
+	/* Limit maximum number of MSI-X vectors so the default configuration
+	 * has some available in the pool. This will allow the user to increase
+	 * the number of vectors in a VF without having to first size-down other
+	 * VFs.
+	 */
+	return max(min(num_vf_msix / num_vfs, max_msix / 2), min_msix);
+}
+
+/**
+ * mlx5_set_msix_vec_count - Set dynamically allocated MSI-X on the VF
+ * @dev: PF to work on
+ * @function_id: Internal PCI VF function IDd
+ * @msix_vec_count: Number of MSI-X vectors to set
+ */
+int mlx5_set_msix_vec_count(struct mlx5_core_dev *dev, int function_id,
+			    int msix_vec_count)
+{
+	int query_sz = MLX5_ST_SZ_BYTES(query_hca_cap_out);
+	int set_sz = MLX5_ST_SZ_BYTES(set_hca_cap_in);
+	void *hca_cap = NULL, *query_cap = NULL, *cap;
+	int num_vf_msix, min_msix, max_msix;
+	int ret;
+
+	num_vf_msix = MLX5_CAP_GEN_MAX(dev, num_total_dynamic_vf_msix);
+	if (!num_vf_msix)
+		return 0;
+
+	if (!MLX5_CAP_GEN(dev, vport_group_manager) || !mlx5_core_is_pf(dev))
+		return -EOPNOTSUPP;
+
+	min_msix = MLX5_CAP_GEN(dev, min_dynamic_vf_msix_table_size);
+	max_msix = MLX5_CAP_GEN(dev, max_dynamic_vf_msix_table_size);
+
+	if (msix_vec_count < min_msix)
+		return -EINVAL;
+
+	if (msix_vec_count > max_msix)
+		return -EOVERFLOW;
+
+	query_cap = kvzalloc(query_sz, GFP_KERNEL);
+	hca_cap = kvzalloc(set_sz, GFP_KERNEL);
+	if (!hca_cap || !query_cap) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	ret = mlx5_vport_get_other_func_cap(dev, function_id, query_cap);
+	if (ret)
+		goto out;
+
+	cap = MLX5_ADDR_OF(set_hca_cap_in, hca_cap, capability);
+	memcpy(cap, MLX5_ADDR_OF(query_hca_cap_out, query_cap, capability),
+	       MLX5_UN_SZ_BYTES(hca_cap_union));
+	MLX5_SET(cmd_hca_cap, cap, dynamic_msix_table_size, msix_vec_count);
+
+	MLX5_SET(set_hca_cap_in, hca_cap, opcode, MLX5_CMD_OP_SET_HCA_CAP);
+	MLX5_SET(set_hca_cap_in, hca_cap, other_function, 1);
+	MLX5_SET(set_hca_cap_in, hca_cap, function_id, function_id);
+
+	MLX5_SET(set_hca_cap_in, hca_cap, op_mod,
+		 MLX5_SET_HCA_CAP_OP_MOD_GENERAL_DEVICE << 1);
+	ret = mlx5_cmd_exec_in(dev, set_hca_cap, hca_cap);
+out:
+	kvfree(hca_cap);
+	kvfree(query_cap);
+	return ret;
+}
+
+/* mlx5_system_free_irq - Free an IRQ
+ * @irq: IRQ to free
+ *
+ * Free the IRQ and other resources such as rmap from the system.
+ * BUT doesn't free or remove reference from mlx5.
+ * This function is very important for the shutdown flow, where we need to
+ * cleanup system resoruces but keep mlx5 objects alive,
+ * see mlx5_irq_table_free_irqs().
+ */
+static void mlx5_system_free_irq(struct mlx5_irq *irq)
+{
+	/* free_irq requires that affinity_hint and rmap will be cleared
+	 * before calling it. This is why there is asymmetry with set_rmap
+	 * which should be called after alloc_irq but before request_irq.
+	 */
+	irq_update_affinity_hint(irq->irqn, NULL);
+	free_irq(irq->irqn, &irq->nh);
+}
+
+static void irq_release(struct mlx5_irq *irq)
+{
+	struct mlx5_irq_pool *pool = irq->pool;
+
+	xa_erase(&pool->irqs, irq->index);
+	mlx5_system_free_irq(irq);
+	free_cpumask_var(irq->mask);
+	kfree(irq);
+}
+
+int mlx5_irq_put(struct mlx5_irq *irq)
+{
+	struct mlx5_irq_pool *pool = irq->pool;
+	int ret = 0;
+
+	mutex_lock(&pool->lock);
+	irq->refcount--;
+	if (!irq->refcount) {
+		irq_release(irq);
+		ret = 1;
+	}
+	mutex_unlock(&pool->lock);
+	return ret;
+}
+
+int mlx5_irq_read_locked(struct mlx5_irq *irq)
+{
+	lockdep_assert_held(&irq->pool->lock);
+	return irq->refcount;
+}
+
+int mlx5_irq_get_locked(struct mlx5_irq *irq)
+{
+	lockdep_assert_held(&irq->pool->lock);
+	if (WARN_ON_ONCE(!irq->refcount))
+		return 0;
+	irq->refcount++;
+	return 1;
+}
+
+static int irq_get(struct mlx5_irq *irq)
+{
+	int err;
+
+	mutex_lock(&irq->pool->lock);
+	err = mlx5_irq_get_locked(irq);
+	mutex_unlock(&irq->pool->lock);
+	return err;
+}
+
+static irqreturn_t irq_int_handler(int irq, void *nh)
+{
+	atomic_notifier_call_chain(nh, 0, NULL);
+	return IRQ_HANDLED;
+}
+
+static void irq_sf_set_name(struct mlx5_irq_pool *pool, char *name, int vecidx)
+{
+	snprintf(name, MLX5_MAX_IRQ_NAME, "%s%d", pool->name, vecidx);
+}
+
+static void irq_set_name(struct mlx5_irq_pool *pool, char *name, int vecidx)
+{
+	if (!pool->xa_num_irqs.max) {
+		/* in case we only have a single irq for the device */
+		snprintf(name, MLX5_MAX_IRQ_NAME, "mlx5_combined%d", vecidx);
+		return;
+	}
+
+	if (vecidx == pool->xa_num_irqs.max) {
+		snprintf(name, MLX5_MAX_IRQ_NAME, "mlx5_async%d", vecidx);
+		return;
+	}
+
+	snprintf(name, MLX5_MAX_IRQ_NAME, "mlx5_comp%d", vecidx);
+}
+
+struct mlx5_irq *mlx5_irq_alloc(struct mlx5_irq_pool *pool, int i,
+				const struct cpumask *affinity)
+{
+	struct mlx5_core_dev *dev = pool->dev;
+	char name[MLX5_MAX_IRQ_NAME];
+	struct mlx5_irq *irq;
+	int err;
+
+	irq = kzalloc(sizeof(*irq), GFP_KERNEL);
+	if (!irq)
+		return ERR_PTR(-ENOMEM);
+	irq->irqn = pci_irq_vector(dev->pdev, i);
+	if (!mlx5_irq_pool_is_sf_pool(pool))
+		irq_set_name(pool, name, i);
+	else
+		irq_sf_set_name(pool, name, i);
+	ATOMIC_INIT_NOTIFIER_HEAD(&irq->nh);
+	snprintf(irq->name, MLX5_MAX_IRQ_FORMATTED_NAME,
+		 MLX5_IRQ_NAME_FORMAT_STR, name, pci_name(dev->pdev));
+	err = request_irq(irq->irqn, irq_int_handler, 0, irq->name,
+			  &irq->nh);
+	if (err) {
+		mlx5_core_err(dev, "Failed to request irq. err = %d\n", err);
+		goto err_req_irq;
+	}
+	if (!zalloc_cpumask_var(&irq->mask, GFP_KERNEL)) {
+		mlx5_core_warn(dev, "zalloc_cpumask_var failed\n");
+		err = -ENOMEM;
+		goto err_cpumask;
+	}
+	if (affinity) {
+		cpumask_copy(irq->mask, affinity);
+		irq_set_affinity_and_hint(irq->irqn, irq->mask);
+	}
+	irq->pool = pool;
+	irq->refcount = 1;
+	irq->index = i;
+	err = xa_err(xa_store(&pool->irqs, irq->index, irq, GFP_KERNEL));
+	if (err) {
+		mlx5_core_err(dev, "Failed to alloc xa entry for irq(%u). err = %d\n",
+			      irq->index, err);
+		goto err_xa;
+	}
+	return irq;
+err_xa:
+	irq_update_affinity_hint(irq->irqn, NULL);
+	free_cpumask_var(irq->mask);
+err_cpumask:
+	free_irq(irq->irqn, &irq->nh);
+err_req_irq:
+	kfree(irq);
+	return ERR_PTR(err);
+}
+
+int mlx5_irq_attach_nb(struct mlx5_irq *irq, struct notifier_block *nb)
+{
+	int ret;
+
+	ret = irq_get(irq);
+	if (!ret)
+		/* Something very bad happens here, we are enabling EQ
+		 * on non-existing IRQ.
+		 */
+		return -ENOENT;
+	ret = atomic_notifier_chain_register(&irq->nh, nb);
+	if (ret)
+		mlx5_irq_put(irq);
+	return ret;
+}
+
+int mlx5_irq_detach_nb(struct mlx5_irq *irq, struct notifier_block *nb)
+{
+	int err = 0;
+
+	err = atomic_notifier_chain_unregister(&irq->nh, nb);
+	mlx5_irq_put(irq);
+	return err;
+}
+
+struct cpumask *mlx5_irq_get_affinity_mask(struct mlx5_irq *irq)
+{
+	return irq->mask;
+}
+
+int mlx5_irq_get_index(struct mlx5_irq *irq)
+{
+	return irq->index;
+}
+
+/* irq_pool API */
+
+/* requesting an irq from a given pool according to given index */
+static struct mlx5_irq *
+irq_pool_request_vector(struct mlx5_irq_pool *pool, int vecidx,
+			struct cpumask *affinity)
+{
+	struct mlx5_irq *irq;
+
+	mutex_lock(&pool->lock);
+	irq = xa_load(&pool->irqs, vecidx);
+	if (irq) {
+		mlx5_irq_get_locked(irq);
+		goto unlock;
+	}
+	irq = mlx5_irq_alloc(pool, vecidx, affinity);
+unlock:
+	mutex_unlock(&pool->lock);
+	return irq;
+}
+
+static struct mlx5_irq_pool *sf_ctrl_irq_pool_get(struct mlx5_irq_table *irq_table)
+{
+	return irq_table->sf_ctrl_pool;
+}
+
+static struct mlx5_irq_pool *sf_irq_pool_get(struct mlx5_irq_table *irq_table)
+{
+	return irq_table->sf_comp_pool;
+}
+
+struct mlx5_irq_pool *mlx5_irq_pool_get(struct mlx5_core_dev *dev)
+{
+	struct mlx5_irq_table *irq_table = mlx5_irq_table_get(dev);
+	struct mlx5_irq_pool *pool = NULL;
+
+	if (mlx5_core_is_sf(dev))
+		pool = sf_irq_pool_get(irq_table);
+
+	/* In some configs, there won't be a pool of SFs IRQs. Hence, returning
+	 * the PF IRQs pool in case the SF pool doesn't exist.
+	 */
+	return pool ? pool : irq_table->pf_pool;
+}
+
+static struct mlx5_irq_pool *ctrl_irq_pool_get(struct mlx5_core_dev *dev)
+{
+	struct mlx5_irq_table *irq_table = mlx5_irq_table_get(dev);
+	struct mlx5_irq_pool *pool = NULL;
+
+	if (mlx5_core_is_sf(dev))
+		pool = sf_ctrl_irq_pool_get(irq_table);
+
+	/* In some configs, there won't be a pool of SFs IRQs. Hence, returning
+	 * the PF IRQs pool in case the SF pool doesn't exist.
+	 */
+	return pool ? pool : irq_table->pf_pool;
+}
+
+/**
+ * mlx5_irqs_release - release one or more IRQs back to the system.
+ * @irqs: IRQs to be released.
+ * @nirqs: number of IRQs to be released.
+ */
+static void mlx5_irqs_release(struct mlx5_irq **irqs, int nirqs)
+{
+	int i;
+
+	for (i = 0; i < nirqs; i++) {
+		synchronize_irq(irqs[i]->irqn);
+		mlx5_irq_put(irqs[i]);
+	}
+}
+
+/**
+ * mlx5_ctrl_irq_release - release a ctrl IRQ back to the system.
+ * @ctrl_irq: ctrl IRQ to be released.
+ */
+void mlx5_ctrl_irq_release(struct mlx5_irq *ctrl_irq)
+{
+	mlx5_irqs_release(&ctrl_irq, 1);
+}
+
+/**
+ * mlx5_ctrl_irq_request - request a ctrl IRQ for mlx5 device.
+ * @dev: mlx5 device that requesting the IRQ.
+ *
+ * This function returns a pointer to IRQ, or ERR_PTR in case of error.
+ */
+struct mlx5_irq *mlx5_ctrl_irq_request(struct mlx5_core_dev *dev)
+{
+	struct mlx5_irq_pool *pool = ctrl_irq_pool_get(dev);
+	cpumask_var_t req_mask;
+	struct mlx5_irq *irq;
+
+	if (!zalloc_cpumask_var(&req_mask, GFP_KERNEL))
+		return ERR_PTR(-ENOMEM);
+	cpumask_copy(req_mask, cpu_online_mask);
+	if (!mlx5_irq_pool_is_sf_pool(pool)) {
+		/* In case we are allocating a control IRQ for PF/VF */
+		if (!pool->xa_num_irqs.max) {
+			cpumask_clear(req_mask);
+			/* In case we only have a single IRQ for PF/VF */
+			cpumask_set_cpu(cpumask_first(cpu_online_mask), req_mask);
+		}
+		/* Allocate the IRQ in the last index of the pool */
+		irq = irq_pool_request_vector(pool, pool->xa_num_irqs.max, req_mask);
+	} else {
+		irq = mlx5_irq_affinity_request(pool, req_mask);
+	}
+
+	free_cpumask_var(req_mask);
+	return irq;
+}
+
+/**
+ * mlx5_irq_request - request an IRQ for mlx5 PF/VF device.
+ * @dev: mlx5 device that requesting the IRQ.
+ * @vecidx: vector index of the IRQ. This argument is ignore if affinity is
+ * provided.
+ * @affinity: cpumask requested for this IRQ.
+ *
+ * This function returns a pointer to IRQ, or ERR_PTR in case of error.
+ */
+struct mlx5_irq *mlx5_irq_request(struct mlx5_core_dev *dev, u16 vecidx,
+				  struct cpumask *affinity)
+{
+	struct mlx5_irq_table *irq_table = mlx5_irq_table_get(dev);
+	struct mlx5_irq_pool *pool;
+	struct mlx5_irq *irq;
+
+	pool = irq_table->pf_pool;
+	irq = irq_pool_request_vector(pool, vecidx, affinity);
+	if (IS_ERR(irq))
+		return irq;
+	mlx5_core_dbg(dev, "irq %u mapped to cpu %*pbl, %u EQs on this irq\n",
+		      irq->irqn, cpumask_pr_args(affinity),
+		      irq->refcount / MLX5_EQ_REFS_PER_IRQ);
+	return irq;
+}
+
+/**
+ * mlx5_irqs_release_vectors - release one or more IRQs back to the system.
+ * @irqs: IRQs to be released.
+ * @nirqs: number of IRQs to be released.
+ */
+void mlx5_irqs_release_vectors(struct mlx5_irq **irqs, int nirqs)
+{
+	mlx5_irqs_release(irqs, nirqs);
+}
+
+/**
+ * mlx5_irqs_request_vectors - request one or more IRQs for mlx5 device.
+ * @dev: mlx5 device that is requesting the IRQs.
+ * @cpus: CPUs array for binding the IRQs
+ * @nirqs: number of IRQs to request.
+ * @irqs: an output array of IRQs pointers.
+ *
+ * Each IRQ is bound to at most 1 CPU.
+ * This function is requests nirqs IRQs, starting from @vecidx.
+ *
+ * This function returns the number of IRQs requested, (which might be smaller than
+ * @nirqs), if successful, or a negative error code in case of an error.
+ */
+int mlx5_irqs_request_vectors(struct mlx5_core_dev *dev, u16 *cpus, int nirqs,
+			      struct mlx5_irq **irqs)
+{
+	cpumask_var_t req_mask;
+	struct mlx5_irq *irq;
+	int i;
+
+	if (!zalloc_cpumask_var(&req_mask, GFP_KERNEL))
+		return -ENOMEM;
+	for (i = 0; i < nirqs; i++) {
+		cpumask_set_cpu(cpus[i], req_mask);
+		irq = mlx5_irq_request(dev, i, req_mask);
+		if (IS_ERR(irq))
+			break;
+		cpumask_clear(req_mask);
+		irqs[i] = irq;
+	}
+
+	free_cpumask_var(req_mask);
+	return i ? i : PTR_ERR(irq);
+}
+
+static struct mlx5_irq_pool *
+irq_pool_alloc(struct mlx5_core_dev *dev, int start, int size, char *name,
+	       u32 min_threshold, u32 max_threshold)
+{
+	struct mlx5_irq_pool *pool = kvzalloc(sizeof(*pool), GFP_KERNEL);
+
+	if (!pool)
+		return ERR_PTR(-ENOMEM);
+	pool->dev = dev;
+	mutex_init(&pool->lock);
+	xa_init_flags(&pool->irqs, XA_FLAGS_ALLOC);
+	pool->xa_num_irqs.min = start;
+	pool->xa_num_irqs.max = start + size - 1;
+	if (name)
+		snprintf(pool->name, MLX5_MAX_IRQ_NAME - MLX5_MAX_IRQ_IDX_CHARS,
+			 "%s", name);
+	pool->min_threshold = min_threshold * MLX5_EQ_REFS_PER_IRQ;
+	pool->max_threshold = max_threshold * MLX5_EQ_REFS_PER_IRQ;
+	mlx5_core_dbg(dev, "pool->name = %s, pool->size = %d, pool->start = %d",
+		      name, size, start);
+	return pool;
+}
+
+static void irq_pool_free(struct mlx5_irq_pool *pool)
+{
+	struct mlx5_irq *irq;
+	unsigned long index;
+
+	/* There are cases in which we are destrying the irq_table before
+	 * freeing all the IRQs, fast teardown for example. Hence, free the irqs
+	 * which might not have been freed.
+	 */
+	xa_for_each(&pool->irqs, index, irq)
+		irq_release(irq);
+	xa_destroy(&pool->irqs);
+	mutex_destroy(&pool->lock);
+	kfree(pool->irqs_per_cpu);
+	kvfree(pool);
+}
+
+static int irq_pools_init(struct mlx5_core_dev *dev, int sf_vec, int pf_vec)
+{
+	struct mlx5_irq_table *table = dev->priv.irq_table;
+	int num_sf_ctrl_by_msix;
+	int num_sf_ctrl_by_sfs;
+	int num_sf_ctrl;
+	int err;
+
+	/* init pf_pool */
+	table->pf_pool = irq_pool_alloc(dev, 0, pf_vec, NULL,
+					MLX5_EQ_SHARE_IRQ_MIN_COMP,
+					MLX5_EQ_SHARE_IRQ_MAX_COMP);
+	if (IS_ERR(table->pf_pool))
+		return PTR_ERR(table->pf_pool);
+	if (!mlx5_sf_max_functions(dev))
+		return 0;
+	if (sf_vec < MLX5_IRQ_VEC_COMP_BASE_SF) {
+		mlx5_core_dbg(dev, "Not enught IRQs for SFs. SF may run at lower performance\n");
+		return 0;
+	}
+
+	/* init sf_ctrl_pool */
+	num_sf_ctrl_by_msix = DIV_ROUND_UP(sf_vec, MLX5_COMP_EQS_PER_SF);
+	num_sf_ctrl_by_sfs = DIV_ROUND_UP(mlx5_sf_max_functions(dev),
+					  MLX5_SFS_PER_CTRL_IRQ);
+	num_sf_ctrl = min_t(int, num_sf_ctrl_by_msix, num_sf_ctrl_by_sfs);
+	num_sf_ctrl = min_t(int, MLX5_IRQ_CTRL_SF_MAX, num_sf_ctrl);
+	table->sf_ctrl_pool = irq_pool_alloc(dev, pf_vec, num_sf_ctrl,
+					     "mlx5_sf_ctrl",
+					     MLX5_EQ_SHARE_IRQ_MIN_CTRL,
+					     MLX5_EQ_SHARE_IRQ_MAX_CTRL);
+	if (IS_ERR(table->sf_ctrl_pool)) {
+		err = PTR_ERR(table->sf_ctrl_pool);
+		goto err_pf;
+	}
+	/* init sf_comp_pool */
+	table->sf_comp_pool = irq_pool_alloc(dev, pf_vec + num_sf_ctrl,
+					     sf_vec - num_sf_ctrl, "mlx5_sf_comp",
+					     MLX5_EQ_SHARE_IRQ_MIN_COMP,
+					     MLX5_EQ_SHARE_IRQ_MAX_COMP);
+	if (IS_ERR(table->sf_comp_pool)) {
+		err = PTR_ERR(table->sf_comp_pool);
+		goto err_sf_ctrl;
+	}
+
+	table->sf_comp_pool->irqs_per_cpu = kcalloc(nr_cpu_ids, sizeof(u16), GFP_KERNEL);
+	if (!table->sf_comp_pool->irqs_per_cpu) {
+		err = -ENOMEM;
+		goto err_irqs_per_cpu;
+	}
+
+	return 0;
+
+err_irqs_per_cpu:
+	irq_pool_free(table->sf_comp_pool);
+err_sf_ctrl:
+	irq_pool_free(table->sf_ctrl_pool);
+err_pf:
+	irq_pool_free(table->pf_pool);
+	return err;
+}
+
+static void irq_pools_destroy(struct mlx5_irq_table *table)
+{
+	if (table->sf_ctrl_pool) {
+		irq_pool_free(table->sf_comp_pool);
+		irq_pool_free(table->sf_ctrl_pool);
+	}
+	irq_pool_free(table->pf_pool);
+}
+
+static void mlx5_irq_pool_free_irqs(struct mlx5_irq_pool *pool)
+{
+	struct mlx5_irq *irq;
+	unsigned long index;
+
+	xa_for_each(&pool->irqs, index, irq)
+		mlx5_system_free_irq(irq);
+}
+
+static void mlx5_irq_pools_free_irqs(struct mlx5_irq_table *table)
+{
+	if (table->sf_ctrl_pool) {
+		mlx5_irq_pool_free_irqs(table->sf_comp_pool);
+		mlx5_irq_pool_free_irqs(table->sf_ctrl_pool);
+	}
+	mlx5_irq_pool_free_irqs(table->pf_pool);
+}
+
+/* irq_table API */
+
+int mlx5_irq_table_init(struct mlx5_core_dev *dev)
+{
+	struct mlx5_irq_table *irq_table;
+
+	if (mlx5_core_is_sf(dev))
+		return 0;
+
+	irq_table = kvzalloc_node(sizeof(*irq_table), GFP_KERNEL,
+				  dev->priv.numa_node);
+	if (!irq_table)
+		return -ENOMEM;
+
+	dev->priv.irq_table = irq_table;
+	return 0;
+}
+
+void mlx5_irq_table_cleanup(struct mlx5_core_dev *dev)
+{
+	if (mlx5_core_is_sf(dev))
+		return;
+
+	kvfree(dev->priv.irq_table);
+}
+
+int mlx5_irq_table_get_num_comp(struct mlx5_irq_table *table)
+{
+	if (!table->pf_pool->xa_num_irqs.max)
+		return 1;
+	return table->pf_pool->xa_num_irqs.max - table->pf_pool->xa_num_irqs.min;
+}
+
+int mlx5_irq_table_create(struct mlx5_core_dev *dev)
+{
+	int num_eqs = MLX5_CAP_GEN(dev, max_num_eqs) ?
+		      MLX5_CAP_GEN(dev, max_num_eqs) :
+		      1 << MLX5_CAP_GEN(dev, log_max_eq);
+	int total_vec;
+	int pf_vec;
+	int err;
+
+	if (mlx5_core_is_sf(dev))
+		return 0;
+
+	pf_vec = MLX5_CAP_GEN(dev, num_ports) * num_online_cpus() + 1;
+	pf_vec = min_t(int, pf_vec, num_eqs);
+
+	total_vec = pf_vec;
+	if (mlx5_sf_max_functions(dev))
+		total_vec += MLX5_IRQ_CTRL_SF_MAX +
+			MLX5_COMP_EQS_PER_SF * mlx5_sf_max_functions(dev);
+
+	total_vec = pci_alloc_irq_vectors(dev->pdev, 1, total_vec, PCI_IRQ_MSIX);
+	if (total_vec < 0)
+		return total_vec;
+	pf_vec = min(pf_vec, total_vec);
+
+	err = irq_pools_init(dev, total_vec - pf_vec, pf_vec);
+	if (err)
+		pci_free_irq_vectors(dev->pdev);
+
+	return err;
+}
+
+void mlx5_irq_table_destroy(struct mlx5_core_dev *dev)
+{
+	struct mlx5_irq_table *table = dev->priv.irq_table;
+
+	if (mlx5_core_is_sf(dev))
+		return;
+
+	/* There are cases where IRQs still will be in used when we reaching
+	 * to here. Hence, making sure all the irqs are released.
+	 */
+	irq_pools_destroy(table);
+	pci_free_irq_vectors(dev->pdev);
+}
+
+void mlx5_irq_table_free_irqs(struct mlx5_core_dev *dev)
+{
+	struct mlx5_irq_table *table = dev->priv.irq_table;
+
+	if (mlx5_core_is_sf(dev))
+		return;
+
+	mlx5_irq_pools_free_irqs(table);
+	pci_free_irq_vectors(dev->pdev);
+}
+
+int mlx5_irq_table_get_sfs_vec(struct mlx5_irq_table *table)
+{
+	if (table->sf_comp_pool)
+		return min_t(int, num_online_cpus(),
+			     table->sf_comp_pool->xa_num_irqs.max -
+			     table->sf_comp_pool->xa_num_irqs.min + 1);
+	else
+		return mlx5_irq_table_get_num_comp(table);
+}
+
+struct mlx5_irq_table *mlx5_irq_table_get(struct mlx5_core_dev *dev)
+{
+#ifdef CONFIG_MLX5_SF
+	if (mlx5_core_is_sf(dev))
+		return dev->priv.parent_mdev->priv.irq_table;
+#endif
+	return dev->priv.irq_table;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/pci_irq.h b/drivers/net/ethernet/mellanox/mlx5/core/pci_irq.h
new file mode 100644
index 000000000..404717930
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/pci_irq.h
@@ -0,0 +1,42 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. All rights reserved. */
+
+#ifndef __PCI_IRQ_H__
+#define __PCI_IRQ_H__
+
+#include <linux/mlx5/driver.h>
+
+#define MLX5_MAX_IRQ_NAME (32)
+#define MLX5_IRQ_NAME_FORMAT_STR ("%s@pci:%s")
+#define MLX5_MAX_IRQ_FORMATTED_NAME \
+	(MLX5_MAX_IRQ_NAME + sizeof(MLX5_IRQ_NAME_FORMAT_STR))
+/* max irq_index is 2047, so four chars */
+#define MLX5_MAX_IRQ_IDX_CHARS (4)
+#define MLX5_EQ_REFS_PER_IRQ (2)
+
+struct mlx5_irq;
+
+struct mlx5_irq_pool {
+	char name[MLX5_MAX_IRQ_NAME - MLX5_MAX_IRQ_IDX_CHARS];
+	struct xa_limit xa_num_irqs;
+	struct mutex lock; /* sync IRQs creations */
+	struct xarray irqs;
+	u32 max_threshold;
+	u32 min_threshold;
+	u16 *irqs_per_cpu;
+	struct mlx5_core_dev *dev;
+};
+
+struct mlx5_irq_pool *mlx5_irq_pool_get(struct mlx5_core_dev *dev);
+static inline bool mlx5_irq_pool_is_sf_pool(struct mlx5_irq_pool *pool)
+{
+	return !strncmp("mlx5_sf", pool->name, strlen("mlx5_sf"));
+}
+
+struct mlx5_irq *mlx5_irq_alloc(struct mlx5_irq_pool *pool, int i,
+				const struct cpumask *affinity);
+int mlx5_irq_get_locked(struct mlx5_irq *irq);
+int mlx5_irq_read_locked(struct mlx5_irq *irq);
+int mlx5_irq_put(struct mlx5_irq *irq);
+
+#endif /* __PCI_IRQ_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/pd.c b/drivers/net/ethernet/mellanox/mlx5/core/pd.c
new file mode 100644
index 000000000..ee5ffdeb9
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/pd.c
@@ -0,0 +1,59 @@
+/*
+ * Copyright (c) 2013-2015, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/kernel.h>
+#include <linux/mlx5/driver.h>
+#include "mlx5_core.h"
+
+int mlx5_core_alloc_pd(struct mlx5_core_dev *dev, u32 *pdn)
+{
+	u32 out[MLX5_ST_SZ_DW(alloc_pd_out)] = {};
+	u32 in[MLX5_ST_SZ_DW(alloc_pd_in)] = {};
+	int err;
+
+	MLX5_SET(alloc_pd_in, in, opcode, MLX5_CMD_OP_ALLOC_PD);
+	err = mlx5_cmd_exec_inout(dev, alloc_pd, in, out);
+	if (!err)
+		*pdn = MLX5_GET(alloc_pd_out, out, pd);
+	return err;
+}
+EXPORT_SYMBOL(mlx5_core_alloc_pd);
+
+int mlx5_core_dealloc_pd(struct mlx5_core_dev *dev, u32 pdn)
+{
+	u32 in[MLX5_ST_SZ_DW(dealloc_pd_in)] = {};
+
+	MLX5_SET(dealloc_pd_in, in, opcode, MLX5_CMD_OP_DEALLOC_PD);
+	MLX5_SET(dealloc_pd_in, in, pd, pdn);
+	return mlx5_cmd_exec_in(dev, dealloc_pd, in);
+}
+EXPORT_SYMBOL(mlx5_core_dealloc_pd);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/port.c b/drivers/net/ethernet/mellanox/mlx5/core/port.c
new file mode 100644
index 000000000..a1548e6bf
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/port.c
@@ -0,0 +1,1056 @@
+/*
+ * Copyright (c) 2013-2015, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/mlx5/port.h>
+#include "mlx5_core.h"
+
+/* calling with verbose false will not print error to log */
+int mlx5_access_reg(struct mlx5_core_dev *dev, void *data_in, int size_in,
+		    void *data_out, int size_out, u16 reg_id, int arg,
+		    int write, bool verbose)
+{
+	int outlen = MLX5_ST_SZ_BYTES(access_register_out) + size_out;
+	int inlen = MLX5_ST_SZ_BYTES(access_register_in) + size_in;
+	int err = -ENOMEM;
+	u32 *out = NULL;
+	u32 *in = NULL;
+	void *data;
+
+	in = kvzalloc(inlen, GFP_KERNEL);
+	out = kvzalloc(outlen, GFP_KERNEL);
+	if (!in || !out)
+		goto out;
+
+	data = MLX5_ADDR_OF(access_register_in, in, register_data);
+	memcpy(data, data_in, size_in);
+
+	MLX5_SET(access_register_in, in, opcode, MLX5_CMD_OP_ACCESS_REG);
+	MLX5_SET(access_register_in, in, op_mod, !write);
+	MLX5_SET(access_register_in, in, argument, arg);
+	MLX5_SET(access_register_in, in, register_id, reg_id);
+
+	err = mlx5_cmd_do(dev, in, inlen, out, outlen);
+	if (verbose)
+		err = mlx5_cmd_check(dev, err, in, out);
+	if (err)
+		goto out;
+
+	data = MLX5_ADDR_OF(access_register_out, out, register_data);
+	memcpy(data_out, data, size_out);
+
+out:
+	kvfree(out);
+	kvfree(in);
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx5_access_reg);
+
+int mlx5_core_access_reg(struct mlx5_core_dev *dev, void *data_in,
+			 int size_in, void *data_out, int size_out,
+			 u16 reg_id, int arg, int write)
+{
+	return mlx5_access_reg(dev, data_in, size_in, data_out, size_out,
+			       reg_id, arg, write, true);
+}
+EXPORT_SYMBOL_GPL(mlx5_core_access_reg);
+
+int mlx5_query_pcam_reg(struct mlx5_core_dev *dev, u32 *pcam, u8 feature_group,
+			u8 access_reg_group)
+{
+	u32 in[MLX5_ST_SZ_DW(pcam_reg)] = {0};
+	int sz = MLX5_ST_SZ_BYTES(pcam_reg);
+
+	MLX5_SET(pcam_reg, in, feature_group, feature_group);
+	MLX5_SET(pcam_reg, in, access_reg_group, access_reg_group);
+
+	return mlx5_core_access_reg(dev, in, sz, pcam, sz, MLX5_REG_PCAM, 0, 0);
+}
+
+int mlx5_query_mcam_reg(struct mlx5_core_dev *dev, u32 *mcam, u8 feature_group,
+			u8 access_reg_group)
+{
+	u32 in[MLX5_ST_SZ_DW(mcam_reg)] = {0};
+	int sz = MLX5_ST_SZ_BYTES(mcam_reg);
+
+	MLX5_SET(mcam_reg, in, feature_group, feature_group);
+	MLX5_SET(mcam_reg, in, access_reg_group, access_reg_group);
+
+	return mlx5_core_access_reg(dev, in, sz, mcam, sz, MLX5_REG_MCAM, 0, 0);
+}
+
+int mlx5_query_qcam_reg(struct mlx5_core_dev *mdev, u32 *qcam,
+			u8 feature_group, u8 access_reg_group)
+{
+	u32 in[MLX5_ST_SZ_DW(qcam_reg)] = {};
+	int sz = MLX5_ST_SZ_BYTES(qcam_reg);
+
+	MLX5_SET(qcam_reg, in, feature_group, feature_group);
+	MLX5_SET(qcam_reg, in, access_reg_group, access_reg_group);
+
+	return mlx5_core_access_reg(mdev, in, sz, qcam, sz, MLX5_REG_QCAM, 0, 0);
+}
+
+struct mlx5_reg_pcap {
+	u8			rsvd0;
+	u8			port_num;
+	u8			rsvd1[2];
+	__be32			caps_127_96;
+	__be32			caps_95_64;
+	__be32			caps_63_32;
+	__be32			caps_31_0;
+};
+
+int mlx5_set_port_caps(struct mlx5_core_dev *dev, u8 port_num, u32 caps)
+{
+	struct mlx5_reg_pcap in;
+	struct mlx5_reg_pcap out;
+
+	memset(&in, 0, sizeof(in));
+	in.caps_127_96 = cpu_to_be32(caps);
+	in.port_num = port_num;
+
+	return mlx5_core_access_reg(dev, &in, sizeof(in), &out,
+				    sizeof(out), MLX5_REG_PCAP, 0, 1);
+}
+EXPORT_SYMBOL_GPL(mlx5_set_port_caps);
+
+int mlx5_query_port_ptys(struct mlx5_core_dev *dev, u32 *ptys,
+			 int ptys_size, int proto_mask, u8 local_port)
+{
+	u32 in[MLX5_ST_SZ_DW(ptys_reg)] = {0};
+
+	MLX5_SET(ptys_reg, in, local_port, local_port);
+	MLX5_SET(ptys_reg, in, proto_mask, proto_mask);
+	return mlx5_core_access_reg(dev, in, sizeof(in), ptys,
+				    ptys_size, MLX5_REG_PTYS, 0, 0);
+}
+EXPORT_SYMBOL_GPL(mlx5_query_port_ptys);
+
+int mlx5_set_port_beacon(struct mlx5_core_dev *dev, u16 beacon_duration)
+{
+	u32 in[MLX5_ST_SZ_DW(mlcr_reg)]  = {0};
+	u32 out[MLX5_ST_SZ_DW(mlcr_reg)];
+
+	MLX5_SET(mlcr_reg, in, local_port, 1);
+	MLX5_SET(mlcr_reg, in, beacon_duration, beacon_duration);
+	return mlx5_core_access_reg(dev, in, sizeof(in), out,
+				    sizeof(out), MLX5_REG_MLCR, 0, 1);
+}
+
+int mlx5_query_ib_port_oper(struct mlx5_core_dev *dev, u16 *link_width_oper,
+			    u16 *proto_oper, u8 local_port)
+{
+	u32 out[MLX5_ST_SZ_DW(ptys_reg)];
+	int err;
+
+	err = mlx5_query_port_ptys(dev, out, sizeof(out), MLX5_PTYS_IB,
+				   local_port);
+	if (err)
+		return err;
+
+	*link_width_oper = MLX5_GET(ptys_reg, out, ib_link_width_oper);
+	*proto_oper = MLX5_GET(ptys_reg, out, ib_proto_oper);
+
+	return 0;
+}
+EXPORT_SYMBOL(mlx5_query_ib_port_oper);
+
+/* This function should be used after setting a port register only */
+void mlx5_toggle_port_link(struct mlx5_core_dev *dev)
+{
+	enum mlx5_port_status ps;
+
+	mlx5_query_port_admin_status(dev, &ps);
+	mlx5_set_port_admin_status(dev, MLX5_PORT_DOWN);
+	if (ps == MLX5_PORT_UP)
+		mlx5_set_port_admin_status(dev, MLX5_PORT_UP);
+}
+EXPORT_SYMBOL_GPL(mlx5_toggle_port_link);
+
+int mlx5_set_port_admin_status(struct mlx5_core_dev *dev,
+			       enum mlx5_port_status status)
+{
+	u32 in[MLX5_ST_SZ_DW(paos_reg)] = {0};
+	u32 out[MLX5_ST_SZ_DW(paos_reg)];
+
+	MLX5_SET(paos_reg, in, local_port, 1);
+	MLX5_SET(paos_reg, in, admin_status, status);
+	MLX5_SET(paos_reg, in, ase, 1);
+	return mlx5_core_access_reg(dev, in, sizeof(in), out,
+				    sizeof(out), MLX5_REG_PAOS, 0, 1);
+}
+EXPORT_SYMBOL_GPL(mlx5_set_port_admin_status);
+
+int mlx5_query_port_admin_status(struct mlx5_core_dev *dev,
+				 enum mlx5_port_status *status)
+{
+	u32 in[MLX5_ST_SZ_DW(paos_reg)] = {0};
+	u32 out[MLX5_ST_SZ_DW(paos_reg)];
+	int err;
+
+	MLX5_SET(paos_reg, in, local_port, 1);
+	err = mlx5_core_access_reg(dev, in, sizeof(in), out,
+				   sizeof(out), MLX5_REG_PAOS, 0, 0);
+	if (err)
+		return err;
+	*status = MLX5_GET(paos_reg, out, admin_status);
+	return 0;
+}
+EXPORT_SYMBOL_GPL(mlx5_query_port_admin_status);
+
+static void mlx5_query_port_mtu(struct mlx5_core_dev *dev, u16 *admin_mtu,
+				u16 *max_mtu, u16 *oper_mtu, u8 port)
+{
+	u32 in[MLX5_ST_SZ_DW(pmtu_reg)] = {0};
+	u32 out[MLX5_ST_SZ_DW(pmtu_reg)];
+
+	MLX5_SET(pmtu_reg, in, local_port, port);
+	mlx5_core_access_reg(dev, in, sizeof(in), out,
+			     sizeof(out), MLX5_REG_PMTU, 0, 0);
+
+	if (max_mtu)
+		*max_mtu  = MLX5_GET(pmtu_reg, out, max_mtu);
+	if (oper_mtu)
+		*oper_mtu = MLX5_GET(pmtu_reg, out, oper_mtu);
+	if (admin_mtu)
+		*admin_mtu = MLX5_GET(pmtu_reg, out, admin_mtu);
+}
+
+int mlx5_set_port_mtu(struct mlx5_core_dev *dev, u16 mtu, u8 port)
+{
+	u32 in[MLX5_ST_SZ_DW(pmtu_reg)] = {0};
+	u32 out[MLX5_ST_SZ_DW(pmtu_reg)];
+
+	MLX5_SET(pmtu_reg, in, admin_mtu, mtu);
+	MLX5_SET(pmtu_reg, in, local_port, port);
+	return mlx5_core_access_reg(dev, in, sizeof(in), out,
+				   sizeof(out), MLX5_REG_PMTU, 0, 1);
+}
+EXPORT_SYMBOL_GPL(mlx5_set_port_mtu);
+
+void mlx5_query_port_max_mtu(struct mlx5_core_dev *dev, u16 *max_mtu,
+			     u8 port)
+{
+	mlx5_query_port_mtu(dev, NULL, max_mtu, NULL, port);
+}
+EXPORT_SYMBOL_GPL(mlx5_query_port_max_mtu);
+
+void mlx5_query_port_oper_mtu(struct mlx5_core_dev *dev, u16 *oper_mtu,
+			      u8 port)
+{
+	mlx5_query_port_mtu(dev, NULL, NULL, oper_mtu, port);
+}
+EXPORT_SYMBOL_GPL(mlx5_query_port_oper_mtu);
+
+static int mlx5_query_module_num(struct mlx5_core_dev *dev, int *module_num)
+{
+	u32 in[MLX5_ST_SZ_DW(pmlp_reg)] = {0};
+	u32 out[MLX5_ST_SZ_DW(pmlp_reg)];
+	int err;
+
+	MLX5_SET(pmlp_reg, in, local_port, 1);
+	err = mlx5_core_access_reg(dev, in, sizeof(in), out, sizeof(out),
+				   MLX5_REG_PMLP, 0, 0);
+	if (err)
+		return err;
+
+	*module_num = MLX5_GET(lane_2_module_mapping,
+			       MLX5_ADDR_OF(pmlp_reg, out, lane0_module_mapping),
+			       module);
+
+	return 0;
+}
+
+static int mlx5_query_module_id(struct mlx5_core_dev *dev, int module_num,
+				u8 *module_id)
+{
+	u32 in[MLX5_ST_SZ_DW(mcia_reg)] = {};
+	u32 out[MLX5_ST_SZ_DW(mcia_reg)];
+	int err, status;
+	u8 *ptr;
+
+	MLX5_SET(mcia_reg, in, i2c_device_address, MLX5_I2C_ADDR_LOW);
+	MLX5_SET(mcia_reg, in, module, module_num);
+	MLX5_SET(mcia_reg, in, device_address, 0);
+	MLX5_SET(mcia_reg, in, page_number, 0);
+	MLX5_SET(mcia_reg, in, size, 1);
+	MLX5_SET(mcia_reg, in, l, 0);
+
+	err = mlx5_core_access_reg(dev, in, sizeof(in), out,
+				   sizeof(out), MLX5_REG_MCIA, 0, 0);
+	if (err)
+		return err;
+
+	status = MLX5_GET(mcia_reg, out, status);
+	if (status) {
+		mlx5_core_err(dev, "query_mcia_reg failed: status: 0x%x\n",
+			      status);
+		return -EIO;
+	}
+	ptr = MLX5_ADDR_OF(mcia_reg, out, dword_0);
+
+	*module_id = ptr[0];
+
+	return 0;
+}
+
+static int mlx5_qsfp_eeprom_page(u16 offset)
+{
+	if (offset < MLX5_EEPROM_PAGE_LENGTH)
+		/* Addresses between 0-255 - page 00 */
+		return 0;
+
+	/* Addresses between 256 - 639 belongs to pages 01, 02 and 03
+	 * For example, offset = 400 belongs to page 02:
+	 * 1 + ((400 - 256)/128) = 2
+	 */
+	return 1 + ((offset - MLX5_EEPROM_PAGE_LENGTH) /
+		    MLX5_EEPROM_HIGH_PAGE_LENGTH);
+}
+
+static int mlx5_qsfp_eeprom_high_page_offset(int page_num)
+{
+	if (!page_num) /* Page 0 always start from low page */
+		return 0;
+
+	/* High page */
+	return page_num * MLX5_EEPROM_HIGH_PAGE_LENGTH;
+}
+
+static void mlx5_qsfp_eeprom_params_set(u16 *i2c_addr, int *page_num, u16 *offset)
+{
+	*i2c_addr = MLX5_I2C_ADDR_LOW;
+	*page_num = mlx5_qsfp_eeprom_page(*offset);
+	*offset -=  mlx5_qsfp_eeprom_high_page_offset(*page_num);
+}
+
+static void mlx5_sfp_eeprom_params_set(u16 *i2c_addr, int *page_num, u16 *offset)
+{
+	*i2c_addr = MLX5_I2C_ADDR_LOW;
+	*page_num = 0;
+
+	if (*offset < MLX5_EEPROM_PAGE_LENGTH)
+		return;
+
+	*i2c_addr = MLX5_I2C_ADDR_HIGH;
+	*offset -= MLX5_EEPROM_PAGE_LENGTH;
+}
+
+static int mlx5_mcia_max_bytes(struct mlx5_core_dev *dev)
+{
+	/* mcia supports either 12 dwords or 32 dwords */
+	return (MLX5_CAP_MCAM_FEATURE(dev, mcia_32dwords) ? 32 : 12) * sizeof(u32);
+}
+
+static int mlx5_query_mcia(struct mlx5_core_dev *dev,
+			   struct mlx5_module_eeprom_query_params *params, u8 *data)
+{
+	u32 in[MLX5_ST_SZ_DW(mcia_reg)] = {};
+	u32 out[MLX5_ST_SZ_DW(mcia_reg)];
+	int status, err;
+	void *ptr;
+	u16 size;
+
+	size = min_t(int, params->size, mlx5_mcia_max_bytes(dev));
+
+	MLX5_SET(mcia_reg, in, l, 0);
+	MLX5_SET(mcia_reg, in, size, size);
+	MLX5_SET(mcia_reg, in, module, params->module_number);
+	MLX5_SET(mcia_reg, in, device_address, params->offset);
+	MLX5_SET(mcia_reg, in, page_number, params->page);
+	MLX5_SET(mcia_reg, in, i2c_device_address, params->i2c_address);
+
+	err = mlx5_core_access_reg(dev, in, sizeof(in), out,
+				   sizeof(out), MLX5_REG_MCIA, 0, 0);
+	if (err)
+		return err;
+
+	status = MLX5_GET(mcia_reg, out, status);
+	if (status) {
+		mlx5_core_err(dev, "query_mcia_reg failed: status: 0x%x\n",
+			      status);
+		return -EIO;
+	}
+
+	ptr = MLX5_ADDR_OF(mcia_reg, out, dword_0);
+	memcpy(data, ptr, size);
+
+	return size;
+}
+
+int mlx5_query_module_eeprom(struct mlx5_core_dev *dev,
+			     u16 offset, u16 size, u8 *data)
+{
+	struct mlx5_module_eeprom_query_params query = {0};
+	u8 module_id;
+	int err;
+
+	err = mlx5_query_module_num(dev, &query.module_number);
+	if (err)
+		return err;
+
+	err = mlx5_query_module_id(dev, query.module_number, &module_id);
+	if (err)
+		return err;
+
+	switch (module_id) {
+	case MLX5_MODULE_ID_SFP:
+		mlx5_sfp_eeprom_params_set(&query.i2c_address, &query.page, &offset);
+		break;
+	case MLX5_MODULE_ID_QSFP:
+	case MLX5_MODULE_ID_QSFP_PLUS:
+	case MLX5_MODULE_ID_QSFP28:
+		mlx5_qsfp_eeprom_params_set(&query.i2c_address, &query.page, &offset);
+		break;
+	default:
+		mlx5_core_err(dev, "Module ID not recognized: 0x%x\n", module_id);
+		return -EINVAL;
+	}
+
+	if (offset + size > MLX5_EEPROM_PAGE_LENGTH)
+		/* Cross pages read, read until offset 256 in low page */
+		size = MLX5_EEPROM_PAGE_LENGTH - offset;
+
+	query.size = size;
+	query.offset = offset;
+
+	return mlx5_query_mcia(dev, &query, data);
+}
+EXPORT_SYMBOL_GPL(mlx5_query_module_eeprom);
+
+int mlx5_query_module_eeprom_by_page(struct mlx5_core_dev *dev,
+				     struct mlx5_module_eeprom_query_params *params,
+				     u8 *data)
+{
+	int err;
+
+	err = mlx5_query_module_num(dev, &params->module_number);
+	if (err)
+		return err;
+
+	if (params->i2c_address != MLX5_I2C_ADDR_HIGH &&
+	    params->i2c_address != MLX5_I2C_ADDR_LOW) {
+		mlx5_core_err(dev, "I2C address not recognized: 0x%x\n", params->i2c_address);
+		return -EINVAL;
+	}
+
+	return mlx5_query_mcia(dev, params, data);
+}
+EXPORT_SYMBOL_GPL(mlx5_query_module_eeprom_by_page);
+
+static int mlx5_query_port_pvlc(struct mlx5_core_dev *dev, u32 *pvlc,
+				int pvlc_size,  u8 local_port)
+{
+	u32 in[MLX5_ST_SZ_DW(pvlc_reg)] = {0};
+
+	MLX5_SET(pvlc_reg, in, local_port, local_port);
+	return mlx5_core_access_reg(dev, in, sizeof(in), pvlc,
+				    pvlc_size, MLX5_REG_PVLC, 0, 0);
+}
+
+int mlx5_query_port_vl_hw_cap(struct mlx5_core_dev *dev,
+			      u8 *vl_hw_cap, u8 local_port)
+{
+	u32 out[MLX5_ST_SZ_DW(pvlc_reg)];
+	int err;
+
+	err = mlx5_query_port_pvlc(dev, out, sizeof(out), local_port);
+	if (err)
+		return err;
+
+	*vl_hw_cap = MLX5_GET(pvlc_reg, out, vl_hw_cap);
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(mlx5_query_port_vl_hw_cap);
+
+static int mlx5_query_pfcc_reg(struct mlx5_core_dev *dev, u32 *out,
+			       u32 out_size)
+{
+	u32 in[MLX5_ST_SZ_DW(pfcc_reg)] = {0};
+
+	MLX5_SET(pfcc_reg, in, local_port, 1);
+
+	return mlx5_core_access_reg(dev, in, sizeof(in), out,
+				    out_size, MLX5_REG_PFCC, 0, 0);
+}
+
+int mlx5_set_port_pause(struct mlx5_core_dev *dev, u32 rx_pause, u32 tx_pause)
+{
+	u32 in[MLX5_ST_SZ_DW(pfcc_reg)] = {0};
+	u32 out[MLX5_ST_SZ_DW(pfcc_reg)];
+
+	MLX5_SET(pfcc_reg, in, local_port, 1);
+	MLX5_SET(pfcc_reg, in, pptx, tx_pause);
+	MLX5_SET(pfcc_reg, in, pprx, rx_pause);
+
+	return mlx5_core_access_reg(dev, in, sizeof(in), out,
+				    sizeof(out), MLX5_REG_PFCC, 0, 1);
+}
+EXPORT_SYMBOL_GPL(mlx5_set_port_pause);
+
+int mlx5_query_port_pause(struct mlx5_core_dev *dev,
+			  u32 *rx_pause, u32 *tx_pause)
+{
+	u32 out[MLX5_ST_SZ_DW(pfcc_reg)];
+	int err;
+
+	err = mlx5_query_pfcc_reg(dev, out, sizeof(out));
+	if (err)
+		return err;
+
+	if (rx_pause)
+		*rx_pause = MLX5_GET(pfcc_reg, out, pprx);
+
+	if (tx_pause)
+		*tx_pause = MLX5_GET(pfcc_reg, out, pptx);
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(mlx5_query_port_pause);
+
+int mlx5_set_port_stall_watermark(struct mlx5_core_dev *dev,
+				  u16 stall_critical_watermark,
+				  u16 stall_minor_watermark)
+{
+	u32 in[MLX5_ST_SZ_DW(pfcc_reg)] = {0};
+	u32 out[MLX5_ST_SZ_DW(pfcc_reg)];
+
+	MLX5_SET(pfcc_reg, in, local_port, 1);
+	MLX5_SET(pfcc_reg, in, pptx_mask_n, 1);
+	MLX5_SET(pfcc_reg, in, pprx_mask_n, 1);
+	MLX5_SET(pfcc_reg, in, ppan_mask_n, 1);
+	MLX5_SET(pfcc_reg, in, critical_stall_mask, 1);
+	MLX5_SET(pfcc_reg, in, minor_stall_mask, 1);
+	MLX5_SET(pfcc_reg, in, device_stall_critical_watermark,
+		 stall_critical_watermark);
+	MLX5_SET(pfcc_reg, in, device_stall_minor_watermark, stall_minor_watermark);
+
+	return mlx5_core_access_reg(dev, in, sizeof(in), out,
+				    sizeof(out), MLX5_REG_PFCC, 0, 1);
+}
+
+int mlx5_query_port_stall_watermark(struct mlx5_core_dev *dev,
+				    u16 *stall_critical_watermark,
+				    u16 *stall_minor_watermark)
+{
+	u32 out[MLX5_ST_SZ_DW(pfcc_reg)];
+	int err;
+
+	err = mlx5_query_pfcc_reg(dev, out, sizeof(out));
+	if (err)
+		return err;
+
+	if (stall_critical_watermark)
+		*stall_critical_watermark = MLX5_GET(pfcc_reg, out,
+						     device_stall_critical_watermark);
+
+	if (stall_minor_watermark)
+		*stall_minor_watermark = MLX5_GET(pfcc_reg, out,
+						  device_stall_minor_watermark);
+
+	return 0;
+}
+
+int mlx5_set_port_pfc(struct mlx5_core_dev *dev, u8 pfc_en_tx, u8 pfc_en_rx)
+{
+	u32 in[MLX5_ST_SZ_DW(pfcc_reg)] = {0};
+	u32 out[MLX5_ST_SZ_DW(pfcc_reg)];
+
+	MLX5_SET(pfcc_reg, in, local_port, 1);
+	MLX5_SET(pfcc_reg, in, pfctx, pfc_en_tx);
+	MLX5_SET(pfcc_reg, in, pfcrx, pfc_en_rx);
+	MLX5_SET_TO_ONES(pfcc_reg, in, prio_mask_tx);
+	MLX5_SET_TO_ONES(pfcc_reg, in, prio_mask_rx);
+
+	return mlx5_core_access_reg(dev, in, sizeof(in), out,
+				    sizeof(out), MLX5_REG_PFCC, 0, 1);
+}
+EXPORT_SYMBOL_GPL(mlx5_set_port_pfc);
+
+int mlx5_query_port_pfc(struct mlx5_core_dev *dev, u8 *pfc_en_tx, u8 *pfc_en_rx)
+{
+	u32 out[MLX5_ST_SZ_DW(pfcc_reg)];
+	int err;
+
+	err = mlx5_query_pfcc_reg(dev, out, sizeof(out));
+	if (err)
+		return err;
+
+	if (pfc_en_tx)
+		*pfc_en_tx = MLX5_GET(pfcc_reg, out, pfctx);
+
+	if (pfc_en_rx)
+		*pfc_en_rx = MLX5_GET(pfcc_reg, out, pfcrx);
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(mlx5_query_port_pfc);
+
+int mlx5_max_tc(struct mlx5_core_dev *mdev)
+{
+	u8 num_tc = MLX5_CAP_GEN(mdev, max_tc) ? : 8;
+
+	return num_tc - 1;
+}
+
+int mlx5_query_port_dcbx_param(struct mlx5_core_dev *mdev, u32 *out)
+{
+	u32 in[MLX5_ST_SZ_DW(dcbx_param)] = {0};
+
+	MLX5_SET(dcbx_param, in, port_number, 1);
+
+	return  mlx5_core_access_reg(mdev, in, sizeof(in), out,
+				    sizeof(in), MLX5_REG_DCBX_PARAM, 0, 0);
+}
+
+int mlx5_set_port_dcbx_param(struct mlx5_core_dev *mdev, u32 *in)
+{
+	u32 out[MLX5_ST_SZ_DW(dcbx_param)];
+
+	MLX5_SET(dcbx_param, in, port_number, 1);
+
+	return mlx5_core_access_reg(mdev, in, sizeof(out), out,
+				    sizeof(out), MLX5_REG_DCBX_PARAM, 0, 1);
+}
+
+int mlx5_set_port_prio_tc(struct mlx5_core_dev *mdev, u8 *prio_tc)
+{
+	u32 in[MLX5_ST_SZ_DW(qtct_reg)] = {0};
+	u32 out[MLX5_ST_SZ_DW(qtct_reg)];
+	int err;
+	int i;
+
+	for (i = 0; i < 8; i++) {
+		if (prio_tc[i] > mlx5_max_tc(mdev))
+			return -EINVAL;
+
+		MLX5_SET(qtct_reg, in, prio, i);
+		MLX5_SET(qtct_reg, in, tclass, prio_tc[i]);
+
+		err = mlx5_core_access_reg(mdev, in, sizeof(in), out,
+					   sizeof(out), MLX5_REG_QTCT, 0, 1);
+		if (err)
+			return err;
+	}
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(mlx5_set_port_prio_tc);
+
+int mlx5_query_port_prio_tc(struct mlx5_core_dev *mdev,
+			    u8 prio, u8 *tc)
+{
+	u32 in[MLX5_ST_SZ_DW(qtct_reg)];
+	u32 out[MLX5_ST_SZ_DW(qtct_reg)];
+	int err;
+
+	memset(in, 0, sizeof(in));
+	memset(out, 0, sizeof(out));
+
+	MLX5_SET(qtct_reg, in, port_number, 1);
+	MLX5_SET(qtct_reg, in, prio, prio);
+
+	err = mlx5_core_access_reg(mdev, in, sizeof(in), out,
+				   sizeof(out), MLX5_REG_QTCT, 0, 0);
+	if (!err)
+		*tc = MLX5_GET(qtct_reg, out, tclass);
+
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx5_query_port_prio_tc);
+
+static int mlx5_set_port_qetcr_reg(struct mlx5_core_dev *mdev, u32 *in,
+				   int inlen)
+{
+	u32 out[MLX5_ST_SZ_DW(qetc_reg)];
+
+	if (!MLX5_CAP_GEN(mdev, ets))
+		return -EOPNOTSUPP;
+
+	return mlx5_core_access_reg(mdev, in, inlen, out, sizeof(out),
+				    MLX5_REG_QETCR, 0, 1);
+}
+
+static int mlx5_query_port_qetcr_reg(struct mlx5_core_dev *mdev, u32 *out,
+				     int outlen)
+{
+	u32 in[MLX5_ST_SZ_DW(qetc_reg)];
+
+	if (!MLX5_CAP_GEN(mdev, ets))
+		return -EOPNOTSUPP;
+
+	memset(in, 0, sizeof(in));
+	return mlx5_core_access_reg(mdev, in, sizeof(in), out, outlen,
+				    MLX5_REG_QETCR, 0, 0);
+}
+
+int mlx5_set_port_tc_group(struct mlx5_core_dev *mdev, u8 *tc_group)
+{
+	u32 in[MLX5_ST_SZ_DW(qetc_reg)] = {0};
+	int i;
+
+	for (i = 0; i <= mlx5_max_tc(mdev); i++) {
+		MLX5_SET(qetc_reg, in, tc_configuration[i].g, 1);
+		MLX5_SET(qetc_reg, in, tc_configuration[i].group, tc_group[i]);
+	}
+
+	return mlx5_set_port_qetcr_reg(mdev, in, sizeof(in));
+}
+EXPORT_SYMBOL_GPL(mlx5_set_port_tc_group);
+
+int mlx5_query_port_tc_group(struct mlx5_core_dev *mdev,
+			     u8 tc, u8 *tc_group)
+{
+	u32 out[MLX5_ST_SZ_DW(qetc_reg)];
+	void *ets_tcn_conf;
+	int err;
+
+	err = mlx5_query_port_qetcr_reg(mdev, out, sizeof(out));
+	if (err)
+		return err;
+
+	ets_tcn_conf = MLX5_ADDR_OF(qetc_reg, out,
+				    tc_configuration[tc]);
+
+	*tc_group = MLX5_GET(ets_tcn_config_reg, ets_tcn_conf,
+			     group);
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(mlx5_query_port_tc_group);
+
+int mlx5_set_port_tc_bw_alloc(struct mlx5_core_dev *mdev, u8 *tc_bw)
+{
+	u32 in[MLX5_ST_SZ_DW(qetc_reg)] = {0};
+	int i;
+
+	for (i = 0; i <= mlx5_max_tc(mdev); i++) {
+		MLX5_SET(qetc_reg, in, tc_configuration[i].b, 1);
+		MLX5_SET(qetc_reg, in, tc_configuration[i].bw_allocation, tc_bw[i]);
+	}
+
+	return mlx5_set_port_qetcr_reg(mdev, in, sizeof(in));
+}
+EXPORT_SYMBOL_GPL(mlx5_set_port_tc_bw_alloc);
+
+int mlx5_query_port_tc_bw_alloc(struct mlx5_core_dev *mdev,
+				u8 tc, u8 *bw_pct)
+{
+	u32 out[MLX5_ST_SZ_DW(qetc_reg)];
+	void *ets_tcn_conf;
+	int err;
+
+	err = mlx5_query_port_qetcr_reg(mdev, out, sizeof(out));
+	if (err)
+		return err;
+
+	ets_tcn_conf = MLX5_ADDR_OF(qetc_reg, out,
+				    tc_configuration[tc]);
+
+	*bw_pct = MLX5_GET(ets_tcn_config_reg, ets_tcn_conf,
+			   bw_allocation);
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(mlx5_query_port_tc_bw_alloc);
+
+int mlx5_modify_port_ets_rate_limit(struct mlx5_core_dev *mdev,
+				    u8 *max_bw_value,
+				    u8 *max_bw_units)
+{
+	u32 in[MLX5_ST_SZ_DW(qetc_reg)] = {0};
+	void *ets_tcn_conf;
+	int i;
+
+	MLX5_SET(qetc_reg, in, port_number, 1);
+
+	for (i = 0; i <= mlx5_max_tc(mdev); i++) {
+		ets_tcn_conf = MLX5_ADDR_OF(qetc_reg, in, tc_configuration[i]);
+
+		MLX5_SET(ets_tcn_config_reg, ets_tcn_conf, r, 1);
+		MLX5_SET(ets_tcn_config_reg, ets_tcn_conf, max_bw_units,
+			 max_bw_units[i]);
+		MLX5_SET(ets_tcn_config_reg, ets_tcn_conf, max_bw_value,
+			 max_bw_value[i]);
+	}
+
+	return mlx5_set_port_qetcr_reg(mdev, in, sizeof(in));
+}
+EXPORT_SYMBOL_GPL(mlx5_modify_port_ets_rate_limit);
+
+int mlx5_query_port_ets_rate_limit(struct mlx5_core_dev *mdev,
+				   u8 *max_bw_value,
+				   u8 *max_bw_units)
+{
+	u32 out[MLX5_ST_SZ_DW(qetc_reg)];
+	void *ets_tcn_conf;
+	int err;
+	int i;
+
+	err = mlx5_query_port_qetcr_reg(mdev, out, sizeof(out));
+	if (err)
+		return err;
+
+	for (i = 0; i <= mlx5_max_tc(mdev); i++) {
+		ets_tcn_conf = MLX5_ADDR_OF(qetc_reg, out, tc_configuration[i]);
+
+		max_bw_value[i] = MLX5_GET(ets_tcn_config_reg, ets_tcn_conf,
+					   max_bw_value);
+		max_bw_units[i] = MLX5_GET(ets_tcn_config_reg, ets_tcn_conf,
+					   max_bw_units);
+	}
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(mlx5_query_port_ets_rate_limit);
+
+int mlx5_set_port_wol(struct mlx5_core_dev *mdev, u8 wol_mode)
+{
+	u32 in[MLX5_ST_SZ_DW(set_wol_rol_in)] = {};
+
+	MLX5_SET(set_wol_rol_in, in, opcode, MLX5_CMD_OP_SET_WOL_ROL);
+	MLX5_SET(set_wol_rol_in, in, wol_mode_valid, 1);
+	MLX5_SET(set_wol_rol_in, in, wol_mode, wol_mode);
+	return mlx5_cmd_exec_in(mdev, set_wol_rol, in);
+}
+EXPORT_SYMBOL_GPL(mlx5_set_port_wol);
+
+int mlx5_query_port_wol(struct mlx5_core_dev *mdev, u8 *wol_mode)
+{
+	u32 out[MLX5_ST_SZ_DW(query_wol_rol_out)] = {};
+	u32 in[MLX5_ST_SZ_DW(query_wol_rol_in)] = {};
+	int err;
+
+	MLX5_SET(query_wol_rol_in, in, opcode, MLX5_CMD_OP_QUERY_WOL_ROL);
+	err = mlx5_cmd_exec_inout(mdev, query_wol_rol, in, out);
+	if (!err)
+		*wol_mode = MLX5_GET(query_wol_rol_out, out, wol_mode);
+
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx5_query_port_wol);
+
+int mlx5_query_ports_check(struct mlx5_core_dev *mdev, u32 *out, int outlen)
+{
+	u32 in[MLX5_ST_SZ_DW(pcmr_reg)] = {0};
+
+	MLX5_SET(pcmr_reg, in, local_port, 1);
+	return mlx5_core_access_reg(mdev, in, sizeof(in), out,
+				    outlen, MLX5_REG_PCMR, 0, 0);
+}
+
+int mlx5_set_ports_check(struct mlx5_core_dev *mdev, u32 *in, int inlen)
+{
+	u32 out[MLX5_ST_SZ_DW(pcmr_reg)];
+
+	return mlx5_core_access_reg(mdev, in, inlen, out,
+				    sizeof(out), MLX5_REG_PCMR, 0, 1);
+}
+
+int mlx5_set_port_fcs(struct mlx5_core_dev *mdev, u8 enable)
+{
+	u32 in[MLX5_ST_SZ_DW(pcmr_reg)] = {0};
+	int err;
+
+	err = mlx5_query_ports_check(mdev, in, sizeof(in));
+	if (err)
+		return err;
+	MLX5_SET(pcmr_reg, in, local_port, 1);
+	MLX5_SET(pcmr_reg, in, fcs_chk, enable);
+	return mlx5_set_ports_check(mdev, in, sizeof(in));
+}
+
+void mlx5_query_port_fcs(struct mlx5_core_dev *mdev, bool *supported,
+			 bool *enabled)
+{
+	u32 out[MLX5_ST_SZ_DW(pcmr_reg)];
+	/* Default values for FW which do not support MLX5_REG_PCMR */
+	*supported = false;
+	*enabled = true;
+
+	if (!MLX5_CAP_GEN(mdev, ports_check))
+		return;
+
+	if (mlx5_query_ports_check(mdev, out, sizeof(out)))
+		return;
+
+	*supported = !!(MLX5_GET(pcmr_reg, out, fcs_cap));
+	*enabled = !!(MLX5_GET(pcmr_reg, out, fcs_chk));
+}
+
+int mlx5_query_mtpps(struct mlx5_core_dev *mdev, u32 *mtpps, u32 mtpps_size)
+{
+	u32 in[MLX5_ST_SZ_DW(mtpps_reg)] = {0};
+
+	return mlx5_core_access_reg(mdev, in, sizeof(in), mtpps,
+				    mtpps_size, MLX5_REG_MTPPS, 0, 0);
+}
+
+int mlx5_set_mtpps(struct mlx5_core_dev *mdev, u32 *mtpps, u32 mtpps_size)
+{
+	u32 out[MLX5_ST_SZ_DW(mtpps_reg)] = {0};
+
+	return mlx5_core_access_reg(mdev, mtpps, mtpps_size, out,
+				    sizeof(out), MLX5_REG_MTPPS, 0, 1);
+}
+
+int mlx5_query_mtppse(struct mlx5_core_dev *mdev, u8 pin, u8 *arm, u8 *mode)
+{
+	u32 out[MLX5_ST_SZ_DW(mtppse_reg)] = {0};
+	u32 in[MLX5_ST_SZ_DW(mtppse_reg)] = {0};
+	int err = 0;
+
+	MLX5_SET(mtppse_reg, in, pin, pin);
+
+	err = mlx5_core_access_reg(mdev, in, sizeof(in), out,
+				   sizeof(out), MLX5_REG_MTPPSE, 0, 0);
+	if (err)
+		return err;
+
+	*arm = MLX5_GET(mtppse_reg, in, event_arm);
+	*mode = MLX5_GET(mtppse_reg, in, event_generation_mode);
+
+	return err;
+}
+
+int mlx5_set_mtppse(struct mlx5_core_dev *mdev, u8 pin, u8 arm, u8 mode)
+{
+	u32 out[MLX5_ST_SZ_DW(mtppse_reg)] = {0};
+	u32 in[MLX5_ST_SZ_DW(mtppse_reg)] = {0};
+
+	MLX5_SET(mtppse_reg, in, pin, pin);
+	MLX5_SET(mtppse_reg, in, event_arm, arm);
+	MLX5_SET(mtppse_reg, in, event_generation_mode, mode);
+
+	return mlx5_core_access_reg(mdev, in, sizeof(in), out,
+				    sizeof(out), MLX5_REG_MTPPSE, 0, 1);
+}
+
+int mlx5_set_trust_state(struct mlx5_core_dev *mdev, u8 trust_state)
+{
+	u32 out[MLX5_ST_SZ_DW(qpts_reg)] = {};
+	u32 in[MLX5_ST_SZ_DW(qpts_reg)] = {};
+	int err;
+
+	MLX5_SET(qpts_reg, in, local_port, 1);
+	MLX5_SET(qpts_reg, in, trust_state, trust_state);
+
+	err = mlx5_core_access_reg(mdev, in, sizeof(in), out,
+				   sizeof(out), MLX5_REG_QPTS, 0, 1);
+	return err;
+}
+
+int mlx5_query_trust_state(struct mlx5_core_dev *mdev, u8 *trust_state)
+{
+	u32 out[MLX5_ST_SZ_DW(qpts_reg)] = {};
+	u32 in[MLX5_ST_SZ_DW(qpts_reg)] = {};
+	int err;
+
+	MLX5_SET(qpts_reg, in, local_port, 1);
+
+	err = mlx5_core_access_reg(mdev, in, sizeof(in), out,
+				   sizeof(out), MLX5_REG_QPTS, 0, 0);
+	if (!err)
+		*trust_state = MLX5_GET(qpts_reg, out, trust_state);
+
+	return err;
+}
+
+int mlx5_set_dscp2prio(struct mlx5_core_dev *mdev, u8 dscp, u8 prio)
+{
+	int sz = MLX5_ST_SZ_BYTES(qpdpm_reg);
+	void *qpdpm_dscp;
+	void *out;
+	void *in;
+	int err;
+
+	in = kzalloc(sz, GFP_KERNEL);
+	out = kzalloc(sz, GFP_KERNEL);
+	if (!in || !out) {
+		err = -ENOMEM;
+		goto out;
+	}
+
+	MLX5_SET(qpdpm_reg, in, local_port, 1);
+	err = mlx5_core_access_reg(mdev, in, sz, out, sz, MLX5_REG_QPDPM, 0, 0);
+	if (err)
+		goto out;
+
+	memcpy(in, out, sz);
+	MLX5_SET(qpdpm_reg, in, local_port, 1);
+
+	/* Update the corresponding dscp entry */
+	qpdpm_dscp = MLX5_ADDR_OF(qpdpm_reg, in, dscp[dscp]);
+	MLX5_SET16(qpdpm_dscp_reg, qpdpm_dscp, prio, prio);
+	MLX5_SET16(qpdpm_dscp_reg, qpdpm_dscp, e, 1);
+	err = mlx5_core_access_reg(mdev, in, sz, out, sz, MLX5_REG_QPDPM, 0, 1);
+
+out:
+	kfree(in);
+	kfree(out);
+	return err;
+}
+
+/* dscp2prio[i]: priority that dscp i mapped to */
+#define MLX5E_SUPPORTED_DSCP 64
+int mlx5_query_dscp2prio(struct mlx5_core_dev *mdev, u8 *dscp2prio)
+{
+	int sz = MLX5_ST_SZ_BYTES(qpdpm_reg);
+	void *qpdpm_dscp;
+	void *out;
+	void *in;
+	int err;
+	int i;
+
+	in = kzalloc(sz, GFP_KERNEL);
+	out = kzalloc(sz, GFP_KERNEL);
+	if (!in || !out) {
+		err = -ENOMEM;
+		goto out;
+	}
+
+	MLX5_SET(qpdpm_reg, in, local_port, 1);
+	err = mlx5_core_access_reg(mdev, in, sz, out, sz, MLX5_REG_QPDPM, 0, 0);
+	if (err)
+		goto out;
+
+	for (i = 0; i < (MLX5E_SUPPORTED_DSCP); i++) {
+		qpdpm_dscp = MLX5_ADDR_OF(qpdpm_reg, out, dscp[i]);
+		dscp2prio[i] = MLX5_GET16(qpdpm_dscp_reg, qpdpm_dscp, prio);
+	}
+
+out:
+	kfree(in);
+	kfree(out);
+	return err;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/qos.c b/drivers/net/ethernet/mellanox/mlx5/core/qos.c
new file mode 100644
index 000000000..8bce730b5
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/qos.c
@@ -0,0 +1,84 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2020, Mellanox Technologies inc. All rights reserved. */
+
+#include "qos.h"
+
+#define MLX5_QOS_DEFAULT_DWRR_UID 0
+
+bool mlx5_qos_is_supported(struct mlx5_core_dev *mdev)
+{
+	if (!MLX5_CAP_GEN(mdev, qos))
+		return false;
+	if (!MLX5_CAP_QOS(mdev, nic_sq_scheduling))
+		return false;
+	if (!MLX5_CAP_QOS(mdev, nic_bw_share))
+		return false;
+	if (!MLX5_CAP_QOS(mdev, nic_rate_limit))
+		return false;
+	return true;
+}
+
+int mlx5_qos_max_leaf_nodes(struct mlx5_core_dev *mdev)
+{
+	return 1 << MLX5_CAP_QOS(mdev, log_max_qos_nic_queue_group);
+}
+
+int mlx5_qos_create_leaf_node(struct mlx5_core_dev *mdev, u32 parent_id,
+			      u32 bw_share, u32 max_avg_bw, u32 *id)
+{
+	u32 sched_ctx[MLX5_ST_SZ_DW(scheduling_context)] = {0};
+
+	MLX5_SET(scheduling_context, sched_ctx, parent_element_id, parent_id);
+	MLX5_SET(scheduling_context, sched_ctx, element_type,
+		 SCHEDULING_CONTEXT_ELEMENT_TYPE_QUEUE_GROUP);
+	MLX5_SET(scheduling_context, sched_ctx, bw_share, bw_share);
+	MLX5_SET(scheduling_context, sched_ctx, max_average_bw, max_avg_bw);
+
+	return mlx5_create_scheduling_element_cmd(mdev, SCHEDULING_HIERARCHY_NIC,
+						  sched_ctx, id);
+}
+
+int mlx5_qos_create_inner_node(struct mlx5_core_dev *mdev, u32 parent_id,
+			       u32 bw_share, u32 max_avg_bw, u32 *id)
+{
+	u32 sched_ctx[MLX5_ST_SZ_DW(scheduling_context)] = {0};
+	void *attr;
+
+	MLX5_SET(scheduling_context, sched_ctx, parent_element_id, parent_id);
+	MLX5_SET(scheduling_context, sched_ctx, element_type,
+		 SCHEDULING_CONTEXT_ELEMENT_TYPE_TSAR);
+	MLX5_SET(scheduling_context, sched_ctx, bw_share, bw_share);
+	MLX5_SET(scheduling_context, sched_ctx, max_average_bw, max_avg_bw);
+
+	attr = MLX5_ADDR_OF(scheduling_context, sched_ctx, element_attributes);
+	MLX5_SET(tsar_element, attr, tsar_type, TSAR_ELEMENT_TSAR_TYPE_DWRR);
+
+	return mlx5_create_scheduling_element_cmd(mdev, SCHEDULING_HIERARCHY_NIC,
+						  sched_ctx, id);
+}
+
+int mlx5_qos_create_root_node(struct mlx5_core_dev *mdev, u32 *id)
+{
+	return mlx5_qos_create_inner_node(mdev, MLX5_QOS_DEFAULT_DWRR_UID, 0, 0, id);
+}
+
+int mlx5_qos_update_node(struct mlx5_core_dev *mdev,
+			 u32 bw_share, u32 max_avg_bw, u32 id)
+{
+	u32 sched_ctx[MLX5_ST_SZ_DW(scheduling_context)] = {0};
+	u32 bitmask = 0;
+
+	MLX5_SET(scheduling_context, sched_ctx, bw_share, bw_share);
+	MLX5_SET(scheduling_context, sched_ctx, max_average_bw, max_avg_bw);
+
+	bitmask |= MODIFY_SCHEDULING_ELEMENT_IN_MODIFY_BITMASK_BW_SHARE;
+	bitmask |= MODIFY_SCHEDULING_ELEMENT_IN_MODIFY_BITMASK_MAX_AVERAGE_BW;
+
+	return mlx5_modify_scheduling_element_cmd(mdev, SCHEDULING_HIERARCHY_NIC,
+						  sched_ctx, id, bitmask);
+}
+
+int mlx5_qos_destroy_node(struct mlx5_core_dev *mdev, u32 id)
+{
+	return mlx5_destroy_scheduling_element_cmd(mdev, SCHEDULING_HIERARCHY_NIC, id);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/qos.h b/drivers/net/ethernet/mellanox/mlx5/core/qos.h
new file mode 100644
index 000000000..624ce822b
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/qos.h
@@ -0,0 +1,30 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2020, Mellanox Technologies inc. All rights reserved. */
+
+#ifndef __MLX5_QOS_H
+#define __MLX5_QOS_H
+
+#include "mlx5_core.h"
+
+#define MLX5_DEBUG_QOS_MASK BIT(4)
+
+#define qos_err(mdev, fmt, ...) \
+	mlx5_core_err(mdev, "QoS: " fmt, ##__VA_ARGS__)
+#define qos_warn(mdev, fmt, ...) \
+	mlx5_core_warn(mdev, "QoS: " fmt, ##__VA_ARGS__)
+#define qos_dbg(mdev, fmt, ...) \
+	mlx5_core_dbg_mask(mdev, MLX5_DEBUG_QOS_MASK, "QoS: " fmt, ##__VA_ARGS__)
+
+bool mlx5_qos_is_supported(struct mlx5_core_dev *mdev);
+int mlx5_qos_max_leaf_nodes(struct mlx5_core_dev *mdev);
+
+int mlx5_qos_create_leaf_node(struct mlx5_core_dev *mdev, u32 parent_id,
+			      u32 bw_share, u32 max_avg_bw, u32 *id);
+int mlx5_qos_create_inner_node(struct mlx5_core_dev *mdev, u32 parent_id,
+			       u32 bw_share, u32 max_avg_bw, u32 *id);
+int mlx5_qos_create_root_node(struct mlx5_core_dev *mdev, u32 *id);
+int mlx5_qos_update_node(struct mlx5_core_dev *mdev, u32 bw_share,
+			 u32 max_avg_bw, u32 id);
+int mlx5_qos_destroy_node(struct mlx5_core_dev *mdev, u32 id);
+
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/rdma.c b/drivers/net/ethernet/mellanox/mlx5/core/rdma.c
new file mode 100644
index 000000000..540cf05f6
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/rdma.c
@@ -0,0 +1,186 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2019 Mellanox Technologies */
+
+#include <linux/mlx5/vport.h>
+#include <rdma/ib_verbs.h>
+#include <net/addrconf.h>
+
+#include "lib/mlx5.h"
+#include "eswitch.h"
+#include "fs_core.h"
+#include "rdma.h"
+
+static void mlx5_rdma_disable_roce_steering(struct mlx5_core_dev *dev)
+{
+	struct mlx5_core_roce *roce = &dev->priv.roce;
+
+	mlx5_del_flow_rules(roce->allow_rule);
+	mlx5_destroy_flow_group(roce->fg);
+	mlx5_destroy_flow_table(roce->ft);
+}
+
+static int mlx5_rdma_enable_roce_steering(struct mlx5_core_dev *dev)
+{
+	int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
+	struct mlx5_core_roce *roce = &dev->priv.roce;
+	struct mlx5_flow_handle *flow_rule = NULL;
+	struct mlx5_flow_table_attr ft_attr = {};
+	struct mlx5_flow_namespace *ns = NULL;
+	struct mlx5_flow_act flow_act = {};
+	struct mlx5_flow_spec *spec;
+	struct mlx5_flow_table *ft;
+	struct mlx5_flow_group *fg;
+	void *match_criteria;
+	u32 *flow_group_in;
+	void *misc;
+	int err;
+
+	if (!(MLX5_CAP_FLOWTABLE_RDMA_RX(dev, ft_support) &&
+	      MLX5_CAP_FLOWTABLE_RDMA_RX(dev, table_miss_action_domain)))
+		return -EOPNOTSUPP;
+
+	flow_group_in = kvzalloc(inlen, GFP_KERNEL);
+	if (!flow_group_in)
+		return -ENOMEM;
+	spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
+	if (!spec) {
+		kvfree(flow_group_in);
+		return -ENOMEM;
+	}
+
+	ns = mlx5_get_flow_namespace(dev, MLX5_FLOW_NAMESPACE_RDMA_RX_KERNEL);
+	if (!ns) {
+		mlx5_core_err(dev, "Failed to get RDMA RX namespace");
+		err = -EOPNOTSUPP;
+		goto free;
+	}
+
+	ft_attr.max_fte = 1;
+	ft = mlx5_create_flow_table(ns, &ft_attr);
+	if (IS_ERR(ft)) {
+		mlx5_core_err(dev, "Failed to create RDMA RX flow table");
+		err = PTR_ERR(ft);
+		goto free;
+	}
+
+	MLX5_SET(create_flow_group_in, flow_group_in, match_criteria_enable,
+		 MLX5_MATCH_MISC_PARAMETERS);
+	match_criteria = MLX5_ADDR_OF(create_flow_group_in, flow_group_in,
+				      match_criteria);
+	MLX5_SET_TO_ONES(fte_match_param, match_criteria,
+			 misc_parameters.source_port);
+
+	fg = mlx5_create_flow_group(ft, flow_group_in);
+	if (IS_ERR(fg)) {
+		err = PTR_ERR(fg);
+		mlx5_core_err(dev, "Failed to create RDMA RX flow group err(%d)\n", err);
+		goto destroy_flow_table;
+	}
+
+	spec->match_criteria_enable = MLX5_MATCH_MISC_PARAMETERS;
+	misc = MLX5_ADDR_OF(fte_match_param, spec->match_value,
+			    misc_parameters);
+	MLX5_SET(fte_match_set_misc, misc, source_port,
+		 dev->priv.eswitch->manager_vport);
+	misc = MLX5_ADDR_OF(fte_match_param, spec->match_criteria,
+			    misc_parameters);
+	MLX5_SET_TO_ONES(fte_match_set_misc, misc, source_port);
+
+	flow_act.action = MLX5_FLOW_CONTEXT_ACTION_ALLOW;
+	flow_rule = mlx5_add_flow_rules(ft, spec, &flow_act, NULL, 0);
+	if (IS_ERR(flow_rule)) {
+		err = PTR_ERR(flow_rule);
+		mlx5_core_err(dev, "Failed to add RoCE allow rule, err=%d\n",
+			      err);
+		goto destroy_flow_group;
+	}
+
+	kvfree(spec);
+	kvfree(flow_group_in);
+	roce->ft = ft;
+	roce->fg = fg;
+	roce->allow_rule = flow_rule;
+
+	return 0;
+
+destroy_flow_group:
+	mlx5_destroy_flow_group(fg);
+destroy_flow_table:
+	mlx5_destroy_flow_table(ft);
+free:
+	kvfree(spec);
+	kvfree(flow_group_in);
+	return err;
+}
+
+static void mlx5_rdma_del_roce_addr(struct mlx5_core_dev *dev)
+{
+	mlx5_core_roce_gid_set(dev, 0, 0, 0,
+			       NULL, NULL, false, 0, 1);
+}
+
+static void mlx5_rdma_make_default_gid(struct mlx5_core_dev *dev, union ib_gid *gid)
+{
+	u8 hw_id[ETH_ALEN];
+
+	mlx5_query_mac_address(dev, hw_id);
+	gid->global.subnet_prefix = cpu_to_be64(0xfe80000000000000LL);
+	addrconf_addr_eui48(&gid->raw[8], hw_id);
+}
+
+static int mlx5_rdma_add_roce_addr(struct mlx5_core_dev *dev)
+{
+	union ib_gid gid;
+	u8 mac[ETH_ALEN];
+
+	mlx5_rdma_make_default_gid(dev, &gid);
+	return mlx5_core_roce_gid_set(dev, 0,
+				      MLX5_ROCE_VERSION_1,
+				      0, gid.raw, mac,
+				      false, 0, 1);
+}
+
+void mlx5_rdma_disable_roce(struct mlx5_core_dev *dev)
+{
+	struct mlx5_core_roce *roce = &dev->priv.roce;
+
+	if (!roce->ft)
+		return;
+
+	mlx5_rdma_disable_roce_steering(dev);
+	mlx5_rdma_del_roce_addr(dev);
+	mlx5_nic_vport_disable_roce(dev);
+}
+
+void mlx5_rdma_enable_roce(struct mlx5_core_dev *dev)
+{
+	int err;
+
+	if (!MLX5_CAP_GEN(dev, roce))
+		return;
+
+	err = mlx5_nic_vport_enable_roce(dev);
+	if (err) {
+		mlx5_core_err(dev, "Failed to enable RoCE: %d\n", err);
+		return;
+	}
+
+	err = mlx5_rdma_add_roce_addr(dev);
+	if (err) {
+		mlx5_core_err(dev, "Failed to add RoCE address: %d\n", err);
+		goto disable_roce;
+	}
+
+	err = mlx5_rdma_enable_roce_steering(dev);
+	if (err) {
+		mlx5_core_err(dev, "Failed to enable RoCE steering: %d\n", err);
+		goto del_roce_addr;
+	}
+
+	return;
+
+del_roce_addr:
+	mlx5_rdma_del_roce_addr(dev);
+disable_roce:
+	mlx5_nic_vport_disable_roce(dev);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/rdma.h b/drivers/net/ethernet/mellanox/mlx5/core/rdma.h
new file mode 100644
index 000000000..750cff2a7
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/rdma.h
@@ -0,0 +1,20 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2019 Mellanox Technologies. */
+
+#ifndef __MLX5_RDMA_H__
+#define __MLX5_RDMA_H__
+
+#include "mlx5_core.h"
+
+#ifdef CONFIG_MLX5_ESWITCH
+
+void mlx5_rdma_enable_roce(struct mlx5_core_dev *dev);
+void mlx5_rdma_disable_roce(struct mlx5_core_dev *dev);
+
+#else /* CONFIG_MLX5_ESWITCH */
+
+static inline void mlx5_rdma_enable_roce(struct mlx5_core_dev *dev) {}
+static inline void mlx5_rdma_disable_roce(struct mlx5_core_dev *dev) {}
+
+#endif /* CONFIG_MLX5_ESWITCH */
+#endif /* __MLX5_RDMA_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/rl.c b/drivers/net/ethernet/mellanox/mlx5/core/rl.c
new file mode 100644
index 000000000..9f8b4005f
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/rl.c
@@ -0,0 +1,398 @@
+/*
+ * Copyright (c) 2013-2016, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/kernel.h>
+#include <linux/mlx5/driver.h>
+#include "mlx5_core.h"
+
+/* Scheduling element fw management */
+int mlx5_create_scheduling_element_cmd(struct mlx5_core_dev *dev, u8 hierarchy,
+				       void *ctx, u32 *element_id)
+{
+	u32 out[MLX5_ST_SZ_DW(create_scheduling_element_in)] = {};
+	u32 in[MLX5_ST_SZ_DW(create_scheduling_element_in)] = {};
+	void *schedc;
+	int err;
+
+	schedc = MLX5_ADDR_OF(create_scheduling_element_in, in,
+			      scheduling_context);
+	MLX5_SET(create_scheduling_element_in, in, opcode,
+		 MLX5_CMD_OP_CREATE_SCHEDULING_ELEMENT);
+	MLX5_SET(create_scheduling_element_in, in, scheduling_hierarchy,
+		 hierarchy);
+	memcpy(schedc, ctx, MLX5_ST_SZ_BYTES(scheduling_context));
+
+	err = mlx5_cmd_exec_inout(dev, create_scheduling_element, in, out);
+	if (err)
+		return err;
+
+	*element_id = MLX5_GET(create_scheduling_element_out, out,
+			       scheduling_element_id);
+	return 0;
+}
+
+int mlx5_modify_scheduling_element_cmd(struct mlx5_core_dev *dev, u8 hierarchy,
+				       void *ctx, u32 element_id,
+				       u32 modify_bitmask)
+{
+	u32 in[MLX5_ST_SZ_DW(modify_scheduling_element_in)] = {};
+	void *schedc;
+
+	schedc = MLX5_ADDR_OF(modify_scheduling_element_in, in,
+			      scheduling_context);
+	MLX5_SET(modify_scheduling_element_in, in, opcode,
+		 MLX5_CMD_OP_MODIFY_SCHEDULING_ELEMENT);
+	MLX5_SET(modify_scheduling_element_in, in, scheduling_element_id,
+		 element_id);
+	MLX5_SET(modify_scheduling_element_in, in, modify_bitmask,
+		 modify_bitmask);
+	MLX5_SET(modify_scheduling_element_in, in, scheduling_hierarchy,
+		 hierarchy);
+	memcpy(schedc, ctx, MLX5_ST_SZ_BYTES(scheduling_context));
+
+	return mlx5_cmd_exec_in(dev, modify_scheduling_element, in);
+}
+
+int mlx5_destroy_scheduling_element_cmd(struct mlx5_core_dev *dev, u8 hierarchy,
+					u32 element_id)
+{
+	u32 in[MLX5_ST_SZ_DW(destroy_scheduling_element_in)] = {};
+
+	MLX5_SET(destroy_scheduling_element_in, in, opcode,
+		 MLX5_CMD_OP_DESTROY_SCHEDULING_ELEMENT);
+	MLX5_SET(destroy_scheduling_element_in, in, scheduling_element_id,
+		 element_id);
+	MLX5_SET(destroy_scheduling_element_in, in, scheduling_hierarchy,
+		 hierarchy);
+
+	return mlx5_cmd_exec_in(dev, destroy_scheduling_element, in);
+}
+
+static bool mlx5_rl_are_equal_raw(struct mlx5_rl_entry *entry, void *rl_in,
+				  u16 uid)
+{
+	return (!memcmp(entry->rl_raw, rl_in, sizeof(entry->rl_raw)) &&
+		entry->uid == uid);
+}
+
+/* Finds an entry where we can register the given rate
+ * If the rate already exists, return the entry where it is registered,
+ * otherwise return the first available entry.
+ * If the table is full, return NULL
+ */
+static struct mlx5_rl_entry *find_rl_entry(struct mlx5_rl_table *table,
+					   void *rl_in, u16 uid, bool dedicated)
+{
+	struct mlx5_rl_entry *ret_entry = NULL;
+	bool empty_found = false;
+	int i;
+
+	lockdep_assert_held(&table->rl_lock);
+	WARN_ON(!table->rl_entry);
+
+	for (i = 0; i < table->max_size; i++) {
+		if (dedicated) {
+			if (!table->rl_entry[i].refcount)
+				return &table->rl_entry[i];
+			continue;
+		}
+
+		if (table->rl_entry[i].refcount) {
+			if (table->rl_entry[i].dedicated)
+				continue;
+			if (mlx5_rl_are_equal_raw(&table->rl_entry[i], rl_in,
+						  uid))
+				return &table->rl_entry[i];
+		} else if (!empty_found) {
+			empty_found = true;
+			ret_entry = &table->rl_entry[i];
+		}
+	}
+
+	return ret_entry;
+}
+
+static int mlx5_set_pp_rate_limit_cmd(struct mlx5_core_dev *dev,
+				      struct mlx5_rl_entry *entry, bool set)
+{
+	u32 in[MLX5_ST_SZ_DW(set_pp_rate_limit_in)] = {};
+	void *pp_context;
+
+	pp_context = MLX5_ADDR_OF(set_pp_rate_limit_in, in, ctx);
+	MLX5_SET(set_pp_rate_limit_in, in, opcode,
+		 MLX5_CMD_OP_SET_PP_RATE_LIMIT);
+	MLX5_SET(set_pp_rate_limit_in, in, uid, entry->uid);
+	MLX5_SET(set_pp_rate_limit_in, in, rate_limit_index, entry->index);
+	if (set)
+		memcpy(pp_context, entry->rl_raw, sizeof(entry->rl_raw));
+	return mlx5_cmd_exec_in(dev, set_pp_rate_limit, in);
+}
+
+bool mlx5_rl_is_in_range(struct mlx5_core_dev *dev, u32 rate)
+{
+	struct mlx5_rl_table *table = &dev->priv.rl_table;
+
+	return (rate <= table->max_rate && rate >= table->min_rate);
+}
+EXPORT_SYMBOL(mlx5_rl_is_in_range);
+
+bool mlx5_rl_are_equal(struct mlx5_rate_limit *rl_0,
+		       struct mlx5_rate_limit *rl_1)
+{
+	return ((rl_0->rate == rl_1->rate) &&
+		(rl_0->max_burst_sz == rl_1->max_burst_sz) &&
+		(rl_0->typical_pkt_sz == rl_1->typical_pkt_sz));
+}
+EXPORT_SYMBOL(mlx5_rl_are_equal);
+
+static int mlx5_rl_table_get(struct mlx5_rl_table *table)
+{
+	int i;
+
+	lockdep_assert_held(&table->rl_lock);
+
+	if (table->rl_entry) {
+		table->refcount++;
+		return 0;
+	}
+
+	table->rl_entry = kcalloc(table->max_size, sizeof(struct mlx5_rl_entry),
+				  GFP_KERNEL);
+	if (!table->rl_entry)
+		return -ENOMEM;
+
+	/* The index represents the index in HW rate limit table
+	 * Index 0 is reserved for unlimited rate
+	 */
+	for (i = 0; i < table->max_size; i++)
+		table->rl_entry[i].index = i + 1;
+
+	table->refcount++;
+	return 0;
+}
+
+static void mlx5_rl_table_put(struct mlx5_rl_table *table)
+{
+	lockdep_assert_held(&table->rl_lock);
+	if (--table->refcount)
+		return;
+
+	kfree(table->rl_entry);
+	table->rl_entry = NULL;
+}
+
+static void mlx5_rl_table_free(struct mlx5_core_dev *dev, struct mlx5_rl_table *table)
+{
+	int i;
+
+	if (!table->rl_entry)
+		return;
+
+	/* Clear all configured rates */
+	for (i = 0; i < table->max_size; i++)
+		if (table->rl_entry[i].refcount)
+			mlx5_set_pp_rate_limit_cmd(dev, &table->rl_entry[i], false);
+	kfree(table->rl_entry);
+}
+
+static void mlx5_rl_entry_get(struct mlx5_rl_entry *entry)
+{
+	entry->refcount++;
+}
+
+static void
+mlx5_rl_entry_put(struct mlx5_core_dev *dev, struct mlx5_rl_entry *entry)
+{
+	entry->refcount--;
+	if (!entry->refcount)
+		mlx5_set_pp_rate_limit_cmd(dev, entry, false);
+}
+
+int mlx5_rl_add_rate_raw(struct mlx5_core_dev *dev, void *rl_in, u16 uid,
+			 bool dedicated_entry, u16 *index)
+{
+	struct mlx5_rl_table *table = &dev->priv.rl_table;
+	struct mlx5_rl_entry *entry;
+	u32 rate;
+	int err;
+
+	if (!table->max_size)
+		return -EOPNOTSUPP;
+
+	rate = MLX5_GET(set_pp_rate_limit_context, rl_in, rate_limit);
+	if (!rate || !mlx5_rl_is_in_range(dev, rate)) {
+		mlx5_core_err(dev, "Invalid rate: %u, should be %u to %u\n",
+			      rate, table->min_rate, table->max_rate);
+		return -EINVAL;
+	}
+
+	mutex_lock(&table->rl_lock);
+	err = mlx5_rl_table_get(table);
+	if (err)
+		goto out;
+
+	entry = find_rl_entry(table, rl_in, uid, dedicated_entry);
+	if (!entry) {
+		mlx5_core_err(dev, "Max number of %u rates reached\n",
+			      table->max_size);
+		err = -ENOSPC;
+		goto rl_err;
+	}
+	if (!entry->refcount) {
+		/* new rate limit */
+		memcpy(entry->rl_raw, rl_in, sizeof(entry->rl_raw));
+		entry->uid = uid;
+		err = mlx5_set_pp_rate_limit_cmd(dev, entry, true);
+		if (err) {
+			mlx5_core_err(
+				dev,
+				"Failed configuring rate limit(err %d): rate %u, max_burst_sz %u, typical_pkt_sz %u\n",
+				err, rate,
+				MLX5_GET(set_pp_rate_limit_context, rl_in,
+					 burst_upper_bound),
+				MLX5_GET(set_pp_rate_limit_context, rl_in,
+					 typical_packet_size));
+			goto rl_err;
+		}
+
+		entry->dedicated = dedicated_entry;
+	}
+	mlx5_rl_entry_get(entry);
+	*index = entry->index;
+	mutex_unlock(&table->rl_lock);
+	return 0;
+
+rl_err:
+	mlx5_rl_table_put(table);
+out:
+	mutex_unlock(&table->rl_lock);
+	return err;
+}
+EXPORT_SYMBOL(mlx5_rl_add_rate_raw);
+
+void mlx5_rl_remove_rate_raw(struct mlx5_core_dev *dev, u16 index)
+{
+	struct mlx5_rl_table *table = &dev->priv.rl_table;
+	struct mlx5_rl_entry *entry;
+
+	mutex_lock(&table->rl_lock);
+	entry = &table->rl_entry[index - 1];
+	mlx5_rl_entry_put(dev, entry);
+	mlx5_rl_table_put(table);
+	mutex_unlock(&table->rl_lock);
+}
+EXPORT_SYMBOL(mlx5_rl_remove_rate_raw);
+
+int mlx5_rl_add_rate(struct mlx5_core_dev *dev, u16 *index,
+		     struct mlx5_rate_limit *rl)
+{
+	u8 rl_raw[MLX5_ST_SZ_BYTES(set_pp_rate_limit_context)] = {};
+
+	MLX5_SET(set_pp_rate_limit_context, rl_raw, rate_limit, rl->rate);
+	MLX5_SET(set_pp_rate_limit_context, rl_raw, burst_upper_bound,
+		 rl->max_burst_sz);
+	MLX5_SET(set_pp_rate_limit_context, rl_raw, typical_packet_size,
+		 rl->typical_pkt_sz);
+
+	return mlx5_rl_add_rate_raw(dev, rl_raw,
+				    MLX5_CAP_QOS(dev, packet_pacing_uid) ?
+					MLX5_SHARED_RESOURCE_UID : 0,
+				    false, index);
+}
+EXPORT_SYMBOL(mlx5_rl_add_rate);
+
+void mlx5_rl_remove_rate(struct mlx5_core_dev *dev, struct mlx5_rate_limit *rl)
+{
+	u8 rl_raw[MLX5_ST_SZ_BYTES(set_pp_rate_limit_context)] = {};
+	struct mlx5_rl_table *table = &dev->priv.rl_table;
+	struct mlx5_rl_entry *entry = NULL;
+
+	/* 0 is a reserved value for unlimited rate */
+	if (rl->rate == 0)
+		return;
+
+	MLX5_SET(set_pp_rate_limit_context, rl_raw, rate_limit, rl->rate);
+	MLX5_SET(set_pp_rate_limit_context, rl_raw, burst_upper_bound,
+		 rl->max_burst_sz);
+	MLX5_SET(set_pp_rate_limit_context, rl_raw, typical_packet_size,
+		 rl->typical_pkt_sz);
+
+	mutex_lock(&table->rl_lock);
+	entry = find_rl_entry(table, rl_raw,
+			      MLX5_CAP_QOS(dev, packet_pacing_uid) ?
+				MLX5_SHARED_RESOURCE_UID : 0, false);
+	if (!entry || !entry->refcount) {
+		mlx5_core_warn(dev, "Rate %u, max_burst_sz %u typical_pkt_sz %u are not configured\n",
+			       rl->rate, rl->max_burst_sz, rl->typical_pkt_sz);
+		goto out;
+	}
+	mlx5_rl_entry_put(dev, entry);
+	mlx5_rl_table_put(table);
+out:
+	mutex_unlock(&table->rl_lock);
+}
+EXPORT_SYMBOL(mlx5_rl_remove_rate);
+
+int mlx5_init_rl_table(struct mlx5_core_dev *dev)
+{
+	struct mlx5_rl_table *table = &dev->priv.rl_table;
+
+	if (!MLX5_CAP_GEN(dev, qos) || !MLX5_CAP_QOS(dev, packet_pacing)) {
+		table->max_size = 0;
+		return 0;
+	}
+
+	mutex_init(&table->rl_lock);
+
+	/* First entry is reserved for unlimited rate */
+	table->max_size = MLX5_CAP_QOS(dev, packet_pacing_rate_table_size) - 1;
+	table->max_rate = MLX5_CAP_QOS(dev, packet_pacing_max_rate);
+	table->min_rate = MLX5_CAP_QOS(dev, packet_pacing_min_rate);
+
+	mlx5_core_info(dev, "Rate limit: %u rates are supported, range: %uMbps to %uMbps\n",
+		       table->max_size,
+		       table->min_rate >> 10,
+		       table->max_rate >> 10);
+
+	return 0;
+}
+
+void mlx5_cleanup_rl_table(struct mlx5_core_dev *dev)
+{
+	struct mlx5_rl_table *table = &dev->priv.rl_table;
+
+	if (!MLX5_CAP_GEN(dev, qos) || !MLX5_CAP_QOS(dev, packet_pacing))
+		return;
+
+	mlx5_rl_table_free(dev, table);
+	mutex_destroy(&table->rl_lock);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/sf/cmd.c b/drivers/net/ethernet/mellanox/mlx5/core/sf/cmd.c
new file mode 100644
index 000000000..a8d75c2f0
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/sf/cmd.c
@@ -0,0 +1,49 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2020 Mellanox Technologies Ltd */
+
+#include <linux/mlx5/driver.h>
+#include "priv.h"
+
+int mlx5_cmd_alloc_sf(struct mlx5_core_dev *dev, u16 function_id)
+{
+	u32 out[MLX5_ST_SZ_DW(alloc_sf_out)] = {};
+	u32 in[MLX5_ST_SZ_DW(alloc_sf_in)] = {};
+
+	MLX5_SET(alloc_sf_in, in, opcode, MLX5_CMD_OP_ALLOC_SF);
+	MLX5_SET(alloc_sf_in, in, function_id, function_id);
+
+	return mlx5_cmd_exec(dev, in, sizeof(in), out, sizeof(out));
+}
+
+int mlx5_cmd_dealloc_sf(struct mlx5_core_dev *dev, u16 function_id)
+{
+	u32 out[MLX5_ST_SZ_DW(dealloc_sf_out)] = {};
+	u32 in[MLX5_ST_SZ_DW(dealloc_sf_in)] = {};
+
+	MLX5_SET(dealloc_sf_in, in, opcode, MLX5_CMD_OP_DEALLOC_SF);
+	MLX5_SET(dealloc_sf_in, in, function_id, function_id);
+
+	return mlx5_cmd_exec(dev, in, sizeof(in), out, sizeof(out));
+}
+
+int mlx5_cmd_sf_enable_hca(struct mlx5_core_dev *dev, u16 func_id)
+{
+	u32 out[MLX5_ST_SZ_DW(enable_hca_out)] = {};
+	u32 in[MLX5_ST_SZ_DW(enable_hca_in)] = {};
+
+	MLX5_SET(enable_hca_in, in, opcode, MLX5_CMD_OP_ENABLE_HCA);
+	MLX5_SET(enable_hca_in, in, function_id, func_id);
+	MLX5_SET(enable_hca_in, in, embedded_cpu_function, 0);
+	return mlx5_cmd_exec(dev, &in, sizeof(in), &out, sizeof(out));
+}
+
+int mlx5_cmd_sf_disable_hca(struct mlx5_core_dev *dev, u16 func_id)
+{
+	u32 out[MLX5_ST_SZ_DW(disable_hca_out)] = {};
+	u32 in[MLX5_ST_SZ_DW(disable_hca_in)] = {};
+
+	MLX5_SET(disable_hca_in, in, opcode, MLX5_CMD_OP_DISABLE_HCA);
+	MLX5_SET(disable_hca_in, in, function_id, func_id);
+	MLX5_SET(enable_hca_in, in, embedded_cpu_function, 0);
+	return mlx5_cmd_exec(dev, in, sizeof(in), out, sizeof(out));
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/sf/dev/dev.c b/drivers/net/ethernet/mellanox/mlx5/core/sf/dev/dev.c
new file mode 100644
index 000000000..8e2abbab0
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/sf/dev/dev.c
@@ -0,0 +1,380 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2020 Mellanox Technologies Ltd */
+
+#include <linux/mlx5/driver.h>
+#include <linux/mlx5/device.h>
+#include "mlx5_core.h"
+#include "dev.h"
+#include "sf/vhca_event.h"
+#include "sf/sf.h"
+#include "sf/mlx5_ifc_vhca_event.h"
+#include "ecpf.h"
+#define CREATE_TRACE_POINTS
+#include "diag/dev_tracepoint.h"
+
+struct mlx5_sf_dev_table {
+	struct xarray devices;
+	unsigned int max_sfs;
+	phys_addr_t base_address;
+	u64 sf_bar_length;
+	struct notifier_block nb;
+	struct mutex table_lock; /* Serializes sf life cycle and vhca state change handler */
+	struct workqueue_struct *active_wq;
+	struct work_struct work;
+	u8 stop_active_wq:1;
+	struct mlx5_core_dev *dev;
+};
+
+static bool mlx5_sf_dev_supported(const struct mlx5_core_dev *dev)
+{
+	return MLX5_CAP_GEN(dev, sf) && mlx5_vhca_event_supported(dev);
+}
+
+bool mlx5_sf_dev_allocated(const struct mlx5_core_dev *dev)
+{
+	struct mlx5_sf_dev_table *table = dev->priv.sf_dev_table;
+
+	return table && !xa_empty(&table->devices);
+}
+
+static ssize_t sfnum_show(struct device *dev, struct device_attribute *attr, char *buf)
+{
+	struct auxiliary_device *adev = container_of(dev, struct auxiliary_device, dev);
+	struct mlx5_sf_dev *sf_dev = container_of(adev, struct mlx5_sf_dev, adev);
+
+	return sysfs_emit(buf, "%u\n", sf_dev->sfnum);
+}
+static DEVICE_ATTR_RO(sfnum);
+
+static struct attribute *sf_device_attrs[] = {
+	&dev_attr_sfnum.attr,
+	NULL,
+};
+
+static const struct attribute_group sf_attr_group = {
+	.attrs = sf_device_attrs,
+};
+
+static const struct attribute_group *sf_attr_groups[2] = {
+	&sf_attr_group,
+	NULL
+};
+
+static void mlx5_sf_dev_release(struct device *device)
+{
+	struct auxiliary_device *adev = container_of(device, struct auxiliary_device, dev);
+	struct mlx5_sf_dev *sf_dev = container_of(adev, struct mlx5_sf_dev, adev);
+
+	mlx5_adev_idx_free(adev->id);
+	kfree(sf_dev);
+}
+
+static void mlx5_sf_dev_remove(struct mlx5_core_dev *dev, struct mlx5_sf_dev *sf_dev)
+{
+	int id;
+
+	id = sf_dev->adev.id;
+	trace_mlx5_sf_dev_del(dev, sf_dev, id);
+
+	auxiliary_device_delete(&sf_dev->adev);
+	auxiliary_device_uninit(&sf_dev->adev);
+}
+
+static void mlx5_sf_dev_add(struct mlx5_core_dev *dev, u16 sf_index, u16 fn_id, u32 sfnum)
+{
+	struct mlx5_sf_dev_table *table = dev->priv.sf_dev_table;
+	struct mlx5_sf_dev *sf_dev;
+	struct pci_dev *pdev;
+	int err;
+	int id;
+
+	id = mlx5_adev_idx_alloc();
+	if (id < 0) {
+		err = id;
+		goto add_err;
+	}
+
+	sf_dev = kzalloc(sizeof(*sf_dev), GFP_KERNEL);
+	if (!sf_dev) {
+		mlx5_adev_idx_free(id);
+		err = -ENOMEM;
+		goto add_err;
+	}
+	pdev = dev->pdev;
+	sf_dev->adev.id = id;
+	sf_dev->adev.name = MLX5_SF_DEV_ID_NAME;
+	sf_dev->adev.dev.release = mlx5_sf_dev_release;
+	sf_dev->adev.dev.parent = &pdev->dev;
+	sf_dev->adev.dev.groups = sf_attr_groups;
+	sf_dev->sfnum = sfnum;
+	sf_dev->parent_mdev = dev;
+	sf_dev->fn_id = fn_id;
+
+	if (!table->max_sfs) {
+		mlx5_adev_idx_free(id);
+		kfree(sf_dev);
+		err = -EOPNOTSUPP;
+		goto add_err;
+	}
+	sf_dev->bar_base_addr = table->base_address + (sf_index * table->sf_bar_length);
+
+	trace_mlx5_sf_dev_add(dev, sf_dev, id);
+
+	err = auxiliary_device_init(&sf_dev->adev);
+	if (err) {
+		mlx5_adev_idx_free(id);
+		kfree(sf_dev);
+		goto add_err;
+	}
+
+	err = auxiliary_device_add(&sf_dev->adev);
+	if (err) {
+		put_device(&sf_dev->adev.dev);
+		goto add_err;
+	}
+
+	err = xa_insert(&table->devices, sf_index, sf_dev, GFP_KERNEL);
+	if (err)
+		goto xa_err;
+	return;
+
+xa_err:
+	mlx5_sf_dev_remove(dev, sf_dev);
+add_err:
+	mlx5_core_err(dev, "SF DEV: fail device add for index=%d sfnum=%d err=%d\n",
+		      sf_index, sfnum, err);
+}
+
+static void mlx5_sf_dev_del(struct mlx5_core_dev *dev, struct mlx5_sf_dev *sf_dev, u16 sf_index)
+{
+	struct mlx5_sf_dev_table *table = dev->priv.sf_dev_table;
+
+	xa_erase(&table->devices, sf_index);
+	mlx5_sf_dev_remove(dev, sf_dev);
+}
+
+static int
+mlx5_sf_dev_state_change_handler(struct notifier_block *nb, unsigned long event_code, void *data)
+{
+	struct mlx5_sf_dev_table *table = container_of(nb, struct mlx5_sf_dev_table, nb);
+	const struct mlx5_vhca_state_event *event = data;
+	struct mlx5_sf_dev *sf_dev;
+	u16 max_functions;
+	u16 sf_index;
+	u16 base_id;
+
+	max_functions = mlx5_sf_max_functions(table->dev);
+	if (!max_functions)
+		return 0;
+
+	base_id = MLX5_CAP_GEN(table->dev, sf_base_id);
+	if (event->function_id < base_id || event->function_id >= (base_id + max_functions))
+		return 0;
+
+	sf_index = event->function_id - base_id;
+	mutex_lock(&table->table_lock);
+	sf_dev = xa_load(&table->devices, sf_index);
+	switch (event->new_vhca_state) {
+	case MLX5_VHCA_STATE_INVALID:
+	case MLX5_VHCA_STATE_ALLOCATED:
+		if (sf_dev)
+			mlx5_sf_dev_del(table->dev, sf_dev, sf_index);
+		break;
+	case MLX5_VHCA_STATE_TEARDOWN_REQUEST:
+		if (sf_dev)
+			mlx5_sf_dev_del(table->dev, sf_dev, sf_index);
+		else
+			mlx5_core_err(table->dev,
+				      "SF DEV: teardown state for invalid dev index=%d fn_id=0x%x\n",
+				      sf_index, event->sw_function_id);
+		break;
+	case MLX5_VHCA_STATE_ACTIVE:
+		if (!sf_dev)
+			mlx5_sf_dev_add(table->dev, sf_index, event->function_id,
+					event->sw_function_id);
+		break;
+	default:
+		break;
+	}
+	mutex_unlock(&table->table_lock);
+	return 0;
+}
+
+static int mlx5_sf_dev_vhca_arm_all(struct mlx5_sf_dev_table *table)
+{
+	struct mlx5_core_dev *dev = table->dev;
+	u16 max_functions;
+	u16 function_id;
+	int err = 0;
+	int i;
+
+	max_functions = mlx5_sf_max_functions(dev);
+	function_id = MLX5_CAP_GEN(dev, sf_base_id);
+	/* Arm the vhca context as the vhca event notifier */
+	for (i = 0; i < max_functions; i++) {
+		err = mlx5_vhca_event_arm(dev, function_id);
+		if (err)
+			return err;
+
+		function_id++;
+	}
+	return 0;
+}
+
+static void mlx5_sf_dev_add_active_work(struct work_struct *work)
+{
+	struct mlx5_sf_dev_table *table = container_of(work, struct mlx5_sf_dev_table, work);
+	u32 out[MLX5_ST_SZ_DW(query_vhca_state_out)] = {};
+	struct mlx5_core_dev *dev = table->dev;
+	u16 max_functions;
+	u16 function_id;
+	u16 sw_func_id;
+	int err = 0;
+	u8 state;
+	int i;
+
+	max_functions = mlx5_sf_max_functions(dev);
+	function_id = MLX5_CAP_GEN(dev, sf_base_id);
+	for (i = 0; i < max_functions; i++, function_id++) {
+		if (table->stop_active_wq)
+			return;
+		err = mlx5_cmd_query_vhca_state(dev, function_id, out, sizeof(out));
+		if (err)
+			/* A failure of specific vhca doesn't mean others will
+			 * fail as well.
+			 */
+			continue;
+		state = MLX5_GET(query_vhca_state_out, out, vhca_state_context.vhca_state);
+		if (state != MLX5_VHCA_STATE_ACTIVE)
+			continue;
+
+		sw_func_id = MLX5_GET(query_vhca_state_out, out, vhca_state_context.sw_function_id);
+		mutex_lock(&table->table_lock);
+		/* Don't probe device which is already probe */
+		if (!xa_load(&table->devices, i))
+			mlx5_sf_dev_add(dev, i, function_id, sw_func_id);
+		/* There is a race where SF got inactive after the query
+		 * above. e.g.: the query returns that the state of the
+		 * SF is active, and after that the eswitch manager set it to
+		 * inactive.
+		 * This case cannot be managed in SW, since the probing of the
+		 * SF is on one system, and the inactivation is on a different
+		 * system.
+		 * If the inactive is done after the SF perform init_hca(),
+		 * the SF will fully probe and then removed. If it was
+		 * done before init_hca(), the SF probe will fail.
+		 */
+		mutex_unlock(&table->table_lock);
+	}
+}
+
+/* In case SFs are generated externally, probe active SFs */
+static int mlx5_sf_dev_queue_active_work(struct mlx5_sf_dev_table *table)
+{
+	if (MLX5_CAP_GEN(table->dev, eswitch_manager))
+		return 0; /* the table is local */
+
+	/* Use a workqueue to probe active SFs, which are in large
+	 * quantity and may take up to minutes to probe.
+	 */
+	table->active_wq = create_singlethread_workqueue("mlx5_active_sf");
+	if (!table->active_wq)
+		return -ENOMEM;
+	INIT_WORK(&table->work, &mlx5_sf_dev_add_active_work);
+	queue_work(table->active_wq, &table->work);
+	return 0;
+}
+
+static void mlx5_sf_dev_destroy_active_work(struct mlx5_sf_dev_table *table)
+{
+	if (table->active_wq) {
+		table->stop_active_wq = true;
+		destroy_workqueue(table->active_wq);
+	}
+}
+
+void mlx5_sf_dev_table_create(struct mlx5_core_dev *dev)
+{
+	struct mlx5_sf_dev_table *table;
+	unsigned int max_sfs;
+	int err;
+
+	if (!mlx5_sf_dev_supported(dev) || !mlx5_vhca_event_supported(dev))
+		return;
+
+	table = kzalloc(sizeof(*table), GFP_KERNEL);
+	if (!table) {
+		err = -ENOMEM;
+		goto table_err;
+	}
+
+	table->nb.notifier_call = mlx5_sf_dev_state_change_handler;
+	table->dev = dev;
+	if (MLX5_CAP_GEN(dev, max_num_sf))
+		max_sfs = MLX5_CAP_GEN(dev, max_num_sf);
+	else
+		max_sfs = 1 << MLX5_CAP_GEN(dev, log_max_sf);
+	table->sf_bar_length = 1 << (MLX5_CAP_GEN(dev, log_min_sf_size) + 12);
+	table->base_address = pci_resource_start(dev->pdev, 2);
+	table->max_sfs = max_sfs;
+	xa_init(&table->devices);
+	mutex_init(&table->table_lock);
+	dev->priv.sf_dev_table = table;
+
+	err = mlx5_vhca_event_notifier_register(dev, &table->nb);
+	if (err)
+		goto vhca_err;
+
+	err = mlx5_sf_dev_queue_active_work(table);
+	if (err)
+		goto add_active_err;
+
+	err = mlx5_sf_dev_vhca_arm_all(table);
+	if (err)
+		goto arm_err;
+	mlx5_core_dbg(dev, "SF DEV: max sf devices=%d\n", max_sfs);
+	return;
+
+arm_err:
+	mlx5_sf_dev_destroy_active_work(table);
+add_active_err:
+	mlx5_vhca_event_notifier_unregister(dev, &table->nb);
+vhca_err:
+	table->max_sfs = 0;
+	kfree(table);
+	dev->priv.sf_dev_table = NULL;
+table_err:
+	mlx5_core_err(dev, "SF DEV table create err = %d\n", err);
+}
+
+static void mlx5_sf_dev_destroy_all(struct mlx5_sf_dev_table *table)
+{
+	struct mlx5_sf_dev *sf_dev;
+	unsigned long index;
+
+	xa_for_each(&table->devices, index, sf_dev) {
+		xa_erase(&table->devices, index);
+		mlx5_sf_dev_remove(table->dev, sf_dev);
+	}
+}
+
+void mlx5_sf_dev_table_destroy(struct mlx5_core_dev *dev)
+{
+	struct mlx5_sf_dev_table *table = dev->priv.sf_dev_table;
+
+	if (!table)
+		return;
+
+	mlx5_sf_dev_destroy_active_work(table);
+	mlx5_vhca_event_notifier_unregister(dev, &table->nb);
+	mutex_destroy(&table->table_lock);
+
+	/* Now that event handler is not running, it is safe to destroy
+	 * the sf device without race.
+	 */
+	mlx5_sf_dev_destroy_all(table);
+
+	WARN_ON(!xa_empty(&table->devices));
+	kfree(table);
+	dev->priv.sf_dev_table = NULL;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/sf/dev/dev.h b/drivers/net/ethernet/mellanox/mlx5/core/sf/dev/dev.h
new file mode 100644
index 000000000..2a66a427e
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/sf/dev/dev.h
@@ -0,0 +1,56 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2020 Mellanox Technologies Ltd */
+
+#ifndef __MLX5_SF_DEV_H__
+#define __MLX5_SF_DEV_H__
+
+#ifdef CONFIG_MLX5_SF
+
+#include <linux/auxiliary_bus.h>
+
+#define MLX5_SF_DEV_ID_NAME "sf"
+
+struct mlx5_sf_dev {
+	struct auxiliary_device adev;
+	struct mlx5_core_dev *parent_mdev;
+	struct mlx5_core_dev *mdev;
+	phys_addr_t bar_base_addr;
+	u32 sfnum;
+	u16 fn_id;
+};
+
+void mlx5_sf_dev_table_create(struct mlx5_core_dev *dev);
+void mlx5_sf_dev_table_destroy(struct mlx5_core_dev *dev);
+
+int mlx5_sf_driver_register(void);
+void mlx5_sf_driver_unregister(void);
+
+bool mlx5_sf_dev_allocated(const struct mlx5_core_dev *dev);
+
+#else
+
+static inline void mlx5_sf_dev_table_create(struct mlx5_core_dev *dev)
+{
+}
+
+static inline void mlx5_sf_dev_table_destroy(struct mlx5_core_dev *dev)
+{
+}
+
+static inline int mlx5_sf_driver_register(void)
+{
+	return 0;
+}
+
+static inline void mlx5_sf_driver_unregister(void)
+{
+}
+
+static inline bool mlx5_sf_dev_allocated(const struct mlx5_core_dev *dev)
+{
+	return false;
+}
+
+#endif
+
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/sf/dev/diag/dev_tracepoint.h b/drivers/net/ethernet/mellanox/mlx5/core/sf/dev/diag/dev_tracepoint.h
new file mode 100644
index 000000000..7f7c9af5d
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/sf/dev/diag/dev_tracepoint.h
@@ -0,0 +1,58 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. All rights reserved. */
+
+#undef TRACE_SYSTEM
+#define TRACE_SYSTEM mlx5
+
+#if !defined(_MLX5_SF_DEV_TP_) || defined(TRACE_HEADER_MULTI_READ)
+#define _MLX5_SF_DEV_TP_
+
+#include <linux/tracepoint.h>
+#include <linux/mlx5/driver.h>
+#include "../../dev/dev.h"
+
+DECLARE_EVENT_CLASS(mlx5_sf_dev_template,
+		    TP_PROTO(const struct mlx5_core_dev *dev,
+			     const struct mlx5_sf_dev *sfdev,
+			     int aux_id),
+		    TP_ARGS(dev, sfdev, aux_id),
+		    TP_STRUCT__entry(__string(devname, dev_name(dev->device))
+				     __field(const struct mlx5_sf_dev*, sfdev)
+				     __field(int, aux_id)
+				     __field(u16, hw_fn_id)
+				     __field(u32, sfnum)
+		    ),
+		    TP_fast_assign(__assign_str(devname, dev_name(dev->device));
+				   __entry->sfdev = sfdev;
+				   __entry->aux_id = aux_id;
+				   __entry->hw_fn_id = sfdev->fn_id;
+				   __entry->sfnum = sfdev->sfnum;
+		    ),
+		    TP_printk("(%s) sfdev=%pK aux_id=%d hw_id=0x%x sfnum=%u\n",
+			      __get_str(devname), __entry->sfdev,
+			      __entry->aux_id, __entry->hw_fn_id,
+			      __entry->sfnum)
+);
+
+DEFINE_EVENT(mlx5_sf_dev_template, mlx5_sf_dev_add,
+	     TP_PROTO(const struct mlx5_core_dev *dev,
+		      const struct mlx5_sf_dev *sfdev,
+		      int aux_id),
+	     TP_ARGS(dev, sfdev, aux_id)
+	     );
+
+DEFINE_EVENT(mlx5_sf_dev_template, mlx5_sf_dev_del,
+	     TP_PROTO(const struct mlx5_core_dev *dev,
+		      const struct mlx5_sf_dev *sfdev,
+		      int aux_id),
+	     TP_ARGS(dev, sfdev, aux_id)
+	     );
+
+#endif /* _MLX5_SF_DEV_TP_ */
+
+/* This part must be outside protection */
+#undef TRACE_INCLUDE_PATH
+#define TRACE_INCLUDE_PATH sf/dev/diag
+#undef TRACE_INCLUDE_FILE
+#define TRACE_INCLUDE_FILE dev_tracepoint
+#include <trace/define_trace.h>
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/sf/dev/driver.c b/drivers/net/ethernet/mellanox/mlx5/core/sf/dev/driver.c
new file mode 100644
index 000000000..2424cdf9c
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/sf/dev/driver.c
@@ -0,0 +1,104 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2020 Mellanox Technologies Ltd */
+
+#include <linux/mlx5/driver.h>
+#include <linux/mlx5/device.h>
+#include "mlx5_core.h"
+#include "dev.h"
+#include "devlink.h"
+
+static int mlx5_sf_dev_probe(struct auxiliary_device *adev, const struct auxiliary_device_id *id)
+{
+	struct mlx5_sf_dev *sf_dev = container_of(adev, struct mlx5_sf_dev, adev);
+	struct mlx5_core_dev *mdev;
+	struct devlink *devlink;
+	int err;
+
+	devlink = mlx5_devlink_alloc(&adev->dev);
+	if (!devlink)
+		return -ENOMEM;
+
+	mdev = devlink_priv(devlink);
+	mdev->device = &adev->dev;
+	mdev->pdev = sf_dev->parent_mdev->pdev;
+	mdev->bar_addr = sf_dev->bar_base_addr;
+	mdev->iseg_base = sf_dev->bar_base_addr;
+	mdev->coredev_type = MLX5_COREDEV_SF;
+	mdev->priv.parent_mdev = sf_dev->parent_mdev;
+	mdev->priv.adev_idx = adev->id;
+	sf_dev->mdev = mdev;
+
+	err = mlx5_mdev_init(mdev, MLX5_DEFAULT_PROF);
+	if (err) {
+		mlx5_core_warn(mdev, "mlx5_mdev_init on err=%d\n", err);
+		goto mdev_err;
+	}
+
+	mdev->iseg = ioremap(mdev->iseg_base, sizeof(*mdev->iseg));
+	if (!mdev->iseg) {
+		mlx5_core_warn(mdev, "remap error\n");
+		err = -ENOMEM;
+		goto remap_err;
+	}
+
+	err = mlx5_init_one(mdev);
+	if (err) {
+		mlx5_core_warn(mdev, "mlx5_init_one err=%d\n", err);
+		goto init_one_err;
+	}
+	devlink_register(devlink);
+	return 0;
+
+init_one_err:
+	iounmap(mdev->iseg);
+remap_err:
+	mlx5_mdev_uninit(mdev);
+mdev_err:
+	mlx5_devlink_free(devlink);
+	return err;
+}
+
+static void mlx5_sf_dev_remove(struct auxiliary_device *adev)
+{
+	struct mlx5_sf_dev *sf_dev = container_of(adev, struct mlx5_sf_dev, adev);
+	struct devlink *devlink = priv_to_devlink(sf_dev->mdev);
+
+	mlx5_drain_health_wq(sf_dev->mdev);
+	devlink_unregister(devlink);
+	mlx5_uninit_one(sf_dev->mdev);
+	iounmap(sf_dev->mdev->iseg);
+	mlx5_mdev_uninit(sf_dev->mdev);
+	mlx5_devlink_free(devlink);
+}
+
+static void mlx5_sf_dev_shutdown(struct auxiliary_device *adev)
+{
+	struct mlx5_sf_dev *sf_dev = container_of(adev, struct mlx5_sf_dev, adev);
+
+	mlx5_unload_one(sf_dev->mdev, false);
+}
+
+static const struct auxiliary_device_id mlx5_sf_dev_id_table[] = {
+	{ .name = MLX5_ADEV_NAME "." MLX5_SF_DEV_ID_NAME, },
+	{ },
+};
+
+MODULE_DEVICE_TABLE(auxiliary, mlx5_sf_dev_id_table);
+
+static struct auxiliary_driver mlx5_sf_driver = {
+	.name = MLX5_SF_DEV_ID_NAME,
+	.probe = mlx5_sf_dev_probe,
+	.remove = mlx5_sf_dev_remove,
+	.shutdown = mlx5_sf_dev_shutdown,
+	.id_table = mlx5_sf_dev_id_table,
+};
+
+int mlx5_sf_driver_register(void)
+{
+	return auxiliary_driver_register(&mlx5_sf_driver);
+}
+
+void mlx5_sf_driver_unregister(void)
+{
+	auxiliary_driver_unregister(&mlx5_sf_driver);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/sf/devlink.c b/drivers/net/ethernet/mellanox/mlx5/core/sf/devlink.c
new file mode 100644
index 000000000..7d955a4d9
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/sf/devlink.c
@@ -0,0 +1,571 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2020 Mellanox Technologies Ltd */
+
+#include <linux/mlx5/driver.h>
+#include "eswitch.h"
+#include "priv.h"
+#include "sf/dev/dev.h"
+#include "mlx5_ifc_vhca_event.h"
+#include "vhca_event.h"
+#include "ecpf.h"
+#define CREATE_TRACE_POINTS
+#include "diag/sf_tracepoint.h"
+
+struct mlx5_sf {
+	struct devlink_port dl_port;
+	unsigned int port_index;
+	u32 controller;
+	u16 id;
+	u16 hw_fn_id;
+	u16 hw_state;
+};
+
+struct mlx5_sf_table {
+	struct mlx5_core_dev *dev; /* To refer from notifier context. */
+	struct xarray port_indices; /* port index based lookup. */
+	refcount_t refcount;
+	struct completion disable_complete;
+	struct mutex sf_state_lock; /* Serializes sf state among user cmds & vhca event handler. */
+	struct notifier_block esw_nb;
+	struct notifier_block vhca_nb;
+	u8 ecpu: 1;
+};
+
+static struct mlx5_sf *
+mlx5_sf_lookup_by_index(struct mlx5_sf_table *table, unsigned int port_index)
+{
+	return xa_load(&table->port_indices, port_index);
+}
+
+static struct mlx5_sf *
+mlx5_sf_lookup_by_function_id(struct mlx5_sf_table *table, unsigned int fn_id)
+{
+	unsigned long index;
+	struct mlx5_sf *sf;
+
+	xa_for_each(&table->port_indices, index, sf) {
+		if (sf->hw_fn_id == fn_id)
+			return sf;
+	}
+	return NULL;
+}
+
+static int mlx5_sf_id_insert(struct mlx5_sf_table *table, struct mlx5_sf *sf)
+{
+	return xa_insert(&table->port_indices, sf->port_index, sf, GFP_KERNEL);
+}
+
+static void mlx5_sf_id_erase(struct mlx5_sf_table *table, struct mlx5_sf *sf)
+{
+	xa_erase(&table->port_indices, sf->port_index);
+}
+
+static struct mlx5_sf *
+mlx5_sf_alloc(struct mlx5_sf_table *table, struct mlx5_eswitch *esw,
+	      u32 controller, u32 sfnum, struct netlink_ext_ack *extack)
+{
+	unsigned int dl_port_index;
+	struct mlx5_sf *sf;
+	u16 hw_fn_id;
+	int id_err;
+	int err;
+
+	if (!mlx5_esw_offloads_controller_valid(esw, controller)) {
+		NL_SET_ERR_MSG_MOD(extack, "Invalid controller number");
+		return ERR_PTR(-EINVAL);
+	}
+
+	id_err = mlx5_sf_hw_table_sf_alloc(table->dev, controller, sfnum);
+	if (id_err < 0) {
+		err = id_err;
+		goto id_err;
+	}
+
+	sf = kzalloc(sizeof(*sf), GFP_KERNEL);
+	if (!sf) {
+		err = -ENOMEM;
+		goto alloc_err;
+	}
+	sf->id = id_err;
+	hw_fn_id = mlx5_sf_sw_to_hw_id(table->dev, controller, sf->id);
+	dl_port_index = mlx5_esw_vport_to_devlink_port_index(table->dev, hw_fn_id);
+	sf->port_index = dl_port_index;
+	sf->hw_fn_id = hw_fn_id;
+	sf->hw_state = MLX5_VHCA_STATE_ALLOCATED;
+	sf->controller = controller;
+
+	err = mlx5_sf_id_insert(table, sf);
+	if (err)
+		goto insert_err;
+
+	return sf;
+
+insert_err:
+	kfree(sf);
+alloc_err:
+	mlx5_sf_hw_table_sf_free(table->dev, controller, id_err);
+id_err:
+	if (err == -EEXIST)
+		NL_SET_ERR_MSG_MOD(extack, "SF already exist. Choose different sfnum");
+	return ERR_PTR(err);
+}
+
+static void mlx5_sf_free(struct mlx5_sf_table *table, struct mlx5_sf *sf)
+{
+	mlx5_sf_id_erase(table, sf);
+	mlx5_sf_hw_table_sf_free(table->dev, sf->controller, sf->id);
+	trace_mlx5_sf_free(table->dev, sf->port_index, sf->controller, sf->hw_fn_id);
+	kfree(sf);
+}
+
+static struct mlx5_sf_table *mlx5_sf_table_try_get(struct mlx5_core_dev *dev)
+{
+	struct mlx5_sf_table *table = dev->priv.sf_table;
+
+	if (!table)
+		return NULL;
+
+	return refcount_inc_not_zero(&table->refcount) ? table : NULL;
+}
+
+static void mlx5_sf_table_put(struct mlx5_sf_table *table)
+{
+	if (refcount_dec_and_test(&table->refcount))
+		complete(&table->disable_complete);
+}
+
+static enum devlink_port_fn_state mlx5_sf_to_devlink_state(u8 hw_state)
+{
+	switch (hw_state) {
+	case MLX5_VHCA_STATE_ACTIVE:
+	case MLX5_VHCA_STATE_IN_USE:
+		return DEVLINK_PORT_FN_STATE_ACTIVE;
+	case MLX5_VHCA_STATE_INVALID:
+	case MLX5_VHCA_STATE_ALLOCATED:
+	case MLX5_VHCA_STATE_TEARDOWN_REQUEST:
+	default:
+		return DEVLINK_PORT_FN_STATE_INACTIVE;
+	}
+}
+
+static enum devlink_port_fn_opstate mlx5_sf_to_devlink_opstate(u8 hw_state)
+{
+	switch (hw_state) {
+	case MLX5_VHCA_STATE_IN_USE:
+	case MLX5_VHCA_STATE_TEARDOWN_REQUEST:
+		return DEVLINK_PORT_FN_OPSTATE_ATTACHED;
+	case MLX5_VHCA_STATE_INVALID:
+	case MLX5_VHCA_STATE_ALLOCATED:
+	case MLX5_VHCA_STATE_ACTIVE:
+	default:
+		return DEVLINK_PORT_FN_OPSTATE_DETACHED;
+	}
+}
+
+static bool mlx5_sf_is_active(const struct mlx5_sf *sf)
+{
+	return sf->hw_state == MLX5_VHCA_STATE_ACTIVE || sf->hw_state == MLX5_VHCA_STATE_IN_USE;
+}
+
+int mlx5_devlink_sf_port_fn_state_get(struct devlink_port *dl_port,
+				      enum devlink_port_fn_state *state,
+				      enum devlink_port_fn_opstate *opstate,
+				      struct netlink_ext_ack *extack)
+{
+	struct mlx5_core_dev *dev = devlink_priv(dl_port->devlink);
+	struct mlx5_sf_table *table;
+	struct mlx5_sf *sf;
+	int err = 0;
+
+	table = mlx5_sf_table_try_get(dev);
+	if (!table)
+		return -EOPNOTSUPP;
+
+	sf = mlx5_sf_lookup_by_index(table, dl_port->index);
+	if (!sf) {
+		err = -EOPNOTSUPP;
+		goto sf_err;
+	}
+	mutex_lock(&table->sf_state_lock);
+	*state = mlx5_sf_to_devlink_state(sf->hw_state);
+	*opstate = mlx5_sf_to_devlink_opstate(sf->hw_state);
+	mutex_unlock(&table->sf_state_lock);
+sf_err:
+	mlx5_sf_table_put(table);
+	return err;
+}
+
+static int mlx5_sf_activate(struct mlx5_core_dev *dev, struct mlx5_sf *sf,
+			    struct netlink_ext_ack *extack)
+{
+	int err;
+
+	if (mlx5_sf_is_active(sf))
+		return 0;
+	if (sf->hw_state != MLX5_VHCA_STATE_ALLOCATED) {
+		NL_SET_ERR_MSG_MOD(extack, "SF is inactivated but it is still attached");
+		return -EBUSY;
+	}
+
+	err = mlx5_cmd_sf_enable_hca(dev, sf->hw_fn_id);
+	if (err)
+		return err;
+
+	sf->hw_state = MLX5_VHCA_STATE_ACTIVE;
+	trace_mlx5_sf_activate(dev, sf->port_index, sf->controller, sf->hw_fn_id);
+	return 0;
+}
+
+static int mlx5_sf_deactivate(struct mlx5_core_dev *dev, struct mlx5_sf *sf)
+{
+	int err;
+
+	if (!mlx5_sf_is_active(sf))
+		return 0;
+
+	err = mlx5_cmd_sf_disable_hca(dev, sf->hw_fn_id);
+	if (err)
+		return err;
+
+	sf->hw_state = MLX5_VHCA_STATE_TEARDOWN_REQUEST;
+	trace_mlx5_sf_deactivate(dev, sf->port_index, sf->controller, sf->hw_fn_id);
+	return 0;
+}
+
+static int mlx5_sf_state_set(struct mlx5_core_dev *dev, struct mlx5_sf_table *table,
+			     struct mlx5_sf *sf,
+			     enum devlink_port_fn_state state,
+			     struct netlink_ext_ack *extack)
+{
+	int err = 0;
+
+	mutex_lock(&table->sf_state_lock);
+	if (state == mlx5_sf_to_devlink_state(sf->hw_state))
+		goto out;
+	if (state == DEVLINK_PORT_FN_STATE_ACTIVE)
+		err = mlx5_sf_activate(dev, sf, extack);
+	else if (state == DEVLINK_PORT_FN_STATE_INACTIVE)
+		err = mlx5_sf_deactivate(dev, sf);
+	else
+		err = -EINVAL;
+out:
+	mutex_unlock(&table->sf_state_lock);
+	return err;
+}
+
+int mlx5_devlink_sf_port_fn_state_set(struct devlink_port *dl_port,
+				      enum devlink_port_fn_state state,
+				      struct netlink_ext_ack *extack)
+{
+	struct mlx5_core_dev *dev = devlink_priv(dl_port->devlink);
+	struct mlx5_sf_table *table;
+	struct mlx5_sf *sf;
+	int err;
+
+	table = mlx5_sf_table_try_get(dev);
+	if (!table) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "Port state set is only supported in eswitch switchdev mode or SF ports are disabled.");
+		return -EOPNOTSUPP;
+	}
+	sf = mlx5_sf_lookup_by_index(table, dl_port->index);
+	if (!sf) {
+		err = -ENODEV;
+		goto out;
+	}
+
+	err = mlx5_sf_state_set(dev, table, sf, state, extack);
+out:
+	mlx5_sf_table_put(table);
+	return err;
+}
+
+static int mlx5_sf_add(struct mlx5_core_dev *dev, struct mlx5_sf_table *table,
+		       const struct devlink_port_new_attrs *new_attr,
+		       struct netlink_ext_ack *extack,
+		       unsigned int *new_port_index)
+{
+	struct mlx5_eswitch *esw = dev->priv.eswitch;
+	struct mlx5_sf *sf;
+	int err;
+
+	sf = mlx5_sf_alloc(table, esw, new_attr->controller, new_attr->sfnum, extack);
+	if (IS_ERR(sf))
+		return PTR_ERR(sf);
+
+	err = mlx5_esw_offloads_sf_vport_enable(esw, &sf->dl_port, sf->hw_fn_id,
+						new_attr->controller, new_attr->sfnum);
+	if (err)
+		goto esw_err;
+	*new_port_index = sf->port_index;
+	trace_mlx5_sf_add(dev, sf->port_index, sf->controller, sf->hw_fn_id, new_attr->sfnum);
+	return 0;
+
+esw_err:
+	mlx5_sf_free(table, sf);
+	return err;
+}
+
+static int
+mlx5_sf_new_check_attr(struct mlx5_core_dev *dev, const struct devlink_port_new_attrs *new_attr,
+		       struct netlink_ext_ack *extack)
+{
+	if (new_attr->flavour != DEVLINK_PORT_FLAVOUR_PCI_SF) {
+		NL_SET_ERR_MSG_MOD(extack, "Driver supports only SF port addition");
+		return -EOPNOTSUPP;
+	}
+	if (new_attr->port_index_valid) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "Driver does not support user defined port index assignment");
+		return -EOPNOTSUPP;
+	}
+	if (!new_attr->sfnum_valid) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "User must provide unique sfnum. Driver does not support auto assignment");
+		return -EOPNOTSUPP;
+	}
+	if (new_attr->controller_valid && new_attr->controller &&
+	    !mlx5_core_is_ecpf_esw_manager(dev)) {
+		NL_SET_ERR_MSG_MOD(extack, "External controller is unsupported");
+		return -EOPNOTSUPP;
+	}
+	if (new_attr->pfnum != mlx5_get_dev_index(dev)) {
+		NL_SET_ERR_MSG_MOD(extack, "Invalid pfnum supplied");
+		return -EOPNOTSUPP;
+	}
+	return 0;
+}
+
+int mlx5_devlink_sf_port_new(struct devlink *devlink,
+			     const struct devlink_port_new_attrs *new_attr,
+			     struct netlink_ext_ack *extack,
+			     unsigned int *new_port_index)
+{
+	struct mlx5_core_dev *dev = devlink_priv(devlink);
+	struct mlx5_sf_table *table;
+	int err;
+
+	err = mlx5_sf_new_check_attr(dev, new_attr, extack);
+	if (err)
+		return err;
+
+	table = mlx5_sf_table_try_get(dev);
+	if (!table) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "Port add is only supported in eswitch switchdev mode or SF ports are disabled.");
+		return -EOPNOTSUPP;
+	}
+	err = mlx5_sf_add(dev, table, new_attr, extack, new_port_index);
+	mlx5_sf_table_put(table);
+	return err;
+}
+
+static void mlx5_sf_dealloc(struct mlx5_sf_table *table, struct mlx5_sf *sf)
+{
+	if (sf->hw_state == MLX5_VHCA_STATE_ALLOCATED) {
+		mlx5_sf_free(table, sf);
+	} else if (mlx5_sf_is_active(sf)) {
+		/* Even if its active, it is treated as in_use because by the time,
+		 * it is disabled here, it may getting used. So it is safe to
+		 * always look for the event to ensure that it is recycled only after
+		 * firmware gives confirmation that it is detached by the driver.
+		 */
+		mlx5_cmd_sf_disable_hca(table->dev, sf->hw_fn_id);
+		mlx5_sf_hw_table_sf_deferred_free(table->dev, sf->controller, sf->id);
+		kfree(sf);
+	} else {
+		mlx5_sf_hw_table_sf_deferred_free(table->dev, sf->controller, sf->id);
+		kfree(sf);
+	}
+}
+
+int mlx5_devlink_sf_port_del(struct devlink *devlink, unsigned int port_index,
+			     struct netlink_ext_ack *extack)
+{
+	struct mlx5_core_dev *dev = devlink_priv(devlink);
+	struct mlx5_eswitch *esw = dev->priv.eswitch;
+	struct mlx5_sf_table *table;
+	struct mlx5_sf *sf;
+	int err = 0;
+
+	table = mlx5_sf_table_try_get(dev);
+	if (!table) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "Port del is only supported in eswitch switchdev mode or SF ports are disabled.");
+		return -EOPNOTSUPP;
+	}
+	sf = mlx5_sf_lookup_by_index(table, port_index);
+	if (!sf) {
+		err = -ENODEV;
+		goto sf_err;
+	}
+
+	mlx5_esw_offloads_sf_vport_disable(esw, sf->hw_fn_id);
+	mlx5_sf_id_erase(table, sf);
+
+	mutex_lock(&table->sf_state_lock);
+	mlx5_sf_dealloc(table, sf);
+	mutex_unlock(&table->sf_state_lock);
+sf_err:
+	mlx5_sf_table_put(table);
+	return err;
+}
+
+static bool mlx5_sf_state_update_check(const struct mlx5_sf *sf, u8 new_state)
+{
+	if (sf->hw_state == MLX5_VHCA_STATE_ACTIVE && new_state == MLX5_VHCA_STATE_IN_USE)
+		return true;
+
+	if (sf->hw_state == MLX5_VHCA_STATE_IN_USE && new_state == MLX5_VHCA_STATE_ACTIVE)
+		return true;
+
+	if (sf->hw_state == MLX5_VHCA_STATE_TEARDOWN_REQUEST &&
+	    new_state == MLX5_VHCA_STATE_ALLOCATED)
+		return true;
+
+	return false;
+}
+
+static int mlx5_sf_vhca_event(struct notifier_block *nb, unsigned long opcode, void *data)
+{
+	struct mlx5_sf_table *table = container_of(nb, struct mlx5_sf_table, vhca_nb);
+	const struct mlx5_vhca_state_event *event = data;
+	bool update = false;
+	struct mlx5_sf *sf;
+
+	table = mlx5_sf_table_try_get(table->dev);
+	if (!table)
+		return 0;
+
+	mutex_lock(&table->sf_state_lock);
+	sf = mlx5_sf_lookup_by_function_id(table, event->function_id);
+	if (!sf)
+		goto sf_err;
+
+	/* When driver is attached or detached to a function, an event
+	 * notifies such state change.
+	 */
+	update = mlx5_sf_state_update_check(sf, event->new_vhca_state);
+	if (update)
+		sf->hw_state = event->new_vhca_state;
+	trace_mlx5_sf_update_state(table->dev, sf->port_index, sf->controller,
+				   sf->hw_fn_id, sf->hw_state);
+sf_err:
+	mutex_unlock(&table->sf_state_lock);
+	mlx5_sf_table_put(table);
+	return 0;
+}
+
+static void mlx5_sf_table_enable(struct mlx5_sf_table *table)
+{
+	init_completion(&table->disable_complete);
+	refcount_set(&table->refcount, 1);
+}
+
+static void mlx5_sf_deactivate_all(struct mlx5_sf_table *table)
+{
+	struct mlx5_eswitch *esw = table->dev->priv.eswitch;
+	unsigned long index;
+	struct mlx5_sf *sf;
+
+	/* At this point, no new user commands can start and no vhca event can
+	 * arrive. It is safe to destroy all user created SFs.
+	 */
+	xa_for_each(&table->port_indices, index, sf) {
+		mlx5_esw_offloads_sf_vport_disable(esw, sf->hw_fn_id);
+		mlx5_sf_id_erase(table, sf);
+		mlx5_sf_dealloc(table, sf);
+	}
+}
+
+static void mlx5_sf_table_disable(struct mlx5_sf_table *table)
+{
+	if (!refcount_read(&table->refcount))
+		return;
+
+	/* Balances with refcount_set; drop the reference so that new user cmd cannot start
+	 * and new vhca event handler cannot run.
+	 */
+	mlx5_sf_table_put(table);
+	wait_for_completion(&table->disable_complete);
+
+	mlx5_sf_deactivate_all(table);
+}
+
+static int mlx5_sf_esw_event(struct notifier_block *nb, unsigned long event, void *data)
+{
+	struct mlx5_sf_table *table = container_of(nb, struct mlx5_sf_table, esw_nb);
+	const struct mlx5_esw_event_info *mode = data;
+
+	switch (mode->new_mode) {
+	case MLX5_ESWITCH_OFFLOADS:
+		mlx5_sf_table_enable(table);
+		break;
+	case MLX5_ESWITCH_LEGACY:
+		mlx5_sf_table_disable(table);
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static bool mlx5_sf_table_supported(const struct mlx5_core_dev *dev)
+{
+	return dev->priv.eswitch && MLX5_ESWITCH_MANAGER(dev) &&
+	       mlx5_sf_hw_table_supported(dev);
+}
+
+int mlx5_sf_table_init(struct mlx5_core_dev *dev)
+{
+	struct mlx5_sf_table *table;
+	int err;
+
+	if (!mlx5_sf_table_supported(dev) || !mlx5_vhca_event_supported(dev))
+		return 0;
+
+	table = kzalloc(sizeof(*table), GFP_KERNEL);
+	if (!table)
+		return -ENOMEM;
+
+	mutex_init(&table->sf_state_lock);
+	table->dev = dev;
+	xa_init(&table->port_indices);
+	dev->priv.sf_table = table;
+	refcount_set(&table->refcount, 0);
+	table->esw_nb.notifier_call = mlx5_sf_esw_event;
+	err = mlx5_esw_event_notifier_register(dev->priv.eswitch, &table->esw_nb);
+	if (err)
+		goto reg_err;
+
+	table->vhca_nb.notifier_call = mlx5_sf_vhca_event;
+	err = mlx5_vhca_event_notifier_register(table->dev, &table->vhca_nb);
+	if (err)
+		goto vhca_err;
+
+	return 0;
+
+vhca_err:
+	mlx5_esw_event_notifier_unregister(dev->priv.eswitch, &table->esw_nb);
+reg_err:
+	mutex_destroy(&table->sf_state_lock);
+	kfree(table);
+	dev->priv.sf_table = NULL;
+	return err;
+}
+
+void mlx5_sf_table_cleanup(struct mlx5_core_dev *dev)
+{
+	struct mlx5_sf_table *table = dev->priv.sf_table;
+
+	if (!table)
+		return;
+
+	mlx5_vhca_event_notifier_unregister(table->dev, &table->vhca_nb);
+	mlx5_esw_event_notifier_unregister(dev->priv.eswitch, &table->esw_nb);
+	WARN_ON(refcount_read(&table->refcount));
+	mutex_destroy(&table->sf_state_lock);
+	WARN_ON(!xa_empty(&table->port_indices));
+	kfree(table);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/sf/diag/sf_tracepoint.h b/drivers/net/ethernet/mellanox/mlx5/core/sf/diag/sf_tracepoint.h
new file mode 100644
index 000000000..8bf1cd909
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/sf/diag/sf_tracepoint.h
@@ -0,0 +1,173 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. All rights reserved. */
+
+#undef TRACE_SYSTEM
+#define TRACE_SYSTEM mlx5
+
+#if !defined(_MLX5_SF_TP_) || defined(TRACE_HEADER_MULTI_READ)
+#define _MLX5_SF_TP_
+
+#include <linux/tracepoint.h>
+#include <linux/mlx5/driver.h>
+#include "sf/vhca_event.h"
+
+TRACE_EVENT(mlx5_sf_add,
+	    TP_PROTO(const struct mlx5_core_dev *dev,
+		     unsigned int port_index,
+		     u32 controller,
+		     u16 hw_fn_id,
+		     u32 sfnum),
+	    TP_ARGS(dev, port_index, controller, hw_fn_id, sfnum),
+	    TP_STRUCT__entry(__string(devname, dev_name(dev->device))
+			     __field(unsigned int, port_index)
+			     __field(u32, controller)
+			     __field(u16, hw_fn_id)
+			     __field(u32, sfnum)
+			    ),
+	    TP_fast_assign(__assign_str(devname, dev_name(dev->device));
+		    __entry->port_index = port_index;
+		    __entry->controller = controller;
+		    __entry->hw_fn_id = hw_fn_id;
+		    __entry->sfnum = sfnum;
+	    ),
+	    TP_printk("(%s) port_index=%u controller=%u hw_id=0x%x sfnum=%u\n",
+		      __get_str(devname), __entry->port_index, __entry->controller,
+		      __entry->hw_fn_id, __entry->sfnum)
+);
+
+TRACE_EVENT(mlx5_sf_free,
+	    TP_PROTO(const struct mlx5_core_dev *dev,
+		     unsigned int port_index,
+		     u32 controller,
+		     u16 hw_fn_id),
+	    TP_ARGS(dev, port_index, controller, hw_fn_id),
+	    TP_STRUCT__entry(__string(devname, dev_name(dev->device))
+			     __field(unsigned int, port_index)
+			     __field(u32, controller)
+			     __field(u16, hw_fn_id)
+			    ),
+	    TP_fast_assign(__assign_str(devname, dev_name(dev->device));
+		    __entry->port_index = port_index;
+		    __entry->controller = controller;
+		    __entry->hw_fn_id = hw_fn_id;
+	    ),
+	    TP_printk("(%s) port_index=%u controller=%u hw_id=0x%x\n",
+		      __get_str(devname), __entry->port_index, __entry->controller,
+		      __entry->hw_fn_id)
+);
+
+TRACE_EVENT(mlx5_sf_hwc_alloc,
+	    TP_PROTO(const struct mlx5_core_dev *dev,
+		     u32 controller,
+		     u16 hw_fn_id,
+		     u32 sfnum),
+	    TP_ARGS(dev, controller, hw_fn_id, sfnum),
+	    TP_STRUCT__entry(__string(devname, dev_name(dev->device))
+			     __field(u32, controller)
+			     __field(u16, hw_fn_id)
+			     __field(u32, sfnum)
+			    ),
+	    TP_fast_assign(__assign_str(devname, dev_name(dev->device));
+		    __entry->controller = controller;
+		    __entry->hw_fn_id = hw_fn_id;
+		    __entry->sfnum = sfnum;
+	    ),
+	    TP_printk("(%s) controller=%u hw_id=0x%x sfnum=%u\n",
+		      __get_str(devname), __entry->controller, __entry->hw_fn_id,
+		      __entry->sfnum)
+);
+
+TRACE_EVENT(mlx5_sf_hwc_free,
+	    TP_PROTO(const struct mlx5_core_dev *dev,
+		     u16 hw_fn_id),
+	    TP_ARGS(dev, hw_fn_id),
+	    TP_STRUCT__entry(__string(devname, dev_name(dev->device))
+			     __field(u16, hw_fn_id)
+			    ),
+	    TP_fast_assign(__assign_str(devname, dev_name(dev->device));
+		    __entry->hw_fn_id = hw_fn_id;
+	    ),
+	    TP_printk("(%s) hw_id=0x%x\n", __get_str(devname), __entry->hw_fn_id)
+);
+
+TRACE_EVENT(mlx5_sf_hwc_deferred_free,
+	    TP_PROTO(const struct mlx5_core_dev *dev,
+		     u16 hw_fn_id),
+	    TP_ARGS(dev, hw_fn_id),
+	    TP_STRUCT__entry(__string(devname, dev_name(dev->device))
+			     __field(u16, hw_fn_id)
+			    ),
+	    TP_fast_assign(__assign_str(devname, dev_name(dev->device));
+		    __entry->hw_fn_id = hw_fn_id;
+	    ),
+	    TP_printk("(%s) hw_id=0x%x\n", __get_str(devname), __entry->hw_fn_id)
+);
+
+DECLARE_EVENT_CLASS(mlx5_sf_state_template,
+		    TP_PROTO(const struct mlx5_core_dev *dev,
+			     u32 port_index,
+			     u32 controller,
+			     u16 hw_fn_id),
+		    TP_ARGS(dev, port_index, controller, hw_fn_id),
+		    TP_STRUCT__entry(__string(devname, dev_name(dev->device))
+				     __field(unsigned int, port_index)
+				     __field(u32, controller)
+				     __field(u16, hw_fn_id)),
+		    TP_fast_assign(__assign_str(devname, dev_name(dev->device));
+				   __entry->port_index = port_index;
+				   __entry->controller = controller;
+				   __entry->hw_fn_id = hw_fn_id;
+		    ),
+		    TP_printk("(%s) port_index=%u controller=%u hw_id=0x%x\n",
+			      __get_str(devname), __entry->port_index, __entry->controller,
+			      __entry->hw_fn_id)
+);
+
+DEFINE_EVENT(mlx5_sf_state_template, mlx5_sf_activate,
+	     TP_PROTO(const struct mlx5_core_dev *dev,
+		      u32 port_index,
+		      u32 controller,
+		      u16 hw_fn_id),
+	     TP_ARGS(dev, port_index, controller, hw_fn_id)
+	     );
+
+DEFINE_EVENT(mlx5_sf_state_template, mlx5_sf_deactivate,
+	     TP_PROTO(const struct mlx5_core_dev *dev,
+		      u32 port_index,
+		      u32 controller,
+		      u16 hw_fn_id),
+	     TP_ARGS(dev, port_index, controller, hw_fn_id)
+	     );
+
+TRACE_EVENT(mlx5_sf_update_state,
+	    TP_PROTO(const struct mlx5_core_dev *dev,
+		     unsigned int port_index,
+		     u32 controller,
+		     u16 hw_fn_id,
+		     u8 state),
+	    TP_ARGS(dev, port_index, controller, hw_fn_id, state),
+	    TP_STRUCT__entry(__string(devname, dev_name(dev->device))
+			     __field(unsigned int, port_index)
+			     __field(u32, controller)
+			     __field(u16, hw_fn_id)
+			     __field(u8, state)
+			    ),
+	    TP_fast_assign(__assign_str(devname, dev_name(dev->device));
+		    __entry->port_index = port_index;
+		    __entry->controller = controller;
+		    __entry->hw_fn_id = hw_fn_id;
+		    __entry->state = state;
+	    ),
+	    TP_printk("(%s) port_index=%u controller=%u hw_id=0x%x state=%u\n",
+		      __get_str(devname), __entry->port_index, __entry->controller,
+		      __entry->hw_fn_id, __entry->state)
+);
+
+#endif /* _MLX5_SF_TP_ */
+
+/* This part must be outside protection */
+#undef TRACE_INCLUDE_PATH
+#define TRACE_INCLUDE_PATH sf/diag
+#undef TRACE_INCLUDE_FILE
+#define TRACE_INCLUDE_FILE sf_tracepoint
+#include <trace/define_trace.h>
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/sf/diag/vhca_tracepoint.h b/drivers/net/ethernet/mellanox/mlx5/core/sf/diag/vhca_tracepoint.h
new file mode 100644
index 000000000..fd814a190
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/sf/diag/vhca_tracepoint.h
@@ -0,0 +1,40 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. All rights reserved. */
+
+#undef TRACE_SYSTEM
+#define TRACE_SYSTEM mlx5
+
+#if !defined(_MLX5_SF_VHCA_TP_) || defined(TRACE_HEADER_MULTI_READ)
+#define _MLX5_SF_VHCA_TP_
+
+#include <linux/tracepoint.h>
+#include <linux/mlx5/driver.h>
+#include "sf/vhca_event.h"
+
+TRACE_EVENT(mlx5_sf_vhca_event,
+	    TP_PROTO(const struct mlx5_core_dev *dev,
+		     const struct mlx5_vhca_state_event *event),
+	    TP_ARGS(dev, event),
+	    TP_STRUCT__entry(__string(devname, dev_name(dev->device))
+			     __field(u16, hw_fn_id)
+			     __field(u32, sfnum)
+			     __field(u8, vhca_state)
+			    ),
+	    TP_fast_assign(__assign_str(devname, dev_name(dev->device));
+		    __entry->hw_fn_id = event->function_id;
+		    __entry->sfnum = event->sw_function_id;
+		    __entry->vhca_state = event->new_vhca_state;
+	    ),
+	    TP_printk("(%s) hw_id=0x%x sfnum=%u vhca_state=%d\n",
+		      __get_str(devname), __entry->hw_fn_id,
+		      __entry->sfnum, __entry->vhca_state)
+);
+
+#endif /* _MLX5_SF_VHCA_TP_ */
+
+/* This part must be outside protection */
+#undef TRACE_INCLUDE_PATH
+#define TRACE_INCLUDE_PATH sf/diag
+#undef TRACE_INCLUDE_FILE
+#define TRACE_INCLUDE_FILE vhca_tracepoint
+#include <trace/define_trace.h>
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/sf/hw_table.c b/drivers/net/ethernet/mellanox/mlx5/core/sf/hw_table.c
new file mode 100644
index 000000000..17aa34898
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/sf/hw_table.c
@@ -0,0 +1,364 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2020 Mellanox Technologies Ltd */
+#include <linux/mlx5/driver.h>
+#include "vhca_event.h"
+#include "priv.h"
+#include "sf.h"
+#include "mlx5_ifc_vhca_event.h"
+#include "ecpf.h"
+#include "mlx5_core.h"
+#include "eswitch.h"
+#include "diag/sf_tracepoint.h"
+
+struct mlx5_sf_hw {
+	u32 usr_sfnum;
+	u8 allocated: 1;
+	u8 pending_delete: 1;
+};
+
+struct mlx5_sf_hwc_table {
+	struct mlx5_sf_hw *sfs;
+	int max_fn;
+	u16 start_fn_id;
+};
+
+enum mlx5_sf_hwc_index {
+	MLX5_SF_HWC_LOCAL,
+	MLX5_SF_HWC_EXTERNAL,
+	MLX5_SF_HWC_MAX,
+};
+
+struct mlx5_sf_hw_table {
+	struct mlx5_core_dev *dev;
+	struct mutex table_lock; /* Serializes sf deletion and vhca state change handler. */
+	struct notifier_block vhca_nb;
+	struct mlx5_sf_hwc_table hwc[MLX5_SF_HWC_MAX];
+};
+
+static struct mlx5_sf_hwc_table *
+mlx5_sf_controller_to_hwc(struct mlx5_core_dev *dev, u32 controller)
+{
+	int idx = !!controller;
+
+	return &dev->priv.sf_hw_table->hwc[idx];
+}
+
+u16 mlx5_sf_sw_to_hw_id(struct mlx5_core_dev *dev, u32 controller, u16 sw_id)
+{
+	struct mlx5_sf_hwc_table *hwc;
+
+	hwc = mlx5_sf_controller_to_hwc(dev, controller);
+	return hwc->start_fn_id + sw_id;
+}
+
+static u16 mlx5_sf_hw_to_sw_id(struct mlx5_sf_hwc_table *hwc, u16 hw_id)
+{
+	return hw_id - hwc->start_fn_id;
+}
+
+static struct mlx5_sf_hwc_table *
+mlx5_sf_table_fn_to_hwc(struct mlx5_sf_hw_table *table, u16 fn_id)
+{
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(table->hwc); i++) {
+		if (table->hwc[i].max_fn &&
+		    fn_id >= table->hwc[i].start_fn_id &&
+		    fn_id < (table->hwc[i].start_fn_id + table->hwc[i].max_fn))
+			return &table->hwc[i];
+	}
+	return NULL;
+}
+
+static int mlx5_sf_hw_table_id_alloc(struct mlx5_sf_hw_table *table, u32 controller,
+				     u32 usr_sfnum)
+{
+	struct mlx5_sf_hwc_table *hwc;
+	int free_idx = -1;
+	int i;
+
+	hwc = mlx5_sf_controller_to_hwc(table->dev, controller);
+	if (!hwc->sfs)
+		return -ENOSPC;
+
+	for (i = 0; i < hwc->max_fn; i++) {
+		if (!hwc->sfs[i].allocated && free_idx == -1) {
+			free_idx = i;
+			continue;
+		}
+
+		if (hwc->sfs[i].allocated && hwc->sfs[i].usr_sfnum == usr_sfnum)
+			return -EEXIST;
+	}
+
+	if (free_idx == -1)
+		return -ENOSPC;
+
+	hwc->sfs[free_idx].usr_sfnum = usr_sfnum;
+	hwc->sfs[free_idx].allocated = true;
+	return free_idx;
+}
+
+static void mlx5_sf_hw_table_id_free(struct mlx5_sf_hw_table *table, u32 controller, int id)
+{
+	struct mlx5_sf_hwc_table *hwc;
+
+	hwc = mlx5_sf_controller_to_hwc(table->dev, controller);
+	hwc->sfs[id].allocated = false;
+	hwc->sfs[id].pending_delete = false;
+}
+
+int mlx5_sf_hw_table_sf_alloc(struct mlx5_core_dev *dev, u32 controller, u32 usr_sfnum)
+{
+	struct mlx5_sf_hw_table *table = dev->priv.sf_hw_table;
+	u16 hw_fn_id;
+	int sw_id;
+	int err;
+
+	if (!table)
+		return -EOPNOTSUPP;
+
+	mutex_lock(&table->table_lock);
+	sw_id = mlx5_sf_hw_table_id_alloc(table, controller, usr_sfnum);
+	if (sw_id < 0) {
+		err = sw_id;
+		goto exist_err;
+	}
+
+	hw_fn_id = mlx5_sf_sw_to_hw_id(dev, controller, sw_id);
+	err = mlx5_cmd_alloc_sf(dev, hw_fn_id);
+	if (err)
+		goto err;
+
+	err = mlx5_modify_vhca_sw_id(dev, hw_fn_id, usr_sfnum);
+	if (err)
+		goto vhca_err;
+
+	if (controller) {
+		/* If this SF is for external controller, SF manager
+		 * needs to arm firmware to receive the events.
+		 */
+		err = mlx5_vhca_event_arm(dev, hw_fn_id);
+		if (err)
+			goto vhca_err;
+	}
+
+	trace_mlx5_sf_hwc_alloc(dev, controller, hw_fn_id, usr_sfnum);
+	mutex_unlock(&table->table_lock);
+	return sw_id;
+
+vhca_err:
+	mlx5_cmd_dealloc_sf(dev, hw_fn_id);
+err:
+	mlx5_sf_hw_table_id_free(table, controller, sw_id);
+exist_err:
+	mutex_unlock(&table->table_lock);
+	return err;
+}
+
+void mlx5_sf_hw_table_sf_free(struct mlx5_core_dev *dev, u32 controller, u16 id)
+{
+	struct mlx5_sf_hw_table *table = dev->priv.sf_hw_table;
+	u16 hw_fn_id;
+
+	mutex_lock(&table->table_lock);
+	hw_fn_id = mlx5_sf_sw_to_hw_id(dev, controller, id);
+	mlx5_cmd_dealloc_sf(dev, hw_fn_id);
+	mlx5_sf_hw_table_id_free(table, controller, id);
+	mutex_unlock(&table->table_lock);
+}
+
+static void mlx5_sf_hw_table_hwc_sf_free(struct mlx5_core_dev *dev,
+					 struct mlx5_sf_hwc_table *hwc, int idx)
+{
+	mlx5_cmd_dealloc_sf(dev, hwc->start_fn_id + idx);
+	hwc->sfs[idx].allocated = false;
+	hwc->sfs[idx].pending_delete = false;
+	trace_mlx5_sf_hwc_free(dev, hwc->start_fn_id + idx);
+}
+
+void mlx5_sf_hw_table_sf_deferred_free(struct mlx5_core_dev *dev, u32 controller, u16 id)
+{
+	struct mlx5_sf_hw_table *table = dev->priv.sf_hw_table;
+	u32 out[MLX5_ST_SZ_DW(query_vhca_state_out)] = {};
+	struct mlx5_sf_hwc_table *hwc;
+	u16 hw_fn_id;
+	u8 state;
+	int err;
+
+	hw_fn_id = mlx5_sf_sw_to_hw_id(dev, controller, id);
+	hwc = mlx5_sf_controller_to_hwc(dev, controller);
+	mutex_lock(&table->table_lock);
+	err = mlx5_cmd_query_vhca_state(dev, hw_fn_id, out, sizeof(out));
+	if (err)
+		goto err;
+	state = MLX5_GET(query_vhca_state_out, out, vhca_state_context.vhca_state);
+	if (state == MLX5_VHCA_STATE_ALLOCATED) {
+		mlx5_cmd_dealloc_sf(dev, hw_fn_id);
+		hwc->sfs[id].allocated = false;
+	} else {
+		hwc->sfs[id].pending_delete = true;
+		trace_mlx5_sf_hwc_deferred_free(dev, hw_fn_id);
+	}
+err:
+	mutex_unlock(&table->table_lock);
+}
+
+static void mlx5_sf_hw_table_hwc_dealloc_all(struct mlx5_core_dev *dev,
+					     struct mlx5_sf_hwc_table *hwc)
+{
+	int i;
+
+	for (i = 0; i < hwc->max_fn; i++) {
+		if (hwc->sfs[i].allocated)
+			mlx5_sf_hw_table_hwc_sf_free(dev, hwc, i);
+	}
+}
+
+static void mlx5_sf_hw_table_dealloc_all(struct mlx5_sf_hw_table *table)
+{
+	mlx5_sf_hw_table_hwc_dealloc_all(table->dev, &table->hwc[MLX5_SF_HWC_EXTERNAL]);
+	mlx5_sf_hw_table_hwc_dealloc_all(table->dev, &table->hwc[MLX5_SF_HWC_LOCAL]);
+}
+
+static int mlx5_sf_hw_table_hwc_init(struct mlx5_sf_hwc_table *hwc, u16 max_fn, u16 base_id)
+{
+	struct mlx5_sf_hw *sfs;
+
+	if (!max_fn)
+		return 0;
+
+	sfs = kcalloc(max_fn, sizeof(*sfs), GFP_KERNEL);
+	if (!sfs)
+		return -ENOMEM;
+
+	hwc->sfs = sfs;
+	hwc->max_fn = max_fn;
+	hwc->start_fn_id = base_id;
+	return 0;
+}
+
+static void mlx5_sf_hw_table_hwc_cleanup(struct mlx5_sf_hwc_table *hwc)
+{
+	kfree(hwc->sfs);
+}
+
+int mlx5_sf_hw_table_init(struct mlx5_core_dev *dev)
+{
+	struct mlx5_sf_hw_table *table;
+	u16 max_ext_fn = 0;
+	u16 ext_base_id = 0;
+	u16 max_fn = 0;
+	u16 base_id;
+	int err;
+
+	if (!mlx5_vhca_event_supported(dev))
+		return 0;
+
+	if (mlx5_sf_supported(dev))
+		max_fn = mlx5_sf_max_functions(dev);
+
+	err = mlx5_esw_sf_max_hpf_functions(dev, &max_ext_fn, &ext_base_id);
+	if (err)
+		return err;
+
+	if (!max_fn && !max_ext_fn)
+		return 0;
+
+	table = kzalloc(sizeof(*table), GFP_KERNEL);
+	if (!table)
+		return -ENOMEM;
+
+	mutex_init(&table->table_lock);
+	table->dev = dev;
+	dev->priv.sf_hw_table = table;
+
+	base_id = mlx5_sf_start_function_id(dev);
+	err = mlx5_sf_hw_table_hwc_init(&table->hwc[MLX5_SF_HWC_LOCAL], max_fn, base_id);
+	if (err)
+		goto table_err;
+
+	err = mlx5_sf_hw_table_hwc_init(&table->hwc[MLX5_SF_HWC_EXTERNAL],
+					max_ext_fn, ext_base_id);
+	if (err)
+		goto ext_err;
+
+	mlx5_core_dbg(dev, "SF HW table: max sfs = %d, ext sfs = %d\n", max_fn, max_ext_fn);
+	return 0;
+
+ext_err:
+	mlx5_sf_hw_table_hwc_cleanup(&table->hwc[MLX5_SF_HWC_LOCAL]);
+table_err:
+	mutex_destroy(&table->table_lock);
+	kfree(table);
+	return err;
+}
+
+void mlx5_sf_hw_table_cleanup(struct mlx5_core_dev *dev)
+{
+	struct mlx5_sf_hw_table *table = dev->priv.sf_hw_table;
+
+	if (!table)
+		return;
+
+	mutex_destroy(&table->table_lock);
+	mlx5_sf_hw_table_hwc_cleanup(&table->hwc[MLX5_SF_HWC_EXTERNAL]);
+	mlx5_sf_hw_table_hwc_cleanup(&table->hwc[MLX5_SF_HWC_LOCAL]);
+	kfree(table);
+}
+
+static int mlx5_sf_hw_vhca_event(struct notifier_block *nb, unsigned long opcode, void *data)
+{
+	struct mlx5_sf_hw_table *table = container_of(nb, struct mlx5_sf_hw_table, vhca_nb);
+	const struct mlx5_vhca_state_event *event = data;
+	struct mlx5_sf_hwc_table *hwc;
+	struct mlx5_sf_hw *sf_hw;
+	u16 sw_id;
+
+	if (event->new_vhca_state != MLX5_VHCA_STATE_ALLOCATED)
+		return 0;
+
+	hwc = mlx5_sf_table_fn_to_hwc(table, event->function_id);
+	if (!hwc)
+		return 0;
+
+	sw_id = mlx5_sf_hw_to_sw_id(hwc, event->function_id);
+	sf_hw = &hwc->sfs[sw_id];
+
+	mutex_lock(&table->table_lock);
+	/* SF driver notified through firmware that SF is finally detached.
+	 * Hence recycle the sf hardware id for reuse.
+	 */
+	if (sf_hw->allocated && sf_hw->pending_delete)
+		mlx5_sf_hw_table_hwc_sf_free(table->dev, hwc, sw_id);
+	mutex_unlock(&table->table_lock);
+	return 0;
+}
+
+int mlx5_sf_hw_table_create(struct mlx5_core_dev *dev)
+{
+	struct mlx5_sf_hw_table *table = dev->priv.sf_hw_table;
+
+	if (!table)
+		return 0;
+
+	table->vhca_nb.notifier_call = mlx5_sf_hw_vhca_event;
+	return mlx5_vhca_event_notifier_register(dev, &table->vhca_nb);
+}
+
+void mlx5_sf_hw_table_destroy(struct mlx5_core_dev *dev)
+{
+	struct mlx5_sf_hw_table *table = dev->priv.sf_hw_table;
+
+	if (!table)
+		return;
+
+	mlx5_vhca_event_notifier_unregister(dev, &table->vhca_nb);
+	/* Dealloc SFs whose firmware event has been missed. */
+	mlx5_sf_hw_table_dealloc_all(table);
+}
+
+bool mlx5_sf_hw_table_supported(const struct mlx5_core_dev *dev)
+{
+	return !!dev->priv.sf_hw_table;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/sf/mlx5_ifc_vhca_event.h b/drivers/net/ethernet/mellanox/mlx5/core/sf/mlx5_ifc_vhca_event.h
new file mode 100644
index 000000000..4fc870140
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/sf/mlx5_ifc_vhca_event.h
@@ -0,0 +1,82 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2020 Mellanox Technologies Ltd */
+
+#ifndef __MLX5_IFC_VHCA_EVENT_H__
+#define __MLX5_IFC_VHCA_EVENT_H__
+
+enum mlx5_ifc_vhca_state {
+	MLX5_VHCA_STATE_INVALID = 0x0,
+	MLX5_VHCA_STATE_ALLOCATED = 0x1,
+	MLX5_VHCA_STATE_ACTIVE = 0x2,
+	MLX5_VHCA_STATE_IN_USE = 0x3,
+	MLX5_VHCA_STATE_TEARDOWN_REQUEST = 0x4,
+};
+
+struct mlx5_ifc_vhca_state_context_bits {
+	u8         arm_change_event[0x1];
+	u8         reserved_at_1[0xb];
+	u8         vhca_state[0x4];
+	u8         reserved_at_10[0x10];
+
+	u8         sw_function_id[0x20];
+
+	u8         reserved_at_40[0x40];
+};
+
+struct mlx5_ifc_query_vhca_state_out_bits {
+	u8         status[0x8];
+	u8         reserved_at_8[0x18];
+
+	u8         syndrome[0x20];
+
+	u8         reserved_at_40[0x40];
+
+	struct mlx5_ifc_vhca_state_context_bits vhca_state_context;
+};
+
+struct mlx5_ifc_query_vhca_state_in_bits {
+	u8         opcode[0x10];
+	u8         uid[0x10];
+
+	u8         reserved_at_20[0x10];
+	u8         op_mod[0x10];
+
+	u8         embedded_cpu_function[0x1];
+	u8         reserved_at_41[0xf];
+	u8         function_id[0x10];
+
+	u8         reserved_at_60[0x20];
+};
+
+struct mlx5_ifc_vhca_state_field_select_bits {
+	u8         reserved_at_0[0x1e];
+	u8         sw_function_id[0x1];
+	u8         arm_change_event[0x1];
+};
+
+struct mlx5_ifc_modify_vhca_state_out_bits {
+	u8         status[0x8];
+	u8         reserved_at_8[0x18];
+
+	u8         syndrome[0x20];
+
+	u8         reserved_at_40[0x40];
+};
+
+struct mlx5_ifc_modify_vhca_state_in_bits {
+	u8         opcode[0x10];
+	u8         uid[0x10];
+
+	u8         reserved_at_20[0x10];
+	u8         op_mod[0x10];
+
+	u8         embedded_cpu_function[0x1];
+	u8         reserved_at_41[0xf];
+	u8         function_id[0x10];
+
+	struct mlx5_ifc_vhca_state_field_select_bits vhca_state_field_select;
+
+	struct mlx5_ifc_vhca_state_context_bits vhca_state_context;
+};
+
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/sf/priv.h b/drivers/net/ethernet/mellanox/mlx5/core/sf/priv.h
new file mode 100644
index 000000000..7114f3fc3
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/sf/priv.h
@@ -0,0 +1,22 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2020 Mellanox Technologies Ltd */
+
+#ifndef __MLX5_SF_PRIV_H__
+#define __MLX5_SF_PRIV_H__
+
+#include <linux/mlx5/driver.h>
+
+int mlx5_cmd_alloc_sf(struct mlx5_core_dev *dev, u16 function_id);
+int mlx5_cmd_dealloc_sf(struct mlx5_core_dev *dev, u16 function_id);
+
+int mlx5_cmd_sf_enable_hca(struct mlx5_core_dev *dev, u16 func_id);
+int mlx5_cmd_sf_disable_hca(struct mlx5_core_dev *dev, u16 func_id);
+
+u16 mlx5_sf_sw_to_hw_id(struct mlx5_core_dev *dev, u32 controller, u16 sw_id);
+
+int mlx5_sf_hw_table_sf_alloc(struct mlx5_core_dev *dev, u32 controller, u32 usr_sfnum);
+void mlx5_sf_hw_table_sf_free(struct mlx5_core_dev *dev, u32 controller, u16 id);
+void mlx5_sf_hw_table_sf_deferred_free(struct mlx5_core_dev *dev, u32 controller, u16 id);
+bool mlx5_sf_hw_table_supported(const struct mlx5_core_dev *dev);
+
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/sf/sf.h b/drivers/net/ethernet/mellanox/mlx5/core/sf/sf.h
new file mode 100644
index 000000000..3a480e06e
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/sf/sf.h
@@ -0,0 +1,65 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2020 Mellanox Technologies Ltd */
+
+#ifndef __MLX5_SF_H__
+#define __MLX5_SF_H__
+
+#include <linux/mlx5/driver.h>
+#include "lib/sf.h"
+
+#ifdef CONFIG_MLX5_SF_MANAGER
+
+int mlx5_sf_hw_table_init(struct mlx5_core_dev *dev);
+void mlx5_sf_hw_table_cleanup(struct mlx5_core_dev *dev);
+
+int mlx5_sf_hw_table_create(struct mlx5_core_dev *dev);
+void mlx5_sf_hw_table_destroy(struct mlx5_core_dev *dev);
+
+int mlx5_sf_table_init(struct mlx5_core_dev *dev);
+void mlx5_sf_table_cleanup(struct mlx5_core_dev *dev);
+
+int mlx5_devlink_sf_port_new(struct devlink *devlink,
+			     const struct devlink_port_new_attrs *add_attr,
+			     struct netlink_ext_ack *extack,
+			     unsigned int *new_port_index);
+int mlx5_devlink_sf_port_del(struct devlink *devlink, unsigned int port_index,
+			     struct netlink_ext_ack *extack);
+int mlx5_devlink_sf_port_fn_state_get(struct devlink_port *dl_port,
+				      enum devlink_port_fn_state *state,
+				      enum devlink_port_fn_opstate *opstate,
+				      struct netlink_ext_ack *extack);
+int mlx5_devlink_sf_port_fn_state_set(struct devlink_port *dl_port,
+				      enum devlink_port_fn_state state,
+				      struct netlink_ext_ack *extack);
+#else
+
+static inline int mlx5_sf_hw_table_init(struct mlx5_core_dev *dev)
+{
+	return 0;
+}
+
+static inline void mlx5_sf_hw_table_cleanup(struct mlx5_core_dev *dev)
+{
+}
+
+static inline int mlx5_sf_hw_table_create(struct mlx5_core_dev *dev)
+{
+	return 0;
+}
+
+static inline void mlx5_sf_hw_table_destroy(struct mlx5_core_dev *dev)
+{
+}
+
+static inline int mlx5_sf_table_init(struct mlx5_core_dev *dev)
+{
+	return 0;
+}
+
+static inline void mlx5_sf_table_cleanup(struct mlx5_core_dev *dev)
+{
+}
+
+#endif
+
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/sf/vhca_event.c b/drivers/net/ethernet/mellanox/mlx5/core/sf/vhca_event.c
new file mode 100644
index 000000000..d908fba96
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/sf/vhca_event.c
@@ -0,0 +1,191 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2020 Mellanox Technologies Ltd */
+
+#include <linux/mlx5/driver.h>
+#include "mlx5_ifc_vhca_event.h"
+#include "mlx5_core.h"
+#include "vhca_event.h"
+#include "ecpf.h"
+#define CREATE_TRACE_POINTS
+#include "diag/vhca_tracepoint.h"
+
+struct mlx5_vhca_state_notifier {
+	struct mlx5_core_dev *dev;
+	struct mlx5_nb nb;
+	struct blocking_notifier_head n_head;
+};
+
+struct mlx5_vhca_event_work {
+	struct work_struct work;
+	struct mlx5_vhca_state_notifier *notifier;
+	struct mlx5_vhca_state_event event;
+};
+
+int mlx5_cmd_query_vhca_state(struct mlx5_core_dev *dev, u16 function_id, u32 *out, u32 outlen)
+{
+	u32 in[MLX5_ST_SZ_DW(query_vhca_state_in)] = {};
+
+	MLX5_SET(query_vhca_state_in, in, opcode, MLX5_CMD_OP_QUERY_VHCA_STATE);
+	MLX5_SET(query_vhca_state_in, in, function_id, function_id);
+	MLX5_SET(query_vhca_state_in, in, embedded_cpu_function, 0);
+
+	return mlx5_cmd_exec(dev, in, sizeof(in), out, outlen);
+}
+
+static int mlx5_cmd_modify_vhca_state(struct mlx5_core_dev *dev, u16 function_id,
+				      u32 *in, u32 inlen)
+{
+	u32 out[MLX5_ST_SZ_DW(modify_vhca_state_out)] = {};
+
+	MLX5_SET(modify_vhca_state_in, in, opcode, MLX5_CMD_OP_MODIFY_VHCA_STATE);
+	MLX5_SET(modify_vhca_state_in, in, function_id, function_id);
+	MLX5_SET(modify_vhca_state_in, in, embedded_cpu_function, 0);
+
+	return mlx5_cmd_exec(dev, in, inlen, out, sizeof(out));
+}
+
+int mlx5_modify_vhca_sw_id(struct mlx5_core_dev *dev, u16 function_id, u32 sw_fn_id)
+{
+	u32 out[MLX5_ST_SZ_DW(modify_vhca_state_out)] = {};
+	u32 in[MLX5_ST_SZ_DW(modify_vhca_state_in)] = {};
+
+	MLX5_SET(modify_vhca_state_in, in, opcode, MLX5_CMD_OP_MODIFY_VHCA_STATE);
+	MLX5_SET(modify_vhca_state_in, in, function_id, function_id);
+	MLX5_SET(modify_vhca_state_in, in, embedded_cpu_function, 0);
+	MLX5_SET(modify_vhca_state_in, in, vhca_state_field_select.sw_function_id, 1);
+	MLX5_SET(modify_vhca_state_in, in, vhca_state_context.sw_function_id, sw_fn_id);
+
+	return mlx5_cmd_exec_inout(dev, modify_vhca_state, in, out);
+}
+
+int mlx5_vhca_event_arm(struct mlx5_core_dev *dev, u16 function_id)
+{
+	u32 in[MLX5_ST_SZ_DW(modify_vhca_state_in)] = {};
+
+	MLX5_SET(modify_vhca_state_in, in, vhca_state_context.arm_change_event, 1);
+	MLX5_SET(modify_vhca_state_in, in, vhca_state_field_select.arm_change_event, 1);
+
+	return mlx5_cmd_modify_vhca_state(dev, function_id, in, sizeof(in));
+}
+
+static void
+mlx5_vhca_event_notify(struct mlx5_core_dev *dev, struct mlx5_vhca_state_event *event)
+{
+	u32 out[MLX5_ST_SZ_DW(query_vhca_state_out)] = {};
+	int err;
+
+	err = mlx5_cmd_query_vhca_state(dev, event->function_id, out, sizeof(out));
+	if (err)
+		return;
+
+	event->sw_function_id = MLX5_GET(query_vhca_state_out, out,
+					 vhca_state_context.sw_function_id);
+	event->new_vhca_state = MLX5_GET(query_vhca_state_out, out,
+					 vhca_state_context.vhca_state);
+
+	mlx5_vhca_event_arm(dev, event->function_id);
+	trace_mlx5_sf_vhca_event(dev, event);
+
+	blocking_notifier_call_chain(&dev->priv.vhca_state_notifier->n_head, 0, event);
+}
+
+static void mlx5_vhca_state_work_handler(struct work_struct *_work)
+{
+	struct mlx5_vhca_event_work *work = container_of(_work, struct mlx5_vhca_event_work, work);
+	struct mlx5_vhca_state_notifier *notifier = work->notifier;
+	struct mlx5_core_dev *dev = notifier->dev;
+
+	mlx5_vhca_event_notify(dev, &work->event);
+	kfree(work);
+}
+
+static int
+mlx5_vhca_state_change_notifier(struct notifier_block *nb, unsigned long type, void *data)
+{
+	struct mlx5_vhca_state_notifier *notifier =
+				mlx5_nb_cof(nb, struct mlx5_vhca_state_notifier, nb);
+	struct mlx5_vhca_event_work *work;
+	struct mlx5_eqe *eqe = data;
+
+	work = kzalloc(sizeof(*work), GFP_ATOMIC);
+	if (!work)
+		return NOTIFY_DONE;
+	INIT_WORK(&work->work, &mlx5_vhca_state_work_handler);
+	work->notifier = notifier;
+	work->event.function_id = be16_to_cpu(eqe->data.vhca_state.function_id);
+	mlx5_events_work_enqueue(notifier->dev, &work->work);
+	return NOTIFY_OK;
+}
+
+void mlx5_vhca_state_cap_handle(struct mlx5_core_dev *dev, void *set_hca_cap)
+{
+	if (!mlx5_vhca_event_supported(dev))
+		return;
+
+	MLX5_SET(cmd_hca_cap, set_hca_cap, vhca_state, 1);
+	MLX5_SET(cmd_hca_cap, set_hca_cap, event_on_vhca_state_allocated, 1);
+	MLX5_SET(cmd_hca_cap, set_hca_cap, event_on_vhca_state_active, 1);
+	MLX5_SET(cmd_hca_cap, set_hca_cap, event_on_vhca_state_in_use, 1);
+	MLX5_SET(cmd_hca_cap, set_hca_cap, event_on_vhca_state_teardown_request, 1);
+}
+
+int mlx5_vhca_event_init(struct mlx5_core_dev *dev)
+{
+	struct mlx5_vhca_state_notifier *notifier;
+
+	if (!mlx5_vhca_event_supported(dev))
+		return 0;
+
+	notifier = kzalloc(sizeof(*notifier), GFP_KERNEL);
+	if (!notifier)
+		return -ENOMEM;
+
+	dev->priv.vhca_state_notifier = notifier;
+	notifier->dev = dev;
+	BLOCKING_INIT_NOTIFIER_HEAD(&notifier->n_head);
+	MLX5_NB_INIT(&notifier->nb, mlx5_vhca_state_change_notifier, VHCA_STATE_CHANGE);
+	return 0;
+}
+
+void mlx5_vhca_event_cleanup(struct mlx5_core_dev *dev)
+{
+	if (!mlx5_vhca_event_supported(dev))
+		return;
+
+	kfree(dev->priv.vhca_state_notifier);
+	dev->priv.vhca_state_notifier = NULL;
+}
+
+void mlx5_vhca_event_start(struct mlx5_core_dev *dev)
+{
+	struct mlx5_vhca_state_notifier *notifier;
+
+	if (!dev->priv.vhca_state_notifier)
+		return;
+
+	notifier = dev->priv.vhca_state_notifier;
+	mlx5_eq_notifier_register(dev, &notifier->nb);
+}
+
+void mlx5_vhca_event_stop(struct mlx5_core_dev *dev)
+{
+	struct mlx5_vhca_state_notifier *notifier;
+
+	if (!dev->priv.vhca_state_notifier)
+		return;
+
+	notifier = dev->priv.vhca_state_notifier;
+	mlx5_eq_notifier_unregister(dev, &notifier->nb);
+}
+
+int mlx5_vhca_event_notifier_register(struct mlx5_core_dev *dev, struct notifier_block *nb)
+{
+	if (!dev->priv.vhca_state_notifier)
+		return -EOPNOTSUPP;
+	return blocking_notifier_chain_register(&dev->priv.vhca_state_notifier->n_head, nb);
+}
+
+void mlx5_vhca_event_notifier_unregister(struct mlx5_core_dev *dev, struct notifier_block *nb)
+{
+	blocking_notifier_chain_unregister(&dev->priv.vhca_state_notifier->n_head, nb);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/sf/vhca_event.h b/drivers/net/ethernet/mellanox/mlx5/core/sf/vhca_event.h
new file mode 100644
index 000000000..013cdfe90
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/sf/vhca_event.h
@@ -0,0 +1,56 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2020 Mellanox Technologies Ltd */
+
+#ifndef __MLX5_VHCA_EVENT_H__
+#define __MLX5_VHCA_EVENT_H__
+
+#ifdef CONFIG_MLX5_SF
+
+struct mlx5_vhca_state_event {
+	u16 function_id;
+	u16 sw_function_id;
+	u8 new_vhca_state;
+};
+
+static inline bool mlx5_vhca_event_supported(const struct mlx5_core_dev *dev)
+{
+	return MLX5_CAP_GEN_MAX(dev, vhca_state);
+}
+
+void mlx5_vhca_state_cap_handle(struct mlx5_core_dev *dev, void *set_hca_cap);
+int mlx5_vhca_event_init(struct mlx5_core_dev *dev);
+void mlx5_vhca_event_cleanup(struct mlx5_core_dev *dev);
+void mlx5_vhca_event_start(struct mlx5_core_dev *dev);
+void mlx5_vhca_event_stop(struct mlx5_core_dev *dev);
+int mlx5_vhca_event_notifier_register(struct mlx5_core_dev *dev, struct notifier_block *nb);
+void mlx5_vhca_event_notifier_unregister(struct mlx5_core_dev *dev, struct notifier_block *nb);
+int mlx5_modify_vhca_sw_id(struct mlx5_core_dev *dev, u16 function_id, u32 sw_fn_id);
+int mlx5_vhca_event_arm(struct mlx5_core_dev *dev, u16 function_id);
+int mlx5_cmd_query_vhca_state(struct mlx5_core_dev *dev, u16 function_id,
+			      u32 *out, u32 outlen);
+#else
+
+static inline void mlx5_vhca_state_cap_handle(struct mlx5_core_dev *dev, void *set_hca_cap)
+{
+}
+
+static inline int mlx5_vhca_event_init(struct mlx5_core_dev *dev)
+{
+	return 0;
+}
+
+static inline void mlx5_vhca_event_cleanup(struct mlx5_core_dev *dev)
+{
+}
+
+static inline void mlx5_vhca_event_start(struct mlx5_core_dev *dev)
+{
+}
+
+static inline void mlx5_vhca_event_stop(struct mlx5_core_dev *dev)
+{
+}
+
+#endif
+
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/sriov.c b/drivers/net/ethernet/mellanox/mlx5/core/sriov.c
new file mode 100644
index 000000000..5f2195e65
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/sriov.c
@@ -0,0 +1,354 @@
+/*
+ * Copyright (c) 2014, Mellanox Technologies inc.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/pci.h>
+#include <linux/mlx5/driver.h>
+#include <linux/mlx5/vport.h>
+#include "mlx5_core.h"
+#include "mlx5_irq.h"
+#include "eswitch.h"
+
+static int sriov_restore_guids(struct mlx5_core_dev *dev, int vf)
+{
+	struct mlx5_core_sriov *sriov = &dev->priv.sriov;
+	struct mlx5_hca_vport_context *in;
+	int err = 0;
+
+	/* Restore sriov guid and policy settings */
+	if (sriov->vfs_ctx[vf].node_guid ||
+	    sriov->vfs_ctx[vf].port_guid ||
+	    sriov->vfs_ctx[vf].policy != MLX5_POLICY_INVALID) {
+		in = kzalloc(sizeof(*in), GFP_KERNEL);
+		if (!in)
+			return -ENOMEM;
+
+		in->node_guid = sriov->vfs_ctx[vf].node_guid;
+		in->port_guid = sriov->vfs_ctx[vf].port_guid;
+		in->policy = sriov->vfs_ctx[vf].policy;
+		in->field_select =
+			!!(in->port_guid) * MLX5_HCA_VPORT_SEL_PORT_GUID |
+			!!(in->node_guid) * MLX5_HCA_VPORT_SEL_NODE_GUID |
+			!!(in->policy) * MLX5_HCA_VPORT_SEL_STATE_POLICY;
+
+		err = mlx5_core_modify_hca_vport_context(dev, 1, 1, vf + 1, in);
+		if (err)
+			mlx5_core_warn(dev, "modify vport context failed, unable to restore VF %d settings\n", vf);
+
+		kfree(in);
+	}
+
+	return err;
+}
+
+static int mlx5_device_enable_sriov(struct mlx5_core_dev *dev, int num_vfs)
+{
+	struct mlx5_core_sriov *sriov = &dev->priv.sriov;
+	int err, vf, num_msix_count;
+
+	if (!MLX5_ESWITCH_MANAGER(dev))
+		goto enable_vfs_hca;
+
+	err = mlx5_eswitch_enable(dev->priv.eswitch, num_vfs);
+	if (err) {
+		mlx5_core_warn(dev,
+			       "failed to enable eswitch SRIOV (%d)\n", err);
+		return err;
+	}
+
+enable_vfs_hca:
+	num_msix_count = mlx5_get_default_msix_vec_count(dev, num_vfs);
+	for (vf = 0; vf < num_vfs; vf++) {
+		/* Notify the VF before its enablement to let it set
+		 * some stuff.
+		 */
+		blocking_notifier_call_chain(&sriov->vfs_ctx[vf].notifier,
+					     MLX5_PF_NOTIFY_ENABLE_VF, dev);
+		err = mlx5_core_enable_hca(dev, vf + 1);
+		if (err) {
+			mlx5_core_warn(dev, "failed to enable VF %d (%d)\n", vf, err);
+			continue;
+		}
+
+		err = mlx5_set_msix_vec_count(dev, vf + 1, num_msix_count);
+		if (err) {
+			mlx5_core_warn(dev,
+				       "failed to set MSI-X vector counts VF %d, err %d\n",
+				       vf, err);
+			continue;
+		}
+
+		sriov->vfs_ctx[vf].enabled = 1;
+		if (MLX5_CAP_GEN(dev, port_type) == MLX5_CAP_PORT_TYPE_IB) {
+			err = sriov_restore_guids(dev, vf);
+			if (err) {
+				mlx5_core_warn(dev,
+					       "failed to restore VF %d settings, err %d\n",
+					       vf, err);
+				continue;
+			}
+		}
+		mlx5_core_dbg(dev, "successfully enabled VF* %d\n", vf);
+	}
+
+	return 0;
+}
+
+static void
+mlx5_device_disable_sriov(struct mlx5_core_dev *dev, int num_vfs, bool clear_vf)
+{
+	struct mlx5_core_sriov *sriov = &dev->priv.sriov;
+	int err;
+	int vf;
+
+	for (vf = num_vfs - 1; vf >= 0; vf--) {
+		if (!sriov->vfs_ctx[vf].enabled)
+			continue;
+		/* Notify the VF before its disablement to let it clean
+		 * some resources.
+		 */
+		blocking_notifier_call_chain(&sriov->vfs_ctx[vf].notifier,
+					     MLX5_PF_NOTIFY_DISABLE_VF, dev);
+		err = mlx5_core_disable_hca(dev, vf + 1);
+		if (err) {
+			mlx5_core_warn(dev, "failed to disable VF %d\n", vf);
+			continue;
+		}
+		sriov->vfs_ctx[vf].enabled = 0;
+	}
+
+	mlx5_eswitch_disable_sriov(dev->priv.eswitch, clear_vf);
+
+	/* For ECPFs, skip waiting for host VF pages until ECPF is destroyed */
+	if (mlx5_core_is_ecpf(dev))
+		return;
+
+	if (mlx5_wait_for_pages(dev, &dev->priv.page_counters[MLX5_VF]))
+		mlx5_core_warn(dev, "timeout reclaiming VFs pages\n");
+}
+
+static int mlx5_sriov_enable(struct pci_dev *pdev, int num_vfs)
+{
+	struct mlx5_core_dev *dev  = pci_get_drvdata(pdev);
+	struct devlink *devlink = priv_to_devlink(dev);
+	int err;
+
+	devl_lock(devlink);
+	err = mlx5_device_enable_sriov(dev, num_vfs);
+	devl_unlock(devlink);
+	if (err) {
+		mlx5_core_warn(dev, "mlx5_device_enable_sriov failed : %d\n", err);
+		return err;
+	}
+
+	err = pci_enable_sriov(pdev, num_vfs);
+	if (err) {
+		mlx5_core_warn(dev, "pci_enable_sriov failed : %d\n", err);
+		mlx5_device_disable_sriov(dev, num_vfs, true);
+	}
+	return err;
+}
+
+void mlx5_sriov_disable(struct pci_dev *pdev)
+{
+	struct mlx5_core_dev *dev  = pci_get_drvdata(pdev);
+	struct devlink *devlink = priv_to_devlink(dev);
+	int num_vfs = pci_num_vf(dev->pdev);
+
+	pci_disable_sriov(pdev);
+	devl_lock(devlink);
+	mlx5_device_disable_sriov(dev, num_vfs, true);
+	devl_unlock(devlink);
+}
+
+int mlx5_core_sriov_configure(struct pci_dev *pdev, int num_vfs)
+{
+	struct mlx5_core_dev *dev  = pci_get_drvdata(pdev);
+	struct mlx5_core_sriov *sriov = &dev->priv.sriov;
+	int err = 0;
+
+	mlx5_core_dbg(dev, "requested num_vfs %d\n", num_vfs);
+
+	if (num_vfs)
+		err = mlx5_sriov_enable(pdev, num_vfs);
+	else
+		mlx5_sriov_disable(pdev);
+
+	if (!err)
+		sriov->num_vfs = num_vfs;
+	return err ? err : num_vfs;
+}
+
+int mlx5_core_sriov_set_msix_vec_count(struct pci_dev *vf, int msix_vec_count)
+{
+	struct pci_dev *pf = pci_physfn(vf);
+	struct mlx5_core_sriov *sriov;
+	struct mlx5_core_dev *dev;
+	int num_vf_msix, id;
+
+	dev = pci_get_drvdata(pf);
+	num_vf_msix = MLX5_CAP_GEN_MAX(dev, num_total_dynamic_vf_msix);
+	if (!num_vf_msix)
+		return -EOPNOTSUPP;
+
+	if (!msix_vec_count)
+		msix_vec_count =
+			mlx5_get_default_msix_vec_count(dev, pci_num_vf(pf));
+
+	sriov = &dev->priv.sriov;
+	id = pci_iov_vf_id(vf);
+	if (id < 0 || !sriov->vfs_ctx[id].enabled)
+		return -EINVAL;
+
+	return mlx5_set_msix_vec_count(dev, id + 1, msix_vec_count);
+}
+
+int mlx5_sriov_attach(struct mlx5_core_dev *dev)
+{
+	if (!mlx5_core_is_pf(dev) || !pci_num_vf(dev->pdev))
+		return 0;
+
+	/* If sriov VFs exist in PCI level, enable them in device level */
+	return mlx5_device_enable_sriov(dev, pci_num_vf(dev->pdev));
+}
+
+void mlx5_sriov_detach(struct mlx5_core_dev *dev)
+{
+	if (!mlx5_core_is_pf(dev))
+		return;
+
+	mlx5_device_disable_sriov(dev, pci_num_vf(dev->pdev), false);
+}
+
+static u16 mlx5_get_max_vfs(struct mlx5_core_dev *dev)
+{
+	u16 host_total_vfs;
+	const u32 *out;
+
+	if (mlx5_core_is_ecpf_esw_manager(dev)) {
+		out = mlx5_esw_query_functions(dev);
+
+		/* Old FW doesn't support getting total_vfs from esw func
+		 * but supports getting it from pci_sriov.
+		 */
+		if (IS_ERR(out))
+			goto done;
+		host_total_vfs = MLX5_GET(query_esw_functions_out, out,
+					  host_params_context.host_total_vfs);
+		kvfree(out);
+		return host_total_vfs;
+	}
+
+done:
+	return pci_sriov_get_totalvfs(dev->pdev);
+}
+
+int mlx5_sriov_init(struct mlx5_core_dev *dev)
+{
+	struct mlx5_core_sriov *sriov = &dev->priv.sriov;
+	struct pci_dev *pdev = dev->pdev;
+	int total_vfs, i;
+
+	if (!mlx5_core_is_pf(dev))
+		return 0;
+
+	total_vfs = pci_sriov_get_totalvfs(pdev);
+	sriov->max_vfs = mlx5_get_max_vfs(dev);
+	sriov->num_vfs = pci_num_vf(pdev);
+	sriov->vfs_ctx = kcalloc(total_vfs, sizeof(*sriov->vfs_ctx), GFP_KERNEL);
+	if (!sriov->vfs_ctx)
+		return -ENOMEM;
+
+	for (i = 0; i < total_vfs; i++)
+		BLOCKING_INIT_NOTIFIER_HEAD(&sriov->vfs_ctx[i].notifier);
+
+	return 0;
+}
+
+void mlx5_sriov_cleanup(struct mlx5_core_dev *dev)
+{
+	struct mlx5_core_sriov *sriov = &dev->priv.sriov;
+
+	if (!mlx5_core_is_pf(dev))
+		return;
+
+	kfree(sriov->vfs_ctx);
+}
+
+/**
+ * mlx5_sriov_blocking_notifier_unregister - Unregister a VF from
+ * a notification block chain.
+ *
+ * @mdev: The mlx5 core device.
+ * @vf_id: The VF id.
+ * @nb: The notifier block to be unregistered.
+ */
+void mlx5_sriov_blocking_notifier_unregister(struct mlx5_core_dev *mdev,
+					     int vf_id,
+					     struct notifier_block *nb)
+{
+	struct mlx5_vf_context *vfs_ctx;
+	struct mlx5_core_sriov *sriov;
+
+	sriov = &mdev->priv.sriov;
+	if (WARN_ON(vf_id < 0 || vf_id >= sriov->num_vfs))
+		return;
+
+	vfs_ctx = &sriov->vfs_ctx[vf_id];
+	blocking_notifier_chain_unregister(&vfs_ctx->notifier, nb);
+}
+EXPORT_SYMBOL(mlx5_sriov_blocking_notifier_unregister);
+
+/**
+ * mlx5_sriov_blocking_notifier_register - Register a VF notification
+ * block chain.
+ *
+ * @mdev: The mlx5 core device.
+ * @vf_id: The VF id.
+ * @nb: The notifier block to be called upon the VF events.
+ *
+ * Returns 0 on success or an error code.
+ */
+int mlx5_sriov_blocking_notifier_register(struct mlx5_core_dev *mdev,
+					  int vf_id,
+					  struct notifier_block *nb)
+{
+	struct mlx5_vf_context *vfs_ctx;
+	struct mlx5_core_sriov *sriov;
+
+	sriov = &mdev->priv.sriov;
+	if (vf_id < 0 || vf_id >= sriov->num_vfs)
+		return -EINVAL;
+
+	vfs_ctx = &sriov->vfs_ctx[vf_id];
+	return blocking_notifier_chain_register(&vfs_ctx->notifier, nb);
+}
+EXPORT_SYMBOL(mlx5_sriov_blocking_notifier_register);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/Makefile b/drivers/net/ethernet/mellanox/mlx5/core/steering/Makefile
new file mode 100644
index 000000000..c78512eed
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/Makefile
@@ -0,0 +1,2 @@
+# SPDX-License-Identifier: GPL-2.0-only
+subdir-ccflags-y += -I$(src)/..
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_action.c b/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_action.c
new file mode 100644
index 000000000..bf7517725
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_action.c
@@ -0,0 +1,2003 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2019 Mellanox Technologies. */
+
+#include "dr_types.h"
+#include "dr_ste.h"
+
+enum dr_action_domain {
+	DR_ACTION_DOMAIN_NIC_INGRESS,
+	DR_ACTION_DOMAIN_NIC_EGRESS,
+	DR_ACTION_DOMAIN_FDB_INGRESS,
+	DR_ACTION_DOMAIN_FDB_EGRESS,
+	DR_ACTION_DOMAIN_MAX,
+};
+
+enum dr_action_valid_state {
+	DR_ACTION_STATE_ERR,
+	DR_ACTION_STATE_NO_ACTION,
+	DR_ACTION_STATE_ENCAP,
+	DR_ACTION_STATE_DECAP,
+	DR_ACTION_STATE_MODIFY_HDR,
+	DR_ACTION_STATE_POP_VLAN,
+	DR_ACTION_STATE_PUSH_VLAN,
+	DR_ACTION_STATE_NON_TERM,
+	DR_ACTION_STATE_TERM,
+	DR_ACTION_STATE_ASO,
+	DR_ACTION_STATE_MAX,
+};
+
+static const char * const action_type_to_str[] = {
+	[DR_ACTION_TYP_TNL_L2_TO_L2] = "DR_ACTION_TYP_TNL_L2_TO_L2",
+	[DR_ACTION_TYP_L2_TO_TNL_L2] = "DR_ACTION_TYP_L2_TO_TNL_L2",
+	[DR_ACTION_TYP_TNL_L3_TO_L2] = "DR_ACTION_TYP_TNL_L3_TO_L2",
+	[DR_ACTION_TYP_L2_TO_TNL_L3] = "DR_ACTION_TYP_L2_TO_TNL_L3",
+	[DR_ACTION_TYP_DROP] = "DR_ACTION_TYP_DROP",
+	[DR_ACTION_TYP_QP] = "DR_ACTION_TYP_QP",
+	[DR_ACTION_TYP_FT] = "DR_ACTION_TYP_FT",
+	[DR_ACTION_TYP_CTR] = "DR_ACTION_TYP_CTR",
+	[DR_ACTION_TYP_TAG] = "DR_ACTION_TYP_TAG",
+	[DR_ACTION_TYP_MODIFY_HDR] = "DR_ACTION_TYP_MODIFY_HDR",
+	[DR_ACTION_TYP_VPORT] = "DR_ACTION_TYP_VPORT",
+	[DR_ACTION_TYP_POP_VLAN] = "DR_ACTION_TYP_POP_VLAN",
+	[DR_ACTION_TYP_PUSH_VLAN] = "DR_ACTION_TYP_PUSH_VLAN",
+	[DR_ACTION_TYP_SAMPLER] = "DR_ACTION_TYP_SAMPLER",
+	[DR_ACTION_TYP_INSERT_HDR] = "DR_ACTION_TYP_INSERT_HDR",
+	[DR_ACTION_TYP_REMOVE_HDR] = "DR_ACTION_TYP_REMOVE_HDR",
+	[DR_ACTION_TYP_ASO_FLOW_METER] = "DR_ACTION_TYP_ASO_FLOW_METER",
+	[DR_ACTION_TYP_MAX] = "DR_ACTION_UNKNOWN",
+};
+
+static const char *dr_action_id_to_str(enum mlx5dr_action_type action_id)
+{
+	if (action_id > DR_ACTION_TYP_MAX)
+		action_id = DR_ACTION_TYP_MAX;
+	return action_type_to_str[action_id];
+}
+
+static const enum dr_action_valid_state
+next_action_state[DR_ACTION_DOMAIN_MAX][DR_ACTION_STATE_MAX][DR_ACTION_TYP_MAX] = {
+	[DR_ACTION_DOMAIN_NIC_INGRESS] = {
+		[DR_ACTION_STATE_NO_ACTION] = {
+			[DR_ACTION_TYP_DROP]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_QP]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_FT]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_SAMPLER]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_TAG]		= DR_ACTION_STATE_NON_TERM,
+			[DR_ACTION_TYP_CTR]		= DR_ACTION_STATE_NON_TERM,
+			[DR_ACTION_TYP_TNL_L2_TO_L2]	= DR_ACTION_STATE_DECAP,
+			[DR_ACTION_TYP_TNL_L3_TO_L2]	= DR_ACTION_STATE_DECAP,
+			[DR_ACTION_TYP_L2_TO_TNL_L2]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_L2_TO_TNL_L3]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_INSERT_HDR]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_REMOVE_HDR]	= DR_ACTION_STATE_DECAP,
+			[DR_ACTION_TYP_MODIFY_HDR]	= DR_ACTION_STATE_MODIFY_HDR,
+			[DR_ACTION_TYP_POP_VLAN]	= DR_ACTION_STATE_POP_VLAN,
+			[DR_ACTION_TYP_PUSH_VLAN]	= DR_ACTION_STATE_PUSH_VLAN,
+			[DR_ACTION_TYP_ASO_FLOW_METER]	= DR_ACTION_STATE_ASO,
+		},
+		[DR_ACTION_STATE_DECAP] = {
+			[DR_ACTION_TYP_DROP]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_QP]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_FT]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_SAMPLER]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_TAG]		= DR_ACTION_STATE_DECAP,
+			[DR_ACTION_TYP_CTR]		= DR_ACTION_STATE_DECAP,
+			[DR_ACTION_TYP_L2_TO_TNL_L2]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_L2_TO_TNL_L3]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_INSERT_HDR]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_MODIFY_HDR]	= DR_ACTION_STATE_MODIFY_HDR,
+			[DR_ACTION_TYP_POP_VLAN]	= DR_ACTION_STATE_POP_VLAN,
+			[DR_ACTION_TYP_PUSH_VLAN]	= DR_ACTION_STATE_PUSH_VLAN,
+			[DR_ACTION_TYP_ASO_FLOW_METER]	= DR_ACTION_STATE_ASO,
+		},
+		[DR_ACTION_STATE_ENCAP] = {
+			[DR_ACTION_TYP_DROP]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_QP]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_FT]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_SAMPLER]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_TAG]		= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_CTR]		= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_ASO_FLOW_METER]	= DR_ACTION_STATE_ASO,
+		},
+		[DR_ACTION_STATE_MODIFY_HDR] = {
+			[DR_ACTION_TYP_DROP]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_QP]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_FT]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_SAMPLER]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_TAG]		= DR_ACTION_STATE_MODIFY_HDR,
+			[DR_ACTION_TYP_CTR]		= DR_ACTION_STATE_MODIFY_HDR,
+			[DR_ACTION_TYP_L2_TO_TNL_L2]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_L2_TO_TNL_L3]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_INSERT_HDR]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_PUSH_VLAN]	= DR_ACTION_STATE_PUSH_VLAN,
+			[DR_ACTION_TYP_ASO_FLOW_METER]	= DR_ACTION_STATE_ASO,
+		},
+		[DR_ACTION_STATE_POP_VLAN] = {
+			[DR_ACTION_TYP_DROP]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_QP]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_FT]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_SAMPLER]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_TAG]		= DR_ACTION_STATE_POP_VLAN,
+			[DR_ACTION_TYP_CTR]		= DR_ACTION_STATE_POP_VLAN,
+			[DR_ACTION_TYP_POP_VLAN]	= DR_ACTION_STATE_POP_VLAN,
+			[DR_ACTION_TYP_MODIFY_HDR]	= DR_ACTION_STATE_MODIFY_HDR,
+			[DR_ACTION_TYP_L2_TO_TNL_L2]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_L2_TO_TNL_L3]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_INSERT_HDR]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_ASO_FLOW_METER]	= DR_ACTION_STATE_ASO,
+		},
+		[DR_ACTION_STATE_PUSH_VLAN] = {
+			[DR_ACTION_TYP_QP]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_FT]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_SAMPLER]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_TAG]		= DR_ACTION_STATE_PUSH_VLAN,
+			[DR_ACTION_TYP_CTR]		= DR_ACTION_STATE_PUSH_VLAN,
+			[DR_ACTION_TYP_PUSH_VLAN]	= DR_ACTION_STATE_PUSH_VLAN,
+			[DR_ACTION_TYP_L2_TO_TNL_L2]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_L2_TO_TNL_L3]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_ASO_FLOW_METER]	= DR_ACTION_STATE_ASO,
+		},
+		[DR_ACTION_STATE_NON_TERM] = {
+			[DR_ACTION_TYP_DROP]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_QP]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_FT]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_SAMPLER]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_TAG]		= DR_ACTION_STATE_NON_TERM,
+			[DR_ACTION_TYP_CTR]		= DR_ACTION_STATE_NON_TERM,
+			[DR_ACTION_TYP_TNL_L2_TO_L2]	= DR_ACTION_STATE_DECAP,
+			[DR_ACTION_TYP_TNL_L3_TO_L2]	= DR_ACTION_STATE_DECAP,
+			[DR_ACTION_TYP_L2_TO_TNL_L2]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_L2_TO_TNL_L3]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_INSERT_HDR]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_REMOVE_HDR]	= DR_ACTION_STATE_DECAP,
+			[DR_ACTION_TYP_MODIFY_HDR]	= DR_ACTION_STATE_MODIFY_HDR,
+			[DR_ACTION_TYP_POP_VLAN]	= DR_ACTION_STATE_POP_VLAN,
+			[DR_ACTION_TYP_PUSH_VLAN]	= DR_ACTION_STATE_PUSH_VLAN,
+			[DR_ACTION_TYP_ASO_FLOW_METER]	= DR_ACTION_STATE_ASO,
+		},
+		[DR_ACTION_STATE_ASO] = {
+			[DR_ACTION_TYP_DROP]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_QP]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_FT]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_CTR]             = DR_ACTION_STATE_ASO,
+		},
+		[DR_ACTION_STATE_TERM] = {
+			[DR_ACTION_TYP_CTR]		= DR_ACTION_STATE_TERM,
+		},
+	},
+	[DR_ACTION_DOMAIN_NIC_EGRESS] = {
+		[DR_ACTION_STATE_NO_ACTION] = {
+			[DR_ACTION_TYP_DROP]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_FT]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_SAMPLER]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_CTR]		= DR_ACTION_STATE_NON_TERM,
+			[DR_ACTION_TYP_L2_TO_TNL_L2]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_L2_TO_TNL_L3]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_INSERT_HDR]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_REMOVE_HDR]	= DR_ACTION_STATE_DECAP,
+			[DR_ACTION_TYP_MODIFY_HDR]	= DR_ACTION_STATE_MODIFY_HDR,
+			[DR_ACTION_TYP_PUSH_VLAN]	= DR_ACTION_STATE_PUSH_VLAN,
+			[DR_ACTION_TYP_POP_VLAN]	= DR_ACTION_STATE_POP_VLAN,
+			[DR_ACTION_TYP_ASO_FLOW_METER]	= DR_ACTION_STATE_ASO,
+		},
+		[DR_ACTION_STATE_DECAP] = {
+			[DR_ACTION_TYP_DROP]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_FT]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_SAMPLER]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_CTR]		= DR_ACTION_STATE_DECAP,
+			[DR_ACTION_TYP_ASO_FLOW_METER]	= DR_ACTION_STATE_ASO,
+		},
+		[DR_ACTION_STATE_ENCAP] = {
+			[DR_ACTION_TYP_DROP]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_FT]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_SAMPLER]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_CTR]		= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_ASO_FLOW_METER]	= DR_ACTION_STATE_ASO,
+		},
+		[DR_ACTION_STATE_MODIFY_HDR] = {
+			[DR_ACTION_TYP_DROP]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_FT]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_SAMPLER]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_CTR]		= DR_ACTION_STATE_MODIFY_HDR,
+			[DR_ACTION_TYP_L2_TO_TNL_L2]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_L2_TO_TNL_L3]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_INSERT_HDR]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_PUSH_VLAN]	= DR_ACTION_STATE_PUSH_VLAN,
+			[DR_ACTION_TYP_ASO_FLOW_METER]	= DR_ACTION_STATE_ASO,
+		},
+		[DR_ACTION_STATE_POP_VLAN] = {
+			[DR_ACTION_TYP_FT]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_SAMPLER]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_CTR]		= DR_ACTION_STATE_POP_VLAN,
+			[DR_ACTION_TYP_POP_VLAN]	= DR_ACTION_STATE_POP_VLAN,
+			[DR_ACTION_TYP_PUSH_VLAN]	= DR_ACTION_STATE_PUSH_VLAN,
+			[DR_ACTION_TYP_MODIFY_HDR]	= DR_ACTION_STATE_MODIFY_HDR,
+			[DR_ACTION_TYP_L2_TO_TNL_L2]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_L2_TO_TNL_L3]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_INSERT_HDR]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_ASO_FLOW_METER]	= DR_ACTION_STATE_ASO,
+		},
+		[DR_ACTION_STATE_PUSH_VLAN] = {
+			[DR_ACTION_TYP_DROP]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_FT]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_SAMPLER]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_CTR]		= DR_ACTION_STATE_PUSH_VLAN,
+			[DR_ACTION_TYP_PUSH_VLAN]	= DR_ACTION_STATE_PUSH_VLAN,
+			[DR_ACTION_TYP_L2_TO_TNL_L2]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_L2_TO_TNL_L3]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_INSERT_HDR]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_ASO_FLOW_METER]	= DR_ACTION_STATE_ASO,
+		},
+		[DR_ACTION_STATE_NON_TERM] = {
+			[DR_ACTION_TYP_DROP]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_FT]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_SAMPLER]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_CTR]		= DR_ACTION_STATE_NON_TERM,
+			[DR_ACTION_TYP_L2_TO_TNL_L2]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_L2_TO_TNL_L3]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_INSERT_HDR]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_REMOVE_HDR]	= DR_ACTION_STATE_DECAP,
+			[DR_ACTION_TYP_MODIFY_HDR]	= DR_ACTION_STATE_MODIFY_HDR,
+			[DR_ACTION_TYP_PUSH_VLAN]	= DR_ACTION_STATE_PUSH_VLAN,
+			[DR_ACTION_TYP_POP_VLAN]	= DR_ACTION_STATE_POP_VLAN,
+			[DR_ACTION_TYP_ASO_FLOW_METER]	= DR_ACTION_STATE_ASO,
+		},
+		[DR_ACTION_STATE_ASO] = {
+			[DR_ACTION_TYP_L2_TO_TNL_L2]    = DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_L2_TO_TNL_L3]    = DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_MODIFY_HDR]      = DR_ACTION_STATE_MODIFY_HDR,
+			[DR_ACTION_TYP_PUSH_VLAN]       = DR_ACTION_STATE_PUSH_VLAN,
+			[DR_ACTION_TYP_CTR]             = DR_ACTION_STATE_ASO,
+			[DR_ACTION_TYP_DROP]            = DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_FT]              = DR_ACTION_STATE_TERM,
+		},
+		[DR_ACTION_STATE_TERM] = {
+			[DR_ACTION_TYP_CTR]		= DR_ACTION_STATE_TERM,
+		},
+	},
+	[DR_ACTION_DOMAIN_FDB_INGRESS] = {
+		[DR_ACTION_STATE_NO_ACTION] = {
+			[DR_ACTION_TYP_DROP]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_FT]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_SAMPLER]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_CTR]		= DR_ACTION_STATE_NON_TERM,
+			[DR_ACTION_TYP_TNL_L2_TO_L2]	= DR_ACTION_STATE_DECAP,
+			[DR_ACTION_TYP_TNL_L3_TO_L2]	= DR_ACTION_STATE_DECAP,
+			[DR_ACTION_TYP_L2_TO_TNL_L2]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_L2_TO_TNL_L3]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_INSERT_HDR]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_REMOVE_HDR]	= DR_ACTION_STATE_DECAP,
+			[DR_ACTION_TYP_MODIFY_HDR]	= DR_ACTION_STATE_MODIFY_HDR,
+			[DR_ACTION_TYP_PUSH_VLAN]	= DR_ACTION_STATE_PUSH_VLAN,
+			[DR_ACTION_TYP_POP_VLAN]	= DR_ACTION_STATE_POP_VLAN,
+			[DR_ACTION_TYP_VPORT]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_ASO_FLOW_METER]	= DR_ACTION_STATE_ASO,
+		},
+		[DR_ACTION_STATE_DECAP] = {
+			[DR_ACTION_TYP_DROP]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_FT]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_CTR]		= DR_ACTION_STATE_DECAP,
+			[DR_ACTION_TYP_SAMPLER]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_MODIFY_HDR]	= DR_ACTION_STATE_MODIFY_HDR,
+			[DR_ACTION_TYP_VPORT]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_L2_TO_TNL_L2]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_L2_TO_TNL_L3]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_INSERT_HDR]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_POP_VLAN]	= DR_ACTION_STATE_POP_VLAN,
+			[DR_ACTION_TYP_PUSH_VLAN]	= DR_ACTION_STATE_PUSH_VLAN,
+			[DR_ACTION_TYP_ASO_FLOW_METER]	= DR_ACTION_STATE_ASO,
+		},
+		[DR_ACTION_STATE_ENCAP] = {
+			[DR_ACTION_TYP_DROP]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_QP]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_FT]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_VPORT]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_SAMPLER]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_CTR]		= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_ASO_FLOW_METER]	= DR_ACTION_STATE_ASO,
+		},
+		[DR_ACTION_STATE_MODIFY_HDR] = {
+			[DR_ACTION_TYP_DROP]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_FT]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_SAMPLER]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_CTR]		= DR_ACTION_STATE_MODIFY_HDR,
+			[DR_ACTION_TYP_VPORT]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_L2_TO_TNL_L2]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_L2_TO_TNL_L3]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_INSERT_HDR]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_PUSH_VLAN]	= DR_ACTION_STATE_PUSH_VLAN,
+			[DR_ACTION_TYP_ASO_FLOW_METER]	= DR_ACTION_STATE_ASO,
+		},
+		[DR_ACTION_STATE_POP_VLAN] = {
+			[DR_ACTION_TYP_DROP]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_FT]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_SAMPLER]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_POP_VLAN]	= DR_ACTION_STATE_POP_VLAN,
+			[DR_ACTION_TYP_CTR]		= DR_ACTION_STATE_POP_VLAN,
+			[DR_ACTION_TYP_VPORT]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_MODIFY_HDR]	= DR_ACTION_STATE_MODIFY_HDR,
+			[DR_ACTION_TYP_L2_TO_TNL_L2]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_L2_TO_TNL_L3]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_INSERT_HDR]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_ASO_FLOW_METER]	= DR_ACTION_STATE_ASO,
+		},
+		[DR_ACTION_STATE_PUSH_VLAN] = {
+			[DR_ACTION_TYP_DROP]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_FT]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_SAMPLER]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_PUSH_VLAN]	= DR_ACTION_STATE_PUSH_VLAN,
+			[DR_ACTION_TYP_CTR]		= DR_ACTION_STATE_PUSH_VLAN,
+			[DR_ACTION_TYP_VPORT]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_MODIFY_HDR]	= DR_ACTION_STATE_MODIFY_HDR,
+			[DR_ACTION_TYP_L2_TO_TNL_L2]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_L2_TO_TNL_L3]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_INSERT_HDR]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_ASO_FLOW_METER]	= DR_ACTION_STATE_ASO,
+		},
+		[DR_ACTION_STATE_NON_TERM] = {
+			[DR_ACTION_TYP_DROP]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_FT]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_SAMPLER]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_CTR]		= DR_ACTION_STATE_NON_TERM,
+			[DR_ACTION_TYP_TNL_L2_TO_L2]	= DR_ACTION_STATE_DECAP,
+			[DR_ACTION_TYP_TNL_L3_TO_L2]	= DR_ACTION_STATE_DECAP,
+			[DR_ACTION_TYP_L2_TO_TNL_L2]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_L2_TO_TNL_L3]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_INSERT_HDR]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_REMOVE_HDR]	= DR_ACTION_STATE_DECAP,
+			[DR_ACTION_TYP_MODIFY_HDR]	= DR_ACTION_STATE_MODIFY_HDR,
+			[DR_ACTION_TYP_POP_VLAN]	= DR_ACTION_STATE_POP_VLAN,
+			[DR_ACTION_TYP_PUSH_VLAN]	= DR_ACTION_STATE_PUSH_VLAN,
+			[DR_ACTION_TYP_VPORT]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_ASO_FLOW_METER]	= DR_ACTION_STATE_ASO,
+		},
+		[DR_ACTION_STATE_ASO] = {
+			[DR_ACTION_TYP_DROP]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_FT]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_VPORT]           = DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_CTR]             = DR_ACTION_STATE_ASO,
+		},
+		[DR_ACTION_STATE_TERM] = {
+			[DR_ACTION_TYP_CTR]		= DR_ACTION_STATE_TERM,
+		},
+	},
+	[DR_ACTION_DOMAIN_FDB_EGRESS] = {
+		[DR_ACTION_STATE_NO_ACTION] = {
+			[DR_ACTION_TYP_DROP]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_FT]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_SAMPLER]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_CTR]		= DR_ACTION_STATE_NON_TERM,
+			[DR_ACTION_TYP_MODIFY_HDR]	= DR_ACTION_STATE_MODIFY_HDR,
+			[DR_ACTION_TYP_L2_TO_TNL_L2]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_L2_TO_TNL_L3]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_INSERT_HDR]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_REMOVE_HDR]	= DR_ACTION_STATE_DECAP,
+			[DR_ACTION_TYP_PUSH_VLAN]	= DR_ACTION_STATE_PUSH_VLAN,
+			[DR_ACTION_TYP_POP_VLAN]	= DR_ACTION_STATE_POP_VLAN,
+			[DR_ACTION_TYP_VPORT]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_ASO_FLOW_METER]	= DR_ACTION_STATE_ASO,
+		},
+		[DR_ACTION_STATE_DECAP] = {
+			[DR_ACTION_TYP_DROP]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_FT]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_CTR]		= DR_ACTION_STATE_DECAP,
+			[DR_ACTION_TYP_VPORT]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_SAMPLER]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_ASO_FLOW_METER]  = DR_ACTION_STATE_ASO,
+		},
+		[DR_ACTION_STATE_ENCAP] = {
+			[DR_ACTION_TYP_DROP]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_FT]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_CTR]		= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_SAMPLER]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_VPORT]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_ASO_FLOW_METER]  = DR_ACTION_STATE_ASO,
+		},
+		[DR_ACTION_STATE_MODIFY_HDR] = {
+			[DR_ACTION_TYP_DROP]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_FT]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_SAMPLER]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_CTR]		= DR_ACTION_STATE_MODIFY_HDR,
+			[DR_ACTION_TYP_L2_TO_TNL_L2]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_L2_TO_TNL_L3]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_INSERT_HDR]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_PUSH_VLAN]	= DR_ACTION_STATE_PUSH_VLAN,
+			[DR_ACTION_TYP_VPORT]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_ASO_FLOW_METER]	= DR_ACTION_STATE_ASO,
+		},
+		[DR_ACTION_STATE_POP_VLAN] = {
+			[DR_ACTION_TYP_FT]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_SAMPLER]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_CTR]		= DR_ACTION_STATE_POP_VLAN,
+			[DR_ACTION_TYP_POP_VLAN]	= DR_ACTION_STATE_POP_VLAN,
+			[DR_ACTION_TYP_PUSH_VLAN]	= DR_ACTION_STATE_PUSH_VLAN,
+			[DR_ACTION_TYP_MODIFY_HDR]	= DR_ACTION_STATE_MODIFY_HDR,
+			[DR_ACTION_TYP_L2_TO_TNL_L2]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_L2_TO_TNL_L3]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_INSERT_HDR]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_VPORT]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_ASO_FLOW_METER]  = DR_ACTION_STATE_ASO,
+		},
+		[DR_ACTION_STATE_PUSH_VLAN] = {
+			[DR_ACTION_TYP_DROP]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_FT]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_SAMPLER]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_PUSH_VLAN]	= DR_ACTION_STATE_PUSH_VLAN,
+			[DR_ACTION_TYP_CTR]		= DR_ACTION_STATE_PUSH_VLAN,
+			[DR_ACTION_TYP_L2_TO_TNL_L2]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_L2_TO_TNL_L3]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_INSERT_HDR]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_VPORT]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_ASO_FLOW_METER]  = DR_ACTION_STATE_ASO,
+		},
+		[DR_ACTION_STATE_NON_TERM] = {
+			[DR_ACTION_TYP_DROP]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_FT]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_SAMPLER]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_CTR]		= DR_ACTION_STATE_NON_TERM,
+			[DR_ACTION_TYP_MODIFY_HDR]	= DR_ACTION_STATE_MODIFY_HDR,
+			[DR_ACTION_TYP_L2_TO_TNL_L2]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_L2_TO_TNL_L3]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_INSERT_HDR]	= DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_REMOVE_HDR]	= DR_ACTION_STATE_DECAP,
+			[DR_ACTION_TYP_PUSH_VLAN]	= DR_ACTION_STATE_PUSH_VLAN,
+			[DR_ACTION_TYP_POP_VLAN]	= DR_ACTION_STATE_POP_VLAN,
+			[DR_ACTION_TYP_VPORT]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_ASO_FLOW_METER]  = DR_ACTION_STATE_ASO,
+		},
+		[DR_ACTION_STATE_ASO] = {
+			[DR_ACTION_TYP_L2_TO_TNL_L2]    = DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_L2_TO_TNL_L3]    = DR_ACTION_STATE_ENCAP,
+			[DR_ACTION_TYP_MODIFY_HDR]      = DR_ACTION_STATE_MODIFY_HDR,
+			[DR_ACTION_TYP_PUSH_VLAN]       = DR_ACTION_STATE_PUSH_VLAN,
+			[DR_ACTION_TYP_DROP]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_FT]		= DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_VPORT]           = DR_ACTION_STATE_TERM,
+			[DR_ACTION_TYP_CTR]             = DR_ACTION_STATE_ASO,
+		},
+		[DR_ACTION_STATE_TERM] = {
+			[DR_ACTION_TYP_CTR]		= DR_ACTION_STATE_TERM,
+		},
+	},
+};
+
+static int
+dr_action_reformat_to_action_type(enum mlx5dr_action_reformat_type reformat_type,
+				  enum mlx5dr_action_type *action_type)
+{
+	switch (reformat_type) {
+	case DR_ACTION_REFORMAT_TYP_TNL_L2_TO_L2:
+		*action_type = DR_ACTION_TYP_TNL_L2_TO_L2;
+		break;
+	case DR_ACTION_REFORMAT_TYP_L2_TO_TNL_L2:
+		*action_type = DR_ACTION_TYP_L2_TO_TNL_L2;
+		break;
+	case DR_ACTION_REFORMAT_TYP_TNL_L3_TO_L2:
+		*action_type = DR_ACTION_TYP_TNL_L3_TO_L2;
+		break;
+	case DR_ACTION_REFORMAT_TYP_L2_TO_TNL_L3:
+		*action_type = DR_ACTION_TYP_L2_TO_TNL_L3;
+		break;
+	case DR_ACTION_REFORMAT_TYP_INSERT_HDR:
+		*action_type = DR_ACTION_TYP_INSERT_HDR;
+		break;
+	case DR_ACTION_REFORMAT_TYP_REMOVE_HDR:
+		*action_type = DR_ACTION_TYP_REMOVE_HDR;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+/* Apply the actions on the rule STE array starting from the last_ste.
+ * Actions might require more than one STE, new_num_stes will return
+ * the new size of the STEs array, rule with actions.
+ */
+static void dr_actions_apply(struct mlx5dr_domain *dmn,
+			     enum mlx5dr_domain_nic_type nic_type,
+			     u8 *action_type_set,
+			     u8 *last_ste,
+			     struct mlx5dr_ste_actions_attr *attr,
+			     u32 *new_num_stes)
+{
+	struct mlx5dr_ste_ctx *ste_ctx = dmn->ste_ctx;
+	u32 added_stes = 0;
+
+	if (nic_type == DR_DOMAIN_NIC_TYPE_RX)
+		mlx5dr_ste_set_actions_rx(ste_ctx, dmn, action_type_set,
+					  last_ste, attr, &added_stes);
+	else
+		mlx5dr_ste_set_actions_tx(ste_ctx, dmn, action_type_set,
+					  last_ste, attr, &added_stes);
+
+	*new_num_stes += added_stes;
+}
+
+static enum dr_action_domain
+dr_action_get_action_domain(enum mlx5dr_domain_type domain,
+			    enum mlx5dr_domain_nic_type nic_type)
+{
+	switch (domain) {
+	case MLX5DR_DOMAIN_TYPE_NIC_RX:
+		return DR_ACTION_DOMAIN_NIC_INGRESS;
+	case MLX5DR_DOMAIN_TYPE_NIC_TX:
+		return DR_ACTION_DOMAIN_NIC_EGRESS;
+	case MLX5DR_DOMAIN_TYPE_FDB:
+		if (nic_type == DR_DOMAIN_NIC_TYPE_RX)
+			return DR_ACTION_DOMAIN_FDB_INGRESS;
+		return DR_ACTION_DOMAIN_FDB_EGRESS;
+	default:
+		WARN_ON(true);
+		return DR_ACTION_DOMAIN_MAX;
+	}
+}
+
+static
+int dr_action_validate_and_get_next_state(enum dr_action_domain action_domain,
+					  u32 action_type,
+					  u32 *state)
+{
+	u32 cur_state = *state;
+
+	/* Check action state machine is valid */
+	*state = next_action_state[action_domain][cur_state][action_type];
+
+	if (*state == DR_ACTION_STATE_ERR)
+		return -EOPNOTSUPP;
+
+	return 0;
+}
+
+static int dr_action_handle_cs_recalc(struct mlx5dr_domain *dmn,
+				      struct mlx5dr_action *dest_action,
+				      u64 *final_icm_addr)
+{
+	int ret;
+
+	switch (dest_action->action_type) {
+	case DR_ACTION_TYP_FT:
+		/* Allow destination flow table only if table is a terminating
+		 * table, since there is an *assumption* that in such case FW
+		 * will recalculate the CS.
+		 */
+		if (dest_action->dest_tbl->is_fw_tbl) {
+			*final_icm_addr = dest_action->dest_tbl->fw_tbl.rx_icm_addr;
+		} else {
+			mlx5dr_dbg(dmn,
+				   "Destination FT should be terminating when modify TTL is used\n");
+			return -EINVAL;
+		}
+		break;
+
+	case DR_ACTION_TYP_VPORT:
+		/* If destination is vport we will get the FW flow table
+		 * that recalculates the CS and forwards to the vport.
+		 */
+		ret = mlx5dr_domain_get_recalc_cs_ft_addr(dest_action->vport->dmn,
+							  dest_action->vport->caps->num,
+							  final_icm_addr);
+		if (ret) {
+			mlx5dr_err(dmn, "Failed to get FW cs recalc flow table\n");
+			return ret;
+		}
+		break;
+
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static void dr_action_modify_ttl_adjust(struct mlx5dr_domain *dmn,
+					struct mlx5dr_ste_actions_attr *attr,
+					bool rx_rule,
+					bool *recalc_cs_required)
+{
+	*recalc_cs_required = false;
+
+	/* if device supports csum recalculation - no adjustment needed */
+	if (mlx5dr_ste_supp_ttl_cs_recalc(&dmn->info.caps))
+		return;
+
+	/* no adjustment needed on TX rules */
+	if (!rx_rule)
+		return;
+
+	if (!MLX5_CAP_ESW_FLOWTABLE(dmn->mdev, fdb_ipv4_ttl_modify)) {
+		/* Ignore the modify TTL action.
+		 * It is always kept as last HW action.
+		 */
+		attr->modify_actions--;
+		return;
+	}
+
+	if (dmn->type == MLX5DR_DOMAIN_TYPE_FDB)
+		/* Due to a HW bug on some devices, modifying TTL on RX flows
+		 * will cause an incorrect checksum calculation. In such cases
+		 * we will use a FW table to recalculate the checksum.
+		 */
+		*recalc_cs_required = true;
+}
+
+static void dr_action_print_sequence(struct mlx5dr_domain *dmn,
+				     struct mlx5dr_action *actions[],
+				     int last_idx)
+{
+	int i;
+
+	for (i = 0; i <= last_idx; i++)
+		mlx5dr_err(dmn, "< %s (%d) > ",
+			   dr_action_id_to_str(actions[i]->action_type),
+			   actions[i]->action_type);
+}
+
+#define WITH_VLAN_NUM_HW_ACTIONS 6
+
+int mlx5dr_actions_build_ste_arr(struct mlx5dr_matcher *matcher,
+				 struct mlx5dr_matcher_rx_tx *nic_matcher,
+				 struct mlx5dr_action *actions[],
+				 u32 num_actions,
+				 u8 *ste_arr,
+				 u32 *new_hw_ste_arr_sz)
+{
+	struct mlx5dr_domain_rx_tx *nic_dmn = nic_matcher->nic_tbl->nic_dmn;
+	bool rx_rule = nic_dmn->type == DR_DOMAIN_NIC_TYPE_RX;
+	struct mlx5dr_domain *dmn = matcher->tbl->dmn;
+	u8 action_type_set[DR_ACTION_TYP_MAX] = {};
+	struct mlx5dr_ste_actions_attr attr = {};
+	struct mlx5dr_action *dest_action = NULL;
+	u32 state = DR_ACTION_STATE_NO_ACTION;
+	enum dr_action_domain action_domain;
+	bool recalc_cs_required = false;
+	u8 *last_ste;
+	int i, ret;
+
+	attr.gvmi = dmn->info.caps.gvmi;
+	attr.hit_gvmi = dmn->info.caps.gvmi;
+	attr.final_icm_addr = nic_dmn->default_icm_addr;
+	action_domain = dr_action_get_action_domain(dmn->type, nic_dmn->type);
+
+	for (i = 0; i < num_actions; i++) {
+		struct mlx5dr_action_dest_tbl *dest_tbl;
+		struct mlx5dr_icm_chunk *chunk;
+		struct mlx5dr_action *action;
+		int max_actions_type = 1;
+		u32 action_type;
+
+		action = actions[i];
+		action_type = action->action_type;
+
+		switch (action_type) {
+		case DR_ACTION_TYP_DROP:
+			attr.final_icm_addr = nic_dmn->drop_icm_addr;
+			attr.hit_gvmi = nic_dmn->drop_icm_addr >> 48;
+			break;
+		case DR_ACTION_TYP_FT:
+			dest_action = action;
+			dest_tbl = action->dest_tbl;
+			if (!dest_tbl->is_fw_tbl) {
+				if (dest_tbl->tbl->dmn != dmn) {
+					mlx5dr_err(dmn,
+						   "Destination table belongs to a different domain\n");
+					return -EINVAL;
+				}
+				if (dest_tbl->tbl->level <= matcher->tbl->level) {
+					mlx5_core_dbg_once(dmn->mdev,
+							   "Connecting table to a lower/same level destination table\n");
+					mlx5dr_dbg(dmn,
+						   "Connecting table at level %d to a destination table at level %d\n",
+						   matcher->tbl->level,
+						   dest_tbl->tbl->level);
+				}
+				chunk = rx_rule ? dest_tbl->tbl->rx.s_anchor->chunk :
+					dest_tbl->tbl->tx.s_anchor->chunk;
+				attr.final_icm_addr = mlx5dr_icm_pool_get_chunk_icm_addr(chunk);
+			} else {
+				struct mlx5dr_cmd_query_flow_table_details output;
+				int ret;
+
+				/* get the relevant addresses */
+				if (!action->dest_tbl->fw_tbl.rx_icm_addr) {
+					ret = mlx5dr_cmd_query_flow_table(dmn->mdev,
+									  dest_tbl->fw_tbl.type,
+									  dest_tbl->fw_tbl.id,
+									  &output);
+					if (!ret) {
+						dest_tbl->fw_tbl.tx_icm_addr =
+							output.sw_owner_icm_root_1;
+						dest_tbl->fw_tbl.rx_icm_addr =
+							output.sw_owner_icm_root_0;
+					} else {
+						mlx5dr_err(dmn,
+							   "Failed mlx5_cmd_query_flow_table ret: %d\n",
+							   ret);
+						return ret;
+					}
+				}
+				attr.final_icm_addr = rx_rule ?
+					dest_tbl->fw_tbl.rx_icm_addr :
+					dest_tbl->fw_tbl.tx_icm_addr;
+			}
+			break;
+		case DR_ACTION_TYP_QP:
+			mlx5dr_info(dmn, "Domain doesn't support QP\n");
+			return -EOPNOTSUPP;
+		case DR_ACTION_TYP_CTR:
+			attr.ctr_id = action->ctr->ctr_id +
+				action->ctr->offset;
+			break;
+		case DR_ACTION_TYP_TAG:
+			attr.flow_tag = action->flow_tag->flow_tag;
+			break;
+		case DR_ACTION_TYP_TNL_L2_TO_L2:
+			break;
+		case DR_ACTION_TYP_TNL_L3_TO_L2:
+			attr.decap_index = action->rewrite->index;
+			attr.decap_actions = action->rewrite->num_of_actions;
+			attr.decap_with_vlan =
+				attr.decap_actions == WITH_VLAN_NUM_HW_ACTIONS;
+			break;
+		case DR_ACTION_TYP_MODIFY_HDR:
+			attr.modify_index = action->rewrite->index;
+			attr.modify_actions = action->rewrite->num_of_actions;
+			if (action->rewrite->modify_ttl)
+				dr_action_modify_ttl_adjust(dmn, &attr, rx_rule,
+							    &recalc_cs_required);
+			break;
+		case DR_ACTION_TYP_L2_TO_TNL_L2:
+		case DR_ACTION_TYP_L2_TO_TNL_L3:
+			if (rx_rule &&
+			    !(dmn->ste_ctx->actions_caps & DR_STE_CTX_ACTION_CAP_RX_ENCAP)) {
+				mlx5dr_info(dmn, "Device doesn't support Encap on RX\n");
+				return -EOPNOTSUPP;
+			}
+			attr.reformat.size = action->reformat->size;
+			attr.reformat.id = action->reformat->id;
+			break;
+		case DR_ACTION_TYP_SAMPLER:
+			attr.final_icm_addr = rx_rule ? action->sampler->rx_icm_addr :
+							action->sampler->tx_icm_addr;
+			break;
+		case DR_ACTION_TYP_VPORT:
+			if (unlikely(rx_rule && action->vport->caps->num == MLX5_VPORT_UPLINK)) {
+				/* can't go to uplink on RX rule - dropping instead */
+				attr.final_icm_addr = nic_dmn->drop_icm_addr;
+				attr.hit_gvmi = nic_dmn->drop_icm_addr >> 48;
+			} else {
+				attr.hit_gvmi = action->vport->caps->vhca_gvmi;
+				dest_action = action;
+				attr.final_icm_addr = rx_rule ?
+						      action->vport->caps->icm_address_rx :
+						      action->vport->caps->icm_address_tx;
+			}
+			break;
+		case DR_ACTION_TYP_POP_VLAN:
+			if (!rx_rule && !(dmn->ste_ctx->actions_caps &
+					  DR_STE_CTX_ACTION_CAP_TX_POP)) {
+				mlx5dr_dbg(dmn, "Device doesn't support POP VLAN action on TX\n");
+				return -EOPNOTSUPP;
+			}
+
+			max_actions_type = MLX5DR_MAX_VLANS;
+			attr.vlans.count++;
+			break;
+		case DR_ACTION_TYP_PUSH_VLAN:
+			if (rx_rule && !(dmn->ste_ctx->actions_caps &
+					 DR_STE_CTX_ACTION_CAP_RX_PUSH)) {
+				mlx5dr_dbg(dmn, "Device doesn't support PUSH VLAN action on RX\n");
+				return -EOPNOTSUPP;
+			}
+
+			max_actions_type = MLX5DR_MAX_VLANS;
+			if (attr.vlans.count == MLX5DR_MAX_VLANS) {
+				mlx5dr_dbg(dmn, "Max VLAN push/pop count exceeded\n");
+				return -EINVAL;
+			}
+
+			attr.vlans.headers[attr.vlans.count++] = action->push_vlan->vlan_hdr;
+			break;
+		case DR_ACTION_TYP_INSERT_HDR:
+		case DR_ACTION_TYP_REMOVE_HDR:
+			attr.reformat.size = action->reformat->size;
+			attr.reformat.id = action->reformat->id;
+			attr.reformat.param_0 = action->reformat->param_0;
+			attr.reformat.param_1 = action->reformat->param_1;
+			break;
+		case DR_ACTION_TYP_ASO_FLOW_METER:
+			attr.aso_flow_meter.obj_id = action->aso->obj_id;
+			attr.aso_flow_meter.offset = action->aso->offset;
+			attr.aso_flow_meter.dest_reg_id = action->aso->dest_reg_id;
+			attr.aso_flow_meter.init_color = action->aso->init_color;
+			break;
+		default:
+			mlx5dr_err(dmn, "Unsupported action type %d\n", action_type);
+			return -EINVAL;
+		}
+
+		/* Check action duplication */
+		if (++action_type_set[action_type] > max_actions_type) {
+			mlx5dr_err(dmn, "Action type %d supports only max %d time(s)\n",
+				   action_type, max_actions_type);
+			return -EINVAL;
+		}
+
+		/* Check action state machine is valid */
+		if (dr_action_validate_and_get_next_state(action_domain,
+							  action_type,
+							  &state)) {
+			mlx5dr_err(dmn, "Invalid action (gvmi: %d, is_rx: %d) sequence provided:",
+				   attr.gvmi, rx_rule);
+			dr_action_print_sequence(dmn, actions, i);
+			return -EOPNOTSUPP;
+		}
+	}
+
+	*new_hw_ste_arr_sz = nic_matcher->num_of_builders;
+	last_ste = ste_arr + DR_STE_SIZE * (nic_matcher->num_of_builders - 1);
+
+	if (recalc_cs_required && dest_action) {
+		ret = dr_action_handle_cs_recalc(dmn, dest_action, &attr.final_icm_addr);
+		if (ret) {
+			mlx5dr_err(dmn,
+				   "Failed to handle checksum recalculation err %d\n",
+				   ret);
+			return ret;
+		}
+	}
+
+	dr_actions_apply(dmn,
+			 nic_dmn->type,
+			 action_type_set,
+			 last_ste,
+			 &attr,
+			 new_hw_ste_arr_sz);
+
+	return 0;
+}
+
+static unsigned int action_size[DR_ACTION_TYP_MAX] = {
+	[DR_ACTION_TYP_TNL_L2_TO_L2] = sizeof(struct mlx5dr_action_reformat),
+	[DR_ACTION_TYP_L2_TO_TNL_L2] = sizeof(struct mlx5dr_action_reformat),
+	[DR_ACTION_TYP_TNL_L3_TO_L2] = sizeof(struct mlx5dr_action_rewrite),
+	[DR_ACTION_TYP_L2_TO_TNL_L3] = sizeof(struct mlx5dr_action_reformat),
+	[DR_ACTION_TYP_FT]           = sizeof(struct mlx5dr_action_dest_tbl),
+	[DR_ACTION_TYP_CTR]          = sizeof(struct mlx5dr_action_ctr),
+	[DR_ACTION_TYP_TAG]          = sizeof(struct mlx5dr_action_flow_tag),
+	[DR_ACTION_TYP_MODIFY_HDR]   = sizeof(struct mlx5dr_action_rewrite),
+	[DR_ACTION_TYP_VPORT]        = sizeof(struct mlx5dr_action_vport),
+	[DR_ACTION_TYP_PUSH_VLAN]    = sizeof(struct mlx5dr_action_push_vlan),
+	[DR_ACTION_TYP_INSERT_HDR]   = sizeof(struct mlx5dr_action_reformat),
+	[DR_ACTION_TYP_REMOVE_HDR]   = sizeof(struct mlx5dr_action_reformat),
+	[DR_ACTION_TYP_SAMPLER]      = sizeof(struct mlx5dr_action_sampler),
+	[DR_ACTION_TYP_ASO_FLOW_METER] = sizeof(struct mlx5dr_action_aso_flow_meter),
+};
+
+static struct mlx5dr_action *
+dr_action_create_generic(enum mlx5dr_action_type action_type)
+{
+	struct mlx5dr_action *action;
+	int extra_size;
+
+	if (action_type < DR_ACTION_TYP_MAX)
+		extra_size = action_size[action_type];
+	else
+		return NULL;
+
+	action = kzalloc(sizeof(*action) + extra_size, GFP_KERNEL);
+	if (!action)
+		return NULL;
+
+	action->action_type = action_type;
+	refcount_set(&action->refcount, 1);
+	action->data = action + 1;
+
+	return action;
+}
+
+struct mlx5dr_action *mlx5dr_action_create_drop(void)
+{
+	return dr_action_create_generic(DR_ACTION_TYP_DROP);
+}
+
+struct mlx5dr_action *
+mlx5dr_action_create_dest_table_num(struct mlx5dr_domain *dmn, u32 table_num)
+{
+	struct mlx5dr_action *action;
+
+	action = dr_action_create_generic(DR_ACTION_TYP_FT);
+	if (!action)
+		return NULL;
+
+	action->dest_tbl->is_fw_tbl = true;
+	action->dest_tbl->fw_tbl.dmn = dmn;
+	action->dest_tbl->fw_tbl.id = table_num;
+	action->dest_tbl->fw_tbl.type = FS_FT_FDB;
+	refcount_inc(&dmn->refcount);
+
+	return action;
+}
+
+struct mlx5dr_action *
+mlx5dr_action_create_dest_table(struct mlx5dr_table *tbl)
+{
+	struct mlx5dr_action *action;
+
+	refcount_inc(&tbl->refcount);
+
+	action = dr_action_create_generic(DR_ACTION_TYP_FT);
+	if (!action)
+		goto dec_ref;
+
+	action->dest_tbl->tbl = tbl;
+
+	return action;
+
+dec_ref:
+	refcount_dec(&tbl->refcount);
+	return NULL;
+}
+
+struct mlx5dr_action *
+mlx5dr_action_create_mult_dest_tbl(struct mlx5dr_domain *dmn,
+				   struct mlx5dr_action_dest *dests,
+				   u32 num_of_dests,
+				   bool ignore_flow_level,
+				   u32 flow_source)
+{
+	struct mlx5dr_cmd_flow_destination_hw_info *hw_dests;
+	struct mlx5dr_action **ref_actions;
+	struct mlx5dr_action *action;
+	bool reformat_req = false;
+	u32 num_of_ref = 0;
+	u32 ref_act_cnt;
+	int ret;
+	int i;
+
+	if (dmn->type != MLX5DR_DOMAIN_TYPE_FDB) {
+		mlx5dr_err(dmn, "Multiple destination support is for FDB only\n");
+		return NULL;
+	}
+
+	hw_dests = kcalloc(num_of_dests, sizeof(*hw_dests), GFP_KERNEL);
+	if (!hw_dests)
+		return NULL;
+
+	if (unlikely(check_mul_overflow(num_of_dests, 2u, &ref_act_cnt)))
+		goto free_hw_dests;
+
+	ref_actions = kcalloc(ref_act_cnt, sizeof(*ref_actions), GFP_KERNEL);
+	if (!ref_actions)
+		goto free_hw_dests;
+
+	for (i = 0; i < num_of_dests; i++) {
+		struct mlx5dr_action *reformat_action = dests[i].reformat;
+		struct mlx5dr_action *dest_action = dests[i].dest;
+
+		ref_actions[num_of_ref++] = dest_action;
+
+		switch (dest_action->action_type) {
+		case DR_ACTION_TYP_VPORT:
+			hw_dests[i].vport.flags = MLX5_FLOW_DEST_VPORT_VHCA_ID;
+			hw_dests[i].type = MLX5_FLOW_DESTINATION_TYPE_VPORT;
+			hw_dests[i].vport.num = dest_action->vport->caps->num;
+			hw_dests[i].vport.vhca_id = dest_action->vport->caps->vhca_gvmi;
+			if (reformat_action) {
+				reformat_req = true;
+				hw_dests[i].vport.reformat_id =
+					reformat_action->reformat->id;
+				ref_actions[num_of_ref++] = reformat_action;
+				hw_dests[i].vport.flags |= MLX5_FLOW_DEST_VPORT_REFORMAT_ID;
+			}
+			break;
+
+		case DR_ACTION_TYP_FT:
+			hw_dests[i].type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
+			if (dest_action->dest_tbl->is_fw_tbl)
+				hw_dests[i].ft_id = dest_action->dest_tbl->fw_tbl.id;
+			else
+				hw_dests[i].ft_id = dest_action->dest_tbl->tbl->table_id;
+			break;
+
+		default:
+			mlx5dr_dbg(dmn, "Invalid multiple destinations action\n");
+			goto free_ref_actions;
+		}
+	}
+
+	action = dr_action_create_generic(DR_ACTION_TYP_FT);
+	if (!action)
+		goto free_ref_actions;
+
+	ret = mlx5dr_fw_create_md_tbl(dmn,
+				      hw_dests,
+				      num_of_dests,
+				      reformat_req,
+				      &action->dest_tbl->fw_tbl.id,
+				      &action->dest_tbl->fw_tbl.group_id,
+				      ignore_flow_level,
+				      flow_source);
+	if (ret)
+		goto free_action;
+
+	refcount_inc(&dmn->refcount);
+
+	for (i = 0; i < num_of_ref; i++)
+		refcount_inc(&ref_actions[i]->refcount);
+
+	action->dest_tbl->is_fw_tbl = true;
+	action->dest_tbl->fw_tbl.dmn = dmn;
+	action->dest_tbl->fw_tbl.type = FS_FT_FDB;
+	action->dest_tbl->fw_tbl.ref_actions = ref_actions;
+	action->dest_tbl->fw_tbl.num_of_ref_actions = num_of_ref;
+
+	kfree(hw_dests);
+
+	return action;
+
+free_action:
+	kfree(action);
+free_ref_actions:
+	kfree(ref_actions);
+free_hw_dests:
+	kfree(hw_dests);
+	return NULL;
+}
+
+struct mlx5dr_action *
+mlx5dr_action_create_dest_flow_fw_table(struct mlx5dr_domain *dmn,
+					struct mlx5_flow_table *ft)
+{
+	struct mlx5dr_action *action;
+
+	action = dr_action_create_generic(DR_ACTION_TYP_FT);
+	if (!action)
+		return NULL;
+
+	action->dest_tbl->is_fw_tbl = 1;
+	action->dest_tbl->fw_tbl.type = ft->type;
+	action->dest_tbl->fw_tbl.id = ft->id;
+	action->dest_tbl->fw_tbl.dmn = dmn;
+
+	refcount_inc(&dmn->refcount);
+
+	return action;
+}
+
+struct mlx5dr_action *
+mlx5dr_action_create_flow_counter(u32 counter_id)
+{
+	struct mlx5dr_action *action;
+
+	action = dr_action_create_generic(DR_ACTION_TYP_CTR);
+	if (!action)
+		return NULL;
+
+	action->ctr->ctr_id = counter_id;
+
+	return action;
+}
+
+struct mlx5dr_action *mlx5dr_action_create_tag(u32 tag_value)
+{
+	struct mlx5dr_action *action;
+
+	action = dr_action_create_generic(DR_ACTION_TYP_TAG);
+	if (!action)
+		return NULL;
+
+	action->flow_tag->flow_tag = tag_value & 0xffffff;
+
+	return action;
+}
+
+struct mlx5dr_action *
+mlx5dr_action_create_flow_sampler(struct mlx5dr_domain *dmn, u32 sampler_id)
+{
+	struct mlx5dr_action *action;
+	u64 icm_rx, icm_tx;
+	int ret;
+
+	ret = mlx5dr_cmd_query_flow_sampler(dmn->mdev, sampler_id,
+					    &icm_rx, &icm_tx);
+	if (ret)
+		return NULL;
+
+	action = dr_action_create_generic(DR_ACTION_TYP_SAMPLER);
+	if (!action)
+		return NULL;
+
+	action->sampler->dmn = dmn;
+	action->sampler->sampler_id = sampler_id;
+	action->sampler->rx_icm_addr = icm_rx;
+	action->sampler->tx_icm_addr = icm_tx;
+
+	refcount_inc(&dmn->refcount);
+	return action;
+}
+
+static int
+dr_action_verify_reformat_params(enum mlx5dr_action_type reformat_type,
+				 struct mlx5dr_domain *dmn,
+				 u8 reformat_param_0,
+				 u8 reformat_param_1,
+				 size_t data_sz,
+				 void *data)
+{
+	if (reformat_type == DR_ACTION_TYP_INSERT_HDR) {
+		if ((!data && data_sz) || (data && !data_sz) ||
+		    MLX5_CAP_GEN_2(dmn->mdev, max_reformat_insert_size) < data_sz ||
+		    MLX5_CAP_GEN_2(dmn->mdev, max_reformat_insert_offset) < reformat_param_1) {
+			mlx5dr_dbg(dmn, "Invalid reformat parameters for INSERT_HDR\n");
+			goto out_err;
+		}
+	} else if (reformat_type == DR_ACTION_TYP_REMOVE_HDR) {
+		if (data ||
+		    MLX5_CAP_GEN_2(dmn->mdev, max_reformat_remove_size) < data_sz ||
+		    MLX5_CAP_GEN_2(dmn->mdev, max_reformat_remove_offset) < reformat_param_1) {
+			mlx5dr_dbg(dmn, "Invalid reformat parameters for REMOVE_HDR\n");
+			goto out_err;
+		}
+	} else if (reformat_param_0 || reformat_param_1 ||
+		   reformat_type > DR_ACTION_TYP_REMOVE_HDR) {
+		mlx5dr_dbg(dmn, "Invalid reformat parameters\n");
+		goto out_err;
+	}
+
+	if (dmn->type == MLX5DR_DOMAIN_TYPE_FDB)
+		return 0;
+
+	if (dmn->type == MLX5DR_DOMAIN_TYPE_NIC_RX) {
+		if (reformat_type != DR_ACTION_TYP_TNL_L2_TO_L2 &&
+		    reformat_type != DR_ACTION_TYP_TNL_L3_TO_L2) {
+			mlx5dr_dbg(dmn, "Action reformat type not support on RX domain\n");
+			goto out_err;
+		}
+	} else if (dmn->type == MLX5DR_DOMAIN_TYPE_NIC_TX) {
+		if (reformat_type != DR_ACTION_TYP_L2_TO_TNL_L2 &&
+		    reformat_type != DR_ACTION_TYP_L2_TO_TNL_L3) {
+			mlx5dr_dbg(dmn, "Action reformat type not support on TX domain\n");
+			goto out_err;
+		}
+	}
+
+	return 0;
+
+out_err:
+	return -EINVAL;
+}
+
+#define ACTION_CACHE_LINE_SIZE 64
+
+static int
+dr_action_create_reformat_action(struct mlx5dr_domain *dmn,
+				 u8 reformat_param_0, u8 reformat_param_1,
+				 size_t data_sz, void *data,
+				 struct mlx5dr_action *action)
+{
+	u32 reformat_id;
+	int ret;
+
+	switch (action->action_type) {
+	case DR_ACTION_TYP_L2_TO_TNL_L2:
+	case DR_ACTION_TYP_L2_TO_TNL_L3:
+	{
+		enum mlx5_reformat_ctx_type rt;
+
+		if (action->action_type == DR_ACTION_TYP_L2_TO_TNL_L2)
+			rt = MLX5_REFORMAT_TYPE_L2_TO_L2_TUNNEL;
+		else
+			rt = MLX5_REFORMAT_TYPE_L2_TO_L3_TUNNEL;
+
+		ret = mlx5dr_cmd_create_reformat_ctx(dmn->mdev, rt, 0, 0,
+						     data_sz, data,
+						     &reformat_id);
+		if (ret)
+			return ret;
+
+		action->reformat->id = reformat_id;
+		action->reformat->size = data_sz;
+		return 0;
+	}
+	case DR_ACTION_TYP_TNL_L2_TO_L2:
+	{
+		return 0;
+	}
+	case DR_ACTION_TYP_TNL_L3_TO_L2:
+	{
+		u8 *hw_actions;
+		int ret;
+
+		hw_actions = kzalloc(ACTION_CACHE_LINE_SIZE, GFP_KERNEL);
+		if (!hw_actions)
+			return -ENOMEM;
+
+		ret = mlx5dr_ste_set_action_decap_l3_list(dmn->ste_ctx,
+							  data, data_sz,
+							  hw_actions,
+							  ACTION_CACHE_LINE_SIZE,
+							  &action->rewrite->num_of_actions);
+		if (ret) {
+			mlx5dr_dbg(dmn, "Failed creating decap l3 action list\n");
+			kfree(hw_actions);
+			return ret;
+		}
+
+		action->rewrite->chunk = mlx5dr_icm_alloc_chunk(dmn->action_icm_pool,
+								DR_CHUNK_SIZE_8);
+		if (!action->rewrite->chunk) {
+			mlx5dr_dbg(dmn, "Failed allocating modify header chunk\n");
+			kfree(hw_actions);
+			return -ENOMEM;
+		}
+
+		action->rewrite->data = (void *)hw_actions;
+		action->rewrite->index = (mlx5dr_icm_pool_get_chunk_icm_addr
+					  (action->rewrite->chunk) -
+					 dmn->info.caps.hdr_modify_icm_addr) /
+					 ACTION_CACHE_LINE_SIZE;
+
+		ret = mlx5dr_send_postsend_action(dmn, action);
+		if (ret) {
+			mlx5dr_dbg(dmn, "Writing decap l3 actions to ICM failed\n");
+			mlx5dr_icm_free_chunk(action->rewrite->chunk);
+			kfree(hw_actions);
+			return ret;
+		}
+		return 0;
+	}
+	case DR_ACTION_TYP_INSERT_HDR:
+		ret = mlx5dr_cmd_create_reformat_ctx(dmn->mdev,
+						     MLX5_REFORMAT_TYPE_INSERT_HDR,
+						     reformat_param_0,
+						     reformat_param_1,
+						     data_sz, data,
+						     &reformat_id);
+		if (ret)
+			return ret;
+
+		action->reformat->id = reformat_id;
+		action->reformat->size = data_sz;
+		action->reformat->param_0 = reformat_param_0;
+		action->reformat->param_1 = reformat_param_1;
+		return 0;
+	case DR_ACTION_TYP_REMOVE_HDR:
+		action->reformat->id = 0;
+		action->reformat->size = data_sz;
+		action->reformat->param_0 = reformat_param_0;
+		action->reformat->param_1 = reformat_param_1;
+		return 0;
+	default:
+		mlx5dr_info(dmn, "Reformat type is not supported %d\n", action->action_type);
+		return -EINVAL;
+	}
+}
+
+#define CVLAN_ETHERTYPE 0x8100
+#define SVLAN_ETHERTYPE 0x88a8
+
+struct mlx5dr_action *mlx5dr_action_create_pop_vlan(void)
+{
+	return dr_action_create_generic(DR_ACTION_TYP_POP_VLAN);
+}
+
+struct mlx5dr_action *mlx5dr_action_create_push_vlan(struct mlx5dr_domain *dmn,
+						     __be32 vlan_hdr)
+{
+	u32 vlan_hdr_h = ntohl(vlan_hdr);
+	u16 ethertype = vlan_hdr_h >> 16;
+	struct mlx5dr_action *action;
+
+	if (ethertype != SVLAN_ETHERTYPE && ethertype != CVLAN_ETHERTYPE) {
+		mlx5dr_dbg(dmn, "Invalid vlan ethertype\n");
+		return NULL;
+	}
+
+	action = dr_action_create_generic(DR_ACTION_TYP_PUSH_VLAN);
+	if (!action)
+		return NULL;
+
+	action->push_vlan->vlan_hdr = vlan_hdr_h;
+	return action;
+}
+
+struct mlx5dr_action *
+mlx5dr_action_create_packet_reformat(struct mlx5dr_domain *dmn,
+				     enum mlx5dr_action_reformat_type reformat_type,
+				     u8 reformat_param_0,
+				     u8 reformat_param_1,
+				     size_t data_sz,
+				     void *data)
+{
+	enum mlx5dr_action_type action_type;
+	struct mlx5dr_action *action;
+	int ret;
+
+	refcount_inc(&dmn->refcount);
+
+	/* General checks */
+	ret = dr_action_reformat_to_action_type(reformat_type, &action_type);
+	if (ret) {
+		mlx5dr_dbg(dmn, "Invalid reformat_type provided\n");
+		goto dec_ref;
+	}
+
+	ret = dr_action_verify_reformat_params(action_type, dmn,
+					       reformat_param_0, reformat_param_1,
+					       data_sz, data);
+	if (ret)
+		goto dec_ref;
+
+	action = dr_action_create_generic(action_type);
+	if (!action)
+		goto dec_ref;
+
+	action->reformat->dmn = dmn;
+
+	ret = dr_action_create_reformat_action(dmn,
+					       reformat_param_0,
+					       reformat_param_1,
+					       data_sz,
+					       data,
+					       action);
+	if (ret) {
+		mlx5dr_dbg(dmn, "Failed creating reformat action %d\n", ret);
+		goto free_action;
+	}
+
+	return action;
+
+free_action:
+	kfree(action);
+dec_ref:
+	refcount_dec(&dmn->refcount);
+	return NULL;
+}
+
+static int
+dr_action_modify_sw_to_hw_add(struct mlx5dr_domain *dmn,
+			      __be64 *sw_action,
+			      __be64 *hw_action,
+			      const struct mlx5dr_ste_action_modify_field **ret_hw_info)
+{
+	const struct mlx5dr_ste_action_modify_field *hw_action_info;
+	u8 max_length;
+	u16 sw_field;
+	u32 data;
+
+	/* Get SW modify action data */
+	sw_field = MLX5_GET(set_action_in, sw_action, field);
+	data = MLX5_GET(set_action_in, sw_action, data);
+
+	/* Convert SW data to HW modify action format */
+	hw_action_info = mlx5dr_ste_conv_modify_hdr_sw_field(dmn->ste_ctx, sw_field);
+	if (!hw_action_info) {
+		mlx5dr_dbg(dmn, "Modify add action invalid field given\n");
+		return -EINVAL;
+	}
+
+	max_length = hw_action_info->end - hw_action_info->start + 1;
+
+	mlx5dr_ste_set_action_add(dmn->ste_ctx,
+				  hw_action,
+				  hw_action_info->hw_field,
+				  hw_action_info->start,
+				  max_length,
+				  data);
+
+	*ret_hw_info = hw_action_info;
+
+	return 0;
+}
+
+static int
+dr_action_modify_sw_to_hw_set(struct mlx5dr_domain *dmn,
+			      __be64 *sw_action,
+			      __be64 *hw_action,
+			      const struct mlx5dr_ste_action_modify_field **ret_hw_info)
+{
+	const struct mlx5dr_ste_action_modify_field *hw_action_info;
+	u8 offset, length, max_length;
+	u16 sw_field;
+	u32 data;
+
+	/* Get SW modify action data */
+	length = MLX5_GET(set_action_in, sw_action, length);
+	offset = MLX5_GET(set_action_in, sw_action, offset);
+	sw_field = MLX5_GET(set_action_in, sw_action, field);
+	data = MLX5_GET(set_action_in, sw_action, data);
+
+	/* Convert SW data to HW modify action format */
+	hw_action_info = mlx5dr_ste_conv_modify_hdr_sw_field(dmn->ste_ctx, sw_field);
+	if (!hw_action_info) {
+		mlx5dr_dbg(dmn, "Modify set action invalid field given\n");
+		return -EINVAL;
+	}
+
+	/* PRM defines that length zero specific length of 32bits */
+	length = length ? length : 32;
+
+	max_length = hw_action_info->end - hw_action_info->start + 1;
+
+	if (length + offset > max_length) {
+		mlx5dr_dbg(dmn, "Modify action length + offset exceeds limit\n");
+		return -EINVAL;
+	}
+
+	mlx5dr_ste_set_action_set(dmn->ste_ctx,
+				  hw_action,
+				  hw_action_info->hw_field,
+				  hw_action_info->start + offset,
+				  length,
+				  data);
+
+	*ret_hw_info = hw_action_info;
+
+	return 0;
+}
+
+static int
+dr_action_modify_sw_to_hw_copy(struct mlx5dr_domain *dmn,
+			       __be64 *sw_action,
+			       __be64 *hw_action,
+			       const struct mlx5dr_ste_action_modify_field **ret_dst_hw_info,
+			       const struct mlx5dr_ste_action_modify_field **ret_src_hw_info)
+{
+	u8 src_offset, dst_offset, src_max_length, dst_max_length, length;
+	const struct mlx5dr_ste_action_modify_field *hw_dst_action_info;
+	const struct mlx5dr_ste_action_modify_field *hw_src_action_info;
+	u16 src_field, dst_field;
+
+	/* Get SW modify action data */
+	src_field = MLX5_GET(copy_action_in, sw_action, src_field);
+	dst_field = MLX5_GET(copy_action_in, sw_action, dst_field);
+	src_offset = MLX5_GET(copy_action_in, sw_action, src_offset);
+	dst_offset = MLX5_GET(copy_action_in, sw_action, dst_offset);
+	length = MLX5_GET(copy_action_in, sw_action, length);
+
+	/* Convert SW data to HW modify action format */
+	hw_src_action_info = mlx5dr_ste_conv_modify_hdr_sw_field(dmn->ste_ctx, src_field);
+	hw_dst_action_info = mlx5dr_ste_conv_modify_hdr_sw_field(dmn->ste_ctx, dst_field);
+	if (!hw_src_action_info || !hw_dst_action_info) {
+		mlx5dr_dbg(dmn, "Modify copy action invalid field given\n");
+		return -EINVAL;
+	}
+
+	/* PRM defines that length zero specific length of 32bits */
+	length = length ? length : 32;
+
+	src_max_length = hw_src_action_info->end -
+			 hw_src_action_info->start + 1;
+	dst_max_length = hw_dst_action_info->end -
+			 hw_dst_action_info->start + 1;
+
+	if (length + src_offset > src_max_length ||
+	    length + dst_offset > dst_max_length) {
+		mlx5dr_dbg(dmn, "Modify action length + offset exceeds limit\n");
+		return -EINVAL;
+	}
+
+	mlx5dr_ste_set_action_copy(dmn->ste_ctx,
+				   hw_action,
+				   hw_dst_action_info->hw_field,
+				   hw_dst_action_info->start + dst_offset,
+				   length,
+				   hw_src_action_info->hw_field,
+				   hw_src_action_info->start + src_offset);
+
+	*ret_dst_hw_info = hw_dst_action_info;
+	*ret_src_hw_info = hw_src_action_info;
+
+	return 0;
+}
+
+static int
+dr_action_modify_sw_to_hw(struct mlx5dr_domain *dmn,
+			  __be64 *sw_action,
+			  __be64 *hw_action,
+			  const struct mlx5dr_ste_action_modify_field **ret_dst_hw_info,
+			  const struct mlx5dr_ste_action_modify_field **ret_src_hw_info)
+{
+	u8 action;
+	int ret;
+
+	*hw_action = 0;
+	*ret_src_hw_info = NULL;
+
+	/* Get SW modify action type */
+	action = MLX5_GET(set_action_in, sw_action, action_type);
+
+	switch (action) {
+	case MLX5_ACTION_TYPE_SET:
+		ret = dr_action_modify_sw_to_hw_set(dmn, sw_action,
+						    hw_action,
+						    ret_dst_hw_info);
+		break;
+
+	case MLX5_ACTION_TYPE_ADD:
+		ret = dr_action_modify_sw_to_hw_add(dmn, sw_action,
+						    hw_action,
+						    ret_dst_hw_info);
+		break;
+
+	case MLX5_ACTION_TYPE_COPY:
+		ret = dr_action_modify_sw_to_hw_copy(dmn, sw_action,
+						     hw_action,
+						     ret_dst_hw_info,
+						     ret_src_hw_info);
+		break;
+
+	default:
+		mlx5dr_info(dmn, "Unsupported action_type for modify action\n");
+		ret = -EOPNOTSUPP;
+	}
+
+	return ret;
+}
+
+static int
+dr_action_modify_check_set_field_limitation(struct mlx5dr_action *action,
+					    const __be64 *sw_action)
+{
+	u16 sw_field = MLX5_GET(set_action_in, sw_action, field);
+	struct mlx5dr_domain *dmn = action->rewrite->dmn;
+
+	if (sw_field == MLX5_ACTION_IN_FIELD_METADATA_REG_A) {
+		action->rewrite->allow_rx = 0;
+		if (dmn->type != MLX5DR_DOMAIN_TYPE_NIC_TX) {
+			mlx5dr_dbg(dmn, "Unsupported field %d for RX/FDB set action\n",
+				   sw_field);
+			return -EINVAL;
+		}
+	} else if (sw_field == MLX5_ACTION_IN_FIELD_METADATA_REG_B) {
+		action->rewrite->allow_tx = 0;
+		if (dmn->type != MLX5DR_DOMAIN_TYPE_NIC_RX) {
+			mlx5dr_dbg(dmn, "Unsupported field %d for TX/FDB set action\n",
+				   sw_field);
+			return -EINVAL;
+		}
+	}
+
+	if (!action->rewrite->allow_rx && !action->rewrite->allow_tx) {
+		mlx5dr_dbg(dmn, "Modify SET actions not supported on both RX and TX\n");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int
+dr_action_modify_check_add_field_limitation(struct mlx5dr_action *action,
+					    const __be64 *sw_action)
+{
+	u16 sw_field = MLX5_GET(set_action_in, sw_action, field);
+	struct mlx5dr_domain *dmn = action->rewrite->dmn;
+
+	if (sw_field != MLX5_ACTION_IN_FIELD_OUT_IP_TTL &&
+	    sw_field != MLX5_ACTION_IN_FIELD_OUT_IPV6_HOPLIMIT &&
+	    sw_field != MLX5_ACTION_IN_FIELD_OUT_TCP_SEQ_NUM &&
+	    sw_field != MLX5_ACTION_IN_FIELD_OUT_TCP_ACK_NUM) {
+		mlx5dr_dbg(dmn, "Unsupported field %d for add action\n",
+			   sw_field);
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int
+dr_action_modify_check_copy_field_limitation(struct mlx5dr_action *action,
+					     const __be64 *sw_action)
+{
+	struct mlx5dr_domain *dmn = action->rewrite->dmn;
+	u16 sw_fields[2];
+	int i;
+
+	sw_fields[0] = MLX5_GET(copy_action_in, sw_action, src_field);
+	sw_fields[1] = MLX5_GET(copy_action_in, sw_action, dst_field);
+
+	for (i = 0; i < 2; i++) {
+		if (sw_fields[i] == MLX5_ACTION_IN_FIELD_METADATA_REG_A) {
+			action->rewrite->allow_rx = 0;
+			if (dmn->type != MLX5DR_DOMAIN_TYPE_NIC_TX) {
+				mlx5dr_dbg(dmn, "Unsupported field %d for RX/FDB set action\n",
+					   sw_fields[i]);
+				return -EINVAL;
+			}
+		} else if (sw_fields[i] == MLX5_ACTION_IN_FIELD_METADATA_REG_B) {
+			action->rewrite->allow_tx = 0;
+			if (dmn->type != MLX5DR_DOMAIN_TYPE_NIC_RX) {
+				mlx5dr_dbg(dmn, "Unsupported field %d for TX/FDB set action\n",
+					   sw_fields[i]);
+				return -EINVAL;
+			}
+		}
+	}
+
+	if (!action->rewrite->allow_rx && !action->rewrite->allow_tx) {
+		mlx5dr_dbg(dmn, "Modify copy actions not supported on both RX and TX\n");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int
+dr_action_modify_check_field_limitation(struct mlx5dr_action *action,
+					const __be64 *sw_action)
+{
+	struct mlx5dr_domain *dmn = action->rewrite->dmn;
+	u8 action_type;
+	int ret;
+
+	action_type = MLX5_GET(set_action_in, sw_action, action_type);
+
+	switch (action_type) {
+	case MLX5_ACTION_TYPE_SET:
+		ret = dr_action_modify_check_set_field_limitation(action,
+								  sw_action);
+		break;
+
+	case MLX5_ACTION_TYPE_ADD:
+		ret = dr_action_modify_check_add_field_limitation(action,
+								  sw_action);
+		break;
+
+	case MLX5_ACTION_TYPE_COPY:
+		ret = dr_action_modify_check_copy_field_limitation(action,
+								   sw_action);
+		break;
+
+	default:
+		mlx5dr_info(dmn, "Unsupported action %d modify action\n",
+			    action_type);
+		ret = -EOPNOTSUPP;
+	}
+
+	return ret;
+}
+
+static bool
+dr_action_modify_check_is_ttl_modify(const void *sw_action)
+{
+	u16 sw_field = MLX5_GET(set_action_in, sw_action, field);
+
+	return sw_field == MLX5_ACTION_IN_FIELD_OUT_IP_TTL;
+}
+
+static int dr_actions_convert_modify_header(struct mlx5dr_action *action,
+					    u32 max_hw_actions,
+					    u32 num_sw_actions,
+					    __be64 sw_actions[],
+					    __be64 hw_actions[],
+					    u32 *num_hw_actions,
+					    bool *modify_ttl)
+{
+	const struct mlx5dr_ste_action_modify_field *hw_dst_action_info;
+	const struct mlx5dr_ste_action_modify_field *hw_src_action_info;
+	struct mlx5dr_domain *dmn = action->rewrite->dmn;
+	__be64 *modify_ttl_sw_action = NULL;
+	int ret, i, hw_idx = 0;
+	__be64 *sw_action;
+	__be64 hw_action;
+	u16 hw_field = 0;
+	u32 l3_type = 0;
+	u32 l4_type = 0;
+
+	*modify_ttl = false;
+
+	action->rewrite->allow_rx = 1;
+	action->rewrite->allow_tx = 1;
+
+	for (i = 0; i < num_sw_actions || modify_ttl_sw_action; i++) {
+		/* modify TTL is handled separately, as a last action */
+		if (i == num_sw_actions) {
+			sw_action = modify_ttl_sw_action;
+			modify_ttl_sw_action = NULL;
+		} else {
+			sw_action = &sw_actions[i];
+		}
+
+		ret = dr_action_modify_check_field_limitation(action,
+							      sw_action);
+		if (ret)
+			return ret;
+
+		if (!(*modify_ttl) &&
+		    dr_action_modify_check_is_ttl_modify(sw_action)) {
+			modify_ttl_sw_action = sw_action;
+			*modify_ttl = true;
+			continue;
+		}
+
+		/* Convert SW action to HW action */
+		ret = dr_action_modify_sw_to_hw(dmn,
+						sw_action,
+						&hw_action,
+						&hw_dst_action_info,
+						&hw_src_action_info);
+		if (ret)
+			return ret;
+
+		/* Due to a HW limitation we cannot modify 2 different L3 types */
+		if (l3_type && hw_dst_action_info->l3_type &&
+		    hw_dst_action_info->l3_type != l3_type) {
+			mlx5dr_dbg(dmn, "Action list can't support two different L3 types\n");
+			return -EINVAL;
+		}
+		if (hw_dst_action_info->l3_type)
+			l3_type = hw_dst_action_info->l3_type;
+
+		/* Due to a HW limitation we cannot modify two different L4 types */
+		if (l4_type && hw_dst_action_info->l4_type &&
+		    hw_dst_action_info->l4_type != l4_type) {
+			mlx5dr_dbg(dmn, "Action list can't support two different L4 types\n");
+			return -EINVAL;
+		}
+		if (hw_dst_action_info->l4_type)
+			l4_type = hw_dst_action_info->l4_type;
+
+		/* HW reads and executes two actions at once this means we
+		 * need to create a gap if two actions access the same field
+		 */
+		if ((hw_idx % 2) && (hw_field == hw_dst_action_info->hw_field ||
+				     (hw_src_action_info &&
+				      hw_field == hw_src_action_info->hw_field))) {
+			/* Check if after gap insertion the total number of HW
+			 * modify actions doesn't exceeds the limit
+			 */
+			hw_idx++;
+			if (hw_idx >= max_hw_actions) {
+				mlx5dr_dbg(dmn, "Modify header action number exceeds HW limit\n");
+				return -EINVAL;
+			}
+		}
+		hw_field = hw_dst_action_info->hw_field;
+
+		hw_actions[hw_idx] = hw_action;
+		hw_idx++;
+	}
+
+	/* if the resulting HW actions list is empty, add NOP action */
+	if (!hw_idx)
+		hw_idx++;
+
+	*num_hw_actions = hw_idx;
+
+	return 0;
+}
+
+static int dr_action_create_modify_action(struct mlx5dr_domain *dmn,
+					  size_t actions_sz,
+					  __be64 actions[],
+					  struct mlx5dr_action *action)
+{
+	struct mlx5dr_icm_chunk *chunk;
+	u32 max_hw_actions;
+	u32 num_hw_actions;
+	u32 num_sw_actions;
+	__be64 *hw_actions;
+	bool modify_ttl;
+	int ret;
+
+	num_sw_actions = actions_sz / DR_MODIFY_ACTION_SIZE;
+	max_hw_actions = mlx5dr_icm_pool_chunk_size_to_entries(DR_CHUNK_SIZE_16);
+
+	if (num_sw_actions > max_hw_actions) {
+		mlx5dr_dbg(dmn, "Max number of actions %d exceeds limit %d\n",
+			   num_sw_actions, max_hw_actions);
+		return -EINVAL;
+	}
+
+	chunk = mlx5dr_icm_alloc_chunk(dmn->action_icm_pool, DR_CHUNK_SIZE_16);
+	if (!chunk)
+		return -ENOMEM;
+
+	hw_actions = kcalloc(1, max_hw_actions * DR_MODIFY_ACTION_SIZE, GFP_KERNEL);
+	if (!hw_actions) {
+		ret = -ENOMEM;
+		goto free_chunk;
+	}
+
+	ret = dr_actions_convert_modify_header(action,
+					       max_hw_actions,
+					       num_sw_actions,
+					       actions,
+					       hw_actions,
+					       &num_hw_actions,
+					       &modify_ttl);
+	if (ret)
+		goto free_hw_actions;
+
+	action->rewrite->chunk = chunk;
+	action->rewrite->modify_ttl = modify_ttl;
+	action->rewrite->data = (u8 *)hw_actions;
+	action->rewrite->num_of_actions = num_hw_actions;
+	action->rewrite->index = (mlx5dr_icm_pool_get_chunk_icm_addr(chunk) -
+				  dmn->info.caps.hdr_modify_icm_addr) /
+				  ACTION_CACHE_LINE_SIZE;
+
+	ret = mlx5dr_send_postsend_action(dmn, action);
+	if (ret)
+		goto free_hw_actions;
+
+	return 0;
+
+free_hw_actions:
+	kfree(hw_actions);
+free_chunk:
+	mlx5dr_icm_free_chunk(chunk);
+	return ret;
+}
+
+struct mlx5dr_action *
+mlx5dr_action_create_modify_header(struct mlx5dr_domain *dmn,
+				   u32 flags,
+				   size_t actions_sz,
+				   __be64 actions[])
+{
+	struct mlx5dr_action *action;
+	int ret = 0;
+
+	refcount_inc(&dmn->refcount);
+
+	if (actions_sz % DR_MODIFY_ACTION_SIZE) {
+		mlx5dr_dbg(dmn, "Invalid modify actions size provided\n");
+		goto dec_ref;
+	}
+
+	action = dr_action_create_generic(DR_ACTION_TYP_MODIFY_HDR);
+	if (!action)
+		goto dec_ref;
+
+	action->rewrite->dmn = dmn;
+
+	ret = dr_action_create_modify_action(dmn,
+					     actions_sz,
+					     actions,
+					     action);
+	if (ret) {
+		mlx5dr_dbg(dmn, "Failed creating modify header action %d\n", ret);
+		goto free_action;
+	}
+
+	return action;
+
+free_action:
+	kfree(action);
+dec_ref:
+	refcount_dec(&dmn->refcount);
+	return NULL;
+}
+
+struct mlx5dr_action *
+mlx5dr_action_create_dest_vport(struct mlx5dr_domain *dmn,
+				u16 vport, u8 vhca_id_valid,
+				u16 vhca_id)
+{
+	struct mlx5dr_cmd_vport_cap *vport_cap;
+	struct mlx5dr_domain *vport_dmn;
+	struct mlx5dr_action *action;
+	u8 peer_vport;
+
+	peer_vport = vhca_id_valid && (vhca_id != dmn->info.caps.gvmi);
+	vport_dmn = peer_vport ? dmn->peer_dmn : dmn;
+	if (!vport_dmn) {
+		mlx5dr_dbg(dmn, "No peer vport domain for given vhca_id\n");
+		return NULL;
+	}
+
+	if (vport_dmn->type != MLX5DR_DOMAIN_TYPE_FDB) {
+		mlx5dr_dbg(dmn, "Domain doesn't support vport actions\n");
+		return NULL;
+	}
+
+	vport_cap = mlx5dr_domain_get_vport_cap(vport_dmn, vport);
+	if (!vport_cap) {
+		mlx5dr_err(dmn,
+			   "Failed to get vport 0x%x caps - vport is disabled or invalid\n",
+			   vport);
+		return NULL;
+	}
+
+	action = dr_action_create_generic(DR_ACTION_TYP_VPORT);
+	if (!action)
+		return NULL;
+
+	action->vport->dmn = vport_dmn;
+	action->vport->caps = vport_cap;
+
+	return action;
+}
+
+struct mlx5dr_action *
+mlx5dr_action_create_aso(struct mlx5dr_domain *dmn, u32 obj_id,
+			 u8 dest_reg_id, u8 aso_type,
+			 u8 init_color, u8 meter_id)
+{
+	struct mlx5dr_action *action;
+
+	if (aso_type != MLX5_EXE_ASO_FLOW_METER)
+		return NULL;
+
+	if (init_color > MLX5_FLOW_METER_COLOR_UNDEFINED)
+		return NULL;
+
+	action = dr_action_create_generic(DR_ACTION_TYP_ASO_FLOW_METER);
+	if (!action)
+		return NULL;
+
+	action->aso->obj_id = obj_id;
+	action->aso->offset = meter_id;
+	action->aso->dest_reg_id = dest_reg_id;
+	action->aso->init_color = init_color;
+	action->aso->dmn = dmn;
+
+	refcount_inc(&dmn->refcount);
+
+	return action;
+}
+
+int mlx5dr_action_destroy(struct mlx5dr_action *action)
+{
+	if (WARN_ON_ONCE(refcount_read(&action->refcount) > 1))
+		return -EBUSY;
+
+	switch (action->action_type) {
+	case DR_ACTION_TYP_FT:
+		if (action->dest_tbl->is_fw_tbl)
+			refcount_dec(&action->dest_tbl->fw_tbl.dmn->refcount);
+		else
+			refcount_dec(&action->dest_tbl->tbl->refcount);
+
+		if (action->dest_tbl->is_fw_tbl &&
+		    action->dest_tbl->fw_tbl.num_of_ref_actions) {
+			struct mlx5dr_action **ref_actions;
+			int i;
+
+			ref_actions = action->dest_tbl->fw_tbl.ref_actions;
+			for (i = 0; i < action->dest_tbl->fw_tbl.num_of_ref_actions; i++)
+				refcount_dec(&ref_actions[i]->refcount);
+
+			kfree(ref_actions);
+
+			mlx5dr_fw_destroy_md_tbl(action->dest_tbl->fw_tbl.dmn,
+						 action->dest_tbl->fw_tbl.id,
+						 action->dest_tbl->fw_tbl.group_id);
+		}
+		break;
+	case DR_ACTION_TYP_TNL_L2_TO_L2:
+	case DR_ACTION_TYP_REMOVE_HDR:
+		refcount_dec(&action->reformat->dmn->refcount);
+		break;
+	case DR_ACTION_TYP_TNL_L3_TO_L2:
+		mlx5dr_icm_free_chunk(action->rewrite->chunk);
+		refcount_dec(&action->rewrite->dmn->refcount);
+		break;
+	case DR_ACTION_TYP_L2_TO_TNL_L2:
+	case DR_ACTION_TYP_L2_TO_TNL_L3:
+	case DR_ACTION_TYP_INSERT_HDR:
+		mlx5dr_cmd_destroy_reformat_ctx((action->reformat->dmn)->mdev,
+						action->reformat->id);
+		refcount_dec(&action->reformat->dmn->refcount);
+		break;
+	case DR_ACTION_TYP_MODIFY_HDR:
+		mlx5dr_icm_free_chunk(action->rewrite->chunk);
+		kfree(action->rewrite->data);
+		refcount_dec(&action->rewrite->dmn->refcount);
+		break;
+	case DR_ACTION_TYP_SAMPLER:
+		refcount_dec(&action->sampler->dmn->refcount);
+		break;
+	case DR_ACTION_TYP_ASO_FLOW_METER:
+		refcount_dec(&action->aso->dmn->refcount);
+		break;
+	default:
+		break;
+	}
+
+	kfree(action);
+	return 0;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_buddy.c b/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_buddy.c
new file mode 100644
index 000000000..7df11a019
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_buddy.c
@@ -0,0 +1,170 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2004 Topspin Communications. All rights reserved.
+ * Copyright (c) 2005 - 2008 Mellanox Technologies. All rights reserved.
+ * Copyright (c) 2006 - 2007 Cisco Systems, Inc. All rights reserved.
+ * Copyright (c) 2020 NVIDIA CORPORATION. All rights reserved.
+ */
+
+#include "dr_types.h"
+
+int mlx5dr_buddy_init(struct mlx5dr_icm_buddy_mem *buddy,
+		      unsigned int max_order)
+{
+	int i;
+
+	buddy->max_order = max_order;
+
+	INIT_LIST_HEAD(&buddy->list_node);
+	INIT_LIST_HEAD(&buddy->used_list);
+	INIT_LIST_HEAD(&buddy->hot_list);
+
+	buddy->bitmap = kcalloc(buddy->max_order + 1,
+				sizeof(*buddy->bitmap),
+				GFP_KERNEL);
+	buddy->num_free = kcalloc(buddy->max_order + 1,
+				  sizeof(*buddy->num_free),
+				  GFP_KERNEL);
+
+	if (!buddy->bitmap || !buddy->num_free)
+		goto err_free_all;
+
+	/* Allocating max_order bitmaps, one for each order */
+
+	for (i = 0; i <= buddy->max_order; ++i) {
+		unsigned int size = 1 << (buddy->max_order - i);
+
+		buddy->bitmap[i] = bitmap_zalloc(size, GFP_KERNEL);
+		if (!buddy->bitmap[i])
+			goto err_out_free_each_bit_per_order;
+	}
+
+	/* In the beginning, we have only one order that is available for
+	 * use (the biggest one), so mark the first bit in both bitmaps.
+	 */
+
+	bitmap_set(buddy->bitmap[buddy->max_order], 0, 1);
+
+	buddy->num_free[buddy->max_order] = 1;
+
+	return 0;
+
+err_out_free_each_bit_per_order:
+	for (i = 0; i <= buddy->max_order; ++i)
+		bitmap_free(buddy->bitmap[i]);
+
+err_free_all:
+	kfree(buddy->num_free);
+	kfree(buddy->bitmap);
+	return -ENOMEM;
+}
+
+void mlx5dr_buddy_cleanup(struct mlx5dr_icm_buddy_mem *buddy)
+{
+	int i;
+
+	list_del(&buddy->list_node);
+
+	for (i = 0; i <= buddy->max_order; ++i)
+		bitmap_free(buddy->bitmap[i]);
+
+	kfree(buddy->num_free);
+	kfree(buddy->bitmap);
+}
+
+static int dr_buddy_find_free_seg(struct mlx5dr_icm_buddy_mem *buddy,
+				  unsigned int start_order,
+				  unsigned int *segment,
+				  unsigned int *order)
+{
+	unsigned int seg, order_iter, m;
+
+	for (order_iter = start_order;
+	     order_iter <= buddy->max_order; ++order_iter) {
+		if (!buddy->num_free[order_iter])
+			continue;
+
+		m = 1 << (buddy->max_order - order_iter);
+		seg = find_first_bit(buddy->bitmap[order_iter], m);
+
+		if (WARN(seg >= m,
+			 "ICM Buddy: failed finding free mem for order %d\n",
+			 order_iter))
+			return -ENOMEM;
+
+		break;
+	}
+
+	if (order_iter > buddy->max_order)
+		return -ENOMEM;
+
+	*segment = seg;
+	*order = order_iter;
+	return 0;
+}
+
+/**
+ * mlx5dr_buddy_alloc_mem() - Update second level bitmap.
+ * @buddy: Buddy to update.
+ * @order: Order of the buddy to update.
+ * @segment: Segment number.
+ *
+ * This function finds the first area of the ICM memory managed by this buddy.
+ * It uses the data structures of the buddy system in order to find the first
+ * area of free place, starting from the current order till the maximum order
+ * in the system.
+ *
+ * Return: 0 when segment is set, non-zero error status otherwise.
+ *
+ * The function returns the location (segment) in the whole buddy ICM memory
+ * area - the index of the memory segment that is available for use.
+ */
+int mlx5dr_buddy_alloc_mem(struct mlx5dr_icm_buddy_mem *buddy,
+			   unsigned int order,
+			   unsigned int *segment)
+{
+	unsigned int seg, order_iter;
+	int err;
+
+	err = dr_buddy_find_free_seg(buddy, order, &seg, &order_iter);
+	if (err)
+		return err;
+
+	bitmap_clear(buddy->bitmap[order_iter], seg, 1);
+	--buddy->num_free[order_iter];
+
+	/* If we found free memory in some order that is bigger than the
+	 * required order, we need to split every order between the required
+	 * order and the order that we found into two parts, and mark accordingly.
+	 */
+	while (order_iter > order) {
+		--order_iter;
+		seg <<= 1;
+		bitmap_set(buddy->bitmap[order_iter], seg ^ 1, 1);
+		++buddy->num_free[order_iter];
+	}
+
+	seg <<= order;
+	*segment = seg;
+
+	return 0;
+}
+
+void mlx5dr_buddy_free_mem(struct mlx5dr_icm_buddy_mem *buddy,
+			   unsigned int seg, unsigned int order)
+{
+	seg >>= order;
+
+	/* Whenever a segment is free,
+	 * the mem is added to the buddy that gave it.
+	 */
+	while (test_bit(seg ^ 1, buddy->bitmap[order])) {
+		bitmap_clear(buddy->bitmap[order], seg ^ 1, 1);
+		--buddy->num_free[order];
+		seg >>= 1;
+		++order;
+	}
+	bitmap_set(buddy->bitmap[order], seg, 1);
+
+	++buddy->num_free[order];
+}
+
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_cmd.c b/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_cmd.c
new file mode 100644
index 000000000..d7b1a230b
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_cmd.c
@@ -0,0 +1,824 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2019 Mellanox Technologies. */
+
+#include "dr_types.h"
+
+int mlx5dr_cmd_query_esw_vport_context(struct mlx5_core_dev *mdev,
+				       bool other_vport,
+				       u16 vport_number,
+				       u64 *icm_address_rx,
+				       u64 *icm_address_tx)
+{
+	u32 out[MLX5_ST_SZ_DW(query_esw_vport_context_out)] = {};
+	u32 in[MLX5_ST_SZ_DW(query_esw_vport_context_in)] = {};
+	int err;
+
+	MLX5_SET(query_esw_vport_context_in, in, opcode,
+		 MLX5_CMD_OP_QUERY_ESW_VPORT_CONTEXT);
+	MLX5_SET(query_esw_vport_context_in, in, other_vport, other_vport);
+	MLX5_SET(query_esw_vport_context_in, in, vport_number, vport_number);
+
+	err = mlx5_cmd_exec_inout(mdev, query_esw_vport_context, in, out);
+	if (err)
+		return err;
+
+	*icm_address_rx =
+		MLX5_GET64(query_esw_vport_context_out, out,
+			   esw_vport_context.sw_steering_vport_icm_address_rx);
+	*icm_address_tx =
+		MLX5_GET64(query_esw_vport_context_out, out,
+			   esw_vport_context.sw_steering_vport_icm_address_tx);
+	return 0;
+}
+
+int mlx5dr_cmd_query_gvmi(struct mlx5_core_dev *mdev, bool other_vport,
+			  u16 vport_number, u16 *gvmi)
+{
+	u32 in[MLX5_ST_SZ_DW(query_hca_cap_in)] = {};
+	int out_size;
+	void *out;
+	int err;
+
+	out_size = MLX5_ST_SZ_BYTES(query_hca_cap_out);
+	out = kzalloc(out_size, GFP_KERNEL);
+	if (!out)
+		return -ENOMEM;
+
+	MLX5_SET(query_hca_cap_in, in, opcode, MLX5_CMD_OP_QUERY_HCA_CAP);
+	MLX5_SET(query_hca_cap_in, in, other_function, other_vport);
+	MLX5_SET(query_hca_cap_in, in, function_id, vport_number);
+	MLX5_SET(query_hca_cap_in, in, op_mod,
+		 MLX5_SET_HCA_CAP_OP_MOD_GENERAL_DEVICE << 1 |
+		 HCA_CAP_OPMOD_GET_CUR);
+
+	err = mlx5_cmd_exec_inout(mdev, query_hca_cap, in, out);
+	if (err) {
+		kfree(out);
+		return err;
+	}
+
+	*gvmi = MLX5_GET(query_hca_cap_out, out, capability.cmd_hca_cap.vhca_id);
+
+	kfree(out);
+	return 0;
+}
+
+int mlx5dr_cmd_query_esw_caps(struct mlx5_core_dev *mdev,
+			      struct mlx5dr_esw_caps *caps)
+{
+	caps->drop_icm_address_rx =
+		MLX5_CAP64_ESW_FLOWTABLE(mdev,
+					 sw_steering_fdb_action_drop_icm_address_rx);
+	caps->drop_icm_address_tx =
+		MLX5_CAP64_ESW_FLOWTABLE(mdev,
+					 sw_steering_fdb_action_drop_icm_address_tx);
+	caps->uplink_icm_address_rx =
+		MLX5_CAP64_ESW_FLOWTABLE(mdev,
+					 sw_steering_uplink_icm_address_rx);
+	caps->uplink_icm_address_tx =
+		MLX5_CAP64_ESW_FLOWTABLE(mdev,
+					 sw_steering_uplink_icm_address_tx);
+	caps->sw_owner_v2 = MLX5_CAP_ESW_FLOWTABLE_FDB(mdev, sw_owner_v2);
+	if (!caps->sw_owner_v2)
+		caps->sw_owner = MLX5_CAP_ESW_FLOWTABLE_FDB(mdev, sw_owner);
+
+	return 0;
+}
+
+static int dr_cmd_query_nic_vport_roce_en(struct mlx5_core_dev *mdev,
+					  u16 vport, bool *roce_en)
+{
+	u32 out[MLX5_ST_SZ_DW(query_nic_vport_context_out)] = {};
+	u32 in[MLX5_ST_SZ_DW(query_nic_vport_context_in)] = {};
+	int err;
+
+	MLX5_SET(query_nic_vport_context_in, in, opcode,
+		 MLX5_CMD_OP_QUERY_NIC_VPORT_CONTEXT);
+	MLX5_SET(query_nic_vport_context_in, in, vport_number, vport);
+	MLX5_SET(query_nic_vport_context_in, in, other_vport, !!vport);
+
+	err = mlx5_cmd_exec(mdev, in, sizeof(in), out, sizeof(out));
+	if (err)
+		return err;
+
+	*roce_en = MLX5_GET(query_nic_vport_context_out, out,
+			    nic_vport_context.roce_en);
+	return 0;
+}
+
+int mlx5dr_cmd_query_device(struct mlx5_core_dev *mdev,
+			    struct mlx5dr_cmd_caps *caps)
+{
+	bool roce_en;
+	int err;
+
+	caps->prio_tag_required	= MLX5_CAP_GEN(mdev, prio_tag_required);
+	caps->eswitch_manager	= MLX5_CAP_GEN(mdev, eswitch_manager);
+	caps->gvmi		= MLX5_CAP_GEN(mdev, vhca_id);
+	caps->flex_protocols	= MLX5_CAP_GEN(mdev, flex_parser_protocols);
+	caps->sw_format_ver	= MLX5_CAP_GEN(mdev, steering_format_version);
+	caps->roce_caps.fl_rc_qp_when_roce_disabled =
+		MLX5_CAP_GEN(mdev, fl_rc_qp_when_roce_disabled);
+
+	if (MLX5_CAP_GEN(mdev, roce)) {
+		err = dr_cmd_query_nic_vport_roce_en(mdev, 0, &roce_en);
+		if (err)
+			return err;
+
+		caps->roce_caps.roce_en = roce_en;
+		caps->roce_caps.fl_rc_qp_when_roce_disabled |=
+			MLX5_CAP_ROCE(mdev, fl_rc_qp_when_roce_disabled);
+		caps->roce_caps.fl_rc_qp_when_roce_enabled =
+			MLX5_CAP_ROCE(mdev, fl_rc_qp_when_roce_enabled);
+	}
+
+	caps->isolate_vl_tc = MLX5_CAP_GEN(mdev, isolate_vl_tc_new);
+
+	/* geneve_tlv_option_0_exist is the indication of
+	 * STE support for lookup type flex_parser_ok
+	 */
+	caps->flex_parser_ok_bits_supp =
+		MLX5_CAP_FLOWTABLE(mdev,
+				   flow_table_properties_nic_receive.ft_field_support.geneve_tlv_option_0_exist);
+
+	if (caps->flex_protocols & MLX5_FLEX_PARSER_ICMP_V4_ENABLED) {
+		caps->flex_parser_id_icmp_dw0 = MLX5_CAP_GEN(mdev, flex_parser_id_icmp_dw0);
+		caps->flex_parser_id_icmp_dw1 = MLX5_CAP_GEN(mdev, flex_parser_id_icmp_dw1);
+	}
+
+	if (caps->flex_protocols & MLX5_FLEX_PARSER_ICMP_V6_ENABLED) {
+		caps->flex_parser_id_icmpv6_dw0 =
+			MLX5_CAP_GEN(mdev, flex_parser_id_icmpv6_dw0);
+		caps->flex_parser_id_icmpv6_dw1 =
+			MLX5_CAP_GEN(mdev, flex_parser_id_icmpv6_dw1);
+	}
+
+	if (caps->flex_protocols & MLX5_FLEX_PARSER_GENEVE_TLV_OPTION_0_ENABLED)
+		caps->flex_parser_id_geneve_tlv_option_0 =
+			MLX5_CAP_GEN(mdev, flex_parser_id_geneve_tlv_option_0);
+
+	if (caps->flex_protocols & MLX5_FLEX_PARSER_MPLS_OVER_GRE_ENABLED)
+		caps->flex_parser_id_mpls_over_gre =
+			MLX5_CAP_GEN(mdev, flex_parser_id_outer_first_mpls_over_gre);
+
+	if (caps->flex_protocols & MLX5_FLEX_PARSER_MPLS_OVER_UDP_ENABLED)
+		caps->flex_parser_id_mpls_over_udp =
+			MLX5_CAP_GEN(mdev, flex_parser_id_outer_first_mpls_over_udp_label);
+
+	if (caps->flex_protocols & MLX5_FLEX_PARSER_GTPU_DW_0_ENABLED)
+		caps->flex_parser_id_gtpu_dw_0 =
+			MLX5_CAP_GEN(mdev, flex_parser_id_gtpu_dw_0);
+
+	if (caps->flex_protocols & MLX5_FLEX_PARSER_GTPU_TEID_ENABLED)
+		caps->flex_parser_id_gtpu_teid =
+			MLX5_CAP_GEN(mdev, flex_parser_id_gtpu_teid);
+
+	if (caps->flex_protocols & MLX5_FLEX_PARSER_GTPU_DW_2_ENABLED)
+		caps->flex_parser_id_gtpu_dw_2 =
+			MLX5_CAP_GEN(mdev, flex_parser_id_gtpu_dw_2);
+
+	if (caps->flex_protocols & MLX5_FLEX_PARSER_GTPU_FIRST_EXT_DW_0_ENABLED)
+		caps->flex_parser_id_gtpu_first_ext_dw_0 =
+			MLX5_CAP_GEN(mdev, flex_parser_id_gtpu_first_ext_dw_0);
+
+	caps->nic_rx_drop_address =
+		MLX5_CAP64_FLOWTABLE(mdev, sw_steering_nic_rx_action_drop_icm_address);
+	caps->nic_tx_drop_address =
+		MLX5_CAP64_FLOWTABLE(mdev, sw_steering_nic_tx_action_drop_icm_address);
+	caps->nic_tx_allow_address =
+		MLX5_CAP64_FLOWTABLE(mdev, sw_steering_nic_tx_action_allow_icm_address);
+
+	caps->rx_sw_owner_v2 = MLX5_CAP_FLOWTABLE_NIC_RX(mdev, sw_owner_v2);
+	caps->tx_sw_owner_v2 = MLX5_CAP_FLOWTABLE_NIC_TX(mdev, sw_owner_v2);
+
+	if (!caps->rx_sw_owner_v2)
+		caps->rx_sw_owner = MLX5_CAP_FLOWTABLE_NIC_RX(mdev, sw_owner);
+	if (!caps->tx_sw_owner_v2)
+		caps->tx_sw_owner = MLX5_CAP_FLOWTABLE_NIC_TX(mdev, sw_owner);
+
+	caps->max_ft_level = MLX5_CAP_FLOWTABLE_NIC_RX(mdev, max_ft_level);
+
+	caps->log_icm_size = MLX5_CAP_DEV_MEM(mdev, log_steering_sw_icm_size);
+	caps->hdr_modify_icm_addr =
+		MLX5_CAP64_DEV_MEM(mdev, header_modify_sw_icm_start_address);
+
+	caps->roce_min_src_udp = MLX5_CAP_ROCE(mdev, r_roce_min_src_udp_port);
+
+	caps->is_ecpf = mlx5_core_is_ecpf_esw_manager(mdev);
+
+	return 0;
+}
+
+int mlx5dr_cmd_query_flow_table(struct mlx5_core_dev *dev,
+				enum fs_flow_table_type type,
+				u32 table_id,
+				struct mlx5dr_cmd_query_flow_table_details *output)
+{
+	u32 out[MLX5_ST_SZ_DW(query_flow_table_out)] = {};
+	u32 in[MLX5_ST_SZ_DW(query_flow_table_in)] = {};
+	int err;
+
+	MLX5_SET(query_flow_table_in, in, opcode,
+		 MLX5_CMD_OP_QUERY_FLOW_TABLE);
+
+	MLX5_SET(query_flow_table_in, in, table_type, type);
+	MLX5_SET(query_flow_table_in, in, table_id, table_id);
+
+	err = mlx5_cmd_exec_inout(dev, query_flow_table, in, out);
+	if (err)
+		return err;
+
+	output->status = MLX5_GET(query_flow_table_out, out, status);
+	output->level = MLX5_GET(query_flow_table_out, out, flow_table_context.level);
+
+	output->sw_owner_icm_root_1 = MLX5_GET64(query_flow_table_out, out,
+						 flow_table_context.sw_owner_icm_root_1);
+	output->sw_owner_icm_root_0 = MLX5_GET64(query_flow_table_out, out,
+						 flow_table_context.sw_owner_icm_root_0);
+
+	return 0;
+}
+
+int mlx5dr_cmd_query_flow_sampler(struct mlx5_core_dev *dev,
+				  u32 sampler_id,
+				  u64 *rx_icm_addr,
+				  u64 *tx_icm_addr)
+{
+	u32 out[MLX5_ST_SZ_DW(query_sampler_obj_out)] = {};
+	u32 in[MLX5_ST_SZ_DW(general_obj_in_cmd_hdr)] = {};
+	void *attr;
+	int ret;
+
+	MLX5_SET(general_obj_in_cmd_hdr, in, opcode,
+		 MLX5_CMD_OP_QUERY_GENERAL_OBJECT);
+	MLX5_SET(general_obj_in_cmd_hdr, in, obj_type,
+		 MLX5_GENERAL_OBJECT_TYPES_SAMPLER);
+	MLX5_SET(general_obj_in_cmd_hdr, in, obj_id, sampler_id);
+
+	ret = mlx5_cmd_exec(dev, in, sizeof(in), out, sizeof(out));
+	if (ret)
+		return ret;
+
+	attr = MLX5_ADDR_OF(query_sampler_obj_out, out, sampler_object);
+
+	*rx_icm_addr = MLX5_GET64(sampler_obj, attr,
+				  sw_steering_icm_address_rx);
+	*tx_icm_addr = MLX5_GET64(sampler_obj, attr,
+				  sw_steering_icm_address_tx);
+
+	return 0;
+}
+
+int mlx5dr_cmd_sync_steering(struct mlx5_core_dev *mdev)
+{
+	u32 in[MLX5_ST_SZ_DW(sync_steering_in)] = {};
+
+	MLX5_SET(sync_steering_in, in, opcode, MLX5_CMD_OP_SYNC_STEERING);
+
+	return mlx5_cmd_exec_in(mdev, sync_steering, in);
+}
+
+int mlx5dr_cmd_set_fte_modify_and_vport(struct mlx5_core_dev *mdev,
+					u32 table_type,
+					u32 table_id,
+					u32 group_id,
+					u32 modify_header_id,
+					u16 vport)
+{
+	u32 out[MLX5_ST_SZ_DW(set_fte_out)] = {};
+	void *in_flow_context;
+	unsigned int inlen;
+	void *in_dests;
+	u32 *in;
+	int err;
+
+	inlen = MLX5_ST_SZ_BYTES(set_fte_in) +
+		1 * MLX5_ST_SZ_BYTES(dest_format_struct); /* One destination only */
+
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in)
+		return -ENOMEM;
+
+	MLX5_SET(set_fte_in, in, opcode, MLX5_CMD_OP_SET_FLOW_TABLE_ENTRY);
+	MLX5_SET(set_fte_in, in, table_type, table_type);
+	MLX5_SET(set_fte_in, in, table_id, table_id);
+
+	in_flow_context = MLX5_ADDR_OF(set_fte_in, in, flow_context);
+	MLX5_SET(flow_context, in_flow_context, group_id, group_id);
+	MLX5_SET(flow_context, in_flow_context, modify_header_id, modify_header_id);
+	MLX5_SET(flow_context, in_flow_context, destination_list_size, 1);
+	MLX5_SET(flow_context, in_flow_context, action,
+		 MLX5_FLOW_CONTEXT_ACTION_FWD_DEST |
+		 MLX5_FLOW_CONTEXT_ACTION_MOD_HDR);
+
+	in_dests = MLX5_ADDR_OF(flow_context, in_flow_context, destination);
+	MLX5_SET(dest_format_struct, in_dests, destination_type,
+		 MLX5_IFC_FLOW_DESTINATION_TYPE_VPORT);
+	MLX5_SET(dest_format_struct, in_dests, destination_id, vport);
+
+	err = mlx5_cmd_exec(mdev, in, inlen, out, sizeof(out));
+	kvfree(in);
+
+	return err;
+}
+
+int mlx5dr_cmd_del_flow_table_entry(struct mlx5_core_dev *mdev,
+				    u32 table_type,
+				    u32 table_id)
+{
+	u32 in[MLX5_ST_SZ_DW(delete_fte_in)] = {};
+
+	MLX5_SET(delete_fte_in, in, opcode, MLX5_CMD_OP_DELETE_FLOW_TABLE_ENTRY);
+	MLX5_SET(delete_fte_in, in, table_type, table_type);
+	MLX5_SET(delete_fte_in, in, table_id, table_id);
+
+	return mlx5_cmd_exec_in(mdev, delete_fte, in);
+}
+
+int mlx5dr_cmd_alloc_modify_header(struct mlx5_core_dev *mdev,
+				   u32 table_type,
+				   u8 num_of_actions,
+				   u64 *actions,
+				   u32 *modify_header_id)
+{
+	u32 out[MLX5_ST_SZ_DW(alloc_modify_header_context_out)] = {};
+	void *p_actions;
+	u32 inlen;
+	u32 *in;
+	int err;
+
+	inlen = MLX5_ST_SZ_BYTES(alloc_modify_header_context_in) +
+		 num_of_actions * sizeof(u64);
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in)
+		return -ENOMEM;
+
+	MLX5_SET(alloc_modify_header_context_in, in, opcode,
+		 MLX5_CMD_OP_ALLOC_MODIFY_HEADER_CONTEXT);
+	MLX5_SET(alloc_modify_header_context_in, in, table_type, table_type);
+	MLX5_SET(alloc_modify_header_context_in, in, num_of_actions, num_of_actions);
+	p_actions = MLX5_ADDR_OF(alloc_modify_header_context_in, in, actions);
+	memcpy(p_actions, actions, num_of_actions * sizeof(u64));
+
+	err = mlx5_cmd_exec(mdev, in, inlen, out, sizeof(out));
+	if (err)
+		goto out;
+
+	*modify_header_id = MLX5_GET(alloc_modify_header_context_out, out,
+				     modify_header_id);
+out:
+	kvfree(in);
+	return err;
+}
+
+int mlx5dr_cmd_dealloc_modify_header(struct mlx5_core_dev *mdev,
+				     u32 modify_header_id)
+{
+	u32 in[MLX5_ST_SZ_DW(dealloc_modify_header_context_in)] = {};
+
+	MLX5_SET(dealloc_modify_header_context_in, in, opcode,
+		 MLX5_CMD_OP_DEALLOC_MODIFY_HEADER_CONTEXT);
+	MLX5_SET(dealloc_modify_header_context_in, in, modify_header_id,
+		 modify_header_id);
+
+	return mlx5_cmd_exec_in(mdev, dealloc_modify_header_context, in);
+}
+
+int mlx5dr_cmd_create_empty_flow_group(struct mlx5_core_dev *mdev,
+				       u32 table_type,
+				       u32 table_id,
+				       u32 *group_id)
+{
+	u32 out[MLX5_ST_SZ_DW(create_flow_group_out)] = {};
+	int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
+	u32 *in;
+	int err;
+
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in)
+		return -ENOMEM;
+
+	MLX5_SET(create_flow_group_in, in, opcode, MLX5_CMD_OP_CREATE_FLOW_GROUP);
+	MLX5_SET(create_flow_group_in, in, table_type, table_type);
+	MLX5_SET(create_flow_group_in, in, table_id, table_id);
+
+	err = mlx5_cmd_exec_inout(mdev, create_flow_group, in, out);
+	if (err)
+		goto out;
+
+	*group_id = MLX5_GET(create_flow_group_out, out, group_id);
+
+out:
+	kvfree(in);
+	return err;
+}
+
+int mlx5dr_cmd_destroy_flow_group(struct mlx5_core_dev *mdev,
+				  u32 table_type,
+				  u32 table_id,
+				  u32 group_id)
+{
+	u32 in[MLX5_ST_SZ_DW(destroy_flow_group_in)] = {};
+
+	MLX5_SET(destroy_flow_group_in, in, opcode,
+		 MLX5_CMD_OP_DESTROY_FLOW_GROUP);
+	MLX5_SET(destroy_flow_group_in, in, table_type, table_type);
+	MLX5_SET(destroy_flow_group_in, in, table_id, table_id);
+	MLX5_SET(destroy_flow_group_in, in, group_id, group_id);
+
+	return mlx5_cmd_exec_in(mdev, destroy_flow_group, in);
+}
+
+int mlx5dr_cmd_create_flow_table(struct mlx5_core_dev *mdev,
+				 struct mlx5dr_cmd_create_flow_table_attr *attr,
+				 u64 *fdb_rx_icm_addr,
+				 u32 *table_id)
+{
+	u32 out[MLX5_ST_SZ_DW(create_flow_table_out)] = {};
+	u32 in[MLX5_ST_SZ_DW(create_flow_table_in)] = {};
+	void *ft_mdev;
+	int err;
+
+	MLX5_SET(create_flow_table_in, in, opcode, MLX5_CMD_OP_CREATE_FLOW_TABLE);
+	MLX5_SET(create_flow_table_in, in, table_type, attr->table_type);
+	MLX5_SET(create_flow_table_in, in, uid, attr->uid);
+
+	ft_mdev = MLX5_ADDR_OF(create_flow_table_in, in, flow_table_context);
+	MLX5_SET(flow_table_context, ft_mdev, termination_table, attr->term_tbl);
+	MLX5_SET(flow_table_context, ft_mdev, sw_owner, attr->sw_owner);
+	MLX5_SET(flow_table_context, ft_mdev, level, attr->level);
+
+	if (attr->sw_owner) {
+		/* icm_addr_0 used for FDB RX / NIC TX / NIC_RX
+		 * icm_addr_1 used for FDB TX
+		 */
+		if (attr->table_type == MLX5_FLOW_TABLE_TYPE_NIC_RX) {
+			MLX5_SET64(flow_table_context, ft_mdev,
+				   sw_owner_icm_root_0, attr->icm_addr_rx);
+		} else if (attr->table_type == MLX5_FLOW_TABLE_TYPE_NIC_TX) {
+			MLX5_SET64(flow_table_context, ft_mdev,
+				   sw_owner_icm_root_0, attr->icm_addr_tx);
+		} else if (attr->table_type == MLX5_FLOW_TABLE_TYPE_FDB) {
+			MLX5_SET64(flow_table_context, ft_mdev,
+				   sw_owner_icm_root_0, attr->icm_addr_rx);
+			MLX5_SET64(flow_table_context, ft_mdev,
+				   sw_owner_icm_root_1, attr->icm_addr_tx);
+		}
+	}
+
+	MLX5_SET(create_flow_table_in, in, flow_table_context.decap_en,
+		 attr->decap_en);
+	MLX5_SET(create_flow_table_in, in, flow_table_context.reformat_en,
+		 attr->reformat_en);
+
+	err = mlx5_cmd_exec_inout(mdev, create_flow_table, in, out);
+	if (err)
+		return err;
+
+	*table_id = MLX5_GET(create_flow_table_out, out, table_id);
+	if (!attr->sw_owner && attr->table_type == MLX5_FLOW_TABLE_TYPE_FDB &&
+	    fdb_rx_icm_addr)
+		*fdb_rx_icm_addr =
+		(u64)MLX5_GET(create_flow_table_out, out, icm_address_31_0) |
+		(u64)MLX5_GET(create_flow_table_out, out, icm_address_39_32) << 32 |
+		(u64)MLX5_GET(create_flow_table_out, out, icm_address_63_40) << 40;
+
+	return 0;
+}
+
+int mlx5dr_cmd_destroy_flow_table(struct mlx5_core_dev *mdev,
+				  u32 table_id,
+				  u32 table_type)
+{
+	u32 in[MLX5_ST_SZ_DW(destroy_flow_table_in)] = {};
+
+	MLX5_SET(destroy_flow_table_in, in, opcode,
+		 MLX5_CMD_OP_DESTROY_FLOW_TABLE);
+	MLX5_SET(destroy_flow_table_in, in, table_type, table_type);
+	MLX5_SET(destroy_flow_table_in, in, table_id, table_id);
+
+	return mlx5_cmd_exec_in(mdev, destroy_flow_table, in);
+}
+
+int mlx5dr_cmd_create_reformat_ctx(struct mlx5_core_dev *mdev,
+				   enum mlx5_reformat_ctx_type rt,
+				   u8 reformat_param_0,
+				   u8 reformat_param_1,
+				   size_t reformat_size,
+				   void *reformat_data,
+				   u32 *reformat_id)
+{
+	u32 out[MLX5_ST_SZ_DW(alloc_packet_reformat_context_out)] = {};
+	size_t inlen, cmd_data_sz, cmd_total_sz;
+	void *prctx;
+	void *pdata;
+	void *in;
+	int err;
+
+	cmd_total_sz = MLX5_ST_SZ_BYTES(alloc_packet_reformat_context_in);
+	cmd_data_sz = MLX5_FLD_SZ_BYTES(alloc_packet_reformat_context_in,
+					packet_reformat_context.reformat_data);
+	inlen = ALIGN(cmd_total_sz + reformat_size - cmd_data_sz, 4);
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in)
+		return -ENOMEM;
+
+	MLX5_SET(alloc_packet_reformat_context_in, in, opcode,
+		 MLX5_CMD_OP_ALLOC_PACKET_REFORMAT_CONTEXT);
+
+	prctx = MLX5_ADDR_OF(alloc_packet_reformat_context_in, in, packet_reformat_context);
+	pdata = MLX5_ADDR_OF(packet_reformat_context_in, prctx, reformat_data);
+
+	MLX5_SET(packet_reformat_context_in, prctx, reformat_type, rt);
+	MLX5_SET(packet_reformat_context_in, prctx, reformat_param_0, reformat_param_0);
+	MLX5_SET(packet_reformat_context_in, prctx, reformat_param_1, reformat_param_1);
+	MLX5_SET(packet_reformat_context_in, prctx, reformat_data_size, reformat_size);
+	if (reformat_data && reformat_size)
+		memcpy(pdata, reformat_data, reformat_size);
+
+	err = mlx5_cmd_exec(mdev, in, inlen, out, sizeof(out));
+	if (err)
+		goto err_free_in;
+
+	*reformat_id = MLX5_GET(alloc_packet_reformat_context_out, out, packet_reformat_id);
+
+err_free_in:
+	kvfree(in);
+	return err;
+}
+
+void mlx5dr_cmd_destroy_reformat_ctx(struct mlx5_core_dev *mdev,
+				     u32 reformat_id)
+{
+	u32 in[MLX5_ST_SZ_DW(dealloc_packet_reformat_context_in)] = {};
+
+	MLX5_SET(dealloc_packet_reformat_context_in, in, opcode,
+		 MLX5_CMD_OP_DEALLOC_PACKET_REFORMAT_CONTEXT);
+	MLX5_SET(dealloc_packet_reformat_context_in, in, packet_reformat_id,
+		 reformat_id);
+
+	mlx5_cmd_exec_in(mdev, dealloc_packet_reformat_context, in);
+}
+
+int mlx5dr_cmd_query_gid(struct mlx5_core_dev *mdev, u8 vhca_port_num,
+			 u16 index, struct mlx5dr_cmd_gid_attr *attr)
+{
+	u32 out[MLX5_ST_SZ_DW(query_roce_address_out)] = {};
+	u32 in[MLX5_ST_SZ_DW(query_roce_address_in)] = {};
+	int err;
+
+	MLX5_SET(query_roce_address_in, in, opcode,
+		 MLX5_CMD_OP_QUERY_ROCE_ADDRESS);
+
+	MLX5_SET(query_roce_address_in, in, roce_address_index, index);
+	MLX5_SET(query_roce_address_in, in, vhca_port_num, vhca_port_num);
+
+	err = mlx5_cmd_exec_inout(mdev, query_roce_address, in, out);
+	if (err)
+		return err;
+
+	memcpy(&attr->gid,
+	       MLX5_ADDR_OF(query_roce_address_out,
+			    out, roce_address.source_l3_address),
+	       sizeof(attr->gid));
+	memcpy(attr->mac,
+	       MLX5_ADDR_OF(query_roce_address_out, out,
+			    roce_address.source_mac_47_32),
+	       sizeof(attr->mac));
+
+	if (MLX5_GET(query_roce_address_out, out,
+		     roce_address.roce_version) == MLX5_ROCE_VERSION_2)
+		attr->roce_ver = MLX5_ROCE_VERSION_2;
+	else
+		attr->roce_ver = MLX5_ROCE_VERSION_1;
+
+	return 0;
+}
+
+static int mlx5dr_cmd_set_extended_dest(struct mlx5_core_dev *dev,
+					struct mlx5dr_cmd_fte_info *fte,
+					bool *extended_dest)
+{
+	int fw_log_max_fdb_encap_uplink = MLX5_CAP_ESW(dev, log_max_fdb_encap_uplink);
+	int num_fwd_destinations = 0;
+	int num_encap = 0;
+	int i;
+
+	*extended_dest = false;
+	if (!(fte->action.action & MLX5_FLOW_CONTEXT_ACTION_FWD_DEST))
+		return 0;
+	for (i = 0; i < fte->dests_size; i++) {
+		if (fte->dest_arr[i].type == MLX5_FLOW_DESTINATION_TYPE_COUNTER ||
+		    fte->dest_arr[i].type == MLX5_FLOW_DESTINATION_TYPE_NONE)
+			continue;
+		if ((fte->dest_arr[i].type == MLX5_FLOW_DESTINATION_TYPE_VPORT ||
+		     fte->dest_arr[i].type == MLX5_FLOW_DESTINATION_TYPE_UPLINK) &&
+		    fte->dest_arr[i].vport.flags & MLX5_FLOW_DEST_VPORT_REFORMAT_ID)
+			num_encap++;
+		num_fwd_destinations++;
+	}
+
+	if (num_fwd_destinations > 1 && num_encap > 0)
+		*extended_dest = true;
+
+	if (*extended_dest && !fw_log_max_fdb_encap_uplink) {
+		mlx5_core_warn(dev, "FW does not support extended destination");
+		return -EOPNOTSUPP;
+	}
+	if (num_encap > (1 << fw_log_max_fdb_encap_uplink)) {
+		mlx5_core_warn(dev, "FW does not support more than %d encaps",
+			       1 << fw_log_max_fdb_encap_uplink);
+		return -EOPNOTSUPP;
+	}
+
+	return 0;
+}
+
+int mlx5dr_cmd_set_fte(struct mlx5_core_dev *dev,
+		       int opmod, int modify_mask,
+		       struct mlx5dr_cmd_ft_info *ft,
+		       u32 group_id,
+		       struct mlx5dr_cmd_fte_info *fte)
+{
+	u32 out[MLX5_ST_SZ_DW(set_fte_out)] = {};
+	void *in_flow_context, *vlan;
+	bool extended_dest = false;
+	void *in_match_value;
+	unsigned int inlen;
+	int dst_cnt_size;
+	void *in_dests;
+	u32 *in;
+	int err;
+	int i;
+
+	if (mlx5dr_cmd_set_extended_dest(dev, fte, &extended_dest))
+		return -EOPNOTSUPP;
+
+	if (!extended_dest)
+		dst_cnt_size = MLX5_ST_SZ_BYTES(dest_format_struct);
+	else
+		dst_cnt_size = MLX5_ST_SZ_BYTES(extended_dest_format);
+
+	inlen = MLX5_ST_SZ_BYTES(set_fte_in) + fte->dests_size * dst_cnt_size;
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in)
+		return -ENOMEM;
+
+	MLX5_SET(set_fte_in, in, opcode, MLX5_CMD_OP_SET_FLOW_TABLE_ENTRY);
+	MLX5_SET(set_fte_in, in, op_mod, opmod);
+	MLX5_SET(set_fte_in, in, modify_enable_mask, modify_mask);
+	MLX5_SET(set_fte_in, in, table_type, ft->type);
+	MLX5_SET(set_fte_in, in, table_id, ft->id);
+	MLX5_SET(set_fte_in, in, flow_index, fte->index);
+	MLX5_SET(set_fte_in, in, ignore_flow_level, fte->ignore_flow_level);
+	if (ft->vport) {
+		MLX5_SET(set_fte_in, in, vport_number, ft->vport);
+		MLX5_SET(set_fte_in, in, other_vport, 1);
+	}
+
+	in_flow_context = MLX5_ADDR_OF(set_fte_in, in, flow_context);
+	MLX5_SET(flow_context, in_flow_context, group_id, group_id);
+
+	MLX5_SET(flow_context, in_flow_context, flow_tag,
+		 fte->flow_context.flow_tag);
+	MLX5_SET(flow_context, in_flow_context, flow_source,
+		 fte->flow_context.flow_source);
+
+	MLX5_SET(flow_context, in_flow_context, extended_destination,
+		 extended_dest);
+	if (extended_dest) {
+		u32 action;
+
+		action = fte->action.action &
+			~MLX5_FLOW_CONTEXT_ACTION_PACKET_REFORMAT;
+		MLX5_SET(flow_context, in_flow_context, action, action);
+	} else {
+		MLX5_SET(flow_context, in_flow_context, action,
+			 fte->action.action);
+		if (fte->action.pkt_reformat)
+			MLX5_SET(flow_context, in_flow_context, packet_reformat_id,
+				 fte->action.pkt_reformat->id);
+	}
+	if (fte->action.modify_hdr)
+		MLX5_SET(flow_context, in_flow_context, modify_header_id,
+			 fte->action.modify_hdr->id);
+
+	vlan = MLX5_ADDR_OF(flow_context, in_flow_context, push_vlan);
+
+	MLX5_SET(vlan, vlan, ethtype, fte->action.vlan[0].ethtype);
+	MLX5_SET(vlan, vlan, vid, fte->action.vlan[0].vid);
+	MLX5_SET(vlan, vlan, prio, fte->action.vlan[0].prio);
+
+	vlan = MLX5_ADDR_OF(flow_context, in_flow_context, push_vlan_2);
+
+	MLX5_SET(vlan, vlan, ethtype, fte->action.vlan[1].ethtype);
+	MLX5_SET(vlan, vlan, vid, fte->action.vlan[1].vid);
+	MLX5_SET(vlan, vlan, prio, fte->action.vlan[1].prio);
+
+	in_match_value = MLX5_ADDR_OF(flow_context, in_flow_context,
+				      match_value);
+	memcpy(in_match_value, fte->val, sizeof(u32) * MLX5_ST_SZ_DW_MATCH_PARAM);
+
+	in_dests = MLX5_ADDR_OF(flow_context, in_flow_context, destination);
+	if (fte->action.action & MLX5_FLOW_CONTEXT_ACTION_FWD_DEST) {
+		int list_size = 0;
+
+		for (i = 0; i < fte->dests_size; i++) {
+			enum mlx5_flow_destination_type type = fte->dest_arr[i].type;
+			enum mlx5_ifc_flow_destination_type ifc_type;
+			unsigned int id;
+
+			if (type == MLX5_FLOW_DESTINATION_TYPE_COUNTER)
+				continue;
+
+			switch (type) {
+			case MLX5_FLOW_DESTINATION_TYPE_NONE:
+				continue;
+			case MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE_NUM:
+				id = fte->dest_arr[i].ft_num;
+				ifc_type = MLX5_IFC_FLOW_DESTINATION_TYPE_FLOW_TABLE;
+				break;
+			case MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE:
+				id = fte->dest_arr[i].ft_id;
+				ifc_type = MLX5_IFC_FLOW_DESTINATION_TYPE_FLOW_TABLE;
+
+				break;
+			case MLX5_FLOW_DESTINATION_TYPE_UPLINK:
+			case MLX5_FLOW_DESTINATION_TYPE_VPORT:
+				if (type == MLX5_FLOW_DESTINATION_TYPE_VPORT) {
+					id = fte->dest_arr[i].vport.num;
+					MLX5_SET(dest_format_struct, in_dests,
+						 destination_eswitch_owner_vhca_id_valid,
+						 !!(fte->dest_arr[i].vport.flags &
+						    MLX5_FLOW_DEST_VPORT_VHCA_ID));
+					ifc_type = MLX5_IFC_FLOW_DESTINATION_TYPE_VPORT;
+				} else {
+					id = 0;
+					ifc_type = MLX5_IFC_FLOW_DESTINATION_TYPE_UPLINK;
+					MLX5_SET(dest_format_struct, in_dests,
+						 destination_eswitch_owner_vhca_id_valid, 1);
+				}
+				MLX5_SET(dest_format_struct, in_dests,
+					 destination_eswitch_owner_vhca_id,
+					 fte->dest_arr[i].vport.vhca_id);
+				if (extended_dest && (fte->dest_arr[i].vport.flags &
+						    MLX5_FLOW_DEST_VPORT_REFORMAT_ID)) {
+					MLX5_SET(dest_format_struct, in_dests,
+						 packet_reformat,
+						 !!(fte->dest_arr[i].vport.flags &
+						    MLX5_FLOW_DEST_VPORT_REFORMAT_ID));
+					MLX5_SET(extended_dest_format, in_dests,
+						 packet_reformat_id,
+						 fte->dest_arr[i].vport.reformat_id);
+				}
+				break;
+			case MLX5_FLOW_DESTINATION_TYPE_FLOW_SAMPLER:
+				id = fte->dest_arr[i].sampler_id;
+				ifc_type = MLX5_IFC_FLOW_DESTINATION_TYPE_FLOW_SAMPLER;
+				break;
+			default:
+				id = fte->dest_arr[i].tir_num;
+				ifc_type = MLX5_IFC_FLOW_DESTINATION_TYPE_TIR;
+			}
+
+			MLX5_SET(dest_format_struct, in_dests, destination_type,
+				 ifc_type);
+			MLX5_SET(dest_format_struct, in_dests, destination_id, id);
+			in_dests += dst_cnt_size;
+			list_size++;
+		}
+
+		MLX5_SET(flow_context, in_flow_context, destination_list_size,
+			 list_size);
+	}
+
+	if (fte->action.action & MLX5_FLOW_CONTEXT_ACTION_COUNT) {
+		int max_list_size = BIT(MLX5_CAP_FLOWTABLE_TYPE(dev,
+					log_max_flow_counter,
+					ft->type));
+		int list_size = 0;
+
+		for (i = 0; i < fte->dests_size; i++) {
+			if (fte->dest_arr[i].type !=
+			    MLX5_FLOW_DESTINATION_TYPE_COUNTER)
+				continue;
+
+			MLX5_SET(flow_counter_list, in_dests, flow_counter_id,
+				 fte->dest_arr[i].counter_id);
+			in_dests += dst_cnt_size;
+			list_size++;
+		}
+		if (list_size > max_list_size) {
+			err = -EINVAL;
+			goto err_out;
+		}
+
+		MLX5_SET(flow_context, in_flow_context, flow_counter_list_size,
+			 list_size);
+	}
+
+	err = mlx5_cmd_exec(dev, in, inlen, out, sizeof(out));
+err_out:
+	kvfree(in);
+	return err;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_dbg.c b/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_dbg.c
new file mode 100644
index 000000000..7adcf0eec
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_dbg.c
@@ -0,0 +1,657 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+// Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+
+#include <linux/debugfs.h>
+#include <linux/kernel.h>
+#include <linux/seq_file.h>
+#include "dr_types.h"
+
+#define DR_DBG_PTR_TO_ID(p) ((u64)(uintptr_t)(p) & 0xFFFFFFFFULL)
+
+enum dr_dump_rec_type {
+	DR_DUMP_REC_TYPE_DOMAIN = 3000,
+	DR_DUMP_REC_TYPE_DOMAIN_INFO_FLEX_PARSER = 3001,
+	DR_DUMP_REC_TYPE_DOMAIN_INFO_DEV_ATTR = 3002,
+	DR_DUMP_REC_TYPE_DOMAIN_INFO_VPORT = 3003,
+	DR_DUMP_REC_TYPE_DOMAIN_INFO_CAPS = 3004,
+	DR_DUMP_REC_TYPE_DOMAIN_SEND_RING = 3005,
+
+	DR_DUMP_REC_TYPE_TABLE = 3100,
+	DR_DUMP_REC_TYPE_TABLE_RX = 3101,
+	DR_DUMP_REC_TYPE_TABLE_TX = 3102,
+
+	DR_DUMP_REC_TYPE_MATCHER = 3200,
+	DR_DUMP_REC_TYPE_MATCHER_MASK_DEPRECATED = 3201,
+	DR_DUMP_REC_TYPE_MATCHER_RX = 3202,
+	DR_DUMP_REC_TYPE_MATCHER_TX = 3203,
+	DR_DUMP_REC_TYPE_MATCHER_BUILDER = 3204,
+	DR_DUMP_REC_TYPE_MATCHER_MASK = 3205,
+
+	DR_DUMP_REC_TYPE_RULE = 3300,
+	DR_DUMP_REC_TYPE_RULE_RX_ENTRY_V0 = 3301,
+	DR_DUMP_REC_TYPE_RULE_TX_ENTRY_V0 = 3302,
+	DR_DUMP_REC_TYPE_RULE_RX_ENTRY_V1 = 3303,
+	DR_DUMP_REC_TYPE_RULE_TX_ENTRY_V1 = 3304,
+
+	DR_DUMP_REC_TYPE_ACTION_ENCAP_L2 = 3400,
+	DR_DUMP_REC_TYPE_ACTION_ENCAP_L3 = 3401,
+	DR_DUMP_REC_TYPE_ACTION_MODIFY_HDR = 3402,
+	DR_DUMP_REC_TYPE_ACTION_DROP = 3403,
+	DR_DUMP_REC_TYPE_ACTION_QP = 3404,
+	DR_DUMP_REC_TYPE_ACTION_FT = 3405,
+	DR_DUMP_REC_TYPE_ACTION_CTR = 3406,
+	DR_DUMP_REC_TYPE_ACTION_TAG = 3407,
+	DR_DUMP_REC_TYPE_ACTION_VPORT = 3408,
+	DR_DUMP_REC_TYPE_ACTION_DECAP_L2 = 3409,
+	DR_DUMP_REC_TYPE_ACTION_DECAP_L3 = 3410,
+	DR_DUMP_REC_TYPE_ACTION_DEVX_TIR = 3411,
+	DR_DUMP_REC_TYPE_ACTION_PUSH_VLAN = 3412,
+	DR_DUMP_REC_TYPE_ACTION_POP_VLAN = 3413,
+	DR_DUMP_REC_TYPE_ACTION_SAMPLER = 3415,
+	DR_DUMP_REC_TYPE_ACTION_INSERT_HDR = 3420,
+	DR_DUMP_REC_TYPE_ACTION_REMOVE_HDR = 3421
+};
+
+void mlx5dr_dbg_tbl_add(struct mlx5dr_table *tbl)
+{
+	mutex_lock(&tbl->dmn->dump_info.dbg_mutex);
+	list_add_tail(&tbl->dbg_node, &tbl->dmn->dbg_tbl_list);
+	mutex_unlock(&tbl->dmn->dump_info.dbg_mutex);
+}
+
+void mlx5dr_dbg_tbl_del(struct mlx5dr_table *tbl)
+{
+	mutex_lock(&tbl->dmn->dump_info.dbg_mutex);
+	list_del(&tbl->dbg_node);
+	mutex_unlock(&tbl->dmn->dump_info.dbg_mutex);
+}
+
+void mlx5dr_dbg_rule_add(struct mlx5dr_rule *rule)
+{
+	struct mlx5dr_domain *dmn = rule->matcher->tbl->dmn;
+
+	mutex_lock(&dmn->dump_info.dbg_mutex);
+	list_add_tail(&rule->dbg_node, &rule->matcher->dbg_rule_list);
+	mutex_unlock(&dmn->dump_info.dbg_mutex);
+}
+
+void mlx5dr_dbg_rule_del(struct mlx5dr_rule *rule)
+{
+	struct mlx5dr_domain *dmn = rule->matcher->tbl->dmn;
+
+	mutex_lock(&dmn->dump_info.dbg_mutex);
+	list_del(&rule->dbg_node);
+	mutex_unlock(&dmn->dump_info.dbg_mutex);
+}
+
+static u64 dr_dump_icm_to_idx(u64 icm_addr)
+{
+	return (icm_addr >> 6) & 0xffffffff;
+}
+
+#define DR_HEX_SIZE 256
+
+static void
+dr_dump_hex_print(char hex[DR_HEX_SIZE], char *src, u32 size)
+{
+	if (WARN_ON_ONCE(DR_HEX_SIZE < 2 * size + 1))
+		size = DR_HEX_SIZE / 2 - 1; /* truncate */
+
+	bin2hex(hex, src, size);
+	hex[2 * size] = 0; /* NULL-terminate */
+}
+
+static int
+dr_dump_rule_action_mem(struct seq_file *file, const u64 rule_id,
+			struct mlx5dr_rule_action_member *action_mem)
+{
+	struct mlx5dr_action *action = action_mem->action;
+	const u64 action_id = DR_DBG_PTR_TO_ID(action);
+
+	switch (action->action_type) {
+	case DR_ACTION_TYP_DROP:
+		seq_printf(file, "%d,0x%llx,0x%llx\n",
+			   DR_DUMP_REC_TYPE_ACTION_DROP, action_id, rule_id);
+		break;
+	case DR_ACTION_TYP_FT:
+		if (action->dest_tbl->is_fw_tbl)
+			seq_printf(file, "%d,0x%llx,0x%llx,0x%x,0x%x\n",
+				   DR_DUMP_REC_TYPE_ACTION_FT, action_id,
+				   rule_id, action->dest_tbl->fw_tbl.id,
+				   -1);
+		else
+			seq_printf(file, "%d,0x%llx,0x%llx,0x%x,0x%llx\n",
+				   DR_DUMP_REC_TYPE_ACTION_FT, action_id,
+				   rule_id, action->dest_tbl->tbl->table_id,
+				   DR_DBG_PTR_TO_ID(action->dest_tbl->tbl));
+
+		break;
+	case DR_ACTION_TYP_CTR:
+		seq_printf(file, "%d,0x%llx,0x%llx,0x%x\n",
+			   DR_DUMP_REC_TYPE_ACTION_CTR, action_id, rule_id,
+			   action->ctr->ctr_id + action->ctr->offset);
+		break;
+	case DR_ACTION_TYP_TAG:
+		seq_printf(file, "%d,0x%llx,0x%llx,0x%x\n",
+			   DR_DUMP_REC_TYPE_ACTION_TAG, action_id, rule_id,
+			   action->flow_tag->flow_tag);
+		break;
+	case DR_ACTION_TYP_MODIFY_HDR:
+		seq_printf(file, "%d,0x%llx,0x%llx,0x%x\n",
+			   DR_DUMP_REC_TYPE_ACTION_MODIFY_HDR, action_id,
+			   rule_id, action->rewrite->index);
+		break;
+	case DR_ACTION_TYP_VPORT:
+		seq_printf(file, "%d,0x%llx,0x%llx,0x%x\n",
+			   DR_DUMP_REC_TYPE_ACTION_VPORT, action_id, rule_id,
+			   action->vport->caps->num);
+		break;
+	case DR_ACTION_TYP_TNL_L2_TO_L2:
+		seq_printf(file, "%d,0x%llx,0x%llx\n",
+			   DR_DUMP_REC_TYPE_ACTION_DECAP_L2, action_id,
+			   rule_id);
+		break;
+	case DR_ACTION_TYP_TNL_L3_TO_L2:
+		seq_printf(file, "%d,0x%llx,0x%llx,0x%x\n",
+			   DR_DUMP_REC_TYPE_ACTION_DECAP_L3, action_id,
+			   rule_id, action->rewrite->index);
+		break;
+	case DR_ACTION_TYP_L2_TO_TNL_L2:
+		seq_printf(file, "%d,0x%llx,0x%llx,0x%x\n",
+			   DR_DUMP_REC_TYPE_ACTION_ENCAP_L2, action_id,
+			   rule_id, action->reformat->id);
+		break;
+	case DR_ACTION_TYP_L2_TO_TNL_L3:
+		seq_printf(file, "%d,0x%llx,0x%llx,0x%x\n",
+			   DR_DUMP_REC_TYPE_ACTION_ENCAP_L3, action_id,
+			   rule_id, action->reformat->id);
+		break;
+	case DR_ACTION_TYP_POP_VLAN:
+		seq_printf(file, "%d,0x%llx,0x%llx\n",
+			   DR_DUMP_REC_TYPE_ACTION_POP_VLAN, action_id,
+			   rule_id);
+		break;
+	case DR_ACTION_TYP_PUSH_VLAN:
+		seq_printf(file, "%d,0x%llx,0x%llx,0x%x\n",
+			   DR_DUMP_REC_TYPE_ACTION_PUSH_VLAN, action_id,
+			   rule_id, action->push_vlan->vlan_hdr);
+		break;
+	case DR_ACTION_TYP_INSERT_HDR:
+		seq_printf(file, "%d,0x%llx,0x%llx,0x%x,0x%x,0x%x\n",
+			   DR_DUMP_REC_TYPE_ACTION_INSERT_HDR, action_id,
+			   rule_id, action->reformat->id,
+			   action->reformat->param_0,
+			   action->reformat->param_1);
+		break;
+	case DR_ACTION_TYP_REMOVE_HDR:
+		seq_printf(file, "%d,0x%llx,0x%llx,0x%x,0x%x,0x%x\n",
+			   DR_DUMP_REC_TYPE_ACTION_REMOVE_HDR, action_id,
+			   rule_id, action->reformat->id,
+			   action->reformat->param_0,
+			   action->reformat->param_1);
+		break;
+	case DR_ACTION_TYP_SAMPLER:
+		seq_printf(file,
+			   "%d,0x%llx,0x%llx,0x%x,0x%x,0x%x,0x%llx,0x%llx\n",
+			   DR_DUMP_REC_TYPE_ACTION_SAMPLER, action_id, rule_id,
+			   0, 0, action->sampler->sampler_id,
+			   action->sampler->rx_icm_addr,
+			   action->sampler->tx_icm_addr);
+		break;
+	default:
+		return 0;
+	}
+
+	return 0;
+}
+
+static int
+dr_dump_rule_mem(struct seq_file *file, struct mlx5dr_ste *ste,
+		 bool is_rx, const u64 rule_id, u8 format_ver)
+{
+	char hw_ste_dump[DR_HEX_SIZE];
+	u32 mem_rec_type;
+
+	if (format_ver == MLX5_STEERING_FORMAT_CONNECTX_5) {
+		mem_rec_type = is_rx ? DR_DUMP_REC_TYPE_RULE_RX_ENTRY_V0 :
+				       DR_DUMP_REC_TYPE_RULE_TX_ENTRY_V0;
+	} else {
+		mem_rec_type = is_rx ? DR_DUMP_REC_TYPE_RULE_RX_ENTRY_V1 :
+				       DR_DUMP_REC_TYPE_RULE_TX_ENTRY_V1;
+	}
+
+	dr_dump_hex_print(hw_ste_dump, (char *)mlx5dr_ste_get_hw_ste(ste),
+			  DR_STE_SIZE_REDUCED);
+
+	seq_printf(file, "%d,0x%llx,0x%llx,%s\n", mem_rec_type,
+		   dr_dump_icm_to_idx(mlx5dr_ste_get_icm_addr(ste)), rule_id,
+		   hw_ste_dump);
+
+	return 0;
+}
+
+static int
+dr_dump_rule_rx_tx(struct seq_file *file, struct mlx5dr_rule_rx_tx *rule_rx_tx,
+		   bool is_rx, const u64 rule_id, u8 format_ver)
+{
+	struct mlx5dr_ste *ste_arr[DR_RULE_MAX_STES + DR_ACTION_MAX_STES];
+	struct mlx5dr_ste *curr_ste = rule_rx_tx->last_rule_ste;
+	int ret, i;
+
+	if (mlx5dr_rule_get_reverse_rule_members(ste_arr, curr_ste, &i))
+		return 0;
+
+	while (i--) {
+		ret = dr_dump_rule_mem(file, ste_arr[i], is_rx, rule_id,
+				       format_ver);
+		if (ret < 0)
+			return ret;
+	}
+
+	return 0;
+}
+
+static int dr_dump_rule(struct seq_file *file, struct mlx5dr_rule *rule)
+{
+	struct mlx5dr_rule_action_member *action_mem;
+	const u64 rule_id = DR_DBG_PTR_TO_ID(rule);
+	struct mlx5dr_rule_rx_tx *rx = &rule->rx;
+	struct mlx5dr_rule_rx_tx *tx = &rule->tx;
+	u8 format_ver;
+	int ret;
+
+	format_ver = rule->matcher->tbl->dmn->info.caps.sw_format_ver;
+
+	seq_printf(file, "%d,0x%llx,0x%llx\n", DR_DUMP_REC_TYPE_RULE, rule_id,
+		   DR_DBG_PTR_TO_ID(rule->matcher));
+
+	if (rx->nic_matcher) {
+		ret = dr_dump_rule_rx_tx(file, rx, true, rule_id, format_ver);
+		if (ret < 0)
+			return ret;
+	}
+
+	if (tx->nic_matcher) {
+		ret = dr_dump_rule_rx_tx(file, tx, false, rule_id, format_ver);
+		if (ret < 0)
+			return ret;
+	}
+
+	list_for_each_entry(action_mem, &rule->rule_actions_list, list) {
+		ret = dr_dump_rule_action_mem(file, rule_id, action_mem);
+		if (ret < 0)
+			return ret;
+	}
+
+	return 0;
+}
+
+static int
+dr_dump_matcher_mask(struct seq_file *file, struct mlx5dr_match_param *mask,
+		     u8 criteria, const u64 matcher_id)
+{
+	char dump[DR_HEX_SIZE];
+
+	seq_printf(file, "%d,0x%llx,", DR_DUMP_REC_TYPE_MATCHER_MASK,
+		   matcher_id);
+
+	if (criteria & DR_MATCHER_CRITERIA_OUTER) {
+		dr_dump_hex_print(dump, (char *)&mask->outer, sizeof(mask->outer));
+		seq_printf(file, "%s,", dump);
+	} else {
+		seq_puts(file, ",");
+	}
+
+	if (criteria & DR_MATCHER_CRITERIA_INNER) {
+		dr_dump_hex_print(dump, (char *)&mask->inner, sizeof(mask->inner));
+		seq_printf(file, "%s,", dump);
+	} else {
+		seq_puts(file, ",");
+	}
+
+	if (criteria & DR_MATCHER_CRITERIA_MISC) {
+		dr_dump_hex_print(dump, (char *)&mask->misc, sizeof(mask->misc));
+		seq_printf(file, "%s,", dump);
+	} else {
+		seq_puts(file, ",");
+	}
+
+	if (criteria & DR_MATCHER_CRITERIA_MISC2) {
+		dr_dump_hex_print(dump, (char *)&mask->misc2, sizeof(mask->misc2));
+		seq_printf(file, "%s,", dump);
+	} else {
+		seq_puts(file, ",");
+	}
+
+	if (criteria & DR_MATCHER_CRITERIA_MISC3) {
+		dr_dump_hex_print(dump, (char *)&mask->misc3, sizeof(mask->misc3));
+		seq_printf(file, "%s\n", dump);
+	} else {
+		seq_puts(file, ",\n");
+	}
+
+	return 0;
+}
+
+static int
+dr_dump_matcher_builder(struct seq_file *file, struct mlx5dr_ste_build *builder,
+			u32 index, bool is_rx, const u64 matcher_id)
+{
+	seq_printf(file, "%d,0x%llx,%d,%d,0x%x\n",
+		   DR_DUMP_REC_TYPE_MATCHER_BUILDER, matcher_id, index, is_rx,
+		   builder->lu_type);
+
+	return 0;
+}
+
+static int
+dr_dump_matcher_rx_tx(struct seq_file *file, bool is_rx,
+		      struct mlx5dr_matcher_rx_tx *matcher_rx_tx,
+		      const u64 matcher_id)
+{
+	enum dr_dump_rec_type rec_type;
+	u64 s_icm_addr, e_icm_addr;
+	int i, ret;
+
+	rec_type = is_rx ? DR_DUMP_REC_TYPE_MATCHER_RX :
+			   DR_DUMP_REC_TYPE_MATCHER_TX;
+
+	s_icm_addr = mlx5dr_icm_pool_get_chunk_icm_addr(matcher_rx_tx->s_htbl->chunk);
+	e_icm_addr = mlx5dr_icm_pool_get_chunk_icm_addr(matcher_rx_tx->e_anchor->chunk);
+	seq_printf(file, "%d,0x%llx,0x%llx,%d,0x%llx,0x%llx\n",
+		   rec_type, DR_DBG_PTR_TO_ID(matcher_rx_tx),
+		   matcher_id, matcher_rx_tx->num_of_builders,
+		   dr_dump_icm_to_idx(s_icm_addr),
+		   dr_dump_icm_to_idx(e_icm_addr));
+
+	for (i = 0; i < matcher_rx_tx->num_of_builders; i++) {
+		ret = dr_dump_matcher_builder(file,
+					      &matcher_rx_tx->ste_builder[i],
+					      i, is_rx, matcher_id);
+		if (ret < 0)
+			return ret;
+	}
+
+	return 0;
+}
+
+static int
+dr_dump_matcher(struct seq_file *file, struct mlx5dr_matcher *matcher)
+{
+	struct mlx5dr_matcher_rx_tx *rx = &matcher->rx;
+	struct mlx5dr_matcher_rx_tx *tx = &matcher->tx;
+	u64 matcher_id;
+	int ret;
+
+	matcher_id = DR_DBG_PTR_TO_ID(matcher);
+
+	seq_printf(file, "%d,0x%llx,0x%llx,%d\n", DR_DUMP_REC_TYPE_MATCHER,
+		   matcher_id, DR_DBG_PTR_TO_ID(matcher->tbl), matcher->prio);
+
+	ret = dr_dump_matcher_mask(file, &matcher->mask,
+				   matcher->match_criteria, matcher_id);
+	if (ret < 0)
+		return ret;
+
+	if (rx->nic_tbl) {
+		ret = dr_dump_matcher_rx_tx(file, true, rx, matcher_id);
+		if (ret < 0)
+			return ret;
+	}
+
+	if (tx->nic_tbl) {
+		ret = dr_dump_matcher_rx_tx(file, false, tx, matcher_id);
+		if (ret < 0)
+			return ret;
+	}
+
+	return 0;
+}
+
+static int
+dr_dump_matcher_all(struct seq_file *file, struct mlx5dr_matcher *matcher)
+{
+	struct mlx5dr_rule *rule;
+	int ret;
+
+	ret = dr_dump_matcher(file, matcher);
+	if (ret < 0)
+		return ret;
+
+	list_for_each_entry(rule, &matcher->dbg_rule_list, dbg_node) {
+		ret = dr_dump_rule(file, rule);
+		if (ret < 0)
+			return ret;
+	}
+
+	return 0;
+}
+
+static int
+dr_dump_table_rx_tx(struct seq_file *file, bool is_rx,
+		    struct mlx5dr_table_rx_tx *table_rx_tx,
+		    const u64 table_id)
+{
+	enum dr_dump_rec_type rec_type;
+	u64 s_icm_addr;
+
+	rec_type = is_rx ? DR_DUMP_REC_TYPE_TABLE_RX :
+			   DR_DUMP_REC_TYPE_TABLE_TX;
+
+	s_icm_addr = mlx5dr_icm_pool_get_chunk_icm_addr(table_rx_tx->s_anchor->chunk);
+	seq_printf(file, "%d,0x%llx,0x%llx\n", rec_type, table_id,
+		   dr_dump_icm_to_idx(s_icm_addr));
+
+	return 0;
+}
+
+static int dr_dump_table(struct seq_file *file, struct mlx5dr_table *table)
+{
+	struct mlx5dr_table_rx_tx *rx = &table->rx;
+	struct mlx5dr_table_rx_tx *tx = &table->tx;
+	int ret;
+
+	seq_printf(file, "%d,0x%llx,0x%llx,%d,%d\n", DR_DUMP_REC_TYPE_TABLE,
+		   DR_DBG_PTR_TO_ID(table), DR_DBG_PTR_TO_ID(table->dmn),
+		   table->table_type, table->level);
+
+	if (rx->nic_dmn) {
+		ret = dr_dump_table_rx_tx(file, true, rx,
+					  DR_DBG_PTR_TO_ID(table));
+		if (ret < 0)
+			return ret;
+	}
+
+	if (tx->nic_dmn) {
+		ret = dr_dump_table_rx_tx(file, false, tx,
+					  DR_DBG_PTR_TO_ID(table));
+		if (ret < 0)
+			return ret;
+	}
+	return 0;
+}
+
+static int dr_dump_table_all(struct seq_file *file, struct mlx5dr_table *tbl)
+{
+	struct mlx5dr_matcher *matcher;
+	int ret;
+
+	ret = dr_dump_table(file, tbl);
+	if (ret < 0)
+		return ret;
+
+	list_for_each_entry(matcher, &tbl->matcher_list, list_node) {
+		ret = dr_dump_matcher_all(file, matcher);
+		if (ret < 0)
+			return ret;
+	}
+	return 0;
+}
+
+static int
+dr_dump_send_ring(struct seq_file *file, struct mlx5dr_send_ring *ring,
+		  const u64 domain_id)
+{
+	seq_printf(file, "%d,0x%llx,0x%llx,0x%x,0x%x\n",
+		   DR_DUMP_REC_TYPE_DOMAIN_SEND_RING, DR_DBG_PTR_TO_ID(ring),
+		   domain_id, ring->cq->mcq.cqn, ring->qp->qpn);
+	return 0;
+}
+
+static int
+dr_dump_domain_info_flex_parser(struct seq_file *file,
+				const char *flex_parser_name,
+				const u8 flex_parser_value,
+				const u64 domain_id)
+{
+	seq_printf(file, "%d,0x%llx,%s,0x%x\n",
+		   DR_DUMP_REC_TYPE_DOMAIN_INFO_FLEX_PARSER, domain_id,
+		   flex_parser_name, flex_parser_value);
+	return 0;
+}
+
+static int
+dr_dump_domain_info_caps(struct seq_file *file, struct mlx5dr_cmd_caps *caps,
+			 const u64 domain_id)
+{
+	struct mlx5dr_cmd_vport_cap *vport_caps;
+	unsigned long i, vports_num;
+
+	xa_for_each(&caps->vports.vports_caps_xa, vports_num, vport_caps)
+		; /* count the number of vports in xarray */
+
+	seq_printf(file, "%d,0x%llx,0x%x,0x%llx,0x%llx,0x%x,%lu,%d\n",
+		   DR_DUMP_REC_TYPE_DOMAIN_INFO_CAPS, domain_id, caps->gvmi,
+		   caps->nic_rx_drop_address, caps->nic_tx_drop_address,
+		   caps->flex_protocols, vports_num, caps->eswitch_manager);
+
+	xa_for_each(&caps->vports.vports_caps_xa, i, vport_caps) {
+		vport_caps = xa_load(&caps->vports.vports_caps_xa, i);
+
+		seq_printf(file, "%d,0x%llx,%lu,0x%x,0x%llx,0x%llx\n",
+			   DR_DUMP_REC_TYPE_DOMAIN_INFO_VPORT, domain_id, i,
+			   vport_caps->vport_gvmi, vport_caps->icm_address_rx,
+			   vport_caps->icm_address_tx);
+	}
+	return 0;
+}
+
+static int
+dr_dump_domain_info(struct seq_file *file, struct mlx5dr_domain_info *info,
+		    const u64 domain_id)
+{
+	int ret;
+
+	ret = dr_dump_domain_info_caps(file, &info->caps, domain_id);
+	if (ret < 0)
+		return ret;
+
+	ret = dr_dump_domain_info_flex_parser(file, "icmp_dw0",
+					      info->caps.flex_parser_id_icmp_dw0,
+					      domain_id);
+	if (ret < 0)
+		return ret;
+
+	ret = dr_dump_domain_info_flex_parser(file, "icmp_dw1",
+					      info->caps.flex_parser_id_icmp_dw1,
+					      domain_id);
+	if (ret < 0)
+		return ret;
+
+	ret = dr_dump_domain_info_flex_parser(file, "icmpv6_dw0",
+					      info->caps.flex_parser_id_icmpv6_dw0,
+					      domain_id);
+	if (ret < 0)
+		return ret;
+
+	ret = dr_dump_domain_info_flex_parser(file, "icmpv6_dw1",
+					      info->caps.flex_parser_id_icmpv6_dw1,
+					      domain_id);
+	if (ret < 0)
+		return ret;
+
+	return 0;
+}
+
+static int
+dr_dump_domain(struct seq_file *file, struct mlx5dr_domain *dmn)
+{
+	u64 domain_id = DR_DBG_PTR_TO_ID(dmn);
+	int ret;
+
+	seq_printf(file, "%d,0x%llx,%d,0%x,%d,%s\n", DR_DUMP_REC_TYPE_DOMAIN,
+		   domain_id, dmn->type, dmn->info.caps.gvmi,
+		   dmn->info.supp_sw_steering, pci_name(dmn->mdev->pdev));
+
+	ret = dr_dump_domain_info(file, &dmn->info, domain_id);
+	if (ret < 0)
+		return ret;
+
+	if (dmn->info.supp_sw_steering) {
+		ret = dr_dump_send_ring(file, dmn->send_ring, domain_id);
+		if (ret < 0)
+			return ret;
+	}
+
+	return 0;
+}
+
+static int dr_dump_domain_all(struct seq_file *file, struct mlx5dr_domain *dmn)
+{
+	struct mlx5dr_table *tbl;
+	int ret;
+
+	mutex_lock(&dmn->dump_info.dbg_mutex);
+	mlx5dr_domain_lock(dmn);
+
+	ret = dr_dump_domain(file, dmn);
+	if (ret < 0)
+		goto unlock_mutex;
+
+	list_for_each_entry(tbl, &dmn->dbg_tbl_list, dbg_node) {
+		ret = dr_dump_table_all(file, tbl);
+		if (ret < 0)
+			break;
+	}
+
+unlock_mutex:
+	mlx5dr_domain_unlock(dmn);
+	mutex_unlock(&dmn->dump_info.dbg_mutex);
+	return ret;
+}
+
+static int dr_dump_show(struct seq_file *file, void *priv)
+{
+	return dr_dump_domain_all(file, file->private);
+}
+DEFINE_SHOW_ATTRIBUTE(dr_dump);
+
+void mlx5dr_dbg_init_dump(struct mlx5dr_domain *dmn)
+{
+	struct mlx5_core_dev *dev = dmn->mdev;
+	char file_name[128];
+
+	if (dmn->type != MLX5DR_DOMAIN_TYPE_FDB) {
+		mlx5_core_warn(dev,
+			       "Steering dump is not supported for NIC RX/TX domains\n");
+		return;
+	}
+
+	dmn->dump_info.steering_debugfs =
+		debugfs_create_dir("steering", mlx5_debugfs_get_dev_root(dev));
+	dmn->dump_info.fdb_debugfs =
+		debugfs_create_dir("fdb", dmn->dump_info.steering_debugfs);
+
+	sprintf(file_name, "dmn_%p", dmn);
+	debugfs_create_file(file_name, 0444, dmn->dump_info.fdb_debugfs,
+			    dmn, &dr_dump_fops);
+
+	INIT_LIST_HEAD(&dmn->dbg_tbl_list);
+	mutex_init(&dmn->dump_info.dbg_mutex);
+}
+
+void mlx5dr_dbg_uninit_dump(struct mlx5dr_domain *dmn)
+{
+	debugfs_remove_recursive(dmn->dump_info.steering_debugfs);
+	mutex_destroy(&dmn->dump_info.dbg_mutex);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_dbg.h b/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_dbg.h
new file mode 100644
index 000000000..def6cf853
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_dbg.h
@@ -0,0 +1,15 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2021, NVIDIA CORPORATION & AFFILIATES. All rights reserved. */
+
+struct mlx5dr_dbg_dump_info {
+	struct mutex dbg_mutex; /* protect dbg lists */
+	struct dentry *steering_debugfs;
+	struct dentry *fdb_debugfs;
+};
+
+void mlx5dr_dbg_init_dump(struct mlx5dr_domain *dmn);
+void mlx5dr_dbg_uninit_dump(struct mlx5dr_domain *dmn);
+void mlx5dr_dbg_tbl_add(struct mlx5dr_table *tbl);
+void mlx5dr_dbg_tbl_del(struct mlx5dr_table *tbl);
+void mlx5dr_dbg_rule_add(struct mlx5dr_rule *rule);
+void mlx5dr_dbg_rule_del(struct mlx5dr_rule *rule);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_domain.c b/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_domain.c
new file mode 100644
index 000000000..fc6ae49b5
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_domain.c
@@ -0,0 +1,463 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2019 Mellanox Technologies. */
+
+#include <linux/mlx5/eswitch.h>
+#include <linux/err.h>
+#include "dr_types.h"
+
+#define DR_DOMAIN_SW_STEERING_SUPPORTED(dmn, dmn_type)	\
+	((dmn)->info.caps.dmn_type##_sw_owner ||	\
+	 ((dmn)->info.caps.dmn_type##_sw_owner_v2 &&	\
+	  (dmn)->info.caps.sw_format_ver <= MLX5_STEERING_FORMAT_CONNECTX_7))
+
+static void dr_domain_init_csum_recalc_fts(struct mlx5dr_domain *dmn)
+{
+	/* Per vport cached FW FT for checksum recalculation, this
+	 * recalculation is needed due to a HW bug in STEv0.
+	 */
+	xa_init(&dmn->csum_fts_xa);
+}
+
+static void dr_domain_uninit_csum_recalc_fts(struct mlx5dr_domain *dmn)
+{
+	struct mlx5dr_fw_recalc_cs_ft *recalc_cs_ft;
+	unsigned long i;
+
+	xa_for_each(&dmn->csum_fts_xa, i, recalc_cs_ft) {
+		if (recalc_cs_ft)
+			mlx5dr_fw_destroy_recalc_cs_ft(dmn, recalc_cs_ft);
+	}
+
+	xa_destroy(&dmn->csum_fts_xa);
+}
+
+int mlx5dr_domain_get_recalc_cs_ft_addr(struct mlx5dr_domain *dmn,
+					u16 vport_num,
+					u64 *rx_icm_addr)
+{
+	struct mlx5dr_fw_recalc_cs_ft *recalc_cs_ft;
+	int ret;
+
+	recalc_cs_ft = xa_load(&dmn->csum_fts_xa, vport_num);
+	if (!recalc_cs_ft) {
+		/* Table hasn't been created yet */
+		recalc_cs_ft = mlx5dr_fw_create_recalc_cs_ft(dmn, vport_num);
+		if (!recalc_cs_ft)
+			return -EINVAL;
+
+		ret = xa_err(xa_store(&dmn->csum_fts_xa, vport_num,
+				      recalc_cs_ft, GFP_KERNEL));
+		if (ret)
+			return ret;
+	}
+
+	*rx_icm_addr = recalc_cs_ft->rx_icm_addr;
+
+	return 0;
+}
+
+static int dr_domain_init_resources(struct mlx5dr_domain *dmn)
+{
+	int ret;
+
+	dmn->ste_ctx = mlx5dr_ste_get_ctx(dmn->info.caps.sw_format_ver);
+	if (!dmn->ste_ctx) {
+		mlx5dr_err(dmn, "SW Steering on this device is unsupported\n");
+		return -EOPNOTSUPP;
+	}
+
+	ret = mlx5_core_alloc_pd(dmn->mdev, &dmn->pdn);
+	if (ret) {
+		mlx5dr_err(dmn, "Couldn't allocate PD, ret: %d", ret);
+		return ret;
+	}
+
+	dmn->uar = mlx5_get_uars_page(dmn->mdev);
+	if (IS_ERR(dmn->uar)) {
+		mlx5dr_err(dmn, "Couldn't allocate UAR\n");
+		ret = PTR_ERR(dmn->uar);
+		goto clean_pd;
+	}
+
+	dmn->ste_icm_pool = mlx5dr_icm_pool_create(dmn, DR_ICM_TYPE_STE);
+	if (!dmn->ste_icm_pool) {
+		mlx5dr_err(dmn, "Couldn't get icm memory\n");
+		ret = -ENOMEM;
+		goto clean_uar;
+	}
+
+	dmn->action_icm_pool = mlx5dr_icm_pool_create(dmn, DR_ICM_TYPE_MODIFY_ACTION);
+	if (!dmn->action_icm_pool) {
+		mlx5dr_err(dmn, "Couldn't get action icm memory\n");
+		ret = -ENOMEM;
+		goto free_ste_icm_pool;
+	}
+
+	ret = mlx5dr_send_ring_alloc(dmn);
+	if (ret) {
+		mlx5dr_err(dmn, "Couldn't create send-ring\n");
+		goto free_action_icm_pool;
+	}
+
+	return 0;
+
+free_action_icm_pool:
+	mlx5dr_icm_pool_destroy(dmn->action_icm_pool);
+free_ste_icm_pool:
+	mlx5dr_icm_pool_destroy(dmn->ste_icm_pool);
+clean_uar:
+	mlx5_put_uars_page(dmn->mdev, dmn->uar);
+clean_pd:
+	mlx5_core_dealloc_pd(dmn->mdev, dmn->pdn);
+
+	return ret;
+}
+
+static void dr_domain_uninit_resources(struct mlx5dr_domain *dmn)
+{
+	mlx5dr_send_ring_free(dmn, dmn->send_ring);
+	mlx5dr_icm_pool_destroy(dmn->action_icm_pool);
+	mlx5dr_icm_pool_destroy(dmn->ste_icm_pool);
+	mlx5_put_uars_page(dmn->mdev, dmn->uar);
+	mlx5_core_dealloc_pd(dmn->mdev, dmn->pdn);
+}
+
+static void dr_domain_fill_uplink_caps(struct mlx5dr_domain *dmn,
+				       struct mlx5dr_cmd_vport_cap *uplink_vport)
+{
+	struct mlx5dr_esw_caps *esw_caps = &dmn->info.caps.esw_caps;
+
+	uplink_vport->num = MLX5_VPORT_UPLINK;
+	uplink_vport->icm_address_rx = esw_caps->uplink_icm_address_rx;
+	uplink_vport->icm_address_tx = esw_caps->uplink_icm_address_tx;
+	uplink_vport->vport_gvmi = 0;
+	uplink_vport->vhca_gvmi = dmn->info.caps.gvmi;
+}
+
+static int dr_domain_query_vport(struct mlx5dr_domain *dmn,
+				 u16 vport_number,
+				 bool other_vport,
+				 struct mlx5dr_cmd_vport_cap *vport_caps)
+{
+	int ret;
+
+	ret = mlx5dr_cmd_query_esw_vport_context(dmn->mdev,
+						 other_vport,
+						 vport_number,
+						 &vport_caps->icm_address_rx,
+						 &vport_caps->icm_address_tx);
+	if (ret)
+		return ret;
+
+	ret = mlx5dr_cmd_query_gvmi(dmn->mdev,
+				    other_vport,
+				    vport_number,
+				    &vport_caps->vport_gvmi);
+	if (ret)
+		return ret;
+
+	vport_caps->num = vport_number;
+	vport_caps->vhca_gvmi = dmn->info.caps.gvmi;
+
+	return 0;
+}
+
+static int dr_domain_query_esw_mngr(struct mlx5dr_domain *dmn)
+{
+	return dr_domain_query_vport(dmn, 0, false,
+				     &dmn->info.caps.vports.esw_manager_caps);
+}
+
+static void dr_domain_query_uplink(struct mlx5dr_domain *dmn)
+{
+	dr_domain_fill_uplink_caps(dmn, &dmn->info.caps.vports.uplink_caps);
+}
+
+static struct mlx5dr_cmd_vport_cap *
+dr_domain_add_vport_cap(struct mlx5dr_domain *dmn, u16 vport)
+{
+	struct mlx5dr_cmd_caps *caps = &dmn->info.caps;
+	struct mlx5dr_cmd_vport_cap *vport_caps;
+	int ret;
+
+	vport_caps = kvzalloc(sizeof(*vport_caps), GFP_KERNEL);
+	if (!vport_caps)
+		return NULL;
+
+	ret = dr_domain_query_vport(dmn, vport, true, vport_caps);
+	if (ret) {
+		kvfree(vport_caps);
+		return NULL;
+	}
+
+	ret = xa_insert(&caps->vports.vports_caps_xa, vport,
+			vport_caps, GFP_KERNEL);
+	if (ret) {
+		mlx5dr_dbg(dmn, "Couldn't insert new vport into xarray (%d)\n", ret);
+		kvfree(vport_caps);
+		return ERR_PTR(ret);
+	}
+
+	return vport_caps;
+}
+
+static bool dr_domain_is_esw_mgr_vport(struct mlx5dr_domain *dmn, u16 vport)
+{
+	struct mlx5dr_cmd_caps *caps = &dmn->info.caps;
+
+	return (caps->is_ecpf && vport == MLX5_VPORT_ECPF) ||
+	       (!caps->is_ecpf && vport == 0);
+}
+
+struct mlx5dr_cmd_vport_cap *
+mlx5dr_domain_get_vport_cap(struct mlx5dr_domain *dmn, u16 vport)
+{
+	struct mlx5dr_cmd_caps *caps = &dmn->info.caps;
+	struct mlx5dr_cmd_vport_cap *vport_caps;
+
+	if (dr_domain_is_esw_mgr_vport(dmn, vport))
+		return &caps->vports.esw_manager_caps;
+
+	if (vport == MLX5_VPORT_UPLINK)
+		return &caps->vports.uplink_caps;
+
+vport_load:
+	vport_caps = xa_load(&caps->vports.vports_caps_xa, vport);
+	if (vport_caps)
+		return vport_caps;
+
+	vport_caps = dr_domain_add_vport_cap(dmn, vport);
+	if (PTR_ERR(vport_caps) == -EBUSY)
+		/* caps were already stored by another thread */
+		goto vport_load;
+
+	return vport_caps;
+}
+
+static void dr_domain_clear_vports(struct mlx5dr_domain *dmn)
+{
+	struct mlx5dr_cmd_vport_cap *vport_caps;
+	unsigned long i;
+
+	xa_for_each(&dmn->info.caps.vports.vports_caps_xa, i, vport_caps) {
+		vport_caps = xa_erase(&dmn->info.caps.vports.vports_caps_xa, i);
+		kvfree(vport_caps);
+	}
+}
+
+static int dr_domain_query_fdb_caps(struct mlx5_core_dev *mdev,
+				    struct mlx5dr_domain *dmn)
+{
+	int ret;
+
+	if (!dmn->info.caps.eswitch_manager)
+		return -EOPNOTSUPP;
+
+	ret = mlx5dr_cmd_query_esw_caps(mdev, &dmn->info.caps.esw_caps);
+	if (ret)
+		return ret;
+
+	dmn->info.caps.fdb_sw_owner = dmn->info.caps.esw_caps.sw_owner;
+	dmn->info.caps.fdb_sw_owner_v2 = dmn->info.caps.esw_caps.sw_owner_v2;
+	dmn->info.caps.esw_rx_drop_address = dmn->info.caps.esw_caps.drop_icm_address_rx;
+	dmn->info.caps.esw_tx_drop_address = dmn->info.caps.esw_caps.drop_icm_address_tx;
+
+	xa_init(&dmn->info.caps.vports.vports_caps_xa);
+
+	/* Query eswitch manager and uplink vports only. Rest of the
+	 * vports (vport 0, VFs and SFs) will be queried dynamically.
+	 */
+
+	ret = dr_domain_query_esw_mngr(dmn);
+	if (ret) {
+		mlx5dr_err(dmn, "Failed to query eswitch manager vport caps (err: %d)", ret);
+		goto free_vports_caps_xa;
+	}
+
+	dr_domain_query_uplink(dmn);
+
+	return 0;
+
+free_vports_caps_xa:
+	xa_destroy(&dmn->info.caps.vports.vports_caps_xa);
+
+	return ret;
+}
+
+static int dr_domain_caps_init(struct mlx5_core_dev *mdev,
+			       struct mlx5dr_domain *dmn)
+{
+	struct mlx5dr_cmd_vport_cap *vport_cap;
+	int ret;
+
+	if (MLX5_CAP_GEN(mdev, port_type) != MLX5_CAP_PORT_TYPE_ETH) {
+		mlx5dr_err(dmn, "Failed to allocate domain, bad link type\n");
+		return -EOPNOTSUPP;
+	}
+
+	ret = mlx5dr_cmd_query_device(mdev, &dmn->info.caps);
+	if (ret)
+		return ret;
+
+	ret = dr_domain_query_fdb_caps(mdev, dmn);
+	if (ret)
+		return ret;
+
+	switch (dmn->type) {
+	case MLX5DR_DOMAIN_TYPE_NIC_RX:
+		if (!DR_DOMAIN_SW_STEERING_SUPPORTED(dmn, rx))
+			return -ENOTSUPP;
+
+		dmn->info.supp_sw_steering = true;
+		dmn->info.rx.type = DR_DOMAIN_NIC_TYPE_RX;
+		dmn->info.rx.default_icm_addr = dmn->info.caps.nic_rx_drop_address;
+		dmn->info.rx.drop_icm_addr = dmn->info.caps.nic_rx_drop_address;
+		break;
+	case MLX5DR_DOMAIN_TYPE_NIC_TX:
+		if (!DR_DOMAIN_SW_STEERING_SUPPORTED(dmn, tx))
+			return -ENOTSUPP;
+
+		dmn->info.supp_sw_steering = true;
+		dmn->info.tx.type = DR_DOMAIN_NIC_TYPE_TX;
+		dmn->info.tx.default_icm_addr = dmn->info.caps.nic_tx_allow_address;
+		dmn->info.tx.drop_icm_addr = dmn->info.caps.nic_tx_drop_address;
+		break;
+	case MLX5DR_DOMAIN_TYPE_FDB:
+		if (!dmn->info.caps.eswitch_manager)
+			return -ENOTSUPP;
+
+		if (!DR_DOMAIN_SW_STEERING_SUPPORTED(dmn, fdb))
+			return -ENOTSUPP;
+
+		dmn->info.rx.type = DR_DOMAIN_NIC_TYPE_RX;
+		dmn->info.tx.type = DR_DOMAIN_NIC_TYPE_TX;
+		vport_cap = &dmn->info.caps.vports.esw_manager_caps;
+
+		dmn->info.supp_sw_steering = true;
+		dmn->info.tx.default_icm_addr = vport_cap->icm_address_tx;
+		dmn->info.rx.default_icm_addr = vport_cap->icm_address_rx;
+		dmn->info.rx.drop_icm_addr = dmn->info.caps.esw_rx_drop_address;
+		dmn->info.tx.drop_icm_addr = dmn->info.caps.esw_tx_drop_address;
+		break;
+	default:
+		mlx5dr_err(dmn, "Invalid domain\n");
+		ret = -EINVAL;
+		break;
+	}
+
+	return ret;
+}
+
+static void dr_domain_caps_uninit(struct mlx5dr_domain *dmn)
+{
+	dr_domain_clear_vports(dmn);
+	xa_destroy(&dmn->info.caps.vports.vports_caps_xa);
+}
+
+struct mlx5dr_domain *
+mlx5dr_domain_create(struct mlx5_core_dev *mdev, enum mlx5dr_domain_type type)
+{
+	struct mlx5dr_domain *dmn;
+	int ret;
+
+	if (type > MLX5DR_DOMAIN_TYPE_FDB)
+		return NULL;
+
+	dmn = kzalloc(sizeof(*dmn), GFP_KERNEL);
+	if (!dmn)
+		return NULL;
+
+	dmn->mdev = mdev;
+	dmn->type = type;
+	refcount_set(&dmn->refcount, 1);
+	mutex_init(&dmn->info.rx.mutex);
+	mutex_init(&dmn->info.tx.mutex);
+
+	if (dr_domain_caps_init(mdev, dmn)) {
+		mlx5dr_err(dmn, "Failed init domain, no caps\n");
+		goto free_domain;
+	}
+
+	dmn->info.max_log_action_icm_sz = DR_CHUNK_SIZE_4K;
+	dmn->info.max_log_sw_icm_sz = min_t(u32, DR_CHUNK_SIZE_1024K,
+					    dmn->info.caps.log_icm_size);
+
+	if (!dmn->info.supp_sw_steering) {
+		mlx5dr_err(dmn, "SW steering is not supported\n");
+		goto uninit_caps;
+	}
+
+	/* Allocate resources */
+	ret = dr_domain_init_resources(dmn);
+	if (ret) {
+		mlx5dr_err(dmn, "Failed init domain resources\n");
+		goto uninit_caps;
+	}
+
+	dr_domain_init_csum_recalc_fts(dmn);
+	mlx5dr_dbg_init_dump(dmn);
+	return dmn;
+
+uninit_caps:
+	dr_domain_caps_uninit(dmn);
+free_domain:
+	kfree(dmn);
+	return NULL;
+}
+
+/* Assure synchronization of the device steering tables with updates made by SW
+ * insertion.
+ */
+int mlx5dr_domain_sync(struct mlx5dr_domain *dmn, u32 flags)
+{
+	int ret = 0;
+
+	if (flags & MLX5DR_DOMAIN_SYNC_FLAGS_SW) {
+		mlx5dr_domain_lock(dmn);
+		ret = mlx5dr_send_ring_force_drain(dmn);
+		mlx5dr_domain_unlock(dmn);
+		if (ret) {
+			mlx5dr_err(dmn, "Force drain failed flags: %d, ret: %d\n",
+				   flags, ret);
+			return ret;
+		}
+	}
+
+	if (flags & MLX5DR_DOMAIN_SYNC_FLAGS_HW)
+		ret = mlx5dr_cmd_sync_steering(dmn->mdev);
+
+	return ret;
+}
+
+int mlx5dr_domain_destroy(struct mlx5dr_domain *dmn)
+{
+	if (WARN_ON_ONCE(refcount_read(&dmn->refcount) > 1))
+		return -EBUSY;
+
+	/* make sure resources are not used by the hardware */
+	mlx5dr_cmd_sync_steering(dmn->mdev);
+	mlx5dr_dbg_uninit_dump(dmn);
+	dr_domain_uninit_csum_recalc_fts(dmn);
+	dr_domain_uninit_resources(dmn);
+	dr_domain_caps_uninit(dmn);
+	mutex_destroy(&dmn->info.tx.mutex);
+	mutex_destroy(&dmn->info.rx.mutex);
+	kfree(dmn);
+	return 0;
+}
+
+void mlx5dr_domain_set_peer(struct mlx5dr_domain *dmn,
+			    struct mlx5dr_domain *peer_dmn)
+{
+	mlx5dr_domain_lock(dmn);
+
+	if (dmn->peer_dmn)
+		refcount_dec(&dmn->peer_dmn->refcount);
+
+	dmn->peer_dmn = peer_dmn;
+
+	if (dmn->peer_dmn)
+		refcount_inc(&dmn->peer_dmn->refcount);
+
+	mlx5dr_domain_unlock(dmn);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_fw.c b/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_fw.c
new file mode 100644
index 000000000..f05ef0cd5
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_fw.c
@@ -0,0 +1,171 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2019 Mellanox Technologies. */
+
+#include <linux/types.h>
+#include "dr_types.h"
+
+struct mlx5dr_fw_recalc_cs_ft *
+mlx5dr_fw_create_recalc_cs_ft(struct mlx5dr_domain *dmn, u16 vport_num)
+{
+	struct mlx5dr_cmd_create_flow_table_attr ft_attr = {};
+	struct mlx5dr_fw_recalc_cs_ft *recalc_cs_ft;
+	u32 table_id, group_id, modify_hdr_id;
+	u64 rx_icm_addr, modify_ttl_action;
+	int ret;
+
+	recalc_cs_ft = kzalloc(sizeof(*recalc_cs_ft), GFP_KERNEL);
+	if (!recalc_cs_ft)
+		return NULL;
+
+	ft_attr.table_type = MLX5_FLOW_TABLE_TYPE_FDB;
+	ft_attr.level = dmn->info.caps.max_ft_level - 1;
+	ft_attr.term_tbl = true;
+
+	ret = mlx5dr_cmd_create_flow_table(dmn->mdev,
+					   &ft_attr,
+					   &rx_icm_addr,
+					   &table_id);
+	if (ret) {
+		mlx5dr_err(dmn, "Failed creating TTL W/A FW flow table %d\n", ret);
+		goto free_ttl_tbl;
+	}
+
+	ret = mlx5dr_cmd_create_empty_flow_group(dmn->mdev,
+						 MLX5_FLOW_TABLE_TYPE_FDB,
+						 table_id, &group_id);
+	if (ret) {
+		mlx5dr_err(dmn, "Failed creating TTL W/A FW flow group %d\n", ret);
+		goto destroy_flow_table;
+	}
+
+	/* Modify TTL action by adding zero to trigger CS recalculation */
+	modify_ttl_action = 0;
+	MLX5_SET(set_action_in, &modify_ttl_action, action_type, MLX5_ACTION_TYPE_ADD);
+	MLX5_SET(set_action_in, &modify_ttl_action, field, MLX5_ACTION_IN_FIELD_OUT_IP_TTL);
+
+	ret = mlx5dr_cmd_alloc_modify_header(dmn->mdev, MLX5_FLOW_TABLE_TYPE_FDB, 1,
+					     &modify_ttl_action,
+					     &modify_hdr_id);
+	if (ret) {
+		mlx5dr_err(dmn, "Failed modify header TTL %d\n", ret);
+		goto destroy_flow_group;
+	}
+
+	ret = mlx5dr_cmd_set_fte_modify_and_vport(dmn->mdev,
+						  MLX5_FLOW_TABLE_TYPE_FDB,
+						  table_id, group_id, modify_hdr_id,
+						  vport_num);
+	if (ret) {
+		mlx5dr_err(dmn, "Failed setting TTL W/A flow table entry %d\n", ret);
+		goto dealloc_modify_header;
+	}
+
+	recalc_cs_ft->modify_hdr_id = modify_hdr_id;
+	recalc_cs_ft->rx_icm_addr = rx_icm_addr;
+	recalc_cs_ft->table_id = table_id;
+	recalc_cs_ft->group_id = group_id;
+
+	return recalc_cs_ft;
+
+dealloc_modify_header:
+	mlx5dr_cmd_dealloc_modify_header(dmn->mdev, modify_hdr_id);
+destroy_flow_group:
+	mlx5dr_cmd_destroy_flow_group(dmn->mdev,
+				      MLX5_FLOW_TABLE_TYPE_FDB,
+				      table_id, group_id);
+destroy_flow_table:
+	mlx5dr_cmd_destroy_flow_table(dmn->mdev, table_id, MLX5_FLOW_TABLE_TYPE_FDB);
+free_ttl_tbl:
+	kfree(recalc_cs_ft);
+	return NULL;
+}
+
+void mlx5dr_fw_destroy_recalc_cs_ft(struct mlx5dr_domain *dmn,
+				    struct mlx5dr_fw_recalc_cs_ft *recalc_cs_ft)
+{
+	mlx5dr_cmd_del_flow_table_entry(dmn->mdev,
+					MLX5_FLOW_TABLE_TYPE_FDB,
+					recalc_cs_ft->table_id);
+	mlx5dr_cmd_dealloc_modify_header(dmn->mdev, recalc_cs_ft->modify_hdr_id);
+	mlx5dr_cmd_destroy_flow_group(dmn->mdev,
+				      MLX5_FLOW_TABLE_TYPE_FDB,
+				      recalc_cs_ft->table_id,
+				      recalc_cs_ft->group_id);
+	mlx5dr_cmd_destroy_flow_table(dmn->mdev,
+				      recalc_cs_ft->table_id,
+				      MLX5_FLOW_TABLE_TYPE_FDB);
+
+	kfree(recalc_cs_ft);
+}
+
+int mlx5dr_fw_create_md_tbl(struct mlx5dr_domain *dmn,
+			    struct mlx5dr_cmd_flow_destination_hw_info *dest,
+			    int num_dest,
+			    bool reformat_req,
+			    u32 *tbl_id,
+			    u32 *group_id,
+			    bool ignore_flow_level,
+			    u32 flow_source)
+{
+	struct mlx5dr_cmd_create_flow_table_attr ft_attr = {};
+	struct mlx5dr_cmd_fte_info fte_info = {};
+	u32 val[MLX5_ST_SZ_DW_MATCH_PARAM] = {};
+	struct mlx5dr_cmd_ft_info ft_info = {};
+	int ret;
+
+	ft_attr.table_type = MLX5_FLOW_TABLE_TYPE_FDB;
+	ft_attr.level = min_t(int, dmn->info.caps.max_ft_level - 2,
+			      MLX5_FT_MAX_MULTIPATH_LEVEL);
+	ft_attr.reformat_en = reformat_req;
+	ft_attr.decap_en = reformat_req;
+
+	ret = mlx5dr_cmd_create_flow_table(dmn->mdev, &ft_attr, NULL, tbl_id);
+	if (ret) {
+		mlx5dr_err(dmn, "Failed creating multi dest FW flow table %d\n", ret);
+		return ret;
+	}
+
+	ret = mlx5dr_cmd_create_empty_flow_group(dmn->mdev,
+						 MLX5_FLOW_TABLE_TYPE_FDB,
+						 *tbl_id, group_id);
+	if (ret) {
+		mlx5dr_err(dmn, "Failed creating multi dest FW flow group %d\n", ret);
+		goto free_flow_table;
+	}
+
+	ft_info.id = *tbl_id;
+	ft_info.type = FS_FT_FDB;
+	fte_info.action.action = MLX5_FLOW_CONTEXT_ACTION_FWD_DEST;
+	fte_info.dests_size = num_dest;
+	fte_info.val = val;
+	fte_info.dest_arr = dest;
+	fte_info.ignore_flow_level = ignore_flow_level;
+	fte_info.flow_context.flow_source = flow_source;
+
+	ret = mlx5dr_cmd_set_fte(dmn->mdev, 0, 0, &ft_info, *group_id, &fte_info);
+	if (ret) {
+		mlx5dr_err(dmn, "Failed setting fte into table %d\n", ret);
+		goto free_flow_group;
+	}
+
+	return 0;
+
+free_flow_group:
+	mlx5dr_cmd_destroy_flow_group(dmn->mdev, MLX5_FLOW_TABLE_TYPE_FDB,
+				      *tbl_id, *group_id);
+free_flow_table:
+	mlx5dr_cmd_destroy_flow_table(dmn->mdev, *tbl_id,
+				      MLX5_FLOW_TABLE_TYPE_FDB);
+	return ret;
+}
+
+void mlx5dr_fw_destroy_md_tbl(struct mlx5dr_domain *dmn,
+			      u32 tbl_id, u32 group_id)
+{
+	mlx5dr_cmd_del_flow_table_entry(dmn->mdev, FS_FT_FDB, tbl_id);
+	mlx5dr_cmd_destroy_flow_group(dmn->mdev,
+				      MLX5_FLOW_TABLE_TYPE_FDB,
+				      tbl_id, group_id);
+	mlx5dr_cmd_destroy_flow_table(dmn->mdev, tbl_id,
+				      MLX5_FLOW_TABLE_TYPE_FDB);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_icm_pool.c b/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_icm_pool.c
new file mode 100644
index 000000000..4ca67fa24
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_icm_pool.c
@@ -0,0 +1,520 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2019 Mellanox Technologies. */
+
+#include "dr_types.h"
+
+#define DR_ICM_MODIFY_HDR_ALIGN_BASE 64
+
+struct mlx5dr_icm_pool {
+	enum mlx5dr_icm_type icm_type;
+	enum mlx5dr_icm_chunk_size max_log_chunk_sz;
+	struct mlx5dr_domain *dmn;
+	/* memory management */
+	struct mutex mutex; /* protect the ICM pool and ICM buddy */
+	struct list_head buddy_mem_list;
+	u64 hot_memory_size;
+};
+
+struct mlx5dr_icm_dm {
+	u32 obj_id;
+	enum mlx5_sw_icm_type type;
+	phys_addr_t addr;
+	size_t length;
+};
+
+struct mlx5dr_icm_mr {
+	u32 mkey;
+	struct mlx5dr_icm_dm dm;
+	struct mlx5dr_domain *dmn;
+	size_t length;
+	u64 icm_start_addr;
+};
+
+static int dr_icm_create_dm_mkey(struct mlx5_core_dev *mdev,
+				 u32 pd, u64 length, u64 start_addr, int mode,
+				 u32 *mkey)
+{
+	u32 inlen = MLX5_ST_SZ_BYTES(create_mkey_in);
+	u32 in[MLX5_ST_SZ_DW(create_mkey_in)] = {};
+	void *mkc;
+
+	mkc = MLX5_ADDR_OF(create_mkey_in, in, memory_key_mkey_entry);
+
+	MLX5_SET(mkc, mkc, access_mode_1_0, mode);
+	MLX5_SET(mkc, mkc, access_mode_4_2, (mode >> 2) & 0x7);
+	MLX5_SET(mkc, mkc, lw, 1);
+	MLX5_SET(mkc, mkc, lr, 1);
+	if (mode == MLX5_MKC_ACCESS_MODE_SW_ICM) {
+		MLX5_SET(mkc, mkc, rw, 1);
+		MLX5_SET(mkc, mkc, rr, 1);
+	}
+
+	MLX5_SET64(mkc, mkc, len, length);
+	MLX5_SET(mkc, mkc, pd, pd);
+	MLX5_SET(mkc, mkc, qpn, 0xffffff);
+	MLX5_SET64(mkc, mkc, start_addr, start_addr);
+
+	return mlx5_core_create_mkey(mdev, mkey, in, inlen);
+}
+
+u64 mlx5dr_icm_pool_get_chunk_mr_addr(struct mlx5dr_icm_chunk *chunk)
+{
+	u32 offset = mlx5dr_icm_pool_dm_type_to_entry_size(chunk->buddy_mem->pool->icm_type);
+
+	return (u64)offset * chunk->seg;
+}
+
+u32 mlx5dr_icm_pool_get_chunk_rkey(struct mlx5dr_icm_chunk *chunk)
+{
+	return chunk->buddy_mem->icm_mr->mkey;
+}
+
+u64 mlx5dr_icm_pool_get_chunk_icm_addr(struct mlx5dr_icm_chunk *chunk)
+{
+	u32 size = mlx5dr_icm_pool_dm_type_to_entry_size(chunk->buddy_mem->pool->icm_type);
+
+	return (u64)chunk->buddy_mem->icm_mr->icm_start_addr + size * chunk->seg;
+}
+
+u32 mlx5dr_icm_pool_get_chunk_byte_size(struct mlx5dr_icm_chunk *chunk)
+{
+	return mlx5dr_icm_pool_chunk_size_to_byte(chunk->size,
+			chunk->buddy_mem->pool->icm_type);
+}
+
+u32 mlx5dr_icm_pool_get_chunk_num_of_entries(struct mlx5dr_icm_chunk *chunk)
+{
+	return mlx5dr_icm_pool_chunk_size_to_entries(chunk->size);
+}
+
+static struct mlx5dr_icm_mr *
+dr_icm_pool_mr_create(struct mlx5dr_icm_pool *pool)
+{
+	struct mlx5_core_dev *mdev = pool->dmn->mdev;
+	enum mlx5_sw_icm_type dm_type;
+	struct mlx5dr_icm_mr *icm_mr;
+	size_t log_align_base;
+	int err;
+
+	icm_mr = kvzalloc(sizeof(*icm_mr), GFP_KERNEL);
+	if (!icm_mr)
+		return NULL;
+
+	icm_mr->dmn = pool->dmn;
+
+	icm_mr->dm.length = mlx5dr_icm_pool_chunk_size_to_byte(pool->max_log_chunk_sz,
+							       pool->icm_type);
+
+	if (pool->icm_type == DR_ICM_TYPE_STE) {
+		dm_type = MLX5_SW_ICM_TYPE_STEERING;
+		log_align_base = ilog2(icm_mr->dm.length);
+	} else {
+		dm_type = MLX5_SW_ICM_TYPE_HEADER_MODIFY;
+		/* Align base is 64B */
+		log_align_base = ilog2(DR_ICM_MODIFY_HDR_ALIGN_BASE);
+	}
+	icm_mr->dm.type = dm_type;
+
+	err = mlx5_dm_sw_icm_alloc(mdev, icm_mr->dm.type, icm_mr->dm.length,
+				   log_align_base, 0, &icm_mr->dm.addr,
+				   &icm_mr->dm.obj_id);
+	if (err) {
+		mlx5dr_err(pool->dmn, "Failed to allocate SW ICM memory, err (%d)\n", err);
+		goto free_icm_mr;
+	}
+
+	/* Register device memory */
+	err = dr_icm_create_dm_mkey(mdev, pool->dmn->pdn,
+				    icm_mr->dm.length,
+				    icm_mr->dm.addr,
+				    MLX5_MKC_ACCESS_MODE_SW_ICM,
+				    &icm_mr->mkey);
+	if (err) {
+		mlx5dr_err(pool->dmn, "Failed to create SW ICM MKEY, err (%d)\n", err);
+		goto free_dm;
+	}
+
+	icm_mr->icm_start_addr = icm_mr->dm.addr;
+
+	if (icm_mr->icm_start_addr & (BIT(log_align_base) - 1)) {
+		mlx5dr_err(pool->dmn, "Failed to get Aligned ICM mem (asked: %zu)\n",
+			   log_align_base);
+		goto free_mkey;
+	}
+
+	return icm_mr;
+
+free_mkey:
+	mlx5_core_destroy_mkey(mdev, icm_mr->mkey);
+free_dm:
+	mlx5_dm_sw_icm_dealloc(mdev, icm_mr->dm.type, icm_mr->dm.length, 0,
+			       icm_mr->dm.addr, icm_mr->dm.obj_id);
+free_icm_mr:
+	kvfree(icm_mr);
+	return NULL;
+}
+
+static void dr_icm_pool_mr_destroy(struct mlx5dr_icm_mr *icm_mr)
+{
+	struct mlx5_core_dev *mdev = icm_mr->dmn->mdev;
+	struct mlx5dr_icm_dm *dm = &icm_mr->dm;
+
+	mlx5_core_destroy_mkey(mdev, icm_mr->mkey);
+	mlx5_dm_sw_icm_dealloc(mdev, dm->type, dm->length, 0,
+			       dm->addr, dm->obj_id);
+	kvfree(icm_mr);
+}
+
+static int dr_icm_buddy_get_ste_size(struct mlx5dr_icm_buddy_mem *buddy)
+{
+	/* We support only one type of STE size, both for ConnectX-5 and later
+	 * devices. Once the support for match STE which has a larger tag is
+	 * added (32B instead of 16B), the STE size for devices later than
+	 * ConnectX-5 needs to account for that.
+	 */
+	return DR_STE_SIZE_REDUCED;
+}
+
+static void dr_icm_chunk_ste_init(struct mlx5dr_icm_chunk *chunk, int offset)
+{
+	struct mlx5dr_icm_buddy_mem *buddy = chunk->buddy_mem;
+	int index = offset / DR_STE_SIZE;
+
+	chunk->ste_arr = &buddy->ste_arr[index];
+	chunk->miss_list = &buddy->miss_list[index];
+	chunk->hw_ste_arr = buddy->hw_ste_arr +
+			    index * dr_icm_buddy_get_ste_size(buddy);
+}
+
+static void dr_icm_chunk_ste_cleanup(struct mlx5dr_icm_chunk *chunk)
+{
+	int num_of_entries = mlx5dr_icm_pool_get_chunk_num_of_entries(chunk);
+	struct mlx5dr_icm_buddy_mem *buddy = chunk->buddy_mem;
+
+	memset(chunk->hw_ste_arr, 0,
+	       num_of_entries * dr_icm_buddy_get_ste_size(buddy));
+	memset(chunk->ste_arr, 0,
+	       num_of_entries * sizeof(chunk->ste_arr[0]));
+}
+
+static enum mlx5dr_icm_type
+get_chunk_icm_type(struct mlx5dr_icm_chunk *chunk)
+{
+	return chunk->buddy_mem->pool->icm_type;
+}
+
+static void dr_icm_chunk_destroy(struct mlx5dr_icm_chunk *chunk,
+				 struct mlx5dr_icm_buddy_mem *buddy)
+{
+	enum mlx5dr_icm_type icm_type = get_chunk_icm_type(chunk);
+
+	buddy->used_memory -= mlx5dr_icm_pool_get_chunk_byte_size(chunk);
+	list_del(&chunk->chunk_list);
+
+	if (icm_type == DR_ICM_TYPE_STE)
+		dr_icm_chunk_ste_cleanup(chunk);
+
+	kvfree(chunk);
+}
+
+static int dr_icm_buddy_init_ste_cache(struct mlx5dr_icm_buddy_mem *buddy)
+{
+	int num_of_entries =
+		mlx5dr_icm_pool_chunk_size_to_entries(buddy->pool->max_log_chunk_sz);
+
+	buddy->ste_arr = kvcalloc(num_of_entries,
+				  sizeof(struct mlx5dr_ste), GFP_KERNEL);
+	if (!buddy->ste_arr)
+		return -ENOMEM;
+
+	/* Preallocate full STE size on non-ConnectX-5 devices since
+	 * we need to support both full and reduced with the same cache.
+	 */
+	buddy->hw_ste_arr = kvcalloc(num_of_entries,
+				     dr_icm_buddy_get_ste_size(buddy), GFP_KERNEL);
+	if (!buddy->hw_ste_arr)
+		goto free_ste_arr;
+
+	buddy->miss_list = kvmalloc(num_of_entries * sizeof(struct list_head), GFP_KERNEL);
+	if (!buddy->miss_list)
+		goto free_hw_ste_arr;
+
+	return 0;
+
+free_hw_ste_arr:
+	kvfree(buddy->hw_ste_arr);
+free_ste_arr:
+	kvfree(buddy->ste_arr);
+	return -ENOMEM;
+}
+
+static void dr_icm_buddy_cleanup_ste_cache(struct mlx5dr_icm_buddy_mem *buddy)
+{
+	kvfree(buddy->ste_arr);
+	kvfree(buddy->hw_ste_arr);
+	kvfree(buddy->miss_list);
+}
+
+static int dr_icm_buddy_create(struct mlx5dr_icm_pool *pool)
+{
+	struct mlx5dr_icm_buddy_mem *buddy;
+	struct mlx5dr_icm_mr *icm_mr;
+
+	icm_mr = dr_icm_pool_mr_create(pool);
+	if (!icm_mr)
+		return -ENOMEM;
+
+	buddy = kvzalloc(sizeof(*buddy), GFP_KERNEL);
+	if (!buddy)
+		goto free_mr;
+
+	if (mlx5dr_buddy_init(buddy, pool->max_log_chunk_sz))
+		goto err_free_buddy;
+
+	buddy->icm_mr = icm_mr;
+	buddy->pool = pool;
+
+	if (pool->icm_type == DR_ICM_TYPE_STE) {
+		/* Reduce allocations by preallocating and reusing the STE structures */
+		if (dr_icm_buddy_init_ste_cache(buddy))
+			goto err_cleanup_buddy;
+	}
+
+	/* add it to the -start- of the list in order to search in it first */
+	list_add(&buddy->list_node, &pool->buddy_mem_list);
+
+	return 0;
+
+err_cleanup_buddy:
+	mlx5dr_buddy_cleanup(buddy);
+err_free_buddy:
+	kvfree(buddy);
+free_mr:
+	dr_icm_pool_mr_destroy(icm_mr);
+	return -ENOMEM;
+}
+
+static void dr_icm_buddy_destroy(struct mlx5dr_icm_buddy_mem *buddy)
+{
+	struct mlx5dr_icm_chunk *chunk, *next;
+
+	list_for_each_entry_safe(chunk, next, &buddy->hot_list, chunk_list)
+		dr_icm_chunk_destroy(chunk, buddy);
+
+	list_for_each_entry_safe(chunk, next, &buddy->used_list, chunk_list)
+		dr_icm_chunk_destroy(chunk, buddy);
+
+	dr_icm_pool_mr_destroy(buddy->icm_mr);
+
+	mlx5dr_buddy_cleanup(buddy);
+
+	if (buddy->pool->icm_type == DR_ICM_TYPE_STE)
+		dr_icm_buddy_cleanup_ste_cache(buddy);
+
+	kvfree(buddy);
+}
+
+static struct mlx5dr_icm_chunk *
+dr_icm_chunk_create(struct mlx5dr_icm_pool *pool,
+		    enum mlx5dr_icm_chunk_size chunk_size,
+		    struct mlx5dr_icm_buddy_mem *buddy_mem_pool,
+		    unsigned int seg)
+{
+	struct mlx5dr_icm_chunk *chunk;
+	int offset;
+
+	chunk = kvzalloc(sizeof(*chunk), GFP_KERNEL);
+	if (!chunk)
+		return NULL;
+
+	offset = mlx5dr_icm_pool_dm_type_to_entry_size(pool->icm_type) * seg;
+
+	chunk->seg = seg;
+	chunk->size = chunk_size;
+	chunk->buddy_mem = buddy_mem_pool;
+
+	if (pool->icm_type == DR_ICM_TYPE_STE)
+		dr_icm_chunk_ste_init(chunk, offset);
+
+	buddy_mem_pool->used_memory += mlx5dr_icm_pool_get_chunk_byte_size(chunk);
+	INIT_LIST_HEAD(&chunk->chunk_list);
+
+	/* chunk now is part of the used_list */
+	list_add_tail(&chunk->chunk_list, &buddy_mem_pool->used_list);
+
+	return chunk;
+}
+
+static bool dr_icm_pool_is_sync_required(struct mlx5dr_icm_pool *pool)
+{
+	int allow_hot_size;
+
+	/* sync when hot memory reaches half of the pool size */
+	allow_hot_size =
+		mlx5dr_icm_pool_chunk_size_to_byte(pool->max_log_chunk_sz,
+						   pool->icm_type) / 2;
+
+	return pool->hot_memory_size > allow_hot_size;
+}
+
+static int dr_icm_pool_sync_all_buddy_pools(struct mlx5dr_icm_pool *pool)
+{
+	struct mlx5dr_icm_buddy_mem *buddy, *tmp_buddy;
+	u32 num_entries;
+	int err;
+
+	err = mlx5dr_cmd_sync_steering(pool->dmn->mdev);
+	if (err) {
+		mlx5dr_err(pool->dmn, "Failed to sync to HW (err: %d)\n", err);
+		return err;
+	}
+
+	list_for_each_entry_safe(buddy, tmp_buddy, &pool->buddy_mem_list, list_node) {
+		struct mlx5dr_icm_chunk *chunk, *tmp_chunk;
+
+		list_for_each_entry_safe(chunk, tmp_chunk, &buddy->hot_list, chunk_list) {
+			num_entries = mlx5dr_icm_pool_get_chunk_num_of_entries(chunk);
+			mlx5dr_buddy_free_mem(buddy, chunk->seg, ilog2(num_entries));
+			pool->hot_memory_size -= mlx5dr_icm_pool_get_chunk_byte_size(chunk);
+			dr_icm_chunk_destroy(chunk, buddy);
+		}
+
+		if (!buddy->used_memory && pool->icm_type == DR_ICM_TYPE_STE)
+			dr_icm_buddy_destroy(buddy);
+	}
+
+	return 0;
+}
+
+static int dr_icm_handle_buddies_get_mem(struct mlx5dr_icm_pool *pool,
+					 enum mlx5dr_icm_chunk_size chunk_size,
+					 struct mlx5dr_icm_buddy_mem **buddy,
+					 unsigned int *seg)
+{
+	struct mlx5dr_icm_buddy_mem *buddy_mem_pool;
+	bool new_mem = false;
+	int err;
+
+alloc_buddy_mem:
+	/* find the next free place from the buddy list */
+	list_for_each_entry(buddy_mem_pool, &pool->buddy_mem_list, list_node) {
+		err = mlx5dr_buddy_alloc_mem(buddy_mem_pool,
+					     chunk_size, seg);
+		if (!err)
+			goto found;
+
+		if (WARN_ON(new_mem)) {
+			/* We have new memory pool, first in the list */
+			mlx5dr_err(pool->dmn,
+				   "No memory for order: %d\n",
+				   chunk_size);
+			goto out;
+		}
+	}
+
+	/* no more available allocators in that pool, create new */
+	err = dr_icm_buddy_create(pool);
+	if (err) {
+		mlx5dr_err(pool->dmn,
+			   "Failed creating buddy for order %d\n",
+			   chunk_size);
+		goto out;
+	}
+
+	/* mark we have new memory, first in list */
+	new_mem = true;
+	goto alloc_buddy_mem;
+
+found:
+	*buddy = buddy_mem_pool;
+out:
+	return err;
+}
+
+/* Allocate an ICM chunk, each chunk holds a piece of ICM memory and
+ * also memory used for HW STE management for optimizations.
+ */
+struct mlx5dr_icm_chunk *
+mlx5dr_icm_alloc_chunk(struct mlx5dr_icm_pool *pool,
+		       enum mlx5dr_icm_chunk_size chunk_size)
+{
+	struct mlx5dr_icm_chunk *chunk = NULL;
+	struct mlx5dr_icm_buddy_mem *buddy;
+	unsigned int seg;
+	int ret;
+
+	if (chunk_size > pool->max_log_chunk_sz)
+		return NULL;
+
+	mutex_lock(&pool->mutex);
+	/* find mem, get back the relevant buddy pool and seg in that mem */
+	ret = dr_icm_handle_buddies_get_mem(pool, chunk_size, &buddy, &seg);
+	if (ret)
+		goto out;
+
+	chunk = dr_icm_chunk_create(pool, chunk_size, buddy, seg);
+	if (!chunk)
+		goto out_err;
+
+	goto out;
+
+out_err:
+	mlx5dr_buddy_free_mem(buddy, seg, chunk_size);
+out:
+	mutex_unlock(&pool->mutex);
+	return chunk;
+}
+
+void mlx5dr_icm_free_chunk(struct mlx5dr_icm_chunk *chunk)
+{
+	struct mlx5dr_icm_buddy_mem *buddy = chunk->buddy_mem;
+	struct mlx5dr_icm_pool *pool = buddy->pool;
+
+	/* move the memory to the waiting list AKA "hot" */
+	mutex_lock(&pool->mutex);
+	list_move_tail(&chunk->chunk_list, &buddy->hot_list);
+	pool->hot_memory_size += mlx5dr_icm_pool_get_chunk_byte_size(chunk);
+
+	/* Check if we have chunks that are waiting for sync-ste */
+	if (dr_icm_pool_is_sync_required(pool))
+		dr_icm_pool_sync_all_buddy_pools(pool);
+
+	mutex_unlock(&pool->mutex);
+}
+
+struct mlx5dr_icm_pool *mlx5dr_icm_pool_create(struct mlx5dr_domain *dmn,
+					       enum mlx5dr_icm_type icm_type)
+{
+	enum mlx5dr_icm_chunk_size max_log_chunk_sz;
+	struct mlx5dr_icm_pool *pool;
+
+	if (icm_type == DR_ICM_TYPE_STE)
+		max_log_chunk_sz = dmn->info.max_log_sw_icm_sz;
+	else
+		max_log_chunk_sz = dmn->info.max_log_action_icm_sz;
+
+	pool = kvzalloc(sizeof(*pool), GFP_KERNEL);
+	if (!pool)
+		return NULL;
+
+	pool->dmn = dmn;
+	pool->icm_type = icm_type;
+	pool->max_log_chunk_sz = max_log_chunk_sz;
+
+	INIT_LIST_HEAD(&pool->buddy_mem_list);
+
+	mutex_init(&pool->mutex);
+
+	return pool;
+}
+
+void mlx5dr_icm_pool_destroy(struct mlx5dr_icm_pool *pool)
+{
+	struct mlx5dr_icm_buddy_mem *buddy, *tmp_buddy;
+
+	list_for_each_entry_safe(buddy, tmp_buddy, &pool->buddy_mem_list, list_node)
+		dr_icm_buddy_destroy(buddy);
+
+	mutex_destroy(&pool->mutex);
+	kvfree(pool);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_matcher.c b/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_matcher.c
new file mode 100644
index 000000000..0726848eb
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_matcher.c
@@ -0,0 +1,1108 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2019 Mellanox Technologies. */
+
+#include "dr_types.h"
+
+static bool dr_mask_is_smac_set(struct mlx5dr_match_spec *spec)
+{
+	return (spec->smac_47_16 || spec->smac_15_0);
+}
+
+static bool dr_mask_is_dmac_set(struct mlx5dr_match_spec *spec)
+{
+	return (spec->dmac_47_16 || spec->dmac_15_0);
+}
+
+static bool dr_mask_is_l3_base_set(struct mlx5dr_match_spec *spec)
+{
+	return (spec->ip_protocol || spec->frag || spec->tcp_flags ||
+		spec->ip_ecn || spec->ip_dscp);
+}
+
+static bool dr_mask_is_tcp_udp_base_set(struct mlx5dr_match_spec *spec)
+{
+	return (spec->tcp_sport || spec->tcp_dport ||
+		spec->udp_sport || spec->udp_dport);
+}
+
+static bool dr_mask_is_ipv4_set(struct mlx5dr_match_spec *spec)
+{
+	return (spec->dst_ip_31_0 || spec->src_ip_31_0);
+}
+
+static bool dr_mask_is_ipv4_5_tuple_set(struct mlx5dr_match_spec *spec)
+{
+	return (dr_mask_is_l3_base_set(spec) ||
+		dr_mask_is_tcp_udp_base_set(spec) ||
+		dr_mask_is_ipv4_set(spec));
+}
+
+static bool dr_mask_is_eth_l2_tnl_set(struct mlx5dr_match_misc *misc)
+{
+	return misc->vxlan_vni;
+}
+
+static bool dr_mask_is_ttl_set(struct mlx5dr_match_spec *spec)
+{
+	return spec->ttl_hoplimit;
+}
+
+static bool dr_mask_is_ipv4_ihl_set(struct mlx5dr_match_spec *spec)
+{
+	return spec->ipv4_ihl;
+}
+
+#define DR_MASK_IS_L2_DST(_spec, _misc, _inner_outer) (_spec.first_vid || \
+	(_spec).first_cfi || (_spec).first_prio || (_spec).cvlan_tag || \
+	(_spec).svlan_tag || (_spec).dmac_47_16 || (_spec).dmac_15_0 || \
+	(_spec).ethertype || (_spec).ip_version || \
+	(_misc)._inner_outer##_second_vid || \
+	(_misc)._inner_outer##_second_cfi || \
+	(_misc)._inner_outer##_second_prio || \
+	(_misc)._inner_outer##_second_cvlan_tag || \
+	(_misc)._inner_outer##_second_svlan_tag)
+
+#define DR_MASK_IS_ETH_L4_SET(_spec, _misc, _inner_outer) ( \
+	dr_mask_is_l3_base_set(&(_spec)) || \
+	dr_mask_is_tcp_udp_base_set(&(_spec)) || \
+	dr_mask_is_ttl_set(&(_spec)) || \
+	(_misc)._inner_outer##_ipv6_flow_label)
+
+#define DR_MASK_IS_ETH_L4_MISC_SET(_misc3, _inner_outer) ( \
+	(_misc3)._inner_outer##_tcp_seq_num || \
+	(_misc3)._inner_outer##_tcp_ack_num)
+
+#define DR_MASK_IS_FIRST_MPLS_SET(_misc2, _inner_outer) ( \
+	(_misc2)._inner_outer##_first_mpls_label || \
+	(_misc2)._inner_outer##_first_mpls_exp || \
+	(_misc2)._inner_outer##_first_mpls_s_bos || \
+	(_misc2)._inner_outer##_first_mpls_ttl)
+
+static bool dr_mask_is_tnl_gre_set(struct mlx5dr_match_misc *misc)
+{
+	return (misc->gre_key_h || misc->gre_key_l ||
+		misc->gre_protocol || misc->gre_c_present ||
+		misc->gre_k_present || misc->gre_s_present);
+}
+
+#define DR_MASK_IS_OUTER_MPLS_OVER_GRE_SET(_misc) (\
+	(_misc)->outer_first_mpls_over_gre_label || \
+	(_misc)->outer_first_mpls_over_gre_exp || \
+	(_misc)->outer_first_mpls_over_gre_s_bos || \
+	(_misc)->outer_first_mpls_over_gre_ttl)
+
+#define DR_MASK_IS_OUTER_MPLS_OVER_UDP_SET(_misc) (\
+	(_misc)->outer_first_mpls_over_udp_label || \
+	(_misc)->outer_first_mpls_over_udp_exp || \
+	(_misc)->outer_first_mpls_over_udp_s_bos || \
+	(_misc)->outer_first_mpls_over_udp_ttl)
+
+static bool
+dr_mask_is_vxlan_gpe_set(struct mlx5dr_match_misc3 *misc3)
+{
+	return (misc3->outer_vxlan_gpe_vni ||
+		misc3->outer_vxlan_gpe_next_protocol ||
+		misc3->outer_vxlan_gpe_flags);
+}
+
+static bool
+dr_matcher_supp_vxlan_gpe(struct mlx5dr_cmd_caps *caps)
+{
+	return (caps->sw_format_ver >= MLX5_STEERING_FORMAT_CONNECTX_6DX) ||
+	       (caps->flex_protocols & MLX5_FLEX_PARSER_VXLAN_GPE_ENABLED);
+}
+
+static bool
+dr_mask_is_tnl_vxlan_gpe(struct mlx5dr_match_param *mask,
+			 struct mlx5dr_domain *dmn)
+{
+	return dr_mask_is_vxlan_gpe_set(&mask->misc3) &&
+	       dr_matcher_supp_vxlan_gpe(&dmn->info.caps);
+}
+
+static bool dr_mask_is_tnl_geneve_set(struct mlx5dr_match_misc *misc)
+{
+	return misc->geneve_vni ||
+	       misc->geneve_oam ||
+	       misc->geneve_protocol_type ||
+	       misc->geneve_opt_len;
+}
+
+static bool dr_mask_is_tnl_geneve_tlv_opt(struct mlx5dr_match_misc3 *misc3)
+{
+	return misc3->geneve_tlv_option_0_data;
+}
+
+static bool
+dr_matcher_supp_flex_parser_ok(struct mlx5dr_cmd_caps *caps)
+{
+	return caps->flex_parser_ok_bits_supp;
+}
+
+static bool dr_mask_is_tnl_geneve_tlv_opt_exist_set(struct mlx5dr_match_misc *misc,
+						    struct mlx5dr_domain *dmn)
+{
+	return dr_matcher_supp_flex_parser_ok(&dmn->info.caps) &&
+	       misc->geneve_tlv_option_0_exist;
+}
+
+static bool
+dr_matcher_supp_tnl_geneve(struct mlx5dr_cmd_caps *caps)
+{
+	return (caps->sw_format_ver >= MLX5_STEERING_FORMAT_CONNECTX_6DX) ||
+	       (caps->flex_protocols & MLX5_FLEX_PARSER_GENEVE_ENABLED);
+}
+
+static bool
+dr_mask_is_tnl_geneve(struct mlx5dr_match_param *mask,
+		      struct mlx5dr_domain *dmn)
+{
+	return dr_mask_is_tnl_geneve_set(&mask->misc) &&
+	       dr_matcher_supp_tnl_geneve(&dmn->info.caps);
+}
+
+static bool dr_mask_is_tnl_gtpu_set(struct mlx5dr_match_misc3 *misc3)
+{
+	return misc3->gtpu_msg_flags || misc3->gtpu_msg_type || misc3->gtpu_teid;
+}
+
+static bool dr_matcher_supp_tnl_gtpu(struct mlx5dr_cmd_caps *caps)
+{
+	return caps->flex_protocols & MLX5_FLEX_PARSER_GTPU_ENABLED;
+}
+
+static bool dr_mask_is_tnl_gtpu(struct mlx5dr_match_param *mask,
+				struct mlx5dr_domain *dmn)
+{
+	return dr_mask_is_tnl_gtpu_set(&mask->misc3) &&
+	       dr_matcher_supp_tnl_gtpu(&dmn->info.caps);
+}
+
+static int dr_matcher_supp_tnl_gtpu_dw_0(struct mlx5dr_cmd_caps *caps)
+{
+	return caps->flex_protocols & MLX5_FLEX_PARSER_GTPU_DW_0_ENABLED;
+}
+
+static bool dr_mask_is_tnl_gtpu_dw_0(struct mlx5dr_match_param *mask,
+				     struct mlx5dr_domain *dmn)
+{
+	return mask->misc3.gtpu_dw_0 &&
+	       dr_matcher_supp_tnl_gtpu_dw_0(&dmn->info.caps);
+}
+
+static int dr_matcher_supp_tnl_gtpu_teid(struct mlx5dr_cmd_caps *caps)
+{
+	return caps->flex_protocols & MLX5_FLEX_PARSER_GTPU_TEID_ENABLED;
+}
+
+static bool dr_mask_is_tnl_gtpu_teid(struct mlx5dr_match_param *mask,
+				     struct mlx5dr_domain *dmn)
+{
+	return mask->misc3.gtpu_teid &&
+	       dr_matcher_supp_tnl_gtpu_teid(&dmn->info.caps);
+}
+
+static int dr_matcher_supp_tnl_gtpu_dw_2(struct mlx5dr_cmd_caps *caps)
+{
+	return caps->flex_protocols & MLX5_FLEX_PARSER_GTPU_DW_2_ENABLED;
+}
+
+static bool dr_mask_is_tnl_gtpu_dw_2(struct mlx5dr_match_param *mask,
+				     struct mlx5dr_domain *dmn)
+{
+	return mask->misc3.gtpu_dw_2 &&
+	       dr_matcher_supp_tnl_gtpu_dw_2(&dmn->info.caps);
+}
+
+static int dr_matcher_supp_tnl_gtpu_first_ext(struct mlx5dr_cmd_caps *caps)
+{
+	return caps->flex_protocols & MLX5_FLEX_PARSER_GTPU_FIRST_EXT_DW_0_ENABLED;
+}
+
+static bool dr_mask_is_tnl_gtpu_first_ext(struct mlx5dr_match_param *mask,
+					  struct mlx5dr_domain *dmn)
+{
+	return mask->misc3.gtpu_first_ext_dw_0 &&
+	       dr_matcher_supp_tnl_gtpu_first_ext(&dmn->info.caps);
+}
+
+static bool dr_mask_is_tnl_gtpu_flex_parser_0(struct mlx5dr_match_param *mask,
+					      struct mlx5dr_domain *dmn)
+{
+	struct mlx5dr_cmd_caps *caps = &dmn->info.caps;
+
+	return (dr_is_flex_parser_0_id(caps->flex_parser_id_gtpu_dw_0) &&
+		dr_mask_is_tnl_gtpu_dw_0(mask, dmn)) ||
+	       (dr_is_flex_parser_0_id(caps->flex_parser_id_gtpu_teid) &&
+		dr_mask_is_tnl_gtpu_teid(mask, dmn)) ||
+	       (dr_is_flex_parser_0_id(caps->flex_parser_id_gtpu_dw_2) &&
+		dr_mask_is_tnl_gtpu_dw_2(mask, dmn)) ||
+	       (dr_is_flex_parser_0_id(caps->flex_parser_id_gtpu_first_ext_dw_0) &&
+		dr_mask_is_tnl_gtpu_first_ext(mask, dmn));
+}
+
+static bool dr_mask_is_tnl_gtpu_flex_parser_1(struct mlx5dr_match_param *mask,
+					      struct mlx5dr_domain *dmn)
+{
+	struct mlx5dr_cmd_caps *caps = &dmn->info.caps;
+
+	return (dr_is_flex_parser_1_id(caps->flex_parser_id_gtpu_dw_0) &&
+		dr_mask_is_tnl_gtpu_dw_0(mask, dmn)) ||
+	       (dr_is_flex_parser_1_id(caps->flex_parser_id_gtpu_teid) &&
+		dr_mask_is_tnl_gtpu_teid(mask, dmn)) ||
+	       (dr_is_flex_parser_1_id(caps->flex_parser_id_gtpu_dw_2) &&
+		dr_mask_is_tnl_gtpu_dw_2(mask, dmn)) ||
+	       (dr_is_flex_parser_1_id(caps->flex_parser_id_gtpu_first_ext_dw_0) &&
+		dr_mask_is_tnl_gtpu_first_ext(mask, dmn));
+}
+
+static bool dr_mask_is_tnl_gtpu_any(struct mlx5dr_match_param *mask,
+				    struct mlx5dr_domain *dmn)
+{
+	return dr_mask_is_tnl_gtpu_flex_parser_0(mask, dmn) ||
+	       dr_mask_is_tnl_gtpu_flex_parser_1(mask, dmn) ||
+	       dr_mask_is_tnl_gtpu(mask, dmn);
+}
+
+static int dr_matcher_supp_icmp_v4(struct mlx5dr_cmd_caps *caps)
+{
+	return (caps->sw_format_ver >= MLX5_STEERING_FORMAT_CONNECTX_6DX) ||
+	       (caps->flex_protocols & MLX5_FLEX_PARSER_ICMP_V4_ENABLED);
+}
+
+static int dr_matcher_supp_icmp_v6(struct mlx5dr_cmd_caps *caps)
+{
+	return (caps->sw_format_ver >= MLX5_STEERING_FORMAT_CONNECTX_6DX) ||
+	       (caps->flex_protocols & MLX5_FLEX_PARSER_ICMP_V6_ENABLED);
+}
+
+static bool dr_mask_is_icmpv6_set(struct mlx5dr_match_misc3 *misc3)
+{
+	return (misc3->icmpv6_type || misc3->icmpv6_code ||
+		misc3->icmpv6_header_data);
+}
+
+static bool dr_mask_is_icmp(struct mlx5dr_match_param *mask,
+			    struct mlx5dr_domain *dmn)
+{
+	if (DR_MASK_IS_ICMPV4_SET(&mask->misc3))
+		return dr_matcher_supp_icmp_v4(&dmn->info.caps);
+	else if (dr_mask_is_icmpv6_set(&mask->misc3))
+		return dr_matcher_supp_icmp_v6(&dmn->info.caps);
+
+	return false;
+}
+
+static bool dr_mask_is_wqe_metadata_set(struct mlx5dr_match_misc2 *misc2)
+{
+	return misc2->metadata_reg_a;
+}
+
+static bool dr_mask_is_reg_c_0_3_set(struct mlx5dr_match_misc2 *misc2)
+{
+	return (misc2->metadata_reg_c_0 || misc2->metadata_reg_c_1 ||
+		misc2->metadata_reg_c_2 || misc2->metadata_reg_c_3);
+}
+
+static bool dr_mask_is_reg_c_4_7_set(struct mlx5dr_match_misc2 *misc2)
+{
+	return (misc2->metadata_reg_c_4 || misc2->metadata_reg_c_5 ||
+		misc2->metadata_reg_c_6 || misc2->metadata_reg_c_7);
+}
+
+static bool dr_mask_is_gvmi_or_qpn_set(struct mlx5dr_match_misc *misc)
+{
+	return (misc->source_sqn || misc->source_port);
+}
+
+static bool dr_mask_is_flex_parser_id_0_3_set(u32 flex_parser_id,
+					      u32 flex_parser_value)
+{
+	if (flex_parser_id)
+		return flex_parser_id <= DR_STE_MAX_FLEX_0_ID;
+
+	/* Using flex_parser 0 means that id is zero, thus value must be set. */
+	return flex_parser_value;
+}
+
+static bool dr_mask_is_flex_parser_0_3_set(struct mlx5dr_match_misc4 *misc4)
+{
+	return (dr_mask_is_flex_parser_id_0_3_set(misc4->prog_sample_field_id_0,
+						  misc4->prog_sample_field_value_0) ||
+		dr_mask_is_flex_parser_id_0_3_set(misc4->prog_sample_field_id_1,
+						  misc4->prog_sample_field_value_1) ||
+		dr_mask_is_flex_parser_id_0_3_set(misc4->prog_sample_field_id_2,
+						  misc4->prog_sample_field_value_2) ||
+		dr_mask_is_flex_parser_id_0_3_set(misc4->prog_sample_field_id_3,
+						  misc4->prog_sample_field_value_3));
+}
+
+static bool dr_mask_is_flex_parser_id_4_7_set(u32 flex_parser_id)
+{
+	return flex_parser_id > DR_STE_MAX_FLEX_0_ID &&
+	       flex_parser_id <= DR_STE_MAX_FLEX_1_ID;
+}
+
+static bool dr_mask_is_flex_parser_4_7_set(struct mlx5dr_match_misc4 *misc4)
+{
+	return (dr_mask_is_flex_parser_id_4_7_set(misc4->prog_sample_field_id_0) ||
+		dr_mask_is_flex_parser_id_4_7_set(misc4->prog_sample_field_id_1) ||
+		dr_mask_is_flex_parser_id_4_7_set(misc4->prog_sample_field_id_2) ||
+		dr_mask_is_flex_parser_id_4_7_set(misc4->prog_sample_field_id_3));
+}
+
+static int dr_matcher_supp_tnl_mpls_over_gre(struct mlx5dr_cmd_caps *caps)
+{
+	return caps->flex_protocols & MLX5_FLEX_PARSER_MPLS_OVER_GRE_ENABLED;
+}
+
+static bool dr_mask_is_tnl_mpls_over_gre(struct mlx5dr_match_param *mask,
+					 struct mlx5dr_domain *dmn)
+{
+	return DR_MASK_IS_OUTER_MPLS_OVER_GRE_SET(&mask->misc2) &&
+	       dr_matcher_supp_tnl_mpls_over_gre(&dmn->info.caps);
+}
+
+static int dr_matcher_supp_tnl_mpls_over_udp(struct mlx5dr_cmd_caps *caps)
+{
+	return caps->flex_protocols & MLX5_FLEX_PARSER_MPLS_OVER_UDP_ENABLED;
+}
+
+static bool dr_mask_is_tnl_mpls_over_udp(struct mlx5dr_match_param *mask,
+					 struct mlx5dr_domain *dmn)
+{
+	return DR_MASK_IS_OUTER_MPLS_OVER_UDP_SET(&mask->misc2) &&
+	       dr_matcher_supp_tnl_mpls_over_udp(&dmn->info.caps);
+}
+
+static bool dr_mask_is_tnl_header_0_1_set(struct mlx5dr_match_misc5 *misc5)
+{
+	return misc5->tunnel_header_0 || misc5->tunnel_header_1;
+}
+
+int mlx5dr_matcher_select_builders(struct mlx5dr_matcher *matcher,
+				   struct mlx5dr_matcher_rx_tx *nic_matcher,
+				   enum mlx5dr_ipv outer_ipv,
+				   enum mlx5dr_ipv inner_ipv)
+{
+	nic_matcher->ste_builder =
+		nic_matcher->ste_builder_arr[outer_ipv][inner_ipv];
+	nic_matcher->num_of_builders =
+		nic_matcher->num_of_builders_arr[outer_ipv][inner_ipv];
+
+	if (!nic_matcher->num_of_builders) {
+		mlx5dr_dbg(matcher->tbl->dmn,
+			   "Rule not supported on this matcher due to IP related fields\n");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int dr_matcher_set_ste_builders(struct mlx5dr_matcher *matcher,
+				       struct mlx5dr_matcher_rx_tx *nic_matcher,
+				       enum mlx5dr_ipv outer_ipv,
+				       enum mlx5dr_ipv inner_ipv)
+{
+	struct mlx5dr_domain_rx_tx *nic_dmn = nic_matcher->nic_tbl->nic_dmn;
+	struct mlx5dr_domain *dmn = matcher->tbl->dmn;
+	struct mlx5dr_ste_ctx *ste_ctx = dmn->ste_ctx;
+	struct mlx5dr_match_param mask = {};
+	bool allow_empty_match = false;
+	struct mlx5dr_ste_build *sb;
+	bool inner, rx;
+	int idx = 0;
+	int ret, i;
+
+	sb = nic_matcher->ste_builder_arr[outer_ipv][inner_ipv];
+	rx = nic_dmn->type == DR_DOMAIN_NIC_TYPE_RX;
+
+	/* Create a temporary mask to track and clear used mask fields */
+	if (matcher->match_criteria & DR_MATCHER_CRITERIA_OUTER)
+		mask.outer = matcher->mask.outer;
+
+	if (matcher->match_criteria & DR_MATCHER_CRITERIA_MISC)
+		mask.misc = matcher->mask.misc;
+
+	if (matcher->match_criteria & DR_MATCHER_CRITERIA_INNER)
+		mask.inner = matcher->mask.inner;
+
+	if (matcher->match_criteria & DR_MATCHER_CRITERIA_MISC2)
+		mask.misc2 = matcher->mask.misc2;
+
+	if (matcher->match_criteria & DR_MATCHER_CRITERIA_MISC3)
+		mask.misc3 = matcher->mask.misc3;
+
+	if (matcher->match_criteria & DR_MATCHER_CRITERIA_MISC4)
+		mask.misc4 = matcher->mask.misc4;
+
+	if (matcher->match_criteria & DR_MATCHER_CRITERIA_MISC5)
+		mask.misc5 = matcher->mask.misc5;
+
+	ret = mlx5dr_ste_build_pre_check(dmn, matcher->match_criteria,
+					 &matcher->mask, NULL);
+	if (ret)
+		return ret;
+
+	/* Optimize RX pipe by reducing source port match, since
+	 * the FDB RX part is connected only to the wire.
+	 */
+	if (dmn->type == MLX5DR_DOMAIN_TYPE_FDB &&
+	    rx && mask.misc.source_port) {
+		mask.misc.source_port = 0;
+		mask.misc.source_eswitch_owner_vhca_id = 0;
+		allow_empty_match = true;
+	}
+
+	/* Outer */
+	if (matcher->match_criteria & (DR_MATCHER_CRITERIA_OUTER |
+				       DR_MATCHER_CRITERIA_MISC |
+				       DR_MATCHER_CRITERIA_MISC2 |
+				       DR_MATCHER_CRITERIA_MISC3 |
+				       DR_MATCHER_CRITERIA_MISC5)) {
+		inner = false;
+
+		if (dr_mask_is_wqe_metadata_set(&mask.misc2))
+			mlx5dr_ste_build_general_purpose(ste_ctx, &sb[idx++],
+							 &mask, inner, rx);
+
+		if (dr_mask_is_reg_c_0_3_set(&mask.misc2))
+			mlx5dr_ste_build_register_0(ste_ctx, &sb[idx++],
+						    &mask, inner, rx);
+
+		if (dr_mask_is_reg_c_4_7_set(&mask.misc2))
+			mlx5dr_ste_build_register_1(ste_ctx, &sb[idx++],
+						    &mask, inner, rx);
+
+		if (dr_mask_is_gvmi_or_qpn_set(&mask.misc) &&
+		    (dmn->type == MLX5DR_DOMAIN_TYPE_FDB ||
+		     dmn->type == MLX5DR_DOMAIN_TYPE_NIC_RX)) {
+			mlx5dr_ste_build_src_gvmi_qpn(ste_ctx, &sb[idx++],
+						      &mask, dmn, inner, rx);
+		}
+
+		if (dr_mask_is_smac_set(&mask.outer) &&
+		    dr_mask_is_dmac_set(&mask.outer)) {
+			mlx5dr_ste_build_eth_l2_src_dst(ste_ctx, &sb[idx++],
+							&mask, inner, rx);
+		}
+
+		if (dr_mask_is_smac_set(&mask.outer))
+			mlx5dr_ste_build_eth_l2_src(ste_ctx, &sb[idx++],
+						    &mask, inner, rx);
+
+		if (DR_MASK_IS_L2_DST(mask.outer, mask.misc, outer))
+			mlx5dr_ste_build_eth_l2_dst(ste_ctx, &sb[idx++],
+						    &mask, inner, rx);
+
+		if (outer_ipv == DR_RULE_IPV6) {
+			if (DR_MASK_IS_DST_IP_SET(&mask.outer))
+				mlx5dr_ste_build_eth_l3_ipv6_dst(ste_ctx, &sb[idx++],
+								 &mask, inner, rx);
+
+			if (DR_MASK_IS_SRC_IP_SET(&mask.outer))
+				mlx5dr_ste_build_eth_l3_ipv6_src(ste_ctx, &sb[idx++],
+								 &mask, inner, rx);
+
+			if (DR_MASK_IS_ETH_L4_SET(mask.outer, mask.misc, outer))
+				mlx5dr_ste_build_eth_ipv6_l3_l4(ste_ctx, &sb[idx++],
+								&mask, inner, rx);
+		} else {
+			if (dr_mask_is_ipv4_5_tuple_set(&mask.outer))
+				mlx5dr_ste_build_eth_l3_ipv4_5_tuple(ste_ctx, &sb[idx++],
+								     &mask, inner, rx);
+
+			if (dr_mask_is_ttl_set(&mask.outer) ||
+			    dr_mask_is_ipv4_ihl_set(&mask.outer))
+				mlx5dr_ste_build_eth_l3_ipv4_misc(ste_ctx, &sb[idx++],
+								  &mask, inner, rx);
+		}
+
+		if (dr_mask_is_tnl_vxlan_gpe(&mask, dmn))
+			mlx5dr_ste_build_tnl_vxlan_gpe(ste_ctx, &sb[idx++],
+						       &mask, inner, rx);
+		else if (dr_mask_is_tnl_geneve(&mask, dmn)) {
+			mlx5dr_ste_build_tnl_geneve(ste_ctx, &sb[idx++],
+						    &mask, inner, rx);
+			if (dr_mask_is_tnl_geneve_tlv_opt(&mask.misc3))
+				mlx5dr_ste_build_tnl_geneve_tlv_opt(ste_ctx, &sb[idx++],
+								    &mask, &dmn->info.caps,
+								    inner, rx);
+			if (dr_mask_is_tnl_geneve_tlv_opt_exist_set(&mask.misc, dmn))
+				mlx5dr_ste_build_tnl_geneve_tlv_opt_exist(ste_ctx, &sb[idx++],
+									  &mask, &dmn->info.caps,
+									  inner, rx);
+		} else if (dr_mask_is_tnl_gtpu_any(&mask, dmn)) {
+			if (dr_mask_is_tnl_gtpu_flex_parser_0(&mask, dmn))
+				mlx5dr_ste_build_tnl_gtpu_flex_parser_0(ste_ctx, &sb[idx++],
+									&mask, &dmn->info.caps,
+									inner, rx);
+
+			if (dr_mask_is_tnl_gtpu_flex_parser_1(&mask, dmn))
+				mlx5dr_ste_build_tnl_gtpu_flex_parser_1(ste_ctx, &sb[idx++],
+									&mask, &dmn->info.caps,
+									inner, rx);
+
+			if (dr_mask_is_tnl_gtpu(&mask, dmn))
+				mlx5dr_ste_build_tnl_gtpu(ste_ctx, &sb[idx++],
+							  &mask, inner, rx);
+		} else if (dr_mask_is_tnl_header_0_1_set(&mask.misc5)) {
+			mlx5dr_ste_build_tnl_header_0_1(ste_ctx, &sb[idx++],
+							&mask, inner, rx);
+		}
+
+		if (DR_MASK_IS_ETH_L4_MISC_SET(mask.misc3, outer))
+			mlx5dr_ste_build_eth_l4_misc(ste_ctx, &sb[idx++],
+						     &mask, inner, rx);
+
+		if (DR_MASK_IS_FIRST_MPLS_SET(mask.misc2, outer))
+			mlx5dr_ste_build_mpls(ste_ctx, &sb[idx++],
+					      &mask, inner, rx);
+
+		if (dr_mask_is_tnl_mpls_over_gre(&mask, dmn))
+			mlx5dr_ste_build_tnl_mpls_over_gre(ste_ctx, &sb[idx++],
+							   &mask, &dmn->info.caps,
+							   inner, rx);
+		else if (dr_mask_is_tnl_mpls_over_udp(&mask, dmn))
+			mlx5dr_ste_build_tnl_mpls_over_udp(ste_ctx, &sb[idx++],
+							   &mask, &dmn->info.caps,
+							   inner, rx);
+
+		if (dr_mask_is_icmp(&mask, dmn))
+			mlx5dr_ste_build_icmp(ste_ctx, &sb[idx++],
+					      &mask, &dmn->info.caps,
+					      inner, rx);
+
+		if (dr_mask_is_tnl_gre_set(&mask.misc))
+			mlx5dr_ste_build_tnl_gre(ste_ctx, &sb[idx++],
+						 &mask, inner, rx);
+	}
+
+	/* Inner */
+	if (matcher->match_criteria & (DR_MATCHER_CRITERIA_INNER |
+				       DR_MATCHER_CRITERIA_MISC |
+				       DR_MATCHER_CRITERIA_MISC2 |
+				       DR_MATCHER_CRITERIA_MISC3)) {
+		inner = true;
+
+		if (dr_mask_is_eth_l2_tnl_set(&mask.misc))
+			mlx5dr_ste_build_eth_l2_tnl(ste_ctx, &sb[idx++],
+						    &mask, inner, rx);
+
+		if (dr_mask_is_smac_set(&mask.inner) &&
+		    dr_mask_is_dmac_set(&mask.inner)) {
+			mlx5dr_ste_build_eth_l2_src_dst(ste_ctx, &sb[idx++],
+							&mask, inner, rx);
+		}
+
+		if (dr_mask_is_smac_set(&mask.inner))
+			mlx5dr_ste_build_eth_l2_src(ste_ctx, &sb[idx++],
+						    &mask, inner, rx);
+
+		if (DR_MASK_IS_L2_DST(mask.inner, mask.misc, inner))
+			mlx5dr_ste_build_eth_l2_dst(ste_ctx, &sb[idx++],
+						    &mask, inner, rx);
+
+		if (inner_ipv == DR_RULE_IPV6) {
+			if (DR_MASK_IS_DST_IP_SET(&mask.inner))
+				mlx5dr_ste_build_eth_l3_ipv6_dst(ste_ctx, &sb[idx++],
+								 &mask, inner, rx);
+
+			if (DR_MASK_IS_SRC_IP_SET(&mask.inner))
+				mlx5dr_ste_build_eth_l3_ipv6_src(ste_ctx, &sb[idx++],
+								 &mask, inner, rx);
+
+			if (DR_MASK_IS_ETH_L4_SET(mask.inner, mask.misc, inner))
+				mlx5dr_ste_build_eth_ipv6_l3_l4(ste_ctx, &sb[idx++],
+								&mask, inner, rx);
+		} else {
+			if (dr_mask_is_ipv4_5_tuple_set(&mask.inner))
+				mlx5dr_ste_build_eth_l3_ipv4_5_tuple(ste_ctx, &sb[idx++],
+								     &mask, inner, rx);
+
+			if (dr_mask_is_ttl_set(&mask.inner) ||
+			    dr_mask_is_ipv4_ihl_set(&mask.inner))
+				mlx5dr_ste_build_eth_l3_ipv4_misc(ste_ctx, &sb[idx++],
+								  &mask, inner, rx);
+		}
+
+		if (DR_MASK_IS_ETH_L4_MISC_SET(mask.misc3, inner))
+			mlx5dr_ste_build_eth_l4_misc(ste_ctx, &sb[idx++],
+						     &mask, inner, rx);
+
+		if (DR_MASK_IS_FIRST_MPLS_SET(mask.misc2, inner))
+			mlx5dr_ste_build_mpls(ste_ctx, &sb[idx++],
+					      &mask, inner, rx);
+
+		if (dr_mask_is_tnl_mpls_over_gre(&mask, dmn))
+			mlx5dr_ste_build_tnl_mpls_over_gre(ste_ctx, &sb[idx++],
+							   &mask, &dmn->info.caps,
+							   inner, rx);
+		else if (dr_mask_is_tnl_mpls_over_udp(&mask, dmn))
+			mlx5dr_ste_build_tnl_mpls_over_udp(ste_ctx, &sb[idx++],
+							   &mask, &dmn->info.caps,
+							   inner, rx);
+	}
+
+	if (matcher->match_criteria & DR_MATCHER_CRITERIA_MISC4) {
+		if (dr_mask_is_flex_parser_0_3_set(&mask.misc4))
+			mlx5dr_ste_build_flex_parser_0(ste_ctx, &sb[idx++],
+						       &mask, false, rx);
+
+		if (dr_mask_is_flex_parser_4_7_set(&mask.misc4))
+			mlx5dr_ste_build_flex_parser_1(ste_ctx, &sb[idx++],
+						       &mask, false, rx);
+	}
+
+	/* Empty matcher, takes all */
+	if ((!idx && allow_empty_match) ||
+	    matcher->match_criteria == DR_MATCHER_CRITERIA_EMPTY)
+		mlx5dr_ste_build_empty_always_hit(&sb[idx++], rx);
+
+	if (idx == 0) {
+		mlx5dr_err(dmn, "Cannot generate any valid rules from mask\n");
+		return -EINVAL;
+	}
+
+	/* Check that all mask fields were consumed */
+	for (i = 0; i < sizeof(struct mlx5dr_match_param); i++) {
+		if (((u8 *)&mask)[i] != 0) {
+			mlx5dr_dbg(dmn, "Mask contains unsupported parameters\n");
+			return -EOPNOTSUPP;
+		}
+	}
+
+	nic_matcher->ste_builder = sb;
+	nic_matcher->num_of_builders_arr[outer_ipv][inner_ipv] = idx;
+
+	return 0;
+}
+
+static int dr_nic_matcher_connect(struct mlx5dr_domain *dmn,
+				  struct mlx5dr_matcher_rx_tx *curr_nic_matcher,
+				  struct mlx5dr_matcher_rx_tx *next_nic_matcher,
+				  struct mlx5dr_matcher_rx_tx *prev_nic_matcher)
+{
+	struct mlx5dr_table_rx_tx *nic_tbl = curr_nic_matcher->nic_tbl;
+	struct mlx5dr_domain_rx_tx *nic_dmn = nic_tbl->nic_dmn;
+	struct mlx5dr_htbl_connect_info info;
+	struct mlx5dr_ste_htbl *prev_htbl;
+	int ret;
+
+	/* Connect end anchor hash table to next_htbl or to the default address */
+	if (next_nic_matcher) {
+		info.type = CONNECT_HIT;
+		info.hit_next_htbl = next_nic_matcher->s_htbl;
+	} else {
+		info.type = CONNECT_MISS;
+		info.miss_icm_addr = nic_tbl->default_icm_addr;
+	}
+	ret = mlx5dr_ste_htbl_init_and_postsend(dmn, nic_dmn,
+						curr_nic_matcher->e_anchor,
+						&info, info.type == CONNECT_HIT);
+	if (ret)
+		return ret;
+
+	/* Connect start hash table to end anchor */
+	info.type = CONNECT_MISS;
+	info.miss_icm_addr = mlx5dr_icm_pool_get_chunk_icm_addr(curr_nic_matcher->e_anchor->chunk);
+	ret = mlx5dr_ste_htbl_init_and_postsend(dmn, nic_dmn,
+						curr_nic_matcher->s_htbl,
+						&info, false);
+	if (ret)
+		return ret;
+
+	/* Connect previous hash table to matcher start hash table */
+	if (prev_nic_matcher)
+		prev_htbl = prev_nic_matcher->e_anchor;
+	else
+		prev_htbl = nic_tbl->s_anchor;
+
+	info.type = CONNECT_HIT;
+	info.hit_next_htbl = curr_nic_matcher->s_htbl;
+	ret = mlx5dr_ste_htbl_init_and_postsend(dmn, nic_dmn, prev_htbl,
+						&info, true);
+	if (ret)
+		return ret;
+
+	/* Update the pointing ste and next hash table */
+	curr_nic_matcher->s_htbl->pointing_ste = prev_htbl->chunk->ste_arr;
+	prev_htbl->chunk->ste_arr[0].next_htbl = curr_nic_matcher->s_htbl;
+
+	if (next_nic_matcher) {
+		next_nic_matcher->s_htbl->pointing_ste =
+			curr_nic_matcher->e_anchor->chunk->ste_arr;
+		curr_nic_matcher->e_anchor->chunk->ste_arr[0].next_htbl =
+			next_nic_matcher->s_htbl;
+	}
+
+	return 0;
+}
+
+int mlx5dr_matcher_add_to_tbl_nic(struct mlx5dr_domain *dmn,
+				  struct mlx5dr_matcher_rx_tx *nic_matcher)
+{
+	struct mlx5dr_matcher_rx_tx *next_nic_matcher, *prev_nic_matcher, *tmp_nic_matcher;
+	struct mlx5dr_table_rx_tx *nic_tbl = nic_matcher->nic_tbl;
+	bool first = true;
+	int ret;
+
+	/* If the nic matcher is already on its parent nic table list,
+	 * then it is already connected to the chain of nic matchers.
+	 */
+	if (!list_empty(&nic_matcher->list_node))
+		return 0;
+
+	next_nic_matcher = NULL;
+	list_for_each_entry(tmp_nic_matcher, &nic_tbl->nic_matcher_list, list_node) {
+		if (tmp_nic_matcher->prio >= nic_matcher->prio) {
+			next_nic_matcher = tmp_nic_matcher;
+			break;
+		}
+		first = false;
+	}
+
+	prev_nic_matcher = NULL;
+	if (next_nic_matcher && !first)
+		prev_nic_matcher = list_prev_entry(next_nic_matcher, list_node);
+	else if (!first)
+		prev_nic_matcher = list_last_entry(&nic_tbl->nic_matcher_list,
+						   struct mlx5dr_matcher_rx_tx,
+						   list_node);
+
+	ret = dr_nic_matcher_connect(dmn, nic_matcher,
+				     next_nic_matcher, prev_nic_matcher);
+	if (ret)
+		return ret;
+
+	if (prev_nic_matcher)
+		list_add(&nic_matcher->list_node, &prev_nic_matcher->list_node);
+	else if (next_nic_matcher)
+		list_add_tail(&nic_matcher->list_node, &next_nic_matcher->list_node);
+	else
+		list_add(&nic_matcher->list_node, &nic_matcher->nic_tbl->nic_matcher_list);
+
+	return ret;
+}
+
+static void dr_matcher_uninit_nic(struct mlx5dr_matcher_rx_tx *nic_matcher)
+{
+	mlx5dr_htbl_put(nic_matcher->s_htbl);
+	mlx5dr_htbl_put(nic_matcher->e_anchor);
+}
+
+static void dr_matcher_uninit_fdb(struct mlx5dr_matcher *matcher)
+{
+	dr_matcher_uninit_nic(&matcher->rx);
+	dr_matcher_uninit_nic(&matcher->tx);
+}
+
+static void dr_matcher_uninit(struct mlx5dr_matcher *matcher)
+{
+	struct mlx5dr_domain *dmn = matcher->tbl->dmn;
+
+	switch (dmn->type) {
+	case MLX5DR_DOMAIN_TYPE_NIC_RX:
+		dr_matcher_uninit_nic(&matcher->rx);
+		break;
+	case MLX5DR_DOMAIN_TYPE_NIC_TX:
+		dr_matcher_uninit_nic(&matcher->tx);
+		break;
+	case MLX5DR_DOMAIN_TYPE_FDB:
+		dr_matcher_uninit_fdb(matcher);
+		break;
+	default:
+		WARN_ON(true);
+		break;
+	}
+}
+
+static int dr_matcher_set_all_ste_builders(struct mlx5dr_matcher *matcher,
+					   struct mlx5dr_matcher_rx_tx *nic_matcher)
+{
+	struct mlx5dr_domain *dmn = matcher->tbl->dmn;
+
+	dr_matcher_set_ste_builders(matcher, nic_matcher, DR_RULE_IPV4, DR_RULE_IPV4);
+	dr_matcher_set_ste_builders(matcher, nic_matcher, DR_RULE_IPV4, DR_RULE_IPV6);
+	dr_matcher_set_ste_builders(matcher, nic_matcher, DR_RULE_IPV6, DR_RULE_IPV4);
+	dr_matcher_set_ste_builders(matcher, nic_matcher, DR_RULE_IPV6, DR_RULE_IPV6);
+
+	if (!nic_matcher->ste_builder) {
+		mlx5dr_err(dmn, "Cannot generate IPv4 or IPv6 rules with given mask\n");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int dr_matcher_init_nic(struct mlx5dr_matcher *matcher,
+			       struct mlx5dr_matcher_rx_tx *nic_matcher)
+{
+	struct mlx5dr_domain *dmn = matcher->tbl->dmn;
+	int ret;
+
+	nic_matcher->prio = matcher->prio;
+	INIT_LIST_HEAD(&nic_matcher->list_node);
+
+	ret = dr_matcher_set_all_ste_builders(matcher, nic_matcher);
+	if (ret)
+		return ret;
+
+	nic_matcher->e_anchor = mlx5dr_ste_htbl_alloc(dmn->ste_icm_pool,
+						      DR_CHUNK_SIZE_1,
+						      MLX5DR_STE_LU_TYPE_DONT_CARE,
+						      0);
+	if (!nic_matcher->e_anchor)
+		return -ENOMEM;
+
+	nic_matcher->s_htbl = mlx5dr_ste_htbl_alloc(dmn->ste_icm_pool,
+						    DR_CHUNK_SIZE_1,
+						    nic_matcher->ste_builder[0].lu_type,
+						    nic_matcher->ste_builder[0].byte_mask);
+	if (!nic_matcher->s_htbl) {
+		ret = -ENOMEM;
+		goto free_e_htbl;
+	}
+
+	/* make sure the tables exist while empty */
+	mlx5dr_htbl_get(nic_matcher->s_htbl);
+	mlx5dr_htbl_get(nic_matcher->e_anchor);
+
+	return 0;
+
+free_e_htbl:
+	mlx5dr_ste_htbl_free(nic_matcher->e_anchor);
+	return ret;
+}
+
+static int dr_matcher_init_fdb(struct mlx5dr_matcher *matcher)
+{
+	int ret;
+
+	ret = dr_matcher_init_nic(matcher, &matcher->rx);
+	if (ret)
+		return ret;
+
+	ret = dr_matcher_init_nic(matcher, &matcher->tx);
+	if (ret)
+		goto uninit_nic_rx;
+
+	return 0;
+
+uninit_nic_rx:
+	dr_matcher_uninit_nic(&matcher->rx);
+	return ret;
+}
+
+static int dr_matcher_copy_param(struct mlx5dr_matcher *matcher,
+				 struct mlx5dr_match_parameters *mask)
+{
+	struct mlx5dr_domain *dmn = matcher->tbl->dmn;
+	struct mlx5dr_match_parameters consumed_mask;
+	int i, ret = 0;
+
+	if (matcher->match_criteria >= DR_MATCHER_CRITERIA_MAX) {
+		mlx5dr_err(dmn, "Invalid match criteria attribute\n");
+		return -EINVAL;
+	}
+
+	if (mask) {
+		if (mask->match_sz > DR_SZ_MATCH_PARAM) {
+			mlx5dr_err(dmn, "Invalid match size attribute\n");
+			return -EINVAL;
+		}
+
+		consumed_mask.match_buf = kzalloc(mask->match_sz, GFP_KERNEL);
+		if (!consumed_mask.match_buf)
+			return -ENOMEM;
+
+		consumed_mask.match_sz = mask->match_sz;
+		memcpy(consumed_mask.match_buf, mask->match_buf, mask->match_sz);
+		mlx5dr_ste_copy_param(matcher->match_criteria,
+				      &matcher->mask, &consumed_mask, true);
+
+		/* Check that all mask data was consumed */
+		for (i = 0; i < consumed_mask.match_sz; i++) {
+			if (!((u8 *)consumed_mask.match_buf)[i])
+				continue;
+
+			mlx5dr_dbg(dmn,
+				   "Match param mask contains unsupported parameters\n");
+			ret = -EOPNOTSUPP;
+			break;
+		}
+
+		kfree(consumed_mask.match_buf);
+	}
+
+	return ret;
+}
+
+static int dr_matcher_init(struct mlx5dr_matcher *matcher,
+			   struct mlx5dr_match_parameters *mask)
+{
+	struct mlx5dr_table *tbl = matcher->tbl;
+	struct mlx5dr_domain *dmn = tbl->dmn;
+	int ret;
+
+	ret = dr_matcher_copy_param(matcher, mask);
+	if (ret)
+		return ret;
+
+	switch (dmn->type) {
+	case MLX5DR_DOMAIN_TYPE_NIC_RX:
+		matcher->rx.nic_tbl = &tbl->rx;
+		ret = dr_matcher_init_nic(matcher, &matcher->rx);
+		break;
+	case MLX5DR_DOMAIN_TYPE_NIC_TX:
+		matcher->tx.nic_tbl = &tbl->tx;
+		ret = dr_matcher_init_nic(matcher, &matcher->tx);
+		break;
+	case MLX5DR_DOMAIN_TYPE_FDB:
+		matcher->rx.nic_tbl = &tbl->rx;
+		matcher->tx.nic_tbl = &tbl->tx;
+		ret = dr_matcher_init_fdb(matcher);
+		break;
+	default:
+		WARN_ON(true);
+		ret = -EINVAL;
+	}
+
+	return ret;
+}
+
+static void dr_matcher_add_to_dbg_list(struct mlx5dr_matcher *matcher)
+{
+	mutex_lock(&matcher->tbl->dmn->dump_info.dbg_mutex);
+	list_add(&matcher->list_node, &matcher->tbl->matcher_list);
+	mutex_unlock(&matcher->tbl->dmn->dump_info.dbg_mutex);
+}
+
+static void dr_matcher_remove_from_dbg_list(struct mlx5dr_matcher *matcher)
+{
+	mutex_lock(&matcher->tbl->dmn->dump_info.dbg_mutex);
+	list_del(&matcher->list_node);
+	mutex_unlock(&matcher->tbl->dmn->dump_info.dbg_mutex);
+}
+
+struct mlx5dr_matcher *
+mlx5dr_matcher_create(struct mlx5dr_table *tbl,
+		      u32 priority,
+		      u8 match_criteria_enable,
+		      struct mlx5dr_match_parameters *mask)
+{
+	struct mlx5dr_matcher *matcher;
+	int ret;
+
+	refcount_inc(&tbl->refcount);
+
+	matcher = kzalloc(sizeof(*matcher), GFP_KERNEL);
+	if (!matcher)
+		goto dec_ref;
+
+	matcher->tbl = tbl;
+	matcher->prio = priority;
+	matcher->match_criteria = match_criteria_enable;
+	refcount_set(&matcher->refcount, 1);
+	INIT_LIST_HEAD(&matcher->list_node);
+	INIT_LIST_HEAD(&matcher->dbg_rule_list);
+
+	mlx5dr_domain_lock(tbl->dmn);
+
+	ret = dr_matcher_init(matcher, mask);
+	if (ret)
+		goto free_matcher;
+
+	dr_matcher_add_to_dbg_list(matcher);
+
+	mlx5dr_domain_unlock(tbl->dmn);
+
+	return matcher;
+
+free_matcher:
+	mlx5dr_domain_unlock(tbl->dmn);
+	kfree(matcher);
+dec_ref:
+	refcount_dec(&tbl->refcount);
+	return NULL;
+}
+
+static int dr_matcher_disconnect_nic(struct mlx5dr_domain *dmn,
+				     struct mlx5dr_table_rx_tx *nic_tbl,
+				     struct mlx5dr_matcher_rx_tx *next_nic_matcher,
+				     struct mlx5dr_matcher_rx_tx *prev_nic_matcher)
+{
+	struct mlx5dr_domain_rx_tx *nic_dmn = nic_tbl->nic_dmn;
+	struct mlx5dr_htbl_connect_info info;
+	struct mlx5dr_ste_htbl *prev_anchor;
+
+	if (prev_nic_matcher)
+		prev_anchor = prev_nic_matcher->e_anchor;
+	else
+		prev_anchor = nic_tbl->s_anchor;
+
+	/* Connect previous anchor hash table to next matcher or to the default address */
+	if (next_nic_matcher) {
+		info.type = CONNECT_HIT;
+		info.hit_next_htbl = next_nic_matcher->s_htbl;
+		next_nic_matcher->s_htbl->pointing_ste = prev_anchor->chunk->ste_arr;
+		prev_anchor->chunk->ste_arr[0].next_htbl = next_nic_matcher->s_htbl;
+	} else {
+		info.type = CONNECT_MISS;
+		info.miss_icm_addr = nic_tbl->default_icm_addr;
+		prev_anchor->chunk->ste_arr[0].next_htbl = NULL;
+	}
+
+	return mlx5dr_ste_htbl_init_and_postsend(dmn, nic_dmn, prev_anchor,
+						 &info, true);
+}
+
+int mlx5dr_matcher_remove_from_tbl_nic(struct mlx5dr_domain *dmn,
+				       struct mlx5dr_matcher_rx_tx *nic_matcher)
+{
+	struct mlx5dr_matcher_rx_tx *prev_nic_matcher, *next_nic_matcher;
+	struct mlx5dr_table_rx_tx *nic_tbl = nic_matcher->nic_tbl;
+	int ret;
+
+	/* If the nic matcher is not on its parent nic table list,
+	 * then it is detached - no need to disconnect it.
+	 */
+	if (list_empty(&nic_matcher->list_node))
+		return 0;
+
+	if (list_is_last(&nic_matcher->list_node, &nic_tbl->nic_matcher_list))
+		next_nic_matcher = NULL;
+	else
+		next_nic_matcher = list_next_entry(nic_matcher, list_node);
+
+	if (nic_matcher->list_node.prev == &nic_tbl->nic_matcher_list)
+		prev_nic_matcher = NULL;
+	else
+		prev_nic_matcher = list_prev_entry(nic_matcher, list_node);
+
+	ret = dr_matcher_disconnect_nic(dmn, nic_tbl, next_nic_matcher, prev_nic_matcher);
+	if (ret)
+		return ret;
+
+	list_del_init(&nic_matcher->list_node);
+	return 0;
+}
+
+int mlx5dr_matcher_destroy(struct mlx5dr_matcher *matcher)
+{
+	struct mlx5dr_table *tbl = matcher->tbl;
+
+	if (WARN_ON_ONCE(refcount_read(&matcher->refcount) > 1))
+		return -EBUSY;
+
+	mlx5dr_domain_lock(tbl->dmn);
+
+	dr_matcher_remove_from_dbg_list(matcher);
+	dr_matcher_uninit(matcher);
+	refcount_dec(&matcher->tbl->refcount);
+
+	mlx5dr_domain_unlock(tbl->dmn);
+	kfree(matcher);
+
+	return 0;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_rule.c b/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_rule.c
new file mode 100644
index 000000000..91ff19f67
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_rule.c
@@ -0,0 +1,1334 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2019 Mellanox Technologies. */
+
+#include "dr_types.h"
+
+#define DR_RULE_MAX_STE_CHAIN (DR_RULE_MAX_STES + DR_ACTION_MAX_STES)
+
+static int dr_rule_append_to_miss_list(struct mlx5dr_ste_ctx *ste_ctx,
+				       struct mlx5dr_ste *new_last_ste,
+				       struct list_head *miss_list,
+				       struct list_head *send_list)
+{
+	struct mlx5dr_ste_send_info *ste_info_last;
+	struct mlx5dr_ste *last_ste;
+
+	/* The new entry will be inserted after the last */
+	last_ste = list_last_entry(miss_list, struct mlx5dr_ste, miss_list_node);
+	WARN_ON(!last_ste);
+
+	ste_info_last = kzalloc(sizeof(*ste_info_last), GFP_KERNEL);
+	if (!ste_info_last)
+		return -ENOMEM;
+
+	mlx5dr_ste_set_miss_addr(ste_ctx, mlx5dr_ste_get_hw_ste(last_ste),
+				 mlx5dr_ste_get_icm_addr(new_last_ste));
+	list_add_tail(&new_last_ste->miss_list_node, miss_list);
+
+	mlx5dr_send_fill_and_append_ste_send_info(last_ste, DR_STE_SIZE_CTRL,
+						  0, mlx5dr_ste_get_hw_ste(last_ste),
+						  ste_info_last, send_list, true);
+
+	return 0;
+}
+
+static struct mlx5dr_ste *
+dr_rule_create_collision_htbl(struct mlx5dr_matcher *matcher,
+			      struct mlx5dr_matcher_rx_tx *nic_matcher,
+			      u8 *hw_ste)
+{
+	struct mlx5dr_domain *dmn = matcher->tbl->dmn;
+	struct mlx5dr_ste_ctx *ste_ctx = dmn->ste_ctx;
+	struct mlx5dr_ste_htbl *new_htbl;
+	struct mlx5dr_ste *ste;
+	u64 icm_addr;
+
+	/* Create new table for miss entry */
+	new_htbl = mlx5dr_ste_htbl_alloc(dmn->ste_icm_pool,
+					 DR_CHUNK_SIZE_1,
+					 MLX5DR_STE_LU_TYPE_DONT_CARE,
+					 0);
+	if (!new_htbl) {
+		mlx5dr_dbg(dmn, "Failed allocating collision table\n");
+		return NULL;
+	}
+
+	/* One and only entry, never grows */
+	ste = new_htbl->chunk->ste_arr;
+	icm_addr = mlx5dr_icm_pool_get_chunk_icm_addr(nic_matcher->e_anchor->chunk);
+	mlx5dr_ste_set_miss_addr(ste_ctx, hw_ste, icm_addr);
+	mlx5dr_htbl_get(new_htbl);
+
+	return ste;
+}
+
+static struct mlx5dr_ste *
+dr_rule_create_collision_entry(struct mlx5dr_matcher *matcher,
+			       struct mlx5dr_matcher_rx_tx *nic_matcher,
+			       u8 *hw_ste,
+			       struct mlx5dr_ste *orig_ste)
+{
+	struct mlx5dr_ste *ste;
+
+	ste = dr_rule_create_collision_htbl(matcher, nic_matcher, hw_ste);
+	if (!ste) {
+		mlx5dr_dbg(matcher->tbl->dmn, "Failed creating collision entry\n");
+		return NULL;
+	}
+
+	ste->ste_chain_location = orig_ste->ste_chain_location;
+	ste->htbl->pointing_ste = orig_ste->htbl->pointing_ste;
+
+	/* In collision entry, all members share the same miss_list_head */
+	ste->htbl->chunk->miss_list = mlx5dr_ste_get_miss_list(orig_ste);
+
+	/* Next table */
+	if (mlx5dr_ste_create_next_htbl(matcher, nic_matcher, ste, hw_ste,
+					DR_CHUNK_SIZE_1)) {
+		mlx5dr_dbg(matcher->tbl->dmn, "Failed allocating table\n");
+		goto free_tbl;
+	}
+
+	return ste;
+
+free_tbl:
+	mlx5dr_ste_free(ste, matcher, nic_matcher);
+	return NULL;
+}
+
+static int
+dr_rule_handle_one_ste_in_update_list(struct mlx5dr_ste_send_info *ste_info,
+				      struct mlx5dr_domain *dmn)
+{
+	int ret;
+
+	list_del(&ste_info->send_list);
+
+	/* Copy data to ste, only reduced size or control, the last 16B (mask)
+	 * is already written to the hw.
+	 */
+	if (ste_info->size == DR_STE_SIZE_CTRL)
+		memcpy(mlx5dr_ste_get_hw_ste(ste_info->ste),
+		       ste_info->data, DR_STE_SIZE_CTRL);
+	else
+		memcpy(mlx5dr_ste_get_hw_ste(ste_info->ste),
+		       ste_info->data, DR_STE_SIZE_REDUCED);
+
+	ret = mlx5dr_send_postsend_ste(dmn, ste_info->ste, ste_info->data,
+				       ste_info->size, ste_info->offset);
+	if (ret)
+		goto out;
+
+out:
+	kfree(ste_info);
+	return ret;
+}
+
+static int dr_rule_send_update_list(struct list_head *send_ste_list,
+				    struct mlx5dr_domain *dmn,
+				    bool is_reverse)
+{
+	struct mlx5dr_ste_send_info *ste_info, *tmp_ste_info;
+	int ret;
+
+	if (is_reverse) {
+		list_for_each_entry_safe_reverse(ste_info, tmp_ste_info,
+						 send_ste_list, send_list) {
+			ret = dr_rule_handle_one_ste_in_update_list(ste_info,
+								    dmn);
+			if (ret)
+				return ret;
+		}
+	} else {
+		list_for_each_entry_safe(ste_info, tmp_ste_info,
+					 send_ste_list, send_list) {
+			ret = dr_rule_handle_one_ste_in_update_list(ste_info,
+								    dmn);
+			if (ret)
+				return ret;
+		}
+	}
+
+	return 0;
+}
+
+static struct mlx5dr_ste *
+dr_rule_find_ste_in_miss_list(struct list_head *miss_list, u8 *hw_ste)
+{
+	struct mlx5dr_ste *ste;
+
+	if (list_empty(miss_list))
+		return NULL;
+
+	/* Check if hw_ste is present in the list */
+	list_for_each_entry(ste, miss_list, miss_list_node) {
+		if (mlx5dr_ste_equal_tag(mlx5dr_ste_get_hw_ste(ste), hw_ste))
+			return ste;
+	}
+
+	return NULL;
+}
+
+static struct mlx5dr_ste *
+dr_rule_rehash_handle_collision(struct mlx5dr_matcher *matcher,
+				struct mlx5dr_matcher_rx_tx *nic_matcher,
+				struct list_head *update_list,
+				struct mlx5dr_ste *col_ste,
+				u8 *hw_ste)
+{
+	struct mlx5dr_domain *dmn = matcher->tbl->dmn;
+	struct mlx5dr_ste *new_ste;
+	int ret;
+
+	new_ste = dr_rule_create_collision_htbl(matcher, nic_matcher, hw_ste);
+	if (!new_ste)
+		return NULL;
+
+	/* Update collision pointing STE */
+	new_ste->htbl->pointing_ste = col_ste->htbl->pointing_ste;
+
+	/* In collision entry, all members share the same miss_list_head */
+	new_ste->htbl->chunk->miss_list = mlx5dr_ste_get_miss_list(col_ste);
+
+	/* Update the previous from the list */
+	ret = dr_rule_append_to_miss_list(dmn->ste_ctx, new_ste,
+					  mlx5dr_ste_get_miss_list(col_ste),
+					  update_list);
+	if (ret) {
+		mlx5dr_dbg(dmn, "Failed update dup entry\n");
+		goto err_exit;
+	}
+
+	return new_ste;
+
+err_exit:
+	mlx5dr_ste_free(new_ste, matcher, nic_matcher);
+	return NULL;
+}
+
+static void dr_rule_rehash_copy_ste_ctrl(struct mlx5dr_matcher *matcher,
+					 struct mlx5dr_matcher_rx_tx *nic_matcher,
+					 struct mlx5dr_ste *cur_ste,
+					 struct mlx5dr_ste *new_ste)
+{
+	new_ste->next_htbl = cur_ste->next_htbl;
+	new_ste->ste_chain_location = cur_ste->ste_chain_location;
+
+	if (new_ste->next_htbl)
+		new_ste->next_htbl->pointing_ste = new_ste;
+
+	/* We need to copy the refcount since this ste
+	 * may have been traversed several times
+	 */
+	new_ste->refcount = cur_ste->refcount;
+
+	/* Link old STEs rule to the new ste */
+	mlx5dr_rule_set_last_member(cur_ste->rule_rx_tx, new_ste, false);
+}
+
+static struct mlx5dr_ste *
+dr_rule_rehash_copy_ste(struct mlx5dr_matcher *matcher,
+			struct mlx5dr_matcher_rx_tx *nic_matcher,
+			struct mlx5dr_ste *cur_ste,
+			struct mlx5dr_ste_htbl *new_htbl,
+			struct list_head *update_list)
+{
+	struct mlx5dr_domain *dmn = matcher->tbl->dmn;
+	struct mlx5dr_ste_send_info *ste_info;
+	bool use_update_list = false;
+	u8 hw_ste[DR_STE_SIZE] = {};
+	struct mlx5dr_ste *new_ste;
+	u64 icm_addr;
+	int new_idx;
+	u8 sb_idx;
+
+	/* Copy STE mask from the matcher */
+	sb_idx = cur_ste->ste_chain_location - 1;
+	mlx5dr_ste_set_bit_mask(hw_ste, nic_matcher->ste_builder[sb_idx].bit_mask);
+
+	/* Copy STE control and tag */
+	icm_addr = mlx5dr_icm_pool_get_chunk_icm_addr(nic_matcher->e_anchor->chunk);
+	memcpy(hw_ste, mlx5dr_ste_get_hw_ste(cur_ste), DR_STE_SIZE_REDUCED);
+	mlx5dr_ste_set_miss_addr(dmn->ste_ctx, hw_ste, icm_addr);
+
+	new_idx = mlx5dr_ste_calc_hash_index(hw_ste, new_htbl);
+	new_ste = &new_htbl->chunk->ste_arr[new_idx];
+
+	if (mlx5dr_ste_is_not_used(new_ste)) {
+		mlx5dr_htbl_get(new_htbl);
+		list_add_tail(&new_ste->miss_list_node,
+			      mlx5dr_ste_get_miss_list(new_ste));
+	} else {
+		new_ste = dr_rule_rehash_handle_collision(matcher,
+							  nic_matcher,
+							  update_list,
+							  new_ste,
+							  hw_ste);
+		if (!new_ste) {
+			mlx5dr_dbg(dmn, "Failed adding collision entry, index: %d\n",
+				   new_idx);
+			return NULL;
+		}
+		new_htbl->ctrl.num_of_collisions++;
+		use_update_list = true;
+	}
+
+	memcpy(mlx5dr_ste_get_hw_ste(new_ste), hw_ste, DR_STE_SIZE_REDUCED);
+
+	new_htbl->ctrl.num_of_valid_entries++;
+
+	if (use_update_list) {
+		ste_info = kzalloc(sizeof(*ste_info), GFP_KERNEL);
+		if (!ste_info)
+			goto err_exit;
+
+		mlx5dr_send_fill_and_append_ste_send_info(new_ste, DR_STE_SIZE, 0,
+							  hw_ste, ste_info,
+							  update_list, true);
+	}
+
+	dr_rule_rehash_copy_ste_ctrl(matcher, nic_matcher, cur_ste, new_ste);
+
+	return new_ste;
+
+err_exit:
+	mlx5dr_ste_free(new_ste, matcher, nic_matcher);
+	return NULL;
+}
+
+static int dr_rule_rehash_copy_miss_list(struct mlx5dr_matcher *matcher,
+					 struct mlx5dr_matcher_rx_tx *nic_matcher,
+					 struct list_head *cur_miss_list,
+					 struct mlx5dr_ste_htbl *new_htbl,
+					 struct list_head *update_list)
+{
+	struct mlx5dr_ste *tmp_ste, *cur_ste, *new_ste;
+
+	if (list_empty(cur_miss_list))
+		return 0;
+
+	list_for_each_entry_safe(cur_ste, tmp_ste, cur_miss_list, miss_list_node) {
+		new_ste = dr_rule_rehash_copy_ste(matcher,
+						  nic_matcher,
+						  cur_ste,
+						  new_htbl,
+						  update_list);
+		if (!new_ste)
+			goto err_insert;
+
+		list_del(&cur_ste->miss_list_node);
+		mlx5dr_htbl_put(cur_ste->htbl);
+	}
+	return 0;
+
+err_insert:
+	mlx5dr_err(matcher->tbl->dmn, "Fatal error during resize\n");
+	WARN_ON(true);
+	return -EINVAL;
+}
+
+static int dr_rule_rehash_copy_htbl(struct mlx5dr_matcher *matcher,
+				    struct mlx5dr_matcher_rx_tx *nic_matcher,
+				    struct mlx5dr_ste_htbl *cur_htbl,
+				    struct mlx5dr_ste_htbl *new_htbl,
+				    struct list_head *update_list)
+{
+	struct mlx5dr_ste *cur_ste;
+	int cur_entries;
+	int err = 0;
+	int i;
+
+	cur_entries = mlx5dr_icm_pool_chunk_size_to_entries(cur_htbl->chunk->size);
+
+	if (cur_entries < 1) {
+		mlx5dr_dbg(matcher->tbl->dmn, "Invalid number of entries\n");
+		return -EINVAL;
+	}
+
+	for (i = 0; i < cur_entries; i++) {
+		cur_ste = &cur_htbl->chunk->ste_arr[i];
+		if (mlx5dr_ste_is_not_used(cur_ste)) /* Empty, nothing to copy */
+			continue;
+
+		err = dr_rule_rehash_copy_miss_list(matcher,
+						    nic_matcher,
+						    mlx5dr_ste_get_miss_list(cur_ste),
+						    new_htbl,
+						    update_list);
+		if (err)
+			goto clean_copy;
+	}
+
+clean_copy:
+	return err;
+}
+
+static struct mlx5dr_ste_htbl *
+dr_rule_rehash_htbl(struct mlx5dr_rule *rule,
+		    struct mlx5dr_rule_rx_tx *nic_rule,
+		    struct mlx5dr_ste_htbl *cur_htbl,
+		    u8 ste_location,
+		    struct list_head *update_list,
+		    enum mlx5dr_icm_chunk_size new_size)
+{
+	struct mlx5dr_ste_send_info *del_ste_info, *tmp_ste_info;
+	struct mlx5dr_matcher *matcher = rule->matcher;
+	struct mlx5dr_domain *dmn = matcher->tbl->dmn;
+	struct mlx5dr_matcher_rx_tx *nic_matcher;
+	struct mlx5dr_ste_send_info *ste_info;
+	struct mlx5dr_htbl_connect_info info;
+	struct mlx5dr_domain_rx_tx *nic_dmn;
+	u8 formatted_ste[DR_STE_SIZE] = {};
+	LIST_HEAD(rehash_table_send_list);
+	struct mlx5dr_ste *ste_to_update;
+	struct mlx5dr_ste_htbl *new_htbl;
+	int err;
+
+	nic_matcher = nic_rule->nic_matcher;
+	nic_dmn = nic_matcher->nic_tbl->nic_dmn;
+
+	ste_info = kzalloc(sizeof(*ste_info), GFP_KERNEL);
+	if (!ste_info)
+		return NULL;
+
+	new_htbl = mlx5dr_ste_htbl_alloc(dmn->ste_icm_pool,
+					 new_size,
+					 cur_htbl->lu_type,
+					 cur_htbl->byte_mask);
+	if (!new_htbl) {
+		mlx5dr_err(dmn, "Failed to allocate new hash table\n");
+		goto free_ste_info;
+	}
+
+	/* Write new table to HW */
+	info.type = CONNECT_MISS;
+	info.miss_icm_addr = mlx5dr_icm_pool_get_chunk_icm_addr(nic_matcher->e_anchor->chunk);
+	mlx5dr_ste_set_formatted_ste(dmn->ste_ctx,
+				     dmn->info.caps.gvmi,
+				     nic_dmn->type,
+				     new_htbl,
+				     formatted_ste,
+				     &info);
+
+	new_htbl->pointing_ste = cur_htbl->pointing_ste;
+	new_htbl->pointing_ste->next_htbl = new_htbl;
+	err = dr_rule_rehash_copy_htbl(matcher,
+				       nic_matcher,
+				       cur_htbl,
+				       new_htbl,
+				       &rehash_table_send_list);
+	if (err)
+		goto free_new_htbl;
+
+	if (mlx5dr_send_postsend_htbl(dmn, new_htbl, formatted_ste,
+				      nic_matcher->ste_builder[ste_location - 1].bit_mask)) {
+		mlx5dr_err(dmn, "Failed writing table to HW\n");
+		goto free_new_htbl;
+	}
+
+	/* Writing to the hw is done in regular order of rehash_table_send_list,
+	 * in order to have the origin data written before the miss address of
+	 * collision entries, if exists.
+	 */
+	if (dr_rule_send_update_list(&rehash_table_send_list, dmn, false)) {
+		mlx5dr_err(dmn, "Failed updating table to HW\n");
+		goto free_ste_list;
+	}
+
+	/* Connect previous hash table to current */
+	if (ste_location == 1) {
+		/* The previous table is an anchor, anchors size is always one STE */
+		struct mlx5dr_ste_htbl *prev_htbl = cur_htbl->pointing_ste->htbl;
+
+		/* On matcher s_anchor we keep an extra refcount */
+		mlx5dr_htbl_get(new_htbl);
+		mlx5dr_htbl_put(cur_htbl);
+
+		nic_matcher->s_htbl = new_htbl;
+
+		/* It is safe to operate dr_ste_set_hit_addr on the hw_ste here
+		 * (48B len) which works only on first 32B
+		 */
+		mlx5dr_ste_set_hit_addr(dmn->ste_ctx,
+					prev_htbl->chunk->hw_ste_arr,
+					mlx5dr_icm_pool_get_chunk_icm_addr(new_htbl->chunk),
+					mlx5dr_icm_pool_get_chunk_num_of_entries(new_htbl->chunk));
+
+		ste_to_update = &prev_htbl->chunk->ste_arr[0];
+	} else {
+		mlx5dr_ste_set_hit_addr_by_next_htbl(dmn->ste_ctx,
+						     mlx5dr_ste_get_hw_ste(cur_htbl->pointing_ste),
+						     new_htbl);
+		ste_to_update = cur_htbl->pointing_ste;
+	}
+
+	mlx5dr_send_fill_and_append_ste_send_info(ste_to_update, DR_STE_SIZE_CTRL,
+						  0, mlx5dr_ste_get_hw_ste(ste_to_update),
+						  ste_info, update_list, false);
+
+	return new_htbl;
+
+free_ste_list:
+	/* Clean all ste_info's from the new table */
+	list_for_each_entry_safe(del_ste_info, tmp_ste_info,
+				 &rehash_table_send_list, send_list) {
+		list_del(&del_ste_info->send_list);
+		kfree(del_ste_info);
+	}
+
+free_new_htbl:
+	mlx5dr_ste_htbl_free(new_htbl);
+free_ste_info:
+	kfree(ste_info);
+	mlx5dr_info(dmn, "Failed creating rehash table\n");
+	return NULL;
+}
+
+static struct mlx5dr_ste_htbl *dr_rule_rehash(struct mlx5dr_rule *rule,
+					      struct mlx5dr_rule_rx_tx *nic_rule,
+					      struct mlx5dr_ste_htbl *cur_htbl,
+					      u8 ste_location,
+					      struct list_head *update_list)
+{
+	struct mlx5dr_domain *dmn = rule->matcher->tbl->dmn;
+	enum mlx5dr_icm_chunk_size new_size;
+
+	new_size = mlx5dr_icm_next_higher_chunk(cur_htbl->chunk->size);
+	new_size = min_t(u32, new_size, dmn->info.max_log_sw_icm_sz);
+
+	if (new_size == cur_htbl->chunk->size)
+		return NULL; /* Skip rehash, we already at the max size */
+
+	return dr_rule_rehash_htbl(rule, nic_rule, cur_htbl, ste_location,
+				   update_list, new_size);
+}
+
+static struct mlx5dr_ste *
+dr_rule_handle_collision(struct mlx5dr_matcher *matcher,
+			 struct mlx5dr_matcher_rx_tx *nic_matcher,
+			 struct mlx5dr_ste *ste,
+			 u8 *hw_ste,
+			 struct list_head *miss_list,
+			 struct list_head *send_list)
+{
+	struct mlx5dr_domain *dmn = matcher->tbl->dmn;
+	struct mlx5dr_ste_ctx *ste_ctx = dmn->ste_ctx;
+	struct mlx5dr_ste_send_info *ste_info;
+	struct mlx5dr_ste *new_ste;
+
+	ste_info = kzalloc(sizeof(*ste_info), GFP_KERNEL);
+	if (!ste_info)
+		return NULL;
+
+	new_ste = dr_rule_create_collision_entry(matcher, nic_matcher, hw_ste, ste);
+	if (!new_ste)
+		goto free_send_info;
+
+	if (dr_rule_append_to_miss_list(ste_ctx, new_ste,
+					miss_list, send_list)) {
+		mlx5dr_dbg(dmn, "Failed to update prev miss_list\n");
+		goto err_exit;
+	}
+
+	mlx5dr_send_fill_and_append_ste_send_info(new_ste, DR_STE_SIZE, 0, hw_ste,
+						  ste_info, send_list, false);
+
+	ste->htbl->ctrl.num_of_collisions++;
+	ste->htbl->ctrl.num_of_valid_entries++;
+
+	return new_ste;
+
+err_exit:
+	mlx5dr_ste_free(new_ste, matcher, nic_matcher);
+free_send_info:
+	kfree(ste_info);
+	return NULL;
+}
+
+static void dr_rule_remove_action_members(struct mlx5dr_rule *rule)
+{
+	struct mlx5dr_rule_action_member *action_mem;
+	struct mlx5dr_rule_action_member *tmp;
+
+	list_for_each_entry_safe(action_mem, tmp, &rule->rule_actions_list, list) {
+		list_del(&action_mem->list);
+		refcount_dec(&action_mem->action->refcount);
+		kvfree(action_mem);
+	}
+}
+
+static int dr_rule_add_action_members(struct mlx5dr_rule *rule,
+				      size_t num_actions,
+				      struct mlx5dr_action *actions[])
+{
+	struct mlx5dr_rule_action_member *action_mem;
+	int i;
+
+	for (i = 0; i < num_actions; i++) {
+		action_mem = kvzalloc(sizeof(*action_mem), GFP_KERNEL);
+		if (!action_mem)
+			goto free_action_members;
+
+		action_mem->action = actions[i];
+		INIT_LIST_HEAD(&action_mem->list);
+		list_add_tail(&action_mem->list, &rule->rule_actions_list);
+		refcount_inc(&action_mem->action->refcount);
+	}
+
+	return 0;
+
+free_action_members:
+	dr_rule_remove_action_members(rule);
+	return -ENOMEM;
+}
+
+void mlx5dr_rule_set_last_member(struct mlx5dr_rule_rx_tx *nic_rule,
+				 struct mlx5dr_ste *ste,
+				 bool force)
+{
+	/* Update rule member is usually done for the last STE or during rule
+	 * creation to recover from mid-creation failure (for this peruse the
+	 * force flag is used)
+	 */
+	if (ste->next_htbl && !force)
+		return;
+
+	/* Update is required since each rule keeps track of its last STE */
+	ste->rule_rx_tx = nic_rule;
+	nic_rule->last_rule_ste = ste;
+}
+
+static struct mlx5dr_ste *dr_rule_get_pointed_ste(struct mlx5dr_ste *curr_ste)
+{
+	struct mlx5dr_ste *first_ste;
+
+	first_ste = list_first_entry(mlx5dr_ste_get_miss_list(curr_ste),
+				     struct mlx5dr_ste, miss_list_node);
+
+	return first_ste->htbl->pointing_ste;
+}
+
+int mlx5dr_rule_get_reverse_rule_members(struct mlx5dr_ste **ste_arr,
+					 struct mlx5dr_ste *curr_ste,
+					 int *num_of_stes)
+{
+	bool first = false;
+
+	*num_of_stes = 0;
+
+	if (!curr_ste)
+		return -ENOENT;
+
+	/* Iterate from last to first */
+	while (!first) {
+		first = curr_ste->ste_chain_location == 1;
+		ste_arr[*num_of_stes] = curr_ste;
+		*num_of_stes += 1;
+		curr_ste = dr_rule_get_pointed_ste(curr_ste);
+	}
+
+	return 0;
+}
+
+static void dr_rule_clean_rule_members(struct mlx5dr_rule *rule,
+				       struct mlx5dr_rule_rx_tx *nic_rule)
+{
+	struct mlx5dr_ste *ste_arr[DR_RULE_MAX_STES + DR_ACTION_MAX_STES];
+	struct mlx5dr_ste *curr_ste = nic_rule->last_rule_ste;
+	int i;
+
+	if (mlx5dr_rule_get_reverse_rule_members(ste_arr, curr_ste, &i))
+		return;
+
+	while (i--)
+		mlx5dr_ste_put(ste_arr[i], rule->matcher, nic_rule->nic_matcher);
+}
+
+static u16 dr_get_bits_per_mask(u16 byte_mask)
+{
+	u16 bits = 0;
+
+	while (byte_mask) {
+		byte_mask = byte_mask & (byte_mask - 1);
+		bits++;
+	}
+
+	return bits;
+}
+
+static bool dr_rule_need_enlarge_hash(struct mlx5dr_ste_htbl *htbl,
+				      struct mlx5dr_domain *dmn,
+				      struct mlx5dr_domain_rx_tx *nic_dmn)
+{
+	struct mlx5dr_ste_htbl_ctrl *ctrl = &htbl->ctrl;
+	int threshold;
+
+	if (dmn->info.max_log_sw_icm_sz <= htbl->chunk->size)
+		return false;
+
+	if (!mlx5dr_ste_htbl_may_grow(htbl))
+		return false;
+
+	if (dr_get_bits_per_mask(htbl->byte_mask) * BITS_PER_BYTE <= htbl->chunk->size)
+		return false;
+
+	threshold = mlx5dr_ste_htbl_increase_threshold(htbl);
+	if (ctrl->num_of_collisions >= threshold &&
+	    (ctrl->num_of_valid_entries - ctrl->num_of_collisions) >= threshold)
+		return true;
+
+	return false;
+}
+
+static int dr_rule_handle_action_stes(struct mlx5dr_rule *rule,
+				      struct mlx5dr_rule_rx_tx *nic_rule,
+				      struct list_head *send_ste_list,
+				      struct mlx5dr_ste *last_ste,
+				      u8 *hw_ste_arr,
+				      u32 new_hw_ste_arr_sz)
+{
+	struct mlx5dr_matcher_rx_tx *nic_matcher = nic_rule->nic_matcher;
+	struct mlx5dr_ste_send_info *ste_info_arr[DR_ACTION_MAX_STES];
+	u8 num_of_builders = nic_matcher->num_of_builders;
+	struct mlx5dr_matcher *matcher = rule->matcher;
+	struct mlx5dr_domain *dmn = matcher->tbl->dmn;
+	u8 *curr_hw_ste, *prev_hw_ste;
+	struct mlx5dr_ste *action_ste;
+	int i, k;
+
+	/* Two cases:
+	 * 1. num_of_builders is equal to new_hw_ste_arr_sz, the action in the ste
+	 * 2. num_of_builders is less then new_hw_ste_arr_sz, new ste was added
+	 *    to support the action.
+	 */
+
+	for (i = num_of_builders, k = 0; i < new_hw_ste_arr_sz; i++, k++) {
+		curr_hw_ste = hw_ste_arr + i * DR_STE_SIZE;
+		prev_hw_ste = (i == 0) ? curr_hw_ste : hw_ste_arr + ((i - 1) * DR_STE_SIZE);
+		action_ste = dr_rule_create_collision_htbl(matcher,
+							   nic_matcher,
+							   curr_hw_ste);
+		if (!action_ste)
+			return -ENOMEM;
+
+		mlx5dr_ste_get(action_ste);
+
+		action_ste->htbl->pointing_ste = last_ste;
+		last_ste->next_htbl = action_ste->htbl;
+		last_ste = action_ste;
+
+		/* While free ste we go over the miss list, so add this ste to the list */
+		list_add_tail(&action_ste->miss_list_node,
+			      mlx5dr_ste_get_miss_list(action_ste));
+
+		ste_info_arr[k] = kzalloc(sizeof(*ste_info_arr[k]),
+					  GFP_KERNEL);
+		if (!ste_info_arr[k])
+			goto err_exit;
+
+		/* Point current ste to the new action */
+		mlx5dr_ste_set_hit_addr_by_next_htbl(dmn->ste_ctx,
+						     prev_hw_ste,
+						     action_ste->htbl);
+
+		mlx5dr_rule_set_last_member(nic_rule, action_ste, true);
+
+		mlx5dr_send_fill_and_append_ste_send_info(action_ste, DR_STE_SIZE, 0,
+							  curr_hw_ste,
+							  ste_info_arr[k],
+							  send_ste_list, false);
+	}
+
+	last_ste->next_htbl = NULL;
+
+	return 0;
+
+err_exit:
+	mlx5dr_ste_put(action_ste, matcher, nic_matcher);
+	return -ENOMEM;
+}
+
+static int dr_rule_handle_empty_entry(struct mlx5dr_matcher *matcher,
+				      struct mlx5dr_matcher_rx_tx *nic_matcher,
+				      struct mlx5dr_ste_htbl *cur_htbl,
+				      struct mlx5dr_ste *ste,
+				      u8 ste_location,
+				      u8 *hw_ste,
+				      struct list_head *miss_list,
+				      struct list_head *send_list)
+{
+	struct mlx5dr_domain *dmn = matcher->tbl->dmn;
+	struct mlx5dr_ste_send_info *ste_info;
+	u64 icm_addr;
+
+	/* Take ref on table, only on first time this ste is used */
+	mlx5dr_htbl_get(cur_htbl);
+
+	/* new entry -> new branch */
+	list_add_tail(&ste->miss_list_node, miss_list);
+
+	icm_addr = mlx5dr_icm_pool_get_chunk_icm_addr(nic_matcher->e_anchor->chunk);
+	mlx5dr_ste_set_miss_addr(dmn->ste_ctx, hw_ste, icm_addr);
+
+	ste->ste_chain_location = ste_location;
+
+	ste_info = kzalloc(sizeof(*ste_info), GFP_KERNEL);
+	if (!ste_info)
+		goto clean_ste_setting;
+
+	if (mlx5dr_ste_create_next_htbl(matcher,
+					nic_matcher,
+					ste,
+					hw_ste,
+					DR_CHUNK_SIZE_1)) {
+		mlx5dr_dbg(dmn, "Failed allocating table\n");
+		goto clean_ste_info;
+	}
+
+	cur_htbl->ctrl.num_of_valid_entries++;
+
+	mlx5dr_send_fill_and_append_ste_send_info(ste, DR_STE_SIZE, 0, hw_ste,
+						  ste_info, send_list, false);
+
+	return 0;
+
+clean_ste_info:
+	kfree(ste_info);
+clean_ste_setting:
+	list_del_init(&ste->miss_list_node);
+	mlx5dr_htbl_put(cur_htbl);
+
+	return -ENOMEM;
+}
+
+static struct mlx5dr_ste *
+dr_rule_handle_ste_branch(struct mlx5dr_rule *rule,
+			  struct mlx5dr_rule_rx_tx *nic_rule,
+			  struct list_head *send_ste_list,
+			  struct mlx5dr_ste_htbl *cur_htbl,
+			  u8 *hw_ste,
+			  u8 ste_location,
+			  struct mlx5dr_ste_htbl **put_htbl)
+{
+	struct mlx5dr_matcher *matcher = rule->matcher;
+	struct mlx5dr_domain *dmn = matcher->tbl->dmn;
+	struct mlx5dr_matcher_rx_tx *nic_matcher;
+	struct mlx5dr_domain_rx_tx *nic_dmn;
+	struct mlx5dr_ste_htbl *new_htbl;
+	struct mlx5dr_ste *matched_ste;
+	struct list_head *miss_list;
+	bool skip_rehash = false;
+	struct mlx5dr_ste *ste;
+	int index;
+
+	nic_matcher = nic_rule->nic_matcher;
+	nic_dmn = nic_matcher->nic_tbl->nic_dmn;
+
+again:
+	index = mlx5dr_ste_calc_hash_index(hw_ste, cur_htbl);
+	miss_list = &cur_htbl->chunk->miss_list[index];
+	ste = &cur_htbl->chunk->ste_arr[index];
+
+	if (mlx5dr_ste_is_not_used(ste)) {
+		if (dr_rule_handle_empty_entry(matcher, nic_matcher, cur_htbl,
+					       ste, ste_location,
+					       hw_ste, miss_list,
+					       send_ste_list))
+			return NULL;
+	} else {
+		/* Hash table index in use, check if this ste is in the miss list */
+		matched_ste = dr_rule_find_ste_in_miss_list(miss_list, hw_ste);
+		if (matched_ste) {
+			/* If it is last STE in the chain, and has the same tag
+			 * it means that all the previous stes are the same,
+			 * if so, this rule is duplicated.
+			 */
+			if (!mlx5dr_ste_is_last_in_rule(nic_matcher, ste_location))
+				return matched_ste;
+
+			mlx5dr_dbg(dmn, "Duplicate rule inserted\n");
+		}
+
+		if (!skip_rehash && dr_rule_need_enlarge_hash(cur_htbl, dmn, nic_dmn)) {
+			/* Hash table index in use, try to resize of the hash */
+			skip_rehash = true;
+
+			/* Hold the table till we update.
+			 * Release in dr_rule_create_rule()
+			 */
+			*put_htbl = cur_htbl;
+			mlx5dr_htbl_get(cur_htbl);
+
+			new_htbl = dr_rule_rehash(rule, nic_rule, cur_htbl,
+						  ste_location, send_ste_list);
+			if (!new_htbl) {
+				mlx5dr_err(dmn, "Failed creating rehash table, htbl-log_size: %d\n",
+					   cur_htbl->chunk->size);
+				mlx5dr_htbl_put(cur_htbl);
+			} else {
+				cur_htbl = new_htbl;
+			}
+			goto again;
+		} else {
+			/* Hash table index in use, add another collision (miss) */
+			ste = dr_rule_handle_collision(matcher,
+						       nic_matcher,
+						       ste,
+						       hw_ste,
+						       miss_list,
+						       send_ste_list);
+			if (!ste) {
+				mlx5dr_dbg(dmn, "failed adding collision entry, index: %d\n",
+					   index);
+				return NULL;
+			}
+		}
+	}
+	return ste;
+}
+
+static bool dr_rule_cmp_value_to_mask(u8 *mask, u8 *value,
+				      u32 s_idx, u32 e_idx)
+{
+	u32 i;
+
+	for (i = s_idx; i < e_idx; i++) {
+		if (value[i] & ~mask[i]) {
+			pr_info("Rule parameters contains a value not specified by mask\n");
+			return false;
+		}
+	}
+	return true;
+}
+
+static bool dr_rule_verify(struct mlx5dr_matcher *matcher,
+			   struct mlx5dr_match_parameters *value,
+			   struct mlx5dr_match_param *param)
+{
+	u8 match_criteria = matcher->match_criteria;
+	size_t value_size = value->match_sz;
+	u8 *mask_p = (u8 *)&matcher->mask;
+	u8 *param_p = (u8 *)param;
+	u32 s_idx, e_idx;
+
+	if (!value_size ||
+	    (value_size > DR_SZ_MATCH_PARAM || (value_size % sizeof(u32)))) {
+		mlx5dr_err(matcher->tbl->dmn, "Rule parameters length is incorrect\n");
+		return false;
+	}
+
+	mlx5dr_ste_copy_param(matcher->match_criteria, param, value, false);
+
+	if (match_criteria & DR_MATCHER_CRITERIA_OUTER) {
+		s_idx = offsetof(struct mlx5dr_match_param, outer);
+		e_idx = min(s_idx + sizeof(param->outer), value_size);
+
+		if (!dr_rule_cmp_value_to_mask(mask_p, param_p, s_idx, e_idx)) {
+			mlx5dr_err(matcher->tbl->dmn, "Rule outer parameters contains a value not specified by mask\n");
+			return false;
+		}
+	}
+
+	if (match_criteria & DR_MATCHER_CRITERIA_MISC) {
+		s_idx = offsetof(struct mlx5dr_match_param, misc);
+		e_idx = min(s_idx + sizeof(param->misc), value_size);
+
+		if (!dr_rule_cmp_value_to_mask(mask_p, param_p, s_idx, e_idx)) {
+			mlx5dr_err(matcher->tbl->dmn, "Rule misc parameters contains a value not specified by mask\n");
+			return false;
+		}
+	}
+
+	if (match_criteria & DR_MATCHER_CRITERIA_INNER) {
+		s_idx = offsetof(struct mlx5dr_match_param, inner);
+		e_idx = min(s_idx + sizeof(param->inner), value_size);
+
+		if (!dr_rule_cmp_value_to_mask(mask_p, param_p, s_idx, e_idx)) {
+			mlx5dr_err(matcher->tbl->dmn, "Rule inner parameters contains a value not specified by mask\n");
+			return false;
+		}
+	}
+
+	if (match_criteria & DR_MATCHER_CRITERIA_MISC2) {
+		s_idx = offsetof(struct mlx5dr_match_param, misc2);
+		e_idx = min(s_idx + sizeof(param->misc2), value_size);
+
+		if (!dr_rule_cmp_value_to_mask(mask_p, param_p, s_idx, e_idx)) {
+			mlx5dr_err(matcher->tbl->dmn, "Rule misc2 parameters contains a value not specified by mask\n");
+			return false;
+		}
+	}
+
+	if (match_criteria & DR_MATCHER_CRITERIA_MISC3) {
+		s_idx = offsetof(struct mlx5dr_match_param, misc3);
+		e_idx = min(s_idx + sizeof(param->misc3), value_size);
+
+		if (!dr_rule_cmp_value_to_mask(mask_p, param_p, s_idx, e_idx)) {
+			mlx5dr_err(matcher->tbl->dmn, "Rule misc3 parameters contains a value not specified by mask\n");
+			return false;
+		}
+	}
+
+	if (match_criteria & DR_MATCHER_CRITERIA_MISC4) {
+		s_idx = offsetof(struct mlx5dr_match_param, misc4);
+		e_idx = min(s_idx + sizeof(param->misc4), value_size);
+
+		if (!dr_rule_cmp_value_to_mask(mask_p, param_p, s_idx, e_idx)) {
+			mlx5dr_err(matcher->tbl->dmn,
+				   "Rule misc4 parameters contains a value not specified by mask\n");
+			return false;
+		}
+	}
+
+	if (match_criteria & DR_MATCHER_CRITERIA_MISC5) {
+		s_idx = offsetof(struct mlx5dr_match_param, misc5);
+		e_idx = min(s_idx + sizeof(param->misc5), value_size);
+
+		if (!dr_rule_cmp_value_to_mask(mask_p, param_p, s_idx, e_idx)) {
+			mlx5dr_err(matcher->tbl->dmn, "Rule misc5 parameters contains a value not specified by mask\n");
+			return false;
+		}
+	}
+	return true;
+}
+
+static int dr_rule_destroy_rule_nic(struct mlx5dr_rule *rule,
+				    struct mlx5dr_rule_rx_tx *nic_rule)
+{
+	/* Check if this nic rule was actually created, or was it skipped
+	 * and only the other type of the RX/TX nic rule was created.
+	 */
+	if (!nic_rule->last_rule_ste)
+		return 0;
+
+	mlx5dr_domain_nic_lock(nic_rule->nic_matcher->nic_tbl->nic_dmn);
+	dr_rule_clean_rule_members(rule, nic_rule);
+
+	nic_rule->nic_matcher->rules--;
+	if (!nic_rule->nic_matcher->rules)
+		mlx5dr_matcher_remove_from_tbl_nic(rule->matcher->tbl->dmn,
+						   nic_rule->nic_matcher);
+
+	mlx5dr_domain_nic_unlock(nic_rule->nic_matcher->nic_tbl->nic_dmn);
+
+	return 0;
+}
+
+static int dr_rule_destroy_rule_fdb(struct mlx5dr_rule *rule)
+{
+	dr_rule_destroy_rule_nic(rule, &rule->rx);
+	dr_rule_destroy_rule_nic(rule, &rule->tx);
+	return 0;
+}
+
+static int dr_rule_destroy_rule(struct mlx5dr_rule *rule)
+{
+	struct mlx5dr_domain *dmn = rule->matcher->tbl->dmn;
+
+	mlx5dr_dbg_rule_del(rule);
+
+	switch (dmn->type) {
+	case MLX5DR_DOMAIN_TYPE_NIC_RX:
+		dr_rule_destroy_rule_nic(rule, &rule->rx);
+		break;
+	case MLX5DR_DOMAIN_TYPE_NIC_TX:
+		dr_rule_destroy_rule_nic(rule, &rule->tx);
+		break;
+	case MLX5DR_DOMAIN_TYPE_FDB:
+		dr_rule_destroy_rule_fdb(rule);
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	dr_rule_remove_action_members(rule);
+	kfree(rule);
+	return 0;
+}
+
+static enum mlx5dr_ipv dr_rule_get_ipv(struct mlx5dr_match_spec *spec)
+{
+	if (spec->ip_version == 6 || spec->ethertype == ETH_P_IPV6)
+		return DR_RULE_IPV6;
+
+	return DR_RULE_IPV4;
+}
+
+static bool dr_rule_skip(enum mlx5dr_domain_type domain,
+			 enum mlx5dr_domain_nic_type nic_type,
+			 struct mlx5dr_match_param *mask,
+			 struct mlx5dr_match_param *value,
+			 u32 flow_source)
+{
+	bool rx = nic_type == DR_DOMAIN_NIC_TYPE_RX;
+
+	if (domain != MLX5DR_DOMAIN_TYPE_FDB)
+		return false;
+
+	if (mask->misc.source_port) {
+		if (rx && value->misc.source_port != MLX5_VPORT_UPLINK)
+			return true;
+
+		if (!rx && value->misc.source_port == MLX5_VPORT_UPLINK)
+			return true;
+	}
+
+	if (rx && flow_source == MLX5_FLOW_CONTEXT_FLOW_SOURCE_LOCAL_VPORT)
+		return true;
+
+	if (!rx && flow_source == MLX5_FLOW_CONTEXT_FLOW_SOURCE_UPLINK)
+		return true;
+
+	return false;
+}
+
+static int
+dr_rule_create_rule_nic(struct mlx5dr_rule *rule,
+			struct mlx5dr_rule_rx_tx *nic_rule,
+			struct mlx5dr_match_param *param,
+			size_t num_actions,
+			struct mlx5dr_action *actions[])
+{
+	struct mlx5dr_ste_send_info *ste_info, *tmp_ste_info;
+	struct mlx5dr_matcher *matcher = rule->matcher;
+	struct mlx5dr_domain *dmn = matcher->tbl->dmn;
+	struct mlx5dr_matcher_rx_tx *nic_matcher;
+	struct mlx5dr_domain_rx_tx *nic_dmn;
+	struct mlx5dr_ste_htbl *htbl = NULL;
+	struct mlx5dr_ste_htbl *cur_htbl;
+	struct mlx5dr_ste *ste = NULL;
+	LIST_HEAD(send_ste_list);
+	u8 *hw_ste_arr = NULL;
+	u32 new_hw_ste_arr_sz;
+	int ret, i;
+
+	nic_matcher = nic_rule->nic_matcher;
+	nic_dmn = nic_matcher->nic_tbl->nic_dmn;
+
+	if (dr_rule_skip(dmn->type, nic_dmn->type, &matcher->mask, param,
+			 rule->flow_source))
+		return 0;
+
+	hw_ste_arr = kzalloc(DR_RULE_MAX_STE_CHAIN * DR_STE_SIZE, GFP_KERNEL);
+	if (!hw_ste_arr)
+		return -ENOMEM;
+
+	mlx5dr_domain_nic_lock(nic_dmn);
+
+	ret = mlx5dr_matcher_add_to_tbl_nic(dmn, nic_matcher);
+	if (ret)
+		goto free_hw_ste;
+
+	ret = mlx5dr_matcher_select_builders(matcher,
+					     nic_matcher,
+					     dr_rule_get_ipv(&param->outer),
+					     dr_rule_get_ipv(&param->inner));
+	if (ret)
+		goto remove_from_nic_tbl;
+
+	/* Set the tag values inside the ste array */
+	ret = mlx5dr_ste_build_ste_arr(matcher, nic_matcher, param, hw_ste_arr);
+	if (ret)
+		goto remove_from_nic_tbl;
+
+	/* Set the actions values/addresses inside the ste array */
+	ret = mlx5dr_actions_build_ste_arr(matcher, nic_matcher, actions,
+					   num_actions, hw_ste_arr,
+					   &new_hw_ste_arr_sz);
+	if (ret)
+		goto remove_from_nic_tbl;
+
+	cur_htbl = nic_matcher->s_htbl;
+
+	/* Go over the array of STEs, and build dr_ste accordingly.
+	 * The loop is over only the builders which are equal or less to the
+	 * number of stes, in case we have actions that lives in other stes.
+	 */
+	for (i = 0; i < nic_matcher->num_of_builders; i++) {
+		/* Calculate CRC and keep new ste entry */
+		u8 *cur_hw_ste_ent = hw_ste_arr + (i * DR_STE_SIZE);
+
+		ste = dr_rule_handle_ste_branch(rule,
+						nic_rule,
+						&send_ste_list,
+						cur_htbl,
+						cur_hw_ste_ent,
+						i + 1,
+						&htbl);
+		if (!ste) {
+			mlx5dr_err(dmn, "Failed creating next branch\n");
+			ret = -ENOENT;
+			goto free_rule;
+		}
+
+		cur_htbl = ste->next_htbl;
+
+		mlx5dr_ste_get(ste);
+		mlx5dr_rule_set_last_member(nic_rule, ste, true);
+	}
+
+	/* Connect actions */
+	ret = dr_rule_handle_action_stes(rule, nic_rule, &send_ste_list,
+					 ste, hw_ste_arr, new_hw_ste_arr_sz);
+	if (ret) {
+		mlx5dr_dbg(dmn, "Failed apply actions\n");
+		goto free_rule;
+	}
+	ret = dr_rule_send_update_list(&send_ste_list, dmn, true);
+	if (ret) {
+		mlx5dr_err(dmn, "Failed sending ste!\n");
+		goto free_rule;
+	}
+
+	if (htbl)
+		mlx5dr_htbl_put(htbl);
+
+	nic_matcher->rules++;
+
+	mlx5dr_domain_nic_unlock(nic_dmn);
+
+	kfree(hw_ste_arr);
+
+	return 0;
+
+free_rule:
+	dr_rule_clean_rule_members(rule, nic_rule);
+	/* Clean all ste_info's */
+	list_for_each_entry_safe(ste_info, tmp_ste_info, &send_ste_list, send_list) {
+		list_del(&ste_info->send_list);
+		kfree(ste_info);
+	}
+
+remove_from_nic_tbl:
+	if (!nic_matcher->rules)
+		mlx5dr_matcher_remove_from_tbl_nic(dmn, nic_matcher);
+
+free_hw_ste:
+	mlx5dr_domain_nic_unlock(nic_dmn);
+	kfree(hw_ste_arr);
+	return ret;
+}
+
+static int
+dr_rule_create_rule_fdb(struct mlx5dr_rule *rule,
+			struct mlx5dr_match_param *param,
+			size_t num_actions,
+			struct mlx5dr_action *actions[])
+{
+	struct mlx5dr_match_param copy_param = {};
+	int ret;
+
+	/* Copy match_param since they will be consumed during the first
+	 * nic_rule insertion.
+	 */
+	memcpy(&copy_param, param, sizeof(struct mlx5dr_match_param));
+
+	ret = dr_rule_create_rule_nic(rule, &rule->rx, param,
+				      num_actions, actions);
+	if (ret)
+		return ret;
+
+	ret = dr_rule_create_rule_nic(rule, &rule->tx, &copy_param,
+				      num_actions, actions);
+	if (ret)
+		goto destroy_rule_nic_rx;
+
+	return 0;
+
+destroy_rule_nic_rx:
+	dr_rule_destroy_rule_nic(rule, &rule->rx);
+	return ret;
+}
+
+static struct mlx5dr_rule *
+dr_rule_create_rule(struct mlx5dr_matcher *matcher,
+		    struct mlx5dr_match_parameters *value,
+		    size_t num_actions,
+		    struct mlx5dr_action *actions[],
+		    u32 flow_source)
+{
+	struct mlx5dr_domain *dmn = matcher->tbl->dmn;
+	struct mlx5dr_match_param param = {};
+	struct mlx5dr_rule *rule;
+	int ret;
+
+	if (!dr_rule_verify(matcher, value, &param))
+		return NULL;
+
+	rule = kzalloc(sizeof(*rule), GFP_KERNEL);
+	if (!rule)
+		return NULL;
+
+	rule->matcher = matcher;
+	rule->flow_source = flow_source;
+	INIT_LIST_HEAD(&rule->rule_actions_list);
+
+	ret = dr_rule_add_action_members(rule, num_actions, actions);
+	if (ret)
+		goto free_rule;
+
+	switch (dmn->type) {
+	case MLX5DR_DOMAIN_TYPE_NIC_RX:
+		rule->rx.nic_matcher = &matcher->rx;
+		ret = dr_rule_create_rule_nic(rule, &rule->rx, &param,
+					      num_actions, actions);
+		break;
+	case MLX5DR_DOMAIN_TYPE_NIC_TX:
+		rule->tx.nic_matcher = &matcher->tx;
+		ret = dr_rule_create_rule_nic(rule, &rule->tx, &param,
+					      num_actions, actions);
+		break;
+	case MLX5DR_DOMAIN_TYPE_FDB:
+		rule->rx.nic_matcher = &matcher->rx;
+		rule->tx.nic_matcher = &matcher->tx;
+		ret = dr_rule_create_rule_fdb(rule, &param,
+					      num_actions, actions);
+		break;
+	default:
+		ret = -EINVAL;
+		break;
+	}
+
+	if (ret)
+		goto remove_action_members;
+
+	INIT_LIST_HEAD(&rule->dbg_node);
+	mlx5dr_dbg_rule_add(rule);
+	return rule;
+
+remove_action_members:
+	dr_rule_remove_action_members(rule);
+free_rule:
+	kfree(rule);
+	mlx5dr_err(dmn, "Failed creating rule\n");
+	return NULL;
+}
+
+struct mlx5dr_rule *mlx5dr_rule_create(struct mlx5dr_matcher *matcher,
+				       struct mlx5dr_match_parameters *value,
+				       size_t num_actions,
+				       struct mlx5dr_action *actions[],
+				       u32 flow_source)
+{
+	struct mlx5dr_rule *rule;
+
+	refcount_inc(&matcher->refcount);
+
+	rule = dr_rule_create_rule(matcher, value, num_actions, actions, flow_source);
+	if (!rule)
+		refcount_dec(&matcher->refcount);
+
+	return rule;
+}
+
+int mlx5dr_rule_destroy(struct mlx5dr_rule *rule)
+{
+	struct mlx5dr_matcher *matcher = rule->matcher;
+	int ret;
+
+	ret = dr_rule_destroy_rule(rule);
+	if (!ret)
+		refcount_dec(&matcher->refcount);
+
+	return ret;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_send.c b/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_send.c
new file mode 100644
index 000000000..ef19a66f5
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_send.c
@@ -0,0 +1,1056 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2019 Mellanox Technologies. */
+
+#include <linux/smp.h>
+#include "dr_types.h"
+
+#define QUEUE_SIZE 128
+#define SIGNAL_PER_DIV_QUEUE 16
+#define TH_NUMS_TO_DRAIN 2
+
+enum { CQ_OK = 0, CQ_EMPTY = -1, CQ_POLL_ERR = -2 };
+
+struct dr_data_seg {
+	u64 addr;
+	u32 length;
+	u32 lkey;
+	unsigned int send_flags;
+};
+
+struct postsend_info {
+	struct dr_data_seg write;
+	struct dr_data_seg read;
+	u64 remote_addr;
+	u32 rkey;
+};
+
+struct dr_qp_rtr_attr {
+	struct mlx5dr_cmd_gid_attr dgid_attr;
+	enum ib_mtu mtu;
+	u32 qp_num;
+	u16 port_num;
+	u8 min_rnr_timer;
+	u8 sgid_index;
+	u16 udp_src_port;
+	u8 fl:1;
+};
+
+struct dr_qp_rts_attr {
+	u8 timeout;
+	u8 retry_cnt;
+	u8 rnr_retry;
+};
+
+struct dr_qp_init_attr {
+	u32 cqn;
+	u32 pdn;
+	u32 max_send_wr;
+	struct mlx5_uars_page *uar;
+	u8 isolate_vl_tc:1;
+};
+
+static int dr_parse_cqe(struct mlx5dr_cq *dr_cq, struct mlx5_cqe64 *cqe64)
+{
+	unsigned int idx;
+	u8 opcode;
+
+	opcode = get_cqe_opcode(cqe64);
+	if (opcode == MLX5_CQE_REQ_ERR) {
+		idx = be16_to_cpu(cqe64->wqe_counter) &
+			(dr_cq->qp->sq.wqe_cnt - 1);
+		dr_cq->qp->sq.cc = dr_cq->qp->sq.wqe_head[idx] + 1;
+	} else if (opcode == MLX5_CQE_RESP_ERR) {
+		++dr_cq->qp->sq.cc;
+	} else {
+		idx = be16_to_cpu(cqe64->wqe_counter) &
+			(dr_cq->qp->sq.wqe_cnt - 1);
+		dr_cq->qp->sq.cc = dr_cq->qp->sq.wqe_head[idx] + 1;
+
+		return CQ_OK;
+	}
+
+	return CQ_POLL_ERR;
+}
+
+static int dr_cq_poll_one(struct mlx5dr_cq *dr_cq)
+{
+	struct mlx5_cqe64 *cqe64;
+	int err;
+
+	cqe64 = mlx5_cqwq_get_cqe(&dr_cq->wq);
+	if (!cqe64)
+		return CQ_EMPTY;
+
+	mlx5_cqwq_pop(&dr_cq->wq);
+	err = dr_parse_cqe(dr_cq, cqe64);
+	mlx5_cqwq_update_db_record(&dr_cq->wq);
+
+	return err;
+}
+
+static int dr_poll_cq(struct mlx5dr_cq *dr_cq, int ne)
+{
+	int npolled;
+	int err = 0;
+
+	for (npolled = 0; npolled < ne; ++npolled) {
+		err = dr_cq_poll_one(dr_cq);
+		if (err != CQ_OK)
+			break;
+	}
+
+	return err == CQ_POLL_ERR ? err : npolled;
+}
+
+static struct mlx5dr_qp *dr_create_rc_qp(struct mlx5_core_dev *mdev,
+					 struct dr_qp_init_attr *attr)
+{
+	u32 out[MLX5_ST_SZ_DW(create_qp_out)] = {};
+	u32 temp_qpc[MLX5_ST_SZ_DW(qpc)] = {};
+	struct mlx5_wq_param wqp;
+	struct mlx5dr_qp *dr_qp;
+	int inlen;
+	void *qpc;
+	void *in;
+	int err;
+
+	dr_qp = kzalloc(sizeof(*dr_qp), GFP_KERNEL);
+	if (!dr_qp)
+		return NULL;
+
+	wqp.buf_numa_node = mdev->priv.numa_node;
+	wqp.db_numa_node = mdev->priv.numa_node;
+
+	dr_qp->rq.pc = 0;
+	dr_qp->rq.cc = 0;
+	dr_qp->rq.wqe_cnt = 4;
+	dr_qp->sq.pc = 0;
+	dr_qp->sq.cc = 0;
+	dr_qp->sq.wqe_cnt = roundup_pow_of_two(attr->max_send_wr);
+
+	MLX5_SET(qpc, temp_qpc, log_rq_stride, ilog2(MLX5_SEND_WQE_DS) - 4);
+	MLX5_SET(qpc, temp_qpc, log_rq_size, ilog2(dr_qp->rq.wqe_cnt));
+	MLX5_SET(qpc, temp_qpc, log_sq_size, ilog2(dr_qp->sq.wqe_cnt));
+	err = mlx5_wq_qp_create(mdev, &wqp, temp_qpc, &dr_qp->wq,
+				&dr_qp->wq_ctrl);
+	if (err) {
+		mlx5_core_warn(mdev, "Can't create QP WQ\n");
+		goto err_wq;
+	}
+
+	dr_qp->sq.wqe_head = kcalloc(dr_qp->sq.wqe_cnt,
+				     sizeof(dr_qp->sq.wqe_head[0]),
+				     GFP_KERNEL);
+
+	if (!dr_qp->sq.wqe_head) {
+		mlx5_core_warn(mdev, "Can't allocate wqe head\n");
+		goto err_wqe_head;
+	}
+
+	inlen = MLX5_ST_SZ_BYTES(create_qp_in) +
+		MLX5_FLD_SZ_BYTES(create_qp_in, pas[0]) *
+		dr_qp->wq_ctrl.buf.npages;
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in) {
+		err = -ENOMEM;
+		goto err_in;
+	}
+
+	qpc = MLX5_ADDR_OF(create_qp_in, in, qpc);
+	MLX5_SET(qpc, qpc, st, MLX5_QP_ST_RC);
+	MLX5_SET(qpc, qpc, pm_state, MLX5_QP_PM_MIGRATED);
+	MLX5_SET(qpc, qpc, isolate_vl_tc, attr->isolate_vl_tc);
+	MLX5_SET(qpc, qpc, pd, attr->pdn);
+	MLX5_SET(qpc, qpc, uar_page, attr->uar->index);
+	MLX5_SET(qpc, qpc, log_page_size,
+		 dr_qp->wq_ctrl.buf.page_shift - MLX5_ADAPTER_PAGE_SHIFT);
+	MLX5_SET(qpc, qpc, fre, 1);
+	MLX5_SET(qpc, qpc, rlky, 1);
+	MLX5_SET(qpc, qpc, cqn_snd, attr->cqn);
+	MLX5_SET(qpc, qpc, cqn_rcv, attr->cqn);
+	MLX5_SET(qpc, qpc, log_rq_stride, ilog2(MLX5_SEND_WQE_DS) - 4);
+	MLX5_SET(qpc, qpc, log_rq_size, ilog2(dr_qp->rq.wqe_cnt));
+	MLX5_SET(qpc, qpc, rq_type, MLX5_NON_ZERO_RQ);
+	MLX5_SET(qpc, qpc, log_sq_size, ilog2(dr_qp->sq.wqe_cnt));
+	MLX5_SET(qpc, qpc, ts_format, mlx5_get_qp_default_ts(mdev));
+	MLX5_SET64(qpc, qpc, dbr_addr, dr_qp->wq_ctrl.db.dma);
+	if (MLX5_CAP_GEN(mdev, cqe_version) == 1)
+		MLX5_SET(qpc, qpc, user_index, 0xFFFFFF);
+	mlx5_fill_page_frag_array(&dr_qp->wq_ctrl.buf,
+				  (__be64 *)MLX5_ADDR_OF(create_qp_in,
+							 in, pas));
+
+	MLX5_SET(create_qp_in, in, opcode, MLX5_CMD_OP_CREATE_QP);
+	err = mlx5_cmd_exec(mdev, in, inlen, out, sizeof(out));
+	dr_qp->qpn = MLX5_GET(create_qp_out, out, qpn);
+	kvfree(in);
+	if (err)
+		goto err_in;
+	dr_qp->uar = attr->uar;
+
+	return dr_qp;
+
+err_in:
+	kfree(dr_qp->sq.wqe_head);
+err_wqe_head:
+	mlx5_wq_destroy(&dr_qp->wq_ctrl);
+err_wq:
+	kfree(dr_qp);
+	return NULL;
+}
+
+static void dr_destroy_qp(struct mlx5_core_dev *mdev,
+			  struct mlx5dr_qp *dr_qp)
+{
+	u32 in[MLX5_ST_SZ_DW(destroy_qp_in)] = {};
+
+	MLX5_SET(destroy_qp_in, in, opcode, MLX5_CMD_OP_DESTROY_QP);
+	MLX5_SET(destroy_qp_in, in, qpn, dr_qp->qpn);
+	mlx5_cmd_exec_in(mdev, destroy_qp, in);
+
+	kfree(dr_qp->sq.wqe_head);
+	mlx5_wq_destroy(&dr_qp->wq_ctrl);
+	kfree(dr_qp);
+}
+
+static void dr_cmd_notify_hw(struct mlx5dr_qp *dr_qp, void *ctrl)
+{
+	dma_wmb();
+	*dr_qp->wq.sq.db = cpu_to_be32(dr_qp->sq.pc & 0xffff);
+
+	/* After wmb() the hw aware of new work */
+	wmb();
+
+	mlx5_write64(ctrl, dr_qp->uar->map + MLX5_BF_OFFSET);
+}
+
+static void dr_rdma_segments(struct mlx5dr_qp *dr_qp, u64 remote_addr,
+			     u32 rkey, struct dr_data_seg *data_seg,
+			     u32 opcode, bool notify_hw)
+{
+	struct mlx5_wqe_raddr_seg *wq_raddr;
+	struct mlx5_wqe_ctrl_seg *wq_ctrl;
+	struct mlx5_wqe_data_seg *wq_dseg;
+	unsigned int size;
+	unsigned int idx;
+
+	size = sizeof(*wq_ctrl) / 16 + sizeof(*wq_dseg) / 16 +
+		sizeof(*wq_raddr) / 16;
+
+	idx = dr_qp->sq.pc & (dr_qp->sq.wqe_cnt - 1);
+
+	wq_ctrl = mlx5_wq_cyc_get_wqe(&dr_qp->wq.sq, idx);
+	wq_ctrl->imm = 0;
+	wq_ctrl->fm_ce_se = (data_seg->send_flags) ?
+		MLX5_WQE_CTRL_CQ_UPDATE : 0;
+	wq_ctrl->opmod_idx_opcode = cpu_to_be32(((dr_qp->sq.pc & 0xffff) << 8) |
+						opcode);
+	wq_ctrl->qpn_ds = cpu_to_be32(size | dr_qp->qpn << 8);
+	wq_raddr = (void *)(wq_ctrl + 1);
+	wq_raddr->raddr = cpu_to_be64(remote_addr);
+	wq_raddr->rkey = cpu_to_be32(rkey);
+	wq_raddr->reserved = 0;
+
+	wq_dseg = (void *)(wq_raddr + 1);
+	wq_dseg->byte_count = cpu_to_be32(data_seg->length);
+	wq_dseg->lkey = cpu_to_be32(data_seg->lkey);
+	wq_dseg->addr = cpu_to_be64(data_seg->addr);
+
+	dr_qp->sq.wqe_head[idx] = dr_qp->sq.pc++;
+
+	if (notify_hw)
+		dr_cmd_notify_hw(dr_qp, wq_ctrl);
+}
+
+static void dr_post_send(struct mlx5dr_qp *dr_qp, struct postsend_info *send_info)
+{
+	dr_rdma_segments(dr_qp, send_info->remote_addr, send_info->rkey,
+			 &send_info->write, MLX5_OPCODE_RDMA_WRITE, false);
+	dr_rdma_segments(dr_qp, send_info->remote_addr, send_info->rkey,
+			 &send_info->read, MLX5_OPCODE_RDMA_READ, true);
+}
+
+/**
+ * mlx5dr_send_fill_and_append_ste_send_info: Add data to be sent
+ * with send_list parameters:
+ *
+ *     @ste:       The data that attached to this specific ste
+ *     @size:      of data to write
+ *     @offset:    of the data from start of the hw_ste entry
+ *     @data:      data
+ *     @ste_info:  ste to be sent with send_list
+ *     @send_list: to append into it
+ *     @copy_data: if true indicates that the data should be kept because
+ *                 it's not backuped any where (like in re-hash).
+ *                 if false, it lets the data to be updated after
+ *                 it was added to the list.
+ */
+void mlx5dr_send_fill_and_append_ste_send_info(struct mlx5dr_ste *ste, u16 size,
+					       u16 offset, u8 *data,
+					       struct mlx5dr_ste_send_info *ste_info,
+					       struct list_head *send_list,
+					       bool copy_data)
+{
+	ste_info->size = size;
+	ste_info->ste = ste;
+	ste_info->offset = offset;
+
+	if (copy_data) {
+		memcpy(ste_info->data_cont, data, size);
+		ste_info->data = ste_info->data_cont;
+	} else {
+		ste_info->data = data;
+	}
+
+	list_add_tail(&ste_info->send_list, send_list);
+}
+
+/* The function tries to consume one wc each time, unless the queue is full, in
+ * that case, which means that the hw is behind the sw in a full queue len
+ * the function will drain the cq till it empty.
+ */
+static int dr_handle_pending_wc(struct mlx5dr_domain *dmn,
+				struct mlx5dr_send_ring *send_ring)
+{
+	bool is_drain = false;
+	int ne;
+
+	if (send_ring->pending_wqe < send_ring->signal_th)
+		return 0;
+
+	/* Queue is full start drain it */
+	if (send_ring->pending_wqe >=
+	    dmn->send_ring->signal_th * TH_NUMS_TO_DRAIN)
+		is_drain = true;
+
+	do {
+		ne = dr_poll_cq(send_ring->cq, 1);
+		if (unlikely(ne < 0)) {
+			mlx5_core_warn_once(dmn->mdev, "SMFS QPN 0x%x is disabled/limited",
+					    send_ring->qp->qpn);
+			send_ring->err_state = true;
+			return ne;
+		} else if (ne == 1) {
+			send_ring->pending_wqe -= send_ring->signal_th;
+		}
+	} while (is_drain && send_ring->pending_wqe);
+
+	return 0;
+}
+
+static void dr_fill_data_segs(struct mlx5dr_send_ring *send_ring,
+			      struct postsend_info *send_info)
+{
+	send_ring->pending_wqe++;
+
+	if (send_ring->pending_wqe % send_ring->signal_th == 0)
+		send_info->write.send_flags |= IB_SEND_SIGNALED;
+
+	send_ring->pending_wqe++;
+	send_info->read.length = send_info->write.length;
+	/* Read into the same write area */
+	send_info->read.addr = (uintptr_t)send_info->write.addr;
+	send_info->read.lkey = send_ring->mr->mkey;
+
+	if (send_ring->pending_wqe % send_ring->signal_th == 0)
+		send_info->read.send_flags = IB_SEND_SIGNALED;
+	else
+		send_info->read.send_flags = 0;
+}
+
+static int dr_postsend_icm_data(struct mlx5dr_domain *dmn,
+				struct postsend_info *send_info)
+{
+	struct mlx5dr_send_ring *send_ring = dmn->send_ring;
+	u32 buff_offset;
+	int ret;
+
+	if (unlikely(dmn->mdev->state == MLX5_DEVICE_STATE_INTERNAL_ERROR ||
+		     send_ring->err_state)) {
+		mlx5_core_dbg_once(dmn->mdev,
+				   "Skipping post send: QP err state: %d, device state: %d\n",
+				   send_ring->err_state, dmn->mdev->state);
+		return 0;
+	}
+
+	spin_lock(&send_ring->lock);
+
+	ret = dr_handle_pending_wc(dmn, send_ring);
+	if (ret)
+		goto out_unlock;
+
+	if (send_info->write.length > dmn->info.max_inline_size) {
+		buff_offset = (send_ring->tx_head &
+			       (dmn->send_ring->signal_th - 1)) *
+			send_ring->max_post_send_size;
+		/* Copy to ring mr */
+		memcpy(send_ring->buf + buff_offset,
+		       (void *)(uintptr_t)send_info->write.addr,
+		       send_info->write.length);
+		send_info->write.addr = (uintptr_t)send_ring->mr->dma_addr + buff_offset;
+		send_info->write.lkey = send_ring->mr->mkey;
+	}
+
+	send_ring->tx_head++;
+	dr_fill_data_segs(send_ring, send_info);
+	dr_post_send(send_ring->qp, send_info);
+
+out_unlock:
+	spin_unlock(&send_ring->lock);
+	return ret;
+}
+
+static int dr_get_tbl_copy_details(struct mlx5dr_domain *dmn,
+				   struct mlx5dr_ste_htbl *htbl,
+				   u8 **data,
+				   u32 *byte_size,
+				   int *iterations,
+				   int *num_stes)
+{
+	u32 chunk_byte_size = mlx5dr_icm_pool_get_chunk_byte_size(htbl->chunk);
+	int alloc_size;
+
+	if (chunk_byte_size > dmn->send_ring->max_post_send_size) {
+		*iterations = chunk_byte_size / dmn->send_ring->max_post_send_size;
+		*byte_size = dmn->send_ring->max_post_send_size;
+		alloc_size = *byte_size;
+		*num_stes = *byte_size / DR_STE_SIZE;
+	} else {
+		*iterations = 1;
+		*num_stes = mlx5dr_icm_pool_get_chunk_num_of_entries(htbl->chunk);
+		alloc_size = *num_stes * DR_STE_SIZE;
+	}
+
+	*data = kvzalloc(alloc_size, GFP_KERNEL);
+	if (!*data)
+		return -ENOMEM;
+
+	return 0;
+}
+
+/**
+ * mlx5dr_send_postsend_ste: write size bytes into offset from the hw cm.
+ *
+ *     @dmn:    Domain
+ *     @ste:    The ste struct that contains the data (at
+ *              least part of it)
+ *     @data:   The real data to send size data
+ *     @size:   for writing.
+ *     @offset: The offset from the icm mapped data to
+ *              start write to this for write only part of the
+ *              buffer.
+ *
+ * Return: 0 on success.
+ */
+int mlx5dr_send_postsend_ste(struct mlx5dr_domain *dmn, struct mlx5dr_ste *ste,
+			     u8 *data, u16 size, u16 offset)
+{
+	struct postsend_info send_info = {};
+
+	mlx5dr_ste_prepare_for_postsend(dmn->ste_ctx, data, size);
+
+	send_info.write.addr = (uintptr_t)data;
+	send_info.write.length = size;
+	send_info.write.lkey = 0;
+	send_info.remote_addr = mlx5dr_ste_get_mr_addr(ste) + offset;
+	send_info.rkey = mlx5dr_icm_pool_get_chunk_rkey(ste->htbl->chunk);
+
+	return dr_postsend_icm_data(dmn, &send_info);
+}
+
+int mlx5dr_send_postsend_htbl(struct mlx5dr_domain *dmn,
+			      struct mlx5dr_ste_htbl *htbl,
+			      u8 *formatted_ste, u8 *mask)
+{
+	u32 byte_size = mlx5dr_icm_pool_get_chunk_byte_size(htbl->chunk);
+	int num_stes_per_iter;
+	int iterations;
+	u8 *data;
+	int ret;
+	int i;
+	int j;
+
+	ret = dr_get_tbl_copy_details(dmn, htbl, &data, &byte_size,
+				      &iterations, &num_stes_per_iter);
+	if (ret)
+		return ret;
+
+	mlx5dr_ste_prepare_for_postsend(dmn->ste_ctx, formatted_ste, DR_STE_SIZE);
+
+	/* Send the data iteration times */
+	for (i = 0; i < iterations; i++) {
+		u32 ste_index = i * (byte_size / DR_STE_SIZE);
+		struct postsend_info send_info = {};
+
+		/* Copy all ste's on the data buffer
+		 * need to add the bit_mask
+		 */
+		for (j = 0; j < num_stes_per_iter; j++) {
+			struct mlx5dr_ste *ste = &htbl->chunk->ste_arr[ste_index + j];
+			u32 ste_off = j * DR_STE_SIZE;
+
+			if (mlx5dr_ste_is_not_used(ste)) {
+				memcpy(data + ste_off,
+				       formatted_ste, DR_STE_SIZE);
+			} else {
+				/* Copy data */
+				memcpy(data + ste_off,
+				       htbl->chunk->hw_ste_arr +
+				       DR_STE_SIZE_REDUCED * (ste_index + j),
+				       DR_STE_SIZE_REDUCED);
+				/* Copy bit_mask */
+				memcpy(data + ste_off + DR_STE_SIZE_REDUCED,
+				       mask, DR_STE_SIZE_MASK);
+				/* Only when we have mask we need to re-arrange the STE */
+				mlx5dr_ste_prepare_for_postsend(dmn->ste_ctx,
+								data + (j * DR_STE_SIZE),
+								DR_STE_SIZE);
+			}
+		}
+
+		send_info.write.addr = (uintptr_t)data;
+		send_info.write.length = byte_size;
+		send_info.write.lkey = 0;
+		send_info.remote_addr =
+			mlx5dr_ste_get_mr_addr(htbl->chunk->ste_arr + ste_index);
+		send_info.rkey = mlx5dr_icm_pool_get_chunk_rkey(htbl->chunk);
+
+		ret = dr_postsend_icm_data(dmn, &send_info);
+		if (ret)
+			goto out_free;
+	}
+
+out_free:
+	kvfree(data);
+	return ret;
+}
+
+/* Initialize htble with default STEs */
+int mlx5dr_send_postsend_formatted_htbl(struct mlx5dr_domain *dmn,
+					struct mlx5dr_ste_htbl *htbl,
+					u8 *ste_init_data,
+					bool update_hw_ste)
+{
+	u32 byte_size = mlx5dr_icm_pool_get_chunk_byte_size(htbl->chunk);
+	int iterations;
+	int num_stes;
+	u8 *copy_dst;
+	u8 *data;
+	int ret;
+	int i;
+
+	ret = dr_get_tbl_copy_details(dmn, htbl, &data, &byte_size,
+				      &iterations, &num_stes);
+	if (ret)
+		return ret;
+
+	if (update_hw_ste) {
+		/* Copy the reduced STE to hash table ste_arr */
+		for (i = 0; i < num_stes; i++) {
+			copy_dst = htbl->chunk->hw_ste_arr + i * DR_STE_SIZE_REDUCED;
+			memcpy(copy_dst, ste_init_data, DR_STE_SIZE_REDUCED);
+		}
+	}
+
+	mlx5dr_ste_prepare_for_postsend(dmn->ste_ctx, ste_init_data, DR_STE_SIZE);
+
+	/* Copy the same STE on the data buffer */
+	for (i = 0; i < num_stes; i++) {
+		copy_dst = data + i * DR_STE_SIZE;
+		memcpy(copy_dst, ste_init_data, DR_STE_SIZE);
+	}
+
+	/* Send the data iteration times */
+	for (i = 0; i < iterations; i++) {
+		u8 ste_index = i * (byte_size / DR_STE_SIZE);
+		struct postsend_info send_info = {};
+
+		send_info.write.addr = (uintptr_t)data;
+		send_info.write.length = byte_size;
+		send_info.write.lkey = 0;
+		send_info.remote_addr =
+			mlx5dr_ste_get_mr_addr(htbl->chunk->ste_arr + ste_index);
+		send_info.rkey = mlx5dr_icm_pool_get_chunk_rkey(htbl->chunk);
+
+		ret = dr_postsend_icm_data(dmn, &send_info);
+		if (ret)
+			goto out_free;
+	}
+
+out_free:
+	kvfree(data);
+	return ret;
+}
+
+int mlx5dr_send_postsend_action(struct mlx5dr_domain *dmn,
+				struct mlx5dr_action *action)
+{
+	struct postsend_info send_info = {};
+	int ret;
+
+	send_info.write.addr = (uintptr_t)action->rewrite->data;
+	send_info.write.length = action->rewrite->num_of_actions *
+				 DR_MODIFY_ACTION_SIZE;
+	send_info.write.lkey = 0;
+	send_info.remote_addr =
+		mlx5dr_icm_pool_get_chunk_mr_addr(action->rewrite->chunk);
+	send_info.rkey = mlx5dr_icm_pool_get_chunk_rkey(action->rewrite->chunk);
+
+	ret = dr_postsend_icm_data(dmn, &send_info);
+
+	return ret;
+}
+
+static int dr_modify_qp_rst2init(struct mlx5_core_dev *mdev,
+				 struct mlx5dr_qp *dr_qp,
+				 int port)
+{
+	u32 in[MLX5_ST_SZ_DW(rst2init_qp_in)] = {};
+	void *qpc;
+
+	qpc = MLX5_ADDR_OF(rst2init_qp_in, in, qpc);
+
+	MLX5_SET(qpc, qpc, primary_address_path.vhca_port_num, port);
+	MLX5_SET(qpc, qpc, pm_state, MLX5_QPC_PM_STATE_MIGRATED);
+	MLX5_SET(qpc, qpc, rre, 1);
+	MLX5_SET(qpc, qpc, rwe, 1);
+
+	MLX5_SET(rst2init_qp_in, in, opcode, MLX5_CMD_OP_RST2INIT_QP);
+	MLX5_SET(rst2init_qp_in, in, qpn, dr_qp->qpn);
+
+	return mlx5_cmd_exec_in(mdev, rst2init_qp, in);
+}
+
+static int dr_cmd_modify_qp_rtr2rts(struct mlx5_core_dev *mdev,
+				    struct mlx5dr_qp *dr_qp,
+				    struct dr_qp_rts_attr *attr)
+{
+	u32 in[MLX5_ST_SZ_DW(rtr2rts_qp_in)] = {};
+	void *qpc;
+
+	qpc  = MLX5_ADDR_OF(rtr2rts_qp_in, in, qpc);
+
+	MLX5_SET(rtr2rts_qp_in, in, qpn, dr_qp->qpn);
+
+	MLX5_SET(qpc, qpc, retry_count, attr->retry_cnt);
+	MLX5_SET(qpc, qpc, rnr_retry, attr->rnr_retry);
+	MLX5_SET(qpc, qpc, primary_address_path.ack_timeout, 0x8); /* ~1ms */
+
+	MLX5_SET(rtr2rts_qp_in, in, opcode, MLX5_CMD_OP_RTR2RTS_QP);
+	MLX5_SET(rtr2rts_qp_in, in, qpn, dr_qp->qpn);
+
+	return mlx5_cmd_exec_in(mdev, rtr2rts_qp, in);
+}
+
+static int dr_cmd_modify_qp_init2rtr(struct mlx5_core_dev *mdev,
+				     struct mlx5dr_qp *dr_qp,
+				     struct dr_qp_rtr_attr *attr)
+{
+	u32 in[MLX5_ST_SZ_DW(init2rtr_qp_in)] = {};
+	void *qpc;
+
+	qpc = MLX5_ADDR_OF(init2rtr_qp_in, in, qpc);
+
+	MLX5_SET(init2rtr_qp_in, in, qpn, dr_qp->qpn);
+
+	MLX5_SET(qpc, qpc, mtu, attr->mtu);
+	MLX5_SET(qpc, qpc, log_msg_max, DR_CHUNK_SIZE_MAX - 1);
+	MLX5_SET(qpc, qpc, remote_qpn, attr->qp_num);
+	memcpy(MLX5_ADDR_OF(qpc, qpc, primary_address_path.rmac_47_32),
+	       attr->dgid_attr.mac, sizeof(attr->dgid_attr.mac));
+	memcpy(MLX5_ADDR_OF(qpc, qpc, primary_address_path.rgid_rip),
+	       attr->dgid_attr.gid, sizeof(attr->dgid_attr.gid));
+	MLX5_SET(qpc, qpc, primary_address_path.src_addr_index,
+		 attr->sgid_index);
+
+	if (attr->dgid_attr.roce_ver == MLX5_ROCE_VERSION_2)
+		MLX5_SET(qpc, qpc, primary_address_path.udp_sport,
+			 attr->udp_src_port);
+
+	MLX5_SET(qpc, qpc, primary_address_path.vhca_port_num, attr->port_num);
+	MLX5_SET(qpc, qpc, primary_address_path.fl, attr->fl);
+	MLX5_SET(qpc, qpc, min_rnr_nak, 1);
+
+	MLX5_SET(init2rtr_qp_in, in, opcode, MLX5_CMD_OP_INIT2RTR_QP);
+	MLX5_SET(init2rtr_qp_in, in, qpn, dr_qp->qpn);
+
+	return mlx5_cmd_exec_in(mdev, init2rtr_qp, in);
+}
+
+static bool dr_send_allow_fl(struct mlx5dr_cmd_caps *caps)
+{
+	/* Check whether RC RoCE QP creation with force loopback is allowed.
+	 * There are two separate capability bits for this:
+	 *  - force loopback when RoCE is enabled
+	 *  - force loopback when RoCE is disabled
+	 */
+	return ((caps->roce_caps.roce_en &&
+		 caps->roce_caps.fl_rc_qp_when_roce_enabled) ||
+		(!caps->roce_caps.roce_en &&
+		 caps->roce_caps.fl_rc_qp_when_roce_disabled));
+}
+
+static int dr_prepare_qp_to_rts(struct mlx5dr_domain *dmn)
+{
+	struct mlx5dr_qp *dr_qp = dmn->send_ring->qp;
+	struct dr_qp_rts_attr rts_attr = {};
+	struct dr_qp_rtr_attr rtr_attr = {};
+	enum ib_mtu mtu = IB_MTU_1024;
+	u16 gid_index = 0;
+	int port = 1;
+	int ret;
+
+	/* Init */
+	ret = dr_modify_qp_rst2init(dmn->mdev, dr_qp, port);
+	if (ret) {
+		mlx5dr_err(dmn, "Failed modify QP rst2init\n");
+		return ret;
+	}
+
+	/* RTR */
+	rtr_attr.mtu		= mtu;
+	rtr_attr.qp_num		= dr_qp->qpn;
+	rtr_attr.min_rnr_timer	= 12;
+	rtr_attr.port_num	= port;
+	rtr_attr.udp_src_port	= dmn->info.caps.roce_min_src_udp;
+
+	/* If QP creation with force loopback is allowed, then there
+	 * is no need for GID index when creating the QP.
+	 * Otherwise we query GID attributes and use GID index.
+	 */
+	rtr_attr.fl = dr_send_allow_fl(&dmn->info.caps);
+	if (!rtr_attr.fl) {
+		ret = mlx5dr_cmd_query_gid(dmn->mdev, port, gid_index,
+					   &rtr_attr.dgid_attr);
+		if (ret)
+			return ret;
+
+		rtr_attr.sgid_index = gid_index;
+	}
+
+	ret = dr_cmd_modify_qp_init2rtr(dmn->mdev, dr_qp, &rtr_attr);
+	if (ret) {
+		mlx5dr_err(dmn, "Failed modify QP init2rtr\n");
+		return ret;
+	}
+
+	/* RTS */
+	rts_attr.timeout	= 14;
+	rts_attr.retry_cnt	= 7;
+	rts_attr.rnr_retry	= 7;
+
+	ret = dr_cmd_modify_qp_rtr2rts(dmn->mdev, dr_qp, &rts_attr);
+	if (ret) {
+		mlx5dr_err(dmn, "Failed modify QP rtr2rts\n");
+		return ret;
+	}
+
+	return 0;
+}
+
+static void dr_cq_complete(struct mlx5_core_cq *mcq,
+			   struct mlx5_eqe *eqe)
+{
+	pr_err("CQ completion CQ: #%u\n", mcq->cqn);
+}
+
+static struct mlx5dr_cq *dr_create_cq(struct mlx5_core_dev *mdev,
+				      struct mlx5_uars_page *uar,
+				      size_t ncqe)
+{
+	u32 temp_cqc[MLX5_ST_SZ_DW(cqc)] = {};
+	u32 out[MLX5_ST_SZ_DW(create_cq_out)];
+	struct mlx5_wq_param wqp;
+	struct mlx5_cqe64 *cqe;
+	struct mlx5dr_cq *cq;
+	int inlen, err, eqn;
+	void *cqc, *in;
+	__be64 *pas;
+	int vector;
+	u32 i;
+
+	cq = kzalloc(sizeof(*cq), GFP_KERNEL);
+	if (!cq)
+		return NULL;
+
+	ncqe = roundup_pow_of_two(ncqe);
+	MLX5_SET(cqc, temp_cqc, log_cq_size, ilog2(ncqe));
+
+	wqp.buf_numa_node = mdev->priv.numa_node;
+	wqp.db_numa_node = mdev->priv.numa_node;
+
+	err = mlx5_cqwq_create(mdev, &wqp, temp_cqc, &cq->wq,
+			       &cq->wq_ctrl);
+	if (err)
+		goto out;
+
+	for (i = 0; i < mlx5_cqwq_get_size(&cq->wq); i++) {
+		cqe = mlx5_cqwq_get_wqe(&cq->wq, i);
+		cqe->op_own = MLX5_CQE_INVALID << 4 | MLX5_CQE_OWNER_MASK;
+	}
+
+	inlen = MLX5_ST_SZ_BYTES(create_cq_in) +
+		sizeof(u64) * cq->wq_ctrl.buf.npages;
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in)
+		goto err_cqwq;
+
+	vector = raw_smp_processor_id() % mlx5_comp_vectors_count(mdev);
+	err = mlx5_vector2eqn(mdev, vector, &eqn);
+	if (err) {
+		kvfree(in);
+		goto err_cqwq;
+	}
+
+	cqc = MLX5_ADDR_OF(create_cq_in, in, cq_context);
+	MLX5_SET(cqc, cqc, log_cq_size, ilog2(ncqe));
+	MLX5_SET(cqc, cqc, c_eqn_or_apu_element, eqn);
+	MLX5_SET(cqc, cqc, uar_page, uar->index);
+	MLX5_SET(cqc, cqc, log_page_size, cq->wq_ctrl.buf.page_shift -
+		 MLX5_ADAPTER_PAGE_SHIFT);
+	MLX5_SET64(cqc, cqc, dbr_addr, cq->wq_ctrl.db.dma);
+
+	pas = (__be64 *)MLX5_ADDR_OF(create_cq_in, in, pas);
+	mlx5_fill_page_frag_array(&cq->wq_ctrl.buf, pas);
+
+	cq->mcq.comp  = dr_cq_complete;
+
+	err = mlx5_core_create_cq(mdev, &cq->mcq, in, inlen, out, sizeof(out));
+	kvfree(in);
+
+	if (err)
+		goto err_cqwq;
+
+	cq->mcq.cqe_sz = 64;
+	cq->mcq.set_ci_db = cq->wq_ctrl.db.db;
+	cq->mcq.arm_db = cq->wq_ctrl.db.db + 1;
+	*cq->mcq.set_ci_db = 0;
+
+	/* set no-zero value, in order to avoid the HW to run db-recovery on
+	 * CQ that used in polling mode.
+	 */
+	*cq->mcq.arm_db = cpu_to_be32(2 << 28);
+
+	cq->mcq.vector = 0;
+	cq->mcq.uar = uar;
+
+	return cq;
+
+err_cqwq:
+	mlx5_wq_destroy(&cq->wq_ctrl);
+out:
+	kfree(cq);
+	return NULL;
+}
+
+static void dr_destroy_cq(struct mlx5_core_dev *mdev, struct mlx5dr_cq *cq)
+{
+	mlx5_core_destroy_cq(mdev, &cq->mcq);
+	mlx5_wq_destroy(&cq->wq_ctrl);
+	kfree(cq);
+}
+
+static int dr_create_mkey(struct mlx5_core_dev *mdev, u32 pdn, u32 *mkey)
+{
+	u32 in[MLX5_ST_SZ_DW(create_mkey_in)] = {};
+	void *mkc;
+
+	mkc = MLX5_ADDR_OF(create_mkey_in, in, memory_key_mkey_entry);
+	MLX5_SET(mkc, mkc, access_mode_1_0, MLX5_MKC_ACCESS_MODE_PA);
+	MLX5_SET(mkc, mkc, a, 1);
+	MLX5_SET(mkc, mkc, rw, 1);
+	MLX5_SET(mkc, mkc, rr, 1);
+	MLX5_SET(mkc, mkc, lw, 1);
+	MLX5_SET(mkc, mkc, lr, 1);
+
+	MLX5_SET(mkc, mkc, pd, pdn);
+	MLX5_SET(mkc, mkc, length64, 1);
+	MLX5_SET(mkc, mkc, qpn, 0xffffff);
+
+	return mlx5_core_create_mkey(mdev, mkey, in, sizeof(in));
+}
+
+static struct mlx5dr_mr *dr_reg_mr(struct mlx5_core_dev *mdev,
+				   u32 pdn, void *buf, size_t size)
+{
+	struct mlx5dr_mr *mr = kzalloc(sizeof(*mr), GFP_KERNEL);
+	struct device *dma_device;
+	dma_addr_t dma_addr;
+	int err;
+
+	if (!mr)
+		return NULL;
+
+	dma_device = mlx5_core_dma_dev(mdev);
+	dma_addr = dma_map_single(dma_device, buf, size,
+				  DMA_BIDIRECTIONAL);
+	err = dma_mapping_error(dma_device, dma_addr);
+	if (err) {
+		mlx5_core_warn(mdev, "Can't dma buf\n");
+		kfree(mr);
+		return NULL;
+	}
+
+	err = dr_create_mkey(mdev, pdn, &mr->mkey);
+	if (err) {
+		mlx5_core_warn(mdev, "Can't create mkey\n");
+		dma_unmap_single(dma_device, dma_addr, size,
+				 DMA_BIDIRECTIONAL);
+		kfree(mr);
+		return NULL;
+	}
+
+	mr->dma_addr = dma_addr;
+	mr->size = size;
+	mr->addr = buf;
+
+	return mr;
+}
+
+static void dr_dereg_mr(struct mlx5_core_dev *mdev, struct mlx5dr_mr *mr)
+{
+	mlx5_core_destroy_mkey(mdev, mr->mkey);
+	dma_unmap_single(mlx5_core_dma_dev(mdev), mr->dma_addr, mr->size,
+			 DMA_BIDIRECTIONAL);
+	kfree(mr);
+}
+
+int mlx5dr_send_ring_alloc(struct mlx5dr_domain *dmn)
+{
+	struct dr_qp_init_attr init_attr = {};
+	int cq_size;
+	int size;
+	int ret;
+
+	dmn->send_ring = kzalloc(sizeof(*dmn->send_ring), GFP_KERNEL);
+	if (!dmn->send_ring)
+		return -ENOMEM;
+
+	cq_size = QUEUE_SIZE + 1;
+	dmn->send_ring->cq = dr_create_cq(dmn->mdev, dmn->uar, cq_size);
+	if (!dmn->send_ring->cq) {
+		mlx5dr_err(dmn, "Failed creating CQ\n");
+		ret = -ENOMEM;
+		goto free_send_ring;
+	}
+
+	init_attr.cqn = dmn->send_ring->cq->mcq.cqn;
+	init_attr.pdn = dmn->pdn;
+	init_attr.uar = dmn->uar;
+	init_attr.max_send_wr = QUEUE_SIZE;
+
+	/* Isolated VL is applicable only if force loopback is supported */
+	if (dr_send_allow_fl(&dmn->info.caps))
+		init_attr.isolate_vl_tc = dmn->info.caps.isolate_vl_tc;
+
+	spin_lock_init(&dmn->send_ring->lock);
+
+	dmn->send_ring->qp = dr_create_rc_qp(dmn->mdev, &init_attr);
+	if (!dmn->send_ring->qp)  {
+		mlx5dr_err(dmn, "Failed creating QP\n");
+		ret = -ENOMEM;
+		goto clean_cq;
+	}
+
+	dmn->send_ring->cq->qp = dmn->send_ring->qp;
+
+	dmn->info.max_send_wr = QUEUE_SIZE;
+	dmn->info.max_inline_size = min(dmn->send_ring->qp->max_inline_data,
+					DR_STE_SIZE);
+
+	dmn->send_ring->signal_th = dmn->info.max_send_wr /
+		SIGNAL_PER_DIV_QUEUE;
+
+	/* Prepare qp to be used */
+	ret = dr_prepare_qp_to_rts(dmn);
+	if (ret)
+		goto clean_qp;
+
+	dmn->send_ring->max_post_send_size =
+		mlx5dr_icm_pool_chunk_size_to_byte(DR_CHUNK_SIZE_1K,
+						   DR_ICM_TYPE_STE);
+
+	/* Allocating the max size as a buffer for writing */
+	size = dmn->send_ring->signal_th * dmn->send_ring->max_post_send_size;
+	dmn->send_ring->buf = kzalloc(size, GFP_KERNEL);
+	if (!dmn->send_ring->buf) {
+		ret = -ENOMEM;
+		goto clean_qp;
+	}
+
+	dmn->send_ring->buf_size = size;
+
+	dmn->send_ring->mr = dr_reg_mr(dmn->mdev,
+				       dmn->pdn, dmn->send_ring->buf, size);
+	if (!dmn->send_ring->mr) {
+		ret = -ENOMEM;
+		goto free_mem;
+	}
+
+	dmn->send_ring->sync_mr = dr_reg_mr(dmn->mdev,
+					    dmn->pdn, dmn->send_ring->sync_buff,
+					    MIN_READ_SYNC);
+	if (!dmn->send_ring->sync_mr) {
+		ret = -ENOMEM;
+		goto clean_mr;
+	}
+
+	return 0;
+
+clean_mr:
+	dr_dereg_mr(dmn->mdev, dmn->send_ring->mr);
+free_mem:
+	kfree(dmn->send_ring->buf);
+clean_qp:
+	dr_destroy_qp(dmn->mdev, dmn->send_ring->qp);
+clean_cq:
+	dr_destroy_cq(dmn->mdev, dmn->send_ring->cq);
+free_send_ring:
+	kfree(dmn->send_ring);
+
+	return ret;
+}
+
+void mlx5dr_send_ring_free(struct mlx5dr_domain *dmn,
+			   struct mlx5dr_send_ring *send_ring)
+{
+	dr_destroy_qp(dmn->mdev, send_ring->qp);
+	dr_destroy_cq(dmn->mdev, send_ring->cq);
+	dr_dereg_mr(dmn->mdev, send_ring->sync_mr);
+	dr_dereg_mr(dmn->mdev, send_ring->mr);
+	kfree(send_ring->buf);
+	kfree(send_ring);
+}
+
+int mlx5dr_send_ring_force_drain(struct mlx5dr_domain *dmn)
+{
+	struct mlx5dr_send_ring *send_ring = dmn->send_ring;
+	struct postsend_info send_info = {};
+	u8 data[DR_STE_SIZE];
+	int num_of_sends_req;
+	int ret;
+	int i;
+
+	/* Sending this amount of requests makes sure we will get drain */
+	num_of_sends_req = send_ring->signal_th * TH_NUMS_TO_DRAIN / 2;
+
+	/* Send fake requests forcing the last to be signaled */
+	send_info.write.addr = (uintptr_t)data;
+	send_info.write.length = DR_STE_SIZE;
+	send_info.write.lkey = 0;
+	/* Using the sync_mr in order to write/read */
+	send_info.remote_addr = (uintptr_t)send_ring->sync_mr->addr;
+	send_info.rkey = send_ring->sync_mr->mkey;
+
+	for (i = 0; i < num_of_sends_req; i++) {
+		ret = dr_postsend_icm_data(dmn, &send_info);
+		if (ret)
+			return ret;
+	}
+
+	spin_lock(&send_ring->lock);
+	ret = dr_handle_pending_wc(dmn, send_ring);
+	spin_unlock(&send_ring->lock);
+
+	return ret;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_ste.c b/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_ste.c
new file mode 100644
index 000000000..7815a629d
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_ste.c
@@ -0,0 +1,1390 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2019 Mellanox Technologies. */
+
+#include <linux/types.h>
+#include <linux/crc32.h>
+#include "dr_ste.h"
+
+struct dr_hw_ste_format {
+	u8 ctrl[DR_STE_SIZE_CTRL];
+	u8 tag[DR_STE_SIZE_TAG];
+	u8 mask[DR_STE_SIZE_MASK];
+};
+
+static u32 dr_ste_crc32_calc(const void *input_data, size_t length)
+{
+	u32 crc = crc32(0, input_data, length);
+
+	return (__force u32)((crc >> 24) & 0xff) | ((crc << 8) & 0xff0000) |
+			    ((crc >> 8) & 0xff00) | ((crc << 24) & 0xff000000);
+}
+
+bool mlx5dr_ste_supp_ttl_cs_recalc(struct mlx5dr_cmd_caps *caps)
+{
+	return caps->sw_format_ver > MLX5_STEERING_FORMAT_CONNECTX_5;
+}
+
+u32 mlx5dr_ste_calc_hash_index(u8 *hw_ste_p, struct mlx5dr_ste_htbl *htbl)
+{
+	u32 num_entries = mlx5dr_icm_pool_get_chunk_num_of_entries(htbl->chunk);
+	struct dr_hw_ste_format *hw_ste = (struct dr_hw_ste_format *)hw_ste_p;
+	u8 masked[DR_STE_SIZE_TAG] = {};
+	u32 crc32, index;
+	u16 bit;
+	int i;
+
+	/* Don't calculate CRC if the result is predicted */
+	if (num_entries == 1 || htbl->byte_mask == 0)
+		return 0;
+
+	/* Mask tag using byte mask, bit per byte */
+	bit = 1 << (DR_STE_SIZE_TAG - 1);
+	for (i = 0; i < DR_STE_SIZE_TAG; i++) {
+		if (htbl->byte_mask & bit)
+			masked[i] = hw_ste->tag[i];
+
+		bit = bit >> 1;
+	}
+
+	crc32 = dr_ste_crc32_calc(masked, DR_STE_SIZE_TAG);
+	index = crc32 & (num_entries - 1);
+
+	return index;
+}
+
+u16 mlx5dr_ste_conv_bit_to_byte_mask(u8 *bit_mask)
+{
+	u16 byte_mask = 0;
+	int i;
+
+	for (i = 0; i < DR_STE_SIZE_MASK; i++) {
+		byte_mask = byte_mask << 1;
+		if (bit_mask[i] == 0xff)
+			byte_mask |= 1;
+	}
+	return byte_mask;
+}
+
+static u8 *dr_ste_get_tag(u8 *hw_ste_p)
+{
+	struct dr_hw_ste_format *hw_ste = (struct dr_hw_ste_format *)hw_ste_p;
+
+	return hw_ste->tag;
+}
+
+void mlx5dr_ste_set_bit_mask(u8 *hw_ste_p, u8 *bit_mask)
+{
+	struct dr_hw_ste_format *hw_ste = (struct dr_hw_ste_format *)hw_ste_p;
+
+	memcpy(hw_ste->mask, bit_mask, DR_STE_SIZE_MASK);
+}
+
+static void dr_ste_set_always_hit(struct dr_hw_ste_format *hw_ste)
+{
+	memset(&hw_ste->tag, 0, sizeof(hw_ste->tag));
+	memset(&hw_ste->mask, 0, sizeof(hw_ste->mask));
+}
+
+static void dr_ste_set_always_miss(struct dr_hw_ste_format *hw_ste)
+{
+	hw_ste->tag[0] = 0xdc;
+	hw_ste->mask[0] = 0;
+}
+
+void mlx5dr_ste_set_miss_addr(struct mlx5dr_ste_ctx *ste_ctx,
+			      u8 *hw_ste_p, u64 miss_addr)
+{
+	ste_ctx->set_miss_addr(hw_ste_p, miss_addr);
+}
+
+static void dr_ste_always_miss_addr(struct mlx5dr_ste_ctx *ste_ctx,
+				    u8 *hw_ste, u64 miss_addr)
+{
+	ste_ctx->set_next_lu_type(hw_ste, MLX5DR_STE_LU_TYPE_DONT_CARE);
+	ste_ctx->set_miss_addr(hw_ste, miss_addr);
+	dr_ste_set_always_miss((struct dr_hw_ste_format *)hw_ste);
+}
+
+void mlx5dr_ste_set_hit_addr(struct mlx5dr_ste_ctx *ste_ctx,
+			     u8 *hw_ste, u64 icm_addr, u32 ht_size)
+{
+	ste_ctx->set_hit_addr(hw_ste, icm_addr, ht_size);
+}
+
+u64 mlx5dr_ste_get_icm_addr(struct mlx5dr_ste *ste)
+{
+	u64 base_icm_addr = mlx5dr_icm_pool_get_chunk_icm_addr(ste->htbl->chunk);
+	u32 index = ste - ste->htbl->chunk->ste_arr;
+
+	return base_icm_addr + DR_STE_SIZE * index;
+}
+
+u64 mlx5dr_ste_get_mr_addr(struct mlx5dr_ste *ste)
+{
+	u32 index = ste - ste->htbl->chunk->ste_arr;
+
+	return mlx5dr_icm_pool_get_chunk_mr_addr(ste->htbl->chunk) + DR_STE_SIZE * index;
+}
+
+u8 *mlx5dr_ste_get_hw_ste(struct mlx5dr_ste *ste)
+{
+	u64 index = ste - ste->htbl->chunk->ste_arr;
+
+	return ste->htbl->chunk->hw_ste_arr + DR_STE_SIZE_REDUCED * index;
+}
+
+struct list_head *mlx5dr_ste_get_miss_list(struct mlx5dr_ste *ste)
+{
+	u32 index = ste - ste->htbl->chunk->ste_arr;
+
+	return &ste->htbl->chunk->miss_list[index];
+}
+
+static void dr_ste_always_hit_htbl(struct mlx5dr_ste_ctx *ste_ctx,
+				   u8 *hw_ste,
+				   struct mlx5dr_ste_htbl *next_htbl)
+{
+	struct mlx5dr_icm_chunk *chunk = next_htbl->chunk;
+
+	ste_ctx->set_byte_mask(hw_ste, next_htbl->byte_mask);
+	ste_ctx->set_next_lu_type(hw_ste, next_htbl->lu_type);
+	ste_ctx->set_hit_addr(hw_ste, mlx5dr_icm_pool_get_chunk_icm_addr(chunk),
+			      mlx5dr_icm_pool_get_chunk_num_of_entries(chunk));
+
+	dr_ste_set_always_hit((struct dr_hw_ste_format *)hw_ste);
+}
+
+bool mlx5dr_ste_is_last_in_rule(struct mlx5dr_matcher_rx_tx *nic_matcher,
+				u8 ste_location)
+{
+	return ste_location == nic_matcher->num_of_builders;
+}
+
+/* Replace relevant fields, except of:
+ * htbl - keep the origin htbl
+ * miss_list + list - already took the src from the list.
+ * icm_addr/mr_addr - depends on the hosting table.
+ *
+ * Before:
+ * | a | -> | b | -> | c | ->
+ *
+ * After:
+ * | a | -> | c | ->
+ * While the data that was in b copied to a.
+ */
+static void dr_ste_replace(struct mlx5dr_ste *dst, struct mlx5dr_ste *src)
+{
+	memcpy(mlx5dr_ste_get_hw_ste(dst), mlx5dr_ste_get_hw_ste(src),
+	       DR_STE_SIZE_REDUCED);
+	dst->next_htbl = src->next_htbl;
+	if (dst->next_htbl)
+		dst->next_htbl->pointing_ste = dst;
+
+	dst->refcount = src->refcount;
+}
+
+/* Free ste which is the head and the only one in miss_list */
+static void
+dr_ste_remove_head_ste(struct mlx5dr_ste_ctx *ste_ctx,
+		       struct mlx5dr_ste *ste,
+		       struct mlx5dr_matcher_rx_tx *nic_matcher,
+		       struct mlx5dr_ste_send_info *ste_info_head,
+		       struct list_head *send_ste_list,
+		       struct mlx5dr_ste_htbl *stats_tbl)
+{
+	u8 tmp_data_ste[DR_STE_SIZE] = {};
+	u64 miss_addr;
+
+	miss_addr = mlx5dr_icm_pool_get_chunk_icm_addr(nic_matcher->e_anchor->chunk);
+
+	/* Use temp ste because dr_ste_always_miss_addr
+	 * touches bit_mask area which doesn't exist at ste->hw_ste.
+	 * Need to use a full-sized (DR_STE_SIZE) hw_ste.
+	 */
+	memcpy(tmp_data_ste, mlx5dr_ste_get_hw_ste(ste), DR_STE_SIZE_REDUCED);
+	dr_ste_always_miss_addr(ste_ctx, tmp_data_ste, miss_addr);
+	memcpy(mlx5dr_ste_get_hw_ste(ste), tmp_data_ste, DR_STE_SIZE_REDUCED);
+
+	list_del_init(&ste->miss_list_node);
+
+	/* Write full STE size in order to have "always_miss" */
+	mlx5dr_send_fill_and_append_ste_send_info(ste, DR_STE_SIZE,
+						  0, tmp_data_ste,
+						  ste_info_head,
+						  send_ste_list,
+						  true /* Copy data */);
+
+	stats_tbl->ctrl.num_of_valid_entries--;
+}
+
+/* Free ste which is the head but NOT the only one in miss_list:
+ * |_ste_| --> |_next_ste_| -->|__| -->|__| -->/0
+ */
+static void
+dr_ste_replace_head_ste(struct mlx5dr_matcher_rx_tx *nic_matcher,
+			struct mlx5dr_ste *ste,
+			struct mlx5dr_ste *next_ste,
+			struct mlx5dr_ste_send_info *ste_info_head,
+			struct list_head *send_ste_list,
+			struct mlx5dr_ste_htbl *stats_tbl)
+
+{
+	struct mlx5dr_ste_htbl *next_miss_htbl;
+	u8 hw_ste[DR_STE_SIZE] = {};
+	int sb_idx;
+
+	next_miss_htbl = next_ste->htbl;
+
+	/* Remove from the miss_list the next_ste before copy */
+	list_del_init(&next_ste->miss_list_node);
+
+	/* Move data from next into ste */
+	dr_ste_replace(ste, next_ste);
+
+	/* Update the rule on STE change */
+	mlx5dr_rule_set_last_member(next_ste->rule_rx_tx, ste, false);
+
+	/* Copy all 64 hw_ste bytes */
+	memcpy(hw_ste, mlx5dr_ste_get_hw_ste(ste), DR_STE_SIZE_REDUCED);
+	sb_idx = ste->ste_chain_location - 1;
+	mlx5dr_ste_set_bit_mask(hw_ste,
+				nic_matcher->ste_builder[sb_idx].bit_mask);
+
+	/* Del the htbl that contains the next_ste.
+	 * The origin htbl stay with the same number of entries.
+	 */
+	mlx5dr_htbl_put(next_miss_htbl);
+
+	mlx5dr_send_fill_and_append_ste_send_info(ste, DR_STE_SIZE,
+						  0, hw_ste,
+						  ste_info_head,
+						  send_ste_list,
+						  true /* Copy data */);
+
+	stats_tbl->ctrl.num_of_collisions--;
+	stats_tbl->ctrl.num_of_valid_entries--;
+}
+
+/* Free ste that is located in the middle of the miss list:
+ * |__| -->|_prev_ste_|->|_ste_|-->|_next_ste_|
+ */
+static void dr_ste_remove_middle_ste(struct mlx5dr_ste_ctx *ste_ctx,
+				     struct mlx5dr_ste *ste,
+				     struct mlx5dr_ste_send_info *ste_info,
+				     struct list_head *send_ste_list,
+				     struct mlx5dr_ste_htbl *stats_tbl)
+{
+	struct mlx5dr_ste *prev_ste;
+	u64 miss_addr;
+
+	prev_ste = list_prev_entry(ste, miss_list_node);
+	if (WARN_ON(!prev_ste))
+		return;
+
+	miss_addr = ste_ctx->get_miss_addr(mlx5dr_ste_get_hw_ste(ste));
+	ste_ctx->set_miss_addr(mlx5dr_ste_get_hw_ste(prev_ste), miss_addr);
+
+	mlx5dr_send_fill_and_append_ste_send_info(prev_ste, DR_STE_SIZE_CTRL, 0,
+						  mlx5dr_ste_get_hw_ste(prev_ste),
+						  ste_info, send_ste_list,
+						  true /* Copy data*/);
+
+	list_del_init(&ste->miss_list_node);
+
+	stats_tbl->ctrl.num_of_valid_entries--;
+	stats_tbl->ctrl.num_of_collisions--;
+}
+
+void mlx5dr_ste_free(struct mlx5dr_ste *ste,
+		     struct mlx5dr_matcher *matcher,
+		     struct mlx5dr_matcher_rx_tx *nic_matcher)
+{
+	struct mlx5dr_ste_send_info *cur_ste_info, *tmp_ste_info;
+	struct mlx5dr_domain *dmn = matcher->tbl->dmn;
+	struct mlx5dr_ste_ctx *ste_ctx = dmn->ste_ctx;
+	struct mlx5dr_ste_send_info ste_info_head;
+	struct mlx5dr_ste *next_ste, *first_ste;
+	bool put_on_origin_table = true;
+	struct mlx5dr_ste_htbl *stats_tbl;
+	LIST_HEAD(send_ste_list);
+
+	first_ste = list_first_entry(mlx5dr_ste_get_miss_list(ste),
+				     struct mlx5dr_ste, miss_list_node);
+	stats_tbl = first_ste->htbl;
+
+	/* Two options:
+	 * 1. ste is head:
+	 *	a. head ste is the only ste in the miss list
+	 *	b. head ste is not the only ste in the miss-list
+	 * 2. ste is not head
+	 */
+	if (first_ste == ste) { /* Ste is the head */
+		struct mlx5dr_ste *last_ste;
+
+		last_ste = list_last_entry(mlx5dr_ste_get_miss_list(ste),
+					   struct mlx5dr_ste, miss_list_node);
+		if (last_ste == first_ste)
+			next_ste = NULL;
+		else
+			next_ste = list_next_entry(ste, miss_list_node);
+
+		if (!next_ste) {
+			/* One and only entry in the list */
+			dr_ste_remove_head_ste(ste_ctx, ste,
+					       nic_matcher,
+					       &ste_info_head,
+					       &send_ste_list,
+					       stats_tbl);
+		} else {
+			/* First but not only entry in the list */
+			dr_ste_replace_head_ste(nic_matcher, ste,
+						next_ste, &ste_info_head,
+						&send_ste_list, stats_tbl);
+			put_on_origin_table = false;
+		}
+	} else { /* Ste in the middle of the list */
+		dr_ste_remove_middle_ste(ste_ctx, ste,
+					 &ste_info_head, &send_ste_list,
+					 stats_tbl);
+	}
+
+	/* Update HW */
+	list_for_each_entry_safe(cur_ste_info, tmp_ste_info,
+				 &send_ste_list, send_list) {
+		list_del(&cur_ste_info->send_list);
+		mlx5dr_send_postsend_ste(dmn, cur_ste_info->ste,
+					 cur_ste_info->data, cur_ste_info->size,
+					 cur_ste_info->offset);
+	}
+
+	if (put_on_origin_table)
+		mlx5dr_htbl_put(ste->htbl);
+}
+
+bool mlx5dr_ste_equal_tag(void *src, void *dst)
+{
+	struct dr_hw_ste_format *s_hw_ste = (struct dr_hw_ste_format *)src;
+	struct dr_hw_ste_format *d_hw_ste = (struct dr_hw_ste_format *)dst;
+
+	return !memcmp(s_hw_ste->tag, d_hw_ste->tag, DR_STE_SIZE_TAG);
+}
+
+void mlx5dr_ste_set_hit_addr_by_next_htbl(struct mlx5dr_ste_ctx *ste_ctx,
+					  u8 *hw_ste,
+					  struct mlx5dr_ste_htbl *next_htbl)
+{
+	u64 icm_addr = mlx5dr_icm_pool_get_chunk_icm_addr(next_htbl->chunk);
+	u32 num_entries =
+		mlx5dr_icm_pool_get_chunk_num_of_entries(next_htbl->chunk);
+
+	ste_ctx->set_hit_addr(hw_ste, icm_addr, num_entries);
+}
+
+void mlx5dr_ste_prepare_for_postsend(struct mlx5dr_ste_ctx *ste_ctx,
+				     u8 *hw_ste_p, u32 ste_size)
+{
+	if (ste_ctx->prepare_for_postsend)
+		ste_ctx->prepare_for_postsend(hw_ste_p, ste_size);
+}
+
+/* Init one ste as a pattern for ste data array */
+void mlx5dr_ste_set_formatted_ste(struct mlx5dr_ste_ctx *ste_ctx,
+				  u16 gvmi,
+				  enum mlx5dr_domain_nic_type nic_type,
+				  struct mlx5dr_ste_htbl *htbl,
+				  u8 *formatted_ste,
+				  struct mlx5dr_htbl_connect_info *connect_info)
+{
+	bool is_rx = nic_type == DR_DOMAIN_NIC_TYPE_RX;
+	u8 tmp_hw_ste[DR_STE_SIZE] = {0};
+
+	ste_ctx->ste_init(formatted_ste, htbl->lu_type, is_rx, gvmi);
+
+	/* Use temp ste because dr_ste_always_miss_addr/hit_htbl
+	 * touches bit_mask area which doesn't exist at ste->hw_ste.
+	 * Need to use a full-sized (DR_STE_SIZE) hw_ste.
+	 */
+	memcpy(tmp_hw_ste, formatted_ste, DR_STE_SIZE_REDUCED);
+	if (connect_info->type == CONNECT_HIT)
+		dr_ste_always_hit_htbl(ste_ctx, tmp_hw_ste,
+				       connect_info->hit_next_htbl);
+	else
+		dr_ste_always_miss_addr(ste_ctx, tmp_hw_ste,
+					connect_info->miss_icm_addr);
+	memcpy(formatted_ste, tmp_hw_ste, DR_STE_SIZE_REDUCED);
+}
+
+int mlx5dr_ste_htbl_init_and_postsend(struct mlx5dr_domain *dmn,
+				      struct mlx5dr_domain_rx_tx *nic_dmn,
+				      struct mlx5dr_ste_htbl *htbl,
+				      struct mlx5dr_htbl_connect_info *connect_info,
+				      bool update_hw_ste)
+{
+	u8 formatted_ste[DR_STE_SIZE] = {};
+
+	mlx5dr_ste_set_formatted_ste(dmn->ste_ctx,
+				     dmn->info.caps.gvmi,
+				     nic_dmn->type,
+				     htbl,
+				     formatted_ste,
+				     connect_info);
+
+	return mlx5dr_send_postsend_formatted_htbl(dmn, htbl, formatted_ste, update_hw_ste);
+}
+
+int mlx5dr_ste_create_next_htbl(struct mlx5dr_matcher *matcher,
+				struct mlx5dr_matcher_rx_tx *nic_matcher,
+				struct mlx5dr_ste *ste,
+				u8 *cur_hw_ste,
+				enum mlx5dr_icm_chunk_size log_table_size)
+{
+	struct mlx5dr_domain_rx_tx *nic_dmn = nic_matcher->nic_tbl->nic_dmn;
+	struct mlx5dr_domain *dmn = matcher->tbl->dmn;
+	struct mlx5dr_ste_ctx *ste_ctx = dmn->ste_ctx;
+	struct mlx5dr_htbl_connect_info info;
+	struct mlx5dr_ste_htbl *next_htbl;
+
+	if (!mlx5dr_ste_is_last_in_rule(nic_matcher, ste->ste_chain_location)) {
+		u16 next_lu_type;
+		u16 byte_mask;
+
+		next_lu_type = ste_ctx->get_next_lu_type(cur_hw_ste);
+		byte_mask = ste_ctx->get_byte_mask(cur_hw_ste);
+
+		next_htbl = mlx5dr_ste_htbl_alloc(dmn->ste_icm_pool,
+						  log_table_size,
+						  next_lu_type,
+						  byte_mask);
+		if (!next_htbl) {
+			mlx5dr_dbg(dmn, "Failed allocating table\n");
+			return -ENOMEM;
+		}
+
+		/* Write new table to HW */
+		info.type = CONNECT_MISS;
+		info.miss_icm_addr =
+			mlx5dr_icm_pool_get_chunk_icm_addr(nic_matcher->e_anchor->chunk);
+		if (mlx5dr_ste_htbl_init_and_postsend(dmn, nic_dmn, next_htbl,
+						      &info, false)) {
+			mlx5dr_info(dmn, "Failed writing table to HW\n");
+			goto free_table;
+		}
+
+		mlx5dr_ste_set_hit_addr_by_next_htbl(ste_ctx,
+						     cur_hw_ste, next_htbl);
+		ste->next_htbl = next_htbl;
+		next_htbl->pointing_ste = ste;
+	}
+
+	return 0;
+
+free_table:
+	mlx5dr_ste_htbl_free(next_htbl);
+	return -ENOENT;
+}
+
+struct mlx5dr_ste_htbl *mlx5dr_ste_htbl_alloc(struct mlx5dr_icm_pool *pool,
+					      enum mlx5dr_icm_chunk_size chunk_size,
+					      u16 lu_type, u16 byte_mask)
+{
+	struct mlx5dr_icm_chunk *chunk;
+	struct mlx5dr_ste_htbl *htbl;
+	u32 num_entries;
+	int i;
+
+	htbl = kzalloc(sizeof(*htbl), GFP_KERNEL);
+	if (!htbl)
+		return NULL;
+
+	chunk = mlx5dr_icm_alloc_chunk(pool, chunk_size);
+	if (!chunk)
+		goto out_free_htbl;
+
+	htbl->chunk = chunk;
+	htbl->lu_type = lu_type;
+	htbl->byte_mask = byte_mask;
+	htbl->refcount = 0;
+	num_entries = mlx5dr_icm_pool_get_chunk_num_of_entries(chunk);
+
+	for (i = 0; i < num_entries; i++) {
+		struct mlx5dr_ste *ste = &chunk->ste_arr[i];
+
+		ste->htbl = htbl;
+		ste->refcount = 0;
+		INIT_LIST_HEAD(&ste->miss_list_node);
+		INIT_LIST_HEAD(&chunk->miss_list[i]);
+	}
+
+	return htbl;
+
+out_free_htbl:
+	kfree(htbl);
+	return NULL;
+}
+
+int mlx5dr_ste_htbl_free(struct mlx5dr_ste_htbl *htbl)
+{
+	if (htbl->refcount)
+		return -EBUSY;
+
+	mlx5dr_icm_free_chunk(htbl->chunk);
+	kfree(htbl);
+	return 0;
+}
+
+void mlx5dr_ste_set_actions_tx(struct mlx5dr_ste_ctx *ste_ctx,
+			       struct mlx5dr_domain *dmn,
+			       u8 *action_type_set,
+			       u8 *hw_ste_arr,
+			       struct mlx5dr_ste_actions_attr *attr,
+			       u32 *added_stes)
+{
+	ste_ctx->set_actions_tx(dmn, action_type_set, ste_ctx->actions_caps,
+				hw_ste_arr, attr, added_stes);
+}
+
+void mlx5dr_ste_set_actions_rx(struct mlx5dr_ste_ctx *ste_ctx,
+			       struct mlx5dr_domain *dmn,
+			       u8 *action_type_set,
+			       u8 *hw_ste_arr,
+			       struct mlx5dr_ste_actions_attr *attr,
+			       u32 *added_stes)
+{
+	ste_ctx->set_actions_rx(dmn, action_type_set, ste_ctx->actions_caps,
+				hw_ste_arr, attr, added_stes);
+}
+
+const struct mlx5dr_ste_action_modify_field *
+mlx5dr_ste_conv_modify_hdr_sw_field(struct mlx5dr_ste_ctx *ste_ctx, u16 sw_field)
+{
+	const struct mlx5dr_ste_action_modify_field *hw_field;
+
+	if (sw_field >= ste_ctx->modify_field_arr_sz)
+		return NULL;
+
+	hw_field = &ste_ctx->modify_field_arr[sw_field];
+	if (!hw_field->end && !hw_field->start)
+		return NULL;
+
+	return hw_field;
+}
+
+void mlx5dr_ste_set_action_set(struct mlx5dr_ste_ctx *ste_ctx,
+			       __be64 *hw_action,
+			       u8 hw_field,
+			       u8 shifter,
+			       u8 length,
+			       u32 data)
+{
+	ste_ctx->set_action_set((u8 *)hw_action,
+				hw_field, shifter, length, data);
+}
+
+void mlx5dr_ste_set_action_add(struct mlx5dr_ste_ctx *ste_ctx,
+			       __be64 *hw_action,
+			       u8 hw_field,
+			       u8 shifter,
+			       u8 length,
+			       u32 data)
+{
+	ste_ctx->set_action_add((u8 *)hw_action,
+				hw_field, shifter, length, data);
+}
+
+void mlx5dr_ste_set_action_copy(struct mlx5dr_ste_ctx *ste_ctx,
+				__be64 *hw_action,
+				u8 dst_hw_field,
+				u8 dst_shifter,
+				u8 dst_len,
+				u8 src_hw_field,
+				u8 src_shifter)
+{
+	ste_ctx->set_action_copy((u8 *)hw_action,
+				 dst_hw_field, dst_shifter, dst_len,
+				 src_hw_field, src_shifter);
+}
+
+int mlx5dr_ste_set_action_decap_l3_list(struct mlx5dr_ste_ctx *ste_ctx,
+					void *data, u32 data_sz,
+					u8 *hw_action, u32 hw_action_sz,
+					u16 *used_hw_action_num)
+{
+	/* Only Ethernet frame is supported, with VLAN (18) or without (14) */
+	if (data_sz != HDR_LEN_L2 && data_sz != HDR_LEN_L2_W_VLAN)
+		return -EINVAL;
+
+	return ste_ctx->set_action_decap_l3_list(data, data_sz,
+						 hw_action, hw_action_sz,
+						 used_hw_action_num);
+}
+
+static int dr_ste_build_pre_check_spec(struct mlx5dr_domain *dmn,
+				       struct mlx5dr_match_spec *spec)
+{
+	if (spec->ip_version) {
+		if (spec->ip_version != 0xf) {
+			mlx5dr_err(dmn,
+				   "Partial ip_version mask with src/dst IP is not supported\n");
+			return -EINVAL;
+		}
+	} else if (spec->ethertype != 0xffff &&
+		   (DR_MASK_IS_SRC_IP_SET(spec) || DR_MASK_IS_DST_IP_SET(spec))) {
+		mlx5dr_err(dmn,
+			   "Partial/no ethertype mask with src/dst IP is not supported\n");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+int mlx5dr_ste_build_pre_check(struct mlx5dr_domain *dmn,
+			       u8 match_criteria,
+			       struct mlx5dr_match_param *mask,
+			       struct mlx5dr_match_param *value)
+{
+	if (value)
+		return 0;
+
+	if (match_criteria & DR_MATCHER_CRITERIA_MISC) {
+		if (mask->misc.source_port && mask->misc.source_port != 0xffff) {
+			mlx5dr_err(dmn,
+				   "Partial mask source_port is not supported\n");
+			return -EINVAL;
+		}
+		if (mask->misc.source_eswitch_owner_vhca_id &&
+		    mask->misc.source_eswitch_owner_vhca_id != 0xffff) {
+			mlx5dr_err(dmn,
+				   "Partial mask source_eswitch_owner_vhca_id is not supported\n");
+			return -EINVAL;
+		}
+	}
+
+	if ((match_criteria & DR_MATCHER_CRITERIA_OUTER) &&
+	    dr_ste_build_pre_check_spec(dmn, &mask->outer))
+		return -EINVAL;
+
+	if ((match_criteria & DR_MATCHER_CRITERIA_INNER) &&
+	    dr_ste_build_pre_check_spec(dmn, &mask->inner))
+		return -EINVAL;
+
+	return 0;
+}
+
+int mlx5dr_ste_build_ste_arr(struct mlx5dr_matcher *matcher,
+			     struct mlx5dr_matcher_rx_tx *nic_matcher,
+			     struct mlx5dr_match_param *value,
+			     u8 *ste_arr)
+{
+	struct mlx5dr_domain_rx_tx *nic_dmn = nic_matcher->nic_tbl->nic_dmn;
+	bool is_rx = nic_dmn->type == DR_DOMAIN_NIC_TYPE_RX;
+	struct mlx5dr_domain *dmn = matcher->tbl->dmn;
+	struct mlx5dr_ste_ctx *ste_ctx = dmn->ste_ctx;
+	struct mlx5dr_ste_build *sb;
+	int ret, i;
+
+	ret = mlx5dr_ste_build_pre_check(dmn, matcher->match_criteria,
+					 &matcher->mask, value);
+	if (ret)
+		return ret;
+
+	sb = nic_matcher->ste_builder;
+	for (i = 0; i < nic_matcher->num_of_builders; i++) {
+		ste_ctx->ste_init(ste_arr,
+				  sb->lu_type,
+				  is_rx,
+				  dmn->info.caps.gvmi);
+
+		mlx5dr_ste_set_bit_mask(ste_arr, sb->bit_mask);
+
+		ret = sb->ste_build_tag_func(value, sb, dr_ste_get_tag(ste_arr));
+		if (ret)
+			return ret;
+
+		/* Connect the STEs */
+		if (i < (nic_matcher->num_of_builders - 1)) {
+			/* Need the next builder for these fields,
+			 * not relevant for the last ste in the chain.
+			 */
+			sb++;
+			ste_ctx->set_next_lu_type(ste_arr, sb->lu_type);
+			ste_ctx->set_byte_mask(ste_arr, sb->byte_mask);
+		}
+		ste_arr += DR_STE_SIZE;
+	}
+	return 0;
+}
+
+#define IFC_GET_CLR(typ, p, fld, clear) ({ \
+	void *__p = (p); \
+	u32 __t = MLX5_GET(typ, __p, fld); \
+	if (clear) \
+		MLX5_SET(typ, __p, fld, 0); \
+	__t; \
+})
+
+#define memcpy_and_clear(to, from, len, clear) ({ \
+	void *__to = (to), *__from = (from); \
+	size_t __len = (len); \
+	memcpy(__to, __from, __len); \
+	if (clear) \
+		memset(__from, 0, __len); \
+})
+
+static void dr_ste_copy_mask_misc(char *mask, struct mlx5dr_match_misc *spec, bool clr)
+{
+	spec->gre_c_present = IFC_GET_CLR(fte_match_set_misc, mask, gre_c_present, clr);
+	spec->gre_k_present = IFC_GET_CLR(fte_match_set_misc, mask, gre_k_present, clr);
+	spec->gre_s_present = IFC_GET_CLR(fte_match_set_misc, mask, gre_s_present, clr);
+	spec->source_vhca_port = IFC_GET_CLR(fte_match_set_misc, mask, source_vhca_port, clr);
+	spec->source_sqn = IFC_GET_CLR(fte_match_set_misc, mask, source_sqn, clr);
+
+	spec->source_port = IFC_GET_CLR(fte_match_set_misc, mask, source_port, clr);
+	spec->source_eswitch_owner_vhca_id =
+		IFC_GET_CLR(fte_match_set_misc, mask, source_eswitch_owner_vhca_id, clr);
+
+	spec->outer_second_prio = IFC_GET_CLR(fte_match_set_misc, mask, outer_second_prio, clr);
+	spec->outer_second_cfi = IFC_GET_CLR(fte_match_set_misc, mask, outer_second_cfi, clr);
+	spec->outer_second_vid = IFC_GET_CLR(fte_match_set_misc, mask, outer_second_vid, clr);
+	spec->inner_second_prio = IFC_GET_CLR(fte_match_set_misc, mask, inner_second_prio, clr);
+	spec->inner_second_cfi = IFC_GET_CLR(fte_match_set_misc, mask, inner_second_cfi, clr);
+	spec->inner_second_vid = IFC_GET_CLR(fte_match_set_misc, mask, inner_second_vid, clr);
+
+	spec->outer_second_cvlan_tag =
+		IFC_GET_CLR(fte_match_set_misc, mask, outer_second_cvlan_tag, clr);
+	spec->inner_second_cvlan_tag =
+		IFC_GET_CLR(fte_match_set_misc, mask, inner_second_cvlan_tag, clr);
+	spec->outer_second_svlan_tag =
+		IFC_GET_CLR(fte_match_set_misc, mask, outer_second_svlan_tag, clr);
+	spec->inner_second_svlan_tag =
+		IFC_GET_CLR(fte_match_set_misc, mask, inner_second_svlan_tag, clr);
+	spec->gre_protocol = IFC_GET_CLR(fte_match_set_misc, mask, gre_protocol, clr);
+
+	spec->gre_key_h = IFC_GET_CLR(fte_match_set_misc, mask, gre_key.nvgre.hi, clr);
+	spec->gre_key_l = IFC_GET_CLR(fte_match_set_misc, mask, gre_key.nvgre.lo, clr);
+
+	spec->vxlan_vni = IFC_GET_CLR(fte_match_set_misc, mask, vxlan_vni, clr);
+
+	spec->geneve_vni = IFC_GET_CLR(fte_match_set_misc, mask, geneve_vni, clr);
+	spec->geneve_tlv_option_0_exist =
+		IFC_GET_CLR(fte_match_set_misc, mask, geneve_tlv_option_0_exist, clr);
+	spec->geneve_oam = IFC_GET_CLR(fte_match_set_misc, mask, geneve_oam, clr);
+
+	spec->outer_ipv6_flow_label =
+		IFC_GET_CLR(fte_match_set_misc, mask, outer_ipv6_flow_label, clr);
+
+	spec->inner_ipv6_flow_label =
+		IFC_GET_CLR(fte_match_set_misc, mask, inner_ipv6_flow_label, clr);
+
+	spec->geneve_opt_len = IFC_GET_CLR(fte_match_set_misc, mask, geneve_opt_len, clr);
+	spec->geneve_protocol_type =
+		IFC_GET_CLR(fte_match_set_misc, mask, geneve_protocol_type, clr);
+
+	spec->bth_dst_qp = IFC_GET_CLR(fte_match_set_misc, mask, bth_dst_qp, clr);
+}
+
+static void dr_ste_copy_mask_spec(char *mask, struct mlx5dr_match_spec *spec, bool clr)
+{
+	__be32 raw_ip[4];
+
+	spec->smac_47_16 = IFC_GET_CLR(fte_match_set_lyr_2_4, mask, smac_47_16, clr);
+
+	spec->smac_15_0 = IFC_GET_CLR(fte_match_set_lyr_2_4, mask, smac_15_0, clr);
+	spec->ethertype = IFC_GET_CLR(fte_match_set_lyr_2_4, mask, ethertype, clr);
+
+	spec->dmac_47_16 = IFC_GET_CLR(fte_match_set_lyr_2_4, mask, dmac_47_16, clr);
+
+	spec->dmac_15_0 = IFC_GET_CLR(fte_match_set_lyr_2_4, mask, dmac_15_0, clr);
+	spec->first_prio = IFC_GET_CLR(fte_match_set_lyr_2_4, mask, first_prio, clr);
+	spec->first_cfi = IFC_GET_CLR(fte_match_set_lyr_2_4, mask, first_cfi, clr);
+	spec->first_vid = IFC_GET_CLR(fte_match_set_lyr_2_4, mask, first_vid, clr);
+
+	spec->ip_protocol = IFC_GET_CLR(fte_match_set_lyr_2_4, mask, ip_protocol, clr);
+	spec->ip_dscp = IFC_GET_CLR(fte_match_set_lyr_2_4, mask, ip_dscp, clr);
+	spec->ip_ecn = IFC_GET_CLR(fte_match_set_lyr_2_4, mask, ip_ecn, clr);
+	spec->cvlan_tag = IFC_GET_CLR(fte_match_set_lyr_2_4, mask, cvlan_tag, clr);
+	spec->svlan_tag = IFC_GET_CLR(fte_match_set_lyr_2_4, mask, svlan_tag, clr);
+	spec->frag = IFC_GET_CLR(fte_match_set_lyr_2_4, mask, frag, clr);
+	spec->ip_version = IFC_GET_CLR(fte_match_set_lyr_2_4, mask, ip_version, clr);
+	spec->tcp_flags = IFC_GET_CLR(fte_match_set_lyr_2_4, mask, tcp_flags, clr);
+	spec->tcp_sport = IFC_GET_CLR(fte_match_set_lyr_2_4, mask, tcp_sport, clr);
+	spec->tcp_dport = IFC_GET_CLR(fte_match_set_lyr_2_4, mask, tcp_dport, clr);
+
+	spec->ipv4_ihl = IFC_GET_CLR(fte_match_set_lyr_2_4, mask, ipv4_ihl, clr);
+	spec->ttl_hoplimit = IFC_GET_CLR(fte_match_set_lyr_2_4, mask, ttl_hoplimit, clr);
+
+	spec->udp_sport = IFC_GET_CLR(fte_match_set_lyr_2_4, mask, udp_sport, clr);
+	spec->udp_dport = IFC_GET_CLR(fte_match_set_lyr_2_4, mask, udp_dport, clr);
+
+	memcpy_and_clear(raw_ip, MLX5_ADDR_OF(fte_match_set_lyr_2_4, mask,
+					      src_ipv4_src_ipv6.ipv6_layout.ipv6),
+			 sizeof(raw_ip), clr);
+
+	spec->src_ip_127_96 = be32_to_cpu(raw_ip[0]);
+	spec->src_ip_95_64 = be32_to_cpu(raw_ip[1]);
+	spec->src_ip_63_32 = be32_to_cpu(raw_ip[2]);
+	spec->src_ip_31_0 = be32_to_cpu(raw_ip[3]);
+
+	memcpy_and_clear(raw_ip, MLX5_ADDR_OF(fte_match_set_lyr_2_4, mask,
+					      dst_ipv4_dst_ipv6.ipv6_layout.ipv6),
+			 sizeof(raw_ip), clr);
+
+	spec->dst_ip_127_96 = be32_to_cpu(raw_ip[0]);
+	spec->dst_ip_95_64 = be32_to_cpu(raw_ip[1]);
+	spec->dst_ip_63_32 = be32_to_cpu(raw_ip[2]);
+	spec->dst_ip_31_0 = be32_to_cpu(raw_ip[3]);
+}
+
+static void dr_ste_copy_mask_misc2(char *mask, struct mlx5dr_match_misc2 *spec, bool clr)
+{
+	spec->outer_first_mpls_label =
+		IFC_GET_CLR(fte_match_set_misc2, mask, outer_first_mpls.mpls_label, clr);
+	spec->outer_first_mpls_exp =
+		IFC_GET_CLR(fte_match_set_misc2, mask, outer_first_mpls.mpls_exp, clr);
+	spec->outer_first_mpls_s_bos =
+		IFC_GET_CLR(fte_match_set_misc2, mask, outer_first_mpls.mpls_s_bos, clr);
+	spec->outer_first_mpls_ttl =
+		IFC_GET_CLR(fte_match_set_misc2, mask, outer_first_mpls.mpls_ttl, clr);
+	spec->inner_first_mpls_label =
+		IFC_GET_CLR(fte_match_set_misc2, mask, inner_first_mpls.mpls_label, clr);
+	spec->inner_first_mpls_exp =
+		IFC_GET_CLR(fte_match_set_misc2, mask, inner_first_mpls.mpls_exp, clr);
+	spec->inner_first_mpls_s_bos =
+		IFC_GET_CLR(fte_match_set_misc2, mask, inner_first_mpls.mpls_s_bos, clr);
+	spec->inner_first_mpls_ttl =
+		IFC_GET_CLR(fte_match_set_misc2, mask, inner_first_mpls.mpls_ttl, clr);
+	spec->outer_first_mpls_over_gre_label =
+		IFC_GET_CLR(fte_match_set_misc2, mask, outer_first_mpls_over_gre.mpls_label, clr);
+	spec->outer_first_mpls_over_gre_exp =
+		IFC_GET_CLR(fte_match_set_misc2, mask, outer_first_mpls_over_gre.mpls_exp, clr);
+	spec->outer_first_mpls_over_gre_s_bos =
+		IFC_GET_CLR(fte_match_set_misc2, mask, outer_first_mpls_over_gre.mpls_s_bos, clr);
+	spec->outer_first_mpls_over_gre_ttl =
+		IFC_GET_CLR(fte_match_set_misc2, mask, outer_first_mpls_over_gre.mpls_ttl, clr);
+	spec->outer_first_mpls_over_udp_label =
+		IFC_GET_CLR(fte_match_set_misc2, mask, outer_first_mpls_over_udp.mpls_label, clr);
+	spec->outer_first_mpls_over_udp_exp =
+		IFC_GET_CLR(fte_match_set_misc2, mask, outer_first_mpls_over_udp.mpls_exp, clr);
+	spec->outer_first_mpls_over_udp_s_bos =
+		IFC_GET_CLR(fte_match_set_misc2, mask, outer_first_mpls_over_udp.mpls_s_bos, clr);
+	spec->outer_first_mpls_over_udp_ttl =
+		IFC_GET_CLR(fte_match_set_misc2, mask, outer_first_mpls_over_udp.mpls_ttl, clr);
+	spec->metadata_reg_c_7 = IFC_GET_CLR(fte_match_set_misc2, mask, metadata_reg_c_7, clr);
+	spec->metadata_reg_c_6 = IFC_GET_CLR(fte_match_set_misc2, mask, metadata_reg_c_6, clr);
+	spec->metadata_reg_c_5 = IFC_GET_CLR(fte_match_set_misc2, mask, metadata_reg_c_5, clr);
+	spec->metadata_reg_c_4 = IFC_GET_CLR(fte_match_set_misc2, mask, metadata_reg_c_4, clr);
+	spec->metadata_reg_c_3 = IFC_GET_CLR(fte_match_set_misc2, mask, metadata_reg_c_3, clr);
+	spec->metadata_reg_c_2 = IFC_GET_CLR(fte_match_set_misc2, mask, metadata_reg_c_2, clr);
+	spec->metadata_reg_c_1 = IFC_GET_CLR(fte_match_set_misc2, mask, metadata_reg_c_1, clr);
+	spec->metadata_reg_c_0 = IFC_GET_CLR(fte_match_set_misc2, mask, metadata_reg_c_0, clr);
+	spec->metadata_reg_a = IFC_GET_CLR(fte_match_set_misc2, mask, metadata_reg_a, clr);
+}
+
+static void dr_ste_copy_mask_misc3(char *mask, struct mlx5dr_match_misc3 *spec, bool clr)
+{
+	spec->inner_tcp_seq_num = IFC_GET_CLR(fte_match_set_misc3, mask, inner_tcp_seq_num, clr);
+	spec->outer_tcp_seq_num = IFC_GET_CLR(fte_match_set_misc3, mask, outer_tcp_seq_num, clr);
+	spec->inner_tcp_ack_num = IFC_GET_CLR(fte_match_set_misc3, mask, inner_tcp_ack_num, clr);
+	spec->outer_tcp_ack_num = IFC_GET_CLR(fte_match_set_misc3, mask, outer_tcp_ack_num, clr);
+	spec->outer_vxlan_gpe_vni =
+		IFC_GET_CLR(fte_match_set_misc3, mask, outer_vxlan_gpe_vni, clr);
+	spec->outer_vxlan_gpe_next_protocol =
+		IFC_GET_CLR(fte_match_set_misc3, mask, outer_vxlan_gpe_next_protocol, clr);
+	spec->outer_vxlan_gpe_flags =
+		IFC_GET_CLR(fte_match_set_misc3, mask, outer_vxlan_gpe_flags, clr);
+	spec->icmpv4_header_data = IFC_GET_CLR(fte_match_set_misc3, mask, icmp_header_data, clr);
+	spec->icmpv6_header_data =
+		IFC_GET_CLR(fte_match_set_misc3, mask, icmpv6_header_data, clr);
+	spec->icmpv4_type = IFC_GET_CLR(fte_match_set_misc3, mask, icmp_type, clr);
+	spec->icmpv4_code = IFC_GET_CLR(fte_match_set_misc3, mask, icmp_code, clr);
+	spec->icmpv6_type = IFC_GET_CLR(fte_match_set_misc3, mask, icmpv6_type, clr);
+	spec->icmpv6_code = IFC_GET_CLR(fte_match_set_misc3, mask, icmpv6_code, clr);
+	spec->geneve_tlv_option_0_data =
+		IFC_GET_CLR(fte_match_set_misc3, mask, geneve_tlv_option_0_data, clr);
+	spec->gtpu_teid = IFC_GET_CLR(fte_match_set_misc3, mask, gtpu_teid, clr);
+	spec->gtpu_msg_flags = IFC_GET_CLR(fte_match_set_misc3, mask, gtpu_msg_flags, clr);
+	spec->gtpu_msg_type = IFC_GET_CLR(fte_match_set_misc3, mask, gtpu_msg_type, clr);
+	spec->gtpu_dw_0 = IFC_GET_CLR(fte_match_set_misc3, mask, gtpu_dw_0, clr);
+	spec->gtpu_dw_2 = IFC_GET_CLR(fte_match_set_misc3, mask, gtpu_dw_2, clr);
+	spec->gtpu_first_ext_dw_0 =
+		IFC_GET_CLR(fte_match_set_misc3, mask, gtpu_first_ext_dw_0, clr);
+}
+
+static void dr_ste_copy_mask_misc4(char *mask, struct mlx5dr_match_misc4 *spec, bool clr)
+{
+	spec->prog_sample_field_id_0 =
+		IFC_GET_CLR(fte_match_set_misc4, mask, prog_sample_field_id_0, clr);
+	spec->prog_sample_field_value_0 =
+		IFC_GET_CLR(fte_match_set_misc4, mask, prog_sample_field_value_0, clr);
+	spec->prog_sample_field_id_1 =
+		IFC_GET_CLR(fte_match_set_misc4, mask, prog_sample_field_id_1, clr);
+	spec->prog_sample_field_value_1 =
+		IFC_GET_CLR(fte_match_set_misc4, mask, prog_sample_field_value_1, clr);
+	spec->prog_sample_field_id_2 =
+		IFC_GET_CLR(fte_match_set_misc4, mask, prog_sample_field_id_2, clr);
+	spec->prog_sample_field_value_2 =
+		IFC_GET_CLR(fte_match_set_misc4, mask, prog_sample_field_value_2, clr);
+	spec->prog_sample_field_id_3 =
+		IFC_GET_CLR(fte_match_set_misc4, mask, prog_sample_field_id_3, clr);
+	spec->prog_sample_field_value_3 =
+		IFC_GET_CLR(fte_match_set_misc4, mask, prog_sample_field_value_3, clr);
+}
+
+static void dr_ste_copy_mask_misc5(char *mask, struct mlx5dr_match_misc5 *spec, bool clr)
+{
+	spec->macsec_tag_0 =
+		IFC_GET_CLR(fte_match_set_misc5, mask, macsec_tag_0, clr);
+	spec->macsec_tag_1 =
+		IFC_GET_CLR(fte_match_set_misc5, mask, macsec_tag_1, clr);
+	spec->macsec_tag_2 =
+		IFC_GET_CLR(fte_match_set_misc5, mask, macsec_tag_2, clr);
+	spec->macsec_tag_3 =
+		IFC_GET_CLR(fte_match_set_misc5, mask, macsec_tag_3, clr);
+	spec->tunnel_header_0 =
+		IFC_GET_CLR(fte_match_set_misc5, mask, tunnel_header_0, clr);
+	spec->tunnel_header_1 =
+		IFC_GET_CLR(fte_match_set_misc5, mask, tunnel_header_1, clr);
+	spec->tunnel_header_2 =
+		IFC_GET_CLR(fte_match_set_misc5, mask, tunnel_header_2, clr);
+	spec->tunnel_header_3 =
+		IFC_GET_CLR(fte_match_set_misc5, mask, tunnel_header_3, clr);
+}
+
+void mlx5dr_ste_copy_param(u8 match_criteria,
+			   struct mlx5dr_match_param *set_param,
+			   struct mlx5dr_match_parameters *mask,
+			   bool clr)
+{
+	u8 tail_param[MLX5_ST_SZ_BYTES(fte_match_set_lyr_2_4)] = {};
+	u8 *data = (u8 *)mask->match_buf;
+	size_t param_location;
+	void *buff;
+
+	if (match_criteria & DR_MATCHER_CRITERIA_OUTER) {
+		if (mask->match_sz < sizeof(struct mlx5dr_match_spec)) {
+			memcpy(tail_param, data, mask->match_sz);
+			buff = tail_param;
+		} else {
+			buff = mask->match_buf;
+		}
+		dr_ste_copy_mask_spec(buff, &set_param->outer, clr);
+	}
+	param_location = sizeof(struct mlx5dr_match_spec);
+
+	if (match_criteria & DR_MATCHER_CRITERIA_MISC) {
+		if (mask->match_sz < param_location +
+		    sizeof(struct mlx5dr_match_misc)) {
+			memcpy(tail_param, data + param_location,
+			       mask->match_sz - param_location);
+			buff = tail_param;
+		} else {
+			buff = data + param_location;
+		}
+		dr_ste_copy_mask_misc(buff, &set_param->misc, clr);
+	}
+	param_location += sizeof(struct mlx5dr_match_misc);
+
+	if (match_criteria & DR_MATCHER_CRITERIA_INNER) {
+		if (mask->match_sz < param_location +
+		    sizeof(struct mlx5dr_match_spec)) {
+			memcpy(tail_param, data + param_location,
+			       mask->match_sz - param_location);
+			buff = tail_param;
+		} else {
+			buff = data + param_location;
+		}
+		dr_ste_copy_mask_spec(buff, &set_param->inner, clr);
+	}
+	param_location += sizeof(struct mlx5dr_match_spec);
+
+	if (match_criteria & DR_MATCHER_CRITERIA_MISC2) {
+		if (mask->match_sz < param_location +
+		    sizeof(struct mlx5dr_match_misc2)) {
+			memcpy(tail_param, data + param_location,
+			       mask->match_sz - param_location);
+			buff = tail_param;
+		} else {
+			buff = data + param_location;
+		}
+		dr_ste_copy_mask_misc2(buff, &set_param->misc2, clr);
+	}
+
+	param_location += sizeof(struct mlx5dr_match_misc2);
+
+	if (match_criteria & DR_MATCHER_CRITERIA_MISC3) {
+		if (mask->match_sz < param_location +
+		    sizeof(struct mlx5dr_match_misc3)) {
+			memcpy(tail_param, data + param_location,
+			       mask->match_sz - param_location);
+			buff = tail_param;
+		} else {
+			buff = data + param_location;
+		}
+		dr_ste_copy_mask_misc3(buff, &set_param->misc3, clr);
+	}
+
+	param_location += sizeof(struct mlx5dr_match_misc3);
+
+	if (match_criteria & DR_MATCHER_CRITERIA_MISC4) {
+		if (mask->match_sz < param_location +
+		    sizeof(struct mlx5dr_match_misc4)) {
+			memcpy(tail_param, data + param_location,
+			       mask->match_sz - param_location);
+			buff = tail_param;
+		} else {
+			buff = data + param_location;
+		}
+		dr_ste_copy_mask_misc4(buff, &set_param->misc4, clr);
+	}
+
+	param_location += sizeof(struct mlx5dr_match_misc4);
+
+	if (match_criteria & DR_MATCHER_CRITERIA_MISC5) {
+		if (mask->match_sz < param_location +
+		    sizeof(struct mlx5dr_match_misc5)) {
+			memcpy(tail_param, data + param_location,
+			       mask->match_sz - param_location);
+			buff = tail_param;
+		} else {
+			buff = data + param_location;
+		}
+		dr_ste_copy_mask_misc5(buff, &set_param->misc5, clr);
+	}
+}
+
+void mlx5dr_ste_build_eth_l2_src_dst(struct mlx5dr_ste_ctx *ste_ctx,
+				     struct mlx5dr_ste_build *sb,
+				     struct mlx5dr_match_param *mask,
+				     bool inner, bool rx)
+{
+	sb->rx = rx;
+	sb->inner = inner;
+	ste_ctx->build_eth_l2_src_dst_init(sb, mask);
+}
+
+void mlx5dr_ste_build_eth_l3_ipv6_dst(struct mlx5dr_ste_ctx *ste_ctx,
+				      struct mlx5dr_ste_build *sb,
+				      struct mlx5dr_match_param *mask,
+				      bool inner, bool rx)
+{
+	sb->rx = rx;
+	sb->inner = inner;
+	ste_ctx->build_eth_l3_ipv6_dst_init(sb, mask);
+}
+
+void mlx5dr_ste_build_eth_l3_ipv6_src(struct mlx5dr_ste_ctx *ste_ctx,
+				      struct mlx5dr_ste_build *sb,
+				      struct mlx5dr_match_param *mask,
+				      bool inner, bool rx)
+{
+	sb->rx = rx;
+	sb->inner = inner;
+	ste_ctx->build_eth_l3_ipv6_src_init(sb, mask);
+}
+
+void mlx5dr_ste_build_eth_l3_ipv4_5_tuple(struct mlx5dr_ste_ctx *ste_ctx,
+					  struct mlx5dr_ste_build *sb,
+					  struct mlx5dr_match_param *mask,
+					  bool inner, bool rx)
+{
+	sb->rx = rx;
+	sb->inner = inner;
+	ste_ctx->build_eth_l3_ipv4_5_tuple_init(sb, mask);
+}
+
+void mlx5dr_ste_build_eth_l2_src(struct mlx5dr_ste_ctx *ste_ctx,
+				 struct mlx5dr_ste_build *sb,
+				 struct mlx5dr_match_param *mask,
+				 bool inner, bool rx)
+{
+	sb->rx = rx;
+	sb->inner = inner;
+	ste_ctx->build_eth_l2_src_init(sb, mask);
+}
+
+void mlx5dr_ste_build_eth_l2_dst(struct mlx5dr_ste_ctx *ste_ctx,
+				 struct mlx5dr_ste_build *sb,
+				 struct mlx5dr_match_param *mask,
+				 bool inner, bool rx)
+{
+	sb->rx = rx;
+	sb->inner = inner;
+	ste_ctx->build_eth_l2_dst_init(sb, mask);
+}
+
+void mlx5dr_ste_build_eth_l2_tnl(struct mlx5dr_ste_ctx *ste_ctx,
+				 struct mlx5dr_ste_build *sb,
+				 struct mlx5dr_match_param *mask, bool inner, bool rx)
+{
+	sb->rx = rx;
+	sb->inner = inner;
+	ste_ctx->build_eth_l2_tnl_init(sb, mask);
+}
+
+void mlx5dr_ste_build_eth_l3_ipv4_misc(struct mlx5dr_ste_ctx *ste_ctx,
+				       struct mlx5dr_ste_build *sb,
+				       struct mlx5dr_match_param *mask,
+				       bool inner, bool rx)
+{
+	sb->rx = rx;
+	sb->inner = inner;
+	ste_ctx->build_eth_l3_ipv4_misc_init(sb, mask);
+}
+
+void mlx5dr_ste_build_eth_ipv6_l3_l4(struct mlx5dr_ste_ctx *ste_ctx,
+				     struct mlx5dr_ste_build *sb,
+				     struct mlx5dr_match_param *mask,
+				     bool inner, bool rx)
+{
+	sb->rx = rx;
+	sb->inner = inner;
+	ste_ctx->build_eth_ipv6_l3_l4_init(sb, mask);
+}
+
+static int dr_ste_build_empty_always_hit_tag(struct mlx5dr_match_param *value,
+					     struct mlx5dr_ste_build *sb,
+					     u8 *tag)
+{
+	return 0;
+}
+
+void mlx5dr_ste_build_empty_always_hit(struct mlx5dr_ste_build *sb, bool rx)
+{
+	sb->rx = rx;
+	sb->lu_type = MLX5DR_STE_LU_TYPE_DONT_CARE;
+	sb->byte_mask = 0;
+	sb->ste_build_tag_func = &dr_ste_build_empty_always_hit_tag;
+}
+
+void mlx5dr_ste_build_mpls(struct mlx5dr_ste_ctx *ste_ctx,
+			   struct mlx5dr_ste_build *sb,
+			   struct mlx5dr_match_param *mask,
+			   bool inner, bool rx)
+{
+	sb->rx = rx;
+	sb->inner = inner;
+	ste_ctx->build_mpls_init(sb, mask);
+}
+
+void mlx5dr_ste_build_tnl_gre(struct mlx5dr_ste_ctx *ste_ctx,
+			      struct mlx5dr_ste_build *sb,
+			      struct mlx5dr_match_param *mask,
+			      bool inner, bool rx)
+{
+	sb->rx = rx;
+	sb->inner = inner;
+	ste_ctx->build_tnl_gre_init(sb, mask);
+}
+
+void mlx5dr_ste_build_tnl_mpls_over_gre(struct mlx5dr_ste_ctx *ste_ctx,
+					struct mlx5dr_ste_build *sb,
+					struct mlx5dr_match_param *mask,
+					struct mlx5dr_cmd_caps *caps,
+					bool inner, bool rx)
+{
+	sb->rx = rx;
+	sb->inner = inner;
+	sb->caps = caps;
+	return ste_ctx->build_tnl_mpls_over_gre_init(sb, mask);
+}
+
+void mlx5dr_ste_build_tnl_mpls_over_udp(struct mlx5dr_ste_ctx *ste_ctx,
+					struct mlx5dr_ste_build *sb,
+					struct mlx5dr_match_param *mask,
+					struct mlx5dr_cmd_caps *caps,
+					bool inner, bool rx)
+{
+	sb->rx = rx;
+	sb->inner = inner;
+	sb->caps = caps;
+	return ste_ctx->build_tnl_mpls_over_udp_init(sb, mask);
+}
+
+void mlx5dr_ste_build_icmp(struct mlx5dr_ste_ctx *ste_ctx,
+			   struct mlx5dr_ste_build *sb,
+			   struct mlx5dr_match_param *mask,
+			   struct mlx5dr_cmd_caps *caps,
+			   bool inner, bool rx)
+{
+	sb->rx = rx;
+	sb->inner = inner;
+	sb->caps = caps;
+	ste_ctx->build_icmp_init(sb, mask);
+}
+
+void mlx5dr_ste_build_general_purpose(struct mlx5dr_ste_ctx *ste_ctx,
+				      struct mlx5dr_ste_build *sb,
+				      struct mlx5dr_match_param *mask,
+				      bool inner, bool rx)
+{
+	sb->rx = rx;
+	sb->inner = inner;
+	ste_ctx->build_general_purpose_init(sb, mask);
+}
+
+void mlx5dr_ste_build_eth_l4_misc(struct mlx5dr_ste_ctx *ste_ctx,
+				  struct mlx5dr_ste_build *sb,
+				  struct mlx5dr_match_param *mask,
+				  bool inner, bool rx)
+{
+	sb->rx = rx;
+	sb->inner = inner;
+	ste_ctx->build_eth_l4_misc_init(sb, mask);
+}
+
+void mlx5dr_ste_build_tnl_vxlan_gpe(struct mlx5dr_ste_ctx *ste_ctx,
+				    struct mlx5dr_ste_build *sb,
+				    struct mlx5dr_match_param *mask,
+				    bool inner, bool rx)
+{
+	sb->rx = rx;
+	sb->inner = inner;
+	ste_ctx->build_tnl_vxlan_gpe_init(sb, mask);
+}
+
+void mlx5dr_ste_build_tnl_geneve(struct mlx5dr_ste_ctx *ste_ctx,
+				 struct mlx5dr_ste_build *sb,
+				 struct mlx5dr_match_param *mask,
+				 bool inner, bool rx)
+{
+	sb->rx = rx;
+	sb->inner = inner;
+	ste_ctx->build_tnl_geneve_init(sb, mask);
+}
+
+void mlx5dr_ste_build_tnl_geneve_tlv_opt(struct mlx5dr_ste_ctx *ste_ctx,
+					 struct mlx5dr_ste_build *sb,
+					 struct mlx5dr_match_param *mask,
+					 struct mlx5dr_cmd_caps *caps,
+					 bool inner, bool rx)
+{
+	sb->rx = rx;
+	sb->caps = caps;
+	sb->inner = inner;
+	ste_ctx->build_tnl_geneve_tlv_opt_init(sb, mask);
+}
+
+void mlx5dr_ste_build_tnl_geneve_tlv_opt_exist(struct mlx5dr_ste_ctx *ste_ctx,
+					       struct mlx5dr_ste_build *sb,
+					       struct mlx5dr_match_param *mask,
+					       struct mlx5dr_cmd_caps *caps,
+					       bool inner, bool rx)
+{
+	if (!ste_ctx->build_tnl_geneve_tlv_opt_exist_init)
+		return;
+
+	sb->rx = rx;
+	sb->caps = caps;
+	sb->inner = inner;
+	ste_ctx->build_tnl_geneve_tlv_opt_exist_init(sb, mask);
+}
+
+void mlx5dr_ste_build_tnl_gtpu(struct mlx5dr_ste_ctx *ste_ctx,
+			       struct mlx5dr_ste_build *sb,
+			       struct mlx5dr_match_param *mask,
+			       bool inner, bool rx)
+{
+	sb->rx = rx;
+	sb->inner = inner;
+	ste_ctx->build_tnl_gtpu_init(sb, mask);
+}
+
+void mlx5dr_ste_build_tnl_gtpu_flex_parser_0(struct mlx5dr_ste_ctx *ste_ctx,
+					     struct mlx5dr_ste_build *sb,
+					     struct mlx5dr_match_param *mask,
+					     struct mlx5dr_cmd_caps *caps,
+					     bool inner, bool rx)
+{
+	sb->rx = rx;
+	sb->caps = caps;
+	sb->inner = inner;
+	ste_ctx->build_tnl_gtpu_flex_parser_0_init(sb, mask);
+}
+
+void mlx5dr_ste_build_tnl_gtpu_flex_parser_1(struct mlx5dr_ste_ctx *ste_ctx,
+					     struct mlx5dr_ste_build *sb,
+					     struct mlx5dr_match_param *mask,
+					     struct mlx5dr_cmd_caps *caps,
+					     bool inner, bool rx)
+{
+	sb->rx = rx;
+	sb->caps = caps;
+	sb->inner = inner;
+	ste_ctx->build_tnl_gtpu_flex_parser_1_init(sb, mask);
+}
+
+void mlx5dr_ste_build_register_0(struct mlx5dr_ste_ctx *ste_ctx,
+				 struct mlx5dr_ste_build *sb,
+				 struct mlx5dr_match_param *mask,
+				 bool inner, bool rx)
+{
+	sb->rx = rx;
+	sb->inner = inner;
+	ste_ctx->build_register_0_init(sb, mask);
+}
+
+void mlx5dr_ste_build_register_1(struct mlx5dr_ste_ctx *ste_ctx,
+				 struct mlx5dr_ste_build *sb,
+				 struct mlx5dr_match_param *mask,
+				 bool inner, bool rx)
+{
+	sb->rx = rx;
+	sb->inner = inner;
+	ste_ctx->build_register_1_init(sb, mask);
+}
+
+void mlx5dr_ste_build_src_gvmi_qpn(struct mlx5dr_ste_ctx *ste_ctx,
+				   struct mlx5dr_ste_build *sb,
+				   struct mlx5dr_match_param *mask,
+				   struct mlx5dr_domain *dmn,
+				   bool inner, bool rx)
+{
+	/* Set vhca_id_valid before we reset source_eswitch_owner_vhca_id */
+	sb->vhca_id_valid = mask->misc.source_eswitch_owner_vhca_id;
+
+	sb->rx = rx;
+	sb->dmn = dmn;
+	sb->inner = inner;
+	ste_ctx->build_src_gvmi_qpn_init(sb, mask);
+}
+
+void mlx5dr_ste_build_flex_parser_0(struct mlx5dr_ste_ctx *ste_ctx,
+				    struct mlx5dr_ste_build *sb,
+				    struct mlx5dr_match_param *mask,
+				    bool inner, bool rx)
+{
+	sb->rx = rx;
+	sb->inner = inner;
+	ste_ctx->build_flex_parser_0_init(sb, mask);
+}
+
+void mlx5dr_ste_build_flex_parser_1(struct mlx5dr_ste_ctx *ste_ctx,
+				    struct mlx5dr_ste_build *sb,
+				    struct mlx5dr_match_param *mask,
+				    bool inner, bool rx)
+{
+	sb->rx = rx;
+	sb->inner = inner;
+	ste_ctx->build_flex_parser_1_init(sb, mask);
+}
+
+void mlx5dr_ste_build_tnl_header_0_1(struct mlx5dr_ste_ctx *ste_ctx,
+				     struct mlx5dr_ste_build *sb,
+				     struct mlx5dr_match_param *mask,
+				     bool inner, bool rx)
+{
+	sb->rx = rx;
+	sb->inner = inner;
+	ste_ctx->build_tnl_header_0_1_init(sb, mask);
+}
+
+struct mlx5dr_ste_ctx *mlx5dr_ste_get_ctx(u8 version)
+{
+	if (version == MLX5_STEERING_FORMAT_CONNECTX_5)
+		return mlx5dr_ste_get_ctx_v0();
+	else if (version == MLX5_STEERING_FORMAT_CONNECTX_6DX)
+		return mlx5dr_ste_get_ctx_v1();
+	else if (version == MLX5_STEERING_FORMAT_CONNECTX_7)
+		return mlx5dr_ste_get_ctx_v2();
+
+	return NULL;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_ste.h b/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_ste.h
new file mode 100644
index 000000000..17513baff
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_ste.h
@@ -0,0 +1,206 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2020 NVIDIA CORPORATION. All rights reserved. */
+
+#ifndef	_DR_STE_
+#define	_DR_STE_
+
+#include "dr_types.h"
+
+#define STE_IPV4 0x1
+#define STE_IPV6 0x2
+#define STE_TCP 0x1
+#define STE_UDP 0x2
+#define STE_SPI 0x3
+#define IP_VERSION_IPV4 0x4
+#define IP_VERSION_IPV6 0x6
+#define STE_SVLAN 0x1
+#define STE_CVLAN 0x2
+#define HDR_LEN_L2_MACS   0xC
+#define HDR_LEN_L2_VLAN   0x4
+#define HDR_LEN_L2_ETHER  0x2
+#define HDR_LEN_L2        (HDR_LEN_L2_MACS + HDR_LEN_L2_ETHER)
+#define HDR_LEN_L2_W_VLAN (HDR_LEN_L2 + HDR_LEN_L2_VLAN)
+
+/* Set to STE a specific value using DR_STE_SET */
+#define DR_STE_SET_VAL(lookup_type, tag, t_fname, spec, s_fname, value) do { \
+	if ((spec)->s_fname) { \
+		MLX5_SET(ste_##lookup_type, tag, t_fname, value); \
+		(spec)->s_fname = 0; \
+	} \
+} while (0)
+
+/* Set to STE spec->s_fname to tag->t_fname set spec->s_fname as used */
+#define DR_STE_SET_TAG(lookup_type, tag, t_fname, spec, s_fname) \
+	DR_STE_SET_VAL(lookup_type, tag, t_fname, spec, s_fname, spec->s_fname)
+
+/* Set to STE -1 to tag->t_fname and set spec->s_fname as used */
+#define DR_STE_SET_ONES(lookup_type, tag, t_fname, spec, s_fname) \
+	DR_STE_SET_VAL(lookup_type, tag, t_fname, spec, s_fname, -1)
+
+#define DR_STE_SET_TCP_FLAGS(lookup_type, tag, spec) do { \
+	MLX5_SET(ste_##lookup_type, tag, tcp_ns, !!((spec)->tcp_flags & (1 << 8))); \
+	MLX5_SET(ste_##lookup_type, tag, tcp_cwr, !!((spec)->tcp_flags & (1 << 7))); \
+	MLX5_SET(ste_##lookup_type, tag, tcp_ece, !!((spec)->tcp_flags & (1 << 6))); \
+	MLX5_SET(ste_##lookup_type, tag, tcp_urg, !!((spec)->tcp_flags & (1 << 5))); \
+	MLX5_SET(ste_##lookup_type, tag, tcp_ack, !!((spec)->tcp_flags & (1 << 4))); \
+	MLX5_SET(ste_##lookup_type, tag, tcp_psh, !!((spec)->tcp_flags & (1 << 3))); \
+	MLX5_SET(ste_##lookup_type, tag, tcp_rst, !!((spec)->tcp_flags & (1 << 2))); \
+	MLX5_SET(ste_##lookup_type, tag, tcp_syn, !!((spec)->tcp_flags & (1 << 1))); \
+	MLX5_SET(ste_##lookup_type, tag, tcp_fin, !!((spec)->tcp_flags & (1 << 0))); \
+} while (0)
+
+#define DR_STE_SET_MPLS(lookup_type, mask, in_out, tag) do { \
+	struct mlx5dr_match_misc2 *_mask = mask; \
+	u8 *_tag = tag; \
+	DR_STE_SET_TAG(lookup_type, _tag, mpls0_label, _mask, \
+		       in_out##_first_mpls_label);\
+	DR_STE_SET_TAG(lookup_type, _tag, mpls0_s_bos, _mask, \
+		       in_out##_first_mpls_s_bos); \
+	DR_STE_SET_TAG(lookup_type, _tag, mpls0_exp, _mask, \
+		       in_out##_first_mpls_exp); \
+	DR_STE_SET_TAG(lookup_type, _tag, mpls0_ttl, _mask, \
+		       in_out##_first_mpls_ttl); \
+} while (0)
+
+#define DR_STE_SET_FLEX_PARSER_FIELD(tag, fname, caps, spec) do { \
+	u8 parser_id = (caps)->flex_parser_id_##fname; \
+	u8 *parser_ptr = dr_ste_calc_flex_parser_offset(tag, parser_id); \
+	*(__be32 *)parser_ptr = cpu_to_be32((spec)->fname);\
+	(spec)->fname = 0;\
+} while (0)
+
+#define DR_STE_IS_OUTER_MPLS_OVER_GRE_SET(_misc) (\
+	(_misc)->outer_first_mpls_over_gre_label || \
+	(_misc)->outer_first_mpls_over_gre_exp || \
+	(_misc)->outer_first_mpls_over_gre_s_bos || \
+	(_misc)->outer_first_mpls_over_gre_ttl)
+
+#define DR_STE_IS_OUTER_MPLS_OVER_UDP_SET(_misc) (\
+	(_misc)->outer_first_mpls_over_udp_label || \
+	(_misc)->outer_first_mpls_over_udp_exp || \
+	(_misc)->outer_first_mpls_over_udp_s_bos || \
+	(_misc)->outer_first_mpls_over_udp_ttl)
+
+enum dr_ste_action_modify_type_l3 {
+	DR_STE_ACTION_MDFY_TYPE_L3_NONE	= 0x0,
+	DR_STE_ACTION_MDFY_TYPE_L3_IPV4	= 0x1,
+	DR_STE_ACTION_MDFY_TYPE_L3_IPV6	= 0x2,
+};
+
+enum dr_ste_action_modify_type_l4 {
+	DR_STE_ACTION_MDFY_TYPE_L4_NONE	= 0x0,
+	DR_STE_ACTION_MDFY_TYPE_L4_TCP	= 0x1,
+	DR_STE_ACTION_MDFY_TYPE_L4_UDP	= 0x2,
+};
+
+enum {
+	HDR_MPLS_OFFSET_LABEL	= 12,
+	HDR_MPLS_OFFSET_EXP	= 9,
+	HDR_MPLS_OFFSET_S_BOS	= 8,
+	HDR_MPLS_OFFSET_TTL	= 0,
+};
+
+u16 mlx5dr_ste_conv_bit_to_byte_mask(u8 *bit_mask);
+
+static inline u8 *
+dr_ste_calc_flex_parser_offset(u8 *tag, u8 parser_id)
+{
+	/* Calculate tag byte offset based on flex parser id */
+	return tag + 4 * (3 - (parser_id % 4));
+}
+
+#define DR_STE_CTX_BUILDER(fname) \
+	((*build_##fname##_init)(struct mlx5dr_ste_build *sb, \
+				 struct mlx5dr_match_param *mask))
+
+struct mlx5dr_ste_ctx {
+	/* Builders */
+	void DR_STE_CTX_BUILDER(eth_l2_src_dst);
+	void DR_STE_CTX_BUILDER(eth_l3_ipv6_src);
+	void DR_STE_CTX_BUILDER(eth_l3_ipv6_dst);
+	void DR_STE_CTX_BUILDER(eth_l3_ipv4_5_tuple);
+	void DR_STE_CTX_BUILDER(eth_l2_src);
+	void DR_STE_CTX_BUILDER(eth_l2_dst);
+	void DR_STE_CTX_BUILDER(eth_l2_tnl);
+	void DR_STE_CTX_BUILDER(eth_l3_ipv4_misc);
+	void DR_STE_CTX_BUILDER(eth_ipv6_l3_l4);
+	void DR_STE_CTX_BUILDER(mpls);
+	void DR_STE_CTX_BUILDER(tnl_gre);
+	void DR_STE_CTX_BUILDER(tnl_mpls);
+	void DR_STE_CTX_BUILDER(tnl_mpls_over_gre);
+	void DR_STE_CTX_BUILDER(tnl_mpls_over_udp);
+	void DR_STE_CTX_BUILDER(icmp);
+	void DR_STE_CTX_BUILDER(general_purpose);
+	void DR_STE_CTX_BUILDER(eth_l4_misc);
+	void DR_STE_CTX_BUILDER(tnl_vxlan_gpe);
+	void DR_STE_CTX_BUILDER(tnl_geneve);
+	void DR_STE_CTX_BUILDER(tnl_geneve_tlv_opt);
+	void DR_STE_CTX_BUILDER(tnl_geneve_tlv_opt_exist);
+	void DR_STE_CTX_BUILDER(register_0);
+	void DR_STE_CTX_BUILDER(register_1);
+	void DR_STE_CTX_BUILDER(src_gvmi_qpn);
+	void DR_STE_CTX_BUILDER(flex_parser_0);
+	void DR_STE_CTX_BUILDER(flex_parser_1);
+	void DR_STE_CTX_BUILDER(tnl_gtpu);
+	void DR_STE_CTX_BUILDER(tnl_header_0_1);
+	void DR_STE_CTX_BUILDER(tnl_gtpu_flex_parser_0);
+	void DR_STE_CTX_BUILDER(tnl_gtpu_flex_parser_1);
+
+	/* Getters and Setters */
+	void (*ste_init)(u8 *hw_ste_p, u16 lu_type,
+			 bool is_rx, u16 gvmi);
+	void (*set_next_lu_type)(u8 *hw_ste_p, u16 lu_type);
+	u16  (*get_next_lu_type)(u8 *hw_ste_p);
+	void (*set_miss_addr)(u8 *hw_ste_p, u64 miss_addr);
+	u64  (*get_miss_addr)(u8 *hw_ste_p);
+	void (*set_hit_addr)(u8 *hw_ste_p, u64 icm_addr, u32 ht_size);
+	void (*set_byte_mask)(u8 *hw_ste_p, u16 byte_mask);
+	u16  (*get_byte_mask)(u8 *hw_ste_p);
+
+	/* Actions */
+	u32 actions_caps;
+	void (*set_actions_rx)(struct mlx5dr_domain *dmn,
+			       u8 *action_type_set,
+			       u32 actions_caps,
+			       u8 *hw_ste_arr,
+			       struct mlx5dr_ste_actions_attr *attr,
+			       u32 *added_stes);
+	void (*set_actions_tx)(struct mlx5dr_domain *dmn,
+			       u8 *action_type_set,
+			       u32 actions_caps,
+			       u8 *hw_ste_arr,
+			       struct mlx5dr_ste_actions_attr *attr,
+			       u32 *added_stes);
+	u32 modify_field_arr_sz;
+	const struct mlx5dr_ste_action_modify_field *modify_field_arr;
+	void (*set_action_set)(u8 *hw_action,
+			       u8 hw_field,
+			       u8 shifter,
+			       u8 length,
+			       u32 data);
+	void (*set_action_add)(u8 *hw_action,
+			       u8 hw_field,
+			       u8 shifter,
+			       u8 length,
+			       u32 data);
+	void (*set_action_copy)(u8 *hw_action,
+				u8 dst_hw_field,
+				u8 dst_shifter,
+				u8 dst_len,
+				u8 src_hw_field,
+				u8 src_shifter);
+	int (*set_action_decap_l3_list)(void *data,
+					u32 data_sz,
+					u8 *hw_action,
+					u32 hw_action_sz,
+					u16 *used_hw_action_num);
+
+	/* Send */
+	void (*prepare_for_postsend)(u8 *hw_ste_p, u32 ste_size);
+};
+
+struct mlx5dr_ste_ctx *mlx5dr_ste_get_ctx_v0(void);
+struct mlx5dr_ste_ctx *mlx5dr_ste_get_ctx_v1(void);
+struct mlx5dr_ste_ctx *mlx5dr_ste_get_ctx_v2(void);
+
+#endif  /* _DR_STE_ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_ste_v0.c b/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_ste_v0.c
new file mode 100644
index 000000000..2010d4ac6
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_ste_v0.c
@@ -0,0 +1,1960 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2020 NVIDIA CORPORATION. All rights reserved. */
+
+#include <linux/types.h>
+#include <linux/crc32.h>
+#include "dr_ste.h"
+
+#define SVLAN_ETHERTYPE		0x88a8
+#define DR_STE_ENABLE_FLOW_TAG	BIT(31)
+
+enum dr_ste_v0_entry_type {
+	DR_STE_TYPE_TX          = 1,
+	DR_STE_TYPE_RX          = 2,
+	DR_STE_TYPE_MODIFY_PKT  = 6,
+};
+
+enum dr_ste_v0_action_tunl {
+	DR_STE_TUNL_ACTION_NONE		= 0,
+	DR_STE_TUNL_ACTION_ENABLE	= 1,
+	DR_STE_TUNL_ACTION_DECAP	= 2,
+	DR_STE_TUNL_ACTION_L3_DECAP	= 3,
+	DR_STE_TUNL_ACTION_POP_VLAN	= 4,
+};
+
+enum dr_ste_v0_action_type {
+	DR_STE_ACTION_TYPE_PUSH_VLAN	= 1,
+	DR_STE_ACTION_TYPE_ENCAP_L3	= 3,
+	DR_STE_ACTION_TYPE_ENCAP	= 4,
+};
+
+enum dr_ste_v0_action_mdfy_op {
+	DR_STE_ACTION_MDFY_OP_COPY	= 0x1,
+	DR_STE_ACTION_MDFY_OP_SET	= 0x2,
+	DR_STE_ACTION_MDFY_OP_ADD	= 0x3,
+};
+
+#define DR_STE_CALC_LU_TYPE(lookup_type, rx, inner) \
+	((inner) ? DR_STE_V0_LU_TYPE_##lookup_type##_I : \
+		   (rx) ? DR_STE_V0_LU_TYPE_##lookup_type##_D : \
+			  DR_STE_V0_LU_TYPE_##lookup_type##_O)
+
+enum {
+	DR_STE_V0_LU_TYPE_NOP				= 0x00,
+	DR_STE_V0_LU_TYPE_SRC_GVMI_AND_QP		= 0x05,
+	DR_STE_V0_LU_TYPE_ETHL2_TUNNELING_I		= 0x0a,
+	DR_STE_V0_LU_TYPE_ETHL2_DST_O			= 0x06,
+	DR_STE_V0_LU_TYPE_ETHL2_DST_I			= 0x07,
+	DR_STE_V0_LU_TYPE_ETHL2_DST_D			= 0x1b,
+	DR_STE_V0_LU_TYPE_ETHL2_SRC_O			= 0x08,
+	DR_STE_V0_LU_TYPE_ETHL2_SRC_I			= 0x09,
+	DR_STE_V0_LU_TYPE_ETHL2_SRC_D			= 0x1c,
+	DR_STE_V0_LU_TYPE_ETHL2_SRC_DST_O		= 0x36,
+	DR_STE_V0_LU_TYPE_ETHL2_SRC_DST_I		= 0x37,
+	DR_STE_V0_LU_TYPE_ETHL2_SRC_DST_D		= 0x38,
+	DR_STE_V0_LU_TYPE_ETHL3_IPV6_DST_O		= 0x0d,
+	DR_STE_V0_LU_TYPE_ETHL3_IPV6_DST_I		= 0x0e,
+	DR_STE_V0_LU_TYPE_ETHL3_IPV6_DST_D		= 0x1e,
+	DR_STE_V0_LU_TYPE_ETHL3_IPV6_SRC_O		= 0x0f,
+	DR_STE_V0_LU_TYPE_ETHL3_IPV6_SRC_I		= 0x10,
+	DR_STE_V0_LU_TYPE_ETHL3_IPV6_SRC_D		= 0x1f,
+	DR_STE_V0_LU_TYPE_ETHL3_IPV4_5_TUPLE_O		= 0x11,
+	DR_STE_V0_LU_TYPE_ETHL3_IPV4_5_TUPLE_I		= 0x12,
+	DR_STE_V0_LU_TYPE_ETHL3_IPV4_5_TUPLE_D		= 0x20,
+	DR_STE_V0_LU_TYPE_ETHL3_IPV4_MISC_O		= 0x29,
+	DR_STE_V0_LU_TYPE_ETHL3_IPV4_MISC_I		= 0x2a,
+	DR_STE_V0_LU_TYPE_ETHL3_IPV4_MISC_D		= 0x2b,
+	DR_STE_V0_LU_TYPE_ETHL4_O			= 0x13,
+	DR_STE_V0_LU_TYPE_ETHL4_I			= 0x14,
+	DR_STE_V0_LU_TYPE_ETHL4_D			= 0x21,
+	DR_STE_V0_LU_TYPE_ETHL4_MISC_O			= 0x2c,
+	DR_STE_V0_LU_TYPE_ETHL4_MISC_I			= 0x2d,
+	DR_STE_V0_LU_TYPE_ETHL4_MISC_D			= 0x2e,
+	DR_STE_V0_LU_TYPE_MPLS_FIRST_O			= 0x15,
+	DR_STE_V0_LU_TYPE_MPLS_FIRST_I			= 0x24,
+	DR_STE_V0_LU_TYPE_MPLS_FIRST_D			= 0x25,
+	DR_STE_V0_LU_TYPE_GRE				= 0x16,
+	DR_STE_V0_LU_TYPE_FLEX_PARSER_0			= 0x22,
+	DR_STE_V0_LU_TYPE_FLEX_PARSER_1			= 0x23,
+	DR_STE_V0_LU_TYPE_FLEX_PARSER_TNL_HEADER	= 0x19,
+	DR_STE_V0_LU_TYPE_GENERAL_PURPOSE		= 0x18,
+	DR_STE_V0_LU_TYPE_STEERING_REGISTERS_0		= 0x2f,
+	DR_STE_V0_LU_TYPE_STEERING_REGISTERS_1		= 0x30,
+	DR_STE_V0_LU_TYPE_TUNNEL_HEADER			= 0x34,
+	DR_STE_V0_LU_TYPE_DONT_CARE			= MLX5DR_STE_LU_TYPE_DONT_CARE,
+};
+
+enum {
+	DR_STE_V0_ACTION_MDFY_FLD_L2_0		= 0,
+	DR_STE_V0_ACTION_MDFY_FLD_L2_1		= 1,
+	DR_STE_V0_ACTION_MDFY_FLD_L2_2		= 2,
+	DR_STE_V0_ACTION_MDFY_FLD_L3_0		= 3,
+	DR_STE_V0_ACTION_MDFY_FLD_L3_1		= 4,
+	DR_STE_V0_ACTION_MDFY_FLD_L3_2		= 5,
+	DR_STE_V0_ACTION_MDFY_FLD_L3_3		= 6,
+	DR_STE_V0_ACTION_MDFY_FLD_L3_4		= 7,
+	DR_STE_V0_ACTION_MDFY_FLD_L4_0		= 8,
+	DR_STE_V0_ACTION_MDFY_FLD_L4_1		= 9,
+	DR_STE_V0_ACTION_MDFY_FLD_MPLS		= 10,
+	DR_STE_V0_ACTION_MDFY_FLD_L2_TNL_0	= 11,
+	DR_STE_V0_ACTION_MDFY_FLD_REG_0		= 12,
+	DR_STE_V0_ACTION_MDFY_FLD_REG_1		= 13,
+	DR_STE_V0_ACTION_MDFY_FLD_REG_2		= 14,
+	DR_STE_V0_ACTION_MDFY_FLD_REG_3		= 15,
+	DR_STE_V0_ACTION_MDFY_FLD_L4_2		= 16,
+	DR_STE_V0_ACTION_MDFY_FLD_FLEX_0	= 17,
+	DR_STE_V0_ACTION_MDFY_FLD_FLEX_1	= 18,
+	DR_STE_V0_ACTION_MDFY_FLD_FLEX_2	= 19,
+	DR_STE_V0_ACTION_MDFY_FLD_FLEX_3	= 20,
+	DR_STE_V0_ACTION_MDFY_FLD_L2_TNL_1	= 21,
+	DR_STE_V0_ACTION_MDFY_FLD_METADATA	= 22,
+	DR_STE_V0_ACTION_MDFY_FLD_RESERVED	= 23,
+};
+
+static const struct mlx5dr_ste_action_modify_field dr_ste_v0_action_modify_field_arr[] = {
+	[MLX5_ACTION_IN_FIELD_OUT_SMAC_47_16] = {
+		.hw_field = DR_STE_V0_ACTION_MDFY_FLD_L2_1, .start = 16, .end = 47,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_SMAC_15_0] = {
+		.hw_field = DR_STE_V0_ACTION_MDFY_FLD_L2_1, .start = 0, .end = 15,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_ETHERTYPE] = {
+		.hw_field = DR_STE_V0_ACTION_MDFY_FLD_L2_2, .start = 32, .end = 47,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_DMAC_47_16] = {
+		.hw_field = DR_STE_V0_ACTION_MDFY_FLD_L2_0, .start = 16, .end = 47,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_DMAC_15_0] = {
+		.hw_field = DR_STE_V0_ACTION_MDFY_FLD_L2_0, .start = 0, .end = 15,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_IP_DSCP] = {
+		.hw_field = DR_STE_V0_ACTION_MDFY_FLD_L3_1, .start = 0, .end = 5,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_TCP_FLAGS] = {
+		.hw_field = DR_STE_V0_ACTION_MDFY_FLD_L4_0, .start = 48, .end = 56,
+		.l4_type = DR_STE_ACTION_MDFY_TYPE_L4_TCP,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_TCP_SPORT] = {
+		.hw_field = DR_STE_V0_ACTION_MDFY_FLD_L4_0, .start = 0, .end = 15,
+		.l4_type = DR_STE_ACTION_MDFY_TYPE_L4_TCP,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_TCP_DPORT] = {
+		.hw_field = DR_STE_V0_ACTION_MDFY_FLD_L4_0, .start = 16, .end = 31,
+		.l4_type = DR_STE_ACTION_MDFY_TYPE_L4_TCP,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_IP_TTL] = {
+		.hw_field = DR_STE_V0_ACTION_MDFY_FLD_L3_1, .start = 8, .end = 15,
+		.l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV4,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_IPV6_HOPLIMIT] = {
+		.hw_field = DR_STE_V0_ACTION_MDFY_FLD_L3_1, .start = 8, .end = 15,
+		.l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV6,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_UDP_SPORT] = {
+		.hw_field = DR_STE_V0_ACTION_MDFY_FLD_L4_0, .start = 0, .end = 15,
+		.l4_type = DR_STE_ACTION_MDFY_TYPE_L4_UDP,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_UDP_DPORT] = {
+		.hw_field = DR_STE_V0_ACTION_MDFY_FLD_L4_0, .start = 16, .end = 31,
+		.l4_type = DR_STE_ACTION_MDFY_TYPE_L4_UDP,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_SIPV6_127_96] = {
+		.hw_field = DR_STE_V0_ACTION_MDFY_FLD_L3_3, .start = 32, .end = 63,
+		.l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV6,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_SIPV6_95_64] = {
+		.hw_field = DR_STE_V0_ACTION_MDFY_FLD_L3_3, .start = 0, .end = 31,
+		.l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV6,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_SIPV6_63_32] = {
+		.hw_field = DR_STE_V0_ACTION_MDFY_FLD_L3_4, .start = 32, .end = 63,
+		.l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV6,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_SIPV6_31_0] = {
+		.hw_field = DR_STE_V0_ACTION_MDFY_FLD_L3_4, .start = 0, .end = 31,
+		.l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV6,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_DIPV6_127_96] = {
+		.hw_field = DR_STE_V0_ACTION_MDFY_FLD_L3_0, .start = 32, .end = 63,
+		.l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV6,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_DIPV6_95_64] = {
+		.hw_field = DR_STE_V0_ACTION_MDFY_FLD_L3_0, .start = 0, .end = 31,
+		.l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV6,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_DIPV6_63_32] = {
+		.hw_field = DR_STE_V0_ACTION_MDFY_FLD_L3_2, .start = 32, .end = 63,
+		.l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV6,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_DIPV6_31_0] = {
+		.hw_field = DR_STE_V0_ACTION_MDFY_FLD_L3_2, .start = 0, .end = 31,
+		.l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV6,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_SIPV4] = {
+		.hw_field = DR_STE_V0_ACTION_MDFY_FLD_L3_0, .start = 0, .end = 31,
+		.l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV4,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_DIPV4] = {
+		.hw_field = DR_STE_V0_ACTION_MDFY_FLD_L3_0, .start = 32, .end = 63,
+		.l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV4,
+	},
+	[MLX5_ACTION_IN_FIELD_METADATA_REG_A] = {
+		.hw_field = DR_STE_V0_ACTION_MDFY_FLD_METADATA, .start = 0, .end = 31,
+	},
+	[MLX5_ACTION_IN_FIELD_METADATA_REG_B] = {
+		.hw_field = DR_STE_V0_ACTION_MDFY_FLD_METADATA, .start = 32, .end = 63,
+	},
+	[MLX5_ACTION_IN_FIELD_METADATA_REG_C_0] = {
+		.hw_field = DR_STE_V0_ACTION_MDFY_FLD_REG_0, .start = 32, .end = 63,
+	},
+	[MLX5_ACTION_IN_FIELD_METADATA_REG_C_1] = {
+		.hw_field = DR_STE_V0_ACTION_MDFY_FLD_REG_0, .start = 0, .end = 31,
+	},
+	[MLX5_ACTION_IN_FIELD_METADATA_REG_C_2] = {
+		.hw_field = DR_STE_V0_ACTION_MDFY_FLD_REG_1, .start = 32, .end = 63,
+	},
+	[MLX5_ACTION_IN_FIELD_METADATA_REG_C_3] = {
+		.hw_field = DR_STE_V0_ACTION_MDFY_FLD_REG_1, .start = 0, .end = 31,
+	},
+	[MLX5_ACTION_IN_FIELD_METADATA_REG_C_4] = {
+		.hw_field = DR_STE_V0_ACTION_MDFY_FLD_REG_2, .start = 32, .end = 63,
+	},
+	[MLX5_ACTION_IN_FIELD_METADATA_REG_C_5] = {
+		.hw_field = DR_STE_V0_ACTION_MDFY_FLD_REG_2, .start = 0, .end = 31,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_TCP_SEQ_NUM] = {
+		.hw_field = DR_STE_V0_ACTION_MDFY_FLD_L4_1, .start = 32, .end = 63,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_TCP_ACK_NUM] = {
+		.hw_field = DR_STE_V0_ACTION_MDFY_FLD_L4_1, .start = 0, .end = 31,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_FIRST_VID] = {
+		.hw_field = DR_STE_V0_ACTION_MDFY_FLD_L2_2, .start = 0, .end = 15,
+	},
+};
+
+static void dr_ste_v0_set_entry_type(u8 *hw_ste_p, u8 entry_type)
+{
+	MLX5_SET(ste_general, hw_ste_p, entry_type, entry_type);
+}
+
+static u8 dr_ste_v0_get_entry_type(u8 *hw_ste_p)
+{
+	return MLX5_GET(ste_general, hw_ste_p, entry_type);
+}
+
+static void dr_ste_v0_set_miss_addr(u8 *hw_ste_p, u64 miss_addr)
+{
+	u64 index = miss_addr >> 6;
+
+	/* Miss address for TX and RX STEs located in the same offsets */
+	MLX5_SET(ste_rx_steering_mult, hw_ste_p, miss_address_39_32, index >> 26);
+	MLX5_SET(ste_rx_steering_mult, hw_ste_p, miss_address_31_6, index);
+}
+
+static u64 dr_ste_v0_get_miss_addr(u8 *hw_ste_p)
+{
+	u64 index =
+		((u64)MLX5_GET(ste_rx_steering_mult, hw_ste_p, miss_address_31_6) |
+		 ((u64)MLX5_GET(ste_rx_steering_mult, hw_ste_p, miss_address_39_32)) << 26);
+
+	return index << 6;
+}
+
+static void dr_ste_v0_set_byte_mask(u8 *hw_ste_p, u16 byte_mask)
+{
+	MLX5_SET(ste_general, hw_ste_p, byte_mask, byte_mask);
+}
+
+static u16 dr_ste_v0_get_byte_mask(u8 *hw_ste_p)
+{
+	return MLX5_GET(ste_general, hw_ste_p, byte_mask);
+}
+
+static void dr_ste_v0_set_lu_type(u8 *hw_ste_p, u16 lu_type)
+{
+	MLX5_SET(ste_general, hw_ste_p, entry_sub_type, lu_type);
+}
+
+static void dr_ste_v0_set_next_lu_type(u8 *hw_ste_p, u16 lu_type)
+{
+	MLX5_SET(ste_general, hw_ste_p, next_lu_type, lu_type);
+}
+
+static u16 dr_ste_v0_get_next_lu_type(u8 *hw_ste_p)
+{
+	return MLX5_GET(ste_general, hw_ste_p, next_lu_type);
+}
+
+static void dr_ste_v0_set_hit_gvmi(u8 *hw_ste_p, u16 gvmi)
+{
+	MLX5_SET(ste_general, hw_ste_p, next_table_base_63_48, gvmi);
+}
+
+static void dr_ste_v0_set_hit_addr(u8 *hw_ste_p, u64 icm_addr, u32 ht_size)
+{
+	u64 index = (icm_addr >> 5) | ht_size;
+
+	MLX5_SET(ste_general, hw_ste_p, next_table_base_39_32_size, index >> 27);
+	MLX5_SET(ste_general, hw_ste_p, next_table_base_31_5_size, index);
+}
+
+static void dr_ste_v0_init_full(u8 *hw_ste_p, u16 lu_type,
+				enum dr_ste_v0_entry_type entry_type, u16 gvmi)
+{
+	dr_ste_v0_set_entry_type(hw_ste_p, entry_type);
+	dr_ste_v0_set_lu_type(hw_ste_p, lu_type);
+	dr_ste_v0_set_next_lu_type(hw_ste_p, MLX5DR_STE_LU_TYPE_DONT_CARE);
+
+	/* Set GVMI once, this is the same for RX/TX
+	 * bits 63_48 of next table base / miss address encode the next GVMI
+	 */
+	MLX5_SET(ste_rx_steering_mult, hw_ste_p, gvmi, gvmi);
+	MLX5_SET(ste_rx_steering_mult, hw_ste_p, next_table_base_63_48, gvmi);
+	MLX5_SET(ste_rx_steering_mult, hw_ste_p, miss_address_63_48, gvmi);
+}
+
+static void dr_ste_v0_init(u8 *hw_ste_p, u16 lu_type,
+			   bool is_rx, u16 gvmi)
+{
+	enum dr_ste_v0_entry_type entry_type;
+
+	entry_type = is_rx ? DR_STE_TYPE_RX : DR_STE_TYPE_TX;
+	dr_ste_v0_init_full(hw_ste_p, lu_type, entry_type, gvmi);
+}
+
+static void dr_ste_v0_rx_set_flow_tag(u8 *hw_ste_p, u32 flow_tag)
+{
+	MLX5_SET(ste_rx_steering_mult, hw_ste_p, qp_list_pointer,
+		 DR_STE_ENABLE_FLOW_TAG | flow_tag);
+}
+
+static void dr_ste_v0_set_counter_id(u8 *hw_ste_p, u32 ctr_id)
+{
+	/* This can be used for both rx_steering_mult and for sx_transmit */
+	MLX5_SET(ste_rx_steering_mult, hw_ste_p, counter_trigger_15_0, ctr_id);
+	MLX5_SET(ste_rx_steering_mult, hw_ste_p, counter_trigger_23_16, ctr_id >> 16);
+}
+
+static void dr_ste_v0_set_go_back_bit(u8 *hw_ste_p)
+{
+	MLX5_SET(ste_sx_transmit, hw_ste_p, go_back, 1);
+}
+
+static void dr_ste_v0_set_tx_push_vlan(u8 *hw_ste_p, u32 vlan_hdr,
+				       bool go_back)
+{
+	MLX5_SET(ste_sx_transmit, hw_ste_p, action_type,
+		 DR_STE_ACTION_TYPE_PUSH_VLAN);
+	MLX5_SET(ste_sx_transmit, hw_ste_p, encap_pointer_vlan_data, vlan_hdr);
+	/* Due to HW limitation we need to set this bit, otherwise reformat +
+	 * push vlan will not work.
+	 */
+	if (go_back)
+		dr_ste_v0_set_go_back_bit(hw_ste_p);
+}
+
+static void dr_ste_v0_set_tx_encap(void *hw_ste_p, u32 reformat_id,
+				   int size, bool encap_l3)
+{
+	MLX5_SET(ste_sx_transmit, hw_ste_p, action_type,
+		 encap_l3 ? DR_STE_ACTION_TYPE_ENCAP_L3 : DR_STE_ACTION_TYPE_ENCAP);
+	/* The hardware expects here size in words (2 byte) */
+	MLX5_SET(ste_sx_transmit, hw_ste_p, action_description, size / 2);
+	MLX5_SET(ste_sx_transmit, hw_ste_p, encap_pointer_vlan_data, reformat_id);
+}
+
+static void dr_ste_v0_set_rx_decap(u8 *hw_ste_p)
+{
+	MLX5_SET(ste_rx_steering_mult, hw_ste_p, tunneling_action,
+		 DR_STE_TUNL_ACTION_DECAP);
+	MLX5_SET(ste_rx_steering_mult, hw_ste_p, fail_on_error, 1);
+}
+
+static void dr_ste_v0_set_rx_pop_vlan(u8 *hw_ste_p)
+{
+	MLX5_SET(ste_rx_steering_mult, hw_ste_p, tunneling_action,
+		 DR_STE_TUNL_ACTION_POP_VLAN);
+}
+
+static void dr_ste_v0_set_rx_decap_l3(u8 *hw_ste_p, bool vlan)
+{
+	MLX5_SET(ste_rx_steering_mult, hw_ste_p, tunneling_action,
+		 DR_STE_TUNL_ACTION_L3_DECAP);
+	MLX5_SET(ste_modify_packet, hw_ste_p, action_description, vlan ? 1 : 0);
+	MLX5_SET(ste_rx_steering_mult, hw_ste_p, fail_on_error, 1);
+}
+
+static void dr_ste_v0_set_rewrite_actions(u8 *hw_ste_p, u16 num_of_actions,
+					  u32 re_write_index)
+{
+	MLX5_SET(ste_modify_packet, hw_ste_p, number_of_re_write_actions,
+		 num_of_actions);
+	MLX5_SET(ste_modify_packet, hw_ste_p, header_re_write_actions_pointer,
+		 re_write_index);
+}
+
+static void dr_ste_v0_arr_init_next(u8 **last_ste,
+				    u32 *added_stes,
+				    enum dr_ste_v0_entry_type entry_type,
+				    u16 gvmi)
+{
+	(*added_stes)++;
+	*last_ste += DR_STE_SIZE;
+	dr_ste_v0_init_full(*last_ste, MLX5DR_STE_LU_TYPE_DONT_CARE,
+			    entry_type, gvmi);
+}
+
+static void
+dr_ste_v0_set_actions_tx(struct mlx5dr_domain *dmn,
+			 u8 *action_type_set,
+			 u32 actions_caps,
+			 u8 *last_ste,
+			 struct mlx5dr_ste_actions_attr *attr,
+			 u32 *added_stes)
+{
+	bool encap = action_type_set[DR_ACTION_TYP_L2_TO_TNL_L2] ||
+		action_type_set[DR_ACTION_TYP_L2_TO_TNL_L3];
+
+	/* We want to make sure the modify header comes before L2
+	 * encapsulation. The reason for that is that we support
+	 * modify headers for outer headers only
+	 */
+	if (action_type_set[DR_ACTION_TYP_MODIFY_HDR] && attr->modify_actions) {
+		dr_ste_v0_set_entry_type(last_ste, DR_STE_TYPE_MODIFY_PKT);
+		dr_ste_v0_set_rewrite_actions(last_ste,
+					      attr->modify_actions,
+					      attr->modify_index);
+	}
+
+	if (action_type_set[DR_ACTION_TYP_PUSH_VLAN]) {
+		int i;
+
+		for (i = 0; i < attr->vlans.count; i++) {
+			if (i || action_type_set[DR_ACTION_TYP_MODIFY_HDR])
+				dr_ste_v0_arr_init_next(&last_ste,
+							added_stes,
+							DR_STE_TYPE_TX,
+							attr->gvmi);
+
+			dr_ste_v0_set_tx_push_vlan(last_ste,
+						   attr->vlans.headers[i],
+						   encap);
+		}
+	}
+
+	if (encap) {
+		/* Modify header and encapsulation require a different STEs.
+		 * Since modify header STE format doesn't support encapsulation
+		 * tunneling_action.
+		 */
+		if (action_type_set[DR_ACTION_TYP_MODIFY_HDR] ||
+		    action_type_set[DR_ACTION_TYP_PUSH_VLAN])
+			dr_ste_v0_arr_init_next(&last_ste,
+						added_stes,
+						DR_STE_TYPE_TX,
+						attr->gvmi);
+
+		dr_ste_v0_set_tx_encap(last_ste,
+				       attr->reformat.id,
+				       attr->reformat.size,
+				       action_type_set[DR_ACTION_TYP_L2_TO_TNL_L3]);
+		/* Whenever prio_tag_required enabled, we can be sure that the
+		 * previous table (ACL) already push vlan to our packet,
+		 * And due to HW limitation we need to set this bit, otherwise
+		 * push vlan + reformat will not work.
+		 */
+		if (MLX5_CAP_GEN(dmn->mdev, prio_tag_required))
+			dr_ste_v0_set_go_back_bit(last_ste);
+	}
+
+	if (action_type_set[DR_ACTION_TYP_CTR])
+		dr_ste_v0_set_counter_id(last_ste, attr->ctr_id);
+
+	dr_ste_v0_set_hit_gvmi(last_ste, attr->hit_gvmi);
+	dr_ste_v0_set_hit_addr(last_ste, attr->final_icm_addr, 1);
+}
+
+static void
+dr_ste_v0_set_actions_rx(struct mlx5dr_domain *dmn,
+			 u8 *action_type_set,
+			 u32 actions_caps,
+			 u8 *last_ste,
+			 struct mlx5dr_ste_actions_attr *attr,
+			 u32 *added_stes)
+{
+	if (action_type_set[DR_ACTION_TYP_CTR])
+		dr_ste_v0_set_counter_id(last_ste, attr->ctr_id);
+
+	if (action_type_set[DR_ACTION_TYP_TNL_L3_TO_L2]) {
+		dr_ste_v0_set_entry_type(last_ste, DR_STE_TYPE_MODIFY_PKT);
+		dr_ste_v0_set_rx_decap_l3(last_ste, attr->decap_with_vlan);
+		dr_ste_v0_set_rewrite_actions(last_ste,
+					      attr->decap_actions,
+					      attr->decap_index);
+	}
+
+	if (action_type_set[DR_ACTION_TYP_TNL_L2_TO_L2])
+		dr_ste_v0_set_rx_decap(last_ste);
+
+	if (action_type_set[DR_ACTION_TYP_POP_VLAN]) {
+		int i;
+
+		for (i = 0; i < attr->vlans.count; i++) {
+			if (i ||
+			    action_type_set[DR_ACTION_TYP_TNL_L2_TO_L2] ||
+			    action_type_set[DR_ACTION_TYP_TNL_L3_TO_L2])
+				dr_ste_v0_arr_init_next(&last_ste,
+							added_stes,
+							DR_STE_TYPE_RX,
+							attr->gvmi);
+
+			dr_ste_v0_set_rx_pop_vlan(last_ste);
+		}
+	}
+
+	if (action_type_set[DR_ACTION_TYP_MODIFY_HDR] && attr->modify_actions) {
+		if (dr_ste_v0_get_entry_type(last_ste) == DR_STE_TYPE_MODIFY_PKT)
+			dr_ste_v0_arr_init_next(&last_ste,
+						added_stes,
+						DR_STE_TYPE_MODIFY_PKT,
+						attr->gvmi);
+		else
+			dr_ste_v0_set_entry_type(last_ste, DR_STE_TYPE_MODIFY_PKT);
+
+		dr_ste_v0_set_rewrite_actions(last_ste,
+					      attr->modify_actions,
+					      attr->modify_index);
+	}
+
+	if (action_type_set[DR_ACTION_TYP_TAG]) {
+		if (dr_ste_v0_get_entry_type(last_ste) == DR_STE_TYPE_MODIFY_PKT)
+			dr_ste_v0_arr_init_next(&last_ste,
+						added_stes,
+						DR_STE_TYPE_RX,
+						attr->gvmi);
+
+		dr_ste_v0_rx_set_flow_tag(last_ste, attr->flow_tag);
+	}
+
+	dr_ste_v0_set_hit_gvmi(last_ste, attr->hit_gvmi);
+	dr_ste_v0_set_hit_addr(last_ste, attr->final_icm_addr, 1);
+}
+
+static void dr_ste_v0_set_action_set(u8 *hw_action,
+				     u8 hw_field,
+				     u8 shifter,
+				     u8 length,
+				     u32 data)
+{
+	length = (length == 32) ? 0 : length;
+	MLX5_SET(dr_action_hw_set, hw_action, opcode, DR_STE_ACTION_MDFY_OP_SET);
+	MLX5_SET(dr_action_hw_set, hw_action, destination_field_code, hw_field);
+	MLX5_SET(dr_action_hw_set, hw_action, destination_left_shifter, shifter);
+	MLX5_SET(dr_action_hw_set, hw_action, destination_length, length);
+	MLX5_SET(dr_action_hw_set, hw_action, inline_data, data);
+}
+
+static void dr_ste_v0_set_action_add(u8 *hw_action,
+				     u8 hw_field,
+				     u8 shifter,
+				     u8 length,
+				     u32 data)
+{
+	length = (length == 32) ? 0 : length;
+	MLX5_SET(dr_action_hw_set, hw_action, opcode, DR_STE_ACTION_MDFY_OP_ADD);
+	MLX5_SET(dr_action_hw_set, hw_action, destination_field_code, hw_field);
+	MLX5_SET(dr_action_hw_set, hw_action, destination_left_shifter, shifter);
+	MLX5_SET(dr_action_hw_set, hw_action, destination_length, length);
+	MLX5_SET(dr_action_hw_set, hw_action, inline_data, data);
+}
+
+static void dr_ste_v0_set_action_copy(u8 *hw_action,
+				      u8 dst_hw_field,
+				      u8 dst_shifter,
+				      u8 dst_len,
+				      u8 src_hw_field,
+				      u8 src_shifter)
+{
+	MLX5_SET(dr_action_hw_copy, hw_action, opcode, DR_STE_ACTION_MDFY_OP_COPY);
+	MLX5_SET(dr_action_hw_copy, hw_action, destination_field_code, dst_hw_field);
+	MLX5_SET(dr_action_hw_copy, hw_action, destination_left_shifter, dst_shifter);
+	MLX5_SET(dr_action_hw_copy, hw_action, destination_length, dst_len);
+	MLX5_SET(dr_action_hw_copy, hw_action, source_field_code, src_hw_field);
+	MLX5_SET(dr_action_hw_copy, hw_action, source_left_shifter, src_shifter);
+}
+
+#define DR_STE_DECAP_L3_MIN_ACTION_NUM	5
+
+static int
+dr_ste_v0_set_action_decap_l3_list(void *data, u32 data_sz,
+				   u8 *hw_action, u32 hw_action_sz,
+				   u16 *used_hw_action_num)
+{
+	struct mlx5_ifc_l2_hdr_bits *l2_hdr = data;
+	u32 hw_action_num;
+	int required_actions;
+	u32 hdr_fld_4b;
+	u16 hdr_fld_2b;
+	u16 vlan_type;
+	bool vlan;
+
+	vlan = (data_sz != HDR_LEN_L2);
+	hw_action_num = hw_action_sz / MLX5_ST_SZ_BYTES(dr_action_hw_set);
+	required_actions = DR_STE_DECAP_L3_MIN_ACTION_NUM + !!vlan;
+
+	if (hw_action_num < required_actions)
+		return -ENOMEM;
+
+	/* dmac_47_16 */
+	MLX5_SET(dr_action_hw_set, hw_action,
+		 opcode, DR_STE_ACTION_MDFY_OP_SET);
+	MLX5_SET(dr_action_hw_set, hw_action,
+		 destination_length, 0);
+	MLX5_SET(dr_action_hw_set, hw_action,
+		 destination_field_code, DR_STE_V0_ACTION_MDFY_FLD_L2_0);
+	MLX5_SET(dr_action_hw_set, hw_action,
+		 destination_left_shifter, 16);
+	hdr_fld_4b = MLX5_GET(l2_hdr, l2_hdr, dmac_47_16);
+	MLX5_SET(dr_action_hw_set, hw_action,
+		 inline_data, hdr_fld_4b);
+	hw_action += MLX5_ST_SZ_BYTES(dr_action_hw_set);
+
+	/* smac_47_16 */
+	MLX5_SET(dr_action_hw_set, hw_action,
+		 opcode, DR_STE_ACTION_MDFY_OP_SET);
+	MLX5_SET(dr_action_hw_set, hw_action,
+		 destination_length, 0);
+	MLX5_SET(dr_action_hw_set, hw_action,
+		 destination_field_code, DR_STE_V0_ACTION_MDFY_FLD_L2_1);
+	MLX5_SET(dr_action_hw_set, hw_action, destination_left_shifter, 16);
+	hdr_fld_4b = (MLX5_GET(l2_hdr, l2_hdr, smac_31_0) >> 16 |
+		      MLX5_GET(l2_hdr, l2_hdr, smac_47_32) << 16);
+	MLX5_SET(dr_action_hw_set, hw_action, inline_data, hdr_fld_4b);
+	hw_action += MLX5_ST_SZ_BYTES(dr_action_hw_set);
+
+	/* dmac_15_0 */
+	MLX5_SET(dr_action_hw_set, hw_action,
+		 opcode, DR_STE_ACTION_MDFY_OP_SET);
+	MLX5_SET(dr_action_hw_set, hw_action,
+		 destination_length, 16);
+	MLX5_SET(dr_action_hw_set, hw_action,
+		 destination_field_code, DR_STE_V0_ACTION_MDFY_FLD_L2_0);
+	MLX5_SET(dr_action_hw_set, hw_action,
+		 destination_left_shifter, 0);
+	hdr_fld_2b = MLX5_GET(l2_hdr, l2_hdr, dmac_15_0);
+	MLX5_SET(dr_action_hw_set, hw_action,
+		 inline_data, hdr_fld_2b);
+	hw_action += MLX5_ST_SZ_BYTES(dr_action_hw_set);
+
+	/* ethertype + (optional) vlan */
+	MLX5_SET(dr_action_hw_set, hw_action,
+		 opcode, DR_STE_ACTION_MDFY_OP_SET);
+	MLX5_SET(dr_action_hw_set, hw_action,
+		 destination_field_code, DR_STE_V0_ACTION_MDFY_FLD_L2_2);
+	MLX5_SET(dr_action_hw_set, hw_action,
+		 destination_left_shifter, 32);
+	if (!vlan) {
+		hdr_fld_2b = MLX5_GET(l2_hdr, l2_hdr, ethertype);
+		MLX5_SET(dr_action_hw_set, hw_action, inline_data, hdr_fld_2b);
+		MLX5_SET(dr_action_hw_set, hw_action, destination_length, 16);
+	} else {
+		hdr_fld_2b = MLX5_GET(l2_hdr, l2_hdr, ethertype);
+		vlan_type = hdr_fld_2b == SVLAN_ETHERTYPE ? DR_STE_SVLAN : DR_STE_CVLAN;
+		hdr_fld_2b = MLX5_GET(l2_hdr, l2_hdr, vlan);
+		hdr_fld_4b = (vlan_type << 16) | hdr_fld_2b;
+		MLX5_SET(dr_action_hw_set, hw_action, inline_data, hdr_fld_4b);
+		MLX5_SET(dr_action_hw_set, hw_action, destination_length, 18);
+	}
+	hw_action += MLX5_ST_SZ_BYTES(dr_action_hw_set);
+
+	/* smac_15_0 */
+	MLX5_SET(dr_action_hw_set, hw_action,
+		 opcode, DR_STE_ACTION_MDFY_OP_SET);
+	MLX5_SET(dr_action_hw_set, hw_action,
+		 destination_length, 16);
+	MLX5_SET(dr_action_hw_set, hw_action,
+		 destination_field_code, DR_STE_V0_ACTION_MDFY_FLD_L2_1);
+	MLX5_SET(dr_action_hw_set, hw_action,
+		 destination_left_shifter, 0);
+	hdr_fld_2b = MLX5_GET(l2_hdr, l2_hdr, smac_31_0);
+	MLX5_SET(dr_action_hw_set, hw_action, inline_data, hdr_fld_2b);
+	hw_action += MLX5_ST_SZ_BYTES(dr_action_hw_set);
+
+	if (vlan) {
+		MLX5_SET(dr_action_hw_set, hw_action,
+			 opcode, DR_STE_ACTION_MDFY_OP_SET);
+		hdr_fld_2b = MLX5_GET(l2_hdr, l2_hdr, vlan_type);
+		MLX5_SET(dr_action_hw_set, hw_action,
+			 inline_data, hdr_fld_2b);
+		MLX5_SET(dr_action_hw_set, hw_action,
+			 destination_length, 16);
+		MLX5_SET(dr_action_hw_set, hw_action,
+			 destination_field_code, DR_STE_V0_ACTION_MDFY_FLD_L2_2);
+		MLX5_SET(dr_action_hw_set, hw_action,
+			 destination_left_shifter, 0);
+	}
+
+	*used_hw_action_num = required_actions;
+
+	return 0;
+}
+
+static void
+dr_ste_v0_build_eth_l2_src_dst_bit_mask(struct mlx5dr_match_param *value,
+					bool inner, u8 *bit_mask)
+{
+	struct mlx5dr_match_spec *mask = inner ? &value->inner : &value->outer;
+
+	DR_STE_SET_TAG(eth_l2_src_dst, bit_mask, dmac_47_16, mask, dmac_47_16);
+	DR_STE_SET_TAG(eth_l2_src_dst, bit_mask, dmac_15_0, mask, dmac_15_0);
+
+	if (mask->smac_47_16 || mask->smac_15_0) {
+		MLX5_SET(ste_eth_l2_src_dst, bit_mask, smac_47_32,
+			 mask->smac_47_16 >> 16);
+		MLX5_SET(ste_eth_l2_src_dst, bit_mask, smac_31_0,
+			 mask->smac_47_16 << 16 | mask->smac_15_0);
+		mask->smac_47_16 = 0;
+		mask->smac_15_0 = 0;
+	}
+
+	DR_STE_SET_TAG(eth_l2_src_dst, bit_mask, first_vlan_id, mask, first_vid);
+	DR_STE_SET_TAG(eth_l2_src_dst, bit_mask, first_cfi, mask, first_cfi);
+	DR_STE_SET_TAG(eth_l2_src_dst, bit_mask, first_priority, mask, first_prio);
+	DR_STE_SET_ONES(eth_l2_src_dst, bit_mask, l3_type, mask, ip_version);
+
+	if (mask->cvlan_tag) {
+		MLX5_SET(ste_eth_l2_src_dst, bit_mask, first_vlan_qualifier, -1);
+		mask->cvlan_tag = 0;
+	} else if (mask->svlan_tag) {
+		MLX5_SET(ste_eth_l2_src_dst, bit_mask, first_vlan_qualifier, -1);
+		mask->svlan_tag = 0;
+	}
+}
+
+static int
+dr_ste_v0_build_eth_l2_src_dst_tag(struct mlx5dr_match_param *value,
+				   struct mlx5dr_ste_build *sb,
+				   u8 *tag)
+{
+	struct mlx5dr_match_spec *spec = sb->inner ? &value->inner : &value->outer;
+
+	DR_STE_SET_TAG(eth_l2_src_dst, tag, dmac_47_16, spec, dmac_47_16);
+	DR_STE_SET_TAG(eth_l2_src_dst, tag, dmac_15_0, spec, dmac_15_0);
+
+	if (spec->smac_47_16 || spec->smac_15_0) {
+		MLX5_SET(ste_eth_l2_src_dst, tag, smac_47_32,
+			 spec->smac_47_16 >> 16);
+		MLX5_SET(ste_eth_l2_src_dst, tag, smac_31_0,
+			 spec->smac_47_16 << 16 | spec->smac_15_0);
+		spec->smac_47_16 = 0;
+		spec->smac_15_0 = 0;
+	}
+
+	if (spec->ip_version) {
+		if (spec->ip_version == IP_VERSION_IPV4) {
+			MLX5_SET(ste_eth_l2_src_dst, tag, l3_type, STE_IPV4);
+			spec->ip_version = 0;
+		} else if (spec->ip_version == IP_VERSION_IPV6) {
+			MLX5_SET(ste_eth_l2_src_dst, tag, l3_type, STE_IPV6);
+			spec->ip_version = 0;
+		} else {
+			return -EINVAL;
+		}
+	}
+
+	DR_STE_SET_TAG(eth_l2_src_dst, tag, first_vlan_id, spec, first_vid);
+	DR_STE_SET_TAG(eth_l2_src_dst, tag, first_cfi, spec, first_cfi);
+	DR_STE_SET_TAG(eth_l2_src_dst, tag, first_priority, spec, first_prio);
+
+	if (spec->cvlan_tag) {
+		MLX5_SET(ste_eth_l2_src_dst, tag, first_vlan_qualifier, DR_STE_CVLAN);
+		spec->cvlan_tag = 0;
+	} else if (spec->svlan_tag) {
+		MLX5_SET(ste_eth_l2_src_dst, tag, first_vlan_qualifier, DR_STE_SVLAN);
+		spec->svlan_tag = 0;
+	}
+	return 0;
+}
+
+static void
+dr_ste_v0_build_eth_l2_src_dst_init(struct mlx5dr_ste_build *sb,
+				    struct mlx5dr_match_param *mask)
+{
+	dr_ste_v0_build_eth_l2_src_dst_bit_mask(mask, sb->inner, sb->bit_mask);
+
+	sb->lu_type = DR_STE_CALC_LU_TYPE(ETHL2_SRC_DST, sb->rx, sb->inner);
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v0_build_eth_l2_src_dst_tag;
+}
+
+static int
+dr_ste_v0_build_eth_l3_ipv6_dst_tag(struct mlx5dr_match_param *value,
+				    struct mlx5dr_ste_build *sb,
+				    u8 *tag)
+{
+	struct mlx5dr_match_spec *spec = sb->inner ? &value->inner : &value->outer;
+
+	DR_STE_SET_TAG(eth_l3_ipv6_dst, tag, dst_ip_127_96, spec, dst_ip_127_96);
+	DR_STE_SET_TAG(eth_l3_ipv6_dst, tag, dst_ip_95_64, spec, dst_ip_95_64);
+	DR_STE_SET_TAG(eth_l3_ipv6_dst, tag, dst_ip_63_32, spec, dst_ip_63_32);
+	DR_STE_SET_TAG(eth_l3_ipv6_dst, tag, dst_ip_31_0, spec, dst_ip_31_0);
+
+	return 0;
+}
+
+static void
+dr_ste_v0_build_eth_l3_ipv6_dst_init(struct mlx5dr_ste_build *sb,
+				     struct mlx5dr_match_param *mask)
+{
+	dr_ste_v0_build_eth_l3_ipv6_dst_tag(mask, sb, sb->bit_mask);
+
+	sb->lu_type = DR_STE_CALC_LU_TYPE(ETHL3_IPV6_DST, sb->rx, sb->inner);
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v0_build_eth_l3_ipv6_dst_tag;
+}
+
+static int
+dr_ste_v0_build_eth_l3_ipv6_src_tag(struct mlx5dr_match_param *value,
+				    struct mlx5dr_ste_build *sb,
+				    u8 *tag)
+{
+	struct mlx5dr_match_spec *spec = sb->inner ? &value->inner : &value->outer;
+
+	DR_STE_SET_TAG(eth_l3_ipv6_src, tag, src_ip_127_96, spec, src_ip_127_96);
+	DR_STE_SET_TAG(eth_l3_ipv6_src, tag, src_ip_95_64, spec, src_ip_95_64);
+	DR_STE_SET_TAG(eth_l3_ipv6_src, tag, src_ip_63_32, spec, src_ip_63_32);
+	DR_STE_SET_TAG(eth_l3_ipv6_src, tag, src_ip_31_0, spec, src_ip_31_0);
+
+	return 0;
+}
+
+static void
+dr_ste_v0_build_eth_l3_ipv6_src_init(struct mlx5dr_ste_build *sb,
+				     struct mlx5dr_match_param *mask)
+{
+	dr_ste_v0_build_eth_l3_ipv6_src_tag(mask, sb, sb->bit_mask);
+
+	sb->lu_type = DR_STE_CALC_LU_TYPE(ETHL3_IPV6_SRC, sb->rx, sb->inner);
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v0_build_eth_l3_ipv6_src_tag;
+}
+
+static int
+dr_ste_v0_build_eth_l3_ipv4_5_tuple_tag(struct mlx5dr_match_param *value,
+					struct mlx5dr_ste_build *sb,
+					u8 *tag)
+{
+	struct mlx5dr_match_spec *spec = sb->inner ? &value->inner : &value->outer;
+
+	DR_STE_SET_TAG(eth_l3_ipv4_5_tuple, tag, destination_address, spec, dst_ip_31_0);
+	DR_STE_SET_TAG(eth_l3_ipv4_5_tuple, tag, source_address, spec, src_ip_31_0);
+	DR_STE_SET_TAG(eth_l3_ipv4_5_tuple, tag, destination_port, spec, tcp_dport);
+	DR_STE_SET_TAG(eth_l3_ipv4_5_tuple, tag, destination_port, spec, udp_dport);
+	DR_STE_SET_TAG(eth_l3_ipv4_5_tuple, tag, source_port, spec, tcp_sport);
+	DR_STE_SET_TAG(eth_l3_ipv4_5_tuple, tag, source_port, spec, udp_sport);
+	DR_STE_SET_TAG(eth_l3_ipv4_5_tuple, tag, protocol, spec, ip_protocol);
+	DR_STE_SET_TAG(eth_l3_ipv4_5_tuple, tag, fragmented, spec, frag);
+	DR_STE_SET_TAG(eth_l3_ipv4_5_tuple, tag, dscp, spec, ip_dscp);
+	DR_STE_SET_TAG(eth_l3_ipv4_5_tuple, tag, ecn, spec, ip_ecn);
+
+	if (spec->tcp_flags) {
+		DR_STE_SET_TCP_FLAGS(eth_l3_ipv4_5_tuple, tag, spec);
+		spec->tcp_flags = 0;
+	}
+
+	return 0;
+}
+
+static void
+dr_ste_v0_build_eth_l3_ipv4_5_tuple_init(struct mlx5dr_ste_build *sb,
+					 struct mlx5dr_match_param *mask)
+{
+	dr_ste_v0_build_eth_l3_ipv4_5_tuple_tag(mask, sb, sb->bit_mask);
+
+	sb->lu_type = DR_STE_CALC_LU_TYPE(ETHL3_IPV4_5_TUPLE, sb->rx, sb->inner);
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v0_build_eth_l3_ipv4_5_tuple_tag;
+}
+
+static void
+dr_ste_v0_build_eth_l2_src_or_dst_bit_mask(struct mlx5dr_match_param *value,
+					   bool inner, u8 *bit_mask)
+{
+	struct mlx5dr_match_spec *mask = inner ? &value->inner : &value->outer;
+	struct mlx5dr_match_misc *misc_mask = &value->misc;
+
+	DR_STE_SET_TAG(eth_l2_src, bit_mask, first_vlan_id, mask, first_vid);
+	DR_STE_SET_TAG(eth_l2_src, bit_mask, first_cfi, mask, first_cfi);
+	DR_STE_SET_TAG(eth_l2_src, bit_mask, first_priority, mask, first_prio);
+	DR_STE_SET_TAG(eth_l2_src, bit_mask, ip_fragmented, mask, frag);
+	DR_STE_SET_TAG(eth_l2_src, bit_mask, l3_ethertype, mask, ethertype);
+	DR_STE_SET_ONES(eth_l2_src, bit_mask, l3_type, mask, ip_version);
+
+	if (mask->svlan_tag || mask->cvlan_tag) {
+		MLX5_SET(ste_eth_l2_src, bit_mask, first_vlan_qualifier, -1);
+		mask->cvlan_tag = 0;
+		mask->svlan_tag = 0;
+	}
+
+	if (inner) {
+		if (misc_mask->inner_second_cvlan_tag ||
+		    misc_mask->inner_second_svlan_tag) {
+			MLX5_SET(ste_eth_l2_src, bit_mask, second_vlan_qualifier, -1);
+			misc_mask->inner_second_cvlan_tag = 0;
+			misc_mask->inner_second_svlan_tag = 0;
+		}
+
+		DR_STE_SET_TAG(eth_l2_src, bit_mask,
+			       second_vlan_id, misc_mask, inner_second_vid);
+		DR_STE_SET_TAG(eth_l2_src, bit_mask,
+			       second_cfi, misc_mask, inner_second_cfi);
+		DR_STE_SET_TAG(eth_l2_src, bit_mask,
+			       second_priority, misc_mask, inner_second_prio);
+	} else {
+		if (misc_mask->outer_second_cvlan_tag ||
+		    misc_mask->outer_second_svlan_tag) {
+			MLX5_SET(ste_eth_l2_src, bit_mask, second_vlan_qualifier, -1);
+			misc_mask->outer_second_cvlan_tag = 0;
+			misc_mask->outer_second_svlan_tag = 0;
+		}
+
+		DR_STE_SET_TAG(eth_l2_src, bit_mask,
+			       second_vlan_id, misc_mask, outer_second_vid);
+		DR_STE_SET_TAG(eth_l2_src, bit_mask,
+			       second_cfi, misc_mask, outer_second_cfi);
+		DR_STE_SET_TAG(eth_l2_src, bit_mask,
+			       second_priority, misc_mask, outer_second_prio);
+	}
+}
+
+static int
+dr_ste_v0_build_eth_l2_src_or_dst_tag(struct mlx5dr_match_param *value,
+				      bool inner, u8 *tag)
+{
+	struct mlx5dr_match_spec *spec = inner ? &value->inner : &value->outer;
+	struct mlx5dr_match_misc *misc_spec = &value->misc;
+
+	DR_STE_SET_TAG(eth_l2_src, tag, first_vlan_id, spec, first_vid);
+	DR_STE_SET_TAG(eth_l2_src, tag, first_cfi, spec, first_cfi);
+	DR_STE_SET_TAG(eth_l2_src, tag, first_priority, spec, first_prio);
+	DR_STE_SET_TAG(eth_l2_src, tag, ip_fragmented, spec, frag);
+	DR_STE_SET_TAG(eth_l2_src, tag, l3_ethertype, spec, ethertype);
+
+	if (spec->ip_version) {
+		if (spec->ip_version == IP_VERSION_IPV4) {
+			MLX5_SET(ste_eth_l2_src, tag, l3_type, STE_IPV4);
+			spec->ip_version = 0;
+		} else if (spec->ip_version == IP_VERSION_IPV6) {
+			MLX5_SET(ste_eth_l2_src, tag, l3_type, STE_IPV6);
+			spec->ip_version = 0;
+		} else {
+			return -EINVAL;
+		}
+	}
+
+	if (spec->cvlan_tag) {
+		MLX5_SET(ste_eth_l2_src, tag, first_vlan_qualifier, DR_STE_CVLAN);
+		spec->cvlan_tag = 0;
+	} else if (spec->svlan_tag) {
+		MLX5_SET(ste_eth_l2_src, tag, first_vlan_qualifier, DR_STE_SVLAN);
+		spec->svlan_tag = 0;
+	}
+
+	if (inner) {
+		if (misc_spec->inner_second_cvlan_tag) {
+			MLX5_SET(ste_eth_l2_src, tag, second_vlan_qualifier, DR_STE_CVLAN);
+			misc_spec->inner_second_cvlan_tag = 0;
+		} else if (misc_spec->inner_second_svlan_tag) {
+			MLX5_SET(ste_eth_l2_src, tag, second_vlan_qualifier, DR_STE_SVLAN);
+			misc_spec->inner_second_svlan_tag = 0;
+		}
+
+		DR_STE_SET_TAG(eth_l2_src, tag, second_vlan_id, misc_spec, inner_second_vid);
+		DR_STE_SET_TAG(eth_l2_src, tag, second_cfi, misc_spec, inner_second_cfi);
+		DR_STE_SET_TAG(eth_l2_src, tag, second_priority, misc_spec, inner_second_prio);
+	} else {
+		if (misc_spec->outer_second_cvlan_tag) {
+			MLX5_SET(ste_eth_l2_src, tag, second_vlan_qualifier, DR_STE_CVLAN);
+			misc_spec->outer_second_cvlan_tag = 0;
+		} else if (misc_spec->outer_second_svlan_tag) {
+			MLX5_SET(ste_eth_l2_src, tag, second_vlan_qualifier, DR_STE_SVLAN);
+			misc_spec->outer_second_svlan_tag = 0;
+		}
+		DR_STE_SET_TAG(eth_l2_src, tag, second_vlan_id, misc_spec, outer_second_vid);
+		DR_STE_SET_TAG(eth_l2_src, tag, second_cfi, misc_spec, outer_second_cfi);
+		DR_STE_SET_TAG(eth_l2_src, tag, second_priority, misc_spec, outer_second_prio);
+	}
+
+	return 0;
+}
+
+static void
+dr_ste_v0_build_eth_l2_src_bit_mask(struct mlx5dr_match_param *value,
+				    bool inner, u8 *bit_mask)
+{
+	struct mlx5dr_match_spec *mask = inner ? &value->inner : &value->outer;
+
+	DR_STE_SET_TAG(eth_l2_src, bit_mask, smac_47_16, mask, smac_47_16);
+	DR_STE_SET_TAG(eth_l2_src, bit_mask, smac_15_0, mask, smac_15_0);
+
+	dr_ste_v0_build_eth_l2_src_or_dst_bit_mask(value, inner, bit_mask);
+}
+
+static int
+dr_ste_v0_build_eth_l2_src_tag(struct mlx5dr_match_param *value,
+			       struct mlx5dr_ste_build *sb,
+			       u8 *tag)
+{
+	struct mlx5dr_match_spec *spec = sb->inner ? &value->inner : &value->outer;
+
+	DR_STE_SET_TAG(eth_l2_src, tag, smac_47_16, spec, smac_47_16);
+	DR_STE_SET_TAG(eth_l2_src, tag, smac_15_0, spec, smac_15_0);
+
+	return dr_ste_v0_build_eth_l2_src_or_dst_tag(value, sb->inner, tag);
+}
+
+static void
+dr_ste_v0_build_eth_l2_src_init(struct mlx5dr_ste_build *sb,
+				struct mlx5dr_match_param *mask)
+{
+	dr_ste_v0_build_eth_l2_src_bit_mask(mask, sb->inner, sb->bit_mask);
+	sb->lu_type = DR_STE_CALC_LU_TYPE(ETHL2_SRC, sb->rx, sb->inner);
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v0_build_eth_l2_src_tag;
+}
+
+static void
+dr_ste_v0_build_eth_l2_dst_bit_mask(struct mlx5dr_match_param *value,
+				    struct mlx5dr_ste_build *sb,
+				    u8 *bit_mask)
+{
+	struct mlx5dr_match_spec *mask = sb->inner ? &value->inner : &value->outer;
+
+	DR_STE_SET_TAG(eth_l2_dst, bit_mask, dmac_47_16, mask, dmac_47_16);
+	DR_STE_SET_TAG(eth_l2_dst, bit_mask, dmac_15_0, mask, dmac_15_0);
+
+	dr_ste_v0_build_eth_l2_src_or_dst_bit_mask(value, sb->inner, bit_mask);
+}
+
+static int
+dr_ste_v0_build_eth_l2_dst_tag(struct mlx5dr_match_param *value,
+			       struct mlx5dr_ste_build *sb,
+			       u8 *tag)
+{
+	struct mlx5dr_match_spec *spec = sb->inner ? &value->inner : &value->outer;
+
+	DR_STE_SET_TAG(eth_l2_dst, tag, dmac_47_16, spec, dmac_47_16);
+	DR_STE_SET_TAG(eth_l2_dst, tag, dmac_15_0, spec, dmac_15_0);
+
+	return dr_ste_v0_build_eth_l2_src_or_dst_tag(value, sb->inner, tag);
+}
+
+static void
+dr_ste_v0_build_eth_l2_dst_init(struct mlx5dr_ste_build *sb,
+				struct mlx5dr_match_param *mask)
+{
+	dr_ste_v0_build_eth_l2_dst_bit_mask(mask, sb, sb->bit_mask);
+
+	sb->lu_type = DR_STE_CALC_LU_TYPE(ETHL2_DST, sb->rx, sb->inner);
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v0_build_eth_l2_dst_tag;
+}
+
+static void
+dr_ste_v0_build_eth_l2_tnl_bit_mask(struct mlx5dr_match_param *value,
+				    bool inner, u8 *bit_mask)
+{
+	struct mlx5dr_match_spec *mask = inner ? &value->inner : &value->outer;
+	struct mlx5dr_match_misc *misc = &value->misc;
+
+	DR_STE_SET_TAG(eth_l2_tnl, bit_mask, dmac_47_16, mask, dmac_47_16);
+	DR_STE_SET_TAG(eth_l2_tnl, bit_mask, dmac_15_0, mask, dmac_15_0);
+	DR_STE_SET_TAG(eth_l2_tnl, bit_mask, first_vlan_id, mask, first_vid);
+	DR_STE_SET_TAG(eth_l2_tnl, bit_mask, first_cfi, mask, first_cfi);
+	DR_STE_SET_TAG(eth_l2_tnl, bit_mask, first_priority, mask, first_prio);
+	DR_STE_SET_TAG(eth_l2_tnl, bit_mask, ip_fragmented, mask, frag);
+	DR_STE_SET_TAG(eth_l2_tnl, bit_mask, l3_ethertype, mask, ethertype);
+	DR_STE_SET_ONES(eth_l2_tnl, bit_mask, l3_type, mask, ip_version);
+
+	if (misc->vxlan_vni) {
+		MLX5_SET(ste_eth_l2_tnl, bit_mask,
+			 l2_tunneling_network_id, (misc->vxlan_vni << 8));
+		misc->vxlan_vni = 0;
+	}
+
+	if (mask->svlan_tag || mask->cvlan_tag) {
+		MLX5_SET(ste_eth_l2_tnl, bit_mask, first_vlan_qualifier, -1);
+		mask->cvlan_tag = 0;
+		mask->svlan_tag = 0;
+	}
+}
+
+static int
+dr_ste_v0_build_eth_l2_tnl_tag(struct mlx5dr_match_param *value,
+			       struct mlx5dr_ste_build *sb,
+			       u8 *tag)
+{
+	struct mlx5dr_match_spec *spec = sb->inner ? &value->inner : &value->outer;
+	struct mlx5dr_match_misc *misc = &value->misc;
+
+	DR_STE_SET_TAG(eth_l2_tnl, tag, dmac_47_16, spec, dmac_47_16);
+	DR_STE_SET_TAG(eth_l2_tnl, tag, dmac_15_0, spec, dmac_15_0);
+	DR_STE_SET_TAG(eth_l2_tnl, tag, first_vlan_id, spec, first_vid);
+	DR_STE_SET_TAG(eth_l2_tnl, tag, first_cfi, spec, first_cfi);
+	DR_STE_SET_TAG(eth_l2_tnl, tag, ip_fragmented, spec, frag);
+	DR_STE_SET_TAG(eth_l2_tnl, tag, first_priority, spec, first_prio);
+	DR_STE_SET_TAG(eth_l2_tnl, tag, l3_ethertype, spec, ethertype);
+
+	if (misc->vxlan_vni) {
+		MLX5_SET(ste_eth_l2_tnl, tag, l2_tunneling_network_id,
+			 (misc->vxlan_vni << 8));
+		misc->vxlan_vni = 0;
+	}
+
+	if (spec->cvlan_tag) {
+		MLX5_SET(ste_eth_l2_tnl, tag, first_vlan_qualifier, DR_STE_CVLAN);
+		spec->cvlan_tag = 0;
+	} else if (spec->svlan_tag) {
+		MLX5_SET(ste_eth_l2_tnl, tag, first_vlan_qualifier, DR_STE_SVLAN);
+		spec->svlan_tag = 0;
+	}
+
+	if (spec->ip_version) {
+		if (spec->ip_version == IP_VERSION_IPV4) {
+			MLX5_SET(ste_eth_l2_tnl, tag, l3_type, STE_IPV4);
+			spec->ip_version = 0;
+		} else if (spec->ip_version == IP_VERSION_IPV6) {
+			MLX5_SET(ste_eth_l2_tnl, tag, l3_type, STE_IPV6);
+			spec->ip_version = 0;
+		} else {
+			return -EINVAL;
+		}
+	}
+
+	return 0;
+}
+
+static void
+dr_ste_v0_build_eth_l2_tnl_init(struct mlx5dr_ste_build *sb,
+				struct mlx5dr_match_param *mask)
+{
+	dr_ste_v0_build_eth_l2_tnl_bit_mask(mask, sb->inner, sb->bit_mask);
+
+	sb->lu_type = DR_STE_V0_LU_TYPE_ETHL2_TUNNELING_I;
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v0_build_eth_l2_tnl_tag;
+}
+
+static int
+dr_ste_v0_build_eth_l3_ipv4_misc_tag(struct mlx5dr_match_param *value,
+				     struct mlx5dr_ste_build *sb,
+				     u8 *tag)
+{
+	struct mlx5dr_match_spec *spec = sb->inner ? &value->inner : &value->outer;
+
+	DR_STE_SET_TAG(eth_l3_ipv4_misc, tag, time_to_live, spec, ttl_hoplimit);
+	DR_STE_SET_TAG(eth_l3_ipv4_misc, tag, ihl, spec, ipv4_ihl);
+
+	return 0;
+}
+
+static void
+dr_ste_v0_build_eth_l3_ipv4_misc_init(struct mlx5dr_ste_build *sb,
+				      struct mlx5dr_match_param *mask)
+{
+	dr_ste_v0_build_eth_l3_ipv4_misc_tag(mask, sb, sb->bit_mask);
+
+	sb->lu_type = DR_STE_CALC_LU_TYPE(ETHL3_IPV4_MISC, sb->rx, sb->inner);
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v0_build_eth_l3_ipv4_misc_tag;
+}
+
+static int
+dr_ste_v0_build_eth_ipv6_l3_l4_tag(struct mlx5dr_match_param *value,
+				   struct mlx5dr_ste_build *sb,
+				   u8 *tag)
+{
+	struct mlx5dr_match_spec *spec = sb->inner ? &value->inner : &value->outer;
+	struct mlx5dr_match_misc *misc = &value->misc;
+
+	DR_STE_SET_TAG(eth_l4, tag, dst_port, spec, tcp_dport);
+	DR_STE_SET_TAG(eth_l4, tag, src_port, spec, tcp_sport);
+	DR_STE_SET_TAG(eth_l4, tag, dst_port, spec, udp_dport);
+	DR_STE_SET_TAG(eth_l4, tag, src_port, spec, udp_sport);
+	DR_STE_SET_TAG(eth_l4, tag, protocol, spec, ip_protocol);
+	DR_STE_SET_TAG(eth_l4, tag, fragmented, spec, frag);
+	DR_STE_SET_TAG(eth_l4, tag, dscp, spec, ip_dscp);
+	DR_STE_SET_TAG(eth_l4, tag, ecn, spec, ip_ecn);
+	DR_STE_SET_TAG(eth_l4, tag, ipv6_hop_limit, spec, ttl_hoplimit);
+
+	if (sb->inner)
+		DR_STE_SET_TAG(eth_l4, tag, flow_label, misc, inner_ipv6_flow_label);
+	else
+		DR_STE_SET_TAG(eth_l4, tag, flow_label, misc, outer_ipv6_flow_label);
+
+	if (spec->tcp_flags) {
+		DR_STE_SET_TCP_FLAGS(eth_l4, tag, spec);
+		spec->tcp_flags = 0;
+	}
+
+	return 0;
+}
+
+static void
+dr_ste_v0_build_eth_ipv6_l3_l4_init(struct mlx5dr_ste_build *sb,
+				    struct mlx5dr_match_param *mask)
+{
+	dr_ste_v0_build_eth_ipv6_l3_l4_tag(mask, sb, sb->bit_mask);
+
+	sb->lu_type = DR_STE_CALC_LU_TYPE(ETHL4, sb->rx, sb->inner);
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v0_build_eth_ipv6_l3_l4_tag;
+}
+
+static int
+dr_ste_v0_build_mpls_tag(struct mlx5dr_match_param *value,
+			 struct mlx5dr_ste_build *sb,
+			 u8 *tag)
+{
+	struct mlx5dr_match_misc2 *misc2 = &value->misc2;
+
+	if (sb->inner)
+		DR_STE_SET_MPLS(mpls, misc2, inner, tag);
+	else
+		DR_STE_SET_MPLS(mpls, misc2, outer, tag);
+
+	return 0;
+}
+
+static void
+dr_ste_v0_build_mpls_init(struct mlx5dr_ste_build *sb,
+			  struct mlx5dr_match_param *mask)
+{
+	dr_ste_v0_build_mpls_tag(mask, sb, sb->bit_mask);
+
+	sb->lu_type = DR_STE_CALC_LU_TYPE(MPLS_FIRST, sb->rx, sb->inner);
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v0_build_mpls_tag;
+}
+
+static int
+dr_ste_v0_build_tnl_gre_tag(struct mlx5dr_match_param *value,
+			    struct mlx5dr_ste_build *sb,
+			    u8 *tag)
+{
+	struct  mlx5dr_match_misc *misc = &value->misc;
+
+	DR_STE_SET_TAG(gre, tag, gre_protocol, misc, gre_protocol);
+
+	DR_STE_SET_TAG(gre, tag, gre_k_present, misc, gre_k_present);
+	DR_STE_SET_TAG(gre, tag, gre_key_h, misc, gre_key_h);
+	DR_STE_SET_TAG(gre, tag, gre_key_l, misc, gre_key_l);
+
+	DR_STE_SET_TAG(gre, tag, gre_c_present, misc, gre_c_present);
+
+	DR_STE_SET_TAG(gre, tag, gre_s_present, misc, gre_s_present);
+
+	return 0;
+}
+
+static void
+dr_ste_v0_build_tnl_gre_init(struct mlx5dr_ste_build *sb,
+			     struct mlx5dr_match_param *mask)
+{
+	dr_ste_v0_build_tnl_gre_tag(mask, sb, sb->bit_mask);
+
+	sb->lu_type = DR_STE_V0_LU_TYPE_GRE;
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v0_build_tnl_gre_tag;
+}
+
+static int
+dr_ste_v0_build_tnl_mpls_tag(struct mlx5dr_match_param *value,
+			     struct mlx5dr_ste_build *sb,
+			     u8 *tag)
+{
+	struct mlx5dr_match_misc2 *misc_2 = &value->misc2;
+	u32 mpls_hdr;
+
+	if (DR_STE_IS_OUTER_MPLS_OVER_GRE_SET(misc_2)) {
+		mpls_hdr = misc_2->outer_first_mpls_over_gre_label << HDR_MPLS_OFFSET_LABEL;
+		misc_2->outer_first_mpls_over_gre_label = 0;
+		mpls_hdr |= misc_2->outer_first_mpls_over_gre_exp << HDR_MPLS_OFFSET_EXP;
+		misc_2->outer_first_mpls_over_gre_exp = 0;
+		mpls_hdr |= misc_2->outer_first_mpls_over_gre_s_bos << HDR_MPLS_OFFSET_S_BOS;
+		misc_2->outer_first_mpls_over_gre_s_bos = 0;
+		mpls_hdr |= misc_2->outer_first_mpls_over_gre_ttl << HDR_MPLS_OFFSET_TTL;
+		misc_2->outer_first_mpls_over_gre_ttl = 0;
+	} else {
+		mpls_hdr = misc_2->outer_first_mpls_over_udp_label << HDR_MPLS_OFFSET_LABEL;
+		misc_2->outer_first_mpls_over_udp_label = 0;
+		mpls_hdr |= misc_2->outer_first_mpls_over_udp_exp << HDR_MPLS_OFFSET_EXP;
+		misc_2->outer_first_mpls_over_udp_exp = 0;
+		mpls_hdr |= misc_2->outer_first_mpls_over_udp_s_bos << HDR_MPLS_OFFSET_S_BOS;
+		misc_2->outer_first_mpls_over_udp_s_bos = 0;
+		mpls_hdr |= misc_2->outer_first_mpls_over_udp_ttl << HDR_MPLS_OFFSET_TTL;
+		misc_2->outer_first_mpls_over_udp_ttl = 0;
+	}
+
+	MLX5_SET(ste_flex_parser_0, tag, flex_parser_3, mpls_hdr);
+	return 0;
+}
+
+static void
+dr_ste_v0_build_tnl_mpls_init(struct mlx5dr_ste_build *sb,
+			      struct mlx5dr_match_param *mask)
+{
+	dr_ste_v0_build_tnl_mpls_tag(mask, sb, sb->bit_mask);
+
+	sb->lu_type = DR_STE_V0_LU_TYPE_FLEX_PARSER_0;
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v0_build_tnl_mpls_tag;
+}
+
+static int
+dr_ste_v0_build_tnl_mpls_over_udp_tag(struct mlx5dr_match_param *value,
+				      struct mlx5dr_ste_build *sb,
+				      u8 *tag)
+{
+	struct mlx5dr_match_misc2 *misc2 = &value->misc2;
+	u8 *parser_ptr;
+	u8 parser_id;
+	u32 mpls_hdr;
+
+	mpls_hdr = misc2->outer_first_mpls_over_udp_label << HDR_MPLS_OFFSET_LABEL;
+	misc2->outer_first_mpls_over_udp_label = 0;
+	mpls_hdr |= misc2->outer_first_mpls_over_udp_exp << HDR_MPLS_OFFSET_EXP;
+	misc2->outer_first_mpls_over_udp_exp = 0;
+	mpls_hdr |= misc2->outer_first_mpls_over_udp_s_bos << HDR_MPLS_OFFSET_S_BOS;
+	misc2->outer_first_mpls_over_udp_s_bos = 0;
+	mpls_hdr |= misc2->outer_first_mpls_over_udp_ttl << HDR_MPLS_OFFSET_TTL;
+	misc2->outer_first_mpls_over_udp_ttl = 0;
+
+	parser_id = sb->caps->flex_parser_id_mpls_over_udp;
+	parser_ptr = dr_ste_calc_flex_parser_offset(tag, parser_id);
+	*(__be32 *)parser_ptr = cpu_to_be32(mpls_hdr);
+
+	return 0;
+}
+
+static void
+dr_ste_v0_build_tnl_mpls_over_udp_init(struct mlx5dr_ste_build *sb,
+				       struct mlx5dr_match_param *mask)
+{
+	dr_ste_v0_build_tnl_mpls_over_udp_tag(mask, sb, sb->bit_mask);
+	/* STEs with lookup type FLEX_PARSER_{0/1} includes
+	 * flex parsers_{0-3}/{4-7} respectively.
+	 */
+	sb->lu_type = sb->caps->flex_parser_id_mpls_over_udp > DR_STE_MAX_FLEX_0_ID ?
+		      DR_STE_V0_LU_TYPE_FLEX_PARSER_1 :
+		      DR_STE_V0_LU_TYPE_FLEX_PARSER_0;
+
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v0_build_tnl_mpls_over_udp_tag;
+}
+
+static int
+dr_ste_v0_build_tnl_mpls_over_gre_tag(struct mlx5dr_match_param *value,
+				      struct mlx5dr_ste_build *sb,
+				      u8 *tag)
+{
+	struct mlx5dr_match_misc2 *misc2 = &value->misc2;
+	u8 *parser_ptr;
+	u8 parser_id;
+	u32 mpls_hdr;
+
+	mpls_hdr = misc2->outer_first_mpls_over_gre_label << HDR_MPLS_OFFSET_LABEL;
+	misc2->outer_first_mpls_over_gre_label = 0;
+	mpls_hdr |= misc2->outer_first_mpls_over_gre_exp << HDR_MPLS_OFFSET_EXP;
+	misc2->outer_first_mpls_over_gre_exp = 0;
+	mpls_hdr |= misc2->outer_first_mpls_over_gre_s_bos << HDR_MPLS_OFFSET_S_BOS;
+	misc2->outer_first_mpls_over_gre_s_bos = 0;
+	mpls_hdr |= misc2->outer_first_mpls_over_gre_ttl << HDR_MPLS_OFFSET_TTL;
+	misc2->outer_first_mpls_over_gre_ttl = 0;
+
+	parser_id = sb->caps->flex_parser_id_mpls_over_gre;
+	parser_ptr = dr_ste_calc_flex_parser_offset(tag, parser_id);
+	*(__be32 *)parser_ptr = cpu_to_be32(mpls_hdr);
+
+	return 0;
+}
+
+static void
+dr_ste_v0_build_tnl_mpls_over_gre_init(struct mlx5dr_ste_build *sb,
+				       struct mlx5dr_match_param *mask)
+{
+	dr_ste_v0_build_tnl_mpls_over_gre_tag(mask, sb, sb->bit_mask);
+
+	/* STEs with lookup type FLEX_PARSER_{0/1} includes
+	 * flex parsers_{0-3}/{4-7} respectively.
+	 */
+	sb->lu_type = sb->caps->flex_parser_id_mpls_over_gre > DR_STE_MAX_FLEX_0_ID ?
+		      DR_STE_V0_LU_TYPE_FLEX_PARSER_1 :
+		      DR_STE_V0_LU_TYPE_FLEX_PARSER_0;
+
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v0_build_tnl_mpls_over_gre_tag;
+}
+
+#define ICMP_TYPE_OFFSET_FIRST_DW	24
+#define ICMP_CODE_OFFSET_FIRST_DW	16
+
+static int
+dr_ste_v0_build_icmp_tag(struct mlx5dr_match_param *value,
+			 struct mlx5dr_ste_build *sb,
+			 u8 *tag)
+{
+	struct mlx5dr_match_misc3 *misc_3 = &value->misc3;
+	u32 *icmp_header_data;
+	int dw0_location;
+	int dw1_location;
+	u8 *parser_ptr;
+	u8 *icmp_type;
+	u8 *icmp_code;
+	bool is_ipv4;
+	u32 icmp_hdr;
+
+	is_ipv4 = DR_MASK_IS_ICMPV4_SET(misc_3);
+	if (is_ipv4) {
+		icmp_header_data	= &misc_3->icmpv4_header_data;
+		icmp_type		= &misc_3->icmpv4_type;
+		icmp_code		= &misc_3->icmpv4_code;
+		dw0_location		= sb->caps->flex_parser_id_icmp_dw0;
+		dw1_location		= sb->caps->flex_parser_id_icmp_dw1;
+	} else {
+		icmp_header_data	= &misc_3->icmpv6_header_data;
+		icmp_type		= &misc_3->icmpv6_type;
+		icmp_code		= &misc_3->icmpv6_code;
+		dw0_location		= sb->caps->flex_parser_id_icmpv6_dw0;
+		dw1_location		= sb->caps->flex_parser_id_icmpv6_dw1;
+	}
+
+	parser_ptr = dr_ste_calc_flex_parser_offset(tag, dw0_location);
+	icmp_hdr = (*icmp_type << ICMP_TYPE_OFFSET_FIRST_DW) |
+		   (*icmp_code << ICMP_CODE_OFFSET_FIRST_DW);
+	*(__be32 *)parser_ptr = cpu_to_be32(icmp_hdr);
+	*icmp_code = 0;
+	*icmp_type = 0;
+
+	parser_ptr = dr_ste_calc_flex_parser_offset(tag, dw1_location);
+	*(__be32 *)parser_ptr = cpu_to_be32(*icmp_header_data);
+	*icmp_header_data = 0;
+
+	return 0;
+}
+
+static void
+dr_ste_v0_build_icmp_init(struct mlx5dr_ste_build *sb,
+			  struct mlx5dr_match_param *mask)
+{
+	u8 parser_id;
+	bool is_ipv4;
+
+	dr_ste_v0_build_icmp_tag(mask, sb, sb->bit_mask);
+
+	/* STEs with lookup type FLEX_PARSER_{0/1} includes
+	 * flex parsers_{0-3}/{4-7} respectively.
+	 */
+	is_ipv4 = DR_MASK_IS_ICMPV4_SET(&mask->misc3);
+	parser_id = is_ipv4 ? sb->caps->flex_parser_id_icmp_dw0 :
+		    sb->caps->flex_parser_id_icmpv6_dw0;
+	sb->lu_type = parser_id > DR_STE_MAX_FLEX_0_ID ?
+		      DR_STE_V0_LU_TYPE_FLEX_PARSER_1 :
+		      DR_STE_V0_LU_TYPE_FLEX_PARSER_0;
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v0_build_icmp_tag;
+}
+
+static int
+dr_ste_v0_build_general_purpose_tag(struct mlx5dr_match_param *value,
+				    struct mlx5dr_ste_build *sb,
+				    u8 *tag)
+{
+	struct mlx5dr_match_misc2 *misc_2 = &value->misc2;
+
+	DR_STE_SET_TAG(general_purpose, tag, general_purpose_lookup_field,
+		       misc_2, metadata_reg_a);
+
+	return 0;
+}
+
+static void
+dr_ste_v0_build_general_purpose_init(struct mlx5dr_ste_build *sb,
+				     struct mlx5dr_match_param *mask)
+{
+	dr_ste_v0_build_general_purpose_tag(mask, sb, sb->bit_mask);
+
+	sb->lu_type = DR_STE_V0_LU_TYPE_GENERAL_PURPOSE;
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v0_build_general_purpose_tag;
+}
+
+static int
+dr_ste_v0_build_eth_l4_misc_tag(struct mlx5dr_match_param *value,
+				struct mlx5dr_ste_build *sb,
+				u8 *tag)
+{
+	struct mlx5dr_match_misc3 *misc3 = &value->misc3;
+
+	if (sb->inner) {
+		DR_STE_SET_TAG(eth_l4_misc, tag, seq_num, misc3, inner_tcp_seq_num);
+		DR_STE_SET_TAG(eth_l4_misc, tag, ack_num, misc3, inner_tcp_ack_num);
+	} else {
+		DR_STE_SET_TAG(eth_l4_misc, tag, seq_num, misc3, outer_tcp_seq_num);
+		DR_STE_SET_TAG(eth_l4_misc, tag, ack_num, misc3, outer_tcp_ack_num);
+	}
+
+	return 0;
+}
+
+static void
+dr_ste_v0_build_eth_l4_misc_init(struct mlx5dr_ste_build *sb,
+				 struct mlx5dr_match_param *mask)
+{
+	dr_ste_v0_build_eth_l4_misc_tag(mask, sb, sb->bit_mask);
+
+	sb->lu_type = DR_STE_CALC_LU_TYPE(ETHL4_MISC, sb->rx, sb->inner);
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v0_build_eth_l4_misc_tag;
+}
+
+static int
+dr_ste_v0_build_flex_parser_tnl_vxlan_gpe_tag(struct mlx5dr_match_param *value,
+					      struct mlx5dr_ste_build *sb,
+					      u8 *tag)
+{
+	struct mlx5dr_match_misc3 *misc3 = &value->misc3;
+
+	DR_STE_SET_TAG(flex_parser_tnl_vxlan_gpe, tag,
+		       outer_vxlan_gpe_flags, misc3,
+		       outer_vxlan_gpe_flags);
+	DR_STE_SET_TAG(flex_parser_tnl_vxlan_gpe, tag,
+		       outer_vxlan_gpe_next_protocol, misc3,
+		       outer_vxlan_gpe_next_protocol);
+	DR_STE_SET_TAG(flex_parser_tnl_vxlan_gpe, tag,
+		       outer_vxlan_gpe_vni, misc3,
+		       outer_vxlan_gpe_vni);
+
+	return 0;
+}
+
+static void
+dr_ste_v0_build_flex_parser_tnl_vxlan_gpe_init(struct mlx5dr_ste_build *sb,
+					       struct mlx5dr_match_param *mask)
+{
+	dr_ste_v0_build_flex_parser_tnl_vxlan_gpe_tag(mask, sb, sb->bit_mask);
+	sb->lu_type = DR_STE_V0_LU_TYPE_FLEX_PARSER_TNL_HEADER;
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v0_build_flex_parser_tnl_vxlan_gpe_tag;
+}
+
+static int
+dr_ste_v0_build_flex_parser_tnl_geneve_tag(struct mlx5dr_match_param *value,
+					   struct mlx5dr_ste_build *sb,
+					   u8 *tag)
+{
+	struct mlx5dr_match_misc *misc = &value->misc;
+
+	DR_STE_SET_TAG(flex_parser_tnl_geneve, tag,
+		       geneve_protocol_type, misc, geneve_protocol_type);
+	DR_STE_SET_TAG(flex_parser_tnl_geneve, tag,
+		       geneve_oam, misc, geneve_oam);
+	DR_STE_SET_TAG(flex_parser_tnl_geneve, tag,
+		       geneve_opt_len, misc, geneve_opt_len);
+	DR_STE_SET_TAG(flex_parser_tnl_geneve, tag,
+		       geneve_vni, misc, geneve_vni);
+
+	return 0;
+}
+
+static void
+dr_ste_v0_build_flex_parser_tnl_geneve_init(struct mlx5dr_ste_build *sb,
+					    struct mlx5dr_match_param *mask)
+{
+	dr_ste_v0_build_flex_parser_tnl_geneve_tag(mask, sb, sb->bit_mask);
+	sb->lu_type = DR_STE_V0_LU_TYPE_FLEX_PARSER_TNL_HEADER;
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v0_build_flex_parser_tnl_geneve_tag;
+}
+
+static int
+dr_ste_v0_build_register_0_tag(struct mlx5dr_match_param *value,
+			       struct mlx5dr_ste_build *sb,
+			       u8 *tag)
+{
+	struct mlx5dr_match_misc2 *misc2 = &value->misc2;
+
+	DR_STE_SET_TAG(register_0, tag, register_0_h, misc2, metadata_reg_c_0);
+	DR_STE_SET_TAG(register_0, tag, register_0_l, misc2, metadata_reg_c_1);
+	DR_STE_SET_TAG(register_0, tag, register_1_h, misc2, metadata_reg_c_2);
+	DR_STE_SET_TAG(register_0, tag, register_1_l, misc2, metadata_reg_c_3);
+
+	return 0;
+}
+
+static void
+dr_ste_v0_build_register_0_init(struct mlx5dr_ste_build *sb,
+				struct mlx5dr_match_param *mask)
+{
+	dr_ste_v0_build_register_0_tag(mask, sb, sb->bit_mask);
+
+	sb->lu_type = DR_STE_V0_LU_TYPE_STEERING_REGISTERS_0;
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v0_build_register_0_tag;
+}
+
+static int
+dr_ste_v0_build_register_1_tag(struct mlx5dr_match_param *value,
+			       struct mlx5dr_ste_build *sb,
+			       u8 *tag)
+{
+	struct mlx5dr_match_misc2 *misc2 = &value->misc2;
+
+	DR_STE_SET_TAG(register_1, tag, register_2_h, misc2, metadata_reg_c_4);
+	DR_STE_SET_TAG(register_1, tag, register_2_l, misc2, metadata_reg_c_5);
+	DR_STE_SET_TAG(register_1, tag, register_3_h, misc2, metadata_reg_c_6);
+	DR_STE_SET_TAG(register_1, tag, register_3_l, misc2, metadata_reg_c_7);
+
+	return 0;
+}
+
+static void
+dr_ste_v0_build_register_1_init(struct mlx5dr_ste_build *sb,
+				struct mlx5dr_match_param *mask)
+{
+	dr_ste_v0_build_register_1_tag(mask, sb, sb->bit_mask);
+
+	sb->lu_type = DR_STE_V0_LU_TYPE_STEERING_REGISTERS_1;
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v0_build_register_1_tag;
+}
+
+static void
+dr_ste_v0_build_src_gvmi_qpn_bit_mask(struct mlx5dr_match_param *value,
+				      u8 *bit_mask)
+{
+	struct mlx5dr_match_misc *misc_mask = &value->misc;
+
+	DR_STE_SET_ONES(src_gvmi_qp, bit_mask, source_gvmi, misc_mask, source_port);
+	DR_STE_SET_ONES(src_gvmi_qp, bit_mask, source_qp, misc_mask, source_sqn);
+	misc_mask->source_eswitch_owner_vhca_id = 0;
+}
+
+static int
+dr_ste_v0_build_src_gvmi_qpn_tag(struct mlx5dr_match_param *value,
+				 struct mlx5dr_ste_build *sb,
+				 u8 *tag)
+{
+	struct mlx5dr_match_misc *misc = &value->misc;
+	struct mlx5dr_cmd_vport_cap *vport_cap;
+	struct mlx5dr_domain *dmn = sb->dmn;
+	struct mlx5dr_domain *vport_dmn;
+	u8 *bit_mask = sb->bit_mask;
+	bool source_gvmi_set;
+
+	DR_STE_SET_TAG(src_gvmi_qp, tag, source_qp, misc, source_sqn);
+
+	if (sb->vhca_id_valid) {
+		/* Find port GVMI based on the eswitch_owner_vhca_id */
+		if (misc->source_eswitch_owner_vhca_id == dmn->info.caps.gvmi)
+			vport_dmn = dmn;
+		else if (dmn->peer_dmn && (misc->source_eswitch_owner_vhca_id ==
+					   dmn->peer_dmn->info.caps.gvmi))
+			vport_dmn = dmn->peer_dmn;
+		else
+			return -EINVAL;
+
+		misc->source_eswitch_owner_vhca_id = 0;
+	} else {
+		vport_dmn = dmn;
+	}
+
+	source_gvmi_set = MLX5_GET(ste_src_gvmi_qp, bit_mask, source_gvmi);
+	if (source_gvmi_set) {
+		vport_cap = mlx5dr_domain_get_vport_cap(vport_dmn,
+							misc->source_port);
+		if (!vport_cap) {
+			mlx5dr_err(dmn, "Vport 0x%x is disabled or invalid\n",
+				   misc->source_port);
+			return -EINVAL;
+		}
+
+		if (vport_cap->vport_gvmi)
+			MLX5_SET(ste_src_gvmi_qp, tag, source_gvmi, vport_cap->vport_gvmi);
+
+		misc->source_port = 0;
+	}
+
+	return 0;
+}
+
+static void
+dr_ste_v0_build_src_gvmi_qpn_init(struct mlx5dr_ste_build *sb,
+				  struct mlx5dr_match_param *mask)
+{
+	dr_ste_v0_build_src_gvmi_qpn_bit_mask(mask, sb->bit_mask);
+
+	sb->lu_type = DR_STE_V0_LU_TYPE_SRC_GVMI_AND_QP;
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v0_build_src_gvmi_qpn_tag;
+}
+
+static void dr_ste_v0_set_flex_parser(u32 *misc4_field_id,
+				      u32 *misc4_field_value,
+				      bool *parser_is_used,
+				      u8 *tag)
+{
+	u32 id = *misc4_field_id;
+	u8 *parser_ptr;
+
+	if (id >= DR_NUM_OF_FLEX_PARSERS || parser_is_used[id])
+		return;
+
+	parser_is_used[id] = true;
+	parser_ptr = dr_ste_calc_flex_parser_offset(tag, id);
+
+	*(__be32 *)parser_ptr = cpu_to_be32(*misc4_field_value);
+	*misc4_field_id = 0;
+	*misc4_field_value = 0;
+}
+
+static int dr_ste_v0_build_flex_parser_tag(struct mlx5dr_match_param *value,
+					   struct mlx5dr_ste_build *sb,
+					   u8 *tag)
+{
+	struct mlx5dr_match_misc4 *misc_4_mask = &value->misc4;
+	bool parser_is_used[DR_NUM_OF_FLEX_PARSERS] = {};
+
+	dr_ste_v0_set_flex_parser(&misc_4_mask->prog_sample_field_id_0,
+				  &misc_4_mask->prog_sample_field_value_0,
+				  parser_is_used, tag);
+
+	dr_ste_v0_set_flex_parser(&misc_4_mask->prog_sample_field_id_1,
+				  &misc_4_mask->prog_sample_field_value_1,
+				  parser_is_used, tag);
+
+	dr_ste_v0_set_flex_parser(&misc_4_mask->prog_sample_field_id_2,
+				  &misc_4_mask->prog_sample_field_value_2,
+				  parser_is_used, tag);
+
+	dr_ste_v0_set_flex_parser(&misc_4_mask->prog_sample_field_id_3,
+				  &misc_4_mask->prog_sample_field_value_3,
+				  parser_is_used, tag);
+
+	return 0;
+}
+
+static void dr_ste_v0_build_flex_parser_0_init(struct mlx5dr_ste_build *sb,
+					       struct mlx5dr_match_param *mask)
+{
+	sb->lu_type = DR_STE_V0_LU_TYPE_FLEX_PARSER_0;
+	dr_ste_v0_build_flex_parser_tag(mask, sb, sb->bit_mask);
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v0_build_flex_parser_tag;
+}
+
+static void dr_ste_v0_build_flex_parser_1_init(struct mlx5dr_ste_build *sb,
+					       struct mlx5dr_match_param *mask)
+{
+	sb->lu_type = DR_STE_V0_LU_TYPE_FLEX_PARSER_1;
+	dr_ste_v0_build_flex_parser_tag(mask, sb, sb->bit_mask);
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v0_build_flex_parser_tag;
+}
+
+static int
+dr_ste_v0_build_flex_parser_tnl_geneve_tlv_opt_tag(struct mlx5dr_match_param *value,
+						   struct mlx5dr_ste_build *sb,
+						   u8 *tag)
+{
+	struct mlx5dr_match_misc3 *misc3 = &value->misc3;
+	u8 parser_id = sb->caps->flex_parser_id_geneve_tlv_option_0;
+	u8 *parser_ptr = dr_ste_calc_flex_parser_offset(tag, parser_id);
+
+	MLX5_SET(ste_flex_parser_0, parser_ptr, flex_parser_3,
+		 misc3->geneve_tlv_option_0_data);
+	misc3->geneve_tlv_option_0_data = 0;
+
+	return 0;
+}
+
+static void
+dr_ste_v0_build_flex_parser_tnl_geneve_tlv_opt_init(struct mlx5dr_ste_build *sb,
+						    struct mlx5dr_match_param *mask)
+{
+	dr_ste_v0_build_flex_parser_tnl_geneve_tlv_opt_tag(mask, sb, sb->bit_mask);
+
+	/* STEs with lookup type FLEX_PARSER_{0/1} includes
+	 * flex parsers_{0-3}/{4-7} respectively.
+	 */
+	sb->lu_type = sb->caps->flex_parser_id_geneve_tlv_option_0 > 3 ?
+		DR_STE_V0_LU_TYPE_FLEX_PARSER_1 :
+		DR_STE_V0_LU_TYPE_FLEX_PARSER_0;
+
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v0_build_flex_parser_tnl_geneve_tlv_opt_tag;
+}
+
+static int dr_ste_v0_build_flex_parser_tnl_gtpu_tag(struct mlx5dr_match_param *value,
+						    struct mlx5dr_ste_build *sb,
+						    u8 *tag)
+{
+	struct mlx5dr_match_misc3 *misc3 = &value->misc3;
+
+	DR_STE_SET_TAG(flex_parser_tnl_gtpu, tag,
+		       gtpu_msg_flags, misc3,
+		       gtpu_msg_flags);
+	DR_STE_SET_TAG(flex_parser_tnl_gtpu, tag,
+		       gtpu_msg_type, misc3,
+		       gtpu_msg_type);
+	DR_STE_SET_TAG(flex_parser_tnl_gtpu, tag,
+		       gtpu_teid, misc3,
+		       gtpu_teid);
+
+	return 0;
+}
+
+static void dr_ste_v0_build_flex_parser_tnl_gtpu_init(struct mlx5dr_ste_build *sb,
+						      struct mlx5dr_match_param *mask)
+{
+	dr_ste_v0_build_flex_parser_tnl_gtpu_tag(mask, sb, sb->bit_mask);
+
+	sb->lu_type = DR_STE_V0_LU_TYPE_FLEX_PARSER_TNL_HEADER;
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v0_build_flex_parser_tnl_gtpu_tag;
+}
+
+static int
+dr_ste_v0_build_tnl_gtpu_flex_parser_0_tag(struct mlx5dr_match_param *value,
+					   struct mlx5dr_ste_build *sb,
+					   u8 *tag)
+{
+	if (dr_is_flex_parser_0_id(sb->caps->flex_parser_id_gtpu_dw_0))
+		DR_STE_SET_FLEX_PARSER_FIELD(tag, gtpu_dw_0, sb->caps, &value->misc3);
+	if (dr_is_flex_parser_0_id(sb->caps->flex_parser_id_gtpu_teid))
+		DR_STE_SET_FLEX_PARSER_FIELD(tag, gtpu_teid, sb->caps, &value->misc3);
+	if (dr_is_flex_parser_0_id(sb->caps->flex_parser_id_gtpu_dw_2))
+		DR_STE_SET_FLEX_PARSER_FIELD(tag, gtpu_dw_2, sb->caps, &value->misc3);
+	if (dr_is_flex_parser_0_id(sb->caps->flex_parser_id_gtpu_first_ext_dw_0))
+		DR_STE_SET_FLEX_PARSER_FIELD(tag, gtpu_first_ext_dw_0, sb->caps, &value->misc3);
+	return 0;
+}
+
+static void
+dr_ste_v0_build_tnl_gtpu_flex_parser_0_init(struct mlx5dr_ste_build *sb,
+					    struct mlx5dr_match_param *mask)
+{
+	dr_ste_v0_build_tnl_gtpu_flex_parser_0_tag(mask, sb, sb->bit_mask);
+
+	sb->lu_type = DR_STE_V0_LU_TYPE_FLEX_PARSER_0;
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v0_build_tnl_gtpu_flex_parser_0_tag;
+}
+
+static int
+dr_ste_v0_build_tnl_gtpu_flex_parser_1_tag(struct mlx5dr_match_param *value,
+					   struct mlx5dr_ste_build *sb,
+					   u8 *tag)
+{
+	if (dr_is_flex_parser_1_id(sb->caps->flex_parser_id_gtpu_dw_0))
+		DR_STE_SET_FLEX_PARSER_FIELD(tag, gtpu_dw_0, sb->caps, &value->misc3);
+	if (dr_is_flex_parser_1_id(sb->caps->flex_parser_id_gtpu_teid))
+		DR_STE_SET_FLEX_PARSER_FIELD(tag, gtpu_teid, sb->caps, &value->misc3);
+	if (dr_is_flex_parser_1_id(sb->caps->flex_parser_id_gtpu_dw_2))
+		DR_STE_SET_FLEX_PARSER_FIELD(tag, gtpu_dw_2, sb->caps, &value->misc3);
+	if (dr_is_flex_parser_1_id(sb->caps->flex_parser_id_gtpu_first_ext_dw_0))
+		DR_STE_SET_FLEX_PARSER_FIELD(tag, gtpu_first_ext_dw_0, sb->caps, &value->misc3);
+	return 0;
+}
+
+static void
+dr_ste_v0_build_tnl_gtpu_flex_parser_1_init(struct mlx5dr_ste_build *sb,
+					    struct mlx5dr_match_param *mask)
+{
+	dr_ste_v0_build_tnl_gtpu_flex_parser_1_tag(mask, sb, sb->bit_mask);
+
+	sb->lu_type = DR_STE_V0_LU_TYPE_FLEX_PARSER_1;
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v0_build_tnl_gtpu_flex_parser_1_tag;
+}
+
+static int dr_ste_v0_build_tnl_header_0_1_tag(struct mlx5dr_match_param *value,
+					      struct mlx5dr_ste_build *sb,
+					      uint8_t *tag)
+{
+	struct mlx5dr_match_misc5 *misc5 = &value->misc5;
+
+	DR_STE_SET_TAG(tunnel_header, tag, tunnel_header_0, misc5, tunnel_header_0);
+	DR_STE_SET_TAG(tunnel_header, tag, tunnel_header_1, misc5, tunnel_header_1);
+
+	return 0;
+}
+
+static void dr_ste_v0_build_tnl_header_0_1_init(struct mlx5dr_ste_build *sb,
+						struct mlx5dr_match_param *mask)
+{
+	sb->lu_type = DR_STE_V0_LU_TYPE_TUNNEL_HEADER;
+	dr_ste_v0_build_tnl_header_0_1_tag(mask, sb, sb->bit_mask);
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v0_build_tnl_header_0_1_tag;
+}
+
+static struct mlx5dr_ste_ctx ste_ctx_v0 = {
+	/* Builders */
+	.build_eth_l2_src_dst_init	= &dr_ste_v0_build_eth_l2_src_dst_init,
+	.build_eth_l3_ipv6_src_init	= &dr_ste_v0_build_eth_l3_ipv6_src_init,
+	.build_eth_l3_ipv6_dst_init	= &dr_ste_v0_build_eth_l3_ipv6_dst_init,
+	.build_eth_l3_ipv4_5_tuple_init	= &dr_ste_v0_build_eth_l3_ipv4_5_tuple_init,
+	.build_eth_l2_src_init		= &dr_ste_v0_build_eth_l2_src_init,
+	.build_eth_l2_dst_init		= &dr_ste_v0_build_eth_l2_dst_init,
+	.build_eth_l2_tnl_init		= &dr_ste_v0_build_eth_l2_tnl_init,
+	.build_eth_l3_ipv4_misc_init	= &dr_ste_v0_build_eth_l3_ipv4_misc_init,
+	.build_eth_ipv6_l3_l4_init	= &dr_ste_v0_build_eth_ipv6_l3_l4_init,
+	.build_mpls_init		= &dr_ste_v0_build_mpls_init,
+	.build_tnl_gre_init		= &dr_ste_v0_build_tnl_gre_init,
+	.build_tnl_mpls_init		= &dr_ste_v0_build_tnl_mpls_init,
+	.build_tnl_mpls_over_udp_init	= &dr_ste_v0_build_tnl_mpls_over_udp_init,
+	.build_tnl_mpls_over_gre_init	= &dr_ste_v0_build_tnl_mpls_over_gre_init,
+	.build_icmp_init		= &dr_ste_v0_build_icmp_init,
+	.build_general_purpose_init	= &dr_ste_v0_build_general_purpose_init,
+	.build_eth_l4_misc_init		= &dr_ste_v0_build_eth_l4_misc_init,
+	.build_tnl_vxlan_gpe_init	= &dr_ste_v0_build_flex_parser_tnl_vxlan_gpe_init,
+	.build_tnl_geneve_init		= &dr_ste_v0_build_flex_parser_tnl_geneve_init,
+	.build_tnl_geneve_tlv_opt_init	= &dr_ste_v0_build_flex_parser_tnl_geneve_tlv_opt_init,
+	.build_register_0_init		= &dr_ste_v0_build_register_0_init,
+	.build_register_1_init		= &dr_ste_v0_build_register_1_init,
+	.build_src_gvmi_qpn_init	= &dr_ste_v0_build_src_gvmi_qpn_init,
+	.build_flex_parser_0_init	= &dr_ste_v0_build_flex_parser_0_init,
+	.build_flex_parser_1_init	= &dr_ste_v0_build_flex_parser_1_init,
+	.build_tnl_gtpu_init		= &dr_ste_v0_build_flex_parser_tnl_gtpu_init,
+	.build_tnl_header_0_1_init	= &dr_ste_v0_build_tnl_header_0_1_init,
+	.build_tnl_gtpu_flex_parser_0_init   = &dr_ste_v0_build_tnl_gtpu_flex_parser_0_init,
+	.build_tnl_gtpu_flex_parser_1_init   = &dr_ste_v0_build_tnl_gtpu_flex_parser_1_init,
+
+	/* Getters and Setters */
+	.ste_init			= &dr_ste_v0_init,
+	.set_next_lu_type		= &dr_ste_v0_set_next_lu_type,
+	.get_next_lu_type		= &dr_ste_v0_get_next_lu_type,
+	.set_miss_addr			= &dr_ste_v0_set_miss_addr,
+	.get_miss_addr			= &dr_ste_v0_get_miss_addr,
+	.set_hit_addr			= &dr_ste_v0_set_hit_addr,
+	.set_byte_mask			= &dr_ste_v0_set_byte_mask,
+	.get_byte_mask			= &dr_ste_v0_get_byte_mask,
+
+	/* Actions */
+	.actions_caps			= DR_STE_CTX_ACTION_CAP_NONE,
+	.set_actions_rx			= &dr_ste_v0_set_actions_rx,
+	.set_actions_tx			= &dr_ste_v0_set_actions_tx,
+	.modify_field_arr_sz		= ARRAY_SIZE(dr_ste_v0_action_modify_field_arr),
+	.modify_field_arr		= dr_ste_v0_action_modify_field_arr,
+	.set_action_set			= &dr_ste_v0_set_action_set,
+	.set_action_add			= &dr_ste_v0_set_action_add,
+	.set_action_copy		= &dr_ste_v0_set_action_copy,
+	.set_action_decap_l3_list	= &dr_ste_v0_set_action_decap_l3_list,
+};
+
+struct mlx5dr_ste_ctx *mlx5dr_ste_get_ctx_v0(void)
+{
+	return &ste_ctx_v0;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_ste_v1.c b/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_ste_v1.c
new file mode 100644
index 000000000..ee677a5c7
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_ste_v1.c
@@ -0,0 +1,2172 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2020 NVIDIA CORPORATION. All rights reserved. */
+
+#include <linux/types.h>
+#include "mlx5_ifc_dr_ste_v1.h"
+#include "dr_ste_v1.h"
+
+#define DR_STE_CALC_DFNR_TYPE(lookup_type, inner) \
+	((inner) ? DR_STE_V1_LU_TYPE_##lookup_type##_I : \
+		   DR_STE_V1_LU_TYPE_##lookup_type##_O)
+
+enum dr_ste_v1_entry_format {
+	DR_STE_V1_TYPE_BWC_BYTE	= 0x0,
+	DR_STE_V1_TYPE_BWC_DW	= 0x1,
+	DR_STE_V1_TYPE_MATCH	= 0x2,
+};
+
+/* Lookup type is built from 2B: [ Definer mode 1B ][ Definer index 1B ] */
+enum {
+	DR_STE_V1_LU_TYPE_NOP				= 0x0000,
+	DR_STE_V1_LU_TYPE_ETHL2_TNL			= 0x0002,
+	DR_STE_V1_LU_TYPE_IBL3_EXT			= 0x0102,
+	DR_STE_V1_LU_TYPE_ETHL2_O			= 0x0003,
+	DR_STE_V1_LU_TYPE_IBL4				= 0x0103,
+	DR_STE_V1_LU_TYPE_ETHL2_I			= 0x0004,
+	DR_STE_V1_LU_TYPE_SRC_QP_GVMI			= 0x0104,
+	DR_STE_V1_LU_TYPE_ETHL2_SRC_O			= 0x0005,
+	DR_STE_V1_LU_TYPE_ETHL2_HEADERS_O		= 0x0105,
+	DR_STE_V1_LU_TYPE_ETHL2_SRC_I			= 0x0006,
+	DR_STE_V1_LU_TYPE_ETHL2_HEADERS_I		= 0x0106,
+	DR_STE_V1_LU_TYPE_ETHL3_IPV4_5_TUPLE_O		= 0x0007,
+	DR_STE_V1_LU_TYPE_IPV6_DES_O			= 0x0107,
+	DR_STE_V1_LU_TYPE_ETHL3_IPV4_5_TUPLE_I		= 0x0008,
+	DR_STE_V1_LU_TYPE_IPV6_DES_I			= 0x0108,
+	DR_STE_V1_LU_TYPE_ETHL4_O			= 0x0009,
+	DR_STE_V1_LU_TYPE_IPV6_SRC_O			= 0x0109,
+	DR_STE_V1_LU_TYPE_ETHL4_I			= 0x000a,
+	DR_STE_V1_LU_TYPE_IPV6_SRC_I			= 0x010a,
+	DR_STE_V1_LU_TYPE_ETHL2_SRC_DST_O		= 0x000b,
+	DR_STE_V1_LU_TYPE_MPLS_O			= 0x010b,
+	DR_STE_V1_LU_TYPE_ETHL2_SRC_DST_I		= 0x000c,
+	DR_STE_V1_LU_TYPE_MPLS_I			= 0x010c,
+	DR_STE_V1_LU_TYPE_ETHL3_IPV4_MISC_O		= 0x000d,
+	DR_STE_V1_LU_TYPE_GRE				= 0x010d,
+	DR_STE_V1_LU_TYPE_FLEX_PARSER_TNL_HEADER	= 0x000e,
+	DR_STE_V1_LU_TYPE_GENERAL_PURPOSE		= 0x010e,
+	DR_STE_V1_LU_TYPE_ETHL3_IPV4_MISC_I		= 0x000f,
+	DR_STE_V1_LU_TYPE_STEERING_REGISTERS_0		= 0x010f,
+	DR_STE_V1_LU_TYPE_STEERING_REGISTERS_1		= 0x0110,
+	DR_STE_V1_LU_TYPE_FLEX_PARSER_OK		= 0x0011,
+	DR_STE_V1_LU_TYPE_FLEX_PARSER_0			= 0x0111,
+	DR_STE_V1_LU_TYPE_FLEX_PARSER_1			= 0x0112,
+	DR_STE_V1_LU_TYPE_ETHL4_MISC_O			= 0x0113,
+	DR_STE_V1_LU_TYPE_ETHL4_MISC_I			= 0x0114,
+	DR_STE_V1_LU_TYPE_INVALID			= 0x00ff,
+	DR_STE_V1_LU_TYPE_DONT_CARE			= MLX5DR_STE_LU_TYPE_DONT_CARE,
+};
+
+enum dr_ste_v1_header_anchors {
+	DR_STE_HEADER_ANCHOR_START_OUTER		= 0x00,
+	DR_STE_HEADER_ANCHOR_1ST_VLAN			= 0x02,
+	DR_STE_HEADER_ANCHOR_IPV6_IPV4			= 0x07,
+	DR_STE_HEADER_ANCHOR_INNER_MAC			= 0x13,
+	DR_STE_HEADER_ANCHOR_INNER_IPV6_IPV4		= 0x19,
+};
+
+enum dr_ste_v1_action_size {
+	DR_STE_ACTION_SINGLE_SZ = 4,
+	DR_STE_ACTION_DOUBLE_SZ = 8,
+	DR_STE_ACTION_TRIPLE_SZ = 12,
+};
+
+enum dr_ste_v1_action_insert_ptr_attr {
+	DR_STE_V1_ACTION_INSERT_PTR_ATTR_NONE = 0,  /* Regular push header (e.g. push vlan) */
+	DR_STE_V1_ACTION_INSERT_PTR_ATTR_ENCAP = 1, /* Encapsulation / Tunneling */
+	DR_STE_V1_ACTION_INSERT_PTR_ATTR_ESP = 2,   /* IPsec */
+};
+
+enum dr_ste_v1_action_id {
+	DR_STE_V1_ACTION_ID_NOP				= 0x00,
+	DR_STE_V1_ACTION_ID_COPY			= 0x05,
+	DR_STE_V1_ACTION_ID_SET				= 0x06,
+	DR_STE_V1_ACTION_ID_ADD				= 0x07,
+	DR_STE_V1_ACTION_ID_REMOVE_BY_SIZE		= 0x08,
+	DR_STE_V1_ACTION_ID_REMOVE_HEADER_TO_HEADER	= 0x09,
+	DR_STE_V1_ACTION_ID_INSERT_INLINE		= 0x0a,
+	DR_STE_V1_ACTION_ID_INSERT_POINTER		= 0x0b,
+	DR_STE_V1_ACTION_ID_FLOW_TAG			= 0x0c,
+	DR_STE_V1_ACTION_ID_QUEUE_ID_SEL		= 0x0d,
+	DR_STE_V1_ACTION_ID_ACCELERATED_LIST		= 0x0e,
+	DR_STE_V1_ACTION_ID_MODIFY_LIST			= 0x0f,
+	DR_STE_V1_ACTION_ID_ASO				= 0x12,
+	DR_STE_V1_ACTION_ID_TRAILER			= 0x13,
+	DR_STE_V1_ACTION_ID_COUNTER_ID			= 0x14,
+	DR_STE_V1_ACTION_ID_MAX				= 0x21,
+	/* use for special cases */
+	DR_STE_V1_ACTION_ID_SPECIAL_ENCAP_L3		= 0x22,
+};
+
+enum {
+	DR_STE_V1_ACTION_MDFY_FLD_L2_OUT_0		= 0x00,
+	DR_STE_V1_ACTION_MDFY_FLD_L2_OUT_1		= 0x01,
+	DR_STE_V1_ACTION_MDFY_FLD_L2_OUT_2		= 0x02,
+	DR_STE_V1_ACTION_MDFY_FLD_SRC_L2_OUT_0		= 0x08,
+	DR_STE_V1_ACTION_MDFY_FLD_SRC_L2_OUT_1		= 0x09,
+	DR_STE_V1_ACTION_MDFY_FLD_L3_OUT_0		= 0x0e,
+	DR_STE_V1_ACTION_MDFY_FLD_L4_OUT_0		= 0x18,
+	DR_STE_V1_ACTION_MDFY_FLD_L4_OUT_1		= 0x19,
+	DR_STE_V1_ACTION_MDFY_FLD_IPV4_OUT_0		= 0x40,
+	DR_STE_V1_ACTION_MDFY_FLD_IPV4_OUT_1		= 0x41,
+	DR_STE_V1_ACTION_MDFY_FLD_IPV6_DST_OUT_0	= 0x44,
+	DR_STE_V1_ACTION_MDFY_FLD_IPV6_DST_OUT_1	= 0x45,
+	DR_STE_V1_ACTION_MDFY_FLD_IPV6_DST_OUT_2	= 0x46,
+	DR_STE_V1_ACTION_MDFY_FLD_IPV6_DST_OUT_3	= 0x47,
+	DR_STE_V1_ACTION_MDFY_FLD_IPV6_SRC_OUT_0	= 0x4c,
+	DR_STE_V1_ACTION_MDFY_FLD_IPV6_SRC_OUT_1	= 0x4d,
+	DR_STE_V1_ACTION_MDFY_FLD_IPV6_SRC_OUT_2	= 0x4e,
+	DR_STE_V1_ACTION_MDFY_FLD_IPV6_SRC_OUT_3	= 0x4f,
+	DR_STE_V1_ACTION_MDFY_FLD_TCP_MISC_0		= 0x5e,
+	DR_STE_V1_ACTION_MDFY_FLD_TCP_MISC_1		= 0x5f,
+	DR_STE_V1_ACTION_MDFY_FLD_CFG_HDR_0_0		= 0x6f,
+	DR_STE_V1_ACTION_MDFY_FLD_CFG_HDR_0_1		= 0x70,
+	DR_STE_V1_ACTION_MDFY_FLD_METADATA_2_CQE	= 0x7b,
+	DR_STE_V1_ACTION_MDFY_FLD_GNRL_PURPOSE		= 0x7c,
+	DR_STE_V1_ACTION_MDFY_FLD_REGISTER_2_0		= 0x8c,
+	DR_STE_V1_ACTION_MDFY_FLD_REGISTER_2_1		= 0x8d,
+	DR_STE_V1_ACTION_MDFY_FLD_REGISTER_1_0		= 0x8e,
+	DR_STE_V1_ACTION_MDFY_FLD_REGISTER_1_1		= 0x8f,
+	DR_STE_V1_ACTION_MDFY_FLD_REGISTER_0_0		= 0x90,
+	DR_STE_V1_ACTION_MDFY_FLD_REGISTER_0_1		= 0x91,
+};
+
+enum dr_ste_v1_aso_ctx_type {
+	DR_STE_V1_ASO_CTX_TYPE_POLICERS = 0x2,
+};
+
+static const struct mlx5dr_ste_action_modify_field dr_ste_v1_action_modify_field_arr[] = {
+	[MLX5_ACTION_IN_FIELD_OUT_SMAC_47_16] = {
+		.hw_field = DR_STE_V1_ACTION_MDFY_FLD_SRC_L2_OUT_0, .start = 0, .end = 31,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_SMAC_15_0] = {
+		.hw_field = DR_STE_V1_ACTION_MDFY_FLD_SRC_L2_OUT_1, .start = 16, .end = 31,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_ETHERTYPE] = {
+		.hw_field = DR_STE_V1_ACTION_MDFY_FLD_L2_OUT_1, .start = 0, .end = 15,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_DMAC_47_16] = {
+		.hw_field = DR_STE_V1_ACTION_MDFY_FLD_L2_OUT_0, .start = 0, .end = 31,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_DMAC_15_0] = {
+		.hw_field = DR_STE_V1_ACTION_MDFY_FLD_L2_OUT_1, .start = 16, .end = 31,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_IP_DSCP] = {
+		.hw_field = DR_STE_V1_ACTION_MDFY_FLD_L3_OUT_0, .start = 18, .end = 23,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_TCP_FLAGS] = {
+		.hw_field = DR_STE_V1_ACTION_MDFY_FLD_L4_OUT_1, .start = 16, .end = 24,
+		.l4_type = DR_STE_ACTION_MDFY_TYPE_L4_TCP,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_TCP_SPORT] = {
+		.hw_field = DR_STE_V1_ACTION_MDFY_FLD_L4_OUT_0, .start = 16, .end = 31,
+		.l4_type = DR_STE_ACTION_MDFY_TYPE_L4_TCP,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_TCP_DPORT] = {
+		.hw_field = DR_STE_V1_ACTION_MDFY_FLD_L4_OUT_0, .start = 0, .end = 15,
+		.l4_type = DR_STE_ACTION_MDFY_TYPE_L4_TCP,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_IP_TTL] = {
+		.hw_field = DR_STE_V1_ACTION_MDFY_FLD_L3_OUT_0, .start = 8, .end = 15,
+		.l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV4,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_IPV6_HOPLIMIT] = {
+		.hw_field = DR_STE_V1_ACTION_MDFY_FLD_L3_OUT_0, .start = 8, .end = 15,
+		.l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV6,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_UDP_SPORT] = {
+		.hw_field = DR_STE_V1_ACTION_MDFY_FLD_L4_OUT_0, .start = 16, .end = 31,
+		.l4_type = DR_STE_ACTION_MDFY_TYPE_L4_UDP,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_UDP_DPORT] = {
+		.hw_field = DR_STE_V1_ACTION_MDFY_FLD_L4_OUT_0, .start = 0, .end = 15,
+		.l4_type = DR_STE_ACTION_MDFY_TYPE_L4_UDP,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_SIPV6_127_96] = {
+		.hw_field = DR_STE_V1_ACTION_MDFY_FLD_IPV6_SRC_OUT_0, .start = 0, .end = 31,
+		.l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV6,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_SIPV6_95_64] = {
+		.hw_field = DR_STE_V1_ACTION_MDFY_FLD_IPV6_SRC_OUT_1, .start = 0, .end = 31,
+		.l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV6,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_SIPV6_63_32] = {
+		.hw_field = DR_STE_V1_ACTION_MDFY_FLD_IPV6_SRC_OUT_2, .start = 0, .end = 31,
+		.l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV6,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_SIPV6_31_0] = {
+		.hw_field = DR_STE_V1_ACTION_MDFY_FLD_IPV6_SRC_OUT_3, .start = 0, .end = 31,
+		.l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV6,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_DIPV6_127_96] = {
+		.hw_field = DR_STE_V1_ACTION_MDFY_FLD_IPV6_DST_OUT_0, .start = 0, .end = 31,
+		.l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV6,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_DIPV6_95_64] = {
+		.hw_field = DR_STE_V1_ACTION_MDFY_FLD_IPV6_DST_OUT_1, .start = 0, .end = 31,
+		.l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV6,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_DIPV6_63_32] = {
+		.hw_field = DR_STE_V1_ACTION_MDFY_FLD_IPV6_DST_OUT_2, .start = 0, .end = 31,
+		.l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV6,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_DIPV6_31_0] = {
+		.hw_field = DR_STE_V1_ACTION_MDFY_FLD_IPV6_DST_OUT_3, .start = 0, .end = 31,
+		.l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV6,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_SIPV4] = {
+		.hw_field = DR_STE_V1_ACTION_MDFY_FLD_IPV4_OUT_0, .start = 0, .end = 31,
+		.l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV4,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_DIPV4] = {
+		.hw_field = DR_STE_V1_ACTION_MDFY_FLD_IPV4_OUT_1, .start = 0, .end = 31,
+		.l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV4,
+	},
+	[MLX5_ACTION_IN_FIELD_METADATA_REG_A] = {
+		.hw_field = DR_STE_V1_ACTION_MDFY_FLD_GNRL_PURPOSE, .start = 0, .end = 31,
+	},
+	[MLX5_ACTION_IN_FIELD_METADATA_REG_B] = {
+		.hw_field = DR_STE_V1_ACTION_MDFY_FLD_METADATA_2_CQE, .start = 0, .end = 31,
+	},
+	[MLX5_ACTION_IN_FIELD_METADATA_REG_C_0] = {
+		.hw_field = DR_STE_V1_ACTION_MDFY_FLD_REGISTER_0_0, .start = 0, .end = 31,
+	},
+	[MLX5_ACTION_IN_FIELD_METADATA_REG_C_1] = {
+		.hw_field = DR_STE_V1_ACTION_MDFY_FLD_REGISTER_0_1, .start = 0, .end = 31,
+	},
+	[MLX5_ACTION_IN_FIELD_METADATA_REG_C_2] = {
+		.hw_field = DR_STE_V1_ACTION_MDFY_FLD_REGISTER_1_0, .start = 0, .end = 31,
+	},
+	[MLX5_ACTION_IN_FIELD_METADATA_REG_C_3] = {
+		.hw_field = DR_STE_V1_ACTION_MDFY_FLD_REGISTER_1_1, .start = 0, .end = 31,
+	},
+	[MLX5_ACTION_IN_FIELD_METADATA_REG_C_4] = {
+		.hw_field = DR_STE_V1_ACTION_MDFY_FLD_REGISTER_2_0, .start = 0, .end = 31,
+	},
+	[MLX5_ACTION_IN_FIELD_METADATA_REG_C_5] = {
+		.hw_field = DR_STE_V1_ACTION_MDFY_FLD_REGISTER_2_1, .start = 0, .end = 31,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_TCP_SEQ_NUM] = {
+		.hw_field = DR_STE_V1_ACTION_MDFY_FLD_TCP_MISC_0, .start = 0, .end = 31,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_TCP_ACK_NUM] = {
+		.hw_field = DR_STE_V1_ACTION_MDFY_FLD_TCP_MISC_1, .start = 0, .end = 31,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_FIRST_VID] = {
+		.hw_field = DR_STE_V1_ACTION_MDFY_FLD_L2_OUT_2, .start = 0, .end = 15,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_EMD_31_0] = {
+		.hw_field = DR_STE_V1_ACTION_MDFY_FLD_CFG_HDR_0_1, .start = 0, .end = 31,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_EMD_47_32] = {
+		.hw_field = DR_STE_V1_ACTION_MDFY_FLD_CFG_HDR_0_0, .start = 0, .end = 15,
+	},
+};
+
+static void dr_ste_v1_set_entry_type(u8 *hw_ste_p, u8 entry_type)
+{
+	MLX5_SET(ste_match_bwc_v1, hw_ste_p, entry_format, entry_type);
+}
+
+void dr_ste_v1_set_miss_addr(u8 *hw_ste_p, u64 miss_addr)
+{
+	u64 index = miss_addr >> 6;
+
+	MLX5_SET(ste_match_bwc_v1, hw_ste_p, miss_address_39_32, index >> 26);
+	MLX5_SET(ste_match_bwc_v1, hw_ste_p, miss_address_31_6, index);
+}
+
+u64 dr_ste_v1_get_miss_addr(u8 *hw_ste_p)
+{
+	u64 index =
+		((u64)MLX5_GET(ste_match_bwc_v1, hw_ste_p, miss_address_31_6) |
+		 ((u64)MLX5_GET(ste_match_bwc_v1, hw_ste_p, miss_address_39_32)) << 26);
+
+	return index << 6;
+}
+
+void dr_ste_v1_set_byte_mask(u8 *hw_ste_p, u16 byte_mask)
+{
+	MLX5_SET(ste_match_bwc_v1, hw_ste_p, byte_mask, byte_mask);
+}
+
+u16 dr_ste_v1_get_byte_mask(u8 *hw_ste_p)
+{
+	return MLX5_GET(ste_match_bwc_v1, hw_ste_p, byte_mask);
+}
+
+static void dr_ste_v1_set_lu_type(u8 *hw_ste_p, u16 lu_type)
+{
+	MLX5_SET(ste_match_bwc_v1, hw_ste_p, entry_format, lu_type >> 8);
+	MLX5_SET(ste_match_bwc_v1, hw_ste_p, match_definer_ctx_idx, lu_type & 0xFF);
+}
+
+void dr_ste_v1_set_next_lu_type(u8 *hw_ste_p, u16 lu_type)
+{
+	MLX5_SET(ste_match_bwc_v1, hw_ste_p, next_entry_format, lu_type >> 8);
+	MLX5_SET(ste_match_bwc_v1, hw_ste_p, hash_definer_ctx_idx, lu_type & 0xFF);
+}
+
+u16 dr_ste_v1_get_next_lu_type(u8 *hw_ste_p)
+{
+	u8 mode = MLX5_GET(ste_match_bwc_v1, hw_ste_p, next_entry_format);
+	u8 index = MLX5_GET(ste_match_bwc_v1, hw_ste_p, hash_definer_ctx_idx);
+
+	return (mode << 8 | index);
+}
+
+static void dr_ste_v1_set_hit_gvmi(u8 *hw_ste_p, u16 gvmi)
+{
+	MLX5_SET(ste_match_bwc_v1, hw_ste_p, next_table_base_63_48, gvmi);
+}
+
+void dr_ste_v1_set_hit_addr(u8 *hw_ste_p, u64 icm_addr, u32 ht_size)
+{
+	u64 index = (icm_addr >> 5) | ht_size;
+
+	MLX5_SET(ste_match_bwc_v1, hw_ste_p, next_table_base_39_32_size, index >> 27);
+	MLX5_SET(ste_match_bwc_v1, hw_ste_p, next_table_base_31_5_size, index);
+}
+
+void dr_ste_v1_init(u8 *hw_ste_p, u16 lu_type, bool is_rx, u16 gvmi)
+{
+	dr_ste_v1_set_lu_type(hw_ste_p, lu_type);
+	dr_ste_v1_set_next_lu_type(hw_ste_p, MLX5DR_STE_LU_TYPE_DONT_CARE);
+
+	MLX5_SET(ste_match_bwc_v1, hw_ste_p, gvmi, gvmi);
+	MLX5_SET(ste_match_bwc_v1, hw_ste_p, next_table_base_63_48, gvmi);
+	MLX5_SET(ste_match_bwc_v1, hw_ste_p, miss_address_63_48, gvmi);
+}
+
+void dr_ste_v1_prepare_for_postsend(u8 *hw_ste_p, u32 ste_size)
+{
+	u8 *tag = hw_ste_p + DR_STE_SIZE_CTRL;
+	u8 *mask = tag + DR_STE_SIZE_TAG;
+	u8 tmp_tag[DR_STE_SIZE_TAG] = {};
+
+	if (ste_size == DR_STE_SIZE_CTRL)
+		return;
+
+	WARN_ON(ste_size != DR_STE_SIZE);
+
+	/* Backup tag */
+	memcpy(tmp_tag, tag, DR_STE_SIZE_TAG);
+
+	/* Swap mask and tag  both are the same size */
+	memcpy(tag, mask, DR_STE_SIZE_MASK);
+	memcpy(mask, tmp_tag, DR_STE_SIZE_TAG);
+}
+
+static void dr_ste_v1_set_rx_flow_tag(u8 *s_action, u32 flow_tag)
+{
+	MLX5_SET(ste_single_action_flow_tag_v1, s_action, action_id,
+		 DR_STE_V1_ACTION_ID_FLOW_TAG);
+	MLX5_SET(ste_single_action_flow_tag_v1, s_action, flow_tag, flow_tag);
+}
+
+static void dr_ste_v1_set_counter_id(u8 *hw_ste_p, u32 ctr_id)
+{
+	MLX5_SET(ste_match_bwc_v1, hw_ste_p, counter_id, ctr_id);
+}
+
+static void dr_ste_v1_set_reparse(u8 *hw_ste_p)
+{
+	MLX5_SET(ste_match_bwc_v1, hw_ste_p, reparse, 1);
+}
+
+static void dr_ste_v1_set_encap(u8 *hw_ste_p, u8 *d_action,
+				u32 reformat_id, int size)
+{
+	MLX5_SET(ste_double_action_insert_with_ptr_v1, d_action, action_id,
+		 DR_STE_V1_ACTION_ID_INSERT_POINTER);
+	/* The hardware expects here size in words (2 byte) */
+	MLX5_SET(ste_double_action_insert_with_ptr_v1, d_action, size, size / 2);
+	MLX5_SET(ste_double_action_insert_with_ptr_v1, d_action, pointer, reformat_id);
+	MLX5_SET(ste_double_action_insert_with_ptr_v1, d_action, attributes,
+		 DR_STE_V1_ACTION_INSERT_PTR_ATTR_ENCAP);
+	dr_ste_v1_set_reparse(hw_ste_p);
+}
+
+static void dr_ste_v1_set_insert_hdr(u8 *hw_ste_p, u8 *d_action,
+				     u32 reformat_id,
+				     u8 anchor, u8 offset,
+				     int size)
+{
+	MLX5_SET(ste_double_action_insert_with_ptr_v1, d_action,
+		 action_id, DR_STE_V1_ACTION_ID_INSERT_POINTER);
+	MLX5_SET(ste_double_action_insert_with_ptr_v1, d_action, start_anchor, anchor);
+
+	/* The hardware expects here size and offset in words (2 byte) */
+	MLX5_SET(ste_double_action_insert_with_ptr_v1, d_action, size, size / 2);
+	MLX5_SET(ste_double_action_insert_with_ptr_v1, d_action, start_offset, offset / 2);
+
+	MLX5_SET(ste_double_action_insert_with_ptr_v1, d_action, pointer, reformat_id);
+	MLX5_SET(ste_double_action_insert_with_ptr_v1, d_action, attributes,
+		 DR_STE_V1_ACTION_INSERT_PTR_ATTR_NONE);
+
+	dr_ste_v1_set_reparse(hw_ste_p);
+}
+
+static void dr_ste_v1_set_remove_hdr(u8 *hw_ste_p, u8 *s_action,
+				     u8 anchor, u8 offset,
+				     int size)
+{
+	MLX5_SET(ste_single_action_remove_header_size_v1, s_action,
+		 action_id, DR_STE_V1_ACTION_ID_REMOVE_BY_SIZE);
+	MLX5_SET(ste_single_action_remove_header_size_v1, s_action, start_anchor, anchor);
+
+	/* The hardware expects here size and offset in words (2 byte) */
+	MLX5_SET(ste_single_action_remove_header_size_v1, s_action, remove_size, size / 2);
+	MLX5_SET(ste_single_action_remove_header_size_v1, s_action, start_offset, offset / 2);
+
+	dr_ste_v1_set_reparse(hw_ste_p);
+}
+
+static void dr_ste_v1_set_push_vlan(u8 *hw_ste_p, u8 *d_action,
+				    u32 vlan_hdr)
+{
+	MLX5_SET(ste_double_action_insert_with_inline_v1, d_action,
+		 action_id, DR_STE_V1_ACTION_ID_INSERT_INLINE);
+	/* The hardware expects offset to vlan header in words (2 byte) */
+	MLX5_SET(ste_double_action_insert_with_inline_v1, d_action,
+		 start_offset, HDR_LEN_L2_MACS >> 1);
+	MLX5_SET(ste_double_action_insert_with_inline_v1, d_action,
+		 inline_data, vlan_hdr);
+
+	dr_ste_v1_set_reparse(hw_ste_p);
+}
+
+static void dr_ste_v1_set_pop_vlan(u8 *hw_ste_p, u8 *s_action, u8 vlans_num)
+{
+	MLX5_SET(ste_single_action_remove_header_size_v1, s_action,
+		 action_id, DR_STE_V1_ACTION_ID_REMOVE_BY_SIZE);
+	MLX5_SET(ste_single_action_remove_header_size_v1, s_action,
+		 start_anchor, DR_STE_HEADER_ANCHOR_1ST_VLAN);
+	/* The hardware expects here size in words (2 byte) */
+	MLX5_SET(ste_single_action_remove_header_size_v1, s_action,
+		 remove_size, (HDR_LEN_L2_VLAN >> 1) * vlans_num);
+
+	dr_ste_v1_set_reparse(hw_ste_p);
+}
+
+static void dr_ste_v1_set_encap_l3(u8 *hw_ste_p,
+				   u8 *frst_s_action,
+				   u8 *scnd_d_action,
+				   u32 reformat_id,
+				   int size)
+{
+	/* Remove L2 headers */
+	MLX5_SET(ste_single_action_remove_header_v1, frst_s_action, action_id,
+		 DR_STE_V1_ACTION_ID_REMOVE_HEADER_TO_HEADER);
+	MLX5_SET(ste_single_action_remove_header_v1, frst_s_action, end_anchor,
+		 DR_STE_HEADER_ANCHOR_IPV6_IPV4);
+
+	/* Encapsulate with given reformat ID */
+	MLX5_SET(ste_double_action_insert_with_ptr_v1, scnd_d_action, action_id,
+		 DR_STE_V1_ACTION_ID_INSERT_POINTER);
+	/* The hardware expects here size in words (2 byte) */
+	MLX5_SET(ste_double_action_insert_with_ptr_v1, scnd_d_action, size, size / 2);
+	MLX5_SET(ste_double_action_insert_with_ptr_v1, scnd_d_action, pointer, reformat_id);
+	MLX5_SET(ste_double_action_insert_with_ptr_v1, scnd_d_action, attributes,
+		 DR_STE_V1_ACTION_INSERT_PTR_ATTR_ENCAP);
+
+	dr_ste_v1_set_reparse(hw_ste_p);
+}
+
+static void dr_ste_v1_set_rx_decap(u8 *hw_ste_p, u8 *s_action)
+{
+	MLX5_SET(ste_single_action_remove_header_v1, s_action, action_id,
+		 DR_STE_V1_ACTION_ID_REMOVE_HEADER_TO_HEADER);
+	MLX5_SET(ste_single_action_remove_header_v1, s_action, decap, 1);
+	MLX5_SET(ste_single_action_remove_header_v1, s_action, vni_to_cqe, 1);
+	MLX5_SET(ste_single_action_remove_header_v1, s_action, end_anchor,
+		 DR_STE_HEADER_ANCHOR_INNER_MAC);
+
+	dr_ste_v1_set_reparse(hw_ste_p);
+}
+
+static void dr_ste_v1_set_rewrite_actions(u8 *hw_ste_p,
+					  u8 *s_action,
+					  u16 num_of_actions,
+					  u32 re_write_index)
+{
+	MLX5_SET(ste_single_action_modify_list_v1, s_action, action_id,
+		 DR_STE_V1_ACTION_ID_MODIFY_LIST);
+	MLX5_SET(ste_single_action_modify_list_v1, s_action, num_of_modify_actions,
+		 num_of_actions);
+	MLX5_SET(ste_single_action_modify_list_v1, s_action, modify_actions_ptr,
+		 re_write_index);
+
+	dr_ste_v1_set_reparse(hw_ste_p);
+}
+
+static void dr_ste_v1_set_aso_flow_meter(u8 *d_action,
+					 u32 object_id,
+					 u32 offset,
+					 u8 dest_reg_id,
+					 u8 init_color)
+{
+	MLX5_SET(ste_double_action_aso_v1, d_action, action_id,
+		 DR_STE_V1_ACTION_ID_ASO);
+	MLX5_SET(ste_double_action_aso_v1, d_action, aso_context_number,
+		 object_id + (offset / MLX5DR_ASO_FLOW_METER_NUM_PER_OBJ));
+	/* Convert reg_c index to HW 64bit index */
+	MLX5_SET(ste_double_action_aso_v1, d_action, dest_reg_id,
+		 (dest_reg_id - 1) / 2);
+	MLX5_SET(ste_double_action_aso_v1, d_action, aso_context_type,
+		 DR_STE_V1_ASO_CTX_TYPE_POLICERS);
+	MLX5_SET(ste_double_action_aso_v1, d_action, flow_meter.line_id,
+		 offset % MLX5DR_ASO_FLOW_METER_NUM_PER_OBJ);
+	MLX5_SET(ste_double_action_aso_v1, d_action, flow_meter.initial_color,
+		 init_color);
+}
+
+static void dr_ste_v1_arr_init_next_match(u8 **last_ste,
+					  u32 *added_stes,
+					  u16 gvmi)
+{
+	u8 *action;
+
+	(*added_stes)++;
+	*last_ste += DR_STE_SIZE;
+	dr_ste_v1_init(*last_ste, MLX5DR_STE_LU_TYPE_DONT_CARE, 0, gvmi);
+	dr_ste_v1_set_entry_type(*last_ste, DR_STE_V1_TYPE_MATCH);
+
+	action = MLX5_ADDR_OF(ste_mask_and_match_v1, *last_ste, action);
+	memset(action, 0, MLX5_FLD_SZ_BYTES(ste_mask_and_match_v1, action));
+}
+
+void dr_ste_v1_set_actions_tx(struct mlx5dr_domain *dmn,
+			      u8 *action_type_set,
+			      u32 actions_caps,
+			      u8 *last_ste,
+			      struct mlx5dr_ste_actions_attr *attr,
+			      u32 *added_stes)
+{
+	u8 *action = MLX5_ADDR_OF(ste_match_bwc_v1, last_ste, action);
+	u8 action_sz = DR_STE_ACTION_DOUBLE_SZ;
+	bool allow_modify_hdr = true;
+	bool allow_encap = true;
+
+	if (action_type_set[DR_ACTION_TYP_POP_VLAN]) {
+		if (action_sz < DR_STE_ACTION_SINGLE_SZ) {
+			dr_ste_v1_arr_init_next_match(&last_ste, added_stes,
+						      attr->gvmi);
+			action = MLX5_ADDR_OF(ste_mask_and_match_v1,
+					      last_ste, action);
+			action_sz = DR_STE_ACTION_TRIPLE_SZ;
+		}
+		dr_ste_v1_set_pop_vlan(last_ste, action, attr->vlans.count);
+		action_sz -= DR_STE_ACTION_SINGLE_SZ;
+		action += DR_STE_ACTION_SINGLE_SZ;
+
+		/* Check if vlan_pop and modify_hdr on same STE is supported */
+		if (!(actions_caps & DR_STE_CTX_ACTION_CAP_POP_MDFY))
+			allow_modify_hdr = false;
+	}
+
+	if (action_type_set[DR_ACTION_TYP_CTR])
+		dr_ste_v1_set_counter_id(last_ste, attr->ctr_id);
+
+	if (action_type_set[DR_ACTION_TYP_MODIFY_HDR]) {
+		if (!allow_modify_hdr || action_sz < DR_STE_ACTION_DOUBLE_SZ) {
+			dr_ste_v1_arr_init_next_match(&last_ste, added_stes,
+						      attr->gvmi);
+			action = MLX5_ADDR_OF(ste_mask_and_match_v1,
+					      last_ste, action);
+			action_sz = DR_STE_ACTION_TRIPLE_SZ;
+		}
+		dr_ste_v1_set_rewrite_actions(last_ste, action,
+					      attr->modify_actions,
+					      attr->modify_index);
+		action_sz -= DR_STE_ACTION_DOUBLE_SZ;
+		action += DR_STE_ACTION_DOUBLE_SZ;
+		allow_encap = false;
+	}
+
+	if (action_type_set[DR_ACTION_TYP_PUSH_VLAN]) {
+		int i;
+
+		for (i = 0; i < attr->vlans.count; i++) {
+			if (action_sz < DR_STE_ACTION_DOUBLE_SZ || !allow_encap) {
+				dr_ste_v1_arr_init_next_match(&last_ste, added_stes, attr->gvmi);
+				action = MLX5_ADDR_OF(ste_mask_and_match_v1, last_ste, action);
+				action_sz = DR_STE_ACTION_TRIPLE_SZ;
+				allow_encap = true;
+			}
+			dr_ste_v1_set_push_vlan(last_ste, action,
+						attr->vlans.headers[i]);
+			action_sz -= DR_STE_ACTION_DOUBLE_SZ;
+			action += DR_STE_ACTION_DOUBLE_SZ;
+		}
+	}
+
+	if (action_type_set[DR_ACTION_TYP_L2_TO_TNL_L2]) {
+		if (!allow_encap || action_sz < DR_STE_ACTION_DOUBLE_SZ) {
+			dr_ste_v1_arr_init_next_match(&last_ste, added_stes, attr->gvmi);
+			action = MLX5_ADDR_OF(ste_mask_and_match_v1, last_ste, action);
+			action_sz = DR_STE_ACTION_TRIPLE_SZ;
+			allow_encap = true;
+		}
+		dr_ste_v1_set_encap(last_ste, action,
+				    attr->reformat.id,
+				    attr->reformat.size);
+		action_sz -= DR_STE_ACTION_DOUBLE_SZ;
+		action += DR_STE_ACTION_DOUBLE_SZ;
+	} else if (action_type_set[DR_ACTION_TYP_L2_TO_TNL_L3]) {
+		u8 *d_action;
+
+		if (action_sz < DR_STE_ACTION_TRIPLE_SZ) {
+			dr_ste_v1_arr_init_next_match(&last_ste, added_stes, attr->gvmi);
+			action = MLX5_ADDR_OF(ste_mask_and_match_v1, last_ste, action);
+			action_sz = DR_STE_ACTION_TRIPLE_SZ;
+		}
+		d_action = action + DR_STE_ACTION_SINGLE_SZ;
+
+		dr_ste_v1_set_encap_l3(last_ste,
+				       action, d_action,
+				       attr->reformat.id,
+				       attr->reformat.size);
+		action_sz -= DR_STE_ACTION_TRIPLE_SZ;
+		action += DR_STE_ACTION_TRIPLE_SZ;
+	} else if (action_type_set[DR_ACTION_TYP_INSERT_HDR]) {
+		if (!allow_encap || action_sz < DR_STE_ACTION_DOUBLE_SZ) {
+			dr_ste_v1_arr_init_next_match(&last_ste, added_stes, attr->gvmi);
+			action = MLX5_ADDR_OF(ste_mask_and_match_v1, last_ste, action);
+			action_sz = DR_STE_ACTION_TRIPLE_SZ;
+		}
+		dr_ste_v1_set_insert_hdr(last_ste, action,
+					 attr->reformat.id,
+					 attr->reformat.param_0,
+					 attr->reformat.param_1,
+					 attr->reformat.size);
+		action_sz -= DR_STE_ACTION_DOUBLE_SZ;
+		action += DR_STE_ACTION_DOUBLE_SZ;
+	} else if (action_type_set[DR_ACTION_TYP_REMOVE_HDR]) {
+		if (action_sz < DR_STE_ACTION_SINGLE_SZ) {
+			dr_ste_v1_arr_init_next_match(&last_ste, added_stes, attr->gvmi);
+			action = MLX5_ADDR_OF(ste_mask_and_match_v1, last_ste, action);
+			action_sz = DR_STE_ACTION_TRIPLE_SZ;
+		}
+		dr_ste_v1_set_remove_hdr(last_ste, action,
+					 attr->reformat.param_0,
+					 attr->reformat.param_1,
+					 attr->reformat.size);
+		action_sz -= DR_STE_ACTION_SINGLE_SZ;
+		action += DR_STE_ACTION_SINGLE_SZ;
+	}
+
+	if (action_type_set[DR_ACTION_TYP_ASO_FLOW_METER]) {
+		if (action_sz < DR_STE_ACTION_DOUBLE_SZ) {
+			dr_ste_v1_arr_init_next_match(&last_ste, added_stes, attr->gvmi);
+			action = MLX5_ADDR_OF(ste_mask_and_match_v1, last_ste, action);
+			action_sz = DR_STE_ACTION_TRIPLE_SZ;
+		}
+		dr_ste_v1_set_aso_flow_meter(action,
+					     attr->aso_flow_meter.obj_id,
+					     attr->aso_flow_meter.offset,
+					     attr->aso_flow_meter.dest_reg_id,
+					     attr->aso_flow_meter.init_color);
+		action_sz -= DR_STE_ACTION_DOUBLE_SZ;
+		action += DR_STE_ACTION_DOUBLE_SZ;
+	}
+
+	dr_ste_v1_set_hit_gvmi(last_ste, attr->hit_gvmi);
+	dr_ste_v1_set_hit_addr(last_ste, attr->final_icm_addr, 1);
+}
+
+void dr_ste_v1_set_actions_rx(struct mlx5dr_domain *dmn,
+			      u8 *action_type_set,
+			      u32 actions_caps,
+			      u8 *last_ste,
+			      struct mlx5dr_ste_actions_attr *attr,
+			      u32 *added_stes)
+{
+	u8 *action = MLX5_ADDR_OF(ste_match_bwc_v1, last_ste, action);
+	u8 action_sz = DR_STE_ACTION_DOUBLE_SZ;
+	bool allow_modify_hdr = true;
+	bool allow_ctr = true;
+
+	if (action_type_set[DR_ACTION_TYP_TNL_L3_TO_L2]) {
+		dr_ste_v1_set_rewrite_actions(last_ste, action,
+					      attr->decap_actions,
+					      attr->decap_index);
+		action_sz -= DR_STE_ACTION_DOUBLE_SZ;
+		action += DR_STE_ACTION_DOUBLE_SZ;
+		allow_modify_hdr = false;
+		allow_ctr = false;
+	} else if (action_type_set[DR_ACTION_TYP_TNL_L2_TO_L2]) {
+		dr_ste_v1_set_rx_decap(last_ste, action);
+		action_sz -= DR_STE_ACTION_SINGLE_SZ;
+		action += DR_STE_ACTION_SINGLE_SZ;
+		allow_modify_hdr = false;
+		allow_ctr = false;
+	}
+
+	if (action_type_set[DR_ACTION_TYP_TAG]) {
+		if (action_sz < DR_STE_ACTION_SINGLE_SZ) {
+			dr_ste_v1_arr_init_next_match(&last_ste, added_stes, attr->gvmi);
+			action = MLX5_ADDR_OF(ste_mask_and_match_v1, last_ste, action);
+			action_sz = DR_STE_ACTION_TRIPLE_SZ;
+			allow_modify_hdr = true;
+			allow_ctr = true;
+		}
+		dr_ste_v1_set_rx_flow_tag(action, attr->flow_tag);
+		action_sz -= DR_STE_ACTION_SINGLE_SZ;
+		action += DR_STE_ACTION_SINGLE_SZ;
+	}
+
+	if (action_type_set[DR_ACTION_TYP_POP_VLAN]) {
+		if (action_sz < DR_STE_ACTION_SINGLE_SZ ||
+		    !allow_modify_hdr) {
+			dr_ste_v1_arr_init_next_match(&last_ste, added_stes, attr->gvmi);
+			action = MLX5_ADDR_OF(ste_mask_and_match_v1, last_ste, action);
+			action_sz = DR_STE_ACTION_TRIPLE_SZ;
+		}
+
+		dr_ste_v1_set_pop_vlan(last_ste, action, attr->vlans.count);
+		action_sz -= DR_STE_ACTION_SINGLE_SZ;
+		action += DR_STE_ACTION_SINGLE_SZ;
+		allow_ctr = false;
+
+		/* Check if vlan_pop and modify_hdr on same STE is supported */
+		if (!(actions_caps & DR_STE_CTX_ACTION_CAP_POP_MDFY))
+			allow_modify_hdr = false;
+	}
+
+	if (action_type_set[DR_ACTION_TYP_MODIFY_HDR]) {
+		/* Modify header and decapsulation must use different STEs */
+		if (!allow_modify_hdr || action_sz < DR_STE_ACTION_DOUBLE_SZ) {
+			dr_ste_v1_arr_init_next_match(&last_ste, added_stes, attr->gvmi);
+			action = MLX5_ADDR_OF(ste_mask_and_match_v1, last_ste, action);
+			action_sz = DR_STE_ACTION_TRIPLE_SZ;
+			allow_modify_hdr = true;
+			allow_ctr = true;
+		}
+		dr_ste_v1_set_rewrite_actions(last_ste, action,
+					      attr->modify_actions,
+					      attr->modify_index);
+		action_sz -= DR_STE_ACTION_DOUBLE_SZ;
+		action += DR_STE_ACTION_DOUBLE_SZ;
+	}
+
+	if (action_type_set[DR_ACTION_TYP_PUSH_VLAN]) {
+		int i;
+
+		for (i = 0; i < attr->vlans.count; i++) {
+			if (action_sz < DR_STE_ACTION_DOUBLE_SZ ||
+			    !allow_modify_hdr) {
+				dr_ste_v1_arr_init_next_match(&last_ste,
+							      added_stes,
+							      attr->gvmi);
+				action = MLX5_ADDR_OF(ste_mask_and_match_v1,
+						      last_ste, action);
+				action_sz = DR_STE_ACTION_TRIPLE_SZ;
+			}
+			dr_ste_v1_set_push_vlan(last_ste, action,
+						attr->vlans.headers[i]);
+			action_sz -= DR_STE_ACTION_DOUBLE_SZ;
+			action += DR_STE_ACTION_DOUBLE_SZ;
+		}
+	}
+
+	if (action_type_set[DR_ACTION_TYP_CTR]) {
+		/* Counter action set after decap and before insert_hdr
+		 * to exclude decaped / encaped header respectively.
+		 */
+		if (!allow_ctr) {
+			dr_ste_v1_arr_init_next_match(&last_ste, added_stes, attr->gvmi);
+			action = MLX5_ADDR_OF(ste_mask_and_match_v1, last_ste, action);
+			action_sz = DR_STE_ACTION_TRIPLE_SZ;
+			allow_modify_hdr = true;
+		}
+		dr_ste_v1_set_counter_id(last_ste, attr->ctr_id);
+		allow_ctr = false;
+	}
+
+	if (action_type_set[DR_ACTION_TYP_L2_TO_TNL_L2]) {
+		if (action_sz < DR_STE_ACTION_DOUBLE_SZ) {
+			dr_ste_v1_arr_init_next_match(&last_ste, added_stes, attr->gvmi);
+			action = MLX5_ADDR_OF(ste_mask_and_match_v1, last_ste, action);
+			action_sz = DR_STE_ACTION_TRIPLE_SZ;
+		}
+		dr_ste_v1_set_encap(last_ste, action,
+				    attr->reformat.id,
+				    attr->reformat.size);
+		action_sz -= DR_STE_ACTION_DOUBLE_SZ;
+		action += DR_STE_ACTION_DOUBLE_SZ;
+		allow_modify_hdr = false;
+	} else if (action_type_set[DR_ACTION_TYP_L2_TO_TNL_L3]) {
+		u8 *d_action;
+
+		if (action_sz < DR_STE_ACTION_TRIPLE_SZ) {
+			dr_ste_v1_arr_init_next_match(&last_ste, added_stes, attr->gvmi);
+			action = MLX5_ADDR_OF(ste_mask_and_match_v1, last_ste, action);
+			action_sz = DR_STE_ACTION_TRIPLE_SZ;
+		}
+
+		d_action = action + DR_STE_ACTION_SINGLE_SZ;
+
+		dr_ste_v1_set_encap_l3(last_ste,
+				       action, d_action,
+				       attr->reformat.id,
+				       attr->reformat.size);
+		action_sz -= DR_STE_ACTION_TRIPLE_SZ;
+		allow_modify_hdr = false;
+	} else if (action_type_set[DR_ACTION_TYP_INSERT_HDR]) {
+		/* Modify header, decap, and encap must use different STEs */
+		if (!allow_modify_hdr || action_sz < DR_STE_ACTION_DOUBLE_SZ) {
+			dr_ste_v1_arr_init_next_match(&last_ste, added_stes, attr->gvmi);
+			action = MLX5_ADDR_OF(ste_mask_and_match_v1, last_ste, action);
+			action_sz = DR_STE_ACTION_TRIPLE_SZ;
+		}
+		dr_ste_v1_set_insert_hdr(last_ste, action,
+					 attr->reformat.id,
+					 attr->reformat.param_0,
+					 attr->reformat.param_1,
+					 attr->reformat.size);
+		action_sz -= DR_STE_ACTION_DOUBLE_SZ;
+		action += DR_STE_ACTION_DOUBLE_SZ;
+		allow_modify_hdr = false;
+	} else if (action_type_set[DR_ACTION_TYP_REMOVE_HDR]) {
+		if (action_sz < DR_STE_ACTION_SINGLE_SZ) {
+			dr_ste_v1_arr_init_next_match(&last_ste, added_stes, attr->gvmi);
+			action = MLX5_ADDR_OF(ste_mask_and_match_v1, last_ste, action);
+			action_sz = DR_STE_ACTION_TRIPLE_SZ;
+			allow_modify_hdr = true;
+			allow_ctr = true;
+		}
+		dr_ste_v1_set_remove_hdr(last_ste, action,
+					 attr->reformat.param_0,
+					 attr->reformat.param_1,
+					 attr->reformat.size);
+		action_sz -= DR_STE_ACTION_SINGLE_SZ;
+		action += DR_STE_ACTION_SINGLE_SZ;
+	}
+
+	if (action_type_set[DR_ACTION_TYP_ASO_FLOW_METER]) {
+		if (action_sz < DR_STE_ACTION_DOUBLE_SZ) {
+			dr_ste_v1_arr_init_next_match(&last_ste, added_stes, attr->gvmi);
+			action = MLX5_ADDR_OF(ste_mask_and_match_v1, last_ste, action);
+			action_sz = DR_STE_ACTION_TRIPLE_SZ;
+		}
+		dr_ste_v1_set_aso_flow_meter(action,
+					     attr->aso_flow_meter.obj_id,
+					     attr->aso_flow_meter.offset,
+					     attr->aso_flow_meter.dest_reg_id,
+					     attr->aso_flow_meter.init_color);
+		action_sz -= DR_STE_ACTION_DOUBLE_SZ;
+		action += DR_STE_ACTION_DOUBLE_SZ;
+	}
+
+	dr_ste_v1_set_hit_gvmi(last_ste, attr->hit_gvmi);
+	dr_ste_v1_set_hit_addr(last_ste, attr->final_icm_addr, 1);
+}
+
+void dr_ste_v1_set_action_set(u8 *d_action,
+			      u8 hw_field,
+			      u8 shifter,
+			      u8 length,
+			      u32 data)
+{
+	shifter += MLX5_MODIFY_HEADER_V1_QW_OFFSET;
+	MLX5_SET(ste_double_action_set_v1, d_action, action_id, DR_STE_V1_ACTION_ID_SET);
+	MLX5_SET(ste_double_action_set_v1, d_action, destination_dw_offset, hw_field);
+	MLX5_SET(ste_double_action_set_v1, d_action, destination_left_shifter, shifter);
+	MLX5_SET(ste_double_action_set_v1, d_action, destination_length, length);
+	MLX5_SET(ste_double_action_set_v1, d_action, inline_data, data);
+}
+
+void dr_ste_v1_set_action_add(u8 *d_action,
+			      u8 hw_field,
+			      u8 shifter,
+			      u8 length,
+			      u32 data)
+{
+	shifter += MLX5_MODIFY_HEADER_V1_QW_OFFSET;
+	MLX5_SET(ste_double_action_add_v1, d_action, action_id, DR_STE_V1_ACTION_ID_ADD);
+	MLX5_SET(ste_double_action_add_v1, d_action, destination_dw_offset, hw_field);
+	MLX5_SET(ste_double_action_add_v1, d_action, destination_left_shifter, shifter);
+	MLX5_SET(ste_double_action_add_v1, d_action, destination_length, length);
+	MLX5_SET(ste_double_action_add_v1, d_action, add_value, data);
+}
+
+void dr_ste_v1_set_action_copy(u8 *d_action,
+			       u8 dst_hw_field,
+			       u8 dst_shifter,
+			       u8 dst_len,
+			       u8 src_hw_field,
+			       u8 src_shifter)
+{
+	dst_shifter += MLX5_MODIFY_HEADER_V1_QW_OFFSET;
+	src_shifter += MLX5_MODIFY_HEADER_V1_QW_OFFSET;
+	MLX5_SET(ste_double_action_copy_v1, d_action, action_id, DR_STE_V1_ACTION_ID_COPY);
+	MLX5_SET(ste_double_action_copy_v1, d_action, destination_dw_offset, dst_hw_field);
+	MLX5_SET(ste_double_action_copy_v1, d_action, destination_left_shifter, dst_shifter);
+	MLX5_SET(ste_double_action_copy_v1, d_action, destination_length, dst_len);
+	MLX5_SET(ste_double_action_copy_v1, d_action, source_dw_offset, src_hw_field);
+	MLX5_SET(ste_double_action_copy_v1, d_action, source_right_shifter, src_shifter);
+}
+
+#define DR_STE_DECAP_L3_ACTION_NUM	8
+#define DR_STE_L2_HDR_MAX_SZ		20
+
+int dr_ste_v1_set_action_decap_l3_list(void *data,
+				       u32 data_sz,
+				       u8 *hw_action,
+				       u32 hw_action_sz,
+				       u16 *used_hw_action_num)
+{
+	u8 padded_data[DR_STE_L2_HDR_MAX_SZ] = {};
+	void *data_ptr = padded_data;
+	u16 used_actions = 0;
+	u32 inline_data_sz;
+	u32 i;
+
+	if (hw_action_sz / DR_STE_ACTION_DOUBLE_SZ < DR_STE_DECAP_L3_ACTION_NUM)
+		return -EINVAL;
+
+	inline_data_sz =
+		MLX5_FLD_SZ_BYTES(ste_double_action_insert_with_inline_v1, inline_data);
+
+	/* Add an alignment padding  */
+	memcpy(padded_data + data_sz % inline_data_sz, data, data_sz);
+
+	/* Remove L2L3 outer headers */
+	MLX5_SET(ste_single_action_remove_header_v1, hw_action, action_id,
+		 DR_STE_V1_ACTION_ID_REMOVE_HEADER_TO_HEADER);
+	MLX5_SET(ste_single_action_remove_header_v1, hw_action, decap, 1);
+	MLX5_SET(ste_single_action_remove_header_v1, hw_action, vni_to_cqe, 1);
+	MLX5_SET(ste_single_action_remove_header_v1, hw_action, end_anchor,
+		 DR_STE_HEADER_ANCHOR_INNER_IPV6_IPV4);
+	hw_action += DR_STE_ACTION_DOUBLE_SZ;
+	used_actions++; /* Remove and NOP are a single double action */
+
+	/* Point to the last dword of the header */
+	data_ptr += (data_sz / inline_data_sz) * inline_data_sz;
+
+	/* Add the new header using inline action 4Byte at a time, the header
+	 * is added in reversed order to the beginning of the packet to avoid
+	 * incorrect parsing by the HW. Since header is 14B or 18B an extra
+	 * two bytes are padded and later removed.
+	 */
+	for (i = 0; i < data_sz / inline_data_sz + 1; i++) {
+		void *addr_inline;
+
+		MLX5_SET(ste_double_action_insert_with_inline_v1, hw_action, action_id,
+			 DR_STE_V1_ACTION_ID_INSERT_INLINE);
+		/* The hardware expects here offset to words (2 bytes) */
+		MLX5_SET(ste_double_action_insert_with_inline_v1, hw_action, start_offset, 0);
+
+		/* Copy bytes one by one to avoid endianness problem */
+		addr_inline = MLX5_ADDR_OF(ste_double_action_insert_with_inline_v1,
+					   hw_action, inline_data);
+		memcpy(addr_inline, data_ptr - i * inline_data_sz, inline_data_sz);
+		hw_action += DR_STE_ACTION_DOUBLE_SZ;
+		used_actions++;
+	}
+
+	/* Remove first 2 extra bytes */
+	MLX5_SET(ste_single_action_remove_header_size_v1, hw_action, action_id,
+		 DR_STE_V1_ACTION_ID_REMOVE_BY_SIZE);
+	MLX5_SET(ste_single_action_remove_header_size_v1, hw_action, start_offset, 0);
+	/* The hardware expects here size in words (2 bytes) */
+	MLX5_SET(ste_single_action_remove_header_size_v1, hw_action, remove_size, 1);
+	used_actions++;
+
+	*used_hw_action_num = used_actions;
+
+	return 0;
+}
+
+static void dr_ste_v1_build_eth_l2_src_dst_bit_mask(struct mlx5dr_match_param *value,
+						    bool inner, u8 *bit_mask)
+{
+	struct mlx5dr_match_spec *mask = inner ? &value->inner : &value->outer;
+
+	DR_STE_SET_TAG(eth_l2_src_dst_v1, bit_mask, dmac_47_16, mask, dmac_47_16);
+	DR_STE_SET_TAG(eth_l2_src_dst_v1, bit_mask, dmac_15_0, mask, dmac_15_0);
+
+	DR_STE_SET_TAG(eth_l2_src_dst_v1, bit_mask, smac_47_16, mask, smac_47_16);
+	DR_STE_SET_TAG(eth_l2_src_dst_v1, bit_mask, smac_15_0, mask, smac_15_0);
+
+	DR_STE_SET_TAG(eth_l2_src_dst_v1, bit_mask, first_vlan_id, mask, first_vid);
+	DR_STE_SET_TAG(eth_l2_src_dst_v1, bit_mask, first_cfi, mask, first_cfi);
+	DR_STE_SET_TAG(eth_l2_src_dst_v1, bit_mask, first_priority, mask, first_prio);
+	DR_STE_SET_ONES(eth_l2_src_dst_v1, bit_mask, l3_type, mask, ip_version);
+
+	if (mask->cvlan_tag) {
+		MLX5_SET(ste_eth_l2_src_dst_v1, bit_mask, first_vlan_qualifier, -1);
+		mask->cvlan_tag = 0;
+	} else if (mask->svlan_tag) {
+		MLX5_SET(ste_eth_l2_src_dst_v1, bit_mask, first_vlan_qualifier, -1);
+		mask->svlan_tag = 0;
+	}
+}
+
+static int dr_ste_v1_build_eth_l2_src_dst_tag(struct mlx5dr_match_param *value,
+					      struct mlx5dr_ste_build *sb,
+					      u8 *tag)
+{
+	struct mlx5dr_match_spec *spec = sb->inner ? &value->inner : &value->outer;
+
+	DR_STE_SET_TAG(eth_l2_src_dst_v1, tag, dmac_47_16, spec, dmac_47_16);
+	DR_STE_SET_TAG(eth_l2_src_dst_v1, tag, dmac_15_0, spec, dmac_15_0);
+
+	DR_STE_SET_TAG(eth_l2_src_dst_v1, tag, smac_47_16, spec, smac_47_16);
+	DR_STE_SET_TAG(eth_l2_src_dst_v1, tag, smac_15_0, spec, smac_15_0);
+
+	if (spec->ip_version == IP_VERSION_IPV4) {
+		MLX5_SET(ste_eth_l2_src_dst_v1, tag, l3_type, STE_IPV4);
+		spec->ip_version = 0;
+	} else if (spec->ip_version == IP_VERSION_IPV6) {
+		MLX5_SET(ste_eth_l2_src_dst_v1, tag, l3_type, STE_IPV6);
+		spec->ip_version = 0;
+	} else if (spec->ip_version) {
+		return -EINVAL;
+	}
+
+	DR_STE_SET_TAG(eth_l2_src_dst_v1, tag, first_vlan_id, spec, first_vid);
+	DR_STE_SET_TAG(eth_l2_src_dst_v1, tag, first_cfi, spec, first_cfi);
+	DR_STE_SET_TAG(eth_l2_src_dst_v1, tag, first_priority, spec, first_prio);
+
+	if (spec->cvlan_tag) {
+		MLX5_SET(ste_eth_l2_src_dst_v1, tag, first_vlan_qualifier, DR_STE_CVLAN);
+		spec->cvlan_tag = 0;
+	} else if (spec->svlan_tag) {
+		MLX5_SET(ste_eth_l2_src_dst_v1, tag, first_vlan_qualifier, DR_STE_SVLAN);
+		spec->svlan_tag = 0;
+	}
+	return 0;
+}
+
+void dr_ste_v1_build_eth_l2_src_dst_init(struct mlx5dr_ste_build *sb,
+					 struct mlx5dr_match_param *mask)
+{
+	dr_ste_v1_build_eth_l2_src_dst_bit_mask(mask, sb->inner, sb->bit_mask);
+
+	sb->lu_type = DR_STE_CALC_DFNR_TYPE(ETHL2_SRC_DST, sb->inner);
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v1_build_eth_l2_src_dst_tag;
+}
+
+static int dr_ste_v1_build_eth_l3_ipv6_dst_tag(struct mlx5dr_match_param *value,
+					       struct mlx5dr_ste_build *sb,
+					       u8 *tag)
+{
+	struct mlx5dr_match_spec *spec = sb->inner ? &value->inner : &value->outer;
+
+	DR_STE_SET_TAG(eth_l3_ipv6_dst, tag, dst_ip_127_96, spec, dst_ip_127_96);
+	DR_STE_SET_TAG(eth_l3_ipv6_dst, tag, dst_ip_95_64, spec, dst_ip_95_64);
+	DR_STE_SET_TAG(eth_l3_ipv6_dst, tag, dst_ip_63_32, spec, dst_ip_63_32);
+	DR_STE_SET_TAG(eth_l3_ipv6_dst, tag, dst_ip_31_0, spec, dst_ip_31_0);
+
+	return 0;
+}
+
+void dr_ste_v1_build_eth_l3_ipv6_dst_init(struct mlx5dr_ste_build *sb,
+					  struct mlx5dr_match_param *mask)
+{
+	dr_ste_v1_build_eth_l3_ipv6_dst_tag(mask, sb, sb->bit_mask);
+
+	sb->lu_type = DR_STE_CALC_DFNR_TYPE(IPV6_DES, sb->inner);
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v1_build_eth_l3_ipv6_dst_tag;
+}
+
+static int dr_ste_v1_build_eth_l3_ipv6_src_tag(struct mlx5dr_match_param *value,
+					       struct mlx5dr_ste_build *sb,
+					       u8 *tag)
+{
+	struct mlx5dr_match_spec *spec = sb->inner ? &value->inner : &value->outer;
+
+	DR_STE_SET_TAG(eth_l3_ipv6_src, tag, src_ip_127_96, spec, src_ip_127_96);
+	DR_STE_SET_TAG(eth_l3_ipv6_src, tag, src_ip_95_64, spec, src_ip_95_64);
+	DR_STE_SET_TAG(eth_l3_ipv6_src, tag, src_ip_63_32, spec, src_ip_63_32);
+	DR_STE_SET_TAG(eth_l3_ipv6_src, tag, src_ip_31_0, spec, src_ip_31_0);
+
+	return 0;
+}
+
+void dr_ste_v1_build_eth_l3_ipv6_src_init(struct mlx5dr_ste_build *sb,
+					  struct mlx5dr_match_param *mask)
+{
+	dr_ste_v1_build_eth_l3_ipv6_src_tag(mask, sb, sb->bit_mask);
+
+	sb->lu_type = DR_STE_CALC_DFNR_TYPE(IPV6_SRC, sb->inner);
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v1_build_eth_l3_ipv6_src_tag;
+}
+
+static int dr_ste_v1_build_eth_l3_ipv4_5_tuple_tag(struct mlx5dr_match_param *value,
+						   struct mlx5dr_ste_build *sb,
+						   u8 *tag)
+{
+	struct mlx5dr_match_spec *spec = sb->inner ? &value->inner : &value->outer;
+
+	DR_STE_SET_TAG(eth_l3_ipv4_5_tuple_v1, tag, destination_address, spec, dst_ip_31_0);
+	DR_STE_SET_TAG(eth_l3_ipv4_5_tuple_v1, tag, source_address, spec, src_ip_31_0);
+	DR_STE_SET_TAG(eth_l3_ipv4_5_tuple_v1, tag, destination_port, spec, tcp_dport);
+	DR_STE_SET_TAG(eth_l3_ipv4_5_tuple_v1, tag, destination_port, spec, udp_dport);
+	DR_STE_SET_TAG(eth_l3_ipv4_5_tuple_v1, tag, source_port, spec, tcp_sport);
+	DR_STE_SET_TAG(eth_l3_ipv4_5_tuple_v1, tag, source_port, spec, udp_sport);
+	DR_STE_SET_TAG(eth_l3_ipv4_5_tuple_v1, tag, protocol, spec, ip_protocol);
+	DR_STE_SET_TAG(eth_l3_ipv4_5_tuple_v1, tag, fragmented, spec, frag);
+	DR_STE_SET_TAG(eth_l3_ipv4_5_tuple_v1, tag, dscp, spec, ip_dscp);
+	DR_STE_SET_TAG(eth_l3_ipv4_5_tuple_v1, tag, ecn, spec, ip_ecn);
+
+	if (spec->tcp_flags) {
+		DR_STE_SET_TCP_FLAGS(eth_l3_ipv4_5_tuple_v1, tag, spec);
+		spec->tcp_flags = 0;
+	}
+
+	return 0;
+}
+
+void dr_ste_v1_build_eth_l3_ipv4_5_tuple_init(struct mlx5dr_ste_build *sb,
+					      struct mlx5dr_match_param *mask)
+{
+	dr_ste_v1_build_eth_l3_ipv4_5_tuple_tag(mask, sb, sb->bit_mask);
+
+	sb->lu_type = DR_STE_CALC_DFNR_TYPE(ETHL3_IPV4_5_TUPLE, sb->inner);
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v1_build_eth_l3_ipv4_5_tuple_tag;
+}
+
+static void dr_ste_v1_build_eth_l2_src_or_dst_bit_mask(struct mlx5dr_match_param *value,
+						       bool inner, u8 *bit_mask)
+{
+	struct mlx5dr_match_spec *mask = inner ? &value->inner : &value->outer;
+	struct mlx5dr_match_misc *misc_mask = &value->misc;
+
+	DR_STE_SET_TAG(eth_l2_src_v1, bit_mask, first_vlan_id, mask, first_vid);
+	DR_STE_SET_TAG(eth_l2_src_v1, bit_mask, first_cfi, mask, first_cfi);
+	DR_STE_SET_TAG(eth_l2_src_v1, bit_mask, first_priority, mask, first_prio);
+	DR_STE_SET_TAG(eth_l2_src_v1, bit_mask, ip_fragmented, mask, frag);
+	DR_STE_SET_TAG(eth_l2_src_v1, bit_mask, l3_ethertype, mask, ethertype);
+	DR_STE_SET_ONES(eth_l2_src_v1, bit_mask, l3_type, mask, ip_version);
+
+	if (mask->svlan_tag || mask->cvlan_tag) {
+		MLX5_SET(ste_eth_l2_src_v1, bit_mask, first_vlan_qualifier, -1);
+		mask->cvlan_tag = 0;
+		mask->svlan_tag = 0;
+	}
+
+	if (inner) {
+		if (misc_mask->inner_second_cvlan_tag ||
+		    misc_mask->inner_second_svlan_tag) {
+			MLX5_SET(ste_eth_l2_src_v1, bit_mask, second_vlan_qualifier, -1);
+			misc_mask->inner_second_cvlan_tag = 0;
+			misc_mask->inner_second_svlan_tag = 0;
+		}
+
+		DR_STE_SET_TAG(eth_l2_src_v1, bit_mask,
+			       second_vlan_id, misc_mask, inner_second_vid);
+		DR_STE_SET_TAG(eth_l2_src_v1, bit_mask,
+			       second_cfi, misc_mask, inner_second_cfi);
+		DR_STE_SET_TAG(eth_l2_src_v1, bit_mask,
+			       second_priority, misc_mask, inner_second_prio);
+	} else {
+		if (misc_mask->outer_second_cvlan_tag ||
+		    misc_mask->outer_second_svlan_tag) {
+			MLX5_SET(ste_eth_l2_src_v1, bit_mask, second_vlan_qualifier, -1);
+			misc_mask->outer_second_cvlan_tag = 0;
+			misc_mask->outer_second_svlan_tag = 0;
+		}
+
+		DR_STE_SET_TAG(eth_l2_src_v1, bit_mask,
+			       second_vlan_id, misc_mask, outer_second_vid);
+		DR_STE_SET_TAG(eth_l2_src_v1, bit_mask,
+			       second_cfi, misc_mask, outer_second_cfi);
+		DR_STE_SET_TAG(eth_l2_src_v1, bit_mask,
+			       second_priority, misc_mask, outer_second_prio);
+	}
+}
+
+static int dr_ste_v1_build_eth_l2_src_or_dst_tag(struct mlx5dr_match_param *value,
+						 bool inner, u8 *tag)
+{
+	struct mlx5dr_match_spec *spec = inner ? &value->inner : &value->outer;
+	struct mlx5dr_match_misc *misc_spec = &value->misc;
+
+	DR_STE_SET_TAG(eth_l2_src_v1, tag, first_vlan_id, spec, first_vid);
+	DR_STE_SET_TAG(eth_l2_src_v1, tag, first_cfi, spec, first_cfi);
+	DR_STE_SET_TAG(eth_l2_src_v1, tag, first_priority, spec, first_prio);
+	DR_STE_SET_TAG(eth_l2_src_v1, tag, ip_fragmented, spec, frag);
+	DR_STE_SET_TAG(eth_l2_src_v1, tag, l3_ethertype, spec, ethertype);
+
+	if (spec->ip_version == IP_VERSION_IPV4) {
+		MLX5_SET(ste_eth_l2_src_v1, tag, l3_type, STE_IPV4);
+		spec->ip_version = 0;
+	} else if (spec->ip_version == IP_VERSION_IPV6) {
+		MLX5_SET(ste_eth_l2_src_v1, tag, l3_type, STE_IPV6);
+		spec->ip_version = 0;
+	} else if (spec->ip_version) {
+		return -EINVAL;
+	}
+
+	if (spec->cvlan_tag) {
+		MLX5_SET(ste_eth_l2_src_v1, tag, first_vlan_qualifier, DR_STE_CVLAN);
+		spec->cvlan_tag = 0;
+	} else if (spec->svlan_tag) {
+		MLX5_SET(ste_eth_l2_src_v1, tag, first_vlan_qualifier, DR_STE_SVLAN);
+		spec->svlan_tag = 0;
+	}
+
+	if (inner) {
+		if (misc_spec->inner_second_cvlan_tag) {
+			MLX5_SET(ste_eth_l2_src_v1, tag, second_vlan_qualifier, DR_STE_CVLAN);
+			misc_spec->inner_second_cvlan_tag = 0;
+		} else if (misc_spec->inner_second_svlan_tag) {
+			MLX5_SET(ste_eth_l2_src_v1, tag, second_vlan_qualifier, DR_STE_SVLAN);
+			misc_spec->inner_second_svlan_tag = 0;
+		}
+
+		DR_STE_SET_TAG(eth_l2_src_v1, tag, second_vlan_id, misc_spec, inner_second_vid);
+		DR_STE_SET_TAG(eth_l2_src_v1, tag, second_cfi, misc_spec, inner_second_cfi);
+		DR_STE_SET_TAG(eth_l2_src_v1, tag, second_priority, misc_spec, inner_second_prio);
+	} else {
+		if (misc_spec->outer_second_cvlan_tag) {
+			MLX5_SET(ste_eth_l2_src_v1, tag, second_vlan_qualifier, DR_STE_CVLAN);
+			misc_spec->outer_second_cvlan_tag = 0;
+		} else if (misc_spec->outer_second_svlan_tag) {
+			MLX5_SET(ste_eth_l2_src_v1, tag, second_vlan_qualifier, DR_STE_SVLAN);
+			misc_spec->outer_second_svlan_tag = 0;
+		}
+		DR_STE_SET_TAG(eth_l2_src_v1, tag, second_vlan_id, misc_spec, outer_second_vid);
+		DR_STE_SET_TAG(eth_l2_src_v1, tag, second_cfi, misc_spec, outer_second_cfi);
+		DR_STE_SET_TAG(eth_l2_src_v1, tag, second_priority, misc_spec, outer_second_prio);
+	}
+
+	return 0;
+}
+
+static void dr_ste_v1_build_eth_l2_src_bit_mask(struct mlx5dr_match_param *value,
+						bool inner, u8 *bit_mask)
+{
+	struct mlx5dr_match_spec *mask = inner ? &value->inner : &value->outer;
+
+	DR_STE_SET_TAG(eth_l2_src_v1, bit_mask, smac_47_16, mask, smac_47_16);
+	DR_STE_SET_TAG(eth_l2_src_v1, bit_mask, smac_15_0, mask, smac_15_0);
+
+	dr_ste_v1_build_eth_l2_src_or_dst_bit_mask(value, inner, bit_mask);
+}
+
+static int dr_ste_v1_build_eth_l2_src_tag(struct mlx5dr_match_param *value,
+					  struct mlx5dr_ste_build *sb,
+					  u8 *tag)
+{
+	struct mlx5dr_match_spec *spec = sb->inner ? &value->inner : &value->outer;
+
+	DR_STE_SET_TAG(eth_l2_src_v1, tag, smac_47_16, spec, smac_47_16);
+	DR_STE_SET_TAG(eth_l2_src_v1, tag, smac_15_0, spec, smac_15_0);
+
+	return dr_ste_v1_build_eth_l2_src_or_dst_tag(value, sb->inner, tag);
+}
+
+void dr_ste_v1_build_eth_l2_src_init(struct mlx5dr_ste_build *sb,
+				     struct mlx5dr_match_param *mask)
+{
+	dr_ste_v1_build_eth_l2_src_bit_mask(mask, sb->inner, sb->bit_mask);
+
+	sb->lu_type = DR_STE_CALC_DFNR_TYPE(ETHL2_SRC, sb->inner);
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v1_build_eth_l2_src_tag;
+}
+
+static void dr_ste_v1_build_eth_l2_dst_bit_mask(struct mlx5dr_match_param *value,
+						bool inner, u8 *bit_mask)
+{
+	struct mlx5dr_match_spec *mask = inner ? &value->inner : &value->outer;
+
+	DR_STE_SET_TAG(eth_l2_dst_v1, bit_mask, dmac_47_16, mask, dmac_47_16);
+	DR_STE_SET_TAG(eth_l2_dst_v1, bit_mask, dmac_15_0, mask, dmac_15_0);
+
+	dr_ste_v1_build_eth_l2_src_or_dst_bit_mask(value, inner, bit_mask);
+}
+
+static int dr_ste_v1_build_eth_l2_dst_tag(struct mlx5dr_match_param *value,
+					  struct mlx5dr_ste_build *sb,
+					  u8 *tag)
+{
+	struct mlx5dr_match_spec *spec = sb->inner ? &value->inner : &value->outer;
+
+	DR_STE_SET_TAG(eth_l2_dst_v1, tag, dmac_47_16, spec, dmac_47_16);
+	DR_STE_SET_TAG(eth_l2_dst_v1, tag, dmac_15_0, spec, dmac_15_0);
+
+	return dr_ste_v1_build_eth_l2_src_or_dst_tag(value, sb->inner, tag);
+}
+
+void dr_ste_v1_build_eth_l2_dst_init(struct mlx5dr_ste_build *sb,
+				     struct mlx5dr_match_param *mask)
+{
+	dr_ste_v1_build_eth_l2_dst_bit_mask(mask, sb->inner, sb->bit_mask);
+
+	sb->lu_type = DR_STE_CALC_DFNR_TYPE(ETHL2, sb->inner);
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v1_build_eth_l2_dst_tag;
+}
+
+static void dr_ste_v1_build_eth_l2_tnl_bit_mask(struct mlx5dr_match_param *value,
+						bool inner, u8 *bit_mask)
+{
+	struct mlx5dr_match_spec *mask = inner ? &value->inner : &value->outer;
+	struct mlx5dr_match_misc *misc = &value->misc;
+
+	DR_STE_SET_TAG(eth_l2_tnl_v1, bit_mask, dmac_47_16, mask, dmac_47_16);
+	DR_STE_SET_TAG(eth_l2_tnl_v1, bit_mask, dmac_15_0, mask, dmac_15_0);
+	DR_STE_SET_TAG(eth_l2_tnl_v1, bit_mask, first_vlan_id, mask, first_vid);
+	DR_STE_SET_TAG(eth_l2_tnl_v1, bit_mask, first_cfi, mask, first_cfi);
+	DR_STE_SET_TAG(eth_l2_tnl_v1, bit_mask, first_priority, mask, first_prio);
+	DR_STE_SET_TAG(eth_l2_tnl_v1, bit_mask, ip_fragmented, mask, frag);
+	DR_STE_SET_TAG(eth_l2_tnl_v1, bit_mask, l3_ethertype, mask, ethertype);
+	DR_STE_SET_ONES(eth_l2_tnl_v1, bit_mask, l3_type, mask, ip_version);
+
+	if (misc->vxlan_vni) {
+		MLX5_SET(ste_eth_l2_tnl_v1, bit_mask,
+			 l2_tunneling_network_id, (misc->vxlan_vni << 8));
+		misc->vxlan_vni = 0;
+	}
+
+	if (mask->svlan_tag || mask->cvlan_tag) {
+		MLX5_SET(ste_eth_l2_tnl_v1, bit_mask, first_vlan_qualifier, -1);
+		mask->cvlan_tag = 0;
+		mask->svlan_tag = 0;
+	}
+}
+
+static int dr_ste_v1_build_eth_l2_tnl_tag(struct mlx5dr_match_param *value,
+					  struct mlx5dr_ste_build *sb,
+					  u8 *tag)
+{
+	struct mlx5dr_match_spec *spec = sb->inner ? &value->inner : &value->outer;
+	struct mlx5dr_match_misc *misc = &value->misc;
+
+	DR_STE_SET_TAG(eth_l2_tnl_v1, tag, dmac_47_16, spec, dmac_47_16);
+	DR_STE_SET_TAG(eth_l2_tnl_v1, tag, dmac_15_0, spec, dmac_15_0);
+	DR_STE_SET_TAG(eth_l2_tnl_v1, tag, first_vlan_id, spec, first_vid);
+	DR_STE_SET_TAG(eth_l2_tnl_v1, tag, first_cfi, spec, first_cfi);
+	DR_STE_SET_TAG(eth_l2_tnl_v1, tag, ip_fragmented, spec, frag);
+	DR_STE_SET_TAG(eth_l2_tnl_v1, tag, first_priority, spec, first_prio);
+	DR_STE_SET_TAG(eth_l2_tnl_v1, tag, l3_ethertype, spec, ethertype);
+
+	if (misc->vxlan_vni) {
+		MLX5_SET(ste_eth_l2_tnl_v1, tag, l2_tunneling_network_id,
+			 (misc->vxlan_vni << 8));
+		misc->vxlan_vni = 0;
+	}
+
+	if (spec->cvlan_tag) {
+		MLX5_SET(ste_eth_l2_tnl_v1, tag, first_vlan_qualifier, DR_STE_CVLAN);
+		spec->cvlan_tag = 0;
+	} else if (spec->svlan_tag) {
+		MLX5_SET(ste_eth_l2_tnl_v1, tag, first_vlan_qualifier, DR_STE_SVLAN);
+		spec->svlan_tag = 0;
+	}
+
+	if (spec->ip_version == IP_VERSION_IPV4) {
+		MLX5_SET(ste_eth_l2_tnl_v1, tag, l3_type, STE_IPV4);
+		spec->ip_version = 0;
+	} else if (spec->ip_version == IP_VERSION_IPV6) {
+		MLX5_SET(ste_eth_l2_tnl_v1, tag, l3_type, STE_IPV6);
+		spec->ip_version = 0;
+	} else if (spec->ip_version) {
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+void dr_ste_v1_build_eth_l2_tnl_init(struct mlx5dr_ste_build *sb,
+				     struct mlx5dr_match_param *mask)
+{
+	dr_ste_v1_build_eth_l2_tnl_bit_mask(mask, sb->inner, sb->bit_mask);
+
+	sb->lu_type = DR_STE_V1_LU_TYPE_ETHL2_TNL;
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v1_build_eth_l2_tnl_tag;
+}
+
+static int dr_ste_v1_build_eth_l3_ipv4_misc_tag(struct mlx5dr_match_param *value,
+						struct mlx5dr_ste_build *sb,
+						u8 *tag)
+{
+	struct mlx5dr_match_spec *spec = sb->inner ? &value->inner : &value->outer;
+
+	DR_STE_SET_TAG(eth_l3_ipv4_misc_v1, tag, time_to_live, spec, ttl_hoplimit);
+	DR_STE_SET_TAG(eth_l3_ipv4_misc_v1, tag, ihl, spec, ipv4_ihl);
+
+	return 0;
+}
+
+void dr_ste_v1_build_eth_l3_ipv4_misc_init(struct mlx5dr_ste_build *sb,
+					   struct mlx5dr_match_param *mask)
+{
+	dr_ste_v1_build_eth_l3_ipv4_misc_tag(mask, sb, sb->bit_mask);
+
+	sb->lu_type = DR_STE_CALC_DFNR_TYPE(ETHL3_IPV4_MISC, sb->inner);
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v1_build_eth_l3_ipv4_misc_tag;
+}
+
+static int dr_ste_v1_build_eth_ipv6_l3_l4_tag(struct mlx5dr_match_param *value,
+					      struct mlx5dr_ste_build *sb,
+					      u8 *tag)
+{
+	struct mlx5dr_match_spec *spec = sb->inner ? &value->inner : &value->outer;
+	struct mlx5dr_match_misc *misc = &value->misc;
+
+	DR_STE_SET_TAG(eth_l4_v1, tag, dst_port, spec, tcp_dport);
+	DR_STE_SET_TAG(eth_l4_v1, tag, src_port, spec, tcp_sport);
+	DR_STE_SET_TAG(eth_l4_v1, tag, dst_port, spec, udp_dport);
+	DR_STE_SET_TAG(eth_l4_v1, tag, src_port, spec, udp_sport);
+	DR_STE_SET_TAG(eth_l4_v1, tag, protocol, spec, ip_protocol);
+	DR_STE_SET_TAG(eth_l4_v1, tag, fragmented, spec, frag);
+	DR_STE_SET_TAG(eth_l4_v1, tag, dscp, spec, ip_dscp);
+	DR_STE_SET_TAG(eth_l4_v1, tag, ecn, spec, ip_ecn);
+	DR_STE_SET_TAG(eth_l4_v1, tag, ipv6_hop_limit, spec, ttl_hoplimit);
+
+	if (sb->inner)
+		DR_STE_SET_TAG(eth_l4_v1, tag, flow_label, misc, inner_ipv6_flow_label);
+	else
+		DR_STE_SET_TAG(eth_l4_v1, tag, flow_label, misc, outer_ipv6_flow_label);
+
+	if (spec->tcp_flags) {
+		DR_STE_SET_TCP_FLAGS(eth_l4_v1, tag, spec);
+		spec->tcp_flags = 0;
+	}
+
+	return 0;
+}
+
+void dr_ste_v1_build_eth_ipv6_l3_l4_init(struct mlx5dr_ste_build *sb,
+					 struct mlx5dr_match_param *mask)
+{
+	dr_ste_v1_build_eth_ipv6_l3_l4_tag(mask, sb, sb->bit_mask);
+
+	sb->lu_type = DR_STE_CALC_DFNR_TYPE(ETHL4, sb->inner);
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v1_build_eth_ipv6_l3_l4_tag;
+}
+
+static int dr_ste_v1_build_mpls_tag(struct mlx5dr_match_param *value,
+				    struct mlx5dr_ste_build *sb,
+				    u8 *tag)
+{
+	struct mlx5dr_match_misc2 *misc2 = &value->misc2;
+
+	if (sb->inner)
+		DR_STE_SET_MPLS(mpls_v1, misc2, inner, tag);
+	else
+		DR_STE_SET_MPLS(mpls_v1, misc2, outer, tag);
+
+	return 0;
+}
+
+void dr_ste_v1_build_mpls_init(struct mlx5dr_ste_build *sb,
+			       struct mlx5dr_match_param *mask)
+{
+	dr_ste_v1_build_mpls_tag(mask, sb, sb->bit_mask);
+
+	sb->lu_type = DR_STE_CALC_DFNR_TYPE(MPLS, sb->inner);
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v1_build_mpls_tag;
+}
+
+static int dr_ste_v1_build_tnl_gre_tag(struct mlx5dr_match_param *value,
+				       struct mlx5dr_ste_build *sb,
+				       u8 *tag)
+{
+	struct  mlx5dr_match_misc *misc = &value->misc;
+
+	DR_STE_SET_TAG(gre_v1, tag, gre_protocol, misc, gre_protocol);
+	DR_STE_SET_TAG(gre_v1, tag, gre_k_present, misc, gre_k_present);
+	DR_STE_SET_TAG(gre_v1, tag, gre_key_h, misc, gre_key_h);
+	DR_STE_SET_TAG(gre_v1, tag, gre_key_l, misc, gre_key_l);
+
+	DR_STE_SET_TAG(gre_v1, tag, gre_c_present, misc, gre_c_present);
+	DR_STE_SET_TAG(gre_v1, tag, gre_s_present, misc, gre_s_present);
+
+	return 0;
+}
+
+void dr_ste_v1_build_tnl_gre_init(struct mlx5dr_ste_build *sb,
+				  struct mlx5dr_match_param *mask)
+{
+	dr_ste_v1_build_tnl_gre_tag(mask, sb, sb->bit_mask);
+
+	sb->lu_type = DR_STE_V1_LU_TYPE_GRE;
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v1_build_tnl_gre_tag;
+}
+
+static int dr_ste_v1_build_tnl_mpls_tag(struct mlx5dr_match_param *value,
+					struct mlx5dr_ste_build *sb,
+					u8 *tag)
+{
+	struct mlx5dr_match_misc2 *misc2 = &value->misc2;
+
+	if (DR_STE_IS_OUTER_MPLS_OVER_GRE_SET(misc2)) {
+		DR_STE_SET_TAG(mpls_v1, tag, mpls0_label,
+			       misc2, outer_first_mpls_over_gre_label);
+
+		DR_STE_SET_TAG(mpls_v1, tag, mpls0_exp,
+			       misc2, outer_first_mpls_over_gre_exp);
+
+		DR_STE_SET_TAG(mpls_v1, tag, mpls0_s_bos,
+			       misc2, outer_first_mpls_over_gre_s_bos);
+
+		DR_STE_SET_TAG(mpls_v1, tag, mpls0_ttl,
+			       misc2, outer_first_mpls_over_gre_ttl);
+	} else {
+		DR_STE_SET_TAG(mpls_v1, tag, mpls0_label,
+			       misc2, outer_first_mpls_over_udp_label);
+
+		DR_STE_SET_TAG(mpls_v1, tag, mpls0_exp,
+			       misc2, outer_first_mpls_over_udp_exp);
+
+		DR_STE_SET_TAG(mpls_v1, tag, mpls0_s_bos,
+			       misc2, outer_first_mpls_over_udp_s_bos);
+
+		DR_STE_SET_TAG(mpls_v1, tag, mpls0_ttl,
+			       misc2, outer_first_mpls_over_udp_ttl);
+	}
+
+	return 0;
+}
+
+void dr_ste_v1_build_tnl_mpls_init(struct mlx5dr_ste_build *sb,
+				   struct mlx5dr_match_param *mask)
+{
+	dr_ste_v1_build_tnl_mpls_tag(mask, sb, sb->bit_mask);
+
+	sb->lu_type = DR_STE_V1_LU_TYPE_MPLS_I;
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v1_build_tnl_mpls_tag;
+}
+
+static int dr_ste_v1_build_tnl_mpls_over_udp_tag(struct mlx5dr_match_param *value,
+						 struct mlx5dr_ste_build *sb,
+						 u8 *tag)
+{
+	struct mlx5dr_match_misc2 *misc2 = &value->misc2;
+	u8 *parser_ptr;
+	u8 parser_id;
+	u32 mpls_hdr;
+
+	mpls_hdr = misc2->outer_first_mpls_over_udp_label << HDR_MPLS_OFFSET_LABEL;
+	misc2->outer_first_mpls_over_udp_label = 0;
+	mpls_hdr |= misc2->outer_first_mpls_over_udp_exp << HDR_MPLS_OFFSET_EXP;
+	misc2->outer_first_mpls_over_udp_exp = 0;
+	mpls_hdr |= misc2->outer_first_mpls_over_udp_s_bos << HDR_MPLS_OFFSET_S_BOS;
+	misc2->outer_first_mpls_over_udp_s_bos = 0;
+	mpls_hdr |= misc2->outer_first_mpls_over_udp_ttl << HDR_MPLS_OFFSET_TTL;
+	misc2->outer_first_mpls_over_udp_ttl = 0;
+
+	parser_id = sb->caps->flex_parser_id_mpls_over_udp;
+	parser_ptr = dr_ste_calc_flex_parser_offset(tag, parser_id);
+	*(__be32 *)parser_ptr = cpu_to_be32(mpls_hdr);
+
+	return 0;
+}
+
+void dr_ste_v1_build_tnl_mpls_over_udp_init(struct mlx5dr_ste_build *sb,
+					    struct mlx5dr_match_param *mask)
+{
+	dr_ste_v1_build_tnl_mpls_over_udp_tag(mask, sb, sb->bit_mask);
+
+	/* STEs with lookup type FLEX_PARSER_{0/1} includes
+	 * flex parsers_{0-3}/{4-7} respectively.
+	 */
+	sb->lu_type = sb->caps->flex_parser_id_mpls_over_udp > DR_STE_MAX_FLEX_0_ID ?
+		      DR_STE_V1_LU_TYPE_FLEX_PARSER_1 :
+		      DR_STE_V1_LU_TYPE_FLEX_PARSER_0;
+
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v1_build_tnl_mpls_over_udp_tag;
+}
+
+static int dr_ste_v1_build_tnl_mpls_over_gre_tag(struct mlx5dr_match_param *value,
+						 struct mlx5dr_ste_build *sb,
+						 u8 *tag)
+{
+	struct mlx5dr_match_misc2 *misc2 = &value->misc2;
+	u8 *parser_ptr;
+	u8 parser_id;
+	u32 mpls_hdr;
+
+	mpls_hdr = misc2->outer_first_mpls_over_gre_label << HDR_MPLS_OFFSET_LABEL;
+	misc2->outer_first_mpls_over_gre_label = 0;
+	mpls_hdr |= misc2->outer_first_mpls_over_gre_exp << HDR_MPLS_OFFSET_EXP;
+	misc2->outer_first_mpls_over_gre_exp = 0;
+	mpls_hdr |= misc2->outer_first_mpls_over_gre_s_bos << HDR_MPLS_OFFSET_S_BOS;
+	misc2->outer_first_mpls_over_gre_s_bos = 0;
+	mpls_hdr |= misc2->outer_first_mpls_over_gre_ttl << HDR_MPLS_OFFSET_TTL;
+	misc2->outer_first_mpls_over_gre_ttl = 0;
+
+	parser_id = sb->caps->flex_parser_id_mpls_over_gre;
+	parser_ptr = dr_ste_calc_flex_parser_offset(tag, parser_id);
+	*(__be32 *)parser_ptr = cpu_to_be32(mpls_hdr);
+
+	return 0;
+}
+
+void dr_ste_v1_build_tnl_mpls_over_gre_init(struct mlx5dr_ste_build *sb,
+					    struct mlx5dr_match_param *mask)
+{
+	dr_ste_v1_build_tnl_mpls_over_gre_tag(mask, sb, sb->bit_mask);
+
+	/* STEs with lookup type FLEX_PARSER_{0/1} includes
+	 * flex parsers_{0-3}/{4-7} respectively.
+	 */
+	sb->lu_type = sb->caps->flex_parser_id_mpls_over_gre > DR_STE_MAX_FLEX_0_ID ?
+		      DR_STE_V1_LU_TYPE_FLEX_PARSER_1 :
+		      DR_STE_V1_LU_TYPE_FLEX_PARSER_0;
+
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v1_build_tnl_mpls_over_gre_tag;
+}
+
+static int dr_ste_v1_build_icmp_tag(struct mlx5dr_match_param *value,
+				    struct mlx5dr_ste_build *sb,
+				    u8 *tag)
+{
+	struct mlx5dr_match_misc3 *misc3 = &value->misc3;
+	bool is_ipv4 = DR_MASK_IS_ICMPV4_SET(misc3);
+	u32 *icmp_header_data;
+	u8 *icmp_type;
+	u8 *icmp_code;
+
+	if (is_ipv4) {
+		icmp_header_data	= &misc3->icmpv4_header_data;
+		icmp_type		= &misc3->icmpv4_type;
+		icmp_code		= &misc3->icmpv4_code;
+	} else {
+		icmp_header_data	= &misc3->icmpv6_header_data;
+		icmp_type		= &misc3->icmpv6_type;
+		icmp_code		= &misc3->icmpv6_code;
+	}
+
+	MLX5_SET(ste_icmp_v1, tag, icmp_header_data, *icmp_header_data);
+	MLX5_SET(ste_icmp_v1, tag, icmp_type, *icmp_type);
+	MLX5_SET(ste_icmp_v1, tag, icmp_code, *icmp_code);
+
+	*icmp_header_data = 0;
+	*icmp_type = 0;
+	*icmp_code = 0;
+
+	return 0;
+}
+
+void dr_ste_v1_build_icmp_init(struct mlx5dr_ste_build *sb,
+			       struct mlx5dr_match_param *mask)
+{
+	dr_ste_v1_build_icmp_tag(mask, sb, sb->bit_mask);
+
+	sb->lu_type = DR_STE_V1_LU_TYPE_ETHL4_MISC_O;
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v1_build_icmp_tag;
+}
+
+static int dr_ste_v1_build_general_purpose_tag(struct mlx5dr_match_param *value,
+					       struct mlx5dr_ste_build *sb,
+					       u8 *tag)
+{
+	struct mlx5dr_match_misc2 *misc2 = &value->misc2;
+
+	DR_STE_SET_TAG(general_purpose, tag, general_purpose_lookup_field,
+		       misc2, metadata_reg_a);
+
+	return 0;
+}
+
+void dr_ste_v1_build_general_purpose_init(struct mlx5dr_ste_build *sb,
+					  struct mlx5dr_match_param *mask)
+{
+	dr_ste_v1_build_general_purpose_tag(mask, sb, sb->bit_mask);
+
+	sb->lu_type = DR_STE_V1_LU_TYPE_GENERAL_PURPOSE;
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v1_build_general_purpose_tag;
+}
+
+static int dr_ste_v1_build_eth_l4_misc_tag(struct mlx5dr_match_param *value,
+					   struct mlx5dr_ste_build *sb,
+					   u8 *tag)
+{
+	struct mlx5dr_match_misc3 *misc3 = &value->misc3;
+
+	if (sb->inner) {
+		DR_STE_SET_TAG(eth_l4_misc_v1, tag, seq_num, misc3, inner_tcp_seq_num);
+		DR_STE_SET_TAG(eth_l4_misc_v1, tag, ack_num, misc3, inner_tcp_ack_num);
+	} else {
+		DR_STE_SET_TAG(eth_l4_misc_v1, tag, seq_num, misc3, outer_tcp_seq_num);
+		DR_STE_SET_TAG(eth_l4_misc_v1, tag, ack_num, misc3, outer_tcp_ack_num);
+	}
+
+	return 0;
+}
+
+void dr_ste_v1_build_eth_l4_misc_init(struct mlx5dr_ste_build *sb,
+				      struct mlx5dr_match_param *mask)
+{
+	dr_ste_v1_build_eth_l4_misc_tag(mask, sb, sb->bit_mask);
+
+	sb->lu_type = DR_STE_V1_LU_TYPE_ETHL4_MISC_O;
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v1_build_eth_l4_misc_tag;
+}
+
+static int
+dr_ste_v1_build_flex_parser_tnl_vxlan_gpe_tag(struct mlx5dr_match_param *value,
+					      struct mlx5dr_ste_build *sb,
+					      u8 *tag)
+{
+	struct mlx5dr_match_misc3 *misc3 = &value->misc3;
+
+	DR_STE_SET_TAG(flex_parser_tnl_vxlan_gpe, tag,
+		       outer_vxlan_gpe_flags, misc3,
+		       outer_vxlan_gpe_flags);
+	DR_STE_SET_TAG(flex_parser_tnl_vxlan_gpe, tag,
+		       outer_vxlan_gpe_next_protocol, misc3,
+		       outer_vxlan_gpe_next_protocol);
+	DR_STE_SET_TAG(flex_parser_tnl_vxlan_gpe, tag,
+		       outer_vxlan_gpe_vni, misc3,
+		       outer_vxlan_gpe_vni);
+
+	return 0;
+}
+
+void dr_ste_v1_build_flex_parser_tnl_vxlan_gpe_init(struct mlx5dr_ste_build *sb,
+						    struct mlx5dr_match_param *mask)
+{
+	dr_ste_v1_build_flex_parser_tnl_vxlan_gpe_tag(mask, sb, sb->bit_mask);
+
+	sb->lu_type = DR_STE_V1_LU_TYPE_FLEX_PARSER_TNL_HEADER;
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v1_build_flex_parser_tnl_vxlan_gpe_tag;
+}
+
+static int
+dr_ste_v1_build_flex_parser_tnl_geneve_tag(struct mlx5dr_match_param *value,
+					   struct mlx5dr_ste_build *sb,
+					   u8 *tag)
+{
+	struct mlx5dr_match_misc *misc = &value->misc;
+
+	DR_STE_SET_TAG(flex_parser_tnl_geneve, tag,
+		       geneve_protocol_type, misc, geneve_protocol_type);
+	DR_STE_SET_TAG(flex_parser_tnl_geneve, tag,
+		       geneve_oam, misc, geneve_oam);
+	DR_STE_SET_TAG(flex_parser_tnl_geneve, tag,
+		       geneve_opt_len, misc, geneve_opt_len);
+	DR_STE_SET_TAG(flex_parser_tnl_geneve, tag,
+		       geneve_vni, misc, geneve_vni);
+
+	return 0;
+}
+
+void dr_ste_v1_build_flex_parser_tnl_geneve_init(struct mlx5dr_ste_build *sb,
+						 struct mlx5dr_match_param *mask)
+{
+	dr_ste_v1_build_flex_parser_tnl_geneve_tag(mask, sb, sb->bit_mask);
+
+	sb->lu_type = DR_STE_V1_LU_TYPE_FLEX_PARSER_TNL_HEADER;
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v1_build_flex_parser_tnl_geneve_tag;
+}
+
+static int dr_ste_v1_build_tnl_header_0_1_tag(struct mlx5dr_match_param *value,
+					      struct mlx5dr_ste_build *sb,
+					      uint8_t *tag)
+{
+	struct mlx5dr_match_misc5 *misc5 = &value->misc5;
+
+	DR_STE_SET_TAG(tunnel_header, tag, tunnel_header_0, misc5, tunnel_header_0);
+	DR_STE_SET_TAG(tunnel_header, tag, tunnel_header_1, misc5, tunnel_header_1);
+
+	return 0;
+}
+
+void dr_ste_v1_build_tnl_header_0_1_init(struct mlx5dr_ste_build *sb,
+					 struct mlx5dr_match_param *mask)
+{
+	sb->lu_type = DR_STE_V1_LU_TYPE_FLEX_PARSER_TNL_HEADER;
+	dr_ste_v1_build_tnl_header_0_1_tag(mask, sb, sb->bit_mask);
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v1_build_tnl_header_0_1_tag;
+}
+
+static int dr_ste_v1_build_register_0_tag(struct mlx5dr_match_param *value,
+					  struct mlx5dr_ste_build *sb,
+					  u8 *tag)
+{
+	struct mlx5dr_match_misc2 *misc2 = &value->misc2;
+
+	DR_STE_SET_TAG(register_0, tag, register_0_h, misc2, metadata_reg_c_0);
+	DR_STE_SET_TAG(register_0, tag, register_0_l, misc2, metadata_reg_c_1);
+	DR_STE_SET_TAG(register_0, tag, register_1_h, misc2, metadata_reg_c_2);
+	DR_STE_SET_TAG(register_0, tag, register_1_l, misc2, metadata_reg_c_3);
+
+	return 0;
+}
+
+void dr_ste_v1_build_register_0_init(struct mlx5dr_ste_build *sb,
+				     struct mlx5dr_match_param *mask)
+{
+	dr_ste_v1_build_register_0_tag(mask, sb, sb->bit_mask);
+
+	sb->lu_type = DR_STE_V1_LU_TYPE_STEERING_REGISTERS_0;
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v1_build_register_0_tag;
+}
+
+static int dr_ste_v1_build_register_1_tag(struct mlx5dr_match_param *value,
+					  struct mlx5dr_ste_build *sb,
+					  u8 *tag)
+{
+	struct mlx5dr_match_misc2 *misc2 = &value->misc2;
+
+	DR_STE_SET_TAG(register_1, tag, register_2_h, misc2, metadata_reg_c_4);
+	DR_STE_SET_TAG(register_1, tag, register_2_l, misc2, metadata_reg_c_5);
+	DR_STE_SET_TAG(register_1, tag, register_3_h, misc2, metadata_reg_c_6);
+	DR_STE_SET_TAG(register_1, tag, register_3_l, misc2, metadata_reg_c_7);
+
+	return 0;
+}
+
+void dr_ste_v1_build_register_1_init(struct mlx5dr_ste_build *sb,
+				     struct mlx5dr_match_param *mask)
+{
+	dr_ste_v1_build_register_1_tag(mask, sb, sb->bit_mask);
+
+	sb->lu_type = DR_STE_V1_LU_TYPE_STEERING_REGISTERS_1;
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v1_build_register_1_tag;
+}
+
+static void dr_ste_v1_build_src_gvmi_qpn_bit_mask(struct mlx5dr_match_param *value,
+						  u8 *bit_mask)
+{
+	struct mlx5dr_match_misc *misc_mask = &value->misc;
+
+	DR_STE_SET_ONES(src_gvmi_qp_v1, bit_mask, source_gvmi, misc_mask, source_port);
+	DR_STE_SET_ONES(src_gvmi_qp_v1, bit_mask, source_qp, misc_mask, source_sqn);
+	misc_mask->source_eswitch_owner_vhca_id = 0;
+}
+
+static int dr_ste_v1_build_src_gvmi_qpn_tag(struct mlx5dr_match_param *value,
+					    struct mlx5dr_ste_build *sb,
+					    u8 *tag)
+{
+	struct mlx5dr_match_misc *misc = &value->misc;
+	struct mlx5dr_cmd_vport_cap *vport_cap;
+	struct mlx5dr_domain *dmn = sb->dmn;
+	struct mlx5dr_domain *vport_dmn;
+	u8 *bit_mask = sb->bit_mask;
+
+	DR_STE_SET_TAG(src_gvmi_qp_v1, tag, source_qp, misc, source_sqn);
+
+	if (sb->vhca_id_valid) {
+		/* Find port GVMI based on the eswitch_owner_vhca_id */
+		if (misc->source_eswitch_owner_vhca_id == dmn->info.caps.gvmi)
+			vport_dmn = dmn;
+		else if (dmn->peer_dmn && (misc->source_eswitch_owner_vhca_id ==
+					   dmn->peer_dmn->info.caps.gvmi))
+			vport_dmn = dmn->peer_dmn;
+		else
+			return -EINVAL;
+
+		misc->source_eswitch_owner_vhca_id = 0;
+	} else {
+		vport_dmn = dmn;
+	}
+
+	if (!MLX5_GET(ste_src_gvmi_qp_v1, bit_mask, source_gvmi))
+		return 0;
+
+	vport_cap = mlx5dr_domain_get_vport_cap(vport_dmn, misc->source_port);
+	if (!vport_cap) {
+		mlx5dr_err(dmn, "Vport 0x%x is disabled or invalid\n",
+			   misc->source_port);
+		return -EINVAL;
+	}
+
+	if (vport_cap->vport_gvmi)
+		MLX5_SET(ste_src_gvmi_qp_v1, tag, source_gvmi, vport_cap->vport_gvmi);
+
+	misc->source_port = 0;
+	return 0;
+}
+
+void dr_ste_v1_build_src_gvmi_qpn_init(struct mlx5dr_ste_build *sb,
+				       struct mlx5dr_match_param *mask)
+{
+	dr_ste_v1_build_src_gvmi_qpn_bit_mask(mask, sb->bit_mask);
+
+	sb->lu_type = DR_STE_V1_LU_TYPE_SRC_QP_GVMI;
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v1_build_src_gvmi_qpn_tag;
+}
+
+static void dr_ste_v1_set_flex_parser(u32 *misc4_field_id,
+				      u32 *misc4_field_value,
+				      bool *parser_is_used,
+				      u8 *tag)
+{
+	u32 id = *misc4_field_id;
+	u8 *parser_ptr;
+
+	if (id >= DR_NUM_OF_FLEX_PARSERS || parser_is_used[id])
+		return;
+
+	parser_is_used[id] = true;
+	parser_ptr = dr_ste_calc_flex_parser_offset(tag, id);
+
+	*(__be32 *)parser_ptr = cpu_to_be32(*misc4_field_value);
+	*misc4_field_id = 0;
+	*misc4_field_value = 0;
+}
+
+static int dr_ste_v1_build_felx_parser_tag(struct mlx5dr_match_param *value,
+					   struct mlx5dr_ste_build *sb,
+					   u8 *tag)
+{
+	struct mlx5dr_match_misc4 *misc_4_mask = &value->misc4;
+	bool parser_is_used[DR_NUM_OF_FLEX_PARSERS] = {};
+
+	dr_ste_v1_set_flex_parser(&misc_4_mask->prog_sample_field_id_0,
+				  &misc_4_mask->prog_sample_field_value_0,
+				  parser_is_used, tag);
+
+	dr_ste_v1_set_flex_parser(&misc_4_mask->prog_sample_field_id_1,
+				  &misc_4_mask->prog_sample_field_value_1,
+				  parser_is_used, tag);
+
+	dr_ste_v1_set_flex_parser(&misc_4_mask->prog_sample_field_id_2,
+				  &misc_4_mask->prog_sample_field_value_2,
+				  parser_is_used, tag);
+
+	dr_ste_v1_set_flex_parser(&misc_4_mask->prog_sample_field_id_3,
+				  &misc_4_mask->prog_sample_field_value_3,
+				  parser_is_used, tag);
+
+	return 0;
+}
+
+void dr_ste_v1_build_flex_parser_0_init(struct mlx5dr_ste_build *sb,
+					struct mlx5dr_match_param *mask)
+{
+	sb->lu_type = DR_STE_V1_LU_TYPE_FLEX_PARSER_0;
+	dr_ste_v1_build_felx_parser_tag(mask, sb, sb->bit_mask);
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v1_build_felx_parser_tag;
+}
+
+void dr_ste_v1_build_flex_parser_1_init(struct mlx5dr_ste_build *sb,
+					struct mlx5dr_match_param *mask)
+{
+	sb->lu_type = DR_STE_V1_LU_TYPE_FLEX_PARSER_1;
+	dr_ste_v1_build_felx_parser_tag(mask, sb, sb->bit_mask);
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v1_build_felx_parser_tag;
+}
+
+static int
+dr_ste_v1_build_flex_parser_tnl_geneve_tlv_opt_tag(struct mlx5dr_match_param *value,
+						   struct mlx5dr_ste_build *sb,
+						   u8 *tag)
+{
+	struct mlx5dr_match_misc3 *misc3 = &value->misc3;
+	u8 parser_id = sb->caps->flex_parser_id_geneve_tlv_option_0;
+	u8 *parser_ptr = dr_ste_calc_flex_parser_offset(tag, parser_id);
+
+	MLX5_SET(ste_flex_parser_0, parser_ptr, flex_parser_3,
+		 misc3->geneve_tlv_option_0_data);
+	misc3->geneve_tlv_option_0_data = 0;
+
+	return 0;
+}
+
+void
+dr_ste_v1_build_flex_parser_tnl_geneve_tlv_opt_init(struct mlx5dr_ste_build *sb,
+						    struct mlx5dr_match_param *mask)
+{
+	dr_ste_v1_build_flex_parser_tnl_geneve_tlv_opt_tag(mask, sb, sb->bit_mask);
+
+	/* STEs with lookup type FLEX_PARSER_{0/1} includes
+	 * flex parsers_{0-3}/{4-7} respectively.
+	 */
+	sb->lu_type = sb->caps->flex_parser_id_geneve_tlv_option_0 > 3 ?
+		      DR_STE_V1_LU_TYPE_FLEX_PARSER_1 :
+		      DR_STE_V1_LU_TYPE_FLEX_PARSER_0;
+
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v1_build_flex_parser_tnl_geneve_tlv_opt_tag;
+}
+
+static int
+dr_ste_v1_build_flex_parser_tnl_geneve_tlv_opt_exist_tag(struct mlx5dr_match_param *value,
+							 struct mlx5dr_ste_build *sb,
+							 uint8_t *tag)
+{
+	u8 parser_id = sb->caps->flex_parser_id_geneve_tlv_option_0;
+	struct mlx5dr_match_misc *misc = &value->misc;
+
+	if (misc->geneve_tlv_option_0_exist) {
+		MLX5_SET(ste_flex_parser_ok, tag, flex_parsers_ok, 1 << parser_id);
+		misc->geneve_tlv_option_0_exist = 0;
+	}
+
+	return 0;
+}
+
+void
+dr_ste_v1_build_flex_parser_tnl_geneve_tlv_opt_exist_init(struct mlx5dr_ste_build *sb,
+							  struct mlx5dr_match_param *mask)
+{
+	sb->lu_type = DR_STE_V1_LU_TYPE_FLEX_PARSER_OK;
+	dr_ste_v1_build_flex_parser_tnl_geneve_tlv_opt_exist_tag(mask, sb, sb->bit_mask);
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v1_build_flex_parser_tnl_geneve_tlv_opt_exist_tag;
+}
+
+static int dr_ste_v1_build_flex_parser_tnl_gtpu_tag(struct mlx5dr_match_param *value,
+						    struct mlx5dr_ste_build *sb,
+						    u8 *tag)
+{
+	struct mlx5dr_match_misc3 *misc3 = &value->misc3;
+
+	DR_STE_SET_TAG(flex_parser_tnl_gtpu, tag, gtpu_msg_flags, misc3, gtpu_msg_flags);
+	DR_STE_SET_TAG(flex_parser_tnl_gtpu, tag, gtpu_msg_type, misc3, gtpu_msg_type);
+	DR_STE_SET_TAG(flex_parser_tnl_gtpu, tag, gtpu_teid, misc3, gtpu_teid);
+
+	return 0;
+}
+
+void dr_ste_v1_build_flex_parser_tnl_gtpu_init(struct mlx5dr_ste_build *sb,
+					       struct mlx5dr_match_param *mask)
+{
+	dr_ste_v1_build_flex_parser_tnl_gtpu_tag(mask, sb, sb->bit_mask);
+
+	sb->lu_type = DR_STE_V1_LU_TYPE_FLEX_PARSER_TNL_HEADER;
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v1_build_flex_parser_tnl_gtpu_tag;
+}
+
+static int
+dr_ste_v1_build_tnl_gtpu_flex_parser_0_tag(struct mlx5dr_match_param *value,
+					   struct mlx5dr_ste_build *sb,
+					   u8 *tag)
+{
+	if (dr_is_flex_parser_0_id(sb->caps->flex_parser_id_gtpu_dw_0))
+		DR_STE_SET_FLEX_PARSER_FIELD(tag, gtpu_dw_0, sb->caps, &value->misc3);
+	if (dr_is_flex_parser_0_id(sb->caps->flex_parser_id_gtpu_teid))
+		DR_STE_SET_FLEX_PARSER_FIELD(tag, gtpu_teid, sb->caps, &value->misc3);
+	if (dr_is_flex_parser_0_id(sb->caps->flex_parser_id_gtpu_dw_2))
+		DR_STE_SET_FLEX_PARSER_FIELD(tag, gtpu_dw_2, sb->caps, &value->misc3);
+	if (dr_is_flex_parser_0_id(sb->caps->flex_parser_id_gtpu_first_ext_dw_0))
+		DR_STE_SET_FLEX_PARSER_FIELD(tag, gtpu_first_ext_dw_0, sb->caps, &value->misc3);
+	return 0;
+}
+
+void
+dr_ste_v1_build_tnl_gtpu_flex_parser_0_init(struct mlx5dr_ste_build *sb,
+					    struct mlx5dr_match_param *mask)
+{
+	dr_ste_v1_build_tnl_gtpu_flex_parser_0_tag(mask, sb, sb->bit_mask);
+
+	sb->lu_type = DR_STE_V1_LU_TYPE_FLEX_PARSER_0;
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v1_build_tnl_gtpu_flex_parser_0_tag;
+}
+
+static int
+dr_ste_v1_build_tnl_gtpu_flex_parser_1_tag(struct mlx5dr_match_param *value,
+					   struct mlx5dr_ste_build *sb,
+					   u8 *tag)
+{
+	if (dr_is_flex_parser_1_id(sb->caps->flex_parser_id_gtpu_dw_0))
+		DR_STE_SET_FLEX_PARSER_FIELD(tag, gtpu_dw_0, sb->caps, &value->misc3);
+	if (dr_is_flex_parser_1_id(sb->caps->flex_parser_id_gtpu_teid))
+		DR_STE_SET_FLEX_PARSER_FIELD(tag, gtpu_teid, sb->caps, &value->misc3);
+	if (dr_is_flex_parser_1_id(sb->caps->flex_parser_id_gtpu_dw_2))
+		DR_STE_SET_FLEX_PARSER_FIELD(tag, gtpu_dw_2, sb->caps, &value->misc3);
+	if (dr_is_flex_parser_1_id(sb->caps->flex_parser_id_gtpu_first_ext_dw_0))
+		DR_STE_SET_FLEX_PARSER_FIELD(tag, gtpu_first_ext_dw_0, sb->caps, &value->misc3);
+	return 0;
+}
+
+void
+dr_ste_v1_build_tnl_gtpu_flex_parser_1_init(struct mlx5dr_ste_build *sb,
+					    struct mlx5dr_match_param *mask)
+{
+	dr_ste_v1_build_tnl_gtpu_flex_parser_1_tag(mask, sb, sb->bit_mask);
+
+	sb->lu_type = DR_STE_V1_LU_TYPE_FLEX_PARSER_1;
+	sb->byte_mask = mlx5dr_ste_conv_bit_to_byte_mask(sb->bit_mask);
+	sb->ste_build_tag_func = &dr_ste_v1_build_tnl_gtpu_flex_parser_1_tag;
+}
+
+static struct mlx5dr_ste_ctx ste_ctx_v1 = {
+	/* Builders */
+	.build_eth_l2_src_dst_init	= &dr_ste_v1_build_eth_l2_src_dst_init,
+	.build_eth_l3_ipv6_src_init	= &dr_ste_v1_build_eth_l3_ipv6_src_init,
+	.build_eth_l3_ipv6_dst_init	= &dr_ste_v1_build_eth_l3_ipv6_dst_init,
+	.build_eth_l3_ipv4_5_tuple_init	= &dr_ste_v1_build_eth_l3_ipv4_5_tuple_init,
+	.build_eth_l2_src_init		= &dr_ste_v1_build_eth_l2_src_init,
+	.build_eth_l2_dst_init		= &dr_ste_v1_build_eth_l2_dst_init,
+	.build_eth_l2_tnl_init		= &dr_ste_v1_build_eth_l2_tnl_init,
+	.build_eth_l3_ipv4_misc_init	= &dr_ste_v1_build_eth_l3_ipv4_misc_init,
+	.build_eth_ipv6_l3_l4_init	= &dr_ste_v1_build_eth_ipv6_l3_l4_init,
+	.build_mpls_init		= &dr_ste_v1_build_mpls_init,
+	.build_tnl_gre_init		= &dr_ste_v1_build_tnl_gre_init,
+	.build_tnl_mpls_init		= &dr_ste_v1_build_tnl_mpls_init,
+	.build_tnl_mpls_over_udp_init	= &dr_ste_v1_build_tnl_mpls_over_udp_init,
+	.build_tnl_mpls_over_gre_init	= &dr_ste_v1_build_tnl_mpls_over_gre_init,
+	.build_icmp_init		= &dr_ste_v1_build_icmp_init,
+	.build_general_purpose_init	= &dr_ste_v1_build_general_purpose_init,
+	.build_eth_l4_misc_init		= &dr_ste_v1_build_eth_l4_misc_init,
+	.build_tnl_vxlan_gpe_init	= &dr_ste_v1_build_flex_parser_tnl_vxlan_gpe_init,
+	.build_tnl_geneve_init		= &dr_ste_v1_build_flex_parser_tnl_geneve_init,
+	.build_tnl_geneve_tlv_opt_init	= &dr_ste_v1_build_flex_parser_tnl_geneve_tlv_opt_init,
+	.build_tnl_geneve_tlv_opt_exist_init = &dr_ste_v1_build_flex_parser_tnl_geneve_tlv_opt_exist_init,
+	.build_register_0_init		= &dr_ste_v1_build_register_0_init,
+	.build_register_1_init		= &dr_ste_v1_build_register_1_init,
+	.build_src_gvmi_qpn_init	= &dr_ste_v1_build_src_gvmi_qpn_init,
+	.build_flex_parser_0_init	= &dr_ste_v1_build_flex_parser_0_init,
+	.build_flex_parser_1_init	= &dr_ste_v1_build_flex_parser_1_init,
+	.build_tnl_gtpu_init		= &dr_ste_v1_build_flex_parser_tnl_gtpu_init,
+	.build_tnl_header_0_1_init	= &dr_ste_v1_build_tnl_header_0_1_init,
+	.build_tnl_gtpu_flex_parser_0_init = &dr_ste_v1_build_tnl_gtpu_flex_parser_0_init,
+	.build_tnl_gtpu_flex_parser_1_init = &dr_ste_v1_build_tnl_gtpu_flex_parser_1_init,
+
+	/* Getters and Setters */
+	.ste_init			= &dr_ste_v1_init,
+	.set_next_lu_type		= &dr_ste_v1_set_next_lu_type,
+	.get_next_lu_type		= &dr_ste_v1_get_next_lu_type,
+	.set_miss_addr			= &dr_ste_v1_set_miss_addr,
+	.get_miss_addr			= &dr_ste_v1_get_miss_addr,
+	.set_hit_addr			= &dr_ste_v1_set_hit_addr,
+	.set_byte_mask			= &dr_ste_v1_set_byte_mask,
+	.get_byte_mask			= &dr_ste_v1_get_byte_mask,
+	/* Actions */
+	.actions_caps			= DR_STE_CTX_ACTION_CAP_TX_POP |
+					  DR_STE_CTX_ACTION_CAP_RX_PUSH |
+					  DR_STE_CTX_ACTION_CAP_RX_ENCAP |
+					  DR_STE_CTX_ACTION_CAP_POP_MDFY,
+	.set_actions_rx			= &dr_ste_v1_set_actions_rx,
+	.set_actions_tx			= &dr_ste_v1_set_actions_tx,
+	.modify_field_arr_sz		= ARRAY_SIZE(dr_ste_v1_action_modify_field_arr),
+	.modify_field_arr		= dr_ste_v1_action_modify_field_arr,
+	.set_action_set			= &dr_ste_v1_set_action_set,
+	.set_action_add			= &dr_ste_v1_set_action_add,
+	.set_action_copy		= &dr_ste_v1_set_action_copy,
+	.set_action_decap_l3_list	= &dr_ste_v1_set_action_decap_l3_list,
+	/* Send */
+	.prepare_for_postsend		= &dr_ste_v1_prepare_for_postsend,
+};
+
+struct mlx5dr_ste_ctx *mlx5dr_ste_get_ctx_v1(void)
+{
+	return &ste_ctx_v1;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_ste_v1.h b/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_ste_v1.h
new file mode 100644
index 000000000..8a1d49790
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_ste_v1.h
@@ -0,0 +1,94 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2022, NVIDIA CORPORATION & AFFILIATES. All rights reserved. */
+
+#ifndef	_DR_STE_V1_
+#define	_DR_STE_V1_
+
+#include "dr_types.h"
+#include "dr_ste.h"
+
+void dr_ste_v1_set_miss_addr(u8 *hw_ste_p, u64 miss_addr);
+u64 dr_ste_v1_get_miss_addr(u8 *hw_ste_p);
+void dr_ste_v1_set_byte_mask(u8 *hw_ste_p, u16 byte_mask);
+u16 dr_ste_v1_get_byte_mask(u8 *hw_ste_p);
+void dr_ste_v1_set_next_lu_type(u8 *hw_ste_p, u16 lu_type);
+u16 dr_ste_v1_get_next_lu_type(u8 *hw_ste_p);
+void dr_ste_v1_set_hit_addr(u8 *hw_ste_p, u64 icm_addr, u32 ht_size);
+void dr_ste_v1_init(u8 *hw_ste_p, u16 lu_type, bool is_rx, u16 gvmi);
+void dr_ste_v1_prepare_for_postsend(u8 *hw_ste_p, u32 ste_size);
+void dr_ste_v1_set_actions_tx(struct mlx5dr_domain *dmn, u8 *action_type_set,
+			      u32 actions_caps, u8 *last_ste,
+			      struct mlx5dr_ste_actions_attr *attr, u32 *added_stes);
+void dr_ste_v1_set_actions_rx(struct mlx5dr_domain *dmn, u8 *action_type_set,
+			      u32 actions_caps, u8 *last_ste,
+			      struct mlx5dr_ste_actions_attr *attr, u32 *added_stes);
+void dr_ste_v1_set_action_set(u8 *d_action, u8 hw_field, u8 shifter,
+			      u8 length, u32 data);
+void dr_ste_v1_set_action_add(u8 *d_action, u8 hw_field, u8 shifter,
+			      u8 length, u32 data);
+void dr_ste_v1_set_action_copy(u8 *d_action, u8 dst_hw_field, u8 dst_shifter,
+			       u8 dst_len, u8 src_hw_field, u8 src_shifter);
+int dr_ste_v1_set_action_decap_l3_list(void *data, u32 data_sz, u8 *hw_action,
+				       u32 hw_action_sz, u16 *used_hw_action_num);
+void dr_ste_v1_build_eth_l2_src_dst_init(struct mlx5dr_ste_build *sb,
+					 struct mlx5dr_match_param *mask);
+void dr_ste_v1_build_eth_l3_ipv6_dst_init(struct mlx5dr_ste_build *sb,
+					  struct mlx5dr_match_param *mask);
+void dr_ste_v1_build_eth_l3_ipv6_src_init(struct mlx5dr_ste_build *sb,
+					  struct mlx5dr_match_param *mask);
+void dr_ste_v1_build_eth_l3_ipv4_5_tuple_init(struct mlx5dr_ste_build *sb,
+					      struct mlx5dr_match_param *mask);
+void dr_ste_v1_build_eth_l2_src_init(struct mlx5dr_ste_build *sb,
+				     struct mlx5dr_match_param *mask);
+void dr_ste_v1_build_eth_l2_dst_init(struct mlx5dr_ste_build *sb,
+				     struct mlx5dr_match_param *mask);
+void dr_ste_v1_build_eth_l2_tnl_init(struct mlx5dr_ste_build *sb,
+				     struct mlx5dr_match_param *mask);
+void dr_ste_v1_build_eth_l3_ipv4_misc_init(struct mlx5dr_ste_build *sb,
+					   struct mlx5dr_match_param *mask);
+void dr_ste_v1_build_eth_ipv6_l3_l4_init(struct mlx5dr_ste_build *sb,
+					 struct mlx5dr_match_param *mask);
+void dr_ste_v1_build_mpls_init(struct mlx5dr_ste_build *sb,
+			       struct mlx5dr_match_param *mask);
+void dr_ste_v1_build_tnl_gre_init(struct mlx5dr_ste_build *sb,
+				  struct mlx5dr_match_param *mask);
+void dr_ste_v1_build_tnl_mpls_init(struct mlx5dr_ste_build *sb,
+				   struct mlx5dr_match_param *mask);
+void dr_ste_v1_build_tnl_mpls_over_udp_init(struct mlx5dr_ste_build *sb,
+					    struct mlx5dr_match_param *mask);
+void dr_ste_v1_build_tnl_mpls_over_gre_init(struct mlx5dr_ste_build *sb,
+					    struct mlx5dr_match_param *mask);
+void dr_ste_v1_build_icmp_init(struct mlx5dr_ste_build *sb,
+			       struct mlx5dr_match_param *mask);
+void dr_ste_v1_build_general_purpose_init(struct mlx5dr_ste_build *sb,
+					  struct mlx5dr_match_param *mask);
+void dr_ste_v1_build_eth_l4_misc_init(struct mlx5dr_ste_build *sb,
+				      struct mlx5dr_match_param *mask);
+void dr_ste_v1_build_flex_parser_tnl_vxlan_gpe_init(struct mlx5dr_ste_build *sb,
+						    struct mlx5dr_match_param *mask);
+void dr_ste_v1_build_flex_parser_tnl_geneve_init(struct mlx5dr_ste_build *sb,
+						 struct mlx5dr_match_param *mask);
+void dr_ste_v1_build_tnl_header_0_1_init(struct mlx5dr_ste_build *sb,
+					 struct mlx5dr_match_param *mask);
+void dr_ste_v1_build_register_0_init(struct mlx5dr_ste_build *sb,
+				     struct mlx5dr_match_param *mask);
+void dr_ste_v1_build_register_1_init(struct mlx5dr_ste_build *sb,
+				     struct mlx5dr_match_param *mask);
+void dr_ste_v1_build_src_gvmi_qpn_init(struct mlx5dr_ste_build *sb,
+				       struct mlx5dr_match_param *mask);
+void dr_ste_v1_build_flex_parser_0_init(struct mlx5dr_ste_build *sb,
+					struct mlx5dr_match_param *mask);
+void dr_ste_v1_build_flex_parser_1_init(struct mlx5dr_ste_build *sb,
+					struct mlx5dr_match_param *mask);
+void dr_ste_v1_build_flex_parser_tnl_geneve_tlv_opt_init(struct mlx5dr_ste_build *sb,
+							 struct mlx5dr_match_param *mask);
+void dr_ste_v1_build_flex_parser_tnl_geneve_tlv_opt_exist_init(struct mlx5dr_ste_build *sb,
+							       struct mlx5dr_match_param *mask);
+void dr_ste_v1_build_flex_parser_tnl_gtpu_init(struct mlx5dr_ste_build *sb,
+					       struct mlx5dr_match_param *mask);
+void dr_ste_v1_build_tnl_gtpu_flex_parser_0_init(struct mlx5dr_ste_build *sb,
+						 struct mlx5dr_match_param *mask);
+void dr_ste_v1_build_tnl_gtpu_flex_parser_1_init(struct mlx5dr_ste_build *sb,
+						 struct mlx5dr_match_param *mask);
+
+#endif  /* _DR_STE_V1_ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_ste_v2.c b/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_ste_v2.c
new file mode 100644
index 000000000..c60fddd12
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_ste_v2.c
@@ -0,0 +1,231 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2022, NVIDIA CORPORATION & AFFILIATES. All rights reserved. */
+
+#include "dr_ste_v1.h"
+
+enum {
+	DR_STE_V2_ACTION_MDFY_FLD_L2_OUT_0		= 0x00,
+	DR_STE_V2_ACTION_MDFY_FLD_L2_OUT_1		= 0x01,
+	DR_STE_V2_ACTION_MDFY_FLD_L2_OUT_2		= 0x02,
+	DR_STE_V2_ACTION_MDFY_FLD_SRC_L2_OUT_0		= 0x08,
+	DR_STE_V2_ACTION_MDFY_FLD_SRC_L2_OUT_1		= 0x09,
+	DR_STE_V2_ACTION_MDFY_FLD_L3_OUT_0		= 0x0e,
+	DR_STE_V2_ACTION_MDFY_FLD_L4_OUT_0		= 0x18,
+	DR_STE_V2_ACTION_MDFY_FLD_L4_OUT_1		= 0x19,
+	DR_STE_V2_ACTION_MDFY_FLD_IPV4_OUT_0		= 0x40,
+	DR_STE_V2_ACTION_MDFY_FLD_IPV4_OUT_1		= 0x41,
+	DR_STE_V2_ACTION_MDFY_FLD_IPV6_DST_OUT_0	= 0x44,
+	DR_STE_V2_ACTION_MDFY_FLD_IPV6_DST_OUT_1	= 0x45,
+	DR_STE_V2_ACTION_MDFY_FLD_IPV6_DST_OUT_2	= 0x46,
+	DR_STE_V2_ACTION_MDFY_FLD_IPV6_DST_OUT_3	= 0x47,
+	DR_STE_V2_ACTION_MDFY_FLD_IPV6_SRC_OUT_0	= 0x4c,
+	DR_STE_V2_ACTION_MDFY_FLD_IPV6_SRC_OUT_1	= 0x4d,
+	DR_STE_V2_ACTION_MDFY_FLD_IPV6_SRC_OUT_2	= 0x4e,
+	DR_STE_V2_ACTION_MDFY_FLD_IPV6_SRC_OUT_3	= 0x4f,
+	DR_STE_V2_ACTION_MDFY_FLD_TCP_MISC_0		= 0x5e,
+	DR_STE_V2_ACTION_MDFY_FLD_TCP_MISC_1		= 0x5f,
+	DR_STE_V2_ACTION_MDFY_FLD_CFG_HDR_0_0		= 0x6f,
+	DR_STE_V2_ACTION_MDFY_FLD_CFG_HDR_0_1		= 0x70,
+	DR_STE_V2_ACTION_MDFY_FLD_METADATA_2_CQE	= 0x7b,
+	DR_STE_V2_ACTION_MDFY_FLD_GNRL_PURPOSE		= 0x7c,
+	DR_STE_V2_ACTION_MDFY_FLD_REGISTER_2_0		= 0x90,
+	DR_STE_V2_ACTION_MDFY_FLD_REGISTER_2_1		= 0x91,
+	DR_STE_V2_ACTION_MDFY_FLD_REGISTER_1_0		= 0x92,
+	DR_STE_V2_ACTION_MDFY_FLD_REGISTER_1_1		= 0x93,
+	DR_STE_V2_ACTION_MDFY_FLD_REGISTER_0_0		= 0x94,
+	DR_STE_V2_ACTION_MDFY_FLD_REGISTER_0_1		= 0x95,
+};
+
+static const struct mlx5dr_ste_action_modify_field dr_ste_v2_action_modify_field_arr[] = {
+	[MLX5_ACTION_IN_FIELD_OUT_SMAC_47_16] = {
+		.hw_field = DR_STE_V2_ACTION_MDFY_FLD_SRC_L2_OUT_0, .start = 0, .end = 31,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_SMAC_15_0] = {
+		.hw_field = DR_STE_V2_ACTION_MDFY_FLD_SRC_L2_OUT_1, .start = 16, .end = 31,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_ETHERTYPE] = {
+		.hw_field = DR_STE_V2_ACTION_MDFY_FLD_L2_OUT_1, .start = 0, .end = 15,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_DMAC_47_16] = {
+		.hw_field = DR_STE_V2_ACTION_MDFY_FLD_L2_OUT_0, .start = 0, .end = 31,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_DMAC_15_0] = {
+		.hw_field = DR_STE_V2_ACTION_MDFY_FLD_L2_OUT_1, .start = 16, .end = 31,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_IP_DSCP] = {
+		.hw_field = DR_STE_V2_ACTION_MDFY_FLD_L3_OUT_0, .start = 18, .end = 23,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_TCP_FLAGS] = {
+		.hw_field = DR_STE_V2_ACTION_MDFY_FLD_L4_OUT_1, .start = 16, .end = 24,
+		.l4_type = DR_STE_ACTION_MDFY_TYPE_L4_TCP,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_TCP_SPORT] = {
+		.hw_field = DR_STE_V2_ACTION_MDFY_FLD_L4_OUT_0, .start = 16, .end = 31,
+		.l4_type = DR_STE_ACTION_MDFY_TYPE_L4_TCP,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_TCP_DPORT] = {
+		.hw_field = DR_STE_V2_ACTION_MDFY_FLD_L4_OUT_0, .start = 0, .end = 15,
+		.l4_type = DR_STE_ACTION_MDFY_TYPE_L4_TCP,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_IP_TTL] = {
+		.hw_field = DR_STE_V2_ACTION_MDFY_FLD_L3_OUT_0, .start = 8, .end = 15,
+		.l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV4,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_IPV6_HOPLIMIT] = {
+		.hw_field = DR_STE_V2_ACTION_MDFY_FLD_L3_OUT_0, .start = 8, .end = 15,
+		.l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV6,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_UDP_SPORT] = {
+		.hw_field = DR_STE_V2_ACTION_MDFY_FLD_L4_OUT_0, .start = 16, .end = 31,
+		.l4_type = DR_STE_ACTION_MDFY_TYPE_L4_UDP,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_UDP_DPORT] = {
+		.hw_field = DR_STE_V2_ACTION_MDFY_FLD_L4_OUT_0, .start = 0, .end = 15,
+		.l4_type = DR_STE_ACTION_MDFY_TYPE_L4_UDP,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_SIPV6_127_96] = {
+		.hw_field = DR_STE_V2_ACTION_MDFY_FLD_IPV6_SRC_OUT_0, .start = 0, .end = 31,
+		.l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV6,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_SIPV6_95_64] = {
+		.hw_field = DR_STE_V2_ACTION_MDFY_FLD_IPV6_SRC_OUT_1, .start = 0, .end = 31,
+		.l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV6,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_SIPV6_63_32] = {
+		.hw_field = DR_STE_V2_ACTION_MDFY_FLD_IPV6_SRC_OUT_2, .start = 0, .end = 31,
+		.l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV6,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_SIPV6_31_0] = {
+		.hw_field = DR_STE_V2_ACTION_MDFY_FLD_IPV6_SRC_OUT_3, .start = 0, .end = 31,
+		.l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV6,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_DIPV6_127_96] = {
+		.hw_field = DR_STE_V2_ACTION_MDFY_FLD_IPV6_DST_OUT_0, .start = 0, .end = 31,
+		.l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV6,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_DIPV6_95_64] = {
+		.hw_field = DR_STE_V2_ACTION_MDFY_FLD_IPV6_DST_OUT_1, .start = 0, .end = 31,
+		.l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV6,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_DIPV6_63_32] = {
+		.hw_field = DR_STE_V2_ACTION_MDFY_FLD_IPV6_DST_OUT_2, .start = 0, .end = 31,
+		.l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV6,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_DIPV6_31_0] = {
+		.hw_field = DR_STE_V2_ACTION_MDFY_FLD_IPV6_DST_OUT_3, .start = 0, .end = 31,
+		.l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV6,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_SIPV4] = {
+		.hw_field = DR_STE_V2_ACTION_MDFY_FLD_IPV4_OUT_0, .start = 0, .end = 31,
+		.l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV4,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_DIPV4] = {
+		.hw_field = DR_STE_V2_ACTION_MDFY_FLD_IPV4_OUT_1, .start = 0, .end = 31,
+		.l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV4,
+	},
+	[MLX5_ACTION_IN_FIELD_METADATA_REG_A] = {
+		.hw_field = DR_STE_V2_ACTION_MDFY_FLD_GNRL_PURPOSE, .start = 0, .end = 31,
+	},
+	[MLX5_ACTION_IN_FIELD_METADATA_REG_B] = {
+		.hw_field = DR_STE_V2_ACTION_MDFY_FLD_METADATA_2_CQE, .start = 0, .end = 31,
+	},
+	[MLX5_ACTION_IN_FIELD_METADATA_REG_C_0] = {
+		.hw_field = DR_STE_V2_ACTION_MDFY_FLD_REGISTER_0_0, .start = 0, .end = 31,
+	},
+	[MLX5_ACTION_IN_FIELD_METADATA_REG_C_1] = {
+		.hw_field = DR_STE_V2_ACTION_MDFY_FLD_REGISTER_0_1, .start = 0, .end = 31,
+	},
+	[MLX5_ACTION_IN_FIELD_METADATA_REG_C_2] = {
+		.hw_field = DR_STE_V2_ACTION_MDFY_FLD_REGISTER_1_0, .start = 0, .end = 31,
+	},
+	[MLX5_ACTION_IN_FIELD_METADATA_REG_C_3] = {
+		.hw_field = DR_STE_V2_ACTION_MDFY_FLD_REGISTER_1_1, .start = 0, .end = 31,
+	},
+	[MLX5_ACTION_IN_FIELD_METADATA_REG_C_4] = {
+		.hw_field = DR_STE_V2_ACTION_MDFY_FLD_REGISTER_2_0, .start = 0, .end = 31,
+	},
+	[MLX5_ACTION_IN_FIELD_METADATA_REG_C_5] = {
+		.hw_field = DR_STE_V2_ACTION_MDFY_FLD_REGISTER_2_1, .start = 0, .end = 31,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_TCP_SEQ_NUM] = {
+		.hw_field = DR_STE_V2_ACTION_MDFY_FLD_TCP_MISC_0, .start = 0, .end = 31,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_TCP_ACK_NUM] = {
+		.hw_field = DR_STE_V2_ACTION_MDFY_FLD_TCP_MISC_1, .start = 0, .end = 31,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_FIRST_VID] = {
+		.hw_field = DR_STE_V2_ACTION_MDFY_FLD_L2_OUT_2, .start = 0, .end = 15,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_EMD_31_0] = {
+		.hw_field = DR_STE_V2_ACTION_MDFY_FLD_CFG_HDR_0_1, .start = 0, .end = 31,
+	},
+	[MLX5_ACTION_IN_FIELD_OUT_EMD_47_32] = {
+		.hw_field = DR_STE_V2_ACTION_MDFY_FLD_CFG_HDR_0_0, .start = 0, .end = 15,
+	},
+};
+
+static struct mlx5dr_ste_ctx ste_ctx_v2 = {
+	/* Builders */
+	.build_eth_l2_src_dst_init	= &dr_ste_v1_build_eth_l2_src_dst_init,
+	.build_eth_l3_ipv6_src_init	= &dr_ste_v1_build_eth_l3_ipv6_src_init,
+	.build_eth_l3_ipv6_dst_init	= &dr_ste_v1_build_eth_l3_ipv6_dst_init,
+	.build_eth_l3_ipv4_5_tuple_init	= &dr_ste_v1_build_eth_l3_ipv4_5_tuple_init,
+	.build_eth_l2_src_init		= &dr_ste_v1_build_eth_l2_src_init,
+	.build_eth_l2_dst_init		= &dr_ste_v1_build_eth_l2_dst_init,
+	.build_eth_l2_tnl_init		= &dr_ste_v1_build_eth_l2_tnl_init,
+	.build_eth_l3_ipv4_misc_init	= &dr_ste_v1_build_eth_l3_ipv4_misc_init,
+	.build_eth_ipv6_l3_l4_init	= &dr_ste_v1_build_eth_ipv6_l3_l4_init,
+	.build_mpls_init		= &dr_ste_v1_build_mpls_init,
+	.build_tnl_gre_init		= &dr_ste_v1_build_tnl_gre_init,
+	.build_tnl_mpls_init		= &dr_ste_v1_build_tnl_mpls_init,
+	.build_tnl_mpls_over_udp_init	= &dr_ste_v1_build_tnl_mpls_over_udp_init,
+	.build_tnl_mpls_over_gre_init	= &dr_ste_v1_build_tnl_mpls_over_gre_init,
+	.build_icmp_init		= &dr_ste_v1_build_icmp_init,
+	.build_general_purpose_init	= &dr_ste_v1_build_general_purpose_init,
+	.build_eth_l4_misc_init		= &dr_ste_v1_build_eth_l4_misc_init,
+	.build_tnl_vxlan_gpe_init	= &dr_ste_v1_build_flex_parser_tnl_vxlan_gpe_init,
+	.build_tnl_geneve_init		= &dr_ste_v1_build_flex_parser_tnl_geneve_init,
+	.build_tnl_geneve_tlv_opt_init	= &dr_ste_v1_build_flex_parser_tnl_geneve_tlv_opt_init,
+	.build_tnl_geneve_tlv_opt_exist_init =
+				  &dr_ste_v1_build_flex_parser_tnl_geneve_tlv_opt_exist_init,
+	.build_register_0_init		= &dr_ste_v1_build_register_0_init,
+	.build_register_1_init		= &dr_ste_v1_build_register_1_init,
+	.build_src_gvmi_qpn_init	= &dr_ste_v1_build_src_gvmi_qpn_init,
+	.build_flex_parser_0_init	= &dr_ste_v1_build_flex_parser_0_init,
+	.build_flex_parser_1_init	= &dr_ste_v1_build_flex_parser_1_init,
+	.build_tnl_gtpu_init		= &dr_ste_v1_build_flex_parser_tnl_gtpu_init,
+	.build_tnl_header_0_1_init	= &dr_ste_v1_build_tnl_header_0_1_init,
+	.build_tnl_gtpu_flex_parser_0_init = &dr_ste_v1_build_tnl_gtpu_flex_parser_0_init,
+	.build_tnl_gtpu_flex_parser_1_init = &dr_ste_v1_build_tnl_gtpu_flex_parser_1_init,
+
+	/* Getters and Setters */
+	.ste_init			= &dr_ste_v1_init,
+	.set_next_lu_type		= &dr_ste_v1_set_next_lu_type,
+	.get_next_lu_type		= &dr_ste_v1_get_next_lu_type,
+	.set_miss_addr			= &dr_ste_v1_set_miss_addr,
+	.get_miss_addr			= &dr_ste_v1_get_miss_addr,
+	.set_hit_addr			= &dr_ste_v1_set_hit_addr,
+	.set_byte_mask			= &dr_ste_v1_set_byte_mask,
+	.get_byte_mask			= &dr_ste_v1_get_byte_mask,
+
+	/* Actions */
+	.actions_caps			= DR_STE_CTX_ACTION_CAP_TX_POP |
+					  DR_STE_CTX_ACTION_CAP_RX_PUSH |
+					  DR_STE_CTX_ACTION_CAP_RX_ENCAP,
+	.set_actions_rx			= &dr_ste_v1_set_actions_rx,
+	.set_actions_tx			= &dr_ste_v1_set_actions_tx,
+	.modify_field_arr_sz		= ARRAY_SIZE(dr_ste_v2_action_modify_field_arr),
+	.modify_field_arr		= dr_ste_v2_action_modify_field_arr,
+	.set_action_set			= &dr_ste_v1_set_action_set,
+	.set_action_add			= &dr_ste_v1_set_action_add,
+	.set_action_copy		= &dr_ste_v1_set_action_copy,
+	.set_action_decap_l3_list	= &dr_ste_v1_set_action_decap_l3_list,
+
+	/* Send */
+	.prepare_for_postsend		= &dr_ste_v1_prepare_for_postsend,
+};
+
+struct mlx5dr_ste_ctx *mlx5dr_ste_get_ctx_v2(void)
+{
+	return &ste_ctx_v2;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_table.c b/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_table.c
new file mode 100644
index 000000000..f68461b13
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_table.c
@@ -0,0 +1,319 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2019 Mellanox Technologies. */
+
+#include "dr_types.h"
+
+static int dr_table_set_miss_action_nic(struct mlx5dr_domain *dmn,
+					struct mlx5dr_table_rx_tx *nic_tbl,
+					struct mlx5dr_action *action)
+{
+	struct mlx5dr_matcher_rx_tx *last_nic_matcher = NULL;
+	struct mlx5dr_htbl_connect_info info;
+	struct mlx5dr_ste_htbl *last_htbl;
+	struct mlx5dr_icm_chunk *chunk;
+	int ret;
+
+	if (!list_empty(&nic_tbl->nic_matcher_list))
+		last_nic_matcher = list_last_entry(&nic_tbl->nic_matcher_list,
+						   struct mlx5dr_matcher_rx_tx,
+						   list_node);
+
+	if (last_nic_matcher)
+		last_htbl = last_nic_matcher->e_anchor;
+	else
+		last_htbl = nic_tbl->s_anchor;
+
+	if (action) {
+		chunk = nic_tbl->nic_dmn->type == DR_DOMAIN_NIC_TYPE_RX ?
+			action->dest_tbl->tbl->rx.s_anchor->chunk :
+			action->dest_tbl->tbl->tx.s_anchor->chunk;
+		nic_tbl->default_icm_addr = mlx5dr_icm_pool_get_chunk_icm_addr(chunk);
+	} else {
+		nic_tbl->default_icm_addr = nic_tbl->nic_dmn->default_icm_addr;
+	}
+
+	info.type = CONNECT_MISS;
+	info.miss_icm_addr = nic_tbl->default_icm_addr;
+
+	ret = mlx5dr_ste_htbl_init_and_postsend(dmn, nic_tbl->nic_dmn,
+						last_htbl, &info, true);
+	if (ret)
+		mlx5dr_dbg(dmn, "Failed to set NIC RX/TX miss action, ret %d\n", ret);
+
+	return ret;
+}
+
+int mlx5dr_table_set_miss_action(struct mlx5dr_table *tbl,
+				 struct mlx5dr_action *action)
+{
+	int ret = -EOPNOTSUPP;
+
+	if (action && action->action_type != DR_ACTION_TYP_FT)
+		return -EOPNOTSUPP;
+
+	mlx5dr_domain_lock(tbl->dmn);
+
+	if (tbl->dmn->type == MLX5DR_DOMAIN_TYPE_NIC_RX ||
+	    tbl->dmn->type == MLX5DR_DOMAIN_TYPE_FDB) {
+		ret = dr_table_set_miss_action_nic(tbl->dmn, &tbl->rx, action);
+		if (ret)
+			goto out;
+	}
+
+	if (tbl->dmn->type == MLX5DR_DOMAIN_TYPE_NIC_TX ||
+	    tbl->dmn->type == MLX5DR_DOMAIN_TYPE_FDB) {
+		ret = dr_table_set_miss_action_nic(tbl->dmn, &tbl->tx, action);
+		if (ret)
+			goto out;
+	}
+
+	if (ret)
+		goto out;
+
+	/* Release old action */
+	if (tbl->miss_action)
+		refcount_dec(&tbl->miss_action->refcount);
+
+	/* Set new miss action */
+	tbl->miss_action = action;
+	if (tbl->miss_action)
+		refcount_inc(&action->refcount);
+
+out:
+	mlx5dr_domain_unlock(tbl->dmn);
+	return ret;
+}
+
+static void dr_table_uninit_nic(struct mlx5dr_table_rx_tx *nic_tbl)
+{
+	mlx5dr_htbl_put(nic_tbl->s_anchor);
+}
+
+static void dr_table_uninit_fdb(struct mlx5dr_table *tbl)
+{
+	dr_table_uninit_nic(&tbl->rx);
+	dr_table_uninit_nic(&tbl->tx);
+}
+
+static void dr_table_uninit(struct mlx5dr_table *tbl)
+{
+	mlx5dr_domain_lock(tbl->dmn);
+
+	switch (tbl->dmn->type) {
+	case MLX5DR_DOMAIN_TYPE_NIC_RX:
+		dr_table_uninit_nic(&tbl->rx);
+		break;
+	case MLX5DR_DOMAIN_TYPE_NIC_TX:
+		dr_table_uninit_nic(&tbl->tx);
+		break;
+	case MLX5DR_DOMAIN_TYPE_FDB:
+		dr_table_uninit_fdb(tbl);
+		break;
+	default:
+		WARN_ON(true);
+		break;
+	}
+
+	mlx5dr_domain_unlock(tbl->dmn);
+}
+
+static int dr_table_init_nic(struct mlx5dr_domain *dmn,
+			     struct mlx5dr_table_rx_tx *nic_tbl)
+{
+	struct mlx5dr_domain_rx_tx *nic_dmn = nic_tbl->nic_dmn;
+	struct mlx5dr_htbl_connect_info info;
+	int ret;
+
+	INIT_LIST_HEAD(&nic_tbl->nic_matcher_list);
+
+	nic_tbl->default_icm_addr = nic_dmn->default_icm_addr;
+
+	nic_tbl->s_anchor = mlx5dr_ste_htbl_alloc(dmn->ste_icm_pool,
+						  DR_CHUNK_SIZE_1,
+						  MLX5DR_STE_LU_TYPE_DONT_CARE,
+						  0);
+	if (!nic_tbl->s_anchor) {
+		mlx5dr_err(dmn, "Failed allocating htbl\n");
+		return -ENOMEM;
+	}
+
+	info.type = CONNECT_MISS;
+	info.miss_icm_addr = nic_dmn->default_icm_addr;
+	ret = mlx5dr_ste_htbl_init_and_postsend(dmn, nic_dmn,
+						nic_tbl->s_anchor,
+						&info, true);
+	if (ret) {
+		mlx5dr_err(dmn, "Failed int and send htbl\n");
+		goto free_s_anchor;
+	}
+
+	mlx5dr_htbl_get(nic_tbl->s_anchor);
+
+	return 0;
+
+free_s_anchor:
+	mlx5dr_ste_htbl_free(nic_tbl->s_anchor);
+	return ret;
+}
+
+static int dr_table_init_fdb(struct mlx5dr_table *tbl)
+{
+	int ret;
+
+	ret = dr_table_init_nic(tbl->dmn, &tbl->rx);
+	if (ret)
+		return ret;
+
+	ret = dr_table_init_nic(tbl->dmn, &tbl->tx);
+	if (ret)
+		goto destroy_rx;
+
+	return 0;
+
+destroy_rx:
+	dr_table_uninit_nic(&tbl->rx);
+	return ret;
+}
+
+static int dr_table_init(struct mlx5dr_table *tbl)
+{
+	int ret = 0;
+
+	INIT_LIST_HEAD(&tbl->matcher_list);
+
+	mlx5dr_domain_lock(tbl->dmn);
+
+	switch (tbl->dmn->type) {
+	case MLX5DR_DOMAIN_TYPE_NIC_RX:
+		tbl->table_type = MLX5_FLOW_TABLE_TYPE_NIC_RX;
+		tbl->rx.nic_dmn = &tbl->dmn->info.rx;
+		ret = dr_table_init_nic(tbl->dmn, &tbl->rx);
+		break;
+	case MLX5DR_DOMAIN_TYPE_NIC_TX:
+		tbl->table_type = MLX5_FLOW_TABLE_TYPE_NIC_TX;
+		tbl->tx.nic_dmn = &tbl->dmn->info.tx;
+		ret = dr_table_init_nic(tbl->dmn, &tbl->tx);
+		break;
+	case MLX5DR_DOMAIN_TYPE_FDB:
+		tbl->table_type = MLX5_FLOW_TABLE_TYPE_FDB;
+		tbl->rx.nic_dmn = &tbl->dmn->info.rx;
+		tbl->tx.nic_dmn = &tbl->dmn->info.tx;
+		ret = dr_table_init_fdb(tbl);
+		break;
+	default:
+		WARN_ON(true);
+		break;
+	}
+
+	mlx5dr_domain_unlock(tbl->dmn);
+
+	return ret;
+}
+
+static int dr_table_destroy_sw_owned_tbl(struct mlx5dr_table *tbl)
+{
+	return mlx5dr_cmd_destroy_flow_table(tbl->dmn->mdev,
+					     tbl->table_id,
+					     tbl->table_type);
+}
+
+static int dr_table_create_sw_owned_tbl(struct mlx5dr_table *tbl, u16 uid)
+{
+	bool en_encap = !!(tbl->flags & MLX5_FLOW_TABLE_TUNNEL_EN_REFORMAT);
+	bool en_decap = !!(tbl->flags & MLX5_FLOW_TABLE_TUNNEL_EN_DECAP);
+	struct mlx5dr_cmd_create_flow_table_attr ft_attr = {};
+	u64 icm_addr_rx = 0;
+	u64 icm_addr_tx = 0;
+	int ret;
+
+	if (tbl->rx.s_anchor)
+		icm_addr_rx = mlx5dr_icm_pool_get_chunk_icm_addr(tbl->rx.s_anchor->chunk);
+
+	if (tbl->tx.s_anchor)
+		icm_addr_tx = mlx5dr_icm_pool_get_chunk_icm_addr(tbl->tx.s_anchor->chunk);
+
+	ft_attr.table_type = tbl->table_type;
+	ft_attr.icm_addr_rx = icm_addr_rx;
+	ft_attr.icm_addr_tx = icm_addr_tx;
+	ft_attr.level = tbl->dmn->info.caps.max_ft_level - 1;
+	ft_attr.sw_owner = true;
+	ft_attr.decap_en = en_decap;
+	ft_attr.reformat_en = en_encap;
+	ft_attr.uid = uid;
+
+	ret = mlx5dr_cmd_create_flow_table(tbl->dmn->mdev, &ft_attr,
+					   NULL, &tbl->table_id);
+
+	return ret;
+}
+
+struct mlx5dr_table *mlx5dr_table_create(struct mlx5dr_domain *dmn, u32 level,
+					 u32 flags, u16 uid)
+{
+	struct mlx5dr_table *tbl;
+	int ret;
+
+	refcount_inc(&dmn->refcount);
+
+	tbl = kzalloc(sizeof(*tbl), GFP_KERNEL);
+	if (!tbl)
+		goto dec_ref;
+
+	tbl->dmn = dmn;
+	tbl->level = level;
+	tbl->flags = flags;
+	refcount_set(&tbl->refcount, 1);
+
+	ret = dr_table_init(tbl);
+	if (ret)
+		goto free_tbl;
+
+	ret = dr_table_create_sw_owned_tbl(tbl, uid);
+	if (ret)
+		goto uninit_tbl;
+
+	INIT_LIST_HEAD(&tbl->dbg_node);
+	mlx5dr_dbg_tbl_add(tbl);
+	return tbl;
+
+uninit_tbl:
+	dr_table_uninit(tbl);
+free_tbl:
+	kfree(tbl);
+dec_ref:
+	refcount_dec(&dmn->refcount);
+	return NULL;
+}
+
+int mlx5dr_table_destroy(struct mlx5dr_table *tbl)
+{
+	int ret;
+
+	if (WARN_ON_ONCE(refcount_read(&tbl->refcount) > 1))
+		return -EBUSY;
+
+	mlx5dr_dbg_tbl_del(tbl);
+	ret = dr_table_destroy_sw_owned_tbl(tbl);
+	if (ret)
+		return ret;
+
+	dr_table_uninit(tbl);
+
+	if (tbl->miss_action)
+		refcount_dec(&tbl->miss_action->refcount);
+
+	refcount_dec(&tbl->dmn->refcount);
+	kfree(tbl);
+
+	return ret;
+}
+
+u32 mlx5dr_table_get_id(struct mlx5dr_table *tbl)
+{
+	return tbl->table_id;
+}
+
+struct mlx5dr_table *mlx5dr_table_get_from_fs_ft(struct mlx5_flow_table *ft)
+{
+	return ft->fs_dr_table.dr_table;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_types.h b/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_types.h
new file mode 100644
index 000000000..1777a1e50
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/dr_types.h
@@ -0,0 +1,1472 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2019, Mellanox Technologies */
+
+#ifndef	_DR_TYPES_
+#define	_DR_TYPES_
+
+#include <linux/mlx5/vport.h>
+#include <linux/refcount.h>
+#include "fs_core.h"
+#include "wq.h"
+#include "lib/mlx5.h"
+#include "mlx5_ifc_dr.h"
+#include "mlx5dr.h"
+#include "dr_dbg.h"
+
+#define DR_RULE_MAX_STES 18
+#define DR_ACTION_MAX_STES 5
+#define DR_STE_SVLAN 0x1
+#define DR_STE_CVLAN 0x2
+#define DR_SZ_MATCH_PARAM (MLX5_ST_SZ_DW_MATCH_PARAM * 4)
+#define DR_NUM_OF_FLEX_PARSERS 8
+#define DR_STE_MAX_FLEX_0_ID 3
+#define DR_STE_MAX_FLEX_1_ID 7
+
+#define mlx5dr_err(dmn, arg...) mlx5_core_err((dmn)->mdev, ##arg)
+#define mlx5dr_info(dmn, arg...) mlx5_core_info((dmn)->mdev, ##arg)
+#define mlx5dr_dbg(dmn, arg...) mlx5_core_dbg((dmn)->mdev, ##arg)
+
+static inline bool dr_is_flex_parser_0_id(u8 parser_id)
+{
+	return parser_id <= DR_STE_MAX_FLEX_0_ID;
+}
+
+static inline bool dr_is_flex_parser_1_id(u8 parser_id)
+{
+	return parser_id > DR_STE_MAX_FLEX_0_ID;
+}
+
+enum mlx5dr_icm_chunk_size {
+	DR_CHUNK_SIZE_1,
+	DR_CHUNK_SIZE_MIN = DR_CHUNK_SIZE_1, /* keep updated when changing */
+	DR_CHUNK_SIZE_2,
+	DR_CHUNK_SIZE_4,
+	DR_CHUNK_SIZE_8,
+	DR_CHUNK_SIZE_16,
+	DR_CHUNK_SIZE_32,
+	DR_CHUNK_SIZE_64,
+	DR_CHUNK_SIZE_128,
+	DR_CHUNK_SIZE_256,
+	DR_CHUNK_SIZE_512,
+	DR_CHUNK_SIZE_1K,
+	DR_CHUNK_SIZE_2K,
+	DR_CHUNK_SIZE_4K,
+	DR_CHUNK_SIZE_8K,
+	DR_CHUNK_SIZE_16K,
+	DR_CHUNK_SIZE_32K,
+	DR_CHUNK_SIZE_64K,
+	DR_CHUNK_SIZE_128K,
+	DR_CHUNK_SIZE_256K,
+	DR_CHUNK_SIZE_512K,
+	DR_CHUNK_SIZE_1024K,
+	DR_CHUNK_SIZE_2048K,
+	DR_CHUNK_SIZE_MAX,
+};
+
+enum mlx5dr_icm_type {
+	DR_ICM_TYPE_STE,
+	DR_ICM_TYPE_MODIFY_ACTION,
+};
+
+static inline enum mlx5dr_icm_chunk_size
+mlx5dr_icm_next_higher_chunk(enum mlx5dr_icm_chunk_size chunk)
+{
+	chunk += 2;
+	if (chunk < DR_CHUNK_SIZE_MAX)
+		return chunk;
+
+	return DR_CHUNK_SIZE_MAX;
+}
+
+enum {
+	DR_STE_SIZE = 64,
+	DR_STE_SIZE_CTRL = 32,
+	DR_STE_SIZE_TAG = 16,
+	DR_STE_SIZE_MASK = 16,
+	DR_STE_SIZE_REDUCED = DR_STE_SIZE - DR_STE_SIZE_MASK,
+};
+
+enum mlx5dr_ste_ctx_action_cap {
+	DR_STE_CTX_ACTION_CAP_NONE = 0,
+	DR_STE_CTX_ACTION_CAP_TX_POP   = 1 << 0,
+	DR_STE_CTX_ACTION_CAP_RX_PUSH  = 1 << 1,
+	DR_STE_CTX_ACTION_CAP_RX_ENCAP = 1 << 2,
+	DR_STE_CTX_ACTION_CAP_POP_MDFY = 1 << 3,
+};
+
+enum {
+	DR_MODIFY_ACTION_SIZE = 8,
+};
+
+enum mlx5dr_matcher_criteria {
+	DR_MATCHER_CRITERIA_EMPTY = 0,
+	DR_MATCHER_CRITERIA_OUTER = 1 << 0,
+	DR_MATCHER_CRITERIA_MISC = 1 << 1,
+	DR_MATCHER_CRITERIA_INNER = 1 << 2,
+	DR_MATCHER_CRITERIA_MISC2 = 1 << 3,
+	DR_MATCHER_CRITERIA_MISC3 = 1 << 4,
+	DR_MATCHER_CRITERIA_MISC4 = 1 << 5,
+	DR_MATCHER_CRITERIA_MISC5 = 1 << 6,
+	DR_MATCHER_CRITERIA_MAX = 1 << 7,
+};
+
+enum mlx5dr_action_type {
+	DR_ACTION_TYP_TNL_L2_TO_L2,
+	DR_ACTION_TYP_L2_TO_TNL_L2,
+	DR_ACTION_TYP_TNL_L3_TO_L2,
+	DR_ACTION_TYP_L2_TO_TNL_L3,
+	DR_ACTION_TYP_DROP,
+	DR_ACTION_TYP_QP,
+	DR_ACTION_TYP_FT,
+	DR_ACTION_TYP_CTR,
+	DR_ACTION_TYP_TAG,
+	DR_ACTION_TYP_MODIFY_HDR,
+	DR_ACTION_TYP_VPORT,
+	DR_ACTION_TYP_POP_VLAN,
+	DR_ACTION_TYP_PUSH_VLAN,
+	DR_ACTION_TYP_INSERT_HDR,
+	DR_ACTION_TYP_REMOVE_HDR,
+	DR_ACTION_TYP_SAMPLER,
+	DR_ACTION_TYP_ASO_FLOW_METER,
+	DR_ACTION_TYP_MAX,
+};
+
+enum mlx5dr_ipv {
+	DR_RULE_IPV4,
+	DR_RULE_IPV6,
+	DR_RULE_IPV_MAX,
+};
+
+struct mlx5dr_icm_pool;
+struct mlx5dr_icm_chunk;
+struct mlx5dr_icm_buddy_mem;
+struct mlx5dr_ste_htbl;
+struct mlx5dr_match_param;
+struct mlx5dr_cmd_caps;
+struct mlx5dr_rule_rx_tx;
+struct mlx5dr_matcher_rx_tx;
+struct mlx5dr_ste_ctx;
+
+struct mlx5dr_ste {
+	/* refcount: indicates the num of rules that using this ste */
+	u32 refcount;
+
+	/* this ste is part of a rule, located in ste's chain */
+	u8 ste_chain_location;
+
+	/* attached to the miss_list head at each htbl entry */
+	struct list_head miss_list_node;
+
+	/* this ste is member of htbl */
+	struct mlx5dr_ste_htbl *htbl;
+
+	struct mlx5dr_ste_htbl *next_htbl;
+
+	/* The rule this STE belongs to */
+	struct mlx5dr_rule_rx_tx *rule_rx_tx;
+};
+
+struct mlx5dr_ste_htbl_ctrl {
+	/* total number of valid entries belonging to this hash table. This
+	 * includes the non collision and collision entries
+	 */
+	unsigned int num_of_valid_entries;
+
+	/* total number of collisions entries attached to this table */
+	unsigned int num_of_collisions;
+};
+
+struct mlx5dr_ste_htbl {
+	u16 lu_type;
+	u16 byte_mask;
+	u32 refcount;
+	struct mlx5dr_icm_chunk *chunk;
+	struct mlx5dr_ste *pointing_ste;
+	struct mlx5dr_ste_htbl_ctrl ctrl;
+};
+
+struct mlx5dr_ste_send_info {
+	struct mlx5dr_ste *ste;
+	struct list_head send_list;
+	u16 size;
+	u16 offset;
+	u8 data_cont[DR_STE_SIZE];
+	u8 *data;
+};
+
+void mlx5dr_send_fill_and_append_ste_send_info(struct mlx5dr_ste *ste, u16 size,
+					       u16 offset, u8 *data,
+					       struct mlx5dr_ste_send_info *ste_info,
+					       struct list_head *send_list,
+					       bool copy_data);
+
+struct mlx5dr_ste_build {
+	u8 inner:1;
+	u8 rx:1;
+	u8 vhca_id_valid:1;
+	struct mlx5dr_domain *dmn;
+	struct mlx5dr_cmd_caps *caps;
+	u16 lu_type;
+	u16 byte_mask;
+	u8 bit_mask[DR_STE_SIZE_MASK];
+	int (*ste_build_tag_func)(struct mlx5dr_match_param *spec,
+				  struct mlx5dr_ste_build *sb,
+				  u8 *tag);
+};
+
+struct mlx5dr_ste_htbl *
+mlx5dr_ste_htbl_alloc(struct mlx5dr_icm_pool *pool,
+		      enum mlx5dr_icm_chunk_size chunk_size,
+		      u16 lu_type, u16 byte_mask);
+
+int mlx5dr_ste_htbl_free(struct mlx5dr_ste_htbl *htbl);
+
+static inline void mlx5dr_htbl_put(struct mlx5dr_ste_htbl *htbl)
+{
+	htbl->refcount--;
+	if (!htbl->refcount)
+		mlx5dr_ste_htbl_free(htbl);
+}
+
+static inline void mlx5dr_htbl_get(struct mlx5dr_ste_htbl *htbl)
+{
+	htbl->refcount++;
+}
+
+/* STE utils */
+u32 mlx5dr_ste_calc_hash_index(u8 *hw_ste_p, struct mlx5dr_ste_htbl *htbl);
+void mlx5dr_ste_set_miss_addr(struct mlx5dr_ste_ctx *ste_ctx,
+			      u8 *hw_ste, u64 miss_addr);
+void mlx5dr_ste_set_hit_addr(struct mlx5dr_ste_ctx *ste_ctx,
+			     u8 *hw_ste, u64 icm_addr, u32 ht_size);
+void mlx5dr_ste_set_hit_addr_by_next_htbl(struct mlx5dr_ste_ctx *ste_ctx,
+					  u8 *hw_ste,
+					  struct mlx5dr_ste_htbl *next_htbl);
+void mlx5dr_ste_set_bit_mask(u8 *hw_ste_p, u8 *bit_mask);
+bool mlx5dr_ste_is_last_in_rule(struct mlx5dr_matcher_rx_tx *nic_matcher,
+				u8 ste_location);
+u64 mlx5dr_ste_get_icm_addr(struct mlx5dr_ste *ste);
+u64 mlx5dr_ste_get_mr_addr(struct mlx5dr_ste *ste);
+struct list_head *mlx5dr_ste_get_miss_list(struct mlx5dr_ste *ste);
+
+#define MLX5DR_MAX_VLANS 2
+
+struct mlx5dr_ste_actions_attr {
+	u32	modify_index;
+	u16	modify_actions;
+	u32	decap_index;
+	u16	decap_actions;
+	u8	decap_with_vlan:1;
+	u64	final_icm_addr;
+	u32	flow_tag;
+	u32	ctr_id;
+	u16	gvmi;
+	u16	hit_gvmi;
+	struct {
+		u32	id;
+		u32	size;
+		u8	param_0;
+		u8	param_1;
+	} reformat;
+	struct {
+		int	count;
+		u32	headers[MLX5DR_MAX_VLANS];
+	} vlans;
+
+	struct {
+		u32 obj_id;
+		u32 offset;
+		u8 dest_reg_id;
+		u8 init_color;
+	} aso_flow_meter;
+};
+
+void mlx5dr_ste_set_actions_rx(struct mlx5dr_ste_ctx *ste_ctx,
+			       struct mlx5dr_domain *dmn,
+			       u8 *action_type_set,
+			       u8 *last_ste,
+			       struct mlx5dr_ste_actions_attr *attr,
+			       u32 *added_stes);
+void mlx5dr_ste_set_actions_tx(struct mlx5dr_ste_ctx *ste_ctx,
+			       struct mlx5dr_domain *dmn,
+			       u8 *action_type_set,
+			       u8 *last_ste,
+			       struct mlx5dr_ste_actions_attr *attr,
+			       u32 *added_stes);
+
+void mlx5dr_ste_set_action_set(struct mlx5dr_ste_ctx *ste_ctx,
+			       __be64 *hw_action,
+			       u8 hw_field,
+			       u8 shifter,
+			       u8 length,
+			       u32 data);
+void mlx5dr_ste_set_action_add(struct mlx5dr_ste_ctx *ste_ctx,
+			       __be64 *hw_action,
+			       u8 hw_field,
+			       u8 shifter,
+			       u8 length,
+			       u32 data);
+void mlx5dr_ste_set_action_copy(struct mlx5dr_ste_ctx *ste_ctx,
+				__be64 *hw_action,
+				u8 dst_hw_field,
+				u8 dst_shifter,
+				u8 dst_len,
+				u8 src_hw_field,
+				u8 src_shifter);
+int mlx5dr_ste_set_action_decap_l3_list(struct mlx5dr_ste_ctx *ste_ctx,
+					void *data,
+					u32 data_sz,
+					u8 *hw_action,
+					u32 hw_action_sz,
+					u16 *used_hw_action_num);
+
+const struct mlx5dr_ste_action_modify_field *
+mlx5dr_ste_conv_modify_hdr_sw_field(struct mlx5dr_ste_ctx *ste_ctx, u16 sw_field);
+
+struct mlx5dr_ste_ctx *mlx5dr_ste_get_ctx(u8 version);
+void mlx5dr_ste_free(struct mlx5dr_ste *ste,
+		     struct mlx5dr_matcher *matcher,
+		     struct mlx5dr_matcher_rx_tx *nic_matcher);
+static inline void mlx5dr_ste_put(struct mlx5dr_ste *ste,
+				  struct mlx5dr_matcher *matcher,
+				  struct mlx5dr_matcher_rx_tx *nic_matcher)
+{
+	ste->refcount--;
+	if (!ste->refcount)
+		mlx5dr_ste_free(ste, matcher, nic_matcher);
+}
+
+/* initial as 0, increased only when ste appears in a new rule */
+static inline void mlx5dr_ste_get(struct mlx5dr_ste *ste)
+{
+	ste->refcount++;
+}
+
+static inline bool mlx5dr_ste_is_not_used(struct mlx5dr_ste *ste)
+{
+	return !ste->refcount;
+}
+
+bool mlx5dr_ste_equal_tag(void *src, void *dst);
+int mlx5dr_ste_create_next_htbl(struct mlx5dr_matcher *matcher,
+				struct mlx5dr_matcher_rx_tx *nic_matcher,
+				struct mlx5dr_ste *ste,
+				u8 *cur_hw_ste,
+				enum mlx5dr_icm_chunk_size log_table_size);
+
+/* STE build functions */
+int mlx5dr_ste_build_pre_check(struct mlx5dr_domain *dmn,
+			       u8 match_criteria,
+			       struct mlx5dr_match_param *mask,
+			       struct mlx5dr_match_param *value);
+int mlx5dr_ste_build_ste_arr(struct mlx5dr_matcher *matcher,
+			     struct mlx5dr_matcher_rx_tx *nic_matcher,
+			     struct mlx5dr_match_param *value,
+			     u8 *ste_arr);
+void mlx5dr_ste_build_eth_l2_src_dst(struct mlx5dr_ste_ctx *ste_ctx,
+				     struct mlx5dr_ste_build *builder,
+				     struct mlx5dr_match_param *mask,
+				     bool inner, bool rx);
+void mlx5dr_ste_build_eth_l3_ipv4_5_tuple(struct mlx5dr_ste_ctx *ste_ctx,
+					  struct mlx5dr_ste_build *sb,
+					  struct mlx5dr_match_param *mask,
+					  bool inner, bool rx);
+void mlx5dr_ste_build_eth_l3_ipv4_misc(struct mlx5dr_ste_ctx *ste_ctx,
+				       struct mlx5dr_ste_build *sb,
+				       struct mlx5dr_match_param *mask,
+				       bool inner, bool rx);
+void mlx5dr_ste_build_eth_l3_ipv6_dst(struct mlx5dr_ste_ctx *ste_ctx,
+				      struct mlx5dr_ste_build *sb,
+				      struct mlx5dr_match_param *mask,
+				      bool inner, bool rx);
+void mlx5dr_ste_build_eth_l3_ipv6_src(struct mlx5dr_ste_ctx *ste_ctx,
+				      struct mlx5dr_ste_build *sb,
+				      struct mlx5dr_match_param *mask,
+				      bool inner, bool rx);
+void mlx5dr_ste_build_eth_l2_src(struct mlx5dr_ste_ctx *ste_ctx,
+				 struct mlx5dr_ste_build *sb,
+				 struct mlx5dr_match_param *mask,
+				 bool inner, bool rx);
+void mlx5dr_ste_build_eth_l2_dst(struct mlx5dr_ste_ctx *ste_ctx,
+				 struct mlx5dr_ste_build *sb,
+				 struct mlx5dr_match_param *mask,
+				 bool inner, bool rx);
+void mlx5dr_ste_build_eth_l2_tnl(struct mlx5dr_ste_ctx *ste_ctx,
+				 struct mlx5dr_ste_build *sb,
+				 struct mlx5dr_match_param *mask,
+				 bool inner, bool rx);
+void mlx5dr_ste_build_eth_ipv6_l3_l4(struct mlx5dr_ste_ctx *ste_ctx,
+				     struct mlx5dr_ste_build *sb,
+				     struct mlx5dr_match_param *mask,
+				     bool inner, bool rx);
+void mlx5dr_ste_build_eth_l4_misc(struct mlx5dr_ste_ctx *ste_ctx,
+				  struct mlx5dr_ste_build *sb,
+				  struct mlx5dr_match_param *mask,
+				  bool inner, bool rx);
+void mlx5dr_ste_build_tnl_gre(struct mlx5dr_ste_ctx *ste_ctx,
+			      struct mlx5dr_ste_build *sb,
+			      struct mlx5dr_match_param *mask,
+			      bool inner, bool rx);
+void mlx5dr_ste_build_mpls(struct mlx5dr_ste_ctx *ste_ctx,
+			   struct mlx5dr_ste_build *sb,
+			   struct mlx5dr_match_param *mask,
+			   bool inner, bool rx);
+void mlx5dr_ste_build_tnl_mpls(struct mlx5dr_ste_ctx *ste_ctx,
+			       struct mlx5dr_ste_build *sb,
+			       struct mlx5dr_match_param *mask,
+			       bool inner, bool rx);
+void mlx5dr_ste_build_tnl_mpls_over_gre(struct mlx5dr_ste_ctx *ste_ctx,
+					struct mlx5dr_ste_build *sb,
+					struct mlx5dr_match_param *mask,
+					struct mlx5dr_cmd_caps *caps,
+					bool inner, bool rx);
+void mlx5dr_ste_build_tnl_mpls_over_udp(struct mlx5dr_ste_ctx *ste_ctx,
+					struct mlx5dr_ste_build *sb,
+					struct mlx5dr_match_param *mask,
+					struct mlx5dr_cmd_caps *caps,
+					bool inner, bool rx);
+void mlx5dr_ste_build_icmp(struct mlx5dr_ste_ctx *ste_ctx,
+			   struct mlx5dr_ste_build *sb,
+			   struct mlx5dr_match_param *mask,
+			   struct mlx5dr_cmd_caps *caps,
+			   bool inner, bool rx);
+void mlx5dr_ste_build_tnl_vxlan_gpe(struct mlx5dr_ste_ctx *ste_ctx,
+				    struct mlx5dr_ste_build *sb,
+				    struct mlx5dr_match_param *mask,
+				    bool inner, bool rx);
+void mlx5dr_ste_build_tnl_geneve(struct mlx5dr_ste_ctx *ste_ctx,
+				 struct mlx5dr_ste_build *sb,
+				 struct mlx5dr_match_param *mask,
+				 bool inner, bool rx);
+void mlx5dr_ste_build_tnl_geneve_tlv_opt(struct mlx5dr_ste_ctx *ste_ctx,
+					 struct mlx5dr_ste_build *sb,
+					 struct mlx5dr_match_param *mask,
+					 struct mlx5dr_cmd_caps *caps,
+					 bool inner, bool rx);
+void mlx5dr_ste_build_tnl_geneve_tlv_opt_exist(struct mlx5dr_ste_ctx *ste_ctx,
+					       struct mlx5dr_ste_build *sb,
+					       struct mlx5dr_match_param *mask,
+					       struct mlx5dr_cmd_caps *caps,
+					       bool inner, bool rx);
+void mlx5dr_ste_build_tnl_gtpu(struct mlx5dr_ste_ctx *ste_ctx,
+			       struct mlx5dr_ste_build *sb,
+			       struct mlx5dr_match_param *mask,
+			       bool inner, bool rx);
+void mlx5dr_ste_build_tnl_gtpu_flex_parser_0(struct mlx5dr_ste_ctx *ste_ctx,
+					     struct mlx5dr_ste_build *sb,
+					     struct mlx5dr_match_param *mask,
+					     struct mlx5dr_cmd_caps *caps,
+					     bool inner, bool rx);
+void mlx5dr_ste_build_tnl_gtpu_flex_parser_1(struct mlx5dr_ste_ctx *ste_ctx,
+					     struct mlx5dr_ste_build *sb,
+					     struct mlx5dr_match_param *mask,
+					     struct mlx5dr_cmd_caps *caps,
+					     bool inner, bool rx);
+void mlx5dr_ste_build_tnl_header_0_1(struct mlx5dr_ste_ctx *ste_ctx,
+				     struct mlx5dr_ste_build *sb,
+				     struct mlx5dr_match_param *mask,
+				     bool inner, bool rx);
+void mlx5dr_ste_build_general_purpose(struct mlx5dr_ste_ctx *ste_ctx,
+				      struct mlx5dr_ste_build *sb,
+				      struct mlx5dr_match_param *mask,
+				      bool inner, bool rx);
+void mlx5dr_ste_build_register_0(struct mlx5dr_ste_ctx *ste_ctx,
+				 struct mlx5dr_ste_build *sb,
+				 struct mlx5dr_match_param *mask,
+				 bool inner, bool rx);
+void mlx5dr_ste_build_register_1(struct mlx5dr_ste_ctx *ste_ctx,
+				 struct mlx5dr_ste_build *sb,
+				 struct mlx5dr_match_param *mask,
+				 bool inner, bool rx);
+void mlx5dr_ste_build_src_gvmi_qpn(struct mlx5dr_ste_ctx *ste_ctx,
+				   struct mlx5dr_ste_build *sb,
+				   struct mlx5dr_match_param *mask,
+				   struct mlx5dr_domain *dmn,
+				   bool inner, bool rx);
+void mlx5dr_ste_build_flex_parser_0(struct mlx5dr_ste_ctx *ste_ctx,
+				    struct mlx5dr_ste_build *sb,
+				    struct mlx5dr_match_param *mask,
+				    bool inner, bool rx);
+void mlx5dr_ste_build_flex_parser_1(struct mlx5dr_ste_ctx *ste_ctx,
+				    struct mlx5dr_ste_build *sb,
+				    struct mlx5dr_match_param *mask,
+				    bool inner, bool rx);
+void mlx5dr_ste_build_empty_always_hit(struct mlx5dr_ste_build *sb, bool rx);
+
+/* Actions utils */
+int mlx5dr_actions_build_ste_arr(struct mlx5dr_matcher *matcher,
+				 struct mlx5dr_matcher_rx_tx *nic_matcher,
+				 struct mlx5dr_action *actions[],
+				 u32 num_actions,
+				 u8 *ste_arr,
+				 u32 *new_hw_ste_arr_sz);
+
+struct mlx5dr_match_spec {
+	u32 smac_47_16;		/* Source MAC address of incoming packet */
+	/* Incoming packet Ethertype - this is the Ethertype
+	 * following the last VLAN tag of the packet
+	 */
+	u32 smac_15_0:16;	/* Source MAC address of incoming packet */
+	u32 ethertype:16;
+
+	u32 dmac_47_16;		/* Destination MAC address of incoming packet */
+
+	u32 dmac_15_0:16;	/* Destination MAC address of incoming packet */
+	/* Priority of first VLAN tag in the incoming packet.
+	 * Valid only when cvlan_tag==1 or svlan_tag==1
+	 */
+	u32 first_prio:3;
+	/* CFI bit of first VLAN tag in the incoming packet.
+	 * Valid only when cvlan_tag==1 or svlan_tag==1
+	 */
+	u32 first_cfi:1;
+	/* VLAN ID of first VLAN tag in the incoming packet.
+	 * Valid only when cvlan_tag==1 or svlan_tag==1
+	 */
+	u32 first_vid:12;
+
+	u32 ip_protocol:8;	/* IP protocol */
+	/* Differentiated Services Code Point derived from
+	 * Traffic Class/TOS field of IPv6/v4
+	 */
+	u32 ip_dscp:6;
+	/* Explicit Congestion Notification derived from
+	 * Traffic Class/TOS field of IPv6/v4
+	 */
+	u32 ip_ecn:2;
+	/* The first vlan in the packet is c-vlan (0x8100).
+	 * cvlan_tag and svlan_tag cannot be set together
+	 */
+	u32 cvlan_tag:1;
+	/* The first vlan in the packet is s-vlan (0x8a88).
+	 * cvlan_tag and svlan_tag cannot be set together
+	 */
+	u32 svlan_tag:1;
+	u32 frag:1;		/* Packet is an IP fragment */
+	u32 ip_version:4;	/* IP version */
+	/* TCP flags. ;Bit 0: FIN;Bit 1: SYN;Bit 2: RST;Bit 3: PSH;Bit 4: ACK;
+	 *             Bit 5: URG;Bit 6: ECE;Bit 7: CWR;Bit 8: NS
+	 */
+	u32 tcp_flags:9;
+
+	/* TCP source port.;tcp and udp sport/dport are mutually exclusive */
+	u32 tcp_sport:16;
+	/* TCP destination port.
+	 * tcp and udp sport/dport are mutually exclusive
+	 */
+	u32 tcp_dport:16;
+
+	u32 reserved_auto1:16;
+	u32 ipv4_ihl:4;
+	u32 reserved_auto2:4;
+	u32 ttl_hoplimit:8;
+
+	/* UDP source port.;tcp and udp sport/dport are mutually exclusive */
+	u32 udp_sport:16;
+	/* UDP destination port.;tcp and udp sport/dport are mutually exclusive */
+	u32 udp_dport:16;
+
+	/* IPv6 source address of incoming packets
+	 * For IPv4 address use bits 31:0 (rest of the bits are reserved)
+	 * This field should be qualified by an appropriate ethertype
+	 */
+	u32 src_ip_127_96;
+	/* IPv6 source address of incoming packets
+	 * For IPv4 address use bits 31:0 (rest of the bits are reserved)
+	 * This field should be qualified by an appropriate ethertype
+	 */
+	u32 src_ip_95_64;
+	/* IPv6 source address of incoming packets
+	 * For IPv4 address use bits 31:0 (rest of the bits are reserved)
+	 * This field should be qualified by an appropriate ethertype
+	 */
+	u32 src_ip_63_32;
+	/* IPv6 source address of incoming packets
+	 * For IPv4 address use bits 31:0 (rest of the bits are reserved)
+	 * This field should be qualified by an appropriate ethertype
+	 */
+	u32 src_ip_31_0;
+	/* IPv6 destination address of incoming packets
+	 * For IPv4 address use bits 31:0 (rest of the bits are reserved)
+	 * This field should be qualified by an appropriate ethertype
+	 */
+	u32 dst_ip_127_96;
+	/* IPv6 destination address of incoming packets
+	 * For IPv4 address use bits 31:0 (rest of the bits are reserved)
+	 * This field should be qualified by an appropriate ethertype
+	 */
+	u32 dst_ip_95_64;
+	/* IPv6 destination address of incoming packets
+	 * For IPv4 address use bits 31:0 (rest of the bits are reserved)
+	 * This field should be qualified by an appropriate ethertype
+	 */
+	u32 dst_ip_63_32;
+	/* IPv6 destination address of incoming packets
+	 * For IPv4 address use bits 31:0 (rest of the bits are reserved)
+	 * This field should be qualified by an appropriate ethertype
+	 */
+	u32 dst_ip_31_0;
+};
+
+struct mlx5dr_match_misc {
+	/* used with GRE, checksum exist when gre_c_present == 1 */
+	u32 gre_c_present:1;
+	u32 reserved_auto1:1;
+	/* used with GRE, key exist when gre_k_present == 1 */
+	u32 gre_k_present:1;
+	/* used with GRE, sequence number exist when gre_s_present == 1 */
+	u32 gre_s_present:1;
+	u32 source_vhca_port:4;
+	u32 source_sqn:24;		/* Source SQN */
+
+	u32 source_eswitch_owner_vhca_id:16;
+	/* Source port.;0xffff determines wire port */
+	u32 source_port:16;
+
+	/* Priority of second VLAN tag in the outer header of the incoming packet.
+	 * Valid only when outer_second_cvlan_tag ==1 or outer_second_svlan_tag ==1
+	 */
+	u32 outer_second_prio:3;
+	/* CFI bit of first VLAN tag in the outer header of the incoming packet.
+	 * Valid only when outer_second_cvlan_tag ==1 or outer_second_svlan_tag ==1
+	 */
+	u32 outer_second_cfi:1;
+	/* VLAN ID of first VLAN tag the outer header of the incoming packet.
+	 * Valid only when outer_second_cvlan_tag ==1 or outer_second_svlan_tag ==1
+	 */
+	u32 outer_second_vid:12;
+	/* Priority of second VLAN tag in the inner header of the incoming packet.
+	 * Valid only when inner_second_cvlan_tag ==1 or inner_second_svlan_tag ==1
+	 */
+	u32 inner_second_prio:3;
+	/* CFI bit of first VLAN tag in the inner header of the incoming packet.
+	 * Valid only when inner_second_cvlan_tag ==1 or inner_second_svlan_tag ==1
+	 */
+	u32 inner_second_cfi:1;
+	/* VLAN ID of first VLAN tag the inner header of the incoming packet.
+	 * Valid only when inner_second_cvlan_tag ==1 or inner_second_svlan_tag ==1
+	 */
+	u32 inner_second_vid:12;
+
+	u32 outer_second_cvlan_tag:1;
+	u32 inner_second_cvlan_tag:1;
+	/* The second vlan in the outer header of the packet is c-vlan (0x8100).
+	 * outer_second_cvlan_tag and outer_second_svlan_tag cannot be set together
+	 */
+	u32 outer_second_svlan_tag:1;
+	/* The second vlan in the inner header of the packet is c-vlan (0x8100).
+	 * inner_second_cvlan_tag and inner_second_svlan_tag cannot be set together
+	 */
+	u32 inner_second_svlan_tag:1;
+	/* The second vlan in the outer header of the packet is s-vlan (0x8a88).
+	 * outer_second_cvlan_tag and outer_second_svlan_tag cannot be set together
+	 */
+	u32 reserved_auto2:12;
+	/* The second vlan in the inner header of the packet is s-vlan (0x8a88).
+	 * inner_second_cvlan_tag and inner_second_svlan_tag cannot be set together
+	 */
+	u32 gre_protocol:16;		/* GRE Protocol (outer) */
+
+	u32 gre_key_h:24;		/* GRE Key[31:8] (outer) */
+	u32 gre_key_l:8;		/* GRE Key [7:0] (outer) */
+
+	u32 vxlan_vni:24;		/* VXLAN VNI (outer) */
+	u32 reserved_auto3:8;
+
+	u32 geneve_vni:24;		/* GENEVE VNI field (outer) */
+	u32 reserved_auto4:6;
+	u32 geneve_tlv_option_0_exist:1;
+	u32 geneve_oam:1;		/* GENEVE OAM field (outer) */
+
+	u32 reserved_auto5:12;
+	u32 outer_ipv6_flow_label:20;	/* Flow label of incoming IPv6 packet (outer) */
+
+	u32 reserved_auto6:12;
+	u32 inner_ipv6_flow_label:20;	/* Flow label of incoming IPv6 packet (inner) */
+
+	u32 reserved_auto7:10;
+	u32 geneve_opt_len:6;		/* GENEVE OptLen (outer) */
+	u32 geneve_protocol_type:16;	/* GENEVE protocol type (outer) */
+
+	u32 reserved_auto8:8;
+	u32 bth_dst_qp:24;		/* Destination QP in BTH header */
+
+	u32 reserved_auto9;
+	u32 outer_esp_spi;
+	u32 reserved_auto10[3];
+};
+
+struct mlx5dr_match_misc2 {
+	u32 outer_first_mpls_label:20;		/* First MPLS LABEL (outer) */
+	u32 outer_first_mpls_exp:3;		/* First MPLS EXP (outer) */
+	u32 outer_first_mpls_s_bos:1;		/* First MPLS S_BOS (outer) */
+	u32 outer_first_mpls_ttl:8;		/* First MPLS TTL (outer) */
+
+	u32 inner_first_mpls_label:20;		/* First MPLS LABEL (inner) */
+	u32 inner_first_mpls_exp:3;		/* First MPLS EXP (inner) */
+	u32 inner_first_mpls_s_bos:1;		/* First MPLS S_BOS (inner) */
+	u32 inner_first_mpls_ttl:8;		/* First MPLS TTL (inner) */
+
+	u32 outer_first_mpls_over_gre_label:20;	/* last MPLS LABEL (outer) */
+	u32 outer_first_mpls_over_gre_exp:3;	/* last MPLS EXP (outer) */
+	u32 outer_first_mpls_over_gre_s_bos:1;	/* last MPLS S_BOS (outer) */
+	u32 outer_first_mpls_over_gre_ttl:8;	/* last MPLS TTL (outer) */
+
+	u32 outer_first_mpls_over_udp_label:20;	/* last MPLS LABEL (outer) */
+	u32 outer_first_mpls_over_udp_exp:3;	/* last MPLS EXP (outer) */
+	u32 outer_first_mpls_over_udp_s_bos:1;	/* last MPLS S_BOS (outer) */
+	u32 outer_first_mpls_over_udp_ttl:8;	/* last MPLS TTL (outer) */
+
+	u32 metadata_reg_c_7;			/* metadata_reg_c_7 */
+	u32 metadata_reg_c_6;			/* metadata_reg_c_6 */
+	u32 metadata_reg_c_5;			/* metadata_reg_c_5 */
+	u32 metadata_reg_c_4;			/* metadata_reg_c_4 */
+	u32 metadata_reg_c_3;			/* metadata_reg_c_3 */
+	u32 metadata_reg_c_2;			/* metadata_reg_c_2 */
+	u32 metadata_reg_c_1;			/* metadata_reg_c_1 */
+	u32 metadata_reg_c_0;			/* metadata_reg_c_0 */
+	u32 metadata_reg_a;			/* metadata_reg_a */
+	u32 reserved_auto1[3];
+};
+
+struct mlx5dr_match_misc3 {
+	u32 inner_tcp_seq_num;
+	u32 outer_tcp_seq_num;
+	u32 inner_tcp_ack_num;
+	u32 outer_tcp_ack_num;
+
+	u32 reserved_auto1:8;
+	u32 outer_vxlan_gpe_vni:24;
+
+	u32 outer_vxlan_gpe_next_protocol:8;
+	u32 outer_vxlan_gpe_flags:8;
+	u32 reserved_auto2:16;
+
+	u32 icmpv4_header_data;
+	u32 icmpv6_header_data;
+
+	u8 icmpv4_type;
+	u8 icmpv4_code;
+	u8 icmpv6_type;
+	u8 icmpv6_code;
+
+	u32 geneve_tlv_option_0_data;
+
+	u32 gtpu_teid;
+
+	u8 gtpu_msg_type;
+	u8 gtpu_msg_flags;
+	u32 reserved_auto3:16;
+
+	u32 gtpu_dw_2;
+	u32 gtpu_first_ext_dw_0;
+	u32 gtpu_dw_0;
+	u32 reserved_auto4;
+};
+
+struct mlx5dr_match_misc4 {
+	u32 prog_sample_field_value_0;
+	u32 prog_sample_field_id_0;
+	u32 prog_sample_field_value_1;
+	u32 prog_sample_field_id_1;
+	u32 prog_sample_field_value_2;
+	u32 prog_sample_field_id_2;
+	u32 prog_sample_field_value_3;
+	u32 prog_sample_field_id_3;
+	u32 reserved_auto1[8];
+};
+
+struct mlx5dr_match_misc5 {
+	u32 macsec_tag_0;
+	u32 macsec_tag_1;
+	u32 macsec_tag_2;
+	u32 macsec_tag_3;
+	u32 tunnel_header_0;
+	u32 tunnel_header_1;
+	u32 tunnel_header_2;
+	u32 tunnel_header_3;
+};
+
+struct mlx5dr_match_param {
+	struct mlx5dr_match_spec outer;
+	struct mlx5dr_match_misc misc;
+	struct mlx5dr_match_spec inner;
+	struct mlx5dr_match_misc2 misc2;
+	struct mlx5dr_match_misc3 misc3;
+	struct mlx5dr_match_misc4 misc4;
+	struct mlx5dr_match_misc5 misc5;
+};
+
+#define DR_MASK_IS_ICMPV4_SET(_misc3) ((_misc3)->icmpv4_type || \
+				       (_misc3)->icmpv4_code || \
+				       (_misc3)->icmpv4_header_data)
+
+#define DR_MASK_IS_SRC_IP_SET(_spec) ((_spec)->src_ip_127_96 || \
+				      (_spec)->src_ip_95_64  || \
+				      (_spec)->src_ip_63_32  || \
+				      (_spec)->src_ip_31_0)
+
+#define DR_MASK_IS_DST_IP_SET(_spec) ((_spec)->dst_ip_127_96 || \
+				      (_spec)->dst_ip_95_64  || \
+				      (_spec)->dst_ip_63_32  || \
+				      (_spec)->dst_ip_31_0)
+
+struct mlx5dr_esw_caps {
+	u64 drop_icm_address_rx;
+	u64 drop_icm_address_tx;
+	u64 uplink_icm_address_rx;
+	u64 uplink_icm_address_tx;
+	u8 sw_owner:1;
+	u8 sw_owner_v2:1;
+};
+
+struct mlx5dr_cmd_vport_cap {
+	u16 vport_gvmi;
+	u16 vhca_gvmi;
+	u16 num;
+	u64 icm_address_rx;
+	u64 icm_address_tx;
+};
+
+struct mlx5dr_roce_cap {
+	u8 roce_en:1;
+	u8 fl_rc_qp_when_roce_disabled:1;
+	u8 fl_rc_qp_when_roce_enabled:1;
+};
+
+struct mlx5dr_vports {
+	struct mlx5dr_cmd_vport_cap esw_manager_caps;
+	struct mlx5dr_cmd_vport_cap uplink_caps;
+	struct xarray vports_caps_xa;
+};
+
+struct mlx5dr_cmd_caps {
+	u16 gvmi;
+	u64 nic_rx_drop_address;
+	u64 nic_tx_drop_address;
+	u64 nic_tx_allow_address;
+	u64 esw_rx_drop_address;
+	u64 esw_tx_drop_address;
+	u32 log_icm_size;
+	u64 hdr_modify_icm_addr;
+	u32 flex_protocols;
+	u8 flex_parser_id_icmp_dw0;
+	u8 flex_parser_id_icmp_dw1;
+	u8 flex_parser_id_icmpv6_dw0;
+	u8 flex_parser_id_icmpv6_dw1;
+	u8 flex_parser_id_geneve_tlv_option_0;
+	u8 flex_parser_id_mpls_over_gre;
+	u8 flex_parser_id_mpls_over_udp;
+	u8 flex_parser_id_gtpu_dw_0;
+	u8 flex_parser_id_gtpu_teid;
+	u8 flex_parser_id_gtpu_dw_2;
+	u8 flex_parser_id_gtpu_first_ext_dw_0;
+	u8 flex_parser_ok_bits_supp;
+	u8 max_ft_level;
+	u16 roce_min_src_udp;
+	u8 sw_format_ver;
+	bool eswitch_manager;
+	bool rx_sw_owner;
+	bool tx_sw_owner;
+	bool fdb_sw_owner;
+	u8 rx_sw_owner_v2:1;
+	u8 tx_sw_owner_v2:1;
+	u8 fdb_sw_owner_v2:1;
+	struct mlx5dr_esw_caps esw_caps;
+	struct mlx5dr_vports vports;
+	bool prio_tag_required;
+	struct mlx5dr_roce_cap roce_caps;
+	u8 is_ecpf:1;
+	u8 isolate_vl_tc:1;
+};
+
+enum mlx5dr_domain_nic_type {
+	DR_DOMAIN_NIC_TYPE_RX,
+	DR_DOMAIN_NIC_TYPE_TX,
+};
+
+struct mlx5dr_domain_rx_tx {
+	u64 drop_icm_addr;
+	u64 default_icm_addr;
+	enum mlx5dr_domain_nic_type type;
+	struct mutex mutex; /* protect rx/tx domain */
+};
+
+struct mlx5dr_domain_info {
+	bool supp_sw_steering;
+	u32 max_inline_size;
+	u32 max_send_wr;
+	u32 max_log_sw_icm_sz;
+	u32 max_log_action_icm_sz;
+	struct mlx5dr_domain_rx_tx rx;
+	struct mlx5dr_domain_rx_tx tx;
+	struct mlx5dr_cmd_caps caps;
+};
+
+struct mlx5dr_domain {
+	struct mlx5dr_domain *peer_dmn;
+	struct mlx5_core_dev *mdev;
+	u32 pdn;
+	struct mlx5_uars_page *uar;
+	enum mlx5dr_domain_type type;
+	refcount_t refcount;
+	struct mlx5dr_icm_pool *ste_icm_pool;
+	struct mlx5dr_icm_pool *action_icm_pool;
+	struct mlx5dr_send_ring *send_ring;
+	struct mlx5dr_domain_info info;
+	struct xarray csum_fts_xa;
+	struct mlx5dr_ste_ctx *ste_ctx;
+	struct list_head dbg_tbl_list;
+	struct mlx5dr_dbg_dump_info dump_info;
+};
+
+struct mlx5dr_table_rx_tx {
+	struct mlx5dr_ste_htbl *s_anchor;
+	struct mlx5dr_domain_rx_tx *nic_dmn;
+	u64 default_icm_addr;
+	struct list_head nic_matcher_list;
+};
+
+struct mlx5dr_table {
+	struct mlx5dr_domain *dmn;
+	struct mlx5dr_table_rx_tx rx;
+	struct mlx5dr_table_rx_tx tx;
+	u32 level;
+	u32 table_type;
+	u32 table_id;
+	u32 flags;
+	struct list_head matcher_list;
+	struct mlx5dr_action *miss_action;
+	refcount_t refcount;
+	struct list_head dbg_node;
+};
+
+struct mlx5dr_matcher_rx_tx {
+	struct mlx5dr_ste_htbl *s_htbl;
+	struct mlx5dr_ste_htbl *e_anchor;
+	struct mlx5dr_ste_build *ste_builder;
+	struct mlx5dr_ste_build ste_builder_arr[DR_RULE_IPV_MAX]
+					       [DR_RULE_IPV_MAX]
+					       [DR_RULE_MAX_STES];
+	u8 num_of_builders;
+	u8 num_of_builders_arr[DR_RULE_IPV_MAX][DR_RULE_IPV_MAX];
+	u64 default_icm_addr;
+	struct mlx5dr_table_rx_tx *nic_tbl;
+	u32 prio;
+	struct list_head list_node;
+	u32 rules;
+};
+
+struct mlx5dr_matcher {
+	struct mlx5dr_table *tbl;
+	struct mlx5dr_matcher_rx_tx rx;
+	struct mlx5dr_matcher_rx_tx tx;
+	struct list_head list_node; /* Used for both matchers and dbg managing */
+	u32 prio;
+	struct mlx5dr_match_param mask;
+	u8 match_criteria;
+	refcount_t refcount;
+	struct list_head dbg_rule_list;
+};
+
+struct mlx5dr_ste_action_modify_field {
+	u16 hw_field;
+	u8 start;
+	u8 end;
+	u8 l3_type;
+	u8 l4_type;
+};
+
+struct mlx5dr_action_rewrite {
+	struct mlx5dr_domain *dmn;
+	struct mlx5dr_icm_chunk *chunk;
+	u8 *data;
+	u16 num_of_actions;
+	u32 index;
+	u8 allow_rx:1;
+	u8 allow_tx:1;
+	u8 modify_ttl:1;
+};
+
+struct mlx5dr_action_reformat {
+	struct mlx5dr_domain *dmn;
+	u32 id;
+	u32 size;
+	u8 param_0;
+	u8 param_1;
+};
+
+struct mlx5dr_action_sampler {
+	struct mlx5dr_domain *dmn;
+	u64 rx_icm_addr;
+	u64 tx_icm_addr;
+	u32 sampler_id;
+};
+
+struct mlx5dr_action_dest_tbl {
+	u8 is_fw_tbl:1;
+	union {
+		struct mlx5dr_table *tbl;
+		struct {
+			struct mlx5dr_domain *dmn;
+			u32 id;
+			u32 group_id;
+			enum fs_flow_table_type type;
+			u64 rx_icm_addr;
+			u64 tx_icm_addr;
+			struct mlx5dr_action **ref_actions;
+			u32 num_of_ref_actions;
+		} fw_tbl;
+	};
+};
+
+struct mlx5dr_action_ctr {
+	u32 ctr_id;
+	u32 offset;
+};
+
+struct mlx5dr_action_vport {
+	struct mlx5dr_domain *dmn;
+	struct mlx5dr_cmd_vport_cap *caps;
+};
+
+struct mlx5dr_action_push_vlan {
+	u32 vlan_hdr; /* tpid_pcp_dei_vid */
+};
+
+struct mlx5dr_action_flow_tag {
+	u32 flow_tag;
+};
+
+struct mlx5dr_rule_action_member {
+	struct mlx5dr_action *action;
+	struct list_head list;
+};
+
+struct mlx5dr_action_aso_flow_meter {
+	struct mlx5dr_domain *dmn;
+	u32 obj_id;
+	u32 offset;
+	u8 dest_reg_id;
+	u8 init_color;
+};
+
+struct mlx5dr_action {
+	enum mlx5dr_action_type action_type;
+	refcount_t refcount;
+
+	union {
+		void *data;
+		struct mlx5dr_action_rewrite *rewrite;
+		struct mlx5dr_action_reformat *reformat;
+		struct mlx5dr_action_sampler *sampler;
+		struct mlx5dr_action_dest_tbl *dest_tbl;
+		struct mlx5dr_action_ctr *ctr;
+		struct mlx5dr_action_vport *vport;
+		struct mlx5dr_action_push_vlan *push_vlan;
+		struct mlx5dr_action_flow_tag *flow_tag;
+		struct mlx5dr_action_aso_flow_meter *aso;
+	};
+};
+
+enum mlx5dr_connect_type {
+	CONNECT_HIT	= 1,
+	CONNECT_MISS	= 2,
+};
+
+struct mlx5dr_htbl_connect_info {
+	enum mlx5dr_connect_type type;
+	union {
+		struct mlx5dr_ste_htbl *hit_next_htbl;
+		u64 miss_icm_addr;
+	};
+};
+
+struct mlx5dr_rule_rx_tx {
+	struct mlx5dr_matcher_rx_tx *nic_matcher;
+	struct mlx5dr_ste *last_rule_ste;
+};
+
+struct mlx5dr_rule {
+	struct mlx5dr_matcher *matcher;
+	struct mlx5dr_rule_rx_tx rx;
+	struct mlx5dr_rule_rx_tx tx;
+	struct list_head rule_actions_list;
+	struct list_head dbg_node;
+	u32 flow_source;
+};
+
+void mlx5dr_rule_set_last_member(struct mlx5dr_rule_rx_tx *nic_rule,
+				 struct mlx5dr_ste *ste,
+				 bool force);
+int mlx5dr_rule_get_reverse_rule_members(struct mlx5dr_ste **ste_arr,
+					 struct mlx5dr_ste *curr_ste,
+					 int *num_of_stes);
+
+struct mlx5dr_icm_chunk {
+	struct mlx5dr_icm_buddy_mem *buddy_mem;
+	struct list_head chunk_list;
+
+	/* indicates the index of this chunk in the whole memory,
+	 * used for deleting the chunk from the buddy
+	 */
+	unsigned int seg;
+	enum mlx5dr_icm_chunk_size size;
+
+	/* Memory optimisation */
+	struct mlx5dr_ste *ste_arr;
+	u8 *hw_ste_arr;
+	struct list_head *miss_list;
+};
+
+static inline void mlx5dr_domain_nic_lock(struct mlx5dr_domain_rx_tx *nic_dmn)
+{
+	mutex_lock(&nic_dmn->mutex);
+}
+
+static inline void mlx5dr_domain_nic_unlock(struct mlx5dr_domain_rx_tx *nic_dmn)
+{
+	mutex_unlock(&nic_dmn->mutex);
+}
+
+static inline void mlx5dr_domain_lock(struct mlx5dr_domain *dmn)
+{
+	mlx5dr_domain_nic_lock(&dmn->info.rx);
+	mlx5dr_domain_nic_lock(&dmn->info.tx);
+}
+
+static inline void mlx5dr_domain_unlock(struct mlx5dr_domain *dmn)
+{
+	mlx5dr_domain_nic_unlock(&dmn->info.tx);
+	mlx5dr_domain_nic_unlock(&dmn->info.rx);
+}
+
+int mlx5dr_matcher_add_to_tbl_nic(struct mlx5dr_domain *dmn,
+				  struct mlx5dr_matcher_rx_tx *nic_matcher);
+int mlx5dr_matcher_remove_from_tbl_nic(struct mlx5dr_domain *dmn,
+				       struct mlx5dr_matcher_rx_tx *nic_matcher);
+
+int mlx5dr_matcher_select_builders(struct mlx5dr_matcher *matcher,
+				   struct mlx5dr_matcher_rx_tx *nic_matcher,
+				   enum mlx5dr_ipv outer_ipv,
+				   enum mlx5dr_ipv inner_ipv);
+
+u64 mlx5dr_icm_pool_get_chunk_mr_addr(struct mlx5dr_icm_chunk *chunk);
+u32 mlx5dr_icm_pool_get_chunk_rkey(struct mlx5dr_icm_chunk *chunk);
+u64 mlx5dr_icm_pool_get_chunk_icm_addr(struct mlx5dr_icm_chunk *chunk);
+u32 mlx5dr_icm_pool_get_chunk_num_of_entries(struct mlx5dr_icm_chunk *chunk);
+u32 mlx5dr_icm_pool_get_chunk_byte_size(struct mlx5dr_icm_chunk *chunk);
+u8 *mlx5dr_ste_get_hw_ste(struct mlx5dr_ste *ste);
+
+static inline int
+mlx5dr_icm_pool_dm_type_to_entry_size(enum mlx5dr_icm_type icm_type)
+{
+	if (icm_type == DR_ICM_TYPE_STE)
+		return DR_STE_SIZE;
+
+	return DR_MODIFY_ACTION_SIZE;
+}
+
+static inline u32
+mlx5dr_icm_pool_chunk_size_to_entries(enum mlx5dr_icm_chunk_size chunk_size)
+{
+	return 1 << chunk_size;
+}
+
+static inline int
+mlx5dr_icm_pool_chunk_size_to_byte(enum mlx5dr_icm_chunk_size chunk_size,
+				   enum mlx5dr_icm_type icm_type)
+{
+	int num_of_entries;
+	int entry_size;
+
+	entry_size = mlx5dr_icm_pool_dm_type_to_entry_size(icm_type);
+	num_of_entries = mlx5dr_icm_pool_chunk_size_to_entries(chunk_size);
+
+	return entry_size * num_of_entries;
+}
+
+static inline int
+mlx5dr_ste_htbl_increase_threshold(struct mlx5dr_ste_htbl *htbl)
+{
+	int num_of_entries =
+		mlx5dr_icm_pool_chunk_size_to_entries(htbl->chunk->size);
+
+	/* Threshold is 50%, one is added to table of size 1 */
+	return (num_of_entries + 1) / 2;
+}
+
+static inline bool
+mlx5dr_ste_htbl_may_grow(struct mlx5dr_ste_htbl *htbl)
+{
+	if (htbl->chunk->size == DR_CHUNK_SIZE_MAX - 1 || !htbl->byte_mask)
+		return false;
+
+	return true;
+}
+
+struct mlx5dr_cmd_vport_cap *
+mlx5dr_domain_get_vport_cap(struct mlx5dr_domain *dmn, u16 vport);
+
+struct mlx5dr_cmd_query_flow_table_details {
+	u8 status;
+	u8 level;
+	u64 sw_owner_icm_root_1;
+	u64 sw_owner_icm_root_0;
+};
+
+struct mlx5dr_cmd_create_flow_table_attr {
+	u32 table_type;
+	u16 uid;
+	u64 icm_addr_rx;
+	u64 icm_addr_tx;
+	u8 level;
+	bool sw_owner;
+	bool term_tbl;
+	bool decap_en;
+	bool reformat_en;
+};
+
+/* internal API functions */
+int mlx5dr_cmd_query_device(struct mlx5_core_dev *mdev,
+			    struct mlx5dr_cmd_caps *caps);
+int mlx5dr_cmd_query_esw_vport_context(struct mlx5_core_dev *mdev,
+				       bool other_vport, u16 vport_number,
+				       u64 *icm_address_rx,
+				       u64 *icm_address_tx);
+int mlx5dr_cmd_query_gvmi(struct mlx5_core_dev *mdev,
+			  bool other_vport, u16 vport_number, u16 *gvmi);
+int mlx5dr_cmd_query_esw_caps(struct mlx5_core_dev *mdev,
+			      struct mlx5dr_esw_caps *caps);
+int mlx5dr_cmd_query_flow_sampler(struct mlx5_core_dev *dev,
+				  u32 sampler_id,
+				  u64 *rx_icm_addr,
+				  u64 *tx_icm_addr);
+int mlx5dr_cmd_sync_steering(struct mlx5_core_dev *mdev);
+int mlx5dr_cmd_set_fte_modify_and_vport(struct mlx5_core_dev *mdev,
+					u32 table_type,
+					u32 table_id,
+					u32 group_id,
+					u32 modify_header_id,
+					u16 vport_id);
+int mlx5dr_cmd_del_flow_table_entry(struct mlx5_core_dev *mdev,
+				    u32 table_type,
+				    u32 table_id);
+int mlx5dr_cmd_alloc_modify_header(struct mlx5_core_dev *mdev,
+				   u32 table_type,
+				   u8 num_of_actions,
+				   u64 *actions,
+				   u32 *modify_header_id);
+int mlx5dr_cmd_dealloc_modify_header(struct mlx5_core_dev *mdev,
+				     u32 modify_header_id);
+int mlx5dr_cmd_create_empty_flow_group(struct mlx5_core_dev *mdev,
+				       u32 table_type,
+				       u32 table_id,
+				       u32 *group_id);
+int mlx5dr_cmd_destroy_flow_group(struct mlx5_core_dev *mdev,
+				  u32 table_type,
+				  u32 table_id,
+				  u32 group_id);
+int mlx5dr_cmd_create_flow_table(struct mlx5_core_dev *mdev,
+				 struct mlx5dr_cmd_create_flow_table_attr *attr,
+				 u64 *fdb_rx_icm_addr,
+				 u32 *table_id);
+int mlx5dr_cmd_destroy_flow_table(struct mlx5_core_dev *mdev,
+				  u32 table_id,
+				  u32 table_type);
+int mlx5dr_cmd_query_flow_table(struct mlx5_core_dev *dev,
+				enum fs_flow_table_type type,
+				u32 table_id,
+				struct mlx5dr_cmd_query_flow_table_details *output);
+int mlx5dr_cmd_create_reformat_ctx(struct mlx5_core_dev *mdev,
+				   enum mlx5_reformat_ctx_type rt,
+				   u8 reformat_param_0,
+				   u8 reformat_param_1,
+				   size_t reformat_size,
+				   void *reformat_data,
+				   u32 *reformat_id);
+void mlx5dr_cmd_destroy_reformat_ctx(struct mlx5_core_dev *mdev,
+				     u32 reformat_id);
+
+struct mlx5dr_cmd_gid_attr {
+	u8 gid[16];
+	u8 mac[6];
+	u32 roce_ver;
+};
+
+int mlx5dr_cmd_query_gid(struct mlx5_core_dev *mdev, u8 vhca_port_num,
+			 u16 index, struct mlx5dr_cmd_gid_attr *attr);
+
+struct mlx5dr_icm_pool *mlx5dr_icm_pool_create(struct mlx5dr_domain *dmn,
+					       enum mlx5dr_icm_type icm_type);
+void mlx5dr_icm_pool_destroy(struct mlx5dr_icm_pool *pool);
+
+struct mlx5dr_icm_chunk *
+mlx5dr_icm_alloc_chunk(struct mlx5dr_icm_pool *pool,
+		       enum mlx5dr_icm_chunk_size chunk_size);
+void mlx5dr_icm_free_chunk(struct mlx5dr_icm_chunk *chunk);
+
+void mlx5dr_ste_prepare_for_postsend(struct mlx5dr_ste_ctx *ste_ctx,
+				     u8 *hw_ste_p, u32 ste_size);
+int mlx5dr_ste_htbl_init_and_postsend(struct mlx5dr_domain *dmn,
+				      struct mlx5dr_domain_rx_tx *nic_dmn,
+				      struct mlx5dr_ste_htbl *htbl,
+				      struct mlx5dr_htbl_connect_info *connect_info,
+				      bool update_hw_ste);
+void mlx5dr_ste_set_formatted_ste(struct mlx5dr_ste_ctx *ste_ctx,
+				  u16 gvmi,
+				  enum mlx5dr_domain_nic_type nic_type,
+				  struct mlx5dr_ste_htbl *htbl,
+				  u8 *formatted_ste,
+				  struct mlx5dr_htbl_connect_info *connect_info);
+void mlx5dr_ste_copy_param(u8 match_criteria,
+			   struct mlx5dr_match_param *set_param,
+			   struct mlx5dr_match_parameters *mask,
+			   bool clear);
+
+struct mlx5dr_qp {
+	struct mlx5_core_dev *mdev;
+	struct mlx5_wq_qp wq;
+	struct mlx5_uars_page *uar;
+	struct mlx5_wq_ctrl wq_ctrl;
+	u32 qpn;
+	struct {
+		unsigned int pc;
+		unsigned int cc;
+		unsigned int size;
+		unsigned int *wqe_head;
+		unsigned int wqe_cnt;
+	} sq;
+	struct {
+		unsigned int pc;
+		unsigned int cc;
+		unsigned int size;
+		unsigned int wqe_cnt;
+	} rq;
+	int max_inline_data;
+};
+
+struct mlx5dr_cq {
+	struct mlx5_core_dev *mdev;
+	struct mlx5_cqwq wq;
+	struct mlx5_wq_ctrl wq_ctrl;
+	struct mlx5_core_cq mcq;
+	struct mlx5dr_qp *qp;
+};
+
+struct mlx5dr_mr {
+	struct mlx5_core_dev *mdev;
+	u32 mkey;
+	dma_addr_t dma_addr;
+	void *addr;
+	size_t size;
+};
+
+#define MAX_SEND_CQE		64
+#define MIN_READ_SYNC		64
+
+struct mlx5dr_send_ring {
+	struct mlx5dr_cq *cq;
+	struct mlx5dr_qp *qp;
+	struct mlx5dr_mr *mr;
+	/* How much wqes are waiting for completion */
+	u32 pending_wqe;
+	/* Signal request per this trash hold value */
+	u16 signal_th;
+	/* Each post_send_size less than max_post_send_size */
+	u32 max_post_send_size;
+	/* manage the send queue */
+	u32 tx_head;
+	void *buf;
+	u32 buf_size;
+	u8 sync_buff[MIN_READ_SYNC];
+	struct mlx5dr_mr *sync_mr;
+	spinlock_t lock; /* Protect the data path of the send ring */
+	bool err_state; /* send_ring is not usable in err state */
+};
+
+int mlx5dr_send_ring_alloc(struct mlx5dr_domain *dmn);
+void mlx5dr_send_ring_free(struct mlx5dr_domain *dmn,
+			   struct mlx5dr_send_ring *send_ring);
+int mlx5dr_send_ring_force_drain(struct mlx5dr_domain *dmn);
+int mlx5dr_send_postsend_ste(struct mlx5dr_domain *dmn,
+			     struct mlx5dr_ste *ste,
+			     u8 *data,
+			     u16 size,
+			     u16 offset);
+int mlx5dr_send_postsend_htbl(struct mlx5dr_domain *dmn,
+			      struct mlx5dr_ste_htbl *htbl,
+			      u8 *formatted_ste, u8 *mask);
+int mlx5dr_send_postsend_formatted_htbl(struct mlx5dr_domain *dmn,
+					struct mlx5dr_ste_htbl *htbl,
+					u8 *ste_init_data,
+					bool update_hw_ste);
+int mlx5dr_send_postsend_action(struct mlx5dr_domain *dmn,
+				struct mlx5dr_action *action);
+
+struct mlx5dr_cmd_ft_info {
+	u32 id;
+	u16 vport;
+	enum fs_flow_table_type type;
+};
+
+struct mlx5dr_cmd_flow_destination_hw_info {
+	enum mlx5_flow_destination_type type;
+	union {
+		u32 tir_num;
+		u32 ft_num;
+		u32 ft_id;
+		u32 counter_id;
+		u32 sampler_id;
+		struct {
+			u16 num;
+			u16 vhca_id;
+			u32 reformat_id;
+			u8 flags;
+		} vport;
+	};
+};
+
+struct mlx5dr_cmd_fte_info {
+	u32 dests_size;
+	u32 index;
+	struct mlx5_flow_context flow_context;
+	u32 *val;
+	struct mlx5_flow_act action;
+	struct mlx5dr_cmd_flow_destination_hw_info *dest_arr;
+	bool ignore_flow_level;
+};
+
+int mlx5dr_cmd_set_fte(struct mlx5_core_dev *dev,
+		       int opmod, int modify_mask,
+		       struct mlx5dr_cmd_ft_info *ft,
+		       u32 group_id,
+		       struct mlx5dr_cmd_fte_info *fte);
+
+bool mlx5dr_ste_supp_ttl_cs_recalc(struct mlx5dr_cmd_caps *caps);
+
+struct mlx5dr_fw_recalc_cs_ft {
+	u64 rx_icm_addr;
+	u32 table_id;
+	u32 group_id;
+	u32 modify_hdr_id;
+};
+
+struct mlx5dr_fw_recalc_cs_ft *
+mlx5dr_fw_create_recalc_cs_ft(struct mlx5dr_domain *dmn, u16 vport_num);
+void mlx5dr_fw_destroy_recalc_cs_ft(struct mlx5dr_domain *dmn,
+				    struct mlx5dr_fw_recalc_cs_ft *recalc_cs_ft);
+int mlx5dr_domain_get_recalc_cs_ft_addr(struct mlx5dr_domain *dmn,
+					u16 vport_num,
+					u64 *rx_icm_addr);
+int mlx5dr_fw_create_md_tbl(struct mlx5dr_domain *dmn,
+			    struct mlx5dr_cmd_flow_destination_hw_info *dest,
+			    int num_dest,
+			    bool reformat_req,
+			    u32 *tbl_id,
+			    u32 *group_id,
+			    bool ignore_flow_level,
+			    u32 flow_source);
+void mlx5dr_fw_destroy_md_tbl(struct mlx5dr_domain *dmn, u32 tbl_id,
+			      u32 group_id);
+#endif  /* _DR_TYPES_H_ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/fs_dr.c b/drivers/net/ethernet/mellanox/mlx5/core/steering/fs_dr.c
new file mode 100644
index 000000000..13b6d4721
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/fs_dr.c
@@ -0,0 +1,820 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2019 Mellanox Technologies */
+
+#include <linux/mlx5/vport.h>
+#include "mlx5_core.h"
+#include "fs_core.h"
+#include "fs_cmd.h"
+#include "mlx5dr.h"
+#include "fs_dr.h"
+
+static bool mlx5_dr_is_fw_table(u32 flags)
+{
+	if (flags & MLX5_FLOW_TABLE_TERMINATION)
+		return true;
+
+	return false;
+}
+
+static int mlx5_cmd_dr_update_root_ft(struct mlx5_flow_root_namespace *ns,
+				      struct mlx5_flow_table *ft,
+				      u32 underlay_qpn,
+				      bool disconnect)
+{
+	return mlx5_fs_cmd_get_fw_cmds()->update_root_ft(ns, ft, underlay_qpn,
+							 disconnect);
+}
+
+static int set_miss_action(struct mlx5_flow_root_namespace *ns,
+			   struct mlx5_flow_table *ft,
+			   struct mlx5_flow_table *next_ft)
+{
+	struct mlx5dr_action *old_miss_action;
+	struct mlx5dr_action *action = NULL;
+	struct mlx5dr_table *next_tbl;
+	int err;
+
+	next_tbl = next_ft ? next_ft->fs_dr_table.dr_table : NULL;
+	if (next_tbl) {
+		action = mlx5dr_action_create_dest_table(next_tbl);
+		if (!action)
+			return -EINVAL;
+	}
+	old_miss_action = ft->fs_dr_table.miss_action;
+	err = mlx5dr_table_set_miss_action(ft->fs_dr_table.dr_table, action);
+	if (err && action) {
+		err = mlx5dr_action_destroy(action);
+		if (err)
+			mlx5_core_err(ns->dev,
+				      "Failed to destroy action (%d)\n", err);
+		action = NULL;
+	}
+	ft->fs_dr_table.miss_action = action;
+	if (old_miss_action) {
+		err = mlx5dr_action_destroy(old_miss_action);
+		if (err)
+			mlx5_core_err(ns->dev, "Failed to destroy action (%d)\n",
+				      err);
+	}
+
+	return err;
+}
+
+static int mlx5_cmd_dr_create_flow_table(struct mlx5_flow_root_namespace *ns,
+					 struct mlx5_flow_table *ft,
+					 struct mlx5_flow_table_attr *ft_attr,
+					 struct mlx5_flow_table *next_ft)
+{
+	struct mlx5dr_table *tbl;
+	u32 flags;
+	int err;
+
+	if (mlx5_dr_is_fw_table(ft->flags))
+		return mlx5_fs_cmd_get_fw_cmds()->create_flow_table(ns, ft,
+								    ft_attr,
+								    next_ft);
+	flags = ft->flags;
+	/* turn off encap/decap if not supported for sw-str by fw */
+	if (!MLX5_CAP_FLOWTABLE(ns->dev, sw_owner_reformat_supported))
+		flags = ft->flags & ~(MLX5_FLOW_TABLE_TUNNEL_EN_REFORMAT |
+				      MLX5_FLOW_TABLE_TUNNEL_EN_DECAP);
+
+	tbl = mlx5dr_table_create(ns->fs_dr_domain.dr_domain, ft->level, flags,
+				  ft_attr->uid);
+	if (!tbl) {
+		mlx5_core_err(ns->dev, "Failed creating dr flow_table\n");
+		return -EINVAL;
+	}
+
+	ft->fs_dr_table.dr_table = tbl;
+	ft->id = mlx5dr_table_get_id(tbl);
+
+	if (next_ft) {
+		err = set_miss_action(ns, ft, next_ft);
+		if (err) {
+			mlx5dr_table_destroy(tbl);
+			ft->fs_dr_table.dr_table = NULL;
+			return err;
+		}
+	}
+
+	ft->max_fte = INT_MAX;
+
+	return 0;
+}
+
+static int mlx5_cmd_dr_destroy_flow_table(struct mlx5_flow_root_namespace *ns,
+					  struct mlx5_flow_table *ft)
+{
+	struct mlx5dr_action *action = ft->fs_dr_table.miss_action;
+	int err;
+
+	if (mlx5_dr_is_fw_table(ft->flags))
+		return mlx5_fs_cmd_get_fw_cmds()->destroy_flow_table(ns, ft);
+
+	err = mlx5dr_table_destroy(ft->fs_dr_table.dr_table);
+	if (err) {
+		mlx5_core_err(ns->dev, "Failed to destroy flow_table (%d)\n",
+			      err);
+		return err;
+	}
+	if (action) {
+		err = mlx5dr_action_destroy(action);
+		if (err) {
+			mlx5_core_err(ns->dev, "Failed to destroy action(%d)\n",
+				      err);
+			return err;
+		}
+	}
+
+	return err;
+}
+
+static int mlx5_cmd_dr_modify_flow_table(struct mlx5_flow_root_namespace *ns,
+					 struct mlx5_flow_table *ft,
+					 struct mlx5_flow_table *next_ft)
+{
+	if (mlx5_dr_is_fw_table(ft->flags))
+		return mlx5_fs_cmd_get_fw_cmds()->modify_flow_table(ns, ft, next_ft);
+
+	return set_miss_action(ns, ft, next_ft);
+}
+
+static int mlx5_cmd_dr_create_flow_group(struct mlx5_flow_root_namespace *ns,
+					 struct mlx5_flow_table *ft,
+					 u32 *in,
+					 struct mlx5_flow_group *fg)
+{
+	struct mlx5dr_matcher *matcher;
+	u32 priority = MLX5_GET(create_flow_group_in, in,
+				start_flow_index);
+	u8 match_criteria_enable = MLX5_GET(create_flow_group_in,
+					    in,
+					    match_criteria_enable);
+	struct mlx5dr_match_parameters mask;
+
+	if (mlx5_dr_is_fw_table(ft->flags))
+		return mlx5_fs_cmd_get_fw_cmds()->create_flow_group(ns, ft, in,
+								    fg);
+
+	mask.match_buf = MLX5_ADDR_OF(create_flow_group_in,
+				      in, match_criteria);
+	mask.match_sz = sizeof(fg->mask.match_criteria);
+
+	matcher = mlx5dr_matcher_create(ft->fs_dr_table.dr_table,
+					priority,
+					match_criteria_enable,
+					&mask);
+	if (!matcher) {
+		mlx5_core_err(ns->dev, "Failed creating matcher\n");
+		return -EINVAL;
+	}
+
+	fg->fs_dr_matcher.dr_matcher = matcher;
+	return 0;
+}
+
+static int mlx5_cmd_dr_destroy_flow_group(struct mlx5_flow_root_namespace *ns,
+					  struct mlx5_flow_table *ft,
+					  struct mlx5_flow_group *fg)
+{
+	if (mlx5_dr_is_fw_table(ft->flags))
+		return mlx5_fs_cmd_get_fw_cmds()->destroy_flow_group(ns, ft, fg);
+
+	return mlx5dr_matcher_destroy(fg->fs_dr_matcher.dr_matcher);
+}
+
+static struct mlx5dr_action *create_vport_action(struct mlx5dr_domain *domain,
+						 struct mlx5_flow_rule *dst)
+{
+	struct mlx5_flow_destination *dest_attr = &dst->dest_attr;
+
+	return mlx5dr_action_create_dest_vport(domain, dest_attr->vport.num,
+					       dest_attr->vport.flags &
+					       MLX5_FLOW_DEST_VPORT_VHCA_ID,
+					       dest_attr->vport.vhca_id);
+}
+
+static struct mlx5dr_action *create_uplink_action(struct mlx5dr_domain *domain,
+						  struct mlx5_flow_rule *dst)
+{
+	struct mlx5_flow_destination *dest_attr = &dst->dest_attr;
+
+	return mlx5dr_action_create_dest_vport(domain, MLX5_VPORT_UPLINK, 1,
+					       dest_attr->vport.vhca_id);
+}
+
+static struct mlx5dr_action *create_ft_action(struct mlx5dr_domain *domain,
+					      struct mlx5_flow_rule *dst)
+{
+	struct mlx5_flow_table *dest_ft = dst->dest_attr.ft;
+
+	if (mlx5_dr_is_fw_table(dest_ft->flags))
+		return mlx5dr_action_create_dest_flow_fw_table(domain, dest_ft);
+	return mlx5dr_action_create_dest_table(dest_ft->fs_dr_table.dr_table);
+}
+
+static struct mlx5dr_action *create_action_push_vlan(struct mlx5dr_domain *domain,
+						     struct mlx5_fs_vlan *vlan)
+{
+	u16 n_ethtype = vlan->ethtype;
+	u8  prio = vlan->prio;
+	u16 vid = vlan->vid;
+	u32 vlan_hdr;
+
+	vlan_hdr = (u32)n_ethtype << 16 | (u32)(prio) << 12 |  (u32)vid;
+	return mlx5dr_action_create_push_vlan(domain, htonl(vlan_hdr));
+}
+
+static bool contain_vport_reformat_action(struct mlx5_flow_rule *dst)
+{
+	return (dst->dest_attr.type == MLX5_FLOW_DESTINATION_TYPE_VPORT ||
+		dst->dest_attr.type == MLX5_FLOW_DESTINATION_TYPE_UPLINK) &&
+		dst->dest_attr.vport.flags & MLX5_FLOW_DEST_VPORT_REFORMAT_ID;
+}
+
+/* We want to support a rule with 32 destinations, which means we need to
+ * account for 32 destinations plus usually a counter plus one more action
+ * for a multi-destination flow table.
+ */
+#define MLX5_FLOW_CONTEXT_ACTION_MAX  34
+static int mlx5_cmd_dr_create_fte(struct mlx5_flow_root_namespace *ns,
+				  struct mlx5_flow_table *ft,
+				  struct mlx5_flow_group *group,
+				  struct fs_fte *fte)
+{
+	struct mlx5dr_domain *domain = ns->fs_dr_domain.dr_domain;
+	struct mlx5dr_action_dest *term_actions;
+	struct mlx5dr_match_parameters params;
+	struct mlx5_core_dev *dev = ns->dev;
+	struct mlx5dr_action **fs_dr_actions;
+	struct mlx5dr_action *tmp_action;
+	struct mlx5dr_action **actions;
+	bool delay_encap_set = false;
+	struct mlx5dr_rule *rule;
+	struct mlx5_flow_rule *dst;
+	int fs_dr_num_actions = 0;
+	int num_term_actions = 0;
+	int num_actions = 0;
+	size_t match_sz;
+	int err = 0;
+	int i;
+
+	if (mlx5_dr_is_fw_table(ft->flags))
+		return mlx5_fs_cmd_get_fw_cmds()->create_fte(ns, ft, group, fte);
+
+	actions = kcalloc(MLX5_FLOW_CONTEXT_ACTION_MAX, sizeof(*actions),
+			  GFP_KERNEL);
+	if (!actions) {
+		err = -ENOMEM;
+		goto out_err;
+	}
+
+	fs_dr_actions = kcalloc(MLX5_FLOW_CONTEXT_ACTION_MAX,
+				sizeof(*fs_dr_actions), GFP_KERNEL);
+	if (!fs_dr_actions) {
+		err = -ENOMEM;
+		goto free_actions_alloc;
+	}
+
+	term_actions = kcalloc(MLX5_FLOW_CONTEXT_ACTION_MAX,
+			       sizeof(*term_actions), GFP_KERNEL);
+	if (!term_actions) {
+		err = -ENOMEM;
+		goto free_fs_dr_actions_alloc;
+	}
+
+	match_sz = sizeof(fte->val);
+
+	/* Drop reformat action bit if destination vport set with reformat */
+	if (fte->action.action & MLX5_FLOW_CONTEXT_ACTION_FWD_DEST) {
+		list_for_each_entry(dst, &fte->node.children, node.list) {
+			if (!contain_vport_reformat_action(dst))
+				continue;
+
+			fte->action.action &= ~MLX5_FLOW_CONTEXT_ACTION_PACKET_REFORMAT;
+			break;
+		}
+	}
+
+	/* The order of the actions are must to be keep, only the following
+	 * order is supported by SW steering:
+	 * TX: modify header -> push vlan -> encap
+	 * RX: decap -> pop vlan -> modify header
+	 */
+	if (fte->action.action & MLX5_FLOW_CONTEXT_ACTION_DECAP) {
+		enum mlx5dr_action_reformat_type decap_type =
+			DR_ACTION_REFORMAT_TYP_TNL_L2_TO_L2;
+
+		tmp_action = mlx5dr_action_create_packet_reformat(domain,
+								  decap_type,
+								  0, 0, 0,
+								  NULL);
+		if (!tmp_action) {
+			err = -ENOMEM;
+			goto free_actions;
+		}
+		fs_dr_actions[fs_dr_num_actions++] = tmp_action;
+		actions[num_actions++] = tmp_action;
+	}
+
+	if (fte->action.action & MLX5_FLOW_CONTEXT_ACTION_PACKET_REFORMAT) {
+		bool is_decap = fte->action.pkt_reformat->reformat_type ==
+			MLX5_REFORMAT_TYPE_L3_TUNNEL_TO_L2;
+
+		if (is_decap)
+			actions[num_actions++] =
+				fte->action.pkt_reformat->action.dr_action;
+		else
+			delay_encap_set = true;
+	}
+
+	if (fte->action.action & MLX5_FLOW_CONTEXT_ACTION_VLAN_POP) {
+		tmp_action =
+			mlx5dr_action_create_pop_vlan();
+		if (!tmp_action) {
+			err = -ENOMEM;
+			goto free_actions;
+		}
+		fs_dr_actions[fs_dr_num_actions++] = tmp_action;
+		actions[num_actions++] = tmp_action;
+	}
+
+	if (fte->action.action & MLX5_FLOW_CONTEXT_ACTION_VLAN_POP_2) {
+		tmp_action =
+			mlx5dr_action_create_pop_vlan();
+		if (!tmp_action) {
+			err = -ENOMEM;
+			goto free_actions;
+		}
+		fs_dr_actions[fs_dr_num_actions++] = tmp_action;
+		actions[num_actions++] = tmp_action;
+	}
+
+	if (fte->action.action & MLX5_FLOW_CONTEXT_ACTION_MOD_HDR)
+		actions[num_actions++] =
+			fte->action.modify_hdr->action.dr_action;
+
+	if (fte->action.action & MLX5_FLOW_CONTEXT_ACTION_VLAN_PUSH) {
+		tmp_action = create_action_push_vlan(domain, &fte->action.vlan[0]);
+		if (!tmp_action) {
+			err = -ENOMEM;
+			goto free_actions;
+		}
+		fs_dr_actions[fs_dr_num_actions++] = tmp_action;
+		actions[num_actions++] = tmp_action;
+	}
+
+	if (fte->action.action & MLX5_FLOW_CONTEXT_ACTION_VLAN_PUSH_2) {
+		tmp_action = create_action_push_vlan(domain, &fte->action.vlan[1]);
+		if (!tmp_action) {
+			err = -ENOMEM;
+			goto free_actions;
+		}
+		fs_dr_actions[fs_dr_num_actions++] = tmp_action;
+		actions[num_actions++] = tmp_action;
+	}
+
+	if (delay_encap_set)
+		actions[num_actions++] =
+			fte->action.pkt_reformat->action.dr_action;
+
+	/* The order of the actions below is not important */
+
+	if (fte->action.action & MLX5_FLOW_CONTEXT_ACTION_DROP) {
+		tmp_action = mlx5dr_action_create_drop();
+		if (!tmp_action) {
+			err = -ENOMEM;
+			goto free_actions;
+		}
+		fs_dr_actions[fs_dr_num_actions++] = tmp_action;
+		term_actions[num_term_actions++].dest = tmp_action;
+	}
+
+	if (fte->flow_context.flow_tag) {
+		tmp_action =
+			mlx5dr_action_create_tag(fte->flow_context.flow_tag);
+		if (!tmp_action) {
+			err = -ENOMEM;
+			goto free_actions;
+		}
+		fs_dr_actions[fs_dr_num_actions++] = tmp_action;
+		actions[num_actions++] = tmp_action;
+	}
+
+	if (fte->action.action & MLX5_FLOW_CONTEXT_ACTION_FWD_DEST) {
+		list_for_each_entry(dst, &fte->node.children, node.list) {
+			enum mlx5_flow_destination_type type = dst->dest_attr.type;
+			u32 id;
+
+			if (fs_dr_num_actions == MLX5_FLOW_CONTEXT_ACTION_MAX ||
+			    num_term_actions == MLX5_FLOW_CONTEXT_ACTION_MAX) {
+				err = -EOPNOTSUPP;
+				goto free_actions;
+			}
+
+			if (type == MLX5_FLOW_DESTINATION_TYPE_COUNTER)
+				continue;
+
+			switch (type) {
+			case MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE:
+				tmp_action = create_ft_action(domain, dst);
+				if (!tmp_action) {
+					err = -ENOMEM;
+					goto free_actions;
+				}
+				fs_dr_actions[fs_dr_num_actions++] = tmp_action;
+				term_actions[num_term_actions++].dest = tmp_action;
+				break;
+			case MLX5_FLOW_DESTINATION_TYPE_UPLINK:
+			case MLX5_FLOW_DESTINATION_TYPE_VPORT:
+				tmp_action = type == MLX5_FLOW_DESTINATION_TYPE_VPORT ?
+					     create_vport_action(domain, dst) :
+					     create_uplink_action(domain, dst);
+				if (!tmp_action) {
+					err = -ENOMEM;
+					goto free_actions;
+				}
+				fs_dr_actions[fs_dr_num_actions++] = tmp_action;
+				term_actions[num_term_actions].dest = tmp_action;
+
+				if (dst->dest_attr.vport.flags &
+				    MLX5_FLOW_DEST_VPORT_REFORMAT_ID)
+					term_actions[num_term_actions].reformat =
+						dst->dest_attr.vport.pkt_reformat->action.dr_action;
+
+				num_term_actions++;
+				break;
+			case MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE_NUM:
+				id = dst->dest_attr.ft_num;
+				tmp_action = mlx5dr_action_create_dest_table_num(domain,
+										 id);
+				if (!tmp_action) {
+					err = -ENOMEM;
+					goto free_actions;
+				}
+				fs_dr_actions[fs_dr_num_actions++] = tmp_action;
+				term_actions[num_term_actions++].dest = tmp_action;
+				break;
+			case MLX5_FLOW_DESTINATION_TYPE_FLOW_SAMPLER:
+				id = dst->dest_attr.sampler_id;
+				tmp_action = mlx5dr_action_create_flow_sampler(domain,
+									       id);
+				if (!tmp_action) {
+					err = -ENOMEM;
+					goto free_actions;
+				}
+				fs_dr_actions[fs_dr_num_actions++] = tmp_action;
+				term_actions[num_term_actions++].dest = tmp_action;
+				break;
+			default:
+				err = -EOPNOTSUPP;
+				goto free_actions;
+			}
+		}
+	}
+
+	if (fte->action.action & MLX5_FLOW_CONTEXT_ACTION_COUNT) {
+		list_for_each_entry(dst, &fte->node.children, node.list) {
+			u32 id;
+
+			if (dst->dest_attr.type !=
+			    MLX5_FLOW_DESTINATION_TYPE_COUNTER)
+				continue;
+
+			if (num_actions == MLX5_FLOW_CONTEXT_ACTION_MAX ||
+			    fs_dr_num_actions == MLX5_FLOW_CONTEXT_ACTION_MAX) {
+				err = -EOPNOTSUPP;
+				goto free_actions;
+			}
+
+			id = dst->dest_attr.counter_id;
+			tmp_action =
+				mlx5dr_action_create_flow_counter(id);
+			if (!tmp_action) {
+				err = -ENOMEM;
+				goto free_actions;
+			}
+
+			fs_dr_actions[fs_dr_num_actions++] = tmp_action;
+			actions[num_actions++] = tmp_action;
+		}
+	}
+
+	if (fte->action.action & MLX5_FLOW_CONTEXT_ACTION_EXECUTE_ASO) {
+		if (fte->action.exe_aso.type != MLX5_EXE_ASO_FLOW_METER) {
+			err = -EOPNOTSUPP;
+			goto free_actions;
+		}
+
+		tmp_action =
+			mlx5dr_action_create_aso(domain,
+						 fte->action.exe_aso.object_id,
+						 fte->action.exe_aso.return_reg_id,
+						 fte->action.exe_aso.type,
+						 fte->action.exe_aso.flow_meter.init_color,
+						 fte->action.exe_aso.flow_meter.meter_idx);
+		if (!tmp_action) {
+			err = -ENOMEM;
+			goto free_actions;
+		}
+		fs_dr_actions[fs_dr_num_actions++] = tmp_action;
+		actions[num_actions++] = tmp_action;
+	}
+
+	params.match_sz = match_sz;
+	params.match_buf = (u64 *)fte->val;
+	if (num_term_actions == 1) {
+		if (term_actions->reformat) {
+			if (num_actions == MLX5_FLOW_CONTEXT_ACTION_MAX) {
+				err = -EOPNOTSUPP;
+				goto free_actions;
+			}
+			actions[num_actions++] = term_actions->reformat;
+		}
+
+		if (num_actions == MLX5_FLOW_CONTEXT_ACTION_MAX) {
+			err = -EOPNOTSUPP;
+			goto free_actions;
+		}
+		actions[num_actions++] = term_actions->dest;
+	} else if (num_term_actions > 1) {
+		bool ignore_flow_level =
+			!!(fte->action.flags & FLOW_ACT_IGNORE_FLOW_LEVEL);
+		u32 flow_source = fte->flow_context.flow_source;
+
+		if (num_actions == MLX5_FLOW_CONTEXT_ACTION_MAX ||
+		    fs_dr_num_actions == MLX5_FLOW_CONTEXT_ACTION_MAX) {
+			err = -EOPNOTSUPP;
+			goto free_actions;
+		}
+		tmp_action = mlx5dr_action_create_mult_dest_tbl(domain,
+								term_actions,
+								num_term_actions,
+								ignore_flow_level,
+								flow_source);
+		if (!tmp_action) {
+			err = -EOPNOTSUPP;
+			goto free_actions;
+		}
+		fs_dr_actions[fs_dr_num_actions++] = tmp_action;
+		actions[num_actions++] = tmp_action;
+	}
+
+	rule = mlx5dr_rule_create(group->fs_dr_matcher.dr_matcher,
+				  &params,
+				  num_actions,
+				  actions,
+				  fte->flow_context.flow_source);
+	if (!rule) {
+		err = -EINVAL;
+		goto free_actions;
+	}
+
+	kfree(term_actions);
+	kfree(actions);
+
+	fte->fs_dr_rule.dr_rule = rule;
+	fte->fs_dr_rule.num_actions = fs_dr_num_actions;
+	fte->fs_dr_rule.dr_actions = fs_dr_actions;
+
+	return 0;
+
+free_actions:
+	/* Free in reverse order to handle action dependencies */
+	for (i = fs_dr_num_actions - 1; i >= 0; i--)
+		if (!IS_ERR_OR_NULL(fs_dr_actions[i]))
+			mlx5dr_action_destroy(fs_dr_actions[i]);
+
+	kfree(term_actions);
+free_fs_dr_actions_alloc:
+	kfree(fs_dr_actions);
+free_actions_alloc:
+	kfree(actions);
+out_err:
+	mlx5_core_err(dev, "Failed to create dr rule err(%d)\n", err);
+	return err;
+}
+
+static int mlx5_cmd_dr_packet_reformat_alloc(struct mlx5_flow_root_namespace *ns,
+					     struct mlx5_pkt_reformat_params *params,
+					     enum mlx5_flow_namespace_type namespace,
+					     struct mlx5_pkt_reformat *pkt_reformat)
+{
+	struct mlx5dr_domain *dr_domain = ns->fs_dr_domain.dr_domain;
+	struct mlx5dr_action *action;
+	int dr_reformat;
+
+	switch (params->type) {
+	case MLX5_REFORMAT_TYPE_L2_TO_VXLAN:
+	case MLX5_REFORMAT_TYPE_L2_TO_NVGRE:
+	case MLX5_REFORMAT_TYPE_L2_TO_L2_TUNNEL:
+		dr_reformat = DR_ACTION_REFORMAT_TYP_L2_TO_TNL_L2;
+		break;
+	case MLX5_REFORMAT_TYPE_L3_TUNNEL_TO_L2:
+		dr_reformat = DR_ACTION_REFORMAT_TYP_TNL_L3_TO_L2;
+		break;
+	case MLX5_REFORMAT_TYPE_L2_TO_L3_TUNNEL:
+		dr_reformat = DR_ACTION_REFORMAT_TYP_L2_TO_TNL_L3;
+		break;
+	case MLX5_REFORMAT_TYPE_INSERT_HDR:
+		dr_reformat = DR_ACTION_REFORMAT_TYP_INSERT_HDR;
+		break;
+	case MLX5_REFORMAT_TYPE_REMOVE_HDR:
+		dr_reformat = DR_ACTION_REFORMAT_TYP_REMOVE_HDR;
+		break;
+	default:
+		mlx5_core_err(ns->dev, "Packet-reformat not supported(%d)\n",
+			      params->type);
+		return -EOPNOTSUPP;
+	}
+
+	action = mlx5dr_action_create_packet_reformat(dr_domain,
+						      dr_reformat,
+						      params->param_0,
+						      params->param_1,
+						      params->size,
+						      params->data);
+	if (!action) {
+		mlx5_core_err(ns->dev, "Failed allocating packet-reformat action\n");
+		return -EINVAL;
+	}
+
+	pkt_reformat->action.dr_action = action;
+
+	return 0;
+}
+
+static void mlx5_cmd_dr_packet_reformat_dealloc(struct mlx5_flow_root_namespace *ns,
+						struct mlx5_pkt_reformat *pkt_reformat)
+{
+	mlx5dr_action_destroy(pkt_reformat->action.dr_action);
+}
+
+static int mlx5_cmd_dr_modify_header_alloc(struct mlx5_flow_root_namespace *ns,
+					   u8 namespace, u8 num_actions,
+					   void *modify_actions,
+					   struct mlx5_modify_hdr *modify_hdr)
+{
+	struct mlx5dr_domain *dr_domain = ns->fs_dr_domain.dr_domain;
+	struct mlx5dr_action *action;
+	size_t actions_sz;
+
+	actions_sz = MLX5_UN_SZ_BYTES(set_add_copy_action_in_auto) *
+		num_actions;
+	action = mlx5dr_action_create_modify_header(dr_domain, 0,
+						    actions_sz,
+						    modify_actions);
+	if (!action) {
+		mlx5_core_err(ns->dev, "Failed allocating modify-header action\n");
+		return -EINVAL;
+	}
+
+	modify_hdr->action.dr_action = action;
+
+	return 0;
+}
+
+static void mlx5_cmd_dr_modify_header_dealloc(struct mlx5_flow_root_namespace *ns,
+					      struct mlx5_modify_hdr *modify_hdr)
+{
+	mlx5dr_action_destroy(modify_hdr->action.dr_action);
+}
+
+static int
+mlx5_cmd_dr_destroy_match_definer(struct mlx5_flow_root_namespace *ns,
+				  int definer_id)
+{
+	return -EOPNOTSUPP;
+}
+
+static int mlx5_cmd_dr_create_match_definer(struct mlx5_flow_root_namespace *ns,
+					    u16 format_id, u32 *match_mask)
+{
+	return -EOPNOTSUPP;
+}
+
+static int mlx5_cmd_dr_delete_fte(struct mlx5_flow_root_namespace *ns,
+				  struct mlx5_flow_table *ft,
+				  struct fs_fte *fte)
+{
+	struct mlx5_fs_dr_rule *rule = &fte->fs_dr_rule;
+	int err;
+	int i;
+
+	if (mlx5_dr_is_fw_table(ft->flags))
+		return mlx5_fs_cmd_get_fw_cmds()->delete_fte(ns, ft, fte);
+
+	err = mlx5dr_rule_destroy(rule->dr_rule);
+	if (err)
+		return err;
+
+	/* Free in reverse order to handle action dependencies */
+	for (i = rule->num_actions - 1; i >= 0; i--)
+		if (!IS_ERR_OR_NULL(rule->dr_actions[i]))
+			mlx5dr_action_destroy(rule->dr_actions[i]);
+
+	kfree(rule->dr_actions);
+	return 0;
+}
+
+static int mlx5_cmd_dr_update_fte(struct mlx5_flow_root_namespace *ns,
+				  struct mlx5_flow_table *ft,
+				  struct mlx5_flow_group *group,
+				  int modify_mask,
+				  struct fs_fte *fte)
+{
+	struct fs_fte fte_tmp = {};
+	int ret;
+
+	if (mlx5_dr_is_fw_table(ft->flags))
+		return mlx5_fs_cmd_get_fw_cmds()->update_fte(ns, ft, group, modify_mask, fte);
+
+	/* Backup current dr rule details */
+	fte_tmp.fs_dr_rule = fte->fs_dr_rule;
+	memset(&fte->fs_dr_rule, 0, sizeof(struct mlx5_fs_dr_rule));
+
+	/* First add the new updated rule, then delete the old rule */
+	ret = mlx5_cmd_dr_create_fte(ns, ft, group, fte);
+	if (ret)
+		goto restore_fte;
+
+	ret = mlx5_cmd_dr_delete_fte(ns, ft, &fte_tmp);
+	WARN_ONCE(ret, "dr update fte duplicate rule deletion failed\n");
+	return ret;
+
+restore_fte:
+	fte->fs_dr_rule = fte_tmp.fs_dr_rule;
+	return ret;
+}
+
+static int mlx5_cmd_dr_set_peer(struct mlx5_flow_root_namespace *ns,
+				struct mlx5_flow_root_namespace *peer_ns)
+{
+	struct mlx5dr_domain *peer_domain = NULL;
+
+	if (peer_ns)
+		peer_domain = peer_ns->fs_dr_domain.dr_domain;
+	mlx5dr_domain_set_peer(ns->fs_dr_domain.dr_domain,
+			       peer_domain);
+	return 0;
+}
+
+static int mlx5_cmd_dr_create_ns(struct mlx5_flow_root_namespace *ns)
+{
+	ns->fs_dr_domain.dr_domain =
+		mlx5dr_domain_create(ns->dev,
+				     MLX5DR_DOMAIN_TYPE_FDB);
+	if (!ns->fs_dr_domain.dr_domain) {
+		mlx5_core_err(ns->dev, "Failed to create dr flow namespace\n");
+		return -EOPNOTSUPP;
+	}
+	return 0;
+}
+
+static int mlx5_cmd_dr_destroy_ns(struct mlx5_flow_root_namespace *ns)
+{
+	return mlx5dr_domain_destroy(ns->fs_dr_domain.dr_domain);
+}
+
+static u32 mlx5_cmd_dr_get_capabilities(struct mlx5_flow_root_namespace *ns,
+					enum fs_flow_table_type ft_type)
+{
+	if (ft_type != FS_FT_FDB ||
+	    MLX5_CAP_GEN(ns->dev, steering_format_version) == MLX5_STEERING_FORMAT_CONNECTX_5)
+		return 0;
+
+	return MLX5_FLOW_STEERING_CAP_VLAN_PUSH_ON_RX | MLX5_FLOW_STEERING_CAP_VLAN_POP_ON_TX;
+}
+
+bool mlx5_fs_dr_is_supported(struct mlx5_core_dev *dev)
+{
+	return mlx5dr_is_supported(dev);
+}
+
+static const struct mlx5_flow_cmds mlx5_flow_cmds_dr = {
+	.create_flow_table = mlx5_cmd_dr_create_flow_table,
+	.destroy_flow_table = mlx5_cmd_dr_destroy_flow_table,
+	.modify_flow_table = mlx5_cmd_dr_modify_flow_table,
+	.create_flow_group = mlx5_cmd_dr_create_flow_group,
+	.destroy_flow_group = mlx5_cmd_dr_destroy_flow_group,
+	.create_fte = mlx5_cmd_dr_create_fte,
+	.update_fte = mlx5_cmd_dr_update_fte,
+	.delete_fte = mlx5_cmd_dr_delete_fte,
+	.update_root_ft = mlx5_cmd_dr_update_root_ft,
+	.packet_reformat_alloc = mlx5_cmd_dr_packet_reformat_alloc,
+	.packet_reformat_dealloc = mlx5_cmd_dr_packet_reformat_dealloc,
+	.modify_header_alloc = mlx5_cmd_dr_modify_header_alloc,
+	.modify_header_dealloc = mlx5_cmd_dr_modify_header_dealloc,
+	.create_match_definer = mlx5_cmd_dr_create_match_definer,
+	.destroy_match_definer = mlx5_cmd_dr_destroy_match_definer,
+	.set_peer = mlx5_cmd_dr_set_peer,
+	.create_ns = mlx5_cmd_dr_create_ns,
+	.destroy_ns = mlx5_cmd_dr_destroy_ns,
+	.get_capabilities = mlx5_cmd_dr_get_capabilities,
+};
+
+const struct mlx5_flow_cmds *mlx5_fs_cmd_get_dr_cmds(void)
+{
+		return &mlx5_flow_cmds_dr;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/fs_dr.h b/drivers/net/ethernet/mellanox/mlx5/core/steering/fs_dr.h
new file mode 100644
index 000000000..d16862206
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/fs_dr.h
@@ -0,0 +1,56 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+ * Copyright (c) 2019 Mellanox Technologies
+ */
+
+#ifndef _MLX5_FS_DR_
+#define _MLX5_FS_DR_
+
+#include "mlx5dr.h"
+
+struct mlx5_flow_root_namespace;
+struct fs_fte;
+
+struct mlx5_fs_dr_action {
+	struct mlx5dr_action *dr_action;
+};
+
+struct mlx5_fs_dr_rule {
+	struct mlx5dr_rule    *dr_rule;
+	/* Only actions created by fs_dr */
+	struct mlx5dr_action  **dr_actions;
+	int                      num_actions;
+};
+
+struct mlx5_fs_dr_domain {
+	struct mlx5dr_domain	*dr_domain;
+};
+
+struct mlx5_fs_dr_matcher {
+	struct mlx5dr_matcher *dr_matcher;
+};
+
+struct mlx5_fs_dr_table {
+	struct mlx5dr_table  *dr_table;
+	struct mlx5dr_action *miss_action;
+};
+
+#ifdef CONFIG_MLX5_SW_STEERING
+
+bool mlx5_fs_dr_is_supported(struct mlx5_core_dev *dev);
+
+const struct mlx5_flow_cmds *mlx5_fs_cmd_get_dr_cmds(void);
+
+#else
+
+static inline const struct mlx5_flow_cmds *mlx5_fs_cmd_get_dr_cmds(void)
+{
+	return NULL;
+}
+
+static inline bool mlx5_fs_dr_is_supported(struct mlx5_core_dev *dev)
+{
+	return false;
+}
+
+#endif /* CONFIG_MLX5_SW_STEERING */
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/mlx5_ifc_dr.h b/drivers/net/ethernet/mellanox/mlx5/core/steering/mlx5_ifc_dr.h
new file mode 100644
index 000000000..fb078fa0f
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/mlx5_ifc_dr.h
@@ -0,0 +1,603 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2019, Mellanox Technologies */
+
+#ifndef MLX5_IFC_DR_H
+#define MLX5_IFC_DR_H
+
+enum {
+	MLX5DR_STE_LU_TYPE_DONT_CARE			= 0x0f,
+};
+
+struct mlx5_ifc_ste_general_bits {
+	u8         entry_type[0x4];
+	u8         reserved_at_4[0x4];
+	u8         entry_sub_type[0x8];
+	u8         byte_mask[0x10];
+
+	u8         next_table_base_63_48[0x10];
+	u8         next_lu_type[0x8];
+	u8         next_table_base_39_32_size[0x8];
+
+	u8         next_table_base_31_5_size[0x1b];
+	u8         linear_hash_enable[0x1];
+	u8         reserved_at_5c[0x2];
+	u8         next_table_rank[0x2];
+
+	u8         reserved_at_60[0xa0];
+	u8         tag_value[0x60];
+	u8         bit_mask[0x60];
+};
+
+struct mlx5_ifc_ste_sx_transmit_bits {
+	u8         entry_type[0x4];
+	u8         reserved_at_4[0x4];
+	u8         entry_sub_type[0x8];
+	u8         byte_mask[0x10];
+
+	u8         next_table_base_63_48[0x10];
+	u8         next_lu_type[0x8];
+	u8         next_table_base_39_32_size[0x8];
+
+	u8         next_table_base_31_5_size[0x1b];
+	u8         linear_hash_enable[0x1];
+	u8         reserved_at_5c[0x2];
+	u8         next_table_rank[0x2];
+
+	u8         sx_wire[0x1];
+	u8         sx_func_lb[0x1];
+	u8         sx_sniffer[0x1];
+	u8         sx_wire_enable[0x1];
+	u8         sx_func_lb_enable[0x1];
+	u8         sx_sniffer_enable[0x1];
+	u8         action_type[0x3];
+	u8         reserved_at_69[0x1];
+	u8         action_description[0x6];
+	u8         gvmi[0x10];
+
+	u8         encap_pointer_vlan_data[0x20];
+
+	u8         loopback_syndome_en[0x8];
+	u8         loopback_syndome[0x8];
+	u8         counter_trigger[0x10];
+
+	u8         miss_address_63_48[0x10];
+	u8         counter_trigger_23_16[0x8];
+	u8         miss_address_39_32[0x8];
+
+	u8         miss_address_31_6[0x1a];
+	u8         learning_point[0x1];
+	u8         go_back[0x1];
+	u8         match_polarity[0x1];
+	u8         mask_mode[0x1];
+	u8         miss_rank[0x2];
+};
+
+struct mlx5_ifc_ste_rx_steering_mult_bits {
+	u8         entry_type[0x4];
+	u8         reserved_at_4[0x4];
+	u8         entry_sub_type[0x8];
+	u8         byte_mask[0x10];
+
+	u8         next_table_base_63_48[0x10];
+	u8         next_lu_type[0x8];
+	u8         next_table_base_39_32_size[0x8];
+
+	u8         next_table_base_31_5_size[0x1b];
+	u8         linear_hash_enable[0x1];
+	u8         reserved_at_[0x2];
+	u8         next_table_rank[0x2];
+
+	u8         member_count[0x10];
+	u8         gvmi[0x10];
+
+	u8         qp_list_pointer[0x20];
+
+	u8         reserved_at_a0[0x1];
+	u8         tunneling_action[0x3];
+	u8         action_description[0x4];
+	u8         reserved_at_a8[0x8];
+	u8         counter_trigger_15_0[0x10];
+
+	u8         miss_address_63_48[0x10];
+	u8         counter_trigger_23_16[0x08];
+	u8         miss_address_39_32[0x8];
+
+	u8         miss_address_31_6[0x1a];
+	u8         learning_point[0x1];
+	u8         fail_on_error[0x1];
+	u8         match_polarity[0x1];
+	u8         mask_mode[0x1];
+	u8         miss_rank[0x2];
+};
+
+struct mlx5_ifc_ste_modify_packet_bits {
+	u8         entry_type[0x4];
+	u8         reserved_at_4[0x4];
+	u8         entry_sub_type[0x8];
+	u8         byte_mask[0x10];
+
+	u8         next_table_base_63_48[0x10];
+	u8         next_lu_type[0x8];
+	u8         next_table_base_39_32_size[0x8];
+
+	u8         next_table_base_31_5_size[0x1b];
+	u8         linear_hash_enable[0x1];
+	u8         reserved_at_[0x2];
+	u8         next_table_rank[0x2];
+
+	u8         number_of_re_write_actions[0x10];
+	u8         gvmi[0x10];
+
+	u8         header_re_write_actions_pointer[0x20];
+
+	u8         reserved_at_a0[0x1];
+	u8         tunneling_action[0x3];
+	u8         action_description[0x4];
+	u8         reserved_at_a8[0x8];
+	u8         counter_trigger_15_0[0x10];
+
+	u8         miss_address_63_48[0x10];
+	u8         counter_trigger_23_16[0x08];
+	u8         miss_address_39_32[0x8];
+
+	u8         miss_address_31_6[0x1a];
+	u8         learning_point[0x1];
+	u8         fail_on_error[0x1];
+	u8         match_polarity[0x1];
+	u8         mask_mode[0x1];
+	u8         miss_rank[0x2];
+};
+
+struct mlx5_ifc_ste_eth_l2_src_bits {
+	u8         smac_47_16[0x20];
+
+	u8         smac_15_0[0x10];
+	u8         l3_ethertype[0x10];
+
+	u8         qp_type[0x2];
+	u8         ethertype_filter[0x1];
+	u8         reserved_at_43[0x1];
+	u8         sx_sniffer[0x1];
+	u8         force_lb[0x1];
+	u8         functional_lb[0x1];
+	u8         port[0x1];
+	u8         reserved_at_48[0x4];
+	u8         first_priority[0x3];
+	u8         first_cfi[0x1];
+	u8         first_vlan_qualifier[0x2];
+	u8         reserved_at_52[0x2];
+	u8         first_vlan_id[0xc];
+
+	u8         ip_fragmented[0x1];
+	u8         tcp_syn[0x1];
+	u8         encp_type[0x2];
+	u8         l3_type[0x2];
+	u8         l4_type[0x2];
+	u8         reserved_at_68[0x4];
+	u8         second_priority[0x3];
+	u8         second_cfi[0x1];
+	u8         second_vlan_qualifier[0x2];
+	u8         reserved_at_72[0x2];
+	u8         second_vlan_id[0xc];
+};
+
+struct mlx5_ifc_ste_eth_l2_dst_bits {
+	u8         dmac_47_16[0x20];
+
+	u8         dmac_15_0[0x10];
+	u8         l3_ethertype[0x10];
+
+	u8         qp_type[0x2];
+	u8         ethertype_filter[0x1];
+	u8         reserved_at_43[0x1];
+	u8         sx_sniffer[0x1];
+	u8         force_lb[0x1];
+	u8         functional_lb[0x1];
+	u8         port[0x1];
+	u8         reserved_at_48[0x4];
+	u8         first_priority[0x3];
+	u8         first_cfi[0x1];
+	u8         first_vlan_qualifier[0x2];
+	u8         reserved_at_52[0x2];
+	u8         first_vlan_id[0xc];
+
+	u8         ip_fragmented[0x1];
+	u8         tcp_syn[0x1];
+	u8         encp_type[0x2];
+	u8         l3_type[0x2];
+	u8         l4_type[0x2];
+	u8         reserved_at_68[0x4];
+	u8         second_priority[0x3];
+	u8         second_cfi[0x1];
+	u8         second_vlan_qualifier[0x2];
+	u8         reserved_at_72[0x2];
+	u8         second_vlan_id[0xc];
+};
+
+struct mlx5_ifc_ste_eth_l2_src_dst_bits {
+	u8         dmac_47_16[0x20];
+
+	u8         dmac_15_0[0x10];
+	u8         smac_47_32[0x10];
+
+	u8         smac_31_0[0x20];
+
+	u8         sx_sniffer[0x1];
+	u8         force_lb[0x1];
+	u8         functional_lb[0x1];
+	u8         port[0x1];
+	u8         l3_type[0x2];
+	u8         reserved_at_66[0x6];
+	u8         first_priority[0x3];
+	u8         first_cfi[0x1];
+	u8         first_vlan_qualifier[0x2];
+	u8         reserved_at_72[0x2];
+	u8         first_vlan_id[0xc];
+};
+
+struct mlx5_ifc_ste_eth_l3_ipv4_5_tuple_bits {
+	u8         destination_address[0x20];
+
+	u8         source_address[0x20];
+
+	u8         source_port[0x10];
+	u8         destination_port[0x10];
+
+	u8         fragmented[0x1];
+	u8         first_fragment[0x1];
+	u8         reserved_at_62[0x2];
+	u8         reserved_at_64[0x1];
+	u8         ecn[0x2];
+	u8         tcp_ns[0x1];
+	u8         tcp_cwr[0x1];
+	u8         tcp_ece[0x1];
+	u8         tcp_urg[0x1];
+	u8         tcp_ack[0x1];
+	u8         tcp_psh[0x1];
+	u8         tcp_rst[0x1];
+	u8         tcp_syn[0x1];
+	u8         tcp_fin[0x1];
+	u8         dscp[0x6];
+	u8         reserved_at_76[0x2];
+	u8         protocol[0x8];
+};
+
+struct mlx5_ifc_ste_eth_l3_ipv6_dst_bits {
+	u8         dst_ip_127_96[0x20];
+
+	u8         dst_ip_95_64[0x20];
+
+	u8         dst_ip_63_32[0x20];
+
+	u8         dst_ip_31_0[0x20];
+};
+
+struct mlx5_ifc_ste_eth_l2_tnl_bits {
+	u8         dmac_47_16[0x20];
+
+	u8         dmac_15_0[0x10];
+	u8         l3_ethertype[0x10];
+
+	u8         l2_tunneling_network_id[0x20];
+
+	u8         ip_fragmented[0x1];
+	u8         tcp_syn[0x1];
+	u8         encp_type[0x2];
+	u8         l3_type[0x2];
+	u8         l4_type[0x2];
+	u8         first_priority[0x3];
+	u8         first_cfi[0x1];
+	u8         reserved_at_6c[0x3];
+	u8         gre_key_flag[0x1];
+	u8         first_vlan_qualifier[0x2];
+	u8         reserved_at_72[0x2];
+	u8         first_vlan_id[0xc];
+};
+
+struct mlx5_ifc_ste_eth_l3_ipv6_src_bits {
+	u8         src_ip_127_96[0x20];
+
+	u8         src_ip_95_64[0x20];
+
+	u8         src_ip_63_32[0x20];
+
+	u8         src_ip_31_0[0x20];
+};
+
+struct mlx5_ifc_ste_eth_l3_ipv4_misc_bits {
+	u8         version[0x4];
+	u8         ihl[0x4];
+	u8         reserved_at_8[0x8];
+	u8         total_length[0x10];
+
+	u8         identification[0x10];
+	u8         flags[0x3];
+	u8         fragment_offset[0xd];
+
+	u8         time_to_live[0x8];
+	u8         reserved_at_48[0x8];
+	u8         checksum[0x10];
+
+	u8         reserved_at_60[0x20];
+};
+
+struct mlx5_ifc_ste_eth_l4_bits {
+	u8         fragmented[0x1];
+	u8         first_fragment[0x1];
+	u8         reserved_at_2[0x6];
+	u8         protocol[0x8];
+	u8         dst_port[0x10];
+
+	u8         ipv6_version[0x4];
+	u8         reserved_at_24[0x1];
+	u8         ecn[0x2];
+	u8         tcp_ns[0x1];
+	u8         tcp_cwr[0x1];
+	u8         tcp_ece[0x1];
+	u8         tcp_urg[0x1];
+	u8         tcp_ack[0x1];
+	u8         tcp_psh[0x1];
+	u8         tcp_rst[0x1];
+	u8         tcp_syn[0x1];
+	u8         tcp_fin[0x1];
+	u8         src_port[0x10];
+
+	u8         ipv6_payload_length[0x10];
+	u8         ipv6_hop_limit[0x8];
+	u8         dscp[0x6];
+	u8         reserved_at_5e[0x2];
+
+	u8         tcp_data_offset[0x4];
+	u8         reserved_at_64[0x8];
+	u8         flow_label[0x14];
+};
+
+struct mlx5_ifc_ste_eth_l4_misc_bits {
+	u8         checksum[0x10];
+	u8         length[0x10];
+
+	u8         seq_num[0x20];
+
+	u8         ack_num[0x20];
+
+	u8         urgent_pointer[0x10];
+	u8         window_size[0x10];
+};
+
+struct mlx5_ifc_ste_mpls_bits {
+	u8         mpls0_label[0x14];
+	u8         mpls0_exp[0x3];
+	u8         mpls0_s_bos[0x1];
+	u8         mpls0_ttl[0x8];
+
+	u8         mpls1_label[0x20];
+
+	u8         mpls2_label[0x20];
+
+	u8         reserved_at_60[0x16];
+	u8         mpls4_s_bit[0x1];
+	u8         mpls4_qualifier[0x1];
+	u8         mpls3_s_bit[0x1];
+	u8         mpls3_qualifier[0x1];
+	u8         mpls2_s_bit[0x1];
+	u8         mpls2_qualifier[0x1];
+	u8         mpls1_s_bit[0x1];
+	u8         mpls1_qualifier[0x1];
+	u8         mpls0_s_bit[0x1];
+	u8         mpls0_qualifier[0x1];
+};
+
+struct mlx5_ifc_ste_register_0_bits {
+	u8         register_0_h[0x20];
+
+	u8         register_0_l[0x20];
+
+	u8         register_1_h[0x20];
+
+	u8         register_1_l[0x20];
+};
+
+struct mlx5_ifc_ste_register_1_bits {
+	u8         register_2_h[0x20];
+
+	u8         register_2_l[0x20];
+
+	u8         register_3_h[0x20];
+
+	u8         register_3_l[0x20];
+};
+
+struct mlx5_ifc_ste_gre_bits {
+	u8         gre_c_present[0x1];
+	u8         reserved_at_30[0x1];
+	u8         gre_k_present[0x1];
+	u8         gre_s_present[0x1];
+	u8         strict_src_route[0x1];
+	u8         recur[0x3];
+	u8         flags[0x5];
+	u8         version[0x3];
+	u8         gre_protocol[0x10];
+
+	u8         checksum[0x10];
+	u8         offset[0x10];
+
+	u8         gre_key_h[0x18];
+	u8         gre_key_l[0x8];
+
+	u8         seq_num[0x20];
+};
+
+struct mlx5_ifc_ste_flex_parser_0_bits {
+	u8         flex_parser_3[0x20];
+
+	u8         flex_parser_2[0x20];
+
+	u8         flex_parser_1[0x20];
+
+	u8         flex_parser_0[0x20];
+};
+
+struct mlx5_ifc_ste_flex_parser_1_bits {
+	u8         flex_parser_7[0x20];
+
+	u8         flex_parser_6[0x20];
+
+	u8         flex_parser_5[0x20];
+
+	u8         flex_parser_4[0x20];
+};
+
+struct mlx5_ifc_ste_flex_parser_ok_bits {
+	u8         flex_parser_3[0x20];
+	u8         flex_parser_2[0x20];
+	u8         flex_parsers_ok[0x8];
+	u8         reserved_at_48[0x18];
+	u8         flex_parser_0[0x20];
+};
+
+struct mlx5_ifc_ste_flex_parser_tnl_bits {
+	u8         flex_parser_tunneling_header_63_32[0x20];
+
+	u8         flex_parser_tunneling_header_31_0[0x20];
+
+	u8         reserved_at_40[0x40];
+};
+
+struct mlx5_ifc_ste_flex_parser_tnl_vxlan_gpe_bits {
+	u8         outer_vxlan_gpe_flags[0x8];
+	u8         reserved_at_8[0x10];
+	u8         outer_vxlan_gpe_next_protocol[0x8];
+
+	u8         outer_vxlan_gpe_vni[0x18];
+	u8         reserved_at_38[0x8];
+
+	u8         reserved_at_40[0x40];
+};
+
+struct mlx5_ifc_ste_flex_parser_tnl_geneve_bits {
+	u8         reserved_at_0[0x2];
+	u8         geneve_opt_len[0x6];
+	u8         geneve_oam[0x1];
+	u8         reserved_at_9[0x7];
+	u8         geneve_protocol_type[0x10];
+
+	u8         geneve_vni[0x18];
+	u8         reserved_at_38[0x8];
+
+	u8         reserved_at_40[0x40];
+};
+
+struct mlx5_ifc_ste_flex_parser_tnl_gtpu_bits {
+	u8	   reserved_at_0[0x5];
+	u8	   gtpu_msg_flags[0x3];
+	u8	   gtpu_msg_type[0x8];
+	u8	   reserved_at_10[0x10];
+
+	u8	   gtpu_teid[0x20];
+
+	u8	   reserved_at_40[0x40];
+};
+
+struct mlx5_ifc_ste_tunnel_header_bits {
+	u8	   tunnel_header_0[0x20];
+
+	u8	   tunnel_header_1[0x20];
+
+	u8	   reserved_at_40[0x40];
+};
+
+struct mlx5_ifc_ste_general_purpose_bits {
+	u8         general_purpose_lookup_field[0x20];
+
+	u8         reserved_at_20[0x20];
+
+	u8         reserved_at_40[0x20];
+
+	u8         reserved_at_60[0x20];
+};
+
+struct mlx5_ifc_ste_src_gvmi_qp_bits {
+	u8         loopback_syndrome[0x8];
+	u8         reserved_at_8[0x8];
+	u8         source_gvmi[0x10];
+
+	u8         reserved_at_20[0x5];
+	u8         force_lb[0x1];
+	u8         functional_lb[0x1];
+	u8         source_is_requestor[0x1];
+	u8         source_qp[0x18];
+
+	u8         reserved_at_40[0x20];
+
+	u8         reserved_at_60[0x20];
+};
+
+struct mlx5_ifc_l2_hdr_bits {
+	u8         dmac_47_16[0x20];
+
+	u8         dmac_15_0[0x10];
+	u8         smac_47_32[0x10];
+
+	u8         smac_31_0[0x20];
+
+	u8         ethertype[0x10];
+	u8         vlan_type[0x10];
+
+	u8         vlan[0x10];
+	u8         reserved_at_90[0x10];
+};
+
+/* Both HW set and HW add share the same HW format with different opcodes */
+struct mlx5_ifc_dr_action_hw_set_bits {
+	u8         opcode[0x8];
+	u8         destination_field_code[0x8];
+	u8         reserved_at_10[0x2];
+	u8         destination_left_shifter[0x6];
+	u8         reserved_at_18[0x3];
+	u8         destination_length[0x5];
+
+	u8         inline_data[0x20];
+};
+
+struct mlx5_ifc_dr_action_hw_copy_bits {
+	u8         opcode[0x8];
+	u8         destination_field_code[0x8];
+	u8         reserved_at_10[0x2];
+	u8         destination_left_shifter[0x6];
+	u8         reserved_at_18[0x2];
+	u8         destination_length[0x6];
+
+	u8         reserved_at_20[0x8];
+	u8         source_field_code[0x8];
+	u8         reserved_at_30[0x2];
+	u8         source_left_shifter[0x6];
+	u8         reserved_at_38[0x8];
+};
+
+enum {
+	MLX5DR_ASO_FLOW_METER_NUM_PER_OBJ = 2,
+};
+
+struct mlx5_ifc_ste_aso_flow_meter_action_bits {
+	u8         reserved_at_0[0xc];
+	u8         action[0x1];
+	u8         initial_color[0x2];
+	u8         line_id[0x1];
+};
+
+struct mlx5_ifc_ste_double_action_aso_v1_bits {
+	u8         action_id[0x8];
+	u8         aso_context_number[0x18];
+
+	u8         dest_reg_id[0x2];
+	u8         change_ordering_tag[0x1];
+	u8         aso_check_ordering[0x1];
+	u8         aso_context_type[0x4];
+	u8         reserved_at_28[0x8];
+	union {
+		u8 aso_fields[0x10];
+		struct mlx5_ifc_ste_aso_flow_meter_action_bits flow_meter;
+	};
+};
+
+#endif /* MLX5_IFC_DR_H */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/mlx5_ifc_dr_ste_v1.h b/drivers/net/ethernet/mellanox/mlx5/core/steering/mlx5_ifc_dr_ste_v1.h
new file mode 100644
index 000000000..34c2bd17a
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/mlx5_ifc_dr_ste_v1.h
@@ -0,0 +1,434 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2020 NVIDIA CORPORATION. All rights reserved. */
+
+#ifndef MLX5_IFC_DR_STE_V1_H
+#define MLX5_IFC_DR_STE_V1_H
+
+enum mlx5_ifc_ste_v1_modify_hdr_offset {
+	MLX5_MODIFY_HEADER_V1_QW_OFFSET = 0x20,
+};
+
+struct mlx5_ifc_ste_single_action_flow_tag_v1_bits {
+	u8         action_id[0x8];
+	u8         flow_tag[0x18];
+};
+
+struct mlx5_ifc_ste_single_action_modify_list_v1_bits {
+	u8         action_id[0x8];
+	u8         num_of_modify_actions[0x8];
+	u8         modify_actions_ptr[0x10];
+};
+
+struct mlx5_ifc_ste_single_action_remove_header_v1_bits {
+	u8         action_id[0x8];
+	u8         reserved_at_8[0x2];
+	u8         start_anchor[0x6];
+	u8         reserved_at_10[0x2];
+	u8         end_anchor[0x6];
+	u8         reserved_at_18[0x4];
+	u8         decap[0x1];
+	u8         vni_to_cqe[0x1];
+	u8         qos_profile[0x2];
+};
+
+struct mlx5_ifc_ste_single_action_remove_header_size_v1_bits {
+	u8         action_id[0x8];
+	u8         reserved_at_8[0x2];
+	u8         start_anchor[0x6];
+	u8         outer_l4_remove[0x1];
+	u8         reserved_at_11[0x1];
+	u8         start_offset[0x7];
+	u8         reserved_at_18[0x1];
+	u8         remove_size[0x6];
+};
+
+struct mlx5_ifc_ste_double_action_copy_v1_bits {
+	u8         action_id[0x8];
+	u8         destination_dw_offset[0x8];
+	u8         reserved_at_10[0x2];
+	u8         destination_left_shifter[0x6];
+	u8         reserved_at_17[0x2];
+	u8         destination_length[0x6];
+
+	u8         reserved_at_20[0x8];
+	u8         source_dw_offset[0x8];
+	u8         reserved_at_30[0x2];
+	u8         source_right_shifter[0x6];
+	u8         reserved_at_38[0x8];
+};
+
+struct mlx5_ifc_ste_double_action_set_v1_bits {
+	u8         action_id[0x8];
+	u8         destination_dw_offset[0x8];
+	u8         reserved_at_10[0x2];
+	u8         destination_left_shifter[0x6];
+	u8         reserved_at_18[0x2];
+	u8         destination_length[0x6];
+
+	u8         inline_data[0x20];
+};
+
+struct mlx5_ifc_ste_double_action_add_v1_bits {
+	u8         action_id[0x8];
+	u8         destination_dw_offset[0x8];
+	u8         reserved_at_10[0x2];
+	u8         destination_left_shifter[0x6];
+	u8         reserved_at_18[0x2];
+	u8         destination_length[0x6];
+
+	u8         add_value[0x20];
+};
+
+struct mlx5_ifc_ste_double_action_insert_with_inline_v1_bits {
+	u8         action_id[0x8];
+	u8         reserved_at_8[0x2];
+	u8         start_anchor[0x6];
+	u8         start_offset[0x7];
+	u8         reserved_at_17[0x9];
+
+	u8         inline_data[0x20];
+};
+
+struct mlx5_ifc_ste_double_action_insert_with_ptr_v1_bits {
+	u8         action_id[0x8];
+	u8         reserved_at_8[0x2];
+	u8         start_anchor[0x6];
+	u8         start_offset[0x7];
+	u8         size[0x6];
+	u8         attributes[0x3];
+
+	u8         pointer[0x20];
+};
+
+struct mlx5_ifc_ste_double_action_modify_action_list_v1_bits {
+	u8         action_id[0x8];
+	u8         modify_actions_pattern_pointer[0x18];
+
+	u8         number_of_modify_actions[0x8];
+	u8         modify_actions_argument_pointer[0x18];
+};
+
+struct mlx5_ifc_ste_match_bwc_v1_bits {
+	u8         entry_format[0x8];
+	u8         counter_id[0x18];
+
+	u8         miss_address_63_48[0x10];
+	u8         match_definer_ctx_idx[0x8];
+	u8         miss_address_39_32[0x8];
+
+	u8         miss_address_31_6[0x1a];
+	u8         reserved_at_5a[0x1];
+	u8         match_polarity[0x1];
+	u8         reparse[0x1];
+	u8         reserved_at_5d[0x3];
+
+	u8         next_table_base_63_48[0x10];
+	u8         hash_definer_ctx_idx[0x8];
+	u8         next_table_base_39_32_size[0x8];
+
+	u8         next_table_base_31_5_size[0x1b];
+	u8         hash_type[0x2];
+	u8         hash_after_actions[0x1];
+	u8         reserved_at_9e[0x2];
+
+	u8         byte_mask[0x10];
+	u8         next_entry_format[0x1];
+	u8         mask_mode[0x1];
+	u8         gvmi[0xe];
+
+	u8         action[0x40];
+};
+
+struct mlx5_ifc_ste_mask_and_match_v1_bits {
+	u8         entry_format[0x8];
+	u8         counter_id[0x18];
+
+	u8         miss_address_63_48[0x10];
+	u8         match_definer_ctx_idx[0x8];
+	u8         miss_address_39_32[0x8];
+
+	u8         miss_address_31_6[0x1a];
+	u8         reserved_at_5a[0x1];
+	u8         match_polarity[0x1];
+	u8         reparse[0x1];
+	u8         reserved_at_5d[0x3];
+
+	u8         next_table_base_63_48[0x10];
+	u8         hash_definer_ctx_idx[0x8];
+	u8         next_table_base_39_32_size[0x8];
+
+	u8         next_table_base_31_5_size[0x1b];
+	u8         hash_type[0x2];
+	u8         hash_after_actions[0x1];
+	u8         reserved_at_9e[0x2];
+
+	u8         action[0x60];
+};
+
+struct mlx5_ifc_ste_eth_l2_src_v1_bits {
+	u8         reserved_at_0[0x1];
+	u8         sx_sniffer[0x1];
+	u8         functional_loopback[0x1];
+	u8         ip_fragmented[0x1];
+	u8         qp_type[0x2];
+	u8         encapsulation_type[0x2];
+	u8         port[0x2];
+	u8         l3_type[0x2];
+	u8         l4_type[0x2];
+	u8         first_vlan_qualifier[0x2];
+	u8         first_priority[0x3];
+	u8         first_cfi[0x1];
+	u8         first_vlan_id[0xc];
+
+	u8         smac_47_16[0x20];
+
+	u8         smac_15_0[0x10];
+	u8         l3_ethertype[0x10];
+
+	u8         reserved_at_60[0x6];
+	u8         tcp_syn[0x1];
+	u8         reserved_at_67[0x3];
+	u8         force_loopback[0x1];
+	u8         l2_ok[0x1];
+	u8         l3_ok[0x1];
+	u8         l4_ok[0x1];
+	u8         second_vlan_qualifier[0x2];
+
+	u8         second_priority[0x3];
+	u8         second_cfi[0x1];
+	u8         second_vlan_id[0xc];
+};
+
+struct mlx5_ifc_ste_eth_l2_dst_v1_bits {
+	u8         reserved_at_0[0x1];
+	u8         sx_sniffer[0x1];
+	u8         functional_lb[0x1];
+	u8         ip_fragmented[0x1];
+	u8         qp_type[0x2];
+	u8         encapsulation_type[0x2];
+	u8         port[0x2];
+	u8         l3_type[0x2];
+	u8         l4_type[0x2];
+	u8         first_vlan_qualifier[0x2];
+	u8         first_priority[0x3];
+	u8         first_cfi[0x1];
+	u8         first_vlan_id[0xc];
+
+	u8         dmac_47_16[0x20];
+
+	u8         dmac_15_0[0x10];
+	u8         l3_ethertype[0x10];
+
+	u8         reserved_at_60[0x6];
+	u8         tcp_syn[0x1];
+	u8         reserved_at_67[0x3];
+	u8         force_lb[0x1];
+	u8         l2_ok[0x1];
+	u8         l3_ok[0x1];
+	u8         l4_ok[0x1];
+	u8         second_vlan_qualifier[0x2];
+	u8         second_priority[0x3];
+	u8         second_cfi[0x1];
+	u8         second_vlan_id[0xc];
+};
+
+struct mlx5_ifc_ste_eth_l2_src_dst_v1_bits {
+	u8         dmac_47_16[0x20];
+
+	u8         smac_47_16[0x20];
+
+	u8         dmac_15_0[0x10];
+	u8         reserved_at_50[0x2];
+	u8         functional_lb[0x1];
+	u8         reserved_at_53[0x5];
+	u8         port[0x2];
+	u8         l3_type[0x2];
+	u8         reserved_at_5c[0x2];
+	u8         first_vlan_qualifier[0x2];
+
+	u8         first_priority[0x3];
+	u8         first_cfi[0x1];
+	u8         first_vlan_id[0xc];
+	u8         smac_15_0[0x10];
+};
+
+struct mlx5_ifc_ste_eth_l3_ipv4_5_tuple_v1_bits {
+	u8         source_address[0x20];
+
+	u8         destination_address[0x20];
+
+	u8         source_port[0x10];
+	u8         destination_port[0x10];
+
+	u8         reserved_at_60[0x4];
+	u8         l4_ok[0x1];
+	u8         l3_ok[0x1];
+	u8         fragmented[0x1];
+	u8         tcp_ns[0x1];
+	u8         tcp_cwr[0x1];
+	u8         tcp_ece[0x1];
+	u8         tcp_urg[0x1];
+	u8         tcp_ack[0x1];
+	u8         tcp_psh[0x1];
+	u8         tcp_rst[0x1];
+	u8         tcp_syn[0x1];
+	u8         tcp_fin[0x1];
+	u8         dscp[0x6];
+	u8         ecn[0x2];
+	u8         protocol[0x8];
+};
+
+struct mlx5_ifc_ste_eth_l2_tnl_v1_bits {
+	u8         l2_tunneling_network_id[0x20];
+
+	u8         dmac_47_16[0x20];
+
+	u8         dmac_15_0[0x10];
+	u8         l3_ethertype[0x10];
+
+	u8         reserved_at_60[0x3];
+	u8         ip_fragmented[0x1];
+	u8         reserved_at_64[0x2];
+	u8         encp_type[0x2];
+	u8         reserved_at_68[0x2];
+	u8         l3_type[0x2];
+	u8         l4_type[0x2];
+	u8         first_vlan_qualifier[0x2];
+	u8         first_priority[0x3];
+	u8         first_cfi[0x1];
+	u8         first_vlan_id[0xc];
+};
+
+struct mlx5_ifc_ste_eth_l3_ipv4_misc_v1_bits {
+	u8         identification[0x10];
+	u8         flags[0x3];
+	u8         fragment_offset[0xd];
+
+	u8         total_length[0x10];
+	u8         checksum[0x10];
+
+	u8         version[0x4];
+	u8         ihl[0x4];
+	u8         time_to_live[0x8];
+	u8         reserved_at_50[0x10];
+
+	u8         reserved_at_60[0x1c];
+	u8         voq_internal_prio[0x4];
+};
+
+struct mlx5_ifc_ste_eth_l4_v1_bits {
+	u8         ipv6_version[0x4];
+	u8         reserved_at_4[0x4];
+	u8         dscp[0x6];
+	u8         ecn[0x2];
+	u8         ipv6_hop_limit[0x8];
+	u8         protocol[0x8];
+
+	u8         src_port[0x10];
+	u8         dst_port[0x10];
+
+	u8         first_fragment[0x1];
+	u8         reserved_at_41[0xb];
+	u8         flow_label[0x14];
+
+	u8         tcp_data_offset[0x4];
+	u8         l4_ok[0x1];
+	u8         l3_ok[0x1];
+	u8         fragmented[0x1];
+	u8         tcp_ns[0x1];
+	u8         tcp_cwr[0x1];
+	u8         tcp_ece[0x1];
+	u8         tcp_urg[0x1];
+	u8         tcp_ack[0x1];
+	u8         tcp_psh[0x1];
+	u8         tcp_rst[0x1];
+	u8         tcp_syn[0x1];
+	u8         tcp_fin[0x1];
+	u8         ipv6_paylen[0x10];
+};
+
+struct mlx5_ifc_ste_eth_l4_misc_v1_bits {
+	u8         window_size[0x10];
+	u8         urgent_pointer[0x10];
+
+	u8         ack_num[0x20];
+
+	u8         seq_num[0x20];
+
+	u8         length[0x10];
+	u8         checksum[0x10];
+};
+
+struct mlx5_ifc_ste_mpls_v1_bits {
+	u8         reserved_at_0[0x15];
+	u8         mpls_ok[0x1];
+	u8         mpls4_s_bit[0x1];
+	u8         mpls4_qualifier[0x1];
+	u8         mpls3_s_bit[0x1];
+	u8         mpls3_qualifier[0x1];
+	u8         mpls2_s_bit[0x1];
+	u8         mpls2_qualifier[0x1];
+	u8         mpls1_s_bit[0x1];
+	u8         mpls1_qualifier[0x1];
+	u8         mpls0_s_bit[0x1];
+	u8         mpls0_qualifier[0x1];
+
+	u8         mpls0_label[0x14];
+	u8         mpls0_exp[0x3];
+	u8         mpls0_s_bos[0x1];
+	u8         mpls0_ttl[0x8];
+
+	u8         mpls1_label[0x20];
+
+	u8         mpls2_label[0x20];
+};
+
+struct mlx5_ifc_ste_gre_v1_bits {
+	u8         gre_c_present[0x1];
+	u8         reserved_at_1[0x1];
+	u8         gre_k_present[0x1];
+	u8         gre_s_present[0x1];
+	u8         strict_src_route[0x1];
+	u8         recur[0x3];
+	u8         flags[0x5];
+	u8         version[0x3];
+	u8         gre_protocol[0x10];
+
+	u8         reserved_at_20[0x20];
+
+	u8         gre_key_h[0x18];
+	u8         gre_key_l[0x8];
+
+	u8         reserved_at_60[0x20];
+};
+
+struct mlx5_ifc_ste_src_gvmi_qp_v1_bits {
+	u8         loopback_synd[0x8];
+	u8         reserved_at_8[0x7];
+	u8         functional_lb[0x1];
+	u8         source_gvmi[0x10];
+
+	u8         force_lb[0x1];
+	u8         reserved_at_21[0x1];
+	u8         source_is_requestor[0x1];
+	u8         reserved_at_23[0x5];
+	u8         source_qp[0x18];
+
+	u8         reserved_at_40[0x20];
+
+	u8         reserved_at_60[0x20];
+};
+
+struct mlx5_ifc_ste_icmp_v1_bits {
+	u8         icmp_payload_data[0x20];
+
+	u8         icmp_header_data[0x20];
+
+	u8         icmp_type[0x8];
+	u8         icmp_code[0x8];
+	u8         reserved_at_50[0x10];
+
+	u8         reserved_at_60[0x20];
+};
+
+#endif /* MLX5_IFC_DR_STE_V1_H */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/mlx5dr.h b/drivers/net/ethernet/mellanox/mlx5/core/steering/mlx5dr.h
new file mode 100644
index 000000000..226a0d7bb
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/mlx5dr.h
@@ -0,0 +1,192 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2019, Mellanox Technologies */
+
+#ifndef _MLX5DR_H_
+#define _MLX5DR_H_
+
+struct mlx5dr_domain;
+struct mlx5dr_table;
+struct mlx5dr_matcher;
+struct mlx5dr_rule;
+struct mlx5dr_action;
+
+enum mlx5dr_domain_type {
+	MLX5DR_DOMAIN_TYPE_NIC_RX,
+	MLX5DR_DOMAIN_TYPE_NIC_TX,
+	MLX5DR_DOMAIN_TYPE_FDB,
+};
+
+enum mlx5dr_domain_sync_flags {
+	MLX5DR_DOMAIN_SYNC_FLAGS_SW = 1 << 0,
+	MLX5DR_DOMAIN_SYNC_FLAGS_HW = 1 << 1,
+};
+
+enum mlx5dr_action_reformat_type {
+	DR_ACTION_REFORMAT_TYP_TNL_L2_TO_L2,
+	DR_ACTION_REFORMAT_TYP_L2_TO_TNL_L2,
+	DR_ACTION_REFORMAT_TYP_TNL_L3_TO_L2,
+	DR_ACTION_REFORMAT_TYP_L2_TO_TNL_L3,
+	DR_ACTION_REFORMAT_TYP_INSERT_HDR,
+	DR_ACTION_REFORMAT_TYP_REMOVE_HDR,
+};
+
+struct mlx5dr_match_parameters {
+	size_t match_sz;
+	u64 *match_buf; /* Device spec format */
+};
+
+struct mlx5dr_action_dest {
+	struct mlx5dr_action *dest;
+	struct mlx5dr_action *reformat;
+};
+
+struct mlx5dr_domain *
+mlx5dr_domain_create(struct mlx5_core_dev *mdev, enum mlx5dr_domain_type type);
+
+int mlx5dr_domain_destroy(struct mlx5dr_domain *domain);
+
+int mlx5dr_domain_sync(struct mlx5dr_domain *domain, u32 flags);
+
+void mlx5dr_domain_set_peer(struct mlx5dr_domain *dmn,
+			    struct mlx5dr_domain *peer_dmn);
+
+struct mlx5dr_table *
+mlx5dr_table_create(struct mlx5dr_domain *domain, u32 level, u32 flags,
+		    u16 uid);
+
+struct mlx5dr_table *
+mlx5dr_table_get_from_fs_ft(struct mlx5_flow_table *ft);
+
+int mlx5dr_table_destroy(struct mlx5dr_table *table);
+
+u32 mlx5dr_table_get_id(struct mlx5dr_table *table);
+
+struct mlx5dr_matcher *
+mlx5dr_matcher_create(struct mlx5dr_table *table,
+		      u32 priority,
+		      u8 match_criteria_enable,
+		      struct mlx5dr_match_parameters *mask);
+
+int mlx5dr_matcher_destroy(struct mlx5dr_matcher *matcher);
+
+struct mlx5dr_rule *
+mlx5dr_rule_create(struct mlx5dr_matcher *matcher,
+		   struct mlx5dr_match_parameters *value,
+		   size_t num_actions,
+		   struct mlx5dr_action *actions[],
+		   u32 flow_source);
+
+int mlx5dr_rule_destroy(struct mlx5dr_rule *rule);
+
+int mlx5dr_table_set_miss_action(struct mlx5dr_table *tbl,
+				 struct mlx5dr_action *action);
+
+struct mlx5dr_action *
+mlx5dr_action_create_dest_table_num(struct mlx5dr_domain *dmn, u32 table_num);
+
+struct mlx5dr_action *
+mlx5dr_action_create_dest_table(struct mlx5dr_table *table);
+
+struct mlx5dr_action *
+mlx5dr_action_create_dest_flow_fw_table(struct mlx5dr_domain *domain,
+					struct mlx5_flow_table *ft);
+
+struct mlx5dr_action *
+mlx5dr_action_create_dest_vport(struct mlx5dr_domain *domain,
+				u16 vport, u8 vhca_id_valid,
+				u16 vhca_id);
+
+struct mlx5dr_action *
+mlx5dr_action_create_mult_dest_tbl(struct mlx5dr_domain *dmn,
+				   struct mlx5dr_action_dest *dests,
+				   u32 num_of_dests,
+				   bool ignore_flow_level,
+				   u32 flow_source);
+
+struct mlx5dr_action *mlx5dr_action_create_drop(void);
+
+struct mlx5dr_action *mlx5dr_action_create_tag(u32 tag_value);
+
+struct mlx5dr_action *
+mlx5dr_action_create_flow_sampler(struct mlx5dr_domain *dmn, u32 sampler_id);
+
+struct mlx5dr_action *
+mlx5dr_action_create_flow_counter(u32 counter_id);
+
+struct mlx5dr_action *
+mlx5dr_action_create_packet_reformat(struct mlx5dr_domain *dmn,
+				     enum mlx5dr_action_reformat_type reformat_type,
+				     u8 reformat_param_0,
+				     u8 reformat_param_1,
+				     size_t data_sz,
+				     void *data);
+
+struct mlx5dr_action *
+mlx5dr_action_create_modify_header(struct mlx5dr_domain *domain,
+				   u32 flags,
+				   size_t actions_sz,
+				   __be64 actions[]);
+
+struct mlx5dr_action *mlx5dr_action_create_pop_vlan(void);
+
+struct mlx5dr_action *
+mlx5dr_action_create_push_vlan(struct mlx5dr_domain *domain, __be32 vlan_hdr);
+
+struct mlx5dr_action *
+mlx5dr_action_create_aso(struct mlx5dr_domain *dmn,
+			 u32 obj_id,
+			 u8 return_reg_id,
+			 u8 aso_type,
+			 u8 init_color,
+			 u8 meter_id);
+
+int mlx5dr_action_destroy(struct mlx5dr_action *action);
+
+static inline bool
+mlx5dr_is_supported(struct mlx5_core_dev *dev)
+{
+	return MLX5_CAP_GEN(dev, roce) &&
+	       (MLX5_CAP_ESW_FLOWTABLE_FDB(dev, sw_owner) ||
+		(MLX5_CAP_ESW_FLOWTABLE_FDB(dev, sw_owner_v2) &&
+		 (MLX5_CAP_GEN(dev, steering_format_version) <=
+		  MLX5_STEERING_FORMAT_CONNECTX_7)));
+}
+
+/* buddy functions & structure */
+
+struct mlx5dr_icm_mr;
+
+struct mlx5dr_icm_buddy_mem {
+	unsigned long		**bitmap;
+	unsigned int		*num_free;
+	u32			max_order;
+	struct list_head	list_node;
+	struct mlx5dr_icm_mr	*icm_mr;
+	struct mlx5dr_icm_pool	*pool;
+
+	/* This is the list of used chunks. HW may be accessing this memory */
+	struct list_head	used_list;
+	u64			used_memory;
+
+	/* Hardware may be accessing this memory but at some future,
+	 * undetermined time, it might cease to do so.
+	 * sync_ste command sets them free.
+	 */
+	struct list_head	hot_list;
+
+	/* Memory optimisation */
+	struct mlx5dr_ste	*ste_arr;
+	struct list_head	*miss_list;
+	u8			*hw_ste_arr;
+};
+
+int mlx5dr_buddy_init(struct mlx5dr_icm_buddy_mem *buddy,
+		      unsigned int max_order);
+void mlx5dr_buddy_cleanup(struct mlx5dr_icm_buddy_mem *buddy);
+int mlx5dr_buddy_alloc_mem(struct mlx5dr_icm_buddy_mem *buddy,
+			   unsigned int order,
+			   unsigned int *segment);
+void mlx5dr_buddy_free_mem(struct mlx5dr_icm_buddy_mem *buddy,
+			   unsigned int seg, unsigned int order);
+
+#endif /* _MLX5DR_H_ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/transobj.c b/drivers/net/ethernet/mellanox/mlx5/core/transobj.c
new file mode 100644
index 000000000..b6931bbe5
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/transobj.c
@@ -0,0 +1,505 @@
+/*
+ * Copyright (c) 2013-2015, Mellanox Technologies, Ltd.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/mlx5/driver.h>
+#include "mlx5_core.h"
+#include <linux/mlx5/transobj.h>
+
+int mlx5_core_alloc_transport_domain(struct mlx5_core_dev *dev, u32 *tdn)
+{
+	u32 out[MLX5_ST_SZ_DW(alloc_transport_domain_out)] = {};
+	u32 in[MLX5_ST_SZ_DW(alloc_transport_domain_in)] = {};
+	int err;
+
+	MLX5_SET(alloc_transport_domain_in, in, opcode,
+		 MLX5_CMD_OP_ALLOC_TRANSPORT_DOMAIN);
+
+	err = mlx5_cmd_exec_inout(dev, alloc_transport_domain, in, out);
+	if (!err)
+		*tdn = MLX5_GET(alloc_transport_domain_out, out,
+				transport_domain);
+
+	return err;
+}
+EXPORT_SYMBOL(mlx5_core_alloc_transport_domain);
+
+void mlx5_core_dealloc_transport_domain(struct mlx5_core_dev *dev, u32 tdn)
+{
+	u32 in[MLX5_ST_SZ_DW(dealloc_transport_domain_in)] = {};
+
+	MLX5_SET(dealloc_transport_domain_in, in, opcode,
+		 MLX5_CMD_OP_DEALLOC_TRANSPORT_DOMAIN);
+	MLX5_SET(dealloc_transport_domain_in, in, transport_domain, tdn);
+	mlx5_cmd_exec_in(dev, dealloc_transport_domain, in);
+}
+EXPORT_SYMBOL(mlx5_core_dealloc_transport_domain);
+
+int mlx5_core_create_rq(struct mlx5_core_dev *dev, u32 *in, int inlen, u32 *rqn)
+{
+	u32 out[MLX5_ST_SZ_DW(create_rq_out)] = {};
+	int err;
+
+	MLX5_SET(create_rq_in, in, opcode, MLX5_CMD_OP_CREATE_RQ);
+	err = mlx5_cmd_exec(dev, in, inlen, out, sizeof(out));
+	if (!err)
+		*rqn = MLX5_GET(create_rq_out, out, rqn);
+
+	return err;
+}
+EXPORT_SYMBOL(mlx5_core_create_rq);
+
+int mlx5_core_modify_rq(struct mlx5_core_dev *dev, u32 rqn, u32 *in)
+{
+	MLX5_SET(modify_rq_in, in, rqn, rqn);
+	MLX5_SET(modify_rq_in, in, opcode, MLX5_CMD_OP_MODIFY_RQ);
+
+	return mlx5_cmd_exec_in(dev, modify_rq, in);
+}
+EXPORT_SYMBOL(mlx5_core_modify_rq);
+
+void mlx5_core_destroy_rq(struct mlx5_core_dev *dev, u32 rqn)
+{
+	u32 in[MLX5_ST_SZ_DW(destroy_rq_in)] = {};
+
+	MLX5_SET(destroy_rq_in, in, opcode, MLX5_CMD_OP_DESTROY_RQ);
+	MLX5_SET(destroy_rq_in, in, rqn, rqn);
+	mlx5_cmd_exec_in(dev, destroy_rq, in);
+}
+EXPORT_SYMBOL(mlx5_core_destroy_rq);
+
+int mlx5_core_query_rq(struct mlx5_core_dev *dev, u32 rqn, u32 *out)
+{
+	u32 in[MLX5_ST_SZ_DW(query_rq_in)] = {};
+
+	MLX5_SET(query_rq_in, in, opcode, MLX5_CMD_OP_QUERY_RQ);
+	MLX5_SET(query_rq_in, in, rqn, rqn);
+
+	return mlx5_cmd_exec_inout(dev, query_rq, in, out);
+}
+EXPORT_SYMBOL(mlx5_core_query_rq);
+
+int mlx5_core_create_sq(struct mlx5_core_dev *dev, u32 *in, int inlen, u32 *sqn)
+{
+	u32 out[MLX5_ST_SZ_DW(create_sq_out)] = {};
+	int err;
+
+	MLX5_SET(create_sq_in, in, opcode, MLX5_CMD_OP_CREATE_SQ);
+	err = mlx5_cmd_exec(dev, in, inlen, out, sizeof(out));
+	if (!err)
+		*sqn = MLX5_GET(create_sq_out, out, sqn);
+
+	return err;
+}
+
+int mlx5_core_modify_sq(struct mlx5_core_dev *dev, u32 sqn, u32 *in)
+{
+	MLX5_SET(modify_sq_in, in, sqn, sqn);
+	MLX5_SET(modify_sq_in, in, opcode, MLX5_CMD_OP_MODIFY_SQ);
+	return mlx5_cmd_exec_in(dev, modify_sq, in);
+}
+EXPORT_SYMBOL(mlx5_core_modify_sq);
+
+void mlx5_core_destroy_sq(struct mlx5_core_dev *dev, u32 sqn)
+{
+	u32 in[MLX5_ST_SZ_DW(destroy_sq_in)] = {};
+
+	MLX5_SET(destroy_sq_in, in, opcode, MLX5_CMD_OP_DESTROY_SQ);
+	MLX5_SET(destroy_sq_in, in, sqn, sqn);
+	mlx5_cmd_exec_in(dev, destroy_sq, in);
+}
+
+int mlx5_core_query_sq(struct mlx5_core_dev *dev, u32 sqn, u32 *out)
+{
+	u32 in[MLX5_ST_SZ_DW(query_sq_in)] = {};
+
+	MLX5_SET(query_sq_in, in, opcode, MLX5_CMD_OP_QUERY_SQ);
+	MLX5_SET(query_sq_in, in, sqn, sqn);
+	return mlx5_cmd_exec_inout(dev, query_sq, in, out);
+}
+EXPORT_SYMBOL(mlx5_core_query_sq);
+
+int mlx5_core_query_sq_state(struct mlx5_core_dev *dev, u32 sqn, u8 *state)
+{
+	void *out;
+	void *sqc;
+	int inlen;
+	int err;
+
+	inlen = MLX5_ST_SZ_BYTES(query_sq_out);
+	out = kvzalloc(inlen, GFP_KERNEL);
+	if (!out)
+		return -ENOMEM;
+
+	err = mlx5_core_query_sq(dev, sqn, out);
+	if (err)
+		goto out;
+
+	sqc = MLX5_ADDR_OF(query_sq_out, out, sq_context);
+	*state = MLX5_GET(sqc, sqc, state);
+
+out:
+	kvfree(out);
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx5_core_query_sq_state);
+
+int mlx5_core_create_tir(struct mlx5_core_dev *dev, u32 *in, u32 *tirn)
+{
+	u32 out[MLX5_ST_SZ_DW(create_tir_out)] = {};
+	int err;
+
+	MLX5_SET(create_tir_in, in, opcode, MLX5_CMD_OP_CREATE_TIR);
+	err = mlx5_cmd_exec_inout(dev, create_tir, in, out);
+	if (!err)
+		*tirn = MLX5_GET(create_tir_out, out, tirn);
+
+	return err;
+}
+EXPORT_SYMBOL(mlx5_core_create_tir);
+
+int mlx5_core_modify_tir(struct mlx5_core_dev *dev, u32 tirn, u32 *in)
+{
+	MLX5_SET(modify_tir_in, in, tirn, tirn);
+	MLX5_SET(modify_tir_in, in, opcode, MLX5_CMD_OP_MODIFY_TIR);
+	return mlx5_cmd_exec_in(dev, modify_tir, in);
+}
+
+void mlx5_core_destroy_tir(struct mlx5_core_dev *dev, u32 tirn)
+{
+	u32 in[MLX5_ST_SZ_DW(destroy_tir_in)] = {};
+
+	MLX5_SET(destroy_tir_in, in, opcode, MLX5_CMD_OP_DESTROY_TIR);
+	MLX5_SET(destroy_tir_in, in, tirn, tirn);
+	mlx5_cmd_exec_in(dev, destroy_tir, in);
+}
+EXPORT_SYMBOL(mlx5_core_destroy_tir);
+
+int mlx5_core_create_tis(struct mlx5_core_dev *dev, u32 *in, u32 *tisn)
+{
+	u32 out[MLX5_ST_SZ_DW(create_tis_out)] = {};
+	int err;
+
+	MLX5_SET(create_tis_in, in, opcode, MLX5_CMD_OP_CREATE_TIS);
+	err = mlx5_cmd_exec_inout(dev, create_tis, in, out);
+	if (!err)
+		*tisn = MLX5_GET(create_tis_out, out, tisn);
+
+	return err;
+}
+EXPORT_SYMBOL(mlx5_core_create_tis);
+
+int mlx5_core_modify_tis(struct mlx5_core_dev *dev, u32 tisn, u32 *in)
+{
+	MLX5_SET(modify_tis_in, in, tisn, tisn);
+	MLX5_SET(modify_tis_in, in, opcode, MLX5_CMD_OP_MODIFY_TIS);
+
+	return mlx5_cmd_exec_in(dev, modify_tis, in);
+}
+EXPORT_SYMBOL(mlx5_core_modify_tis);
+
+void mlx5_core_destroy_tis(struct mlx5_core_dev *dev, u32 tisn)
+{
+	u32 in[MLX5_ST_SZ_DW(destroy_tis_in)] = {};
+
+	MLX5_SET(destroy_tis_in, in, opcode, MLX5_CMD_OP_DESTROY_TIS);
+	MLX5_SET(destroy_tis_in, in, tisn, tisn);
+	mlx5_cmd_exec_in(dev, destroy_tis, in);
+}
+EXPORT_SYMBOL(mlx5_core_destroy_tis);
+
+int mlx5_core_create_rqt(struct mlx5_core_dev *dev, u32 *in, int inlen,
+			 u32 *rqtn)
+{
+	u32 out[MLX5_ST_SZ_DW(create_rqt_out)] = {};
+	int err;
+
+	MLX5_SET(create_rqt_in, in, opcode, MLX5_CMD_OP_CREATE_RQT);
+	err = mlx5_cmd_exec(dev, in, inlen, out, sizeof(out));
+	if (!err)
+		*rqtn = MLX5_GET(create_rqt_out, out, rqtn);
+
+	return err;
+}
+EXPORT_SYMBOL(mlx5_core_create_rqt);
+
+int mlx5_core_modify_rqt(struct mlx5_core_dev *dev, u32 rqtn, u32 *in,
+			 int inlen)
+{
+	u32 out[MLX5_ST_SZ_DW(modify_rqt_out)] = {};
+
+	MLX5_SET(modify_rqt_in, in, rqtn, rqtn);
+	MLX5_SET(modify_rqt_in, in, opcode, MLX5_CMD_OP_MODIFY_RQT);
+	return mlx5_cmd_exec(dev, in, inlen, out, sizeof(out));
+}
+
+void mlx5_core_destroy_rqt(struct mlx5_core_dev *dev, u32 rqtn)
+{
+	u32 in[MLX5_ST_SZ_DW(destroy_rqt_in)] = {};
+
+	MLX5_SET(destroy_rqt_in, in, opcode, MLX5_CMD_OP_DESTROY_RQT);
+	MLX5_SET(destroy_rqt_in, in, rqtn, rqtn);
+	mlx5_cmd_exec_in(dev, destroy_rqt, in);
+}
+EXPORT_SYMBOL(mlx5_core_destroy_rqt);
+
+static int mlx5_hairpin_create_rq(struct mlx5_core_dev *mdev,
+				  struct mlx5_hairpin_params *params, u32 *rqn)
+{
+	u32 in[MLX5_ST_SZ_DW(create_rq_in)] = {0};
+	void *rqc, *wq;
+
+	rqc = MLX5_ADDR_OF(create_rq_in, in, ctx);
+	wq  = MLX5_ADDR_OF(rqc, rqc, wq);
+
+	MLX5_SET(rqc, rqc, hairpin, 1);
+	MLX5_SET(rqc, rqc, state, MLX5_RQC_STATE_RST);
+	MLX5_SET(rqc, rqc, counter_set_id, params->q_counter);
+
+	MLX5_SET(wq, wq, log_hairpin_data_sz, params->log_data_size);
+	MLX5_SET(wq, wq, log_hairpin_num_packets, params->log_num_packets);
+
+	return mlx5_core_create_rq(mdev, in, MLX5_ST_SZ_BYTES(create_rq_in), rqn);
+}
+
+static int mlx5_hairpin_create_sq(struct mlx5_core_dev *mdev,
+				  struct mlx5_hairpin_params *params, u32 *sqn)
+{
+	u32 in[MLX5_ST_SZ_DW(create_sq_in)] = {0};
+	void *sqc, *wq;
+
+	sqc = MLX5_ADDR_OF(create_sq_in, in, ctx);
+	wq  = MLX5_ADDR_OF(sqc, sqc, wq);
+
+	MLX5_SET(sqc, sqc, hairpin, 1);
+	MLX5_SET(sqc, sqc, state, MLX5_SQC_STATE_RST);
+
+	MLX5_SET(wq, wq, log_hairpin_data_sz, params->log_data_size);
+	MLX5_SET(wq, wq, log_hairpin_num_packets, params->log_num_packets);
+
+	return mlx5_core_create_sq(mdev, in, MLX5_ST_SZ_BYTES(create_sq_in), sqn);
+}
+
+static int mlx5_hairpin_create_queues(struct mlx5_hairpin *hp,
+				      struct mlx5_hairpin_params *params)
+{
+	int i, j, err;
+
+	for (i = 0; i < hp->num_channels; i++) {
+		err = mlx5_hairpin_create_rq(hp->func_mdev, params, &hp->rqn[i]);
+		if (err)
+			goto out_err_rq;
+	}
+
+	for (i = 0; i < hp->num_channels; i++) {
+		err = mlx5_hairpin_create_sq(hp->peer_mdev, params, &hp->sqn[i]);
+		if (err)
+			goto out_err_sq;
+	}
+
+	return 0;
+
+out_err_sq:
+	for (j = 0; j < i; j++)
+		mlx5_core_destroy_sq(hp->peer_mdev, hp->sqn[j]);
+	i = hp->num_channels;
+out_err_rq:
+	for (j = 0; j < i; j++)
+		mlx5_core_destroy_rq(hp->func_mdev, hp->rqn[j]);
+	return err;
+}
+
+static void mlx5_hairpin_destroy_queues(struct mlx5_hairpin *hp)
+{
+	int i;
+
+	for (i = 0; i < hp->num_channels; i++) {
+		mlx5_core_destroy_rq(hp->func_mdev, hp->rqn[i]);
+		if (!hp->peer_gone)
+			mlx5_core_destroy_sq(hp->peer_mdev, hp->sqn[i]);
+	}
+}
+
+static int mlx5_hairpin_modify_rq(struct mlx5_core_dev *func_mdev, u32 rqn,
+				  int curr_state, int next_state,
+				  u16 peer_vhca, u32 peer_sq)
+{
+	u32 in[MLX5_ST_SZ_DW(modify_rq_in)] = {};
+	void *rqc;
+
+	rqc = MLX5_ADDR_OF(modify_rq_in, in, ctx);
+
+	if (next_state == MLX5_RQC_STATE_RDY) {
+		MLX5_SET(rqc, rqc, hairpin_peer_sq, peer_sq);
+		MLX5_SET(rqc, rqc, hairpin_peer_vhca, peer_vhca);
+	}
+
+	MLX5_SET(modify_rq_in, in, rq_state, curr_state);
+	MLX5_SET(rqc, rqc, state, next_state);
+
+	return mlx5_core_modify_rq(func_mdev, rqn, in);
+}
+
+static int mlx5_hairpin_modify_sq(struct mlx5_core_dev *peer_mdev, u32 sqn,
+				  int curr_state, int next_state,
+				  u16 peer_vhca, u32 peer_rq)
+{
+	u32 in[MLX5_ST_SZ_DW(modify_sq_in)] = {0};
+	void *sqc;
+
+	sqc = MLX5_ADDR_OF(modify_sq_in, in, ctx);
+
+	if (next_state == MLX5_SQC_STATE_RDY) {
+		MLX5_SET(sqc, sqc, hairpin_peer_rq, peer_rq);
+		MLX5_SET(sqc, sqc, hairpin_peer_vhca, peer_vhca);
+	}
+
+	MLX5_SET(modify_sq_in, in, sq_state, curr_state);
+	MLX5_SET(sqc, sqc, state, next_state);
+
+	return mlx5_core_modify_sq(peer_mdev, sqn, in);
+}
+
+static int mlx5_hairpin_pair_queues(struct mlx5_hairpin *hp)
+{
+	int i, j, err;
+
+	/* set peer SQs */
+	for (i = 0; i < hp->num_channels; i++) {
+		err = mlx5_hairpin_modify_sq(hp->peer_mdev, hp->sqn[i],
+					     MLX5_SQC_STATE_RST, MLX5_SQC_STATE_RDY,
+					     MLX5_CAP_GEN(hp->func_mdev, vhca_id), hp->rqn[i]);
+		if (err)
+			goto err_modify_sq;
+	}
+
+	/* set func RQs */
+	for (i = 0; i < hp->num_channels; i++) {
+		err = mlx5_hairpin_modify_rq(hp->func_mdev, hp->rqn[i],
+					     MLX5_RQC_STATE_RST, MLX5_RQC_STATE_RDY,
+					     MLX5_CAP_GEN(hp->peer_mdev, vhca_id), hp->sqn[i]);
+		if (err)
+			goto err_modify_rq;
+	}
+
+	return 0;
+
+err_modify_rq:
+	for (j = 0; j < i; j++)
+		mlx5_hairpin_modify_rq(hp->func_mdev, hp->rqn[j], MLX5_RQC_STATE_RDY,
+				       MLX5_RQC_STATE_RST, 0, 0);
+	i = hp->num_channels;
+err_modify_sq:
+	for (j = 0; j < i; j++)
+		mlx5_hairpin_modify_sq(hp->peer_mdev, hp->sqn[j], MLX5_SQC_STATE_RDY,
+				       MLX5_SQC_STATE_RST, 0, 0);
+	return err;
+}
+
+static void mlx5_hairpin_unpair_peer_sq(struct mlx5_hairpin *hp)
+{
+	int i;
+
+	for (i = 0; i < hp->num_channels; i++)
+		mlx5_hairpin_modify_sq(hp->peer_mdev, hp->sqn[i], MLX5_SQC_STATE_RDY,
+				       MLX5_SQC_STATE_RST, 0, 0);
+}
+
+static void mlx5_hairpin_unpair_queues(struct mlx5_hairpin *hp)
+{
+	int i;
+
+	/* unset func RQs */
+	for (i = 0; i < hp->num_channels; i++)
+		mlx5_hairpin_modify_rq(hp->func_mdev, hp->rqn[i], MLX5_RQC_STATE_RDY,
+				       MLX5_RQC_STATE_RST, 0, 0);
+	/* unset peer SQs */
+	if (!hp->peer_gone)
+		mlx5_hairpin_unpair_peer_sq(hp);
+}
+
+struct mlx5_hairpin *
+mlx5_core_hairpin_create(struct mlx5_core_dev *func_mdev,
+			 struct mlx5_core_dev *peer_mdev,
+			 struct mlx5_hairpin_params *params)
+{
+	struct mlx5_hairpin *hp;
+	int size, err;
+
+	size = sizeof(*hp) + params->num_channels * 2 * sizeof(u32);
+	hp = kzalloc(size, GFP_KERNEL);
+	if (!hp)
+		return ERR_PTR(-ENOMEM);
+
+	hp->func_mdev = func_mdev;
+	hp->peer_mdev = peer_mdev;
+	hp->num_channels = params->num_channels;
+
+	hp->rqn = (void *)hp + sizeof(*hp);
+	hp->sqn = hp->rqn + params->num_channels;
+
+	/* alloc and pair func --> peer hairpin */
+	err = mlx5_hairpin_create_queues(hp, params);
+	if (err)
+		goto err_create_queues;
+
+	err = mlx5_hairpin_pair_queues(hp);
+	if (err)
+		goto err_pair_queues;
+
+	return hp;
+
+err_pair_queues:
+	mlx5_hairpin_destroy_queues(hp);
+err_create_queues:
+	kfree(hp);
+	return ERR_PTR(err);
+}
+
+void mlx5_core_hairpin_destroy(struct mlx5_hairpin *hp)
+{
+	mlx5_hairpin_unpair_queues(hp);
+	mlx5_hairpin_destroy_queues(hp);
+	kfree(hp);
+}
+
+void mlx5_core_hairpin_clear_dead_peer(struct mlx5_hairpin *hp)
+{
+	int i;
+
+	mlx5_hairpin_unpair_peer_sq(hp);
+
+	/* destroy peer SQ */
+	for (i = 0; i < hp->num_channels; i++)
+		mlx5_core_destroy_sq(hp->peer_mdev, hp->sqn[i]);
+
+	hp->peer_gone = true;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/uar.c b/drivers/net/ethernet/mellanox/mlx5/core/uar.c
new file mode 100644
index 000000000..8455e79bc
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/uar.c
@@ -0,0 +1,328 @@
+/*
+ * Copyright (c) 2013-2015, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/kernel.h>
+#include <linux/io-mapping.h>
+#include <linux/mlx5/driver.h>
+#include "mlx5_core.h"
+
+static int mlx5_cmd_alloc_uar(struct mlx5_core_dev *dev, u32 *uarn)
+{
+	u32 out[MLX5_ST_SZ_DW(alloc_uar_out)] = {};
+	u32 in[MLX5_ST_SZ_DW(alloc_uar_in)] = {};
+	int err;
+
+	MLX5_SET(alloc_uar_in, in, opcode, MLX5_CMD_OP_ALLOC_UAR);
+	err = mlx5_cmd_exec_inout(dev, alloc_uar, in, out);
+	if (err)
+		return err;
+
+	*uarn = MLX5_GET(alloc_uar_out, out, uar);
+	return 0;
+}
+
+static int mlx5_cmd_free_uar(struct mlx5_core_dev *dev, u32 uarn)
+{
+	u32 in[MLX5_ST_SZ_DW(dealloc_uar_in)] = {};
+
+	MLX5_SET(dealloc_uar_in, in, opcode, MLX5_CMD_OP_DEALLOC_UAR);
+	MLX5_SET(dealloc_uar_in, in, uar, uarn);
+	return mlx5_cmd_exec_in(dev, dealloc_uar, in);
+}
+
+static int uars_per_sys_page(struct mlx5_core_dev *mdev)
+{
+	if (MLX5_CAP_GEN(mdev, uar_4k))
+		return MLX5_CAP_GEN(mdev, num_of_uars_per_page);
+
+	return 1;
+}
+
+static u64 uar2pfn(struct mlx5_core_dev *mdev, u32 index)
+{
+	u32 system_page_index;
+
+	if (MLX5_CAP_GEN(mdev, uar_4k))
+		system_page_index = index >> (PAGE_SHIFT - MLX5_ADAPTER_PAGE_SHIFT);
+	else
+		system_page_index = index;
+
+	return (mdev->bar_addr >> PAGE_SHIFT) + system_page_index;
+}
+
+static void up_rel_func(struct kref *kref)
+{
+	struct mlx5_uars_page *up = container_of(kref, struct mlx5_uars_page, ref_count);
+
+	list_del(&up->list);
+	iounmap(up->map);
+	if (mlx5_cmd_free_uar(up->mdev, up->index))
+		mlx5_core_warn(up->mdev, "failed to free uar index %d\n", up->index);
+	bitmap_free(up->reg_bitmap);
+	bitmap_free(up->fp_bitmap);
+	kfree(up);
+}
+
+static struct mlx5_uars_page *alloc_uars_page(struct mlx5_core_dev *mdev,
+					      bool map_wc)
+{
+	struct mlx5_uars_page *up;
+	int err = -ENOMEM;
+	phys_addr_t pfn;
+	int bfregs;
+	int node;
+	int i;
+
+	bfregs = uars_per_sys_page(mdev) * MLX5_BFREGS_PER_UAR;
+	node = mdev->priv.numa_node;
+	up = kzalloc_node(sizeof(*up), GFP_KERNEL, node);
+	if (!up)
+		return ERR_PTR(err);
+
+	up->mdev = mdev;
+	up->reg_bitmap = bitmap_zalloc_node(bfregs, GFP_KERNEL, node);
+	if (!up->reg_bitmap)
+		goto error1;
+
+	up->fp_bitmap = bitmap_zalloc_node(bfregs, GFP_KERNEL, node);
+	if (!up->fp_bitmap)
+		goto error1;
+
+	for (i = 0; i < bfregs; i++)
+		if ((i % MLX5_BFREGS_PER_UAR) < MLX5_NON_FP_BFREGS_PER_UAR)
+			set_bit(i, up->reg_bitmap);
+		else
+			set_bit(i, up->fp_bitmap);
+
+	up->bfregs = bfregs;
+	up->fp_avail = bfregs * MLX5_FP_BFREGS_PER_UAR / MLX5_BFREGS_PER_UAR;
+	up->reg_avail = bfregs * MLX5_NON_FP_BFREGS_PER_UAR / MLX5_BFREGS_PER_UAR;
+
+	err = mlx5_cmd_alloc_uar(mdev, &up->index);
+	if (err) {
+		mlx5_core_warn(mdev, "mlx5_cmd_alloc_uar() failed, %d\n", err);
+		goto error1;
+	}
+
+	pfn = uar2pfn(mdev, up->index);
+	if (map_wc) {
+		up->map = ioremap_wc(pfn << PAGE_SHIFT, PAGE_SIZE);
+		if (!up->map) {
+			err = -EAGAIN;
+			goto error2;
+		}
+	} else {
+		up->map = ioremap(pfn << PAGE_SHIFT, PAGE_SIZE);
+		if (!up->map) {
+			err = -ENOMEM;
+			goto error2;
+		}
+	}
+	kref_init(&up->ref_count);
+	mlx5_core_dbg(mdev, "allocated UAR page: index %d, total bfregs %d\n",
+		      up->index, up->bfregs);
+	return up;
+
+error2:
+	if (mlx5_cmd_free_uar(mdev, up->index))
+		mlx5_core_warn(mdev, "failed to free uar index %d\n", up->index);
+error1:
+	bitmap_free(up->fp_bitmap);
+	bitmap_free(up->reg_bitmap);
+	kfree(up);
+	return ERR_PTR(err);
+}
+
+struct mlx5_uars_page *mlx5_get_uars_page(struct mlx5_core_dev *mdev)
+{
+	struct mlx5_uars_page *ret;
+
+	mutex_lock(&mdev->priv.bfregs.reg_head.lock);
+	if (!list_empty(&mdev->priv.bfregs.reg_head.list)) {
+		ret = list_first_entry(&mdev->priv.bfregs.reg_head.list,
+				       struct mlx5_uars_page, list);
+		kref_get(&ret->ref_count);
+		goto out;
+	}
+	ret = alloc_uars_page(mdev, false);
+	if (IS_ERR(ret))
+		goto out;
+	list_add(&ret->list, &mdev->priv.bfregs.reg_head.list);
+out:
+	mutex_unlock(&mdev->priv.bfregs.reg_head.lock);
+
+	return ret;
+}
+EXPORT_SYMBOL(mlx5_get_uars_page);
+
+void mlx5_put_uars_page(struct mlx5_core_dev *mdev, struct mlx5_uars_page *up)
+{
+	mutex_lock(&mdev->priv.bfregs.reg_head.lock);
+	kref_put(&up->ref_count, up_rel_func);
+	mutex_unlock(&mdev->priv.bfregs.reg_head.lock);
+}
+EXPORT_SYMBOL(mlx5_put_uars_page);
+
+static unsigned long map_offset(struct mlx5_core_dev *mdev, int dbi)
+{
+	/* return the offset in bytes from the start of the page to the
+	 * blue flame area of the UAR
+	 */
+	return dbi / MLX5_BFREGS_PER_UAR * MLX5_ADAPTER_PAGE_SIZE +
+	       (dbi % MLX5_BFREGS_PER_UAR) *
+	       (1 << MLX5_CAP_GEN(mdev, log_bf_reg_size)) + MLX5_BF_OFFSET;
+}
+
+static int alloc_bfreg(struct mlx5_core_dev *mdev, struct mlx5_sq_bfreg *bfreg,
+		       bool map_wc, bool fast_path)
+{
+	struct mlx5_bfreg_data *bfregs;
+	struct mlx5_uars_page *up;
+	struct list_head *head;
+	unsigned long *bitmap;
+	unsigned int *avail;
+	struct mutex *lock;  /* pointer to right mutex */
+	int dbi;
+
+	bfregs = &mdev->priv.bfregs;
+	if (map_wc) {
+		head = &bfregs->wc_head.list;
+		lock = &bfregs->wc_head.lock;
+	} else {
+		head = &bfregs->reg_head.list;
+		lock = &bfregs->reg_head.lock;
+	}
+	mutex_lock(lock);
+	if (list_empty(head)) {
+		up = alloc_uars_page(mdev, map_wc);
+		if (IS_ERR(up)) {
+			mutex_unlock(lock);
+			return PTR_ERR(up);
+		}
+		list_add(&up->list, head);
+	} else {
+		up = list_entry(head->next, struct mlx5_uars_page, list);
+		kref_get(&up->ref_count);
+	}
+	if (fast_path) {
+		bitmap = up->fp_bitmap;
+		avail = &up->fp_avail;
+	} else {
+		bitmap = up->reg_bitmap;
+		avail = &up->reg_avail;
+	}
+	dbi = find_first_bit(bitmap, up->bfregs);
+	clear_bit(dbi, bitmap);
+	(*avail)--;
+	if (!(*avail))
+		list_del(&up->list);
+
+	bfreg->map = up->map + map_offset(mdev, dbi);
+	bfreg->up = up;
+	bfreg->wc = map_wc;
+	bfreg->index = up->index + dbi / MLX5_BFREGS_PER_UAR;
+	mutex_unlock(lock);
+
+	return 0;
+}
+
+int mlx5_alloc_bfreg(struct mlx5_core_dev *mdev, struct mlx5_sq_bfreg *bfreg,
+		     bool map_wc, bool fast_path)
+{
+	int err;
+
+	err = alloc_bfreg(mdev, bfreg, map_wc, fast_path);
+	if (!err)
+		return 0;
+
+	if (err == -EAGAIN && map_wc)
+		return alloc_bfreg(mdev, bfreg, false, fast_path);
+
+	return err;
+}
+EXPORT_SYMBOL(mlx5_alloc_bfreg);
+
+static unsigned int addr_to_dbi_in_syspage(struct mlx5_core_dev *dev,
+					   struct mlx5_uars_page *up,
+					   struct mlx5_sq_bfreg *bfreg)
+{
+	unsigned int uar_idx;
+	unsigned int bfreg_idx;
+	unsigned int bf_reg_size;
+
+	bf_reg_size = 1 << MLX5_CAP_GEN(dev, log_bf_reg_size);
+
+	uar_idx = (bfreg->map - up->map) >> MLX5_ADAPTER_PAGE_SHIFT;
+	bfreg_idx = (((uintptr_t)bfreg->map % MLX5_ADAPTER_PAGE_SIZE) - MLX5_BF_OFFSET) / bf_reg_size;
+
+	return uar_idx * MLX5_BFREGS_PER_UAR + bfreg_idx;
+}
+
+void mlx5_free_bfreg(struct mlx5_core_dev *mdev, struct mlx5_sq_bfreg *bfreg)
+{
+	struct mlx5_bfreg_data *bfregs;
+	struct mlx5_uars_page *up;
+	struct mutex *lock; /* pointer to right mutex */
+	unsigned int dbi;
+	bool fp;
+	unsigned int *avail;
+	unsigned long *bitmap;
+	struct list_head *head;
+
+	bfregs = &mdev->priv.bfregs;
+	if (bfreg->wc) {
+		head = &bfregs->wc_head.list;
+		lock = &bfregs->wc_head.lock;
+	} else {
+		head = &bfregs->reg_head.list;
+		lock = &bfregs->reg_head.lock;
+	}
+	up = bfreg->up;
+	dbi = addr_to_dbi_in_syspage(mdev, up, bfreg);
+	fp = (dbi % MLX5_BFREGS_PER_UAR) >= MLX5_NON_FP_BFREGS_PER_UAR;
+	if (fp) {
+		avail = &up->fp_avail;
+		bitmap = up->fp_bitmap;
+	} else {
+		avail = &up->reg_avail;
+		bitmap = up->reg_bitmap;
+	}
+	mutex_lock(lock);
+	(*avail)++;
+	set_bit(dbi, bitmap);
+	if (*avail == 1)
+		list_add_tail(&up->list, head);
+
+	kref_put(&up->ref_count, up_rel_func);
+	mutex_unlock(lock);
+}
+EXPORT_SYMBOL(mlx5_free_bfreg);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/vport.c b/drivers/net/ethernet/mellanox/mlx5/core/vport.c
new file mode 100644
index 000000000..3f68e3198
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/vport.c
@@ -0,0 +1,1173 @@
+/*
+ * Copyright (c) 2013-2015, Mellanox Technologies, Ltd.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/export.h>
+#include <linux/etherdevice.h>
+#include <linux/mlx5/driver.h>
+#include <linux/mlx5/vport.h>
+#include <linux/mlx5/eswitch.h>
+#include "mlx5_core.h"
+#include "sf/sf.h"
+
+/* Mutex to hold while enabling or disabling RoCE */
+static DEFINE_MUTEX(mlx5_roce_en_lock);
+
+u8 mlx5_query_vport_state(struct mlx5_core_dev *mdev, u8 opmod, u16 vport)
+{
+	u32 out[MLX5_ST_SZ_DW(query_vport_state_out)] = {};
+	u32 in[MLX5_ST_SZ_DW(query_vport_state_in)] = {};
+	int err;
+
+	MLX5_SET(query_vport_state_in, in, opcode,
+		 MLX5_CMD_OP_QUERY_VPORT_STATE);
+	MLX5_SET(query_vport_state_in, in, op_mod, opmod);
+	MLX5_SET(query_vport_state_in, in, vport_number, vport);
+	if (vport)
+		MLX5_SET(query_vport_state_in, in, other_vport, 1);
+
+	err = mlx5_cmd_exec_inout(mdev, query_vport_state, in, out);
+	if (err)
+		return 0;
+
+	return MLX5_GET(query_vport_state_out, out, state);
+}
+
+int mlx5_modify_vport_admin_state(struct mlx5_core_dev *mdev, u8 opmod,
+				  u16 vport, u8 other_vport, u8 state)
+{
+	u32 in[MLX5_ST_SZ_DW(modify_vport_state_in)] = {};
+
+	MLX5_SET(modify_vport_state_in, in, opcode,
+		 MLX5_CMD_OP_MODIFY_VPORT_STATE);
+	MLX5_SET(modify_vport_state_in, in, op_mod, opmod);
+	MLX5_SET(modify_vport_state_in, in, vport_number, vport);
+	MLX5_SET(modify_vport_state_in, in, other_vport, other_vport);
+	MLX5_SET(modify_vport_state_in, in, admin_state, state);
+
+	return mlx5_cmd_exec_in(mdev, modify_vport_state, in);
+}
+
+static int mlx5_query_nic_vport_context(struct mlx5_core_dev *mdev, u16 vport,
+					u32 *out)
+{
+	u32 in[MLX5_ST_SZ_DW(query_nic_vport_context_in)] = {};
+
+	MLX5_SET(query_nic_vport_context_in, in, opcode,
+		 MLX5_CMD_OP_QUERY_NIC_VPORT_CONTEXT);
+	MLX5_SET(query_nic_vport_context_in, in, vport_number, vport);
+	if (vport)
+		MLX5_SET(query_nic_vport_context_in, in, other_vport, 1);
+
+	return mlx5_cmd_exec_inout(mdev, query_nic_vport_context, in, out);
+}
+
+int mlx5_query_nic_vport_min_inline(struct mlx5_core_dev *mdev,
+				    u16 vport, u8 *min_inline)
+{
+	u32 out[MLX5_ST_SZ_DW(query_nic_vport_context_out)] = {};
+	int err;
+
+	err = mlx5_query_nic_vport_context(mdev, vport, out);
+	if (!err)
+		*min_inline = MLX5_GET(query_nic_vport_context_out, out,
+				       nic_vport_context.min_wqe_inline_mode);
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx5_query_nic_vport_min_inline);
+
+void mlx5_query_min_inline(struct mlx5_core_dev *mdev,
+			   u8 *min_inline_mode)
+{
+	switch (MLX5_CAP_ETH(mdev, wqe_inline_mode)) {
+	case MLX5_CAP_INLINE_MODE_VPORT_CONTEXT:
+		if (!mlx5_query_nic_vport_min_inline(mdev, 0, min_inline_mode))
+			break;
+		fallthrough;
+	case MLX5_CAP_INLINE_MODE_L2:
+		*min_inline_mode = MLX5_INLINE_MODE_L2;
+		break;
+	case MLX5_CAP_INLINE_MODE_NOT_REQUIRED:
+		*min_inline_mode = MLX5_INLINE_MODE_NONE;
+		break;
+	}
+}
+EXPORT_SYMBOL_GPL(mlx5_query_min_inline);
+
+int mlx5_modify_nic_vport_min_inline(struct mlx5_core_dev *mdev,
+				     u16 vport, u8 min_inline)
+{
+	u32 in[MLX5_ST_SZ_DW(modify_nic_vport_context_in)] = {};
+	void *nic_vport_ctx;
+
+	MLX5_SET(modify_nic_vport_context_in, in,
+		 field_select.min_inline, 1);
+	MLX5_SET(modify_nic_vport_context_in, in, vport_number, vport);
+	MLX5_SET(modify_nic_vport_context_in, in, other_vport, 1);
+
+	nic_vport_ctx = MLX5_ADDR_OF(modify_nic_vport_context_in,
+				     in, nic_vport_context);
+	MLX5_SET(nic_vport_context, nic_vport_ctx,
+		 min_wqe_inline_mode, min_inline);
+	MLX5_SET(modify_nic_vport_context_in, in, opcode,
+		 MLX5_CMD_OP_MODIFY_NIC_VPORT_CONTEXT);
+
+	return mlx5_cmd_exec_in(mdev, modify_nic_vport_context, in);
+}
+
+int mlx5_query_nic_vport_mac_address(struct mlx5_core_dev *mdev,
+				     u16 vport, bool other, u8 *addr)
+{
+	u32 out[MLX5_ST_SZ_DW(query_nic_vport_context_out)] = {};
+	u32 in[MLX5_ST_SZ_DW(query_nic_vport_context_in)] = {};
+	u8 *out_addr;
+	int err;
+
+	out_addr = MLX5_ADDR_OF(query_nic_vport_context_out, out,
+				nic_vport_context.permanent_address);
+
+	MLX5_SET(query_nic_vport_context_in, in, opcode,
+		 MLX5_CMD_OP_QUERY_NIC_VPORT_CONTEXT);
+	MLX5_SET(query_nic_vport_context_in, in, vport_number, vport);
+	MLX5_SET(query_nic_vport_context_in, in, other_vport, other);
+
+	err = mlx5_cmd_exec_inout(mdev, query_nic_vport_context, in, out);
+	if (!err)
+		ether_addr_copy(addr, &out_addr[2]);
+
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx5_query_nic_vport_mac_address);
+
+int mlx5_query_mac_address(struct mlx5_core_dev *mdev, u8 *addr)
+{
+	return mlx5_query_nic_vport_mac_address(mdev, 0, false, addr);
+}
+EXPORT_SYMBOL_GPL(mlx5_query_mac_address);
+
+int mlx5_modify_nic_vport_mac_address(struct mlx5_core_dev *mdev,
+				      u16 vport, const u8 *addr)
+{
+	void *in;
+	int inlen = MLX5_ST_SZ_BYTES(modify_nic_vport_context_in);
+	int err;
+	void *nic_vport_ctx;
+	u8 *perm_mac;
+
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in)
+		return -ENOMEM;
+
+	MLX5_SET(modify_nic_vport_context_in, in,
+		 field_select.permanent_address, 1);
+	MLX5_SET(modify_nic_vport_context_in, in, vport_number, vport);
+	MLX5_SET(modify_nic_vport_context_in, in, other_vport, 1);
+
+	nic_vport_ctx = MLX5_ADDR_OF(modify_nic_vport_context_in,
+				     in, nic_vport_context);
+	perm_mac = MLX5_ADDR_OF(nic_vport_context, nic_vport_ctx,
+				permanent_address);
+
+	ether_addr_copy(&perm_mac[2], addr);
+	MLX5_SET(modify_nic_vport_context_in, in, opcode,
+		 MLX5_CMD_OP_MODIFY_NIC_VPORT_CONTEXT);
+
+	err = mlx5_cmd_exec_in(mdev, modify_nic_vport_context, in);
+
+	kvfree(in);
+
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx5_modify_nic_vport_mac_address);
+
+int mlx5_query_nic_vport_mtu(struct mlx5_core_dev *mdev, u16 *mtu)
+{
+	int outlen = MLX5_ST_SZ_BYTES(query_nic_vport_context_out);
+	u32 *out;
+	int err;
+
+	out = kvzalloc(outlen, GFP_KERNEL);
+	if (!out)
+		return -ENOMEM;
+
+	err = mlx5_query_nic_vport_context(mdev, 0, out);
+	if (!err)
+		*mtu = MLX5_GET(query_nic_vport_context_out, out,
+				nic_vport_context.mtu);
+
+	kvfree(out);
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx5_query_nic_vport_mtu);
+
+int mlx5_modify_nic_vport_mtu(struct mlx5_core_dev *mdev, u16 mtu)
+{
+	int inlen = MLX5_ST_SZ_BYTES(modify_nic_vport_context_in);
+	void *in;
+	int err;
+
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in)
+		return -ENOMEM;
+
+	MLX5_SET(modify_nic_vport_context_in, in, field_select.mtu, 1);
+	MLX5_SET(modify_nic_vport_context_in, in, nic_vport_context.mtu, mtu);
+	MLX5_SET(modify_nic_vport_context_in, in, opcode,
+		 MLX5_CMD_OP_MODIFY_NIC_VPORT_CONTEXT);
+
+	err = mlx5_cmd_exec_in(mdev, modify_nic_vport_context, in);
+
+	kvfree(in);
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx5_modify_nic_vport_mtu);
+
+int mlx5_query_nic_vport_mac_list(struct mlx5_core_dev *dev,
+				  u16 vport,
+				  enum mlx5_list_type list_type,
+				  u8 addr_list[][ETH_ALEN],
+				  int *list_size)
+{
+	u32 in[MLX5_ST_SZ_DW(query_nic_vport_context_in)] = {0};
+	void *nic_vport_ctx;
+	int max_list_size;
+	int req_list_size;
+	int out_sz;
+	void *out;
+	int err;
+	int i;
+
+	req_list_size = *list_size;
+
+	max_list_size = list_type == MLX5_NVPRT_LIST_TYPE_UC ?
+		1 << MLX5_CAP_GEN(dev, log_max_current_uc_list) :
+		1 << MLX5_CAP_GEN(dev, log_max_current_mc_list);
+
+	if (req_list_size > max_list_size) {
+		mlx5_core_warn(dev, "Requested list size (%d) > (%d) max_list_size\n",
+			       req_list_size, max_list_size);
+		req_list_size = max_list_size;
+	}
+
+	out_sz = MLX5_ST_SZ_BYTES(query_nic_vport_context_out) +
+			req_list_size * MLX5_ST_SZ_BYTES(mac_address_layout);
+
+	out = kvzalloc(out_sz, GFP_KERNEL);
+	if (!out)
+		return -ENOMEM;
+
+	MLX5_SET(query_nic_vport_context_in, in, opcode,
+		 MLX5_CMD_OP_QUERY_NIC_VPORT_CONTEXT);
+	MLX5_SET(query_nic_vport_context_in, in, allowed_list_type, list_type);
+	MLX5_SET(query_nic_vport_context_in, in, vport_number, vport);
+	MLX5_SET(query_nic_vport_context_in, in, other_vport, 1);
+
+	err = mlx5_cmd_exec(dev, in, sizeof(in), out, out_sz);
+	if (err)
+		goto out;
+
+	nic_vport_ctx = MLX5_ADDR_OF(query_nic_vport_context_out, out,
+				     nic_vport_context);
+	req_list_size = MLX5_GET(nic_vport_context, nic_vport_ctx,
+				 allowed_list_size);
+
+	*list_size = req_list_size;
+	for (i = 0; i < req_list_size; i++) {
+		u8 *mac_addr = MLX5_ADDR_OF(nic_vport_context,
+					nic_vport_ctx,
+					current_uc_mac_address[i]) + 2;
+		ether_addr_copy(addr_list[i], mac_addr);
+	}
+out:
+	kvfree(out);
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx5_query_nic_vport_mac_list);
+
+int mlx5_modify_nic_vport_mac_list(struct mlx5_core_dev *dev,
+				   enum mlx5_list_type list_type,
+				   u8 addr_list[][ETH_ALEN],
+				   int list_size)
+{
+	u32 out[MLX5_ST_SZ_DW(modify_nic_vport_context_out)] = {};
+	void *nic_vport_ctx;
+	int max_list_size;
+	int in_sz;
+	void *in;
+	int err;
+	int i;
+
+	max_list_size = list_type == MLX5_NVPRT_LIST_TYPE_UC ?
+		 1 << MLX5_CAP_GEN(dev, log_max_current_uc_list) :
+		 1 << MLX5_CAP_GEN(dev, log_max_current_mc_list);
+
+	if (list_size > max_list_size)
+		return -ENOSPC;
+
+	in_sz = MLX5_ST_SZ_BYTES(modify_nic_vport_context_in) +
+		list_size * MLX5_ST_SZ_BYTES(mac_address_layout);
+
+	in = kvzalloc(in_sz, GFP_KERNEL);
+	if (!in)
+		return -ENOMEM;
+
+	MLX5_SET(modify_nic_vport_context_in, in, opcode,
+		 MLX5_CMD_OP_MODIFY_NIC_VPORT_CONTEXT);
+	MLX5_SET(modify_nic_vport_context_in, in,
+		 field_select.addresses_list, 1);
+
+	nic_vport_ctx = MLX5_ADDR_OF(modify_nic_vport_context_in, in,
+				     nic_vport_context);
+
+	MLX5_SET(nic_vport_context, nic_vport_ctx,
+		 allowed_list_type, list_type);
+	MLX5_SET(nic_vport_context, nic_vport_ctx,
+		 allowed_list_size, list_size);
+
+	for (i = 0; i < list_size; i++) {
+		u8 *curr_mac = MLX5_ADDR_OF(nic_vport_context,
+					    nic_vport_ctx,
+					    current_uc_mac_address[i]) + 2;
+		ether_addr_copy(curr_mac, addr_list[i]);
+	}
+
+	err = mlx5_cmd_exec(dev, in, in_sz, out, sizeof(out));
+	kvfree(in);
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx5_modify_nic_vport_mac_list);
+
+int mlx5_modify_nic_vport_vlans(struct mlx5_core_dev *dev,
+				u16 vlans[],
+				int list_size)
+{
+	u32 out[MLX5_ST_SZ_DW(modify_nic_vport_context_out)];
+	void *nic_vport_ctx;
+	int max_list_size;
+	int in_sz;
+	void *in;
+	int err;
+	int i;
+
+	max_list_size = 1 << MLX5_CAP_GEN(dev, log_max_vlan_list);
+
+	if (list_size > max_list_size)
+		return -ENOSPC;
+
+	in_sz = MLX5_ST_SZ_BYTES(modify_nic_vport_context_in) +
+		list_size * MLX5_ST_SZ_BYTES(vlan_layout);
+
+	memset(out, 0, sizeof(out));
+	in = kvzalloc(in_sz, GFP_KERNEL);
+	if (!in)
+		return -ENOMEM;
+
+	MLX5_SET(modify_nic_vport_context_in, in, opcode,
+		 MLX5_CMD_OP_MODIFY_NIC_VPORT_CONTEXT);
+	MLX5_SET(modify_nic_vport_context_in, in,
+		 field_select.addresses_list, 1);
+
+	nic_vport_ctx = MLX5_ADDR_OF(modify_nic_vport_context_in, in,
+				     nic_vport_context);
+
+	MLX5_SET(nic_vport_context, nic_vport_ctx,
+		 allowed_list_type, MLX5_NVPRT_LIST_TYPE_VLAN);
+	MLX5_SET(nic_vport_context, nic_vport_ctx,
+		 allowed_list_size, list_size);
+
+	for (i = 0; i < list_size; i++) {
+		void *vlan_addr = MLX5_ADDR_OF(nic_vport_context,
+					       nic_vport_ctx,
+					       current_uc_mac_address[i]);
+		MLX5_SET(vlan_layout, vlan_addr, vlan, vlans[i]);
+	}
+
+	err = mlx5_cmd_exec(dev, in, in_sz, out, sizeof(out));
+	kvfree(in);
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx5_modify_nic_vport_vlans);
+
+int mlx5_query_nic_vport_system_image_guid(struct mlx5_core_dev *mdev,
+					   u64 *system_image_guid)
+{
+	u32 *out;
+	int outlen = MLX5_ST_SZ_BYTES(query_nic_vport_context_out);
+	int err;
+
+	out = kvzalloc(outlen, GFP_KERNEL);
+	if (!out)
+		return -ENOMEM;
+
+	err = mlx5_query_nic_vport_context(mdev, 0, out);
+	if (err)
+		goto out;
+
+	*system_image_guid = MLX5_GET64(query_nic_vport_context_out, out,
+					nic_vport_context.system_image_guid);
+out:
+	kvfree(out);
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx5_query_nic_vport_system_image_guid);
+
+int mlx5_query_nic_vport_node_guid(struct mlx5_core_dev *mdev, u64 *node_guid)
+{
+	u32 *out;
+	int outlen = MLX5_ST_SZ_BYTES(query_nic_vport_context_out);
+
+	out = kvzalloc(outlen, GFP_KERNEL);
+	if (!out)
+		return -ENOMEM;
+
+	mlx5_query_nic_vport_context(mdev, 0, out);
+
+	*node_guid = MLX5_GET64(query_nic_vport_context_out, out,
+				nic_vport_context.node_guid);
+
+	kvfree(out);
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(mlx5_query_nic_vport_node_guid);
+
+int mlx5_modify_nic_vport_node_guid(struct mlx5_core_dev *mdev,
+				    u16 vport, u64 node_guid)
+{
+	int inlen = MLX5_ST_SZ_BYTES(modify_nic_vport_context_in);
+	void *nic_vport_context;
+	void *in;
+	int err;
+
+	if (!MLX5_CAP_GEN(mdev, vport_group_manager))
+		return -EACCES;
+
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in)
+		return -ENOMEM;
+
+	MLX5_SET(modify_nic_vport_context_in, in,
+		 field_select.node_guid, 1);
+	MLX5_SET(modify_nic_vport_context_in, in, vport_number, vport);
+	MLX5_SET(modify_nic_vport_context_in, in, other_vport, 1);
+
+	nic_vport_context = MLX5_ADDR_OF(modify_nic_vport_context_in,
+					 in, nic_vport_context);
+	MLX5_SET64(nic_vport_context, nic_vport_context, node_guid, node_guid);
+	MLX5_SET(modify_nic_vport_context_in, in, opcode,
+		 MLX5_CMD_OP_MODIFY_NIC_VPORT_CONTEXT);
+
+	err = mlx5_cmd_exec_in(mdev, modify_nic_vport_context, in);
+
+	kvfree(in);
+
+	return err;
+}
+
+int mlx5_query_nic_vport_qkey_viol_cntr(struct mlx5_core_dev *mdev,
+					u16 *qkey_viol_cntr)
+{
+	u32 *out;
+	int outlen = MLX5_ST_SZ_BYTES(query_nic_vport_context_out);
+
+	out = kvzalloc(outlen, GFP_KERNEL);
+	if (!out)
+		return -ENOMEM;
+
+	mlx5_query_nic_vport_context(mdev, 0, out);
+
+	*qkey_viol_cntr = MLX5_GET(query_nic_vport_context_out, out,
+				   nic_vport_context.qkey_violation_counter);
+
+	kvfree(out);
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(mlx5_query_nic_vport_qkey_viol_cntr);
+
+int mlx5_query_hca_vport_gid(struct mlx5_core_dev *dev, u8 other_vport,
+			     u8 port_num, u16  vf_num, u16 gid_index,
+			     union ib_gid *gid)
+{
+	int in_sz = MLX5_ST_SZ_BYTES(query_hca_vport_gid_in);
+	int out_sz = MLX5_ST_SZ_BYTES(query_hca_vport_gid_out);
+	int is_group_manager;
+	void *out = NULL;
+	void *in = NULL;
+	union ib_gid *tmp;
+	int tbsz;
+	int nout;
+	int err;
+
+	is_group_manager = MLX5_CAP_GEN(dev, vport_group_manager);
+	tbsz = mlx5_get_gid_table_len(MLX5_CAP_GEN(dev, gid_table_size));
+	mlx5_core_dbg(dev, "vf_num %d, index %d, gid_table_size %d\n",
+		      vf_num, gid_index, tbsz);
+
+	if (gid_index > tbsz && gid_index != 0xffff)
+		return -EINVAL;
+
+	if (gid_index == 0xffff)
+		nout = tbsz;
+	else
+		nout = 1;
+
+	out_sz += nout * sizeof(*gid);
+
+	in = kvzalloc(in_sz, GFP_KERNEL);
+	out = kvzalloc(out_sz, GFP_KERNEL);
+	if (!in || !out) {
+		err = -ENOMEM;
+		goto out;
+	}
+
+	MLX5_SET(query_hca_vport_gid_in, in, opcode, MLX5_CMD_OP_QUERY_HCA_VPORT_GID);
+	if (other_vport) {
+		if (is_group_manager) {
+			MLX5_SET(query_hca_vport_gid_in, in, vport_number, vf_num);
+			MLX5_SET(query_hca_vport_gid_in, in, other_vport, 1);
+		} else {
+			err = -EPERM;
+			goto out;
+		}
+	}
+	MLX5_SET(query_hca_vport_gid_in, in, gid_index, gid_index);
+
+	if (MLX5_CAP_GEN(dev, num_ports) == 2)
+		MLX5_SET(query_hca_vport_gid_in, in, port_num, port_num);
+
+	err = mlx5_cmd_exec(dev, in, in_sz, out, out_sz);
+	if (err)
+		goto out;
+
+	tmp = out + MLX5_ST_SZ_BYTES(query_hca_vport_gid_out);
+	gid->global.subnet_prefix = tmp->global.subnet_prefix;
+	gid->global.interface_id = tmp->global.interface_id;
+
+out:
+	kvfree(in);
+	kvfree(out);
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx5_query_hca_vport_gid);
+
+int mlx5_query_hca_vport_pkey(struct mlx5_core_dev *dev, u8 other_vport,
+			      u8 port_num, u16 vf_num, u16 pkey_index,
+			      u16 *pkey)
+{
+	int in_sz = MLX5_ST_SZ_BYTES(query_hca_vport_pkey_in);
+	int out_sz = MLX5_ST_SZ_BYTES(query_hca_vport_pkey_out);
+	int is_group_manager;
+	void *out = NULL;
+	void *in = NULL;
+	void *pkarr;
+	int nout;
+	int tbsz;
+	int err;
+	int i;
+
+	is_group_manager = MLX5_CAP_GEN(dev, vport_group_manager);
+
+	tbsz = mlx5_to_sw_pkey_sz(MLX5_CAP_GEN(dev, pkey_table_size));
+	if (pkey_index > tbsz && pkey_index != 0xffff)
+		return -EINVAL;
+
+	if (pkey_index == 0xffff)
+		nout = tbsz;
+	else
+		nout = 1;
+
+	out_sz += nout * MLX5_ST_SZ_BYTES(pkey);
+
+	in = kvzalloc(in_sz, GFP_KERNEL);
+	out = kvzalloc(out_sz, GFP_KERNEL);
+	if (!in || !out) {
+		err = -ENOMEM;
+		goto out;
+	}
+
+	MLX5_SET(query_hca_vport_pkey_in, in, opcode, MLX5_CMD_OP_QUERY_HCA_VPORT_PKEY);
+	if (other_vport) {
+		if (is_group_manager) {
+			MLX5_SET(query_hca_vport_pkey_in, in, vport_number, vf_num);
+			MLX5_SET(query_hca_vport_pkey_in, in, other_vport, 1);
+		} else {
+			err = -EPERM;
+			goto out;
+		}
+	}
+	MLX5_SET(query_hca_vport_pkey_in, in, pkey_index, pkey_index);
+
+	if (MLX5_CAP_GEN(dev, num_ports) == 2)
+		MLX5_SET(query_hca_vport_pkey_in, in, port_num, port_num);
+
+	err = mlx5_cmd_exec(dev, in, in_sz, out, out_sz);
+	if (err)
+		goto out;
+
+	pkarr = MLX5_ADDR_OF(query_hca_vport_pkey_out, out, pkey);
+	for (i = 0; i < nout; i++, pkey++, pkarr += MLX5_ST_SZ_BYTES(pkey))
+		*pkey = MLX5_GET_PR(pkey, pkarr, pkey);
+
+out:
+	kvfree(in);
+	kvfree(out);
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx5_query_hca_vport_pkey);
+
+int mlx5_query_hca_vport_context(struct mlx5_core_dev *dev,
+				 u8 other_vport, u8 port_num,
+				 u16 vf_num,
+				 struct mlx5_hca_vport_context *rep)
+{
+	int out_sz = MLX5_ST_SZ_BYTES(query_hca_vport_context_out);
+	int in[MLX5_ST_SZ_DW(query_hca_vport_context_in)] = {};
+	int is_group_manager;
+	void *out;
+	void *ctx;
+	int err;
+
+	is_group_manager = MLX5_CAP_GEN(dev, vport_group_manager);
+
+	out = kvzalloc(out_sz, GFP_KERNEL);
+	if (!out)
+		return -ENOMEM;
+
+	MLX5_SET(query_hca_vport_context_in, in, opcode, MLX5_CMD_OP_QUERY_HCA_VPORT_CONTEXT);
+
+	if (other_vport) {
+		if (is_group_manager) {
+			MLX5_SET(query_hca_vport_context_in, in, other_vport, 1);
+			MLX5_SET(query_hca_vport_context_in, in, vport_number, vf_num);
+		} else {
+			err = -EPERM;
+			goto ex;
+		}
+	}
+
+	if (MLX5_CAP_GEN(dev, num_ports) == 2)
+		MLX5_SET(query_hca_vport_context_in, in, port_num, port_num);
+
+	err = mlx5_cmd_exec_inout(dev, query_hca_vport_context, in, out);
+	if (err)
+		goto ex;
+
+	ctx = MLX5_ADDR_OF(query_hca_vport_context_out, out, hca_vport_context);
+	rep->field_select = MLX5_GET_PR(hca_vport_context, ctx, field_select);
+	rep->sm_virt_aware = MLX5_GET_PR(hca_vport_context, ctx, sm_virt_aware);
+	rep->has_smi = MLX5_GET_PR(hca_vport_context, ctx, has_smi);
+	rep->has_raw = MLX5_GET_PR(hca_vport_context, ctx, has_raw);
+	rep->policy = MLX5_GET_PR(hca_vport_context, ctx, vport_state_policy);
+	rep->phys_state = MLX5_GET_PR(hca_vport_context, ctx,
+				      port_physical_state);
+	rep->vport_state = MLX5_GET_PR(hca_vport_context, ctx, vport_state);
+	rep->port_physical_state = MLX5_GET_PR(hca_vport_context, ctx,
+					       port_physical_state);
+	rep->port_guid = MLX5_GET64_PR(hca_vport_context, ctx, port_guid);
+	rep->node_guid = MLX5_GET64_PR(hca_vport_context, ctx, node_guid);
+	rep->cap_mask1 = MLX5_GET_PR(hca_vport_context, ctx, cap_mask1);
+	rep->cap_mask1_perm = MLX5_GET_PR(hca_vport_context, ctx,
+					  cap_mask1_field_select);
+	rep->cap_mask2 = MLX5_GET_PR(hca_vport_context, ctx, cap_mask2);
+	rep->cap_mask2_perm = MLX5_GET_PR(hca_vport_context, ctx,
+					  cap_mask2_field_select);
+	rep->lid = MLX5_GET_PR(hca_vport_context, ctx, lid);
+	rep->init_type_reply = MLX5_GET_PR(hca_vport_context, ctx,
+					   init_type_reply);
+	rep->lmc = MLX5_GET_PR(hca_vport_context, ctx, lmc);
+	rep->subnet_timeout = MLX5_GET_PR(hca_vport_context, ctx,
+					  subnet_timeout);
+	rep->sm_lid = MLX5_GET_PR(hca_vport_context, ctx, sm_lid);
+	rep->sm_sl = MLX5_GET_PR(hca_vport_context, ctx, sm_sl);
+	rep->qkey_violation_counter = MLX5_GET_PR(hca_vport_context, ctx,
+						  qkey_violation_counter);
+	rep->pkey_violation_counter = MLX5_GET_PR(hca_vport_context, ctx,
+						  pkey_violation_counter);
+	rep->grh_required = MLX5_GET_PR(hca_vport_context, ctx, grh_required);
+	rep->sys_image_guid = MLX5_GET64_PR(hca_vport_context, ctx,
+					    system_image_guid);
+
+ex:
+	kvfree(out);
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx5_query_hca_vport_context);
+
+int mlx5_query_hca_vport_system_image_guid(struct mlx5_core_dev *dev,
+					   u64 *sys_image_guid)
+{
+	struct mlx5_hca_vport_context *rep;
+	int err;
+
+	rep = kvzalloc(sizeof(*rep), GFP_KERNEL);
+	if (!rep)
+		return -ENOMEM;
+
+	err = mlx5_query_hca_vport_context(dev, 0, 1, 0, rep);
+	if (!err)
+		*sys_image_guid = rep->sys_image_guid;
+
+	kvfree(rep);
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx5_query_hca_vport_system_image_guid);
+
+int mlx5_query_hca_vport_node_guid(struct mlx5_core_dev *dev,
+				   u64 *node_guid)
+{
+	struct mlx5_hca_vport_context *rep;
+	int err;
+
+	rep = kvzalloc(sizeof(*rep), GFP_KERNEL);
+	if (!rep)
+		return -ENOMEM;
+
+	err = mlx5_query_hca_vport_context(dev, 0, 1, 0, rep);
+	if (!err)
+		*node_guid = rep->node_guid;
+
+	kvfree(rep);
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx5_query_hca_vport_node_guid);
+
+int mlx5_query_nic_vport_promisc(struct mlx5_core_dev *mdev,
+				 u16 vport,
+				 int *promisc_uc,
+				 int *promisc_mc,
+				 int *promisc_all)
+{
+	u32 *out;
+	int outlen = MLX5_ST_SZ_BYTES(query_nic_vport_context_out);
+	int err;
+
+	out = kvzalloc(outlen, GFP_KERNEL);
+	if (!out)
+		return -ENOMEM;
+
+	err = mlx5_query_nic_vport_context(mdev, vport, out);
+	if (err)
+		goto out;
+
+	*promisc_uc = MLX5_GET(query_nic_vport_context_out, out,
+			       nic_vport_context.promisc_uc);
+	*promisc_mc = MLX5_GET(query_nic_vport_context_out, out,
+			       nic_vport_context.promisc_mc);
+	*promisc_all = MLX5_GET(query_nic_vport_context_out, out,
+				nic_vport_context.promisc_all);
+
+out:
+	kvfree(out);
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx5_query_nic_vport_promisc);
+
+int mlx5_modify_nic_vport_promisc(struct mlx5_core_dev *mdev,
+				  int promisc_uc,
+				  int promisc_mc,
+				  int promisc_all)
+{
+	void *in;
+	int inlen = MLX5_ST_SZ_BYTES(modify_nic_vport_context_in);
+	int err;
+
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in)
+		return -ENOMEM;
+
+	MLX5_SET(modify_nic_vport_context_in, in, field_select.promisc, 1);
+	MLX5_SET(modify_nic_vport_context_in, in,
+		 nic_vport_context.promisc_uc, promisc_uc);
+	MLX5_SET(modify_nic_vport_context_in, in,
+		 nic_vport_context.promisc_mc, promisc_mc);
+	MLX5_SET(modify_nic_vport_context_in, in,
+		 nic_vport_context.promisc_all, promisc_all);
+	MLX5_SET(modify_nic_vport_context_in, in, opcode,
+		 MLX5_CMD_OP_MODIFY_NIC_VPORT_CONTEXT);
+
+	err = mlx5_cmd_exec_in(mdev, modify_nic_vport_context, in);
+
+	kvfree(in);
+
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx5_modify_nic_vport_promisc);
+
+enum {
+	UC_LOCAL_LB,
+	MC_LOCAL_LB
+};
+
+int mlx5_nic_vport_update_local_lb(struct mlx5_core_dev *mdev, bool enable)
+{
+	int inlen = MLX5_ST_SZ_BYTES(modify_nic_vport_context_in);
+	void *in;
+	int err;
+
+	if (!MLX5_CAP_GEN(mdev, disable_local_lb_mc) &&
+	    !MLX5_CAP_GEN(mdev, disable_local_lb_uc))
+		return 0;
+
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in)
+		return -ENOMEM;
+
+	MLX5_SET(modify_nic_vport_context_in, in,
+		 nic_vport_context.disable_mc_local_lb, !enable);
+	MLX5_SET(modify_nic_vport_context_in, in,
+		 nic_vport_context.disable_uc_local_lb, !enable);
+
+	if (MLX5_CAP_GEN(mdev, disable_local_lb_mc))
+		MLX5_SET(modify_nic_vport_context_in, in,
+			 field_select.disable_mc_local_lb, 1);
+
+	if (MLX5_CAP_GEN(mdev, disable_local_lb_uc))
+		MLX5_SET(modify_nic_vport_context_in, in,
+			 field_select.disable_uc_local_lb, 1);
+	MLX5_SET(modify_nic_vport_context_in, in, opcode,
+		 MLX5_CMD_OP_MODIFY_NIC_VPORT_CONTEXT);
+
+	err = mlx5_cmd_exec_in(mdev, modify_nic_vport_context, in);
+
+	if (!err)
+		mlx5_core_dbg(mdev, "%s local_lb\n",
+			      enable ? "enable" : "disable");
+
+	kvfree(in);
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx5_nic_vport_update_local_lb);
+
+int mlx5_nic_vport_query_local_lb(struct mlx5_core_dev *mdev, bool *status)
+{
+	int outlen = MLX5_ST_SZ_BYTES(query_nic_vport_context_out);
+	u32 *out;
+	int value;
+	int err;
+
+	out = kvzalloc(outlen, GFP_KERNEL);
+	if (!out)
+		return -ENOMEM;
+
+	err = mlx5_query_nic_vport_context(mdev, 0, out);
+	if (err)
+		goto out;
+
+	value = MLX5_GET(query_nic_vport_context_out, out,
+			 nic_vport_context.disable_mc_local_lb) << MC_LOCAL_LB;
+
+	value |= MLX5_GET(query_nic_vport_context_out, out,
+			  nic_vport_context.disable_uc_local_lb) << UC_LOCAL_LB;
+
+	*status = !value;
+
+out:
+	kvfree(out);
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx5_nic_vport_query_local_lb);
+
+enum mlx5_vport_roce_state {
+	MLX5_VPORT_ROCE_DISABLED = 0,
+	MLX5_VPORT_ROCE_ENABLED  = 1,
+};
+
+static int mlx5_nic_vport_update_roce_state(struct mlx5_core_dev *mdev,
+					    enum mlx5_vport_roce_state state)
+{
+	void *in;
+	int inlen = MLX5_ST_SZ_BYTES(modify_nic_vport_context_in);
+	int err;
+
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in)
+		return -ENOMEM;
+
+	MLX5_SET(modify_nic_vport_context_in, in, field_select.roce_en, 1);
+	MLX5_SET(modify_nic_vport_context_in, in, nic_vport_context.roce_en,
+		 state);
+	MLX5_SET(modify_nic_vport_context_in, in, opcode,
+		 MLX5_CMD_OP_MODIFY_NIC_VPORT_CONTEXT);
+
+	err = mlx5_cmd_exec_in(mdev, modify_nic_vport_context, in);
+
+	kvfree(in);
+
+	return err;
+}
+
+int mlx5_nic_vport_enable_roce(struct mlx5_core_dev *mdev)
+{
+	int err = 0;
+
+	mutex_lock(&mlx5_roce_en_lock);
+	if (!mdev->roce.roce_en)
+		err = mlx5_nic_vport_update_roce_state(mdev, MLX5_VPORT_ROCE_ENABLED);
+
+	if (!err)
+		mdev->roce.roce_en++;
+	mutex_unlock(&mlx5_roce_en_lock);
+
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx5_nic_vport_enable_roce);
+
+int mlx5_nic_vport_disable_roce(struct mlx5_core_dev *mdev)
+{
+	int err = 0;
+
+	mutex_lock(&mlx5_roce_en_lock);
+	if (mdev->roce.roce_en) {
+		mdev->roce.roce_en--;
+		if (mdev->roce.roce_en == 0)
+			err = mlx5_nic_vport_update_roce_state(mdev, MLX5_VPORT_ROCE_DISABLED);
+
+		if (err)
+			mdev->roce.roce_en++;
+	}
+	mutex_unlock(&mlx5_roce_en_lock);
+	return err;
+}
+EXPORT_SYMBOL(mlx5_nic_vport_disable_roce);
+
+int mlx5_core_query_vport_counter(struct mlx5_core_dev *dev, u8 other_vport,
+				  int vf, u8 port_num, void *out)
+{
+	int in_sz = MLX5_ST_SZ_BYTES(query_vport_counter_in);
+	int is_group_manager;
+	void *in;
+	int err;
+
+	is_group_manager = MLX5_CAP_GEN(dev, vport_group_manager);
+	in = kvzalloc(in_sz, GFP_KERNEL);
+	if (!in) {
+		err = -ENOMEM;
+		return err;
+	}
+
+	MLX5_SET(query_vport_counter_in, in, opcode,
+		 MLX5_CMD_OP_QUERY_VPORT_COUNTER);
+	if (other_vport) {
+		if (is_group_manager) {
+			MLX5_SET(query_vport_counter_in, in, other_vport, 1);
+			MLX5_SET(query_vport_counter_in, in, vport_number, vf + 1);
+		} else {
+			err = -EPERM;
+			goto free;
+		}
+	}
+	if (MLX5_CAP_GEN(dev, num_ports) == 2)
+		MLX5_SET(query_vport_counter_in, in, port_num, port_num);
+
+	err = mlx5_cmd_exec_inout(dev, query_vport_counter, in, out);
+free:
+	kvfree(in);
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx5_core_query_vport_counter);
+
+int mlx5_query_vport_down_stats(struct mlx5_core_dev *mdev, u16 vport,
+				u8 other_vport, u64 *rx_discard_vport_down,
+				u64 *tx_discard_vport_down)
+{
+	u32 out[MLX5_ST_SZ_DW(query_vnic_env_out)] = {};
+	u32 in[MLX5_ST_SZ_DW(query_vnic_env_in)] = {};
+	int err;
+
+	MLX5_SET(query_vnic_env_in, in, opcode,
+		 MLX5_CMD_OP_QUERY_VNIC_ENV);
+	MLX5_SET(query_vnic_env_in, in, op_mod, 0);
+	MLX5_SET(query_vnic_env_in, in, vport_number, vport);
+	MLX5_SET(query_vnic_env_in, in, other_vport, other_vport);
+
+	err = mlx5_cmd_exec_inout(mdev, query_vnic_env, in, out);
+	if (err)
+		return err;
+
+	*rx_discard_vport_down = MLX5_GET64(query_vnic_env_out, out,
+					    vport_env.receive_discard_vport_down);
+	*tx_discard_vport_down = MLX5_GET64(query_vnic_env_out, out,
+					    vport_env.transmit_discard_vport_down);
+	return 0;
+}
+
+int mlx5_core_modify_hca_vport_context(struct mlx5_core_dev *dev,
+				       u8 other_vport, u8 port_num,
+				       int vf,
+				       struct mlx5_hca_vport_context *req)
+{
+	int in_sz = MLX5_ST_SZ_BYTES(modify_hca_vport_context_in);
+	int is_group_manager;
+	void *ctx;
+	void *in;
+	int err;
+
+	mlx5_core_dbg(dev, "vf %d\n", vf);
+	is_group_manager = MLX5_CAP_GEN(dev, vport_group_manager);
+	in = kvzalloc(in_sz, GFP_KERNEL);
+	if (!in)
+		return -ENOMEM;
+
+	MLX5_SET(modify_hca_vport_context_in, in, opcode, MLX5_CMD_OP_MODIFY_HCA_VPORT_CONTEXT);
+	if (other_vport) {
+		if (is_group_manager) {
+			MLX5_SET(modify_hca_vport_context_in, in, other_vport, 1);
+			MLX5_SET(modify_hca_vport_context_in, in, vport_number, vf);
+		} else {
+			err = -EPERM;
+			goto ex;
+		}
+	}
+
+	if (MLX5_CAP_GEN(dev, num_ports) > 1)
+		MLX5_SET(modify_hca_vport_context_in, in, port_num, port_num);
+
+	ctx = MLX5_ADDR_OF(modify_hca_vport_context_in, in, hca_vport_context);
+	MLX5_SET(hca_vport_context, ctx, field_select, req->field_select);
+	if (req->field_select & MLX5_HCA_VPORT_SEL_STATE_POLICY)
+		MLX5_SET(hca_vport_context, ctx, vport_state_policy,
+			 req->policy);
+	if (req->field_select & MLX5_HCA_VPORT_SEL_PORT_GUID)
+		MLX5_SET64(hca_vport_context, ctx, port_guid, req->port_guid);
+	if (req->field_select & MLX5_HCA_VPORT_SEL_NODE_GUID)
+		MLX5_SET64(hca_vport_context, ctx, node_guid, req->node_guid);
+	MLX5_SET(hca_vport_context, ctx, cap_mask1, req->cap_mask1);
+	MLX5_SET(hca_vport_context, ctx, cap_mask1_field_select,
+		 req->cap_mask1_perm);
+	err = mlx5_cmd_exec_in(dev, modify_hca_vport_context, in);
+ex:
+	kvfree(in);
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx5_core_modify_hca_vport_context);
+
+int mlx5_nic_vport_affiliate_multiport(struct mlx5_core_dev *master_mdev,
+				       struct mlx5_core_dev *port_mdev)
+{
+	int inlen = MLX5_ST_SZ_BYTES(modify_nic_vport_context_in);
+	void *in;
+	int err;
+
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in)
+		return -ENOMEM;
+
+	err = mlx5_nic_vport_enable_roce(port_mdev);
+	if (err)
+		goto free;
+
+	MLX5_SET(modify_nic_vport_context_in, in, field_select.affiliation, 1);
+	if (MLX5_CAP_GEN_2(master_mdev, sw_vhca_id_valid)) {
+		MLX5_SET(modify_nic_vport_context_in, in,
+			 nic_vport_context.vhca_id_type, VHCA_ID_TYPE_SW);
+		MLX5_SET(modify_nic_vport_context_in, in,
+			 nic_vport_context.affiliated_vhca_id,
+			 MLX5_CAP_GEN_2(master_mdev, sw_vhca_id));
+	} else {
+		MLX5_SET(modify_nic_vport_context_in, in,
+			 nic_vport_context.affiliated_vhca_id,
+			 MLX5_CAP_GEN(master_mdev, vhca_id));
+	}
+	MLX5_SET(modify_nic_vport_context_in, in,
+		 nic_vport_context.affiliation_criteria,
+		 MLX5_CAP_GEN(port_mdev, affiliate_nic_vport_criteria));
+	MLX5_SET(modify_nic_vport_context_in, in, opcode,
+		 MLX5_CMD_OP_MODIFY_NIC_VPORT_CONTEXT);
+
+	err = mlx5_cmd_exec_in(port_mdev, modify_nic_vport_context, in);
+	if (err)
+		mlx5_nic_vport_disable_roce(port_mdev);
+
+free:
+	kvfree(in);
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx5_nic_vport_affiliate_multiport);
+
+int mlx5_nic_vport_unaffiliate_multiport(struct mlx5_core_dev *port_mdev)
+{
+	int inlen = MLX5_ST_SZ_BYTES(modify_nic_vport_context_in);
+	void *in;
+	int err;
+
+	in = kvzalloc(inlen, GFP_KERNEL);
+	if (!in)
+		return -ENOMEM;
+
+	MLX5_SET(modify_nic_vport_context_in, in, field_select.affiliation, 1);
+	MLX5_SET(modify_nic_vport_context_in, in,
+		 nic_vport_context.affiliated_vhca_id, 0);
+	MLX5_SET(modify_nic_vport_context_in, in,
+		 nic_vport_context.affiliation_criteria, 0);
+	MLX5_SET(modify_nic_vport_context_in, in, opcode,
+		 MLX5_CMD_OP_MODIFY_NIC_VPORT_CONTEXT);
+
+	err = mlx5_cmd_exec_in(port_mdev, modify_nic_vport_context, in);
+	if (!err)
+		mlx5_nic_vport_disable_roce(port_mdev);
+
+	kvfree(in);
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx5_nic_vport_unaffiliate_multiport);
+
+u64 mlx5_query_nic_system_image_guid(struct mlx5_core_dev *mdev)
+{
+	int port_type_cap = MLX5_CAP_GEN(mdev, port_type);
+	u64 tmp;
+	int err;
+
+	if (mdev->sys_image_guid)
+		return mdev->sys_image_guid;
+
+	if (port_type_cap == MLX5_CAP_PORT_TYPE_ETH)
+		err = mlx5_query_nic_vport_system_image_guid(mdev, &tmp);
+	else
+		err = mlx5_query_hca_vport_system_image_guid(mdev, &tmp);
+
+	mdev->sys_image_guid = err ? 0 : tmp;
+
+	return mdev->sys_image_guid;
+}
+EXPORT_SYMBOL_GPL(mlx5_query_nic_system_image_guid);
+
+int mlx5_vport_get_other_func_cap(struct mlx5_core_dev *dev, u16 function_id, void *out)
+{
+	u16 opmod = (MLX5_CAP_GENERAL << 1) | (HCA_CAP_OPMOD_GET_MAX & 0x01);
+	u8 in[MLX5_ST_SZ_BYTES(query_hca_cap_in)] = {};
+
+	MLX5_SET(query_hca_cap_in, in, opcode, MLX5_CMD_OP_QUERY_HCA_CAP);
+	MLX5_SET(query_hca_cap_in, in, op_mod, opmod);
+	MLX5_SET(query_hca_cap_in, in, function_id, function_id);
+	MLX5_SET(query_hca_cap_in, in, other_function, true);
+	return mlx5_cmd_exec_inout(dev, query_hca_cap, in, out);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/wq.c b/drivers/net/ethernet/mellanox/mlx5/core/wq.c
new file mode 100644
index 000000000..3091dd014
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/wq.c
@@ -0,0 +1,261 @@
+/*
+ * Copyright (c) 2013-2015, Mellanox Technologies, Ltd.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/mlx5/driver.h>
+#include "wq.h"
+#include "mlx5_core.h"
+
+int mlx5_wq_cyc_create(struct mlx5_core_dev *mdev, struct mlx5_wq_param *param,
+		       void *wqc, struct mlx5_wq_cyc *wq,
+		       struct mlx5_wq_ctrl *wq_ctrl)
+{
+	u8 log_wq_stride = MLX5_GET(wq, wqc, log_wq_stride);
+	u8 log_wq_sz     = MLX5_GET(wq, wqc, log_wq_sz);
+	struct mlx5_frag_buf_ctrl *fbc = &wq->fbc;
+	int err;
+
+	err = mlx5_db_alloc_node(mdev, &wq_ctrl->db, param->db_numa_node);
+	if (err) {
+		mlx5_core_warn(mdev, "mlx5_db_alloc_node() failed, %d\n", err);
+		return err;
+	}
+
+	wq->db  = wq_ctrl->db.db;
+
+	err = mlx5_frag_buf_alloc_node(mdev, wq_get_byte_sz(log_wq_sz, log_wq_stride),
+				       &wq_ctrl->buf, param->buf_numa_node);
+	if (err) {
+		mlx5_core_warn(mdev, "mlx5_frag_buf_alloc_node() failed, %d\n", err);
+		goto err_db_free;
+	}
+
+	mlx5_init_fbc(wq_ctrl->buf.frags, log_wq_stride, log_wq_sz, fbc);
+	wq->sz = mlx5_wq_cyc_get_size(wq);
+
+	wq_ctrl->mdev = mdev;
+
+	return 0;
+
+err_db_free:
+	mlx5_db_free(mdev, &wq_ctrl->db);
+
+	return err;
+}
+
+void mlx5_wq_cyc_wqe_dump(struct mlx5_wq_cyc *wq, u16 ix, u8 nstrides)
+{
+	size_t len;
+	void *wqe;
+
+	if (!net_ratelimit())
+		return;
+
+	nstrides = max_t(u8, nstrides, 1);
+
+	len = nstrides << wq->fbc.log_stride;
+	wqe = mlx5_wq_cyc_get_wqe(wq, ix);
+
+	pr_info("WQE DUMP: WQ size %d WQ cur size %d, WQE index 0x%x, len: %zu\n",
+		mlx5_wq_cyc_get_size(wq), wq->cur_sz, ix, len);
+	print_hex_dump(KERN_WARNING, "", DUMP_PREFIX_OFFSET, 16, 1, wqe, len, false);
+}
+
+void mlx5_wq_cyc_reset(struct mlx5_wq_cyc *wq)
+{
+	wq->wqe_ctr = 0;
+	wq->cur_sz = 0;
+	mlx5_wq_cyc_update_db_record(wq);
+}
+
+int mlx5_wq_qp_create(struct mlx5_core_dev *mdev, struct mlx5_wq_param *param,
+		      void *qpc, struct mlx5_wq_qp *wq,
+		      struct mlx5_wq_ctrl *wq_ctrl)
+{
+	u8 log_rq_stride = MLX5_GET(qpc, qpc, log_rq_stride) + 4;
+	u8 log_rq_sz     = MLX5_GET(qpc, qpc, log_rq_size);
+	u8 log_sq_stride = ilog2(MLX5_SEND_WQE_BB);
+	u8 log_sq_sz     = MLX5_GET(qpc, qpc, log_sq_size);
+
+	u32 rq_byte_size;
+	int err;
+
+
+
+	err = mlx5_db_alloc_node(mdev, &wq_ctrl->db, param->db_numa_node);
+	if (err) {
+		mlx5_core_warn(mdev, "mlx5_db_alloc_node() failed, %d\n", err);
+		return err;
+	}
+
+	err = mlx5_frag_buf_alloc_node(mdev,
+				       wq_get_byte_sz(log_rq_sz, log_rq_stride) +
+				       wq_get_byte_sz(log_sq_sz, log_sq_stride),
+				       &wq_ctrl->buf, param->buf_numa_node);
+	if (err) {
+		mlx5_core_warn(mdev, "mlx5_frag_buf_alloc_node() failed, %d\n", err);
+		goto err_db_free;
+	}
+
+	mlx5_init_fbc(wq_ctrl->buf.frags, log_rq_stride, log_rq_sz, &wq->rq.fbc);
+
+	rq_byte_size = wq_get_byte_sz(log_rq_sz, log_rq_stride);
+
+	if (rq_byte_size < PAGE_SIZE) {
+		/* SQ starts within the same page of the RQ */
+		u16 sq_strides_offset = rq_byte_size / MLX5_SEND_WQE_BB;
+
+		mlx5_init_fbc_offset(wq_ctrl->buf.frags,
+				     log_sq_stride, log_sq_sz, sq_strides_offset,
+				     &wq->sq.fbc);
+	} else {
+		u16 rq_npages = rq_byte_size >> PAGE_SHIFT;
+
+		mlx5_init_fbc(wq_ctrl->buf.frags + rq_npages,
+			      log_sq_stride, log_sq_sz, &wq->sq.fbc);
+	}
+
+	wq->rq.db  = &wq_ctrl->db.db[MLX5_RCV_DBR];
+	wq->sq.db  = &wq_ctrl->db.db[MLX5_SND_DBR];
+
+	wq_ctrl->mdev = mdev;
+
+	return 0;
+
+err_db_free:
+	mlx5_db_free(mdev, &wq_ctrl->db);
+
+	return err;
+}
+
+int mlx5_cqwq_create(struct mlx5_core_dev *mdev, struct mlx5_wq_param *param,
+		     void *cqc, struct mlx5_cqwq *wq,
+		     struct mlx5_wq_ctrl *wq_ctrl)
+{
+	/* CQE_STRIDE_128 and CQE_STRIDE_128_PAD both mean 128B stride */
+	u8 log_wq_stride = MLX5_GET(cqc, cqc, cqe_sz) == CQE_STRIDE_64 ? 6 : 7;
+	u8 log_wq_sz     = MLX5_GET(cqc, cqc, log_cq_size);
+	int err;
+
+	err = mlx5_db_alloc_node(mdev, &wq_ctrl->db, param->db_numa_node);
+	if (err) {
+		mlx5_core_warn(mdev, "mlx5_db_alloc_node() failed, %d\n", err);
+		return err;
+	}
+
+	wq->db  = wq_ctrl->db.db;
+
+	err = mlx5_frag_buf_alloc_node(mdev, wq_get_byte_sz(log_wq_sz, log_wq_stride),
+				       &wq_ctrl->buf,
+				       param->buf_numa_node);
+	if (err) {
+		mlx5_core_warn(mdev, "mlx5_frag_buf_alloc_node() failed, %d\n",
+			       err);
+		goto err_db_free;
+	}
+
+	mlx5_init_fbc(wq_ctrl->buf.frags, log_wq_stride, log_wq_sz, &wq->fbc);
+
+	wq_ctrl->mdev = mdev;
+
+	return 0;
+
+err_db_free:
+	mlx5_db_free(mdev, &wq_ctrl->db);
+
+	return err;
+}
+
+static void mlx5_wq_ll_init_list(struct mlx5_wq_ll *wq)
+{
+	struct mlx5_wqe_srq_next_seg *next_seg;
+	int i;
+
+	for (i = 0; i < wq->fbc.sz_m1; i++) {
+		next_seg = mlx5_wq_ll_get_wqe(wq, i);
+		next_seg->next_wqe_index = cpu_to_be16(i + 1);
+	}
+	next_seg = mlx5_wq_ll_get_wqe(wq, i);
+	wq->tail_next = &next_seg->next_wqe_index;
+}
+
+int mlx5_wq_ll_create(struct mlx5_core_dev *mdev, struct mlx5_wq_param *param,
+		      void *wqc, struct mlx5_wq_ll *wq,
+		      struct mlx5_wq_ctrl *wq_ctrl)
+{
+	u8 log_wq_stride = MLX5_GET(wq, wqc, log_wq_stride);
+	u8 log_wq_sz     = MLX5_GET(wq, wqc, log_wq_sz);
+	struct mlx5_frag_buf_ctrl *fbc = &wq->fbc;
+	int err;
+
+	err = mlx5_db_alloc_node(mdev, &wq_ctrl->db, param->db_numa_node);
+	if (err) {
+		mlx5_core_warn(mdev, "mlx5_db_alloc_node() failed, %d\n", err);
+		return err;
+	}
+
+	wq->db  = wq_ctrl->db.db;
+
+	err = mlx5_frag_buf_alloc_node(mdev, wq_get_byte_sz(log_wq_sz, log_wq_stride),
+				       &wq_ctrl->buf, param->buf_numa_node);
+	if (err) {
+		mlx5_core_warn(mdev, "mlx5_frag_buf_alloc_node() failed, %d\n", err);
+		goto err_db_free;
+	}
+
+	mlx5_init_fbc(wq_ctrl->buf.frags, log_wq_stride, log_wq_sz, fbc);
+
+	mlx5_wq_ll_init_list(wq);
+	wq_ctrl->mdev = mdev;
+
+	return 0;
+
+err_db_free:
+	mlx5_db_free(mdev, &wq_ctrl->db);
+
+	return err;
+}
+
+void mlx5_wq_ll_reset(struct mlx5_wq_ll *wq)
+{
+	wq->head = 0;
+	wq->wqe_ctr = 0;
+	wq->cur_sz = 0;
+	mlx5_wq_ll_init_list(wq);
+	mlx5_wq_ll_update_db_record(wq);
+}
+
+void mlx5_wq_destroy(struct mlx5_wq_ctrl *wq_ctrl)
+{
+	mlx5_frag_buf_free(wq_ctrl->mdev, &wq_ctrl->buf);
+	mlx5_db_free(wq_ctrl->mdev, &wq_ctrl->db);
+}
+
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/wq.h b/drivers/net/ethernet/mellanox/mlx5/core/wq.h
new file mode 100644
index 000000000..4d629e5dd
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/wq.h
@@ -0,0 +1,308 @@
+/*
+ * Copyright (c) 2013-2015, Mellanox Technologies, Ltd.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef __MLX5_WQ_H__
+#define __MLX5_WQ_H__
+
+#include <linux/mlx5/mlx5_ifc.h>
+#include <linux/mlx5/cq.h>
+#include <linux/mlx5/qp.h>
+
+struct mlx5_wq_param {
+	int		buf_numa_node;
+	int		db_numa_node;
+};
+
+struct mlx5_wq_ctrl {
+	struct mlx5_core_dev	*mdev;
+	struct mlx5_frag_buf	buf;
+	struct mlx5_db		db;
+};
+
+struct mlx5_wq_cyc {
+	struct mlx5_frag_buf_ctrl fbc;
+	__be32			*db;
+	u16			sz;
+	u16			wqe_ctr;
+	u16			cur_sz;
+};
+
+struct mlx5_wq_qp {
+	struct mlx5_wq_cyc	rq;
+	struct mlx5_wq_cyc	sq;
+};
+
+struct mlx5_cqwq {
+	struct mlx5_frag_buf_ctrl fbc;
+	__be32			  *db;
+	u32			  cc; /* consumer counter */
+};
+
+struct mlx5_wq_ll {
+	struct mlx5_frag_buf_ctrl fbc;
+	__be32			*db;
+	__be16			*tail_next;
+	u16			head;
+	u16			wqe_ctr;
+	u16			cur_sz;
+};
+
+int mlx5_wq_cyc_create(struct mlx5_core_dev *mdev, struct mlx5_wq_param *param,
+		       void *wqc, struct mlx5_wq_cyc *wq,
+		       struct mlx5_wq_ctrl *wq_ctrl);
+void mlx5_wq_cyc_wqe_dump(struct mlx5_wq_cyc *wq, u16 ix, u8 nstrides);
+void mlx5_wq_cyc_reset(struct mlx5_wq_cyc *wq);
+
+int mlx5_wq_qp_create(struct mlx5_core_dev *mdev, struct mlx5_wq_param *param,
+		      void *qpc, struct mlx5_wq_qp *wq,
+		      struct mlx5_wq_ctrl *wq_ctrl);
+
+int mlx5_cqwq_create(struct mlx5_core_dev *mdev, struct mlx5_wq_param *param,
+		     void *cqc, struct mlx5_cqwq *wq,
+		     struct mlx5_wq_ctrl *wq_ctrl);
+
+int mlx5_wq_ll_create(struct mlx5_core_dev *mdev, struct mlx5_wq_param *param,
+		      void *wqc, struct mlx5_wq_ll *wq,
+		      struct mlx5_wq_ctrl *wq_ctrl);
+void mlx5_wq_ll_reset(struct mlx5_wq_ll *wq);
+
+void mlx5_wq_destroy(struct mlx5_wq_ctrl *wq_ctrl);
+
+static inline u32 mlx5_wq_cyc_get_size(struct mlx5_wq_cyc *wq)
+{
+	return (u32)wq->fbc.sz_m1 + 1;
+}
+
+static inline int mlx5_wq_cyc_is_full(struct mlx5_wq_cyc *wq)
+{
+	return wq->cur_sz == wq->sz;
+}
+
+static inline int mlx5_wq_cyc_missing(struct mlx5_wq_cyc *wq)
+{
+	return wq->sz - wq->cur_sz;
+}
+
+static inline int mlx5_wq_cyc_is_empty(struct mlx5_wq_cyc *wq)
+{
+	return !wq->cur_sz;
+}
+
+static inline void mlx5_wq_cyc_push(struct mlx5_wq_cyc *wq)
+{
+	wq->wqe_ctr++;
+	wq->cur_sz++;
+}
+
+static inline void mlx5_wq_cyc_push_n(struct mlx5_wq_cyc *wq, u16 n)
+{
+	wq->wqe_ctr += n;
+	wq->cur_sz += n;
+}
+
+static inline void mlx5_wq_cyc_pop(struct mlx5_wq_cyc *wq)
+{
+	wq->cur_sz--;
+}
+
+static inline void mlx5_wq_cyc_update_db_record(struct mlx5_wq_cyc *wq)
+{
+	*wq->db = cpu_to_be32(wq->wqe_ctr);
+}
+
+static inline u16 mlx5_wq_cyc_ctr2ix(struct mlx5_wq_cyc *wq, u16 ctr)
+{
+	return ctr & wq->fbc.sz_m1;
+}
+
+static inline u16 mlx5_wq_cyc_get_head(struct mlx5_wq_cyc *wq)
+{
+	return mlx5_wq_cyc_ctr2ix(wq, wq->wqe_ctr);
+}
+
+static inline u16 mlx5_wq_cyc_get_tail(struct mlx5_wq_cyc *wq)
+{
+	return mlx5_wq_cyc_ctr2ix(wq, wq->wqe_ctr - wq->cur_sz);
+}
+
+static inline void *mlx5_wq_cyc_get_wqe(struct mlx5_wq_cyc *wq, u16 ix)
+{
+	return mlx5_frag_buf_get_wqe(&wq->fbc, ix);
+}
+
+static inline u16 mlx5_wq_cyc_get_contig_wqebbs(struct mlx5_wq_cyc *wq, u16 ix)
+{
+	return mlx5_frag_buf_get_idx_last_contig_stride(&wq->fbc, ix) - ix + 1;
+}
+
+static inline int mlx5_wq_cyc_cc_bigger(u16 cc1, u16 cc2)
+{
+	int equal   = (cc1 == cc2);
+	int smaller = 0x8000 & (cc1 - cc2);
+
+	return !equal && !smaller;
+}
+
+static inline u16 mlx5_wq_cyc_get_counter(struct mlx5_wq_cyc *wq)
+{
+	return wq->wqe_ctr;
+}
+
+static inline u32 mlx5_cqwq_get_size(struct mlx5_cqwq *wq)
+{
+	return wq->fbc.sz_m1 + 1;
+}
+
+static inline u8 mlx5_cqwq_get_log_stride_size(struct mlx5_cqwq *wq)
+{
+	return wq->fbc.log_stride;
+}
+
+static inline u32 mlx5_cqwq_ctr2ix(struct mlx5_cqwq *wq, u32 ctr)
+{
+	return ctr & wq->fbc.sz_m1;
+}
+
+static inline u32 mlx5_cqwq_get_ci(struct mlx5_cqwq *wq)
+{
+	return mlx5_cqwq_ctr2ix(wq, wq->cc);
+}
+
+static inline struct mlx5_cqe64 *mlx5_cqwq_get_wqe(struct mlx5_cqwq *wq, u32 ix)
+{
+	struct mlx5_cqe64 *cqe = mlx5_frag_buf_get_wqe(&wq->fbc, ix);
+
+	/* For 128B CQEs the data is in the last 64B */
+	cqe += wq->fbc.log_stride == 7;
+
+	return cqe;
+}
+
+static inline u32 mlx5_cqwq_get_ctr_wrap_cnt(struct mlx5_cqwq *wq, u32 ctr)
+{
+	return ctr >> wq->fbc.log_sz;
+}
+
+static inline u32 mlx5_cqwq_get_wrap_cnt(struct mlx5_cqwq *wq)
+{
+	return mlx5_cqwq_get_ctr_wrap_cnt(wq, wq->cc);
+}
+
+static inline void mlx5_cqwq_pop(struct mlx5_cqwq *wq)
+{
+	wq->cc++;
+}
+
+static inline void mlx5_cqwq_update_db_record(struct mlx5_cqwq *wq)
+{
+	*wq->db = cpu_to_be32(wq->cc & 0xffffff);
+}
+
+static inline struct mlx5_cqe64 *mlx5_cqwq_get_cqe(struct mlx5_cqwq *wq)
+{
+	u32 ci = mlx5_cqwq_get_ci(wq);
+	struct mlx5_cqe64 *cqe = mlx5_cqwq_get_wqe(wq, ci);
+	u8 cqe_ownership_bit = cqe->op_own & MLX5_CQE_OWNER_MASK;
+	u8 sw_ownership_val = mlx5_cqwq_get_wrap_cnt(wq) & 1;
+
+	if (cqe_ownership_bit != sw_ownership_val)
+		return NULL;
+
+	/* ensure cqe content is read after cqe ownership bit */
+	dma_rmb();
+
+	return cqe;
+}
+
+static inline u32 mlx5_wq_ll_get_size(struct mlx5_wq_ll *wq)
+{
+	return (u32)wq->fbc.sz_m1 + 1;
+}
+
+static inline int mlx5_wq_ll_is_full(struct mlx5_wq_ll *wq)
+{
+	return wq->cur_sz == wq->fbc.sz_m1;
+}
+
+static inline int mlx5_wq_ll_is_empty(struct mlx5_wq_ll *wq)
+{
+	return !wq->cur_sz;
+}
+
+static inline int mlx5_wq_ll_missing(struct mlx5_wq_ll *wq)
+{
+	return wq->fbc.sz_m1 - wq->cur_sz;
+}
+
+static inline void *mlx5_wq_ll_get_wqe(struct mlx5_wq_ll *wq, u16 ix)
+{
+	return mlx5_frag_buf_get_wqe(&wq->fbc, ix);
+}
+
+static inline u16 mlx5_wq_ll_get_wqe_next_ix(struct mlx5_wq_ll *wq, u16 ix)
+{
+	struct mlx5_wqe_srq_next_seg *wqe = mlx5_wq_ll_get_wqe(wq, ix);
+
+	return be16_to_cpu(wqe->next_wqe_index);
+}
+
+static inline void mlx5_wq_ll_push(struct mlx5_wq_ll *wq, u16 head_next)
+{
+	wq->head = head_next;
+	wq->wqe_ctr++;
+	wq->cur_sz++;
+}
+
+static inline void mlx5_wq_ll_pop(struct mlx5_wq_ll *wq, __be16 ix,
+				  __be16 *next_tail_next)
+{
+	*wq->tail_next = ix;
+	wq->tail_next = next_tail_next;
+	wq->cur_sz--;
+}
+
+static inline void mlx5_wq_ll_update_db_record(struct mlx5_wq_ll *wq)
+{
+	*wq->db = cpu_to_be32(wq->wqe_ctr);
+}
+
+static inline u16 mlx5_wq_ll_get_head(struct mlx5_wq_ll *wq)
+{
+	return wq->head;
+}
+
+static inline u16 mlx5_wq_ll_get_counter(struct mlx5_wq_ll *wq)
+{
+	return wq->wqe_ctr;
+}
+
+#endif /* __MLX5_WQ_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlxbf_gige/Kconfig b/drivers/net/ethernet/mellanox/mlxbf_gige/Kconfig
new file mode 100644
index 000000000..4cdebafaf
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxbf_gige/Kconfig
@@ -0,0 +1,13 @@
+# SPDX-License-Identifier: GPL-2.0-only OR BSD-3-Clause
+#
+# Mellanox GigE driver configuration
+#
+
+config MLXBF_GIGE
+	tristate "Mellanox Technologies BlueField Gigabit Ethernet support"
+	depends on (ARM64 && ACPI) || COMPILE_TEST
+	select PHYLIB
+	help
+	  The second generation BlueField SoC from Mellanox Technologies
+	  supports an out-of-band Gigabit Ethernet management port to the
+	  Arm subsystem.
diff --git a/drivers/net/ethernet/mellanox/mlxbf_gige/Makefile b/drivers/net/ethernet/mellanox/mlxbf_gige/Makefile
new file mode 100644
index 000000000..a97c2bef8
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxbf_gige/Makefile
@@ -0,0 +1,10 @@
+# SPDX-License-Identifier: GPL-2.0-only OR BSD-3-Clause
+
+obj-$(CONFIG_MLXBF_GIGE) += mlxbf_gige.o
+
+mlxbf_gige-y := mlxbf_gige_ethtool.o \
+		mlxbf_gige_intr.o \
+		mlxbf_gige_main.o \
+		mlxbf_gige_mdio.o \
+		mlxbf_gige_rx.o   \
+		mlxbf_gige_tx.o
diff --git a/drivers/net/ethernet/mellanox/mlxbf_gige/mlxbf_gige.h b/drivers/net/ethernet/mellanox/mlxbf_gige/mlxbf_gige.h
new file mode 100644
index 000000000..5a1027b07
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxbf_gige/mlxbf_gige.h
@@ -0,0 +1,177 @@
+/* SPDX-License-Identifier: GPL-2.0-only OR BSD-3-Clause */
+
+/* Header file for Gigabit Ethernet driver for Mellanox BlueField SoC
+ * - this file contains software data structures and any chip-specific
+ *   data structures (e.g. TX WQE format) that are memory resident.
+ *
+ * Copyright (C) 2020-2021 NVIDIA CORPORATION & AFFILIATES
+ */
+
+#ifndef __MLXBF_GIGE_H__
+#define __MLXBF_GIGE_H__
+
+#include <linux/io-64-nonatomic-lo-hi.h>
+#include <linux/irqreturn.h>
+#include <linux/netdevice.h>
+#include <linux/irq.h>
+
+/* The silicon design supports a maximum RX ring size of
+ * 32K entries. Based on current testing this maximum size
+ * is not required to be supported.  Instead the RX ring
+ * will be capped at a realistic value of 1024 entries.
+ */
+#define MLXBF_GIGE_MIN_RXQ_SZ     32
+#define MLXBF_GIGE_MAX_RXQ_SZ     1024
+#define MLXBF_GIGE_DEFAULT_RXQ_SZ 128
+
+#define MLXBF_GIGE_MIN_TXQ_SZ     4
+#define MLXBF_GIGE_MAX_TXQ_SZ     256
+#define MLXBF_GIGE_DEFAULT_TXQ_SZ 128
+
+#define MLXBF_GIGE_DEFAULT_BUF_SZ 2048
+
+#define MLXBF_GIGE_DMA_PAGE_SZ    4096
+#define MLXBF_GIGE_DMA_PAGE_SHIFT 12
+
+/* There are four individual MAC RX filters. Currently
+ * two of them are being used: one for the broadcast MAC
+ * (index 0) and one for local MAC (index 1)
+ */
+#define MLXBF_GIGE_BCAST_MAC_FILTER_IDX 0
+#define MLXBF_GIGE_LOCAL_MAC_FILTER_IDX 1
+
+/* Define for broadcast MAC literal */
+#define BCAST_MAC_ADDR 0xFFFFFFFFFFFF
+
+/* There are three individual interrupts:
+ *   1) Errors, "OOB" interrupt line
+ *   2) Receive Packet, "OOB_LLU" interrupt line
+ *   3) LLU and PLU Events, "OOB_PLU" interrupt line
+ */
+#define MLXBF_GIGE_ERROR_INTR_IDX       0
+#define MLXBF_GIGE_RECEIVE_PKT_INTR_IDX 1
+#define MLXBF_GIGE_LLU_PLU_INTR_IDX     2
+
+struct mlxbf_gige_stats {
+	u64 hw_access_errors;
+	u64 tx_invalid_checksums;
+	u64 tx_small_frames;
+	u64 tx_index_errors;
+	u64 sw_config_errors;
+	u64 sw_access_errors;
+	u64 rx_truncate_errors;
+	u64 rx_mac_errors;
+	u64 rx_din_dropped_pkts;
+	u64 tx_fifo_full;
+	u64 rx_filter_passed_pkts;
+	u64 rx_filter_discard_pkts;
+};
+
+struct mlxbf_gige {
+	void __iomem *base;
+	void __iomem *llu_base;
+	void __iomem *plu_base;
+	struct device *dev;
+	struct net_device *netdev;
+	struct platform_device *pdev;
+	void __iomem *mdio_io;
+	void __iomem *clk_io;
+	struct mii_bus *mdiobus;
+	spinlock_t lock;      /* for packet processing indices */
+	u16 rx_q_entries;
+	u16 tx_q_entries;
+	u64 *tx_wqe_base;
+	dma_addr_t tx_wqe_base_dma;
+	u64 *tx_wqe_next;
+	u64 *tx_cc;
+	dma_addr_t tx_cc_dma;
+	dma_addr_t *rx_wqe_base;
+	dma_addr_t rx_wqe_base_dma;
+	u64 *rx_cqe_base;
+	dma_addr_t rx_cqe_base_dma;
+	u16 tx_pi;
+	u16 prev_tx_ci;
+	struct sk_buff *rx_skb[MLXBF_GIGE_MAX_RXQ_SZ];
+	struct sk_buff *tx_skb[MLXBF_GIGE_MAX_TXQ_SZ];
+	int error_irq;
+	int rx_irq;
+	int llu_plu_irq;
+	int phy_irq;
+	int hw_phy_irq;
+	bool promisc_enabled;
+	u8 valid_polarity;
+	struct napi_struct napi;
+	struct mlxbf_gige_stats stats;
+};
+
+/* Rx Work Queue Element definitions */
+#define MLXBF_GIGE_RX_WQE_SZ                   8
+
+/* Rx Completion Queue Element definitions */
+#define MLXBF_GIGE_RX_CQE_SZ                   8
+#define MLXBF_GIGE_RX_CQE_PKT_LEN_MASK         GENMASK(10, 0)
+#define MLXBF_GIGE_RX_CQE_VALID_MASK           GENMASK(11, 11)
+#define MLXBF_GIGE_RX_CQE_PKT_STATUS_MASK      GENMASK(15, 12)
+#define MLXBF_GIGE_RX_CQE_PKT_STATUS_MAC_ERR   GENMASK(12, 12)
+#define MLXBF_GIGE_RX_CQE_PKT_STATUS_TRUNCATED GENMASK(13, 13)
+#define MLXBF_GIGE_RX_CQE_CHKSUM_MASK          GENMASK(31, 16)
+
+/* Tx Work Queue Element definitions */
+#define MLXBF_GIGE_TX_WQE_SZ_QWORDS            2
+#define MLXBF_GIGE_TX_WQE_SZ                   16
+#define MLXBF_GIGE_TX_WQE_PKT_LEN_MASK         GENMASK(10, 0)
+#define MLXBF_GIGE_TX_WQE_UPDATE_MASK          GENMASK(31, 31)
+#define MLXBF_GIGE_TX_WQE_CHKSUM_LEN_MASK      GENMASK(42, 32)
+#define MLXBF_GIGE_TX_WQE_CHKSUM_START_MASK    GENMASK(55, 48)
+#define MLXBF_GIGE_TX_WQE_CHKSUM_OFFSET_MASK   GENMASK(63, 56)
+
+/* Macro to return packet length of specified TX WQE */
+#define MLXBF_GIGE_TX_WQE_PKT_LEN(tx_wqe_addr) \
+	(*((tx_wqe_addr) + 1) & MLXBF_GIGE_TX_WQE_PKT_LEN_MASK)
+
+/* Tx Completion Count */
+#define MLXBF_GIGE_TX_CC_SZ                    8
+
+/* List of resources in ACPI table */
+enum mlxbf_gige_res {
+	MLXBF_GIGE_RES_MAC,
+	MLXBF_GIGE_RES_MDIO9,
+	MLXBF_GIGE_RES_GPIO0,
+	MLXBF_GIGE_RES_LLU,
+	MLXBF_GIGE_RES_PLU,
+	MLXBF_GIGE_RES_CLK
+};
+
+/* Version of register data returned by mlxbf_gige_get_regs() */
+#define MLXBF_GIGE_REGS_VERSION 1
+
+int mlxbf_gige_mdio_probe(struct platform_device *pdev,
+			  struct mlxbf_gige *priv);
+void mlxbf_gige_mdio_remove(struct mlxbf_gige *priv);
+irqreturn_t mlxbf_gige_mdio_handle_phy_interrupt(int irq, void *dev_id);
+void mlxbf_gige_mdio_enable_phy_int(struct mlxbf_gige *priv);
+
+void mlxbf_gige_set_mac_rx_filter(struct mlxbf_gige *priv,
+				  unsigned int index, u64 dmac);
+void mlxbf_gige_get_mac_rx_filter(struct mlxbf_gige *priv,
+				  unsigned int index, u64 *dmac);
+void mlxbf_gige_enable_promisc(struct mlxbf_gige *priv);
+void mlxbf_gige_disable_promisc(struct mlxbf_gige *priv);
+int mlxbf_gige_rx_init(struct mlxbf_gige *priv);
+void mlxbf_gige_rx_deinit(struct mlxbf_gige *priv);
+int mlxbf_gige_tx_init(struct mlxbf_gige *priv);
+void mlxbf_gige_tx_deinit(struct mlxbf_gige *priv);
+bool mlxbf_gige_handle_tx_complete(struct mlxbf_gige *priv);
+netdev_tx_t mlxbf_gige_start_xmit(struct sk_buff *skb,
+				  struct net_device *netdev);
+struct sk_buff *mlxbf_gige_alloc_skb(struct mlxbf_gige *priv,
+				     unsigned int map_len,
+				     dma_addr_t *buf_dma,
+				     enum dma_data_direction dir);
+int mlxbf_gige_request_irqs(struct mlxbf_gige *priv);
+void mlxbf_gige_free_irqs(struct mlxbf_gige *priv);
+int mlxbf_gige_poll(struct napi_struct *napi, int budget);
+extern const struct ethtool_ops mlxbf_gige_ethtool_ops;
+void mlxbf_gige_update_tx_wqe_next(struct mlxbf_gige *priv);
+
+#endif /* !defined(__MLXBF_GIGE_H__) */
diff --git a/drivers/net/ethernet/mellanox/mlxbf_gige/mlxbf_gige_ethtool.c b/drivers/net/ethernet/mellanox/mlxbf_gige/mlxbf_gige_ethtool.c
new file mode 100644
index 000000000..41ebef25a
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxbf_gige/mlxbf_gige_ethtool.c
@@ -0,0 +1,138 @@
+// SPDX-License-Identifier: GPL-2.0-only OR BSD-3-Clause
+
+/* Ethtool support for Mellanox Gigabit Ethernet driver
+ *
+ * Copyright (C) 2020-2021 NVIDIA CORPORATION & AFFILIATES
+ */
+
+#include <linux/phy.h>
+
+#include "mlxbf_gige.h"
+#include "mlxbf_gige_regs.h"
+
+/* Start of struct ethtool_ops functions */
+static int mlxbf_gige_get_regs_len(struct net_device *netdev)
+{
+	return MLXBF_GIGE_MMIO_REG_SZ;
+}
+
+static void mlxbf_gige_get_regs(struct net_device *netdev,
+				struct ethtool_regs *regs, void *p)
+{
+	struct mlxbf_gige *priv = netdev_priv(netdev);
+
+	regs->version = MLXBF_GIGE_REGS_VERSION;
+
+	/* Read entire MMIO register space and store results
+	 * into the provided buffer. By design, a read to an
+	 * offset without an existing register will be
+	 * acknowledged and return zero.
+	 */
+	memcpy_fromio(p, priv->base, MLXBF_GIGE_MMIO_REG_SZ);
+}
+
+static void
+mlxbf_gige_get_ringparam(struct net_device *netdev,
+			 struct ethtool_ringparam *ering,
+			 struct kernel_ethtool_ringparam *kernel_ering,
+			 struct netlink_ext_ack *extack)
+{
+	struct mlxbf_gige *priv = netdev_priv(netdev);
+
+	ering->rx_max_pending = MLXBF_GIGE_MAX_RXQ_SZ;
+	ering->tx_max_pending = MLXBF_GIGE_MAX_TXQ_SZ;
+	ering->rx_pending = priv->rx_q_entries;
+	ering->tx_pending = priv->tx_q_entries;
+}
+
+static const struct {
+	const char string[ETH_GSTRING_LEN];
+} mlxbf_gige_ethtool_stats_keys[] = {
+	{ "hw_access_errors" },
+	{ "tx_invalid_checksums" },
+	{ "tx_small_frames" },
+	{ "tx_index_errors" },
+	{ "sw_config_errors" },
+	{ "sw_access_errors" },
+	{ "rx_truncate_errors" },
+	{ "rx_mac_errors" },
+	{ "rx_din_dropped_pkts" },
+	{ "tx_fifo_full" },
+	{ "rx_filter_passed_pkts" },
+	{ "rx_filter_discard_pkts" },
+};
+
+static int mlxbf_gige_get_sset_count(struct net_device *netdev, int stringset)
+{
+	if (stringset != ETH_SS_STATS)
+		return -EOPNOTSUPP;
+	return ARRAY_SIZE(mlxbf_gige_ethtool_stats_keys);
+}
+
+static void mlxbf_gige_get_strings(struct net_device *netdev, u32 stringset,
+				   u8 *buf)
+{
+	if (stringset != ETH_SS_STATS)
+		return;
+	memcpy(buf, &mlxbf_gige_ethtool_stats_keys,
+	       sizeof(mlxbf_gige_ethtool_stats_keys));
+}
+
+static void mlxbf_gige_get_ethtool_stats(struct net_device *netdev,
+					 struct ethtool_stats *estats,
+					 u64 *data)
+{
+	struct mlxbf_gige *priv = netdev_priv(netdev);
+
+	/* Fill data array with interface statistics
+	 *
+	 * NOTE: the data writes must be in
+	 *       sync with the strings shown in
+	 *       the mlxbf_gige_ethtool_stats_keys[] array
+	 *
+	 * NOTE2: certain statistics below are zeroed upon
+	 *        port disable, so the calculation below
+	 *        must include the "cached" value of the stat
+	 *        plus the value read directly from hardware.
+	 *        Cached statistics are currently:
+	 *          rx_din_dropped_pkts
+	 *          rx_filter_passed_pkts
+	 *          rx_filter_discard_pkts
+	 */
+	*data++ = priv->stats.hw_access_errors;
+	*data++ = priv->stats.tx_invalid_checksums;
+	*data++ = priv->stats.tx_small_frames;
+	*data++ = priv->stats.tx_index_errors;
+	*data++ = priv->stats.sw_config_errors;
+	*data++ = priv->stats.sw_access_errors;
+	*data++ = priv->stats.rx_truncate_errors;
+	*data++ = priv->stats.rx_mac_errors;
+	*data++ = (priv->stats.rx_din_dropped_pkts +
+		   readq(priv->base + MLXBF_GIGE_RX_DIN_DROP_COUNTER));
+	*data++ = priv->stats.tx_fifo_full;
+	*data++ = (priv->stats.rx_filter_passed_pkts +
+		   readq(priv->base + MLXBF_GIGE_RX_PASS_COUNTER_ALL));
+	*data++ = (priv->stats.rx_filter_discard_pkts +
+		   readq(priv->base + MLXBF_GIGE_RX_DISC_COUNTER_ALL));
+}
+
+static void mlxbf_gige_get_pauseparam(struct net_device *netdev,
+				      struct ethtool_pauseparam *pause)
+{
+	pause->autoneg = AUTONEG_DISABLE;
+	pause->rx_pause = 1;
+	pause->tx_pause = 1;
+}
+
+const struct ethtool_ops mlxbf_gige_ethtool_ops = {
+	.get_link		= ethtool_op_get_link,
+	.get_ringparam		= mlxbf_gige_get_ringparam,
+	.get_regs_len           = mlxbf_gige_get_regs_len,
+	.get_regs               = mlxbf_gige_get_regs,
+	.get_strings            = mlxbf_gige_get_strings,
+	.get_sset_count         = mlxbf_gige_get_sset_count,
+	.get_ethtool_stats      = mlxbf_gige_get_ethtool_stats,
+	.nway_reset		= phy_ethtool_nway_reset,
+	.get_pauseparam		= mlxbf_gige_get_pauseparam,
+	.get_link_ksettings	= phy_ethtool_get_link_ksettings,
+};
diff --git a/drivers/net/ethernet/mellanox/mlxbf_gige/mlxbf_gige_intr.c b/drivers/net/ethernet/mellanox/mlxbf_gige/mlxbf_gige_intr.c
new file mode 100644
index 000000000..5b3519f0c
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxbf_gige/mlxbf_gige_intr.c
@@ -0,0 +1,133 @@
+// SPDX-License-Identifier: GPL-2.0-only OR BSD-3-Clause
+
+/* Interrupt related logic for Mellanox Gigabit Ethernet driver
+ *
+ * Copyright (C) 2020-2021 NVIDIA CORPORATION & AFFILIATES
+ */
+
+#include <linux/interrupt.h>
+
+#include "mlxbf_gige.h"
+#include "mlxbf_gige_regs.h"
+
+static irqreturn_t mlxbf_gige_error_intr(int irq, void *dev_id)
+{
+	struct mlxbf_gige *priv;
+	u64 int_status;
+
+	priv = dev_id;
+
+	int_status = readq(priv->base + MLXBF_GIGE_INT_STATUS);
+
+	if (int_status & MLXBF_GIGE_INT_STATUS_HW_ACCESS_ERROR)
+		priv->stats.hw_access_errors++;
+
+	if (int_status & MLXBF_GIGE_INT_STATUS_TX_CHECKSUM_INPUTS) {
+		priv->stats.tx_invalid_checksums++;
+		/* This error condition is latched into MLXBF_GIGE_INT_STATUS
+		 * when the GigE silicon operates on the offending
+		 * TX WQE. The write to MLXBF_GIGE_INT_STATUS at the bottom
+		 * of this routine clears this error condition.
+		 */
+	}
+
+	if (int_status & MLXBF_GIGE_INT_STATUS_TX_SMALL_FRAME_SIZE) {
+		priv->stats.tx_small_frames++;
+		/* This condition happens when the networking stack invokes
+		 * this driver's "start_xmit()" method with a packet whose
+		 * size < 60 bytes.  The GigE silicon will automatically pad
+		 * this small frame up to a minimum-sized frame before it is
+		 * sent. The "tx_small_frame" condition is latched into the
+		 * MLXBF_GIGE_INT_STATUS register when the GigE silicon
+		 * operates on the offending TX WQE. The write to
+		 * MLXBF_GIGE_INT_STATUS at the bottom of this routine
+		 * clears this condition.
+		 */
+	}
+
+	if (int_status & MLXBF_GIGE_INT_STATUS_TX_PI_CI_EXCEED_WQ_SIZE)
+		priv->stats.tx_index_errors++;
+
+	if (int_status & MLXBF_GIGE_INT_STATUS_SW_CONFIG_ERROR)
+		priv->stats.sw_config_errors++;
+
+	if (int_status & MLXBF_GIGE_INT_STATUS_SW_ACCESS_ERROR)
+		priv->stats.sw_access_errors++;
+
+	/* Clear all error interrupts by writing '1' back to
+	 * all the asserted bits in INT_STATUS.  Do not write
+	 * '1' back to 'receive packet' bit, since that is
+	 * managed separately.
+	 */
+
+	int_status &= ~MLXBF_GIGE_INT_STATUS_RX_RECEIVE_PACKET;
+
+	writeq(int_status, priv->base + MLXBF_GIGE_INT_STATUS);
+
+	return IRQ_HANDLED;
+}
+
+static irqreturn_t mlxbf_gige_rx_intr(int irq, void *dev_id)
+{
+	struct mlxbf_gige *priv;
+
+	priv = dev_id;
+
+	/* NOTE: GigE silicon automatically disables "packet rx" interrupt by
+	 *       setting MLXBF_GIGE_INT_MASK bit0 upon triggering the interrupt
+	 *       to the ARM cores.  Software needs to re-enable "packet rx"
+	 *       interrupts by clearing MLXBF_GIGE_INT_MASK bit0.
+	 */
+
+	napi_schedule(&priv->napi);
+
+	return IRQ_HANDLED;
+}
+
+static irqreturn_t mlxbf_gige_llu_plu_intr(int irq, void *dev_id)
+{
+	return IRQ_HANDLED;
+}
+
+int mlxbf_gige_request_irqs(struct mlxbf_gige *priv)
+{
+	int err;
+
+	err = request_irq(priv->error_irq, mlxbf_gige_error_intr, 0,
+			  "mlxbf_gige_error", priv);
+	if (err) {
+		dev_err(priv->dev, "Request error_irq failure\n");
+		return err;
+	}
+
+	err = request_irq(priv->rx_irq, mlxbf_gige_rx_intr, 0,
+			  "mlxbf_gige_rx", priv);
+	if (err) {
+		dev_err(priv->dev, "Request rx_irq failure\n");
+		goto free_error_irq;
+	}
+
+	err = request_irq(priv->llu_plu_irq, mlxbf_gige_llu_plu_intr, 0,
+			  "mlxbf_gige_llu_plu", priv);
+	if (err) {
+		dev_err(priv->dev, "Request llu_plu_irq failure\n");
+		goto free_rx_irq;
+	}
+
+	return 0;
+
+free_rx_irq:
+	free_irq(priv->rx_irq, priv);
+
+free_error_irq:
+	free_irq(priv->error_irq, priv);
+
+	return err;
+}
+
+void mlxbf_gige_free_irqs(struct mlxbf_gige *priv)
+{
+	free_irq(priv->error_irq, priv);
+	free_irq(priv->rx_irq, priv);
+	free_irq(priv->llu_plu_irq, priv);
+}
diff --git a/drivers/net/ethernet/mellanox/mlxbf_gige/mlxbf_gige_main.c b/drivers/net/ethernet/mellanox/mlxbf_gige/mlxbf_gige_main.c
new file mode 100644
index 000000000..83c465939
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxbf_gige/mlxbf_gige_main.c
@@ -0,0 +1,438 @@
+// SPDX-License-Identifier: GPL-2.0-only OR BSD-3-Clause
+
+/* Gigabit Ethernet driver for Mellanox BlueField SoC
+ *
+ * Copyright (C) 2020-2021 NVIDIA CORPORATION & AFFILIATES
+ */
+
+#include <linux/acpi.h>
+#include <linux/device.h>
+#include <linux/dma-mapping.h>
+#include <linux/etherdevice.h>
+#include <linux/interrupt.h>
+#include <linux/iopoll.h>
+#include <linux/module.h>
+#include <linux/phy.h>
+#include <linux/platform_device.h>
+#include <linux/skbuff.h>
+
+#include "mlxbf_gige.h"
+#include "mlxbf_gige_regs.h"
+
+/* Allocate SKB whose payload pointer aligns with the Bluefield
+ * hardware DMA limitation, i.e. DMA operation can't cross
+ * a 4KB boundary.  A maximum packet size of 2KB is assumed in the
+ * alignment formula.  The alignment logic overallocates an SKB,
+ * and then adjusts the headroom so that the SKB data pointer is
+ * naturally aligned to a 2KB boundary.
+ */
+struct sk_buff *mlxbf_gige_alloc_skb(struct mlxbf_gige *priv,
+				     unsigned int map_len,
+				     dma_addr_t *buf_dma,
+				     enum dma_data_direction dir)
+{
+	struct sk_buff *skb;
+	u64 addr, offset;
+
+	/* Overallocate the SKB so that any headroom adjustment (to
+	 * provide 2KB natural alignment) does not exceed payload area
+	 */
+	skb = netdev_alloc_skb(priv->netdev, MLXBF_GIGE_DEFAULT_BUF_SZ * 2);
+	if (!skb)
+		return NULL;
+
+	/* Adjust the headroom so that skb->data is naturally aligned to
+	 * a 2KB boundary, which is the maximum packet size supported.
+	 */
+	addr = (long)skb->data;
+	offset = (addr + MLXBF_GIGE_DEFAULT_BUF_SZ - 1) &
+		~(MLXBF_GIGE_DEFAULT_BUF_SZ - 1);
+	offset -= addr;
+	if (offset)
+		skb_reserve(skb, offset);
+
+	/* Return streaming DMA mapping to caller */
+	*buf_dma = dma_map_single(priv->dev, skb->data, map_len, dir);
+	if (dma_mapping_error(priv->dev, *buf_dma)) {
+		dev_kfree_skb(skb);
+		*buf_dma = (dma_addr_t)0;
+		return NULL;
+	}
+
+	return skb;
+}
+
+static void mlxbf_gige_initial_mac(struct mlxbf_gige *priv)
+{
+	u8 mac[ETH_ALEN];
+	u64 local_mac;
+
+	eth_zero_addr(mac);
+	mlxbf_gige_get_mac_rx_filter(priv, MLXBF_GIGE_LOCAL_MAC_FILTER_IDX,
+				     &local_mac);
+	u64_to_ether_addr(local_mac, mac);
+
+	if (is_valid_ether_addr(mac)) {
+		eth_hw_addr_set(priv->netdev, mac);
+	} else {
+		/* Provide a random MAC if for some reason the device has
+		 * not been configured with a valid MAC address already.
+		 */
+		eth_hw_addr_random(priv->netdev);
+	}
+
+	local_mac = ether_addr_to_u64(priv->netdev->dev_addr);
+	mlxbf_gige_set_mac_rx_filter(priv, MLXBF_GIGE_LOCAL_MAC_FILTER_IDX,
+				     local_mac);
+}
+
+static void mlxbf_gige_cache_stats(struct mlxbf_gige *priv)
+{
+	struct mlxbf_gige_stats *p;
+
+	/* Cache stats that will be cleared by clean port operation */
+	p = &priv->stats;
+	p->rx_din_dropped_pkts += readq(priv->base +
+					MLXBF_GIGE_RX_DIN_DROP_COUNTER);
+	p->rx_filter_passed_pkts += readq(priv->base +
+					  MLXBF_GIGE_RX_PASS_COUNTER_ALL);
+	p->rx_filter_discard_pkts += readq(priv->base +
+					   MLXBF_GIGE_RX_DISC_COUNTER_ALL);
+}
+
+static int mlxbf_gige_clean_port(struct mlxbf_gige *priv)
+{
+	u64 control;
+	u64 temp;
+	int err;
+
+	/* Set the CLEAN_PORT_EN bit to trigger SW reset */
+	control = readq(priv->base + MLXBF_GIGE_CONTROL);
+	control |= MLXBF_GIGE_CONTROL_CLEAN_PORT_EN;
+	writeq(control, priv->base + MLXBF_GIGE_CONTROL);
+
+	/* Ensure completion of "clean port" write before polling status */
+	mb();
+
+	err = readq_poll_timeout_atomic(priv->base + MLXBF_GIGE_STATUS, temp,
+					(temp & MLXBF_GIGE_STATUS_READY),
+					100, 100000);
+
+	/* Clear the CLEAN_PORT_EN bit at end of this loop */
+	control = readq(priv->base + MLXBF_GIGE_CONTROL);
+	control &= ~MLXBF_GIGE_CONTROL_CLEAN_PORT_EN;
+	writeq(control, priv->base + MLXBF_GIGE_CONTROL);
+
+	return err;
+}
+
+static int mlxbf_gige_open(struct net_device *netdev)
+{
+	struct mlxbf_gige *priv = netdev_priv(netdev);
+	struct phy_device *phydev = netdev->phydev;
+	u64 control;
+	u64 int_en;
+	int err;
+
+	/* Perform general init of GigE block */
+	control = readq(priv->base + MLXBF_GIGE_CONTROL);
+	control |= MLXBF_GIGE_CONTROL_PORT_EN;
+	writeq(control, priv->base + MLXBF_GIGE_CONTROL);
+
+	err = mlxbf_gige_request_irqs(priv);
+	if (err)
+		return err;
+	mlxbf_gige_cache_stats(priv);
+	err = mlxbf_gige_clean_port(priv);
+	if (err)
+		goto free_irqs;
+
+	/* Clear driver's valid_polarity to match hardware,
+	 * since the above call to clean_port() resets the
+	 * receive polarity used by hardware.
+	 */
+	priv->valid_polarity = 0;
+
+	phy_start(phydev);
+
+	err = mlxbf_gige_tx_init(priv);
+	if (err)
+		goto free_irqs;
+	err = mlxbf_gige_rx_init(priv);
+	if (err)
+		goto tx_deinit;
+
+	netif_napi_add(netdev, &priv->napi, mlxbf_gige_poll);
+	napi_enable(&priv->napi);
+	netif_start_queue(netdev);
+
+	/* Set bits in INT_EN that we care about */
+	int_en = MLXBF_GIGE_INT_EN_HW_ACCESS_ERROR |
+		 MLXBF_GIGE_INT_EN_TX_CHECKSUM_INPUTS |
+		 MLXBF_GIGE_INT_EN_TX_SMALL_FRAME_SIZE |
+		 MLXBF_GIGE_INT_EN_TX_PI_CI_EXCEED_WQ_SIZE |
+		 MLXBF_GIGE_INT_EN_SW_CONFIG_ERROR |
+		 MLXBF_GIGE_INT_EN_SW_ACCESS_ERROR |
+		 MLXBF_GIGE_INT_EN_RX_RECEIVE_PACKET;
+
+	/* Ensure completion of all initialization before enabling interrupts */
+	mb();
+
+	writeq(int_en, priv->base + MLXBF_GIGE_INT_EN);
+
+	return 0;
+
+tx_deinit:
+	mlxbf_gige_tx_deinit(priv);
+
+free_irqs:
+	mlxbf_gige_free_irqs(priv);
+	return err;
+}
+
+static int mlxbf_gige_stop(struct net_device *netdev)
+{
+	struct mlxbf_gige *priv = netdev_priv(netdev);
+
+	writeq(0, priv->base + MLXBF_GIGE_INT_EN);
+	netif_stop_queue(netdev);
+	napi_disable(&priv->napi);
+	netif_napi_del(&priv->napi);
+	mlxbf_gige_free_irqs(priv);
+
+	phy_stop(netdev->phydev);
+
+	mlxbf_gige_rx_deinit(priv);
+	mlxbf_gige_tx_deinit(priv);
+	mlxbf_gige_cache_stats(priv);
+	mlxbf_gige_clean_port(priv);
+
+	return 0;
+}
+
+static int mlxbf_gige_eth_ioctl(struct net_device *netdev,
+			       struct ifreq *ifr, int cmd)
+{
+	if (!(netif_running(netdev)))
+		return -EINVAL;
+
+	return phy_mii_ioctl(netdev->phydev, ifr, cmd);
+}
+
+static void mlxbf_gige_set_rx_mode(struct net_device *netdev)
+{
+	struct mlxbf_gige *priv = netdev_priv(netdev);
+	bool new_promisc_enabled;
+
+	new_promisc_enabled = netdev->flags & IFF_PROMISC;
+
+	/* Only write to the hardware registers if the new setting
+	 * of promiscuous mode is different from the current one.
+	 */
+	if (new_promisc_enabled != priv->promisc_enabled) {
+		priv->promisc_enabled = new_promisc_enabled;
+
+		if (new_promisc_enabled)
+			mlxbf_gige_enable_promisc(priv);
+		else
+			mlxbf_gige_disable_promisc(priv);
+	}
+}
+
+static void mlxbf_gige_get_stats64(struct net_device *netdev,
+				   struct rtnl_link_stats64 *stats)
+{
+	struct mlxbf_gige *priv = netdev_priv(netdev);
+
+	netdev_stats_to_stats64(stats, &netdev->stats);
+
+	stats->rx_length_errors = priv->stats.rx_truncate_errors;
+	stats->rx_fifo_errors = priv->stats.rx_din_dropped_pkts +
+				readq(priv->base + MLXBF_GIGE_RX_DIN_DROP_COUNTER);
+	stats->rx_crc_errors = priv->stats.rx_mac_errors;
+	stats->rx_errors = stats->rx_length_errors +
+			   stats->rx_fifo_errors +
+			   stats->rx_crc_errors;
+
+	stats->tx_fifo_errors = priv->stats.tx_fifo_full;
+	stats->tx_errors = stats->tx_fifo_errors;
+}
+
+static const struct net_device_ops mlxbf_gige_netdev_ops = {
+	.ndo_open		= mlxbf_gige_open,
+	.ndo_stop		= mlxbf_gige_stop,
+	.ndo_start_xmit		= mlxbf_gige_start_xmit,
+	.ndo_set_mac_address	= eth_mac_addr,
+	.ndo_validate_addr	= eth_validate_addr,
+	.ndo_eth_ioctl		= mlxbf_gige_eth_ioctl,
+	.ndo_set_rx_mode        = mlxbf_gige_set_rx_mode,
+	.ndo_get_stats64        = mlxbf_gige_get_stats64,
+};
+
+static void mlxbf_gige_adjust_link(struct net_device *netdev)
+{
+	struct phy_device *phydev = netdev->phydev;
+
+	phy_print_status(phydev);
+}
+
+static int mlxbf_gige_probe(struct platform_device *pdev)
+{
+	struct phy_device *phydev;
+	struct net_device *netdev;
+	struct mlxbf_gige *priv;
+	void __iomem *llu_base;
+	void __iomem *plu_base;
+	void __iomem *base;
+	int addr, phy_irq;
+	int err;
+
+	base = devm_platform_ioremap_resource(pdev, MLXBF_GIGE_RES_MAC);
+	if (IS_ERR(base))
+		return PTR_ERR(base);
+
+	llu_base = devm_platform_ioremap_resource(pdev, MLXBF_GIGE_RES_LLU);
+	if (IS_ERR(llu_base))
+		return PTR_ERR(llu_base);
+
+	plu_base = devm_platform_ioremap_resource(pdev, MLXBF_GIGE_RES_PLU);
+	if (IS_ERR(plu_base))
+		return PTR_ERR(plu_base);
+
+	netdev = devm_alloc_etherdev(&pdev->dev, sizeof(*priv));
+	if (!netdev)
+		return -ENOMEM;
+
+	SET_NETDEV_DEV(netdev, &pdev->dev);
+	netdev->netdev_ops = &mlxbf_gige_netdev_ops;
+	netdev->ethtool_ops = &mlxbf_gige_ethtool_ops;
+	priv = netdev_priv(netdev);
+	priv->netdev = netdev;
+
+	platform_set_drvdata(pdev, priv);
+	priv->dev = &pdev->dev;
+	priv->pdev = pdev;
+
+	spin_lock_init(&priv->lock);
+
+	/* Attach MDIO device */
+	err = mlxbf_gige_mdio_probe(pdev, priv);
+	if (err)
+		return err;
+
+	priv->base = base;
+	priv->llu_base = llu_base;
+	priv->plu_base = plu_base;
+
+	priv->rx_q_entries = MLXBF_GIGE_DEFAULT_RXQ_SZ;
+	priv->tx_q_entries = MLXBF_GIGE_DEFAULT_TXQ_SZ;
+
+	/* Write initial MAC address to hardware */
+	mlxbf_gige_initial_mac(priv);
+
+	err = dma_set_mask_and_coherent(&pdev->dev, DMA_BIT_MASK(64));
+	if (err) {
+		dev_err(&pdev->dev, "DMA configuration failed: 0x%x\n", err);
+		goto out;
+	}
+
+	priv->error_irq = platform_get_irq(pdev, MLXBF_GIGE_ERROR_INTR_IDX);
+	priv->rx_irq = platform_get_irq(pdev, MLXBF_GIGE_RECEIVE_PKT_INTR_IDX);
+	priv->llu_plu_irq = platform_get_irq(pdev, MLXBF_GIGE_LLU_PLU_INTR_IDX);
+
+	phy_irq = acpi_dev_gpio_irq_get_by(ACPI_COMPANION(&pdev->dev), "phy-gpios", 0);
+	if (phy_irq < 0) {
+		dev_err(&pdev->dev, "Error getting PHY irq. Use polling instead");
+		phy_irq = PHY_POLL;
+	}
+
+	phydev = phy_find_first(priv->mdiobus);
+	if (!phydev) {
+		err = -ENODEV;
+		goto out;
+	}
+
+	addr = phydev->mdio.addr;
+	priv->mdiobus->irq[addr] = phy_irq;
+	phydev->irq = phy_irq;
+
+	err = phy_connect_direct(netdev, phydev,
+				 mlxbf_gige_adjust_link,
+				 PHY_INTERFACE_MODE_GMII);
+	if (err) {
+		dev_err(&pdev->dev, "Could not attach to PHY\n");
+		goto out;
+	}
+
+	/* MAC only supports 1000T full duplex mode */
+	phy_remove_link_mode(phydev, ETHTOOL_LINK_MODE_1000baseT_Half_BIT);
+	phy_remove_link_mode(phydev, ETHTOOL_LINK_MODE_100baseT_Full_BIT);
+	phy_remove_link_mode(phydev, ETHTOOL_LINK_MODE_100baseT_Half_BIT);
+	phy_remove_link_mode(phydev, ETHTOOL_LINK_MODE_10baseT_Full_BIT);
+	phy_remove_link_mode(phydev, ETHTOOL_LINK_MODE_10baseT_Half_BIT);
+
+	/* Only symmetric pause with flow control enabled is supported so no
+	 * need to negotiate pause.
+	 */
+	linkmode_clear_bit(ETHTOOL_LINK_MODE_Pause_BIT, phydev->advertising);
+	linkmode_clear_bit(ETHTOOL_LINK_MODE_Asym_Pause_BIT, phydev->advertising);
+
+	/* Display information about attached PHY device */
+	phy_attached_info(phydev);
+
+	err = register_netdev(netdev);
+	if (err) {
+		dev_err(&pdev->dev, "Failed to register netdev\n");
+		phy_disconnect(phydev);
+		goto out;
+	}
+
+	return 0;
+
+out:
+	mlxbf_gige_mdio_remove(priv);
+	return err;
+}
+
+static int mlxbf_gige_remove(struct platform_device *pdev)
+{
+	struct mlxbf_gige *priv = platform_get_drvdata(pdev);
+
+	unregister_netdev(priv->netdev);
+	phy_disconnect(priv->netdev->phydev);
+	mlxbf_gige_mdio_remove(priv);
+	platform_set_drvdata(pdev, NULL);
+
+	return 0;
+}
+
+static void mlxbf_gige_shutdown(struct platform_device *pdev)
+{
+	struct mlxbf_gige *priv = platform_get_drvdata(pdev);
+
+	writeq(0, priv->base + MLXBF_GIGE_INT_EN);
+	mlxbf_gige_clean_port(priv);
+}
+
+static const struct acpi_device_id __maybe_unused mlxbf_gige_acpi_match[] = {
+	{ "MLNXBF17", 0 },
+	{},
+};
+MODULE_DEVICE_TABLE(acpi, mlxbf_gige_acpi_match);
+
+static struct platform_driver mlxbf_gige_driver = {
+	.probe = mlxbf_gige_probe,
+	.remove = mlxbf_gige_remove,
+	.shutdown = mlxbf_gige_shutdown,
+	.driver = {
+		.name = KBUILD_MODNAME,
+		.acpi_match_table = ACPI_PTR(mlxbf_gige_acpi_match),
+	},
+};
+
+module_platform_driver(mlxbf_gige_driver);
+
+MODULE_DESCRIPTION("Mellanox BlueField SoC Gigabit Ethernet Driver");
+MODULE_AUTHOR("David Thompson <davthompson@nvidia.com>");
+MODULE_AUTHOR("Asmaa Mnebhi <asmaa@nvidia.com>");
+MODULE_LICENSE("Dual BSD/GPL");
diff --git a/drivers/net/ethernet/mellanox/mlxbf_gige/mlxbf_gige_mdio.c b/drivers/net/ethernet/mellanox/mlxbf_gige/mlxbf_gige_mdio.c
new file mode 100644
index 000000000..aa780b161
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxbf_gige/mlxbf_gige_mdio.c
@@ -0,0 +1,278 @@
+// SPDX-License-Identifier: GPL-2.0-only OR BSD-3-Clause
+
+/* MDIO support for Mellanox Gigabit Ethernet driver
+ *
+ * Copyright (C) 2020-2021 NVIDIA CORPORATION & AFFILIATES
+ */
+
+#include <linux/acpi.h>
+#include <linux/bitfield.h>
+#include <linux/delay.h>
+#include <linux/device.h>
+#include <linux/err.h>
+#include <linux/io.h>
+#include <linux/iopoll.h>
+#include <linux/ioport.h>
+#include <linux/irqreturn.h>
+#include <linux/jiffies.h>
+#include <linux/module.h>
+#include <linux/mod_devicetable.h>
+#include <linux/phy.h>
+#include <linux/platform_device.h>
+#include <linux/property.h>
+
+#include "mlxbf_gige.h"
+#include "mlxbf_gige_regs.h"
+
+#define MLXBF_GIGE_MDIO_GW_OFFSET	0x0
+#define MLXBF_GIGE_MDIO_CFG_OFFSET	0x4
+
+#define MLXBF_GIGE_MDIO_FREQ_REFERENCE 156250000ULL
+#define MLXBF_GIGE_MDIO_COREPLL_CONST  16384ULL
+#define MLXBF_GIGE_MDC_CLK_NS          400
+#define MLXBF_GIGE_MDIO_PLL_I1CLK_REG1 0x4
+#define MLXBF_GIGE_MDIO_PLL_I1CLK_REG2 0x8
+#define MLXBF_GIGE_MDIO_CORE_F_SHIFT   0
+#define MLXBF_GIGE_MDIO_CORE_F_MASK    GENMASK(25, 0)
+#define MLXBF_GIGE_MDIO_CORE_R_SHIFT   26
+#define MLXBF_GIGE_MDIO_CORE_R_MASK    GENMASK(31, 26)
+#define MLXBF_GIGE_MDIO_CORE_OD_SHIFT  0
+#define MLXBF_GIGE_MDIO_CORE_OD_MASK   GENMASK(3, 0)
+
+/* Support clause 22 */
+#define MLXBF_GIGE_MDIO_CL22_ST1	0x1
+#define MLXBF_GIGE_MDIO_CL22_WRITE	0x1
+#define MLXBF_GIGE_MDIO_CL22_READ	0x2
+
+/* Busy bit is set by software and cleared by hardware */
+#define MLXBF_GIGE_MDIO_SET_BUSY	0x1
+
+/* MDIO GW register bits */
+#define MLXBF_GIGE_MDIO_GW_AD_MASK	GENMASK(15, 0)
+#define MLXBF_GIGE_MDIO_GW_DEVAD_MASK	GENMASK(20, 16)
+#define MLXBF_GIGE_MDIO_GW_PARTAD_MASK	GENMASK(25, 21)
+#define MLXBF_GIGE_MDIO_GW_OPCODE_MASK	GENMASK(27, 26)
+#define MLXBF_GIGE_MDIO_GW_ST1_MASK	GENMASK(28, 28)
+#define MLXBF_GIGE_MDIO_GW_BUSY_MASK	GENMASK(30, 30)
+
+/* MDIO config register bits */
+#define MLXBF_GIGE_MDIO_CFG_MDIO_MODE_MASK		GENMASK(1, 0)
+#define MLXBF_GIGE_MDIO_CFG_MDIO3_3_MASK		GENMASK(2, 2)
+#define MLXBF_GIGE_MDIO_CFG_MDIO_FULL_DRIVE_MASK	GENMASK(4, 4)
+#define MLXBF_GIGE_MDIO_CFG_MDC_PERIOD_MASK		GENMASK(15, 8)
+#define MLXBF_GIGE_MDIO_CFG_MDIO_IN_SAMP_MASK		GENMASK(23, 16)
+#define MLXBF_GIGE_MDIO_CFG_MDIO_OUT_SAMP_MASK		GENMASK(31, 24)
+
+#define MLXBF_GIGE_MDIO_CFG_VAL (FIELD_PREP(MLXBF_GIGE_MDIO_CFG_MDIO_MODE_MASK, 1) | \
+				 FIELD_PREP(MLXBF_GIGE_MDIO_CFG_MDIO3_3_MASK, 1) | \
+				 FIELD_PREP(MLXBF_GIGE_MDIO_CFG_MDIO_FULL_DRIVE_MASK, 1) | \
+				 FIELD_PREP(MLXBF_GIGE_MDIO_CFG_MDIO_IN_SAMP_MASK, 6) | \
+				 FIELD_PREP(MLXBF_GIGE_MDIO_CFG_MDIO_OUT_SAMP_MASK, 13))
+
+#define MLXBF_GIGE_BF2_COREPLL_ADDR 0x02800c30
+#define MLXBF_GIGE_BF2_COREPLL_SIZE 0x0000000c
+
+static struct resource corepll_params[] = {
+	[MLXBF_GIGE_VERSION_BF2] = {
+		.start = MLXBF_GIGE_BF2_COREPLL_ADDR,
+		.end = MLXBF_GIGE_BF2_COREPLL_ADDR + MLXBF_GIGE_BF2_COREPLL_SIZE - 1,
+		.name = "COREPLL_RES"
+	},
+};
+
+/* Returns core clock i1clk in Hz */
+static u64 calculate_i1clk(struct mlxbf_gige *priv)
+{
+	u8 core_od, core_r;
+	u64 freq_output;
+	u32 reg1, reg2;
+	u32 core_f;
+
+	reg1 = readl(priv->clk_io + MLXBF_GIGE_MDIO_PLL_I1CLK_REG1);
+	reg2 = readl(priv->clk_io + MLXBF_GIGE_MDIO_PLL_I1CLK_REG2);
+
+	core_f = (reg1 & MLXBF_GIGE_MDIO_CORE_F_MASK) >>
+		MLXBF_GIGE_MDIO_CORE_F_SHIFT;
+	core_r = (reg1 & MLXBF_GIGE_MDIO_CORE_R_MASK) >>
+		MLXBF_GIGE_MDIO_CORE_R_SHIFT;
+	core_od = (reg2 & MLXBF_GIGE_MDIO_CORE_OD_MASK) >>
+		MLXBF_GIGE_MDIO_CORE_OD_SHIFT;
+
+	/* Compute PLL output frequency as follow:
+	 *
+	 *                                     CORE_F / 16384
+	 * freq_output = freq_reference * ----------------------------
+	 *                              (CORE_R + 1) * (CORE_OD + 1)
+	 */
+	freq_output = div_u64((MLXBF_GIGE_MDIO_FREQ_REFERENCE * core_f),
+			      MLXBF_GIGE_MDIO_COREPLL_CONST);
+	freq_output = div_u64(freq_output, (core_r + 1) * (core_od + 1));
+
+	return freq_output;
+}
+
+/* Formula for encoding the MDIO period. The encoded value is
+ * passed to the MDIO config register.
+ *
+ * mdc_clk = 2*(val + 1)*(core clock in sec)
+ *
+ * i1clk is in Hz:
+ * 400 ns = 2*(val + 1)*(1/i1clk)
+ *
+ * val = (((400/10^9) / (1/i1clk) / 2) - 1)
+ * val = (400/2 * i1clk)/10^9 - 1
+ */
+static u8 mdio_period_map(struct mlxbf_gige *priv)
+{
+	u8 mdio_period;
+	u64 i1clk;
+
+	i1clk = calculate_i1clk(priv);
+
+	mdio_period = div_u64((MLXBF_GIGE_MDC_CLK_NS >> 1) * i1clk, 1000000000) - 1;
+
+	return mdio_period;
+}
+
+static u32 mlxbf_gige_mdio_create_cmd(u16 data, int phy_add,
+				      int phy_reg, u32 opcode)
+{
+	u32 gw_reg = 0;
+
+	gw_reg |= FIELD_PREP(MLXBF_GIGE_MDIO_GW_AD_MASK, data);
+	gw_reg |= FIELD_PREP(MLXBF_GIGE_MDIO_GW_DEVAD_MASK, phy_reg);
+	gw_reg |= FIELD_PREP(MLXBF_GIGE_MDIO_GW_PARTAD_MASK, phy_add);
+	gw_reg |= FIELD_PREP(MLXBF_GIGE_MDIO_GW_OPCODE_MASK, opcode);
+	gw_reg |= FIELD_PREP(MLXBF_GIGE_MDIO_GW_ST1_MASK,
+			     MLXBF_GIGE_MDIO_CL22_ST1);
+	gw_reg |= FIELD_PREP(MLXBF_GIGE_MDIO_GW_BUSY_MASK,
+			     MLXBF_GIGE_MDIO_SET_BUSY);
+
+	return gw_reg;
+}
+
+static int mlxbf_gige_mdio_read(struct mii_bus *bus, int phy_add, int phy_reg)
+{
+	struct mlxbf_gige *priv = bus->priv;
+	u32 cmd;
+	int ret;
+	u32 val;
+
+	if (phy_reg & MII_ADDR_C45)
+		return -EOPNOTSUPP;
+
+	/* Send mdio read request */
+	cmd = mlxbf_gige_mdio_create_cmd(0, phy_add, phy_reg, MLXBF_GIGE_MDIO_CL22_READ);
+
+	writel(cmd, priv->mdio_io + MLXBF_GIGE_MDIO_GW_OFFSET);
+
+	ret = readl_poll_timeout_atomic(priv->mdio_io + MLXBF_GIGE_MDIO_GW_OFFSET,
+					val, !(val & MLXBF_GIGE_MDIO_GW_BUSY_MASK),
+					5, 1000000);
+
+	if (ret) {
+		writel(0, priv->mdio_io + MLXBF_GIGE_MDIO_GW_OFFSET);
+		return ret;
+	}
+
+	ret = readl(priv->mdio_io + MLXBF_GIGE_MDIO_GW_OFFSET);
+	/* Only return ad bits of the gw register */
+	ret &= MLXBF_GIGE_MDIO_GW_AD_MASK;
+
+	/* The MDIO lock is set on read. To release it, clear gw register */
+	writel(0, priv->mdio_io + MLXBF_GIGE_MDIO_GW_OFFSET);
+
+	return ret;
+}
+
+static int mlxbf_gige_mdio_write(struct mii_bus *bus, int phy_add,
+				 int phy_reg, u16 val)
+{
+	struct mlxbf_gige *priv = bus->priv;
+	u32 temp;
+	u32 cmd;
+	int ret;
+
+	if (phy_reg & MII_ADDR_C45)
+		return -EOPNOTSUPP;
+
+	/* Send mdio write request */
+	cmd = mlxbf_gige_mdio_create_cmd(val, phy_add, phy_reg,
+					 MLXBF_GIGE_MDIO_CL22_WRITE);
+	writel(cmd, priv->mdio_io + MLXBF_GIGE_MDIO_GW_OFFSET);
+
+	/* If the poll timed out, drop the request */
+	ret = readl_poll_timeout_atomic(priv->mdio_io + MLXBF_GIGE_MDIO_GW_OFFSET,
+					temp, !(temp & MLXBF_GIGE_MDIO_GW_BUSY_MASK),
+					5, 1000000);
+
+	/* The MDIO lock is set on read. To release it, clear gw register */
+	writel(0, priv->mdio_io + MLXBF_GIGE_MDIO_GW_OFFSET);
+
+	return ret;
+}
+
+static void mlxbf_gige_mdio_cfg(struct mlxbf_gige *priv)
+{
+	u8 mdio_period;
+	u32 val;
+
+	mdio_period = mdio_period_map(priv);
+
+	val = MLXBF_GIGE_MDIO_CFG_VAL;
+	val |= FIELD_PREP(MLXBF_GIGE_MDIO_CFG_MDC_PERIOD_MASK, mdio_period);
+	writel(val, priv->mdio_io + MLXBF_GIGE_MDIO_CFG_OFFSET);
+}
+
+int mlxbf_gige_mdio_probe(struct platform_device *pdev, struct mlxbf_gige *priv)
+{
+	struct device *dev = &pdev->dev;
+	struct resource *res;
+	int ret;
+
+	priv->mdio_io = devm_platform_ioremap_resource(pdev, MLXBF_GIGE_RES_MDIO9);
+	if (IS_ERR(priv->mdio_io))
+		return PTR_ERR(priv->mdio_io);
+
+	/* clk resource shared with other drivers so cannot use
+	 * devm_platform_ioremap_resource
+	 */
+	res = platform_get_resource(pdev, IORESOURCE_MEM, MLXBF_GIGE_RES_CLK);
+	if (!res) {
+		/* For backward compatibility with older ACPI tables, also keep
+		 * CLK resource internal to the driver.
+		 */
+		res = &corepll_params[MLXBF_GIGE_VERSION_BF2];
+	}
+
+	priv->clk_io = devm_ioremap(dev, res->start, resource_size(res));
+	if (!priv->clk_io)
+		return -ENOMEM;
+
+	mlxbf_gige_mdio_cfg(priv);
+
+	priv->mdiobus = devm_mdiobus_alloc(dev);
+	if (!priv->mdiobus) {
+		dev_err(dev, "Failed to alloc MDIO bus\n");
+		return -ENOMEM;
+	}
+
+	priv->mdiobus->name = "mlxbf-mdio";
+	priv->mdiobus->read = mlxbf_gige_mdio_read;
+	priv->mdiobus->write = mlxbf_gige_mdio_write;
+	priv->mdiobus->parent = dev;
+	priv->mdiobus->priv = priv;
+	snprintf(priv->mdiobus->id, MII_BUS_ID_SIZE, "%s",
+		 dev_name(dev));
+
+	ret = mdiobus_register(priv->mdiobus);
+	if (ret)
+		dev_err(dev, "Failed to register MDIO bus\n");
+
+	return ret;
+}
+
+void mlxbf_gige_mdio_remove(struct mlxbf_gige *priv)
+{
+	mdiobus_unregister(priv->mdiobus);
+}
diff --git a/drivers/net/ethernet/mellanox/mlxbf_gige/mlxbf_gige_regs.h b/drivers/net/ethernet/mellanox/mlxbf_gige/mlxbf_gige_regs.h
new file mode 100644
index 000000000..7be3a7939
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxbf_gige/mlxbf_gige_regs.h
@@ -0,0 +1,80 @@
+/* SPDX-License-Identifier: GPL-2.0-only OR BSD-3-Clause */
+
+/* Header file for Mellanox BlueField GigE register defines
+ *
+ * Copyright (C) 2020-2021 NVIDIA CORPORATION & AFFILIATES
+ */
+
+#ifndef __MLXBF_GIGE_REGS_H__
+#define __MLXBF_GIGE_REGS_H__
+
+#define MLXBF_GIGE_VERSION                            0x0000
+#define MLXBF_GIGE_VERSION_BF2                        0x0
+#define MLXBF_GIGE_STATUS                             0x0010
+#define MLXBF_GIGE_STATUS_READY                       BIT(0)
+#define MLXBF_GIGE_INT_STATUS                         0x0028
+#define MLXBF_GIGE_INT_STATUS_RX_RECEIVE_PACKET       BIT(0)
+#define MLXBF_GIGE_INT_STATUS_RX_MAC_ERROR            BIT(1)
+#define MLXBF_GIGE_INT_STATUS_RX_TRN_ERROR            BIT(2)
+#define MLXBF_GIGE_INT_STATUS_SW_ACCESS_ERROR         BIT(3)
+#define MLXBF_GIGE_INT_STATUS_SW_CONFIG_ERROR         BIT(4)
+#define MLXBF_GIGE_INT_STATUS_TX_PI_CI_EXCEED_WQ_SIZE BIT(5)
+#define MLXBF_GIGE_INT_STATUS_TX_SMALL_FRAME_SIZE     BIT(6)
+#define MLXBF_GIGE_INT_STATUS_TX_CHECKSUM_INPUTS      BIT(7)
+#define MLXBF_GIGE_INT_STATUS_HW_ACCESS_ERROR         BIT(8)
+#define MLXBF_GIGE_INT_EN                             0x0030
+#define MLXBF_GIGE_INT_EN_RX_RECEIVE_PACKET           BIT(0)
+#define MLXBF_GIGE_INT_EN_RX_MAC_ERROR                BIT(1)
+#define MLXBF_GIGE_INT_EN_RX_TRN_ERROR                BIT(2)
+#define MLXBF_GIGE_INT_EN_SW_ACCESS_ERROR             BIT(3)
+#define MLXBF_GIGE_INT_EN_SW_CONFIG_ERROR             BIT(4)
+#define MLXBF_GIGE_INT_EN_TX_PI_CI_EXCEED_WQ_SIZE     BIT(5)
+#define MLXBF_GIGE_INT_EN_TX_SMALL_FRAME_SIZE         BIT(6)
+#define MLXBF_GIGE_INT_EN_TX_CHECKSUM_INPUTS          BIT(7)
+#define MLXBF_GIGE_INT_EN_HW_ACCESS_ERROR             BIT(8)
+#define MLXBF_GIGE_INT_MASK                           0x0038
+#define MLXBF_GIGE_INT_MASK_RX_RECEIVE_PACKET         BIT(0)
+#define MLXBF_GIGE_CONTROL                            0x0040
+#define MLXBF_GIGE_CONTROL_PORT_EN                    BIT(0)
+#define MLXBF_GIGE_CONTROL_MAC_ID_RANGE_EN            BIT(1)
+#define MLXBF_GIGE_CONTROL_EN_SPECIFIC_MAC            BIT(4)
+#define MLXBF_GIGE_CONTROL_CLEAN_PORT_EN              BIT(31)
+#define MLXBF_GIGE_RX_WQ_BASE                         0x0200
+#define MLXBF_GIGE_RX_WQE_SIZE_LOG2                   0x0208
+#define MLXBF_GIGE_RX_WQE_SIZE_LOG2_RESET_VAL         7
+#define MLXBF_GIGE_RX_CQ_BASE                         0x0210
+#define MLXBF_GIGE_TX_WQ_BASE                         0x0218
+#define MLXBF_GIGE_TX_WQ_SIZE_LOG2                    0x0220
+#define MLXBF_GIGE_TX_WQ_SIZE_LOG2_RESET_VAL          7
+#define MLXBF_GIGE_TX_CI_UPDATE_ADDRESS               0x0228
+#define MLXBF_GIGE_RX_WQE_PI                          0x0230
+#define MLXBF_GIGE_TX_PRODUCER_INDEX                  0x0238
+#define MLXBF_GIGE_RX_MAC_FILTER                      0x0240
+#define MLXBF_GIGE_RX_MAC_FILTER_STRIDE               0x0008
+#define MLXBF_GIGE_RX_DIN_DROP_COUNTER                0x0260
+#define MLXBF_GIGE_TX_CONSUMER_INDEX                  0x0310
+#define MLXBF_GIGE_TX_CONTROL                         0x0318
+#define MLXBF_GIGE_TX_CONTROL_GRACEFUL_STOP           BIT(0)
+#define MLXBF_GIGE_TX_STATUS                          0x0388
+#define MLXBF_GIGE_TX_STATUS_DATA_FIFO_FULL           BIT(1)
+#define MLXBF_GIGE_RX_MAC_FILTER_DMAC_RANGE_START     0x0520
+#define MLXBF_GIGE_RX_MAC_FILTER_DMAC_RANGE_END       0x0528
+#define MLXBF_GIGE_RX_MAC_FILTER_COUNT_DISC           0x0540
+#define MLXBF_GIGE_RX_MAC_FILTER_COUNT_DISC_EN        BIT(0)
+#define MLXBF_GIGE_RX_MAC_FILTER_COUNT_PASS           0x0548
+#define MLXBF_GIGE_RX_MAC_FILTER_COUNT_PASS_EN        BIT(0)
+#define MLXBF_GIGE_RX_PASS_COUNTER_ALL                0x0550
+#define MLXBF_GIGE_RX_DISC_COUNTER_ALL                0x0560
+#define MLXBF_GIGE_RX                                 0x0578
+#define MLXBF_GIGE_RX_STRIP_CRC_EN                    BIT(1)
+#define MLXBF_GIGE_RX_DMA                             0x0580
+#define MLXBF_GIGE_RX_DMA_EN                          BIT(0)
+#define MLXBF_GIGE_RX_CQE_PACKET_CI                   0x05b0
+#define MLXBF_GIGE_MAC_CFG                            0x05e8
+
+/* NOTE: MLXBF_GIGE_MAC_CFG is the last defined register offset,
+ * so use that plus size of single register to derive total size
+ */
+#define MLXBF_GIGE_MMIO_REG_SZ                        (MLXBF_GIGE_MAC_CFG + 8)
+
+#endif /* !defined(__MLXBF_GIGE_REGS_H__) */
diff --git a/drivers/net/ethernet/mellanox/mlxbf_gige/mlxbf_gige_rx.c b/drivers/net/ethernet/mellanox/mlxbf_gige/mlxbf_gige_rx.c
new file mode 100644
index 000000000..699984358
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxbf_gige/mlxbf_gige_rx.c
@@ -0,0 +1,326 @@
+// SPDX-License-Identifier: GPL-2.0-only OR BSD-3-Clause
+
+/* Packet receive logic for Mellanox Gigabit Ethernet driver
+ *
+ * Copyright (C) 2020-2021 NVIDIA CORPORATION & AFFILIATES
+ */
+
+#include <linux/etherdevice.h>
+#include <linux/skbuff.h>
+
+#include "mlxbf_gige.h"
+#include "mlxbf_gige_regs.h"
+
+void mlxbf_gige_set_mac_rx_filter(struct mlxbf_gige *priv,
+				  unsigned int index, u64 dmac)
+{
+	void __iomem *base = priv->base;
+	u64 control;
+
+	/* Write destination MAC to specified MAC RX filter */
+	writeq(dmac, base + MLXBF_GIGE_RX_MAC_FILTER +
+	       (index * MLXBF_GIGE_RX_MAC_FILTER_STRIDE));
+
+	/* Enable MAC receive filter mask for specified index */
+	control = readq(base + MLXBF_GIGE_CONTROL);
+	control |= (MLXBF_GIGE_CONTROL_EN_SPECIFIC_MAC << index);
+	writeq(control, base + MLXBF_GIGE_CONTROL);
+}
+
+void mlxbf_gige_get_mac_rx_filter(struct mlxbf_gige *priv,
+				  unsigned int index, u64 *dmac)
+{
+	void __iomem *base = priv->base;
+
+	/* Read destination MAC from specified MAC RX filter */
+	*dmac = readq(base + MLXBF_GIGE_RX_MAC_FILTER +
+		      (index * MLXBF_GIGE_RX_MAC_FILTER_STRIDE));
+}
+
+void mlxbf_gige_enable_promisc(struct mlxbf_gige *priv)
+{
+	void __iomem *base = priv->base;
+	u64 control;
+	u64 end_mac;
+
+	/* Enable MAC_ID_RANGE match functionality */
+	control = readq(base + MLXBF_GIGE_CONTROL);
+	control |= MLXBF_GIGE_CONTROL_MAC_ID_RANGE_EN;
+	writeq(control, base + MLXBF_GIGE_CONTROL);
+
+	/* Set start of destination MAC range check to 0 */
+	writeq(0, base + MLXBF_GIGE_RX_MAC_FILTER_DMAC_RANGE_START);
+
+	/* Set end of destination MAC range check to all FFs */
+	end_mac = BCAST_MAC_ADDR;
+	writeq(end_mac, base + MLXBF_GIGE_RX_MAC_FILTER_DMAC_RANGE_END);
+}
+
+void mlxbf_gige_disable_promisc(struct mlxbf_gige *priv)
+{
+	void __iomem *base = priv->base;
+	u64 control;
+
+	/* Disable MAC_ID_RANGE match functionality */
+	control = readq(base + MLXBF_GIGE_CONTROL);
+	control &= ~MLXBF_GIGE_CONTROL_MAC_ID_RANGE_EN;
+	writeq(control, base + MLXBF_GIGE_CONTROL);
+
+	/* NOTE: no need to change DMAC_RANGE_START or END;
+	 * those values are ignored since MAC_ID_RANGE_EN=0
+	 */
+}
+
+/* Receive Initialization
+ * 1) Configures RX MAC filters via MMIO registers
+ * 2) Allocates RX WQE array using coherent DMA mapping
+ * 3) Initializes each element of RX WQE array with a receive
+ *    buffer pointer (also using coherent DMA mapping)
+ * 4) Allocates RX CQE array using coherent DMA mapping
+ * 5) Completes other misc receive initialization
+ */
+int mlxbf_gige_rx_init(struct mlxbf_gige *priv)
+{
+	size_t wq_size, cq_size;
+	dma_addr_t *rx_wqe_ptr;
+	dma_addr_t rx_buf_dma;
+	u64 data;
+	int i, j;
+
+	/* Configure MAC RX filter #0 to allow RX of broadcast pkts */
+	mlxbf_gige_set_mac_rx_filter(priv, MLXBF_GIGE_BCAST_MAC_FILTER_IDX,
+				     BCAST_MAC_ADDR);
+
+	wq_size = MLXBF_GIGE_RX_WQE_SZ * priv->rx_q_entries;
+	priv->rx_wqe_base = dma_alloc_coherent(priv->dev, wq_size,
+					       &priv->rx_wqe_base_dma,
+					       GFP_KERNEL);
+	if (!priv->rx_wqe_base)
+		return -ENOMEM;
+
+	/* Initialize 'rx_wqe_ptr' to point to first RX WQE in array
+	 * Each RX WQE is simply a receive buffer pointer, so walk
+	 * the entire array, allocating a 2KB buffer for each element
+	 */
+	rx_wqe_ptr = priv->rx_wqe_base;
+
+	for (i = 0; i < priv->rx_q_entries; i++) {
+		priv->rx_skb[i] = mlxbf_gige_alloc_skb(priv, MLXBF_GIGE_DEFAULT_BUF_SZ,
+						       &rx_buf_dma, DMA_FROM_DEVICE);
+		if (!priv->rx_skb[i])
+			goto free_wqe_and_skb;
+		*rx_wqe_ptr++ = rx_buf_dma;
+	}
+
+	/* Write RX WQE base address into MMIO reg */
+	writeq(priv->rx_wqe_base_dma, priv->base + MLXBF_GIGE_RX_WQ_BASE);
+
+	cq_size = MLXBF_GIGE_RX_CQE_SZ * priv->rx_q_entries;
+	priv->rx_cqe_base = dma_alloc_coherent(priv->dev, cq_size,
+					       &priv->rx_cqe_base_dma,
+					       GFP_KERNEL);
+	if (!priv->rx_cqe_base)
+		goto free_wqe_and_skb;
+
+	for (i = 0; i < priv->rx_q_entries; i++)
+		priv->rx_cqe_base[i] |= MLXBF_GIGE_RX_CQE_VALID_MASK;
+
+	/* Write RX CQE base address into MMIO reg */
+	writeq(priv->rx_cqe_base_dma, priv->base + MLXBF_GIGE_RX_CQ_BASE);
+
+	/* Write RX_WQE_PI with current number of replenished buffers */
+	writeq(priv->rx_q_entries, priv->base + MLXBF_GIGE_RX_WQE_PI);
+
+	/* Enable removal of CRC during RX */
+	data = readq(priv->base + MLXBF_GIGE_RX);
+	data |= MLXBF_GIGE_RX_STRIP_CRC_EN;
+	writeq(data, priv->base + MLXBF_GIGE_RX);
+
+	/* Enable RX MAC filter pass and discard counters */
+	writeq(MLXBF_GIGE_RX_MAC_FILTER_COUNT_DISC_EN,
+	       priv->base + MLXBF_GIGE_RX_MAC_FILTER_COUNT_DISC);
+	writeq(MLXBF_GIGE_RX_MAC_FILTER_COUNT_PASS_EN,
+	       priv->base + MLXBF_GIGE_RX_MAC_FILTER_COUNT_PASS);
+
+	writeq(ilog2(priv->rx_q_entries),
+	       priv->base + MLXBF_GIGE_RX_WQE_SIZE_LOG2);
+
+	/* Clear MLXBF_GIGE_INT_MASK 'receive pkt' bit to
+	 * indicate readiness to receive interrupts
+	 */
+	data = readq(priv->base + MLXBF_GIGE_INT_MASK);
+	data &= ~MLXBF_GIGE_INT_MASK_RX_RECEIVE_PACKET;
+	writeq(data, priv->base + MLXBF_GIGE_INT_MASK);
+
+	/* Enable RX DMA to write new packets to memory */
+	data = readq(priv->base + MLXBF_GIGE_RX_DMA);
+	data |= MLXBF_GIGE_RX_DMA_EN;
+	writeq(data, priv->base + MLXBF_GIGE_RX_DMA);
+
+	return 0;
+
+free_wqe_and_skb:
+	rx_wqe_ptr = priv->rx_wqe_base;
+	for (j = 0; j < i; j++) {
+		dma_unmap_single(priv->dev, *rx_wqe_ptr,
+				 MLXBF_GIGE_DEFAULT_BUF_SZ, DMA_FROM_DEVICE);
+		dev_kfree_skb(priv->rx_skb[j]);
+		rx_wqe_ptr++;
+	}
+	dma_free_coherent(priv->dev, wq_size,
+			  priv->rx_wqe_base, priv->rx_wqe_base_dma);
+	return -ENOMEM;
+}
+
+/* Receive Deinitialization
+ * This routine will free allocations done by mlxbf_gige_rx_init(),
+ * namely the RX WQE and RX CQE arrays, as well as all RX buffers
+ */
+void mlxbf_gige_rx_deinit(struct mlxbf_gige *priv)
+{
+	dma_addr_t *rx_wqe_ptr;
+	size_t size;
+	u64 data;
+	int i;
+
+	/* Disable RX DMA to prevent packet transfers to memory */
+	data = readq(priv->base + MLXBF_GIGE_RX_DMA);
+	data &= ~MLXBF_GIGE_RX_DMA_EN;
+	writeq(data, priv->base + MLXBF_GIGE_RX_DMA);
+
+	rx_wqe_ptr = priv->rx_wqe_base;
+
+	for (i = 0; i < priv->rx_q_entries; i++) {
+		dma_unmap_single(priv->dev, *rx_wqe_ptr, MLXBF_GIGE_DEFAULT_BUF_SZ,
+				 DMA_FROM_DEVICE);
+		dev_kfree_skb(priv->rx_skb[i]);
+		rx_wqe_ptr++;
+	}
+
+	size = MLXBF_GIGE_RX_WQE_SZ * priv->rx_q_entries;
+	dma_free_coherent(priv->dev, size,
+			  priv->rx_wqe_base, priv->rx_wqe_base_dma);
+
+	size = MLXBF_GIGE_RX_CQE_SZ * priv->rx_q_entries;
+	dma_free_coherent(priv->dev, size,
+			  priv->rx_cqe_base, priv->rx_cqe_base_dma);
+
+	priv->rx_wqe_base = NULL;
+	priv->rx_wqe_base_dma = 0;
+	priv->rx_cqe_base = NULL;
+	priv->rx_cqe_base_dma = 0;
+	writeq(0, priv->base + MLXBF_GIGE_RX_WQ_BASE);
+	writeq(0, priv->base + MLXBF_GIGE_RX_CQ_BASE);
+}
+
+static bool mlxbf_gige_rx_packet(struct mlxbf_gige *priv, int *rx_pkts)
+{
+	struct net_device *netdev = priv->netdev;
+	struct sk_buff *skb = NULL, *rx_skb;
+	u16 rx_pi_rem, rx_ci_rem;
+	dma_addr_t *rx_wqe_addr;
+	dma_addr_t rx_buf_dma;
+	u64 *rx_cqe_addr;
+	u64 datalen;
+	u64 rx_cqe;
+	u16 rx_ci;
+	u16 rx_pi;
+
+	/* Index into RX buffer array is rx_pi w/wrap based on RX_CQE_SIZE */
+	rx_pi = readq(priv->base + MLXBF_GIGE_RX_WQE_PI);
+	rx_pi_rem = rx_pi % priv->rx_q_entries;
+
+	rx_wqe_addr = priv->rx_wqe_base + rx_pi_rem;
+	rx_cqe_addr = priv->rx_cqe_base + rx_pi_rem;
+	rx_cqe = *rx_cqe_addr;
+
+	if ((!!(rx_cqe & MLXBF_GIGE_RX_CQE_VALID_MASK)) != priv->valid_polarity)
+		return false;
+
+	if ((rx_cqe & MLXBF_GIGE_RX_CQE_PKT_STATUS_MASK) == 0) {
+		/* Packet is OK, increment stats */
+		datalen = rx_cqe & MLXBF_GIGE_RX_CQE_PKT_LEN_MASK;
+		netdev->stats.rx_packets++;
+		netdev->stats.rx_bytes += datalen;
+
+		skb = priv->rx_skb[rx_pi_rem];
+
+		/* Alloc another RX SKB for this same index */
+		rx_skb = mlxbf_gige_alloc_skb(priv, MLXBF_GIGE_DEFAULT_BUF_SZ,
+					      &rx_buf_dma, DMA_FROM_DEVICE);
+		if (!rx_skb)
+			return false;
+		priv->rx_skb[rx_pi_rem] = rx_skb;
+		dma_unmap_single(priv->dev, *rx_wqe_addr,
+				 MLXBF_GIGE_DEFAULT_BUF_SZ, DMA_FROM_DEVICE);
+
+		skb_put(skb, datalen);
+
+		skb->ip_summed = CHECKSUM_NONE; /* device did not checksum packet */
+
+		skb->protocol = eth_type_trans(skb, netdev);
+
+		*rx_wqe_addr = rx_buf_dma;
+	} else if (rx_cqe & MLXBF_GIGE_RX_CQE_PKT_STATUS_MAC_ERR) {
+		priv->stats.rx_mac_errors++;
+	} else if (rx_cqe & MLXBF_GIGE_RX_CQE_PKT_STATUS_TRUNCATED) {
+		priv->stats.rx_truncate_errors++;
+	}
+
+	/* Read receive consumer index before replenish so that this routine
+	 * returns accurate return value even if packet is received into
+	 * just-replenished buffer prior to exiting this routine.
+	 */
+	rx_ci = readq(priv->base + MLXBF_GIGE_RX_CQE_PACKET_CI);
+	rx_ci_rem = rx_ci % priv->rx_q_entries;
+
+	/* Let hardware know we've replenished one buffer */
+	rx_pi++;
+
+	/* Ensure completion of all writes before notifying HW of replenish */
+	wmb();
+	writeq(rx_pi, priv->base + MLXBF_GIGE_RX_WQE_PI);
+
+	(*rx_pkts)++;
+
+	rx_pi_rem = rx_pi % priv->rx_q_entries;
+	if (rx_pi_rem == 0)
+		priv->valid_polarity ^= 1;
+
+	if (skb)
+		netif_receive_skb(skb);
+
+	return rx_pi_rem != rx_ci_rem;
+}
+
+/* Driver poll() function called by NAPI infrastructure */
+int mlxbf_gige_poll(struct napi_struct *napi, int budget)
+{
+	struct mlxbf_gige *priv;
+	bool remaining_pkts;
+	int work_done = 0;
+	u64 data;
+
+	priv = container_of(napi, struct mlxbf_gige, napi);
+
+	mlxbf_gige_handle_tx_complete(priv);
+
+	do {
+		remaining_pkts = mlxbf_gige_rx_packet(priv, &work_done);
+	} while (remaining_pkts && work_done < budget);
+
+	/* If amount of work done < budget, turn off NAPI polling
+	 * via napi_complete_done(napi, work_done) and then
+	 * re-enable interrupts.
+	 */
+	if (work_done < budget && napi_complete_done(napi, work_done)) {
+		/* Clear MLXBF_GIGE_INT_MASK 'receive pkt' bit to
+		 * indicate receive readiness
+		 */
+		data = readq(priv->base + MLXBF_GIGE_INT_MASK);
+		data &= ~MLXBF_GIGE_INT_MASK_RX_RECEIVE_PACKET;
+		writeq(data, priv->base + MLXBF_GIGE_INT_MASK);
+	}
+
+	return work_done;
+}
diff --git a/drivers/net/ethernet/mellanox/mlxbf_gige/mlxbf_gige_tx.c b/drivers/net/ethernet/mellanox/mlxbf_gige/mlxbf_gige_tx.c
new file mode 100644
index 000000000..04982e888
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxbf_gige/mlxbf_gige_tx.c
@@ -0,0 +1,284 @@
+// SPDX-License-Identifier: GPL-2.0-only OR BSD-3-Clause
+
+/* Packet transmit logic for Mellanox Gigabit Ethernet driver
+ *
+ * Copyright (C) 2020-2021 NVIDIA CORPORATION & AFFILIATES
+ */
+
+#include <linux/skbuff.h>
+
+#include "mlxbf_gige.h"
+#include "mlxbf_gige_regs.h"
+
+/* Transmit Initialization
+ * 1) Allocates TX WQE array using coherent DMA mapping
+ * 2) Allocates TX completion counter using coherent DMA mapping
+ */
+int mlxbf_gige_tx_init(struct mlxbf_gige *priv)
+{
+	size_t size;
+
+	size = MLXBF_GIGE_TX_WQE_SZ * priv->tx_q_entries;
+	priv->tx_wqe_base = dma_alloc_coherent(priv->dev, size,
+					       &priv->tx_wqe_base_dma,
+					       GFP_KERNEL);
+	if (!priv->tx_wqe_base)
+		return -ENOMEM;
+
+	priv->tx_wqe_next = priv->tx_wqe_base;
+
+	/* Write TX WQE base address into MMIO reg */
+	writeq(priv->tx_wqe_base_dma, priv->base + MLXBF_GIGE_TX_WQ_BASE);
+
+	/* Allocate address for TX completion count */
+	priv->tx_cc = dma_alloc_coherent(priv->dev, MLXBF_GIGE_TX_CC_SZ,
+					 &priv->tx_cc_dma, GFP_KERNEL);
+	if (!priv->tx_cc) {
+		dma_free_coherent(priv->dev, size,
+				  priv->tx_wqe_base, priv->tx_wqe_base_dma);
+		return -ENOMEM;
+	}
+
+	/* Write TX CC base address into MMIO reg */
+	writeq(priv->tx_cc_dma, priv->base + MLXBF_GIGE_TX_CI_UPDATE_ADDRESS);
+
+	writeq(ilog2(priv->tx_q_entries),
+	       priv->base + MLXBF_GIGE_TX_WQ_SIZE_LOG2);
+
+	priv->prev_tx_ci = 0;
+	priv->tx_pi = 0;
+
+	return 0;
+}
+
+/* Transmit Deinitialization
+ * This routine will free allocations done by mlxbf_gige_tx_init(),
+ * namely the TX WQE array and the TX completion counter
+ */
+void mlxbf_gige_tx_deinit(struct mlxbf_gige *priv)
+{
+	u64 *tx_wqe_addr;
+	size_t size;
+	int i;
+
+	tx_wqe_addr = priv->tx_wqe_base;
+
+	for (i = 0; i < priv->tx_q_entries; i++) {
+		if (priv->tx_skb[i]) {
+			dma_unmap_single(priv->dev, *tx_wqe_addr,
+					 priv->tx_skb[i]->len, DMA_TO_DEVICE);
+			dev_kfree_skb(priv->tx_skb[i]);
+			priv->tx_skb[i] = NULL;
+		}
+		tx_wqe_addr += 2;
+	}
+
+	size = MLXBF_GIGE_TX_WQE_SZ * priv->tx_q_entries;
+	dma_free_coherent(priv->dev, size,
+			  priv->tx_wqe_base, priv->tx_wqe_base_dma);
+
+	dma_free_coherent(priv->dev, MLXBF_GIGE_TX_CC_SZ,
+			  priv->tx_cc, priv->tx_cc_dma);
+
+	priv->tx_wqe_base = NULL;
+	priv->tx_wqe_base_dma = 0;
+	priv->tx_cc = NULL;
+	priv->tx_cc_dma = 0;
+	priv->tx_wqe_next = NULL;
+	writeq(0, priv->base + MLXBF_GIGE_TX_WQ_BASE);
+	writeq(0, priv->base + MLXBF_GIGE_TX_CI_UPDATE_ADDRESS);
+}
+
+/* Function that returns status of TX ring:
+ *          0: TX ring is full, i.e. there are no
+ *             available un-used entries in TX ring.
+ *   non-null: TX ring is not full, i.e. there are
+ *             some available entries in TX ring.
+ *             The non-null value is a measure of
+ *             how many TX entries are available, but
+ *             it is not the exact number of available
+ *             entries (see below).
+ *
+ * The algorithm makes the assumption that if
+ * (prev_tx_ci == tx_pi) then the TX ring is empty.
+ * An empty ring actually has (tx_q_entries-1)
+ * entries, which allows the algorithm to differentiate
+ * the case of an empty ring vs. a full ring.
+ */
+static u16 mlxbf_gige_tx_buffs_avail(struct mlxbf_gige *priv)
+{
+	unsigned long flags;
+	u16 avail;
+
+	spin_lock_irqsave(&priv->lock, flags);
+
+	if (priv->prev_tx_ci == priv->tx_pi)
+		avail = priv->tx_q_entries - 1;
+	else
+		avail = ((priv->tx_q_entries + priv->prev_tx_ci - priv->tx_pi)
+			  % priv->tx_q_entries) - 1;
+
+	spin_unlock_irqrestore(&priv->lock, flags);
+
+	return avail;
+}
+
+bool mlxbf_gige_handle_tx_complete(struct mlxbf_gige *priv)
+{
+	struct net_device_stats *stats;
+	u16 tx_wqe_index;
+	u64 *tx_wqe_addr;
+	u64 tx_status;
+	u16 tx_ci;
+
+	tx_status = readq(priv->base + MLXBF_GIGE_TX_STATUS);
+	if (tx_status & MLXBF_GIGE_TX_STATUS_DATA_FIFO_FULL)
+		priv->stats.tx_fifo_full++;
+	tx_ci = readq(priv->base + MLXBF_GIGE_TX_CONSUMER_INDEX);
+	stats = &priv->netdev->stats;
+
+	/* Transmit completion logic needs to loop until the completion
+	 * index (in SW) equals TX consumer index (from HW).  These
+	 * parameters are unsigned 16-bit values and the wrap case needs
+	 * to be supported, that is TX consumer index wrapped from 0xFFFF
+	 * to 0 while TX completion index is still < 0xFFFF.
+	 */
+	for (; priv->prev_tx_ci != tx_ci; priv->prev_tx_ci++) {
+		tx_wqe_index = priv->prev_tx_ci % priv->tx_q_entries;
+		/* Each TX WQE is 16 bytes. The 8 MSB store the 2KB TX
+		 * buffer address and the 8 LSB contain information
+		 * about the TX WQE.
+		 */
+		tx_wqe_addr = priv->tx_wqe_base +
+			       (tx_wqe_index * MLXBF_GIGE_TX_WQE_SZ_QWORDS);
+
+		stats->tx_packets++;
+		stats->tx_bytes += MLXBF_GIGE_TX_WQE_PKT_LEN(tx_wqe_addr);
+
+		dma_unmap_single(priv->dev, *tx_wqe_addr,
+				 priv->tx_skb[tx_wqe_index]->len, DMA_TO_DEVICE);
+		dev_consume_skb_any(priv->tx_skb[tx_wqe_index]);
+		priv->tx_skb[tx_wqe_index] = NULL;
+
+		/* Ensure completion of updates across all cores */
+		mb();
+	}
+
+	/* Since the TX ring was likely just drained, check if TX queue
+	 * had previously been stopped and now that there are TX buffers
+	 * available the TX queue can be awakened.
+	 */
+	if (netif_queue_stopped(priv->netdev) &&
+	    mlxbf_gige_tx_buffs_avail(priv))
+		netif_wake_queue(priv->netdev);
+
+	return true;
+}
+
+/* Function to advance the tx_wqe_next pointer to next TX WQE */
+void mlxbf_gige_update_tx_wqe_next(struct mlxbf_gige *priv)
+{
+	/* Advance tx_wqe_next pointer */
+	priv->tx_wqe_next += MLXBF_GIGE_TX_WQE_SZ_QWORDS;
+
+	/* Check if 'next' pointer is beyond end of TX ring */
+	/* If so, set 'next' back to 'base' pointer of ring */
+	if (priv->tx_wqe_next == (priv->tx_wqe_base +
+				  (priv->tx_q_entries * MLXBF_GIGE_TX_WQE_SZ_QWORDS)))
+		priv->tx_wqe_next = priv->tx_wqe_base;
+}
+
+netdev_tx_t mlxbf_gige_start_xmit(struct sk_buff *skb,
+				  struct net_device *netdev)
+{
+	struct mlxbf_gige *priv = netdev_priv(netdev);
+	long buff_addr, start_dma_page, end_dma_page;
+	struct sk_buff *tx_skb;
+	dma_addr_t tx_buf_dma;
+	unsigned long flags;
+	u64 *tx_wqe_addr;
+	u64 word2;
+
+	/* If needed, linearize TX SKB as hardware DMA expects this */
+	if (skb->len > MLXBF_GIGE_DEFAULT_BUF_SZ || skb_linearize(skb)) {
+		dev_kfree_skb(skb);
+		netdev->stats.tx_dropped++;
+		return NETDEV_TX_OK;
+	}
+
+	buff_addr = (long)skb->data;
+	start_dma_page = buff_addr >> MLXBF_GIGE_DMA_PAGE_SHIFT;
+	end_dma_page   = (buff_addr + skb->len - 1) >> MLXBF_GIGE_DMA_PAGE_SHIFT;
+
+	/* Verify that payload pointer and data length of SKB to be
+	 * transmitted does not violate the hardware DMA limitation.
+	 */
+	if (start_dma_page != end_dma_page) {
+		/* DMA operation would fail as-is, alloc new aligned SKB */
+		tx_skb = mlxbf_gige_alloc_skb(priv, skb->len,
+					      &tx_buf_dma, DMA_TO_DEVICE);
+		if (!tx_skb) {
+			/* Free original skb, could not alloc new aligned SKB */
+			dev_kfree_skb(skb);
+			netdev->stats.tx_dropped++;
+			return NETDEV_TX_OK;
+		}
+
+		skb_put_data(tx_skb, skb->data, skb->len);
+
+		/* Free the original SKB */
+		dev_kfree_skb(skb);
+	} else {
+		tx_skb = skb;
+		tx_buf_dma = dma_map_single(priv->dev, skb->data,
+					    skb->len, DMA_TO_DEVICE);
+		if (dma_mapping_error(priv->dev, tx_buf_dma)) {
+			dev_kfree_skb(skb);
+			netdev->stats.tx_dropped++;
+			return NETDEV_TX_OK;
+		}
+	}
+
+	/* Get address of TX WQE */
+	tx_wqe_addr = priv->tx_wqe_next;
+
+	mlxbf_gige_update_tx_wqe_next(priv);
+
+	/* Put PA of buffer address into first 64-bit word of TX WQE */
+	*tx_wqe_addr = tx_buf_dma;
+
+	/* Set TX WQE pkt_len appropriately
+	 * NOTE: GigE silicon will automatically pad up to
+	 *       minimum packet length if needed.
+	 */
+	word2 = tx_skb->len & MLXBF_GIGE_TX_WQE_PKT_LEN_MASK;
+
+	/* Write entire 2nd word of TX WQE */
+	*(tx_wqe_addr + 1) = word2;
+
+	spin_lock_irqsave(&priv->lock, flags);
+	priv->tx_skb[priv->tx_pi % priv->tx_q_entries] = tx_skb;
+	priv->tx_pi++;
+	spin_unlock_irqrestore(&priv->lock, flags);
+
+	if (!netdev_xmit_more()) {
+		/* Create memory barrier before write to TX PI */
+		wmb();
+		writeq(priv->tx_pi, priv->base + MLXBF_GIGE_TX_PRODUCER_INDEX);
+	}
+
+	/* Check if the last TX entry was just used */
+	if (!mlxbf_gige_tx_buffs_avail(priv)) {
+		/* TX ring is full, inform stack */
+		netif_stop_queue(netdev);
+
+		/* Since there is no separate "TX complete" interrupt, need
+		 * to explicitly schedule NAPI poll.  This will trigger logic
+		 * which processes TX completions, and will hopefully drain
+		 * the TX ring allowing the TX queue to be awakened.
+		 */
+		napi_schedule(&priv->napi);
+	}
+
+	return NETDEV_TX_OK;
+}
diff --git a/drivers/net/ethernet/mellanox/mlxfw/Kconfig b/drivers/net/ethernet/mellanox/mlxfw/Kconfig
new file mode 100644
index 000000000..c339f3c46
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxfw/Kconfig
@@ -0,0 +1,15 @@
+# SPDX-License-Identifier: GPL-2.0-only
+#
+# Mellanox firmware flash library configuration
+#
+
+config MLXFW
+	tristate "Mellanox Technologies firmware flash module"
+	help
+	  This driver supports Mellanox Technologies Firmware
+	  flashing common logic.
+
+	  To compile this driver as a module, choose M here: the
+	  module will be called mlxfw.
+	select XZ_DEC
+	select NET_DEVLINK
diff --git a/drivers/net/ethernet/mellanox/mlxfw/Makefile b/drivers/net/ethernet/mellanox/mlxfw/Makefile
new file mode 100644
index 000000000..36007cd24
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxfw/Makefile
@@ -0,0 +1,3 @@
+# SPDX-License-Identifier: GPL-2.0-only
+obj-$(CONFIG_MLXFW)	+= mlxfw.o
+mlxfw-objs		:= mlxfw_fsm.o mlxfw_mfa2_tlv_multi.o mlxfw_mfa2.o
diff --git a/drivers/net/ethernet/mellanox/mlxfw/mlxfw.h b/drivers/net/ethernet/mellanox/mlxfw/mlxfw.h
new file mode 100644
index 000000000..e6475ea77
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxfw/mlxfw.h
@@ -0,0 +1,115 @@
+/* SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0 */
+/* Copyright (c) 2017-2019 Mellanox Technologies. All rights reserved */
+
+#ifndef _MLXFW_H
+#define _MLXFW_H
+
+#include <linux/firmware.h>
+#include <linux/netlink.h>
+#include <linux/device.h>
+#include <net/devlink.h>
+
+struct mlxfw_dev {
+	const struct mlxfw_dev_ops *ops;
+	const char *psid;
+	u16 psid_size;
+	struct devlink *devlink;
+};
+
+static inline
+struct device *mlxfw_dev_dev(struct mlxfw_dev *mlxfw_dev)
+{
+	return devlink_to_dev(mlxfw_dev->devlink);
+}
+
+#define MLXFW_PRFX "mlxfw: "
+
+#define mlxfw_info(mlxfw_dev, fmt, ...) \
+	dev_info(mlxfw_dev_dev(mlxfw_dev), MLXFW_PRFX fmt, ## __VA_ARGS__)
+#define mlxfw_err(mlxfw_dev, fmt, ...) \
+	dev_err(mlxfw_dev_dev(mlxfw_dev), MLXFW_PRFX fmt, ## __VA_ARGS__)
+#define mlxfw_dbg(mlxfw_dev, fmt, ...) \
+	dev_dbg(mlxfw_dev_dev(mlxfw_dev), MLXFW_PRFX fmt, ## __VA_ARGS__)
+
+enum mlxfw_fsm_state {
+	MLXFW_FSM_STATE_IDLE,
+	MLXFW_FSM_STATE_LOCKED,
+	MLXFW_FSM_STATE_INITIALIZE,
+	MLXFW_FSM_STATE_DOWNLOAD,
+	MLXFW_FSM_STATE_VERIFY,
+	MLXFW_FSM_STATE_APPLY,
+	MLXFW_FSM_STATE_ACTIVATE,
+};
+
+enum mlxfw_fsm_state_err {
+	MLXFW_FSM_STATE_ERR_OK,
+	MLXFW_FSM_STATE_ERR_ERROR,
+	MLXFW_FSM_STATE_ERR_REJECTED_DIGEST_ERR,
+	MLXFW_FSM_STATE_ERR_REJECTED_NOT_APPLICABLE,
+	MLXFW_FSM_STATE_ERR_REJECTED_UNKNOWN_KEY,
+	MLXFW_FSM_STATE_ERR_REJECTED_AUTH_FAILED,
+	MLXFW_FSM_STATE_ERR_REJECTED_UNSIGNED,
+	MLXFW_FSM_STATE_ERR_REJECTED_KEY_NOT_APPLICABLE,
+	MLXFW_FSM_STATE_ERR_REJECTED_BAD_FORMAT,
+	MLXFW_FSM_STATE_ERR_BLOCKED_PENDING_RESET,
+	MLXFW_FSM_STATE_ERR_MAX,
+};
+
+enum mlxfw_fsm_reactivate_status {
+	MLXFW_FSM_REACTIVATE_STATUS_OK,
+	MLXFW_FSM_REACTIVATE_STATUS_BUSY,
+	MLXFW_FSM_REACTIVATE_STATUS_PROHIBITED_FW_VER_ERR,
+	MLXFW_FSM_REACTIVATE_STATUS_FIRST_PAGE_COPY_FAILED,
+	MLXFW_FSM_REACTIVATE_STATUS_FIRST_PAGE_ERASE_FAILED,
+	MLXFW_FSM_REACTIVATE_STATUS_FIRST_PAGE_RESTORE_FAILED,
+	MLXFW_FSM_REACTIVATE_STATUS_CANDIDATE_FW_DEACTIVATION_FAILED,
+	MLXFW_FSM_REACTIVATE_STATUS_FW_ALREADY_ACTIVATED,
+	MLXFW_FSM_REACTIVATE_STATUS_ERR_DEVICE_RESET_REQUIRED,
+	MLXFW_FSM_REACTIVATE_STATUS_ERR_FW_PROGRAMMING_NEEDED,
+	MLXFW_FSM_REACTIVATE_STATUS_MAX,
+};
+
+struct mlxfw_dev_ops {
+	int (*component_query)(struct mlxfw_dev *mlxfw_dev, u16 component_index,
+			       u32 *p_max_size, u8 *p_align_bits,
+			       u16 *p_max_write_size);
+
+	int (*fsm_lock)(struct mlxfw_dev *mlxfw_dev, u32 *fwhandle);
+
+	int (*fsm_component_update)(struct mlxfw_dev *mlxfw_dev, u32 fwhandle,
+				    u16 component_index, u32 component_size);
+
+	int (*fsm_block_download)(struct mlxfw_dev *mlxfw_dev, u32 fwhandle,
+				  u8 *data, u16 size, u32 offset);
+
+	int (*fsm_component_verify)(struct mlxfw_dev *mlxfw_dev, u32 fwhandle,
+				    u16 component_index);
+
+	int (*fsm_activate)(struct mlxfw_dev *mlxfw_dev, u32 fwhandle);
+
+	int (*fsm_reactivate)(struct mlxfw_dev *mlxfw_dev, u8 *status);
+
+	int (*fsm_query_state)(struct mlxfw_dev *mlxfw_dev, u32 fwhandle,
+			       enum mlxfw_fsm_state *fsm_state,
+			       enum mlxfw_fsm_state_err *fsm_state_err);
+
+	void (*fsm_cancel)(struct mlxfw_dev *mlxfw_dev, u32 fwhandle);
+
+	void (*fsm_release)(struct mlxfw_dev *mlxfw_dev, u32 fwhandle);
+};
+
+#if IS_REACHABLE(CONFIG_MLXFW)
+int mlxfw_firmware_flash(struct mlxfw_dev *mlxfw_dev,
+			 const struct firmware *firmware,
+			 struct netlink_ext_ack *extack);
+#else
+static inline
+int mlxfw_firmware_flash(struct mlxfw_dev *mlxfw_dev,
+			 const struct firmware *firmware,
+			 struct netlink_ext_ack *extack)
+{
+	return -EOPNOTSUPP;
+}
+#endif
+
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlxfw/mlxfw_fsm.c b/drivers/net/ethernet/mellanox/mlxfw/mlxfw_fsm.c
new file mode 100644
index 000000000..46245e0b2
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxfw/mlxfw_fsm.c
@@ -0,0 +1,436 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2017-2019 Mellanox Technologies. All rights reserved */
+
+#define pr_fmt(fmt) "mlxfw: " fmt
+
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/delay.h>
+
+#include "mlxfw.h"
+#include "mlxfw_mfa2.h"
+
+#define MLXFW_FSM_STATE_WAIT_CYCLE_MS 200
+#define MLXFW_FSM_STATE_WAIT_TIMEOUT_MS 30000
+#define MLXFW_FSM_STATE_WAIT_ROUNDS \
+	(MLXFW_FSM_STATE_WAIT_TIMEOUT_MS / MLXFW_FSM_STATE_WAIT_CYCLE_MS)
+#define MLXFW_FSM_MAX_COMPONENT_SIZE (10 * (1 << 20))
+
+static const int mlxfw_fsm_state_errno[] = {
+	[MLXFW_FSM_STATE_ERR_ERROR] = -EIO,
+	[MLXFW_FSM_STATE_ERR_REJECTED_DIGEST_ERR] = -EBADMSG,
+	[MLXFW_FSM_STATE_ERR_REJECTED_NOT_APPLICABLE] = -ENOENT,
+	[MLXFW_FSM_STATE_ERR_REJECTED_UNKNOWN_KEY] = -ENOKEY,
+	[MLXFW_FSM_STATE_ERR_REJECTED_AUTH_FAILED] = -EACCES,
+	[MLXFW_FSM_STATE_ERR_REJECTED_UNSIGNED] = -EKEYREVOKED,
+	[MLXFW_FSM_STATE_ERR_REJECTED_KEY_NOT_APPLICABLE] = -EKEYREJECTED,
+	[MLXFW_FSM_STATE_ERR_REJECTED_BAD_FORMAT] = -ENOEXEC,
+	[MLXFW_FSM_STATE_ERR_BLOCKED_PENDING_RESET] = -EBUSY,
+	[MLXFW_FSM_STATE_ERR_MAX] = -EINVAL
+};
+
+#define MLXFW_ERR_PRFX "Firmware flash failed: "
+#define MLXFW_ERR_MSG(fwdev, extack, msg, err) do { \
+	mlxfw_err(fwdev, "%s, err (%d)\n", MLXFW_ERR_PRFX msg, err); \
+	NL_SET_ERR_MSG_MOD(extack, MLXFW_ERR_PRFX msg); \
+} while (0)
+
+static int mlxfw_fsm_state_err(struct mlxfw_dev *mlxfw_dev,
+			       struct netlink_ext_ack *extack,
+			       enum mlxfw_fsm_state_err err)
+{
+	enum mlxfw_fsm_state_err fsm_state_err;
+
+	fsm_state_err = min_t(enum mlxfw_fsm_state_err, err,
+			      MLXFW_FSM_STATE_ERR_MAX);
+
+	switch (fsm_state_err) {
+	case MLXFW_FSM_STATE_ERR_ERROR:
+		MLXFW_ERR_MSG(mlxfw_dev, extack, "general error", err);
+		break;
+	case MLXFW_FSM_STATE_ERR_REJECTED_DIGEST_ERR:
+		MLXFW_ERR_MSG(mlxfw_dev, extack, "component hash mismatch", err);
+		break;
+	case MLXFW_FSM_STATE_ERR_REJECTED_NOT_APPLICABLE:
+		MLXFW_ERR_MSG(mlxfw_dev, extack, "component not applicable", err);
+		break;
+	case MLXFW_FSM_STATE_ERR_REJECTED_UNKNOWN_KEY:
+		MLXFW_ERR_MSG(mlxfw_dev, extack, "unknown key", err);
+		break;
+	case MLXFW_FSM_STATE_ERR_REJECTED_AUTH_FAILED:
+		MLXFW_ERR_MSG(mlxfw_dev, extack, "authentication failed", err);
+		break;
+	case MLXFW_FSM_STATE_ERR_REJECTED_UNSIGNED:
+		MLXFW_ERR_MSG(mlxfw_dev, extack, "component was not signed", err);
+		break;
+	case MLXFW_FSM_STATE_ERR_REJECTED_KEY_NOT_APPLICABLE:
+		MLXFW_ERR_MSG(mlxfw_dev, extack, "key not applicable", err);
+		break;
+	case MLXFW_FSM_STATE_ERR_REJECTED_BAD_FORMAT:
+		MLXFW_ERR_MSG(mlxfw_dev, extack, "bad format", err);
+		break;
+	case MLXFW_FSM_STATE_ERR_BLOCKED_PENDING_RESET:
+		MLXFW_ERR_MSG(mlxfw_dev, extack, "pending reset", err);
+		break;
+	case MLXFW_FSM_STATE_ERR_OK:
+	case MLXFW_FSM_STATE_ERR_MAX:
+		MLXFW_ERR_MSG(mlxfw_dev, extack, "unknown error", err);
+		break;
+	}
+
+	return mlxfw_fsm_state_errno[fsm_state_err];
+};
+
+static int mlxfw_fsm_state_wait(struct mlxfw_dev *mlxfw_dev, u32 fwhandle,
+				enum mlxfw_fsm_state fsm_state,
+				struct netlink_ext_ack *extack)
+{
+	enum mlxfw_fsm_state_err fsm_state_err;
+	enum mlxfw_fsm_state curr_fsm_state;
+	int times;
+	int err;
+
+	times = MLXFW_FSM_STATE_WAIT_ROUNDS;
+retry:
+	err = mlxfw_dev->ops->fsm_query_state(mlxfw_dev, fwhandle,
+					      &curr_fsm_state, &fsm_state_err);
+	if (err) {
+		MLXFW_ERR_MSG(mlxfw_dev, extack, "FSM state query failed", err);
+		return err;
+	}
+
+	if (fsm_state_err != MLXFW_FSM_STATE_ERR_OK)
+		return mlxfw_fsm_state_err(mlxfw_dev, extack, fsm_state_err);
+
+	if (curr_fsm_state != fsm_state) {
+		if (--times == 0) {
+			MLXFW_ERR_MSG(mlxfw_dev, extack,
+				      "Timeout reached on FSM state change", -ETIMEDOUT);
+			return -ETIMEDOUT;
+		}
+		msleep(MLXFW_FSM_STATE_WAIT_CYCLE_MS);
+		goto retry;
+	}
+	return 0;
+}
+
+static int
+mlxfw_fsm_reactivate_err(struct mlxfw_dev *mlxfw_dev,
+			 struct netlink_ext_ack *extack, u8 err)
+{
+	enum mlxfw_fsm_reactivate_status status;
+
+#define MXFW_REACT_PRFX "Reactivate FSM: "
+#define MLXFW_REACT_ERR(msg, err) \
+	MLXFW_ERR_MSG(mlxfw_dev, extack, MXFW_REACT_PRFX msg, err)
+
+	status = min_t(enum mlxfw_fsm_reactivate_status, err,
+		       MLXFW_FSM_REACTIVATE_STATUS_MAX);
+
+	switch (status) {
+	case MLXFW_FSM_REACTIVATE_STATUS_BUSY:
+		MLXFW_REACT_ERR("busy", err);
+		break;
+	case MLXFW_FSM_REACTIVATE_STATUS_PROHIBITED_FW_VER_ERR:
+		MLXFW_REACT_ERR("prohibited fw ver", err);
+		break;
+	case MLXFW_FSM_REACTIVATE_STATUS_FIRST_PAGE_COPY_FAILED:
+		MLXFW_REACT_ERR("first page copy failed", err);
+		break;
+	case MLXFW_FSM_REACTIVATE_STATUS_FIRST_PAGE_ERASE_FAILED:
+		MLXFW_REACT_ERR("first page erase failed", err);
+		break;
+	case MLXFW_FSM_REACTIVATE_STATUS_FIRST_PAGE_RESTORE_FAILED:
+		MLXFW_REACT_ERR("first page restore failed", err);
+		break;
+	case MLXFW_FSM_REACTIVATE_STATUS_CANDIDATE_FW_DEACTIVATION_FAILED:
+		MLXFW_REACT_ERR("candidate fw deactivation failed", err);
+		break;
+	case MLXFW_FSM_REACTIVATE_STATUS_ERR_DEVICE_RESET_REQUIRED:
+		MLXFW_REACT_ERR("device reset required", err);
+		break;
+	case MLXFW_FSM_REACTIVATE_STATUS_ERR_FW_PROGRAMMING_NEEDED:
+		MLXFW_REACT_ERR("fw programming needed", err);
+		break;
+	case MLXFW_FSM_REACTIVATE_STATUS_FW_ALREADY_ACTIVATED:
+		MLXFW_REACT_ERR("fw already activated", err);
+		break;
+	case MLXFW_FSM_REACTIVATE_STATUS_OK:
+	case MLXFW_FSM_REACTIVATE_STATUS_MAX:
+		MLXFW_REACT_ERR("unexpected error", err);
+		break;
+	}
+	return -EREMOTEIO;
+};
+
+static int mlxfw_fsm_reactivate(struct mlxfw_dev *mlxfw_dev,
+				struct netlink_ext_ack *extack,
+				bool *supported)
+{
+	u8 status;
+	int err;
+
+	if (!mlxfw_dev->ops->fsm_reactivate)
+		return 0;
+
+	err = mlxfw_dev->ops->fsm_reactivate(mlxfw_dev, &status);
+	if (err == -EOPNOTSUPP) {
+		*supported = false;
+		return 0;
+	}
+
+	if (err) {
+		MLXFW_ERR_MSG(mlxfw_dev, extack,
+			      "Could not reactivate firmware flash", err);
+		return err;
+	}
+
+	if (status == MLXFW_FSM_REACTIVATE_STATUS_OK ||
+	    status == MLXFW_FSM_REACTIVATE_STATUS_FW_ALREADY_ACTIVATED)
+		return 0;
+
+	return mlxfw_fsm_reactivate_err(mlxfw_dev, extack, status);
+}
+
+static void mlxfw_status_notify(struct mlxfw_dev *mlxfw_dev,
+				const char *msg, const char *comp_name,
+				u32 done_bytes, u32 total_bytes)
+{
+	devlink_flash_update_status_notify(mlxfw_dev->devlink, msg, comp_name,
+					   done_bytes, total_bytes);
+}
+
+#define MLXFW_ALIGN_DOWN(x, align_bits) ((x) & ~((1 << (align_bits)) - 1))
+#define MLXFW_ALIGN_UP(x, align_bits) \
+		MLXFW_ALIGN_DOWN((x) + ((1 << (align_bits)) - 1), (align_bits))
+
+static int mlxfw_flash_component(struct mlxfw_dev *mlxfw_dev,
+				 u32 fwhandle,
+				 struct mlxfw_mfa2_component *comp,
+				 bool reactivate_supp,
+				 struct netlink_ext_ack *extack)
+{
+	u16 comp_max_write_size;
+	u8 comp_align_bits;
+	u32 comp_max_size;
+	char comp_name[8];
+	u16 block_size;
+	u8 *block_ptr;
+	u32 offset;
+	int err;
+
+	sprintf(comp_name, "%u", comp->index);
+
+	err = mlxfw_dev->ops->component_query(mlxfw_dev, comp->index,
+					      &comp_max_size, &comp_align_bits,
+					      &comp_max_write_size);
+	if (err) {
+		MLXFW_ERR_MSG(mlxfw_dev, extack, "FSM component query failed", err);
+		return err;
+	}
+
+	comp_max_size = min_t(u32, comp_max_size, MLXFW_FSM_MAX_COMPONENT_SIZE);
+	if (comp->data_size > comp_max_size) {
+		MLXFW_ERR_MSG(mlxfw_dev, extack,
+			      "Component size is bigger than limit", -EINVAL);
+		return -EINVAL;
+	}
+
+	comp_max_write_size = MLXFW_ALIGN_DOWN(comp_max_write_size,
+					       comp_align_bits);
+
+	mlxfw_dbg(mlxfw_dev, "Component update\n");
+	mlxfw_status_notify(mlxfw_dev, "Updating component", comp_name, 0, 0);
+	err = mlxfw_dev->ops->fsm_component_update(mlxfw_dev, fwhandle,
+						   comp->index,
+						   comp->data_size);
+	if (err) {
+		if (!reactivate_supp)
+			MLXFW_ERR_MSG(mlxfw_dev, extack,
+				      "FSM component update failed, FW reactivate is not supported",
+				      err);
+		else
+			MLXFW_ERR_MSG(mlxfw_dev, extack,
+				      "FSM component update failed", err);
+		return err;
+	}
+
+	err = mlxfw_fsm_state_wait(mlxfw_dev, fwhandle,
+				   MLXFW_FSM_STATE_DOWNLOAD, extack);
+	if (err)
+		goto err_out;
+
+	mlxfw_dbg(mlxfw_dev, "Component download\n");
+	mlxfw_status_notify(mlxfw_dev, "Downloading component",
+			    comp_name, 0, comp->data_size);
+	for (offset = 0;
+	     offset < MLXFW_ALIGN_UP(comp->data_size, comp_align_bits);
+	     offset += comp_max_write_size) {
+		block_ptr = comp->data + offset;
+		block_size = (u16) min_t(u32, comp->data_size - offset,
+					 comp_max_write_size);
+		err = mlxfw_dev->ops->fsm_block_download(mlxfw_dev, fwhandle,
+							 block_ptr, block_size,
+							 offset);
+		if (err) {
+			MLXFW_ERR_MSG(mlxfw_dev, extack,
+				      "Component download failed", err);
+			goto err_out;
+		}
+		mlxfw_status_notify(mlxfw_dev, "Downloading component",
+				    comp_name, offset + block_size,
+				    comp->data_size);
+	}
+
+	mlxfw_dbg(mlxfw_dev, "Component verify\n");
+	mlxfw_status_notify(mlxfw_dev, "Verifying component", comp_name, 0, 0);
+	err = mlxfw_dev->ops->fsm_component_verify(mlxfw_dev, fwhandle,
+						   comp->index);
+	if (err) {
+		MLXFW_ERR_MSG(mlxfw_dev, extack,
+			      "FSM component verify failed", err);
+		goto err_out;
+	}
+
+	err = mlxfw_fsm_state_wait(mlxfw_dev, fwhandle,
+				   MLXFW_FSM_STATE_LOCKED, extack);
+	if (err)
+		goto err_out;
+	return 0;
+
+err_out:
+	mlxfw_dev->ops->fsm_cancel(mlxfw_dev, fwhandle);
+	return err;
+}
+
+static int mlxfw_flash_components(struct mlxfw_dev *mlxfw_dev, u32 fwhandle,
+				  struct mlxfw_mfa2_file *mfa2_file,
+				  bool reactivate_supp,
+				  struct netlink_ext_ack *extack)
+{
+	u32 component_count;
+	int err;
+	int i;
+
+	err = mlxfw_mfa2_file_component_count(mfa2_file, mlxfw_dev->psid,
+					      mlxfw_dev->psid_size,
+					      &component_count);
+	if (err) {
+		MLXFW_ERR_MSG(mlxfw_dev, extack,
+			      "Could not find device PSID in MFA2 file", err);
+		return err;
+	}
+
+	for (i = 0; i < component_count; i++) {
+		struct mlxfw_mfa2_component *comp;
+
+		comp = mlxfw_mfa2_file_component_get(mfa2_file, mlxfw_dev->psid,
+						     mlxfw_dev->psid_size, i);
+		if (IS_ERR(comp)) {
+			err = PTR_ERR(comp);
+			MLXFW_ERR_MSG(mlxfw_dev, extack,
+				      "Failed to get MFA2 component", err);
+			return err;
+		}
+
+		mlxfw_info(mlxfw_dev, "Flashing component type %d\n",
+			   comp->index);
+		err = mlxfw_flash_component(mlxfw_dev, fwhandle, comp,
+					    reactivate_supp, extack);
+		mlxfw_mfa2_file_component_put(comp);
+		if (err)
+			return err;
+	}
+	return 0;
+}
+
+int mlxfw_firmware_flash(struct mlxfw_dev *mlxfw_dev,
+			 const struct firmware *firmware,
+			 struct netlink_ext_ack *extack)
+{
+	struct mlxfw_mfa2_file *mfa2_file;
+	bool reactivate_supp = true;
+	u32 fwhandle;
+	int err;
+
+	if (!mlxfw_mfa2_check(firmware)) {
+		MLXFW_ERR_MSG(mlxfw_dev, extack,
+			      "Firmware file is not MFA2", -EINVAL);
+		return -EINVAL;
+	}
+
+	mfa2_file = mlxfw_mfa2_file_init(firmware);
+	if (IS_ERR(mfa2_file)) {
+		err = PTR_ERR(mfa2_file);
+		MLXFW_ERR_MSG(mlxfw_dev, extack,
+			      "Failed to initialize MFA2 firmware file", err);
+		return err;
+	}
+
+	mlxfw_info(mlxfw_dev, "Initialize firmware flash process\n");
+	mlxfw_status_notify(mlxfw_dev, "Initializing firmware flash process",
+			    NULL, 0, 0);
+	err = mlxfw_dev->ops->fsm_lock(mlxfw_dev, &fwhandle);
+	if (err) {
+		MLXFW_ERR_MSG(mlxfw_dev, extack,
+			      "Could not lock the firmware FSM", err);
+		goto err_fsm_lock;
+	}
+
+	err = mlxfw_fsm_state_wait(mlxfw_dev, fwhandle,
+				   MLXFW_FSM_STATE_LOCKED, extack);
+	if (err)
+		goto err_state_wait_idle_to_locked;
+
+	err = mlxfw_fsm_reactivate(mlxfw_dev, extack, &reactivate_supp);
+	if (err)
+		goto err_fsm_reactivate;
+
+	err = mlxfw_fsm_state_wait(mlxfw_dev, fwhandle,
+				   MLXFW_FSM_STATE_LOCKED, extack);
+	if (err)
+		goto err_state_wait_reactivate_to_locked;
+
+	err = mlxfw_flash_components(mlxfw_dev, fwhandle, mfa2_file,
+				     reactivate_supp, extack);
+	if (err)
+		goto err_flash_components;
+
+	mlxfw_dbg(mlxfw_dev, "Activate image\n");
+	mlxfw_status_notify(mlxfw_dev, "Activating image", NULL, 0, 0);
+	err = mlxfw_dev->ops->fsm_activate(mlxfw_dev, fwhandle);
+	if (err) {
+		MLXFW_ERR_MSG(mlxfw_dev, extack,
+			      "Could not activate the downloaded image", err);
+		goto err_fsm_activate;
+	}
+
+	err = mlxfw_fsm_state_wait(mlxfw_dev, fwhandle,
+				   MLXFW_FSM_STATE_LOCKED, extack);
+	if (err)
+		goto err_state_wait_activate_to_locked;
+
+	mlxfw_dbg(mlxfw_dev, "Handle release\n");
+	mlxfw_dev->ops->fsm_release(mlxfw_dev, fwhandle);
+
+	mlxfw_info(mlxfw_dev, "Firmware flash done\n");
+	mlxfw_status_notify(mlxfw_dev, "Firmware flash done", NULL, 0, 0);
+	mlxfw_mfa2_file_fini(mfa2_file);
+	return 0;
+
+err_state_wait_activate_to_locked:
+err_fsm_activate:
+err_flash_components:
+err_state_wait_reactivate_to_locked:
+err_fsm_reactivate:
+err_state_wait_idle_to_locked:
+	mlxfw_dev->ops->fsm_release(mlxfw_dev, fwhandle);
+err_fsm_lock:
+	mlxfw_mfa2_file_fini(mfa2_file);
+	return err;
+}
+EXPORT_SYMBOL(mlxfw_firmware_flash);
+
+MODULE_LICENSE("Dual BSD/GPL");
+MODULE_AUTHOR("Yotam Gigi <yotamg@mellanox.com>");
+MODULE_DESCRIPTION("Mellanox firmware flash lib");
diff --git a/drivers/net/ethernet/mellanox/mlxfw/mlxfw_mfa2.c b/drivers/net/ethernet/mellanox/mlxfw/mlxfw_mfa2.c
new file mode 100644
index 000000000..e6f677e42
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxfw/mlxfw_mfa2.c
@@ -0,0 +1,589 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2017-2019 Mellanox Technologies. All rights reserved */
+
+#define pr_fmt(fmt) "mlxfw_mfa2: " fmt
+
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/netlink.h>
+#include <linux/vmalloc.h>
+#include <linux/xz.h>
+#include "mlxfw_mfa2.h"
+#include "mlxfw_mfa2_file.h"
+#include "mlxfw_mfa2_tlv.h"
+#include "mlxfw_mfa2_format.h"
+#include "mlxfw_mfa2_tlv_multi.h"
+
+/*               MFA2 FILE
+ *  +----------------------------------+
+ *  |        MFA2 finger print         |
+ *  +----------------------------------+
+ *  |   package descriptor multi_tlv   |
+ *  | +------------------------------+ |     +-----------------+
+ *  | |    package descriptor tlv    +-----> |num_devices=n    |
+ *  | +------------------------------+ |     |num_components=m |
+ *  +----------------------------------+     |CB offset        |
+ *  |    device descriptor multi_tlv   |     |...              |
+ *  | +------------------------------+ |     |                 |
+ *  | |           PSID tlv           | |     +-----------------+
+ *  | +------------------------------+ |
+ *  | |     component index tlv      | |
+ *  | +------------------------------+ |
+ *  +----------------------------------+
+ *  |  component descriptor multi_tlv  |
+ *  | +------------------------------+ |     +-----------------+
+ *  | |  component descriptor tlv    +-----> |Among others:    |
+ *  | +------------------------------+ |     |CB offset=o      |
+ *  +----------------------------------+     |comp index=i     |
+ *  |                                  |     |...              |
+ *  |                                  |     |                 |
+ *  |                                  |     +-----------------+
+ *  |        COMPONENT BLOCK (CB)      |
+ *  |                                  |
+ *  |                                  |
+ *  |                                  |
+ *  +----------------------------------+
+ *
+ * On the top level, an MFA2 file contains:
+ *  - Fingerprint
+ *  - Several multi_tlvs (TLVs of type MLXFW_MFA2_TLV_MULTI, as defined in
+ *    mlxfw_mfa2_format.h)
+ *  - Compresses content block
+ *
+ * The first multi_tlv
+ * -------------------
+ * The first multi TLV is treated as package descriptor, and expected to have a
+ * first TLV child of type MLXFW_MFA2_TLV_PACKAGE_DESCRIPTOR which contains all
+ * the global information needed to parse the file. Among others, it contains
+ * the number of device descriptors and component descriptor following this
+ * multi TLV.
+ *
+ * The device descriptor multi_tlv
+ * -------------------------------
+ * The multi TLVs following the package descriptor are treated as device
+ * descriptor, and are expected to have the following children:
+ *  - PSID TLV child of type MLXFW_MFA2_TLV_PSID containing that device PSID.
+ *  - Component index of type MLXFW_MFA2_TLV_COMPONENT_PTR that contains that
+ *    device component index.
+ *
+ * The component descriptor multi_tlv
+ * ----------------------------------
+ * The multi TLVs following the device descriptor multi TLVs are treated as
+ * component descriptor, and are expected to have a first child of type
+ * MLXFW_MFA2_TLV_COMPONENT_DESCRIPTOR that contains mostly the component index,
+ * needed for the flash process and the offset to the binary within the
+ * component block.
+ */
+
+static const u8 mlxfw_mfa2_fingerprint[] = "MLNX.MFA2.XZ.00!";
+static const int mlxfw_mfa2_fingerprint_len =
+			sizeof(mlxfw_mfa2_fingerprint) - 1;
+
+static const u8 mlxfw_mfa2_comp_magic[] = "#BIN.COMPONENT!#";
+static const int mlxfw_mfa2_comp_magic_len = sizeof(mlxfw_mfa2_comp_magic) - 1;
+
+bool mlxfw_mfa2_check(const struct firmware *fw)
+{
+	if (fw->size < sizeof(mlxfw_mfa2_fingerprint))
+		return false;
+
+	return memcmp(fw->data, mlxfw_mfa2_fingerprint,
+		      mlxfw_mfa2_fingerprint_len) == 0;
+}
+
+static bool
+mlxfw_mfa2_tlv_multi_validate(const struct mlxfw_mfa2_file *mfa2_file,
+			      const struct mlxfw_mfa2_tlv_multi *multi)
+{
+	const struct mlxfw_mfa2_tlv *tlv;
+	u16 idx;
+
+	/* Check that all children are valid */
+	mlxfw_mfa2_tlv_multi_foreach(mfa2_file, tlv, idx, multi) {
+		if (!tlv) {
+			pr_err("Multi has invalid child");
+			return false;
+		}
+	}
+	return true;
+}
+
+static bool
+mlxfw_mfa2_file_dev_validate(const struct mlxfw_mfa2_file *mfa2_file,
+			     const struct mlxfw_mfa2_tlv *dev_tlv,
+			     u16 dev_idx)
+{
+	const struct mlxfw_mfa2_tlv_component_ptr *cptr;
+	const struct mlxfw_mfa2_tlv_multi *multi;
+	const struct mlxfw_mfa2_tlv_psid *psid;
+	const struct mlxfw_mfa2_tlv *tlv;
+	u16 cptr_count;
+	u16 cptr_idx;
+	int err;
+
+	pr_debug("Device %d\n", dev_idx);
+
+	multi = mlxfw_mfa2_tlv_multi_get(mfa2_file, dev_tlv);
+	if (!multi) {
+		pr_err("Device %d is not a valid TLV error\n", dev_idx);
+		return false;
+	}
+
+	if (!mlxfw_mfa2_tlv_multi_validate(mfa2_file, multi))
+		return false;
+
+	/* Validate the device has PSID tlv */
+	tlv = mlxfw_mfa2_tlv_multi_child_find(mfa2_file, multi,
+					      MLXFW_MFA2_TLV_PSID, 0);
+	if (!tlv) {
+		pr_err("Device %d does not have PSID\n", dev_idx);
+		return false;
+	}
+
+	psid = mlxfw_mfa2_tlv_psid_get(mfa2_file, tlv);
+	if (!psid) {
+		pr_err("Device %d PSID TLV is not valid\n", dev_idx);
+		return false;
+	}
+
+	print_hex_dump_debug("  -- Device PSID ", DUMP_PREFIX_NONE, 16, 16,
+			     psid->psid, be16_to_cpu(tlv->len), true);
+
+	/* Validate the device has COMPONENT_PTR */
+	err = mlxfw_mfa2_tlv_multi_child_count(mfa2_file, multi,
+					       MLXFW_MFA2_TLV_COMPONENT_PTR,
+					       &cptr_count);
+	if (err)
+		return false;
+
+	if (cptr_count == 0) {
+		pr_err("Device %d has no components\n", dev_idx);
+		return false;
+	}
+
+	for (cptr_idx = 0; cptr_idx < cptr_count; cptr_idx++) {
+		tlv = mlxfw_mfa2_tlv_multi_child_find(mfa2_file, multi,
+						      MLXFW_MFA2_TLV_COMPONENT_PTR,
+						      cptr_idx);
+		if (!tlv)
+			return false;
+
+		cptr = mlxfw_mfa2_tlv_component_ptr_get(mfa2_file, tlv);
+		if (!cptr) {
+			pr_err("Device %d COMPONENT_PTR TLV is not valid\n",
+			       dev_idx);
+			return false;
+		}
+
+		pr_debug("  -- Component index %d\n",
+			 be16_to_cpu(cptr->component_index));
+	}
+	return true;
+}
+
+static bool
+mlxfw_mfa2_file_comp_validate(const struct mlxfw_mfa2_file *mfa2_file,
+			      const struct mlxfw_mfa2_tlv *comp_tlv,
+			      u16 comp_idx)
+{
+	const struct mlxfw_mfa2_tlv_component_descriptor *cdesc;
+	const struct mlxfw_mfa2_tlv_multi *multi;
+	const struct mlxfw_mfa2_tlv *tlv;
+
+	pr_debug("Component %d\n", comp_idx);
+
+	multi = mlxfw_mfa2_tlv_multi_get(mfa2_file, comp_tlv);
+	if (!multi) {
+		pr_err("Component %d is not a valid TLV error\n", comp_idx);
+		return false;
+	}
+
+	if (!mlxfw_mfa2_tlv_multi_validate(mfa2_file, multi))
+		return false;
+
+	/* Check that component have COMPONENT_DESCRIPTOR as first child */
+	tlv = mlxfw_mfa2_tlv_multi_child(mfa2_file, multi);
+	if (!tlv) {
+		pr_err("Component descriptor %d multi TLV error\n", comp_idx);
+		return false;
+	}
+
+	cdesc = mlxfw_mfa2_tlv_component_descriptor_get(mfa2_file, tlv);
+	if (!cdesc) {
+		pr_err("Component %d does not have a valid descriptor\n",
+		       comp_idx);
+		return false;
+	}
+	pr_debug("  -- Component type %d\n", be16_to_cpu(cdesc->identifier));
+	pr_debug("  -- Offset 0x%llx and size %d\n",
+		 ((u64) be32_to_cpu(cdesc->cb_offset_h) << 32)
+		 | be32_to_cpu(cdesc->cb_offset_l), be32_to_cpu(cdesc->size));
+
+	return true;
+}
+
+static bool mlxfw_mfa2_file_validate(const struct mlxfw_mfa2_file *mfa2_file)
+{
+	const struct mlxfw_mfa2_tlv *tlv;
+	u16 idx;
+
+	pr_debug("Validating file\n");
+
+	/* check that all the devices exist */
+	mlxfw_mfa2_tlv_foreach(mfa2_file, tlv, idx, mfa2_file->first_dev,
+			       mfa2_file->dev_count) {
+		if (!tlv) {
+			pr_err("Device TLV error\n");
+			return false;
+		}
+
+		/* Check each device */
+		if (!mlxfw_mfa2_file_dev_validate(mfa2_file, tlv, idx))
+			return false;
+	}
+
+	/* check that all the components exist */
+	mlxfw_mfa2_tlv_foreach(mfa2_file, tlv, idx, mfa2_file->first_component,
+			       mfa2_file->component_count) {
+		if (!tlv) {
+			pr_err("Device TLV error\n");
+			return false;
+		}
+
+		/* Check each component */
+		if (!mlxfw_mfa2_file_comp_validate(mfa2_file, tlv, idx))
+			return false;
+	}
+	return true;
+}
+
+struct mlxfw_mfa2_file *mlxfw_mfa2_file_init(const struct firmware *fw)
+{
+	const struct mlxfw_mfa2_tlv_package_descriptor *pd;
+	const struct mlxfw_mfa2_tlv_multi *multi;
+	const struct mlxfw_mfa2_tlv *multi_child;
+	const struct mlxfw_mfa2_tlv *first_tlv;
+	struct mlxfw_mfa2_file *mfa2_file;
+	const void *first_tlv_ptr;
+	const void *cb_top_ptr;
+
+	mfa2_file = kzalloc(sizeof(*mfa2_file), GFP_KERNEL);
+	if (!mfa2_file)
+		return ERR_PTR(-ENOMEM);
+
+	mfa2_file->fw = fw;
+	first_tlv_ptr = fw->data + NLA_ALIGN(mlxfw_mfa2_fingerprint_len);
+	first_tlv = mlxfw_mfa2_tlv_get(mfa2_file, first_tlv_ptr);
+	if (!first_tlv) {
+		pr_err("Could not parse package descriptor TLV\n");
+		goto err_out;
+	}
+
+	multi = mlxfw_mfa2_tlv_multi_get(mfa2_file, first_tlv);
+	if (!multi) {
+		pr_err("First TLV is not of valid multi type\n");
+		goto err_out;
+	}
+
+	multi_child = mlxfw_mfa2_tlv_multi_child(mfa2_file, multi);
+	if (!multi_child)
+		goto err_out;
+
+	pd = mlxfw_mfa2_tlv_package_descriptor_get(mfa2_file, multi_child);
+	if (!pd) {
+		pr_err("Could not parse package descriptor TLV\n");
+		goto err_out;
+	}
+
+	mfa2_file->first_dev = mlxfw_mfa2_tlv_next(mfa2_file, first_tlv);
+	if (!mfa2_file->first_dev) {
+		pr_err("First device TLV is not valid\n");
+		goto err_out;
+	}
+
+	mfa2_file->dev_count = be16_to_cpu(pd->num_devices);
+	mfa2_file->first_component = mlxfw_mfa2_tlv_advance(mfa2_file,
+							    mfa2_file->first_dev,
+							    mfa2_file->dev_count);
+	mfa2_file->component_count = be16_to_cpu(pd->num_components);
+	mfa2_file->cb = fw->data + NLA_ALIGN(be32_to_cpu(pd->cb_offset));
+	if (!mlxfw_mfa2_valid_ptr(mfa2_file, mfa2_file->cb)) {
+		pr_err("Component block is out side the file\n");
+		goto err_out;
+	}
+	mfa2_file->cb_archive_size = be32_to_cpu(pd->cb_archive_size);
+	cb_top_ptr = mfa2_file->cb + mfa2_file->cb_archive_size - 1;
+	if (!mlxfw_mfa2_valid_ptr(mfa2_file, cb_top_ptr)) {
+		pr_err("Component block size is too big\n");
+		goto err_out;
+	}
+
+	if (!mlxfw_mfa2_file_validate(mfa2_file))
+		goto err_out;
+	return mfa2_file;
+err_out:
+	kfree(mfa2_file);
+	return ERR_PTR(-EINVAL);
+}
+
+static const struct mlxfw_mfa2_tlv_multi *
+mlxfw_mfa2_tlv_dev_get(const struct mlxfw_mfa2_file *mfa2_file,
+		       const char *psid, u16 psid_size)
+{
+	const struct mlxfw_mfa2_tlv_psid *tlv_psid;
+	const struct mlxfw_mfa2_tlv_multi *dev_multi;
+	const struct mlxfw_mfa2_tlv *dev_tlv;
+	const struct mlxfw_mfa2_tlv *tlv;
+	u32 idx;
+
+	/* for each device tlv */
+	mlxfw_mfa2_tlv_foreach(mfa2_file, dev_tlv, idx, mfa2_file->first_dev,
+			       mfa2_file->dev_count) {
+		if (!dev_tlv)
+			return NULL;
+
+		dev_multi = mlxfw_mfa2_tlv_multi_get(mfa2_file, dev_tlv);
+		if (!dev_multi)
+			return NULL;
+
+		/* find psid child and compare */
+		tlv = mlxfw_mfa2_tlv_multi_child_find(mfa2_file, dev_multi,
+						      MLXFW_MFA2_TLV_PSID, 0);
+		if (!tlv)
+			return NULL;
+		if (be16_to_cpu(tlv->len) != psid_size)
+			continue;
+
+		tlv_psid = mlxfw_mfa2_tlv_psid_get(mfa2_file, tlv);
+		if (!tlv_psid)
+			return NULL;
+
+		if (memcmp(psid, tlv_psid->psid, psid_size) == 0)
+			return dev_multi;
+	}
+
+	return NULL;
+}
+
+int mlxfw_mfa2_file_component_count(const struct mlxfw_mfa2_file *mfa2_file,
+				    const char *psid, u32 psid_size,
+				    u32 *p_count)
+{
+	const struct mlxfw_mfa2_tlv_multi *dev_multi;
+	u16 count;
+	int err;
+
+	dev_multi = mlxfw_mfa2_tlv_dev_get(mfa2_file, psid, psid_size);
+	if (!dev_multi)
+		return -EINVAL;
+
+	err = mlxfw_mfa2_tlv_multi_child_count(mfa2_file, dev_multi,
+					       MLXFW_MFA2_TLV_COMPONENT_PTR,
+					       &count);
+	if (err)
+		return err;
+
+	*p_count = count;
+	return 0;
+}
+
+static int mlxfw_mfa2_xz_dec_run(struct xz_dec *xz_dec, struct xz_buf *xz_buf,
+				 bool *finished)
+{
+	enum xz_ret xz_ret;
+
+	xz_ret = xz_dec_run(xz_dec, xz_buf);
+
+	switch (xz_ret) {
+	case XZ_STREAM_END:
+		*finished = true;
+		return 0;
+	case XZ_OK:
+		*finished = false;
+		return 0;
+	case XZ_MEM_ERROR:
+		pr_err("xz no memory\n");
+		return -ENOMEM;
+	case XZ_DATA_ERROR:
+		pr_err("xz file corrupted\n");
+		return -EINVAL;
+	case XZ_FORMAT_ERROR:
+		pr_err("xz format not found\n");
+		return -EINVAL;
+	case XZ_OPTIONS_ERROR:
+		pr_err("unsupported xz option\n");
+		return -EINVAL;
+	case XZ_MEMLIMIT_ERROR:
+		pr_err("xz dictionary too small\n");
+		return -EINVAL;
+	default:
+		pr_err("xz error %d\n", xz_ret);
+		return -EINVAL;
+	}
+}
+
+static int mlxfw_mfa2_file_cb_offset_xz(const struct mlxfw_mfa2_file *mfa2_file,
+					off_t off, size_t size, u8 *buf)
+{
+	struct xz_dec *xz_dec;
+	struct xz_buf dec_buf;
+	off_t curr_off = 0;
+	bool finished;
+	int err;
+
+	xz_dec = xz_dec_init(XZ_DYNALLOC, (u32) -1);
+	if (!xz_dec)
+		return -EINVAL;
+
+	dec_buf.in_size = mfa2_file->cb_archive_size;
+	dec_buf.in = mfa2_file->cb;
+	dec_buf.in_pos = 0;
+	dec_buf.out = buf;
+
+	/* decode up to the offset */
+	do {
+		dec_buf.out_pos = 0;
+		dec_buf.out_size = min_t(size_t, size, off - curr_off);
+		if (dec_buf.out_size == 0)
+			break;
+
+		err = mlxfw_mfa2_xz_dec_run(xz_dec, &dec_buf, &finished);
+		if (err)
+			goto out;
+		if (finished) {
+			pr_err("xz section too short\n");
+			err = -EINVAL;
+			goto out;
+		}
+		curr_off += dec_buf.out_pos;
+	} while (curr_off != off);
+
+	/* decode the needed section */
+	dec_buf.out_pos = 0;
+	dec_buf.out_size = size;
+	err = mlxfw_mfa2_xz_dec_run(xz_dec, &dec_buf, &finished);
+out:
+	xz_dec_end(xz_dec);
+	return err;
+}
+
+static const struct mlxfw_mfa2_tlv_component_descriptor *
+mlxfw_mfa2_file_component_tlv_get(const struct mlxfw_mfa2_file *mfa2_file,
+				  u16 comp_index)
+{
+	const struct mlxfw_mfa2_tlv_multi *multi;
+	const struct mlxfw_mfa2_tlv *multi_child;
+	const struct mlxfw_mfa2_tlv *comp_tlv;
+
+	if (comp_index > mfa2_file->component_count)
+		return NULL;
+
+	comp_tlv = mlxfw_mfa2_tlv_advance(mfa2_file, mfa2_file->first_component,
+					  comp_index);
+	if (!comp_tlv)
+		return NULL;
+
+	multi = mlxfw_mfa2_tlv_multi_get(mfa2_file, comp_tlv);
+	if (!multi)
+		return NULL;
+
+	multi_child = mlxfw_mfa2_tlv_multi_child(mfa2_file, multi);
+	if (!multi_child)
+		return NULL;
+
+	return mlxfw_mfa2_tlv_component_descriptor_get(mfa2_file, multi_child);
+}
+
+struct mlxfw_mfa2_comp_data {
+	struct mlxfw_mfa2_component comp;
+	u8 buff[];
+};
+
+static const struct mlxfw_mfa2_tlv_component_descriptor *
+mlxfw_mfa2_file_component_find(const struct mlxfw_mfa2_file *mfa2_file,
+			       const char *psid, int psid_size,
+			       int component_index)
+{
+	const struct mlxfw_mfa2_tlv_component_ptr *cptr;
+	const struct mlxfw_mfa2_tlv_multi *dev_multi;
+	const struct mlxfw_mfa2_tlv *cptr_tlv;
+	u16 comp_idx;
+
+	dev_multi = mlxfw_mfa2_tlv_dev_get(mfa2_file, psid, psid_size);
+	if (!dev_multi)
+		return NULL;
+
+	cptr_tlv = mlxfw_mfa2_tlv_multi_child_find(mfa2_file, dev_multi,
+						   MLXFW_MFA2_TLV_COMPONENT_PTR,
+						   component_index);
+	if (!cptr_tlv)
+		return NULL;
+
+	cptr = mlxfw_mfa2_tlv_component_ptr_get(mfa2_file, cptr_tlv);
+	if (!cptr)
+		return NULL;
+
+	comp_idx = be16_to_cpu(cptr->component_index);
+	return mlxfw_mfa2_file_component_tlv_get(mfa2_file, comp_idx);
+}
+
+struct mlxfw_mfa2_component *
+mlxfw_mfa2_file_component_get(const struct mlxfw_mfa2_file *mfa2_file,
+			      const char *psid, int psid_size,
+			      int component_index)
+{
+	const struct mlxfw_mfa2_tlv_component_descriptor *comp;
+	struct mlxfw_mfa2_comp_data *comp_data;
+	u32 comp_buf_size;
+	off_t cb_offset;
+	u32 comp_size;
+	int err;
+
+	comp = mlxfw_mfa2_file_component_find(mfa2_file, psid, psid_size,
+					      component_index);
+	if (!comp)
+		return ERR_PTR(-EINVAL);
+
+	cb_offset = (u64) be32_to_cpu(comp->cb_offset_h) << 32 |
+		    be32_to_cpu(comp->cb_offset_l);
+	comp_size = be32_to_cpu(comp->size);
+	comp_buf_size = comp_size + mlxfw_mfa2_comp_magic_len;
+
+	comp_data = vzalloc(sizeof(*comp_data) + comp_buf_size);
+	if (!comp_data)
+		return ERR_PTR(-ENOMEM);
+	comp_data->comp.data_size = comp_size;
+	comp_data->comp.index = be16_to_cpu(comp->identifier);
+	err = mlxfw_mfa2_file_cb_offset_xz(mfa2_file, cb_offset, comp_buf_size,
+					   comp_data->buff);
+	if (err) {
+		pr_err("Component could not be reached in CB\n");
+		goto err_out;
+	}
+
+	if (memcmp(comp_data->buff, mlxfw_mfa2_comp_magic,
+		   mlxfw_mfa2_comp_magic_len) != 0) {
+		pr_err("Component has wrong magic\n");
+		err = -EINVAL;
+		goto err_out;
+	}
+
+	comp_data->comp.data = comp_data->buff + mlxfw_mfa2_comp_magic_len;
+	return &comp_data->comp;
+err_out:
+	vfree(comp_data);
+	return ERR_PTR(err);
+}
+
+void mlxfw_mfa2_file_component_put(struct mlxfw_mfa2_component *comp)
+{
+	const struct mlxfw_mfa2_comp_data *comp_data;
+
+	comp_data = container_of(comp, struct mlxfw_mfa2_comp_data, comp);
+	vfree(comp_data);
+}
+
+void mlxfw_mfa2_file_fini(struct mlxfw_mfa2_file *mfa2_file)
+{
+	kfree(mfa2_file);
+}
diff --git a/drivers/net/ethernet/mellanox/mlxfw/mlxfw_mfa2.h b/drivers/net/ethernet/mellanox/mlxfw/mlxfw_mfa2.h
new file mode 100644
index 000000000..5bba6ad79
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxfw/mlxfw_mfa2.h
@@ -0,0 +1,35 @@
+/* SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0 */
+/* Copyright (c) 2017-2019 Mellanox Technologies. All rights reserved */
+
+#ifndef _MLXFW_MFA2_H
+#define _MLXFW_MFA2_H
+
+#include <linux/firmware.h>
+#include "mlxfw.h"
+
+struct mlxfw_mfa2_component {
+	u16 index;
+	u32 data_size;
+	u8 *data;
+};
+
+struct mlxfw_mfa2_file;
+
+bool mlxfw_mfa2_check(const struct firmware *fw);
+
+struct mlxfw_mfa2_file *mlxfw_mfa2_file_init(const struct firmware *fw);
+
+int mlxfw_mfa2_file_component_count(const struct mlxfw_mfa2_file *mfa2_file,
+				    const char *psid, u32 psid_size,
+				    u32 *p_count);
+
+struct mlxfw_mfa2_component *
+mlxfw_mfa2_file_component_get(const struct mlxfw_mfa2_file *mfa2_file,
+			      const char *psid, int psid_size,
+			      int component_index);
+
+void mlxfw_mfa2_file_component_put(struct mlxfw_mfa2_component *component);
+
+void mlxfw_mfa2_file_fini(struct mlxfw_mfa2_file *mfa2_file);
+
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlxfw/mlxfw_mfa2_file.h b/drivers/net/ethernet/mellanox/mlxfw/mlxfw_mfa2_file.h
new file mode 100644
index 000000000..874c0a247
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxfw/mlxfw_mfa2_file.h
@@ -0,0 +1,29 @@
+/* SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0 */
+/* Copyright (c) 2017-2019 Mellanox Technologies. All rights reserved */
+
+#ifndef _MLXFW_MFA2_FILE_H
+#define _MLXFW_MFA2_FILE_H
+
+#include <linux/firmware.h>
+#include <linux/kernel.h>
+
+struct mlxfw_mfa2_file {
+	const struct firmware *fw;
+	const struct mlxfw_mfa2_tlv *first_dev;
+	u16 dev_count;
+	const struct mlxfw_mfa2_tlv *first_component;
+	u16 component_count;
+	const void *cb; /* components block */
+	u32 cb_archive_size; /* size of compressed components block */
+};
+
+static inline bool mlxfw_mfa2_valid_ptr(const struct mlxfw_mfa2_file *mfa2_file,
+					const void *ptr)
+{
+	const void *valid_to = mfa2_file->fw->data + mfa2_file->fw->size;
+	const void *valid_from = mfa2_file->fw->data;
+
+	return ptr > valid_from && ptr < valid_to;
+}
+
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlxfw/mlxfw_mfa2_format.h b/drivers/net/ethernet/mellanox/mlxfw/mlxfw_mfa2_format.h
new file mode 100644
index 000000000..b001e5258
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxfw/mlxfw_mfa2_format.h
@@ -0,0 +1,73 @@
+/* SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0 */
+/* Copyright (c) 2017-2019 Mellanox Technologies. All rights reserved */
+
+#ifndef _MLXFW_MFA2_FORMAT_H
+#define _MLXFW_MFA2_FORMAT_H
+
+#include "mlxfw_mfa2_file.h"
+#include "mlxfw_mfa2_tlv.h"
+
+enum mlxfw_mfa2_tlv_type {
+	MLXFW_MFA2_TLV_MULTI_PART = 0x01,
+	MLXFW_MFA2_TLV_PACKAGE_DESCRIPTOR = 0x02,
+	MLXFW_MFA2_TLV_COMPONENT_DESCRIPTOR = 0x04,
+	MLXFW_MFA2_TLV_COMPONENT_PTR = 0x22,
+	MLXFW_MFA2_TLV_PSID = 0x2A,
+};
+
+enum mlxfw_mfa2_compression_type {
+	MLXFW_MFA2_COMPRESSION_TYPE_NONE,
+	MLXFW_MFA2_COMPRESSION_TYPE_XZ,
+};
+
+struct mlxfw_mfa2_tlv_package_descriptor {
+	__be16 num_components;
+	__be16 num_devices;
+	__be32 cb_offset;
+	__be32 cb_archive_size;
+	__be32 cb_size_h;
+	__be32 cb_size_l;
+	u8 padding[3];
+	u8 cv_compression;
+	__be32 user_data_offset;
+} __packed;
+
+MLXFW_MFA2_TLV(package_descriptor, struct mlxfw_mfa2_tlv_package_descriptor,
+	       MLXFW_MFA2_TLV_PACKAGE_DESCRIPTOR);
+
+struct mlxfw_mfa2_tlv_multi {
+	__be16 num_extensions;
+	__be16 total_len;
+} __packed;
+
+MLXFW_MFA2_TLV(multi, struct mlxfw_mfa2_tlv_multi,
+	       MLXFW_MFA2_TLV_MULTI_PART);
+
+struct mlxfw_mfa2_tlv_psid {
+	u8 psid[0];
+} __packed;
+
+MLXFW_MFA2_TLV_VARSIZE(psid, struct mlxfw_mfa2_tlv_psid,
+		       MLXFW_MFA2_TLV_PSID);
+
+struct mlxfw_mfa2_tlv_component_ptr {
+	__be16 storage_id;
+	__be16 component_index;
+	__be32 storage_address;
+} __packed;
+
+MLXFW_MFA2_TLV(component_ptr, struct mlxfw_mfa2_tlv_component_ptr,
+	       MLXFW_MFA2_TLV_COMPONENT_PTR);
+
+struct mlxfw_mfa2_tlv_component_descriptor {
+	__be16 pldm_classification;
+	__be16 identifier;
+	__be32 cb_offset_h;
+	__be32 cb_offset_l;
+	__be32 size;
+} __packed;
+
+MLXFW_MFA2_TLV(component_descriptor, struct mlxfw_mfa2_tlv_component_descriptor,
+	       MLXFW_MFA2_TLV_COMPONENT_DESCRIPTOR);
+
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlxfw/mlxfw_mfa2_tlv.h b/drivers/net/ethernet/mellanox/mlxfw/mlxfw_mfa2_tlv.h
new file mode 100644
index 000000000..2014a5de5
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxfw/mlxfw_mfa2_tlv.h
@@ -0,0 +1,67 @@
+/* SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0 */
+/* Copyright (c) 2017-2019 Mellanox Technologies. All rights reserved */
+
+#ifndef _MLXFW_MFA2_TLV_H
+#define _MLXFW_MFA2_TLV_H
+
+#include <linux/kernel.h>
+#include "mlxfw_mfa2_file.h"
+
+struct mlxfw_mfa2_tlv {
+	u8 version;
+	u8 type;
+	__be16 len;
+	u8 data[];
+} __packed;
+
+static inline const struct mlxfw_mfa2_tlv *
+mlxfw_mfa2_tlv_get(const struct mlxfw_mfa2_file *mfa2_file, const void *ptr)
+{
+	if (!mlxfw_mfa2_valid_ptr(mfa2_file, ptr) ||
+	    !mlxfw_mfa2_valid_ptr(mfa2_file, ptr + sizeof(struct mlxfw_mfa2_tlv)))
+		return NULL;
+	return ptr;
+}
+
+static inline const void *
+mlxfw_mfa2_tlv_payload_get(const struct mlxfw_mfa2_file *mfa2_file,
+			   const struct mlxfw_mfa2_tlv *tlv, u8 payload_type,
+			   size_t payload_size, bool varsize)
+{
+	void *tlv_top;
+
+	tlv_top = (void *) tlv + be16_to_cpu(tlv->len) - 1;
+	if (!mlxfw_mfa2_valid_ptr(mfa2_file, tlv) ||
+	    !mlxfw_mfa2_valid_ptr(mfa2_file, tlv_top))
+		return NULL;
+	if (tlv->type != payload_type)
+		return NULL;
+	if (varsize && (be16_to_cpu(tlv->len) < payload_size))
+		return NULL;
+	if (!varsize && (be16_to_cpu(tlv->len) != payload_size))
+		return NULL;
+
+	return tlv->data;
+}
+
+#define MLXFW_MFA2_TLV(name, payload_type, tlv_type)			       \
+static inline const payload_type *					       \
+mlxfw_mfa2_tlv_ ## name ## _get(const struct mlxfw_mfa2_file *mfa2_file,       \
+				const struct mlxfw_mfa2_tlv *tlv)	       \
+{									       \
+	return mlxfw_mfa2_tlv_payload_get(mfa2_file, tlv,		       \
+					  tlv_type, sizeof(payload_type),      \
+					  false);			       \
+}
+
+#define MLXFW_MFA2_TLV_VARSIZE(name, payload_type, tlv_type)		       \
+static inline const payload_type *					       \
+mlxfw_mfa2_tlv_ ## name ## _get(const struct mlxfw_mfa2_file *mfa2_file,       \
+				const struct mlxfw_mfa2_tlv *tlv)	       \
+{									       \
+	return mlxfw_mfa2_tlv_payload_get(mfa2_file, tlv,		       \
+					  tlv_type, sizeof(payload_type),      \
+					  true);			       \
+}
+
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlxfw/mlxfw_mfa2_tlv_multi.c b/drivers/net/ethernet/mellanox/mlxfw/mlxfw_mfa2_tlv_multi.c
new file mode 100644
index 000000000..972c571b4
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxfw/mlxfw_mfa2_tlv_multi.c
@@ -0,0 +1,97 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2017-2019 Mellanox Technologies. All rights reserved */
+
+#define pr_fmt(fmt) "MFA2: " fmt
+
+#include "mlxfw_mfa2_tlv_multi.h"
+#include <uapi/linux/netlink.h>
+
+#define MLXFW_MFA2_TLV_TOTAL_SIZE(tlv) \
+	NLA_ALIGN(sizeof(*(tlv)) + be16_to_cpu((tlv)->len))
+
+const struct mlxfw_mfa2_tlv *
+mlxfw_mfa2_tlv_multi_child(const struct mlxfw_mfa2_file *mfa2_file,
+			   const struct mlxfw_mfa2_tlv_multi *multi)
+{
+	size_t multi_len;
+
+	multi_len = NLA_ALIGN(sizeof(struct mlxfw_mfa2_tlv_multi));
+	return mlxfw_mfa2_tlv_get(mfa2_file, (void *) multi + multi_len);
+}
+
+const struct mlxfw_mfa2_tlv *
+mlxfw_mfa2_tlv_next(const struct mlxfw_mfa2_file *mfa2_file,
+		    const struct mlxfw_mfa2_tlv *tlv)
+{
+	const struct mlxfw_mfa2_tlv_multi *multi;
+	u16 tlv_len;
+	void *next;
+
+	tlv_len = MLXFW_MFA2_TLV_TOTAL_SIZE(tlv);
+
+	if (tlv->type == MLXFW_MFA2_TLV_MULTI_PART) {
+		multi = mlxfw_mfa2_tlv_multi_get(mfa2_file, tlv);
+		if (!multi)
+			return NULL;
+		tlv_len = NLA_ALIGN(tlv_len + be16_to_cpu(multi->total_len));
+	}
+
+	next = (void *) tlv + tlv_len;
+	return mlxfw_mfa2_tlv_get(mfa2_file, next);
+}
+
+const struct mlxfw_mfa2_tlv *
+mlxfw_mfa2_tlv_advance(const struct mlxfw_mfa2_file *mfa2_file,
+		       const struct mlxfw_mfa2_tlv *from_tlv, u16 count)
+{
+	const struct mlxfw_mfa2_tlv *tlv;
+	u16 idx;
+
+	mlxfw_mfa2_tlv_foreach(mfa2_file, tlv, idx, from_tlv, count)
+		if (!tlv)
+			return NULL;
+	return tlv;
+}
+
+const struct mlxfw_mfa2_tlv *
+mlxfw_mfa2_tlv_multi_child_find(const struct mlxfw_mfa2_file *mfa2_file,
+				const struct mlxfw_mfa2_tlv_multi *multi,
+				enum mlxfw_mfa2_tlv_type type, u16 index)
+{
+	const struct mlxfw_mfa2_tlv *tlv;
+	u16 skip = 0;
+	u16 idx;
+
+	mlxfw_mfa2_tlv_multi_foreach(mfa2_file, tlv, idx, multi) {
+		if (!tlv) {
+			pr_err("TLV parsing error\n");
+			return NULL;
+		}
+		if (tlv->type == type)
+			if (skip++ == index)
+				return tlv;
+	}
+	return NULL;
+}
+
+int mlxfw_mfa2_tlv_multi_child_count(const struct mlxfw_mfa2_file *mfa2_file,
+				     const struct mlxfw_mfa2_tlv_multi *multi,
+				     enum mlxfw_mfa2_tlv_type type,
+				     u16 *p_count)
+{
+	const struct mlxfw_mfa2_tlv *tlv;
+	u16 count = 0;
+	u16 idx;
+
+	mlxfw_mfa2_tlv_multi_foreach(mfa2_file, tlv, idx, multi) {
+		if (!tlv) {
+			pr_err("TLV parsing error\n");
+			return -EINVAL;
+		}
+
+		if (tlv->type == type)
+			count++;
+	}
+	*p_count = count;
+	return 0;
+}
diff --git a/drivers/net/ethernet/mellanox/mlxfw/mlxfw_mfa2_tlv_multi.h b/drivers/net/ethernet/mellanox/mlxfw/mlxfw_mfa2_tlv_multi.h
new file mode 100644
index 000000000..633284eed
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxfw/mlxfw_mfa2_tlv_multi.h
@@ -0,0 +1,41 @@
+/* SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0 */
+/* Copyright (c) 2017-2019 Mellanox Technologies. All rights reserved */
+
+#ifndef _MLXFW_MFA2_TLV_MULTI_H
+#define _MLXFW_MFA2_TLV_MULTI_H
+
+#include "mlxfw_mfa2_tlv.h"
+#include "mlxfw_mfa2_format.h"
+#include "mlxfw_mfa2_file.h"
+
+const struct mlxfw_mfa2_tlv *
+mlxfw_mfa2_tlv_multi_child(const struct mlxfw_mfa2_file *mfa2_file,
+			   const struct mlxfw_mfa2_tlv_multi *multi);
+
+const struct mlxfw_mfa2_tlv *
+mlxfw_mfa2_tlv_next(const struct mlxfw_mfa2_file *mfa2_file,
+		    const struct mlxfw_mfa2_tlv *tlv);
+
+const struct mlxfw_mfa2_tlv *
+mlxfw_mfa2_tlv_advance(const struct mlxfw_mfa2_file *mfa2_file,
+		       const struct mlxfw_mfa2_tlv *from_tlv, u16 count);
+
+const struct mlxfw_mfa2_tlv *
+mlxfw_mfa2_tlv_multi_child_find(const struct mlxfw_mfa2_file *mfa2_file,
+				const struct mlxfw_mfa2_tlv_multi *multi,
+				enum mlxfw_mfa2_tlv_type type, u16 index);
+
+int mlxfw_mfa2_tlv_multi_child_count(const struct mlxfw_mfa2_file *mfa2_file,
+				     const struct mlxfw_mfa2_tlv_multi *multi,
+				     enum mlxfw_mfa2_tlv_type type,
+				     u16 *p_count);
+
+#define mlxfw_mfa2_tlv_foreach(mfa2_file, tlv, idx, from_tlv, count) \
+	for (idx = 0, tlv = from_tlv; idx < (count); \
+	     idx++, tlv = mlxfw_mfa2_tlv_next(mfa2_file, tlv))
+
+#define mlxfw_mfa2_tlv_multi_foreach(mfa2_file, tlv, idx, multi) \
+	mlxfw_mfa2_tlv_foreach(mfa2_file, tlv, idx, \
+			       mlxfw_mfa2_tlv_multi_child(mfa2_file, multi), \
+			       be16_to_cpu(multi->num_extensions) + 1)
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlxsw/Kconfig b/drivers/net/ethernet/mellanox/mlxsw/Kconfig
new file mode 100644
index 000000000..a510bf2cf
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/Kconfig
@@ -0,0 +1,92 @@
+# SPDX-License-Identifier: GPL-2.0-only
+#
+# Mellanox switch drivers configuration
+#
+
+config MLXSW_CORE
+	tristate "Mellanox Technologies Switch ASICs support"
+	select NET_DEVLINK
+	select MLXFW
+	select AUXILIARY_BUS
+	help
+	  This driver supports Mellanox Technologies Switch ASICs family.
+
+	  To compile this driver as a module, choose M here: the
+	  module will be called mlxsw_core.
+
+config MLXSW_CORE_HWMON
+	bool "HWMON support for Mellanox Technologies Switch ASICs"
+	depends on MLXSW_CORE && HWMON
+	depends on !(MLXSW_CORE=y && HWMON=m)
+	default y
+	help
+	  Say Y here if you want to expose HWMON interface on mlxsw devices.
+
+config MLXSW_CORE_THERMAL
+	bool "Thermal zone support for Mellanox Technologies Switch ASICs"
+	depends on MLXSW_CORE && THERMAL
+	default y
+	help
+	 Say Y here if you want to automatically control fans speed according
+	 ambient temperature reported by ASIC.
+
+config MLXSW_PCI
+	tristate "PCI bus implementation for Mellanox Technologies Switch ASICs"
+	depends on PCI && HAS_IOMEM && MLXSW_CORE
+	default m
+	help
+	  This is PCI bus implementation for Mellanox Technologies Switch ASICs.
+
+	  To compile this driver as a module, choose M here: the
+	  module will be called mlxsw_pci.
+
+config MLXSW_I2C
+	tristate "I2C bus implementation for Mellanox Technologies Switch ASICs"
+	depends on I2C && MLXSW_CORE
+	default m
+	help
+	  This is I2C bus implementation for Mellanox Technologies Switch ASICs.
+
+	  To compile this driver as a module, choose M here: the
+	  module will be called mlxsw_i2c.
+
+config MLXSW_SPECTRUM
+	tristate "Mellanox Technologies Spectrum family support"
+	depends on MLXSW_CORE && MLXSW_PCI && NET_SWITCHDEV && VLAN_8021Q
+	depends on PSAMPLE || PSAMPLE=n
+	depends on BRIDGE || BRIDGE=n
+	depends on IPV6 || IPV6=n
+	depends on NET_IPGRE || NET_IPGRE=n
+	depends on IPV6_GRE || IPV6_GRE=n
+	depends on VXLAN || VXLAN=n
+	depends on PTP_1588_CLOCK_OPTIONAL
+	select GENERIC_ALLOCATOR
+	select PARMAN
+	select OBJAGG
+	select NET_PTP_CLASSIFY if PTP_1588_CLOCK
+	default m
+	help
+	  This driver supports Mellanox Technologies
+	  Spectrum/Spectrum-2/Spectrum-3/Spectrum-4 Ethernet Switch ASICs.
+
+	  To compile this driver as a module, choose M here: the
+	  module will be called mlxsw_spectrum.
+
+config MLXSW_SPECTRUM_DCB
+	bool "Data Center Bridging (DCB) support"
+	depends on MLXSW_SPECTRUM && DCB
+	default y
+	help
+	  Say Y here if you want to use Data Center Bridging (DCB) in the
+	  driver.
+
+config MLXSW_MINIMAL
+	tristate "Mellanox Technologies minimal I2C support"
+	depends on MLXSW_CORE && MLXSW_I2C
+	default m
+	help
+	  This driver supports I2C access for Mellanox Technologies Switch
+	  ASICs.
+
+	  To compile this driver as a module, choose M here: the
+	  module will be called mlxsw_minimal.
diff --git a/drivers/net/ethernet/mellanox/mlxsw/Makefile b/drivers/net/ethernet/mellanox/mlxsw/Makefile
new file mode 100644
index 000000000..3ca9fce75
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/Makefile
@@ -0,0 +1,36 @@
+# SPDX-License-Identifier: GPL-2.0
+obj-$(CONFIG_MLXSW_CORE)	+= mlxsw_core.o
+mlxsw_core-objs			:= core.o core_acl_flex_keys.o \
+				   core_acl_flex_actions.o core_env.o \
+				   core_linecards.o core_linecard_dev.o
+mlxsw_core-$(CONFIG_MLXSW_CORE_HWMON) += core_hwmon.o
+mlxsw_core-$(CONFIG_MLXSW_CORE_THERMAL) += core_thermal.o
+obj-$(CONFIG_MLXSW_PCI)		+= mlxsw_pci.o
+mlxsw_pci-objs			:= pci.o
+obj-$(CONFIG_MLXSW_I2C)		+= mlxsw_i2c.o
+mlxsw_i2c-objs			:= i2c.o
+obj-$(CONFIG_MLXSW_SPECTRUM)	+= mlxsw_spectrum.o
+mlxsw_spectrum-objs		:= spectrum.o spectrum_buffers.o \
+				   spectrum_switchdev.o spectrum_router.o \
+				   spectrum1_kvdl.o spectrum2_kvdl.o \
+				   spectrum_kvdl.o \
+				   spectrum_acl_tcam.o spectrum_acl_ctcam.o \
+				   spectrum_acl_atcam.o spectrum_acl_erp.o \
+				   spectrum1_acl_tcam.o spectrum2_acl_tcam.o \
+				   spectrum_acl_bloom_filter.o spectrum_acl.o \
+				   spectrum_flow.o spectrum_matchall.o \
+				   spectrum_flower.o spectrum_cnt.o \
+				   spectrum_fid.o spectrum_ipip.o \
+				   spectrum_acl_flex_actions.o \
+				   spectrum_acl_flex_keys.o \
+				   spectrum1_mr_tcam.o spectrum2_mr_tcam.o \
+				   spectrum_mr_tcam.o spectrum_mr.o \
+				   spectrum_qdisc.o spectrum_span.o \
+				   spectrum_nve.o spectrum_nve_vxlan.o \
+				   spectrum_dpipe.o spectrum_trap.o \
+				   spectrum_ethtool.o spectrum_policer.o \
+				   spectrum_pgt.o
+mlxsw_spectrum-$(CONFIG_MLXSW_SPECTRUM_DCB)	+= spectrum_dcb.o
+mlxsw_spectrum-$(CONFIG_PTP_1588_CLOCK)		+= spectrum_ptp.o
+obj-$(CONFIG_MLXSW_MINIMAL)	+= mlxsw_minimal.o
+mlxsw_minimal-objs		:= minimal.o
diff --git a/drivers/net/ethernet/mellanox/mlxsw/cmd.h b/drivers/net/ethernet/mellanox/mlxsw/cmd.h
new file mode 100644
index 000000000..09bef04b1
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/cmd.h
@@ -0,0 +1,1281 @@
+/* SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0 */
+/* Copyright (c) 2015-2018 Mellanox Technologies. All rights reserved */
+
+#ifndef _MLXSW_CMD_H
+#define _MLXSW_CMD_H
+
+#include "item.h"
+
+#define MLXSW_CMD_MBOX_SIZE	4096
+
+static inline char *mlxsw_cmd_mbox_alloc(void)
+{
+	return kzalloc(MLXSW_CMD_MBOX_SIZE, GFP_KERNEL);
+}
+
+static inline void mlxsw_cmd_mbox_free(char *mbox)
+{
+	kfree(mbox);
+}
+
+static inline void mlxsw_cmd_mbox_zero(char *mbox)
+{
+	memset(mbox, 0, MLXSW_CMD_MBOX_SIZE);
+}
+
+struct mlxsw_core;
+
+int mlxsw_cmd_exec(struct mlxsw_core *mlxsw_core, u16 opcode, u8 opcode_mod,
+		   u32 in_mod, bool out_mbox_direct, bool reset_ok,
+		   char *in_mbox, size_t in_mbox_size,
+		   char *out_mbox, size_t out_mbox_size);
+
+static inline int mlxsw_cmd_exec_in(struct mlxsw_core *mlxsw_core, u16 opcode,
+				    u8 opcode_mod, u32 in_mod, char *in_mbox,
+				    size_t in_mbox_size)
+{
+	return mlxsw_cmd_exec(mlxsw_core, opcode, opcode_mod, in_mod, false,
+			      false, in_mbox, in_mbox_size, NULL, 0);
+}
+
+static inline int mlxsw_cmd_exec_out(struct mlxsw_core *mlxsw_core, u16 opcode,
+				     u8 opcode_mod, u32 in_mod,
+				     bool out_mbox_direct,
+				     char *out_mbox, size_t out_mbox_size)
+{
+	return mlxsw_cmd_exec(mlxsw_core, opcode, opcode_mod, in_mod,
+			      out_mbox_direct, false, NULL, 0,
+			      out_mbox, out_mbox_size);
+}
+
+static inline int mlxsw_cmd_exec_none(struct mlxsw_core *mlxsw_core, u16 opcode,
+				      u8 opcode_mod, u32 in_mod)
+{
+	return mlxsw_cmd_exec(mlxsw_core, opcode, opcode_mod, in_mod, false,
+			      false, NULL, 0, NULL, 0);
+}
+
+enum mlxsw_cmd_opcode {
+	MLXSW_CMD_OPCODE_QUERY_FW		= 0x004,
+	MLXSW_CMD_OPCODE_QUERY_BOARDINFO	= 0x006,
+	MLXSW_CMD_OPCODE_QUERY_AQ_CAP		= 0x003,
+	MLXSW_CMD_OPCODE_MAP_FA			= 0xFFF,
+	MLXSW_CMD_OPCODE_UNMAP_FA		= 0xFFE,
+	MLXSW_CMD_OPCODE_CONFIG_PROFILE		= 0x100,
+	MLXSW_CMD_OPCODE_ACCESS_REG		= 0x040,
+	MLXSW_CMD_OPCODE_SW2HW_DQ		= 0x201,
+	MLXSW_CMD_OPCODE_HW2SW_DQ		= 0x202,
+	MLXSW_CMD_OPCODE_2ERR_DQ		= 0x01E,
+	MLXSW_CMD_OPCODE_QUERY_DQ		= 0x022,
+	MLXSW_CMD_OPCODE_SW2HW_CQ		= 0x016,
+	MLXSW_CMD_OPCODE_HW2SW_CQ		= 0x017,
+	MLXSW_CMD_OPCODE_QUERY_CQ		= 0x018,
+	MLXSW_CMD_OPCODE_SW2HW_EQ		= 0x013,
+	MLXSW_CMD_OPCODE_HW2SW_EQ		= 0x014,
+	MLXSW_CMD_OPCODE_QUERY_EQ		= 0x015,
+	MLXSW_CMD_OPCODE_QUERY_RESOURCES	= 0x101,
+};
+
+static inline const char *mlxsw_cmd_opcode_str(u16 opcode)
+{
+	switch (opcode) {
+	case MLXSW_CMD_OPCODE_QUERY_FW:
+		return "QUERY_FW";
+	case MLXSW_CMD_OPCODE_QUERY_BOARDINFO:
+		return "QUERY_BOARDINFO";
+	case MLXSW_CMD_OPCODE_QUERY_AQ_CAP:
+		return "QUERY_AQ_CAP";
+	case MLXSW_CMD_OPCODE_MAP_FA:
+		return "MAP_FA";
+	case MLXSW_CMD_OPCODE_UNMAP_FA:
+		return "UNMAP_FA";
+	case MLXSW_CMD_OPCODE_CONFIG_PROFILE:
+		return "CONFIG_PROFILE";
+	case MLXSW_CMD_OPCODE_ACCESS_REG:
+		return "ACCESS_REG";
+	case MLXSW_CMD_OPCODE_SW2HW_DQ:
+		return "SW2HW_DQ";
+	case MLXSW_CMD_OPCODE_HW2SW_DQ:
+		return "HW2SW_DQ";
+	case MLXSW_CMD_OPCODE_2ERR_DQ:
+		return "2ERR_DQ";
+	case MLXSW_CMD_OPCODE_QUERY_DQ:
+		return "QUERY_DQ";
+	case MLXSW_CMD_OPCODE_SW2HW_CQ:
+		return "SW2HW_CQ";
+	case MLXSW_CMD_OPCODE_HW2SW_CQ:
+		return "HW2SW_CQ";
+	case MLXSW_CMD_OPCODE_QUERY_CQ:
+		return "QUERY_CQ";
+	case MLXSW_CMD_OPCODE_SW2HW_EQ:
+		return "SW2HW_EQ";
+	case MLXSW_CMD_OPCODE_HW2SW_EQ:
+		return "HW2SW_EQ";
+	case MLXSW_CMD_OPCODE_QUERY_EQ:
+		return "QUERY_EQ";
+	case MLXSW_CMD_OPCODE_QUERY_RESOURCES:
+		return "QUERY_RESOURCES";
+	default:
+		return "*UNKNOWN*";
+	}
+}
+
+enum mlxsw_cmd_status {
+	/* Command execution succeeded. */
+	MLXSW_CMD_STATUS_OK		= 0x00,
+	/* Internal error (e.g. bus error) occurred while processing command. */
+	MLXSW_CMD_STATUS_INTERNAL_ERR	= 0x01,
+	/* Operation/command not supported or opcode modifier not supported. */
+	MLXSW_CMD_STATUS_BAD_OP		= 0x02,
+	/* Parameter not supported, parameter out of range. */
+	MLXSW_CMD_STATUS_BAD_PARAM	= 0x03,
+	/* System was not enabled or bad system state. */
+	MLXSW_CMD_STATUS_BAD_SYS_STATE	= 0x04,
+	/* Attempt to access reserved or unallocated resource, or resource in
+	 * inappropriate ownership.
+	 */
+	MLXSW_CMD_STATUS_BAD_RESOURCE	= 0x05,
+	/* Requested resource is currently executing a command. */
+	MLXSW_CMD_STATUS_RESOURCE_BUSY	= 0x06,
+	/* Required capability exceeds device limits. */
+	MLXSW_CMD_STATUS_EXCEED_LIM	= 0x08,
+	/* Resource is not in the appropriate state or ownership. */
+	MLXSW_CMD_STATUS_BAD_RES_STATE	= 0x09,
+	/* Index out of range (might be beyond table size or attempt to
+	 * access a reserved resource).
+	 */
+	MLXSW_CMD_STATUS_BAD_INDEX	= 0x0A,
+	/* NVMEM checksum/CRC failed. */
+	MLXSW_CMD_STATUS_BAD_NVMEM	= 0x0B,
+	/* Device is currently running reset */
+	MLXSW_CMD_STATUS_RUNNING_RESET	= 0x26,
+	/* Bad management packet (silently discarded). */
+	MLXSW_CMD_STATUS_BAD_PKT	= 0x30,
+};
+
+static inline const char *mlxsw_cmd_status_str(u8 status)
+{
+	switch (status) {
+	case MLXSW_CMD_STATUS_OK:
+		return "OK";
+	case MLXSW_CMD_STATUS_INTERNAL_ERR:
+		return "INTERNAL_ERR";
+	case MLXSW_CMD_STATUS_BAD_OP:
+		return "BAD_OP";
+	case MLXSW_CMD_STATUS_BAD_PARAM:
+		return "BAD_PARAM";
+	case MLXSW_CMD_STATUS_BAD_SYS_STATE:
+		return "BAD_SYS_STATE";
+	case MLXSW_CMD_STATUS_BAD_RESOURCE:
+		return "BAD_RESOURCE";
+	case MLXSW_CMD_STATUS_RESOURCE_BUSY:
+		return "RESOURCE_BUSY";
+	case MLXSW_CMD_STATUS_EXCEED_LIM:
+		return "EXCEED_LIM";
+	case MLXSW_CMD_STATUS_BAD_RES_STATE:
+		return "BAD_RES_STATE";
+	case MLXSW_CMD_STATUS_BAD_INDEX:
+		return "BAD_INDEX";
+	case MLXSW_CMD_STATUS_BAD_NVMEM:
+		return "BAD_NVMEM";
+	case MLXSW_CMD_STATUS_RUNNING_RESET:
+		return "RUNNING_RESET";
+	case MLXSW_CMD_STATUS_BAD_PKT:
+		return "BAD_PKT";
+	default:
+		return "*UNKNOWN*";
+	}
+}
+
+/* QUERY_FW - Query Firmware
+ * -------------------------
+ * OpMod == 0, INMmod == 0
+ * -----------------------
+ * The QUERY_FW command retrieves information related to firmware, command
+ * interface version and the amount of resources that should be allocated to
+ * the firmware.
+ */
+
+static inline int mlxsw_cmd_query_fw(struct mlxsw_core *mlxsw_core,
+				     char *out_mbox)
+{
+	return mlxsw_cmd_exec_out(mlxsw_core, MLXSW_CMD_OPCODE_QUERY_FW,
+				  0, 0, false, out_mbox, MLXSW_CMD_MBOX_SIZE);
+}
+
+/* cmd_mbox_query_fw_fw_pages
+ * Amount of physical memory to be allocatedfor firmware usage in 4KB pages.
+ */
+MLXSW_ITEM32(cmd_mbox, query_fw, fw_pages, 0x00, 16, 16);
+
+/* cmd_mbox_query_fw_fw_rev_major
+ * Firmware Revision - Major
+ */
+MLXSW_ITEM32(cmd_mbox, query_fw, fw_rev_major, 0x00, 0, 16);
+
+/* cmd_mbox_query_fw_fw_rev_subminor
+ * Firmware Sub-minor version (Patch level)
+ */
+MLXSW_ITEM32(cmd_mbox, query_fw, fw_rev_subminor, 0x04, 16, 16);
+
+/* cmd_mbox_query_fw_fw_rev_minor
+ * Firmware Revision - Minor
+ */
+MLXSW_ITEM32(cmd_mbox, query_fw, fw_rev_minor, 0x04, 0, 16);
+
+/* cmd_mbox_query_fw_core_clk
+ * Internal Clock Frequency (in MHz)
+ */
+MLXSW_ITEM32(cmd_mbox, query_fw, core_clk, 0x08, 16, 16);
+
+/* cmd_mbox_query_fw_cmd_interface_rev
+ * Command Interface Interpreter Revision ID. This number is bumped up
+ * every time a non-backward-compatible change is done for the command
+ * interface. The current cmd_interface_rev is 1.
+ */
+MLXSW_ITEM32(cmd_mbox, query_fw, cmd_interface_rev, 0x08, 0, 16);
+
+/* cmd_mbox_query_fw_dt
+ * If set, Debug Trace is supported
+ */
+MLXSW_ITEM32(cmd_mbox, query_fw, dt, 0x0C, 31, 1);
+
+/* cmd_mbox_query_fw_api_version
+ * Indicates the version of the API, to enable software querying
+ * for compatibility. The current api_version is 1.
+ */
+MLXSW_ITEM32(cmd_mbox, query_fw, api_version, 0x0C, 0, 16);
+
+/* cmd_mbox_query_fw_fw_hour
+ * Firmware timestamp - hour
+ */
+MLXSW_ITEM32(cmd_mbox, query_fw, fw_hour, 0x10, 24, 8);
+
+/* cmd_mbox_query_fw_fw_minutes
+ * Firmware timestamp - minutes
+ */
+MLXSW_ITEM32(cmd_mbox, query_fw, fw_minutes, 0x10, 16, 8);
+
+/* cmd_mbox_query_fw_fw_seconds
+ * Firmware timestamp - seconds
+ */
+MLXSW_ITEM32(cmd_mbox, query_fw, fw_seconds, 0x10, 8, 8);
+
+/* cmd_mbox_query_fw_fw_year
+ * Firmware timestamp - year
+ */
+MLXSW_ITEM32(cmd_mbox, query_fw, fw_year, 0x14, 16, 16);
+
+/* cmd_mbox_query_fw_fw_month
+ * Firmware timestamp - month
+ */
+MLXSW_ITEM32(cmd_mbox, query_fw, fw_month, 0x14, 8, 8);
+
+/* cmd_mbox_query_fw_fw_day
+ * Firmware timestamp - day
+ */
+MLXSW_ITEM32(cmd_mbox, query_fw, fw_day, 0x14, 0, 8);
+
+/* cmd_mbox_query_fw_clr_int_base_offset
+ * Clear Interrupt register's offset from clr_int_bar register
+ * in PCI address space.
+ */
+MLXSW_ITEM64(cmd_mbox, query_fw, clr_int_base_offset, 0x20, 0, 64);
+
+/* cmd_mbox_query_fw_clr_int_bar
+ * PCI base address register (BAR) where clr_int register is located.
+ * 00 - BAR 0-1 (64 bit BAR)
+ */
+MLXSW_ITEM32(cmd_mbox, query_fw, clr_int_bar, 0x28, 30, 2);
+
+/* cmd_mbox_query_fw_error_buf_offset
+ * Read Only buffer for internal error reports of offset
+ * from error_buf_bar register in PCI address space).
+ */
+MLXSW_ITEM64(cmd_mbox, query_fw, error_buf_offset, 0x30, 0, 64);
+
+/* cmd_mbox_query_fw_error_buf_size
+ * Internal error buffer size in DWORDs
+ */
+MLXSW_ITEM32(cmd_mbox, query_fw, error_buf_size, 0x38, 0, 32);
+
+/* cmd_mbox_query_fw_error_int_bar
+ * PCI base address register (BAR) where error buffer
+ * register is located.
+ * 00 - BAR 0-1 (64 bit BAR)
+ */
+MLXSW_ITEM32(cmd_mbox, query_fw, error_int_bar, 0x3C, 30, 2);
+
+/* cmd_mbox_query_fw_doorbell_page_offset
+ * Offset of the doorbell page
+ */
+MLXSW_ITEM64(cmd_mbox, query_fw, doorbell_page_offset, 0x40, 0, 64);
+
+/* cmd_mbox_query_fw_doorbell_page_bar
+ * PCI base address register (BAR) of the doorbell page
+ * 00 - BAR 0-1 (64 bit BAR)
+ */
+MLXSW_ITEM32(cmd_mbox, query_fw, doorbell_page_bar, 0x48, 30, 2);
+
+/* cmd_mbox_query_fw_free_running_clock_offset
+ * The offset of the free running clock page
+ */
+MLXSW_ITEM64(cmd_mbox, query_fw, free_running_clock_offset, 0x50, 0, 64);
+
+/* cmd_mbox_query_fw_fr_rn_clk_bar
+ * PCI base address register (BAR) of the free running clock page
+ * 0: BAR 0
+ * 1: 64 bit BAR
+ */
+MLXSW_ITEM32(cmd_mbox, query_fw, fr_rn_clk_bar, 0x58, 30, 2);
+
+/* cmd_mbox_query_fw_utc_sec_offset
+ * The offset of the UTC_Sec page
+ */
+MLXSW_ITEM64(cmd_mbox, query_fw, utc_sec_offset, 0x70, 0, 64);
+
+/* cmd_mbox_query_fw_utc_sec_bar
+ * PCI base address register (BAR) of the UTC_Sec page
+ * 0: BAR 0
+ * 1: 64 bit BAR
+ * Reserved on SwitchX/-2, Switch-IB/2, Spectrum-1
+ */
+MLXSW_ITEM32(cmd_mbox, query_fw, utc_sec_bar, 0x78, 30, 2);
+
+/* cmd_mbox_query_fw_utc_nsec_offset
+ * The offset of the UTC_nSec page
+ */
+MLXSW_ITEM64(cmd_mbox, query_fw, utc_nsec_offset, 0x80, 0, 64);
+
+/* cmd_mbox_query_fw_utc_nsec_bar
+ * PCI base address register (BAR) of the UTC_nSec page
+ * 0: BAR 0
+ * 1: 64 bit BAR
+ * Reserved on SwitchX/-2, Switch-IB/2, Spectrum-1
+ */
+MLXSW_ITEM32(cmd_mbox, query_fw, utc_nsec_bar, 0x88, 30, 2);
+
+/* QUERY_BOARDINFO - Query Board Information
+ * -----------------------------------------
+ * OpMod == 0 (N/A), INMmod == 0 (N/A)
+ * -----------------------------------
+ * The QUERY_BOARDINFO command retrieves adapter specific parameters.
+ */
+
+static inline int mlxsw_cmd_boardinfo(struct mlxsw_core *mlxsw_core,
+				      char *out_mbox)
+{
+	return mlxsw_cmd_exec_out(mlxsw_core, MLXSW_CMD_OPCODE_QUERY_BOARDINFO,
+				  0, 0, false, out_mbox, MLXSW_CMD_MBOX_SIZE);
+}
+
+/* cmd_mbox_boardinfo_intapin
+ * When PCIe interrupt messages are being used, this value is used for clearing
+ * an interrupt. When using MSI-X, this register is not used.
+ */
+MLXSW_ITEM32(cmd_mbox, boardinfo, intapin, 0x10, 24, 8);
+
+/* cmd_mbox_boardinfo_vsd_vendor_id
+ * PCISIG Vendor ID (www.pcisig.com/membership/vid_search) of the vendor
+ * specifying/formatting the VSD. The vsd_vendor_id identifies the management
+ * domain of the VSD/PSID data. Different vendors may choose different VSD/PSID
+ * format and encoding as long as they use their assigned vsd_vendor_id.
+ */
+MLXSW_ITEM32(cmd_mbox, boardinfo, vsd_vendor_id, 0x1C, 0, 16);
+
+/* cmd_mbox_boardinfo_vsd
+ * Vendor Specific Data. The VSD string that is burnt to the Flash
+ * with the firmware.
+ */
+#define MLXSW_CMD_BOARDINFO_VSD_LEN 208
+MLXSW_ITEM_BUF(cmd_mbox, boardinfo, vsd, 0x20, MLXSW_CMD_BOARDINFO_VSD_LEN);
+
+/* cmd_mbox_boardinfo_psid
+ * The PSID field is a 16-ascii (byte) character string which acts as
+ * the board ID. The PSID format is used in conjunction with
+ * Mellanox vsd_vendor_id (15B3h).
+ */
+#define MLXSW_CMD_BOARDINFO_PSID_LEN 16
+MLXSW_ITEM_BUF(cmd_mbox, boardinfo, psid, 0xF0, MLXSW_CMD_BOARDINFO_PSID_LEN);
+
+/* QUERY_AQ_CAP - Query Asynchronous Queues Capabilities
+ * -----------------------------------------------------
+ * OpMod == 0 (N/A), INMmod == 0 (N/A)
+ * -----------------------------------
+ * The QUERY_AQ_CAP command returns the device asynchronous queues
+ * capabilities supported.
+ */
+
+static inline int mlxsw_cmd_query_aq_cap(struct mlxsw_core *mlxsw_core,
+					 char *out_mbox)
+{
+	return mlxsw_cmd_exec_out(mlxsw_core, MLXSW_CMD_OPCODE_QUERY_AQ_CAP,
+				  0, 0, false, out_mbox, MLXSW_CMD_MBOX_SIZE);
+}
+
+/* cmd_mbox_query_aq_cap_log_max_sdq_sz
+ * Log (base 2) of max WQEs allowed on SDQ.
+ */
+MLXSW_ITEM32(cmd_mbox, query_aq_cap, log_max_sdq_sz, 0x00, 24, 8);
+
+/* cmd_mbox_query_aq_cap_max_num_sdqs
+ * Maximum number of SDQs.
+ */
+MLXSW_ITEM32(cmd_mbox, query_aq_cap, max_num_sdqs, 0x00, 0, 8);
+
+/* cmd_mbox_query_aq_cap_log_max_rdq_sz
+ * Log (base 2) of max WQEs allowed on RDQ.
+ */
+MLXSW_ITEM32(cmd_mbox, query_aq_cap, log_max_rdq_sz, 0x04, 24, 8);
+
+/* cmd_mbox_query_aq_cap_max_num_rdqs
+ * Maximum number of RDQs.
+ */
+MLXSW_ITEM32(cmd_mbox, query_aq_cap, max_num_rdqs, 0x04, 0, 8);
+
+/* cmd_mbox_query_aq_cap_log_max_cq_sz
+ * Log (base 2) of the Maximum CQEs allowed in a CQ for CQEv0 and CQEv1.
+ */
+MLXSW_ITEM32(cmd_mbox, query_aq_cap, log_max_cq_sz, 0x08, 24, 8);
+
+/* cmd_mbox_query_aq_cap_log_max_cqv2_sz
+ * Log (base 2) of the Maximum CQEs allowed in a CQ for CQEv2.
+ */
+MLXSW_ITEM32(cmd_mbox, query_aq_cap, log_max_cqv2_sz, 0x08, 16, 8);
+
+/* cmd_mbox_query_aq_cap_max_num_cqs
+ * Maximum number of CQs.
+ */
+MLXSW_ITEM32(cmd_mbox, query_aq_cap, max_num_cqs, 0x08, 0, 8);
+
+/* cmd_mbox_query_aq_cap_log_max_eq_sz
+ * Log (base 2) of max EQEs allowed on EQ.
+ */
+MLXSW_ITEM32(cmd_mbox, query_aq_cap, log_max_eq_sz, 0x0C, 24, 8);
+
+/* cmd_mbox_query_aq_cap_max_num_eqs
+ * Maximum number of EQs.
+ */
+MLXSW_ITEM32(cmd_mbox, query_aq_cap, max_num_eqs, 0x0C, 0, 8);
+
+/* cmd_mbox_query_aq_cap_max_sg_sq
+ * The maximum S/G list elements in an DSQ. DSQ must not contain
+ * more S/G entries than indicated here.
+ */
+MLXSW_ITEM32(cmd_mbox, query_aq_cap, max_sg_sq, 0x10, 8, 8);
+
+/* cmd_mbox_query_aq_cap_
+ * The maximum S/G list elements in an DRQ. DRQ must not contain
+ * more S/G entries than indicated here.
+ */
+MLXSW_ITEM32(cmd_mbox, query_aq_cap, max_sg_rq, 0x10, 0, 8);
+
+/* MAP_FA - Map Firmware Area
+ * --------------------------
+ * OpMod == 0 (N/A), INMmod == Number of VPM entries
+ * -------------------------------------------------
+ * The MAP_FA command passes physical pages to the switch. These pages
+ * are used to store the device firmware. MAP_FA can be executed multiple
+ * times until all the firmware area is mapped (the size that should be
+ * mapped is retrieved through the QUERY_FW command). All required pages
+ * must be mapped to finish the initialization phase. Physical memory
+ * passed in this command must be pinned.
+ */
+
+#define MLXSW_CMD_MAP_FA_VPM_ENTRIES_MAX 32
+
+static inline int mlxsw_cmd_map_fa(struct mlxsw_core *mlxsw_core,
+				   char *in_mbox, u32 vpm_entries_count)
+{
+	return mlxsw_cmd_exec_in(mlxsw_core, MLXSW_CMD_OPCODE_MAP_FA,
+				 0, vpm_entries_count,
+				 in_mbox, MLXSW_CMD_MBOX_SIZE);
+}
+
+/* cmd_mbox_map_fa_pa
+ * Physical Address.
+ */
+MLXSW_ITEM64_INDEXED(cmd_mbox, map_fa, pa, 0x00, 12, 52, 0x08, 0x00, true);
+
+/* cmd_mbox_map_fa_log2size
+ * Log (base 2) of the size in 4KB pages of the physical and contiguous memory
+ * that starts at PA_L/H.
+ */
+MLXSW_ITEM32_INDEXED(cmd_mbox, map_fa, log2size, 0x00, 0, 5, 0x08, 0x04, false);
+
+/* UNMAP_FA - Unmap Firmware Area
+ * ------------------------------
+ * OpMod == 0 (N/A), INMmod == 0 (N/A)
+ * -----------------------------------
+ * The UNMAP_FA command unload the firmware and unmaps all the
+ * firmware area. After this command is completed the device will not access
+ * the pages that were mapped to the firmware area. After executing UNMAP_FA
+ * command, software reset must be done prior to execution of MAP_FW command.
+ */
+
+static inline int mlxsw_cmd_unmap_fa(struct mlxsw_core *mlxsw_core)
+{
+	return mlxsw_cmd_exec_none(mlxsw_core, MLXSW_CMD_OPCODE_UNMAP_FA, 0, 0);
+}
+
+/* QUERY_RESOURCES - Query chip resources
+ * --------------------------------------
+ * OpMod == 0 (N/A) , INMmod is index
+ * ----------------------------------
+ * The QUERY_RESOURCES command retrieves information related to chip resources
+ * by resource ID. Every command returns 32 entries. INmod is being use as base.
+ * for example, index 1 will return entries 32-63. When the tables end and there
+ * are no more sources in the table, will return resource id 0xFFF to indicate
+ * it.
+ */
+
+#define MLXSW_CMD_QUERY_RESOURCES_TABLE_END_ID 0xffff
+#define MLXSW_CMD_QUERY_RESOURCES_MAX_QUERIES 100
+#define MLXSW_CMD_QUERY_RESOURCES_PER_QUERY 32
+
+static inline int mlxsw_cmd_query_resources(struct mlxsw_core *mlxsw_core,
+					    char *out_mbox, int index)
+{
+	return mlxsw_cmd_exec_out(mlxsw_core, MLXSW_CMD_OPCODE_QUERY_RESOURCES,
+				  0, index, false, out_mbox,
+				  MLXSW_CMD_MBOX_SIZE);
+}
+
+/* cmd_mbox_query_resource_id
+ * The resource id. 0xFFFF indicates table's end.
+ */
+MLXSW_ITEM32_INDEXED(cmd_mbox, query_resource, id, 0x00, 16, 16, 0x8, 0, false);
+
+/* cmd_mbox_query_resource_data
+ * The resource
+ */
+MLXSW_ITEM64_INDEXED(cmd_mbox, query_resource, data,
+		     0x00, 0, 40, 0x8, 0, false);
+
+/* CONFIG_PROFILE (Set) - Configure Switch Profile
+ * ------------------------------
+ * OpMod == 1 (Set), INMmod == 0 (N/A)
+ * -----------------------------------
+ * The CONFIG_PROFILE command sets the switch profile. The command can be
+ * executed on the device only once at startup in order to allocate and
+ * configure all switch resources and prepare it for operational mode.
+ * It is not possible to change the device profile after the chip is
+ * in operational mode.
+ * Failure of the CONFIG_PROFILE command leaves the hardware in an indeterminate
+ * state therefore it is required to perform software reset to the device
+ * following an unsuccessful completion of the command. It is required
+ * to perform software reset to the device to change an existing profile.
+ */
+
+static inline int mlxsw_cmd_config_profile_set(struct mlxsw_core *mlxsw_core,
+					       char *in_mbox)
+{
+	return mlxsw_cmd_exec_in(mlxsw_core, MLXSW_CMD_OPCODE_CONFIG_PROFILE,
+				 1, 0, in_mbox, MLXSW_CMD_MBOX_SIZE);
+}
+
+/* cmd_mbox_config_profile_set_max_vepa_channels
+ * Capability bit. Setting a bit to 1 configures the profile
+ * according to the mailbox contents.
+ */
+MLXSW_ITEM32(cmd_mbox, config_profile, set_max_vepa_channels, 0x0C, 0, 1);
+
+/* cmd_mbox_config_profile_set_max_lag
+ * Capability bit. Setting a bit to 1 configures the profile
+ * according to the mailbox contents.
+ */
+MLXSW_ITEM32(cmd_mbox, config_profile, set_max_lag, 0x0C, 1, 1);
+
+/* cmd_mbox_config_profile_set_max_port_per_lag
+ * Capability bit. Setting a bit to 1 configures the profile
+ * according to the mailbox contents.
+ */
+MLXSW_ITEM32(cmd_mbox, config_profile, set_max_port_per_lag, 0x0C, 2, 1);
+
+/* cmd_mbox_config_profile_set_max_mid
+ * Capability bit. Setting a bit to 1 configures the profile
+ * according to the mailbox contents.
+ */
+MLXSW_ITEM32(cmd_mbox, config_profile, set_max_mid, 0x0C, 3, 1);
+
+/* cmd_mbox_config_profile_set_max_pgt
+ * Capability bit. Setting a bit to 1 configures the profile
+ * according to the mailbox contents.
+ */
+MLXSW_ITEM32(cmd_mbox, config_profile, set_max_pgt, 0x0C, 4, 1);
+
+/* cmd_mbox_config_profile_set_max_system_port
+ * Capability bit. Setting a bit to 1 configures the profile
+ * according to the mailbox contents.
+ */
+MLXSW_ITEM32(cmd_mbox, config_profile, set_max_system_port, 0x0C, 5, 1);
+
+/* cmd_mbox_config_profile_set_max_vlan_groups
+ * Capability bit. Setting a bit to 1 configures the profile
+ * according to the mailbox contents.
+ */
+MLXSW_ITEM32(cmd_mbox, config_profile, set_max_vlan_groups, 0x0C, 6, 1);
+
+/* cmd_mbox_config_profile_set_max_regions
+ * Capability bit. Setting a bit to 1 configures the profile
+ * according to the mailbox contents.
+ */
+MLXSW_ITEM32(cmd_mbox, config_profile, set_max_regions, 0x0C, 7, 1);
+
+/* cmd_mbox_config_profile_set_flood_mode
+ * Capability bit. Setting a bit to 1 configures the profile
+ * according to the mailbox contents.
+ */
+MLXSW_ITEM32(cmd_mbox, config_profile, set_flood_mode, 0x0C, 8, 1);
+
+/* cmd_mbox_config_profile_set_max_flood_tables
+ * Capability bit. Setting a bit to 1 configures the profile
+ * according to the mailbox contents.
+ */
+MLXSW_ITEM32(cmd_mbox, config_profile, set_flood_tables, 0x0C, 9, 1);
+
+/* cmd_mbox_config_profile_set_max_ib_mc
+ * Capability bit. Setting a bit to 1 configures the profile
+ * according to the mailbox contents.
+ */
+MLXSW_ITEM32(cmd_mbox, config_profile, set_max_ib_mc, 0x0C, 12, 1);
+
+/* cmd_mbox_config_profile_set_max_pkey
+ * Capability bit. Setting a bit to 1 configures the profile
+ * according to the mailbox contents.
+ */
+MLXSW_ITEM32(cmd_mbox, config_profile, set_max_pkey, 0x0C, 13, 1);
+
+/* cmd_mbox_config_profile_set_adaptive_routing_group_cap
+ * Capability bit. Setting a bit to 1 configures the profile
+ * according to the mailbox contents.
+ */
+MLXSW_ITEM32(cmd_mbox, config_profile,
+	     set_adaptive_routing_group_cap, 0x0C, 14, 1);
+
+/* cmd_mbox_config_profile_set_ar_sec
+ * Capability bit. Setting a bit to 1 configures the profile
+ * according to the mailbox contents.
+ */
+MLXSW_ITEM32(cmd_mbox, config_profile, set_ar_sec, 0x0C, 15, 1);
+
+/* cmd_mbox_config_set_ubridge
+ * Capability bit. Setting a bit to 1 configures the profile
+ * according to the mailbox contents.
+ */
+MLXSW_ITEM32(cmd_mbox, config_profile, set_ubridge, 0x0C, 22, 1);
+
+/* cmd_mbox_config_set_kvd_linear_size
+ * Capability bit. Setting a bit to 1 configures the profile
+ * according to the mailbox contents.
+ */
+MLXSW_ITEM32(cmd_mbox, config_profile, set_kvd_linear_size, 0x0C, 24, 1);
+
+/* cmd_mbox_config_set_kvd_hash_single_size
+ * Capability bit. Setting a bit to 1 configures the profile
+ * according to the mailbox contents.
+ */
+MLXSW_ITEM32(cmd_mbox, config_profile, set_kvd_hash_single_size, 0x0C, 25, 1);
+
+/* cmd_mbox_config_set_kvd_hash_double_size
+ * Capability bit. Setting a bit to 1 configures the profile
+ * according to the mailbox contents.
+ */
+MLXSW_ITEM32(cmd_mbox, config_profile, set_kvd_hash_double_size, 0x0C, 26, 1);
+
+/* cmd_mbox_config_set_cqe_version
+ * Capability bit. Setting a bit to 1 configures the profile
+ * according to the mailbox contents.
+ */
+MLXSW_ITEM32(cmd_mbox, config_profile, set_cqe_version, 0x08, 0, 1);
+
+/* cmd_mbox_config_set_cqe_time_stamp_type
+ * Capability bit. Setting a bit to 1 configures the profile
+ * according to the mailbox contents.
+ */
+MLXSW_ITEM32(cmd_mbox, config_profile, set_cqe_time_stamp_type, 0x08, 2, 1);
+
+/* cmd_mbox_config_profile_max_vepa_channels
+ * Maximum number of VEPA channels per port (0 through 16)
+ * 0 - multi-channel VEPA is disabled
+ */
+MLXSW_ITEM32(cmd_mbox, config_profile, max_vepa_channels, 0x10, 0, 8);
+
+/* cmd_mbox_config_profile_max_lag
+ * Maximum number of LAG IDs requested.
+ * Reserved when Spectrum-1/2/3, supported from Spectrum-4 and above.
+ * For Spectrum-4, firmware sets 128 for values between 1-128 and 256 for values
+ * between 129-256.
+ */
+MLXSW_ITEM32(cmd_mbox, config_profile, max_lag, 0x14, 0, 16);
+
+/* cmd_mbox_config_profile_max_port_per_lag
+ * Maximum number of ports per LAG requested.
+ */
+MLXSW_ITEM32(cmd_mbox, config_profile, max_port_per_lag, 0x18, 0, 16);
+
+/* cmd_mbox_config_profile_max_mid
+ * Maximum Multicast IDs.
+ * Multicast IDs are allocated from 0 to max_mid-1
+ */
+MLXSW_ITEM32(cmd_mbox, config_profile, max_mid, 0x1C, 0, 16);
+
+/* cmd_mbox_config_profile_max_pgt
+ * Maximum records in the Port Group Table per Switch Partition.
+ * Port Group Table indexes are from 0 to max_pgt-1
+ */
+MLXSW_ITEM32(cmd_mbox, config_profile, max_pgt, 0x20, 0, 16);
+
+/* cmd_mbox_config_profile_max_system_port
+ * The maximum number of system ports that can be allocated.
+ */
+MLXSW_ITEM32(cmd_mbox, config_profile, max_system_port, 0x24, 0, 16);
+
+/* cmd_mbox_config_profile_max_vlan_groups
+ * Maximum number VLAN Groups for VLAN binding.
+ */
+MLXSW_ITEM32(cmd_mbox, config_profile, max_vlan_groups, 0x28, 0, 12);
+
+/* cmd_mbox_config_profile_max_regions
+ * Maximum number of TCAM Regions.
+ */
+MLXSW_ITEM32(cmd_mbox, config_profile, max_regions, 0x2C, 0, 16);
+
+/* cmd_mbox_config_profile_max_flood_tables
+ * Maximum number of single-entry flooding tables. Different flooding tables
+ * can be associated with different packet types.
+ */
+MLXSW_ITEM32(cmd_mbox, config_profile, max_flood_tables, 0x30, 16, 4);
+
+/* cmd_mbox_config_profile_max_vid_flood_tables
+ * Maximum number of per-vid flooding tables. Flooding tables are associated
+ * to the different packet types for the different switch partitions.
+ * Table size is 4K entries covering all VID space.
+ */
+MLXSW_ITEM32(cmd_mbox, config_profile, max_vid_flood_tables, 0x30, 8, 4);
+
+enum mlxsw_cmd_mbox_config_profile_flood_mode {
+	/* Mixed mode, where:
+	 * max_flood_tables indicates the number of single-entry tables.
+	 * max_vid_flood_tables indicates the number of per-VID tables.
+	 * max_fid_offset_flood_tables indicates the number of FID-offset
+	 * tables. max_fid_flood_tables indicates the number of per-FID tables.
+	 * Reserved when unified bridge model is used.
+	 */
+	MLXSW_CMD_MBOX_CONFIG_PROFILE_FLOOD_MODE_MIXED = 3,
+	/* Controlled flood tables. Reserved when legacy bridge model is
+	 * used.
+	 */
+	MLXSW_CMD_MBOX_CONFIG_PROFILE_FLOOD_MODE_CONTROLLED = 4,
+};
+
+/* cmd_mbox_config_profile_flood_mode
+ * Flooding mode to use.
+ */
+MLXSW_ITEM32(cmd_mbox, config_profile, flood_mode, 0x30, 0, 3);
+
+/* cmd_mbox_config_profile_max_fid_offset_flood_tables
+ * Maximum number of FID-offset flooding tables.
+ */
+MLXSW_ITEM32(cmd_mbox, config_profile,
+	     max_fid_offset_flood_tables, 0x34, 24, 4);
+
+/* cmd_mbox_config_profile_fid_offset_flood_table_size
+ * The size (number of entries) of each FID-offset flood table.
+ */
+MLXSW_ITEM32(cmd_mbox, config_profile,
+	     fid_offset_flood_table_size, 0x34, 0, 16);
+
+/* cmd_mbox_config_profile_max_fid_flood_tables
+ * Maximum number of per-FID flooding tables.
+ *
+ * Note: This flooding tables cover special FIDs only (vFIDs), starting at
+ * FID value 4K and higher.
+ */
+MLXSW_ITEM32(cmd_mbox, config_profile, max_fid_flood_tables, 0x38, 24, 4);
+
+/* cmd_mbox_config_profile_fid_flood_table_size
+ * The size (number of entries) of each per-FID table.
+ */
+MLXSW_ITEM32(cmd_mbox, config_profile, fid_flood_table_size, 0x38, 0, 16);
+
+/* cmd_mbox_config_profile_max_ib_mc
+ * Maximum number of multicast FDB records for InfiniBand
+ * FDB (in 512 chunks) per InfiniBand switch partition.
+ */
+MLXSW_ITEM32(cmd_mbox, config_profile, max_ib_mc, 0x40, 0, 15);
+
+/* cmd_mbox_config_profile_max_pkey
+ * Maximum per port PKEY table size (for PKEY enforcement)
+ */
+MLXSW_ITEM32(cmd_mbox, config_profile, max_pkey, 0x44, 0, 15);
+
+/* cmd_mbox_config_profile_ar_sec
+ * Primary/secondary capability
+ * Describes the number of adaptive routing sub-groups
+ * 0 - disable primary/secondary (single group)
+ * 1 - enable primary/secondary (2 sub-groups)
+ * 2 - 3 sub-groups: Not supported in SwitchX, SwitchX-2
+ * 3 - 4 sub-groups: Not supported in SwitchX, SwitchX-2
+ */
+MLXSW_ITEM32(cmd_mbox, config_profile, ar_sec, 0x4C, 24, 2);
+
+/* cmd_mbox_config_profile_adaptive_routing_group_cap
+ * Adaptive Routing Group Capability. Indicates the number of AR groups
+ * supported. Note that when Primary/secondary is enabled, each
+ * primary/secondary couple consumes 2 adaptive routing entries.
+ */
+MLXSW_ITEM32(cmd_mbox, config_profile, adaptive_routing_group_cap, 0x4C, 0, 16);
+
+/* cmd_mbox_config_profile_arn
+ * Adaptive Routing Notification Enable
+ * Not supported in SwitchX, SwitchX-2
+ */
+MLXSW_ITEM32(cmd_mbox, config_profile, arn, 0x50, 31, 1);
+
+/* cmd_mbox_config_profile_ubridge
+ * Unified Bridge
+ * 0 - non unified bridge
+ * 1 - unified bridge
+ */
+MLXSW_ITEM32(cmd_mbox, config_profile, ubridge, 0x50, 4, 1);
+
+/* cmd_mbox_config_kvd_linear_size
+ * KVD Linear Size
+ * Valid for Spectrum only
+ * Allowed values are 128*N where N=0 or higher
+ */
+MLXSW_ITEM32(cmd_mbox, config_profile, kvd_linear_size, 0x54, 0, 24);
+
+/* cmd_mbox_config_kvd_hash_single_size
+ * KVD Hash single-entries size
+ * Valid for Spectrum only
+ * Allowed values are 128*N where N=0 or higher
+ * Must be greater or equal to cap_min_kvd_hash_single_size
+ * Must be smaller or equal to cap_kvd_size - kvd_linear_size
+ */
+MLXSW_ITEM32(cmd_mbox, config_profile, kvd_hash_single_size, 0x58, 0, 24);
+
+/* cmd_mbox_config_kvd_hash_double_size
+ * KVD Hash double-entries size (units of single-size entries)
+ * Valid for Spectrum only
+ * Allowed values are 128*N where N=0 or higher
+ * Must be either 0 or greater or equal to cap_min_kvd_hash_double_size
+ * Must be smaller or equal to cap_kvd_size - kvd_linear_size
+ */
+MLXSW_ITEM32(cmd_mbox, config_profile, kvd_hash_double_size, 0x5C, 0, 24);
+
+/* cmd_mbox_config_profile_swid_config_mask
+ * Modify Switch Partition Configuration mask. When set, the configu-
+ * ration value for the Switch Partition are taken from the mailbox.
+ * When clear, the current configuration values are used.
+ * Bit 0 - set type
+ * Bit 1 - properties
+ * Other - reserved
+ */
+MLXSW_ITEM32_INDEXED(cmd_mbox, config_profile, swid_config_mask,
+		     0x60, 24, 8, 0x08, 0x00, false);
+
+/* cmd_mbox_config_profile_swid_config_type
+ * Switch Partition type.
+ * 0000 - disabled (Switch Partition does not exist)
+ * 0001 - InfiniBand
+ * 0010 - Ethernet
+ * 1000 - router port (SwitchX-2 only)
+ * Other - reserved
+ */
+MLXSW_ITEM32_INDEXED(cmd_mbox, config_profile, swid_config_type,
+		     0x60, 20, 4, 0x08, 0x00, false);
+
+/* cmd_mbox_config_profile_swid_config_properties
+ * Switch Partition properties.
+ */
+MLXSW_ITEM32_INDEXED(cmd_mbox, config_profile, swid_config_properties,
+		     0x60, 0, 8, 0x08, 0x00, false);
+
+enum mlxsw_cmd_mbox_config_profile_cqe_time_stamp_type {
+	/* uSec - 1.024uSec (default). Only bits 15:0 are valid. */
+	MLXSW_CMD_MBOX_CONFIG_PROFILE_CQE_TIME_STAMP_TYPE_USEC,
+	/* FRC - Free Running Clock, units of 1nSec.
+	 * Reserved when SwitchX/-2, Switch-IB/2 and Spectrum-1.
+	 */
+	MLXSW_CMD_MBOX_CONFIG_PROFILE_CQE_TIME_STAMP_TYPE_FRC,
+	/* UTC. time_stamp[37:30] = Sec, time_stamp[29:0] = nSec.
+	 * Reserved when SwitchX/2, Switch-IB/2 and Spectrum-1.
+	 */
+	MLXSW_CMD_MBOX_CONFIG_PROFILE_CQE_TIME_STAMP_TYPE_UTC,
+};
+
+/* cmd_mbox_config_profile_cqe_time_stamp_type
+ * CQE time_stamp_type for non-mirror-packets.
+ * Configured if set_cqe_time_stamp_type is set.
+ * Reserved when SwitchX/-2, Switch-IB/2 and Spectrum-1.
+ */
+MLXSW_ITEM32(cmd_mbox, config_profile, cqe_time_stamp_type, 0xB0, 8, 2);
+
+/* cmd_mbox_config_profile_cqe_version
+ * CQE version:
+ * 0: CQE version is 0
+ * 1: CQE version is either 1 or 2
+ * CQE ver 1 or 2 is configured by Completion Queue Context field cqe_ver.
+ */
+MLXSW_ITEM32(cmd_mbox, config_profile, cqe_version, 0xB0, 0, 8);
+
+/* ACCESS_REG - Access EMAD Supported Register
+ * ----------------------------------
+ * OpMod == 0 (N/A), INMmod == 0 (N/A)
+ * -------------------------------------
+ * The ACCESS_REG command supports accessing device registers. This access
+ * is mainly used for bootstrapping.
+ */
+
+static inline int mlxsw_cmd_access_reg(struct mlxsw_core *mlxsw_core,
+				       bool reset_ok,
+				       char *in_mbox, char *out_mbox)
+{
+	return mlxsw_cmd_exec(mlxsw_core, MLXSW_CMD_OPCODE_ACCESS_REG,
+			      0, 0, false, reset_ok,
+			      in_mbox, MLXSW_CMD_MBOX_SIZE,
+			      out_mbox, MLXSW_CMD_MBOX_SIZE);
+}
+
+/* SW2HW_DQ - Software to Hardware DQ
+ * ----------------------------------
+ * OpMod == 0 (send DQ) / OpMod == 1 (receive DQ)
+ * INMmod == DQ number
+ * ----------------------------------------------
+ * The SW2HW_DQ command transitions a descriptor queue from software to
+ * hardware ownership. The command enables posting WQEs and ringing DoorBells
+ * on the descriptor queue.
+ */
+
+static inline int __mlxsw_cmd_sw2hw_dq(struct mlxsw_core *mlxsw_core,
+				       char *in_mbox, u32 dq_number,
+				       u8 opcode_mod)
+{
+	return mlxsw_cmd_exec_in(mlxsw_core, MLXSW_CMD_OPCODE_SW2HW_DQ,
+				 opcode_mod, dq_number,
+				 in_mbox, MLXSW_CMD_MBOX_SIZE);
+}
+
+enum {
+	MLXSW_CMD_OPCODE_MOD_SDQ = 0,
+	MLXSW_CMD_OPCODE_MOD_RDQ = 1,
+};
+
+static inline int mlxsw_cmd_sw2hw_sdq(struct mlxsw_core *mlxsw_core,
+				      char *in_mbox, u32 dq_number)
+{
+	return __mlxsw_cmd_sw2hw_dq(mlxsw_core, in_mbox, dq_number,
+				    MLXSW_CMD_OPCODE_MOD_SDQ);
+}
+
+static inline int mlxsw_cmd_sw2hw_rdq(struct mlxsw_core *mlxsw_core,
+				      char *in_mbox, u32 dq_number)
+{
+	return __mlxsw_cmd_sw2hw_dq(mlxsw_core, in_mbox, dq_number,
+				    MLXSW_CMD_OPCODE_MOD_RDQ);
+}
+
+/* cmd_mbox_sw2hw_dq_cq
+ * Number of the CQ that this Descriptor Queue reports completions to.
+ */
+MLXSW_ITEM32(cmd_mbox, sw2hw_dq, cq, 0x00, 24, 8);
+
+enum mlxsw_cmd_mbox_sw2hw_dq_sdq_lp {
+	MLXSW_CMD_MBOX_SW2HW_DQ_SDQ_LP_WQE,
+	MLXSW_CMD_MBOX_SW2HW_DQ_SDQ_LP_IGNORE_WQE,
+};
+
+/* cmd_mbox_sw2hw_dq_sdq_lp
+ * SDQ local Processing
+ * 0: local processing by wqe.lp
+ * 1: local processing (ignoring wqe.lp)
+ */
+MLXSW_ITEM32(cmd_mbox, sw2hw_dq, sdq_lp, 0x00, 23, 1);
+
+/* cmd_mbox_sw2hw_dq_sdq_tclass
+ * SDQ: CPU Egress TClass
+ * RDQ: Reserved
+ */
+MLXSW_ITEM32(cmd_mbox, sw2hw_dq, sdq_tclass, 0x00, 16, 6);
+
+/* cmd_mbox_sw2hw_dq_log2_dq_sz
+ * Log (base 2) of the Descriptor Queue size in 4KB pages.
+ */
+MLXSW_ITEM32(cmd_mbox, sw2hw_dq, log2_dq_sz, 0x00, 0, 6);
+
+/* cmd_mbox_sw2hw_dq_pa
+ * Physical Address.
+ */
+MLXSW_ITEM64_INDEXED(cmd_mbox, sw2hw_dq, pa, 0x10, 12, 52, 0x08, 0x00, true);
+
+/* HW2SW_DQ - Hardware to Software DQ
+ * ----------------------------------
+ * OpMod == 0 (send DQ) / OpMod == 1 (receive DQ)
+ * INMmod == DQ number
+ * ----------------------------------------------
+ * The HW2SW_DQ command transitions a descriptor queue from hardware to
+ * software ownership. Incoming packets on the DQ are silently discarded,
+ * SW should not post descriptors on nonoperational DQs.
+ */
+
+static inline int __mlxsw_cmd_hw2sw_dq(struct mlxsw_core *mlxsw_core,
+				       u32 dq_number, u8 opcode_mod)
+{
+	return mlxsw_cmd_exec_none(mlxsw_core, MLXSW_CMD_OPCODE_HW2SW_DQ,
+				   opcode_mod, dq_number);
+}
+
+static inline int mlxsw_cmd_hw2sw_sdq(struct mlxsw_core *mlxsw_core,
+				      u32 dq_number)
+{
+	return __mlxsw_cmd_hw2sw_dq(mlxsw_core, dq_number,
+				    MLXSW_CMD_OPCODE_MOD_SDQ);
+}
+
+static inline int mlxsw_cmd_hw2sw_rdq(struct mlxsw_core *mlxsw_core,
+				      u32 dq_number)
+{
+	return __mlxsw_cmd_hw2sw_dq(mlxsw_core, dq_number,
+				    MLXSW_CMD_OPCODE_MOD_RDQ);
+}
+
+/* 2ERR_DQ - To Error DQ
+ * ---------------------
+ * OpMod == 0 (send DQ) / OpMod == 1 (receive DQ)
+ * INMmod == DQ number
+ * ----------------------------------------------
+ * The 2ERR_DQ command transitions the DQ into the error state from the state
+ * in which it has been. While the command is executed, some in-process
+ * descriptors may complete. Once the DQ transitions into the error state,
+ * if there are posted descriptors on the RDQ/SDQ, the hardware writes
+ * a completion with error (flushed) for all descriptors posted in the RDQ/SDQ.
+ * When the command is completed successfully, the DQ is already in
+ * the error state.
+ */
+
+static inline int __mlxsw_cmd_2err_dq(struct mlxsw_core *mlxsw_core,
+				      u32 dq_number, u8 opcode_mod)
+{
+	return mlxsw_cmd_exec_none(mlxsw_core, MLXSW_CMD_OPCODE_2ERR_DQ,
+				   opcode_mod, dq_number);
+}
+
+static inline int mlxsw_cmd_2err_sdq(struct mlxsw_core *mlxsw_core,
+				     u32 dq_number)
+{
+	return __mlxsw_cmd_2err_dq(mlxsw_core, dq_number,
+				   MLXSW_CMD_OPCODE_MOD_SDQ);
+}
+
+static inline int mlxsw_cmd_2err_rdq(struct mlxsw_core *mlxsw_core,
+				     u32 dq_number)
+{
+	return __mlxsw_cmd_2err_dq(mlxsw_core, dq_number,
+				   MLXSW_CMD_OPCODE_MOD_RDQ);
+}
+
+/* QUERY_DQ - Query DQ
+ * ---------------------
+ * OpMod == 0 (send DQ) / OpMod == 1 (receive DQ)
+ * INMmod == DQ number
+ * ----------------------------------------------
+ * The QUERY_DQ command retrieves a snapshot of DQ parameters from the hardware.
+ *
+ * Note: Output mailbox has the same format as SW2HW_DQ.
+ */
+
+static inline int __mlxsw_cmd_query_dq(struct mlxsw_core *mlxsw_core,
+				       char *out_mbox, u32 dq_number,
+				       u8 opcode_mod)
+{
+	return mlxsw_cmd_exec_out(mlxsw_core, MLXSW_CMD_OPCODE_2ERR_DQ,
+				  opcode_mod, dq_number, false,
+				  out_mbox, MLXSW_CMD_MBOX_SIZE);
+}
+
+static inline int mlxsw_cmd_query_sdq(struct mlxsw_core *mlxsw_core,
+				      char *out_mbox, u32 dq_number)
+{
+	return __mlxsw_cmd_query_dq(mlxsw_core, out_mbox, dq_number,
+				    MLXSW_CMD_OPCODE_MOD_SDQ);
+}
+
+static inline int mlxsw_cmd_query_rdq(struct mlxsw_core *mlxsw_core,
+				      char *out_mbox, u32 dq_number)
+{
+	return __mlxsw_cmd_query_dq(mlxsw_core, out_mbox, dq_number,
+				    MLXSW_CMD_OPCODE_MOD_RDQ);
+}
+
+/* SW2HW_CQ - Software to Hardware CQ
+ * ----------------------------------
+ * OpMod == 0 (N/A), INMmod == CQ number
+ * -------------------------------------
+ * The SW2HW_CQ command transfers ownership of a CQ context entry from software
+ * to hardware. The command takes the CQ context entry from the input mailbox
+ * and stores it in the CQC in the ownership of the hardware. The command fails
+ * if the requested CQC entry is already in the ownership of the hardware.
+ */
+
+static inline int mlxsw_cmd_sw2hw_cq(struct mlxsw_core *mlxsw_core,
+				     char *in_mbox, u32 cq_number)
+{
+	return mlxsw_cmd_exec_in(mlxsw_core, MLXSW_CMD_OPCODE_SW2HW_CQ,
+				 0, cq_number, in_mbox, MLXSW_CMD_MBOX_SIZE);
+}
+
+enum mlxsw_cmd_mbox_sw2hw_cq_cqe_ver {
+	MLXSW_CMD_MBOX_SW2HW_CQ_CQE_VER_1,
+	MLXSW_CMD_MBOX_SW2HW_CQ_CQE_VER_2,
+};
+
+/* cmd_mbox_sw2hw_cq_cqe_ver
+ * CQE Version.
+ */
+MLXSW_ITEM32(cmd_mbox, sw2hw_cq, cqe_ver, 0x00, 28, 4);
+
+/* cmd_mbox_sw2hw_cq_c_eqn
+ * Event Queue this CQ reports completion events to.
+ */
+MLXSW_ITEM32(cmd_mbox, sw2hw_cq, c_eqn, 0x00, 24, 1);
+
+/* cmd_mbox_sw2hw_cq_st
+ * Event delivery state machine
+ * 0x0 - FIRED
+ * 0x1 - ARMED (Request for Notification)
+ */
+MLXSW_ITEM32(cmd_mbox, sw2hw_cq, st, 0x00, 8, 1);
+
+/* cmd_mbox_sw2hw_cq_log_cq_size
+ * Log (base 2) of the CQ size (in entries).
+ */
+MLXSW_ITEM32(cmd_mbox, sw2hw_cq, log_cq_size, 0x00, 0, 4);
+
+/* cmd_mbox_sw2hw_cq_producer_counter
+ * Producer Counter. The counter is incremented for each CQE that is
+ * written by the HW to the CQ.
+ * Maintained by HW (valid for the QUERY_CQ command only)
+ */
+MLXSW_ITEM32(cmd_mbox, sw2hw_cq, producer_counter, 0x04, 0, 16);
+
+/* cmd_mbox_sw2hw_cq_pa
+ * Physical Address.
+ */
+MLXSW_ITEM64_INDEXED(cmd_mbox, sw2hw_cq, pa, 0x10, 11, 53, 0x08, 0x00, true);
+
+/* HW2SW_CQ - Hardware to Software CQ
+ * ----------------------------------
+ * OpMod == 0 (N/A), INMmod == CQ number
+ * -------------------------------------
+ * The HW2SW_CQ command transfers ownership of a CQ context entry from hardware
+ * to software. The CQC entry is invalidated as a result of this command.
+ */
+
+static inline int mlxsw_cmd_hw2sw_cq(struct mlxsw_core *mlxsw_core,
+				     u32 cq_number)
+{
+	return mlxsw_cmd_exec_none(mlxsw_core, MLXSW_CMD_OPCODE_HW2SW_CQ,
+				   0, cq_number);
+}
+
+/* QUERY_CQ - Query CQ
+ * ----------------------------------
+ * OpMod == 0 (N/A), INMmod == CQ number
+ * -------------------------------------
+ * The QUERY_CQ command retrieves a snapshot of the current CQ context entry.
+ * The command stores the snapshot in the output mailbox in the software format.
+ * Note that the CQ context state and values are not affected by the QUERY_CQ
+ * command. The QUERY_CQ command is for debug purposes only.
+ *
+ * Note: Output mailbox has the same format as SW2HW_CQ.
+ */
+
+static inline int mlxsw_cmd_query_cq(struct mlxsw_core *mlxsw_core,
+				     char *out_mbox, u32 cq_number)
+{
+	return mlxsw_cmd_exec_out(mlxsw_core, MLXSW_CMD_OPCODE_QUERY_CQ,
+				  0, cq_number, false,
+				  out_mbox, MLXSW_CMD_MBOX_SIZE);
+}
+
+/* SW2HW_EQ - Software to Hardware EQ
+ * ----------------------------------
+ * OpMod == 0 (N/A), INMmod == EQ number
+ * -------------------------------------
+ * The SW2HW_EQ command transfers ownership of an EQ context entry from software
+ * to hardware. The command takes the EQ context entry from the input mailbox
+ * and stores it in the EQC in the ownership of the hardware. The command fails
+ * if the requested EQC entry is already in the ownership of the hardware.
+ */
+
+static inline int mlxsw_cmd_sw2hw_eq(struct mlxsw_core *mlxsw_core,
+				     char *in_mbox, u32 eq_number)
+{
+	return mlxsw_cmd_exec_in(mlxsw_core, MLXSW_CMD_OPCODE_SW2HW_EQ,
+				 0, eq_number, in_mbox, MLXSW_CMD_MBOX_SIZE);
+}
+
+/* cmd_mbox_sw2hw_eq_int_msix
+ * When set, MSI-X cycles will be generated by this EQ.
+ * When cleared, an interrupt will be generated by this EQ.
+ */
+MLXSW_ITEM32(cmd_mbox, sw2hw_eq, int_msix, 0x00, 24, 1);
+
+/* cmd_mbox_sw2hw_eq_st
+ * Event delivery state machine
+ * 0x0 - FIRED
+ * 0x1 - ARMED (Request for Notification)
+ * 0x11 - Always ARMED
+ * other - reserved
+ */
+MLXSW_ITEM32(cmd_mbox, sw2hw_eq, st, 0x00, 8, 2);
+
+/* cmd_mbox_sw2hw_eq_log_eq_size
+ * Log (base 2) of the EQ size (in entries).
+ */
+MLXSW_ITEM32(cmd_mbox, sw2hw_eq, log_eq_size, 0x00, 0, 4);
+
+/* cmd_mbox_sw2hw_eq_producer_counter
+ * Producer Counter. The counter is incremented for each EQE that is written
+ * by the HW to the EQ.
+ * Maintained by HW (valid for the QUERY_EQ command only)
+ */
+MLXSW_ITEM32(cmd_mbox, sw2hw_eq, producer_counter, 0x04, 0, 16);
+
+/* cmd_mbox_sw2hw_eq_pa
+ * Physical Address.
+ */
+MLXSW_ITEM64_INDEXED(cmd_mbox, sw2hw_eq, pa, 0x10, 11, 53, 0x08, 0x00, true);
+
+/* HW2SW_EQ - Hardware to Software EQ
+ * ----------------------------------
+ * OpMod == 0 (N/A), INMmod == EQ number
+ * -------------------------------------
+ */
+
+static inline int mlxsw_cmd_hw2sw_eq(struct mlxsw_core *mlxsw_core,
+				     u32 eq_number)
+{
+	return mlxsw_cmd_exec_none(mlxsw_core, MLXSW_CMD_OPCODE_HW2SW_EQ,
+				   0, eq_number);
+}
+
+/* QUERY_EQ - Query EQ
+ * ----------------------------------
+ * OpMod == 0 (N/A), INMmod == EQ number
+ * -------------------------------------
+ *
+ * Note: Output mailbox has the same format as SW2HW_EQ.
+ */
+
+static inline int mlxsw_cmd_query_eq(struct mlxsw_core *mlxsw_core,
+				     char *out_mbox, u32 eq_number)
+{
+	return mlxsw_cmd_exec_out(mlxsw_core, MLXSW_CMD_OPCODE_QUERY_EQ,
+				  0, eq_number, false,
+				  out_mbox, MLXSW_CMD_MBOX_SIZE);
+}
+
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlxsw/core.c b/drivers/net/ethernet/mellanox/mlxsw/core.c
new file mode 100644
index 000000000..e2a985ec2
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/core.c
@@ -0,0 +1,3443 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2015-2018 Mellanox Technologies. All rights reserved */
+
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/device.h>
+#include <linux/export.h>
+#include <linux/err.h>
+#include <linux/if_link.h>
+#include <linux/netdevice.h>
+#include <linux/completion.h>
+#include <linux/skbuff.h>
+#include <linux/etherdevice.h>
+#include <linux/types.h>
+#include <linux/string.h>
+#include <linux/gfp.h>
+#include <linux/random.h>
+#include <linux/jiffies.h>
+#include <linux/mutex.h>
+#include <linux/rcupdate.h>
+#include <linux/slab.h>
+#include <linux/workqueue.h>
+#include <linux/firmware.h>
+#include <asm/byteorder.h>
+#include <net/devlink.h>
+#include <trace/events/devlink.h>
+
+#include "core.h"
+#include "core_env.h"
+#include "item.h"
+#include "cmd.h"
+#include "port.h"
+#include "trap.h"
+#include "emad.h"
+#include "reg.h"
+#include "resources.h"
+#include "../mlxfw/mlxfw.h"
+
+static LIST_HEAD(mlxsw_core_driver_list);
+static DEFINE_SPINLOCK(mlxsw_core_driver_list_lock);
+
+static const char mlxsw_core_driver_name[] = "mlxsw_core";
+
+static struct workqueue_struct *mlxsw_wq;
+static struct workqueue_struct *mlxsw_owq;
+
+struct mlxsw_core_port {
+	struct devlink_port devlink_port;
+	void *port_driver_priv;
+	u16 local_port;
+	struct mlxsw_linecard *linecard;
+};
+
+void *mlxsw_core_port_driver_priv(struct mlxsw_core_port *mlxsw_core_port)
+{
+	return mlxsw_core_port->port_driver_priv;
+}
+EXPORT_SYMBOL(mlxsw_core_port_driver_priv);
+
+static bool mlxsw_core_port_check(struct mlxsw_core_port *mlxsw_core_port)
+{
+	return mlxsw_core_port->port_driver_priv != NULL;
+}
+
+struct mlxsw_core {
+	struct mlxsw_driver *driver;
+	const struct mlxsw_bus *bus;
+	void *bus_priv;
+	const struct mlxsw_bus_info *bus_info;
+	struct workqueue_struct *emad_wq;
+	struct list_head rx_listener_list;
+	struct list_head event_listener_list;
+	struct list_head irq_event_handler_list;
+	struct mutex irq_event_handler_lock; /* Locks access to handlers list */
+	struct {
+		atomic64_t tid;
+		struct list_head trans_list;
+		spinlock_t trans_list_lock; /* protects trans_list writes */
+		bool use_emad;
+		bool enable_string_tlv;
+	} emad;
+	struct {
+		u16 *mapping; /* lag_id+port_index to local_port mapping */
+	} lag;
+	struct mlxsw_res res;
+	struct mlxsw_hwmon *hwmon;
+	struct mlxsw_thermal *thermal;
+	struct mlxsw_linecards *linecards;
+	struct mlxsw_core_port *ports;
+	unsigned int max_ports;
+	atomic_t active_ports_count;
+	bool fw_flash_in_progress;
+	struct {
+		struct devlink_health_reporter *fw_fatal;
+	} health;
+	struct mlxsw_env *env;
+	unsigned long driver_priv[];
+	/* driver_priv has to be always the last item */
+};
+
+struct mlxsw_linecards *mlxsw_core_linecards(struct mlxsw_core *mlxsw_core)
+{
+	return mlxsw_core->linecards;
+}
+
+void mlxsw_core_linecards_set(struct mlxsw_core *mlxsw_core,
+			      struct mlxsw_linecards *linecards)
+{
+	mlxsw_core->linecards = linecards;
+}
+
+#define MLXSW_PORT_MAX_PORTS_DEFAULT	0x40
+
+static u64 mlxsw_ports_occ_get(void *priv)
+{
+	struct mlxsw_core *mlxsw_core = priv;
+
+	return atomic_read(&mlxsw_core->active_ports_count);
+}
+
+static int mlxsw_core_resources_ports_register(struct mlxsw_core *mlxsw_core)
+{
+	struct devlink *devlink = priv_to_devlink(mlxsw_core);
+	struct devlink_resource_size_params ports_num_params;
+	u32 max_ports;
+
+	max_ports = mlxsw_core->max_ports - 1;
+	devlink_resource_size_params_init(&ports_num_params, max_ports,
+					  max_ports, 1,
+					  DEVLINK_RESOURCE_UNIT_ENTRY);
+
+	return devl_resource_register(devlink,
+				      DEVLINK_RESOURCE_GENERIC_NAME_PORTS,
+				      max_ports, MLXSW_CORE_RESOURCE_PORTS,
+				      DEVLINK_RESOURCE_ID_PARENT_TOP,
+				      &ports_num_params);
+}
+
+static int mlxsw_ports_init(struct mlxsw_core *mlxsw_core, bool reload)
+{
+	struct devlink *devlink = priv_to_devlink(mlxsw_core);
+	int err;
+
+	/* Switch ports are numbered from 1 to queried value */
+	if (MLXSW_CORE_RES_VALID(mlxsw_core, MAX_SYSTEM_PORT))
+		mlxsw_core->max_ports = MLXSW_CORE_RES_GET(mlxsw_core,
+							   MAX_SYSTEM_PORT) + 1;
+	else
+		mlxsw_core->max_ports = MLXSW_PORT_MAX_PORTS_DEFAULT + 1;
+
+	mlxsw_core->ports = kcalloc(mlxsw_core->max_ports,
+				    sizeof(struct mlxsw_core_port), GFP_KERNEL);
+	if (!mlxsw_core->ports)
+		return -ENOMEM;
+
+	if (!reload) {
+		err = mlxsw_core_resources_ports_register(mlxsw_core);
+		if (err)
+			goto err_resources_ports_register;
+	}
+	atomic_set(&mlxsw_core->active_ports_count, 0);
+	devl_resource_occ_get_register(devlink, MLXSW_CORE_RESOURCE_PORTS,
+				       mlxsw_ports_occ_get, mlxsw_core);
+
+	return 0;
+
+err_resources_ports_register:
+	kfree(mlxsw_core->ports);
+	return err;
+}
+
+static void mlxsw_ports_fini(struct mlxsw_core *mlxsw_core, bool reload)
+{
+	struct devlink *devlink = priv_to_devlink(mlxsw_core);
+
+	devl_resource_occ_get_unregister(devlink, MLXSW_CORE_RESOURCE_PORTS);
+	if (!reload)
+		devl_resources_unregister(priv_to_devlink(mlxsw_core));
+
+	kfree(mlxsw_core->ports);
+}
+
+unsigned int mlxsw_core_max_ports(const struct mlxsw_core *mlxsw_core)
+{
+	return mlxsw_core->max_ports;
+}
+EXPORT_SYMBOL(mlxsw_core_max_ports);
+
+int mlxsw_core_max_lag(struct mlxsw_core *mlxsw_core, u16 *p_max_lag)
+{
+	struct mlxsw_driver *driver = mlxsw_core->driver;
+
+	if (driver->profile->used_max_lag) {
+		*p_max_lag = driver->profile->max_lag;
+		return 0;
+	}
+
+	if (!MLXSW_CORE_RES_VALID(mlxsw_core, MAX_LAG))
+		return -EIO;
+
+	*p_max_lag = MLXSW_CORE_RES_GET(mlxsw_core, MAX_LAG);
+	return 0;
+}
+EXPORT_SYMBOL(mlxsw_core_max_lag);
+
+void *mlxsw_core_driver_priv(struct mlxsw_core *mlxsw_core)
+{
+	return mlxsw_core->driver_priv;
+}
+EXPORT_SYMBOL(mlxsw_core_driver_priv);
+
+bool
+mlxsw_core_fw_rev_minor_subminor_validate(const struct mlxsw_fw_rev *rev,
+					  const struct mlxsw_fw_rev *req_rev)
+{
+	return rev->minor > req_rev->minor ||
+	       (rev->minor == req_rev->minor &&
+		rev->subminor >= req_rev->subminor);
+}
+EXPORT_SYMBOL(mlxsw_core_fw_rev_minor_subminor_validate);
+
+struct mlxsw_rx_listener_item {
+	struct list_head list;
+	struct mlxsw_rx_listener rxl;
+	void *priv;
+	bool enabled;
+};
+
+struct mlxsw_event_listener_item {
+	struct list_head list;
+	struct mlxsw_core *mlxsw_core;
+	struct mlxsw_event_listener el;
+	void *priv;
+};
+
+static const u8 mlxsw_core_trap_groups[] = {
+	MLXSW_REG_HTGT_TRAP_GROUP_EMAD,
+	MLXSW_REG_HTGT_TRAP_GROUP_CORE_EVENT,
+};
+
+static int mlxsw_core_trap_groups_set(struct mlxsw_core *mlxsw_core)
+{
+	char htgt_pl[MLXSW_REG_HTGT_LEN];
+	int err;
+	int i;
+
+	if (!(mlxsw_core->bus->features & MLXSW_BUS_F_TXRX))
+		return 0;
+
+	for (i = 0; i < ARRAY_SIZE(mlxsw_core_trap_groups); i++) {
+		mlxsw_reg_htgt_pack(htgt_pl, mlxsw_core_trap_groups[i],
+				    MLXSW_REG_HTGT_INVALID_POLICER,
+				    MLXSW_REG_HTGT_DEFAULT_PRIORITY,
+				    MLXSW_REG_HTGT_DEFAULT_TC);
+		err = mlxsw_reg_write(mlxsw_core, MLXSW_REG(htgt), htgt_pl);
+		if (err)
+			return err;
+	}
+	return 0;
+}
+
+/******************
+ * EMAD processing
+ ******************/
+
+/* emad_eth_hdr_dmac
+ * Destination MAC in EMAD's Ethernet header.
+ * Must be set to 01:02:c9:00:00:01
+ */
+MLXSW_ITEM_BUF(emad, eth_hdr, dmac, 0x00, 6);
+
+/* emad_eth_hdr_smac
+ * Source MAC in EMAD's Ethernet header.
+ * Must be set to 00:02:c9:01:02:03
+ */
+MLXSW_ITEM_BUF(emad, eth_hdr, smac, 0x06, 6);
+
+/* emad_eth_hdr_ethertype
+ * Ethertype in EMAD's Ethernet header.
+ * Must be set to 0x8932
+ */
+MLXSW_ITEM32(emad, eth_hdr, ethertype, 0x0C, 16, 16);
+
+/* emad_eth_hdr_mlx_proto
+ * Mellanox protocol.
+ * Must be set to 0x0.
+ */
+MLXSW_ITEM32(emad, eth_hdr, mlx_proto, 0x0C, 8, 8);
+
+/* emad_eth_hdr_ver
+ * Mellanox protocol version.
+ * Must be set to 0x0.
+ */
+MLXSW_ITEM32(emad, eth_hdr, ver, 0x0C, 4, 4);
+
+/* emad_op_tlv_type
+ * Type of the TLV.
+ * Must be set to 0x1 (operation TLV).
+ */
+MLXSW_ITEM32(emad, op_tlv, type, 0x00, 27, 5);
+
+/* emad_op_tlv_len
+ * Length of the operation TLV in u32.
+ * Must be set to 0x4.
+ */
+MLXSW_ITEM32(emad, op_tlv, len, 0x00, 16, 11);
+
+/* emad_op_tlv_dr
+ * Direct route bit. Setting to 1 indicates the EMAD is a direct route
+ * EMAD. DR TLV must follow.
+ *
+ * Note: Currently not supported and must not be set.
+ */
+MLXSW_ITEM32(emad, op_tlv, dr, 0x00, 15, 1);
+
+/* emad_op_tlv_status
+ * Returned status in case of EMAD response. Must be set to 0 in case
+ * of EMAD request.
+ * 0x0 - success
+ * 0x1 - device is busy. Requester should retry
+ * 0x2 - Mellanox protocol version not supported
+ * 0x3 - unknown TLV
+ * 0x4 - register not supported
+ * 0x5 - operation class not supported
+ * 0x6 - EMAD method not supported
+ * 0x7 - bad parameter (e.g. port out of range)
+ * 0x8 - resource not available
+ * 0x9 - message receipt acknowledgment. Requester should retry
+ * 0x70 - internal error
+ */
+MLXSW_ITEM32(emad, op_tlv, status, 0x00, 8, 7);
+
+/* emad_op_tlv_register_id
+ * Register ID of register within register TLV.
+ */
+MLXSW_ITEM32(emad, op_tlv, register_id, 0x04, 16, 16);
+
+/* emad_op_tlv_r
+ * Response bit. Setting to 1 indicates Response, otherwise request.
+ */
+MLXSW_ITEM32(emad, op_tlv, r, 0x04, 15, 1);
+
+/* emad_op_tlv_method
+ * EMAD method type.
+ * 0x1 - query
+ * 0x2 - write
+ * 0x3 - send (currently not supported)
+ * 0x4 - event
+ */
+MLXSW_ITEM32(emad, op_tlv, method, 0x04, 8, 7);
+
+/* emad_op_tlv_class
+ * EMAD operation class. Must be set to 0x1 (REG_ACCESS).
+ */
+MLXSW_ITEM32(emad, op_tlv, class, 0x04, 0, 8);
+
+/* emad_op_tlv_tid
+ * EMAD transaction ID. Used for pairing request and response EMADs.
+ */
+MLXSW_ITEM64(emad, op_tlv, tid, 0x08, 0, 64);
+
+/* emad_string_tlv_type
+ * Type of the TLV.
+ * Must be set to 0x2 (string TLV).
+ */
+MLXSW_ITEM32(emad, string_tlv, type, 0x00, 27, 5);
+
+/* emad_string_tlv_len
+ * Length of the string TLV in u32.
+ */
+MLXSW_ITEM32(emad, string_tlv, len, 0x00, 16, 11);
+
+#define MLXSW_EMAD_STRING_TLV_STRING_LEN 128
+
+/* emad_string_tlv_string
+ * String provided by the device's firmware in case of erroneous register access
+ */
+MLXSW_ITEM_BUF(emad, string_tlv, string, 0x04,
+	       MLXSW_EMAD_STRING_TLV_STRING_LEN);
+
+/* emad_reg_tlv_type
+ * Type of the TLV.
+ * Must be set to 0x3 (register TLV).
+ */
+MLXSW_ITEM32(emad, reg_tlv, type, 0x00, 27, 5);
+
+/* emad_reg_tlv_len
+ * Length of the operation TLV in u32.
+ */
+MLXSW_ITEM32(emad, reg_tlv, len, 0x00, 16, 11);
+
+/* emad_end_tlv_type
+ * Type of the TLV.
+ * Must be set to 0x0 (end TLV).
+ */
+MLXSW_ITEM32(emad, end_tlv, type, 0x00, 27, 5);
+
+/* emad_end_tlv_len
+ * Length of the end TLV in u32.
+ * Must be set to 1.
+ */
+MLXSW_ITEM32(emad, end_tlv, len, 0x00, 16, 11);
+
+enum mlxsw_core_reg_access_type {
+	MLXSW_CORE_REG_ACCESS_TYPE_QUERY,
+	MLXSW_CORE_REG_ACCESS_TYPE_WRITE,
+};
+
+static inline const char *
+mlxsw_core_reg_access_type_str(enum mlxsw_core_reg_access_type type)
+{
+	switch (type) {
+	case MLXSW_CORE_REG_ACCESS_TYPE_QUERY:
+		return "query";
+	case MLXSW_CORE_REG_ACCESS_TYPE_WRITE:
+		return "write";
+	}
+	BUG();
+}
+
+static void mlxsw_emad_pack_end_tlv(char *end_tlv)
+{
+	mlxsw_emad_end_tlv_type_set(end_tlv, MLXSW_EMAD_TLV_TYPE_END);
+	mlxsw_emad_end_tlv_len_set(end_tlv, MLXSW_EMAD_END_TLV_LEN);
+}
+
+static void mlxsw_emad_pack_reg_tlv(char *reg_tlv,
+				    const struct mlxsw_reg_info *reg,
+				    char *payload)
+{
+	mlxsw_emad_reg_tlv_type_set(reg_tlv, MLXSW_EMAD_TLV_TYPE_REG);
+	mlxsw_emad_reg_tlv_len_set(reg_tlv, reg->len / sizeof(u32) + 1);
+	memcpy(reg_tlv + sizeof(u32), payload, reg->len);
+}
+
+static void mlxsw_emad_pack_string_tlv(char *string_tlv)
+{
+	mlxsw_emad_string_tlv_type_set(string_tlv, MLXSW_EMAD_TLV_TYPE_STRING);
+	mlxsw_emad_string_tlv_len_set(string_tlv, MLXSW_EMAD_STRING_TLV_LEN);
+}
+
+static void mlxsw_emad_pack_op_tlv(char *op_tlv,
+				   const struct mlxsw_reg_info *reg,
+				   enum mlxsw_core_reg_access_type type,
+				   u64 tid)
+{
+	mlxsw_emad_op_tlv_type_set(op_tlv, MLXSW_EMAD_TLV_TYPE_OP);
+	mlxsw_emad_op_tlv_len_set(op_tlv, MLXSW_EMAD_OP_TLV_LEN);
+	mlxsw_emad_op_tlv_dr_set(op_tlv, 0);
+	mlxsw_emad_op_tlv_status_set(op_tlv, 0);
+	mlxsw_emad_op_tlv_register_id_set(op_tlv, reg->id);
+	mlxsw_emad_op_tlv_r_set(op_tlv, MLXSW_EMAD_OP_TLV_REQUEST);
+	if (type == MLXSW_CORE_REG_ACCESS_TYPE_QUERY)
+		mlxsw_emad_op_tlv_method_set(op_tlv,
+					     MLXSW_EMAD_OP_TLV_METHOD_QUERY);
+	else
+		mlxsw_emad_op_tlv_method_set(op_tlv,
+					     MLXSW_EMAD_OP_TLV_METHOD_WRITE);
+	mlxsw_emad_op_tlv_class_set(op_tlv,
+				    MLXSW_EMAD_OP_TLV_CLASS_REG_ACCESS);
+	mlxsw_emad_op_tlv_tid_set(op_tlv, tid);
+}
+
+static int mlxsw_emad_construct_eth_hdr(struct sk_buff *skb)
+{
+	char *eth_hdr = skb_push(skb, MLXSW_EMAD_ETH_HDR_LEN);
+
+	mlxsw_emad_eth_hdr_dmac_memcpy_to(eth_hdr, MLXSW_EMAD_EH_DMAC);
+	mlxsw_emad_eth_hdr_smac_memcpy_to(eth_hdr, MLXSW_EMAD_EH_SMAC);
+	mlxsw_emad_eth_hdr_ethertype_set(eth_hdr, MLXSW_EMAD_EH_ETHERTYPE);
+	mlxsw_emad_eth_hdr_mlx_proto_set(eth_hdr, MLXSW_EMAD_EH_MLX_PROTO);
+	mlxsw_emad_eth_hdr_ver_set(eth_hdr, MLXSW_EMAD_EH_PROTO_VERSION);
+
+	skb_reset_mac_header(skb);
+
+	return 0;
+}
+
+static void mlxsw_emad_construct(struct sk_buff *skb,
+				 const struct mlxsw_reg_info *reg,
+				 char *payload,
+				 enum mlxsw_core_reg_access_type type,
+				 u64 tid, bool enable_string_tlv)
+{
+	char *buf;
+
+	buf = skb_push(skb, MLXSW_EMAD_END_TLV_LEN * sizeof(u32));
+	mlxsw_emad_pack_end_tlv(buf);
+
+	buf = skb_push(skb, reg->len + sizeof(u32));
+	mlxsw_emad_pack_reg_tlv(buf, reg, payload);
+
+	if (enable_string_tlv) {
+		buf = skb_push(skb, MLXSW_EMAD_STRING_TLV_LEN * sizeof(u32));
+		mlxsw_emad_pack_string_tlv(buf);
+	}
+
+	buf = skb_push(skb, MLXSW_EMAD_OP_TLV_LEN * sizeof(u32));
+	mlxsw_emad_pack_op_tlv(buf, reg, type, tid);
+
+	mlxsw_emad_construct_eth_hdr(skb);
+}
+
+struct mlxsw_emad_tlv_offsets {
+	u16 op_tlv;
+	u16 string_tlv;
+	u16 reg_tlv;
+};
+
+static bool mlxsw_emad_tlv_is_string_tlv(const char *tlv)
+{
+	u8 tlv_type = mlxsw_emad_string_tlv_type_get(tlv);
+
+	return tlv_type == MLXSW_EMAD_TLV_TYPE_STRING;
+}
+
+static void mlxsw_emad_tlv_parse(struct sk_buff *skb)
+{
+	struct mlxsw_emad_tlv_offsets *offsets =
+		(struct mlxsw_emad_tlv_offsets *) skb->cb;
+
+	offsets->op_tlv = MLXSW_EMAD_ETH_HDR_LEN;
+	offsets->string_tlv = 0;
+	offsets->reg_tlv = MLXSW_EMAD_ETH_HDR_LEN +
+			   MLXSW_EMAD_OP_TLV_LEN * sizeof(u32);
+
+	/* If string TLV is present, it must come after the operation TLV. */
+	if (mlxsw_emad_tlv_is_string_tlv(skb->data + offsets->reg_tlv)) {
+		offsets->string_tlv = offsets->reg_tlv;
+		offsets->reg_tlv += MLXSW_EMAD_STRING_TLV_LEN * sizeof(u32);
+	}
+}
+
+static char *mlxsw_emad_op_tlv(const struct sk_buff *skb)
+{
+	struct mlxsw_emad_tlv_offsets *offsets =
+		(struct mlxsw_emad_tlv_offsets *) skb->cb;
+
+	return ((char *) (skb->data + offsets->op_tlv));
+}
+
+static char *mlxsw_emad_string_tlv(const struct sk_buff *skb)
+{
+	struct mlxsw_emad_tlv_offsets *offsets =
+		(struct mlxsw_emad_tlv_offsets *) skb->cb;
+
+	if (!offsets->string_tlv)
+		return NULL;
+
+	return ((char *) (skb->data + offsets->string_tlv));
+}
+
+static char *mlxsw_emad_reg_tlv(const struct sk_buff *skb)
+{
+	struct mlxsw_emad_tlv_offsets *offsets =
+		(struct mlxsw_emad_tlv_offsets *) skb->cb;
+
+	return ((char *) (skb->data + offsets->reg_tlv));
+}
+
+static char *mlxsw_emad_reg_payload(const char *reg_tlv)
+{
+	return ((char *) (reg_tlv + sizeof(u32)));
+}
+
+static char *mlxsw_emad_reg_payload_cmd(const char *mbox)
+{
+	return ((char *) (mbox + (MLXSW_EMAD_OP_TLV_LEN + 1) * sizeof(u32)));
+}
+
+static u64 mlxsw_emad_get_tid(const struct sk_buff *skb)
+{
+	char *op_tlv;
+
+	op_tlv = mlxsw_emad_op_tlv(skb);
+	return mlxsw_emad_op_tlv_tid_get(op_tlv);
+}
+
+static bool mlxsw_emad_is_resp(const struct sk_buff *skb)
+{
+	char *op_tlv;
+
+	op_tlv = mlxsw_emad_op_tlv(skb);
+	return (mlxsw_emad_op_tlv_r_get(op_tlv) == MLXSW_EMAD_OP_TLV_RESPONSE);
+}
+
+static int mlxsw_emad_process_status(char *op_tlv,
+				     enum mlxsw_emad_op_tlv_status *p_status)
+{
+	*p_status = mlxsw_emad_op_tlv_status_get(op_tlv);
+
+	switch (*p_status) {
+	case MLXSW_EMAD_OP_TLV_STATUS_SUCCESS:
+		return 0;
+	case MLXSW_EMAD_OP_TLV_STATUS_BUSY:
+	case MLXSW_EMAD_OP_TLV_STATUS_MESSAGE_RECEIPT_ACK:
+		return -EAGAIN;
+	case MLXSW_EMAD_OP_TLV_STATUS_VERSION_NOT_SUPPORTED:
+	case MLXSW_EMAD_OP_TLV_STATUS_UNKNOWN_TLV:
+	case MLXSW_EMAD_OP_TLV_STATUS_REGISTER_NOT_SUPPORTED:
+	case MLXSW_EMAD_OP_TLV_STATUS_CLASS_NOT_SUPPORTED:
+	case MLXSW_EMAD_OP_TLV_STATUS_METHOD_NOT_SUPPORTED:
+	case MLXSW_EMAD_OP_TLV_STATUS_BAD_PARAMETER:
+	case MLXSW_EMAD_OP_TLV_STATUS_RESOURCE_NOT_AVAILABLE:
+	case MLXSW_EMAD_OP_TLV_STATUS_INTERNAL_ERROR:
+	default:
+		return -EIO;
+	}
+}
+
+static int
+mlxsw_emad_process_status_skb(struct sk_buff *skb,
+			      enum mlxsw_emad_op_tlv_status *p_status)
+{
+	return mlxsw_emad_process_status(mlxsw_emad_op_tlv(skb), p_status);
+}
+
+struct mlxsw_reg_trans {
+	struct list_head list;
+	struct list_head bulk_list;
+	struct mlxsw_core *core;
+	struct sk_buff *tx_skb;
+	struct mlxsw_tx_info tx_info;
+	struct delayed_work timeout_dw;
+	unsigned int retries;
+	u64 tid;
+	struct completion completion;
+	atomic_t active;
+	mlxsw_reg_trans_cb_t *cb;
+	unsigned long cb_priv;
+	const struct mlxsw_reg_info *reg;
+	enum mlxsw_core_reg_access_type type;
+	int err;
+	char *emad_err_string;
+	enum mlxsw_emad_op_tlv_status emad_status;
+	struct rcu_head rcu;
+};
+
+static void mlxsw_emad_process_string_tlv(const struct sk_buff *skb,
+					  struct mlxsw_reg_trans *trans)
+{
+	char *string_tlv;
+	char *string;
+
+	string_tlv = mlxsw_emad_string_tlv(skb);
+	if (!string_tlv)
+		return;
+
+	trans->emad_err_string = kzalloc(MLXSW_EMAD_STRING_TLV_STRING_LEN,
+					 GFP_ATOMIC);
+	if (!trans->emad_err_string)
+		return;
+
+	string = mlxsw_emad_string_tlv_string_data(string_tlv);
+	strscpy(trans->emad_err_string, string,
+		MLXSW_EMAD_STRING_TLV_STRING_LEN);
+}
+
+#define MLXSW_EMAD_TIMEOUT_DURING_FW_FLASH_MS	3000
+#define MLXSW_EMAD_TIMEOUT_MS			200
+
+static void mlxsw_emad_trans_timeout_schedule(struct mlxsw_reg_trans *trans)
+{
+	unsigned long timeout = msecs_to_jiffies(MLXSW_EMAD_TIMEOUT_MS);
+
+	if (trans->core->fw_flash_in_progress)
+		timeout = msecs_to_jiffies(MLXSW_EMAD_TIMEOUT_DURING_FW_FLASH_MS);
+
+	queue_delayed_work(trans->core->emad_wq, &trans->timeout_dw,
+			   timeout << trans->retries);
+}
+
+static int mlxsw_emad_transmit(struct mlxsw_core *mlxsw_core,
+			       struct mlxsw_reg_trans *trans)
+{
+	struct sk_buff *skb;
+	int err;
+
+	skb = skb_clone(trans->tx_skb, GFP_KERNEL);
+	if (!skb)
+		return -ENOMEM;
+
+	trace_devlink_hwmsg(priv_to_devlink(mlxsw_core), false, 0,
+			    skb->data + mlxsw_core->driver->txhdr_len,
+			    skb->len - mlxsw_core->driver->txhdr_len);
+
+	atomic_set(&trans->active, 1);
+	err = mlxsw_core_skb_transmit(mlxsw_core, skb, &trans->tx_info);
+	if (err) {
+		dev_kfree_skb(skb);
+		return err;
+	}
+	mlxsw_emad_trans_timeout_schedule(trans);
+	return 0;
+}
+
+static void mlxsw_emad_trans_finish(struct mlxsw_reg_trans *trans, int err)
+{
+	struct mlxsw_core *mlxsw_core = trans->core;
+
+	dev_kfree_skb(trans->tx_skb);
+	spin_lock_bh(&mlxsw_core->emad.trans_list_lock);
+	list_del_rcu(&trans->list);
+	spin_unlock_bh(&mlxsw_core->emad.trans_list_lock);
+	trans->err = err;
+	complete(&trans->completion);
+}
+
+static void mlxsw_emad_transmit_retry(struct mlxsw_core *mlxsw_core,
+				      struct mlxsw_reg_trans *trans)
+{
+	int err;
+
+	if (trans->retries < MLXSW_EMAD_MAX_RETRY) {
+		trans->retries++;
+		err = mlxsw_emad_transmit(trans->core, trans);
+		if (err == 0)
+			return;
+
+		if (!atomic_dec_and_test(&trans->active))
+			return;
+	} else {
+		err = -EIO;
+	}
+	mlxsw_emad_trans_finish(trans, err);
+}
+
+static void mlxsw_emad_trans_timeout_work(struct work_struct *work)
+{
+	struct mlxsw_reg_trans *trans = container_of(work,
+						     struct mlxsw_reg_trans,
+						     timeout_dw.work);
+
+	if (!atomic_dec_and_test(&trans->active))
+		return;
+
+	mlxsw_emad_transmit_retry(trans->core, trans);
+}
+
+static void mlxsw_emad_process_response(struct mlxsw_core *mlxsw_core,
+					struct mlxsw_reg_trans *trans,
+					struct sk_buff *skb)
+{
+	int err;
+
+	if (!atomic_dec_and_test(&trans->active))
+		return;
+
+	err = mlxsw_emad_process_status_skb(skb, &trans->emad_status);
+	if (err == -EAGAIN) {
+		mlxsw_emad_transmit_retry(mlxsw_core, trans);
+	} else {
+		if (err == 0) {
+			char *reg_tlv = mlxsw_emad_reg_tlv(skb);
+
+			if (trans->cb)
+				trans->cb(mlxsw_core,
+					  mlxsw_emad_reg_payload(reg_tlv),
+					  trans->reg->len, trans->cb_priv);
+		} else {
+			mlxsw_emad_process_string_tlv(skb, trans);
+		}
+		mlxsw_emad_trans_finish(trans, err);
+	}
+}
+
+/* called with rcu read lock held */
+static void mlxsw_emad_rx_listener_func(struct sk_buff *skb, u16 local_port,
+					void *priv)
+{
+	struct mlxsw_core *mlxsw_core = priv;
+	struct mlxsw_reg_trans *trans;
+
+	trace_devlink_hwmsg(priv_to_devlink(mlxsw_core), true, 0,
+			    skb->data, skb->len);
+
+	mlxsw_emad_tlv_parse(skb);
+
+	if (!mlxsw_emad_is_resp(skb))
+		goto free_skb;
+
+	list_for_each_entry_rcu(trans, &mlxsw_core->emad.trans_list, list) {
+		if (mlxsw_emad_get_tid(skb) == trans->tid) {
+			mlxsw_emad_process_response(mlxsw_core, trans, skb);
+			break;
+		}
+	}
+
+free_skb:
+	dev_kfree_skb(skb);
+}
+
+static const struct mlxsw_listener mlxsw_emad_rx_listener =
+	MLXSW_RXL(mlxsw_emad_rx_listener_func, ETHEMAD, TRAP_TO_CPU, false,
+		  EMAD, DISCARD);
+
+static int mlxsw_emad_init(struct mlxsw_core *mlxsw_core)
+{
+	struct workqueue_struct *emad_wq;
+	u64 tid;
+	int err;
+
+	if (!(mlxsw_core->bus->features & MLXSW_BUS_F_TXRX))
+		return 0;
+
+	emad_wq = alloc_workqueue("mlxsw_core_emad", 0, 0);
+	if (!emad_wq)
+		return -ENOMEM;
+	mlxsw_core->emad_wq = emad_wq;
+
+	/* Set the upper 32 bits of the transaction ID field to a random
+	 * number. This allows us to discard EMADs addressed to other
+	 * devices.
+	 */
+	get_random_bytes(&tid, 4);
+	tid <<= 32;
+	atomic64_set(&mlxsw_core->emad.tid, tid);
+
+	INIT_LIST_HEAD(&mlxsw_core->emad.trans_list);
+	spin_lock_init(&mlxsw_core->emad.trans_list_lock);
+
+	err = mlxsw_core_trap_register(mlxsw_core, &mlxsw_emad_rx_listener,
+				       mlxsw_core);
+	if (err)
+		goto err_trap_register;
+
+	mlxsw_core->emad.use_emad = true;
+
+	return 0;
+
+err_trap_register:
+	destroy_workqueue(mlxsw_core->emad_wq);
+	return err;
+}
+
+static void mlxsw_emad_fini(struct mlxsw_core *mlxsw_core)
+{
+
+	if (!(mlxsw_core->bus->features & MLXSW_BUS_F_TXRX))
+		return;
+
+	mlxsw_core->emad.use_emad = false;
+	mlxsw_core_trap_unregister(mlxsw_core, &mlxsw_emad_rx_listener,
+				   mlxsw_core);
+	destroy_workqueue(mlxsw_core->emad_wq);
+}
+
+static struct sk_buff *mlxsw_emad_alloc(const struct mlxsw_core *mlxsw_core,
+					u16 reg_len, bool enable_string_tlv)
+{
+	struct sk_buff *skb;
+	u16 emad_len;
+
+	emad_len = (reg_len + sizeof(u32) + MLXSW_EMAD_ETH_HDR_LEN +
+		    (MLXSW_EMAD_OP_TLV_LEN + MLXSW_EMAD_END_TLV_LEN) *
+		    sizeof(u32) + mlxsw_core->driver->txhdr_len);
+	if (enable_string_tlv)
+		emad_len += MLXSW_EMAD_STRING_TLV_LEN * sizeof(u32);
+	if (emad_len > MLXSW_EMAD_MAX_FRAME_LEN)
+		return NULL;
+
+	skb = netdev_alloc_skb(NULL, emad_len);
+	if (!skb)
+		return NULL;
+	memset(skb->data, 0, emad_len);
+	skb_reserve(skb, emad_len);
+
+	return skb;
+}
+
+static int mlxsw_emad_reg_access(struct mlxsw_core *mlxsw_core,
+				 const struct mlxsw_reg_info *reg,
+				 char *payload,
+				 enum mlxsw_core_reg_access_type type,
+				 struct mlxsw_reg_trans *trans,
+				 struct list_head *bulk_list,
+				 mlxsw_reg_trans_cb_t *cb,
+				 unsigned long cb_priv, u64 tid)
+{
+	bool enable_string_tlv;
+	struct sk_buff *skb;
+	int err;
+
+	dev_dbg(mlxsw_core->bus_info->dev, "EMAD reg access (tid=%llx,reg_id=%x(%s),type=%s)\n",
+		tid, reg->id, mlxsw_reg_id_str(reg->id),
+		mlxsw_core_reg_access_type_str(type));
+
+	/* Since this can be changed during emad_reg_access, read it once and
+	 * use the value all the way.
+	 */
+	enable_string_tlv = mlxsw_core->emad.enable_string_tlv;
+
+	skb = mlxsw_emad_alloc(mlxsw_core, reg->len, enable_string_tlv);
+	if (!skb)
+		return -ENOMEM;
+
+	list_add_tail(&trans->bulk_list, bulk_list);
+	trans->core = mlxsw_core;
+	trans->tx_skb = skb;
+	trans->tx_info.local_port = MLXSW_PORT_CPU_PORT;
+	trans->tx_info.is_emad = true;
+	INIT_DELAYED_WORK(&trans->timeout_dw, mlxsw_emad_trans_timeout_work);
+	trans->tid = tid;
+	init_completion(&trans->completion);
+	trans->cb = cb;
+	trans->cb_priv = cb_priv;
+	trans->reg = reg;
+	trans->type = type;
+
+	mlxsw_emad_construct(skb, reg, payload, type, trans->tid,
+			     enable_string_tlv);
+	mlxsw_core->driver->txhdr_construct(skb, &trans->tx_info);
+
+	spin_lock_bh(&mlxsw_core->emad.trans_list_lock);
+	list_add_tail_rcu(&trans->list, &mlxsw_core->emad.trans_list);
+	spin_unlock_bh(&mlxsw_core->emad.trans_list_lock);
+	err = mlxsw_emad_transmit(mlxsw_core, trans);
+	if (err)
+		goto err_out;
+	return 0;
+
+err_out:
+	spin_lock_bh(&mlxsw_core->emad.trans_list_lock);
+	list_del_rcu(&trans->list);
+	spin_unlock_bh(&mlxsw_core->emad.trans_list_lock);
+	list_del(&trans->bulk_list);
+	dev_kfree_skb(trans->tx_skb);
+	return err;
+}
+
+/*****************
+ * Core functions
+ *****************/
+
+int mlxsw_core_driver_register(struct mlxsw_driver *mlxsw_driver)
+{
+	spin_lock(&mlxsw_core_driver_list_lock);
+	list_add_tail(&mlxsw_driver->list, &mlxsw_core_driver_list);
+	spin_unlock(&mlxsw_core_driver_list_lock);
+	return 0;
+}
+EXPORT_SYMBOL(mlxsw_core_driver_register);
+
+void mlxsw_core_driver_unregister(struct mlxsw_driver *mlxsw_driver)
+{
+	spin_lock(&mlxsw_core_driver_list_lock);
+	list_del(&mlxsw_driver->list);
+	spin_unlock(&mlxsw_core_driver_list_lock);
+}
+EXPORT_SYMBOL(mlxsw_core_driver_unregister);
+
+static struct mlxsw_driver *__driver_find(const char *kind)
+{
+	struct mlxsw_driver *mlxsw_driver;
+
+	list_for_each_entry(mlxsw_driver, &mlxsw_core_driver_list, list) {
+		if (strcmp(mlxsw_driver->kind, kind) == 0)
+			return mlxsw_driver;
+	}
+	return NULL;
+}
+
+static struct mlxsw_driver *mlxsw_core_driver_get(const char *kind)
+{
+	struct mlxsw_driver *mlxsw_driver;
+
+	spin_lock(&mlxsw_core_driver_list_lock);
+	mlxsw_driver = __driver_find(kind);
+	spin_unlock(&mlxsw_core_driver_list_lock);
+	return mlxsw_driver;
+}
+
+int mlxsw_core_fw_flash(struct mlxsw_core *mlxsw_core,
+			struct mlxfw_dev *mlxfw_dev,
+			const struct firmware *firmware,
+			struct netlink_ext_ack *extack)
+{
+	int err;
+
+	mlxsw_core->fw_flash_in_progress = true;
+	err = mlxfw_firmware_flash(mlxfw_dev, firmware, extack);
+	mlxsw_core->fw_flash_in_progress = false;
+
+	return err;
+}
+
+struct mlxsw_core_fw_info {
+	struct mlxfw_dev mlxfw_dev;
+	struct mlxsw_core *mlxsw_core;
+};
+
+static int mlxsw_core_fw_component_query(struct mlxfw_dev *mlxfw_dev,
+					 u16 component_index, u32 *p_max_size,
+					 u8 *p_align_bits, u16 *p_max_write_size)
+{
+	struct mlxsw_core_fw_info *mlxsw_core_fw_info =
+		container_of(mlxfw_dev, struct mlxsw_core_fw_info, mlxfw_dev);
+	struct mlxsw_core *mlxsw_core = mlxsw_core_fw_info->mlxsw_core;
+	char mcqi_pl[MLXSW_REG_MCQI_LEN];
+	int err;
+
+	mlxsw_reg_mcqi_pack(mcqi_pl, component_index);
+	err = mlxsw_reg_query(mlxsw_core, MLXSW_REG(mcqi), mcqi_pl);
+	if (err)
+		return err;
+	mlxsw_reg_mcqi_unpack(mcqi_pl, p_max_size, p_align_bits, p_max_write_size);
+
+	*p_align_bits = max_t(u8, *p_align_bits, 2);
+	*p_max_write_size = min_t(u16, *p_max_write_size, MLXSW_REG_MCDA_MAX_DATA_LEN);
+	return 0;
+}
+
+static int mlxsw_core_fw_fsm_lock(struct mlxfw_dev *mlxfw_dev, u32 *fwhandle)
+{
+	struct mlxsw_core_fw_info *mlxsw_core_fw_info =
+		container_of(mlxfw_dev, struct mlxsw_core_fw_info, mlxfw_dev);
+	struct mlxsw_core *mlxsw_core = mlxsw_core_fw_info->mlxsw_core;
+	char mcc_pl[MLXSW_REG_MCC_LEN];
+	u8 control_state;
+	int err;
+
+	mlxsw_reg_mcc_pack(mcc_pl, 0, 0, 0, 0);
+	err = mlxsw_reg_query(mlxsw_core, MLXSW_REG(mcc), mcc_pl);
+	if (err)
+		return err;
+
+	mlxsw_reg_mcc_unpack(mcc_pl, fwhandle, NULL, &control_state);
+	if (control_state != MLXFW_FSM_STATE_IDLE)
+		return -EBUSY;
+
+	mlxsw_reg_mcc_pack(mcc_pl, MLXSW_REG_MCC_INSTRUCTION_LOCK_UPDATE_HANDLE, 0, *fwhandle, 0);
+	return mlxsw_reg_write(mlxsw_core, MLXSW_REG(mcc), mcc_pl);
+}
+
+static int mlxsw_core_fw_fsm_component_update(struct mlxfw_dev *mlxfw_dev, u32 fwhandle,
+					      u16 component_index, u32 component_size)
+{
+	struct mlxsw_core_fw_info *mlxsw_core_fw_info =
+		container_of(mlxfw_dev, struct mlxsw_core_fw_info, mlxfw_dev);
+	struct mlxsw_core *mlxsw_core = mlxsw_core_fw_info->mlxsw_core;
+	char mcc_pl[MLXSW_REG_MCC_LEN];
+
+	mlxsw_reg_mcc_pack(mcc_pl, MLXSW_REG_MCC_INSTRUCTION_UPDATE_COMPONENT,
+			   component_index, fwhandle, component_size);
+	return mlxsw_reg_write(mlxsw_core, MLXSW_REG(mcc), mcc_pl);
+}
+
+static int mlxsw_core_fw_fsm_block_download(struct mlxfw_dev *mlxfw_dev, u32 fwhandle,
+					    u8 *data, u16 size, u32 offset)
+{
+	struct mlxsw_core_fw_info *mlxsw_core_fw_info =
+		container_of(mlxfw_dev, struct mlxsw_core_fw_info, mlxfw_dev);
+	struct mlxsw_core *mlxsw_core = mlxsw_core_fw_info->mlxsw_core;
+	char mcda_pl[MLXSW_REG_MCDA_LEN];
+
+	mlxsw_reg_mcda_pack(mcda_pl, fwhandle, offset, size, data);
+	return mlxsw_reg_write(mlxsw_core, MLXSW_REG(mcda), mcda_pl);
+}
+
+static int mlxsw_core_fw_fsm_component_verify(struct mlxfw_dev *mlxfw_dev, u32 fwhandle,
+					      u16 component_index)
+{
+	struct mlxsw_core_fw_info *mlxsw_core_fw_info =
+		container_of(mlxfw_dev, struct mlxsw_core_fw_info, mlxfw_dev);
+	struct mlxsw_core *mlxsw_core = mlxsw_core_fw_info->mlxsw_core;
+	char mcc_pl[MLXSW_REG_MCC_LEN];
+
+	mlxsw_reg_mcc_pack(mcc_pl, MLXSW_REG_MCC_INSTRUCTION_VERIFY_COMPONENT,
+			   component_index, fwhandle, 0);
+	return mlxsw_reg_write(mlxsw_core, MLXSW_REG(mcc), mcc_pl);
+}
+
+static int mlxsw_core_fw_fsm_activate(struct mlxfw_dev *mlxfw_dev, u32 fwhandle)
+{
+	struct mlxsw_core_fw_info *mlxsw_core_fw_info =
+		container_of(mlxfw_dev, struct mlxsw_core_fw_info, mlxfw_dev);
+	struct mlxsw_core *mlxsw_core = mlxsw_core_fw_info->mlxsw_core;
+	char mcc_pl[MLXSW_REG_MCC_LEN];
+
+	mlxsw_reg_mcc_pack(mcc_pl, MLXSW_REG_MCC_INSTRUCTION_ACTIVATE, 0, fwhandle, 0);
+	return mlxsw_reg_write(mlxsw_core, MLXSW_REG(mcc), mcc_pl);
+}
+
+static int mlxsw_core_fw_fsm_query_state(struct mlxfw_dev *mlxfw_dev, u32 fwhandle,
+					 enum mlxfw_fsm_state *fsm_state,
+					 enum mlxfw_fsm_state_err *fsm_state_err)
+{
+	struct mlxsw_core_fw_info *mlxsw_core_fw_info =
+		container_of(mlxfw_dev, struct mlxsw_core_fw_info, mlxfw_dev);
+	struct mlxsw_core *mlxsw_core = mlxsw_core_fw_info->mlxsw_core;
+	char mcc_pl[MLXSW_REG_MCC_LEN];
+	u8 control_state;
+	u8 error_code;
+	int err;
+
+	mlxsw_reg_mcc_pack(mcc_pl, 0, 0, fwhandle, 0);
+	err = mlxsw_reg_query(mlxsw_core, MLXSW_REG(mcc), mcc_pl);
+	if (err)
+		return err;
+
+	mlxsw_reg_mcc_unpack(mcc_pl, NULL, &error_code, &control_state);
+	*fsm_state = control_state;
+	*fsm_state_err = min_t(enum mlxfw_fsm_state_err, error_code, MLXFW_FSM_STATE_ERR_MAX);
+	return 0;
+}
+
+static void mlxsw_core_fw_fsm_cancel(struct mlxfw_dev *mlxfw_dev, u32 fwhandle)
+{
+	struct mlxsw_core_fw_info *mlxsw_core_fw_info =
+		container_of(mlxfw_dev, struct mlxsw_core_fw_info, mlxfw_dev);
+	struct mlxsw_core *mlxsw_core = mlxsw_core_fw_info->mlxsw_core;
+	char mcc_pl[MLXSW_REG_MCC_LEN];
+
+	mlxsw_reg_mcc_pack(mcc_pl, MLXSW_REG_MCC_INSTRUCTION_CANCEL, 0, fwhandle, 0);
+	mlxsw_reg_write(mlxsw_core, MLXSW_REG(mcc), mcc_pl);
+}
+
+static void mlxsw_core_fw_fsm_release(struct mlxfw_dev *mlxfw_dev, u32 fwhandle)
+{
+	struct mlxsw_core_fw_info *mlxsw_core_fw_info =
+		container_of(mlxfw_dev, struct mlxsw_core_fw_info, mlxfw_dev);
+	struct mlxsw_core *mlxsw_core = mlxsw_core_fw_info->mlxsw_core;
+	char mcc_pl[MLXSW_REG_MCC_LEN];
+
+	mlxsw_reg_mcc_pack(mcc_pl, MLXSW_REG_MCC_INSTRUCTION_RELEASE_UPDATE_HANDLE, 0, fwhandle, 0);
+	mlxsw_reg_write(mlxsw_core, MLXSW_REG(mcc), mcc_pl);
+}
+
+static const struct mlxfw_dev_ops mlxsw_core_fw_mlxsw_dev_ops = {
+	.component_query	= mlxsw_core_fw_component_query,
+	.fsm_lock		= mlxsw_core_fw_fsm_lock,
+	.fsm_component_update	= mlxsw_core_fw_fsm_component_update,
+	.fsm_block_download	= mlxsw_core_fw_fsm_block_download,
+	.fsm_component_verify	= mlxsw_core_fw_fsm_component_verify,
+	.fsm_activate		= mlxsw_core_fw_fsm_activate,
+	.fsm_query_state	= mlxsw_core_fw_fsm_query_state,
+	.fsm_cancel		= mlxsw_core_fw_fsm_cancel,
+	.fsm_release		= mlxsw_core_fw_fsm_release,
+};
+
+static int mlxsw_core_dev_fw_flash(struct mlxsw_core *mlxsw_core,
+				   const struct firmware *firmware,
+				   struct netlink_ext_ack *extack)
+{
+	struct mlxsw_core_fw_info mlxsw_core_fw_info = {
+		.mlxfw_dev = {
+			.ops = &mlxsw_core_fw_mlxsw_dev_ops,
+			.psid = mlxsw_core->bus_info->psid,
+			.psid_size = strlen(mlxsw_core->bus_info->psid),
+			.devlink = priv_to_devlink(mlxsw_core),
+		},
+		.mlxsw_core = mlxsw_core
+	};
+
+	return mlxsw_core_fw_flash(mlxsw_core, &mlxsw_core_fw_info.mlxfw_dev,
+				   firmware, extack);
+}
+
+static int mlxsw_core_fw_rev_validate(struct mlxsw_core *mlxsw_core,
+				      const struct mlxsw_bus_info *mlxsw_bus_info,
+				      const struct mlxsw_fw_rev *req_rev,
+				      const char *filename)
+{
+	const struct mlxsw_fw_rev *rev = &mlxsw_bus_info->fw_rev;
+	union devlink_param_value value;
+	const struct firmware *firmware;
+	int err;
+
+	/* Don't check if driver does not require it */
+	if (!req_rev || !filename)
+		return 0;
+
+	/* Don't check if devlink 'fw_load_policy' param is 'flash' */
+	err = devlink_param_driverinit_value_get(priv_to_devlink(mlxsw_core),
+						 DEVLINK_PARAM_GENERIC_ID_FW_LOAD_POLICY,
+						 &value);
+	if (err)
+		return err;
+	if (value.vu8 == DEVLINK_PARAM_FW_LOAD_POLICY_VALUE_FLASH)
+		return 0;
+
+	/* Validate driver & FW are compatible */
+	if (rev->major != req_rev->major) {
+		WARN(1, "Mismatch in major FW version [%d:%d] is never expected; Please contact support\n",
+		     rev->major, req_rev->major);
+		return -EINVAL;
+	}
+	if (mlxsw_core_fw_rev_minor_subminor_validate(rev, req_rev))
+		return 0;
+
+	dev_err(mlxsw_bus_info->dev, "The firmware version %d.%d.%d is incompatible with the driver (required >= %d.%d.%d)\n",
+		rev->major, rev->minor, rev->subminor, req_rev->major,
+		req_rev->minor, req_rev->subminor);
+	dev_info(mlxsw_bus_info->dev, "Flashing firmware using file %s\n", filename);
+
+	err = request_firmware_direct(&firmware, filename, mlxsw_bus_info->dev);
+	if (err) {
+		dev_err(mlxsw_bus_info->dev, "Could not request firmware file %s\n", filename);
+		return err;
+	}
+
+	err = mlxsw_core_dev_fw_flash(mlxsw_core, firmware, NULL);
+	release_firmware(firmware);
+	if (err)
+		dev_err(mlxsw_bus_info->dev, "Could not upgrade firmware\n");
+
+	/* On FW flash success, tell the caller FW reset is needed
+	 * if current FW supports it.
+	 */
+	if (rev->minor >= req_rev->can_reset_minor)
+		return err ? err : -EAGAIN;
+	else
+		return 0;
+}
+
+static int mlxsw_core_fw_flash_update(struct mlxsw_core *mlxsw_core,
+				      struct devlink_flash_update_params *params,
+				      struct netlink_ext_ack *extack)
+{
+	return mlxsw_core_dev_fw_flash(mlxsw_core, params->fw, extack);
+}
+
+static int mlxsw_core_devlink_param_fw_load_policy_validate(struct devlink *devlink, u32 id,
+							    union devlink_param_value val,
+							    struct netlink_ext_ack *extack)
+{
+	if (val.vu8 != DEVLINK_PARAM_FW_LOAD_POLICY_VALUE_DRIVER &&
+	    val.vu8 != DEVLINK_PARAM_FW_LOAD_POLICY_VALUE_FLASH) {
+		NL_SET_ERR_MSG_MOD(extack, "'fw_load_policy' must be 'driver' or 'flash'");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static const struct devlink_param mlxsw_core_fw_devlink_params[] = {
+	DEVLINK_PARAM_GENERIC(FW_LOAD_POLICY, BIT(DEVLINK_PARAM_CMODE_DRIVERINIT), NULL, NULL,
+			      mlxsw_core_devlink_param_fw_load_policy_validate),
+};
+
+static int mlxsw_core_fw_params_register(struct mlxsw_core *mlxsw_core)
+{
+	struct devlink *devlink = priv_to_devlink(mlxsw_core);
+	union devlink_param_value value;
+	int err;
+
+	err = devlink_params_register(devlink, mlxsw_core_fw_devlink_params,
+				      ARRAY_SIZE(mlxsw_core_fw_devlink_params));
+	if (err)
+		return err;
+
+	value.vu8 = DEVLINK_PARAM_FW_LOAD_POLICY_VALUE_DRIVER;
+	devlink_param_driverinit_value_set(devlink, DEVLINK_PARAM_GENERIC_ID_FW_LOAD_POLICY, value);
+	return 0;
+}
+
+static void mlxsw_core_fw_params_unregister(struct mlxsw_core *mlxsw_core)
+{
+	devlink_params_unregister(priv_to_devlink(mlxsw_core), mlxsw_core_fw_devlink_params,
+				  ARRAY_SIZE(mlxsw_core_fw_devlink_params));
+}
+
+static void *__dl_port(struct devlink_port *devlink_port)
+{
+	return container_of(devlink_port, struct mlxsw_core_port, devlink_port);
+}
+
+static int mlxsw_devlink_port_split(struct devlink *devlink,
+				    struct devlink_port *port,
+				    unsigned int count,
+				    struct netlink_ext_ack *extack)
+{
+	struct mlxsw_core_port *mlxsw_core_port = __dl_port(port);
+	struct mlxsw_core *mlxsw_core = devlink_priv(devlink);
+
+	if (!mlxsw_core->driver->port_split)
+		return -EOPNOTSUPP;
+	return mlxsw_core->driver->port_split(mlxsw_core,
+					      mlxsw_core_port->local_port,
+					      count, extack);
+}
+
+static int mlxsw_devlink_port_unsplit(struct devlink *devlink,
+				      struct devlink_port *port,
+				      struct netlink_ext_ack *extack)
+{
+	struct mlxsw_core_port *mlxsw_core_port = __dl_port(port);
+	struct mlxsw_core *mlxsw_core = devlink_priv(devlink);
+
+	if (!mlxsw_core->driver->port_unsplit)
+		return -EOPNOTSUPP;
+	return mlxsw_core->driver->port_unsplit(mlxsw_core,
+						mlxsw_core_port->local_port,
+						extack);
+}
+
+static int
+mlxsw_devlink_sb_pool_get(struct devlink *devlink,
+			  unsigned int sb_index, u16 pool_index,
+			  struct devlink_sb_pool_info *pool_info)
+{
+	struct mlxsw_core *mlxsw_core = devlink_priv(devlink);
+	struct mlxsw_driver *mlxsw_driver = mlxsw_core->driver;
+
+	if (!mlxsw_driver->sb_pool_get)
+		return -EOPNOTSUPP;
+	return mlxsw_driver->sb_pool_get(mlxsw_core, sb_index,
+					 pool_index, pool_info);
+}
+
+static int
+mlxsw_devlink_sb_pool_set(struct devlink *devlink,
+			  unsigned int sb_index, u16 pool_index, u32 size,
+			  enum devlink_sb_threshold_type threshold_type,
+			  struct netlink_ext_ack *extack)
+{
+	struct mlxsw_core *mlxsw_core = devlink_priv(devlink);
+	struct mlxsw_driver *mlxsw_driver = mlxsw_core->driver;
+
+	if (!mlxsw_driver->sb_pool_set)
+		return -EOPNOTSUPP;
+	return mlxsw_driver->sb_pool_set(mlxsw_core, sb_index,
+					 pool_index, size, threshold_type,
+					 extack);
+}
+
+static int mlxsw_devlink_sb_port_pool_get(struct devlink_port *devlink_port,
+					  unsigned int sb_index, u16 pool_index,
+					  u32 *p_threshold)
+{
+	struct mlxsw_core *mlxsw_core = devlink_priv(devlink_port->devlink);
+	struct mlxsw_driver *mlxsw_driver = mlxsw_core->driver;
+	struct mlxsw_core_port *mlxsw_core_port = __dl_port(devlink_port);
+
+	if (!mlxsw_driver->sb_port_pool_get ||
+	    !mlxsw_core_port_check(mlxsw_core_port))
+		return -EOPNOTSUPP;
+	return mlxsw_driver->sb_port_pool_get(mlxsw_core_port, sb_index,
+					      pool_index, p_threshold);
+}
+
+static int mlxsw_devlink_sb_port_pool_set(struct devlink_port *devlink_port,
+					  unsigned int sb_index, u16 pool_index,
+					  u32 threshold,
+					  struct netlink_ext_ack *extack)
+{
+	struct mlxsw_core *mlxsw_core = devlink_priv(devlink_port->devlink);
+	struct mlxsw_driver *mlxsw_driver = mlxsw_core->driver;
+	struct mlxsw_core_port *mlxsw_core_port = __dl_port(devlink_port);
+
+	if (!mlxsw_driver->sb_port_pool_set ||
+	    !mlxsw_core_port_check(mlxsw_core_port))
+		return -EOPNOTSUPP;
+	return mlxsw_driver->sb_port_pool_set(mlxsw_core_port, sb_index,
+					      pool_index, threshold, extack);
+}
+
+static int
+mlxsw_devlink_sb_tc_pool_bind_get(struct devlink_port *devlink_port,
+				  unsigned int sb_index, u16 tc_index,
+				  enum devlink_sb_pool_type pool_type,
+				  u16 *p_pool_index, u32 *p_threshold)
+{
+	struct mlxsw_core *mlxsw_core = devlink_priv(devlink_port->devlink);
+	struct mlxsw_driver *mlxsw_driver = mlxsw_core->driver;
+	struct mlxsw_core_port *mlxsw_core_port = __dl_port(devlink_port);
+
+	if (!mlxsw_driver->sb_tc_pool_bind_get ||
+	    !mlxsw_core_port_check(mlxsw_core_port))
+		return -EOPNOTSUPP;
+	return mlxsw_driver->sb_tc_pool_bind_get(mlxsw_core_port, sb_index,
+						 tc_index, pool_type,
+						 p_pool_index, p_threshold);
+}
+
+static int
+mlxsw_devlink_sb_tc_pool_bind_set(struct devlink_port *devlink_port,
+				  unsigned int sb_index, u16 tc_index,
+				  enum devlink_sb_pool_type pool_type,
+				  u16 pool_index, u32 threshold,
+				  struct netlink_ext_ack *extack)
+{
+	struct mlxsw_core *mlxsw_core = devlink_priv(devlink_port->devlink);
+	struct mlxsw_driver *mlxsw_driver = mlxsw_core->driver;
+	struct mlxsw_core_port *mlxsw_core_port = __dl_port(devlink_port);
+
+	if (!mlxsw_driver->sb_tc_pool_bind_set ||
+	    !mlxsw_core_port_check(mlxsw_core_port))
+		return -EOPNOTSUPP;
+	return mlxsw_driver->sb_tc_pool_bind_set(mlxsw_core_port, sb_index,
+						 tc_index, pool_type,
+						 pool_index, threshold, extack);
+}
+
+static int mlxsw_devlink_sb_occ_snapshot(struct devlink *devlink,
+					 unsigned int sb_index)
+{
+	struct mlxsw_core *mlxsw_core = devlink_priv(devlink);
+	struct mlxsw_driver *mlxsw_driver = mlxsw_core->driver;
+
+	if (!mlxsw_driver->sb_occ_snapshot)
+		return -EOPNOTSUPP;
+	return mlxsw_driver->sb_occ_snapshot(mlxsw_core, sb_index);
+}
+
+static int mlxsw_devlink_sb_occ_max_clear(struct devlink *devlink,
+					  unsigned int sb_index)
+{
+	struct mlxsw_core *mlxsw_core = devlink_priv(devlink);
+	struct mlxsw_driver *mlxsw_driver = mlxsw_core->driver;
+
+	if (!mlxsw_driver->sb_occ_max_clear)
+		return -EOPNOTSUPP;
+	return mlxsw_driver->sb_occ_max_clear(mlxsw_core, sb_index);
+}
+
+static int
+mlxsw_devlink_sb_occ_port_pool_get(struct devlink_port *devlink_port,
+				   unsigned int sb_index, u16 pool_index,
+				   u32 *p_cur, u32 *p_max)
+{
+	struct mlxsw_core *mlxsw_core = devlink_priv(devlink_port->devlink);
+	struct mlxsw_driver *mlxsw_driver = mlxsw_core->driver;
+	struct mlxsw_core_port *mlxsw_core_port = __dl_port(devlink_port);
+
+	if (!mlxsw_driver->sb_occ_port_pool_get ||
+	    !mlxsw_core_port_check(mlxsw_core_port))
+		return -EOPNOTSUPP;
+	return mlxsw_driver->sb_occ_port_pool_get(mlxsw_core_port, sb_index,
+						  pool_index, p_cur, p_max);
+}
+
+static int
+mlxsw_devlink_sb_occ_tc_port_bind_get(struct devlink_port *devlink_port,
+				      unsigned int sb_index, u16 tc_index,
+				      enum devlink_sb_pool_type pool_type,
+				      u32 *p_cur, u32 *p_max)
+{
+	struct mlxsw_core *mlxsw_core = devlink_priv(devlink_port->devlink);
+	struct mlxsw_driver *mlxsw_driver = mlxsw_core->driver;
+	struct mlxsw_core_port *mlxsw_core_port = __dl_port(devlink_port);
+
+	if (!mlxsw_driver->sb_occ_tc_port_bind_get ||
+	    !mlxsw_core_port_check(mlxsw_core_port))
+		return -EOPNOTSUPP;
+	return mlxsw_driver->sb_occ_tc_port_bind_get(mlxsw_core_port,
+						     sb_index, tc_index,
+						     pool_type, p_cur, p_max);
+}
+
+static int
+mlxsw_devlink_info_get(struct devlink *devlink, struct devlink_info_req *req,
+		       struct netlink_ext_ack *extack)
+{
+	struct mlxsw_core *mlxsw_core = devlink_priv(devlink);
+	char fw_info_psid[MLXSW_REG_MGIR_FW_INFO_PSID_SIZE];
+	u32 hw_rev, fw_major, fw_minor, fw_sub_minor;
+	char mgir_pl[MLXSW_REG_MGIR_LEN];
+	char buf[32];
+	int err;
+
+	err = devlink_info_driver_name_put(req,
+					   mlxsw_core->bus_info->device_kind);
+	if (err)
+		return err;
+
+	mlxsw_reg_mgir_pack(mgir_pl);
+	err = mlxsw_reg_query(mlxsw_core, MLXSW_REG(mgir), mgir_pl);
+	if (err)
+		return err;
+	mlxsw_reg_mgir_unpack(mgir_pl, &hw_rev, fw_info_psid, &fw_major,
+			      &fw_minor, &fw_sub_minor);
+
+	sprintf(buf, "%X", hw_rev);
+	err = devlink_info_version_fixed_put(req, "hw.revision", buf);
+	if (err)
+		return err;
+
+	err = devlink_info_version_fixed_put(req,
+					     DEVLINK_INFO_VERSION_GENERIC_FW_PSID,
+					     fw_info_psid);
+	if (err)
+		return err;
+
+	sprintf(buf, "%d.%d.%d", fw_major, fw_minor, fw_sub_minor);
+	err = devlink_info_version_running_put(req, "fw.version", buf);
+	if (err)
+		return err;
+
+	return devlink_info_version_running_put(req,
+						DEVLINK_INFO_VERSION_GENERIC_FW,
+						buf);
+}
+
+static int
+mlxsw_devlink_core_bus_device_reload_down(struct devlink *devlink,
+					  bool netns_change, enum devlink_reload_action action,
+					  enum devlink_reload_limit limit,
+					  struct netlink_ext_ack *extack)
+{
+	struct mlxsw_core *mlxsw_core = devlink_priv(devlink);
+
+	if (!(mlxsw_core->bus->features & MLXSW_BUS_F_RESET))
+		return -EOPNOTSUPP;
+
+	mlxsw_core_bus_device_unregister(mlxsw_core, true);
+	return 0;
+}
+
+static int
+mlxsw_devlink_core_bus_device_reload_up(struct devlink *devlink, enum devlink_reload_action action,
+					enum devlink_reload_limit limit, u32 *actions_performed,
+					struct netlink_ext_ack *extack)
+{
+	struct mlxsw_core *mlxsw_core = devlink_priv(devlink);
+	int err;
+
+	*actions_performed = BIT(DEVLINK_RELOAD_ACTION_DRIVER_REINIT) |
+			     BIT(DEVLINK_RELOAD_ACTION_FW_ACTIVATE);
+	err = mlxsw_core_bus_device_register(mlxsw_core->bus_info,
+					     mlxsw_core->bus,
+					     mlxsw_core->bus_priv, true,
+					     devlink, extack);
+	return err;
+}
+
+static int mlxsw_devlink_flash_update(struct devlink *devlink,
+				      struct devlink_flash_update_params *params,
+				      struct netlink_ext_ack *extack)
+{
+	struct mlxsw_core *mlxsw_core = devlink_priv(devlink);
+
+	return mlxsw_core_fw_flash_update(mlxsw_core, params, extack);
+}
+
+static int mlxsw_devlink_trap_init(struct devlink *devlink,
+				   const struct devlink_trap *trap,
+				   void *trap_ctx)
+{
+	struct mlxsw_core *mlxsw_core = devlink_priv(devlink);
+	struct mlxsw_driver *mlxsw_driver = mlxsw_core->driver;
+
+	if (!mlxsw_driver->trap_init)
+		return -EOPNOTSUPP;
+	return mlxsw_driver->trap_init(mlxsw_core, trap, trap_ctx);
+}
+
+static void mlxsw_devlink_trap_fini(struct devlink *devlink,
+				    const struct devlink_trap *trap,
+				    void *trap_ctx)
+{
+	struct mlxsw_core *mlxsw_core = devlink_priv(devlink);
+	struct mlxsw_driver *mlxsw_driver = mlxsw_core->driver;
+
+	if (!mlxsw_driver->trap_fini)
+		return;
+	mlxsw_driver->trap_fini(mlxsw_core, trap, trap_ctx);
+}
+
+static int mlxsw_devlink_trap_action_set(struct devlink *devlink,
+					 const struct devlink_trap *trap,
+					 enum devlink_trap_action action,
+					 struct netlink_ext_ack *extack)
+{
+	struct mlxsw_core *mlxsw_core = devlink_priv(devlink);
+	struct mlxsw_driver *mlxsw_driver = mlxsw_core->driver;
+
+	if (!mlxsw_driver->trap_action_set)
+		return -EOPNOTSUPP;
+	return mlxsw_driver->trap_action_set(mlxsw_core, trap, action, extack);
+}
+
+static int
+mlxsw_devlink_trap_group_init(struct devlink *devlink,
+			      const struct devlink_trap_group *group)
+{
+	struct mlxsw_core *mlxsw_core = devlink_priv(devlink);
+	struct mlxsw_driver *mlxsw_driver = mlxsw_core->driver;
+
+	if (!mlxsw_driver->trap_group_init)
+		return -EOPNOTSUPP;
+	return mlxsw_driver->trap_group_init(mlxsw_core, group);
+}
+
+static int
+mlxsw_devlink_trap_group_set(struct devlink *devlink,
+			     const struct devlink_trap_group *group,
+			     const struct devlink_trap_policer *policer,
+			     struct netlink_ext_ack *extack)
+{
+	struct mlxsw_core *mlxsw_core = devlink_priv(devlink);
+	struct mlxsw_driver *mlxsw_driver = mlxsw_core->driver;
+
+	if (!mlxsw_driver->trap_group_set)
+		return -EOPNOTSUPP;
+	return mlxsw_driver->trap_group_set(mlxsw_core, group, policer, extack);
+}
+
+static int
+mlxsw_devlink_trap_policer_init(struct devlink *devlink,
+				const struct devlink_trap_policer *policer)
+{
+	struct mlxsw_core *mlxsw_core = devlink_priv(devlink);
+	struct mlxsw_driver *mlxsw_driver = mlxsw_core->driver;
+
+	if (!mlxsw_driver->trap_policer_init)
+		return -EOPNOTSUPP;
+	return mlxsw_driver->trap_policer_init(mlxsw_core, policer);
+}
+
+static void
+mlxsw_devlink_trap_policer_fini(struct devlink *devlink,
+				const struct devlink_trap_policer *policer)
+{
+	struct mlxsw_core *mlxsw_core = devlink_priv(devlink);
+	struct mlxsw_driver *mlxsw_driver = mlxsw_core->driver;
+
+	if (!mlxsw_driver->trap_policer_fini)
+		return;
+	mlxsw_driver->trap_policer_fini(mlxsw_core, policer);
+}
+
+static int
+mlxsw_devlink_trap_policer_set(struct devlink *devlink,
+			       const struct devlink_trap_policer *policer,
+			       u64 rate, u64 burst,
+			       struct netlink_ext_ack *extack)
+{
+	struct mlxsw_core *mlxsw_core = devlink_priv(devlink);
+	struct mlxsw_driver *mlxsw_driver = mlxsw_core->driver;
+
+	if (!mlxsw_driver->trap_policer_set)
+		return -EOPNOTSUPP;
+	return mlxsw_driver->trap_policer_set(mlxsw_core, policer, rate, burst,
+					      extack);
+}
+
+static int
+mlxsw_devlink_trap_policer_counter_get(struct devlink *devlink,
+				       const struct devlink_trap_policer *policer,
+				       u64 *p_drops)
+{
+	struct mlxsw_core *mlxsw_core = devlink_priv(devlink);
+	struct mlxsw_driver *mlxsw_driver = mlxsw_core->driver;
+
+	if (!mlxsw_driver->trap_policer_counter_get)
+		return -EOPNOTSUPP;
+	return mlxsw_driver->trap_policer_counter_get(mlxsw_core, policer,
+						      p_drops);
+}
+
+static const struct devlink_ops mlxsw_devlink_ops = {
+	.reload_actions		= BIT(DEVLINK_RELOAD_ACTION_DRIVER_REINIT) |
+				  BIT(DEVLINK_RELOAD_ACTION_FW_ACTIVATE),
+	.reload_down		= mlxsw_devlink_core_bus_device_reload_down,
+	.reload_up		= mlxsw_devlink_core_bus_device_reload_up,
+	.port_split			= mlxsw_devlink_port_split,
+	.port_unsplit			= mlxsw_devlink_port_unsplit,
+	.sb_pool_get			= mlxsw_devlink_sb_pool_get,
+	.sb_pool_set			= mlxsw_devlink_sb_pool_set,
+	.sb_port_pool_get		= mlxsw_devlink_sb_port_pool_get,
+	.sb_port_pool_set		= mlxsw_devlink_sb_port_pool_set,
+	.sb_tc_pool_bind_get		= mlxsw_devlink_sb_tc_pool_bind_get,
+	.sb_tc_pool_bind_set		= mlxsw_devlink_sb_tc_pool_bind_set,
+	.sb_occ_snapshot		= mlxsw_devlink_sb_occ_snapshot,
+	.sb_occ_max_clear		= mlxsw_devlink_sb_occ_max_clear,
+	.sb_occ_port_pool_get		= mlxsw_devlink_sb_occ_port_pool_get,
+	.sb_occ_tc_port_bind_get	= mlxsw_devlink_sb_occ_tc_port_bind_get,
+	.info_get			= mlxsw_devlink_info_get,
+	.flash_update			= mlxsw_devlink_flash_update,
+	.trap_init			= mlxsw_devlink_trap_init,
+	.trap_fini			= mlxsw_devlink_trap_fini,
+	.trap_action_set		= mlxsw_devlink_trap_action_set,
+	.trap_group_init		= mlxsw_devlink_trap_group_init,
+	.trap_group_set			= mlxsw_devlink_trap_group_set,
+	.trap_policer_init		= mlxsw_devlink_trap_policer_init,
+	.trap_policer_fini		= mlxsw_devlink_trap_policer_fini,
+	.trap_policer_set		= mlxsw_devlink_trap_policer_set,
+	.trap_policer_counter_get	= mlxsw_devlink_trap_policer_counter_get,
+};
+
+static int mlxsw_core_params_register(struct mlxsw_core *mlxsw_core)
+{
+	int err;
+
+	err = mlxsw_core_fw_params_register(mlxsw_core);
+	if (err)
+		return err;
+
+	if (mlxsw_core->driver->params_register) {
+		err = mlxsw_core->driver->params_register(mlxsw_core);
+		if (err)
+			goto err_params_register;
+	}
+	return 0;
+
+err_params_register:
+	mlxsw_core_fw_params_unregister(mlxsw_core);
+	return err;
+}
+
+static void mlxsw_core_params_unregister(struct mlxsw_core *mlxsw_core)
+{
+	mlxsw_core_fw_params_unregister(mlxsw_core);
+	if (mlxsw_core->driver->params_register)
+		mlxsw_core->driver->params_unregister(mlxsw_core);
+}
+
+struct mlxsw_core_health_event {
+	struct mlxsw_core *mlxsw_core;
+	char mfde_pl[MLXSW_REG_MFDE_LEN];
+	struct work_struct work;
+};
+
+static void mlxsw_core_health_event_work(struct work_struct *work)
+{
+	struct mlxsw_core_health_event *event;
+	struct mlxsw_core *mlxsw_core;
+
+	event = container_of(work, struct mlxsw_core_health_event, work);
+	mlxsw_core = event->mlxsw_core;
+	devlink_health_report(mlxsw_core->health.fw_fatal, "FW fatal event occurred",
+			      event->mfde_pl);
+	kfree(event);
+}
+
+static void mlxsw_core_health_listener_func(const struct mlxsw_reg_info *reg,
+					    char *mfde_pl, void *priv)
+{
+	struct mlxsw_core_health_event *event;
+	struct mlxsw_core *mlxsw_core = priv;
+
+	event = kmalloc(sizeof(*event), GFP_ATOMIC);
+	if (!event)
+		return;
+	event->mlxsw_core = mlxsw_core;
+	memcpy(event->mfde_pl, mfde_pl, sizeof(event->mfde_pl));
+	INIT_WORK(&event->work, mlxsw_core_health_event_work);
+	mlxsw_core_schedule_work(&event->work);
+}
+
+static const struct mlxsw_listener mlxsw_core_health_listener =
+	MLXSW_CORE_EVENTL(mlxsw_core_health_listener_func, MFDE);
+
+static int
+mlxsw_core_health_fw_fatal_dump_fatal_cause(const char *mfde_pl,
+					    struct devlink_fmsg *fmsg)
+{
+	u32 val, tile_v;
+	int err;
+
+	val = mlxsw_reg_mfde_fatal_cause_id_get(mfde_pl);
+	err = devlink_fmsg_u32_pair_put(fmsg, "cause_id", val);
+	if (err)
+		return err;
+	tile_v = mlxsw_reg_mfde_fatal_cause_tile_v_get(mfde_pl);
+	if (tile_v) {
+		val = mlxsw_reg_mfde_fatal_cause_tile_index_get(mfde_pl);
+		err = devlink_fmsg_u8_pair_put(fmsg, "tile_index", val);
+		if (err)
+			return err;
+	}
+
+	return 0;
+}
+
+static int
+mlxsw_core_health_fw_fatal_dump_fw_assert(const char *mfde_pl,
+					  struct devlink_fmsg *fmsg)
+{
+	u32 val, tile_v;
+	int err;
+
+	val = mlxsw_reg_mfde_fw_assert_var0_get(mfde_pl);
+	err = devlink_fmsg_u32_pair_put(fmsg, "var0", val);
+	if (err)
+		return err;
+	val = mlxsw_reg_mfde_fw_assert_var1_get(mfde_pl);
+	err = devlink_fmsg_u32_pair_put(fmsg, "var1", val);
+	if (err)
+		return err;
+	val = mlxsw_reg_mfde_fw_assert_var2_get(mfde_pl);
+	err = devlink_fmsg_u32_pair_put(fmsg, "var2", val);
+	if (err)
+		return err;
+	val = mlxsw_reg_mfde_fw_assert_var3_get(mfde_pl);
+	err = devlink_fmsg_u32_pair_put(fmsg, "var3", val);
+	if (err)
+		return err;
+	val = mlxsw_reg_mfde_fw_assert_var4_get(mfde_pl);
+	err = devlink_fmsg_u32_pair_put(fmsg, "var4", val);
+	if (err)
+		return err;
+	val = mlxsw_reg_mfde_fw_assert_existptr_get(mfde_pl);
+	err = devlink_fmsg_u32_pair_put(fmsg, "existptr", val);
+	if (err)
+		return err;
+	val = mlxsw_reg_mfde_fw_assert_callra_get(mfde_pl);
+	err = devlink_fmsg_u32_pair_put(fmsg, "callra", val);
+	if (err)
+		return err;
+	val = mlxsw_reg_mfde_fw_assert_oe_get(mfde_pl);
+	err = devlink_fmsg_bool_pair_put(fmsg, "old_event", val);
+	if (err)
+		return err;
+	tile_v = mlxsw_reg_mfde_fw_assert_tile_v_get(mfde_pl);
+	if (tile_v) {
+		val = mlxsw_reg_mfde_fw_assert_tile_index_get(mfde_pl);
+		err = devlink_fmsg_u8_pair_put(fmsg, "tile_index", val);
+		if (err)
+			return err;
+	}
+	val = mlxsw_reg_mfde_fw_assert_ext_synd_get(mfde_pl);
+	err = devlink_fmsg_u32_pair_put(fmsg, "ext_synd", val);
+	if (err)
+		return err;
+
+	return 0;
+}
+
+static int
+mlxsw_core_health_fw_fatal_dump_kvd_im_stop(const char *mfde_pl,
+					    struct devlink_fmsg *fmsg)
+{
+	u32 val;
+	int err;
+
+	val = mlxsw_reg_mfde_kvd_im_stop_oe_get(mfde_pl);
+	err = devlink_fmsg_bool_pair_put(fmsg, "old_event", val);
+	if (err)
+		return err;
+	val = mlxsw_reg_mfde_kvd_im_stop_pipes_mask_get(mfde_pl);
+	return devlink_fmsg_u32_pair_put(fmsg, "pipes_mask", val);
+}
+
+static int
+mlxsw_core_health_fw_fatal_dump_crspace_to(const char *mfde_pl,
+					   struct devlink_fmsg *fmsg)
+{
+	u32 val;
+	int err;
+
+	val = mlxsw_reg_mfde_crspace_to_log_address_get(mfde_pl);
+	err = devlink_fmsg_u32_pair_put(fmsg, "log_address", val);
+	if (err)
+		return err;
+	val = mlxsw_reg_mfde_crspace_to_oe_get(mfde_pl);
+	err = devlink_fmsg_bool_pair_put(fmsg, "old_event", val);
+	if (err)
+		return err;
+	val = mlxsw_reg_mfde_crspace_to_log_id_get(mfde_pl);
+	err = devlink_fmsg_u8_pair_put(fmsg, "log_irisc_id", val);
+	if (err)
+		return err;
+	val = mlxsw_reg_mfde_crspace_to_log_ip_get(mfde_pl);
+	err = devlink_fmsg_u64_pair_put(fmsg, "log_ip", val);
+	if (err)
+		return err;
+
+	return 0;
+}
+
+static int mlxsw_core_health_fw_fatal_dump(struct devlink_health_reporter *reporter,
+					   struct devlink_fmsg *fmsg, void *priv_ctx,
+					   struct netlink_ext_ack *extack)
+{
+	char *mfde_pl = priv_ctx;
+	char *val_str;
+	u8 event_id;
+	u32 val;
+	int err;
+
+	if (!priv_ctx)
+		/* User-triggered dumps are not possible */
+		return -EOPNOTSUPP;
+
+	val = mlxsw_reg_mfde_irisc_id_get(mfde_pl);
+	err = devlink_fmsg_u8_pair_put(fmsg, "irisc_id", val);
+	if (err)
+		return err;
+	err = devlink_fmsg_arr_pair_nest_start(fmsg, "event");
+	if (err)
+		return err;
+
+	event_id = mlxsw_reg_mfde_event_id_get(mfde_pl);
+	err = devlink_fmsg_u32_pair_put(fmsg, "id", event_id);
+	if (err)
+		return err;
+	switch (event_id) {
+	case MLXSW_REG_MFDE_EVENT_ID_CRSPACE_TO:
+		val_str = "CR space timeout";
+		break;
+	case MLXSW_REG_MFDE_EVENT_ID_KVD_IM_STOP:
+		val_str = "KVD insertion machine stopped";
+		break;
+	case MLXSW_REG_MFDE_EVENT_ID_TEST:
+		val_str = "Test";
+		break;
+	case MLXSW_REG_MFDE_EVENT_ID_FW_ASSERT:
+		val_str = "FW assert";
+		break;
+	case MLXSW_REG_MFDE_EVENT_ID_FATAL_CAUSE:
+		val_str = "Fatal cause";
+		break;
+	default:
+		val_str = NULL;
+	}
+	if (val_str) {
+		err = devlink_fmsg_string_pair_put(fmsg, "desc", val_str);
+		if (err)
+			return err;
+	}
+
+	err = devlink_fmsg_arr_pair_nest_end(fmsg);
+	if (err)
+		return err;
+
+	err = devlink_fmsg_arr_pair_nest_start(fmsg, "severity");
+	if (err)
+		return err;
+
+	val = mlxsw_reg_mfde_severity_get(mfde_pl);
+	err = devlink_fmsg_u8_pair_put(fmsg, "id", val);
+	if (err)
+		return err;
+	switch (val) {
+	case MLXSW_REG_MFDE_SEVERITY_FATL:
+		val_str = "Fatal";
+		break;
+	case MLXSW_REG_MFDE_SEVERITY_NRML:
+		val_str = "Normal";
+		break;
+	case MLXSW_REG_MFDE_SEVERITY_INTR:
+		val_str = "Debug";
+		break;
+	default:
+		val_str = NULL;
+	}
+	if (val_str) {
+		err = devlink_fmsg_string_pair_put(fmsg, "desc", val_str);
+		if (err)
+			return err;
+	}
+
+	err = devlink_fmsg_arr_pair_nest_end(fmsg);
+	if (err)
+		return err;
+
+	val = mlxsw_reg_mfde_method_get(mfde_pl);
+	switch (val) {
+	case MLXSW_REG_MFDE_METHOD_QUERY:
+		val_str = "query";
+		break;
+	case MLXSW_REG_MFDE_METHOD_WRITE:
+		val_str = "write";
+		break;
+	default:
+		val_str = NULL;
+	}
+	if (val_str) {
+		err = devlink_fmsg_string_pair_put(fmsg, "method", val_str);
+		if (err)
+			return err;
+	}
+
+	val = mlxsw_reg_mfde_long_process_get(mfde_pl);
+	err = devlink_fmsg_bool_pair_put(fmsg, "long_process", val);
+	if (err)
+		return err;
+
+	val = mlxsw_reg_mfde_command_type_get(mfde_pl);
+	switch (val) {
+	case MLXSW_REG_MFDE_COMMAND_TYPE_MAD:
+		val_str = "mad";
+		break;
+	case MLXSW_REG_MFDE_COMMAND_TYPE_EMAD:
+		val_str = "emad";
+		break;
+	case MLXSW_REG_MFDE_COMMAND_TYPE_CMDIF:
+		val_str = "cmdif";
+		break;
+	default:
+		val_str = NULL;
+	}
+	if (val_str) {
+		err = devlink_fmsg_string_pair_put(fmsg, "command_type", val_str);
+		if (err)
+			return err;
+	}
+
+	val = mlxsw_reg_mfde_reg_attr_id_get(mfde_pl);
+	err = devlink_fmsg_u32_pair_put(fmsg, "reg_attr_id", val);
+	if (err)
+		return err;
+
+	switch (event_id) {
+	case MLXSW_REG_MFDE_EVENT_ID_CRSPACE_TO:
+		return mlxsw_core_health_fw_fatal_dump_crspace_to(mfde_pl,
+								  fmsg);
+	case MLXSW_REG_MFDE_EVENT_ID_KVD_IM_STOP:
+		return mlxsw_core_health_fw_fatal_dump_kvd_im_stop(mfde_pl,
+								   fmsg);
+	case MLXSW_REG_MFDE_EVENT_ID_FW_ASSERT:
+		return mlxsw_core_health_fw_fatal_dump_fw_assert(mfde_pl, fmsg);
+	case MLXSW_REG_MFDE_EVENT_ID_FATAL_CAUSE:
+		return mlxsw_core_health_fw_fatal_dump_fatal_cause(mfde_pl,
+								   fmsg);
+	}
+
+	return 0;
+}
+
+static int
+mlxsw_core_health_fw_fatal_test(struct devlink_health_reporter *reporter,
+				struct netlink_ext_ack *extack)
+{
+	struct mlxsw_core *mlxsw_core = devlink_health_reporter_priv(reporter);
+	char mfgd_pl[MLXSW_REG_MFGD_LEN];
+	int err;
+
+	/* Read the register first to make sure no other bits are changed. */
+	err = mlxsw_reg_query(mlxsw_core, MLXSW_REG(mfgd), mfgd_pl);
+	if (err)
+		return err;
+	mlxsw_reg_mfgd_trigger_test_set(mfgd_pl, true);
+	return mlxsw_reg_write(mlxsw_core, MLXSW_REG(mfgd), mfgd_pl);
+}
+
+static const struct devlink_health_reporter_ops
+mlxsw_core_health_fw_fatal_ops = {
+	.name = "fw_fatal",
+	.dump = mlxsw_core_health_fw_fatal_dump,
+	.test = mlxsw_core_health_fw_fatal_test,
+};
+
+static int mlxsw_core_health_fw_fatal_config(struct mlxsw_core *mlxsw_core,
+					     bool enable)
+{
+	char mfgd_pl[MLXSW_REG_MFGD_LEN];
+	int err;
+
+	/* Read the register first to make sure no other bits are changed. */
+	err = mlxsw_reg_query(mlxsw_core, MLXSW_REG(mfgd), mfgd_pl);
+	if (err)
+		return err;
+	mlxsw_reg_mfgd_fatal_event_mode_set(mfgd_pl, enable);
+	return mlxsw_reg_write(mlxsw_core, MLXSW_REG(mfgd), mfgd_pl);
+}
+
+static int mlxsw_core_health_init(struct mlxsw_core *mlxsw_core)
+{
+	struct devlink *devlink = priv_to_devlink(mlxsw_core);
+	struct devlink_health_reporter *fw_fatal;
+	int err;
+
+	if (!(mlxsw_core->bus->features & MLXSW_BUS_F_TXRX))
+		return 0;
+
+	fw_fatal = devlink_health_reporter_create(devlink, &mlxsw_core_health_fw_fatal_ops,
+						  0, mlxsw_core);
+	if (IS_ERR(fw_fatal)) {
+		dev_err(mlxsw_core->bus_info->dev, "Failed to create fw fatal reporter");
+		return PTR_ERR(fw_fatal);
+	}
+	mlxsw_core->health.fw_fatal = fw_fatal;
+
+	err = mlxsw_core_trap_register(mlxsw_core, &mlxsw_core_health_listener, mlxsw_core);
+	if (err)
+		goto err_trap_register;
+
+	err = mlxsw_core_health_fw_fatal_config(mlxsw_core, true);
+	if (err)
+		goto err_fw_fatal_config;
+
+	return 0;
+
+err_fw_fatal_config:
+	mlxsw_core_trap_unregister(mlxsw_core, &mlxsw_core_health_listener, mlxsw_core);
+err_trap_register:
+	devlink_health_reporter_destroy(mlxsw_core->health.fw_fatal);
+	return err;
+}
+
+static void mlxsw_core_health_fini(struct mlxsw_core *mlxsw_core)
+{
+	if (!(mlxsw_core->bus->features & MLXSW_BUS_F_TXRX))
+		return;
+
+	mlxsw_core_health_fw_fatal_config(mlxsw_core, false);
+	mlxsw_core_trap_unregister(mlxsw_core, &mlxsw_core_health_listener, mlxsw_core);
+	/* Make sure there is no more event work scheduled */
+	mlxsw_core_flush_owq();
+	devlink_health_reporter_destroy(mlxsw_core->health.fw_fatal);
+}
+
+static void mlxsw_core_irq_event_handler_init(struct mlxsw_core *mlxsw_core)
+{
+	INIT_LIST_HEAD(&mlxsw_core->irq_event_handler_list);
+	mutex_init(&mlxsw_core->irq_event_handler_lock);
+}
+
+static void mlxsw_core_irq_event_handler_fini(struct mlxsw_core *mlxsw_core)
+{
+	mutex_destroy(&mlxsw_core->irq_event_handler_lock);
+	WARN_ON(!list_empty(&mlxsw_core->irq_event_handler_list));
+}
+
+static int
+__mlxsw_core_bus_device_register(const struct mlxsw_bus_info *mlxsw_bus_info,
+				 const struct mlxsw_bus *mlxsw_bus,
+				 void *bus_priv, bool reload,
+				 struct devlink *devlink,
+				 struct netlink_ext_ack *extack)
+{
+	const char *device_kind = mlxsw_bus_info->device_kind;
+	struct mlxsw_core *mlxsw_core;
+	struct mlxsw_driver *mlxsw_driver;
+	size_t alloc_size;
+	u16 max_lag;
+	int err;
+
+	mlxsw_driver = mlxsw_core_driver_get(device_kind);
+	if (!mlxsw_driver)
+		return -EINVAL;
+
+	if (!reload) {
+		alloc_size = sizeof(*mlxsw_core) + mlxsw_driver->priv_size;
+		devlink = devlink_alloc(&mlxsw_devlink_ops, alloc_size,
+					mlxsw_bus_info->dev);
+		if (!devlink) {
+			err = -ENOMEM;
+			goto err_devlink_alloc;
+		}
+		devl_lock(devlink);
+	}
+
+	mlxsw_core = devlink_priv(devlink);
+	INIT_LIST_HEAD(&mlxsw_core->rx_listener_list);
+	INIT_LIST_HEAD(&mlxsw_core->event_listener_list);
+	mlxsw_core->driver = mlxsw_driver;
+	mlxsw_core->bus = mlxsw_bus;
+	mlxsw_core->bus_priv = bus_priv;
+	mlxsw_core->bus_info = mlxsw_bus_info;
+	mlxsw_core_irq_event_handler_init(mlxsw_core);
+
+	err = mlxsw_bus->init(bus_priv, mlxsw_core, mlxsw_driver->profile,
+			      &mlxsw_core->res);
+	if (err)
+		goto err_bus_init;
+
+	if (mlxsw_driver->resources_register && !reload) {
+		err = mlxsw_driver->resources_register(mlxsw_core);
+		if (err)
+			goto err_register_resources;
+	}
+
+	err = mlxsw_ports_init(mlxsw_core, reload);
+	if (err)
+		goto err_ports_init;
+
+	err = mlxsw_core_max_lag(mlxsw_core, &max_lag);
+	if (!err && MLXSW_CORE_RES_VALID(mlxsw_core, MAX_LAG_MEMBERS)) {
+		alloc_size = sizeof(*mlxsw_core->lag.mapping) * max_lag *
+			MLXSW_CORE_RES_GET(mlxsw_core, MAX_LAG_MEMBERS);
+		mlxsw_core->lag.mapping = kzalloc(alloc_size, GFP_KERNEL);
+		if (!mlxsw_core->lag.mapping) {
+			err = -ENOMEM;
+			goto err_alloc_lag_mapping;
+		}
+	}
+
+	err = mlxsw_core_trap_groups_set(mlxsw_core);
+	if (err)
+		goto err_trap_groups_set;
+
+	err = mlxsw_emad_init(mlxsw_core);
+	if (err)
+		goto err_emad_init;
+
+	if (!reload) {
+		err = mlxsw_core_params_register(mlxsw_core);
+		if (err)
+			goto err_register_params;
+	}
+
+	err = mlxsw_core_fw_rev_validate(mlxsw_core, mlxsw_bus_info, mlxsw_driver->fw_req_rev,
+					 mlxsw_driver->fw_filename);
+	if (err)
+		goto err_fw_rev_validate;
+
+	err = mlxsw_linecards_init(mlxsw_core, mlxsw_bus_info);
+	if (err)
+		goto err_linecards_init;
+
+	err = mlxsw_core_health_init(mlxsw_core);
+	if (err)
+		goto err_health_init;
+
+	err = mlxsw_hwmon_init(mlxsw_core, mlxsw_bus_info, &mlxsw_core->hwmon);
+	if (err)
+		goto err_hwmon_init;
+
+	err = mlxsw_thermal_init(mlxsw_core, mlxsw_bus_info,
+				 &mlxsw_core->thermal);
+	if (err)
+		goto err_thermal_init;
+
+	err = mlxsw_env_init(mlxsw_core, mlxsw_bus_info, &mlxsw_core->env);
+	if (err)
+		goto err_env_init;
+
+	if (mlxsw_driver->init) {
+		err = mlxsw_driver->init(mlxsw_core, mlxsw_bus_info, extack);
+		if (err)
+			goto err_driver_init;
+	}
+
+	if (!reload) {
+		devlink_set_features(devlink, DEVLINK_F_RELOAD);
+		devl_unlock(devlink);
+		devlink_register(devlink);
+	}
+	return 0;
+
+err_driver_init:
+	mlxsw_env_fini(mlxsw_core->env);
+err_env_init:
+	mlxsw_thermal_fini(mlxsw_core->thermal);
+err_thermal_init:
+	mlxsw_hwmon_fini(mlxsw_core->hwmon);
+err_hwmon_init:
+	mlxsw_core_health_fini(mlxsw_core);
+err_health_init:
+	mlxsw_linecards_fini(mlxsw_core);
+err_linecards_init:
+err_fw_rev_validate:
+	if (!reload)
+		mlxsw_core_params_unregister(mlxsw_core);
+err_register_params:
+	mlxsw_emad_fini(mlxsw_core);
+err_emad_init:
+err_trap_groups_set:
+	kfree(mlxsw_core->lag.mapping);
+err_alloc_lag_mapping:
+	mlxsw_ports_fini(mlxsw_core, reload);
+err_ports_init:
+	if (!reload)
+		devl_resources_unregister(devlink);
+err_register_resources:
+	mlxsw_bus->fini(bus_priv);
+err_bus_init:
+	mlxsw_core_irq_event_handler_fini(mlxsw_core);
+	if (!reload) {
+		devl_unlock(devlink);
+		devlink_free(devlink);
+	}
+err_devlink_alloc:
+	return err;
+}
+
+int mlxsw_core_bus_device_register(const struct mlxsw_bus_info *mlxsw_bus_info,
+				   const struct mlxsw_bus *mlxsw_bus,
+				   void *bus_priv, bool reload,
+				   struct devlink *devlink,
+				   struct netlink_ext_ack *extack)
+{
+	bool called_again = false;
+	int err;
+
+again:
+	err = __mlxsw_core_bus_device_register(mlxsw_bus_info, mlxsw_bus,
+					       bus_priv, reload,
+					       devlink, extack);
+	/* -EAGAIN is returned in case the FW was updated. FW needs
+	 * a reset, so lets try to call __mlxsw_core_bus_device_register()
+	 * again.
+	 */
+	if (err == -EAGAIN && !called_again) {
+		called_again = true;
+		goto again;
+	}
+
+	return err;
+}
+EXPORT_SYMBOL(mlxsw_core_bus_device_register);
+
+void mlxsw_core_bus_device_unregister(struct mlxsw_core *mlxsw_core,
+				      bool reload)
+{
+	struct devlink *devlink = priv_to_devlink(mlxsw_core);
+
+	if (!reload) {
+		devlink_unregister(devlink);
+		devl_lock(devlink);
+	}
+
+	if (devlink_is_reload_failed(devlink)) {
+		if (!reload)
+			/* Only the parts that were not de-initialized in the
+			 * failed reload attempt need to be de-initialized.
+			 */
+			goto reload_fail_deinit;
+		else
+			return;
+	}
+
+	if (mlxsw_core->driver->fini)
+		mlxsw_core->driver->fini(mlxsw_core);
+	mlxsw_env_fini(mlxsw_core->env);
+	mlxsw_thermal_fini(mlxsw_core->thermal);
+	mlxsw_hwmon_fini(mlxsw_core->hwmon);
+	mlxsw_core_health_fini(mlxsw_core);
+	mlxsw_linecards_fini(mlxsw_core);
+	if (!reload)
+		mlxsw_core_params_unregister(mlxsw_core);
+	mlxsw_emad_fini(mlxsw_core);
+	kfree(mlxsw_core->lag.mapping);
+	mlxsw_ports_fini(mlxsw_core, reload);
+	if (!reload)
+		devl_resources_unregister(devlink);
+	mlxsw_core->bus->fini(mlxsw_core->bus_priv);
+	mlxsw_core_irq_event_handler_fini(mlxsw_core);
+	if (!reload) {
+		devl_unlock(devlink);
+		devlink_free(devlink);
+	}
+
+	return;
+
+reload_fail_deinit:
+	mlxsw_core_params_unregister(mlxsw_core);
+	devl_resources_unregister(devlink);
+	devl_unlock(devlink);
+	devlink_free(devlink);
+}
+EXPORT_SYMBOL(mlxsw_core_bus_device_unregister);
+
+bool mlxsw_core_skb_transmit_busy(struct mlxsw_core *mlxsw_core,
+				  const struct mlxsw_tx_info *tx_info)
+{
+	return mlxsw_core->bus->skb_transmit_busy(mlxsw_core->bus_priv,
+						  tx_info);
+}
+EXPORT_SYMBOL(mlxsw_core_skb_transmit_busy);
+
+int mlxsw_core_skb_transmit(struct mlxsw_core *mlxsw_core, struct sk_buff *skb,
+			    const struct mlxsw_tx_info *tx_info)
+{
+	return mlxsw_core->bus->skb_transmit(mlxsw_core->bus_priv, skb,
+					     tx_info);
+}
+EXPORT_SYMBOL(mlxsw_core_skb_transmit);
+
+void mlxsw_core_ptp_transmitted(struct mlxsw_core *mlxsw_core,
+				struct sk_buff *skb, u16 local_port)
+{
+	if (mlxsw_core->driver->ptp_transmitted)
+		mlxsw_core->driver->ptp_transmitted(mlxsw_core, skb,
+						    local_port);
+}
+EXPORT_SYMBOL(mlxsw_core_ptp_transmitted);
+
+static bool __is_rx_listener_equal(const struct mlxsw_rx_listener *rxl_a,
+				   const struct mlxsw_rx_listener *rxl_b)
+{
+	return (rxl_a->func == rxl_b->func &&
+		rxl_a->local_port == rxl_b->local_port &&
+		rxl_a->trap_id == rxl_b->trap_id &&
+		rxl_a->mirror_reason == rxl_b->mirror_reason);
+}
+
+static struct mlxsw_rx_listener_item *
+__find_rx_listener_item(struct mlxsw_core *mlxsw_core,
+			const struct mlxsw_rx_listener *rxl)
+{
+	struct mlxsw_rx_listener_item *rxl_item;
+
+	list_for_each_entry(rxl_item, &mlxsw_core->rx_listener_list, list) {
+		if (__is_rx_listener_equal(&rxl_item->rxl, rxl))
+			return rxl_item;
+	}
+	return NULL;
+}
+
+int mlxsw_core_rx_listener_register(struct mlxsw_core *mlxsw_core,
+				    const struct mlxsw_rx_listener *rxl,
+				    void *priv, bool enabled)
+{
+	struct mlxsw_rx_listener_item *rxl_item;
+
+	rxl_item = __find_rx_listener_item(mlxsw_core, rxl);
+	if (rxl_item)
+		return -EEXIST;
+	rxl_item = kmalloc(sizeof(*rxl_item), GFP_KERNEL);
+	if (!rxl_item)
+		return -ENOMEM;
+	rxl_item->rxl = *rxl;
+	rxl_item->priv = priv;
+	rxl_item->enabled = enabled;
+
+	list_add_rcu(&rxl_item->list, &mlxsw_core->rx_listener_list);
+	return 0;
+}
+EXPORT_SYMBOL(mlxsw_core_rx_listener_register);
+
+void mlxsw_core_rx_listener_unregister(struct mlxsw_core *mlxsw_core,
+				       const struct mlxsw_rx_listener *rxl)
+{
+	struct mlxsw_rx_listener_item *rxl_item;
+
+	rxl_item = __find_rx_listener_item(mlxsw_core, rxl);
+	if (!rxl_item)
+		return;
+	list_del_rcu(&rxl_item->list);
+	synchronize_rcu();
+	kfree(rxl_item);
+}
+EXPORT_SYMBOL(mlxsw_core_rx_listener_unregister);
+
+static void
+mlxsw_core_rx_listener_state_set(struct mlxsw_core *mlxsw_core,
+				 const struct mlxsw_rx_listener *rxl,
+				 bool enabled)
+{
+	struct mlxsw_rx_listener_item *rxl_item;
+
+	rxl_item = __find_rx_listener_item(mlxsw_core, rxl);
+	if (WARN_ON(!rxl_item))
+		return;
+	rxl_item->enabled = enabled;
+}
+
+static void mlxsw_core_event_listener_func(struct sk_buff *skb, u16 local_port,
+					   void *priv)
+{
+	struct mlxsw_event_listener_item *event_listener_item = priv;
+	struct mlxsw_core *mlxsw_core;
+	struct mlxsw_reg_info reg;
+	char *payload;
+	char *reg_tlv;
+	char *op_tlv;
+
+	mlxsw_core = event_listener_item->mlxsw_core;
+	trace_devlink_hwmsg(priv_to_devlink(mlxsw_core), true, 0,
+			    skb->data, skb->len);
+
+	mlxsw_emad_tlv_parse(skb);
+	op_tlv = mlxsw_emad_op_tlv(skb);
+	reg_tlv = mlxsw_emad_reg_tlv(skb);
+
+	reg.id = mlxsw_emad_op_tlv_register_id_get(op_tlv);
+	reg.len = (mlxsw_emad_reg_tlv_len_get(reg_tlv) - 1) * sizeof(u32);
+	payload = mlxsw_emad_reg_payload(reg_tlv);
+	event_listener_item->el.func(&reg, payload, event_listener_item->priv);
+	dev_kfree_skb(skb);
+}
+
+static bool __is_event_listener_equal(const struct mlxsw_event_listener *el_a,
+				      const struct mlxsw_event_listener *el_b)
+{
+	return (el_a->func == el_b->func &&
+		el_a->trap_id == el_b->trap_id);
+}
+
+static struct mlxsw_event_listener_item *
+__find_event_listener_item(struct mlxsw_core *mlxsw_core,
+			   const struct mlxsw_event_listener *el)
+{
+	struct mlxsw_event_listener_item *el_item;
+
+	list_for_each_entry(el_item, &mlxsw_core->event_listener_list, list) {
+		if (__is_event_listener_equal(&el_item->el, el))
+			return el_item;
+	}
+	return NULL;
+}
+
+int mlxsw_core_event_listener_register(struct mlxsw_core *mlxsw_core,
+				       const struct mlxsw_event_listener *el,
+				       void *priv)
+{
+	int err;
+	struct mlxsw_event_listener_item *el_item;
+	const struct mlxsw_rx_listener rxl = {
+		.func = mlxsw_core_event_listener_func,
+		.local_port = MLXSW_PORT_DONT_CARE,
+		.trap_id = el->trap_id,
+	};
+
+	el_item = __find_event_listener_item(mlxsw_core, el);
+	if (el_item)
+		return -EEXIST;
+	el_item = kmalloc(sizeof(*el_item), GFP_KERNEL);
+	if (!el_item)
+		return -ENOMEM;
+	el_item->mlxsw_core = mlxsw_core;
+	el_item->el = *el;
+	el_item->priv = priv;
+
+	err = mlxsw_core_rx_listener_register(mlxsw_core, &rxl, el_item, true);
+	if (err)
+		goto err_rx_listener_register;
+
+	/* No reason to save item if we did not manage to register an RX
+	 * listener for it.
+	 */
+	list_add_rcu(&el_item->list, &mlxsw_core->event_listener_list);
+
+	return 0;
+
+err_rx_listener_register:
+	kfree(el_item);
+	return err;
+}
+EXPORT_SYMBOL(mlxsw_core_event_listener_register);
+
+void mlxsw_core_event_listener_unregister(struct mlxsw_core *mlxsw_core,
+					  const struct mlxsw_event_listener *el)
+{
+	struct mlxsw_event_listener_item *el_item;
+	const struct mlxsw_rx_listener rxl = {
+		.func = mlxsw_core_event_listener_func,
+		.local_port = MLXSW_PORT_DONT_CARE,
+		.trap_id = el->trap_id,
+	};
+
+	el_item = __find_event_listener_item(mlxsw_core, el);
+	if (!el_item)
+		return;
+	mlxsw_core_rx_listener_unregister(mlxsw_core, &rxl);
+	list_del(&el_item->list);
+	kfree(el_item);
+}
+EXPORT_SYMBOL(mlxsw_core_event_listener_unregister);
+
+static int mlxsw_core_listener_register(struct mlxsw_core *mlxsw_core,
+					const struct mlxsw_listener *listener,
+					void *priv, bool enabled)
+{
+	if (listener->is_event) {
+		WARN_ON(!enabled);
+		return mlxsw_core_event_listener_register(mlxsw_core,
+						&listener->event_listener,
+						priv);
+	} else {
+		return mlxsw_core_rx_listener_register(mlxsw_core,
+						&listener->rx_listener,
+						priv, enabled);
+	}
+}
+
+static void mlxsw_core_listener_unregister(struct mlxsw_core *mlxsw_core,
+				      const struct mlxsw_listener *listener,
+				      void *priv)
+{
+	if (listener->is_event)
+		mlxsw_core_event_listener_unregister(mlxsw_core,
+						     &listener->event_listener);
+	else
+		mlxsw_core_rx_listener_unregister(mlxsw_core,
+						  &listener->rx_listener);
+}
+
+int mlxsw_core_trap_register(struct mlxsw_core *mlxsw_core,
+			     const struct mlxsw_listener *listener, void *priv)
+{
+	enum mlxsw_reg_htgt_trap_group trap_group;
+	enum mlxsw_reg_hpkt_action action;
+	char hpkt_pl[MLXSW_REG_HPKT_LEN];
+	int err;
+
+	if (!(mlxsw_core->bus->features & MLXSW_BUS_F_TXRX))
+		return 0;
+
+	err = mlxsw_core_listener_register(mlxsw_core, listener, priv,
+					   listener->enabled_on_register);
+	if (err)
+		return err;
+
+	action = listener->enabled_on_register ? listener->en_action :
+						 listener->dis_action;
+	trap_group = listener->enabled_on_register ? listener->en_trap_group :
+						     listener->dis_trap_group;
+	mlxsw_reg_hpkt_pack(hpkt_pl, action, listener->trap_id,
+			    trap_group, listener->is_ctrl);
+	err = mlxsw_reg_write(mlxsw_core,  MLXSW_REG(hpkt), hpkt_pl);
+	if (err)
+		goto err_trap_set;
+
+	return 0;
+
+err_trap_set:
+	mlxsw_core_listener_unregister(mlxsw_core, listener, priv);
+	return err;
+}
+EXPORT_SYMBOL(mlxsw_core_trap_register);
+
+void mlxsw_core_trap_unregister(struct mlxsw_core *mlxsw_core,
+				const struct mlxsw_listener *listener,
+				void *priv)
+{
+	char hpkt_pl[MLXSW_REG_HPKT_LEN];
+
+	if (!(mlxsw_core->bus->features & MLXSW_BUS_F_TXRX))
+		return;
+
+	if (!listener->is_event) {
+		mlxsw_reg_hpkt_pack(hpkt_pl, listener->dis_action,
+				    listener->trap_id, listener->dis_trap_group,
+				    listener->is_ctrl);
+		mlxsw_reg_write(mlxsw_core, MLXSW_REG(hpkt), hpkt_pl);
+	}
+
+	mlxsw_core_listener_unregister(mlxsw_core, listener, priv);
+}
+EXPORT_SYMBOL(mlxsw_core_trap_unregister);
+
+int mlxsw_core_traps_register(struct mlxsw_core *mlxsw_core,
+			      const struct mlxsw_listener *listeners,
+			      size_t listeners_count, void *priv)
+{
+	int i, err;
+
+	for (i = 0; i < listeners_count; i++) {
+		err = mlxsw_core_trap_register(mlxsw_core,
+					       &listeners[i],
+					       priv);
+		if (err)
+			goto err_listener_register;
+	}
+	return 0;
+
+err_listener_register:
+	for (i--; i >= 0; i--) {
+		mlxsw_core_trap_unregister(mlxsw_core,
+					   &listeners[i],
+					   priv);
+	}
+	return err;
+}
+EXPORT_SYMBOL(mlxsw_core_traps_register);
+
+void mlxsw_core_traps_unregister(struct mlxsw_core *mlxsw_core,
+				 const struct mlxsw_listener *listeners,
+				 size_t listeners_count, void *priv)
+{
+	int i;
+
+	for (i = 0; i < listeners_count; i++) {
+		mlxsw_core_trap_unregister(mlxsw_core,
+					   &listeners[i],
+					   priv);
+	}
+}
+EXPORT_SYMBOL(mlxsw_core_traps_unregister);
+
+int mlxsw_core_trap_state_set(struct mlxsw_core *mlxsw_core,
+			      const struct mlxsw_listener *listener,
+			      bool enabled)
+{
+	enum mlxsw_reg_htgt_trap_group trap_group;
+	enum mlxsw_reg_hpkt_action action;
+	char hpkt_pl[MLXSW_REG_HPKT_LEN];
+	int err;
+
+	/* Not supported for event listener */
+	if (WARN_ON(listener->is_event))
+		return -EINVAL;
+
+	action = enabled ? listener->en_action : listener->dis_action;
+	trap_group = enabled ? listener->en_trap_group :
+			       listener->dis_trap_group;
+	mlxsw_reg_hpkt_pack(hpkt_pl, action, listener->trap_id,
+			    trap_group, listener->is_ctrl);
+	err = mlxsw_reg_write(mlxsw_core, MLXSW_REG(hpkt), hpkt_pl);
+	if (err)
+		return err;
+
+	mlxsw_core_rx_listener_state_set(mlxsw_core, &listener->rx_listener,
+					 enabled);
+	return 0;
+}
+EXPORT_SYMBOL(mlxsw_core_trap_state_set);
+
+static u64 mlxsw_core_tid_get(struct mlxsw_core *mlxsw_core)
+{
+	return atomic64_inc_return(&mlxsw_core->emad.tid);
+}
+
+static int mlxsw_core_reg_access_emad(struct mlxsw_core *mlxsw_core,
+				      const struct mlxsw_reg_info *reg,
+				      char *payload,
+				      enum mlxsw_core_reg_access_type type,
+				      struct list_head *bulk_list,
+				      mlxsw_reg_trans_cb_t *cb,
+				      unsigned long cb_priv)
+{
+	u64 tid = mlxsw_core_tid_get(mlxsw_core);
+	struct mlxsw_reg_trans *trans;
+	int err;
+
+	trans = kzalloc(sizeof(*trans), GFP_KERNEL);
+	if (!trans)
+		return -ENOMEM;
+
+	err = mlxsw_emad_reg_access(mlxsw_core, reg, payload, type, trans,
+				    bulk_list, cb, cb_priv, tid);
+	if (err) {
+		kfree_rcu(trans, rcu);
+		return err;
+	}
+	return 0;
+}
+
+int mlxsw_reg_trans_query(struct mlxsw_core *mlxsw_core,
+			  const struct mlxsw_reg_info *reg, char *payload,
+			  struct list_head *bulk_list,
+			  mlxsw_reg_trans_cb_t *cb, unsigned long cb_priv)
+{
+	return mlxsw_core_reg_access_emad(mlxsw_core, reg, payload,
+					  MLXSW_CORE_REG_ACCESS_TYPE_QUERY,
+					  bulk_list, cb, cb_priv);
+}
+EXPORT_SYMBOL(mlxsw_reg_trans_query);
+
+int mlxsw_reg_trans_write(struct mlxsw_core *mlxsw_core,
+			  const struct mlxsw_reg_info *reg, char *payload,
+			  struct list_head *bulk_list,
+			  mlxsw_reg_trans_cb_t *cb, unsigned long cb_priv)
+{
+	return mlxsw_core_reg_access_emad(mlxsw_core, reg, payload,
+					  MLXSW_CORE_REG_ACCESS_TYPE_WRITE,
+					  bulk_list, cb, cb_priv);
+}
+EXPORT_SYMBOL(mlxsw_reg_trans_write);
+
+#define MLXSW_REG_TRANS_ERR_STRING_SIZE	256
+
+static int mlxsw_reg_trans_wait(struct mlxsw_reg_trans *trans)
+{
+	char err_string[MLXSW_REG_TRANS_ERR_STRING_SIZE];
+	struct mlxsw_core *mlxsw_core = trans->core;
+	int err;
+
+	wait_for_completion(&trans->completion);
+	cancel_delayed_work_sync(&trans->timeout_dw);
+	err = trans->err;
+
+	if (trans->retries)
+		dev_warn(mlxsw_core->bus_info->dev, "EMAD retries (%d/%d) (tid=%llx)\n",
+			 trans->retries, MLXSW_EMAD_MAX_RETRY, trans->tid);
+	if (err) {
+		dev_err(mlxsw_core->bus_info->dev, "EMAD reg access failed (tid=%llx,reg_id=%x(%s),type=%s,status=%x(%s))\n",
+			trans->tid, trans->reg->id,
+			mlxsw_reg_id_str(trans->reg->id),
+			mlxsw_core_reg_access_type_str(trans->type),
+			trans->emad_status,
+			mlxsw_emad_op_tlv_status_str(trans->emad_status));
+
+		snprintf(err_string, MLXSW_REG_TRANS_ERR_STRING_SIZE,
+			 "(tid=%llx,reg_id=%x(%s)) %s (%s)\n", trans->tid,
+			 trans->reg->id, mlxsw_reg_id_str(trans->reg->id),
+			 mlxsw_emad_op_tlv_status_str(trans->emad_status),
+			 trans->emad_err_string ? trans->emad_err_string : "");
+
+		trace_devlink_hwerr(priv_to_devlink(mlxsw_core),
+				    trans->emad_status, err_string);
+
+		kfree(trans->emad_err_string);
+	}
+
+	list_del(&trans->bulk_list);
+	kfree_rcu(trans, rcu);
+	return err;
+}
+
+int mlxsw_reg_trans_bulk_wait(struct list_head *bulk_list)
+{
+	struct mlxsw_reg_trans *trans;
+	struct mlxsw_reg_trans *tmp;
+	int sum_err = 0;
+	int err;
+
+	list_for_each_entry_safe(trans, tmp, bulk_list, bulk_list) {
+		err = mlxsw_reg_trans_wait(trans);
+		if (err && sum_err == 0)
+			sum_err = err; /* first error to be returned */
+	}
+	return sum_err;
+}
+EXPORT_SYMBOL(mlxsw_reg_trans_bulk_wait);
+
+struct mlxsw_core_irq_event_handler_item {
+	struct list_head list;
+	void (*cb)(struct mlxsw_core *mlxsw_core);
+};
+
+int mlxsw_core_irq_event_handler_register(struct mlxsw_core *mlxsw_core,
+					  mlxsw_irq_event_cb_t cb)
+{
+	struct mlxsw_core_irq_event_handler_item *item;
+
+	item = kzalloc(sizeof(*item), GFP_KERNEL);
+	if (!item)
+		return -ENOMEM;
+	item->cb = cb;
+	mutex_lock(&mlxsw_core->irq_event_handler_lock);
+	list_add_tail(&item->list, &mlxsw_core->irq_event_handler_list);
+	mutex_unlock(&mlxsw_core->irq_event_handler_lock);
+	return 0;
+}
+EXPORT_SYMBOL(mlxsw_core_irq_event_handler_register);
+
+void mlxsw_core_irq_event_handler_unregister(struct mlxsw_core *mlxsw_core,
+					     mlxsw_irq_event_cb_t cb)
+{
+	struct mlxsw_core_irq_event_handler_item *item, *tmp;
+
+	mutex_lock(&mlxsw_core->irq_event_handler_lock);
+	list_for_each_entry_safe(item, tmp,
+				 &mlxsw_core->irq_event_handler_list, list) {
+		if (item->cb == cb) {
+			list_del(&item->list);
+			kfree(item);
+		}
+	}
+	mutex_unlock(&mlxsw_core->irq_event_handler_lock);
+}
+EXPORT_SYMBOL(mlxsw_core_irq_event_handler_unregister);
+
+void mlxsw_core_irq_event_handlers_call(struct mlxsw_core *mlxsw_core)
+{
+	struct mlxsw_core_irq_event_handler_item *item;
+
+	mutex_lock(&mlxsw_core->irq_event_handler_lock);
+	list_for_each_entry(item, &mlxsw_core->irq_event_handler_list, list) {
+		if (item->cb)
+			item->cb(mlxsw_core);
+	}
+	mutex_unlock(&mlxsw_core->irq_event_handler_lock);
+}
+EXPORT_SYMBOL(mlxsw_core_irq_event_handlers_call);
+
+static int mlxsw_core_reg_access_cmd(struct mlxsw_core *mlxsw_core,
+				     const struct mlxsw_reg_info *reg,
+				     char *payload,
+				     enum mlxsw_core_reg_access_type type)
+{
+	enum mlxsw_emad_op_tlv_status status;
+	int err, n_retry;
+	bool reset_ok;
+	char *in_mbox, *out_mbox, *tmp;
+
+	dev_dbg(mlxsw_core->bus_info->dev, "Reg cmd access (reg_id=%x(%s),type=%s)\n",
+		reg->id, mlxsw_reg_id_str(reg->id),
+		mlxsw_core_reg_access_type_str(type));
+
+	in_mbox = mlxsw_cmd_mbox_alloc();
+	if (!in_mbox)
+		return -ENOMEM;
+
+	out_mbox = mlxsw_cmd_mbox_alloc();
+	if (!out_mbox) {
+		err = -ENOMEM;
+		goto free_in_mbox;
+	}
+
+	mlxsw_emad_pack_op_tlv(in_mbox, reg, type,
+			       mlxsw_core_tid_get(mlxsw_core));
+	tmp = in_mbox + MLXSW_EMAD_OP_TLV_LEN * sizeof(u32);
+	mlxsw_emad_pack_reg_tlv(tmp, reg, payload);
+
+	/* There is a special treatment needed for MRSR (reset) register.
+	 * The command interface will return error after the command
+	 * is executed, so tell the lower layer to expect it
+	 * and cope accordingly.
+	 */
+	reset_ok = reg->id == MLXSW_REG_MRSR_ID;
+
+	n_retry = 0;
+retry:
+	err = mlxsw_cmd_access_reg(mlxsw_core, reset_ok, in_mbox, out_mbox);
+	if (!err) {
+		err = mlxsw_emad_process_status(out_mbox, &status);
+		if (err) {
+			if (err == -EAGAIN && n_retry++ < MLXSW_EMAD_MAX_RETRY)
+				goto retry;
+			dev_err(mlxsw_core->bus_info->dev, "Reg cmd access status failed (status=%x(%s))\n",
+				status, mlxsw_emad_op_tlv_status_str(status));
+		}
+	}
+
+	if (!err)
+		memcpy(payload, mlxsw_emad_reg_payload_cmd(out_mbox),
+		       reg->len);
+
+	mlxsw_cmd_mbox_free(out_mbox);
+free_in_mbox:
+	mlxsw_cmd_mbox_free(in_mbox);
+	if (err)
+		dev_err(mlxsw_core->bus_info->dev, "Reg cmd access failed (reg_id=%x(%s),type=%s)\n",
+			reg->id, mlxsw_reg_id_str(reg->id),
+			mlxsw_core_reg_access_type_str(type));
+	return err;
+}
+
+static void mlxsw_core_reg_access_cb(struct mlxsw_core *mlxsw_core,
+				     char *payload, size_t payload_len,
+				     unsigned long cb_priv)
+{
+	char *orig_payload = (char *) cb_priv;
+
+	memcpy(orig_payload, payload, payload_len);
+}
+
+static int mlxsw_core_reg_access(struct mlxsw_core *mlxsw_core,
+				 const struct mlxsw_reg_info *reg,
+				 char *payload,
+				 enum mlxsw_core_reg_access_type type)
+{
+	LIST_HEAD(bulk_list);
+	int err;
+
+	/* During initialization EMAD interface is not available to us,
+	 * so we default to command interface. We switch to EMAD interface
+	 * after setting the appropriate traps.
+	 */
+	if (!mlxsw_core->emad.use_emad)
+		return mlxsw_core_reg_access_cmd(mlxsw_core, reg,
+						 payload, type);
+
+	err = mlxsw_core_reg_access_emad(mlxsw_core, reg,
+					 payload, type, &bulk_list,
+					 mlxsw_core_reg_access_cb,
+					 (unsigned long) payload);
+	if (err)
+		return err;
+	return mlxsw_reg_trans_bulk_wait(&bulk_list);
+}
+
+int mlxsw_reg_query(struct mlxsw_core *mlxsw_core,
+		    const struct mlxsw_reg_info *reg, char *payload)
+{
+	return mlxsw_core_reg_access(mlxsw_core, reg, payload,
+				     MLXSW_CORE_REG_ACCESS_TYPE_QUERY);
+}
+EXPORT_SYMBOL(mlxsw_reg_query);
+
+int mlxsw_reg_write(struct mlxsw_core *mlxsw_core,
+		    const struct mlxsw_reg_info *reg, char *payload)
+{
+	return mlxsw_core_reg_access(mlxsw_core, reg, payload,
+				     MLXSW_CORE_REG_ACCESS_TYPE_WRITE);
+}
+EXPORT_SYMBOL(mlxsw_reg_write);
+
+void mlxsw_core_skb_receive(struct mlxsw_core *mlxsw_core, struct sk_buff *skb,
+			    struct mlxsw_rx_info *rx_info)
+{
+	struct mlxsw_rx_listener_item *rxl_item;
+	const struct mlxsw_rx_listener *rxl;
+	u16 local_port;
+	bool found = false;
+
+	if (rx_info->is_lag) {
+		dev_dbg_ratelimited(mlxsw_core->bus_info->dev, "%s: lag_id = %d, lag_port_index = 0x%x\n",
+				    __func__, rx_info->u.lag_id,
+				    rx_info->trap_id);
+		/* Upper layer does not care if the skb came from LAG or not,
+		 * so just get the local_port for the lag port and push it up.
+		 */
+		local_port = mlxsw_core_lag_mapping_get(mlxsw_core,
+							rx_info->u.lag_id,
+							rx_info->lag_port_index);
+	} else {
+		local_port = rx_info->u.sys_port;
+	}
+
+	dev_dbg_ratelimited(mlxsw_core->bus_info->dev, "%s: local_port = %d, trap_id = 0x%x\n",
+			    __func__, local_port, rx_info->trap_id);
+
+	if ((rx_info->trap_id >= MLXSW_TRAP_ID_MAX) ||
+	    (local_port >= mlxsw_core->max_ports))
+		goto drop;
+
+	rcu_read_lock();
+	list_for_each_entry_rcu(rxl_item, &mlxsw_core->rx_listener_list, list) {
+		rxl = &rxl_item->rxl;
+		if ((rxl->local_port == MLXSW_PORT_DONT_CARE ||
+		     rxl->local_port == local_port) &&
+		    rxl->trap_id == rx_info->trap_id &&
+		    rxl->mirror_reason == rx_info->mirror_reason) {
+			if (rxl_item->enabled)
+				found = true;
+			break;
+		}
+	}
+	if (!found) {
+		rcu_read_unlock();
+		goto drop;
+	}
+
+	rxl->func(skb, local_port, rxl_item->priv);
+	rcu_read_unlock();
+	return;
+
+drop:
+	dev_kfree_skb(skb);
+}
+EXPORT_SYMBOL(mlxsw_core_skb_receive);
+
+static int mlxsw_core_lag_mapping_index(struct mlxsw_core *mlxsw_core,
+					u16 lag_id, u8 port_index)
+{
+	return MLXSW_CORE_RES_GET(mlxsw_core, MAX_LAG_MEMBERS) * lag_id +
+	       port_index;
+}
+
+void mlxsw_core_lag_mapping_set(struct mlxsw_core *mlxsw_core,
+				u16 lag_id, u8 port_index, u16 local_port)
+{
+	int index = mlxsw_core_lag_mapping_index(mlxsw_core,
+						 lag_id, port_index);
+
+	mlxsw_core->lag.mapping[index] = local_port;
+}
+EXPORT_SYMBOL(mlxsw_core_lag_mapping_set);
+
+u16 mlxsw_core_lag_mapping_get(struct mlxsw_core *mlxsw_core,
+			       u16 lag_id, u8 port_index)
+{
+	int index = mlxsw_core_lag_mapping_index(mlxsw_core,
+						 lag_id, port_index);
+
+	return mlxsw_core->lag.mapping[index];
+}
+EXPORT_SYMBOL(mlxsw_core_lag_mapping_get);
+
+void mlxsw_core_lag_mapping_clear(struct mlxsw_core *mlxsw_core,
+				  u16 lag_id, u16 local_port)
+{
+	int i;
+
+	for (i = 0; i < MLXSW_CORE_RES_GET(mlxsw_core, MAX_LAG_MEMBERS); i++) {
+		int index = mlxsw_core_lag_mapping_index(mlxsw_core,
+							 lag_id, i);
+
+		if (mlxsw_core->lag.mapping[index] == local_port)
+			mlxsw_core->lag.mapping[index] = 0;
+	}
+}
+EXPORT_SYMBOL(mlxsw_core_lag_mapping_clear);
+
+bool mlxsw_core_res_valid(struct mlxsw_core *mlxsw_core,
+			  enum mlxsw_res_id res_id)
+{
+	return mlxsw_res_valid(&mlxsw_core->res, res_id);
+}
+EXPORT_SYMBOL(mlxsw_core_res_valid);
+
+u64 mlxsw_core_res_get(struct mlxsw_core *mlxsw_core,
+		       enum mlxsw_res_id res_id)
+{
+	return mlxsw_res_get(&mlxsw_core->res, res_id);
+}
+EXPORT_SYMBOL(mlxsw_core_res_get);
+
+static int __mlxsw_core_port_init(struct mlxsw_core *mlxsw_core, u16 local_port,
+				  enum devlink_port_flavour flavour,
+				  u8 slot_index, u32 port_number, bool split,
+				  u32 split_port_subnumber,
+				  bool splittable, u32 lanes,
+				  const unsigned char *switch_id,
+				  unsigned char switch_id_len)
+{
+	struct devlink *devlink = priv_to_devlink(mlxsw_core);
+	struct mlxsw_core_port *mlxsw_core_port =
+					&mlxsw_core->ports[local_port];
+	struct devlink_port *devlink_port = &mlxsw_core_port->devlink_port;
+	struct devlink_port_attrs attrs = {};
+	int err;
+
+	attrs.split = split;
+	attrs.lanes = lanes;
+	attrs.splittable = splittable;
+	attrs.flavour = flavour;
+	attrs.phys.port_number = port_number;
+	attrs.phys.split_subport_number = split_port_subnumber;
+	memcpy(attrs.switch_id.id, switch_id, switch_id_len);
+	attrs.switch_id.id_len = switch_id_len;
+	mlxsw_core_port->local_port = local_port;
+	devlink_port_attrs_set(devlink_port, &attrs);
+	if (slot_index) {
+		struct mlxsw_linecard *linecard;
+
+		linecard = mlxsw_linecard_get(mlxsw_core->linecards,
+					      slot_index);
+		mlxsw_core_port->linecard = linecard;
+		devlink_port_linecard_set(devlink_port,
+					  linecard->devlink_linecard);
+	}
+	err = devl_port_register(devlink, devlink_port, local_port);
+	if (err)
+		memset(mlxsw_core_port, 0, sizeof(*mlxsw_core_port));
+	return err;
+}
+
+static void __mlxsw_core_port_fini(struct mlxsw_core *mlxsw_core, u16 local_port)
+{
+	struct mlxsw_core_port *mlxsw_core_port =
+					&mlxsw_core->ports[local_port];
+	struct devlink_port *devlink_port = &mlxsw_core_port->devlink_port;
+
+	devl_port_unregister(devlink_port);
+	memset(mlxsw_core_port, 0, sizeof(*mlxsw_core_port));
+}
+
+int mlxsw_core_port_init(struct mlxsw_core *mlxsw_core, u16 local_port,
+			 u8 slot_index, u32 port_number, bool split,
+			 u32 split_port_subnumber,
+			 bool splittable, u32 lanes,
+			 const unsigned char *switch_id,
+			 unsigned char switch_id_len)
+{
+	int err;
+
+	err = __mlxsw_core_port_init(mlxsw_core, local_port,
+				     DEVLINK_PORT_FLAVOUR_PHYSICAL, slot_index,
+				     port_number, split, split_port_subnumber,
+				     splittable, lanes,
+				     switch_id, switch_id_len);
+	if (err)
+		return err;
+
+	atomic_inc(&mlxsw_core->active_ports_count);
+	return 0;
+}
+EXPORT_SYMBOL(mlxsw_core_port_init);
+
+void mlxsw_core_port_fini(struct mlxsw_core *mlxsw_core, u16 local_port)
+{
+	atomic_dec(&mlxsw_core->active_ports_count);
+
+	__mlxsw_core_port_fini(mlxsw_core, local_port);
+}
+EXPORT_SYMBOL(mlxsw_core_port_fini);
+
+int mlxsw_core_cpu_port_init(struct mlxsw_core *mlxsw_core,
+			     void *port_driver_priv,
+			     const unsigned char *switch_id,
+			     unsigned char switch_id_len)
+{
+	struct mlxsw_core_port *mlxsw_core_port =
+				&mlxsw_core->ports[MLXSW_PORT_CPU_PORT];
+	int err;
+
+	err = __mlxsw_core_port_init(mlxsw_core, MLXSW_PORT_CPU_PORT,
+				     DEVLINK_PORT_FLAVOUR_CPU,
+				     0, 0, false, 0, false, 0,
+				     switch_id, switch_id_len);
+	if (err)
+		return err;
+
+	mlxsw_core_port->port_driver_priv = port_driver_priv;
+	return 0;
+}
+EXPORT_SYMBOL(mlxsw_core_cpu_port_init);
+
+void mlxsw_core_cpu_port_fini(struct mlxsw_core *mlxsw_core)
+{
+	__mlxsw_core_port_fini(mlxsw_core, MLXSW_PORT_CPU_PORT);
+}
+EXPORT_SYMBOL(mlxsw_core_cpu_port_fini);
+
+void mlxsw_core_port_eth_set(struct mlxsw_core *mlxsw_core, u16 local_port,
+			     void *port_driver_priv, struct net_device *dev)
+{
+	struct mlxsw_core_port *mlxsw_core_port =
+					&mlxsw_core->ports[local_port];
+	struct devlink_port *devlink_port = &mlxsw_core_port->devlink_port;
+
+	mlxsw_core_port->port_driver_priv = port_driver_priv;
+	devlink_port_type_eth_set(devlink_port, dev);
+}
+EXPORT_SYMBOL(mlxsw_core_port_eth_set);
+
+void mlxsw_core_port_clear(struct mlxsw_core *mlxsw_core, u16 local_port,
+			   void *port_driver_priv)
+{
+	struct mlxsw_core_port *mlxsw_core_port =
+					&mlxsw_core->ports[local_port];
+	struct devlink_port *devlink_port = &mlxsw_core_port->devlink_port;
+
+	mlxsw_core_port->port_driver_priv = port_driver_priv;
+	devlink_port_type_clear(devlink_port);
+}
+EXPORT_SYMBOL(mlxsw_core_port_clear);
+
+struct devlink_port *
+mlxsw_core_port_devlink_port_get(struct mlxsw_core *mlxsw_core,
+				 u16 local_port)
+{
+	struct mlxsw_core_port *mlxsw_core_port =
+					&mlxsw_core->ports[local_port];
+	struct devlink_port *devlink_port = &mlxsw_core_port->devlink_port;
+
+	return devlink_port;
+}
+EXPORT_SYMBOL(mlxsw_core_port_devlink_port_get);
+
+struct mlxsw_linecard *
+mlxsw_core_port_linecard_get(struct mlxsw_core *mlxsw_core,
+			     u16 local_port)
+{
+	struct mlxsw_core_port *mlxsw_core_port =
+					&mlxsw_core->ports[local_port];
+
+	return mlxsw_core_port->linecard;
+}
+
+void mlxsw_core_ports_remove_selected(struct mlxsw_core *mlxsw_core,
+				      bool (*selector)(void *priv, u16 local_port),
+				      void *priv)
+{
+	if (WARN_ON_ONCE(!mlxsw_core->driver->ports_remove_selected))
+		return;
+	mlxsw_core->driver->ports_remove_selected(mlxsw_core, selector, priv);
+}
+
+struct mlxsw_env *mlxsw_core_env(const struct mlxsw_core *mlxsw_core)
+{
+	return mlxsw_core->env;
+}
+
+static void mlxsw_core_buf_dump_dbg(struct mlxsw_core *mlxsw_core,
+				    const char *buf, size_t size)
+{
+	__be32 *m = (__be32 *) buf;
+	int i;
+	int count = size / sizeof(__be32);
+
+	for (i = count - 1; i >= 0; i--)
+		if (m[i])
+			break;
+	i++;
+	count = i ? i : 1;
+	for (i = 0; i < count; i += 4)
+		dev_dbg(mlxsw_core->bus_info->dev, "%04x - %08x %08x %08x %08x\n",
+			i * 4, be32_to_cpu(m[i]), be32_to_cpu(m[i + 1]),
+			be32_to_cpu(m[i + 2]), be32_to_cpu(m[i + 3]));
+}
+
+int mlxsw_cmd_exec(struct mlxsw_core *mlxsw_core, u16 opcode, u8 opcode_mod,
+		   u32 in_mod, bool out_mbox_direct, bool reset_ok,
+		   char *in_mbox, size_t in_mbox_size,
+		   char *out_mbox, size_t out_mbox_size)
+{
+	u8 status;
+	int err;
+
+	BUG_ON(in_mbox_size % sizeof(u32) || out_mbox_size % sizeof(u32));
+	if (!mlxsw_core->bus->cmd_exec)
+		return -EOPNOTSUPP;
+
+	dev_dbg(mlxsw_core->bus_info->dev, "Cmd exec (opcode=%x(%s),opcode_mod=%x,in_mod=%x)\n",
+		opcode, mlxsw_cmd_opcode_str(opcode), opcode_mod, in_mod);
+	if (in_mbox) {
+		dev_dbg(mlxsw_core->bus_info->dev, "Input mailbox:\n");
+		mlxsw_core_buf_dump_dbg(mlxsw_core, in_mbox, in_mbox_size);
+	}
+
+	err = mlxsw_core->bus->cmd_exec(mlxsw_core->bus_priv, opcode,
+					opcode_mod, in_mod, out_mbox_direct,
+					in_mbox, in_mbox_size,
+					out_mbox, out_mbox_size, &status);
+
+	if (!err && out_mbox) {
+		dev_dbg(mlxsw_core->bus_info->dev, "Output mailbox:\n");
+		mlxsw_core_buf_dump_dbg(mlxsw_core, out_mbox, out_mbox_size);
+	}
+
+	if (reset_ok && err == -EIO &&
+	    status == MLXSW_CMD_STATUS_RUNNING_RESET) {
+		err = 0;
+	} else if (err == -EIO && status != MLXSW_CMD_STATUS_OK) {
+		dev_err(mlxsw_core->bus_info->dev, "Cmd exec failed (opcode=%x(%s),opcode_mod=%x,in_mod=%x,status=%x(%s))\n",
+			opcode, mlxsw_cmd_opcode_str(opcode), opcode_mod,
+			in_mod, status, mlxsw_cmd_status_str(status));
+	} else if (err == -ETIMEDOUT) {
+		dev_err(mlxsw_core->bus_info->dev, "Cmd exec timed-out (opcode=%x(%s),opcode_mod=%x,in_mod=%x)\n",
+			opcode, mlxsw_cmd_opcode_str(opcode), opcode_mod,
+			in_mod);
+	}
+
+	return err;
+}
+EXPORT_SYMBOL(mlxsw_cmd_exec);
+
+int mlxsw_core_schedule_dw(struct delayed_work *dwork, unsigned long delay)
+{
+	return queue_delayed_work(mlxsw_wq, dwork, delay);
+}
+EXPORT_SYMBOL(mlxsw_core_schedule_dw);
+
+bool mlxsw_core_schedule_work(struct work_struct *work)
+{
+	return queue_work(mlxsw_owq, work);
+}
+EXPORT_SYMBOL(mlxsw_core_schedule_work);
+
+void mlxsw_core_flush_owq(void)
+{
+	flush_workqueue(mlxsw_owq);
+}
+EXPORT_SYMBOL(mlxsw_core_flush_owq);
+
+int mlxsw_core_kvd_sizes_get(struct mlxsw_core *mlxsw_core,
+			     const struct mlxsw_config_profile *profile,
+			     u64 *p_single_size, u64 *p_double_size,
+			     u64 *p_linear_size)
+{
+	struct mlxsw_driver *driver = mlxsw_core->driver;
+
+	if (!driver->kvd_sizes_get)
+		return -EINVAL;
+
+	return driver->kvd_sizes_get(mlxsw_core, profile,
+				     p_single_size, p_double_size,
+				     p_linear_size);
+}
+EXPORT_SYMBOL(mlxsw_core_kvd_sizes_get);
+
+int mlxsw_core_resources_query(struct mlxsw_core *mlxsw_core, char *mbox,
+			       struct mlxsw_res *res)
+{
+	int index, i;
+	u64 data;
+	u16 id;
+	int err;
+
+	mlxsw_cmd_mbox_zero(mbox);
+
+	for (index = 0; index < MLXSW_CMD_QUERY_RESOURCES_MAX_QUERIES;
+	     index++) {
+		err = mlxsw_cmd_query_resources(mlxsw_core, mbox, index);
+		if (err)
+			return err;
+
+		for (i = 0; i < MLXSW_CMD_QUERY_RESOURCES_PER_QUERY; i++) {
+			id = mlxsw_cmd_mbox_query_resource_id_get(mbox, i);
+			data = mlxsw_cmd_mbox_query_resource_data_get(mbox, i);
+
+			if (id == MLXSW_CMD_QUERY_RESOURCES_TABLE_END_ID)
+				return 0;
+
+			mlxsw_res_parse(res, id, data);
+		}
+	}
+
+	/* If after MLXSW_RESOURCES_QUERY_MAX_QUERIES we still didn't get
+	 * MLXSW_RESOURCES_TABLE_END_ID, something went bad in the FW.
+	 */
+	return -EIO;
+}
+EXPORT_SYMBOL(mlxsw_core_resources_query);
+
+u32 mlxsw_core_read_frc_h(struct mlxsw_core *mlxsw_core)
+{
+	return mlxsw_core->bus->read_frc_h(mlxsw_core->bus_priv);
+}
+EXPORT_SYMBOL(mlxsw_core_read_frc_h);
+
+u32 mlxsw_core_read_frc_l(struct mlxsw_core *mlxsw_core)
+{
+	return mlxsw_core->bus->read_frc_l(mlxsw_core->bus_priv);
+}
+EXPORT_SYMBOL(mlxsw_core_read_frc_l);
+
+u32 mlxsw_core_read_utc_sec(struct mlxsw_core *mlxsw_core)
+{
+	return mlxsw_core->bus->read_utc_sec(mlxsw_core->bus_priv);
+}
+EXPORT_SYMBOL(mlxsw_core_read_utc_sec);
+
+u32 mlxsw_core_read_utc_nsec(struct mlxsw_core *mlxsw_core)
+{
+	return mlxsw_core->bus->read_utc_nsec(mlxsw_core->bus_priv);
+}
+EXPORT_SYMBOL(mlxsw_core_read_utc_nsec);
+
+bool mlxsw_core_sdq_supports_cqe_v2(struct mlxsw_core *mlxsw_core)
+{
+	return mlxsw_core->driver->sdq_supports_cqe_v2;
+}
+EXPORT_SYMBOL(mlxsw_core_sdq_supports_cqe_v2);
+
+void mlxsw_core_emad_string_tlv_enable(struct mlxsw_core *mlxsw_core)
+{
+	mlxsw_core->emad.enable_string_tlv = true;
+}
+EXPORT_SYMBOL(mlxsw_core_emad_string_tlv_enable);
+
+static int __init mlxsw_core_module_init(void)
+{
+	int err;
+
+	err = mlxsw_linecard_driver_register();
+	if (err)
+		return err;
+
+	mlxsw_wq = alloc_workqueue(mlxsw_core_driver_name, 0, 0);
+	if (!mlxsw_wq) {
+		err = -ENOMEM;
+		goto err_alloc_workqueue;
+	}
+	mlxsw_owq = alloc_ordered_workqueue("%s_ordered", 0,
+					    mlxsw_core_driver_name);
+	if (!mlxsw_owq) {
+		err = -ENOMEM;
+		goto err_alloc_ordered_workqueue;
+	}
+	return 0;
+
+err_alloc_ordered_workqueue:
+	destroy_workqueue(mlxsw_wq);
+err_alloc_workqueue:
+	mlxsw_linecard_driver_unregister();
+	return err;
+}
+
+static void __exit mlxsw_core_module_exit(void)
+{
+	destroy_workqueue(mlxsw_owq);
+	destroy_workqueue(mlxsw_wq);
+	mlxsw_linecard_driver_unregister();
+}
+
+module_init(mlxsw_core_module_init);
+module_exit(mlxsw_core_module_exit);
+
+MODULE_LICENSE("Dual BSD/GPL");
+MODULE_AUTHOR("Jiri Pirko <jiri@mellanox.com>");
+MODULE_DESCRIPTION("Mellanox switch device core driver");
diff --git a/drivers/net/ethernet/mellanox/mlxsw/core.h b/drivers/net/ethernet/mellanox/mlxsw/core.h
new file mode 100644
index 000000000..ca0c3d2be
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/core.h
@@ -0,0 +1,674 @@
+/* SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0 */
+/* Copyright (c) 2015-2018 Mellanox Technologies. All rights reserved */
+
+#ifndef _MLXSW_CORE_H
+#define _MLXSW_CORE_H
+
+#include <linux/module.h>
+#include <linux/device.h>
+#include <linux/slab.h>
+#include <linux/gfp.h>
+#include <linux/types.h>
+#include <linux/skbuff.h>
+#include <linux/workqueue.h>
+#include <linux/net_namespace.h>
+#include <linux/auxiliary_bus.h>
+#include <net/devlink.h>
+
+#include "trap.h"
+#include "reg.h"
+#include "cmd.h"
+#include "resources.h"
+#include "../mlxfw/mlxfw.h"
+
+enum mlxsw_core_resource_id {
+	MLXSW_CORE_RESOURCE_PORTS = 1,
+	MLXSW_CORE_RESOURCE_MAX,
+};
+
+struct mlxsw_core;
+struct mlxsw_core_port;
+struct mlxsw_driver;
+struct mlxsw_bus;
+struct mlxsw_bus_info;
+struct mlxsw_fw_rev;
+
+unsigned int mlxsw_core_max_ports(const struct mlxsw_core *mlxsw_core);
+
+int mlxsw_core_max_lag(struct mlxsw_core *mlxsw_core, u16 *p_max_lag);
+
+void *mlxsw_core_driver_priv(struct mlxsw_core *mlxsw_core);
+
+struct mlxsw_linecards *mlxsw_core_linecards(struct mlxsw_core *mlxsw_core);
+
+void mlxsw_core_linecards_set(struct mlxsw_core *mlxsw_core,
+			      struct mlxsw_linecards *linecard);
+
+bool
+mlxsw_core_fw_rev_minor_subminor_validate(const struct mlxsw_fw_rev *rev,
+					  const struct mlxsw_fw_rev *req_rev);
+
+int mlxsw_core_driver_register(struct mlxsw_driver *mlxsw_driver);
+void mlxsw_core_driver_unregister(struct mlxsw_driver *mlxsw_driver);
+
+int mlxsw_core_fw_flash(struct mlxsw_core *mlxsw_core,
+			struct mlxfw_dev *mlxfw_dev,
+			const struct firmware *firmware,
+			struct netlink_ext_ack *extack);
+
+int mlxsw_core_bus_device_register(const struct mlxsw_bus_info *mlxsw_bus_info,
+				   const struct mlxsw_bus *mlxsw_bus,
+				   void *bus_priv, bool reload,
+				   struct devlink *devlink,
+				   struct netlink_ext_ack *extack);
+void mlxsw_core_bus_device_unregister(struct mlxsw_core *mlxsw_core, bool reload);
+
+struct mlxsw_tx_info {
+	u16 local_port;
+	bool is_emad;
+};
+
+struct mlxsw_rx_md_info {
+	u32 cookie_index;
+	u32 latency;
+	u32 tx_congestion;
+	union {
+		/* Valid when 'tx_port_valid' is set. */
+		u16 tx_sys_port;
+		u16 tx_lag_id;
+	};
+	u16 tx_lag_port_index; /* Valid when 'tx_port_is_lag' is set. */
+	u8 tx_tc;
+	u8 latency_valid:1,
+	   tx_congestion_valid:1,
+	   tx_tc_valid:1,
+	   tx_port_valid:1,
+	   tx_port_is_lag:1,
+	   unused:3;
+};
+
+bool mlxsw_core_skb_transmit_busy(struct mlxsw_core *mlxsw_core,
+				  const struct mlxsw_tx_info *tx_info);
+int mlxsw_core_skb_transmit(struct mlxsw_core *mlxsw_core, struct sk_buff *skb,
+			    const struct mlxsw_tx_info *tx_info);
+void mlxsw_core_ptp_transmitted(struct mlxsw_core *mlxsw_core,
+				struct sk_buff *skb, u16 local_port);
+
+struct mlxsw_rx_listener {
+	void (*func)(struct sk_buff *skb, u16 local_port, void *priv);
+	u16 local_port;
+	u8 mirror_reason;
+	u16 trap_id;
+};
+
+struct mlxsw_event_listener {
+	void (*func)(const struct mlxsw_reg_info *reg,
+		     char *payload, void *priv);
+	enum mlxsw_event_trap_id trap_id;
+};
+
+struct mlxsw_listener {
+	u16 trap_id;
+	union {
+		struct mlxsw_rx_listener rx_listener;
+		struct mlxsw_event_listener event_listener;
+	};
+	enum mlxsw_reg_hpkt_action en_action; /* Action when enabled */
+	enum mlxsw_reg_hpkt_action dis_action; /* Action when disabled */
+	u8 en_trap_group; /* Trap group when enabled */
+	u8 dis_trap_group; /* Trap group when disabled */
+	u8 is_ctrl:1, /* should go via control buffer or not */
+	   is_event:1,
+	   enabled_on_register:1; /* Trap should be enabled when listener
+				   * is registered.
+				   */
+};
+
+#define __MLXSW_RXL(_func, _trap_id, _en_action, _is_ctrl, _en_trap_group,	\
+		    _dis_action, _enabled_on_register, _dis_trap_group,		\
+		    _mirror_reason)						\
+	{									\
+		.trap_id = MLXSW_TRAP_ID_##_trap_id,				\
+		.rx_listener =							\
+		{								\
+			.func = _func,						\
+			.local_port = MLXSW_PORT_DONT_CARE,			\
+			.mirror_reason = _mirror_reason,			\
+			.trap_id = MLXSW_TRAP_ID_##_trap_id,			\
+		},								\
+		.en_action = MLXSW_REG_HPKT_ACTION_##_en_action,		\
+		.dis_action = MLXSW_REG_HPKT_ACTION_##_dis_action,		\
+		.en_trap_group = MLXSW_REG_HTGT_TRAP_GROUP_##_en_trap_group,	\
+		.dis_trap_group = MLXSW_REG_HTGT_TRAP_GROUP_##_dis_trap_group,	\
+		.is_ctrl = _is_ctrl,						\
+		.enabled_on_register = _enabled_on_register,			\
+	}
+
+#define MLXSW_RXL(_func, _trap_id, _en_action, _is_ctrl, _trap_group,		\
+		  _dis_action)							\
+	__MLXSW_RXL(_func, _trap_id, _en_action, _is_ctrl, _trap_group,		\
+		    _dis_action, true, _trap_group, 0)
+
+#define MLXSW_RXL_DIS(_func, _trap_id, _en_action, _is_ctrl, _en_trap_group,	\
+		      _dis_action, _dis_trap_group)				\
+	__MLXSW_RXL(_func, _trap_id, _en_action, _is_ctrl, _en_trap_group,	\
+		    _dis_action, false, _dis_trap_group, 0)
+
+#define MLXSW_RXL_MIRROR(_func, _session_id, _trap_group, _mirror_reason)	\
+	__MLXSW_RXL(_func, MIRROR_SESSION##_session_id,	TRAP_TO_CPU, false,	\
+		    _trap_group, TRAP_TO_CPU, true, _trap_group,		\
+		    _mirror_reason)
+
+#define MLXSW_EVENTL(_func, _trap_id, _trap_group)				\
+	{									\
+		.trap_id = MLXSW_TRAP_ID_##_trap_id,				\
+		.event_listener =						\
+		{								\
+			.func = _func,						\
+			.trap_id = MLXSW_TRAP_ID_##_trap_id,			\
+		},								\
+		.en_action = MLXSW_REG_HPKT_ACTION_TRAP_TO_CPU,			\
+		.en_trap_group = MLXSW_REG_HTGT_TRAP_GROUP_##_trap_group,	\
+		.is_event = true,						\
+		.enabled_on_register = true,					\
+	}
+
+#define MLXSW_CORE_EVENTL(_func, _trap_id)		\
+	MLXSW_EVENTL(_func, _trap_id, CORE_EVENT)
+
+int mlxsw_core_rx_listener_register(struct mlxsw_core *mlxsw_core,
+				    const struct mlxsw_rx_listener *rxl,
+				    void *priv, bool enabled);
+void mlxsw_core_rx_listener_unregister(struct mlxsw_core *mlxsw_core,
+				       const struct mlxsw_rx_listener *rxl);
+
+int mlxsw_core_event_listener_register(struct mlxsw_core *mlxsw_core,
+				       const struct mlxsw_event_listener *el,
+				       void *priv);
+void mlxsw_core_event_listener_unregister(struct mlxsw_core *mlxsw_core,
+					  const struct mlxsw_event_listener *el);
+
+int mlxsw_core_trap_register(struct mlxsw_core *mlxsw_core,
+			     const struct mlxsw_listener *listener,
+			     void *priv);
+void mlxsw_core_trap_unregister(struct mlxsw_core *mlxsw_core,
+				const struct mlxsw_listener *listener,
+				void *priv);
+int mlxsw_core_traps_register(struct mlxsw_core *mlxsw_core,
+			      const struct mlxsw_listener *listeners,
+			      size_t listeners_count, void *priv);
+void mlxsw_core_traps_unregister(struct mlxsw_core *mlxsw_core,
+				 const struct mlxsw_listener *listeners,
+				 size_t listeners_count, void *priv);
+int mlxsw_core_trap_state_set(struct mlxsw_core *mlxsw_core,
+			      const struct mlxsw_listener *listener,
+			      bool enabled);
+
+typedef void mlxsw_reg_trans_cb_t(struct mlxsw_core *mlxsw_core, char *payload,
+				  size_t payload_len, unsigned long cb_priv);
+
+int mlxsw_reg_trans_query(struct mlxsw_core *mlxsw_core,
+			  const struct mlxsw_reg_info *reg, char *payload,
+			  struct list_head *bulk_list,
+			  mlxsw_reg_trans_cb_t *cb, unsigned long cb_priv);
+int mlxsw_reg_trans_write(struct mlxsw_core *mlxsw_core,
+			  const struct mlxsw_reg_info *reg, char *payload,
+			  struct list_head *bulk_list,
+			  mlxsw_reg_trans_cb_t *cb, unsigned long cb_priv);
+int mlxsw_reg_trans_bulk_wait(struct list_head *bulk_list);
+
+typedef void mlxsw_irq_event_cb_t(struct mlxsw_core *mlxsw_core);
+
+int mlxsw_core_irq_event_handler_register(struct mlxsw_core *mlxsw_core,
+					  mlxsw_irq_event_cb_t cb);
+void mlxsw_core_irq_event_handler_unregister(struct mlxsw_core *mlxsw_core,
+					     mlxsw_irq_event_cb_t cb);
+void mlxsw_core_irq_event_handlers_call(struct mlxsw_core *mlxsw_core);
+
+int mlxsw_reg_query(struct mlxsw_core *mlxsw_core,
+		    const struct mlxsw_reg_info *reg, char *payload);
+int mlxsw_reg_write(struct mlxsw_core *mlxsw_core,
+		    const struct mlxsw_reg_info *reg, char *payload);
+
+struct mlxsw_rx_info {
+	bool is_lag;
+	union {
+		u16 sys_port;
+		u16 lag_id;
+	} u;
+	u16 lag_port_index;
+	u8 mirror_reason;
+	int trap_id;
+};
+
+void mlxsw_core_skb_receive(struct mlxsw_core *mlxsw_core, struct sk_buff *skb,
+			    struct mlxsw_rx_info *rx_info);
+
+void mlxsw_core_lag_mapping_set(struct mlxsw_core *mlxsw_core,
+				u16 lag_id, u8 port_index, u16 local_port);
+u16 mlxsw_core_lag_mapping_get(struct mlxsw_core *mlxsw_core,
+			       u16 lag_id, u8 port_index);
+void mlxsw_core_lag_mapping_clear(struct mlxsw_core *mlxsw_core,
+				  u16 lag_id, u16 local_port);
+
+void *mlxsw_core_port_driver_priv(struct mlxsw_core_port *mlxsw_core_port);
+int mlxsw_core_port_init(struct mlxsw_core *mlxsw_core, u16 local_port,
+			 u8 slot_index, u32 port_number, bool split,
+			 u32 split_port_subnumber,
+			 bool splittable, u32 lanes,
+			 const unsigned char *switch_id,
+			 unsigned char switch_id_len);
+void mlxsw_core_port_fini(struct mlxsw_core *mlxsw_core, u16 local_port);
+int mlxsw_core_cpu_port_init(struct mlxsw_core *mlxsw_core,
+			     void *port_driver_priv,
+			     const unsigned char *switch_id,
+			     unsigned char switch_id_len);
+void mlxsw_core_cpu_port_fini(struct mlxsw_core *mlxsw_core);
+void mlxsw_core_port_eth_set(struct mlxsw_core *mlxsw_core, u16 local_port,
+			     void *port_driver_priv, struct net_device *dev);
+void mlxsw_core_port_clear(struct mlxsw_core *mlxsw_core, u16 local_port,
+			   void *port_driver_priv);
+struct devlink_port *
+mlxsw_core_port_devlink_port_get(struct mlxsw_core *mlxsw_core,
+				 u16 local_port);
+struct mlxsw_linecard *
+mlxsw_core_port_linecard_get(struct mlxsw_core *mlxsw_core,
+			     u16 local_port);
+void mlxsw_core_ports_remove_selected(struct mlxsw_core *mlxsw_core,
+				      bool (*selector)(void *priv,
+						       u16 local_port),
+				      void *priv);
+struct mlxsw_env *mlxsw_core_env(const struct mlxsw_core *mlxsw_core);
+
+int mlxsw_core_schedule_dw(struct delayed_work *dwork, unsigned long delay);
+bool mlxsw_core_schedule_work(struct work_struct *work);
+void mlxsw_core_flush_owq(void);
+int mlxsw_core_resources_query(struct mlxsw_core *mlxsw_core, char *mbox,
+			       struct mlxsw_res *res);
+
+#define MLXSW_CONFIG_PROFILE_SWID_COUNT 8
+
+struct mlxsw_swid_config {
+	u8	used_type:1,
+		used_properties:1;
+	u8	type;
+	u8	properties;
+};
+
+struct mlxsw_config_profile {
+	u16	used_max_vepa_channels:1,
+		used_max_lag:1,
+		used_max_mid:1,
+		used_max_pgt:1,
+		used_max_system_port:1,
+		used_max_vlan_groups:1,
+		used_max_regions:1,
+		used_flood_tables:1,
+		used_flood_mode:1,
+		used_max_ib_mc:1,
+		used_max_pkey:1,
+		used_ar_sec:1,
+		used_adaptive_routing_group_cap:1,
+		used_ubridge:1,
+		used_kvd_sizes:1,
+		used_cqe_time_stamp_type:1;
+	u8	max_vepa_channels;
+	u16	max_lag;
+	u16	max_mid;
+	u16	max_pgt;
+	u16	max_system_port;
+	u16	max_vlan_groups;
+	u16	max_regions;
+	u8	max_flood_tables;
+	u8	max_vid_flood_tables;
+	u8	flood_mode;
+	u8	max_fid_offset_flood_tables;
+	u16	fid_offset_flood_table_size;
+	u8	max_fid_flood_tables;
+	u16	fid_flood_table_size;
+	u16	max_ib_mc;
+	u16	max_pkey;
+	u8	ar_sec;
+	u16	adaptive_routing_group_cap;
+	u8	arn;
+	u8	ubridge;
+	u32	kvd_linear_size;
+	u8	kvd_hash_single_parts;
+	u8	kvd_hash_double_parts;
+	u8	cqe_time_stamp_type;
+	struct mlxsw_swid_config swid_config[MLXSW_CONFIG_PROFILE_SWID_COUNT];
+};
+
+struct mlxsw_driver {
+	struct list_head list;
+	const char *kind;
+	size_t priv_size;
+	const struct mlxsw_fw_rev *fw_req_rev;
+	const char *fw_filename;
+	int (*init)(struct mlxsw_core *mlxsw_core,
+		    const struct mlxsw_bus_info *mlxsw_bus_info,
+		    struct netlink_ext_ack *extack);
+	void (*fini)(struct mlxsw_core *mlxsw_core);
+	int (*port_split)(struct mlxsw_core *mlxsw_core, u16 local_port,
+			  unsigned int count, struct netlink_ext_ack *extack);
+	int (*port_unsplit)(struct mlxsw_core *mlxsw_core, u16 local_port,
+			    struct netlink_ext_ack *extack);
+	void (*ports_remove_selected)(struct mlxsw_core *mlxsw_core,
+				      bool (*selector)(void *priv,
+						       u16 local_port),
+				      void *priv);
+	int (*sb_pool_get)(struct mlxsw_core *mlxsw_core,
+			   unsigned int sb_index, u16 pool_index,
+			   struct devlink_sb_pool_info *pool_info);
+	int (*sb_pool_set)(struct mlxsw_core *mlxsw_core,
+			   unsigned int sb_index, u16 pool_index, u32 size,
+			   enum devlink_sb_threshold_type threshold_type,
+			   struct netlink_ext_ack *extack);
+	int (*sb_port_pool_get)(struct mlxsw_core_port *mlxsw_core_port,
+				unsigned int sb_index, u16 pool_index,
+				u32 *p_threshold);
+	int (*sb_port_pool_set)(struct mlxsw_core_port *mlxsw_core_port,
+				unsigned int sb_index, u16 pool_index,
+				u32 threshold, struct netlink_ext_ack *extack);
+	int (*sb_tc_pool_bind_get)(struct mlxsw_core_port *mlxsw_core_port,
+				   unsigned int sb_index, u16 tc_index,
+				   enum devlink_sb_pool_type pool_type,
+				   u16 *p_pool_index, u32 *p_threshold);
+	int (*sb_tc_pool_bind_set)(struct mlxsw_core_port *mlxsw_core_port,
+				   unsigned int sb_index, u16 tc_index,
+				   enum devlink_sb_pool_type pool_type,
+				   u16 pool_index, u32 threshold,
+				   struct netlink_ext_ack *extack);
+	int (*sb_occ_snapshot)(struct mlxsw_core *mlxsw_core,
+			       unsigned int sb_index);
+	int (*sb_occ_max_clear)(struct mlxsw_core *mlxsw_core,
+				unsigned int sb_index);
+	int (*sb_occ_port_pool_get)(struct mlxsw_core_port *mlxsw_core_port,
+				    unsigned int sb_index, u16 pool_index,
+				    u32 *p_cur, u32 *p_max);
+	int (*sb_occ_tc_port_bind_get)(struct mlxsw_core_port *mlxsw_core_port,
+				       unsigned int sb_index, u16 tc_index,
+				       enum devlink_sb_pool_type pool_type,
+				       u32 *p_cur, u32 *p_max);
+	int (*trap_init)(struct mlxsw_core *mlxsw_core,
+			 const struct devlink_trap *trap, void *trap_ctx);
+	void (*trap_fini)(struct mlxsw_core *mlxsw_core,
+			  const struct devlink_trap *trap, void *trap_ctx);
+	int (*trap_action_set)(struct mlxsw_core *mlxsw_core,
+			       const struct devlink_trap *trap,
+			       enum devlink_trap_action action,
+			       struct netlink_ext_ack *extack);
+	int (*trap_group_init)(struct mlxsw_core *mlxsw_core,
+			       const struct devlink_trap_group *group);
+	int (*trap_group_set)(struct mlxsw_core *mlxsw_core,
+			      const struct devlink_trap_group *group,
+			      const struct devlink_trap_policer *policer,
+			      struct netlink_ext_ack *extack);
+	int (*trap_policer_init)(struct mlxsw_core *mlxsw_core,
+				 const struct devlink_trap_policer *policer);
+	void (*trap_policer_fini)(struct mlxsw_core *mlxsw_core,
+				  const struct devlink_trap_policer *policer);
+	int (*trap_policer_set)(struct mlxsw_core *mlxsw_core,
+				const struct devlink_trap_policer *policer,
+				u64 rate, u64 burst,
+				struct netlink_ext_ack *extack);
+	int (*trap_policer_counter_get)(struct mlxsw_core *mlxsw_core,
+					const struct devlink_trap_policer *policer,
+					u64 *p_drops);
+	void (*txhdr_construct)(struct sk_buff *skb,
+				const struct mlxsw_tx_info *tx_info);
+	int (*resources_register)(struct mlxsw_core *mlxsw_core);
+	int (*kvd_sizes_get)(struct mlxsw_core *mlxsw_core,
+			     const struct mlxsw_config_profile *profile,
+			     u64 *p_single_size, u64 *p_double_size,
+			     u64 *p_linear_size);
+	int (*params_register)(struct mlxsw_core *mlxsw_core);
+	void (*params_unregister)(struct mlxsw_core *mlxsw_core);
+
+	/* Notify a driver that a timestamped packet was transmitted. Driver
+	 * is responsible for freeing the passed-in SKB.
+	 */
+	void (*ptp_transmitted)(struct mlxsw_core *mlxsw_core,
+				struct sk_buff *skb, u16 local_port);
+
+	u8 txhdr_len;
+	const struct mlxsw_config_profile *profile;
+	bool sdq_supports_cqe_v2;
+};
+
+int mlxsw_core_kvd_sizes_get(struct mlxsw_core *mlxsw_core,
+			     const struct mlxsw_config_profile *profile,
+			     u64 *p_single_size, u64 *p_double_size,
+			     u64 *p_linear_size);
+
+u32 mlxsw_core_read_frc_h(struct mlxsw_core *mlxsw_core);
+u32 mlxsw_core_read_frc_l(struct mlxsw_core *mlxsw_core);
+
+u32 mlxsw_core_read_utc_sec(struct mlxsw_core *mlxsw_core);
+u32 mlxsw_core_read_utc_nsec(struct mlxsw_core *mlxsw_core);
+
+bool mlxsw_core_sdq_supports_cqe_v2(struct mlxsw_core *mlxsw_core);
+
+void mlxsw_core_emad_string_tlv_enable(struct mlxsw_core *mlxsw_core);
+
+bool mlxsw_core_res_valid(struct mlxsw_core *mlxsw_core,
+			  enum mlxsw_res_id res_id);
+
+#define MLXSW_CORE_RES_VALID(mlxsw_core, short_res_id)			\
+	mlxsw_core_res_valid(mlxsw_core, MLXSW_RES_ID_##short_res_id)
+
+u64 mlxsw_core_res_get(struct mlxsw_core *mlxsw_core,
+		       enum mlxsw_res_id res_id);
+
+#define MLXSW_CORE_RES_GET(mlxsw_core, short_res_id)			\
+	mlxsw_core_res_get(mlxsw_core, MLXSW_RES_ID_##short_res_id)
+
+static inline struct net *mlxsw_core_net(struct mlxsw_core *mlxsw_core)
+{
+	return devlink_net(priv_to_devlink(mlxsw_core));
+}
+
+#define MLXSW_BUS_F_TXRX	BIT(0)
+#define MLXSW_BUS_F_RESET	BIT(1)
+
+struct mlxsw_bus {
+	const char *kind;
+	int (*init)(void *bus_priv, struct mlxsw_core *mlxsw_core,
+		    const struct mlxsw_config_profile *profile,
+		    struct mlxsw_res *res);
+	void (*fini)(void *bus_priv);
+	bool (*skb_transmit_busy)(void *bus_priv,
+				  const struct mlxsw_tx_info *tx_info);
+	int (*skb_transmit)(void *bus_priv, struct sk_buff *skb,
+			    const struct mlxsw_tx_info *tx_info);
+	int (*cmd_exec)(void *bus_priv, u16 opcode, u8 opcode_mod,
+			u32 in_mod, bool out_mbox_direct,
+			char *in_mbox, size_t in_mbox_size,
+			char *out_mbox, size_t out_mbox_size,
+			u8 *p_status);
+	u32 (*read_frc_h)(void *bus_priv);
+	u32 (*read_frc_l)(void *bus_priv);
+	u32 (*read_utc_sec)(void *bus_priv);
+	u32 (*read_utc_nsec)(void *bus_priv);
+	u8 features;
+};
+
+struct mlxsw_fw_rev {
+	u16 major;
+	u16 minor;
+	u16 subminor;
+	u16 can_reset_minor;
+};
+
+struct mlxsw_bus_info {
+	const char *device_kind;
+	const char *device_name;
+	struct device *dev;
+	struct mlxsw_fw_rev fw_rev;
+	u8 vsd[MLXSW_CMD_BOARDINFO_VSD_LEN];
+	u8 psid[MLXSW_CMD_BOARDINFO_PSID_LEN];
+	u8 low_frequency:1,
+	   read_clock_capable:1;
+};
+
+struct mlxsw_hwmon;
+
+#ifdef CONFIG_MLXSW_CORE_HWMON
+
+int mlxsw_hwmon_init(struct mlxsw_core *mlxsw_core,
+		     const struct mlxsw_bus_info *mlxsw_bus_info,
+		     struct mlxsw_hwmon **p_hwmon);
+void mlxsw_hwmon_fini(struct mlxsw_hwmon *mlxsw_hwmon);
+
+#else
+
+static inline int mlxsw_hwmon_init(struct mlxsw_core *mlxsw_core,
+				   const struct mlxsw_bus_info *mlxsw_bus_info,
+				   struct mlxsw_hwmon **p_hwmon)
+{
+	return 0;
+}
+
+static inline void mlxsw_hwmon_fini(struct mlxsw_hwmon *mlxsw_hwmon)
+{
+}
+
+#endif
+
+struct mlxsw_thermal;
+
+#ifdef CONFIG_MLXSW_CORE_THERMAL
+
+int mlxsw_thermal_init(struct mlxsw_core *mlxsw_core,
+		       const struct mlxsw_bus_info *mlxsw_bus_info,
+		       struct mlxsw_thermal **p_thermal);
+void mlxsw_thermal_fini(struct mlxsw_thermal *thermal);
+
+#else
+
+static inline int mlxsw_thermal_init(struct mlxsw_core *mlxsw_core,
+				     const struct mlxsw_bus_info *mlxsw_bus_info,
+				     struct mlxsw_thermal **p_thermal)
+{
+	return 0;
+}
+
+static inline void mlxsw_thermal_fini(struct mlxsw_thermal *thermal)
+{
+}
+
+#endif
+
+enum mlxsw_devlink_param_id {
+	MLXSW_DEVLINK_PARAM_ID_BASE = DEVLINK_PARAM_GENERIC_ID_MAX,
+	MLXSW_DEVLINK_PARAM_ID_ACL_REGION_REHASH_INTERVAL,
+};
+
+struct mlxsw_cqe_ts {
+	u8 sec;
+	u32 nsec;
+};
+
+struct mlxsw_skb_cb {
+	union {
+		struct mlxsw_tx_info tx_info;
+		struct mlxsw_rx_md_info rx_md_info;
+	};
+	struct mlxsw_cqe_ts cqe_ts;
+};
+
+static inline struct mlxsw_skb_cb *mlxsw_skb_cb(struct sk_buff *skb)
+{
+	BUILD_BUG_ON(sizeof(mlxsw_skb_cb) > sizeof(skb->cb));
+	return (struct mlxsw_skb_cb *) skb->cb;
+}
+
+struct mlxsw_linecards;
+
+enum mlxsw_linecard_status_event_type {
+	MLXSW_LINECARD_STATUS_EVENT_TYPE_PROVISION,
+	MLXSW_LINECARD_STATUS_EVENT_TYPE_UNPROVISION,
+};
+
+struct mlxsw_linecard_bdev;
+
+struct mlxsw_linecard_device_info {
+	u16 fw_major;
+	u16 fw_minor;
+	u16 fw_sub_minor;
+	char psid[MLXSW_REG_MGIR_FW_INFO_PSID_SIZE];
+};
+
+struct mlxsw_linecard {
+	u8 slot_index;
+	struct mlxsw_linecards *linecards;
+	struct devlink_linecard *devlink_linecard;
+	struct mutex lock; /* Locks accesses to the linecard structure */
+	char name[MLXSW_REG_MDDQ_SLOT_ASCII_NAME_LEN];
+	char mbct_pl[MLXSW_REG_MBCT_LEN]; /* Too big for stack */
+	enum mlxsw_linecard_status_event_type status_event_type_to;
+	struct delayed_work status_event_to_dw;
+	u8 provisioned:1,
+	   ready:1,
+	   active:1;
+	u16 hw_revision;
+	u16 ini_version;
+	struct mlxsw_linecard_bdev *bdev;
+	struct {
+		struct mlxsw_linecard_device_info info;
+		u8 index;
+	} device;
+};
+
+struct mlxsw_linecard_types_info;
+
+struct mlxsw_linecards {
+	struct mlxsw_core *mlxsw_core;
+	const struct mlxsw_bus_info *bus_info;
+	u8 count;
+	struct mlxsw_linecard_types_info *types_info;
+	struct list_head event_ops_list;
+	struct mutex event_ops_list_lock; /* Locks accesses to event ops list */
+	struct mlxsw_linecard linecards[];
+};
+
+static inline struct mlxsw_linecard *
+mlxsw_linecard_get(struct mlxsw_linecards *linecards, u8 slot_index)
+{
+	return &linecards->linecards[slot_index - 1];
+}
+
+int mlxsw_linecard_devlink_info_get(struct mlxsw_linecard *linecard,
+				    struct devlink_info_req *req,
+				    struct netlink_ext_ack *extack);
+int mlxsw_linecard_flash_update(struct devlink *linecard_devlink,
+				struct mlxsw_linecard *linecard,
+				const struct firmware *firmware,
+				struct netlink_ext_ack *extack);
+
+int mlxsw_linecards_init(struct mlxsw_core *mlxsw_core,
+			 const struct mlxsw_bus_info *bus_info);
+void mlxsw_linecards_fini(struct mlxsw_core *mlxsw_core);
+
+typedef void mlxsw_linecards_event_op_t(struct mlxsw_core *mlxsw_core,
+					u8 slot_index, void *priv);
+
+struct mlxsw_linecards_event_ops {
+	mlxsw_linecards_event_op_t *got_active;
+	mlxsw_linecards_event_op_t *got_inactive;
+};
+
+int mlxsw_linecards_event_ops_register(struct mlxsw_core *mlxsw_core,
+				       struct mlxsw_linecards_event_ops *ops,
+				       void *priv);
+void mlxsw_linecards_event_ops_unregister(struct mlxsw_core *mlxsw_core,
+					  struct mlxsw_linecards_event_ops *ops,
+					  void *priv);
+
+int mlxsw_linecard_bdev_add(struct mlxsw_linecard *linecard);
+void mlxsw_linecard_bdev_del(struct mlxsw_linecard *linecard);
+
+int mlxsw_linecard_driver_register(void);
+void mlxsw_linecard_driver_unregister(void);
+
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlxsw/core_acl_flex_actions.c b/drivers/net/ethernet/mellanox/mlxsw/core_acl_flex_actions.c
new file mode 100644
index 000000000..9dfe71481
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/core_acl_flex_actions.c
@@ -0,0 +1,2218 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2017-2018 Mellanox Technologies. All rights reserved */
+
+#include <linux/kernel.h>
+#include <linux/types.h>
+#include <linux/slab.h>
+#include <linux/errno.h>
+#include <linux/rhashtable.h>
+#include <linux/list.h>
+#include <linux/idr.h>
+#include <linux/refcount.h>
+#include <net/flow_offload.h>
+
+#include "item.h"
+#include "trap.h"
+#include "core_acl_flex_actions.h"
+
+enum mlxsw_afa_set_type {
+	MLXSW_AFA_SET_TYPE_NEXT,
+	MLXSW_AFA_SET_TYPE_GOTO,
+};
+
+/* afa_set_type
+ * Type of the record at the end of the action set.
+ */
+MLXSW_ITEM32(afa, set, type, 0xA0, 28, 4);
+
+/* afa_set_next_action_set_ptr
+ * A pointer to the next action set in the KVD Centralized database.
+ */
+MLXSW_ITEM32(afa, set, next_action_set_ptr, 0xA4, 0, 24);
+
+/* afa_set_goto_g
+ * group - When set, the binding is of an ACL group. When cleared,
+ * the binding is of an ACL.
+ * Must be set to 1 for Spectrum.
+ */
+MLXSW_ITEM32(afa, set, goto_g, 0xA4, 29, 1);
+
+enum mlxsw_afa_set_goto_binding_cmd {
+	/* continue go the next binding point */
+	MLXSW_AFA_SET_GOTO_BINDING_CMD_NONE,
+	/* jump to the next binding point no return */
+	MLXSW_AFA_SET_GOTO_BINDING_CMD_JUMP,
+	/* terminate the acl binding */
+	MLXSW_AFA_SET_GOTO_BINDING_CMD_TERM = 4,
+};
+
+/* afa_set_goto_binding_cmd */
+MLXSW_ITEM32(afa, set, goto_binding_cmd, 0xA4, 24, 3);
+
+/* afa_set_goto_next_binding
+ * ACL/ACL group identifier. If the g bit is set, this field should hold
+ * the acl_group_id, else it should hold the acl_id.
+ */
+MLXSW_ITEM32(afa, set, goto_next_binding, 0xA4, 0, 16);
+
+/* afa_all_action_type
+ * Action Type.
+ */
+MLXSW_ITEM32(afa, all, action_type, 0x00, 24, 6);
+
+struct mlxsw_afa {
+	unsigned int max_acts_per_set;
+	const struct mlxsw_afa_ops *ops;
+	void *ops_priv;
+	struct rhashtable set_ht;
+	struct rhashtable fwd_entry_ht;
+	struct rhashtable cookie_ht;
+	struct rhashtable policer_ht;
+	struct idr cookie_idr;
+	struct list_head policer_list;
+};
+
+#define MLXSW_AFA_SET_LEN 0xA8
+
+struct mlxsw_afa_set_ht_key {
+	char enc_actions[MLXSW_AFA_SET_LEN]; /* Encoded set */
+	bool is_first;
+};
+
+/* Set structure holds one action set record. It contains up to three
+ * actions (depends on size of particular actions). The set is either
+ * put directly to a rule, or it is stored in KVD linear area.
+ * To prevent duplicate entries in KVD linear area, a hashtable is
+ * used to track sets that were previously inserted and may be shared.
+ */
+
+struct mlxsw_afa_set {
+	struct rhash_head ht_node;
+	struct mlxsw_afa_set_ht_key ht_key;
+	u32 kvdl_index;
+	u8 shared:1, /* Inserted in hashtable (doesn't mean that
+		      * kvdl_index is valid).
+		      */
+	   has_trap:1,
+	   has_police:1;
+	unsigned int ref_count;
+	struct mlxsw_afa_set *next; /* Pointer to the next set. */
+	struct mlxsw_afa_set *prev; /* Pointer to the previous set,
+				     * note that set may have multiple
+				     * sets from multiple blocks
+				     * pointing at it. This is only
+				     * usable until commit.
+				     */
+};
+
+static const struct rhashtable_params mlxsw_afa_set_ht_params = {
+	.key_len = sizeof(struct mlxsw_afa_set_ht_key),
+	.key_offset = offsetof(struct mlxsw_afa_set, ht_key),
+	.head_offset = offsetof(struct mlxsw_afa_set, ht_node),
+	.automatic_shrinking = true,
+};
+
+struct mlxsw_afa_fwd_entry_ht_key {
+	u16 local_port;
+};
+
+struct mlxsw_afa_fwd_entry {
+	struct rhash_head ht_node;
+	struct mlxsw_afa_fwd_entry_ht_key ht_key;
+	u32 kvdl_index;
+	unsigned int ref_count;
+};
+
+static const struct rhashtable_params mlxsw_afa_fwd_entry_ht_params = {
+	.key_len = sizeof(struct mlxsw_afa_fwd_entry_ht_key),
+	.key_offset = offsetof(struct mlxsw_afa_fwd_entry, ht_key),
+	.head_offset = offsetof(struct mlxsw_afa_fwd_entry, ht_node),
+	.automatic_shrinking = true,
+};
+
+struct mlxsw_afa_cookie {
+	struct rhash_head ht_node;
+	refcount_t ref_count;
+	struct rcu_head rcu;
+	u32 cookie_index;
+	struct flow_action_cookie fa_cookie;
+};
+
+static u32 mlxsw_afa_cookie_hash(const struct flow_action_cookie *fa_cookie,
+				 u32 seed)
+{
+	return jhash2((u32 *) fa_cookie->cookie,
+		      fa_cookie->cookie_len / sizeof(u32), seed);
+}
+
+static u32 mlxsw_afa_cookie_key_hashfn(const void *data, u32 len, u32 seed)
+{
+	const struct flow_action_cookie *fa_cookie = data;
+
+	return mlxsw_afa_cookie_hash(fa_cookie, seed);
+}
+
+static u32 mlxsw_afa_cookie_obj_hashfn(const void *data, u32 len, u32 seed)
+{
+	const struct mlxsw_afa_cookie *cookie = data;
+
+	return mlxsw_afa_cookie_hash(&cookie->fa_cookie, seed);
+}
+
+static int mlxsw_afa_cookie_obj_cmpfn(struct rhashtable_compare_arg *arg,
+				      const void *obj)
+{
+	const struct flow_action_cookie *fa_cookie = arg->key;
+	const struct mlxsw_afa_cookie *cookie = obj;
+
+	if (cookie->fa_cookie.cookie_len == fa_cookie->cookie_len)
+		return memcmp(cookie->fa_cookie.cookie, fa_cookie->cookie,
+			      fa_cookie->cookie_len);
+	return 1;
+}
+
+static const struct rhashtable_params mlxsw_afa_cookie_ht_params = {
+	.head_offset = offsetof(struct mlxsw_afa_cookie, ht_node),
+	.hashfn	= mlxsw_afa_cookie_key_hashfn,
+	.obj_hashfn = mlxsw_afa_cookie_obj_hashfn,
+	.obj_cmpfn = mlxsw_afa_cookie_obj_cmpfn,
+	.automatic_shrinking = true,
+};
+
+struct mlxsw_afa_policer {
+	struct rhash_head ht_node;
+	struct list_head list; /* Member of policer_list */
+	refcount_t ref_count;
+	u32 fa_index;
+	u16 policer_index;
+};
+
+static const struct rhashtable_params mlxsw_afa_policer_ht_params = {
+	.key_len = sizeof(u32),
+	.key_offset = offsetof(struct mlxsw_afa_policer, fa_index),
+	.head_offset = offsetof(struct mlxsw_afa_policer, ht_node),
+	.automatic_shrinking = true,
+};
+
+struct mlxsw_afa *mlxsw_afa_create(unsigned int max_acts_per_set,
+				   const struct mlxsw_afa_ops *ops,
+				   void *ops_priv)
+{
+	struct mlxsw_afa *mlxsw_afa;
+	int err;
+
+	mlxsw_afa = kzalloc(sizeof(*mlxsw_afa), GFP_KERNEL);
+	if (!mlxsw_afa)
+		return ERR_PTR(-ENOMEM);
+	err = rhashtable_init(&mlxsw_afa->set_ht, &mlxsw_afa_set_ht_params);
+	if (err)
+		goto err_set_rhashtable_init;
+	err = rhashtable_init(&mlxsw_afa->fwd_entry_ht,
+			      &mlxsw_afa_fwd_entry_ht_params);
+	if (err)
+		goto err_fwd_entry_rhashtable_init;
+	err = rhashtable_init(&mlxsw_afa->cookie_ht,
+			      &mlxsw_afa_cookie_ht_params);
+	if (err)
+		goto err_cookie_rhashtable_init;
+	err = rhashtable_init(&mlxsw_afa->policer_ht,
+			      &mlxsw_afa_policer_ht_params);
+	if (err)
+		goto err_policer_rhashtable_init;
+	idr_init(&mlxsw_afa->cookie_idr);
+	INIT_LIST_HEAD(&mlxsw_afa->policer_list);
+	mlxsw_afa->max_acts_per_set = max_acts_per_set;
+	mlxsw_afa->ops = ops;
+	mlxsw_afa->ops_priv = ops_priv;
+	return mlxsw_afa;
+
+err_policer_rhashtable_init:
+	rhashtable_destroy(&mlxsw_afa->cookie_ht);
+err_cookie_rhashtable_init:
+	rhashtable_destroy(&mlxsw_afa->fwd_entry_ht);
+err_fwd_entry_rhashtable_init:
+	rhashtable_destroy(&mlxsw_afa->set_ht);
+err_set_rhashtable_init:
+	kfree(mlxsw_afa);
+	return ERR_PTR(err);
+}
+EXPORT_SYMBOL(mlxsw_afa_create);
+
+void mlxsw_afa_destroy(struct mlxsw_afa *mlxsw_afa)
+{
+	WARN_ON(!list_empty(&mlxsw_afa->policer_list));
+	WARN_ON(!idr_is_empty(&mlxsw_afa->cookie_idr));
+	idr_destroy(&mlxsw_afa->cookie_idr);
+	rhashtable_destroy(&mlxsw_afa->policer_ht);
+	rhashtable_destroy(&mlxsw_afa->cookie_ht);
+	rhashtable_destroy(&mlxsw_afa->fwd_entry_ht);
+	rhashtable_destroy(&mlxsw_afa->set_ht);
+	kfree(mlxsw_afa);
+}
+EXPORT_SYMBOL(mlxsw_afa_destroy);
+
+static void mlxsw_afa_set_goto_set(struct mlxsw_afa_set *set,
+				   enum mlxsw_afa_set_goto_binding_cmd cmd,
+				   u16 group_id)
+{
+	char *actions = set->ht_key.enc_actions;
+
+	mlxsw_afa_set_type_set(actions, MLXSW_AFA_SET_TYPE_GOTO);
+	mlxsw_afa_set_goto_g_set(actions, true);
+	mlxsw_afa_set_goto_binding_cmd_set(actions, cmd);
+	mlxsw_afa_set_goto_next_binding_set(actions, group_id);
+}
+
+static void mlxsw_afa_set_next_set(struct mlxsw_afa_set *set,
+				   u32 next_set_kvdl_index)
+{
+	char *actions = set->ht_key.enc_actions;
+
+	mlxsw_afa_set_type_set(actions, MLXSW_AFA_SET_TYPE_NEXT);
+	mlxsw_afa_set_next_action_set_ptr_set(actions, next_set_kvdl_index);
+}
+
+static struct mlxsw_afa_set *mlxsw_afa_set_create(bool is_first)
+{
+	struct mlxsw_afa_set *set;
+
+	set = kzalloc(sizeof(*set), GFP_KERNEL);
+	if (!set)
+		return NULL;
+	/* Need to initialize the set to pass by default */
+	mlxsw_afa_set_goto_set(set, MLXSW_AFA_SET_GOTO_BINDING_CMD_TERM, 0);
+	set->ht_key.is_first = is_first;
+	set->ref_count = 1;
+	return set;
+}
+
+static void mlxsw_afa_set_destroy(struct mlxsw_afa_set *set)
+{
+	kfree(set);
+}
+
+static int mlxsw_afa_set_share(struct mlxsw_afa *mlxsw_afa,
+			       struct mlxsw_afa_set *set)
+{
+	int err;
+
+	err = rhashtable_insert_fast(&mlxsw_afa->set_ht, &set->ht_node,
+				     mlxsw_afa_set_ht_params);
+	if (err)
+		return err;
+	err = mlxsw_afa->ops->kvdl_set_add(mlxsw_afa->ops_priv,
+					   &set->kvdl_index,
+					   set->ht_key.enc_actions,
+					   set->ht_key.is_first);
+	if (err)
+		goto err_kvdl_set_add;
+	set->shared = true;
+	set->prev = NULL;
+	return 0;
+
+err_kvdl_set_add:
+	rhashtable_remove_fast(&mlxsw_afa->set_ht, &set->ht_node,
+			       mlxsw_afa_set_ht_params);
+	return err;
+}
+
+static void mlxsw_afa_set_unshare(struct mlxsw_afa *mlxsw_afa,
+				  struct mlxsw_afa_set *set)
+{
+	mlxsw_afa->ops->kvdl_set_del(mlxsw_afa->ops_priv,
+				     set->kvdl_index,
+				     set->ht_key.is_first);
+	rhashtable_remove_fast(&mlxsw_afa->set_ht, &set->ht_node,
+			       mlxsw_afa_set_ht_params);
+	set->shared = false;
+}
+
+static void mlxsw_afa_set_put(struct mlxsw_afa *mlxsw_afa,
+			      struct mlxsw_afa_set *set)
+{
+	if (--set->ref_count)
+		return;
+	if (set->shared)
+		mlxsw_afa_set_unshare(mlxsw_afa, set);
+	mlxsw_afa_set_destroy(set);
+}
+
+static struct mlxsw_afa_set *mlxsw_afa_set_get(struct mlxsw_afa *mlxsw_afa,
+					       struct mlxsw_afa_set *orig_set)
+{
+	struct mlxsw_afa_set *set;
+	int err;
+
+	/* There is a hashtable of sets maintained. If a set with the exact
+	 * same encoding exists, we reuse it. Otherwise, the current set
+	 * is shared by making it available to others using the hash table.
+	 */
+	set = rhashtable_lookup_fast(&mlxsw_afa->set_ht, &orig_set->ht_key,
+				     mlxsw_afa_set_ht_params);
+	if (set) {
+		set->ref_count++;
+		mlxsw_afa_set_put(mlxsw_afa, orig_set);
+	} else {
+		set = orig_set;
+		err = mlxsw_afa_set_share(mlxsw_afa, set);
+		if (err)
+			return ERR_PTR(err);
+	}
+	return set;
+}
+
+/* Block structure holds a list of action sets. One action block
+ * represents one chain of actions executed upon match of a rule.
+ */
+
+struct mlxsw_afa_block {
+	struct mlxsw_afa *afa;
+	bool finished;
+	struct mlxsw_afa_set *first_set;
+	struct mlxsw_afa_set *cur_set;
+	unsigned int cur_act_index; /* In current set. */
+	struct list_head resource_list; /* List of resources held by actions
+					 * in this block.
+					 */
+};
+
+struct mlxsw_afa_resource {
+	struct list_head list;
+	void (*destructor)(struct mlxsw_afa_block *block,
+			   struct mlxsw_afa_resource *resource);
+};
+
+static void mlxsw_afa_resource_add(struct mlxsw_afa_block *block,
+				   struct mlxsw_afa_resource *resource)
+{
+	list_add(&resource->list, &block->resource_list);
+}
+
+static void mlxsw_afa_resource_del(struct mlxsw_afa_resource *resource)
+{
+	list_del(&resource->list);
+}
+
+static void mlxsw_afa_resources_destroy(struct mlxsw_afa_block *block)
+{
+	struct mlxsw_afa_resource *resource, *tmp;
+
+	list_for_each_entry_safe(resource, tmp, &block->resource_list, list) {
+		resource->destructor(block, resource);
+	}
+}
+
+struct mlxsw_afa_block *mlxsw_afa_block_create(struct mlxsw_afa *mlxsw_afa)
+{
+	struct mlxsw_afa_block *block;
+
+	block = kzalloc(sizeof(*block), GFP_KERNEL);
+	if (!block)
+		return ERR_PTR(-ENOMEM);
+	INIT_LIST_HEAD(&block->resource_list);
+	block->afa = mlxsw_afa;
+
+	/* At least one action set is always present, so just create it here */
+	block->first_set = mlxsw_afa_set_create(true);
+	if (!block->first_set)
+		goto err_first_set_create;
+
+	/* In case user instructs to have dummy first set, we leave it
+	 * empty here and create another, real, set right away.
+	 */
+	if (mlxsw_afa->ops->dummy_first_set) {
+		block->cur_set = mlxsw_afa_set_create(false);
+		if (!block->cur_set)
+			goto err_second_set_create;
+		block->cur_set->prev = block->first_set;
+		block->first_set->next = block->cur_set;
+	} else {
+		block->cur_set = block->first_set;
+	}
+
+	return block;
+
+err_second_set_create:
+	mlxsw_afa_set_destroy(block->first_set);
+err_first_set_create:
+	kfree(block);
+	return ERR_PTR(-ENOMEM);
+}
+EXPORT_SYMBOL(mlxsw_afa_block_create);
+
+void mlxsw_afa_block_destroy(struct mlxsw_afa_block *block)
+{
+	struct mlxsw_afa_set *set = block->first_set;
+	struct mlxsw_afa_set *next_set;
+
+	do {
+		next_set = set->next;
+		mlxsw_afa_set_put(block->afa, set);
+		set = next_set;
+	} while (set);
+	mlxsw_afa_resources_destroy(block);
+	kfree(block);
+}
+EXPORT_SYMBOL(mlxsw_afa_block_destroy);
+
+int mlxsw_afa_block_commit(struct mlxsw_afa_block *block)
+{
+	struct mlxsw_afa_set *set = block->cur_set;
+	struct mlxsw_afa_set *prev_set;
+
+	block->cur_set = NULL;
+	block->finished = true;
+
+	/* Go over all linked sets starting from last
+	 * and try to find existing set in the hash table.
+	 * In case it is not there, assign a KVD linear index
+	 * and insert it.
+	 */
+	do {
+		prev_set = set->prev;
+		set = mlxsw_afa_set_get(block->afa, set);
+		if (IS_ERR(set))
+			/* No rollback is needed since the chain is
+			 * in consistent state and mlxsw_afa_block_destroy
+			 * will take care of putting it away.
+			 */
+			return PTR_ERR(set);
+		if (prev_set) {
+			prev_set->next = set;
+			mlxsw_afa_set_next_set(prev_set, set->kvdl_index);
+			set = prev_set;
+		}
+	} while (prev_set);
+
+	block->first_set = set;
+	return 0;
+}
+EXPORT_SYMBOL(mlxsw_afa_block_commit);
+
+char *mlxsw_afa_block_first_set(struct mlxsw_afa_block *block)
+{
+	return block->first_set->ht_key.enc_actions;
+}
+EXPORT_SYMBOL(mlxsw_afa_block_first_set);
+
+char *mlxsw_afa_block_cur_set(struct mlxsw_afa_block *block)
+{
+	return block->cur_set->ht_key.enc_actions;
+}
+EXPORT_SYMBOL(mlxsw_afa_block_cur_set);
+
+u32 mlxsw_afa_block_first_kvdl_index(struct mlxsw_afa_block *block)
+{
+	/* First set is never in KVD linear. So the first set
+	 * with valid KVD linear index is always the second one.
+	 */
+	if (WARN_ON(!block->first_set->next))
+		return 0;
+	return block->first_set->next->kvdl_index;
+}
+EXPORT_SYMBOL(mlxsw_afa_block_first_kvdl_index);
+
+int mlxsw_afa_block_activity_get(struct mlxsw_afa_block *block, bool *activity)
+{
+	u32 kvdl_index = mlxsw_afa_block_first_kvdl_index(block);
+
+	return block->afa->ops->kvdl_set_activity_get(block->afa->ops_priv,
+						      kvdl_index, activity);
+}
+EXPORT_SYMBOL(mlxsw_afa_block_activity_get);
+
+int mlxsw_afa_block_continue(struct mlxsw_afa_block *block)
+{
+	if (block->finished)
+		return -EINVAL;
+	mlxsw_afa_set_goto_set(block->cur_set,
+			       MLXSW_AFA_SET_GOTO_BINDING_CMD_NONE, 0);
+	block->finished = true;
+	return 0;
+}
+EXPORT_SYMBOL(mlxsw_afa_block_continue);
+
+int mlxsw_afa_block_jump(struct mlxsw_afa_block *block, u16 group_id)
+{
+	if (block->finished)
+		return -EINVAL;
+	mlxsw_afa_set_goto_set(block->cur_set,
+			       MLXSW_AFA_SET_GOTO_BINDING_CMD_JUMP, group_id);
+	block->finished = true;
+	return 0;
+}
+EXPORT_SYMBOL(mlxsw_afa_block_jump);
+
+int mlxsw_afa_block_terminate(struct mlxsw_afa_block *block)
+{
+	if (block->finished)
+		return -EINVAL;
+	mlxsw_afa_set_goto_set(block->cur_set,
+			       MLXSW_AFA_SET_GOTO_BINDING_CMD_TERM, 0);
+	block->finished = true;
+	return 0;
+}
+EXPORT_SYMBOL(mlxsw_afa_block_terminate);
+
+static struct mlxsw_afa_fwd_entry *
+mlxsw_afa_fwd_entry_create(struct mlxsw_afa *mlxsw_afa, u16 local_port)
+{
+	struct mlxsw_afa_fwd_entry *fwd_entry;
+	int err;
+
+	fwd_entry = kzalloc(sizeof(*fwd_entry), GFP_KERNEL);
+	if (!fwd_entry)
+		return ERR_PTR(-ENOMEM);
+	fwd_entry->ht_key.local_port = local_port;
+	fwd_entry->ref_count = 1;
+
+	err = rhashtable_insert_fast(&mlxsw_afa->fwd_entry_ht,
+				     &fwd_entry->ht_node,
+				     mlxsw_afa_fwd_entry_ht_params);
+	if (err)
+		goto err_rhashtable_insert;
+
+	err = mlxsw_afa->ops->kvdl_fwd_entry_add(mlxsw_afa->ops_priv,
+						 &fwd_entry->kvdl_index,
+						 local_port);
+	if (err)
+		goto err_kvdl_fwd_entry_add;
+	return fwd_entry;
+
+err_kvdl_fwd_entry_add:
+	rhashtable_remove_fast(&mlxsw_afa->fwd_entry_ht, &fwd_entry->ht_node,
+			       mlxsw_afa_fwd_entry_ht_params);
+err_rhashtable_insert:
+	kfree(fwd_entry);
+	return ERR_PTR(err);
+}
+
+static void mlxsw_afa_fwd_entry_destroy(struct mlxsw_afa *mlxsw_afa,
+					struct mlxsw_afa_fwd_entry *fwd_entry)
+{
+	mlxsw_afa->ops->kvdl_fwd_entry_del(mlxsw_afa->ops_priv,
+					   fwd_entry->kvdl_index);
+	rhashtable_remove_fast(&mlxsw_afa->fwd_entry_ht, &fwd_entry->ht_node,
+			       mlxsw_afa_fwd_entry_ht_params);
+	kfree(fwd_entry);
+}
+
+static struct mlxsw_afa_fwd_entry *
+mlxsw_afa_fwd_entry_get(struct mlxsw_afa *mlxsw_afa, u16 local_port)
+{
+	struct mlxsw_afa_fwd_entry_ht_key ht_key = {0};
+	struct mlxsw_afa_fwd_entry *fwd_entry;
+
+	ht_key.local_port = local_port;
+	fwd_entry = rhashtable_lookup_fast(&mlxsw_afa->fwd_entry_ht, &ht_key,
+					   mlxsw_afa_fwd_entry_ht_params);
+	if (fwd_entry) {
+		fwd_entry->ref_count++;
+		return fwd_entry;
+	}
+	return mlxsw_afa_fwd_entry_create(mlxsw_afa, local_port);
+}
+
+static void mlxsw_afa_fwd_entry_put(struct mlxsw_afa *mlxsw_afa,
+				    struct mlxsw_afa_fwd_entry *fwd_entry)
+{
+	if (--fwd_entry->ref_count)
+		return;
+	mlxsw_afa_fwd_entry_destroy(mlxsw_afa, fwd_entry);
+}
+
+struct mlxsw_afa_fwd_entry_ref {
+	struct mlxsw_afa_resource resource;
+	struct mlxsw_afa_fwd_entry *fwd_entry;
+};
+
+static void
+mlxsw_afa_fwd_entry_ref_destroy(struct mlxsw_afa_block *block,
+				struct mlxsw_afa_fwd_entry_ref *fwd_entry_ref)
+{
+	mlxsw_afa_resource_del(&fwd_entry_ref->resource);
+	mlxsw_afa_fwd_entry_put(block->afa, fwd_entry_ref->fwd_entry);
+	kfree(fwd_entry_ref);
+}
+
+static void
+mlxsw_afa_fwd_entry_ref_destructor(struct mlxsw_afa_block *block,
+				   struct mlxsw_afa_resource *resource)
+{
+	struct mlxsw_afa_fwd_entry_ref *fwd_entry_ref;
+
+	fwd_entry_ref = container_of(resource, struct mlxsw_afa_fwd_entry_ref,
+				     resource);
+	mlxsw_afa_fwd_entry_ref_destroy(block, fwd_entry_ref);
+}
+
+static struct mlxsw_afa_fwd_entry_ref *
+mlxsw_afa_fwd_entry_ref_create(struct mlxsw_afa_block *block, u16 local_port)
+{
+	struct mlxsw_afa_fwd_entry_ref *fwd_entry_ref;
+	struct mlxsw_afa_fwd_entry *fwd_entry;
+	int err;
+
+	fwd_entry_ref = kzalloc(sizeof(*fwd_entry_ref), GFP_KERNEL);
+	if (!fwd_entry_ref)
+		return ERR_PTR(-ENOMEM);
+	fwd_entry = mlxsw_afa_fwd_entry_get(block->afa, local_port);
+	if (IS_ERR(fwd_entry)) {
+		err = PTR_ERR(fwd_entry);
+		goto err_fwd_entry_get;
+	}
+	fwd_entry_ref->fwd_entry = fwd_entry;
+	fwd_entry_ref->resource.destructor = mlxsw_afa_fwd_entry_ref_destructor;
+	mlxsw_afa_resource_add(block, &fwd_entry_ref->resource);
+	return fwd_entry_ref;
+
+err_fwd_entry_get:
+	kfree(fwd_entry_ref);
+	return ERR_PTR(err);
+}
+
+struct mlxsw_afa_counter {
+	struct mlxsw_afa_resource resource;
+	u32 counter_index;
+};
+
+static void
+mlxsw_afa_counter_destroy(struct mlxsw_afa_block *block,
+			  struct mlxsw_afa_counter *counter)
+{
+	mlxsw_afa_resource_del(&counter->resource);
+	block->afa->ops->counter_index_put(block->afa->ops_priv,
+					   counter->counter_index);
+	kfree(counter);
+}
+
+static void
+mlxsw_afa_counter_destructor(struct mlxsw_afa_block *block,
+			     struct mlxsw_afa_resource *resource)
+{
+	struct mlxsw_afa_counter *counter;
+
+	counter = container_of(resource, struct mlxsw_afa_counter, resource);
+	mlxsw_afa_counter_destroy(block, counter);
+}
+
+static struct mlxsw_afa_counter *
+mlxsw_afa_counter_create(struct mlxsw_afa_block *block)
+{
+	struct mlxsw_afa_counter *counter;
+	int err;
+
+	counter = kzalloc(sizeof(*counter), GFP_KERNEL);
+	if (!counter)
+		return ERR_PTR(-ENOMEM);
+
+	err = block->afa->ops->counter_index_get(block->afa->ops_priv,
+						 &counter->counter_index);
+	if (err)
+		goto err_counter_index_get;
+	counter->resource.destructor = mlxsw_afa_counter_destructor;
+	mlxsw_afa_resource_add(block, &counter->resource);
+	return counter;
+
+err_counter_index_get:
+	kfree(counter);
+	return ERR_PTR(err);
+}
+
+/* 20 bits is a maximum that hardware can handle in trap with userdef action
+ * and carry along with the trapped packet.
+ */
+#define MLXSW_AFA_COOKIE_INDEX_BITS 20
+#define MLXSW_AFA_COOKIE_INDEX_MAX ((1 << MLXSW_AFA_COOKIE_INDEX_BITS) - 1)
+
+static struct mlxsw_afa_cookie *
+mlxsw_afa_cookie_create(struct mlxsw_afa *mlxsw_afa,
+			const struct flow_action_cookie *fa_cookie)
+{
+	struct mlxsw_afa_cookie *cookie;
+	u32 cookie_index;
+	int err;
+
+	cookie = kzalloc(sizeof(*cookie) + fa_cookie->cookie_len, GFP_KERNEL);
+	if (!cookie)
+		return ERR_PTR(-ENOMEM);
+	refcount_set(&cookie->ref_count, 1);
+	cookie->fa_cookie = *fa_cookie;
+	memcpy(cookie->fa_cookie.cookie, fa_cookie->cookie,
+	       fa_cookie->cookie_len);
+
+	err = rhashtable_insert_fast(&mlxsw_afa->cookie_ht, &cookie->ht_node,
+				     mlxsw_afa_cookie_ht_params);
+	if (err)
+		goto err_rhashtable_insert;
+
+	/* Start cookie indexes with 1. Leave the 0 index unused. Packets
+	 * that come from the HW which are not dropped by drop-with-cookie
+	 * action are going to pass cookie_index 0 to lookup.
+	 */
+	cookie_index = 1;
+	err = idr_alloc_u32(&mlxsw_afa->cookie_idr, cookie, &cookie_index,
+			    MLXSW_AFA_COOKIE_INDEX_MAX, GFP_KERNEL);
+	if (err)
+		goto err_idr_alloc;
+	cookie->cookie_index = cookie_index;
+	return cookie;
+
+err_idr_alloc:
+	rhashtable_remove_fast(&mlxsw_afa->cookie_ht, &cookie->ht_node,
+			       mlxsw_afa_cookie_ht_params);
+err_rhashtable_insert:
+	kfree(cookie);
+	return ERR_PTR(err);
+}
+
+static void mlxsw_afa_cookie_destroy(struct mlxsw_afa *mlxsw_afa,
+				     struct mlxsw_afa_cookie *cookie)
+{
+	idr_remove(&mlxsw_afa->cookie_idr, cookie->cookie_index);
+	rhashtable_remove_fast(&mlxsw_afa->cookie_ht, &cookie->ht_node,
+			       mlxsw_afa_cookie_ht_params);
+	kfree_rcu(cookie, rcu);
+}
+
+static struct mlxsw_afa_cookie *
+mlxsw_afa_cookie_get(struct mlxsw_afa *mlxsw_afa,
+		     const struct flow_action_cookie *fa_cookie)
+{
+	struct mlxsw_afa_cookie *cookie;
+
+	cookie = rhashtable_lookup_fast(&mlxsw_afa->cookie_ht, fa_cookie,
+					mlxsw_afa_cookie_ht_params);
+	if (cookie) {
+		refcount_inc(&cookie->ref_count);
+		return cookie;
+	}
+	return mlxsw_afa_cookie_create(mlxsw_afa, fa_cookie);
+}
+
+static void mlxsw_afa_cookie_put(struct mlxsw_afa *mlxsw_afa,
+				 struct mlxsw_afa_cookie *cookie)
+{
+	if (!refcount_dec_and_test(&cookie->ref_count))
+		return;
+	mlxsw_afa_cookie_destroy(mlxsw_afa, cookie);
+}
+
+/* RCU read lock must be held */
+const struct flow_action_cookie *
+mlxsw_afa_cookie_lookup(struct mlxsw_afa *mlxsw_afa, u32 cookie_index)
+{
+	struct mlxsw_afa_cookie *cookie;
+
+	/* 0 index means no cookie */
+	if (!cookie_index)
+		return NULL;
+	cookie = idr_find(&mlxsw_afa->cookie_idr, cookie_index);
+	if (!cookie)
+		return NULL;
+	return &cookie->fa_cookie;
+}
+EXPORT_SYMBOL(mlxsw_afa_cookie_lookup);
+
+struct mlxsw_afa_cookie_ref {
+	struct mlxsw_afa_resource resource;
+	struct mlxsw_afa_cookie *cookie;
+};
+
+static void
+mlxsw_afa_cookie_ref_destroy(struct mlxsw_afa_block *block,
+			     struct mlxsw_afa_cookie_ref *cookie_ref)
+{
+	mlxsw_afa_resource_del(&cookie_ref->resource);
+	mlxsw_afa_cookie_put(block->afa, cookie_ref->cookie);
+	kfree(cookie_ref);
+}
+
+static void
+mlxsw_afa_cookie_ref_destructor(struct mlxsw_afa_block *block,
+				struct mlxsw_afa_resource *resource)
+{
+	struct mlxsw_afa_cookie_ref *cookie_ref;
+
+	cookie_ref = container_of(resource, struct mlxsw_afa_cookie_ref,
+				  resource);
+	mlxsw_afa_cookie_ref_destroy(block, cookie_ref);
+}
+
+static struct mlxsw_afa_cookie_ref *
+mlxsw_afa_cookie_ref_create(struct mlxsw_afa_block *block,
+			    const struct flow_action_cookie *fa_cookie)
+{
+	struct mlxsw_afa_cookie_ref *cookie_ref;
+	struct mlxsw_afa_cookie *cookie;
+	int err;
+
+	cookie_ref = kzalloc(sizeof(*cookie_ref), GFP_KERNEL);
+	if (!cookie_ref)
+		return ERR_PTR(-ENOMEM);
+	cookie = mlxsw_afa_cookie_get(block->afa, fa_cookie);
+	if (IS_ERR(cookie)) {
+		err = PTR_ERR(cookie);
+		goto err_cookie_get;
+	}
+	cookie_ref->cookie = cookie;
+	cookie_ref->resource.destructor = mlxsw_afa_cookie_ref_destructor;
+	mlxsw_afa_resource_add(block, &cookie_ref->resource);
+	return cookie_ref;
+
+err_cookie_get:
+	kfree(cookie_ref);
+	return ERR_PTR(err);
+}
+
+static struct mlxsw_afa_policer *
+mlxsw_afa_policer_create(struct mlxsw_afa *mlxsw_afa, u32 fa_index,
+			 u64 rate_bytes_ps, u32 burst,
+			 struct netlink_ext_ack *extack)
+{
+	struct mlxsw_afa_policer *policer;
+	int err;
+
+	policer = kzalloc(sizeof(*policer), GFP_KERNEL);
+	if (!policer)
+		return ERR_PTR(-ENOMEM);
+
+	err = mlxsw_afa->ops->policer_add(mlxsw_afa->ops_priv, rate_bytes_ps,
+					  burst, &policer->policer_index,
+					  extack);
+	if (err)
+		goto err_policer_add;
+
+	refcount_set(&policer->ref_count, 1);
+	policer->fa_index = fa_index;
+
+	err = rhashtable_insert_fast(&mlxsw_afa->policer_ht, &policer->ht_node,
+				     mlxsw_afa_policer_ht_params);
+	if (err)
+		goto err_rhashtable_insert;
+
+	list_add_tail(&policer->list, &mlxsw_afa->policer_list);
+
+	return policer;
+
+err_rhashtable_insert:
+	mlxsw_afa->ops->policer_del(mlxsw_afa->ops_priv,
+				    policer->policer_index);
+err_policer_add:
+	kfree(policer);
+	return ERR_PTR(err);
+}
+
+static void mlxsw_afa_policer_destroy(struct mlxsw_afa *mlxsw_afa,
+				      struct mlxsw_afa_policer *policer)
+{
+	list_del(&policer->list);
+	rhashtable_remove_fast(&mlxsw_afa->policer_ht, &policer->ht_node,
+			       mlxsw_afa_policer_ht_params);
+	mlxsw_afa->ops->policer_del(mlxsw_afa->ops_priv,
+				    policer->policer_index);
+	kfree(policer);
+}
+
+static struct mlxsw_afa_policer *
+mlxsw_afa_policer_get(struct mlxsw_afa *mlxsw_afa, u32 fa_index,
+		      u64 rate_bytes_ps, u32 burst,
+		      struct netlink_ext_ack *extack)
+{
+	struct mlxsw_afa_policer *policer;
+
+	policer = rhashtable_lookup_fast(&mlxsw_afa->policer_ht, &fa_index,
+					 mlxsw_afa_policer_ht_params);
+	if (policer) {
+		refcount_inc(&policer->ref_count);
+		return policer;
+	}
+
+	return mlxsw_afa_policer_create(mlxsw_afa, fa_index, rate_bytes_ps,
+					burst, extack);
+}
+
+static void mlxsw_afa_policer_put(struct mlxsw_afa *mlxsw_afa,
+				  struct mlxsw_afa_policer *policer)
+{
+	if (!refcount_dec_and_test(&policer->ref_count))
+		return;
+	mlxsw_afa_policer_destroy(mlxsw_afa, policer);
+}
+
+struct mlxsw_afa_policer_ref {
+	struct mlxsw_afa_resource resource;
+	struct mlxsw_afa_policer *policer;
+};
+
+static void
+mlxsw_afa_policer_ref_destroy(struct mlxsw_afa_block *block,
+			      struct mlxsw_afa_policer_ref *policer_ref)
+{
+	mlxsw_afa_resource_del(&policer_ref->resource);
+	mlxsw_afa_policer_put(block->afa, policer_ref->policer);
+	kfree(policer_ref);
+}
+
+static void
+mlxsw_afa_policer_ref_destructor(struct mlxsw_afa_block *block,
+				 struct mlxsw_afa_resource *resource)
+{
+	struct mlxsw_afa_policer_ref *policer_ref;
+
+	policer_ref = container_of(resource, struct mlxsw_afa_policer_ref,
+				   resource);
+	mlxsw_afa_policer_ref_destroy(block, policer_ref);
+}
+
+static struct mlxsw_afa_policer_ref *
+mlxsw_afa_policer_ref_create(struct mlxsw_afa_block *block, u32 fa_index,
+			     u64 rate_bytes_ps, u32 burst,
+			     struct netlink_ext_ack *extack)
+{
+	struct mlxsw_afa_policer_ref *policer_ref;
+	struct mlxsw_afa_policer *policer;
+	int err;
+
+	policer_ref = kzalloc(sizeof(*policer_ref), GFP_KERNEL);
+	if (!policer_ref)
+		return ERR_PTR(-ENOMEM);
+
+	policer = mlxsw_afa_policer_get(block->afa, fa_index, rate_bytes_ps,
+					burst, extack);
+	if (IS_ERR(policer)) {
+		err = PTR_ERR(policer);
+		goto err_policer_get;
+	}
+
+	policer_ref->policer = policer;
+	policer_ref->resource.destructor = mlxsw_afa_policer_ref_destructor;
+	mlxsw_afa_resource_add(block, &policer_ref->resource);
+
+	return policer_ref;
+
+err_policer_get:
+	kfree(policer_ref);
+	return ERR_PTR(err);
+}
+
+#define MLXSW_AFA_ONE_ACTION_LEN 32
+#define MLXSW_AFA_PAYLOAD_OFFSET 4
+
+enum mlxsw_afa_action_type {
+	MLXSW_AFA_ACTION_TYPE_TRAP,
+	MLXSW_AFA_ACTION_TYPE_POLICE,
+	MLXSW_AFA_ACTION_TYPE_OTHER,
+};
+
+static bool
+mlxsw_afa_block_need_split(const struct mlxsw_afa_block *block,
+			   enum mlxsw_afa_action_type type)
+{
+	struct mlxsw_afa_set *cur_set = block->cur_set;
+
+	/* Due to a hardware limitation, police action cannot be in the same
+	 * action set with MLXSW_AFA_TRAP_CODE or MLXSW_AFA_TRAPWU_CODE
+	 * actions. Work around this limitation by creating a new action set
+	 * and place the new action there.
+	 */
+	return (cur_set->has_trap && type == MLXSW_AFA_ACTION_TYPE_POLICE) ||
+	       (cur_set->has_police && type == MLXSW_AFA_ACTION_TYPE_TRAP);
+}
+
+static char *mlxsw_afa_block_append_action_ext(struct mlxsw_afa_block *block,
+					       u8 action_code, u8 action_size,
+					       enum mlxsw_afa_action_type type)
+{
+	char *oneact;
+	char *actions;
+
+	if (block->finished)
+		return ERR_PTR(-EINVAL);
+	if (block->cur_act_index + action_size > block->afa->max_acts_per_set ||
+	    mlxsw_afa_block_need_split(block, type)) {
+		struct mlxsw_afa_set *set;
+
+		/* The appended action won't fit into the current action set,
+		 * so create a new set.
+		 */
+		set = mlxsw_afa_set_create(false);
+		if (!set)
+			return ERR_PTR(-ENOBUFS);
+		set->prev = block->cur_set;
+		block->cur_act_index = 0;
+		block->cur_set->next = set;
+		block->cur_set = set;
+	}
+
+	switch (type) {
+	case MLXSW_AFA_ACTION_TYPE_TRAP:
+		block->cur_set->has_trap = true;
+		break;
+	case MLXSW_AFA_ACTION_TYPE_POLICE:
+		block->cur_set->has_police = true;
+		break;
+	default:
+		break;
+	}
+
+	actions = block->cur_set->ht_key.enc_actions;
+	oneact = actions + block->cur_act_index * MLXSW_AFA_ONE_ACTION_LEN;
+	block->cur_act_index += action_size;
+	mlxsw_afa_all_action_type_set(oneact, action_code);
+	return oneact + MLXSW_AFA_PAYLOAD_OFFSET;
+}
+
+static char *mlxsw_afa_block_append_action(struct mlxsw_afa_block *block,
+					   u8 action_code, u8 action_size)
+{
+	return mlxsw_afa_block_append_action_ext(block, action_code,
+						 action_size,
+						 MLXSW_AFA_ACTION_TYPE_OTHER);
+}
+
+/* VLAN Action
+ * -----------
+ * VLAN action is used for manipulating VLANs. It can be used to implement QinQ,
+ * VLAN translation, change of PCP bits of the VLAN tag, push, pop as swap VLANs
+ * and more.
+ */
+
+#define MLXSW_AFA_VLAN_CODE 0x02
+#define MLXSW_AFA_VLAN_SIZE 1
+
+enum mlxsw_afa_vlan_vlan_tag_cmd {
+	MLXSW_AFA_VLAN_VLAN_TAG_CMD_NOP,
+	MLXSW_AFA_VLAN_VLAN_TAG_CMD_PUSH_TAG,
+	MLXSW_AFA_VLAN_VLAN_TAG_CMD_POP_TAG,
+};
+
+enum mlxsw_afa_vlan_cmd {
+	MLXSW_AFA_VLAN_CMD_NOP,
+	MLXSW_AFA_VLAN_CMD_SET_OUTER,
+	MLXSW_AFA_VLAN_CMD_SET_INNER,
+	MLXSW_AFA_VLAN_CMD_COPY_OUTER_TO_INNER,
+	MLXSW_AFA_VLAN_CMD_COPY_INNER_TO_OUTER,
+	MLXSW_AFA_VLAN_CMD_SWAP,
+};
+
+/* afa_vlan_vlan_tag_cmd
+ * Tag command: push, pop, nop VLAN header.
+ */
+MLXSW_ITEM32(afa, vlan, vlan_tag_cmd, 0x00, 29, 3);
+
+/* afa_vlan_vid_cmd */
+MLXSW_ITEM32(afa, vlan, vid_cmd, 0x04, 29, 3);
+
+/* afa_vlan_vid */
+MLXSW_ITEM32(afa, vlan, vid, 0x04, 0, 12);
+
+/* afa_vlan_ethertype_cmd */
+MLXSW_ITEM32(afa, vlan, ethertype_cmd, 0x08, 29, 3);
+
+/* afa_vlan_ethertype
+ * Index to EtherTypes in Switch VLAN EtherType Register (SVER).
+ */
+MLXSW_ITEM32(afa, vlan, ethertype, 0x08, 24, 3);
+
+/* afa_vlan_pcp_cmd */
+MLXSW_ITEM32(afa, vlan, pcp_cmd, 0x08, 13, 3);
+
+/* afa_vlan_pcp */
+MLXSW_ITEM32(afa, vlan, pcp, 0x08, 8, 3);
+
+static inline void
+mlxsw_afa_vlan_pack(char *payload,
+		    enum mlxsw_afa_vlan_vlan_tag_cmd vlan_tag_cmd,
+		    enum mlxsw_afa_vlan_cmd vid_cmd, u16 vid,
+		    enum mlxsw_afa_vlan_cmd pcp_cmd, u8 pcp,
+		    enum mlxsw_afa_vlan_cmd ethertype_cmd, u8 ethertype)
+{
+	mlxsw_afa_vlan_vlan_tag_cmd_set(payload, vlan_tag_cmd);
+	mlxsw_afa_vlan_vid_cmd_set(payload, vid_cmd);
+	mlxsw_afa_vlan_vid_set(payload, vid);
+	mlxsw_afa_vlan_pcp_cmd_set(payload, pcp_cmd);
+	mlxsw_afa_vlan_pcp_set(payload, pcp);
+	mlxsw_afa_vlan_ethertype_cmd_set(payload, ethertype_cmd);
+	mlxsw_afa_vlan_ethertype_set(payload, ethertype);
+}
+
+int mlxsw_afa_block_append_vlan_modify(struct mlxsw_afa_block *block,
+				       u16 vid, u8 pcp, u8 et,
+				       struct netlink_ext_ack *extack)
+{
+	char *act = mlxsw_afa_block_append_action(block,
+						  MLXSW_AFA_VLAN_CODE,
+						  MLXSW_AFA_VLAN_SIZE);
+
+	if (IS_ERR(act)) {
+		NL_SET_ERR_MSG_MOD(extack, "Cannot append vlan_modify action");
+		return PTR_ERR(act);
+	}
+	mlxsw_afa_vlan_pack(act, MLXSW_AFA_VLAN_VLAN_TAG_CMD_NOP,
+			    MLXSW_AFA_VLAN_CMD_SET_OUTER, vid,
+			    MLXSW_AFA_VLAN_CMD_SET_OUTER, pcp,
+			    MLXSW_AFA_VLAN_CMD_SET_OUTER, et);
+	return 0;
+}
+EXPORT_SYMBOL(mlxsw_afa_block_append_vlan_modify);
+
+/* Trap Action / Trap With Userdef Action
+ * --------------------------------------
+ * The Trap action enables trapping / mirroring packets to the CPU
+ * as well as discarding packets.
+ * The ACL Trap / Discard separates the forward/discard control from CPU
+ * trap control. In addition, the Trap / Discard action enables activating
+ * SPAN (port mirroring).
+ *
+ * The Trap with userdef action has the same functionality as
+ * the Trap action with addition of user defined value that can be set
+ * and used by higher layer applications.
+ */
+
+#define MLXSW_AFA_TRAP_CODE 0x03
+#define MLXSW_AFA_TRAP_SIZE 1
+
+#define MLXSW_AFA_TRAPWU_CODE 0x04
+#define MLXSW_AFA_TRAPWU_SIZE 2
+
+enum mlxsw_afa_trap_trap_action {
+	MLXSW_AFA_TRAP_TRAP_ACTION_NOP = 0,
+	MLXSW_AFA_TRAP_TRAP_ACTION_TRAP = 2,
+};
+
+/* afa_trap_trap_action
+ * Trap Action.
+ */
+MLXSW_ITEM32(afa, trap, trap_action, 0x00, 24, 4);
+
+enum mlxsw_afa_trap_forward_action {
+	MLXSW_AFA_TRAP_FORWARD_ACTION_FORWARD = 1,
+	MLXSW_AFA_TRAP_FORWARD_ACTION_DISCARD = 3,
+};
+
+/* afa_trap_forward_action
+ * Forward Action.
+ */
+MLXSW_ITEM32(afa, trap, forward_action, 0x00, 0, 4);
+
+/* afa_trap_trap_id
+ * Trap ID to configure.
+ */
+MLXSW_ITEM32(afa, trap, trap_id, 0x04, 0, 9);
+
+/* afa_trap_mirror_agent
+ * Mirror agent.
+ */
+MLXSW_ITEM32(afa, trap, mirror_agent, 0x08, 29, 3);
+
+/* afa_trap_mirror_enable
+ * Mirror enable.
+ */
+MLXSW_ITEM32(afa, trap, mirror_enable, 0x08, 24, 1);
+
+/* user_def_val
+ * Value for the SW usage. Can be used to pass information of which
+ * rule has caused a trap. This may be overwritten by later traps.
+ * This field does a set on the packet's user_def_val only if this
+ * is the first trap_id or if the trap_id has replaced the previous
+ * packet's trap_id.
+ */
+MLXSW_ITEM32(afa, trap, user_def_val, 0x0C, 0, 20);
+
+static inline void
+mlxsw_afa_trap_pack(char *payload,
+		    enum mlxsw_afa_trap_trap_action trap_action,
+		    enum mlxsw_afa_trap_forward_action forward_action,
+		    u16 trap_id)
+{
+	mlxsw_afa_trap_trap_action_set(payload, trap_action);
+	mlxsw_afa_trap_forward_action_set(payload, forward_action);
+	mlxsw_afa_trap_trap_id_set(payload, trap_id);
+}
+
+static inline void
+mlxsw_afa_trapwu_pack(char *payload,
+		      enum mlxsw_afa_trap_trap_action trap_action,
+		      enum mlxsw_afa_trap_forward_action forward_action,
+		      u16 trap_id, u32 user_def_val)
+{
+	mlxsw_afa_trap_pack(payload, trap_action, forward_action, trap_id);
+	mlxsw_afa_trap_user_def_val_set(payload, user_def_val);
+}
+
+static inline void
+mlxsw_afa_trap_mirror_pack(char *payload, bool mirror_enable,
+			   u8 mirror_agent)
+{
+	mlxsw_afa_trap_mirror_enable_set(payload, mirror_enable);
+	mlxsw_afa_trap_mirror_agent_set(payload, mirror_agent);
+}
+
+static char *mlxsw_afa_block_append_action_trap(struct mlxsw_afa_block *block,
+						u8 action_code, u8 action_size)
+{
+	return mlxsw_afa_block_append_action_ext(block, action_code,
+						 action_size,
+						 MLXSW_AFA_ACTION_TYPE_TRAP);
+}
+
+static int mlxsw_afa_block_append_drop_plain(struct mlxsw_afa_block *block,
+					     bool ingress)
+{
+	char *act = mlxsw_afa_block_append_action_trap(block,
+						       MLXSW_AFA_TRAP_CODE,
+						       MLXSW_AFA_TRAP_SIZE);
+
+	if (IS_ERR(act))
+		return PTR_ERR(act);
+	mlxsw_afa_trap_pack(act, MLXSW_AFA_TRAP_TRAP_ACTION_TRAP,
+			    MLXSW_AFA_TRAP_FORWARD_ACTION_DISCARD,
+			    ingress ? MLXSW_TRAP_ID_DISCARD_INGRESS_ACL :
+				      MLXSW_TRAP_ID_DISCARD_EGRESS_ACL);
+	return 0;
+}
+
+static int
+mlxsw_afa_block_append_drop_with_cookie(struct mlxsw_afa_block *block,
+					bool ingress,
+					const struct flow_action_cookie *fa_cookie,
+					struct netlink_ext_ack *extack)
+{
+	struct mlxsw_afa_cookie_ref *cookie_ref;
+	u32 cookie_index;
+	char *act;
+	int err;
+
+	cookie_ref = mlxsw_afa_cookie_ref_create(block, fa_cookie);
+	if (IS_ERR(cookie_ref)) {
+		NL_SET_ERR_MSG_MOD(extack, "Cannot create cookie for drop action");
+		return PTR_ERR(cookie_ref);
+	}
+	cookie_index = cookie_ref->cookie->cookie_index;
+
+	act = mlxsw_afa_block_append_action_trap(block, MLXSW_AFA_TRAPWU_CODE,
+						 MLXSW_AFA_TRAPWU_SIZE);
+	if (IS_ERR(act)) {
+		NL_SET_ERR_MSG_MOD(extack, "Cannot append drop with cookie action");
+		err = PTR_ERR(act);
+		goto err_append_action;
+	}
+	mlxsw_afa_trapwu_pack(act, MLXSW_AFA_TRAP_TRAP_ACTION_TRAP,
+			      MLXSW_AFA_TRAP_FORWARD_ACTION_DISCARD,
+			      ingress ? MLXSW_TRAP_ID_DISCARD_INGRESS_ACL :
+					MLXSW_TRAP_ID_DISCARD_EGRESS_ACL,
+			      cookie_index);
+	return 0;
+
+err_append_action:
+	mlxsw_afa_cookie_ref_destroy(block, cookie_ref);
+	return err;
+}
+
+int mlxsw_afa_block_append_drop(struct mlxsw_afa_block *block, bool ingress,
+				const struct flow_action_cookie *fa_cookie,
+				struct netlink_ext_ack *extack)
+{
+	return fa_cookie ?
+	       mlxsw_afa_block_append_drop_with_cookie(block, ingress,
+						       fa_cookie, extack) :
+	       mlxsw_afa_block_append_drop_plain(block, ingress);
+}
+EXPORT_SYMBOL(mlxsw_afa_block_append_drop);
+
+int mlxsw_afa_block_append_trap(struct mlxsw_afa_block *block, u16 trap_id)
+{
+	char *act = mlxsw_afa_block_append_action_trap(block,
+						       MLXSW_AFA_TRAP_CODE,
+						       MLXSW_AFA_TRAP_SIZE);
+
+	if (IS_ERR(act))
+		return PTR_ERR(act);
+	mlxsw_afa_trap_pack(act, MLXSW_AFA_TRAP_TRAP_ACTION_TRAP,
+			    MLXSW_AFA_TRAP_FORWARD_ACTION_DISCARD, trap_id);
+	return 0;
+}
+EXPORT_SYMBOL(mlxsw_afa_block_append_trap);
+
+int mlxsw_afa_block_append_trap_and_forward(struct mlxsw_afa_block *block,
+					    u16 trap_id)
+{
+	char *act = mlxsw_afa_block_append_action_trap(block,
+						       MLXSW_AFA_TRAP_CODE,
+						       MLXSW_AFA_TRAP_SIZE);
+
+	if (IS_ERR(act))
+		return PTR_ERR(act);
+	mlxsw_afa_trap_pack(act, MLXSW_AFA_TRAP_TRAP_ACTION_TRAP,
+			    MLXSW_AFA_TRAP_FORWARD_ACTION_FORWARD, trap_id);
+	return 0;
+}
+EXPORT_SYMBOL(mlxsw_afa_block_append_trap_and_forward);
+
+struct mlxsw_afa_mirror {
+	struct mlxsw_afa_resource resource;
+	int span_id;
+	u16 local_in_port;
+	bool ingress;
+};
+
+static void
+mlxsw_afa_mirror_destroy(struct mlxsw_afa_block *block,
+			 struct mlxsw_afa_mirror *mirror)
+{
+	mlxsw_afa_resource_del(&mirror->resource);
+	block->afa->ops->mirror_del(block->afa->ops_priv,
+				    mirror->local_in_port,
+				    mirror->span_id,
+				    mirror->ingress);
+	kfree(mirror);
+}
+
+static void
+mlxsw_afa_mirror_destructor(struct mlxsw_afa_block *block,
+			    struct mlxsw_afa_resource *resource)
+{
+	struct mlxsw_afa_mirror *mirror;
+
+	mirror = container_of(resource, struct mlxsw_afa_mirror, resource);
+	mlxsw_afa_mirror_destroy(block, mirror);
+}
+
+static struct mlxsw_afa_mirror *
+mlxsw_afa_mirror_create(struct mlxsw_afa_block *block, u16 local_in_port,
+			const struct net_device *out_dev, bool ingress)
+{
+	struct mlxsw_afa_mirror *mirror;
+	int err;
+
+	mirror = kzalloc(sizeof(*mirror), GFP_KERNEL);
+	if (!mirror)
+		return ERR_PTR(-ENOMEM);
+
+	err = block->afa->ops->mirror_add(block->afa->ops_priv,
+					  local_in_port, out_dev,
+					  ingress, &mirror->span_id);
+	if (err)
+		goto err_mirror_add;
+
+	mirror->ingress = ingress;
+	mirror->local_in_port = local_in_port;
+	mirror->resource.destructor = mlxsw_afa_mirror_destructor;
+	mlxsw_afa_resource_add(block, &mirror->resource);
+	return mirror;
+
+err_mirror_add:
+	kfree(mirror);
+	return ERR_PTR(err);
+}
+
+static int
+mlxsw_afa_block_append_allocated_mirror(struct mlxsw_afa_block *block,
+					u8 mirror_agent)
+{
+	char *act = mlxsw_afa_block_append_action_trap(block,
+						       MLXSW_AFA_TRAP_CODE,
+						       MLXSW_AFA_TRAP_SIZE);
+
+	if (IS_ERR(act))
+		return PTR_ERR(act);
+	mlxsw_afa_trap_pack(act, MLXSW_AFA_TRAP_TRAP_ACTION_NOP,
+			    MLXSW_AFA_TRAP_FORWARD_ACTION_FORWARD, 0);
+	mlxsw_afa_trap_mirror_pack(act, true, mirror_agent);
+	return 0;
+}
+
+int
+mlxsw_afa_block_append_mirror(struct mlxsw_afa_block *block, u16 local_in_port,
+			      const struct net_device *out_dev, bool ingress,
+			      struct netlink_ext_ack *extack)
+{
+	struct mlxsw_afa_mirror *mirror;
+	int err;
+
+	mirror = mlxsw_afa_mirror_create(block, local_in_port, out_dev,
+					 ingress);
+	if (IS_ERR(mirror)) {
+		NL_SET_ERR_MSG_MOD(extack, "Cannot create mirror action");
+		return PTR_ERR(mirror);
+	}
+	err = mlxsw_afa_block_append_allocated_mirror(block, mirror->span_id);
+	if (err) {
+		NL_SET_ERR_MSG_MOD(extack, "Cannot append mirror action");
+		goto err_append_allocated_mirror;
+	}
+
+	return 0;
+
+err_append_allocated_mirror:
+	mlxsw_afa_mirror_destroy(block, mirror);
+	return err;
+}
+EXPORT_SYMBOL(mlxsw_afa_block_append_mirror);
+
+/* QoS Action
+ * ----------
+ * The QOS_ACTION is used for manipulating the QoS attributes of a packet. It
+ * can be used to change the DCSP, ECN, Color and Switch Priority of the packet.
+ * Note that PCP field can be changed using the VLAN action.
+ */
+
+#define MLXSW_AFA_QOS_CODE 0x06
+#define MLXSW_AFA_QOS_SIZE 1
+
+enum mlxsw_afa_qos_ecn_cmd {
+	/* Do nothing */
+	MLXSW_AFA_QOS_ECN_CMD_NOP,
+	/* Set ECN to afa_qos_ecn */
+	MLXSW_AFA_QOS_ECN_CMD_SET,
+};
+
+/* afa_qos_ecn_cmd
+ */
+MLXSW_ITEM32(afa, qos, ecn_cmd, 0x04, 29, 3);
+
+/* afa_qos_ecn
+ * ECN value.
+ */
+MLXSW_ITEM32(afa, qos, ecn, 0x04, 24, 2);
+
+enum mlxsw_afa_qos_dscp_cmd {
+	/* Do nothing */
+	MLXSW_AFA_QOS_DSCP_CMD_NOP,
+	/* Set DSCP 3 LSB bits according to dscp[2:0] */
+	MLXSW_AFA_QOS_DSCP_CMD_SET_3LSB,
+	/* Set DSCP 3 MSB bits according to dscp[5:3] */
+	MLXSW_AFA_QOS_DSCP_CMD_SET_3MSB,
+	/* Set DSCP 6 bits according to dscp[5:0] */
+	MLXSW_AFA_QOS_DSCP_CMD_SET_ALL,
+};
+
+/* afa_qos_dscp_cmd
+ * DSCP command.
+ */
+MLXSW_ITEM32(afa, qos, dscp_cmd, 0x04, 14, 2);
+
+/* afa_qos_dscp
+ * DSCP value.
+ */
+MLXSW_ITEM32(afa, qos, dscp, 0x04, 0, 6);
+
+enum mlxsw_afa_qos_switch_prio_cmd {
+	/* Do nothing */
+	MLXSW_AFA_QOS_SWITCH_PRIO_CMD_NOP,
+	/* Set Switch Priority to afa_qos_switch_prio */
+	MLXSW_AFA_QOS_SWITCH_PRIO_CMD_SET,
+};
+
+/* afa_qos_switch_prio_cmd
+ */
+MLXSW_ITEM32(afa, qos, switch_prio_cmd, 0x08, 14, 2);
+
+/* afa_qos_switch_prio
+ * Switch Priority.
+ */
+MLXSW_ITEM32(afa, qos, switch_prio, 0x08, 0, 4);
+
+enum mlxsw_afa_qos_dscp_rw {
+	MLXSW_AFA_QOS_DSCP_RW_PRESERVE,
+	MLXSW_AFA_QOS_DSCP_RW_SET,
+	MLXSW_AFA_QOS_DSCP_RW_CLEAR,
+};
+
+/* afa_qos_dscp_rw
+ * DSCP Re-write Enable. Controlling the rewrite_enable for DSCP.
+ */
+MLXSW_ITEM32(afa, qos, dscp_rw, 0x0C, 30, 2);
+
+static inline void
+mlxsw_afa_qos_ecn_pack(char *payload,
+		       enum mlxsw_afa_qos_ecn_cmd ecn_cmd, u8 ecn)
+{
+	mlxsw_afa_qos_ecn_cmd_set(payload, ecn_cmd);
+	mlxsw_afa_qos_ecn_set(payload, ecn);
+}
+
+static inline void
+mlxsw_afa_qos_dscp_pack(char *payload,
+			enum mlxsw_afa_qos_dscp_cmd dscp_cmd, u8 dscp)
+{
+	mlxsw_afa_qos_dscp_cmd_set(payload, dscp_cmd);
+	mlxsw_afa_qos_dscp_set(payload, dscp);
+}
+
+static inline void
+mlxsw_afa_qos_switch_prio_pack(char *payload,
+			       enum mlxsw_afa_qos_switch_prio_cmd prio_cmd,
+			       u8 prio)
+{
+	mlxsw_afa_qos_switch_prio_cmd_set(payload, prio_cmd);
+	mlxsw_afa_qos_switch_prio_set(payload, prio);
+}
+
+static int __mlxsw_afa_block_append_qos_dsfield(struct mlxsw_afa_block *block,
+						bool set_dscp, u8 dscp,
+						bool set_ecn, u8 ecn,
+						struct netlink_ext_ack *extack)
+{
+	char *act = mlxsw_afa_block_append_action(block,
+						  MLXSW_AFA_QOS_CODE,
+						  MLXSW_AFA_QOS_SIZE);
+
+	if (IS_ERR(act)) {
+		NL_SET_ERR_MSG_MOD(extack, "Cannot append QOS action");
+		return PTR_ERR(act);
+	}
+
+	if (set_ecn)
+		mlxsw_afa_qos_ecn_pack(act, MLXSW_AFA_QOS_ECN_CMD_SET, ecn);
+	if (set_dscp) {
+		mlxsw_afa_qos_dscp_pack(act, MLXSW_AFA_QOS_DSCP_CMD_SET_ALL,
+					dscp);
+		mlxsw_afa_qos_dscp_rw_set(act, MLXSW_AFA_QOS_DSCP_RW_CLEAR);
+	}
+
+	return 0;
+}
+
+int mlxsw_afa_block_append_qos_dsfield(struct mlxsw_afa_block *block,
+				       u8 dsfield,
+				       struct netlink_ext_ack *extack)
+{
+	return __mlxsw_afa_block_append_qos_dsfield(block,
+						    true, dsfield >> 2,
+						    true, dsfield & 0x03,
+						    extack);
+}
+EXPORT_SYMBOL(mlxsw_afa_block_append_qos_dsfield);
+
+int mlxsw_afa_block_append_qos_dscp(struct mlxsw_afa_block *block,
+				    u8 dscp, struct netlink_ext_ack *extack)
+{
+	return __mlxsw_afa_block_append_qos_dsfield(block,
+						    true, dscp,
+						    false, 0,
+						    extack);
+}
+EXPORT_SYMBOL(mlxsw_afa_block_append_qos_dscp);
+
+int mlxsw_afa_block_append_qos_ecn(struct mlxsw_afa_block *block,
+				   u8 ecn, struct netlink_ext_ack *extack)
+{
+	return __mlxsw_afa_block_append_qos_dsfield(block,
+						    false, 0,
+						    true, ecn,
+						    extack);
+}
+EXPORT_SYMBOL(mlxsw_afa_block_append_qos_ecn);
+
+int mlxsw_afa_block_append_qos_switch_prio(struct mlxsw_afa_block *block,
+					   u8 prio,
+					   struct netlink_ext_ack *extack)
+{
+	char *act = mlxsw_afa_block_append_action(block,
+						  MLXSW_AFA_QOS_CODE,
+						  MLXSW_AFA_QOS_SIZE);
+
+	if (IS_ERR(act)) {
+		NL_SET_ERR_MSG_MOD(extack, "Cannot append QOS action");
+		return PTR_ERR(act);
+	}
+	mlxsw_afa_qos_switch_prio_pack(act, MLXSW_AFA_QOS_SWITCH_PRIO_CMD_SET,
+				       prio);
+	return 0;
+}
+EXPORT_SYMBOL(mlxsw_afa_block_append_qos_switch_prio);
+
+/* Forwarding Action
+ * -----------------
+ * Forwarding Action can be used to implement Policy Based Switching (PBS)
+ * as well as OpenFlow related "Output" action.
+ */
+
+#define MLXSW_AFA_FORWARD_CODE 0x07
+#define MLXSW_AFA_FORWARD_SIZE 1
+
+enum mlxsw_afa_forward_type {
+	/* PBS, Policy Based Switching */
+	MLXSW_AFA_FORWARD_TYPE_PBS,
+	/* Output, OpenFlow output type */
+	MLXSW_AFA_FORWARD_TYPE_OUTPUT,
+};
+
+/* afa_forward_type */
+MLXSW_ITEM32(afa, forward, type, 0x00, 24, 2);
+
+/* afa_forward_pbs_ptr
+ * A pointer to the PBS entry configured by PPBS register.
+ * Reserved when in_port is set.
+ */
+MLXSW_ITEM32(afa, forward, pbs_ptr, 0x08, 0, 24);
+
+/* afa_forward_in_port
+ * Packet is forwarded back to the ingress port.
+ */
+MLXSW_ITEM32(afa, forward, in_port, 0x0C, 0, 1);
+
+static inline void
+mlxsw_afa_forward_pack(char *payload, enum mlxsw_afa_forward_type type,
+		       u32 pbs_ptr, bool in_port)
+{
+	mlxsw_afa_forward_type_set(payload, type);
+	mlxsw_afa_forward_pbs_ptr_set(payload, pbs_ptr);
+	mlxsw_afa_forward_in_port_set(payload, in_port);
+}
+
+int mlxsw_afa_block_append_fwd(struct mlxsw_afa_block *block,
+			       u16 local_port, bool in_port,
+			       struct netlink_ext_ack *extack)
+{
+	struct mlxsw_afa_fwd_entry_ref *fwd_entry_ref;
+	u32 kvdl_index;
+	char *act;
+	int err;
+
+	if (in_port) {
+		NL_SET_ERR_MSG_MOD(extack, "Forwarding to ingress port is not supported");
+		return -EOPNOTSUPP;
+	}
+	fwd_entry_ref = mlxsw_afa_fwd_entry_ref_create(block, local_port);
+	if (IS_ERR(fwd_entry_ref)) {
+		NL_SET_ERR_MSG_MOD(extack, "Cannot create forward action");
+		return PTR_ERR(fwd_entry_ref);
+	}
+	kvdl_index = fwd_entry_ref->fwd_entry->kvdl_index;
+
+	act = mlxsw_afa_block_append_action(block, MLXSW_AFA_FORWARD_CODE,
+					    MLXSW_AFA_FORWARD_SIZE);
+	if (IS_ERR(act)) {
+		NL_SET_ERR_MSG_MOD(extack, "Cannot append forward action");
+		err = PTR_ERR(act);
+		goto err_append_action;
+	}
+	mlxsw_afa_forward_pack(act, MLXSW_AFA_FORWARD_TYPE_PBS,
+			       kvdl_index, in_port);
+	return 0;
+
+err_append_action:
+	mlxsw_afa_fwd_entry_ref_destroy(block, fwd_entry_ref);
+	return err;
+}
+EXPORT_SYMBOL(mlxsw_afa_block_append_fwd);
+
+/* Policing and Counting Action
+ * ----------------------------
+ * Policing and Counting action is used for binding policer and counter
+ * to ACL rules.
+ */
+
+#define MLXSW_AFA_POLCNT_CODE 0x08
+#define MLXSW_AFA_POLCNT_SIZE 1
+
+enum {
+	MLXSW_AFA_POLCNT_COUNTER,
+	MLXSW_AFA_POLCNT_POLICER,
+};
+
+/* afa_polcnt_c_p
+ * Counter or policer.
+ * Indicates whether the action binds a policer or a counter to the flow.
+ * 0: Counter
+ * 1: Policer
+ */
+MLXSW_ITEM32(afa, polcnt, c_p, 0x00, 31, 1);
+
+enum mlxsw_afa_polcnt_counter_set_type {
+	/* No count */
+	MLXSW_AFA_POLCNT_COUNTER_SET_TYPE_NO_COUNT = 0x00,
+	/* Count packets and bytes */
+	MLXSW_AFA_POLCNT_COUNTER_SET_TYPE_PACKETS_BYTES = 0x03,
+	/* Count only packets */
+	MLXSW_AFA_POLCNT_COUNTER_SET_TYPE_PACKETS = 0x05,
+};
+
+/* afa_polcnt_counter_set_type
+ * Counter set type for flow counters.
+ */
+MLXSW_ITEM32(afa, polcnt, counter_set_type, 0x04, 24, 8);
+
+/* afa_polcnt_counter_index
+ * Counter index for flow counters.
+ */
+MLXSW_ITEM32(afa, polcnt, counter_index, 0x04, 0, 24);
+
+/* afa_polcnt_pid
+ * Policer ID.
+ * Reserved when c_p = 0
+ */
+MLXSW_ITEM32(afa, polcnt, pid, 0x08, 0, 14);
+
+static inline void
+mlxsw_afa_polcnt_pack(char *payload,
+		      enum mlxsw_afa_polcnt_counter_set_type set_type,
+		      u32 counter_index)
+{
+	mlxsw_afa_polcnt_c_p_set(payload, MLXSW_AFA_POLCNT_COUNTER);
+	mlxsw_afa_polcnt_counter_set_type_set(payload, set_type);
+	mlxsw_afa_polcnt_counter_index_set(payload, counter_index);
+}
+
+static void mlxsw_afa_polcnt_policer_pack(char *payload, u16 policer_index)
+{
+	mlxsw_afa_polcnt_c_p_set(payload, MLXSW_AFA_POLCNT_POLICER);
+	mlxsw_afa_polcnt_pid_set(payload, policer_index);
+}
+
+int mlxsw_afa_block_append_allocated_counter(struct mlxsw_afa_block *block,
+					     u32 counter_index)
+{
+	char *act = mlxsw_afa_block_append_action(block, MLXSW_AFA_POLCNT_CODE,
+						  MLXSW_AFA_POLCNT_SIZE);
+	if (IS_ERR(act))
+		return PTR_ERR(act);
+	mlxsw_afa_polcnt_pack(act, MLXSW_AFA_POLCNT_COUNTER_SET_TYPE_PACKETS_BYTES,
+			      counter_index);
+	return 0;
+}
+EXPORT_SYMBOL(mlxsw_afa_block_append_allocated_counter);
+
+int mlxsw_afa_block_append_counter(struct mlxsw_afa_block *block,
+				   u32 *p_counter_index,
+				   struct netlink_ext_ack *extack)
+{
+	struct mlxsw_afa_counter *counter;
+	u32 counter_index;
+	int err;
+
+	counter = mlxsw_afa_counter_create(block);
+	if (IS_ERR(counter)) {
+		NL_SET_ERR_MSG_MOD(extack, "Cannot create count action");
+		return PTR_ERR(counter);
+	}
+	counter_index = counter->counter_index;
+
+	err = mlxsw_afa_block_append_allocated_counter(block, counter_index);
+	if (err) {
+		NL_SET_ERR_MSG_MOD(extack, "Cannot append count action");
+		goto err_append_allocated_counter;
+	}
+	if (p_counter_index)
+		*p_counter_index = counter_index;
+	return 0;
+
+err_append_allocated_counter:
+	mlxsw_afa_counter_destroy(block, counter);
+	return err;
+}
+EXPORT_SYMBOL(mlxsw_afa_block_append_counter);
+
+int mlxsw_afa_block_append_police(struct mlxsw_afa_block *block,
+				  u32 fa_index, u64 rate_bytes_ps, u32 burst,
+				  u16 *p_policer_index,
+				  struct netlink_ext_ack *extack)
+{
+	struct mlxsw_afa_policer_ref *policer_ref;
+	char *act;
+	int err;
+
+	policer_ref = mlxsw_afa_policer_ref_create(block, fa_index,
+						   rate_bytes_ps,
+						   burst, extack);
+	if (IS_ERR(policer_ref))
+		return PTR_ERR(policer_ref);
+	*p_policer_index = policer_ref->policer->policer_index;
+
+	act = mlxsw_afa_block_append_action_ext(block, MLXSW_AFA_POLCNT_CODE,
+						MLXSW_AFA_POLCNT_SIZE,
+						MLXSW_AFA_ACTION_TYPE_POLICE);
+	if (IS_ERR(act)) {
+		NL_SET_ERR_MSG_MOD(extack, "Cannot append police action");
+		err = PTR_ERR(act);
+		goto err_append_action;
+	}
+	mlxsw_afa_polcnt_policer_pack(act, *p_policer_index);
+
+	return 0;
+
+err_append_action:
+	mlxsw_afa_policer_ref_destroy(block, policer_ref);
+	return err;
+}
+EXPORT_SYMBOL(mlxsw_afa_block_append_police);
+
+/* Virtual Router and Forwarding Domain Action
+ * -------------------------------------------
+ * Virtual Switch action is used for manipulate the Virtual Router (VR),
+ * MPLS label space and the Forwarding Identifier (FID).
+ */
+
+#define MLXSW_AFA_VIRFWD_CODE 0x0E
+#define MLXSW_AFA_VIRFWD_SIZE 1
+
+enum mlxsw_afa_virfwd_fid_cmd {
+	/* Do nothing */
+	MLXSW_AFA_VIRFWD_FID_CMD_NOOP,
+	/* Set the Forwarding Identifier (FID) to fid */
+	MLXSW_AFA_VIRFWD_FID_CMD_SET,
+};
+
+/* afa_virfwd_fid_cmd */
+MLXSW_ITEM32(afa, virfwd, fid_cmd, 0x08, 29, 3);
+
+/* afa_virfwd_fid
+ * The FID value.
+ */
+MLXSW_ITEM32(afa, virfwd, fid, 0x08, 0, 16);
+
+static inline void mlxsw_afa_virfwd_pack(char *payload,
+					 enum mlxsw_afa_virfwd_fid_cmd fid_cmd,
+					 u16 fid)
+{
+	mlxsw_afa_virfwd_fid_cmd_set(payload, fid_cmd);
+	mlxsw_afa_virfwd_fid_set(payload, fid);
+}
+
+int mlxsw_afa_block_append_fid_set(struct mlxsw_afa_block *block, u16 fid,
+				   struct netlink_ext_ack *extack)
+{
+	char *act = mlxsw_afa_block_append_action(block,
+						  MLXSW_AFA_VIRFWD_CODE,
+						  MLXSW_AFA_VIRFWD_SIZE);
+	if (IS_ERR(act)) {
+		NL_SET_ERR_MSG_MOD(extack, "Cannot append fid_set action");
+		return PTR_ERR(act);
+	}
+	mlxsw_afa_virfwd_pack(act, MLXSW_AFA_VIRFWD_FID_CMD_SET, fid);
+	return 0;
+}
+EXPORT_SYMBOL(mlxsw_afa_block_append_fid_set);
+
+/* MC Routing Action
+ * -----------------
+ * The Multicast router action. Can be used by RMFT_V2 - Router Multicast
+ * Forwarding Table Version 2 Register.
+ */
+
+#define MLXSW_AFA_MCROUTER_CODE 0x10
+#define MLXSW_AFA_MCROUTER_SIZE 2
+
+enum mlxsw_afa_mcrouter_rpf_action {
+	MLXSW_AFA_MCROUTER_RPF_ACTION_NOP,
+	MLXSW_AFA_MCROUTER_RPF_ACTION_TRAP,
+	MLXSW_AFA_MCROUTER_RPF_ACTION_DISCARD_ERROR,
+};
+
+/* afa_mcrouter_rpf_action */
+MLXSW_ITEM32(afa, mcrouter, rpf_action, 0x00, 28, 3);
+
+/* afa_mcrouter_expected_irif */
+MLXSW_ITEM32(afa, mcrouter, expected_irif, 0x00, 0, 16);
+
+/* afa_mcrouter_min_mtu */
+MLXSW_ITEM32(afa, mcrouter, min_mtu, 0x08, 0, 16);
+
+enum mlxsw_afa_mrouter_vrmid {
+	MLXSW_AFA_MCROUTER_VRMID_INVALID,
+	MLXSW_AFA_MCROUTER_VRMID_VALID
+};
+
+/* afa_mcrouter_vrmid
+ * Valid RMID: rigr_rmid_index is used as RMID
+ */
+MLXSW_ITEM32(afa, mcrouter, vrmid, 0x0C, 31, 1);
+
+/* afa_mcrouter_rigr_rmid_index
+ * When the vrmid field is set to invalid, the field is used as pointer to
+ * Router Interface Group (RIGR) Table in the KVD linear.
+ * When the vrmid is set to valid, the field is used as RMID index, ranged
+ * from 0 to max_mid - 1. The index is to the Port Group Table.
+ */
+MLXSW_ITEM32(afa, mcrouter, rigr_rmid_index, 0x0C, 0, 24);
+
+static inline void
+mlxsw_afa_mcrouter_pack(char *payload,
+			enum mlxsw_afa_mcrouter_rpf_action rpf_action,
+			u16 expected_irif, u16 min_mtu,
+			enum mlxsw_afa_mrouter_vrmid vrmid, u32 rigr_rmid_index)
+
+{
+	mlxsw_afa_mcrouter_rpf_action_set(payload, rpf_action);
+	mlxsw_afa_mcrouter_expected_irif_set(payload, expected_irif);
+	mlxsw_afa_mcrouter_min_mtu_set(payload, min_mtu);
+	mlxsw_afa_mcrouter_vrmid_set(payload, vrmid);
+	mlxsw_afa_mcrouter_rigr_rmid_index_set(payload, rigr_rmid_index);
+}
+
+int mlxsw_afa_block_append_mcrouter(struct mlxsw_afa_block *block,
+				    u16 expected_irif, u16 min_mtu,
+				    bool rmid_valid, u32 kvdl_index)
+{
+	char *act = mlxsw_afa_block_append_action(block,
+						  MLXSW_AFA_MCROUTER_CODE,
+						  MLXSW_AFA_MCROUTER_SIZE);
+	if (IS_ERR(act))
+		return PTR_ERR(act);
+	mlxsw_afa_mcrouter_pack(act, MLXSW_AFA_MCROUTER_RPF_ACTION_TRAP,
+				expected_irif, min_mtu, rmid_valid, kvdl_index);
+	return 0;
+}
+EXPORT_SYMBOL(mlxsw_afa_block_append_mcrouter);
+
+/* SIP DIP Action
+ * --------------
+ * The SIP_DIP_ACTION is used for modifying the SIP and DIP fields of the
+ * packet, e.g. for NAT. The L3 checksum is updated. Also, if the L4 is TCP or
+ * if the L4 is UDP and the checksum field is not zero, then the L4 checksum is
+ * updated.
+ */
+
+#define MLXSW_AFA_IP_CODE 0x11
+#define MLXSW_AFA_IP_SIZE 2
+
+enum mlxsw_afa_ip_s_d {
+	/* ip refers to dip */
+	MLXSW_AFA_IP_S_D_DIP,
+	/* ip refers to sip */
+	MLXSW_AFA_IP_S_D_SIP,
+};
+
+/* afa_ip_s_d
+ * Source or destination.
+ */
+MLXSW_ITEM32(afa, ip, s_d, 0x00, 31, 1);
+
+enum mlxsw_afa_ip_m_l {
+	/* LSB: ip[63:0] refers to ip[63:0] */
+	MLXSW_AFA_IP_M_L_LSB,
+	/* MSB: ip[63:0] refers to ip[127:64] */
+	MLXSW_AFA_IP_M_L_MSB,
+};
+
+/* afa_ip_m_l
+ * MSB or LSB.
+ */
+MLXSW_ITEM32(afa, ip, m_l, 0x00, 30, 1);
+
+/* afa_ip_ip_63_32
+ * Bits [63:32] in the IP address to change to.
+ */
+MLXSW_ITEM32(afa, ip, ip_63_32, 0x08, 0, 32);
+
+/* afa_ip_ip_31_0
+ * Bits [31:0] in the IP address to change to.
+ */
+MLXSW_ITEM32(afa, ip, ip_31_0, 0x0C, 0, 32);
+
+static void mlxsw_afa_ip_pack(char *payload, enum mlxsw_afa_ip_s_d s_d,
+			      enum mlxsw_afa_ip_m_l m_l, u32 ip_31_0,
+			      u32 ip_63_32)
+{
+	mlxsw_afa_ip_s_d_set(payload, s_d);
+	mlxsw_afa_ip_m_l_set(payload, m_l);
+	mlxsw_afa_ip_ip_31_0_set(payload, ip_31_0);
+	mlxsw_afa_ip_ip_63_32_set(payload, ip_63_32);
+}
+
+int mlxsw_afa_block_append_ip(struct mlxsw_afa_block *block, bool is_dip,
+			      bool is_lsb, u32 val_31_0, u32 val_63_32,
+			      struct netlink_ext_ack *extack)
+{
+	enum mlxsw_afa_ip_s_d s_d = is_dip ? MLXSW_AFA_IP_S_D_DIP :
+					     MLXSW_AFA_IP_S_D_SIP;
+	enum mlxsw_afa_ip_m_l m_l = is_lsb ? MLXSW_AFA_IP_M_L_LSB :
+					     MLXSW_AFA_IP_M_L_MSB;
+	char *act = mlxsw_afa_block_append_action(block,
+						  MLXSW_AFA_IP_CODE,
+						  MLXSW_AFA_IP_SIZE);
+
+	if (IS_ERR(act)) {
+		NL_SET_ERR_MSG_MOD(extack, "Cannot append IP action");
+		return PTR_ERR(act);
+	}
+
+	mlxsw_afa_ip_pack(act, s_d, m_l, val_31_0, val_63_32);
+	return 0;
+}
+EXPORT_SYMBOL(mlxsw_afa_block_append_ip);
+
+/* L4 Port Action
+ * --------------
+ * The L4_PORT_ACTION is used for modifying the sport and dport fields of the packet, e.g. for NAT.
+ * If (the L4 is TCP) or if (the L4 is UDP and checksum field!=0) then the L4 checksum is updated.
+ */
+
+#define MLXSW_AFA_L4PORT_CODE 0x12
+#define MLXSW_AFA_L4PORT_SIZE 1
+
+enum mlxsw_afa_l4port_s_d {
+	/* configure src_l4_port */
+	MLXSW_AFA_L4PORT_S_D_SRC,
+	/* configure dst_l4_port */
+	MLXSW_AFA_L4PORT_S_D_DST,
+};
+
+/* afa_l4port_s_d
+ * Source or destination.
+ */
+MLXSW_ITEM32(afa, l4port, s_d, 0x00, 31, 1);
+
+/* afa_l4port_l4_port
+ * Number of port to change to.
+ */
+MLXSW_ITEM32(afa, l4port, l4_port, 0x08, 0, 16);
+
+static void mlxsw_afa_l4port_pack(char *payload, enum mlxsw_afa_l4port_s_d s_d, u16 l4_port)
+{
+	mlxsw_afa_l4port_s_d_set(payload, s_d);
+	mlxsw_afa_l4port_l4_port_set(payload, l4_port);
+}
+
+int mlxsw_afa_block_append_l4port(struct mlxsw_afa_block *block, bool is_dport, u16 l4_port,
+				  struct netlink_ext_ack *extack)
+{
+	enum mlxsw_afa_l4port_s_d s_d = is_dport ? MLXSW_AFA_L4PORT_S_D_DST :
+						   MLXSW_AFA_L4PORT_S_D_SRC;
+	char *act = mlxsw_afa_block_append_action(block,
+						  MLXSW_AFA_L4PORT_CODE,
+						  MLXSW_AFA_L4PORT_SIZE);
+
+	if (IS_ERR(act)) {
+		NL_SET_ERR_MSG_MOD(extack, "Cannot append L4_PORT action");
+		return PTR_ERR(act);
+	}
+
+	mlxsw_afa_l4port_pack(act, s_d, l4_port);
+	return 0;
+}
+EXPORT_SYMBOL(mlxsw_afa_block_append_l4port);
+
+/* Mirror Sampler Action
+ * ---------------------
+ * The SAMPLER_ACTION is used to mirror packets with a probability (sampling).
+ */
+
+#define MLXSW_AFA_SAMPLER_CODE 0x13
+#define MLXSW_AFA_SAMPLER_SIZE 1
+
+/* afa_sampler_mirror_agent
+ * Mirror (SPAN) agent.
+ */
+MLXSW_ITEM32(afa, sampler, mirror_agent, 0x04, 0, 3);
+
+#define MLXSW_AFA_SAMPLER_RATE_MAX (BIT(24) - 1)
+
+/* afa_sampler_mirror_probability_rate
+ * Mirroring probability.
+ * Valid values are 1 to 2^24 - 1
+ */
+MLXSW_ITEM32(afa, sampler, mirror_probability_rate, 0x08, 0, 24);
+
+static void mlxsw_afa_sampler_pack(char *payload, u8 mirror_agent, u32 rate)
+{
+	mlxsw_afa_sampler_mirror_agent_set(payload, mirror_agent);
+	mlxsw_afa_sampler_mirror_probability_rate_set(payload, rate);
+}
+
+struct mlxsw_afa_sampler {
+	struct mlxsw_afa_resource resource;
+	int span_id;
+	u16 local_port;
+	bool ingress;
+};
+
+static void mlxsw_afa_sampler_destroy(struct mlxsw_afa_block *block,
+				      struct mlxsw_afa_sampler *sampler)
+{
+	mlxsw_afa_resource_del(&sampler->resource);
+	block->afa->ops->sampler_del(block->afa->ops_priv, sampler->local_port,
+				     sampler->span_id, sampler->ingress);
+	kfree(sampler);
+}
+
+static void mlxsw_afa_sampler_destructor(struct mlxsw_afa_block *block,
+					 struct mlxsw_afa_resource *resource)
+{
+	struct mlxsw_afa_sampler *sampler;
+
+	sampler = container_of(resource, struct mlxsw_afa_sampler, resource);
+	mlxsw_afa_sampler_destroy(block, sampler);
+}
+
+static struct mlxsw_afa_sampler *
+mlxsw_afa_sampler_create(struct mlxsw_afa_block *block, u16 local_port,
+			 struct psample_group *psample_group, u32 rate,
+			 u32 trunc_size, bool truncate, bool ingress,
+			 struct netlink_ext_ack *extack)
+{
+	struct mlxsw_afa_sampler *sampler;
+	int err;
+
+	sampler = kzalloc(sizeof(*sampler), GFP_KERNEL);
+	if (!sampler)
+		return ERR_PTR(-ENOMEM);
+
+	err = block->afa->ops->sampler_add(block->afa->ops_priv, local_port,
+					   psample_group, rate, trunc_size,
+					   truncate, ingress, &sampler->span_id,
+					   extack);
+	if (err)
+		goto err_sampler_add;
+
+	sampler->ingress = ingress;
+	sampler->local_port = local_port;
+	sampler->resource.destructor = mlxsw_afa_sampler_destructor;
+	mlxsw_afa_resource_add(block, &sampler->resource);
+	return sampler;
+
+err_sampler_add:
+	kfree(sampler);
+	return ERR_PTR(err);
+}
+
+static int
+mlxsw_afa_block_append_allocated_sampler(struct mlxsw_afa_block *block,
+					 u8 mirror_agent, u32 rate)
+{
+	char *act = mlxsw_afa_block_append_action(block, MLXSW_AFA_SAMPLER_CODE,
+						  MLXSW_AFA_SAMPLER_SIZE);
+
+	if (IS_ERR(act))
+		return PTR_ERR(act);
+	mlxsw_afa_sampler_pack(act, mirror_agent, rate);
+	return 0;
+}
+
+int mlxsw_afa_block_append_sampler(struct mlxsw_afa_block *block, u16 local_port,
+				   struct psample_group *psample_group,
+				   u32 rate, u32 trunc_size, bool truncate,
+				   bool ingress,
+				   struct netlink_ext_ack *extack)
+{
+	struct mlxsw_afa_sampler *sampler;
+	int err;
+
+	if (rate > MLXSW_AFA_SAMPLER_RATE_MAX) {
+		NL_SET_ERR_MSG_MOD(extack, "Sampling rate is too high");
+		return -EINVAL;
+	}
+
+	sampler = mlxsw_afa_sampler_create(block, local_port, psample_group,
+					   rate, trunc_size, truncate, ingress,
+					   extack);
+	if (IS_ERR(sampler))
+		return PTR_ERR(sampler);
+
+	err = mlxsw_afa_block_append_allocated_sampler(block, sampler->span_id,
+						       rate);
+	if (err) {
+		NL_SET_ERR_MSG_MOD(extack, "Cannot append sampler action");
+		goto err_append_allocated_sampler;
+	}
+
+	return 0;
+
+err_append_allocated_sampler:
+	mlxsw_afa_sampler_destroy(block, sampler);
+	return err;
+}
+EXPORT_SYMBOL(mlxsw_afa_block_append_sampler);
diff --git a/drivers/net/ethernet/mellanox/mlxsw/core_acl_flex_actions.h b/drivers/net/ethernet/mellanox/mlxsw/core_acl_flex_actions.h
new file mode 100644
index 000000000..db58037be
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/core_acl_flex_actions.h
@@ -0,0 +1,110 @@
+/* SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0 */
+/* Copyright (c) 2017-2018 Mellanox Technologies. All rights reserved */
+
+#ifndef _MLXSW_CORE_ACL_FLEX_ACTIONS_H
+#define _MLXSW_CORE_ACL_FLEX_ACTIONS_H
+
+#include <linux/types.h>
+#include <linux/netdevice.h>
+#include <net/flow_offload.h>
+
+struct mlxsw_afa;
+struct mlxsw_afa_block;
+
+struct mlxsw_afa_ops {
+	int (*kvdl_set_add)(void *priv, u32 *p_kvdl_index,
+			    char *enc_actions, bool is_first);
+	void (*kvdl_set_del)(void *priv, u32 kvdl_index, bool is_first);
+	int (*kvdl_set_activity_get)(void *priv, u32 kvdl_index,
+				     bool *activity);
+	int (*kvdl_fwd_entry_add)(void *priv, u32 *p_kvdl_index, u16 local_port);
+	void (*kvdl_fwd_entry_del)(void *priv, u32 kvdl_index);
+	int (*counter_index_get)(void *priv, unsigned int *p_counter_index);
+	void (*counter_index_put)(void *priv, unsigned int counter_index);
+	int (*mirror_add)(void *priv, u16 local_in_port,
+			  const struct net_device *out_dev,
+			  bool ingress, int *p_span_id);
+	void (*mirror_del)(void *priv, u16 local_in_port, int span_id,
+			   bool ingress);
+	int (*policer_add)(void *priv, u64 rate_bytes_ps, u32 burst,
+			   u16 *p_policer_index,
+			   struct netlink_ext_ack *extack);
+	void (*policer_del)(void *priv, u16 policer_index);
+	int (*sampler_add)(void *priv, u16 local_port,
+			   struct psample_group *psample_group, u32 rate,
+			   u32 trunc_size, bool truncate, bool ingress,
+			   int *p_span_id, struct netlink_ext_ack *extack);
+	void (*sampler_del)(void *priv, u16 local_port, int span_id,
+			    bool ingress);
+	bool dummy_first_set;
+};
+
+struct mlxsw_afa *mlxsw_afa_create(unsigned int max_acts_per_set,
+				   const struct mlxsw_afa_ops *ops,
+				   void *ops_priv);
+void mlxsw_afa_destroy(struct mlxsw_afa *mlxsw_afa);
+struct mlxsw_afa_block *mlxsw_afa_block_create(struct mlxsw_afa *mlxsw_afa);
+void mlxsw_afa_block_destroy(struct mlxsw_afa_block *block);
+int mlxsw_afa_block_commit(struct mlxsw_afa_block *block);
+char *mlxsw_afa_block_first_set(struct mlxsw_afa_block *block);
+char *mlxsw_afa_block_cur_set(struct mlxsw_afa_block *block);
+u32 mlxsw_afa_block_first_kvdl_index(struct mlxsw_afa_block *block);
+int mlxsw_afa_block_activity_get(struct mlxsw_afa_block *block, bool *activity);
+int mlxsw_afa_block_continue(struct mlxsw_afa_block *block);
+int mlxsw_afa_block_jump(struct mlxsw_afa_block *block, u16 group_id);
+int mlxsw_afa_block_terminate(struct mlxsw_afa_block *block);
+const struct flow_action_cookie *
+mlxsw_afa_cookie_lookup(struct mlxsw_afa *mlxsw_afa, u32 cookie_index);
+int mlxsw_afa_block_append_drop(struct mlxsw_afa_block *block, bool ingress,
+				const struct flow_action_cookie *fa_cookie,
+				struct netlink_ext_ack *extack);
+int mlxsw_afa_block_append_trap(struct mlxsw_afa_block *block, u16 trap_id);
+int mlxsw_afa_block_append_trap_and_forward(struct mlxsw_afa_block *block,
+					    u16 trap_id);
+int mlxsw_afa_block_append_mirror(struct mlxsw_afa_block *block,
+				  u16 local_in_port,
+				  const struct net_device *out_dev,
+				  bool ingress,
+				  struct netlink_ext_ack *extack);
+int mlxsw_afa_block_append_fwd(struct mlxsw_afa_block *block,
+			       u16 local_port, bool in_port,
+			       struct netlink_ext_ack *extack);
+int mlxsw_afa_block_append_vlan_modify(struct mlxsw_afa_block *block,
+				       u16 vid, u8 pcp, u8 et,
+				       struct netlink_ext_ack *extack);
+int mlxsw_afa_block_append_qos_switch_prio(struct mlxsw_afa_block *block,
+					   u8 prio,
+					   struct netlink_ext_ack *extack);
+int mlxsw_afa_block_append_qos_dsfield(struct mlxsw_afa_block *block,
+				       u8 dsfield,
+				       struct netlink_ext_ack *extack);
+int mlxsw_afa_block_append_qos_dscp(struct mlxsw_afa_block *block,
+				    u8 dscp, struct netlink_ext_ack *extack);
+int mlxsw_afa_block_append_qos_ecn(struct mlxsw_afa_block *block,
+				   u8 ecn, struct netlink_ext_ack *extack);
+int mlxsw_afa_block_append_allocated_counter(struct mlxsw_afa_block *block,
+					     u32 counter_index);
+int mlxsw_afa_block_append_counter(struct mlxsw_afa_block *block,
+				   u32 *p_counter_index,
+				   struct netlink_ext_ack *extack);
+int mlxsw_afa_block_append_fid_set(struct mlxsw_afa_block *block, u16 fid,
+				   struct netlink_ext_ack *extack);
+int mlxsw_afa_block_append_mcrouter(struct mlxsw_afa_block *block,
+				    u16 expected_irif, u16 min_mtu,
+				    bool rmid_valid, u32 kvdl_index);
+int mlxsw_afa_block_append_ip(struct mlxsw_afa_block *block, bool is_dip,
+			      bool is_lsb, u32 val_31_0, u32 val_63_32,
+			      struct netlink_ext_ack *extack);
+int mlxsw_afa_block_append_l4port(struct mlxsw_afa_block *block, bool is_dport, u16 l4_port,
+				  struct netlink_ext_ack *extack);
+int mlxsw_afa_block_append_police(struct mlxsw_afa_block *block,
+				  u32 fa_index, u64 rate_bytes_ps, u32 burst,
+				  u16 *p_policer_index,
+				  struct netlink_ext_ack *extack);
+int mlxsw_afa_block_append_sampler(struct mlxsw_afa_block *block, u16 local_port,
+				   struct psample_group *psample_group,
+				   u32 rate, u32 trunc_size, bool truncate,
+				   bool ingress,
+				   struct netlink_ext_ack *extack);
+
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlxsw/core_acl_flex_keys.c b/drivers/net/ethernet/mellanox/mlxsw/core_acl_flex_keys.c
new file mode 100644
index 000000000..f208a237d
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/core_acl_flex_keys.c
@@ -0,0 +1,506 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2017-2018 Mellanox Technologies. All rights reserved */
+
+#include <linux/kernel.h>
+#include <linux/slab.h>
+#include <linux/list.h>
+#include <linux/errno.h>
+
+#include "item.h"
+#include "core_acl_flex_keys.h"
+
+/* For the purpose of the driver, define an internal storage scratchpad
+ * that will be used to store key/mask values. For each defined element type
+ * define an internal storage geometry.
+ *
+ * When adding new elements, MLXSW_AFK_ELEMENT_STORAGE_SIZE must be increased
+ * accordingly.
+ */
+static const struct mlxsw_afk_element_info mlxsw_afk_element_infos[] = {
+	MLXSW_AFK_ELEMENT_INFO_U32(SRC_SYS_PORT, 0x00, 16, 16),
+	MLXSW_AFK_ELEMENT_INFO_BUF(DMAC_32_47, 0x04, 2),
+	MLXSW_AFK_ELEMENT_INFO_BUF(DMAC_0_31, 0x06, 4),
+	MLXSW_AFK_ELEMENT_INFO_BUF(SMAC_32_47, 0x0A, 2),
+	MLXSW_AFK_ELEMENT_INFO_BUF(SMAC_0_31, 0x0C, 4),
+	MLXSW_AFK_ELEMENT_INFO_U32(ETHERTYPE, 0x00, 0, 16),
+	MLXSW_AFK_ELEMENT_INFO_U32(IP_PROTO, 0x10, 0, 8),
+	MLXSW_AFK_ELEMENT_INFO_U32(VID, 0x10, 8, 12),
+	MLXSW_AFK_ELEMENT_INFO_U32(PCP, 0x10, 20, 3),
+	MLXSW_AFK_ELEMENT_INFO_U32(TCP_FLAGS, 0x10, 23, 9),
+	MLXSW_AFK_ELEMENT_INFO_U32(DST_L4_PORT, 0x14, 0, 16),
+	MLXSW_AFK_ELEMENT_INFO_U32(SRC_L4_PORT, 0x14, 16, 16),
+	MLXSW_AFK_ELEMENT_INFO_U32(IP_TTL_, 0x18, 0, 8),
+	MLXSW_AFK_ELEMENT_INFO_U32(IP_ECN, 0x18, 9, 2),
+	MLXSW_AFK_ELEMENT_INFO_U32(IP_DSCP, 0x18, 11, 6),
+	MLXSW_AFK_ELEMENT_INFO_U32(VIRT_ROUTER_MSB, 0x18, 17, 4),
+	MLXSW_AFK_ELEMENT_INFO_U32(VIRT_ROUTER_LSB, 0x18, 21, 8),
+	MLXSW_AFK_ELEMENT_INFO_BUF(SRC_IP_96_127, 0x20, 4),
+	MLXSW_AFK_ELEMENT_INFO_BUF(SRC_IP_64_95, 0x24, 4),
+	MLXSW_AFK_ELEMENT_INFO_BUF(SRC_IP_32_63, 0x28, 4),
+	MLXSW_AFK_ELEMENT_INFO_BUF(SRC_IP_0_31, 0x2C, 4),
+	MLXSW_AFK_ELEMENT_INFO_BUF(DST_IP_96_127, 0x30, 4),
+	MLXSW_AFK_ELEMENT_INFO_BUF(DST_IP_64_95, 0x34, 4),
+	MLXSW_AFK_ELEMENT_INFO_BUF(DST_IP_32_63, 0x38, 4),
+	MLXSW_AFK_ELEMENT_INFO_BUF(DST_IP_0_31, 0x3C, 4),
+};
+
+struct mlxsw_afk {
+	struct list_head key_info_list;
+	unsigned int max_blocks;
+	const struct mlxsw_afk_ops *ops;
+	const struct mlxsw_afk_block *blocks;
+	unsigned int blocks_count;
+};
+
+static bool mlxsw_afk_blocks_check(struct mlxsw_afk *mlxsw_afk)
+{
+	int i;
+	int j;
+
+	for (i = 0; i < mlxsw_afk->blocks_count; i++) {
+		const struct mlxsw_afk_block *block = &mlxsw_afk->blocks[i];
+
+		for (j = 0; j < block->instances_count; j++) {
+			const struct mlxsw_afk_element_info *elinfo;
+			struct mlxsw_afk_element_inst *elinst;
+
+			elinst = &block->instances[j];
+			elinfo = &mlxsw_afk_element_infos[elinst->element];
+			if (elinst->type != elinfo->type ||
+			    (!elinst->avoid_size_check &&
+			     elinst->item.size.bits !=
+			     elinfo->item.size.bits))
+				return false;
+		}
+	}
+	return true;
+}
+
+struct mlxsw_afk *mlxsw_afk_create(unsigned int max_blocks,
+				   const struct mlxsw_afk_ops *ops)
+{
+	struct mlxsw_afk *mlxsw_afk;
+
+	mlxsw_afk = kzalloc(sizeof(*mlxsw_afk), GFP_KERNEL);
+	if (!mlxsw_afk)
+		return NULL;
+	INIT_LIST_HEAD(&mlxsw_afk->key_info_list);
+	mlxsw_afk->max_blocks = max_blocks;
+	mlxsw_afk->ops = ops;
+	mlxsw_afk->blocks = ops->blocks;
+	mlxsw_afk->blocks_count = ops->blocks_count;
+	WARN_ON(!mlxsw_afk_blocks_check(mlxsw_afk));
+	return mlxsw_afk;
+}
+EXPORT_SYMBOL(mlxsw_afk_create);
+
+void mlxsw_afk_destroy(struct mlxsw_afk *mlxsw_afk)
+{
+	WARN_ON(!list_empty(&mlxsw_afk->key_info_list));
+	kfree(mlxsw_afk);
+}
+EXPORT_SYMBOL(mlxsw_afk_destroy);
+
+struct mlxsw_afk_key_info {
+	struct list_head list;
+	unsigned int ref_count;
+	unsigned int blocks_count;
+	int element_to_block[MLXSW_AFK_ELEMENT_MAX]; /* index is element, value
+						      * is index inside "blocks"
+						      */
+	struct mlxsw_afk_element_usage elusage;
+	const struct mlxsw_afk_block *blocks[];
+};
+
+static bool
+mlxsw_afk_key_info_elements_eq(struct mlxsw_afk_key_info *key_info,
+			       struct mlxsw_afk_element_usage *elusage)
+{
+	return memcmp(&key_info->elusage, elusage, sizeof(*elusage)) == 0;
+}
+
+static struct mlxsw_afk_key_info *
+mlxsw_afk_key_info_find(struct mlxsw_afk *mlxsw_afk,
+			struct mlxsw_afk_element_usage *elusage)
+{
+	struct mlxsw_afk_key_info *key_info;
+
+	list_for_each_entry(key_info, &mlxsw_afk->key_info_list, list) {
+		if (mlxsw_afk_key_info_elements_eq(key_info, elusage))
+			return key_info;
+	}
+	return NULL;
+}
+
+struct mlxsw_afk_picker {
+	DECLARE_BITMAP(element, MLXSW_AFK_ELEMENT_MAX);
+	unsigned int total;
+};
+
+static void mlxsw_afk_picker_count_hits(struct mlxsw_afk *mlxsw_afk,
+					struct mlxsw_afk_picker *picker,
+					enum mlxsw_afk_element element)
+{
+	int i;
+	int j;
+
+	for (i = 0; i < mlxsw_afk->blocks_count; i++) {
+		const struct mlxsw_afk_block *block = &mlxsw_afk->blocks[i];
+
+		for (j = 0; j < block->instances_count; j++) {
+			struct mlxsw_afk_element_inst *elinst;
+
+			elinst = &block->instances[j];
+			if (elinst->element == element) {
+				__set_bit(element, picker[i].element);
+				picker[i].total++;
+			}
+		}
+	}
+}
+
+static void mlxsw_afk_picker_subtract_hits(struct mlxsw_afk *mlxsw_afk,
+					   struct mlxsw_afk_picker *picker,
+					   int block_index)
+{
+	DECLARE_BITMAP(hits_element, MLXSW_AFK_ELEMENT_MAX);
+	int i;
+	int j;
+
+	memcpy(&hits_element, &picker[block_index].element,
+	       sizeof(hits_element));
+
+	for (i = 0; i < mlxsw_afk->blocks_count; i++) {
+		for_each_set_bit(j, hits_element, MLXSW_AFK_ELEMENT_MAX) {
+			if (__test_and_clear_bit(j, picker[i].element))
+				picker[i].total--;
+		}
+	}
+}
+
+static int mlxsw_afk_picker_most_hits_get(struct mlxsw_afk *mlxsw_afk,
+					  struct mlxsw_afk_picker *picker)
+{
+	int most_index = -EINVAL; /* Should never happen to return this */
+	int most_hits = 0;
+	int i;
+
+	for (i = 0; i < mlxsw_afk->blocks_count; i++) {
+		if (picker[i].total > most_hits) {
+			most_hits = picker[i].total;
+			most_index = i;
+		}
+	}
+	return most_index;
+}
+
+static int mlxsw_afk_picker_key_info_add(struct mlxsw_afk *mlxsw_afk,
+					 struct mlxsw_afk_picker *picker,
+					 int block_index,
+					 struct mlxsw_afk_key_info *key_info)
+{
+	enum mlxsw_afk_element element;
+
+	if (key_info->blocks_count == mlxsw_afk->max_blocks)
+		return -EINVAL;
+
+	for_each_set_bit(element, picker[block_index].element,
+			 MLXSW_AFK_ELEMENT_MAX) {
+		key_info->element_to_block[element] = key_info->blocks_count;
+		mlxsw_afk_element_usage_add(&key_info->elusage, element);
+	}
+
+	key_info->blocks[key_info->blocks_count] =
+					&mlxsw_afk->blocks[block_index];
+	key_info->blocks_count++;
+	return 0;
+}
+
+static int mlxsw_afk_picker(struct mlxsw_afk *mlxsw_afk,
+			    struct mlxsw_afk_key_info *key_info,
+			    struct mlxsw_afk_element_usage *elusage)
+{
+	struct mlxsw_afk_picker *picker;
+	enum mlxsw_afk_element element;
+	int err;
+
+	picker = kcalloc(mlxsw_afk->blocks_count, sizeof(*picker), GFP_KERNEL);
+	if (!picker)
+		return -ENOMEM;
+
+	/* Since the same elements could be present in multiple blocks,
+	 * we must find out optimal block list in order to make the
+	 * block count as low as possible.
+	 *
+	 * First, we count hits. We go over all available blocks and count
+	 * how many of requested elements are covered by each.
+	 *
+	 * Then in loop, we find block with most hits and add it to
+	 * output key_info. Then we have to subtract this block hits so
+	 * the next iteration will find most suitable block for
+	 * the rest of requested elements.
+	 */
+
+	mlxsw_afk_element_usage_for_each(element, elusage)
+		mlxsw_afk_picker_count_hits(mlxsw_afk, picker, element);
+
+	do {
+		int block_index;
+
+		block_index = mlxsw_afk_picker_most_hits_get(mlxsw_afk, picker);
+		if (block_index < 0) {
+			err = block_index;
+			goto out;
+		}
+		err = mlxsw_afk_picker_key_info_add(mlxsw_afk, picker,
+						    block_index, key_info);
+		if (err)
+			goto out;
+		mlxsw_afk_picker_subtract_hits(mlxsw_afk, picker, block_index);
+	} while (!mlxsw_afk_key_info_elements_eq(key_info, elusage));
+
+	err = 0;
+out:
+	kfree(picker);
+	return err;
+}
+
+static struct mlxsw_afk_key_info *
+mlxsw_afk_key_info_create(struct mlxsw_afk *mlxsw_afk,
+			  struct mlxsw_afk_element_usage *elusage)
+{
+	struct mlxsw_afk_key_info *key_info;
+	int err;
+
+	key_info = kzalloc(struct_size(key_info, blocks, mlxsw_afk->max_blocks),
+			   GFP_KERNEL);
+	if (!key_info)
+		return ERR_PTR(-ENOMEM);
+	err = mlxsw_afk_picker(mlxsw_afk, key_info, elusage);
+	if (err)
+		goto err_picker;
+	list_add(&key_info->list, &mlxsw_afk->key_info_list);
+	key_info->ref_count = 1;
+	return key_info;
+
+err_picker:
+	kfree(key_info);
+	return ERR_PTR(err);
+}
+
+static void mlxsw_afk_key_info_destroy(struct mlxsw_afk_key_info *key_info)
+{
+	list_del(&key_info->list);
+	kfree(key_info);
+}
+
+struct mlxsw_afk_key_info *
+mlxsw_afk_key_info_get(struct mlxsw_afk *mlxsw_afk,
+		       struct mlxsw_afk_element_usage *elusage)
+{
+	struct mlxsw_afk_key_info *key_info;
+
+	key_info = mlxsw_afk_key_info_find(mlxsw_afk, elusage);
+	if (key_info) {
+		key_info->ref_count++;
+		return key_info;
+	}
+	return mlxsw_afk_key_info_create(mlxsw_afk, elusage);
+}
+EXPORT_SYMBOL(mlxsw_afk_key_info_get);
+
+void mlxsw_afk_key_info_put(struct mlxsw_afk_key_info *key_info)
+{
+	if (--key_info->ref_count)
+		return;
+	mlxsw_afk_key_info_destroy(key_info);
+}
+EXPORT_SYMBOL(mlxsw_afk_key_info_put);
+
+bool mlxsw_afk_key_info_subset(struct mlxsw_afk_key_info *key_info,
+			       struct mlxsw_afk_element_usage *elusage)
+{
+	return mlxsw_afk_element_usage_subset(elusage, &key_info->elusage);
+}
+EXPORT_SYMBOL(mlxsw_afk_key_info_subset);
+
+static const struct mlxsw_afk_element_inst *
+mlxsw_afk_block_elinst_get(const struct mlxsw_afk_block *block,
+			   enum mlxsw_afk_element element)
+{
+	int i;
+
+	for (i = 0; i < block->instances_count; i++) {
+		struct mlxsw_afk_element_inst *elinst;
+
+		elinst = &block->instances[i];
+		if (elinst->element == element)
+			return elinst;
+	}
+	return NULL;
+}
+
+static const struct mlxsw_afk_element_inst *
+mlxsw_afk_key_info_elinst_get(struct mlxsw_afk_key_info *key_info,
+			      enum mlxsw_afk_element element,
+			      int *p_block_index)
+{
+	const struct mlxsw_afk_element_inst *elinst;
+	const struct mlxsw_afk_block *block;
+	int block_index;
+
+	if (WARN_ON(!test_bit(element, key_info->elusage.usage)))
+		return NULL;
+	block_index = key_info->element_to_block[element];
+	block = key_info->blocks[block_index];
+
+	elinst = mlxsw_afk_block_elinst_get(block, element);
+	if (WARN_ON(!elinst))
+		return NULL;
+
+	*p_block_index = block_index;
+	return elinst;
+}
+
+u16
+mlxsw_afk_key_info_block_encoding_get(const struct mlxsw_afk_key_info *key_info,
+				      int block_index)
+{
+	return key_info->blocks[block_index]->encoding;
+}
+EXPORT_SYMBOL(mlxsw_afk_key_info_block_encoding_get);
+
+unsigned int
+mlxsw_afk_key_info_blocks_count_get(const struct mlxsw_afk_key_info *key_info)
+{
+	return key_info->blocks_count;
+}
+EXPORT_SYMBOL(mlxsw_afk_key_info_blocks_count_get);
+
+void mlxsw_afk_values_add_u32(struct mlxsw_afk_element_values *values,
+			      enum mlxsw_afk_element element,
+			      u32 key_value, u32 mask_value)
+{
+	const struct mlxsw_afk_element_info *elinfo =
+				&mlxsw_afk_element_infos[element];
+	const struct mlxsw_item *storage_item = &elinfo->item;
+
+	if (!mask_value)
+		return;
+	if (WARN_ON(elinfo->type != MLXSW_AFK_ELEMENT_TYPE_U32))
+		return;
+	__mlxsw_item_set32(values->storage.key, storage_item, 0, key_value);
+	__mlxsw_item_set32(values->storage.mask, storage_item, 0, mask_value);
+	mlxsw_afk_element_usage_add(&values->elusage, element);
+}
+EXPORT_SYMBOL(mlxsw_afk_values_add_u32);
+
+void mlxsw_afk_values_add_buf(struct mlxsw_afk_element_values *values,
+			      enum mlxsw_afk_element element,
+			      const char *key_value, const char *mask_value,
+			      unsigned int len)
+{
+	const struct mlxsw_afk_element_info *elinfo =
+				&mlxsw_afk_element_infos[element];
+	const struct mlxsw_item *storage_item = &elinfo->item;
+
+	if (!memchr_inv(mask_value, 0, len)) /* If mask is zero */
+		return;
+	if (WARN_ON(elinfo->type != MLXSW_AFK_ELEMENT_TYPE_BUF) ||
+	    WARN_ON(elinfo->item.size.bytes != len))
+		return;
+	__mlxsw_item_memcpy_to(values->storage.key, key_value,
+			       storage_item, 0);
+	__mlxsw_item_memcpy_to(values->storage.mask, mask_value,
+			       storage_item, 0);
+	mlxsw_afk_element_usage_add(&values->elusage, element);
+}
+EXPORT_SYMBOL(mlxsw_afk_values_add_buf);
+
+static void mlxsw_sp_afk_encode_u32(const struct mlxsw_item *storage_item,
+				    const struct mlxsw_item *output_item,
+				    char *storage, char *output, int diff)
+{
+	u32 value;
+
+	value = __mlxsw_item_get32(storage, storage_item, 0);
+	__mlxsw_item_set32(output, output_item, 0, value + diff);
+}
+
+static void mlxsw_sp_afk_encode_buf(const struct mlxsw_item *storage_item,
+				    const struct mlxsw_item *output_item,
+				    char *storage, char *output)
+{
+	char *storage_data = __mlxsw_item_data(storage, storage_item, 0);
+	char *output_data = __mlxsw_item_data(output, output_item, 0);
+	size_t len = output_item->size.bytes;
+
+	memcpy(output_data, storage_data, len);
+}
+
+static void
+mlxsw_sp_afk_encode_one(const struct mlxsw_afk_element_inst *elinst,
+			char *output, char *storage, int u32_diff)
+{
+	const struct mlxsw_item *output_item = &elinst->item;
+	const struct mlxsw_afk_element_info *elinfo;
+	const struct mlxsw_item *storage_item;
+
+	elinfo = &mlxsw_afk_element_infos[elinst->element];
+	storage_item = &elinfo->item;
+	if (elinst->type == MLXSW_AFK_ELEMENT_TYPE_U32)
+		mlxsw_sp_afk_encode_u32(storage_item, output_item,
+					storage, output, u32_diff);
+	else if (elinst->type == MLXSW_AFK_ELEMENT_TYPE_BUF)
+		mlxsw_sp_afk_encode_buf(storage_item, output_item,
+					storage, output);
+}
+
+#define MLXSW_SP_AFK_KEY_BLOCK_MAX_SIZE 16
+
+void mlxsw_afk_encode(struct mlxsw_afk *mlxsw_afk,
+		      struct mlxsw_afk_key_info *key_info,
+		      struct mlxsw_afk_element_values *values,
+		      char *key, char *mask)
+{
+	unsigned int blocks_count =
+			mlxsw_afk_key_info_blocks_count_get(key_info);
+	char block_mask[MLXSW_SP_AFK_KEY_BLOCK_MAX_SIZE];
+	char block_key[MLXSW_SP_AFK_KEY_BLOCK_MAX_SIZE];
+	const struct mlxsw_afk_element_inst *elinst;
+	enum mlxsw_afk_element element;
+	int block_index, i;
+
+	for (i = 0; i < blocks_count; i++) {
+		memset(block_key, 0, MLXSW_SP_AFK_KEY_BLOCK_MAX_SIZE);
+		memset(block_mask, 0, MLXSW_SP_AFK_KEY_BLOCK_MAX_SIZE);
+
+		mlxsw_afk_element_usage_for_each(element, &values->elusage) {
+			elinst = mlxsw_afk_key_info_elinst_get(key_info,
+							       element,
+							       &block_index);
+			if (!elinst || block_index != i)
+				continue;
+
+			mlxsw_sp_afk_encode_one(elinst, block_key,
+						values->storage.key,
+						elinst->u32_key_diff);
+			mlxsw_sp_afk_encode_one(elinst, block_mask,
+						values->storage.mask, 0);
+		}
+
+		mlxsw_afk->ops->encode_block(key, i, block_key);
+		mlxsw_afk->ops->encode_block(mask, i, block_mask);
+	}
+}
+EXPORT_SYMBOL(mlxsw_afk_encode);
+
+void mlxsw_afk_clear(struct mlxsw_afk *mlxsw_afk, char *key,
+		     int block_start, int block_end)
+{
+	int i;
+
+	for (i = block_start; i <= block_end; i++)
+		mlxsw_afk->ops->clear_block(key, i);
+}
+EXPORT_SYMBOL(mlxsw_afk_clear);
diff --git a/drivers/net/ethernet/mellanox/mlxsw/core_acl_flex_keys.h b/drivers/net/ethernet/mellanox/mlxsw/core_acl_flex_keys.h
new file mode 100644
index 000000000..3a037fe47
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/core_acl_flex_keys.h
@@ -0,0 +1,222 @@
+/* SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0 */
+/* Copyright (c) 2017-2018 Mellanox Technologies. All rights reserved */
+
+#ifndef _MLXSW_CORE_ACL_FLEX_KEYS_H
+#define _MLXSW_CORE_ACL_FLEX_KEYS_H
+
+#include <linux/types.h>
+#include <linux/bitmap.h>
+
+#include "item.h"
+
+enum mlxsw_afk_element {
+	MLXSW_AFK_ELEMENT_SRC_SYS_PORT,
+	MLXSW_AFK_ELEMENT_DMAC_32_47,
+	MLXSW_AFK_ELEMENT_DMAC_0_31,
+	MLXSW_AFK_ELEMENT_SMAC_32_47,
+	MLXSW_AFK_ELEMENT_SMAC_0_31,
+	MLXSW_AFK_ELEMENT_ETHERTYPE,
+	MLXSW_AFK_ELEMENT_IP_PROTO,
+	MLXSW_AFK_ELEMENT_SRC_IP_96_127,
+	MLXSW_AFK_ELEMENT_SRC_IP_64_95,
+	MLXSW_AFK_ELEMENT_SRC_IP_32_63,
+	MLXSW_AFK_ELEMENT_SRC_IP_0_31,
+	MLXSW_AFK_ELEMENT_DST_IP_96_127,
+	MLXSW_AFK_ELEMENT_DST_IP_64_95,
+	MLXSW_AFK_ELEMENT_DST_IP_32_63,
+	MLXSW_AFK_ELEMENT_DST_IP_0_31,
+	MLXSW_AFK_ELEMENT_DST_L4_PORT,
+	MLXSW_AFK_ELEMENT_SRC_L4_PORT,
+	MLXSW_AFK_ELEMENT_VID,
+	MLXSW_AFK_ELEMENT_PCP,
+	MLXSW_AFK_ELEMENT_TCP_FLAGS,
+	MLXSW_AFK_ELEMENT_IP_TTL_,
+	MLXSW_AFK_ELEMENT_IP_ECN,
+	MLXSW_AFK_ELEMENT_IP_DSCP,
+	MLXSW_AFK_ELEMENT_VIRT_ROUTER_MSB,
+	MLXSW_AFK_ELEMENT_VIRT_ROUTER_LSB,
+	MLXSW_AFK_ELEMENT_MAX,
+};
+
+enum mlxsw_afk_element_type {
+	MLXSW_AFK_ELEMENT_TYPE_U32,
+	MLXSW_AFK_ELEMENT_TYPE_BUF,
+};
+
+struct mlxsw_afk_element_info {
+	enum mlxsw_afk_element element; /* element ID */
+	enum mlxsw_afk_element_type type;
+	struct mlxsw_item item; /* element geometry in internal storage */
+};
+
+#define MLXSW_AFK_ELEMENT_INFO(_type, _element, _offset, _shift, _size)		\
+	[MLXSW_AFK_ELEMENT_##_element] = {					\
+		.element = MLXSW_AFK_ELEMENT_##_element,			\
+		.type = _type,							\
+		.item = {							\
+			.offset = _offset,					\
+			.shift = _shift,					\
+			.size = {.bits = _size},				\
+			.name = #_element,					\
+		},								\
+	}
+
+#define MLXSW_AFK_ELEMENT_INFO_U32(_element, _offset, _shift, _size)		\
+	MLXSW_AFK_ELEMENT_INFO(MLXSW_AFK_ELEMENT_TYPE_U32,			\
+			       _element, _offset, _shift, _size)
+
+#define MLXSW_AFK_ELEMENT_INFO_BUF(_element, _offset, _size)			\
+	MLXSW_AFK_ELEMENT_INFO(MLXSW_AFK_ELEMENT_TYPE_BUF,			\
+			       _element, _offset, 0, _size)
+
+#define MLXSW_AFK_ELEMENT_STORAGE_SIZE 0x40
+
+struct mlxsw_afk_element_inst { /* element instance in actual block */
+	enum mlxsw_afk_element element;
+	enum mlxsw_afk_element_type type;
+	struct mlxsw_item item; /* element geometry in block */
+	int u32_key_diff; /* in case value needs to be adjusted before write
+			   * this diff is here to handle that
+			   */
+	bool avoid_size_check;
+};
+
+#define MLXSW_AFK_ELEMENT_INST(_type, _element, _offset,			\
+			       _shift, _size, _u32_key_diff, _avoid_size_check)	\
+	{									\
+		.element = MLXSW_AFK_ELEMENT_##_element,			\
+		.type = _type,							\
+		.item = {							\
+			.offset = _offset,					\
+			.shift = _shift,					\
+			.size = {.bits = _size},				\
+			.name = #_element,					\
+		},								\
+		.u32_key_diff = _u32_key_diff,					\
+		.avoid_size_check = _avoid_size_check,				\
+	}
+
+#define MLXSW_AFK_ELEMENT_INST_U32(_element, _offset, _shift, _size)		\
+	MLXSW_AFK_ELEMENT_INST(MLXSW_AFK_ELEMENT_TYPE_U32,			\
+			       _element, _offset, _shift, _size, 0, false)
+
+#define MLXSW_AFK_ELEMENT_INST_EXT_U32(_element, _offset,			\
+				       _shift, _size, _key_diff,		\
+				       _avoid_size_check)			\
+	MLXSW_AFK_ELEMENT_INST(MLXSW_AFK_ELEMENT_TYPE_U32,			\
+			       _element, _offset, _shift, _size,		\
+			       _key_diff, _avoid_size_check)
+
+#define MLXSW_AFK_ELEMENT_INST_BUF(_element, _offset, _size)			\
+	MLXSW_AFK_ELEMENT_INST(MLXSW_AFK_ELEMENT_TYPE_BUF,			\
+			       _element, _offset, 0, _size, 0, false)
+
+struct mlxsw_afk_block {
+	u16 encoding; /* block ID */
+	struct mlxsw_afk_element_inst *instances;
+	unsigned int instances_count;
+};
+
+#define MLXSW_AFK_BLOCK(_encoding, _instances)					\
+	{									\
+		.encoding = _encoding,						\
+		.instances = _instances,					\
+		.instances_count = ARRAY_SIZE(_instances),			\
+	}
+
+struct mlxsw_afk_element_usage {
+	DECLARE_BITMAP(usage, MLXSW_AFK_ELEMENT_MAX);
+};
+
+#define mlxsw_afk_element_usage_for_each(element, elusage)			\
+	for_each_set_bit(element, (elusage)->usage, MLXSW_AFK_ELEMENT_MAX)
+
+static inline void
+mlxsw_afk_element_usage_add(struct mlxsw_afk_element_usage *elusage,
+			    enum mlxsw_afk_element element)
+{
+	__set_bit(element, elusage->usage);
+}
+
+static inline void
+mlxsw_afk_element_usage_zero(struct mlxsw_afk_element_usage *elusage)
+{
+	bitmap_zero(elusage->usage, MLXSW_AFK_ELEMENT_MAX);
+}
+
+static inline void
+mlxsw_afk_element_usage_fill(struct mlxsw_afk_element_usage *elusage,
+			     const enum mlxsw_afk_element *elements,
+			     unsigned int elements_count)
+{
+	int i;
+
+	mlxsw_afk_element_usage_zero(elusage);
+	for (i = 0; i < elements_count; i++)
+		mlxsw_afk_element_usage_add(elusage, elements[i]);
+}
+
+static inline bool
+mlxsw_afk_element_usage_subset(struct mlxsw_afk_element_usage *elusage_small,
+			       struct mlxsw_afk_element_usage *elusage_big)
+{
+	int i;
+
+	for (i = 0; i < MLXSW_AFK_ELEMENT_MAX; i++)
+		if (test_bit(i, elusage_small->usage) &&
+		    !test_bit(i, elusage_big->usage))
+			return false;
+	return true;
+}
+
+struct mlxsw_afk;
+
+struct mlxsw_afk_ops {
+	const struct mlxsw_afk_block *blocks;
+	unsigned int blocks_count;
+	void (*encode_block)(char *output, int block_index, char *block);
+	void (*clear_block)(char *output, int block_index);
+};
+
+struct mlxsw_afk *mlxsw_afk_create(unsigned int max_blocks,
+				   const struct mlxsw_afk_ops *ops);
+void mlxsw_afk_destroy(struct mlxsw_afk *mlxsw_afk);
+
+struct mlxsw_afk_key_info;
+
+struct mlxsw_afk_key_info *
+mlxsw_afk_key_info_get(struct mlxsw_afk *mlxsw_afk,
+		       struct mlxsw_afk_element_usage *elusage);
+void mlxsw_afk_key_info_put(struct mlxsw_afk_key_info *key_info);
+bool mlxsw_afk_key_info_subset(struct mlxsw_afk_key_info *key_info,
+			       struct mlxsw_afk_element_usage *elusage);
+
+u16
+mlxsw_afk_key_info_block_encoding_get(const struct mlxsw_afk_key_info *key_info,
+				      int block_index);
+unsigned int
+mlxsw_afk_key_info_blocks_count_get(const struct mlxsw_afk_key_info *key_info);
+
+struct mlxsw_afk_element_values {
+	struct mlxsw_afk_element_usage elusage;
+	struct {
+		char key[MLXSW_AFK_ELEMENT_STORAGE_SIZE];
+		char mask[MLXSW_AFK_ELEMENT_STORAGE_SIZE];
+	} storage;
+};
+
+void mlxsw_afk_values_add_u32(struct mlxsw_afk_element_values *values,
+			      enum mlxsw_afk_element element,
+			      u32 key_value, u32 mask_value);
+void mlxsw_afk_values_add_buf(struct mlxsw_afk_element_values *values,
+			      enum mlxsw_afk_element element,
+			      const char *key_value, const char *mask_value,
+			      unsigned int len);
+void mlxsw_afk_encode(struct mlxsw_afk *mlxsw_afk,
+		      struct mlxsw_afk_key_info *key_info,
+		      struct mlxsw_afk_element_values *values,
+		      char *key, char *mask);
+void mlxsw_afk_clear(struct mlxsw_afk *mlxsw_afk, char *key,
+		     int block_start, int block_end);
+
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlxsw/core_env.c b/drivers/net/ethernet/mellanox/mlxsw/core_env.c
new file mode 100644
index 000000000..0107cbc32
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/core_env.c
@@ -0,0 +1,1464 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2018 Mellanox Technologies. All rights reserved */
+
+#include <linux/kernel.h>
+#include <linux/err.h>
+#include <linux/ethtool.h>
+#include <linux/sfp.h>
+#include <linux/mutex.h>
+
+#include "core.h"
+#include "core_env.h"
+#include "item.h"
+#include "reg.h"
+
+struct mlxsw_env_module_info {
+	u64 module_overheat_counter;
+	bool is_overheat;
+	int num_ports_mapped;
+	int num_ports_up;
+	enum ethtool_module_power_mode_policy power_mode_policy;
+	enum mlxsw_reg_pmtm_module_type type;
+};
+
+struct mlxsw_env_line_card {
+	u8 module_count;
+	bool active;
+	struct mlxsw_env_module_info module_info[];
+};
+
+struct mlxsw_env {
+	struct mlxsw_core *core;
+	const struct mlxsw_bus_info *bus_info;
+	u8 max_module_count; /* Maximum number of modules per-slot. */
+	u8 num_of_slots; /* Including the main board. */
+	struct mutex line_cards_lock; /* Protects line cards. */
+	struct mlxsw_env_line_card *line_cards[];
+};
+
+static bool __mlxsw_env_linecard_is_active(struct mlxsw_env *mlxsw_env,
+					   u8 slot_index)
+{
+	return mlxsw_env->line_cards[slot_index]->active;
+}
+
+static bool mlxsw_env_linecard_is_active(struct mlxsw_env *mlxsw_env,
+					 u8 slot_index)
+{
+	bool active;
+
+	mutex_lock(&mlxsw_env->line_cards_lock);
+	active = __mlxsw_env_linecard_is_active(mlxsw_env, slot_index);
+	mutex_unlock(&mlxsw_env->line_cards_lock);
+
+	return active;
+}
+
+static struct
+mlxsw_env_module_info *mlxsw_env_module_info_get(struct mlxsw_core *mlxsw_core,
+						 u8 slot_index, u8 module)
+{
+	struct mlxsw_env *mlxsw_env = mlxsw_core_env(mlxsw_core);
+
+	return &mlxsw_env->line_cards[slot_index]->module_info[module];
+}
+
+static int __mlxsw_env_validate_module_type(struct mlxsw_core *core,
+					    u8 slot_index, u8 module)
+{
+	struct mlxsw_env *mlxsw_env = mlxsw_core_env(core);
+	struct mlxsw_env_module_info *module_info;
+	int err;
+
+	if (!__mlxsw_env_linecard_is_active(mlxsw_env, slot_index))
+		return 0;
+
+	module_info = mlxsw_env_module_info_get(core, slot_index, module);
+	switch (module_info->type) {
+	case MLXSW_REG_PMTM_MODULE_TYPE_TWISTED_PAIR:
+		err = -EINVAL;
+		break;
+	default:
+		err = 0;
+	}
+
+	return err;
+}
+
+static int mlxsw_env_validate_module_type(struct mlxsw_core *core,
+					  u8 slot_index, u8 module)
+{
+	struct mlxsw_env *mlxsw_env = mlxsw_core_env(core);
+	int err;
+
+	mutex_lock(&mlxsw_env->line_cards_lock);
+	err = __mlxsw_env_validate_module_type(core, slot_index, module);
+	mutex_unlock(&mlxsw_env->line_cards_lock);
+
+	return err;
+}
+
+static int
+mlxsw_env_validate_cable_ident(struct mlxsw_core *core, u8 slot_index, int id,
+			       bool *qsfp, bool *cmis)
+{
+	char mcia_pl[MLXSW_REG_MCIA_LEN];
+	char *eeprom_tmp;
+	u8 ident;
+	int err;
+
+	err = mlxsw_env_validate_module_type(core, slot_index, id);
+	if (err)
+		return err;
+
+	mlxsw_reg_mcia_pack(mcia_pl, slot_index, id, 0,
+			    MLXSW_REG_MCIA_PAGE0_LO_OFF, 0, 1,
+			    MLXSW_REG_MCIA_I2C_ADDR_LOW);
+	err = mlxsw_reg_query(core, MLXSW_REG(mcia), mcia_pl);
+	if (err)
+		return err;
+	eeprom_tmp = mlxsw_reg_mcia_eeprom_data(mcia_pl);
+	ident = eeprom_tmp[0];
+	*cmis = false;
+	switch (ident) {
+	case MLXSW_REG_MCIA_EEPROM_MODULE_INFO_ID_SFP:
+		*qsfp = false;
+		break;
+	case MLXSW_REG_MCIA_EEPROM_MODULE_INFO_ID_QSFP:
+	case MLXSW_REG_MCIA_EEPROM_MODULE_INFO_ID_QSFP_PLUS:
+	case MLXSW_REG_MCIA_EEPROM_MODULE_INFO_ID_QSFP28:
+		*qsfp = true;
+		break;
+	case MLXSW_REG_MCIA_EEPROM_MODULE_INFO_ID_QSFP_DD:
+	case MLXSW_REG_MCIA_EEPROM_MODULE_INFO_ID_OSFP:
+		*qsfp = true;
+		*cmis = true;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int
+mlxsw_env_query_module_eeprom(struct mlxsw_core *mlxsw_core, u8 slot_index,
+			      int module, u16 offset, u16 size, void *data,
+			      bool qsfp, unsigned int *p_read_size)
+{
+	char mcia_pl[MLXSW_REG_MCIA_LEN];
+	char *eeprom_tmp;
+	u16 i2c_addr;
+	u8 page = 0;
+	int status;
+	int err;
+
+	/* MCIA register accepts buffer size <= 48. Page of size 128 should be
+	 * read by chunks of size 48, 48, 32. Align the size of the last chunk
+	 * to avoid reading after the end of the page.
+	 */
+	size = min_t(u16, size, MLXSW_REG_MCIA_EEPROM_SIZE);
+
+	if (offset < MLXSW_REG_MCIA_EEPROM_PAGE_LENGTH &&
+	    offset + size > MLXSW_REG_MCIA_EEPROM_PAGE_LENGTH)
+		/* Cross pages read, read until offset 256 in low page */
+		size = MLXSW_REG_MCIA_EEPROM_PAGE_LENGTH - offset;
+
+	i2c_addr = MLXSW_REG_MCIA_I2C_ADDR_LOW;
+	if (offset >= MLXSW_REG_MCIA_EEPROM_PAGE_LENGTH) {
+		if (qsfp) {
+			/* When reading upper pages 1, 2 and 3 the offset
+			 * starts at 128. Please refer to "QSFP+ Memory Map"
+			 * figure in SFF-8436 specification and to "CMIS Module
+			 * Memory Map" figure in CMIS specification for
+			 * graphical depiction.
+			 */
+			page = MLXSW_REG_MCIA_PAGE_GET(offset);
+			offset -= MLXSW_REG_MCIA_EEPROM_UP_PAGE_LENGTH * page;
+			if (offset + size > MLXSW_REG_MCIA_EEPROM_PAGE_LENGTH)
+				size = MLXSW_REG_MCIA_EEPROM_PAGE_LENGTH - offset;
+		} else {
+			/* When reading upper pages 1, 2 and 3 the offset
+			 * starts at 0 and I2C high address is used. Please refer
+			 * to "Memory Organization" figure in SFF-8472
+			 * specification for graphical depiction.
+			 */
+			i2c_addr = MLXSW_REG_MCIA_I2C_ADDR_HIGH;
+			offset -= MLXSW_REG_MCIA_EEPROM_PAGE_LENGTH;
+		}
+	}
+
+	mlxsw_reg_mcia_pack(mcia_pl, slot_index, module, 0, page, offset, size,
+			    i2c_addr);
+
+	err = mlxsw_reg_query(mlxsw_core, MLXSW_REG(mcia), mcia_pl);
+	if (err)
+		return err;
+
+	status = mlxsw_reg_mcia_status_get(mcia_pl);
+	if (status)
+		return -EIO;
+
+	eeprom_tmp = mlxsw_reg_mcia_eeprom_data(mcia_pl);
+	memcpy(data, eeprom_tmp, size);
+	*p_read_size = size;
+
+	return 0;
+}
+
+int
+mlxsw_env_module_temp_thresholds_get(struct mlxsw_core *core, u8 slot_index,
+				     int module, int off, int *temp)
+{
+	unsigned int module_temp, module_crit, module_emerg;
+	union {
+		u8 buf[MLXSW_REG_MCIA_TH_ITEM_SIZE];
+		u16 temp;
+	} temp_thresh;
+	char mcia_pl[MLXSW_REG_MCIA_LEN] = {0};
+	char mtmp_pl[MLXSW_REG_MTMP_LEN];
+	char *eeprom_tmp;
+	bool qsfp, cmis;
+	int page;
+	int err;
+
+	mlxsw_reg_mtmp_pack(mtmp_pl, slot_index,
+			    MLXSW_REG_MTMP_MODULE_INDEX_MIN + module, false,
+			    false);
+	err = mlxsw_reg_query(core, MLXSW_REG(mtmp), mtmp_pl);
+	if (err)
+		return err;
+	mlxsw_reg_mtmp_unpack(mtmp_pl, &module_temp, NULL, &module_crit,
+			      &module_emerg, NULL);
+	if (!module_temp) {
+		*temp = 0;
+		return 0;
+	}
+
+	/* Validate if threshold reading is available through MTMP register,
+	 * otherwise fallback to read through MCIA.
+	 */
+	if (module_emerg) {
+		*temp = off == SFP_TEMP_HIGH_WARN ? module_crit : module_emerg;
+		return 0;
+	}
+
+	/* Read Free Side Device Temperature Thresholds from page 03h
+	 * (MSB at lower byte address).
+	 * Bytes:
+	 * 128-129 - Temp High Alarm (SFP_TEMP_HIGH_ALARM);
+	 * 130-131 - Temp Low Alarm (SFP_TEMP_LOW_ALARM);
+	 * 132-133 - Temp High Warning (SFP_TEMP_HIGH_WARN);
+	 * 134-135 - Temp Low Warning (SFP_TEMP_LOW_WARN);
+	 */
+
+	/* Validate module identifier value. */
+	err = mlxsw_env_validate_cable_ident(core, slot_index, module, &qsfp,
+					     &cmis);
+	if (err)
+		return err;
+
+	if (qsfp) {
+		/* For QSFP/CMIS module-defined thresholds are located in page
+		 * 02h, otherwise in page 03h.
+		 */
+		if (cmis)
+			page = MLXSW_REG_MCIA_TH_PAGE_CMIS_NUM;
+		else
+			page = MLXSW_REG_MCIA_TH_PAGE_NUM;
+		mlxsw_reg_mcia_pack(mcia_pl, slot_index, module, 0, page,
+				    MLXSW_REG_MCIA_TH_PAGE_OFF + off,
+				    MLXSW_REG_MCIA_TH_ITEM_SIZE,
+				    MLXSW_REG_MCIA_I2C_ADDR_LOW);
+	} else {
+		mlxsw_reg_mcia_pack(mcia_pl, slot_index, module, 0,
+				    MLXSW_REG_MCIA_PAGE0_LO,
+				    off, MLXSW_REG_MCIA_TH_ITEM_SIZE,
+				    MLXSW_REG_MCIA_I2C_ADDR_HIGH);
+	}
+
+	err = mlxsw_reg_query(core, MLXSW_REG(mcia), mcia_pl);
+	if (err)
+		return err;
+
+	eeprom_tmp = mlxsw_reg_mcia_eeprom_data(mcia_pl);
+	memcpy(temp_thresh.buf, eeprom_tmp, MLXSW_REG_MCIA_TH_ITEM_SIZE);
+	*temp = temp_thresh.temp * 1000;
+
+	return 0;
+}
+
+int mlxsw_env_get_module_info(struct net_device *netdev,
+			      struct mlxsw_core *mlxsw_core, u8 slot_index,
+			      int module, struct ethtool_modinfo *modinfo)
+{
+	struct mlxsw_env *mlxsw_env = mlxsw_core_env(mlxsw_core);
+	u8 module_info[MLXSW_REG_MCIA_EEPROM_MODULE_INFO_SIZE];
+	u16 offset = MLXSW_REG_MCIA_EEPROM_MODULE_INFO_SIZE;
+	u8 module_rev_id, module_id, diag_mon;
+	unsigned int read_size;
+	int err;
+
+	if (!mlxsw_env_linecard_is_active(mlxsw_env, slot_index)) {
+		netdev_err(netdev, "Cannot read EEPROM of module on an inactive line card\n");
+		return -EIO;
+	}
+
+	err = mlxsw_env_validate_module_type(mlxsw_core, slot_index, module);
+	if (err) {
+		netdev_err(netdev,
+			   "EEPROM is not equipped on port module type");
+		return err;
+	}
+
+	err = mlxsw_env_query_module_eeprom(mlxsw_core, slot_index, module, 0,
+					    offset, module_info, false,
+					    &read_size);
+	if (err)
+		return err;
+
+	if (read_size < offset)
+		return -EIO;
+
+	module_rev_id = module_info[MLXSW_REG_MCIA_EEPROM_MODULE_INFO_REV_ID];
+	module_id = module_info[MLXSW_REG_MCIA_EEPROM_MODULE_INFO_ID];
+
+	switch (module_id) {
+	case MLXSW_REG_MCIA_EEPROM_MODULE_INFO_ID_QSFP:
+		modinfo->type       = ETH_MODULE_SFF_8436;
+		modinfo->eeprom_len = ETH_MODULE_SFF_8436_MAX_LEN;
+		break;
+	case MLXSW_REG_MCIA_EEPROM_MODULE_INFO_ID_QSFP_PLUS:
+	case MLXSW_REG_MCIA_EEPROM_MODULE_INFO_ID_QSFP28:
+		if (module_id == MLXSW_REG_MCIA_EEPROM_MODULE_INFO_ID_QSFP28 ||
+		    module_rev_id >=
+		    MLXSW_REG_MCIA_EEPROM_MODULE_INFO_REV_ID_8636) {
+			modinfo->type       = ETH_MODULE_SFF_8636;
+			modinfo->eeprom_len = ETH_MODULE_SFF_8636_MAX_LEN;
+		} else {
+			modinfo->type       = ETH_MODULE_SFF_8436;
+			modinfo->eeprom_len = ETH_MODULE_SFF_8436_MAX_LEN;
+		}
+		break;
+	case MLXSW_REG_MCIA_EEPROM_MODULE_INFO_ID_SFP:
+		/* Verify if transceiver provides diagnostic monitoring page */
+		err = mlxsw_env_query_module_eeprom(mlxsw_core, slot_index,
+						    module, SFP_DIAGMON, 1,
+						    &diag_mon, false,
+						    &read_size);
+		if (err)
+			return err;
+
+		if (read_size < 1)
+			return -EIO;
+
+		modinfo->type       = ETH_MODULE_SFF_8472;
+		if (diag_mon)
+			modinfo->eeprom_len = ETH_MODULE_SFF_8472_LEN;
+		else
+			modinfo->eeprom_len = ETH_MODULE_SFF_8472_LEN / 2;
+		break;
+	case MLXSW_REG_MCIA_EEPROM_MODULE_INFO_ID_QSFP_DD:
+	case MLXSW_REG_MCIA_EEPROM_MODULE_INFO_ID_OSFP:
+		/* Use SFF_8636 as base type. ethtool should recognize specific
+		 * type through the identifier value.
+		 */
+		modinfo->type       = ETH_MODULE_SFF_8636;
+		/* Verify if module EEPROM is a flat memory. In case of flat
+		 * memory only page 00h (0-255 bytes) can be read. Otherwise
+		 * upper pages 01h and 02h can also be read. Upper pages 10h
+		 * and 11h are currently not supported by the driver.
+		 */
+		if (module_info[MLXSW_REG_MCIA_EEPROM_MODULE_INFO_TYPE_ID] &
+		    MLXSW_REG_MCIA_EEPROM_CMIS_FLAT_MEMORY)
+			modinfo->eeprom_len = ETH_MODULE_SFF_8636_LEN;
+		else
+			modinfo->eeprom_len = ETH_MODULE_SFF_8472_LEN;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	return 0;
+}
+EXPORT_SYMBOL(mlxsw_env_get_module_info);
+
+int mlxsw_env_get_module_eeprom(struct net_device *netdev,
+				struct mlxsw_core *mlxsw_core, u8 slot_index,
+				int module, struct ethtool_eeprom *ee,
+				u8 *data)
+{
+	struct mlxsw_env *mlxsw_env = mlxsw_core_env(mlxsw_core);
+	int offset = ee->offset;
+	unsigned int read_size;
+	bool qsfp, cmis;
+	int i = 0;
+	int err;
+
+	if (!ee->len)
+		return -EINVAL;
+
+	if (!mlxsw_env_linecard_is_active(mlxsw_env, slot_index)) {
+		netdev_err(netdev, "Cannot read EEPROM of module on an inactive line card\n");
+		return -EIO;
+	}
+
+	memset(data, 0, ee->len);
+	/* Validate module identifier value. */
+	err = mlxsw_env_validate_cable_ident(mlxsw_core, slot_index, module,
+					     &qsfp, &cmis);
+	if (err)
+		return err;
+
+	while (i < ee->len) {
+		err = mlxsw_env_query_module_eeprom(mlxsw_core, slot_index,
+						    module, offset,
+						    ee->len - i, data + i,
+						    qsfp, &read_size);
+		if (err) {
+			netdev_err(netdev, "Eeprom query failed\n");
+			return err;
+		}
+
+		i += read_size;
+		offset += read_size;
+	}
+
+	return 0;
+}
+EXPORT_SYMBOL(mlxsw_env_get_module_eeprom);
+
+static int mlxsw_env_mcia_status_process(const char *mcia_pl,
+					 struct netlink_ext_ack *extack)
+{
+	u8 status = mlxsw_reg_mcia_status_get(mcia_pl);
+
+	switch (status) {
+	case MLXSW_REG_MCIA_STATUS_GOOD:
+		return 0;
+	case MLXSW_REG_MCIA_STATUS_NO_EEPROM_MODULE:
+		NL_SET_ERR_MSG_MOD(extack, "No response from module's EEPROM");
+		return -EIO;
+	case MLXSW_REG_MCIA_STATUS_MODULE_NOT_SUPPORTED:
+		NL_SET_ERR_MSG_MOD(extack, "Module type not supported by the device");
+		return -EOPNOTSUPP;
+	case MLXSW_REG_MCIA_STATUS_MODULE_NOT_CONNECTED:
+		NL_SET_ERR_MSG_MOD(extack, "No module present indication");
+		return -EIO;
+	case MLXSW_REG_MCIA_STATUS_I2C_ERROR:
+		NL_SET_ERR_MSG_MOD(extack, "Error occurred while trying to access module's EEPROM using I2C");
+		return -EIO;
+	case MLXSW_REG_MCIA_STATUS_MODULE_DISABLED:
+		NL_SET_ERR_MSG_MOD(extack, "Module is disabled");
+		return -EIO;
+	default:
+		NL_SET_ERR_MSG_MOD(extack, "Unknown error");
+		return -EIO;
+	}
+}
+
+int
+mlxsw_env_get_module_eeprom_by_page(struct mlxsw_core *mlxsw_core,
+				    u8 slot_index, u8 module,
+				    const struct ethtool_module_eeprom *page,
+				    struct netlink_ext_ack *extack)
+{
+	struct mlxsw_env *mlxsw_env = mlxsw_core_env(mlxsw_core);
+	u32 bytes_read = 0;
+	u16 device_addr;
+	int err;
+
+	if (!mlxsw_env_linecard_is_active(mlxsw_env, slot_index)) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "Cannot read EEPROM of module on an inactive line card");
+		return -EIO;
+	}
+
+	err = mlxsw_env_validate_module_type(mlxsw_core, slot_index, module);
+	if (err) {
+		NL_SET_ERR_MSG_MOD(extack, "EEPROM is not equipped on port module type");
+		return err;
+	}
+
+	/* Offset cannot be larger than 2 * ETH_MODULE_EEPROM_PAGE_LEN */
+	device_addr = page->offset;
+
+	while (bytes_read < page->length) {
+		char mcia_pl[MLXSW_REG_MCIA_LEN];
+		char *eeprom_tmp;
+		u8 size;
+
+		size = min_t(u8, page->length - bytes_read,
+			     MLXSW_REG_MCIA_EEPROM_SIZE);
+
+		mlxsw_reg_mcia_pack(mcia_pl, slot_index, module, 0, page->page,
+				    device_addr + bytes_read, size,
+				    page->i2c_address);
+		mlxsw_reg_mcia_bank_number_set(mcia_pl, page->bank);
+
+		err = mlxsw_reg_query(mlxsw_core, MLXSW_REG(mcia), mcia_pl);
+		if (err) {
+			NL_SET_ERR_MSG_MOD(extack, "Failed to access module's EEPROM");
+			return err;
+		}
+
+		err = mlxsw_env_mcia_status_process(mcia_pl, extack);
+		if (err)
+			return err;
+
+		eeprom_tmp = mlxsw_reg_mcia_eeprom_data(mcia_pl);
+		memcpy(page->data + bytes_read, eeprom_tmp, size);
+		bytes_read += size;
+	}
+
+	return bytes_read;
+}
+EXPORT_SYMBOL(mlxsw_env_get_module_eeprom_by_page);
+
+static int mlxsw_env_module_reset(struct mlxsw_core *mlxsw_core, u8 slot_index,
+				  u8 module)
+{
+	char pmaos_pl[MLXSW_REG_PMAOS_LEN];
+
+	mlxsw_reg_pmaos_pack(pmaos_pl, slot_index, module);
+	mlxsw_reg_pmaos_rst_set(pmaos_pl, true);
+
+	return mlxsw_reg_write(mlxsw_core, MLXSW_REG(pmaos), pmaos_pl);
+}
+
+int mlxsw_env_reset_module(struct net_device *netdev,
+			   struct mlxsw_core *mlxsw_core, u8 slot_index,
+			   u8 module, u32 *flags)
+{
+	struct mlxsw_env *mlxsw_env = mlxsw_core_env(mlxsw_core);
+	struct mlxsw_env_module_info *module_info;
+	u32 req = *flags;
+	int err;
+
+	if (!(req & ETH_RESET_PHY) &&
+	    !(req & (ETH_RESET_PHY << ETH_RESET_SHARED_SHIFT)))
+		return 0;
+
+	if (!mlxsw_env_linecard_is_active(mlxsw_env, slot_index)) {
+		netdev_err(netdev, "Cannot reset module on an inactive line card\n");
+		return -EIO;
+	}
+
+	mutex_lock(&mlxsw_env->line_cards_lock);
+
+	err = __mlxsw_env_validate_module_type(mlxsw_core, slot_index, module);
+	if (err) {
+		netdev_err(netdev, "Reset module is not supported on port module type\n");
+		goto out;
+	}
+
+	module_info = mlxsw_env_module_info_get(mlxsw_core, slot_index, module);
+	if (module_info->num_ports_up) {
+		netdev_err(netdev, "Cannot reset module when ports using it are administratively up\n");
+		err = -EINVAL;
+		goto out;
+	}
+
+	if (module_info->num_ports_mapped > 1 &&
+	    !(req & (ETH_RESET_PHY << ETH_RESET_SHARED_SHIFT))) {
+		netdev_err(netdev, "Cannot reset module without \"phy-shared\" flag when shared by multiple ports\n");
+		err = -EINVAL;
+		goto out;
+	}
+
+	err = mlxsw_env_module_reset(mlxsw_core, slot_index, module);
+	if (err) {
+		netdev_err(netdev, "Failed to reset module\n");
+		goto out;
+	}
+
+	*flags &= ~(ETH_RESET_PHY | (ETH_RESET_PHY << ETH_RESET_SHARED_SHIFT));
+
+out:
+	mutex_unlock(&mlxsw_env->line_cards_lock);
+	return err;
+}
+EXPORT_SYMBOL(mlxsw_env_reset_module);
+
+int
+mlxsw_env_get_module_power_mode(struct mlxsw_core *mlxsw_core, u8 slot_index,
+				u8 module,
+				struct ethtool_module_power_mode_params *params,
+				struct netlink_ext_ack *extack)
+{
+	struct mlxsw_env *mlxsw_env = mlxsw_core_env(mlxsw_core);
+	struct mlxsw_env_module_info *module_info;
+	char mcion_pl[MLXSW_REG_MCION_LEN];
+	u32 status_bits;
+	int err = 0;
+
+	mutex_lock(&mlxsw_env->line_cards_lock);
+
+	err = __mlxsw_env_validate_module_type(mlxsw_core, slot_index, module);
+	if (err) {
+		NL_SET_ERR_MSG_MOD(extack, "Power mode is not supported on port module type");
+		goto out;
+	}
+
+	module_info = mlxsw_env_module_info_get(mlxsw_core, slot_index, module);
+	params->policy = module_info->power_mode_policy;
+
+	/* Avoid accessing an inactive line card, as it will result in an error. */
+	if (!__mlxsw_env_linecard_is_active(mlxsw_env, slot_index))
+		goto out;
+
+	mlxsw_reg_mcion_pack(mcion_pl, slot_index, module);
+	err = mlxsw_reg_query(mlxsw_core, MLXSW_REG(mcion), mcion_pl);
+	if (err) {
+		NL_SET_ERR_MSG_MOD(extack, "Failed to retrieve module's power mode");
+		goto out;
+	}
+
+	status_bits = mlxsw_reg_mcion_module_status_bits_get(mcion_pl);
+	if (!(status_bits & MLXSW_REG_MCION_MODULE_STATUS_BITS_PRESENT_MASK))
+		goto out;
+
+	if (status_bits & MLXSW_REG_MCION_MODULE_STATUS_BITS_LOW_POWER_MASK)
+		params->mode = ETHTOOL_MODULE_POWER_MODE_LOW;
+	else
+		params->mode = ETHTOOL_MODULE_POWER_MODE_HIGH;
+
+out:
+	mutex_unlock(&mlxsw_env->line_cards_lock);
+	return err;
+}
+EXPORT_SYMBOL(mlxsw_env_get_module_power_mode);
+
+static int mlxsw_env_module_enable_set(struct mlxsw_core *mlxsw_core,
+				       u8 slot_index, u8 module, bool enable)
+{
+	enum mlxsw_reg_pmaos_admin_status admin_status;
+	char pmaos_pl[MLXSW_REG_PMAOS_LEN];
+
+	mlxsw_reg_pmaos_pack(pmaos_pl, slot_index, module);
+	admin_status = enable ? MLXSW_REG_PMAOS_ADMIN_STATUS_ENABLED :
+				MLXSW_REG_PMAOS_ADMIN_STATUS_DISABLED;
+	mlxsw_reg_pmaos_admin_status_set(pmaos_pl, admin_status);
+	mlxsw_reg_pmaos_ase_set(pmaos_pl, true);
+
+	return mlxsw_reg_write(mlxsw_core, MLXSW_REG(pmaos), pmaos_pl);
+}
+
+static int mlxsw_env_module_low_power_set(struct mlxsw_core *mlxsw_core,
+					  u8 slot_index, u8 module,
+					  bool low_power)
+{
+	u16 eeprom_override_mask, eeprom_override;
+	char pmmp_pl[MLXSW_REG_PMMP_LEN];
+
+	mlxsw_reg_pmmp_pack(pmmp_pl, slot_index, module);
+	mlxsw_reg_pmmp_sticky_set(pmmp_pl, true);
+	/* Mask all the bits except low power mode. */
+	eeprom_override_mask = ~MLXSW_REG_PMMP_EEPROM_OVERRIDE_LOW_POWER_MASK;
+	mlxsw_reg_pmmp_eeprom_override_mask_set(pmmp_pl, eeprom_override_mask);
+	eeprom_override = low_power ? MLXSW_REG_PMMP_EEPROM_OVERRIDE_LOW_POWER_MASK :
+				      0;
+	mlxsw_reg_pmmp_eeprom_override_set(pmmp_pl, eeprom_override);
+
+	return mlxsw_reg_write(mlxsw_core, MLXSW_REG(pmmp), pmmp_pl);
+}
+
+static int __mlxsw_env_set_module_power_mode(struct mlxsw_core *mlxsw_core,
+					     u8 slot_index, u8 module,
+					     bool low_power,
+					     struct netlink_ext_ack *extack)
+{
+	struct mlxsw_env *mlxsw_env = mlxsw_core_env(mlxsw_core);
+	int err;
+
+	/* Avoid accessing an inactive line card, as it will result in an error.
+	 * Cached configuration will be applied by mlxsw_env_got_active() when
+	 * line card becomes active.
+	 */
+	if (!__mlxsw_env_linecard_is_active(mlxsw_env, slot_index))
+		return 0;
+
+	err = mlxsw_env_module_enable_set(mlxsw_core, slot_index, module, false);
+	if (err) {
+		NL_SET_ERR_MSG_MOD(extack, "Failed to disable module");
+		return err;
+	}
+
+	err = mlxsw_env_module_low_power_set(mlxsw_core, slot_index, module,
+					     low_power);
+	if (err) {
+		NL_SET_ERR_MSG_MOD(extack, "Failed to set module's power mode");
+		goto err_module_low_power_set;
+	}
+
+	err = mlxsw_env_module_enable_set(mlxsw_core, slot_index, module, true);
+	if (err) {
+		NL_SET_ERR_MSG_MOD(extack, "Failed to enable module");
+		goto err_module_enable_set;
+	}
+
+	return 0;
+
+err_module_enable_set:
+	mlxsw_env_module_low_power_set(mlxsw_core, slot_index, module,
+				       !low_power);
+err_module_low_power_set:
+	mlxsw_env_module_enable_set(mlxsw_core, slot_index, module, true);
+	return err;
+}
+
+static int
+mlxsw_env_set_module_power_mode_apply(struct mlxsw_core *mlxsw_core,
+				      u8 slot_index, u8 module,
+				      enum ethtool_module_power_mode_policy policy,
+				      struct netlink_ext_ack *extack)
+{
+	struct mlxsw_env_module_info *module_info;
+	bool low_power;
+	int err = 0;
+
+	err = __mlxsw_env_validate_module_type(mlxsw_core, slot_index, module);
+	if (err) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "Power mode set is not supported on port module type");
+		goto out;
+	}
+
+	module_info = mlxsw_env_module_info_get(mlxsw_core, slot_index, module);
+	if (module_info->power_mode_policy == policy)
+		goto out;
+
+	/* If any ports are up, we are already in high power mode. */
+	if (module_info->num_ports_up)
+		goto out_set_policy;
+
+	low_power = policy == ETHTOOL_MODULE_POWER_MODE_POLICY_AUTO;
+	err = __mlxsw_env_set_module_power_mode(mlxsw_core, slot_index, module,
+						low_power, extack);
+	if (err)
+		goto out;
+
+out_set_policy:
+	module_info->power_mode_policy = policy;
+out:
+	return err;
+}
+
+int
+mlxsw_env_set_module_power_mode(struct mlxsw_core *mlxsw_core, u8 slot_index,
+				u8 module,
+				enum ethtool_module_power_mode_policy policy,
+				struct netlink_ext_ack *extack)
+{
+	struct mlxsw_env *mlxsw_env = mlxsw_core_env(mlxsw_core);
+	int err;
+
+	if (policy != ETHTOOL_MODULE_POWER_MODE_POLICY_HIGH &&
+	    policy != ETHTOOL_MODULE_POWER_MODE_POLICY_AUTO) {
+		NL_SET_ERR_MSG_MOD(extack, "Unsupported power mode policy");
+		return -EOPNOTSUPP;
+	}
+
+	mutex_lock(&mlxsw_env->line_cards_lock);
+	err = mlxsw_env_set_module_power_mode_apply(mlxsw_core, slot_index,
+						    module, policy, extack);
+	mutex_unlock(&mlxsw_env->line_cards_lock);
+
+	return err;
+}
+EXPORT_SYMBOL(mlxsw_env_set_module_power_mode);
+
+static int mlxsw_env_module_has_temp_sensor(struct mlxsw_core *mlxsw_core,
+					    u8 slot_index, u8 module,
+					    bool *p_has_temp_sensor)
+{
+	char mtbr_pl[MLXSW_REG_MTBR_LEN];
+	u16 temp;
+	int err;
+
+	mlxsw_reg_mtbr_pack(mtbr_pl, slot_index,
+			    MLXSW_REG_MTBR_BASE_MODULE_INDEX + module, 1);
+	err = mlxsw_reg_query(mlxsw_core, MLXSW_REG(mtbr), mtbr_pl);
+	if (err)
+		return err;
+
+	mlxsw_reg_mtbr_temp_unpack(mtbr_pl, 0, &temp, NULL);
+
+	switch (temp) {
+	case MLXSW_REG_MTBR_BAD_SENS_INFO:
+	case MLXSW_REG_MTBR_NO_CONN:
+	case MLXSW_REG_MTBR_NO_TEMP_SENS:
+	case MLXSW_REG_MTBR_INDEX_NA:
+		*p_has_temp_sensor = false;
+		break;
+	default:
+		*p_has_temp_sensor = temp ? true : false;
+	}
+	return 0;
+}
+
+static int
+mlxsw_env_temp_event_set(struct mlxsw_core *mlxsw_core, u8 slot_index,
+			 u16 sensor_index, bool enable)
+{
+	char mtmp_pl[MLXSW_REG_MTMP_LEN] = {0};
+	enum mlxsw_reg_mtmp_tee tee;
+	int err, threshold_hi;
+
+	mlxsw_reg_mtmp_slot_index_set(mtmp_pl, slot_index);
+	mlxsw_reg_mtmp_sensor_index_set(mtmp_pl, sensor_index);
+	err = mlxsw_reg_query(mlxsw_core, MLXSW_REG(mtmp), mtmp_pl);
+	if (err)
+		return err;
+
+	if (enable) {
+		err = mlxsw_env_module_temp_thresholds_get(mlxsw_core,
+							   slot_index,
+							   sensor_index -
+							   MLXSW_REG_MTMP_MODULE_INDEX_MIN,
+							   SFP_TEMP_HIGH_WARN,
+							   &threshold_hi);
+		/* In case it is not possible to query the module's threshold,
+		 * use the default value.
+		 */
+		if (err)
+			threshold_hi = MLXSW_REG_MTMP_THRESH_HI;
+		else
+			/* mlxsw_env_module_temp_thresholds_get() multiplies
+			 * Celsius degrees by 1000 whereas MTMP expects
+			 * temperature in 0.125 Celsius degrees units.
+			 * Convert threshold_hi to correct units.
+			 */
+			threshold_hi = threshold_hi / 1000 * 8;
+
+		mlxsw_reg_mtmp_temperature_threshold_hi_set(mtmp_pl, threshold_hi);
+		mlxsw_reg_mtmp_temperature_threshold_lo_set(mtmp_pl, threshold_hi -
+							    MLXSW_REG_MTMP_HYSTERESIS_TEMP);
+	}
+	tee = enable ? MLXSW_REG_MTMP_TEE_GENERATE_EVENT : MLXSW_REG_MTMP_TEE_NO_EVENT;
+	mlxsw_reg_mtmp_tee_set(mtmp_pl, tee);
+	return mlxsw_reg_write(mlxsw_core, MLXSW_REG(mtmp), mtmp_pl);
+}
+
+static int mlxsw_env_module_temp_event_enable(struct mlxsw_core *mlxsw_core,
+					      u8 slot_index)
+{
+	struct mlxsw_env *mlxsw_env = mlxsw_core_env(mlxsw_core);
+	int i, err, sensor_index;
+	bool has_temp_sensor;
+
+	for (i = 0; i < mlxsw_env->line_cards[slot_index]->module_count; i++) {
+		err = mlxsw_env_module_has_temp_sensor(mlxsw_core, slot_index,
+						       i, &has_temp_sensor);
+		if (err)
+			return err;
+
+		if (!has_temp_sensor)
+			continue;
+
+		sensor_index = i + MLXSW_REG_MTMP_MODULE_INDEX_MIN;
+		err = mlxsw_env_temp_event_set(mlxsw_core, slot_index,
+					       sensor_index, true);
+		if (err)
+			return err;
+	}
+
+	return 0;
+}
+
+struct mlxsw_env_module_temp_warn_event {
+	struct mlxsw_env *mlxsw_env;
+	char mtwe_pl[MLXSW_REG_MTWE_LEN];
+	struct work_struct work;
+};
+
+static void mlxsw_env_mtwe_event_work(struct work_struct *work)
+{
+	struct mlxsw_env_module_temp_warn_event *event;
+	struct mlxsw_env_module_info *module_info;
+	struct mlxsw_env *mlxsw_env;
+	int i, sensor_warning;
+	bool is_overheat;
+
+	event = container_of(work, struct mlxsw_env_module_temp_warn_event,
+			     work);
+	mlxsw_env = event->mlxsw_env;
+
+	for (i = 0; i < mlxsw_env->max_module_count; i++) {
+		/* 64-127 of sensor_index are mapped to the port modules
+		 * sequentially (module 0 is mapped to sensor_index 64,
+		 * module 1 to sensor_index 65 and so on)
+		 */
+		sensor_warning =
+			mlxsw_reg_mtwe_sensor_warning_get(event->mtwe_pl,
+							  i + MLXSW_REG_MTMP_MODULE_INDEX_MIN);
+		mutex_lock(&mlxsw_env->line_cards_lock);
+		/* MTWE only supports main board. */
+		module_info = mlxsw_env_module_info_get(mlxsw_env->core, 0, i);
+		is_overheat = module_info->is_overheat;
+
+		if ((is_overheat && sensor_warning) ||
+		    (!is_overheat && !sensor_warning)) {
+			/* Current state is "warning" and MTWE still reports
+			 * warning OR current state in "no warning" and MTWE
+			 * does not report warning.
+			 */
+			mutex_unlock(&mlxsw_env->line_cards_lock);
+			continue;
+		} else if (is_overheat && !sensor_warning) {
+			/* MTWE reports "no warning", turn is_overheat off.
+			 */
+			module_info->is_overheat = false;
+			mutex_unlock(&mlxsw_env->line_cards_lock);
+		} else {
+			/* Current state is "no warning" and MTWE reports
+			 * "warning", increase the counter and turn is_overheat
+			 * on.
+			 */
+			module_info->is_overheat = true;
+			module_info->module_overheat_counter++;
+			mutex_unlock(&mlxsw_env->line_cards_lock);
+		}
+	}
+
+	kfree(event);
+}
+
+static void
+mlxsw_env_mtwe_listener_func(const struct mlxsw_reg_info *reg, char *mtwe_pl,
+			     void *priv)
+{
+	struct mlxsw_env_module_temp_warn_event *event;
+	struct mlxsw_env *mlxsw_env = priv;
+
+	event = kmalloc(sizeof(*event), GFP_ATOMIC);
+	if (!event)
+		return;
+
+	event->mlxsw_env = mlxsw_env;
+	memcpy(event->mtwe_pl, mtwe_pl, MLXSW_REG_MTWE_LEN);
+	INIT_WORK(&event->work, mlxsw_env_mtwe_event_work);
+	mlxsw_core_schedule_work(&event->work);
+}
+
+static const struct mlxsw_listener mlxsw_env_temp_warn_listener =
+	MLXSW_CORE_EVENTL(mlxsw_env_mtwe_listener_func, MTWE);
+
+static int mlxsw_env_temp_warn_event_register(struct mlxsw_core *mlxsw_core)
+{
+	struct mlxsw_env *mlxsw_env = mlxsw_core_env(mlxsw_core);
+
+	return mlxsw_core_trap_register(mlxsw_core,
+					&mlxsw_env_temp_warn_listener,
+					mlxsw_env);
+}
+
+static void mlxsw_env_temp_warn_event_unregister(struct mlxsw_env *mlxsw_env)
+{
+	mlxsw_core_trap_unregister(mlxsw_env->core,
+				   &mlxsw_env_temp_warn_listener, mlxsw_env);
+}
+
+struct mlxsw_env_module_plug_unplug_event {
+	struct mlxsw_env *mlxsw_env;
+	u8 slot_index;
+	u8 module;
+	struct work_struct work;
+};
+
+static void mlxsw_env_pmpe_event_work(struct work_struct *work)
+{
+	struct mlxsw_env_module_plug_unplug_event *event;
+	struct mlxsw_env_module_info *module_info;
+	struct mlxsw_env *mlxsw_env;
+	bool has_temp_sensor;
+	u16 sensor_index;
+	int err;
+
+	event = container_of(work, struct mlxsw_env_module_plug_unplug_event,
+			     work);
+	mlxsw_env = event->mlxsw_env;
+
+	mutex_lock(&mlxsw_env->line_cards_lock);
+	module_info = mlxsw_env_module_info_get(mlxsw_env->core,
+						event->slot_index,
+						event->module);
+	module_info->is_overheat = false;
+	mutex_unlock(&mlxsw_env->line_cards_lock);
+
+	err = mlxsw_env_module_has_temp_sensor(mlxsw_env->core,
+					       event->slot_index,
+					       event->module,
+					       &has_temp_sensor);
+	/* Do not disable events on modules without sensors or faulty sensors
+	 * because FW returns errors.
+	 */
+	if (err)
+		goto out;
+
+	if (!has_temp_sensor)
+		goto out;
+
+	sensor_index = event->module + MLXSW_REG_MTMP_MODULE_INDEX_MIN;
+	mlxsw_env_temp_event_set(mlxsw_env->core, event->slot_index,
+				 sensor_index, true);
+
+out:
+	kfree(event);
+}
+
+static void
+mlxsw_env_pmpe_listener_func(const struct mlxsw_reg_info *reg, char *pmpe_pl,
+			     void *priv)
+{
+	u8 slot_index = mlxsw_reg_pmpe_slot_index_get(pmpe_pl);
+	struct mlxsw_env_module_plug_unplug_event *event;
+	enum mlxsw_reg_pmpe_module_status module_status;
+	u8 module = mlxsw_reg_pmpe_module_get(pmpe_pl);
+	struct mlxsw_env *mlxsw_env = priv;
+
+	if (WARN_ON_ONCE(module >= mlxsw_env->max_module_count ||
+			 slot_index >= mlxsw_env->num_of_slots))
+		return;
+
+	module_status = mlxsw_reg_pmpe_module_status_get(pmpe_pl);
+	if (module_status != MLXSW_REG_PMPE_MODULE_STATUS_PLUGGED_ENABLED)
+		return;
+
+	event = kmalloc(sizeof(*event), GFP_ATOMIC);
+	if (!event)
+		return;
+
+	event->mlxsw_env = mlxsw_env;
+	event->slot_index = slot_index;
+	event->module = module;
+	INIT_WORK(&event->work, mlxsw_env_pmpe_event_work);
+	mlxsw_core_schedule_work(&event->work);
+}
+
+static const struct mlxsw_listener mlxsw_env_module_plug_listener =
+	MLXSW_CORE_EVENTL(mlxsw_env_pmpe_listener_func, PMPE);
+
+static int
+mlxsw_env_module_plug_event_register(struct mlxsw_core *mlxsw_core)
+{
+	struct mlxsw_env *mlxsw_env = mlxsw_core_env(mlxsw_core);
+
+	return mlxsw_core_trap_register(mlxsw_core,
+					&mlxsw_env_module_plug_listener,
+					mlxsw_env);
+}
+
+static void
+mlxsw_env_module_plug_event_unregister(struct mlxsw_env *mlxsw_env)
+{
+	mlxsw_core_trap_unregister(mlxsw_env->core,
+				   &mlxsw_env_module_plug_listener,
+				   mlxsw_env);
+}
+
+static int
+mlxsw_env_module_oper_state_event_enable(struct mlxsw_core *mlxsw_core,
+					 u8 slot_index)
+{
+	struct mlxsw_env *mlxsw_env = mlxsw_core_env(mlxsw_core);
+	int i, err;
+
+	for (i = 0; i < mlxsw_env->line_cards[slot_index]->module_count; i++) {
+		char pmaos_pl[MLXSW_REG_PMAOS_LEN];
+
+		mlxsw_reg_pmaos_pack(pmaos_pl, slot_index, i);
+		mlxsw_reg_pmaos_e_set(pmaos_pl,
+				      MLXSW_REG_PMAOS_E_GENERATE_EVENT);
+		mlxsw_reg_pmaos_ee_set(pmaos_pl, true);
+		err = mlxsw_reg_write(mlxsw_core, MLXSW_REG(pmaos), pmaos_pl);
+		if (err)
+			return err;
+	}
+	return 0;
+}
+
+int
+mlxsw_env_module_overheat_counter_get(struct mlxsw_core *mlxsw_core, u8 slot_index,
+				      u8 module, u64 *p_counter)
+{
+	struct mlxsw_env *mlxsw_env = mlxsw_core_env(mlxsw_core);
+	struct mlxsw_env_module_info *module_info;
+
+	mutex_lock(&mlxsw_env->line_cards_lock);
+	module_info = mlxsw_env_module_info_get(mlxsw_core, slot_index, module);
+	*p_counter = module_info->module_overheat_counter;
+	mutex_unlock(&mlxsw_env->line_cards_lock);
+
+	return 0;
+}
+EXPORT_SYMBOL(mlxsw_env_module_overheat_counter_get);
+
+void mlxsw_env_module_port_map(struct mlxsw_core *mlxsw_core, u8 slot_index,
+			       u8 module)
+{
+	struct mlxsw_env *mlxsw_env = mlxsw_core_env(mlxsw_core);
+	struct mlxsw_env_module_info *module_info;
+
+	mutex_lock(&mlxsw_env->line_cards_lock);
+	module_info = mlxsw_env_module_info_get(mlxsw_core, slot_index, module);
+	module_info->num_ports_mapped++;
+	mutex_unlock(&mlxsw_env->line_cards_lock);
+}
+EXPORT_SYMBOL(mlxsw_env_module_port_map);
+
+void mlxsw_env_module_port_unmap(struct mlxsw_core *mlxsw_core, u8 slot_index,
+				 u8 module)
+{
+	struct mlxsw_env *mlxsw_env = mlxsw_core_env(mlxsw_core);
+	struct mlxsw_env_module_info *module_info;
+
+	mutex_lock(&mlxsw_env->line_cards_lock);
+	module_info = mlxsw_env_module_info_get(mlxsw_core, slot_index, module);
+	module_info->num_ports_mapped--;
+	mutex_unlock(&mlxsw_env->line_cards_lock);
+}
+EXPORT_SYMBOL(mlxsw_env_module_port_unmap);
+
+int mlxsw_env_module_port_up(struct mlxsw_core *mlxsw_core, u8 slot_index,
+			     u8 module)
+{
+	struct mlxsw_env *mlxsw_env = mlxsw_core_env(mlxsw_core);
+	struct mlxsw_env_module_info *module_info;
+	int err = 0;
+
+	mutex_lock(&mlxsw_env->line_cards_lock);
+
+	module_info = mlxsw_env_module_info_get(mlxsw_core, slot_index, module);
+	if (module_info->power_mode_policy !=
+	    ETHTOOL_MODULE_POWER_MODE_POLICY_AUTO)
+		goto out_inc;
+
+	if (module_info->num_ports_up != 0)
+		goto out_inc;
+
+	/* Transition to high power mode following first port using the module
+	 * being put administratively up.
+	 */
+	err = __mlxsw_env_set_module_power_mode(mlxsw_core, slot_index, module,
+						false, NULL);
+	if (err)
+		goto out_unlock;
+
+out_inc:
+	module_info->num_ports_up++;
+out_unlock:
+	mutex_unlock(&mlxsw_env->line_cards_lock);
+	return err;
+}
+EXPORT_SYMBOL(mlxsw_env_module_port_up);
+
+void mlxsw_env_module_port_down(struct mlxsw_core *mlxsw_core, u8 slot_index,
+				u8 module)
+{
+	struct mlxsw_env *mlxsw_env = mlxsw_core_env(mlxsw_core);
+	struct mlxsw_env_module_info *module_info;
+
+	mutex_lock(&mlxsw_env->line_cards_lock);
+
+	module_info = mlxsw_env_module_info_get(mlxsw_core, slot_index, module);
+	module_info->num_ports_up--;
+
+	if (module_info->power_mode_policy !=
+	    ETHTOOL_MODULE_POWER_MODE_POLICY_AUTO)
+		goto out_unlock;
+
+	if (module_info->num_ports_up != 0)
+		goto out_unlock;
+
+	/* Transition to low power mode following last port using the module
+	 * being put administratively down.
+	 */
+	__mlxsw_env_set_module_power_mode(mlxsw_core, slot_index, module, true,
+					  NULL);
+
+out_unlock:
+	mutex_unlock(&mlxsw_env->line_cards_lock);
+}
+EXPORT_SYMBOL(mlxsw_env_module_port_down);
+
+static int mlxsw_env_line_cards_alloc(struct mlxsw_env *env)
+{
+	struct mlxsw_env_module_info *module_info;
+	int i, j;
+
+	for (i = 0; i < env->num_of_slots; i++) {
+		env->line_cards[i] = kzalloc(struct_size(env->line_cards[i],
+							 module_info,
+							 env->max_module_count),
+							 GFP_KERNEL);
+		if (!env->line_cards[i])
+			goto kzalloc_err;
+
+		/* Firmware defaults to high power mode policy where modules
+		 * are transitioned to high power mode following plug-in.
+		 */
+		for (j = 0; j < env->max_module_count; j++) {
+			module_info = &env->line_cards[i]->module_info[j];
+			module_info->power_mode_policy =
+					ETHTOOL_MODULE_POWER_MODE_POLICY_HIGH;
+		}
+	}
+
+	return 0;
+
+kzalloc_err:
+	for (i--; i >= 0; i--)
+		kfree(env->line_cards[i]);
+	return -ENOMEM;
+}
+
+static void mlxsw_env_line_cards_free(struct mlxsw_env *env)
+{
+	int i = env->num_of_slots;
+
+	for (i--; i >= 0; i--)
+		kfree(env->line_cards[i]);
+}
+
+static int
+mlxsw_env_module_event_enable(struct mlxsw_env *mlxsw_env, u8 slot_index)
+{
+	int err;
+
+	err = mlxsw_env_module_oper_state_event_enable(mlxsw_env->core,
+						       slot_index);
+	if (err)
+		return err;
+
+	err = mlxsw_env_module_temp_event_enable(mlxsw_env->core, slot_index);
+	if (err)
+		return err;
+
+	return 0;
+}
+
+static void
+mlxsw_env_module_event_disable(struct mlxsw_env *mlxsw_env, u8 slot_index)
+{
+}
+
+static int
+mlxsw_env_module_type_set(struct mlxsw_core *mlxsw_core, u8 slot_index)
+{
+	struct mlxsw_env *mlxsw_env = mlxsw_core_env(mlxsw_core);
+	int i;
+
+	for (i = 0; i < mlxsw_env->line_cards[slot_index]->module_count; i++) {
+		struct mlxsw_env_module_info *module_info;
+		char pmtm_pl[MLXSW_REG_PMTM_LEN];
+		int err;
+
+		mlxsw_reg_pmtm_pack(pmtm_pl, slot_index, i);
+		err = mlxsw_reg_query(mlxsw_core, MLXSW_REG(pmtm), pmtm_pl);
+		if (err)
+			return err;
+
+		module_info = mlxsw_env_module_info_get(mlxsw_core, slot_index,
+							i);
+		module_info->type = mlxsw_reg_pmtm_module_type_get(pmtm_pl);
+	}
+
+	return 0;
+}
+
+static void
+mlxsw_env_linecard_modules_power_mode_apply(struct mlxsw_core *mlxsw_core,
+					    struct mlxsw_env *env,
+					    u8 slot_index)
+{
+	int i;
+
+	for (i = 0; i < env->line_cards[slot_index]->module_count; i++) {
+		enum ethtool_module_power_mode_policy policy;
+		struct mlxsw_env_module_info *module_info;
+		struct netlink_ext_ack extack;
+		int err;
+
+		module_info = &env->line_cards[slot_index]->module_info[i];
+		policy = module_info->power_mode_policy;
+		err = mlxsw_env_set_module_power_mode_apply(mlxsw_core,
+							    slot_index, i,
+							    policy, &extack);
+		if (err)
+			dev_err(env->bus_info->dev, "%s\n", extack._msg);
+	}
+}
+
+static void
+mlxsw_env_got_active(struct mlxsw_core *mlxsw_core, u8 slot_index, void *priv)
+{
+	struct mlxsw_env *mlxsw_env = priv;
+	char mgpir_pl[MLXSW_REG_MGPIR_LEN];
+	int err;
+
+	mutex_lock(&mlxsw_env->line_cards_lock);
+	if (__mlxsw_env_linecard_is_active(mlxsw_env, slot_index))
+		goto out_unlock;
+
+	mlxsw_reg_mgpir_pack(mgpir_pl, slot_index);
+	err = mlxsw_reg_query(mlxsw_env->core, MLXSW_REG(mgpir), mgpir_pl);
+	if (err)
+		goto out_unlock;
+
+	mlxsw_reg_mgpir_unpack(mgpir_pl, NULL, NULL, NULL,
+			       &mlxsw_env->line_cards[slot_index]->module_count,
+			       NULL);
+
+	err = mlxsw_env_module_event_enable(mlxsw_env, slot_index);
+	if (err) {
+		dev_err(mlxsw_env->bus_info->dev, "Failed to enable port module events for line card in slot %d\n",
+			slot_index);
+		goto err_mlxsw_env_module_event_enable;
+	}
+	err = mlxsw_env_module_type_set(mlxsw_env->core, slot_index);
+	if (err) {
+		dev_err(mlxsw_env->bus_info->dev, "Failed to set modules' type for line card in slot %d\n",
+			slot_index);
+		goto err_type_set;
+	}
+
+	mlxsw_env->line_cards[slot_index]->active = true;
+	/* Apply power mode policy. */
+	mlxsw_env_linecard_modules_power_mode_apply(mlxsw_core, mlxsw_env,
+						    slot_index);
+	mutex_unlock(&mlxsw_env->line_cards_lock);
+
+	return;
+
+err_type_set:
+	mlxsw_env_module_event_disable(mlxsw_env, slot_index);
+err_mlxsw_env_module_event_enable:
+out_unlock:
+	mutex_unlock(&mlxsw_env->line_cards_lock);
+}
+
+static void
+mlxsw_env_got_inactive(struct mlxsw_core *mlxsw_core, u8 slot_index,
+		       void *priv)
+{
+	struct mlxsw_env *mlxsw_env = priv;
+
+	mutex_lock(&mlxsw_env->line_cards_lock);
+	if (!__mlxsw_env_linecard_is_active(mlxsw_env, slot_index))
+		goto out_unlock;
+	mlxsw_env->line_cards[slot_index]->active = false;
+	mlxsw_env_module_event_disable(mlxsw_env, slot_index);
+	mlxsw_env->line_cards[slot_index]->module_count = 0;
+out_unlock:
+	mutex_unlock(&mlxsw_env->line_cards_lock);
+}
+
+static struct mlxsw_linecards_event_ops mlxsw_env_event_ops = {
+	.got_active = mlxsw_env_got_active,
+	.got_inactive = mlxsw_env_got_inactive,
+};
+
+int mlxsw_env_init(struct mlxsw_core *mlxsw_core,
+		   const struct mlxsw_bus_info *bus_info,
+		   struct mlxsw_env **p_env)
+{
+	u8 module_count, num_of_slots, max_module_count;
+	char mgpir_pl[MLXSW_REG_MGPIR_LEN];
+	struct mlxsw_env *env;
+	int err;
+
+	mlxsw_reg_mgpir_pack(mgpir_pl, 0);
+	err = mlxsw_reg_query(mlxsw_core, MLXSW_REG(mgpir), mgpir_pl);
+	if (err)
+		return err;
+
+	mlxsw_reg_mgpir_unpack(mgpir_pl, NULL, NULL, NULL, &module_count,
+			       &num_of_slots);
+	/* If the system is modular, get the maximum number of modules per-slot.
+	 * Otherwise, get the maximum number of modules on the main board.
+	 */
+	max_module_count = num_of_slots ?
+			   mlxsw_reg_mgpir_max_modules_per_slot_get(mgpir_pl) :
+			   module_count;
+
+	env = kzalloc(struct_size(env, line_cards, num_of_slots + 1),
+		      GFP_KERNEL);
+	if (!env)
+		return -ENOMEM;
+
+	env->core = mlxsw_core;
+	env->bus_info = bus_info;
+	env->num_of_slots = num_of_slots + 1;
+	env->max_module_count = max_module_count;
+	err = mlxsw_env_line_cards_alloc(env);
+	if (err)
+		goto err_mlxsw_env_line_cards_alloc;
+
+	mutex_init(&env->line_cards_lock);
+	*p_env = env;
+
+	err = mlxsw_linecards_event_ops_register(env->core,
+						 &mlxsw_env_event_ops, env);
+	if (err)
+		goto err_linecards_event_ops_register;
+
+	err = mlxsw_env_temp_warn_event_register(mlxsw_core);
+	if (err)
+		goto err_temp_warn_event_register;
+
+	err = mlxsw_env_module_plug_event_register(mlxsw_core);
+	if (err)
+		goto err_module_plug_event_register;
+
+	/* Set 'module_count' only for main board. Actual count for line card
+	 * is to be set after line card is activated.
+	 */
+	env->line_cards[0]->module_count = num_of_slots ? 0 : module_count;
+	/* Enable events only for main board. Line card events are to be
+	 * configured only after line card is activated. Before that, access to
+	 * modules on line cards is not allowed.
+	 */
+	err = mlxsw_env_module_event_enable(env, 0);
+	if (err)
+		goto err_mlxsw_env_module_event_enable;
+
+	err = mlxsw_env_module_type_set(mlxsw_core, 0);
+	if (err)
+		goto err_type_set;
+
+	env->line_cards[0]->active = true;
+
+	return 0;
+
+err_type_set:
+	mlxsw_env_module_event_disable(env, 0);
+err_mlxsw_env_module_event_enable:
+	mlxsw_env_module_plug_event_unregister(env);
+err_module_plug_event_register:
+	mlxsw_env_temp_warn_event_unregister(env);
+err_temp_warn_event_register:
+	mlxsw_linecards_event_ops_unregister(env->core,
+					     &mlxsw_env_event_ops, env);
+err_linecards_event_ops_register:
+	mutex_destroy(&env->line_cards_lock);
+	mlxsw_env_line_cards_free(env);
+err_mlxsw_env_line_cards_alloc:
+	kfree(env);
+	return err;
+}
+
+void mlxsw_env_fini(struct mlxsw_env *env)
+{
+	env->line_cards[0]->active = false;
+	mlxsw_env_module_event_disable(env, 0);
+	mlxsw_env_module_plug_event_unregister(env);
+	/* Make sure there is no more event work scheduled. */
+	mlxsw_core_flush_owq();
+	mlxsw_env_temp_warn_event_unregister(env);
+	mlxsw_linecards_event_ops_unregister(env->core,
+					     &mlxsw_env_event_ops, env);
+	mutex_destroy(&env->line_cards_lock);
+	mlxsw_env_line_cards_free(env);
+	kfree(env);
+}
diff --git a/drivers/net/ethernet/mellanox/mlxsw/core_env.h b/drivers/net/ethernet/mellanox/mlxsw/core_env.h
new file mode 100644
index 000000000..a197e3ae0
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/core_env.h
@@ -0,0 +1,68 @@
+/* SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0 */
+/* Copyright (c) 2018 Mellanox Technologies. All rights reserved */
+
+#ifndef _MLXSW_CORE_ENV_H
+#define _MLXSW_CORE_ENV_H
+
+#include <linux/ethtool.h>
+
+struct ethtool_modinfo;
+struct ethtool_eeprom;
+
+int mlxsw_env_module_temp_thresholds_get(struct mlxsw_core *core,
+					 u8 slot_index, int module, int off,
+					 int *temp);
+
+int mlxsw_env_get_module_info(struct net_device *netdev,
+			      struct mlxsw_core *mlxsw_core, u8 slot_index,
+			      int module, struct ethtool_modinfo *modinfo);
+
+int mlxsw_env_get_module_eeprom(struct net_device *netdev,
+				struct mlxsw_core *mlxsw_core, u8 slot_index,
+				int module, struct ethtool_eeprom *ee,
+				u8 *data);
+
+int
+mlxsw_env_get_module_eeprom_by_page(struct mlxsw_core *mlxsw_core,
+				    u8 slot_index, u8 module,
+				    const struct ethtool_module_eeprom *page,
+				    struct netlink_ext_ack *extack);
+
+int mlxsw_env_reset_module(struct net_device *netdev,
+			   struct mlxsw_core *mlxsw_core, u8 slot_index,
+			   u8 module, u32 *flags);
+
+int
+mlxsw_env_get_module_power_mode(struct mlxsw_core *mlxsw_core, u8 slot_index,
+				u8 module,
+				struct ethtool_module_power_mode_params *params,
+				struct netlink_ext_ack *extack);
+
+int
+mlxsw_env_set_module_power_mode(struct mlxsw_core *mlxsw_core, u8 slot_index,
+				u8 module,
+				enum ethtool_module_power_mode_policy policy,
+				struct netlink_ext_ack *extack);
+
+int
+mlxsw_env_module_overheat_counter_get(struct mlxsw_core *mlxsw_core, u8 slot_index,
+				      u8 module, u64 *p_counter);
+
+void mlxsw_env_module_port_map(struct mlxsw_core *mlxsw_core, u8 slot_index,
+			       u8 module);
+
+void mlxsw_env_module_port_unmap(struct mlxsw_core *mlxsw_core, u8 slot_index,
+				 u8 module);
+
+int mlxsw_env_module_port_up(struct mlxsw_core *mlxsw_core, u8 slot_index,
+			     u8 module);
+
+void mlxsw_env_module_port_down(struct mlxsw_core *mlxsw_core, u8 slot_index,
+				u8 module);
+
+int mlxsw_env_init(struct mlxsw_core *core,
+		   const struct mlxsw_bus_info *bus_info,
+		   struct mlxsw_env **p_env);
+void mlxsw_env_fini(struct mlxsw_env *env);
+
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlxsw/core_hwmon.c b/drivers/net/ethernet/mellanox/mlxsw/core_hwmon.c
new file mode 100644
index 000000000..0fd290d77
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/core_hwmon.c
@@ -0,0 +1,941 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2015-2018 Mellanox Technologies. All rights reserved */
+
+#include <linux/kernel.h>
+#include <linux/types.h>
+#include <linux/device.h>
+#include <linux/sysfs.h>
+#include <linux/hwmon.h>
+#include <linux/err.h>
+#include <linux/sfp.h>
+
+#include "core.h"
+#include "core_env.h"
+
+#define MLXSW_HWMON_SENSORS_MAX_COUNT 64
+#define MLXSW_HWMON_MODULES_MAX_COUNT 64
+#define MLXSW_HWMON_GEARBOXES_MAX_COUNT 32
+
+#define MLXSW_HWMON_ATTR_PER_SENSOR 3
+#define MLXSW_HWMON_ATTR_PER_MODULE 7
+#define MLXSW_HWMON_ATTR_PER_GEARBOX 4
+#define MLXSW_HWMON_DEV_NAME_LEN_MAX 16
+
+#define MLXSW_HWMON_ATTR_COUNT (MLXSW_HWMON_SENSORS_MAX_COUNT * MLXSW_HWMON_ATTR_PER_SENSOR + \
+				MLXSW_HWMON_MODULES_MAX_COUNT * MLXSW_HWMON_ATTR_PER_MODULE + \
+				MLXSW_HWMON_GEARBOXES_MAX_COUNT * MLXSW_HWMON_ATTR_PER_GEARBOX + \
+				MLXSW_MFCR_TACHOS_MAX + MLXSW_MFCR_PWMS_MAX)
+
+struct mlxsw_hwmon_attr {
+	struct device_attribute dev_attr;
+	struct mlxsw_hwmon_dev *mlxsw_hwmon_dev;
+	unsigned int type_index;
+	char name[32];
+};
+
+static int mlxsw_hwmon_get_attr_index(int index, int count)
+{
+	if (index >= count)
+		return index % count + MLXSW_REG_MTMP_GBOX_INDEX_MIN;
+
+	return index;
+}
+
+struct mlxsw_hwmon_dev {
+	char name[MLXSW_HWMON_DEV_NAME_LEN_MAX];
+	struct mlxsw_hwmon *hwmon;
+	struct device *hwmon_dev;
+	struct attribute_group group;
+	const struct attribute_group *groups[2];
+	struct attribute *attrs[MLXSW_HWMON_ATTR_COUNT + 1];
+	struct mlxsw_hwmon_attr hwmon_attrs[MLXSW_HWMON_ATTR_COUNT];
+	unsigned int attrs_count;
+	u8 sensor_count;
+	u8 module_sensor_max;
+	u8 slot_index;
+	bool active;
+};
+
+struct mlxsw_hwmon {
+	struct mlxsw_core *core;
+	const struct mlxsw_bus_info *bus_info;
+	struct mlxsw_hwmon_dev line_cards[];
+};
+
+static ssize_t mlxsw_hwmon_temp_show(struct device *dev,
+				     struct device_attribute *attr,
+				     char *buf)
+{
+	struct mlxsw_hwmon_attr *mlxsw_hwmon_attr =
+			container_of(attr, struct mlxsw_hwmon_attr, dev_attr);
+	struct mlxsw_hwmon_dev *mlxsw_hwmon_dev = mlxsw_hwmon_attr->mlxsw_hwmon_dev;
+	struct mlxsw_hwmon *mlxsw_hwmon = mlxsw_hwmon_dev->hwmon;
+	char mtmp_pl[MLXSW_REG_MTMP_LEN];
+	int temp, index;
+	int err;
+
+	index = mlxsw_hwmon_get_attr_index(mlxsw_hwmon_attr->type_index,
+					   mlxsw_hwmon_dev->module_sensor_max);
+	mlxsw_reg_mtmp_pack(mtmp_pl, mlxsw_hwmon_dev->slot_index, index, false,
+			    false);
+	err = mlxsw_reg_query(mlxsw_hwmon->core, MLXSW_REG(mtmp), mtmp_pl);
+	if (err) {
+		dev_err(mlxsw_hwmon->bus_info->dev, "Failed to query temp sensor\n");
+		return err;
+	}
+	mlxsw_reg_mtmp_unpack(mtmp_pl, &temp, NULL, NULL, NULL, NULL);
+	return sprintf(buf, "%d\n", temp);
+}
+
+static ssize_t mlxsw_hwmon_temp_max_show(struct device *dev,
+					 struct device_attribute *attr,
+					 char *buf)
+{
+	struct mlxsw_hwmon_attr *mlxsw_hwmon_attr =
+			container_of(attr, struct mlxsw_hwmon_attr, dev_attr);
+	struct mlxsw_hwmon_dev *mlxsw_hwmon_dev = mlxsw_hwmon_attr->mlxsw_hwmon_dev;
+	struct mlxsw_hwmon *mlxsw_hwmon = mlxsw_hwmon_dev->hwmon;
+	char mtmp_pl[MLXSW_REG_MTMP_LEN];
+	int temp_max, index;
+	int err;
+
+	index = mlxsw_hwmon_get_attr_index(mlxsw_hwmon_attr->type_index,
+					   mlxsw_hwmon_dev->module_sensor_max);
+	mlxsw_reg_mtmp_pack(mtmp_pl, mlxsw_hwmon_dev->slot_index, index, false,
+			    false);
+	err = mlxsw_reg_query(mlxsw_hwmon->core, MLXSW_REG(mtmp), mtmp_pl);
+	if (err) {
+		dev_err(mlxsw_hwmon->bus_info->dev, "Failed to query temp sensor\n");
+		return err;
+	}
+	mlxsw_reg_mtmp_unpack(mtmp_pl, NULL, &temp_max, NULL, NULL, NULL);
+	return sprintf(buf, "%d\n", temp_max);
+}
+
+static ssize_t mlxsw_hwmon_temp_rst_store(struct device *dev,
+					  struct device_attribute *attr,
+					  const char *buf, size_t len)
+{
+	struct mlxsw_hwmon_attr *mlxsw_hwmon_attr =
+			container_of(attr, struct mlxsw_hwmon_attr, dev_attr);
+	struct mlxsw_hwmon_dev *mlxsw_hwmon_dev = mlxsw_hwmon_attr->mlxsw_hwmon_dev;
+	struct mlxsw_hwmon *mlxsw_hwmon = mlxsw_hwmon_dev->hwmon;
+	char mtmp_pl[MLXSW_REG_MTMP_LEN];
+	unsigned long val;
+	int index;
+	int err;
+
+	err = kstrtoul(buf, 10, &val);
+	if (err)
+		return err;
+	if (val != 1)
+		return -EINVAL;
+
+	index = mlxsw_hwmon_get_attr_index(mlxsw_hwmon_attr->type_index,
+					   mlxsw_hwmon_dev->module_sensor_max);
+
+	mlxsw_reg_mtmp_slot_index_set(mtmp_pl, mlxsw_hwmon_dev->slot_index);
+	mlxsw_reg_mtmp_sensor_index_set(mtmp_pl, index);
+	err = mlxsw_reg_query(mlxsw_hwmon->core, MLXSW_REG(mtmp), mtmp_pl);
+	if (err)
+		return err;
+	mlxsw_reg_mtmp_mte_set(mtmp_pl, true);
+	mlxsw_reg_mtmp_mtr_set(mtmp_pl, true);
+	err = mlxsw_reg_write(mlxsw_hwmon->core, MLXSW_REG(mtmp), mtmp_pl);
+	if (err) {
+		dev_err(mlxsw_hwmon->bus_info->dev, "Failed to reset temp sensor history\n");
+		return err;
+	}
+	return len;
+}
+
+static ssize_t mlxsw_hwmon_fan_rpm_show(struct device *dev,
+					struct device_attribute *attr,
+					char *buf)
+{
+	struct mlxsw_hwmon_attr *mlxsw_hwmon_attr =
+			container_of(attr, struct mlxsw_hwmon_attr, dev_attr);
+	struct mlxsw_hwmon_dev *mlxsw_hwmon_dev = mlxsw_hwmon_attr->mlxsw_hwmon_dev;
+	struct mlxsw_hwmon *mlxsw_hwmon = mlxsw_hwmon_dev->hwmon;
+	char mfsm_pl[MLXSW_REG_MFSM_LEN];
+	int err;
+
+	mlxsw_reg_mfsm_pack(mfsm_pl, mlxsw_hwmon_attr->type_index);
+	err = mlxsw_reg_query(mlxsw_hwmon->core, MLXSW_REG(mfsm), mfsm_pl);
+	if (err) {
+		dev_err(mlxsw_hwmon->bus_info->dev, "Failed to query fan\n");
+		return err;
+	}
+	return sprintf(buf, "%u\n", mlxsw_reg_mfsm_rpm_get(mfsm_pl));
+}
+
+static ssize_t mlxsw_hwmon_fan_fault_show(struct device *dev,
+					  struct device_attribute *attr,
+					  char *buf)
+{
+	struct mlxsw_hwmon_attr *mlxsw_hwmon_attr =
+			container_of(attr, struct mlxsw_hwmon_attr, dev_attr);
+	struct mlxsw_hwmon_dev *mlxsw_hwmon_dev = mlxsw_hwmon_attr->mlxsw_hwmon_dev;
+	struct mlxsw_hwmon *mlxsw_hwmon = mlxsw_hwmon_dev->hwmon;
+	char fore_pl[MLXSW_REG_FORE_LEN];
+	bool fault;
+	int err;
+
+	err = mlxsw_reg_query(mlxsw_hwmon->core, MLXSW_REG(fore), fore_pl);
+	if (err) {
+		dev_err(mlxsw_hwmon->bus_info->dev, "Failed to query fan\n");
+		return err;
+	}
+	mlxsw_reg_fore_unpack(fore_pl, mlxsw_hwmon_attr->type_index, &fault);
+
+	return sprintf(buf, "%u\n", fault);
+}
+
+static ssize_t mlxsw_hwmon_pwm_show(struct device *dev,
+				    struct device_attribute *attr,
+				    char *buf)
+{
+	struct mlxsw_hwmon_attr *mlxsw_hwmon_attr =
+			container_of(attr, struct mlxsw_hwmon_attr, dev_attr);
+	struct mlxsw_hwmon_dev *mlxsw_hwmon_dev = mlxsw_hwmon_attr->mlxsw_hwmon_dev;
+	struct mlxsw_hwmon *mlxsw_hwmon = mlxsw_hwmon_dev->hwmon;
+	char mfsc_pl[MLXSW_REG_MFSC_LEN];
+	int err;
+
+	mlxsw_reg_mfsc_pack(mfsc_pl, mlxsw_hwmon_attr->type_index, 0);
+	err = mlxsw_reg_query(mlxsw_hwmon->core, MLXSW_REG(mfsc), mfsc_pl);
+	if (err) {
+		dev_err(mlxsw_hwmon->bus_info->dev, "Failed to query PWM\n");
+		return err;
+	}
+	return sprintf(buf, "%u\n",
+		       mlxsw_reg_mfsc_pwm_duty_cycle_get(mfsc_pl));
+}
+
+static ssize_t mlxsw_hwmon_pwm_store(struct device *dev,
+				     struct device_attribute *attr,
+				     const char *buf, size_t len)
+{
+	struct mlxsw_hwmon_attr *mlxsw_hwmon_attr =
+			container_of(attr, struct mlxsw_hwmon_attr, dev_attr);
+	struct mlxsw_hwmon_dev *mlxsw_hwmon_dev = mlxsw_hwmon_attr->mlxsw_hwmon_dev;
+	struct mlxsw_hwmon *mlxsw_hwmon = mlxsw_hwmon_dev->hwmon;
+	char mfsc_pl[MLXSW_REG_MFSC_LEN];
+	unsigned long val;
+	int err;
+
+	err = kstrtoul(buf, 10, &val);
+	if (err)
+		return err;
+	if (val > 255)
+		return -EINVAL;
+
+	mlxsw_reg_mfsc_pack(mfsc_pl, mlxsw_hwmon_attr->type_index, val);
+	err = mlxsw_reg_write(mlxsw_hwmon->core, MLXSW_REG(mfsc), mfsc_pl);
+	if (err) {
+		dev_err(mlxsw_hwmon->bus_info->dev, "Failed to write PWM\n");
+		return err;
+	}
+	return len;
+}
+
+static int mlxsw_hwmon_module_temp_get(struct device *dev,
+				       struct device_attribute *attr,
+				       int *p_temp)
+{
+	struct mlxsw_hwmon_attr *mlxsw_hwmon_attr =
+			container_of(attr, struct mlxsw_hwmon_attr, dev_attr);
+	struct mlxsw_hwmon_dev *mlxsw_hwmon_dev = mlxsw_hwmon_attr->mlxsw_hwmon_dev;
+	struct mlxsw_hwmon *mlxsw_hwmon = mlxsw_hwmon_dev->hwmon;
+	char mtmp_pl[MLXSW_REG_MTMP_LEN];
+	u8 module;
+	int err;
+
+	module = mlxsw_hwmon_attr->type_index - mlxsw_hwmon_dev->sensor_count;
+	mlxsw_reg_mtmp_pack(mtmp_pl, mlxsw_hwmon_dev->slot_index,
+			    MLXSW_REG_MTMP_MODULE_INDEX_MIN + module, false,
+			    false);
+	err = mlxsw_reg_query(mlxsw_hwmon->core, MLXSW_REG(mtmp), mtmp_pl);
+	if (err) {
+		dev_err(dev, "Failed to query module temperature\n");
+		return err;
+	}
+	mlxsw_reg_mtmp_unpack(mtmp_pl, p_temp, NULL, NULL, NULL, NULL);
+
+	return 0;
+}
+
+static ssize_t mlxsw_hwmon_module_temp_show(struct device *dev,
+					    struct device_attribute *attr,
+					    char *buf)
+{
+	int err, temp;
+
+	err = mlxsw_hwmon_module_temp_get(dev, attr, &temp);
+	if (err)
+		return err;
+
+	return sprintf(buf, "%d\n", temp);
+}
+
+static ssize_t mlxsw_hwmon_module_temp_fault_show(struct device *dev,
+						  struct device_attribute *attr,
+						  char *buf)
+{
+	struct mlxsw_hwmon_attr *mlxsw_hwmon_attr =
+			container_of(attr, struct mlxsw_hwmon_attr, dev_attr);
+	struct mlxsw_hwmon_dev *mlxsw_hwmon_dev = mlxsw_hwmon_attr->mlxsw_hwmon_dev;
+	struct mlxsw_hwmon *mlxsw_hwmon = mlxsw_hwmon_dev->hwmon;
+	char mtbr_pl[MLXSW_REG_MTBR_LEN] = {0};
+	u8 module, fault;
+	u16 temp;
+	int err;
+
+	module = mlxsw_hwmon_attr->type_index - mlxsw_hwmon_dev->sensor_count;
+	mlxsw_reg_mtbr_pack(mtbr_pl, mlxsw_hwmon_dev->slot_index,
+			    MLXSW_REG_MTBR_BASE_MODULE_INDEX + module, 1);
+	err = mlxsw_reg_query(mlxsw_hwmon->core, MLXSW_REG(mtbr), mtbr_pl);
+	if (err) {
+		dev_err(dev, "Failed to query module temperature sensor\n");
+		return err;
+	}
+
+	mlxsw_reg_mtbr_temp_unpack(mtbr_pl, 0, &temp, NULL);
+
+	/* Update status and temperature cache. */
+	switch (temp) {
+	case MLXSW_REG_MTBR_BAD_SENS_INFO:
+		/* Untrusted cable is connected. Reading temperature from its
+		 * sensor is faulty.
+		 */
+		fault = 1;
+		break;
+	case MLXSW_REG_MTBR_NO_CONN:
+	case MLXSW_REG_MTBR_NO_TEMP_SENS:
+	case MLXSW_REG_MTBR_INDEX_NA:
+	default:
+		fault = 0;
+		break;
+	}
+
+	return sprintf(buf, "%u\n", fault);
+}
+
+static int mlxsw_hwmon_module_temp_critical_get(struct device *dev,
+						struct device_attribute *attr,
+						int *p_temp)
+{
+	struct mlxsw_hwmon_attr *mlxsw_hwmon_attr =
+			container_of(attr, struct mlxsw_hwmon_attr, dev_attr);
+	struct mlxsw_hwmon_dev *mlxsw_hwmon_dev = mlxsw_hwmon_attr->mlxsw_hwmon_dev;
+	struct mlxsw_hwmon *mlxsw_hwmon = mlxsw_hwmon_dev->hwmon;
+	u8 module;
+	int err;
+
+	module = mlxsw_hwmon_attr->type_index - mlxsw_hwmon_dev->sensor_count;
+	err = mlxsw_env_module_temp_thresholds_get(mlxsw_hwmon->core,
+						   mlxsw_hwmon_dev->slot_index,
+						   module, SFP_TEMP_HIGH_WARN,
+						   p_temp);
+	if (err) {
+		dev_err(dev, "Failed to query module temperature thresholds\n");
+		return err;
+	}
+
+	return 0;
+}
+
+static ssize_t
+mlxsw_hwmon_module_temp_critical_show(struct device *dev,
+				      struct device_attribute *attr, char *buf)
+{
+	int err, temp;
+
+	err = mlxsw_hwmon_module_temp_critical_get(dev, attr, &temp);
+	if (err)
+		return err;
+
+	return sprintf(buf, "%u\n", temp);
+}
+
+static int mlxsw_hwmon_module_temp_emergency_get(struct device *dev,
+						 struct device_attribute *attr,
+						 int *p_temp)
+{
+	struct mlxsw_hwmon_attr *mlxsw_hwmon_attr =
+			container_of(attr, struct mlxsw_hwmon_attr, dev_attr);
+	struct mlxsw_hwmon_dev *mlxsw_hwmon_dev = mlxsw_hwmon_attr->mlxsw_hwmon_dev;
+	struct mlxsw_hwmon *mlxsw_hwmon = mlxsw_hwmon_dev->hwmon;
+	u8 module;
+	int err;
+
+	module = mlxsw_hwmon_attr->type_index - mlxsw_hwmon_dev->sensor_count;
+	err = mlxsw_env_module_temp_thresholds_get(mlxsw_hwmon->core,
+						   mlxsw_hwmon_dev->slot_index,
+						   module, SFP_TEMP_HIGH_ALARM,
+						   p_temp);
+	if (err) {
+		dev_err(dev, "Failed to query module temperature thresholds\n");
+		return err;
+	}
+
+	return 0;
+}
+
+static ssize_t
+mlxsw_hwmon_module_temp_emergency_show(struct device *dev,
+				       struct device_attribute *attr,
+				       char *buf)
+{
+	int err, temp;
+
+	err = mlxsw_hwmon_module_temp_emergency_get(dev, attr, &temp);
+	if (err)
+		return err;
+
+	return sprintf(buf, "%u\n", temp);
+}
+
+static ssize_t
+mlxsw_hwmon_module_temp_label_show(struct device *dev,
+				   struct device_attribute *attr,
+				   char *buf)
+{
+	struct mlxsw_hwmon_attr *mlxsw_hwmon_attr =
+			container_of(attr, struct mlxsw_hwmon_attr, dev_attr);
+
+	return sprintf(buf, "front panel %03u\n",
+		       mlxsw_hwmon_attr->type_index + 1 -
+		       mlxsw_hwmon_attr->mlxsw_hwmon_dev->sensor_count);
+}
+
+static ssize_t
+mlxsw_hwmon_gbox_temp_label_show(struct device *dev,
+				 struct device_attribute *attr,
+				 char *buf)
+{
+	struct mlxsw_hwmon_attr *mlxsw_hwmon_attr =
+			container_of(attr, struct mlxsw_hwmon_attr, dev_attr);
+	struct mlxsw_hwmon_dev *mlxsw_hwmon_dev = mlxsw_hwmon_attr->mlxsw_hwmon_dev;
+	int index = mlxsw_hwmon_attr->type_index -
+		    mlxsw_hwmon_dev->module_sensor_max + 1;
+
+	return sprintf(buf, "gearbox %03u\n", index);
+}
+
+static ssize_t mlxsw_hwmon_temp_critical_alarm_show(struct device *dev,
+						    struct device_attribute *attr,
+						    char *buf)
+{
+	int err, temp, emergency_temp, critic_temp;
+
+	err = mlxsw_hwmon_module_temp_get(dev, attr, &temp);
+	if (err)
+		return err;
+
+	if (temp <= 0)
+		return sprintf(buf, "%d\n", false);
+
+	err = mlxsw_hwmon_module_temp_emergency_get(dev, attr, &emergency_temp);
+	if (err)
+		return err;
+
+	if (temp >= emergency_temp)
+		return sprintf(buf, "%d\n", false);
+
+	err = mlxsw_hwmon_module_temp_critical_get(dev, attr, &critic_temp);
+	if (err)
+		return err;
+
+	return sprintf(buf, "%d\n", temp >= critic_temp);
+}
+
+static ssize_t mlxsw_hwmon_temp_emergency_alarm_show(struct device *dev,
+						     struct device_attribute *attr,
+						     char *buf)
+{
+	int err, temp, emergency_temp;
+
+	err = mlxsw_hwmon_module_temp_get(dev, attr, &temp);
+	if (err)
+		return err;
+
+	if (temp <= 0)
+		return sprintf(buf, "%d\n", false);
+
+	err = mlxsw_hwmon_module_temp_emergency_get(dev, attr, &emergency_temp);
+	if (err)
+		return err;
+
+	return sprintf(buf, "%d\n", temp >= emergency_temp);
+}
+
+enum mlxsw_hwmon_attr_type {
+	MLXSW_HWMON_ATTR_TYPE_TEMP,
+	MLXSW_HWMON_ATTR_TYPE_TEMP_MAX,
+	MLXSW_HWMON_ATTR_TYPE_TEMP_RST,
+	MLXSW_HWMON_ATTR_TYPE_FAN_RPM,
+	MLXSW_HWMON_ATTR_TYPE_FAN_FAULT,
+	MLXSW_HWMON_ATTR_TYPE_PWM,
+	MLXSW_HWMON_ATTR_TYPE_TEMP_MODULE,
+	MLXSW_HWMON_ATTR_TYPE_TEMP_MODULE_FAULT,
+	MLXSW_HWMON_ATTR_TYPE_TEMP_MODULE_CRIT,
+	MLXSW_HWMON_ATTR_TYPE_TEMP_MODULE_EMERG,
+	MLXSW_HWMON_ATTR_TYPE_TEMP_MODULE_LABEL,
+	MLXSW_HWMON_ATTR_TYPE_TEMP_GBOX_LABEL,
+	MLXSW_HWMON_ATTR_TYPE_TEMP_CRIT_ALARM,
+	MLXSW_HWMON_ATTR_TYPE_TEMP_EMERGENCY_ALARM,
+};
+
+static void mlxsw_hwmon_attr_add(struct mlxsw_hwmon_dev *mlxsw_hwmon_dev,
+				 enum mlxsw_hwmon_attr_type attr_type,
+				 unsigned int type_index, unsigned int num)
+{
+	struct mlxsw_hwmon_attr *mlxsw_hwmon_attr;
+	unsigned int attr_index;
+
+	attr_index = mlxsw_hwmon_dev->attrs_count;
+	mlxsw_hwmon_attr = &mlxsw_hwmon_dev->hwmon_attrs[attr_index];
+
+	switch (attr_type) {
+	case MLXSW_HWMON_ATTR_TYPE_TEMP:
+		mlxsw_hwmon_attr->dev_attr.show = mlxsw_hwmon_temp_show;
+		mlxsw_hwmon_attr->dev_attr.attr.mode = 0444;
+		snprintf(mlxsw_hwmon_attr->name, sizeof(mlxsw_hwmon_attr->name),
+			 "temp%u_input", num + 1);
+		break;
+	case MLXSW_HWMON_ATTR_TYPE_TEMP_MAX:
+		mlxsw_hwmon_attr->dev_attr.show = mlxsw_hwmon_temp_max_show;
+		mlxsw_hwmon_attr->dev_attr.attr.mode = 0444;
+		snprintf(mlxsw_hwmon_attr->name, sizeof(mlxsw_hwmon_attr->name),
+			 "temp%u_highest", num + 1);
+		break;
+	case MLXSW_HWMON_ATTR_TYPE_TEMP_RST:
+		mlxsw_hwmon_attr->dev_attr.store = mlxsw_hwmon_temp_rst_store;
+		mlxsw_hwmon_attr->dev_attr.attr.mode = 0200;
+		snprintf(mlxsw_hwmon_attr->name, sizeof(mlxsw_hwmon_attr->name),
+			 "temp%u_reset_history", num + 1);
+		break;
+	case MLXSW_HWMON_ATTR_TYPE_FAN_RPM:
+		mlxsw_hwmon_attr->dev_attr.show = mlxsw_hwmon_fan_rpm_show;
+		mlxsw_hwmon_attr->dev_attr.attr.mode = 0444;
+		snprintf(mlxsw_hwmon_attr->name, sizeof(mlxsw_hwmon_attr->name),
+			 "fan%u_input", num + 1);
+		break;
+	case MLXSW_HWMON_ATTR_TYPE_FAN_FAULT:
+		mlxsw_hwmon_attr->dev_attr.show = mlxsw_hwmon_fan_fault_show;
+		mlxsw_hwmon_attr->dev_attr.attr.mode = 0444;
+		snprintf(mlxsw_hwmon_attr->name, sizeof(mlxsw_hwmon_attr->name),
+			 "fan%u_fault", num + 1);
+		break;
+	case MLXSW_HWMON_ATTR_TYPE_PWM:
+		mlxsw_hwmon_attr->dev_attr.show = mlxsw_hwmon_pwm_show;
+		mlxsw_hwmon_attr->dev_attr.store = mlxsw_hwmon_pwm_store;
+		mlxsw_hwmon_attr->dev_attr.attr.mode = 0644;
+		snprintf(mlxsw_hwmon_attr->name, sizeof(mlxsw_hwmon_attr->name),
+			 "pwm%u", num + 1);
+		break;
+	case MLXSW_HWMON_ATTR_TYPE_TEMP_MODULE:
+		mlxsw_hwmon_attr->dev_attr.show = mlxsw_hwmon_module_temp_show;
+		mlxsw_hwmon_attr->dev_attr.attr.mode = 0444;
+		snprintf(mlxsw_hwmon_attr->name, sizeof(mlxsw_hwmon_attr->name),
+			 "temp%u_input", num + 1);
+		break;
+	case MLXSW_HWMON_ATTR_TYPE_TEMP_MODULE_FAULT:
+		mlxsw_hwmon_attr->dev_attr.show =
+					mlxsw_hwmon_module_temp_fault_show;
+		mlxsw_hwmon_attr->dev_attr.attr.mode = 0444;
+		snprintf(mlxsw_hwmon_attr->name, sizeof(mlxsw_hwmon_attr->name),
+			 "temp%u_fault", num + 1);
+		break;
+	case MLXSW_HWMON_ATTR_TYPE_TEMP_MODULE_CRIT:
+		mlxsw_hwmon_attr->dev_attr.show =
+			mlxsw_hwmon_module_temp_critical_show;
+		mlxsw_hwmon_attr->dev_attr.attr.mode = 0444;
+		snprintf(mlxsw_hwmon_attr->name, sizeof(mlxsw_hwmon_attr->name),
+			 "temp%u_crit", num + 1);
+		break;
+	case MLXSW_HWMON_ATTR_TYPE_TEMP_MODULE_EMERG:
+		mlxsw_hwmon_attr->dev_attr.show =
+			mlxsw_hwmon_module_temp_emergency_show;
+		mlxsw_hwmon_attr->dev_attr.attr.mode = 0444;
+		snprintf(mlxsw_hwmon_attr->name, sizeof(mlxsw_hwmon_attr->name),
+			 "temp%u_emergency", num + 1);
+		break;
+	case MLXSW_HWMON_ATTR_TYPE_TEMP_MODULE_LABEL:
+		mlxsw_hwmon_attr->dev_attr.show =
+			mlxsw_hwmon_module_temp_label_show;
+		mlxsw_hwmon_attr->dev_attr.attr.mode = 0444;
+		snprintf(mlxsw_hwmon_attr->name, sizeof(mlxsw_hwmon_attr->name),
+			 "temp%u_label", num + 1);
+		break;
+	case MLXSW_HWMON_ATTR_TYPE_TEMP_GBOX_LABEL:
+		mlxsw_hwmon_attr->dev_attr.show =
+			mlxsw_hwmon_gbox_temp_label_show;
+		mlxsw_hwmon_attr->dev_attr.attr.mode = 0444;
+		snprintf(mlxsw_hwmon_attr->name, sizeof(mlxsw_hwmon_attr->name),
+			 "temp%u_label", num + 1);
+		break;
+	case MLXSW_HWMON_ATTR_TYPE_TEMP_CRIT_ALARM:
+		mlxsw_hwmon_attr->dev_attr.show =
+			mlxsw_hwmon_temp_critical_alarm_show;
+		mlxsw_hwmon_attr->dev_attr.attr.mode = 0444;
+		snprintf(mlxsw_hwmon_attr->name, sizeof(mlxsw_hwmon_attr->name),
+			 "temp%u_crit_alarm", num + 1);
+		break;
+	case MLXSW_HWMON_ATTR_TYPE_TEMP_EMERGENCY_ALARM:
+		mlxsw_hwmon_attr->dev_attr.show =
+			mlxsw_hwmon_temp_emergency_alarm_show;
+		mlxsw_hwmon_attr->dev_attr.attr.mode = 0444;
+		snprintf(mlxsw_hwmon_attr->name, sizeof(mlxsw_hwmon_attr->name),
+			 "temp%u_emergency_alarm", num + 1);
+		break;
+	default:
+		WARN_ON(1);
+	}
+
+	mlxsw_hwmon_attr->type_index = type_index;
+	mlxsw_hwmon_attr->mlxsw_hwmon_dev = mlxsw_hwmon_dev;
+	mlxsw_hwmon_attr->dev_attr.attr.name = mlxsw_hwmon_attr->name;
+	sysfs_attr_init(&mlxsw_hwmon_attr->dev_attr.attr);
+
+	mlxsw_hwmon_dev->attrs[attr_index] = &mlxsw_hwmon_attr->dev_attr.attr;
+	mlxsw_hwmon_dev->attrs_count++;
+}
+
+static int mlxsw_hwmon_temp_init(struct mlxsw_hwmon_dev *mlxsw_hwmon_dev)
+{
+	struct mlxsw_hwmon *mlxsw_hwmon = mlxsw_hwmon_dev->hwmon;
+	char mtcap_pl[MLXSW_REG_MTCAP_LEN] = {0};
+	int i;
+	int err;
+
+	err = mlxsw_reg_query(mlxsw_hwmon->core, MLXSW_REG(mtcap), mtcap_pl);
+	if (err) {
+		dev_err(mlxsw_hwmon->bus_info->dev, "Failed to get number of temp sensors\n");
+		return err;
+	}
+	mlxsw_hwmon_dev->sensor_count = mlxsw_reg_mtcap_sensor_count_get(mtcap_pl);
+	for (i = 0; i < mlxsw_hwmon_dev->sensor_count; i++) {
+		char mtmp_pl[MLXSW_REG_MTMP_LEN] = {0};
+
+		mlxsw_reg_mtmp_slot_index_set(mtmp_pl,
+					      mlxsw_hwmon_dev->slot_index);
+		mlxsw_reg_mtmp_sensor_index_set(mtmp_pl, i);
+		err = mlxsw_reg_query(mlxsw_hwmon->core, MLXSW_REG(mtmp),
+				      mtmp_pl);
+		if (err)
+			return err;
+		mlxsw_reg_mtmp_mte_set(mtmp_pl, true);
+		mlxsw_reg_mtmp_mtr_set(mtmp_pl, true);
+		err = mlxsw_reg_write(mlxsw_hwmon->core,
+				      MLXSW_REG(mtmp), mtmp_pl);
+		if (err) {
+			dev_err(mlxsw_hwmon->bus_info->dev, "Failed to setup temp sensor number %d\n",
+				i);
+			return err;
+		}
+		mlxsw_hwmon_attr_add(mlxsw_hwmon_dev,
+				     MLXSW_HWMON_ATTR_TYPE_TEMP, i, i);
+		mlxsw_hwmon_attr_add(mlxsw_hwmon_dev,
+				     MLXSW_HWMON_ATTR_TYPE_TEMP_MAX, i, i);
+		mlxsw_hwmon_attr_add(mlxsw_hwmon_dev,
+				     MLXSW_HWMON_ATTR_TYPE_TEMP_RST, i, i);
+	}
+	return 0;
+}
+
+static int mlxsw_hwmon_fans_init(struct mlxsw_hwmon_dev *mlxsw_hwmon_dev)
+{
+	struct mlxsw_hwmon *mlxsw_hwmon = mlxsw_hwmon_dev->hwmon;
+	char mfcr_pl[MLXSW_REG_MFCR_LEN] = {0};
+	enum mlxsw_reg_mfcr_pwm_frequency freq;
+	unsigned int type_index;
+	unsigned int num;
+	u16 tacho_active;
+	u8 pwm_active;
+	int err;
+
+	err = mlxsw_reg_query(mlxsw_hwmon->core, MLXSW_REG(mfcr), mfcr_pl);
+	if (err) {
+		dev_err(mlxsw_hwmon->bus_info->dev, "Failed to get to probe PWMs and Tachometers\n");
+		return err;
+	}
+	mlxsw_reg_mfcr_unpack(mfcr_pl, &freq, &tacho_active, &pwm_active);
+	num = 0;
+	for (type_index = 0; type_index < MLXSW_MFCR_TACHOS_MAX; type_index++) {
+		if (tacho_active & BIT(type_index)) {
+			mlxsw_hwmon_attr_add(mlxsw_hwmon_dev,
+					     MLXSW_HWMON_ATTR_TYPE_FAN_RPM,
+					     type_index, num);
+			mlxsw_hwmon_attr_add(mlxsw_hwmon_dev,
+					     MLXSW_HWMON_ATTR_TYPE_FAN_FAULT,
+					     type_index, num++);
+		}
+	}
+	num = 0;
+	for (type_index = 0; type_index < MLXSW_MFCR_PWMS_MAX; type_index++) {
+		if (pwm_active & BIT(type_index))
+			mlxsw_hwmon_attr_add(mlxsw_hwmon_dev,
+					     MLXSW_HWMON_ATTR_TYPE_PWM,
+					     type_index, num++);
+	}
+	return 0;
+}
+
+static int mlxsw_hwmon_module_init(struct mlxsw_hwmon_dev *mlxsw_hwmon_dev)
+{
+	struct mlxsw_hwmon *mlxsw_hwmon = mlxsw_hwmon_dev->hwmon;
+	char mgpir_pl[MLXSW_REG_MGPIR_LEN];
+	u8 module_sensor_max;
+	int i, err;
+
+	mlxsw_reg_mgpir_pack(mgpir_pl, mlxsw_hwmon_dev->slot_index);
+	err = mlxsw_reg_query(mlxsw_hwmon->core, MLXSW_REG(mgpir), mgpir_pl);
+	if (err)
+		return err;
+
+	mlxsw_reg_mgpir_unpack(mgpir_pl, NULL, NULL, NULL,
+			       &module_sensor_max, NULL);
+
+	/* Add extra attributes for module temperature. Sensor index is
+	 * assigned to sensor_count value, while all indexed before
+	 * sensor_count are already utilized by the sensors connected through
+	 * mtmp register by mlxsw_hwmon_temp_init().
+	 */
+	mlxsw_hwmon_dev->module_sensor_max = mlxsw_hwmon_dev->sensor_count +
+					     module_sensor_max;
+	for (i = mlxsw_hwmon_dev->sensor_count;
+	     i < mlxsw_hwmon_dev->module_sensor_max; i++) {
+		mlxsw_hwmon_attr_add(mlxsw_hwmon_dev,
+				     MLXSW_HWMON_ATTR_TYPE_TEMP_MODULE, i, i);
+		mlxsw_hwmon_attr_add(mlxsw_hwmon_dev,
+				     MLXSW_HWMON_ATTR_TYPE_TEMP_MODULE_FAULT,
+				     i, i);
+		mlxsw_hwmon_attr_add(mlxsw_hwmon_dev,
+				     MLXSW_HWMON_ATTR_TYPE_TEMP_MODULE_CRIT, i,
+				     i);
+		mlxsw_hwmon_attr_add(mlxsw_hwmon_dev,
+				     MLXSW_HWMON_ATTR_TYPE_TEMP_MODULE_EMERG,
+				     i, i);
+		mlxsw_hwmon_attr_add(mlxsw_hwmon_dev,
+				     MLXSW_HWMON_ATTR_TYPE_TEMP_MODULE_LABEL,
+				     i, i);
+		mlxsw_hwmon_attr_add(mlxsw_hwmon_dev,
+				     MLXSW_HWMON_ATTR_TYPE_TEMP_CRIT_ALARM,
+				     i, i);
+		mlxsw_hwmon_attr_add(mlxsw_hwmon_dev,
+				     MLXSW_HWMON_ATTR_TYPE_TEMP_EMERGENCY_ALARM,
+				     i, i);
+	}
+
+	return 0;
+}
+
+static int mlxsw_hwmon_gearbox_init(struct mlxsw_hwmon_dev *mlxsw_hwmon_dev)
+{
+	struct mlxsw_hwmon *mlxsw_hwmon = mlxsw_hwmon_dev->hwmon;
+	enum mlxsw_reg_mgpir_device_type device_type;
+	int index, max_index, sensor_index;
+	char mgpir_pl[MLXSW_REG_MGPIR_LEN];
+	char mtmp_pl[MLXSW_REG_MTMP_LEN];
+	u8 gbox_num;
+	int err;
+
+	mlxsw_reg_mgpir_pack(mgpir_pl, mlxsw_hwmon_dev->slot_index);
+	err = mlxsw_reg_query(mlxsw_hwmon->core, MLXSW_REG(mgpir), mgpir_pl);
+	if (err)
+		return err;
+
+	mlxsw_reg_mgpir_unpack(mgpir_pl, &gbox_num, &device_type, NULL, NULL,
+			       NULL);
+	if (device_type != MLXSW_REG_MGPIR_DEVICE_TYPE_GEARBOX_DIE ||
+	    !gbox_num)
+		return 0;
+
+	index = mlxsw_hwmon_dev->module_sensor_max;
+	max_index = mlxsw_hwmon_dev->module_sensor_max + gbox_num;
+	while (index < max_index) {
+		sensor_index = index % mlxsw_hwmon_dev->module_sensor_max +
+			       MLXSW_REG_MTMP_GBOX_INDEX_MIN;
+		mlxsw_reg_mtmp_pack(mtmp_pl, mlxsw_hwmon_dev->slot_index,
+				    sensor_index, true, true);
+		err = mlxsw_reg_write(mlxsw_hwmon->core,
+				      MLXSW_REG(mtmp), mtmp_pl);
+		if (err) {
+			dev_err(mlxsw_hwmon->bus_info->dev, "Failed to setup temp sensor number %d\n",
+				sensor_index);
+			return err;
+		}
+		mlxsw_hwmon_attr_add(mlxsw_hwmon_dev,
+				     MLXSW_HWMON_ATTR_TYPE_TEMP, index, index);
+		mlxsw_hwmon_attr_add(mlxsw_hwmon_dev,
+				     MLXSW_HWMON_ATTR_TYPE_TEMP_MAX, index,
+				     index);
+		mlxsw_hwmon_attr_add(mlxsw_hwmon_dev,
+				     MLXSW_HWMON_ATTR_TYPE_TEMP_RST, index,
+				     index);
+		mlxsw_hwmon_attr_add(mlxsw_hwmon_dev,
+				     MLXSW_HWMON_ATTR_TYPE_TEMP_GBOX_LABEL,
+				     index, index);
+		index++;
+	}
+
+	return 0;
+}
+
+static void
+mlxsw_hwmon_got_active(struct mlxsw_core *mlxsw_core, u8 slot_index,
+		       void *priv)
+{
+	struct mlxsw_hwmon *hwmon = priv;
+	struct mlxsw_hwmon_dev *linecard;
+	struct device *dev;
+	int err;
+
+	dev = hwmon->bus_info->dev;
+	linecard = &hwmon->line_cards[slot_index];
+	if (linecard->active)
+		return;
+	/* For the main board, module sensor indexes start from 1, sensor index
+	 * 0 is used for the ASIC. Use the same numbering for line cards.
+	 */
+	linecard->sensor_count = 1;
+	linecard->slot_index = slot_index;
+	linecard->hwmon = hwmon;
+	err = mlxsw_hwmon_module_init(linecard);
+	if (err) {
+		dev_err(dev, "Failed to configure hwmon objects for line card modules in slot %d\n",
+			slot_index);
+		return;
+	}
+
+	err = mlxsw_hwmon_gearbox_init(linecard);
+	if (err) {
+		dev_err(dev, "Failed to configure hwmon objects for line card gearboxes in slot %d\n",
+			slot_index);
+		return;
+	}
+
+	linecard->groups[0] = &linecard->group;
+	linecard->group.attrs = linecard->attrs;
+	sprintf(linecard->name, "%s#%02u", "linecard", slot_index);
+	linecard->hwmon_dev =
+		hwmon_device_register_with_groups(dev, linecard->name,
+						  linecard, linecard->groups);
+	if (IS_ERR(linecard->hwmon_dev)) {
+		dev_err(dev, "Failed to register hwmon objects for line card in slot %d\n",
+			slot_index);
+		return;
+	}
+
+	linecard->active = true;
+}
+
+static void
+mlxsw_hwmon_got_inactive(struct mlxsw_core *mlxsw_core, u8 slot_index,
+			 void *priv)
+{
+	struct mlxsw_hwmon *hwmon = priv;
+	struct mlxsw_hwmon_dev *linecard;
+
+	linecard = &hwmon->line_cards[slot_index];
+	if (!linecard->active)
+		return;
+	linecard->active = false;
+	hwmon_device_unregister(linecard->hwmon_dev);
+	/* Reset attributes counter */
+	linecard->attrs_count = 0;
+}
+
+static struct mlxsw_linecards_event_ops mlxsw_hwmon_event_ops = {
+	.got_active = mlxsw_hwmon_got_active,
+	.got_inactive = mlxsw_hwmon_got_inactive,
+};
+
+int mlxsw_hwmon_init(struct mlxsw_core *mlxsw_core,
+		     const struct mlxsw_bus_info *mlxsw_bus_info,
+		     struct mlxsw_hwmon **p_hwmon)
+{
+	char mgpir_pl[MLXSW_REG_MGPIR_LEN];
+	struct mlxsw_hwmon *mlxsw_hwmon;
+	struct device *hwmon_dev;
+	u8 num_of_slots;
+	int err;
+
+	mlxsw_reg_mgpir_pack(mgpir_pl, 0);
+	err = mlxsw_reg_query(mlxsw_core, MLXSW_REG(mgpir), mgpir_pl);
+	if (err)
+		return err;
+
+	mlxsw_reg_mgpir_unpack(mgpir_pl, NULL, NULL, NULL, NULL,
+			       &num_of_slots);
+
+	mlxsw_hwmon = kzalloc(struct_size(mlxsw_hwmon, line_cards,
+					  num_of_slots + 1), GFP_KERNEL);
+	if (!mlxsw_hwmon)
+		return -ENOMEM;
+
+	mlxsw_hwmon->core = mlxsw_core;
+	mlxsw_hwmon->bus_info = mlxsw_bus_info;
+	mlxsw_hwmon->line_cards[0].hwmon = mlxsw_hwmon;
+	mlxsw_hwmon->line_cards[0].slot_index = 0;
+
+	err = mlxsw_hwmon_temp_init(&mlxsw_hwmon->line_cards[0]);
+	if (err)
+		goto err_temp_init;
+
+	err = mlxsw_hwmon_fans_init(&mlxsw_hwmon->line_cards[0]);
+	if (err)
+		goto err_fans_init;
+
+	err = mlxsw_hwmon_module_init(&mlxsw_hwmon->line_cards[0]);
+	if (err)
+		goto err_temp_module_init;
+
+	err = mlxsw_hwmon_gearbox_init(&mlxsw_hwmon->line_cards[0]);
+	if (err)
+		goto err_temp_gearbox_init;
+
+	mlxsw_hwmon->line_cards[0].groups[0] = &mlxsw_hwmon->line_cards[0].group;
+	mlxsw_hwmon->line_cards[0].group.attrs = mlxsw_hwmon->line_cards[0].attrs;
+
+	hwmon_dev = hwmon_device_register_with_groups(mlxsw_bus_info->dev,
+						      "mlxsw",
+						      &mlxsw_hwmon->line_cards[0],
+						      mlxsw_hwmon->line_cards[0].groups);
+	if (IS_ERR(hwmon_dev)) {
+		err = PTR_ERR(hwmon_dev);
+		goto err_hwmon_register;
+	}
+
+	err = mlxsw_linecards_event_ops_register(mlxsw_hwmon->core,
+						 &mlxsw_hwmon_event_ops,
+						 mlxsw_hwmon);
+	if (err)
+		goto err_linecards_event_ops_register;
+
+	mlxsw_hwmon->line_cards[0].hwmon_dev = hwmon_dev;
+	mlxsw_hwmon->line_cards[0].active = true;
+	*p_hwmon = mlxsw_hwmon;
+	return 0;
+
+err_linecards_event_ops_register:
+	hwmon_device_unregister(mlxsw_hwmon->line_cards[0].hwmon_dev);
+err_hwmon_register:
+err_temp_gearbox_init:
+err_temp_module_init:
+err_fans_init:
+err_temp_init:
+	kfree(mlxsw_hwmon);
+	return err;
+}
+
+void mlxsw_hwmon_fini(struct mlxsw_hwmon *mlxsw_hwmon)
+{
+	mlxsw_hwmon->line_cards[0].active = false;
+	mlxsw_linecards_event_ops_unregister(mlxsw_hwmon->core,
+					     &mlxsw_hwmon_event_ops, mlxsw_hwmon);
+	hwmon_device_unregister(mlxsw_hwmon->line_cards[0].hwmon_dev);
+	kfree(mlxsw_hwmon);
+}
diff --git a/drivers/net/ethernet/mellanox/mlxsw/core_linecard_dev.c b/drivers/net/ethernet/mellanox/mlxsw/core_linecard_dev.c
new file mode 100644
index 000000000..af37e650a
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/core_linecard_dev.c
@@ -0,0 +1,183 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2022 NVIDIA Corporation and Mellanox Technologies. All rights reserved */
+
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/err.h>
+#include <linux/types.h>
+#include <linux/auxiliary_bus.h>
+#include <linux/idr.h>
+#include <linux/gfp.h>
+#include <linux/slab.h>
+#include <net/devlink.h>
+#include "core.h"
+
+#define MLXSW_LINECARD_DEV_ID_NAME "lc"
+
+struct mlxsw_linecard_dev {
+	struct mlxsw_linecard *linecard;
+};
+
+struct mlxsw_linecard_bdev {
+	struct auxiliary_device adev;
+	struct mlxsw_linecard *linecard;
+	struct mlxsw_linecard_dev *linecard_dev;
+};
+
+static DEFINE_IDA(mlxsw_linecard_bdev_ida);
+
+static int mlxsw_linecard_bdev_id_alloc(void)
+{
+	return ida_alloc(&mlxsw_linecard_bdev_ida, GFP_KERNEL);
+}
+
+static void mlxsw_linecard_bdev_id_free(int id)
+{
+	ida_free(&mlxsw_linecard_bdev_ida, id);
+}
+
+static void mlxsw_linecard_bdev_release(struct device *device)
+{
+	struct auxiliary_device *adev =
+			container_of(device, struct auxiliary_device, dev);
+	struct mlxsw_linecard_bdev *linecard_bdev =
+			container_of(adev, struct mlxsw_linecard_bdev, adev);
+
+	mlxsw_linecard_bdev_id_free(adev->id);
+	kfree(linecard_bdev);
+}
+
+int mlxsw_linecard_bdev_add(struct mlxsw_linecard *linecard)
+{
+	struct mlxsw_linecard_bdev *linecard_bdev;
+	int err;
+	int id;
+
+	id = mlxsw_linecard_bdev_id_alloc();
+	if (id < 0)
+		return id;
+
+	linecard_bdev = kzalloc(sizeof(*linecard_bdev), GFP_KERNEL);
+	if (!linecard_bdev) {
+		mlxsw_linecard_bdev_id_free(id);
+		return -ENOMEM;
+	}
+	linecard_bdev->adev.id = id;
+	linecard_bdev->adev.name = MLXSW_LINECARD_DEV_ID_NAME;
+	linecard_bdev->adev.dev.release = mlxsw_linecard_bdev_release;
+	linecard_bdev->adev.dev.parent = linecard->linecards->bus_info->dev;
+	linecard_bdev->linecard = linecard;
+
+	err = auxiliary_device_init(&linecard_bdev->adev);
+	if (err) {
+		mlxsw_linecard_bdev_id_free(id);
+		kfree(linecard_bdev);
+		return err;
+	}
+
+	err = auxiliary_device_add(&linecard_bdev->adev);
+	if (err) {
+		auxiliary_device_uninit(&linecard_bdev->adev);
+		return err;
+	}
+
+	linecard->bdev = linecard_bdev;
+	return 0;
+}
+
+void mlxsw_linecard_bdev_del(struct mlxsw_linecard *linecard)
+{
+	struct mlxsw_linecard_bdev *linecard_bdev = linecard->bdev;
+
+	if (!linecard_bdev)
+		/* Unprovisioned line cards do not have an auxiliary device. */
+		return;
+	auxiliary_device_delete(&linecard_bdev->adev);
+	auxiliary_device_uninit(&linecard_bdev->adev);
+	linecard->bdev = NULL;
+}
+
+static int mlxsw_linecard_dev_devlink_info_get(struct devlink *devlink,
+					       struct devlink_info_req *req,
+					       struct netlink_ext_ack *extack)
+{
+	struct mlxsw_linecard_dev *linecard_dev = devlink_priv(devlink);
+	struct mlxsw_linecard *linecard = linecard_dev->linecard;
+
+	return mlxsw_linecard_devlink_info_get(linecard, req, extack);
+}
+
+static int
+mlxsw_linecard_dev_devlink_flash_update(struct devlink *devlink,
+					struct devlink_flash_update_params *params,
+					struct netlink_ext_ack *extack)
+{
+	struct mlxsw_linecard_dev *linecard_dev = devlink_priv(devlink);
+	struct mlxsw_linecard *linecard = linecard_dev->linecard;
+
+	return mlxsw_linecard_flash_update(devlink, linecard,
+					   params->fw, extack);
+}
+
+static const struct devlink_ops mlxsw_linecard_dev_devlink_ops = {
+	.info_get			= mlxsw_linecard_dev_devlink_info_get,
+	.flash_update			= mlxsw_linecard_dev_devlink_flash_update,
+};
+
+static int mlxsw_linecard_bdev_probe(struct auxiliary_device *adev,
+				     const struct auxiliary_device_id *id)
+{
+	struct mlxsw_linecard_bdev *linecard_bdev =
+			container_of(adev, struct mlxsw_linecard_bdev, adev);
+	struct mlxsw_linecard *linecard = linecard_bdev->linecard;
+	struct mlxsw_linecard_dev *linecard_dev;
+	struct devlink *devlink;
+
+	devlink = devlink_alloc(&mlxsw_linecard_dev_devlink_ops,
+				sizeof(*linecard_dev), &adev->dev);
+	if (!devlink)
+		return -ENOMEM;
+	linecard_dev = devlink_priv(devlink);
+	linecard_dev->linecard = linecard_bdev->linecard;
+	linecard_bdev->linecard_dev = linecard_dev;
+
+	devlink_register(devlink);
+	devlink_linecard_nested_dl_set(linecard->devlink_linecard, devlink);
+	return 0;
+}
+
+static void mlxsw_linecard_bdev_remove(struct auxiliary_device *adev)
+{
+	struct mlxsw_linecard_bdev *linecard_bdev =
+			container_of(adev, struct mlxsw_linecard_bdev, adev);
+	struct devlink *devlink = priv_to_devlink(linecard_bdev->linecard_dev);
+	struct mlxsw_linecard *linecard = linecard_bdev->linecard;
+
+	devlink_linecard_nested_dl_set(linecard->devlink_linecard, NULL);
+	devlink_unregister(devlink);
+	devlink_free(devlink);
+}
+
+static const struct auxiliary_device_id mlxsw_linecard_bdev_id_table[] = {
+	{ .name = KBUILD_MODNAME "." MLXSW_LINECARD_DEV_ID_NAME },
+	{},
+};
+
+MODULE_DEVICE_TABLE(auxiliary, mlxsw_linecard_bdev_id_table);
+
+static struct auxiliary_driver mlxsw_linecard_driver = {
+	.name = MLXSW_LINECARD_DEV_ID_NAME,
+	.probe = mlxsw_linecard_bdev_probe,
+	.remove = mlxsw_linecard_bdev_remove,
+	.id_table = mlxsw_linecard_bdev_id_table,
+};
+
+int mlxsw_linecard_driver_register(void)
+{
+	return auxiliary_driver_register(&mlxsw_linecard_driver);
+}
+
+void mlxsw_linecard_driver_unregister(void)
+{
+	auxiliary_driver_unregister(&mlxsw_linecard_driver);
+}
diff --git a/drivers/net/ethernet/mellanox/mlxsw/core_linecards.c b/drivers/net/ethernet/mellanox/mlxsw/core_linecards.c
new file mode 100644
index 000000000..83d2dc91b
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/core_linecards.c
@@ -0,0 +1,1601 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2022 NVIDIA Corporation and Mellanox Technologies. All rights reserved */
+
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/err.h>
+#include <linux/types.h>
+#include <linux/string.h>
+#include <linux/workqueue.h>
+#include <linux/gfp.h>
+#include <linux/slab.h>
+#include <linux/list.h>
+#include <linux/vmalloc.h>
+
+#include "core.h"
+#include "../mlxfw/mlxfw.h"
+
+struct mlxsw_linecard_ini_file {
+	__le16 size;
+	union {
+		u8 data[0];
+		struct {
+			__be16 hw_revision;
+			__be16 ini_version;
+			u8 __dontcare[3];
+			u8 type;
+			u8 name[20];
+		} format;
+	};
+};
+
+struct mlxsw_linecard_types_info {
+	struct mlxsw_linecard_ini_file **ini_files;
+	unsigned int count;
+	size_t data_size;
+	char *data;
+};
+
+#define MLXSW_LINECARD_STATUS_EVENT_TO (10 * MSEC_PER_SEC)
+
+static void
+mlxsw_linecard_status_event_to_schedule(struct mlxsw_linecard *linecard,
+					enum mlxsw_linecard_status_event_type status_event_type)
+{
+	cancel_delayed_work_sync(&linecard->status_event_to_dw);
+	linecard->status_event_type_to = status_event_type;
+	mlxsw_core_schedule_dw(&linecard->status_event_to_dw,
+			       msecs_to_jiffies(MLXSW_LINECARD_STATUS_EVENT_TO));
+}
+
+static void
+mlxsw_linecard_status_event_done(struct mlxsw_linecard *linecard,
+				 enum mlxsw_linecard_status_event_type status_event_type)
+{
+	if (linecard->status_event_type_to == status_event_type)
+		cancel_delayed_work_sync(&linecard->status_event_to_dw);
+}
+
+static const char *
+mlxsw_linecard_types_lookup(struct mlxsw_linecards *linecards, u8 card_type)
+{
+	struct mlxsw_linecard_types_info *types_info;
+	struct mlxsw_linecard_ini_file *ini_file;
+	int i;
+
+	types_info = linecards->types_info;
+	if (!types_info)
+		return NULL;
+	for (i = 0; i < types_info->count; i++) {
+		ini_file = linecards->types_info->ini_files[i];
+		if (ini_file->format.type == card_type)
+			return ini_file->format.name;
+	}
+	return NULL;
+}
+
+static const char *mlxsw_linecard_type_name(struct mlxsw_linecard *linecard)
+{
+	struct mlxsw_core *mlxsw_core = linecard->linecards->mlxsw_core;
+	char mddq_pl[MLXSW_REG_MDDQ_LEN];
+	int err;
+
+	mlxsw_reg_mddq_slot_name_pack(mddq_pl, linecard->slot_index);
+	err = mlxsw_reg_query(mlxsw_core, MLXSW_REG(mddq), mddq_pl);
+	if (err)
+		return ERR_PTR(err);
+	mlxsw_reg_mddq_slot_name_unpack(mddq_pl, linecard->name);
+	return linecard->name;
+}
+
+struct mlxsw_linecard_device_fw_info {
+	struct mlxfw_dev mlxfw_dev;
+	struct mlxsw_core *mlxsw_core;
+	struct mlxsw_linecard *linecard;
+};
+
+static int mlxsw_linecard_device_fw_component_query(struct mlxfw_dev *mlxfw_dev,
+						    u16 component_index,
+						    u32 *p_max_size,
+						    u8 *p_align_bits,
+						    u16 *p_max_write_size)
+{
+	struct mlxsw_linecard_device_fw_info *info =
+		container_of(mlxfw_dev, struct mlxsw_linecard_device_fw_info,
+			     mlxfw_dev);
+	struct mlxsw_linecard *linecard = info->linecard;
+	struct mlxsw_core *mlxsw_core = info->mlxsw_core;
+	char mddt_pl[MLXSW_REG_MDDT_LEN];
+	char *mcqi_pl;
+	int err;
+
+	mlxsw_reg_mddt_pack(mddt_pl, linecard->slot_index,
+			    linecard->device.index,
+			    MLXSW_REG_MDDT_METHOD_QUERY,
+			    MLXSW_REG(mcqi), &mcqi_pl);
+
+	mlxsw_reg_mcqi_pack(mcqi_pl, component_index);
+	err = mlxsw_reg_query(mlxsw_core, MLXSW_REG(mddt), mddt_pl);
+	if (err)
+		return err;
+	mlxsw_reg_mcqi_unpack(mcqi_pl, p_max_size, p_align_bits,
+			      p_max_write_size);
+
+	*p_align_bits = max_t(u8, *p_align_bits, 2);
+	*p_max_write_size = min_t(u16, *p_max_write_size,
+				  MLXSW_REG_MCDA_MAX_DATA_LEN);
+	return 0;
+}
+
+static int mlxsw_linecard_device_fw_fsm_lock(struct mlxfw_dev *mlxfw_dev,
+					     u32 *fwhandle)
+{
+	struct mlxsw_linecard_device_fw_info *info =
+		container_of(mlxfw_dev, struct mlxsw_linecard_device_fw_info,
+			     mlxfw_dev);
+	struct mlxsw_linecard *linecard = info->linecard;
+	struct mlxsw_core *mlxsw_core = info->mlxsw_core;
+	char mddt_pl[MLXSW_REG_MDDT_LEN];
+	u8 control_state;
+	char *mcc_pl;
+	int err;
+
+	mlxsw_reg_mddt_pack(mddt_pl, linecard->slot_index,
+			    linecard->device.index,
+			    MLXSW_REG_MDDT_METHOD_QUERY,
+			    MLXSW_REG(mcc), &mcc_pl);
+	mlxsw_reg_mcc_pack(mcc_pl, 0, 0, 0, 0);
+	err = mlxsw_reg_query(mlxsw_core, MLXSW_REG(mddt), mddt_pl);
+	if (err)
+		return err;
+
+	mlxsw_reg_mcc_unpack(mcc_pl, fwhandle, NULL, &control_state);
+	if (control_state != MLXFW_FSM_STATE_IDLE)
+		return -EBUSY;
+
+	mlxsw_reg_mddt_pack(mddt_pl, linecard->slot_index,
+			    linecard->device.index,
+			    MLXSW_REG_MDDT_METHOD_WRITE,
+			    MLXSW_REG(mcc), &mcc_pl);
+	mlxsw_reg_mcc_pack(mcc_pl, MLXSW_REG_MCC_INSTRUCTION_LOCK_UPDATE_HANDLE,
+			   0, *fwhandle, 0);
+	return mlxsw_reg_write(mlxsw_core, MLXSW_REG(mddt), mddt_pl);
+}
+
+static int
+mlxsw_linecard_device_fw_fsm_component_update(struct mlxfw_dev *mlxfw_dev,
+					      u32 fwhandle,
+					      u16 component_index,
+					      u32 component_size)
+{
+	struct mlxsw_linecard_device_fw_info *info =
+		container_of(mlxfw_dev, struct mlxsw_linecard_device_fw_info,
+			     mlxfw_dev);
+	struct mlxsw_linecard *linecard = info->linecard;
+	struct mlxsw_core *mlxsw_core = info->mlxsw_core;
+	char mddt_pl[MLXSW_REG_MDDT_LEN];
+	char *mcc_pl;
+
+	mlxsw_reg_mddt_pack(mddt_pl, linecard->slot_index,
+			    linecard->device.index,
+			    MLXSW_REG_MDDT_METHOD_WRITE,
+			    MLXSW_REG(mcc), &mcc_pl);
+	mlxsw_reg_mcc_pack(mcc_pl, MLXSW_REG_MCC_INSTRUCTION_UPDATE_COMPONENT,
+			   component_index, fwhandle, component_size);
+	return mlxsw_reg_write(mlxsw_core, MLXSW_REG(mddt), mddt_pl);
+}
+
+static int
+mlxsw_linecard_device_fw_fsm_block_download(struct mlxfw_dev *mlxfw_dev,
+					    u32 fwhandle, u8 *data,
+					    u16 size, u32 offset)
+{
+	struct mlxsw_linecard_device_fw_info *info =
+		container_of(mlxfw_dev, struct mlxsw_linecard_device_fw_info,
+			     mlxfw_dev);
+	struct mlxsw_linecard *linecard = info->linecard;
+	struct mlxsw_core *mlxsw_core = info->mlxsw_core;
+	char mddt_pl[MLXSW_REG_MDDT_LEN];
+	char *mcda_pl;
+
+	mlxsw_reg_mddt_pack(mddt_pl, linecard->slot_index,
+			    linecard->device.index,
+			    MLXSW_REG_MDDT_METHOD_WRITE,
+			    MLXSW_REG(mcda), &mcda_pl);
+	mlxsw_reg_mcda_pack(mcda_pl, fwhandle, offset, size, data);
+	return mlxsw_reg_write(mlxsw_core, MLXSW_REG(mddt), mddt_pl);
+}
+
+static int
+mlxsw_linecard_device_fw_fsm_component_verify(struct mlxfw_dev *mlxfw_dev,
+					      u32 fwhandle, u16 component_index)
+{
+	struct mlxsw_linecard_device_fw_info *info =
+		container_of(mlxfw_dev, struct mlxsw_linecard_device_fw_info,
+			     mlxfw_dev);
+	struct mlxsw_linecard *linecard = info->linecard;
+	struct mlxsw_core *mlxsw_core = info->mlxsw_core;
+	char mddt_pl[MLXSW_REG_MDDT_LEN];
+	char *mcc_pl;
+
+	mlxsw_reg_mddt_pack(mddt_pl, linecard->slot_index,
+			    linecard->device.index,
+			    MLXSW_REG_MDDT_METHOD_WRITE,
+			    MLXSW_REG(mcc), &mcc_pl);
+	mlxsw_reg_mcc_pack(mcc_pl, MLXSW_REG_MCC_INSTRUCTION_VERIFY_COMPONENT,
+			   component_index, fwhandle, 0);
+	return mlxsw_reg_write(mlxsw_core, MLXSW_REG(mddt), mddt_pl);
+}
+
+static int mlxsw_linecard_device_fw_fsm_activate(struct mlxfw_dev *mlxfw_dev,
+						 u32 fwhandle)
+{
+	struct mlxsw_linecard_device_fw_info *info =
+		container_of(mlxfw_dev, struct mlxsw_linecard_device_fw_info,
+			     mlxfw_dev);
+	struct mlxsw_linecard *linecard = info->linecard;
+	struct mlxsw_core *mlxsw_core = info->mlxsw_core;
+	char mddt_pl[MLXSW_REG_MDDT_LEN];
+	char *mcc_pl;
+
+	mlxsw_reg_mddt_pack(mddt_pl, linecard->slot_index,
+			    linecard->device.index,
+			    MLXSW_REG_MDDT_METHOD_WRITE,
+			    MLXSW_REG(mcc), &mcc_pl);
+	mlxsw_reg_mcc_pack(mcc_pl, MLXSW_REG_MCC_INSTRUCTION_ACTIVATE,
+			   0, fwhandle, 0);
+	return mlxsw_reg_write(mlxsw_core, MLXSW_REG(mddt), mddt_pl);
+}
+
+static int
+mlxsw_linecard_device_fw_fsm_query_state(struct mlxfw_dev *mlxfw_dev,
+					 u32 fwhandle,
+					 enum mlxfw_fsm_state *fsm_state,
+					 enum mlxfw_fsm_state_err *fsm_state_err)
+{
+	struct mlxsw_linecard_device_fw_info *info =
+		container_of(mlxfw_dev, struct mlxsw_linecard_device_fw_info,
+			     mlxfw_dev);
+	struct mlxsw_linecard *linecard = info->linecard;
+	struct mlxsw_core *mlxsw_core = info->mlxsw_core;
+	char mddt_pl[MLXSW_REG_MDDT_LEN];
+	u8 control_state;
+	u8 error_code;
+	char *mcc_pl;
+	int err;
+
+	mlxsw_reg_mddt_pack(mddt_pl, linecard->slot_index,
+			    linecard->device.index,
+			    MLXSW_REG_MDDT_METHOD_QUERY,
+			    MLXSW_REG(mcc), &mcc_pl);
+	mlxsw_reg_mcc_pack(mcc_pl, 0, 0, fwhandle, 0);
+	err = mlxsw_reg_query(mlxsw_core, MLXSW_REG(mddt), mddt_pl);
+	if (err)
+		return err;
+
+	mlxsw_reg_mcc_unpack(mcc_pl, NULL, &error_code, &control_state);
+	*fsm_state = control_state;
+	*fsm_state_err = min_t(enum mlxfw_fsm_state_err, error_code,
+			       MLXFW_FSM_STATE_ERR_MAX);
+	return 0;
+}
+
+static void mlxsw_linecard_device_fw_fsm_cancel(struct mlxfw_dev *mlxfw_dev,
+						u32 fwhandle)
+{
+	struct mlxsw_linecard_device_fw_info *info =
+		container_of(mlxfw_dev, struct mlxsw_linecard_device_fw_info,
+			     mlxfw_dev);
+	struct mlxsw_linecard *linecard = info->linecard;
+	struct mlxsw_core *mlxsw_core = info->mlxsw_core;
+	char mddt_pl[MLXSW_REG_MDDT_LEN];
+	char *mcc_pl;
+
+	mlxsw_reg_mddt_pack(mddt_pl, linecard->slot_index,
+			    linecard->device.index,
+			    MLXSW_REG_MDDT_METHOD_WRITE,
+			    MLXSW_REG(mcc), &mcc_pl);
+	mlxsw_reg_mcc_pack(mcc_pl, MLXSW_REG_MCC_INSTRUCTION_CANCEL,
+			   0, fwhandle, 0);
+	mlxsw_reg_write(mlxsw_core, MLXSW_REG(mddt), mddt_pl);
+}
+
+static void mlxsw_linecard_device_fw_fsm_release(struct mlxfw_dev *mlxfw_dev,
+						 u32 fwhandle)
+{
+	struct mlxsw_linecard_device_fw_info *info =
+		container_of(mlxfw_dev, struct mlxsw_linecard_device_fw_info,
+			     mlxfw_dev);
+	struct mlxsw_linecard *linecard = info->linecard;
+	struct mlxsw_core *mlxsw_core = info->mlxsw_core;
+	char mddt_pl[MLXSW_REG_MDDT_LEN];
+	char *mcc_pl;
+
+	mlxsw_reg_mddt_pack(mddt_pl, linecard->slot_index,
+			    linecard->device.index,
+			    MLXSW_REG_MDDT_METHOD_WRITE,
+			    MLXSW_REG(mcc), &mcc_pl);
+	mlxsw_reg_mcc_pack(mcc_pl,
+			   MLXSW_REG_MCC_INSTRUCTION_RELEASE_UPDATE_HANDLE,
+			   0, fwhandle, 0);
+	mlxsw_reg_write(mlxsw_core, MLXSW_REG(mddt), mddt_pl);
+}
+
+static const struct mlxfw_dev_ops mlxsw_linecard_device_dev_ops = {
+	.component_query	= mlxsw_linecard_device_fw_component_query,
+	.fsm_lock		= mlxsw_linecard_device_fw_fsm_lock,
+	.fsm_component_update	= mlxsw_linecard_device_fw_fsm_component_update,
+	.fsm_block_download	= mlxsw_linecard_device_fw_fsm_block_download,
+	.fsm_component_verify	= mlxsw_linecard_device_fw_fsm_component_verify,
+	.fsm_activate		= mlxsw_linecard_device_fw_fsm_activate,
+	.fsm_query_state	= mlxsw_linecard_device_fw_fsm_query_state,
+	.fsm_cancel		= mlxsw_linecard_device_fw_fsm_cancel,
+	.fsm_release		= mlxsw_linecard_device_fw_fsm_release,
+};
+
+int mlxsw_linecard_flash_update(struct devlink *linecard_devlink,
+				struct mlxsw_linecard *linecard,
+				const struct firmware *firmware,
+				struct netlink_ext_ack *extack)
+{
+	struct mlxsw_core *mlxsw_core = linecard->linecards->mlxsw_core;
+	struct mlxsw_linecard_device_fw_info info = {
+		.mlxfw_dev = {
+			.ops = &mlxsw_linecard_device_dev_ops,
+			.psid = linecard->device.info.psid,
+			.psid_size = strlen(linecard->device.info.psid),
+			.devlink = linecard_devlink,
+		},
+		.mlxsw_core = mlxsw_core,
+		.linecard = linecard,
+	};
+	int err;
+
+	mutex_lock(&linecard->lock);
+	if (!linecard->active) {
+		NL_SET_ERR_MSG_MOD(extack, "Only active line cards can be flashed");
+		err = -EINVAL;
+		goto unlock;
+	}
+	err = mlxsw_core_fw_flash(mlxsw_core, &info.mlxfw_dev,
+				  firmware, extack);
+unlock:
+	mutex_unlock(&linecard->lock);
+	return err;
+}
+
+static int mlxsw_linecard_device_psid_get(struct mlxsw_linecard *linecard,
+					  u8 device_index, char *psid)
+{
+	struct mlxsw_core *mlxsw_core = linecard->linecards->mlxsw_core;
+	char mddt_pl[MLXSW_REG_MDDT_LEN];
+	char *mgir_pl;
+	int err;
+
+	mlxsw_reg_mddt_pack(mddt_pl, linecard->slot_index, device_index,
+			    MLXSW_REG_MDDT_METHOD_QUERY,
+			    MLXSW_REG(mgir), &mgir_pl);
+
+	mlxsw_reg_mgir_pack(mgir_pl);
+	err = mlxsw_reg_query(mlxsw_core, MLXSW_REG(mddt), mddt_pl);
+	if (err)
+		return err;
+
+	mlxsw_reg_mgir_fw_info_psid_memcpy_from(mgir_pl, psid);
+	return 0;
+}
+
+static int mlxsw_linecard_device_info_update(struct mlxsw_linecard *linecard)
+{
+	struct mlxsw_core *mlxsw_core = linecard->linecards->mlxsw_core;
+	bool flashable_found = false;
+	u8 msg_seq = 0;
+
+	do {
+		struct mlxsw_linecard_device_info info;
+		char mddq_pl[MLXSW_REG_MDDQ_LEN];
+		bool flash_owner;
+		bool data_valid;
+		u8 device_index;
+		int err;
+
+		mlxsw_reg_mddq_device_info_pack(mddq_pl, linecard->slot_index,
+						msg_seq);
+		err = mlxsw_reg_query(mlxsw_core, MLXSW_REG(mddq), mddq_pl);
+		if (err)
+			return err;
+		mlxsw_reg_mddq_device_info_unpack(mddq_pl, &msg_seq,
+						  &data_valid, &flash_owner,
+						  &device_index,
+						  &info.fw_major,
+						  &info.fw_minor,
+						  &info.fw_sub_minor);
+		if (!data_valid)
+			break;
+		if (!flash_owner) /* We care only about flashable ones. */
+			continue;
+		if (flashable_found) {
+			dev_warn_once(linecard->linecards->bus_info->dev, "linecard %u: More flashable devices present, exposing only the first one\n",
+				      linecard->slot_index);
+			return 0;
+		}
+
+		err = mlxsw_linecard_device_psid_get(linecard, device_index,
+						     info.psid);
+		if (err)
+			return err;
+
+		linecard->device.info = info;
+		linecard->device.index = device_index;
+		flashable_found = true;
+	} while (msg_seq);
+
+	return 0;
+}
+
+static void mlxsw_linecard_provision_fail(struct mlxsw_linecard *linecard)
+{
+	linecard->provisioned = false;
+	linecard->ready = false;
+	linecard->active = false;
+	devlink_linecard_provision_fail(linecard->devlink_linecard);
+}
+
+struct mlxsw_linecards_event_ops_item {
+	struct list_head list;
+	const struct mlxsw_linecards_event_ops *event_ops;
+	void *priv;
+};
+
+static void
+mlxsw_linecard_event_op_call(struct mlxsw_linecard *linecard,
+			     mlxsw_linecards_event_op_t *op, void *priv)
+{
+	struct mlxsw_core *mlxsw_core = linecard->linecards->mlxsw_core;
+
+	if (!op)
+		return;
+	op(mlxsw_core, linecard->slot_index, priv);
+}
+
+static void
+mlxsw_linecard_active_ops_call(struct mlxsw_linecard *linecard)
+{
+	struct mlxsw_linecards *linecards = linecard->linecards;
+	struct mlxsw_linecards_event_ops_item *item;
+
+	mutex_lock(&linecards->event_ops_list_lock);
+	list_for_each_entry(item, &linecards->event_ops_list, list)
+		mlxsw_linecard_event_op_call(linecard,
+					     item->event_ops->got_active,
+					     item->priv);
+	mutex_unlock(&linecards->event_ops_list_lock);
+}
+
+static void
+mlxsw_linecard_inactive_ops_call(struct mlxsw_linecard *linecard)
+{
+	struct mlxsw_linecards *linecards = linecard->linecards;
+	struct mlxsw_linecards_event_ops_item *item;
+
+	mutex_lock(&linecards->event_ops_list_lock);
+	list_for_each_entry(item, &linecards->event_ops_list, list)
+		mlxsw_linecard_event_op_call(linecard,
+					     item->event_ops->got_inactive,
+					     item->priv);
+	mutex_unlock(&linecards->event_ops_list_lock);
+}
+
+static void
+mlxsw_linecards_event_ops_register_call(struct mlxsw_linecards *linecards,
+					const struct mlxsw_linecards_event_ops_item *item)
+{
+	struct mlxsw_linecard *linecard;
+	int i;
+
+	for (i = 0; i < linecards->count; i++) {
+		linecard = mlxsw_linecard_get(linecards, i + 1);
+		mutex_lock(&linecard->lock);
+		if (linecard->active)
+			mlxsw_linecard_event_op_call(linecard,
+						     item->event_ops->got_active,
+						     item->priv);
+		mutex_unlock(&linecard->lock);
+	}
+}
+
+static void
+mlxsw_linecards_event_ops_unregister_call(struct mlxsw_linecards *linecards,
+					  const struct mlxsw_linecards_event_ops_item *item)
+{
+	struct mlxsw_linecard *linecard;
+	int i;
+
+	for (i = 0; i < linecards->count; i++) {
+		linecard = mlxsw_linecard_get(linecards, i + 1);
+		mutex_lock(&linecard->lock);
+		if (linecard->active)
+			mlxsw_linecard_event_op_call(linecard,
+						     item->event_ops->got_inactive,
+						     item->priv);
+		mutex_unlock(&linecard->lock);
+	}
+}
+
+int mlxsw_linecards_event_ops_register(struct mlxsw_core *mlxsw_core,
+				       struct mlxsw_linecards_event_ops *ops,
+				       void *priv)
+{
+	struct mlxsw_linecards *linecards = mlxsw_core_linecards(mlxsw_core);
+	struct mlxsw_linecards_event_ops_item *item;
+
+	if (!linecards)
+		return 0;
+	item = kzalloc(sizeof(*item), GFP_KERNEL);
+	if (!item)
+		return -ENOMEM;
+	item->event_ops = ops;
+	item->priv = priv;
+
+	mutex_lock(&linecards->event_ops_list_lock);
+	list_add_tail(&item->list, &linecards->event_ops_list);
+	mutex_unlock(&linecards->event_ops_list_lock);
+	mlxsw_linecards_event_ops_register_call(linecards, item);
+	return 0;
+}
+EXPORT_SYMBOL(mlxsw_linecards_event_ops_register);
+
+void mlxsw_linecards_event_ops_unregister(struct mlxsw_core *mlxsw_core,
+					  struct mlxsw_linecards_event_ops *ops,
+					  void *priv)
+{
+	struct mlxsw_linecards *linecards = mlxsw_core_linecards(mlxsw_core);
+	struct mlxsw_linecards_event_ops_item *item, *tmp;
+	bool found = false;
+
+	if (!linecards)
+		return;
+	mutex_lock(&linecards->event_ops_list_lock);
+	list_for_each_entry_safe(item, tmp, &linecards->event_ops_list, list) {
+		if (item->event_ops == ops && item->priv == priv) {
+			list_del(&item->list);
+			found = true;
+			break;
+		}
+	}
+	mutex_unlock(&linecards->event_ops_list_lock);
+
+	if (!found)
+		return;
+	mlxsw_linecards_event_ops_unregister_call(linecards, item);
+	kfree(item);
+}
+EXPORT_SYMBOL(mlxsw_linecards_event_ops_unregister);
+
+int mlxsw_linecard_devlink_info_get(struct mlxsw_linecard *linecard,
+				    struct devlink_info_req *req,
+				    struct netlink_ext_ack *extack)
+{
+	char buf[32];
+	int err;
+
+	mutex_lock(&linecard->lock);
+	if (WARN_ON(!linecard->provisioned)) {
+		err = -EOPNOTSUPP;
+		goto unlock;
+	}
+
+	sprintf(buf, "%d", linecard->hw_revision);
+	err = devlink_info_version_fixed_put(req, "hw.revision", buf);
+	if (err)
+		goto unlock;
+
+	sprintf(buf, "%d", linecard->ini_version);
+	err = devlink_info_version_running_put(req, "ini.version", buf);
+	if (err)
+		goto unlock;
+
+	if (linecard->active) {
+		struct mlxsw_linecard_device_info *info = &linecard->device.info;
+
+		err = devlink_info_version_fixed_put(req,
+						     DEVLINK_INFO_VERSION_GENERIC_FW_PSID,
+						     info->psid);
+
+		sprintf(buf, "%u.%u.%u", info->fw_major, info->fw_minor,
+			info->fw_sub_minor);
+		err = devlink_info_version_running_put(req,
+						       DEVLINK_INFO_VERSION_GENERIC_FW,
+						       buf);
+		if (err)
+			goto unlock;
+	}
+
+unlock:
+	mutex_unlock(&linecard->lock);
+	return err;
+}
+
+static int
+mlxsw_linecard_provision_set(struct mlxsw_linecard *linecard, u8 card_type,
+			     u16 hw_revision, u16 ini_version)
+{
+	struct mlxsw_linecards *linecards = linecard->linecards;
+	const char *type;
+	int err;
+
+	type = mlxsw_linecard_types_lookup(linecards, card_type);
+	mlxsw_linecard_status_event_done(linecard,
+					 MLXSW_LINECARD_STATUS_EVENT_TYPE_PROVISION);
+	if (!type) {
+		/* It is possible for a line card to be provisioned before
+		 * driver initialization. Due to a missing INI bundle file
+		 * or an outdated one, the queried card's type might not
+		 * be recognized by the driver. In this case, try to query
+		 * the card's name from the device.
+		 */
+		type = mlxsw_linecard_type_name(linecard);
+		if (IS_ERR(type)) {
+			mlxsw_linecard_provision_fail(linecard);
+			return PTR_ERR(type);
+		}
+	}
+	linecard->provisioned = true;
+	linecard->hw_revision = hw_revision;
+	linecard->ini_version = ini_version;
+
+	err = mlxsw_linecard_bdev_add(linecard);
+	if (err) {
+		linecard->provisioned = false;
+		mlxsw_linecard_provision_fail(linecard);
+		return err;
+	}
+
+	devlink_linecard_provision_set(linecard->devlink_linecard, type);
+	return 0;
+}
+
+static void mlxsw_linecard_provision_clear(struct mlxsw_linecard *linecard)
+{
+	mlxsw_linecard_status_event_done(linecard,
+					 MLXSW_LINECARD_STATUS_EVENT_TYPE_UNPROVISION);
+	mlxsw_linecard_bdev_del(linecard);
+	linecard->provisioned = false;
+	devlink_linecard_provision_clear(linecard->devlink_linecard);
+}
+
+static int mlxsw_linecard_ready_set(struct mlxsw_linecard *linecard)
+{
+	struct mlxsw_core *mlxsw_core = linecard->linecards->mlxsw_core;
+	char mddc_pl[MLXSW_REG_MDDC_LEN];
+	int err;
+
+	err = mlxsw_linecard_device_info_update(linecard);
+	if (err)
+		return err;
+
+	mlxsw_reg_mddc_pack(mddc_pl, linecard->slot_index, false, true);
+	err = mlxsw_reg_write(mlxsw_core, MLXSW_REG(mddc), mddc_pl);
+	if (err)
+		return err;
+	linecard->ready = true;
+	return 0;
+}
+
+static int mlxsw_linecard_ready_clear(struct mlxsw_linecard *linecard)
+{
+	struct mlxsw_core *mlxsw_core = linecard->linecards->mlxsw_core;
+	char mddc_pl[MLXSW_REG_MDDC_LEN];
+	int err;
+
+	mlxsw_reg_mddc_pack(mddc_pl, linecard->slot_index, false, false);
+	err = mlxsw_reg_write(mlxsw_core, MLXSW_REG(mddc), mddc_pl);
+	if (err)
+		return err;
+	linecard->ready = false;
+	return 0;
+}
+
+static void mlxsw_linecard_active_set(struct mlxsw_linecard *linecard)
+{
+	mlxsw_linecard_active_ops_call(linecard);
+	linecard->active = true;
+	devlink_linecard_activate(linecard->devlink_linecard);
+}
+
+static void mlxsw_linecard_active_clear(struct mlxsw_linecard *linecard)
+{
+	mlxsw_linecard_inactive_ops_call(linecard);
+	linecard->active = false;
+	devlink_linecard_deactivate(linecard->devlink_linecard);
+}
+
+static int mlxsw_linecard_status_process(struct mlxsw_linecards *linecards,
+					 struct mlxsw_linecard *linecard,
+					 const char *mddq_pl)
+{
+	enum mlxsw_reg_mddq_slot_info_ready ready;
+	bool provisioned, sr_valid, active;
+	u16 ini_version, hw_revision;
+	u8 slot_index, card_type;
+	int err = 0;
+
+	mlxsw_reg_mddq_slot_info_unpack(mddq_pl, &slot_index, &provisioned,
+					&sr_valid, &ready, &active,
+					&hw_revision, &ini_version,
+					&card_type);
+
+	if (linecard) {
+		if (WARN_ON(slot_index != linecard->slot_index))
+			return -EINVAL;
+	} else {
+		if (WARN_ON(slot_index > linecards->count))
+			return -EINVAL;
+		linecard = mlxsw_linecard_get(linecards, slot_index);
+	}
+
+	mutex_lock(&linecard->lock);
+
+	if (provisioned && linecard->provisioned != provisioned) {
+		err = mlxsw_linecard_provision_set(linecard, card_type,
+						   hw_revision, ini_version);
+		if (err)
+			goto out;
+	}
+
+	if (ready == MLXSW_REG_MDDQ_SLOT_INFO_READY_READY && !linecard->ready) {
+		err = mlxsw_linecard_ready_set(linecard);
+		if (err)
+			goto out;
+	}
+
+	if (active && linecard->active != active)
+		mlxsw_linecard_active_set(linecard);
+
+	if (!active && linecard->active != active)
+		mlxsw_linecard_active_clear(linecard);
+
+	if (ready != MLXSW_REG_MDDQ_SLOT_INFO_READY_READY &&
+	    linecard->ready) {
+		err = mlxsw_linecard_ready_clear(linecard);
+		if (err)
+			goto out;
+	}
+
+	if (!provisioned && linecard->provisioned != provisioned)
+		mlxsw_linecard_provision_clear(linecard);
+
+out:
+	mutex_unlock(&linecard->lock);
+	return err;
+}
+
+static int mlxsw_linecard_status_get_and_process(struct mlxsw_core *mlxsw_core,
+						 struct mlxsw_linecards *linecards,
+						 struct mlxsw_linecard *linecard)
+{
+	char mddq_pl[MLXSW_REG_MDDQ_LEN];
+	int err;
+
+	mlxsw_reg_mddq_slot_info_pack(mddq_pl, linecard->slot_index, false);
+	err = mlxsw_reg_query(mlxsw_core, MLXSW_REG(mddq), mddq_pl);
+	if (err)
+		return err;
+
+	return mlxsw_linecard_status_process(linecards, linecard, mddq_pl);
+}
+
+static void mlxsw_linecards_irq_event_handler(struct mlxsw_core *mlxsw_core)
+{
+	struct mlxsw_linecards *linecards = mlxsw_core_linecards(mlxsw_core);
+	int i;
+
+	/* Handle change of line card active state. */
+	for (i = 0; i < linecards->count; i++) {
+		struct mlxsw_linecard *linecard = mlxsw_linecard_get(linecards,
+								     i + 1);
+
+		mlxsw_linecard_status_get_and_process(mlxsw_core, linecards,
+						      linecard);
+	}
+}
+
+static const char * const mlxsw_linecard_status_event_type_name[] = {
+	[MLXSW_LINECARD_STATUS_EVENT_TYPE_PROVISION] = "provision",
+	[MLXSW_LINECARD_STATUS_EVENT_TYPE_UNPROVISION] = "unprovision",
+};
+
+static void mlxsw_linecard_status_event_to_work(struct work_struct *work)
+{
+	struct mlxsw_linecard *linecard =
+		container_of(work, struct mlxsw_linecard,
+			     status_event_to_dw.work);
+
+	mutex_lock(&linecard->lock);
+	dev_err(linecard->linecards->bus_info->dev, "linecard %u: Timeout reached waiting on %s status event",
+		linecard->slot_index,
+		mlxsw_linecard_status_event_type_name[linecard->status_event_type_to]);
+	mlxsw_linecard_provision_fail(linecard);
+	mutex_unlock(&linecard->lock);
+}
+
+static int __mlxsw_linecard_fix_fsm_state(struct mlxsw_linecard *linecard)
+{
+	dev_info(linecard->linecards->bus_info->dev, "linecard %u: Clearing FSM state error",
+		 linecard->slot_index);
+	mlxsw_reg_mbct_pack(linecard->mbct_pl, linecard->slot_index,
+			    MLXSW_REG_MBCT_OP_CLEAR_ERRORS, false);
+	return mlxsw_reg_write(linecard->linecards->mlxsw_core,
+			       MLXSW_REG(mbct), linecard->mbct_pl);
+}
+
+static int mlxsw_linecard_fix_fsm_state(struct mlxsw_linecard *linecard,
+					enum mlxsw_reg_mbct_fsm_state fsm_state)
+{
+	if (fsm_state != MLXSW_REG_MBCT_FSM_STATE_ERROR)
+		return 0;
+	return __mlxsw_linecard_fix_fsm_state(linecard);
+}
+
+static int
+mlxsw_linecard_query_ini_status(struct mlxsw_linecard *linecard,
+				enum mlxsw_reg_mbct_status *status,
+				enum mlxsw_reg_mbct_fsm_state *fsm_state,
+				struct netlink_ext_ack *extack)
+{
+	int err;
+
+	mlxsw_reg_mbct_pack(linecard->mbct_pl, linecard->slot_index,
+			    MLXSW_REG_MBCT_OP_QUERY_STATUS, false);
+	err = mlxsw_reg_query(linecard->linecards->mlxsw_core, MLXSW_REG(mbct),
+			      linecard->mbct_pl);
+	if (err) {
+		NL_SET_ERR_MSG_MOD(extack, "Failed to query linecard INI status");
+		return err;
+	}
+	mlxsw_reg_mbct_unpack(linecard->mbct_pl, NULL, status, fsm_state);
+	return err;
+}
+
+static int
+mlxsw_linecard_ini_transfer(struct mlxsw_core *mlxsw_core,
+			    struct mlxsw_linecard *linecard,
+			    const struct mlxsw_linecard_ini_file *ini_file,
+			    struct netlink_ext_ack *extack)
+{
+	enum mlxsw_reg_mbct_fsm_state fsm_state;
+	enum mlxsw_reg_mbct_status status;
+	size_t size_left;
+	const u8 *data;
+	int err;
+
+	size_left = le16_to_cpu(ini_file->size);
+	data = ini_file->data;
+	while (size_left) {
+		size_t data_size = MLXSW_REG_MBCT_DATA_LEN;
+		bool is_last = false;
+
+		if (size_left <= MLXSW_REG_MBCT_DATA_LEN) {
+			data_size = size_left;
+			is_last = true;
+		}
+
+		mlxsw_reg_mbct_pack(linecard->mbct_pl, linecard->slot_index,
+				    MLXSW_REG_MBCT_OP_DATA_TRANSFER, false);
+		mlxsw_reg_mbct_dt_pack(linecard->mbct_pl, data_size,
+				       is_last, data);
+		err = mlxsw_reg_write(mlxsw_core, MLXSW_REG(mbct),
+				      linecard->mbct_pl);
+		if (err) {
+			NL_SET_ERR_MSG_MOD(extack, "Failed to issue linecard INI data transfer");
+			return err;
+		}
+		mlxsw_reg_mbct_unpack(linecard->mbct_pl, NULL,
+				      &status, &fsm_state);
+		if ((!is_last && status != MLXSW_REG_MBCT_STATUS_PART_DATA) ||
+		    (is_last && status != MLXSW_REG_MBCT_STATUS_LAST_DATA)) {
+			NL_SET_ERR_MSG_MOD(extack, "Failed to transfer linecard INI data");
+			mlxsw_linecard_fix_fsm_state(linecard, fsm_state);
+			return -EINVAL;
+		}
+		size_left -= data_size;
+		data += data_size;
+	}
+
+	return 0;
+}
+
+static int
+mlxsw_linecard_ini_erase(struct mlxsw_core *mlxsw_core,
+			 struct mlxsw_linecard *linecard,
+			 struct netlink_ext_ack *extack)
+{
+	enum mlxsw_reg_mbct_fsm_state fsm_state;
+	enum mlxsw_reg_mbct_status status;
+	int err;
+
+	mlxsw_reg_mbct_pack(linecard->mbct_pl, linecard->slot_index,
+			    MLXSW_REG_MBCT_OP_ERASE_INI_IMAGE, false);
+	err = mlxsw_reg_write(mlxsw_core, MLXSW_REG(mbct),
+			      linecard->mbct_pl);
+	if (err) {
+		NL_SET_ERR_MSG_MOD(extack, "Failed to issue linecard INI erase");
+		return err;
+	}
+	mlxsw_reg_mbct_unpack(linecard->mbct_pl, NULL, &status, &fsm_state);
+	switch (status) {
+	case MLXSW_REG_MBCT_STATUS_ERASE_COMPLETE:
+		break;
+	default:
+		/* Should not happen */
+		fallthrough;
+	case MLXSW_REG_MBCT_STATUS_ERASE_FAILED:
+		NL_SET_ERR_MSG_MOD(extack, "Failed to erase linecard INI");
+		goto fix_fsm_err_out;
+	case MLXSW_REG_MBCT_STATUS_ERROR_INI_IN_USE:
+		NL_SET_ERR_MSG_MOD(extack, "Failed to erase linecard INI while being used");
+		goto fix_fsm_err_out;
+	}
+	return 0;
+
+fix_fsm_err_out:
+	mlxsw_linecard_fix_fsm_state(linecard, fsm_state);
+	return -EINVAL;
+}
+
+static void mlxsw_linecard_bct_process(struct mlxsw_core *mlxsw_core,
+				       const char *mbct_pl)
+{
+	struct mlxsw_linecards *linecards = mlxsw_core_linecards(mlxsw_core);
+	enum mlxsw_reg_mbct_fsm_state fsm_state;
+	enum mlxsw_reg_mbct_status status;
+	struct mlxsw_linecard *linecard;
+	u8 slot_index;
+
+	mlxsw_reg_mbct_unpack(mbct_pl, &slot_index, &status, &fsm_state);
+	if (WARN_ON(slot_index > linecards->count))
+		return;
+	linecard = mlxsw_linecard_get(linecards, slot_index);
+	mutex_lock(&linecard->lock);
+	if (status == MLXSW_REG_MBCT_STATUS_ACTIVATION_FAILED) {
+		dev_err(linecards->bus_info->dev, "linecard %u: Failed to activate INI",
+			linecard->slot_index);
+		goto fix_fsm_out;
+	}
+	mutex_unlock(&linecard->lock);
+	return;
+
+fix_fsm_out:
+	mlxsw_linecard_fix_fsm_state(linecard, fsm_state);
+	mlxsw_linecard_provision_fail(linecard);
+	mutex_unlock(&linecard->lock);
+}
+
+static int
+mlxsw_linecard_ini_activate(struct mlxsw_core *mlxsw_core,
+			    struct mlxsw_linecard *linecard,
+			    struct netlink_ext_ack *extack)
+{
+	enum mlxsw_reg_mbct_fsm_state fsm_state;
+	enum mlxsw_reg_mbct_status status;
+	int err;
+
+	mlxsw_reg_mbct_pack(linecard->mbct_pl, linecard->slot_index,
+			    MLXSW_REG_MBCT_OP_ACTIVATE, true);
+	err = mlxsw_reg_write(mlxsw_core, MLXSW_REG(mbct), linecard->mbct_pl);
+	if (err) {
+		NL_SET_ERR_MSG_MOD(extack, "Failed to issue linecard INI activation");
+		return err;
+	}
+	mlxsw_reg_mbct_unpack(linecard->mbct_pl, NULL, &status, &fsm_state);
+	if (status == MLXSW_REG_MBCT_STATUS_ACTIVATION_FAILED) {
+		NL_SET_ERR_MSG_MOD(extack, "Failed to activate linecard INI");
+		goto fix_fsm_err_out;
+	}
+
+	return 0;
+
+fix_fsm_err_out:
+	mlxsw_linecard_fix_fsm_state(linecard, fsm_state);
+	return -EINVAL;
+}
+
+#define MLXSW_LINECARD_INI_WAIT_RETRIES 10
+#define MLXSW_LINECARD_INI_WAIT_MS 500
+
+static int
+mlxsw_linecard_ini_in_use_wait(struct mlxsw_core *mlxsw_core,
+			       struct mlxsw_linecard *linecard,
+			       struct netlink_ext_ack *extack)
+{
+	enum mlxsw_reg_mbct_fsm_state fsm_state;
+	enum mlxsw_reg_mbct_status status;
+	unsigned int ini_wait_retries = 0;
+	int err;
+
+query_ini_status:
+	err = mlxsw_linecard_query_ini_status(linecard, &status,
+					      &fsm_state, extack);
+	if (err)
+		return err;
+
+	switch (fsm_state) {
+	case MLXSW_REG_MBCT_FSM_STATE_INI_IN_USE:
+		if (ini_wait_retries++ > MLXSW_LINECARD_INI_WAIT_RETRIES) {
+			NL_SET_ERR_MSG_MOD(extack, "Failed to wait for linecard INI to be unused");
+			return -EINVAL;
+		}
+		mdelay(MLXSW_LINECARD_INI_WAIT_MS);
+		goto query_ini_status;
+	default:
+		break;
+	}
+	return 0;
+}
+
+static bool mlxsw_linecard_port_selector(void *priv, u16 local_port)
+{
+	struct mlxsw_linecard *linecard = priv;
+	struct mlxsw_core *mlxsw_core;
+
+	mlxsw_core = linecard->linecards->mlxsw_core;
+	return linecard == mlxsw_core_port_linecard_get(mlxsw_core, local_port);
+}
+
+static int mlxsw_linecard_provision(struct devlink_linecard *devlink_linecard,
+				    void *priv, const char *type,
+				    const void *type_priv,
+				    struct netlink_ext_ack *extack)
+{
+	const struct mlxsw_linecard_ini_file *ini_file = type_priv;
+	struct mlxsw_linecard *linecard = priv;
+	struct mlxsw_core *mlxsw_core;
+	int err;
+
+	mutex_lock(&linecard->lock);
+
+	mlxsw_core = linecard->linecards->mlxsw_core;
+
+	err = mlxsw_linecard_ini_erase(mlxsw_core, linecard, extack);
+	if (err)
+		goto err_out;
+
+	err = mlxsw_linecard_ini_transfer(mlxsw_core, linecard,
+					  ini_file, extack);
+	if (err)
+		goto err_out;
+
+	mlxsw_linecard_status_event_to_schedule(linecard,
+						MLXSW_LINECARD_STATUS_EVENT_TYPE_PROVISION);
+	err = mlxsw_linecard_ini_activate(mlxsw_core, linecard, extack);
+	if (err)
+		goto err_out;
+
+	goto out;
+
+err_out:
+	mlxsw_linecard_provision_fail(linecard);
+out:
+	mutex_unlock(&linecard->lock);
+	return err;
+}
+
+static int mlxsw_linecard_unprovision(struct devlink_linecard *devlink_linecard,
+				      void *priv,
+				      struct netlink_ext_ack *extack)
+{
+	struct mlxsw_linecard *linecard = priv;
+	struct mlxsw_core *mlxsw_core;
+	int err;
+
+	mutex_lock(&linecard->lock);
+
+	mlxsw_core = linecard->linecards->mlxsw_core;
+
+	mlxsw_core_ports_remove_selected(mlxsw_core,
+					 mlxsw_linecard_port_selector,
+					 linecard);
+
+	err = mlxsw_linecard_ini_in_use_wait(mlxsw_core, linecard, extack);
+	if (err)
+		goto err_out;
+
+	mlxsw_linecard_status_event_to_schedule(linecard,
+						MLXSW_LINECARD_STATUS_EVENT_TYPE_UNPROVISION);
+	err = mlxsw_linecard_ini_erase(mlxsw_core, linecard, extack);
+	if (err)
+		goto err_out;
+
+	goto out;
+
+err_out:
+	mlxsw_linecard_provision_fail(linecard);
+out:
+	mutex_unlock(&linecard->lock);
+	return err;
+}
+
+static bool mlxsw_linecard_same_provision(struct devlink_linecard *devlink_linecard,
+					  void *priv, const char *type,
+					  const void *type_priv)
+{
+	const struct mlxsw_linecard_ini_file *ini_file = type_priv;
+	struct mlxsw_linecard *linecard = priv;
+	bool ret;
+
+	mutex_lock(&linecard->lock);
+	ret = linecard->hw_revision == be16_to_cpu(ini_file->format.hw_revision) &&
+	      linecard->ini_version == be16_to_cpu(ini_file->format.ini_version);
+	mutex_unlock(&linecard->lock);
+	return ret;
+}
+
+static unsigned int
+mlxsw_linecard_types_count(struct devlink_linecard *devlink_linecard,
+			   void *priv)
+{
+	struct mlxsw_linecard *linecard = priv;
+
+	return linecard->linecards->types_info ?
+	       linecard->linecards->types_info->count : 0;
+}
+
+static void mlxsw_linecard_types_get(struct devlink_linecard *devlink_linecard,
+				     void *priv, unsigned int index,
+				     const char **type, const void **type_priv)
+{
+	struct mlxsw_linecard_types_info *types_info;
+	struct mlxsw_linecard_ini_file *ini_file;
+	struct mlxsw_linecard *linecard = priv;
+
+	types_info = linecard->linecards->types_info;
+	if (WARN_ON_ONCE(!types_info))
+		return;
+	ini_file = types_info->ini_files[index];
+	*type = ini_file->format.name;
+	*type_priv = ini_file;
+}
+
+static const struct devlink_linecard_ops mlxsw_linecard_ops = {
+	.provision = mlxsw_linecard_provision,
+	.unprovision = mlxsw_linecard_unprovision,
+	.same_provision = mlxsw_linecard_same_provision,
+	.types_count = mlxsw_linecard_types_count,
+	.types_get = mlxsw_linecard_types_get,
+};
+
+struct mlxsw_linecard_status_event {
+	struct mlxsw_core *mlxsw_core;
+	char mddq_pl[MLXSW_REG_MDDQ_LEN];
+	struct work_struct work;
+};
+
+static void mlxsw_linecard_status_event_work(struct work_struct *work)
+{
+	struct mlxsw_linecard_status_event *event;
+	struct mlxsw_linecards *linecards;
+	struct mlxsw_core *mlxsw_core;
+
+	event = container_of(work, struct mlxsw_linecard_status_event, work);
+	mlxsw_core = event->mlxsw_core;
+	linecards = mlxsw_core_linecards(mlxsw_core);
+	mlxsw_linecard_status_process(linecards, NULL, event->mddq_pl);
+	kfree(event);
+}
+
+static void
+mlxsw_linecard_status_listener_func(const struct mlxsw_reg_info *reg,
+				    char *mddq_pl, void *priv)
+{
+	struct mlxsw_linecard_status_event *event;
+	struct mlxsw_core *mlxsw_core = priv;
+
+	event = kmalloc(sizeof(*event), GFP_ATOMIC);
+	if (!event)
+		return;
+	event->mlxsw_core = mlxsw_core;
+	memcpy(event->mddq_pl, mddq_pl, sizeof(event->mddq_pl));
+	INIT_WORK(&event->work, mlxsw_linecard_status_event_work);
+	mlxsw_core_schedule_work(&event->work);
+}
+
+struct mlxsw_linecard_bct_event {
+	struct mlxsw_core *mlxsw_core;
+	char mbct_pl[MLXSW_REG_MBCT_LEN];
+	struct work_struct work;
+};
+
+static void mlxsw_linecard_bct_event_work(struct work_struct *work)
+{
+	struct mlxsw_linecard_bct_event *event;
+	struct mlxsw_core *mlxsw_core;
+
+	event = container_of(work, struct mlxsw_linecard_bct_event, work);
+	mlxsw_core = event->mlxsw_core;
+	mlxsw_linecard_bct_process(mlxsw_core, event->mbct_pl);
+	kfree(event);
+}
+
+static void
+mlxsw_linecard_bct_listener_func(const struct mlxsw_reg_info *reg,
+				 char *mbct_pl, void *priv)
+{
+	struct mlxsw_linecard_bct_event *event;
+	struct mlxsw_core *mlxsw_core = priv;
+
+	event = kmalloc(sizeof(*event), GFP_ATOMIC);
+	if (!event)
+		return;
+	event->mlxsw_core = mlxsw_core;
+	memcpy(event->mbct_pl, mbct_pl, sizeof(event->mbct_pl));
+	INIT_WORK(&event->work, mlxsw_linecard_bct_event_work);
+	mlxsw_core_schedule_work(&event->work);
+}
+
+static const struct mlxsw_listener mlxsw_linecard_listener[] = {
+	MLXSW_CORE_EVENTL(mlxsw_linecard_status_listener_func, DSDSC),
+	MLXSW_CORE_EVENTL(mlxsw_linecard_bct_listener_func, BCTOE),
+};
+
+static int mlxsw_linecard_event_delivery_set(struct mlxsw_core *mlxsw_core,
+					     struct mlxsw_linecard *linecard,
+					     bool enable)
+{
+	char mddq_pl[MLXSW_REG_MDDQ_LEN];
+
+	mlxsw_reg_mddq_slot_info_pack(mddq_pl, linecard->slot_index, enable);
+	return mlxsw_reg_write(mlxsw_core, MLXSW_REG(mddq), mddq_pl);
+}
+
+static int mlxsw_linecard_init(struct mlxsw_core *mlxsw_core,
+			       struct mlxsw_linecards *linecards,
+			       u8 slot_index)
+{
+	struct devlink_linecard *devlink_linecard;
+	struct mlxsw_linecard *linecard;
+
+	linecard = mlxsw_linecard_get(linecards, slot_index);
+	linecard->slot_index = slot_index;
+	linecard->linecards = linecards;
+	mutex_init(&linecard->lock);
+
+	devlink_linecard = devlink_linecard_create(priv_to_devlink(mlxsw_core),
+						   slot_index, &mlxsw_linecard_ops,
+						   linecard);
+	if (IS_ERR(devlink_linecard))
+		return PTR_ERR(devlink_linecard);
+
+	linecard->devlink_linecard = devlink_linecard;
+	INIT_DELAYED_WORK(&linecard->status_event_to_dw,
+			  &mlxsw_linecard_status_event_to_work);
+
+	return 0;
+}
+
+static void mlxsw_linecard_fini(struct mlxsw_core *mlxsw_core,
+				struct mlxsw_linecards *linecards,
+				u8 slot_index)
+{
+	struct mlxsw_linecard *linecard;
+
+	linecard = mlxsw_linecard_get(linecards, slot_index);
+	cancel_delayed_work_sync(&linecard->status_event_to_dw);
+	/* Make sure all scheduled events are processed */
+	mlxsw_core_flush_owq();
+	if (linecard->active)
+		mlxsw_linecard_active_clear(linecard);
+	mlxsw_linecard_bdev_del(linecard);
+	devlink_linecard_destroy(linecard->devlink_linecard);
+	mutex_destroy(&linecard->lock);
+}
+
+static int
+mlxsw_linecard_event_delivery_init(struct mlxsw_core *mlxsw_core,
+				   struct mlxsw_linecards *linecards,
+				   u8 slot_index)
+{
+	struct mlxsw_linecard *linecard;
+	int err;
+
+	linecard = mlxsw_linecard_get(linecards, slot_index);
+	err = mlxsw_linecard_event_delivery_set(mlxsw_core, linecard, true);
+	if (err)
+		return err;
+
+	err = mlxsw_linecard_status_get_and_process(mlxsw_core, linecards,
+						    linecard);
+	if (err)
+		goto err_status_get_and_process;
+
+	return 0;
+
+err_status_get_and_process:
+	mlxsw_linecard_event_delivery_set(mlxsw_core, linecard, false);
+	return err;
+}
+
+static void
+mlxsw_linecard_event_delivery_fini(struct mlxsw_core *mlxsw_core,
+				   struct mlxsw_linecards *linecards,
+				   u8 slot_index)
+{
+	struct mlxsw_linecard *linecard;
+
+	linecard = mlxsw_linecard_get(linecards, slot_index);
+	mlxsw_linecard_event_delivery_set(mlxsw_core, linecard, false);
+}
+
+/*       LINECARDS INI BUNDLE FILE
+ *  +----------------------------------+
+ *  |        MAGIC ("NVLCINI+")        |
+ *  +----------------------------------+     +--------------------+
+ *  |  INI 0                           +---> | __le16 size        |
+ *  +----------------------------------+     | __be16 hw_revision |
+ *  |  INI 1                           |     | __be16 ini_version |
+ *  +----------------------------------+     | u8 __dontcare[3]   |
+ *  |  ...                             |     | u8 type            |
+ *  +----------------------------------+     | u8 name[20]        |
+ *  |  INI N                           |     | ...                |
+ *  +----------------------------------+     +--------------------+
+ */
+
+#define MLXSW_LINECARDS_INI_BUNDLE_MAGIC "NVLCINI+"
+
+static int
+mlxsw_linecard_types_file_validate(struct mlxsw_linecards *linecards,
+				   struct mlxsw_linecard_types_info *types_info)
+{
+	size_t magic_size = strlen(MLXSW_LINECARDS_INI_BUNDLE_MAGIC);
+	struct mlxsw_linecard_ini_file *ini_file;
+	size_t size = types_info->data_size;
+	const u8 *data = types_info->data;
+	unsigned int count = 0;
+	u16 ini_file_size;
+
+	if (size < magic_size) {
+		dev_warn(linecards->bus_info->dev, "Invalid linecards INIs file size, smaller than magic size\n");
+		return -EINVAL;
+	}
+	if (memcmp(data, MLXSW_LINECARDS_INI_BUNDLE_MAGIC, magic_size)) {
+		dev_warn(linecards->bus_info->dev, "Invalid linecards INIs file magic pattern\n");
+		return -EINVAL;
+	}
+
+	data += magic_size;
+	size -= magic_size;
+
+	while (size > 0) {
+		if (size < sizeof(*ini_file)) {
+			dev_warn(linecards->bus_info->dev, "Linecards INIs file contains INI which is smaller than bare minimum\n");
+			return -EINVAL;
+		}
+		ini_file = (struct mlxsw_linecard_ini_file *) data;
+		ini_file_size = le16_to_cpu(ini_file->size);
+		if (ini_file_size + sizeof(__le16) > size) {
+			dev_warn(linecards->bus_info->dev, "Linecards INIs file appears to be truncated\n");
+			return -EINVAL;
+		}
+		if (ini_file_size % 4) {
+			dev_warn(linecards->bus_info->dev, "Linecards INIs file contains INI with invalid size\n");
+			return -EINVAL;
+		}
+		data += ini_file_size + sizeof(__le16);
+		size -= ini_file_size + sizeof(__le16);
+		count++;
+	}
+	if (!count) {
+		dev_warn(linecards->bus_info->dev, "Linecards INIs file does not contain any INI\n");
+		return -EINVAL;
+	}
+	types_info->count = count;
+	return 0;
+}
+
+static void
+mlxsw_linecard_types_file_parse(struct mlxsw_linecard_types_info *types_info)
+{
+	size_t magic_size = strlen(MLXSW_LINECARDS_INI_BUNDLE_MAGIC);
+	size_t size = types_info->data_size - magic_size;
+	const u8 *data = types_info->data + magic_size;
+	struct mlxsw_linecard_ini_file *ini_file;
+	unsigned int count = 0;
+	u16 ini_file_size;
+	int i;
+
+	while (size) {
+		ini_file = (struct mlxsw_linecard_ini_file *) data;
+		ini_file_size = le16_to_cpu(ini_file->size);
+		for (i = 0; i < ini_file_size / 4; i++) {
+			u32 *val = &((u32 *) ini_file->data)[i];
+
+			*val = swab32(*val);
+		}
+		types_info->ini_files[count] = ini_file;
+		data += ini_file_size + sizeof(__le16);
+		size -= ini_file_size + sizeof(__le16);
+		count++;
+	}
+}
+
+#define MLXSW_LINECARDS_INI_BUNDLE_FILENAME_FMT \
+	"mellanox/lc_ini_bundle_%u_%u.bin"
+#define MLXSW_LINECARDS_INI_BUNDLE_FILENAME_LEN \
+	(sizeof(MLXSW_LINECARDS_INI_BUNDLE_FILENAME_FMT) + 4)
+
+static int mlxsw_linecard_types_init(struct mlxsw_core *mlxsw_core,
+				     struct mlxsw_linecards *linecards)
+{
+	const struct mlxsw_fw_rev *rev = &linecards->bus_info->fw_rev;
+	char filename[MLXSW_LINECARDS_INI_BUNDLE_FILENAME_LEN];
+	struct mlxsw_linecard_types_info *types_info;
+	const struct firmware *firmware;
+	int err;
+
+	err = snprintf(filename, sizeof(filename),
+		       MLXSW_LINECARDS_INI_BUNDLE_FILENAME_FMT,
+		       rev->minor, rev->subminor);
+	WARN_ON(err >= sizeof(filename));
+
+	err = request_firmware_direct(&firmware, filename,
+				      linecards->bus_info->dev);
+	if (err) {
+		dev_warn(linecards->bus_info->dev, "Could not request linecards INI file \"%s\", provisioning will not be possible\n",
+			 filename);
+		return 0;
+	}
+
+	types_info = kzalloc(sizeof(*types_info), GFP_KERNEL);
+	if (!types_info) {
+		release_firmware(firmware);
+		return -ENOMEM;
+	}
+	linecards->types_info = types_info;
+
+	types_info->data_size = firmware->size;
+	types_info->data = vmalloc(types_info->data_size);
+	if (!types_info->data) {
+		err = -ENOMEM;
+		release_firmware(firmware);
+		goto err_data_alloc;
+	}
+	memcpy(types_info->data, firmware->data, types_info->data_size);
+	release_firmware(firmware);
+
+	err = mlxsw_linecard_types_file_validate(linecards, types_info);
+	if (err) {
+		err = 0;
+		goto err_type_file_file_validate;
+	}
+
+	types_info->ini_files = kmalloc_array(types_info->count,
+					      sizeof(struct mlxsw_linecard_ini_file *),
+					      GFP_KERNEL);
+	if (!types_info->ini_files) {
+		err = -ENOMEM;
+		goto err_ini_files_alloc;
+	}
+
+	mlxsw_linecard_types_file_parse(types_info);
+
+	return 0;
+
+err_ini_files_alloc:
+err_type_file_file_validate:
+	vfree(types_info->data);
+err_data_alloc:
+	kfree(types_info);
+	return err;
+}
+
+static void mlxsw_linecard_types_fini(struct mlxsw_linecards *linecards)
+{
+	struct mlxsw_linecard_types_info *types_info = linecards->types_info;
+
+	if (!types_info)
+		return;
+	kfree(types_info->ini_files);
+	vfree(types_info->data);
+	kfree(types_info);
+}
+
+int mlxsw_linecards_init(struct mlxsw_core *mlxsw_core,
+			 const struct mlxsw_bus_info *bus_info)
+{
+	char mgpir_pl[MLXSW_REG_MGPIR_LEN];
+	struct mlxsw_linecards *linecards;
+	u8 slot_count;
+	int err;
+	int i;
+
+	mlxsw_reg_mgpir_pack(mgpir_pl, 0);
+	err = mlxsw_reg_query(mlxsw_core, MLXSW_REG(mgpir), mgpir_pl);
+	if (err)
+		return err;
+
+	mlxsw_reg_mgpir_unpack(mgpir_pl, NULL, NULL, NULL,
+			       NULL, &slot_count);
+	if (!slot_count)
+		return 0;
+
+	linecards = vzalloc(struct_size(linecards, linecards, slot_count));
+	if (!linecards)
+		return -ENOMEM;
+	linecards->count = slot_count;
+	linecards->mlxsw_core = mlxsw_core;
+	linecards->bus_info = bus_info;
+	INIT_LIST_HEAD(&linecards->event_ops_list);
+	mutex_init(&linecards->event_ops_list_lock);
+
+	err = mlxsw_linecard_types_init(mlxsw_core, linecards);
+	if (err)
+		goto err_types_init;
+
+	err = mlxsw_core_traps_register(mlxsw_core, mlxsw_linecard_listener,
+					ARRAY_SIZE(mlxsw_linecard_listener),
+					mlxsw_core);
+	if (err)
+		goto err_traps_register;
+
+	err = mlxsw_core_irq_event_handler_register(mlxsw_core,
+						    mlxsw_linecards_irq_event_handler);
+	if (err)
+		goto err_irq_event_handler_register;
+
+	mlxsw_core_linecards_set(mlxsw_core, linecards);
+
+	for (i = 0; i < linecards->count; i++) {
+		err = mlxsw_linecard_init(mlxsw_core, linecards, i + 1);
+		if (err)
+			goto err_linecard_init;
+	}
+
+	for (i = 0; i < linecards->count; i++) {
+		err = mlxsw_linecard_event_delivery_init(mlxsw_core, linecards,
+							 i + 1);
+		if (err)
+			goto err_linecard_event_delivery_init;
+	}
+
+	return 0;
+
+err_linecard_event_delivery_init:
+	for (i--; i >= 0; i--)
+		mlxsw_linecard_event_delivery_fini(mlxsw_core, linecards, i + 1);
+	i = linecards->count;
+err_linecard_init:
+	for (i--; i >= 0; i--)
+		mlxsw_linecard_fini(mlxsw_core, linecards, i + 1);
+	mlxsw_core_irq_event_handler_unregister(mlxsw_core,
+						mlxsw_linecards_irq_event_handler);
+err_irq_event_handler_register:
+	mlxsw_core_traps_unregister(mlxsw_core, mlxsw_linecard_listener,
+				    ARRAY_SIZE(mlxsw_linecard_listener),
+				    mlxsw_core);
+err_traps_register:
+	mlxsw_linecard_types_fini(linecards);
+err_types_init:
+	vfree(linecards);
+	return err;
+}
+
+void mlxsw_linecards_fini(struct mlxsw_core *mlxsw_core)
+{
+	struct mlxsw_linecards *linecards = mlxsw_core_linecards(mlxsw_core);
+	int i;
+
+	if (!linecards)
+		return;
+	for (i = 0; i < linecards->count; i++)
+		mlxsw_linecard_event_delivery_fini(mlxsw_core, linecards, i + 1);
+	for (i = 0; i < linecards->count; i++)
+		mlxsw_linecard_fini(mlxsw_core, linecards, i + 1);
+	mlxsw_core_irq_event_handler_unregister(mlxsw_core,
+						mlxsw_linecards_irq_event_handler);
+	mlxsw_core_traps_unregister(mlxsw_core, mlxsw_linecard_listener,
+				    ARRAY_SIZE(mlxsw_linecard_listener),
+				    mlxsw_core);
+	mlxsw_linecard_types_fini(linecards);
+	mutex_destroy(&linecards->event_ops_list_lock);
+	WARN_ON(!list_empty(&linecards->event_ops_list));
+	vfree(linecards);
+}
diff --git a/drivers/net/ethernet/mellanox/mlxsw/core_thermal.c b/drivers/net/ethernet/mellanox/mlxsw/core_thermal.c
new file mode 100644
index 000000000..ef5e61708
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/core_thermal.c
@@ -0,0 +1,1063 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2016-2018 Mellanox Technologies. All rights reserved
+ * Copyright (c) 2016 Ivan Vecera <cera@cera.cz>
+ */
+
+#include <linux/kernel.h>
+#include <linux/types.h>
+#include <linux/device.h>
+#include <linux/sysfs.h>
+#include <linux/thermal.h>
+#include <linux/err.h>
+#include <linux/sfp.h>
+
+#include "core.h"
+#include "core_env.h"
+
+#define MLXSW_THERMAL_POLL_INT	1000	/* ms */
+#define MLXSW_THERMAL_SLOW_POLL_INT	20000	/* ms */
+#define MLXSW_THERMAL_ASIC_TEMP_NORM	75000	/* 75C */
+#define MLXSW_THERMAL_ASIC_TEMP_HIGH	85000	/* 85C */
+#define MLXSW_THERMAL_ASIC_TEMP_HOT	105000	/* 105C */
+#define MLXSW_THERMAL_HYSTERESIS_TEMP	5000	/* 5C */
+#define MLXSW_THERMAL_MODULE_TEMP_SHIFT	(MLXSW_THERMAL_HYSTERESIS_TEMP * 2)
+#define MLXSW_THERMAL_MAX_STATE	10
+#define MLXSW_THERMAL_MIN_STATE	2
+#define MLXSW_THERMAL_MAX_DUTY	255
+
+/* External cooling devices, allowed for binding to mlxsw thermal zones. */
+static char * const mlxsw_thermal_external_allowed_cdev[] = {
+	"mlxreg_fan",
+};
+
+enum mlxsw_thermal_trips {
+	MLXSW_THERMAL_TEMP_TRIP_NORM,
+	MLXSW_THERMAL_TEMP_TRIP_HIGH,
+	MLXSW_THERMAL_TEMP_TRIP_HOT,
+};
+
+struct mlxsw_thermal_trip {
+	int	type;
+	int	temp;
+	int	hyst;
+	int	min_state;
+	int	max_state;
+};
+
+static const struct mlxsw_thermal_trip default_thermal_trips[] = {
+	{	/* In range - 0-40% PWM */
+		.type		= THERMAL_TRIP_ACTIVE,
+		.temp		= MLXSW_THERMAL_ASIC_TEMP_NORM,
+		.hyst		= MLXSW_THERMAL_HYSTERESIS_TEMP,
+		.min_state	= 0,
+		.max_state	= (4 * MLXSW_THERMAL_MAX_STATE) / 10,
+	},
+	{
+		/* In range - 40-100% PWM */
+		.type		= THERMAL_TRIP_ACTIVE,
+		.temp		= MLXSW_THERMAL_ASIC_TEMP_HIGH,
+		.hyst		= MLXSW_THERMAL_HYSTERESIS_TEMP,
+		.min_state	= (4 * MLXSW_THERMAL_MAX_STATE) / 10,
+		.max_state	= MLXSW_THERMAL_MAX_STATE,
+	},
+	{	/* Warning */
+		.type		= THERMAL_TRIP_HOT,
+		.temp		= MLXSW_THERMAL_ASIC_TEMP_HOT,
+		.min_state	= MLXSW_THERMAL_MAX_STATE,
+		.max_state	= MLXSW_THERMAL_MAX_STATE,
+	},
+};
+
+#define MLXSW_THERMAL_NUM_TRIPS	ARRAY_SIZE(default_thermal_trips)
+
+/* Make sure all trips are writable */
+#define MLXSW_THERMAL_TRIP_MASK	(BIT(MLXSW_THERMAL_NUM_TRIPS) - 1)
+
+struct mlxsw_thermal;
+
+struct mlxsw_thermal_module {
+	struct mlxsw_thermal *parent;
+	struct thermal_zone_device *tzdev;
+	struct mlxsw_thermal_trip trips[MLXSW_THERMAL_NUM_TRIPS];
+	int module; /* Module or gearbox number */
+	u8 slot_index;
+};
+
+struct mlxsw_thermal_area {
+	struct mlxsw_thermal_module *tz_module_arr;
+	u8 tz_module_num;
+	struct mlxsw_thermal_module *tz_gearbox_arr;
+	u8 tz_gearbox_num;
+	u8 slot_index;
+	bool active;
+};
+
+struct mlxsw_thermal {
+	struct mlxsw_core *core;
+	const struct mlxsw_bus_info *bus_info;
+	struct thermal_zone_device *tzdev;
+	int polling_delay;
+	struct thermal_cooling_device *cdevs[MLXSW_MFCR_PWMS_MAX];
+	struct mlxsw_thermal_trip trips[MLXSW_THERMAL_NUM_TRIPS];
+	struct mlxsw_thermal_area line_cards[];
+};
+
+static inline u8 mlxsw_state_to_duty(int state)
+{
+	return DIV_ROUND_CLOSEST(state * MLXSW_THERMAL_MAX_DUTY,
+				 MLXSW_THERMAL_MAX_STATE);
+}
+
+static inline int mlxsw_duty_to_state(u8 duty)
+{
+	return DIV_ROUND_CLOSEST(duty * MLXSW_THERMAL_MAX_STATE,
+				 MLXSW_THERMAL_MAX_DUTY);
+}
+
+static int mlxsw_get_cooling_device_idx(struct mlxsw_thermal *thermal,
+					struct thermal_cooling_device *cdev)
+{
+	int i;
+
+	for (i = 0; i < MLXSW_MFCR_PWMS_MAX; i++)
+		if (thermal->cdevs[i] == cdev)
+			return i;
+
+	/* Allow mlxsw thermal zone binding to an external cooling device */
+	for (i = 0; i < ARRAY_SIZE(mlxsw_thermal_external_allowed_cdev); i++) {
+		if (!strcmp(cdev->type, mlxsw_thermal_external_allowed_cdev[i]))
+			return 0;
+	}
+
+	return -ENODEV;
+}
+
+static void
+mlxsw_thermal_module_trips_reset(struct mlxsw_thermal_module *tz)
+{
+	tz->trips[MLXSW_THERMAL_TEMP_TRIP_NORM].temp = 0;
+	tz->trips[MLXSW_THERMAL_TEMP_TRIP_HIGH].temp = 0;
+	tz->trips[MLXSW_THERMAL_TEMP_TRIP_HOT].temp = 0;
+}
+
+static int
+mlxsw_thermal_module_trips_update(struct device *dev, struct mlxsw_core *core,
+				  struct mlxsw_thermal_module *tz,
+				  int crit_temp, int emerg_temp)
+{
+	int err;
+
+	/* Do not try to query temperature thresholds directly from the module's
+	 * EEPROM if we got valid thresholds from MTMP.
+	 */
+	if (!emerg_temp || !crit_temp) {
+		err = mlxsw_env_module_temp_thresholds_get(core, tz->slot_index,
+							   tz->module,
+							   SFP_TEMP_HIGH_WARN,
+							   &crit_temp);
+		if (err)
+			return err;
+
+		err = mlxsw_env_module_temp_thresholds_get(core, tz->slot_index,
+							   tz->module,
+							   SFP_TEMP_HIGH_ALARM,
+							   &emerg_temp);
+		if (err)
+			return err;
+	}
+
+	if (crit_temp > emerg_temp) {
+		dev_warn(dev, "%s : Critical threshold %d is above emergency threshold %d\n",
+			 tz->tzdev->type, crit_temp, emerg_temp);
+		return 0;
+	}
+
+	/* According to the system thermal requirements, the thermal zones are
+	 * defined with three trip points. The critical and emergency
+	 * temperature thresholds, provided by QSFP module are set as "active"
+	 * and "hot" trip points, "normal" trip point is derived from "active"
+	 * by subtracting double hysteresis value.
+	 */
+	if (crit_temp >= MLXSW_THERMAL_MODULE_TEMP_SHIFT)
+		tz->trips[MLXSW_THERMAL_TEMP_TRIP_NORM].temp = crit_temp -
+					MLXSW_THERMAL_MODULE_TEMP_SHIFT;
+	else
+		tz->trips[MLXSW_THERMAL_TEMP_TRIP_NORM].temp = crit_temp;
+	tz->trips[MLXSW_THERMAL_TEMP_TRIP_HIGH].temp = crit_temp;
+	tz->trips[MLXSW_THERMAL_TEMP_TRIP_HOT].temp = emerg_temp;
+
+	return 0;
+}
+
+static int mlxsw_thermal_bind(struct thermal_zone_device *tzdev,
+			      struct thermal_cooling_device *cdev)
+{
+	struct mlxsw_thermal *thermal = tzdev->devdata;
+	struct device *dev = thermal->bus_info->dev;
+	int i, err;
+
+	/* If the cooling device is one of ours bind it */
+	if (mlxsw_get_cooling_device_idx(thermal, cdev) < 0)
+		return 0;
+
+	for (i = 0; i < MLXSW_THERMAL_NUM_TRIPS; i++) {
+		const struct mlxsw_thermal_trip *trip = &thermal->trips[i];
+
+		err = thermal_zone_bind_cooling_device(tzdev, i, cdev,
+						       trip->max_state,
+						       trip->min_state,
+						       THERMAL_WEIGHT_DEFAULT);
+		if (err < 0) {
+			dev_err(dev, "Failed to bind cooling device to trip %d\n", i);
+			return err;
+		}
+	}
+	return 0;
+}
+
+static int mlxsw_thermal_unbind(struct thermal_zone_device *tzdev,
+				struct thermal_cooling_device *cdev)
+{
+	struct mlxsw_thermal *thermal = tzdev->devdata;
+	struct device *dev = thermal->bus_info->dev;
+	int i;
+	int err;
+
+	/* If the cooling device is our one unbind it */
+	if (mlxsw_get_cooling_device_idx(thermal, cdev) < 0)
+		return 0;
+
+	for (i = 0; i < MLXSW_THERMAL_NUM_TRIPS; i++) {
+		err = thermal_zone_unbind_cooling_device(tzdev, i, cdev);
+		if (err < 0) {
+			dev_err(dev, "Failed to unbind cooling device\n");
+			return err;
+		}
+	}
+	return 0;
+}
+
+static int mlxsw_thermal_get_temp(struct thermal_zone_device *tzdev,
+				  int *p_temp)
+{
+	struct mlxsw_thermal *thermal = tzdev->devdata;
+	struct device *dev = thermal->bus_info->dev;
+	char mtmp_pl[MLXSW_REG_MTMP_LEN];
+	int temp;
+	int err;
+
+	mlxsw_reg_mtmp_pack(mtmp_pl, 0, 0, false, false);
+
+	err = mlxsw_reg_query(thermal->core, MLXSW_REG(mtmp), mtmp_pl);
+	if (err) {
+		dev_err(dev, "Failed to query temp sensor\n");
+		return err;
+	}
+	mlxsw_reg_mtmp_unpack(mtmp_pl, &temp, NULL, NULL, NULL, NULL);
+
+	*p_temp = temp;
+	return 0;
+}
+
+static int mlxsw_thermal_get_trip_type(struct thermal_zone_device *tzdev,
+				       int trip,
+				       enum thermal_trip_type *p_type)
+{
+	struct mlxsw_thermal *thermal = tzdev->devdata;
+
+	if (trip < 0 || trip >= MLXSW_THERMAL_NUM_TRIPS)
+		return -EINVAL;
+
+	*p_type = thermal->trips[trip].type;
+	return 0;
+}
+
+static int mlxsw_thermal_get_trip_temp(struct thermal_zone_device *tzdev,
+				       int trip, int *p_temp)
+{
+	struct mlxsw_thermal *thermal = tzdev->devdata;
+
+	if (trip < 0 || trip >= MLXSW_THERMAL_NUM_TRIPS)
+		return -EINVAL;
+
+	*p_temp = thermal->trips[trip].temp;
+	return 0;
+}
+
+static int mlxsw_thermal_set_trip_temp(struct thermal_zone_device *tzdev,
+				       int trip, int temp)
+{
+	struct mlxsw_thermal *thermal = tzdev->devdata;
+
+	if (trip < 0 || trip >= MLXSW_THERMAL_NUM_TRIPS)
+		return -EINVAL;
+
+	thermal->trips[trip].temp = temp;
+	return 0;
+}
+
+static int mlxsw_thermal_get_trip_hyst(struct thermal_zone_device *tzdev,
+				       int trip, int *p_hyst)
+{
+	struct mlxsw_thermal *thermal = tzdev->devdata;
+
+	*p_hyst = thermal->trips[trip].hyst;
+	return 0;
+}
+
+static int mlxsw_thermal_set_trip_hyst(struct thermal_zone_device *tzdev,
+				       int trip, int hyst)
+{
+	struct mlxsw_thermal *thermal = tzdev->devdata;
+
+	thermal->trips[trip].hyst = hyst;
+	return 0;
+}
+
+static struct thermal_zone_params mlxsw_thermal_params = {
+	.no_hwmon = true,
+};
+
+static struct thermal_zone_device_ops mlxsw_thermal_ops = {
+	.bind = mlxsw_thermal_bind,
+	.unbind = mlxsw_thermal_unbind,
+	.get_temp = mlxsw_thermal_get_temp,
+	.get_trip_type	= mlxsw_thermal_get_trip_type,
+	.get_trip_temp	= mlxsw_thermal_get_trip_temp,
+	.set_trip_temp	= mlxsw_thermal_set_trip_temp,
+	.get_trip_hyst	= mlxsw_thermal_get_trip_hyst,
+	.set_trip_hyst	= mlxsw_thermal_set_trip_hyst,
+};
+
+static int mlxsw_thermal_module_bind(struct thermal_zone_device *tzdev,
+				     struct thermal_cooling_device *cdev)
+{
+	struct mlxsw_thermal_module *tz = tzdev->devdata;
+	struct mlxsw_thermal *thermal = tz->parent;
+	int i, j, err;
+
+	/* If the cooling device is one of ours bind it */
+	if (mlxsw_get_cooling_device_idx(thermal, cdev) < 0)
+		return 0;
+
+	for (i = 0; i < MLXSW_THERMAL_NUM_TRIPS; i++) {
+		const struct mlxsw_thermal_trip *trip = &tz->trips[i];
+
+		err = thermal_zone_bind_cooling_device(tzdev, i, cdev,
+						       trip->max_state,
+						       trip->min_state,
+						       THERMAL_WEIGHT_DEFAULT);
+		if (err < 0)
+			goto err_thermal_zone_bind_cooling_device;
+	}
+	return 0;
+
+err_thermal_zone_bind_cooling_device:
+	for (j = i - 1; j >= 0; j--)
+		thermal_zone_unbind_cooling_device(tzdev, j, cdev);
+	return err;
+}
+
+static int mlxsw_thermal_module_unbind(struct thermal_zone_device *tzdev,
+				       struct thermal_cooling_device *cdev)
+{
+	struct mlxsw_thermal_module *tz = tzdev->devdata;
+	struct mlxsw_thermal *thermal = tz->parent;
+	int i;
+	int err;
+
+	/* If the cooling device is one of ours unbind it */
+	if (mlxsw_get_cooling_device_idx(thermal, cdev) < 0)
+		return 0;
+
+	for (i = 0; i < MLXSW_THERMAL_NUM_TRIPS; i++) {
+		err = thermal_zone_unbind_cooling_device(tzdev, i, cdev);
+		WARN_ON(err);
+	}
+	return err;
+}
+
+static void
+mlxsw_thermal_module_temp_and_thresholds_get(struct mlxsw_core *core,
+					     u8 slot_index, u16 sensor_index,
+					     int *p_temp, int *p_crit_temp,
+					     int *p_emerg_temp)
+{
+	char mtmp_pl[MLXSW_REG_MTMP_LEN];
+	int err;
+
+	/* Read module temperature and thresholds. */
+	mlxsw_reg_mtmp_pack(mtmp_pl, slot_index, sensor_index,
+			    false, false);
+	err = mlxsw_reg_query(core, MLXSW_REG(mtmp), mtmp_pl);
+	if (err) {
+		/* Set temperature and thresholds to zero to avoid passing
+		 * uninitialized data back to the caller.
+		 */
+		*p_temp = 0;
+		*p_crit_temp = 0;
+		*p_emerg_temp = 0;
+
+		return;
+	}
+	mlxsw_reg_mtmp_unpack(mtmp_pl, p_temp, NULL, p_crit_temp, p_emerg_temp,
+			      NULL);
+}
+
+static int mlxsw_thermal_module_temp_get(struct thermal_zone_device *tzdev,
+					 int *p_temp)
+{
+	struct mlxsw_thermal_module *tz = tzdev->devdata;
+	struct mlxsw_thermal *thermal = tz->parent;
+	int temp, crit_temp, emerg_temp;
+	struct device *dev;
+	u16 sensor_index;
+
+	dev = thermal->bus_info->dev;
+	sensor_index = MLXSW_REG_MTMP_MODULE_INDEX_MIN + tz->module;
+
+	/* Read module temperature and thresholds. */
+	mlxsw_thermal_module_temp_and_thresholds_get(thermal->core,
+						     tz->slot_index,
+						     sensor_index, &temp,
+						     &crit_temp, &emerg_temp);
+	*p_temp = temp;
+
+	if (!temp)
+		return 0;
+
+	/* Update trip points. */
+	mlxsw_thermal_module_trips_update(dev, thermal->core, tz,
+					  crit_temp, emerg_temp);
+
+	return 0;
+}
+
+static int
+mlxsw_thermal_module_trip_type_get(struct thermal_zone_device *tzdev, int trip,
+				   enum thermal_trip_type *p_type)
+{
+	struct mlxsw_thermal_module *tz = tzdev->devdata;
+
+	if (trip < 0 || trip >= MLXSW_THERMAL_NUM_TRIPS)
+		return -EINVAL;
+
+	*p_type = tz->trips[trip].type;
+	return 0;
+}
+
+static int
+mlxsw_thermal_module_trip_temp_get(struct thermal_zone_device *tzdev,
+				   int trip, int *p_temp)
+{
+	struct mlxsw_thermal_module *tz = tzdev->devdata;
+
+	if (trip < 0 || trip >= MLXSW_THERMAL_NUM_TRIPS)
+		return -EINVAL;
+
+	*p_temp = tz->trips[trip].temp;
+	return 0;
+}
+
+static int
+mlxsw_thermal_module_trip_temp_set(struct thermal_zone_device *tzdev,
+				   int trip, int temp)
+{
+	struct mlxsw_thermal_module *tz = tzdev->devdata;
+
+	if (trip < 0 || trip >= MLXSW_THERMAL_NUM_TRIPS)
+		return -EINVAL;
+
+	tz->trips[trip].temp = temp;
+	return 0;
+}
+
+static int
+mlxsw_thermal_module_trip_hyst_get(struct thermal_zone_device *tzdev, int trip,
+				   int *p_hyst)
+{
+	struct mlxsw_thermal_module *tz = tzdev->devdata;
+
+	*p_hyst = tz->trips[trip].hyst;
+	return 0;
+}
+
+static int
+mlxsw_thermal_module_trip_hyst_set(struct thermal_zone_device *tzdev, int trip,
+				   int hyst)
+{
+	struct mlxsw_thermal_module *tz = tzdev->devdata;
+
+	tz->trips[trip].hyst = hyst;
+	return 0;
+}
+
+static struct thermal_zone_device_ops mlxsw_thermal_module_ops = {
+	.bind		= mlxsw_thermal_module_bind,
+	.unbind		= mlxsw_thermal_module_unbind,
+	.get_temp	= mlxsw_thermal_module_temp_get,
+	.get_trip_type	= mlxsw_thermal_module_trip_type_get,
+	.get_trip_temp	= mlxsw_thermal_module_trip_temp_get,
+	.set_trip_temp	= mlxsw_thermal_module_trip_temp_set,
+	.get_trip_hyst	= mlxsw_thermal_module_trip_hyst_get,
+	.set_trip_hyst	= mlxsw_thermal_module_trip_hyst_set,
+};
+
+static int mlxsw_thermal_gearbox_temp_get(struct thermal_zone_device *tzdev,
+					  int *p_temp)
+{
+	struct mlxsw_thermal_module *tz = tzdev->devdata;
+	struct mlxsw_thermal *thermal = tz->parent;
+	char mtmp_pl[MLXSW_REG_MTMP_LEN];
+	u16 index;
+	int temp;
+	int err;
+
+	index = MLXSW_REG_MTMP_GBOX_INDEX_MIN + tz->module;
+	mlxsw_reg_mtmp_pack(mtmp_pl, tz->slot_index, index, false, false);
+
+	err = mlxsw_reg_query(thermal->core, MLXSW_REG(mtmp), mtmp_pl);
+	if (err)
+		return err;
+
+	mlxsw_reg_mtmp_unpack(mtmp_pl, &temp, NULL, NULL, NULL, NULL);
+
+	*p_temp = temp;
+	return 0;
+}
+
+static struct thermal_zone_device_ops mlxsw_thermal_gearbox_ops = {
+	.bind		= mlxsw_thermal_module_bind,
+	.unbind		= mlxsw_thermal_module_unbind,
+	.get_temp	= mlxsw_thermal_gearbox_temp_get,
+	.get_trip_type	= mlxsw_thermal_module_trip_type_get,
+	.get_trip_temp	= mlxsw_thermal_module_trip_temp_get,
+	.set_trip_temp	= mlxsw_thermal_module_trip_temp_set,
+	.get_trip_hyst	= mlxsw_thermal_module_trip_hyst_get,
+	.set_trip_hyst	= mlxsw_thermal_module_trip_hyst_set,
+};
+
+static int mlxsw_thermal_get_max_state(struct thermal_cooling_device *cdev,
+				       unsigned long *p_state)
+{
+	*p_state = MLXSW_THERMAL_MAX_STATE;
+	return 0;
+}
+
+static int mlxsw_thermal_get_cur_state(struct thermal_cooling_device *cdev,
+				       unsigned long *p_state)
+
+{
+	struct mlxsw_thermal *thermal = cdev->devdata;
+	struct device *dev = thermal->bus_info->dev;
+	char mfsc_pl[MLXSW_REG_MFSC_LEN];
+	int err, idx;
+	u8 duty;
+
+	idx = mlxsw_get_cooling_device_idx(thermal, cdev);
+	if (idx < 0)
+		return idx;
+
+	mlxsw_reg_mfsc_pack(mfsc_pl, idx, 0);
+	err = mlxsw_reg_query(thermal->core, MLXSW_REG(mfsc), mfsc_pl);
+	if (err) {
+		dev_err(dev, "Failed to query PWM duty\n");
+		return err;
+	}
+
+	duty = mlxsw_reg_mfsc_pwm_duty_cycle_get(mfsc_pl);
+	*p_state = mlxsw_duty_to_state(duty);
+	return 0;
+}
+
+static int mlxsw_thermal_set_cur_state(struct thermal_cooling_device *cdev,
+				       unsigned long state)
+
+{
+	struct mlxsw_thermal *thermal = cdev->devdata;
+	struct device *dev = thermal->bus_info->dev;
+	char mfsc_pl[MLXSW_REG_MFSC_LEN];
+	int idx;
+	int err;
+
+	if (state > MLXSW_THERMAL_MAX_STATE)
+		return -EINVAL;
+
+	idx = mlxsw_get_cooling_device_idx(thermal, cdev);
+	if (idx < 0)
+		return idx;
+
+	/* Normalize the state to the valid speed range. */
+	state = max_t(unsigned long, MLXSW_THERMAL_MIN_STATE, state);
+	mlxsw_reg_mfsc_pack(mfsc_pl, idx, mlxsw_state_to_duty(state));
+	err = mlxsw_reg_write(thermal->core, MLXSW_REG(mfsc), mfsc_pl);
+	if (err) {
+		dev_err(dev, "Failed to write PWM duty\n");
+		return err;
+	}
+	return 0;
+}
+
+static const struct thermal_cooling_device_ops mlxsw_cooling_ops = {
+	.get_max_state	= mlxsw_thermal_get_max_state,
+	.get_cur_state	= mlxsw_thermal_get_cur_state,
+	.set_cur_state	= mlxsw_thermal_set_cur_state,
+};
+
+static int
+mlxsw_thermal_module_tz_init(struct mlxsw_thermal_module *module_tz)
+{
+	char tz_name[THERMAL_NAME_LENGTH];
+	int err;
+
+	if (module_tz->slot_index)
+		snprintf(tz_name, sizeof(tz_name), "mlxsw-lc%d-module%d",
+			 module_tz->slot_index, module_tz->module + 1);
+	else
+		snprintf(tz_name, sizeof(tz_name), "mlxsw-module%d",
+			 module_tz->module + 1);
+	module_tz->tzdev = thermal_zone_device_register(tz_name,
+							MLXSW_THERMAL_NUM_TRIPS,
+							MLXSW_THERMAL_TRIP_MASK,
+							module_tz,
+							&mlxsw_thermal_module_ops,
+							&mlxsw_thermal_params,
+							0,
+							module_tz->parent->polling_delay);
+	if (IS_ERR(module_tz->tzdev)) {
+		err = PTR_ERR(module_tz->tzdev);
+		return err;
+	}
+
+	err = thermal_zone_device_enable(module_tz->tzdev);
+	if (err)
+		thermal_zone_device_unregister(module_tz->tzdev);
+
+	return err;
+}
+
+static void mlxsw_thermal_module_tz_fini(struct thermal_zone_device *tzdev)
+{
+	thermal_zone_device_unregister(tzdev);
+}
+
+static int
+mlxsw_thermal_module_init(struct device *dev, struct mlxsw_core *core,
+			  struct mlxsw_thermal *thermal,
+			  struct mlxsw_thermal_area *area, u8 module)
+{
+	struct mlxsw_thermal_module *module_tz;
+	int dummy_temp, crit_temp, emerg_temp;
+	u16 sensor_index;
+
+	sensor_index = MLXSW_REG_MTMP_MODULE_INDEX_MIN + module;
+	module_tz = &area->tz_module_arr[module];
+	/* Skip if parent is already set (case of port split). */
+	if (module_tz->parent)
+		return 0;
+	module_tz->module = module;
+	module_tz->slot_index = area->slot_index;
+	module_tz->parent = thermal;
+	memcpy(module_tz->trips, default_thermal_trips,
+	       sizeof(thermal->trips));
+	/* Initialize all trip point. */
+	mlxsw_thermal_module_trips_reset(module_tz);
+	/* Read module temperature and thresholds. */
+	mlxsw_thermal_module_temp_and_thresholds_get(core, area->slot_index,
+						     sensor_index, &dummy_temp,
+						     &crit_temp, &emerg_temp);
+	/* Update trip point according to the module data. */
+	return mlxsw_thermal_module_trips_update(dev, core, module_tz,
+						 crit_temp, emerg_temp);
+}
+
+static void mlxsw_thermal_module_fini(struct mlxsw_thermal_module *module_tz)
+{
+	if (module_tz && module_tz->tzdev) {
+		mlxsw_thermal_module_tz_fini(module_tz->tzdev);
+		module_tz->tzdev = NULL;
+		module_tz->parent = NULL;
+	}
+}
+
+static int
+mlxsw_thermal_modules_init(struct device *dev, struct mlxsw_core *core,
+			   struct mlxsw_thermal *thermal,
+			   struct mlxsw_thermal_area *area)
+{
+	struct mlxsw_thermal_module *module_tz;
+	char mgpir_pl[MLXSW_REG_MGPIR_LEN];
+	int i, err;
+
+	mlxsw_reg_mgpir_pack(mgpir_pl, area->slot_index);
+	err = mlxsw_reg_query(core, MLXSW_REG(mgpir), mgpir_pl);
+	if (err)
+		return err;
+
+	mlxsw_reg_mgpir_unpack(mgpir_pl, NULL, NULL, NULL,
+			       &area->tz_module_num, NULL);
+
+	/* For modular system module counter could be zero. */
+	if (!area->tz_module_num)
+		return 0;
+
+	area->tz_module_arr = kcalloc(area->tz_module_num,
+				      sizeof(*area->tz_module_arr),
+				      GFP_KERNEL);
+	if (!area->tz_module_arr)
+		return -ENOMEM;
+
+	for (i = 0; i < area->tz_module_num; i++) {
+		err = mlxsw_thermal_module_init(dev, core, thermal, area, i);
+		if (err)
+			goto err_thermal_module_init;
+	}
+
+	for (i = 0; i < area->tz_module_num; i++) {
+		module_tz = &area->tz_module_arr[i];
+		if (!module_tz->parent)
+			continue;
+		err = mlxsw_thermal_module_tz_init(module_tz);
+		if (err)
+			goto err_thermal_module_tz_init;
+	}
+
+	return 0;
+
+err_thermal_module_tz_init:
+err_thermal_module_init:
+	for (i = area->tz_module_num - 1; i >= 0; i--)
+		mlxsw_thermal_module_fini(&area->tz_module_arr[i]);
+	kfree(area->tz_module_arr);
+	return err;
+}
+
+static void
+mlxsw_thermal_modules_fini(struct mlxsw_thermal *thermal,
+			   struct mlxsw_thermal_area *area)
+{
+	int i;
+
+	for (i = area->tz_module_num - 1; i >= 0; i--)
+		mlxsw_thermal_module_fini(&area->tz_module_arr[i]);
+	kfree(area->tz_module_arr);
+}
+
+static int
+mlxsw_thermal_gearbox_tz_init(struct mlxsw_thermal_module *gearbox_tz)
+{
+	char tz_name[THERMAL_NAME_LENGTH];
+	int ret;
+
+	if (gearbox_tz->slot_index)
+		snprintf(tz_name, sizeof(tz_name), "mlxsw-lc%d-gearbox%d",
+			 gearbox_tz->slot_index, gearbox_tz->module + 1);
+	else
+		snprintf(tz_name, sizeof(tz_name), "mlxsw-gearbox%d",
+			 gearbox_tz->module + 1);
+	gearbox_tz->tzdev = thermal_zone_device_register(tz_name,
+						MLXSW_THERMAL_NUM_TRIPS,
+						MLXSW_THERMAL_TRIP_MASK,
+						gearbox_tz,
+						&mlxsw_thermal_gearbox_ops,
+						&mlxsw_thermal_params, 0,
+						gearbox_tz->parent->polling_delay);
+	if (IS_ERR(gearbox_tz->tzdev))
+		return PTR_ERR(gearbox_tz->tzdev);
+
+	ret = thermal_zone_device_enable(gearbox_tz->tzdev);
+	if (ret)
+		thermal_zone_device_unregister(gearbox_tz->tzdev);
+
+	return ret;
+}
+
+static void
+mlxsw_thermal_gearbox_tz_fini(struct mlxsw_thermal_module *gearbox_tz)
+{
+	thermal_zone_device_unregister(gearbox_tz->tzdev);
+}
+
+static int
+mlxsw_thermal_gearboxes_init(struct device *dev, struct mlxsw_core *core,
+			     struct mlxsw_thermal *thermal,
+			     struct mlxsw_thermal_area *area)
+{
+	enum mlxsw_reg_mgpir_device_type device_type;
+	struct mlxsw_thermal_module *gearbox_tz;
+	char mgpir_pl[MLXSW_REG_MGPIR_LEN];
+	u8 gbox_num;
+	int i;
+	int err;
+
+	mlxsw_reg_mgpir_pack(mgpir_pl, area->slot_index);
+	err = mlxsw_reg_query(core, MLXSW_REG(mgpir), mgpir_pl);
+	if (err)
+		return err;
+
+	mlxsw_reg_mgpir_unpack(mgpir_pl, &gbox_num, &device_type, NULL,
+			       NULL, NULL);
+	if (device_type != MLXSW_REG_MGPIR_DEVICE_TYPE_GEARBOX_DIE ||
+	    !gbox_num)
+		return 0;
+
+	area->tz_gearbox_num = gbox_num;
+	area->tz_gearbox_arr = kcalloc(area->tz_gearbox_num,
+				       sizeof(*area->tz_gearbox_arr),
+				       GFP_KERNEL);
+	if (!area->tz_gearbox_arr)
+		return -ENOMEM;
+
+	for (i = 0; i < area->tz_gearbox_num; i++) {
+		gearbox_tz = &area->tz_gearbox_arr[i];
+		memcpy(gearbox_tz->trips, default_thermal_trips,
+		       sizeof(thermal->trips));
+		gearbox_tz->module = i;
+		gearbox_tz->parent = thermal;
+		gearbox_tz->slot_index = area->slot_index;
+		err = mlxsw_thermal_gearbox_tz_init(gearbox_tz);
+		if (err)
+			goto err_thermal_gearbox_tz_init;
+	}
+
+	return 0;
+
+err_thermal_gearbox_tz_init:
+	for (i--; i >= 0; i--)
+		mlxsw_thermal_gearbox_tz_fini(&area->tz_gearbox_arr[i]);
+	kfree(area->tz_gearbox_arr);
+	return err;
+}
+
+static void
+mlxsw_thermal_gearboxes_fini(struct mlxsw_thermal *thermal,
+			     struct mlxsw_thermal_area *area)
+{
+	int i;
+
+	for (i = area->tz_gearbox_num - 1; i >= 0; i--)
+		mlxsw_thermal_gearbox_tz_fini(&area->tz_gearbox_arr[i]);
+	kfree(area->tz_gearbox_arr);
+}
+
+static void
+mlxsw_thermal_got_active(struct mlxsw_core *mlxsw_core, u8 slot_index,
+			 void *priv)
+{
+	struct mlxsw_thermal *thermal = priv;
+	struct mlxsw_thermal_area *linecard;
+	int err;
+
+	linecard = &thermal->line_cards[slot_index];
+
+	if (linecard->active)
+		return;
+
+	linecard->slot_index = slot_index;
+	err = mlxsw_thermal_modules_init(thermal->bus_info->dev, thermal->core,
+					 thermal, linecard);
+	if (err) {
+		dev_err(thermal->bus_info->dev, "Failed to configure thermal objects for line card modules in slot %d\n",
+			slot_index);
+		return;
+	}
+
+	err = mlxsw_thermal_gearboxes_init(thermal->bus_info->dev,
+					   thermal->core, thermal, linecard);
+	if (err) {
+		dev_err(thermal->bus_info->dev, "Failed to configure thermal objects for line card gearboxes in slot %d\n",
+			slot_index);
+		goto err_thermal_linecard_gearboxes_init;
+	}
+
+	linecard->active = true;
+
+	return;
+
+err_thermal_linecard_gearboxes_init:
+	mlxsw_thermal_modules_fini(thermal, linecard);
+}
+
+static void
+mlxsw_thermal_got_inactive(struct mlxsw_core *mlxsw_core, u8 slot_index,
+			   void *priv)
+{
+	struct mlxsw_thermal *thermal = priv;
+	struct mlxsw_thermal_area *linecard;
+
+	linecard = &thermal->line_cards[slot_index];
+	if (!linecard->active)
+		return;
+	linecard->active = false;
+	mlxsw_thermal_gearboxes_fini(thermal, linecard);
+	mlxsw_thermal_modules_fini(thermal, linecard);
+}
+
+static struct mlxsw_linecards_event_ops mlxsw_thermal_event_ops = {
+	.got_active = mlxsw_thermal_got_active,
+	.got_inactive = mlxsw_thermal_got_inactive,
+};
+
+int mlxsw_thermal_init(struct mlxsw_core *core,
+		       const struct mlxsw_bus_info *bus_info,
+		       struct mlxsw_thermal **p_thermal)
+{
+	char mfcr_pl[MLXSW_REG_MFCR_LEN] = { 0 };
+	enum mlxsw_reg_mfcr_pwm_frequency freq;
+	struct device *dev = bus_info->dev;
+	char mgpir_pl[MLXSW_REG_MGPIR_LEN];
+	struct mlxsw_thermal *thermal;
+	u8 pwm_active, num_of_slots;
+	u16 tacho_active;
+	int err, i;
+
+	mlxsw_reg_mgpir_pack(mgpir_pl, 0);
+	err = mlxsw_reg_query(core, MLXSW_REG(mgpir), mgpir_pl);
+	if (err)
+		return err;
+
+	mlxsw_reg_mgpir_unpack(mgpir_pl, NULL, NULL, NULL, NULL,
+			       &num_of_slots);
+
+	thermal = kzalloc(struct_size(thermal, line_cards, num_of_slots + 1),
+			  GFP_KERNEL);
+	if (!thermal)
+		return -ENOMEM;
+
+	thermal->core = core;
+	thermal->bus_info = bus_info;
+	memcpy(thermal->trips, default_thermal_trips, sizeof(thermal->trips));
+	thermal->line_cards[0].slot_index = 0;
+
+	err = mlxsw_reg_query(thermal->core, MLXSW_REG(mfcr), mfcr_pl);
+	if (err) {
+		dev_err(dev, "Failed to probe PWMs\n");
+		goto err_reg_query;
+	}
+	mlxsw_reg_mfcr_unpack(mfcr_pl, &freq, &tacho_active, &pwm_active);
+
+	for (i = 0; i < MLXSW_MFCR_TACHOS_MAX; i++) {
+		if (tacho_active & BIT(i)) {
+			char mfsl_pl[MLXSW_REG_MFSL_LEN];
+
+			mlxsw_reg_mfsl_pack(mfsl_pl, i, 0, 0);
+
+			/* We need to query the register to preserve maximum */
+			err = mlxsw_reg_query(thermal->core, MLXSW_REG(mfsl),
+					      mfsl_pl);
+			if (err)
+				goto err_reg_query;
+
+			/* set the minimal RPMs to 0 */
+			mlxsw_reg_mfsl_tach_min_set(mfsl_pl, 0);
+			err = mlxsw_reg_write(thermal->core, MLXSW_REG(mfsl),
+					      mfsl_pl);
+			if (err)
+				goto err_reg_write;
+		}
+	}
+	for (i = 0; i < MLXSW_MFCR_PWMS_MAX; i++) {
+		if (pwm_active & BIT(i)) {
+			struct thermal_cooling_device *cdev;
+
+			cdev = thermal_cooling_device_register("mlxsw_fan",
+							       thermal,
+							       &mlxsw_cooling_ops);
+			if (IS_ERR(cdev)) {
+				err = PTR_ERR(cdev);
+				dev_err(dev, "Failed to register cooling device\n");
+				goto err_thermal_cooling_device_register;
+			}
+			thermal->cdevs[i] = cdev;
+		}
+	}
+
+	thermal->polling_delay = bus_info->low_frequency ?
+				 MLXSW_THERMAL_SLOW_POLL_INT :
+				 MLXSW_THERMAL_POLL_INT;
+
+	thermal->tzdev = thermal_zone_device_register("mlxsw",
+						      MLXSW_THERMAL_NUM_TRIPS,
+						      MLXSW_THERMAL_TRIP_MASK,
+						      thermal,
+						      &mlxsw_thermal_ops,
+						      &mlxsw_thermal_params, 0,
+						      thermal->polling_delay);
+	if (IS_ERR(thermal->tzdev)) {
+		err = PTR_ERR(thermal->tzdev);
+		dev_err(dev, "Failed to register thermal zone\n");
+		goto err_thermal_zone_device_register;
+	}
+
+	err = mlxsw_thermal_modules_init(dev, core, thermal,
+					 &thermal->line_cards[0]);
+	if (err)
+		goto err_thermal_modules_init;
+
+	err = mlxsw_thermal_gearboxes_init(dev, core, thermal,
+					   &thermal->line_cards[0]);
+	if (err)
+		goto err_thermal_gearboxes_init;
+
+	err = mlxsw_linecards_event_ops_register(core,
+						 &mlxsw_thermal_event_ops,
+						 thermal);
+	if (err)
+		goto err_linecards_event_ops_register;
+
+	err = thermal_zone_device_enable(thermal->tzdev);
+	if (err)
+		goto err_thermal_zone_device_enable;
+
+	thermal->line_cards[0].active = true;
+	*p_thermal = thermal;
+	return 0;
+
+err_thermal_zone_device_enable:
+	mlxsw_linecards_event_ops_unregister(thermal->core,
+					     &mlxsw_thermal_event_ops,
+					     thermal);
+err_linecards_event_ops_register:
+	mlxsw_thermal_gearboxes_fini(thermal, &thermal->line_cards[0]);
+err_thermal_gearboxes_init:
+	mlxsw_thermal_modules_fini(thermal, &thermal->line_cards[0]);
+err_thermal_modules_init:
+	if (thermal->tzdev) {
+		thermal_zone_device_unregister(thermal->tzdev);
+		thermal->tzdev = NULL;
+	}
+err_thermal_zone_device_register:
+err_thermal_cooling_device_register:
+	for (i = 0; i < MLXSW_MFCR_PWMS_MAX; i++)
+		if (thermal->cdevs[i])
+			thermal_cooling_device_unregister(thermal->cdevs[i]);
+err_reg_write:
+err_reg_query:
+	kfree(thermal);
+	return err;
+}
+
+void mlxsw_thermal_fini(struct mlxsw_thermal *thermal)
+{
+	int i;
+
+	thermal->line_cards[0].active = false;
+	mlxsw_linecards_event_ops_unregister(thermal->core,
+					     &mlxsw_thermal_event_ops,
+					     thermal);
+	mlxsw_thermal_gearboxes_fini(thermal, &thermal->line_cards[0]);
+	mlxsw_thermal_modules_fini(thermal, &thermal->line_cards[0]);
+	if (thermal->tzdev) {
+		thermal_zone_device_unregister(thermal->tzdev);
+		thermal->tzdev = NULL;
+	}
+
+	for (i = 0; i < MLXSW_MFCR_PWMS_MAX; i++) {
+		if (thermal->cdevs[i]) {
+			thermal_cooling_device_unregister(thermal->cdevs[i]);
+			thermal->cdevs[i] = NULL;
+		}
+	}
+
+	kfree(thermal);
+}
diff --git a/drivers/net/ethernet/mellanox/mlxsw/emad.h b/drivers/net/ethernet/mellanox/mlxsw/emad.h
new file mode 100644
index 000000000..acfbbec52
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/emad.h
@@ -0,0 +1,96 @@
+/* SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0 */
+/* Copyright (c) 2015-2018 Mellanox Technologies. All rights reserved */
+
+#ifndef _MLXSW_EMAD_H
+#define _MLXSW_EMAD_H
+
+#define MLXSW_EMAD_MAX_FRAME_LEN 1518	/* Length in u8 */
+#define MLXSW_EMAD_MAX_RETRY 5
+
+/* EMAD Ethernet header */
+#define MLXSW_EMAD_ETH_HDR_LEN 0x10	/* Length in u8 */
+#define MLXSW_EMAD_EH_DMAC "\x01\x02\xc9\x00\x00\x01"
+#define MLXSW_EMAD_EH_SMAC "\x00\x02\xc9\x01\x02\x03"
+#define MLXSW_EMAD_EH_ETHERTYPE 0x8932
+#define MLXSW_EMAD_EH_MLX_PROTO 0
+#define MLXSW_EMAD_EH_PROTO_VERSION 0
+
+/* EMAD TLV Types */
+enum {
+	MLXSW_EMAD_TLV_TYPE_END,
+	MLXSW_EMAD_TLV_TYPE_OP,
+	MLXSW_EMAD_TLV_TYPE_STRING,
+	MLXSW_EMAD_TLV_TYPE_REG,
+};
+
+/* OP TLV */
+#define MLXSW_EMAD_OP_TLV_LEN 4		/* Length in u32 */
+
+enum {
+	MLXSW_EMAD_OP_TLV_CLASS_REG_ACCESS = 1,
+	MLXSW_EMAD_OP_TLV_CLASS_IPC = 2,
+};
+
+enum mlxsw_emad_op_tlv_status {
+	MLXSW_EMAD_OP_TLV_STATUS_SUCCESS,
+	MLXSW_EMAD_OP_TLV_STATUS_BUSY,
+	MLXSW_EMAD_OP_TLV_STATUS_VERSION_NOT_SUPPORTED,
+	MLXSW_EMAD_OP_TLV_STATUS_UNKNOWN_TLV,
+	MLXSW_EMAD_OP_TLV_STATUS_REGISTER_NOT_SUPPORTED,
+	MLXSW_EMAD_OP_TLV_STATUS_CLASS_NOT_SUPPORTED,
+	MLXSW_EMAD_OP_TLV_STATUS_METHOD_NOT_SUPPORTED,
+	MLXSW_EMAD_OP_TLV_STATUS_BAD_PARAMETER,
+	MLXSW_EMAD_OP_TLV_STATUS_RESOURCE_NOT_AVAILABLE,
+	MLXSW_EMAD_OP_TLV_STATUS_MESSAGE_RECEIPT_ACK,
+	MLXSW_EMAD_OP_TLV_STATUS_INTERNAL_ERROR = 0x70,
+};
+
+static inline char *mlxsw_emad_op_tlv_status_str(u8 status)
+{
+	switch (status) {
+	case MLXSW_EMAD_OP_TLV_STATUS_SUCCESS:
+		return "operation performed";
+	case MLXSW_EMAD_OP_TLV_STATUS_BUSY:
+		return "device is busy";
+	case MLXSW_EMAD_OP_TLV_STATUS_VERSION_NOT_SUPPORTED:
+		return "version not supported";
+	case MLXSW_EMAD_OP_TLV_STATUS_UNKNOWN_TLV:
+		return "unknown TLV";
+	case MLXSW_EMAD_OP_TLV_STATUS_REGISTER_NOT_SUPPORTED:
+		return "register not supported";
+	case MLXSW_EMAD_OP_TLV_STATUS_CLASS_NOT_SUPPORTED:
+		return "class not supported";
+	case MLXSW_EMAD_OP_TLV_STATUS_METHOD_NOT_SUPPORTED:
+		return "method not supported";
+	case MLXSW_EMAD_OP_TLV_STATUS_BAD_PARAMETER:
+		return "bad parameter";
+	case MLXSW_EMAD_OP_TLV_STATUS_RESOURCE_NOT_AVAILABLE:
+		return "resource not available";
+	case MLXSW_EMAD_OP_TLV_STATUS_MESSAGE_RECEIPT_ACK:
+		return "acknowledged. retransmit";
+	case MLXSW_EMAD_OP_TLV_STATUS_INTERNAL_ERROR:
+		return "internal error";
+	default:
+		return "*UNKNOWN*";
+	}
+}
+
+enum {
+	MLXSW_EMAD_OP_TLV_REQUEST,
+	MLXSW_EMAD_OP_TLV_RESPONSE
+};
+
+enum {
+	MLXSW_EMAD_OP_TLV_METHOD_QUERY = 1,
+	MLXSW_EMAD_OP_TLV_METHOD_WRITE = 2,
+	MLXSW_EMAD_OP_TLV_METHOD_SEND = 3,
+	MLXSW_EMAD_OP_TLV_METHOD_EVENT = 5,
+};
+
+/* STRING TLV */
+#define MLXSW_EMAD_STRING_TLV_LEN 33	/* Length in u32 */
+
+/* END TLV */
+#define MLXSW_EMAD_END_TLV_LEN 1	/* Length in u32 */
+
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlxsw/i2c.c b/drivers/net/ethernet/mellanox/mlxsw/i2c.c
new file mode 100644
index 000000000..3beefc167
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/i2c.c
@@ -0,0 +1,769 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2016-2018 Mellanox Technologies. All rights reserved */
+
+#include <linux/err.h>
+#include <linux/i2c.h>
+#include <linux/init.h>
+#include <linux/jiffies.h>
+#include <linux/kernel.h>
+#include <linux/mutex.h>
+#include <linux/module.h>
+#include <linux/mod_devicetable.h>
+#include <linux/platform_data/mlxreg.h>
+#include <linux/slab.h>
+
+#include "cmd.h"
+#include "core.h"
+#include "i2c.h"
+#include "resources.h"
+
+#define MLXSW_I2C_CIR2_BASE		0x72000
+#define MLXSW_I2C_CIR_STATUS_OFF	0x18
+#define MLXSW_I2C_CIR2_OFF_STATUS	(MLXSW_I2C_CIR2_BASE + \
+					 MLXSW_I2C_CIR_STATUS_OFF)
+#define MLXSW_I2C_OPMOD_SHIFT		12
+#define MLXSW_I2C_EVENT_BIT_SHIFT	22
+#define MLXSW_I2C_GO_BIT_SHIFT		23
+#define MLXSW_I2C_CIR_CTRL_STATUS_SHIFT	24
+#define MLXSW_I2C_EVENT_BIT		BIT(MLXSW_I2C_EVENT_BIT_SHIFT)
+#define MLXSW_I2C_GO_BIT		BIT(MLXSW_I2C_GO_BIT_SHIFT)
+#define MLXSW_I2C_GO_OPMODE		BIT(MLXSW_I2C_OPMOD_SHIFT)
+#define MLXSW_I2C_SET_IMM_CMD		(MLXSW_I2C_GO_OPMODE | \
+					 MLXSW_CMD_OPCODE_QUERY_FW)
+#define MLXSW_I2C_PUSH_IMM_CMD		(MLXSW_I2C_GO_BIT | \
+					 MLXSW_I2C_SET_IMM_CMD)
+#define MLXSW_I2C_SET_CMD		(MLXSW_CMD_OPCODE_ACCESS_REG)
+#define MLXSW_I2C_PUSH_CMD		(MLXSW_I2C_GO_BIT | MLXSW_I2C_SET_CMD)
+#define MLXSW_I2C_TLV_HDR_SIZE		0x10
+#define MLXSW_I2C_ADDR_WIDTH		4
+#define MLXSW_I2C_PUSH_CMD_SIZE		(MLXSW_I2C_ADDR_WIDTH + 4)
+#define MLXSW_I2C_SET_EVENT_CMD		(MLXSW_I2C_EVENT_BIT)
+#define MLXSW_I2C_PUSH_EVENT_CMD	(MLXSW_I2C_GO_BIT | \
+					 MLXSW_I2C_SET_EVENT_CMD)
+#define MLXSW_I2C_READ_SEMA_SIZE	4
+#define MLXSW_I2C_PREP_SIZE		(MLXSW_I2C_ADDR_WIDTH + 28)
+#define MLXSW_I2C_MBOX_SIZE		20
+#define MLXSW_I2C_MBOX_OUT_PARAM_OFF	12
+#define MLXSW_I2C_MBOX_OFFSET_BITS	20
+#define MLXSW_I2C_MBOX_SIZE_BITS	12
+#define MLXSW_I2C_ADDR_BUF_SIZE		4
+#define MLXSW_I2C_BLK_DEF		32
+#define MLXSW_I2C_BLK_MAX		100
+#define MLXSW_I2C_RETRY			5
+#define MLXSW_I2C_TIMEOUT_MSECS		5000
+#define MLXSW_I2C_MAX_DATA_SIZE		256
+
+/* Driver can be initialized by kernel platform driver or from the user
+ * space. In the first case IRQ line number is passed through the platform
+ * data, otherwise default IRQ line is to be used. Default IRQ is relevant
+ * only for specific I2C slave address, allowing 3.4 MHz I2C path to the chip
+ * (special hardware feature for I2C acceleration).
+ */
+#define MLXSW_I2C_DEFAULT_IRQ		17
+#define MLXSW_FAST_I2C_SLAVE		0x37
+
+/**
+ * struct mlxsw_i2c - device private data:
+ * @cmd: command attributes;
+ * @cmd.mb_size_in: input mailbox size;
+ * @cmd.mb_off_in: input mailbox offset in register space;
+ * @cmd.mb_size_out: output mailbox size;
+ * @cmd.mb_off_out: output mailbox offset in register space;
+ * @cmd.lock: command execution lock;
+ * @dev: I2C device;
+ * @core: switch core pointer;
+ * @bus_info: bus info block;
+ * @block_size: maximum block size allowed to pass to under layer;
+ * @pdata: device platform data;
+ * @irq_work: interrupts work item;
+ * @irq: IRQ line number;
+ */
+struct mlxsw_i2c {
+	struct {
+		u32 mb_size_in;
+		u32 mb_off_in;
+		u32 mb_size_out;
+		u32 mb_off_out;
+		struct mutex lock;
+	} cmd;
+	struct device *dev;
+	struct mlxsw_core *core;
+	struct mlxsw_bus_info bus_info;
+	u16 block_size;
+	struct mlxreg_core_hotplug_platform_data *pdata;
+	struct work_struct irq_work;
+	int irq;
+};
+
+#define MLXSW_I2C_READ_MSG(_client, _addr_buf, _buf, _len) {	\
+	{ .addr = (_client)->addr,				\
+	  .buf = (_addr_buf),					\
+	  .len = MLXSW_I2C_ADDR_BUF_SIZE,			\
+	  .flags = 0 },						\
+	{ .addr = (_client)->addr,				\
+	  .buf = (_buf),					\
+	  .len = (_len),					\
+	  .flags = I2C_M_RD } }
+
+#define MLXSW_I2C_WRITE_MSG(_client, _buf, _len)		\
+	{ .addr = (_client)->addr,				\
+	  .buf = (u8 *)(_buf),					\
+	  .len = (_len),					\
+	  .flags = 0 }
+
+/* Routine converts in and out mail boxes offset and size. */
+static inline void
+mlxsw_i2c_convert_mbox(struct mlxsw_i2c *mlxsw_i2c, u8 *buf)
+{
+	u32 tmp;
+
+	/* Local in/out mailboxes: 20 bits for offset, 12 for size */
+	tmp = be32_to_cpup((__be32 *) buf);
+	mlxsw_i2c->cmd.mb_off_in = tmp &
+				   GENMASK(MLXSW_I2C_MBOX_OFFSET_BITS - 1, 0);
+	mlxsw_i2c->cmd.mb_size_in = (tmp & GENMASK(31,
+					MLXSW_I2C_MBOX_OFFSET_BITS)) >>
+					MLXSW_I2C_MBOX_OFFSET_BITS;
+
+	tmp = be32_to_cpup((__be32 *) (buf + MLXSW_I2C_ADDR_WIDTH));
+	mlxsw_i2c->cmd.mb_off_out = tmp &
+				    GENMASK(MLXSW_I2C_MBOX_OFFSET_BITS - 1, 0);
+	mlxsw_i2c->cmd.mb_size_out = (tmp & GENMASK(31,
+					MLXSW_I2C_MBOX_OFFSET_BITS)) >>
+					MLXSW_I2C_MBOX_OFFSET_BITS;
+}
+
+/* Routine obtains register size from mail box buffer. */
+static inline int mlxsw_i2c_get_reg_size(u8 *in_mbox)
+{
+	u16  tmp = be16_to_cpup((__be16 *) (in_mbox + MLXSW_I2C_TLV_HDR_SIZE));
+
+	return (tmp & 0x7ff) * 4 + MLXSW_I2C_TLV_HDR_SIZE;
+}
+
+/* Routine sets I2C device internal offset in the transaction buffer. */
+static inline void mlxsw_i2c_set_slave_addr(u8 *buf, u32 off)
+{
+	__be32 *val = (__be32 *) buf;
+
+	*val = htonl(off);
+}
+
+/* Routine waits until go bit is cleared. */
+static int mlxsw_i2c_wait_go_bit(struct i2c_client *client,
+				 struct mlxsw_i2c *mlxsw_i2c, u8 *p_status)
+{
+	u8 addr_buf[MLXSW_I2C_ADDR_BUF_SIZE];
+	u8 buf[MLXSW_I2C_READ_SEMA_SIZE];
+	int len = MLXSW_I2C_READ_SEMA_SIZE;
+	struct i2c_msg read_sema[] =
+		MLXSW_I2C_READ_MSG(client, addr_buf, buf, len);
+	bool wait_done = false;
+	unsigned long end;
+	int i = 0, err;
+
+	mlxsw_i2c_set_slave_addr(addr_buf, MLXSW_I2C_CIR2_OFF_STATUS);
+
+	end = jiffies + msecs_to_jiffies(MLXSW_I2C_TIMEOUT_MSECS);
+	do {
+		u32 ctrl;
+
+		err = i2c_transfer(client->adapter, read_sema,
+				   ARRAY_SIZE(read_sema));
+
+		ctrl = be32_to_cpu(*(__be32 *) buf);
+		if (err == ARRAY_SIZE(read_sema)) {
+			if (!(ctrl & MLXSW_I2C_GO_BIT)) {
+				wait_done = true;
+				*p_status = ctrl >>
+					    MLXSW_I2C_CIR_CTRL_STATUS_SHIFT;
+				break;
+			}
+		}
+		cond_resched();
+	} while ((time_before(jiffies, end)) || (i++ < MLXSW_I2C_RETRY));
+
+	if (wait_done) {
+		if (*p_status)
+			err = -EIO;
+	} else {
+		return -ETIMEDOUT;
+	}
+
+	return err > 0 ? 0 : err;
+}
+
+/* Routine posts a command to ASIC through mail box. */
+static int mlxsw_i2c_write_cmd(struct i2c_client *client,
+			       struct mlxsw_i2c *mlxsw_i2c,
+			       int immediate)
+{
+	__be32 push_cmd_buf[MLXSW_I2C_PUSH_CMD_SIZE / 4] = {
+		0, cpu_to_be32(MLXSW_I2C_PUSH_IMM_CMD)
+	};
+	__be32 prep_cmd_buf[MLXSW_I2C_PREP_SIZE / 4] = {
+		0, 0, 0, 0, 0, 0,
+		cpu_to_be32(client->adapter->nr & 0xffff),
+		cpu_to_be32(MLXSW_I2C_SET_IMM_CMD)
+	};
+	struct i2c_msg push_cmd =
+		MLXSW_I2C_WRITE_MSG(client, push_cmd_buf,
+				    MLXSW_I2C_PUSH_CMD_SIZE);
+	struct i2c_msg prep_cmd =
+		MLXSW_I2C_WRITE_MSG(client, prep_cmd_buf, MLXSW_I2C_PREP_SIZE);
+	int err;
+
+	if (!immediate) {
+		push_cmd_buf[1] = cpu_to_be32(MLXSW_I2C_PUSH_CMD);
+		prep_cmd_buf[7] = cpu_to_be32(MLXSW_I2C_SET_CMD);
+	}
+	mlxsw_i2c_set_slave_addr((u8 *)prep_cmd_buf,
+				 MLXSW_I2C_CIR2_BASE);
+	mlxsw_i2c_set_slave_addr((u8 *)push_cmd_buf,
+				 MLXSW_I2C_CIR2_OFF_STATUS);
+
+	/* Prepare Command Interface Register for transaction */
+	err = i2c_transfer(client->adapter, &prep_cmd, 1);
+	if (err < 0)
+		return err;
+	else if (err != 1)
+		return -EIO;
+
+	/* Write out Command Interface Register GO bit to push transaction */
+	err = i2c_transfer(client->adapter, &push_cmd, 1);
+	if (err < 0)
+		return err;
+	else if (err != 1)
+		return -EIO;
+
+	return 0;
+}
+
+/* Routine posts initialization command to ASIC through mail box. */
+static int
+mlxsw_i2c_write_init_cmd(struct i2c_client *client,
+			 struct mlxsw_i2c *mlxsw_i2c, u16 opcode, u32 in_mod)
+{
+	__be32 push_cmd_buf[MLXSW_I2C_PUSH_CMD_SIZE / 4] = {
+		0, cpu_to_be32(MLXSW_I2C_PUSH_EVENT_CMD)
+	};
+	__be32 prep_cmd_buf[MLXSW_I2C_PREP_SIZE / 4] = {
+		0, 0, 0, 0, 0, 0,
+		cpu_to_be32(client->adapter->nr & 0xffff),
+		cpu_to_be32(MLXSW_I2C_SET_EVENT_CMD)
+	};
+	struct i2c_msg push_cmd =
+		MLXSW_I2C_WRITE_MSG(client, push_cmd_buf,
+				    MLXSW_I2C_PUSH_CMD_SIZE);
+	struct i2c_msg prep_cmd =
+		MLXSW_I2C_WRITE_MSG(client, prep_cmd_buf, MLXSW_I2C_PREP_SIZE);
+	u8 status;
+	int err;
+
+	push_cmd_buf[1] = cpu_to_be32(MLXSW_I2C_PUSH_EVENT_CMD | opcode);
+	prep_cmd_buf[3] = cpu_to_be32(in_mod);
+	prep_cmd_buf[7] = cpu_to_be32(MLXSW_I2C_GO_BIT | opcode);
+	mlxsw_i2c_set_slave_addr((u8 *)prep_cmd_buf,
+				 MLXSW_I2C_CIR2_BASE);
+	mlxsw_i2c_set_slave_addr((u8 *)push_cmd_buf,
+				 MLXSW_I2C_CIR2_OFF_STATUS);
+
+	/* Prepare Command Interface Register for transaction */
+	err = i2c_transfer(client->adapter, &prep_cmd, 1);
+	if (err < 0)
+		return err;
+	else if (err != 1)
+		return -EIO;
+
+	/* Write out Command Interface Register GO bit to push transaction */
+	err = i2c_transfer(client->adapter, &push_cmd, 1);
+	if (err < 0)
+		return err;
+	else if (err != 1)
+		return -EIO;
+
+	/* Wait until go bit is cleared. */
+	err = mlxsw_i2c_wait_go_bit(client, mlxsw_i2c, &status);
+	if (err) {
+		dev_err(&client->dev, "HW semaphore is not released");
+		return err;
+	}
+
+	/* Validate transaction completion status. */
+	if (status) {
+		dev_err(&client->dev, "Bad transaction completion status %x\n",
+			status);
+		return -EIO;
+	}
+
+	return 0;
+}
+
+/* Routine obtains mail box offsets from ASIC register space. */
+static int mlxsw_i2c_get_mbox(struct i2c_client *client,
+			      struct mlxsw_i2c *mlxsw_i2c)
+{
+	u8 addr_buf[MLXSW_I2C_ADDR_BUF_SIZE];
+	u8 buf[MLXSW_I2C_MBOX_SIZE];
+	struct i2c_msg mbox_cmd[] =
+		MLXSW_I2C_READ_MSG(client, addr_buf, buf, MLXSW_I2C_MBOX_SIZE);
+	int err;
+
+	/* Read mail boxes offsets. */
+	mlxsw_i2c_set_slave_addr(addr_buf, MLXSW_I2C_CIR2_BASE);
+	err = i2c_transfer(client->adapter, mbox_cmd, 2);
+	if (err != 2) {
+		dev_err(&client->dev, "Could not obtain mail boxes\n");
+		if (!err)
+			return -EIO;
+		else
+			return err;
+	}
+
+	/* Convert mail boxes. */
+	mlxsw_i2c_convert_mbox(mlxsw_i2c, &buf[MLXSW_I2C_MBOX_OUT_PARAM_OFF]);
+
+	return err;
+}
+
+/* Routine sends I2C write transaction to ASIC device. */
+static int
+mlxsw_i2c_write(struct device *dev, size_t in_mbox_size, u8 *in_mbox, int num,
+		u8 *p_status)
+{
+	struct i2c_client *client = to_i2c_client(dev);
+	struct mlxsw_i2c *mlxsw_i2c = i2c_get_clientdata(client);
+	unsigned long timeout = msecs_to_jiffies(MLXSW_I2C_TIMEOUT_MSECS);
+	int off = mlxsw_i2c->cmd.mb_off_in, chunk_size, i, j;
+	unsigned long end;
+	u8 *tran_buf;
+	struct i2c_msg write_tran =
+		MLXSW_I2C_WRITE_MSG(client, NULL, MLXSW_I2C_PUSH_CMD_SIZE);
+	int err;
+
+	tran_buf = kmalloc(mlxsw_i2c->block_size + MLXSW_I2C_ADDR_BUF_SIZE,
+			   GFP_KERNEL);
+	if (!tran_buf)
+		return -ENOMEM;
+
+	write_tran.buf = tran_buf;
+	for (i = 0; i < num; i++) {
+		chunk_size = (in_mbox_size > mlxsw_i2c->block_size) ?
+			     mlxsw_i2c->block_size : in_mbox_size;
+		write_tran.len = MLXSW_I2C_ADDR_WIDTH + chunk_size;
+		mlxsw_i2c_set_slave_addr(tran_buf, off);
+		memcpy(&tran_buf[MLXSW_I2C_ADDR_BUF_SIZE], in_mbox +
+		       mlxsw_i2c->block_size * i, chunk_size);
+
+		j = 0;
+		end = jiffies + timeout;
+		do {
+			err = i2c_transfer(client->adapter, &write_tran, 1);
+			if (err == 1)
+				break;
+
+			cond_resched();
+		} while ((time_before(jiffies, end)) ||
+			 (j++ < MLXSW_I2C_RETRY));
+
+		if (err != 1) {
+			if (!err) {
+				err = -EIO;
+				goto mlxsw_i2c_write_exit;
+			}
+		}
+
+		off += chunk_size;
+		in_mbox_size -= chunk_size;
+	}
+
+	/* Prepare and write out Command Interface Register for transaction. */
+	err = mlxsw_i2c_write_cmd(client, mlxsw_i2c, 0);
+	if (err) {
+		dev_err(&client->dev, "Could not start transaction");
+		err = -EIO;
+		goto mlxsw_i2c_write_exit;
+	}
+
+	/* Wait until go bit is cleared. */
+	err = mlxsw_i2c_wait_go_bit(client, mlxsw_i2c, p_status);
+	if (err) {
+		dev_err(&client->dev, "HW semaphore is not released");
+		goto mlxsw_i2c_write_exit;
+	}
+
+	/* Validate transaction completion status. */
+	if (*p_status) {
+		dev_err(&client->dev, "Bad transaction completion status %x\n",
+			*p_status);
+		err = -EIO;
+	}
+
+mlxsw_i2c_write_exit:
+	kfree(tran_buf);
+	return err;
+}
+
+/* Routine executes I2C command. */
+static int
+mlxsw_i2c_cmd(struct device *dev, u16 opcode, u32 in_mod, size_t in_mbox_size,
+	      u8 *in_mbox, size_t out_mbox_size, u8 *out_mbox, u8 *status)
+{
+	struct i2c_client *client = to_i2c_client(dev);
+	struct mlxsw_i2c *mlxsw_i2c = i2c_get_clientdata(client);
+	unsigned long timeout = msecs_to_jiffies(MLXSW_I2C_TIMEOUT_MSECS);
+	u8 tran_buf[MLXSW_I2C_ADDR_BUF_SIZE];
+	int num, chunk_size, reg_size, i, j;
+	int off = mlxsw_i2c->cmd.mb_off_out;
+	unsigned long end;
+	struct i2c_msg read_tran[] =
+		MLXSW_I2C_READ_MSG(client, tran_buf, NULL, 0);
+	int err;
+
+	WARN_ON(in_mbox_size % sizeof(u32) || out_mbox_size % sizeof(u32));
+
+	if (in_mbox) {
+		reg_size = mlxsw_i2c_get_reg_size(in_mbox);
+		num = reg_size / mlxsw_i2c->block_size;
+		if (reg_size % mlxsw_i2c->block_size)
+			num++;
+
+		if (mutex_lock_interruptible(&mlxsw_i2c->cmd.lock) < 0) {
+			dev_err(&client->dev, "Could not acquire lock");
+			return -EINVAL;
+		}
+
+		err = mlxsw_i2c_write(dev, reg_size, in_mbox, num, status);
+		if (err)
+			goto cmd_fail;
+
+		/* No out mailbox is case of write transaction. */
+		if (!out_mbox) {
+			mutex_unlock(&mlxsw_i2c->cmd.lock);
+			return 0;
+		}
+	} else {
+		/* No input mailbox is case of initialization query command. */
+		reg_size = MLXSW_I2C_MAX_DATA_SIZE;
+		num = DIV_ROUND_UP(reg_size, mlxsw_i2c->block_size);
+
+		if (mutex_lock_interruptible(&mlxsw_i2c->cmd.lock) < 0) {
+			dev_err(&client->dev, "Could not acquire lock");
+			return -EINVAL;
+		}
+
+		err = mlxsw_i2c_write_init_cmd(client, mlxsw_i2c, opcode,
+					       in_mod);
+		if (err)
+			goto cmd_fail;
+	}
+
+	/* Send read transaction to get output mailbox content. */
+	read_tran[1].buf = out_mbox;
+	for (i = 0; i < num; i++) {
+		chunk_size = (reg_size > mlxsw_i2c->block_size) ?
+			     mlxsw_i2c->block_size : reg_size;
+		read_tran[1].len = chunk_size;
+		mlxsw_i2c_set_slave_addr(tran_buf, off);
+
+		j = 0;
+		end = jiffies + timeout;
+		do {
+			err = i2c_transfer(client->adapter, read_tran,
+					   ARRAY_SIZE(read_tran));
+			if (err == ARRAY_SIZE(read_tran))
+				break;
+
+			cond_resched();
+		} while ((time_before(jiffies, end)) ||
+			 (j++ < MLXSW_I2C_RETRY));
+
+		if (err != ARRAY_SIZE(read_tran)) {
+			if (!err)
+				err = -EIO;
+
+			goto cmd_fail;
+		}
+
+		off += chunk_size;
+		reg_size -= chunk_size;
+		read_tran[1].buf += chunk_size;
+	}
+
+	mutex_unlock(&mlxsw_i2c->cmd.lock);
+
+	return 0;
+
+cmd_fail:
+	mutex_unlock(&mlxsw_i2c->cmd.lock);
+	return err;
+}
+
+static int mlxsw_i2c_cmd_exec(void *bus_priv, u16 opcode, u8 opcode_mod,
+			      u32 in_mod, bool out_mbox_direct,
+			      char *in_mbox, size_t in_mbox_size,
+			      char *out_mbox, size_t out_mbox_size,
+			      u8 *status)
+{
+	struct mlxsw_i2c *mlxsw_i2c = bus_priv;
+
+	return mlxsw_i2c_cmd(mlxsw_i2c->dev, opcode, in_mod, in_mbox_size,
+			     in_mbox, out_mbox_size, out_mbox, status);
+}
+
+static bool mlxsw_i2c_skb_transmit_busy(void *bus_priv,
+					const struct mlxsw_tx_info *tx_info)
+{
+	return false;
+}
+
+static int mlxsw_i2c_skb_transmit(void *bus_priv, struct sk_buff *skb,
+				  const struct mlxsw_tx_info *tx_info)
+{
+	return 0;
+}
+
+static int
+mlxsw_i2c_init(void *bus_priv, struct mlxsw_core *mlxsw_core,
+	       const struct mlxsw_config_profile *profile,
+	       struct mlxsw_res *res)
+{
+	struct mlxsw_i2c *mlxsw_i2c = bus_priv;
+	char *mbox;
+	int err;
+
+	mlxsw_i2c->core = mlxsw_core;
+
+	mbox = mlxsw_cmd_mbox_alloc();
+	if (!mbox)
+		return -ENOMEM;
+
+	err = mlxsw_cmd_query_fw(mlxsw_core, mbox);
+	if (err)
+		goto mbox_put;
+
+	mlxsw_i2c->bus_info.fw_rev.major =
+		mlxsw_cmd_mbox_query_fw_fw_rev_major_get(mbox);
+	mlxsw_i2c->bus_info.fw_rev.minor =
+		mlxsw_cmd_mbox_query_fw_fw_rev_minor_get(mbox);
+	mlxsw_i2c->bus_info.fw_rev.subminor =
+		mlxsw_cmd_mbox_query_fw_fw_rev_subminor_get(mbox);
+
+	err = mlxsw_core_resources_query(mlxsw_core, mbox, res);
+
+mbox_put:
+	mlxsw_cmd_mbox_free(mbox);
+	return err;
+}
+
+static void mlxsw_i2c_fini(void *bus_priv)
+{
+	struct mlxsw_i2c *mlxsw_i2c = bus_priv;
+
+	mlxsw_i2c->core = NULL;
+}
+
+static void mlxsw_i2c_work_handler(struct work_struct *work)
+{
+	struct mlxsw_i2c *mlxsw_i2c;
+
+	mlxsw_i2c = container_of(work, struct mlxsw_i2c, irq_work);
+	mlxsw_core_irq_event_handlers_call(mlxsw_i2c->core);
+}
+
+static irqreturn_t mlxsw_i2c_irq_handler(int irq, void *dev)
+{
+	struct mlxsw_i2c *mlxsw_i2c = dev;
+
+	mlxsw_core_schedule_work(&mlxsw_i2c->irq_work);
+
+	/* Interrupt handler shares IRQ line with 'main' interrupt handler.
+	 * Return here IRQ_NONE, while main handler will return IRQ_HANDLED.
+	 */
+	return IRQ_NONE;
+}
+
+static int mlxsw_i2c_irq_init(struct mlxsw_i2c *mlxsw_i2c, u8 addr)
+{
+	int err;
+
+	/* Initialize interrupt handler if system hotplug driver is reachable,
+	 * otherwise interrupt line is not enabled and interrupts will not be
+	 * raised to CPU. Also request_irq() call will be not valid.
+	 */
+	if (!IS_REACHABLE(CONFIG_MLXREG_HOTPLUG))
+		return 0;
+
+	/* Set default interrupt line. */
+	if (mlxsw_i2c->pdata && mlxsw_i2c->pdata->irq)
+		mlxsw_i2c->irq = mlxsw_i2c->pdata->irq;
+	else if (addr == MLXSW_FAST_I2C_SLAVE)
+		mlxsw_i2c->irq = MLXSW_I2C_DEFAULT_IRQ;
+
+	if (!mlxsw_i2c->irq)
+		return 0;
+
+	INIT_WORK(&mlxsw_i2c->irq_work, mlxsw_i2c_work_handler);
+	err = request_irq(mlxsw_i2c->irq, mlxsw_i2c_irq_handler,
+			  IRQF_TRIGGER_FALLING | IRQF_SHARED, "mlxsw-i2c",
+			  mlxsw_i2c);
+	if (err) {
+		dev_err(mlxsw_i2c->bus_info.dev, "Failed to request irq: %d\n",
+			err);
+		return err;
+	}
+
+	return 0;
+}
+
+static void mlxsw_i2c_irq_fini(struct mlxsw_i2c *mlxsw_i2c)
+{
+	if (!IS_REACHABLE(CONFIG_MLXREG_HOTPLUG) || !mlxsw_i2c->irq)
+		return;
+	cancel_work_sync(&mlxsw_i2c->irq_work);
+	free_irq(mlxsw_i2c->irq, mlxsw_i2c);
+}
+
+static const struct mlxsw_bus mlxsw_i2c_bus = {
+	.kind			= "i2c",
+	.init			= mlxsw_i2c_init,
+	.fini			= mlxsw_i2c_fini,
+	.skb_transmit_busy	= mlxsw_i2c_skb_transmit_busy,
+	.skb_transmit		= mlxsw_i2c_skb_transmit,
+	.cmd_exec		= mlxsw_i2c_cmd_exec,
+};
+
+static int mlxsw_i2c_probe(struct i2c_client *client,
+			   const struct i2c_device_id *id)
+{
+	const struct i2c_adapter_quirks *quirks = client->adapter->quirks;
+	struct mlxsw_i2c *mlxsw_i2c;
+	u8 status;
+	int err;
+
+	mlxsw_i2c = devm_kzalloc(&client->dev, sizeof(*mlxsw_i2c), GFP_KERNEL);
+	if (!mlxsw_i2c)
+		return -ENOMEM;
+
+	if (quirks) {
+		if ((quirks->max_read_len &&
+		     quirks->max_read_len < MLXSW_I2C_BLK_DEF) ||
+		    (quirks->max_write_len &&
+		     quirks->max_write_len < MLXSW_I2C_BLK_DEF)) {
+			dev_err(&client->dev, "Insufficient transaction buffer length\n");
+			return -EOPNOTSUPP;
+		}
+
+		mlxsw_i2c->block_size = min_t(u16, MLXSW_I2C_BLK_MAX,
+					      min_t(u16, quirks->max_read_len,
+						    quirks->max_write_len));
+	} else {
+		mlxsw_i2c->block_size = MLXSW_I2C_BLK_DEF;
+	}
+
+	i2c_set_clientdata(client, mlxsw_i2c);
+	mutex_init(&mlxsw_i2c->cmd.lock);
+
+	/* In order to use mailboxes through the i2c, special area is reserved
+	 * on the i2c address space that can be used for input and output
+	 * mailboxes. Such mailboxes are called local mailboxes. When using a
+	 * local mailbox, software should specify 0 as the Input/Output
+	 * parameters. The location of the Local Mailbox addresses on the i2c
+	 * space can be retrieved through the QUERY_FW command.
+	 * For this purpose QUERY_FW is to be issued with opcode modifier equal
+	 * 0x01. For such command the output parameter is an immediate value.
+	 * Here QUERY_FW command is invoked for ASIC probing and for getting
+	 * local mailboxes addresses from immedate output parameters.
+	 */
+
+	/* Prepare and write out Command Interface Register for transaction */
+	err = mlxsw_i2c_write_cmd(client, mlxsw_i2c, 1);
+	if (err) {
+		dev_err(&client->dev, "Could not start transaction");
+		goto errout;
+	}
+
+	/* Wait until go bit is cleared. */
+	err = mlxsw_i2c_wait_go_bit(client, mlxsw_i2c, &status);
+	if (err) {
+		dev_err(&client->dev, "HW semaphore is not released");
+		goto errout;
+	}
+
+	/* Validate transaction completion status. */
+	if (status) {
+		dev_err(&client->dev, "Bad transaction completion status %x\n",
+			status);
+		err = -EIO;
+		goto errout;
+	}
+
+	/* Get mailbox offsets. */
+	err = mlxsw_i2c_get_mbox(client, mlxsw_i2c);
+	if (err < 0) {
+		dev_err(&client->dev, "Fail to get mailboxes\n");
+		goto errout;
+	}
+
+	dev_info(&client->dev, "%s mb size=%x off=0x%08x out mb size=%x off=0x%08x\n",
+		 id->name, mlxsw_i2c->cmd.mb_size_in,
+		 mlxsw_i2c->cmd.mb_off_in, mlxsw_i2c->cmd.mb_size_out,
+		 mlxsw_i2c->cmd.mb_off_out);
+
+	/* Register device bus. */
+	mlxsw_i2c->bus_info.device_kind = id->name;
+	mlxsw_i2c->bus_info.device_name = client->name;
+	mlxsw_i2c->bus_info.dev = &client->dev;
+	mlxsw_i2c->bus_info.low_frequency = true;
+	mlxsw_i2c->dev = &client->dev;
+	mlxsw_i2c->pdata = client->dev.platform_data;
+
+	err = mlxsw_i2c_irq_init(mlxsw_i2c, client->addr);
+	if (err)
+		goto errout;
+
+	err = mlxsw_core_bus_device_register(&mlxsw_i2c->bus_info,
+					     &mlxsw_i2c_bus, mlxsw_i2c, false,
+					     NULL, NULL);
+	if (err) {
+		dev_err(&client->dev, "Fail to register core bus\n");
+		goto err_bus_device_register;
+	}
+
+	return 0;
+
+err_bus_device_register:
+	mlxsw_i2c_irq_fini(mlxsw_i2c);
+errout:
+	mutex_destroy(&mlxsw_i2c->cmd.lock);
+	i2c_set_clientdata(client, NULL);
+
+	return err;
+}
+
+static void mlxsw_i2c_remove(struct i2c_client *client)
+{
+	struct mlxsw_i2c *mlxsw_i2c = i2c_get_clientdata(client);
+
+	mlxsw_core_bus_device_unregister(mlxsw_i2c->core, false);
+	mlxsw_i2c_irq_fini(mlxsw_i2c);
+	mutex_destroy(&mlxsw_i2c->cmd.lock);
+}
+
+int mlxsw_i2c_driver_register(struct i2c_driver *i2c_driver)
+{
+	i2c_driver->probe = mlxsw_i2c_probe;
+	i2c_driver->remove = mlxsw_i2c_remove;
+	return i2c_add_driver(i2c_driver);
+}
+EXPORT_SYMBOL(mlxsw_i2c_driver_register);
+
+void mlxsw_i2c_driver_unregister(struct i2c_driver *i2c_driver)
+{
+	i2c_del_driver(i2c_driver);
+}
+EXPORT_SYMBOL(mlxsw_i2c_driver_unregister);
+
+MODULE_AUTHOR("Vadim Pasternak <vadimp@mellanox.com>");
+MODULE_DESCRIPTION("Mellanox switch I2C interface driver");
+MODULE_LICENSE("Dual BSD/GPL");
diff --git a/drivers/net/ethernet/mellanox/mlxsw/i2c.h b/drivers/net/ethernet/mellanox/mlxsw/i2c.h
new file mode 100644
index 000000000..17e059d47
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/i2c.h
@@ -0,0 +1,29 @@
+/* SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0 */
+/* Copyright (c) 2016-2018 Mellanox Technologies. All rights reserved */
+
+#ifndef _MLXSW_I2C_H
+#define _MLXSW_I2C_H
+
+#include <linux/i2c.h>
+
+#if IS_ENABLED(CONFIG_MLXSW_I2C)
+
+int mlxsw_i2c_driver_register(struct i2c_driver *i2c_driver);
+void mlxsw_i2c_driver_unregister(struct i2c_driver *i2c_driver);
+
+#else
+
+static inline int
+mlxsw_i2c_driver_register(struct i2c_driver *i2c_driver)
+{
+	return -ENODEV;
+}
+
+static inline void
+mlxsw_i2c_driver_unregister(struct i2c_driver *i2c_driver)
+{
+}
+
+#endif
+
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlxsw/item.h b/drivers/net/ethernet/mellanox/mlxsw/item.h
new file mode 100644
index 000000000..cfafbeb42
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/item.h
@@ -0,0 +1,553 @@
+/* SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0 */
+/* Copyright (c) 2015-2018 Mellanox Technologies. All rights reserved */
+
+#ifndef _MLXSW_ITEM_H
+#define _MLXSW_ITEM_H
+
+#include <linux/types.h>
+#include <linux/string.h>
+#include <linux/bitops.h>
+
+struct mlxsw_item {
+	unsigned short	offset;		/* bytes in container */
+	short		step;		/* step in bytes for indexed items */
+	unsigned short	in_step_offset; /* offset within one step */
+	unsigned char	shift;		/* shift in bits */
+	unsigned char	element_size;	/* size of element in bit array */
+	bool		no_real_shift;
+	union {
+		unsigned char	bits;
+		unsigned short	bytes;
+	} size;
+	const char	*name;
+};
+
+static inline unsigned int
+__mlxsw_item_offset(const struct mlxsw_item *item, unsigned short index,
+		    size_t typesize)
+{
+	BUG_ON(index && !item->step);
+	if (item->offset % typesize != 0 ||
+	    item->step % typesize != 0 ||
+	    item->in_step_offset % typesize != 0) {
+		pr_err("mlxsw: item bug (name=%s,offset=%x,step=%x,in_step_offset=%x,typesize=%zx)\n",
+		       item->name, item->offset, item->step,
+		       item->in_step_offset, typesize);
+		BUG();
+	}
+
+	return ((item->offset + item->step * index + item->in_step_offset) /
+		typesize);
+}
+
+static inline u8 __mlxsw_item_get8(const char *buf,
+				   const struct mlxsw_item *item,
+				   unsigned short index)
+{
+	unsigned int offset = __mlxsw_item_offset(item, index, sizeof(u8));
+	u8 *b = (u8 *) buf;
+	u8 tmp;
+
+	tmp = b[offset];
+	tmp >>= item->shift;
+	tmp &= GENMASK(item->size.bits - 1, 0);
+	if (item->no_real_shift)
+		tmp <<= item->shift;
+	return tmp;
+}
+
+static inline void __mlxsw_item_set8(char *buf, const struct mlxsw_item *item,
+				     unsigned short index, u8 val)
+{
+	unsigned int offset = __mlxsw_item_offset(item, index,
+						  sizeof(u8));
+	u8 *b = (u8 *) buf;
+	u8 mask = GENMASK(item->size.bits - 1, 0) << item->shift;
+	u8 tmp;
+
+	if (!item->no_real_shift)
+		val <<= item->shift;
+	val &= mask;
+	tmp = b[offset];
+	tmp &= ~mask;
+	tmp |= val;
+	b[offset] = tmp;
+}
+
+static inline u16 __mlxsw_item_get16(const char *buf,
+				     const struct mlxsw_item *item,
+				     unsigned short index)
+{
+	unsigned int offset = __mlxsw_item_offset(item, index, sizeof(u16));
+	__be16 *b = (__be16 *) buf;
+	u16 tmp;
+
+	tmp = be16_to_cpu(b[offset]);
+	tmp >>= item->shift;
+	tmp &= GENMASK(item->size.bits - 1, 0);
+	if (item->no_real_shift)
+		tmp <<= item->shift;
+	return tmp;
+}
+
+static inline void __mlxsw_item_set16(char *buf, const struct mlxsw_item *item,
+				      unsigned short index, u16 val)
+{
+	unsigned int offset = __mlxsw_item_offset(item, index,
+						  sizeof(u16));
+	__be16 *b = (__be16 *) buf;
+	u16 mask = GENMASK(item->size.bits - 1, 0) << item->shift;
+	u16 tmp;
+
+	if (!item->no_real_shift)
+		val <<= item->shift;
+	val &= mask;
+	tmp = be16_to_cpu(b[offset]);
+	tmp &= ~mask;
+	tmp |= val;
+	b[offset] = cpu_to_be16(tmp);
+}
+
+static inline u32 __mlxsw_item_get32(const char *buf,
+				     const struct mlxsw_item *item,
+				     unsigned short index)
+{
+	unsigned int offset = __mlxsw_item_offset(item, index, sizeof(u32));
+	__be32 *b = (__be32 *) buf;
+	u32 tmp;
+
+	tmp = be32_to_cpu(b[offset]);
+	tmp >>= item->shift;
+	tmp &= GENMASK(item->size.bits - 1, 0);
+	if (item->no_real_shift)
+		tmp <<= item->shift;
+	return tmp;
+}
+
+static inline void __mlxsw_item_set32(char *buf, const struct mlxsw_item *item,
+				      unsigned short index, u32 val)
+{
+	unsigned int offset = __mlxsw_item_offset(item, index,
+						  sizeof(u32));
+	__be32 *b = (__be32 *) buf;
+	u32 mask = GENMASK(item->size.bits - 1, 0) << item->shift;
+	u32 tmp;
+
+	if (!item->no_real_shift)
+		val <<= item->shift;
+	val &= mask;
+	tmp = be32_to_cpu(b[offset]);
+	tmp &= ~mask;
+	tmp |= val;
+	b[offset] = cpu_to_be32(tmp);
+}
+
+static inline u64 __mlxsw_item_get64(const char *buf,
+				     const struct mlxsw_item *item,
+				     unsigned short index)
+{
+	unsigned int offset = __mlxsw_item_offset(item, index, sizeof(u64));
+	__be64 *b = (__be64 *) buf;
+	u64 tmp;
+
+	tmp = be64_to_cpu(b[offset]);
+	tmp >>= item->shift;
+	tmp &= GENMASK_ULL(item->size.bits - 1, 0);
+	if (item->no_real_shift)
+		tmp <<= item->shift;
+	return tmp;
+}
+
+static inline void __mlxsw_item_set64(char *buf, const struct mlxsw_item *item,
+				      unsigned short index, u64 val)
+{
+	unsigned int offset = __mlxsw_item_offset(item, index, sizeof(u64));
+	__be64 *b = (__be64 *) buf;
+	u64 mask = GENMASK_ULL(item->size.bits - 1, 0) << item->shift;
+	u64 tmp;
+
+	if (!item->no_real_shift)
+		val <<= item->shift;
+	val &= mask;
+	tmp = be64_to_cpu(b[offset]);
+	tmp &= ~mask;
+	tmp |= val;
+	b[offset] = cpu_to_be64(tmp);
+}
+
+static inline void __mlxsw_item_memcpy_from(const char *buf, char *dst,
+					    const struct mlxsw_item *item,
+					    unsigned short index)
+{
+	unsigned int offset = __mlxsw_item_offset(item, index, sizeof(char));
+
+	memcpy(dst, &buf[offset], item->size.bytes);
+}
+
+static inline void __mlxsw_item_memcpy_to(char *buf, const char *src,
+					  const struct mlxsw_item *item,
+					  unsigned short index)
+{
+	unsigned int offset = __mlxsw_item_offset(item, index, sizeof(char));
+
+	memcpy(&buf[offset], src, item->size.bytes);
+}
+
+static inline char *__mlxsw_item_data(char *buf, const struct mlxsw_item *item,
+				      unsigned short index)
+{
+	unsigned int offset = __mlxsw_item_offset(item, index, sizeof(char));
+
+	return &buf[offset];
+}
+
+static inline u16
+__mlxsw_item_bit_array_offset(const struct mlxsw_item *item,
+			      u16 index, u8 *shift)
+{
+	u16 max_index, be_index;
+	u16 offset;		/* byte offset inside the array */
+	u8 in_byte_index;
+
+	BUG_ON(index && !item->element_size);
+	if (item->offset % sizeof(u32) != 0 ||
+	    BITS_PER_BYTE % item->element_size != 0) {
+		pr_err("mlxsw: item bug (name=%s,offset=%x,element_size=%x)\n",
+		       item->name, item->offset, item->element_size);
+		BUG();
+	}
+
+	max_index = (item->size.bytes << 3) / item->element_size - 1;
+	be_index = max_index - index;
+	offset = be_index * item->element_size >> 3;
+	in_byte_index  = index % (BITS_PER_BYTE / item->element_size);
+	*shift = in_byte_index * item->element_size;
+
+	return item->offset + offset;
+}
+
+static inline u8 __mlxsw_item_bit_array_get(const char *buf,
+					    const struct mlxsw_item *item,
+					    u16 index)
+{
+	u8 shift, tmp;
+	u16 offset = __mlxsw_item_bit_array_offset(item, index, &shift);
+
+	tmp = buf[offset];
+	tmp >>= shift;
+	tmp &= GENMASK(item->element_size - 1, 0);
+	return tmp;
+}
+
+static inline void __mlxsw_item_bit_array_set(char *buf,
+					      const struct mlxsw_item *item,
+					      u16 index, u8 val)
+{
+	u8 shift, tmp;
+	u16 offset = __mlxsw_item_bit_array_offset(item, index, &shift);
+	u8 mask = GENMASK(item->element_size - 1, 0) << shift;
+
+	val <<= shift;
+	val &= mask;
+	tmp = buf[offset];
+	tmp &= ~mask;
+	tmp |= val;
+	buf[offset] = tmp;
+}
+
+#define __ITEM_NAME(_type, _cname, _iname)					\
+	mlxsw_##_type##_##_cname##_##_iname##_item
+
+/* _type: cmd_mbox, reg, etc.
+ * _cname: containter name (e.g. command name, register name)
+ * _iname: item name within the container
+ */
+
+#define MLXSW_ITEM8(_type, _cname, _iname, _offset, _shift, _sizebits)		\
+static struct mlxsw_item __ITEM_NAME(_type, _cname, _iname) = {			\
+	.offset = _offset,							\
+	.shift = _shift,							\
+	.size = {.bits = _sizebits,},						\
+	.name = #_type "_" #_cname "_" #_iname,					\
+};										\
+static inline u8 __maybe_unused							\
+mlxsw_##_type##_##_cname##_##_iname##_get(const char *buf)			\
+{										\
+	return __mlxsw_item_get8(buf, &__ITEM_NAME(_type, _cname, _iname), 0);	\
+}										\
+static inline void __maybe_unused						\
+mlxsw_##_type##_##_cname##_##_iname##_set(char *buf, u8 val)			\
+{										\
+	__mlxsw_item_set8(buf, &__ITEM_NAME(_type, _cname, _iname), 0, val);	\
+}
+
+#define MLXSW_ITEM8_INDEXED(_type, _cname, _iname, _offset, _shift, _sizebits,	\
+			    _step, _instepoffset, _norealshift)			\
+static struct mlxsw_item __ITEM_NAME(_type, _cname, _iname) = {			\
+	.offset = _offset,							\
+	.step = _step,								\
+	.in_step_offset = _instepoffset,					\
+	.shift = _shift,							\
+	.no_real_shift = _norealshift,						\
+	.size = {.bits = _sizebits,},						\
+	.name = #_type "_" #_cname "_" #_iname,					\
+};										\
+static inline u8 __maybe_unused							\
+mlxsw_##_type##_##_cname##_##_iname##_get(const char *buf, unsigned short index)\
+{										\
+	return __mlxsw_item_get8(buf, &__ITEM_NAME(_type, _cname, _iname),	\
+				 index);					\
+}										\
+static inline void __maybe_unused						\
+mlxsw_##_type##_##_cname##_##_iname##_set(char *buf, unsigned short index,	\
+					  u8 val)				\
+{										\
+	__mlxsw_item_set8(buf, &__ITEM_NAME(_type, _cname, _iname),		\
+			  index, val);						\
+}
+
+#define MLXSW_ITEM16(_type, _cname, _iname, _offset, _shift, _sizebits)		\
+static struct mlxsw_item __ITEM_NAME(_type, _cname, _iname) = {			\
+	.offset = _offset,							\
+	.shift = _shift,							\
+	.size = {.bits = _sizebits,},						\
+	.name = #_type "_" #_cname "_" #_iname,					\
+};										\
+static inline u16 __maybe_unused						\
+mlxsw_##_type##_##_cname##_##_iname##_get(const char *buf)			\
+{										\
+	return __mlxsw_item_get16(buf, &__ITEM_NAME(_type, _cname, _iname), 0);	\
+}										\
+static inline void __maybe_unused						\
+mlxsw_##_type##_##_cname##_##_iname##_set(char *buf, u16 val)			\
+{										\
+	__mlxsw_item_set16(buf, &__ITEM_NAME(_type, _cname, _iname), 0, val);	\
+}
+
+#define MLXSW_ITEM16_INDEXED(_type, _cname, _iname, _offset, _shift, _sizebits,	\
+			     _step, _instepoffset, _norealshift)		\
+static struct mlxsw_item __ITEM_NAME(_type, _cname, _iname) = {			\
+	.offset = _offset,							\
+	.step = _step,								\
+	.in_step_offset = _instepoffset,					\
+	.shift = _shift,							\
+	.no_real_shift = _norealshift,						\
+	.size = {.bits = _sizebits,},						\
+	.name = #_type "_" #_cname "_" #_iname,					\
+};										\
+static inline u16 __maybe_unused						\
+mlxsw_##_type##_##_cname##_##_iname##_get(const char *buf, unsigned short index)\
+{										\
+	return __mlxsw_item_get16(buf, &__ITEM_NAME(_type, _cname, _iname),	\
+				  index);					\
+}										\
+static inline void __maybe_unused						\
+mlxsw_##_type##_##_cname##_##_iname##_set(char *buf, unsigned short index,	\
+					  u16 val)				\
+{										\
+	__mlxsw_item_set16(buf, &__ITEM_NAME(_type, _cname, _iname),		\
+			   index, val);						\
+}
+
+#define MLXSW_ITEM32(_type, _cname, _iname, _offset, _shift, _sizebits)		\
+static struct mlxsw_item __ITEM_NAME(_type, _cname, _iname) = {			\
+	.offset = _offset,							\
+	.shift = _shift,							\
+	.size = {.bits = _sizebits,},						\
+	.name = #_type "_" #_cname "_" #_iname,					\
+};										\
+static inline u32 __maybe_unused						\
+mlxsw_##_type##_##_cname##_##_iname##_get(const char *buf)			\
+{										\
+	return __mlxsw_item_get32(buf, &__ITEM_NAME(_type, _cname, _iname), 0);	\
+}										\
+static inline void __maybe_unused						\
+mlxsw_##_type##_##_cname##_##_iname##_set(char *buf, u32 val)			\
+{										\
+	__mlxsw_item_set32(buf, &__ITEM_NAME(_type, _cname, _iname), 0, val);	\
+}
+
+#define LOCAL_PORT_LSB_SIZE 8
+#define LOCAL_PORT_MSB_SIZE 2
+
+#define MLXSW_ITEM32_LP(_type, _cname, _offset1, _shift1, _offset2, _shift2)	\
+static struct mlxsw_item __ITEM_NAME(_type, _cname, local_port) = {		\
+	.offset = _offset1,							\
+	.shift = _shift1,							\
+	.size = {.bits = LOCAL_PORT_LSB_SIZE,},					\
+	.name = #_type "_" #_cname "_local_port",				\
+};										\
+static struct mlxsw_item __ITEM_NAME(_type, _cname, lp_msb) = {			\
+	.offset = _offset2,							\
+	.shift = _shift2,							\
+	.size = {.bits = LOCAL_PORT_MSB_SIZE,},					\
+	.name = #_type "_" #_cname "_lp_msb",					\
+};										\
+static inline u32 __maybe_unused						\
+mlxsw_##_type##_##_cname##_local_port_get(const char *buf)			\
+{										\
+	u32 local_port, lp_msb;							\
+										\
+	local_port = __mlxsw_item_get32(buf, &__ITEM_NAME(_type, _cname,	\
+					local_port), 0);			\
+	lp_msb = __mlxsw_item_get32(buf, &__ITEM_NAME(_type, _cname, lp_msb),	\
+				   0);						\
+	return (lp_msb << LOCAL_PORT_LSB_SIZE) + local_port;			\
+}										\
+static inline void __maybe_unused						\
+mlxsw_##_type##_##_cname##_local_port_set(char *buf, u32 val)			\
+{										\
+	__mlxsw_item_set32(buf, &__ITEM_NAME(_type, _cname, local_port), 0,	\
+			   val & ((1 << LOCAL_PORT_LSB_SIZE) - 1));		\
+	__mlxsw_item_set32(buf, &__ITEM_NAME(_type, _cname, lp_msb), 0,		\
+			   val >> LOCAL_PORT_LSB_SIZE);				\
+}
+
+#define MLXSW_ITEM32_INDEXED(_type, _cname, _iname, _offset, _shift, _sizebits,	\
+			     _step, _instepoffset, _norealshift)		\
+static struct mlxsw_item __ITEM_NAME(_type, _cname, _iname) = {			\
+	.offset = _offset,							\
+	.step = _step,								\
+	.in_step_offset = _instepoffset,					\
+	.shift = _shift,							\
+	.no_real_shift = _norealshift,						\
+	.size = {.bits = _sizebits,},						\
+	.name = #_type "_" #_cname "_" #_iname,					\
+};										\
+static inline u32 __maybe_unused						\
+mlxsw_##_type##_##_cname##_##_iname##_get(const char *buf, unsigned short index)\
+{										\
+	return __mlxsw_item_get32(buf, &__ITEM_NAME(_type, _cname, _iname),	\
+				  index);					\
+}										\
+static inline void __maybe_unused						\
+mlxsw_##_type##_##_cname##_##_iname##_set(char *buf, unsigned short index,	\
+					  u32 val)				\
+{										\
+	__mlxsw_item_set32(buf, &__ITEM_NAME(_type, _cname, _iname),		\
+			   index, val);						\
+}
+
+#define MLXSW_ITEM64(_type, _cname, _iname, _offset, _shift, _sizebits)		\
+static struct mlxsw_item __ITEM_NAME(_type, _cname, _iname) = {			\
+	.offset = _offset,							\
+	.shift = _shift,							\
+	.size = {.bits = _sizebits,},						\
+	.name = #_type "_" #_cname "_" #_iname,					\
+};										\
+static inline u64 __maybe_unused						\
+mlxsw_##_type##_##_cname##_##_iname##_get(const char *buf)			\
+{										\
+	return __mlxsw_item_get64(buf, &__ITEM_NAME(_type, _cname, _iname), 0);	\
+}										\
+static inline void __maybe_unused						\
+mlxsw_##_type##_##_cname##_##_iname##_set(char *buf, u64 val)			\
+{										\
+	__mlxsw_item_set64(buf, &__ITEM_NAME(_type, _cname, _iname), 0,	val);	\
+}
+
+#define MLXSW_ITEM64_INDEXED(_type, _cname, _iname, _offset, _shift,		\
+			     _sizebits, _step, _instepoffset, _norealshift)	\
+static struct mlxsw_item __ITEM_NAME(_type, _cname, _iname) = {			\
+	.offset = _offset,							\
+	.step = _step,								\
+	.in_step_offset = _instepoffset,					\
+	.shift = _shift,							\
+	.no_real_shift = _norealshift,						\
+	.size = {.bits = _sizebits,},						\
+	.name = #_type "_" #_cname "_" #_iname,					\
+};										\
+static inline u64 __maybe_unused						\
+mlxsw_##_type##_##_cname##_##_iname##_get(const char *buf, unsigned short index)\
+{										\
+	return __mlxsw_item_get64(buf, &__ITEM_NAME(_type, _cname, _iname),	\
+				  index);					\
+}										\
+static inline void __maybe_unused						\
+mlxsw_##_type##_##_cname##_##_iname##_set(char *buf, unsigned short index,	\
+					  u64 val)				\
+{										\
+	__mlxsw_item_set64(buf, &__ITEM_NAME(_type, _cname, _iname),		\
+			   index, val);						\
+}
+
+#define MLXSW_ITEM_BUF(_type, _cname, _iname, _offset, _sizebytes)		\
+static struct mlxsw_item __ITEM_NAME(_type, _cname, _iname) = {			\
+	.offset = _offset,							\
+	.size = {.bytes = _sizebytes,},						\
+	.name = #_type "_" #_cname "_" #_iname,					\
+};										\
+static inline void __maybe_unused						\
+mlxsw_##_type##_##_cname##_##_iname##_memcpy_from(const char *buf, char *dst)	\
+{										\
+	__mlxsw_item_memcpy_from(buf, dst,					\
+				 &__ITEM_NAME(_type, _cname, _iname), 0);	\
+}										\
+static inline void __maybe_unused						\
+mlxsw_##_type##_##_cname##_##_iname##_memcpy_to(char *buf, const char *src)	\
+{										\
+	__mlxsw_item_memcpy_to(buf, src,					\
+			       &__ITEM_NAME(_type, _cname, _iname), 0);		\
+}										\
+static inline char * __maybe_unused						\
+mlxsw_##_type##_##_cname##_##_iname##_data(char *buf)				\
+{										\
+	return __mlxsw_item_data(buf, &__ITEM_NAME(_type, _cname, _iname), 0);	\
+}
+
+#define MLXSW_ITEM_BUF_INDEXED(_type, _cname, _iname, _offset, _sizebytes,	\
+			       _step, _instepoffset)				\
+static struct mlxsw_item __ITEM_NAME(_type, _cname, _iname) = {			\
+	.offset = _offset,							\
+	.step = _step,								\
+	.in_step_offset = _instepoffset,					\
+	.size = {.bytes = _sizebytes,},						\
+	.name = #_type "_" #_cname "_" #_iname,					\
+};										\
+static inline void __maybe_unused						\
+mlxsw_##_type##_##_cname##_##_iname##_memcpy_from(const char *buf,		\
+						  unsigned short index,		\
+						  char *dst)			\
+{										\
+	__mlxsw_item_memcpy_from(buf, dst,					\
+				 &__ITEM_NAME(_type, _cname, _iname), index);	\
+}										\
+static inline void __maybe_unused						\
+mlxsw_##_type##_##_cname##_##_iname##_memcpy_to(char *buf,			\
+						unsigned short index,		\
+						const char *src)		\
+{										\
+	__mlxsw_item_memcpy_to(buf, src,					\
+			       &__ITEM_NAME(_type, _cname, _iname), index);	\
+}										\
+static inline char * __maybe_unused						\
+mlxsw_##_type##_##_cname##_##_iname##_data(char *buf, unsigned short index)	\
+{										\
+	return __mlxsw_item_data(buf,						\
+				 &__ITEM_NAME(_type, _cname, _iname), index);	\
+}
+
+#define MLXSW_ITEM_BIT_ARRAY(_type, _cname, _iname, _offset, _sizebytes,	\
+			     _element_size)					\
+static struct mlxsw_item __ITEM_NAME(_type, _cname, _iname) = {			\
+	.offset = _offset,							\
+	.element_size = _element_size,						\
+	.size = {.bytes = _sizebytes,},						\
+	.name = #_type "_" #_cname "_" #_iname,					\
+};										\
+static inline u8 __maybe_unused							\
+mlxsw_##_type##_##_cname##_##_iname##_get(const char *buf, u16 index)		\
+{										\
+	return __mlxsw_item_bit_array_get(buf,					\
+					  &__ITEM_NAME(_type, _cname, _iname),	\
+					  index);				\
+}										\
+static inline void __maybe_unused						\
+mlxsw_##_type##_##_cname##_##_iname##_set(char *buf, u16 index, u8 val)		\
+{										\
+	return __mlxsw_item_bit_array_set(buf,					\
+					  &__ITEM_NAME(_type, _cname, _iname),	\
+					  index, val);				\
+}										\
+
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlxsw/minimal.c b/drivers/net/ethernet/mellanox/mlxsw/minimal.c
new file mode 100644
index 000000000..15116d930
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/minimal.c
@@ -0,0 +1,760 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2016-2019 Mellanox Technologies. All rights reserved */
+
+#include <linux/netdevice.h>
+#include <linux/etherdevice.h>
+#include <linux/ethtool.h>
+#include <linux/i2c.h>
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/mod_devicetable.h>
+#include <linux/types.h>
+
+#include "core.h"
+#include "core_env.h"
+#include "i2c.h"
+
+static const char mlxsw_m_driver_name[] = "mlxsw_minimal";
+
+#define MLXSW_M_FWREV_MINOR	2000
+#define MLXSW_M_FWREV_SUBMINOR	1886
+
+static const struct mlxsw_fw_rev mlxsw_m_fw_rev = {
+	.minor = MLXSW_M_FWREV_MINOR,
+	.subminor = MLXSW_M_FWREV_SUBMINOR,
+};
+
+struct mlxsw_m_port;
+
+struct mlxsw_m_line_card {
+	bool active;
+	int module_to_port[];
+};
+
+struct mlxsw_m {
+	struct mlxsw_m_port **ports;
+	struct mlxsw_core *core;
+	const struct mlxsw_bus_info *bus_info;
+	u8 base_mac[ETH_ALEN];
+	u8 max_ports;
+	u8 max_modules_per_slot; /* Maximum number of modules per-slot. */
+	u8 num_of_slots; /* Including the main board. */
+	struct mlxsw_m_line_card **line_cards;
+};
+
+struct mlxsw_m_port {
+	struct net_device *dev;
+	struct mlxsw_m *mlxsw_m;
+	u16 local_port;
+	u8 slot_index;
+	u8 module;
+	u8 module_offset;
+};
+
+static int mlxsw_m_base_mac_get(struct mlxsw_m *mlxsw_m)
+{
+	char spad_pl[MLXSW_REG_SPAD_LEN] = {0};
+	int err;
+
+	err = mlxsw_reg_query(mlxsw_m->core, MLXSW_REG(spad), spad_pl);
+	if (err)
+		return err;
+	mlxsw_reg_spad_base_mac_memcpy_from(spad_pl, mlxsw_m->base_mac);
+	return 0;
+}
+
+static int mlxsw_m_port_open(struct net_device *dev)
+{
+	struct mlxsw_m_port *mlxsw_m_port = netdev_priv(dev);
+	struct mlxsw_m *mlxsw_m = mlxsw_m_port->mlxsw_m;
+
+	return mlxsw_env_module_port_up(mlxsw_m->core, 0,
+					mlxsw_m_port->module);
+}
+
+static int mlxsw_m_port_stop(struct net_device *dev)
+{
+	struct mlxsw_m_port *mlxsw_m_port = netdev_priv(dev);
+	struct mlxsw_m *mlxsw_m = mlxsw_m_port->mlxsw_m;
+
+	mlxsw_env_module_port_down(mlxsw_m->core, 0, mlxsw_m_port->module);
+	return 0;
+}
+
+static struct devlink_port *
+mlxsw_m_port_get_devlink_port(struct net_device *dev)
+{
+	struct mlxsw_m_port *mlxsw_m_port = netdev_priv(dev);
+	struct mlxsw_m *mlxsw_m = mlxsw_m_port->mlxsw_m;
+
+	return mlxsw_core_port_devlink_port_get(mlxsw_m->core,
+						mlxsw_m_port->local_port);
+}
+
+static const struct net_device_ops mlxsw_m_port_netdev_ops = {
+	.ndo_open		= mlxsw_m_port_open,
+	.ndo_stop		= mlxsw_m_port_stop,
+	.ndo_get_devlink_port	= mlxsw_m_port_get_devlink_port,
+};
+
+static void mlxsw_m_module_get_drvinfo(struct net_device *dev,
+				       struct ethtool_drvinfo *drvinfo)
+{
+	struct mlxsw_m_port *mlxsw_m_port = netdev_priv(dev);
+	struct mlxsw_m *mlxsw_m = mlxsw_m_port->mlxsw_m;
+
+	strscpy(drvinfo->driver, mlxsw_m->bus_info->device_kind,
+		sizeof(drvinfo->driver));
+	snprintf(drvinfo->fw_version, sizeof(drvinfo->fw_version),
+		 "%d.%d.%d",
+		 mlxsw_m->bus_info->fw_rev.major,
+		 mlxsw_m->bus_info->fw_rev.minor,
+		 mlxsw_m->bus_info->fw_rev.subminor);
+	strscpy(drvinfo->bus_info, mlxsw_m->bus_info->device_name,
+		sizeof(drvinfo->bus_info));
+}
+
+static int mlxsw_m_get_module_info(struct net_device *netdev,
+				   struct ethtool_modinfo *modinfo)
+{
+	struct mlxsw_m_port *mlxsw_m_port = netdev_priv(netdev);
+	struct mlxsw_core *core = mlxsw_m_port->mlxsw_m->core;
+
+	return mlxsw_env_get_module_info(netdev, core,
+					 mlxsw_m_port->slot_index,
+					 mlxsw_m_port->module, modinfo);
+}
+
+static int
+mlxsw_m_get_module_eeprom(struct net_device *netdev, struct ethtool_eeprom *ee,
+			  u8 *data)
+{
+	struct mlxsw_m_port *mlxsw_m_port = netdev_priv(netdev);
+	struct mlxsw_core *core = mlxsw_m_port->mlxsw_m->core;
+
+	return mlxsw_env_get_module_eeprom(netdev, core,
+					   mlxsw_m_port->slot_index,
+					   mlxsw_m_port->module, ee, data);
+}
+
+static int
+mlxsw_m_get_module_eeprom_by_page(struct net_device *netdev,
+				  const struct ethtool_module_eeprom *page,
+				  struct netlink_ext_ack *extack)
+{
+	struct mlxsw_m_port *mlxsw_m_port = netdev_priv(netdev);
+	struct mlxsw_core *core = mlxsw_m_port->mlxsw_m->core;
+
+	return mlxsw_env_get_module_eeprom_by_page(core,
+						   mlxsw_m_port->slot_index,
+						   mlxsw_m_port->module,
+						   page, extack);
+}
+
+static int mlxsw_m_reset(struct net_device *netdev, u32 *flags)
+{
+	struct mlxsw_m_port *mlxsw_m_port = netdev_priv(netdev);
+	struct mlxsw_core *core = mlxsw_m_port->mlxsw_m->core;
+
+	return mlxsw_env_reset_module(netdev, core, mlxsw_m_port->slot_index,
+				      mlxsw_m_port->module,
+				      flags);
+}
+
+static int
+mlxsw_m_get_module_power_mode(struct net_device *netdev,
+			      struct ethtool_module_power_mode_params *params,
+			      struct netlink_ext_ack *extack)
+{
+	struct mlxsw_m_port *mlxsw_m_port = netdev_priv(netdev);
+	struct mlxsw_core *core = mlxsw_m_port->mlxsw_m->core;
+
+	return mlxsw_env_get_module_power_mode(core, mlxsw_m_port->slot_index,
+					       mlxsw_m_port->module,
+					       params, extack);
+}
+
+static int
+mlxsw_m_set_module_power_mode(struct net_device *netdev,
+			      const struct ethtool_module_power_mode_params *params,
+			      struct netlink_ext_ack *extack)
+{
+	struct mlxsw_m_port *mlxsw_m_port = netdev_priv(netdev);
+	struct mlxsw_core *core = mlxsw_m_port->mlxsw_m->core;
+
+	return mlxsw_env_set_module_power_mode(core, mlxsw_m_port->slot_index,
+					       mlxsw_m_port->module,
+					       params->policy, extack);
+}
+
+static const struct ethtool_ops mlxsw_m_port_ethtool_ops = {
+	.get_drvinfo		= mlxsw_m_module_get_drvinfo,
+	.get_module_info	= mlxsw_m_get_module_info,
+	.get_module_eeprom	= mlxsw_m_get_module_eeprom,
+	.get_module_eeprom_by_page = mlxsw_m_get_module_eeprom_by_page,
+	.reset			= mlxsw_m_reset,
+	.get_module_power_mode	= mlxsw_m_get_module_power_mode,
+	.set_module_power_mode	= mlxsw_m_set_module_power_mode,
+};
+
+static int
+mlxsw_m_port_module_info_get(struct mlxsw_m *mlxsw_m, u16 local_port,
+			     u8 *p_module, u8 *p_width, u8 *p_slot_index)
+{
+	char pmlp_pl[MLXSW_REG_PMLP_LEN];
+	int err;
+
+	mlxsw_reg_pmlp_pack(pmlp_pl, local_port);
+	err = mlxsw_reg_query(mlxsw_m->core, MLXSW_REG(pmlp), pmlp_pl);
+	if (err)
+		return err;
+	*p_module = mlxsw_reg_pmlp_module_get(pmlp_pl, 0);
+	*p_width = mlxsw_reg_pmlp_width_get(pmlp_pl);
+	*p_slot_index = mlxsw_reg_pmlp_slot_index_get(pmlp_pl, 0);
+
+	return 0;
+}
+
+static int
+mlxsw_m_port_dev_addr_get(struct mlxsw_m_port *mlxsw_m_port)
+{
+	struct mlxsw_m *mlxsw_m = mlxsw_m_port->mlxsw_m;
+	char ppad_pl[MLXSW_REG_PPAD_LEN];
+	u8 addr[ETH_ALEN];
+	int err;
+
+	mlxsw_reg_ppad_pack(ppad_pl, false, 0);
+	err = mlxsw_reg_query(mlxsw_m->core, MLXSW_REG(ppad), ppad_pl);
+	if (err)
+		return err;
+	mlxsw_reg_ppad_mac_memcpy_from(ppad_pl, addr);
+	eth_hw_addr_gen(mlxsw_m_port->dev, addr, mlxsw_m_port->module + 1 +
+			mlxsw_m_port->module_offset);
+	return 0;
+}
+
+static bool mlxsw_m_port_created(struct mlxsw_m *mlxsw_m, u16 local_port)
+{
+	return mlxsw_m->ports[local_port];
+}
+
+static int
+mlxsw_m_port_create(struct mlxsw_m *mlxsw_m, u16 local_port, u8 slot_index,
+		    u8 module)
+{
+	struct mlxsw_m_port *mlxsw_m_port;
+	struct net_device *dev;
+	int err;
+
+	err = mlxsw_core_port_init(mlxsw_m->core, local_port, slot_index,
+				   module + 1, false, 0, false,
+				   0, mlxsw_m->base_mac,
+				   sizeof(mlxsw_m->base_mac));
+	if (err) {
+		dev_err(mlxsw_m->bus_info->dev, "Port %d: Failed to init core port\n",
+			local_port);
+		return err;
+	}
+
+	dev = alloc_etherdev(sizeof(struct mlxsw_m_port));
+	if (!dev) {
+		err = -ENOMEM;
+		goto err_alloc_etherdev;
+	}
+
+	SET_NETDEV_DEV(dev, mlxsw_m->bus_info->dev);
+	dev_net_set(dev, mlxsw_core_net(mlxsw_m->core));
+	mlxsw_m_port = netdev_priv(dev);
+	mlxsw_m_port->dev = dev;
+	mlxsw_m_port->mlxsw_m = mlxsw_m;
+	mlxsw_m_port->local_port = local_port;
+	mlxsw_m_port->module = module;
+	mlxsw_m_port->slot_index = slot_index;
+	/* Add module offset for line card. Offset for main board iz zero.
+	 * For line card in slot #n offset is calculated as (#n - 1)
+	 * multiplied by maximum modules number, which could be found on a line
+	 * card.
+	 */
+	mlxsw_m_port->module_offset = mlxsw_m_port->slot_index ?
+				      (mlxsw_m_port->slot_index - 1) *
+				      mlxsw_m->max_modules_per_slot : 0;
+
+	dev->netdev_ops = &mlxsw_m_port_netdev_ops;
+	dev->ethtool_ops = &mlxsw_m_port_ethtool_ops;
+
+	err = mlxsw_m_port_dev_addr_get(mlxsw_m_port);
+	if (err) {
+		dev_err(mlxsw_m->bus_info->dev, "Port %d: Unable to get port mac address\n",
+			mlxsw_m_port->local_port);
+		goto err_dev_addr_get;
+	}
+
+	netif_carrier_off(dev);
+	mlxsw_m->ports[local_port] = mlxsw_m_port;
+	err = register_netdev(dev);
+	if (err) {
+		dev_err(mlxsw_m->bus_info->dev, "Port %d: Failed to register netdev\n",
+			mlxsw_m_port->local_port);
+		goto err_register_netdev;
+	}
+
+	mlxsw_core_port_eth_set(mlxsw_m->core, mlxsw_m_port->local_port,
+				mlxsw_m_port, dev);
+
+	return 0;
+
+err_register_netdev:
+	mlxsw_m->ports[local_port] = NULL;
+err_dev_addr_get:
+	free_netdev(dev);
+err_alloc_etherdev:
+	mlxsw_core_port_fini(mlxsw_m->core, local_port);
+	return err;
+}
+
+static void mlxsw_m_port_remove(struct mlxsw_m *mlxsw_m, u16 local_port)
+{
+	struct mlxsw_m_port *mlxsw_m_port = mlxsw_m->ports[local_port];
+
+	mlxsw_core_port_clear(mlxsw_m->core, local_port, mlxsw_m);
+	unregister_netdev(mlxsw_m_port->dev); /* This calls ndo_stop */
+	mlxsw_m->ports[local_port] = NULL;
+	free_netdev(mlxsw_m_port->dev);
+	mlxsw_core_port_fini(mlxsw_m->core, local_port);
+}
+
+static int*
+mlxsw_m_port_mapping_get(struct mlxsw_m *mlxsw_m, u8 slot_index, u8 module)
+{
+	return &mlxsw_m->line_cards[slot_index]->module_to_port[module];
+}
+
+static int mlxsw_m_port_module_map(struct mlxsw_m *mlxsw_m, u16 local_port,
+				   u8 *last_module)
+{
+	unsigned int max_ports = mlxsw_core_max_ports(mlxsw_m->core);
+	u8 module, width, slot_index;
+	int *module_to_port;
+	int err;
+
+	/* Fill out to local port mapping array */
+	err = mlxsw_m_port_module_info_get(mlxsw_m, local_port, &module,
+					   &width, &slot_index);
+	if (err)
+		return err;
+
+	/* Skip if line card has been already configured */
+	if (mlxsw_m->line_cards[slot_index]->active)
+		return 0;
+	if (!width)
+		return 0;
+	/* Skip, if port belongs to the cluster */
+	if (module == *last_module)
+		return 0;
+	*last_module = module;
+
+	if (WARN_ON_ONCE(module >= max_ports))
+		return -EINVAL;
+	mlxsw_env_module_port_map(mlxsw_m->core, slot_index, module);
+	module_to_port = mlxsw_m_port_mapping_get(mlxsw_m, slot_index, module);
+	*module_to_port = local_port;
+
+	return 0;
+}
+
+static void
+mlxsw_m_port_module_unmap(struct mlxsw_m *mlxsw_m, u8 slot_index, u8 module)
+{
+	int *module_to_port = mlxsw_m_port_mapping_get(mlxsw_m, slot_index,
+						       module);
+	*module_to_port = -1;
+	mlxsw_env_module_port_unmap(mlxsw_m->core, slot_index, module);
+}
+
+static int mlxsw_m_linecards_init(struct mlxsw_m *mlxsw_m)
+{
+	unsigned int max_ports = mlxsw_core_max_ports(mlxsw_m->core);
+	char mgpir_pl[MLXSW_REG_MGPIR_LEN];
+	u8 num_of_modules;
+	int i, j, err;
+
+	mlxsw_reg_mgpir_pack(mgpir_pl, 0);
+	err = mlxsw_reg_query(mlxsw_m->core, MLXSW_REG(mgpir), mgpir_pl);
+	if (err)
+		return err;
+
+	mlxsw_reg_mgpir_unpack(mgpir_pl, NULL, NULL, NULL, &num_of_modules,
+			       &mlxsw_m->num_of_slots);
+	/* If the system is modular, get the maximum number of modules per-slot.
+	 * Otherwise, get the maximum number of modules on the main board.
+	 */
+	if (mlxsw_m->num_of_slots)
+		mlxsw_m->max_modules_per_slot =
+			mlxsw_reg_mgpir_max_modules_per_slot_get(mgpir_pl);
+	else
+		mlxsw_m->max_modules_per_slot = num_of_modules;
+	/* Add slot for main board. */
+	mlxsw_m->num_of_slots += 1;
+
+	mlxsw_m->ports = kcalloc(max_ports, sizeof(*mlxsw_m->ports),
+				 GFP_KERNEL);
+	if (!mlxsw_m->ports)
+		return -ENOMEM;
+
+	mlxsw_m->line_cards = kcalloc(mlxsw_m->num_of_slots,
+				      sizeof(*mlxsw_m->line_cards),
+				      GFP_KERNEL);
+	if (!mlxsw_m->line_cards) {
+		err = -ENOMEM;
+		goto err_kcalloc;
+	}
+
+	for (i = 0; i < mlxsw_m->num_of_slots; i++) {
+		mlxsw_m->line_cards[i] =
+			kzalloc(struct_size(mlxsw_m->line_cards[i],
+					    module_to_port,
+					    mlxsw_m->max_modules_per_slot),
+				GFP_KERNEL);
+		if (!mlxsw_m->line_cards[i]) {
+			err = -ENOMEM;
+			goto err_kmalloc_array;
+		}
+
+		/* Invalidate the entries of module to local port mapping array. */
+		for (j = 0; j < mlxsw_m->max_modules_per_slot; j++)
+			mlxsw_m->line_cards[i]->module_to_port[j] = -1;
+	}
+
+	return 0;
+
+err_kmalloc_array:
+	for (i--; i >= 0; i--)
+		kfree(mlxsw_m->line_cards[i]);
+	kfree(mlxsw_m->line_cards);
+err_kcalloc:
+	kfree(mlxsw_m->ports);
+	return err;
+}
+
+static void mlxsw_m_linecards_fini(struct mlxsw_m *mlxsw_m)
+{
+	int i = mlxsw_m->num_of_slots;
+
+	for (i--; i >= 0; i--)
+		kfree(mlxsw_m->line_cards[i]);
+	kfree(mlxsw_m->line_cards);
+	kfree(mlxsw_m->ports);
+}
+
+static void
+mlxsw_m_linecard_port_module_unmap(struct mlxsw_m *mlxsw_m, u8 slot_index)
+{
+	int i;
+
+	for (i = mlxsw_m->max_modules_per_slot - 1; i >= 0; i--) {
+		int *module_to_port;
+
+		module_to_port = mlxsw_m_port_mapping_get(mlxsw_m, slot_index, i);
+		if (*module_to_port > 0)
+			mlxsw_m_port_module_unmap(mlxsw_m, slot_index, i);
+	}
+}
+
+static int
+mlxsw_m_linecard_ports_create(struct mlxsw_m *mlxsw_m, u8 slot_index)
+{
+	int *module_to_port;
+	int i, err;
+
+	for (i = 0; i < mlxsw_m->max_modules_per_slot; i++) {
+		module_to_port = mlxsw_m_port_mapping_get(mlxsw_m, slot_index, i);
+		if (*module_to_port > 0) {
+			err = mlxsw_m_port_create(mlxsw_m, *module_to_port,
+						  slot_index, i);
+			if (err)
+				goto err_port_create;
+			/* Mark slot as active */
+			if (!mlxsw_m->line_cards[slot_index]->active)
+				mlxsw_m->line_cards[slot_index]->active = true;
+		}
+	}
+	return 0;
+
+err_port_create:
+	for (i--; i >= 0; i--) {
+		module_to_port = mlxsw_m_port_mapping_get(mlxsw_m, slot_index, i);
+		if (*module_to_port > 0 &&
+		    mlxsw_m_port_created(mlxsw_m, *module_to_port)) {
+			mlxsw_m_port_remove(mlxsw_m, *module_to_port);
+			/* Mark slot as inactive */
+			if (mlxsw_m->line_cards[slot_index]->active)
+				mlxsw_m->line_cards[slot_index]->active = false;
+		}
+	}
+	return err;
+}
+
+static void
+mlxsw_m_linecard_ports_remove(struct mlxsw_m *mlxsw_m, u8 slot_index)
+{
+	int i;
+
+	for (i = 0; i < mlxsw_m->max_modules_per_slot; i++) {
+		int *module_to_port = mlxsw_m_port_mapping_get(mlxsw_m,
+							       slot_index, i);
+
+		if (*module_to_port > 0 &&
+		    mlxsw_m_port_created(mlxsw_m, *module_to_port)) {
+			mlxsw_m_port_remove(mlxsw_m, *module_to_port);
+			mlxsw_m_port_module_unmap(mlxsw_m, slot_index, i);
+		}
+	}
+}
+
+static int mlxsw_m_ports_module_map(struct mlxsw_m *mlxsw_m)
+{
+	unsigned int max_ports = mlxsw_core_max_ports(mlxsw_m->core);
+	u8 last_module = max_ports;
+	int i, err;
+
+	for (i = 1; i < max_ports; i++) {
+		err = mlxsw_m_port_module_map(mlxsw_m, i, &last_module);
+		if (err)
+			return err;
+	}
+
+	return 0;
+}
+
+static int mlxsw_m_ports_create(struct mlxsw_m *mlxsw_m)
+{
+	int err;
+
+	/* Fill out module to local port mapping array */
+	err = mlxsw_m_ports_module_map(mlxsw_m);
+	if (err)
+		goto err_ports_module_map;
+
+	/* Create port objects for each valid entry */
+	err = mlxsw_m_linecard_ports_create(mlxsw_m, 0);
+	if (err)
+		goto err_linecard_ports_create;
+
+	return 0;
+
+err_linecard_ports_create:
+err_ports_module_map:
+	mlxsw_m_linecard_port_module_unmap(mlxsw_m, 0);
+
+	return err;
+}
+
+static void mlxsw_m_ports_remove(struct mlxsw_m *mlxsw_m)
+{
+	mlxsw_m_linecard_ports_remove(mlxsw_m, 0);
+}
+
+static void
+mlxsw_m_ports_remove_selected(struct mlxsw_core *mlxsw_core,
+			      bool (*selector)(void *priv, u16 local_port),
+			      void *priv)
+{
+	struct mlxsw_m *mlxsw_m = mlxsw_core_driver_priv(mlxsw_core);
+	struct mlxsw_linecard *linecard_priv = priv;
+	struct mlxsw_m_line_card *linecard;
+
+	linecard = mlxsw_m->line_cards[linecard_priv->slot_index];
+
+	if (WARN_ON(!linecard->active))
+		return;
+
+	mlxsw_m_linecard_ports_remove(mlxsw_m, linecard_priv->slot_index);
+	linecard->active = false;
+}
+
+static int mlxsw_m_fw_rev_validate(struct mlxsw_m *mlxsw_m)
+{
+	const struct mlxsw_fw_rev *rev = &mlxsw_m->bus_info->fw_rev;
+
+	/* Validate driver and FW are compatible.
+	 * Do not check major version, since it defines chip type, while
+	 * driver is supposed to support any type.
+	 */
+	if (mlxsw_core_fw_rev_minor_subminor_validate(rev, &mlxsw_m_fw_rev))
+		return 0;
+
+	dev_err(mlxsw_m->bus_info->dev, "The firmware version %d.%d.%d is incompatible with the driver (required >= %d.%d.%d)\n",
+		rev->major, rev->minor, rev->subminor, rev->major,
+		mlxsw_m_fw_rev.minor, mlxsw_m_fw_rev.subminor);
+
+	return -EINVAL;
+}
+
+static void
+mlxsw_m_got_active(struct mlxsw_core *mlxsw_core, u8 slot_index, void *priv)
+{
+	struct mlxsw_m_line_card *linecard;
+	struct mlxsw_m *mlxsw_m = priv;
+	int err;
+
+	linecard = mlxsw_m->line_cards[slot_index];
+	/* Skip if line card has been already configured during init */
+	if (linecard->active)
+		return;
+
+	/* Fill out module to local port mapping array */
+	err = mlxsw_m_ports_module_map(mlxsw_m);
+	if (err)
+		goto err_ports_module_map;
+
+	/* Create port objects for each valid entry */
+	err = mlxsw_m_linecard_ports_create(mlxsw_m, slot_index);
+	if (err) {
+		dev_err(mlxsw_m->bus_info->dev, "Failed to create port for line card at slot %d\n",
+			slot_index);
+		goto err_linecard_ports_create;
+	}
+
+	linecard->active = true;
+
+	return;
+
+err_linecard_ports_create:
+err_ports_module_map:
+	mlxsw_m_linecard_port_module_unmap(mlxsw_m, slot_index);
+}
+
+static void
+mlxsw_m_got_inactive(struct mlxsw_core *mlxsw_core, u8 slot_index, void *priv)
+{
+	struct mlxsw_m_line_card *linecard;
+	struct mlxsw_m *mlxsw_m = priv;
+
+	linecard = mlxsw_m->line_cards[slot_index];
+
+	if (WARN_ON(!linecard->active))
+		return;
+
+	mlxsw_m_linecard_ports_remove(mlxsw_m, slot_index);
+	linecard->active = false;
+}
+
+static struct mlxsw_linecards_event_ops mlxsw_m_event_ops = {
+	.got_active = mlxsw_m_got_active,
+	.got_inactive = mlxsw_m_got_inactive,
+};
+
+static int mlxsw_m_init(struct mlxsw_core *mlxsw_core,
+			const struct mlxsw_bus_info *mlxsw_bus_info,
+			struct netlink_ext_ack *extack)
+{
+	struct mlxsw_m *mlxsw_m = mlxsw_core_driver_priv(mlxsw_core);
+	int err;
+
+	mlxsw_m->core = mlxsw_core;
+	mlxsw_m->bus_info = mlxsw_bus_info;
+
+	err = mlxsw_m_fw_rev_validate(mlxsw_m);
+	if (err)
+		return err;
+
+	err = mlxsw_m_base_mac_get(mlxsw_m);
+	if (err) {
+		dev_err(mlxsw_m->bus_info->dev, "Failed to get base mac\n");
+		return err;
+	}
+
+	err = mlxsw_m_linecards_init(mlxsw_m);
+	if (err) {
+		dev_err(mlxsw_m->bus_info->dev, "Failed to create line cards\n");
+		return err;
+	}
+
+	err = mlxsw_linecards_event_ops_register(mlxsw_core,
+						 &mlxsw_m_event_ops, mlxsw_m);
+	if (err) {
+		dev_err(mlxsw_m->bus_info->dev, "Failed to register line cards operations\n");
+		goto linecards_event_ops_register;
+	}
+
+	err = mlxsw_m_ports_create(mlxsw_m);
+	if (err) {
+		dev_err(mlxsw_m->bus_info->dev, "Failed to create ports\n");
+		goto err_ports_create;
+	}
+
+	return 0;
+
+err_ports_create:
+	mlxsw_linecards_event_ops_unregister(mlxsw_core,
+					     &mlxsw_m_event_ops, mlxsw_m);
+linecards_event_ops_register:
+	mlxsw_m_linecards_fini(mlxsw_m);
+	return err;
+}
+
+static void mlxsw_m_fini(struct mlxsw_core *mlxsw_core)
+{
+	struct mlxsw_m *mlxsw_m = mlxsw_core_driver_priv(mlxsw_core);
+
+	mlxsw_m_ports_remove(mlxsw_m);
+	mlxsw_linecards_event_ops_unregister(mlxsw_core,
+					     &mlxsw_m_event_ops, mlxsw_m);
+	mlxsw_m_linecards_fini(mlxsw_m);
+}
+
+static const struct mlxsw_config_profile mlxsw_m_config_profile;
+
+static struct mlxsw_driver mlxsw_m_driver = {
+	.kind			= mlxsw_m_driver_name,
+	.priv_size		= sizeof(struct mlxsw_m),
+	.init			= mlxsw_m_init,
+	.fini			= mlxsw_m_fini,
+	.ports_remove_selected	= mlxsw_m_ports_remove_selected,
+	.profile		= &mlxsw_m_config_profile,
+};
+
+static const struct i2c_device_id mlxsw_m_i2c_id[] = {
+	{ "mlxsw_minimal", 0},
+	{ },
+};
+
+static struct i2c_driver mlxsw_m_i2c_driver = {
+	.driver.name = "mlxsw_minimal",
+	.class = I2C_CLASS_HWMON,
+	.id_table = mlxsw_m_i2c_id,
+};
+
+static int __init mlxsw_m_module_init(void)
+{
+	int err;
+
+	err = mlxsw_core_driver_register(&mlxsw_m_driver);
+	if (err)
+		return err;
+
+	err = mlxsw_i2c_driver_register(&mlxsw_m_i2c_driver);
+	if (err)
+		goto err_i2c_driver_register;
+
+	return 0;
+
+err_i2c_driver_register:
+	mlxsw_core_driver_unregister(&mlxsw_m_driver);
+
+	return err;
+}
+
+static void __exit mlxsw_m_module_exit(void)
+{
+	mlxsw_i2c_driver_unregister(&mlxsw_m_i2c_driver);
+	mlxsw_core_driver_unregister(&mlxsw_m_driver);
+}
+
+module_init(mlxsw_m_module_init);
+module_exit(mlxsw_m_module_exit);
+
+MODULE_LICENSE("Dual BSD/GPL");
+MODULE_AUTHOR("Vadim Pasternak <vadimp@mellanox.com>");
+MODULE_DESCRIPTION("Mellanox minimal driver");
+MODULE_DEVICE_TABLE(i2c, mlxsw_m_i2c_id);
diff --git a/drivers/net/ethernet/mellanox/mlxsw/pci.c b/drivers/net/ethernet/mellanox/mlxsw/pci.c
new file mode 100644
index 000000000..51eea1f05
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/pci.c
@@ -0,0 +1,2068 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2015-2018 Mellanox Technologies. All rights reserved */
+
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/export.h>
+#include <linux/err.h>
+#include <linux/device.h>
+#include <linux/pci.h>
+#include <linux/interrupt.h>
+#include <linux/wait.h>
+#include <linux/types.h>
+#include <linux/skbuff.h>
+#include <linux/if_vlan.h>
+#include <linux/log2.h>
+#include <linux/string.h>
+
+#include "pci_hw.h"
+#include "pci.h"
+#include "core.h"
+#include "cmd.h"
+#include "port.h"
+#include "resources.h"
+
+#define mlxsw_pci_write32(mlxsw_pci, reg, val) \
+	iowrite32be(val, (mlxsw_pci)->hw_addr + (MLXSW_PCI_ ## reg))
+#define mlxsw_pci_read32(mlxsw_pci, reg) \
+	ioread32be((mlxsw_pci)->hw_addr + (MLXSW_PCI_ ## reg))
+
+enum mlxsw_pci_queue_type {
+	MLXSW_PCI_QUEUE_TYPE_SDQ,
+	MLXSW_PCI_QUEUE_TYPE_RDQ,
+	MLXSW_PCI_QUEUE_TYPE_CQ,
+	MLXSW_PCI_QUEUE_TYPE_EQ,
+};
+
+#define MLXSW_PCI_QUEUE_TYPE_COUNT	4
+
+static const u16 mlxsw_pci_doorbell_type_offset[] = {
+	MLXSW_PCI_DOORBELL_SDQ_OFFSET,	/* for type MLXSW_PCI_QUEUE_TYPE_SDQ */
+	MLXSW_PCI_DOORBELL_RDQ_OFFSET,	/* for type MLXSW_PCI_QUEUE_TYPE_RDQ */
+	MLXSW_PCI_DOORBELL_CQ_OFFSET,	/* for type MLXSW_PCI_QUEUE_TYPE_CQ */
+	MLXSW_PCI_DOORBELL_EQ_OFFSET,	/* for type MLXSW_PCI_QUEUE_TYPE_EQ */
+};
+
+static const u16 mlxsw_pci_doorbell_arm_type_offset[] = {
+	0, /* unused */
+	0, /* unused */
+	MLXSW_PCI_DOORBELL_ARM_CQ_OFFSET, /* for type MLXSW_PCI_QUEUE_TYPE_CQ */
+	MLXSW_PCI_DOORBELL_ARM_EQ_OFFSET, /* for type MLXSW_PCI_QUEUE_TYPE_EQ */
+};
+
+struct mlxsw_pci_mem_item {
+	char *buf;
+	dma_addr_t mapaddr;
+	size_t size;
+};
+
+struct mlxsw_pci_queue_elem_info {
+	char *elem; /* pointer to actual dma mapped element mem chunk */
+	union {
+		struct {
+			struct sk_buff *skb;
+		} sdq;
+		struct {
+			struct sk_buff *skb;
+		} rdq;
+	} u;
+};
+
+struct mlxsw_pci_queue {
+	spinlock_t lock; /* for queue accesses */
+	struct mlxsw_pci_mem_item mem_item;
+	struct mlxsw_pci_queue_elem_info *elem_info;
+	u16 producer_counter;
+	u16 consumer_counter;
+	u16 count; /* number of elements in queue */
+	u8 num; /* queue number */
+	u8 elem_size; /* size of one element */
+	enum mlxsw_pci_queue_type type;
+	struct tasklet_struct tasklet; /* queue processing tasklet */
+	struct mlxsw_pci *pci;
+	union {
+		struct {
+			u32 comp_sdq_count;
+			u32 comp_rdq_count;
+			enum mlxsw_pci_cqe_v v;
+		} cq;
+		struct {
+			u32 ev_cmd_count;
+			u32 ev_comp_count;
+			u32 ev_other_count;
+		} eq;
+	} u;
+};
+
+struct mlxsw_pci_queue_type_group {
+	struct mlxsw_pci_queue *q;
+	u8 count; /* number of queues in group */
+};
+
+struct mlxsw_pci {
+	struct pci_dev *pdev;
+	u8 __iomem *hw_addr;
+	u64 free_running_clock_offset;
+	u64 utc_sec_offset;
+	u64 utc_nsec_offset;
+	struct mlxsw_pci_queue_type_group queues[MLXSW_PCI_QUEUE_TYPE_COUNT];
+	u32 doorbell_offset;
+	struct mlxsw_core *core;
+	struct {
+		struct mlxsw_pci_mem_item *items;
+		unsigned int count;
+	} fw_area;
+	struct {
+		struct mlxsw_pci_mem_item out_mbox;
+		struct mlxsw_pci_mem_item in_mbox;
+		struct mutex lock; /* Lock access to command registers */
+		bool nopoll;
+		wait_queue_head_t wait;
+		bool wait_done;
+		struct {
+			u8 status;
+			u64 out_param;
+		} comp;
+	} cmd;
+	struct mlxsw_bus_info bus_info;
+	const struct pci_device_id *id;
+	enum mlxsw_pci_cqe_v max_cqe_ver; /* Maximal supported CQE version */
+	u8 num_sdq_cqs; /* Number of CQs used for SDQs */
+};
+
+static void mlxsw_pci_queue_tasklet_schedule(struct mlxsw_pci_queue *q)
+{
+	tasklet_schedule(&q->tasklet);
+}
+
+static char *__mlxsw_pci_queue_elem_get(struct mlxsw_pci_queue *q,
+					size_t elem_size, int elem_index)
+{
+	return q->mem_item.buf + (elem_size * elem_index);
+}
+
+static struct mlxsw_pci_queue_elem_info *
+mlxsw_pci_queue_elem_info_get(struct mlxsw_pci_queue *q, int elem_index)
+{
+	return &q->elem_info[elem_index];
+}
+
+static struct mlxsw_pci_queue_elem_info *
+mlxsw_pci_queue_elem_info_producer_get(struct mlxsw_pci_queue *q)
+{
+	int index = q->producer_counter & (q->count - 1);
+
+	if ((u16) (q->producer_counter - q->consumer_counter) == q->count)
+		return NULL;
+	return mlxsw_pci_queue_elem_info_get(q, index);
+}
+
+static struct mlxsw_pci_queue_elem_info *
+mlxsw_pci_queue_elem_info_consumer_get(struct mlxsw_pci_queue *q)
+{
+	int index = q->consumer_counter & (q->count - 1);
+
+	return mlxsw_pci_queue_elem_info_get(q, index);
+}
+
+static char *mlxsw_pci_queue_elem_get(struct mlxsw_pci_queue *q, int elem_index)
+{
+	return mlxsw_pci_queue_elem_info_get(q, elem_index)->elem;
+}
+
+static bool mlxsw_pci_elem_hw_owned(struct mlxsw_pci_queue *q, bool owner_bit)
+{
+	return owner_bit != !!(q->consumer_counter & q->count);
+}
+
+static struct mlxsw_pci_queue_type_group *
+mlxsw_pci_queue_type_group_get(struct mlxsw_pci *mlxsw_pci,
+			       enum mlxsw_pci_queue_type q_type)
+{
+	return &mlxsw_pci->queues[q_type];
+}
+
+static u8 __mlxsw_pci_queue_count(struct mlxsw_pci *mlxsw_pci,
+				  enum mlxsw_pci_queue_type q_type)
+{
+	struct mlxsw_pci_queue_type_group *queue_group;
+
+	queue_group = mlxsw_pci_queue_type_group_get(mlxsw_pci, q_type);
+	return queue_group->count;
+}
+
+static u8 mlxsw_pci_sdq_count(struct mlxsw_pci *mlxsw_pci)
+{
+	return __mlxsw_pci_queue_count(mlxsw_pci, MLXSW_PCI_QUEUE_TYPE_SDQ);
+}
+
+static u8 mlxsw_pci_cq_count(struct mlxsw_pci *mlxsw_pci)
+{
+	return __mlxsw_pci_queue_count(mlxsw_pci, MLXSW_PCI_QUEUE_TYPE_CQ);
+}
+
+static struct mlxsw_pci_queue *
+__mlxsw_pci_queue_get(struct mlxsw_pci *mlxsw_pci,
+		      enum mlxsw_pci_queue_type q_type, u8 q_num)
+{
+	return &mlxsw_pci->queues[q_type].q[q_num];
+}
+
+static struct mlxsw_pci_queue *mlxsw_pci_sdq_get(struct mlxsw_pci *mlxsw_pci,
+						 u8 q_num)
+{
+	return __mlxsw_pci_queue_get(mlxsw_pci,
+				     MLXSW_PCI_QUEUE_TYPE_SDQ, q_num);
+}
+
+static struct mlxsw_pci_queue *mlxsw_pci_rdq_get(struct mlxsw_pci *mlxsw_pci,
+						 u8 q_num)
+{
+	return __mlxsw_pci_queue_get(mlxsw_pci,
+				     MLXSW_PCI_QUEUE_TYPE_RDQ, q_num);
+}
+
+static struct mlxsw_pci_queue *mlxsw_pci_cq_get(struct mlxsw_pci *mlxsw_pci,
+						u8 q_num)
+{
+	return __mlxsw_pci_queue_get(mlxsw_pci, MLXSW_PCI_QUEUE_TYPE_CQ, q_num);
+}
+
+static struct mlxsw_pci_queue *mlxsw_pci_eq_get(struct mlxsw_pci *mlxsw_pci,
+						u8 q_num)
+{
+	return __mlxsw_pci_queue_get(mlxsw_pci, MLXSW_PCI_QUEUE_TYPE_EQ, q_num);
+}
+
+static void __mlxsw_pci_queue_doorbell_set(struct mlxsw_pci *mlxsw_pci,
+					   struct mlxsw_pci_queue *q,
+					   u16 val)
+{
+	mlxsw_pci_write32(mlxsw_pci,
+			  DOORBELL(mlxsw_pci->doorbell_offset,
+				   mlxsw_pci_doorbell_type_offset[q->type],
+				   q->num), val);
+}
+
+static void __mlxsw_pci_queue_doorbell_arm_set(struct mlxsw_pci *mlxsw_pci,
+					       struct mlxsw_pci_queue *q,
+					       u16 val)
+{
+	mlxsw_pci_write32(mlxsw_pci,
+			  DOORBELL(mlxsw_pci->doorbell_offset,
+				   mlxsw_pci_doorbell_arm_type_offset[q->type],
+				   q->num), val);
+}
+
+static void mlxsw_pci_queue_doorbell_producer_ring(struct mlxsw_pci *mlxsw_pci,
+						   struct mlxsw_pci_queue *q)
+{
+	wmb(); /* ensure all writes are done before we ring a bell */
+	__mlxsw_pci_queue_doorbell_set(mlxsw_pci, q, q->producer_counter);
+}
+
+static void mlxsw_pci_queue_doorbell_consumer_ring(struct mlxsw_pci *mlxsw_pci,
+						   struct mlxsw_pci_queue *q)
+{
+	wmb(); /* ensure all writes are done before we ring a bell */
+	__mlxsw_pci_queue_doorbell_set(mlxsw_pci, q,
+				       q->consumer_counter + q->count);
+}
+
+static void
+mlxsw_pci_queue_doorbell_arm_consumer_ring(struct mlxsw_pci *mlxsw_pci,
+					   struct mlxsw_pci_queue *q)
+{
+	wmb(); /* ensure all writes are done before we ring a bell */
+	__mlxsw_pci_queue_doorbell_arm_set(mlxsw_pci, q, q->consumer_counter);
+}
+
+static dma_addr_t __mlxsw_pci_queue_page_get(struct mlxsw_pci_queue *q,
+					     int page_index)
+{
+	return q->mem_item.mapaddr + MLXSW_PCI_PAGE_SIZE * page_index;
+}
+
+static int mlxsw_pci_sdq_init(struct mlxsw_pci *mlxsw_pci, char *mbox,
+			      struct mlxsw_pci_queue *q)
+{
+	int tclass;
+	int lp;
+	int i;
+	int err;
+
+	q->producer_counter = 0;
+	q->consumer_counter = 0;
+	tclass = q->num == MLXSW_PCI_SDQ_EMAD_INDEX ? MLXSW_PCI_SDQ_EMAD_TC :
+						      MLXSW_PCI_SDQ_CTL_TC;
+	lp = q->num == MLXSW_PCI_SDQ_EMAD_INDEX ? MLXSW_CMD_MBOX_SW2HW_DQ_SDQ_LP_IGNORE_WQE :
+						  MLXSW_CMD_MBOX_SW2HW_DQ_SDQ_LP_WQE;
+
+	/* Set CQ of same number of this SDQ. */
+	mlxsw_cmd_mbox_sw2hw_dq_cq_set(mbox, q->num);
+	mlxsw_cmd_mbox_sw2hw_dq_sdq_lp_set(mbox, lp);
+	mlxsw_cmd_mbox_sw2hw_dq_sdq_tclass_set(mbox, tclass);
+	mlxsw_cmd_mbox_sw2hw_dq_log2_dq_sz_set(mbox, 3); /* 8 pages */
+	for (i = 0; i < MLXSW_PCI_AQ_PAGES; i++) {
+		dma_addr_t mapaddr = __mlxsw_pci_queue_page_get(q, i);
+
+		mlxsw_cmd_mbox_sw2hw_dq_pa_set(mbox, i, mapaddr);
+	}
+
+	err = mlxsw_cmd_sw2hw_sdq(mlxsw_pci->core, mbox, q->num);
+	if (err)
+		return err;
+	mlxsw_pci_queue_doorbell_producer_ring(mlxsw_pci, q);
+	return 0;
+}
+
+static void mlxsw_pci_sdq_fini(struct mlxsw_pci *mlxsw_pci,
+			       struct mlxsw_pci_queue *q)
+{
+	mlxsw_cmd_hw2sw_sdq(mlxsw_pci->core, q->num);
+}
+
+static int mlxsw_pci_wqe_frag_map(struct mlxsw_pci *mlxsw_pci, char *wqe,
+				  int index, char *frag_data, size_t frag_len,
+				  int direction)
+{
+	struct pci_dev *pdev = mlxsw_pci->pdev;
+	dma_addr_t mapaddr;
+
+	mapaddr = dma_map_single(&pdev->dev, frag_data, frag_len, direction);
+	if (unlikely(dma_mapping_error(&pdev->dev, mapaddr))) {
+		dev_err_ratelimited(&pdev->dev, "failed to dma map tx frag\n");
+		return -EIO;
+	}
+	mlxsw_pci_wqe_address_set(wqe, index, mapaddr);
+	mlxsw_pci_wqe_byte_count_set(wqe, index, frag_len);
+	return 0;
+}
+
+static void mlxsw_pci_wqe_frag_unmap(struct mlxsw_pci *mlxsw_pci, char *wqe,
+				     int index, int direction)
+{
+	struct pci_dev *pdev = mlxsw_pci->pdev;
+	size_t frag_len = mlxsw_pci_wqe_byte_count_get(wqe, index);
+	dma_addr_t mapaddr = mlxsw_pci_wqe_address_get(wqe, index);
+
+	if (!frag_len)
+		return;
+	dma_unmap_single(&pdev->dev, mapaddr, frag_len, direction);
+}
+
+static int mlxsw_pci_rdq_skb_alloc(struct mlxsw_pci *mlxsw_pci,
+				   struct mlxsw_pci_queue_elem_info *elem_info)
+{
+	size_t buf_len = MLXSW_PORT_MAX_MTU;
+	char *wqe = elem_info->elem;
+	struct sk_buff *skb;
+	int err;
+
+	skb = netdev_alloc_skb_ip_align(NULL, buf_len);
+	if (!skb)
+		return -ENOMEM;
+
+	err = mlxsw_pci_wqe_frag_map(mlxsw_pci, wqe, 0, skb->data,
+				     buf_len, DMA_FROM_DEVICE);
+	if (err)
+		goto err_frag_map;
+
+	elem_info->u.rdq.skb = skb;
+	return 0;
+
+err_frag_map:
+	dev_kfree_skb_any(skb);
+	return err;
+}
+
+static void mlxsw_pci_rdq_skb_free(struct mlxsw_pci *mlxsw_pci,
+				   struct mlxsw_pci_queue_elem_info *elem_info)
+{
+	struct sk_buff *skb;
+	char *wqe;
+
+	skb = elem_info->u.rdq.skb;
+	wqe = elem_info->elem;
+
+	mlxsw_pci_wqe_frag_unmap(mlxsw_pci, wqe, 0, DMA_FROM_DEVICE);
+	dev_kfree_skb_any(skb);
+}
+
+static int mlxsw_pci_rdq_init(struct mlxsw_pci *mlxsw_pci, char *mbox,
+			      struct mlxsw_pci_queue *q)
+{
+	struct mlxsw_pci_queue_elem_info *elem_info;
+	u8 sdq_count = mlxsw_pci_sdq_count(mlxsw_pci);
+	int i;
+	int err;
+
+	q->producer_counter = 0;
+	q->consumer_counter = 0;
+
+	/* Set CQ of same number of this RDQ with base
+	 * above SDQ count as the lower ones are assigned to SDQs.
+	 */
+	mlxsw_cmd_mbox_sw2hw_dq_cq_set(mbox, sdq_count + q->num);
+	mlxsw_cmd_mbox_sw2hw_dq_log2_dq_sz_set(mbox, 3); /* 8 pages */
+	for (i = 0; i < MLXSW_PCI_AQ_PAGES; i++) {
+		dma_addr_t mapaddr = __mlxsw_pci_queue_page_get(q, i);
+
+		mlxsw_cmd_mbox_sw2hw_dq_pa_set(mbox, i, mapaddr);
+	}
+
+	err = mlxsw_cmd_sw2hw_rdq(mlxsw_pci->core, mbox, q->num);
+	if (err)
+		return err;
+
+	mlxsw_pci_queue_doorbell_producer_ring(mlxsw_pci, q);
+
+	for (i = 0; i < q->count; i++) {
+		elem_info = mlxsw_pci_queue_elem_info_producer_get(q);
+		BUG_ON(!elem_info);
+		err = mlxsw_pci_rdq_skb_alloc(mlxsw_pci, elem_info);
+		if (err)
+			goto rollback;
+		/* Everything is set up, ring doorbell to pass elem to HW */
+		q->producer_counter++;
+		mlxsw_pci_queue_doorbell_producer_ring(mlxsw_pci, q);
+	}
+
+	return 0;
+
+rollback:
+	for (i--; i >= 0; i--) {
+		elem_info = mlxsw_pci_queue_elem_info_get(q, i);
+		mlxsw_pci_rdq_skb_free(mlxsw_pci, elem_info);
+	}
+	mlxsw_cmd_hw2sw_rdq(mlxsw_pci->core, q->num);
+
+	return err;
+}
+
+static void mlxsw_pci_rdq_fini(struct mlxsw_pci *mlxsw_pci,
+			       struct mlxsw_pci_queue *q)
+{
+	struct mlxsw_pci_queue_elem_info *elem_info;
+	int i;
+
+	mlxsw_cmd_hw2sw_rdq(mlxsw_pci->core, q->num);
+	for (i = 0; i < q->count; i++) {
+		elem_info = mlxsw_pci_queue_elem_info_get(q, i);
+		mlxsw_pci_rdq_skb_free(mlxsw_pci, elem_info);
+	}
+}
+
+static void mlxsw_pci_cq_pre_init(struct mlxsw_pci *mlxsw_pci,
+				  struct mlxsw_pci_queue *q)
+{
+	q->u.cq.v = mlxsw_pci->max_cqe_ver;
+
+	if (q->u.cq.v == MLXSW_PCI_CQE_V2 &&
+	    q->num < mlxsw_pci->num_sdq_cqs &&
+	    !mlxsw_core_sdq_supports_cqe_v2(mlxsw_pci->core))
+		q->u.cq.v = MLXSW_PCI_CQE_V1;
+}
+
+static int mlxsw_pci_cq_init(struct mlxsw_pci *mlxsw_pci, char *mbox,
+			     struct mlxsw_pci_queue *q)
+{
+	int i;
+	int err;
+
+	q->consumer_counter = 0;
+
+	for (i = 0; i < q->count; i++) {
+		char *elem = mlxsw_pci_queue_elem_get(q, i);
+
+		mlxsw_pci_cqe_owner_set(q->u.cq.v, elem, 1);
+	}
+
+	if (q->u.cq.v == MLXSW_PCI_CQE_V1)
+		mlxsw_cmd_mbox_sw2hw_cq_cqe_ver_set(mbox,
+				MLXSW_CMD_MBOX_SW2HW_CQ_CQE_VER_1);
+	else if (q->u.cq.v == MLXSW_PCI_CQE_V2)
+		mlxsw_cmd_mbox_sw2hw_cq_cqe_ver_set(mbox,
+				MLXSW_CMD_MBOX_SW2HW_CQ_CQE_VER_2);
+
+	mlxsw_cmd_mbox_sw2hw_cq_c_eqn_set(mbox, MLXSW_PCI_EQ_COMP_NUM);
+	mlxsw_cmd_mbox_sw2hw_cq_st_set(mbox, 0);
+	mlxsw_cmd_mbox_sw2hw_cq_log_cq_size_set(mbox, ilog2(q->count));
+	for (i = 0; i < MLXSW_PCI_AQ_PAGES; i++) {
+		dma_addr_t mapaddr = __mlxsw_pci_queue_page_get(q, i);
+
+		mlxsw_cmd_mbox_sw2hw_cq_pa_set(mbox, i, mapaddr);
+	}
+	err = mlxsw_cmd_sw2hw_cq(mlxsw_pci->core, mbox, q->num);
+	if (err)
+		return err;
+	mlxsw_pci_queue_doorbell_consumer_ring(mlxsw_pci, q);
+	mlxsw_pci_queue_doorbell_arm_consumer_ring(mlxsw_pci, q);
+	return 0;
+}
+
+static void mlxsw_pci_cq_fini(struct mlxsw_pci *mlxsw_pci,
+			      struct mlxsw_pci_queue *q)
+{
+	mlxsw_cmd_hw2sw_cq(mlxsw_pci->core, q->num);
+}
+
+static unsigned int mlxsw_pci_read32_off(struct mlxsw_pci *mlxsw_pci,
+					 ptrdiff_t off)
+{
+	return ioread32be(mlxsw_pci->hw_addr + off);
+}
+
+static void mlxsw_pci_skb_cb_ts_set(struct mlxsw_pci *mlxsw_pci,
+				    struct sk_buff *skb,
+				    enum mlxsw_pci_cqe_v cqe_v, char *cqe)
+{
+	u8 ts_type;
+
+	if (cqe_v != MLXSW_PCI_CQE_V2)
+		return;
+
+	ts_type = mlxsw_pci_cqe2_time_stamp_type_get(cqe);
+
+	if (ts_type != MLXSW_PCI_CQE_TIME_STAMP_TYPE_UTC &&
+	    ts_type != MLXSW_PCI_CQE_TIME_STAMP_TYPE_MIRROR_UTC)
+		return;
+
+	mlxsw_skb_cb(skb)->cqe_ts.sec = mlxsw_pci_cqe2_time_stamp_sec_get(cqe);
+	mlxsw_skb_cb(skb)->cqe_ts.nsec =
+		mlxsw_pci_cqe2_time_stamp_nsec_get(cqe);
+}
+
+static void mlxsw_pci_cqe_sdq_handle(struct mlxsw_pci *mlxsw_pci,
+				     struct mlxsw_pci_queue *q,
+				     u16 consumer_counter_limit,
+				     enum mlxsw_pci_cqe_v cqe_v,
+				     char *cqe)
+{
+	struct pci_dev *pdev = mlxsw_pci->pdev;
+	struct mlxsw_pci_queue_elem_info *elem_info;
+	struct mlxsw_tx_info tx_info;
+	char *wqe;
+	struct sk_buff *skb;
+	int i;
+
+	spin_lock(&q->lock);
+	elem_info = mlxsw_pci_queue_elem_info_consumer_get(q);
+	tx_info = mlxsw_skb_cb(elem_info->u.sdq.skb)->tx_info;
+	skb = elem_info->u.sdq.skb;
+	wqe = elem_info->elem;
+	for (i = 0; i < MLXSW_PCI_WQE_SG_ENTRIES; i++)
+		mlxsw_pci_wqe_frag_unmap(mlxsw_pci, wqe, i, DMA_TO_DEVICE);
+
+	if (unlikely(!tx_info.is_emad &&
+		     skb_shinfo(skb)->tx_flags & SKBTX_HW_TSTAMP)) {
+		mlxsw_pci_skb_cb_ts_set(mlxsw_pci, skb, cqe_v, cqe);
+		mlxsw_core_ptp_transmitted(mlxsw_pci->core, skb,
+					   tx_info.local_port);
+		skb = NULL;
+	}
+
+	if (skb)
+		dev_kfree_skb_any(skb);
+	elem_info->u.sdq.skb = NULL;
+
+	if (q->consumer_counter++ != consumer_counter_limit)
+		dev_dbg_ratelimited(&pdev->dev, "Consumer counter does not match limit in SDQ\n");
+	spin_unlock(&q->lock);
+}
+
+static void mlxsw_pci_cqe_rdq_md_tx_port_init(struct sk_buff *skb,
+					      const char *cqe)
+{
+	struct mlxsw_skb_cb *cb = mlxsw_skb_cb(skb);
+
+	if (mlxsw_pci_cqe2_tx_lag_get(cqe)) {
+		cb->rx_md_info.tx_port_is_lag = true;
+		cb->rx_md_info.tx_lag_id = mlxsw_pci_cqe2_tx_lag_id_get(cqe);
+		cb->rx_md_info.tx_lag_port_index =
+			mlxsw_pci_cqe2_tx_lag_subport_get(cqe);
+	} else {
+		cb->rx_md_info.tx_port_is_lag = false;
+		cb->rx_md_info.tx_sys_port =
+			mlxsw_pci_cqe2_tx_system_port_get(cqe);
+	}
+
+	if (cb->rx_md_info.tx_sys_port != MLXSW_PCI_CQE2_TX_PORT_MULTI_PORT &&
+	    cb->rx_md_info.tx_sys_port != MLXSW_PCI_CQE2_TX_PORT_INVALID)
+		cb->rx_md_info.tx_port_valid = 1;
+	else
+		cb->rx_md_info.tx_port_valid = 0;
+}
+
+static void mlxsw_pci_cqe_rdq_md_init(struct sk_buff *skb, const char *cqe)
+{
+	struct mlxsw_skb_cb *cb = mlxsw_skb_cb(skb);
+
+	cb->rx_md_info.tx_congestion = mlxsw_pci_cqe2_mirror_cong_get(cqe);
+	if (cb->rx_md_info.tx_congestion != MLXSW_PCI_CQE2_MIRROR_CONG_INVALID)
+		cb->rx_md_info.tx_congestion_valid = 1;
+	else
+		cb->rx_md_info.tx_congestion_valid = 0;
+	cb->rx_md_info.tx_congestion <<= MLXSW_PCI_CQE2_MIRROR_CONG_SHIFT;
+
+	cb->rx_md_info.latency = mlxsw_pci_cqe2_mirror_latency_get(cqe);
+	if (cb->rx_md_info.latency != MLXSW_PCI_CQE2_MIRROR_LATENCY_INVALID)
+		cb->rx_md_info.latency_valid = 1;
+	else
+		cb->rx_md_info.latency_valid = 0;
+
+	cb->rx_md_info.tx_tc = mlxsw_pci_cqe2_mirror_tclass_get(cqe);
+	if (cb->rx_md_info.tx_tc != MLXSW_PCI_CQE2_MIRROR_TCLASS_INVALID)
+		cb->rx_md_info.tx_tc_valid = 1;
+	else
+		cb->rx_md_info.tx_tc_valid = 0;
+
+	mlxsw_pci_cqe_rdq_md_tx_port_init(skb, cqe);
+}
+
+static void mlxsw_pci_cqe_rdq_handle(struct mlxsw_pci *mlxsw_pci,
+				     struct mlxsw_pci_queue *q,
+				     u16 consumer_counter_limit,
+				     enum mlxsw_pci_cqe_v cqe_v, char *cqe)
+{
+	struct pci_dev *pdev = mlxsw_pci->pdev;
+	struct mlxsw_pci_queue_elem_info *elem_info;
+	struct mlxsw_rx_info rx_info = {};
+	char wqe[MLXSW_PCI_WQE_SIZE];
+	struct sk_buff *skb;
+	u16 byte_count;
+	int err;
+
+	elem_info = mlxsw_pci_queue_elem_info_consumer_get(q);
+	skb = elem_info->u.rdq.skb;
+	memcpy(wqe, elem_info->elem, MLXSW_PCI_WQE_SIZE);
+
+	if (q->consumer_counter++ != consumer_counter_limit)
+		dev_dbg_ratelimited(&pdev->dev, "Consumer counter does not match limit in RDQ\n");
+
+	err = mlxsw_pci_rdq_skb_alloc(mlxsw_pci, elem_info);
+	if (err) {
+		dev_err_ratelimited(&pdev->dev, "Failed to alloc skb for RDQ\n");
+		goto out;
+	}
+
+	mlxsw_pci_wqe_frag_unmap(mlxsw_pci, wqe, 0, DMA_FROM_DEVICE);
+
+	if (mlxsw_pci_cqe_lag_get(cqe_v, cqe)) {
+		rx_info.is_lag = true;
+		rx_info.u.lag_id = mlxsw_pci_cqe_lag_id_get(cqe_v, cqe);
+		rx_info.lag_port_index =
+			mlxsw_pci_cqe_lag_subport_get(cqe_v, cqe);
+	} else {
+		rx_info.is_lag = false;
+		rx_info.u.sys_port = mlxsw_pci_cqe_system_port_get(cqe);
+	}
+
+	rx_info.trap_id = mlxsw_pci_cqe_trap_id_get(cqe);
+
+	if (rx_info.trap_id == MLXSW_TRAP_ID_DISCARD_INGRESS_ACL ||
+	    rx_info.trap_id == MLXSW_TRAP_ID_DISCARD_EGRESS_ACL) {
+		u32 cookie_index = 0;
+
+		if (mlxsw_pci->max_cqe_ver >= MLXSW_PCI_CQE_V2)
+			cookie_index = mlxsw_pci_cqe2_user_def_val_orig_pkt_len_get(cqe);
+		mlxsw_skb_cb(skb)->rx_md_info.cookie_index = cookie_index;
+	} else if (rx_info.trap_id >= MLXSW_TRAP_ID_MIRROR_SESSION0 &&
+		   rx_info.trap_id <= MLXSW_TRAP_ID_MIRROR_SESSION7 &&
+		   mlxsw_pci->max_cqe_ver >= MLXSW_PCI_CQE_V2) {
+		rx_info.mirror_reason = mlxsw_pci_cqe2_mirror_reason_get(cqe);
+		mlxsw_pci_cqe_rdq_md_init(skb, cqe);
+	} else if (rx_info.trap_id == MLXSW_TRAP_ID_PKT_SAMPLE &&
+		   mlxsw_pci->max_cqe_ver >= MLXSW_PCI_CQE_V2) {
+		mlxsw_pci_cqe_rdq_md_tx_port_init(skb, cqe);
+	}
+
+	mlxsw_pci_skb_cb_ts_set(mlxsw_pci, skb, cqe_v, cqe);
+
+	byte_count = mlxsw_pci_cqe_byte_count_get(cqe);
+	if (mlxsw_pci_cqe_crc_get(cqe_v, cqe))
+		byte_count -= ETH_FCS_LEN;
+	skb_put(skb, byte_count);
+	mlxsw_core_skb_receive(mlxsw_pci->core, skb, &rx_info);
+
+out:
+	/* Everything is set up, ring doorbell to pass elem to HW */
+	q->producer_counter++;
+	mlxsw_pci_queue_doorbell_producer_ring(mlxsw_pci, q);
+	return;
+}
+
+static char *mlxsw_pci_cq_sw_cqe_get(struct mlxsw_pci_queue *q)
+{
+	struct mlxsw_pci_queue_elem_info *elem_info;
+	char *elem;
+	bool owner_bit;
+
+	elem_info = mlxsw_pci_queue_elem_info_consumer_get(q);
+	elem = elem_info->elem;
+	owner_bit = mlxsw_pci_cqe_owner_get(q->u.cq.v, elem);
+	if (mlxsw_pci_elem_hw_owned(q, owner_bit))
+		return NULL;
+	q->consumer_counter++;
+	rmb(); /* make sure we read owned bit before the rest of elem */
+	return elem;
+}
+
+static void mlxsw_pci_cq_tasklet(struct tasklet_struct *t)
+{
+	struct mlxsw_pci_queue *q = from_tasklet(q, t, tasklet);
+	struct mlxsw_pci *mlxsw_pci = q->pci;
+	char *cqe;
+	int items = 0;
+	int credits = q->count >> 1;
+
+	while ((cqe = mlxsw_pci_cq_sw_cqe_get(q))) {
+		u16 wqe_counter = mlxsw_pci_cqe_wqe_counter_get(cqe);
+		u8 sendq = mlxsw_pci_cqe_sr_get(q->u.cq.v, cqe);
+		u8 dqn = mlxsw_pci_cqe_dqn_get(q->u.cq.v, cqe);
+		char ncqe[MLXSW_PCI_CQE_SIZE_MAX];
+
+		memcpy(ncqe, cqe, q->elem_size);
+		mlxsw_pci_queue_doorbell_consumer_ring(mlxsw_pci, q);
+
+		if (sendq) {
+			struct mlxsw_pci_queue *sdq;
+
+			sdq = mlxsw_pci_sdq_get(mlxsw_pci, dqn);
+			mlxsw_pci_cqe_sdq_handle(mlxsw_pci, sdq,
+						 wqe_counter, q->u.cq.v, ncqe);
+			q->u.cq.comp_sdq_count++;
+		} else {
+			struct mlxsw_pci_queue *rdq;
+
+			rdq = mlxsw_pci_rdq_get(mlxsw_pci, dqn);
+			mlxsw_pci_cqe_rdq_handle(mlxsw_pci, rdq,
+						 wqe_counter, q->u.cq.v, ncqe);
+			q->u.cq.comp_rdq_count++;
+		}
+		if (++items == credits)
+			break;
+	}
+	if (items)
+		mlxsw_pci_queue_doorbell_arm_consumer_ring(mlxsw_pci, q);
+}
+
+static u16 mlxsw_pci_cq_elem_count(const struct mlxsw_pci_queue *q)
+{
+	return q->u.cq.v == MLXSW_PCI_CQE_V2 ? MLXSW_PCI_CQE2_COUNT :
+					       MLXSW_PCI_CQE01_COUNT;
+}
+
+static u8 mlxsw_pci_cq_elem_size(const struct mlxsw_pci_queue *q)
+{
+	return q->u.cq.v == MLXSW_PCI_CQE_V2 ? MLXSW_PCI_CQE2_SIZE :
+					       MLXSW_PCI_CQE01_SIZE;
+}
+
+static int mlxsw_pci_eq_init(struct mlxsw_pci *mlxsw_pci, char *mbox,
+			     struct mlxsw_pci_queue *q)
+{
+	int i;
+	int err;
+
+	q->consumer_counter = 0;
+
+	for (i = 0; i < q->count; i++) {
+		char *elem = mlxsw_pci_queue_elem_get(q, i);
+
+		mlxsw_pci_eqe_owner_set(elem, 1);
+	}
+
+	mlxsw_cmd_mbox_sw2hw_eq_int_msix_set(mbox, 1); /* MSI-X used */
+	mlxsw_cmd_mbox_sw2hw_eq_st_set(mbox, 1); /* armed */
+	mlxsw_cmd_mbox_sw2hw_eq_log_eq_size_set(mbox, ilog2(q->count));
+	for (i = 0; i < MLXSW_PCI_AQ_PAGES; i++) {
+		dma_addr_t mapaddr = __mlxsw_pci_queue_page_get(q, i);
+
+		mlxsw_cmd_mbox_sw2hw_eq_pa_set(mbox, i, mapaddr);
+	}
+	err = mlxsw_cmd_sw2hw_eq(mlxsw_pci->core, mbox, q->num);
+	if (err)
+		return err;
+	mlxsw_pci_queue_doorbell_consumer_ring(mlxsw_pci, q);
+	mlxsw_pci_queue_doorbell_arm_consumer_ring(mlxsw_pci, q);
+	return 0;
+}
+
+static void mlxsw_pci_eq_fini(struct mlxsw_pci *mlxsw_pci,
+			      struct mlxsw_pci_queue *q)
+{
+	mlxsw_cmd_hw2sw_eq(mlxsw_pci->core, q->num);
+}
+
+static void mlxsw_pci_eq_cmd_event(struct mlxsw_pci *mlxsw_pci, char *eqe)
+{
+	mlxsw_pci->cmd.comp.status = mlxsw_pci_eqe_cmd_status_get(eqe);
+	mlxsw_pci->cmd.comp.out_param =
+		((u64) mlxsw_pci_eqe_cmd_out_param_h_get(eqe)) << 32 |
+		mlxsw_pci_eqe_cmd_out_param_l_get(eqe);
+	mlxsw_pci->cmd.wait_done = true;
+	wake_up(&mlxsw_pci->cmd.wait);
+}
+
+static char *mlxsw_pci_eq_sw_eqe_get(struct mlxsw_pci_queue *q)
+{
+	struct mlxsw_pci_queue_elem_info *elem_info;
+	char *elem;
+	bool owner_bit;
+
+	elem_info = mlxsw_pci_queue_elem_info_consumer_get(q);
+	elem = elem_info->elem;
+	owner_bit = mlxsw_pci_eqe_owner_get(elem);
+	if (mlxsw_pci_elem_hw_owned(q, owner_bit))
+		return NULL;
+	q->consumer_counter++;
+	rmb(); /* make sure we read owned bit before the rest of elem */
+	return elem;
+}
+
+static void mlxsw_pci_eq_tasklet(struct tasklet_struct *t)
+{
+	struct mlxsw_pci_queue *q = from_tasklet(q, t, tasklet);
+	struct mlxsw_pci *mlxsw_pci = q->pci;
+	u8 cq_count = mlxsw_pci_cq_count(mlxsw_pci);
+	unsigned long active_cqns[BITS_TO_LONGS(MLXSW_PCI_CQS_MAX)];
+	char *eqe;
+	u8 cqn;
+	bool cq_handle = false;
+	int items = 0;
+	int credits = q->count >> 1;
+
+	memset(&active_cqns, 0, sizeof(active_cqns));
+
+	while ((eqe = mlxsw_pci_eq_sw_eqe_get(q))) {
+
+		/* Command interface completion events are always received on
+		 * queue MLXSW_PCI_EQ_ASYNC_NUM (EQ0) and completion events
+		 * are mapped to queue MLXSW_PCI_EQ_COMP_NUM (EQ1).
+		 */
+		switch (q->num) {
+		case MLXSW_PCI_EQ_ASYNC_NUM:
+			mlxsw_pci_eq_cmd_event(mlxsw_pci, eqe);
+			q->u.eq.ev_cmd_count++;
+			break;
+		case MLXSW_PCI_EQ_COMP_NUM:
+			cqn = mlxsw_pci_eqe_cqn_get(eqe);
+			set_bit(cqn, active_cqns);
+			cq_handle = true;
+			q->u.eq.ev_comp_count++;
+			break;
+		default:
+			q->u.eq.ev_other_count++;
+		}
+		if (++items == credits)
+			break;
+	}
+	if (items) {
+		mlxsw_pci_queue_doorbell_consumer_ring(mlxsw_pci, q);
+		mlxsw_pci_queue_doorbell_arm_consumer_ring(mlxsw_pci, q);
+	}
+
+	if (!cq_handle)
+		return;
+	for_each_set_bit(cqn, active_cqns, cq_count) {
+		q = mlxsw_pci_cq_get(mlxsw_pci, cqn);
+		mlxsw_pci_queue_tasklet_schedule(q);
+	}
+}
+
+struct mlxsw_pci_queue_ops {
+	const char *name;
+	enum mlxsw_pci_queue_type type;
+	void (*pre_init)(struct mlxsw_pci *mlxsw_pci,
+			 struct mlxsw_pci_queue *q);
+	int (*init)(struct mlxsw_pci *mlxsw_pci, char *mbox,
+		    struct mlxsw_pci_queue *q);
+	void (*fini)(struct mlxsw_pci *mlxsw_pci,
+		     struct mlxsw_pci_queue *q);
+	void (*tasklet)(struct tasklet_struct *t);
+	u16 (*elem_count_f)(const struct mlxsw_pci_queue *q);
+	u8 (*elem_size_f)(const struct mlxsw_pci_queue *q);
+	u16 elem_count;
+	u8 elem_size;
+};
+
+static const struct mlxsw_pci_queue_ops mlxsw_pci_sdq_ops = {
+	.type		= MLXSW_PCI_QUEUE_TYPE_SDQ,
+	.init		= mlxsw_pci_sdq_init,
+	.fini		= mlxsw_pci_sdq_fini,
+	.elem_count	= MLXSW_PCI_WQE_COUNT,
+	.elem_size	= MLXSW_PCI_WQE_SIZE,
+};
+
+static const struct mlxsw_pci_queue_ops mlxsw_pci_rdq_ops = {
+	.type		= MLXSW_PCI_QUEUE_TYPE_RDQ,
+	.init		= mlxsw_pci_rdq_init,
+	.fini		= mlxsw_pci_rdq_fini,
+	.elem_count	= MLXSW_PCI_WQE_COUNT,
+	.elem_size	= MLXSW_PCI_WQE_SIZE
+};
+
+static const struct mlxsw_pci_queue_ops mlxsw_pci_cq_ops = {
+	.type		= MLXSW_PCI_QUEUE_TYPE_CQ,
+	.pre_init	= mlxsw_pci_cq_pre_init,
+	.init		= mlxsw_pci_cq_init,
+	.fini		= mlxsw_pci_cq_fini,
+	.tasklet	= mlxsw_pci_cq_tasklet,
+	.elem_count_f	= mlxsw_pci_cq_elem_count,
+	.elem_size_f	= mlxsw_pci_cq_elem_size
+};
+
+static const struct mlxsw_pci_queue_ops mlxsw_pci_eq_ops = {
+	.type		= MLXSW_PCI_QUEUE_TYPE_EQ,
+	.init		= mlxsw_pci_eq_init,
+	.fini		= mlxsw_pci_eq_fini,
+	.tasklet	= mlxsw_pci_eq_tasklet,
+	.elem_count	= MLXSW_PCI_EQE_COUNT,
+	.elem_size	= MLXSW_PCI_EQE_SIZE
+};
+
+static int mlxsw_pci_queue_init(struct mlxsw_pci *mlxsw_pci, char *mbox,
+				const struct mlxsw_pci_queue_ops *q_ops,
+				struct mlxsw_pci_queue *q, u8 q_num)
+{
+	struct mlxsw_pci_mem_item *mem_item = &q->mem_item;
+	int i;
+	int err;
+
+	q->num = q_num;
+	if (q_ops->pre_init)
+		q_ops->pre_init(mlxsw_pci, q);
+
+	spin_lock_init(&q->lock);
+	q->count = q_ops->elem_count_f ? q_ops->elem_count_f(q) :
+					 q_ops->elem_count;
+	q->elem_size = q_ops->elem_size_f ? q_ops->elem_size_f(q) :
+					    q_ops->elem_size;
+	q->type = q_ops->type;
+	q->pci = mlxsw_pci;
+
+	if (q_ops->tasklet)
+		tasklet_setup(&q->tasklet, q_ops->tasklet);
+
+	mem_item->size = MLXSW_PCI_AQ_SIZE;
+	mem_item->buf = dma_alloc_coherent(&mlxsw_pci->pdev->dev,
+					   mem_item->size, &mem_item->mapaddr,
+					   GFP_KERNEL);
+	if (!mem_item->buf)
+		return -ENOMEM;
+
+	q->elem_info = kcalloc(q->count, sizeof(*q->elem_info), GFP_KERNEL);
+	if (!q->elem_info) {
+		err = -ENOMEM;
+		goto err_elem_info_alloc;
+	}
+
+	/* Initialize dma mapped elements info elem_info for
+	 * future easy access.
+	 */
+	for (i = 0; i < q->count; i++) {
+		struct mlxsw_pci_queue_elem_info *elem_info;
+
+		elem_info = mlxsw_pci_queue_elem_info_get(q, i);
+		elem_info->elem =
+			__mlxsw_pci_queue_elem_get(q, q->elem_size, i);
+	}
+
+	mlxsw_cmd_mbox_zero(mbox);
+	err = q_ops->init(mlxsw_pci, mbox, q);
+	if (err)
+		goto err_q_ops_init;
+	return 0;
+
+err_q_ops_init:
+	kfree(q->elem_info);
+err_elem_info_alloc:
+	dma_free_coherent(&mlxsw_pci->pdev->dev, mem_item->size,
+			  mem_item->buf, mem_item->mapaddr);
+	return err;
+}
+
+static void mlxsw_pci_queue_fini(struct mlxsw_pci *mlxsw_pci,
+				 const struct mlxsw_pci_queue_ops *q_ops,
+				 struct mlxsw_pci_queue *q)
+{
+	struct mlxsw_pci_mem_item *mem_item = &q->mem_item;
+
+	q_ops->fini(mlxsw_pci, q);
+	kfree(q->elem_info);
+	dma_free_coherent(&mlxsw_pci->pdev->dev, mem_item->size,
+			  mem_item->buf, mem_item->mapaddr);
+}
+
+static int mlxsw_pci_queue_group_init(struct mlxsw_pci *mlxsw_pci, char *mbox,
+				      const struct mlxsw_pci_queue_ops *q_ops,
+				      u8 num_qs)
+{
+	struct mlxsw_pci_queue_type_group *queue_group;
+	int i;
+	int err;
+
+	queue_group = mlxsw_pci_queue_type_group_get(mlxsw_pci, q_ops->type);
+	queue_group->q = kcalloc(num_qs, sizeof(*queue_group->q), GFP_KERNEL);
+	if (!queue_group->q)
+		return -ENOMEM;
+
+	for (i = 0; i < num_qs; i++) {
+		err = mlxsw_pci_queue_init(mlxsw_pci, mbox, q_ops,
+					   &queue_group->q[i], i);
+		if (err)
+			goto err_queue_init;
+	}
+	queue_group->count = num_qs;
+
+	return 0;
+
+err_queue_init:
+	for (i--; i >= 0; i--)
+		mlxsw_pci_queue_fini(mlxsw_pci, q_ops, &queue_group->q[i]);
+	kfree(queue_group->q);
+	return err;
+}
+
+static void mlxsw_pci_queue_group_fini(struct mlxsw_pci *mlxsw_pci,
+				       const struct mlxsw_pci_queue_ops *q_ops)
+{
+	struct mlxsw_pci_queue_type_group *queue_group;
+	int i;
+
+	queue_group = mlxsw_pci_queue_type_group_get(mlxsw_pci, q_ops->type);
+	for (i = 0; i < queue_group->count; i++)
+		mlxsw_pci_queue_fini(mlxsw_pci, q_ops, &queue_group->q[i]);
+	kfree(queue_group->q);
+}
+
+static int mlxsw_pci_aqs_init(struct mlxsw_pci *mlxsw_pci, char *mbox)
+{
+	struct pci_dev *pdev = mlxsw_pci->pdev;
+	u8 num_sdqs;
+	u8 sdq_log2sz;
+	u8 num_rdqs;
+	u8 rdq_log2sz;
+	u8 num_cqs;
+	u8 cq_log2sz;
+	u8 cqv2_log2sz;
+	u8 num_eqs;
+	u8 eq_log2sz;
+	int err;
+
+	mlxsw_cmd_mbox_zero(mbox);
+	err = mlxsw_cmd_query_aq_cap(mlxsw_pci->core, mbox);
+	if (err)
+		return err;
+
+	num_sdqs = mlxsw_cmd_mbox_query_aq_cap_max_num_sdqs_get(mbox);
+	sdq_log2sz = mlxsw_cmd_mbox_query_aq_cap_log_max_sdq_sz_get(mbox);
+	num_rdqs = mlxsw_cmd_mbox_query_aq_cap_max_num_rdqs_get(mbox);
+	rdq_log2sz = mlxsw_cmd_mbox_query_aq_cap_log_max_rdq_sz_get(mbox);
+	num_cqs = mlxsw_cmd_mbox_query_aq_cap_max_num_cqs_get(mbox);
+	cq_log2sz = mlxsw_cmd_mbox_query_aq_cap_log_max_cq_sz_get(mbox);
+	cqv2_log2sz = mlxsw_cmd_mbox_query_aq_cap_log_max_cqv2_sz_get(mbox);
+	num_eqs = mlxsw_cmd_mbox_query_aq_cap_max_num_eqs_get(mbox);
+	eq_log2sz = mlxsw_cmd_mbox_query_aq_cap_log_max_eq_sz_get(mbox);
+
+	if (num_sdqs + num_rdqs > num_cqs ||
+	    num_sdqs < MLXSW_PCI_SDQS_MIN ||
+	    num_cqs > MLXSW_PCI_CQS_MAX || num_eqs != MLXSW_PCI_EQS_COUNT) {
+		dev_err(&pdev->dev, "Unsupported number of queues\n");
+		return -EINVAL;
+	}
+
+	if ((1 << sdq_log2sz != MLXSW_PCI_WQE_COUNT) ||
+	    (1 << rdq_log2sz != MLXSW_PCI_WQE_COUNT) ||
+	    (1 << cq_log2sz != MLXSW_PCI_CQE01_COUNT) ||
+	    (mlxsw_pci->max_cqe_ver == MLXSW_PCI_CQE_V2 &&
+	     (1 << cqv2_log2sz != MLXSW_PCI_CQE2_COUNT)) ||
+	    (1 << eq_log2sz != MLXSW_PCI_EQE_COUNT)) {
+		dev_err(&pdev->dev, "Unsupported number of async queue descriptors\n");
+		return -EINVAL;
+	}
+
+	mlxsw_pci->num_sdq_cqs = num_sdqs;
+
+	err = mlxsw_pci_queue_group_init(mlxsw_pci, mbox, &mlxsw_pci_eq_ops,
+					 num_eqs);
+	if (err) {
+		dev_err(&pdev->dev, "Failed to initialize event queues\n");
+		return err;
+	}
+
+	err = mlxsw_pci_queue_group_init(mlxsw_pci, mbox, &mlxsw_pci_cq_ops,
+					 num_cqs);
+	if (err) {
+		dev_err(&pdev->dev, "Failed to initialize completion queues\n");
+		goto err_cqs_init;
+	}
+
+	err = mlxsw_pci_queue_group_init(mlxsw_pci, mbox, &mlxsw_pci_sdq_ops,
+					 num_sdqs);
+	if (err) {
+		dev_err(&pdev->dev, "Failed to initialize send descriptor queues\n");
+		goto err_sdqs_init;
+	}
+
+	err = mlxsw_pci_queue_group_init(mlxsw_pci, mbox, &mlxsw_pci_rdq_ops,
+					 num_rdqs);
+	if (err) {
+		dev_err(&pdev->dev, "Failed to initialize receive descriptor queues\n");
+		goto err_rdqs_init;
+	}
+
+	/* We have to poll in command interface until queues are initialized */
+	mlxsw_pci->cmd.nopoll = true;
+	return 0;
+
+err_rdqs_init:
+	mlxsw_pci_queue_group_fini(mlxsw_pci, &mlxsw_pci_sdq_ops);
+err_sdqs_init:
+	mlxsw_pci_queue_group_fini(mlxsw_pci, &mlxsw_pci_cq_ops);
+err_cqs_init:
+	mlxsw_pci_queue_group_fini(mlxsw_pci, &mlxsw_pci_eq_ops);
+	return err;
+}
+
+static void mlxsw_pci_aqs_fini(struct mlxsw_pci *mlxsw_pci)
+{
+	mlxsw_pci->cmd.nopoll = false;
+	mlxsw_pci_queue_group_fini(mlxsw_pci, &mlxsw_pci_rdq_ops);
+	mlxsw_pci_queue_group_fini(mlxsw_pci, &mlxsw_pci_sdq_ops);
+	mlxsw_pci_queue_group_fini(mlxsw_pci, &mlxsw_pci_cq_ops);
+	mlxsw_pci_queue_group_fini(mlxsw_pci, &mlxsw_pci_eq_ops);
+}
+
+static void
+mlxsw_pci_config_profile_swid_config(struct mlxsw_pci *mlxsw_pci,
+				     char *mbox, int index,
+				     const struct mlxsw_swid_config *swid)
+{
+	u8 mask = 0;
+
+	if (swid->used_type) {
+		mlxsw_cmd_mbox_config_profile_swid_config_type_set(
+			mbox, index, swid->type);
+		mask |= 1;
+	}
+	if (swid->used_properties) {
+		mlxsw_cmd_mbox_config_profile_swid_config_properties_set(
+			mbox, index, swid->properties);
+		mask |= 2;
+	}
+	mlxsw_cmd_mbox_config_profile_swid_config_mask_set(mbox, index, mask);
+}
+
+static int
+mlxsw_pci_profile_get_kvd_sizes(const struct mlxsw_pci *mlxsw_pci,
+				const struct mlxsw_config_profile *profile,
+				struct mlxsw_res *res)
+{
+	u64 single_size, double_size, linear_size;
+	int err;
+
+	err = mlxsw_core_kvd_sizes_get(mlxsw_pci->core, profile,
+				       &single_size, &double_size,
+				       &linear_size);
+	if (err)
+		return err;
+
+	MLXSW_RES_SET(res, KVD_SINGLE_SIZE, single_size);
+	MLXSW_RES_SET(res, KVD_DOUBLE_SIZE, double_size);
+	MLXSW_RES_SET(res, KVD_LINEAR_SIZE, linear_size);
+
+	return 0;
+}
+
+static int mlxsw_pci_config_profile(struct mlxsw_pci *mlxsw_pci, char *mbox,
+				    const struct mlxsw_config_profile *profile,
+				    struct mlxsw_res *res)
+{
+	int i;
+	int err;
+
+	mlxsw_cmd_mbox_zero(mbox);
+
+	if (profile->used_max_vepa_channels) {
+		mlxsw_cmd_mbox_config_profile_set_max_vepa_channels_set(
+			mbox, 1);
+		mlxsw_cmd_mbox_config_profile_max_vepa_channels_set(
+			mbox, profile->max_vepa_channels);
+	}
+	if (profile->used_max_lag) {
+		mlxsw_cmd_mbox_config_profile_set_max_lag_set(mbox, 1);
+		mlxsw_cmd_mbox_config_profile_max_lag_set(mbox,
+							  profile->max_lag);
+	}
+	if (profile->used_max_mid) {
+		mlxsw_cmd_mbox_config_profile_set_max_mid_set(
+			mbox, 1);
+		mlxsw_cmd_mbox_config_profile_max_mid_set(
+			mbox, profile->max_mid);
+	}
+	if (profile->used_max_pgt) {
+		mlxsw_cmd_mbox_config_profile_set_max_pgt_set(
+			mbox, 1);
+		mlxsw_cmd_mbox_config_profile_max_pgt_set(
+			mbox, profile->max_pgt);
+	}
+	if (profile->used_max_system_port) {
+		mlxsw_cmd_mbox_config_profile_set_max_system_port_set(
+			mbox, 1);
+		mlxsw_cmd_mbox_config_profile_max_system_port_set(
+			mbox, profile->max_system_port);
+	}
+	if (profile->used_max_vlan_groups) {
+		mlxsw_cmd_mbox_config_profile_set_max_vlan_groups_set(
+			mbox, 1);
+		mlxsw_cmd_mbox_config_profile_max_vlan_groups_set(
+			mbox, profile->max_vlan_groups);
+	}
+	if (profile->used_max_regions) {
+		mlxsw_cmd_mbox_config_profile_set_max_regions_set(
+			mbox, 1);
+		mlxsw_cmd_mbox_config_profile_max_regions_set(
+			mbox, profile->max_regions);
+	}
+	if (profile->used_flood_tables) {
+		mlxsw_cmd_mbox_config_profile_set_flood_tables_set(
+			mbox, 1);
+		mlxsw_cmd_mbox_config_profile_max_flood_tables_set(
+			mbox, profile->max_flood_tables);
+		mlxsw_cmd_mbox_config_profile_max_vid_flood_tables_set(
+			mbox, profile->max_vid_flood_tables);
+		mlxsw_cmd_mbox_config_profile_max_fid_offset_flood_tables_set(
+			mbox, profile->max_fid_offset_flood_tables);
+		mlxsw_cmd_mbox_config_profile_fid_offset_flood_table_size_set(
+			mbox, profile->fid_offset_flood_table_size);
+		mlxsw_cmd_mbox_config_profile_max_fid_flood_tables_set(
+			mbox, profile->max_fid_flood_tables);
+		mlxsw_cmd_mbox_config_profile_fid_flood_table_size_set(
+			mbox, profile->fid_flood_table_size);
+	}
+	if (profile->used_flood_mode) {
+		mlxsw_cmd_mbox_config_profile_set_flood_mode_set(
+			mbox, 1);
+		mlxsw_cmd_mbox_config_profile_flood_mode_set(
+			mbox, profile->flood_mode);
+	}
+	if (profile->used_max_ib_mc) {
+		mlxsw_cmd_mbox_config_profile_set_max_ib_mc_set(
+			mbox, 1);
+		mlxsw_cmd_mbox_config_profile_max_ib_mc_set(
+			mbox, profile->max_ib_mc);
+	}
+	if (profile->used_max_pkey) {
+		mlxsw_cmd_mbox_config_profile_set_max_pkey_set(
+			mbox, 1);
+		mlxsw_cmd_mbox_config_profile_max_pkey_set(
+			mbox, profile->max_pkey);
+	}
+	if (profile->used_ar_sec) {
+		mlxsw_cmd_mbox_config_profile_set_ar_sec_set(
+			mbox, 1);
+		mlxsw_cmd_mbox_config_profile_ar_sec_set(
+			mbox, profile->ar_sec);
+	}
+	if (profile->used_adaptive_routing_group_cap) {
+		mlxsw_cmd_mbox_config_profile_set_adaptive_routing_group_cap_set(
+			mbox, 1);
+		mlxsw_cmd_mbox_config_profile_adaptive_routing_group_cap_set(
+			mbox, profile->adaptive_routing_group_cap);
+	}
+	if (profile->used_ubridge) {
+		mlxsw_cmd_mbox_config_profile_set_ubridge_set(mbox, 1);
+		mlxsw_cmd_mbox_config_profile_ubridge_set(mbox,
+							  profile->ubridge);
+	}
+	if (profile->used_kvd_sizes && MLXSW_RES_VALID(res, KVD_SIZE)) {
+		err = mlxsw_pci_profile_get_kvd_sizes(mlxsw_pci, profile, res);
+		if (err)
+			return err;
+
+		mlxsw_cmd_mbox_config_profile_set_kvd_linear_size_set(mbox, 1);
+		mlxsw_cmd_mbox_config_profile_kvd_linear_size_set(mbox,
+					MLXSW_RES_GET(res, KVD_LINEAR_SIZE));
+		mlxsw_cmd_mbox_config_profile_set_kvd_hash_single_size_set(mbox,
+									   1);
+		mlxsw_cmd_mbox_config_profile_kvd_hash_single_size_set(mbox,
+					MLXSW_RES_GET(res, KVD_SINGLE_SIZE));
+		mlxsw_cmd_mbox_config_profile_set_kvd_hash_double_size_set(
+								mbox, 1);
+		mlxsw_cmd_mbox_config_profile_kvd_hash_double_size_set(mbox,
+					MLXSW_RES_GET(res, KVD_DOUBLE_SIZE));
+	}
+
+	for (i = 0; i < MLXSW_CONFIG_PROFILE_SWID_COUNT; i++)
+		mlxsw_pci_config_profile_swid_config(mlxsw_pci, mbox, i,
+						     &profile->swid_config[i]);
+
+	if (mlxsw_pci->max_cqe_ver > MLXSW_PCI_CQE_V0) {
+		mlxsw_cmd_mbox_config_profile_set_cqe_version_set(mbox, 1);
+		mlxsw_cmd_mbox_config_profile_cqe_version_set(mbox, 1);
+	}
+
+	if (profile->used_cqe_time_stamp_type) {
+		mlxsw_cmd_mbox_config_profile_set_cqe_time_stamp_type_set(mbox,
+									  1);
+		mlxsw_cmd_mbox_config_profile_cqe_time_stamp_type_set(mbox,
+					profile->cqe_time_stamp_type);
+	}
+
+	return mlxsw_cmd_config_profile_set(mlxsw_pci->core, mbox);
+}
+
+static int mlxsw_pci_boardinfo(struct mlxsw_pci *mlxsw_pci, char *mbox)
+{
+	struct mlxsw_bus_info *bus_info = &mlxsw_pci->bus_info;
+	int err;
+
+	mlxsw_cmd_mbox_zero(mbox);
+	err = mlxsw_cmd_boardinfo(mlxsw_pci->core, mbox);
+	if (err)
+		return err;
+	mlxsw_cmd_mbox_boardinfo_vsd_memcpy_from(mbox, bus_info->vsd);
+	mlxsw_cmd_mbox_boardinfo_psid_memcpy_from(mbox, bus_info->psid);
+	return 0;
+}
+
+static int mlxsw_pci_fw_area_init(struct mlxsw_pci *mlxsw_pci, char *mbox,
+				  u16 num_pages)
+{
+	struct mlxsw_pci_mem_item *mem_item;
+	int nent = 0;
+	int i;
+	int err;
+
+	mlxsw_pci->fw_area.items = kcalloc(num_pages, sizeof(*mem_item),
+					   GFP_KERNEL);
+	if (!mlxsw_pci->fw_area.items)
+		return -ENOMEM;
+	mlxsw_pci->fw_area.count = num_pages;
+
+	mlxsw_cmd_mbox_zero(mbox);
+	for (i = 0; i < num_pages; i++) {
+		mem_item = &mlxsw_pci->fw_area.items[i];
+
+		mem_item->size = MLXSW_PCI_PAGE_SIZE;
+		mem_item->buf = dma_alloc_coherent(&mlxsw_pci->pdev->dev,
+						   mem_item->size,
+						   &mem_item->mapaddr, GFP_KERNEL);
+		if (!mem_item->buf) {
+			err = -ENOMEM;
+			goto err_alloc;
+		}
+		mlxsw_cmd_mbox_map_fa_pa_set(mbox, nent, mem_item->mapaddr);
+		mlxsw_cmd_mbox_map_fa_log2size_set(mbox, nent, 0); /* 1 page */
+		if (++nent == MLXSW_CMD_MAP_FA_VPM_ENTRIES_MAX) {
+			err = mlxsw_cmd_map_fa(mlxsw_pci->core, mbox, nent);
+			if (err)
+				goto err_cmd_map_fa;
+			nent = 0;
+			mlxsw_cmd_mbox_zero(mbox);
+		}
+	}
+
+	if (nent) {
+		err = mlxsw_cmd_map_fa(mlxsw_pci->core, mbox, nent);
+		if (err)
+			goto err_cmd_map_fa;
+	}
+
+	return 0;
+
+err_cmd_map_fa:
+err_alloc:
+	for (i--; i >= 0; i--) {
+		mem_item = &mlxsw_pci->fw_area.items[i];
+
+		dma_free_coherent(&mlxsw_pci->pdev->dev, mem_item->size,
+				  mem_item->buf, mem_item->mapaddr);
+	}
+	kfree(mlxsw_pci->fw_area.items);
+	return err;
+}
+
+static void mlxsw_pci_fw_area_fini(struct mlxsw_pci *mlxsw_pci)
+{
+	struct mlxsw_pci_mem_item *mem_item;
+	int i;
+
+	mlxsw_cmd_unmap_fa(mlxsw_pci->core);
+
+	for (i = 0; i < mlxsw_pci->fw_area.count; i++) {
+		mem_item = &mlxsw_pci->fw_area.items[i];
+
+		dma_free_coherent(&mlxsw_pci->pdev->dev, mem_item->size,
+				  mem_item->buf, mem_item->mapaddr);
+	}
+	kfree(mlxsw_pci->fw_area.items);
+}
+
+static irqreturn_t mlxsw_pci_eq_irq_handler(int irq, void *dev_id)
+{
+	struct mlxsw_pci *mlxsw_pci = dev_id;
+	struct mlxsw_pci_queue *q;
+	int i;
+
+	for (i = 0; i < MLXSW_PCI_EQS_COUNT; i++) {
+		q = mlxsw_pci_eq_get(mlxsw_pci, i);
+		mlxsw_pci_queue_tasklet_schedule(q);
+	}
+	return IRQ_HANDLED;
+}
+
+static int mlxsw_pci_mbox_alloc(struct mlxsw_pci *mlxsw_pci,
+				struct mlxsw_pci_mem_item *mbox)
+{
+	struct pci_dev *pdev = mlxsw_pci->pdev;
+	int err = 0;
+
+	mbox->size = MLXSW_CMD_MBOX_SIZE;
+	mbox->buf = dma_alloc_coherent(&pdev->dev, MLXSW_CMD_MBOX_SIZE,
+				       &mbox->mapaddr, GFP_KERNEL);
+	if (!mbox->buf) {
+		dev_err(&pdev->dev, "Failed allocating memory for mailbox\n");
+		err = -ENOMEM;
+	}
+
+	return err;
+}
+
+static void mlxsw_pci_mbox_free(struct mlxsw_pci *mlxsw_pci,
+				struct mlxsw_pci_mem_item *mbox)
+{
+	struct pci_dev *pdev = mlxsw_pci->pdev;
+
+	dma_free_coherent(&pdev->dev, MLXSW_CMD_MBOX_SIZE, mbox->buf,
+			  mbox->mapaddr);
+}
+
+static int mlxsw_pci_sys_ready_wait(struct mlxsw_pci *mlxsw_pci,
+				    const struct pci_device_id *id,
+				    u32 *p_sys_status)
+{
+	unsigned long end;
+	u32 val;
+
+	/* We must wait for the HW to become responsive. */
+	msleep(MLXSW_PCI_SW_RESET_WAIT_MSECS);
+
+	end = jiffies + msecs_to_jiffies(MLXSW_PCI_SW_RESET_TIMEOUT_MSECS);
+	do {
+		val = mlxsw_pci_read32(mlxsw_pci, FW_READY);
+		if ((val & MLXSW_PCI_FW_READY_MASK) == MLXSW_PCI_FW_READY_MAGIC)
+			return 0;
+		cond_resched();
+	} while (time_before(jiffies, end));
+
+	*p_sys_status = val & MLXSW_PCI_FW_READY_MASK;
+
+	return -EBUSY;
+}
+
+static int mlxsw_pci_sw_reset(struct mlxsw_pci *mlxsw_pci,
+			      const struct pci_device_id *id)
+{
+	struct pci_dev *pdev = mlxsw_pci->pdev;
+	char mrsr_pl[MLXSW_REG_MRSR_LEN];
+	u32 sys_status;
+	int err;
+
+	err = mlxsw_pci_sys_ready_wait(mlxsw_pci, id, &sys_status);
+	if (err) {
+		dev_err(&pdev->dev, "Failed to reach system ready status before reset. Status is 0x%x\n",
+			sys_status);
+		return err;
+	}
+
+	mlxsw_reg_mrsr_pack(mrsr_pl);
+	err = mlxsw_reg_write(mlxsw_pci->core, MLXSW_REG(mrsr), mrsr_pl);
+	if (err)
+		return err;
+
+	err = mlxsw_pci_sys_ready_wait(mlxsw_pci, id, &sys_status);
+	if (err) {
+		dev_err(&pdev->dev, "Failed to reach system ready status after reset. Status is 0x%x\n",
+			sys_status);
+		return err;
+	}
+
+	return 0;
+}
+
+static int mlxsw_pci_alloc_irq_vectors(struct mlxsw_pci *mlxsw_pci)
+{
+	int err;
+
+	err = pci_alloc_irq_vectors(mlxsw_pci->pdev, 1, 1, PCI_IRQ_MSIX);
+	if (err < 0)
+		dev_err(&mlxsw_pci->pdev->dev, "MSI-X init failed\n");
+	return err;
+}
+
+static void mlxsw_pci_free_irq_vectors(struct mlxsw_pci *mlxsw_pci)
+{
+	pci_free_irq_vectors(mlxsw_pci->pdev);
+}
+
+static int mlxsw_pci_init(void *bus_priv, struct mlxsw_core *mlxsw_core,
+			  const struct mlxsw_config_profile *profile,
+			  struct mlxsw_res *res)
+{
+	struct mlxsw_pci *mlxsw_pci = bus_priv;
+	struct pci_dev *pdev = mlxsw_pci->pdev;
+	char *mbox;
+	u16 num_pages;
+	int err;
+
+	mlxsw_pci->core = mlxsw_core;
+
+	mbox = mlxsw_cmd_mbox_alloc();
+	if (!mbox)
+		return -ENOMEM;
+
+	err = mlxsw_pci_sw_reset(mlxsw_pci, mlxsw_pci->id);
+	if (err)
+		goto err_sw_reset;
+
+	err = mlxsw_pci_alloc_irq_vectors(mlxsw_pci);
+	if (err < 0) {
+		dev_err(&pdev->dev, "MSI-X init failed\n");
+		goto err_alloc_irq;
+	}
+
+	err = mlxsw_cmd_query_fw(mlxsw_core, mbox);
+	if (err)
+		goto err_query_fw;
+
+	mlxsw_pci->bus_info.fw_rev.major =
+		mlxsw_cmd_mbox_query_fw_fw_rev_major_get(mbox);
+	mlxsw_pci->bus_info.fw_rev.minor =
+		mlxsw_cmd_mbox_query_fw_fw_rev_minor_get(mbox);
+	mlxsw_pci->bus_info.fw_rev.subminor =
+		mlxsw_cmd_mbox_query_fw_fw_rev_subminor_get(mbox);
+
+	if (mlxsw_cmd_mbox_query_fw_cmd_interface_rev_get(mbox) != 1) {
+		dev_err(&pdev->dev, "Unsupported cmd interface revision ID queried from hw\n");
+		err = -EINVAL;
+		goto err_iface_rev;
+	}
+	if (mlxsw_cmd_mbox_query_fw_doorbell_page_bar_get(mbox) != 0) {
+		dev_err(&pdev->dev, "Unsupported doorbell page bar queried from hw\n");
+		err = -EINVAL;
+		goto err_doorbell_page_bar;
+	}
+
+	mlxsw_pci->doorbell_offset =
+		mlxsw_cmd_mbox_query_fw_doorbell_page_offset_get(mbox);
+
+	if (mlxsw_cmd_mbox_query_fw_fr_rn_clk_bar_get(mbox) != 0) {
+		dev_err(&pdev->dev, "Unsupported free running clock BAR queried from hw\n");
+		err = -EINVAL;
+		goto err_fr_rn_clk_bar;
+	}
+
+	mlxsw_pci->free_running_clock_offset =
+		mlxsw_cmd_mbox_query_fw_free_running_clock_offset_get(mbox);
+
+	if (mlxsw_cmd_mbox_query_fw_utc_sec_bar_get(mbox) != 0) {
+		dev_err(&pdev->dev, "Unsupported UTC sec BAR queried from hw\n");
+		err = -EINVAL;
+		goto err_utc_sec_bar;
+	}
+
+	mlxsw_pci->utc_sec_offset =
+		mlxsw_cmd_mbox_query_fw_utc_sec_offset_get(mbox);
+
+	if (mlxsw_cmd_mbox_query_fw_utc_nsec_bar_get(mbox) != 0) {
+		dev_err(&pdev->dev, "Unsupported UTC nsec BAR queried from hw\n");
+		err = -EINVAL;
+		goto err_utc_nsec_bar;
+	}
+
+	mlxsw_pci->utc_nsec_offset =
+		mlxsw_cmd_mbox_query_fw_utc_nsec_offset_get(mbox);
+
+	num_pages = mlxsw_cmd_mbox_query_fw_fw_pages_get(mbox);
+	err = mlxsw_pci_fw_area_init(mlxsw_pci, mbox, num_pages);
+	if (err)
+		goto err_fw_area_init;
+
+	err = mlxsw_pci_boardinfo(mlxsw_pci, mbox);
+	if (err)
+		goto err_boardinfo;
+
+	err = mlxsw_core_resources_query(mlxsw_core, mbox, res);
+	if (err)
+		goto err_query_resources;
+
+	if (MLXSW_CORE_RES_VALID(mlxsw_core, CQE_V2) &&
+	    MLXSW_CORE_RES_GET(mlxsw_core, CQE_V2))
+		mlxsw_pci->max_cqe_ver = MLXSW_PCI_CQE_V2;
+	else if (MLXSW_CORE_RES_VALID(mlxsw_core, CQE_V1) &&
+		 MLXSW_CORE_RES_GET(mlxsw_core, CQE_V1))
+		mlxsw_pci->max_cqe_ver = MLXSW_PCI_CQE_V1;
+	else if ((MLXSW_CORE_RES_VALID(mlxsw_core, CQE_V0) &&
+		  MLXSW_CORE_RES_GET(mlxsw_core, CQE_V0)) ||
+		 !MLXSW_CORE_RES_VALID(mlxsw_core, CQE_V0)) {
+		mlxsw_pci->max_cqe_ver = MLXSW_PCI_CQE_V0;
+	} else {
+		dev_err(&pdev->dev, "Invalid supported CQE version combination reported\n");
+		goto err_cqe_v_check;
+	}
+
+	err = mlxsw_pci_config_profile(mlxsw_pci, mbox, profile, res);
+	if (err)
+		goto err_config_profile;
+
+	/* Some resources depend on unified bridge model, which is configured
+	 * as part of config_profile. Query the resources again to get correct
+	 * values.
+	 */
+	err = mlxsw_core_resources_query(mlxsw_core, mbox, res);
+	if (err)
+		goto err_requery_resources;
+
+	err = mlxsw_pci_aqs_init(mlxsw_pci, mbox);
+	if (err)
+		goto err_aqs_init;
+
+	err = request_irq(pci_irq_vector(pdev, 0),
+			  mlxsw_pci_eq_irq_handler, 0,
+			  mlxsw_pci->bus_info.device_kind, mlxsw_pci);
+	if (err) {
+		dev_err(&pdev->dev, "IRQ request failed\n");
+		goto err_request_eq_irq;
+	}
+
+	goto mbox_put;
+
+err_request_eq_irq:
+	mlxsw_pci_aqs_fini(mlxsw_pci);
+err_aqs_init:
+err_requery_resources:
+err_config_profile:
+err_cqe_v_check:
+err_query_resources:
+err_boardinfo:
+	mlxsw_pci_fw_area_fini(mlxsw_pci);
+err_fw_area_init:
+err_utc_nsec_bar:
+err_utc_sec_bar:
+err_fr_rn_clk_bar:
+err_doorbell_page_bar:
+err_iface_rev:
+err_query_fw:
+	mlxsw_pci_free_irq_vectors(mlxsw_pci);
+err_alloc_irq:
+err_sw_reset:
+mbox_put:
+	mlxsw_cmd_mbox_free(mbox);
+	return err;
+}
+
+static void mlxsw_pci_fini(void *bus_priv)
+{
+	struct mlxsw_pci *mlxsw_pci = bus_priv;
+
+	free_irq(pci_irq_vector(mlxsw_pci->pdev, 0), mlxsw_pci);
+	mlxsw_pci_aqs_fini(mlxsw_pci);
+	mlxsw_pci_fw_area_fini(mlxsw_pci);
+	mlxsw_pci_free_irq_vectors(mlxsw_pci);
+}
+
+static struct mlxsw_pci_queue *
+mlxsw_pci_sdq_pick(struct mlxsw_pci *mlxsw_pci,
+		   const struct mlxsw_tx_info *tx_info)
+{
+	u8 ctl_sdq_count = mlxsw_pci_sdq_count(mlxsw_pci) - 1;
+	u8 sdqn;
+
+	if (tx_info->is_emad) {
+		sdqn = MLXSW_PCI_SDQ_EMAD_INDEX;
+	} else {
+		BUILD_BUG_ON(MLXSW_PCI_SDQ_EMAD_INDEX != 0);
+		sdqn = 1 + (tx_info->local_port % ctl_sdq_count);
+	}
+
+	return mlxsw_pci_sdq_get(mlxsw_pci, sdqn);
+}
+
+static bool mlxsw_pci_skb_transmit_busy(void *bus_priv,
+					const struct mlxsw_tx_info *tx_info)
+{
+	struct mlxsw_pci *mlxsw_pci = bus_priv;
+	struct mlxsw_pci_queue *q = mlxsw_pci_sdq_pick(mlxsw_pci, tx_info);
+
+	return !mlxsw_pci_queue_elem_info_producer_get(q);
+}
+
+static int mlxsw_pci_skb_transmit(void *bus_priv, struct sk_buff *skb,
+				  const struct mlxsw_tx_info *tx_info)
+{
+	struct mlxsw_pci *mlxsw_pci = bus_priv;
+	struct mlxsw_pci_queue *q;
+	struct mlxsw_pci_queue_elem_info *elem_info;
+	char *wqe;
+	int i;
+	int err;
+
+	if (skb_shinfo(skb)->nr_frags > MLXSW_PCI_WQE_SG_ENTRIES - 1) {
+		err = skb_linearize(skb);
+		if (err)
+			return err;
+	}
+
+	q = mlxsw_pci_sdq_pick(mlxsw_pci, tx_info);
+	spin_lock_bh(&q->lock);
+	elem_info = mlxsw_pci_queue_elem_info_producer_get(q);
+	if (!elem_info) {
+		/* queue is full */
+		err = -EAGAIN;
+		goto unlock;
+	}
+	mlxsw_skb_cb(skb)->tx_info = *tx_info;
+	elem_info->u.sdq.skb = skb;
+
+	wqe = elem_info->elem;
+	mlxsw_pci_wqe_c_set(wqe, 1); /* always report completion */
+	mlxsw_pci_wqe_lp_set(wqe, 0);
+	mlxsw_pci_wqe_type_set(wqe, MLXSW_PCI_WQE_TYPE_ETHERNET);
+
+	err = mlxsw_pci_wqe_frag_map(mlxsw_pci, wqe, 0, skb->data,
+				     skb_headlen(skb), DMA_TO_DEVICE);
+	if (err)
+		goto unlock;
+
+	for (i = 0; i < skb_shinfo(skb)->nr_frags; i++) {
+		const skb_frag_t *frag = &skb_shinfo(skb)->frags[i];
+
+		err = mlxsw_pci_wqe_frag_map(mlxsw_pci, wqe, i + 1,
+					     skb_frag_address(frag),
+					     skb_frag_size(frag),
+					     DMA_TO_DEVICE);
+		if (err)
+			goto unmap_frags;
+	}
+
+	if (unlikely(skb_shinfo(skb)->tx_flags & SKBTX_HW_TSTAMP))
+		skb_shinfo(skb)->tx_flags |= SKBTX_IN_PROGRESS;
+
+	/* Set unused sq entries byte count to zero. */
+	for (i++; i < MLXSW_PCI_WQE_SG_ENTRIES; i++)
+		mlxsw_pci_wqe_byte_count_set(wqe, i, 0);
+
+	/* Everything is set up, ring producer doorbell to get HW going */
+	q->producer_counter++;
+	mlxsw_pci_queue_doorbell_producer_ring(mlxsw_pci, q);
+
+	goto unlock;
+
+unmap_frags:
+	for (; i >= 0; i--)
+		mlxsw_pci_wqe_frag_unmap(mlxsw_pci, wqe, i, DMA_TO_DEVICE);
+unlock:
+	spin_unlock_bh(&q->lock);
+	return err;
+}
+
+static int mlxsw_pci_cmd_exec(void *bus_priv, u16 opcode, u8 opcode_mod,
+			      u32 in_mod, bool out_mbox_direct,
+			      char *in_mbox, size_t in_mbox_size,
+			      char *out_mbox, size_t out_mbox_size,
+			      u8 *p_status)
+{
+	struct mlxsw_pci *mlxsw_pci = bus_priv;
+	dma_addr_t in_mapaddr = 0, out_mapaddr = 0;
+	bool evreq = mlxsw_pci->cmd.nopoll;
+	unsigned long timeout = msecs_to_jiffies(MLXSW_PCI_CIR_TIMEOUT_MSECS);
+	bool *p_wait_done = &mlxsw_pci->cmd.wait_done;
+	int err;
+
+	*p_status = MLXSW_CMD_STATUS_OK;
+
+	err = mutex_lock_interruptible(&mlxsw_pci->cmd.lock);
+	if (err)
+		return err;
+
+	if (in_mbox) {
+		memcpy(mlxsw_pci->cmd.in_mbox.buf, in_mbox, in_mbox_size);
+		in_mapaddr = mlxsw_pci->cmd.in_mbox.mapaddr;
+	}
+	mlxsw_pci_write32(mlxsw_pci, CIR_IN_PARAM_HI, upper_32_bits(in_mapaddr));
+	mlxsw_pci_write32(mlxsw_pci, CIR_IN_PARAM_LO, lower_32_bits(in_mapaddr));
+
+	if (out_mbox)
+		out_mapaddr = mlxsw_pci->cmd.out_mbox.mapaddr;
+	mlxsw_pci_write32(mlxsw_pci, CIR_OUT_PARAM_HI, upper_32_bits(out_mapaddr));
+	mlxsw_pci_write32(mlxsw_pci, CIR_OUT_PARAM_LO, lower_32_bits(out_mapaddr));
+
+	mlxsw_pci_write32(mlxsw_pci, CIR_IN_MODIFIER, in_mod);
+	mlxsw_pci_write32(mlxsw_pci, CIR_TOKEN, 0);
+
+	*p_wait_done = false;
+
+	wmb(); /* all needs to be written before we write control register */
+	mlxsw_pci_write32(mlxsw_pci, CIR_CTRL,
+			  MLXSW_PCI_CIR_CTRL_GO_BIT |
+			  (evreq ? MLXSW_PCI_CIR_CTRL_EVREQ_BIT : 0) |
+			  (opcode_mod << MLXSW_PCI_CIR_CTRL_OPCODE_MOD_SHIFT) |
+			  opcode);
+
+	if (!evreq) {
+		unsigned long end;
+
+		end = jiffies + timeout;
+		do {
+			u32 ctrl = mlxsw_pci_read32(mlxsw_pci, CIR_CTRL);
+
+			if (!(ctrl & MLXSW_PCI_CIR_CTRL_GO_BIT)) {
+				*p_wait_done = true;
+				*p_status = ctrl >> MLXSW_PCI_CIR_CTRL_STATUS_SHIFT;
+				break;
+			}
+			cond_resched();
+		} while (time_before(jiffies, end));
+	} else {
+		wait_event_timeout(mlxsw_pci->cmd.wait, *p_wait_done, timeout);
+		*p_status = mlxsw_pci->cmd.comp.status;
+	}
+
+	err = 0;
+	if (*p_wait_done) {
+		if (*p_status)
+			err = -EIO;
+	} else {
+		err = -ETIMEDOUT;
+	}
+
+	if (!err && out_mbox && out_mbox_direct) {
+		/* Some commands don't use output param as address to mailbox
+		 * but they store output directly into registers. In that case,
+		 * copy registers into mbox buffer.
+		 */
+		__be32 tmp;
+
+		if (!evreq) {
+			tmp = cpu_to_be32(mlxsw_pci_read32(mlxsw_pci,
+							   CIR_OUT_PARAM_HI));
+			memcpy(out_mbox, &tmp, sizeof(tmp));
+			tmp = cpu_to_be32(mlxsw_pci_read32(mlxsw_pci,
+							   CIR_OUT_PARAM_LO));
+			memcpy(out_mbox + sizeof(tmp), &tmp, sizeof(tmp));
+		}
+	} else if (!err && out_mbox) {
+		memcpy(out_mbox, mlxsw_pci->cmd.out_mbox.buf, out_mbox_size);
+	}
+
+	mutex_unlock(&mlxsw_pci->cmd.lock);
+
+	return err;
+}
+
+static u32 mlxsw_pci_read_frc_h(void *bus_priv)
+{
+	struct mlxsw_pci *mlxsw_pci = bus_priv;
+	u64 frc_offset_h;
+
+	frc_offset_h = mlxsw_pci->free_running_clock_offset;
+	return mlxsw_pci_read32_off(mlxsw_pci, frc_offset_h);
+}
+
+static u32 mlxsw_pci_read_frc_l(void *bus_priv)
+{
+	struct mlxsw_pci *mlxsw_pci = bus_priv;
+	u64 frc_offset_l;
+
+	frc_offset_l = mlxsw_pci->free_running_clock_offset + 4;
+	return mlxsw_pci_read32_off(mlxsw_pci, frc_offset_l);
+}
+
+static u32 mlxsw_pci_read_utc_sec(void *bus_priv)
+{
+	struct mlxsw_pci *mlxsw_pci = bus_priv;
+
+	return mlxsw_pci_read32_off(mlxsw_pci, mlxsw_pci->utc_sec_offset);
+}
+
+static u32 mlxsw_pci_read_utc_nsec(void *bus_priv)
+{
+	struct mlxsw_pci *mlxsw_pci = bus_priv;
+
+	return mlxsw_pci_read32_off(mlxsw_pci, mlxsw_pci->utc_nsec_offset);
+}
+
+static const struct mlxsw_bus mlxsw_pci_bus = {
+	.kind			= "pci",
+	.init			= mlxsw_pci_init,
+	.fini			= mlxsw_pci_fini,
+	.skb_transmit_busy	= mlxsw_pci_skb_transmit_busy,
+	.skb_transmit		= mlxsw_pci_skb_transmit,
+	.cmd_exec		= mlxsw_pci_cmd_exec,
+	.read_frc_h		= mlxsw_pci_read_frc_h,
+	.read_frc_l		= mlxsw_pci_read_frc_l,
+	.read_utc_sec		= mlxsw_pci_read_utc_sec,
+	.read_utc_nsec		= mlxsw_pci_read_utc_nsec,
+	.features		= MLXSW_BUS_F_TXRX | MLXSW_BUS_F_RESET,
+};
+
+static int mlxsw_pci_cmd_init(struct mlxsw_pci *mlxsw_pci)
+{
+	int err;
+
+	mutex_init(&mlxsw_pci->cmd.lock);
+	init_waitqueue_head(&mlxsw_pci->cmd.wait);
+
+	err = mlxsw_pci_mbox_alloc(mlxsw_pci, &mlxsw_pci->cmd.in_mbox);
+	if (err)
+		goto err_in_mbox_alloc;
+
+	err = mlxsw_pci_mbox_alloc(mlxsw_pci, &mlxsw_pci->cmd.out_mbox);
+	if (err)
+		goto err_out_mbox_alloc;
+
+	return 0;
+
+err_out_mbox_alloc:
+	mlxsw_pci_mbox_free(mlxsw_pci, &mlxsw_pci->cmd.in_mbox);
+err_in_mbox_alloc:
+	mutex_destroy(&mlxsw_pci->cmd.lock);
+	return err;
+}
+
+static void mlxsw_pci_cmd_fini(struct mlxsw_pci *mlxsw_pci)
+{
+	mlxsw_pci_mbox_free(mlxsw_pci, &mlxsw_pci->cmd.out_mbox);
+	mlxsw_pci_mbox_free(mlxsw_pci, &mlxsw_pci->cmd.in_mbox);
+	mutex_destroy(&mlxsw_pci->cmd.lock);
+}
+
+static int mlxsw_pci_probe(struct pci_dev *pdev, const struct pci_device_id *id)
+{
+	const char *driver_name = dev_driver_string(&pdev->dev);
+	struct mlxsw_pci *mlxsw_pci;
+	int err;
+
+	mlxsw_pci = kzalloc(sizeof(*mlxsw_pci), GFP_KERNEL);
+	if (!mlxsw_pci)
+		return -ENOMEM;
+
+	err = pci_enable_device(pdev);
+	if (err) {
+		dev_err(&pdev->dev, "pci_enable_device failed\n");
+		goto err_pci_enable_device;
+	}
+
+	err = pci_request_regions(pdev, driver_name);
+	if (err) {
+		dev_err(&pdev->dev, "pci_request_regions failed\n");
+		goto err_pci_request_regions;
+	}
+
+	err = dma_set_mask_and_coherent(&pdev->dev, DMA_BIT_MASK(64));
+	if (err) {
+		err = dma_set_mask(&pdev->dev, DMA_BIT_MASK(32));
+		if (err) {
+			dev_err(&pdev->dev, "dma_set_mask failed\n");
+			goto err_pci_set_dma_mask;
+		}
+	}
+
+	if (pci_resource_len(pdev, 0) < MLXSW_PCI_BAR0_SIZE) {
+		dev_err(&pdev->dev, "invalid PCI region size\n");
+		err = -EINVAL;
+		goto err_pci_resource_len_check;
+	}
+
+	mlxsw_pci->hw_addr = ioremap(pci_resource_start(pdev, 0),
+				     pci_resource_len(pdev, 0));
+	if (!mlxsw_pci->hw_addr) {
+		dev_err(&pdev->dev, "ioremap failed\n");
+		err = -EIO;
+		goto err_ioremap;
+	}
+	pci_set_master(pdev);
+
+	mlxsw_pci->pdev = pdev;
+	pci_set_drvdata(pdev, mlxsw_pci);
+
+	err = mlxsw_pci_cmd_init(mlxsw_pci);
+	if (err)
+		goto err_pci_cmd_init;
+
+	mlxsw_pci->bus_info.device_kind = driver_name;
+	mlxsw_pci->bus_info.device_name = pci_name(mlxsw_pci->pdev);
+	mlxsw_pci->bus_info.dev = &pdev->dev;
+	mlxsw_pci->bus_info.read_clock_capable = true;
+	mlxsw_pci->id = id;
+
+	err = mlxsw_core_bus_device_register(&mlxsw_pci->bus_info,
+					     &mlxsw_pci_bus, mlxsw_pci, false,
+					     NULL, NULL);
+	if (err) {
+		dev_err(&pdev->dev, "cannot register bus device\n");
+		goto err_bus_device_register;
+	}
+
+	return 0;
+
+err_bus_device_register:
+	mlxsw_pci_cmd_fini(mlxsw_pci);
+err_pci_cmd_init:
+	iounmap(mlxsw_pci->hw_addr);
+err_ioremap:
+err_pci_resource_len_check:
+err_pci_set_dma_mask:
+	pci_release_regions(pdev);
+err_pci_request_regions:
+	pci_disable_device(pdev);
+err_pci_enable_device:
+	kfree(mlxsw_pci);
+	return err;
+}
+
+static void mlxsw_pci_remove(struct pci_dev *pdev)
+{
+	struct mlxsw_pci *mlxsw_pci = pci_get_drvdata(pdev);
+
+	mlxsw_core_bus_device_unregister(mlxsw_pci->core, false);
+	mlxsw_pci_cmd_fini(mlxsw_pci);
+	iounmap(mlxsw_pci->hw_addr);
+	pci_release_regions(mlxsw_pci->pdev);
+	pci_disable_device(mlxsw_pci->pdev);
+	kfree(mlxsw_pci);
+}
+
+int mlxsw_pci_driver_register(struct pci_driver *pci_driver)
+{
+	pci_driver->probe = mlxsw_pci_probe;
+	pci_driver->remove = mlxsw_pci_remove;
+	pci_driver->shutdown = mlxsw_pci_remove;
+	return pci_register_driver(pci_driver);
+}
+EXPORT_SYMBOL(mlxsw_pci_driver_register);
+
+void mlxsw_pci_driver_unregister(struct pci_driver *pci_driver)
+{
+	pci_unregister_driver(pci_driver);
+}
+EXPORT_SYMBOL(mlxsw_pci_driver_unregister);
+
+static int __init mlxsw_pci_module_init(void)
+{
+	return 0;
+}
+
+static void __exit mlxsw_pci_module_exit(void)
+{
+}
+
+module_init(mlxsw_pci_module_init);
+module_exit(mlxsw_pci_module_exit);
+
+MODULE_LICENSE("Dual BSD/GPL");
+MODULE_AUTHOR("Jiri Pirko <jiri@mellanox.com>");
+MODULE_DESCRIPTION("Mellanox switch PCI interface driver");
diff --git a/drivers/net/ethernet/mellanox/mlxsw/pci.h b/drivers/net/ethernet/mellanox/mlxsw/pci.h
new file mode 100644
index 000000000..cacc2f9fa
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/pci.h
@@ -0,0 +1,34 @@
+/* SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0 */
+/* Copyright (c) 2016-2018 Mellanox Technologies. All rights reserved */
+
+#ifndef _MLXSW_PCI_H
+#define _MLXSW_PCI_H
+
+#include <linux/pci.h>
+
+#define PCI_DEVICE_ID_MELLANOX_SPECTRUM		0xcb84
+#define PCI_DEVICE_ID_MELLANOX_SPECTRUM2	0xcf6c
+#define PCI_DEVICE_ID_MELLANOX_SPECTRUM3	0xcf70
+#define PCI_DEVICE_ID_MELLANOX_SPECTRUM4	0xcf80
+
+#if IS_ENABLED(CONFIG_MLXSW_PCI)
+
+int mlxsw_pci_driver_register(struct pci_driver *pci_driver);
+void mlxsw_pci_driver_unregister(struct pci_driver *pci_driver);
+
+#else
+
+static inline int
+mlxsw_pci_driver_register(struct pci_driver *pci_driver)
+{
+	return 0;
+}
+
+static inline void
+mlxsw_pci_driver_unregister(struct pci_driver *pci_driver)
+{
+}
+
+#endif
+
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlxsw/pci_hw.h b/drivers/net/ethernet/mellanox/mlxsw/pci_hw.h
new file mode 100644
index 000000000..7cdf0ce24
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/pci_hw.h
@@ -0,0 +1,416 @@
+/* SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0 */
+/* Copyright (c) 2015-2018 Mellanox Technologies. All rights reserved */
+
+#ifndef _MLXSW_PCI_HW_H
+#define _MLXSW_PCI_HW_H
+
+#include <linux/bitops.h>
+
+#include "item.h"
+
+#define MLXSW_PCI_BAR0_SIZE		(1024 * 1024) /* 1MB */
+#define MLXSW_PCI_PAGE_SIZE		4096
+
+#define MLXSW_PCI_CIR_BASE			0x71000
+#define MLXSW_PCI_CIR_IN_PARAM_HI		MLXSW_PCI_CIR_BASE
+#define MLXSW_PCI_CIR_IN_PARAM_LO		(MLXSW_PCI_CIR_BASE + 0x04)
+#define MLXSW_PCI_CIR_IN_MODIFIER		(MLXSW_PCI_CIR_BASE + 0x08)
+#define MLXSW_PCI_CIR_OUT_PARAM_HI		(MLXSW_PCI_CIR_BASE + 0x0C)
+#define MLXSW_PCI_CIR_OUT_PARAM_LO		(MLXSW_PCI_CIR_BASE + 0x10)
+#define MLXSW_PCI_CIR_TOKEN			(MLXSW_PCI_CIR_BASE + 0x14)
+#define MLXSW_PCI_CIR_CTRL			(MLXSW_PCI_CIR_BASE + 0x18)
+#define MLXSW_PCI_CIR_CTRL_GO_BIT		BIT(23)
+#define MLXSW_PCI_CIR_CTRL_EVREQ_BIT		BIT(22)
+#define MLXSW_PCI_CIR_CTRL_OPCODE_MOD_SHIFT	12
+#define MLXSW_PCI_CIR_CTRL_STATUS_SHIFT		24
+#define MLXSW_PCI_CIR_TIMEOUT_MSECS		1000
+
+#define MLXSW_PCI_SW_RESET_TIMEOUT_MSECS	900000
+#define MLXSW_PCI_SW_RESET_WAIT_MSECS		400
+#define MLXSW_PCI_FW_READY			0xA1844
+#define MLXSW_PCI_FW_READY_MASK			0xFFFF
+#define MLXSW_PCI_FW_READY_MAGIC		0x5E
+
+#define MLXSW_PCI_DOORBELL_SDQ_OFFSET		0x000
+#define MLXSW_PCI_DOORBELL_RDQ_OFFSET		0x200
+#define MLXSW_PCI_DOORBELL_CQ_OFFSET		0x400
+#define MLXSW_PCI_DOORBELL_EQ_OFFSET		0x600
+#define MLXSW_PCI_DOORBELL_ARM_CQ_OFFSET	0x800
+#define MLXSW_PCI_DOORBELL_ARM_EQ_OFFSET	0xA00
+
+#define MLXSW_PCI_DOORBELL(offset, type_offset, num)	\
+	((offset) + (type_offset) + (num) * 4)
+
+#define MLXSW_PCI_CQS_MAX	96
+#define MLXSW_PCI_EQS_COUNT	2
+#define MLXSW_PCI_EQ_ASYNC_NUM	0
+#define MLXSW_PCI_EQ_COMP_NUM	1
+
+#define MLXSW_PCI_SDQS_MIN	2 /* EMAD and control traffic */
+#define MLXSW_PCI_SDQ_EMAD_INDEX	0
+#define MLXSW_PCI_SDQ_EMAD_TC	0
+#define MLXSW_PCI_SDQ_CTL_TC	3
+
+#define MLXSW_PCI_AQ_PAGES	8
+#define MLXSW_PCI_AQ_SIZE	(MLXSW_PCI_PAGE_SIZE * MLXSW_PCI_AQ_PAGES)
+#define MLXSW_PCI_WQE_SIZE	32 /* 32 bytes per element */
+#define MLXSW_PCI_CQE01_SIZE	16 /* 16 bytes per element */
+#define MLXSW_PCI_CQE2_SIZE	32 /* 32 bytes per element */
+#define MLXSW_PCI_CQE_SIZE_MAX	MLXSW_PCI_CQE2_SIZE
+#define MLXSW_PCI_EQE_SIZE	16 /* 16 bytes per element */
+#define MLXSW_PCI_WQE_COUNT	(MLXSW_PCI_AQ_SIZE / MLXSW_PCI_WQE_SIZE)
+#define MLXSW_PCI_CQE01_COUNT	(MLXSW_PCI_AQ_SIZE / MLXSW_PCI_CQE01_SIZE)
+#define MLXSW_PCI_CQE2_COUNT	(MLXSW_PCI_AQ_SIZE / MLXSW_PCI_CQE2_SIZE)
+#define MLXSW_PCI_EQE_COUNT	(MLXSW_PCI_AQ_SIZE / MLXSW_PCI_EQE_SIZE)
+#define MLXSW_PCI_EQE_UPDATE_COUNT	0x80
+
+#define MLXSW_PCI_WQE_SG_ENTRIES	3
+#define MLXSW_PCI_WQE_TYPE_ETHERNET	0xA
+
+/* pci_wqe_c
+ * If set it indicates that a completion should be reported upon
+ * execution of this descriptor.
+ */
+MLXSW_ITEM32(pci, wqe, c, 0x00, 31, 1);
+
+/* pci_wqe_lp
+ * Local Processing, set if packet should be processed by the local
+ * switch hardware:
+ * For Ethernet EMAD (Direct Route and non Direct Route) -
+ * must be set if packet destination is local device
+ * For InfiniBand CTL - must be set if packet destination is local device
+ * Otherwise it must be clear
+ * Local Process packets must not exceed the size of 2K (including payload
+ * and headers).
+ */
+MLXSW_ITEM32(pci, wqe, lp, 0x00, 30, 1);
+
+/* pci_wqe_type
+ * Packet type.
+ */
+MLXSW_ITEM32(pci, wqe, type, 0x00, 23, 4);
+
+/* pci_wqe_byte_count
+ * Size of i-th scatter/gather entry, 0 if entry is unused.
+ */
+MLXSW_ITEM16_INDEXED(pci, wqe, byte_count, 0x02, 0, 14, 0x02, 0x00, false);
+
+/* pci_wqe_address
+ * Physical address of i-th scatter/gather entry.
+ * Gather Entries must be 2Byte aligned.
+ */
+MLXSW_ITEM64_INDEXED(pci, wqe, address, 0x08, 0, 64, 0x8, 0x0, false);
+
+enum mlxsw_pci_cqe_v {
+	MLXSW_PCI_CQE_V0,
+	MLXSW_PCI_CQE_V1,
+	MLXSW_PCI_CQE_V2,
+};
+
+#define mlxsw_pci_cqe_item_helpers(name, v0, v1, v2)				\
+static inline u32 mlxsw_pci_cqe_##name##_get(enum mlxsw_pci_cqe_v v, char *cqe)	\
+{										\
+	switch (v) {								\
+	default:								\
+	case MLXSW_PCI_CQE_V0:							\
+		return mlxsw_pci_cqe##v0##_##name##_get(cqe);			\
+	case MLXSW_PCI_CQE_V1:							\
+		return mlxsw_pci_cqe##v1##_##name##_get(cqe);			\
+	case MLXSW_PCI_CQE_V2:							\
+		return mlxsw_pci_cqe##v2##_##name##_get(cqe);			\
+	}									\
+}										\
+static inline void mlxsw_pci_cqe_##name##_set(enum mlxsw_pci_cqe_v v,		\
+					      char *cqe, u32 val)		\
+{										\
+	switch (v) {								\
+	default:								\
+	case MLXSW_PCI_CQE_V0:							\
+		mlxsw_pci_cqe##v0##_##name##_set(cqe, val);			\
+		break;								\
+	case MLXSW_PCI_CQE_V1:							\
+		mlxsw_pci_cqe##v1##_##name##_set(cqe, val);			\
+		break;								\
+	case MLXSW_PCI_CQE_V2:							\
+		mlxsw_pci_cqe##v2##_##name##_set(cqe, val);			\
+		break;								\
+	}									\
+}
+
+/* pci_cqe_lag
+ * Packet arrives from a port which is a LAG
+ */
+MLXSW_ITEM32(pci, cqe0, lag, 0x00, 23, 1);
+MLXSW_ITEM32(pci, cqe12, lag, 0x00, 24, 1);
+mlxsw_pci_cqe_item_helpers(lag, 0, 12, 12);
+
+/* pci_cqe_system_port/lag_id
+ * When lag=0: System port on which the packet was received
+ * When lag=1:
+ * bits [15:4] LAG ID on which the packet was received
+ * bits [3:0] sub_port on which the packet was received
+ */
+MLXSW_ITEM32(pci, cqe, system_port, 0x00, 0, 16);
+MLXSW_ITEM32(pci, cqe0, lag_id, 0x00, 4, 12);
+MLXSW_ITEM32(pci, cqe12, lag_id, 0x00, 0, 16);
+mlxsw_pci_cqe_item_helpers(lag_id, 0, 12, 12);
+MLXSW_ITEM32(pci, cqe0, lag_subport, 0x00, 0, 4);
+MLXSW_ITEM32(pci, cqe12, lag_subport, 0x00, 16, 8);
+mlxsw_pci_cqe_item_helpers(lag_subport, 0, 12, 12);
+
+/* pci_cqe_wqe_counter
+ * WQE count of the WQEs completed on the associated dqn
+ */
+MLXSW_ITEM32(pci, cqe, wqe_counter, 0x04, 16, 16);
+
+/* pci_cqe_byte_count
+ * Byte count of received packets including additional two
+ * Reserved Bytes that are append to the end of the frame.
+ * Reserved for Send CQE.
+ */
+MLXSW_ITEM32(pci, cqe, byte_count, 0x04, 0, 14);
+
+#define MLXSW_PCI_CQE2_MIRROR_CONG_INVALID	0xFFFF
+
+/* pci_cqe_mirror_cong_high
+ * Congestion level in units of 8KB of the egress traffic class of the original
+ * packet that does mirroring to the CPU. Value of 0xFFFF means that the
+ * congestion level is invalid.
+ */
+MLXSW_ITEM32(pci, cqe2, mirror_cong_high, 0x08, 16, 4);
+
+/* pci_cqe_trap_id
+ * Trap ID that captured the packet.
+ */
+MLXSW_ITEM32(pci, cqe, trap_id, 0x08, 0, 10);
+
+/* pci_cqe_crc
+ * Length include CRC. Indicates the length field includes
+ * the packet's CRC.
+ */
+MLXSW_ITEM32(pci, cqe0, crc, 0x0C, 8, 1);
+MLXSW_ITEM32(pci, cqe12, crc, 0x0C, 9, 1);
+mlxsw_pci_cqe_item_helpers(crc, 0, 12, 12);
+
+/* pci_cqe_e
+ * CQE with Error.
+ */
+MLXSW_ITEM32(pci, cqe0, e, 0x0C, 7, 1);
+MLXSW_ITEM32(pci, cqe12, e, 0x00, 27, 1);
+mlxsw_pci_cqe_item_helpers(e, 0, 12, 12);
+
+/* pci_cqe_sr
+ * 1 - Send Queue
+ * 0 - Receive Queue
+ */
+MLXSW_ITEM32(pci, cqe0, sr, 0x0C, 6, 1);
+MLXSW_ITEM32(pci, cqe12, sr, 0x00, 26, 1);
+mlxsw_pci_cqe_item_helpers(sr, 0, 12, 12);
+
+/* pci_cqe_dqn
+ * Descriptor Queue (DQ) Number.
+ */
+MLXSW_ITEM32(pci, cqe0, dqn, 0x0C, 1, 5);
+MLXSW_ITEM32(pci, cqe12, dqn, 0x0C, 1, 6);
+mlxsw_pci_cqe_item_helpers(dqn, 0, 12, 12);
+
+/* pci_cqe_time_stamp_low
+ * Time stamp of the CQE
+ * Format according to time_stamp_type:
+ * 0: uSec - 1.024uSec (default for devices which do not support
+ * time_stamp_type). Only bits 15:0 are valid
+ * 1: FRC - Free Running Clock - units of 1nSec
+ * 2: UTC - time_stamp[37:30] = Sec
+ *	  - time_stamp[29:0] = nSec
+ * 3: Mirror_UTC. UTC time stamp of the original packet that has
+ * MIRROR_SESSION traps
+ *   - time_stamp[37:30] = Sec
+ *   - time_stamp[29:0] = nSec
+ *   Formats 0..2 are configured by
+ *   CONFIG_PROFILE.cqe_time_stamp_type for PTP traps
+ *   Format 3 is used for MIRROR_SESSION traps
+ *   Note that Spectrum does not reveal FRC, UTC and Mirror_UTC
+ */
+MLXSW_ITEM32(pci, cqe2, time_stamp_low, 0x0C, 16, 16);
+
+#define MLXSW_PCI_CQE2_MIRROR_TCLASS_INVALID	0x1F
+
+/* pci_cqe_mirror_tclass
+ * The egress traffic class of the original packet that does mirroring to the
+ * CPU. Value of 0x1F means that the traffic class is invalid.
+ */
+MLXSW_ITEM32(pci, cqe2, mirror_tclass, 0x10, 27, 5);
+
+/* pci_cqe_tx_lag
+ * The Tx port of a packet that is mirrored / sampled to the CPU is a LAG.
+ */
+MLXSW_ITEM32(pci, cqe2, tx_lag, 0x10, 24, 1);
+
+/* pci_cqe_tx_lag_subport
+ * The port index within the LAG of a packet that is mirrored / sampled to the
+ * CPU. Reserved when tx_lag is 0.
+ */
+MLXSW_ITEM32(pci, cqe2, tx_lag_subport, 0x10, 16, 8);
+
+#define MLXSW_PCI_CQE2_TX_PORT_MULTI_PORT	0xFFFE
+#define MLXSW_PCI_CQE2_TX_PORT_INVALID		0xFFFF
+
+/* pci_cqe_tx_lag_id
+ * The Tx LAG ID of the original packet that is mirrored / sampled to the CPU.
+ * Value of 0xFFFE means multi-port. Value fo 0xFFFF means that the Tx LAG ID
+ * is invalid. Reserved when tx_lag is 0.
+ */
+MLXSW_ITEM32(pci, cqe2, tx_lag_id, 0x10, 0, 16);
+
+/* pci_cqe_tx_system_port
+ * The Tx port of the original packet that is mirrored / sampled to the CPU.
+ * Value of 0xFFFE means multi-port. Value fo 0xFFFF means that the Tx port is
+ * invalid. Reserved when tx_lag is 1.
+ */
+MLXSW_ITEM32(pci, cqe2, tx_system_port, 0x10, 0, 16);
+
+/* pci_cqe_mirror_cong_low
+ * Congestion level in units of 8KB of the egress traffic class of the original
+ * packet that does mirroring to the CPU. Value of 0xFFFF means that the
+ * congestion level is invalid.
+ */
+MLXSW_ITEM32(pci, cqe2, mirror_cong_low, 0x14, 20, 12);
+
+#define MLXSW_PCI_CQE2_MIRROR_CONG_SHIFT	13	/* Units of 8KB. */
+
+static inline u16 mlxsw_pci_cqe2_mirror_cong_get(const char *cqe)
+{
+	u16 cong_high = mlxsw_pci_cqe2_mirror_cong_high_get(cqe);
+	u16 cong_low = mlxsw_pci_cqe2_mirror_cong_low_get(cqe);
+
+	return cong_high << 12 | cong_low;
+}
+
+/* pci_cqe_user_def_val_orig_pkt_len
+ * When trap_id is an ACL: User defined value from policy engine action.
+ */
+MLXSW_ITEM32(pci, cqe2, user_def_val_orig_pkt_len, 0x14, 0, 20);
+
+/* pci_cqe_mirror_reason
+ * Mirror reason.
+ */
+MLXSW_ITEM32(pci, cqe2, mirror_reason, 0x18, 24, 8);
+
+enum mlxsw_pci_cqe_time_stamp_type {
+	MLXSW_PCI_CQE_TIME_STAMP_TYPE_USEC,
+	MLXSW_PCI_CQE_TIME_STAMP_TYPE_FRC,
+	MLXSW_PCI_CQE_TIME_STAMP_TYPE_UTC,
+	MLXSW_PCI_CQE_TIME_STAMP_TYPE_MIRROR_UTC,
+};
+
+/* pci_cqe_time_stamp_type
+ * Time stamp type:
+ * 0: uSec - 1.024uSec (default for devices which do not support
+ * time_stamp_type)
+ * 1: FRC - Free Running Clock - units of 1nSec
+ * 2: UTC
+ * 3: Mirror_UTC. UTC time stamp of the original packet that has
+ * MIRROR_SESSION traps
+ */
+MLXSW_ITEM32(pci, cqe2, time_stamp_type, 0x18, 22, 2);
+
+#define MLXSW_PCI_CQE2_MIRROR_LATENCY_INVALID	0xFFFFFF
+
+/* pci_cqe_time_stamp_high
+ * Time stamp of the CQE
+ * Format according to time_stamp_type:
+ * 0: uSec - 1.024uSec (default for devices which do not support
+ * time_stamp_type). Only bits 15:0 are valid
+ * 1: FRC - Free Running Clock - units of 1nSec
+ * 2: UTC - time_stamp[37:30] = Sec
+ *	  - time_stamp[29:0] = nSec
+ * 3: Mirror_UTC. UTC time stamp of the original packet that has
+ * MIRROR_SESSION traps
+ *   - time_stamp[37:30] = Sec
+ *   - time_stamp[29:0] = nSec
+ *   Formats 0..2 are configured by
+ *   CONFIG_PROFILE.cqe_time_stamp_type for PTP traps
+ *   Format 3 is used for MIRROR_SESSION traps
+ *   Note that Spectrum does not reveal FRC, UTC and Mirror_UTC
+ */
+MLXSW_ITEM32(pci, cqe2, time_stamp_high, 0x18, 0, 22);
+
+static inline u64 mlxsw_pci_cqe2_time_stamp_get(const char *cqe)
+{
+	u64 ts_high = mlxsw_pci_cqe2_time_stamp_high_get(cqe);
+	u64 ts_low = mlxsw_pci_cqe2_time_stamp_low_get(cqe);
+
+	return ts_high << 16 | ts_low;
+}
+
+static inline u8 mlxsw_pci_cqe2_time_stamp_sec_get(const char *cqe)
+{
+	u64 full_ts = mlxsw_pci_cqe2_time_stamp_get(cqe);
+
+	return full_ts >> 30 & 0xFF;
+}
+
+static inline u32 mlxsw_pci_cqe2_time_stamp_nsec_get(const char *cqe)
+{
+	u64 full_ts = mlxsw_pci_cqe2_time_stamp_get(cqe);
+
+	return full_ts & 0x3FFFFFFF;
+}
+
+/* pci_cqe_mirror_latency
+ * End-to-end latency of the original packet that does mirroring to the CPU.
+ * Value of 0xFFFFFF means that the latency is invalid. Units are according to
+ * MOGCR.mirror_latency_units.
+ */
+MLXSW_ITEM32(pci, cqe2, mirror_latency, 0x1C, 8, 24);
+
+/* pci_cqe_owner
+ * Ownership bit.
+ */
+MLXSW_ITEM32(pci, cqe01, owner, 0x0C, 0, 1);
+MLXSW_ITEM32(pci, cqe2, owner, 0x1C, 0, 1);
+mlxsw_pci_cqe_item_helpers(owner, 01, 01, 2);
+
+/* pci_eqe_event_type
+ * Event type.
+ */
+MLXSW_ITEM32(pci, eqe, event_type, 0x0C, 24, 8);
+#define MLXSW_PCI_EQE_EVENT_TYPE_COMP	0x00
+#define MLXSW_PCI_EQE_EVENT_TYPE_CMD	0x0A
+
+/* pci_eqe_event_sub_type
+ * Event type.
+ */
+MLXSW_ITEM32(pci, eqe, event_sub_type, 0x0C, 16, 8);
+
+/* pci_eqe_cqn
+ * Completion Queue that triggered this EQE.
+ */
+MLXSW_ITEM32(pci, eqe, cqn, 0x0C, 8, 7);
+
+/* pci_eqe_owner
+ * Ownership bit.
+ */
+MLXSW_ITEM32(pci, eqe, owner, 0x0C, 0, 1);
+
+/* pci_eqe_cmd_token
+ * Command completion event - token
+ */
+MLXSW_ITEM32(pci, eqe, cmd_token, 0x00, 16, 16);
+
+/* pci_eqe_cmd_status
+ * Command completion event - status
+ */
+MLXSW_ITEM32(pci, eqe, cmd_status, 0x00, 0, 8);
+
+/* pci_eqe_cmd_out_param_h
+ * Command completion event - output parameter - higher part
+ */
+MLXSW_ITEM32(pci, eqe, cmd_out_param_h, 0x04, 0, 32);
+
+/* pci_eqe_cmd_out_param_l
+ * Command completion event - output parameter - lower part
+ */
+MLXSW_ITEM32(pci, eqe, cmd_out_param_l, 0x08, 0, 32);
+
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlxsw/port.h b/drivers/net/ethernet/mellanox/mlxsw/port.h
new file mode 100644
index 000000000..ac4d4ea51
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/port.h
@@ -0,0 +1,38 @@
+/* SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0 */
+/* Copyright (c) 2015-2018 Mellanox Technologies. All rights reserved */
+
+#ifndef _MLXSW_PORT_H
+#define _MLXSW_PORT_H
+
+#include <linux/types.h>
+
+#define MLXSW_PORT_MAX_MTU		10000
+
+#define MLXSW_PORT_DEFAULT_VID		1
+
+#define MLXSW_PORT_SWID_DISABLED_PORT	255
+#define MLXSW_PORT_SWID_ALL_SWIDS	254
+#define MLXSW_PORT_SWID_TYPE_IB		1
+#define MLXSW_PORT_SWID_TYPE_ETH	2
+
+#define MLXSW_PORT_MAX_IB_PHY_PORTS	36
+#define MLXSW_PORT_MAX_IB_PORTS		(MLXSW_PORT_MAX_IB_PHY_PORTS + 1)
+
+#define MLXSW_PORT_CPU_PORT		0x0
+
+#define MLXSW_PORT_DONT_CARE		0xFF
+
+enum mlxsw_port_admin_status {
+	MLXSW_PORT_ADMIN_STATUS_UP = 1,
+	MLXSW_PORT_ADMIN_STATUS_DOWN = 2,
+	MLXSW_PORT_ADMIN_STATUS_UP_ONCE = 3,
+	MLXSW_PORT_ADMIN_STATUS_DISABLED = 4,
+};
+
+enum mlxsw_reg_pude_oper_status {
+	MLXSW_PORT_OPER_STATUS_UP = 1,
+	MLXSW_PORT_OPER_STATUS_DOWN = 2,
+	MLXSW_PORT_OPER_STATUS_FAILURE = 4,	/* Can be set to up again. */
+};
+
+#endif /* _MLXSW_PORT_H */
diff --git a/drivers/net/ethernet/mellanox/mlxsw/reg.h b/drivers/net/ethernet/mellanox/mlxsw/reg.h
new file mode 100644
index 000000000..a34ff19c5
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/reg.h
@@ -0,0 +1,12937 @@
+/* SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0 */
+/* Copyright (c) 2015-2018 Mellanox Technologies. All rights reserved */
+
+#ifndef _MLXSW_REG_H
+#define _MLXSW_REG_H
+
+#include <linux/kernel.h>
+#include <linux/string.h>
+#include <linux/bitops.h>
+#include <linux/if_vlan.h>
+
+#include "item.h"
+#include "port.h"
+
+struct mlxsw_reg_info {
+	u16 id;
+	u16 len; /* In u8 */
+	const char *name;
+};
+
+#define MLXSW_REG_DEFINE(_name, _id, _len)				\
+static const struct mlxsw_reg_info mlxsw_reg_##_name = {		\
+	.id = _id,							\
+	.len = _len,							\
+	.name = #_name,							\
+}
+
+#define MLXSW_REG(type) (&mlxsw_reg_##type)
+#define MLXSW_REG_LEN(type) MLXSW_REG(type)->len
+#define MLXSW_REG_ZERO(type, payload) memset(payload, 0, MLXSW_REG(type)->len)
+
+/* SGCR - Switch General Configuration Register
+ * --------------------------------------------
+ * This register is used for configuration of the switch capabilities.
+ */
+#define MLXSW_REG_SGCR_ID 0x2000
+#define MLXSW_REG_SGCR_LEN 0x10
+
+MLXSW_REG_DEFINE(sgcr, MLXSW_REG_SGCR_ID, MLXSW_REG_SGCR_LEN);
+
+/* reg_sgcr_llb
+ * Link Local Broadcast (Default=0)
+ * When set, all Link Local packets (224.0.0.X) will be treated as broadcast
+ * packets and ignore the IGMP snooping entries.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, sgcr, llb, 0x04, 0, 1);
+
+static inline void mlxsw_reg_sgcr_pack(char *payload, bool llb)
+{
+	MLXSW_REG_ZERO(sgcr, payload);
+	mlxsw_reg_sgcr_llb_set(payload, !!llb);
+}
+
+/* SPAD - Switch Physical Address Register
+ * ---------------------------------------
+ * The SPAD register configures the switch physical MAC address.
+ */
+#define MLXSW_REG_SPAD_ID 0x2002
+#define MLXSW_REG_SPAD_LEN 0x10
+
+MLXSW_REG_DEFINE(spad, MLXSW_REG_SPAD_ID, MLXSW_REG_SPAD_LEN);
+
+/* reg_spad_base_mac
+ * Base MAC address for the switch partitions.
+ * Per switch partition MAC address is equal to:
+ * base_mac + swid
+ * Access: RW
+ */
+MLXSW_ITEM_BUF(reg, spad, base_mac, 0x02, 6);
+
+/* SSPR - Switch System Port Record Register
+ * -----------------------------------------
+ * Configures the system port to local port mapping.
+ */
+#define MLXSW_REG_SSPR_ID 0x2008
+#define MLXSW_REG_SSPR_LEN 0x8
+
+MLXSW_REG_DEFINE(sspr, MLXSW_REG_SSPR_ID, MLXSW_REG_SSPR_LEN);
+
+/* reg_sspr_m
+ * Master - if set, then the record describes the master system port.
+ * This is needed in case a local port is mapped into several system ports
+ * (for multipathing). That number will be reported as the source system
+ * port when packets are forwarded to the CPU. Only one master port is allowed
+ * per local port.
+ *
+ * Note: Must be set for Spectrum.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, sspr, m, 0x00, 31, 1);
+
+/* reg_sspr_local_port
+ * Local port number.
+ *
+ * Access: RW
+ */
+MLXSW_ITEM32_LP(reg, sspr, 0x00, 16, 0x00, 12);
+
+/* reg_sspr_system_port
+ * Unique identifier within the stacking domain that represents all the ports
+ * that are available in the system (external ports).
+ *
+ * Currently, only single-ASIC configurations are supported, so we default to
+ * 1:1 mapping between system ports and local ports.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, sspr, system_port, 0x04, 0, 16);
+
+static inline void mlxsw_reg_sspr_pack(char *payload, u16 local_port)
+{
+	MLXSW_REG_ZERO(sspr, payload);
+	mlxsw_reg_sspr_m_set(payload, 1);
+	mlxsw_reg_sspr_local_port_set(payload, local_port);
+	mlxsw_reg_sspr_system_port_set(payload, local_port);
+}
+
+/* SFDAT - Switch Filtering Database Aging Time
+ * --------------------------------------------
+ * Controls the Switch aging time. Aging time is able to be set per Switch
+ * Partition.
+ */
+#define MLXSW_REG_SFDAT_ID 0x2009
+#define MLXSW_REG_SFDAT_LEN 0x8
+
+MLXSW_REG_DEFINE(sfdat, MLXSW_REG_SFDAT_ID, MLXSW_REG_SFDAT_LEN);
+
+/* reg_sfdat_swid
+ * Switch partition ID.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, sfdat, swid, 0x00, 24, 8);
+
+/* reg_sfdat_age_time
+ * Aging time in seconds
+ * Min - 10 seconds
+ * Max - 1,000,000 seconds
+ * Default is 300 seconds.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, sfdat, age_time, 0x04, 0, 20);
+
+static inline void mlxsw_reg_sfdat_pack(char *payload, u32 age_time)
+{
+	MLXSW_REG_ZERO(sfdat, payload);
+	mlxsw_reg_sfdat_swid_set(payload, 0);
+	mlxsw_reg_sfdat_age_time_set(payload, age_time);
+}
+
+/* SFD - Switch Filtering Database
+ * -------------------------------
+ * The following register defines the access to the filtering database.
+ * The register supports querying, adding, removing and modifying the database.
+ * The access is optimized for bulk updates in which case more than one
+ * FDB record is present in the same command.
+ */
+#define MLXSW_REG_SFD_ID 0x200A
+#define MLXSW_REG_SFD_BASE_LEN 0x10 /* base length, without records */
+#define MLXSW_REG_SFD_REC_LEN 0x10 /* record length */
+#define MLXSW_REG_SFD_REC_MAX_COUNT 64
+#define MLXSW_REG_SFD_LEN (MLXSW_REG_SFD_BASE_LEN +	\
+			   MLXSW_REG_SFD_REC_LEN * MLXSW_REG_SFD_REC_MAX_COUNT)
+
+MLXSW_REG_DEFINE(sfd, MLXSW_REG_SFD_ID, MLXSW_REG_SFD_LEN);
+
+/* reg_sfd_swid
+ * Switch partition ID for queries. Reserved on Write.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, sfd, swid, 0x00, 24, 8);
+
+enum mlxsw_reg_sfd_op {
+	/* Dump entire FDB a (process according to record_locator) */
+	MLXSW_REG_SFD_OP_QUERY_DUMP = 0,
+	/* Query records by {MAC, VID/FID} value */
+	MLXSW_REG_SFD_OP_QUERY_QUERY = 1,
+	/* Query and clear activity. Query records by {MAC, VID/FID} value */
+	MLXSW_REG_SFD_OP_QUERY_QUERY_AND_CLEAR_ACTIVITY = 2,
+	/* Test. Response indicates if each of the records could be
+	 * added to the FDB.
+	 */
+	MLXSW_REG_SFD_OP_WRITE_TEST = 0,
+	/* Add/modify. Aged-out records cannot be added. This command removes
+	 * the learning notification of the {MAC, VID/FID}. Response includes
+	 * the entries that were added to the FDB.
+	 */
+	MLXSW_REG_SFD_OP_WRITE_EDIT = 1,
+	/* Remove record by {MAC, VID/FID}. This command also removes
+	 * the learning notification and aged-out notifications
+	 * of the {MAC, VID/FID}. The response provides current (pre-removal)
+	 * entries as non-aged-out.
+	 */
+	MLXSW_REG_SFD_OP_WRITE_REMOVE = 2,
+	/* Remove learned notification by {MAC, VID/FID}. The response provides
+	 * the removed learning notification.
+	 */
+	MLXSW_REG_SFD_OP_WRITE_REMOVE_NOTIFICATION = 2,
+};
+
+/* reg_sfd_op
+ * Operation.
+ * Access: OP
+ */
+MLXSW_ITEM32(reg, sfd, op, 0x04, 30, 2);
+
+/* reg_sfd_record_locator
+ * Used for querying the FDB. Use record_locator=0 to initiate the
+ * query. When a record is returned, a new record_locator is
+ * returned to be used in the subsequent query.
+ * Reserved for database update.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, sfd, record_locator, 0x04, 0, 30);
+
+/* reg_sfd_num_rec
+ * Request: Number of records to read/add/modify/remove
+ * Response: Number of records read/added/replaced/removed
+ * See above description for more details.
+ * Ranges 0..64
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, sfd, num_rec, 0x08, 0, 8);
+
+static inline void mlxsw_reg_sfd_pack(char *payload, enum mlxsw_reg_sfd_op op,
+				      u32 record_locator)
+{
+	MLXSW_REG_ZERO(sfd, payload);
+	mlxsw_reg_sfd_op_set(payload, op);
+	mlxsw_reg_sfd_record_locator_set(payload, record_locator);
+}
+
+/* reg_sfd_rec_swid
+ * Switch partition ID.
+ * Access: Index
+ */
+MLXSW_ITEM32_INDEXED(reg, sfd, rec_swid, MLXSW_REG_SFD_BASE_LEN, 24, 8,
+		     MLXSW_REG_SFD_REC_LEN, 0x00, false);
+
+enum mlxsw_reg_sfd_rec_type {
+	MLXSW_REG_SFD_REC_TYPE_UNICAST = 0x0,
+	MLXSW_REG_SFD_REC_TYPE_UNICAST_LAG = 0x1,
+	MLXSW_REG_SFD_REC_TYPE_MULTICAST = 0x2,
+	MLXSW_REG_SFD_REC_TYPE_UNICAST_TUNNEL = 0xC,
+};
+
+/* reg_sfd_rec_type
+ * FDB record type.
+ * Access: RW
+ */
+MLXSW_ITEM32_INDEXED(reg, sfd, rec_type, MLXSW_REG_SFD_BASE_LEN, 20, 4,
+		     MLXSW_REG_SFD_REC_LEN, 0x00, false);
+
+enum mlxsw_reg_sfd_rec_policy {
+	/* Replacement disabled, aging disabled. */
+	MLXSW_REG_SFD_REC_POLICY_STATIC_ENTRY = 0,
+	/* (mlag remote): Replacement enabled, aging disabled,
+	 * learning notification enabled on this port.
+	 */
+	MLXSW_REG_SFD_REC_POLICY_DYNAMIC_ENTRY_MLAG = 1,
+	/* (ingress device): Replacement enabled, aging enabled. */
+	MLXSW_REG_SFD_REC_POLICY_DYNAMIC_ENTRY_INGRESS = 3,
+};
+
+/* reg_sfd_rec_policy
+ * Policy.
+ * Access: RW
+ */
+MLXSW_ITEM32_INDEXED(reg, sfd, rec_policy, MLXSW_REG_SFD_BASE_LEN, 18, 2,
+		     MLXSW_REG_SFD_REC_LEN, 0x00, false);
+
+/* reg_sfd_rec_a
+ * Activity. Set for new static entries. Set for static entries if a frame SMAC
+ * lookup hits on the entry.
+ * To clear the a bit, use "query and clear activity" op.
+ * Access: RO
+ */
+MLXSW_ITEM32_INDEXED(reg, sfd, rec_a, MLXSW_REG_SFD_BASE_LEN, 16, 1,
+		     MLXSW_REG_SFD_REC_LEN, 0x00, false);
+
+/* reg_sfd_rec_mac
+ * MAC address.
+ * Access: Index
+ */
+MLXSW_ITEM_BUF_INDEXED(reg, sfd, rec_mac, MLXSW_REG_SFD_BASE_LEN, 6,
+		       MLXSW_REG_SFD_REC_LEN, 0x02);
+
+enum mlxsw_reg_sfd_rec_action {
+	/* forward */
+	MLXSW_REG_SFD_REC_ACTION_NOP = 0,
+	/* forward and trap, trap_id is FDB_TRAP */
+	MLXSW_REG_SFD_REC_ACTION_MIRROR_TO_CPU = 1,
+	/* trap and do not forward, trap_id is FDB_TRAP */
+	MLXSW_REG_SFD_REC_ACTION_TRAP = 2,
+	/* forward to IP router */
+	MLXSW_REG_SFD_REC_ACTION_FORWARD_IP_ROUTER = 3,
+	MLXSW_REG_SFD_REC_ACTION_DISCARD_ERROR = 15,
+};
+
+/* reg_sfd_rec_action
+ * Action to apply on the packet.
+ * Note: Dynamic entries can only be configured with NOP action.
+ * Access: RW
+ */
+MLXSW_ITEM32_INDEXED(reg, sfd, rec_action, MLXSW_REG_SFD_BASE_LEN, 28, 4,
+		     MLXSW_REG_SFD_REC_LEN, 0x0C, false);
+
+/* reg_sfd_uc_sub_port
+ * VEPA channel on local port.
+ * Valid only if local port is a non-stacking port. Must be 0 if multichannel
+ * VEPA is not enabled.
+ * Access: RW
+ */
+MLXSW_ITEM32_INDEXED(reg, sfd, uc_sub_port, MLXSW_REG_SFD_BASE_LEN, 16, 8,
+		     MLXSW_REG_SFD_REC_LEN, 0x08, false);
+
+/* reg_sfd_uc_set_vid
+ * Set VID.
+ * 0 - Do not update VID.
+ * 1 - Set VID.
+ * For Spectrum-2 when set_vid=0 and smpe_valid=1, the smpe will modify the vid.
+ * Access: RW
+ *
+ * Note: Reserved when legacy bridge model is used.
+ */
+MLXSW_ITEM32_INDEXED(reg, sfd, uc_set_vid, MLXSW_REG_SFD_BASE_LEN, 31, 1,
+		     MLXSW_REG_SFD_REC_LEN, 0x08, false);
+
+/* reg_sfd_uc_fid_vid
+ * Filtering ID or VLAN ID
+ * For SwitchX and SwitchX-2:
+ * - Dynamic entries (policy 2,3) use FID
+ * - Static entries (policy 0) use VID
+ * - When independent learning is configured, VID=FID
+ * For Spectrum: use FID for both Dynamic and Static entries.
+ * VID should not be used.
+ * Access: Index
+ */
+MLXSW_ITEM32_INDEXED(reg, sfd, uc_fid_vid, MLXSW_REG_SFD_BASE_LEN, 0, 16,
+		     MLXSW_REG_SFD_REC_LEN, 0x08, false);
+
+/* reg_sfd_uc_vid
+ * New VID when set_vid=1.
+ * Access: RW
+ *
+ * Note: Reserved when legacy bridge model is used and when set_vid=0.
+ */
+MLXSW_ITEM32_INDEXED(reg, sfd, uc_vid, MLXSW_REG_SFD_BASE_LEN, 16, 12,
+		     MLXSW_REG_SFD_REC_LEN, 0x0C, false);
+
+/* reg_sfd_uc_system_port
+ * Unique port identifier for the final destination of the packet.
+ * Access: RW
+ */
+MLXSW_ITEM32_INDEXED(reg, sfd, uc_system_port, MLXSW_REG_SFD_BASE_LEN, 0, 16,
+		     MLXSW_REG_SFD_REC_LEN, 0x0C, false);
+
+static inline void mlxsw_reg_sfd_rec_pack(char *payload, int rec_index,
+					  enum mlxsw_reg_sfd_rec_type rec_type,
+					  const char *mac,
+					  enum mlxsw_reg_sfd_rec_action action)
+{
+	u8 num_rec = mlxsw_reg_sfd_num_rec_get(payload);
+
+	if (rec_index >= num_rec)
+		mlxsw_reg_sfd_num_rec_set(payload, rec_index + 1);
+	mlxsw_reg_sfd_rec_swid_set(payload, rec_index, 0);
+	mlxsw_reg_sfd_rec_type_set(payload, rec_index, rec_type);
+	mlxsw_reg_sfd_rec_mac_memcpy_to(payload, rec_index, mac);
+	mlxsw_reg_sfd_rec_action_set(payload, rec_index, action);
+}
+
+static inline void mlxsw_reg_sfd_uc_pack(char *payload, int rec_index,
+					 enum mlxsw_reg_sfd_rec_policy policy,
+					 const char *mac, u16 fid_vid, u16 vid,
+					 enum mlxsw_reg_sfd_rec_action action,
+					 u16 local_port)
+{
+	mlxsw_reg_sfd_rec_pack(payload, rec_index,
+			       MLXSW_REG_SFD_REC_TYPE_UNICAST, mac, action);
+	mlxsw_reg_sfd_rec_policy_set(payload, rec_index, policy);
+	mlxsw_reg_sfd_uc_sub_port_set(payload, rec_index, 0);
+	mlxsw_reg_sfd_uc_fid_vid_set(payload, rec_index, fid_vid);
+	mlxsw_reg_sfd_uc_set_vid_set(payload, rec_index, vid ? true : false);
+	mlxsw_reg_sfd_uc_vid_set(payload, rec_index, vid);
+	mlxsw_reg_sfd_uc_system_port_set(payload, rec_index, local_port);
+}
+
+/* reg_sfd_uc_lag_sub_port
+ * LAG sub port.
+ * Must be 0 if multichannel VEPA is not enabled.
+ * Access: RW
+ */
+MLXSW_ITEM32_INDEXED(reg, sfd, uc_lag_sub_port, MLXSW_REG_SFD_BASE_LEN, 16, 8,
+		     MLXSW_REG_SFD_REC_LEN, 0x08, false);
+
+/* reg_sfd_uc_lag_set_vid
+ * Set VID.
+ * 0 - Do not update VID.
+ * 1 - Set VID.
+ * For Spectrum-2 when set_vid=0 and smpe_valid=1, the smpe will modify the vid.
+ * Access: RW
+ *
+ * Note: Reserved when legacy bridge model is used.
+ */
+MLXSW_ITEM32_INDEXED(reg, sfd, uc_lag_set_vid, MLXSW_REG_SFD_BASE_LEN, 31, 1,
+		     MLXSW_REG_SFD_REC_LEN, 0x08, false);
+
+/* reg_sfd_uc_lag_fid_vid
+ * Filtering ID or VLAN ID
+ * For SwitchX and SwitchX-2:
+ * - Dynamic entries (policy 2,3) use FID
+ * - Static entries (policy 0) use VID
+ * - When independent learning is configured, VID=FID
+ * For Spectrum: use FID for both Dynamic and Static entries.
+ * VID should not be used.
+ * Access: Index
+ */
+MLXSW_ITEM32_INDEXED(reg, sfd, uc_lag_fid_vid, MLXSW_REG_SFD_BASE_LEN, 0, 16,
+		     MLXSW_REG_SFD_REC_LEN, 0x08, false);
+
+/* reg_sfd_uc_lag_lag_vid
+ * New vlan ID.
+ * Access: RW
+ *
+ * Note: Reserved when legacy bridge model is used and set_vid=0.
+ */
+MLXSW_ITEM32_INDEXED(reg, sfd, uc_lag_lag_vid, MLXSW_REG_SFD_BASE_LEN, 16, 12,
+		     MLXSW_REG_SFD_REC_LEN, 0x0C, false);
+
+/* reg_sfd_uc_lag_lag_id
+ * LAG Identifier - pointer into the LAG descriptor table.
+ * Access: RW
+ */
+MLXSW_ITEM32_INDEXED(reg, sfd, uc_lag_lag_id, MLXSW_REG_SFD_BASE_LEN, 0, 10,
+		     MLXSW_REG_SFD_REC_LEN, 0x0C, false);
+
+static inline void
+mlxsw_reg_sfd_uc_lag_pack(char *payload, int rec_index,
+			  enum mlxsw_reg_sfd_rec_policy policy,
+			  const char *mac, u16 fid_vid,
+			  enum mlxsw_reg_sfd_rec_action action, u16 lag_vid,
+			  u16 lag_id)
+{
+	mlxsw_reg_sfd_rec_pack(payload, rec_index,
+			       MLXSW_REG_SFD_REC_TYPE_UNICAST_LAG,
+			       mac, action);
+	mlxsw_reg_sfd_rec_policy_set(payload, rec_index, policy);
+	mlxsw_reg_sfd_uc_lag_sub_port_set(payload, rec_index, 0);
+	mlxsw_reg_sfd_uc_lag_fid_vid_set(payload, rec_index, fid_vid);
+	mlxsw_reg_sfd_uc_lag_set_vid_set(payload, rec_index, true);
+	mlxsw_reg_sfd_uc_lag_lag_vid_set(payload, rec_index, lag_vid);
+	mlxsw_reg_sfd_uc_lag_lag_id_set(payload, rec_index, lag_id);
+}
+
+/* reg_sfd_mc_pgi
+ *
+ * Multicast port group index - index into the port group table.
+ * Value 0x1FFF indicates the pgi should point to the MID entry.
+ * For Spectrum this value must be set to 0x1FFF
+ * Access: RW
+ */
+MLXSW_ITEM32_INDEXED(reg, sfd, mc_pgi, MLXSW_REG_SFD_BASE_LEN, 16, 13,
+		     MLXSW_REG_SFD_REC_LEN, 0x08, false);
+
+/* reg_sfd_mc_fid_vid
+ *
+ * Filtering ID or VLAN ID
+ * Access: Index
+ */
+MLXSW_ITEM32_INDEXED(reg, sfd, mc_fid_vid, MLXSW_REG_SFD_BASE_LEN, 0, 16,
+		     MLXSW_REG_SFD_REC_LEN, 0x08, false);
+
+/* reg_sfd_mc_mid
+ *
+ * Multicast identifier - global identifier that represents the multicast
+ * group across all devices.
+ * Access: RW
+ */
+MLXSW_ITEM32_INDEXED(reg, sfd, mc_mid, MLXSW_REG_SFD_BASE_LEN, 0, 16,
+		     MLXSW_REG_SFD_REC_LEN, 0x0C, false);
+
+static inline void
+mlxsw_reg_sfd_mc_pack(char *payload, int rec_index,
+		      const char *mac, u16 fid_vid,
+		      enum mlxsw_reg_sfd_rec_action action, u16 mid)
+{
+	mlxsw_reg_sfd_rec_pack(payload, rec_index,
+			       MLXSW_REG_SFD_REC_TYPE_MULTICAST, mac, action);
+	mlxsw_reg_sfd_mc_pgi_set(payload, rec_index, 0x1FFF);
+	mlxsw_reg_sfd_mc_fid_vid_set(payload, rec_index, fid_vid);
+	mlxsw_reg_sfd_mc_mid_set(payload, rec_index, mid);
+}
+
+/* reg_sfd_uc_tunnel_uip_msb
+ * When protocol is IPv4, the most significant byte of the underlay IPv4
+ * destination IP.
+ * When protocol is IPv6, reserved.
+ * Access: RW
+ */
+MLXSW_ITEM32_INDEXED(reg, sfd, uc_tunnel_uip_msb, MLXSW_REG_SFD_BASE_LEN, 24,
+		     8, MLXSW_REG_SFD_REC_LEN, 0x08, false);
+
+/* reg_sfd_uc_tunnel_fid
+ * Filtering ID.
+ * Access: Index
+ */
+MLXSW_ITEM32_INDEXED(reg, sfd, uc_tunnel_fid, MLXSW_REG_SFD_BASE_LEN, 0, 16,
+		     MLXSW_REG_SFD_REC_LEN, 0x08, false);
+
+enum mlxsw_reg_sfd_uc_tunnel_protocol {
+	MLXSW_REG_SFD_UC_TUNNEL_PROTOCOL_IPV4,
+	MLXSW_REG_SFD_UC_TUNNEL_PROTOCOL_IPV6,
+};
+
+/* reg_sfd_uc_tunnel_protocol
+ * IP protocol.
+ * Access: RW
+ */
+MLXSW_ITEM32_INDEXED(reg, sfd, uc_tunnel_protocol, MLXSW_REG_SFD_BASE_LEN, 27,
+		     1, MLXSW_REG_SFD_REC_LEN, 0x0C, false);
+
+/* reg_sfd_uc_tunnel_uip_lsb
+ * When protocol is IPv4, the least significant bytes of the underlay
+ * IPv4 destination IP.
+ * When protocol is IPv6, pointer to the underlay IPv6 destination IP
+ * which is configured by RIPS.
+ * Access: RW
+ */
+MLXSW_ITEM32_INDEXED(reg, sfd, uc_tunnel_uip_lsb, MLXSW_REG_SFD_BASE_LEN, 0,
+		     24, MLXSW_REG_SFD_REC_LEN, 0x0C, false);
+
+static inline void
+mlxsw_reg_sfd_uc_tunnel_pack(char *payload, int rec_index,
+			     enum mlxsw_reg_sfd_rec_policy policy,
+			     const char *mac, u16 fid,
+			     enum mlxsw_reg_sfd_rec_action action,
+			     enum mlxsw_reg_sfd_uc_tunnel_protocol proto)
+{
+	mlxsw_reg_sfd_rec_pack(payload, rec_index,
+			       MLXSW_REG_SFD_REC_TYPE_UNICAST_TUNNEL, mac,
+			       action);
+	mlxsw_reg_sfd_rec_policy_set(payload, rec_index, policy);
+	mlxsw_reg_sfd_uc_tunnel_fid_set(payload, rec_index, fid);
+	mlxsw_reg_sfd_uc_tunnel_protocol_set(payload, rec_index, proto);
+}
+
+static inline void
+mlxsw_reg_sfd_uc_tunnel_pack4(char *payload, int rec_index,
+			      enum mlxsw_reg_sfd_rec_policy policy,
+			      const char *mac, u16 fid,
+			      enum mlxsw_reg_sfd_rec_action action, u32 uip)
+{
+	mlxsw_reg_sfd_uc_tunnel_uip_msb_set(payload, rec_index, uip >> 24);
+	mlxsw_reg_sfd_uc_tunnel_uip_lsb_set(payload, rec_index, uip);
+	mlxsw_reg_sfd_uc_tunnel_pack(payload, rec_index, policy, mac, fid,
+				     action,
+				     MLXSW_REG_SFD_UC_TUNNEL_PROTOCOL_IPV4);
+}
+
+static inline void
+mlxsw_reg_sfd_uc_tunnel_pack6(char *payload, int rec_index, const char *mac,
+			      u16 fid, enum mlxsw_reg_sfd_rec_action action,
+			      u32 uip_ptr)
+{
+	mlxsw_reg_sfd_uc_tunnel_uip_lsb_set(payload, rec_index, uip_ptr);
+	/* Only static policy is supported for IPv6 unicast tunnel entry. */
+	mlxsw_reg_sfd_uc_tunnel_pack(payload, rec_index,
+				     MLXSW_REG_SFD_REC_POLICY_STATIC_ENTRY,
+				     mac, fid, action,
+				     MLXSW_REG_SFD_UC_TUNNEL_PROTOCOL_IPV6);
+}
+
+enum mlxsw_reg_tunnel_port {
+	MLXSW_REG_TUNNEL_PORT_NVE,
+	MLXSW_REG_TUNNEL_PORT_VPLS,
+	MLXSW_REG_TUNNEL_PORT_FLEX_TUNNEL0,
+	MLXSW_REG_TUNNEL_PORT_FLEX_TUNNEL1,
+};
+
+/* SFN - Switch FDB Notification Register
+ * -------------------------------------------
+ * The switch provides notifications on newly learned FDB entries and
+ * aged out entries. The notifications can be polled by software.
+ */
+#define MLXSW_REG_SFN_ID 0x200B
+#define MLXSW_REG_SFN_BASE_LEN 0x10 /* base length, without records */
+#define MLXSW_REG_SFN_REC_LEN 0x10 /* record length */
+#define MLXSW_REG_SFN_REC_MAX_COUNT 64
+#define MLXSW_REG_SFN_LEN (MLXSW_REG_SFN_BASE_LEN +	\
+			   MLXSW_REG_SFN_REC_LEN * MLXSW_REG_SFN_REC_MAX_COUNT)
+
+MLXSW_REG_DEFINE(sfn, MLXSW_REG_SFN_ID, MLXSW_REG_SFN_LEN);
+
+/* reg_sfn_swid
+ * Switch partition ID.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, sfn, swid, 0x00, 24, 8);
+
+/* reg_sfn_end
+ * Forces the current session to end.
+ * Access: OP
+ */
+MLXSW_ITEM32(reg, sfn, end, 0x04, 20, 1);
+
+/* reg_sfn_num_rec
+ * Request: Number of learned notifications and aged-out notification
+ * records requested.
+ * Response: Number of notification records returned (must be smaller
+ * than or equal to the value requested)
+ * Ranges 0..64
+ * Access: OP
+ */
+MLXSW_ITEM32(reg, sfn, num_rec, 0x04, 0, 8);
+
+static inline void mlxsw_reg_sfn_pack(char *payload)
+{
+	MLXSW_REG_ZERO(sfn, payload);
+	mlxsw_reg_sfn_swid_set(payload, 0);
+	mlxsw_reg_sfn_end_set(payload, 0);
+	mlxsw_reg_sfn_num_rec_set(payload, MLXSW_REG_SFN_REC_MAX_COUNT);
+}
+
+/* reg_sfn_rec_swid
+ * Switch partition ID.
+ * Access: RO
+ */
+MLXSW_ITEM32_INDEXED(reg, sfn, rec_swid, MLXSW_REG_SFN_BASE_LEN, 24, 8,
+		     MLXSW_REG_SFN_REC_LEN, 0x00, false);
+
+enum mlxsw_reg_sfn_rec_type {
+	/* MAC addresses learned on a regular port. */
+	MLXSW_REG_SFN_REC_TYPE_LEARNED_MAC = 0x5,
+	/* MAC addresses learned on a LAG port. */
+	MLXSW_REG_SFN_REC_TYPE_LEARNED_MAC_LAG = 0x6,
+	/* Aged-out MAC address on a regular port. */
+	MLXSW_REG_SFN_REC_TYPE_AGED_OUT_MAC = 0x7,
+	/* Aged-out MAC address on a LAG port. */
+	MLXSW_REG_SFN_REC_TYPE_AGED_OUT_MAC_LAG = 0x8,
+	/* Learned unicast tunnel record. */
+	MLXSW_REG_SFN_REC_TYPE_LEARNED_UNICAST_TUNNEL = 0xD,
+	/* Aged-out unicast tunnel record. */
+	MLXSW_REG_SFN_REC_TYPE_AGED_OUT_UNICAST_TUNNEL = 0xE,
+};
+
+/* reg_sfn_rec_type
+ * Notification record type.
+ * Access: RO
+ */
+MLXSW_ITEM32_INDEXED(reg, sfn, rec_type, MLXSW_REG_SFN_BASE_LEN, 20, 4,
+		     MLXSW_REG_SFN_REC_LEN, 0x00, false);
+
+/* reg_sfn_rec_mac
+ * MAC address.
+ * Access: RO
+ */
+MLXSW_ITEM_BUF_INDEXED(reg, sfn, rec_mac, MLXSW_REG_SFN_BASE_LEN, 6,
+		       MLXSW_REG_SFN_REC_LEN, 0x02);
+
+/* reg_sfn_mac_sub_port
+ * VEPA channel on the local port.
+ * 0 if multichannel VEPA is not enabled.
+ * Access: RO
+ */
+MLXSW_ITEM32_INDEXED(reg, sfn, mac_sub_port, MLXSW_REG_SFN_BASE_LEN, 16, 8,
+		     MLXSW_REG_SFN_REC_LEN, 0x08, false);
+
+/* reg_sfn_mac_fid
+ * Filtering identifier.
+ * Access: RO
+ */
+MLXSW_ITEM32_INDEXED(reg, sfn, mac_fid, MLXSW_REG_SFN_BASE_LEN, 0, 16,
+		     MLXSW_REG_SFN_REC_LEN, 0x08, false);
+
+/* reg_sfn_mac_system_port
+ * Unique port identifier for the final destination of the packet.
+ * Access: RO
+ */
+MLXSW_ITEM32_INDEXED(reg, sfn, mac_system_port, MLXSW_REG_SFN_BASE_LEN, 0, 16,
+		     MLXSW_REG_SFN_REC_LEN, 0x0C, false);
+
+static inline void mlxsw_reg_sfn_mac_unpack(char *payload, int rec_index,
+					    char *mac, u16 *p_vid,
+					    u16 *p_local_port)
+{
+	mlxsw_reg_sfn_rec_mac_memcpy_from(payload, rec_index, mac);
+	*p_vid = mlxsw_reg_sfn_mac_fid_get(payload, rec_index);
+	*p_local_port = mlxsw_reg_sfn_mac_system_port_get(payload, rec_index);
+}
+
+/* reg_sfn_mac_lag_lag_id
+ * LAG ID (pointer into the LAG descriptor table).
+ * Access: RO
+ */
+MLXSW_ITEM32_INDEXED(reg, sfn, mac_lag_lag_id, MLXSW_REG_SFN_BASE_LEN, 0, 10,
+		     MLXSW_REG_SFN_REC_LEN, 0x0C, false);
+
+static inline void mlxsw_reg_sfn_mac_lag_unpack(char *payload, int rec_index,
+						char *mac, u16 *p_vid,
+						u16 *p_lag_id)
+{
+	mlxsw_reg_sfn_rec_mac_memcpy_from(payload, rec_index, mac);
+	*p_vid = mlxsw_reg_sfn_mac_fid_get(payload, rec_index);
+	*p_lag_id = mlxsw_reg_sfn_mac_lag_lag_id_get(payload, rec_index);
+}
+
+/* reg_sfn_uc_tunnel_uip_msb
+ * When protocol is IPv4, the most significant byte of the underlay IPv4
+ * address of the remote VTEP.
+ * When protocol is IPv6, reserved.
+ * Access: RO
+ */
+MLXSW_ITEM32_INDEXED(reg, sfn, uc_tunnel_uip_msb, MLXSW_REG_SFN_BASE_LEN, 24,
+		     8, MLXSW_REG_SFN_REC_LEN, 0x08, false);
+
+enum mlxsw_reg_sfn_uc_tunnel_protocol {
+	MLXSW_REG_SFN_UC_TUNNEL_PROTOCOL_IPV4,
+	MLXSW_REG_SFN_UC_TUNNEL_PROTOCOL_IPV6,
+};
+
+/* reg_sfn_uc_tunnel_protocol
+ * IP protocol.
+ * Access: RO
+ */
+MLXSW_ITEM32_INDEXED(reg, sfn, uc_tunnel_protocol, MLXSW_REG_SFN_BASE_LEN, 27,
+		     1, MLXSW_REG_SFN_REC_LEN, 0x0C, false);
+
+/* reg_sfn_uc_tunnel_uip_lsb
+ * When protocol is IPv4, the least significant bytes of the underlay
+ * IPv4 address of the remote VTEP.
+ * When protocol is IPv6, ipv6_id to be queried from TNIPSD.
+ * Access: RO
+ */
+MLXSW_ITEM32_INDEXED(reg, sfn, uc_tunnel_uip_lsb, MLXSW_REG_SFN_BASE_LEN, 0,
+		     24, MLXSW_REG_SFN_REC_LEN, 0x0C, false);
+
+/* reg_sfn_uc_tunnel_port
+ * Tunnel port.
+ * Reserved on Spectrum.
+ * Access: RO
+ */
+MLXSW_ITEM32_INDEXED(reg, sfn, tunnel_port, MLXSW_REG_SFN_BASE_LEN, 0, 4,
+		     MLXSW_REG_SFN_REC_LEN, 0x10, false);
+
+static inline void
+mlxsw_reg_sfn_uc_tunnel_unpack(char *payload, int rec_index, char *mac,
+			       u16 *p_fid, u32 *p_uip,
+			       enum mlxsw_reg_sfn_uc_tunnel_protocol *p_proto)
+{
+	u32 uip_msb, uip_lsb;
+
+	mlxsw_reg_sfn_rec_mac_memcpy_from(payload, rec_index, mac);
+	*p_fid = mlxsw_reg_sfn_mac_fid_get(payload, rec_index);
+	uip_msb = mlxsw_reg_sfn_uc_tunnel_uip_msb_get(payload, rec_index);
+	uip_lsb = mlxsw_reg_sfn_uc_tunnel_uip_lsb_get(payload, rec_index);
+	*p_uip = uip_msb << 24 | uip_lsb;
+	*p_proto = mlxsw_reg_sfn_uc_tunnel_protocol_get(payload, rec_index);
+}
+
+/* SPMS - Switch Port MSTP/RSTP State Register
+ * -------------------------------------------
+ * Configures the spanning tree state of a physical port.
+ */
+#define MLXSW_REG_SPMS_ID 0x200D
+#define MLXSW_REG_SPMS_LEN 0x404
+
+MLXSW_REG_DEFINE(spms, MLXSW_REG_SPMS_ID, MLXSW_REG_SPMS_LEN);
+
+/* reg_spms_local_port
+ * Local port number.
+ * Access: Index
+ */
+MLXSW_ITEM32_LP(reg, spms, 0x00, 16, 0x00, 12);
+
+enum mlxsw_reg_spms_state {
+	MLXSW_REG_SPMS_STATE_NO_CHANGE,
+	MLXSW_REG_SPMS_STATE_DISCARDING,
+	MLXSW_REG_SPMS_STATE_LEARNING,
+	MLXSW_REG_SPMS_STATE_FORWARDING,
+};
+
+/* reg_spms_state
+ * Spanning tree state of each VLAN ID (VID) of the local port.
+ * 0 - Do not change spanning tree state (used only when writing).
+ * 1 - Discarding. No learning or forwarding to/from this port (default).
+ * 2 - Learning. Port is learning, but not forwarding.
+ * 3 - Forwarding. Port is learning and forwarding.
+ * Access: RW
+ */
+MLXSW_ITEM_BIT_ARRAY(reg, spms, state, 0x04, 0x400, 2);
+
+static inline void mlxsw_reg_spms_pack(char *payload, u16 local_port)
+{
+	MLXSW_REG_ZERO(spms, payload);
+	mlxsw_reg_spms_local_port_set(payload, local_port);
+}
+
+static inline void mlxsw_reg_spms_vid_pack(char *payload, u16 vid,
+					   enum mlxsw_reg_spms_state state)
+{
+	mlxsw_reg_spms_state_set(payload, vid, state);
+}
+
+/* SPVID - Switch Port VID
+ * -----------------------
+ * The switch port VID configures the default VID for a port.
+ */
+#define MLXSW_REG_SPVID_ID 0x200E
+#define MLXSW_REG_SPVID_LEN 0x08
+
+MLXSW_REG_DEFINE(spvid, MLXSW_REG_SPVID_ID, MLXSW_REG_SPVID_LEN);
+
+/* reg_spvid_tport
+ * Port is tunnel port.
+ * Reserved when SwitchX/-2 or Spectrum-1.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, spvid, tport, 0x00, 24, 1);
+
+/* reg_spvid_local_port
+ * When tport = 0: Local port number. Not supported for CPU port.
+ * When tport = 1: Tunnel port.
+ * Access: Index
+ */
+MLXSW_ITEM32_LP(reg, spvid, 0x00, 16, 0x00, 12);
+
+/* reg_spvid_sub_port
+ * Virtual port within the physical port.
+ * Should be set to 0 when virtual ports are not enabled on the port.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, spvid, sub_port, 0x00, 8, 8);
+
+/* reg_spvid_egr_et_set
+ * When VLAN is pushed at ingress (for untagged packets or for
+ * QinQ push mode) then the EtherType is decided at the egress port.
+ * Reserved when Spectrum-1.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, spvid, egr_et_set, 0x04, 24, 1);
+
+/* reg_spvid_et_vlan
+ * EtherType used for when VLAN is pushed at ingress (for untagged
+ * packets or for QinQ push mode).
+ * 0: ether_type0 - (default)
+ * 1: ether_type1
+ * 2: ether_type2 - Reserved when Spectrum-1, supported by Spectrum-2
+ * Ethertype IDs are configured by SVER.
+ * Reserved when egr_et_set = 1.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, spvid, et_vlan, 0x04, 16, 2);
+
+/* reg_spvid_pvid
+ * Port default VID
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, spvid, pvid, 0x04, 0, 12);
+
+static inline void mlxsw_reg_spvid_pack(char *payload, u16 local_port, u16 pvid,
+					u8 et_vlan)
+{
+	MLXSW_REG_ZERO(spvid, payload);
+	mlxsw_reg_spvid_local_port_set(payload, local_port);
+	mlxsw_reg_spvid_pvid_set(payload, pvid);
+	mlxsw_reg_spvid_et_vlan_set(payload, et_vlan);
+}
+
+/* SPVM - Switch Port VLAN Membership
+ * ----------------------------------
+ * The Switch Port VLAN Membership register configures the VLAN membership
+ * of a port in a VLAN denoted by VID. VLAN membership is managed per
+ * virtual port. The register can be used to add and remove VID(s) from a port.
+ */
+#define MLXSW_REG_SPVM_ID 0x200F
+#define MLXSW_REG_SPVM_BASE_LEN 0x04 /* base length, without records */
+#define MLXSW_REG_SPVM_REC_LEN 0x04 /* record length */
+#define MLXSW_REG_SPVM_REC_MAX_COUNT 255
+#define MLXSW_REG_SPVM_LEN (MLXSW_REG_SPVM_BASE_LEN +	\
+		    MLXSW_REG_SPVM_REC_LEN * MLXSW_REG_SPVM_REC_MAX_COUNT)
+
+MLXSW_REG_DEFINE(spvm, MLXSW_REG_SPVM_ID, MLXSW_REG_SPVM_LEN);
+
+/* reg_spvm_pt
+ * Priority tagged. If this bit is set, packets forwarded to the port with
+ * untagged VLAN membership (u bit is set) will be tagged with priority tag
+ * (VID=0)
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, spvm, pt, 0x00, 31, 1);
+
+/* reg_spvm_pte
+ * Priority Tagged Update Enable. On Write operations, if this bit is cleared,
+ * the pt bit will NOT be updated. To update the pt bit, pte must be set.
+ * Access: WO
+ */
+MLXSW_ITEM32(reg, spvm, pte, 0x00, 30, 1);
+
+/* reg_spvm_local_port
+ * Local port number.
+ * Access: Index
+ */
+MLXSW_ITEM32_LP(reg, spvm, 0x00, 16, 0x00, 12);
+
+/* reg_spvm_sub_port
+ * Virtual port within the physical port.
+ * Should be set to 0 when virtual ports are not enabled on the port.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, spvm, sub_port, 0x00, 8, 8);
+
+/* reg_spvm_num_rec
+ * Number of records to update. Each record contains: i, e, u, vid.
+ * Access: OP
+ */
+MLXSW_ITEM32(reg, spvm, num_rec, 0x00, 0, 8);
+
+/* reg_spvm_rec_i
+ * Ingress membership in VLAN ID.
+ * Access: Index
+ */
+MLXSW_ITEM32_INDEXED(reg, spvm, rec_i,
+		     MLXSW_REG_SPVM_BASE_LEN, 14, 1,
+		     MLXSW_REG_SPVM_REC_LEN, 0, false);
+
+/* reg_spvm_rec_e
+ * Egress membership in VLAN ID.
+ * Access: Index
+ */
+MLXSW_ITEM32_INDEXED(reg, spvm, rec_e,
+		     MLXSW_REG_SPVM_BASE_LEN, 13, 1,
+		     MLXSW_REG_SPVM_REC_LEN, 0, false);
+
+/* reg_spvm_rec_u
+ * Untagged - port is an untagged member - egress transmission uses untagged
+ * frames on VID<n>
+ * Access: Index
+ */
+MLXSW_ITEM32_INDEXED(reg, spvm, rec_u,
+		     MLXSW_REG_SPVM_BASE_LEN, 12, 1,
+		     MLXSW_REG_SPVM_REC_LEN, 0, false);
+
+/* reg_spvm_rec_vid
+ * Egress membership in VLAN ID.
+ * Access: Index
+ */
+MLXSW_ITEM32_INDEXED(reg, spvm, rec_vid,
+		     MLXSW_REG_SPVM_BASE_LEN, 0, 12,
+		     MLXSW_REG_SPVM_REC_LEN, 0, false);
+
+static inline void mlxsw_reg_spvm_pack(char *payload, u16 local_port,
+				       u16 vid_begin, u16 vid_end,
+				       bool is_member, bool untagged)
+{
+	int size = vid_end - vid_begin + 1;
+	int i;
+
+	MLXSW_REG_ZERO(spvm, payload);
+	mlxsw_reg_spvm_local_port_set(payload, local_port);
+	mlxsw_reg_spvm_num_rec_set(payload, size);
+
+	for (i = 0; i < size; i++) {
+		mlxsw_reg_spvm_rec_i_set(payload, i, is_member);
+		mlxsw_reg_spvm_rec_e_set(payload, i, is_member);
+		mlxsw_reg_spvm_rec_u_set(payload, i, untagged);
+		mlxsw_reg_spvm_rec_vid_set(payload, i, vid_begin + i);
+	}
+}
+
+/* SPAFT - Switch Port Acceptable Frame Types
+ * ------------------------------------------
+ * The Switch Port Acceptable Frame Types register configures the frame
+ * admittance of the port.
+ */
+#define MLXSW_REG_SPAFT_ID 0x2010
+#define MLXSW_REG_SPAFT_LEN 0x08
+
+MLXSW_REG_DEFINE(spaft, MLXSW_REG_SPAFT_ID, MLXSW_REG_SPAFT_LEN);
+
+/* reg_spaft_local_port
+ * Local port number.
+ * Access: Index
+ *
+ * Note: CPU port is not supported (all tag types are allowed).
+ */
+MLXSW_ITEM32_LP(reg, spaft, 0x00, 16, 0x00, 12);
+
+/* reg_spaft_sub_port
+ * Virtual port within the physical port.
+ * Should be set to 0 when virtual ports are not enabled on the port.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, spaft, sub_port, 0x00, 8, 8);
+
+/* reg_spaft_allow_untagged
+ * When set, untagged frames on the ingress are allowed (default).
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, spaft, allow_untagged, 0x04, 31, 1);
+
+/* reg_spaft_allow_prio_tagged
+ * When set, priority tagged frames on the ingress are allowed (default).
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, spaft, allow_prio_tagged, 0x04, 30, 1);
+
+/* reg_spaft_allow_tagged
+ * When set, tagged frames on the ingress are allowed (default).
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, spaft, allow_tagged, 0x04, 29, 1);
+
+static inline void mlxsw_reg_spaft_pack(char *payload, u16 local_port,
+					bool allow_untagged)
+{
+	MLXSW_REG_ZERO(spaft, payload);
+	mlxsw_reg_spaft_local_port_set(payload, local_port);
+	mlxsw_reg_spaft_allow_untagged_set(payload, allow_untagged);
+	mlxsw_reg_spaft_allow_prio_tagged_set(payload, allow_untagged);
+	mlxsw_reg_spaft_allow_tagged_set(payload, true);
+}
+
+/* SFGC - Switch Flooding Group Configuration
+ * ------------------------------------------
+ * The following register controls the association of flooding tables and MIDs
+ * to packet types used for flooding.
+ */
+#define MLXSW_REG_SFGC_ID 0x2011
+#define MLXSW_REG_SFGC_LEN 0x14
+
+MLXSW_REG_DEFINE(sfgc, MLXSW_REG_SFGC_ID, MLXSW_REG_SFGC_LEN);
+
+enum mlxsw_reg_sfgc_type {
+	MLXSW_REG_SFGC_TYPE_BROADCAST,
+	MLXSW_REG_SFGC_TYPE_UNKNOWN_UNICAST,
+	MLXSW_REG_SFGC_TYPE_UNREGISTERED_MULTICAST_IPV4,
+	MLXSW_REG_SFGC_TYPE_UNREGISTERED_MULTICAST_IPV6,
+	MLXSW_REG_SFGC_TYPE_RESERVED,
+	MLXSW_REG_SFGC_TYPE_UNREGISTERED_MULTICAST_NON_IP,
+	MLXSW_REG_SFGC_TYPE_IPV4_LINK_LOCAL,
+	MLXSW_REG_SFGC_TYPE_IPV6_ALL_HOST,
+	MLXSW_REG_SFGC_TYPE_MAX,
+};
+
+/* reg_sfgc_type
+ * The traffic type to reach the flooding table.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, sfgc, type, 0x00, 0, 4);
+
+/* bridge_type is used in SFGC and SFMR. */
+enum mlxsw_reg_bridge_type {
+	MLXSW_REG_BRIDGE_TYPE_0 = 0, /* Used for .1q FIDs. */
+	MLXSW_REG_BRIDGE_TYPE_1 = 1, /* Used for .1d FIDs. */
+};
+
+/* reg_sfgc_bridge_type
+ * Access: Index
+ *
+ * Note: SwitchX-2 only supports 802.1Q mode.
+ */
+MLXSW_ITEM32(reg, sfgc, bridge_type, 0x04, 24, 3);
+
+enum mlxsw_flood_table_type {
+	MLXSW_REG_SFGC_TABLE_TYPE_VID = 1,
+	MLXSW_REG_SFGC_TABLE_TYPE_SINGLE = 2,
+	MLXSW_REG_SFGC_TABLE_TYPE_ANY = 0,
+	MLXSW_REG_SFGC_TABLE_TYPE_FID_OFFSET = 3,
+	MLXSW_REG_SFGC_TABLE_TYPE_FID = 4,
+};
+
+/* reg_sfgc_table_type
+ * See mlxsw_flood_table_type
+ * Access: RW
+ *
+ * Note: FID offset and FID types are not supported in SwitchX-2.
+ */
+MLXSW_ITEM32(reg, sfgc, table_type, 0x04, 16, 3);
+
+/* reg_sfgc_flood_table
+ * Flooding table index to associate with the specific type on the specific
+ * switch partition.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, sfgc, flood_table, 0x04, 0, 6);
+
+/* reg_sfgc_counter_set_type
+ * Counter Set Type for flow counters.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, sfgc, counter_set_type, 0x0C, 24, 8);
+
+/* reg_sfgc_counter_index
+ * Counter Index for flow counters.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, sfgc, counter_index, 0x0C, 0, 24);
+
+/* reg_sfgc_mid_base
+ * MID Base.
+ * Access: RW
+ *
+ * Note: Reserved when legacy bridge model is used.
+ */
+MLXSW_ITEM32(reg, sfgc, mid_base, 0x10, 0, 16);
+
+static inline void
+mlxsw_reg_sfgc_pack(char *payload, enum mlxsw_reg_sfgc_type type,
+		    enum mlxsw_reg_bridge_type bridge_type,
+		    enum mlxsw_flood_table_type table_type,
+		    unsigned int flood_table, u16 mid_base)
+{
+	MLXSW_REG_ZERO(sfgc, payload);
+	mlxsw_reg_sfgc_type_set(payload, type);
+	mlxsw_reg_sfgc_bridge_type_set(payload, bridge_type);
+	mlxsw_reg_sfgc_table_type_set(payload, table_type);
+	mlxsw_reg_sfgc_flood_table_set(payload, flood_table);
+	mlxsw_reg_sfgc_mid_base_set(payload, mid_base);
+}
+
+/* SFDF - Switch Filtering DB Flush
+ * --------------------------------
+ * The switch filtering DB flush register is used to flush the FDB.
+ * Note that FDB notifications are flushed as well.
+ */
+#define MLXSW_REG_SFDF_ID 0x2013
+#define MLXSW_REG_SFDF_LEN 0x14
+
+MLXSW_REG_DEFINE(sfdf, MLXSW_REG_SFDF_ID, MLXSW_REG_SFDF_LEN);
+
+/* reg_sfdf_swid
+ * Switch partition ID.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, sfdf, swid, 0x00, 24, 8);
+
+enum mlxsw_reg_sfdf_flush_type {
+	MLXSW_REG_SFDF_FLUSH_PER_SWID,
+	MLXSW_REG_SFDF_FLUSH_PER_FID,
+	MLXSW_REG_SFDF_FLUSH_PER_PORT,
+	MLXSW_REG_SFDF_FLUSH_PER_PORT_AND_FID,
+	MLXSW_REG_SFDF_FLUSH_PER_LAG,
+	MLXSW_REG_SFDF_FLUSH_PER_LAG_AND_FID,
+	MLXSW_REG_SFDF_FLUSH_PER_NVE,
+	MLXSW_REG_SFDF_FLUSH_PER_NVE_AND_FID,
+};
+
+/* reg_sfdf_flush_type
+ * Flush type.
+ * 0 - All SWID dynamic entries are flushed.
+ * 1 - All FID dynamic entries are flushed.
+ * 2 - All dynamic entries pointing to port are flushed.
+ * 3 - All FID dynamic entries pointing to port are flushed.
+ * 4 - All dynamic entries pointing to LAG are flushed.
+ * 5 - All FID dynamic entries pointing to LAG are flushed.
+ * 6 - All entries of type "Unicast Tunnel" or "Multicast Tunnel" are
+ *     flushed.
+ * 7 - All entries of type "Unicast Tunnel" or "Multicast Tunnel" are
+ *     flushed, per FID.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, sfdf, flush_type, 0x04, 28, 4);
+
+/* reg_sfdf_flush_static
+ * Static.
+ * 0 - Flush only dynamic entries.
+ * 1 - Flush both dynamic and static entries.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, sfdf, flush_static, 0x04, 24, 1);
+
+static inline void mlxsw_reg_sfdf_pack(char *payload,
+				       enum mlxsw_reg_sfdf_flush_type type)
+{
+	MLXSW_REG_ZERO(sfdf, payload);
+	mlxsw_reg_sfdf_flush_type_set(payload, type);
+	mlxsw_reg_sfdf_flush_static_set(payload, true);
+}
+
+/* reg_sfdf_fid
+ * FID to flush.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, sfdf, fid, 0x0C, 0, 16);
+
+/* reg_sfdf_system_port
+ * Port to flush.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, sfdf, system_port, 0x0C, 0, 16);
+
+/* reg_sfdf_port_fid_system_port
+ * Port to flush, pointed to by FID.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, sfdf, port_fid_system_port, 0x08, 0, 16);
+
+/* reg_sfdf_lag_id
+ * LAG ID to flush.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, sfdf, lag_id, 0x0C, 0, 10);
+
+/* reg_sfdf_lag_fid_lag_id
+ * LAG ID to flush, pointed to by FID.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, sfdf, lag_fid_lag_id, 0x08, 0, 10);
+
+/* SLDR - Switch LAG Descriptor Register
+ * -----------------------------------------
+ * The switch LAG descriptor register is populated by LAG descriptors.
+ * Each LAG descriptor is indexed by lag_id. The LAG ID runs from 0 to
+ * max_lag-1.
+ */
+#define MLXSW_REG_SLDR_ID 0x2014
+#define MLXSW_REG_SLDR_LEN 0x0C /* counting in only one port in list */
+
+MLXSW_REG_DEFINE(sldr, MLXSW_REG_SLDR_ID, MLXSW_REG_SLDR_LEN);
+
+enum mlxsw_reg_sldr_op {
+	/* Indicates a creation of a new LAG-ID, lag_id must be valid */
+	MLXSW_REG_SLDR_OP_LAG_CREATE,
+	MLXSW_REG_SLDR_OP_LAG_DESTROY,
+	/* Ports that appear in the list have the Distributor enabled */
+	MLXSW_REG_SLDR_OP_LAG_ADD_PORT_LIST,
+	/* Removes ports from the disributor list */
+	MLXSW_REG_SLDR_OP_LAG_REMOVE_PORT_LIST,
+};
+
+/* reg_sldr_op
+ * Operation.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, sldr, op, 0x00, 29, 3);
+
+/* reg_sldr_lag_id
+ * LAG identifier. The lag_id is the index into the LAG descriptor table.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, sldr, lag_id, 0x00, 0, 10);
+
+static inline void mlxsw_reg_sldr_lag_create_pack(char *payload, u8 lag_id)
+{
+	MLXSW_REG_ZERO(sldr, payload);
+	mlxsw_reg_sldr_op_set(payload, MLXSW_REG_SLDR_OP_LAG_CREATE);
+	mlxsw_reg_sldr_lag_id_set(payload, lag_id);
+}
+
+static inline void mlxsw_reg_sldr_lag_destroy_pack(char *payload, u8 lag_id)
+{
+	MLXSW_REG_ZERO(sldr, payload);
+	mlxsw_reg_sldr_op_set(payload, MLXSW_REG_SLDR_OP_LAG_DESTROY);
+	mlxsw_reg_sldr_lag_id_set(payload, lag_id);
+}
+
+/* reg_sldr_num_ports
+ * The number of member ports of the LAG.
+ * Reserved for Create / Destroy operations
+ * For Add / Remove operations - indicates the number of ports in the list.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, sldr, num_ports, 0x04, 24, 8);
+
+/* reg_sldr_system_port
+ * System port.
+ * Access: RW
+ */
+MLXSW_ITEM32_INDEXED(reg, sldr, system_port, 0x08, 0, 16, 4, 0, false);
+
+static inline void mlxsw_reg_sldr_lag_add_port_pack(char *payload, u8 lag_id,
+						    u16 local_port)
+{
+	MLXSW_REG_ZERO(sldr, payload);
+	mlxsw_reg_sldr_op_set(payload, MLXSW_REG_SLDR_OP_LAG_ADD_PORT_LIST);
+	mlxsw_reg_sldr_lag_id_set(payload, lag_id);
+	mlxsw_reg_sldr_num_ports_set(payload, 1);
+	mlxsw_reg_sldr_system_port_set(payload, 0, local_port);
+}
+
+static inline void mlxsw_reg_sldr_lag_remove_port_pack(char *payload, u8 lag_id,
+						       u16 local_port)
+{
+	MLXSW_REG_ZERO(sldr, payload);
+	mlxsw_reg_sldr_op_set(payload, MLXSW_REG_SLDR_OP_LAG_REMOVE_PORT_LIST);
+	mlxsw_reg_sldr_lag_id_set(payload, lag_id);
+	mlxsw_reg_sldr_num_ports_set(payload, 1);
+	mlxsw_reg_sldr_system_port_set(payload, 0, local_port);
+}
+
+/* SLCR - Switch LAG Configuration 2 Register
+ * -------------------------------------------
+ * The Switch LAG Configuration register is used for configuring the
+ * LAG properties of the switch.
+ */
+#define MLXSW_REG_SLCR_ID 0x2015
+#define MLXSW_REG_SLCR_LEN 0x10
+
+MLXSW_REG_DEFINE(slcr, MLXSW_REG_SLCR_ID, MLXSW_REG_SLCR_LEN);
+
+enum mlxsw_reg_slcr_pp {
+	/* Global Configuration (for all ports) */
+	MLXSW_REG_SLCR_PP_GLOBAL,
+	/* Per port configuration, based on local_port field */
+	MLXSW_REG_SLCR_PP_PER_PORT,
+};
+
+/* reg_slcr_pp
+ * Per Port Configuration
+ * Note: Reading at Global mode results in reading port 1 configuration.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, slcr, pp, 0x00, 24, 1);
+
+/* reg_slcr_local_port
+ * Local port number
+ * Supported from CPU port
+ * Not supported from router port
+ * Reserved when pp = Global Configuration
+ * Access: Index
+ */
+MLXSW_ITEM32_LP(reg, slcr, 0x00, 16, 0x00, 12);
+
+enum mlxsw_reg_slcr_type {
+	MLXSW_REG_SLCR_TYPE_CRC, /* default */
+	MLXSW_REG_SLCR_TYPE_XOR,
+	MLXSW_REG_SLCR_TYPE_RANDOM,
+};
+
+/* reg_slcr_type
+ * Hash type
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, slcr, type, 0x00, 0, 4);
+
+/* Ingress port */
+#define MLXSW_REG_SLCR_LAG_HASH_IN_PORT		BIT(0)
+/* SMAC - for IPv4 and IPv6 packets */
+#define MLXSW_REG_SLCR_LAG_HASH_SMAC_IP		BIT(1)
+/* SMAC - for non-IP packets */
+#define MLXSW_REG_SLCR_LAG_HASH_SMAC_NONIP	BIT(2)
+#define MLXSW_REG_SLCR_LAG_HASH_SMAC \
+	(MLXSW_REG_SLCR_LAG_HASH_SMAC_IP | \
+	 MLXSW_REG_SLCR_LAG_HASH_SMAC_NONIP)
+/* DMAC - for IPv4 and IPv6 packets */
+#define MLXSW_REG_SLCR_LAG_HASH_DMAC_IP		BIT(3)
+/* DMAC - for non-IP packets */
+#define MLXSW_REG_SLCR_LAG_HASH_DMAC_NONIP	BIT(4)
+#define MLXSW_REG_SLCR_LAG_HASH_DMAC \
+	(MLXSW_REG_SLCR_LAG_HASH_DMAC_IP | \
+	 MLXSW_REG_SLCR_LAG_HASH_DMAC_NONIP)
+/* Ethertype - for IPv4 and IPv6 packets */
+#define MLXSW_REG_SLCR_LAG_HASH_ETHERTYPE_IP	BIT(5)
+/* Ethertype - for non-IP packets */
+#define MLXSW_REG_SLCR_LAG_HASH_ETHERTYPE_NONIP	BIT(6)
+#define MLXSW_REG_SLCR_LAG_HASH_ETHERTYPE \
+	(MLXSW_REG_SLCR_LAG_HASH_ETHERTYPE_IP | \
+	 MLXSW_REG_SLCR_LAG_HASH_ETHERTYPE_NONIP)
+/* VLAN ID - for IPv4 and IPv6 packets */
+#define MLXSW_REG_SLCR_LAG_HASH_VLANID_IP	BIT(7)
+/* VLAN ID - for non-IP packets */
+#define MLXSW_REG_SLCR_LAG_HASH_VLANID_NONIP	BIT(8)
+#define MLXSW_REG_SLCR_LAG_HASH_VLANID \
+	(MLXSW_REG_SLCR_LAG_HASH_VLANID_IP | \
+	 MLXSW_REG_SLCR_LAG_HASH_VLANID_NONIP)
+/* Source IP address (can be IPv4 or IPv6) */
+#define MLXSW_REG_SLCR_LAG_HASH_SIP		BIT(9)
+/* Destination IP address (can be IPv4 or IPv6) */
+#define MLXSW_REG_SLCR_LAG_HASH_DIP		BIT(10)
+/* TCP/UDP source port */
+#define MLXSW_REG_SLCR_LAG_HASH_SPORT		BIT(11)
+/* TCP/UDP destination port*/
+#define MLXSW_REG_SLCR_LAG_HASH_DPORT		BIT(12)
+/* IPv4 Protocol/IPv6 Next Header */
+#define MLXSW_REG_SLCR_LAG_HASH_IPPROTO		BIT(13)
+/* IPv6 Flow label */
+#define MLXSW_REG_SLCR_LAG_HASH_FLOWLABEL	BIT(14)
+/* SID - FCoE source ID */
+#define MLXSW_REG_SLCR_LAG_HASH_FCOE_SID	BIT(15)
+/* DID - FCoE destination ID */
+#define MLXSW_REG_SLCR_LAG_HASH_FCOE_DID	BIT(16)
+/* OXID - FCoE originator exchange ID */
+#define MLXSW_REG_SLCR_LAG_HASH_FCOE_OXID	BIT(17)
+/* Destination QP number - for RoCE packets */
+#define MLXSW_REG_SLCR_LAG_HASH_ROCE_DQP	BIT(19)
+
+/* reg_slcr_lag_hash
+ * LAG hashing configuration. This is a bitmask, in which each set
+ * bit includes the corresponding item in the LAG hash calculation.
+ * The default lag_hash contains SMAC, DMAC, VLANID and
+ * Ethertype (for all packet types).
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, slcr, lag_hash, 0x04, 0, 20);
+
+/* reg_slcr_seed
+ * LAG seed value. The seed is the same for all ports.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, slcr, seed, 0x08, 0, 32);
+
+static inline void mlxsw_reg_slcr_pack(char *payload, u16 lag_hash, u32 seed)
+{
+	MLXSW_REG_ZERO(slcr, payload);
+	mlxsw_reg_slcr_pp_set(payload, MLXSW_REG_SLCR_PP_GLOBAL);
+	mlxsw_reg_slcr_type_set(payload, MLXSW_REG_SLCR_TYPE_CRC);
+	mlxsw_reg_slcr_lag_hash_set(payload, lag_hash);
+	mlxsw_reg_slcr_seed_set(payload, seed);
+}
+
+/* SLCOR - Switch LAG Collector Register
+ * -------------------------------------
+ * The Switch LAG Collector register controls the Local Port membership
+ * in a LAG and enablement of the collector.
+ */
+#define MLXSW_REG_SLCOR_ID 0x2016
+#define MLXSW_REG_SLCOR_LEN 0x10
+
+MLXSW_REG_DEFINE(slcor, MLXSW_REG_SLCOR_ID, MLXSW_REG_SLCOR_LEN);
+
+enum mlxsw_reg_slcor_col {
+	/* Port is added with collector disabled */
+	MLXSW_REG_SLCOR_COL_LAG_ADD_PORT,
+	MLXSW_REG_SLCOR_COL_LAG_COLLECTOR_ENABLED,
+	MLXSW_REG_SLCOR_COL_LAG_COLLECTOR_DISABLED,
+	MLXSW_REG_SLCOR_COL_LAG_REMOVE_PORT,
+};
+
+/* reg_slcor_col
+ * Collector configuration
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, slcor, col, 0x00, 30, 2);
+
+/* reg_slcor_local_port
+ * Local port number
+ * Not supported for CPU port
+ * Access: Index
+ */
+MLXSW_ITEM32_LP(reg, slcor, 0x00, 16, 0x00, 12);
+
+/* reg_slcor_lag_id
+ * LAG Identifier. Index into the LAG descriptor table.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, slcor, lag_id, 0x00, 0, 10);
+
+/* reg_slcor_port_index
+ * Port index in the LAG list. Only valid on Add Port to LAG col.
+ * Valid range is from 0 to cap_max_lag_members-1
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, slcor, port_index, 0x04, 0, 10);
+
+static inline void mlxsw_reg_slcor_pack(char *payload,
+					u16 local_port, u16 lag_id,
+					enum mlxsw_reg_slcor_col col)
+{
+	MLXSW_REG_ZERO(slcor, payload);
+	mlxsw_reg_slcor_col_set(payload, col);
+	mlxsw_reg_slcor_local_port_set(payload, local_port);
+	mlxsw_reg_slcor_lag_id_set(payload, lag_id);
+}
+
+static inline void mlxsw_reg_slcor_port_add_pack(char *payload,
+						 u16 local_port, u16 lag_id,
+						 u8 port_index)
+{
+	mlxsw_reg_slcor_pack(payload, local_port, lag_id,
+			     MLXSW_REG_SLCOR_COL_LAG_ADD_PORT);
+	mlxsw_reg_slcor_port_index_set(payload, port_index);
+}
+
+static inline void mlxsw_reg_slcor_port_remove_pack(char *payload,
+						    u16 local_port, u16 lag_id)
+{
+	mlxsw_reg_slcor_pack(payload, local_port, lag_id,
+			     MLXSW_REG_SLCOR_COL_LAG_REMOVE_PORT);
+}
+
+static inline void mlxsw_reg_slcor_col_enable_pack(char *payload,
+						   u16 local_port, u16 lag_id)
+{
+	mlxsw_reg_slcor_pack(payload, local_port, lag_id,
+			     MLXSW_REG_SLCOR_COL_LAG_COLLECTOR_ENABLED);
+}
+
+static inline void mlxsw_reg_slcor_col_disable_pack(char *payload,
+						    u16 local_port, u16 lag_id)
+{
+	mlxsw_reg_slcor_pack(payload, local_port, lag_id,
+			     MLXSW_REG_SLCOR_COL_LAG_COLLECTOR_ENABLED);
+}
+
+/* SPMLR - Switch Port MAC Learning Register
+ * -----------------------------------------
+ * Controls the Switch MAC learning policy per port.
+ */
+#define MLXSW_REG_SPMLR_ID 0x2018
+#define MLXSW_REG_SPMLR_LEN 0x8
+
+MLXSW_REG_DEFINE(spmlr, MLXSW_REG_SPMLR_ID, MLXSW_REG_SPMLR_LEN);
+
+/* reg_spmlr_local_port
+ * Local port number.
+ * Access: Index
+ */
+MLXSW_ITEM32_LP(reg, spmlr, 0x00, 16, 0x00, 12);
+
+/* reg_spmlr_sub_port
+ * Virtual port within the physical port.
+ * Should be set to 0 when virtual ports are not enabled on the port.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, spmlr, sub_port, 0x00, 8, 8);
+
+enum mlxsw_reg_spmlr_learn_mode {
+	MLXSW_REG_SPMLR_LEARN_MODE_DISABLE = 0,
+	MLXSW_REG_SPMLR_LEARN_MODE_ENABLE = 2,
+	MLXSW_REG_SPMLR_LEARN_MODE_SEC = 3,
+};
+
+/* reg_spmlr_learn_mode
+ * Learning mode on the port.
+ * 0 - Learning disabled.
+ * 2 - Learning enabled.
+ * 3 - Security mode.
+ *
+ * In security mode the switch does not learn MACs on the port, but uses the
+ * SMAC to see if it exists on another ingress port. If so, the packet is
+ * classified as a bad packet and is discarded unless the software registers
+ * to receive port security error packets usign HPKT.
+ */
+MLXSW_ITEM32(reg, spmlr, learn_mode, 0x04, 30, 2);
+
+static inline void mlxsw_reg_spmlr_pack(char *payload, u16 local_port,
+					enum mlxsw_reg_spmlr_learn_mode mode)
+{
+	MLXSW_REG_ZERO(spmlr, payload);
+	mlxsw_reg_spmlr_local_port_set(payload, local_port);
+	mlxsw_reg_spmlr_sub_port_set(payload, 0);
+	mlxsw_reg_spmlr_learn_mode_set(payload, mode);
+}
+
+/* SVFA - Switch VID to FID Allocation Register
+ * --------------------------------------------
+ * Controls the VID to FID mapping and {Port, VID} to FID mapping for
+ * virtualized ports.
+ */
+#define MLXSW_REG_SVFA_ID 0x201C
+#define MLXSW_REG_SVFA_LEN 0x18
+
+MLXSW_REG_DEFINE(svfa, MLXSW_REG_SVFA_ID, MLXSW_REG_SVFA_LEN);
+
+/* reg_svfa_swid
+ * Switch partition ID.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, svfa, swid, 0x00, 24, 8);
+
+/* reg_svfa_local_port
+ * Local port number.
+ * Access: Index
+ *
+ * Note: Reserved for 802.1Q FIDs.
+ */
+MLXSW_ITEM32_LP(reg, svfa, 0x00, 16, 0x00, 12);
+
+enum mlxsw_reg_svfa_mt {
+	MLXSW_REG_SVFA_MT_VID_TO_FID,
+	MLXSW_REG_SVFA_MT_PORT_VID_TO_FID,
+	MLXSW_REG_SVFA_MT_VNI_TO_FID,
+};
+
+/* reg_svfa_mapping_table
+ * Mapping table:
+ * 0 - VID to FID
+ * 1 - {Port, VID} to FID
+ * Access: Index
+ *
+ * Note: Reserved for SwitchX-2.
+ */
+MLXSW_ITEM32(reg, svfa, mapping_table, 0x00, 8, 3);
+
+/* reg_svfa_v
+ * Valid.
+ * Valid if set.
+ * Access: RW
+ *
+ * Note: Reserved for SwitchX-2.
+ */
+MLXSW_ITEM32(reg, svfa, v, 0x00, 0, 1);
+
+/* reg_svfa_fid
+ * Filtering ID.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, svfa, fid, 0x04, 16, 16);
+
+/* reg_svfa_vid
+ * VLAN ID.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, svfa, vid, 0x04, 0, 12);
+
+/* reg_svfa_counter_set_type
+ * Counter set type for flow counters.
+ * Access: RW
+ *
+ * Note: Reserved for SwitchX-2.
+ */
+MLXSW_ITEM32(reg, svfa, counter_set_type, 0x08, 24, 8);
+
+/* reg_svfa_counter_index
+ * Counter index for flow counters.
+ * Access: RW
+ *
+ * Note: Reserved for SwitchX-2.
+ */
+MLXSW_ITEM32(reg, svfa, counter_index, 0x08, 0, 24);
+
+/* reg_svfa_vni
+ * Virtual Network Identifier.
+ * Access: Index
+ *
+ * Note: Reserved when mapping_table is not 2 (VNI mapping table).
+ */
+MLXSW_ITEM32(reg, svfa, vni, 0x10, 0, 24);
+
+/* reg_svfa_irif_v
+ * Ingress RIF valid.
+ * 0 - Ingress RIF is not valid, no ingress RIF assigned.
+ * 1 - Ingress RIF valid.
+ * Must not be set for a non enabled RIF.
+ * Access: RW
+ *
+ * Note: Reserved when legacy bridge model is used.
+ */
+MLXSW_ITEM32(reg, svfa, irif_v, 0x14, 24, 1);
+
+/* reg_svfa_irif
+ * Ingress RIF (Router Interface).
+ * Range is 0..cap_max_router_interfaces-1.
+ * Access: RW
+ *
+ * Note: Reserved when legacy bridge model is used and when irif_v=0.
+ */
+MLXSW_ITEM32(reg, svfa, irif, 0x14, 0, 16);
+
+static inline void __mlxsw_reg_svfa_pack(char *payload,
+					 enum mlxsw_reg_svfa_mt mt, bool valid,
+					 u16 fid, bool irif_v, u16 irif)
+{
+	MLXSW_REG_ZERO(svfa, payload);
+	mlxsw_reg_svfa_swid_set(payload, 0);
+	mlxsw_reg_svfa_mapping_table_set(payload, mt);
+	mlxsw_reg_svfa_v_set(payload, valid);
+	mlxsw_reg_svfa_fid_set(payload, fid);
+	mlxsw_reg_svfa_irif_v_set(payload, irif_v);
+	mlxsw_reg_svfa_irif_set(payload, irif_v ? irif : 0);
+}
+
+static inline void mlxsw_reg_svfa_port_vid_pack(char *payload, u16 local_port,
+						bool valid, u16 fid, u16 vid,
+						bool irif_v, u16 irif)
+{
+	enum mlxsw_reg_svfa_mt mt = MLXSW_REG_SVFA_MT_PORT_VID_TO_FID;
+
+	__mlxsw_reg_svfa_pack(payload, mt, valid, fid, irif_v, irif);
+	mlxsw_reg_svfa_local_port_set(payload, local_port);
+	mlxsw_reg_svfa_vid_set(payload, vid);
+}
+
+static inline void mlxsw_reg_svfa_vid_pack(char *payload, bool valid, u16 fid,
+					   u16 vid, bool irif_v, u16 irif)
+{
+	enum mlxsw_reg_svfa_mt mt = MLXSW_REG_SVFA_MT_VID_TO_FID;
+
+	__mlxsw_reg_svfa_pack(payload, mt, valid, fid, irif_v, irif);
+	mlxsw_reg_svfa_vid_set(payload, vid);
+}
+
+static inline void mlxsw_reg_svfa_vni_pack(char *payload, bool valid, u16 fid,
+					   u32 vni, bool irif_v, u16 irif)
+{
+	enum mlxsw_reg_svfa_mt mt = MLXSW_REG_SVFA_MT_VNI_TO_FID;
+
+	__mlxsw_reg_svfa_pack(payload, mt, valid, fid, irif_v, irif);
+	mlxsw_reg_svfa_vni_set(payload, vni);
+}
+
+/*  SPVTR - Switch Port VLAN Stacking Register
+ *  ------------------------------------------
+ *  The Switch Port VLAN Stacking register configures the VLAN mode of the port
+ *  to enable VLAN stacking.
+ */
+#define MLXSW_REG_SPVTR_ID 0x201D
+#define MLXSW_REG_SPVTR_LEN 0x10
+
+MLXSW_REG_DEFINE(spvtr, MLXSW_REG_SPVTR_ID, MLXSW_REG_SPVTR_LEN);
+
+/* reg_spvtr_tport
+ * Port is tunnel port.
+ * Access: Index
+ *
+ * Note: Reserved when SwitchX/-2 or Spectrum-1.
+ */
+MLXSW_ITEM32(reg, spvtr, tport, 0x00, 24, 1);
+
+/* reg_spvtr_local_port
+ * When tport = 0: local port number (Not supported from/to CPU).
+ * When tport = 1: tunnel port.
+ * Access: Index
+ */
+MLXSW_ITEM32_LP(reg, spvtr, 0x00, 16, 0x00, 12);
+
+/* reg_spvtr_ippe
+ * Ingress Port Prio Mode Update Enable.
+ * When set, the Port Prio Mode is updated with the provided ipprio_mode field.
+ * Reserved on Get operations.
+ * Access: OP
+ */
+MLXSW_ITEM32(reg, spvtr, ippe, 0x04, 31, 1);
+
+/* reg_spvtr_ipve
+ * Ingress Port VID Mode Update Enable.
+ * When set, the Ingress Port VID Mode is updated with the provided ipvid_mode
+ * field.
+ * Reserved on Get operations.
+ * Access: OP
+ */
+MLXSW_ITEM32(reg, spvtr, ipve, 0x04, 30, 1);
+
+/* reg_spvtr_epve
+ * Egress Port VID Mode Update Enable.
+ * When set, the Egress Port VID Mode is updated with the provided epvid_mode
+ * field.
+ * Access: OP
+ */
+MLXSW_ITEM32(reg, spvtr, epve, 0x04, 29, 1);
+
+/* reg_spvtr_ipprio_mode
+ * Ingress Port Priority Mode.
+ * This controls the PCP and DEI of the new outer VLAN
+ * Note: for SwitchX/-2 the DEI is not affected.
+ * 0: use port default PCP and DEI (configured by QPDPC).
+ * 1: use C-VLAN PCP and DEI.
+ * Has no effect when ipvid_mode = 0.
+ * Reserved when tport = 1.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, spvtr, ipprio_mode, 0x04, 20, 4);
+
+enum mlxsw_reg_spvtr_ipvid_mode {
+	/* IEEE Compliant PVID (default) */
+	MLXSW_REG_SPVTR_IPVID_MODE_IEEE_COMPLIANT_PVID,
+	/* Push VLAN (for VLAN stacking, except prio tagged packets) */
+	MLXSW_REG_SPVTR_IPVID_MODE_PUSH_VLAN_FOR_UNTAGGED_PACKET,
+	/* Always push VLAN (also for prio tagged packets) */
+	MLXSW_REG_SPVTR_IPVID_MODE_ALWAYS_PUSH_VLAN,
+};
+
+/* reg_spvtr_ipvid_mode
+ * Ingress Port VLAN-ID Mode.
+ * For Spectrum family, this affects the values of SPVM.i
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, spvtr, ipvid_mode, 0x04, 16, 4);
+
+enum mlxsw_reg_spvtr_epvid_mode {
+	/* IEEE Compliant VLAN membership */
+	MLXSW_REG_SPVTR_EPVID_MODE_IEEE_COMPLIANT_VLAN_MEMBERSHIP,
+	/* Pop VLAN (for VLAN stacking) */
+	MLXSW_REG_SPVTR_EPVID_MODE_POP_VLAN,
+};
+
+/* reg_spvtr_epvid_mode
+ * Egress Port VLAN-ID Mode.
+ * For Spectrum family, this affects the values of SPVM.e,u,pt.
+ * Access: WO
+ */
+MLXSW_ITEM32(reg, spvtr, epvid_mode, 0x04, 0, 4);
+
+static inline void mlxsw_reg_spvtr_pack(char *payload, bool tport,
+					u16 local_port,
+					enum mlxsw_reg_spvtr_ipvid_mode ipvid_mode)
+{
+	MLXSW_REG_ZERO(spvtr, payload);
+	mlxsw_reg_spvtr_tport_set(payload, tport);
+	mlxsw_reg_spvtr_local_port_set(payload, local_port);
+	mlxsw_reg_spvtr_ipvid_mode_set(payload, ipvid_mode);
+	mlxsw_reg_spvtr_ipve_set(payload, true);
+}
+
+/* SVPE - Switch Virtual-Port Enabling Register
+ * --------------------------------------------
+ * Enables port virtualization.
+ */
+#define MLXSW_REG_SVPE_ID 0x201E
+#define MLXSW_REG_SVPE_LEN 0x4
+
+MLXSW_REG_DEFINE(svpe, MLXSW_REG_SVPE_ID, MLXSW_REG_SVPE_LEN);
+
+/* reg_svpe_local_port
+ * Local port number
+ * Access: Index
+ *
+ * Note: CPU port is not supported (uses VLAN mode only).
+ */
+MLXSW_ITEM32_LP(reg, svpe, 0x00, 16, 0x00, 12);
+
+/* reg_svpe_vp_en
+ * Virtual port enable.
+ * 0 - Disable, VLAN mode (VID to FID).
+ * 1 - Enable, Virtual port mode ({Port, VID} to FID).
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, svpe, vp_en, 0x00, 8, 1);
+
+static inline void mlxsw_reg_svpe_pack(char *payload, u16 local_port,
+				       bool enable)
+{
+	MLXSW_REG_ZERO(svpe, payload);
+	mlxsw_reg_svpe_local_port_set(payload, local_port);
+	mlxsw_reg_svpe_vp_en_set(payload, enable);
+}
+
+/* SFMR - Switch FID Management Register
+ * -------------------------------------
+ * Creates and configures FIDs.
+ */
+#define MLXSW_REG_SFMR_ID 0x201F
+#define MLXSW_REG_SFMR_LEN 0x30
+
+MLXSW_REG_DEFINE(sfmr, MLXSW_REG_SFMR_ID, MLXSW_REG_SFMR_LEN);
+
+enum mlxsw_reg_sfmr_op {
+	MLXSW_REG_SFMR_OP_CREATE_FID,
+	MLXSW_REG_SFMR_OP_DESTROY_FID,
+};
+
+/* reg_sfmr_op
+ * Operation.
+ * 0 - Create or edit FID.
+ * 1 - Destroy FID.
+ * Access: WO
+ */
+MLXSW_ITEM32(reg, sfmr, op, 0x00, 24, 4);
+
+/* reg_sfmr_fid
+ * Filtering ID.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, sfmr, fid, 0x00, 0, 16);
+
+/* reg_sfmr_flood_rsp
+ * Router sub-port flooding table.
+ * 0 - Regular flooding table.
+ * 1 - Router sub-port flooding table. For this FID the flooding is per
+ * router-sub-port local_port. Must not be set for a FID which is not a
+ * router-sub-port and must be set prior to enabling the relevant RIF.
+ * Access: RW
+ *
+ * Note: Reserved when legacy bridge model is used.
+ */
+MLXSW_ITEM32(reg, sfmr, flood_rsp, 0x08, 31, 1);
+
+/* reg_sfmr_flood_bridge_type
+ * Flood bridge type (see SFGC.bridge_type).
+ * 0 - type_0.
+ * 1 - type_1.
+ * Access: RW
+ *
+ * Note: Reserved when legacy bridge model is used and when flood_rsp=1.
+ */
+MLXSW_ITEM32(reg, sfmr, flood_bridge_type, 0x08, 28, 1);
+
+/* reg_sfmr_fid_offset
+ * FID offset.
+ * Used to point into the flooding table selected by SFGC register if
+ * the table is of type FID-Offset. Otherwise, this field is reserved.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, sfmr, fid_offset, 0x08, 0, 16);
+
+/* reg_sfmr_vtfp
+ * Valid Tunnel Flood Pointer.
+ * If not set, then nve_tunnel_flood_ptr is reserved and considered NULL.
+ * Access: RW
+ *
+ * Note: Reserved for 802.1Q FIDs.
+ */
+MLXSW_ITEM32(reg, sfmr, vtfp, 0x0C, 31, 1);
+
+/* reg_sfmr_nve_tunnel_flood_ptr
+ * Underlay Flooding and BC Pointer.
+ * Used as a pointer to the first entry of the group based link lists of
+ * flooding or BC entries (for NVE tunnels).
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, sfmr, nve_tunnel_flood_ptr, 0x0C, 0, 24);
+
+/* reg_sfmr_vv
+ * VNI Valid.
+ * If not set, then vni is reserved.
+ * Access: RW
+ *
+ * Note: Reserved for 802.1Q FIDs.
+ */
+MLXSW_ITEM32(reg, sfmr, vv, 0x10, 31, 1);
+
+/* reg_sfmr_vni
+ * Virtual Network Identifier.
+ * When legacy bridge model is used, a given VNI can only be assigned to one
+ * FID. When unified bridge model is used, it configures only the FID->VNI,
+ * the VNI->FID is done by SVFA.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, sfmr, vni, 0x10, 0, 24);
+
+/* reg_sfmr_irif_v
+ * Ingress RIF valid.
+ * 0 - Ingress RIF is not valid, no ingress RIF assigned.
+ * 1 - Ingress RIF valid.
+ * Must not be set for a non valid RIF.
+ * Access: RW
+ *
+ * Note: Reserved when legacy bridge model is used.
+ */
+MLXSW_ITEM32(reg, sfmr, irif_v, 0x14, 24, 1);
+
+/* reg_sfmr_irif
+ * Ingress RIF (Router Interface).
+ * Range is 0..cap_max_router_interfaces-1.
+ * Access: RW
+ *
+ * Note: Reserved when legacy bridge model is used and when irif_v=0.
+ */
+MLXSW_ITEM32(reg, sfmr, irif, 0x14, 0, 16);
+
+/* reg_sfmr_smpe_valid
+ * SMPE is valid.
+ * Access: RW
+ *
+ * Note: Reserved when legacy bridge model is used, when flood_rsp=1 and on
+ * Spectrum-1.
+ */
+MLXSW_ITEM32(reg, sfmr, smpe_valid, 0x28, 20, 1);
+
+/* reg_sfmr_smpe
+ * Switch multicast port to egress VID.
+ * Range is 0..cap_max_rmpe-1
+ * Access: RW
+ *
+ * Note: Reserved when legacy bridge model is used, when flood_rsp=1 and on
+ * Spectrum-1.
+ */
+MLXSW_ITEM32(reg, sfmr, smpe, 0x28, 0, 16);
+
+static inline void mlxsw_reg_sfmr_pack(char *payload,
+				       enum mlxsw_reg_sfmr_op op, u16 fid,
+				       u16 fid_offset, bool flood_rsp,
+				       enum mlxsw_reg_bridge_type bridge_type,
+				       bool smpe_valid, u16 smpe)
+{
+	MLXSW_REG_ZERO(sfmr, payload);
+	mlxsw_reg_sfmr_op_set(payload, op);
+	mlxsw_reg_sfmr_fid_set(payload, fid);
+	mlxsw_reg_sfmr_fid_offset_set(payload, fid_offset);
+	mlxsw_reg_sfmr_vtfp_set(payload, false);
+	mlxsw_reg_sfmr_vv_set(payload, false);
+	mlxsw_reg_sfmr_flood_rsp_set(payload, flood_rsp);
+	mlxsw_reg_sfmr_flood_bridge_type_set(payload, bridge_type);
+	mlxsw_reg_sfmr_smpe_valid_set(payload, smpe_valid);
+	mlxsw_reg_sfmr_smpe_set(payload, smpe);
+}
+
+/* SPVMLR - Switch Port VLAN MAC Learning Register
+ * -----------------------------------------------
+ * Controls the switch MAC learning policy per {Port, VID}.
+ */
+#define MLXSW_REG_SPVMLR_ID 0x2020
+#define MLXSW_REG_SPVMLR_BASE_LEN 0x04 /* base length, without records */
+#define MLXSW_REG_SPVMLR_REC_LEN 0x04 /* record length */
+#define MLXSW_REG_SPVMLR_REC_MAX_COUNT 255
+#define MLXSW_REG_SPVMLR_LEN (MLXSW_REG_SPVMLR_BASE_LEN + \
+			      MLXSW_REG_SPVMLR_REC_LEN * \
+			      MLXSW_REG_SPVMLR_REC_MAX_COUNT)
+
+MLXSW_REG_DEFINE(spvmlr, MLXSW_REG_SPVMLR_ID, MLXSW_REG_SPVMLR_LEN);
+
+/* reg_spvmlr_local_port
+ * Local ingress port.
+ * Access: Index
+ *
+ * Note: CPU port is not supported.
+ */
+MLXSW_ITEM32_LP(reg, spvmlr, 0x00, 16, 0x00, 12);
+
+/* reg_spvmlr_num_rec
+ * Number of records to update.
+ * Access: OP
+ */
+MLXSW_ITEM32(reg, spvmlr, num_rec, 0x00, 0, 8);
+
+/* reg_spvmlr_rec_learn_enable
+ * 0 - Disable learning for {Port, VID}.
+ * 1 - Enable learning for {Port, VID}.
+ * Access: RW
+ */
+MLXSW_ITEM32_INDEXED(reg, spvmlr, rec_learn_enable, MLXSW_REG_SPVMLR_BASE_LEN,
+		     31, 1, MLXSW_REG_SPVMLR_REC_LEN, 0x00, false);
+
+/* reg_spvmlr_rec_vid
+ * VLAN ID to be added/removed from port or for querying.
+ * Access: Index
+ */
+MLXSW_ITEM32_INDEXED(reg, spvmlr, rec_vid, MLXSW_REG_SPVMLR_BASE_LEN, 0, 12,
+		     MLXSW_REG_SPVMLR_REC_LEN, 0x00, false);
+
+static inline void mlxsw_reg_spvmlr_pack(char *payload, u16 local_port,
+					 u16 vid_begin, u16 vid_end,
+					 bool learn_enable)
+{
+	int num_rec = vid_end - vid_begin + 1;
+	int i;
+
+	WARN_ON(num_rec < 1 || num_rec > MLXSW_REG_SPVMLR_REC_MAX_COUNT);
+
+	MLXSW_REG_ZERO(spvmlr, payload);
+	mlxsw_reg_spvmlr_local_port_set(payload, local_port);
+	mlxsw_reg_spvmlr_num_rec_set(payload, num_rec);
+
+	for (i = 0; i < num_rec; i++) {
+		mlxsw_reg_spvmlr_rec_learn_enable_set(payload, i, learn_enable);
+		mlxsw_reg_spvmlr_rec_vid_set(payload, i, vid_begin + i);
+	}
+}
+
+/* SPVC - Switch Port VLAN Classification Register
+ * -----------------------------------------------
+ * Configures the port to identify packets as untagged / single tagged /
+ * double packets based on the packet EtherTypes.
+ * Ethertype IDs are configured by SVER.
+ */
+#define MLXSW_REG_SPVC_ID 0x2026
+#define MLXSW_REG_SPVC_LEN 0x0C
+
+MLXSW_REG_DEFINE(spvc, MLXSW_REG_SPVC_ID, MLXSW_REG_SPVC_LEN);
+
+/* reg_spvc_local_port
+ * Local port.
+ * Access: Index
+ *
+ * Note: applies both to Rx port and Tx port, so if a packet traverses
+ * through Rx port i and a Tx port j then port i and port j must have the
+ * same configuration.
+ */
+MLXSW_ITEM32_LP(reg, spvc, 0x00, 16, 0x00, 12);
+
+/* reg_spvc_inner_et2
+ * Vlan Tag1 EtherType2 enable.
+ * Packet is initially classified as double VLAN Tag if in addition to
+ * being classified with a tag0 VLAN Tag its tag1 EtherType value is
+ * equal to ether_type2.
+ * 0: disable (default)
+ * 1: enable
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, spvc, inner_et2, 0x08, 17, 1);
+
+/* reg_spvc_et2
+ * Vlan Tag0 EtherType2 enable.
+ * Packet is initially classified as VLAN Tag if its tag0 EtherType is
+ * equal to ether_type2.
+ * 0: disable (default)
+ * 1: enable
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, spvc, et2, 0x08, 16, 1);
+
+/* reg_spvc_inner_et1
+ * Vlan Tag1 EtherType1 enable.
+ * Packet is initially classified as double VLAN Tag if in addition to
+ * being classified with a tag0 VLAN Tag its tag1 EtherType value is
+ * equal to ether_type1.
+ * 0: disable
+ * 1: enable (default)
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, spvc, inner_et1, 0x08, 9, 1);
+
+/* reg_spvc_et1
+ * Vlan Tag0 EtherType1 enable.
+ * Packet is initially classified as VLAN Tag if its tag0 EtherType is
+ * equal to ether_type1.
+ * 0: disable
+ * 1: enable (default)
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, spvc, et1, 0x08, 8, 1);
+
+/* reg_inner_et0
+ * Vlan Tag1 EtherType0 enable.
+ * Packet is initially classified as double VLAN Tag if in addition to
+ * being classified with a tag0 VLAN Tag its tag1 EtherType value is
+ * equal to ether_type0.
+ * 0: disable
+ * 1: enable (default)
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, spvc, inner_et0, 0x08, 1, 1);
+
+/* reg_et0
+ * Vlan Tag0 EtherType0 enable.
+ * Packet is initially classified as VLAN Tag if its tag0 EtherType is
+ * equal to ether_type0.
+ * 0: disable
+ * 1: enable (default)
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, spvc, et0, 0x08, 0, 1);
+
+static inline void mlxsw_reg_spvc_pack(char *payload, u16 local_port, bool et1,
+				       bool et0)
+{
+	MLXSW_REG_ZERO(spvc, payload);
+	mlxsw_reg_spvc_local_port_set(payload, local_port);
+	/* Enable inner_et1 and inner_et0 to enable identification of double
+	 * tagged packets.
+	 */
+	mlxsw_reg_spvc_inner_et1_set(payload, 1);
+	mlxsw_reg_spvc_inner_et0_set(payload, 1);
+	mlxsw_reg_spvc_et1_set(payload, et1);
+	mlxsw_reg_spvc_et0_set(payload, et0);
+}
+
+/* SPEVET - Switch Port Egress VLAN EtherType
+ * ------------------------------------------
+ * The switch port egress VLAN EtherType configures which EtherType to push at
+ * egress for packets incoming through a local port for which 'SPVID.egr_et_set'
+ * is set.
+ */
+#define MLXSW_REG_SPEVET_ID 0x202A
+#define MLXSW_REG_SPEVET_LEN 0x08
+
+MLXSW_REG_DEFINE(spevet, MLXSW_REG_SPEVET_ID, MLXSW_REG_SPEVET_LEN);
+
+/* reg_spevet_local_port
+ * Egress Local port number.
+ * Not supported to CPU port.
+ * Access: Index
+ */
+MLXSW_ITEM32_LP(reg, spevet, 0x00, 16, 0x00, 12);
+
+/* reg_spevet_et_vlan
+ * Egress EtherType VLAN to push when SPVID.egr_et_set field set for the packet:
+ * 0: ether_type0 - (default)
+ * 1: ether_type1
+ * 2: ether_type2
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, spevet, et_vlan, 0x04, 16, 2);
+
+static inline void mlxsw_reg_spevet_pack(char *payload, u16 local_port,
+					 u8 et_vlan)
+{
+	MLXSW_REG_ZERO(spevet, payload);
+	mlxsw_reg_spevet_local_port_set(payload, local_port);
+	mlxsw_reg_spevet_et_vlan_set(payload, et_vlan);
+}
+
+/* SMPE - Switch Multicast Port to Egress VID
+ * ------------------------------------------
+ * The switch multicast port to egress VID maps
+ * {egress_port, SMPE index} -> {VID}.
+ */
+#define MLXSW_REG_SMPE_ID 0x202B
+#define MLXSW_REG_SMPE_LEN 0x0C
+
+MLXSW_REG_DEFINE(smpe, MLXSW_REG_SMPE_ID, MLXSW_REG_SMPE_LEN);
+
+/* reg_smpe_local_port
+ * Local port number.
+ * CPU port is not supported.
+ * Access: Index
+ */
+MLXSW_ITEM32_LP(reg, smpe, 0x00, 16, 0x00, 12);
+
+/* reg_smpe_smpe_index
+ * Switch multicast port to egress VID.
+ * Range is 0..cap_max_rmpe-1.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, smpe, smpe_index, 0x04, 0, 16);
+
+/* reg_smpe_evid
+ * Egress VID.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, smpe, evid, 0x08, 0, 12);
+
+static inline void mlxsw_reg_smpe_pack(char *payload, u16 local_port,
+				       u16 smpe_index, u16 evid)
+{
+	MLXSW_REG_ZERO(smpe, payload);
+	mlxsw_reg_smpe_local_port_set(payload, local_port);
+	mlxsw_reg_smpe_smpe_index_set(payload, smpe_index);
+	mlxsw_reg_smpe_evid_set(payload, evid);
+}
+
+/* SMID-V2 - Switch Multicast ID Version 2 Register
+ * ------------------------------------------------
+ * The MID record maps from a MID (Multicast ID), which is a unique identifier
+ * of the multicast group within the stacking domain, into a list of local
+ * ports into which the packet is replicated.
+ */
+#define MLXSW_REG_SMID2_ID 0x2034
+#define MLXSW_REG_SMID2_LEN 0x120
+
+MLXSW_REG_DEFINE(smid2, MLXSW_REG_SMID2_ID, MLXSW_REG_SMID2_LEN);
+
+/* reg_smid2_swid
+ * Switch partition ID.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, smid2, swid, 0x00, 24, 8);
+
+/* reg_smid2_mid
+ * Multicast identifier - global identifier that represents the multicast group
+ * across all devices.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, smid2, mid, 0x00, 0, 16);
+
+/* reg_smid2_smpe_valid
+ * SMPE is valid.
+ * When not valid, the egress VID will not be modified by the SMPE table.
+ * Access: RW
+ *
+ * Note: Reserved when legacy bridge model is used and on Spectrum-2.
+ */
+MLXSW_ITEM32(reg, smid2, smpe_valid, 0x08, 20, 1);
+
+/* reg_smid2_smpe
+ * Switch multicast port to egress VID.
+ * Access: RW
+ *
+ * Note: Reserved when legacy bridge model is used and on Spectrum-2.
+ */
+MLXSW_ITEM32(reg, smid2, smpe, 0x08, 0, 16);
+
+/* reg_smid2_port
+ * Local port memebership (1 bit per port).
+ * Access: RW
+ */
+MLXSW_ITEM_BIT_ARRAY(reg, smid2, port, 0x20, 0x80, 1);
+
+/* reg_smid2_port_mask
+ * Local port mask (1 bit per port).
+ * Access: WO
+ */
+MLXSW_ITEM_BIT_ARRAY(reg, smid2, port_mask, 0xA0, 0x80, 1);
+
+static inline void mlxsw_reg_smid2_pack(char *payload, u16 mid, u16 port,
+					bool set, bool smpe_valid, u16 smpe)
+{
+	MLXSW_REG_ZERO(smid2, payload);
+	mlxsw_reg_smid2_swid_set(payload, 0);
+	mlxsw_reg_smid2_mid_set(payload, mid);
+	mlxsw_reg_smid2_port_set(payload, port, set);
+	mlxsw_reg_smid2_port_mask_set(payload, port, 1);
+	mlxsw_reg_smid2_smpe_valid_set(payload, smpe_valid);
+	mlxsw_reg_smid2_smpe_set(payload, smpe_valid ? smpe : 0);
+}
+
+/* CWTP - Congetion WRED ECN TClass Profile
+ * ----------------------------------------
+ * Configures the profiles for queues of egress port and traffic class
+ */
+#define MLXSW_REG_CWTP_ID 0x2802
+#define MLXSW_REG_CWTP_BASE_LEN 0x28
+#define MLXSW_REG_CWTP_PROFILE_DATA_REC_LEN 0x08
+#define MLXSW_REG_CWTP_LEN 0x40
+
+MLXSW_REG_DEFINE(cwtp, MLXSW_REG_CWTP_ID, MLXSW_REG_CWTP_LEN);
+
+/* reg_cwtp_local_port
+ * Local port number
+ * Not supported for CPU port
+ * Access: Index
+ */
+MLXSW_ITEM32_LP(reg, cwtp, 0x00, 16, 0x00, 12);
+
+/* reg_cwtp_traffic_class
+ * Traffic Class to configure
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, cwtp, traffic_class, 32, 0, 8);
+
+/* reg_cwtp_profile_min
+ * Minimum Average Queue Size of the profile in cells.
+ * Access: RW
+ */
+MLXSW_ITEM32_INDEXED(reg, cwtp, profile_min, MLXSW_REG_CWTP_BASE_LEN,
+		     0, 20, MLXSW_REG_CWTP_PROFILE_DATA_REC_LEN, 0, false);
+
+/* reg_cwtp_profile_percent
+ * Percentage of WRED and ECN marking for maximum Average Queue size
+ * Range is 0 to 100, units of integer percentage
+ * Access: RW
+ */
+MLXSW_ITEM32_INDEXED(reg, cwtp, profile_percent, MLXSW_REG_CWTP_BASE_LEN,
+		     24, 7, MLXSW_REG_CWTP_PROFILE_DATA_REC_LEN, 4, false);
+
+/* reg_cwtp_profile_max
+ * Maximum Average Queue size of the profile in cells
+ * Access: RW
+ */
+MLXSW_ITEM32_INDEXED(reg, cwtp, profile_max, MLXSW_REG_CWTP_BASE_LEN,
+		     0, 20, MLXSW_REG_CWTP_PROFILE_DATA_REC_LEN, 4, false);
+
+#define MLXSW_REG_CWTP_MIN_VALUE 64
+#define MLXSW_REG_CWTP_MAX_PROFILE 2
+#define MLXSW_REG_CWTP_DEFAULT_PROFILE 1
+
+static inline void mlxsw_reg_cwtp_pack(char *payload, u16 local_port,
+				       u8 traffic_class)
+{
+	int i;
+
+	MLXSW_REG_ZERO(cwtp, payload);
+	mlxsw_reg_cwtp_local_port_set(payload, local_port);
+	mlxsw_reg_cwtp_traffic_class_set(payload, traffic_class);
+
+	for (i = 0; i <= MLXSW_REG_CWTP_MAX_PROFILE; i++) {
+		mlxsw_reg_cwtp_profile_min_set(payload, i,
+					       MLXSW_REG_CWTP_MIN_VALUE);
+		mlxsw_reg_cwtp_profile_max_set(payload, i,
+					       MLXSW_REG_CWTP_MIN_VALUE);
+	}
+}
+
+#define MLXSW_REG_CWTP_PROFILE_TO_INDEX(profile) (profile - 1)
+
+static inline void
+mlxsw_reg_cwtp_profile_pack(char *payload, u8 profile, u32 min, u32 max,
+			    u32 probability)
+{
+	u8 index = MLXSW_REG_CWTP_PROFILE_TO_INDEX(profile);
+
+	mlxsw_reg_cwtp_profile_min_set(payload, index, min);
+	mlxsw_reg_cwtp_profile_max_set(payload, index, max);
+	mlxsw_reg_cwtp_profile_percent_set(payload, index, probability);
+}
+
+/* CWTPM - Congestion WRED ECN TClass and Pool Mapping
+ * ---------------------------------------------------
+ * The CWTPM register maps each egress port and traffic class to profile num.
+ */
+#define MLXSW_REG_CWTPM_ID 0x2803
+#define MLXSW_REG_CWTPM_LEN 0x44
+
+MLXSW_REG_DEFINE(cwtpm, MLXSW_REG_CWTPM_ID, MLXSW_REG_CWTPM_LEN);
+
+/* reg_cwtpm_local_port
+ * Local port number
+ * Not supported for CPU port
+ * Access: Index
+ */
+MLXSW_ITEM32_LP(reg, cwtpm, 0x00, 16, 0x00, 12);
+
+/* reg_cwtpm_traffic_class
+ * Traffic Class to configure
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, cwtpm, traffic_class, 32, 0, 8);
+
+/* reg_cwtpm_ew
+ * Control enablement of WRED for traffic class:
+ * 0 - Disable
+ * 1 - Enable
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, cwtpm, ew, 36, 1, 1);
+
+/* reg_cwtpm_ee
+ * Control enablement of ECN for traffic class:
+ * 0 - Disable
+ * 1 - Enable
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, cwtpm, ee, 36, 0, 1);
+
+/* reg_cwtpm_tcp_g
+ * TCP Green Profile.
+ * Index of the profile within {port, traffic class} to use.
+ * 0 for disabling both WRED and ECN for this type of traffic.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, cwtpm, tcp_g, 52, 0, 2);
+
+/* reg_cwtpm_tcp_y
+ * TCP Yellow Profile.
+ * Index of the profile within {port, traffic class} to use.
+ * 0 for disabling both WRED and ECN for this type of traffic.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, cwtpm, tcp_y, 56, 16, 2);
+
+/* reg_cwtpm_tcp_r
+ * TCP Red Profile.
+ * Index of the profile within {port, traffic class} to use.
+ * 0 for disabling both WRED and ECN for this type of traffic.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, cwtpm, tcp_r, 56, 0, 2);
+
+/* reg_cwtpm_ntcp_g
+ * Non-TCP Green Profile.
+ * Index of the profile within {port, traffic class} to use.
+ * 0 for disabling both WRED and ECN for this type of traffic.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, cwtpm, ntcp_g, 60, 0, 2);
+
+/* reg_cwtpm_ntcp_y
+ * Non-TCP Yellow Profile.
+ * Index of the profile within {port, traffic class} to use.
+ * 0 for disabling both WRED and ECN for this type of traffic.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, cwtpm, ntcp_y, 64, 16, 2);
+
+/* reg_cwtpm_ntcp_r
+ * Non-TCP Red Profile.
+ * Index of the profile within {port, traffic class} to use.
+ * 0 for disabling both WRED and ECN for this type of traffic.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, cwtpm, ntcp_r, 64, 0, 2);
+
+#define MLXSW_REG_CWTPM_RESET_PROFILE 0
+
+static inline void mlxsw_reg_cwtpm_pack(char *payload, u16 local_port,
+					u8 traffic_class, u8 profile,
+					bool wred, bool ecn)
+{
+	MLXSW_REG_ZERO(cwtpm, payload);
+	mlxsw_reg_cwtpm_local_port_set(payload, local_port);
+	mlxsw_reg_cwtpm_traffic_class_set(payload, traffic_class);
+	mlxsw_reg_cwtpm_ew_set(payload, wred);
+	mlxsw_reg_cwtpm_ee_set(payload, ecn);
+	mlxsw_reg_cwtpm_tcp_g_set(payload, profile);
+	mlxsw_reg_cwtpm_tcp_y_set(payload, profile);
+	mlxsw_reg_cwtpm_tcp_r_set(payload, profile);
+	mlxsw_reg_cwtpm_ntcp_g_set(payload, profile);
+	mlxsw_reg_cwtpm_ntcp_y_set(payload, profile);
+	mlxsw_reg_cwtpm_ntcp_r_set(payload, profile);
+}
+
+/* PGCR - Policy-Engine General Configuration Register
+ * ---------------------------------------------------
+ * This register configures general Policy-Engine settings.
+ */
+#define MLXSW_REG_PGCR_ID 0x3001
+#define MLXSW_REG_PGCR_LEN 0x20
+
+MLXSW_REG_DEFINE(pgcr, MLXSW_REG_PGCR_ID, MLXSW_REG_PGCR_LEN);
+
+/* reg_pgcr_default_action_pointer_base
+ * Default action pointer base. Each region has a default action pointer
+ * which is equal to default_action_pointer_base + region_id.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, pgcr, default_action_pointer_base, 0x1C, 0, 24);
+
+static inline void mlxsw_reg_pgcr_pack(char *payload, u32 pointer_base)
+{
+	MLXSW_REG_ZERO(pgcr, payload);
+	mlxsw_reg_pgcr_default_action_pointer_base_set(payload, pointer_base);
+}
+
+/* PPBT - Policy-Engine Port Binding Table
+ * ---------------------------------------
+ * This register is used for configuration of the Port Binding Table.
+ */
+#define MLXSW_REG_PPBT_ID 0x3002
+#define MLXSW_REG_PPBT_LEN 0x14
+
+MLXSW_REG_DEFINE(ppbt, MLXSW_REG_PPBT_ID, MLXSW_REG_PPBT_LEN);
+
+enum mlxsw_reg_pxbt_e {
+	MLXSW_REG_PXBT_E_IACL,
+	MLXSW_REG_PXBT_E_EACL,
+};
+
+/* reg_ppbt_e
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, ppbt, e, 0x00, 31, 1);
+
+enum mlxsw_reg_pxbt_op {
+	MLXSW_REG_PXBT_OP_BIND,
+	MLXSW_REG_PXBT_OP_UNBIND,
+};
+
+/* reg_ppbt_op
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ppbt, op, 0x00, 28, 3);
+
+/* reg_ppbt_local_port
+ * Local port. Not including CPU port.
+ * Access: Index
+ */
+MLXSW_ITEM32_LP(reg, ppbt, 0x00, 16, 0x00, 12);
+
+/* reg_ppbt_g
+ * group - When set, the binding is of an ACL group. When cleared,
+ * the binding is of an ACL.
+ * Must be set to 1 for Spectrum.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ppbt, g, 0x10, 31, 1);
+
+/* reg_ppbt_acl_info
+ * ACL/ACL group identifier. If the g bit is set, this field should hold
+ * the acl_group_id, else it should hold the acl_id.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ppbt, acl_info, 0x10, 0, 16);
+
+static inline void mlxsw_reg_ppbt_pack(char *payload, enum mlxsw_reg_pxbt_e e,
+				       enum mlxsw_reg_pxbt_op op,
+				       u16 local_port, u16 acl_info)
+{
+	MLXSW_REG_ZERO(ppbt, payload);
+	mlxsw_reg_ppbt_e_set(payload, e);
+	mlxsw_reg_ppbt_op_set(payload, op);
+	mlxsw_reg_ppbt_local_port_set(payload, local_port);
+	mlxsw_reg_ppbt_g_set(payload, true);
+	mlxsw_reg_ppbt_acl_info_set(payload, acl_info);
+}
+
+/* PACL - Policy-Engine ACL Register
+ * ---------------------------------
+ * This register is used for configuration of the ACL.
+ */
+#define MLXSW_REG_PACL_ID 0x3004
+#define MLXSW_REG_PACL_LEN 0x70
+
+MLXSW_REG_DEFINE(pacl, MLXSW_REG_PACL_ID, MLXSW_REG_PACL_LEN);
+
+/* reg_pacl_v
+ * Valid. Setting the v bit makes the ACL valid. It should not be cleared
+ * while the ACL is bounded to either a port, VLAN or ACL rule.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, pacl, v, 0x00, 24, 1);
+
+/* reg_pacl_acl_id
+ * An identifier representing the ACL (managed by software)
+ * Range 0 .. cap_max_acl_regions - 1
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, pacl, acl_id, 0x08, 0, 16);
+
+#define MLXSW_REG_PXXX_TCAM_REGION_INFO_LEN 16
+
+/* reg_pacl_tcam_region_info
+ * Opaque object that represents a TCAM region.
+ * Obtained through PTAR register.
+ * Access: RW
+ */
+MLXSW_ITEM_BUF(reg, pacl, tcam_region_info, 0x30,
+	       MLXSW_REG_PXXX_TCAM_REGION_INFO_LEN);
+
+static inline void mlxsw_reg_pacl_pack(char *payload, u16 acl_id,
+				       bool valid, const char *tcam_region_info)
+{
+	MLXSW_REG_ZERO(pacl, payload);
+	mlxsw_reg_pacl_acl_id_set(payload, acl_id);
+	mlxsw_reg_pacl_v_set(payload, valid);
+	mlxsw_reg_pacl_tcam_region_info_memcpy_to(payload, tcam_region_info);
+}
+
+/* PAGT - Policy-Engine ACL Group Table
+ * ------------------------------------
+ * This register is used for configuration of the ACL Group Table.
+ */
+#define MLXSW_REG_PAGT_ID 0x3005
+#define MLXSW_REG_PAGT_BASE_LEN 0x30
+#define MLXSW_REG_PAGT_ACL_LEN 4
+#define MLXSW_REG_PAGT_ACL_MAX_NUM 16
+#define MLXSW_REG_PAGT_LEN (MLXSW_REG_PAGT_BASE_LEN + \
+		MLXSW_REG_PAGT_ACL_MAX_NUM * MLXSW_REG_PAGT_ACL_LEN)
+
+MLXSW_REG_DEFINE(pagt, MLXSW_REG_PAGT_ID, MLXSW_REG_PAGT_LEN);
+
+/* reg_pagt_size
+ * Number of ACLs in the group.
+ * Size 0 invalidates a group.
+ * Range 0 .. cap_max_acl_group_size (hard coded to 16 for now)
+ * Total number of ACLs in all groups must be lower or equal
+ * to cap_max_acl_tot_groups
+ * Note: a group which is binded must not be invalidated
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, pagt, size, 0x00, 0, 8);
+
+/* reg_pagt_acl_group_id
+ * An identifier (numbered from 0..cap_max_acl_groups-1) representing
+ * the ACL Group identifier (managed by software).
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, pagt, acl_group_id, 0x08, 0, 16);
+
+/* reg_pagt_multi
+ * Multi-ACL
+ * 0 - This ACL is the last ACL in the multi-ACL
+ * 1 - This ACL is part of a multi-ACL
+ * Access: RW
+ */
+MLXSW_ITEM32_INDEXED(reg, pagt, multi, 0x30, 31, 1, 0x04, 0x00, false);
+
+/* reg_pagt_acl_id
+ * ACL identifier
+ * Access: RW
+ */
+MLXSW_ITEM32_INDEXED(reg, pagt, acl_id, 0x30, 0, 16, 0x04, 0x00, false);
+
+static inline void mlxsw_reg_pagt_pack(char *payload, u16 acl_group_id)
+{
+	MLXSW_REG_ZERO(pagt, payload);
+	mlxsw_reg_pagt_acl_group_id_set(payload, acl_group_id);
+}
+
+static inline void mlxsw_reg_pagt_acl_id_pack(char *payload, int index,
+					      u16 acl_id, bool multi)
+{
+	u8 size = mlxsw_reg_pagt_size_get(payload);
+
+	if (index >= size)
+		mlxsw_reg_pagt_size_set(payload, index + 1);
+	mlxsw_reg_pagt_multi_set(payload, index, multi);
+	mlxsw_reg_pagt_acl_id_set(payload, index, acl_id);
+}
+
+/* PTAR - Policy-Engine TCAM Allocation Register
+ * ---------------------------------------------
+ * This register is used for allocation of regions in the TCAM.
+ * Note: Query method is not supported on this register.
+ */
+#define MLXSW_REG_PTAR_ID 0x3006
+#define MLXSW_REG_PTAR_BASE_LEN 0x20
+#define MLXSW_REG_PTAR_KEY_ID_LEN 1
+#define MLXSW_REG_PTAR_KEY_ID_MAX_NUM 16
+#define MLXSW_REG_PTAR_LEN (MLXSW_REG_PTAR_BASE_LEN + \
+		MLXSW_REG_PTAR_KEY_ID_MAX_NUM * MLXSW_REG_PTAR_KEY_ID_LEN)
+
+MLXSW_REG_DEFINE(ptar, MLXSW_REG_PTAR_ID, MLXSW_REG_PTAR_LEN);
+
+enum mlxsw_reg_ptar_op {
+	/* allocate a TCAM region */
+	MLXSW_REG_PTAR_OP_ALLOC,
+	/* resize a TCAM region */
+	MLXSW_REG_PTAR_OP_RESIZE,
+	/* deallocate TCAM region */
+	MLXSW_REG_PTAR_OP_FREE,
+	/* test allocation */
+	MLXSW_REG_PTAR_OP_TEST,
+};
+
+/* reg_ptar_op
+ * Access: OP
+ */
+MLXSW_ITEM32(reg, ptar, op, 0x00, 28, 4);
+
+/* reg_ptar_action_set_type
+ * Type of action set to be used on this region.
+ * For Spectrum and Spectrum-2, this is always type 2 - "flexible"
+ * Access: WO
+ */
+MLXSW_ITEM32(reg, ptar, action_set_type, 0x00, 16, 8);
+
+enum mlxsw_reg_ptar_key_type {
+	MLXSW_REG_PTAR_KEY_TYPE_FLEX = 0x50, /* Spetrum */
+	MLXSW_REG_PTAR_KEY_TYPE_FLEX2 = 0x51, /* Spectrum-2 */
+};
+
+/* reg_ptar_key_type
+ * TCAM key type for the region.
+ * Access: WO
+ */
+MLXSW_ITEM32(reg, ptar, key_type, 0x00, 0, 8);
+
+/* reg_ptar_region_size
+ * TCAM region size. When allocating/resizing this is the requested size,
+ * the response is the actual size. Note that actual size may be
+ * larger than requested.
+ * Allowed range 1 .. cap_max_rules-1
+ * Reserved during op deallocate.
+ * Access: WO
+ */
+MLXSW_ITEM32(reg, ptar, region_size, 0x04, 0, 16);
+
+/* reg_ptar_region_id
+ * Region identifier
+ * Range 0 .. cap_max_regions-1
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, ptar, region_id, 0x08, 0, 16);
+
+/* reg_ptar_tcam_region_info
+ * Opaque object that represents the TCAM region.
+ * Returned when allocating a region.
+ * Provided by software for ACL generation and region deallocation and resize.
+ * Access: RW
+ */
+MLXSW_ITEM_BUF(reg, ptar, tcam_region_info, 0x10,
+	       MLXSW_REG_PXXX_TCAM_REGION_INFO_LEN);
+
+/* reg_ptar_flexible_key_id
+ * Identifier of the Flexible Key.
+ * Only valid if key_type == "FLEX_KEY"
+ * The key size will be rounded up to one of the following values:
+ * 9B, 18B, 36B, 54B.
+ * This field is reserved for in resize operation.
+ * Access: WO
+ */
+MLXSW_ITEM8_INDEXED(reg, ptar, flexible_key_id, 0x20, 0, 8,
+		    MLXSW_REG_PTAR_KEY_ID_LEN, 0x00, false);
+
+static inline void mlxsw_reg_ptar_pack(char *payload, enum mlxsw_reg_ptar_op op,
+				       enum mlxsw_reg_ptar_key_type key_type,
+				       u16 region_size, u16 region_id,
+				       const char *tcam_region_info)
+{
+	MLXSW_REG_ZERO(ptar, payload);
+	mlxsw_reg_ptar_op_set(payload, op);
+	mlxsw_reg_ptar_action_set_type_set(payload, 2); /* "flexible" */
+	mlxsw_reg_ptar_key_type_set(payload, key_type);
+	mlxsw_reg_ptar_region_size_set(payload, region_size);
+	mlxsw_reg_ptar_region_id_set(payload, region_id);
+	mlxsw_reg_ptar_tcam_region_info_memcpy_to(payload, tcam_region_info);
+}
+
+static inline void mlxsw_reg_ptar_key_id_pack(char *payload, int index,
+					      u16 key_id)
+{
+	mlxsw_reg_ptar_flexible_key_id_set(payload, index, key_id);
+}
+
+static inline void mlxsw_reg_ptar_unpack(char *payload, char *tcam_region_info)
+{
+	mlxsw_reg_ptar_tcam_region_info_memcpy_from(payload, tcam_region_info);
+}
+
+/* PPBS - Policy-Engine Policy Based Switching Register
+ * ----------------------------------------------------
+ * This register retrieves and sets Policy Based Switching Table entries.
+ */
+#define MLXSW_REG_PPBS_ID 0x300C
+#define MLXSW_REG_PPBS_LEN 0x14
+
+MLXSW_REG_DEFINE(ppbs, MLXSW_REG_PPBS_ID, MLXSW_REG_PPBS_LEN);
+
+/* reg_ppbs_pbs_ptr
+ * Index into the PBS table.
+ * For Spectrum, the index points to the KVD Linear.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, ppbs, pbs_ptr, 0x08, 0, 24);
+
+/* reg_ppbs_system_port
+ * Unique port identifier for the final destination of the packet.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ppbs, system_port, 0x10, 0, 16);
+
+static inline void mlxsw_reg_ppbs_pack(char *payload, u32 pbs_ptr,
+				       u16 system_port)
+{
+	MLXSW_REG_ZERO(ppbs, payload);
+	mlxsw_reg_ppbs_pbs_ptr_set(payload, pbs_ptr);
+	mlxsw_reg_ppbs_system_port_set(payload, system_port);
+}
+
+/* PRCR - Policy-Engine Rules Copy Register
+ * ----------------------------------------
+ * This register is used for accessing rules within a TCAM region.
+ */
+#define MLXSW_REG_PRCR_ID 0x300D
+#define MLXSW_REG_PRCR_LEN 0x40
+
+MLXSW_REG_DEFINE(prcr, MLXSW_REG_PRCR_ID, MLXSW_REG_PRCR_LEN);
+
+enum mlxsw_reg_prcr_op {
+	/* Move rules. Moves the rules from "tcam_region_info" starting
+	 * at offset "offset" to "dest_tcam_region_info"
+	 * at offset "dest_offset."
+	 */
+	MLXSW_REG_PRCR_OP_MOVE,
+	/* Copy rules. Copies the rules from "tcam_region_info" starting
+	 * at offset "offset" to "dest_tcam_region_info"
+	 * at offset "dest_offset."
+	 */
+	MLXSW_REG_PRCR_OP_COPY,
+};
+
+/* reg_prcr_op
+ * Access: OP
+ */
+MLXSW_ITEM32(reg, prcr, op, 0x00, 28, 4);
+
+/* reg_prcr_offset
+ * Offset within the source region to copy/move from.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, prcr, offset, 0x00, 0, 16);
+
+/* reg_prcr_size
+ * The number of rules to copy/move.
+ * Access: WO
+ */
+MLXSW_ITEM32(reg, prcr, size, 0x04, 0, 16);
+
+/* reg_prcr_tcam_region_info
+ * Opaque object that represents the source TCAM region.
+ * Access: Index
+ */
+MLXSW_ITEM_BUF(reg, prcr, tcam_region_info, 0x10,
+	       MLXSW_REG_PXXX_TCAM_REGION_INFO_LEN);
+
+/* reg_prcr_dest_offset
+ * Offset within the source region to copy/move to.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, prcr, dest_offset, 0x20, 0, 16);
+
+/* reg_prcr_dest_tcam_region_info
+ * Opaque object that represents the destination TCAM region.
+ * Access: Index
+ */
+MLXSW_ITEM_BUF(reg, prcr, dest_tcam_region_info, 0x30,
+	       MLXSW_REG_PXXX_TCAM_REGION_INFO_LEN);
+
+static inline void mlxsw_reg_prcr_pack(char *payload, enum mlxsw_reg_prcr_op op,
+				       const char *src_tcam_region_info,
+				       u16 src_offset,
+				       const char *dest_tcam_region_info,
+				       u16 dest_offset, u16 size)
+{
+	MLXSW_REG_ZERO(prcr, payload);
+	mlxsw_reg_prcr_op_set(payload, op);
+	mlxsw_reg_prcr_offset_set(payload, src_offset);
+	mlxsw_reg_prcr_size_set(payload, size);
+	mlxsw_reg_prcr_tcam_region_info_memcpy_to(payload,
+						  src_tcam_region_info);
+	mlxsw_reg_prcr_dest_offset_set(payload, dest_offset);
+	mlxsw_reg_prcr_dest_tcam_region_info_memcpy_to(payload,
+						       dest_tcam_region_info);
+}
+
+/* PEFA - Policy-Engine Extended Flexible Action Register
+ * ------------------------------------------------------
+ * This register is used for accessing an extended flexible action entry
+ * in the central KVD Linear Database.
+ */
+#define MLXSW_REG_PEFA_ID 0x300F
+#define MLXSW_REG_PEFA_LEN 0xB0
+
+MLXSW_REG_DEFINE(pefa, MLXSW_REG_PEFA_ID, MLXSW_REG_PEFA_LEN);
+
+/* reg_pefa_index
+ * Index in the KVD Linear Centralized Database.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, pefa, index, 0x00, 0, 24);
+
+/* reg_pefa_a
+ * Index in the KVD Linear Centralized Database.
+ * Activity
+ * For a new entry: set if ca=0, clear if ca=1
+ * Set if a packet lookup has hit on the specific entry
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, pefa, a, 0x04, 29, 1);
+
+/* reg_pefa_ca
+ * Clear activity
+ * When write: activity is according to this field
+ * When read: after reading the activity is cleared according to ca
+ * Access: OP
+ */
+MLXSW_ITEM32(reg, pefa, ca, 0x04, 24, 1);
+
+#define MLXSW_REG_FLEX_ACTION_SET_LEN 0xA8
+
+/* reg_pefa_flex_action_set
+ * Action-set to perform when rule is matched.
+ * Must be zero padded if action set is shorter.
+ * Access: RW
+ */
+MLXSW_ITEM_BUF(reg, pefa, flex_action_set, 0x08, MLXSW_REG_FLEX_ACTION_SET_LEN);
+
+static inline void mlxsw_reg_pefa_pack(char *payload, u32 index, bool ca,
+				       const char *flex_action_set)
+{
+	MLXSW_REG_ZERO(pefa, payload);
+	mlxsw_reg_pefa_index_set(payload, index);
+	mlxsw_reg_pefa_ca_set(payload, ca);
+	if (flex_action_set)
+		mlxsw_reg_pefa_flex_action_set_memcpy_to(payload,
+							 flex_action_set);
+}
+
+static inline void mlxsw_reg_pefa_unpack(char *payload, bool *p_a)
+{
+	*p_a = mlxsw_reg_pefa_a_get(payload);
+}
+
+/* PEMRBT - Policy-Engine Multicast Router Binding Table Register
+ * --------------------------------------------------------------
+ * This register is used for binding Multicast router to an ACL group
+ * that serves the MC router.
+ * This register is not supported by SwitchX/-2 and Spectrum.
+ */
+#define MLXSW_REG_PEMRBT_ID 0x3014
+#define MLXSW_REG_PEMRBT_LEN 0x14
+
+MLXSW_REG_DEFINE(pemrbt, MLXSW_REG_PEMRBT_ID, MLXSW_REG_PEMRBT_LEN);
+
+enum mlxsw_reg_pemrbt_protocol {
+	MLXSW_REG_PEMRBT_PROTO_IPV4,
+	MLXSW_REG_PEMRBT_PROTO_IPV6,
+};
+
+/* reg_pemrbt_protocol
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, pemrbt, protocol, 0x00, 0, 1);
+
+/* reg_pemrbt_group_id
+ * ACL group identifier.
+ * Range 0..cap_max_acl_groups-1
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, pemrbt, group_id, 0x10, 0, 16);
+
+static inline void
+mlxsw_reg_pemrbt_pack(char *payload, enum mlxsw_reg_pemrbt_protocol protocol,
+		      u16 group_id)
+{
+	MLXSW_REG_ZERO(pemrbt, payload);
+	mlxsw_reg_pemrbt_protocol_set(payload, protocol);
+	mlxsw_reg_pemrbt_group_id_set(payload, group_id);
+}
+
+/* PTCE-V2 - Policy-Engine TCAM Entry Register Version 2
+ * -----------------------------------------------------
+ * This register is used for accessing rules within a TCAM region.
+ * It is a new version of PTCE in order to support wider key,
+ * mask and action within a TCAM region. This register is not supported
+ * by SwitchX and SwitchX-2.
+ */
+#define MLXSW_REG_PTCE2_ID 0x3017
+#define MLXSW_REG_PTCE2_LEN 0x1D8
+
+MLXSW_REG_DEFINE(ptce2, MLXSW_REG_PTCE2_ID, MLXSW_REG_PTCE2_LEN);
+
+/* reg_ptce2_v
+ * Valid.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ptce2, v, 0x00, 31, 1);
+
+/* reg_ptce2_a
+ * Activity. Set if a packet lookup has hit on the specific entry.
+ * To clear the "a" bit, use "clear activity" op or "clear on read" op.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, ptce2, a, 0x00, 30, 1);
+
+enum mlxsw_reg_ptce2_op {
+	/* Read operation. */
+	MLXSW_REG_PTCE2_OP_QUERY_READ = 0,
+	/* clear on read operation. Used to read entry
+	 * and clear Activity bit.
+	 */
+	MLXSW_REG_PTCE2_OP_QUERY_CLEAR_ON_READ = 1,
+	/* Write operation. Used to write a new entry to the table.
+	 * All R/W fields are relevant for new entry. Activity bit is set
+	 * for new entries - Note write with v = 0 will delete the entry.
+	 */
+	MLXSW_REG_PTCE2_OP_WRITE_WRITE = 0,
+	/* Update action. Only action set will be updated. */
+	MLXSW_REG_PTCE2_OP_WRITE_UPDATE = 1,
+	/* Clear activity. A bit is cleared for the entry. */
+	MLXSW_REG_PTCE2_OP_WRITE_CLEAR_ACTIVITY = 2,
+};
+
+/* reg_ptce2_op
+ * Access: OP
+ */
+MLXSW_ITEM32(reg, ptce2, op, 0x00, 20, 3);
+
+/* reg_ptce2_offset
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, ptce2, offset, 0x00, 0, 16);
+
+/* reg_ptce2_priority
+ * Priority of the rule, higher values win. The range is 1..cap_kvd_size-1.
+ * Note: priority does not have to be unique per rule.
+ * Within a region, higher priority should have lower offset (no limitation
+ * between regions in a multi-region).
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ptce2, priority, 0x04, 0, 24);
+
+/* reg_ptce2_tcam_region_info
+ * Opaque object that represents the TCAM region.
+ * Access: Index
+ */
+MLXSW_ITEM_BUF(reg, ptce2, tcam_region_info, 0x10,
+	       MLXSW_REG_PXXX_TCAM_REGION_INFO_LEN);
+
+#define MLXSW_REG_PTCEX_FLEX_KEY_BLOCKS_LEN 96
+
+/* reg_ptce2_flex_key_blocks
+ * ACL Key.
+ * Access: RW
+ */
+MLXSW_ITEM_BUF(reg, ptce2, flex_key_blocks, 0x20,
+	       MLXSW_REG_PTCEX_FLEX_KEY_BLOCKS_LEN);
+
+/* reg_ptce2_mask
+ * mask- in the same size as key. A bit that is set directs the TCAM
+ * to compare the corresponding bit in key. A bit that is clear directs
+ * the TCAM to ignore the corresponding bit in key.
+ * Access: RW
+ */
+MLXSW_ITEM_BUF(reg, ptce2, mask, 0x80,
+	       MLXSW_REG_PTCEX_FLEX_KEY_BLOCKS_LEN);
+
+/* reg_ptce2_flex_action_set
+ * ACL action set.
+ * Access: RW
+ */
+MLXSW_ITEM_BUF(reg, ptce2, flex_action_set, 0xE0,
+	       MLXSW_REG_FLEX_ACTION_SET_LEN);
+
+static inline void mlxsw_reg_ptce2_pack(char *payload, bool valid,
+					enum mlxsw_reg_ptce2_op op,
+					const char *tcam_region_info,
+					u16 offset, u32 priority)
+{
+	MLXSW_REG_ZERO(ptce2, payload);
+	mlxsw_reg_ptce2_v_set(payload, valid);
+	mlxsw_reg_ptce2_op_set(payload, op);
+	mlxsw_reg_ptce2_offset_set(payload, offset);
+	mlxsw_reg_ptce2_priority_set(payload, priority);
+	mlxsw_reg_ptce2_tcam_region_info_memcpy_to(payload, tcam_region_info);
+}
+
+/* PERPT - Policy-Engine ERP Table Register
+ * ----------------------------------------
+ * This register adds and removes eRPs from the eRP table.
+ */
+#define MLXSW_REG_PERPT_ID 0x3021
+#define MLXSW_REG_PERPT_LEN 0x80
+
+MLXSW_REG_DEFINE(perpt, MLXSW_REG_PERPT_ID, MLXSW_REG_PERPT_LEN);
+
+/* reg_perpt_erpt_bank
+ * eRP table bank.
+ * Range 0 .. cap_max_erp_table_banks - 1
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, perpt, erpt_bank, 0x00, 16, 4);
+
+/* reg_perpt_erpt_index
+ * Index to eRP table within the eRP bank.
+ * Range is 0 .. cap_max_erp_table_bank_size - 1
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, perpt, erpt_index, 0x00, 0, 8);
+
+enum mlxsw_reg_perpt_key_size {
+	MLXSW_REG_PERPT_KEY_SIZE_2KB,
+	MLXSW_REG_PERPT_KEY_SIZE_4KB,
+	MLXSW_REG_PERPT_KEY_SIZE_8KB,
+	MLXSW_REG_PERPT_KEY_SIZE_12KB,
+};
+
+/* reg_perpt_key_size
+ * Access: OP
+ */
+MLXSW_ITEM32(reg, perpt, key_size, 0x04, 0, 4);
+
+/* reg_perpt_bf_bypass
+ * 0 - The eRP is used only if bloom filter state is set for the given
+ * rule.
+ * 1 - The eRP is used regardless of bloom filter state.
+ * The bypass is an OR condition of region_id or eRP. See PERCR.bf_bypass
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, perpt, bf_bypass, 0x08, 8, 1);
+
+/* reg_perpt_erp_id
+ * eRP ID for use by the rules.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, perpt, erp_id, 0x08, 0, 4);
+
+/* reg_perpt_erpt_base_bank
+ * Base eRP table bank, points to head of erp_vector
+ * Range is 0 .. cap_max_erp_table_banks - 1
+ * Access: OP
+ */
+MLXSW_ITEM32(reg, perpt, erpt_base_bank, 0x0C, 16, 4);
+
+/* reg_perpt_erpt_base_index
+ * Base index to eRP table within the eRP bank
+ * Range is 0 .. cap_max_erp_table_bank_size - 1
+ * Access: OP
+ */
+MLXSW_ITEM32(reg, perpt, erpt_base_index, 0x0C, 0, 8);
+
+/* reg_perpt_erp_index_in_vector
+ * eRP index in the vector.
+ * Access: OP
+ */
+MLXSW_ITEM32(reg, perpt, erp_index_in_vector, 0x10, 0, 4);
+
+/* reg_perpt_erp_vector
+ * eRP vector.
+ * Access: OP
+ */
+MLXSW_ITEM_BIT_ARRAY(reg, perpt, erp_vector, 0x14, 4, 1);
+
+/* reg_perpt_mask
+ * Mask
+ * 0 - A-TCAM will ignore the bit in key
+ * 1 - A-TCAM will compare the bit in key
+ * Access: RW
+ */
+MLXSW_ITEM_BUF(reg, perpt, mask, 0x20, MLXSW_REG_PTCEX_FLEX_KEY_BLOCKS_LEN);
+
+static inline void mlxsw_reg_perpt_erp_vector_pack(char *payload,
+						   unsigned long *erp_vector,
+						   unsigned long size)
+{
+	unsigned long bit;
+
+	for_each_set_bit(bit, erp_vector, size)
+		mlxsw_reg_perpt_erp_vector_set(payload, bit, true);
+}
+
+static inline void
+mlxsw_reg_perpt_pack(char *payload, u8 erpt_bank, u8 erpt_index,
+		     enum mlxsw_reg_perpt_key_size key_size, u8 erp_id,
+		     u8 erpt_base_bank, u8 erpt_base_index, u8 erp_index,
+		     char *mask)
+{
+	MLXSW_REG_ZERO(perpt, payload);
+	mlxsw_reg_perpt_erpt_bank_set(payload, erpt_bank);
+	mlxsw_reg_perpt_erpt_index_set(payload, erpt_index);
+	mlxsw_reg_perpt_key_size_set(payload, key_size);
+	mlxsw_reg_perpt_bf_bypass_set(payload, false);
+	mlxsw_reg_perpt_erp_id_set(payload, erp_id);
+	mlxsw_reg_perpt_erpt_base_bank_set(payload, erpt_base_bank);
+	mlxsw_reg_perpt_erpt_base_index_set(payload, erpt_base_index);
+	mlxsw_reg_perpt_erp_index_in_vector_set(payload, erp_index);
+	mlxsw_reg_perpt_mask_memcpy_to(payload, mask);
+}
+
+/* PERAR - Policy-Engine Region Association Register
+ * -------------------------------------------------
+ * This register associates a hw region for region_id's. Changing on the fly
+ * is supported by the device.
+ */
+#define MLXSW_REG_PERAR_ID 0x3026
+#define MLXSW_REG_PERAR_LEN 0x08
+
+MLXSW_REG_DEFINE(perar, MLXSW_REG_PERAR_ID, MLXSW_REG_PERAR_LEN);
+
+/* reg_perar_region_id
+ * Region identifier
+ * Range 0 .. cap_max_regions-1
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, perar, region_id, 0x00, 0, 16);
+
+static inline unsigned int
+mlxsw_reg_perar_hw_regions_needed(unsigned int block_num)
+{
+	return DIV_ROUND_UP(block_num, 4);
+}
+
+/* reg_perar_hw_region
+ * HW Region
+ * Range 0 .. cap_max_regions-1
+ * Default: hw_region = region_id
+ * For a 8 key block region, 2 consecutive regions are used
+ * For a 12 key block region, 3 consecutive regions are used
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, perar, hw_region, 0x04, 0, 16);
+
+static inline void mlxsw_reg_perar_pack(char *payload, u16 region_id,
+					u16 hw_region)
+{
+	MLXSW_REG_ZERO(perar, payload);
+	mlxsw_reg_perar_region_id_set(payload, region_id);
+	mlxsw_reg_perar_hw_region_set(payload, hw_region);
+}
+
+/* PTCE-V3 - Policy-Engine TCAM Entry Register Version 3
+ * -----------------------------------------------------
+ * This register is a new version of PTCE-V2 in order to support the
+ * A-TCAM. This register is not supported by SwitchX/-2 and Spectrum.
+ */
+#define MLXSW_REG_PTCE3_ID 0x3027
+#define MLXSW_REG_PTCE3_LEN 0xF0
+
+MLXSW_REG_DEFINE(ptce3, MLXSW_REG_PTCE3_ID, MLXSW_REG_PTCE3_LEN);
+
+/* reg_ptce3_v
+ * Valid.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ptce3, v, 0x00, 31, 1);
+
+enum mlxsw_reg_ptce3_op {
+	/* Write operation. Used to write a new entry to the table.
+	 * All R/W fields are relevant for new entry. Activity bit is set
+	 * for new entries. Write with v = 0 will delete the entry. Must
+	 * not be used if an entry exists.
+	 */
+	 MLXSW_REG_PTCE3_OP_WRITE_WRITE = 0,
+	 /* Update operation */
+	 MLXSW_REG_PTCE3_OP_WRITE_UPDATE = 1,
+	 /* Read operation */
+	 MLXSW_REG_PTCE3_OP_QUERY_READ = 0,
+};
+
+/* reg_ptce3_op
+ * Access: OP
+ */
+MLXSW_ITEM32(reg, ptce3, op, 0x00, 20, 3);
+
+/* reg_ptce3_priority
+ * Priority of the rule. Higher values win.
+ * For Spectrum-2 range is 1..cap_kvd_size - 1
+ * Note: Priority does not have to be unique per rule.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ptce3, priority, 0x04, 0, 24);
+
+/* reg_ptce3_tcam_region_info
+ * Opaque object that represents the TCAM region.
+ * Access: Index
+ */
+MLXSW_ITEM_BUF(reg, ptce3, tcam_region_info, 0x10,
+	       MLXSW_REG_PXXX_TCAM_REGION_INFO_LEN);
+
+/* reg_ptce3_flex2_key_blocks
+ * ACL key. The key must be masked according to eRP (if exists) or
+ * according to master mask.
+ * Access: Index
+ */
+MLXSW_ITEM_BUF(reg, ptce3, flex2_key_blocks, 0x20,
+	       MLXSW_REG_PTCEX_FLEX_KEY_BLOCKS_LEN);
+
+/* reg_ptce3_erp_id
+ * eRP ID.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, ptce3, erp_id, 0x80, 0, 4);
+
+/* reg_ptce3_delta_start
+ * Start point of delta_value and delta_mask, in bits. Must not exceed
+ * num_key_blocks * 36 - 8. Reserved when delta_mask = 0.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, ptce3, delta_start, 0x84, 0, 10);
+
+/* reg_ptce3_delta_mask
+ * Delta mask.
+ * 0 - Ignore relevant bit in delta_value
+ * 1 - Compare relevant bit in delta_value
+ * Delta mask must not be set for reserved fields in the key blocks.
+ * Note: No delta when no eRPs. Thus, for regions with
+ * PERERP.erpt_pointer_valid = 0 the delta mask must be 0.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, ptce3, delta_mask, 0x88, 16, 8);
+
+/* reg_ptce3_delta_value
+ * Delta value.
+ * Bits which are masked by delta_mask must be 0.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, ptce3, delta_value, 0x88, 0, 8);
+
+/* reg_ptce3_prune_vector
+ * Pruning vector relative to the PERPT.erp_id.
+ * Used for reducing lookups.
+ * 0 - NEED: Do a lookup using the eRP.
+ * 1 - PRUNE: Do not perform a lookup using the eRP.
+ * Maybe be modified by PEAPBL and PEAPBM.
+ * Note: In Spectrum-2, a region of 8 key blocks must be set to either
+ * all 1's or all 0's.
+ * Access: RW
+ */
+MLXSW_ITEM_BIT_ARRAY(reg, ptce3, prune_vector, 0x90, 4, 1);
+
+/* reg_ptce3_prune_ctcam
+ * Pruning on C-TCAM. Used for reducing lookups.
+ * 0 - NEED: Do a lookup in the C-TCAM.
+ * 1 - PRUNE: Do not perform a lookup in the C-TCAM.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ptce3, prune_ctcam, 0x94, 31, 1);
+
+/* reg_ptce3_large_exists
+ * Large entry key ID exists.
+ * Within the region:
+ * 0 - SINGLE: The large_entry_key_id is not currently in use.
+ * For rule insert: The MSB of the key (blocks 6..11) will be added.
+ * For rule delete: The MSB of the key will be removed.
+ * 1 - NON_SINGLE: The large_entry_key_id is currently in use.
+ * For rule insert: The MSB of the key (blocks 6..11) will not be added.
+ * For rule delete: The MSB of the key will not be removed.
+ * Access: WO
+ */
+MLXSW_ITEM32(reg, ptce3, large_exists, 0x98, 31, 1);
+
+/* reg_ptce3_large_entry_key_id
+ * Large entry key ID.
+ * A key for 12 key blocks rules. Reserved when region has less than 12 key
+ * blocks. Must be different for different keys which have the same common
+ * 6 key blocks (MSB, blocks 6..11) key within a region.
+ * Range is 0..cap_max_pe_large_key_id - 1
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ptce3, large_entry_key_id, 0x98, 0, 24);
+
+/* reg_ptce3_action_pointer
+ * Pointer to action.
+ * Range is 0..cap_max_kvd_action_sets - 1
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ptce3, action_pointer, 0xA0, 0, 24);
+
+static inline void mlxsw_reg_ptce3_pack(char *payload, bool valid,
+					enum mlxsw_reg_ptce3_op op,
+					u32 priority,
+					const char *tcam_region_info,
+					const char *key, u8 erp_id,
+					u16 delta_start, u8 delta_mask,
+					u8 delta_value, bool large_exists,
+					u32 lkey_id, u32 action_pointer)
+{
+	MLXSW_REG_ZERO(ptce3, payload);
+	mlxsw_reg_ptce3_v_set(payload, valid);
+	mlxsw_reg_ptce3_op_set(payload, op);
+	mlxsw_reg_ptce3_priority_set(payload, priority);
+	mlxsw_reg_ptce3_tcam_region_info_memcpy_to(payload, tcam_region_info);
+	mlxsw_reg_ptce3_flex2_key_blocks_memcpy_to(payload, key);
+	mlxsw_reg_ptce3_erp_id_set(payload, erp_id);
+	mlxsw_reg_ptce3_delta_start_set(payload, delta_start);
+	mlxsw_reg_ptce3_delta_mask_set(payload, delta_mask);
+	mlxsw_reg_ptce3_delta_value_set(payload, delta_value);
+	mlxsw_reg_ptce3_large_exists_set(payload, large_exists);
+	mlxsw_reg_ptce3_large_entry_key_id_set(payload, lkey_id);
+	mlxsw_reg_ptce3_action_pointer_set(payload, action_pointer);
+}
+
+/* PERCR - Policy-Engine Region Configuration Register
+ * ---------------------------------------------------
+ * This register configures the region parameters. The region_id must be
+ * allocated.
+ */
+#define MLXSW_REG_PERCR_ID 0x302A
+#define MLXSW_REG_PERCR_LEN 0x80
+
+MLXSW_REG_DEFINE(percr, MLXSW_REG_PERCR_ID, MLXSW_REG_PERCR_LEN);
+
+/* reg_percr_region_id
+ * Region identifier.
+ * Range 0..cap_max_regions-1
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, percr, region_id, 0x00, 0, 16);
+
+/* reg_percr_atcam_ignore_prune
+ * Ignore prune_vector by other A-TCAM rules. Used e.g., for a new rule.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, percr, atcam_ignore_prune, 0x04, 25, 1);
+
+/* reg_percr_ctcam_ignore_prune
+ * Ignore prune_ctcam by other A-TCAM rules. Used e.g., for a new rule.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, percr, ctcam_ignore_prune, 0x04, 24, 1);
+
+/* reg_percr_bf_bypass
+ * Bloom filter bypass.
+ * 0 - Bloom filter is used (default)
+ * 1 - Bloom filter is bypassed. The bypass is an OR condition of
+ * region_id or eRP. See PERPT.bf_bypass
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, percr, bf_bypass, 0x04, 16, 1);
+
+/* reg_percr_master_mask
+ * Master mask. Logical OR mask of all masks of all rules of a region
+ * (both A-TCAM and C-TCAM). When there are no eRPs
+ * (erpt_pointer_valid = 0), then this provides the mask.
+ * Access: RW
+ */
+MLXSW_ITEM_BUF(reg, percr, master_mask, 0x20, 96);
+
+static inline void mlxsw_reg_percr_pack(char *payload, u16 region_id)
+{
+	MLXSW_REG_ZERO(percr, payload);
+	mlxsw_reg_percr_region_id_set(payload, region_id);
+	mlxsw_reg_percr_atcam_ignore_prune_set(payload, false);
+	mlxsw_reg_percr_ctcam_ignore_prune_set(payload, false);
+	mlxsw_reg_percr_bf_bypass_set(payload, false);
+}
+
+/* PERERP - Policy-Engine Region eRP Register
+ * ------------------------------------------
+ * This register configures the region eRP. The region_id must be
+ * allocated.
+ */
+#define MLXSW_REG_PERERP_ID 0x302B
+#define MLXSW_REG_PERERP_LEN 0x1C
+
+MLXSW_REG_DEFINE(pererp, MLXSW_REG_PERERP_ID, MLXSW_REG_PERERP_LEN);
+
+/* reg_pererp_region_id
+ * Region identifier.
+ * Range 0..cap_max_regions-1
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, pererp, region_id, 0x00, 0, 16);
+
+/* reg_pererp_ctcam_le
+ * C-TCAM lookup enable. Reserved when erpt_pointer_valid = 0.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, pererp, ctcam_le, 0x04, 28, 1);
+
+/* reg_pererp_erpt_pointer_valid
+ * erpt_pointer is valid.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, pererp, erpt_pointer_valid, 0x10, 31, 1);
+
+/* reg_pererp_erpt_bank_pointer
+ * Pointer to eRP table bank. May be modified at any time.
+ * Range 0..cap_max_erp_table_banks-1
+ * Reserved when erpt_pointer_valid = 0
+ */
+MLXSW_ITEM32(reg, pererp, erpt_bank_pointer, 0x10, 16, 4);
+
+/* reg_pererp_erpt_pointer
+ * Pointer to eRP table within the eRP bank. Can be changed for an
+ * existing region.
+ * Range 0..cap_max_erp_table_size-1
+ * Reserved when erpt_pointer_valid = 0
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, pererp, erpt_pointer, 0x10, 0, 8);
+
+/* reg_pererp_erpt_vector
+ * Vector of allowed eRP indexes starting from erpt_pointer within the
+ * erpt_bank_pointer. Next entries will be in next bank.
+ * Note that eRP index is used and not eRP ID.
+ * Reserved when erpt_pointer_valid = 0
+ * Access: RW
+ */
+MLXSW_ITEM_BIT_ARRAY(reg, pererp, erpt_vector, 0x14, 4, 1);
+
+/* reg_pererp_master_rp_id
+ * Master RP ID. When there are no eRPs, then this provides the eRP ID
+ * for the lookup. Can be changed for an existing region.
+ * Reserved when erpt_pointer_valid = 1
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, pererp, master_rp_id, 0x18, 0, 4);
+
+static inline void mlxsw_reg_pererp_erp_vector_pack(char *payload,
+						    unsigned long *erp_vector,
+						    unsigned long size)
+{
+	unsigned long bit;
+
+	for_each_set_bit(bit, erp_vector, size)
+		mlxsw_reg_pererp_erpt_vector_set(payload, bit, true);
+}
+
+static inline void mlxsw_reg_pererp_pack(char *payload, u16 region_id,
+					 bool ctcam_le, bool erpt_pointer_valid,
+					 u8 erpt_bank_pointer, u8 erpt_pointer,
+					 u8 master_rp_id)
+{
+	MLXSW_REG_ZERO(pererp, payload);
+	mlxsw_reg_pererp_region_id_set(payload, region_id);
+	mlxsw_reg_pererp_ctcam_le_set(payload, ctcam_le);
+	mlxsw_reg_pererp_erpt_pointer_valid_set(payload, erpt_pointer_valid);
+	mlxsw_reg_pererp_erpt_bank_pointer_set(payload, erpt_bank_pointer);
+	mlxsw_reg_pererp_erpt_pointer_set(payload, erpt_pointer);
+	mlxsw_reg_pererp_master_rp_id_set(payload, master_rp_id);
+}
+
+/* PEABFE - Policy-Engine Algorithmic Bloom Filter Entries Register
+ * ----------------------------------------------------------------
+ * This register configures the Bloom filter entries.
+ */
+#define MLXSW_REG_PEABFE_ID 0x3022
+#define MLXSW_REG_PEABFE_BASE_LEN 0x10
+#define MLXSW_REG_PEABFE_BF_REC_LEN 0x4
+#define MLXSW_REG_PEABFE_BF_REC_MAX_COUNT 256
+#define MLXSW_REG_PEABFE_LEN (MLXSW_REG_PEABFE_BASE_LEN + \
+			      MLXSW_REG_PEABFE_BF_REC_LEN * \
+			      MLXSW_REG_PEABFE_BF_REC_MAX_COUNT)
+
+MLXSW_REG_DEFINE(peabfe, MLXSW_REG_PEABFE_ID, MLXSW_REG_PEABFE_LEN);
+
+/* reg_peabfe_size
+ * Number of BF entries to be updated.
+ * Range 1..256
+ * Access: Op
+ */
+MLXSW_ITEM32(reg, peabfe, size, 0x00, 0, 9);
+
+/* reg_peabfe_bf_entry_state
+ * Bloom filter state
+ * 0 - Clear
+ * 1 - Set
+ * Access: RW
+ */
+MLXSW_ITEM32_INDEXED(reg, peabfe, bf_entry_state,
+		     MLXSW_REG_PEABFE_BASE_LEN,	31, 1,
+		     MLXSW_REG_PEABFE_BF_REC_LEN, 0x00, false);
+
+/* reg_peabfe_bf_entry_bank
+ * Bloom filter bank ID
+ * Range 0..cap_max_erp_table_banks-1
+ * Access: Index
+ */
+MLXSW_ITEM32_INDEXED(reg, peabfe, bf_entry_bank,
+		     MLXSW_REG_PEABFE_BASE_LEN,	24, 4,
+		     MLXSW_REG_PEABFE_BF_REC_LEN, 0x00, false);
+
+/* reg_peabfe_bf_entry_index
+ * Bloom filter entry index
+ * Range 0..2^cap_max_bf_log-1
+ * Access: Index
+ */
+MLXSW_ITEM32_INDEXED(reg, peabfe, bf_entry_index,
+		     MLXSW_REG_PEABFE_BASE_LEN,	0, 24,
+		     MLXSW_REG_PEABFE_BF_REC_LEN, 0x00, false);
+
+static inline void mlxsw_reg_peabfe_pack(char *payload)
+{
+	MLXSW_REG_ZERO(peabfe, payload);
+}
+
+static inline void mlxsw_reg_peabfe_rec_pack(char *payload, int rec_index,
+					     u8 state, u8 bank, u32 bf_index)
+{
+	u8 num_rec = mlxsw_reg_peabfe_size_get(payload);
+
+	if (rec_index >= num_rec)
+		mlxsw_reg_peabfe_size_set(payload, rec_index + 1);
+	mlxsw_reg_peabfe_bf_entry_state_set(payload, rec_index, state);
+	mlxsw_reg_peabfe_bf_entry_bank_set(payload, rec_index, bank);
+	mlxsw_reg_peabfe_bf_entry_index_set(payload, rec_index, bf_index);
+}
+
+/* IEDR - Infrastructure Entry Delete Register
+ * ----------------------------------------------------
+ * This register is used for deleting entries from the entry tables.
+ * It is legitimate to attempt to delete a nonexisting entry (the device will
+ * respond as a good flow).
+ */
+#define MLXSW_REG_IEDR_ID 0x3804
+#define MLXSW_REG_IEDR_BASE_LEN 0x10 /* base length, without records */
+#define MLXSW_REG_IEDR_REC_LEN 0x8 /* record length */
+#define MLXSW_REG_IEDR_REC_MAX_COUNT 64
+#define MLXSW_REG_IEDR_LEN (MLXSW_REG_IEDR_BASE_LEN +	\
+			    MLXSW_REG_IEDR_REC_LEN *	\
+			    MLXSW_REG_IEDR_REC_MAX_COUNT)
+
+MLXSW_REG_DEFINE(iedr, MLXSW_REG_IEDR_ID, MLXSW_REG_IEDR_LEN);
+
+/* reg_iedr_num_rec
+ * Number of records.
+ * Access: OP
+ */
+MLXSW_ITEM32(reg, iedr, num_rec, 0x00, 0, 8);
+
+/* reg_iedr_rec_type
+ * Resource type.
+ * Access: OP
+ */
+MLXSW_ITEM32_INDEXED(reg, iedr, rec_type, MLXSW_REG_IEDR_BASE_LEN, 24, 8,
+		     MLXSW_REG_IEDR_REC_LEN, 0x00, false);
+
+/* reg_iedr_rec_size
+ * Size of entries do be deleted. The unit is 1 entry, regardless of entry type.
+ * Access: OP
+ */
+MLXSW_ITEM32_INDEXED(reg, iedr, rec_size, MLXSW_REG_IEDR_BASE_LEN, 0, 13,
+		     MLXSW_REG_IEDR_REC_LEN, 0x00, false);
+
+/* reg_iedr_rec_index_start
+ * Resource index start.
+ * Access: OP
+ */
+MLXSW_ITEM32_INDEXED(reg, iedr, rec_index_start, MLXSW_REG_IEDR_BASE_LEN, 0, 24,
+		     MLXSW_REG_IEDR_REC_LEN, 0x04, false);
+
+static inline void mlxsw_reg_iedr_pack(char *payload)
+{
+	MLXSW_REG_ZERO(iedr, payload);
+}
+
+static inline void mlxsw_reg_iedr_rec_pack(char *payload, int rec_index,
+					   u8 rec_type, u16 rec_size,
+					   u32 rec_index_start)
+{
+	u8 num_rec = mlxsw_reg_iedr_num_rec_get(payload);
+
+	if (rec_index >= num_rec)
+		mlxsw_reg_iedr_num_rec_set(payload, rec_index + 1);
+	mlxsw_reg_iedr_rec_type_set(payload, rec_index, rec_type);
+	mlxsw_reg_iedr_rec_size_set(payload, rec_index, rec_size);
+	mlxsw_reg_iedr_rec_index_start_set(payload, rec_index, rec_index_start);
+}
+
+/* QPTS - QoS Priority Trust State Register
+ * ----------------------------------------
+ * This register controls the port policy to calculate the switch priority and
+ * packet color based on incoming packet fields.
+ */
+#define MLXSW_REG_QPTS_ID 0x4002
+#define MLXSW_REG_QPTS_LEN 0x8
+
+MLXSW_REG_DEFINE(qpts, MLXSW_REG_QPTS_ID, MLXSW_REG_QPTS_LEN);
+
+/* reg_qpts_local_port
+ * Local port number.
+ * Access: Index
+ *
+ * Note: CPU port is supported.
+ */
+MLXSW_ITEM32_LP(reg, qpts, 0x00, 16, 0x00, 12);
+
+enum mlxsw_reg_qpts_trust_state {
+	MLXSW_REG_QPTS_TRUST_STATE_PCP = 1,
+	MLXSW_REG_QPTS_TRUST_STATE_DSCP = 2, /* For MPLS, trust EXP. */
+};
+
+/* reg_qpts_trust_state
+ * Trust state for a given port.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, qpts, trust_state, 0x04, 0, 3);
+
+static inline void mlxsw_reg_qpts_pack(char *payload, u16 local_port,
+				       enum mlxsw_reg_qpts_trust_state ts)
+{
+	MLXSW_REG_ZERO(qpts, payload);
+
+	mlxsw_reg_qpts_local_port_set(payload, local_port);
+	mlxsw_reg_qpts_trust_state_set(payload, ts);
+}
+
+/* QPCR - QoS Policer Configuration Register
+ * -----------------------------------------
+ * The QPCR register is used to create policers - that limit
+ * the rate of bytes or packets via some trap group.
+ */
+#define MLXSW_REG_QPCR_ID 0x4004
+#define MLXSW_REG_QPCR_LEN 0x28
+
+MLXSW_REG_DEFINE(qpcr, MLXSW_REG_QPCR_ID, MLXSW_REG_QPCR_LEN);
+
+enum mlxsw_reg_qpcr_g {
+	MLXSW_REG_QPCR_G_GLOBAL = 2,
+	MLXSW_REG_QPCR_G_STORM_CONTROL = 3,
+};
+
+/* reg_qpcr_g
+ * The policer type.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, qpcr, g, 0x00, 14, 2);
+
+/* reg_qpcr_pid
+ * Policer ID.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, qpcr, pid, 0x00, 0, 14);
+
+/* reg_qpcr_clear_counter
+ * Clear counters.
+ * Access: OP
+ */
+MLXSW_ITEM32(reg, qpcr, clear_counter, 0x04, 31, 1);
+
+/* reg_qpcr_color_aware
+ * Is the policer aware of colors.
+ * Must be 0 (unaware) for cpu port.
+ * Access: RW for unbounded policer. RO for bounded policer.
+ */
+MLXSW_ITEM32(reg, qpcr, color_aware, 0x04, 15, 1);
+
+/* reg_qpcr_bytes
+ * Is policer limit is for bytes per sec or packets per sec.
+ * 0 - packets
+ * 1 - bytes
+ * Access: RW for unbounded policer. RO for bounded policer.
+ */
+MLXSW_ITEM32(reg, qpcr, bytes, 0x04, 14, 1);
+
+enum mlxsw_reg_qpcr_ir_units {
+	MLXSW_REG_QPCR_IR_UNITS_M,
+	MLXSW_REG_QPCR_IR_UNITS_K,
+};
+
+/* reg_qpcr_ir_units
+ * Policer's units for cir and eir fields (for bytes limits only)
+ * 1 - 10^3
+ * 0 - 10^6
+ * Access: OP
+ */
+MLXSW_ITEM32(reg, qpcr, ir_units, 0x04, 12, 1);
+
+enum mlxsw_reg_qpcr_rate_type {
+	MLXSW_REG_QPCR_RATE_TYPE_SINGLE = 1,
+	MLXSW_REG_QPCR_RATE_TYPE_DOUBLE = 2,
+};
+
+/* reg_qpcr_rate_type
+ * Policer can have one limit (single rate) or 2 limits with specific operation
+ * for packets that exceed the lower rate but not the upper one.
+ * (For cpu port must be single rate)
+ * Access: RW for unbounded policer. RO for bounded policer.
+ */
+MLXSW_ITEM32(reg, qpcr, rate_type, 0x04, 8, 2);
+
+/* reg_qpc_cbs
+ * Policer's committed burst size.
+ * The policer is working with time slices of 50 nano sec. By default every
+ * slice is granted the proportionate share of the committed rate. If we want to
+ * allow a slice to exceed that share (while still keeping the rate per sec) we
+ * can allow burst. The burst size is between the default proportionate share
+ * (and no lower than 8) to 32Gb. (Even though giving a number higher than the
+ * committed rate will result in exceeding the rate). The burst size must be a
+ * log of 2 and will be determined by 2^cbs.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, qpcr, cbs, 0x08, 24, 6);
+
+/* reg_qpcr_cir
+ * Policer's committed rate.
+ * The rate used for sungle rate, the lower rate for double rate.
+ * For bytes limits, the rate will be this value * the unit from ir_units.
+ * (Resolution error is up to 1%).
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, qpcr, cir, 0x0C, 0, 32);
+
+/* reg_qpcr_eir
+ * Policer's exceed rate.
+ * The higher rate for double rate, reserved for single rate.
+ * Lower rate for double rate policer.
+ * For bytes limits, the rate will be this value * the unit from ir_units.
+ * (Resolution error is up to 1%).
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, qpcr, eir, 0x10, 0, 32);
+
+#define MLXSW_REG_QPCR_DOUBLE_RATE_ACTION 2
+
+/* reg_qpcr_exceed_action.
+ * What to do with packets between the 2 limits for double rate.
+ * Access: RW for unbounded policer. RO for bounded policer.
+ */
+MLXSW_ITEM32(reg, qpcr, exceed_action, 0x14, 0, 4);
+
+enum mlxsw_reg_qpcr_action {
+	/* Discard */
+	MLXSW_REG_QPCR_ACTION_DISCARD = 1,
+	/* Forward and set color to red.
+	 * If the packet is intended to cpu port, it will be dropped.
+	 */
+	MLXSW_REG_QPCR_ACTION_FORWARD = 2,
+};
+
+/* reg_qpcr_violate_action
+ * What to do with packets that cross the cir limit (for single rate) or the eir
+ * limit (for double rate).
+ * Access: RW for unbounded policer. RO for bounded policer.
+ */
+MLXSW_ITEM32(reg, qpcr, violate_action, 0x18, 0, 4);
+
+/* reg_qpcr_violate_count
+ * Counts the number of times violate_action happened on this PID.
+ * Access: RW
+ */
+MLXSW_ITEM64(reg, qpcr, violate_count, 0x20, 0, 64);
+
+/* Packets */
+#define MLXSW_REG_QPCR_LOWEST_CIR	1
+#define MLXSW_REG_QPCR_HIGHEST_CIR	(2 * 1000 * 1000 * 1000) /* 2Gpps */
+#define MLXSW_REG_QPCR_LOWEST_CBS	4
+#define MLXSW_REG_QPCR_HIGHEST_CBS	24
+
+/* Bandwidth */
+#define MLXSW_REG_QPCR_LOWEST_CIR_BITS		1024 /* bps */
+#define MLXSW_REG_QPCR_HIGHEST_CIR_BITS		2000000000000ULL /* 2Tbps */
+#define MLXSW_REG_QPCR_LOWEST_CBS_BITS_SP1	4
+#define MLXSW_REG_QPCR_LOWEST_CBS_BITS_SP2	4
+#define MLXSW_REG_QPCR_HIGHEST_CBS_BITS_SP1	25
+#define MLXSW_REG_QPCR_HIGHEST_CBS_BITS_SP2	31
+
+static inline void mlxsw_reg_qpcr_pack(char *payload, u16 pid,
+				       enum mlxsw_reg_qpcr_ir_units ir_units,
+				       bool bytes, u32 cir, u16 cbs)
+{
+	MLXSW_REG_ZERO(qpcr, payload);
+	mlxsw_reg_qpcr_pid_set(payload, pid);
+	mlxsw_reg_qpcr_g_set(payload, MLXSW_REG_QPCR_G_GLOBAL);
+	mlxsw_reg_qpcr_rate_type_set(payload, MLXSW_REG_QPCR_RATE_TYPE_SINGLE);
+	mlxsw_reg_qpcr_violate_action_set(payload,
+					  MLXSW_REG_QPCR_ACTION_DISCARD);
+	mlxsw_reg_qpcr_cir_set(payload, cir);
+	mlxsw_reg_qpcr_ir_units_set(payload, ir_units);
+	mlxsw_reg_qpcr_bytes_set(payload, bytes);
+	mlxsw_reg_qpcr_cbs_set(payload, cbs);
+}
+
+/* QTCT - QoS Switch Traffic Class Table
+ * -------------------------------------
+ * Configures the mapping between the packet switch priority and the
+ * traffic class on the transmit port.
+ */
+#define MLXSW_REG_QTCT_ID 0x400A
+#define MLXSW_REG_QTCT_LEN 0x08
+
+MLXSW_REG_DEFINE(qtct, MLXSW_REG_QTCT_ID, MLXSW_REG_QTCT_LEN);
+
+/* reg_qtct_local_port
+ * Local port number.
+ * Access: Index
+ *
+ * Note: CPU port is not supported.
+ */
+MLXSW_ITEM32_LP(reg, qtct, 0x00, 16, 0x00, 12);
+
+/* reg_qtct_sub_port
+ * Virtual port within the physical port.
+ * Should be set to 0 when virtual ports are not enabled on the port.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, qtct, sub_port, 0x00, 8, 8);
+
+/* reg_qtct_switch_prio
+ * Switch priority.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, qtct, switch_prio, 0x00, 0, 4);
+
+/* reg_qtct_tclass
+ * Traffic class.
+ * Default values:
+ * switch_prio 0 : tclass 1
+ * switch_prio 1 : tclass 0
+ * switch_prio i : tclass i, for i > 1
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, qtct, tclass, 0x04, 0, 4);
+
+static inline void mlxsw_reg_qtct_pack(char *payload, u16 local_port,
+				       u8 switch_prio, u8 tclass)
+{
+	MLXSW_REG_ZERO(qtct, payload);
+	mlxsw_reg_qtct_local_port_set(payload, local_port);
+	mlxsw_reg_qtct_switch_prio_set(payload, switch_prio);
+	mlxsw_reg_qtct_tclass_set(payload, tclass);
+}
+
+/* QEEC - QoS ETS Element Configuration Register
+ * ---------------------------------------------
+ * Configures the ETS elements.
+ */
+#define MLXSW_REG_QEEC_ID 0x400D
+#define MLXSW_REG_QEEC_LEN 0x20
+
+MLXSW_REG_DEFINE(qeec, MLXSW_REG_QEEC_ID, MLXSW_REG_QEEC_LEN);
+
+/* reg_qeec_local_port
+ * Local port number.
+ * Access: Index
+ *
+ * Note: CPU port is supported.
+ */
+MLXSW_ITEM32_LP(reg, qeec, 0x00, 16, 0x00, 12);
+
+enum mlxsw_reg_qeec_hr {
+	MLXSW_REG_QEEC_HR_PORT,
+	MLXSW_REG_QEEC_HR_GROUP,
+	MLXSW_REG_QEEC_HR_SUBGROUP,
+	MLXSW_REG_QEEC_HR_TC,
+};
+
+/* reg_qeec_element_hierarchy
+ * 0 - Port
+ * 1 - Group
+ * 2 - Subgroup
+ * 3 - Traffic Class
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, qeec, element_hierarchy, 0x04, 16, 4);
+
+/* reg_qeec_element_index
+ * The index of the element in the hierarchy.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, qeec, element_index, 0x04, 0, 8);
+
+/* reg_qeec_next_element_index
+ * The index of the next (lower) element in the hierarchy.
+ * Access: RW
+ *
+ * Note: Reserved for element_hierarchy 0.
+ */
+MLXSW_ITEM32(reg, qeec, next_element_index, 0x08, 0, 8);
+
+/* reg_qeec_mise
+ * Min shaper configuration enable. Enables configuration of the min
+ * shaper on this ETS element
+ * 0 - Disable
+ * 1 - Enable
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, qeec, mise, 0x0C, 31, 1);
+
+/* reg_qeec_ptps
+ * PTP shaper
+ * 0: regular shaper mode
+ * 1: PTP oriented shaper
+ * Allowed only for hierarchy 0
+ * Not supported for CPU port
+ * Note that ptps mode may affect the shaper rates of all hierarchies
+ * Supported only on Spectrum-1
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, qeec, ptps, 0x0C, 29, 1);
+
+enum {
+	MLXSW_REG_QEEC_BYTES_MODE,
+	MLXSW_REG_QEEC_PACKETS_MODE,
+};
+
+/* reg_qeec_pb
+ * Packets or bytes mode.
+ * 0 - Bytes mode
+ * 1 - Packets mode
+ * Access: RW
+ *
+ * Note: Used for max shaper configuration. For Spectrum, packets mode
+ * is supported only for traffic classes of CPU port.
+ */
+MLXSW_ITEM32(reg, qeec, pb, 0x0C, 28, 1);
+
+/* The smallest permitted min shaper rate. */
+#define MLXSW_REG_QEEC_MIS_MIN	200000		/* Kbps */
+
+/* reg_qeec_min_shaper_rate
+ * Min shaper information rate.
+ * For CPU port, can only be configured for port hierarchy.
+ * When in bytes mode, value is specified in units of 1000bps.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, qeec, min_shaper_rate, 0x0C, 0, 28);
+
+/* reg_qeec_mase
+ * Max shaper configuration enable. Enables configuration of the max
+ * shaper on this ETS element.
+ * 0 - Disable
+ * 1 - Enable
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, qeec, mase, 0x10, 31, 1);
+
+/* The largest max shaper value possible to disable the shaper. */
+#define MLXSW_REG_QEEC_MAS_DIS	((1u << 31) - 1)	/* Kbps */
+
+/* reg_qeec_max_shaper_rate
+ * Max shaper information rate.
+ * For CPU port, can only be configured for port hierarchy.
+ * When in bytes mode, value is specified in units of 1000bps.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, qeec, max_shaper_rate, 0x10, 0, 31);
+
+/* reg_qeec_de
+ * DWRR configuration enable. Enables configuration of the dwrr and
+ * dwrr_weight.
+ * 0 - Disable
+ * 1 - Enable
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, qeec, de, 0x18, 31, 1);
+
+/* reg_qeec_dwrr
+ * Transmission selection algorithm to use on the link going down from
+ * the ETS element.
+ * 0 - Strict priority
+ * 1 - DWRR
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, qeec, dwrr, 0x18, 15, 1);
+
+/* reg_qeec_dwrr_weight
+ * DWRR weight on the link going down from the ETS element. The
+ * percentage of bandwidth guaranteed to an ETS element within
+ * its hierarchy. The sum of all weights across all ETS elements
+ * within one hierarchy should be equal to 100. Reserved when
+ * transmission selection algorithm is strict priority.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, qeec, dwrr_weight, 0x18, 0, 8);
+
+/* reg_qeec_max_shaper_bs
+ * Max shaper burst size
+ * Burst size is 2^max_shaper_bs * 512 bits
+ * For Spectrum-1: Range is: 5..25
+ * For Spectrum-2: Range is: 11..25
+ * Reserved when ptps = 1
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, qeec, max_shaper_bs, 0x1C, 0, 6);
+
+#define MLXSW_REG_QEEC_HIGHEST_SHAPER_BS	25
+#define MLXSW_REG_QEEC_LOWEST_SHAPER_BS_SP1	5
+#define MLXSW_REG_QEEC_LOWEST_SHAPER_BS_SP2	11
+#define MLXSW_REG_QEEC_LOWEST_SHAPER_BS_SP3	11
+#define MLXSW_REG_QEEC_LOWEST_SHAPER_BS_SP4	11
+
+static inline void mlxsw_reg_qeec_pack(char *payload, u16 local_port,
+				       enum mlxsw_reg_qeec_hr hr, u8 index,
+				       u8 next_index)
+{
+	MLXSW_REG_ZERO(qeec, payload);
+	mlxsw_reg_qeec_local_port_set(payload, local_port);
+	mlxsw_reg_qeec_element_hierarchy_set(payload, hr);
+	mlxsw_reg_qeec_element_index_set(payload, index);
+	mlxsw_reg_qeec_next_element_index_set(payload, next_index);
+}
+
+static inline void mlxsw_reg_qeec_ptps_pack(char *payload, u16 local_port,
+					    bool ptps)
+{
+	MLXSW_REG_ZERO(qeec, payload);
+	mlxsw_reg_qeec_local_port_set(payload, local_port);
+	mlxsw_reg_qeec_element_hierarchy_set(payload, MLXSW_REG_QEEC_HR_PORT);
+	mlxsw_reg_qeec_ptps_set(payload, ptps);
+}
+
+/* QRWE - QoS ReWrite Enable
+ * -------------------------
+ * This register configures the rewrite enable per receive port.
+ */
+#define MLXSW_REG_QRWE_ID 0x400F
+#define MLXSW_REG_QRWE_LEN 0x08
+
+MLXSW_REG_DEFINE(qrwe, MLXSW_REG_QRWE_ID, MLXSW_REG_QRWE_LEN);
+
+/* reg_qrwe_local_port
+ * Local port number.
+ * Access: Index
+ *
+ * Note: CPU port is supported. No support for router port.
+ */
+MLXSW_ITEM32_LP(reg, qrwe, 0x00, 16, 0x00, 12);
+
+/* reg_qrwe_dscp
+ * Whether to enable DSCP rewrite (default is 0, don't rewrite).
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, qrwe, dscp, 0x04, 1, 1);
+
+/* reg_qrwe_pcp
+ * Whether to enable PCP and DEI rewrite (default is 0, don't rewrite).
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, qrwe, pcp, 0x04, 0, 1);
+
+static inline void mlxsw_reg_qrwe_pack(char *payload, u16 local_port,
+				       bool rewrite_pcp, bool rewrite_dscp)
+{
+	MLXSW_REG_ZERO(qrwe, payload);
+	mlxsw_reg_qrwe_local_port_set(payload, local_port);
+	mlxsw_reg_qrwe_pcp_set(payload, rewrite_pcp);
+	mlxsw_reg_qrwe_dscp_set(payload, rewrite_dscp);
+}
+
+/* QPDSM - QoS Priority to DSCP Mapping
+ * ------------------------------------
+ * QoS Priority to DSCP Mapping Register
+ */
+#define MLXSW_REG_QPDSM_ID 0x4011
+#define MLXSW_REG_QPDSM_BASE_LEN 0x04 /* base length, without records */
+#define MLXSW_REG_QPDSM_PRIO_ENTRY_REC_LEN 0x4 /* record length */
+#define MLXSW_REG_QPDSM_PRIO_ENTRY_REC_MAX_COUNT 16
+#define MLXSW_REG_QPDSM_LEN (MLXSW_REG_QPDSM_BASE_LEN +			\
+			     MLXSW_REG_QPDSM_PRIO_ENTRY_REC_LEN *	\
+			     MLXSW_REG_QPDSM_PRIO_ENTRY_REC_MAX_COUNT)
+
+MLXSW_REG_DEFINE(qpdsm, MLXSW_REG_QPDSM_ID, MLXSW_REG_QPDSM_LEN);
+
+/* reg_qpdsm_local_port
+ * Local Port. Supported for data packets from CPU port.
+ * Access: Index
+ */
+MLXSW_ITEM32_LP(reg, qpdsm, 0x00, 16, 0x00, 12);
+
+/* reg_qpdsm_prio_entry_color0_e
+ * Enable update of the entry for color 0 and a given port.
+ * Access: WO
+ */
+MLXSW_ITEM32_INDEXED(reg, qpdsm, prio_entry_color0_e,
+		     MLXSW_REG_QPDSM_BASE_LEN, 31, 1,
+		     MLXSW_REG_QPDSM_PRIO_ENTRY_REC_LEN, 0x00, false);
+
+/* reg_qpdsm_prio_entry_color0_dscp
+ * DSCP field in the outer label of the packet for color 0 and a given port.
+ * Reserved when e=0.
+ * Access: RW
+ */
+MLXSW_ITEM32_INDEXED(reg, qpdsm, prio_entry_color0_dscp,
+		     MLXSW_REG_QPDSM_BASE_LEN, 24, 6,
+		     MLXSW_REG_QPDSM_PRIO_ENTRY_REC_LEN, 0x00, false);
+
+/* reg_qpdsm_prio_entry_color1_e
+ * Enable update of the entry for color 1 and a given port.
+ * Access: WO
+ */
+MLXSW_ITEM32_INDEXED(reg, qpdsm, prio_entry_color1_e,
+		     MLXSW_REG_QPDSM_BASE_LEN, 23, 1,
+		     MLXSW_REG_QPDSM_PRIO_ENTRY_REC_LEN, 0x00, false);
+
+/* reg_qpdsm_prio_entry_color1_dscp
+ * DSCP field in the outer label of the packet for color 1 and a given port.
+ * Reserved when e=0.
+ * Access: RW
+ */
+MLXSW_ITEM32_INDEXED(reg, qpdsm, prio_entry_color1_dscp,
+		     MLXSW_REG_QPDSM_BASE_LEN, 16, 6,
+		     MLXSW_REG_QPDSM_PRIO_ENTRY_REC_LEN, 0x00, false);
+
+/* reg_qpdsm_prio_entry_color2_e
+ * Enable update of the entry for color 2 and a given port.
+ * Access: WO
+ */
+MLXSW_ITEM32_INDEXED(reg, qpdsm, prio_entry_color2_e,
+		     MLXSW_REG_QPDSM_BASE_LEN, 15, 1,
+		     MLXSW_REG_QPDSM_PRIO_ENTRY_REC_LEN, 0x00, false);
+
+/* reg_qpdsm_prio_entry_color2_dscp
+ * DSCP field in the outer label of the packet for color 2 and a given port.
+ * Reserved when e=0.
+ * Access: RW
+ */
+MLXSW_ITEM32_INDEXED(reg, qpdsm, prio_entry_color2_dscp,
+		     MLXSW_REG_QPDSM_BASE_LEN, 8, 6,
+		     MLXSW_REG_QPDSM_PRIO_ENTRY_REC_LEN, 0x00, false);
+
+static inline void mlxsw_reg_qpdsm_pack(char *payload, u16 local_port)
+{
+	MLXSW_REG_ZERO(qpdsm, payload);
+	mlxsw_reg_qpdsm_local_port_set(payload, local_port);
+}
+
+static inline void
+mlxsw_reg_qpdsm_prio_pack(char *payload, unsigned short prio, u8 dscp)
+{
+	mlxsw_reg_qpdsm_prio_entry_color0_e_set(payload, prio, 1);
+	mlxsw_reg_qpdsm_prio_entry_color0_dscp_set(payload, prio, dscp);
+	mlxsw_reg_qpdsm_prio_entry_color1_e_set(payload, prio, 1);
+	mlxsw_reg_qpdsm_prio_entry_color1_dscp_set(payload, prio, dscp);
+	mlxsw_reg_qpdsm_prio_entry_color2_e_set(payload, prio, 1);
+	mlxsw_reg_qpdsm_prio_entry_color2_dscp_set(payload, prio, dscp);
+}
+
+/* QPDP - QoS Port DSCP to Priority Mapping Register
+ * -------------------------------------------------
+ * This register controls the port default Switch Priority and Color. The
+ * default Switch Priority and Color are used for frames where the trust state
+ * uses default values. All member ports of a LAG should be configured with the
+ * same default values.
+ */
+#define MLXSW_REG_QPDP_ID 0x4007
+#define MLXSW_REG_QPDP_LEN 0x8
+
+MLXSW_REG_DEFINE(qpdp, MLXSW_REG_QPDP_ID, MLXSW_REG_QPDP_LEN);
+
+/* reg_qpdp_local_port
+ * Local Port. Supported for data packets from CPU port.
+ * Access: Index
+ */
+MLXSW_ITEM32_LP(reg, qpdp, 0x00, 16, 0x00, 12);
+
+/* reg_qpdp_switch_prio
+ * Default port Switch Priority (default 0)
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, qpdp, switch_prio, 0x04, 0, 4);
+
+static inline void mlxsw_reg_qpdp_pack(char *payload, u16 local_port,
+				       u8 switch_prio)
+{
+	MLXSW_REG_ZERO(qpdp, payload);
+	mlxsw_reg_qpdp_local_port_set(payload, local_port);
+	mlxsw_reg_qpdp_switch_prio_set(payload, switch_prio);
+}
+
+/* QPDPM - QoS Port DSCP to Priority Mapping Register
+ * --------------------------------------------------
+ * This register controls the mapping from DSCP field to
+ * Switch Priority for IP packets.
+ */
+#define MLXSW_REG_QPDPM_ID 0x4013
+#define MLXSW_REG_QPDPM_BASE_LEN 0x4 /* base length, without records */
+#define MLXSW_REG_QPDPM_DSCP_ENTRY_REC_LEN 0x2 /* record length */
+#define MLXSW_REG_QPDPM_DSCP_ENTRY_REC_MAX_COUNT 64
+#define MLXSW_REG_QPDPM_LEN (MLXSW_REG_QPDPM_BASE_LEN +			\
+			     MLXSW_REG_QPDPM_DSCP_ENTRY_REC_LEN *	\
+			     MLXSW_REG_QPDPM_DSCP_ENTRY_REC_MAX_COUNT)
+
+MLXSW_REG_DEFINE(qpdpm, MLXSW_REG_QPDPM_ID, MLXSW_REG_QPDPM_LEN);
+
+/* reg_qpdpm_local_port
+ * Local Port. Supported for data packets from CPU port.
+ * Access: Index
+ */
+MLXSW_ITEM32_LP(reg, qpdpm, 0x00, 16, 0x00, 12);
+
+/* reg_qpdpm_dscp_e
+ * Enable update of the specific entry. When cleared, the switch_prio and color
+ * fields are ignored and the previous switch_prio and color values are
+ * preserved.
+ * Access: WO
+ */
+MLXSW_ITEM16_INDEXED(reg, qpdpm, dscp_entry_e, MLXSW_REG_QPDPM_BASE_LEN, 15, 1,
+		     MLXSW_REG_QPDPM_DSCP_ENTRY_REC_LEN, 0x00, false);
+
+/* reg_qpdpm_dscp_prio
+ * The new Switch Priority value for the relevant DSCP value.
+ * Access: RW
+ */
+MLXSW_ITEM16_INDEXED(reg, qpdpm, dscp_entry_prio,
+		     MLXSW_REG_QPDPM_BASE_LEN, 0, 4,
+		     MLXSW_REG_QPDPM_DSCP_ENTRY_REC_LEN, 0x00, false);
+
+static inline void mlxsw_reg_qpdpm_pack(char *payload, u16 local_port)
+{
+	MLXSW_REG_ZERO(qpdpm, payload);
+	mlxsw_reg_qpdpm_local_port_set(payload, local_port);
+}
+
+static inline void
+mlxsw_reg_qpdpm_dscp_pack(char *payload, unsigned short dscp, u8 prio)
+{
+	mlxsw_reg_qpdpm_dscp_entry_e_set(payload, dscp, 1);
+	mlxsw_reg_qpdpm_dscp_entry_prio_set(payload, dscp, prio);
+}
+
+/* QTCTM - QoS Switch Traffic Class Table is Multicast-Aware Register
+ * ------------------------------------------------------------------
+ * This register configures if the Switch Priority to Traffic Class mapping is
+ * based on Multicast packet indication. If so, then multicast packets will get
+ * a Traffic Class that is plus (cap_max_tclass_data/2) the value configured by
+ * QTCT.
+ * By default, Switch Priority to Traffic Class mapping is not based on
+ * Multicast packet indication.
+ */
+#define MLXSW_REG_QTCTM_ID 0x401A
+#define MLXSW_REG_QTCTM_LEN 0x08
+
+MLXSW_REG_DEFINE(qtctm, MLXSW_REG_QTCTM_ID, MLXSW_REG_QTCTM_LEN);
+
+/* reg_qtctm_local_port
+ * Local port number.
+ * No support for CPU port.
+ * Access: Index
+ */
+MLXSW_ITEM32_LP(reg, qtctm, 0x00, 16, 0x00, 12);
+
+/* reg_qtctm_mc
+ * Multicast Mode
+ * Whether Switch Priority to Traffic Class mapping is based on Multicast packet
+ * indication (default is 0, not based on Multicast packet indication).
+ */
+MLXSW_ITEM32(reg, qtctm, mc, 0x04, 0, 1);
+
+static inline void
+mlxsw_reg_qtctm_pack(char *payload, u16 local_port, bool mc)
+{
+	MLXSW_REG_ZERO(qtctm, payload);
+	mlxsw_reg_qtctm_local_port_set(payload, local_port);
+	mlxsw_reg_qtctm_mc_set(payload, mc);
+}
+
+/* QPSC - QoS PTP Shaper Configuration Register
+ * --------------------------------------------
+ * The QPSC allows advanced configuration of the shapers when QEEC.ptps=1.
+ * Supported only on Spectrum-1.
+ */
+#define MLXSW_REG_QPSC_ID 0x401B
+#define MLXSW_REG_QPSC_LEN 0x28
+
+MLXSW_REG_DEFINE(qpsc, MLXSW_REG_QPSC_ID, MLXSW_REG_QPSC_LEN);
+
+enum mlxsw_reg_qpsc_port_speed {
+	MLXSW_REG_QPSC_PORT_SPEED_100M,
+	MLXSW_REG_QPSC_PORT_SPEED_1G,
+	MLXSW_REG_QPSC_PORT_SPEED_10G,
+	MLXSW_REG_QPSC_PORT_SPEED_25G,
+};
+
+/* reg_qpsc_port_speed
+ * Port speed.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, qpsc, port_speed, 0x00, 0, 4);
+
+/* reg_qpsc_shaper_time_exp
+ * The base-time-interval for updating the shapers tokens (for all hierarchies).
+ * shaper_update_rate = 2 ^ shaper_time_exp * (1 + shaper_time_mantissa) * 32nSec
+ * shaper_rate = 64bit * shaper_inc / shaper_update_rate
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, qpsc, shaper_time_exp, 0x04, 16, 4);
+
+/* reg_qpsc_shaper_time_mantissa
+ * The base-time-interval for updating the shapers tokens (for all hierarchies).
+ * shaper_update_rate = 2 ^ shaper_time_exp * (1 + shaper_time_mantissa) * 32nSec
+ * shaper_rate = 64bit * shaper_inc / shaper_update_rate
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, qpsc, shaper_time_mantissa, 0x04, 0, 5);
+
+/* reg_qpsc_shaper_inc
+ * Number of tokens added to shaper on each update.
+ * Units of 8B.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, qpsc, shaper_inc, 0x08, 0, 5);
+
+/* reg_qpsc_shaper_bs
+ * Max shaper Burst size.
+ * Burst size is 2 ^ max_shaper_bs * 512 [bits]
+ * Range is: 5..25 (from 2KB..2GB)
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, qpsc, shaper_bs, 0x0C, 0, 6);
+
+/* reg_qpsc_ptsc_we
+ * Write enable to port_to_shaper_credits.
+ * Access: WO
+ */
+MLXSW_ITEM32(reg, qpsc, ptsc_we, 0x10, 31, 1);
+
+/* reg_qpsc_port_to_shaper_credits
+ * For split ports: range 1..57
+ * For non-split ports: range 1..112
+ * Written only when ptsc_we is set.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, qpsc, port_to_shaper_credits, 0x10, 0, 8);
+
+/* reg_qpsc_ing_timestamp_inc
+ * Ingress timestamp increment.
+ * 2's complement.
+ * The timestamp of MTPPTR at ingress will be incremented by this value. Global
+ * value for all ports.
+ * Same units as used by MTPPTR.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, qpsc, ing_timestamp_inc, 0x20, 0, 32);
+
+/* reg_qpsc_egr_timestamp_inc
+ * Egress timestamp increment.
+ * 2's complement.
+ * The timestamp of MTPPTR at egress will be incremented by this value. Global
+ * value for all ports.
+ * Same units as used by MTPPTR.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, qpsc, egr_timestamp_inc, 0x24, 0, 32);
+
+static inline void
+mlxsw_reg_qpsc_pack(char *payload, enum mlxsw_reg_qpsc_port_speed port_speed,
+		    u8 shaper_time_exp, u8 shaper_time_mantissa, u8 shaper_inc,
+		    u8 shaper_bs, u8 port_to_shaper_credits,
+		    int ing_timestamp_inc, int egr_timestamp_inc)
+{
+	MLXSW_REG_ZERO(qpsc, payload);
+	mlxsw_reg_qpsc_port_speed_set(payload, port_speed);
+	mlxsw_reg_qpsc_shaper_time_exp_set(payload, shaper_time_exp);
+	mlxsw_reg_qpsc_shaper_time_mantissa_set(payload, shaper_time_mantissa);
+	mlxsw_reg_qpsc_shaper_inc_set(payload, shaper_inc);
+	mlxsw_reg_qpsc_shaper_bs_set(payload, shaper_bs);
+	mlxsw_reg_qpsc_ptsc_we_set(payload, true);
+	mlxsw_reg_qpsc_port_to_shaper_credits_set(payload, port_to_shaper_credits);
+	mlxsw_reg_qpsc_ing_timestamp_inc_set(payload, ing_timestamp_inc);
+	mlxsw_reg_qpsc_egr_timestamp_inc_set(payload, egr_timestamp_inc);
+}
+
+/* PMLP - Ports Module to Local Port Register
+ * ------------------------------------------
+ * Configures the assignment of modules to local ports.
+ */
+#define MLXSW_REG_PMLP_ID 0x5002
+#define MLXSW_REG_PMLP_LEN 0x40
+
+MLXSW_REG_DEFINE(pmlp, MLXSW_REG_PMLP_ID, MLXSW_REG_PMLP_LEN);
+
+/* reg_pmlp_rxtx
+ * 0 - Tx value is used for both Tx and Rx.
+ * 1 - Rx value is taken from a separte field.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, pmlp, rxtx, 0x00, 31, 1);
+
+/* reg_pmlp_local_port
+ * Local port number.
+ * Access: Index
+ */
+MLXSW_ITEM32_LP(reg, pmlp, 0x00, 16, 0x00, 12);
+
+/* reg_pmlp_width
+ * 0 - Unmap local port.
+ * 1 - Lane 0 is used.
+ * 2 - Lanes 0 and 1 are used.
+ * 4 - Lanes 0, 1, 2 and 3 are used.
+ * 8 - Lanes 0-7 are used.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, pmlp, width, 0x00, 0, 8);
+
+/* reg_pmlp_module
+ * Module number.
+ * Access: RW
+ */
+MLXSW_ITEM32_INDEXED(reg, pmlp, module, 0x04, 0, 8, 0x04, 0x00, false);
+
+/* reg_pmlp_slot_index
+ * Module number.
+ * Slot_index
+ * Slot_index = 0 represent the onboard (motherboard).
+ * In case of non-modular system only slot_index = 0 is available.
+ * Access: RW
+ */
+MLXSW_ITEM32_INDEXED(reg, pmlp, slot_index, 0x04, 8, 4, 0x04, 0x00, false);
+
+/* reg_pmlp_tx_lane
+ * Tx Lane. When rxtx field is cleared, this field is used for Rx as well.
+ * Access: RW
+ */
+MLXSW_ITEM32_INDEXED(reg, pmlp, tx_lane, 0x04, 16, 4, 0x04, 0x00, false);
+
+/* reg_pmlp_rx_lane
+ * Rx Lane. When rxtx field is cleared, this field is ignored and Rx lane is
+ * equal to Tx lane.
+ * Access: RW
+ */
+MLXSW_ITEM32_INDEXED(reg, pmlp, rx_lane, 0x04, 24, 4, 0x04, 0x00, false);
+
+static inline void mlxsw_reg_pmlp_pack(char *payload, u16 local_port)
+{
+	MLXSW_REG_ZERO(pmlp, payload);
+	mlxsw_reg_pmlp_local_port_set(payload, local_port);
+}
+
+/* PMTU - Port MTU Register
+ * ------------------------
+ * Configures and reports the port MTU.
+ */
+#define MLXSW_REG_PMTU_ID 0x5003
+#define MLXSW_REG_PMTU_LEN 0x10
+
+MLXSW_REG_DEFINE(pmtu, MLXSW_REG_PMTU_ID, MLXSW_REG_PMTU_LEN);
+
+/* reg_pmtu_local_port
+ * Local port number.
+ * Access: Index
+ */
+MLXSW_ITEM32_LP(reg, pmtu, 0x00, 16, 0x00, 12);
+
+/* reg_pmtu_max_mtu
+ * Maximum MTU.
+ * When port type (e.g. Ethernet) is configured, the relevant MTU is
+ * reported, otherwise the minimum between the max_mtu of the different
+ * types is reported.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, pmtu, max_mtu, 0x04, 16, 16);
+
+/* reg_pmtu_admin_mtu
+ * MTU value to set port to. Must be smaller or equal to max_mtu.
+ * Note: If port type is Infiniband, then port must be disabled, when its
+ * MTU is set.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, pmtu, admin_mtu, 0x08, 16, 16);
+
+/* reg_pmtu_oper_mtu
+ * The actual MTU configured on the port. Packets exceeding this size
+ * will be dropped.
+ * Note: In Ethernet and FC oper_mtu == admin_mtu, however, in Infiniband
+ * oper_mtu might be smaller than admin_mtu.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, pmtu, oper_mtu, 0x0C, 16, 16);
+
+static inline void mlxsw_reg_pmtu_pack(char *payload, u16 local_port,
+				       u16 new_mtu)
+{
+	MLXSW_REG_ZERO(pmtu, payload);
+	mlxsw_reg_pmtu_local_port_set(payload, local_port);
+	mlxsw_reg_pmtu_max_mtu_set(payload, 0);
+	mlxsw_reg_pmtu_admin_mtu_set(payload, new_mtu);
+	mlxsw_reg_pmtu_oper_mtu_set(payload, 0);
+}
+
+/* PTYS - Port Type and Speed Register
+ * -----------------------------------
+ * Configures and reports the port speed type.
+ *
+ * Note: When set while the link is up, the changes will not take effect
+ * until the port transitions from down to up state.
+ */
+#define MLXSW_REG_PTYS_ID 0x5004
+#define MLXSW_REG_PTYS_LEN 0x40
+
+MLXSW_REG_DEFINE(ptys, MLXSW_REG_PTYS_ID, MLXSW_REG_PTYS_LEN);
+
+/* an_disable_admin
+ * Auto negotiation disable administrative configuration
+ * 0 - Device doesn't support AN disable.
+ * 1 - Device supports AN disable.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ptys, an_disable_admin, 0x00, 30, 1);
+
+/* reg_ptys_local_port
+ * Local port number.
+ * Access: Index
+ */
+MLXSW_ITEM32_LP(reg, ptys, 0x00, 16, 0x00, 12);
+
+#define MLXSW_REG_PTYS_PROTO_MASK_IB	BIT(0)
+#define MLXSW_REG_PTYS_PROTO_MASK_ETH	BIT(2)
+
+/* reg_ptys_proto_mask
+ * Protocol mask. Indicates which protocol is used.
+ * 0 - Infiniband.
+ * 1 - Fibre Channel.
+ * 2 - Ethernet.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, ptys, proto_mask, 0x00, 0, 3);
+
+enum {
+	MLXSW_REG_PTYS_AN_STATUS_NA,
+	MLXSW_REG_PTYS_AN_STATUS_OK,
+	MLXSW_REG_PTYS_AN_STATUS_FAIL,
+};
+
+/* reg_ptys_an_status
+ * Autonegotiation status.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, ptys, an_status, 0x04, 28, 4);
+
+#define MLXSW_REG_PTYS_EXT_ETH_SPEED_SGMII_100M				BIT(0)
+#define MLXSW_REG_PTYS_EXT_ETH_SPEED_1000BASE_X_SGMII			BIT(1)
+#define MLXSW_REG_PTYS_EXT_ETH_SPEED_5GBASE_R				BIT(3)
+#define MLXSW_REG_PTYS_EXT_ETH_SPEED_XFI_XAUI_1_10G			BIT(4)
+#define MLXSW_REG_PTYS_EXT_ETH_SPEED_XLAUI_4_XLPPI_4_40G		BIT(5)
+#define MLXSW_REG_PTYS_EXT_ETH_SPEED_25GAUI_1_25GBASE_CR_KR		BIT(6)
+#define MLXSW_REG_PTYS_EXT_ETH_SPEED_50GAUI_2_LAUI_2_50GBASE_CR2_KR2	BIT(7)
+#define MLXSW_REG_PTYS_EXT_ETH_SPEED_50GAUI_1_LAUI_1_50GBASE_CR_KR	BIT(8)
+#define MLXSW_REG_PTYS_EXT_ETH_SPEED_CAUI_4_100GBASE_CR4_KR4		BIT(9)
+#define MLXSW_REG_PTYS_EXT_ETH_SPEED_100GAUI_2_100GBASE_CR2_KR2		BIT(10)
+#define MLXSW_REG_PTYS_EXT_ETH_SPEED_200GAUI_4_200GBASE_CR4_KR4		BIT(12)
+#define MLXSW_REG_PTYS_EXT_ETH_SPEED_400GAUI_8				BIT(15)
+
+/* reg_ptys_ext_eth_proto_cap
+ * Extended Ethernet port supported speeds and protocols.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, ptys, ext_eth_proto_cap, 0x08, 0, 32);
+
+#define MLXSW_REG_PTYS_ETH_SPEED_SGMII			BIT(0)
+#define MLXSW_REG_PTYS_ETH_SPEED_1000BASE_KX		BIT(1)
+#define MLXSW_REG_PTYS_ETH_SPEED_10GBASE_CX4		BIT(2)
+#define MLXSW_REG_PTYS_ETH_SPEED_10GBASE_KX4		BIT(3)
+#define MLXSW_REG_PTYS_ETH_SPEED_10GBASE_KR		BIT(4)
+#define MLXSW_REG_PTYS_ETH_SPEED_40GBASE_CR4		BIT(6)
+#define MLXSW_REG_PTYS_ETH_SPEED_40GBASE_KR4		BIT(7)
+#define MLXSW_REG_PTYS_ETH_SPEED_10GBASE_CR		BIT(12)
+#define MLXSW_REG_PTYS_ETH_SPEED_10GBASE_SR		BIT(13)
+#define MLXSW_REG_PTYS_ETH_SPEED_10GBASE_ER_LR		BIT(14)
+#define MLXSW_REG_PTYS_ETH_SPEED_40GBASE_SR4		BIT(15)
+#define MLXSW_REG_PTYS_ETH_SPEED_40GBASE_LR4_ER4	BIT(16)
+#define MLXSW_REG_PTYS_ETH_SPEED_50GBASE_SR2		BIT(18)
+#define MLXSW_REG_PTYS_ETH_SPEED_50GBASE_KR4		BIT(19)
+#define MLXSW_REG_PTYS_ETH_SPEED_100GBASE_CR4		BIT(20)
+#define MLXSW_REG_PTYS_ETH_SPEED_100GBASE_SR4		BIT(21)
+#define MLXSW_REG_PTYS_ETH_SPEED_100GBASE_KR4		BIT(22)
+#define MLXSW_REG_PTYS_ETH_SPEED_100GBASE_LR4_ER4	BIT(23)
+#define MLXSW_REG_PTYS_ETH_SPEED_100BASE_T		BIT(24)
+#define MLXSW_REG_PTYS_ETH_SPEED_1000BASE_T		BIT(25)
+#define MLXSW_REG_PTYS_ETH_SPEED_25GBASE_CR		BIT(27)
+#define MLXSW_REG_PTYS_ETH_SPEED_25GBASE_KR		BIT(28)
+#define MLXSW_REG_PTYS_ETH_SPEED_25GBASE_SR		BIT(29)
+#define MLXSW_REG_PTYS_ETH_SPEED_50GBASE_CR2		BIT(30)
+#define MLXSW_REG_PTYS_ETH_SPEED_50GBASE_KR2		BIT(31)
+
+/* reg_ptys_eth_proto_cap
+ * Ethernet port supported speeds and protocols.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, ptys, eth_proto_cap, 0x0C, 0, 32);
+
+/* reg_ptys_ext_eth_proto_admin
+ * Extended speed and protocol to set port to.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ptys, ext_eth_proto_admin, 0x14, 0, 32);
+
+/* reg_ptys_eth_proto_admin
+ * Speed and protocol to set port to.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ptys, eth_proto_admin, 0x18, 0, 32);
+
+/* reg_ptys_ext_eth_proto_oper
+ * The extended current speed and protocol configured for the port.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, ptys, ext_eth_proto_oper, 0x20, 0, 32);
+
+/* reg_ptys_eth_proto_oper
+ * The current speed and protocol configured for the port.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, ptys, eth_proto_oper, 0x24, 0, 32);
+
+enum mlxsw_reg_ptys_connector_type {
+	MLXSW_REG_PTYS_CONNECTOR_TYPE_UNKNOWN_OR_NO_CONNECTOR,
+	MLXSW_REG_PTYS_CONNECTOR_TYPE_PORT_NONE,
+	MLXSW_REG_PTYS_CONNECTOR_TYPE_PORT_TP,
+	MLXSW_REG_PTYS_CONNECTOR_TYPE_PORT_AUI,
+	MLXSW_REG_PTYS_CONNECTOR_TYPE_PORT_BNC,
+	MLXSW_REG_PTYS_CONNECTOR_TYPE_PORT_MII,
+	MLXSW_REG_PTYS_CONNECTOR_TYPE_PORT_FIBRE,
+	MLXSW_REG_PTYS_CONNECTOR_TYPE_PORT_DA,
+	MLXSW_REG_PTYS_CONNECTOR_TYPE_PORT_OTHER,
+};
+
+/* reg_ptys_connector_type
+ * Connector type indication.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, ptys, connector_type, 0x2C, 0, 4);
+
+static inline void mlxsw_reg_ptys_eth_pack(char *payload, u16 local_port,
+					   u32 proto_admin, bool autoneg)
+{
+	MLXSW_REG_ZERO(ptys, payload);
+	mlxsw_reg_ptys_local_port_set(payload, local_port);
+	mlxsw_reg_ptys_proto_mask_set(payload, MLXSW_REG_PTYS_PROTO_MASK_ETH);
+	mlxsw_reg_ptys_eth_proto_admin_set(payload, proto_admin);
+	mlxsw_reg_ptys_an_disable_admin_set(payload, !autoneg);
+}
+
+static inline void mlxsw_reg_ptys_ext_eth_pack(char *payload, u16 local_port,
+					       u32 proto_admin, bool autoneg)
+{
+	MLXSW_REG_ZERO(ptys, payload);
+	mlxsw_reg_ptys_local_port_set(payload, local_port);
+	mlxsw_reg_ptys_proto_mask_set(payload, MLXSW_REG_PTYS_PROTO_MASK_ETH);
+	mlxsw_reg_ptys_ext_eth_proto_admin_set(payload, proto_admin);
+	mlxsw_reg_ptys_an_disable_admin_set(payload, !autoneg);
+}
+
+static inline void mlxsw_reg_ptys_eth_unpack(char *payload,
+					     u32 *p_eth_proto_cap,
+					     u32 *p_eth_proto_admin,
+					     u32 *p_eth_proto_oper)
+{
+	if (p_eth_proto_cap)
+		*p_eth_proto_cap =
+			mlxsw_reg_ptys_eth_proto_cap_get(payload);
+	if (p_eth_proto_admin)
+		*p_eth_proto_admin =
+			mlxsw_reg_ptys_eth_proto_admin_get(payload);
+	if (p_eth_proto_oper)
+		*p_eth_proto_oper =
+			mlxsw_reg_ptys_eth_proto_oper_get(payload);
+}
+
+static inline void mlxsw_reg_ptys_ext_eth_unpack(char *payload,
+						 u32 *p_eth_proto_cap,
+						 u32 *p_eth_proto_admin,
+						 u32 *p_eth_proto_oper)
+{
+	if (p_eth_proto_cap)
+		*p_eth_proto_cap =
+			mlxsw_reg_ptys_ext_eth_proto_cap_get(payload);
+	if (p_eth_proto_admin)
+		*p_eth_proto_admin =
+			mlxsw_reg_ptys_ext_eth_proto_admin_get(payload);
+	if (p_eth_proto_oper)
+		*p_eth_proto_oper =
+			mlxsw_reg_ptys_ext_eth_proto_oper_get(payload);
+}
+
+/* PPAD - Port Physical Address Register
+ * -------------------------------------
+ * The PPAD register configures the per port physical MAC address.
+ */
+#define MLXSW_REG_PPAD_ID 0x5005
+#define MLXSW_REG_PPAD_LEN 0x10
+
+MLXSW_REG_DEFINE(ppad, MLXSW_REG_PPAD_ID, MLXSW_REG_PPAD_LEN);
+
+/* reg_ppad_single_base_mac
+ * 0: base_mac, local port should be 0 and mac[7:0] is
+ * reserved. HW will set incremental
+ * 1: single_mac - mac of the local_port
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ppad, single_base_mac, 0x00, 28, 1);
+
+/* reg_ppad_local_port
+ * port number, if single_base_mac = 0 then local_port is reserved
+ * Access: RW
+ */
+MLXSW_ITEM32_LP(reg, ppad, 0x00, 16, 0x00, 24);
+
+/* reg_ppad_mac
+ * If single_base_mac = 0 - base MAC address, mac[7:0] is reserved.
+ * If single_base_mac = 1 - the per port MAC address
+ * Access: RW
+ */
+MLXSW_ITEM_BUF(reg, ppad, mac, 0x02, 6);
+
+static inline void mlxsw_reg_ppad_pack(char *payload, bool single_base_mac,
+				       u16 local_port)
+{
+	MLXSW_REG_ZERO(ppad, payload);
+	mlxsw_reg_ppad_single_base_mac_set(payload, !!single_base_mac);
+	mlxsw_reg_ppad_local_port_set(payload, local_port);
+}
+
+/* PAOS - Ports Administrative and Operational Status Register
+ * -----------------------------------------------------------
+ * Configures and retrieves per port administrative and operational status.
+ */
+#define MLXSW_REG_PAOS_ID 0x5006
+#define MLXSW_REG_PAOS_LEN 0x10
+
+MLXSW_REG_DEFINE(paos, MLXSW_REG_PAOS_ID, MLXSW_REG_PAOS_LEN);
+
+/* reg_paos_swid
+ * Switch partition ID with which to associate the port.
+ * Note: while external ports uses unique local port numbers (and thus swid is
+ * redundant), router ports use the same local port number where swid is the
+ * only indication for the relevant port.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, paos, swid, 0x00, 24, 8);
+
+/* reg_paos_local_port
+ * Local port number.
+ * Access: Index
+ */
+MLXSW_ITEM32_LP(reg, paos, 0x00, 16, 0x00, 12);
+
+/* reg_paos_admin_status
+ * Port administrative state (the desired state of the port):
+ * 1 - Up.
+ * 2 - Down.
+ * 3 - Up once. This means that in case of link failure, the port won't go
+ *     into polling mode, but will wait to be re-enabled by software.
+ * 4 - Disabled by system. Can only be set by hardware.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, paos, admin_status, 0x00, 8, 4);
+
+/* reg_paos_oper_status
+ * Port operational state (the current state):
+ * 1 - Up.
+ * 2 - Down.
+ * 3 - Down by port failure. This means that the device will not let the
+ *     port up again until explicitly specified by software.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, paos, oper_status, 0x00, 0, 4);
+
+/* reg_paos_ase
+ * Admin state update enabled.
+ * Access: WO
+ */
+MLXSW_ITEM32(reg, paos, ase, 0x04, 31, 1);
+
+/* reg_paos_ee
+ * Event update enable. If this bit is set, event generation will be
+ * updated based on the e field.
+ * Access: WO
+ */
+MLXSW_ITEM32(reg, paos, ee, 0x04, 30, 1);
+
+/* reg_paos_e
+ * Event generation on operational state change:
+ * 0 - Do not generate event.
+ * 1 - Generate Event.
+ * 2 - Generate Single Event.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, paos, e, 0x04, 0, 2);
+
+static inline void mlxsw_reg_paos_pack(char *payload, u16 local_port,
+				       enum mlxsw_port_admin_status status)
+{
+	MLXSW_REG_ZERO(paos, payload);
+	mlxsw_reg_paos_swid_set(payload, 0);
+	mlxsw_reg_paos_local_port_set(payload, local_port);
+	mlxsw_reg_paos_admin_status_set(payload, status);
+	mlxsw_reg_paos_oper_status_set(payload, 0);
+	mlxsw_reg_paos_ase_set(payload, 1);
+	mlxsw_reg_paos_ee_set(payload, 1);
+	mlxsw_reg_paos_e_set(payload, 1);
+}
+
+/* PFCC - Ports Flow Control Configuration Register
+ * ------------------------------------------------
+ * Configures and retrieves the per port flow control configuration.
+ */
+#define MLXSW_REG_PFCC_ID 0x5007
+#define MLXSW_REG_PFCC_LEN 0x20
+
+MLXSW_REG_DEFINE(pfcc, MLXSW_REG_PFCC_ID, MLXSW_REG_PFCC_LEN);
+
+/* reg_pfcc_local_port
+ * Local port number.
+ * Access: Index
+ */
+MLXSW_ITEM32_LP(reg, pfcc, 0x00, 16, 0x00, 12);
+
+/* reg_pfcc_pnat
+ * Port number access type. Determines the way local_port is interpreted:
+ * 0 - Local port number.
+ * 1 - IB / label port number.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, pfcc, pnat, 0x00, 14, 2);
+
+/* reg_pfcc_shl_cap
+ * Send to higher layers capabilities:
+ * 0 - No capability of sending Pause and PFC frames to higher layers.
+ * 1 - Device has capability of sending Pause and PFC frames to higher
+ *     layers.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, pfcc, shl_cap, 0x00, 1, 1);
+
+/* reg_pfcc_shl_opr
+ * Send to higher layers operation:
+ * 0 - Pause and PFC frames are handled by the port (default).
+ * 1 - Pause and PFC frames are handled by the port and also sent to
+ *     higher layers. Only valid if shl_cap = 1.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, pfcc, shl_opr, 0x00, 0, 1);
+
+/* reg_pfcc_ppan
+ * Pause policy auto negotiation.
+ * 0 - Disabled. Generate / ignore Pause frames based on pptx / pprtx.
+ * 1 - Enabled. When auto-negotiation is performed, set the Pause policy
+ *     based on the auto-negotiation resolution.
+ * Access: RW
+ *
+ * Note: The auto-negotiation advertisement is set according to pptx and
+ * pprtx. When PFC is set on Tx / Rx, ppan must be set to 0.
+ */
+MLXSW_ITEM32(reg, pfcc, ppan, 0x04, 28, 4);
+
+/* reg_pfcc_prio_mask_tx
+ * Bit per priority indicating if Tx flow control policy should be
+ * updated based on bit pfctx.
+ * Access: WO
+ */
+MLXSW_ITEM32(reg, pfcc, prio_mask_tx, 0x04, 16, 8);
+
+/* reg_pfcc_prio_mask_rx
+ * Bit per priority indicating if Rx flow control policy should be
+ * updated based on bit pfcrx.
+ * Access: WO
+ */
+MLXSW_ITEM32(reg, pfcc, prio_mask_rx, 0x04, 0, 8);
+
+/* reg_pfcc_pptx
+ * Admin Pause policy on Tx.
+ * 0 - Never generate Pause frames (default).
+ * 1 - Generate Pause frames according to Rx buffer threshold.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, pfcc, pptx, 0x08, 31, 1);
+
+/* reg_pfcc_aptx
+ * Active (operational) Pause policy on Tx.
+ * 0 - Never generate Pause frames.
+ * 1 - Generate Pause frames according to Rx buffer threshold.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, pfcc, aptx, 0x08, 30, 1);
+
+/* reg_pfcc_pfctx
+ * Priority based flow control policy on Tx[7:0]. Per-priority bit mask:
+ * 0 - Never generate priority Pause frames on the specified priority
+ *     (default).
+ * 1 - Generate priority Pause frames according to Rx buffer threshold on
+ *     the specified priority.
+ * Access: RW
+ *
+ * Note: pfctx and pptx must be mutually exclusive.
+ */
+MLXSW_ITEM32(reg, pfcc, pfctx, 0x08, 16, 8);
+
+/* reg_pfcc_pprx
+ * Admin Pause policy on Rx.
+ * 0 - Ignore received Pause frames (default).
+ * 1 - Respect received Pause frames.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, pfcc, pprx, 0x0C, 31, 1);
+
+/* reg_pfcc_aprx
+ * Active (operational) Pause policy on Rx.
+ * 0 - Ignore received Pause frames.
+ * 1 - Respect received Pause frames.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, pfcc, aprx, 0x0C, 30, 1);
+
+/* reg_pfcc_pfcrx
+ * Priority based flow control policy on Rx[7:0]. Per-priority bit mask:
+ * 0 - Ignore incoming priority Pause frames on the specified priority
+ *     (default).
+ * 1 - Respect incoming priority Pause frames on the specified priority.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, pfcc, pfcrx, 0x0C, 16, 8);
+
+#define MLXSW_REG_PFCC_ALL_PRIO 0xFF
+
+static inline void mlxsw_reg_pfcc_prio_pack(char *payload, u8 pfc_en)
+{
+	mlxsw_reg_pfcc_prio_mask_tx_set(payload, MLXSW_REG_PFCC_ALL_PRIO);
+	mlxsw_reg_pfcc_prio_mask_rx_set(payload, MLXSW_REG_PFCC_ALL_PRIO);
+	mlxsw_reg_pfcc_pfctx_set(payload, pfc_en);
+	mlxsw_reg_pfcc_pfcrx_set(payload, pfc_en);
+}
+
+static inline void mlxsw_reg_pfcc_pack(char *payload, u16 local_port)
+{
+	MLXSW_REG_ZERO(pfcc, payload);
+	mlxsw_reg_pfcc_local_port_set(payload, local_port);
+}
+
+/* PPCNT - Ports Performance Counters Register
+ * -------------------------------------------
+ * The PPCNT register retrieves per port performance counters.
+ */
+#define MLXSW_REG_PPCNT_ID 0x5008
+#define MLXSW_REG_PPCNT_LEN 0x100
+#define MLXSW_REG_PPCNT_COUNTERS_OFFSET 0x08
+
+MLXSW_REG_DEFINE(ppcnt, MLXSW_REG_PPCNT_ID, MLXSW_REG_PPCNT_LEN);
+
+/* reg_ppcnt_swid
+ * For HCA: must be always 0.
+ * Switch partition ID to associate port with.
+ * Switch partitions are numbered from 0 to 7 inclusively.
+ * Switch partition 254 indicates stacking ports.
+ * Switch partition 255 indicates all switch partitions.
+ * Only valid on Set() operation with local_port=255.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, ppcnt, swid, 0x00, 24, 8);
+
+/* reg_ppcnt_local_port
+ * Local port number.
+ * Access: Index
+ */
+MLXSW_ITEM32_LP(reg, ppcnt, 0x00, 16, 0x00, 12);
+
+/* reg_ppcnt_pnat
+ * Port number access type:
+ * 0 - Local port number
+ * 1 - IB port number
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, ppcnt, pnat, 0x00, 14, 2);
+
+enum mlxsw_reg_ppcnt_grp {
+	MLXSW_REG_PPCNT_IEEE_8023_CNT = 0x0,
+	MLXSW_REG_PPCNT_RFC_2863_CNT = 0x1,
+	MLXSW_REG_PPCNT_RFC_2819_CNT = 0x2,
+	MLXSW_REG_PPCNT_RFC_3635_CNT = 0x3,
+	MLXSW_REG_PPCNT_EXT_CNT = 0x5,
+	MLXSW_REG_PPCNT_DISCARD_CNT = 0x6,
+	MLXSW_REG_PPCNT_PRIO_CNT = 0x10,
+	MLXSW_REG_PPCNT_TC_CNT = 0x11,
+	MLXSW_REG_PPCNT_TC_CONG_CNT = 0x13,
+};
+
+/* reg_ppcnt_grp
+ * Performance counter group.
+ * Group 63 indicates all groups. Only valid on Set() operation with
+ * clr bit set.
+ * 0x0: IEEE 802.3 Counters
+ * 0x1: RFC 2863 Counters
+ * 0x2: RFC 2819 Counters
+ * 0x3: RFC 3635 Counters
+ * 0x5: Ethernet Extended Counters
+ * 0x6: Ethernet Discard Counters
+ * 0x8: Link Level Retransmission Counters
+ * 0x10: Per Priority Counters
+ * 0x11: Per Traffic Class Counters
+ * 0x12: Physical Layer Counters
+ * 0x13: Per Traffic Class Congestion Counters
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, ppcnt, grp, 0x00, 0, 6);
+
+/* reg_ppcnt_clr
+ * Clear counters. Setting the clr bit will reset the counter value
+ * for all counters in the counter group. This bit can be set
+ * for both Set() and Get() operation.
+ * Access: OP
+ */
+MLXSW_ITEM32(reg, ppcnt, clr, 0x04, 31, 1);
+
+/* reg_ppcnt_lp_gl
+ * Local port global variable.
+ * 0: local_port 255 = all ports of the device.
+ * 1: local_port indicates local port number for all ports.
+ * Access: OP
+ */
+MLXSW_ITEM32(reg, ppcnt, lp_gl, 0x04, 30, 1);
+
+/* reg_ppcnt_prio_tc
+ * Priority for counter set that support per priority, valid values: 0-7.
+ * Traffic class for counter set that support per traffic class,
+ * valid values: 0- cap_max_tclass-1 .
+ * For HCA: cap_max_tclass is always 8.
+ * Otherwise must be 0.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, ppcnt, prio_tc, 0x04, 0, 5);
+
+/* Ethernet IEEE 802.3 Counter Group */
+
+/* reg_ppcnt_a_frames_transmitted_ok
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, a_frames_transmitted_ok,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x00, 0, 64);
+
+/* reg_ppcnt_a_frames_received_ok
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, a_frames_received_ok,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x08, 0, 64);
+
+/* reg_ppcnt_a_frame_check_sequence_errors
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, a_frame_check_sequence_errors,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x10, 0, 64);
+
+/* reg_ppcnt_a_alignment_errors
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, a_alignment_errors,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x18, 0, 64);
+
+/* reg_ppcnt_a_octets_transmitted_ok
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, a_octets_transmitted_ok,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x20, 0, 64);
+
+/* reg_ppcnt_a_octets_received_ok
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, a_octets_received_ok,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x28, 0, 64);
+
+/* reg_ppcnt_a_multicast_frames_xmitted_ok
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, a_multicast_frames_xmitted_ok,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x30, 0, 64);
+
+/* reg_ppcnt_a_broadcast_frames_xmitted_ok
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, a_broadcast_frames_xmitted_ok,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x38, 0, 64);
+
+/* reg_ppcnt_a_multicast_frames_received_ok
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, a_multicast_frames_received_ok,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x40, 0, 64);
+
+/* reg_ppcnt_a_broadcast_frames_received_ok
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, a_broadcast_frames_received_ok,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x48, 0, 64);
+
+/* reg_ppcnt_a_in_range_length_errors
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, a_in_range_length_errors,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x50, 0, 64);
+
+/* reg_ppcnt_a_out_of_range_length_field
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, a_out_of_range_length_field,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x58, 0, 64);
+
+/* reg_ppcnt_a_frame_too_long_errors
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, a_frame_too_long_errors,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x60, 0, 64);
+
+/* reg_ppcnt_a_symbol_error_during_carrier
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, a_symbol_error_during_carrier,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x68, 0, 64);
+
+/* reg_ppcnt_a_mac_control_frames_transmitted
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, a_mac_control_frames_transmitted,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x70, 0, 64);
+
+/* reg_ppcnt_a_mac_control_frames_received
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, a_mac_control_frames_received,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x78, 0, 64);
+
+/* reg_ppcnt_a_unsupported_opcodes_received
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, a_unsupported_opcodes_received,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x80, 0, 64);
+
+/* reg_ppcnt_a_pause_mac_ctrl_frames_received
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, a_pause_mac_ctrl_frames_received,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x88, 0, 64);
+
+/* reg_ppcnt_a_pause_mac_ctrl_frames_transmitted
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, a_pause_mac_ctrl_frames_transmitted,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x90, 0, 64);
+
+/* Ethernet RFC 2863 Counter Group */
+
+/* reg_ppcnt_if_in_discards
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, if_in_discards,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x10, 0, 64);
+
+/* reg_ppcnt_if_out_discards
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, if_out_discards,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x38, 0, 64);
+
+/* reg_ppcnt_if_out_errors
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, if_out_errors,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x40, 0, 64);
+
+/* Ethernet RFC 2819 Counter Group */
+
+/* reg_ppcnt_ether_stats_undersize_pkts
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, ether_stats_undersize_pkts,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x30, 0, 64);
+
+/* reg_ppcnt_ether_stats_oversize_pkts
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, ether_stats_oversize_pkts,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x38, 0, 64);
+
+/* reg_ppcnt_ether_stats_fragments
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, ether_stats_fragments,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x40, 0, 64);
+
+/* reg_ppcnt_ether_stats_pkts64octets
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, ether_stats_pkts64octets,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x58, 0, 64);
+
+/* reg_ppcnt_ether_stats_pkts65to127octets
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, ether_stats_pkts65to127octets,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x60, 0, 64);
+
+/* reg_ppcnt_ether_stats_pkts128to255octets
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, ether_stats_pkts128to255octets,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x68, 0, 64);
+
+/* reg_ppcnt_ether_stats_pkts256to511octets
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, ether_stats_pkts256to511octets,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x70, 0, 64);
+
+/* reg_ppcnt_ether_stats_pkts512to1023octets
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, ether_stats_pkts512to1023octets,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x78, 0, 64);
+
+/* reg_ppcnt_ether_stats_pkts1024to1518octets
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, ether_stats_pkts1024to1518octets,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x80, 0, 64);
+
+/* reg_ppcnt_ether_stats_pkts1519to2047octets
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, ether_stats_pkts1519to2047octets,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x88, 0, 64);
+
+/* reg_ppcnt_ether_stats_pkts2048to4095octets
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, ether_stats_pkts2048to4095octets,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x90, 0, 64);
+
+/* reg_ppcnt_ether_stats_pkts4096to8191octets
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, ether_stats_pkts4096to8191octets,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x98, 0, 64);
+
+/* reg_ppcnt_ether_stats_pkts8192to10239octets
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, ether_stats_pkts8192to10239octets,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0xA0, 0, 64);
+
+/* Ethernet RFC 3635 Counter Group */
+
+/* reg_ppcnt_dot3stats_fcs_errors
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, dot3stats_fcs_errors,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x08, 0, 64);
+
+/* reg_ppcnt_dot3stats_symbol_errors
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, dot3stats_symbol_errors,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x60, 0, 64);
+
+/* reg_ppcnt_dot3control_in_unknown_opcodes
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, dot3control_in_unknown_opcodes,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x68, 0, 64);
+
+/* reg_ppcnt_dot3in_pause_frames
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, dot3in_pause_frames,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x70, 0, 64);
+
+/* Ethernet Extended Counter Group Counters */
+
+/* reg_ppcnt_ecn_marked
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, ecn_marked,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x08, 0, 64);
+
+/* Ethernet Discard Counter Group Counters */
+
+/* reg_ppcnt_ingress_general
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, ingress_general,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x00, 0, 64);
+
+/* reg_ppcnt_ingress_policy_engine
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, ingress_policy_engine,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x08, 0, 64);
+
+/* reg_ppcnt_ingress_vlan_membership
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, ingress_vlan_membership,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x10, 0, 64);
+
+/* reg_ppcnt_ingress_tag_frame_type
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, ingress_tag_frame_type,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x18, 0, 64);
+
+/* reg_ppcnt_egress_vlan_membership
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, egress_vlan_membership,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x20, 0, 64);
+
+/* reg_ppcnt_loopback_filter
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, loopback_filter,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x28, 0, 64);
+
+/* reg_ppcnt_egress_general
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, egress_general,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x30, 0, 64);
+
+/* reg_ppcnt_egress_hoq
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, egress_hoq,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x40, 0, 64);
+
+/* reg_ppcnt_egress_policy_engine
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, egress_policy_engine,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x50, 0, 64);
+
+/* reg_ppcnt_ingress_tx_link_down
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, ingress_tx_link_down,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x58, 0, 64);
+
+/* reg_ppcnt_egress_stp_filter
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, egress_stp_filter,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x60, 0, 64);
+
+/* reg_ppcnt_egress_sll
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, egress_sll,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x70, 0, 64);
+
+/* Ethernet Per Priority Group Counters */
+
+/* reg_ppcnt_rx_octets
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, rx_octets,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x00, 0, 64);
+
+/* reg_ppcnt_rx_frames
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, rx_frames,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x20, 0, 64);
+
+/* reg_ppcnt_tx_octets
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, tx_octets,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x28, 0, 64);
+
+/* reg_ppcnt_tx_frames
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, tx_frames,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x48, 0, 64);
+
+/* reg_ppcnt_rx_pause
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, rx_pause,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x50, 0, 64);
+
+/* reg_ppcnt_rx_pause_duration
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, rx_pause_duration,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x58, 0, 64);
+
+/* reg_ppcnt_tx_pause
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, tx_pause,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x60, 0, 64);
+
+/* reg_ppcnt_tx_pause_duration
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, tx_pause_duration,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x68, 0, 64);
+
+/* reg_ppcnt_rx_pause_transition
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, tx_pause_transition,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x70, 0, 64);
+
+/* Ethernet Per Traffic Class Counters */
+
+/* reg_ppcnt_tc_transmit_queue
+ * Contains the transmit queue depth in cells of traffic class
+ * selected by prio_tc and the port selected by local_port.
+ * The field cannot be cleared.
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, tc_transmit_queue,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x00, 0, 64);
+
+/* reg_ppcnt_tc_no_buffer_discard_uc
+ * The number of unicast packets dropped due to lack of shared
+ * buffer resources.
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, tc_no_buffer_discard_uc,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x08, 0, 64);
+
+/* Ethernet Per Traffic Class Congestion Group Counters */
+
+/* reg_ppcnt_wred_discard
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, wred_discard,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x00, 0, 64);
+
+/* reg_ppcnt_ecn_marked_tc
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, ecn_marked_tc,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x08, 0, 64);
+
+static inline void mlxsw_reg_ppcnt_pack(char *payload, u16 local_port,
+					enum mlxsw_reg_ppcnt_grp grp,
+					u8 prio_tc)
+{
+	MLXSW_REG_ZERO(ppcnt, payload);
+	mlxsw_reg_ppcnt_swid_set(payload, 0);
+	mlxsw_reg_ppcnt_local_port_set(payload, local_port);
+	mlxsw_reg_ppcnt_pnat_set(payload, 0);
+	mlxsw_reg_ppcnt_grp_set(payload, grp);
+	mlxsw_reg_ppcnt_clr_set(payload, 0);
+	mlxsw_reg_ppcnt_lp_gl_set(payload, 1);
+	mlxsw_reg_ppcnt_prio_tc_set(payload, prio_tc);
+}
+
+/* PPTB - Port Prio To Buffer Register
+ * -----------------------------------
+ * Configures the switch priority to buffer table.
+ */
+#define MLXSW_REG_PPTB_ID 0x500B
+#define MLXSW_REG_PPTB_LEN 0x10
+
+MLXSW_REG_DEFINE(pptb, MLXSW_REG_PPTB_ID, MLXSW_REG_PPTB_LEN);
+
+enum {
+	MLXSW_REG_PPTB_MM_UM,
+	MLXSW_REG_PPTB_MM_UNICAST,
+	MLXSW_REG_PPTB_MM_MULTICAST,
+};
+
+/* reg_pptb_mm
+ * Mapping mode.
+ * 0 - Map both unicast and multicast packets to the same buffer.
+ * 1 - Map only unicast packets.
+ * 2 - Map only multicast packets.
+ * Access: Index
+ *
+ * Note: SwitchX-2 only supports the first option.
+ */
+MLXSW_ITEM32(reg, pptb, mm, 0x00, 28, 2);
+
+/* reg_pptb_local_port
+ * Local port number.
+ * Access: Index
+ */
+MLXSW_ITEM32_LP(reg, pptb, 0x00, 16, 0x00, 12);
+
+/* reg_pptb_um
+ * Enables the update of the untagged_buf field.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, pptb, um, 0x00, 8, 1);
+
+/* reg_pptb_pm
+ * Enables the update of the prio_to_buff field.
+ * Bit <i> is a flag for updating the mapping for switch priority <i>.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, pptb, pm, 0x00, 0, 8);
+
+/* reg_pptb_prio_to_buff
+ * Mapping of switch priority <i> to one of the allocated receive port
+ * buffers.
+ * Access: RW
+ */
+MLXSW_ITEM_BIT_ARRAY(reg, pptb, prio_to_buff, 0x04, 0x04, 4);
+
+/* reg_pptb_pm_msb
+ * Enables the update of the prio_to_buff field.
+ * Bit <i> is a flag for updating the mapping for switch priority <i+8>.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, pptb, pm_msb, 0x08, 24, 8);
+
+/* reg_pptb_untagged_buff
+ * Mapping of untagged frames to one of the allocated receive port buffers.
+ * Access: RW
+ *
+ * Note: In SwitchX-2 this field must be mapped to buffer 8. Reserved for
+ * Spectrum, as it maps untagged packets based on the default switch priority.
+ */
+MLXSW_ITEM32(reg, pptb, untagged_buff, 0x08, 0, 4);
+
+/* reg_pptb_prio_to_buff_msb
+ * Mapping of switch priority <i+8> to one of the allocated receive port
+ * buffers.
+ * Access: RW
+ */
+MLXSW_ITEM_BIT_ARRAY(reg, pptb, prio_to_buff_msb, 0x0C, 0x04, 4);
+
+#define MLXSW_REG_PPTB_ALL_PRIO 0xFF
+
+static inline void mlxsw_reg_pptb_pack(char *payload, u16 local_port)
+{
+	MLXSW_REG_ZERO(pptb, payload);
+	mlxsw_reg_pptb_mm_set(payload, MLXSW_REG_PPTB_MM_UM);
+	mlxsw_reg_pptb_local_port_set(payload, local_port);
+	mlxsw_reg_pptb_pm_set(payload, MLXSW_REG_PPTB_ALL_PRIO);
+	mlxsw_reg_pptb_pm_msb_set(payload, MLXSW_REG_PPTB_ALL_PRIO);
+}
+
+static inline void mlxsw_reg_pptb_prio_to_buff_pack(char *payload, u8 prio,
+						    u8 buff)
+{
+	mlxsw_reg_pptb_prio_to_buff_set(payload, prio, buff);
+	mlxsw_reg_pptb_prio_to_buff_msb_set(payload, prio, buff);
+}
+
+/* PBMC - Port Buffer Management Control Register
+ * ----------------------------------------------
+ * The PBMC register configures and retrieves the port packet buffer
+ * allocation for different Prios, and the Pause threshold management.
+ */
+#define MLXSW_REG_PBMC_ID 0x500C
+#define MLXSW_REG_PBMC_LEN 0x6C
+
+MLXSW_REG_DEFINE(pbmc, MLXSW_REG_PBMC_ID, MLXSW_REG_PBMC_LEN);
+
+/* reg_pbmc_local_port
+ * Local port number.
+ * Access: Index
+ */
+MLXSW_ITEM32_LP(reg, pbmc, 0x00, 16, 0x00, 12);
+
+/* reg_pbmc_xoff_timer_value
+ * When device generates a pause frame, it uses this value as the pause
+ * timer (time for the peer port to pause in quota-512 bit time).
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, pbmc, xoff_timer_value, 0x04, 16, 16);
+
+/* reg_pbmc_xoff_refresh
+ * The time before a new pause frame should be sent to refresh the pause RW
+ * state. Using the same units as xoff_timer_value above (in quota-512 bit
+ * time).
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, pbmc, xoff_refresh, 0x04, 0, 16);
+
+#define MLXSW_REG_PBMC_PORT_SHARED_BUF_IDX 11
+
+/* reg_pbmc_buf_lossy
+ * The field indicates if the buffer is lossy.
+ * 0 - Lossless
+ * 1 - Lossy
+ * Access: RW
+ */
+MLXSW_ITEM32_INDEXED(reg, pbmc, buf_lossy, 0x0C, 25, 1, 0x08, 0x00, false);
+
+/* reg_pbmc_buf_epsb
+ * Eligible for Port Shared buffer.
+ * If epsb is set, packets assigned to buffer are allowed to insert the port
+ * shared buffer.
+ * When buf_lossy is MLXSW_REG_PBMC_LOSSY_LOSSY this field is reserved.
+ * Access: RW
+ */
+MLXSW_ITEM32_INDEXED(reg, pbmc, buf_epsb, 0x0C, 24, 1, 0x08, 0x00, false);
+
+/* reg_pbmc_buf_size
+ * The part of the packet buffer array is allocated for the specific buffer.
+ * Units are represented in cells.
+ * Access: RW
+ */
+MLXSW_ITEM32_INDEXED(reg, pbmc, buf_size, 0x0C, 0, 16, 0x08, 0x00, false);
+
+/* reg_pbmc_buf_xoff_threshold
+ * Once the amount of data in the buffer goes above this value, device
+ * starts sending PFC frames for all priorities associated with the
+ * buffer. Units are represented in cells. Reserved in case of lossy
+ * buffer.
+ * Access: RW
+ *
+ * Note: In Spectrum, reserved for buffer[9].
+ */
+MLXSW_ITEM32_INDEXED(reg, pbmc, buf_xoff_threshold, 0x0C, 16, 16,
+		     0x08, 0x04, false);
+
+/* reg_pbmc_buf_xon_threshold
+ * When the amount of data in the buffer goes below this value, device
+ * stops sending PFC frames for the priorities associated with the
+ * buffer. Units are represented in cells. Reserved in case of lossy
+ * buffer.
+ * Access: RW
+ *
+ * Note: In Spectrum, reserved for buffer[9].
+ */
+MLXSW_ITEM32_INDEXED(reg, pbmc, buf_xon_threshold, 0x0C, 0, 16,
+		     0x08, 0x04, false);
+
+static inline void mlxsw_reg_pbmc_pack(char *payload, u16 local_port,
+				       u16 xoff_timer_value, u16 xoff_refresh)
+{
+	MLXSW_REG_ZERO(pbmc, payload);
+	mlxsw_reg_pbmc_local_port_set(payload, local_port);
+	mlxsw_reg_pbmc_xoff_timer_value_set(payload, xoff_timer_value);
+	mlxsw_reg_pbmc_xoff_refresh_set(payload, xoff_refresh);
+}
+
+static inline void mlxsw_reg_pbmc_lossy_buffer_pack(char *payload,
+						    int buf_index,
+						    u16 size)
+{
+	mlxsw_reg_pbmc_buf_lossy_set(payload, buf_index, 1);
+	mlxsw_reg_pbmc_buf_epsb_set(payload, buf_index, 0);
+	mlxsw_reg_pbmc_buf_size_set(payload, buf_index, size);
+}
+
+static inline void mlxsw_reg_pbmc_lossless_buffer_pack(char *payload,
+						       int buf_index, u16 size,
+						       u16 threshold)
+{
+	mlxsw_reg_pbmc_buf_lossy_set(payload, buf_index, 0);
+	mlxsw_reg_pbmc_buf_epsb_set(payload, buf_index, 0);
+	mlxsw_reg_pbmc_buf_size_set(payload, buf_index, size);
+	mlxsw_reg_pbmc_buf_xoff_threshold_set(payload, buf_index, threshold);
+	mlxsw_reg_pbmc_buf_xon_threshold_set(payload, buf_index, threshold);
+}
+
+/* PSPA - Port Switch Partition Allocation
+ * ---------------------------------------
+ * Controls the association of a port with a switch partition and enables
+ * configuring ports as stacking ports.
+ */
+#define MLXSW_REG_PSPA_ID 0x500D
+#define MLXSW_REG_PSPA_LEN 0x8
+
+MLXSW_REG_DEFINE(pspa, MLXSW_REG_PSPA_ID, MLXSW_REG_PSPA_LEN);
+
+/* reg_pspa_swid
+ * Switch partition ID.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, pspa, swid, 0x00, 24, 8);
+
+/* reg_pspa_local_port
+ * Local port number.
+ * Access: Index
+ */
+MLXSW_ITEM32_LP(reg, pspa, 0x00, 16, 0x00, 0);
+
+/* reg_pspa_sub_port
+ * Virtual port within the local port. Set to 0 when virtual ports are
+ * disabled on the local port.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, pspa, sub_port, 0x00, 8, 8);
+
+static inline void mlxsw_reg_pspa_pack(char *payload, u8 swid, u16 local_port)
+{
+	MLXSW_REG_ZERO(pspa, payload);
+	mlxsw_reg_pspa_swid_set(payload, swid);
+	mlxsw_reg_pspa_local_port_set(payload, local_port);
+	mlxsw_reg_pspa_sub_port_set(payload, 0);
+}
+
+/* PMAOS - Ports Module Administrative and Operational Status
+ * ----------------------------------------------------------
+ * This register configures and retrieves the per module status.
+ */
+#define MLXSW_REG_PMAOS_ID 0x5012
+#define MLXSW_REG_PMAOS_LEN 0x10
+
+MLXSW_REG_DEFINE(pmaos, MLXSW_REG_PMAOS_ID, MLXSW_REG_PMAOS_LEN);
+
+/* reg_pmaos_rst
+ * Module reset toggle.
+ * Note: Setting reset while module is plugged-in will result in transition to
+ * "initializing" operational state.
+ * Access: OP
+ */
+MLXSW_ITEM32(reg, pmaos, rst, 0x00, 31, 1);
+
+/* reg_pmaos_slot_index
+ * Slot index.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, pmaos, slot_index, 0x00, 24, 4);
+
+/* reg_pmaos_module
+ * Module number.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, pmaos, module, 0x00, 16, 8);
+
+enum mlxsw_reg_pmaos_admin_status {
+	MLXSW_REG_PMAOS_ADMIN_STATUS_ENABLED = 1,
+	MLXSW_REG_PMAOS_ADMIN_STATUS_DISABLED = 2,
+	/* If the module is active and then unplugged, or experienced an error
+	 * event, the operational status should go to "disabled" and can only
+	 * be enabled upon explicit enable command.
+	 */
+	MLXSW_REG_PMAOS_ADMIN_STATUS_ENABLED_ONCE = 3,
+};
+
+/* reg_pmaos_admin_status
+ * Module administrative state (the desired state of the module).
+ * Note: To disable a module, all ports associated with the port must be
+ * administatively down first.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, pmaos, admin_status, 0x00, 8, 4);
+
+/* reg_pmaos_ase
+ * Admin state update enable.
+ * If this bit is set, admin state will be updated based on admin_state field.
+ * Only relevant on Set() operations.
+ * Access: WO
+ */
+MLXSW_ITEM32(reg, pmaos, ase, 0x04, 31, 1);
+
+/* reg_pmaos_ee
+ * Event update enable.
+ * If this bit is set, event generation will be updated based on the e field.
+ * Only relevant on Set operations.
+ * Access: WO
+ */
+MLXSW_ITEM32(reg, pmaos, ee, 0x04, 30, 1);
+
+enum mlxsw_reg_pmaos_e {
+	MLXSW_REG_PMAOS_E_DO_NOT_GENERATE_EVENT,
+	MLXSW_REG_PMAOS_E_GENERATE_EVENT,
+	MLXSW_REG_PMAOS_E_GENERATE_SINGLE_EVENT,
+};
+
+/* reg_pmaos_e
+ * Event Generation on operational state change.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, pmaos, e, 0x04, 0, 2);
+
+static inline void mlxsw_reg_pmaos_pack(char *payload, u8 slot_index, u8 module)
+{
+	MLXSW_REG_ZERO(pmaos, payload);
+	mlxsw_reg_pmaos_slot_index_set(payload, slot_index);
+	mlxsw_reg_pmaos_module_set(payload, module);
+}
+
+/* PPLR - Port Physical Loopback Register
+ * --------------------------------------
+ * This register allows configuration of the port's loopback mode.
+ */
+#define MLXSW_REG_PPLR_ID 0x5018
+#define MLXSW_REG_PPLR_LEN 0x8
+
+MLXSW_REG_DEFINE(pplr, MLXSW_REG_PPLR_ID, MLXSW_REG_PPLR_LEN);
+
+/* reg_pplr_local_port
+ * Local port number.
+ * Access: Index
+ */
+MLXSW_ITEM32_LP(reg, pplr, 0x00, 16, 0x00, 12);
+
+/* Phy local loopback. When set the port's egress traffic is looped back
+ * to the receiver and the port transmitter is disabled.
+ */
+#define MLXSW_REG_PPLR_LB_TYPE_BIT_PHY_LOCAL BIT(1)
+
+/* reg_pplr_lb_en
+ * Loopback enable.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, pplr, lb_en, 0x04, 0, 8);
+
+static inline void mlxsw_reg_pplr_pack(char *payload, u16 local_port,
+				       bool phy_local)
+{
+	MLXSW_REG_ZERO(pplr, payload);
+	mlxsw_reg_pplr_local_port_set(payload, local_port);
+	mlxsw_reg_pplr_lb_en_set(payload,
+				 phy_local ?
+				 MLXSW_REG_PPLR_LB_TYPE_BIT_PHY_LOCAL : 0);
+}
+
+/* PMTDB - Port Module To local DataBase Register
+ * ----------------------------------------------
+ * The PMTDB register allows to query the possible module<->local port
+ * mapping than can be used in PMLP. It does not represent the actual/current
+ * mapping of the local to module. Actual mapping is only defined by PMLP.
+ */
+#define MLXSW_REG_PMTDB_ID 0x501A
+#define MLXSW_REG_PMTDB_LEN 0x40
+
+MLXSW_REG_DEFINE(pmtdb, MLXSW_REG_PMTDB_ID, MLXSW_REG_PMTDB_LEN);
+
+/* reg_pmtdb_slot_index
+ * Slot index (0: Main board).
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, pmtdb, slot_index, 0x00, 24, 4);
+
+/* reg_pmtdb_module
+ * Module number.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, pmtdb, module, 0x00, 16, 8);
+
+/* reg_pmtdb_ports_width
+ * Port's width
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, pmtdb, ports_width, 0x00, 12, 4);
+
+/* reg_pmtdb_num_ports
+ * Number of ports in a single module (split/breakout)
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, pmtdb, num_ports, 0x00, 8, 4);
+
+enum mlxsw_reg_pmtdb_status {
+	MLXSW_REG_PMTDB_STATUS_SUCCESS,
+};
+
+/* reg_pmtdb_status
+ * Status
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, pmtdb, status, 0x00, 0, 4);
+
+/* reg_pmtdb_port_num
+ * The local_port value which can be assigned to the module.
+ * In case of more than one port, port<x> represent the /<x> port of
+ * the module.
+ * Access: RO
+ */
+MLXSW_ITEM16_INDEXED(reg, pmtdb, port_num, 0x04, 0, 10, 0x02, 0x00, false);
+
+static inline void mlxsw_reg_pmtdb_pack(char *payload, u8 slot_index, u8 module,
+					u8 ports_width, u8 num_ports)
+{
+	MLXSW_REG_ZERO(pmtdb, payload);
+	mlxsw_reg_pmtdb_slot_index_set(payload, slot_index);
+	mlxsw_reg_pmtdb_module_set(payload, module);
+	mlxsw_reg_pmtdb_ports_width_set(payload, ports_width);
+	mlxsw_reg_pmtdb_num_ports_set(payload, num_ports);
+}
+
+/* PMECR - Ports Mapping Event Configuration Register
+ * --------------------------------------------------
+ * The PMECR register is used to enable/disable event triggering
+ * in case of local port mapping change.
+ */
+#define MLXSW_REG_PMECR_ID 0x501B
+#define MLXSW_REG_PMECR_LEN 0x20
+
+MLXSW_REG_DEFINE(pmecr, MLXSW_REG_PMECR_ID, MLXSW_REG_PMECR_LEN);
+
+/* reg_pmecr_local_port
+ * Local port number.
+ * Access: Index
+ */
+MLXSW_ITEM32_LP(reg, pmecr, 0x00, 16, 0x00, 12);
+
+/* reg_pmecr_ee
+ * Event update enable. If this bit is set, event generation will be updated
+ * based on the e field. Only relevant on Set operations.
+ * Access: WO
+ */
+MLXSW_ITEM32(reg, pmecr, ee, 0x04, 30, 1);
+
+/* reg_pmecr_eswi
+ * Software ignore enable bit. If this bit is set, the value of swi is used.
+ * If this bit is clear, the value of swi is ignored.
+ * Only relevant on Set operations.
+ * Access: WO
+ */
+MLXSW_ITEM32(reg, pmecr, eswi, 0x04, 24, 1);
+
+/* reg_pmecr_swi
+ * Software ignore. If this bit is set, the device shouldn't generate events
+ * in case of PMLP SET operation but only upon self local port mapping change
+ * (if applicable according to e configuration). This is supplementary
+ * configuration on top of e value.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, pmecr, swi, 0x04, 8, 1);
+
+enum mlxsw_reg_pmecr_e {
+	MLXSW_REG_PMECR_E_DO_NOT_GENERATE_EVENT,
+	MLXSW_REG_PMECR_E_GENERATE_EVENT,
+	MLXSW_REG_PMECR_E_GENERATE_SINGLE_EVENT,
+};
+
+/* reg_pmecr_e
+ * Event generation on local port mapping change.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, pmecr, e, 0x04, 0, 2);
+
+static inline void mlxsw_reg_pmecr_pack(char *payload, u16 local_port,
+					enum mlxsw_reg_pmecr_e e)
+{
+	MLXSW_REG_ZERO(pmecr, payload);
+	mlxsw_reg_pmecr_local_port_set(payload, local_port);
+	mlxsw_reg_pmecr_e_set(payload, e);
+	mlxsw_reg_pmecr_ee_set(payload, true);
+	mlxsw_reg_pmecr_swi_set(payload, true);
+	mlxsw_reg_pmecr_eswi_set(payload, true);
+}
+
+/* PMPE - Port Module Plug/Unplug Event Register
+ * ---------------------------------------------
+ * This register reports any operational status change of a module.
+ * A change in the module’s state will generate an event only if the change
+ * happens after arming the event mechanism. Any changes to the module state
+ * while the event mechanism is not armed will not be reported. Software can
+ * query the PMPE register for module status.
+ */
+#define MLXSW_REG_PMPE_ID 0x5024
+#define MLXSW_REG_PMPE_LEN 0x10
+
+MLXSW_REG_DEFINE(pmpe, MLXSW_REG_PMPE_ID, MLXSW_REG_PMPE_LEN);
+
+/* reg_pmpe_slot_index
+ * Slot index.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, pmpe, slot_index, 0x00, 24, 4);
+
+/* reg_pmpe_module
+ * Module number.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, pmpe, module, 0x00, 16, 8);
+
+enum mlxsw_reg_pmpe_module_status {
+	MLXSW_REG_PMPE_MODULE_STATUS_PLUGGED_ENABLED = 1,
+	MLXSW_REG_PMPE_MODULE_STATUS_UNPLUGGED,
+	MLXSW_REG_PMPE_MODULE_STATUS_PLUGGED_ERROR,
+	MLXSW_REG_PMPE_MODULE_STATUS_PLUGGED_DISABLED,
+};
+
+/* reg_pmpe_module_status
+ * Module status.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, pmpe, module_status, 0x00, 0, 4);
+
+/* reg_pmpe_error_type
+ * Module error details.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, pmpe, error_type, 0x04, 8, 4);
+
+/* PDDR - Port Diagnostics Database Register
+ * -----------------------------------------
+ * The PDDR enables to read the Phy debug database
+ */
+#define MLXSW_REG_PDDR_ID 0x5031
+#define MLXSW_REG_PDDR_LEN 0x100
+
+MLXSW_REG_DEFINE(pddr, MLXSW_REG_PDDR_ID, MLXSW_REG_PDDR_LEN);
+
+/* reg_pddr_local_port
+ * Local port number.
+ * Access: Index
+ */
+MLXSW_ITEM32_LP(reg, pddr, 0x00, 16, 0x00, 12);
+
+enum mlxsw_reg_pddr_page_select {
+	MLXSW_REG_PDDR_PAGE_SELECT_TROUBLESHOOTING_INFO = 1,
+};
+
+/* reg_pddr_page_select
+ * Page select index.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, pddr, page_select, 0x04, 0, 8);
+
+enum mlxsw_reg_pddr_trblsh_group_opcode {
+	/* Monitor opcodes */
+	MLXSW_REG_PDDR_TRBLSH_GROUP_OPCODE_MONITOR,
+};
+
+/* reg_pddr_group_opcode
+ * Group selector.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, pddr, trblsh_group_opcode, 0x08, 0, 16);
+
+/* reg_pddr_status_opcode
+ * Group selector.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, pddr, trblsh_status_opcode, 0x0C, 0, 16);
+
+static inline void mlxsw_reg_pddr_pack(char *payload, u16 local_port,
+				       u8 page_select)
+{
+	MLXSW_REG_ZERO(pddr, payload);
+	mlxsw_reg_pddr_local_port_set(payload, local_port);
+	mlxsw_reg_pddr_page_select_set(payload, page_select);
+}
+
+/* PMMP - Port Module Memory Map Properties Register
+ * -------------------------------------------------
+ * The PMMP register allows to override the module memory map advertisement.
+ * The register can only be set when the module is disabled by PMAOS register.
+ */
+#define MLXSW_REG_PMMP_ID 0x5044
+#define MLXSW_REG_PMMP_LEN 0x2C
+
+MLXSW_REG_DEFINE(pmmp, MLXSW_REG_PMMP_ID, MLXSW_REG_PMMP_LEN);
+
+/* reg_pmmp_module
+ * Module number.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, pmmp, module, 0x00, 16, 8);
+
+/* reg_pmmp_slot_index
+ * Slot index.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, pmmp, slot_index, 0x00, 24, 4);
+
+/* reg_pmmp_sticky
+ * When set, will keep eeprom_override values after plug-out event.
+ * Access: OP
+ */
+MLXSW_ITEM32(reg, pmmp, sticky, 0x00, 0, 1);
+
+/* reg_pmmp_eeprom_override_mask
+ * Write mask bit (negative polarity).
+ * 0 - Allow write
+ * 1 - Ignore write
+ * On write, indicates which of the bits from eeprom_override field are
+ * updated.
+ * Access: WO
+ */
+MLXSW_ITEM32(reg, pmmp, eeprom_override_mask, 0x04, 16, 16);
+
+enum {
+	/* Set module to low power mode */
+	MLXSW_REG_PMMP_EEPROM_OVERRIDE_LOW_POWER_MASK = BIT(8),
+};
+
+/* reg_pmmp_eeprom_override
+ * Override / ignore EEPROM advertisement properties bitmask
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, pmmp, eeprom_override, 0x04, 0, 16);
+
+static inline void mlxsw_reg_pmmp_pack(char *payload, u8 slot_index, u8 module)
+{
+	MLXSW_REG_ZERO(pmmp, payload);
+	mlxsw_reg_pmmp_slot_index_set(payload, slot_index);
+	mlxsw_reg_pmmp_module_set(payload, module);
+}
+
+/* PLLP - Port Local port to Label Port mapping Register
+ * -----------------------------------------------------
+ * The PLLP register returns the mapping from Local Port into Label Port.
+ */
+#define MLXSW_REG_PLLP_ID 0x504A
+#define MLXSW_REG_PLLP_LEN 0x10
+
+MLXSW_REG_DEFINE(pllp, MLXSW_REG_PLLP_ID, MLXSW_REG_PLLP_LEN);
+
+/* reg_pllp_local_port
+ * Local port number.
+ * Access: Index
+ */
+MLXSW_ITEM32_LP(reg, pllp, 0x00, 16, 0x00, 12);
+
+/* reg_pllp_label_port
+ * Front panel label of the port.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, pllp, label_port, 0x00, 0, 8);
+
+/* reg_pllp_split_num
+ * Label split mapping for local_port.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, pllp, split_num, 0x04, 0, 4);
+
+/* reg_pllp_slot_index
+ * Slot index (0: Main board).
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, pllp, slot_index, 0x08, 0, 4);
+
+static inline void mlxsw_reg_pllp_pack(char *payload, u16 local_port)
+{
+	MLXSW_REG_ZERO(pllp, payload);
+	mlxsw_reg_pllp_local_port_set(payload, local_port);
+}
+
+static inline void mlxsw_reg_pllp_unpack(char *payload, u8 *label_port,
+					 u8 *split_num, u8 *slot_index)
+{
+	*label_port = mlxsw_reg_pllp_label_port_get(payload);
+	*split_num = mlxsw_reg_pllp_split_num_get(payload);
+	*slot_index = mlxsw_reg_pllp_slot_index_get(payload);
+}
+
+/* PMTM - Port Module Type Mapping Register
+ * ----------------------------------------
+ * The PMTM register allows query or configuration of module types.
+ * The register can only be set when the module is disabled by PMAOS register
+ */
+#define MLXSW_REG_PMTM_ID 0x5067
+#define MLXSW_REG_PMTM_LEN 0x10
+
+MLXSW_REG_DEFINE(pmtm, MLXSW_REG_PMTM_ID, MLXSW_REG_PMTM_LEN);
+
+/* reg_pmtm_slot_index
+ * Slot index.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, pmtm, slot_index, 0x00, 24, 4);
+
+/* reg_pmtm_module
+ * Module number.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, pmtm, module, 0x00, 16, 8);
+
+enum mlxsw_reg_pmtm_module_type {
+	MLXSW_REG_PMTM_MODULE_TYPE_BACKPLANE_4_LANES = 0,
+	MLXSW_REG_PMTM_MODULE_TYPE_QSFP = 1,
+	MLXSW_REG_PMTM_MODULE_TYPE_SFP = 2,
+	MLXSW_REG_PMTM_MODULE_TYPE_BACKPLANE_SINGLE_LANE = 4,
+	MLXSW_REG_PMTM_MODULE_TYPE_BACKPLANE_2_LANES = 8,
+	MLXSW_REG_PMTM_MODULE_TYPE_CHIP2CHIP4X = 10,
+	MLXSW_REG_PMTM_MODULE_TYPE_CHIP2CHIP2X = 11,
+	MLXSW_REG_PMTM_MODULE_TYPE_CHIP2CHIP1X = 12,
+	MLXSW_REG_PMTM_MODULE_TYPE_QSFP_DD = 14,
+	MLXSW_REG_PMTM_MODULE_TYPE_OSFP = 15,
+	MLXSW_REG_PMTM_MODULE_TYPE_SFP_DD = 16,
+	MLXSW_REG_PMTM_MODULE_TYPE_DSFP = 17,
+	MLXSW_REG_PMTM_MODULE_TYPE_CHIP2CHIP8X = 18,
+	MLXSW_REG_PMTM_MODULE_TYPE_TWISTED_PAIR = 19,
+};
+
+/* reg_pmtm_module_type
+ * Module type.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, pmtm, module_type, 0x04, 0, 5);
+
+static inline void mlxsw_reg_pmtm_pack(char *payload, u8 slot_index, u8 module)
+{
+	MLXSW_REG_ZERO(pmtm, payload);
+	mlxsw_reg_pmtm_slot_index_set(payload, slot_index);
+	mlxsw_reg_pmtm_module_set(payload, module);
+}
+
+/* HTGT - Host Trap Group Table
+ * ----------------------------
+ * Configures the properties for forwarding to CPU.
+ */
+#define MLXSW_REG_HTGT_ID 0x7002
+#define MLXSW_REG_HTGT_LEN 0x20
+
+MLXSW_REG_DEFINE(htgt, MLXSW_REG_HTGT_ID, MLXSW_REG_HTGT_LEN);
+
+/* reg_htgt_swid
+ * Switch partition ID.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, htgt, swid, 0x00, 24, 8);
+
+#define MLXSW_REG_HTGT_PATH_TYPE_LOCAL 0x0	/* For locally attached CPU */
+
+/* reg_htgt_type
+ * CPU path type.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, htgt, type, 0x00, 8, 4);
+
+enum mlxsw_reg_htgt_trap_group {
+	MLXSW_REG_HTGT_TRAP_GROUP_EMAD,
+	MLXSW_REG_HTGT_TRAP_GROUP_CORE_EVENT,
+	MLXSW_REG_HTGT_TRAP_GROUP_SP_STP,
+	MLXSW_REG_HTGT_TRAP_GROUP_SP_LACP,
+	MLXSW_REG_HTGT_TRAP_GROUP_SP_LLDP,
+	MLXSW_REG_HTGT_TRAP_GROUP_SP_MC_SNOOPING,
+	MLXSW_REG_HTGT_TRAP_GROUP_SP_BGP,
+	MLXSW_REG_HTGT_TRAP_GROUP_SP_OSPF,
+	MLXSW_REG_HTGT_TRAP_GROUP_SP_PIM,
+	MLXSW_REG_HTGT_TRAP_GROUP_SP_MULTICAST,
+	MLXSW_REG_HTGT_TRAP_GROUP_SP_NEIGH_DISCOVERY,
+	MLXSW_REG_HTGT_TRAP_GROUP_SP_ROUTER_EXP,
+	MLXSW_REG_HTGT_TRAP_GROUP_SP_EXTERNAL_ROUTE,
+	MLXSW_REG_HTGT_TRAP_GROUP_SP_IP2ME,
+	MLXSW_REG_HTGT_TRAP_GROUP_SP_DHCP,
+	MLXSW_REG_HTGT_TRAP_GROUP_SP_EVENT,
+	MLXSW_REG_HTGT_TRAP_GROUP_SP_IPV6,
+	MLXSW_REG_HTGT_TRAP_GROUP_SP_LBERROR,
+	MLXSW_REG_HTGT_TRAP_GROUP_SP_PTP0,
+	MLXSW_REG_HTGT_TRAP_GROUP_SP_PTP1,
+	MLXSW_REG_HTGT_TRAP_GROUP_SP_VRRP,
+	MLXSW_REG_HTGT_TRAP_GROUP_SP_PKT_SAMPLE,
+	MLXSW_REG_HTGT_TRAP_GROUP_SP_FLOW_LOGGING,
+	MLXSW_REG_HTGT_TRAP_GROUP_SP_FID_MISS,
+	MLXSW_REG_HTGT_TRAP_GROUP_SP_BFD,
+	MLXSW_REG_HTGT_TRAP_GROUP_SP_DUMMY,
+	MLXSW_REG_HTGT_TRAP_GROUP_SP_L2_DISCARDS,
+	MLXSW_REG_HTGT_TRAP_GROUP_SP_L3_DISCARDS,
+	MLXSW_REG_HTGT_TRAP_GROUP_SP_L3_EXCEPTIONS,
+	MLXSW_REG_HTGT_TRAP_GROUP_SP_TUNNEL_DISCARDS,
+	MLXSW_REG_HTGT_TRAP_GROUP_SP_ACL_DISCARDS,
+	MLXSW_REG_HTGT_TRAP_GROUP_SP_BUFFER_DISCARDS,
+
+	__MLXSW_REG_HTGT_TRAP_GROUP_MAX,
+	MLXSW_REG_HTGT_TRAP_GROUP_MAX = __MLXSW_REG_HTGT_TRAP_GROUP_MAX - 1
+};
+
+/* reg_htgt_trap_group
+ * Trap group number. User defined number specifying which trap groups
+ * should be forwarded to the CPU. The mapping between trap IDs and trap
+ * groups is configured using HPKT register.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, htgt, trap_group, 0x00, 0, 8);
+
+enum {
+	MLXSW_REG_HTGT_POLICER_DISABLE,
+	MLXSW_REG_HTGT_POLICER_ENABLE,
+};
+
+/* reg_htgt_pide
+ * Enable policer ID specified using 'pid' field.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, htgt, pide, 0x04, 15, 1);
+
+#define MLXSW_REG_HTGT_INVALID_POLICER 0xff
+
+/* reg_htgt_pid
+ * Policer ID for the trap group.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, htgt, pid, 0x04, 0, 8);
+
+#define MLXSW_REG_HTGT_TRAP_TO_CPU 0x0
+
+/* reg_htgt_mirror_action
+ * Mirror action to use.
+ * 0 - Trap to CPU.
+ * 1 - Trap to CPU and mirror to a mirroring agent.
+ * 2 - Mirror to a mirroring agent and do not trap to CPU.
+ * Access: RW
+ *
+ * Note: Mirroring to a mirroring agent is only supported in Spectrum.
+ */
+MLXSW_ITEM32(reg, htgt, mirror_action, 0x08, 8, 2);
+
+/* reg_htgt_mirroring_agent
+ * Mirroring agent.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, htgt, mirroring_agent, 0x08, 0, 3);
+
+#define MLXSW_REG_HTGT_DEFAULT_PRIORITY 0
+
+/* reg_htgt_priority
+ * Trap group priority.
+ * In case a packet matches multiple classification rules, the packet will
+ * only be trapped once, based on the trap ID associated with the group (via
+ * register HPKT) with the highest priority.
+ * Supported values are 0-7, with 7 represnting the highest priority.
+ * Access: RW
+ *
+ * Note: In SwitchX-2 this field is ignored and the priority value is replaced
+ * by the 'trap_group' field.
+ */
+MLXSW_ITEM32(reg, htgt, priority, 0x0C, 0, 4);
+
+#define MLXSW_REG_HTGT_DEFAULT_TC 7
+
+/* reg_htgt_local_path_cpu_tclass
+ * CPU ingress traffic class for the trap group.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, htgt, local_path_cpu_tclass, 0x10, 16, 6);
+
+enum mlxsw_reg_htgt_local_path_rdq {
+	MLXSW_REG_HTGT_LOCAL_PATH_RDQ_SX2_CTRL = 0x13,
+	MLXSW_REG_HTGT_LOCAL_PATH_RDQ_SX2_RX = 0x14,
+	MLXSW_REG_HTGT_LOCAL_PATH_RDQ_SX2_EMAD = 0x15,
+	MLXSW_REG_HTGT_LOCAL_PATH_RDQ_SIB_EMAD = 0x15,
+};
+/* reg_htgt_local_path_rdq
+ * Receive descriptor queue (RDQ) to use for the trap group.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, htgt, local_path_rdq, 0x10, 0, 6);
+
+static inline void mlxsw_reg_htgt_pack(char *payload, u8 group, u8 policer_id,
+				       u8 priority, u8 tc)
+{
+	MLXSW_REG_ZERO(htgt, payload);
+
+	if (policer_id == MLXSW_REG_HTGT_INVALID_POLICER) {
+		mlxsw_reg_htgt_pide_set(payload,
+					MLXSW_REG_HTGT_POLICER_DISABLE);
+	} else {
+		mlxsw_reg_htgt_pide_set(payload,
+					MLXSW_REG_HTGT_POLICER_ENABLE);
+		mlxsw_reg_htgt_pid_set(payload, policer_id);
+	}
+
+	mlxsw_reg_htgt_type_set(payload, MLXSW_REG_HTGT_PATH_TYPE_LOCAL);
+	mlxsw_reg_htgt_trap_group_set(payload, group);
+	mlxsw_reg_htgt_mirror_action_set(payload, MLXSW_REG_HTGT_TRAP_TO_CPU);
+	mlxsw_reg_htgt_mirroring_agent_set(payload, 0);
+	mlxsw_reg_htgt_priority_set(payload, priority);
+	mlxsw_reg_htgt_local_path_cpu_tclass_set(payload, tc);
+	mlxsw_reg_htgt_local_path_rdq_set(payload, group);
+}
+
+/* HPKT - Host Packet Trap
+ * -----------------------
+ * Configures trap IDs inside trap groups.
+ */
+#define MLXSW_REG_HPKT_ID 0x7003
+#define MLXSW_REG_HPKT_LEN 0x10
+
+MLXSW_REG_DEFINE(hpkt, MLXSW_REG_HPKT_ID, MLXSW_REG_HPKT_LEN);
+
+enum {
+	MLXSW_REG_HPKT_ACK_NOT_REQUIRED,
+	MLXSW_REG_HPKT_ACK_REQUIRED,
+};
+
+/* reg_hpkt_ack
+ * Require acknowledgements from the host for events.
+ * If set, then the device will wait for the event it sent to be acknowledged
+ * by the host. This option is only relevant for event trap IDs.
+ * Access: RW
+ *
+ * Note: Currently not supported by firmware.
+ */
+MLXSW_ITEM32(reg, hpkt, ack, 0x00, 24, 1);
+
+enum mlxsw_reg_hpkt_action {
+	MLXSW_REG_HPKT_ACTION_FORWARD,
+	MLXSW_REG_HPKT_ACTION_TRAP_TO_CPU,
+	MLXSW_REG_HPKT_ACTION_MIRROR_TO_CPU,
+	MLXSW_REG_HPKT_ACTION_DISCARD,
+	MLXSW_REG_HPKT_ACTION_SOFT_DISCARD,
+	MLXSW_REG_HPKT_ACTION_TRAP_AND_SOFT_DISCARD,
+	MLXSW_REG_HPKT_ACTION_TRAP_EXCEPTION_TO_CPU,
+	MLXSW_REG_HPKT_ACTION_SET_FW_DEFAULT = 15,
+};
+
+/* reg_hpkt_action
+ * Action to perform on packet when trapped.
+ * 0 - No action. Forward to CPU based on switching rules.
+ * 1 - Trap to CPU (CPU receives sole copy).
+ * 2 - Mirror to CPU (CPU receives a replica of the packet).
+ * 3 - Discard.
+ * 4 - Soft discard (allow other traps to act on the packet).
+ * 5 - Trap and soft discard (allow other traps to overwrite this trap).
+ * 6 - Trap to CPU (CPU receives sole copy) and count it as error.
+ * 15 - Restore the firmware's default action.
+ * Access: RW
+ *
+ * Note: Must be set to 0 (forward) for event trap IDs, as they are already
+ * addressed to the CPU.
+ */
+MLXSW_ITEM32(reg, hpkt, action, 0x00, 20, 3);
+
+/* reg_hpkt_trap_group
+ * Trap group to associate the trap with.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, hpkt, trap_group, 0x00, 12, 6);
+
+/* reg_hpkt_trap_id
+ * Trap ID.
+ * Access: Index
+ *
+ * Note: A trap ID can only be associated with a single trap group. The device
+ * will associate the trap ID with the last trap group configured.
+ */
+MLXSW_ITEM32(reg, hpkt, trap_id, 0x00, 0, 10);
+
+enum {
+	MLXSW_REG_HPKT_CTRL_PACKET_DEFAULT,
+	MLXSW_REG_HPKT_CTRL_PACKET_NO_BUFFER,
+	MLXSW_REG_HPKT_CTRL_PACKET_USE_BUFFER,
+};
+
+/* reg_hpkt_ctrl
+ * Configure dedicated buffer resources for control packets.
+ * Ignored by SwitchX-2.
+ * 0 - Keep factory defaults.
+ * 1 - Do not use control buffer for this trap ID.
+ * 2 - Use control buffer for this trap ID.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, hpkt, ctrl, 0x04, 16, 2);
+
+static inline void mlxsw_reg_hpkt_pack(char *payload, u8 action, u16 trap_id,
+				       enum mlxsw_reg_htgt_trap_group trap_group,
+				       bool is_ctrl)
+{
+	MLXSW_REG_ZERO(hpkt, payload);
+	mlxsw_reg_hpkt_ack_set(payload, MLXSW_REG_HPKT_ACK_NOT_REQUIRED);
+	mlxsw_reg_hpkt_action_set(payload, action);
+	mlxsw_reg_hpkt_trap_group_set(payload, trap_group);
+	mlxsw_reg_hpkt_trap_id_set(payload, trap_id);
+	mlxsw_reg_hpkt_ctrl_set(payload, is_ctrl ?
+				MLXSW_REG_HPKT_CTRL_PACKET_USE_BUFFER :
+				MLXSW_REG_HPKT_CTRL_PACKET_NO_BUFFER);
+}
+
+/* RGCR - Router General Configuration Register
+ * --------------------------------------------
+ * The register is used for setting up the router configuration.
+ */
+#define MLXSW_REG_RGCR_ID 0x8001
+#define MLXSW_REG_RGCR_LEN 0x28
+
+MLXSW_REG_DEFINE(rgcr, MLXSW_REG_RGCR_ID, MLXSW_REG_RGCR_LEN);
+
+/* reg_rgcr_ipv4_en
+ * IPv4 router enable.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, rgcr, ipv4_en, 0x00, 31, 1);
+
+/* reg_rgcr_ipv6_en
+ * IPv6 router enable.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, rgcr, ipv6_en, 0x00, 30, 1);
+
+/* reg_rgcr_max_router_interfaces
+ * Defines the maximum number of active router interfaces for all virtual
+ * routers.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, rgcr, max_router_interfaces, 0x10, 0, 16);
+
+/* reg_rgcr_usp
+ * Update switch priority and packet color.
+ * 0 - Preserve the value of Switch Priority and packet color.
+ * 1 - Recalculate the value of Switch Priority and packet color.
+ * Access: RW
+ *
+ * Note: Not supported by SwitchX and SwitchX-2.
+ */
+MLXSW_ITEM32(reg, rgcr, usp, 0x18, 20, 1);
+
+/* reg_rgcr_pcp_rw
+ * Indicates how to handle the pcp_rewrite_en value:
+ * 0 - Preserve the value of pcp_rewrite_en.
+ * 2 - Disable PCP rewrite.
+ * 3 - Enable PCP rewrite.
+ * Access: RW
+ *
+ * Note: Not supported by SwitchX and SwitchX-2.
+ */
+MLXSW_ITEM32(reg, rgcr, pcp_rw, 0x18, 16, 2);
+
+/* reg_rgcr_activity_dis
+ * Activity disable:
+ * 0 - Activity will be set when an entry is hit (default).
+ * 1 - Activity will not be set when an entry is hit.
+ *
+ * Bit 0 - Disable activity bit in Router Algorithmic LPM Unicast Entry
+ * (RALUE).
+ * Bit 1 - Disable activity bit in Router Algorithmic LPM Unicast Host
+ * Entry (RAUHT).
+ * Bits 2:7 are reserved.
+ * Access: RW
+ *
+ * Note: Not supported by SwitchX, SwitchX-2 and Switch-IB.
+ */
+MLXSW_ITEM32(reg, rgcr, activity_dis, 0x20, 0, 8);
+
+static inline void mlxsw_reg_rgcr_pack(char *payload, bool ipv4_en,
+				       bool ipv6_en)
+{
+	MLXSW_REG_ZERO(rgcr, payload);
+	mlxsw_reg_rgcr_ipv4_en_set(payload, ipv4_en);
+	mlxsw_reg_rgcr_ipv6_en_set(payload, ipv6_en);
+}
+
+/* RITR - Router Interface Table Register
+ * --------------------------------------
+ * The register is used to configure the router interface table.
+ */
+#define MLXSW_REG_RITR_ID 0x8002
+#define MLXSW_REG_RITR_LEN 0x40
+
+MLXSW_REG_DEFINE(ritr, MLXSW_REG_RITR_ID, MLXSW_REG_RITR_LEN);
+
+/* reg_ritr_enable
+ * Enables routing on the router interface.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ritr, enable, 0x00, 31, 1);
+
+/* reg_ritr_ipv4
+ * IPv4 routing enable. Enables routing of IPv4 traffic on the router
+ * interface.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ritr, ipv4, 0x00, 29, 1);
+
+/* reg_ritr_ipv6
+ * IPv6 routing enable. Enables routing of IPv6 traffic on the router
+ * interface.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ritr, ipv6, 0x00, 28, 1);
+
+/* reg_ritr_ipv4_mc
+ * IPv4 multicast routing enable.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ritr, ipv4_mc, 0x00, 27, 1);
+
+/* reg_ritr_ipv6_mc
+ * IPv6 multicast routing enable.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ritr, ipv6_mc, 0x00, 26, 1);
+
+enum mlxsw_reg_ritr_if_type {
+	/* VLAN interface. */
+	MLXSW_REG_RITR_VLAN_IF,
+	/* FID interface. */
+	MLXSW_REG_RITR_FID_IF,
+	/* Sub-port interface. */
+	MLXSW_REG_RITR_SP_IF,
+	/* Loopback Interface. */
+	MLXSW_REG_RITR_LOOPBACK_IF,
+};
+
+/* reg_ritr_type
+ * Router interface type as per enum mlxsw_reg_ritr_if_type.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ritr, type, 0x00, 23, 3);
+
+enum {
+	MLXSW_REG_RITR_RIF_CREATE,
+	MLXSW_REG_RITR_RIF_DEL,
+};
+
+/* reg_ritr_op
+ * Opcode:
+ * 0 - Create or edit RIF.
+ * 1 - Delete RIF.
+ * Reserved for SwitchX-2. For Spectrum, editing of interface properties
+ * is not supported. An interface must be deleted and re-created in order
+ * to update properties.
+ * Access: WO
+ */
+MLXSW_ITEM32(reg, ritr, op, 0x00, 20, 2);
+
+/* reg_ritr_rif
+ * Router interface index. A pointer to the Router Interface Table.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, ritr, rif, 0x00, 0, 16);
+
+/* reg_ritr_ipv4_fe
+ * IPv4 Forwarding Enable.
+ * Enables routing of IPv4 traffic on the router interface. When disabled,
+ * forwarding is blocked but local traffic (traps and IP2ME) will be enabled.
+ * Not supported in SwitchX-2.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ritr, ipv4_fe, 0x04, 29, 1);
+
+/* reg_ritr_ipv6_fe
+ * IPv6 Forwarding Enable.
+ * Enables routing of IPv6 traffic on the router interface. When disabled,
+ * forwarding is blocked but local traffic (traps and IP2ME) will be enabled.
+ * Not supported in SwitchX-2.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ritr, ipv6_fe, 0x04, 28, 1);
+
+/* reg_ritr_ipv4_mc_fe
+ * IPv4 Multicast Forwarding Enable.
+ * When disabled, forwarding is blocked but local traffic (traps and IP to me)
+ * will be enabled.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ritr, ipv4_mc_fe, 0x04, 27, 1);
+
+/* reg_ritr_ipv6_mc_fe
+ * IPv6 Multicast Forwarding Enable.
+ * When disabled, forwarding is blocked but local traffic (traps and IP to me)
+ * will be enabled.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ritr, ipv6_mc_fe, 0x04, 26, 1);
+
+/* reg_ritr_lb_en
+ * Loop-back filter enable for unicast packets.
+ * If the flag is set then loop-back filter for unicast packets is
+ * implemented on the RIF. Multicast packets are always subject to
+ * loop-back filtering.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ritr, lb_en, 0x04, 24, 1);
+
+/* reg_ritr_virtual_router
+ * Virtual router ID associated with the router interface.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ritr, virtual_router, 0x04, 0, 16);
+
+/* reg_ritr_mtu
+ * Router interface MTU.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ritr, mtu, 0x34, 0, 16);
+
+/* reg_ritr_if_swid
+ * Switch partition ID.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ritr, if_swid, 0x08, 24, 8);
+
+/* reg_ritr_if_mac_profile_id
+ * MAC msb profile ID.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ritr, if_mac_profile_id, 0x10, 16, 4);
+
+/* reg_ritr_if_mac
+ * Router interface MAC address.
+ * In Spectrum, all MAC addresses must have the same 38 MSBits.
+ * Access: RW
+ */
+MLXSW_ITEM_BUF(reg, ritr, if_mac, 0x12, 6);
+
+/* reg_ritr_if_vrrp_id_ipv6
+ * VRRP ID for IPv6
+ * Note: Reserved for RIF types other than VLAN, FID and Sub-port.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ritr, if_vrrp_id_ipv6, 0x1C, 8, 8);
+
+/* reg_ritr_if_vrrp_id_ipv4
+ * VRRP ID for IPv4
+ * Note: Reserved for RIF types other than VLAN, FID and Sub-port.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ritr, if_vrrp_id_ipv4, 0x1C, 0, 8);
+
+/* VLAN Interface */
+
+/* reg_ritr_vlan_if_vlan_id
+ * VLAN ID.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ritr, vlan_if_vlan_id, 0x08, 0, 12);
+
+/* reg_ritr_vlan_if_efid
+ * Egress FID.
+ * Used to connect the RIF to a bridge.
+ * Access: RW
+ *
+ * Note: Reserved when legacy bridge model is used and on Spectrum-1.
+ */
+MLXSW_ITEM32(reg, ritr, vlan_if_efid, 0x0C, 0, 16);
+
+/* FID Interface */
+
+/* reg_ritr_fid_if_fid
+ * Filtering ID. Used to connect a bridge to the router.
+ * When legacy bridge model is used, only FIDs from the vFID range are
+ * supported. When unified bridge model is used, this is the egress FID for
+ * router to bridge.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ritr, fid_if_fid, 0x08, 0, 16);
+
+/* Sub-port Interface */
+
+/* reg_ritr_sp_if_lag
+ * LAG indication. When this bit is set the system_port field holds the
+ * LAG identifier.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ritr, sp_if_lag, 0x08, 24, 1);
+
+/* reg_ritr_sp_system_port
+ * Port unique indentifier. When lag bit is set, this field holds the
+ * lag_id in bits 0:9.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ritr, sp_if_system_port, 0x08, 0, 16);
+
+/* reg_ritr_sp_if_efid
+ * Egress filtering ID.
+ * Used to connect the eRIF to a bridge if eRIF-ACL has modified the DMAC or
+ * the VID.
+ * Access: RW
+ *
+ * Note: Reserved when legacy bridge model is used.
+ */
+MLXSW_ITEM32(reg, ritr, sp_if_efid, 0x0C, 0, 16);
+
+/* reg_ritr_sp_if_vid
+ * VLAN ID.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ritr, sp_if_vid, 0x18, 0, 12);
+
+/* Loopback Interface */
+
+enum mlxsw_reg_ritr_loopback_protocol {
+	/* IPinIP IPv4 underlay Unicast */
+	MLXSW_REG_RITR_LOOPBACK_PROTOCOL_IPIP_IPV4,
+	/* IPinIP IPv6 underlay Unicast */
+	MLXSW_REG_RITR_LOOPBACK_PROTOCOL_IPIP_IPV6,
+	/* IPinIP generic - used for Spectrum-2 underlay RIF */
+	MLXSW_REG_RITR_LOOPBACK_GENERIC,
+};
+
+/* reg_ritr_loopback_protocol
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ritr, loopback_protocol, 0x08, 28, 4);
+
+enum mlxsw_reg_ritr_loopback_ipip_type {
+	/* Tunnel is IPinIP. */
+	MLXSW_REG_RITR_LOOPBACK_IPIP_TYPE_IP_IN_IP,
+	/* Tunnel is GRE, no key. */
+	MLXSW_REG_RITR_LOOPBACK_IPIP_TYPE_IP_IN_GRE_IN_IP,
+	/* Tunnel is GRE, with a key. */
+	MLXSW_REG_RITR_LOOPBACK_IPIP_TYPE_IP_IN_GRE_KEY_IN_IP,
+};
+
+/* reg_ritr_loopback_ipip_type
+ * Encapsulation type.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ritr, loopback_ipip_type, 0x10, 24, 4);
+
+enum mlxsw_reg_ritr_loopback_ipip_options {
+	/* The key is defined by gre_key. */
+	MLXSW_REG_RITR_LOOPBACK_IPIP_OPTIONS_GRE_KEY_PRESET,
+};
+
+/* reg_ritr_loopback_ipip_options
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ritr, loopback_ipip_options, 0x10, 20, 4);
+
+/* reg_ritr_loopback_ipip_uvr
+ * Underlay Virtual Router ID.
+ * Range is 0..cap_max_virtual_routers-1.
+ * Reserved for Spectrum-2.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ritr, loopback_ipip_uvr, 0x10, 0, 16);
+
+/* reg_ritr_loopback_ipip_underlay_rif
+ * Underlay ingress router interface.
+ * Reserved for Spectrum.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ritr, loopback_ipip_underlay_rif, 0x14, 0, 16);
+
+/* reg_ritr_loopback_ipip_usip*
+ * Encapsulation Underlay source IP.
+ * Access: RW
+ */
+MLXSW_ITEM_BUF(reg, ritr, loopback_ipip_usip6, 0x18, 16);
+MLXSW_ITEM32(reg, ritr, loopback_ipip_usip4, 0x24, 0, 32);
+
+/* reg_ritr_loopback_ipip_gre_key
+ * GRE Key.
+ * Reserved when ipip_type is not IP_IN_GRE_KEY_IN_IP.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ritr, loopback_ipip_gre_key, 0x28, 0, 32);
+
+/* Shared between ingress/egress */
+enum mlxsw_reg_ritr_counter_set_type {
+	/* No Count. */
+	MLXSW_REG_RITR_COUNTER_SET_TYPE_NO_COUNT = 0x0,
+	/* Basic. Used for router interfaces, counting the following:
+	 *	- Error and Discard counters.
+	 *	- Unicast, Multicast and Broadcast counters. Sharing the
+	 *	  same set of counters for the different type of traffic
+	 *	  (IPv4, IPv6 and mpls).
+	 */
+	MLXSW_REG_RITR_COUNTER_SET_TYPE_BASIC = 0x9,
+};
+
+/* reg_ritr_ingress_counter_index
+ * Counter Index for flow counter.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ritr, ingress_counter_index, 0x38, 0, 24);
+
+/* reg_ritr_ingress_counter_set_type
+ * Igress Counter Set Type for router interface counter.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ritr, ingress_counter_set_type, 0x38, 24, 8);
+
+/* reg_ritr_egress_counter_index
+ * Counter Index for flow counter.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ritr, egress_counter_index, 0x3C, 0, 24);
+
+/* reg_ritr_egress_counter_set_type
+ * Egress Counter Set Type for router interface counter.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ritr, egress_counter_set_type, 0x3C, 24, 8);
+
+static inline void mlxsw_reg_ritr_counter_pack(char *payload, u32 index,
+					       bool enable, bool egress)
+{
+	enum mlxsw_reg_ritr_counter_set_type set_type;
+
+	if (enable)
+		set_type = MLXSW_REG_RITR_COUNTER_SET_TYPE_BASIC;
+	else
+		set_type = MLXSW_REG_RITR_COUNTER_SET_TYPE_NO_COUNT;
+
+	if (egress) {
+		mlxsw_reg_ritr_egress_counter_set_type_set(payload, set_type);
+		mlxsw_reg_ritr_egress_counter_index_set(payload, index);
+	} else {
+		mlxsw_reg_ritr_ingress_counter_set_type_set(payload, set_type);
+		mlxsw_reg_ritr_ingress_counter_index_set(payload, index);
+	}
+}
+
+static inline void mlxsw_reg_ritr_rif_pack(char *payload, u16 rif)
+{
+	MLXSW_REG_ZERO(ritr, payload);
+	mlxsw_reg_ritr_rif_set(payload, rif);
+}
+
+static inline void mlxsw_reg_ritr_sp_if_pack(char *payload, bool lag,
+					     u16 system_port, u16 efid, u16 vid)
+{
+	mlxsw_reg_ritr_sp_if_lag_set(payload, lag);
+	mlxsw_reg_ritr_sp_if_system_port_set(payload, system_port);
+	mlxsw_reg_ritr_sp_if_efid_set(payload, efid);
+	mlxsw_reg_ritr_sp_if_vid_set(payload, vid);
+}
+
+static inline void mlxsw_reg_ritr_pack(char *payload, bool enable,
+				       enum mlxsw_reg_ritr_if_type type,
+				       u16 rif, u16 vr_id, u16 mtu)
+{
+	bool op = enable ? MLXSW_REG_RITR_RIF_CREATE : MLXSW_REG_RITR_RIF_DEL;
+
+	MLXSW_REG_ZERO(ritr, payload);
+	mlxsw_reg_ritr_enable_set(payload, enable);
+	mlxsw_reg_ritr_ipv4_set(payload, 1);
+	mlxsw_reg_ritr_ipv6_set(payload, 1);
+	mlxsw_reg_ritr_ipv4_mc_set(payload, 1);
+	mlxsw_reg_ritr_ipv6_mc_set(payload, 1);
+	mlxsw_reg_ritr_type_set(payload, type);
+	mlxsw_reg_ritr_op_set(payload, op);
+	mlxsw_reg_ritr_rif_set(payload, rif);
+	mlxsw_reg_ritr_ipv4_fe_set(payload, 1);
+	mlxsw_reg_ritr_ipv6_fe_set(payload, 1);
+	mlxsw_reg_ritr_ipv4_mc_fe_set(payload, 1);
+	mlxsw_reg_ritr_ipv6_mc_fe_set(payload, 1);
+	mlxsw_reg_ritr_lb_en_set(payload, 1);
+	mlxsw_reg_ritr_virtual_router_set(payload, vr_id);
+	mlxsw_reg_ritr_mtu_set(payload, mtu);
+}
+
+static inline void mlxsw_reg_ritr_mac_pack(char *payload, const char *mac)
+{
+	mlxsw_reg_ritr_if_mac_memcpy_to(payload, mac);
+}
+
+static inline void
+mlxsw_reg_ritr_vlan_if_pack(char *payload, bool enable, u16 rif, u16 vr_id,
+			    u16 mtu, const char *mac, u8 mac_profile_id,
+			    u16 vlan_id, u16 efid)
+{
+	enum mlxsw_reg_ritr_if_type type = MLXSW_REG_RITR_VLAN_IF;
+
+	mlxsw_reg_ritr_pack(payload, enable, type, rif, vr_id, mtu);
+	mlxsw_reg_ritr_if_mac_memcpy_to(payload, mac);
+	mlxsw_reg_ritr_if_mac_profile_id_set(payload, mac_profile_id);
+	mlxsw_reg_ritr_vlan_if_vlan_id_set(payload, vlan_id);
+	mlxsw_reg_ritr_vlan_if_efid_set(payload, efid);
+}
+
+static inline void
+mlxsw_reg_ritr_loopback_ipip_common_pack(char *payload,
+			    enum mlxsw_reg_ritr_loopback_ipip_type ipip_type,
+			    enum mlxsw_reg_ritr_loopback_ipip_options options,
+			    u16 uvr_id, u16 underlay_rif, u32 gre_key)
+{
+	mlxsw_reg_ritr_loopback_ipip_type_set(payload, ipip_type);
+	mlxsw_reg_ritr_loopback_ipip_options_set(payload, options);
+	mlxsw_reg_ritr_loopback_ipip_uvr_set(payload, uvr_id);
+	mlxsw_reg_ritr_loopback_ipip_underlay_rif_set(payload, underlay_rif);
+	mlxsw_reg_ritr_loopback_ipip_gre_key_set(payload, gre_key);
+}
+
+static inline void
+mlxsw_reg_ritr_loopback_ipip4_pack(char *payload,
+			    enum mlxsw_reg_ritr_loopback_ipip_type ipip_type,
+			    enum mlxsw_reg_ritr_loopback_ipip_options options,
+			    u16 uvr_id, u16 underlay_rif, u32 usip, u32 gre_key)
+{
+	mlxsw_reg_ritr_loopback_protocol_set(payload,
+				    MLXSW_REG_RITR_LOOPBACK_PROTOCOL_IPIP_IPV4);
+	mlxsw_reg_ritr_loopback_ipip_common_pack(payload, ipip_type, options,
+						 uvr_id, underlay_rif, gre_key);
+	mlxsw_reg_ritr_loopback_ipip_usip4_set(payload, usip);
+}
+
+static inline void
+mlxsw_reg_ritr_loopback_ipip6_pack(char *payload,
+				   enum mlxsw_reg_ritr_loopback_ipip_type ipip_type,
+				   enum mlxsw_reg_ritr_loopback_ipip_options options,
+				   u16 uvr_id, u16 underlay_rif,
+				   const struct in6_addr *usip, u32 gre_key)
+{
+	enum mlxsw_reg_ritr_loopback_protocol protocol =
+		MLXSW_REG_RITR_LOOPBACK_PROTOCOL_IPIP_IPV6;
+
+	mlxsw_reg_ritr_loopback_protocol_set(payload, protocol);
+	mlxsw_reg_ritr_loopback_ipip_common_pack(payload, ipip_type, options,
+						 uvr_id, underlay_rif, gre_key);
+	mlxsw_reg_ritr_loopback_ipip_usip6_memcpy_to(payload,
+						     (const char *)usip);
+}
+
+/* RTAR - Router TCAM Allocation Register
+ * --------------------------------------
+ * This register is used for allocation of regions in the TCAM table.
+ */
+#define MLXSW_REG_RTAR_ID 0x8004
+#define MLXSW_REG_RTAR_LEN 0x20
+
+MLXSW_REG_DEFINE(rtar, MLXSW_REG_RTAR_ID, MLXSW_REG_RTAR_LEN);
+
+enum mlxsw_reg_rtar_op {
+	MLXSW_REG_RTAR_OP_ALLOCATE,
+	MLXSW_REG_RTAR_OP_RESIZE,
+	MLXSW_REG_RTAR_OP_DEALLOCATE,
+};
+
+/* reg_rtar_op
+ * Access: WO
+ */
+MLXSW_ITEM32(reg, rtar, op, 0x00, 28, 4);
+
+enum mlxsw_reg_rtar_key_type {
+	MLXSW_REG_RTAR_KEY_TYPE_IPV4_MULTICAST = 1,
+	MLXSW_REG_RTAR_KEY_TYPE_IPV6_MULTICAST = 3
+};
+
+/* reg_rtar_key_type
+ * TCAM key type for the region.
+ * Access: WO
+ */
+MLXSW_ITEM32(reg, rtar, key_type, 0x00, 0, 8);
+
+/* reg_rtar_region_size
+ * TCAM region size. When allocating/resizing this is the requested
+ * size, the response is the actual size.
+ * Note: Actual size may be larger than requested.
+ * Reserved for op = Deallocate
+ * Access: WO
+ */
+MLXSW_ITEM32(reg, rtar, region_size, 0x04, 0, 16);
+
+static inline void mlxsw_reg_rtar_pack(char *payload,
+				       enum mlxsw_reg_rtar_op op,
+				       enum mlxsw_reg_rtar_key_type key_type,
+				       u16 region_size)
+{
+	MLXSW_REG_ZERO(rtar, payload);
+	mlxsw_reg_rtar_op_set(payload, op);
+	mlxsw_reg_rtar_key_type_set(payload, key_type);
+	mlxsw_reg_rtar_region_size_set(payload, region_size);
+}
+
+/* RATR - Router Adjacency Table Register
+ * --------------------------------------
+ * The RATR register is used to configure the Router Adjacency (next-hop)
+ * Table.
+ */
+#define MLXSW_REG_RATR_ID 0x8008
+#define MLXSW_REG_RATR_LEN 0x2C
+
+MLXSW_REG_DEFINE(ratr, MLXSW_REG_RATR_ID, MLXSW_REG_RATR_LEN);
+
+enum mlxsw_reg_ratr_op {
+	/* Read */
+	MLXSW_REG_RATR_OP_QUERY_READ = 0,
+	/* Read and clear activity */
+	MLXSW_REG_RATR_OP_QUERY_READ_CLEAR = 2,
+	/* Write Adjacency entry */
+	MLXSW_REG_RATR_OP_WRITE_WRITE_ENTRY = 1,
+	/* Write Adjacency entry only if the activity is cleared.
+	 * The write may not succeed if the activity is set. There is not
+	 * direct feedback if the write has succeeded or not, however
+	 * the get will reveal the actual entry (SW can compare the get
+	 * response to the set command).
+	 */
+	MLXSW_REG_RATR_OP_WRITE_WRITE_ENTRY_ON_ACTIVITY = 3,
+};
+
+/* reg_ratr_op
+ * Note that Write operation may also be used for updating
+ * counter_set_type and counter_index. In this case all other
+ * fields must not be updated.
+ * Access: OP
+ */
+MLXSW_ITEM32(reg, ratr, op, 0x00, 28, 4);
+
+/* reg_ratr_v
+ * Valid bit. Indicates if the adjacency entry is valid.
+ * Note: the device may need some time before reusing an invalidated
+ * entry. During this time the entry can not be reused. It is
+ * recommended to use another entry before reusing an invalidated
+ * entry (e.g. software can put it at the end of the list for
+ * reusing). Trying to access an invalidated entry not yet cleared
+ * by the device results with failure indicating "Try Again" status.
+ * When valid is '0' then egress_router_interface,trap_action,
+ * adjacency_parameters and counters are reserved
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ratr, v, 0x00, 24, 1);
+
+/* reg_ratr_a
+ * Activity. Set for new entries. Set if a packet lookup has hit on
+ * the specific entry. To clear the a bit, use "clear activity".
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, ratr, a, 0x00, 16, 1);
+
+enum mlxsw_reg_ratr_type {
+	/* Ethernet */
+	MLXSW_REG_RATR_TYPE_ETHERNET,
+	/* IPoIB Unicast without GRH.
+	 * Reserved for Spectrum.
+	 */
+	MLXSW_REG_RATR_TYPE_IPOIB_UC,
+	/* IPoIB Unicast with GRH. Supported only in table 0 (Ethernet unicast
+	 * adjacency).
+	 * Reserved for Spectrum.
+	 */
+	MLXSW_REG_RATR_TYPE_IPOIB_UC_W_GRH,
+	/* IPoIB Multicast.
+	 * Reserved for Spectrum.
+	 */
+	MLXSW_REG_RATR_TYPE_IPOIB_MC,
+	/* MPLS.
+	 * Reserved for SwitchX/-2.
+	 */
+	MLXSW_REG_RATR_TYPE_MPLS,
+	/* IPinIP Encap.
+	 * Reserved for SwitchX/-2.
+	 */
+	MLXSW_REG_RATR_TYPE_IPIP,
+};
+
+/* reg_ratr_type
+ * Adjacency entry type.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ratr, type, 0x04, 28, 4);
+
+/* reg_ratr_adjacency_index_low
+ * Bits 15:0 of index into the adjacency table.
+ * For SwitchX and SwitchX-2, the adjacency table is linear and
+ * used for adjacency entries only.
+ * For Spectrum, the index is to the KVD linear.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, ratr, adjacency_index_low, 0x04, 0, 16);
+
+/* reg_ratr_egress_router_interface
+ * Range is 0 .. cap_max_router_interfaces - 1
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ratr, egress_router_interface, 0x08, 0, 16);
+
+enum mlxsw_reg_ratr_trap_action {
+	MLXSW_REG_RATR_TRAP_ACTION_NOP,
+	MLXSW_REG_RATR_TRAP_ACTION_TRAP,
+	MLXSW_REG_RATR_TRAP_ACTION_MIRROR_TO_CPU,
+	MLXSW_REG_RATR_TRAP_ACTION_MIRROR,
+	MLXSW_REG_RATR_TRAP_ACTION_DISCARD_ERRORS,
+};
+
+/* reg_ratr_trap_action
+ * see mlxsw_reg_ratr_trap_action
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ratr, trap_action, 0x0C, 28, 4);
+
+/* reg_ratr_adjacency_index_high
+ * Bits 23:16 of the adjacency_index.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, ratr, adjacency_index_high, 0x0C, 16, 8);
+
+enum mlxsw_reg_ratr_trap_id {
+	MLXSW_REG_RATR_TRAP_ID_RTR_EGRESS0,
+	MLXSW_REG_RATR_TRAP_ID_RTR_EGRESS1,
+};
+
+/* reg_ratr_trap_id
+ * Trap ID to be reported to CPU.
+ * Trap-ID is RTR_EGRESS0 or RTR_EGRESS1.
+ * For trap_action of NOP, MIRROR and DISCARD_ERROR
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ratr, trap_id, 0x0C, 0, 8);
+
+/* reg_ratr_eth_destination_mac
+ * MAC address of the destination next-hop.
+ * Access: RW
+ */
+MLXSW_ITEM_BUF(reg, ratr, eth_destination_mac, 0x12, 6);
+
+enum mlxsw_reg_ratr_ipip_type {
+	/* IPv4, address set by mlxsw_reg_ratr_ipip_ipv4_udip. */
+	MLXSW_REG_RATR_IPIP_TYPE_IPV4,
+	/* IPv6, address set by mlxsw_reg_ratr_ipip_ipv6_ptr. */
+	MLXSW_REG_RATR_IPIP_TYPE_IPV6,
+};
+
+/* reg_ratr_ipip_type
+ * Underlay destination ip type.
+ * Note: the type field must match the protocol of the router interface.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ratr, ipip_type, 0x10, 16, 4);
+
+/* reg_ratr_ipip_ipv4_udip
+ * Underlay ipv4 dip.
+ * Reserved when ipip_type is IPv6.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ratr, ipip_ipv4_udip, 0x18, 0, 32);
+
+/* reg_ratr_ipip_ipv6_ptr
+ * Pointer to IPv6 underlay destination ip address.
+ * For Spectrum: Pointer to KVD linear space.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ratr, ipip_ipv6_ptr, 0x1C, 0, 24);
+
+enum mlxsw_reg_flow_counter_set_type {
+	/* No count */
+	MLXSW_REG_FLOW_COUNTER_SET_TYPE_NO_COUNT = 0x00,
+	/* Count packets and bytes */
+	MLXSW_REG_FLOW_COUNTER_SET_TYPE_PACKETS_BYTES = 0x03,
+	/* Count only packets */
+	MLXSW_REG_FLOW_COUNTER_SET_TYPE_PACKETS = 0x05,
+};
+
+/* reg_ratr_counter_set_type
+ * Counter set type for flow counters
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ratr, counter_set_type, 0x28, 24, 8);
+
+/* reg_ratr_counter_index
+ * Counter index for flow counters
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ratr, counter_index, 0x28, 0, 24);
+
+static inline void
+mlxsw_reg_ratr_pack(char *payload,
+		    enum mlxsw_reg_ratr_op op, bool valid,
+		    enum mlxsw_reg_ratr_type type,
+		    u32 adjacency_index, u16 egress_rif)
+{
+	MLXSW_REG_ZERO(ratr, payload);
+	mlxsw_reg_ratr_op_set(payload, op);
+	mlxsw_reg_ratr_v_set(payload, valid);
+	mlxsw_reg_ratr_type_set(payload, type);
+	mlxsw_reg_ratr_adjacency_index_low_set(payload, adjacency_index);
+	mlxsw_reg_ratr_adjacency_index_high_set(payload, adjacency_index >> 16);
+	mlxsw_reg_ratr_egress_router_interface_set(payload, egress_rif);
+}
+
+static inline void mlxsw_reg_ratr_eth_entry_pack(char *payload,
+						 const char *dest_mac)
+{
+	mlxsw_reg_ratr_eth_destination_mac_memcpy_to(payload, dest_mac);
+}
+
+static inline void mlxsw_reg_ratr_ipip4_entry_pack(char *payload, u32 ipv4_udip)
+{
+	mlxsw_reg_ratr_ipip_type_set(payload, MLXSW_REG_RATR_IPIP_TYPE_IPV4);
+	mlxsw_reg_ratr_ipip_ipv4_udip_set(payload, ipv4_udip);
+}
+
+static inline void mlxsw_reg_ratr_ipip6_entry_pack(char *payload, u32 ipv6_ptr)
+{
+	mlxsw_reg_ratr_ipip_type_set(payload, MLXSW_REG_RATR_IPIP_TYPE_IPV6);
+	mlxsw_reg_ratr_ipip_ipv6_ptr_set(payload, ipv6_ptr);
+}
+
+static inline void mlxsw_reg_ratr_counter_pack(char *payload, u64 counter_index,
+					       bool counter_enable)
+{
+	enum mlxsw_reg_flow_counter_set_type set_type;
+
+	if (counter_enable)
+		set_type = MLXSW_REG_FLOW_COUNTER_SET_TYPE_PACKETS_BYTES;
+	else
+		set_type = MLXSW_REG_FLOW_COUNTER_SET_TYPE_NO_COUNT;
+
+	mlxsw_reg_ratr_counter_index_set(payload, counter_index);
+	mlxsw_reg_ratr_counter_set_type_set(payload, set_type);
+}
+
+/* RDPM - Router DSCP to Priority Mapping
+ * --------------------------------------
+ * Controls the mapping from DSCP field to switch priority on routed packets
+ */
+#define MLXSW_REG_RDPM_ID 0x8009
+#define MLXSW_REG_RDPM_BASE_LEN 0x00
+#define MLXSW_REG_RDPM_DSCP_ENTRY_REC_LEN 0x01
+#define MLXSW_REG_RDPM_DSCP_ENTRY_REC_MAX_COUNT 64
+#define MLXSW_REG_RDPM_LEN 0x40
+#define MLXSW_REG_RDPM_LAST_ENTRY (MLXSW_REG_RDPM_BASE_LEN + \
+				   MLXSW_REG_RDPM_LEN - \
+				   MLXSW_REG_RDPM_DSCP_ENTRY_REC_LEN)
+
+MLXSW_REG_DEFINE(rdpm, MLXSW_REG_RDPM_ID, MLXSW_REG_RDPM_LEN);
+
+/* reg_dscp_entry_e
+ * Enable update of the specific entry
+ * Access: Index
+ */
+MLXSW_ITEM8_INDEXED(reg, rdpm, dscp_entry_e, MLXSW_REG_RDPM_LAST_ENTRY, 7, 1,
+		    -MLXSW_REG_RDPM_DSCP_ENTRY_REC_LEN, 0x00, false);
+
+/* reg_dscp_entry_prio
+ * Switch Priority
+ * Access: RW
+ */
+MLXSW_ITEM8_INDEXED(reg, rdpm, dscp_entry_prio, MLXSW_REG_RDPM_LAST_ENTRY, 0, 4,
+		    -MLXSW_REG_RDPM_DSCP_ENTRY_REC_LEN, 0x00, false);
+
+static inline void mlxsw_reg_rdpm_pack(char *payload, unsigned short index,
+				       u8 prio)
+{
+	mlxsw_reg_rdpm_dscp_entry_e_set(payload, index, 1);
+	mlxsw_reg_rdpm_dscp_entry_prio_set(payload, index, prio);
+}
+
+/* RICNT - Router Interface Counter Register
+ * -----------------------------------------
+ * The RICNT register retrieves per port performance counters
+ */
+#define MLXSW_REG_RICNT_ID 0x800B
+#define MLXSW_REG_RICNT_LEN 0x100
+
+MLXSW_REG_DEFINE(ricnt, MLXSW_REG_RICNT_ID, MLXSW_REG_RICNT_LEN);
+
+/* reg_ricnt_counter_index
+ * Counter index
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ricnt, counter_index, 0x04, 0, 24);
+
+enum mlxsw_reg_ricnt_counter_set_type {
+	/* No Count. */
+	MLXSW_REG_RICNT_COUNTER_SET_TYPE_NO_COUNT = 0x00,
+	/* Basic. Used for router interfaces, counting the following:
+	 *	- Error and Discard counters.
+	 *	- Unicast, Multicast and Broadcast counters. Sharing the
+	 *	  same set of counters for the different type of traffic
+	 *	  (IPv4, IPv6 and mpls).
+	 */
+	MLXSW_REG_RICNT_COUNTER_SET_TYPE_BASIC = 0x09,
+};
+
+/* reg_ricnt_counter_set_type
+ * Counter Set Type for router interface counter
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ricnt, counter_set_type, 0x04, 24, 8);
+
+enum mlxsw_reg_ricnt_opcode {
+	/* Nop. Supported only for read access*/
+	MLXSW_REG_RICNT_OPCODE_NOP = 0x00,
+	/* Clear. Setting the clr bit will reset the counter value for
+	 * all counters of the specified Router Interface.
+	 */
+	MLXSW_REG_RICNT_OPCODE_CLEAR = 0x08,
+};
+
+/* reg_ricnt_opcode
+ * Opcode
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ricnt, op, 0x00, 28, 4);
+
+/* reg_ricnt_good_unicast_packets
+ * good unicast packets.
+ * Access: RW
+ */
+MLXSW_ITEM64(reg, ricnt, good_unicast_packets, 0x08, 0, 64);
+
+/* reg_ricnt_good_multicast_packets
+ * good multicast packets.
+ * Access: RW
+ */
+MLXSW_ITEM64(reg, ricnt, good_multicast_packets, 0x10, 0, 64);
+
+/* reg_ricnt_good_broadcast_packets
+ * good broadcast packets
+ * Access: RW
+ */
+MLXSW_ITEM64(reg, ricnt, good_broadcast_packets, 0x18, 0, 64);
+
+/* reg_ricnt_good_unicast_bytes
+ * A count of L3 data and padding octets not including L2 headers
+ * for good unicast frames.
+ * Access: RW
+ */
+MLXSW_ITEM64(reg, ricnt, good_unicast_bytes, 0x20, 0, 64);
+
+/* reg_ricnt_good_multicast_bytes
+ * A count of L3 data and padding octets not including L2 headers
+ * for good multicast frames.
+ * Access: RW
+ */
+MLXSW_ITEM64(reg, ricnt, good_multicast_bytes, 0x28, 0, 64);
+
+/* reg_ritr_good_broadcast_bytes
+ * A count of L3 data and padding octets not including L2 headers
+ * for good broadcast frames.
+ * Access: RW
+ */
+MLXSW_ITEM64(reg, ricnt, good_broadcast_bytes, 0x30, 0, 64);
+
+/* reg_ricnt_error_packets
+ * A count of errored frames that do not pass the router checks.
+ * Access: RW
+ */
+MLXSW_ITEM64(reg, ricnt, error_packets, 0x38, 0, 64);
+
+/* reg_ricnt_discrad_packets
+ * A count of non-errored frames that do not pass the router checks.
+ * Access: RW
+ */
+MLXSW_ITEM64(reg, ricnt, discard_packets, 0x40, 0, 64);
+
+/* reg_ricnt_error_bytes
+ * A count of L3 data and padding octets not including L2 headers
+ * for errored frames.
+ * Access: RW
+ */
+MLXSW_ITEM64(reg, ricnt, error_bytes, 0x48, 0, 64);
+
+/* reg_ricnt_discard_bytes
+ * A count of L3 data and padding octets not including L2 headers
+ * for non-errored frames that do not pass the router checks.
+ * Access: RW
+ */
+MLXSW_ITEM64(reg, ricnt, discard_bytes, 0x50, 0, 64);
+
+static inline void mlxsw_reg_ricnt_pack(char *payload, u32 index,
+					enum mlxsw_reg_ricnt_opcode op)
+{
+	MLXSW_REG_ZERO(ricnt, payload);
+	mlxsw_reg_ricnt_op_set(payload, op);
+	mlxsw_reg_ricnt_counter_index_set(payload, index);
+	mlxsw_reg_ricnt_counter_set_type_set(payload,
+					     MLXSW_REG_RICNT_COUNTER_SET_TYPE_BASIC);
+}
+
+/* RRCR - Router Rules Copy Register Layout
+ * ----------------------------------------
+ * This register is used for moving and copying route entry rules.
+ */
+#define MLXSW_REG_RRCR_ID 0x800F
+#define MLXSW_REG_RRCR_LEN 0x24
+
+MLXSW_REG_DEFINE(rrcr, MLXSW_REG_RRCR_ID, MLXSW_REG_RRCR_LEN);
+
+enum mlxsw_reg_rrcr_op {
+	/* Move rules */
+	MLXSW_REG_RRCR_OP_MOVE,
+	/* Copy rules */
+	MLXSW_REG_RRCR_OP_COPY,
+};
+
+/* reg_rrcr_op
+ * Access: WO
+ */
+MLXSW_ITEM32(reg, rrcr, op, 0x00, 28, 4);
+
+/* reg_rrcr_offset
+ * Offset within the region from which to copy/move.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, rrcr, offset, 0x00, 0, 16);
+
+/* reg_rrcr_size
+ * The number of rules to copy/move.
+ * Access: WO
+ */
+MLXSW_ITEM32(reg, rrcr, size, 0x04, 0, 16);
+
+/* reg_rrcr_table_id
+ * Identifier of the table on which to perform the operation. Encoding is the
+ * same as in RTAR.key_type
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, rrcr, table_id, 0x10, 0, 4);
+
+/* reg_rrcr_dest_offset
+ * Offset within the region to which to copy/move
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, rrcr, dest_offset, 0x20, 0, 16);
+
+static inline void mlxsw_reg_rrcr_pack(char *payload, enum mlxsw_reg_rrcr_op op,
+				       u16 offset, u16 size,
+				       enum mlxsw_reg_rtar_key_type table_id,
+				       u16 dest_offset)
+{
+	MLXSW_REG_ZERO(rrcr, payload);
+	mlxsw_reg_rrcr_op_set(payload, op);
+	mlxsw_reg_rrcr_offset_set(payload, offset);
+	mlxsw_reg_rrcr_size_set(payload, size);
+	mlxsw_reg_rrcr_table_id_set(payload, table_id);
+	mlxsw_reg_rrcr_dest_offset_set(payload, dest_offset);
+}
+
+/* RALTA - Router Algorithmic LPM Tree Allocation Register
+ * -------------------------------------------------------
+ * RALTA is used to allocate the LPM trees of the SHSPM method.
+ */
+#define MLXSW_REG_RALTA_ID 0x8010
+#define MLXSW_REG_RALTA_LEN 0x04
+
+MLXSW_REG_DEFINE(ralta, MLXSW_REG_RALTA_ID, MLXSW_REG_RALTA_LEN);
+
+/* reg_ralta_op
+ * opcode (valid for Write, must be 0 on Read)
+ * 0 - allocate a tree
+ * 1 - deallocate a tree
+ * Access: OP
+ */
+MLXSW_ITEM32(reg, ralta, op, 0x00, 28, 2);
+
+enum mlxsw_reg_ralxx_protocol {
+	MLXSW_REG_RALXX_PROTOCOL_IPV4,
+	MLXSW_REG_RALXX_PROTOCOL_IPV6,
+};
+
+/* reg_ralta_protocol
+ * Protocol.
+ * Deallocation opcode: Reserved.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ralta, protocol, 0x00, 24, 4);
+
+/* reg_ralta_tree_id
+ * An identifier (numbered from 1..cap_shspm_max_trees-1) representing
+ * the tree identifier (managed by software).
+ * Note that tree_id 0 is allocated for a default-route tree.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, ralta, tree_id, 0x00, 0, 8);
+
+static inline void mlxsw_reg_ralta_pack(char *payload, bool alloc,
+					enum mlxsw_reg_ralxx_protocol protocol,
+					u8 tree_id)
+{
+	MLXSW_REG_ZERO(ralta, payload);
+	mlxsw_reg_ralta_op_set(payload, !alloc);
+	mlxsw_reg_ralta_protocol_set(payload, protocol);
+	mlxsw_reg_ralta_tree_id_set(payload, tree_id);
+}
+
+/* RALST - Router Algorithmic LPM Structure Tree Register
+ * ------------------------------------------------------
+ * RALST is used to set and query the structure of an LPM tree.
+ * The structure of the tree must be sorted as a sorted binary tree, while
+ * each node is a bin that is tagged as the length of the prefixes the lookup
+ * will refer to. Therefore, bin X refers to a set of entries with prefixes
+ * of X bits to match with the destination address. The bin 0 indicates
+ * the default action, when there is no match of any prefix.
+ */
+#define MLXSW_REG_RALST_ID 0x8011
+#define MLXSW_REG_RALST_LEN 0x104
+
+MLXSW_REG_DEFINE(ralst, MLXSW_REG_RALST_ID, MLXSW_REG_RALST_LEN);
+
+/* reg_ralst_root_bin
+ * The bin number of the root bin.
+ * 0<root_bin=<(length of IP address)
+ * For a default-route tree configure 0xff
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ralst, root_bin, 0x00, 16, 8);
+
+/* reg_ralst_tree_id
+ * Tree identifier numbered from 1..(cap_shspm_max_trees-1).
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, ralst, tree_id, 0x00, 0, 8);
+
+#define MLXSW_REG_RALST_BIN_NO_CHILD 0xff
+#define MLXSW_REG_RALST_BIN_OFFSET 0x04
+#define MLXSW_REG_RALST_BIN_COUNT 128
+
+/* reg_ralst_left_child_bin
+ * Holding the children of the bin according to the stored tree's structure.
+ * For trees composed of less than 4 blocks, the bins in excess are reserved.
+ * Note that tree_id 0 is allocated for a default-route tree, bins are 0xff
+ * Access: RW
+ */
+MLXSW_ITEM16_INDEXED(reg, ralst, left_child_bin, 0x04, 8, 8, 0x02, 0x00, false);
+
+/* reg_ralst_right_child_bin
+ * Holding the children of the bin according to the stored tree's structure.
+ * For trees composed of less than 4 blocks, the bins in excess are reserved.
+ * Note that tree_id 0 is allocated for a default-route tree, bins are 0xff
+ * Access: RW
+ */
+MLXSW_ITEM16_INDEXED(reg, ralst, right_child_bin, 0x04, 0, 8, 0x02, 0x00,
+		     false);
+
+static inline void mlxsw_reg_ralst_pack(char *payload, u8 root_bin, u8 tree_id)
+{
+	MLXSW_REG_ZERO(ralst, payload);
+
+	/* Initialize all bins to have no left or right child */
+	memset(payload + MLXSW_REG_RALST_BIN_OFFSET,
+	       MLXSW_REG_RALST_BIN_NO_CHILD, MLXSW_REG_RALST_BIN_COUNT * 2);
+
+	mlxsw_reg_ralst_root_bin_set(payload, root_bin);
+	mlxsw_reg_ralst_tree_id_set(payload, tree_id);
+}
+
+static inline void mlxsw_reg_ralst_bin_pack(char *payload, u8 bin_number,
+					    u8 left_child_bin,
+					    u8 right_child_bin)
+{
+	int bin_index = bin_number - 1;
+
+	mlxsw_reg_ralst_left_child_bin_set(payload, bin_index, left_child_bin);
+	mlxsw_reg_ralst_right_child_bin_set(payload, bin_index,
+					    right_child_bin);
+}
+
+/* RALTB - Router Algorithmic LPM Tree Binding Register
+ * ----------------------------------------------------
+ * RALTB is used to bind virtual router and protocol to an allocated LPM tree.
+ */
+#define MLXSW_REG_RALTB_ID 0x8012
+#define MLXSW_REG_RALTB_LEN 0x04
+
+MLXSW_REG_DEFINE(raltb, MLXSW_REG_RALTB_ID, MLXSW_REG_RALTB_LEN);
+
+/* reg_raltb_virtual_router
+ * Virtual Router ID
+ * Range is 0..cap_max_virtual_routers-1
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, raltb, virtual_router, 0x00, 16, 16);
+
+/* reg_raltb_protocol
+ * Protocol.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, raltb, protocol, 0x00, 12, 4);
+
+/* reg_raltb_tree_id
+ * Tree to be used for the {virtual_router, protocol}
+ * Tree identifier numbered from 1..(cap_shspm_max_trees-1).
+ * By default, all Unicast IPv4 and IPv6 are bound to tree_id 0.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, raltb, tree_id, 0x00, 0, 8);
+
+static inline void mlxsw_reg_raltb_pack(char *payload, u16 virtual_router,
+					enum mlxsw_reg_ralxx_protocol protocol,
+					u8 tree_id)
+{
+	MLXSW_REG_ZERO(raltb, payload);
+	mlxsw_reg_raltb_virtual_router_set(payload, virtual_router);
+	mlxsw_reg_raltb_protocol_set(payload, protocol);
+	mlxsw_reg_raltb_tree_id_set(payload, tree_id);
+}
+
+/* RALUE - Router Algorithmic LPM Unicast Entry Register
+ * -----------------------------------------------------
+ * RALUE is used to configure and query LPM entries that serve
+ * the Unicast protocols.
+ */
+#define MLXSW_REG_RALUE_ID 0x8013
+#define MLXSW_REG_RALUE_LEN 0x38
+
+MLXSW_REG_DEFINE(ralue, MLXSW_REG_RALUE_ID, MLXSW_REG_RALUE_LEN);
+
+/* reg_ralue_protocol
+ * Protocol.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, ralue, protocol, 0x00, 24, 4);
+
+enum mlxsw_reg_ralue_op {
+	/* Read operation. If entry doesn't exist, the operation fails. */
+	MLXSW_REG_RALUE_OP_QUERY_READ = 0,
+	/* Clear on read operation. Used to read entry and
+	 * clear Activity bit.
+	 */
+	MLXSW_REG_RALUE_OP_QUERY_CLEAR = 1,
+	/* Write operation. Used to write a new entry to the table. All RW
+	 * fields are written for new entry. Activity bit is set
+	 * for new entries.
+	 */
+	MLXSW_REG_RALUE_OP_WRITE_WRITE = 0,
+	/* Update operation. Used to update an existing route entry and
+	 * only update the RW fields that are detailed in the field
+	 * op_u_mask. If entry doesn't exist, the operation fails.
+	 */
+	MLXSW_REG_RALUE_OP_WRITE_UPDATE = 1,
+	/* Clear activity. The Activity bit (the field a) is cleared
+	 * for the entry.
+	 */
+	MLXSW_REG_RALUE_OP_WRITE_CLEAR = 2,
+	/* Delete operation. Used to delete an existing entry. If entry
+	 * doesn't exist, the operation fails.
+	 */
+	MLXSW_REG_RALUE_OP_WRITE_DELETE = 3,
+};
+
+/* reg_ralue_op
+ * Operation.
+ * Access: OP
+ */
+MLXSW_ITEM32(reg, ralue, op, 0x00, 20, 3);
+
+/* reg_ralue_a
+ * Activity. Set for new entries. Set if a packet lookup has hit on the
+ * specific entry, only if the entry is a route. To clear the a bit, use
+ * "clear activity" op.
+ * Enabled by activity_dis in RGCR
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, ralue, a, 0x00, 16, 1);
+
+/* reg_ralue_virtual_router
+ * Virtual Router ID
+ * Range is 0..cap_max_virtual_routers-1
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, ralue, virtual_router, 0x04, 16, 16);
+
+#define MLXSW_REG_RALUE_OP_U_MASK_ENTRY_TYPE	BIT(0)
+#define MLXSW_REG_RALUE_OP_U_MASK_BMP_LEN	BIT(1)
+#define MLXSW_REG_RALUE_OP_U_MASK_ACTION	BIT(2)
+
+/* reg_ralue_op_u_mask
+ * opcode update mask.
+ * On read operation, this field is reserved.
+ * This field is valid for update opcode, otherwise - reserved.
+ * This field is a bitmask of the fields that should be updated.
+ * Access: WO
+ */
+MLXSW_ITEM32(reg, ralue, op_u_mask, 0x04, 8, 3);
+
+/* reg_ralue_prefix_len
+ * Number of bits in the prefix of the LPM route.
+ * Note that for IPv6 prefixes, if prefix_len>64 the entry consumes
+ * two entries in the physical HW table.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, ralue, prefix_len, 0x08, 0, 8);
+
+/* reg_ralue_dip*
+ * The prefix of the route or of the marker that the object of the LPM
+ * is compared with. The most significant bits of the dip are the prefix.
+ * The least significant bits must be '0' if the prefix_len is smaller
+ * than 128 for IPv6 or smaller than 32 for IPv4.
+ * IPv4 address uses bits dip[31:0] and bits dip[127:32] are reserved.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, ralue, dip4, 0x18, 0, 32);
+MLXSW_ITEM_BUF(reg, ralue, dip6, 0x0C, 16);
+
+enum mlxsw_reg_ralue_entry_type {
+	MLXSW_REG_RALUE_ENTRY_TYPE_MARKER_ENTRY = 1,
+	MLXSW_REG_RALUE_ENTRY_TYPE_ROUTE_ENTRY = 2,
+	MLXSW_REG_RALUE_ENTRY_TYPE_MARKER_AND_ROUTE_ENTRY = 3,
+};
+
+/* reg_ralue_entry_type
+ * Entry type.
+ * Note - for Marker entries, the action_type and action fields are reserved.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ralue, entry_type, 0x1C, 30, 2);
+
+/* reg_ralue_bmp_len
+ * The best match prefix length in the case that there is no match for
+ * longer prefixes.
+ * If (entry_type != MARKER_ENTRY), bmp_len must be equal to prefix_len
+ * Note for any update operation with entry_type modification this
+ * field must be set.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ralue, bmp_len, 0x1C, 16, 8);
+
+enum mlxsw_reg_ralue_action_type {
+	MLXSW_REG_RALUE_ACTION_TYPE_REMOTE,
+	MLXSW_REG_RALUE_ACTION_TYPE_LOCAL,
+	MLXSW_REG_RALUE_ACTION_TYPE_IP2ME,
+};
+
+/* reg_ralue_action_type
+ * Action Type
+ * Indicates how the IP address is connected.
+ * It can be connected to a local subnet through local_erif or can be
+ * on a remote subnet connected through a next-hop router,
+ * or transmitted to the CPU.
+ * Reserved when entry_type = MARKER_ENTRY
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ralue, action_type, 0x1C, 0, 2);
+
+enum mlxsw_reg_ralue_trap_action {
+	MLXSW_REG_RALUE_TRAP_ACTION_NOP,
+	MLXSW_REG_RALUE_TRAP_ACTION_TRAP,
+	MLXSW_REG_RALUE_TRAP_ACTION_MIRROR_TO_CPU,
+	MLXSW_REG_RALUE_TRAP_ACTION_MIRROR,
+	MLXSW_REG_RALUE_TRAP_ACTION_DISCARD_ERROR,
+};
+
+/* reg_ralue_trap_action
+ * Trap action.
+ * For IP2ME action, only NOP and MIRROR are possible.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ralue, trap_action, 0x20, 28, 4);
+
+/* reg_ralue_trap_id
+ * Trap ID to be reported to CPU.
+ * Trap ID is RTR_INGRESS0 or RTR_INGRESS1.
+ * For trap_action of NOP, MIRROR and DISCARD_ERROR, trap_id is reserved.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ralue, trap_id, 0x20, 0, 9);
+
+/* reg_ralue_adjacency_index
+ * Points to the first entry of the group-based ECMP.
+ * Only relevant in case of REMOTE action.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ralue, adjacency_index, 0x24, 0, 24);
+
+/* reg_ralue_ecmp_size
+ * Amount of sequential entries starting
+ * from the adjacency_index (the number of ECMPs).
+ * The valid range is 1-64, 512, 1024, 2048 and 4096.
+ * Reserved when trap_action is TRAP or DISCARD_ERROR.
+ * Only relevant in case of REMOTE action.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ralue, ecmp_size, 0x28, 0, 13);
+
+/* reg_ralue_local_erif
+ * Egress Router Interface.
+ * Only relevant in case of LOCAL action.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ralue, local_erif, 0x24, 0, 16);
+
+/* reg_ralue_ip2me_v
+ * Valid bit for the tunnel_ptr field.
+ * If valid = 0 then trap to CPU as IP2ME trap ID.
+ * If valid = 1 and the packet format allows NVE or IPinIP tunnel
+ * decapsulation then tunnel decapsulation is done.
+ * If valid = 1 and packet format does not allow NVE or IPinIP tunnel
+ * decapsulation then trap as IP2ME trap ID.
+ * Only relevant in case of IP2ME action.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ralue, ip2me_v, 0x24, 31, 1);
+
+/* reg_ralue_ip2me_tunnel_ptr
+ * Tunnel Pointer for NVE or IPinIP tunnel decapsulation.
+ * For Spectrum, pointer to KVD Linear.
+ * Only relevant in case of IP2ME action.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, ralue, ip2me_tunnel_ptr, 0x24, 0, 24);
+
+static inline void mlxsw_reg_ralue_pack(char *payload,
+					enum mlxsw_reg_ralxx_protocol protocol,
+					enum mlxsw_reg_ralue_op op,
+					u16 virtual_router, u8 prefix_len)
+{
+	MLXSW_REG_ZERO(ralue, payload);
+	mlxsw_reg_ralue_protocol_set(payload, protocol);
+	mlxsw_reg_ralue_op_set(payload, op);
+	mlxsw_reg_ralue_virtual_router_set(payload, virtual_router);
+	mlxsw_reg_ralue_prefix_len_set(payload, prefix_len);
+	mlxsw_reg_ralue_entry_type_set(payload,
+				       MLXSW_REG_RALUE_ENTRY_TYPE_ROUTE_ENTRY);
+	mlxsw_reg_ralue_bmp_len_set(payload, prefix_len);
+}
+
+static inline void mlxsw_reg_ralue_pack4(char *payload,
+					 enum mlxsw_reg_ralxx_protocol protocol,
+					 enum mlxsw_reg_ralue_op op,
+					 u16 virtual_router, u8 prefix_len,
+					 u32 dip)
+{
+	mlxsw_reg_ralue_pack(payload, protocol, op, virtual_router, prefix_len);
+	mlxsw_reg_ralue_dip4_set(payload, dip);
+}
+
+static inline void mlxsw_reg_ralue_pack6(char *payload,
+					 enum mlxsw_reg_ralxx_protocol protocol,
+					 enum mlxsw_reg_ralue_op op,
+					 u16 virtual_router, u8 prefix_len,
+					 const void *dip)
+{
+	mlxsw_reg_ralue_pack(payload, protocol, op, virtual_router, prefix_len);
+	mlxsw_reg_ralue_dip6_memcpy_to(payload, dip);
+}
+
+static inline void
+mlxsw_reg_ralue_act_remote_pack(char *payload,
+				enum mlxsw_reg_ralue_trap_action trap_action,
+				u16 trap_id, u32 adjacency_index, u16 ecmp_size)
+{
+	mlxsw_reg_ralue_action_type_set(payload,
+					MLXSW_REG_RALUE_ACTION_TYPE_REMOTE);
+	mlxsw_reg_ralue_trap_action_set(payload, trap_action);
+	mlxsw_reg_ralue_trap_id_set(payload, trap_id);
+	mlxsw_reg_ralue_adjacency_index_set(payload, adjacency_index);
+	mlxsw_reg_ralue_ecmp_size_set(payload, ecmp_size);
+}
+
+static inline void
+mlxsw_reg_ralue_act_local_pack(char *payload,
+			       enum mlxsw_reg_ralue_trap_action trap_action,
+			       u16 trap_id, u16 local_erif)
+{
+	mlxsw_reg_ralue_action_type_set(payload,
+					MLXSW_REG_RALUE_ACTION_TYPE_LOCAL);
+	mlxsw_reg_ralue_trap_action_set(payload, trap_action);
+	mlxsw_reg_ralue_trap_id_set(payload, trap_id);
+	mlxsw_reg_ralue_local_erif_set(payload, local_erif);
+}
+
+static inline void
+mlxsw_reg_ralue_act_ip2me_pack(char *payload)
+{
+	mlxsw_reg_ralue_action_type_set(payload,
+					MLXSW_REG_RALUE_ACTION_TYPE_IP2ME);
+}
+
+static inline void
+mlxsw_reg_ralue_act_ip2me_tun_pack(char *payload, u32 tunnel_ptr)
+{
+	mlxsw_reg_ralue_action_type_set(payload,
+					MLXSW_REG_RALUE_ACTION_TYPE_IP2ME);
+	mlxsw_reg_ralue_ip2me_v_set(payload, 1);
+	mlxsw_reg_ralue_ip2me_tunnel_ptr_set(payload, tunnel_ptr);
+}
+
+/* RAUHT - Router Algorithmic LPM Unicast Host Table Register
+ * ----------------------------------------------------------
+ * The RAUHT register is used to configure and query the Unicast Host table in
+ * devices that implement the Algorithmic LPM.
+ */
+#define MLXSW_REG_RAUHT_ID 0x8014
+#define MLXSW_REG_RAUHT_LEN 0x74
+
+MLXSW_REG_DEFINE(rauht, MLXSW_REG_RAUHT_ID, MLXSW_REG_RAUHT_LEN);
+
+enum mlxsw_reg_rauht_type {
+	MLXSW_REG_RAUHT_TYPE_IPV4,
+	MLXSW_REG_RAUHT_TYPE_IPV6,
+};
+
+/* reg_rauht_type
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, rauht, type, 0x00, 24, 2);
+
+enum mlxsw_reg_rauht_op {
+	MLXSW_REG_RAUHT_OP_QUERY_READ = 0,
+	/* Read operation */
+	MLXSW_REG_RAUHT_OP_QUERY_CLEAR_ON_READ = 1,
+	/* Clear on read operation. Used to read entry and clear
+	 * activity bit.
+	 */
+	MLXSW_REG_RAUHT_OP_WRITE_ADD = 0,
+	/* Add. Used to write a new entry to the table. All R/W fields are
+	 * relevant for new entry. Activity bit is set for new entries.
+	 */
+	MLXSW_REG_RAUHT_OP_WRITE_UPDATE = 1,
+	/* Update action. Used to update an existing route entry and
+	 * only update the following fields:
+	 * trap_action, trap_id, mac, counter_set_type, counter_index
+	 */
+	MLXSW_REG_RAUHT_OP_WRITE_CLEAR_ACTIVITY = 2,
+	/* Clear activity. A bit is cleared for the entry. */
+	MLXSW_REG_RAUHT_OP_WRITE_DELETE = 3,
+	/* Delete entry */
+	MLXSW_REG_RAUHT_OP_WRITE_DELETE_ALL = 4,
+	/* Delete all host entries on a RIF. In this command, dip
+	 * field is reserved.
+	 */
+};
+
+/* reg_rauht_op
+ * Access: OP
+ */
+MLXSW_ITEM32(reg, rauht, op, 0x00, 20, 3);
+
+/* reg_rauht_a
+ * Activity. Set for new entries. Set if a packet lookup has hit on
+ * the specific entry.
+ * To clear the a bit, use "clear activity" op.
+ * Enabled by activity_dis in RGCR
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, rauht, a, 0x00, 16, 1);
+
+/* reg_rauht_rif
+ * Router Interface
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, rauht, rif, 0x00, 0, 16);
+
+/* reg_rauht_dip*
+ * Destination address.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, rauht, dip4, 0x1C, 0x0, 32);
+MLXSW_ITEM_BUF(reg, rauht, dip6, 0x10, 16);
+
+enum mlxsw_reg_rauht_trap_action {
+	MLXSW_REG_RAUHT_TRAP_ACTION_NOP,
+	MLXSW_REG_RAUHT_TRAP_ACTION_TRAP,
+	MLXSW_REG_RAUHT_TRAP_ACTION_MIRROR_TO_CPU,
+	MLXSW_REG_RAUHT_TRAP_ACTION_MIRROR,
+	MLXSW_REG_RAUHT_TRAP_ACTION_DISCARD_ERRORS,
+};
+
+/* reg_rauht_trap_action
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, rauht, trap_action, 0x60, 28, 4);
+
+enum mlxsw_reg_rauht_trap_id {
+	MLXSW_REG_RAUHT_TRAP_ID_RTR_EGRESS0,
+	MLXSW_REG_RAUHT_TRAP_ID_RTR_EGRESS1,
+};
+
+/* reg_rauht_trap_id
+ * Trap ID to be reported to CPU.
+ * Trap-ID is RTR_EGRESS0 or RTR_EGRESS1.
+ * For trap_action of NOP, MIRROR and DISCARD_ERROR,
+ * trap_id is reserved.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, rauht, trap_id, 0x60, 0, 9);
+
+/* reg_rauht_counter_set_type
+ * Counter set type for flow counters
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, rauht, counter_set_type, 0x68, 24, 8);
+
+/* reg_rauht_counter_index
+ * Counter index for flow counters
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, rauht, counter_index, 0x68, 0, 24);
+
+/* reg_rauht_mac
+ * MAC address.
+ * Access: RW
+ */
+MLXSW_ITEM_BUF(reg, rauht, mac, 0x6E, 6);
+
+static inline void mlxsw_reg_rauht_pack(char *payload,
+					enum mlxsw_reg_rauht_op op, u16 rif,
+					const char *mac)
+{
+	MLXSW_REG_ZERO(rauht, payload);
+	mlxsw_reg_rauht_op_set(payload, op);
+	mlxsw_reg_rauht_rif_set(payload, rif);
+	mlxsw_reg_rauht_mac_memcpy_to(payload, mac);
+}
+
+static inline void mlxsw_reg_rauht_pack4(char *payload,
+					 enum mlxsw_reg_rauht_op op, u16 rif,
+					 const char *mac, u32 dip)
+{
+	mlxsw_reg_rauht_pack(payload, op, rif, mac);
+	mlxsw_reg_rauht_dip4_set(payload, dip);
+}
+
+static inline void mlxsw_reg_rauht_pack6(char *payload,
+					 enum mlxsw_reg_rauht_op op, u16 rif,
+					 const char *mac, const char *dip)
+{
+	mlxsw_reg_rauht_pack(payload, op, rif, mac);
+	mlxsw_reg_rauht_type_set(payload, MLXSW_REG_RAUHT_TYPE_IPV6);
+	mlxsw_reg_rauht_dip6_memcpy_to(payload, dip);
+}
+
+static inline void mlxsw_reg_rauht_pack_counter(char *payload,
+						u64 counter_index)
+{
+	mlxsw_reg_rauht_counter_index_set(payload, counter_index);
+	mlxsw_reg_rauht_counter_set_type_set(payload,
+					     MLXSW_REG_FLOW_COUNTER_SET_TYPE_PACKETS_BYTES);
+}
+
+/* RALEU - Router Algorithmic LPM ECMP Update Register
+ * ---------------------------------------------------
+ * The register enables updating the ECMP section in the action for multiple
+ * LPM Unicast entries in a single operation. The update is executed to
+ * all entries of a {virtual router, protocol} tuple using the same ECMP group.
+ */
+#define MLXSW_REG_RALEU_ID 0x8015
+#define MLXSW_REG_RALEU_LEN 0x28
+
+MLXSW_REG_DEFINE(raleu, MLXSW_REG_RALEU_ID, MLXSW_REG_RALEU_LEN);
+
+/* reg_raleu_protocol
+ * Protocol.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, raleu, protocol, 0x00, 24, 4);
+
+/* reg_raleu_virtual_router
+ * Virtual Router ID
+ * Range is 0..cap_max_virtual_routers-1
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, raleu, virtual_router, 0x00, 0, 16);
+
+/* reg_raleu_adjacency_index
+ * Adjacency Index used for matching on the existing entries.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, raleu, adjacency_index, 0x10, 0, 24);
+
+/* reg_raleu_ecmp_size
+ * ECMP Size used for matching on the existing entries.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, raleu, ecmp_size, 0x14, 0, 13);
+
+/* reg_raleu_new_adjacency_index
+ * New Adjacency Index.
+ * Access: WO
+ */
+MLXSW_ITEM32(reg, raleu, new_adjacency_index, 0x20, 0, 24);
+
+/* reg_raleu_new_ecmp_size
+ * New ECMP Size.
+ * Access: WO
+ */
+MLXSW_ITEM32(reg, raleu, new_ecmp_size, 0x24, 0, 13);
+
+static inline void mlxsw_reg_raleu_pack(char *payload,
+					enum mlxsw_reg_ralxx_protocol protocol,
+					u16 virtual_router,
+					u32 adjacency_index, u16 ecmp_size,
+					u32 new_adjacency_index,
+					u16 new_ecmp_size)
+{
+	MLXSW_REG_ZERO(raleu, payload);
+	mlxsw_reg_raleu_protocol_set(payload, protocol);
+	mlxsw_reg_raleu_virtual_router_set(payload, virtual_router);
+	mlxsw_reg_raleu_adjacency_index_set(payload, adjacency_index);
+	mlxsw_reg_raleu_ecmp_size_set(payload, ecmp_size);
+	mlxsw_reg_raleu_new_adjacency_index_set(payload, new_adjacency_index);
+	mlxsw_reg_raleu_new_ecmp_size_set(payload, new_ecmp_size);
+}
+
+/* RAUHTD - Router Algorithmic LPM Unicast Host Table Dump Register
+ * ----------------------------------------------------------------
+ * The RAUHTD register allows dumping entries from the Router Unicast Host
+ * Table. For a given session an entry is dumped no more than one time. The
+ * first RAUHTD access after reset is a new session. A session ends when the
+ * num_rec response is smaller than num_rec request or for IPv4 when the
+ * num_entries is smaller than 4. The clear activity affect the current session
+ * or the last session if a new session has not started.
+ */
+#define MLXSW_REG_RAUHTD_ID 0x8018
+#define MLXSW_REG_RAUHTD_BASE_LEN 0x20
+#define MLXSW_REG_RAUHTD_REC_LEN 0x20
+#define MLXSW_REG_RAUHTD_REC_MAX_NUM 32
+#define MLXSW_REG_RAUHTD_LEN (MLXSW_REG_RAUHTD_BASE_LEN + \
+		MLXSW_REG_RAUHTD_REC_MAX_NUM * MLXSW_REG_RAUHTD_REC_LEN)
+#define MLXSW_REG_RAUHTD_IPV4_ENT_PER_REC 4
+
+MLXSW_REG_DEFINE(rauhtd, MLXSW_REG_RAUHTD_ID, MLXSW_REG_RAUHTD_LEN);
+
+#define MLXSW_REG_RAUHTD_FILTER_A BIT(0)
+#define MLXSW_REG_RAUHTD_FILTER_RIF BIT(3)
+
+/* reg_rauhtd_filter_fields
+ * if a bit is '0' then the relevant field is ignored and dump is done
+ * regardless of the field value
+ * Bit0 - filter by activity: entry_a
+ * Bit3 - filter by entry rip: entry_rif
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, rauhtd, filter_fields, 0x00, 0, 8);
+
+enum mlxsw_reg_rauhtd_op {
+	MLXSW_REG_RAUHTD_OP_DUMP,
+	MLXSW_REG_RAUHTD_OP_DUMP_AND_CLEAR,
+};
+
+/* reg_rauhtd_op
+ * Access: OP
+ */
+MLXSW_ITEM32(reg, rauhtd, op, 0x04, 24, 2);
+
+/* reg_rauhtd_num_rec
+ * At request: number of records requested
+ * At response: number of records dumped
+ * For IPv4, each record has 4 entries at request and up to 4 entries
+ * at response
+ * Range is 0..MLXSW_REG_RAUHTD_REC_MAX_NUM
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, rauhtd, num_rec, 0x04, 0, 8);
+
+/* reg_rauhtd_entry_a
+ * Dump only if activity has value of entry_a
+ * Reserved if filter_fields bit0 is '0'
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, rauhtd, entry_a, 0x08, 16, 1);
+
+enum mlxsw_reg_rauhtd_type {
+	MLXSW_REG_RAUHTD_TYPE_IPV4,
+	MLXSW_REG_RAUHTD_TYPE_IPV6,
+};
+
+/* reg_rauhtd_type
+ * Dump only if record type is:
+ * 0 - IPv4
+ * 1 - IPv6
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, rauhtd, type, 0x08, 0, 4);
+
+/* reg_rauhtd_entry_rif
+ * Dump only if RIF has value of entry_rif
+ * Reserved if filter_fields bit3 is '0'
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, rauhtd, entry_rif, 0x0C, 0, 16);
+
+static inline void mlxsw_reg_rauhtd_pack(char *payload,
+					 enum mlxsw_reg_rauhtd_type type)
+{
+	MLXSW_REG_ZERO(rauhtd, payload);
+	mlxsw_reg_rauhtd_filter_fields_set(payload, MLXSW_REG_RAUHTD_FILTER_A);
+	mlxsw_reg_rauhtd_op_set(payload, MLXSW_REG_RAUHTD_OP_DUMP_AND_CLEAR);
+	mlxsw_reg_rauhtd_num_rec_set(payload, MLXSW_REG_RAUHTD_REC_MAX_NUM);
+	mlxsw_reg_rauhtd_entry_a_set(payload, 1);
+	mlxsw_reg_rauhtd_type_set(payload, type);
+}
+
+/* reg_rauhtd_ipv4_rec_num_entries
+ * Number of valid entries in this record:
+ * 0 - 1 valid entry
+ * 1 - 2 valid entries
+ * 2 - 3 valid entries
+ * 3 - 4 valid entries
+ * Access: RO
+ */
+MLXSW_ITEM32_INDEXED(reg, rauhtd, ipv4_rec_num_entries,
+		     MLXSW_REG_RAUHTD_BASE_LEN, 28, 2,
+		     MLXSW_REG_RAUHTD_REC_LEN, 0x00, false);
+
+/* reg_rauhtd_rec_type
+ * Record type.
+ * 0 - IPv4
+ * 1 - IPv6
+ * Access: RO
+ */
+MLXSW_ITEM32_INDEXED(reg, rauhtd, rec_type, MLXSW_REG_RAUHTD_BASE_LEN, 24, 2,
+		     MLXSW_REG_RAUHTD_REC_LEN, 0x00, false);
+
+#define MLXSW_REG_RAUHTD_IPV4_ENT_LEN 0x8
+
+/* reg_rauhtd_ipv4_ent_a
+ * Activity. Set for new entries. Set if a packet lookup has hit on the
+ * specific entry.
+ * Access: RO
+ */
+MLXSW_ITEM32_INDEXED(reg, rauhtd, ipv4_ent_a, MLXSW_REG_RAUHTD_BASE_LEN, 16, 1,
+		     MLXSW_REG_RAUHTD_IPV4_ENT_LEN, 0x00, false);
+
+/* reg_rauhtd_ipv4_ent_rif
+ * Router interface.
+ * Access: RO
+ */
+MLXSW_ITEM32_INDEXED(reg, rauhtd, ipv4_ent_rif, MLXSW_REG_RAUHTD_BASE_LEN, 0,
+		     16, MLXSW_REG_RAUHTD_IPV4_ENT_LEN, 0x00, false);
+
+/* reg_rauhtd_ipv4_ent_dip
+ * Destination IPv4 address.
+ * Access: RO
+ */
+MLXSW_ITEM32_INDEXED(reg, rauhtd, ipv4_ent_dip, MLXSW_REG_RAUHTD_BASE_LEN, 0,
+		     32, MLXSW_REG_RAUHTD_IPV4_ENT_LEN, 0x04, false);
+
+#define MLXSW_REG_RAUHTD_IPV6_ENT_LEN 0x20
+
+/* reg_rauhtd_ipv6_ent_a
+ * Activity. Set for new entries. Set if a packet lookup has hit on the
+ * specific entry.
+ * Access: RO
+ */
+MLXSW_ITEM32_INDEXED(reg, rauhtd, ipv6_ent_a, MLXSW_REG_RAUHTD_BASE_LEN, 16, 1,
+		     MLXSW_REG_RAUHTD_IPV6_ENT_LEN, 0x00, false);
+
+/* reg_rauhtd_ipv6_ent_rif
+ * Router interface.
+ * Access: RO
+ */
+MLXSW_ITEM32_INDEXED(reg, rauhtd, ipv6_ent_rif, MLXSW_REG_RAUHTD_BASE_LEN, 0,
+		     16, MLXSW_REG_RAUHTD_IPV6_ENT_LEN, 0x00, false);
+
+/* reg_rauhtd_ipv6_ent_dip
+ * Destination IPv6 address.
+ * Access: RO
+ */
+MLXSW_ITEM_BUF_INDEXED(reg, rauhtd, ipv6_ent_dip, MLXSW_REG_RAUHTD_BASE_LEN,
+		       16, MLXSW_REG_RAUHTD_IPV6_ENT_LEN, 0x10);
+
+static inline void mlxsw_reg_rauhtd_ent_ipv4_unpack(char *payload,
+						    int ent_index, u16 *p_rif,
+						    u32 *p_dip)
+{
+	*p_rif = mlxsw_reg_rauhtd_ipv4_ent_rif_get(payload, ent_index);
+	*p_dip = mlxsw_reg_rauhtd_ipv4_ent_dip_get(payload, ent_index);
+}
+
+static inline void mlxsw_reg_rauhtd_ent_ipv6_unpack(char *payload,
+						    int rec_index, u16 *p_rif,
+						    char *p_dip)
+{
+	*p_rif = mlxsw_reg_rauhtd_ipv6_ent_rif_get(payload, rec_index);
+	mlxsw_reg_rauhtd_ipv6_ent_dip_memcpy_from(payload, rec_index, p_dip);
+}
+
+/* RTDP - Routing Tunnel Decap Properties Register
+ * -----------------------------------------------
+ * The RTDP register is used for configuring the tunnel decap properties of NVE
+ * and IPinIP.
+ */
+#define MLXSW_REG_RTDP_ID 0x8020
+#define MLXSW_REG_RTDP_LEN 0x44
+
+MLXSW_REG_DEFINE(rtdp, MLXSW_REG_RTDP_ID, MLXSW_REG_RTDP_LEN);
+
+enum mlxsw_reg_rtdp_type {
+	MLXSW_REG_RTDP_TYPE_NVE,
+	MLXSW_REG_RTDP_TYPE_IPIP,
+};
+
+/* reg_rtdp_type
+ * Type of the RTDP entry as per enum mlxsw_reg_rtdp_type.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, rtdp, type, 0x00, 28, 4);
+
+/* reg_rtdp_tunnel_index
+ * Index to the Decap entry.
+ * For Spectrum, Index to KVD Linear.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, rtdp, tunnel_index, 0x00, 0, 24);
+
+/* reg_rtdp_egress_router_interface
+ * Underlay egress router interface.
+ * Valid range is from 0 to cap_max_router_interfaces - 1
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, rtdp, egress_router_interface, 0x40, 0, 16);
+
+/* IPinIP */
+
+/* reg_rtdp_ipip_irif
+ * Ingress Router Interface for the overlay router
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, rtdp, ipip_irif, 0x04, 16, 16);
+
+enum mlxsw_reg_rtdp_ipip_sip_check {
+	/* No sip checks. */
+	MLXSW_REG_RTDP_IPIP_SIP_CHECK_NO,
+	/* Filter packet if underlay is not IPv4 or if underlay SIP does not
+	 * equal ipv4_usip.
+	 */
+	MLXSW_REG_RTDP_IPIP_SIP_CHECK_FILTER_IPV4,
+	/* Filter packet if underlay is not IPv6 or if underlay SIP does not
+	 * equal ipv6_usip.
+	 */
+	MLXSW_REG_RTDP_IPIP_SIP_CHECK_FILTER_IPV6 = 3,
+};
+
+/* reg_rtdp_ipip_sip_check
+ * SIP check to perform. If decapsulation failed due to these configurations
+ * then trap_id is IPIP_DECAP_ERROR.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, rtdp, ipip_sip_check, 0x04, 0, 3);
+
+/* If set, allow decapsulation of IPinIP (without GRE). */
+#define MLXSW_REG_RTDP_IPIP_TYPE_CHECK_ALLOW_IPIP	BIT(0)
+/* If set, allow decapsulation of IPinGREinIP without a key. */
+#define MLXSW_REG_RTDP_IPIP_TYPE_CHECK_ALLOW_GRE	BIT(1)
+/* If set, allow decapsulation of IPinGREinIP with a key. */
+#define MLXSW_REG_RTDP_IPIP_TYPE_CHECK_ALLOW_GRE_KEY	BIT(2)
+
+/* reg_rtdp_ipip_type_check
+ * Flags as per MLXSW_REG_RTDP_IPIP_TYPE_CHECK_*. If decapsulation failed due to
+ * these configurations then trap_id is IPIP_DECAP_ERROR.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, rtdp, ipip_type_check, 0x08, 24, 3);
+
+/* reg_rtdp_ipip_gre_key_check
+ * Whether GRE key should be checked. When check is enabled:
+ * - A packet received as IPinIP (without GRE) will always pass.
+ * - A packet received as IPinGREinIP without a key will not pass the check.
+ * - A packet received as IPinGREinIP with a key will pass the check only if the
+ *   key in the packet is equal to expected_gre_key.
+ * If decapsulation failed due to GRE key then trap_id is IPIP_DECAP_ERROR.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, rtdp, ipip_gre_key_check, 0x08, 23, 1);
+
+/* reg_rtdp_ipip_ipv4_usip
+ * Underlay IPv4 address for ipv4 source address check.
+ * Reserved when sip_check is not '1'.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, rtdp, ipip_ipv4_usip, 0x0C, 0, 32);
+
+/* reg_rtdp_ipip_ipv6_usip_ptr
+ * This field is valid when sip_check is "sipv6 check explicitly". This is a
+ * pointer to the IPv6 DIP which is configured by RIPS. For Spectrum, the index
+ * is to the KVD linear.
+ * Reserved when sip_check is not MLXSW_REG_RTDP_IPIP_SIP_CHECK_FILTER_IPV6.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, rtdp, ipip_ipv6_usip_ptr, 0x10, 0, 24);
+
+/* reg_rtdp_ipip_expected_gre_key
+ * GRE key for checking.
+ * Reserved when gre_key_check is '0'.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, rtdp, ipip_expected_gre_key, 0x14, 0, 32);
+
+static inline void mlxsw_reg_rtdp_pack(char *payload,
+				       enum mlxsw_reg_rtdp_type type,
+				       u32 tunnel_index)
+{
+	MLXSW_REG_ZERO(rtdp, payload);
+	mlxsw_reg_rtdp_type_set(payload, type);
+	mlxsw_reg_rtdp_tunnel_index_set(payload, tunnel_index);
+}
+
+static inline void
+mlxsw_reg_rtdp_ipip_pack(char *payload, u16 irif,
+			 enum mlxsw_reg_rtdp_ipip_sip_check sip_check,
+			 unsigned int type_check, bool gre_key_check,
+			 u32 expected_gre_key)
+{
+	mlxsw_reg_rtdp_ipip_irif_set(payload, irif);
+	mlxsw_reg_rtdp_ipip_sip_check_set(payload, sip_check);
+	mlxsw_reg_rtdp_ipip_type_check_set(payload, type_check);
+	mlxsw_reg_rtdp_ipip_gre_key_check_set(payload, gre_key_check);
+	mlxsw_reg_rtdp_ipip_expected_gre_key_set(payload, expected_gre_key);
+}
+
+static inline void
+mlxsw_reg_rtdp_ipip4_pack(char *payload, u16 irif,
+			  enum mlxsw_reg_rtdp_ipip_sip_check sip_check,
+			  unsigned int type_check, bool gre_key_check,
+			  u32 ipv4_usip, u32 expected_gre_key)
+{
+	mlxsw_reg_rtdp_ipip_pack(payload, irif, sip_check, type_check,
+				 gre_key_check, expected_gre_key);
+	mlxsw_reg_rtdp_ipip_ipv4_usip_set(payload, ipv4_usip);
+}
+
+static inline void
+mlxsw_reg_rtdp_ipip6_pack(char *payload, u16 irif,
+			  enum mlxsw_reg_rtdp_ipip_sip_check sip_check,
+			  unsigned int type_check, bool gre_key_check,
+			  u32 ipv6_usip_ptr, u32 expected_gre_key)
+{
+	mlxsw_reg_rtdp_ipip_pack(payload, irif, sip_check, type_check,
+				 gre_key_check, expected_gre_key);
+	mlxsw_reg_rtdp_ipip_ipv6_usip_ptr_set(payload, ipv6_usip_ptr);
+}
+
+/* RIPS - Router IP version Six Register
+ * -------------------------------------
+ * The RIPS register is used to store IPv6 addresses for use by the NVE and
+ * IPinIP
+ */
+#define MLXSW_REG_RIPS_ID 0x8021
+#define MLXSW_REG_RIPS_LEN 0x14
+
+MLXSW_REG_DEFINE(rips, MLXSW_REG_RIPS_ID, MLXSW_REG_RIPS_LEN);
+
+/* reg_rips_index
+ * Index to IPv6 address.
+ * For Spectrum, the index is to the KVD linear.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, rips, index, 0x00, 0, 24);
+
+/* reg_rips_ipv6
+ * IPv6 address
+ * Access: RW
+ */
+MLXSW_ITEM_BUF(reg, rips, ipv6, 0x04, 16);
+
+static inline void mlxsw_reg_rips_pack(char *payload, u32 index,
+				       const struct in6_addr *ipv6)
+{
+	MLXSW_REG_ZERO(rips, payload);
+	mlxsw_reg_rips_index_set(payload, index);
+	mlxsw_reg_rips_ipv6_memcpy_to(payload, (const char *)ipv6);
+}
+
+/* RATRAD - Router Adjacency Table Activity Dump Register
+ * ------------------------------------------------------
+ * The RATRAD register is used to dump and optionally clear activity bits of
+ * router adjacency table entries.
+ */
+#define MLXSW_REG_RATRAD_ID 0x8022
+#define MLXSW_REG_RATRAD_LEN 0x210
+
+MLXSW_REG_DEFINE(ratrad, MLXSW_REG_RATRAD_ID, MLXSW_REG_RATRAD_LEN);
+
+enum {
+	/* Read activity */
+	MLXSW_REG_RATRAD_OP_READ_ACTIVITY,
+	/* Read and clear activity */
+	MLXSW_REG_RATRAD_OP_READ_CLEAR_ACTIVITY,
+};
+
+/* reg_ratrad_op
+ * Access: Operation
+ */
+MLXSW_ITEM32(reg, ratrad, op, 0x00, 30, 2);
+
+/* reg_ratrad_ecmp_size
+ * ecmp_size is the amount of sequential entries from adjacency_index. Valid
+ * ranges:
+ * Spectrum-1: 32-64, 512, 1024, 2048, 4096
+ * Spectrum-2/3: 32-128, 256, 512, 1024, 2048, 4096
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, ratrad, ecmp_size, 0x00, 0, 13);
+
+/* reg_ratrad_adjacency_index
+ * Index into the adjacency table.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, ratrad, adjacency_index, 0x04, 0, 24);
+
+/* reg_ratrad_activity_vector
+ * Activity bit per adjacency index.
+ * Bits higher than ecmp_size are reserved.
+ * Access: RO
+ */
+MLXSW_ITEM_BIT_ARRAY(reg, ratrad, activity_vector, 0x10, 0x200, 1);
+
+static inline void mlxsw_reg_ratrad_pack(char *payload, u32 adjacency_index,
+					 u16 ecmp_size)
+{
+	MLXSW_REG_ZERO(ratrad, payload);
+	mlxsw_reg_ratrad_op_set(payload,
+				MLXSW_REG_RATRAD_OP_READ_CLEAR_ACTIVITY);
+	mlxsw_reg_ratrad_ecmp_size_set(payload, ecmp_size);
+	mlxsw_reg_ratrad_adjacency_index_set(payload, adjacency_index);
+}
+
+/* RIGR-V2 - Router Interface Group Register Version 2
+ * ---------------------------------------------------
+ * The RIGR_V2 register is used to add, remove and query egress interface list
+ * of a multicast forwarding entry.
+ */
+#define MLXSW_REG_RIGR2_ID 0x8023
+#define MLXSW_REG_RIGR2_LEN 0xB0
+
+#define MLXSW_REG_RIGR2_MAX_ERIFS 32
+
+MLXSW_REG_DEFINE(rigr2, MLXSW_REG_RIGR2_ID, MLXSW_REG_RIGR2_LEN);
+
+/* reg_rigr2_rigr_index
+ * KVD Linear index.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, rigr2, rigr_index, 0x04, 0, 24);
+
+/* reg_rigr2_vnext
+ * Next RIGR Index is valid.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, rigr2, vnext, 0x08, 31, 1);
+
+/* reg_rigr2_next_rigr_index
+ * Next RIGR Index. The index is to the KVD linear.
+ * Reserved when vnxet = '0'.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, rigr2, next_rigr_index, 0x08, 0, 24);
+
+/* reg_rigr2_vrmid
+ * RMID Index is valid.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, rigr2, vrmid, 0x20, 31, 1);
+
+/* reg_rigr2_rmid_index
+ * RMID Index.
+ * Range 0 .. max_mid - 1
+ * Reserved when vrmid = '0'.
+ * The index is to the Port Group Table (PGT)
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, rigr2, rmid_index, 0x20, 0, 16);
+
+/* reg_rigr2_erif_entry_v
+ * Egress Router Interface is valid.
+ * Note that low-entries must be set if high-entries are set. For
+ * example: if erif_entry[2].v is set then erif_entry[1].v and
+ * erif_entry[0].v must be set.
+ * Index can be from 0 to cap_mc_erif_list_entries-1
+ * Access: RW
+ */
+MLXSW_ITEM32_INDEXED(reg, rigr2, erif_entry_v, 0x24, 31, 1, 4, 0, false);
+
+/* reg_rigr2_erif_entry_erif
+ * Egress Router Interface.
+ * Valid range is from 0 to cap_max_router_interfaces - 1
+ * Index can be from 0 to MLXSW_REG_RIGR2_MAX_ERIFS - 1
+ * Access: RW
+ */
+MLXSW_ITEM32_INDEXED(reg, rigr2, erif_entry_erif, 0x24, 0, 16, 4, 0, false);
+
+static inline void mlxsw_reg_rigr2_pack(char *payload, u32 rigr_index,
+					bool vnext, u32 next_rigr_index)
+{
+	MLXSW_REG_ZERO(rigr2, payload);
+	mlxsw_reg_rigr2_rigr_index_set(payload, rigr_index);
+	mlxsw_reg_rigr2_vnext_set(payload, vnext);
+	mlxsw_reg_rigr2_next_rigr_index_set(payload, next_rigr_index);
+	mlxsw_reg_rigr2_vrmid_set(payload, 0);
+	mlxsw_reg_rigr2_rmid_index_set(payload, 0);
+}
+
+static inline void mlxsw_reg_rigr2_erif_entry_pack(char *payload, int index,
+						   bool v, u16 erif)
+{
+	mlxsw_reg_rigr2_erif_entry_v_set(payload, index, v);
+	mlxsw_reg_rigr2_erif_entry_erif_set(payload, index, erif);
+}
+
+/* RECR-V2 - Router ECMP Configuration Version 2 Register
+ * ------------------------------------------------------
+ */
+#define MLXSW_REG_RECR2_ID 0x8025
+#define MLXSW_REG_RECR2_LEN 0x38
+
+MLXSW_REG_DEFINE(recr2, MLXSW_REG_RECR2_ID, MLXSW_REG_RECR2_LEN);
+
+/* reg_recr2_pp
+ * Per-port configuration
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, recr2, pp, 0x00, 24, 1);
+
+/* reg_recr2_sh
+ * Symmetric hash
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, recr2, sh, 0x00, 8, 1);
+
+/* reg_recr2_seed
+ * Seed
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, recr2, seed, 0x08, 0, 32);
+
+enum {
+	/* Enable IPv4 fields if packet is not TCP and not UDP */
+	MLXSW_REG_RECR2_IPV4_EN_NOT_TCP_NOT_UDP	= 3,
+	/* Enable IPv4 fields if packet is TCP or UDP */
+	MLXSW_REG_RECR2_IPV4_EN_TCP_UDP		= 4,
+	/* Enable IPv6 fields if packet is not TCP and not UDP */
+	MLXSW_REG_RECR2_IPV6_EN_NOT_TCP_NOT_UDP	= 5,
+	/* Enable IPv6 fields if packet is TCP or UDP */
+	MLXSW_REG_RECR2_IPV6_EN_TCP_UDP		= 6,
+	/* Enable TCP/UDP header fields if packet is IPv4 */
+	MLXSW_REG_RECR2_TCP_UDP_EN_IPV4		= 7,
+	/* Enable TCP/UDP header fields if packet is IPv6 */
+	MLXSW_REG_RECR2_TCP_UDP_EN_IPV6		= 8,
+
+	__MLXSW_REG_RECR2_HEADER_CNT,
+};
+
+/* reg_recr2_outer_header_enables
+ * Bit mask where each bit enables a specific layer to be included in
+ * the hash calculation.
+ * Access: RW
+ */
+MLXSW_ITEM_BIT_ARRAY(reg, recr2, outer_header_enables, 0x10, 0x04, 1);
+
+enum {
+	/* IPv4 Source IP */
+	MLXSW_REG_RECR2_IPV4_SIP0			= 9,
+	MLXSW_REG_RECR2_IPV4_SIP3			= 12,
+	/* IPv4 Destination IP */
+	MLXSW_REG_RECR2_IPV4_DIP0			= 13,
+	MLXSW_REG_RECR2_IPV4_DIP3			= 16,
+	/* IP Protocol */
+	MLXSW_REG_RECR2_IPV4_PROTOCOL			= 17,
+	/* IPv6 Source IP */
+	MLXSW_REG_RECR2_IPV6_SIP0_7			= 21,
+	MLXSW_REG_RECR2_IPV6_SIP8			= 29,
+	MLXSW_REG_RECR2_IPV6_SIP15			= 36,
+	/* IPv6 Destination IP */
+	MLXSW_REG_RECR2_IPV6_DIP0_7			= 37,
+	MLXSW_REG_RECR2_IPV6_DIP8			= 45,
+	MLXSW_REG_RECR2_IPV6_DIP15			= 52,
+	/* IPv6 Next Header */
+	MLXSW_REG_RECR2_IPV6_NEXT_HEADER		= 53,
+	/* IPv6 Flow Label */
+	MLXSW_REG_RECR2_IPV6_FLOW_LABEL			= 57,
+	/* TCP/UDP Source Port */
+	MLXSW_REG_RECR2_TCP_UDP_SPORT			= 74,
+	/* TCP/UDP Destination Port */
+	MLXSW_REG_RECR2_TCP_UDP_DPORT			= 75,
+
+	__MLXSW_REG_RECR2_FIELD_CNT,
+};
+
+/* reg_recr2_outer_header_fields_enable
+ * Packet fields to enable for ECMP hash subject to outer_header_enable.
+ * Access: RW
+ */
+MLXSW_ITEM_BIT_ARRAY(reg, recr2, outer_header_fields_enable, 0x14, 0x14, 1);
+
+/* reg_recr2_inner_header_enables
+ * Bit mask where each bit enables a specific inner layer to be included in the
+ * hash calculation. Same values as reg_recr2_outer_header_enables.
+ * Access: RW
+ */
+MLXSW_ITEM_BIT_ARRAY(reg, recr2, inner_header_enables, 0x2C, 0x04, 1);
+
+enum {
+	/* Inner IPv4 Source IP */
+	MLXSW_REG_RECR2_INNER_IPV4_SIP0			= 3,
+	MLXSW_REG_RECR2_INNER_IPV4_SIP3			= 6,
+	/* Inner IPv4 Destination IP */
+	MLXSW_REG_RECR2_INNER_IPV4_DIP0			= 7,
+	MLXSW_REG_RECR2_INNER_IPV4_DIP3			= 10,
+	/* Inner IP Protocol */
+	MLXSW_REG_RECR2_INNER_IPV4_PROTOCOL		= 11,
+	/* Inner IPv6 Source IP */
+	MLXSW_REG_RECR2_INNER_IPV6_SIP0_7		= 12,
+	MLXSW_REG_RECR2_INNER_IPV6_SIP8			= 20,
+	MLXSW_REG_RECR2_INNER_IPV6_SIP15		= 27,
+	/* Inner IPv6 Destination IP */
+	MLXSW_REG_RECR2_INNER_IPV6_DIP0_7		= 28,
+	MLXSW_REG_RECR2_INNER_IPV6_DIP8			= 36,
+	MLXSW_REG_RECR2_INNER_IPV6_DIP15		= 43,
+	/* Inner IPv6 Next Header */
+	MLXSW_REG_RECR2_INNER_IPV6_NEXT_HEADER		= 44,
+	/* Inner IPv6 Flow Label */
+	MLXSW_REG_RECR2_INNER_IPV6_FLOW_LABEL		= 45,
+	/* Inner TCP/UDP Source Port */
+	MLXSW_REG_RECR2_INNER_TCP_UDP_SPORT		= 46,
+	/* Inner TCP/UDP Destination Port */
+	MLXSW_REG_RECR2_INNER_TCP_UDP_DPORT		= 47,
+
+	__MLXSW_REG_RECR2_INNER_FIELD_CNT,
+};
+
+/* reg_recr2_inner_header_fields_enable
+ * Inner packet fields to enable for ECMP hash subject to inner_header_enables.
+ * Access: RW
+ */
+MLXSW_ITEM_BIT_ARRAY(reg, recr2, inner_header_fields_enable, 0x30, 0x08, 1);
+
+static inline void mlxsw_reg_recr2_pack(char *payload, u32 seed)
+{
+	MLXSW_REG_ZERO(recr2, payload);
+	mlxsw_reg_recr2_pp_set(payload, false);
+	mlxsw_reg_recr2_sh_set(payload, true);
+	mlxsw_reg_recr2_seed_set(payload, seed);
+}
+
+/* RMFT-V2 - Router Multicast Forwarding Table Version 2 Register
+ * --------------------------------------------------------------
+ * The RMFT_V2 register is used to configure and query the multicast table.
+ */
+#define MLXSW_REG_RMFT2_ID 0x8027
+#define MLXSW_REG_RMFT2_LEN 0x174
+
+MLXSW_REG_DEFINE(rmft2, MLXSW_REG_RMFT2_ID, MLXSW_REG_RMFT2_LEN);
+
+/* reg_rmft2_v
+ * Valid
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, rmft2, v, 0x00, 31, 1);
+
+enum mlxsw_reg_rmft2_type {
+	MLXSW_REG_RMFT2_TYPE_IPV4,
+	MLXSW_REG_RMFT2_TYPE_IPV6
+};
+
+/* reg_rmft2_type
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, rmft2, type, 0x00, 28, 2);
+
+enum mlxsw_sp_reg_rmft2_op {
+	/* For Write:
+	 * Write operation. Used to write a new entry to the table. All RW
+	 * fields are relevant for new entry. Activity bit is set for new
+	 * entries - Note write with v (Valid) 0 will delete the entry.
+	 * For Query:
+	 * Read operation
+	 */
+	MLXSW_REG_RMFT2_OP_READ_WRITE,
+};
+
+/* reg_rmft2_op
+ * Operation.
+ * Access: OP
+ */
+MLXSW_ITEM32(reg, rmft2, op, 0x00, 20, 2);
+
+/* reg_rmft2_a
+ * Activity. Set for new entries. Set if a packet lookup has hit on the specific
+ * entry.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, rmft2, a, 0x00, 16, 1);
+
+/* reg_rmft2_offset
+ * Offset within the multicast forwarding table to write to.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, rmft2, offset, 0x00, 0, 16);
+
+/* reg_rmft2_virtual_router
+ * Virtual Router ID. Range from 0..cap_max_virtual_routers-1
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, rmft2, virtual_router, 0x04, 0, 16);
+
+enum mlxsw_reg_rmft2_irif_mask {
+	MLXSW_REG_RMFT2_IRIF_MASK_IGNORE,
+	MLXSW_REG_RMFT2_IRIF_MASK_COMPARE
+};
+
+/* reg_rmft2_irif_mask
+ * Ingress RIF mask.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, rmft2, irif_mask, 0x08, 24, 1);
+
+/* reg_rmft2_irif
+ * Ingress RIF index.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, rmft2, irif, 0x08, 0, 16);
+
+/* reg_rmft2_dip{4,6}
+ * Destination IPv4/6 address
+ * Access: RW
+ */
+MLXSW_ITEM_BUF(reg, rmft2, dip6, 0x10, 16);
+MLXSW_ITEM32(reg, rmft2, dip4, 0x1C, 0, 32);
+
+/* reg_rmft2_dip{4,6}_mask
+ * A bit that is set directs the TCAM to compare the corresponding bit in key. A
+ * bit that is clear directs the TCAM to ignore the corresponding bit in key.
+ * Access: RW
+ */
+MLXSW_ITEM_BUF(reg, rmft2, dip6_mask, 0x20, 16);
+MLXSW_ITEM32(reg, rmft2, dip4_mask, 0x2C, 0, 32);
+
+/* reg_rmft2_sip{4,6}
+ * Source IPv4/6 address
+ * Access: RW
+ */
+MLXSW_ITEM_BUF(reg, rmft2, sip6, 0x30, 16);
+MLXSW_ITEM32(reg, rmft2, sip4, 0x3C, 0, 32);
+
+/* reg_rmft2_sip{4,6}_mask
+ * A bit that is set directs the TCAM to compare the corresponding bit in key. A
+ * bit that is clear directs the TCAM to ignore the corresponding bit in key.
+ * Access: RW
+ */
+MLXSW_ITEM_BUF(reg, rmft2, sip6_mask, 0x40, 16);
+MLXSW_ITEM32(reg, rmft2, sip4_mask, 0x4C, 0, 32);
+
+/* reg_rmft2_flexible_action_set
+ * ACL action set. The only supported action types in this field and in any
+ * action-set pointed from here are as follows:
+ * 00h: ACTION_NULL
+ * 01h: ACTION_MAC_TTL, only TTL configuration is supported.
+ * 03h: ACTION_TRAP
+ * 06h: ACTION_QOS
+ * 08h: ACTION_POLICING_MONITORING
+ * 10h: ACTION_ROUTER_MC
+ * Access: RW
+ */
+MLXSW_ITEM_BUF(reg, rmft2, flexible_action_set, 0x80,
+	       MLXSW_REG_FLEX_ACTION_SET_LEN);
+
+static inline void
+mlxsw_reg_rmft2_common_pack(char *payload, bool v, u16 offset,
+			    u16 virtual_router,
+			    enum mlxsw_reg_rmft2_irif_mask irif_mask, u16 irif,
+			    const char *flex_action_set)
+{
+	MLXSW_REG_ZERO(rmft2, payload);
+	mlxsw_reg_rmft2_v_set(payload, v);
+	mlxsw_reg_rmft2_op_set(payload, MLXSW_REG_RMFT2_OP_READ_WRITE);
+	mlxsw_reg_rmft2_offset_set(payload, offset);
+	mlxsw_reg_rmft2_virtual_router_set(payload, virtual_router);
+	mlxsw_reg_rmft2_irif_mask_set(payload, irif_mask);
+	mlxsw_reg_rmft2_irif_set(payload, irif);
+	if (flex_action_set)
+		mlxsw_reg_rmft2_flexible_action_set_memcpy_to(payload,
+							      flex_action_set);
+}
+
+static inline void
+mlxsw_reg_rmft2_ipv4_pack(char *payload, bool v, u16 offset, u16 virtual_router,
+			  enum mlxsw_reg_rmft2_irif_mask irif_mask, u16 irif,
+			  u32 dip4, u32 dip4_mask, u32 sip4, u32 sip4_mask,
+			  const char *flexible_action_set)
+{
+	mlxsw_reg_rmft2_common_pack(payload, v, offset, virtual_router,
+				    irif_mask, irif, flexible_action_set);
+	mlxsw_reg_rmft2_type_set(payload, MLXSW_REG_RMFT2_TYPE_IPV4);
+	mlxsw_reg_rmft2_dip4_set(payload, dip4);
+	mlxsw_reg_rmft2_dip4_mask_set(payload, dip4_mask);
+	mlxsw_reg_rmft2_sip4_set(payload, sip4);
+	mlxsw_reg_rmft2_sip4_mask_set(payload, sip4_mask);
+}
+
+static inline void
+mlxsw_reg_rmft2_ipv6_pack(char *payload, bool v, u16 offset, u16 virtual_router,
+			  enum mlxsw_reg_rmft2_irif_mask irif_mask, u16 irif,
+			  struct in6_addr dip6, struct in6_addr dip6_mask,
+			  struct in6_addr sip6, struct in6_addr sip6_mask,
+			  const char *flexible_action_set)
+{
+	mlxsw_reg_rmft2_common_pack(payload, v, offset, virtual_router,
+				    irif_mask, irif, flexible_action_set);
+	mlxsw_reg_rmft2_type_set(payload, MLXSW_REG_RMFT2_TYPE_IPV6);
+	mlxsw_reg_rmft2_dip6_memcpy_to(payload, (void *)&dip6);
+	mlxsw_reg_rmft2_dip6_mask_memcpy_to(payload, (void *)&dip6_mask);
+	mlxsw_reg_rmft2_sip6_memcpy_to(payload, (void *)&sip6);
+	mlxsw_reg_rmft2_sip6_mask_memcpy_to(payload, (void *)&sip6_mask);
+}
+
+/* REIV - Router Egress Interface to VID Register
+ * ----------------------------------------------
+ * The REIV register maps {eRIF, egress_port} -> VID.
+ * This mapping is done at the egress, after the ACLs.
+ * This mapping always takes effect after router, regardless of cast
+ * (for unicast/multicast/port-base multicast), regardless of eRIF type and
+ * regardless of bridge decisions (e.g. SFD for unicast or SMPE).
+ * Reserved when the RIF is a loopback RIF.
+ *
+ * Note: Reserved when legacy bridge model is used.
+ */
+#define MLXSW_REG_REIV_ID 0x8034
+#define MLXSW_REG_REIV_BASE_LEN 0x20 /* base length, without records */
+#define MLXSW_REG_REIV_REC_LEN 0x04 /* record length */
+#define MLXSW_REG_REIV_REC_MAX_COUNT 256 /* firmware limitation */
+#define MLXSW_REG_REIV_LEN (MLXSW_REG_REIV_BASE_LEN +	\
+			    MLXSW_REG_REIV_REC_LEN *	\
+			    MLXSW_REG_REIV_REC_MAX_COUNT)
+
+MLXSW_REG_DEFINE(reiv, MLXSW_REG_REIV_ID, MLXSW_REG_REIV_LEN);
+
+/* reg_reiv_port_page
+ * Port page - elport_record[0] is 256*port_page.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, reiv, port_page, 0x00, 0, 4);
+
+/* reg_reiv_erif
+ * Egress RIF.
+ * Range is 0..cap_max_router_interfaces-1.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, reiv, erif, 0x04, 0, 16);
+
+/* reg_reiv_rec_update
+ * Update enable (when write):
+ * 0 - Do not update the entry.
+ * 1 - Update the entry.
+ * Access: OP
+ */
+MLXSW_ITEM32_INDEXED(reg, reiv, rec_update, MLXSW_REG_REIV_BASE_LEN, 31, 1,
+		     MLXSW_REG_REIV_REC_LEN, 0x00, false);
+
+/* reg_reiv_rec_evid
+ * Egress VID.
+ * Range is 0..4095.
+ * Access: RW
+ */
+MLXSW_ITEM32_INDEXED(reg, reiv, rec_evid, MLXSW_REG_REIV_BASE_LEN, 0, 12,
+		     MLXSW_REG_REIV_REC_LEN, 0x00, false);
+
+static inline void mlxsw_reg_reiv_pack(char *payload, u8 port_page, u16 erif)
+{
+	MLXSW_REG_ZERO(reiv, payload);
+	mlxsw_reg_reiv_port_page_set(payload, port_page);
+	mlxsw_reg_reiv_erif_set(payload, erif);
+}
+
+/* MFCR - Management Fan Control Register
+ * --------------------------------------
+ * This register controls the settings of the Fan Speed PWM mechanism.
+ */
+#define MLXSW_REG_MFCR_ID 0x9001
+#define MLXSW_REG_MFCR_LEN 0x08
+
+MLXSW_REG_DEFINE(mfcr, MLXSW_REG_MFCR_ID, MLXSW_REG_MFCR_LEN);
+
+enum mlxsw_reg_mfcr_pwm_frequency {
+	MLXSW_REG_MFCR_PWM_FEQ_11HZ = 0x00,
+	MLXSW_REG_MFCR_PWM_FEQ_14_7HZ = 0x01,
+	MLXSW_REG_MFCR_PWM_FEQ_22_1HZ = 0x02,
+	MLXSW_REG_MFCR_PWM_FEQ_1_4KHZ = 0x40,
+	MLXSW_REG_MFCR_PWM_FEQ_5KHZ = 0x41,
+	MLXSW_REG_MFCR_PWM_FEQ_20KHZ = 0x42,
+	MLXSW_REG_MFCR_PWM_FEQ_22_5KHZ = 0x43,
+	MLXSW_REG_MFCR_PWM_FEQ_25KHZ = 0x44,
+};
+
+/* reg_mfcr_pwm_frequency
+ * Controls the frequency of the PWM signal.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mfcr, pwm_frequency, 0x00, 0, 7);
+
+#define MLXSW_MFCR_TACHOS_MAX 10
+
+/* reg_mfcr_tacho_active
+ * Indicates which of the tachometer is active (bit per tachometer).
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mfcr, tacho_active, 0x04, 16, MLXSW_MFCR_TACHOS_MAX);
+
+#define MLXSW_MFCR_PWMS_MAX 5
+
+/* reg_mfcr_pwm_active
+ * Indicates which of the PWM control is active (bit per PWM).
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mfcr, pwm_active, 0x04, 0, MLXSW_MFCR_PWMS_MAX);
+
+static inline void
+mlxsw_reg_mfcr_pack(char *payload,
+		    enum mlxsw_reg_mfcr_pwm_frequency pwm_frequency)
+{
+	MLXSW_REG_ZERO(mfcr, payload);
+	mlxsw_reg_mfcr_pwm_frequency_set(payload, pwm_frequency);
+}
+
+static inline void
+mlxsw_reg_mfcr_unpack(char *payload,
+		      enum mlxsw_reg_mfcr_pwm_frequency *p_pwm_frequency,
+		      u16 *p_tacho_active, u8 *p_pwm_active)
+{
+	*p_pwm_frequency = mlxsw_reg_mfcr_pwm_frequency_get(payload);
+	*p_tacho_active = mlxsw_reg_mfcr_tacho_active_get(payload);
+	*p_pwm_active = mlxsw_reg_mfcr_pwm_active_get(payload);
+}
+
+/* MFSC - Management Fan Speed Control Register
+ * --------------------------------------------
+ * This register controls the settings of the Fan Speed PWM mechanism.
+ */
+#define MLXSW_REG_MFSC_ID 0x9002
+#define MLXSW_REG_MFSC_LEN 0x08
+
+MLXSW_REG_DEFINE(mfsc, MLXSW_REG_MFSC_ID, MLXSW_REG_MFSC_LEN);
+
+/* reg_mfsc_pwm
+ * Fan pwm to control / monitor.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, mfsc, pwm, 0x00, 24, 3);
+
+/* reg_mfsc_pwm_duty_cycle
+ * Controls the duty cycle of the PWM. Value range from 0..255 to
+ * represent duty cycle of 0%...100%.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mfsc, pwm_duty_cycle, 0x04, 0, 8);
+
+static inline void mlxsw_reg_mfsc_pack(char *payload, u8 pwm,
+				       u8 pwm_duty_cycle)
+{
+	MLXSW_REG_ZERO(mfsc, payload);
+	mlxsw_reg_mfsc_pwm_set(payload, pwm);
+	mlxsw_reg_mfsc_pwm_duty_cycle_set(payload, pwm_duty_cycle);
+}
+
+/* MFSM - Management Fan Speed Measurement
+ * ---------------------------------------
+ * This register controls the settings of the Tacho measurements and
+ * enables reading the Tachometer measurements.
+ */
+#define MLXSW_REG_MFSM_ID 0x9003
+#define MLXSW_REG_MFSM_LEN 0x08
+
+MLXSW_REG_DEFINE(mfsm, MLXSW_REG_MFSM_ID, MLXSW_REG_MFSM_LEN);
+
+/* reg_mfsm_tacho
+ * Fan tachometer index.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, mfsm, tacho, 0x00, 24, 4);
+
+/* reg_mfsm_rpm
+ * Fan speed (round per minute).
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mfsm, rpm, 0x04, 0, 16);
+
+static inline void mlxsw_reg_mfsm_pack(char *payload, u8 tacho)
+{
+	MLXSW_REG_ZERO(mfsm, payload);
+	mlxsw_reg_mfsm_tacho_set(payload, tacho);
+}
+
+/* MFSL - Management Fan Speed Limit Register
+ * ------------------------------------------
+ * The Fan Speed Limit register is used to configure the fan speed
+ * event / interrupt notification mechanism. Fan speed threshold are
+ * defined for both under-speed and over-speed.
+ */
+#define MLXSW_REG_MFSL_ID 0x9004
+#define MLXSW_REG_MFSL_LEN 0x0C
+
+MLXSW_REG_DEFINE(mfsl, MLXSW_REG_MFSL_ID, MLXSW_REG_MFSL_LEN);
+
+/* reg_mfsl_tacho
+ * Fan tachometer index.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, mfsl, tacho, 0x00, 24, 4);
+
+/* reg_mfsl_tach_min
+ * Tachometer minimum value (minimum RPM).
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mfsl, tach_min, 0x04, 0, 16);
+
+/* reg_mfsl_tach_max
+ * Tachometer maximum value (maximum RPM).
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mfsl, tach_max, 0x08, 0, 16);
+
+static inline void mlxsw_reg_mfsl_pack(char *payload, u8 tacho,
+				       u16 tach_min, u16 tach_max)
+{
+	MLXSW_REG_ZERO(mfsl, payload);
+	mlxsw_reg_mfsl_tacho_set(payload, tacho);
+	mlxsw_reg_mfsl_tach_min_set(payload, tach_min);
+	mlxsw_reg_mfsl_tach_max_set(payload, tach_max);
+}
+
+static inline void mlxsw_reg_mfsl_unpack(char *payload, u8 tacho,
+					 u16 *p_tach_min, u16 *p_tach_max)
+{
+	if (p_tach_min)
+		*p_tach_min = mlxsw_reg_mfsl_tach_min_get(payload);
+
+	if (p_tach_max)
+		*p_tach_max = mlxsw_reg_mfsl_tach_max_get(payload);
+}
+
+/* FORE - Fan Out of Range Event Register
+ * --------------------------------------
+ * This register reports the status of the controlled fans compared to the
+ * range defined by the MFSL register.
+ */
+#define MLXSW_REG_FORE_ID 0x9007
+#define MLXSW_REG_FORE_LEN 0x0C
+
+MLXSW_REG_DEFINE(fore, MLXSW_REG_FORE_ID, MLXSW_REG_FORE_LEN);
+
+/* fan_under_limit
+ * Fan speed is below the low limit defined in MFSL register. Each bit relates
+ * to a single tachometer and indicates the specific tachometer reading is
+ * below the threshold.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, fore, fan_under_limit, 0x00, 16, 10);
+
+static inline void mlxsw_reg_fore_unpack(char *payload, u8 tacho,
+					 bool *fault)
+{
+	u16 limit;
+
+	if (fault) {
+		limit = mlxsw_reg_fore_fan_under_limit_get(payload);
+		*fault = limit & BIT(tacho);
+	}
+}
+
+/* MTCAP - Management Temperature Capabilities
+ * -------------------------------------------
+ * This register exposes the capabilities of the device and
+ * system temperature sensing.
+ */
+#define MLXSW_REG_MTCAP_ID 0x9009
+#define MLXSW_REG_MTCAP_LEN 0x08
+
+MLXSW_REG_DEFINE(mtcap, MLXSW_REG_MTCAP_ID, MLXSW_REG_MTCAP_LEN);
+
+/* reg_mtcap_sensor_count
+ * Number of sensors supported by the device.
+ * This includes the QSFP module sensors (if exists in the QSFP module).
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mtcap, sensor_count, 0x00, 0, 7);
+
+/* MTMP - Management Temperature
+ * -----------------------------
+ * This register controls the settings of the temperature measurements
+ * and enables reading the temperature measurements. Note that temperature
+ * is in 0.125 degrees Celsius.
+ */
+#define MLXSW_REG_MTMP_ID 0x900A
+#define MLXSW_REG_MTMP_LEN 0x20
+
+MLXSW_REG_DEFINE(mtmp, MLXSW_REG_MTMP_ID, MLXSW_REG_MTMP_LEN);
+
+/* reg_mtmp_slot_index
+ * Slot index (0: Main board).
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, mtmp, slot_index, 0x00, 16, 4);
+
+#define MLXSW_REG_MTMP_MODULE_INDEX_MIN 64
+#define MLXSW_REG_MTMP_GBOX_INDEX_MIN 256
+/* reg_mtmp_sensor_index
+ * Sensors index to access.
+ * 64-127 of sensor_index are mapped to the SFP+/QSFP modules sequentially
+ * (module 0 is mapped to sensor_index 64).
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, mtmp, sensor_index, 0x00, 0, 12);
+
+/* Convert to milli degrees Celsius */
+#define MLXSW_REG_MTMP_TEMP_TO_MC(val) ({ typeof(val) v_ = (val); \
+					  ((v_) >= 0) ? ((v_) * 125) : \
+					  ((s16)((GENMASK(15, 0) + (v_) + 1) \
+					   * 125)); })
+
+/* reg_mtmp_max_operational_temperature
+ * The highest temperature in the nominal operational range. Reading is in
+ * 0.125 Celsius degrees units.
+ * In case of module this is SFF critical temperature threshold.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mtmp, max_operational_temperature, 0x04, 16, 16);
+
+/* reg_mtmp_temperature
+ * Temperature reading from the sensor. Reading is in 0.125 Celsius
+ * degrees units.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mtmp, temperature, 0x04, 0, 16);
+
+/* reg_mtmp_mte
+ * Max Temperature Enable - enables measuring the max temperature on a sensor.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mtmp, mte, 0x08, 31, 1);
+
+/* reg_mtmp_mtr
+ * Max Temperature Reset - clears the value of the max temperature register.
+ * Access: WO
+ */
+MLXSW_ITEM32(reg, mtmp, mtr, 0x08, 30, 1);
+
+/* reg_mtmp_max_temperature
+ * The highest measured temperature from the sensor.
+ * When the bit mte is cleared, the field max_temperature is reserved.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mtmp, max_temperature, 0x08, 0, 16);
+
+/* reg_mtmp_tee
+ * Temperature Event Enable.
+ * 0 - Do not generate event
+ * 1 - Generate event
+ * 2 - Generate single event
+ * Access: RW
+ */
+
+enum mlxsw_reg_mtmp_tee {
+	MLXSW_REG_MTMP_TEE_NO_EVENT,
+	MLXSW_REG_MTMP_TEE_GENERATE_EVENT,
+	MLXSW_REG_MTMP_TEE_GENERATE_SINGLE_EVENT,
+};
+
+MLXSW_ITEM32(reg, mtmp, tee, 0x0C, 30, 2);
+
+#define MLXSW_REG_MTMP_THRESH_HI 0x348	/* 105 Celsius */
+
+/* reg_mtmp_temperature_threshold_hi
+ * High threshold for Temperature Warning Event. In 0.125 Celsius.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mtmp, temperature_threshold_hi, 0x0C, 0, 16);
+
+#define MLXSW_REG_MTMP_HYSTERESIS_TEMP 0x28 /* 5 Celsius */
+/* reg_mtmp_temperature_threshold_lo
+ * Low threshold for Temperature Warning Event. In 0.125 Celsius.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mtmp, temperature_threshold_lo, 0x10, 0, 16);
+
+#define MLXSW_REG_MTMP_SENSOR_NAME_SIZE 8
+
+/* reg_mtmp_sensor_name
+ * Sensor Name
+ * Access: RO
+ */
+MLXSW_ITEM_BUF(reg, mtmp, sensor_name, 0x18, MLXSW_REG_MTMP_SENSOR_NAME_SIZE);
+
+static inline void mlxsw_reg_mtmp_pack(char *payload, u8 slot_index,
+				       u16 sensor_index, bool max_temp_enable,
+				       bool max_temp_reset)
+{
+	MLXSW_REG_ZERO(mtmp, payload);
+	mlxsw_reg_mtmp_slot_index_set(payload, slot_index);
+	mlxsw_reg_mtmp_sensor_index_set(payload, sensor_index);
+	mlxsw_reg_mtmp_mte_set(payload, max_temp_enable);
+	mlxsw_reg_mtmp_mtr_set(payload, max_temp_reset);
+	mlxsw_reg_mtmp_temperature_threshold_hi_set(payload,
+						    MLXSW_REG_MTMP_THRESH_HI);
+}
+
+static inline void mlxsw_reg_mtmp_unpack(char *payload, int *p_temp,
+					 int *p_max_temp, int *p_temp_hi,
+					 int *p_max_oper_temp,
+					 char *sensor_name)
+{
+	s16 temp;
+
+	if (p_temp) {
+		temp = mlxsw_reg_mtmp_temperature_get(payload);
+		*p_temp = MLXSW_REG_MTMP_TEMP_TO_MC(temp);
+	}
+	if (p_max_temp) {
+		temp = mlxsw_reg_mtmp_max_temperature_get(payload);
+		*p_max_temp = MLXSW_REG_MTMP_TEMP_TO_MC(temp);
+	}
+	if (p_temp_hi) {
+		temp = mlxsw_reg_mtmp_temperature_threshold_hi_get(payload);
+		*p_temp_hi = MLXSW_REG_MTMP_TEMP_TO_MC(temp);
+	}
+	if (p_max_oper_temp) {
+		temp = mlxsw_reg_mtmp_max_operational_temperature_get(payload);
+		*p_max_oper_temp = MLXSW_REG_MTMP_TEMP_TO_MC(temp);
+	}
+	if (sensor_name)
+		mlxsw_reg_mtmp_sensor_name_memcpy_from(payload, sensor_name);
+}
+
+/* MTWE - Management Temperature Warning Event
+ * -------------------------------------------
+ * This register is used for over temperature warning.
+ */
+#define MLXSW_REG_MTWE_ID 0x900B
+#define MLXSW_REG_MTWE_LEN 0x10
+
+MLXSW_REG_DEFINE(mtwe, MLXSW_REG_MTWE_ID, MLXSW_REG_MTWE_LEN);
+
+/* reg_mtwe_sensor_warning
+ * Bit vector indicating which of the sensor reading is above threshold.
+ * Address 00h bit31 is sensor_warning[127].
+ * Address 0Ch bit0 is sensor_warning[0].
+ * Access: RO
+ */
+MLXSW_ITEM_BIT_ARRAY(reg, mtwe, sensor_warning, 0x0, 0x10, 1);
+
+/* MTBR - Management Temperature Bulk Register
+ * -------------------------------------------
+ * This register is used for bulk temperature reading.
+ */
+#define MLXSW_REG_MTBR_ID 0x900F
+#define MLXSW_REG_MTBR_BASE_LEN 0x10 /* base length, without records */
+#define MLXSW_REG_MTBR_REC_LEN 0x04 /* record length */
+#define MLXSW_REG_MTBR_REC_MAX_COUNT 47 /* firmware limitation */
+#define MLXSW_REG_MTBR_LEN (MLXSW_REG_MTBR_BASE_LEN +	\
+			    MLXSW_REG_MTBR_REC_LEN *	\
+			    MLXSW_REG_MTBR_REC_MAX_COUNT)
+
+MLXSW_REG_DEFINE(mtbr, MLXSW_REG_MTBR_ID, MLXSW_REG_MTBR_LEN);
+
+/* reg_mtbr_slot_index
+ * Slot index (0: Main board).
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, mtbr, slot_index, 0x00, 16, 4);
+
+/* reg_mtbr_base_sensor_index
+ * Base sensors index to access (0 - ASIC sensor, 1-63 - ambient sensors,
+ * 64-127 are mapped to the SFP+/QSFP modules sequentially).
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, mtbr, base_sensor_index, 0x00, 0, 12);
+
+/* reg_mtbr_num_rec
+ * Request: Number of records to read
+ * Response: Number of records read
+ * See above description for more details.
+ * Range 1..255
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mtbr, num_rec, 0x04, 0, 8);
+
+/* reg_mtbr_rec_max_temp
+ * The highest measured temperature from the sensor.
+ * When the bit mte is cleared, the field max_temperature is reserved.
+ * Access: RO
+ */
+MLXSW_ITEM32_INDEXED(reg, mtbr, rec_max_temp, MLXSW_REG_MTBR_BASE_LEN, 16,
+		     16, MLXSW_REG_MTBR_REC_LEN, 0x00, false);
+
+/* reg_mtbr_rec_temp
+ * Temperature reading from the sensor. Reading is in 0..125 Celsius
+ * degrees units.
+ * Access: RO
+ */
+MLXSW_ITEM32_INDEXED(reg, mtbr, rec_temp, MLXSW_REG_MTBR_BASE_LEN, 0, 16,
+		     MLXSW_REG_MTBR_REC_LEN, 0x00, false);
+
+static inline void mlxsw_reg_mtbr_pack(char *payload, u8 slot_index,
+				       u16 base_sensor_index, u8 num_rec)
+{
+	MLXSW_REG_ZERO(mtbr, payload);
+	mlxsw_reg_mtbr_slot_index_set(payload, slot_index);
+	mlxsw_reg_mtbr_base_sensor_index_set(payload, base_sensor_index);
+	mlxsw_reg_mtbr_num_rec_set(payload, num_rec);
+}
+
+/* Error codes from temperatute reading */
+enum mlxsw_reg_mtbr_temp_status {
+	MLXSW_REG_MTBR_NO_CONN		= 0x8000,
+	MLXSW_REG_MTBR_NO_TEMP_SENS	= 0x8001,
+	MLXSW_REG_MTBR_INDEX_NA		= 0x8002,
+	MLXSW_REG_MTBR_BAD_SENS_INFO	= 0x8003,
+};
+
+/* Base index for reading modules temperature */
+#define MLXSW_REG_MTBR_BASE_MODULE_INDEX 64
+
+static inline void mlxsw_reg_mtbr_temp_unpack(char *payload, int rec_ind,
+					      u16 *p_temp, u16 *p_max_temp)
+{
+	if (p_temp)
+		*p_temp = mlxsw_reg_mtbr_rec_temp_get(payload, rec_ind);
+	if (p_max_temp)
+		*p_max_temp = mlxsw_reg_mtbr_rec_max_temp_get(payload, rec_ind);
+}
+
+/* MCIA - Management Cable Info Access
+ * -----------------------------------
+ * MCIA register is used to access the SFP+ and QSFP connector's EPROM.
+ */
+
+#define MLXSW_REG_MCIA_ID 0x9014
+#define MLXSW_REG_MCIA_LEN 0x40
+
+MLXSW_REG_DEFINE(mcia, MLXSW_REG_MCIA_ID, MLXSW_REG_MCIA_LEN);
+
+/* reg_mcia_l
+ * Lock bit. Setting this bit will lock the access to the specific
+ * cable. Used for updating a full page in a cable EPROM. Any access
+ * other then subsequence writes will fail while the port is locked.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mcia, l, 0x00, 31, 1);
+
+/* reg_mcia_module
+ * Module number.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, mcia, module, 0x00, 16, 8);
+
+/* reg_mcia_slot_index
+ * Slot index (0: Main board)
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, mcia, slot, 0x00, 12, 4);
+
+enum {
+	MLXSW_REG_MCIA_STATUS_GOOD = 0,
+	/* No response from module's EEPROM. */
+	MLXSW_REG_MCIA_STATUS_NO_EEPROM_MODULE = 1,
+	/* Module type not supported by the device. */
+	MLXSW_REG_MCIA_STATUS_MODULE_NOT_SUPPORTED = 2,
+	/* No module present indication. */
+	MLXSW_REG_MCIA_STATUS_MODULE_NOT_CONNECTED = 3,
+	/* Error occurred while trying to access module's EEPROM using I2C. */
+	MLXSW_REG_MCIA_STATUS_I2C_ERROR = 9,
+	/* Module is disabled. */
+	MLXSW_REG_MCIA_STATUS_MODULE_DISABLED = 16,
+};
+
+/* reg_mcia_status
+ * Module status.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mcia, status, 0x00, 0, 8);
+
+/* reg_mcia_i2c_device_address
+ * I2C device address.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mcia, i2c_device_address, 0x04, 24, 8);
+
+/* reg_mcia_page_number
+ * Page number.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mcia, page_number, 0x04, 16, 8);
+
+/* reg_mcia_device_address
+ * Device address.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mcia, device_address, 0x04, 0, 16);
+
+/* reg_mcia_bank_number
+ * Bank number.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, mcia, bank_number, 0x08, 16, 8);
+
+/* reg_mcia_size
+ * Number of bytes to read/write (up to 48 bytes).
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mcia, size, 0x08, 0, 16);
+
+#define MLXSW_REG_MCIA_EEPROM_PAGE_LENGTH	256
+#define MLXSW_REG_MCIA_EEPROM_UP_PAGE_LENGTH	128
+#define MLXSW_REG_MCIA_EEPROM_SIZE		48
+#define MLXSW_REG_MCIA_I2C_ADDR_LOW		0x50
+#define MLXSW_REG_MCIA_I2C_ADDR_HIGH		0x51
+#define MLXSW_REG_MCIA_PAGE0_LO_OFF		0xa0
+#define MLXSW_REG_MCIA_TH_ITEM_SIZE		2
+#define MLXSW_REG_MCIA_TH_PAGE_NUM		3
+#define MLXSW_REG_MCIA_TH_PAGE_CMIS_NUM		2
+#define MLXSW_REG_MCIA_PAGE0_LO			0
+#define MLXSW_REG_MCIA_TH_PAGE_OFF		0x80
+#define MLXSW_REG_MCIA_EEPROM_CMIS_FLAT_MEMORY	BIT(7)
+
+enum mlxsw_reg_mcia_eeprom_module_info_rev_id {
+	MLXSW_REG_MCIA_EEPROM_MODULE_INFO_REV_ID_UNSPC	= 0x00,
+	MLXSW_REG_MCIA_EEPROM_MODULE_INFO_REV_ID_8436	= 0x01,
+	MLXSW_REG_MCIA_EEPROM_MODULE_INFO_REV_ID_8636	= 0x03,
+};
+
+enum mlxsw_reg_mcia_eeprom_module_info_id {
+	MLXSW_REG_MCIA_EEPROM_MODULE_INFO_ID_SFP	= 0x03,
+	MLXSW_REG_MCIA_EEPROM_MODULE_INFO_ID_QSFP	= 0x0C,
+	MLXSW_REG_MCIA_EEPROM_MODULE_INFO_ID_QSFP_PLUS	= 0x0D,
+	MLXSW_REG_MCIA_EEPROM_MODULE_INFO_ID_QSFP28	= 0x11,
+	MLXSW_REG_MCIA_EEPROM_MODULE_INFO_ID_QSFP_DD	= 0x18,
+	MLXSW_REG_MCIA_EEPROM_MODULE_INFO_ID_OSFP	= 0x19,
+};
+
+enum mlxsw_reg_mcia_eeprom_module_info {
+	MLXSW_REG_MCIA_EEPROM_MODULE_INFO_ID,
+	MLXSW_REG_MCIA_EEPROM_MODULE_INFO_REV_ID,
+	MLXSW_REG_MCIA_EEPROM_MODULE_INFO_TYPE_ID,
+	MLXSW_REG_MCIA_EEPROM_MODULE_INFO_SIZE,
+};
+
+/* reg_mcia_eeprom
+ * Bytes to read/write.
+ * Access: RW
+ */
+MLXSW_ITEM_BUF(reg, mcia, eeprom, 0x10, MLXSW_REG_MCIA_EEPROM_SIZE);
+
+/* This is used to access the optional upper pages (1-3) in the QSFP+
+ * memory map. Page 1 is available on offset 256 through 383, page 2 -
+ * on offset 384 through 511, page 3 - on offset 512 through 639.
+ */
+#define MLXSW_REG_MCIA_PAGE_GET(off) (((off) - \
+				MLXSW_REG_MCIA_EEPROM_PAGE_LENGTH) / \
+				MLXSW_REG_MCIA_EEPROM_UP_PAGE_LENGTH + 1)
+
+static inline void mlxsw_reg_mcia_pack(char *payload, u8 slot_index, u8 module,
+				       u8 lock, u8 page_number,
+				       u16 device_addr, u8 size,
+				       u8 i2c_device_addr)
+{
+	MLXSW_REG_ZERO(mcia, payload);
+	mlxsw_reg_mcia_slot_set(payload, slot_index);
+	mlxsw_reg_mcia_module_set(payload, module);
+	mlxsw_reg_mcia_l_set(payload, lock);
+	mlxsw_reg_mcia_page_number_set(payload, page_number);
+	mlxsw_reg_mcia_device_address_set(payload, device_addr);
+	mlxsw_reg_mcia_size_set(payload, size);
+	mlxsw_reg_mcia_i2c_device_address_set(payload, i2c_device_addr);
+}
+
+/* MPAT - Monitoring Port Analyzer Table
+ * -------------------------------------
+ * MPAT Register is used to query and configure the Switch PortAnalyzer Table.
+ * For an enabled analyzer, all fields except e (enable) cannot be modified.
+ */
+#define MLXSW_REG_MPAT_ID 0x901A
+#define MLXSW_REG_MPAT_LEN 0x78
+
+MLXSW_REG_DEFINE(mpat, MLXSW_REG_MPAT_ID, MLXSW_REG_MPAT_LEN);
+
+/* reg_mpat_pa_id
+ * Port Analyzer ID.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, mpat, pa_id, 0x00, 28, 4);
+
+/* reg_mpat_session_id
+ * Mirror Session ID.
+ * Used for MIRROR_SESSION<i> trap.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mpat, session_id, 0x00, 24, 4);
+
+/* reg_mpat_system_port
+ * A unique port identifier for the final destination of the packet.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mpat, system_port, 0x00, 0, 16);
+
+/* reg_mpat_e
+ * Enable. Indicating the Port Analyzer is enabled.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mpat, e, 0x04, 31, 1);
+
+/* reg_mpat_qos
+ * Quality Of Service Mode.
+ * 0: CONFIGURED - QoS parameters (Switch Priority, and encapsulation
+ * PCP, DEI, DSCP or VL) are configured.
+ * 1: MAINTAIN - QoS parameters (Switch Priority, Color) are the
+ * same as in the original packet that has triggered the mirroring. For
+ * SPAN also the pcp,dei are maintained.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mpat, qos, 0x04, 26, 1);
+
+/* reg_mpat_be
+ * Best effort mode. Indicates mirroring traffic should not cause packet
+ * drop or back pressure, but will discard the mirrored packets. Mirrored
+ * packets will be forwarded on a best effort manner.
+ * 0: Do not discard mirrored packets
+ * 1: Discard mirrored packets if causing congestion
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mpat, be, 0x04, 25, 1);
+
+enum mlxsw_reg_mpat_span_type {
+	/* Local SPAN Ethernet.
+	 * The original packet is not encapsulated.
+	 */
+	MLXSW_REG_MPAT_SPAN_TYPE_LOCAL_ETH = 0x0,
+
+	/* Remote SPAN Ethernet VLAN.
+	 * The packet is forwarded to the monitoring port on the monitoring
+	 * VLAN.
+	 */
+	MLXSW_REG_MPAT_SPAN_TYPE_REMOTE_ETH = 0x1,
+
+	/* Encapsulated Remote SPAN Ethernet L3 GRE.
+	 * The packet is encapsulated with GRE header.
+	 */
+	MLXSW_REG_MPAT_SPAN_TYPE_REMOTE_ETH_L3 = 0x3,
+};
+
+/* reg_mpat_span_type
+ * SPAN type.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mpat, span_type, 0x04, 0, 4);
+
+/* reg_mpat_pide
+ * Policer enable.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mpat, pide, 0x0C, 15, 1);
+
+/* reg_mpat_pid
+ * Policer ID.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mpat, pid, 0x0C, 0, 14);
+
+/* Remote SPAN - Ethernet VLAN
+ * - - - - - - - - - - - - - -
+ */
+
+/* reg_mpat_eth_rspan_vid
+ * Encapsulation header VLAN ID.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mpat, eth_rspan_vid, 0x18, 0, 12);
+
+/* Encapsulated Remote SPAN - Ethernet L2
+ * - - - - - - - - - - - - - - - - - - -
+ */
+
+enum mlxsw_reg_mpat_eth_rspan_version {
+	MLXSW_REG_MPAT_ETH_RSPAN_VERSION_NO_HEADER = 15,
+};
+
+/* reg_mpat_eth_rspan_version
+ * RSPAN mirror header version.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mpat, eth_rspan_version, 0x10, 18, 4);
+
+/* reg_mpat_eth_rspan_mac
+ * Destination MAC address.
+ * Access: RW
+ */
+MLXSW_ITEM_BUF(reg, mpat, eth_rspan_mac, 0x12, 6);
+
+/* reg_mpat_eth_rspan_tp
+ * Tag Packet. Indicates whether the mirroring header should be VLAN tagged.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mpat, eth_rspan_tp, 0x18, 16, 1);
+
+/* Encapsulated Remote SPAN - Ethernet L3
+ * - - - - - - - - - - - - - - - - - - -
+ */
+
+enum mlxsw_reg_mpat_eth_rspan_protocol {
+	MLXSW_REG_MPAT_ETH_RSPAN_PROTOCOL_IPV4,
+	MLXSW_REG_MPAT_ETH_RSPAN_PROTOCOL_IPV6,
+};
+
+/* reg_mpat_eth_rspan_protocol
+ * SPAN encapsulation protocol.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mpat, eth_rspan_protocol, 0x18, 24, 4);
+
+/* reg_mpat_eth_rspan_ttl
+ * Encapsulation header Time-to-Live/HopLimit.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mpat, eth_rspan_ttl, 0x1C, 4, 8);
+
+/* reg_mpat_eth_rspan_smac
+ * Source MAC address
+ * Access: RW
+ */
+MLXSW_ITEM_BUF(reg, mpat, eth_rspan_smac, 0x22, 6);
+
+/* reg_mpat_eth_rspan_dip*
+ * Destination IP address. The IP version is configured by protocol.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mpat, eth_rspan_dip4, 0x4C, 0, 32);
+MLXSW_ITEM_BUF(reg, mpat, eth_rspan_dip6, 0x40, 16);
+
+/* reg_mpat_eth_rspan_sip*
+ * Source IP address. The IP version is configured by protocol.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mpat, eth_rspan_sip4, 0x5C, 0, 32);
+MLXSW_ITEM_BUF(reg, mpat, eth_rspan_sip6, 0x50, 16);
+
+static inline void mlxsw_reg_mpat_pack(char *payload, u8 pa_id,
+				       u16 system_port, bool e,
+				       enum mlxsw_reg_mpat_span_type span_type)
+{
+	MLXSW_REG_ZERO(mpat, payload);
+	mlxsw_reg_mpat_pa_id_set(payload, pa_id);
+	mlxsw_reg_mpat_system_port_set(payload, system_port);
+	mlxsw_reg_mpat_e_set(payload, e);
+	mlxsw_reg_mpat_qos_set(payload, 1);
+	mlxsw_reg_mpat_be_set(payload, 1);
+	mlxsw_reg_mpat_span_type_set(payload, span_type);
+}
+
+static inline void mlxsw_reg_mpat_eth_rspan_pack(char *payload, u16 vid)
+{
+	mlxsw_reg_mpat_eth_rspan_vid_set(payload, vid);
+}
+
+static inline void
+mlxsw_reg_mpat_eth_rspan_l2_pack(char *payload,
+				 enum mlxsw_reg_mpat_eth_rspan_version version,
+				 const char *mac,
+				 bool tp)
+{
+	mlxsw_reg_mpat_eth_rspan_version_set(payload, version);
+	mlxsw_reg_mpat_eth_rspan_mac_memcpy_to(payload, mac);
+	mlxsw_reg_mpat_eth_rspan_tp_set(payload, tp);
+}
+
+static inline void
+mlxsw_reg_mpat_eth_rspan_l3_ipv4_pack(char *payload, u8 ttl,
+				      const char *smac,
+				      u32 sip, u32 dip)
+{
+	mlxsw_reg_mpat_eth_rspan_ttl_set(payload, ttl);
+	mlxsw_reg_mpat_eth_rspan_smac_memcpy_to(payload, smac);
+	mlxsw_reg_mpat_eth_rspan_protocol_set(payload,
+				    MLXSW_REG_MPAT_ETH_RSPAN_PROTOCOL_IPV4);
+	mlxsw_reg_mpat_eth_rspan_sip4_set(payload, sip);
+	mlxsw_reg_mpat_eth_rspan_dip4_set(payload, dip);
+}
+
+static inline void
+mlxsw_reg_mpat_eth_rspan_l3_ipv6_pack(char *payload, u8 ttl,
+				      const char *smac,
+				      struct in6_addr sip, struct in6_addr dip)
+{
+	mlxsw_reg_mpat_eth_rspan_ttl_set(payload, ttl);
+	mlxsw_reg_mpat_eth_rspan_smac_memcpy_to(payload, smac);
+	mlxsw_reg_mpat_eth_rspan_protocol_set(payload,
+				    MLXSW_REG_MPAT_ETH_RSPAN_PROTOCOL_IPV6);
+	mlxsw_reg_mpat_eth_rspan_sip6_memcpy_to(payload, (void *)&sip);
+	mlxsw_reg_mpat_eth_rspan_dip6_memcpy_to(payload, (void *)&dip);
+}
+
+/* MPAR - Monitoring Port Analyzer Register
+ * ----------------------------------------
+ * MPAR register is used to query and configure the port analyzer port mirroring
+ * properties.
+ */
+#define MLXSW_REG_MPAR_ID 0x901B
+#define MLXSW_REG_MPAR_LEN 0x0C
+
+MLXSW_REG_DEFINE(mpar, MLXSW_REG_MPAR_ID, MLXSW_REG_MPAR_LEN);
+
+/* reg_mpar_local_port
+ * The local port to mirror the packets from.
+ * Access: Index
+ */
+MLXSW_ITEM32_LP(reg, mpar, 0x00, 16, 0x00, 4);
+
+enum mlxsw_reg_mpar_i_e {
+	MLXSW_REG_MPAR_TYPE_EGRESS,
+	MLXSW_REG_MPAR_TYPE_INGRESS,
+};
+
+/* reg_mpar_i_e
+ * Ingress/Egress
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, mpar, i_e, 0x00, 0, 4);
+
+/* reg_mpar_enable
+ * Enable mirroring
+ * By default, port mirroring is disabled for all ports.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mpar, enable, 0x04, 31, 1);
+
+/* reg_mpar_pa_id
+ * Port Analyzer ID.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mpar, pa_id, 0x04, 0, 4);
+
+#define MLXSW_REG_MPAR_RATE_MAX 3500000000UL
+
+/* reg_mpar_probability_rate
+ * Sampling rate.
+ * Valid values are: 1 to 3.5*10^9
+ * Value of 1 means "sample all". Default is 1.
+ * Reserved when Spectrum-1.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mpar, probability_rate, 0x08, 0, 32);
+
+static inline void mlxsw_reg_mpar_pack(char *payload, u16 local_port,
+				       enum mlxsw_reg_mpar_i_e i_e,
+				       bool enable, u8 pa_id,
+				       u32 probability_rate)
+{
+	MLXSW_REG_ZERO(mpar, payload);
+	mlxsw_reg_mpar_local_port_set(payload, local_port);
+	mlxsw_reg_mpar_enable_set(payload, enable);
+	mlxsw_reg_mpar_i_e_set(payload, i_e);
+	mlxsw_reg_mpar_pa_id_set(payload, pa_id);
+	mlxsw_reg_mpar_probability_rate_set(payload, probability_rate);
+}
+
+/* MGIR - Management General Information Register
+ * ----------------------------------------------
+ * MGIR register allows software to query the hardware and firmware general
+ * information.
+ */
+#define MLXSW_REG_MGIR_ID 0x9020
+#define MLXSW_REG_MGIR_LEN 0x9C
+
+MLXSW_REG_DEFINE(mgir, MLXSW_REG_MGIR_ID, MLXSW_REG_MGIR_LEN);
+
+/* reg_mgir_hw_info_device_hw_revision
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mgir, hw_info_device_hw_revision, 0x0, 16, 16);
+
+#define MLXSW_REG_MGIR_FW_INFO_PSID_SIZE 16
+
+/* reg_mgir_fw_info_psid
+ * PSID (ASCII string).
+ * Access: RO
+ */
+MLXSW_ITEM_BUF(reg, mgir, fw_info_psid, 0x30, MLXSW_REG_MGIR_FW_INFO_PSID_SIZE);
+
+/* reg_mgir_fw_info_extended_major
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mgir, fw_info_extended_major, 0x44, 0, 32);
+
+/* reg_mgir_fw_info_extended_minor
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mgir, fw_info_extended_minor, 0x48, 0, 32);
+
+/* reg_mgir_fw_info_extended_sub_minor
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mgir, fw_info_extended_sub_minor, 0x4C, 0, 32);
+
+static inline void mlxsw_reg_mgir_pack(char *payload)
+{
+	MLXSW_REG_ZERO(mgir, payload);
+}
+
+static inline void
+mlxsw_reg_mgir_unpack(char *payload, u32 *hw_rev, char *fw_info_psid,
+		      u32 *fw_major, u32 *fw_minor, u32 *fw_sub_minor)
+{
+	*hw_rev = mlxsw_reg_mgir_hw_info_device_hw_revision_get(payload);
+	mlxsw_reg_mgir_fw_info_psid_memcpy_from(payload, fw_info_psid);
+	*fw_major = mlxsw_reg_mgir_fw_info_extended_major_get(payload);
+	*fw_minor = mlxsw_reg_mgir_fw_info_extended_minor_get(payload);
+	*fw_sub_minor = mlxsw_reg_mgir_fw_info_extended_sub_minor_get(payload);
+}
+
+/* MRSR - Management Reset and Shutdown Register
+ * ---------------------------------------------
+ * MRSR register is used to reset or shutdown the switch or
+ * the entire system (when applicable).
+ */
+#define MLXSW_REG_MRSR_ID 0x9023
+#define MLXSW_REG_MRSR_LEN 0x08
+
+MLXSW_REG_DEFINE(mrsr, MLXSW_REG_MRSR_ID, MLXSW_REG_MRSR_LEN);
+
+/* reg_mrsr_command
+ * Reset/shutdown command
+ * 0 - do nothing
+ * 1 - software reset
+ * Access: WO
+ */
+MLXSW_ITEM32(reg, mrsr, command, 0x00, 0, 4);
+
+static inline void mlxsw_reg_mrsr_pack(char *payload)
+{
+	MLXSW_REG_ZERO(mrsr, payload);
+	mlxsw_reg_mrsr_command_set(payload, 1);
+}
+
+/* MLCR - Management LED Control Register
+ * --------------------------------------
+ * Controls the system LEDs.
+ */
+#define MLXSW_REG_MLCR_ID 0x902B
+#define MLXSW_REG_MLCR_LEN 0x0C
+
+MLXSW_REG_DEFINE(mlcr, MLXSW_REG_MLCR_ID, MLXSW_REG_MLCR_LEN);
+
+/* reg_mlcr_local_port
+ * Local port number.
+ * Access: RW
+ */
+MLXSW_ITEM32_LP(reg, mlcr, 0x00, 16, 0x00, 24);
+
+#define MLXSW_REG_MLCR_DURATION_MAX 0xFFFF
+
+/* reg_mlcr_beacon_duration
+ * Duration of the beacon to be active, in seconds.
+ * 0x0 - Will turn off the beacon.
+ * 0xFFFF - Will turn on the beacon until explicitly turned off.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mlcr, beacon_duration, 0x04, 0, 16);
+
+/* reg_mlcr_beacon_remain
+ * Remaining duration of the beacon, in seconds.
+ * 0xFFFF indicates an infinite amount of time.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mlcr, beacon_remain, 0x08, 0, 16);
+
+static inline void mlxsw_reg_mlcr_pack(char *payload, u16 local_port,
+				       bool active)
+{
+	MLXSW_REG_ZERO(mlcr, payload);
+	mlxsw_reg_mlcr_local_port_set(payload, local_port);
+	mlxsw_reg_mlcr_beacon_duration_set(payload, active ?
+					   MLXSW_REG_MLCR_DURATION_MAX : 0);
+}
+
+/* MCION - Management Cable IO and Notifications Register
+ * ------------------------------------------------------
+ * The MCION register is used to query transceiver modules' IO pins and other
+ * notifications.
+ */
+#define MLXSW_REG_MCION_ID 0x9052
+#define MLXSW_REG_MCION_LEN 0x18
+
+MLXSW_REG_DEFINE(mcion, MLXSW_REG_MCION_ID, MLXSW_REG_MCION_LEN);
+
+/* reg_mcion_module
+ * Module number.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, mcion, module, 0x00, 16, 8);
+
+/* reg_mcion_slot_index
+ * Slot index.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, mcion, slot_index, 0x00, 12, 4);
+
+enum {
+	MLXSW_REG_MCION_MODULE_STATUS_BITS_PRESENT_MASK = BIT(0),
+	MLXSW_REG_MCION_MODULE_STATUS_BITS_LOW_POWER_MASK = BIT(8),
+};
+
+/* reg_mcion_module_status_bits
+ * Module IO status as defined by SFF.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mcion, module_status_bits, 0x04, 0, 16);
+
+static inline void mlxsw_reg_mcion_pack(char *payload, u8 slot_index, u8 module)
+{
+	MLXSW_REG_ZERO(mcion, payload);
+	mlxsw_reg_mcion_slot_index_set(payload, slot_index);
+	mlxsw_reg_mcion_module_set(payload, module);
+}
+
+/* MTPPS - Management Pulse Per Second Register
+ * --------------------------------------------
+ * This register provides the device PPS capabilities, configure the PPS in and
+ * out modules and holds the PPS in time stamp.
+ */
+#define MLXSW_REG_MTPPS_ID 0x9053
+#define MLXSW_REG_MTPPS_LEN 0x3C
+
+MLXSW_REG_DEFINE(mtpps, MLXSW_REG_MTPPS_ID, MLXSW_REG_MTPPS_LEN);
+
+/* reg_mtpps_enable
+ * Enables the PPS functionality the specific pin.
+ * A boolean variable.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mtpps, enable, 0x20, 31, 1);
+
+enum mlxsw_reg_mtpps_pin_mode {
+	MLXSW_REG_MTPPS_PIN_MODE_VIRTUAL_PIN = 0x2,
+};
+
+/* reg_mtpps_pin_mode
+ * Pin mode to be used. The mode must comply with the supported modes of the
+ * requested pin.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mtpps, pin_mode, 0x20, 8, 4);
+
+#define MLXSW_REG_MTPPS_PIN_SP_VIRTUAL_PIN	7
+
+/* reg_mtpps_pin
+ * Pin to be configured or queried out of the supported pins.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, mtpps, pin, 0x20, 0, 8);
+
+/* reg_mtpps_time_stamp
+ * When pin_mode = pps_in, the latched device time when it was triggered from
+ * the external GPIO pin.
+ * When pin_mode = pps_out or virtual_pin or pps_out_and_virtual_pin, the target
+ * time to generate next output signal.
+ * Time is in units of device clock.
+ * Access: RW
+ */
+MLXSW_ITEM64(reg, mtpps, time_stamp, 0x28, 0, 64);
+
+static inline void
+mlxsw_reg_mtpps_vpin_pack(char *payload, u64 time_stamp)
+{
+	MLXSW_REG_ZERO(mtpps, payload);
+	mlxsw_reg_mtpps_pin_set(payload, MLXSW_REG_MTPPS_PIN_SP_VIRTUAL_PIN);
+	mlxsw_reg_mtpps_pin_mode_set(payload,
+				     MLXSW_REG_MTPPS_PIN_MODE_VIRTUAL_PIN);
+	mlxsw_reg_mtpps_enable_set(payload, true);
+	mlxsw_reg_mtpps_time_stamp_set(payload, time_stamp);
+}
+
+/* MTUTC - Management UTC Register
+ * -------------------------------
+ * Configures the HW UTC counter.
+ */
+#define MLXSW_REG_MTUTC_ID 0x9055
+#define MLXSW_REG_MTUTC_LEN 0x1C
+
+MLXSW_REG_DEFINE(mtutc, MLXSW_REG_MTUTC_ID, MLXSW_REG_MTUTC_LEN);
+
+enum mlxsw_reg_mtutc_operation {
+	MLXSW_REG_MTUTC_OPERATION_SET_TIME_AT_NEXT_SEC = 0,
+	MLXSW_REG_MTUTC_OPERATION_SET_TIME_IMMEDIATE = 1,
+	MLXSW_REG_MTUTC_OPERATION_ADJUST_TIME = 2,
+	MLXSW_REG_MTUTC_OPERATION_ADJUST_FREQ = 3,
+};
+
+/* reg_mtutc_operation
+ * Operation.
+ * Access: OP
+ */
+MLXSW_ITEM32(reg, mtutc, operation, 0x00, 0, 4);
+
+/* reg_mtutc_freq_adjustment
+ * Frequency adjustment: Every PPS the HW frequency will be
+ * adjusted by this value. Units of HW clock, where HW counts
+ * 10^9 HW clocks for 1 HW second. Range is from -50,000,000 to +50,000,000.
+ * In Spectrum-2, the field is reversed, positive values mean to decrease the
+ * frequency.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mtutc, freq_adjustment, 0x04, 0, 32);
+
+#define MLXSW_REG_MTUTC_MAX_FREQ_ADJ (50 * 1000 * 1000)
+
+/* reg_mtutc_utc_sec
+ * UTC seconds.
+ * Access: WO
+ */
+MLXSW_ITEM32(reg, mtutc, utc_sec, 0x10, 0, 32);
+
+/* reg_mtutc_utc_nsec
+ * UTC nSecs.
+ * Range 0..(10^9-1)
+ * Updated when operation is SET_TIME_IMMEDIATE.
+ * Reserved on Spectrum-1.
+ * Access: WO
+ */
+MLXSW_ITEM32(reg, mtutc, utc_nsec, 0x14, 0, 30);
+
+/* reg_mtutc_time_adjustment
+ * Time adjustment.
+ * Units of nSec.
+ * Range is from -32768 to +32767.
+ * Updated when operation is ADJUST_TIME.
+ * Reserved on Spectrum-1.
+ * Access: WO
+ */
+MLXSW_ITEM32(reg, mtutc, time_adjustment, 0x18, 0, 32);
+
+static inline void
+mlxsw_reg_mtutc_pack(char *payload, enum mlxsw_reg_mtutc_operation oper,
+		     u32 freq_adj, u32 utc_sec, u32 utc_nsec, u32 time_adj)
+{
+	MLXSW_REG_ZERO(mtutc, payload);
+	mlxsw_reg_mtutc_operation_set(payload, oper);
+	mlxsw_reg_mtutc_freq_adjustment_set(payload, freq_adj);
+	mlxsw_reg_mtutc_utc_sec_set(payload, utc_sec);
+	mlxsw_reg_mtutc_utc_nsec_set(payload, utc_nsec);
+	mlxsw_reg_mtutc_time_adjustment_set(payload, time_adj);
+}
+
+/* MCQI - Management Component Query Information
+ * ---------------------------------------------
+ * This register allows querying information about firmware components.
+ */
+#define MLXSW_REG_MCQI_ID 0x9061
+#define MLXSW_REG_MCQI_BASE_LEN 0x18
+#define MLXSW_REG_MCQI_CAP_LEN 0x14
+#define MLXSW_REG_MCQI_LEN (MLXSW_REG_MCQI_BASE_LEN + MLXSW_REG_MCQI_CAP_LEN)
+
+MLXSW_REG_DEFINE(mcqi, MLXSW_REG_MCQI_ID, MLXSW_REG_MCQI_LEN);
+
+/* reg_mcqi_component_index
+ * Index of the accessed component.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, mcqi, component_index, 0x00, 0, 16);
+
+enum mlxfw_reg_mcqi_info_type {
+	MLXSW_REG_MCQI_INFO_TYPE_CAPABILITIES,
+};
+
+/* reg_mcqi_info_type
+ * Component properties set.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mcqi, info_type, 0x08, 0, 5);
+
+/* reg_mcqi_offset
+ * The requested/returned data offset from the section start, given in bytes.
+ * Must be DWORD aligned.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mcqi, offset, 0x10, 0, 32);
+
+/* reg_mcqi_data_size
+ * The requested/returned data size, given in bytes. If data_size is not DWORD
+ * aligned, the last bytes are zero padded.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mcqi, data_size, 0x14, 0, 16);
+
+/* reg_mcqi_cap_max_component_size
+ * Maximum size for this component, given in bytes.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mcqi, cap_max_component_size, 0x20, 0, 32);
+
+/* reg_mcqi_cap_log_mcda_word_size
+ * Log 2 of the access word size in bytes. Read and write access must be aligned
+ * to the word size. Write access must be done for an integer number of words.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mcqi, cap_log_mcda_word_size, 0x24, 28, 4);
+
+/* reg_mcqi_cap_mcda_max_write_size
+ * Maximal write size for MCDA register
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mcqi, cap_mcda_max_write_size, 0x24, 0, 16);
+
+static inline void mlxsw_reg_mcqi_pack(char *payload, u16 component_index)
+{
+	MLXSW_REG_ZERO(mcqi, payload);
+	mlxsw_reg_mcqi_component_index_set(payload, component_index);
+	mlxsw_reg_mcqi_info_type_set(payload,
+				     MLXSW_REG_MCQI_INFO_TYPE_CAPABILITIES);
+	mlxsw_reg_mcqi_offset_set(payload, 0);
+	mlxsw_reg_mcqi_data_size_set(payload, MLXSW_REG_MCQI_CAP_LEN);
+}
+
+static inline void mlxsw_reg_mcqi_unpack(char *payload,
+					 u32 *p_cap_max_component_size,
+					 u8 *p_cap_log_mcda_word_size,
+					 u16 *p_cap_mcda_max_write_size)
+{
+	*p_cap_max_component_size =
+		mlxsw_reg_mcqi_cap_max_component_size_get(payload);
+	*p_cap_log_mcda_word_size =
+		mlxsw_reg_mcqi_cap_log_mcda_word_size_get(payload);
+	*p_cap_mcda_max_write_size =
+		mlxsw_reg_mcqi_cap_mcda_max_write_size_get(payload);
+}
+
+/* MCC - Management Component Control
+ * ----------------------------------
+ * Controls the firmware component and updates the FSM.
+ */
+#define MLXSW_REG_MCC_ID 0x9062
+#define MLXSW_REG_MCC_LEN 0x1C
+
+MLXSW_REG_DEFINE(mcc, MLXSW_REG_MCC_ID, MLXSW_REG_MCC_LEN);
+
+enum mlxsw_reg_mcc_instruction {
+	MLXSW_REG_MCC_INSTRUCTION_LOCK_UPDATE_HANDLE = 0x01,
+	MLXSW_REG_MCC_INSTRUCTION_RELEASE_UPDATE_HANDLE = 0x02,
+	MLXSW_REG_MCC_INSTRUCTION_UPDATE_COMPONENT = 0x03,
+	MLXSW_REG_MCC_INSTRUCTION_VERIFY_COMPONENT = 0x04,
+	MLXSW_REG_MCC_INSTRUCTION_ACTIVATE = 0x06,
+	MLXSW_REG_MCC_INSTRUCTION_CANCEL = 0x08,
+};
+
+/* reg_mcc_instruction
+ * Command to be executed by the FSM.
+ * Applicable for write operation only.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mcc, instruction, 0x00, 0, 8);
+
+/* reg_mcc_component_index
+ * Index of the accessed component. Applicable only for commands that
+ * refer to components. Otherwise, this field is reserved.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, mcc, component_index, 0x04, 0, 16);
+
+/* reg_mcc_update_handle
+ * Token representing the current flow executed by the FSM.
+ * Access: WO
+ */
+MLXSW_ITEM32(reg, mcc, update_handle, 0x08, 0, 24);
+
+/* reg_mcc_error_code
+ * Indicates the successful completion of the instruction, or the reason it
+ * failed
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mcc, error_code, 0x0C, 8, 8);
+
+/* reg_mcc_control_state
+ * Current FSM state
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mcc, control_state, 0x0C, 0, 4);
+
+/* reg_mcc_component_size
+ * Component size in bytes. Valid for UPDATE_COMPONENT instruction. Specifying
+ * the size may shorten the update time. Value 0x0 means that size is
+ * unspecified.
+ * Access: WO
+ */
+MLXSW_ITEM32(reg, mcc, component_size, 0x10, 0, 32);
+
+static inline void mlxsw_reg_mcc_pack(char *payload,
+				      enum mlxsw_reg_mcc_instruction instr,
+				      u16 component_index, u32 update_handle,
+				      u32 component_size)
+{
+	MLXSW_REG_ZERO(mcc, payload);
+	mlxsw_reg_mcc_instruction_set(payload, instr);
+	mlxsw_reg_mcc_component_index_set(payload, component_index);
+	mlxsw_reg_mcc_update_handle_set(payload, update_handle);
+	mlxsw_reg_mcc_component_size_set(payload, component_size);
+}
+
+static inline void mlxsw_reg_mcc_unpack(char *payload, u32 *p_update_handle,
+					u8 *p_error_code, u8 *p_control_state)
+{
+	if (p_update_handle)
+		*p_update_handle = mlxsw_reg_mcc_update_handle_get(payload);
+	if (p_error_code)
+		*p_error_code = mlxsw_reg_mcc_error_code_get(payload);
+	if (p_control_state)
+		*p_control_state = mlxsw_reg_mcc_control_state_get(payload);
+}
+
+/* MCDA - Management Component Data Access
+ * ---------------------------------------
+ * This register allows reading and writing a firmware component.
+ */
+#define MLXSW_REG_MCDA_ID 0x9063
+#define MLXSW_REG_MCDA_BASE_LEN 0x10
+#define MLXSW_REG_MCDA_MAX_DATA_LEN 0x80
+#define MLXSW_REG_MCDA_LEN \
+		(MLXSW_REG_MCDA_BASE_LEN + MLXSW_REG_MCDA_MAX_DATA_LEN)
+
+MLXSW_REG_DEFINE(mcda, MLXSW_REG_MCDA_ID, MLXSW_REG_MCDA_LEN);
+
+/* reg_mcda_update_handle
+ * Token representing the current flow executed by the FSM.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mcda, update_handle, 0x00, 0, 24);
+
+/* reg_mcda_offset
+ * Offset of accessed address relative to component start. Accesses must be in
+ * accordance to log_mcda_word_size in MCQI reg.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mcda, offset, 0x04, 0, 32);
+
+/* reg_mcda_size
+ * Size of the data accessed, given in bytes.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mcda, size, 0x08, 0, 16);
+
+/* reg_mcda_data
+ * Data block accessed.
+ * Access: RW
+ */
+MLXSW_ITEM32_INDEXED(reg, mcda, data, 0x10, 0, 32, 4, 0, false);
+
+static inline void mlxsw_reg_mcda_pack(char *payload, u32 update_handle,
+				       u32 offset, u16 size, u8 *data)
+{
+	int i;
+
+	MLXSW_REG_ZERO(mcda, payload);
+	mlxsw_reg_mcda_update_handle_set(payload, update_handle);
+	mlxsw_reg_mcda_offset_set(payload, offset);
+	mlxsw_reg_mcda_size_set(payload, size);
+
+	for (i = 0; i < size / 4; i++)
+		mlxsw_reg_mcda_data_set(payload, i, *(u32 *) &data[i * 4]);
+}
+
+/* MPSC - Monitoring Packet Sampling Configuration Register
+ * --------------------------------------------------------
+ * MPSC Register is used to configure the Packet Sampling mechanism.
+ */
+#define MLXSW_REG_MPSC_ID 0x9080
+#define MLXSW_REG_MPSC_LEN 0x1C
+
+MLXSW_REG_DEFINE(mpsc, MLXSW_REG_MPSC_ID, MLXSW_REG_MPSC_LEN);
+
+/* reg_mpsc_local_port
+ * Local port number
+ * Not supported for CPU port
+ * Access: Index
+ */
+MLXSW_ITEM32_LP(reg, mpsc, 0x00, 16, 0x00, 12);
+
+/* reg_mpsc_e
+ * Enable sampling on port local_port
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mpsc, e, 0x04, 30, 1);
+
+#define MLXSW_REG_MPSC_RATE_MAX 3500000000UL
+
+/* reg_mpsc_rate
+ * Sampling rate = 1 out of rate packets (with randomization around
+ * the point). Valid values are: 1 to MLXSW_REG_MPSC_RATE_MAX
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mpsc, rate, 0x08, 0, 32);
+
+static inline void mlxsw_reg_mpsc_pack(char *payload, u16 local_port, bool e,
+				       u32 rate)
+{
+	MLXSW_REG_ZERO(mpsc, payload);
+	mlxsw_reg_mpsc_local_port_set(payload, local_port);
+	mlxsw_reg_mpsc_e_set(payload, e);
+	mlxsw_reg_mpsc_rate_set(payload, rate);
+}
+
+/* MGPC - Monitoring General Purpose Counter Set Register
+ * The MGPC register retrieves and sets the General Purpose Counter Set.
+ */
+#define MLXSW_REG_MGPC_ID 0x9081
+#define MLXSW_REG_MGPC_LEN 0x18
+
+MLXSW_REG_DEFINE(mgpc, MLXSW_REG_MGPC_ID, MLXSW_REG_MGPC_LEN);
+
+/* reg_mgpc_counter_set_type
+ * Counter set type.
+ * Access: OP
+ */
+MLXSW_ITEM32(reg, mgpc, counter_set_type, 0x00, 24, 8);
+
+/* reg_mgpc_counter_index
+ * Counter index.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, mgpc, counter_index, 0x00, 0, 24);
+
+enum mlxsw_reg_mgpc_opcode {
+	/* Nop */
+	MLXSW_REG_MGPC_OPCODE_NOP = 0x00,
+	/* Clear counters */
+	MLXSW_REG_MGPC_OPCODE_CLEAR = 0x08,
+};
+
+/* reg_mgpc_opcode
+ * Opcode.
+ * Access: OP
+ */
+MLXSW_ITEM32(reg, mgpc, opcode, 0x04, 28, 4);
+
+/* reg_mgpc_byte_counter
+ * Byte counter value.
+ * Access: RW
+ */
+MLXSW_ITEM64(reg, mgpc, byte_counter, 0x08, 0, 64);
+
+/* reg_mgpc_packet_counter
+ * Packet counter value.
+ * Access: RW
+ */
+MLXSW_ITEM64(reg, mgpc, packet_counter, 0x10, 0, 64);
+
+static inline void mlxsw_reg_mgpc_pack(char *payload, u32 counter_index,
+				       enum mlxsw_reg_mgpc_opcode opcode,
+				       enum mlxsw_reg_flow_counter_set_type set_type)
+{
+	MLXSW_REG_ZERO(mgpc, payload);
+	mlxsw_reg_mgpc_counter_index_set(payload, counter_index);
+	mlxsw_reg_mgpc_counter_set_type_set(payload, set_type);
+	mlxsw_reg_mgpc_opcode_set(payload, opcode);
+}
+
+/* MPRS - Monitoring Parsing State Register
+ * ----------------------------------------
+ * The MPRS register is used for setting up the parsing for hash,
+ * policy-engine and routing.
+ */
+#define MLXSW_REG_MPRS_ID 0x9083
+#define MLXSW_REG_MPRS_LEN 0x14
+
+MLXSW_REG_DEFINE(mprs, MLXSW_REG_MPRS_ID, MLXSW_REG_MPRS_LEN);
+
+/* reg_mprs_parsing_depth
+ * Minimum parsing depth.
+ * Need to enlarge parsing depth according to L3, MPLS, tunnels, ACL
+ * rules, traps, hash, etc. Default is 96 bytes. Reserved when SwitchX-2.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mprs, parsing_depth, 0x00, 0, 16);
+
+/* reg_mprs_parsing_en
+ * Parsing enable.
+ * Bit 0 - Enable parsing of NVE of types VxLAN, VxLAN-GPE, GENEVE and
+ * NVGRE. Default is enabled. Reserved when SwitchX-2.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mprs, parsing_en, 0x04, 0, 16);
+
+/* reg_mprs_vxlan_udp_dport
+ * VxLAN UDP destination port.
+ * Used for identifying VxLAN packets and for dport field in
+ * encapsulation. Default is 4789.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mprs, vxlan_udp_dport, 0x10, 0, 16);
+
+static inline void mlxsw_reg_mprs_pack(char *payload, u16 parsing_depth,
+				       u16 vxlan_udp_dport)
+{
+	MLXSW_REG_ZERO(mprs, payload);
+	mlxsw_reg_mprs_parsing_depth_set(payload, parsing_depth);
+	mlxsw_reg_mprs_parsing_en_set(payload, true);
+	mlxsw_reg_mprs_vxlan_udp_dport_set(payload, vxlan_udp_dport);
+}
+
+/* MOGCR - Monitoring Global Configuration Register
+ * ------------------------------------------------
+ */
+#define MLXSW_REG_MOGCR_ID 0x9086
+#define MLXSW_REG_MOGCR_LEN 0x20
+
+MLXSW_REG_DEFINE(mogcr, MLXSW_REG_MOGCR_ID, MLXSW_REG_MOGCR_LEN);
+
+/* reg_mogcr_ptp_iftc
+ * PTP Ingress FIFO Trap Clear
+ * The PTP_ING_FIFO trap provides MTPPTR with clr according
+ * to this value. Default 0.
+ * Reserved when IB switches and when SwitchX/-2, Spectrum-2
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mogcr, ptp_iftc, 0x00, 1, 1);
+
+/* reg_mogcr_ptp_eftc
+ * PTP Egress FIFO Trap Clear
+ * The PTP_EGR_FIFO trap provides MTPPTR with clr according
+ * to this value. Default 0.
+ * Reserved when IB switches and when SwitchX/-2, Spectrum-2
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mogcr, ptp_eftc, 0x00, 0, 1);
+
+/* reg_mogcr_mirroring_pid_base
+ * Base policer id for mirroring policers.
+ * Must have an even value (e.g. 1000, not 1001).
+ * Reserved when SwitchX/-2, Switch-IB/2, Spectrum-1 and Quantum.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mogcr, mirroring_pid_base, 0x0C, 0, 14);
+
+/* MPAGR - Monitoring Port Analyzer Global Register
+ * ------------------------------------------------
+ * This register is used for global port analyzer configurations.
+ * Note: This register is not supported by current FW versions for Spectrum-1.
+ */
+#define MLXSW_REG_MPAGR_ID 0x9089
+#define MLXSW_REG_MPAGR_LEN 0x0C
+
+MLXSW_REG_DEFINE(mpagr, MLXSW_REG_MPAGR_ID, MLXSW_REG_MPAGR_LEN);
+
+enum mlxsw_reg_mpagr_trigger {
+	MLXSW_REG_MPAGR_TRIGGER_EGRESS,
+	MLXSW_REG_MPAGR_TRIGGER_INGRESS,
+	MLXSW_REG_MPAGR_TRIGGER_INGRESS_WRED,
+	MLXSW_REG_MPAGR_TRIGGER_INGRESS_SHARED_BUFFER,
+	MLXSW_REG_MPAGR_TRIGGER_INGRESS_ING_CONG,
+	MLXSW_REG_MPAGR_TRIGGER_INGRESS_EGR_CONG,
+	MLXSW_REG_MPAGR_TRIGGER_EGRESS_ECN,
+	MLXSW_REG_MPAGR_TRIGGER_EGRESS_HIGH_LATENCY,
+};
+
+/* reg_mpagr_trigger
+ * Mirror trigger.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, mpagr, trigger, 0x00, 0, 4);
+
+/* reg_mpagr_pa_id
+ * Port analyzer ID.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mpagr, pa_id, 0x04, 0, 4);
+
+#define MLXSW_REG_MPAGR_RATE_MAX 3500000000UL
+
+/* reg_mpagr_probability_rate
+ * Sampling rate.
+ * Valid values are: 1 to 3.5*10^9
+ * Value of 1 means "sample all". Default is 1.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mpagr, probability_rate, 0x08, 0, 32);
+
+static inline void mlxsw_reg_mpagr_pack(char *payload,
+					enum mlxsw_reg_mpagr_trigger trigger,
+					u8 pa_id, u32 probability_rate)
+{
+	MLXSW_REG_ZERO(mpagr, payload);
+	mlxsw_reg_mpagr_trigger_set(payload, trigger);
+	mlxsw_reg_mpagr_pa_id_set(payload, pa_id);
+	mlxsw_reg_mpagr_probability_rate_set(payload, probability_rate);
+}
+
+/* MOMTE - Monitoring Mirror Trigger Enable Register
+ * -------------------------------------------------
+ * This register is used to configure the mirror enable for different mirror
+ * reasons.
+ */
+#define MLXSW_REG_MOMTE_ID 0x908D
+#define MLXSW_REG_MOMTE_LEN 0x10
+
+MLXSW_REG_DEFINE(momte, MLXSW_REG_MOMTE_ID, MLXSW_REG_MOMTE_LEN);
+
+/* reg_momte_local_port
+ * Local port number.
+ * Access: Index
+ */
+MLXSW_ITEM32_LP(reg, momte, 0x00, 16, 0x00, 12);
+
+enum mlxsw_reg_momte_type {
+	MLXSW_REG_MOMTE_TYPE_WRED = 0x20,
+	MLXSW_REG_MOMTE_TYPE_SHARED_BUFFER_TCLASS = 0x31,
+	MLXSW_REG_MOMTE_TYPE_SHARED_BUFFER_TCLASS_DESCRIPTORS = 0x32,
+	MLXSW_REG_MOMTE_TYPE_SHARED_BUFFER_EGRESS_PORT = 0x33,
+	MLXSW_REG_MOMTE_TYPE_ING_CONG = 0x40,
+	MLXSW_REG_MOMTE_TYPE_EGR_CONG = 0x50,
+	MLXSW_REG_MOMTE_TYPE_ECN = 0x60,
+	MLXSW_REG_MOMTE_TYPE_HIGH_LATENCY = 0x70,
+};
+
+/* reg_momte_type
+ * Type of mirroring.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, momte, type, 0x04, 0, 8);
+
+/* reg_momte_tclass_en
+ * TClass/PG mirror enable. Each bit represents corresponding tclass.
+ * 0: disable (default)
+ * 1: enable
+ * Access: RW
+ */
+MLXSW_ITEM_BIT_ARRAY(reg, momte, tclass_en, 0x08, 0x08, 1);
+
+static inline void mlxsw_reg_momte_pack(char *payload, u16 local_port,
+					enum mlxsw_reg_momte_type type)
+{
+	MLXSW_REG_ZERO(momte, payload);
+	mlxsw_reg_momte_local_port_set(payload, local_port);
+	mlxsw_reg_momte_type_set(payload, type);
+}
+
+/* MTPPPC - Time Precision Packet Port Configuration
+ * -------------------------------------------------
+ * This register serves for configuration of which PTP messages should be
+ * timestamped. This is a global configuration, despite the register name.
+ *
+ * Reserved when Spectrum-2.
+ */
+#define MLXSW_REG_MTPPPC_ID 0x9090
+#define MLXSW_REG_MTPPPC_LEN 0x28
+
+MLXSW_REG_DEFINE(mtpppc, MLXSW_REG_MTPPPC_ID, MLXSW_REG_MTPPPC_LEN);
+
+/* reg_mtpppc_ing_timestamp_message_type
+ * Bitwise vector of PTP message types to timestamp at ingress.
+ * MessageType field as defined by IEEE 1588
+ * Each bit corresponds to a value (e.g. Bit0: Sync, Bit1: Delay_Req)
+ * Default all 0
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mtpppc, ing_timestamp_message_type, 0x08, 0, 16);
+
+/* reg_mtpppc_egr_timestamp_message_type
+ * Bitwise vector of PTP message types to timestamp at egress.
+ * MessageType field as defined by IEEE 1588
+ * Each bit corresponds to a value (e.g. Bit0: Sync, Bit1: Delay_Req)
+ * Default all 0
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mtpppc, egr_timestamp_message_type, 0x0C, 0, 16);
+
+static inline void mlxsw_reg_mtpppc_pack(char *payload, u16 ing, u16 egr)
+{
+	MLXSW_REG_ZERO(mtpppc, payload);
+	mlxsw_reg_mtpppc_ing_timestamp_message_type_set(payload, ing);
+	mlxsw_reg_mtpppc_egr_timestamp_message_type_set(payload, egr);
+}
+
+/* MTPPTR - Time Precision Packet Timestamping Reading
+ * ---------------------------------------------------
+ * The MTPPTR is used for reading the per port PTP timestamp FIFO.
+ * There is a trap for packets which are latched to the timestamp FIFO, thus the
+ * SW knows which FIFO to read. Note that packets enter the FIFO before been
+ * trapped. The sequence number is used to synchronize the timestamp FIFO
+ * entries and the trapped packets.
+ * Reserved when Spectrum-2.
+ */
+
+#define MLXSW_REG_MTPPTR_ID 0x9091
+#define MLXSW_REG_MTPPTR_BASE_LEN 0x10 /* base length, without records */
+#define MLXSW_REG_MTPPTR_REC_LEN 0x10 /* record length */
+#define MLXSW_REG_MTPPTR_REC_MAX_COUNT 4
+#define MLXSW_REG_MTPPTR_LEN (MLXSW_REG_MTPPTR_BASE_LEN +		\
+		    MLXSW_REG_MTPPTR_REC_LEN * MLXSW_REG_MTPPTR_REC_MAX_COUNT)
+
+MLXSW_REG_DEFINE(mtpptr, MLXSW_REG_MTPPTR_ID, MLXSW_REG_MTPPTR_LEN);
+
+/* reg_mtpptr_local_port
+ * Not supported for CPU port.
+ * Access: Index
+ */
+MLXSW_ITEM32_LP(reg, mtpptr, 0x00, 16, 0x00, 12);
+
+enum mlxsw_reg_mtpptr_dir {
+	MLXSW_REG_MTPPTR_DIR_INGRESS,
+	MLXSW_REG_MTPPTR_DIR_EGRESS,
+};
+
+/* reg_mtpptr_dir
+ * Direction.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, mtpptr, dir, 0x00, 0, 1);
+
+/* reg_mtpptr_clr
+ * Clear the records.
+ * Access: OP
+ */
+MLXSW_ITEM32(reg, mtpptr, clr, 0x04, 31, 1);
+
+/* reg_mtpptr_num_rec
+ * Number of valid records in the response
+ * Range 0.. cap_ptp_timestamp_fifo
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mtpptr, num_rec, 0x08, 0, 4);
+
+/* reg_mtpptr_rec_message_type
+ * MessageType field as defined by IEEE 1588 Each bit corresponds to a value
+ * (e.g. Bit0: Sync, Bit1: Delay_Req)
+ * Access: RO
+ */
+MLXSW_ITEM32_INDEXED(reg, mtpptr, rec_message_type,
+		     MLXSW_REG_MTPPTR_BASE_LEN, 8, 4,
+		     MLXSW_REG_MTPPTR_REC_LEN, 0, false);
+
+/* reg_mtpptr_rec_domain_number
+ * DomainNumber field as defined by IEEE 1588
+ * Access: RO
+ */
+MLXSW_ITEM32_INDEXED(reg, mtpptr, rec_domain_number,
+		     MLXSW_REG_MTPPTR_BASE_LEN, 0, 8,
+		     MLXSW_REG_MTPPTR_REC_LEN, 0, false);
+
+/* reg_mtpptr_rec_sequence_id
+ * SequenceId field as defined by IEEE 1588
+ * Access: RO
+ */
+MLXSW_ITEM32_INDEXED(reg, mtpptr, rec_sequence_id,
+		     MLXSW_REG_MTPPTR_BASE_LEN, 0, 16,
+		     MLXSW_REG_MTPPTR_REC_LEN, 0x4, false);
+
+/* reg_mtpptr_rec_timestamp_high
+ * Timestamp of when the PTP packet has passed through the port Units of PLL
+ * clock time.
+ * For Spectrum-1 the PLL clock is 156.25Mhz and PLL clock time is 6.4nSec.
+ * Access: RO
+ */
+MLXSW_ITEM32_INDEXED(reg, mtpptr, rec_timestamp_high,
+		     MLXSW_REG_MTPPTR_BASE_LEN, 0, 32,
+		     MLXSW_REG_MTPPTR_REC_LEN, 0x8, false);
+
+/* reg_mtpptr_rec_timestamp_low
+ * See rec_timestamp_high.
+ * Access: RO
+ */
+MLXSW_ITEM32_INDEXED(reg, mtpptr, rec_timestamp_low,
+		     MLXSW_REG_MTPPTR_BASE_LEN, 0, 32,
+		     MLXSW_REG_MTPPTR_REC_LEN, 0xC, false);
+
+static inline void mlxsw_reg_mtpptr_unpack(const char *payload,
+					   unsigned int rec,
+					   u8 *p_message_type,
+					   u8 *p_domain_number,
+					   u16 *p_sequence_id,
+					   u64 *p_timestamp)
+{
+	u32 timestamp_high, timestamp_low;
+
+	*p_message_type = mlxsw_reg_mtpptr_rec_message_type_get(payload, rec);
+	*p_domain_number = mlxsw_reg_mtpptr_rec_domain_number_get(payload, rec);
+	*p_sequence_id = mlxsw_reg_mtpptr_rec_sequence_id_get(payload, rec);
+	timestamp_high = mlxsw_reg_mtpptr_rec_timestamp_high_get(payload, rec);
+	timestamp_low = mlxsw_reg_mtpptr_rec_timestamp_low_get(payload, rec);
+	*p_timestamp = (u64)timestamp_high << 32 | timestamp_low;
+}
+
+/* MTPTPT - Monitoring Precision Time Protocol Trap Register
+ * ---------------------------------------------------------
+ * This register is used for configuring under which trap to deliver PTP
+ * packets depending on type of the packet.
+ */
+#define MLXSW_REG_MTPTPT_ID 0x9092
+#define MLXSW_REG_MTPTPT_LEN 0x08
+
+MLXSW_REG_DEFINE(mtptpt, MLXSW_REG_MTPTPT_ID, MLXSW_REG_MTPTPT_LEN);
+
+enum mlxsw_reg_mtptpt_trap_id {
+	MLXSW_REG_MTPTPT_TRAP_ID_PTP0,
+	MLXSW_REG_MTPTPT_TRAP_ID_PTP1,
+};
+
+/* reg_mtptpt_trap_id
+ * Trap id.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, mtptpt, trap_id, 0x00, 0, 4);
+
+/* reg_mtptpt_message_type
+ * Bitwise vector of PTP message types to trap. This is a necessary but
+ * non-sufficient condition since need to enable also per port. See MTPPPC.
+ * Message types are defined by IEEE 1588 Each bit corresponds to a value (e.g.
+ * Bit0: Sync, Bit1: Delay_Req)
+ */
+MLXSW_ITEM32(reg, mtptpt, message_type, 0x04, 0, 16);
+
+static inline void mlxsw_reg_mtptpt_pack(char *payload,
+					 enum mlxsw_reg_mtptpt_trap_id trap_id,
+					 u16 message_type)
+{
+	MLXSW_REG_ZERO(mtptpt, payload);
+	mlxsw_reg_mtptpt_trap_id_set(payload, trap_id);
+	mlxsw_reg_mtptpt_message_type_set(payload, message_type);
+}
+
+/* MTPCPC - Monitoring Time Precision Correction Port Configuration Register
+ * -------------------------------------------------------------------------
+ */
+#define MLXSW_REG_MTPCPC_ID 0x9093
+#define MLXSW_REG_MTPCPC_LEN 0x2C
+
+MLXSW_REG_DEFINE(mtpcpc, MLXSW_REG_MTPCPC_ID, MLXSW_REG_MTPCPC_LEN);
+
+/* reg_mtpcpc_pport
+ * Per port:
+ * 0: config is global. When reading - the local_port is 1.
+ * 1: config is per port.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, mtpcpc, pport, 0x00, 31, 1);
+
+/* reg_mtpcpc_local_port
+ * Local port number.
+ * Supported to/from CPU port.
+ * Reserved when pport = 0.
+ * Access: Index
+ */
+MLXSW_ITEM32_LP(reg, mtpcpc, 0x00, 16, 0x00, 12);
+
+/* reg_mtpcpc_ptp_trap_en
+ * Enable PTP traps.
+ * The trap_id is configured by MTPTPT.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mtpcpc, ptp_trap_en, 0x04, 0, 1);
+
+/* reg_mtpcpc_ing_correction_message_type
+ * Bitwise vector of PTP message types to update correction-field at ingress.
+ * MessageType field as defined by IEEE 1588 Each bit corresponds to a value
+ * (e.g. Bit0: Sync, Bit1: Delay_Req). Supported also from CPU port.
+ * Default all 0
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mtpcpc, ing_correction_message_type, 0x10, 0, 16);
+
+/* reg_mtpcpc_egr_correction_message_type
+ * Bitwise vector of PTP message types to update correction-field at egress.
+ * MessageType field as defined by IEEE 1588 Each bit corresponds to a value
+ * (e.g. Bit0: Sync, Bit1: Delay_Req). Supported also from CPU port.
+ * Default all 0
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mtpcpc, egr_correction_message_type, 0x14, 0, 16);
+
+static inline void mlxsw_reg_mtpcpc_pack(char *payload, bool pport,
+					 u16 local_port, bool ptp_trap_en,
+					 u16 ing, u16 egr)
+{
+	MLXSW_REG_ZERO(mtpcpc, payload);
+	mlxsw_reg_mtpcpc_pport_set(payload, pport);
+	mlxsw_reg_mtpcpc_local_port_set(payload, pport ? local_port : 0);
+	mlxsw_reg_mtpcpc_ptp_trap_en_set(payload, ptp_trap_en);
+	mlxsw_reg_mtpcpc_ing_correction_message_type_set(payload, ing);
+	mlxsw_reg_mtpcpc_egr_correction_message_type_set(payload, egr);
+}
+
+/* MFGD - Monitoring FW General Debug Register
+ * -------------------------------------------
+ */
+#define MLXSW_REG_MFGD_ID 0x90F0
+#define MLXSW_REG_MFGD_LEN 0x0C
+
+MLXSW_REG_DEFINE(mfgd, MLXSW_REG_MFGD_ID, MLXSW_REG_MFGD_LEN);
+
+/* reg_mfgd_fw_fatal_event_mode
+ * 0 - don't check FW fatal (default)
+ * 1 - check FW fatal - enable MFDE trap
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mfgd, fatal_event_mode, 0x00, 9, 2);
+
+/* reg_mfgd_trigger_test
+ * Access: WO
+ */
+MLXSW_ITEM32(reg, mfgd, trigger_test, 0x00, 11, 1);
+
+/* MGPIR - Management General Peripheral Information Register
+ * ----------------------------------------------------------
+ * MGPIR register allows software to query the hardware and
+ * firmware general information of peripheral entities.
+ */
+#define MLXSW_REG_MGPIR_ID 0x9100
+#define MLXSW_REG_MGPIR_LEN 0xA0
+
+MLXSW_REG_DEFINE(mgpir, MLXSW_REG_MGPIR_ID, MLXSW_REG_MGPIR_LEN);
+
+enum mlxsw_reg_mgpir_device_type {
+	MLXSW_REG_MGPIR_DEVICE_TYPE_NONE,
+	MLXSW_REG_MGPIR_DEVICE_TYPE_GEARBOX_DIE,
+};
+
+/* mgpir_slot_index
+ * Slot index (0: Main board).
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, mgpir, slot_index, 0x00, 28, 4);
+
+/* mgpir_device_type
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mgpir, device_type, 0x00, 24, 4);
+
+/* mgpir_devices_per_flash
+ * Number of devices of device_type per flash (can be shared by few devices).
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mgpir, devices_per_flash, 0x00, 16, 8);
+
+/* mgpir_num_of_devices
+ * Number of devices of device_type.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mgpir, num_of_devices, 0x00, 0, 8);
+
+/* max_modules_per_slot
+ * Maximum number of modules that can be connected per slot.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mgpir, max_modules_per_slot, 0x04, 16, 8);
+
+/* mgpir_num_of_slots
+ * Number of slots in the system.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mgpir, num_of_slots, 0x04, 8, 8);
+
+/* mgpir_num_of_modules
+ * Number of modules.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mgpir, num_of_modules, 0x04, 0, 8);
+
+static inline void mlxsw_reg_mgpir_pack(char *payload, u8 slot_index)
+{
+	MLXSW_REG_ZERO(mgpir, payload);
+	mlxsw_reg_mgpir_slot_index_set(payload, slot_index);
+}
+
+static inline void
+mlxsw_reg_mgpir_unpack(char *payload, u8 *num_of_devices,
+		       enum mlxsw_reg_mgpir_device_type *device_type,
+		       u8 *devices_per_flash, u8 *num_of_modules,
+		       u8 *num_of_slots)
+{
+	if (num_of_devices)
+		*num_of_devices = mlxsw_reg_mgpir_num_of_devices_get(payload);
+	if (device_type)
+		*device_type = mlxsw_reg_mgpir_device_type_get(payload);
+	if (devices_per_flash)
+		*devices_per_flash =
+				mlxsw_reg_mgpir_devices_per_flash_get(payload);
+	if (num_of_modules)
+		*num_of_modules = mlxsw_reg_mgpir_num_of_modules_get(payload);
+	if (num_of_slots)
+		*num_of_slots = mlxsw_reg_mgpir_num_of_slots_get(payload);
+}
+
+/* MBCT - Management Binary Code Transfer Register
+ * -----------------------------------------------
+ * This register allows to transfer binary codes from the host to
+ * the management FW by transferring it by chunks of maximum 1KB.
+ */
+#define MLXSW_REG_MBCT_ID 0x9120
+#define MLXSW_REG_MBCT_LEN 0x420
+
+MLXSW_REG_DEFINE(mbct, MLXSW_REG_MBCT_ID, MLXSW_REG_MBCT_LEN);
+
+/* reg_mbct_slot_index
+ * Slot index. 0 is reserved.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, mbct, slot_index, 0x00, 0, 4);
+
+/* reg_mbct_data_size
+ * Actual data field size in bytes for the current data transfer.
+ * Access: WO
+ */
+MLXSW_ITEM32(reg, mbct, data_size, 0x04, 0, 11);
+
+enum mlxsw_reg_mbct_op {
+	MLXSW_REG_MBCT_OP_ERASE_INI_IMAGE = 1,
+	MLXSW_REG_MBCT_OP_DATA_TRANSFER, /* Download */
+	MLXSW_REG_MBCT_OP_ACTIVATE,
+	MLXSW_REG_MBCT_OP_CLEAR_ERRORS = 6,
+	MLXSW_REG_MBCT_OP_QUERY_STATUS,
+};
+
+/* reg_mbct_op
+ * Access: WO
+ */
+MLXSW_ITEM32(reg, mbct, op, 0x08, 28, 4);
+
+/* reg_mbct_last
+ * Indicates that the current data field is the last chunk of the INI.
+ * Access: WO
+ */
+MLXSW_ITEM32(reg, mbct, last, 0x08, 26, 1);
+
+/* reg_mbct_oee
+ * Opcode Event Enable. When set a BCTOE event will be sent once the opcode
+ * was executed and the fsm_state has changed.
+ * Access: WO
+ */
+MLXSW_ITEM32(reg, mbct, oee, 0x08, 25, 1);
+
+enum mlxsw_reg_mbct_status {
+	/* Partial data transfer completed successfully and ready for next
+	 * data transfer.
+	 */
+	MLXSW_REG_MBCT_STATUS_PART_DATA = 2,
+	MLXSW_REG_MBCT_STATUS_LAST_DATA,
+	MLXSW_REG_MBCT_STATUS_ERASE_COMPLETE,
+	/* Error - trying to erase INI while it being used. */
+	MLXSW_REG_MBCT_STATUS_ERROR_INI_IN_USE,
+	/* Last data transfer completed, applying magic pattern. */
+	MLXSW_REG_MBCT_STATUS_ERASE_FAILED = 7,
+	MLXSW_REG_MBCT_STATUS_INI_ERROR,
+	MLXSW_REG_MBCT_STATUS_ACTIVATION_FAILED,
+	MLXSW_REG_MBCT_STATUS_ILLEGAL_OPERATION = 11,
+};
+
+/* reg_mbct_status
+ * Status.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mbct, status, 0x0C, 24, 5);
+
+enum mlxsw_reg_mbct_fsm_state {
+	MLXSW_REG_MBCT_FSM_STATE_INI_IN_USE = 5,
+	MLXSW_REG_MBCT_FSM_STATE_ERROR,
+};
+
+/* reg_mbct_fsm_state
+ * FSM state.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mbct, fsm_state,  0x0C, 16, 4);
+
+#define MLXSW_REG_MBCT_DATA_LEN 1024
+
+/* reg_mbct_data
+ * Up to 1KB of data.
+ * Access: WO
+ */
+MLXSW_ITEM_BUF(reg, mbct, data, 0x20, MLXSW_REG_MBCT_DATA_LEN);
+
+static inline void mlxsw_reg_mbct_pack(char *payload, u8 slot_index,
+				       enum mlxsw_reg_mbct_op op, bool oee)
+{
+	MLXSW_REG_ZERO(mbct, payload);
+	mlxsw_reg_mbct_slot_index_set(payload, slot_index);
+	mlxsw_reg_mbct_op_set(payload, op);
+	mlxsw_reg_mbct_oee_set(payload, oee);
+}
+
+static inline void mlxsw_reg_mbct_dt_pack(char *payload,
+					  u16 data_size, bool last,
+					  const char *data)
+{
+	if (WARN_ON(data_size > MLXSW_REG_MBCT_DATA_LEN))
+		return;
+	mlxsw_reg_mbct_data_size_set(payload, data_size);
+	mlxsw_reg_mbct_last_set(payload, last);
+	mlxsw_reg_mbct_data_memcpy_to(payload, data);
+}
+
+static inline void
+mlxsw_reg_mbct_unpack(const char *payload, u8 *p_slot_index,
+		      enum mlxsw_reg_mbct_status *p_status,
+		      enum mlxsw_reg_mbct_fsm_state *p_fsm_state)
+{
+	if (p_slot_index)
+		*p_slot_index = mlxsw_reg_mbct_slot_index_get(payload);
+	*p_status = mlxsw_reg_mbct_status_get(payload);
+	if (p_fsm_state)
+		*p_fsm_state = mlxsw_reg_mbct_fsm_state_get(payload);
+}
+
+/* MDDT - Management DownStream Device Tunneling Register
+ * ------------------------------------------------------
+ * This register allows to deliver query and request messages (PRM registers,
+ * commands) to a DownStream device.
+ */
+#define MLXSW_REG_MDDT_ID 0x9160
+#define MLXSW_REG_MDDT_LEN 0x110
+
+MLXSW_REG_DEFINE(mddt, MLXSW_REG_MDDT_ID, MLXSW_REG_MDDT_LEN);
+
+/* reg_mddt_slot_index
+ * Slot index.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, mddt, slot_index, 0x00, 8, 4);
+
+/* reg_mddt_device_index
+ * Device index.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, mddt, device_index, 0x00, 0, 8);
+
+/* reg_mddt_read_size
+ * Read size in D-Words.
+ * Access: OP
+ */
+MLXSW_ITEM32(reg, mddt, read_size, 0x04, 24, 8);
+
+/* reg_mddt_write_size
+ * Write size in D-Words.
+ * Access: OP
+ */
+MLXSW_ITEM32(reg, mddt, write_size, 0x04, 16, 8);
+
+enum mlxsw_reg_mddt_status {
+	MLXSW_REG_MDDT_STATUS_OK,
+};
+
+/* reg_mddt_status
+ * Return code of the Downstream Device to the register that was sent.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mddt, status, 0x0C, 24, 8);
+
+enum mlxsw_reg_mddt_method {
+	MLXSW_REG_MDDT_METHOD_QUERY,
+	MLXSW_REG_MDDT_METHOD_WRITE,
+};
+
+/* reg_mddt_method
+ * Access: OP
+ */
+MLXSW_ITEM32(reg, mddt, method, 0x0C, 22, 2);
+
+/* reg_mddt_register_id
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, mddt, register_id, 0x0C, 0, 16);
+
+#define MLXSW_REG_MDDT_PAYLOAD_OFFSET 0x0C
+#define MLXSW_REG_MDDT_PRM_REGISTER_HEADER_LEN 4
+
+static inline char *mlxsw_reg_mddt_inner_payload(char *payload)
+{
+	return payload + MLXSW_REG_MDDT_PAYLOAD_OFFSET +
+	       MLXSW_REG_MDDT_PRM_REGISTER_HEADER_LEN;
+}
+
+static inline void mlxsw_reg_mddt_pack(char *payload, u8 slot_index,
+				       u8 device_index,
+				       enum mlxsw_reg_mddt_method method,
+				       const struct mlxsw_reg_info *reg,
+				       char **inner_payload)
+{
+	int len = reg->len + MLXSW_REG_MDDT_PRM_REGISTER_HEADER_LEN;
+
+	if (WARN_ON(len + MLXSW_REG_MDDT_PAYLOAD_OFFSET > MLXSW_REG_MDDT_LEN))
+		len = MLXSW_REG_MDDT_LEN - MLXSW_REG_MDDT_PAYLOAD_OFFSET;
+
+	MLXSW_REG_ZERO(mddt, payload);
+	mlxsw_reg_mddt_slot_index_set(payload, slot_index);
+	mlxsw_reg_mddt_device_index_set(payload, device_index);
+	mlxsw_reg_mddt_method_set(payload, method);
+	mlxsw_reg_mddt_register_id_set(payload, reg->id);
+	mlxsw_reg_mddt_read_size_set(payload, len / 4);
+	mlxsw_reg_mddt_write_size_set(payload, len / 4);
+	*inner_payload = mlxsw_reg_mddt_inner_payload(payload);
+}
+
+/* MDDQ - Management DownStream Device Query Register
+ * --------------------------------------------------
+ * This register allows to query the DownStream device properties. The desired
+ * information is chosen upon the query_type field and is delivered by 32B
+ * of data blocks.
+ */
+#define MLXSW_REG_MDDQ_ID 0x9161
+#define MLXSW_REG_MDDQ_LEN 0x30
+
+MLXSW_REG_DEFINE(mddq, MLXSW_REG_MDDQ_ID, MLXSW_REG_MDDQ_LEN);
+
+/* reg_mddq_sie
+ * Slot info event enable.
+ * When set to '1', each change in the slot_info.provisioned / sr_valid /
+ * active / ready will generate a DSDSC event.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mddq, sie, 0x00, 31, 1);
+
+enum mlxsw_reg_mddq_query_type {
+	MLXSW_REG_MDDQ_QUERY_TYPE_SLOT_INFO = 1,
+	MLXSW_REG_MDDQ_QUERY_TYPE_DEVICE_INFO, /* If there are no devices
+						* on the slot, data_valid
+						* will be '0'.
+						*/
+	MLXSW_REG_MDDQ_QUERY_TYPE_SLOT_NAME,
+};
+
+/* reg_mddq_query_type
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, mddq, query_type, 0x00, 16, 8);
+
+/* reg_mddq_slot_index
+ * Slot index. 0 is reserved.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, mddq, slot_index, 0x00, 0, 4);
+
+/* reg_mddq_response_msg_seq
+ * Response message sequential number. For a specific request, the response
+ * message sequential number is the following one. In addition, the last
+ * message should be 0.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mddq, response_msg_seq, 0x04, 16, 8);
+
+/* reg_mddq_request_msg_seq
+ * Request message sequential number.
+ * The first message number should be 0.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, mddq, request_msg_seq, 0x04, 0, 8);
+
+/* reg_mddq_data_valid
+ * If set, the data in the data field is valid and contain the information
+ * for the queried index.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mddq, data_valid, 0x08, 31, 1);
+
+/* reg_mddq_slot_info_provisioned
+ * If set, the INI file is applied and the card is provisioned.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mddq, slot_info_provisioned, 0x10, 31, 1);
+
+/* reg_mddq_slot_info_sr_valid
+ * If set, Shift Register is valid (after being provisioned) and data
+ * can be sent from the switch ASIC to the line-card CPLD over Shift-Register.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mddq, slot_info_sr_valid, 0x10, 30, 1);
+
+enum mlxsw_reg_mddq_slot_info_ready {
+	MLXSW_REG_MDDQ_SLOT_INFO_READY_NOT_READY,
+	MLXSW_REG_MDDQ_SLOT_INFO_READY_READY,
+	MLXSW_REG_MDDQ_SLOT_INFO_READY_ERROR,
+};
+
+/* reg_mddq_slot_info_lc_ready
+ * If set, the LC is powered on, matching the INI version and a new FW
+ * version can be burnt (if necessary).
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mddq, slot_info_lc_ready, 0x10, 28, 2);
+
+/* reg_mddq_slot_info_active
+ * If set, the FW has completed the MDDC.device_enable command.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mddq, slot_info_active, 0x10, 27, 1);
+
+/* reg_mddq_slot_info_hw_revision
+ * Major user-configured version number of the current INI file.
+ * Valid only when active or ready are '1'.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mddq, slot_info_hw_revision, 0x14, 16, 16);
+
+/* reg_mddq_slot_info_ini_file_version
+ * User-configured version number of the current INI file.
+ * Valid only when active or lc_ready are '1'.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mddq, slot_info_ini_file_version, 0x14, 0, 16);
+
+/* reg_mddq_slot_info_card_type
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mddq, slot_info_card_type, 0x18, 0, 8);
+
+static inline void
+__mlxsw_reg_mddq_pack(char *payload, u8 slot_index,
+		      enum mlxsw_reg_mddq_query_type query_type)
+{
+	MLXSW_REG_ZERO(mddq, payload);
+	mlxsw_reg_mddq_slot_index_set(payload, slot_index);
+	mlxsw_reg_mddq_query_type_set(payload, query_type);
+}
+
+static inline void
+mlxsw_reg_mddq_slot_info_pack(char *payload, u8 slot_index, bool sie)
+{
+	__mlxsw_reg_mddq_pack(payload, slot_index,
+			      MLXSW_REG_MDDQ_QUERY_TYPE_SLOT_INFO);
+	mlxsw_reg_mddq_sie_set(payload, sie);
+}
+
+static inline void
+mlxsw_reg_mddq_slot_info_unpack(const char *payload, u8 *p_slot_index,
+				bool *p_provisioned, bool *p_sr_valid,
+				enum mlxsw_reg_mddq_slot_info_ready *p_lc_ready,
+				bool *p_active, u16 *p_hw_revision,
+				u16 *p_ini_file_version,
+				u8 *p_card_type)
+{
+	*p_slot_index = mlxsw_reg_mddq_slot_index_get(payload);
+	*p_provisioned = mlxsw_reg_mddq_slot_info_provisioned_get(payload);
+	*p_sr_valid = mlxsw_reg_mddq_slot_info_sr_valid_get(payload);
+	*p_lc_ready = mlxsw_reg_mddq_slot_info_lc_ready_get(payload);
+	*p_active = mlxsw_reg_mddq_slot_info_active_get(payload);
+	*p_hw_revision = mlxsw_reg_mddq_slot_info_hw_revision_get(payload);
+	*p_ini_file_version = mlxsw_reg_mddq_slot_info_ini_file_version_get(payload);
+	*p_card_type = mlxsw_reg_mddq_slot_info_card_type_get(payload);
+}
+
+/* reg_mddq_device_info_flash_owner
+ * If set, the device is the flash owner. Otherwise, a shared flash
+ * is used by this device (another device is the flash owner).
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mddq, device_info_flash_owner, 0x10, 30, 1);
+
+/* reg_mddq_device_info_device_index
+ * Device index. The first device should number 0.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mddq, device_info_device_index, 0x10, 0, 8);
+
+/* reg_mddq_device_info_fw_major
+ * Major FW version number.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mddq, device_info_fw_major, 0x14, 16, 16);
+
+/* reg_mddq_device_info_fw_minor
+ * Minor FW version number.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mddq, device_info_fw_minor, 0x18, 16, 16);
+
+/* reg_mddq_device_info_fw_sub_minor
+ * Sub-minor FW version number.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mddq, device_info_fw_sub_minor, 0x18, 0, 16);
+
+static inline void
+mlxsw_reg_mddq_device_info_pack(char *payload, u8 slot_index,
+				u8 request_msg_seq)
+{
+	__mlxsw_reg_mddq_pack(payload, slot_index,
+			      MLXSW_REG_MDDQ_QUERY_TYPE_DEVICE_INFO);
+	mlxsw_reg_mddq_request_msg_seq_set(payload, request_msg_seq);
+}
+
+static inline void
+mlxsw_reg_mddq_device_info_unpack(const char *payload, u8 *p_response_msg_seq,
+				  bool *p_data_valid, bool *p_flash_owner,
+				  u8 *p_device_index, u16 *p_fw_major,
+				  u16 *p_fw_minor, u16 *p_fw_sub_minor)
+{
+	*p_response_msg_seq = mlxsw_reg_mddq_response_msg_seq_get(payload);
+	*p_data_valid = mlxsw_reg_mddq_data_valid_get(payload);
+	*p_flash_owner = mlxsw_reg_mddq_device_info_flash_owner_get(payload);
+	*p_device_index = mlxsw_reg_mddq_device_info_device_index_get(payload);
+	*p_fw_major = mlxsw_reg_mddq_device_info_fw_major_get(payload);
+	*p_fw_minor = mlxsw_reg_mddq_device_info_fw_minor_get(payload);
+	*p_fw_sub_minor = mlxsw_reg_mddq_device_info_fw_sub_minor_get(payload);
+}
+
+#define MLXSW_REG_MDDQ_SLOT_ASCII_NAME_LEN 20
+
+/* reg_mddq_slot_ascii_name
+ * Slot's ASCII name.
+ * Access: RO
+ */
+MLXSW_ITEM_BUF(reg, mddq, slot_ascii_name, 0x10,
+	       MLXSW_REG_MDDQ_SLOT_ASCII_NAME_LEN);
+
+static inline void
+mlxsw_reg_mddq_slot_name_pack(char *payload, u8 slot_index)
+{
+	__mlxsw_reg_mddq_pack(payload, slot_index,
+			      MLXSW_REG_MDDQ_QUERY_TYPE_SLOT_NAME);
+}
+
+static inline void
+mlxsw_reg_mddq_slot_name_unpack(const char *payload, char *slot_ascii_name)
+{
+	mlxsw_reg_mddq_slot_ascii_name_memcpy_from(payload, slot_ascii_name);
+}
+
+/* MDDC - Management DownStream Device Control Register
+ * ----------------------------------------------------
+ * This register allows to control downstream devices and line cards.
+ */
+#define MLXSW_REG_MDDC_ID 0x9163
+#define MLXSW_REG_MDDC_LEN 0x30
+
+MLXSW_REG_DEFINE(mddc, MLXSW_REG_MDDC_ID, MLXSW_REG_MDDC_LEN);
+
+/* reg_mddc_slot_index
+ * Slot index. 0 is reserved.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, mddc, slot_index, 0x00, 0, 4);
+
+/* reg_mddc_rst
+ * Reset request.
+ * Access: OP
+ */
+MLXSW_ITEM32(reg, mddc, rst, 0x04, 29, 1);
+
+/* reg_mddc_device_enable
+ * When set, FW is the manager and allowed to program the downstream device.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mddc, device_enable, 0x04, 28, 1);
+
+static inline void mlxsw_reg_mddc_pack(char *payload, u8 slot_index, bool rst,
+				       bool device_enable)
+{
+	MLXSW_REG_ZERO(mddc, payload);
+	mlxsw_reg_mddc_slot_index_set(payload, slot_index);
+	mlxsw_reg_mddc_rst_set(payload, rst);
+	mlxsw_reg_mddc_device_enable_set(payload, device_enable);
+}
+
+/* MFDE - Monitoring FW Debug Register
+ * -----------------------------------
+ */
+#define MLXSW_REG_MFDE_ID 0x9200
+#define MLXSW_REG_MFDE_LEN 0x30
+
+MLXSW_REG_DEFINE(mfde, MLXSW_REG_MFDE_ID, MLXSW_REG_MFDE_LEN);
+
+/* reg_mfde_irisc_id
+ * Which irisc triggered the event
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mfde, irisc_id, 0x00, 24, 8);
+
+enum mlxsw_reg_mfde_severity {
+	/* Unrecoverable switch behavior */
+	MLXSW_REG_MFDE_SEVERITY_FATL = 2,
+	/* Unexpected state with possible systemic failure */
+	MLXSW_REG_MFDE_SEVERITY_NRML = 3,
+	/* Unexpected state without systemic failure */
+	MLXSW_REG_MFDE_SEVERITY_INTR = 5,
+};
+
+/* reg_mfde_severity
+ * The severity of the event.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mfde, severity, 0x00, 16, 8);
+
+enum mlxsw_reg_mfde_event_id {
+	/* CRspace timeout */
+	MLXSW_REG_MFDE_EVENT_ID_CRSPACE_TO = 1,
+	/* KVD insertion machine stopped */
+	MLXSW_REG_MFDE_EVENT_ID_KVD_IM_STOP,
+	/* Triggered by MFGD.trigger_test */
+	MLXSW_REG_MFDE_EVENT_ID_TEST,
+	/* Triggered when firmware hits an assert */
+	MLXSW_REG_MFDE_EVENT_ID_FW_ASSERT,
+	/* Fatal error interrupt from hardware */
+	MLXSW_REG_MFDE_EVENT_ID_FATAL_CAUSE,
+};
+
+/* reg_mfde_event_id
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mfde, event_id, 0x00, 0, 16);
+
+enum mlxsw_reg_mfde_method {
+	MLXSW_REG_MFDE_METHOD_QUERY,
+	MLXSW_REG_MFDE_METHOD_WRITE,
+};
+
+/* reg_mfde_method
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mfde, method, 0x04, 29, 1);
+
+/* reg_mfde_long_process
+ * Indicates if the command is in long_process mode.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mfde, long_process, 0x04, 28, 1);
+
+enum mlxsw_reg_mfde_command_type {
+	MLXSW_REG_MFDE_COMMAND_TYPE_MAD,
+	MLXSW_REG_MFDE_COMMAND_TYPE_EMAD,
+	MLXSW_REG_MFDE_COMMAND_TYPE_CMDIF,
+};
+
+/* reg_mfde_command_type
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mfde, command_type, 0x04, 24, 2);
+
+/* reg_mfde_reg_attr_id
+ * EMAD - register id, MAD - attibute id
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mfde, reg_attr_id, 0x04, 0, 16);
+
+/* reg_mfde_crspace_to_log_address
+ * crspace address accessed, which resulted in timeout.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mfde, crspace_to_log_address, 0x10, 0, 32);
+
+/* reg_mfde_crspace_to_oe
+ * 0 - New event
+ * 1 - Old event, occurred before MFGD activation.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mfde, crspace_to_oe, 0x14, 24, 1);
+
+/* reg_mfde_crspace_to_log_id
+ * Which irisc triggered the timeout.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mfde, crspace_to_log_id, 0x14, 0, 4);
+
+/* reg_mfde_crspace_to_log_ip
+ * IP (instruction pointer) that triggered the timeout.
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, mfde, crspace_to_log_ip, 0x18, 0, 64);
+
+/* reg_mfde_kvd_im_stop_oe
+ * 0 - New event
+ * 1 - Old event, occurred before MFGD activation.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mfde, kvd_im_stop_oe, 0x10, 24, 1);
+
+/* reg_mfde_kvd_im_stop_pipes_mask
+ * Bit per kvh pipe.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mfde, kvd_im_stop_pipes_mask, 0x10, 0, 16);
+
+/* reg_mfde_fw_assert_var0-4
+ * Variables passed to assert.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mfde, fw_assert_var0, 0x10, 0, 32);
+MLXSW_ITEM32(reg, mfde, fw_assert_var1, 0x14, 0, 32);
+MLXSW_ITEM32(reg, mfde, fw_assert_var2, 0x18, 0, 32);
+MLXSW_ITEM32(reg, mfde, fw_assert_var3, 0x1C, 0, 32);
+MLXSW_ITEM32(reg, mfde, fw_assert_var4, 0x20, 0, 32);
+
+/* reg_mfde_fw_assert_existptr
+ * The instruction pointer when assert was triggered.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mfde, fw_assert_existptr, 0x24, 0, 32);
+
+/* reg_mfde_fw_assert_callra
+ * The return address after triggering assert.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mfde, fw_assert_callra, 0x28, 0, 32);
+
+/* reg_mfde_fw_assert_oe
+ * 0 - New event
+ * 1 - Old event, occurred before MFGD activation.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mfde, fw_assert_oe, 0x2C, 24, 1);
+
+/* reg_mfde_fw_assert_tile_v
+ * 0: The assert was from main
+ * 1: The assert was from a tile
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mfde, fw_assert_tile_v, 0x2C, 23, 1);
+
+/* reg_mfde_fw_assert_tile_index
+ * When tile_v=1, the tile_index that caused the assert.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mfde, fw_assert_tile_index, 0x2C, 16, 6);
+
+/* reg_mfde_fw_assert_ext_synd
+ * A generated one-to-one identifier which is specific per-assert.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mfde, fw_assert_ext_synd, 0x2C, 0, 16);
+
+/* reg_mfde_fatal_cause_id
+ * HW interrupt cause id.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mfde, fatal_cause_id, 0x10, 0, 18);
+
+/* reg_mfde_fatal_cause_tile_v
+ * 0: The assert was from main
+ * 1: The assert was from a tile
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mfde, fatal_cause_tile_v, 0x14, 23, 1);
+
+/* reg_mfde_fatal_cause_tile_index
+ * When tile_v=1, the tile_index that caused the assert.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, mfde, fatal_cause_tile_index, 0x14, 16, 6);
+
+/* TNGCR - Tunneling NVE General Configuration Register
+ * ----------------------------------------------------
+ * The TNGCR register is used for setting up the NVE Tunneling configuration.
+ */
+#define MLXSW_REG_TNGCR_ID 0xA001
+#define MLXSW_REG_TNGCR_LEN 0x44
+
+MLXSW_REG_DEFINE(tngcr, MLXSW_REG_TNGCR_ID, MLXSW_REG_TNGCR_LEN);
+
+enum mlxsw_reg_tngcr_type {
+	MLXSW_REG_TNGCR_TYPE_VXLAN,
+	MLXSW_REG_TNGCR_TYPE_VXLAN_GPE,
+	MLXSW_REG_TNGCR_TYPE_GENEVE,
+	MLXSW_REG_TNGCR_TYPE_NVGRE,
+};
+
+/* reg_tngcr_type
+ * Tunnel type for encapsulation and decapsulation. The types are mutually
+ * exclusive.
+ * Note: For Spectrum the NVE parsing must be enabled in MPRS.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tngcr, type, 0x00, 0, 4);
+
+/* reg_tngcr_nve_valid
+ * The VTEP is valid. Allows adding FDB entries for tunnel encapsulation.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tngcr, nve_valid, 0x04, 31, 1);
+
+/* reg_tngcr_nve_ttl_uc
+ * The TTL for NVE tunnel encapsulation underlay unicast packets.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tngcr, nve_ttl_uc, 0x04, 0, 8);
+
+/* reg_tngcr_nve_ttl_mc
+ * The TTL for NVE tunnel encapsulation underlay multicast packets.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tngcr, nve_ttl_mc, 0x08, 0, 8);
+
+enum {
+	/* Do not copy flow label. Calculate flow label using nve_flh. */
+	MLXSW_REG_TNGCR_FL_NO_COPY,
+	/* Copy flow label from inner packet if packet is IPv6 and
+	 * encapsulation is by IPv6. Otherwise, calculate flow label using
+	 * nve_flh.
+	 */
+	MLXSW_REG_TNGCR_FL_COPY,
+};
+
+/* reg_tngcr_nve_flc
+ * For NVE tunnel encapsulation: Flow label copy from inner packet.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tngcr, nve_flc, 0x0C, 25, 1);
+
+enum {
+	/* Flow label is static. In Spectrum this means '0'. Spectrum-2
+	 * uses {nve_fl_prefix, nve_fl_suffix}.
+	 */
+	MLXSW_REG_TNGCR_FL_NO_HASH,
+	/* 8 LSBs of the flow label are calculated from ECMP hash of the
+	 * inner packet. 12 MSBs are configured by nve_fl_prefix.
+	 */
+	MLXSW_REG_TNGCR_FL_HASH,
+};
+
+/* reg_tngcr_nve_flh
+ * NVE flow label hash.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tngcr, nve_flh, 0x0C, 24, 1);
+
+/* reg_tngcr_nve_fl_prefix
+ * NVE flow label prefix. Constant 12 MSBs of the flow label.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tngcr, nve_fl_prefix, 0x0C, 8, 12);
+
+/* reg_tngcr_nve_fl_suffix
+ * NVE flow label suffix. Constant 8 LSBs of the flow label.
+ * Reserved when nve_flh=1 and for Spectrum.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tngcr, nve_fl_suffix, 0x0C, 0, 8);
+
+enum {
+	/* Source UDP port is fixed (default '0') */
+	MLXSW_REG_TNGCR_UDP_SPORT_NO_HASH,
+	/* Source UDP port is calculated based on hash */
+	MLXSW_REG_TNGCR_UDP_SPORT_HASH,
+};
+
+/* reg_tngcr_nve_udp_sport_type
+ * NVE UDP source port type.
+ * Spectrum uses LAG hash (SLCRv2). Spectrum-2 uses ECMP hash (RECRv2).
+ * When the source UDP port is calculated based on hash, then the 8 LSBs
+ * are calculated from hash the 8 MSBs are configured by
+ * nve_udp_sport_prefix.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tngcr, nve_udp_sport_type, 0x10, 24, 1);
+
+/* reg_tngcr_nve_udp_sport_prefix
+ * NVE UDP source port prefix. Constant 8 MSBs of the UDP source port.
+ * Reserved when NVE type is NVGRE.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tngcr, nve_udp_sport_prefix, 0x10, 8, 8);
+
+/* reg_tngcr_nve_group_size_mc
+ * The amount of sequential linked lists of MC entries. The first linked
+ * list is configured by SFD.underlay_mc_ptr.
+ * Valid values: 1, 2, 4, 8, 16, 32, 64
+ * The linked list are configured by TNUMT.
+ * The hash is set by LAG hash.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tngcr, nve_group_size_mc, 0x18, 0, 8);
+
+/* reg_tngcr_nve_group_size_flood
+ * The amount of sequential linked lists of flooding entries. The first
+ * linked list is configured by SFMR.nve_tunnel_flood_ptr
+ * Valid values: 1, 2, 4, 8, 16, 32, 64
+ * The linked list are configured by TNUMT.
+ * The hash is set by LAG hash.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tngcr, nve_group_size_flood, 0x1C, 0, 8);
+
+/* reg_tngcr_learn_enable
+ * During decapsulation, whether to learn from NVE port.
+ * Reserved when Spectrum-2. See TNPC.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tngcr, learn_enable, 0x20, 31, 1);
+
+/* reg_tngcr_underlay_virtual_router
+ * Underlay virtual router.
+ * Reserved when Spectrum-2.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tngcr, underlay_virtual_router, 0x20, 0, 16);
+
+/* reg_tngcr_underlay_rif
+ * Underlay ingress router interface. RIF type should be loopback generic.
+ * Reserved when Spectrum.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tngcr, underlay_rif, 0x24, 0, 16);
+
+/* reg_tngcr_usipv4
+ * Underlay source IPv4 address of the NVE.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tngcr, usipv4, 0x28, 0, 32);
+
+/* reg_tngcr_usipv6
+ * Underlay source IPv6 address of the NVE. For Spectrum, must not be
+ * modified under traffic of NVE tunneling encapsulation.
+ * Access: RW
+ */
+MLXSW_ITEM_BUF(reg, tngcr, usipv6, 0x30, 16);
+
+static inline void mlxsw_reg_tngcr_pack(char *payload,
+					enum mlxsw_reg_tngcr_type type,
+					bool valid, u8 ttl)
+{
+	MLXSW_REG_ZERO(tngcr, payload);
+	mlxsw_reg_tngcr_type_set(payload, type);
+	mlxsw_reg_tngcr_nve_valid_set(payload, valid);
+	mlxsw_reg_tngcr_nve_ttl_uc_set(payload, ttl);
+	mlxsw_reg_tngcr_nve_ttl_mc_set(payload, ttl);
+	mlxsw_reg_tngcr_nve_flc_set(payload, MLXSW_REG_TNGCR_FL_NO_COPY);
+	mlxsw_reg_tngcr_nve_flh_set(payload, 0);
+	mlxsw_reg_tngcr_nve_udp_sport_type_set(payload,
+					       MLXSW_REG_TNGCR_UDP_SPORT_HASH);
+	mlxsw_reg_tngcr_nve_udp_sport_prefix_set(payload, 0);
+	mlxsw_reg_tngcr_nve_group_size_mc_set(payload, 1);
+	mlxsw_reg_tngcr_nve_group_size_flood_set(payload, 1);
+}
+
+/* TNUMT - Tunneling NVE Underlay Multicast Table Register
+ * -------------------------------------------------------
+ * The TNUMT register is for building the underlay MC table. It is used
+ * for MC, flooding and BC traffic into the NVE tunnel.
+ */
+#define MLXSW_REG_TNUMT_ID 0xA003
+#define MLXSW_REG_TNUMT_LEN 0x20
+
+MLXSW_REG_DEFINE(tnumt, MLXSW_REG_TNUMT_ID, MLXSW_REG_TNUMT_LEN);
+
+enum mlxsw_reg_tnumt_record_type {
+	MLXSW_REG_TNUMT_RECORD_TYPE_IPV4,
+	MLXSW_REG_TNUMT_RECORD_TYPE_IPV6,
+	MLXSW_REG_TNUMT_RECORD_TYPE_LABEL,
+};
+
+/* reg_tnumt_record_type
+ * Record type.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tnumt, record_type, 0x00, 28, 4);
+
+/* reg_tnumt_tunnel_port
+ * Tunnel port.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tnumt, tunnel_port, 0x00, 24, 4);
+
+/* reg_tnumt_underlay_mc_ptr
+ * Index to the underlay multicast table.
+ * For Spectrum the index is to the KVD linear.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, tnumt, underlay_mc_ptr, 0x00, 0, 24);
+
+/* reg_tnumt_vnext
+ * The next_underlay_mc_ptr is valid.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tnumt, vnext, 0x04, 31, 1);
+
+/* reg_tnumt_next_underlay_mc_ptr
+ * The next index to the underlay multicast table.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tnumt, next_underlay_mc_ptr, 0x04, 0, 24);
+
+/* reg_tnumt_record_size
+ * Number of IP addresses in the record.
+ * Range is 1..cap_max_nve_mc_entries_ipv{4,6}
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tnumt, record_size, 0x08, 0, 3);
+
+/* reg_tnumt_udip
+ * The underlay IPv4 addresses. udip[i] is reserved if i >= size
+ * Access: RW
+ */
+MLXSW_ITEM32_INDEXED(reg, tnumt, udip, 0x0C, 0, 32, 0x04, 0x00, false);
+
+/* reg_tnumt_udip_ptr
+ * The pointer to the underlay IPv6 addresses. udip_ptr[i] is reserved if
+ * i >= size. The IPv6 addresses are configured by RIPS.
+ * Access: RW
+ */
+MLXSW_ITEM32_INDEXED(reg, tnumt, udip_ptr, 0x0C, 0, 24, 0x04, 0x00, false);
+
+static inline void mlxsw_reg_tnumt_pack(char *payload,
+					enum mlxsw_reg_tnumt_record_type type,
+					enum mlxsw_reg_tunnel_port tport,
+					u32 underlay_mc_ptr, bool vnext,
+					u32 next_underlay_mc_ptr,
+					u8 record_size)
+{
+	MLXSW_REG_ZERO(tnumt, payload);
+	mlxsw_reg_tnumt_record_type_set(payload, type);
+	mlxsw_reg_tnumt_tunnel_port_set(payload, tport);
+	mlxsw_reg_tnumt_underlay_mc_ptr_set(payload, underlay_mc_ptr);
+	mlxsw_reg_tnumt_vnext_set(payload, vnext);
+	mlxsw_reg_tnumt_next_underlay_mc_ptr_set(payload, next_underlay_mc_ptr);
+	mlxsw_reg_tnumt_record_size_set(payload, record_size);
+}
+
+/* TNQCR - Tunneling NVE QoS Configuration Register
+ * ------------------------------------------------
+ * The TNQCR register configures how QoS is set in encapsulation into the
+ * underlay network.
+ */
+#define MLXSW_REG_TNQCR_ID 0xA010
+#define MLXSW_REG_TNQCR_LEN 0x0C
+
+MLXSW_REG_DEFINE(tnqcr, MLXSW_REG_TNQCR_ID, MLXSW_REG_TNQCR_LEN);
+
+/* reg_tnqcr_enc_set_dscp
+ * For encapsulation: How to set DSCP field:
+ * 0 - Copy the DSCP from the overlay (inner) IP header to the underlay
+ * (outer) IP header. If there is no IP header, use TNQDR.dscp
+ * 1 - Set the DSCP field as TNQDR.dscp
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tnqcr, enc_set_dscp, 0x04, 28, 1);
+
+static inline void mlxsw_reg_tnqcr_pack(char *payload)
+{
+	MLXSW_REG_ZERO(tnqcr, payload);
+	mlxsw_reg_tnqcr_enc_set_dscp_set(payload, 0);
+}
+
+/* TNQDR - Tunneling NVE QoS Default Register
+ * ------------------------------------------
+ * The TNQDR register configures the default QoS settings for NVE
+ * encapsulation.
+ */
+#define MLXSW_REG_TNQDR_ID 0xA011
+#define MLXSW_REG_TNQDR_LEN 0x08
+
+MLXSW_REG_DEFINE(tnqdr, MLXSW_REG_TNQDR_ID, MLXSW_REG_TNQDR_LEN);
+
+/* reg_tnqdr_local_port
+ * Local port number (receive port). CPU port is supported.
+ * Access: Index
+ */
+MLXSW_ITEM32_LP(reg, tnqdr, 0x00, 16, 0x00, 12);
+
+/* reg_tnqdr_dscp
+ * For encapsulation, the default DSCP.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tnqdr, dscp, 0x04, 0, 6);
+
+static inline void mlxsw_reg_tnqdr_pack(char *payload, u16 local_port)
+{
+	MLXSW_REG_ZERO(tnqdr, payload);
+	mlxsw_reg_tnqdr_local_port_set(payload, local_port);
+	mlxsw_reg_tnqdr_dscp_set(payload, 0);
+}
+
+/* TNEEM - Tunneling NVE Encapsulation ECN Mapping Register
+ * --------------------------------------------------------
+ * The TNEEM register maps ECN of the IP header at the ingress to the
+ * encapsulation to the ECN of the underlay network.
+ */
+#define MLXSW_REG_TNEEM_ID 0xA012
+#define MLXSW_REG_TNEEM_LEN 0x0C
+
+MLXSW_REG_DEFINE(tneem, MLXSW_REG_TNEEM_ID, MLXSW_REG_TNEEM_LEN);
+
+/* reg_tneem_overlay_ecn
+ * ECN of the IP header in the overlay network.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, tneem, overlay_ecn, 0x04, 24, 2);
+
+/* reg_tneem_underlay_ecn
+ * ECN of the IP header in the underlay network.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tneem, underlay_ecn, 0x04, 16, 2);
+
+static inline void mlxsw_reg_tneem_pack(char *payload, u8 overlay_ecn,
+					u8 underlay_ecn)
+{
+	MLXSW_REG_ZERO(tneem, payload);
+	mlxsw_reg_tneem_overlay_ecn_set(payload, overlay_ecn);
+	mlxsw_reg_tneem_underlay_ecn_set(payload, underlay_ecn);
+}
+
+/* TNDEM - Tunneling NVE Decapsulation ECN Mapping Register
+ * --------------------------------------------------------
+ * The TNDEM register configures the actions that are done in the
+ * decapsulation.
+ */
+#define MLXSW_REG_TNDEM_ID 0xA013
+#define MLXSW_REG_TNDEM_LEN 0x0C
+
+MLXSW_REG_DEFINE(tndem, MLXSW_REG_TNDEM_ID, MLXSW_REG_TNDEM_LEN);
+
+/* reg_tndem_underlay_ecn
+ * ECN field of the IP header in the underlay network.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, tndem, underlay_ecn, 0x04, 24, 2);
+
+/* reg_tndem_overlay_ecn
+ * ECN field of the IP header in the overlay network.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, tndem, overlay_ecn, 0x04, 16, 2);
+
+/* reg_tndem_eip_ecn
+ * Egress IP ECN. ECN field of the IP header of the packet which goes out
+ * from the decapsulation.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tndem, eip_ecn, 0x04, 8, 2);
+
+/* reg_tndem_trap_en
+ * Trap enable:
+ * 0 - No trap due to decap ECN
+ * 1 - Trap enable with trap_id
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tndem, trap_en, 0x08, 28, 4);
+
+/* reg_tndem_trap_id
+ * Trap ID. Either DECAP_ECN0 or DECAP_ECN1.
+ * Reserved when trap_en is '0'.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tndem, trap_id, 0x08, 0, 9);
+
+static inline void mlxsw_reg_tndem_pack(char *payload, u8 underlay_ecn,
+					u8 overlay_ecn, u8 ecn, bool trap_en,
+					u16 trap_id)
+{
+	MLXSW_REG_ZERO(tndem, payload);
+	mlxsw_reg_tndem_underlay_ecn_set(payload, underlay_ecn);
+	mlxsw_reg_tndem_overlay_ecn_set(payload, overlay_ecn);
+	mlxsw_reg_tndem_eip_ecn_set(payload, ecn);
+	mlxsw_reg_tndem_trap_en_set(payload, trap_en);
+	mlxsw_reg_tndem_trap_id_set(payload, trap_id);
+}
+
+/* TNPC - Tunnel Port Configuration Register
+ * -----------------------------------------
+ * The TNPC register is used for tunnel port configuration.
+ * Reserved when Spectrum.
+ */
+#define MLXSW_REG_TNPC_ID 0xA020
+#define MLXSW_REG_TNPC_LEN 0x18
+
+MLXSW_REG_DEFINE(tnpc, MLXSW_REG_TNPC_ID, MLXSW_REG_TNPC_LEN);
+
+/* reg_tnpc_tunnel_port
+ * Tunnel port.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, tnpc, tunnel_port, 0x00, 0, 4);
+
+/* reg_tnpc_learn_enable_v6
+ * During IPv6 underlay decapsulation, whether to learn from tunnel port.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tnpc, learn_enable_v6, 0x04, 1, 1);
+
+/* reg_tnpc_learn_enable_v4
+ * During IPv4 underlay decapsulation, whether to learn from tunnel port.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tnpc, learn_enable_v4, 0x04, 0, 1);
+
+static inline void mlxsw_reg_tnpc_pack(char *payload,
+				       enum mlxsw_reg_tunnel_port tport,
+				       bool learn_enable)
+{
+	MLXSW_REG_ZERO(tnpc, payload);
+	mlxsw_reg_tnpc_tunnel_port_set(payload, tport);
+	mlxsw_reg_tnpc_learn_enable_v4_set(payload, learn_enable);
+	mlxsw_reg_tnpc_learn_enable_v6_set(payload, learn_enable);
+}
+
+/* TIGCR - Tunneling IPinIP General Configuration Register
+ * -------------------------------------------------------
+ * The TIGCR register is used for setting up the IPinIP Tunnel configuration.
+ */
+#define MLXSW_REG_TIGCR_ID 0xA801
+#define MLXSW_REG_TIGCR_LEN 0x10
+
+MLXSW_REG_DEFINE(tigcr, MLXSW_REG_TIGCR_ID, MLXSW_REG_TIGCR_LEN);
+
+/* reg_tigcr_ipip_ttlc
+ * For IPinIP Tunnel encapsulation: whether to copy the ttl from the packet
+ * header.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tigcr, ttlc, 0x04, 8, 1);
+
+/* reg_tigcr_ipip_ttl_uc
+ * The TTL for IPinIP Tunnel encapsulation of unicast packets if
+ * reg_tigcr_ipip_ttlc is unset.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tigcr, ttl_uc, 0x04, 0, 8);
+
+static inline void mlxsw_reg_tigcr_pack(char *payload, bool ttlc, u8 ttl_uc)
+{
+	MLXSW_REG_ZERO(tigcr, payload);
+	mlxsw_reg_tigcr_ttlc_set(payload, ttlc);
+	mlxsw_reg_tigcr_ttl_uc_set(payload, ttl_uc);
+}
+
+/* TIEEM - Tunneling IPinIP Encapsulation ECN Mapping Register
+ * -----------------------------------------------------------
+ * The TIEEM register maps ECN of the IP header at the ingress to the
+ * encapsulation to the ECN of the underlay network.
+ */
+#define MLXSW_REG_TIEEM_ID 0xA812
+#define MLXSW_REG_TIEEM_LEN 0x0C
+
+MLXSW_REG_DEFINE(tieem, MLXSW_REG_TIEEM_ID, MLXSW_REG_TIEEM_LEN);
+
+/* reg_tieem_overlay_ecn
+ * ECN of the IP header in the overlay network.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, tieem, overlay_ecn, 0x04, 24, 2);
+
+/* reg_tineem_underlay_ecn
+ * ECN of the IP header in the underlay network.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tieem, underlay_ecn, 0x04, 16, 2);
+
+static inline void mlxsw_reg_tieem_pack(char *payload, u8 overlay_ecn,
+					u8 underlay_ecn)
+{
+	MLXSW_REG_ZERO(tieem, payload);
+	mlxsw_reg_tieem_overlay_ecn_set(payload, overlay_ecn);
+	mlxsw_reg_tieem_underlay_ecn_set(payload, underlay_ecn);
+}
+
+/* TIDEM - Tunneling IPinIP Decapsulation ECN Mapping Register
+ * -----------------------------------------------------------
+ * The TIDEM register configures the actions that are done in the
+ * decapsulation.
+ */
+#define MLXSW_REG_TIDEM_ID 0xA813
+#define MLXSW_REG_TIDEM_LEN 0x0C
+
+MLXSW_REG_DEFINE(tidem, MLXSW_REG_TIDEM_ID, MLXSW_REG_TIDEM_LEN);
+
+/* reg_tidem_underlay_ecn
+ * ECN field of the IP header in the underlay network.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, tidem, underlay_ecn, 0x04, 24, 2);
+
+/* reg_tidem_overlay_ecn
+ * ECN field of the IP header in the overlay network.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, tidem, overlay_ecn, 0x04, 16, 2);
+
+/* reg_tidem_eip_ecn
+ * Egress IP ECN. ECN field of the IP header of the packet which goes out
+ * from the decapsulation.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tidem, eip_ecn, 0x04, 8, 2);
+
+/* reg_tidem_trap_en
+ * Trap enable:
+ * 0 - No trap due to decap ECN
+ * 1 - Trap enable with trap_id
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tidem, trap_en, 0x08, 28, 4);
+
+/* reg_tidem_trap_id
+ * Trap ID. Either DECAP_ECN0 or DECAP_ECN1.
+ * Reserved when trap_en is '0'.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tidem, trap_id, 0x08, 0, 9);
+
+static inline void mlxsw_reg_tidem_pack(char *payload, u8 underlay_ecn,
+					u8 overlay_ecn, u8 eip_ecn,
+					bool trap_en, u16 trap_id)
+{
+	MLXSW_REG_ZERO(tidem, payload);
+	mlxsw_reg_tidem_underlay_ecn_set(payload, underlay_ecn);
+	mlxsw_reg_tidem_overlay_ecn_set(payload, overlay_ecn);
+	mlxsw_reg_tidem_eip_ecn_set(payload, eip_ecn);
+	mlxsw_reg_tidem_trap_en_set(payload, trap_en);
+	mlxsw_reg_tidem_trap_id_set(payload, trap_id);
+}
+
+/* SBPR - Shared Buffer Pools Register
+ * -----------------------------------
+ * The SBPR configures and retrieves the shared buffer pools and configuration.
+ */
+#define MLXSW_REG_SBPR_ID 0xB001
+#define MLXSW_REG_SBPR_LEN 0x14
+
+MLXSW_REG_DEFINE(sbpr, MLXSW_REG_SBPR_ID, MLXSW_REG_SBPR_LEN);
+
+/* reg_sbpr_desc
+ * When set, configures descriptor buffer.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, sbpr, desc, 0x00, 31, 1);
+
+/* shared direstion enum for SBPR, SBCM, SBPM */
+enum mlxsw_reg_sbxx_dir {
+	MLXSW_REG_SBXX_DIR_INGRESS,
+	MLXSW_REG_SBXX_DIR_EGRESS,
+};
+
+/* reg_sbpr_dir
+ * Direction.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, sbpr, dir, 0x00, 24, 2);
+
+/* reg_sbpr_pool
+ * Pool index.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, sbpr, pool, 0x00, 0, 4);
+
+/* reg_sbpr_infi_size
+ * Size is infinite.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, sbpr, infi_size, 0x04, 31, 1);
+
+/* reg_sbpr_size
+ * Pool size in buffer cells.
+ * Reserved when infi_size = 1.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, sbpr, size, 0x04, 0, 24);
+
+enum mlxsw_reg_sbpr_mode {
+	MLXSW_REG_SBPR_MODE_STATIC,
+	MLXSW_REG_SBPR_MODE_DYNAMIC,
+};
+
+/* reg_sbpr_mode
+ * Pool quota calculation mode.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, sbpr, mode, 0x08, 0, 4);
+
+static inline void mlxsw_reg_sbpr_pack(char *payload, u8 pool,
+				       enum mlxsw_reg_sbxx_dir dir,
+				       enum mlxsw_reg_sbpr_mode mode, u32 size,
+				       bool infi_size)
+{
+	MLXSW_REG_ZERO(sbpr, payload);
+	mlxsw_reg_sbpr_pool_set(payload, pool);
+	mlxsw_reg_sbpr_dir_set(payload, dir);
+	mlxsw_reg_sbpr_mode_set(payload, mode);
+	mlxsw_reg_sbpr_size_set(payload, size);
+	mlxsw_reg_sbpr_infi_size_set(payload, infi_size);
+}
+
+/* SBCM - Shared Buffer Class Management Register
+ * ----------------------------------------------
+ * The SBCM register configures and retrieves the shared buffer allocation
+ * and configuration according to Port-PG, including the binding to pool
+ * and definition of the associated quota.
+ */
+#define MLXSW_REG_SBCM_ID 0xB002
+#define MLXSW_REG_SBCM_LEN 0x28
+
+MLXSW_REG_DEFINE(sbcm, MLXSW_REG_SBCM_ID, MLXSW_REG_SBCM_LEN);
+
+/* reg_sbcm_local_port
+ * Local port number.
+ * For Ingress: excludes CPU port and Router port
+ * For Egress: excludes IP Router
+ * Access: Index
+ */
+MLXSW_ITEM32_LP(reg, sbcm, 0x00, 16, 0x00, 4);
+
+/* reg_sbcm_pg_buff
+ * PG buffer - Port PG (dir=ingress) / traffic class (dir=egress)
+ * For PG buffer: range is 0..cap_max_pg_buffers - 1
+ * For traffic class: range is 0..cap_max_tclass - 1
+ * Note that when traffic class is in MC aware mode then the traffic
+ * classes which are MC aware cannot be configured.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, sbcm, pg_buff, 0x00, 8, 6);
+
+/* reg_sbcm_dir
+ * Direction.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, sbcm, dir, 0x00, 0, 2);
+
+/* reg_sbcm_min_buff
+ * Minimum buffer size for the limiter, in cells.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, sbcm, min_buff, 0x18, 0, 24);
+
+/* shared max_buff limits for dynamic threshold for SBCM, SBPM */
+#define MLXSW_REG_SBXX_DYN_MAX_BUFF_MIN 1
+#define MLXSW_REG_SBXX_DYN_MAX_BUFF_MAX 14
+
+/* reg_sbcm_infi_max
+ * Max buffer is infinite.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, sbcm, infi_max, 0x1C, 31, 1);
+
+/* reg_sbcm_max_buff
+ * When the pool associated to the port-pg/tclass is configured to
+ * static, Maximum buffer size for the limiter configured in cells.
+ * When the pool associated to the port-pg/tclass is configured to
+ * dynamic, the max_buff holds the "alpha" parameter, supporting
+ * the following values:
+ * 0: 0
+ * i: (1/128)*2^(i-1), for i=1..14
+ * 0xFF: Infinity
+ * Reserved when infi_max = 1.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, sbcm, max_buff, 0x1C, 0, 24);
+
+/* reg_sbcm_pool
+ * Association of the port-priority to a pool.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, sbcm, pool, 0x24, 0, 4);
+
+static inline void mlxsw_reg_sbcm_pack(char *payload, u16 local_port, u8 pg_buff,
+				       enum mlxsw_reg_sbxx_dir dir,
+				       u32 min_buff, u32 max_buff,
+				       bool infi_max, u8 pool)
+{
+	MLXSW_REG_ZERO(sbcm, payload);
+	mlxsw_reg_sbcm_local_port_set(payload, local_port);
+	mlxsw_reg_sbcm_pg_buff_set(payload, pg_buff);
+	mlxsw_reg_sbcm_dir_set(payload, dir);
+	mlxsw_reg_sbcm_min_buff_set(payload, min_buff);
+	mlxsw_reg_sbcm_max_buff_set(payload, max_buff);
+	mlxsw_reg_sbcm_infi_max_set(payload, infi_max);
+	mlxsw_reg_sbcm_pool_set(payload, pool);
+}
+
+/* SBPM - Shared Buffer Port Management Register
+ * ---------------------------------------------
+ * The SBPM register configures and retrieves the shared buffer allocation
+ * and configuration according to Port-Pool, including the definition
+ * of the associated quota.
+ */
+#define MLXSW_REG_SBPM_ID 0xB003
+#define MLXSW_REG_SBPM_LEN 0x28
+
+MLXSW_REG_DEFINE(sbpm, MLXSW_REG_SBPM_ID, MLXSW_REG_SBPM_LEN);
+
+/* reg_sbpm_local_port
+ * Local port number.
+ * For Ingress: excludes CPU port and Router port
+ * For Egress: excludes IP Router
+ * Access: Index
+ */
+MLXSW_ITEM32_LP(reg, sbpm, 0x00, 16, 0x00, 12);
+
+/* reg_sbpm_pool
+ * The pool associated to quota counting on the local_port.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, sbpm, pool, 0x00, 8, 4);
+
+/* reg_sbpm_dir
+ * Direction.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, sbpm, dir, 0x00, 0, 2);
+
+/* reg_sbpm_buff_occupancy
+ * Current buffer occupancy in cells.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, sbpm, buff_occupancy, 0x10, 0, 24);
+
+/* reg_sbpm_clr
+ * Clear Max Buffer Occupancy
+ * When this bit is set, max_buff_occupancy field is cleared (and a
+ * new max value is tracked from the time the clear was performed).
+ * Access: OP
+ */
+MLXSW_ITEM32(reg, sbpm, clr, 0x14, 31, 1);
+
+/* reg_sbpm_max_buff_occupancy
+ * Maximum value of buffer occupancy in cells monitored. Cleared by
+ * writing to the clr field.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, sbpm, max_buff_occupancy, 0x14, 0, 24);
+
+/* reg_sbpm_min_buff
+ * Minimum buffer size for the limiter, in cells.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, sbpm, min_buff, 0x18, 0, 24);
+
+/* reg_sbpm_max_buff
+ * When the pool associated to the port-pg/tclass is configured to
+ * static, Maximum buffer size for the limiter configured in cells.
+ * When the pool associated to the port-pg/tclass is configured to
+ * dynamic, the max_buff holds the "alpha" parameter, supporting
+ * the following values:
+ * 0: 0
+ * i: (1/128)*2^(i-1), for i=1..14
+ * 0xFF: Infinity
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, sbpm, max_buff, 0x1C, 0, 24);
+
+static inline void mlxsw_reg_sbpm_pack(char *payload, u16 local_port, u8 pool,
+				       enum mlxsw_reg_sbxx_dir dir, bool clr,
+				       u32 min_buff, u32 max_buff)
+{
+	MLXSW_REG_ZERO(sbpm, payload);
+	mlxsw_reg_sbpm_local_port_set(payload, local_port);
+	mlxsw_reg_sbpm_pool_set(payload, pool);
+	mlxsw_reg_sbpm_dir_set(payload, dir);
+	mlxsw_reg_sbpm_clr_set(payload, clr);
+	mlxsw_reg_sbpm_min_buff_set(payload, min_buff);
+	mlxsw_reg_sbpm_max_buff_set(payload, max_buff);
+}
+
+static inline void mlxsw_reg_sbpm_unpack(char *payload, u32 *p_buff_occupancy,
+					 u32 *p_max_buff_occupancy)
+{
+	*p_buff_occupancy = mlxsw_reg_sbpm_buff_occupancy_get(payload);
+	*p_max_buff_occupancy = mlxsw_reg_sbpm_max_buff_occupancy_get(payload);
+}
+
+/* SBMM - Shared Buffer Multicast Management Register
+ * --------------------------------------------------
+ * The SBMM register configures and retrieves the shared buffer allocation
+ * and configuration for MC packets according to Switch-Priority, including
+ * the binding to pool and definition of the associated quota.
+ */
+#define MLXSW_REG_SBMM_ID 0xB004
+#define MLXSW_REG_SBMM_LEN 0x28
+
+MLXSW_REG_DEFINE(sbmm, MLXSW_REG_SBMM_ID, MLXSW_REG_SBMM_LEN);
+
+/* reg_sbmm_prio
+ * Switch Priority.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, sbmm, prio, 0x00, 8, 4);
+
+/* reg_sbmm_min_buff
+ * Minimum buffer size for the limiter, in cells.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, sbmm, min_buff, 0x18, 0, 24);
+
+/* reg_sbmm_max_buff
+ * When the pool associated to the port-pg/tclass is configured to
+ * static, Maximum buffer size for the limiter configured in cells.
+ * When the pool associated to the port-pg/tclass is configured to
+ * dynamic, the max_buff holds the "alpha" parameter, supporting
+ * the following values:
+ * 0: 0
+ * i: (1/128)*2^(i-1), for i=1..14
+ * 0xFF: Infinity
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, sbmm, max_buff, 0x1C, 0, 24);
+
+/* reg_sbmm_pool
+ * Association of the port-priority to a pool.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, sbmm, pool, 0x24, 0, 4);
+
+static inline void mlxsw_reg_sbmm_pack(char *payload, u8 prio, u32 min_buff,
+				       u32 max_buff, u8 pool)
+{
+	MLXSW_REG_ZERO(sbmm, payload);
+	mlxsw_reg_sbmm_prio_set(payload, prio);
+	mlxsw_reg_sbmm_min_buff_set(payload, min_buff);
+	mlxsw_reg_sbmm_max_buff_set(payload, max_buff);
+	mlxsw_reg_sbmm_pool_set(payload, pool);
+}
+
+/* SBSR - Shared Buffer Status Register
+ * ------------------------------------
+ * The SBSR register retrieves the shared buffer occupancy according to
+ * Port-Pool. Note that this register enables reading a large amount of data.
+ * It is the user's responsibility to limit the amount of data to ensure the
+ * response can match the maximum transfer unit. In case the response exceeds
+ * the maximum transport unit, it will be truncated with no special notice.
+ */
+#define MLXSW_REG_SBSR_ID 0xB005
+#define MLXSW_REG_SBSR_BASE_LEN 0x5C /* base length, without records */
+#define MLXSW_REG_SBSR_REC_LEN 0x8 /* record length */
+#define MLXSW_REG_SBSR_REC_MAX_COUNT 120
+#define MLXSW_REG_SBSR_LEN (MLXSW_REG_SBSR_BASE_LEN +	\
+			    MLXSW_REG_SBSR_REC_LEN *	\
+			    MLXSW_REG_SBSR_REC_MAX_COUNT)
+
+MLXSW_REG_DEFINE(sbsr, MLXSW_REG_SBSR_ID, MLXSW_REG_SBSR_LEN);
+
+/* reg_sbsr_clr
+ * Clear Max Buffer Occupancy. When this bit is set, the max_buff_occupancy
+ * field is cleared (and a new max value is tracked from the time the clear
+ * was performed).
+ * Access: OP
+ */
+MLXSW_ITEM32(reg, sbsr, clr, 0x00, 31, 1);
+
+#define MLXSW_REG_SBSR_NUM_PORTS_IN_PAGE 256
+
+/* reg_sbsr_port_page
+ * Determines the range of the ports specified in the 'ingress_port_mask'
+ * and 'egress_port_mask' bit masks.
+ * {ingress,egress}_port_mask[x] is (256 * port_page) + x
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, sbsr, port_page, 0x04, 0, 4);
+
+/* reg_sbsr_ingress_port_mask
+ * Bit vector for all ingress network ports.
+ * Indicates which of the ports (for which the relevant bit is set)
+ * are affected by the set operation. Configuration of any other port
+ * does not change.
+ * Access: Index
+ */
+MLXSW_ITEM_BIT_ARRAY(reg, sbsr, ingress_port_mask, 0x10, 0x20, 1);
+
+/* reg_sbsr_pg_buff_mask
+ * Bit vector for all switch priority groups.
+ * Indicates which of the priorities (for which the relevant bit is set)
+ * are affected by the set operation. Configuration of any other priority
+ * does not change.
+ * Range is 0..cap_max_pg_buffers - 1
+ * Access: Index
+ */
+MLXSW_ITEM_BIT_ARRAY(reg, sbsr, pg_buff_mask, 0x30, 0x4, 1);
+
+/* reg_sbsr_egress_port_mask
+ * Bit vector for all egress network ports.
+ * Indicates which of the ports (for which the relevant bit is set)
+ * are affected by the set operation. Configuration of any other port
+ * does not change.
+ * Access: Index
+ */
+MLXSW_ITEM_BIT_ARRAY(reg, sbsr, egress_port_mask, 0x34, 0x20, 1);
+
+/* reg_sbsr_tclass_mask
+ * Bit vector for all traffic classes.
+ * Indicates which of the traffic classes (for which the relevant bit is
+ * set) are affected by the set operation. Configuration of any other
+ * traffic class does not change.
+ * Range is 0..cap_max_tclass - 1
+ * Access: Index
+ */
+MLXSW_ITEM_BIT_ARRAY(reg, sbsr, tclass_mask, 0x54, 0x8, 1);
+
+static inline void mlxsw_reg_sbsr_pack(char *payload, bool clr)
+{
+	MLXSW_REG_ZERO(sbsr, payload);
+	mlxsw_reg_sbsr_clr_set(payload, clr);
+}
+
+/* reg_sbsr_rec_buff_occupancy
+ * Current buffer occupancy in cells.
+ * Access: RO
+ */
+MLXSW_ITEM32_INDEXED(reg, sbsr, rec_buff_occupancy, MLXSW_REG_SBSR_BASE_LEN,
+		     0, 24, MLXSW_REG_SBSR_REC_LEN, 0x00, false);
+
+/* reg_sbsr_rec_max_buff_occupancy
+ * Maximum value of buffer occupancy in cells monitored. Cleared by
+ * writing to the clr field.
+ * Access: RO
+ */
+MLXSW_ITEM32_INDEXED(reg, sbsr, rec_max_buff_occupancy, MLXSW_REG_SBSR_BASE_LEN,
+		     0, 24, MLXSW_REG_SBSR_REC_LEN, 0x04, false);
+
+static inline void mlxsw_reg_sbsr_rec_unpack(char *payload, int rec_index,
+					     u32 *p_buff_occupancy,
+					     u32 *p_max_buff_occupancy)
+{
+	*p_buff_occupancy =
+		mlxsw_reg_sbsr_rec_buff_occupancy_get(payload, rec_index);
+	*p_max_buff_occupancy =
+		mlxsw_reg_sbsr_rec_max_buff_occupancy_get(payload, rec_index);
+}
+
+/* SBIB - Shared Buffer Internal Buffer Register
+ * ---------------------------------------------
+ * The SBIB register configures per port buffers for internal use. The internal
+ * buffers consume memory on the port buffers (note that the port buffers are
+ * used also by PBMC).
+ *
+ * For Spectrum this is used for egress mirroring.
+ */
+#define MLXSW_REG_SBIB_ID 0xB006
+#define MLXSW_REG_SBIB_LEN 0x10
+
+MLXSW_REG_DEFINE(sbib, MLXSW_REG_SBIB_ID, MLXSW_REG_SBIB_LEN);
+
+/* reg_sbib_local_port
+ * Local port number
+ * Not supported for CPU port and router port
+ * Access: Index
+ */
+MLXSW_ITEM32_LP(reg, sbib, 0x00, 16, 0x00, 12);
+
+/* reg_sbib_buff_size
+ * Units represented in cells
+ * Allowed range is 0 to (cap_max_headroom_size - 1)
+ * Default is 0
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, sbib, buff_size, 0x08, 0, 24);
+
+static inline void mlxsw_reg_sbib_pack(char *payload, u16 local_port,
+				       u32 buff_size)
+{
+	MLXSW_REG_ZERO(sbib, payload);
+	mlxsw_reg_sbib_local_port_set(payload, local_port);
+	mlxsw_reg_sbib_buff_size_set(payload, buff_size);
+}
+
+static const struct mlxsw_reg_info *mlxsw_reg_infos[] = {
+	MLXSW_REG(sgcr),
+	MLXSW_REG(spad),
+	MLXSW_REG(sspr),
+	MLXSW_REG(sfdat),
+	MLXSW_REG(sfd),
+	MLXSW_REG(sfn),
+	MLXSW_REG(spms),
+	MLXSW_REG(spvid),
+	MLXSW_REG(spvm),
+	MLXSW_REG(spaft),
+	MLXSW_REG(sfgc),
+	MLXSW_REG(sfdf),
+	MLXSW_REG(sldr),
+	MLXSW_REG(slcr),
+	MLXSW_REG(slcor),
+	MLXSW_REG(spmlr),
+	MLXSW_REG(svfa),
+	MLXSW_REG(spvtr),
+	MLXSW_REG(svpe),
+	MLXSW_REG(sfmr),
+	MLXSW_REG(spvmlr),
+	MLXSW_REG(spvc),
+	MLXSW_REG(spevet),
+	MLXSW_REG(smpe),
+	MLXSW_REG(smid2),
+	MLXSW_REG(cwtp),
+	MLXSW_REG(cwtpm),
+	MLXSW_REG(pgcr),
+	MLXSW_REG(ppbt),
+	MLXSW_REG(pacl),
+	MLXSW_REG(pagt),
+	MLXSW_REG(ptar),
+	MLXSW_REG(ppbs),
+	MLXSW_REG(prcr),
+	MLXSW_REG(pefa),
+	MLXSW_REG(pemrbt),
+	MLXSW_REG(ptce2),
+	MLXSW_REG(perpt),
+	MLXSW_REG(peabfe),
+	MLXSW_REG(perar),
+	MLXSW_REG(ptce3),
+	MLXSW_REG(percr),
+	MLXSW_REG(pererp),
+	MLXSW_REG(iedr),
+	MLXSW_REG(qpts),
+	MLXSW_REG(qpcr),
+	MLXSW_REG(qtct),
+	MLXSW_REG(qeec),
+	MLXSW_REG(qrwe),
+	MLXSW_REG(qpdsm),
+	MLXSW_REG(qpdp),
+	MLXSW_REG(qpdpm),
+	MLXSW_REG(qtctm),
+	MLXSW_REG(qpsc),
+	MLXSW_REG(pmlp),
+	MLXSW_REG(pmtu),
+	MLXSW_REG(ptys),
+	MLXSW_REG(ppad),
+	MLXSW_REG(paos),
+	MLXSW_REG(pfcc),
+	MLXSW_REG(ppcnt),
+	MLXSW_REG(pptb),
+	MLXSW_REG(pbmc),
+	MLXSW_REG(pspa),
+	MLXSW_REG(pmaos),
+	MLXSW_REG(pplr),
+	MLXSW_REG(pmtdb),
+	MLXSW_REG(pmecr),
+	MLXSW_REG(pmpe),
+	MLXSW_REG(pddr),
+	MLXSW_REG(pmmp),
+	MLXSW_REG(pllp),
+	MLXSW_REG(pmtm),
+	MLXSW_REG(htgt),
+	MLXSW_REG(hpkt),
+	MLXSW_REG(rgcr),
+	MLXSW_REG(ritr),
+	MLXSW_REG(rtar),
+	MLXSW_REG(ratr),
+	MLXSW_REG(rtdp),
+	MLXSW_REG(rips),
+	MLXSW_REG(ratrad),
+	MLXSW_REG(rdpm),
+	MLXSW_REG(ricnt),
+	MLXSW_REG(rrcr),
+	MLXSW_REG(ralta),
+	MLXSW_REG(ralst),
+	MLXSW_REG(raltb),
+	MLXSW_REG(ralue),
+	MLXSW_REG(rauht),
+	MLXSW_REG(raleu),
+	MLXSW_REG(rauhtd),
+	MLXSW_REG(rigr2),
+	MLXSW_REG(recr2),
+	MLXSW_REG(rmft2),
+	MLXSW_REG(reiv),
+	MLXSW_REG(mfcr),
+	MLXSW_REG(mfsc),
+	MLXSW_REG(mfsm),
+	MLXSW_REG(mfsl),
+	MLXSW_REG(fore),
+	MLXSW_REG(mtcap),
+	MLXSW_REG(mtmp),
+	MLXSW_REG(mtwe),
+	MLXSW_REG(mtbr),
+	MLXSW_REG(mcia),
+	MLXSW_REG(mpat),
+	MLXSW_REG(mpar),
+	MLXSW_REG(mgir),
+	MLXSW_REG(mrsr),
+	MLXSW_REG(mlcr),
+	MLXSW_REG(mcion),
+	MLXSW_REG(mtpps),
+	MLXSW_REG(mtutc),
+	MLXSW_REG(mpsc),
+	MLXSW_REG(mcqi),
+	MLXSW_REG(mcc),
+	MLXSW_REG(mcda),
+	MLXSW_REG(mgpc),
+	MLXSW_REG(mprs),
+	MLXSW_REG(mogcr),
+	MLXSW_REG(mpagr),
+	MLXSW_REG(momte),
+	MLXSW_REG(mtpppc),
+	MLXSW_REG(mtpptr),
+	MLXSW_REG(mtptpt),
+	MLXSW_REG(mtpcpc),
+	MLXSW_REG(mfgd),
+	MLXSW_REG(mgpir),
+	MLXSW_REG(mbct),
+	MLXSW_REG(mddt),
+	MLXSW_REG(mddq),
+	MLXSW_REG(mddc),
+	MLXSW_REG(mfde),
+	MLXSW_REG(tngcr),
+	MLXSW_REG(tnumt),
+	MLXSW_REG(tnqcr),
+	MLXSW_REG(tnqdr),
+	MLXSW_REG(tneem),
+	MLXSW_REG(tndem),
+	MLXSW_REG(tnpc),
+	MLXSW_REG(tigcr),
+	MLXSW_REG(tieem),
+	MLXSW_REG(tidem),
+	MLXSW_REG(sbpr),
+	MLXSW_REG(sbcm),
+	MLXSW_REG(sbpm),
+	MLXSW_REG(sbmm),
+	MLXSW_REG(sbsr),
+	MLXSW_REG(sbib),
+};
+
+static inline const char *mlxsw_reg_id_str(u16 reg_id)
+{
+	const struct mlxsw_reg_info *reg_info;
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(mlxsw_reg_infos); i++) {
+		reg_info = mlxsw_reg_infos[i];
+		if (reg_info->id == reg_id)
+			return reg_info->name;
+	}
+	return "*UNKNOWN*";
+}
+
+/* PUDE - Port Up / Down Event
+ * ---------------------------
+ * Reports the operational state change of a port.
+ */
+#define MLXSW_REG_PUDE_LEN 0x10
+
+/* reg_pude_swid
+ * Switch partition ID with which to associate the port.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, pude, swid, 0x00, 24, 8);
+
+/* reg_pude_local_port
+ * Local port number.
+ * Access: Index
+ */
+MLXSW_ITEM32_LP(reg, pude, 0x00, 16, 0x00, 12);
+
+/* reg_pude_admin_status
+ * Port administrative state (the desired state).
+ * 1 - Up.
+ * 2 - Down.
+ * 3 - Up once. This means that in case of link failure, the port won't go
+ *     into polling mode, but will wait to be re-enabled by software.
+ * 4 - Disabled by system. Can only be set by hardware.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, pude, admin_status, 0x00, 8, 4);
+
+/* reg_pude_oper_status
+ * Port operatioanl state.
+ * 1 - Up.
+ * 2 - Down.
+ * 3 - Down by port failure. This means that the device will not let the
+ *     port up again until explicitly specified by software.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, pude, oper_status, 0x00, 0, 4);
+
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlxsw/resources.h b/drivers/net/ethernet/mellanox/mlxsw/resources.h
new file mode 100644
index 000000000..19ae0d1c7
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/resources.h
@@ -0,0 +1,168 @@
+/* SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0 */
+/* Copyright (c) 2016-2018 Mellanox Technologies. All rights reserved */
+
+#ifndef _MLXSW_RESOURCES_H
+#define _MLXSW_RESOURCES_H
+
+#include <linux/kernel.h>
+#include <linux/types.h>
+
+enum mlxsw_res_id {
+	MLXSW_RES_ID_KVD_SIZE,
+	MLXSW_RES_ID_KVD_SINGLE_MIN_SIZE,
+	MLXSW_RES_ID_KVD_DOUBLE_MIN_SIZE,
+	MLXSW_RES_ID_PGT_SIZE,
+	MLXSW_RES_ID_MAX_KVD_LINEAR_RANGE,
+	MLXSW_RES_ID_MAX_KVD_ACTION_SETS,
+	MLXSW_RES_ID_MAX_TRAP_GROUPS,
+	MLXSW_RES_ID_CQE_V0,
+	MLXSW_RES_ID_CQE_V1,
+	MLXSW_RES_ID_CQE_V2,
+	MLXSW_RES_ID_COUNTER_POOL_SIZE,
+	MLXSW_RES_ID_COUNTER_BANK_SIZE,
+	MLXSW_RES_ID_MAX_SPAN,
+	MLXSW_RES_ID_COUNTER_SIZE_PACKETS_BYTES,
+	MLXSW_RES_ID_COUNTER_SIZE_ROUTER_BASIC,
+	MLXSW_RES_ID_MAX_SYSTEM_PORT,
+	MLXSW_RES_ID_FID,
+	MLXSW_RES_ID_MAX_LAG,
+	MLXSW_RES_ID_MAX_LAG_MEMBERS,
+	MLXSW_RES_ID_GUARANTEED_SHARED_BUFFER,
+	MLXSW_RES_ID_CELL_SIZE,
+	MLXSW_RES_ID_MAX_HEADROOM_SIZE,
+	MLXSW_RES_ID_ACL_MAX_TCAM_REGIONS,
+	MLXSW_RES_ID_ACL_MAX_TCAM_RULES,
+	MLXSW_RES_ID_ACL_MAX_REGIONS,
+	MLXSW_RES_ID_ACL_MAX_GROUPS,
+	MLXSW_RES_ID_ACL_MAX_GROUP_SIZE,
+	MLXSW_RES_ID_ACL_MAX_DEFAULT_ACTIONS,
+	MLXSW_RES_ID_ACL_FLEX_KEYS,
+	MLXSW_RES_ID_ACL_MAX_ACTION_PER_RULE,
+	MLXSW_RES_ID_ACL_ACTIONS_PER_SET,
+	MLXSW_RES_ID_ACL_MAX_ERPT_BANKS,
+	MLXSW_RES_ID_ACL_MAX_ERPT_BANK_SIZE,
+	MLXSW_RES_ID_ACL_MAX_LARGE_KEY_ID,
+	MLXSW_RES_ID_ACL_ERPT_ENTRIES_2KB,
+	MLXSW_RES_ID_ACL_ERPT_ENTRIES_4KB,
+	MLXSW_RES_ID_ACL_ERPT_ENTRIES_8KB,
+	MLXSW_RES_ID_ACL_ERPT_ENTRIES_12KB,
+	MLXSW_RES_ID_ACL_MAX_BF_LOG,
+	MLXSW_RES_ID_MAX_GLOBAL_POLICERS,
+	MLXSW_RES_ID_MAX_CPU_POLICERS,
+	MLXSW_RES_ID_MAX_VRS,
+	MLXSW_RES_ID_MAX_RIFS,
+	MLXSW_RES_ID_MC_ERIF_LIST_ENTRIES,
+	MLXSW_RES_ID_MAX_RIF_MAC_PROFILES,
+	MLXSW_RES_ID_MAX_LPM_TREES,
+	MLXSW_RES_ID_MAX_NVE_MC_ENTRIES_IPV4,
+	MLXSW_RES_ID_MAX_NVE_MC_ENTRIES_IPV6,
+
+	/* Internal resources.
+	 * Determined by the SW, not queried from the HW.
+	 */
+	MLXSW_RES_ID_KVD_SINGLE_SIZE,
+	MLXSW_RES_ID_KVD_DOUBLE_SIZE,
+	MLXSW_RES_ID_KVD_LINEAR_SIZE,
+
+	__MLXSW_RES_ID_MAX,
+};
+
+static u16 mlxsw_res_ids[] = {
+	[MLXSW_RES_ID_KVD_SIZE] = 0x1001,
+	[MLXSW_RES_ID_KVD_SINGLE_MIN_SIZE] = 0x1002,
+	[MLXSW_RES_ID_KVD_DOUBLE_MIN_SIZE] = 0x1003,
+	[MLXSW_RES_ID_PGT_SIZE] = 0x1004,
+	[MLXSW_RES_ID_MAX_KVD_LINEAR_RANGE] = 0x1005,
+	[MLXSW_RES_ID_MAX_KVD_ACTION_SETS] = 0x1007,
+	[MLXSW_RES_ID_MAX_TRAP_GROUPS] = 0x2201,
+	[MLXSW_RES_ID_CQE_V0] = 0x2210,
+	[MLXSW_RES_ID_CQE_V1] = 0x2211,
+	[MLXSW_RES_ID_CQE_V2] = 0x2212,
+	[MLXSW_RES_ID_COUNTER_POOL_SIZE] = 0x2410,
+	[MLXSW_RES_ID_COUNTER_BANK_SIZE] = 0x2411,
+	[MLXSW_RES_ID_MAX_SPAN] = 0x2420,
+	[MLXSW_RES_ID_COUNTER_SIZE_PACKETS_BYTES] = 0x2443,
+	[MLXSW_RES_ID_COUNTER_SIZE_ROUTER_BASIC] = 0x2449,
+	[MLXSW_RES_ID_MAX_SYSTEM_PORT] = 0x2502,
+	[MLXSW_RES_ID_FID] = 0x2512,
+	[MLXSW_RES_ID_MAX_LAG] = 0x2520,
+	[MLXSW_RES_ID_MAX_LAG_MEMBERS] = 0x2521,
+	[MLXSW_RES_ID_GUARANTEED_SHARED_BUFFER] = 0x2805,	/* Bytes */
+	[MLXSW_RES_ID_CELL_SIZE] = 0x2803,	/* Bytes */
+	[MLXSW_RES_ID_MAX_HEADROOM_SIZE] = 0x2811,	/* Bytes */
+	[MLXSW_RES_ID_ACL_MAX_TCAM_REGIONS] = 0x2901,
+	[MLXSW_RES_ID_ACL_MAX_TCAM_RULES] = 0x2902,
+	[MLXSW_RES_ID_ACL_MAX_REGIONS] = 0x2903,
+	[MLXSW_RES_ID_ACL_MAX_GROUPS] = 0x2904,
+	[MLXSW_RES_ID_ACL_MAX_GROUP_SIZE] = 0x2905,
+	[MLXSW_RES_ID_ACL_MAX_DEFAULT_ACTIONS] = 0x2908,
+	[MLXSW_RES_ID_ACL_FLEX_KEYS] = 0x2910,
+	[MLXSW_RES_ID_ACL_MAX_ACTION_PER_RULE] = 0x2911,
+	[MLXSW_RES_ID_ACL_ACTIONS_PER_SET] = 0x2912,
+	[MLXSW_RES_ID_ACL_MAX_ERPT_BANKS] = 0x2940,
+	[MLXSW_RES_ID_ACL_MAX_ERPT_BANK_SIZE] = 0x2941,
+	[MLXSW_RES_ID_ACL_MAX_LARGE_KEY_ID] = 0x2942,
+	[MLXSW_RES_ID_ACL_ERPT_ENTRIES_2KB] = 0x2950,
+	[MLXSW_RES_ID_ACL_ERPT_ENTRIES_4KB] = 0x2951,
+	[MLXSW_RES_ID_ACL_ERPT_ENTRIES_8KB] = 0x2952,
+	[MLXSW_RES_ID_ACL_ERPT_ENTRIES_12KB] = 0x2953,
+	[MLXSW_RES_ID_ACL_MAX_BF_LOG] = 0x2960,
+	[MLXSW_RES_ID_MAX_GLOBAL_POLICERS] = 0x2A10,
+	[MLXSW_RES_ID_MAX_CPU_POLICERS] = 0x2A13,
+	[MLXSW_RES_ID_MAX_VRS] = 0x2C01,
+	[MLXSW_RES_ID_MAX_RIFS] = 0x2C02,
+	[MLXSW_RES_ID_MC_ERIF_LIST_ENTRIES] = 0x2C10,
+	[MLXSW_RES_ID_MAX_RIF_MAC_PROFILES] = 0x2C14,
+	[MLXSW_RES_ID_MAX_LPM_TREES] = 0x2C30,
+	[MLXSW_RES_ID_MAX_NVE_MC_ENTRIES_IPV4] = 0x2E02,
+	[MLXSW_RES_ID_MAX_NVE_MC_ENTRIES_IPV6] = 0x2E03,
+};
+
+struct mlxsw_res {
+	bool valid[__MLXSW_RES_ID_MAX];
+	u64 values[__MLXSW_RES_ID_MAX];
+};
+
+static inline bool mlxsw_res_valid(struct mlxsw_res *res,
+				   enum mlxsw_res_id res_id)
+{
+	return res->valid[res_id];
+}
+
+#define MLXSW_RES_VALID(res, short_res_id)			\
+	mlxsw_res_valid(res, MLXSW_RES_ID_##short_res_id)
+
+static inline u64 mlxsw_res_get(struct mlxsw_res *res,
+				enum mlxsw_res_id res_id)
+{
+	if (WARN_ON(!res->valid[res_id]))
+		return 0;
+	return res->values[res_id];
+}
+
+#define MLXSW_RES_GET(res, short_res_id)			\
+	mlxsw_res_get(res, MLXSW_RES_ID_##short_res_id)
+
+static inline void mlxsw_res_set(struct mlxsw_res *res,
+				 enum mlxsw_res_id res_id, u64 value)
+{
+	res->valid[res_id] = true;
+	res->values[res_id] = value;
+}
+
+#define MLXSW_RES_SET(res, short_res_id, value)			\
+	mlxsw_res_set(res, MLXSW_RES_ID_##short_res_id, value)
+
+static inline void mlxsw_res_parse(struct mlxsw_res *res, u16 id, u64 value)
+{
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(mlxsw_res_ids); i++) {
+		if (mlxsw_res_ids[i] == id) {
+			mlxsw_res_set(res, i, value);
+			return;
+		}
+	}
+}
+
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
new file mode 100644
index 000000000..67ecdb9e7
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
@@ -0,0 +1,5331 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2015-2018 Mellanox Technologies. All rights reserved */
+
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/types.h>
+#include <linux/pci.h>
+#include <linux/netdevice.h>
+#include <linux/etherdevice.h>
+#include <linux/ethtool.h>
+#include <linux/slab.h>
+#include <linux/device.h>
+#include <linux/skbuff.h>
+#include <linux/if_vlan.h>
+#include <linux/if_bridge.h>
+#include <linux/workqueue.h>
+#include <linux/jiffies.h>
+#include <linux/bitops.h>
+#include <linux/list.h>
+#include <linux/notifier.h>
+#include <linux/dcbnl.h>
+#include <linux/inetdevice.h>
+#include <linux/netlink.h>
+#include <linux/jhash.h>
+#include <linux/log2.h>
+#include <linux/refcount.h>
+#include <linux/rhashtable.h>
+#include <net/switchdev.h>
+#include <net/pkt_cls.h>
+#include <net/netevent.h>
+#include <net/addrconf.h>
+#include <linux/ptp_classify.h>
+
+#include "spectrum.h"
+#include "pci.h"
+#include "core.h"
+#include "core_env.h"
+#include "reg.h"
+#include "port.h"
+#include "trap.h"
+#include "txheader.h"
+#include "spectrum_cnt.h"
+#include "spectrum_dpipe.h"
+#include "spectrum_acl_flex_actions.h"
+#include "spectrum_span.h"
+#include "spectrum_ptp.h"
+#include "spectrum_trap.h"
+
+#define MLXSW_SP_FWREV_MINOR 2010
+#define MLXSW_SP_FWREV_SUBMINOR 1006
+
+#define MLXSW_SP1_FWREV_MAJOR 13
+#define MLXSW_SP1_FWREV_CAN_RESET_MINOR 1702
+
+static const struct mlxsw_fw_rev mlxsw_sp1_fw_rev = {
+	.major = MLXSW_SP1_FWREV_MAJOR,
+	.minor = MLXSW_SP_FWREV_MINOR,
+	.subminor = MLXSW_SP_FWREV_SUBMINOR,
+	.can_reset_minor = MLXSW_SP1_FWREV_CAN_RESET_MINOR,
+};
+
+#define MLXSW_SP1_FW_FILENAME \
+	"mellanox/mlxsw_spectrum-" __stringify(MLXSW_SP1_FWREV_MAJOR) \
+	"." __stringify(MLXSW_SP_FWREV_MINOR) \
+	"." __stringify(MLXSW_SP_FWREV_SUBMINOR) ".mfa2"
+
+#define MLXSW_SP2_FWREV_MAJOR 29
+
+static const struct mlxsw_fw_rev mlxsw_sp2_fw_rev = {
+	.major = MLXSW_SP2_FWREV_MAJOR,
+	.minor = MLXSW_SP_FWREV_MINOR,
+	.subminor = MLXSW_SP_FWREV_SUBMINOR,
+};
+
+#define MLXSW_SP2_FW_FILENAME \
+	"mellanox/mlxsw_spectrum2-" __stringify(MLXSW_SP2_FWREV_MAJOR) \
+	"." __stringify(MLXSW_SP_FWREV_MINOR) \
+	"." __stringify(MLXSW_SP_FWREV_SUBMINOR) ".mfa2"
+
+#define MLXSW_SP3_FWREV_MAJOR 30
+
+static const struct mlxsw_fw_rev mlxsw_sp3_fw_rev = {
+	.major = MLXSW_SP3_FWREV_MAJOR,
+	.minor = MLXSW_SP_FWREV_MINOR,
+	.subminor = MLXSW_SP_FWREV_SUBMINOR,
+};
+
+#define MLXSW_SP3_FW_FILENAME \
+	"mellanox/mlxsw_spectrum3-" __stringify(MLXSW_SP3_FWREV_MAJOR) \
+	"." __stringify(MLXSW_SP_FWREV_MINOR) \
+	"." __stringify(MLXSW_SP_FWREV_SUBMINOR) ".mfa2"
+
+#define MLXSW_SP_LINECARDS_INI_BUNDLE_FILENAME \
+	"mellanox/lc_ini_bundle_" \
+	__stringify(MLXSW_SP_FWREV_MINOR) "_" \
+	__stringify(MLXSW_SP_FWREV_SUBMINOR) ".bin"
+
+static const char mlxsw_sp1_driver_name[] = "mlxsw_spectrum";
+static const char mlxsw_sp2_driver_name[] = "mlxsw_spectrum2";
+static const char mlxsw_sp3_driver_name[] = "mlxsw_spectrum3";
+static const char mlxsw_sp4_driver_name[] = "mlxsw_spectrum4";
+
+static const unsigned char mlxsw_sp1_mac_mask[ETH_ALEN] = {
+	0xff, 0xff, 0xff, 0xff, 0xfc, 0x00
+};
+static const unsigned char mlxsw_sp2_mac_mask[ETH_ALEN] = {
+	0xff, 0xff, 0xff, 0xff, 0xf0, 0x00
+};
+
+/* tx_hdr_version
+ * Tx header version.
+ * Must be set to 1.
+ */
+MLXSW_ITEM32(tx, hdr, version, 0x00, 28, 4);
+
+/* tx_hdr_ctl
+ * Packet control type.
+ * 0 - Ethernet control (e.g. EMADs, LACP)
+ * 1 - Ethernet data
+ */
+MLXSW_ITEM32(tx, hdr, ctl, 0x00, 26, 2);
+
+/* tx_hdr_proto
+ * Packet protocol type. Must be set to 1 (Ethernet).
+ */
+MLXSW_ITEM32(tx, hdr, proto, 0x00, 21, 3);
+
+/* tx_hdr_rx_is_router
+ * Packet is sent from the router. Valid for data packets only.
+ */
+MLXSW_ITEM32(tx, hdr, rx_is_router, 0x00, 19, 1);
+
+/* tx_hdr_fid_valid
+ * Indicates if the 'fid' field is valid and should be used for
+ * forwarding lookup. Valid for data packets only.
+ */
+MLXSW_ITEM32(tx, hdr, fid_valid, 0x00, 16, 1);
+
+/* tx_hdr_swid
+ * Switch partition ID. Must be set to 0.
+ */
+MLXSW_ITEM32(tx, hdr, swid, 0x00, 12, 3);
+
+/* tx_hdr_control_tclass
+ * Indicates if the packet should use the control TClass and not one
+ * of the data TClasses.
+ */
+MLXSW_ITEM32(tx, hdr, control_tclass, 0x00, 6, 1);
+
+/* tx_hdr_etclass
+ * Egress TClass to be used on the egress device on the egress port.
+ */
+MLXSW_ITEM32(tx, hdr, etclass, 0x00, 0, 4);
+
+/* tx_hdr_port_mid
+ * Destination local port for unicast packets.
+ * Destination multicast ID for multicast packets.
+ *
+ * Control packets are directed to a specific egress port, while data
+ * packets are transmitted through the CPU port (0) into the switch partition,
+ * where forwarding rules are applied.
+ */
+MLXSW_ITEM32(tx, hdr, port_mid, 0x04, 16, 16);
+
+/* tx_hdr_fid
+ * Forwarding ID used for L2 forwarding lookup. Valid only if 'fid_valid' is
+ * set, otherwise calculated based on the packet's VID using VID to FID mapping.
+ * Valid for data packets only.
+ */
+MLXSW_ITEM32(tx, hdr, fid, 0x08, 16, 16);
+
+/* tx_hdr_type
+ * 0 - Data packets
+ * 6 - Control packets
+ */
+MLXSW_ITEM32(tx, hdr, type, 0x0C, 0, 4);
+
+int mlxsw_sp_flow_counter_get(struct mlxsw_sp *mlxsw_sp,
+			      unsigned int counter_index, u64 *packets,
+			      u64 *bytes)
+{
+	char mgpc_pl[MLXSW_REG_MGPC_LEN];
+	int err;
+
+	mlxsw_reg_mgpc_pack(mgpc_pl, counter_index, MLXSW_REG_MGPC_OPCODE_NOP,
+			    MLXSW_REG_FLOW_COUNTER_SET_TYPE_PACKETS_BYTES);
+	err = mlxsw_reg_query(mlxsw_sp->core, MLXSW_REG(mgpc), mgpc_pl);
+	if (err)
+		return err;
+	if (packets)
+		*packets = mlxsw_reg_mgpc_packet_counter_get(mgpc_pl);
+	if (bytes)
+		*bytes = mlxsw_reg_mgpc_byte_counter_get(mgpc_pl);
+	return 0;
+}
+
+static int mlxsw_sp_flow_counter_clear(struct mlxsw_sp *mlxsw_sp,
+				       unsigned int counter_index)
+{
+	char mgpc_pl[MLXSW_REG_MGPC_LEN];
+
+	mlxsw_reg_mgpc_pack(mgpc_pl, counter_index, MLXSW_REG_MGPC_OPCODE_CLEAR,
+			    MLXSW_REG_FLOW_COUNTER_SET_TYPE_PACKETS_BYTES);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(mgpc), mgpc_pl);
+}
+
+int mlxsw_sp_flow_counter_alloc(struct mlxsw_sp *mlxsw_sp,
+				unsigned int *p_counter_index)
+{
+	int err;
+
+	err = mlxsw_sp_counter_alloc(mlxsw_sp, MLXSW_SP_COUNTER_SUB_POOL_FLOW,
+				     p_counter_index);
+	if (err)
+		return err;
+	err = mlxsw_sp_flow_counter_clear(mlxsw_sp, *p_counter_index);
+	if (err)
+		goto err_counter_clear;
+	return 0;
+
+err_counter_clear:
+	mlxsw_sp_counter_free(mlxsw_sp, MLXSW_SP_COUNTER_SUB_POOL_FLOW,
+			      *p_counter_index);
+	return err;
+}
+
+void mlxsw_sp_flow_counter_free(struct mlxsw_sp *mlxsw_sp,
+				unsigned int counter_index)
+{
+	 mlxsw_sp_counter_free(mlxsw_sp, MLXSW_SP_COUNTER_SUB_POOL_FLOW,
+			       counter_index);
+}
+
+void mlxsw_sp_txhdr_construct(struct sk_buff *skb,
+			      const struct mlxsw_tx_info *tx_info)
+{
+	char *txhdr = skb_push(skb, MLXSW_TXHDR_LEN);
+
+	memset(txhdr, 0, MLXSW_TXHDR_LEN);
+
+	mlxsw_tx_hdr_version_set(txhdr, MLXSW_TXHDR_VERSION_1);
+	mlxsw_tx_hdr_ctl_set(txhdr, MLXSW_TXHDR_ETH_CTL);
+	mlxsw_tx_hdr_proto_set(txhdr, MLXSW_TXHDR_PROTO_ETH);
+	mlxsw_tx_hdr_swid_set(txhdr, 0);
+	mlxsw_tx_hdr_control_tclass_set(txhdr, 1);
+	mlxsw_tx_hdr_port_mid_set(txhdr, tx_info->local_port);
+	mlxsw_tx_hdr_type_set(txhdr, MLXSW_TXHDR_TYPE_CONTROL);
+}
+
+int
+mlxsw_sp_txhdr_ptp_data_construct(struct mlxsw_core *mlxsw_core,
+				  struct mlxsw_sp_port *mlxsw_sp_port,
+				  struct sk_buff *skb,
+				  const struct mlxsw_tx_info *tx_info)
+{
+	char *txhdr;
+	u16 max_fid;
+	int err;
+
+	if (skb_cow_head(skb, MLXSW_TXHDR_LEN)) {
+		err = -ENOMEM;
+		goto err_skb_cow_head;
+	}
+
+	if (!MLXSW_CORE_RES_VALID(mlxsw_core, FID)) {
+		err = -EIO;
+		goto err_res_valid;
+	}
+	max_fid = MLXSW_CORE_RES_GET(mlxsw_core, FID);
+
+	txhdr = skb_push(skb, MLXSW_TXHDR_LEN);
+	memset(txhdr, 0, MLXSW_TXHDR_LEN);
+
+	mlxsw_tx_hdr_version_set(txhdr, MLXSW_TXHDR_VERSION_1);
+	mlxsw_tx_hdr_proto_set(txhdr, MLXSW_TXHDR_PROTO_ETH);
+	mlxsw_tx_hdr_rx_is_router_set(txhdr, true);
+	mlxsw_tx_hdr_fid_valid_set(txhdr, true);
+	mlxsw_tx_hdr_fid_set(txhdr, max_fid + tx_info->local_port - 1);
+	mlxsw_tx_hdr_type_set(txhdr, MLXSW_TXHDR_TYPE_DATA);
+	return 0;
+
+err_res_valid:
+err_skb_cow_head:
+	this_cpu_inc(mlxsw_sp_port->pcpu_stats->tx_dropped);
+	dev_kfree_skb_any(skb);
+	return err;
+}
+
+static bool mlxsw_sp_skb_requires_ts(struct sk_buff *skb)
+{
+	unsigned int type;
+
+	if (!(skb_shinfo(skb)->tx_flags & SKBTX_HW_TSTAMP))
+		return false;
+
+	type = ptp_classify_raw(skb);
+	return !!ptp_parse_header(skb, type);
+}
+
+static int mlxsw_sp_txhdr_handle(struct mlxsw_core *mlxsw_core,
+				 struct mlxsw_sp_port *mlxsw_sp_port,
+				 struct sk_buff *skb,
+				 const struct mlxsw_tx_info *tx_info)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_core_driver_priv(mlxsw_core);
+
+	/* In Spectrum-2 and Spectrum-3, PTP events that require a time stamp
+	 * need special handling and cannot be transmitted as regular control
+	 * packets.
+	 */
+	if (unlikely(mlxsw_sp_skb_requires_ts(skb)))
+		return mlxsw_sp->ptp_ops->txhdr_construct(mlxsw_core,
+							  mlxsw_sp_port, skb,
+							  tx_info);
+
+	if (skb_cow_head(skb, MLXSW_TXHDR_LEN)) {
+		this_cpu_inc(mlxsw_sp_port->pcpu_stats->tx_dropped);
+		dev_kfree_skb_any(skb);
+		return -ENOMEM;
+	}
+
+	mlxsw_sp_txhdr_construct(skb, tx_info);
+	return 0;
+}
+
+enum mlxsw_reg_spms_state mlxsw_sp_stp_spms_state(u8 state)
+{
+	switch (state) {
+	case BR_STATE_FORWARDING:
+		return MLXSW_REG_SPMS_STATE_FORWARDING;
+	case BR_STATE_LEARNING:
+		return MLXSW_REG_SPMS_STATE_LEARNING;
+	case BR_STATE_LISTENING:
+	case BR_STATE_DISABLED:
+	case BR_STATE_BLOCKING:
+		return MLXSW_REG_SPMS_STATE_DISCARDING;
+	default:
+		BUG();
+	}
+}
+
+int mlxsw_sp_port_vid_stp_set(struct mlxsw_sp_port *mlxsw_sp_port, u16 vid,
+			      u8 state)
+{
+	enum mlxsw_reg_spms_state spms_state = mlxsw_sp_stp_spms_state(state);
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	char *spms_pl;
+	int err;
+
+	spms_pl = kmalloc(MLXSW_REG_SPMS_LEN, GFP_KERNEL);
+	if (!spms_pl)
+		return -ENOMEM;
+	mlxsw_reg_spms_pack(spms_pl, mlxsw_sp_port->local_port);
+	mlxsw_reg_spms_vid_pack(spms_pl, vid, spms_state);
+
+	err = mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(spms), spms_pl);
+	kfree(spms_pl);
+	return err;
+}
+
+static int mlxsw_sp_base_mac_get(struct mlxsw_sp *mlxsw_sp)
+{
+	char spad_pl[MLXSW_REG_SPAD_LEN] = {0};
+	int err;
+
+	err = mlxsw_reg_query(mlxsw_sp->core, MLXSW_REG(spad), spad_pl);
+	if (err)
+		return err;
+	mlxsw_reg_spad_base_mac_memcpy_from(spad_pl, mlxsw_sp->base_mac);
+	return 0;
+}
+
+int mlxsw_sp_port_admin_status_set(struct mlxsw_sp_port *mlxsw_sp_port,
+				   bool is_up)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	char paos_pl[MLXSW_REG_PAOS_LEN];
+
+	mlxsw_reg_paos_pack(paos_pl, mlxsw_sp_port->local_port,
+			    is_up ? MLXSW_PORT_ADMIN_STATUS_UP :
+			    MLXSW_PORT_ADMIN_STATUS_DOWN);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(paos), paos_pl);
+}
+
+static int mlxsw_sp_port_dev_addr_set(struct mlxsw_sp_port *mlxsw_sp_port,
+				      const unsigned char *addr)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	char ppad_pl[MLXSW_REG_PPAD_LEN];
+
+	mlxsw_reg_ppad_pack(ppad_pl, true, mlxsw_sp_port->local_port);
+	mlxsw_reg_ppad_mac_memcpy_to(ppad_pl, addr);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(ppad), ppad_pl);
+}
+
+static int mlxsw_sp_port_dev_addr_init(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+
+	eth_hw_addr_gen(mlxsw_sp_port->dev, mlxsw_sp->base_mac,
+			mlxsw_sp_port->local_port);
+	return mlxsw_sp_port_dev_addr_set(mlxsw_sp_port,
+					  mlxsw_sp_port->dev->dev_addr);
+}
+
+static int mlxsw_sp_port_max_mtu_get(struct mlxsw_sp_port *mlxsw_sp_port, int *p_max_mtu)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	char pmtu_pl[MLXSW_REG_PMTU_LEN];
+	int err;
+
+	mlxsw_reg_pmtu_pack(pmtu_pl, mlxsw_sp_port->local_port, 0);
+	err = mlxsw_reg_query(mlxsw_sp->core, MLXSW_REG(pmtu), pmtu_pl);
+	if (err)
+		return err;
+
+	*p_max_mtu = mlxsw_reg_pmtu_max_mtu_get(pmtu_pl);
+	return 0;
+}
+
+static int mlxsw_sp_port_mtu_set(struct mlxsw_sp_port *mlxsw_sp_port, u16 mtu)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	char pmtu_pl[MLXSW_REG_PMTU_LEN];
+
+	mtu += MLXSW_TXHDR_LEN + ETH_HLEN;
+	if (mtu > mlxsw_sp_port->max_mtu)
+		return -EINVAL;
+
+	mlxsw_reg_pmtu_pack(pmtu_pl, mlxsw_sp_port->local_port, mtu);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(pmtu), pmtu_pl);
+}
+
+static int mlxsw_sp_port_swid_set(struct mlxsw_sp *mlxsw_sp,
+				  u16 local_port, u8 swid)
+{
+	char pspa_pl[MLXSW_REG_PSPA_LEN];
+
+	mlxsw_reg_pspa_pack(pspa_pl, swid, local_port);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(pspa), pspa_pl);
+}
+
+int mlxsw_sp_port_vp_mode_set(struct mlxsw_sp_port *mlxsw_sp_port, bool enable)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	char svpe_pl[MLXSW_REG_SVPE_LEN];
+
+	mlxsw_reg_svpe_pack(svpe_pl, mlxsw_sp_port->local_port, enable);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(svpe), svpe_pl);
+}
+
+int mlxsw_sp_port_vid_learning_set(struct mlxsw_sp_port *mlxsw_sp_port, u16 vid,
+				   bool learn_enable)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	char *spvmlr_pl;
+	int err;
+
+	spvmlr_pl = kmalloc(MLXSW_REG_SPVMLR_LEN, GFP_KERNEL);
+	if (!spvmlr_pl)
+		return -ENOMEM;
+	mlxsw_reg_spvmlr_pack(spvmlr_pl, mlxsw_sp_port->local_port, vid, vid,
+			      learn_enable);
+	err = mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(spvmlr), spvmlr_pl);
+	kfree(spvmlr_pl);
+	return err;
+}
+
+int mlxsw_sp_ethtype_to_sver_type(u16 ethtype, u8 *p_sver_type)
+{
+	switch (ethtype) {
+	case ETH_P_8021Q:
+		*p_sver_type = 0;
+		break;
+	case ETH_P_8021AD:
+		*p_sver_type = 1;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+int mlxsw_sp_port_egress_ethtype_set(struct mlxsw_sp_port *mlxsw_sp_port,
+				     u16 ethtype)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	char spevet_pl[MLXSW_REG_SPEVET_LEN];
+	u8 sver_type;
+	int err;
+
+	err = mlxsw_sp_ethtype_to_sver_type(ethtype, &sver_type);
+	if (err)
+		return err;
+
+	mlxsw_reg_spevet_pack(spevet_pl, mlxsw_sp_port->local_port, sver_type);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(spevet), spevet_pl);
+}
+
+static int __mlxsw_sp_port_pvid_set(struct mlxsw_sp_port *mlxsw_sp_port,
+				    u16 vid, u16 ethtype)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	char spvid_pl[MLXSW_REG_SPVID_LEN];
+	u8 sver_type;
+	int err;
+
+	err = mlxsw_sp_ethtype_to_sver_type(ethtype, &sver_type);
+	if (err)
+		return err;
+
+	mlxsw_reg_spvid_pack(spvid_pl, mlxsw_sp_port->local_port, vid,
+			     sver_type);
+
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(spvid), spvid_pl);
+}
+
+static int mlxsw_sp_port_allow_untagged_set(struct mlxsw_sp_port *mlxsw_sp_port,
+					    bool allow)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	char spaft_pl[MLXSW_REG_SPAFT_LEN];
+
+	mlxsw_reg_spaft_pack(spaft_pl, mlxsw_sp_port->local_port, allow);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(spaft), spaft_pl);
+}
+
+int mlxsw_sp_port_pvid_set(struct mlxsw_sp_port *mlxsw_sp_port, u16 vid,
+			   u16 ethtype)
+{
+	int err;
+
+	if (!vid) {
+		err = mlxsw_sp_port_allow_untagged_set(mlxsw_sp_port, false);
+		if (err)
+			return err;
+	} else {
+		err = __mlxsw_sp_port_pvid_set(mlxsw_sp_port, vid, ethtype);
+		if (err)
+			return err;
+		err = mlxsw_sp_port_allow_untagged_set(mlxsw_sp_port, true);
+		if (err)
+			goto err_port_allow_untagged_set;
+	}
+
+	mlxsw_sp_port->pvid = vid;
+	return 0;
+
+err_port_allow_untagged_set:
+	__mlxsw_sp_port_pvid_set(mlxsw_sp_port, mlxsw_sp_port->pvid, ethtype);
+	return err;
+}
+
+static int
+mlxsw_sp_port_system_port_mapping_set(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	char sspr_pl[MLXSW_REG_SSPR_LEN];
+
+	mlxsw_reg_sspr_pack(sspr_pl, mlxsw_sp_port->local_port);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(sspr), sspr_pl);
+}
+
+static int
+mlxsw_sp_port_module_info_parse(struct mlxsw_sp *mlxsw_sp,
+				u16 local_port, char *pmlp_pl,
+				struct mlxsw_sp_port_mapping *port_mapping)
+{
+	bool separate_rxtx;
+	u8 first_lane;
+	u8 slot_index;
+	u8 module;
+	u8 width;
+	int i;
+
+	module = mlxsw_reg_pmlp_module_get(pmlp_pl, 0);
+	slot_index = mlxsw_reg_pmlp_slot_index_get(pmlp_pl, 0);
+	width = mlxsw_reg_pmlp_width_get(pmlp_pl);
+	separate_rxtx = mlxsw_reg_pmlp_rxtx_get(pmlp_pl);
+	first_lane = mlxsw_reg_pmlp_tx_lane_get(pmlp_pl, 0);
+
+	if (width && !is_power_of_2(width)) {
+		dev_err(mlxsw_sp->bus_info->dev, "Port %d: Unsupported module config: width value is not power of 2\n",
+			local_port);
+		return -EINVAL;
+	}
+
+	for (i = 0; i < width; i++) {
+		if (mlxsw_reg_pmlp_module_get(pmlp_pl, i) != module) {
+			dev_err(mlxsw_sp->bus_info->dev, "Port %d: Unsupported module config: contains multiple modules\n",
+				local_port);
+			return -EINVAL;
+		}
+		if (mlxsw_reg_pmlp_slot_index_get(pmlp_pl, i) != slot_index) {
+			dev_err(mlxsw_sp->bus_info->dev, "Port %d: Unsupported module config: contains multiple slot indexes\n",
+				local_port);
+			return -EINVAL;
+		}
+		if (separate_rxtx &&
+		    mlxsw_reg_pmlp_tx_lane_get(pmlp_pl, i) !=
+		    mlxsw_reg_pmlp_rx_lane_get(pmlp_pl, i)) {
+			dev_err(mlxsw_sp->bus_info->dev, "Port %d: Unsupported module config: TX and RX lane numbers are different\n",
+				local_port);
+			return -EINVAL;
+		}
+		if (mlxsw_reg_pmlp_tx_lane_get(pmlp_pl, i) != i + first_lane) {
+			dev_err(mlxsw_sp->bus_info->dev, "Port %d: Unsupported module config: TX and RX lane numbers are not sequential\n",
+				local_port);
+			return -EINVAL;
+		}
+	}
+
+	port_mapping->module = module;
+	port_mapping->slot_index = slot_index;
+	port_mapping->width = width;
+	port_mapping->module_width = width;
+	port_mapping->lane = mlxsw_reg_pmlp_tx_lane_get(pmlp_pl, 0);
+	return 0;
+}
+
+static int
+mlxsw_sp_port_module_info_get(struct mlxsw_sp *mlxsw_sp, u16 local_port,
+			      struct mlxsw_sp_port_mapping *port_mapping)
+{
+	char pmlp_pl[MLXSW_REG_PMLP_LEN];
+	int err;
+
+	mlxsw_reg_pmlp_pack(pmlp_pl, local_port);
+	err = mlxsw_reg_query(mlxsw_sp->core, MLXSW_REG(pmlp), pmlp_pl);
+	if (err)
+		return err;
+	return mlxsw_sp_port_module_info_parse(mlxsw_sp, local_port,
+					       pmlp_pl, port_mapping);
+}
+
+static int
+mlxsw_sp_port_module_map(struct mlxsw_sp *mlxsw_sp, u16 local_port,
+			 const struct mlxsw_sp_port_mapping *port_mapping)
+{
+	char pmlp_pl[MLXSW_REG_PMLP_LEN];
+	int i, err;
+
+	mlxsw_env_module_port_map(mlxsw_sp->core, port_mapping->slot_index,
+				  port_mapping->module);
+
+	mlxsw_reg_pmlp_pack(pmlp_pl, local_port);
+	mlxsw_reg_pmlp_width_set(pmlp_pl, port_mapping->width);
+	for (i = 0; i < port_mapping->width; i++) {
+		mlxsw_reg_pmlp_slot_index_set(pmlp_pl, i,
+					      port_mapping->slot_index);
+		mlxsw_reg_pmlp_module_set(pmlp_pl, i, port_mapping->module);
+		mlxsw_reg_pmlp_tx_lane_set(pmlp_pl, i, port_mapping->lane + i); /* Rx & Tx */
+	}
+
+	err = mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(pmlp), pmlp_pl);
+	if (err)
+		goto err_pmlp_write;
+	return 0;
+
+err_pmlp_write:
+	mlxsw_env_module_port_unmap(mlxsw_sp->core, port_mapping->slot_index,
+				    port_mapping->module);
+	return err;
+}
+
+static void mlxsw_sp_port_module_unmap(struct mlxsw_sp *mlxsw_sp, u16 local_port,
+				       u8 slot_index, u8 module)
+{
+	char pmlp_pl[MLXSW_REG_PMLP_LEN];
+
+	mlxsw_reg_pmlp_pack(pmlp_pl, local_port);
+	mlxsw_reg_pmlp_width_set(pmlp_pl, 0);
+	mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(pmlp), pmlp_pl);
+	mlxsw_env_module_port_unmap(mlxsw_sp->core, slot_index, module);
+}
+
+static int mlxsw_sp_port_open(struct net_device *dev)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	int err;
+
+	err = mlxsw_env_module_port_up(mlxsw_sp->core,
+				       mlxsw_sp_port->mapping.slot_index,
+				       mlxsw_sp_port->mapping.module);
+	if (err)
+		return err;
+	err = mlxsw_sp_port_admin_status_set(mlxsw_sp_port, true);
+	if (err)
+		goto err_port_admin_status_set;
+	netif_start_queue(dev);
+	return 0;
+
+err_port_admin_status_set:
+	mlxsw_env_module_port_down(mlxsw_sp->core,
+				   mlxsw_sp_port->mapping.slot_index,
+				   mlxsw_sp_port->mapping.module);
+	return err;
+}
+
+static int mlxsw_sp_port_stop(struct net_device *dev)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+
+	netif_stop_queue(dev);
+	mlxsw_sp_port_admin_status_set(mlxsw_sp_port, false);
+	mlxsw_env_module_port_down(mlxsw_sp->core,
+				   mlxsw_sp_port->mapping.slot_index,
+				   mlxsw_sp_port->mapping.module);
+	return 0;
+}
+
+static netdev_tx_t mlxsw_sp_port_xmit(struct sk_buff *skb,
+				      struct net_device *dev)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	struct mlxsw_sp_port_pcpu_stats *pcpu_stats;
+	const struct mlxsw_tx_info tx_info = {
+		.local_port = mlxsw_sp_port->local_port,
+		.is_emad = false,
+	};
+	u64 len;
+	int err;
+
+	memset(skb->cb, 0, sizeof(struct mlxsw_skb_cb));
+
+	if (mlxsw_core_skb_transmit_busy(mlxsw_sp->core, &tx_info))
+		return NETDEV_TX_BUSY;
+
+	if (eth_skb_pad(skb)) {
+		this_cpu_inc(mlxsw_sp_port->pcpu_stats->tx_dropped);
+		return NETDEV_TX_OK;
+	}
+
+	err = mlxsw_sp_txhdr_handle(mlxsw_sp->core, mlxsw_sp_port, skb,
+				    &tx_info);
+	if (err)
+		return NETDEV_TX_OK;
+
+	/* TX header is consumed by HW on the way so we shouldn't count its
+	 * bytes as being sent.
+	 */
+	len = skb->len - MLXSW_TXHDR_LEN;
+
+	/* Due to a race we might fail here because of a full queue. In that
+	 * unlikely case we simply drop the packet.
+	 */
+	err = mlxsw_core_skb_transmit(mlxsw_sp->core, skb, &tx_info);
+
+	if (!err) {
+		pcpu_stats = this_cpu_ptr(mlxsw_sp_port->pcpu_stats);
+		u64_stats_update_begin(&pcpu_stats->syncp);
+		pcpu_stats->tx_packets++;
+		pcpu_stats->tx_bytes += len;
+		u64_stats_update_end(&pcpu_stats->syncp);
+	} else {
+		this_cpu_inc(mlxsw_sp_port->pcpu_stats->tx_dropped);
+		dev_kfree_skb_any(skb);
+	}
+	return NETDEV_TX_OK;
+}
+
+static void mlxsw_sp_set_rx_mode(struct net_device *dev)
+{
+}
+
+static int mlxsw_sp_port_set_mac_address(struct net_device *dev, void *p)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+	struct sockaddr *addr = p;
+	int err;
+
+	if (!is_valid_ether_addr(addr->sa_data))
+		return -EADDRNOTAVAIL;
+
+	err = mlxsw_sp_port_dev_addr_set(mlxsw_sp_port, addr->sa_data);
+	if (err)
+		return err;
+	eth_hw_addr_set(dev, addr->sa_data);
+	return 0;
+}
+
+static int mlxsw_sp_port_change_mtu(struct net_device *dev, int mtu)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+	struct mlxsw_sp_hdroom orig_hdroom;
+	struct mlxsw_sp_hdroom hdroom;
+	int err;
+
+	orig_hdroom = *mlxsw_sp_port->hdroom;
+
+	hdroom = orig_hdroom;
+	hdroom.mtu = mtu;
+	mlxsw_sp_hdroom_bufs_reset_sizes(mlxsw_sp_port, &hdroom);
+
+	err = mlxsw_sp_hdroom_configure(mlxsw_sp_port, &hdroom);
+	if (err) {
+		netdev_err(dev, "Failed to configure port's headroom\n");
+		return err;
+	}
+
+	err = mlxsw_sp_port_mtu_set(mlxsw_sp_port, mtu);
+	if (err)
+		goto err_port_mtu_set;
+	dev->mtu = mtu;
+	return 0;
+
+err_port_mtu_set:
+	mlxsw_sp_hdroom_configure(mlxsw_sp_port, &orig_hdroom);
+	return err;
+}
+
+static int
+mlxsw_sp_port_get_sw_stats64(const struct net_device *dev,
+			     struct rtnl_link_stats64 *stats)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+	struct mlxsw_sp_port_pcpu_stats *p;
+	u64 rx_packets, rx_bytes, tx_packets, tx_bytes;
+	u32 tx_dropped = 0;
+	unsigned int start;
+	int i;
+
+	for_each_possible_cpu(i) {
+		p = per_cpu_ptr(mlxsw_sp_port->pcpu_stats, i);
+		do {
+			start = u64_stats_fetch_begin_irq(&p->syncp);
+			rx_packets	= p->rx_packets;
+			rx_bytes	= p->rx_bytes;
+			tx_packets	= p->tx_packets;
+			tx_bytes	= p->tx_bytes;
+		} while (u64_stats_fetch_retry_irq(&p->syncp, start));
+
+		stats->rx_packets	+= rx_packets;
+		stats->rx_bytes		+= rx_bytes;
+		stats->tx_packets	+= tx_packets;
+		stats->tx_bytes		+= tx_bytes;
+		/* tx_dropped is u32, updated without syncp protection. */
+		tx_dropped	+= p->tx_dropped;
+	}
+	stats->tx_dropped	= tx_dropped;
+	return 0;
+}
+
+static bool mlxsw_sp_port_has_offload_stats(const struct net_device *dev, int attr_id)
+{
+	switch (attr_id) {
+	case IFLA_OFFLOAD_XSTATS_CPU_HIT:
+		return true;
+	}
+
+	return false;
+}
+
+static int mlxsw_sp_port_get_offload_stats(int attr_id, const struct net_device *dev,
+					   void *sp)
+{
+	switch (attr_id) {
+	case IFLA_OFFLOAD_XSTATS_CPU_HIT:
+		return mlxsw_sp_port_get_sw_stats64(dev, sp);
+	}
+
+	return -EINVAL;
+}
+
+int mlxsw_sp_port_get_stats_raw(struct net_device *dev, int grp,
+				int prio, char *ppcnt_pl)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+
+	mlxsw_reg_ppcnt_pack(ppcnt_pl, mlxsw_sp_port->local_port, grp, prio);
+	return mlxsw_reg_query(mlxsw_sp->core, MLXSW_REG(ppcnt), ppcnt_pl);
+}
+
+static int mlxsw_sp_port_get_hw_stats(struct net_device *dev,
+				      struct rtnl_link_stats64 *stats)
+{
+	char ppcnt_pl[MLXSW_REG_PPCNT_LEN];
+	int err;
+
+	err = mlxsw_sp_port_get_stats_raw(dev, MLXSW_REG_PPCNT_IEEE_8023_CNT,
+					  0, ppcnt_pl);
+	if (err)
+		goto out;
+
+	stats->tx_packets =
+		mlxsw_reg_ppcnt_a_frames_transmitted_ok_get(ppcnt_pl);
+	stats->rx_packets =
+		mlxsw_reg_ppcnt_a_frames_received_ok_get(ppcnt_pl);
+	stats->tx_bytes =
+		mlxsw_reg_ppcnt_a_octets_transmitted_ok_get(ppcnt_pl);
+	stats->rx_bytes =
+		mlxsw_reg_ppcnt_a_octets_received_ok_get(ppcnt_pl);
+	stats->multicast =
+		mlxsw_reg_ppcnt_a_multicast_frames_received_ok_get(ppcnt_pl);
+
+	stats->rx_crc_errors =
+		mlxsw_reg_ppcnt_a_frame_check_sequence_errors_get(ppcnt_pl);
+	stats->rx_frame_errors =
+		mlxsw_reg_ppcnt_a_alignment_errors_get(ppcnt_pl);
+
+	stats->rx_length_errors = (
+		mlxsw_reg_ppcnt_a_in_range_length_errors_get(ppcnt_pl) +
+		mlxsw_reg_ppcnt_a_out_of_range_length_field_get(ppcnt_pl) +
+		mlxsw_reg_ppcnt_a_frame_too_long_errors_get(ppcnt_pl));
+
+	stats->rx_errors = (stats->rx_crc_errors +
+		stats->rx_frame_errors + stats->rx_length_errors);
+
+out:
+	return err;
+}
+
+static void
+mlxsw_sp_port_get_hw_xstats(struct net_device *dev,
+			    struct mlxsw_sp_port_xstats *xstats)
+{
+	char ppcnt_pl[MLXSW_REG_PPCNT_LEN];
+	int err, i;
+
+	err = mlxsw_sp_port_get_stats_raw(dev, MLXSW_REG_PPCNT_EXT_CNT, 0,
+					  ppcnt_pl);
+	if (!err)
+		xstats->ecn = mlxsw_reg_ppcnt_ecn_marked_get(ppcnt_pl);
+
+	for (i = 0; i < TC_MAX_QUEUE; i++) {
+		err = mlxsw_sp_port_get_stats_raw(dev,
+						  MLXSW_REG_PPCNT_TC_CONG_CNT,
+						  i, ppcnt_pl);
+		if (err)
+			goto tc_cnt;
+
+		xstats->wred_drop[i] =
+			mlxsw_reg_ppcnt_wred_discard_get(ppcnt_pl);
+		xstats->tc_ecn[i] = mlxsw_reg_ppcnt_ecn_marked_tc_get(ppcnt_pl);
+
+tc_cnt:
+		err = mlxsw_sp_port_get_stats_raw(dev, MLXSW_REG_PPCNT_TC_CNT,
+						  i, ppcnt_pl);
+		if (err)
+			continue;
+
+		xstats->backlog[i] =
+			mlxsw_reg_ppcnt_tc_transmit_queue_get(ppcnt_pl);
+		xstats->tail_drop[i] =
+			mlxsw_reg_ppcnt_tc_no_buffer_discard_uc_get(ppcnt_pl);
+	}
+
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) {
+		err = mlxsw_sp_port_get_stats_raw(dev, MLXSW_REG_PPCNT_PRIO_CNT,
+						  i, ppcnt_pl);
+		if (err)
+			continue;
+
+		xstats->tx_packets[i] = mlxsw_reg_ppcnt_tx_frames_get(ppcnt_pl);
+		xstats->tx_bytes[i] = mlxsw_reg_ppcnt_tx_octets_get(ppcnt_pl);
+	}
+}
+
+static void update_stats_cache(struct work_struct *work)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port =
+		container_of(work, struct mlxsw_sp_port,
+			     periodic_hw_stats.update_dw.work);
+
+	if (!netif_carrier_ok(mlxsw_sp_port->dev))
+		/* Note: mlxsw_sp_port_down_wipe_counters() clears the cache as
+		 * necessary when port goes down.
+		 */
+		goto out;
+
+	mlxsw_sp_port_get_hw_stats(mlxsw_sp_port->dev,
+				   &mlxsw_sp_port->periodic_hw_stats.stats);
+	mlxsw_sp_port_get_hw_xstats(mlxsw_sp_port->dev,
+				    &mlxsw_sp_port->periodic_hw_stats.xstats);
+
+out:
+	mlxsw_core_schedule_dw(&mlxsw_sp_port->periodic_hw_stats.update_dw,
+			       MLXSW_HW_STATS_UPDATE_TIME);
+}
+
+/* Return the stats from a cache that is updated periodically,
+ * as this function might get called in an atomic context.
+ */
+static void
+mlxsw_sp_port_get_stats64(struct net_device *dev,
+			  struct rtnl_link_stats64 *stats)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+
+	memcpy(stats, &mlxsw_sp_port->periodic_hw_stats.stats, sizeof(*stats));
+}
+
+static int __mlxsw_sp_port_vlan_set(struct mlxsw_sp_port *mlxsw_sp_port,
+				    u16 vid_begin, u16 vid_end,
+				    bool is_member, bool untagged)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	char *spvm_pl;
+	int err;
+
+	spvm_pl = kmalloc(MLXSW_REG_SPVM_LEN, GFP_KERNEL);
+	if (!spvm_pl)
+		return -ENOMEM;
+
+	mlxsw_reg_spvm_pack(spvm_pl, mlxsw_sp_port->local_port,	vid_begin,
+			    vid_end, is_member, untagged);
+	err = mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(spvm), spvm_pl);
+	kfree(spvm_pl);
+	return err;
+}
+
+int mlxsw_sp_port_vlan_set(struct mlxsw_sp_port *mlxsw_sp_port, u16 vid_begin,
+			   u16 vid_end, bool is_member, bool untagged)
+{
+	u16 vid, vid_e;
+	int err;
+
+	for (vid = vid_begin; vid <= vid_end;
+	     vid += MLXSW_REG_SPVM_REC_MAX_COUNT) {
+		vid_e = min((u16) (vid + MLXSW_REG_SPVM_REC_MAX_COUNT - 1),
+			    vid_end);
+
+		err = __mlxsw_sp_port_vlan_set(mlxsw_sp_port, vid, vid_e,
+					       is_member, untagged);
+		if (err)
+			return err;
+	}
+
+	return 0;
+}
+
+static void mlxsw_sp_port_vlan_flush(struct mlxsw_sp_port *mlxsw_sp_port,
+				     bool flush_default)
+{
+	struct mlxsw_sp_port_vlan *mlxsw_sp_port_vlan, *tmp;
+
+	list_for_each_entry_safe(mlxsw_sp_port_vlan, tmp,
+				 &mlxsw_sp_port->vlans_list, list) {
+		if (!flush_default &&
+		    mlxsw_sp_port_vlan->vid == MLXSW_SP_DEFAULT_VID)
+			continue;
+		mlxsw_sp_port_vlan_destroy(mlxsw_sp_port_vlan);
+	}
+}
+
+static void
+mlxsw_sp_port_vlan_cleanup(struct mlxsw_sp_port_vlan *mlxsw_sp_port_vlan)
+{
+	if (mlxsw_sp_port_vlan->bridge_port)
+		mlxsw_sp_port_vlan_bridge_leave(mlxsw_sp_port_vlan);
+	else if (mlxsw_sp_port_vlan->fid)
+		mlxsw_sp_port_vlan_router_leave(mlxsw_sp_port_vlan);
+}
+
+struct mlxsw_sp_port_vlan *
+mlxsw_sp_port_vlan_create(struct mlxsw_sp_port *mlxsw_sp_port, u16 vid)
+{
+	struct mlxsw_sp_port_vlan *mlxsw_sp_port_vlan;
+	bool untagged = vid == MLXSW_SP_DEFAULT_VID;
+	int err;
+
+	mlxsw_sp_port_vlan = mlxsw_sp_port_vlan_find_by_vid(mlxsw_sp_port, vid);
+	if (mlxsw_sp_port_vlan)
+		return ERR_PTR(-EEXIST);
+
+	err = mlxsw_sp_port_vlan_set(mlxsw_sp_port, vid, vid, true, untagged);
+	if (err)
+		return ERR_PTR(err);
+
+	mlxsw_sp_port_vlan = kzalloc(sizeof(*mlxsw_sp_port_vlan), GFP_KERNEL);
+	if (!mlxsw_sp_port_vlan) {
+		err = -ENOMEM;
+		goto err_port_vlan_alloc;
+	}
+
+	mlxsw_sp_port_vlan->mlxsw_sp_port = mlxsw_sp_port;
+	mlxsw_sp_port_vlan->vid = vid;
+	list_add(&mlxsw_sp_port_vlan->list, &mlxsw_sp_port->vlans_list);
+
+	return mlxsw_sp_port_vlan;
+
+err_port_vlan_alloc:
+	mlxsw_sp_port_vlan_set(mlxsw_sp_port, vid, vid, false, false);
+	return ERR_PTR(err);
+}
+
+void mlxsw_sp_port_vlan_destroy(struct mlxsw_sp_port_vlan *mlxsw_sp_port_vlan)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = mlxsw_sp_port_vlan->mlxsw_sp_port;
+	u16 vid = mlxsw_sp_port_vlan->vid;
+
+	mlxsw_sp_port_vlan_cleanup(mlxsw_sp_port_vlan);
+	list_del(&mlxsw_sp_port_vlan->list);
+	kfree(mlxsw_sp_port_vlan);
+	mlxsw_sp_port_vlan_set(mlxsw_sp_port, vid, vid, false, false);
+}
+
+static int mlxsw_sp_port_add_vid(struct net_device *dev,
+				 __be16 __always_unused proto, u16 vid)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+
+	/* VLAN 0 is added to HW filter when device goes up, but it is
+	 * reserved in our case, so simply return.
+	 */
+	if (!vid)
+		return 0;
+
+	return PTR_ERR_OR_ZERO(mlxsw_sp_port_vlan_create(mlxsw_sp_port, vid));
+}
+
+static int mlxsw_sp_port_kill_vid(struct net_device *dev,
+				  __be16 __always_unused proto, u16 vid)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+	struct mlxsw_sp_port_vlan *mlxsw_sp_port_vlan;
+
+	/* VLAN 0 is removed from HW filter when device goes down, but
+	 * it is reserved in our case, so simply return.
+	 */
+	if (!vid)
+		return 0;
+
+	mlxsw_sp_port_vlan = mlxsw_sp_port_vlan_find_by_vid(mlxsw_sp_port, vid);
+	if (!mlxsw_sp_port_vlan)
+		return 0;
+	mlxsw_sp_port_vlan_destroy(mlxsw_sp_port_vlan);
+
+	return 0;
+}
+
+static int mlxsw_sp_setup_tc_block(struct mlxsw_sp_port *mlxsw_sp_port,
+				   struct flow_block_offload *f)
+{
+	switch (f->binder_type) {
+	case FLOW_BLOCK_BINDER_TYPE_CLSACT_INGRESS:
+		return mlxsw_sp_setup_tc_block_clsact(mlxsw_sp_port, f, true);
+	case FLOW_BLOCK_BINDER_TYPE_CLSACT_EGRESS:
+		return mlxsw_sp_setup_tc_block_clsact(mlxsw_sp_port, f, false);
+	case FLOW_BLOCK_BINDER_TYPE_RED_EARLY_DROP:
+		return mlxsw_sp_setup_tc_block_qevent_early_drop(mlxsw_sp_port, f);
+	case FLOW_BLOCK_BINDER_TYPE_RED_MARK:
+		return mlxsw_sp_setup_tc_block_qevent_mark(mlxsw_sp_port, f);
+	default:
+		return -EOPNOTSUPP;
+	}
+}
+
+static int mlxsw_sp_setup_tc(struct net_device *dev, enum tc_setup_type type,
+			     void *type_data)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+
+	switch (type) {
+	case TC_SETUP_BLOCK:
+		return mlxsw_sp_setup_tc_block(mlxsw_sp_port, type_data);
+	case TC_SETUP_QDISC_RED:
+		return mlxsw_sp_setup_tc_red(mlxsw_sp_port, type_data);
+	case TC_SETUP_QDISC_PRIO:
+		return mlxsw_sp_setup_tc_prio(mlxsw_sp_port, type_data);
+	case TC_SETUP_QDISC_ETS:
+		return mlxsw_sp_setup_tc_ets(mlxsw_sp_port, type_data);
+	case TC_SETUP_QDISC_TBF:
+		return mlxsw_sp_setup_tc_tbf(mlxsw_sp_port, type_data);
+	case TC_SETUP_QDISC_FIFO:
+		return mlxsw_sp_setup_tc_fifo(mlxsw_sp_port, type_data);
+	default:
+		return -EOPNOTSUPP;
+	}
+}
+
+static int mlxsw_sp_feature_hw_tc(struct net_device *dev, bool enable)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+
+	if (!enable) {
+		if (mlxsw_sp_flow_block_rule_count(mlxsw_sp_port->ing_flow_block) ||
+		    mlxsw_sp_flow_block_rule_count(mlxsw_sp_port->eg_flow_block)) {
+			netdev_err(dev, "Active offloaded tc filters, can't turn hw_tc_offload off\n");
+			return -EINVAL;
+		}
+		mlxsw_sp_flow_block_disable_inc(mlxsw_sp_port->ing_flow_block);
+		mlxsw_sp_flow_block_disable_inc(mlxsw_sp_port->eg_flow_block);
+	} else {
+		mlxsw_sp_flow_block_disable_dec(mlxsw_sp_port->ing_flow_block);
+		mlxsw_sp_flow_block_disable_dec(mlxsw_sp_port->eg_flow_block);
+	}
+	return 0;
+}
+
+static int mlxsw_sp_feature_loopback(struct net_device *dev, bool enable)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+	char pplr_pl[MLXSW_REG_PPLR_LEN];
+	int err;
+
+	if (netif_running(dev))
+		mlxsw_sp_port_admin_status_set(mlxsw_sp_port, false);
+
+	mlxsw_reg_pplr_pack(pplr_pl, mlxsw_sp_port->local_port, enable);
+	err = mlxsw_reg_write(mlxsw_sp_port->mlxsw_sp->core, MLXSW_REG(pplr),
+			      pplr_pl);
+
+	if (netif_running(dev))
+		mlxsw_sp_port_admin_status_set(mlxsw_sp_port, true);
+
+	return err;
+}
+
+typedef int (*mlxsw_sp_feature_handler)(struct net_device *dev, bool enable);
+
+static int mlxsw_sp_handle_feature(struct net_device *dev,
+				   netdev_features_t wanted_features,
+				   netdev_features_t feature,
+				   mlxsw_sp_feature_handler feature_handler)
+{
+	netdev_features_t changes = wanted_features ^ dev->features;
+	bool enable = !!(wanted_features & feature);
+	int err;
+
+	if (!(changes & feature))
+		return 0;
+
+	err = feature_handler(dev, enable);
+	if (err) {
+		netdev_err(dev, "%s feature %pNF failed, err %d\n",
+			   enable ? "Enable" : "Disable", &feature, err);
+		return err;
+	}
+
+	if (enable)
+		dev->features |= feature;
+	else
+		dev->features &= ~feature;
+
+	return 0;
+}
+static int mlxsw_sp_set_features(struct net_device *dev,
+				 netdev_features_t features)
+{
+	netdev_features_t oper_features = dev->features;
+	int err = 0;
+
+	err |= mlxsw_sp_handle_feature(dev, features, NETIF_F_HW_TC,
+				       mlxsw_sp_feature_hw_tc);
+	err |= mlxsw_sp_handle_feature(dev, features, NETIF_F_LOOPBACK,
+				       mlxsw_sp_feature_loopback);
+
+	if (err) {
+		dev->features = oper_features;
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static struct devlink_port *
+mlxsw_sp_port_get_devlink_port(struct net_device *dev)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+
+	return mlxsw_core_port_devlink_port_get(mlxsw_sp->core,
+						mlxsw_sp_port->local_port);
+}
+
+static int mlxsw_sp_port_hwtstamp_set(struct mlxsw_sp_port *mlxsw_sp_port,
+				      struct ifreq *ifr)
+{
+	struct hwtstamp_config config;
+	int err;
+
+	if (copy_from_user(&config, ifr->ifr_data, sizeof(config)))
+		return -EFAULT;
+
+	err = mlxsw_sp_port->mlxsw_sp->ptp_ops->hwtstamp_set(mlxsw_sp_port,
+							     &config);
+	if (err)
+		return err;
+
+	if (copy_to_user(ifr->ifr_data, &config, sizeof(config)))
+		return -EFAULT;
+
+	return 0;
+}
+
+static int mlxsw_sp_port_hwtstamp_get(struct mlxsw_sp_port *mlxsw_sp_port,
+				      struct ifreq *ifr)
+{
+	struct hwtstamp_config config;
+	int err;
+
+	err = mlxsw_sp_port->mlxsw_sp->ptp_ops->hwtstamp_get(mlxsw_sp_port,
+							     &config);
+	if (err)
+		return err;
+
+	if (copy_to_user(ifr->ifr_data, &config, sizeof(config)))
+		return -EFAULT;
+
+	return 0;
+}
+
+static inline void mlxsw_sp_port_ptp_clear(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	struct hwtstamp_config config = {0};
+
+	mlxsw_sp_port->mlxsw_sp->ptp_ops->hwtstamp_set(mlxsw_sp_port, &config);
+}
+
+static int
+mlxsw_sp_port_ioctl(struct net_device *dev, struct ifreq *ifr, int cmd)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+
+	switch (cmd) {
+	case SIOCSHWTSTAMP:
+		return mlxsw_sp_port_hwtstamp_set(mlxsw_sp_port, ifr);
+	case SIOCGHWTSTAMP:
+		return mlxsw_sp_port_hwtstamp_get(mlxsw_sp_port, ifr);
+	default:
+		return -EOPNOTSUPP;
+	}
+}
+
+static const struct net_device_ops mlxsw_sp_port_netdev_ops = {
+	.ndo_open		= mlxsw_sp_port_open,
+	.ndo_stop		= mlxsw_sp_port_stop,
+	.ndo_start_xmit		= mlxsw_sp_port_xmit,
+	.ndo_setup_tc           = mlxsw_sp_setup_tc,
+	.ndo_set_rx_mode	= mlxsw_sp_set_rx_mode,
+	.ndo_set_mac_address	= mlxsw_sp_port_set_mac_address,
+	.ndo_change_mtu		= mlxsw_sp_port_change_mtu,
+	.ndo_get_stats64	= mlxsw_sp_port_get_stats64,
+	.ndo_has_offload_stats	= mlxsw_sp_port_has_offload_stats,
+	.ndo_get_offload_stats	= mlxsw_sp_port_get_offload_stats,
+	.ndo_vlan_rx_add_vid	= mlxsw_sp_port_add_vid,
+	.ndo_vlan_rx_kill_vid	= mlxsw_sp_port_kill_vid,
+	.ndo_set_features	= mlxsw_sp_set_features,
+	.ndo_get_devlink_port	= mlxsw_sp_port_get_devlink_port,
+	.ndo_eth_ioctl		= mlxsw_sp_port_ioctl,
+};
+
+static int
+mlxsw_sp_port_speed_by_width_set(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	u32 eth_proto_cap, eth_proto_admin, eth_proto_oper;
+	const struct mlxsw_sp_port_type_speed_ops *ops;
+	char ptys_pl[MLXSW_REG_PTYS_LEN];
+	u32 eth_proto_cap_masked;
+	int err;
+
+	ops = mlxsw_sp->port_type_speed_ops;
+
+	/* Set advertised speeds to speeds supported by both the driver
+	 * and the device.
+	 */
+	ops->reg_ptys_eth_pack(mlxsw_sp, ptys_pl, mlxsw_sp_port->local_port,
+			       0, false);
+	err = mlxsw_reg_query(mlxsw_sp->core, MLXSW_REG(ptys), ptys_pl);
+	if (err)
+		return err;
+
+	ops->reg_ptys_eth_unpack(mlxsw_sp, ptys_pl, &eth_proto_cap,
+				 &eth_proto_admin, &eth_proto_oper);
+	eth_proto_cap_masked = ops->ptys_proto_cap_masked_get(eth_proto_cap);
+	ops->reg_ptys_eth_pack(mlxsw_sp, ptys_pl, mlxsw_sp_port->local_port,
+			       eth_proto_cap_masked,
+			       mlxsw_sp_port->link.autoneg);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(ptys), ptys_pl);
+}
+
+int mlxsw_sp_port_speed_get(struct mlxsw_sp_port *mlxsw_sp_port, u32 *speed)
+{
+	const struct mlxsw_sp_port_type_speed_ops *port_type_speed_ops;
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	char ptys_pl[MLXSW_REG_PTYS_LEN];
+	u32 eth_proto_oper;
+	int err;
+
+	port_type_speed_ops = mlxsw_sp->port_type_speed_ops;
+	port_type_speed_ops->reg_ptys_eth_pack(mlxsw_sp, ptys_pl,
+					       mlxsw_sp_port->local_port, 0,
+					       false);
+	err = mlxsw_reg_query(mlxsw_sp->core, MLXSW_REG(ptys), ptys_pl);
+	if (err)
+		return err;
+	port_type_speed_ops->reg_ptys_eth_unpack(mlxsw_sp, ptys_pl, NULL, NULL,
+						 &eth_proto_oper);
+	*speed = port_type_speed_ops->from_ptys_speed(mlxsw_sp, eth_proto_oper);
+	return 0;
+}
+
+int mlxsw_sp_port_ets_set(struct mlxsw_sp_port *mlxsw_sp_port,
+			  enum mlxsw_reg_qeec_hr hr, u8 index, u8 next_index,
+			  bool dwrr, u8 dwrr_weight)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	char qeec_pl[MLXSW_REG_QEEC_LEN];
+
+	mlxsw_reg_qeec_pack(qeec_pl, mlxsw_sp_port->local_port, hr, index,
+			    next_index);
+	mlxsw_reg_qeec_de_set(qeec_pl, true);
+	mlxsw_reg_qeec_dwrr_set(qeec_pl, dwrr);
+	mlxsw_reg_qeec_dwrr_weight_set(qeec_pl, dwrr_weight);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(qeec), qeec_pl);
+}
+
+int mlxsw_sp_port_ets_maxrate_set(struct mlxsw_sp_port *mlxsw_sp_port,
+				  enum mlxsw_reg_qeec_hr hr, u8 index,
+				  u8 next_index, u32 maxrate, u8 burst_size)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	char qeec_pl[MLXSW_REG_QEEC_LEN];
+
+	mlxsw_reg_qeec_pack(qeec_pl, mlxsw_sp_port->local_port, hr, index,
+			    next_index);
+	mlxsw_reg_qeec_mase_set(qeec_pl, true);
+	mlxsw_reg_qeec_max_shaper_rate_set(qeec_pl, maxrate);
+	mlxsw_reg_qeec_max_shaper_bs_set(qeec_pl, burst_size);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(qeec), qeec_pl);
+}
+
+static int mlxsw_sp_port_min_bw_set(struct mlxsw_sp_port *mlxsw_sp_port,
+				    enum mlxsw_reg_qeec_hr hr, u8 index,
+				    u8 next_index, u32 minrate)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	char qeec_pl[MLXSW_REG_QEEC_LEN];
+
+	mlxsw_reg_qeec_pack(qeec_pl, mlxsw_sp_port->local_port, hr, index,
+			    next_index);
+	mlxsw_reg_qeec_mise_set(qeec_pl, true);
+	mlxsw_reg_qeec_min_shaper_rate_set(qeec_pl, minrate);
+
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(qeec), qeec_pl);
+}
+
+int mlxsw_sp_port_prio_tc_set(struct mlxsw_sp_port *mlxsw_sp_port,
+			      u8 switch_prio, u8 tclass)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	char qtct_pl[MLXSW_REG_QTCT_LEN];
+
+	mlxsw_reg_qtct_pack(qtct_pl, mlxsw_sp_port->local_port, switch_prio,
+			    tclass);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(qtct), qtct_pl);
+}
+
+static int mlxsw_sp_port_ets_init(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	int err, i;
+
+	/* Setup the elements hierarcy, so that each TC is linked to
+	 * one subgroup, which are all member in the same group.
+	 */
+	err = mlxsw_sp_port_ets_set(mlxsw_sp_port,
+				    MLXSW_REG_QEEC_HR_GROUP, 0, 0, false, 0);
+	if (err)
+		return err;
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) {
+		err = mlxsw_sp_port_ets_set(mlxsw_sp_port,
+					    MLXSW_REG_QEEC_HR_SUBGROUP, i,
+					    0, false, 0);
+		if (err)
+			return err;
+	}
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) {
+		err = mlxsw_sp_port_ets_set(mlxsw_sp_port,
+					    MLXSW_REG_QEEC_HR_TC, i, i,
+					    false, 0);
+		if (err)
+			return err;
+
+		err = mlxsw_sp_port_ets_set(mlxsw_sp_port,
+					    MLXSW_REG_QEEC_HR_TC,
+					    i + 8, i,
+					    true, 100);
+		if (err)
+			return err;
+	}
+
+	/* Make sure the max shaper is disabled in all hierarchies that support
+	 * it. Note that this disables ptps (PTP shaper), but that is intended
+	 * for the initial configuration.
+	 */
+	err = mlxsw_sp_port_ets_maxrate_set(mlxsw_sp_port,
+					    MLXSW_REG_QEEC_HR_PORT, 0, 0,
+					    MLXSW_REG_QEEC_MAS_DIS, 0);
+	if (err)
+		return err;
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) {
+		err = mlxsw_sp_port_ets_maxrate_set(mlxsw_sp_port,
+						    MLXSW_REG_QEEC_HR_SUBGROUP,
+						    i, 0,
+						    MLXSW_REG_QEEC_MAS_DIS, 0);
+		if (err)
+			return err;
+	}
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) {
+		err = mlxsw_sp_port_ets_maxrate_set(mlxsw_sp_port,
+						    MLXSW_REG_QEEC_HR_TC,
+						    i, i,
+						    MLXSW_REG_QEEC_MAS_DIS, 0);
+		if (err)
+			return err;
+
+		err = mlxsw_sp_port_ets_maxrate_set(mlxsw_sp_port,
+						    MLXSW_REG_QEEC_HR_TC,
+						    i + 8, i,
+						    MLXSW_REG_QEEC_MAS_DIS, 0);
+		if (err)
+			return err;
+	}
+
+	/* Configure the min shaper for multicast TCs. */
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) {
+		err = mlxsw_sp_port_min_bw_set(mlxsw_sp_port,
+					       MLXSW_REG_QEEC_HR_TC,
+					       i + 8, i,
+					       MLXSW_REG_QEEC_MIS_MIN);
+		if (err)
+			return err;
+	}
+
+	/* Map all priorities to traffic class 0. */
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) {
+		err = mlxsw_sp_port_prio_tc_set(mlxsw_sp_port, i, 0);
+		if (err)
+			return err;
+	}
+
+	return 0;
+}
+
+static int mlxsw_sp_port_tc_mc_mode_set(struct mlxsw_sp_port *mlxsw_sp_port,
+					bool enable)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	char qtctm_pl[MLXSW_REG_QTCTM_LEN];
+
+	mlxsw_reg_qtctm_pack(qtctm_pl, mlxsw_sp_port->local_port, enable);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(qtctm), qtctm_pl);
+}
+
+static int mlxsw_sp_port_overheat_init_val_set(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	u8 slot_index = mlxsw_sp_port->mapping.slot_index;
+	u8 module = mlxsw_sp_port->mapping.module;
+	u64 overheat_counter;
+	int err;
+
+	err = mlxsw_env_module_overheat_counter_get(mlxsw_sp->core, slot_index,
+						    module, &overheat_counter);
+	if (err)
+		return err;
+
+	mlxsw_sp_port->module_overheat_initial_val = overheat_counter;
+	return 0;
+}
+
+int
+mlxsw_sp_port_vlan_classification_set(struct mlxsw_sp_port *mlxsw_sp_port,
+				      bool is_8021ad_tagged,
+				      bool is_8021q_tagged)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	char spvc_pl[MLXSW_REG_SPVC_LEN];
+
+	mlxsw_reg_spvc_pack(spvc_pl, mlxsw_sp_port->local_port,
+			    is_8021ad_tagged, is_8021q_tagged);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(spvc), spvc_pl);
+}
+
+static int mlxsw_sp_port_label_info_get(struct mlxsw_sp *mlxsw_sp,
+					u16 local_port, u8 *port_number,
+					u8 *split_port_subnumber,
+					u8 *slot_index)
+{
+	char pllp_pl[MLXSW_REG_PLLP_LEN];
+	int err;
+
+	mlxsw_reg_pllp_pack(pllp_pl, local_port);
+	err = mlxsw_reg_query(mlxsw_sp->core, MLXSW_REG(pllp), pllp_pl);
+	if (err)
+		return err;
+	mlxsw_reg_pllp_unpack(pllp_pl, port_number,
+			      split_port_subnumber, slot_index);
+	return 0;
+}
+
+static int mlxsw_sp_port_create(struct mlxsw_sp *mlxsw_sp, u16 local_port,
+				bool split,
+				struct mlxsw_sp_port_mapping *port_mapping)
+{
+	struct mlxsw_sp_port_vlan *mlxsw_sp_port_vlan;
+	struct mlxsw_sp_port *mlxsw_sp_port;
+	u32 lanes = port_mapping->width;
+	u8 split_port_subnumber;
+	struct net_device *dev;
+	u8 port_number;
+	u8 slot_index;
+	bool splittable;
+	int err;
+
+	err = mlxsw_sp_port_module_map(mlxsw_sp, local_port, port_mapping);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Port %d: Failed to map module\n",
+			local_port);
+		return err;
+	}
+
+	err = mlxsw_sp_port_swid_set(mlxsw_sp, local_port, 0);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Port %d: Failed to set SWID\n",
+			local_port);
+		goto err_port_swid_set;
+	}
+
+	err = mlxsw_sp_port_label_info_get(mlxsw_sp, local_port, &port_number,
+					   &split_port_subnumber, &slot_index);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Port %d: Failed to get port label information\n",
+			local_port);
+		goto err_port_label_info_get;
+	}
+
+	splittable = lanes > 1 && !split;
+	err = mlxsw_core_port_init(mlxsw_sp->core, local_port, slot_index,
+				   port_number, split, split_port_subnumber,
+				   splittable, lanes, mlxsw_sp->base_mac,
+				   sizeof(mlxsw_sp->base_mac));
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Port %d: Failed to init core port\n",
+			local_port);
+		goto err_core_port_init;
+	}
+
+	dev = alloc_etherdev(sizeof(struct mlxsw_sp_port));
+	if (!dev) {
+		err = -ENOMEM;
+		goto err_alloc_etherdev;
+	}
+	SET_NETDEV_DEV(dev, mlxsw_sp->bus_info->dev);
+	dev_net_set(dev, mlxsw_sp_net(mlxsw_sp));
+	mlxsw_sp_port = netdev_priv(dev);
+	mlxsw_sp_port->dev = dev;
+	mlxsw_sp_port->mlxsw_sp = mlxsw_sp;
+	mlxsw_sp_port->local_port = local_port;
+	mlxsw_sp_port->pvid = MLXSW_SP_DEFAULT_VID;
+	mlxsw_sp_port->split = split;
+	mlxsw_sp_port->mapping = *port_mapping;
+	mlxsw_sp_port->link.autoneg = 1;
+	INIT_LIST_HEAD(&mlxsw_sp_port->vlans_list);
+
+	mlxsw_sp_port->pcpu_stats =
+		netdev_alloc_pcpu_stats(struct mlxsw_sp_port_pcpu_stats);
+	if (!mlxsw_sp_port->pcpu_stats) {
+		err = -ENOMEM;
+		goto err_alloc_stats;
+	}
+
+	INIT_DELAYED_WORK(&mlxsw_sp_port->periodic_hw_stats.update_dw,
+			  &update_stats_cache);
+
+	dev->netdev_ops = &mlxsw_sp_port_netdev_ops;
+	dev->ethtool_ops = &mlxsw_sp_port_ethtool_ops;
+
+	err = mlxsw_sp_port_dev_addr_init(mlxsw_sp_port);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Port %d: Unable to init port mac address\n",
+			mlxsw_sp_port->local_port);
+		goto err_dev_addr_init;
+	}
+
+	netif_carrier_off(dev);
+
+	dev->features |= NETIF_F_NETNS_LOCAL | NETIF_F_LLTX | NETIF_F_SG |
+			 NETIF_F_HW_VLAN_CTAG_FILTER | NETIF_F_HW_TC;
+	dev->hw_features |= NETIF_F_HW_TC | NETIF_F_LOOPBACK;
+
+	dev->min_mtu = 0;
+	dev->max_mtu = ETH_MAX_MTU;
+
+	/* Each packet needs to have a Tx header (metadata) on top all other
+	 * headers.
+	 */
+	dev->needed_headroom = MLXSW_TXHDR_LEN;
+
+	err = mlxsw_sp_port_system_port_mapping_set(mlxsw_sp_port);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Port %d: Failed to set system port mapping\n",
+			mlxsw_sp_port->local_port);
+		goto err_port_system_port_mapping_set;
+	}
+
+	err = mlxsw_sp_port_speed_by_width_set(mlxsw_sp_port);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Port %d: Failed to enable speeds\n",
+			mlxsw_sp_port->local_port);
+		goto err_port_speed_by_width_set;
+	}
+
+	err = mlxsw_sp->port_type_speed_ops->ptys_max_speed(mlxsw_sp_port,
+							    &mlxsw_sp_port->max_speed);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Port %d: Failed to get maximum speed\n",
+			mlxsw_sp_port->local_port);
+		goto err_max_speed_get;
+	}
+
+	err = mlxsw_sp_port_max_mtu_get(mlxsw_sp_port, &mlxsw_sp_port->max_mtu);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Port %d: Failed to get maximum MTU\n",
+			mlxsw_sp_port->local_port);
+		goto err_port_max_mtu_get;
+	}
+
+	err = mlxsw_sp_port_mtu_set(mlxsw_sp_port, ETH_DATA_LEN);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Port %d: Failed to set MTU\n",
+			mlxsw_sp_port->local_port);
+		goto err_port_mtu_set;
+	}
+
+	err = mlxsw_sp_port_admin_status_set(mlxsw_sp_port, false);
+	if (err)
+		goto err_port_admin_status_set;
+
+	err = mlxsw_sp_port_buffers_init(mlxsw_sp_port);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Port %d: Failed to initialize buffers\n",
+			mlxsw_sp_port->local_port);
+		goto err_port_buffers_init;
+	}
+
+	err = mlxsw_sp_port_ets_init(mlxsw_sp_port);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Port %d: Failed to initialize ETS\n",
+			mlxsw_sp_port->local_port);
+		goto err_port_ets_init;
+	}
+
+	err = mlxsw_sp_port_tc_mc_mode_set(mlxsw_sp_port, true);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Port %d: Failed to initialize TC MC mode\n",
+			mlxsw_sp_port->local_port);
+		goto err_port_tc_mc_mode;
+	}
+
+	/* ETS and buffers must be initialized before DCB. */
+	err = mlxsw_sp_port_dcb_init(mlxsw_sp_port);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Port %d: Failed to initialize DCB\n",
+			mlxsw_sp_port->local_port);
+		goto err_port_dcb_init;
+	}
+
+	err = mlxsw_sp_port_fids_init(mlxsw_sp_port);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Port %d: Failed to initialize FIDs\n",
+			mlxsw_sp_port->local_port);
+		goto err_port_fids_init;
+	}
+
+	err = mlxsw_sp_tc_qdisc_init(mlxsw_sp_port);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Port %d: Failed to initialize TC qdiscs\n",
+			mlxsw_sp_port->local_port);
+		goto err_port_qdiscs_init;
+	}
+
+	err = mlxsw_sp_port_vlan_set(mlxsw_sp_port, 0, VLAN_N_VID - 1, false,
+				     false);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Port %d: Failed to clear VLAN filter\n",
+			mlxsw_sp_port->local_port);
+		goto err_port_vlan_clear;
+	}
+
+	err = mlxsw_sp_port_nve_init(mlxsw_sp_port);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Port %d: Failed to initialize NVE\n",
+			mlxsw_sp_port->local_port);
+		goto err_port_nve_init;
+	}
+
+	err = mlxsw_sp_port_pvid_set(mlxsw_sp_port, MLXSW_SP_DEFAULT_VID,
+				     ETH_P_8021Q);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Port %d: Failed to set PVID\n",
+			mlxsw_sp_port->local_port);
+		goto err_port_pvid_set;
+	}
+
+	mlxsw_sp_port_vlan = mlxsw_sp_port_vlan_create(mlxsw_sp_port,
+						       MLXSW_SP_DEFAULT_VID);
+	if (IS_ERR(mlxsw_sp_port_vlan)) {
+		dev_err(mlxsw_sp->bus_info->dev, "Port %d: Failed to create VID 1\n",
+			mlxsw_sp_port->local_port);
+		err = PTR_ERR(mlxsw_sp_port_vlan);
+		goto err_port_vlan_create;
+	}
+	mlxsw_sp_port->default_vlan = mlxsw_sp_port_vlan;
+
+	/* Set SPVC.et0=true and SPVC.et1=false to make the local port to treat
+	 * only packets with 802.1q header as tagged packets.
+	 */
+	err = mlxsw_sp_port_vlan_classification_set(mlxsw_sp_port, false, true);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Port %d: Failed to set default VLAN classification\n",
+			local_port);
+		goto err_port_vlan_classification_set;
+	}
+
+	INIT_DELAYED_WORK(&mlxsw_sp_port->ptp.shaper_dw,
+			  mlxsw_sp->ptp_ops->shaper_work);
+
+	mlxsw_sp->ports[local_port] = mlxsw_sp_port;
+
+	err = mlxsw_sp_port_overheat_init_val_set(mlxsw_sp_port);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Port %d: Failed to set overheat initial value\n",
+			mlxsw_sp_port->local_port);
+		goto err_port_overheat_init_val_set;
+	}
+
+	err = register_netdev(dev);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Port %d: Failed to register netdev\n",
+			mlxsw_sp_port->local_port);
+		goto err_register_netdev;
+	}
+
+	mlxsw_core_port_eth_set(mlxsw_sp->core, mlxsw_sp_port->local_port,
+				mlxsw_sp_port, dev);
+	mlxsw_core_schedule_dw(&mlxsw_sp_port->periodic_hw_stats.update_dw, 0);
+	return 0;
+
+err_register_netdev:
+err_port_overheat_init_val_set:
+	mlxsw_sp_port_vlan_classification_set(mlxsw_sp_port, true, true);
+err_port_vlan_classification_set:
+	mlxsw_sp->ports[local_port] = NULL;
+	mlxsw_sp_port_vlan_destroy(mlxsw_sp_port_vlan);
+err_port_vlan_create:
+err_port_pvid_set:
+	mlxsw_sp_port_nve_fini(mlxsw_sp_port);
+err_port_nve_init:
+err_port_vlan_clear:
+	mlxsw_sp_tc_qdisc_fini(mlxsw_sp_port);
+err_port_qdiscs_init:
+	mlxsw_sp_port_fids_fini(mlxsw_sp_port);
+err_port_fids_init:
+	mlxsw_sp_port_dcb_fini(mlxsw_sp_port);
+err_port_dcb_init:
+	mlxsw_sp_port_tc_mc_mode_set(mlxsw_sp_port, false);
+err_port_tc_mc_mode:
+err_port_ets_init:
+	mlxsw_sp_port_buffers_fini(mlxsw_sp_port);
+err_port_buffers_init:
+err_port_admin_status_set:
+err_port_mtu_set:
+err_port_max_mtu_get:
+err_max_speed_get:
+err_port_speed_by_width_set:
+err_port_system_port_mapping_set:
+err_dev_addr_init:
+	free_percpu(mlxsw_sp_port->pcpu_stats);
+err_alloc_stats:
+	free_netdev(dev);
+err_alloc_etherdev:
+	mlxsw_core_port_fini(mlxsw_sp->core, local_port);
+err_core_port_init:
+err_port_label_info_get:
+	mlxsw_sp_port_swid_set(mlxsw_sp, local_port,
+			       MLXSW_PORT_SWID_DISABLED_PORT);
+err_port_swid_set:
+	mlxsw_sp_port_module_unmap(mlxsw_sp, local_port,
+				   port_mapping->slot_index,
+				   port_mapping->module);
+	return err;
+}
+
+static void mlxsw_sp_port_remove(struct mlxsw_sp *mlxsw_sp, u16 local_port)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = mlxsw_sp->ports[local_port];
+	u8 slot_index = mlxsw_sp_port->mapping.slot_index;
+	u8 module = mlxsw_sp_port->mapping.module;
+
+	cancel_delayed_work_sync(&mlxsw_sp_port->periodic_hw_stats.update_dw);
+	cancel_delayed_work_sync(&mlxsw_sp_port->ptp.shaper_dw);
+	mlxsw_core_port_clear(mlxsw_sp->core, local_port, mlxsw_sp);
+	unregister_netdev(mlxsw_sp_port->dev); /* This calls ndo_stop */
+	mlxsw_sp_port_ptp_clear(mlxsw_sp_port);
+	mlxsw_sp_port_vlan_classification_set(mlxsw_sp_port, true, true);
+	mlxsw_sp->ports[local_port] = NULL;
+	mlxsw_sp_port_vlan_flush(mlxsw_sp_port, true);
+	mlxsw_sp_port_nve_fini(mlxsw_sp_port);
+	mlxsw_sp_tc_qdisc_fini(mlxsw_sp_port);
+	mlxsw_sp_port_fids_fini(mlxsw_sp_port);
+	mlxsw_sp_port_dcb_fini(mlxsw_sp_port);
+	mlxsw_sp_port_tc_mc_mode_set(mlxsw_sp_port, false);
+	mlxsw_sp_port_buffers_fini(mlxsw_sp_port);
+	free_percpu(mlxsw_sp_port->pcpu_stats);
+	WARN_ON_ONCE(!list_empty(&mlxsw_sp_port->vlans_list));
+	free_netdev(mlxsw_sp_port->dev);
+	mlxsw_core_port_fini(mlxsw_sp->core, local_port);
+	mlxsw_sp_port_swid_set(mlxsw_sp, local_port,
+			       MLXSW_PORT_SWID_DISABLED_PORT);
+	mlxsw_sp_port_module_unmap(mlxsw_sp, local_port, slot_index, module);
+}
+
+static int mlxsw_sp_cpu_port_create(struct mlxsw_sp *mlxsw_sp)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port;
+	int err;
+
+	mlxsw_sp_port = kzalloc(sizeof(*mlxsw_sp_port), GFP_KERNEL);
+	if (!mlxsw_sp_port)
+		return -ENOMEM;
+
+	mlxsw_sp_port->mlxsw_sp = mlxsw_sp;
+	mlxsw_sp_port->local_port = MLXSW_PORT_CPU_PORT;
+
+	err = mlxsw_core_cpu_port_init(mlxsw_sp->core,
+				       mlxsw_sp_port,
+				       mlxsw_sp->base_mac,
+				       sizeof(mlxsw_sp->base_mac));
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Failed to initialize core CPU port\n");
+		goto err_core_cpu_port_init;
+	}
+
+	mlxsw_sp->ports[MLXSW_PORT_CPU_PORT] = mlxsw_sp_port;
+	return 0;
+
+err_core_cpu_port_init:
+	kfree(mlxsw_sp_port);
+	return err;
+}
+
+static void mlxsw_sp_cpu_port_remove(struct mlxsw_sp *mlxsw_sp)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port =
+				mlxsw_sp->ports[MLXSW_PORT_CPU_PORT];
+
+	mlxsw_core_cpu_port_fini(mlxsw_sp->core);
+	mlxsw_sp->ports[MLXSW_PORT_CPU_PORT] = NULL;
+	kfree(mlxsw_sp_port);
+}
+
+static bool mlxsw_sp_local_port_valid(u16 local_port)
+{
+	return local_port != MLXSW_PORT_CPU_PORT;
+}
+
+static bool mlxsw_sp_port_created(struct mlxsw_sp *mlxsw_sp, u16 local_port)
+{
+	if (!mlxsw_sp_local_port_valid(local_port))
+		return false;
+	return mlxsw_sp->ports[local_port] != NULL;
+}
+
+static int mlxsw_sp_port_mapping_event_set(struct mlxsw_sp *mlxsw_sp,
+					   u16 local_port, bool enable)
+{
+	char pmecr_pl[MLXSW_REG_PMECR_LEN];
+
+	mlxsw_reg_pmecr_pack(pmecr_pl, local_port,
+			     enable ? MLXSW_REG_PMECR_E_GENERATE_EVENT :
+				      MLXSW_REG_PMECR_E_DO_NOT_GENERATE_EVENT);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(pmecr), pmecr_pl);
+}
+
+struct mlxsw_sp_port_mapping_event {
+	struct list_head list;
+	char pmlp_pl[MLXSW_REG_PMLP_LEN];
+};
+
+static void mlxsw_sp_port_mapping_events_work(struct work_struct *work)
+{
+	struct mlxsw_sp_port_mapping_event *event, *next_event;
+	struct mlxsw_sp_port_mapping_events *events;
+	struct mlxsw_sp_port_mapping port_mapping;
+	struct mlxsw_sp *mlxsw_sp;
+	struct devlink *devlink;
+	LIST_HEAD(event_queue);
+	u16 local_port;
+	int err;
+
+	events = container_of(work, struct mlxsw_sp_port_mapping_events, work);
+	mlxsw_sp = container_of(events, struct mlxsw_sp, port_mapping_events);
+	devlink = priv_to_devlink(mlxsw_sp->core);
+
+	spin_lock_bh(&events->queue_lock);
+	list_splice_init(&events->queue, &event_queue);
+	spin_unlock_bh(&events->queue_lock);
+
+	list_for_each_entry_safe(event, next_event, &event_queue, list) {
+		local_port = mlxsw_reg_pmlp_local_port_get(event->pmlp_pl);
+		err = mlxsw_sp_port_module_info_parse(mlxsw_sp, local_port,
+						      event->pmlp_pl, &port_mapping);
+		if (err)
+			goto out;
+
+		if (WARN_ON_ONCE(!port_mapping.width))
+			goto out;
+
+		devl_lock(devlink);
+
+		if (!mlxsw_sp_port_created(mlxsw_sp, local_port))
+			mlxsw_sp_port_create(mlxsw_sp, local_port,
+					     false, &port_mapping);
+		else
+			WARN_ON_ONCE(1);
+
+		devl_unlock(devlink);
+
+		mlxsw_sp->port_mapping[local_port] = port_mapping;
+
+out:
+		kfree(event);
+	}
+}
+
+static void
+mlxsw_sp_port_mapping_listener_func(const struct mlxsw_reg_info *reg,
+				    char *pmlp_pl, void *priv)
+{
+	struct mlxsw_sp_port_mapping_events *events;
+	struct mlxsw_sp_port_mapping_event *event;
+	struct mlxsw_sp *mlxsw_sp = priv;
+	u16 local_port;
+
+	local_port = mlxsw_reg_pmlp_local_port_get(pmlp_pl);
+	if (WARN_ON_ONCE(!mlxsw_sp_local_port_is_valid(mlxsw_sp, local_port)))
+		return;
+
+	events = &mlxsw_sp->port_mapping_events;
+	event = kmalloc(sizeof(*event), GFP_ATOMIC);
+	if (!event)
+		return;
+	memcpy(event->pmlp_pl, pmlp_pl, sizeof(event->pmlp_pl));
+	spin_lock(&events->queue_lock);
+	list_add_tail(&event->list, &events->queue);
+	spin_unlock(&events->queue_lock);
+	mlxsw_core_schedule_work(&events->work);
+}
+
+static void
+__mlxsw_sp_port_mapping_events_cancel(struct mlxsw_sp *mlxsw_sp)
+{
+	struct mlxsw_sp_port_mapping_event *event, *next_event;
+	struct mlxsw_sp_port_mapping_events *events;
+
+	events = &mlxsw_sp->port_mapping_events;
+
+	/* Caller needs to make sure that no new event is going to appear. */
+	cancel_work_sync(&events->work);
+	list_for_each_entry_safe(event, next_event, &events->queue, list) {
+		list_del(&event->list);
+		kfree(event);
+	}
+}
+
+static void mlxsw_sp_ports_remove(struct mlxsw_sp *mlxsw_sp)
+{
+	unsigned int max_ports = mlxsw_core_max_ports(mlxsw_sp->core);
+	int i;
+
+	for (i = 1; i < max_ports; i++)
+		mlxsw_sp_port_mapping_event_set(mlxsw_sp, i, false);
+	/* Make sure all scheduled events are processed */
+	__mlxsw_sp_port_mapping_events_cancel(mlxsw_sp);
+
+	for (i = 1; i < max_ports; i++)
+		if (mlxsw_sp_port_created(mlxsw_sp, i))
+			mlxsw_sp_port_remove(mlxsw_sp, i);
+	mlxsw_sp_cpu_port_remove(mlxsw_sp);
+	kfree(mlxsw_sp->ports);
+	mlxsw_sp->ports = NULL;
+}
+
+static void
+mlxsw_sp_ports_remove_selected(struct mlxsw_core *mlxsw_core,
+			       bool (*selector)(void *priv, u16 local_port),
+			       void *priv)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_core_driver_priv(mlxsw_core);
+	unsigned int max_ports = mlxsw_core_max_ports(mlxsw_core);
+	int i;
+
+	for (i = 1; i < max_ports; i++)
+		if (mlxsw_sp_port_created(mlxsw_sp, i) && selector(priv, i))
+			mlxsw_sp_port_remove(mlxsw_sp, i);
+}
+
+static int mlxsw_sp_ports_create(struct mlxsw_sp *mlxsw_sp)
+{
+	unsigned int max_ports = mlxsw_core_max_ports(mlxsw_sp->core);
+	struct mlxsw_sp_port_mapping_events *events;
+	struct mlxsw_sp_port_mapping *port_mapping;
+	size_t alloc_size;
+	int i;
+	int err;
+
+	alloc_size = sizeof(struct mlxsw_sp_port *) * max_ports;
+	mlxsw_sp->ports = kzalloc(alloc_size, GFP_KERNEL);
+	if (!mlxsw_sp->ports)
+		return -ENOMEM;
+
+	events = &mlxsw_sp->port_mapping_events;
+	INIT_LIST_HEAD(&events->queue);
+	spin_lock_init(&events->queue_lock);
+	INIT_WORK(&events->work, mlxsw_sp_port_mapping_events_work);
+
+	for (i = 1; i < max_ports; i++) {
+		err = mlxsw_sp_port_mapping_event_set(mlxsw_sp, i, true);
+		if (err)
+			goto err_event_enable;
+	}
+
+	err = mlxsw_sp_cpu_port_create(mlxsw_sp);
+	if (err)
+		goto err_cpu_port_create;
+
+	for (i = 1; i < max_ports; i++) {
+		port_mapping = &mlxsw_sp->port_mapping[i];
+		if (!port_mapping->width)
+			continue;
+		err = mlxsw_sp_port_create(mlxsw_sp, i, false, port_mapping);
+		if (err)
+			goto err_port_create;
+	}
+	return 0;
+
+err_port_create:
+	for (i--; i >= 1; i--)
+		if (mlxsw_sp_port_created(mlxsw_sp, i))
+			mlxsw_sp_port_remove(mlxsw_sp, i);
+	i = max_ports;
+	mlxsw_sp_cpu_port_remove(mlxsw_sp);
+err_cpu_port_create:
+err_event_enable:
+	for (i--; i >= 1; i--)
+		mlxsw_sp_port_mapping_event_set(mlxsw_sp, i, false);
+	/* Make sure all scheduled events are processed */
+	__mlxsw_sp_port_mapping_events_cancel(mlxsw_sp);
+	kfree(mlxsw_sp->ports);
+	mlxsw_sp->ports = NULL;
+	return err;
+}
+
+static int mlxsw_sp_port_module_info_init(struct mlxsw_sp *mlxsw_sp)
+{
+	unsigned int max_ports = mlxsw_core_max_ports(mlxsw_sp->core);
+	struct mlxsw_sp_port_mapping *port_mapping;
+	int i;
+	int err;
+
+	mlxsw_sp->port_mapping = kcalloc(max_ports,
+					 sizeof(struct mlxsw_sp_port_mapping),
+					 GFP_KERNEL);
+	if (!mlxsw_sp->port_mapping)
+		return -ENOMEM;
+
+	for (i = 1; i < max_ports; i++) {
+		port_mapping = &mlxsw_sp->port_mapping[i];
+		err = mlxsw_sp_port_module_info_get(mlxsw_sp, i, port_mapping);
+		if (err)
+			goto err_port_module_info_get;
+	}
+	return 0;
+
+err_port_module_info_get:
+	kfree(mlxsw_sp->port_mapping);
+	return err;
+}
+
+static void mlxsw_sp_port_module_info_fini(struct mlxsw_sp *mlxsw_sp)
+{
+	kfree(mlxsw_sp->port_mapping);
+}
+
+static int
+mlxsw_sp_port_split_create(struct mlxsw_sp *mlxsw_sp,
+			   struct mlxsw_sp_port_mapping *port_mapping,
+			   unsigned int count, const char *pmtdb_pl)
+{
+	struct mlxsw_sp_port_mapping split_port_mapping;
+	int err, i;
+
+	split_port_mapping = *port_mapping;
+	split_port_mapping.width /= count;
+	for (i = 0; i < count; i++) {
+		u16 s_local_port = mlxsw_reg_pmtdb_port_num_get(pmtdb_pl, i);
+
+		if (!mlxsw_sp_local_port_valid(s_local_port))
+			continue;
+
+		err = mlxsw_sp_port_create(mlxsw_sp, s_local_port,
+					   true, &split_port_mapping);
+		if (err)
+			goto err_port_create;
+		split_port_mapping.lane += split_port_mapping.width;
+	}
+
+	return 0;
+
+err_port_create:
+	for (i--; i >= 0; i--) {
+		u16 s_local_port = mlxsw_reg_pmtdb_port_num_get(pmtdb_pl, i);
+
+		if (mlxsw_sp_port_created(mlxsw_sp, s_local_port))
+			mlxsw_sp_port_remove(mlxsw_sp, s_local_port);
+	}
+	return err;
+}
+
+static void mlxsw_sp_port_unsplit_create(struct mlxsw_sp *mlxsw_sp,
+					 unsigned int count,
+					 const char *pmtdb_pl)
+{
+	struct mlxsw_sp_port_mapping *port_mapping;
+	int i;
+
+	/* Go over original unsplit ports in the gap and recreate them. */
+	for (i = 0; i < count; i++) {
+		u16 local_port = mlxsw_reg_pmtdb_port_num_get(pmtdb_pl, i);
+
+		port_mapping = &mlxsw_sp->port_mapping[local_port];
+		if (!port_mapping->width || !mlxsw_sp_local_port_valid(local_port))
+			continue;
+		mlxsw_sp_port_create(mlxsw_sp, local_port,
+				     false, port_mapping);
+	}
+}
+
+static struct mlxsw_sp_port *
+mlxsw_sp_port_get_by_local_port(struct mlxsw_sp *mlxsw_sp, u16 local_port)
+{
+	if (mlxsw_sp->ports && mlxsw_sp->ports[local_port])
+		return mlxsw_sp->ports[local_port];
+	return NULL;
+}
+
+static int mlxsw_sp_port_split(struct mlxsw_core *mlxsw_core, u16 local_port,
+			       unsigned int count,
+			       struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_core_driver_priv(mlxsw_core);
+	struct mlxsw_sp_port_mapping port_mapping;
+	struct mlxsw_sp_port *mlxsw_sp_port;
+	enum mlxsw_reg_pmtdb_status status;
+	char pmtdb_pl[MLXSW_REG_PMTDB_LEN];
+	int i;
+	int err;
+
+	mlxsw_sp_port = mlxsw_sp_port_get_by_local_port(mlxsw_sp, local_port);
+	if (!mlxsw_sp_port) {
+		dev_err(mlxsw_sp->bus_info->dev, "Port number \"%d\" does not exist\n",
+			local_port);
+		NL_SET_ERR_MSG_MOD(extack, "Port number does not exist");
+		return -EINVAL;
+	}
+
+	if (mlxsw_sp_port->split) {
+		NL_SET_ERR_MSG_MOD(extack, "Port is already split");
+		return -EINVAL;
+	}
+
+	mlxsw_reg_pmtdb_pack(pmtdb_pl, mlxsw_sp_port->mapping.slot_index,
+			     mlxsw_sp_port->mapping.module,
+			     mlxsw_sp_port->mapping.module_width / count,
+			     count);
+	err = mlxsw_reg_query(mlxsw_core, MLXSW_REG(pmtdb), pmtdb_pl);
+	if (err) {
+		NL_SET_ERR_MSG_MOD(extack, "Failed to query split info");
+		return err;
+	}
+
+	status = mlxsw_reg_pmtdb_status_get(pmtdb_pl);
+	if (status != MLXSW_REG_PMTDB_STATUS_SUCCESS) {
+		NL_SET_ERR_MSG_MOD(extack, "Unsupported split configuration");
+		return -EINVAL;
+	}
+
+	port_mapping = mlxsw_sp_port->mapping;
+
+	for (i = 0; i < count; i++) {
+		u16 s_local_port = mlxsw_reg_pmtdb_port_num_get(pmtdb_pl, i);
+
+		if (mlxsw_sp_port_created(mlxsw_sp, s_local_port))
+			mlxsw_sp_port_remove(mlxsw_sp, s_local_port);
+	}
+
+	err = mlxsw_sp_port_split_create(mlxsw_sp, &port_mapping,
+					 count, pmtdb_pl);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Failed to create split ports\n");
+		goto err_port_split_create;
+	}
+
+	return 0;
+
+err_port_split_create:
+	mlxsw_sp_port_unsplit_create(mlxsw_sp, count, pmtdb_pl);
+
+	return err;
+}
+
+static int mlxsw_sp_port_unsplit(struct mlxsw_core *mlxsw_core, u16 local_port,
+				 struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_core_driver_priv(mlxsw_core);
+	struct mlxsw_sp_port *mlxsw_sp_port;
+	char pmtdb_pl[MLXSW_REG_PMTDB_LEN];
+	unsigned int count;
+	int i;
+	int err;
+
+	mlxsw_sp_port = mlxsw_sp_port_get_by_local_port(mlxsw_sp, local_port);
+	if (!mlxsw_sp_port) {
+		dev_err(mlxsw_sp->bus_info->dev, "Port number \"%d\" does not exist\n",
+			local_port);
+		NL_SET_ERR_MSG_MOD(extack, "Port number does not exist");
+		return -EINVAL;
+	}
+
+	if (!mlxsw_sp_port->split) {
+		NL_SET_ERR_MSG_MOD(extack, "Port was not split");
+		return -EINVAL;
+	}
+
+	count = mlxsw_sp_port->mapping.module_width /
+		mlxsw_sp_port->mapping.width;
+
+	mlxsw_reg_pmtdb_pack(pmtdb_pl, mlxsw_sp_port->mapping.slot_index,
+			     mlxsw_sp_port->mapping.module,
+			     mlxsw_sp_port->mapping.module_width / count,
+			     count);
+	err = mlxsw_reg_query(mlxsw_core, MLXSW_REG(pmtdb), pmtdb_pl);
+	if (err) {
+		NL_SET_ERR_MSG_MOD(extack, "Failed to query split info");
+		return err;
+	}
+
+	for (i = 0; i < count; i++) {
+		u16 s_local_port = mlxsw_reg_pmtdb_port_num_get(pmtdb_pl, i);
+
+		if (mlxsw_sp_port_created(mlxsw_sp, s_local_port))
+			mlxsw_sp_port_remove(mlxsw_sp, s_local_port);
+	}
+
+	mlxsw_sp_port_unsplit_create(mlxsw_sp, count, pmtdb_pl);
+
+	return 0;
+}
+
+static void
+mlxsw_sp_port_down_wipe_counters(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	int i;
+
+	for (i = 0; i < TC_MAX_QUEUE; i++)
+		mlxsw_sp_port->periodic_hw_stats.xstats.backlog[i] = 0;
+}
+
+static void mlxsw_sp_pude_event_func(const struct mlxsw_reg_info *reg,
+				     char *pude_pl, void *priv)
+{
+	struct mlxsw_sp *mlxsw_sp = priv;
+	struct mlxsw_sp_port *mlxsw_sp_port;
+	enum mlxsw_reg_pude_oper_status status;
+	u16 local_port;
+
+	local_port = mlxsw_reg_pude_local_port_get(pude_pl);
+
+	if (WARN_ON_ONCE(!mlxsw_sp_local_port_is_valid(mlxsw_sp, local_port)))
+		return;
+	mlxsw_sp_port = mlxsw_sp->ports[local_port];
+	if (!mlxsw_sp_port)
+		return;
+
+	status = mlxsw_reg_pude_oper_status_get(pude_pl);
+	if (status == MLXSW_PORT_OPER_STATUS_UP) {
+		netdev_info(mlxsw_sp_port->dev, "link up\n");
+		netif_carrier_on(mlxsw_sp_port->dev);
+		mlxsw_core_schedule_dw(&mlxsw_sp_port->ptp.shaper_dw, 0);
+	} else {
+		netdev_info(mlxsw_sp_port->dev, "link down\n");
+		netif_carrier_off(mlxsw_sp_port->dev);
+		mlxsw_sp_port_down_wipe_counters(mlxsw_sp_port);
+	}
+}
+
+static void mlxsw_sp1_ptp_fifo_event_func(struct mlxsw_sp *mlxsw_sp,
+					  char *mtpptr_pl, bool ingress)
+{
+	u16 local_port;
+	u8 num_rec;
+	int i;
+
+	local_port = mlxsw_reg_mtpptr_local_port_get(mtpptr_pl);
+	num_rec = mlxsw_reg_mtpptr_num_rec_get(mtpptr_pl);
+	for (i = 0; i < num_rec; i++) {
+		u8 domain_number;
+		u8 message_type;
+		u16 sequence_id;
+		u64 timestamp;
+
+		mlxsw_reg_mtpptr_unpack(mtpptr_pl, i, &message_type,
+					&domain_number, &sequence_id,
+					&timestamp);
+		mlxsw_sp1_ptp_got_timestamp(mlxsw_sp, ingress, local_port,
+					    message_type, domain_number,
+					    sequence_id, timestamp);
+	}
+}
+
+static void mlxsw_sp1_ptp_ing_fifo_event_func(const struct mlxsw_reg_info *reg,
+					      char *mtpptr_pl, void *priv)
+{
+	struct mlxsw_sp *mlxsw_sp = priv;
+
+	mlxsw_sp1_ptp_fifo_event_func(mlxsw_sp, mtpptr_pl, true);
+}
+
+static void mlxsw_sp1_ptp_egr_fifo_event_func(const struct mlxsw_reg_info *reg,
+					      char *mtpptr_pl, void *priv)
+{
+	struct mlxsw_sp *mlxsw_sp = priv;
+
+	mlxsw_sp1_ptp_fifo_event_func(mlxsw_sp, mtpptr_pl, false);
+}
+
+void mlxsw_sp_rx_listener_no_mark_func(struct sk_buff *skb,
+				       u16 local_port, void *priv)
+{
+	struct mlxsw_sp *mlxsw_sp = priv;
+	struct mlxsw_sp_port *mlxsw_sp_port = mlxsw_sp->ports[local_port];
+	struct mlxsw_sp_port_pcpu_stats *pcpu_stats;
+
+	if (unlikely(!mlxsw_sp_port)) {
+		dev_warn_ratelimited(mlxsw_sp->bus_info->dev, "Port %d: skb received for non-existent port\n",
+				     local_port);
+		return;
+	}
+
+	skb->dev = mlxsw_sp_port->dev;
+
+	pcpu_stats = this_cpu_ptr(mlxsw_sp_port->pcpu_stats);
+	u64_stats_update_begin(&pcpu_stats->syncp);
+	pcpu_stats->rx_packets++;
+	pcpu_stats->rx_bytes += skb->len;
+	u64_stats_update_end(&pcpu_stats->syncp);
+
+	skb->protocol = eth_type_trans(skb, skb->dev);
+	netif_receive_skb(skb);
+}
+
+static void mlxsw_sp_rx_listener_mark_func(struct sk_buff *skb, u16 local_port,
+					   void *priv)
+{
+	skb->offload_fwd_mark = 1;
+	return mlxsw_sp_rx_listener_no_mark_func(skb, local_port, priv);
+}
+
+static void mlxsw_sp_rx_listener_l3_mark_func(struct sk_buff *skb,
+					      u16 local_port, void *priv)
+{
+	skb->offload_l3_fwd_mark = 1;
+	skb->offload_fwd_mark = 1;
+	return mlxsw_sp_rx_listener_no_mark_func(skb, local_port, priv);
+}
+
+void mlxsw_sp_ptp_receive(struct mlxsw_sp *mlxsw_sp, struct sk_buff *skb,
+			  u16 local_port)
+{
+	mlxsw_sp->ptp_ops->receive(mlxsw_sp, skb, local_port);
+}
+
+#define MLXSW_SP_RXL_NO_MARK(_trap_id, _action, _trap_group, _is_ctrl)	\
+	MLXSW_RXL(mlxsw_sp_rx_listener_no_mark_func, _trap_id, _action,	\
+		  _is_ctrl, SP_##_trap_group, DISCARD)
+
+#define MLXSW_SP_RXL_MARK(_trap_id, _action, _trap_group, _is_ctrl)	\
+	MLXSW_RXL(mlxsw_sp_rx_listener_mark_func, _trap_id, _action,	\
+		_is_ctrl, SP_##_trap_group, DISCARD)
+
+#define MLXSW_SP_RXL_L3_MARK(_trap_id, _action, _trap_group, _is_ctrl)	\
+	MLXSW_RXL(mlxsw_sp_rx_listener_l3_mark_func, _trap_id, _action,	\
+		_is_ctrl, SP_##_trap_group, DISCARD)
+
+#define MLXSW_SP_EVENTL(_func, _trap_id)		\
+	MLXSW_EVENTL(_func, _trap_id, SP_EVENT)
+
+static const struct mlxsw_listener mlxsw_sp_listener[] = {
+	/* Events */
+	MLXSW_SP_EVENTL(mlxsw_sp_pude_event_func, PUDE),
+	/* L2 traps */
+	MLXSW_SP_RXL_NO_MARK(FID_MISS, TRAP_TO_CPU, FID_MISS, false),
+	/* L3 traps */
+	MLXSW_SP_RXL_MARK(IPV6_UNSPECIFIED_ADDRESS, TRAP_TO_CPU, ROUTER_EXP,
+			  false),
+	MLXSW_SP_RXL_MARK(IPV6_LINK_LOCAL_SRC, TRAP_TO_CPU, ROUTER_EXP, false),
+	MLXSW_SP_RXL_MARK(IPV6_MC_LINK_LOCAL_DEST, TRAP_TO_CPU, ROUTER_EXP,
+			  false),
+	MLXSW_SP_RXL_NO_MARK(DISCARD_ING_ROUTER_SIP_CLASS_E, FORWARD,
+			     ROUTER_EXP, false),
+	MLXSW_SP_RXL_NO_MARK(DISCARD_ING_ROUTER_MC_DMAC, FORWARD,
+			     ROUTER_EXP, false),
+	MLXSW_SP_RXL_NO_MARK(DISCARD_ING_ROUTER_SIP_DIP, FORWARD,
+			     ROUTER_EXP, false),
+	MLXSW_SP_RXL_NO_MARK(DISCARD_ING_ROUTER_DIP_LINK_LOCAL, FORWARD,
+			     ROUTER_EXP, false),
+	/* Multicast Router Traps */
+	MLXSW_SP_RXL_MARK(ACL1, TRAP_TO_CPU, MULTICAST, false),
+	MLXSW_SP_RXL_L3_MARK(ACL2, TRAP_TO_CPU, MULTICAST, false),
+	/* NVE traps */
+	MLXSW_SP_RXL_MARK(NVE_ENCAP_ARP, TRAP_TO_CPU, NEIGH_DISCOVERY, false),
+};
+
+static const struct mlxsw_listener mlxsw_sp1_listener[] = {
+	/* Events */
+	MLXSW_EVENTL(mlxsw_sp1_ptp_egr_fifo_event_func, PTP_EGR_FIFO, SP_PTP0),
+	MLXSW_EVENTL(mlxsw_sp1_ptp_ing_fifo_event_func, PTP_ING_FIFO, SP_PTP0),
+};
+
+static const struct mlxsw_listener mlxsw_sp2_listener[] = {
+	/* Events */
+	MLXSW_SP_EVENTL(mlxsw_sp_port_mapping_listener_func, PMLPE),
+};
+
+static int mlxsw_sp_cpu_policers_set(struct mlxsw_core *mlxsw_core)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_core_driver_priv(mlxsw_core);
+	char qpcr_pl[MLXSW_REG_QPCR_LEN];
+	enum mlxsw_reg_qpcr_ir_units ir_units;
+	int max_cpu_policers;
+	bool is_bytes;
+	u8 burst_size;
+	u32 rate;
+	int i, err;
+
+	if (!MLXSW_CORE_RES_VALID(mlxsw_core, MAX_CPU_POLICERS))
+		return -EIO;
+
+	max_cpu_policers = MLXSW_CORE_RES_GET(mlxsw_core, MAX_CPU_POLICERS);
+
+	ir_units = MLXSW_REG_QPCR_IR_UNITS_M;
+	for (i = 0; i < max_cpu_policers; i++) {
+		is_bytes = false;
+		switch (i) {
+		case MLXSW_REG_HTGT_TRAP_GROUP_SP_ROUTER_EXP:
+		case MLXSW_REG_HTGT_TRAP_GROUP_SP_MULTICAST:
+		case MLXSW_REG_HTGT_TRAP_GROUP_SP_FID_MISS:
+			rate = 1024;
+			burst_size = 7;
+			break;
+		default:
+			continue;
+		}
+
+		__set_bit(i, mlxsw_sp->trap->policers_usage);
+		mlxsw_reg_qpcr_pack(qpcr_pl, i, ir_units, is_bytes, rate,
+				    burst_size);
+		err = mlxsw_reg_write(mlxsw_core, MLXSW_REG(qpcr), qpcr_pl);
+		if (err)
+			return err;
+	}
+
+	return 0;
+}
+
+static int mlxsw_sp_trap_groups_set(struct mlxsw_core *mlxsw_core)
+{
+	char htgt_pl[MLXSW_REG_HTGT_LEN];
+	enum mlxsw_reg_htgt_trap_group i;
+	int max_cpu_policers;
+	int max_trap_groups;
+	u8 priority, tc;
+	u16 policer_id;
+	int err;
+
+	if (!MLXSW_CORE_RES_VALID(mlxsw_core, MAX_TRAP_GROUPS))
+		return -EIO;
+
+	max_trap_groups = MLXSW_CORE_RES_GET(mlxsw_core, MAX_TRAP_GROUPS);
+	max_cpu_policers = MLXSW_CORE_RES_GET(mlxsw_core, MAX_CPU_POLICERS);
+
+	for (i = 0; i < max_trap_groups; i++) {
+		policer_id = i;
+		switch (i) {
+		case MLXSW_REG_HTGT_TRAP_GROUP_SP_ROUTER_EXP:
+		case MLXSW_REG_HTGT_TRAP_GROUP_SP_MULTICAST:
+		case MLXSW_REG_HTGT_TRAP_GROUP_SP_FID_MISS:
+			priority = 1;
+			tc = 1;
+			break;
+		case MLXSW_REG_HTGT_TRAP_GROUP_SP_EVENT:
+			priority = MLXSW_REG_HTGT_DEFAULT_PRIORITY;
+			tc = MLXSW_REG_HTGT_DEFAULT_TC;
+			policer_id = MLXSW_REG_HTGT_INVALID_POLICER;
+			break;
+		default:
+			continue;
+		}
+
+		if (max_cpu_policers <= policer_id &&
+		    policer_id != MLXSW_REG_HTGT_INVALID_POLICER)
+			return -EIO;
+
+		mlxsw_reg_htgt_pack(htgt_pl, i, policer_id, priority, tc);
+		err = mlxsw_reg_write(mlxsw_core, MLXSW_REG(htgt), htgt_pl);
+		if (err)
+			return err;
+	}
+
+	return 0;
+}
+
+static int mlxsw_sp_traps_init(struct mlxsw_sp *mlxsw_sp)
+{
+	struct mlxsw_sp_trap *trap;
+	u64 max_policers;
+	int err;
+
+	if (!MLXSW_CORE_RES_VALID(mlxsw_sp->core, MAX_CPU_POLICERS))
+		return -EIO;
+	max_policers = MLXSW_CORE_RES_GET(mlxsw_sp->core, MAX_CPU_POLICERS);
+	trap = kzalloc(struct_size(trap, policers_usage,
+				   BITS_TO_LONGS(max_policers)), GFP_KERNEL);
+	if (!trap)
+		return -ENOMEM;
+	trap->max_policers = max_policers;
+	mlxsw_sp->trap = trap;
+
+	err = mlxsw_sp_cpu_policers_set(mlxsw_sp->core);
+	if (err)
+		goto err_cpu_policers_set;
+
+	err = mlxsw_sp_trap_groups_set(mlxsw_sp->core);
+	if (err)
+		goto err_trap_groups_set;
+
+	err = mlxsw_core_traps_register(mlxsw_sp->core, mlxsw_sp_listener,
+					ARRAY_SIZE(mlxsw_sp_listener),
+					mlxsw_sp);
+	if (err)
+		goto err_traps_register;
+
+	err = mlxsw_core_traps_register(mlxsw_sp->core, mlxsw_sp->listeners,
+					mlxsw_sp->listeners_count, mlxsw_sp);
+	if (err)
+		goto err_extra_traps_init;
+
+	return 0;
+
+err_extra_traps_init:
+	mlxsw_core_traps_unregister(mlxsw_sp->core, mlxsw_sp_listener,
+				    ARRAY_SIZE(mlxsw_sp_listener),
+				    mlxsw_sp);
+err_traps_register:
+err_trap_groups_set:
+err_cpu_policers_set:
+	kfree(trap);
+	return err;
+}
+
+static void mlxsw_sp_traps_fini(struct mlxsw_sp *mlxsw_sp)
+{
+	mlxsw_core_traps_unregister(mlxsw_sp->core, mlxsw_sp->listeners,
+				    mlxsw_sp->listeners_count,
+				    mlxsw_sp);
+	mlxsw_core_traps_unregister(mlxsw_sp->core, mlxsw_sp_listener,
+				    ARRAY_SIZE(mlxsw_sp_listener), mlxsw_sp);
+	kfree(mlxsw_sp->trap);
+}
+
+#define MLXSW_SP_LAG_SEED_INIT 0xcafecafe
+
+static int mlxsw_sp_lag_init(struct mlxsw_sp *mlxsw_sp)
+{
+	char slcr_pl[MLXSW_REG_SLCR_LEN];
+	u16 max_lag;
+	u32 seed;
+	int err;
+
+	seed = jhash(mlxsw_sp->base_mac, sizeof(mlxsw_sp->base_mac),
+		     MLXSW_SP_LAG_SEED_INIT);
+	mlxsw_reg_slcr_pack(slcr_pl, MLXSW_REG_SLCR_LAG_HASH_SMAC |
+				     MLXSW_REG_SLCR_LAG_HASH_DMAC |
+				     MLXSW_REG_SLCR_LAG_HASH_ETHERTYPE |
+				     MLXSW_REG_SLCR_LAG_HASH_VLANID |
+				     MLXSW_REG_SLCR_LAG_HASH_SIP |
+				     MLXSW_REG_SLCR_LAG_HASH_DIP |
+				     MLXSW_REG_SLCR_LAG_HASH_SPORT |
+				     MLXSW_REG_SLCR_LAG_HASH_DPORT |
+				     MLXSW_REG_SLCR_LAG_HASH_IPPROTO, seed);
+	err = mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(slcr), slcr_pl);
+	if (err)
+		return err;
+
+	err = mlxsw_core_max_lag(mlxsw_sp->core, &max_lag);
+	if (err)
+		return err;
+
+	if (!MLXSW_CORE_RES_VALID(mlxsw_sp->core, MAX_LAG_MEMBERS))
+		return -EIO;
+
+	mlxsw_sp->lags = kcalloc(max_lag, sizeof(struct mlxsw_sp_upper),
+				 GFP_KERNEL);
+	if (!mlxsw_sp->lags)
+		return -ENOMEM;
+
+	return 0;
+}
+
+static void mlxsw_sp_lag_fini(struct mlxsw_sp *mlxsw_sp)
+{
+	kfree(mlxsw_sp->lags);
+}
+
+static const struct mlxsw_sp_ptp_ops mlxsw_sp1_ptp_ops = {
+	.clock_init	= mlxsw_sp1_ptp_clock_init,
+	.clock_fini	= mlxsw_sp1_ptp_clock_fini,
+	.init		= mlxsw_sp1_ptp_init,
+	.fini		= mlxsw_sp1_ptp_fini,
+	.receive	= mlxsw_sp1_ptp_receive,
+	.transmitted	= mlxsw_sp1_ptp_transmitted,
+	.hwtstamp_get	= mlxsw_sp1_ptp_hwtstamp_get,
+	.hwtstamp_set	= mlxsw_sp1_ptp_hwtstamp_set,
+	.shaper_work	= mlxsw_sp1_ptp_shaper_work,
+	.get_ts_info	= mlxsw_sp1_ptp_get_ts_info,
+	.get_stats_count = mlxsw_sp1_get_stats_count,
+	.get_stats_strings = mlxsw_sp1_get_stats_strings,
+	.get_stats	= mlxsw_sp1_get_stats,
+	.txhdr_construct = mlxsw_sp_ptp_txhdr_construct,
+};
+
+static const struct mlxsw_sp_ptp_ops mlxsw_sp2_ptp_ops = {
+	.clock_init	= mlxsw_sp2_ptp_clock_init,
+	.clock_fini	= mlxsw_sp2_ptp_clock_fini,
+	.init		= mlxsw_sp2_ptp_init,
+	.fini		= mlxsw_sp2_ptp_fini,
+	.receive	= mlxsw_sp2_ptp_receive,
+	.transmitted	= mlxsw_sp2_ptp_transmitted,
+	.hwtstamp_get	= mlxsw_sp2_ptp_hwtstamp_get,
+	.hwtstamp_set	= mlxsw_sp2_ptp_hwtstamp_set,
+	.shaper_work	= mlxsw_sp2_ptp_shaper_work,
+	.get_ts_info	= mlxsw_sp2_ptp_get_ts_info,
+	.get_stats_count = mlxsw_sp2_get_stats_count,
+	.get_stats_strings = mlxsw_sp2_get_stats_strings,
+	.get_stats	= mlxsw_sp2_get_stats,
+	.txhdr_construct = mlxsw_sp2_ptp_txhdr_construct,
+};
+
+static const struct mlxsw_sp_ptp_ops mlxsw_sp4_ptp_ops = {
+	.clock_init	= mlxsw_sp2_ptp_clock_init,
+	.clock_fini	= mlxsw_sp2_ptp_clock_fini,
+	.init		= mlxsw_sp2_ptp_init,
+	.fini		= mlxsw_sp2_ptp_fini,
+	.receive	= mlxsw_sp2_ptp_receive,
+	.transmitted	= mlxsw_sp2_ptp_transmitted,
+	.hwtstamp_get	= mlxsw_sp2_ptp_hwtstamp_get,
+	.hwtstamp_set	= mlxsw_sp2_ptp_hwtstamp_set,
+	.shaper_work	= mlxsw_sp2_ptp_shaper_work,
+	.get_ts_info	= mlxsw_sp2_ptp_get_ts_info,
+	.get_stats_count = mlxsw_sp2_get_stats_count,
+	.get_stats_strings = mlxsw_sp2_get_stats_strings,
+	.get_stats	= mlxsw_sp2_get_stats,
+	.txhdr_construct = mlxsw_sp_ptp_txhdr_construct,
+};
+
+struct mlxsw_sp_sample_trigger_node {
+	struct mlxsw_sp_sample_trigger trigger;
+	struct mlxsw_sp_sample_params params;
+	struct rhash_head ht_node;
+	struct rcu_head rcu;
+	refcount_t refcount;
+};
+
+static const struct rhashtable_params mlxsw_sp_sample_trigger_ht_params = {
+	.key_offset = offsetof(struct mlxsw_sp_sample_trigger_node, trigger),
+	.head_offset = offsetof(struct mlxsw_sp_sample_trigger_node, ht_node),
+	.key_len = sizeof(struct mlxsw_sp_sample_trigger),
+	.automatic_shrinking = true,
+};
+
+static void
+mlxsw_sp_sample_trigger_key_init(struct mlxsw_sp_sample_trigger *key,
+				 const struct mlxsw_sp_sample_trigger *trigger)
+{
+	memset(key, 0, sizeof(*key));
+	key->type = trigger->type;
+	key->local_port = trigger->local_port;
+}
+
+/* RCU read lock must be held */
+struct mlxsw_sp_sample_params *
+mlxsw_sp_sample_trigger_params_lookup(struct mlxsw_sp *mlxsw_sp,
+				      const struct mlxsw_sp_sample_trigger *trigger)
+{
+	struct mlxsw_sp_sample_trigger_node *trigger_node;
+	struct mlxsw_sp_sample_trigger key;
+
+	mlxsw_sp_sample_trigger_key_init(&key, trigger);
+	trigger_node = rhashtable_lookup(&mlxsw_sp->sample_trigger_ht, &key,
+					 mlxsw_sp_sample_trigger_ht_params);
+	if (!trigger_node)
+		return NULL;
+
+	return &trigger_node->params;
+}
+
+static int
+mlxsw_sp_sample_trigger_node_init(struct mlxsw_sp *mlxsw_sp,
+				  const struct mlxsw_sp_sample_trigger *trigger,
+				  const struct mlxsw_sp_sample_params *params)
+{
+	struct mlxsw_sp_sample_trigger_node *trigger_node;
+	int err;
+
+	trigger_node = kzalloc(sizeof(*trigger_node), GFP_KERNEL);
+	if (!trigger_node)
+		return -ENOMEM;
+
+	trigger_node->trigger = *trigger;
+	trigger_node->params = *params;
+	refcount_set(&trigger_node->refcount, 1);
+
+	err = rhashtable_insert_fast(&mlxsw_sp->sample_trigger_ht,
+				     &trigger_node->ht_node,
+				     mlxsw_sp_sample_trigger_ht_params);
+	if (err)
+		goto err_rhashtable_insert;
+
+	return 0;
+
+err_rhashtable_insert:
+	kfree(trigger_node);
+	return err;
+}
+
+static void
+mlxsw_sp_sample_trigger_node_fini(struct mlxsw_sp *mlxsw_sp,
+				  struct mlxsw_sp_sample_trigger_node *trigger_node)
+{
+	rhashtable_remove_fast(&mlxsw_sp->sample_trigger_ht,
+			       &trigger_node->ht_node,
+			       mlxsw_sp_sample_trigger_ht_params);
+	kfree_rcu(trigger_node, rcu);
+}
+
+int
+mlxsw_sp_sample_trigger_params_set(struct mlxsw_sp *mlxsw_sp,
+				   const struct mlxsw_sp_sample_trigger *trigger,
+				   const struct mlxsw_sp_sample_params *params,
+				   struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp_sample_trigger_node *trigger_node;
+	struct mlxsw_sp_sample_trigger key;
+
+	ASSERT_RTNL();
+
+	mlxsw_sp_sample_trigger_key_init(&key, trigger);
+
+	trigger_node = rhashtable_lookup_fast(&mlxsw_sp->sample_trigger_ht,
+					      &key,
+					      mlxsw_sp_sample_trigger_ht_params);
+	if (!trigger_node)
+		return mlxsw_sp_sample_trigger_node_init(mlxsw_sp, &key,
+							 params);
+
+	if (trigger_node->trigger.local_port) {
+		NL_SET_ERR_MSG_MOD(extack, "Sampling already enabled on port");
+		return -EINVAL;
+	}
+
+	if (trigger_node->params.psample_group != params->psample_group ||
+	    trigger_node->params.truncate != params->truncate ||
+	    trigger_node->params.rate != params->rate ||
+	    trigger_node->params.trunc_size != params->trunc_size) {
+		NL_SET_ERR_MSG_MOD(extack, "Sampling parameters do not match for an existing sampling trigger");
+		return -EINVAL;
+	}
+
+	refcount_inc(&trigger_node->refcount);
+
+	return 0;
+}
+
+void
+mlxsw_sp_sample_trigger_params_unset(struct mlxsw_sp *mlxsw_sp,
+				     const struct mlxsw_sp_sample_trigger *trigger)
+{
+	struct mlxsw_sp_sample_trigger_node *trigger_node;
+	struct mlxsw_sp_sample_trigger key;
+
+	ASSERT_RTNL();
+
+	mlxsw_sp_sample_trigger_key_init(&key, trigger);
+
+	trigger_node = rhashtable_lookup_fast(&mlxsw_sp->sample_trigger_ht,
+					      &key,
+					      mlxsw_sp_sample_trigger_ht_params);
+	if (!trigger_node)
+		return;
+
+	if (!refcount_dec_and_test(&trigger_node->refcount))
+		return;
+
+	mlxsw_sp_sample_trigger_node_fini(mlxsw_sp, trigger_node);
+}
+
+static int mlxsw_sp_netdevice_event(struct notifier_block *unused,
+				    unsigned long event, void *ptr);
+
+#define MLXSW_SP_DEFAULT_PARSING_DEPTH 96
+#define MLXSW_SP_INCREASED_PARSING_DEPTH 128
+#define MLXSW_SP_DEFAULT_VXLAN_UDP_DPORT 4789
+
+static void mlxsw_sp_parsing_init(struct mlxsw_sp *mlxsw_sp)
+{
+	refcount_set(&mlxsw_sp->parsing.parsing_depth_ref, 0);
+	mlxsw_sp->parsing.parsing_depth = MLXSW_SP_DEFAULT_PARSING_DEPTH;
+	mlxsw_sp->parsing.vxlan_udp_dport = MLXSW_SP_DEFAULT_VXLAN_UDP_DPORT;
+	mutex_init(&mlxsw_sp->parsing.lock);
+}
+
+static void mlxsw_sp_parsing_fini(struct mlxsw_sp *mlxsw_sp)
+{
+	mutex_destroy(&mlxsw_sp->parsing.lock);
+	WARN_ON_ONCE(refcount_read(&mlxsw_sp->parsing.parsing_depth_ref));
+}
+
+struct mlxsw_sp_ipv6_addr_node {
+	struct in6_addr key;
+	struct rhash_head ht_node;
+	u32 kvdl_index;
+	refcount_t refcount;
+};
+
+static const struct rhashtable_params mlxsw_sp_ipv6_addr_ht_params = {
+	.key_offset = offsetof(struct mlxsw_sp_ipv6_addr_node, key),
+	.head_offset = offsetof(struct mlxsw_sp_ipv6_addr_node, ht_node),
+	.key_len = sizeof(struct in6_addr),
+	.automatic_shrinking = true,
+};
+
+static int
+mlxsw_sp_ipv6_addr_init(struct mlxsw_sp *mlxsw_sp, const struct in6_addr *addr6,
+			u32 *p_kvdl_index)
+{
+	struct mlxsw_sp_ipv6_addr_node *node;
+	char rips_pl[MLXSW_REG_RIPS_LEN];
+	int err;
+
+	err = mlxsw_sp_kvdl_alloc(mlxsw_sp,
+				  MLXSW_SP_KVDL_ENTRY_TYPE_IPV6_ADDRESS, 1,
+				  p_kvdl_index);
+	if (err)
+		return err;
+
+	mlxsw_reg_rips_pack(rips_pl, *p_kvdl_index, addr6);
+	err = mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(rips), rips_pl);
+	if (err)
+		goto err_rips_write;
+
+	node = kzalloc(sizeof(*node), GFP_KERNEL);
+	if (!node) {
+		err = -ENOMEM;
+		goto err_node_alloc;
+	}
+
+	node->key = *addr6;
+	node->kvdl_index = *p_kvdl_index;
+	refcount_set(&node->refcount, 1);
+
+	err = rhashtable_insert_fast(&mlxsw_sp->ipv6_addr_ht,
+				     &node->ht_node,
+				     mlxsw_sp_ipv6_addr_ht_params);
+	if (err)
+		goto err_rhashtable_insert;
+
+	return 0;
+
+err_rhashtable_insert:
+	kfree(node);
+err_node_alloc:
+err_rips_write:
+	mlxsw_sp_kvdl_free(mlxsw_sp, MLXSW_SP_KVDL_ENTRY_TYPE_IPV6_ADDRESS, 1,
+			   *p_kvdl_index);
+	return err;
+}
+
+static void mlxsw_sp_ipv6_addr_fini(struct mlxsw_sp *mlxsw_sp,
+				    struct mlxsw_sp_ipv6_addr_node *node)
+{
+	u32 kvdl_index = node->kvdl_index;
+
+	rhashtable_remove_fast(&mlxsw_sp->ipv6_addr_ht, &node->ht_node,
+			       mlxsw_sp_ipv6_addr_ht_params);
+	kfree(node);
+	mlxsw_sp_kvdl_free(mlxsw_sp, MLXSW_SP_KVDL_ENTRY_TYPE_IPV6_ADDRESS, 1,
+			   kvdl_index);
+}
+
+int mlxsw_sp_ipv6_addr_kvdl_index_get(struct mlxsw_sp *mlxsw_sp,
+				      const struct in6_addr *addr6,
+				      u32 *p_kvdl_index)
+{
+	struct mlxsw_sp_ipv6_addr_node *node;
+	int err = 0;
+
+	mutex_lock(&mlxsw_sp->ipv6_addr_ht_lock);
+	node = rhashtable_lookup_fast(&mlxsw_sp->ipv6_addr_ht, addr6,
+				      mlxsw_sp_ipv6_addr_ht_params);
+	if (node) {
+		refcount_inc(&node->refcount);
+		*p_kvdl_index = node->kvdl_index;
+		goto out_unlock;
+	}
+
+	err = mlxsw_sp_ipv6_addr_init(mlxsw_sp, addr6, p_kvdl_index);
+
+out_unlock:
+	mutex_unlock(&mlxsw_sp->ipv6_addr_ht_lock);
+	return err;
+}
+
+void
+mlxsw_sp_ipv6_addr_put(struct mlxsw_sp *mlxsw_sp, const struct in6_addr *addr6)
+{
+	struct mlxsw_sp_ipv6_addr_node *node;
+
+	mutex_lock(&mlxsw_sp->ipv6_addr_ht_lock);
+	node = rhashtable_lookup_fast(&mlxsw_sp->ipv6_addr_ht, addr6,
+				      mlxsw_sp_ipv6_addr_ht_params);
+	if (WARN_ON(!node))
+		goto out_unlock;
+
+	if (!refcount_dec_and_test(&node->refcount))
+		goto out_unlock;
+
+	mlxsw_sp_ipv6_addr_fini(mlxsw_sp, node);
+
+out_unlock:
+	mutex_unlock(&mlxsw_sp->ipv6_addr_ht_lock);
+}
+
+static int mlxsw_sp_ipv6_addr_ht_init(struct mlxsw_sp *mlxsw_sp)
+{
+	int err;
+
+	err = rhashtable_init(&mlxsw_sp->ipv6_addr_ht,
+			      &mlxsw_sp_ipv6_addr_ht_params);
+	if (err)
+		return err;
+
+	mutex_init(&mlxsw_sp->ipv6_addr_ht_lock);
+	return 0;
+}
+
+static void mlxsw_sp_ipv6_addr_ht_fini(struct mlxsw_sp *mlxsw_sp)
+{
+	mutex_destroy(&mlxsw_sp->ipv6_addr_ht_lock);
+	rhashtable_destroy(&mlxsw_sp->ipv6_addr_ht);
+}
+
+static int mlxsw_sp_init(struct mlxsw_core *mlxsw_core,
+			 const struct mlxsw_bus_info *mlxsw_bus_info,
+			 struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_core_driver_priv(mlxsw_core);
+	int err;
+
+	mlxsw_sp->core = mlxsw_core;
+	mlxsw_sp->bus_info = mlxsw_bus_info;
+
+	mlxsw_sp_parsing_init(mlxsw_sp);
+	mlxsw_core_emad_string_tlv_enable(mlxsw_core);
+
+	err = mlxsw_sp_base_mac_get(mlxsw_sp);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Failed to get base mac\n");
+		return err;
+	}
+
+	err = mlxsw_sp_kvdl_init(mlxsw_sp);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Failed to initialize KVDL\n");
+		return err;
+	}
+
+	err = mlxsw_sp_pgt_init(mlxsw_sp);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Failed to initialize PGT\n");
+		goto err_pgt_init;
+	}
+
+	err = mlxsw_sp_fids_init(mlxsw_sp);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Failed to initialize FIDs\n");
+		goto err_fids_init;
+	}
+
+	err = mlxsw_sp_policers_init(mlxsw_sp);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Failed to initialize policers\n");
+		goto err_policers_init;
+	}
+
+	err = mlxsw_sp_traps_init(mlxsw_sp);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Failed to set traps\n");
+		goto err_traps_init;
+	}
+
+	err = mlxsw_sp_devlink_traps_init(mlxsw_sp);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Failed to initialize devlink traps\n");
+		goto err_devlink_traps_init;
+	}
+
+	err = mlxsw_sp_buffers_init(mlxsw_sp);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Failed to initialize buffers\n");
+		goto err_buffers_init;
+	}
+
+	err = mlxsw_sp_lag_init(mlxsw_sp);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Failed to initialize LAG\n");
+		goto err_lag_init;
+	}
+
+	/* Initialize SPAN before router and switchdev, so that those components
+	 * can call mlxsw_sp_span_respin().
+	 */
+	err = mlxsw_sp_span_init(mlxsw_sp);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Failed to init span system\n");
+		goto err_span_init;
+	}
+
+	err = mlxsw_sp_switchdev_init(mlxsw_sp);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Failed to initialize switchdev\n");
+		goto err_switchdev_init;
+	}
+
+	err = mlxsw_sp_counter_pool_init(mlxsw_sp);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Failed to init counter pool\n");
+		goto err_counter_pool_init;
+	}
+
+	err = mlxsw_sp_afa_init(mlxsw_sp);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Failed to initialize ACL actions\n");
+		goto err_afa_init;
+	}
+
+	err = mlxsw_sp_ipv6_addr_ht_init(mlxsw_sp);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Failed to initialize hash table for IPv6 addresses\n");
+		goto err_ipv6_addr_ht_init;
+	}
+
+	err = mlxsw_sp_nve_init(mlxsw_sp);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Failed to initialize NVE\n");
+		goto err_nve_init;
+	}
+
+	err = mlxsw_sp_acl_init(mlxsw_sp);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Failed to initialize ACL\n");
+		goto err_acl_init;
+	}
+
+	err = mlxsw_sp_router_init(mlxsw_sp, extack);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Failed to initialize router\n");
+		goto err_router_init;
+	}
+
+	if (mlxsw_sp->bus_info->read_clock_capable) {
+		/* NULL is a valid return value from clock_init */
+		mlxsw_sp->clock =
+			mlxsw_sp->ptp_ops->clock_init(mlxsw_sp,
+						      mlxsw_sp->bus_info->dev);
+		if (IS_ERR(mlxsw_sp->clock)) {
+			err = PTR_ERR(mlxsw_sp->clock);
+			dev_err(mlxsw_sp->bus_info->dev, "Failed to init ptp clock\n");
+			goto err_ptp_clock_init;
+		}
+	}
+
+	if (mlxsw_sp->clock) {
+		/* NULL is a valid return value from ptp_ops->init */
+		mlxsw_sp->ptp_state = mlxsw_sp->ptp_ops->init(mlxsw_sp);
+		if (IS_ERR(mlxsw_sp->ptp_state)) {
+			err = PTR_ERR(mlxsw_sp->ptp_state);
+			dev_err(mlxsw_sp->bus_info->dev, "Failed to initialize PTP\n");
+			goto err_ptp_init;
+		}
+	}
+
+	/* Initialize netdevice notifier after SPAN is initialized, so that the
+	 * event handler can call SPAN respin.
+	 */
+	mlxsw_sp->netdevice_nb.notifier_call = mlxsw_sp_netdevice_event;
+	err = register_netdevice_notifier_net(mlxsw_sp_net(mlxsw_sp),
+					      &mlxsw_sp->netdevice_nb);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Failed to register netdev notifier\n");
+		goto err_netdev_notifier;
+	}
+
+	err = mlxsw_sp_dpipe_init(mlxsw_sp);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Failed to init pipeline debug\n");
+		goto err_dpipe_init;
+	}
+
+	err = mlxsw_sp_port_module_info_init(mlxsw_sp);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Failed to init port module info\n");
+		goto err_port_module_info_init;
+	}
+
+	err = rhashtable_init(&mlxsw_sp->sample_trigger_ht,
+			      &mlxsw_sp_sample_trigger_ht_params);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Failed to init sampling trigger hashtable\n");
+		goto err_sample_trigger_init;
+	}
+
+	err = mlxsw_sp_ports_create(mlxsw_sp);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Failed to create ports\n");
+		goto err_ports_create;
+	}
+
+	return 0;
+
+err_ports_create:
+	rhashtable_destroy(&mlxsw_sp->sample_trigger_ht);
+err_sample_trigger_init:
+	mlxsw_sp_port_module_info_fini(mlxsw_sp);
+err_port_module_info_init:
+	mlxsw_sp_dpipe_fini(mlxsw_sp);
+err_dpipe_init:
+	unregister_netdevice_notifier_net(mlxsw_sp_net(mlxsw_sp),
+					  &mlxsw_sp->netdevice_nb);
+err_netdev_notifier:
+	if (mlxsw_sp->clock)
+		mlxsw_sp->ptp_ops->fini(mlxsw_sp->ptp_state);
+err_ptp_init:
+	if (mlxsw_sp->clock)
+		mlxsw_sp->ptp_ops->clock_fini(mlxsw_sp->clock);
+err_ptp_clock_init:
+	mlxsw_sp_router_fini(mlxsw_sp);
+err_router_init:
+	mlxsw_sp_acl_fini(mlxsw_sp);
+err_acl_init:
+	mlxsw_sp_nve_fini(mlxsw_sp);
+err_nve_init:
+	mlxsw_sp_ipv6_addr_ht_fini(mlxsw_sp);
+err_ipv6_addr_ht_init:
+	mlxsw_sp_afa_fini(mlxsw_sp);
+err_afa_init:
+	mlxsw_sp_counter_pool_fini(mlxsw_sp);
+err_counter_pool_init:
+	mlxsw_sp_switchdev_fini(mlxsw_sp);
+err_switchdev_init:
+	mlxsw_sp_span_fini(mlxsw_sp);
+err_span_init:
+	mlxsw_sp_lag_fini(mlxsw_sp);
+err_lag_init:
+	mlxsw_sp_buffers_fini(mlxsw_sp);
+err_buffers_init:
+	mlxsw_sp_devlink_traps_fini(mlxsw_sp);
+err_devlink_traps_init:
+	mlxsw_sp_traps_fini(mlxsw_sp);
+err_traps_init:
+	mlxsw_sp_policers_fini(mlxsw_sp);
+err_policers_init:
+	mlxsw_sp_fids_fini(mlxsw_sp);
+err_fids_init:
+	mlxsw_sp_pgt_fini(mlxsw_sp);
+err_pgt_init:
+	mlxsw_sp_kvdl_fini(mlxsw_sp);
+	mlxsw_sp_parsing_fini(mlxsw_sp);
+	return err;
+}
+
+static int mlxsw_sp1_init(struct mlxsw_core *mlxsw_core,
+			  const struct mlxsw_bus_info *mlxsw_bus_info,
+			  struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_core_driver_priv(mlxsw_core);
+
+	mlxsw_sp->switchdev_ops = &mlxsw_sp1_switchdev_ops;
+	mlxsw_sp->kvdl_ops = &mlxsw_sp1_kvdl_ops;
+	mlxsw_sp->afa_ops = &mlxsw_sp1_act_afa_ops;
+	mlxsw_sp->afk_ops = &mlxsw_sp1_afk_ops;
+	mlxsw_sp->mr_tcam_ops = &mlxsw_sp1_mr_tcam_ops;
+	mlxsw_sp->acl_rulei_ops = &mlxsw_sp1_acl_rulei_ops;
+	mlxsw_sp->acl_tcam_ops = &mlxsw_sp1_acl_tcam_ops;
+	mlxsw_sp->nve_ops_arr = mlxsw_sp1_nve_ops_arr;
+	mlxsw_sp->mac_mask = mlxsw_sp1_mac_mask;
+	mlxsw_sp->sb_vals = &mlxsw_sp1_sb_vals;
+	mlxsw_sp->sb_ops = &mlxsw_sp1_sb_ops;
+	mlxsw_sp->port_type_speed_ops = &mlxsw_sp1_port_type_speed_ops;
+	mlxsw_sp->ptp_ops = &mlxsw_sp1_ptp_ops;
+	mlxsw_sp->span_ops = &mlxsw_sp1_span_ops;
+	mlxsw_sp->policer_core_ops = &mlxsw_sp1_policer_core_ops;
+	mlxsw_sp->trap_ops = &mlxsw_sp1_trap_ops;
+	mlxsw_sp->mall_ops = &mlxsw_sp1_mall_ops;
+	mlxsw_sp->router_ops = &mlxsw_sp1_router_ops;
+	mlxsw_sp->listeners = mlxsw_sp1_listener;
+	mlxsw_sp->listeners_count = ARRAY_SIZE(mlxsw_sp1_listener);
+	mlxsw_sp->fid_family_arr = mlxsw_sp1_fid_family_arr;
+	mlxsw_sp->lowest_shaper_bs = MLXSW_REG_QEEC_LOWEST_SHAPER_BS_SP1;
+	mlxsw_sp->pgt_smpe_index_valid = true;
+
+	return mlxsw_sp_init(mlxsw_core, mlxsw_bus_info, extack);
+}
+
+static int mlxsw_sp2_init(struct mlxsw_core *mlxsw_core,
+			  const struct mlxsw_bus_info *mlxsw_bus_info,
+			  struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_core_driver_priv(mlxsw_core);
+
+	mlxsw_sp->switchdev_ops = &mlxsw_sp2_switchdev_ops;
+	mlxsw_sp->kvdl_ops = &mlxsw_sp2_kvdl_ops;
+	mlxsw_sp->afa_ops = &mlxsw_sp2_act_afa_ops;
+	mlxsw_sp->afk_ops = &mlxsw_sp2_afk_ops;
+	mlxsw_sp->mr_tcam_ops = &mlxsw_sp2_mr_tcam_ops;
+	mlxsw_sp->acl_rulei_ops = &mlxsw_sp2_acl_rulei_ops;
+	mlxsw_sp->acl_tcam_ops = &mlxsw_sp2_acl_tcam_ops;
+	mlxsw_sp->acl_bf_ops = &mlxsw_sp2_acl_bf_ops;
+	mlxsw_sp->nve_ops_arr = mlxsw_sp2_nve_ops_arr;
+	mlxsw_sp->mac_mask = mlxsw_sp2_mac_mask;
+	mlxsw_sp->sb_vals = &mlxsw_sp2_sb_vals;
+	mlxsw_sp->sb_ops = &mlxsw_sp2_sb_ops;
+	mlxsw_sp->port_type_speed_ops = &mlxsw_sp2_port_type_speed_ops;
+	mlxsw_sp->ptp_ops = &mlxsw_sp2_ptp_ops;
+	mlxsw_sp->span_ops = &mlxsw_sp2_span_ops;
+	mlxsw_sp->policer_core_ops = &mlxsw_sp2_policer_core_ops;
+	mlxsw_sp->trap_ops = &mlxsw_sp2_trap_ops;
+	mlxsw_sp->mall_ops = &mlxsw_sp2_mall_ops;
+	mlxsw_sp->router_ops = &mlxsw_sp2_router_ops;
+	mlxsw_sp->listeners = mlxsw_sp2_listener;
+	mlxsw_sp->listeners_count = ARRAY_SIZE(mlxsw_sp2_listener);
+	mlxsw_sp->fid_family_arr = mlxsw_sp2_fid_family_arr;
+	mlxsw_sp->lowest_shaper_bs = MLXSW_REG_QEEC_LOWEST_SHAPER_BS_SP2;
+	mlxsw_sp->pgt_smpe_index_valid = false;
+
+	return mlxsw_sp_init(mlxsw_core, mlxsw_bus_info, extack);
+}
+
+static int mlxsw_sp3_init(struct mlxsw_core *mlxsw_core,
+			  const struct mlxsw_bus_info *mlxsw_bus_info,
+			  struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_core_driver_priv(mlxsw_core);
+
+	mlxsw_sp->switchdev_ops = &mlxsw_sp2_switchdev_ops;
+	mlxsw_sp->kvdl_ops = &mlxsw_sp2_kvdl_ops;
+	mlxsw_sp->afa_ops = &mlxsw_sp2_act_afa_ops;
+	mlxsw_sp->afk_ops = &mlxsw_sp2_afk_ops;
+	mlxsw_sp->mr_tcam_ops = &mlxsw_sp2_mr_tcam_ops;
+	mlxsw_sp->acl_rulei_ops = &mlxsw_sp2_acl_rulei_ops;
+	mlxsw_sp->acl_tcam_ops = &mlxsw_sp2_acl_tcam_ops;
+	mlxsw_sp->acl_bf_ops = &mlxsw_sp2_acl_bf_ops;
+	mlxsw_sp->nve_ops_arr = mlxsw_sp2_nve_ops_arr;
+	mlxsw_sp->mac_mask = mlxsw_sp2_mac_mask;
+	mlxsw_sp->sb_vals = &mlxsw_sp2_sb_vals;
+	mlxsw_sp->sb_ops = &mlxsw_sp3_sb_ops;
+	mlxsw_sp->port_type_speed_ops = &mlxsw_sp2_port_type_speed_ops;
+	mlxsw_sp->ptp_ops = &mlxsw_sp2_ptp_ops;
+	mlxsw_sp->span_ops = &mlxsw_sp3_span_ops;
+	mlxsw_sp->policer_core_ops = &mlxsw_sp2_policer_core_ops;
+	mlxsw_sp->trap_ops = &mlxsw_sp2_trap_ops;
+	mlxsw_sp->mall_ops = &mlxsw_sp2_mall_ops;
+	mlxsw_sp->router_ops = &mlxsw_sp2_router_ops;
+	mlxsw_sp->listeners = mlxsw_sp2_listener;
+	mlxsw_sp->listeners_count = ARRAY_SIZE(mlxsw_sp2_listener);
+	mlxsw_sp->fid_family_arr = mlxsw_sp2_fid_family_arr;
+	mlxsw_sp->lowest_shaper_bs = MLXSW_REG_QEEC_LOWEST_SHAPER_BS_SP3;
+	mlxsw_sp->pgt_smpe_index_valid = false;
+
+	return mlxsw_sp_init(mlxsw_core, mlxsw_bus_info, extack);
+}
+
+static int mlxsw_sp4_init(struct mlxsw_core *mlxsw_core,
+			  const struct mlxsw_bus_info *mlxsw_bus_info,
+			  struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_core_driver_priv(mlxsw_core);
+
+	mlxsw_sp->switchdev_ops = &mlxsw_sp2_switchdev_ops;
+	mlxsw_sp->kvdl_ops = &mlxsw_sp2_kvdl_ops;
+	mlxsw_sp->afa_ops = &mlxsw_sp2_act_afa_ops;
+	mlxsw_sp->afk_ops = &mlxsw_sp4_afk_ops;
+	mlxsw_sp->mr_tcam_ops = &mlxsw_sp2_mr_tcam_ops;
+	mlxsw_sp->acl_rulei_ops = &mlxsw_sp2_acl_rulei_ops;
+	mlxsw_sp->acl_tcam_ops = &mlxsw_sp2_acl_tcam_ops;
+	mlxsw_sp->acl_bf_ops = &mlxsw_sp4_acl_bf_ops;
+	mlxsw_sp->nve_ops_arr = mlxsw_sp2_nve_ops_arr;
+	mlxsw_sp->mac_mask = mlxsw_sp2_mac_mask;
+	mlxsw_sp->sb_vals = &mlxsw_sp2_sb_vals;
+	mlxsw_sp->sb_ops = &mlxsw_sp3_sb_ops;
+	mlxsw_sp->port_type_speed_ops = &mlxsw_sp2_port_type_speed_ops;
+	mlxsw_sp->ptp_ops = &mlxsw_sp4_ptp_ops;
+	mlxsw_sp->span_ops = &mlxsw_sp3_span_ops;
+	mlxsw_sp->policer_core_ops = &mlxsw_sp2_policer_core_ops;
+	mlxsw_sp->trap_ops = &mlxsw_sp2_trap_ops;
+	mlxsw_sp->mall_ops = &mlxsw_sp2_mall_ops;
+	mlxsw_sp->router_ops = &mlxsw_sp2_router_ops;
+	mlxsw_sp->listeners = mlxsw_sp2_listener;
+	mlxsw_sp->listeners_count = ARRAY_SIZE(mlxsw_sp2_listener);
+	mlxsw_sp->fid_family_arr = mlxsw_sp2_fid_family_arr;
+	mlxsw_sp->lowest_shaper_bs = MLXSW_REG_QEEC_LOWEST_SHAPER_BS_SP4;
+	mlxsw_sp->pgt_smpe_index_valid = false;
+
+	return mlxsw_sp_init(mlxsw_core, mlxsw_bus_info, extack);
+}
+
+static void mlxsw_sp_fini(struct mlxsw_core *mlxsw_core)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_core_driver_priv(mlxsw_core);
+
+	mlxsw_sp_ports_remove(mlxsw_sp);
+	rhashtable_destroy(&mlxsw_sp->sample_trigger_ht);
+	mlxsw_sp_port_module_info_fini(mlxsw_sp);
+	mlxsw_sp_dpipe_fini(mlxsw_sp);
+	unregister_netdevice_notifier_net(mlxsw_sp_net(mlxsw_sp),
+					  &mlxsw_sp->netdevice_nb);
+	if (mlxsw_sp->clock) {
+		mlxsw_sp->ptp_ops->fini(mlxsw_sp->ptp_state);
+		mlxsw_sp->ptp_ops->clock_fini(mlxsw_sp->clock);
+	}
+	mlxsw_sp_router_fini(mlxsw_sp);
+	mlxsw_sp_acl_fini(mlxsw_sp);
+	mlxsw_sp_nve_fini(mlxsw_sp);
+	mlxsw_sp_ipv6_addr_ht_fini(mlxsw_sp);
+	mlxsw_sp_afa_fini(mlxsw_sp);
+	mlxsw_sp_counter_pool_fini(mlxsw_sp);
+	mlxsw_sp_switchdev_fini(mlxsw_sp);
+	mlxsw_sp_span_fini(mlxsw_sp);
+	mlxsw_sp_lag_fini(mlxsw_sp);
+	mlxsw_sp_buffers_fini(mlxsw_sp);
+	mlxsw_sp_devlink_traps_fini(mlxsw_sp);
+	mlxsw_sp_traps_fini(mlxsw_sp);
+	mlxsw_sp_policers_fini(mlxsw_sp);
+	mlxsw_sp_fids_fini(mlxsw_sp);
+	mlxsw_sp_pgt_fini(mlxsw_sp);
+	mlxsw_sp_kvdl_fini(mlxsw_sp);
+	mlxsw_sp_parsing_fini(mlxsw_sp);
+}
+
+static const struct mlxsw_config_profile mlxsw_sp1_config_profile = {
+	.used_flood_mode                = 1,
+	.flood_mode                     = MLXSW_CMD_MBOX_CONFIG_PROFILE_FLOOD_MODE_CONTROLLED,
+	.used_max_ib_mc			= 1,
+	.max_ib_mc			= 0,
+	.used_max_pkey			= 1,
+	.max_pkey			= 0,
+	.used_ubridge			= 1,
+	.ubridge			= 1,
+	.used_kvd_sizes			= 1,
+	.kvd_hash_single_parts		= 59,
+	.kvd_hash_double_parts		= 41,
+	.kvd_linear_size		= MLXSW_SP_KVD_LINEAR_SIZE,
+	.swid_config			= {
+		{
+			.used_type	= 1,
+			.type		= MLXSW_PORT_SWID_TYPE_ETH,
+		}
+	},
+};
+
+static const struct mlxsw_config_profile mlxsw_sp2_config_profile = {
+	.used_flood_mode                = 1,
+	.flood_mode                     = MLXSW_CMD_MBOX_CONFIG_PROFILE_FLOOD_MODE_CONTROLLED,
+	.used_max_ib_mc			= 1,
+	.max_ib_mc			= 0,
+	.used_max_pkey			= 1,
+	.max_pkey			= 0,
+	.used_ubridge			= 1,
+	.ubridge			= 1,
+	.swid_config			= {
+		{
+			.used_type	= 1,
+			.type		= MLXSW_PORT_SWID_TYPE_ETH,
+		}
+	},
+	.used_cqe_time_stamp_type	= 1,
+	.cqe_time_stamp_type		= MLXSW_CMD_MBOX_CONFIG_PROFILE_CQE_TIME_STAMP_TYPE_UTC,
+};
+
+/* Reduce number of LAGs from full capacity (256) to the maximum supported LAGs
+ * in Spectrum-2/3, to avoid regression in number of free entries in the PGT
+ * table.
+ */
+#define MLXSW_SP4_CONFIG_PROFILE_MAX_LAG 128
+
+static const struct mlxsw_config_profile mlxsw_sp4_config_profile = {
+	.used_max_lag			= 1,
+	.max_lag			= MLXSW_SP4_CONFIG_PROFILE_MAX_LAG,
+	.used_flood_mode                = 1,
+	.flood_mode                     = MLXSW_CMD_MBOX_CONFIG_PROFILE_FLOOD_MODE_CONTROLLED,
+	.used_max_ib_mc			= 1,
+	.max_ib_mc			= 0,
+	.used_max_pkey			= 1,
+	.max_pkey			= 0,
+	.used_ubridge			= 1,
+	.ubridge			= 1,
+	.swid_config			= {
+		{
+			.used_type	= 1,
+			.type		= MLXSW_PORT_SWID_TYPE_ETH,
+		}
+	},
+	.used_cqe_time_stamp_type	= 1,
+	.cqe_time_stamp_type		= MLXSW_CMD_MBOX_CONFIG_PROFILE_CQE_TIME_STAMP_TYPE_UTC,
+};
+
+static void
+mlxsw_sp_resource_size_params_prepare(struct mlxsw_core *mlxsw_core,
+				      struct devlink_resource_size_params *kvd_size_params,
+				      struct devlink_resource_size_params *linear_size_params,
+				      struct devlink_resource_size_params *hash_double_size_params,
+				      struct devlink_resource_size_params *hash_single_size_params)
+{
+	u32 single_size_min = MLXSW_CORE_RES_GET(mlxsw_core,
+						 KVD_SINGLE_MIN_SIZE);
+	u32 double_size_min = MLXSW_CORE_RES_GET(mlxsw_core,
+						 KVD_DOUBLE_MIN_SIZE);
+	u32 kvd_size = MLXSW_CORE_RES_GET(mlxsw_core, KVD_SIZE);
+	u32 linear_size_min = 0;
+
+	devlink_resource_size_params_init(kvd_size_params, kvd_size, kvd_size,
+					  MLXSW_SP_KVD_GRANULARITY,
+					  DEVLINK_RESOURCE_UNIT_ENTRY);
+	devlink_resource_size_params_init(linear_size_params, linear_size_min,
+					  kvd_size - single_size_min -
+					  double_size_min,
+					  MLXSW_SP_KVD_GRANULARITY,
+					  DEVLINK_RESOURCE_UNIT_ENTRY);
+	devlink_resource_size_params_init(hash_double_size_params,
+					  double_size_min,
+					  kvd_size - single_size_min -
+					  linear_size_min,
+					  MLXSW_SP_KVD_GRANULARITY,
+					  DEVLINK_RESOURCE_UNIT_ENTRY);
+	devlink_resource_size_params_init(hash_single_size_params,
+					  single_size_min,
+					  kvd_size - double_size_min -
+					  linear_size_min,
+					  MLXSW_SP_KVD_GRANULARITY,
+					  DEVLINK_RESOURCE_UNIT_ENTRY);
+}
+
+static int mlxsw_sp1_resources_kvd_register(struct mlxsw_core *mlxsw_core)
+{
+	struct devlink *devlink = priv_to_devlink(mlxsw_core);
+	struct devlink_resource_size_params hash_single_size_params;
+	struct devlink_resource_size_params hash_double_size_params;
+	struct devlink_resource_size_params linear_size_params;
+	struct devlink_resource_size_params kvd_size_params;
+	u32 kvd_size, single_size, double_size, linear_size;
+	const struct mlxsw_config_profile *profile;
+	int err;
+
+	profile = &mlxsw_sp1_config_profile;
+	if (!MLXSW_CORE_RES_VALID(mlxsw_core, KVD_SIZE))
+		return -EIO;
+
+	mlxsw_sp_resource_size_params_prepare(mlxsw_core, &kvd_size_params,
+					      &linear_size_params,
+					      &hash_double_size_params,
+					      &hash_single_size_params);
+
+	kvd_size = MLXSW_CORE_RES_GET(mlxsw_core, KVD_SIZE);
+	err = devl_resource_register(devlink, MLXSW_SP_RESOURCE_NAME_KVD,
+				     kvd_size, MLXSW_SP_RESOURCE_KVD,
+				     DEVLINK_RESOURCE_ID_PARENT_TOP,
+				     &kvd_size_params);
+	if (err)
+		return err;
+
+	linear_size = profile->kvd_linear_size;
+	err = devl_resource_register(devlink, MLXSW_SP_RESOURCE_NAME_KVD_LINEAR,
+				     linear_size,
+				     MLXSW_SP_RESOURCE_KVD_LINEAR,
+				     MLXSW_SP_RESOURCE_KVD,
+				     &linear_size_params);
+	if (err)
+		return err;
+
+	err = mlxsw_sp1_kvdl_resources_register(mlxsw_core);
+	if  (err)
+		return err;
+
+	double_size = kvd_size - linear_size;
+	double_size *= profile->kvd_hash_double_parts;
+	double_size /= profile->kvd_hash_double_parts +
+		       profile->kvd_hash_single_parts;
+	double_size = rounddown(double_size, MLXSW_SP_KVD_GRANULARITY);
+	err = devl_resource_register(devlink, MLXSW_SP_RESOURCE_NAME_KVD_HASH_DOUBLE,
+				     double_size,
+				     MLXSW_SP_RESOURCE_KVD_HASH_DOUBLE,
+				     MLXSW_SP_RESOURCE_KVD,
+				     &hash_double_size_params);
+	if (err)
+		return err;
+
+	single_size = kvd_size - double_size - linear_size;
+	err = devl_resource_register(devlink, MLXSW_SP_RESOURCE_NAME_KVD_HASH_SINGLE,
+				     single_size,
+				     MLXSW_SP_RESOURCE_KVD_HASH_SINGLE,
+				     MLXSW_SP_RESOURCE_KVD,
+				     &hash_single_size_params);
+	if (err)
+		return err;
+
+	return 0;
+}
+
+static int mlxsw_sp2_resources_kvd_register(struct mlxsw_core *mlxsw_core)
+{
+	struct devlink *devlink = priv_to_devlink(mlxsw_core);
+	struct devlink_resource_size_params kvd_size_params;
+	u32 kvd_size;
+
+	if (!MLXSW_CORE_RES_VALID(mlxsw_core, KVD_SIZE))
+		return -EIO;
+
+	kvd_size = MLXSW_CORE_RES_GET(mlxsw_core, KVD_SIZE);
+	devlink_resource_size_params_init(&kvd_size_params, kvd_size, kvd_size,
+					  MLXSW_SP_KVD_GRANULARITY,
+					  DEVLINK_RESOURCE_UNIT_ENTRY);
+
+	return devl_resource_register(devlink, MLXSW_SP_RESOURCE_NAME_KVD,
+				      kvd_size, MLXSW_SP_RESOURCE_KVD,
+				      DEVLINK_RESOURCE_ID_PARENT_TOP,
+				      &kvd_size_params);
+}
+
+static int mlxsw_sp_resources_span_register(struct mlxsw_core *mlxsw_core)
+{
+	struct devlink *devlink = priv_to_devlink(mlxsw_core);
+	struct devlink_resource_size_params span_size_params;
+	u32 max_span;
+
+	if (!MLXSW_CORE_RES_VALID(mlxsw_core, MAX_SPAN))
+		return -EIO;
+
+	max_span = MLXSW_CORE_RES_GET(mlxsw_core, MAX_SPAN);
+	devlink_resource_size_params_init(&span_size_params, max_span, max_span,
+					  1, DEVLINK_RESOURCE_UNIT_ENTRY);
+
+	return devl_resource_register(devlink, MLXSW_SP_RESOURCE_NAME_SPAN,
+				      max_span, MLXSW_SP_RESOURCE_SPAN,
+				      DEVLINK_RESOURCE_ID_PARENT_TOP,
+				      &span_size_params);
+}
+
+static int
+mlxsw_sp_resources_rif_mac_profile_register(struct mlxsw_core *mlxsw_core)
+{
+	struct devlink *devlink = priv_to_devlink(mlxsw_core);
+	struct devlink_resource_size_params size_params;
+	u8 max_rif_mac_profiles;
+
+	if (!MLXSW_CORE_RES_VALID(mlxsw_core, MAX_RIF_MAC_PROFILES))
+		max_rif_mac_profiles = 1;
+	else
+		max_rif_mac_profiles = MLXSW_CORE_RES_GET(mlxsw_core,
+							  MAX_RIF_MAC_PROFILES);
+	devlink_resource_size_params_init(&size_params, max_rif_mac_profiles,
+					  max_rif_mac_profiles, 1,
+					  DEVLINK_RESOURCE_UNIT_ENTRY);
+
+	return devl_resource_register(devlink,
+				      "rif_mac_profiles",
+				      max_rif_mac_profiles,
+				      MLXSW_SP_RESOURCE_RIF_MAC_PROFILES,
+				      DEVLINK_RESOURCE_ID_PARENT_TOP,
+				      &size_params);
+}
+
+static int mlxsw_sp_resources_rifs_register(struct mlxsw_core *mlxsw_core)
+{
+	struct devlink *devlink = priv_to_devlink(mlxsw_core);
+	struct devlink_resource_size_params size_params;
+	u64 max_rifs;
+
+	if (!MLXSW_CORE_RES_VALID(mlxsw_core, MAX_RIFS))
+		return -EIO;
+
+	max_rifs = MLXSW_CORE_RES_GET(mlxsw_core, MAX_RIFS);
+	devlink_resource_size_params_init(&size_params, max_rifs, max_rifs,
+					  1, DEVLINK_RESOURCE_UNIT_ENTRY);
+
+	return devl_resource_register(devlink, "rifs", max_rifs,
+				      MLXSW_SP_RESOURCE_RIFS,
+				      DEVLINK_RESOURCE_ID_PARENT_TOP,
+				      &size_params);
+}
+
+static int mlxsw_sp1_resources_register(struct mlxsw_core *mlxsw_core)
+{
+	int err;
+
+	err = mlxsw_sp1_resources_kvd_register(mlxsw_core);
+	if (err)
+		return err;
+
+	err = mlxsw_sp_resources_span_register(mlxsw_core);
+	if (err)
+		goto err_resources_span_register;
+
+	err = mlxsw_sp_counter_resources_register(mlxsw_core);
+	if (err)
+		goto err_resources_counter_register;
+
+	err = mlxsw_sp_policer_resources_register(mlxsw_core);
+	if (err)
+		goto err_policer_resources_register;
+
+	err = mlxsw_sp_resources_rif_mac_profile_register(mlxsw_core);
+	if (err)
+		goto err_resources_rif_mac_profile_register;
+
+	err = mlxsw_sp_resources_rifs_register(mlxsw_core);
+	if (err)
+		goto err_resources_rifs_register;
+
+	return 0;
+
+err_resources_rifs_register:
+err_resources_rif_mac_profile_register:
+err_policer_resources_register:
+err_resources_counter_register:
+err_resources_span_register:
+	devl_resources_unregister(priv_to_devlink(mlxsw_core));
+	return err;
+}
+
+static int mlxsw_sp2_resources_register(struct mlxsw_core *mlxsw_core)
+{
+	int err;
+
+	err = mlxsw_sp2_resources_kvd_register(mlxsw_core);
+	if (err)
+		return err;
+
+	err = mlxsw_sp_resources_span_register(mlxsw_core);
+	if (err)
+		goto err_resources_span_register;
+
+	err = mlxsw_sp_counter_resources_register(mlxsw_core);
+	if (err)
+		goto err_resources_counter_register;
+
+	err = mlxsw_sp_policer_resources_register(mlxsw_core);
+	if (err)
+		goto err_policer_resources_register;
+
+	err = mlxsw_sp_resources_rif_mac_profile_register(mlxsw_core);
+	if (err)
+		goto err_resources_rif_mac_profile_register;
+
+	err = mlxsw_sp_resources_rifs_register(mlxsw_core);
+	if (err)
+		goto err_resources_rifs_register;
+
+	return 0;
+
+err_resources_rifs_register:
+err_resources_rif_mac_profile_register:
+err_policer_resources_register:
+err_resources_counter_register:
+err_resources_span_register:
+	devl_resources_unregister(priv_to_devlink(mlxsw_core));
+	return err;
+}
+
+static int mlxsw_sp_kvd_sizes_get(struct mlxsw_core *mlxsw_core,
+				  const struct mlxsw_config_profile *profile,
+				  u64 *p_single_size, u64 *p_double_size,
+				  u64 *p_linear_size)
+{
+	struct devlink *devlink = priv_to_devlink(mlxsw_core);
+	u32 double_size;
+	int err;
+
+	if (!MLXSW_CORE_RES_VALID(mlxsw_core, KVD_SINGLE_MIN_SIZE) ||
+	    !MLXSW_CORE_RES_VALID(mlxsw_core, KVD_DOUBLE_MIN_SIZE))
+		return -EIO;
+
+	/* The hash part is what left of the kvd without the
+	 * linear part. It is split to the single size and
+	 * double size by the parts ratio from the profile.
+	 * Both sizes must be a multiplications of the
+	 * granularity from the profile. In case the user
+	 * provided the sizes they are obtained via devlink.
+	 */
+	err = devl_resource_size_get(devlink,
+				     MLXSW_SP_RESOURCE_KVD_LINEAR,
+				     p_linear_size);
+	if (err)
+		*p_linear_size = profile->kvd_linear_size;
+
+	err = devl_resource_size_get(devlink,
+				     MLXSW_SP_RESOURCE_KVD_HASH_DOUBLE,
+				     p_double_size);
+	if (err) {
+		double_size = MLXSW_CORE_RES_GET(mlxsw_core, KVD_SIZE) -
+			      *p_linear_size;
+		double_size *= profile->kvd_hash_double_parts;
+		double_size /= profile->kvd_hash_double_parts +
+			       profile->kvd_hash_single_parts;
+		*p_double_size = rounddown(double_size,
+					   MLXSW_SP_KVD_GRANULARITY);
+	}
+
+	err = devl_resource_size_get(devlink,
+				     MLXSW_SP_RESOURCE_KVD_HASH_SINGLE,
+				     p_single_size);
+	if (err)
+		*p_single_size = MLXSW_CORE_RES_GET(mlxsw_core, KVD_SIZE) -
+				 *p_double_size - *p_linear_size;
+
+	/* Check results are legal. */
+	if (*p_single_size < MLXSW_CORE_RES_GET(mlxsw_core, KVD_SINGLE_MIN_SIZE) ||
+	    *p_double_size < MLXSW_CORE_RES_GET(mlxsw_core, KVD_DOUBLE_MIN_SIZE) ||
+	    MLXSW_CORE_RES_GET(mlxsw_core, KVD_SIZE) < *p_linear_size)
+		return -EIO;
+
+	return 0;
+}
+
+static int
+mlxsw_sp_params_acl_region_rehash_intrvl_get(struct devlink *devlink, u32 id,
+					     struct devlink_param_gset_ctx *ctx)
+{
+	struct mlxsw_core *mlxsw_core = devlink_priv(devlink);
+	struct mlxsw_sp *mlxsw_sp = mlxsw_core_driver_priv(mlxsw_core);
+
+	ctx->val.vu32 = mlxsw_sp_acl_region_rehash_intrvl_get(mlxsw_sp);
+	return 0;
+}
+
+static int
+mlxsw_sp_params_acl_region_rehash_intrvl_set(struct devlink *devlink, u32 id,
+					     struct devlink_param_gset_ctx *ctx)
+{
+	struct mlxsw_core *mlxsw_core = devlink_priv(devlink);
+	struct mlxsw_sp *mlxsw_sp = mlxsw_core_driver_priv(mlxsw_core);
+
+	return mlxsw_sp_acl_region_rehash_intrvl_set(mlxsw_sp, ctx->val.vu32);
+}
+
+static const struct devlink_param mlxsw_sp2_devlink_params[] = {
+	DEVLINK_PARAM_DRIVER(MLXSW_DEVLINK_PARAM_ID_ACL_REGION_REHASH_INTERVAL,
+			     "acl_region_rehash_interval",
+			     DEVLINK_PARAM_TYPE_U32,
+			     BIT(DEVLINK_PARAM_CMODE_RUNTIME),
+			     mlxsw_sp_params_acl_region_rehash_intrvl_get,
+			     mlxsw_sp_params_acl_region_rehash_intrvl_set,
+			     NULL),
+};
+
+static int mlxsw_sp2_params_register(struct mlxsw_core *mlxsw_core)
+{
+	struct devlink *devlink = priv_to_devlink(mlxsw_core);
+	union devlink_param_value value;
+	int err;
+
+	err = devlink_params_register(devlink, mlxsw_sp2_devlink_params,
+				      ARRAY_SIZE(mlxsw_sp2_devlink_params));
+	if (err)
+		return err;
+
+	value.vu32 = 0;
+	devlink_param_driverinit_value_set(devlink,
+					   MLXSW_DEVLINK_PARAM_ID_ACL_REGION_REHASH_INTERVAL,
+					   value);
+	return 0;
+}
+
+static void mlxsw_sp2_params_unregister(struct mlxsw_core *mlxsw_core)
+{
+	devlink_params_unregister(priv_to_devlink(mlxsw_core),
+				  mlxsw_sp2_devlink_params,
+				  ARRAY_SIZE(mlxsw_sp2_devlink_params));
+}
+
+static void mlxsw_sp_ptp_transmitted(struct mlxsw_core *mlxsw_core,
+				     struct sk_buff *skb, u16 local_port)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_core_driver_priv(mlxsw_core);
+
+	skb_pull(skb, MLXSW_TXHDR_LEN);
+	mlxsw_sp->ptp_ops->transmitted(mlxsw_sp, skb, local_port);
+}
+
+static struct mlxsw_driver mlxsw_sp1_driver = {
+	.kind				= mlxsw_sp1_driver_name,
+	.priv_size			= sizeof(struct mlxsw_sp),
+	.fw_req_rev			= &mlxsw_sp1_fw_rev,
+	.fw_filename			= MLXSW_SP1_FW_FILENAME,
+	.init				= mlxsw_sp1_init,
+	.fini				= mlxsw_sp_fini,
+	.port_split			= mlxsw_sp_port_split,
+	.port_unsplit			= mlxsw_sp_port_unsplit,
+	.sb_pool_get			= mlxsw_sp_sb_pool_get,
+	.sb_pool_set			= mlxsw_sp_sb_pool_set,
+	.sb_port_pool_get		= mlxsw_sp_sb_port_pool_get,
+	.sb_port_pool_set		= mlxsw_sp_sb_port_pool_set,
+	.sb_tc_pool_bind_get		= mlxsw_sp_sb_tc_pool_bind_get,
+	.sb_tc_pool_bind_set		= mlxsw_sp_sb_tc_pool_bind_set,
+	.sb_occ_snapshot		= mlxsw_sp_sb_occ_snapshot,
+	.sb_occ_max_clear		= mlxsw_sp_sb_occ_max_clear,
+	.sb_occ_port_pool_get		= mlxsw_sp_sb_occ_port_pool_get,
+	.sb_occ_tc_port_bind_get	= mlxsw_sp_sb_occ_tc_port_bind_get,
+	.trap_init			= mlxsw_sp_trap_init,
+	.trap_fini			= mlxsw_sp_trap_fini,
+	.trap_action_set		= mlxsw_sp_trap_action_set,
+	.trap_group_init		= mlxsw_sp_trap_group_init,
+	.trap_group_set			= mlxsw_sp_trap_group_set,
+	.trap_policer_init		= mlxsw_sp_trap_policer_init,
+	.trap_policer_fini		= mlxsw_sp_trap_policer_fini,
+	.trap_policer_set		= mlxsw_sp_trap_policer_set,
+	.trap_policer_counter_get	= mlxsw_sp_trap_policer_counter_get,
+	.txhdr_construct		= mlxsw_sp_txhdr_construct,
+	.resources_register		= mlxsw_sp1_resources_register,
+	.kvd_sizes_get			= mlxsw_sp_kvd_sizes_get,
+	.ptp_transmitted		= mlxsw_sp_ptp_transmitted,
+	.txhdr_len			= MLXSW_TXHDR_LEN,
+	.profile			= &mlxsw_sp1_config_profile,
+	.sdq_supports_cqe_v2		= false,
+};
+
+static struct mlxsw_driver mlxsw_sp2_driver = {
+	.kind				= mlxsw_sp2_driver_name,
+	.priv_size			= sizeof(struct mlxsw_sp),
+	.fw_req_rev			= &mlxsw_sp2_fw_rev,
+	.fw_filename			= MLXSW_SP2_FW_FILENAME,
+	.init				= mlxsw_sp2_init,
+	.fini				= mlxsw_sp_fini,
+	.port_split			= mlxsw_sp_port_split,
+	.port_unsplit			= mlxsw_sp_port_unsplit,
+	.ports_remove_selected		= mlxsw_sp_ports_remove_selected,
+	.sb_pool_get			= mlxsw_sp_sb_pool_get,
+	.sb_pool_set			= mlxsw_sp_sb_pool_set,
+	.sb_port_pool_get		= mlxsw_sp_sb_port_pool_get,
+	.sb_port_pool_set		= mlxsw_sp_sb_port_pool_set,
+	.sb_tc_pool_bind_get		= mlxsw_sp_sb_tc_pool_bind_get,
+	.sb_tc_pool_bind_set		= mlxsw_sp_sb_tc_pool_bind_set,
+	.sb_occ_snapshot		= mlxsw_sp_sb_occ_snapshot,
+	.sb_occ_max_clear		= mlxsw_sp_sb_occ_max_clear,
+	.sb_occ_port_pool_get		= mlxsw_sp_sb_occ_port_pool_get,
+	.sb_occ_tc_port_bind_get	= mlxsw_sp_sb_occ_tc_port_bind_get,
+	.trap_init			= mlxsw_sp_trap_init,
+	.trap_fini			= mlxsw_sp_trap_fini,
+	.trap_action_set		= mlxsw_sp_trap_action_set,
+	.trap_group_init		= mlxsw_sp_trap_group_init,
+	.trap_group_set			= mlxsw_sp_trap_group_set,
+	.trap_policer_init		= mlxsw_sp_trap_policer_init,
+	.trap_policer_fini		= mlxsw_sp_trap_policer_fini,
+	.trap_policer_set		= mlxsw_sp_trap_policer_set,
+	.trap_policer_counter_get	= mlxsw_sp_trap_policer_counter_get,
+	.txhdr_construct		= mlxsw_sp_txhdr_construct,
+	.resources_register		= mlxsw_sp2_resources_register,
+	.params_register		= mlxsw_sp2_params_register,
+	.params_unregister		= mlxsw_sp2_params_unregister,
+	.ptp_transmitted		= mlxsw_sp_ptp_transmitted,
+	.txhdr_len			= MLXSW_TXHDR_LEN,
+	.profile			= &mlxsw_sp2_config_profile,
+	.sdq_supports_cqe_v2		= true,
+};
+
+static struct mlxsw_driver mlxsw_sp3_driver = {
+	.kind				= mlxsw_sp3_driver_name,
+	.priv_size			= sizeof(struct mlxsw_sp),
+	.fw_req_rev			= &mlxsw_sp3_fw_rev,
+	.fw_filename			= MLXSW_SP3_FW_FILENAME,
+	.init				= mlxsw_sp3_init,
+	.fini				= mlxsw_sp_fini,
+	.port_split			= mlxsw_sp_port_split,
+	.port_unsplit			= mlxsw_sp_port_unsplit,
+	.ports_remove_selected		= mlxsw_sp_ports_remove_selected,
+	.sb_pool_get			= mlxsw_sp_sb_pool_get,
+	.sb_pool_set			= mlxsw_sp_sb_pool_set,
+	.sb_port_pool_get		= mlxsw_sp_sb_port_pool_get,
+	.sb_port_pool_set		= mlxsw_sp_sb_port_pool_set,
+	.sb_tc_pool_bind_get		= mlxsw_sp_sb_tc_pool_bind_get,
+	.sb_tc_pool_bind_set		= mlxsw_sp_sb_tc_pool_bind_set,
+	.sb_occ_snapshot		= mlxsw_sp_sb_occ_snapshot,
+	.sb_occ_max_clear		= mlxsw_sp_sb_occ_max_clear,
+	.sb_occ_port_pool_get		= mlxsw_sp_sb_occ_port_pool_get,
+	.sb_occ_tc_port_bind_get	= mlxsw_sp_sb_occ_tc_port_bind_get,
+	.trap_init			= mlxsw_sp_trap_init,
+	.trap_fini			= mlxsw_sp_trap_fini,
+	.trap_action_set		= mlxsw_sp_trap_action_set,
+	.trap_group_init		= mlxsw_sp_trap_group_init,
+	.trap_group_set			= mlxsw_sp_trap_group_set,
+	.trap_policer_init		= mlxsw_sp_trap_policer_init,
+	.trap_policer_fini		= mlxsw_sp_trap_policer_fini,
+	.trap_policer_set		= mlxsw_sp_trap_policer_set,
+	.trap_policer_counter_get	= mlxsw_sp_trap_policer_counter_get,
+	.txhdr_construct		= mlxsw_sp_txhdr_construct,
+	.resources_register		= mlxsw_sp2_resources_register,
+	.params_register		= mlxsw_sp2_params_register,
+	.params_unregister		= mlxsw_sp2_params_unregister,
+	.ptp_transmitted		= mlxsw_sp_ptp_transmitted,
+	.txhdr_len			= MLXSW_TXHDR_LEN,
+	.profile			= &mlxsw_sp2_config_profile,
+	.sdq_supports_cqe_v2		= true,
+};
+
+static struct mlxsw_driver mlxsw_sp4_driver = {
+	.kind				= mlxsw_sp4_driver_name,
+	.priv_size			= sizeof(struct mlxsw_sp),
+	.init				= mlxsw_sp4_init,
+	.fini				= mlxsw_sp_fini,
+	.port_split			= mlxsw_sp_port_split,
+	.port_unsplit			= mlxsw_sp_port_unsplit,
+	.ports_remove_selected		= mlxsw_sp_ports_remove_selected,
+	.sb_pool_get			= mlxsw_sp_sb_pool_get,
+	.sb_pool_set			= mlxsw_sp_sb_pool_set,
+	.sb_port_pool_get		= mlxsw_sp_sb_port_pool_get,
+	.sb_port_pool_set		= mlxsw_sp_sb_port_pool_set,
+	.sb_tc_pool_bind_get		= mlxsw_sp_sb_tc_pool_bind_get,
+	.sb_tc_pool_bind_set		= mlxsw_sp_sb_tc_pool_bind_set,
+	.sb_occ_snapshot		= mlxsw_sp_sb_occ_snapshot,
+	.sb_occ_max_clear		= mlxsw_sp_sb_occ_max_clear,
+	.sb_occ_port_pool_get		= mlxsw_sp_sb_occ_port_pool_get,
+	.sb_occ_tc_port_bind_get	= mlxsw_sp_sb_occ_tc_port_bind_get,
+	.trap_init			= mlxsw_sp_trap_init,
+	.trap_fini			= mlxsw_sp_trap_fini,
+	.trap_action_set		= mlxsw_sp_trap_action_set,
+	.trap_group_init		= mlxsw_sp_trap_group_init,
+	.trap_group_set			= mlxsw_sp_trap_group_set,
+	.trap_policer_init		= mlxsw_sp_trap_policer_init,
+	.trap_policer_fini		= mlxsw_sp_trap_policer_fini,
+	.trap_policer_set		= mlxsw_sp_trap_policer_set,
+	.trap_policer_counter_get	= mlxsw_sp_trap_policer_counter_get,
+	.txhdr_construct		= mlxsw_sp_txhdr_construct,
+	.resources_register		= mlxsw_sp2_resources_register,
+	.params_register		= mlxsw_sp2_params_register,
+	.params_unregister		= mlxsw_sp2_params_unregister,
+	.ptp_transmitted		= mlxsw_sp_ptp_transmitted,
+	.txhdr_len			= MLXSW_TXHDR_LEN,
+	.profile			= &mlxsw_sp4_config_profile,
+	.sdq_supports_cqe_v2		= true,
+};
+
+bool mlxsw_sp_port_dev_check(const struct net_device *dev)
+{
+	return dev->netdev_ops == &mlxsw_sp_port_netdev_ops;
+}
+
+static int mlxsw_sp_lower_dev_walk(struct net_device *lower_dev,
+				   struct netdev_nested_priv *priv)
+{
+	int ret = 0;
+
+	if (mlxsw_sp_port_dev_check(lower_dev)) {
+		priv->data = (void *)netdev_priv(lower_dev);
+		ret = 1;
+	}
+
+	return ret;
+}
+
+struct mlxsw_sp_port *mlxsw_sp_port_dev_lower_find(struct net_device *dev)
+{
+	struct netdev_nested_priv priv = {
+		.data = NULL,
+	};
+
+	if (mlxsw_sp_port_dev_check(dev))
+		return netdev_priv(dev);
+
+	netdev_walk_all_lower_dev(dev, mlxsw_sp_lower_dev_walk, &priv);
+
+	return (struct mlxsw_sp_port *)priv.data;
+}
+
+struct mlxsw_sp *mlxsw_sp_lower_get(struct net_device *dev)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port;
+
+	mlxsw_sp_port = mlxsw_sp_port_dev_lower_find(dev);
+	return mlxsw_sp_port ? mlxsw_sp_port->mlxsw_sp : NULL;
+}
+
+struct mlxsw_sp_port *mlxsw_sp_port_dev_lower_find_rcu(struct net_device *dev)
+{
+	struct netdev_nested_priv priv = {
+		.data = NULL,
+	};
+
+	if (mlxsw_sp_port_dev_check(dev))
+		return netdev_priv(dev);
+
+	netdev_walk_all_lower_dev_rcu(dev, mlxsw_sp_lower_dev_walk,
+				      &priv);
+
+	return (struct mlxsw_sp_port *)priv.data;
+}
+
+struct mlxsw_sp_port *mlxsw_sp_port_lower_dev_hold(struct net_device *dev)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port;
+
+	rcu_read_lock();
+	mlxsw_sp_port = mlxsw_sp_port_dev_lower_find_rcu(dev);
+	if (mlxsw_sp_port)
+		dev_hold(mlxsw_sp_port->dev);
+	rcu_read_unlock();
+	return mlxsw_sp_port;
+}
+
+void mlxsw_sp_port_dev_put(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	dev_put(mlxsw_sp_port->dev);
+}
+
+int mlxsw_sp_parsing_depth_inc(struct mlxsw_sp *mlxsw_sp)
+{
+	char mprs_pl[MLXSW_REG_MPRS_LEN];
+	int err = 0;
+
+	mutex_lock(&mlxsw_sp->parsing.lock);
+
+	if (refcount_inc_not_zero(&mlxsw_sp->parsing.parsing_depth_ref))
+		goto out_unlock;
+
+	mlxsw_reg_mprs_pack(mprs_pl, MLXSW_SP_INCREASED_PARSING_DEPTH,
+			    mlxsw_sp->parsing.vxlan_udp_dport);
+	err = mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(mprs), mprs_pl);
+	if (err)
+		goto out_unlock;
+
+	mlxsw_sp->parsing.parsing_depth = MLXSW_SP_INCREASED_PARSING_DEPTH;
+	refcount_set(&mlxsw_sp->parsing.parsing_depth_ref, 1);
+
+out_unlock:
+	mutex_unlock(&mlxsw_sp->parsing.lock);
+	return err;
+}
+
+void mlxsw_sp_parsing_depth_dec(struct mlxsw_sp *mlxsw_sp)
+{
+	char mprs_pl[MLXSW_REG_MPRS_LEN];
+
+	mutex_lock(&mlxsw_sp->parsing.lock);
+
+	if (!refcount_dec_and_test(&mlxsw_sp->parsing.parsing_depth_ref))
+		goto out_unlock;
+
+	mlxsw_reg_mprs_pack(mprs_pl, MLXSW_SP_DEFAULT_PARSING_DEPTH,
+			    mlxsw_sp->parsing.vxlan_udp_dport);
+	mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(mprs), mprs_pl);
+	mlxsw_sp->parsing.parsing_depth = MLXSW_SP_DEFAULT_PARSING_DEPTH;
+
+out_unlock:
+	mutex_unlock(&mlxsw_sp->parsing.lock);
+}
+
+int mlxsw_sp_parsing_vxlan_udp_dport_set(struct mlxsw_sp *mlxsw_sp,
+					 __be16 udp_dport)
+{
+	char mprs_pl[MLXSW_REG_MPRS_LEN];
+	int err;
+
+	mutex_lock(&mlxsw_sp->parsing.lock);
+
+	mlxsw_reg_mprs_pack(mprs_pl, mlxsw_sp->parsing.parsing_depth,
+			    be16_to_cpu(udp_dport));
+	err = mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(mprs), mprs_pl);
+	if (err)
+		goto out_unlock;
+
+	mlxsw_sp->parsing.vxlan_udp_dport = be16_to_cpu(udp_dport);
+
+out_unlock:
+	mutex_unlock(&mlxsw_sp->parsing.lock);
+	return err;
+}
+
+static void
+mlxsw_sp_port_lag_uppers_cleanup(struct mlxsw_sp_port *mlxsw_sp_port,
+				 struct net_device *lag_dev)
+{
+	struct net_device *br_dev = netdev_master_upper_dev_get(lag_dev);
+	struct net_device *upper_dev;
+	struct list_head *iter;
+
+	if (netif_is_bridge_port(lag_dev))
+		mlxsw_sp_port_bridge_leave(mlxsw_sp_port, lag_dev, br_dev);
+
+	netdev_for_each_upper_dev_rcu(lag_dev, upper_dev, iter) {
+		if (!netif_is_bridge_port(upper_dev))
+			continue;
+		br_dev = netdev_master_upper_dev_get(upper_dev);
+		mlxsw_sp_port_bridge_leave(mlxsw_sp_port, upper_dev, br_dev);
+	}
+}
+
+static int mlxsw_sp_lag_create(struct mlxsw_sp *mlxsw_sp, u16 lag_id)
+{
+	char sldr_pl[MLXSW_REG_SLDR_LEN];
+
+	mlxsw_reg_sldr_lag_create_pack(sldr_pl, lag_id);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(sldr), sldr_pl);
+}
+
+static int mlxsw_sp_lag_destroy(struct mlxsw_sp *mlxsw_sp, u16 lag_id)
+{
+	char sldr_pl[MLXSW_REG_SLDR_LEN];
+
+	mlxsw_reg_sldr_lag_destroy_pack(sldr_pl, lag_id);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(sldr), sldr_pl);
+}
+
+static int mlxsw_sp_lag_col_port_add(struct mlxsw_sp_port *mlxsw_sp_port,
+				     u16 lag_id, u8 port_index)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	char slcor_pl[MLXSW_REG_SLCOR_LEN];
+
+	mlxsw_reg_slcor_port_add_pack(slcor_pl, mlxsw_sp_port->local_port,
+				      lag_id, port_index);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(slcor), slcor_pl);
+}
+
+static int mlxsw_sp_lag_col_port_remove(struct mlxsw_sp_port *mlxsw_sp_port,
+					u16 lag_id)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	char slcor_pl[MLXSW_REG_SLCOR_LEN];
+
+	mlxsw_reg_slcor_port_remove_pack(slcor_pl, mlxsw_sp_port->local_port,
+					 lag_id);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(slcor), slcor_pl);
+}
+
+static int mlxsw_sp_lag_col_port_enable(struct mlxsw_sp_port *mlxsw_sp_port,
+					u16 lag_id)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	char slcor_pl[MLXSW_REG_SLCOR_LEN];
+
+	mlxsw_reg_slcor_col_enable_pack(slcor_pl, mlxsw_sp_port->local_port,
+					lag_id);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(slcor), slcor_pl);
+}
+
+static int mlxsw_sp_lag_col_port_disable(struct mlxsw_sp_port *mlxsw_sp_port,
+					 u16 lag_id)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	char slcor_pl[MLXSW_REG_SLCOR_LEN];
+
+	mlxsw_reg_slcor_col_disable_pack(slcor_pl, mlxsw_sp_port->local_port,
+					 lag_id);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(slcor), slcor_pl);
+}
+
+static int mlxsw_sp_lag_index_get(struct mlxsw_sp *mlxsw_sp,
+				  struct net_device *lag_dev,
+				  u16 *p_lag_id)
+{
+	struct mlxsw_sp_upper *lag;
+	int free_lag_id = -1;
+	u16 max_lag;
+	int err, i;
+
+	err = mlxsw_core_max_lag(mlxsw_sp->core, &max_lag);
+	if (err)
+		return err;
+
+	for (i = 0; i < max_lag; i++) {
+		lag = mlxsw_sp_lag_get(mlxsw_sp, i);
+		if (lag->ref_count) {
+			if (lag->dev == lag_dev) {
+				*p_lag_id = i;
+				return 0;
+			}
+		} else if (free_lag_id < 0) {
+			free_lag_id = i;
+		}
+	}
+	if (free_lag_id < 0)
+		return -EBUSY;
+	*p_lag_id = free_lag_id;
+	return 0;
+}
+
+static bool
+mlxsw_sp_master_lag_check(struct mlxsw_sp *mlxsw_sp,
+			  struct net_device *lag_dev,
+			  struct netdev_lag_upper_info *lag_upper_info,
+			  struct netlink_ext_ack *extack)
+{
+	u16 lag_id;
+
+	if (mlxsw_sp_lag_index_get(mlxsw_sp, lag_dev, &lag_id) != 0) {
+		NL_SET_ERR_MSG_MOD(extack, "Exceeded number of supported LAG devices");
+		return false;
+	}
+	if (lag_upper_info->tx_type != NETDEV_LAG_TX_TYPE_HASH) {
+		NL_SET_ERR_MSG_MOD(extack, "LAG device using unsupported Tx type");
+		return false;
+	}
+	return true;
+}
+
+static int mlxsw_sp_port_lag_index_get(struct mlxsw_sp *mlxsw_sp,
+				       u16 lag_id, u8 *p_port_index)
+{
+	u64 max_lag_members;
+	int i;
+
+	max_lag_members = MLXSW_CORE_RES_GET(mlxsw_sp->core,
+					     MAX_LAG_MEMBERS);
+	for (i = 0; i < max_lag_members; i++) {
+		if (!mlxsw_sp_port_lagged_get(mlxsw_sp, lag_id, i)) {
+			*p_port_index = i;
+			return 0;
+		}
+	}
+	return -EBUSY;
+}
+
+static int mlxsw_sp_port_lag_join(struct mlxsw_sp_port *mlxsw_sp_port,
+				  struct net_device *lag_dev,
+				  struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	struct mlxsw_sp_upper *lag;
+	u16 lag_id;
+	u8 port_index;
+	int err;
+
+	err = mlxsw_sp_lag_index_get(mlxsw_sp, lag_dev, &lag_id);
+	if (err)
+		return err;
+	lag = mlxsw_sp_lag_get(mlxsw_sp, lag_id);
+	if (!lag->ref_count) {
+		err = mlxsw_sp_lag_create(mlxsw_sp, lag_id);
+		if (err)
+			return err;
+		lag->dev = lag_dev;
+	}
+
+	err = mlxsw_sp_port_lag_index_get(mlxsw_sp, lag_id, &port_index);
+	if (err)
+		return err;
+	err = mlxsw_sp_lag_col_port_add(mlxsw_sp_port, lag_id, port_index);
+	if (err)
+		goto err_col_port_add;
+
+	mlxsw_core_lag_mapping_set(mlxsw_sp->core, lag_id, port_index,
+				   mlxsw_sp_port->local_port);
+	mlxsw_sp_port->lag_id = lag_id;
+	mlxsw_sp_port->lagged = 1;
+	lag->ref_count++;
+
+	/* Port is no longer usable as a router interface */
+	if (mlxsw_sp_port->default_vlan->fid)
+		mlxsw_sp_port_vlan_router_leave(mlxsw_sp_port->default_vlan);
+
+	/* Join a router interface configured on the LAG, if exists */
+	err = mlxsw_sp_port_vlan_router_join(mlxsw_sp_port->default_vlan,
+					     lag_dev, extack);
+	if (err)
+		goto err_router_join;
+
+	return 0;
+
+err_router_join:
+	lag->ref_count--;
+	mlxsw_sp_port->lagged = 0;
+	mlxsw_core_lag_mapping_clear(mlxsw_sp->core, lag_id,
+				     mlxsw_sp_port->local_port);
+	mlxsw_sp_lag_col_port_remove(mlxsw_sp_port, lag_id);
+err_col_port_add:
+	if (!lag->ref_count)
+		mlxsw_sp_lag_destroy(mlxsw_sp, lag_id);
+	return err;
+}
+
+static void mlxsw_sp_port_lag_leave(struct mlxsw_sp_port *mlxsw_sp_port,
+				    struct net_device *lag_dev)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	u16 lag_id = mlxsw_sp_port->lag_id;
+	struct mlxsw_sp_upper *lag;
+
+	if (!mlxsw_sp_port->lagged)
+		return;
+	lag = mlxsw_sp_lag_get(mlxsw_sp, lag_id);
+	WARN_ON(lag->ref_count == 0);
+
+	mlxsw_sp_lag_col_port_remove(mlxsw_sp_port, lag_id);
+
+	/* Any VLANs configured on the port are no longer valid */
+	mlxsw_sp_port_vlan_flush(mlxsw_sp_port, false);
+	mlxsw_sp_port_vlan_cleanup(mlxsw_sp_port->default_vlan);
+	/* Make the LAG and its directly linked uppers leave bridges they
+	 * are memeber in
+	 */
+	mlxsw_sp_port_lag_uppers_cleanup(mlxsw_sp_port, lag_dev);
+
+	if (lag->ref_count == 1)
+		mlxsw_sp_lag_destroy(mlxsw_sp, lag_id);
+
+	mlxsw_core_lag_mapping_clear(mlxsw_sp->core, lag_id,
+				     mlxsw_sp_port->local_port);
+	mlxsw_sp_port->lagged = 0;
+	lag->ref_count--;
+
+	/* Make sure untagged frames are allowed to ingress */
+	mlxsw_sp_port_pvid_set(mlxsw_sp_port, MLXSW_SP_DEFAULT_VID,
+			       ETH_P_8021Q);
+}
+
+static int mlxsw_sp_lag_dist_port_add(struct mlxsw_sp_port *mlxsw_sp_port,
+				      u16 lag_id)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	char sldr_pl[MLXSW_REG_SLDR_LEN];
+
+	mlxsw_reg_sldr_lag_add_port_pack(sldr_pl, lag_id,
+					 mlxsw_sp_port->local_port);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(sldr), sldr_pl);
+}
+
+static int mlxsw_sp_lag_dist_port_remove(struct mlxsw_sp_port *mlxsw_sp_port,
+					 u16 lag_id)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	char sldr_pl[MLXSW_REG_SLDR_LEN];
+
+	mlxsw_reg_sldr_lag_remove_port_pack(sldr_pl, lag_id,
+					    mlxsw_sp_port->local_port);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(sldr), sldr_pl);
+}
+
+static int
+mlxsw_sp_port_lag_col_dist_enable(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	int err;
+
+	err = mlxsw_sp_lag_col_port_enable(mlxsw_sp_port,
+					   mlxsw_sp_port->lag_id);
+	if (err)
+		return err;
+
+	err = mlxsw_sp_lag_dist_port_add(mlxsw_sp_port, mlxsw_sp_port->lag_id);
+	if (err)
+		goto err_dist_port_add;
+
+	return 0;
+
+err_dist_port_add:
+	mlxsw_sp_lag_col_port_disable(mlxsw_sp_port, mlxsw_sp_port->lag_id);
+	return err;
+}
+
+static int
+mlxsw_sp_port_lag_col_dist_disable(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	int err;
+
+	err = mlxsw_sp_lag_dist_port_remove(mlxsw_sp_port,
+					    mlxsw_sp_port->lag_id);
+	if (err)
+		return err;
+
+	err = mlxsw_sp_lag_col_port_disable(mlxsw_sp_port,
+					    mlxsw_sp_port->lag_id);
+	if (err)
+		goto err_col_port_disable;
+
+	return 0;
+
+err_col_port_disable:
+	mlxsw_sp_lag_dist_port_add(mlxsw_sp_port, mlxsw_sp_port->lag_id);
+	return err;
+}
+
+static int mlxsw_sp_port_lag_changed(struct mlxsw_sp_port *mlxsw_sp_port,
+				     struct netdev_lag_lower_state_info *info)
+{
+	if (info->tx_enabled)
+		return mlxsw_sp_port_lag_col_dist_enable(mlxsw_sp_port);
+	else
+		return mlxsw_sp_port_lag_col_dist_disable(mlxsw_sp_port);
+}
+
+static int mlxsw_sp_port_stp_set(struct mlxsw_sp_port *mlxsw_sp_port,
+				 bool enable)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	enum mlxsw_reg_spms_state spms_state;
+	char *spms_pl;
+	u16 vid;
+	int err;
+
+	spms_state = enable ? MLXSW_REG_SPMS_STATE_FORWARDING :
+			      MLXSW_REG_SPMS_STATE_DISCARDING;
+
+	spms_pl = kmalloc(MLXSW_REG_SPMS_LEN, GFP_KERNEL);
+	if (!spms_pl)
+		return -ENOMEM;
+	mlxsw_reg_spms_pack(spms_pl, mlxsw_sp_port->local_port);
+
+	for (vid = 0; vid < VLAN_N_VID; vid++)
+		mlxsw_reg_spms_vid_pack(spms_pl, vid, spms_state);
+
+	err = mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(spms), spms_pl);
+	kfree(spms_pl);
+	return err;
+}
+
+static int mlxsw_sp_port_ovs_join(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	u16 vid = 1;
+	int err;
+
+	err = mlxsw_sp_port_vp_mode_set(mlxsw_sp_port, true);
+	if (err)
+		return err;
+	err = mlxsw_sp_port_stp_set(mlxsw_sp_port, true);
+	if (err)
+		goto err_port_stp_set;
+	err = mlxsw_sp_port_vlan_set(mlxsw_sp_port, 1, VLAN_N_VID - 2,
+				     true, false);
+	if (err)
+		goto err_port_vlan_set;
+
+	for (; vid <= VLAN_N_VID - 1; vid++) {
+		err = mlxsw_sp_port_vid_learning_set(mlxsw_sp_port,
+						     vid, false);
+		if (err)
+			goto err_vid_learning_set;
+	}
+
+	return 0;
+
+err_vid_learning_set:
+	for (vid--; vid >= 1; vid--)
+		mlxsw_sp_port_vid_learning_set(mlxsw_sp_port, vid, true);
+err_port_vlan_set:
+	mlxsw_sp_port_stp_set(mlxsw_sp_port, false);
+err_port_stp_set:
+	mlxsw_sp_port_vp_mode_set(mlxsw_sp_port, false);
+	return err;
+}
+
+static void mlxsw_sp_port_ovs_leave(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	u16 vid;
+
+	for (vid = VLAN_N_VID - 1; vid >= 1; vid--)
+		mlxsw_sp_port_vid_learning_set(mlxsw_sp_port,
+					       vid, true);
+
+	mlxsw_sp_port_vlan_set(mlxsw_sp_port, 1, VLAN_N_VID - 2,
+			       false, false);
+	mlxsw_sp_port_stp_set(mlxsw_sp_port, false);
+	mlxsw_sp_port_vp_mode_set(mlxsw_sp_port, false);
+}
+
+static bool mlxsw_sp_bridge_has_multiple_vxlans(struct net_device *br_dev)
+{
+	unsigned int num_vxlans = 0;
+	struct net_device *dev;
+	struct list_head *iter;
+
+	netdev_for_each_lower_dev(br_dev, dev, iter) {
+		if (netif_is_vxlan(dev))
+			num_vxlans++;
+	}
+
+	return num_vxlans > 1;
+}
+
+static bool mlxsw_sp_bridge_vxlan_vlan_is_valid(struct net_device *br_dev)
+{
+	DECLARE_BITMAP(vlans, VLAN_N_VID) = {0};
+	struct net_device *dev;
+	struct list_head *iter;
+
+	netdev_for_each_lower_dev(br_dev, dev, iter) {
+		u16 pvid;
+		int err;
+
+		if (!netif_is_vxlan(dev))
+			continue;
+
+		err = mlxsw_sp_vxlan_mapped_vid(dev, &pvid);
+		if (err || !pvid)
+			continue;
+
+		if (test_and_set_bit(pvid, vlans))
+			return false;
+	}
+
+	return true;
+}
+
+static bool mlxsw_sp_bridge_vxlan_is_valid(struct net_device *br_dev,
+					   struct netlink_ext_ack *extack)
+{
+	if (br_multicast_enabled(br_dev)) {
+		NL_SET_ERR_MSG_MOD(extack, "Multicast can not be enabled on a bridge with a VxLAN device");
+		return false;
+	}
+
+	if (!br_vlan_enabled(br_dev) &&
+	    mlxsw_sp_bridge_has_multiple_vxlans(br_dev)) {
+		NL_SET_ERR_MSG_MOD(extack, "Multiple VxLAN devices are not supported in a VLAN-unaware bridge");
+		return false;
+	}
+
+	if (br_vlan_enabled(br_dev) &&
+	    !mlxsw_sp_bridge_vxlan_vlan_is_valid(br_dev)) {
+		NL_SET_ERR_MSG_MOD(extack, "Multiple VxLAN devices cannot have the same VLAN as PVID and egress untagged");
+		return false;
+	}
+
+	return true;
+}
+
+static int mlxsw_sp_netdevice_port_upper_event(struct net_device *lower_dev,
+					       struct net_device *dev,
+					       unsigned long event, void *ptr)
+{
+	struct netdev_notifier_changeupper_info *info;
+	struct mlxsw_sp_port *mlxsw_sp_port;
+	struct netlink_ext_ack *extack;
+	struct net_device *upper_dev;
+	struct mlxsw_sp *mlxsw_sp;
+	int err = 0;
+	u16 proto;
+
+	mlxsw_sp_port = netdev_priv(dev);
+	mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	info = ptr;
+	extack = netdev_notifier_info_to_extack(&info->info);
+
+	switch (event) {
+	case NETDEV_PRECHANGEUPPER:
+		upper_dev = info->upper_dev;
+		if (!is_vlan_dev(upper_dev) &&
+		    !netif_is_lag_master(upper_dev) &&
+		    !netif_is_bridge_master(upper_dev) &&
+		    !netif_is_ovs_master(upper_dev) &&
+		    !netif_is_macvlan(upper_dev) &&
+		    !netif_is_l3_master(upper_dev)) {
+			NL_SET_ERR_MSG_MOD(extack, "Unknown upper device type");
+			return -EINVAL;
+		}
+		if (!info->linking)
+			break;
+		if (netif_is_bridge_master(upper_dev) &&
+		    !mlxsw_sp_bridge_device_is_offloaded(mlxsw_sp, upper_dev) &&
+		    mlxsw_sp_bridge_has_vxlan(upper_dev) &&
+		    !mlxsw_sp_bridge_vxlan_is_valid(upper_dev, extack))
+			return -EOPNOTSUPP;
+		if (netdev_has_any_upper_dev(upper_dev) &&
+		    (!netif_is_bridge_master(upper_dev) ||
+		     !mlxsw_sp_bridge_device_is_offloaded(mlxsw_sp,
+							  upper_dev))) {
+			NL_SET_ERR_MSG_MOD(extack, "Enslaving a port to a device that already has an upper device is not supported");
+			return -EINVAL;
+		}
+		if (netif_is_lag_master(upper_dev) &&
+		    !mlxsw_sp_master_lag_check(mlxsw_sp, upper_dev,
+					       info->upper_info, extack))
+			return -EINVAL;
+		if (netif_is_lag_master(upper_dev) && vlan_uses_dev(dev)) {
+			NL_SET_ERR_MSG_MOD(extack, "Master device is a LAG master and this device has a VLAN");
+			return -EINVAL;
+		}
+		if (netif_is_lag_port(dev) && is_vlan_dev(upper_dev) &&
+		    !netif_is_lag_master(vlan_dev_real_dev(upper_dev))) {
+			NL_SET_ERR_MSG_MOD(extack, "Can not put a VLAN on a LAG port");
+			return -EINVAL;
+		}
+		if (netif_is_macvlan(upper_dev) &&
+		    !mlxsw_sp_rif_exists(mlxsw_sp, lower_dev)) {
+			NL_SET_ERR_MSG_MOD(extack, "macvlan is only supported on top of router interfaces");
+			return -EOPNOTSUPP;
+		}
+		if (netif_is_ovs_master(upper_dev) && vlan_uses_dev(dev)) {
+			NL_SET_ERR_MSG_MOD(extack, "Master device is an OVS master and this device has a VLAN");
+			return -EINVAL;
+		}
+		if (netif_is_ovs_port(dev) && is_vlan_dev(upper_dev)) {
+			NL_SET_ERR_MSG_MOD(extack, "Can not put a VLAN on an OVS port");
+			return -EINVAL;
+		}
+		if (netif_is_bridge_master(upper_dev)) {
+			br_vlan_get_proto(upper_dev, &proto);
+			if (br_vlan_enabled(upper_dev) &&
+			    proto != ETH_P_8021Q && proto != ETH_P_8021AD) {
+				NL_SET_ERR_MSG_MOD(extack, "Enslaving a port to a bridge with unknown VLAN protocol is not supported");
+				return -EOPNOTSUPP;
+			}
+			if (vlan_uses_dev(lower_dev) &&
+			    br_vlan_enabled(upper_dev) &&
+			    proto == ETH_P_8021AD) {
+				NL_SET_ERR_MSG_MOD(extack, "Enslaving a port that already has a VLAN upper to an 802.1ad bridge is not supported");
+				return -EOPNOTSUPP;
+			}
+		}
+		if (netif_is_bridge_port(lower_dev) && is_vlan_dev(upper_dev)) {
+			struct net_device *br_dev = netdev_master_upper_dev_get(lower_dev);
+
+			if (br_vlan_enabled(br_dev)) {
+				br_vlan_get_proto(br_dev, &proto);
+				if (proto == ETH_P_8021AD) {
+					NL_SET_ERR_MSG_MOD(extack, "VLAN uppers are not supported on a port enslaved to an 802.1ad bridge");
+					return -EOPNOTSUPP;
+				}
+			}
+		}
+		if (is_vlan_dev(upper_dev) &&
+		    ntohs(vlan_dev_vlan_proto(upper_dev)) != ETH_P_8021Q) {
+			NL_SET_ERR_MSG_MOD(extack, "VLAN uppers are only supported with 802.1q VLAN protocol");
+			return -EOPNOTSUPP;
+		}
+		break;
+	case NETDEV_CHANGEUPPER:
+		upper_dev = info->upper_dev;
+		if (netif_is_bridge_master(upper_dev)) {
+			if (info->linking)
+				err = mlxsw_sp_port_bridge_join(mlxsw_sp_port,
+								lower_dev,
+								upper_dev,
+								extack);
+			else
+				mlxsw_sp_port_bridge_leave(mlxsw_sp_port,
+							   lower_dev,
+							   upper_dev);
+		} else if (netif_is_lag_master(upper_dev)) {
+			if (info->linking) {
+				err = mlxsw_sp_port_lag_join(mlxsw_sp_port,
+							     upper_dev, extack);
+			} else {
+				mlxsw_sp_port_lag_col_dist_disable(mlxsw_sp_port);
+				mlxsw_sp_port_lag_leave(mlxsw_sp_port,
+							upper_dev);
+			}
+		} else if (netif_is_ovs_master(upper_dev)) {
+			if (info->linking)
+				err = mlxsw_sp_port_ovs_join(mlxsw_sp_port);
+			else
+				mlxsw_sp_port_ovs_leave(mlxsw_sp_port);
+		} else if (netif_is_macvlan(upper_dev)) {
+			if (!info->linking)
+				mlxsw_sp_rif_macvlan_del(mlxsw_sp, upper_dev);
+		} else if (is_vlan_dev(upper_dev)) {
+			struct net_device *br_dev;
+
+			if (!netif_is_bridge_port(upper_dev))
+				break;
+			if (info->linking)
+				break;
+			br_dev = netdev_master_upper_dev_get(upper_dev);
+			mlxsw_sp_port_bridge_leave(mlxsw_sp_port, upper_dev,
+						   br_dev);
+		}
+		break;
+	}
+
+	return err;
+}
+
+static int mlxsw_sp_netdevice_port_lower_event(struct net_device *dev,
+					       unsigned long event, void *ptr)
+{
+	struct netdev_notifier_changelowerstate_info *info;
+	struct mlxsw_sp_port *mlxsw_sp_port;
+	int err;
+
+	mlxsw_sp_port = netdev_priv(dev);
+	info = ptr;
+
+	switch (event) {
+	case NETDEV_CHANGELOWERSTATE:
+		if (netif_is_lag_port(dev) && mlxsw_sp_port->lagged) {
+			err = mlxsw_sp_port_lag_changed(mlxsw_sp_port,
+							info->lower_state_info);
+			if (err)
+				netdev_err(dev, "Failed to reflect link aggregation lower state change\n");
+		}
+		break;
+	}
+
+	return 0;
+}
+
+static int mlxsw_sp_netdevice_port_event(struct net_device *lower_dev,
+					 struct net_device *port_dev,
+					 unsigned long event, void *ptr)
+{
+	switch (event) {
+	case NETDEV_PRECHANGEUPPER:
+	case NETDEV_CHANGEUPPER:
+		return mlxsw_sp_netdevice_port_upper_event(lower_dev, port_dev,
+							   event, ptr);
+	case NETDEV_CHANGELOWERSTATE:
+		return mlxsw_sp_netdevice_port_lower_event(port_dev, event,
+							   ptr);
+	}
+
+	return 0;
+}
+
+static int mlxsw_sp_netdevice_lag_event(struct net_device *lag_dev,
+					unsigned long event, void *ptr)
+{
+	struct net_device *dev;
+	struct list_head *iter;
+	int ret;
+
+	netdev_for_each_lower_dev(lag_dev, dev, iter) {
+		if (mlxsw_sp_port_dev_check(dev)) {
+			ret = mlxsw_sp_netdevice_port_event(lag_dev, dev, event,
+							    ptr);
+			if (ret)
+				return ret;
+		}
+	}
+
+	return 0;
+}
+
+static int mlxsw_sp_netdevice_port_vlan_event(struct net_device *vlan_dev,
+					      struct net_device *dev,
+					      unsigned long event, void *ptr,
+					      u16 vid)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	struct netdev_notifier_changeupper_info *info = ptr;
+	struct netlink_ext_ack *extack;
+	struct net_device *upper_dev;
+	int err = 0;
+
+	extack = netdev_notifier_info_to_extack(&info->info);
+
+	switch (event) {
+	case NETDEV_PRECHANGEUPPER:
+		upper_dev = info->upper_dev;
+		if (!netif_is_bridge_master(upper_dev) &&
+		    !netif_is_macvlan(upper_dev) &&
+		    !netif_is_l3_master(upper_dev)) {
+			NL_SET_ERR_MSG_MOD(extack, "Unknown upper device type");
+			return -EINVAL;
+		}
+		if (!info->linking)
+			break;
+		if (netif_is_bridge_master(upper_dev) &&
+		    !mlxsw_sp_bridge_device_is_offloaded(mlxsw_sp, upper_dev) &&
+		    mlxsw_sp_bridge_has_vxlan(upper_dev) &&
+		    !mlxsw_sp_bridge_vxlan_is_valid(upper_dev, extack))
+			return -EOPNOTSUPP;
+		if (netdev_has_any_upper_dev(upper_dev) &&
+		    (!netif_is_bridge_master(upper_dev) ||
+		     !mlxsw_sp_bridge_device_is_offloaded(mlxsw_sp,
+							  upper_dev))) {
+			NL_SET_ERR_MSG_MOD(extack, "Enslaving a port to a device that already has an upper device is not supported");
+			return -EINVAL;
+		}
+		if (netif_is_macvlan(upper_dev) &&
+		    !mlxsw_sp_rif_exists(mlxsw_sp, vlan_dev)) {
+			NL_SET_ERR_MSG_MOD(extack, "macvlan is only supported on top of router interfaces");
+			return -EOPNOTSUPP;
+		}
+		break;
+	case NETDEV_CHANGEUPPER:
+		upper_dev = info->upper_dev;
+		if (netif_is_bridge_master(upper_dev)) {
+			if (info->linking)
+				err = mlxsw_sp_port_bridge_join(mlxsw_sp_port,
+								vlan_dev,
+								upper_dev,
+								extack);
+			else
+				mlxsw_sp_port_bridge_leave(mlxsw_sp_port,
+							   vlan_dev,
+							   upper_dev);
+		} else if (netif_is_macvlan(upper_dev)) {
+			if (!info->linking)
+				mlxsw_sp_rif_macvlan_del(mlxsw_sp, upper_dev);
+		}
+		break;
+	}
+
+	return err;
+}
+
+static int mlxsw_sp_netdevice_lag_port_vlan_event(struct net_device *vlan_dev,
+						  struct net_device *lag_dev,
+						  unsigned long event,
+						  void *ptr, u16 vid)
+{
+	struct net_device *dev;
+	struct list_head *iter;
+	int ret;
+
+	netdev_for_each_lower_dev(lag_dev, dev, iter) {
+		if (mlxsw_sp_port_dev_check(dev)) {
+			ret = mlxsw_sp_netdevice_port_vlan_event(vlan_dev, dev,
+								 event, ptr,
+								 vid);
+			if (ret)
+				return ret;
+		}
+	}
+
+	return 0;
+}
+
+static int mlxsw_sp_netdevice_bridge_vlan_event(struct net_device *vlan_dev,
+						struct net_device *br_dev,
+						unsigned long event, void *ptr,
+						u16 vid)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_lower_get(vlan_dev);
+	struct netdev_notifier_changeupper_info *info = ptr;
+	struct netlink_ext_ack *extack;
+	struct net_device *upper_dev;
+
+	if (!mlxsw_sp)
+		return 0;
+
+	extack = netdev_notifier_info_to_extack(&info->info);
+
+	switch (event) {
+	case NETDEV_PRECHANGEUPPER:
+		upper_dev = info->upper_dev;
+		if (!netif_is_macvlan(upper_dev) &&
+		    !netif_is_l3_master(upper_dev)) {
+			NL_SET_ERR_MSG_MOD(extack, "Unknown upper device type");
+			return -EOPNOTSUPP;
+		}
+		if (!info->linking)
+			break;
+		if (netif_is_macvlan(upper_dev) &&
+		    !mlxsw_sp_rif_exists(mlxsw_sp, vlan_dev)) {
+			NL_SET_ERR_MSG_MOD(extack, "macvlan is only supported on top of router interfaces");
+			return -EOPNOTSUPP;
+		}
+		break;
+	case NETDEV_CHANGEUPPER:
+		upper_dev = info->upper_dev;
+		if (info->linking)
+			break;
+		if (netif_is_macvlan(upper_dev))
+			mlxsw_sp_rif_macvlan_del(mlxsw_sp, upper_dev);
+		break;
+	}
+
+	return 0;
+}
+
+static int mlxsw_sp_netdevice_vlan_event(struct net_device *vlan_dev,
+					 unsigned long event, void *ptr)
+{
+	struct net_device *real_dev = vlan_dev_real_dev(vlan_dev);
+	u16 vid = vlan_dev_vlan_id(vlan_dev);
+
+	if (mlxsw_sp_port_dev_check(real_dev))
+		return mlxsw_sp_netdevice_port_vlan_event(vlan_dev, real_dev,
+							  event, ptr, vid);
+	else if (netif_is_lag_master(real_dev))
+		return mlxsw_sp_netdevice_lag_port_vlan_event(vlan_dev,
+							      real_dev, event,
+							      ptr, vid);
+	else if (netif_is_bridge_master(real_dev))
+		return mlxsw_sp_netdevice_bridge_vlan_event(vlan_dev, real_dev,
+							    event, ptr, vid);
+
+	return 0;
+}
+
+static int mlxsw_sp_netdevice_bridge_event(struct net_device *br_dev,
+					   unsigned long event, void *ptr)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_lower_get(br_dev);
+	struct netdev_notifier_changeupper_info *info = ptr;
+	struct netlink_ext_ack *extack;
+	struct net_device *upper_dev;
+	u16 proto;
+
+	if (!mlxsw_sp)
+		return 0;
+
+	extack = netdev_notifier_info_to_extack(&info->info);
+
+	switch (event) {
+	case NETDEV_PRECHANGEUPPER:
+		upper_dev = info->upper_dev;
+		if (!is_vlan_dev(upper_dev) &&
+		    !netif_is_macvlan(upper_dev) &&
+		    !netif_is_l3_master(upper_dev)) {
+			NL_SET_ERR_MSG_MOD(extack, "Unknown upper device type");
+			return -EOPNOTSUPP;
+		}
+		if (!info->linking)
+			break;
+		if (br_vlan_enabled(br_dev)) {
+			br_vlan_get_proto(br_dev, &proto);
+			if (proto == ETH_P_8021AD) {
+				NL_SET_ERR_MSG_MOD(extack, "Upper devices are not supported on top of an 802.1ad bridge");
+				return -EOPNOTSUPP;
+			}
+		}
+		if (is_vlan_dev(upper_dev) &&
+		    ntohs(vlan_dev_vlan_proto(upper_dev)) != ETH_P_8021Q) {
+			NL_SET_ERR_MSG_MOD(extack, "VLAN uppers are only supported with 802.1q VLAN protocol");
+			return -EOPNOTSUPP;
+		}
+		if (netif_is_macvlan(upper_dev) &&
+		    !mlxsw_sp_rif_exists(mlxsw_sp, br_dev)) {
+			NL_SET_ERR_MSG_MOD(extack, "macvlan is only supported on top of router interfaces");
+			return -EOPNOTSUPP;
+		}
+		break;
+	case NETDEV_CHANGEUPPER:
+		upper_dev = info->upper_dev;
+		if (info->linking)
+			break;
+		if (is_vlan_dev(upper_dev))
+			mlxsw_sp_rif_destroy_by_dev(mlxsw_sp, upper_dev);
+		if (netif_is_macvlan(upper_dev))
+			mlxsw_sp_rif_macvlan_del(mlxsw_sp, upper_dev);
+		break;
+	}
+
+	return 0;
+}
+
+static int mlxsw_sp_netdevice_macvlan_event(struct net_device *macvlan_dev,
+					    unsigned long event, void *ptr)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_lower_get(macvlan_dev);
+	struct netdev_notifier_changeupper_info *info = ptr;
+	struct netlink_ext_ack *extack;
+	struct net_device *upper_dev;
+
+	if (!mlxsw_sp || event != NETDEV_PRECHANGEUPPER)
+		return 0;
+
+	extack = netdev_notifier_info_to_extack(&info->info);
+	upper_dev = info->upper_dev;
+
+	if (!netif_is_l3_master(upper_dev)) {
+		NL_SET_ERR_MSG_MOD(extack, "Unknown upper device type");
+		return -EOPNOTSUPP;
+	}
+
+	return 0;
+}
+
+static int mlxsw_sp_netdevice_vxlan_event(struct mlxsw_sp *mlxsw_sp,
+					  struct net_device *dev,
+					  unsigned long event, void *ptr)
+{
+	struct netdev_notifier_changeupper_info *cu_info;
+	struct netdev_notifier_info *info = ptr;
+	struct netlink_ext_ack *extack;
+	struct net_device *upper_dev;
+
+	extack = netdev_notifier_info_to_extack(info);
+
+	switch (event) {
+	case NETDEV_CHANGEUPPER:
+		cu_info = container_of(info,
+				       struct netdev_notifier_changeupper_info,
+				       info);
+		upper_dev = cu_info->upper_dev;
+		if (!netif_is_bridge_master(upper_dev))
+			return 0;
+		if (!mlxsw_sp_lower_get(upper_dev))
+			return 0;
+		if (!mlxsw_sp_bridge_vxlan_is_valid(upper_dev, extack))
+			return -EOPNOTSUPP;
+		if (cu_info->linking) {
+			if (!netif_running(dev))
+				return 0;
+			/* When the bridge is VLAN-aware, the VNI of the VxLAN
+			 * device needs to be mapped to a VLAN, but at this
+			 * point no VLANs are configured on the VxLAN device
+			 */
+			if (br_vlan_enabled(upper_dev))
+				return 0;
+			return mlxsw_sp_bridge_vxlan_join(mlxsw_sp, upper_dev,
+							  dev, 0, extack);
+		} else {
+			/* VLANs were already flushed, which triggered the
+			 * necessary cleanup
+			 */
+			if (br_vlan_enabled(upper_dev))
+				return 0;
+			mlxsw_sp_bridge_vxlan_leave(mlxsw_sp, dev);
+		}
+		break;
+	case NETDEV_PRE_UP:
+		upper_dev = netdev_master_upper_dev_get(dev);
+		if (!upper_dev)
+			return 0;
+		if (!netif_is_bridge_master(upper_dev))
+			return 0;
+		if (!mlxsw_sp_lower_get(upper_dev))
+			return 0;
+		return mlxsw_sp_bridge_vxlan_join(mlxsw_sp, upper_dev, dev, 0,
+						  extack);
+	case NETDEV_DOWN:
+		upper_dev = netdev_master_upper_dev_get(dev);
+		if (!upper_dev)
+			return 0;
+		if (!netif_is_bridge_master(upper_dev))
+			return 0;
+		if (!mlxsw_sp_lower_get(upper_dev))
+			return 0;
+		mlxsw_sp_bridge_vxlan_leave(mlxsw_sp, dev);
+		break;
+	}
+
+	return 0;
+}
+
+static int mlxsw_sp_netdevice_event(struct notifier_block *nb,
+				    unsigned long event, void *ptr)
+{
+	struct net_device *dev = netdev_notifier_info_to_dev(ptr);
+	struct mlxsw_sp_span_entry *span_entry;
+	struct mlxsw_sp *mlxsw_sp;
+	int err = 0;
+
+	mlxsw_sp = container_of(nb, struct mlxsw_sp, netdevice_nb);
+	if (event == NETDEV_UNREGISTER) {
+		span_entry = mlxsw_sp_span_entry_find_by_port(mlxsw_sp, dev);
+		if (span_entry)
+			mlxsw_sp_span_entry_invalidate(mlxsw_sp, span_entry);
+	}
+	mlxsw_sp_span_respin(mlxsw_sp);
+
+	if (netif_is_vxlan(dev))
+		err = mlxsw_sp_netdevice_vxlan_event(mlxsw_sp, dev, event, ptr);
+	else if (mlxsw_sp_port_dev_check(dev))
+		err = mlxsw_sp_netdevice_port_event(dev, dev, event, ptr);
+	else if (netif_is_lag_master(dev))
+		err = mlxsw_sp_netdevice_lag_event(dev, event, ptr);
+	else if (is_vlan_dev(dev))
+		err = mlxsw_sp_netdevice_vlan_event(dev, event, ptr);
+	else if (netif_is_bridge_master(dev))
+		err = mlxsw_sp_netdevice_bridge_event(dev, event, ptr);
+	else if (netif_is_macvlan(dev))
+		err = mlxsw_sp_netdevice_macvlan_event(dev, event, ptr);
+
+	return notifier_from_errno(err);
+}
+
+static struct notifier_block mlxsw_sp_inetaddr_valid_nb __read_mostly = {
+	.notifier_call = mlxsw_sp_inetaddr_valid_event,
+};
+
+static struct notifier_block mlxsw_sp_inet6addr_valid_nb __read_mostly = {
+	.notifier_call = mlxsw_sp_inet6addr_valid_event,
+};
+
+static const struct pci_device_id mlxsw_sp1_pci_id_table[] = {
+	{PCI_VDEVICE(MELLANOX, PCI_DEVICE_ID_MELLANOX_SPECTRUM), 0},
+	{0, },
+};
+
+static struct pci_driver mlxsw_sp1_pci_driver = {
+	.name = mlxsw_sp1_driver_name,
+	.id_table = mlxsw_sp1_pci_id_table,
+};
+
+static const struct pci_device_id mlxsw_sp2_pci_id_table[] = {
+	{PCI_VDEVICE(MELLANOX, PCI_DEVICE_ID_MELLANOX_SPECTRUM2), 0},
+	{0, },
+};
+
+static struct pci_driver mlxsw_sp2_pci_driver = {
+	.name = mlxsw_sp2_driver_name,
+	.id_table = mlxsw_sp2_pci_id_table,
+};
+
+static const struct pci_device_id mlxsw_sp3_pci_id_table[] = {
+	{PCI_VDEVICE(MELLANOX, PCI_DEVICE_ID_MELLANOX_SPECTRUM3), 0},
+	{0, },
+};
+
+static struct pci_driver mlxsw_sp3_pci_driver = {
+	.name = mlxsw_sp3_driver_name,
+	.id_table = mlxsw_sp3_pci_id_table,
+};
+
+static const struct pci_device_id mlxsw_sp4_pci_id_table[] = {
+	{PCI_VDEVICE(MELLANOX, PCI_DEVICE_ID_MELLANOX_SPECTRUM4), 0},
+	{0, },
+};
+
+static struct pci_driver mlxsw_sp4_pci_driver = {
+	.name = mlxsw_sp4_driver_name,
+	.id_table = mlxsw_sp4_pci_id_table,
+};
+
+static int __init mlxsw_sp_module_init(void)
+{
+	int err;
+
+	register_inetaddr_validator_notifier(&mlxsw_sp_inetaddr_valid_nb);
+	register_inet6addr_validator_notifier(&mlxsw_sp_inet6addr_valid_nb);
+
+	err = mlxsw_core_driver_register(&mlxsw_sp1_driver);
+	if (err)
+		goto err_sp1_core_driver_register;
+
+	err = mlxsw_core_driver_register(&mlxsw_sp2_driver);
+	if (err)
+		goto err_sp2_core_driver_register;
+
+	err = mlxsw_core_driver_register(&mlxsw_sp3_driver);
+	if (err)
+		goto err_sp3_core_driver_register;
+
+	err = mlxsw_core_driver_register(&mlxsw_sp4_driver);
+	if (err)
+		goto err_sp4_core_driver_register;
+
+	err = mlxsw_pci_driver_register(&mlxsw_sp1_pci_driver);
+	if (err)
+		goto err_sp1_pci_driver_register;
+
+	err = mlxsw_pci_driver_register(&mlxsw_sp2_pci_driver);
+	if (err)
+		goto err_sp2_pci_driver_register;
+
+	err = mlxsw_pci_driver_register(&mlxsw_sp3_pci_driver);
+	if (err)
+		goto err_sp3_pci_driver_register;
+
+	err = mlxsw_pci_driver_register(&mlxsw_sp4_pci_driver);
+	if (err)
+		goto err_sp4_pci_driver_register;
+
+	return 0;
+
+err_sp4_pci_driver_register:
+	mlxsw_pci_driver_unregister(&mlxsw_sp3_pci_driver);
+err_sp3_pci_driver_register:
+	mlxsw_pci_driver_unregister(&mlxsw_sp2_pci_driver);
+err_sp2_pci_driver_register:
+	mlxsw_pci_driver_unregister(&mlxsw_sp1_pci_driver);
+err_sp1_pci_driver_register:
+	mlxsw_core_driver_unregister(&mlxsw_sp4_driver);
+err_sp4_core_driver_register:
+	mlxsw_core_driver_unregister(&mlxsw_sp3_driver);
+err_sp3_core_driver_register:
+	mlxsw_core_driver_unregister(&mlxsw_sp2_driver);
+err_sp2_core_driver_register:
+	mlxsw_core_driver_unregister(&mlxsw_sp1_driver);
+err_sp1_core_driver_register:
+	unregister_inet6addr_validator_notifier(&mlxsw_sp_inet6addr_valid_nb);
+	unregister_inetaddr_validator_notifier(&mlxsw_sp_inetaddr_valid_nb);
+	return err;
+}
+
+static void __exit mlxsw_sp_module_exit(void)
+{
+	mlxsw_pci_driver_unregister(&mlxsw_sp4_pci_driver);
+	mlxsw_pci_driver_unregister(&mlxsw_sp3_pci_driver);
+	mlxsw_pci_driver_unregister(&mlxsw_sp2_pci_driver);
+	mlxsw_pci_driver_unregister(&mlxsw_sp1_pci_driver);
+	mlxsw_core_driver_unregister(&mlxsw_sp4_driver);
+	mlxsw_core_driver_unregister(&mlxsw_sp3_driver);
+	mlxsw_core_driver_unregister(&mlxsw_sp2_driver);
+	mlxsw_core_driver_unregister(&mlxsw_sp1_driver);
+	unregister_inet6addr_validator_notifier(&mlxsw_sp_inet6addr_valid_nb);
+	unregister_inetaddr_validator_notifier(&mlxsw_sp_inetaddr_valid_nb);
+}
+
+module_init(mlxsw_sp_module_init);
+module_exit(mlxsw_sp_module_exit);
+
+MODULE_LICENSE("Dual BSD/GPL");
+MODULE_AUTHOR("Jiri Pirko <jiri@mellanox.com>");
+MODULE_DESCRIPTION("Mellanox Spectrum driver");
+MODULE_DEVICE_TABLE(pci, mlxsw_sp1_pci_id_table);
+MODULE_DEVICE_TABLE(pci, mlxsw_sp2_pci_id_table);
+MODULE_DEVICE_TABLE(pci, mlxsw_sp3_pci_id_table);
+MODULE_DEVICE_TABLE(pci, mlxsw_sp4_pci_id_table);
+MODULE_FIRMWARE(MLXSW_SP1_FW_FILENAME);
+MODULE_FIRMWARE(MLXSW_SP2_FW_FILENAME);
+MODULE_FIRMWARE(MLXSW_SP3_FW_FILENAME);
+MODULE_FIRMWARE(MLXSW_SP_LINECARDS_INI_BUNDLE_FILENAME);
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum.h b/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
new file mode 100644
index 000000000..c8ff2a6d7
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
@@ -0,0 +1,1493 @@
+/* SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0 */
+/* Copyright (c) 2015-2018 Mellanox Technologies. All rights reserved */
+
+#ifndef _MLXSW_SPECTRUM_H
+#define _MLXSW_SPECTRUM_H
+
+#include <linux/ethtool.h>
+#include <linux/types.h>
+#include <linux/netdevice.h>
+#include <linux/rhashtable.h>
+#include <linux/bitops.h>
+#include <linux/if_bridge.h>
+#include <linux/if_vlan.h>
+#include <linux/list.h>
+#include <linux/dcbnl.h>
+#include <linux/in6.h>
+#include <linux/notifier.h>
+#include <linux/net_namespace.h>
+#include <linux/spinlock.h>
+#include <net/psample.h>
+#include <net/pkt_cls.h>
+#include <net/red.h>
+#include <net/vxlan.h>
+#include <net/flow_offload.h>
+#include <net/inet_ecn.h>
+
+#include "port.h"
+#include "core.h"
+#include "core_acl_flex_keys.h"
+#include "core_acl_flex_actions.h"
+#include "reg.h"
+
+#define MLXSW_SP_DEFAULT_VID (VLAN_N_VID - 1)
+
+#define MLXSW_SP_FID_8021D_MAX 1024
+
+#define MLXSW_SP_MID_MAX 7000
+
+#define MLXSW_SP_KVD_LINEAR_SIZE 98304 /* entries */
+#define MLXSW_SP_KVD_GRANULARITY 128
+
+#define MLXSW_SP_RESOURCE_NAME_KVD "kvd"
+#define MLXSW_SP_RESOURCE_NAME_KVD_LINEAR "linear"
+#define MLXSW_SP_RESOURCE_NAME_KVD_HASH_SINGLE "hash_single"
+#define MLXSW_SP_RESOURCE_NAME_KVD_HASH_DOUBLE "hash_double"
+#define MLXSW_SP_RESOURCE_NAME_KVD_LINEAR_SINGLES "singles"
+#define MLXSW_SP_RESOURCE_NAME_KVD_LINEAR_CHUNKS "chunks"
+#define MLXSW_SP_RESOURCE_NAME_KVD_LINEAR_LARGE_CHUNKS "large_chunks"
+
+#define MLXSW_SP_RESOURCE_NAME_SPAN "span_agents"
+
+#define MLXSW_SP_RESOURCE_NAME_COUNTERS "counters"
+#define MLXSW_SP_RESOURCE_NAME_COUNTERS_FLOW "flow"
+#define MLXSW_SP_RESOURCE_NAME_COUNTERS_RIF "rif"
+
+enum mlxsw_sp_resource_id {
+	MLXSW_SP_RESOURCE_KVD = MLXSW_CORE_RESOURCE_MAX,
+	MLXSW_SP_RESOURCE_KVD_LINEAR,
+	MLXSW_SP_RESOURCE_KVD_HASH_SINGLE,
+	MLXSW_SP_RESOURCE_KVD_HASH_DOUBLE,
+	MLXSW_SP_RESOURCE_KVD_LINEAR_SINGLE,
+	MLXSW_SP_RESOURCE_KVD_LINEAR_CHUNKS,
+	MLXSW_SP_RESOURCE_KVD_LINEAR_LARGE_CHUNKS,
+	MLXSW_SP_RESOURCE_SPAN,
+	MLXSW_SP_RESOURCE_COUNTERS,
+	MLXSW_SP_RESOURCE_COUNTERS_FLOW,
+	MLXSW_SP_RESOURCE_COUNTERS_RIF,
+	MLXSW_SP_RESOURCE_GLOBAL_POLICERS,
+	MLXSW_SP_RESOURCE_SINGLE_RATE_POLICERS,
+	MLXSW_SP_RESOURCE_RIF_MAC_PROFILES,
+	MLXSW_SP_RESOURCE_RIFS,
+};
+
+struct mlxsw_sp_port;
+struct mlxsw_sp_rif;
+struct mlxsw_sp_span_entry;
+enum mlxsw_sp_l3proto;
+union mlxsw_sp_l3addr;
+
+struct mlxsw_sp_upper {
+	struct net_device *dev;
+	unsigned int ref_count;
+};
+
+enum mlxsw_sp_rif_type {
+	MLXSW_SP_RIF_TYPE_SUBPORT,
+	MLXSW_SP_RIF_TYPE_VLAN,
+	MLXSW_SP_RIF_TYPE_FID,
+	MLXSW_SP_RIF_TYPE_IPIP_LB, /* IP-in-IP loopback. */
+	MLXSW_SP_RIF_TYPE_MAX,
+};
+
+struct mlxsw_sp_router_ops;
+
+extern const struct mlxsw_sp_router_ops mlxsw_sp1_router_ops;
+extern const struct mlxsw_sp_router_ops mlxsw_sp2_router_ops;
+
+struct mlxsw_sp_switchdev_ops;
+
+extern const struct mlxsw_sp_switchdev_ops mlxsw_sp1_switchdev_ops;
+extern const struct mlxsw_sp_switchdev_ops mlxsw_sp2_switchdev_ops;
+
+enum mlxsw_sp_fid_type {
+	MLXSW_SP_FID_TYPE_8021Q,
+	MLXSW_SP_FID_TYPE_8021D,
+	MLXSW_SP_FID_TYPE_RFID,
+	MLXSW_SP_FID_TYPE_DUMMY,
+	MLXSW_SP_FID_TYPE_MAX,
+};
+
+enum mlxsw_sp_nve_type {
+	MLXSW_SP_NVE_TYPE_VXLAN,
+};
+
+struct mlxsw_sp_sb;
+struct mlxsw_sp_bridge;
+struct mlxsw_sp_router;
+struct mlxsw_sp_mr;
+struct mlxsw_sp_acl;
+struct mlxsw_sp_counter_pool;
+struct mlxsw_sp_fid_core;
+struct mlxsw_sp_kvdl;
+struct mlxsw_sp_nve;
+struct mlxsw_sp_kvdl_ops;
+struct mlxsw_sp_mr_tcam_ops;
+struct mlxsw_sp_acl_rulei_ops;
+struct mlxsw_sp_acl_tcam_ops;
+struct mlxsw_sp_nve_ops;
+struct mlxsw_sp_sb_ops;
+struct mlxsw_sp_sb_vals;
+struct mlxsw_sp_port_type_speed_ops;
+struct mlxsw_sp_ptp_state;
+struct mlxsw_sp_ptp_ops;
+struct mlxsw_sp_span_ops;
+struct mlxsw_sp_qdisc_state;
+struct mlxsw_sp_mall_entry;
+struct mlxsw_sp_pgt;
+
+struct mlxsw_sp_port_mapping {
+	u8 module;
+	u8 slot_index;
+	u8 width; /* Number of lanes used by the port */
+	u8 module_width; /* Number of lanes in the module (static) */
+	u8 lane;
+};
+
+struct mlxsw_sp_port_mapping_events {
+	struct list_head queue;
+	spinlock_t queue_lock; /* protects queue */
+	struct work_struct work;
+};
+
+struct mlxsw_sp_parsing {
+	refcount_t parsing_depth_ref;
+	u16 parsing_depth;
+	u16 vxlan_udp_dport;
+	struct mutex lock; /* Protects parsing configuration */
+};
+
+struct mlxsw_sp {
+	struct mlxsw_sp_port **ports;
+	struct mlxsw_core *core;
+	const struct mlxsw_bus_info *bus_info;
+	unsigned char base_mac[ETH_ALEN];
+	const unsigned char *mac_mask;
+	struct mlxsw_sp_upper *lags;
+	struct mlxsw_sp_port_mapping *port_mapping;
+	struct mlxsw_sp_port_mapping_events port_mapping_events;
+	struct rhashtable sample_trigger_ht;
+	struct mlxsw_sp_sb *sb;
+	struct mlxsw_sp_bridge *bridge;
+	struct mlxsw_sp_router *router;
+	struct mlxsw_sp_mr *mr;
+	struct mlxsw_afa *afa;
+	struct mlxsw_sp_acl *acl;
+	struct mlxsw_sp_fid_core *fid_core;
+	struct mlxsw_sp_policer_core *policer_core;
+	struct mlxsw_sp_kvdl *kvdl;
+	struct mlxsw_sp_nve *nve;
+	struct notifier_block netdevice_nb;
+	struct mlxsw_sp_ptp_clock *clock;
+	struct mlxsw_sp_ptp_state *ptp_state;
+	struct mlxsw_sp_counter_pool *counter_pool;
+	struct mlxsw_sp_span *span;
+	struct mlxsw_sp_trap *trap;
+	struct mlxsw_sp_parsing parsing;
+	const struct mlxsw_sp_switchdev_ops *switchdev_ops;
+	const struct mlxsw_sp_kvdl_ops *kvdl_ops;
+	const struct mlxsw_afa_ops *afa_ops;
+	const struct mlxsw_afk_ops *afk_ops;
+	const struct mlxsw_sp_mr_tcam_ops *mr_tcam_ops;
+	const struct mlxsw_sp_acl_rulei_ops *acl_rulei_ops;
+	const struct mlxsw_sp_acl_tcam_ops *acl_tcam_ops;
+	const struct mlxsw_sp_acl_bf_ops *acl_bf_ops;
+	const struct mlxsw_sp_nve_ops **nve_ops_arr;
+	const struct mlxsw_sp_sb_vals *sb_vals;
+	const struct mlxsw_sp_sb_ops *sb_ops;
+	const struct mlxsw_sp_port_type_speed_ops *port_type_speed_ops;
+	const struct mlxsw_sp_ptp_ops *ptp_ops;
+	const struct mlxsw_sp_span_ops *span_ops;
+	const struct mlxsw_sp_policer_core_ops *policer_core_ops;
+	const struct mlxsw_sp_trap_ops *trap_ops;
+	const struct mlxsw_sp_mall_ops *mall_ops;
+	const struct mlxsw_sp_router_ops *router_ops;
+	const struct mlxsw_listener *listeners;
+	const struct mlxsw_sp_fid_family **fid_family_arr;
+	size_t listeners_count;
+	u32 lowest_shaper_bs;
+	struct rhashtable ipv6_addr_ht;
+	struct mutex ipv6_addr_ht_lock; /* Protects ipv6_addr_ht */
+	struct mlxsw_sp_pgt *pgt;
+	bool pgt_smpe_index_valid;
+};
+
+struct mlxsw_sp_ptp_ops {
+	struct mlxsw_sp_ptp_clock *
+		(*clock_init)(struct mlxsw_sp *mlxsw_sp, struct device *dev);
+	void (*clock_fini)(struct mlxsw_sp_ptp_clock *clock);
+
+	struct mlxsw_sp_ptp_state *(*init)(struct mlxsw_sp *mlxsw_sp);
+	void (*fini)(struct mlxsw_sp_ptp_state *ptp_state);
+
+	/* Notify a driver that a packet that might be PTP was received. Driver
+	 * is responsible for freeing the passed-in SKB.
+	 */
+	void (*receive)(struct mlxsw_sp *mlxsw_sp, struct sk_buff *skb,
+			u16 local_port);
+
+	/* Notify a driver that a timestamped packet was transmitted. Driver
+	 * is responsible for freeing the passed-in SKB.
+	 */
+	void (*transmitted)(struct mlxsw_sp *mlxsw_sp, struct sk_buff *skb,
+			    u16 local_port);
+
+	int (*hwtstamp_get)(struct mlxsw_sp_port *mlxsw_sp_port,
+			    struct hwtstamp_config *config);
+	int (*hwtstamp_set)(struct mlxsw_sp_port *mlxsw_sp_port,
+			    struct hwtstamp_config *config);
+	void (*shaper_work)(struct work_struct *work);
+	int (*get_ts_info)(struct mlxsw_sp *mlxsw_sp,
+			   struct ethtool_ts_info *info);
+	int (*get_stats_count)(void);
+	void (*get_stats_strings)(u8 **p);
+	void (*get_stats)(struct mlxsw_sp_port *mlxsw_sp_port,
+			  u64 *data, int data_index);
+	int (*txhdr_construct)(struct mlxsw_core *mlxsw_core,
+			       struct mlxsw_sp_port *mlxsw_sp_port,
+			       struct sk_buff *skb,
+			       const struct mlxsw_tx_info *tx_info);
+};
+
+static inline struct mlxsw_sp_upper *
+mlxsw_sp_lag_get(struct mlxsw_sp *mlxsw_sp, u16 lag_id)
+{
+	return &mlxsw_sp->lags[lag_id];
+}
+
+struct mlxsw_sp_port_pcpu_stats {
+	u64			rx_packets;
+	u64			rx_bytes;
+	u64			tx_packets;
+	u64			tx_bytes;
+	struct u64_stats_sync	syncp;
+	u32			tx_dropped;
+};
+
+enum mlxsw_sp_sample_trigger_type {
+	MLXSW_SP_SAMPLE_TRIGGER_TYPE_INGRESS,
+	MLXSW_SP_SAMPLE_TRIGGER_TYPE_EGRESS,
+	MLXSW_SP_SAMPLE_TRIGGER_TYPE_POLICY_ENGINE,
+};
+
+struct mlxsw_sp_sample_trigger {
+	enum mlxsw_sp_sample_trigger_type type;
+	u16 local_port; /* Reserved when trigger type is not ingress / egress. */
+};
+
+struct mlxsw_sp_sample_params {
+	struct psample_group *psample_group;
+	u32 trunc_size;
+	u32 rate;
+	bool truncate;
+};
+
+struct mlxsw_sp_bridge_port;
+struct mlxsw_sp_fid;
+
+struct mlxsw_sp_port_vlan {
+	struct list_head list;
+	struct mlxsw_sp_port *mlxsw_sp_port;
+	struct mlxsw_sp_fid *fid;
+	u16 vid;
+	struct mlxsw_sp_bridge_port *bridge_port;
+	struct list_head bridge_vlan_node;
+};
+
+/* No need an internal lock; At worse - miss a single periodic iteration */
+struct mlxsw_sp_port_xstats {
+	u64 ecn;
+	u64 tc_ecn[TC_MAX_QUEUE];
+	u64 wred_drop[TC_MAX_QUEUE];
+	u64 tail_drop[TC_MAX_QUEUE];
+	u64 backlog[TC_MAX_QUEUE];
+	u64 tx_bytes[IEEE_8021QAZ_MAX_TCS];
+	u64 tx_packets[IEEE_8021QAZ_MAX_TCS];
+};
+
+struct mlxsw_sp_ptp_port_dir_stats {
+	u64 packets;
+	u64 timestamps;
+};
+
+struct mlxsw_sp_ptp_port_stats {
+	struct mlxsw_sp_ptp_port_dir_stats rx_gcd;
+	struct mlxsw_sp_ptp_port_dir_stats tx_gcd;
+};
+
+struct mlxsw_sp_port {
+	struct net_device *dev;
+	struct mlxsw_sp_port_pcpu_stats __percpu *pcpu_stats;
+	struct mlxsw_sp *mlxsw_sp;
+	u16 local_port;
+	u8 lagged:1,
+	   split:1;
+	u16 pvid;
+	u16 lag_id;
+	struct {
+		u8 tx_pause:1,
+		   rx_pause:1,
+		   autoneg:1;
+	} link;
+	struct {
+		struct ieee_ets *ets;
+		struct ieee_maxrate *maxrate;
+		struct ieee_pfc *pfc;
+		enum mlxsw_reg_qpts_trust_state trust_state;
+	} dcb;
+	struct mlxsw_sp_port_mapping mapping; /* mapping is constant during the
+					       * mlxsw_sp_port lifetime, however
+					       * the same localport can have
+					       * different mapping.
+					       */
+	struct {
+		#define MLXSW_HW_STATS_UPDATE_TIME HZ
+		struct rtnl_link_stats64 stats;
+		struct mlxsw_sp_port_xstats xstats;
+		struct delayed_work update_dw;
+	} periodic_hw_stats;
+	struct list_head vlans_list;
+	struct mlxsw_sp_port_vlan *default_vlan;
+	struct mlxsw_sp_qdisc_state *qdisc;
+	unsigned acl_rule_count;
+	struct mlxsw_sp_flow_block *ing_flow_block;
+	struct mlxsw_sp_flow_block *eg_flow_block;
+	struct {
+		struct delayed_work shaper_dw;
+		struct hwtstamp_config hwtstamp_config;
+		u16 ing_types;
+		u16 egr_types;
+		struct mlxsw_sp_ptp_port_stats stats;
+	} ptp;
+	int max_mtu;
+	u32 max_speed;
+	struct mlxsw_sp_hdroom *hdroom;
+	u64 module_overheat_initial_val;
+};
+
+struct mlxsw_sp_port_type_speed_ops {
+	void (*from_ptys_supported_port)(struct mlxsw_sp *mlxsw_sp,
+					 u32 ptys_eth_proto,
+					 struct ethtool_link_ksettings *cmd);
+	void (*from_ptys_link)(struct mlxsw_sp *mlxsw_sp, u32 ptys_eth_proto,
+			       unsigned long *mode);
+	u32 (*from_ptys_speed)(struct mlxsw_sp *mlxsw_sp, u32 ptys_eth_proto);
+	void (*from_ptys_link_mode)(struct mlxsw_sp *mlxsw_sp,
+				    bool carrier_ok, u32 ptys_eth_proto,
+				    struct ethtool_link_ksettings *cmd);
+	int (*ptys_max_speed)(struct mlxsw_sp_port *mlxsw_sp_port, u32 *p_max_speed);
+	u32 (*to_ptys_advert_link)(struct mlxsw_sp *mlxsw_sp,
+				   const struct ethtool_link_ksettings *cmd);
+	u32 (*to_ptys_speed_lanes)(struct mlxsw_sp *mlxsw_sp, u8 width,
+				   const struct ethtool_link_ksettings *cmd);
+	void (*reg_ptys_eth_pack)(struct mlxsw_sp *mlxsw_sp, char *payload,
+				  u16 local_port, u32 proto_admin, bool autoneg);
+	void (*reg_ptys_eth_unpack)(struct mlxsw_sp *mlxsw_sp, char *payload,
+				    u32 *p_eth_proto_cap,
+				    u32 *p_eth_proto_admin,
+				    u32 *p_eth_proto_oper);
+	u32 (*ptys_proto_cap_masked_get)(u32 eth_proto_cap);
+};
+
+struct mlxsw_sp_ports_bitmap {
+	unsigned long *bitmap;
+	unsigned int nbits;
+};
+
+static inline int
+mlxsw_sp_port_bitmap_init(struct mlxsw_sp *mlxsw_sp,
+			  struct mlxsw_sp_ports_bitmap *ports_bm)
+{
+	unsigned int nbits = mlxsw_core_max_ports(mlxsw_sp->core);
+
+	ports_bm->nbits = nbits;
+	ports_bm->bitmap = bitmap_zalloc(nbits, GFP_KERNEL);
+	if (!ports_bm->bitmap)
+		return -ENOMEM;
+
+	return 0;
+}
+
+static inline void
+mlxsw_sp_port_bitmap_fini(struct mlxsw_sp_ports_bitmap *ports_bm)
+{
+	bitmap_free(ports_bm->bitmap);
+}
+
+static inline u8 mlxsw_sp_tunnel_ecn_decap(u8 outer_ecn, u8 inner_ecn,
+					   bool *trap_en)
+{
+	bool set_ce = false;
+
+	*trap_en = !!__INET_ECN_decapsulate(outer_ecn, inner_ecn, &set_ce);
+	if (set_ce)
+		return INET_ECN_CE;
+	else if (outer_ecn == INET_ECN_ECT_1 && inner_ecn == INET_ECN_ECT_0)
+		return INET_ECN_ECT_1;
+	else
+		return inner_ecn;
+}
+
+static inline struct net_device *
+mlxsw_sp_bridge_vxlan_dev_find(struct net_device *br_dev)
+{
+	struct net_device *dev;
+	struct list_head *iter;
+
+	netdev_for_each_lower_dev(br_dev, dev, iter) {
+		if (netif_is_vxlan(dev))
+			return dev;
+	}
+
+	return NULL;
+}
+
+static inline bool mlxsw_sp_bridge_has_vxlan(struct net_device *br_dev)
+{
+	return !!mlxsw_sp_bridge_vxlan_dev_find(br_dev);
+}
+
+static inline int
+mlxsw_sp_vxlan_mapped_vid(const struct net_device *vxlan_dev, u16 *p_vid)
+{
+	struct bridge_vlan_info vinfo;
+	u16 vid = 0;
+	int err;
+
+	err = br_vlan_get_pvid(vxlan_dev, &vid);
+	if (err || !vid)
+		goto out;
+
+	err = br_vlan_get_info(vxlan_dev, vid, &vinfo);
+	if (err || !(vinfo.flags & BRIDGE_VLAN_INFO_UNTAGGED))
+		vid = 0;
+
+out:
+	*p_vid = vid;
+	return err;
+}
+
+static inline bool
+mlxsw_sp_port_is_pause_en(const struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	return mlxsw_sp_port->link.tx_pause || mlxsw_sp_port->link.rx_pause;
+}
+
+static inline struct mlxsw_sp_port *
+mlxsw_sp_port_lagged_get(struct mlxsw_sp *mlxsw_sp, u16 lag_id, u8 port_index)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port;
+	u16 local_port;
+
+	local_port = mlxsw_core_lag_mapping_get(mlxsw_sp->core,
+						lag_id, port_index);
+	mlxsw_sp_port = mlxsw_sp->ports[local_port];
+	return mlxsw_sp_port && mlxsw_sp_port->lagged ? mlxsw_sp_port : NULL;
+}
+
+static inline struct mlxsw_sp_port_vlan *
+mlxsw_sp_port_vlan_find_by_vid(const struct mlxsw_sp_port *mlxsw_sp_port,
+			       u16 vid)
+{
+	struct mlxsw_sp_port_vlan *mlxsw_sp_port_vlan;
+
+	list_for_each_entry(mlxsw_sp_port_vlan, &mlxsw_sp_port->vlans_list,
+			    list) {
+		if (mlxsw_sp_port_vlan->vid == vid)
+			return mlxsw_sp_port_vlan;
+	}
+
+	return NULL;
+}
+
+enum mlxsw_sp_flood_type {
+	MLXSW_SP_FLOOD_TYPE_UC,
+	MLXSW_SP_FLOOD_TYPE_BC,
+	MLXSW_SP_FLOOD_TYPE_MC,
+};
+
+int mlxsw_sp_port_get_stats_raw(struct net_device *dev, int grp,
+				int prio, char *ppcnt_pl);
+int mlxsw_sp_port_admin_status_set(struct mlxsw_sp_port *mlxsw_sp_port,
+				   bool is_up);
+int
+mlxsw_sp_port_vlan_classification_set(struct mlxsw_sp_port *mlxsw_sp_port,
+				      bool is_8021ad_tagged,
+				      bool is_8021q_tagged);
+static inline bool
+mlxsw_sp_local_port_is_valid(struct mlxsw_sp *mlxsw_sp, u16 local_port)
+{
+	unsigned int max_ports = mlxsw_core_max_ports(mlxsw_sp->core);
+
+	return local_port < max_ports && local_port;
+}
+
+/* spectrum_buffers.c */
+struct mlxsw_sp_hdroom_prio {
+	/* Number of port buffer associated with this priority. This is the
+	 * actually configured value.
+	 */
+	u8 buf_idx;
+	/* Value of buf_idx deduced from the DCB ETS configuration. */
+	u8 ets_buf_idx;
+	/* Value of buf_idx taken from the dcbnl_setbuffer configuration. */
+	u8 set_buf_idx;
+	bool lossy;
+};
+
+struct mlxsw_sp_hdroom_buf {
+	u32 thres_cells;
+	u32 size_cells;
+	/* Size requirement form dcbnl_setbuffer. */
+	u32 set_size_cells;
+	bool lossy;
+};
+
+enum mlxsw_sp_hdroom_mode {
+	MLXSW_SP_HDROOM_MODE_DCB,
+	MLXSW_SP_HDROOM_MODE_TC,
+};
+
+#define MLXSW_SP_PB_COUNT 10
+
+struct mlxsw_sp_hdroom {
+	enum mlxsw_sp_hdroom_mode mode;
+
+	struct {
+		struct mlxsw_sp_hdroom_prio prio[IEEE_8021Q_MAX_PRIORITIES];
+	} prios;
+	struct {
+		struct mlxsw_sp_hdroom_buf buf[MLXSW_SP_PB_COUNT];
+	} bufs;
+	struct {
+		/* Size actually configured for the internal buffer. Equal to
+		 * reserve when internal buffer is enabled.
+		 */
+		u32 size_cells;
+		/* Space reserved in the headroom for the internal buffer. Port
+		 * buffers are not allowed to grow into this space.
+		 */
+		u32 reserve_cells;
+		bool enable;
+	} int_buf;
+	int delay_bytes;
+	int mtu;
+};
+
+int mlxsw_sp_buffers_init(struct mlxsw_sp *mlxsw_sp);
+void mlxsw_sp_buffers_fini(struct mlxsw_sp *mlxsw_sp);
+int mlxsw_sp_port_buffers_init(struct mlxsw_sp_port *mlxsw_sp_port);
+void mlxsw_sp_port_buffers_fini(struct mlxsw_sp_port *mlxsw_sp_port);
+int mlxsw_sp_sb_pool_get(struct mlxsw_core *mlxsw_core,
+			 unsigned int sb_index, u16 pool_index,
+			 struct devlink_sb_pool_info *pool_info);
+int mlxsw_sp_sb_pool_set(struct mlxsw_core *mlxsw_core,
+			 unsigned int sb_index, u16 pool_index, u32 size,
+			 enum devlink_sb_threshold_type threshold_type,
+			 struct netlink_ext_ack *extack);
+int mlxsw_sp_sb_port_pool_get(struct mlxsw_core_port *mlxsw_core_port,
+			      unsigned int sb_index, u16 pool_index,
+			      u32 *p_threshold);
+int mlxsw_sp_sb_port_pool_set(struct mlxsw_core_port *mlxsw_core_port,
+			      unsigned int sb_index, u16 pool_index,
+			      u32 threshold, struct netlink_ext_ack *extack);
+int mlxsw_sp_sb_tc_pool_bind_get(struct mlxsw_core_port *mlxsw_core_port,
+				 unsigned int sb_index, u16 tc_index,
+				 enum devlink_sb_pool_type pool_type,
+				 u16 *p_pool_index, u32 *p_threshold);
+int mlxsw_sp_sb_tc_pool_bind_set(struct mlxsw_core_port *mlxsw_core_port,
+				 unsigned int sb_index, u16 tc_index,
+				 enum devlink_sb_pool_type pool_type,
+				 u16 pool_index, u32 threshold,
+				 struct netlink_ext_ack *extack);
+int mlxsw_sp_sb_occ_snapshot(struct mlxsw_core *mlxsw_core,
+			     unsigned int sb_index);
+int mlxsw_sp_sb_occ_max_clear(struct mlxsw_core *mlxsw_core,
+			      unsigned int sb_index);
+int mlxsw_sp_sb_occ_port_pool_get(struct mlxsw_core_port *mlxsw_core_port,
+				  unsigned int sb_index, u16 pool_index,
+				  u32 *p_cur, u32 *p_max);
+int mlxsw_sp_sb_occ_tc_port_bind_get(struct mlxsw_core_port *mlxsw_core_port,
+				     unsigned int sb_index, u16 tc_index,
+				     enum devlink_sb_pool_type pool_type,
+				     u32 *p_cur, u32 *p_max);
+u32 mlxsw_sp_cells_bytes(const struct mlxsw_sp *mlxsw_sp, u32 cells);
+u32 mlxsw_sp_bytes_cells(const struct mlxsw_sp *mlxsw_sp, u32 bytes);
+void mlxsw_sp_hdroom_prios_reset_buf_idx(struct mlxsw_sp_hdroom *hdroom);
+void mlxsw_sp_hdroom_bufs_reset_lossiness(struct mlxsw_sp_hdroom *hdroom);
+void mlxsw_sp_hdroom_bufs_reset_sizes(struct mlxsw_sp_port *mlxsw_sp_port,
+				      struct mlxsw_sp_hdroom *hdroom);
+int mlxsw_sp_hdroom_configure(struct mlxsw_sp_port *mlxsw_sp_port,
+			      const struct mlxsw_sp_hdroom *hdroom);
+struct mlxsw_sp_sample_params *
+mlxsw_sp_sample_trigger_params_lookup(struct mlxsw_sp *mlxsw_sp,
+				      const struct mlxsw_sp_sample_trigger *trigger);
+int
+mlxsw_sp_sample_trigger_params_set(struct mlxsw_sp *mlxsw_sp,
+				   const struct mlxsw_sp_sample_trigger *trigger,
+				   const struct mlxsw_sp_sample_params *params,
+				   struct netlink_ext_ack *extack);
+void
+mlxsw_sp_sample_trigger_params_unset(struct mlxsw_sp *mlxsw_sp,
+				     const struct mlxsw_sp_sample_trigger *trigger);
+int mlxsw_sp_ipv6_addr_kvdl_index_get(struct mlxsw_sp *mlxsw_sp,
+				      const struct in6_addr *addr6,
+				      u32 *p_kvdl_index);
+void
+mlxsw_sp_ipv6_addr_put(struct mlxsw_sp *mlxsw_sp, const struct in6_addr *addr6);
+
+extern const struct mlxsw_sp_sb_vals mlxsw_sp1_sb_vals;
+extern const struct mlxsw_sp_sb_vals mlxsw_sp2_sb_vals;
+
+extern const struct mlxsw_sp_sb_ops mlxsw_sp1_sb_ops;
+extern const struct mlxsw_sp_sb_ops mlxsw_sp2_sb_ops;
+extern const struct mlxsw_sp_sb_ops mlxsw_sp3_sb_ops;
+
+/* spectrum_switchdev.c */
+int mlxsw_sp_switchdev_init(struct mlxsw_sp *mlxsw_sp);
+void mlxsw_sp_switchdev_fini(struct mlxsw_sp *mlxsw_sp);
+int mlxsw_sp_rif_fdb_op(struct mlxsw_sp *mlxsw_sp, const char *mac, u16 fid,
+			bool adding);
+void
+mlxsw_sp_port_vlan_bridge_leave(struct mlxsw_sp_port_vlan *mlxsw_sp_port_vlan);
+int mlxsw_sp_port_bridge_join(struct mlxsw_sp_port *mlxsw_sp_port,
+			      struct net_device *brport_dev,
+			      struct net_device *br_dev,
+			      struct netlink_ext_ack *extack);
+void mlxsw_sp_port_bridge_leave(struct mlxsw_sp_port *mlxsw_sp_port,
+				struct net_device *brport_dev,
+				struct net_device *br_dev);
+bool mlxsw_sp_bridge_device_is_offloaded(const struct mlxsw_sp *mlxsw_sp,
+					 const struct net_device *br_dev);
+int mlxsw_sp_bridge_vxlan_join(struct mlxsw_sp *mlxsw_sp,
+			       const struct net_device *br_dev,
+			       const struct net_device *vxlan_dev, u16 vid,
+			       struct netlink_ext_ack *extack);
+void mlxsw_sp_bridge_vxlan_leave(struct mlxsw_sp *mlxsw_sp,
+				 const struct net_device *vxlan_dev);
+extern struct notifier_block mlxsw_sp_switchdev_notifier;
+
+/* spectrum.c */
+void mlxsw_sp_rx_listener_no_mark_func(struct sk_buff *skb,
+				       u16 local_port, void *priv);
+void mlxsw_sp_ptp_receive(struct mlxsw_sp *mlxsw_sp, struct sk_buff *skb,
+			  u16 local_port);
+int mlxsw_sp_port_speed_get(struct mlxsw_sp_port *mlxsw_sp_port, u32 *speed);
+int mlxsw_sp_port_ets_set(struct mlxsw_sp_port *mlxsw_sp_port,
+			  enum mlxsw_reg_qeec_hr hr, u8 index, u8 next_index,
+			  bool dwrr, u8 dwrr_weight);
+int mlxsw_sp_port_prio_tc_set(struct mlxsw_sp_port *mlxsw_sp_port,
+			      u8 switch_prio, u8 tclass);
+int mlxsw_sp_port_ets_maxrate_set(struct mlxsw_sp_port *mlxsw_sp_port,
+				  enum mlxsw_reg_qeec_hr hr, u8 index,
+				  u8 next_index, u32 maxrate, u8 burst_size);
+enum mlxsw_reg_spms_state mlxsw_sp_stp_spms_state(u8 stp_state);
+int mlxsw_sp_port_vid_stp_set(struct mlxsw_sp_port *mlxsw_sp_port, u16 vid,
+			      u8 state);
+int mlxsw_sp_port_vp_mode_set(struct mlxsw_sp_port *mlxsw_sp_port, bool enable);
+int mlxsw_sp_port_vid_learning_set(struct mlxsw_sp_port *mlxsw_sp_port, u16 vid,
+				   bool learn_enable);
+int mlxsw_sp_ethtype_to_sver_type(u16 ethtype, u8 *p_sver_type);
+int mlxsw_sp_port_egress_ethtype_set(struct mlxsw_sp_port *mlxsw_sp_port,
+				     u16 ethtype);
+int mlxsw_sp_port_pvid_set(struct mlxsw_sp_port *mlxsw_sp_port, u16 vid,
+			   u16 ethtype);
+struct mlxsw_sp_port_vlan *
+mlxsw_sp_port_vlan_create(struct mlxsw_sp_port *mlxsw_sp_port, u16 vid);
+void mlxsw_sp_port_vlan_destroy(struct mlxsw_sp_port_vlan *mlxsw_sp_port_vlan);
+int mlxsw_sp_port_vlan_set(struct mlxsw_sp_port *mlxsw_sp_port, u16 vid_begin,
+			   u16 vid_end, bool is_member, bool untagged);
+int mlxsw_sp_flow_counter_get(struct mlxsw_sp *mlxsw_sp,
+			      unsigned int counter_index, u64 *packets,
+			      u64 *bytes);
+int mlxsw_sp_flow_counter_alloc(struct mlxsw_sp *mlxsw_sp,
+				unsigned int *p_counter_index);
+void mlxsw_sp_flow_counter_free(struct mlxsw_sp *mlxsw_sp,
+				unsigned int counter_index);
+void mlxsw_sp_txhdr_construct(struct sk_buff *skb,
+			      const struct mlxsw_tx_info *tx_info);
+int mlxsw_sp_txhdr_ptp_data_construct(struct mlxsw_core *mlxsw_core,
+				      struct mlxsw_sp_port *mlxsw_sp_port,
+				      struct sk_buff *skb,
+				      const struct mlxsw_tx_info *tx_info);
+bool mlxsw_sp_port_dev_check(const struct net_device *dev);
+struct mlxsw_sp *mlxsw_sp_lower_get(struct net_device *dev);
+struct mlxsw_sp_port *mlxsw_sp_port_dev_lower_find(struct net_device *dev);
+struct mlxsw_sp_port *mlxsw_sp_port_lower_dev_hold(struct net_device *dev);
+void mlxsw_sp_port_dev_put(struct mlxsw_sp_port *mlxsw_sp_port);
+struct mlxsw_sp_port *mlxsw_sp_port_dev_lower_find_rcu(struct net_device *dev);
+int mlxsw_sp_parsing_depth_inc(struct mlxsw_sp *mlxsw_sp);
+void mlxsw_sp_parsing_depth_dec(struct mlxsw_sp *mlxsw_sp);
+int mlxsw_sp_parsing_vxlan_udp_dport_set(struct mlxsw_sp *mlxsw_sp,
+					 __be16 udp_dport);
+
+/* spectrum_dcb.c */
+#ifdef CONFIG_MLXSW_SPECTRUM_DCB
+int mlxsw_sp_port_dcb_init(struct mlxsw_sp_port *mlxsw_sp_port);
+void mlxsw_sp_port_dcb_fini(struct mlxsw_sp_port *mlxsw_sp_port);
+#else
+static inline int mlxsw_sp_port_dcb_init(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	return 0;
+}
+static inline void mlxsw_sp_port_dcb_fini(struct mlxsw_sp_port *mlxsw_sp_port)
+{}
+#endif
+
+/* spectrum_router.c */
+enum mlxsw_sp_l3proto {
+	MLXSW_SP_L3_PROTO_IPV4,
+	MLXSW_SP_L3_PROTO_IPV6,
+#define MLXSW_SP_L3_PROTO_MAX	(MLXSW_SP_L3_PROTO_IPV6 + 1)
+};
+
+union mlxsw_sp_l3addr {
+	__be32 addr4;
+	struct in6_addr addr6;
+};
+
+u16 mlxsw_sp_rif_index(const struct mlxsw_sp_rif *rif);
+int mlxsw_sp_router_init(struct mlxsw_sp *mlxsw_sp,
+			 struct netlink_ext_ack *extack);
+void mlxsw_sp_router_fini(struct mlxsw_sp *mlxsw_sp);
+void mlxsw_sp_rif_macvlan_del(struct mlxsw_sp *mlxsw_sp,
+			      const struct net_device *macvlan_dev);
+int mlxsw_sp_inetaddr_valid_event(struct notifier_block *unused,
+				  unsigned long event, void *ptr);
+int mlxsw_sp_inet6addr_valid_event(struct notifier_block *unused,
+				   unsigned long event, void *ptr);
+int
+mlxsw_sp_port_vlan_router_join(struct mlxsw_sp_port_vlan *mlxsw_sp_port_vlan,
+			       struct net_device *l3_dev,
+			       struct netlink_ext_ack *extack);
+void
+mlxsw_sp_port_vlan_router_leave(struct mlxsw_sp_port_vlan *mlxsw_sp_port_vlan);
+void mlxsw_sp_rif_destroy_by_dev(struct mlxsw_sp *mlxsw_sp,
+				 struct net_device *dev);
+bool mlxsw_sp_rif_exists(struct mlxsw_sp *mlxsw_sp,
+			 const struct net_device *dev);
+u16 mlxsw_sp_rif_vid(struct mlxsw_sp *mlxsw_sp, const struct net_device *dev);
+u16 mlxsw_sp_router_port(const struct mlxsw_sp *mlxsw_sp);
+int mlxsw_sp_router_nve_promote_decap(struct mlxsw_sp *mlxsw_sp, u32 ul_tb_id,
+				      enum mlxsw_sp_l3proto ul_proto,
+				      const union mlxsw_sp_l3addr *ul_sip,
+				      u32 tunnel_index);
+void mlxsw_sp_router_nve_demote_decap(struct mlxsw_sp *mlxsw_sp, u32 ul_tb_id,
+				      enum mlxsw_sp_l3proto ul_proto,
+				      const union mlxsw_sp_l3addr *ul_sip);
+int mlxsw_sp_router_tb_id_vr_id(struct mlxsw_sp *mlxsw_sp, u32 tb_id,
+				u16 *vr_id);
+int mlxsw_sp_router_ul_rif_get(struct mlxsw_sp *mlxsw_sp, u32 ul_tb_id,
+			       u16 *ul_rif_index);
+void mlxsw_sp_router_ul_rif_put(struct mlxsw_sp *mlxsw_sp, u16 ul_rif_index);
+
+/* spectrum_kvdl.c */
+enum mlxsw_sp_kvdl_entry_type {
+	MLXSW_SP_KVDL_ENTRY_TYPE_ADJ,
+	MLXSW_SP_KVDL_ENTRY_TYPE_ACTSET,
+	MLXSW_SP_KVDL_ENTRY_TYPE_PBS,
+	MLXSW_SP_KVDL_ENTRY_TYPE_MCRIGR,
+	MLXSW_SP_KVDL_ENTRY_TYPE_IPV6_ADDRESS,
+	MLXSW_SP_KVDL_ENTRY_TYPE_TNUMT,
+};
+
+static inline unsigned int
+mlxsw_sp_kvdl_entry_size(enum mlxsw_sp_kvdl_entry_type type)
+{
+	switch (type) {
+	case MLXSW_SP_KVDL_ENTRY_TYPE_ADJ:
+	case MLXSW_SP_KVDL_ENTRY_TYPE_ACTSET:
+	case MLXSW_SP_KVDL_ENTRY_TYPE_PBS:
+	case MLXSW_SP_KVDL_ENTRY_TYPE_MCRIGR:
+	case MLXSW_SP_KVDL_ENTRY_TYPE_IPV6_ADDRESS:
+	case MLXSW_SP_KVDL_ENTRY_TYPE_TNUMT:
+	default:
+		return 1;
+	}
+}
+
+struct mlxsw_sp_kvdl_ops {
+	size_t priv_size;
+	int (*init)(struct mlxsw_sp *mlxsw_sp, void *priv);
+	void (*fini)(struct mlxsw_sp *mlxsw_sp, void *priv);
+	int (*alloc)(struct mlxsw_sp *mlxsw_sp, void *priv,
+		     enum mlxsw_sp_kvdl_entry_type type,
+		     unsigned int entry_count, u32 *p_entry_index);
+	void (*free)(struct mlxsw_sp *mlxsw_sp, void *priv,
+		     enum mlxsw_sp_kvdl_entry_type type,
+		     unsigned int entry_count, int entry_index);
+	int (*alloc_size_query)(struct mlxsw_sp *mlxsw_sp, void *priv,
+				enum mlxsw_sp_kvdl_entry_type type,
+				unsigned int entry_count,
+				unsigned int *p_alloc_count);
+	int (*resources_register)(struct mlxsw_sp *mlxsw_sp, void *priv);
+};
+
+int mlxsw_sp_kvdl_init(struct mlxsw_sp *mlxsw_sp);
+void mlxsw_sp_kvdl_fini(struct mlxsw_sp *mlxsw_sp);
+int mlxsw_sp_kvdl_alloc(struct mlxsw_sp *mlxsw_sp,
+			enum mlxsw_sp_kvdl_entry_type type,
+			unsigned int entry_count, u32 *p_entry_index);
+void mlxsw_sp_kvdl_free(struct mlxsw_sp *mlxsw_sp,
+			enum mlxsw_sp_kvdl_entry_type type,
+			unsigned int entry_count, int entry_index);
+int mlxsw_sp_kvdl_alloc_count_query(struct mlxsw_sp *mlxsw_sp,
+				    enum mlxsw_sp_kvdl_entry_type type,
+				    unsigned int entry_count,
+				    unsigned int *p_alloc_count);
+
+/* spectrum1_kvdl.c */
+extern const struct mlxsw_sp_kvdl_ops mlxsw_sp1_kvdl_ops;
+int mlxsw_sp1_kvdl_resources_register(struct mlxsw_core *mlxsw_core);
+
+/* spectrum2_kvdl.c */
+extern const struct mlxsw_sp_kvdl_ops mlxsw_sp2_kvdl_ops;
+
+enum mlxsw_sp_acl_mangle_field {
+	MLXSW_SP_ACL_MANGLE_FIELD_IP_DSFIELD,
+	MLXSW_SP_ACL_MANGLE_FIELD_IP_DSCP,
+	MLXSW_SP_ACL_MANGLE_FIELD_IP_ECN,
+	MLXSW_SP_ACL_MANGLE_FIELD_IP_SPORT,
+	MLXSW_SP_ACL_MANGLE_FIELD_IP_DPORT,
+	MLXSW_SP_ACL_MANGLE_FIELD_IP4_SIP,
+	MLXSW_SP_ACL_MANGLE_FIELD_IP4_DIP,
+	MLXSW_SP_ACL_MANGLE_FIELD_IP6_SIP_1,
+	MLXSW_SP_ACL_MANGLE_FIELD_IP6_SIP_2,
+	MLXSW_SP_ACL_MANGLE_FIELD_IP6_SIP_3,
+	MLXSW_SP_ACL_MANGLE_FIELD_IP6_SIP_4,
+	MLXSW_SP_ACL_MANGLE_FIELD_IP6_DIP_1,
+	MLXSW_SP_ACL_MANGLE_FIELD_IP6_DIP_2,
+	MLXSW_SP_ACL_MANGLE_FIELD_IP6_DIP_3,
+	MLXSW_SP_ACL_MANGLE_FIELD_IP6_DIP_4,
+};
+
+struct mlxsw_sp_acl_rule_info {
+	unsigned int priority;
+	struct mlxsw_afk_element_values values;
+	struct mlxsw_afa_block *act_block;
+	u8 action_created:1,
+	   ingress_bind_blocker:1,
+	   egress_bind_blocker:1,
+	   counter_valid:1,
+	   policer_index_valid:1,
+	   ipv6_valid:1;
+	unsigned int counter_index;
+	u16 policer_index;
+	struct {
+		u32 prev_val;
+		enum mlxsw_sp_acl_mangle_field prev_field;
+	} ipv6;
+};
+
+/* spectrum_flow.c */
+struct mlxsw_sp_flow_block {
+	struct list_head binding_list;
+	struct {
+		struct list_head list;
+		unsigned int min_prio;
+		unsigned int max_prio;
+	} mall;
+	struct mlxsw_sp_acl_ruleset *ruleset_zero;
+	struct mlxsw_sp *mlxsw_sp;
+	unsigned int rule_count;
+	unsigned int disable_count;
+	unsigned int ingress_blocker_rule_count;
+	unsigned int egress_blocker_rule_count;
+	unsigned int ingress_binding_count;
+	unsigned int egress_binding_count;
+	struct net *net;
+};
+
+struct mlxsw_sp_flow_block_binding {
+	struct list_head list;
+	struct mlxsw_sp_port *mlxsw_sp_port;
+	bool ingress;
+};
+
+static inline struct mlxsw_sp *
+mlxsw_sp_flow_block_mlxsw_sp(struct mlxsw_sp_flow_block *block)
+{
+	return block->mlxsw_sp;
+}
+
+static inline unsigned int
+mlxsw_sp_flow_block_rule_count(const struct mlxsw_sp_flow_block *block)
+{
+	return block ? block->rule_count : 0;
+}
+
+static inline void
+mlxsw_sp_flow_block_disable_inc(struct mlxsw_sp_flow_block *block)
+{
+	if (block)
+		block->disable_count++;
+}
+
+static inline void
+mlxsw_sp_flow_block_disable_dec(struct mlxsw_sp_flow_block *block)
+{
+	if (block)
+		block->disable_count--;
+}
+
+static inline bool
+mlxsw_sp_flow_block_disabled(const struct mlxsw_sp_flow_block *block)
+{
+	return block->disable_count;
+}
+
+static inline bool
+mlxsw_sp_flow_block_is_egress_bound(const struct mlxsw_sp_flow_block *block)
+{
+	return block->egress_binding_count;
+}
+
+static inline bool
+mlxsw_sp_flow_block_is_ingress_bound(const struct mlxsw_sp_flow_block *block)
+{
+	return block->ingress_binding_count;
+}
+
+static inline bool
+mlxsw_sp_flow_block_is_mixed_bound(const struct mlxsw_sp_flow_block *block)
+{
+	return block->ingress_binding_count && block->egress_binding_count;
+}
+
+struct mlxsw_sp_flow_block *mlxsw_sp_flow_block_create(struct mlxsw_sp *mlxsw_sp,
+						       struct net *net);
+void mlxsw_sp_flow_block_destroy(struct mlxsw_sp_flow_block *block);
+int mlxsw_sp_setup_tc_block_clsact(struct mlxsw_sp_port *mlxsw_sp_port,
+				   struct flow_block_offload *f,
+				   bool ingress);
+
+/* spectrum_acl.c */
+struct mlxsw_sp_acl_ruleset;
+
+enum mlxsw_sp_acl_profile {
+	MLXSW_SP_ACL_PROFILE_FLOWER,
+	MLXSW_SP_ACL_PROFILE_MR,
+};
+
+struct mlxsw_afk *mlxsw_sp_acl_afk(struct mlxsw_sp_acl *acl);
+
+int mlxsw_sp_acl_ruleset_bind(struct mlxsw_sp *mlxsw_sp,
+			      struct mlxsw_sp_flow_block *block,
+			      struct mlxsw_sp_flow_block_binding *binding);
+void mlxsw_sp_acl_ruleset_unbind(struct mlxsw_sp *mlxsw_sp,
+				 struct mlxsw_sp_flow_block *block,
+				 struct mlxsw_sp_flow_block_binding *binding);
+struct mlxsw_sp_acl_ruleset *
+mlxsw_sp_acl_ruleset_lookup(struct mlxsw_sp *mlxsw_sp,
+			    struct mlxsw_sp_flow_block *block, u32 chain_index,
+			    enum mlxsw_sp_acl_profile profile);
+struct mlxsw_sp_acl_ruleset *
+mlxsw_sp_acl_ruleset_get(struct mlxsw_sp *mlxsw_sp,
+			 struct mlxsw_sp_flow_block *block, u32 chain_index,
+			 enum mlxsw_sp_acl_profile profile,
+			 struct mlxsw_afk_element_usage *tmplt_elusage);
+void mlxsw_sp_acl_ruleset_put(struct mlxsw_sp *mlxsw_sp,
+			      struct mlxsw_sp_acl_ruleset *ruleset);
+u16 mlxsw_sp_acl_ruleset_group_id(struct mlxsw_sp_acl_ruleset *ruleset);
+void mlxsw_sp_acl_ruleset_prio_get(struct mlxsw_sp_acl_ruleset *ruleset,
+				   unsigned int *p_min_prio,
+				   unsigned int *p_max_prio);
+
+struct mlxsw_sp_acl_rule_info *
+mlxsw_sp_acl_rulei_create(struct mlxsw_sp_acl *acl,
+			  struct mlxsw_afa_block *afa_block);
+void mlxsw_sp_acl_rulei_destroy(struct mlxsw_sp_acl_rule_info *rulei);
+int mlxsw_sp_acl_rulei_commit(struct mlxsw_sp_acl_rule_info *rulei);
+void mlxsw_sp_acl_rulei_priority(struct mlxsw_sp_acl_rule_info *rulei,
+				 unsigned int priority);
+void mlxsw_sp_acl_rulei_keymask_u32(struct mlxsw_sp_acl_rule_info *rulei,
+				    enum mlxsw_afk_element element,
+				    u32 key_value, u32 mask_value);
+void mlxsw_sp_acl_rulei_keymask_buf(struct mlxsw_sp_acl_rule_info *rulei,
+				    enum mlxsw_afk_element element,
+				    const char *key_value,
+				    const char *mask_value, unsigned int len);
+int mlxsw_sp_acl_rulei_act_continue(struct mlxsw_sp_acl_rule_info *rulei);
+int mlxsw_sp_acl_rulei_act_jump(struct mlxsw_sp_acl_rule_info *rulei,
+				u16 group_id);
+int mlxsw_sp_acl_rulei_act_terminate(struct mlxsw_sp_acl_rule_info *rulei);
+int mlxsw_sp_acl_rulei_act_drop(struct mlxsw_sp_acl_rule_info *rulei,
+				bool ingress,
+				const struct flow_action_cookie *fa_cookie,
+				struct netlink_ext_ack *extack);
+int mlxsw_sp_acl_rulei_act_trap(struct mlxsw_sp_acl_rule_info *rulei);
+int mlxsw_sp_acl_rulei_act_mirror(struct mlxsw_sp *mlxsw_sp,
+				  struct mlxsw_sp_acl_rule_info *rulei,
+				  struct mlxsw_sp_flow_block *block,
+				  struct net_device *out_dev,
+				  struct netlink_ext_ack *extack);
+int mlxsw_sp_acl_rulei_act_fwd(struct mlxsw_sp *mlxsw_sp,
+			       struct mlxsw_sp_acl_rule_info *rulei,
+			       struct net_device *out_dev,
+			       struct netlink_ext_ack *extack);
+int mlxsw_sp_acl_rulei_act_vlan(struct mlxsw_sp *mlxsw_sp,
+				struct mlxsw_sp_acl_rule_info *rulei,
+				u32 action, u16 vid, u16 proto, u8 prio,
+				struct netlink_ext_ack *extack);
+int mlxsw_sp_acl_rulei_act_priority(struct mlxsw_sp *mlxsw_sp,
+				    struct mlxsw_sp_acl_rule_info *rulei,
+				    u32 prio, struct netlink_ext_ack *extack);
+int mlxsw_sp_acl_rulei_act_mangle(struct mlxsw_sp *mlxsw_sp,
+				  struct mlxsw_sp_acl_rule_info *rulei,
+				  enum flow_action_mangle_base htype,
+				  u32 offset, u32 mask, u32 val,
+				  struct netlink_ext_ack *extack);
+int mlxsw_sp_acl_rulei_act_police(struct mlxsw_sp *mlxsw_sp,
+				  struct mlxsw_sp_acl_rule_info *rulei,
+				  u32 index, u64 rate_bytes_ps,
+				  u32 burst, struct netlink_ext_ack *extack);
+int mlxsw_sp_acl_rulei_act_count(struct mlxsw_sp *mlxsw_sp,
+				 struct mlxsw_sp_acl_rule_info *rulei,
+				 struct netlink_ext_ack *extack);
+int mlxsw_sp_acl_rulei_act_fid_set(struct mlxsw_sp *mlxsw_sp,
+				   struct mlxsw_sp_acl_rule_info *rulei,
+				   u16 fid, struct netlink_ext_ack *extack);
+int mlxsw_sp_acl_rulei_act_sample(struct mlxsw_sp *mlxsw_sp,
+				  struct mlxsw_sp_acl_rule_info *rulei,
+				  struct mlxsw_sp_flow_block *block,
+				  struct psample_group *psample_group, u32 rate,
+				  u32 trunc_size, bool truncate,
+				  struct netlink_ext_ack *extack);
+
+struct mlxsw_sp_acl_rule;
+
+struct mlxsw_sp_acl_rule *
+mlxsw_sp_acl_rule_create(struct mlxsw_sp *mlxsw_sp,
+			 struct mlxsw_sp_acl_ruleset *ruleset,
+			 unsigned long cookie,
+			 struct mlxsw_afa_block *afa_block,
+			 struct netlink_ext_ack *extack);
+void mlxsw_sp_acl_rule_destroy(struct mlxsw_sp *mlxsw_sp,
+			       struct mlxsw_sp_acl_rule *rule);
+int mlxsw_sp_acl_rule_add(struct mlxsw_sp *mlxsw_sp,
+			  struct mlxsw_sp_acl_rule *rule);
+void mlxsw_sp_acl_rule_del(struct mlxsw_sp *mlxsw_sp,
+			   struct mlxsw_sp_acl_rule *rule);
+int mlxsw_sp_acl_rule_action_replace(struct mlxsw_sp *mlxsw_sp,
+				     struct mlxsw_sp_acl_rule *rule,
+				     struct mlxsw_afa_block *afa_block);
+struct mlxsw_sp_acl_rule *
+mlxsw_sp_acl_rule_lookup(struct mlxsw_sp *mlxsw_sp,
+			 struct mlxsw_sp_acl_ruleset *ruleset,
+			 unsigned long cookie);
+struct mlxsw_sp_acl_rule_info *
+mlxsw_sp_acl_rule_rulei(struct mlxsw_sp_acl_rule *rule);
+int mlxsw_sp_acl_rule_get_stats(struct mlxsw_sp *mlxsw_sp,
+				struct mlxsw_sp_acl_rule *rule,
+				u64 *packets, u64 *bytes, u64 *drops,
+				u64 *last_use,
+				enum flow_action_hw_stats *used_hw_stats);
+
+struct mlxsw_sp_fid *mlxsw_sp_acl_dummy_fid(struct mlxsw_sp *mlxsw_sp);
+
+static inline const struct flow_action_cookie *
+mlxsw_sp_acl_act_cookie_lookup(struct mlxsw_sp *mlxsw_sp, u32 cookie_index)
+{
+	return mlxsw_afa_cookie_lookup(mlxsw_sp->afa, cookie_index);
+}
+
+int mlxsw_sp_acl_init(struct mlxsw_sp *mlxsw_sp);
+void mlxsw_sp_acl_fini(struct mlxsw_sp *mlxsw_sp);
+u32 mlxsw_sp_acl_region_rehash_intrvl_get(struct mlxsw_sp *mlxsw_sp);
+int mlxsw_sp_acl_region_rehash_intrvl_set(struct mlxsw_sp *mlxsw_sp, u32 val);
+
+struct mlxsw_sp_acl_mangle_action;
+
+struct mlxsw_sp_acl_rulei_ops {
+	int (*act_mangle_field)(struct mlxsw_sp *mlxsw_sp, struct mlxsw_sp_acl_rule_info *rulei,
+				struct mlxsw_sp_acl_mangle_action *mact, u32 val,
+				struct netlink_ext_ack *extack);
+};
+
+extern struct mlxsw_sp_acl_rulei_ops mlxsw_sp1_acl_rulei_ops;
+extern struct mlxsw_sp_acl_rulei_ops mlxsw_sp2_acl_rulei_ops;
+
+/* spectrum_acl_tcam.c */
+struct mlxsw_sp_acl_tcam;
+struct mlxsw_sp_acl_tcam_region;
+
+struct mlxsw_sp_acl_tcam_ops {
+	enum mlxsw_reg_ptar_key_type key_type;
+	size_t priv_size;
+	int (*init)(struct mlxsw_sp *mlxsw_sp, void *priv,
+		    struct mlxsw_sp_acl_tcam *tcam);
+	void (*fini)(struct mlxsw_sp *mlxsw_sp, void *priv);
+	size_t region_priv_size;
+	int (*region_init)(struct mlxsw_sp *mlxsw_sp, void *region_priv,
+			   void *tcam_priv,
+			   struct mlxsw_sp_acl_tcam_region *region,
+			   void *hints_priv);
+	void (*region_fini)(struct mlxsw_sp *mlxsw_sp, void *region_priv);
+	int (*region_associate)(struct mlxsw_sp *mlxsw_sp,
+				struct mlxsw_sp_acl_tcam_region *region);
+	void * (*region_rehash_hints_get)(void *region_priv);
+	void (*region_rehash_hints_put)(void *hints_priv);
+	size_t chunk_priv_size;
+	void (*chunk_init)(void *region_priv, void *chunk_priv,
+			   unsigned int priority);
+	void (*chunk_fini)(void *chunk_priv);
+	size_t entry_priv_size;
+	int (*entry_add)(struct mlxsw_sp *mlxsw_sp,
+			 void *region_priv, void *chunk_priv,
+			 void *entry_priv,
+			 struct mlxsw_sp_acl_rule_info *rulei);
+	void (*entry_del)(struct mlxsw_sp *mlxsw_sp,
+			  void *region_priv, void *chunk_priv,
+			  void *entry_priv);
+	int (*entry_action_replace)(struct mlxsw_sp *mlxsw_sp,
+				    void *region_priv, void *entry_priv,
+				    struct mlxsw_sp_acl_rule_info *rulei);
+	int (*entry_activity_get)(struct mlxsw_sp *mlxsw_sp,
+				  void *region_priv, void *entry_priv,
+				  bool *activity);
+};
+
+/* spectrum1_acl_tcam.c */
+extern const struct mlxsw_sp_acl_tcam_ops mlxsw_sp1_acl_tcam_ops;
+
+/* spectrum2_acl_tcam.c */
+extern const struct mlxsw_sp_acl_tcam_ops mlxsw_sp2_acl_tcam_ops;
+
+/* spectrum_acl_flex_actions.c */
+extern const struct mlxsw_afa_ops mlxsw_sp1_act_afa_ops;
+extern const struct mlxsw_afa_ops mlxsw_sp2_act_afa_ops;
+
+/* spectrum_acl_flex_keys.c */
+extern const struct mlxsw_afk_ops mlxsw_sp1_afk_ops;
+extern const struct mlxsw_afk_ops mlxsw_sp2_afk_ops;
+extern const struct mlxsw_afk_ops mlxsw_sp4_afk_ops;
+
+/* spectrum_acl_bloom_filter.c */
+extern const struct mlxsw_sp_acl_bf_ops mlxsw_sp2_acl_bf_ops;
+extern const struct mlxsw_sp_acl_bf_ops mlxsw_sp4_acl_bf_ops;
+
+/* spectrum_matchall.c */
+struct mlxsw_sp_mall_ops {
+	int (*sample_add)(struct mlxsw_sp *mlxsw_sp,
+			  struct mlxsw_sp_port *mlxsw_sp_port,
+			  struct mlxsw_sp_mall_entry *mall_entry,
+			  struct netlink_ext_ack *extack);
+	void (*sample_del)(struct mlxsw_sp *mlxsw_sp,
+			   struct mlxsw_sp_port *mlxsw_sp_port,
+			   struct mlxsw_sp_mall_entry *mall_entry);
+};
+
+extern const struct mlxsw_sp_mall_ops mlxsw_sp1_mall_ops;
+extern const struct mlxsw_sp_mall_ops mlxsw_sp2_mall_ops;
+
+enum mlxsw_sp_mall_action_type {
+	MLXSW_SP_MALL_ACTION_TYPE_MIRROR,
+	MLXSW_SP_MALL_ACTION_TYPE_SAMPLE,
+	MLXSW_SP_MALL_ACTION_TYPE_TRAP,
+};
+
+struct mlxsw_sp_mall_mirror_entry {
+	const struct net_device *to_dev;
+	int span_id;
+};
+
+struct mlxsw_sp_mall_trap_entry {
+	int span_id;
+};
+
+struct mlxsw_sp_mall_sample_entry {
+	struct mlxsw_sp_sample_params params;
+	int span_id;	/* Relevant for Spectrum-2 onwards. */
+};
+
+struct mlxsw_sp_mall_entry {
+	struct list_head list;
+	unsigned long cookie;
+	unsigned int priority;
+	enum mlxsw_sp_mall_action_type type;
+	bool ingress;
+	union {
+		struct mlxsw_sp_mall_mirror_entry mirror;
+		struct mlxsw_sp_mall_trap_entry trap;
+		struct mlxsw_sp_mall_sample_entry sample;
+	};
+	struct rcu_head rcu;
+};
+
+int mlxsw_sp_mall_replace(struct mlxsw_sp *mlxsw_sp,
+			  struct mlxsw_sp_flow_block *block,
+			  struct tc_cls_matchall_offload *f);
+void mlxsw_sp_mall_destroy(struct mlxsw_sp_flow_block *block,
+			   struct tc_cls_matchall_offload *f);
+int mlxsw_sp_mall_port_bind(struct mlxsw_sp_flow_block *block,
+			    struct mlxsw_sp_port *mlxsw_sp_port,
+			    struct netlink_ext_ack *extack);
+void mlxsw_sp_mall_port_unbind(struct mlxsw_sp_flow_block *block,
+			       struct mlxsw_sp_port *mlxsw_sp_port);
+int mlxsw_sp_mall_prio_get(struct mlxsw_sp_flow_block *block, u32 chain_index,
+			   unsigned int *p_min_prio, unsigned int *p_max_prio);
+
+/* spectrum_flower.c */
+int mlxsw_sp_flower_replace(struct mlxsw_sp *mlxsw_sp,
+			    struct mlxsw_sp_flow_block *block,
+			    struct flow_cls_offload *f);
+void mlxsw_sp_flower_destroy(struct mlxsw_sp *mlxsw_sp,
+			     struct mlxsw_sp_flow_block *block,
+			     struct flow_cls_offload *f);
+int mlxsw_sp_flower_stats(struct mlxsw_sp *mlxsw_sp,
+			  struct mlxsw_sp_flow_block *block,
+			  struct flow_cls_offload *f);
+int mlxsw_sp_flower_tmplt_create(struct mlxsw_sp *mlxsw_sp,
+				 struct mlxsw_sp_flow_block *block,
+				 struct flow_cls_offload *f);
+void mlxsw_sp_flower_tmplt_destroy(struct mlxsw_sp *mlxsw_sp,
+				   struct mlxsw_sp_flow_block *block,
+				   struct flow_cls_offload *f);
+int mlxsw_sp_flower_prio_get(struct mlxsw_sp *mlxsw_sp,
+			     struct mlxsw_sp_flow_block *block,
+			     u32 chain_index, unsigned int *p_min_prio,
+			     unsigned int *p_max_prio);
+
+/* spectrum_qdisc.c */
+int mlxsw_sp_tc_qdisc_init(struct mlxsw_sp_port *mlxsw_sp_port);
+void mlxsw_sp_tc_qdisc_fini(struct mlxsw_sp_port *mlxsw_sp_port);
+int mlxsw_sp_setup_tc_red(struct mlxsw_sp_port *mlxsw_sp_port,
+			  struct tc_red_qopt_offload *p);
+int mlxsw_sp_setup_tc_prio(struct mlxsw_sp_port *mlxsw_sp_port,
+			   struct tc_prio_qopt_offload *p);
+int mlxsw_sp_setup_tc_ets(struct mlxsw_sp_port *mlxsw_sp_port,
+			  struct tc_ets_qopt_offload *p);
+int mlxsw_sp_setup_tc_tbf(struct mlxsw_sp_port *mlxsw_sp_port,
+			  struct tc_tbf_qopt_offload *p);
+int mlxsw_sp_setup_tc_fifo(struct mlxsw_sp_port *mlxsw_sp_port,
+			   struct tc_fifo_qopt_offload *p);
+int mlxsw_sp_setup_tc_block_qevent_early_drop(struct mlxsw_sp_port *mlxsw_sp_port,
+					      struct flow_block_offload *f);
+int mlxsw_sp_setup_tc_block_qevent_mark(struct mlxsw_sp_port *mlxsw_sp_port,
+					struct flow_block_offload *f);
+
+/* spectrum_fid.c */
+bool mlxsw_sp_fid_is_dummy(struct mlxsw_sp *mlxsw_sp, u16 fid_index);
+struct mlxsw_sp_fid *mlxsw_sp_fid_lookup_by_index(struct mlxsw_sp *mlxsw_sp,
+						  u16 fid_index);
+int mlxsw_sp_fid_nve_ifindex(const struct mlxsw_sp_fid *fid, int *nve_ifindex);
+int mlxsw_sp_fid_nve_type(const struct mlxsw_sp_fid *fid,
+			  enum mlxsw_sp_nve_type *p_type);
+struct mlxsw_sp_fid *mlxsw_sp_fid_lookup_by_vni(struct mlxsw_sp *mlxsw_sp,
+						__be32 vni);
+int mlxsw_sp_fid_vni(const struct mlxsw_sp_fid *fid, __be32 *vni);
+int mlxsw_sp_fid_nve_flood_index_set(struct mlxsw_sp_fid *fid,
+				     u32 nve_flood_index);
+void mlxsw_sp_fid_nve_flood_index_clear(struct mlxsw_sp_fid *fid);
+bool mlxsw_sp_fid_nve_flood_index_is_set(const struct mlxsw_sp_fid *fid);
+int mlxsw_sp_fid_vni_set(struct mlxsw_sp_fid *fid, enum mlxsw_sp_nve_type type,
+			 __be32 vni, int nve_ifindex);
+void mlxsw_sp_fid_vni_clear(struct mlxsw_sp_fid *fid);
+bool mlxsw_sp_fid_vni_is_set(const struct mlxsw_sp_fid *fid);
+void mlxsw_sp_fid_fdb_clear_offload(const struct mlxsw_sp_fid *fid,
+				    const struct net_device *nve_dev);
+int mlxsw_sp_fid_flood_set(struct mlxsw_sp_fid *fid,
+			   enum mlxsw_sp_flood_type packet_type, u16 local_port,
+			   bool member);
+int mlxsw_sp_fid_port_vid_map(struct mlxsw_sp_fid *fid,
+			      struct mlxsw_sp_port *mlxsw_sp_port, u16 vid);
+void mlxsw_sp_fid_port_vid_unmap(struct mlxsw_sp_fid *fid,
+				 struct mlxsw_sp_port *mlxsw_sp_port, u16 vid);
+u16 mlxsw_sp_fid_index(const struct mlxsw_sp_fid *fid);
+enum mlxsw_sp_fid_type mlxsw_sp_fid_type(const struct mlxsw_sp_fid *fid);
+int mlxsw_sp_fid_rif_set(struct mlxsw_sp_fid *fid, struct mlxsw_sp_rif *rif);
+void mlxsw_sp_fid_rif_unset(struct mlxsw_sp_fid *fid);
+struct mlxsw_sp_rif *mlxsw_sp_fid_rif(const struct mlxsw_sp_fid *fid);
+enum mlxsw_sp_rif_type
+mlxsw_sp_fid_type_rif_type(const struct mlxsw_sp *mlxsw_sp,
+			   enum mlxsw_sp_fid_type type);
+u16 mlxsw_sp_fid_8021q_vid(const struct mlxsw_sp_fid *fid);
+struct mlxsw_sp_fid *mlxsw_sp_fid_8021q_get(struct mlxsw_sp *mlxsw_sp, u16 vid);
+struct mlxsw_sp_fid *mlxsw_sp_fid_8021d_get(struct mlxsw_sp *mlxsw_sp,
+					    int br_ifindex);
+struct mlxsw_sp_fid *mlxsw_sp_fid_8021q_lookup(struct mlxsw_sp *mlxsw_sp,
+					       u16 vid);
+struct mlxsw_sp_fid *mlxsw_sp_fid_8021d_lookup(struct mlxsw_sp *mlxsw_sp,
+					       int br_ifindex);
+struct mlxsw_sp_fid *mlxsw_sp_fid_rfid_get(struct mlxsw_sp *mlxsw_sp,
+					   u16 rif_index);
+struct mlxsw_sp_fid *mlxsw_sp_fid_dummy_get(struct mlxsw_sp *mlxsw_sp);
+void mlxsw_sp_fid_put(struct mlxsw_sp_fid *fid);
+int mlxsw_sp_port_fids_init(struct mlxsw_sp_port *mlxsw_sp_port);
+void mlxsw_sp_port_fids_fini(struct mlxsw_sp_port *mlxsw_sp_port);
+int mlxsw_sp_fids_init(struct mlxsw_sp *mlxsw_sp);
+void mlxsw_sp_fids_fini(struct mlxsw_sp *mlxsw_sp);
+
+extern const struct mlxsw_sp_fid_family *mlxsw_sp1_fid_family_arr[];
+extern const struct mlxsw_sp_fid_family *mlxsw_sp2_fid_family_arr[];
+
+/* spectrum_mr.c */
+enum mlxsw_sp_mr_route_prio {
+	MLXSW_SP_MR_ROUTE_PRIO_SG,
+	MLXSW_SP_MR_ROUTE_PRIO_STARG,
+	MLXSW_SP_MR_ROUTE_PRIO_CATCHALL,
+	__MLXSW_SP_MR_ROUTE_PRIO_MAX
+};
+
+#define MLXSW_SP_MR_ROUTE_PRIO_MAX (__MLXSW_SP_MR_ROUTE_PRIO_MAX - 1)
+
+struct mlxsw_sp_mr_route_key;
+
+struct mlxsw_sp_mr_tcam_ops {
+	size_t priv_size;
+	int (*init)(struct mlxsw_sp *mlxsw_sp, void *priv);
+	void (*fini)(void *priv);
+	size_t route_priv_size;
+	int (*route_create)(struct mlxsw_sp *mlxsw_sp, void *priv,
+			    void *route_priv,
+			    struct mlxsw_sp_mr_route_key *key,
+			    struct mlxsw_afa_block *afa_block,
+			    enum mlxsw_sp_mr_route_prio prio);
+	void (*route_destroy)(struct mlxsw_sp *mlxsw_sp, void *priv,
+			      void *route_priv,
+			      struct mlxsw_sp_mr_route_key *key);
+	int (*route_update)(struct mlxsw_sp *mlxsw_sp, void *route_priv,
+			    struct mlxsw_sp_mr_route_key *key,
+			    struct mlxsw_afa_block *afa_block);
+};
+
+/* spectrum1_mr_tcam.c */
+extern const struct mlxsw_sp_mr_tcam_ops mlxsw_sp1_mr_tcam_ops;
+
+/* spectrum2_mr_tcam.c */
+extern const struct mlxsw_sp_mr_tcam_ops mlxsw_sp2_mr_tcam_ops;
+
+/* spectrum_nve.c */
+struct mlxsw_sp_nve_params {
+	enum mlxsw_sp_nve_type type;
+	__be32 vni;
+	const struct net_device *dev;
+	u16 ethertype;
+};
+
+extern const struct mlxsw_sp_nve_ops *mlxsw_sp1_nve_ops_arr[];
+extern const struct mlxsw_sp_nve_ops *mlxsw_sp2_nve_ops_arr[];
+
+int mlxsw_sp_nve_learned_ip_resolve(struct mlxsw_sp *mlxsw_sp, u32 uip,
+				    enum mlxsw_sp_l3proto proto,
+				    union mlxsw_sp_l3addr *addr);
+int mlxsw_sp_nve_flood_ip_add(struct mlxsw_sp *mlxsw_sp,
+			      struct mlxsw_sp_fid *fid,
+			      enum mlxsw_sp_l3proto proto,
+			      union mlxsw_sp_l3addr *addr);
+void mlxsw_sp_nve_flood_ip_del(struct mlxsw_sp *mlxsw_sp,
+			       struct mlxsw_sp_fid *fid,
+			       enum mlxsw_sp_l3proto proto,
+			       union mlxsw_sp_l3addr *addr);
+int mlxsw_sp_nve_ipv6_addr_kvdl_set(struct mlxsw_sp *mlxsw_sp,
+				    const struct in6_addr *addr6,
+				    u32 *p_kvdl_index);
+void mlxsw_sp_nve_ipv6_addr_kvdl_unset(struct mlxsw_sp *mlxsw_sp,
+				       const struct in6_addr *addr6);
+int
+mlxsw_sp_nve_ipv6_addr_map_replace(struct mlxsw_sp *mlxsw_sp, const char *mac,
+				   u16 fid_index,
+				   const struct in6_addr *new_addr6);
+void mlxsw_sp_nve_ipv6_addr_map_del(struct mlxsw_sp *mlxsw_sp, const char *mac,
+				    u16 fid_index);
+int mlxsw_sp_nve_fid_enable(struct mlxsw_sp *mlxsw_sp, struct mlxsw_sp_fid *fid,
+			    struct mlxsw_sp_nve_params *params,
+			    struct netlink_ext_ack *extack);
+void mlxsw_sp_nve_fid_disable(struct mlxsw_sp *mlxsw_sp,
+			      struct mlxsw_sp_fid *fid);
+int mlxsw_sp_port_nve_init(struct mlxsw_sp_port *mlxsw_sp_port);
+void mlxsw_sp_port_nve_fini(struct mlxsw_sp_port *mlxsw_sp_port);
+int mlxsw_sp_nve_init(struct mlxsw_sp *mlxsw_sp);
+void mlxsw_sp_nve_fini(struct mlxsw_sp *mlxsw_sp);
+
+/* spectrum_nve_vxlan.c */
+int mlxsw_sp_nve_inc_parsing_depth_get(struct mlxsw_sp *mlxsw_sp);
+void mlxsw_sp_nve_inc_parsing_depth_put(struct mlxsw_sp *mlxsw_sp);
+
+/* spectrum_trap.c */
+int mlxsw_sp_devlink_traps_init(struct mlxsw_sp *mlxsw_sp);
+void mlxsw_sp_devlink_traps_fini(struct mlxsw_sp *mlxsw_sp);
+int mlxsw_sp_trap_init(struct mlxsw_core *mlxsw_core,
+		       const struct devlink_trap *trap, void *trap_ctx);
+void mlxsw_sp_trap_fini(struct mlxsw_core *mlxsw_core,
+			const struct devlink_trap *trap, void *trap_ctx);
+int mlxsw_sp_trap_action_set(struct mlxsw_core *mlxsw_core,
+			     const struct devlink_trap *trap,
+			     enum devlink_trap_action action,
+			     struct netlink_ext_ack *extack);
+int mlxsw_sp_trap_group_init(struct mlxsw_core *mlxsw_core,
+			     const struct devlink_trap_group *group);
+int mlxsw_sp_trap_group_set(struct mlxsw_core *mlxsw_core,
+			    const struct devlink_trap_group *group,
+			    const struct devlink_trap_policer *policer,
+			    struct netlink_ext_ack *extack);
+int
+mlxsw_sp_trap_policer_init(struct mlxsw_core *mlxsw_core,
+			   const struct devlink_trap_policer *policer);
+void mlxsw_sp_trap_policer_fini(struct mlxsw_core *mlxsw_core,
+				const struct devlink_trap_policer *policer);
+int
+mlxsw_sp_trap_policer_set(struct mlxsw_core *mlxsw_core,
+			  const struct devlink_trap_policer *policer,
+			  u64 rate, u64 burst, struct netlink_ext_ack *extack);
+int
+mlxsw_sp_trap_policer_counter_get(struct mlxsw_core *mlxsw_core,
+				  const struct devlink_trap_policer *policer,
+				  u64 *p_drops);
+int mlxsw_sp_trap_group_policer_hw_id_get(struct mlxsw_sp *mlxsw_sp, u16 id,
+					  bool *p_enabled, u16 *p_hw_id);
+
+static inline struct net *mlxsw_sp_net(struct mlxsw_sp *mlxsw_sp)
+{
+	return mlxsw_core_net(mlxsw_sp->core);
+}
+
+/* spectrum_ethtool.c */
+extern const struct ethtool_ops mlxsw_sp_port_ethtool_ops;
+extern const struct mlxsw_sp_port_type_speed_ops mlxsw_sp1_port_type_speed_ops;
+extern const struct mlxsw_sp_port_type_speed_ops mlxsw_sp2_port_type_speed_ops;
+
+/* spectrum_policer.c */
+extern const struct mlxsw_sp_policer_core_ops mlxsw_sp1_policer_core_ops;
+extern const struct mlxsw_sp_policer_core_ops mlxsw_sp2_policer_core_ops;
+
+enum mlxsw_sp_policer_type {
+	MLXSW_SP_POLICER_TYPE_SINGLE_RATE,
+
+	__MLXSW_SP_POLICER_TYPE_MAX,
+	MLXSW_SP_POLICER_TYPE_MAX = __MLXSW_SP_POLICER_TYPE_MAX - 1,
+};
+
+struct mlxsw_sp_policer_params {
+	u64 rate;
+	u64 burst;
+	bool bytes;
+};
+
+int mlxsw_sp_policer_add(struct mlxsw_sp *mlxsw_sp,
+			 enum mlxsw_sp_policer_type type,
+			 const struct mlxsw_sp_policer_params *params,
+			 struct netlink_ext_ack *extack, u16 *p_policer_index);
+void mlxsw_sp_policer_del(struct mlxsw_sp *mlxsw_sp,
+			  enum mlxsw_sp_policer_type type,
+			  u16 policer_index);
+int mlxsw_sp_policer_drops_counter_get(struct mlxsw_sp *mlxsw_sp,
+				       enum mlxsw_sp_policer_type type,
+				       u16 policer_index, u64 *p_drops);
+int mlxsw_sp_policers_init(struct mlxsw_sp *mlxsw_sp);
+void mlxsw_sp_policers_fini(struct mlxsw_sp *mlxsw_sp);
+int mlxsw_sp_policer_resources_register(struct mlxsw_core *mlxsw_core);
+
+/* spectrum_pgt.c */
+int mlxsw_sp_pgt_mid_alloc(struct mlxsw_sp *mlxsw_sp, u16 *p_mid);
+void mlxsw_sp_pgt_mid_free(struct mlxsw_sp *mlxsw_sp, u16 mid_base);
+int mlxsw_sp_pgt_mid_alloc_range(struct mlxsw_sp *mlxsw_sp, u16 mid_base,
+				 u16 count);
+void mlxsw_sp_pgt_mid_free_range(struct mlxsw_sp *mlxsw_sp, u16 mid_base,
+				 u16 count);
+int mlxsw_sp_pgt_entry_port_set(struct mlxsw_sp *mlxsw_sp, u16 mid,
+				u16 smpe, u16 local_port, bool member);
+int mlxsw_sp_pgt_init(struct mlxsw_sp *mlxsw_sp);
+void mlxsw_sp_pgt_fini(struct mlxsw_sp *mlxsw_sp);
+
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum1_acl_tcam.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum1_acl_tcam.c
new file mode 100644
index 000000000..3a636f753
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum1_acl_tcam.c
@@ -0,0 +1,254 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2017-2018 Mellanox Technologies. All rights reserved */
+
+#include <linux/kernel.h>
+#include <linux/slab.h>
+
+#include "reg.h"
+#include "core.h"
+#include "spectrum.h"
+#include "spectrum_acl_tcam.h"
+
+struct mlxsw_sp1_acl_tcam_region {
+	struct mlxsw_sp_acl_ctcam_region cregion;
+	struct mlxsw_sp_acl_tcam_region *region;
+	struct {
+		struct mlxsw_sp_acl_ctcam_chunk cchunk;
+		struct mlxsw_sp_acl_ctcam_entry centry;
+		struct mlxsw_sp_acl_rule_info *rulei;
+	} catchall;
+};
+
+struct mlxsw_sp1_acl_tcam_chunk {
+	struct mlxsw_sp_acl_ctcam_chunk cchunk;
+};
+
+struct mlxsw_sp1_acl_tcam_entry {
+	struct mlxsw_sp_acl_ctcam_entry centry;
+};
+
+static int
+mlxsw_sp1_acl_ctcam_region_entry_insert(struct mlxsw_sp_acl_ctcam_region *cregion,
+					struct mlxsw_sp_acl_ctcam_entry *centry,
+					const char *mask)
+{
+	return 0;
+}
+
+static void
+mlxsw_sp1_acl_ctcam_region_entry_remove(struct mlxsw_sp_acl_ctcam_region *cregion,
+					struct mlxsw_sp_acl_ctcam_entry *centry)
+{
+}
+
+static const struct mlxsw_sp_acl_ctcam_region_ops
+mlxsw_sp1_acl_ctcam_region_ops = {
+	.entry_insert = mlxsw_sp1_acl_ctcam_region_entry_insert,
+	.entry_remove = mlxsw_sp1_acl_ctcam_region_entry_remove,
+};
+
+static int mlxsw_sp1_acl_tcam_init(struct mlxsw_sp *mlxsw_sp, void *priv,
+				   struct mlxsw_sp_acl_tcam *tcam)
+{
+	return 0;
+}
+
+static void mlxsw_sp1_acl_tcam_fini(struct mlxsw_sp *mlxsw_sp, void *priv)
+{
+}
+
+static int
+mlxsw_sp1_acl_ctcam_region_catchall_add(struct mlxsw_sp *mlxsw_sp,
+					struct mlxsw_sp1_acl_tcam_region *region)
+{
+	struct mlxsw_sp_acl_rule_info *rulei;
+	int err;
+
+	mlxsw_sp_acl_ctcam_chunk_init(&region->cregion,
+				      &region->catchall.cchunk,
+				      MLXSW_SP_ACL_TCAM_CATCHALL_PRIO);
+	rulei = mlxsw_sp_acl_rulei_create(mlxsw_sp->acl, NULL);
+	if (IS_ERR(rulei)) {
+		err = PTR_ERR(rulei);
+		goto err_rulei_create;
+	}
+	err = mlxsw_sp_acl_rulei_act_continue(rulei);
+	if (WARN_ON(err))
+		goto err_rulei_act_continue;
+	err = mlxsw_sp_acl_rulei_commit(rulei);
+	if (err)
+		goto err_rulei_commit;
+	err = mlxsw_sp_acl_ctcam_entry_add(mlxsw_sp, &region->cregion,
+					   &region->catchall.cchunk,
+					   &region->catchall.centry,
+					   rulei, false);
+	if (err)
+		goto err_entry_add;
+	region->catchall.rulei = rulei;
+	return 0;
+
+err_entry_add:
+err_rulei_commit:
+err_rulei_act_continue:
+	mlxsw_sp_acl_rulei_destroy(rulei);
+err_rulei_create:
+	mlxsw_sp_acl_ctcam_chunk_fini(&region->catchall.cchunk);
+	return err;
+}
+
+static void
+mlxsw_sp1_acl_ctcam_region_catchall_del(struct mlxsw_sp *mlxsw_sp,
+					struct mlxsw_sp1_acl_tcam_region *region)
+{
+	struct mlxsw_sp_acl_rule_info *rulei = region->catchall.rulei;
+
+	mlxsw_sp_acl_ctcam_entry_del(mlxsw_sp, &region->cregion,
+				     &region->catchall.cchunk,
+				     &region->catchall.centry);
+	mlxsw_sp_acl_rulei_destroy(rulei);
+	mlxsw_sp_acl_ctcam_chunk_fini(&region->catchall.cchunk);
+}
+
+static int
+mlxsw_sp1_acl_tcam_region_init(struct mlxsw_sp *mlxsw_sp, void *region_priv,
+			       void *tcam_priv,
+			       struct mlxsw_sp_acl_tcam_region *_region,
+			       void *hints_priv)
+{
+	struct mlxsw_sp1_acl_tcam_region *region = region_priv;
+	int err;
+
+	err = mlxsw_sp_acl_ctcam_region_init(mlxsw_sp, &region->cregion,
+					     _region,
+					     &mlxsw_sp1_acl_ctcam_region_ops);
+	if (err)
+		return err;
+	err = mlxsw_sp1_acl_ctcam_region_catchall_add(mlxsw_sp, region);
+	if (err)
+		goto err_catchall_add;
+	region->region = _region;
+	return 0;
+
+err_catchall_add:
+	mlxsw_sp_acl_ctcam_region_fini(&region->cregion);
+	return err;
+}
+
+static void
+mlxsw_sp1_acl_tcam_region_fini(struct mlxsw_sp *mlxsw_sp, void *region_priv)
+{
+	struct mlxsw_sp1_acl_tcam_region *region = region_priv;
+
+	mlxsw_sp1_acl_ctcam_region_catchall_del(mlxsw_sp, region);
+	mlxsw_sp_acl_ctcam_region_fini(&region->cregion);
+}
+
+static int
+mlxsw_sp1_acl_tcam_region_associate(struct mlxsw_sp *mlxsw_sp,
+				    struct mlxsw_sp_acl_tcam_region *region)
+{
+	return 0;
+}
+
+static void mlxsw_sp1_acl_tcam_chunk_init(void *region_priv, void *chunk_priv,
+					  unsigned int priority)
+{
+	struct mlxsw_sp1_acl_tcam_region *region = region_priv;
+	struct mlxsw_sp1_acl_tcam_chunk *chunk = chunk_priv;
+
+	mlxsw_sp_acl_ctcam_chunk_init(&region->cregion, &chunk->cchunk,
+				      priority);
+}
+
+static void mlxsw_sp1_acl_tcam_chunk_fini(void *chunk_priv)
+{
+	struct mlxsw_sp1_acl_tcam_chunk *chunk = chunk_priv;
+
+	mlxsw_sp_acl_ctcam_chunk_fini(&chunk->cchunk);
+}
+
+static int mlxsw_sp1_acl_tcam_entry_add(struct mlxsw_sp *mlxsw_sp,
+					void *region_priv, void *chunk_priv,
+					void *entry_priv,
+					struct mlxsw_sp_acl_rule_info *rulei)
+{
+	struct mlxsw_sp1_acl_tcam_region *region = region_priv;
+	struct mlxsw_sp1_acl_tcam_chunk *chunk = chunk_priv;
+	struct mlxsw_sp1_acl_tcam_entry *entry = entry_priv;
+
+	return mlxsw_sp_acl_ctcam_entry_add(mlxsw_sp, &region->cregion,
+					    &chunk->cchunk, &entry->centry,
+					    rulei, false);
+}
+
+static void mlxsw_sp1_acl_tcam_entry_del(struct mlxsw_sp *mlxsw_sp,
+					 void *region_priv, void *chunk_priv,
+					 void *entry_priv)
+{
+	struct mlxsw_sp1_acl_tcam_region *region = region_priv;
+	struct mlxsw_sp1_acl_tcam_chunk *chunk = chunk_priv;
+	struct mlxsw_sp1_acl_tcam_entry *entry = entry_priv;
+
+	mlxsw_sp_acl_ctcam_entry_del(mlxsw_sp, &region->cregion,
+				     &chunk->cchunk, &entry->centry);
+}
+
+static int
+mlxsw_sp1_acl_tcam_entry_action_replace(struct mlxsw_sp *mlxsw_sp,
+					void *region_priv, void *entry_priv,
+					struct mlxsw_sp_acl_rule_info *rulei)
+{
+	return -EOPNOTSUPP;
+}
+
+static int
+mlxsw_sp1_acl_tcam_region_entry_activity_get(struct mlxsw_sp *mlxsw_sp,
+					     struct mlxsw_sp_acl_tcam_region *_region,
+					     unsigned int offset,
+					     bool *activity)
+{
+	char ptce2_pl[MLXSW_REG_PTCE2_LEN];
+	int err;
+
+	mlxsw_reg_ptce2_pack(ptce2_pl, true, MLXSW_REG_PTCE2_OP_QUERY_CLEAR_ON_READ,
+			     _region->tcam_region_info, offset, 0);
+	err = mlxsw_reg_query(mlxsw_sp->core, MLXSW_REG(ptce2), ptce2_pl);
+	if (err)
+		return err;
+	*activity = mlxsw_reg_ptce2_a_get(ptce2_pl);
+	return 0;
+}
+
+static int
+mlxsw_sp1_acl_tcam_entry_activity_get(struct mlxsw_sp *mlxsw_sp,
+				      void *region_priv, void *entry_priv,
+				      bool *activity)
+{
+	struct mlxsw_sp1_acl_tcam_region *region = region_priv;
+	struct mlxsw_sp1_acl_tcam_entry *entry = entry_priv;
+	unsigned int offset;
+
+	offset = mlxsw_sp_acl_ctcam_entry_offset(&entry->centry);
+	return mlxsw_sp1_acl_tcam_region_entry_activity_get(mlxsw_sp,
+							    region->region,
+							    offset, activity);
+}
+
+const struct mlxsw_sp_acl_tcam_ops mlxsw_sp1_acl_tcam_ops = {
+	.key_type		= MLXSW_REG_PTAR_KEY_TYPE_FLEX,
+	.priv_size		= 0,
+	.init			= mlxsw_sp1_acl_tcam_init,
+	.fini			= mlxsw_sp1_acl_tcam_fini,
+	.region_priv_size	= sizeof(struct mlxsw_sp1_acl_tcam_region),
+	.region_init		= mlxsw_sp1_acl_tcam_region_init,
+	.region_fini		= mlxsw_sp1_acl_tcam_region_fini,
+	.region_associate	= mlxsw_sp1_acl_tcam_region_associate,
+	.chunk_priv_size	= sizeof(struct mlxsw_sp1_acl_tcam_chunk),
+	.chunk_init		= mlxsw_sp1_acl_tcam_chunk_init,
+	.chunk_fini		= mlxsw_sp1_acl_tcam_chunk_fini,
+	.entry_priv_size	= sizeof(struct mlxsw_sp1_acl_tcam_entry),
+	.entry_add		= mlxsw_sp1_acl_tcam_entry_add,
+	.entry_del		= mlxsw_sp1_acl_tcam_entry_del,
+	.entry_action_replace	= mlxsw_sp1_acl_tcam_entry_action_replace,
+	.entry_activity_get	= mlxsw_sp1_acl_tcam_entry_activity_get,
+};
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum1_kvdl.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum1_kvdl.c
new file mode 100644
index 000000000..1e3fc9893
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum1_kvdl.c
@@ -0,0 +1,427 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2018 Mellanox Technologies. All rights reserved */
+
+#include <linux/kernel.h>
+#include <linux/bitops.h>
+
+#include "spectrum.h"
+
+#define MLXSW_SP1_KVDL_SINGLE_BASE 0
+#define MLXSW_SP1_KVDL_SINGLE_SIZE 16384
+#define MLXSW_SP1_KVDL_SINGLE_END \
+	(MLXSW_SP1_KVDL_SINGLE_SIZE + MLXSW_SP1_KVDL_SINGLE_BASE - 1)
+
+#define MLXSW_SP1_KVDL_CHUNKS_BASE \
+	(MLXSW_SP1_KVDL_SINGLE_BASE + MLXSW_SP1_KVDL_SINGLE_SIZE)
+#define MLXSW_SP1_KVDL_CHUNKS_SIZE 49152
+#define MLXSW_SP1_KVDL_CHUNKS_END \
+	(MLXSW_SP1_KVDL_CHUNKS_SIZE + MLXSW_SP1_KVDL_CHUNKS_BASE - 1)
+
+#define MLXSW_SP1_KVDL_LARGE_CHUNKS_BASE \
+	(MLXSW_SP1_KVDL_CHUNKS_BASE + MLXSW_SP1_KVDL_CHUNKS_SIZE)
+#define MLXSW_SP1_KVDL_LARGE_CHUNKS_SIZE \
+	(MLXSW_SP_KVD_LINEAR_SIZE - MLXSW_SP1_KVDL_LARGE_CHUNKS_BASE)
+#define MLXSW_SP1_KVDL_LARGE_CHUNKS_END \
+	(MLXSW_SP1_KVDL_LARGE_CHUNKS_SIZE + MLXSW_SP1_KVDL_LARGE_CHUNKS_BASE - 1)
+
+#define MLXSW_SP1_KVDL_SINGLE_ALLOC_SIZE 1
+#define MLXSW_SP1_KVDL_CHUNKS_ALLOC_SIZE 32
+#define MLXSW_SP1_KVDL_LARGE_CHUNKS_ALLOC_SIZE 512
+
+struct mlxsw_sp1_kvdl_part_info {
+	unsigned int part_index;
+	unsigned int start_index;
+	unsigned int end_index;
+	unsigned int alloc_size;
+	enum mlxsw_sp_resource_id resource_id;
+};
+
+enum mlxsw_sp1_kvdl_part_id {
+	MLXSW_SP1_KVDL_PART_ID_SINGLE,
+	MLXSW_SP1_KVDL_PART_ID_CHUNKS,
+	MLXSW_SP1_KVDL_PART_ID_LARGE_CHUNKS,
+};
+
+#define MLXSW_SP1_KVDL_PART_INFO(id)				\
+[MLXSW_SP1_KVDL_PART_ID_##id] = {				\
+	.start_index = MLXSW_SP1_KVDL_##id##_BASE,		\
+	.end_index = MLXSW_SP1_KVDL_##id##_END,			\
+	.alloc_size = MLXSW_SP1_KVDL_##id##_ALLOC_SIZE,		\
+	.resource_id = MLXSW_SP_RESOURCE_KVD_LINEAR_##id,	\
+}
+
+static const struct mlxsw_sp1_kvdl_part_info mlxsw_sp1_kvdl_parts_info[] = {
+	MLXSW_SP1_KVDL_PART_INFO(SINGLE),
+	MLXSW_SP1_KVDL_PART_INFO(CHUNKS),
+	MLXSW_SP1_KVDL_PART_INFO(LARGE_CHUNKS),
+};
+
+#define MLXSW_SP1_KVDL_PARTS_INFO_LEN ARRAY_SIZE(mlxsw_sp1_kvdl_parts_info)
+
+struct mlxsw_sp1_kvdl_part {
+	struct mlxsw_sp1_kvdl_part_info info;
+	unsigned long usage[];	/* Entries */
+};
+
+struct mlxsw_sp1_kvdl {
+	struct mlxsw_sp1_kvdl_part *parts[MLXSW_SP1_KVDL_PARTS_INFO_LEN];
+};
+
+static struct mlxsw_sp1_kvdl_part *
+mlxsw_sp1_kvdl_alloc_size_part(struct mlxsw_sp1_kvdl *kvdl,
+			       unsigned int alloc_size)
+{
+	struct mlxsw_sp1_kvdl_part *part, *min_part = NULL;
+	int i;
+
+	for (i = 0; i < MLXSW_SP1_KVDL_PARTS_INFO_LEN; i++) {
+		part = kvdl->parts[i];
+		if (alloc_size <= part->info.alloc_size &&
+		    (!min_part ||
+		     part->info.alloc_size <= min_part->info.alloc_size))
+			min_part = part;
+	}
+
+	return min_part ?: ERR_PTR(-ENOBUFS);
+}
+
+static struct mlxsw_sp1_kvdl_part *
+mlxsw_sp1_kvdl_index_part(struct mlxsw_sp1_kvdl *kvdl, u32 kvdl_index)
+{
+	struct mlxsw_sp1_kvdl_part *part;
+	int i;
+
+	for (i = 0; i < MLXSW_SP1_KVDL_PARTS_INFO_LEN; i++) {
+		part = kvdl->parts[i];
+		if (kvdl_index >= part->info.start_index &&
+		    kvdl_index <= part->info.end_index)
+			return part;
+	}
+
+	return ERR_PTR(-EINVAL);
+}
+
+static u32
+mlxsw_sp1_kvdl_to_kvdl_index(const struct mlxsw_sp1_kvdl_part_info *info,
+			     unsigned int entry_index)
+{
+	return info->start_index + entry_index * info->alloc_size;
+}
+
+static unsigned int
+mlxsw_sp1_kvdl_to_entry_index(const struct mlxsw_sp1_kvdl_part_info *info,
+			      u32 kvdl_index)
+{
+	return (kvdl_index - info->start_index) / info->alloc_size;
+}
+
+static int mlxsw_sp1_kvdl_part_alloc(struct mlxsw_sp1_kvdl_part *part,
+				     u32 *p_kvdl_index)
+{
+	const struct mlxsw_sp1_kvdl_part_info *info = &part->info;
+	unsigned int entry_index, nr_entries;
+
+	nr_entries = (info->end_index - info->start_index + 1) /
+		     info->alloc_size;
+	entry_index = find_first_zero_bit(part->usage, nr_entries);
+	if (entry_index == nr_entries)
+		return -ENOBUFS;
+	__set_bit(entry_index, part->usage);
+
+	*p_kvdl_index = mlxsw_sp1_kvdl_to_kvdl_index(info, entry_index);
+
+	return 0;
+}
+
+static void mlxsw_sp1_kvdl_part_free(struct mlxsw_sp1_kvdl_part *part,
+				     u32 kvdl_index)
+{
+	const struct mlxsw_sp1_kvdl_part_info *info = &part->info;
+	unsigned int entry_index;
+
+	entry_index = mlxsw_sp1_kvdl_to_entry_index(info, kvdl_index);
+	__clear_bit(entry_index, part->usage);
+}
+
+static int mlxsw_sp1_kvdl_alloc(struct mlxsw_sp *mlxsw_sp, void *priv,
+				enum mlxsw_sp_kvdl_entry_type type,
+				unsigned int entry_count,
+				u32 *p_entry_index)
+{
+	struct mlxsw_sp1_kvdl *kvdl = priv;
+	struct mlxsw_sp1_kvdl_part *part;
+
+	/* Find partition with smallest allocation size satisfying the
+	 * requested size.
+	 */
+	part = mlxsw_sp1_kvdl_alloc_size_part(kvdl, entry_count);
+	if (IS_ERR(part))
+		return PTR_ERR(part);
+
+	return mlxsw_sp1_kvdl_part_alloc(part, p_entry_index);
+}
+
+static void mlxsw_sp1_kvdl_free(struct mlxsw_sp *mlxsw_sp, void *priv,
+				enum mlxsw_sp_kvdl_entry_type type,
+				unsigned int entry_count, int entry_index)
+{
+	struct mlxsw_sp1_kvdl *kvdl = priv;
+	struct mlxsw_sp1_kvdl_part *part;
+
+	part = mlxsw_sp1_kvdl_index_part(kvdl, entry_index);
+	if (IS_ERR(part))
+		return;
+	mlxsw_sp1_kvdl_part_free(part, entry_index);
+}
+
+static int mlxsw_sp1_kvdl_alloc_size_query(struct mlxsw_sp *mlxsw_sp,
+					   void *priv,
+					   enum mlxsw_sp_kvdl_entry_type type,
+					   unsigned int entry_count,
+					   unsigned int *p_alloc_size)
+{
+	struct mlxsw_sp1_kvdl *kvdl = priv;
+	struct mlxsw_sp1_kvdl_part *part;
+
+	part = mlxsw_sp1_kvdl_alloc_size_part(kvdl, entry_count);
+	if (IS_ERR(part))
+		return PTR_ERR(part);
+
+	*p_alloc_size = part->info.alloc_size;
+
+	return 0;
+}
+
+static void mlxsw_sp1_kvdl_part_update(struct mlxsw_sp1_kvdl_part *part,
+				       struct mlxsw_sp1_kvdl_part *part_prev,
+				       unsigned int size)
+{
+	if (!part_prev) {
+		part->info.end_index = size - 1;
+	} else {
+		part->info.start_index = part_prev->info.end_index + 1;
+		part->info.end_index = part->info.start_index + size - 1;
+	}
+}
+
+static struct mlxsw_sp1_kvdl_part *
+mlxsw_sp1_kvdl_part_init(struct mlxsw_sp *mlxsw_sp,
+			 const struct mlxsw_sp1_kvdl_part_info *info,
+			 struct mlxsw_sp1_kvdl_part *part_prev)
+{
+	struct devlink *devlink = priv_to_devlink(mlxsw_sp->core);
+	struct mlxsw_sp1_kvdl_part *part;
+	bool need_update = true;
+	unsigned int nr_entries;
+	u64 resource_size;
+	int err;
+
+	err = devl_resource_size_get(devlink, info->resource_id,
+				     &resource_size);
+	if (err) {
+		need_update = false;
+		resource_size = info->end_index - info->start_index + 1;
+	}
+
+	nr_entries = div_u64(resource_size, info->alloc_size);
+	part = kzalloc(struct_size(part, usage, BITS_TO_LONGS(nr_entries)),
+		       GFP_KERNEL);
+	if (!part)
+		return ERR_PTR(-ENOMEM);
+
+	memcpy(&part->info, info, sizeof(part->info));
+
+	if (need_update)
+		mlxsw_sp1_kvdl_part_update(part, part_prev, resource_size);
+	return part;
+}
+
+static void mlxsw_sp1_kvdl_part_fini(struct mlxsw_sp1_kvdl_part *part)
+{
+	kfree(part);
+}
+
+static int mlxsw_sp1_kvdl_parts_init(struct mlxsw_sp *mlxsw_sp,
+				     struct mlxsw_sp1_kvdl *kvdl)
+{
+	const struct mlxsw_sp1_kvdl_part_info *info;
+	struct mlxsw_sp1_kvdl_part *part_prev = NULL;
+	int err, i;
+
+	for (i = 0; i < MLXSW_SP1_KVDL_PARTS_INFO_LEN; i++) {
+		info = &mlxsw_sp1_kvdl_parts_info[i];
+		kvdl->parts[i] = mlxsw_sp1_kvdl_part_init(mlxsw_sp, info,
+							  part_prev);
+		if (IS_ERR(kvdl->parts[i])) {
+			err = PTR_ERR(kvdl->parts[i]);
+			goto err_kvdl_part_init;
+		}
+		part_prev = kvdl->parts[i];
+	}
+	return 0;
+
+err_kvdl_part_init:
+	for (i--; i >= 0; i--)
+		mlxsw_sp1_kvdl_part_fini(kvdl->parts[i]);
+	return err;
+}
+
+static void mlxsw_sp1_kvdl_parts_fini(struct mlxsw_sp1_kvdl *kvdl)
+{
+	int i;
+
+	for (i = 0; i < MLXSW_SP1_KVDL_PARTS_INFO_LEN; i++)
+		mlxsw_sp1_kvdl_part_fini(kvdl->parts[i]);
+}
+
+static u64 mlxsw_sp1_kvdl_part_occ(struct mlxsw_sp1_kvdl_part *part)
+{
+	const struct mlxsw_sp1_kvdl_part_info *info = &part->info;
+	unsigned int nr_entries;
+	int bit = -1;
+	u64 occ = 0;
+
+	nr_entries = (info->end_index -
+		      info->start_index + 1) /
+		      info->alloc_size;
+	while ((bit = find_next_bit(part->usage, nr_entries, bit + 1))
+		< nr_entries)
+		occ += info->alloc_size;
+	return occ;
+}
+
+static u64 mlxsw_sp1_kvdl_occ_get(void *priv)
+{
+	const struct mlxsw_sp1_kvdl *kvdl = priv;
+	u64 occ = 0;
+	int i;
+
+	for (i = 0; i < MLXSW_SP1_KVDL_PARTS_INFO_LEN; i++)
+		occ += mlxsw_sp1_kvdl_part_occ(kvdl->parts[i]);
+
+	return occ;
+}
+
+static u64 mlxsw_sp1_kvdl_single_occ_get(void *priv)
+{
+	const struct mlxsw_sp1_kvdl *kvdl = priv;
+	struct mlxsw_sp1_kvdl_part *part;
+
+	part = kvdl->parts[MLXSW_SP1_KVDL_PART_ID_SINGLE];
+	return mlxsw_sp1_kvdl_part_occ(part);
+}
+
+static u64 mlxsw_sp1_kvdl_chunks_occ_get(void *priv)
+{
+	const struct mlxsw_sp1_kvdl *kvdl = priv;
+	struct mlxsw_sp1_kvdl_part *part;
+
+	part = kvdl->parts[MLXSW_SP1_KVDL_PART_ID_CHUNKS];
+	return mlxsw_sp1_kvdl_part_occ(part);
+}
+
+static u64 mlxsw_sp1_kvdl_large_chunks_occ_get(void *priv)
+{
+	const struct mlxsw_sp1_kvdl *kvdl = priv;
+	struct mlxsw_sp1_kvdl_part *part;
+
+	part = kvdl->parts[MLXSW_SP1_KVDL_PART_ID_LARGE_CHUNKS];
+	return mlxsw_sp1_kvdl_part_occ(part);
+}
+
+static int mlxsw_sp1_kvdl_init(struct mlxsw_sp *mlxsw_sp, void *priv)
+{
+	struct devlink *devlink = priv_to_devlink(mlxsw_sp->core);
+	struct mlxsw_sp1_kvdl *kvdl = priv;
+	int err;
+
+	err = mlxsw_sp1_kvdl_parts_init(mlxsw_sp, kvdl);
+	if (err)
+		return err;
+	devl_resource_occ_get_register(devlink,
+				       MLXSW_SP_RESOURCE_KVD_LINEAR,
+				       mlxsw_sp1_kvdl_occ_get,
+				       kvdl);
+	devl_resource_occ_get_register(devlink,
+				       MLXSW_SP_RESOURCE_KVD_LINEAR_SINGLE,
+				       mlxsw_sp1_kvdl_single_occ_get,
+				       kvdl);
+	devl_resource_occ_get_register(devlink,
+				       MLXSW_SP_RESOURCE_KVD_LINEAR_CHUNKS,
+				       mlxsw_sp1_kvdl_chunks_occ_get,
+				       kvdl);
+	devl_resource_occ_get_register(devlink,
+				       MLXSW_SP_RESOURCE_KVD_LINEAR_LARGE_CHUNKS,
+				       mlxsw_sp1_kvdl_large_chunks_occ_get,
+				       kvdl);
+	return 0;
+}
+
+static void mlxsw_sp1_kvdl_fini(struct mlxsw_sp *mlxsw_sp, void *priv)
+{
+	struct devlink *devlink = priv_to_devlink(mlxsw_sp->core);
+	struct mlxsw_sp1_kvdl *kvdl = priv;
+
+	devl_resource_occ_get_unregister(devlink,
+					 MLXSW_SP_RESOURCE_KVD_LINEAR_LARGE_CHUNKS);
+	devl_resource_occ_get_unregister(devlink,
+					 MLXSW_SP_RESOURCE_KVD_LINEAR_CHUNKS);
+	devl_resource_occ_get_unregister(devlink,
+					 MLXSW_SP_RESOURCE_KVD_LINEAR_SINGLE);
+	devl_resource_occ_get_unregister(devlink,
+					 MLXSW_SP_RESOURCE_KVD_LINEAR);
+	mlxsw_sp1_kvdl_parts_fini(kvdl);
+}
+
+const struct mlxsw_sp_kvdl_ops mlxsw_sp1_kvdl_ops = {
+	.priv_size = sizeof(struct mlxsw_sp1_kvdl),
+	.init = mlxsw_sp1_kvdl_init,
+	.fini = mlxsw_sp1_kvdl_fini,
+	.alloc = mlxsw_sp1_kvdl_alloc,
+	.free = mlxsw_sp1_kvdl_free,
+	.alloc_size_query = mlxsw_sp1_kvdl_alloc_size_query,
+};
+
+int mlxsw_sp1_kvdl_resources_register(struct mlxsw_core *mlxsw_core)
+{
+	struct devlink *devlink = priv_to_devlink(mlxsw_core);
+	static struct devlink_resource_size_params size_params;
+	u32 kvdl_max_size;
+	int err;
+
+	kvdl_max_size = MLXSW_CORE_RES_GET(mlxsw_core, KVD_SIZE) -
+			MLXSW_CORE_RES_GET(mlxsw_core, KVD_SINGLE_MIN_SIZE) -
+			MLXSW_CORE_RES_GET(mlxsw_core, KVD_DOUBLE_MIN_SIZE);
+
+	devlink_resource_size_params_init(&size_params, 0, kvdl_max_size,
+					  MLXSW_SP1_KVDL_SINGLE_ALLOC_SIZE,
+					  DEVLINK_RESOURCE_UNIT_ENTRY);
+	err = devl_resource_register(devlink, MLXSW_SP_RESOURCE_NAME_KVD_LINEAR_SINGLES,
+				     MLXSW_SP1_KVDL_SINGLE_SIZE,
+				     MLXSW_SP_RESOURCE_KVD_LINEAR_SINGLE,
+				     MLXSW_SP_RESOURCE_KVD_LINEAR,
+				     &size_params);
+	if (err)
+		return err;
+
+	devlink_resource_size_params_init(&size_params, 0, kvdl_max_size,
+					  MLXSW_SP1_KVDL_CHUNKS_ALLOC_SIZE,
+					  DEVLINK_RESOURCE_UNIT_ENTRY);
+	err = devl_resource_register(devlink, MLXSW_SP_RESOURCE_NAME_KVD_LINEAR_CHUNKS,
+				     MLXSW_SP1_KVDL_CHUNKS_SIZE,
+				     MLXSW_SP_RESOURCE_KVD_LINEAR_CHUNKS,
+				     MLXSW_SP_RESOURCE_KVD_LINEAR,
+				     &size_params);
+	if (err)
+		return err;
+
+	devlink_resource_size_params_init(&size_params, 0, kvdl_max_size,
+					  MLXSW_SP1_KVDL_LARGE_CHUNKS_ALLOC_SIZE,
+					  DEVLINK_RESOURCE_UNIT_ENTRY);
+	err = devl_resource_register(devlink, MLXSW_SP_RESOURCE_NAME_KVD_LINEAR_LARGE_CHUNKS,
+				     MLXSW_SP1_KVDL_LARGE_CHUNKS_SIZE,
+				     MLXSW_SP_RESOURCE_KVD_LINEAR_LARGE_CHUNKS,
+				     MLXSW_SP_RESOURCE_KVD_LINEAR,
+				     &size_params);
+	return err;
+}
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum1_mr_tcam.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum1_mr_tcam.c
new file mode 100644
index 000000000..c8c675369
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum1_mr_tcam.c
@@ -0,0 +1,342 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2017-2018 Mellanox Technologies. All rights reserved */
+
+#include <linux/kernel.h>
+#include <linux/parman.h>
+
+#include "reg.h"
+#include "spectrum.h"
+#include "core_acl_flex_actions.h"
+#include "spectrum_mr.h"
+
+struct mlxsw_sp1_mr_tcam_region {
+	struct mlxsw_sp *mlxsw_sp;
+	enum mlxsw_reg_rtar_key_type rtar_key_type;
+	struct parman *parman;
+	struct parman_prio *parman_prios;
+};
+
+struct mlxsw_sp1_mr_tcam {
+	struct mlxsw_sp1_mr_tcam_region tcam_regions[MLXSW_SP_L3_PROTO_MAX];
+};
+
+struct mlxsw_sp1_mr_tcam_route {
+	struct parman_item parman_item;
+	struct parman_prio *parman_prio;
+};
+
+static int mlxsw_sp1_mr_tcam_route_replace(struct mlxsw_sp *mlxsw_sp,
+					   struct parman_item *parman_item,
+					   struct mlxsw_sp_mr_route_key *key,
+					   struct mlxsw_afa_block *afa_block)
+{
+	char rmft2_pl[MLXSW_REG_RMFT2_LEN];
+
+	switch (key->proto) {
+	case MLXSW_SP_L3_PROTO_IPV4:
+		mlxsw_reg_rmft2_ipv4_pack(rmft2_pl, true, parman_item->index,
+					  key->vrid,
+					  MLXSW_REG_RMFT2_IRIF_MASK_IGNORE, 0,
+					  ntohl(key->group.addr4),
+					  ntohl(key->group_mask.addr4),
+					  ntohl(key->source.addr4),
+					  ntohl(key->source_mask.addr4),
+					  mlxsw_afa_block_first_set(afa_block));
+		break;
+	case MLXSW_SP_L3_PROTO_IPV6:
+		mlxsw_reg_rmft2_ipv6_pack(rmft2_pl, true, parman_item->index,
+					  key->vrid,
+					  MLXSW_REG_RMFT2_IRIF_MASK_IGNORE, 0,
+					  key->group.addr6,
+					  key->group_mask.addr6,
+					  key->source.addr6,
+					  key->source_mask.addr6,
+					  mlxsw_afa_block_first_set(afa_block));
+	}
+
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(rmft2), rmft2_pl);
+}
+
+static int mlxsw_sp1_mr_tcam_route_remove(struct mlxsw_sp *mlxsw_sp,
+					  struct parman_item *parman_item,
+					  struct mlxsw_sp_mr_route_key *key)
+{
+	struct in6_addr zero_addr = IN6ADDR_ANY_INIT;
+	char rmft2_pl[MLXSW_REG_RMFT2_LEN];
+
+	switch (key->proto) {
+	case MLXSW_SP_L3_PROTO_IPV4:
+		mlxsw_reg_rmft2_ipv4_pack(rmft2_pl, false, parman_item->index,
+					  key->vrid, 0, 0, 0, 0, 0, 0, NULL);
+		break;
+	case MLXSW_SP_L3_PROTO_IPV6:
+		mlxsw_reg_rmft2_ipv6_pack(rmft2_pl, false, parman_item->index,
+					  key->vrid, 0, 0, zero_addr, zero_addr,
+					  zero_addr, zero_addr, NULL);
+		break;
+	}
+
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(rmft2), rmft2_pl);
+}
+
+static struct mlxsw_sp1_mr_tcam_region *
+mlxsw_sp1_mr_tcam_protocol_region(struct mlxsw_sp1_mr_tcam *mr_tcam,
+				  enum mlxsw_sp_l3proto proto)
+{
+	return &mr_tcam->tcam_regions[proto];
+}
+
+static int
+mlxsw_sp1_mr_tcam_route_parman_item_add(struct mlxsw_sp1_mr_tcam *mr_tcam,
+					struct mlxsw_sp1_mr_tcam_route *route,
+					struct mlxsw_sp_mr_route_key *key,
+					enum mlxsw_sp_mr_route_prio prio)
+{
+	struct mlxsw_sp1_mr_tcam_region *tcam_region;
+	int err;
+
+	tcam_region = mlxsw_sp1_mr_tcam_protocol_region(mr_tcam, key->proto);
+	err = parman_item_add(tcam_region->parman,
+			      &tcam_region->parman_prios[prio],
+			      &route->parman_item);
+	if (err)
+		return err;
+
+	route->parman_prio = &tcam_region->parman_prios[prio];
+	return 0;
+}
+
+static void
+mlxsw_sp1_mr_tcam_route_parman_item_remove(struct mlxsw_sp1_mr_tcam *mr_tcam,
+					   struct mlxsw_sp1_mr_tcam_route *route,
+					   struct mlxsw_sp_mr_route_key *key)
+{
+	struct mlxsw_sp1_mr_tcam_region *tcam_region;
+
+	tcam_region = mlxsw_sp1_mr_tcam_protocol_region(mr_tcam, key->proto);
+	parman_item_remove(tcam_region->parman,
+			   route->parman_prio, &route->parman_item);
+}
+
+static int
+mlxsw_sp1_mr_tcam_route_create(struct mlxsw_sp *mlxsw_sp, void *priv,
+			       void *route_priv,
+			       struct mlxsw_sp_mr_route_key *key,
+			       struct mlxsw_afa_block *afa_block,
+			       enum mlxsw_sp_mr_route_prio prio)
+{
+	struct mlxsw_sp1_mr_tcam_route *route = route_priv;
+	struct mlxsw_sp1_mr_tcam *mr_tcam = priv;
+	int err;
+
+	err = mlxsw_sp1_mr_tcam_route_parman_item_add(mr_tcam, route,
+						      key, prio);
+	if (err)
+		return err;
+
+	err = mlxsw_sp1_mr_tcam_route_replace(mlxsw_sp, &route->parman_item,
+					      key, afa_block);
+	if (err)
+		goto err_route_replace;
+	return 0;
+
+err_route_replace:
+	mlxsw_sp1_mr_tcam_route_parman_item_remove(mr_tcam, route, key);
+	return err;
+}
+
+static void
+mlxsw_sp1_mr_tcam_route_destroy(struct mlxsw_sp *mlxsw_sp, void *priv,
+				void *route_priv,
+				struct mlxsw_sp_mr_route_key *key)
+{
+	struct mlxsw_sp1_mr_tcam_route *route = route_priv;
+	struct mlxsw_sp1_mr_tcam *mr_tcam = priv;
+
+	mlxsw_sp1_mr_tcam_route_remove(mlxsw_sp, &route->parman_item, key);
+	mlxsw_sp1_mr_tcam_route_parman_item_remove(mr_tcam, route, key);
+}
+
+static int
+mlxsw_sp1_mr_tcam_route_update(struct mlxsw_sp *mlxsw_sp,
+			       void *route_priv,
+			       struct mlxsw_sp_mr_route_key *key,
+			       struct mlxsw_afa_block *afa_block)
+{
+	struct mlxsw_sp1_mr_tcam_route *route = route_priv;
+
+	return mlxsw_sp1_mr_tcam_route_replace(mlxsw_sp, &route->parman_item,
+					       key, afa_block);
+}
+
+#define MLXSW_SP1_MR_TCAM_REGION_BASE_COUNT 16
+#define MLXSW_SP1_MR_TCAM_REGION_RESIZE_STEP 16
+
+static int
+mlxsw_sp1_mr_tcam_region_alloc(struct mlxsw_sp1_mr_tcam_region *mr_tcam_region)
+{
+	struct mlxsw_sp *mlxsw_sp = mr_tcam_region->mlxsw_sp;
+	char rtar_pl[MLXSW_REG_RTAR_LEN];
+
+	mlxsw_reg_rtar_pack(rtar_pl, MLXSW_REG_RTAR_OP_ALLOCATE,
+			    mr_tcam_region->rtar_key_type,
+			    MLXSW_SP1_MR_TCAM_REGION_BASE_COUNT);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(rtar), rtar_pl);
+}
+
+static void
+mlxsw_sp1_mr_tcam_region_free(struct mlxsw_sp1_mr_tcam_region *mr_tcam_region)
+{
+	struct mlxsw_sp *mlxsw_sp = mr_tcam_region->mlxsw_sp;
+	char rtar_pl[MLXSW_REG_RTAR_LEN];
+
+	mlxsw_reg_rtar_pack(rtar_pl, MLXSW_REG_RTAR_OP_DEALLOCATE,
+			    mr_tcam_region->rtar_key_type, 0);
+	mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(rtar), rtar_pl);
+}
+
+static int mlxsw_sp1_mr_tcam_region_parman_resize(void *priv,
+						  unsigned long new_count)
+{
+	struct mlxsw_sp1_mr_tcam_region *mr_tcam_region = priv;
+	struct mlxsw_sp *mlxsw_sp = mr_tcam_region->mlxsw_sp;
+	char rtar_pl[MLXSW_REG_RTAR_LEN];
+	u64 max_tcam_rules;
+
+	max_tcam_rules = MLXSW_CORE_RES_GET(mlxsw_sp->core, ACL_MAX_TCAM_RULES);
+	if (new_count > max_tcam_rules)
+		return -EINVAL;
+	mlxsw_reg_rtar_pack(rtar_pl, MLXSW_REG_RTAR_OP_RESIZE,
+			    mr_tcam_region->rtar_key_type, new_count);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(rtar), rtar_pl);
+}
+
+static void mlxsw_sp1_mr_tcam_region_parman_move(void *priv,
+						 unsigned long from_index,
+						 unsigned long to_index,
+						 unsigned long count)
+{
+	struct mlxsw_sp1_mr_tcam_region *mr_tcam_region = priv;
+	struct mlxsw_sp *mlxsw_sp = mr_tcam_region->mlxsw_sp;
+	char rrcr_pl[MLXSW_REG_RRCR_LEN];
+
+	mlxsw_reg_rrcr_pack(rrcr_pl, MLXSW_REG_RRCR_OP_MOVE,
+			    from_index, count,
+			    mr_tcam_region->rtar_key_type, to_index);
+	mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(rrcr), rrcr_pl);
+}
+
+static const struct parman_ops mlxsw_sp1_mr_tcam_region_parman_ops = {
+	.base_count	= MLXSW_SP1_MR_TCAM_REGION_BASE_COUNT,
+	.resize_step	= MLXSW_SP1_MR_TCAM_REGION_RESIZE_STEP,
+	.resize		= mlxsw_sp1_mr_tcam_region_parman_resize,
+	.move		= mlxsw_sp1_mr_tcam_region_parman_move,
+	.algo		= PARMAN_ALGO_TYPE_LSORT,
+};
+
+static int
+mlxsw_sp1_mr_tcam_region_init(struct mlxsw_sp *mlxsw_sp,
+			      struct mlxsw_sp1_mr_tcam_region *mr_tcam_region,
+			      enum mlxsw_reg_rtar_key_type rtar_key_type)
+{
+	struct parman_prio *parman_prios;
+	struct parman *parman;
+	int err;
+	int i;
+
+	mr_tcam_region->rtar_key_type = rtar_key_type;
+	mr_tcam_region->mlxsw_sp = mlxsw_sp;
+
+	err = mlxsw_sp1_mr_tcam_region_alloc(mr_tcam_region);
+	if (err)
+		return err;
+
+	parman = parman_create(&mlxsw_sp1_mr_tcam_region_parman_ops,
+			       mr_tcam_region);
+	if (!parman) {
+		err = -ENOMEM;
+		goto err_parman_create;
+	}
+	mr_tcam_region->parman = parman;
+
+	parman_prios = kmalloc_array(MLXSW_SP_MR_ROUTE_PRIO_MAX + 1,
+				     sizeof(*parman_prios), GFP_KERNEL);
+	if (!parman_prios) {
+		err = -ENOMEM;
+		goto err_parman_prios_alloc;
+	}
+	mr_tcam_region->parman_prios = parman_prios;
+
+	for (i = 0; i < MLXSW_SP_MR_ROUTE_PRIO_MAX + 1; i++)
+		parman_prio_init(mr_tcam_region->parman,
+				 &mr_tcam_region->parman_prios[i], i);
+	return 0;
+
+err_parman_prios_alloc:
+	parman_destroy(parman);
+err_parman_create:
+	mlxsw_sp1_mr_tcam_region_free(mr_tcam_region);
+	return err;
+}
+
+static void
+mlxsw_sp1_mr_tcam_region_fini(struct mlxsw_sp1_mr_tcam_region *mr_tcam_region)
+{
+	int i;
+
+	for (i = 0; i < MLXSW_SP_MR_ROUTE_PRIO_MAX + 1; i++)
+		parman_prio_fini(&mr_tcam_region->parman_prios[i]);
+	kfree(mr_tcam_region->parman_prios);
+	parman_destroy(mr_tcam_region->parman);
+	mlxsw_sp1_mr_tcam_region_free(mr_tcam_region);
+}
+
+static int mlxsw_sp1_mr_tcam_init(struct mlxsw_sp *mlxsw_sp, void *priv)
+{
+	struct mlxsw_sp1_mr_tcam *mr_tcam = priv;
+	struct mlxsw_sp1_mr_tcam_region *region = &mr_tcam->tcam_regions[0];
+	u32 rtar_key;
+	int err;
+
+	if (!MLXSW_CORE_RES_VALID(mlxsw_sp->core, ACL_MAX_TCAM_RULES))
+		return -EIO;
+
+	rtar_key = MLXSW_REG_RTAR_KEY_TYPE_IPV4_MULTICAST;
+	err = mlxsw_sp1_mr_tcam_region_init(mlxsw_sp,
+					    &region[MLXSW_SP_L3_PROTO_IPV4],
+					    rtar_key);
+	if (err)
+		return err;
+
+	rtar_key = MLXSW_REG_RTAR_KEY_TYPE_IPV6_MULTICAST;
+	err = mlxsw_sp1_mr_tcam_region_init(mlxsw_sp,
+					    &region[MLXSW_SP_L3_PROTO_IPV6],
+					    rtar_key);
+	if (err)
+		goto err_ipv6_region_init;
+
+	return 0;
+
+err_ipv6_region_init:
+	mlxsw_sp1_mr_tcam_region_fini(&region[MLXSW_SP_L3_PROTO_IPV4]);
+	return err;
+}
+
+static void mlxsw_sp1_mr_tcam_fini(void *priv)
+{
+	struct mlxsw_sp1_mr_tcam *mr_tcam = priv;
+	struct mlxsw_sp1_mr_tcam_region *region = &mr_tcam->tcam_regions[0];
+
+	mlxsw_sp1_mr_tcam_region_fini(&region[MLXSW_SP_L3_PROTO_IPV6]);
+	mlxsw_sp1_mr_tcam_region_fini(&region[MLXSW_SP_L3_PROTO_IPV4]);
+}
+
+const struct mlxsw_sp_mr_tcam_ops mlxsw_sp1_mr_tcam_ops = {
+	.priv_size = sizeof(struct mlxsw_sp1_mr_tcam),
+	.init = mlxsw_sp1_mr_tcam_init,
+	.fini = mlxsw_sp1_mr_tcam_fini,
+	.route_priv_size = sizeof(struct mlxsw_sp1_mr_tcam_route),
+	.route_create = mlxsw_sp1_mr_tcam_route_create,
+	.route_destroy = mlxsw_sp1_mr_tcam_route_destroy,
+	.route_update = mlxsw_sp1_mr_tcam_route_update,
+};
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum2_acl_tcam.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum2_acl_tcam.c
new file mode 100644
index 000000000..5b0210862
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum2_acl_tcam.c
@@ -0,0 +1,280 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2018 Mellanox Technologies. All rights reserved */
+
+#include <linux/kernel.h>
+
+#include "spectrum.h"
+#include "spectrum_acl_tcam.h"
+#include "core_acl_flex_actions.h"
+
+struct mlxsw_sp2_acl_tcam {
+	struct mlxsw_sp_acl_atcam atcam;
+	u32 kvdl_index;
+	unsigned int kvdl_count;
+};
+
+struct mlxsw_sp2_acl_tcam_region {
+	struct mlxsw_sp_acl_atcam_region aregion;
+	struct mlxsw_sp_acl_tcam_region *region;
+};
+
+struct mlxsw_sp2_acl_tcam_chunk {
+	struct mlxsw_sp_acl_atcam_chunk achunk;
+};
+
+struct mlxsw_sp2_acl_tcam_entry {
+	struct mlxsw_sp_acl_atcam_entry aentry;
+	struct mlxsw_afa_block *act_block;
+};
+
+static int
+mlxsw_sp2_acl_ctcam_region_entry_insert(struct mlxsw_sp_acl_ctcam_region *cregion,
+					struct mlxsw_sp_acl_ctcam_entry *centry,
+					const char *mask)
+{
+	struct mlxsw_sp_acl_atcam_region *aregion;
+	struct mlxsw_sp_acl_atcam_entry *aentry;
+	struct mlxsw_sp_acl_erp_mask *erp_mask;
+
+	aregion = mlxsw_sp_acl_tcam_cregion_aregion(cregion);
+	aentry = mlxsw_sp_acl_tcam_centry_aentry(centry);
+
+	erp_mask = mlxsw_sp_acl_erp_mask_get(aregion, mask, true);
+	if (IS_ERR(erp_mask))
+		return PTR_ERR(erp_mask);
+	aentry->erp_mask = erp_mask;
+
+	return 0;
+}
+
+static void
+mlxsw_sp2_acl_ctcam_region_entry_remove(struct mlxsw_sp_acl_ctcam_region *cregion,
+					struct mlxsw_sp_acl_ctcam_entry *centry)
+{
+	struct mlxsw_sp_acl_atcam_region *aregion;
+	struct mlxsw_sp_acl_atcam_entry *aentry;
+
+	aregion = mlxsw_sp_acl_tcam_cregion_aregion(cregion);
+	aentry = mlxsw_sp_acl_tcam_centry_aentry(centry);
+
+	mlxsw_sp_acl_erp_mask_put(aregion, aentry->erp_mask);
+}
+
+static const struct mlxsw_sp_acl_ctcam_region_ops
+mlxsw_sp2_acl_ctcam_region_ops = {
+	.entry_insert = mlxsw_sp2_acl_ctcam_region_entry_insert,
+	.entry_remove = mlxsw_sp2_acl_ctcam_region_entry_remove,
+};
+
+static int mlxsw_sp2_acl_tcam_init(struct mlxsw_sp *mlxsw_sp, void *priv,
+				   struct mlxsw_sp_acl_tcam *_tcam)
+{
+	struct mlxsw_sp2_acl_tcam *tcam = priv;
+	struct mlxsw_afa_block *afa_block;
+	char pefa_pl[MLXSW_REG_PEFA_LEN];
+	char pgcr_pl[MLXSW_REG_PGCR_LEN];
+	char *enc_actions;
+	int i;
+	int err;
+
+	/* Some TCAM regions are not exposed to the host and used internally
+	 * by the device. Allocate KVDL entries for the default actions of
+	 * these regions to avoid the host from overwriting them.
+	 */
+	tcam->kvdl_count = _tcam->max_regions;
+	if (MLXSW_CORE_RES_VALID(mlxsw_sp->core, ACL_MAX_DEFAULT_ACTIONS))
+		tcam->kvdl_count = MLXSW_CORE_RES_GET(mlxsw_sp->core,
+						      ACL_MAX_DEFAULT_ACTIONS);
+	err = mlxsw_sp_kvdl_alloc(mlxsw_sp, MLXSW_SP_KVDL_ENTRY_TYPE_ACTSET,
+				  tcam->kvdl_count, &tcam->kvdl_index);
+	if (err)
+		return err;
+
+	/* Create flex action block, set default action (continue)
+	 * but don't commit. We need just the current set encoding
+	 * to be written using PEFA register to all indexes for all regions.
+	 */
+	afa_block = mlxsw_afa_block_create(mlxsw_sp->afa);
+	if (IS_ERR(afa_block)) {
+		err = PTR_ERR(afa_block);
+		goto err_afa_block;
+	}
+	err = mlxsw_afa_block_continue(afa_block);
+	if (WARN_ON(err))
+		goto err_afa_block_continue;
+	enc_actions = mlxsw_afa_block_cur_set(afa_block);
+
+	/* Only write to KVDL entries used by TCAM regions exposed to the
+	 * host.
+	 */
+	for (i = 0; i < _tcam->max_regions; i++) {
+		mlxsw_reg_pefa_pack(pefa_pl, tcam->kvdl_index + i,
+				    true, enc_actions);
+		err = mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(pefa), pefa_pl);
+		if (err)
+			goto err_pefa_write;
+	}
+	mlxsw_reg_pgcr_pack(pgcr_pl, tcam->kvdl_index);
+	err = mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(pgcr), pgcr_pl);
+	if (err)
+		goto err_pgcr_write;
+
+	err = mlxsw_sp_acl_atcam_init(mlxsw_sp, &tcam->atcam);
+	if (err)
+		goto err_atcam_init;
+
+	mlxsw_afa_block_destroy(afa_block);
+	return 0;
+
+err_atcam_init:
+err_pgcr_write:
+err_pefa_write:
+err_afa_block_continue:
+	mlxsw_afa_block_destroy(afa_block);
+err_afa_block:
+	mlxsw_sp_kvdl_free(mlxsw_sp, MLXSW_SP_KVDL_ENTRY_TYPE_ACTSET,
+			   tcam->kvdl_count, tcam->kvdl_index);
+	return err;
+}
+
+static void mlxsw_sp2_acl_tcam_fini(struct mlxsw_sp *mlxsw_sp, void *priv)
+{
+	struct mlxsw_sp2_acl_tcam *tcam = priv;
+
+	mlxsw_sp_acl_atcam_fini(mlxsw_sp, &tcam->atcam);
+	mlxsw_sp_kvdl_free(mlxsw_sp, MLXSW_SP_KVDL_ENTRY_TYPE_ACTSET,
+			   tcam->kvdl_count, tcam->kvdl_index);
+}
+
+static int
+mlxsw_sp2_acl_tcam_region_init(struct mlxsw_sp *mlxsw_sp, void *region_priv,
+			       void *tcam_priv,
+			       struct mlxsw_sp_acl_tcam_region *_region,
+			       void *hints_priv)
+{
+	struct mlxsw_sp2_acl_tcam_region *region = region_priv;
+	struct mlxsw_sp2_acl_tcam *tcam = tcam_priv;
+
+	region->region = _region;
+
+	return mlxsw_sp_acl_atcam_region_init(mlxsw_sp, &tcam->atcam,
+					      &region->aregion,
+					      _region, hints_priv,
+					      &mlxsw_sp2_acl_ctcam_region_ops);
+}
+
+static void
+mlxsw_sp2_acl_tcam_region_fini(struct mlxsw_sp *mlxsw_sp, void *region_priv)
+{
+	struct mlxsw_sp2_acl_tcam_region *region = region_priv;
+
+	mlxsw_sp_acl_atcam_region_fini(&region->aregion);
+}
+
+static int
+mlxsw_sp2_acl_tcam_region_associate(struct mlxsw_sp *mlxsw_sp,
+				    struct mlxsw_sp_acl_tcam_region *region)
+{
+	return mlxsw_sp_acl_atcam_region_associate(mlxsw_sp, region->id);
+}
+
+static void *mlxsw_sp2_acl_tcam_region_rehash_hints_get(void *region_priv)
+{
+	struct mlxsw_sp2_acl_tcam_region *region = region_priv;
+
+	return mlxsw_sp_acl_atcam_rehash_hints_get(&region->aregion);
+}
+
+static void mlxsw_sp2_acl_tcam_region_rehash_hints_put(void *hints_priv)
+{
+	mlxsw_sp_acl_atcam_rehash_hints_put(hints_priv);
+}
+
+static void mlxsw_sp2_acl_tcam_chunk_init(void *region_priv, void *chunk_priv,
+					  unsigned int priority)
+{
+	struct mlxsw_sp2_acl_tcam_region *region = region_priv;
+	struct mlxsw_sp2_acl_tcam_chunk *chunk = chunk_priv;
+
+	mlxsw_sp_acl_atcam_chunk_init(&region->aregion, &chunk->achunk,
+				      priority);
+}
+
+static void mlxsw_sp2_acl_tcam_chunk_fini(void *chunk_priv)
+{
+	struct mlxsw_sp2_acl_tcam_chunk *chunk = chunk_priv;
+
+	mlxsw_sp_acl_atcam_chunk_fini(&chunk->achunk);
+}
+
+static int mlxsw_sp2_acl_tcam_entry_add(struct mlxsw_sp *mlxsw_sp,
+					void *region_priv, void *chunk_priv,
+					void *entry_priv,
+					struct mlxsw_sp_acl_rule_info *rulei)
+{
+	struct mlxsw_sp2_acl_tcam_region *region = region_priv;
+	struct mlxsw_sp2_acl_tcam_chunk *chunk = chunk_priv;
+	struct mlxsw_sp2_acl_tcam_entry *entry = entry_priv;
+
+	entry->act_block = rulei->act_block;
+	return mlxsw_sp_acl_atcam_entry_add(mlxsw_sp, &region->aregion,
+					    &chunk->achunk, &entry->aentry,
+					    rulei);
+}
+
+static void mlxsw_sp2_acl_tcam_entry_del(struct mlxsw_sp *mlxsw_sp,
+					 void *region_priv, void *chunk_priv,
+					 void *entry_priv)
+{
+	struct mlxsw_sp2_acl_tcam_region *region = region_priv;
+	struct mlxsw_sp2_acl_tcam_chunk *chunk = chunk_priv;
+	struct mlxsw_sp2_acl_tcam_entry *entry = entry_priv;
+
+	mlxsw_sp_acl_atcam_entry_del(mlxsw_sp, &region->aregion, &chunk->achunk,
+				     &entry->aentry);
+}
+
+static int
+mlxsw_sp2_acl_tcam_entry_action_replace(struct mlxsw_sp *mlxsw_sp,
+					void *region_priv, void *entry_priv,
+					struct mlxsw_sp_acl_rule_info *rulei)
+{
+	struct mlxsw_sp2_acl_tcam_region *region = region_priv;
+	struct mlxsw_sp2_acl_tcam_entry *entry = entry_priv;
+
+	entry->act_block = rulei->act_block;
+	return mlxsw_sp_acl_atcam_entry_action_replace(mlxsw_sp,
+						       &region->aregion,
+						       &entry->aentry, rulei);
+}
+
+static int
+mlxsw_sp2_acl_tcam_entry_activity_get(struct mlxsw_sp *mlxsw_sp,
+				      void *region_priv, void *entry_priv,
+				      bool *activity)
+{
+	struct mlxsw_sp2_acl_tcam_entry *entry = entry_priv;
+
+	return mlxsw_afa_block_activity_get(entry->act_block, activity);
+}
+
+const struct mlxsw_sp_acl_tcam_ops mlxsw_sp2_acl_tcam_ops = {
+	.key_type		= MLXSW_REG_PTAR_KEY_TYPE_FLEX2,
+	.priv_size		= sizeof(struct mlxsw_sp2_acl_tcam),
+	.init			= mlxsw_sp2_acl_tcam_init,
+	.fini			= mlxsw_sp2_acl_tcam_fini,
+	.region_priv_size	= sizeof(struct mlxsw_sp2_acl_tcam_region),
+	.region_init		= mlxsw_sp2_acl_tcam_region_init,
+	.region_fini		= mlxsw_sp2_acl_tcam_region_fini,
+	.region_associate	= mlxsw_sp2_acl_tcam_region_associate,
+	.region_rehash_hints_get = mlxsw_sp2_acl_tcam_region_rehash_hints_get,
+	.region_rehash_hints_put = mlxsw_sp2_acl_tcam_region_rehash_hints_put,
+	.chunk_priv_size	= sizeof(struct mlxsw_sp2_acl_tcam_chunk),
+	.chunk_init		= mlxsw_sp2_acl_tcam_chunk_init,
+	.chunk_fini		= mlxsw_sp2_acl_tcam_chunk_fini,
+	.entry_priv_size	= sizeof(struct mlxsw_sp2_acl_tcam_entry),
+	.entry_add		= mlxsw_sp2_acl_tcam_entry_add,
+	.entry_del		= mlxsw_sp2_acl_tcam_entry_del,
+	.entry_action_replace	= mlxsw_sp2_acl_tcam_entry_action_replace,
+	.entry_activity_get	= mlxsw_sp2_acl_tcam_entry_activity_get,
+};
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum2_kvdl.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum2_kvdl.c
new file mode 100644
index 000000000..24ff305a2
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum2_kvdl.c
@@ -0,0 +1,273 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2018 Mellanox Technologies. All rights reserved */
+
+#include <linux/kernel.h>
+#include <linux/bitops.h>
+
+#include "spectrum.h"
+#include "core.h"
+#include "reg.h"
+#include "resources.h"
+
+struct mlxsw_sp2_kvdl_part_info {
+	u8 res_type;
+	/* For each defined partititon we need to know how many
+	 * usage bits we need and how many indexes there are
+	 * represented by a single bit. This could be got from FW
+	 * querying appropriate resources. So have the resource
+	 * ids for this purpose in partition definition.
+	 */
+	enum mlxsw_res_id usage_bit_count_res_id;
+	enum mlxsw_res_id index_range_res_id;
+};
+
+#define MLXSW_SP2_KVDL_PART_INFO(_entry_type, _res_type,			\
+				 _usage_bit_count_res_id, _index_range_res_id)	\
+[MLXSW_SP_KVDL_ENTRY_TYPE_##_entry_type] = {					\
+	.res_type = _res_type,							\
+	.usage_bit_count_res_id = MLXSW_RES_ID_##_usage_bit_count_res_id,	\
+	.index_range_res_id = MLXSW_RES_ID_##_index_range_res_id,		\
+}
+
+static const struct mlxsw_sp2_kvdl_part_info mlxsw_sp2_kvdl_parts_info[] = {
+	MLXSW_SP2_KVDL_PART_INFO(ADJ, 0x21, KVD_SIZE, MAX_KVD_LINEAR_RANGE),
+	MLXSW_SP2_KVDL_PART_INFO(ACTSET, 0x23, MAX_KVD_ACTION_SETS,
+				 MAX_KVD_ACTION_SETS),
+	MLXSW_SP2_KVDL_PART_INFO(PBS, 0x24, KVD_SIZE, KVD_SIZE),
+	MLXSW_SP2_KVDL_PART_INFO(MCRIGR, 0x26, KVD_SIZE, KVD_SIZE),
+	MLXSW_SP2_KVDL_PART_INFO(IPV6_ADDRESS, 0x28, KVD_SIZE, KVD_SIZE),
+	MLXSW_SP2_KVDL_PART_INFO(TNUMT, 0x29, KVD_SIZE, KVD_SIZE),
+};
+
+#define MLXSW_SP2_KVDL_PARTS_INFO_LEN ARRAY_SIZE(mlxsw_sp2_kvdl_parts_info)
+
+struct mlxsw_sp2_kvdl_part {
+	const struct mlxsw_sp2_kvdl_part_info *info;
+	unsigned int usage_bit_count;
+	unsigned int indexes_per_usage_bit;
+	unsigned int last_allocated_bit;
+	unsigned long usage[];	/* Usage bits */
+};
+
+struct mlxsw_sp2_kvdl {
+	struct mlxsw_sp2_kvdl_part *parts[MLXSW_SP2_KVDL_PARTS_INFO_LEN];
+};
+
+static int mlxsw_sp2_kvdl_part_find_zero_bits(struct mlxsw_sp2_kvdl_part *part,
+					      unsigned int bit_count,
+					      unsigned int *p_bit)
+{
+	unsigned int start_bit;
+	unsigned int bit;
+	unsigned int i;
+	bool wrap = false;
+
+	start_bit = part->last_allocated_bit + 1;
+	if (start_bit == part->usage_bit_count)
+		start_bit = 0;
+	bit = start_bit;
+again:
+	bit = find_next_zero_bit(part->usage, part->usage_bit_count, bit);
+	if (!wrap && bit + bit_count >= part->usage_bit_count) {
+		wrap = true;
+		bit = 0;
+		goto again;
+	}
+	if (wrap && bit + bit_count >= start_bit)
+		return -ENOBUFS;
+	for (i = 0; i < bit_count; i++) {
+		if (test_bit(bit + i, part->usage)) {
+			bit += bit_count;
+			goto again;
+		}
+	}
+	*p_bit = bit;
+	return 0;
+}
+
+static int mlxsw_sp2_kvdl_part_alloc(struct mlxsw_sp2_kvdl_part *part,
+				     unsigned int size,
+				     u32 *p_kvdl_index)
+{
+	unsigned int bit_count;
+	unsigned int bit;
+	unsigned int i;
+	int err;
+
+	bit_count = DIV_ROUND_UP(size, part->indexes_per_usage_bit);
+	err = mlxsw_sp2_kvdl_part_find_zero_bits(part, bit_count, &bit);
+	if (err)
+		return err;
+	for (i = 0; i < bit_count; i++)
+		__set_bit(bit + i, part->usage);
+	*p_kvdl_index = bit * part->indexes_per_usage_bit;
+	return 0;
+}
+
+static int mlxsw_sp2_kvdl_rec_del(struct mlxsw_sp *mlxsw_sp, u8 res_type,
+				  u16 size, u32 kvdl_index)
+{
+	char *iedr_pl;
+	int err;
+
+	iedr_pl = kmalloc(MLXSW_REG_IEDR_LEN, GFP_KERNEL);
+	if (!iedr_pl)
+		return -ENOMEM;
+
+	mlxsw_reg_iedr_pack(iedr_pl);
+	mlxsw_reg_iedr_rec_pack(iedr_pl, 0, res_type, size, kvdl_index);
+	err = mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(iedr), iedr_pl);
+	kfree(iedr_pl);
+	return err;
+}
+
+static void mlxsw_sp2_kvdl_part_free(struct mlxsw_sp *mlxsw_sp,
+				     struct mlxsw_sp2_kvdl_part *part,
+				     unsigned int size, u32 kvdl_index)
+{
+	unsigned int bit_count;
+	unsigned int bit;
+	unsigned int i;
+	int err;
+
+	/* We need to ask FW to delete previously used KVD linear index */
+	err = mlxsw_sp2_kvdl_rec_del(mlxsw_sp, part->info->res_type,
+				     size, kvdl_index);
+	if (err)
+		return;
+
+	bit_count = DIV_ROUND_UP(size, part->indexes_per_usage_bit);
+	bit = kvdl_index / part->indexes_per_usage_bit;
+	for (i = 0; i < bit_count; i++)
+		__clear_bit(bit + i, part->usage);
+}
+
+static int mlxsw_sp2_kvdl_alloc(struct mlxsw_sp *mlxsw_sp, void *priv,
+				enum mlxsw_sp_kvdl_entry_type type,
+				unsigned int entry_count,
+				u32 *p_entry_index)
+{
+	unsigned int size = entry_count * mlxsw_sp_kvdl_entry_size(type);
+	struct mlxsw_sp2_kvdl *kvdl = priv;
+	struct mlxsw_sp2_kvdl_part *part = kvdl->parts[type];
+
+	return mlxsw_sp2_kvdl_part_alloc(part, size, p_entry_index);
+}
+
+static void mlxsw_sp2_kvdl_free(struct mlxsw_sp *mlxsw_sp, void *priv,
+				enum mlxsw_sp_kvdl_entry_type type,
+				unsigned int entry_count,
+				int entry_index)
+{
+	unsigned int size = entry_count * mlxsw_sp_kvdl_entry_size(type);
+	struct mlxsw_sp2_kvdl *kvdl = priv;
+	struct mlxsw_sp2_kvdl_part *part = kvdl->parts[type];
+
+	return mlxsw_sp2_kvdl_part_free(mlxsw_sp, part, size, entry_index);
+}
+
+static int mlxsw_sp2_kvdl_alloc_size_query(struct mlxsw_sp *mlxsw_sp,
+					   void *priv,
+					   enum mlxsw_sp_kvdl_entry_type type,
+					   unsigned int entry_count,
+					   unsigned int *p_alloc_count)
+{
+	*p_alloc_count = entry_count;
+	return 0;
+}
+
+static struct mlxsw_sp2_kvdl_part *
+mlxsw_sp2_kvdl_part_init(struct mlxsw_sp *mlxsw_sp,
+			 const struct mlxsw_sp2_kvdl_part_info *info)
+{
+	unsigned int indexes_per_usage_bit;
+	struct mlxsw_sp2_kvdl_part *part;
+	unsigned int index_range;
+	unsigned int usage_bit_count;
+	size_t usage_size;
+
+	if (!mlxsw_core_res_valid(mlxsw_sp->core,
+				  info->usage_bit_count_res_id) ||
+	    !mlxsw_core_res_valid(mlxsw_sp->core,
+				  info->index_range_res_id))
+		return ERR_PTR(-EIO);
+	usage_bit_count = mlxsw_core_res_get(mlxsw_sp->core,
+					     info->usage_bit_count_res_id);
+	index_range = mlxsw_core_res_get(mlxsw_sp->core,
+					 info->index_range_res_id);
+
+	/* For some partitions, one usage bit represents a group of indexes.
+	 * That's why we compute the number of indexes per usage bit here,
+	 * according to queried resources.
+	 */
+	indexes_per_usage_bit = index_range / usage_bit_count;
+
+	usage_size = BITS_TO_LONGS(usage_bit_count) * sizeof(unsigned long);
+	part = kzalloc(sizeof(*part) + usage_size, GFP_KERNEL);
+	if (!part)
+		return ERR_PTR(-ENOMEM);
+	part->info = info;
+	part->usage_bit_count = usage_bit_count;
+	part->indexes_per_usage_bit = indexes_per_usage_bit;
+	part->last_allocated_bit = usage_bit_count - 1;
+	return part;
+}
+
+static void mlxsw_sp2_kvdl_part_fini(struct mlxsw_sp2_kvdl_part *part)
+{
+	kfree(part);
+}
+
+static int mlxsw_sp2_kvdl_parts_init(struct mlxsw_sp *mlxsw_sp,
+				     struct mlxsw_sp2_kvdl *kvdl)
+{
+	const struct mlxsw_sp2_kvdl_part_info *info;
+	int i;
+	int err;
+
+	for (i = 0; i < MLXSW_SP2_KVDL_PARTS_INFO_LEN; i++) {
+		info = &mlxsw_sp2_kvdl_parts_info[i];
+		kvdl->parts[i] = mlxsw_sp2_kvdl_part_init(mlxsw_sp, info);
+		if (IS_ERR(kvdl->parts[i])) {
+			err = PTR_ERR(kvdl->parts[i]);
+			goto err_kvdl_part_init;
+		}
+	}
+	return 0;
+
+err_kvdl_part_init:
+	for (i--; i >= 0; i--)
+		mlxsw_sp2_kvdl_part_fini(kvdl->parts[i]);
+	return err;
+}
+
+static void mlxsw_sp2_kvdl_parts_fini(struct mlxsw_sp2_kvdl *kvdl)
+{
+	int i;
+
+	for (i = 0; i < MLXSW_SP2_KVDL_PARTS_INFO_LEN; i++)
+		mlxsw_sp2_kvdl_part_fini(kvdl->parts[i]);
+}
+
+static int mlxsw_sp2_kvdl_init(struct mlxsw_sp *mlxsw_sp, void *priv)
+{
+	struct mlxsw_sp2_kvdl *kvdl = priv;
+
+	return mlxsw_sp2_kvdl_parts_init(mlxsw_sp, kvdl);
+}
+
+static void mlxsw_sp2_kvdl_fini(struct mlxsw_sp *mlxsw_sp, void *priv)
+{
+	struct mlxsw_sp2_kvdl *kvdl = priv;
+
+	mlxsw_sp2_kvdl_parts_fini(kvdl);
+}
+
+const struct mlxsw_sp_kvdl_ops mlxsw_sp2_kvdl_ops = {
+	.priv_size = sizeof(struct mlxsw_sp2_kvdl),
+	.init = mlxsw_sp2_kvdl_init,
+	.fini = mlxsw_sp2_kvdl_fini,
+	.alloc = mlxsw_sp2_kvdl_alloc,
+	.free = mlxsw_sp2_kvdl_free,
+	.alloc_size_query = mlxsw_sp2_kvdl_alloc_size_query,
+};
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum2_mr_tcam.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum2_mr_tcam.c
new file mode 100644
index 000000000..b1178b7a7
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum2_mr_tcam.c
@@ -0,0 +1,330 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2018 Mellanox Technologies. All rights reserved */
+
+#include <linux/kernel.h>
+
+#include "core_acl_flex_actions.h"
+#include "spectrum.h"
+#include "spectrum_mr.h"
+
+struct mlxsw_sp2_mr_tcam {
+	struct mlxsw_sp *mlxsw_sp;
+	struct mlxsw_sp_flow_block *flow_block;
+	struct mlxsw_sp_acl_ruleset *ruleset4;
+	struct mlxsw_sp_acl_ruleset *ruleset6;
+};
+
+struct mlxsw_sp2_mr_route {
+	struct mlxsw_sp2_mr_tcam *mr_tcam;
+};
+
+static struct mlxsw_sp_acl_ruleset *
+mlxsw_sp2_mr_tcam_proto_ruleset(struct mlxsw_sp2_mr_tcam *mr_tcam,
+				enum mlxsw_sp_l3proto proto)
+{
+	switch (proto) {
+	case MLXSW_SP_L3_PROTO_IPV4:
+		return mr_tcam->ruleset4;
+	case MLXSW_SP_L3_PROTO_IPV6:
+		return mr_tcam->ruleset6;
+	}
+	return NULL;
+}
+
+static int mlxsw_sp2_mr_tcam_bind_group(struct mlxsw_sp *mlxsw_sp,
+					enum mlxsw_reg_pemrbt_protocol protocol,
+					struct mlxsw_sp_acl_ruleset *ruleset)
+{
+	char pemrbt_pl[MLXSW_REG_PEMRBT_LEN];
+	u16 group_id;
+
+	group_id = mlxsw_sp_acl_ruleset_group_id(ruleset);
+
+	mlxsw_reg_pemrbt_pack(pemrbt_pl, protocol, group_id);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(pemrbt), pemrbt_pl);
+}
+
+static const enum mlxsw_afk_element mlxsw_sp2_mr_tcam_usage_ipv4[] = {
+		MLXSW_AFK_ELEMENT_VIRT_ROUTER_MSB,
+		MLXSW_AFK_ELEMENT_VIRT_ROUTER_LSB,
+		MLXSW_AFK_ELEMENT_SRC_IP_0_31,
+		MLXSW_AFK_ELEMENT_DST_IP_0_31,
+};
+
+static int mlxsw_sp2_mr_tcam_ipv4_init(struct mlxsw_sp2_mr_tcam *mr_tcam)
+{
+	struct mlxsw_afk_element_usage elusage;
+	int err;
+
+	/* Initialize IPv4 ACL group. */
+	mlxsw_afk_element_usage_fill(&elusage,
+				     mlxsw_sp2_mr_tcam_usage_ipv4,
+				     ARRAY_SIZE(mlxsw_sp2_mr_tcam_usage_ipv4));
+	mr_tcam->ruleset4 = mlxsw_sp_acl_ruleset_get(mr_tcam->mlxsw_sp,
+						     mr_tcam->flow_block,
+						     MLXSW_SP_L3_PROTO_IPV4,
+						     MLXSW_SP_ACL_PROFILE_MR,
+						     &elusage);
+
+	if (IS_ERR(mr_tcam->ruleset4))
+		return PTR_ERR(mr_tcam->ruleset4);
+
+	/* MC Router groups should be bound before routes are inserted. */
+	err = mlxsw_sp2_mr_tcam_bind_group(mr_tcam->mlxsw_sp,
+					   MLXSW_REG_PEMRBT_PROTO_IPV4,
+					   mr_tcam->ruleset4);
+	if (err)
+		goto err_bind_group;
+
+	return 0;
+
+err_bind_group:
+	mlxsw_sp_acl_ruleset_put(mr_tcam->mlxsw_sp, mr_tcam->ruleset4);
+	return err;
+}
+
+static void mlxsw_sp2_mr_tcam_ipv4_fini(struct mlxsw_sp2_mr_tcam *mr_tcam)
+{
+	mlxsw_sp_acl_ruleset_put(mr_tcam->mlxsw_sp, mr_tcam->ruleset4);
+}
+
+static const enum mlxsw_afk_element mlxsw_sp2_mr_tcam_usage_ipv6[] = {
+		MLXSW_AFK_ELEMENT_VIRT_ROUTER_MSB,
+		MLXSW_AFK_ELEMENT_VIRT_ROUTER_LSB,
+		MLXSW_AFK_ELEMENT_SRC_IP_96_127,
+		MLXSW_AFK_ELEMENT_SRC_IP_64_95,
+		MLXSW_AFK_ELEMENT_SRC_IP_32_63,
+		MLXSW_AFK_ELEMENT_SRC_IP_0_31,
+		MLXSW_AFK_ELEMENT_DST_IP_96_127,
+		MLXSW_AFK_ELEMENT_DST_IP_64_95,
+		MLXSW_AFK_ELEMENT_DST_IP_32_63,
+		MLXSW_AFK_ELEMENT_DST_IP_0_31,
+};
+
+static int mlxsw_sp2_mr_tcam_ipv6_init(struct mlxsw_sp2_mr_tcam *mr_tcam)
+{
+	struct mlxsw_afk_element_usage elusage;
+	int err;
+
+	/* Initialize IPv6 ACL group */
+	mlxsw_afk_element_usage_fill(&elusage,
+				     mlxsw_sp2_mr_tcam_usage_ipv6,
+				     ARRAY_SIZE(mlxsw_sp2_mr_tcam_usage_ipv6));
+	mr_tcam->ruleset6 = mlxsw_sp_acl_ruleset_get(mr_tcam->mlxsw_sp,
+						     mr_tcam->flow_block,
+						     MLXSW_SP_L3_PROTO_IPV6,
+						     MLXSW_SP_ACL_PROFILE_MR,
+						     &elusage);
+
+	if (IS_ERR(mr_tcam->ruleset6))
+		return PTR_ERR(mr_tcam->ruleset6);
+
+	/* MC Router groups should be bound before routes are inserted. */
+	err = mlxsw_sp2_mr_tcam_bind_group(mr_tcam->mlxsw_sp,
+					   MLXSW_REG_PEMRBT_PROTO_IPV6,
+					   mr_tcam->ruleset6);
+	if (err)
+		goto err_bind_group;
+
+	return 0;
+
+err_bind_group:
+	mlxsw_sp_acl_ruleset_put(mr_tcam->mlxsw_sp, mr_tcam->ruleset6);
+	return err;
+}
+
+static void mlxsw_sp2_mr_tcam_ipv6_fini(struct mlxsw_sp2_mr_tcam *mr_tcam)
+{
+	mlxsw_sp_acl_ruleset_put(mr_tcam->mlxsw_sp, mr_tcam->ruleset6);
+}
+
+static void
+mlxsw_sp2_mr_tcam_rule_parse4(struct mlxsw_sp_acl_rule_info *rulei,
+			      struct mlxsw_sp_mr_route_key *key)
+{
+	mlxsw_sp_acl_rulei_keymask_buf(rulei, MLXSW_AFK_ELEMENT_SRC_IP_0_31,
+				       (char *) &key->source.addr4,
+				       (char *) &key->source_mask.addr4, 4);
+	mlxsw_sp_acl_rulei_keymask_buf(rulei, MLXSW_AFK_ELEMENT_DST_IP_0_31,
+				       (char *) &key->group.addr4,
+				       (char *) &key->group_mask.addr4, 4);
+}
+
+static void
+mlxsw_sp2_mr_tcam_rule_parse6(struct mlxsw_sp_acl_rule_info *rulei,
+			      struct mlxsw_sp_mr_route_key *key)
+{
+	mlxsw_sp_acl_rulei_keymask_buf(rulei, MLXSW_AFK_ELEMENT_SRC_IP_96_127,
+				       &key->source.addr6.s6_addr[0x0],
+				       &key->source_mask.addr6.s6_addr[0x0], 4);
+	mlxsw_sp_acl_rulei_keymask_buf(rulei, MLXSW_AFK_ELEMENT_SRC_IP_64_95,
+				       &key->source.addr6.s6_addr[0x4],
+				       &key->source_mask.addr6.s6_addr[0x4], 4);
+	mlxsw_sp_acl_rulei_keymask_buf(rulei, MLXSW_AFK_ELEMENT_SRC_IP_32_63,
+				       &key->source.addr6.s6_addr[0x8],
+				       &key->source_mask.addr6.s6_addr[0x8], 4);
+	mlxsw_sp_acl_rulei_keymask_buf(rulei, MLXSW_AFK_ELEMENT_SRC_IP_0_31,
+				       &key->source.addr6.s6_addr[0xc],
+				       &key->source_mask.addr6.s6_addr[0xc], 4);
+	mlxsw_sp_acl_rulei_keymask_buf(rulei, MLXSW_AFK_ELEMENT_DST_IP_96_127,
+				       &key->group.addr6.s6_addr[0x0],
+				       &key->group_mask.addr6.s6_addr[0x0], 4);
+	mlxsw_sp_acl_rulei_keymask_buf(rulei, MLXSW_AFK_ELEMENT_DST_IP_64_95,
+				       &key->group.addr6.s6_addr[0x4],
+				       &key->group_mask.addr6.s6_addr[0x4], 4);
+	mlxsw_sp_acl_rulei_keymask_buf(rulei, MLXSW_AFK_ELEMENT_DST_IP_32_63,
+				       &key->group.addr6.s6_addr[0x8],
+				       &key->group_mask.addr6.s6_addr[0x8], 4);
+	mlxsw_sp_acl_rulei_keymask_buf(rulei, MLXSW_AFK_ELEMENT_DST_IP_0_31,
+				       &key->group.addr6.s6_addr[0xc],
+				       &key->group_mask.addr6.s6_addr[0xc], 4);
+}
+
+static void
+mlxsw_sp2_mr_tcam_rule_parse(struct mlxsw_sp_acl_rule *rule,
+			     struct mlxsw_sp_mr_route_key *key,
+			     unsigned int priority)
+{
+	struct mlxsw_sp_acl_rule_info *rulei;
+
+	rulei = mlxsw_sp_acl_rule_rulei(rule);
+	rulei->priority = priority;
+	mlxsw_sp_acl_rulei_keymask_u32(rulei, MLXSW_AFK_ELEMENT_VIRT_ROUTER_LSB,
+				       key->vrid, GENMASK(7, 0));
+	mlxsw_sp_acl_rulei_keymask_u32(rulei,
+				       MLXSW_AFK_ELEMENT_VIRT_ROUTER_MSB,
+				       key->vrid >> 8, GENMASK(3, 0));
+	switch (key->proto) {
+	case MLXSW_SP_L3_PROTO_IPV4:
+		return mlxsw_sp2_mr_tcam_rule_parse4(rulei, key);
+	case MLXSW_SP_L3_PROTO_IPV6:
+		return mlxsw_sp2_mr_tcam_rule_parse6(rulei, key);
+	}
+}
+
+static int
+mlxsw_sp2_mr_tcam_route_create(struct mlxsw_sp *mlxsw_sp, void *priv,
+			       void *route_priv,
+			       struct mlxsw_sp_mr_route_key *key,
+			       struct mlxsw_afa_block *afa_block,
+			       enum mlxsw_sp_mr_route_prio prio)
+{
+	struct mlxsw_sp2_mr_route *mr_route = route_priv;
+	struct mlxsw_sp2_mr_tcam *mr_tcam = priv;
+	struct mlxsw_sp_acl_ruleset *ruleset;
+	struct mlxsw_sp_acl_rule *rule;
+	int err;
+
+	mr_route->mr_tcam = mr_tcam;
+	ruleset = mlxsw_sp2_mr_tcam_proto_ruleset(mr_tcam, key->proto);
+	if (WARN_ON(!ruleset))
+		return -EINVAL;
+
+	rule = mlxsw_sp_acl_rule_create(mlxsw_sp, ruleset,
+					(unsigned long) route_priv, afa_block,
+					NULL);
+	if (IS_ERR(rule))
+		return PTR_ERR(rule);
+
+	mlxsw_sp2_mr_tcam_rule_parse(rule, key, prio);
+	err = mlxsw_sp_acl_rule_add(mlxsw_sp, rule);
+	if (err)
+		goto err_rule_add;
+
+	return 0;
+
+err_rule_add:
+	mlxsw_sp_acl_rule_destroy(mlxsw_sp, rule);
+	return err;
+}
+
+static void
+mlxsw_sp2_mr_tcam_route_destroy(struct mlxsw_sp *mlxsw_sp, void *priv,
+				void *route_priv,
+				struct mlxsw_sp_mr_route_key *key)
+{
+	struct mlxsw_sp2_mr_tcam *mr_tcam = priv;
+	struct mlxsw_sp_acl_ruleset *ruleset;
+	struct mlxsw_sp_acl_rule *rule;
+
+	ruleset = mlxsw_sp2_mr_tcam_proto_ruleset(mr_tcam, key->proto);
+	if (WARN_ON(!ruleset))
+		return;
+
+	rule = mlxsw_sp_acl_rule_lookup(mlxsw_sp, ruleset,
+					(unsigned long) route_priv);
+	if (WARN_ON(!rule))
+		return;
+
+	mlxsw_sp_acl_rule_del(mlxsw_sp, rule);
+	mlxsw_sp_acl_rule_destroy(mlxsw_sp, rule);
+}
+
+static int
+mlxsw_sp2_mr_tcam_route_update(struct mlxsw_sp *mlxsw_sp,
+			       void *route_priv,
+			       struct mlxsw_sp_mr_route_key *key,
+			       struct mlxsw_afa_block *afa_block)
+{
+	struct mlxsw_sp2_mr_route *mr_route = route_priv;
+	struct mlxsw_sp2_mr_tcam *mr_tcam = mr_route->mr_tcam;
+	struct mlxsw_sp_acl_ruleset *ruleset;
+	struct mlxsw_sp_acl_rule *rule;
+
+	ruleset = mlxsw_sp2_mr_tcam_proto_ruleset(mr_tcam, key->proto);
+	if (WARN_ON(!ruleset))
+		return -EINVAL;
+
+	rule = mlxsw_sp_acl_rule_lookup(mlxsw_sp, ruleset,
+					(unsigned long) route_priv);
+	if (WARN_ON(!rule))
+		return -EINVAL;
+
+	return mlxsw_sp_acl_rule_action_replace(mlxsw_sp, rule, afa_block);
+}
+
+static int mlxsw_sp2_mr_tcam_init(struct mlxsw_sp *mlxsw_sp, void *priv)
+{
+	struct mlxsw_sp2_mr_tcam *mr_tcam = priv;
+	int err;
+
+	mr_tcam->mlxsw_sp = mlxsw_sp;
+	mr_tcam->flow_block = mlxsw_sp_flow_block_create(mlxsw_sp, NULL);
+	if (!mr_tcam->flow_block)
+		return -ENOMEM;
+
+	err = mlxsw_sp2_mr_tcam_ipv4_init(mr_tcam);
+	if (err)
+		goto err_ipv4_init;
+
+	err = mlxsw_sp2_mr_tcam_ipv6_init(mr_tcam);
+	if (err)
+		goto err_ipv6_init;
+
+	return 0;
+
+err_ipv6_init:
+	mlxsw_sp2_mr_tcam_ipv4_fini(mr_tcam);
+err_ipv4_init:
+	mlxsw_sp_flow_block_destroy(mr_tcam->flow_block);
+	return err;
+}
+
+static void mlxsw_sp2_mr_tcam_fini(void *priv)
+{
+	struct mlxsw_sp2_mr_tcam *mr_tcam = priv;
+
+	mlxsw_sp2_mr_tcam_ipv6_fini(mr_tcam);
+	mlxsw_sp2_mr_tcam_ipv4_fini(mr_tcam);
+	mlxsw_sp_flow_block_destroy(mr_tcam->flow_block);
+}
+
+const struct mlxsw_sp_mr_tcam_ops mlxsw_sp2_mr_tcam_ops = {
+	.priv_size = sizeof(struct mlxsw_sp2_mr_tcam),
+	.init = mlxsw_sp2_mr_tcam_init,
+	.fini = mlxsw_sp2_mr_tcam_fini,
+	.route_priv_size = sizeof(struct mlxsw_sp2_mr_route),
+	.route_create = mlxsw_sp2_mr_tcam_route_create,
+	.route_destroy = mlxsw_sp2_mr_tcam_route_destroy,
+	.route_update = mlxsw_sp2_mr_tcam_route_update,
+};
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_acl.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_acl.c
new file mode 100644
index 000000000..6c5af0185
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_acl.c
@@ -0,0 +1,1124 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2017-2018 Mellanox Technologies. All rights reserved */
+
+#include <linux/kernel.h>
+#include <linux/slab.h>
+#include <linux/errno.h>
+#include <linux/list.h>
+#include <linux/string.h>
+#include <linux/rhashtable.h>
+#include <linux/netdevice.h>
+#include <linux/mutex.h>
+#include <net/net_namespace.h>
+#include <net/tc_act/tc_vlan.h>
+
+#include "reg.h"
+#include "core.h"
+#include "resources.h"
+#include "spectrum.h"
+#include "core_acl_flex_keys.h"
+#include "core_acl_flex_actions.h"
+#include "spectrum_acl_tcam.h"
+
+struct mlxsw_sp_acl {
+	struct mlxsw_sp *mlxsw_sp;
+	struct mlxsw_afk *afk;
+	struct mlxsw_sp_fid *dummy_fid;
+	struct rhashtable ruleset_ht;
+	struct list_head rules;
+	struct mutex rules_lock; /* Protects rules list */
+	struct {
+		struct delayed_work dw;
+		unsigned long interval;	/* ms */
+#define MLXSW_SP_ACL_RULE_ACTIVITY_UPDATE_PERIOD_MS 1000
+	} rule_activity_update;
+	struct mlxsw_sp_acl_tcam tcam;
+};
+
+struct mlxsw_afk *mlxsw_sp_acl_afk(struct mlxsw_sp_acl *acl)
+{
+	return acl->afk;
+}
+
+struct mlxsw_sp_acl_ruleset_ht_key {
+	struct mlxsw_sp_flow_block *block;
+	u32 chain_index;
+	const struct mlxsw_sp_acl_profile_ops *ops;
+};
+
+struct mlxsw_sp_acl_ruleset {
+	struct rhash_head ht_node; /* Member of acl HT */
+	struct mlxsw_sp_acl_ruleset_ht_key ht_key;
+	struct rhashtable rule_ht;
+	unsigned int ref_count;
+	unsigned int min_prio;
+	unsigned int max_prio;
+	unsigned long priv[];
+	/* priv has to be always the last item */
+};
+
+struct mlxsw_sp_acl_rule {
+	struct rhash_head ht_node; /* Member of rule HT */
+	struct list_head list;
+	unsigned long cookie; /* HT key */
+	struct mlxsw_sp_acl_ruleset *ruleset;
+	struct mlxsw_sp_acl_rule_info *rulei;
+	u64 last_used;
+	u64 last_packets;
+	u64 last_bytes;
+	u64 last_drops;
+	unsigned long priv[];
+	/* priv has to be always the last item */
+};
+
+static const struct rhashtable_params mlxsw_sp_acl_ruleset_ht_params = {
+	.key_len = sizeof(struct mlxsw_sp_acl_ruleset_ht_key),
+	.key_offset = offsetof(struct mlxsw_sp_acl_ruleset, ht_key),
+	.head_offset = offsetof(struct mlxsw_sp_acl_ruleset, ht_node),
+	.automatic_shrinking = true,
+};
+
+static const struct rhashtable_params mlxsw_sp_acl_rule_ht_params = {
+	.key_len = sizeof(unsigned long),
+	.key_offset = offsetof(struct mlxsw_sp_acl_rule, cookie),
+	.head_offset = offsetof(struct mlxsw_sp_acl_rule, ht_node),
+	.automatic_shrinking = true,
+};
+
+struct mlxsw_sp_fid *mlxsw_sp_acl_dummy_fid(struct mlxsw_sp *mlxsw_sp)
+{
+	return mlxsw_sp->acl->dummy_fid;
+}
+
+static bool
+mlxsw_sp_acl_ruleset_is_singular(const struct mlxsw_sp_acl_ruleset *ruleset)
+{
+	/* We hold a reference on ruleset ourselves */
+	return ruleset->ref_count == 2;
+}
+
+int mlxsw_sp_acl_ruleset_bind(struct mlxsw_sp *mlxsw_sp,
+			      struct mlxsw_sp_flow_block *block,
+			      struct mlxsw_sp_flow_block_binding *binding)
+{
+	struct mlxsw_sp_acl_ruleset *ruleset = block->ruleset_zero;
+	const struct mlxsw_sp_acl_profile_ops *ops = ruleset->ht_key.ops;
+
+	return ops->ruleset_bind(mlxsw_sp, ruleset->priv,
+				 binding->mlxsw_sp_port, binding->ingress);
+}
+
+void mlxsw_sp_acl_ruleset_unbind(struct mlxsw_sp *mlxsw_sp,
+				 struct mlxsw_sp_flow_block *block,
+				 struct mlxsw_sp_flow_block_binding *binding)
+{
+	struct mlxsw_sp_acl_ruleset *ruleset = block->ruleset_zero;
+	const struct mlxsw_sp_acl_profile_ops *ops = ruleset->ht_key.ops;
+
+	ops->ruleset_unbind(mlxsw_sp, ruleset->priv,
+			    binding->mlxsw_sp_port, binding->ingress);
+}
+
+static int
+mlxsw_sp_acl_ruleset_block_bind(struct mlxsw_sp *mlxsw_sp,
+				struct mlxsw_sp_acl_ruleset *ruleset,
+				struct mlxsw_sp_flow_block *block)
+{
+	struct mlxsw_sp_flow_block_binding *binding;
+	int err;
+
+	block->ruleset_zero = ruleset;
+	list_for_each_entry(binding, &block->binding_list, list) {
+		err = mlxsw_sp_acl_ruleset_bind(mlxsw_sp, block, binding);
+		if (err)
+			goto rollback;
+	}
+	return 0;
+
+rollback:
+	list_for_each_entry_continue_reverse(binding, &block->binding_list,
+					     list)
+		mlxsw_sp_acl_ruleset_unbind(mlxsw_sp, block, binding);
+	block->ruleset_zero = NULL;
+
+	return err;
+}
+
+static void
+mlxsw_sp_acl_ruleset_block_unbind(struct mlxsw_sp *mlxsw_sp,
+				  struct mlxsw_sp_acl_ruleset *ruleset,
+				  struct mlxsw_sp_flow_block *block)
+{
+	struct mlxsw_sp_flow_block_binding *binding;
+
+	list_for_each_entry(binding, &block->binding_list, list)
+		mlxsw_sp_acl_ruleset_unbind(mlxsw_sp, block, binding);
+	block->ruleset_zero = NULL;
+}
+
+static struct mlxsw_sp_acl_ruleset *
+mlxsw_sp_acl_ruleset_create(struct mlxsw_sp *mlxsw_sp,
+			    struct mlxsw_sp_flow_block *block, u32 chain_index,
+			    const struct mlxsw_sp_acl_profile_ops *ops,
+			    struct mlxsw_afk_element_usage *tmplt_elusage)
+{
+	struct mlxsw_sp_acl *acl = mlxsw_sp->acl;
+	struct mlxsw_sp_acl_ruleset *ruleset;
+	size_t alloc_size;
+	int err;
+
+	alloc_size = sizeof(*ruleset) + ops->ruleset_priv_size;
+	ruleset = kzalloc(alloc_size, GFP_KERNEL);
+	if (!ruleset)
+		return ERR_PTR(-ENOMEM);
+	ruleset->ref_count = 1;
+	ruleset->ht_key.block = block;
+	ruleset->ht_key.chain_index = chain_index;
+	ruleset->ht_key.ops = ops;
+
+	err = rhashtable_init(&ruleset->rule_ht, &mlxsw_sp_acl_rule_ht_params);
+	if (err)
+		goto err_rhashtable_init;
+
+	err = ops->ruleset_add(mlxsw_sp, &acl->tcam, ruleset->priv,
+			       tmplt_elusage, &ruleset->min_prio,
+			       &ruleset->max_prio);
+	if (err)
+		goto err_ops_ruleset_add;
+
+	err = rhashtable_insert_fast(&acl->ruleset_ht, &ruleset->ht_node,
+				     mlxsw_sp_acl_ruleset_ht_params);
+	if (err)
+		goto err_ht_insert;
+
+	return ruleset;
+
+err_ht_insert:
+	ops->ruleset_del(mlxsw_sp, ruleset->priv);
+err_ops_ruleset_add:
+	rhashtable_destroy(&ruleset->rule_ht);
+err_rhashtable_init:
+	kfree(ruleset);
+	return ERR_PTR(err);
+}
+
+static void mlxsw_sp_acl_ruleset_destroy(struct mlxsw_sp *mlxsw_sp,
+					 struct mlxsw_sp_acl_ruleset *ruleset)
+{
+	const struct mlxsw_sp_acl_profile_ops *ops = ruleset->ht_key.ops;
+	struct mlxsw_sp_acl *acl = mlxsw_sp->acl;
+
+	rhashtable_remove_fast(&acl->ruleset_ht, &ruleset->ht_node,
+			       mlxsw_sp_acl_ruleset_ht_params);
+	ops->ruleset_del(mlxsw_sp, ruleset->priv);
+	rhashtable_destroy(&ruleset->rule_ht);
+	kfree(ruleset);
+}
+
+static void mlxsw_sp_acl_ruleset_ref_inc(struct mlxsw_sp_acl_ruleset *ruleset)
+{
+	ruleset->ref_count++;
+}
+
+static void mlxsw_sp_acl_ruleset_ref_dec(struct mlxsw_sp *mlxsw_sp,
+					 struct mlxsw_sp_acl_ruleset *ruleset)
+{
+	if (--ruleset->ref_count)
+		return;
+	mlxsw_sp_acl_ruleset_destroy(mlxsw_sp, ruleset);
+}
+
+static struct mlxsw_sp_acl_ruleset *
+__mlxsw_sp_acl_ruleset_lookup(struct mlxsw_sp_acl *acl,
+			      struct mlxsw_sp_flow_block *block, u32 chain_index,
+			      const struct mlxsw_sp_acl_profile_ops *ops)
+{
+	struct mlxsw_sp_acl_ruleset_ht_key ht_key;
+
+	memset(&ht_key, 0, sizeof(ht_key));
+	ht_key.block = block;
+	ht_key.chain_index = chain_index;
+	ht_key.ops = ops;
+	return rhashtable_lookup_fast(&acl->ruleset_ht, &ht_key,
+				      mlxsw_sp_acl_ruleset_ht_params);
+}
+
+struct mlxsw_sp_acl_ruleset *
+mlxsw_sp_acl_ruleset_lookup(struct mlxsw_sp *mlxsw_sp,
+			    struct mlxsw_sp_flow_block *block, u32 chain_index,
+			    enum mlxsw_sp_acl_profile profile)
+{
+	const struct mlxsw_sp_acl_profile_ops *ops;
+	struct mlxsw_sp_acl *acl = mlxsw_sp->acl;
+	struct mlxsw_sp_acl_ruleset *ruleset;
+
+	ops = mlxsw_sp_acl_tcam_profile_ops(mlxsw_sp, profile);
+	if (!ops)
+		return ERR_PTR(-EINVAL);
+	ruleset = __mlxsw_sp_acl_ruleset_lookup(acl, block, chain_index, ops);
+	if (!ruleset)
+		return ERR_PTR(-ENOENT);
+	return ruleset;
+}
+
+struct mlxsw_sp_acl_ruleset *
+mlxsw_sp_acl_ruleset_get(struct mlxsw_sp *mlxsw_sp,
+			 struct mlxsw_sp_flow_block *block, u32 chain_index,
+			 enum mlxsw_sp_acl_profile profile,
+			 struct mlxsw_afk_element_usage *tmplt_elusage)
+{
+	const struct mlxsw_sp_acl_profile_ops *ops;
+	struct mlxsw_sp_acl *acl = mlxsw_sp->acl;
+	struct mlxsw_sp_acl_ruleset *ruleset;
+
+	ops = mlxsw_sp_acl_tcam_profile_ops(mlxsw_sp, profile);
+	if (!ops)
+		return ERR_PTR(-EINVAL);
+
+	ruleset = __mlxsw_sp_acl_ruleset_lookup(acl, block, chain_index, ops);
+	if (ruleset) {
+		mlxsw_sp_acl_ruleset_ref_inc(ruleset);
+		return ruleset;
+	}
+	return mlxsw_sp_acl_ruleset_create(mlxsw_sp, block, chain_index, ops,
+					   tmplt_elusage);
+}
+
+void mlxsw_sp_acl_ruleset_put(struct mlxsw_sp *mlxsw_sp,
+			      struct mlxsw_sp_acl_ruleset *ruleset)
+{
+	mlxsw_sp_acl_ruleset_ref_dec(mlxsw_sp, ruleset);
+}
+
+u16 mlxsw_sp_acl_ruleset_group_id(struct mlxsw_sp_acl_ruleset *ruleset)
+{
+	const struct mlxsw_sp_acl_profile_ops *ops = ruleset->ht_key.ops;
+
+	return ops->ruleset_group_id(ruleset->priv);
+}
+
+void mlxsw_sp_acl_ruleset_prio_get(struct mlxsw_sp_acl_ruleset *ruleset,
+				   unsigned int *p_min_prio,
+				   unsigned int *p_max_prio)
+{
+	*p_min_prio = ruleset->min_prio;
+	*p_max_prio = ruleset->max_prio;
+}
+
+struct mlxsw_sp_acl_rule_info *
+mlxsw_sp_acl_rulei_create(struct mlxsw_sp_acl *acl,
+			  struct mlxsw_afa_block *afa_block)
+{
+	struct mlxsw_sp_acl_rule_info *rulei;
+	int err;
+
+	rulei = kzalloc(sizeof(*rulei), GFP_KERNEL);
+	if (!rulei)
+		return ERR_PTR(-ENOMEM);
+
+	if (afa_block) {
+		rulei->act_block = afa_block;
+		return rulei;
+	}
+
+	rulei->act_block = mlxsw_afa_block_create(acl->mlxsw_sp->afa);
+	if (IS_ERR(rulei->act_block)) {
+		err = PTR_ERR(rulei->act_block);
+		goto err_afa_block_create;
+	}
+	rulei->action_created = 1;
+	return rulei;
+
+err_afa_block_create:
+	kfree(rulei);
+	return ERR_PTR(err);
+}
+
+void mlxsw_sp_acl_rulei_destroy(struct mlxsw_sp_acl_rule_info *rulei)
+{
+	if (rulei->action_created)
+		mlxsw_afa_block_destroy(rulei->act_block);
+	kfree(rulei);
+}
+
+int mlxsw_sp_acl_rulei_commit(struct mlxsw_sp_acl_rule_info *rulei)
+{
+	return mlxsw_afa_block_commit(rulei->act_block);
+}
+
+void mlxsw_sp_acl_rulei_priority(struct mlxsw_sp_acl_rule_info *rulei,
+				 unsigned int priority)
+{
+	rulei->priority = priority;
+}
+
+void mlxsw_sp_acl_rulei_keymask_u32(struct mlxsw_sp_acl_rule_info *rulei,
+				    enum mlxsw_afk_element element,
+				    u32 key_value, u32 mask_value)
+{
+	mlxsw_afk_values_add_u32(&rulei->values, element,
+				 key_value, mask_value);
+}
+
+void mlxsw_sp_acl_rulei_keymask_buf(struct mlxsw_sp_acl_rule_info *rulei,
+				    enum mlxsw_afk_element element,
+				    const char *key_value,
+				    const char *mask_value, unsigned int len)
+{
+	mlxsw_afk_values_add_buf(&rulei->values, element,
+				 key_value, mask_value, len);
+}
+
+int mlxsw_sp_acl_rulei_act_continue(struct mlxsw_sp_acl_rule_info *rulei)
+{
+	return mlxsw_afa_block_continue(rulei->act_block);
+}
+
+int mlxsw_sp_acl_rulei_act_jump(struct mlxsw_sp_acl_rule_info *rulei,
+				u16 group_id)
+{
+	return mlxsw_afa_block_jump(rulei->act_block, group_id);
+}
+
+int mlxsw_sp_acl_rulei_act_terminate(struct mlxsw_sp_acl_rule_info *rulei)
+{
+	return mlxsw_afa_block_terminate(rulei->act_block);
+}
+
+int mlxsw_sp_acl_rulei_act_drop(struct mlxsw_sp_acl_rule_info *rulei,
+				bool ingress,
+				const struct flow_action_cookie *fa_cookie,
+				struct netlink_ext_ack *extack)
+{
+	return mlxsw_afa_block_append_drop(rulei->act_block, ingress,
+					   fa_cookie, extack);
+}
+
+int mlxsw_sp_acl_rulei_act_trap(struct mlxsw_sp_acl_rule_info *rulei)
+{
+	return mlxsw_afa_block_append_trap(rulei->act_block,
+					   MLXSW_TRAP_ID_ACL0);
+}
+
+int mlxsw_sp_acl_rulei_act_fwd(struct mlxsw_sp *mlxsw_sp,
+			       struct mlxsw_sp_acl_rule_info *rulei,
+			       struct net_device *out_dev,
+			       struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port;
+	u16 local_port;
+	bool in_port;
+
+	if (out_dev) {
+		if (!mlxsw_sp_port_dev_check(out_dev)) {
+			NL_SET_ERR_MSG_MOD(extack, "Invalid output device");
+			return -EINVAL;
+		}
+		mlxsw_sp_port = netdev_priv(out_dev);
+		if (mlxsw_sp_port->mlxsw_sp != mlxsw_sp) {
+			NL_SET_ERR_MSG_MOD(extack, "Invalid output device");
+			return -EINVAL;
+		}
+		local_port = mlxsw_sp_port->local_port;
+		in_port = false;
+	} else {
+		/* If out_dev is NULL, the caller wants to
+		 * set forward to ingress port.
+		 */
+		local_port = 0;
+		in_port = true;
+	}
+	return mlxsw_afa_block_append_fwd(rulei->act_block,
+					  local_port, in_port, extack);
+}
+
+int mlxsw_sp_acl_rulei_act_mirror(struct mlxsw_sp *mlxsw_sp,
+				  struct mlxsw_sp_acl_rule_info *rulei,
+				  struct mlxsw_sp_flow_block *block,
+				  struct net_device *out_dev,
+				  struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp_flow_block_binding *binding;
+	struct mlxsw_sp_port *in_port;
+
+	if (!list_is_singular(&block->binding_list)) {
+		NL_SET_ERR_MSG_MOD(extack, "Only a single mirror source is allowed");
+		return -EOPNOTSUPP;
+	}
+	binding = list_first_entry(&block->binding_list,
+				   struct mlxsw_sp_flow_block_binding, list);
+	in_port = binding->mlxsw_sp_port;
+
+	return mlxsw_afa_block_append_mirror(rulei->act_block,
+					     in_port->local_port,
+					     out_dev,
+					     binding->ingress,
+					     extack);
+}
+
+int mlxsw_sp_acl_rulei_act_vlan(struct mlxsw_sp *mlxsw_sp,
+				struct mlxsw_sp_acl_rule_info *rulei,
+				u32 action, u16 vid, u16 proto, u8 prio,
+				struct netlink_ext_ack *extack)
+{
+	u8 ethertype;
+
+	if (action == FLOW_ACTION_VLAN_MANGLE) {
+		switch (proto) {
+		case ETH_P_8021Q:
+			ethertype = 0;
+			break;
+		case ETH_P_8021AD:
+			ethertype = 1;
+			break;
+		default:
+			NL_SET_ERR_MSG_MOD(extack, "Unsupported VLAN protocol");
+			dev_err(mlxsw_sp->bus_info->dev, "Unsupported VLAN protocol %#04x\n",
+				proto);
+			return -EINVAL;
+		}
+
+		return mlxsw_afa_block_append_vlan_modify(rulei->act_block,
+							  vid, prio, ethertype,
+							  extack);
+	} else {
+		NL_SET_ERR_MSG_MOD(extack, "Unsupported VLAN action");
+		dev_err(mlxsw_sp->bus_info->dev, "Unsupported VLAN action\n");
+		return -EINVAL;
+	}
+}
+
+int mlxsw_sp_acl_rulei_act_priority(struct mlxsw_sp *mlxsw_sp,
+				    struct mlxsw_sp_acl_rule_info *rulei,
+				    u32 prio, struct netlink_ext_ack *extack)
+{
+	/* Even though both Linux and Spectrum switches support 16 priorities,
+	 * spectrum_qdisc only processes the first eight priomap elements, and
+	 * the DCB and PFC features are tied to 8 priorities as well. Therefore
+	 * bounce attempts to prioritize packets to higher priorities.
+	 */
+	if (prio >= IEEE_8021QAZ_MAX_TCS) {
+		NL_SET_ERR_MSG_MOD(extack, "Only priorities 0..7 are supported");
+		return -EINVAL;
+	}
+	return mlxsw_afa_block_append_qos_switch_prio(rulei->act_block, prio,
+						      extack);
+}
+
+struct mlxsw_sp_acl_mangle_action {
+	enum flow_action_mangle_base htype;
+	/* Offset is u32-aligned. */
+	u32 offset;
+	/* Mask bits are unset for the modified field. */
+	u32 mask;
+	/* Shift required to extract the set value. */
+	u32 shift;
+	enum mlxsw_sp_acl_mangle_field field;
+};
+
+#define MLXSW_SP_ACL_MANGLE_ACTION(_htype, _offset, _mask, _shift, _field) \
+	{								\
+		.htype = _htype,					\
+		.offset = _offset,					\
+		.mask = _mask,						\
+		.shift = _shift,					\
+		.field = MLXSW_SP_ACL_MANGLE_FIELD_##_field,		\
+	}
+
+#define MLXSW_SP_ACL_MANGLE_ACTION_IP4(_offset, _mask, _shift, _field) \
+	MLXSW_SP_ACL_MANGLE_ACTION(FLOW_ACT_MANGLE_HDR_TYPE_IP4,       \
+				   _offset, _mask, _shift, _field)
+
+#define MLXSW_SP_ACL_MANGLE_ACTION_IP6(_offset, _mask, _shift, _field) \
+	MLXSW_SP_ACL_MANGLE_ACTION(FLOW_ACT_MANGLE_HDR_TYPE_IP6,       \
+				   _offset, _mask, _shift, _field)
+
+#define MLXSW_SP_ACL_MANGLE_ACTION_TCP(_offset, _mask, _shift, _field) \
+	MLXSW_SP_ACL_MANGLE_ACTION(FLOW_ACT_MANGLE_HDR_TYPE_TCP, _offset, _mask, _shift, _field)
+
+#define MLXSW_SP_ACL_MANGLE_ACTION_UDP(_offset, _mask, _shift, _field) \
+	MLXSW_SP_ACL_MANGLE_ACTION(FLOW_ACT_MANGLE_HDR_TYPE_UDP, _offset, _mask, _shift, _field)
+
+static struct mlxsw_sp_acl_mangle_action mlxsw_sp_acl_mangle_actions[] = {
+	MLXSW_SP_ACL_MANGLE_ACTION_IP4(0, 0xff00ffff, 16, IP_DSFIELD),
+	MLXSW_SP_ACL_MANGLE_ACTION_IP4(0, 0xff03ffff, 18, IP_DSCP),
+	MLXSW_SP_ACL_MANGLE_ACTION_IP4(0, 0xfffcffff, 16, IP_ECN),
+
+	MLXSW_SP_ACL_MANGLE_ACTION_IP6(0, 0xf00fffff, 20, IP_DSFIELD),
+	MLXSW_SP_ACL_MANGLE_ACTION_IP6(0, 0xf03fffff, 22, IP_DSCP),
+	MLXSW_SP_ACL_MANGLE_ACTION_IP6(0, 0xffcfffff, 20, IP_ECN),
+
+	MLXSW_SP_ACL_MANGLE_ACTION_TCP(0, 0x0000ffff, 16, IP_SPORT),
+	MLXSW_SP_ACL_MANGLE_ACTION_TCP(0, 0xffff0000, 0,  IP_DPORT),
+
+	MLXSW_SP_ACL_MANGLE_ACTION_UDP(0, 0x0000ffff, 16, IP_SPORT),
+	MLXSW_SP_ACL_MANGLE_ACTION_UDP(0, 0xffff0000, 0,  IP_DPORT),
+
+	MLXSW_SP_ACL_MANGLE_ACTION_IP4(12, 0x00000000, 0, IP4_SIP),
+	MLXSW_SP_ACL_MANGLE_ACTION_IP4(16, 0x00000000, 0, IP4_DIP),
+
+	MLXSW_SP_ACL_MANGLE_ACTION_IP6(8, 0x00000000, 0, IP6_SIP_1),
+	MLXSW_SP_ACL_MANGLE_ACTION_IP6(12, 0x00000000, 0, IP6_SIP_2),
+	MLXSW_SP_ACL_MANGLE_ACTION_IP6(16, 0x00000000, 0, IP6_SIP_3),
+	MLXSW_SP_ACL_MANGLE_ACTION_IP6(20, 0x00000000, 0, IP6_SIP_4),
+	MLXSW_SP_ACL_MANGLE_ACTION_IP6(24, 0x00000000, 0, IP6_DIP_1),
+	MLXSW_SP_ACL_MANGLE_ACTION_IP6(28, 0x00000000, 0, IP6_DIP_2),
+	MLXSW_SP_ACL_MANGLE_ACTION_IP6(32, 0x00000000, 0, IP6_DIP_3),
+	MLXSW_SP_ACL_MANGLE_ACTION_IP6(36, 0x00000000, 0, IP6_DIP_4),
+};
+
+static int
+mlxsw_sp_acl_rulei_act_mangle_field(struct mlxsw_sp *mlxsw_sp,
+				    struct mlxsw_sp_acl_rule_info *rulei,
+				    struct mlxsw_sp_acl_mangle_action *mact,
+				    u32 val, struct netlink_ext_ack *extack)
+{
+	switch (mact->field) {
+	case MLXSW_SP_ACL_MANGLE_FIELD_IP_DSFIELD:
+		return mlxsw_afa_block_append_qos_dsfield(rulei->act_block,
+							  val, extack);
+	case MLXSW_SP_ACL_MANGLE_FIELD_IP_DSCP:
+		return mlxsw_afa_block_append_qos_dscp(rulei->act_block,
+						       val, extack);
+	case MLXSW_SP_ACL_MANGLE_FIELD_IP_ECN:
+		return mlxsw_afa_block_append_qos_ecn(rulei->act_block,
+						      val, extack);
+	default:
+		return -EOPNOTSUPP;
+	}
+}
+
+static int mlxsw_sp1_acl_rulei_act_mangle_field(struct mlxsw_sp *mlxsw_sp,
+						struct mlxsw_sp_acl_rule_info *rulei,
+						struct mlxsw_sp_acl_mangle_action *mact,
+						u32 val, struct netlink_ext_ack *extack)
+{
+	int err;
+
+	err = mlxsw_sp_acl_rulei_act_mangle_field(mlxsw_sp, rulei, mact, val, extack);
+	if (err != -EOPNOTSUPP)
+		return err;
+
+	NL_SET_ERR_MSG_MOD(extack, "Unsupported mangle field");
+	return err;
+}
+
+static int
+mlxsw_sp2_acl_rulei_act_mangle_field_ip_odd(struct mlxsw_sp_acl_rule_info *rulei,
+					    enum mlxsw_sp_acl_mangle_field field,
+					    u32 val, struct netlink_ext_ack *extack)
+{
+	if (!rulei->ipv6_valid) {
+		rulei->ipv6.prev_val = val;
+		rulei->ipv6_valid = true;
+		rulei->ipv6.prev_field = field;
+		return 0;
+	}
+
+	NL_SET_ERR_MSG_MOD(extack, "Unsupported mangle field order");
+	return -EOPNOTSUPP;
+}
+
+static int mlxsw_sp2_acl_rulei_act_mangle_field(struct mlxsw_sp *mlxsw_sp,
+						struct mlxsw_sp_acl_rule_info *rulei,
+						struct mlxsw_sp_acl_mangle_action *mact,
+						u32 val, struct netlink_ext_ack *extack)
+{
+	int err;
+
+	err = mlxsw_sp_acl_rulei_act_mangle_field(mlxsw_sp, rulei, mact, val, extack);
+	if (err != -EOPNOTSUPP)
+		return err;
+
+	switch (mact->field) {
+	case MLXSW_SP_ACL_MANGLE_FIELD_IP_SPORT:
+		return mlxsw_afa_block_append_l4port(rulei->act_block, false, val, extack);
+	case MLXSW_SP_ACL_MANGLE_FIELD_IP_DPORT:
+		return mlxsw_afa_block_append_l4port(rulei->act_block, true, val, extack);
+	/* IPv4 fields */
+	case MLXSW_SP_ACL_MANGLE_FIELD_IP4_SIP:
+		return mlxsw_afa_block_append_ip(rulei->act_block, false,
+						 true, val, 0, extack);
+	case MLXSW_SP_ACL_MANGLE_FIELD_IP4_DIP:
+		return mlxsw_afa_block_append_ip(rulei->act_block, true,
+						 true, val, 0, extack);
+	/* IPv6 fields */
+	case MLXSW_SP_ACL_MANGLE_FIELD_IP6_SIP_1:
+	case MLXSW_SP_ACL_MANGLE_FIELD_IP6_SIP_3:
+	case MLXSW_SP_ACL_MANGLE_FIELD_IP6_DIP_1:
+	case MLXSW_SP_ACL_MANGLE_FIELD_IP6_DIP_3:
+		return mlxsw_sp2_acl_rulei_act_mangle_field_ip_odd(rulei,
+								   mact->field,
+								   val, extack);
+	case MLXSW_SP_ACL_MANGLE_FIELD_IP6_SIP_2:
+		if (rulei->ipv6_valid &&
+		    rulei->ipv6.prev_field == MLXSW_SP_ACL_MANGLE_FIELD_IP6_SIP_1) {
+			rulei->ipv6_valid = false;
+			return mlxsw_afa_block_append_ip(rulei->act_block,
+							 false, false, val,
+							 rulei->ipv6.prev_val,
+							 extack);
+		}
+		break;
+	case MLXSW_SP_ACL_MANGLE_FIELD_IP6_SIP_4:
+		if (rulei->ipv6_valid &&
+		    rulei->ipv6.prev_field == MLXSW_SP_ACL_MANGLE_FIELD_IP6_SIP_3) {
+			rulei->ipv6_valid = false;
+			return mlxsw_afa_block_append_ip(rulei->act_block,
+							 false, true, val,
+							 rulei->ipv6.prev_val,
+							 extack);
+		}
+		break;
+	case MLXSW_SP_ACL_MANGLE_FIELD_IP6_DIP_2:
+		if (rulei->ipv6_valid &&
+		    rulei->ipv6.prev_field == MLXSW_SP_ACL_MANGLE_FIELD_IP6_DIP_1) {
+			rulei->ipv6_valid = false;
+			return mlxsw_afa_block_append_ip(rulei->act_block,
+							 true, false, val,
+							 rulei->ipv6.prev_val,
+							 extack);
+		}
+		break;
+	case MLXSW_SP_ACL_MANGLE_FIELD_IP6_DIP_4:
+		if (rulei->ipv6_valid &&
+		    rulei->ipv6.prev_field == MLXSW_SP_ACL_MANGLE_FIELD_IP6_DIP_3) {
+			rulei->ipv6_valid = false;
+			return mlxsw_afa_block_append_ip(rulei->act_block,
+							 true, true, val,
+							 rulei->ipv6.prev_val,
+							 extack);
+		}
+		break;
+	default:
+		break;
+	}
+
+	NL_SET_ERR_MSG_MOD(extack, "Unsupported mangle field");
+	return err;
+}
+
+int mlxsw_sp_acl_rulei_act_mangle(struct mlxsw_sp *mlxsw_sp,
+				  struct mlxsw_sp_acl_rule_info *rulei,
+				  enum flow_action_mangle_base htype,
+				  u32 offset, u32 mask, u32 val,
+				  struct netlink_ext_ack *extack)
+{
+	const struct mlxsw_sp_acl_rulei_ops *acl_rulei_ops = mlxsw_sp->acl_rulei_ops;
+	struct mlxsw_sp_acl_mangle_action *mact;
+	size_t i;
+
+	for (i = 0; i < ARRAY_SIZE(mlxsw_sp_acl_mangle_actions); ++i) {
+		mact = &mlxsw_sp_acl_mangle_actions[i];
+		if (mact->htype == htype &&
+		    mact->offset == offset &&
+		    mact->mask == mask) {
+			val >>= mact->shift;
+			return acl_rulei_ops->act_mangle_field(mlxsw_sp,
+							       rulei, mact,
+							       val, extack);
+		}
+	}
+
+	NL_SET_ERR_MSG_MOD(extack, "Unknown mangle field");
+	return -EINVAL;
+}
+
+int mlxsw_sp_acl_rulei_act_police(struct mlxsw_sp *mlxsw_sp,
+				  struct mlxsw_sp_acl_rule_info *rulei,
+				  u32 index, u64 rate_bytes_ps,
+				  u32 burst, struct netlink_ext_ack *extack)
+{
+	int err;
+
+	err = mlxsw_afa_block_append_police(rulei->act_block, index,
+					    rate_bytes_ps, burst,
+					    &rulei->policer_index, extack);
+	if (err)
+		return err;
+
+	rulei->policer_index_valid = true;
+
+	return 0;
+}
+
+int mlxsw_sp_acl_rulei_act_count(struct mlxsw_sp *mlxsw_sp,
+				 struct mlxsw_sp_acl_rule_info *rulei,
+				 struct netlink_ext_ack *extack)
+{
+	int err;
+
+	err = mlxsw_afa_block_append_counter(rulei->act_block,
+					     &rulei->counter_index, extack);
+	if (err)
+		return err;
+	rulei->counter_valid = true;
+	return 0;
+}
+
+int mlxsw_sp_acl_rulei_act_fid_set(struct mlxsw_sp *mlxsw_sp,
+				   struct mlxsw_sp_acl_rule_info *rulei,
+				   u16 fid, struct netlink_ext_ack *extack)
+{
+	return mlxsw_afa_block_append_fid_set(rulei->act_block, fid, extack);
+}
+
+int mlxsw_sp_acl_rulei_act_sample(struct mlxsw_sp *mlxsw_sp,
+				  struct mlxsw_sp_acl_rule_info *rulei,
+				  struct mlxsw_sp_flow_block *block,
+				  struct psample_group *psample_group, u32 rate,
+				  u32 trunc_size, bool truncate,
+				  struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp_flow_block_binding *binding;
+	struct mlxsw_sp_port *mlxsw_sp_port;
+
+	if (!list_is_singular(&block->binding_list)) {
+		NL_SET_ERR_MSG_MOD(extack, "Only a single sampling source is allowed");
+		return -EOPNOTSUPP;
+	}
+	binding = list_first_entry(&block->binding_list,
+				   struct mlxsw_sp_flow_block_binding, list);
+	mlxsw_sp_port = binding->mlxsw_sp_port;
+
+	return mlxsw_afa_block_append_sampler(rulei->act_block,
+					      mlxsw_sp_port->local_port,
+					      psample_group, rate, trunc_size,
+					      truncate, binding->ingress,
+					      extack);
+}
+
+struct mlxsw_sp_acl_rule *
+mlxsw_sp_acl_rule_create(struct mlxsw_sp *mlxsw_sp,
+			 struct mlxsw_sp_acl_ruleset *ruleset,
+			 unsigned long cookie,
+			 struct mlxsw_afa_block *afa_block,
+			 struct netlink_ext_ack *extack)
+{
+	const struct mlxsw_sp_acl_profile_ops *ops = ruleset->ht_key.ops;
+	struct mlxsw_sp_acl_rule *rule;
+	int err;
+
+	mlxsw_sp_acl_ruleset_ref_inc(ruleset);
+	rule = kzalloc(sizeof(*rule) + ops->rule_priv_size,
+		       GFP_KERNEL);
+	if (!rule) {
+		err = -ENOMEM;
+		goto err_alloc;
+	}
+	rule->cookie = cookie;
+	rule->ruleset = ruleset;
+
+	rule->rulei = mlxsw_sp_acl_rulei_create(mlxsw_sp->acl, afa_block);
+	if (IS_ERR(rule->rulei)) {
+		err = PTR_ERR(rule->rulei);
+		goto err_rulei_create;
+	}
+
+	return rule;
+
+err_rulei_create:
+	kfree(rule);
+err_alloc:
+	mlxsw_sp_acl_ruleset_ref_dec(mlxsw_sp, ruleset);
+	return ERR_PTR(err);
+}
+
+void mlxsw_sp_acl_rule_destroy(struct mlxsw_sp *mlxsw_sp,
+			       struct mlxsw_sp_acl_rule *rule)
+{
+	struct mlxsw_sp_acl_ruleset *ruleset = rule->ruleset;
+
+	mlxsw_sp_acl_rulei_destroy(rule->rulei);
+	kfree(rule);
+	mlxsw_sp_acl_ruleset_ref_dec(mlxsw_sp, ruleset);
+}
+
+int mlxsw_sp_acl_rule_add(struct mlxsw_sp *mlxsw_sp,
+			  struct mlxsw_sp_acl_rule *rule)
+{
+	struct mlxsw_sp_acl_ruleset *ruleset = rule->ruleset;
+	const struct mlxsw_sp_acl_profile_ops *ops = ruleset->ht_key.ops;
+	struct mlxsw_sp_flow_block *block = ruleset->ht_key.block;
+	int err;
+
+	err = ops->rule_add(mlxsw_sp, ruleset->priv, rule->priv, rule->rulei);
+	if (err)
+		return err;
+
+	err = rhashtable_insert_fast(&ruleset->rule_ht, &rule->ht_node,
+				     mlxsw_sp_acl_rule_ht_params);
+	if (err)
+		goto err_rhashtable_insert;
+
+	if (!ruleset->ht_key.chain_index &&
+	    mlxsw_sp_acl_ruleset_is_singular(ruleset)) {
+		/* We only need ruleset with chain index 0, the implicit
+		 * one, to be directly bound to device. The rest of the
+		 * rulesets are bound by "Goto action set".
+		 */
+		err = mlxsw_sp_acl_ruleset_block_bind(mlxsw_sp, ruleset, block);
+		if (err)
+			goto err_ruleset_block_bind;
+	}
+
+	mutex_lock(&mlxsw_sp->acl->rules_lock);
+	list_add_tail(&rule->list, &mlxsw_sp->acl->rules);
+	mutex_unlock(&mlxsw_sp->acl->rules_lock);
+	block->rule_count++;
+	block->ingress_blocker_rule_count += rule->rulei->ingress_bind_blocker;
+	block->egress_blocker_rule_count += rule->rulei->egress_bind_blocker;
+	return 0;
+
+err_ruleset_block_bind:
+	rhashtable_remove_fast(&ruleset->rule_ht, &rule->ht_node,
+			       mlxsw_sp_acl_rule_ht_params);
+err_rhashtable_insert:
+	ops->rule_del(mlxsw_sp, rule->priv);
+	return err;
+}
+
+void mlxsw_sp_acl_rule_del(struct mlxsw_sp *mlxsw_sp,
+			   struct mlxsw_sp_acl_rule *rule)
+{
+	struct mlxsw_sp_acl_ruleset *ruleset = rule->ruleset;
+	const struct mlxsw_sp_acl_profile_ops *ops = ruleset->ht_key.ops;
+	struct mlxsw_sp_flow_block *block = ruleset->ht_key.block;
+
+	block->egress_blocker_rule_count -= rule->rulei->egress_bind_blocker;
+	block->ingress_blocker_rule_count -= rule->rulei->ingress_bind_blocker;
+	block->rule_count--;
+	mutex_lock(&mlxsw_sp->acl->rules_lock);
+	list_del(&rule->list);
+	mutex_unlock(&mlxsw_sp->acl->rules_lock);
+	if (!ruleset->ht_key.chain_index &&
+	    mlxsw_sp_acl_ruleset_is_singular(ruleset))
+		mlxsw_sp_acl_ruleset_block_unbind(mlxsw_sp, ruleset, block);
+	rhashtable_remove_fast(&ruleset->rule_ht, &rule->ht_node,
+			       mlxsw_sp_acl_rule_ht_params);
+	ops->rule_del(mlxsw_sp, rule->priv);
+}
+
+int mlxsw_sp_acl_rule_action_replace(struct mlxsw_sp *mlxsw_sp,
+				     struct mlxsw_sp_acl_rule *rule,
+				     struct mlxsw_afa_block *afa_block)
+{
+	struct mlxsw_sp_acl_ruleset *ruleset = rule->ruleset;
+	const struct mlxsw_sp_acl_profile_ops *ops = ruleset->ht_key.ops;
+	struct mlxsw_sp_acl_rule_info *rulei;
+
+	rulei = mlxsw_sp_acl_rule_rulei(rule);
+	rulei->act_block = afa_block;
+
+	return ops->rule_action_replace(mlxsw_sp, rule->priv, rule->rulei);
+}
+
+struct mlxsw_sp_acl_rule *
+mlxsw_sp_acl_rule_lookup(struct mlxsw_sp *mlxsw_sp,
+			 struct mlxsw_sp_acl_ruleset *ruleset,
+			 unsigned long cookie)
+{
+	return rhashtable_lookup_fast(&ruleset->rule_ht, &cookie,
+				       mlxsw_sp_acl_rule_ht_params);
+}
+
+struct mlxsw_sp_acl_rule_info *
+mlxsw_sp_acl_rule_rulei(struct mlxsw_sp_acl_rule *rule)
+{
+	return rule->rulei;
+}
+
+static int mlxsw_sp_acl_rule_activity_update(struct mlxsw_sp *mlxsw_sp,
+					     struct mlxsw_sp_acl_rule *rule)
+{
+	struct mlxsw_sp_acl_ruleset *ruleset = rule->ruleset;
+	const struct mlxsw_sp_acl_profile_ops *ops = ruleset->ht_key.ops;
+	bool active;
+	int err;
+
+	err = ops->rule_activity_get(mlxsw_sp, rule->priv, &active);
+	if (err)
+		return err;
+	if (active)
+		rule->last_used = jiffies;
+	return 0;
+}
+
+static int mlxsw_sp_acl_rules_activity_update(struct mlxsw_sp_acl *acl)
+{
+	struct mlxsw_sp_acl_rule *rule;
+	int err;
+
+	mutex_lock(&acl->rules_lock);
+	list_for_each_entry(rule, &acl->rules, list) {
+		err = mlxsw_sp_acl_rule_activity_update(acl->mlxsw_sp,
+							rule);
+		if (err)
+			goto err_rule_update;
+	}
+	mutex_unlock(&acl->rules_lock);
+	return 0;
+
+err_rule_update:
+	mutex_unlock(&acl->rules_lock);
+	return err;
+}
+
+static void mlxsw_sp_acl_rule_activity_work_schedule(struct mlxsw_sp_acl *acl)
+{
+	unsigned long interval = acl->rule_activity_update.interval;
+
+	mlxsw_core_schedule_dw(&acl->rule_activity_update.dw,
+			       msecs_to_jiffies(interval));
+}
+
+static void mlxsw_sp_acl_rule_activity_update_work(struct work_struct *work)
+{
+	struct mlxsw_sp_acl *acl = container_of(work, struct mlxsw_sp_acl,
+						rule_activity_update.dw.work);
+	int err;
+
+	err = mlxsw_sp_acl_rules_activity_update(acl);
+	if (err)
+		dev_err(acl->mlxsw_sp->bus_info->dev, "Could not update acl activity");
+
+	mlxsw_sp_acl_rule_activity_work_schedule(acl);
+}
+
+int mlxsw_sp_acl_rule_get_stats(struct mlxsw_sp *mlxsw_sp,
+				struct mlxsw_sp_acl_rule *rule,
+				u64 *packets, u64 *bytes, u64 *drops,
+				u64 *last_use,
+				enum flow_action_hw_stats *used_hw_stats)
+
+{
+	enum mlxsw_sp_policer_type type = MLXSW_SP_POLICER_TYPE_SINGLE_RATE;
+	struct mlxsw_sp_acl_rule_info *rulei;
+	u64 current_packets = 0;
+	u64 current_bytes = 0;
+	u64 current_drops = 0;
+	int err;
+
+	rulei = mlxsw_sp_acl_rule_rulei(rule);
+	if (rulei->counter_valid) {
+		err = mlxsw_sp_flow_counter_get(mlxsw_sp, rulei->counter_index,
+						&current_packets,
+						&current_bytes);
+		if (err)
+			return err;
+		*used_hw_stats = FLOW_ACTION_HW_STATS_IMMEDIATE;
+	}
+	if (rulei->policer_index_valid) {
+		err = mlxsw_sp_policer_drops_counter_get(mlxsw_sp, type,
+							 rulei->policer_index,
+							 &current_drops);
+		if (err)
+			return err;
+	}
+	*packets = current_packets - rule->last_packets;
+	*bytes = current_bytes - rule->last_bytes;
+	*drops = current_drops - rule->last_drops;
+	*last_use = rule->last_used;
+
+	rule->last_bytes = current_bytes;
+	rule->last_packets = current_packets;
+	rule->last_drops = current_drops;
+
+	return 0;
+}
+
+int mlxsw_sp_acl_init(struct mlxsw_sp *mlxsw_sp)
+{
+	struct mlxsw_sp_fid *fid;
+	struct mlxsw_sp_acl *acl;
+	size_t alloc_size;
+	int err;
+
+	alloc_size = sizeof(*acl) + mlxsw_sp_acl_tcam_priv_size(mlxsw_sp);
+	acl = kzalloc(alloc_size, GFP_KERNEL);
+	if (!acl)
+		return -ENOMEM;
+	mlxsw_sp->acl = acl;
+	acl->mlxsw_sp = mlxsw_sp;
+	acl->afk = mlxsw_afk_create(MLXSW_CORE_RES_GET(mlxsw_sp->core,
+						       ACL_FLEX_KEYS),
+				    mlxsw_sp->afk_ops);
+	if (!acl->afk) {
+		err = -ENOMEM;
+		goto err_afk_create;
+	}
+
+	err = rhashtable_init(&acl->ruleset_ht,
+			      &mlxsw_sp_acl_ruleset_ht_params);
+	if (err)
+		goto err_rhashtable_init;
+
+	fid = mlxsw_sp_fid_dummy_get(mlxsw_sp);
+	if (IS_ERR(fid)) {
+		err = PTR_ERR(fid);
+		goto err_fid_get;
+	}
+	acl->dummy_fid = fid;
+
+	INIT_LIST_HEAD(&acl->rules);
+	mutex_init(&acl->rules_lock);
+	err = mlxsw_sp_acl_tcam_init(mlxsw_sp, &acl->tcam);
+	if (err)
+		goto err_acl_ops_init;
+
+	/* Create the delayed work for the rule activity_update */
+	INIT_DELAYED_WORK(&acl->rule_activity_update.dw,
+			  mlxsw_sp_acl_rule_activity_update_work);
+	acl->rule_activity_update.interval = MLXSW_SP_ACL_RULE_ACTIVITY_UPDATE_PERIOD_MS;
+	mlxsw_core_schedule_dw(&acl->rule_activity_update.dw, 0);
+	return 0;
+
+err_acl_ops_init:
+	mutex_destroy(&acl->rules_lock);
+	mlxsw_sp_fid_put(fid);
+err_fid_get:
+	rhashtable_destroy(&acl->ruleset_ht);
+err_rhashtable_init:
+	mlxsw_afk_destroy(acl->afk);
+err_afk_create:
+	kfree(acl);
+	return err;
+}
+
+void mlxsw_sp_acl_fini(struct mlxsw_sp *mlxsw_sp)
+{
+	struct mlxsw_sp_acl *acl = mlxsw_sp->acl;
+
+	cancel_delayed_work_sync(&mlxsw_sp->acl->rule_activity_update.dw);
+	mlxsw_sp_acl_tcam_fini(mlxsw_sp, &acl->tcam);
+	mutex_destroy(&acl->rules_lock);
+	WARN_ON(!list_empty(&acl->rules));
+	mlxsw_sp_fid_put(acl->dummy_fid);
+	rhashtable_destroy(&acl->ruleset_ht);
+	mlxsw_afk_destroy(acl->afk);
+	kfree(acl);
+}
+
+u32 mlxsw_sp_acl_region_rehash_intrvl_get(struct mlxsw_sp *mlxsw_sp)
+{
+	struct mlxsw_sp_acl *acl = mlxsw_sp->acl;
+
+	return mlxsw_sp_acl_tcam_vregion_rehash_intrvl_get(mlxsw_sp,
+							   &acl->tcam);
+}
+
+int mlxsw_sp_acl_region_rehash_intrvl_set(struct mlxsw_sp *mlxsw_sp, u32 val)
+{
+	struct mlxsw_sp_acl *acl = mlxsw_sp->acl;
+
+	return mlxsw_sp_acl_tcam_vregion_rehash_intrvl_set(mlxsw_sp,
+							   &acl->tcam, val);
+}
+
+struct mlxsw_sp_acl_rulei_ops mlxsw_sp1_acl_rulei_ops = {
+	.act_mangle_field = mlxsw_sp1_acl_rulei_act_mangle_field,
+};
+
+struct mlxsw_sp_acl_rulei_ops mlxsw_sp2_acl_rulei_ops = {
+	.act_mangle_field = mlxsw_sp2_acl_rulei_act_mangle_field,
+};
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_acl_atcam.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_acl_atcam.c
new file mode 100644
index 000000000..4b713832f
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_acl_atcam.c
@@ -0,0 +1,646 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2018 Mellanox Technologies. All rights reserved */
+
+#include <linux/kernel.h>
+#include <linux/err.h>
+#include <linux/errno.h>
+#include <linux/gfp.h>
+#include <linux/refcount.h>
+#include <linux/rhashtable.h>
+#define CREATE_TRACE_POINTS
+#include <trace/events/mlxsw.h>
+
+#include "reg.h"
+#include "core.h"
+#include "spectrum.h"
+#include "spectrum_acl_tcam.h"
+#include "core_acl_flex_keys.h"
+
+#define MLXSW_SP_ACL_ATCAM_LKEY_ID_BLOCK_CLEAR_START	0
+#define MLXSW_SP_ACL_ATCAM_LKEY_ID_BLOCK_CLEAR_END	5
+
+struct mlxsw_sp_acl_atcam_lkey_id_ht_key {
+	char enc_key[MLXSW_REG_PTCEX_FLEX_KEY_BLOCKS_LEN]; /* MSB blocks */
+	u8 erp_id;
+};
+
+struct mlxsw_sp_acl_atcam_lkey_id {
+	struct rhash_head ht_node;
+	struct mlxsw_sp_acl_atcam_lkey_id_ht_key ht_key;
+	refcount_t refcnt;
+	u32 id;
+};
+
+struct mlxsw_sp_acl_atcam_region_ops {
+	int (*init)(struct mlxsw_sp_acl_atcam_region *aregion);
+	void (*fini)(struct mlxsw_sp_acl_atcam_region *aregion);
+	struct mlxsw_sp_acl_atcam_lkey_id *
+		(*lkey_id_get)(struct mlxsw_sp_acl_atcam_region *aregion,
+			       char *enc_key, u8 erp_id);
+	void (*lkey_id_put)(struct mlxsw_sp_acl_atcam_region *aregion,
+			    struct mlxsw_sp_acl_atcam_lkey_id *lkey_id);
+};
+
+struct mlxsw_sp_acl_atcam_region_generic {
+	struct mlxsw_sp_acl_atcam_lkey_id dummy_lkey_id;
+};
+
+struct mlxsw_sp_acl_atcam_region_12kb {
+	struct rhashtable lkey_ht;
+	unsigned int max_lkey_id;
+	unsigned long *used_lkey_id;
+};
+
+static const struct rhashtable_params mlxsw_sp_acl_atcam_lkey_id_ht_params = {
+	.key_len = sizeof(struct mlxsw_sp_acl_atcam_lkey_id_ht_key),
+	.key_offset = offsetof(struct mlxsw_sp_acl_atcam_lkey_id, ht_key),
+	.head_offset = offsetof(struct mlxsw_sp_acl_atcam_lkey_id, ht_node),
+};
+
+static const struct rhashtable_params mlxsw_sp_acl_atcam_entries_ht_params = {
+	.key_len = sizeof(struct mlxsw_sp_acl_atcam_entry_ht_key),
+	.key_offset = offsetof(struct mlxsw_sp_acl_atcam_entry, ht_key),
+	.head_offset = offsetof(struct mlxsw_sp_acl_atcam_entry, ht_node),
+};
+
+static bool
+mlxsw_sp_acl_atcam_is_centry(const struct mlxsw_sp_acl_atcam_entry *aentry)
+{
+	return mlxsw_sp_acl_erp_mask_is_ctcam(aentry->erp_mask);
+}
+
+static int
+mlxsw_sp_acl_atcam_region_generic_init(struct mlxsw_sp_acl_atcam_region *aregion)
+{
+	struct mlxsw_sp_acl_atcam_region_generic *region_generic;
+
+	region_generic = kzalloc(sizeof(*region_generic), GFP_KERNEL);
+	if (!region_generic)
+		return -ENOMEM;
+
+	refcount_set(&region_generic->dummy_lkey_id.refcnt, 1);
+	aregion->priv = region_generic;
+
+	return 0;
+}
+
+static void
+mlxsw_sp_acl_atcam_region_generic_fini(struct mlxsw_sp_acl_atcam_region *aregion)
+{
+	kfree(aregion->priv);
+}
+
+static struct mlxsw_sp_acl_atcam_lkey_id *
+mlxsw_sp_acl_atcam_generic_lkey_id_get(struct mlxsw_sp_acl_atcam_region *aregion,
+				       char *enc_key, u8 erp_id)
+{
+	struct mlxsw_sp_acl_atcam_region_generic *region_generic;
+
+	region_generic = aregion->priv;
+	return &region_generic->dummy_lkey_id;
+}
+
+static void
+mlxsw_sp_acl_atcam_generic_lkey_id_put(struct mlxsw_sp_acl_atcam_region *aregion,
+				       struct mlxsw_sp_acl_atcam_lkey_id *lkey_id)
+{
+}
+
+static const struct mlxsw_sp_acl_atcam_region_ops
+mlxsw_sp_acl_atcam_region_generic_ops = {
+	.init		= mlxsw_sp_acl_atcam_region_generic_init,
+	.fini		= mlxsw_sp_acl_atcam_region_generic_fini,
+	.lkey_id_get	= mlxsw_sp_acl_atcam_generic_lkey_id_get,
+	.lkey_id_put	= mlxsw_sp_acl_atcam_generic_lkey_id_put,
+};
+
+static int
+mlxsw_sp_acl_atcam_region_12kb_init(struct mlxsw_sp_acl_atcam_region *aregion)
+{
+	struct mlxsw_sp *mlxsw_sp = aregion->region->mlxsw_sp;
+	struct mlxsw_sp_acl_atcam_region_12kb *region_12kb;
+	u64 max_lkey_id;
+	int err;
+
+	if (!MLXSW_CORE_RES_VALID(mlxsw_sp->core, ACL_MAX_LARGE_KEY_ID))
+		return -EIO;
+
+	max_lkey_id = MLXSW_CORE_RES_GET(mlxsw_sp->core, ACL_MAX_LARGE_KEY_ID);
+	region_12kb = kzalloc(sizeof(*region_12kb), GFP_KERNEL);
+	if (!region_12kb)
+		return -ENOMEM;
+
+	region_12kb->used_lkey_id = bitmap_zalloc(max_lkey_id, GFP_KERNEL);
+	if (!region_12kb->used_lkey_id) {
+		err = -ENOMEM;
+		goto err_used_lkey_id_alloc;
+	}
+
+	err = rhashtable_init(&region_12kb->lkey_ht,
+			      &mlxsw_sp_acl_atcam_lkey_id_ht_params);
+	if (err)
+		goto err_rhashtable_init;
+
+	region_12kb->max_lkey_id = max_lkey_id;
+	aregion->priv = region_12kb;
+
+	return 0;
+
+err_rhashtable_init:
+	bitmap_free(region_12kb->used_lkey_id);
+err_used_lkey_id_alloc:
+	kfree(region_12kb);
+	return err;
+}
+
+static void
+mlxsw_sp_acl_atcam_region_12kb_fini(struct mlxsw_sp_acl_atcam_region *aregion)
+{
+	struct mlxsw_sp_acl_atcam_region_12kb *region_12kb = aregion->priv;
+
+	rhashtable_destroy(&region_12kb->lkey_ht);
+	bitmap_free(region_12kb->used_lkey_id);
+	kfree(region_12kb);
+}
+
+static struct mlxsw_sp_acl_atcam_lkey_id *
+mlxsw_sp_acl_atcam_lkey_id_create(struct mlxsw_sp_acl_atcam_region *aregion,
+				  struct mlxsw_sp_acl_atcam_lkey_id_ht_key *ht_key)
+{
+	struct mlxsw_sp_acl_atcam_region_12kb *region_12kb = aregion->priv;
+	struct mlxsw_sp_acl_atcam_lkey_id *lkey_id;
+	u32 id;
+	int err;
+
+	id = find_first_zero_bit(region_12kb->used_lkey_id,
+				 region_12kb->max_lkey_id);
+	if (id < region_12kb->max_lkey_id)
+		__set_bit(id, region_12kb->used_lkey_id);
+	else
+		return ERR_PTR(-ENOBUFS);
+
+	lkey_id = kzalloc(sizeof(*lkey_id), GFP_KERNEL);
+	if (!lkey_id) {
+		err = -ENOMEM;
+		goto err_lkey_id_alloc;
+	}
+
+	lkey_id->id = id;
+	memcpy(&lkey_id->ht_key, ht_key, sizeof(*ht_key));
+	refcount_set(&lkey_id->refcnt, 1);
+
+	err = rhashtable_insert_fast(&region_12kb->lkey_ht,
+				     &lkey_id->ht_node,
+				     mlxsw_sp_acl_atcam_lkey_id_ht_params);
+	if (err)
+		goto err_rhashtable_insert;
+
+	return lkey_id;
+
+err_rhashtable_insert:
+	kfree(lkey_id);
+err_lkey_id_alloc:
+	__clear_bit(id, region_12kb->used_lkey_id);
+	return ERR_PTR(err);
+}
+
+static void
+mlxsw_sp_acl_atcam_lkey_id_destroy(struct mlxsw_sp_acl_atcam_region *aregion,
+				   struct mlxsw_sp_acl_atcam_lkey_id *lkey_id)
+{
+	struct mlxsw_sp_acl_atcam_region_12kb *region_12kb = aregion->priv;
+	u32 id = lkey_id->id;
+
+	rhashtable_remove_fast(&region_12kb->lkey_ht, &lkey_id->ht_node,
+			       mlxsw_sp_acl_atcam_lkey_id_ht_params);
+	kfree(lkey_id);
+	__clear_bit(id, region_12kb->used_lkey_id);
+}
+
+static struct mlxsw_sp_acl_atcam_lkey_id *
+mlxsw_sp_acl_atcam_12kb_lkey_id_get(struct mlxsw_sp_acl_atcam_region *aregion,
+				    char *enc_key, u8 erp_id)
+{
+	struct mlxsw_sp_acl_atcam_region_12kb *region_12kb = aregion->priv;
+	struct mlxsw_sp_acl_tcam_region *region = aregion->region;
+	struct mlxsw_sp_acl_atcam_lkey_id_ht_key ht_key = {{ 0 } };
+	struct mlxsw_sp *mlxsw_sp = region->mlxsw_sp;
+	struct mlxsw_afk *afk = mlxsw_sp_acl_afk(mlxsw_sp->acl);
+	struct mlxsw_sp_acl_atcam_lkey_id *lkey_id;
+
+	memcpy(ht_key.enc_key, enc_key, sizeof(ht_key.enc_key));
+	mlxsw_afk_clear(afk, ht_key.enc_key,
+			MLXSW_SP_ACL_ATCAM_LKEY_ID_BLOCK_CLEAR_START,
+			MLXSW_SP_ACL_ATCAM_LKEY_ID_BLOCK_CLEAR_END);
+	ht_key.erp_id = erp_id;
+	lkey_id = rhashtable_lookup_fast(&region_12kb->lkey_ht, &ht_key,
+					 mlxsw_sp_acl_atcam_lkey_id_ht_params);
+	if (lkey_id) {
+		refcount_inc(&lkey_id->refcnt);
+		return lkey_id;
+	}
+
+	return mlxsw_sp_acl_atcam_lkey_id_create(aregion, &ht_key);
+}
+
+static void
+mlxsw_sp_acl_atcam_12kb_lkey_id_put(struct mlxsw_sp_acl_atcam_region *aregion,
+				    struct mlxsw_sp_acl_atcam_lkey_id *lkey_id)
+{
+	if (refcount_dec_and_test(&lkey_id->refcnt))
+		mlxsw_sp_acl_atcam_lkey_id_destroy(aregion, lkey_id);
+}
+
+static const struct mlxsw_sp_acl_atcam_region_ops
+mlxsw_sp_acl_atcam_region_12kb_ops = {
+	.init		= mlxsw_sp_acl_atcam_region_12kb_init,
+	.fini		= mlxsw_sp_acl_atcam_region_12kb_fini,
+	.lkey_id_get	= mlxsw_sp_acl_atcam_12kb_lkey_id_get,
+	.lkey_id_put	= mlxsw_sp_acl_atcam_12kb_lkey_id_put,
+};
+
+static const struct mlxsw_sp_acl_atcam_region_ops *
+mlxsw_sp_acl_atcam_region_ops_arr[] = {
+	[MLXSW_SP_ACL_ATCAM_REGION_TYPE_2KB]	=
+		&mlxsw_sp_acl_atcam_region_generic_ops,
+	[MLXSW_SP_ACL_ATCAM_REGION_TYPE_4KB]	=
+		&mlxsw_sp_acl_atcam_region_generic_ops,
+	[MLXSW_SP_ACL_ATCAM_REGION_TYPE_8KB]	=
+		&mlxsw_sp_acl_atcam_region_generic_ops,
+	[MLXSW_SP_ACL_ATCAM_REGION_TYPE_12KB]	=
+		&mlxsw_sp_acl_atcam_region_12kb_ops,
+};
+
+int mlxsw_sp_acl_atcam_region_associate(struct mlxsw_sp *mlxsw_sp,
+					u16 region_id)
+{
+	char perar_pl[MLXSW_REG_PERAR_LEN];
+	/* For now, just assume that every region has 12 key blocks */
+	u16 hw_region = region_id * 3;
+	u64 max_regions;
+
+	max_regions = MLXSW_CORE_RES_GET(mlxsw_sp->core, ACL_MAX_REGIONS);
+	if (hw_region >= max_regions)
+		return -ENOBUFS;
+
+	mlxsw_reg_perar_pack(perar_pl, region_id, hw_region);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(perar), perar_pl);
+}
+
+static void
+mlxsw_sp_acl_atcam_region_type_init(struct mlxsw_sp_acl_atcam_region *aregion)
+{
+	struct mlxsw_sp_acl_tcam_region *region = aregion->region;
+	enum mlxsw_sp_acl_atcam_region_type region_type;
+	unsigned int blocks_count;
+
+	/* We already know the blocks count can not exceed the maximum
+	 * blocks count.
+	 */
+	blocks_count = mlxsw_afk_key_info_blocks_count_get(region->key_info);
+	if (blocks_count <= 2)
+		region_type = MLXSW_SP_ACL_ATCAM_REGION_TYPE_2KB;
+	else if (blocks_count <= 4)
+		region_type = MLXSW_SP_ACL_ATCAM_REGION_TYPE_4KB;
+	else if (blocks_count <= 8)
+		region_type = MLXSW_SP_ACL_ATCAM_REGION_TYPE_8KB;
+	else
+		region_type = MLXSW_SP_ACL_ATCAM_REGION_TYPE_12KB;
+
+	aregion->type = region_type;
+	aregion->ops = mlxsw_sp_acl_atcam_region_ops_arr[region_type];
+}
+
+int
+mlxsw_sp_acl_atcam_region_init(struct mlxsw_sp *mlxsw_sp,
+			       struct mlxsw_sp_acl_atcam *atcam,
+			       struct mlxsw_sp_acl_atcam_region *aregion,
+			       struct mlxsw_sp_acl_tcam_region *region,
+			       void *hints_priv,
+			       const struct mlxsw_sp_acl_ctcam_region_ops *ops)
+{
+	int err;
+
+	aregion->region = region;
+	aregion->atcam = atcam;
+	mlxsw_sp_acl_atcam_region_type_init(aregion);
+	INIT_LIST_HEAD(&aregion->entries_list);
+
+	err = rhashtable_init(&aregion->entries_ht,
+			      &mlxsw_sp_acl_atcam_entries_ht_params);
+	if (err)
+		return err;
+	err = aregion->ops->init(aregion);
+	if (err)
+		goto err_ops_init;
+	err = mlxsw_sp_acl_erp_region_init(aregion, hints_priv);
+	if (err)
+		goto err_erp_region_init;
+	err = mlxsw_sp_acl_ctcam_region_init(mlxsw_sp, &aregion->cregion,
+					     region, ops);
+	if (err)
+		goto err_ctcam_region_init;
+
+	return 0;
+
+err_ctcam_region_init:
+	mlxsw_sp_acl_erp_region_fini(aregion);
+err_erp_region_init:
+	aregion->ops->fini(aregion);
+err_ops_init:
+	rhashtable_destroy(&aregion->entries_ht);
+	return err;
+}
+
+void mlxsw_sp_acl_atcam_region_fini(struct mlxsw_sp_acl_atcam_region *aregion)
+{
+	mlxsw_sp_acl_ctcam_region_fini(&aregion->cregion);
+	mlxsw_sp_acl_erp_region_fini(aregion);
+	aregion->ops->fini(aregion);
+	rhashtable_destroy(&aregion->entries_ht);
+	WARN_ON(!list_empty(&aregion->entries_list));
+}
+
+void mlxsw_sp_acl_atcam_chunk_init(struct mlxsw_sp_acl_atcam_region *aregion,
+				   struct mlxsw_sp_acl_atcam_chunk *achunk,
+				   unsigned int priority)
+{
+	mlxsw_sp_acl_ctcam_chunk_init(&aregion->cregion, &achunk->cchunk,
+				      priority);
+}
+
+void mlxsw_sp_acl_atcam_chunk_fini(struct mlxsw_sp_acl_atcam_chunk *achunk)
+{
+	mlxsw_sp_acl_ctcam_chunk_fini(&achunk->cchunk);
+}
+
+static int
+mlxsw_sp_acl_atcam_region_entry_insert(struct mlxsw_sp *mlxsw_sp,
+				       struct mlxsw_sp_acl_atcam_region *aregion,
+				       struct mlxsw_sp_acl_atcam_entry *aentry,
+				       struct mlxsw_sp_acl_rule_info *rulei)
+{
+	struct mlxsw_sp_acl_tcam_region *region = aregion->region;
+	u8 erp_id = mlxsw_sp_acl_erp_mask_erp_id(aentry->erp_mask);
+	struct mlxsw_sp_acl_atcam_lkey_id *lkey_id;
+	char ptce3_pl[MLXSW_REG_PTCE3_LEN];
+	u32 kvdl_index, priority;
+	int err;
+
+	err = mlxsw_sp_acl_tcam_priority_get(mlxsw_sp, rulei, &priority, true);
+	if (err)
+		return err;
+
+	lkey_id = aregion->ops->lkey_id_get(aregion, aentry->enc_key, erp_id);
+	if (IS_ERR(lkey_id))
+		return PTR_ERR(lkey_id);
+	aentry->lkey_id = lkey_id;
+
+	kvdl_index = mlxsw_afa_block_first_kvdl_index(rulei->act_block);
+	mlxsw_reg_ptce3_pack(ptce3_pl, true, MLXSW_REG_PTCE3_OP_WRITE_WRITE,
+			     priority, region->tcam_region_info,
+			     aentry->enc_key, erp_id,
+			     aentry->delta_info.start,
+			     aentry->delta_info.mask,
+			     aentry->delta_info.value,
+			     refcount_read(&lkey_id->refcnt) != 1, lkey_id->id,
+			     kvdl_index);
+	err = mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(ptce3), ptce3_pl);
+	if (err)
+		goto err_ptce3_write;
+
+	return 0;
+
+err_ptce3_write:
+	aregion->ops->lkey_id_put(aregion, lkey_id);
+	return err;
+}
+
+static void
+mlxsw_sp_acl_atcam_region_entry_remove(struct mlxsw_sp *mlxsw_sp,
+				       struct mlxsw_sp_acl_atcam_region *aregion,
+				       struct mlxsw_sp_acl_atcam_entry *aentry)
+{
+	struct mlxsw_sp_acl_atcam_lkey_id *lkey_id = aentry->lkey_id;
+	struct mlxsw_sp_acl_tcam_region *region = aregion->region;
+	u8 erp_id = mlxsw_sp_acl_erp_mask_erp_id(aentry->erp_mask);
+	char ptce3_pl[MLXSW_REG_PTCE3_LEN];
+
+	mlxsw_reg_ptce3_pack(ptce3_pl, false, MLXSW_REG_PTCE3_OP_WRITE_WRITE, 0,
+			     region->tcam_region_info,
+			     aentry->enc_key, erp_id,
+			     aentry->delta_info.start,
+			     aentry->delta_info.mask,
+			     aentry->delta_info.value,
+			     refcount_read(&lkey_id->refcnt) != 1,
+			     lkey_id->id, 0);
+	mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(ptce3), ptce3_pl);
+	aregion->ops->lkey_id_put(aregion, lkey_id);
+}
+
+static int
+mlxsw_sp_acl_atcam_region_entry_action_replace(struct mlxsw_sp *mlxsw_sp,
+					       struct mlxsw_sp_acl_atcam_region *aregion,
+					       struct mlxsw_sp_acl_atcam_entry *aentry,
+					       struct mlxsw_sp_acl_rule_info *rulei)
+{
+	struct mlxsw_sp_acl_atcam_lkey_id *lkey_id = aentry->lkey_id;
+	u8 erp_id = mlxsw_sp_acl_erp_mask_erp_id(aentry->erp_mask);
+	struct mlxsw_sp_acl_tcam_region *region = aregion->region;
+	char ptce3_pl[MLXSW_REG_PTCE3_LEN];
+	u32 kvdl_index, priority;
+	int err;
+
+	err = mlxsw_sp_acl_tcam_priority_get(mlxsw_sp, rulei, &priority, true);
+	if (err)
+		return err;
+	kvdl_index = mlxsw_afa_block_first_kvdl_index(rulei->act_block);
+	mlxsw_reg_ptce3_pack(ptce3_pl, true, MLXSW_REG_PTCE3_OP_WRITE_UPDATE,
+			     priority, region->tcam_region_info,
+			     aentry->enc_key, erp_id,
+			     aentry->delta_info.start,
+			     aentry->delta_info.mask,
+			     aentry->delta_info.value,
+			     refcount_read(&lkey_id->refcnt) != 1, lkey_id->id,
+			     kvdl_index);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(ptce3), ptce3_pl);
+}
+
+static int
+__mlxsw_sp_acl_atcam_entry_add(struct mlxsw_sp *mlxsw_sp,
+			       struct mlxsw_sp_acl_atcam_region *aregion,
+			       struct mlxsw_sp_acl_atcam_entry *aentry,
+			       struct mlxsw_sp_acl_rule_info *rulei)
+{
+	struct mlxsw_sp_acl_tcam_region *region = aregion->region;
+	char mask[MLXSW_REG_PTCEX_FLEX_KEY_BLOCKS_LEN] = { 0 };
+	struct mlxsw_afk *afk = mlxsw_sp_acl_afk(mlxsw_sp->acl);
+	const struct mlxsw_sp_acl_erp_delta *delta;
+	struct mlxsw_sp_acl_erp_mask *erp_mask;
+	int err;
+
+	mlxsw_afk_encode(afk, region->key_info, &rulei->values,
+			 aentry->ht_key.full_enc_key, mask);
+
+	erp_mask = mlxsw_sp_acl_erp_mask_get(aregion, mask, false);
+	if (IS_ERR(erp_mask))
+		return PTR_ERR(erp_mask);
+	aentry->erp_mask = erp_mask;
+	aentry->ht_key.erp_id = mlxsw_sp_acl_erp_mask_erp_id(erp_mask);
+	memcpy(aentry->enc_key, aentry->ht_key.full_enc_key,
+	       sizeof(aentry->enc_key));
+
+	/* Compute all needed delta information and clear the delta bits
+	 * from the encrypted key.
+	 */
+	delta = mlxsw_sp_acl_erp_delta(aentry->erp_mask);
+	aentry->delta_info.start = mlxsw_sp_acl_erp_delta_start(delta);
+	aentry->delta_info.mask = mlxsw_sp_acl_erp_delta_mask(delta);
+	aentry->delta_info.value =
+		mlxsw_sp_acl_erp_delta_value(delta,
+					     aentry->ht_key.full_enc_key);
+	mlxsw_sp_acl_erp_delta_clear(delta, aentry->enc_key);
+
+	/* Add rule to the list of A-TCAM rules, assuming this
+	 * rule is intended to A-TCAM. In case this rule does
+	 * not fit into A-TCAM it will be removed from the list.
+	 */
+	list_add(&aentry->list, &aregion->entries_list);
+
+	/* We can't insert identical rules into the A-TCAM, so fail and
+	 * let the rule spill into C-TCAM
+	 */
+	err = rhashtable_lookup_insert_fast(&aregion->entries_ht,
+					    &aentry->ht_node,
+					    mlxsw_sp_acl_atcam_entries_ht_params);
+	if (err)
+		goto err_rhashtable_insert;
+
+	/* Bloom filter must be updated here, before inserting the rule into
+	 * the A-TCAM.
+	 */
+	err = mlxsw_sp_acl_erp_bf_insert(mlxsw_sp, aregion, erp_mask, aentry);
+	if (err)
+		goto err_bf_insert;
+
+	err = mlxsw_sp_acl_atcam_region_entry_insert(mlxsw_sp, aregion, aentry,
+						     rulei);
+	if (err)
+		goto err_rule_insert;
+
+	return 0;
+
+err_rule_insert:
+	mlxsw_sp_acl_erp_bf_remove(mlxsw_sp, aregion, erp_mask, aentry);
+err_bf_insert:
+	rhashtable_remove_fast(&aregion->entries_ht, &aentry->ht_node,
+			       mlxsw_sp_acl_atcam_entries_ht_params);
+err_rhashtable_insert:
+	list_del(&aentry->list);
+	mlxsw_sp_acl_erp_mask_put(aregion, erp_mask);
+	return err;
+}
+
+static void
+__mlxsw_sp_acl_atcam_entry_del(struct mlxsw_sp *mlxsw_sp,
+			       struct mlxsw_sp_acl_atcam_region *aregion,
+			       struct mlxsw_sp_acl_atcam_entry *aentry)
+{
+	mlxsw_sp_acl_atcam_region_entry_remove(mlxsw_sp, aregion, aentry);
+	mlxsw_sp_acl_erp_bf_remove(mlxsw_sp, aregion, aentry->erp_mask, aentry);
+	rhashtable_remove_fast(&aregion->entries_ht, &aentry->ht_node,
+			       mlxsw_sp_acl_atcam_entries_ht_params);
+	list_del(&aentry->list);
+	mlxsw_sp_acl_erp_mask_put(aregion, aentry->erp_mask);
+}
+
+static int
+__mlxsw_sp_acl_atcam_entry_action_replace(struct mlxsw_sp *mlxsw_sp,
+					  struct mlxsw_sp_acl_atcam_region *aregion,
+					  struct mlxsw_sp_acl_atcam_entry *aentry,
+					  struct mlxsw_sp_acl_rule_info *rulei)
+{
+	return mlxsw_sp_acl_atcam_region_entry_action_replace(mlxsw_sp, aregion,
+							      aentry, rulei);
+}
+
+int mlxsw_sp_acl_atcam_entry_add(struct mlxsw_sp *mlxsw_sp,
+				 struct mlxsw_sp_acl_atcam_region *aregion,
+				 struct mlxsw_sp_acl_atcam_chunk *achunk,
+				 struct mlxsw_sp_acl_atcam_entry *aentry,
+				 struct mlxsw_sp_acl_rule_info *rulei)
+{
+	int err;
+
+	err = __mlxsw_sp_acl_atcam_entry_add(mlxsw_sp, aregion, aentry, rulei);
+	if (!err)
+		return 0;
+
+	/* It is possible we failed to add the rule to the A-TCAM due to
+	 * exceeded number of masks. Try to spill into C-TCAM.
+	 */
+	trace_mlxsw_sp_acl_atcam_entry_add_ctcam_spill(mlxsw_sp, aregion);
+	err = mlxsw_sp_acl_ctcam_entry_add(mlxsw_sp, &aregion->cregion,
+					   &achunk->cchunk, &aentry->centry,
+					   rulei, true);
+	if (!err)
+		return 0;
+
+	return err;
+}
+
+void mlxsw_sp_acl_atcam_entry_del(struct mlxsw_sp *mlxsw_sp,
+				  struct mlxsw_sp_acl_atcam_region *aregion,
+				  struct mlxsw_sp_acl_atcam_chunk *achunk,
+				  struct mlxsw_sp_acl_atcam_entry *aentry)
+{
+	if (mlxsw_sp_acl_atcam_is_centry(aentry))
+		mlxsw_sp_acl_ctcam_entry_del(mlxsw_sp, &aregion->cregion,
+					     &achunk->cchunk, &aentry->centry);
+	else
+		__mlxsw_sp_acl_atcam_entry_del(mlxsw_sp, aregion, aentry);
+}
+
+int
+mlxsw_sp_acl_atcam_entry_action_replace(struct mlxsw_sp *mlxsw_sp,
+					struct mlxsw_sp_acl_atcam_region *aregion,
+					struct mlxsw_sp_acl_atcam_entry *aentry,
+					struct mlxsw_sp_acl_rule_info *rulei)
+{
+	int err;
+
+	if (mlxsw_sp_acl_atcam_is_centry(aentry))
+		err = mlxsw_sp_acl_ctcam_entry_action_replace(mlxsw_sp,
+							      &aregion->cregion,
+							      &aentry->centry,
+							      rulei);
+	else
+		err = __mlxsw_sp_acl_atcam_entry_action_replace(mlxsw_sp,
+								aregion, aentry,
+								rulei);
+
+	return err;
+}
+
+int mlxsw_sp_acl_atcam_init(struct mlxsw_sp *mlxsw_sp,
+			    struct mlxsw_sp_acl_atcam *atcam)
+{
+	return mlxsw_sp_acl_erps_init(mlxsw_sp, atcam);
+}
+
+void mlxsw_sp_acl_atcam_fini(struct mlxsw_sp *mlxsw_sp,
+			     struct mlxsw_sp_acl_atcam *atcam)
+{
+	mlxsw_sp_acl_erps_fini(mlxsw_sp, atcam);
+}
+
+void *
+mlxsw_sp_acl_atcam_rehash_hints_get(struct mlxsw_sp_acl_atcam_region *aregion)
+{
+	return mlxsw_sp_acl_erp_rehash_hints_get(aregion);
+}
+
+void mlxsw_sp_acl_atcam_rehash_hints_put(void *hints_priv)
+{
+	mlxsw_sp_acl_erp_rehash_hints_put(hints_priv);
+}
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_acl_bloom_filter.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_acl_bloom_filter.c
new file mode 100644
index 000000000..95f63fcf4
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_acl_bloom_filter.c
@@ -0,0 +1,522 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2018 Mellanox Technologies. All rights reserved */
+
+#include <linux/errno.h>
+#include <linux/gfp.h>
+#include <linux/kernel.h>
+#include <linux/refcount.h>
+#include <linux/mutex.h>
+
+#include "spectrum.h"
+#include "spectrum_acl_tcam.h"
+
+struct mlxsw_sp_acl_bf {
+	struct mutex lock; /* Protects Bloom Filter updates. */
+	unsigned int bank_size;
+	refcount_t refcnt[];
+};
+
+/* Bloom filter uses a crc-16 hash over chunks of data which contain 4 key
+ * blocks, eRP ID and region ID. In Spectrum-2 and above, region key is combined
+ * of up to 12 key blocks, so there can be up to 3 chunks in the Bloom filter
+ * key, depending on the actual number of key blocks used in the region.
+ * The layout of the Bloom filter key is as follows:
+ *
+ * +-------------------------+------------------------+------------------------+
+ * | Chunk 2 Key blocks 11-8 | Chunk 1 Key blocks 7-4 | Chunk 0 Key blocks 3-0 |
+ * +-------------------------+------------------------+------------------------+
+ */
+#define MLXSW_BLOOM_KEY_CHUNKS 3
+
+/* Spectrum-2 and Spectrum-3 chunks */
+#define MLXSW_SP2_BLOOM_KEY_LEN 69
+
+/* Each chunk size is 23 bytes. 18 bytes of it contain 4 key blocks, each is
+ * 36 bits, 2 bytes which hold eRP ID and region ID, and 3 bytes of zero
+ * padding.
+ * The layout of each chunk is as follows:
+ *
+ * +---------+----------------------+-----------------------------------+
+ * | 3 bytes |        2 bytes       |              18 bytes             |
+ * +---------+-----------+----------+-----------------------------------+
+ * | 183:158 |  157:148  | 147:144  |               143:0               |
+ * +---------+-----------+----------+-----------------------------------+
+ * |    0    | region ID |  eRP ID  |      4 Key blocks (18 Bytes)      |
+ * +---------+-----------+----------+-----------------------------------+
+ */
+#define MLXSW_SP2_BLOOM_CHUNK_PAD_BYTES 3
+#define MLXSW_SP2_BLOOM_CHUNK_KEY_BYTES 18
+#define MLXSW_SP2_BLOOM_KEY_CHUNK_BYTES 23
+
+/* The offset of the key block within a chunk is 5 bytes as it comes after
+ * 3 bytes of zero padding and 16 bits of region ID and eRP ID.
+ */
+#define MLXSW_SP2_BLOOM_CHUNK_KEY_OFFSET 5
+
+/* This table is just the CRC of each possible byte which is used for
+ * Spectrum-{2-3}. It is computed, Msbit first, for the Bloom filter
+ * polynomial which is 0x8529 (1 + x^3 + x^5 + x^8 + x^10 + x^15 and
+ * the implicit x^16).
+ */
+static const u16 mlxsw_sp2_acl_bf_crc16_tab[256] = {
+0x0000, 0x8529, 0x8f7b, 0x0a52, 0x9bdf, 0x1ef6, 0x14a4, 0x918d,
+0xb297, 0x37be, 0x3dec, 0xb8c5, 0x2948, 0xac61, 0xa633, 0x231a,
+0xe007, 0x652e, 0x6f7c, 0xea55, 0x7bd8, 0xfef1, 0xf4a3, 0x718a,
+0x5290, 0xd7b9, 0xddeb, 0x58c2, 0xc94f, 0x4c66, 0x4634, 0xc31d,
+0x4527, 0xc00e, 0xca5c, 0x4f75, 0xdef8, 0x5bd1, 0x5183, 0xd4aa,
+0xf7b0, 0x7299, 0x78cb, 0xfde2, 0x6c6f, 0xe946, 0xe314, 0x663d,
+0xa520, 0x2009, 0x2a5b, 0xaf72, 0x3eff, 0xbbd6, 0xb184, 0x34ad,
+0x17b7, 0x929e, 0x98cc, 0x1de5, 0x8c68, 0x0941, 0x0313, 0x863a,
+0x8a4e, 0x0f67, 0x0535, 0x801c, 0x1191, 0x94b8, 0x9eea, 0x1bc3,
+0x38d9, 0xbdf0, 0xb7a2, 0x328b, 0xa306, 0x262f, 0x2c7d, 0xa954,
+0x6a49, 0xef60, 0xe532, 0x601b, 0xf196, 0x74bf, 0x7eed, 0xfbc4,
+0xd8de, 0x5df7, 0x57a5, 0xd28c, 0x4301, 0xc628, 0xcc7a, 0x4953,
+0xcf69, 0x4a40, 0x4012, 0xc53b, 0x54b6, 0xd19f, 0xdbcd, 0x5ee4,
+0x7dfe, 0xf8d7, 0xf285, 0x77ac, 0xe621, 0x6308, 0x695a, 0xec73,
+0x2f6e, 0xaa47, 0xa015, 0x253c, 0xb4b1, 0x3198, 0x3bca, 0xbee3,
+0x9df9, 0x18d0, 0x1282, 0x97ab, 0x0626, 0x830f, 0x895d, 0x0c74,
+0x91b5, 0x149c, 0x1ece, 0x9be7, 0x0a6a, 0x8f43, 0x8511, 0x0038,
+0x2322, 0xa60b, 0xac59, 0x2970, 0xb8fd, 0x3dd4, 0x3786, 0xb2af,
+0x71b2, 0xf49b, 0xfec9, 0x7be0, 0xea6d, 0x6f44, 0x6516, 0xe03f,
+0xc325, 0x460c, 0x4c5e, 0xc977, 0x58fa, 0xddd3, 0xd781, 0x52a8,
+0xd492, 0x51bb, 0x5be9, 0xdec0, 0x4f4d, 0xca64, 0xc036, 0x451f,
+0x6605, 0xe32c, 0xe97e, 0x6c57, 0xfdda, 0x78f3, 0x72a1, 0xf788,
+0x3495, 0xb1bc, 0xbbee, 0x3ec7, 0xaf4a, 0x2a63, 0x2031, 0xa518,
+0x8602, 0x032b, 0x0979, 0x8c50, 0x1ddd, 0x98f4, 0x92a6, 0x178f,
+0x1bfb, 0x9ed2, 0x9480, 0x11a9, 0x8024, 0x050d, 0x0f5f, 0x8a76,
+0xa96c, 0x2c45, 0x2617, 0xa33e, 0x32b3, 0xb79a, 0xbdc8, 0x38e1,
+0xfbfc, 0x7ed5, 0x7487, 0xf1ae, 0x6023, 0xe50a, 0xef58, 0x6a71,
+0x496b, 0xcc42, 0xc610, 0x4339, 0xd2b4, 0x579d, 0x5dcf, 0xd8e6,
+0x5edc, 0xdbf5, 0xd1a7, 0x548e, 0xc503, 0x402a, 0x4a78, 0xcf51,
+0xec4b, 0x6962, 0x6330, 0xe619, 0x7794, 0xf2bd, 0xf8ef, 0x7dc6,
+0xbedb, 0x3bf2, 0x31a0, 0xb489, 0x2504, 0xa02d, 0xaa7f, 0x2f56,
+0x0c4c, 0x8965, 0x8337, 0x061e, 0x9793, 0x12ba, 0x18e8, 0x9dc1,
+};
+
+/* Spectrum-4 chunks */
+#define MLXSW_SP4_BLOOM_KEY_LEN 60
+
+/* In Spectrum-4, there is no padding. Each chunk size is 20 bytes.
+ * 18 bytes of it contain 4 key blocks, each is 36 bits, and 2 bytes which hold
+ * eRP ID and region ID.
+ * The layout of each chunk is as follows:
+ *
+ * +----------------------+-----------------------------------+
+ * |        2 bytes       |              18 bytes             |
+ * +-----------+----------+-----------------------------------+
+ * |  157:148  | 147:144  |               143:0               |
+ * +---------+-----------+----------+-------------------------+
+ * | region ID |  eRP ID  |      4 Key blocks (18 Bytes)      |
+ * +-----------+----------+-----------------------------------+
+ */
+
+#define MLXSW_SP4_BLOOM_CHUNK_PAD_BYTES 0
+#define MLXSW_SP4_BLOOM_CHUNK_KEY_BYTES 18
+#define MLXSW_SP4_BLOOM_KEY_CHUNK_BYTES 20
+
+/* The offset of the key block within a chunk is 2 bytes as it comes after
+ * 16 bits of region ID and eRP ID.
+ */
+#define MLXSW_SP4_BLOOM_CHUNK_KEY_OFFSET 2
+
+/* For Spectrum-4, two hash functions are used, CRC-10 and CRC-6 based.
+ * The result is combination of the two calculations -
+ * 6 bit column are MSB (result of CRC-6),
+ * 10 bit row are LSB (result of CRC-10).
+ */
+
+/* This table is just the CRC of each possible byte which is used for
+ * Spectrum-4. It is computed, Msbit first, for the Bloom filter
+ * polynomial which is 0x1b (1 + x^1 + x^3 + x^4 and the implicit x^10).
+ */
+static const u16 mlxsw_sp4_acl_bf_crc10_tab[256] = {
+0x0000, 0x001b, 0x0036, 0x002d, 0x006c, 0x0077, 0x005a, 0x0041,
+0x00d8, 0x00c3, 0x00ee, 0x00f5, 0x00b4, 0x00af, 0x0082, 0x0099,
+0x01b0, 0x01ab, 0x0186, 0x019d, 0x01dc, 0x01c7, 0x01ea, 0x01f1,
+0x0168, 0x0173, 0x015e, 0x0145, 0x0104, 0x011f, 0x0132, 0x0129,
+0x0360, 0x037b, 0x0356, 0x034d, 0x030c, 0x0317, 0x033a, 0x0321,
+0x03b8, 0x03a3, 0x038e, 0x0395, 0x03d4, 0x03cf, 0x03e2, 0x03f9,
+0x02d0, 0x02cb, 0x02e6, 0x02fd, 0x02bc, 0x02a7, 0x028a, 0x0291,
+0x0208, 0x0213, 0x023e, 0x0225, 0x0264, 0x027f, 0x0252, 0x0249,
+0x02db, 0x02c0, 0x02ed, 0x02f6, 0x02b7, 0x02ac, 0x0281, 0x029a,
+0x0203, 0x0218, 0x0235, 0x022e, 0x026f, 0x0274, 0x0259, 0x0242,
+0x036b, 0x0370, 0x035d, 0x0346, 0x0307, 0x031c, 0x0331, 0x032a,
+0x03b3, 0x03a8, 0x0385, 0x039e, 0x03df, 0x03c4, 0x03e9, 0x03f2,
+0x01bb, 0x01a0, 0x018d, 0x0196, 0x01d7, 0x01cc, 0x01e1, 0x01fa,
+0x0163, 0x0178, 0x0155, 0x014e, 0x010f, 0x0114, 0x0139, 0x0122,
+0x000b, 0x0010, 0x003d, 0x0026, 0x0067, 0x007c, 0x0051, 0x004a,
+0x00d3, 0x00c8, 0x00e5, 0x00fe, 0x00bf, 0x00a4, 0x0089, 0x0092,
+0x01ad, 0x01b6, 0x019b, 0x0180, 0x01c1, 0x01da, 0x01f7, 0x01ec,
+0x0175, 0x016e, 0x0143, 0x0158, 0x0119, 0x0102, 0x012f, 0x0134,
+0x001d, 0x0006, 0x002b, 0x0030, 0x0071, 0x006a, 0x0047, 0x005c,
+0x00c5, 0x00de, 0x00f3, 0x00e8, 0x00a9, 0x00b2, 0x009f, 0x0084,
+0x02cd, 0x02d6, 0x02fb, 0x02e0, 0x02a1, 0x02ba, 0x0297, 0x028c,
+0x0215, 0x020e, 0x0223, 0x0238, 0x0279, 0x0262, 0x024f, 0x0254,
+0x037d, 0x0366, 0x034b, 0x0350, 0x0311, 0x030a, 0x0327, 0x033c,
+0x03a5, 0x03be, 0x0393, 0x0388, 0x03c9, 0x03d2, 0x03ff, 0x03e4,
+0x0376, 0x036d, 0x0340, 0x035b, 0x031a, 0x0301, 0x032c, 0x0337,
+0x03ae, 0x03b5, 0x0398, 0x0383, 0x03c2, 0x03d9, 0x03f4, 0x03ef,
+0x02c6, 0x02dd, 0x02f0, 0x02eb, 0x02aa, 0x02b1, 0x029c, 0x0287,
+0x021e, 0x0205, 0x0228, 0x0233, 0x0272, 0x0269, 0x0244, 0x025f,
+0x0016, 0x000d, 0x0020, 0x003b, 0x007a, 0x0061, 0x004c, 0x0057,
+0x00ce, 0x00d5, 0x00f8, 0x00e3, 0x00a2, 0x00b9, 0x0094, 0x008f,
+0x01a6, 0x01bd, 0x0190, 0x018b, 0x01ca, 0x01d1, 0x01fc, 0x01e7,
+0x017e, 0x0165, 0x0148, 0x0153, 0x0112, 0x0109, 0x0124, 0x013f,
+};
+
+/* This table is just the CRC of each possible byte which is used for
+ * Spectrum-4. It is computed, Msbit first, for the Bloom filter
+ * polynomial which is 0x2d (1 + x^2+ x^3 + x^5 and the implicit x^6).
+ */
+static const u8 mlxsw_sp4_acl_bf_crc6_tab[256] = {
+0x00, 0x2d, 0x37, 0x1a, 0x03, 0x2e, 0x34, 0x19,
+0x06, 0x2b, 0x31, 0x1c, 0x05, 0x28, 0x32, 0x1f,
+0x0c, 0x21, 0x3b, 0x16, 0x0f, 0x22, 0x38, 0x15,
+0x0a, 0x27, 0x3d, 0x10, 0x09, 0x24, 0x3e, 0x13,
+0x18, 0x35, 0x2f, 0x02, 0x1b, 0x36, 0x2c, 0x01,
+0x1e, 0x33, 0x29, 0x04, 0x1d, 0x30, 0x2a, 0x07,
+0x14, 0x39, 0x23, 0x0e, 0x17, 0x3a, 0x20, 0x0d,
+0x12, 0x3f, 0x25, 0x08, 0x11, 0x3c, 0x26, 0x0b,
+0x30, 0x1d, 0x07, 0x2a, 0x33, 0x1e, 0x04, 0x29,
+0x36, 0x1b, 0x01, 0x2c, 0x35, 0x18, 0x02, 0x2f,
+0x3c, 0x11, 0x0b, 0x26, 0x3f, 0x12, 0x08, 0x25,
+0x3a, 0x17, 0x0d, 0x20, 0x39, 0x14, 0x0e, 0x23,
+0x28, 0x05, 0x1f, 0x32, 0x2b, 0x06, 0x1c, 0x31,
+0x2e, 0x03, 0x19, 0x34, 0x2d, 0x00, 0x1a, 0x37,
+0x24, 0x09, 0x13, 0x3e, 0x27, 0x0a, 0x10, 0x3d,
+0x22, 0x0f, 0x15, 0x38, 0x21, 0x0c, 0x16, 0x3b,
+0x0d, 0x20, 0x3a, 0x17, 0x0e, 0x23, 0x39, 0x14,
+0x0b, 0x26, 0x3c, 0x11, 0x08, 0x25, 0x3f, 0x12,
+0x01, 0x2c, 0x36, 0x1b, 0x02, 0x2f, 0x35, 0x18,
+0x07, 0x2a, 0x30, 0x1d, 0x04, 0x29, 0x33, 0x1e,
+0x15, 0x38, 0x22, 0x0f, 0x16, 0x3b, 0x21, 0x0c,
+0x13, 0x3e, 0x24, 0x09, 0x10, 0x3d, 0x27, 0x0a,
+0x19, 0x34, 0x2e, 0x03, 0x1a, 0x37, 0x2d, 0x00,
+0x1f, 0x32, 0x28, 0x05, 0x1c, 0x31, 0x2b, 0x06,
+0x3d, 0x10, 0x0a, 0x27, 0x3e, 0x13, 0x09, 0x24,
+0x3b, 0x16, 0x0c, 0x21, 0x38, 0x15, 0x0f, 0x22,
+0x31, 0x1c, 0x06, 0x2b, 0x32, 0x1f, 0x05, 0x28,
+0x37, 0x1a, 0x00, 0x2d, 0x34, 0x19, 0x03, 0x2e,
+0x25, 0x08, 0x12, 0x3f, 0x26, 0x0b, 0x11, 0x3c,
+0x23, 0x0e, 0x14, 0x39, 0x20, 0x0d, 0x17, 0x3a,
+0x29, 0x04, 0x1e, 0x33, 0x2a, 0x07, 0x1d, 0x30,
+0x2f, 0x02, 0x18, 0x35, 0x2c, 0x01, 0x1b, 0x36,
+};
+
+/* Each chunk contains 4 key blocks. Chunk 2 uses key blocks 11-8,
+ * and we need to populate it with 4 key blocks copied from the entry encoded
+ * key. The original keys layout is same for Spectrum-{2,3,4}.
+ * Since the encoded key contains a 2 bytes padding, key block 11 starts at
+ * offset 2. block 7 that is used in chunk 1 starts at offset 20 as 4 key blocks
+ * take 18 bytes. See 'MLXSW_SP2_AFK_BLOCK_LAYOUT' for more details.
+ * This array defines key offsets for easy access when copying key blocks from
+ * entry key to Bloom filter chunk.
+ */
+static const u8 chunk_key_offsets[MLXSW_BLOOM_KEY_CHUNKS] = {2, 20, 38};
+
+static u16 mlxsw_sp2_acl_bf_crc16_byte(u16 crc, u8 c)
+{
+	return (crc << 8) ^ mlxsw_sp2_acl_bf_crc16_tab[(crc >> 8) ^ c];
+}
+
+static u16 mlxsw_sp2_acl_bf_crc(const u8 *buffer, size_t len)
+{
+	u16 crc = 0;
+
+	while (len--)
+		crc = mlxsw_sp2_acl_bf_crc16_byte(crc, *buffer++);
+	return crc;
+}
+
+static void
+__mlxsw_sp_acl_bf_key_encode(struct mlxsw_sp_acl_atcam_region *aregion,
+			     struct mlxsw_sp_acl_atcam_entry *aentry,
+			     char *output, u8 *len, u8 max_chunks, u8 pad_bytes,
+			     u8 key_offset, u8 chunk_key_len, u8 chunk_len)
+{
+	struct mlxsw_afk_key_info *key_info = aregion->region->key_info;
+	u8 chunk_index, chunk_count, block_count;
+	char *chunk = output;
+	__be16 erp_region_id;
+
+	block_count = mlxsw_afk_key_info_blocks_count_get(key_info);
+	chunk_count = 1 + ((block_count - 1) >> 2);
+	erp_region_id = cpu_to_be16(aentry->ht_key.erp_id |
+				   (aregion->region->id << 4));
+	for (chunk_index = max_chunks - chunk_count; chunk_index < max_chunks;
+	     chunk_index++) {
+		memset(chunk, 0, pad_bytes);
+		memcpy(chunk + pad_bytes, &erp_region_id,
+		       sizeof(erp_region_id));
+		memcpy(chunk + key_offset,
+		       &aentry->enc_key[chunk_key_offsets[chunk_index]],
+		       chunk_key_len);
+		chunk += chunk_len;
+	}
+	*len = chunk_count * chunk_len;
+}
+
+static void
+mlxsw_sp2_acl_bf_key_encode(struct mlxsw_sp_acl_atcam_region *aregion,
+			    struct mlxsw_sp_acl_atcam_entry *aentry,
+			    char *output, u8 *len)
+{
+	__mlxsw_sp_acl_bf_key_encode(aregion, aentry, output, len,
+				     MLXSW_BLOOM_KEY_CHUNKS,
+				     MLXSW_SP2_BLOOM_CHUNK_PAD_BYTES,
+				     MLXSW_SP2_BLOOM_CHUNK_KEY_OFFSET,
+				     MLXSW_SP2_BLOOM_CHUNK_KEY_BYTES,
+				     MLXSW_SP2_BLOOM_KEY_CHUNK_BYTES);
+}
+
+static unsigned int
+mlxsw_sp2_acl_bf_index_get(struct mlxsw_sp_acl_bf *bf,
+			   struct mlxsw_sp_acl_atcam_region *aregion,
+			   struct mlxsw_sp_acl_atcam_entry *aentry)
+{
+	char bf_key[MLXSW_SP2_BLOOM_KEY_LEN];
+	u8 bf_size;
+
+	mlxsw_sp2_acl_bf_key_encode(aregion, aentry, bf_key, &bf_size);
+	return mlxsw_sp2_acl_bf_crc(bf_key, bf_size);
+}
+
+static u16 mlxsw_sp4_acl_bf_crc10_byte(u16 crc, u8 c)
+{
+	u8 index = ((crc >> 2) ^ c) & 0xff;
+
+	return ((crc << 8) ^ mlxsw_sp4_acl_bf_crc10_tab[index]) & 0x3ff;
+}
+
+static u16 mlxsw_sp4_acl_bf_crc6_byte(u16 crc, u8 c)
+{
+	u8 index = (crc ^ c) & 0xff;
+
+	return ((crc << 6) ^ (mlxsw_sp4_acl_bf_crc6_tab[index] << 2)) & 0xfc;
+}
+
+static u16 mlxsw_sp4_acl_bf_crc(const u8 *buffer, size_t len)
+{
+	u16 crc_row = 0, crc_col = 0;
+
+	while (len--) {
+		crc_row = mlxsw_sp4_acl_bf_crc10_byte(crc_row, *buffer);
+		crc_col = mlxsw_sp4_acl_bf_crc6_byte(crc_col, *buffer);
+		buffer++;
+	}
+
+	crc_col >>= 2;
+
+	/* 6 bit column are MSB, 10 bit row are LSB */
+	return (crc_col << 10) | crc_row;
+}
+
+static void right_shift_array(char *arr, u8 len, u8 shift_bits)
+{
+	u8 byte_mask = 0xff >> shift_bits;
+	int i;
+
+	if (WARN_ON(!shift_bits || shift_bits >= 8))
+		return;
+
+	for (i = len - 1; i >= 0; i--) {
+		/* The first iteration looks like out-of-bounds access,
+		 * but actually references a buffer that the array is shifted
+		 * into. This move is legal as we never send the last chunk to
+		 * this function.
+		 */
+		arr[i + 1] &= byte_mask;
+		arr[i + 1] |= arr[i] << (8 - shift_bits);
+		arr[i] = arr[i] >> shift_bits;
+	}
+}
+
+static void mlxsw_sp4_bf_key_shift_chunks(u8 chunk_count, char *output)
+{
+	/* The chunks are suppoosed to be continuous, with no padding.
+	 * Since region ID and eRP ID use 14 bits, and not fully 2 bytes,
+	 * and in Spectrum-4 there is no padding, it is necessary to shift some
+	 * chunks 2 bits right.
+	 */
+	switch (chunk_count) {
+	case 2:
+		/* The chunks are copied as follow:
+		 * +-------------+-----------------+
+		 * | Chunk 0     |   Chunk 1       |
+		 * | IDs  | keys |(**) IDs  | keys |
+		 * +-------------+-----------------+
+		 * In (**), there are two unused bits, therefore, chunk 0 needs
+		 * to be shifted two bits right.
+		 */
+		right_shift_array(output, MLXSW_SP4_BLOOM_KEY_CHUNK_BYTES, 2);
+		break;
+	case 3:
+		/* The chunks are copied as follow:
+		 * +-------------+-----------------+-----------------+
+		 * | Chunk 0     |   Chunk 1       |   Chunk 2       |
+		 * | IDs  | keys |(**) IDs  | keys |(**) IDs  | keys |
+		 * +-------------+-----------------+-----------------+
+		 * In (**), there are two unused bits, therefore, chunk 1 needs
+		 * to be shifted two bits right and chunk 0 needs to be shifted
+		 * four bits right.
+		 */
+		right_shift_array(output + MLXSW_SP4_BLOOM_KEY_CHUNK_BYTES,
+				  MLXSW_SP4_BLOOM_KEY_CHUNK_BYTES, 2);
+		right_shift_array(output, MLXSW_SP4_BLOOM_KEY_CHUNK_BYTES, 4);
+		break;
+	default:
+		WARN_ON(chunk_count > MLXSW_BLOOM_KEY_CHUNKS);
+	}
+}
+
+static void
+mlxsw_sp4_acl_bf_key_encode(struct mlxsw_sp_acl_atcam_region *aregion,
+			    struct mlxsw_sp_acl_atcam_entry *aentry,
+			    char *output, u8 *len)
+{
+	struct mlxsw_afk_key_info *key_info = aregion->region->key_info;
+	u8 block_count = mlxsw_afk_key_info_blocks_count_get(key_info);
+	u8 chunk_count = 1 + ((block_count - 1) >> 2);
+
+	__mlxsw_sp_acl_bf_key_encode(aregion, aentry, output, len,
+				     MLXSW_BLOOM_KEY_CHUNKS,
+				     MLXSW_SP4_BLOOM_CHUNK_PAD_BYTES,
+				     MLXSW_SP4_BLOOM_CHUNK_KEY_OFFSET,
+				     MLXSW_SP4_BLOOM_CHUNK_KEY_BYTES,
+				     MLXSW_SP4_BLOOM_KEY_CHUNK_BYTES);
+	mlxsw_sp4_bf_key_shift_chunks(chunk_count, output);
+}
+
+static unsigned int
+mlxsw_sp4_acl_bf_index_get(struct mlxsw_sp_acl_bf *bf,
+			   struct mlxsw_sp_acl_atcam_region *aregion,
+			   struct mlxsw_sp_acl_atcam_entry *aentry)
+{
+	char bf_key[MLXSW_SP4_BLOOM_KEY_LEN] = {};
+	u8 bf_size;
+
+	mlxsw_sp4_acl_bf_key_encode(aregion, aentry, bf_key, &bf_size);
+	return mlxsw_sp4_acl_bf_crc(bf_key, bf_size);
+}
+
+static unsigned int
+mlxsw_sp_acl_bf_rule_count_index_get(struct mlxsw_sp_acl_bf *bf,
+				     unsigned int erp_bank,
+				     unsigned int bf_index)
+{
+	return erp_bank * bf->bank_size + bf_index;
+}
+
+int
+mlxsw_sp_acl_bf_entry_add(struct mlxsw_sp *mlxsw_sp,
+			  struct mlxsw_sp_acl_bf *bf,
+			  struct mlxsw_sp_acl_atcam_region *aregion,
+			  unsigned int erp_bank,
+			  struct mlxsw_sp_acl_atcam_entry *aentry)
+{
+	unsigned int rule_index;
+	char *peabfe_pl;
+	u16 bf_index;
+	int err;
+
+	mutex_lock(&bf->lock);
+
+	bf_index = mlxsw_sp->acl_bf_ops->index_get(bf, aregion, aentry);
+	rule_index = mlxsw_sp_acl_bf_rule_count_index_get(bf, erp_bank,
+							  bf_index);
+
+	if (refcount_inc_not_zero(&bf->refcnt[rule_index])) {
+		err = 0;
+		goto unlock;
+	}
+
+	peabfe_pl = kmalloc(MLXSW_REG_PEABFE_LEN, GFP_KERNEL);
+	if (!peabfe_pl) {
+		err = -ENOMEM;
+		goto unlock;
+	}
+
+	mlxsw_reg_peabfe_pack(peabfe_pl);
+	mlxsw_reg_peabfe_rec_pack(peabfe_pl, 0, 1, erp_bank, bf_index);
+	err = mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(peabfe), peabfe_pl);
+	kfree(peabfe_pl);
+	if (err)
+		goto unlock;
+
+	refcount_set(&bf->refcnt[rule_index], 1);
+	err = 0;
+
+unlock:
+	mutex_unlock(&bf->lock);
+	return err;
+}
+
+void
+mlxsw_sp_acl_bf_entry_del(struct mlxsw_sp *mlxsw_sp,
+			  struct mlxsw_sp_acl_bf *bf,
+			  struct mlxsw_sp_acl_atcam_region *aregion,
+			  unsigned int erp_bank,
+			  struct mlxsw_sp_acl_atcam_entry *aentry)
+{
+	unsigned int rule_index;
+	char *peabfe_pl;
+	u16 bf_index;
+
+	mutex_lock(&bf->lock);
+
+	bf_index = mlxsw_sp->acl_bf_ops->index_get(bf, aregion, aentry);
+	rule_index = mlxsw_sp_acl_bf_rule_count_index_get(bf, erp_bank,
+							  bf_index);
+
+	if (refcount_dec_and_test(&bf->refcnt[rule_index])) {
+		peabfe_pl = kmalloc(MLXSW_REG_PEABFE_LEN, GFP_KERNEL);
+		if (!peabfe_pl)
+			goto unlock;
+
+		mlxsw_reg_peabfe_pack(peabfe_pl);
+		mlxsw_reg_peabfe_rec_pack(peabfe_pl, 0, 0, erp_bank, bf_index);
+		mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(peabfe), peabfe_pl);
+		kfree(peabfe_pl);
+	}
+
+unlock:
+	mutex_unlock(&bf->lock);
+}
+
+struct mlxsw_sp_acl_bf *
+mlxsw_sp_acl_bf_init(struct mlxsw_sp *mlxsw_sp, unsigned int num_erp_banks)
+{
+	struct mlxsw_sp_acl_bf *bf;
+	unsigned int bf_bank_size;
+
+	if (!MLXSW_CORE_RES_VALID(mlxsw_sp->core, ACL_MAX_BF_LOG))
+		return ERR_PTR(-EIO);
+
+	/* Bloom filter size per erp_table_bank
+	 * is 2^ACL_MAX_BF_LOG
+	 */
+	bf_bank_size = 1 << MLXSW_CORE_RES_GET(mlxsw_sp->core, ACL_MAX_BF_LOG);
+	bf = kzalloc(struct_size(bf, refcnt, size_mul(bf_bank_size, num_erp_banks)),
+		     GFP_KERNEL);
+	if (!bf)
+		return ERR_PTR(-ENOMEM);
+
+	bf->bank_size = bf_bank_size;
+	mutex_init(&bf->lock);
+
+	return bf;
+}
+
+void mlxsw_sp_acl_bf_fini(struct mlxsw_sp_acl_bf *bf)
+{
+	mutex_destroy(&bf->lock);
+	kfree(bf);
+}
+
+const struct mlxsw_sp_acl_bf_ops mlxsw_sp2_acl_bf_ops = {
+	.index_get = mlxsw_sp2_acl_bf_index_get,
+};
+
+const struct mlxsw_sp_acl_bf_ops mlxsw_sp4_acl_bf_ops = {
+	.index_get = mlxsw_sp4_acl_bf_index_get,
+};
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_acl_ctcam.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_acl_ctcam.c
new file mode 100644
index 000000000..05680a7e6
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_acl_ctcam.c
@@ -0,0 +1,233 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2017-2018 Mellanox Technologies. All rights reserved */
+
+#include <linux/kernel.h>
+#include <linux/errno.h>
+#include <linux/parman.h>
+
+#include "reg.h"
+#include "core.h"
+#include "spectrum.h"
+#include "spectrum_acl_tcam.h"
+
+static int
+mlxsw_sp_acl_ctcam_region_resize(struct mlxsw_sp *mlxsw_sp,
+				 struct mlxsw_sp_acl_tcam_region *region,
+				 u16 new_size)
+{
+	char ptar_pl[MLXSW_REG_PTAR_LEN];
+
+	mlxsw_reg_ptar_pack(ptar_pl, MLXSW_REG_PTAR_OP_RESIZE,
+			    region->key_type, new_size, region->id,
+			    region->tcam_region_info);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(ptar), ptar_pl);
+}
+
+static void
+mlxsw_sp_acl_ctcam_region_move(struct mlxsw_sp *mlxsw_sp,
+			       struct mlxsw_sp_acl_tcam_region *region,
+			       u16 src_offset, u16 dst_offset, u16 size)
+{
+	char prcr_pl[MLXSW_REG_PRCR_LEN];
+
+	mlxsw_reg_prcr_pack(prcr_pl, MLXSW_REG_PRCR_OP_MOVE,
+			    region->tcam_region_info, src_offset,
+			    region->tcam_region_info, dst_offset, size);
+	mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(prcr), prcr_pl);
+}
+
+static int
+mlxsw_sp_acl_ctcam_region_entry_insert(struct mlxsw_sp *mlxsw_sp,
+				       struct mlxsw_sp_acl_ctcam_region *cregion,
+				       struct mlxsw_sp_acl_ctcam_entry *centry,
+				       struct mlxsw_sp_acl_rule_info *rulei,
+				       bool fillup_priority)
+{
+	struct mlxsw_sp_acl_tcam_region *region = cregion->region;
+	struct mlxsw_afk *afk = mlxsw_sp_acl_afk(mlxsw_sp->acl);
+	char ptce2_pl[MLXSW_REG_PTCE2_LEN];
+	char *act_set;
+	u32 priority;
+	char *mask;
+	char *key;
+	int err;
+
+	err = mlxsw_sp_acl_tcam_priority_get(mlxsw_sp, rulei, &priority,
+					     fillup_priority);
+	if (err)
+		return err;
+
+	mlxsw_reg_ptce2_pack(ptce2_pl, true, MLXSW_REG_PTCE2_OP_WRITE_WRITE,
+			     region->tcam_region_info,
+			     centry->parman_item.index, priority);
+	key = mlxsw_reg_ptce2_flex_key_blocks_data(ptce2_pl);
+	mask = mlxsw_reg_ptce2_mask_data(ptce2_pl);
+	mlxsw_afk_encode(afk, region->key_info, &rulei->values, key, mask);
+
+	err = cregion->ops->entry_insert(cregion, centry, mask);
+	if (err)
+		return err;
+
+	/* Only the first action set belongs here, the rest is in KVD */
+	act_set = mlxsw_afa_block_first_set(rulei->act_block);
+	mlxsw_reg_ptce2_flex_action_set_memcpy_to(ptce2_pl, act_set);
+
+	err = mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(ptce2), ptce2_pl);
+	if (err)
+		goto err_ptce2_write;
+
+	return 0;
+
+err_ptce2_write:
+	cregion->ops->entry_remove(cregion, centry);
+	return err;
+}
+
+static void
+mlxsw_sp_acl_ctcam_region_entry_remove(struct mlxsw_sp *mlxsw_sp,
+				       struct mlxsw_sp_acl_ctcam_region *cregion,
+				       struct mlxsw_sp_acl_ctcam_entry *centry)
+{
+	char ptce2_pl[MLXSW_REG_PTCE2_LEN];
+
+	mlxsw_reg_ptce2_pack(ptce2_pl, false, MLXSW_REG_PTCE2_OP_WRITE_WRITE,
+			     cregion->region->tcam_region_info,
+			     centry->parman_item.index, 0);
+	mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(ptce2), ptce2_pl);
+	cregion->ops->entry_remove(cregion, centry);
+}
+
+static int
+mlxsw_sp_acl_ctcam_region_entry_action_replace(struct mlxsw_sp *mlxsw_sp,
+					       struct mlxsw_sp_acl_ctcam_region *cregion,
+					       struct mlxsw_sp_acl_ctcam_entry *centry,
+					       struct mlxsw_afa_block *afa_block,
+					       unsigned int priority)
+{
+	char ptce2_pl[MLXSW_REG_PTCE2_LEN];
+	char *act_set;
+
+	mlxsw_reg_ptce2_pack(ptce2_pl, true, MLXSW_REG_PTCE2_OP_WRITE_UPDATE,
+			     cregion->region->tcam_region_info,
+			     centry->parman_item.index, priority);
+
+	act_set = mlxsw_afa_block_first_set(afa_block);
+	mlxsw_reg_ptce2_flex_action_set_memcpy_to(ptce2_pl, act_set);
+
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(ptce2), ptce2_pl);
+}
+
+
+static int mlxsw_sp_acl_ctcam_region_parman_resize(void *priv,
+						   unsigned long new_count)
+{
+	struct mlxsw_sp_acl_ctcam_region *cregion = priv;
+	struct mlxsw_sp_acl_tcam_region *region = cregion->region;
+	struct mlxsw_sp *mlxsw_sp = region->mlxsw_sp;
+	u64 max_tcam_rules;
+
+	max_tcam_rules = MLXSW_CORE_RES_GET(mlxsw_sp->core, ACL_MAX_TCAM_RULES);
+	if (new_count > max_tcam_rules)
+		return -EINVAL;
+	return mlxsw_sp_acl_ctcam_region_resize(mlxsw_sp, region, new_count);
+}
+
+static void mlxsw_sp_acl_ctcam_region_parman_move(void *priv,
+						  unsigned long from_index,
+						  unsigned long to_index,
+						  unsigned long count)
+{
+	struct mlxsw_sp_acl_ctcam_region *cregion = priv;
+	struct mlxsw_sp_acl_tcam_region *region = cregion->region;
+	struct mlxsw_sp *mlxsw_sp = region->mlxsw_sp;
+
+	mlxsw_sp_acl_ctcam_region_move(mlxsw_sp, region,
+				       from_index, to_index, count);
+}
+
+static const struct parman_ops mlxsw_sp_acl_ctcam_region_parman_ops = {
+	.base_count	= MLXSW_SP_ACL_TCAM_REGION_BASE_COUNT,
+	.resize_step	= MLXSW_SP_ACL_TCAM_REGION_RESIZE_STEP,
+	.resize		= mlxsw_sp_acl_ctcam_region_parman_resize,
+	.move		= mlxsw_sp_acl_ctcam_region_parman_move,
+	.algo		= PARMAN_ALGO_TYPE_LSORT,
+};
+
+int
+mlxsw_sp_acl_ctcam_region_init(struct mlxsw_sp *mlxsw_sp,
+			       struct mlxsw_sp_acl_ctcam_region *cregion,
+			       struct mlxsw_sp_acl_tcam_region *region,
+			       const struct mlxsw_sp_acl_ctcam_region_ops *ops)
+{
+	cregion->region = region;
+	cregion->ops = ops;
+	cregion->parman = parman_create(&mlxsw_sp_acl_ctcam_region_parman_ops,
+					cregion);
+	if (!cregion->parman)
+		return -ENOMEM;
+	return 0;
+}
+
+void mlxsw_sp_acl_ctcam_region_fini(struct mlxsw_sp_acl_ctcam_region *cregion)
+{
+	parman_destroy(cregion->parman);
+}
+
+void mlxsw_sp_acl_ctcam_chunk_init(struct mlxsw_sp_acl_ctcam_region *cregion,
+				   struct mlxsw_sp_acl_ctcam_chunk *cchunk,
+				   unsigned int priority)
+{
+	parman_prio_init(cregion->parman, &cchunk->parman_prio, priority);
+}
+
+void mlxsw_sp_acl_ctcam_chunk_fini(struct mlxsw_sp_acl_ctcam_chunk *cchunk)
+{
+	parman_prio_fini(&cchunk->parman_prio);
+}
+
+int mlxsw_sp_acl_ctcam_entry_add(struct mlxsw_sp *mlxsw_sp,
+				 struct mlxsw_sp_acl_ctcam_region *cregion,
+				 struct mlxsw_sp_acl_ctcam_chunk *cchunk,
+				 struct mlxsw_sp_acl_ctcam_entry *centry,
+				 struct mlxsw_sp_acl_rule_info *rulei,
+				 bool fillup_priority)
+{
+	int err;
+
+	err = parman_item_add(cregion->parman, &cchunk->parman_prio,
+			      &centry->parman_item);
+	if (err)
+		return err;
+
+	err = mlxsw_sp_acl_ctcam_region_entry_insert(mlxsw_sp, cregion, centry,
+						     rulei, fillup_priority);
+	if (err)
+		goto err_rule_insert;
+	return 0;
+
+err_rule_insert:
+	parman_item_remove(cregion->parman, &cchunk->parman_prio,
+			   &centry->parman_item);
+	return err;
+}
+
+void mlxsw_sp_acl_ctcam_entry_del(struct mlxsw_sp *mlxsw_sp,
+				  struct mlxsw_sp_acl_ctcam_region *cregion,
+				  struct mlxsw_sp_acl_ctcam_chunk *cchunk,
+				  struct mlxsw_sp_acl_ctcam_entry *centry)
+{
+	mlxsw_sp_acl_ctcam_region_entry_remove(mlxsw_sp, cregion, centry);
+	parman_item_remove(cregion->parman, &cchunk->parman_prio,
+			   &centry->parman_item);
+}
+
+int mlxsw_sp_acl_ctcam_entry_action_replace(struct mlxsw_sp *mlxsw_sp,
+					    struct mlxsw_sp_acl_ctcam_region *cregion,
+					    struct mlxsw_sp_acl_ctcam_entry *centry,
+					    struct mlxsw_sp_acl_rule_info *rulei)
+{
+	return mlxsw_sp_acl_ctcam_region_entry_action_replace(mlxsw_sp, cregion,
+							      centry,
+							      rulei->act_block,
+							      rulei->priority);
+}
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_acl_erp.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_acl_erp.c
new file mode 100644
index 000000000..d231f4d28
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_acl_erp.c
@@ -0,0 +1,1601 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2018 Mellanox Technologies. All rights reserved */
+
+#include <linux/bitmap.h>
+#include <linux/errno.h>
+#include <linux/genalloc.h>
+#include <linux/gfp.h>
+#include <linux/kernel.h>
+#include <linux/list.h>
+#include <linux/mutex.h>
+#include <linux/objagg.h>
+#include <linux/rtnetlink.h>
+#include <linux/slab.h>
+
+#include "core.h"
+#include "reg.h"
+#include "spectrum.h"
+#include "spectrum_acl_tcam.h"
+
+/* gen_pool_alloc() returns 0 when allocation fails, so use an offset */
+#define MLXSW_SP_ACL_ERP_GENALLOC_OFFSET 0x100
+#define MLXSW_SP_ACL_ERP_MAX_PER_REGION 16
+
+struct mlxsw_sp_acl_erp_core {
+	unsigned int erpt_entries_size[MLXSW_SP_ACL_ATCAM_REGION_TYPE_MAX + 1];
+	struct gen_pool *erp_tables;
+	struct mlxsw_sp *mlxsw_sp;
+	struct mlxsw_sp_acl_bf *bf;
+	unsigned int num_erp_banks;
+};
+
+struct mlxsw_sp_acl_erp_key {
+	char mask[MLXSW_REG_PTCEX_FLEX_KEY_BLOCKS_LEN];
+#define __MASK_LEN 0x38
+#define __MASK_IDX(i) (__MASK_LEN - (i) - 1)
+	bool ctcam;
+};
+
+struct mlxsw_sp_acl_erp {
+	struct mlxsw_sp_acl_erp_key key;
+	u8 id;
+	u8 index;
+	DECLARE_BITMAP(mask_bitmap, MLXSW_SP_ACL_TCAM_MASK_LEN);
+	struct list_head list;
+	struct mlxsw_sp_acl_erp_table *erp_table;
+};
+
+struct mlxsw_sp_acl_erp_master_mask {
+	DECLARE_BITMAP(bitmap, MLXSW_SP_ACL_TCAM_MASK_LEN);
+	unsigned int count[MLXSW_SP_ACL_TCAM_MASK_LEN];
+};
+
+struct mlxsw_sp_acl_erp_table {
+	struct mlxsw_sp_acl_erp_master_mask master_mask;
+	DECLARE_BITMAP(erp_id_bitmap, MLXSW_SP_ACL_ERP_MAX_PER_REGION);
+	DECLARE_BITMAP(erp_index_bitmap, MLXSW_SP_ACL_ERP_MAX_PER_REGION);
+	struct list_head atcam_erps_list;
+	struct mlxsw_sp_acl_erp_core *erp_core;
+	struct mlxsw_sp_acl_atcam_region *aregion;
+	const struct mlxsw_sp_acl_erp_table_ops *ops;
+	unsigned long base_index;
+	unsigned int num_atcam_erps;
+	unsigned int num_max_atcam_erps;
+	unsigned int num_ctcam_erps;
+	unsigned int num_deltas;
+	struct objagg *objagg;
+	struct mutex objagg_lock; /* guards objagg manipulation */
+};
+
+struct mlxsw_sp_acl_erp_table_ops {
+	struct mlxsw_sp_acl_erp *
+		(*erp_create)(struct mlxsw_sp_acl_erp_table *erp_table,
+			      struct mlxsw_sp_acl_erp_key *key);
+	void (*erp_destroy)(struct mlxsw_sp_acl_erp_table *erp_table,
+			    struct mlxsw_sp_acl_erp *erp);
+};
+
+static struct mlxsw_sp_acl_erp *
+mlxsw_sp_acl_erp_mask_create(struct mlxsw_sp_acl_erp_table *erp_table,
+			     struct mlxsw_sp_acl_erp_key *key);
+static void
+mlxsw_sp_acl_erp_mask_destroy(struct mlxsw_sp_acl_erp_table *erp_table,
+			      struct mlxsw_sp_acl_erp *erp);
+static struct mlxsw_sp_acl_erp *
+mlxsw_sp_acl_erp_second_mask_create(struct mlxsw_sp_acl_erp_table *erp_table,
+				    struct mlxsw_sp_acl_erp_key *key);
+static void
+mlxsw_sp_acl_erp_second_mask_destroy(struct mlxsw_sp_acl_erp_table *erp_table,
+				     struct mlxsw_sp_acl_erp *erp);
+static struct mlxsw_sp_acl_erp *
+mlxsw_sp_acl_erp_first_mask_create(struct mlxsw_sp_acl_erp_table *erp_table,
+				   struct mlxsw_sp_acl_erp_key *key);
+static void
+mlxsw_sp_acl_erp_first_mask_destroy(struct mlxsw_sp_acl_erp_table *erp_table,
+				    struct mlxsw_sp_acl_erp *erp);
+static void
+mlxsw_sp_acl_erp_no_mask_destroy(struct mlxsw_sp_acl_erp_table *erp_table,
+				 struct mlxsw_sp_acl_erp *erp);
+
+static const struct mlxsw_sp_acl_erp_table_ops erp_multiple_masks_ops = {
+	.erp_create = mlxsw_sp_acl_erp_mask_create,
+	.erp_destroy = mlxsw_sp_acl_erp_mask_destroy,
+};
+
+static const struct mlxsw_sp_acl_erp_table_ops erp_two_masks_ops = {
+	.erp_create = mlxsw_sp_acl_erp_mask_create,
+	.erp_destroy = mlxsw_sp_acl_erp_second_mask_destroy,
+};
+
+static const struct mlxsw_sp_acl_erp_table_ops erp_single_mask_ops = {
+	.erp_create = mlxsw_sp_acl_erp_second_mask_create,
+	.erp_destroy = mlxsw_sp_acl_erp_first_mask_destroy,
+};
+
+static const struct mlxsw_sp_acl_erp_table_ops erp_no_mask_ops = {
+	.erp_create = mlxsw_sp_acl_erp_first_mask_create,
+	.erp_destroy = mlxsw_sp_acl_erp_no_mask_destroy,
+};
+
+static bool
+mlxsw_sp_acl_erp_table_is_used(const struct mlxsw_sp_acl_erp_table *erp_table)
+{
+	return erp_table->ops != &erp_single_mask_ops &&
+	       erp_table->ops != &erp_no_mask_ops;
+}
+
+static unsigned int
+mlxsw_sp_acl_erp_bank_get(const struct mlxsw_sp_acl_erp *erp)
+{
+	return erp->index % erp->erp_table->erp_core->num_erp_banks;
+}
+
+static unsigned int
+mlxsw_sp_acl_erp_table_entry_size(const struct mlxsw_sp_acl_erp_table *erp_table)
+{
+	struct mlxsw_sp_acl_atcam_region *aregion = erp_table->aregion;
+	struct mlxsw_sp_acl_erp_core *erp_core = erp_table->erp_core;
+
+	return erp_core->erpt_entries_size[aregion->type];
+}
+
+static int mlxsw_sp_acl_erp_id_get(struct mlxsw_sp_acl_erp_table *erp_table,
+				   u8 *p_id)
+{
+	u8 id;
+
+	id = find_first_zero_bit(erp_table->erp_id_bitmap,
+				 MLXSW_SP_ACL_ERP_MAX_PER_REGION);
+	if (id < MLXSW_SP_ACL_ERP_MAX_PER_REGION) {
+		__set_bit(id, erp_table->erp_id_bitmap);
+		*p_id = id;
+		return 0;
+	}
+
+	return -ENOBUFS;
+}
+
+static void mlxsw_sp_acl_erp_id_put(struct mlxsw_sp_acl_erp_table *erp_table,
+				    u8 id)
+{
+	__clear_bit(id, erp_table->erp_id_bitmap);
+}
+
+static void
+mlxsw_sp_acl_erp_master_mask_bit_set(unsigned long bit,
+				     struct mlxsw_sp_acl_erp_master_mask *mask)
+{
+	if (mask->count[bit]++ == 0)
+		__set_bit(bit, mask->bitmap);
+}
+
+static void
+mlxsw_sp_acl_erp_master_mask_bit_clear(unsigned long bit,
+				       struct mlxsw_sp_acl_erp_master_mask *mask)
+{
+	if (--mask->count[bit] == 0)
+		__clear_bit(bit, mask->bitmap);
+}
+
+static int
+mlxsw_sp_acl_erp_master_mask_update(struct mlxsw_sp_acl_erp_table *erp_table)
+{
+	struct mlxsw_sp_acl_tcam_region *region = erp_table->aregion->region;
+	struct mlxsw_sp *mlxsw_sp = region->mlxsw_sp;
+	char percr_pl[MLXSW_REG_PERCR_LEN];
+	char *master_mask;
+
+	mlxsw_reg_percr_pack(percr_pl, region->id);
+	master_mask = mlxsw_reg_percr_master_mask_data(percr_pl);
+	bitmap_to_arr32((u32 *) master_mask, erp_table->master_mask.bitmap,
+			MLXSW_SP_ACL_TCAM_MASK_LEN);
+
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(percr), percr_pl);
+}
+
+static int
+mlxsw_sp_acl_erp_master_mask_set(struct mlxsw_sp_acl_erp_table *erp_table,
+				 struct mlxsw_sp_acl_erp_key *key)
+{
+	DECLARE_BITMAP(mask_bitmap, MLXSW_SP_ACL_TCAM_MASK_LEN);
+	unsigned long bit;
+	int err;
+
+	bitmap_from_arr32(mask_bitmap, (u32 *) key->mask,
+			  MLXSW_SP_ACL_TCAM_MASK_LEN);
+	for_each_set_bit(bit, mask_bitmap, MLXSW_SP_ACL_TCAM_MASK_LEN)
+		mlxsw_sp_acl_erp_master_mask_bit_set(bit,
+						     &erp_table->master_mask);
+
+	err = mlxsw_sp_acl_erp_master_mask_update(erp_table);
+	if (err)
+		goto err_master_mask_update;
+
+	return 0;
+
+err_master_mask_update:
+	for_each_set_bit(bit, mask_bitmap, MLXSW_SP_ACL_TCAM_MASK_LEN)
+		mlxsw_sp_acl_erp_master_mask_bit_clear(bit,
+						       &erp_table->master_mask);
+	return err;
+}
+
+static int
+mlxsw_sp_acl_erp_master_mask_clear(struct mlxsw_sp_acl_erp_table *erp_table,
+				   struct mlxsw_sp_acl_erp_key *key)
+{
+	DECLARE_BITMAP(mask_bitmap, MLXSW_SP_ACL_TCAM_MASK_LEN);
+	unsigned long bit;
+	int err;
+
+	bitmap_from_arr32(mask_bitmap, (u32 *) key->mask,
+			  MLXSW_SP_ACL_TCAM_MASK_LEN);
+	for_each_set_bit(bit, mask_bitmap, MLXSW_SP_ACL_TCAM_MASK_LEN)
+		mlxsw_sp_acl_erp_master_mask_bit_clear(bit,
+						       &erp_table->master_mask);
+
+	err = mlxsw_sp_acl_erp_master_mask_update(erp_table);
+	if (err)
+		goto err_master_mask_update;
+
+	return 0;
+
+err_master_mask_update:
+	for_each_set_bit(bit, mask_bitmap, MLXSW_SP_ACL_TCAM_MASK_LEN)
+		mlxsw_sp_acl_erp_master_mask_bit_set(bit,
+						     &erp_table->master_mask);
+	return err;
+}
+
+static struct mlxsw_sp_acl_erp *
+mlxsw_sp_acl_erp_generic_create(struct mlxsw_sp_acl_erp_table *erp_table,
+				struct mlxsw_sp_acl_erp_key *key)
+{
+	struct mlxsw_sp_acl_erp *erp;
+	int err;
+
+	erp = kzalloc(sizeof(*erp), GFP_KERNEL);
+	if (!erp)
+		return ERR_PTR(-ENOMEM);
+
+	err = mlxsw_sp_acl_erp_id_get(erp_table, &erp->id);
+	if (err)
+		goto err_erp_id_get;
+
+	memcpy(&erp->key, key, sizeof(*key));
+	list_add(&erp->list, &erp_table->atcam_erps_list);
+	erp_table->num_atcam_erps++;
+	erp->erp_table = erp_table;
+
+	err = mlxsw_sp_acl_erp_master_mask_set(erp_table, &erp->key);
+	if (err)
+		goto err_master_mask_set;
+
+	return erp;
+
+err_master_mask_set:
+	erp_table->num_atcam_erps--;
+	list_del(&erp->list);
+	mlxsw_sp_acl_erp_id_put(erp_table, erp->id);
+err_erp_id_get:
+	kfree(erp);
+	return ERR_PTR(err);
+}
+
+static void
+mlxsw_sp_acl_erp_generic_destroy(struct mlxsw_sp_acl_erp *erp)
+{
+	struct mlxsw_sp_acl_erp_table *erp_table = erp->erp_table;
+
+	mlxsw_sp_acl_erp_master_mask_clear(erp_table, &erp->key);
+	erp_table->num_atcam_erps--;
+	list_del(&erp->list);
+	mlxsw_sp_acl_erp_id_put(erp_table, erp->id);
+	kfree(erp);
+}
+
+static int
+mlxsw_sp_acl_erp_table_alloc(struct mlxsw_sp_acl_erp_core *erp_core,
+			     unsigned int num_erps,
+			     enum mlxsw_sp_acl_atcam_region_type region_type,
+			     unsigned long *p_index)
+{
+	unsigned int num_rows, entry_size;
+	unsigned long index;
+
+	/* We only allow allocations of entire rows */
+	if (num_erps % erp_core->num_erp_banks != 0)
+		return -EINVAL;
+
+	entry_size = erp_core->erpt_entries_size[region_type];
+	num_rows = num_erps / erp_core->num_erp_banks;
+
+	index = gen_pool_alloc(erp_core->erp_tables, num_rows * entry_size);
+	if (!index)
+		return -ENOBUFS;
+
+	*p_index = index - MLXSW_SP_ACL_ERP_GENALLOC_OFFSET;
+
+	return 0;
+}
+
+static void
+mlxsw_sp_acl_erp_table_free(struct mlxsw_sp_acl_erp_core *erp_core,
+			    unsigned int num_erps,
+			    enum mlxsw_sp_acl_atcam_region_type region_type,
+			    unsigned long index)
+{
+	unsigned long base_index;
+	unsigned int entry_size;
+	size_t size;
+
+	entry_size = erp_core->erpt_entries_size[region_type];
+	base_index = index + MLXSW_SP_ACL_ERP_GENALLOC_OFFSET;
+	size = num_erps / erp_core->num_erp_banks * entry_size;
+	gen_pool_free(erp_core->erp_tables, base_index, size);
+}
+
+static struct mlxsw_sp_acl_erp *
+mlxsw_sp_acl_erp_table_master_rp(struct mlxsw_sp_acl_erp_table *erp_table)
+{
+	if (!list_is_singular(&erp_table->atcam_erps_list))
+		return NULL;
+
+	return list_first_entry(&erp_table->atcam_erps_list,
+				struct mlxsw_sp_acl_erp, list);
+}
+
+static int mlxsw_sp_acl_erp_index_get(struct mlxsw_sp_acl_erp_table *erp_table,
+				      u8 *p_index)
+{
+	u8 index;
+
+	index = find_first_zero_bit(erp_table->erp_index_bitmap,
+				    erp_table->num_max_atcam_erps);
+	if (index < erp_table->num_max_atcam_erps) {
+		__set_bit(index, erp_table->erp_index_bitmap);
+		*p_index = index;
+		return 0;
+	}
+
+	return -ENOBUFS;
+}
+
+static void mlxsw_sp_acl_erp_index_put(struct mlxsw_sp_acl_erp_table *erp_table,
+				       u8 index)
+{
+	__clear_bit(index, erp_table->erp_index_bitmap);
+}
+
+static void
+mlxsw_sp_acl_erp_table_locate(const struct mlxsw_sp_acl_erp_table *erp_table,
+			      const struct mlxsw_sp_acl_erp *erp,
+			      u8 *p_erpt_bank, u8 *p_erpt_index)
+{
+	unsigned int entry_size = mlxsw_sp_acl_erp_table_entry_size(erp_table);
+	struct mlxsw_sp_acl_erp_core *erp_core = erp_table->erp_core;
+	unsigned int row;
+
+	*p_erpt_bank = erp->index % erp_core->num_erp_banks;
+	row = erp->index / erp_core->num_erp_banks;
+	*p_erpt_index = erp_table->base_index + row * entry_size;
+}
+
+static int
+mlxsw_sp_acl_erp_table_erp_add(struct mlxsw_sp_acl_erp_table *erp_table,
+			       struct mlxsw_sp_acl_erp *erp)
+{
+	struct mlxsw_sp *mlxsw_sp = erp_table->erp_core->mlxsw_sp;
+	enum mlxsw_reg_perpt_key_size key_size;
+	char perpt_pl[MLXSW_REG_PERPT_LEN];
+	u8 erpt_bank, erpt_index;
+
+	mlxsw_sp_acl_erp_table_locate(erp_table, erp, &erpt_bank, &erpt_index);
+	key_size = (enum mlxsw_reg_perpt_key_size) erp_table->aregion->type;
+	mlxsw_reg_perpt_pack(perpt_pl, erpt_bank, erpt_index, key_size, erp->id,
+			     0, erp_table->base_index, erp->index,
+			     erp->key.mask);
+	mlxsw_reg_perpt_erp_vector_pack(perpt_pl, erp_table->erp_index_bitmap,
+					MLXSW_SP_ACL_ERP_MAX_PER_REGION);
+	mlxsw_reg_perpt_erp_vector_set(perpt_pl, erp->index, true);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(perpt), perpt_pl);
+}
+
+static void mlxsw_sp_acl_erp_table_erp_del(struct mlxsw_sp_acl_erp *erp)
+{
+	char empty_mask[MLXSW_REG_PTCEX_FLEX_KEY_BLOCKS_LEN] = { 0 };
+	struct mlxsw_sp_acl_erp_table *erp_table = erp->erp_table;
+	struct mlxsw_sp *mlxsw_sp = erp_table->erp_core->mlxsw_sp;
+	enum mlxsw_reg_perpt_key_size key_size;
+	char perpt_pl[MLXSW_REG_PERPT_LEN];
+	u8 erpt_bank, erpt_index;
+
+	mlxsw_sp_acl_erp_table_locate(erp_table, erp, &erpt_bank, &erpt_index);
+	key_size = (enum mlxsw_reg_perpt_key_size) erp_table->aregion->type;
+	mlxsw_reg_perpt_pack(perpt_pl, erpt_bank, erpt_index, key_size, erp->id,
+			     0, erp_table->base_index, erp->index, empty_mask);
+	mlxsw_reg_perpt_erp_vector_pack(perpt_pl, erp_table->erp_index_bitmap,
+					MLXSW_SP_ACL_ERP_MAX_PER_REGION);
+	mlxsw_reg_perpt_erp_vector_set(perpt_pl, erp->index, false);
+	mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(perpt), perpt_pl);
+}
+
+static int
+mlxsw_sp_acl_erp_table_enable(struct mlxsw_sp_acl_erp_table *erp_table,
+			      bool ctcam_le)
+{
+	struct mlxsw_sp_acl_tcam_region *region = erp_table->aregion->region;
+	struct mlxsw_sp *mlxsw_sp = erp_table->erp_core->mlxsw_sp;
+	char pererp_pl[MLXSW_REG_PERERP_LEN];
+
+	mlxsw_reg_pererp_pack(pererp_pl, region->id, ctcam_le, true, 0,
+			      erp_table->base_index, 0);
+	mlxsw_reg_pererp_erp_vector_pack(pererp_pl, erp_table->erp_index_bitmap,
+					 MLXSW_SP_ACL_ERP_MAX_PER_REGION);
+
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(pererp), pererp_pl);
+}
+
+static void
+mlxsw_sp_acl_erp_table_disable(struct mlxsw_sp_acl_erp_table *erp_table)
+{
+	struct mlxsw_sp_acl_tcam_region *region = erp_table->aregion->region;
+	struct mlxsw_sp *mlxsw_sp = erp_table->erp_core->mlxsw_sp;
+	char pererp_pl[MLXSW_REG_PERERP_LEN];
+	struct mlxsw_sp_acl_erp *master_rp;
+
+	master_rp = mlxsw_sp_acl_erp_table_master_rp(erp_table);
+	/* It is possible we do not have a master RP when we disable the
+	 * table when there are no rules in the A-TCAM and the last C-TCAM
+	 * rule is deleted
+	 */
+	mlxsw_reg_pererp_pack(pererp_pl, region->id, false, false, 0, 0,
+			      master_rp ? master_rp->id : 0);
+	mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(pererp), pererp_pl);
+}
+
+static int
+mlxsw_sp_acl_erp_table_relocate(struct mlxsw_sp_acl_erp_table *erp_table)
+{
+	struct mlxsw_sp_acl_erp *erp;
+	int err;
+
+	list_for_each_entry(erp, &erp_table->atcam_erps_list, list) {
+		err = mlxsw_sp_acl_erp_table_erp_add(erp_table, erp);
+		if (err)
+			goto err_table_erp_add;
+	}
+
+	return 0;
+
+err_table_erp_add:
+	list_for_each_entry_continue_reverse(erp, &erp_table->atcam_erps_list,
+					     list)
+		mlxsw_sp_acl_erp_table_erp_del(erp);
+	return err;
+}
+
+static int
+mlxsw_sp_acl_erp_table_expand(struct mlxsw_sp_acl_erp_table *erp_table)
+{
+	unsigned int num_erps, old_num_erps = erp_table->num_max_atcam_erps;
+	struct mlxsw_sp_acl_erp_core *erp_core = erp_table->erp_core;
+	unsigned long old_base_index = erp_table->base_index;
+	bool ctcam_le = erp_table->num_ctcam_erps > 0;
+	int err;
+
+	if (erp_table->num_atcam_erps < erp_table->num_max_atcam_erps)
+		return 0;
+
+	if (erp_table->num_max_atcam_erps == MLXSW_SP_ACL_ERP_MAX_PER_REGION)
+		return -ENOBUFS;
+
+	num_erps = old_num_erps + erp_core->num_erp_banks;
+	err = mlxsw_sp_acl_erp_table_alloc(erp_core, num_erps,
+					   erp_table->aregion->type,
+					   &erp_table->base_index);
+	if (err)
+		return err;
+	erp_table->num_max_atcam_erps = num_erps;
+
+	err = mlxsw_sp_acl_erp_table_relocate(erp_table);
+	if (err)
+		goto err_table_relocate;
+
+	err = mlxsw_sp_acl_erp_table_enable(erp_table, ctcam_le);
+	if (err)
+		goto err_table_enable;
+
+	mlxsw_sp_acl_erp_table_free(erp_core, old_num_erps,
+				    erp_table->aregion->type, old_base_index);
+
+	return 0;
+
+err_table_enable:
+err_table_relocate:
+	erp_table->num_max_atcam_erps = old_num_erps;
+	mlxsw_sp_acl_erp_table_free(erp_core, num_erps,
+				    erp_table->aregion->type,
+				    erp_table->base_index);
+	erp_table->base_index = old_base_index;
+	return err;
+}
+
+static int
+mlxsw_acl_erp_table_bf_add(struct mlxsw_sp_acl_erp_table *erp_table,
+			   struct mlxsw_sp_acl_erp *erp)
+{
+	struct mlxsw_sp_acl_atcam_region *aregion = erp_table->aregion;
+	unsigned int erp_bank = mlxsw_sp_acl_erp_bank_get(erp);
+	struct mlxsw_sp_acl_atcam_entry *aentry;
+	int err;
+
+	list_for_each_entry(aentry, &aregion->entries_list, list) {
+		err = mlxsw_sp_acl_bf_entry_add(aregion->region->mlxsw_sp,
+						erp_table->erp_core->bf,
+						aregion, erp_bank, aentry);
+		if (err)
+			goto bf_entry_add_err;
+	}
+
+	return 0;
+
+bf_entry_add_err:
+	list_for_each_entry_continue_reverse(aentry, &aregion->entries_list,
+					     list)
+		mlxsw_sp_acl_bf_entry_del(aregion->region->mlxsw_sp,
+					  erp_table->erp_core->bf,
+					  aregion, erp_bank, aentry);
+	return err;
+}
+
+static void
+mlxsw_acl_erp_table_bf_del(struct mlxsw_sp_acl_erp_table *erp_table,
+			   struct mlxsw_sp_acl_erp *erp)
+{
+	struct mlxsw_sp_acl_atcam_region *aregion = erp_table->aregion;
+	unsigned int erp_bank = mlxsw_sp_acl_erp_bank_get(erp);
+	struct mlxsw_sp_acl_atcam_entry *aentry;
+
+	list_for_each_entry_reverse(aentry, &aregion->entries_list, list)
+		mlxsw_sp_acl_bf_entry_del(aregion->region->mlxsw_sp,
+					  erp_table->erp_core->bf,
+					  aregion, erp_bank, aentry);
+}
+
+static int
+mlxsw_sp_acl_erp_region_table_trans(struct mlxsw_sp_acl_erp_table *erp_table)
+{
+	struct mlxsw_sp_acl_erp_core *erp_core = erp_table->erp_core;
+	struct mlxsw_sp_acl_erp *master_rp;
+	int err;
+
+	/* Initially, allocate a single eRP row. Expand later as needed */
+	err = mlxsw_sp_acl_erp_table_alloc(erp_core, erp_core->num_erp_banks,
+					   erp_table->aregion->type,
+					   &erp_table->base_index);
+	if (err)
+		return err;
+	erp_table->num_max_atcam_erps = erp_core->num_erp_banks;
+
+	/* Transition the sole RP currently configured (the master RP)
+	 * to the eRP table
+	 */
+	master_rp = mlxsw_sp_acl_erp_table_master_rp(erp_table);
+	if (!master_rp) {
+		err = -EINVAL;
+		goto err_table_master_rp;
+	}
+
+	/* Make sure the master RP is using a valid index, as
+	 * only a single eRP row is currently allocated.
+	 */
+	master_rp->index = 0;
+	__set_bit(master_rp->index, erp_table->erp_index_bitmap);
+
+	err = mlxsw_sp_acl_erp_table_erp_add(erp_table, master_rp);
+	if (err)
+		goto err_table_master_rp_add;
+
+	/* Update Bloom filter before enabling eRP table, as rules
+	 * on the master RP were not set to Bloom filter up to this
+	 * point.
+	 */
+	err = mlxsw_acl_erp_table_bf_add(erp_table, master_rp);
+	if (err)
+		goto err_table_bf_add;
+
+	err = mlxsw_sp_acl_erp_table_enable(erp_table, false);
+	if (err)
+		goto err_table_enable;
+
+	return 0;
+
+err_table_enable:
+	mlxsw_acl_erp_table_bf_del(erp_table, master_rp);
+err_table_bf_add:
+	mlxsw_sp_acl_erp_table_erp_del(master_rp);
+err_table_master_rp_add:
+	__clear_bit(master_rp->index, erp_table->erp_index_bitmap);
+err_table_master_rp:
+	mlxsw_sp_acl_erp_table_free(erp_core, erp_table->num_max_atcam_erps,
+				    erp_table->aregion->type,
+				    erp_table->base_index);
+	return err;
+}
+
+static void
+mlxsw_sp_acl_erp_region_master_mask_trans(struct mlxsw_sp_acl_erp_table *erp_table)
+{
+	struct mlxsw_sp_acl_erp_core *erp_core = erp_table->erp_core;
+	struct mlxsw_sp_acl_erp *master_rp;
+
+	mlxsw_sp_acl_erp_table_disable(erp_table);
+	master_rp = mlxsw_sp_acl_erp_table_master_rp(erp_table);
+	if (!master_rp)
+		return;
+	mlxsw_acl_erp_table_bf_del(erp_table, master_rp);
+	mlxsw_sp_acl_erp_table_erp_del(master_rp);
+	__clear_bit(master_rp->index, erp_table->erp_index_bitmap);
+	mlxsw_sp_acl_erp_table_free(erp_core, erp_table->num_max_atcam_erps,
+				    erp_table->aregion->type,
+				    erp_table->base_index);
+}
+
+static int
+mlxsw_sp_acl_erp_region_erp_add(struct mlxsw_sp_acl_erp_table *erp_table,
+				struct mlxsw_sp_acl_erp *erp)
+{
+	struct mlxsw_sp_acl_tcam_region *region = erp_table->aregion->region;
+	struct mlxsw_sp *mlxsw_sp = erp_table->erp_core->mlxsw_sp;
+	bool ctcam_le = erp_table->num_ctcam_erps > 0;
+	char pererp_pl[MLXSW_REG_PERERP_LEN];
+
+	mlxsw_reg_pererp_pack(pererp_pl, region->id, ctcam_le, true, 0,
+			      erp_table->base_index, 0);
+	mlxsw_reg_pererp_erp_vector_pack(pererp_pl, erp_table->erp_index_bitmap,
+					 MLXSW_SP_ACL_ERP_MAX_PER_REGION);
+	mlxsw_reg_pererp_erpt_vector_set(pererp_pl, erp->index, true);
+
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(pererp), pererp_pl);
+}
+
+static void mlxsw_sp_acl_erp_region_erp_del(struct mlxsw_sp_acl_erp *erp)
+{
+	struct mlxsw_sp_acl_erp_table *erp_table = erp->erp_table;
+	struct mlxsw_sp_acl_tcam_region *region = erp_table->aregion->region;
+	struct mlxsw_sp *mlxsw_sp = erp_table->erp_core->mlxsw_sp;
+	bool ctcam_le = erp_table->num_ctcam_erps > 0;
+	char pererp_pl[MLXSW_REG_PERERP_LEN];
+
+	mlxsw_reg_pererp_pack(pererp_pl, region->id, ctcam_le, true, 0,
+			      erp_table->base_index, 0);
+	mlxsw_reg_pererp_erp_vector_pack(pererp_pl, erp_table->erp_index_bitmap,
+					 MLXSW_SP_ACL_ERP_MAX_PER_REGION);
+	mlxsw_reg_pererp_erpt_vector_set(pererp_pl, erp->index, false);
+
+	mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(pererp), pererp_pl);
+}
+
+static int
+mlxsw_sp_acl_erp_region_ctcam_enable(struct mlxsw_sp_acl_erp_table *erp_table)
+{
+	/* No need to re-enable lookup in the C-TCAM */
+	if (erp_table->num_ctcam_erps > 1)
+		return 0;
+
+	return mlxsw_sp_acl_erp_table_enable(erp_table, true);
+}
+
+static void
+mlxsw_sp_acl_erp_region_ctcam_disable(struct mlxsw_sp_acl_erp_table *erp_table)
+{
+	/* Only disable C-TCAM lookup when last C-TCAM eRP is deleted */
+	if (erp_table->num_ctcam_erps > 1)
+		return;
+
+	mlxsw_sp_acl_erp_table_enable(erp_table, false);
+}
+
+static int
+__mlxsw_sp_acl_erp_table_other_inc(struct mlxsw_sp_acl_erp_table *erp_table,
+				   unsigned int *inc_num)
+{
+	int err;
+
+	/* If there are C-TCAM eRP or deltas in use we need to transition
+	 * the region to use eRP table, if it is not already done
+	 */
+	if (!mlxsw_sp_acl_erp_table_is_used(erp_table)) {
+		err = mlxsw_sp_acl_erp_region_table_trans(erp_table);
+		if (err)
+			return err;
+	}
+
+	/* When C-TCAM or deltas are used, the eRP table must be used */
+	if (erp_table->ops != &erp_multiple_masks_ops)
+		erp_table->ops = &erp_multiple_masks_ops;
+
+	(*inc_num)++;
+
+	return 0;
+}
+
+static int mlxsw_sp_acl_erp_ctcam_inc(struct mlxsw_sp_acl_erp_table *erp_table)
+{
+	return __mlxsw_sp_acl_erp_table_other_inc(erp_table,
+						  &erp_table->num_ctcam_erps);
+}
+
+static int mlxsw_sp_acl_erp_delta_inc(struct mlxsw_sp_acl_erp_table *erp_table)
+{
+	return __mlxsw_sp_acl_erp_table_other_inc(erp_table,
+						  &erp_table->num_deltas);
+}
+
+static void
+__mlxsw_sp_acl_erp_table_other_dec(struct mlxsw_sp_acl_erp_table *erp_table,
+				   unsigned int *dec_num)
+{
+	(*dec_num)--;
+
+	/* If there are no C-TCAM eRP or deltas in use, the state we
+	 * transition to depends on the number of A-TCAM eRPs currently
+	 * in use.
+	 */
+	if (erp_table->num_ctcam_erps > 0 || erp_table->num_deltas > 0)
+		return;
+
+	switch (erp_table->num_atcam_erps) {
+	case 2:
+		/* Keep using the eRP table, but correctly set the
+		 * operations pointer so that when an A-TCAM eRP is
+		 * deleted we will transition to use the master mask
+		 */
+		erp_table->ops = &erp_two_masks_ops;
+		break;
+	case 1:
+		/* We only kept the eRP table because we had C-TCAM
+		 * eRPs in use. Now that the last C-TCAM eRP is gone we
+		 * can stop using the table and transition to use the
+		 * master mask
+		 */
+		mlxsw_sp_acl_erp_region_master_mask_trans(erp_table);
+		erp_table->ops = &erp_single_mask_ops;
+		break;
+	case 0:
+		/* There are no more eRPs of any kind used by the region
+		 * so free its eRP table and transition to initial state
+		 */
+		mlxsw_sp_acl_erp_table_disable(erp_table);
+		mlxsw_sp_acl_erp_table_free(erp_table->erp_core,
+					    erp_table->num_max_atcam_erps,
+					    erp_table->aregion->type,
+					    erp_table->base_index);
+		erp_table->ops = &erp_no_mask_ops;
+		break;
+	default:
+		break;
+	}
+}
+
+static void mlxsw_sp_acl_erp_ctcam_dec(struct mlxsw_sp_acl_erp_table *erp_table)
+{
+	__mlxsw_sp_acl_erp_table_other_dec(erp_table,
+					   &erp_table->num_ctcam_erps);
+}
+
+static void mlxsw_sp_acl_erp_delta_dec(struct mlxsw_sp_acl_erp_table *erp_table)
+{
+	__mlxsw_sp_acl_erp_table_other_dec(erp_table,
+					   &erp_table->num_deltas);
+}
+
+static struct mlxsw_sp_acl_erp *
+mlxsw_sp_acl_erp_ctcam_mask_create(struct mlxsw_sp_acl_erp_table *erp_table,
+				   struct mlxsw_sp_acl_erp_key *key)
+{
+	struct mlxsw_sp_acl_erp *erp;
+	int err;
+
+	erp = kzalloc(sizeof(*erp), GFP_KERNEL);
+	if (!erp)
+		return ERR_PTR(-ENOMEM);
+
+	memcpy(&erp->key, key, sizeof(*key));
+	bitmap_from_arr32(erp->mask_bitmap, (u32 *) key->mask,
+			  MLXSW_SP_ACL_TCAM_MASK_LEN);
+
+	err = mlxsw_sp_acl_erp_ctcam_inc(erp_table);
+	if (err)
+		goto err_erp_ctcam_inc;
+
+	erp->erp_table = erp_table;
+
+	err = mlxsw_sp_acl_erp_master_mask_set(erp_table, &erp->key);
+	if (err)
+		goto err_master_mask_set;
+
+	err = mlxsw_sp_acl_erp_region_ctcam_enable(erp_table);
+	if (err)
+		goto err_erp_region_ctcam_enable;
+
+	return erp;
+
+err_erp_region_ctcam_enable:
+	mlxsw_sp_acl_erp_master_mask_clear(erp_table, &erp->key);
+err_master_mask_set:
+	mlxsw_sp_acl_erp_ctcam_dec(erp_table);
+err_erp_ctcam_inc:
+	kfree(erp);
+	return ERR_PTR(err);
+}
+
+static void
+mlxsw_sp_acl_erp_ctcam_mask_destroy(struct mlxsw_sp_acl_erp *erp)
+{
+	struct mlxsw_sp_acl_erp_table *erp_table = erp->erp_table;
+
+	mlxsw_sp_acl_erp_region_ctcam_disable(erp_table);
+	mlxsw_sp_acl_erp_master_mask_clear(erp_table, &erp->key);
+	mlxsw_sp_acl_erp_ctcam_dec(erp_table);
+	kfree(erp);
+}
+
+static struct mlxsw_sp_acl_erp *
+mlxsw_sp_acl_erp_mask_create(struct mlxsw_sp_acl_erp_table *erp_table,
+			     struct mlxsw_sp_acl_erp_key *key)
+{
+	struct mlxsw_sp_acl_erp *erp;
+	int err;
+
+	if (key->ctcam)
+		return mlxsw_sp_acl_erp_ctcam_mask_create(erp_table, key);
+
+	/* Expand the eRP table for the new eRP, if needed */
+	err = mlxsw_sp_acl_erp_table_expand(erp_table);
+	if (err)
+		return ERR_PTR(err);
+
+	erp = mlxsw_sp_acl_erp_generic_create(erp_table, key);
+	if (IS_ERR(erp))
+		return erp;
+
+	err = mlxsw_sp_acl_erp_index_get(erp_table, &erp->index);
+	if (err)
+		goto err_erp_index_get;
+
+	err = mlxsw_sp_acl_erp_table_erp_add(erp_table, erp);
+	if (err)
+		goto err_table_erp_add;
+
+	err = mlxsw_sp_acl_erp_region_erp_add(erp_table, erp);
+	if (err)
+		goto err_region_erp_add;
+
+	erp_table->ops = &erp_multiple_masks_ops;
+
+	return erp;
+
+err_region_erp_add:
+	mlxsw_sp_acl_erp_table_erp_del(erp);
+err_table_erp_add:
+	mlxsw_sp_acl_erp_index_put(erp_table, erp->index);
+err_erp_index_get:
+	mlxsw_sp_acl_erp_generic_destroy(erp);
+	return ERR_PTR(err);
+}
+
+static void
+mlxsw_sp_acl_erp_mask_destroy(struct mlxsw_sp_acl_erp_table *erp_table,
+			      struct mlxsw_sp_acl_erp *erp)
+{
+	if (erp->key.ctcam)
+		return mlxsw_sp_acl_erp_ctcam_mask_destroy(erp);
+
+	mlxsw_sp_acl_erp_region_erp_del(erp);
+	mlxsw_sp_acl_erp_table_erp_del(erp);
+	mlxsw_sp_acl_erp_index_put(erp_table, erp->index);
+	mlxsw_sp_acl_erp_generic_destroy(erp);
+
+	if (erp_table->num_atcam_erps == 2 && erp_table->num_ctcam_erps == 0 &&
+	    erp_table->num_deltas == 0)
+		erp_table->ops = &erp_two_masks_ops;
+}
+
+static struct mlxsw_sp_acl_erp *
+mlxsw_sp_acl_erp_second_mask_create(struct mlxsw_sp_acl_erp_table *erp_table,
+				    struct mlxsw_sp_acl_erp_key *key)
+{
+	struct mlxsw_sp_acl_erp *erp;
+	int err;
+
+	if (key->ctcam)
+		return mlxsw_sp_acl_erp_ctcam_mask_create(erp_table, key);
+
+	/* Transition to use eRP table instead of master mask */
+	err = mlxsw_sp_acl_erp_region_table_trans(erp_table);
+	if (err)
+		return ERR_PTR(err);
+
+	erp = mlxsw_sp_acl_erp_generic_create(erp_table, key);
+	if (IS_ERR(erp)) {
+		err = PTR_ERR(erp);
+		goto err_erp_create;
+	}
+
+	err = mlxsw_sp_acl_erp_index_get(erp_table, &erp->index);
+	if (err)
+		goto err_erp_index_get;
+
+	err = mlxsw_sp_acl_erp_table_erp_add(erp_table, erp);
+	if (err)
+		goto err_table_erp_add;
+
+	err = mlxsw_sp_acl_erp_region_erp_add(erp_table, erp);
+	if (err)
+		goto err_region_erp_add;
+
+	erp_table->ops = &erp_two_masks_ops;
+
+	return erp;
+
+err_region_erp_add:
+	mlxsw_sp_acl_erp_table_erp_del(erp);
+err_table_erp_add:
+	mlxsw_sp_acl_erp_index_put(erp_table, erp->index);
+err_erp_index_get:
+	mlxsw_sp_acl_erp_generic_destroy(erp);
+err_erp_create:
+	mlxsw_sp_acl_erp_region_master_mask_trans(erp_table);
+	return ERR_PTR(err);
+}
+
+static void
+mlxsw_sp_acl_erp_second_mask_destroy(struct mlxsw_sp_acl_erp_table *erp_table,
+				     struct mlxsw_sp_acl_erp *erp)
+{
+	if (erp->key.ctcam)
+		return mlxsw_sp_acl_erp_ctcam_mask_destroy(erp);
+
+	mlxsw_sp_acl_erp_region_erp_del(erp);
+	mlxsw_sp_acl_erp_table_erp_del(erp);
+	mlxsw_sp_acl_erp_index_put(erp_table, erp->index);
+	mlxsw_sp_acl_erp_generic_destroy(erp);
+	/* Transition to use master mask instead of eRP table */
+	mlxsw_sp_acl_erp_region_master_mask_trans(erp_table);
+
+	erp_table->ops = &erp_single_mask_ops;
+}
+
+static struct mlxsw_sp_acl_erp *
+mlxsw_sp_acl_erp_first_mask_create(struct mlxsw_sp_acl_erp_table *erp_table,
+				   struct mlxsw_sp_acl_erp_key *key)
+{
+	struct mlxsw_sp_acl_erp *erp;
+
+	if (key->ctcam)
+		return ERR_PTR(-EINVAL);
+
+	erp = mlxsw_sp_acl_erp_generic_create(erp_table, key);
+	if (IS_ERR(erp))
+		return erp;
+
+	erp_table->ops = &erp_single_mask_ops;
+
+	return erp;
+}
+
+static void
+mlxsw_sp_acl_erp_first_mask_destroy(struct mlxsw_sp_acl_erp_table *erp_table,
+				    struct mlxsw_sp_acl_erp *erp)
+{
+	mlxsw_sp_acl_erp_generic_destroy(erp);
+	erp_table->ops = &erp_no_mask_ops;
+}
+
+static void
+mlxsw_sp_acl_erp_no_mask_destroy(struct mlxsw_sp_acl_erp_table *erp_table,
+				 struct mlxsw_sp_acl_erp *erp)
+{
+	WARN_ON(1);
+}
+
+struct mlxsw_sp_acl_erp_mask *
+mlxsw_sp_acl_erp_mask_get(struct mlxsw_sp_acl_atcam_region *aregion,
+			  const char *mask, bool ctcam)
+{
+	struct mlxsw_sp_acl_erp_table *erp_table = aregion->erp_table;
+	struct mlxsw_sp_acl_erp_key key;
+	struct objagg_obj *objagg_obj;
+
+	memcpy(key.mask, mask, MLXSW_REG_PTCEX_FLEX_KEY_BLOCKS_LEN);
+	key.ctcam = ctcam;
+	mutex_lock(&erp_table->objagg_lock);
+	objagg_obj = objagg_obj_get(erp_table->objagg, &key);
+	mutex_unlock(&erp_table->objagg_lock);
+	if (IS_ERR(objagg_obj))
+		return ERR_CAST(objagg_obj);
+	return (struct mlxsw_sp_acl_erp_mask *) objagg_obj;
+}
+
+void mlxsw_sp_acl_erp_mask_put(struct mlxsw_sp_acl_atcam_region *aregion,
+			       struct mlxsw_sp_acl_erp_mask *erp_mask)
+{
+	struct objagg_obj *objagg_obj = (struct objagg_obj *) erp_mask;
+	struct mlxsw_sp_acl_erp_table *erp_table = aregion->erp_table;
+
+	mutex_lock(&erp_table->objagg_lock);
+	objagg_obj_put(erp_table->objagg, objagg_obj);
+	mutex_unlock(&erp_table->objagg_lock);
+}
+
+int mlxsw_sp_acl_erp_bf_insert(struct mlxsw_sp *mlxsw_sp,
+			       struct mlxsw_sp_acl_atcam_region *aregion,
+			       struct mlxsw_sp_acl_erp_mask *erp_mask,
+			       struct mlxsw_sp_acl_atcam_entry *aentry)
+{
+	struct objagg_obj *objagg_obj = (struct objagg_obj *) erp_mask;
+	const struct mlxsw_sp_acl_erp *erp = objagg_obj_root_priv(objagg_obj);
+	unsigned int erp_bank;
+
+	if (!mlxsw_sp_acl_erp_table_is_used(erp->erp_table))
+		return 0;
+
+	erp_bank = mlxsw_sp_acl_erp_bank_get(erp);
+	return mlxsw_sp_acl_bf_entry_add(mlxsw_sp,
+					erp->erp_table->erp_core->bf,
+					aregion, erp_bank, aentry);
+}
+
+void mlxsw_sp_acl_erp_bf_remove(struct mlxsw_sp *mlxsw_sp,
+				struct mlxsw_sp_acl_atcam_region *aregion,
+				struct mlxsw_sp_acl_erp_mask *erp_mask,
+				struct mlxsw_sp_acl_atcam_entry *aentry)
+{
+	struct objagg_obj *objagg_obj = (struct objagg_obj *) erp_mask;
+	const struct mlxsw_sp_acl_erp *erp = objagg_obj_root_priv(objagg_obj);
+	unsigned int erp_bank;
+
+	if (!mlxsw_sp_acl_erp_table_is_used(erp->erp_table))
+		return;
+
+	erp_bank = mlxsw_sp_acl_erp_bank_get(erp);
+	mlxsw_sp_acl_bf_entry_del(mlxsw_sp,
+				  erp->erp_table->erp_core->bf,
+				  aregion, erp_bank, aentry);
+}
+
+bool
+mlxsw_sp_acl_erp_mask_is_ctcam(const struct mlxsw_sp_acl_erp_mask *erp_mask)
+{
+	struct objagg_obj *objagg_obj = (struct objagg_obj *) erp_mask;
+	const struct mlxsw_sp_acl_erp_key *key = objagg_obj_raw(objagg_obj);
+
+	return key->ctcam;
+}
+
+u8 mlxsw_sp_acl_erp_mask_erp_id(const struct mlxsw_sp_acl_erp_mask *erp_mask)
+{
+	struct objagg_obj *objagg_obj = (struct objagg_obj *) erp_mask;
+	const struct mlxsw_sp_acl_erp *erp = objagg_obj_root_priv(objagg_obj);
+
+	return erp->id;
+}
+
+struct mlxsw_sp_acl_erp_delta {
+	struct mlxsw_sp_acl_erp_key key;
+	u16 start;
+	u8 mask;
+};
+
+u16 mlxsw_sp_acl_erp_delta_start(const struct mlxsw_sp_acl_erp_delta *delta)
+{
+	return delta->start;
+}
+
+u8 mlxsw_sp_acl_erp_delta_mask(const struct mlxsw_sp_acl_erp_delta *delta)
+{
+	return delta->mask;
+}
+
+u8 mlxsw_sp_acl_erp_delta_value(const struct mlxsw_sp_acl_erp_delta *delta,
+				const char *enc_key)
+{
+	u16 start = delta->start;
+	u8 mask = delta->mask;
+	u16 tmp;
+
+	if (!mask)
+		return 0;
+
+	tmp = (unsigned char) enc_key[__MASK_IDX(start / 8)];
+	if (start / 8 + 1 < __MASK_LEN)
+		tmp |= (unsigned char) enc_key[__MASK_IDX(start / 8 + 1)] << 8;
+	tmp >>= start % 8;
+	tmp &= mask;
+	return tmp;
+}
+
+void mlxsw_sp_acl_erp_delta_clear(const struct mlxsw_sp_acl_erp_delta *delta,
+				  const char *enc_key)
+{
+	u16 start = delta->start;
+	u8 mask = delta->mask;
+	unsigned char *byte;
+	u16 tmp;
+
+	tmp = mask;
+	tmp <<= start % 8;
+	tmp = ~tmp;
+
+	byte = (unsigned char *) &enc_key[__MASK_IDX(start / 8)];
+	*byte &= tmp & 0xff;
+	if (start / 8 + 1 < __MASK_LEN) {
+		byte = (unsigned char *) &enc_key[__MASK_IDX(start / 8 + 1)];
+		*byte &= (tmp >> 8) & 0xff;
+	}
+}
+
+static const struct mlxsw_sp_acl_erp_delta
+mlxsw_sp_acl_erp_delta_default = {};
+
+const struct mlxsw_sp_acl_erp_delta *
+mlxsw_sp_acl_erp_delta(const struct mlxsw_sp_acl_erp_mask *erp_mask)
+{
+	struct objagg_obj *objagg_obj = (struct objagg_obj *) erp_mask;
+	const struct mlxsw_sp_acl_erp_delta *delta;
+
+	delta = objagg_obj_delta_priv(objagg_obj);
+	if (!delta)
+		delta = &mlxsw_sp_acl_erp_delta_default;
+	return delta;
+}
+
+static int
+mlxsw_sp_acl_erp_delta_fill(const struct mlxsw_sp_acl_erp_key *parent_key,
+			    const struct mlxsw_sp_acl_erp_key *key,
+			    u16 *delta_start, u8 *delta_mask)
+{
+	int offset = 0;
+	int si = -1;
+	u16 pmask;
+	u16 mask;
+	int i;
+
+	/* The difference between 2 masks can be up to 8 consecutive bits. */
+	for (i = 0; i < __MASK_LEN; i++) {
+		if (parent_key->mask[__MASK_IDX(i)] == key->mask[__MASK_IDX(i)])
+			continue;
+		if (si == -1)
+			si = i;
+		else if (si != i - 1)
+			return -EINVAL;
+	}
+	if (si == -1) {
+		/* The masks are the same, this can happen in case eRPs with
+		 * the same mask were created in both A-TCAM and C-TCAM.
+		 * The only possible condition under which this can happen
+		 * is identical rule insertion. Delta is not possible here.
+		 */
+		return -EINVAL;
+	}
+	pmask = (unsigned char) parent_key->mask[__MASK_IDX(si)];
+	mask = (unsigned char) key->mask[__MASK_IDX(si)];
+	if (si + 1 < __MASK_LEN) {
+		pmask |= (unsigned char) parent_key->mask[__MASK_IDX(si + 1)] << 8;
+		mask |= (unsigned char) key->mask[__MASK_IDX(si + 1)] << 8;
+	}
+
+	if ((pmask ^ mask) & pmask)
+		return -EINVAL;
+	mask &= ~pmask;
+	while (!(mask & (1 << offset)))
+		offset++;
+	while (!(mask & 1))
+		mask >>= 1;
+	if (mask & 0xff00)
+		return -EINVAL;
+
+	*delta_start = si * 8 + offset;
+	*delta_mask = mask;
+
+	return 0;
+}
+
+static bool mlxsw_sp_acl_erp_delta_check(void *priv, const void *parent_obj,
+					 const void *obj)
+{
+	const struct mlxsw_sp_acl_erp_key *parent_key = parent_obj;
+	const struct mlxsw_sp_acl_erp_key *key = obj;
+	u16 delta_start;
+	u8 delta_mask;
+	int err;
+
+	err = mlxsw_sp_acl_erp_delta_fill(parent_key, key,
+					  &delta_start, &delta_mask);
+	return err ? false : true;
+}
+
+static int mlxsw_sp_acl_erp_hints_obj_cmp(const void *obj1, const void *obj2)
+{
+	const struct mlxsw_sp_acl_erp_key *key1 = obj1;
+	const struct mlxsw_sp_acl_erp_key *key2 = obj2;
+
+	/* For hints purposes, two objects are considered equal
+	 * in case the masks are the same. Does not matter what
+	 * the "ctcam" value is.
+	 */
+	return memcmp(key1->mask, key2->mask, sizeof(key1->mask));
+}
+
+static void *mlxsw_sp_acl_erp_delta_create(void *priv, void *parent_obj,
+					   void *obj)
+{
+	struct mlxsw_sp_acl_erp_key *parent_key = parent_obj;
+	struct mlxsw_sp_acl_atcam_region *aregion = priv;
+	struct mlxsw_sp_acl_erp_table *erp_table = aregion->erp_table;
+	struct mlxsw_sp_acl_erp_key *key = obj;
+	struct mlxsw_sp_acl_erp_delta *delta;
+	u16 delta_start;
+	u8 delta_mask;
+	int err;
+
+	if (parent_key->ctcam || key->ctcam)
+		return ERR_PTR(-EINVAL);
+	err = mlxsw_sp_acl_erp_delta_fill(parent_key, key,
+					  &delta_start, &delta_mask);
+	if (err)
+		return ERR_PTR(-EINVAL);
+
+	delta = kzalloc(sizeof(*delta), GFP_KERNEL);
+	if (!delta)
+		return ERR_PTR(-ENOMEM);
+	delta->start = delta_start;
+	delta->mask = delta_mask;
+
+	err = mlxsw_sp_acl_erp_delta_inc(erp_table);
+	if (err)
+		goto err_erp_delta_inc;
+
+	memcpy(&delta->key, key, sizeof(*key));
+	err = mlxsw_sp_acl_erp_master_mask_set(erp_table, &delta->key);
+	if (err)
+		goto err_master_mask_set;
+
+	return delta;
+
+err_master_mask_set:
+	mlxsw_sp_acl_erp_delta_dec(erp_table);
+err_erp_delta_inc:
+	kfree(delta);
+	return ERR_PTR(err);
+}
+
+static void mlxsw_sp_acl_erp_delta_destroy(void *priv, void *delta_priv)
+{
+	struct mlxsw_sp_acl_erp_delta *delta = delta_priv;
+	struct mlxsw_sp_acl_atcam_region *aregion = priv;
+	struct mlxsw_sp_acl_erp_table *erp_table = aregion->erp_table;
+
+	mlxsw_sp_acl_erp_master_mask_clear(erp_table, &delta->key);
+	mlxsw_sp_acl_erp_delta_dec(erp_table);
+	kfree(delta);
+}
+
+static void *mlxsw_sp_acl_erp_root_create(void *priv, void *obj,
+					  unsigned int root_id)
+{
+	struct mlxsw_sp_acl_atcam_region *aregion = priv;
+	struct mlxsw_sp_acl_erp_table *erp_table = aregion->erp_table;
+	struct mlxsw_sp_acl_erp_key *key = obj;
+
+	if (!key->ctcam &&
+	    root_id != OBJAGG_OBJ_ROOT_ID_INVALID &&
+	    root_id >= MLXSW_SP_ACL_ERP_MAX_PER_REGION)
+		return ERR_PTR(-ENOBUFS);
+	return erp_table->ops->erp_create(erp_table, key);
+}
+
+static void mlxsw_sp_acl_erp_root_destroy(void *priv, void *root_priv)
+{
+	struct mlxsw_sp_acl_atcam_region *aregion = priv;
+	struct mlxsw_sp_acl_erp_table *erp_table = aregion->erp_table;
+
+	erp_table->ops->erp_destroy(erp_table, root_priv);
+}
+
+static const struct objagg_ops mlxsw_sp_acl_erp_objagg_ops = {
+	.obj_size = sizeof(struct mlxsw_sp_acl_erp_key),
+	.delta_check = mlxsw_sp_acl_erp_delta_check,
+	.hints_obj_cmp = mlxsw_sp_acl_erp_hints_obj_cmp,
+	.delta_create = mlxsw_sp_acl_erp_delta_create,
+	.delta_destroy = mlxsw_sp_acl_erp_delta_destroy,
+	.root_create = mlxsw_sp_acl_erp_root_create,
+	.root_destroy = mlxsw_sp_acl_erp_root_destroy,
+};
+
+static struct mlxsw_sp_acl_erp_table *
+mlxsw_sp_acl_erp_table_create(struct mlxsw_sp_acl_atcam_region *aregion,
+			      struct objagg_hints *hints)
+{
+	struct mlxsw_sp_acl_erp_table *erp_table;
+	int err;
+
+	erp_table = kzalloc(sizeof(*erp_table), GFP_KERNEL);
+	if (!erp_table)
+		return ERR_PTR(-ENOMEM);
+
+	erp_table->objagg = objagg_create(&mlxsw_sp_acl_erp_objagg_ops,
+					  hints, aregion);
+	if (IS_ERR(erp_table->objagg)) {
+		err = PTR_ERR(erp_table->objagg);
+		goto err_objagg_create;
+	}
+
+	erp_table->erp_core = aregion->atcam->erp_core;
+	erp_table->ops = &erp_no_mask_ops;
+	INIT_LIST_HEAD(&erp_table->atcam_erps_list);
+	erp_table->aregion = aregion;
+	mutex_init(&erp_table->objagg_lock);
+
+	return erp_table;
+
+err_objagg_create:
+	kfree(erp_table);
+	return ERR_PTR(err);
+}
+
+static void
+mlxsw_sp_acl_erp_table_destroy(struct mlxsw_sp_acl_erp_table *erp_table)
+{
+	WARN_ON(!list_empty(&erp_table->atcam_erps_list));
+	mutex_destroy(&erp_table->objagg_lock);
+	objagg_destroy(erp_table->objagg);
+	kfree(erp_table);
+}
+
+static int
+mlxsw_sp_acl_erp_master_mask_init(struct mlxsw_sp_acl_atcam_region *aregion)
+{
+	struct mlxsw_sp *mlxsw_sp = aregion->region->mlxsw_sp;
+	char percr_pl[MLXSW_REG_PERCR_LEN];
+
+	mlxsw_reg_percr_pack(percr_pl, aregion->region->id);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(percr), percr_pl);
+}
+
+static int
+mlxsw_sp_acl_erp_region_param_init(struct mlxsw_sp_acl_atcam_region *aregion)
+{
+	struct mlxsw_sp *mlxsw_sp = aregion->region->mlxsw_sp;
+	char pererp_pl[MLXSW_REG_PERERP_LEN];
+
+	mlxsw_reg_pererp_pack(pererp_pl, aregion->region->id, false, false, 0,
+			      0, 0);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(pererp), pererp_pl);
+}
+
+static int
+mlxsw_sp_acl_erp_hints_check(struct mlxsw_sp *mlxsw_sp,
+			     struct mlxsw_sp_acl_atcam_region *aregion,
+			     struct objagg_hints *hints, bool *p_rehash_needed)
+{
+	struct mlxsw_sp_acl_erp_table *erp_table = aregion->erp_table;
+	const struct objagg_stats *ostats;
+	const struct objagg_stats *hstats;
+	int err;
+
+	*p_rehash_needed = false;
+
+	mutex_lock(&erp_table->objagg_lock);
+	ostats = objagg_stats_get(erp_table->objagg);
+	mutex_unlock(&erp_table->objagg_lock);
+	if (IS_ERR(ostats)) {
+		dev_err_ratelimited(mlxsw_sp->bus_info->dev, "Failed to get ERP stats\n");
+		return PTR_ERR(ostats);
+	}
+
+	hstats = objagg_hints_stats_get(hints);
+	if (IS_ERR(hstats)) {
+		dev_err_ratelimited(mlxsw_sp->bus_info->dev, "Failed to get ERP hints stats\n");
+		err = PTR_ERR(hstats);
+		goto err_hints_stats_get;
+	}
+
+	/* Very basic criterion for now. */
+	if (hstats->root_count < ostats->root_count)
+		*p_rehash_needed = true;
+
+	err = 0;
+
+	objagg_stats_put(hstats);
+err_hints_stats_get:
+	objagg_stats_put(ostats);
+	return err;
+}
+
+void *
+mlxsw_sp_acl_erp_rehash_hints_get(struct mlxsw_sp_acl_atcam_region *aregion)
+{
+	struct mlxsw_sp_acl_erp_table *erp_table = aregion->erp_table;
+	struct mlxsw_sp *mlxsw_sp = aregion->region->mlxsw_sp;
+	struct objagg_hints *hints;
+	bool rehash_needed;
+	int err;
+
+	mutex_lock(&erp_table->objagg_lock);
+	hints = objagg_hints_get(erp_table->objagg,
+				 OBJAGG_OPT_ALGO_SIMPLE_GREEDY);
+	mutex_unlock(&erp_table->objagg_lock);
+	if (IS_ERR(hints)) {
+		dev_err_ratelimited(mlxsw_sp->bus_info->dev, "Failed to create ERP hints\n");
+		return ERR_CAST(hints);
+	}
+	err = mlxsw_sp_acl_erp_hints_check(mlxsw_sp, aregion, hints,
+					   &rehash_needed);
+	if (err)
+		goto errout;
+
+	if (!rehash_needed) {
+		err = -EAGAIN;
+		goto errout;
+	}
+	return hints;
+
+errout:
+	objagg_hints_put(hints);
+	return ERR_PTR(err);
+}
+
+void mlxsw_sp_acl_erp_rehash_hints_put(void *hints_priv)
+{
+	struct objagg_hints *hints = hints_priv;
+
+	objagg_hints_put(hints);
+}
+
+int mlxsw_sp_acl_erp_region_init(struct mlxsw_sp_acl_atcam_region *aregion,
+				 void *hints_priv)
+{
+	struct mlxsw_sp_acl_erp_table *erp_table;
+	struct objagg_hints *hints = hints_priv;
+	int err;
+
+	erp_table = mlxsw_sp_acl_erp_table_create(aregion, hints);
+	if (IS_ERR(erp_table))
+		return PTR_ERR(erp_table);
+	aregion->erp_table = erp_table;
+
+	/* Initialize the region's master mask to all zeroes */
+	err = mlxsw_sp_acl_erp_master_mask_init(aregion);
+	if (err)
+		goto err_erp_master_mask_init;
+
+	/* Initialize the region to not use the eRP table */
+	err = mlxsw_sp_acl_erp_region_param_init(aregion);
+	if (err)
+		goto err_erp_region_param_init;
+
+	return 0;
+
+err_erp_region_param_init:
+err_erp_master_mask_init:
+	mlxsw_sp_acl_erp_table_destroy(erp_table);
+	return err;
+}
+
+void mlxsw_sp_acl_erp_region_fini(struct mlxsw_sp_acl_atcam_region *aregion)
+{
+	mlxsw_sp_acl_erp_table_destroy(aregion->erp_table);
+}
+
+static int
+mlxsw_sp_acl_erp_tables_sizes_query(struct mlxsw_sp *mlxsw_sp,
+				    struct mlxsw_sp_acl_erp_core *erp_core)
+{
+	unsigned int size;
+
+	if (!MLXSW_CORE_RES_VALID(mlxsw_sp->core, ACL_ERPT_ENTRIES_2KB) ||
+	    !MLXSW_CORE_RES_VALID(mlxsw_sp->core, ACL_ERPT_ENTRIES_4KB) ||
+	    !MLXSW_CORE_RES_VALID(mlxsw_sp->core, ACL_ERPT_ENTRIES_8KB) ||
+	    !MLXSW_CORE_RES_VALID(mlxsw_sp->core, ACL_ERPT_ENTRIES_12KB))
+		return -EIO;
+
+	size = MLXSW_CORE_RES_GET(mlxsw_sp->core, ACL_ERPT_ENTRIES_2KB);
+	erp_core->erpt_entries_size[MLXSW_SP_ACL_ATCAM_REGION_TYPE_2KB] = size;
+
+	size = MLXSW_CORE_RES_GET(mlxsw_sp->core, ACL_ERPT_ENTRIES_4KB);
+	erp_core->erpt_entries_size[MLXSW_SP_ACL_ATCAM_REGION_TYPE_4KB] = size;
+
+	size = MLXSW_CORE_RES_GET(mlxsw_sp->core, ACL_ERPT_ENTRIES_8KB);
+	erp_core->erpt_entries_size[MLXSW_SP_ACL_ATCAM_REGION_TYPE_8KB] = size;
+
+	size = MLXSW_CORE_RES_GET(mlxsw_sp->core, ACL_ERPT_ENTRIES_12KB);
+	erp_core->erpt_entries_size[MLXSW_SP_ACL_ATCAM_REGION_TYPE_12KB] = size;
+
+	return 0;
+}
+
+static int mlxsw_sp_acl_erp_tables_init(struct mlxsw_sp *mlxsw_sp,
+					struct mlxsw_sp_acl_erp_core *erp_core)
+{
+	unsigned int erpt_bank_size;
+	int err;
+
+	if (!MLXSW_CORE_RES_VALID(mlxsw_sp->core, ACL_MAX_ERPT_BANK_SIZE) ||
+	    !MLXSW_CORE_RES_VALID(mlxsw_sp->core, ACL_MAX_ERPT_BANKS))
+		return -EIO;
+	erpt_bank_size = MLXSW_CORE_RES_GET(mlxsw_sp->core,
+					    ACL_MAX_ERPT_BANK_SIZE);
+	erp_core->num_erp_banks = MLXSW_CORE_RES_GET(mlxsw_sp->core,
+						     ACL_MAX_ERPT_BANKS);
+
+	erp_core->erp_tables = gen_pool_create(0, -1);
+	if (!erp_core->erp_tables)
+		return -ENOMEM;
+	gen_pool_set_algo(erp_core->erp_tables, gen_pool_best_fit, NULL);
+
+	err = gen_pool_add(erp_core->erp_tables,
+			   MLXSW_SP_ACL_ERP_GENALLOC_OFFSET, erpt_bank_size,
+			   -1);
+	if (err)
+		goto err_gen_pool_add;
+
+	erp_core->bf = mlxsw_sp_acl_bf_init(mlxsw_sp, erp_core->num_erp_banks);
+	if (IS_ERR(erp_core->bf)) {
+		err = PTR_ERR(erp_core->bf);
+		goto err_bf_init;
+	}
+
+	/* Different regions require masks of different sizes */
+	err = mlxsw_sp_acl_erp_tables_sizes_query(mlxsw_sp, erp_core);
+	if (err)
+		goto err_erp_tables_sizes_query;
+
+	return 0;
+
+err_erp_tables_sizes_query:
+	mlxsw_sp_acl_bf_fini(erp_core->bf);
+err_bf_init:
+err_gen_pool_add:
+	gen_pool_destroy(erp_core->erp_tables);
+	return err;
+}
+
+static void mlxsw_sp_acl_erp_tables_fini(struct mlxsw_sp *mlxsw_sp,
+					 struct mlxsw_sp_acl_erp_core *erp_core)
+{
+	mlxsw_sp_acl_bf_fini(erp_core->bf);
+	gen_pool_destroy(erp_core->erp_tables);
+}
+
+int mlxsw_sp_acl_erps_init(struct mlxsw_sp *mlxsw_sp,
+			   struct mlxsw_sp_acl_atcam *atcam)
+{
+	struct mlxsw_sp_acl_erp_core *erp_core;
+	int err;
+
+	erp_core = kzalloc(sizeof(*erp_core), GFP_KERNEL);
+	if (!erp_core)
+		return -ENOMEM;
+	erp_core->mlxsw_sp = mlxsw_sp;
+	atcam->erp_core = erp_core;
+
+	err = mlxsw_sp_acl_erp_tables_init(mlxsw_sp, erp_core);
+	if (err)
+		goto err_erp_tables_init;
+
+	return 0;
+
+err_erp_tables_init:
+	kfree(erp_core);
+	return err;
+}
+
+void mlxsw_sp_acl_erps_fini(struct mlxsw_sp *mlxsw_sp,
+			    struct mlxsw_sp_acl_atcam *atcam)
+{
+	mlxsw_sp_acl_erp_tables_fini(mlxsw_sp, atcam->erp_core);
+	kfree(atcam->erp_core);
+}
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_acl_flex_actions.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_acl_flex_actions.c
new file mode 100644
index 000000000..50806594d
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_acl_flex_actions.c
@@ -0,0 +1,318 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2017-2018 Mellanox Technologies. All rights reserved */
+
+#include "spectrum_acl_flex_actions.h"
+#include "core_acl_flex_actions.h"
+#include "spectrum_span.h"
+
+static int mlxsw_sp_act_kvdl_set_add(void *priv, u32 *p_kvdl_index,
+				     char *enc_actions, bool is_first, bool ca)
+{
+	struct mlxsw_sp *mlxsw_sp = priv;
+	char pefa_pl[MLXSW_REG_PEFA_LEN];
+	u32 kvdl_index;
+	int err;
+
+	/* The first action set of a TCAM entry is stored directly in TCAM,
+	 * not KVD linear area.
+	 */
+	if (is_first)
+		return 0;
+
+	err = mlxsw_sp_kvdl_alloc(mlxsw_sp, MLXSW_SP_KVDL_ENTRY_TYPE_ACTSET,
+				  1, &kvdl_index);
+	if (err)
+		return err;
+	mlxsw_reg_pefa_pack(pefa_pl, kvdl_index, ca, enc_actions);
+	err = mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(pefa), pefa_pl);
+	if (err)
+		goto err_pefa_write;
+	*p_kvdl_index = kvdl_index;
+	return 0;
+
+err_pefa_write:
+	mlxsw_sp_kvdl_free(mlxsw_sp, MLXSW_SP_KVDL_ENTRY_TYPE_ACTSET,
+			   1, kvdl_index);
+	return err;
+}
+
+static int mlxsw_sp1_act_kvdl_set_add(void *priv, u32 *p_kvdl_index,
+				      char *enc_actions, bool is_first)
+{
+	return mlxsw_sp_act_kvdl_set_add(priv, p_kvdl_index, enc_actions,
+					 is_first, false);
+}
+
+static int mlxsw_sp2_act_kvdl_set_add(void *priv, u32 *p_kvdl_index,
+				      char *enc_actions, bool is_first)
+{
+	return mlxsw_sp_act_kvdl_set_add(priv, p_kvdl_index, enc_actions,
+					 is_first, true);
+}
+
+static void mlxsw_sp_act_kvdl_set_del(void *priv, u32 kvdl_index,
+				      bool is_first)
+{
+	struct mlxsw_sp *mlxsw_sp = priv;
+
+	if (is_first)
+		return;
+	mlxsw_sp_kvdl_free(mlxsw_sp, MLXSW_SP_KVDL_ENTRY_TYPE_ACTSET,
+			   1, kvdl_index);
+}
+
+static int mlxsw_sp1_act_kvdl_set_activity_get(void *priv, u32 kvdl_index,
+					       bool *activity)
+{
+	return -EOPNOTSUPP;
+}
+
+static int mlxsw_sp2_act_kvdl_set_activity_get(void *priv, u32 kvdl_index,
+					       bool *activity)
+{
+	struct mlxsw_sp *mlxsw_sp = priv;
+	char pefa_pl[MLXSW_REG_PEFA_LEN];
+	int err;
+
+	mlxsw_reg_pefa_pack(pefa_pl, kvdl_index, true, NULL);
+	err = mlxsw_reg_query(mlxsw_sp->core, MLXSW_REG(pefa), pefa_pl);
+	if (err)
+		return err;
+	mlxsw_reg_pefa_unpack(pefa_pl, activity);
+	return 0;
+}
+
+static int mlxsw_sp_act_kvdl_fwd_entry_add(void *priv, u32 *p_kvdl_index,
+					   u16 local_port)
+{
+	struct mlxsw_sp *mlxsw_sp = priv;
+	char ppbs_pl[MLXSW_REG_PPBS_LEN];
+	u32 kvdl_index;
+	int err;
+
+	err = mlxsw_sp_kvdl_alloc(mlxsw_sp, MLXSW_SP_KVDL_ENTRY_TYPE_PBS,
+				  1, &kvdl_index);
+	if (err)
+		return err;
+	mlxsw_reg_ppbs_pack(ppbs_pl, kvdl_index, local_port);
+	err = mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(ppbs), ppbs_pl);
+	if (err)
+		goto err_ppbs_write;
+	*p_kvdl_index = kvdl_index;
+	return 0;
+
+err_ppbs_write:
+	mlxsw_sp_kvdl_free(mlxsw_sp, MLXSW_SP_KVDL_ENTRY_TYPE_PBS,
+			   1, kvdl_index);
+	return err;
+}
+
+static void mlxsw_sp_act_kvdl_fwd_entry_del(void *priv, u32 kvdl_index)
+{
+	struct mlxsw_sp *mlxsw_sp = priv;
+
+	mlxsw_sp_kvdl_free(mlxsw_sp, MLXSW_SP_KVDL_ENTRY_TYPE_PBS,
+			   1, kvdl_index);
+}
+
+static int
+mlxsw_sp_act_counter_index_get(void *priv, unsigned int *p_counter_index)
+{
+	struct mlxsw_sp *mlxsw_sp = priv;
+
+	return mlxsw_sp_flow_counter_alloc(mlxsw_sp, p_counter_index);
+}
+
+static void
+mlxsw_sp_act_counter_index_put(void *priv, unsigned int counter_index)
+{
+	struct mlxsw_sp *mlxsw_sp = priv;
+
+	mlxsw_sp_flow_counter_free(mlxsw_sp, counter_index);
+}
+
+static int
+mlxsw_sp_act_mirror_add(void *priv, u16 local_in_port,
+			const struct net_device *out_dev,
+			bool ingress, int *p_span_id)
+{
+	struct mlxsw_sp_span_agent_parms agent_parms = {};
+	struct mlxsw_sp_port *mlxsw_sp_port;
+	struct mlxsw_sp *mlxsw_sp = priv;
+	int err;
+
+	agent_parms.to_dev = out_dev;
+	err = mlxsw_sp_span_agent_get(mlxsw_sp, p_span_id, &agent_parms);
+	if (err)
+		return err;
+
+	mlxsw_sp_port = mlxsw_sp->ports[local_in_port];
+	err = mlxsw_sp_span_analyzed_port_get(mlxsw_sp_port, ingress);
+	if (err)
+		goto err_analyzed_port_get;
+
+	return 0;
+
+err_analyzed_port_get:
+	mlxsw_sp_span_agent_put(mlxsw_sp, *p_span_id);
+	return err;
+}
+
+static void
+mlxsw_sp_act_mirror_del(void *priv, u16 local_in_port, int span_id, bool ingress)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port;
+	struct mlxsw_sp *mlxsw_sp = priv;
+
+	mlxsw_sp_port = mlxsw_sp->ports[local_in_port];
+	mlxsw_sp_span_analyzed_port_put(mlxsw_sp_port, ingress);
+	mlxsw_sp_span_agent_put(mlxsw_sp, span_id);
+}
+
+static int mlxsw_sp_act_policer_add(void *priv, u64 rate_bytes_ps, u32 burst,
+				    u16 *p_policer_index,
+				    struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp_policer_params params;
+	struct mlxsw_sp *mlxsw_sp = priv;
+
+	params.rate = rate_bytes_ps;
+	params.burst = burst;
+	params.bytes = true;
+	return mlxsw_sp_policer_add(mlxsw_sp,
+				    MLXSW_SP_POLICER_TYPE_SINGLE_RATE,
+				    &params, extack, p_policer_index);
+}
+
+static void mlxsw_sp_act_policer_del(void *priv, u16 policer_index)
+{
+	struct mlxsw_sp *mlxsw_sp = priv;
+
+	mlxsw_sp_policer_del(mlxsw_sp, MLXSW_SP_POLICER_TYPE_SINGLE_RATE,
+			     policer_index);
+}
+
+static int mlxsw_sp1_act_sampler_add(void *priv, u16 local_port,
+				     struct psample_group *psample_group,
+				     u32 rate, u32 trunc_size, bool truncate,
+				     bool ingress, int *p_span_id,
+				     struct netlink_ext_ack *extack)
+{
+	NL_SET_ERR_MSG_MOD(extack, "Sampling action is not supported on Spectrum-1");
+	return -EOPNOTSUPP;
+}
+
+static void mlxsw_sp1_act_sampler_del(void *priv, u16 local_port, int span_id,
+				      bool ingress)
+{
+	WARN_ON_ONCE(1);
+}
+
+const struct mlxsw_afa_ops mlxsw_sp1_act_afa_ops = {
+	.kvdl_set_add		= mlxsw_sp1_act_kvdl_set_add,
+	.kvdl_set_del		= mlxsw_sp_act_kvdl_set_del,
+	.kvdl_set_activity_get	= mlxsw_sp1_act_kvdl_set_activity_get,
+	.kvdl_fwd_entry_add	= mlxsw_sp_act_kvdl_fwd_entry_add,
+	.kvdl_fwd_entry_del	= mlxsw_sp_act_kvdl_fwd_entry_del,
+	.counter_index_get	= mlxsw_sp_act_counter_index_get,
+	.counter_index_put	= mlxsw_sp_act_counter_index_put,
+	.mirror_add		= mlxsw_sp_act_mirror_add,
+	.mirror_del		= mlxsw_sp_act_mirror_del,
+	.policer_add		= mlxsw_sp_act_policer_add,
+	.policer_del		= mlxsw_sp_act_policer_del,
+	.sampler_add		= mlxsw_sp1_act_sampler_add,
+	.sampler_del		= mlxsw_sp1_act_sampler_del,
+};
+
+static int mlxsw_sp2_act_sampler_add(void *priv, u16 local_port,
+				     struct psample_group *psample_group,
+				     u32 rate, u32 trunc_size, bool truncate,
+				     bool ingress, int *p_span_id,
+				     struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp_span_agent_parms agent_parms = {
+		.session_id = MLXSW_SP_SPAN_SESSION_ID_SAMPLING,
+	};
+	struct mlxsw_sp_sample_trigger trigger = {
+		.type = MLXSW_SP_SAMPLE_TRIGGER_TYPE_POLICY_ENGINE,
+	};
+	struct mlxsw_sp_sample_params params;
+	struct mlxsw_sp_port *mlxsw_sp_port;
+	struct mlxsw_sp *mlxsw_sp = priv;
+	int err;
+
+	params.psample_group = psample_group;
+	params.trunc_size = trunc_size;
+	params.rate = rate;
+	params.truncate = truncate;
+	err = mlxsw_sp_sample_trigger_params_set(mlxsw_sp, &trigger, &params,
+						 extack);
+	if (err)
+		return err;
+
+	err = mlxsw_sp_span_agent_get(mlxsw_sp, p_span_id, &agent_parms);
+	if (err) {
+		NL_SET_ERR_MSG_MOD(extack, "Failed to get SPAN agent");
+		goto err_span_agent_get;
+	}
+
+	mlxsw_sp_port = mlxsw_sp->ports[local_port];
+	err = mlxsw_sp_span_analyzed_port_get(mlxsw_sp_port, ingress);
+	if (err) {
+		NL_SET_ERR_MSG_MOD(extack, "Failed to get analyzed port");
+		goto err_analyzed_port_get;
+	}
+
+	return 0;
+
+err_analyzed_port_get:
+	mlxsw_sp_span_agent_put(mlxsw_sp, *p_span_id);
+err_span_agent_get:
+	mlxsw_sp_sample_trigger_params_unset(mlxsw_sp, &trigger);
+	return err;
+}
+
+static void mlxsw_sp2_act_sampler_del(void *priv, u16 local_port, int span_id,
+				      bool ingress)
+{
+	struct mlxsw_sp_sample_trigger trigger = {
+		.type = MLXSW_SP_SAMPLE_TRIGGER_TYPE_POLICY_ENGINE,
+	};
+	struct mlxsw_sp_port *mlxsw_sp_port;
+	struct mlxsw_sp *mlxsw_sp = priv;
+
+	mlxsw_sp_port = mlxsw_sp->ports[local_port];
+	mlxsw_sp_span_analyzed_port_put(mlxsw_sp_port, ingress);
+	mlxsw_sp_span_agent_put(mlxsw_sp, span_id);
+	mlxsw_sp_sample_trigger_params_unset(mlxsw_sp, &trigger);
+}
+
+const struct mlxsw_afa_ops mlxsw_sp2_act_afa_ops = {
+	.kvdl_set_add		= mlxsw_sp2_act_kvdl_set_add,
+	.kvdl_set_del		= mlxsw_sp_act_kvdl_set_del,
+	.kvdl_set_activity_get	= mlxsw_sp2_act_kvdl_set_activity_get,
+	.kvdl_fwd_entry_add	= mlxsw_sp_act_kvdl_fwd_entry_add,
+	.kvdl_fwd_entry_del	= mlxsw_sp_act_kvdl_fwd_entry_del,
+	.counter_index_get	= mlxsw_sp_act_counter_index_get,
+	.counter_index_put	= mlxsw_sp_act_counter_index_put,
+	.mirror_add		= mlxsw_sp_act_mirror_add,
+	.mirror_del		= mlxsw_sp_act_mirror_del,
+	.policer_add		= mlxsw_sp_act_policer_add,
+	.policer_del		= mlxsw_sp_act_policer_del,
+	.sampler_add		= mlxsw_sp2_act_sampler_add,
+	.sampler_del		= mlxsw_sp2_act_sampler_del,
+	.dummy_first_set	= true,
+};
+
+int mlxsw_sp_afa_init(struct mlxsw_sp *mlxsw_sp)
+{
+	mlxsw_sp->afa = mlxsw_afa_create(MLXSW_CORE_RES_GET(mlxsw_sp->core,
+							    ACL_ACTIONS_PER_SET),
+					 mlxsw_sp->afa_ops, mlxsw_sp);
+	return PTR_ERR_OR_ZERO(mlxsw_sp->afa);
+}
+
+void mlxsw_sp_afa_fini(struct mlxsw_sp *mlxsw_sp)
+{
+	mlxsw_afa_destroy(mlxsw_sp->afa);
+}
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_acl_flex_actions.h b/drivers/net/ethernet/mellanox/mlxsw/spectrum_acl_flex_actions.h
new file mode 100644
index 000000000..fe436d816
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_acl_flex_actions.h
@@ -0,0 +1,12 @@
+/* SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0 */
+/* Copyright (c) 2017-2018 Mellanox Technologies. All rights reserved */
+
+#ifndef _MLXSW_SPECTRUM_ACL_FLEX_ACTIONS_H
+#define _MLXSW_SPECTRUM_ACL_FLEX_ACTIONS_H
+
+#include "spectrum.h"
+
+int mlxsw_sp_afa_init(struct mlxsw_sp *mlxsw_sp);
+void mlxsw_sp_afa_fini(struct mlxsw_sp *mlxsw_sp);
+
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_acl_flex_keys.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_acl_flex_keys.c
new file mode 100644
index 000000000..173808c09
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_acl_flex_keys.c
@@ -0,0 +1,355 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2017-2018 Mellanox Technologies. All rights reserved */
+
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include "spectrum.h"
+#include "item.h"
+#include "core_acl_flex_keys.h"
+
+static struct mlxsw_afk_element_inst mlxsw_sp_afk_element_info_l2_dmac[] = {
+	MLXSW_AFK_ELEMENT_INST_BUF(DMAC_32_47, 0x00, 2),
+	MLXSW_AFK_ELEMENT_INST_BUF(DMAC_0_31, 0x02, 4),
+	MLXSW_AFK_ELEMENT_INST_U32(PCP, 0x08, 13, 3),
+	MLXSW_AFK_ELEMENT_INST_U32(VID, 0x08, 0, 12),
+	MLXSW_AFK_ELEMENT_INST_U32(SRC_SYS_PORT, 0x0C, 0, 16),
+};
+
+static struct mlxsw_afk_element_inst mlxsw_sp_afk_element_info_l2_smac[] = {
+	MLXSW_AFK_ELEMENT_INST_BUF(SMAC_32_47, 0x00, 2),
+	MLXSW_AFK_ELEMENT_INST_BUF(SMAC_0_31, 0x02, 4),
+	MLXSW_AFK_ELEMENT_INST_U32(PCP, 0x08, 13, 3),
+	MLXSW_AFK_ELEMENT_INST_U32(VID, 0x08, 0, 12),
+	MLXSW_AFK_ELEMENT_INST_U32(SRC_SYS_PORT, 0x0C, 0, 16),
+};
+
+static struct mlxsw_afk_element_inst mlxsw_sp_afk_element_info_l2_smac_ex[] = {
+	MLXSW_AFK_ELEMENT_INST_BUF(SMAC_32_47, 0x02, 2),
+	MLXSW_AFK_ELEMENT_INST_BUF(SMAC_0_31, 0x04, 4),
+	MLXSW_AFK_ELEMENT_INST_U32(ETHERTYPE, 0x0C, 0, 16),
+};
+
+static struct mlxsw_afk_element_inst mlxsw_sp_afk_element_info_ipv4_sip[] = {
+	MLXSW_AFK_ELEMENT_INST_BUF(SRC_IP_0_31, 0x00, 4),
+	MLXSW_AFK_ELEMENT_INST_U32(IP_PROTO, 0x08, 0, 8),
+	MLXSW_AFK_ELEMENT_INST_U32(SRC_SYS_PORT, 0x0C, 0, 16),
+};
+
+static struct mlxsw_afk_element_inst mlxsw_sp_afk_element_info_ipv4_dip[] = {
+	MLXSW_AFK_ELEMENT_INST_BUF(DST_IP_0_31, 0x00, 4),
+	MLXSW_AFK_ELEMENT_INST_U32(IP_PROTO, 0x08, 0, 8),
+	MLXSW_AFK_ELEMENT_INST_U32(SRC_SYS_PORT, 0x0C, 0, 16),
+};
+
+static struct mlxsw_afk_element_inst mlxsw_sp_afk_element_info_ipv4[] = {
+	MLXSW_AFK_ELEMENT_INST_BUF(SRC_IP_0_31, 0x00, 4),
+	MLXSW_AFK_ELEMENT_INST_U32(IP_ECN, 0x04, 4, 2),
+	MLXSW_AFK_ELEMENT_INST_U32(IP_TTL_, 0x04, 24, 8),
+	MLXSW_AFK_ELEMENT_INST_U32(IP_DSCP, 0x08, 0, 6),
+	MLXSW_AFK_ELEMENT_INST_U32(TCP_FLAGS, 0x08, 8, 9), /* TCP_CONTROL+TCP_ECN */
+};
+
+static struct mlxsw_afk_element_inst mlxsw_sp_afk_element_info_ipv4_ex[] = {
+	MLXSW_AFK_ELEMENT_INST_U32(VID, 0x00, 0, 12),
+	MLXSW_AFK_ELEMENT_INST_U32(PCP, 0x08, 29, 3),
+	MLXSW_AFK_ELEMENT_INST_U32(SRC_L4_PORT, 0x08, 0, 16),
+	MLXSW_AFK_ELEMENT_INST_U32(DST_L4_PORT, 0x0C, 0, 16),
+};
+
+static struct mlxsw_afk_element_inst mlxsw_sp_afk_element_info_ipv6_dip[] = {
+	MLXSW_AFK_ELEMENT_INST_BUF(DST_IP_32_63, 0x00, 4),
+	MLXSW_AFK_ELEMENT_INST_BUF(DST_IP_0_31, 0x04, 4),
+};
+
+static struct mlxsw_afk_element_inst mlxsw_sp_afk_element_info_ipv6_ex1[] = {
+	MLXSW_AFK_ELEMENT_INST_BUF(DST_IP_96_127, 0x00, 4),
+	MLXSW_AFK_ELEMENT_INST_BUF(DST_IP_64_95, 0x04, 4),
+	MLXSW_AFK_ELEMENT_INST_U32(IP_PROTO, 0x08, 0, 8),
+};
+
+static struct mlxsw_afk_element_inst mlxsw_sp_afk_element_info_ipv6_sip[] = {
+	MLXSW_AFK_ELEMENT_INST_BUF(SRC_IP_32_63, 0x00, 4),
+	MLXSW_AFK_ELEMENT_INST_BUF(SRC_IP_0_31, 0x04, 4),
+};
+
+static struct mlxsw_afk_element_inst mlxsw_sp_afk_element_info_ipv6_sip_ex[] = {
+	MLXSW_AFK_ELEMENT_INST_BUF(SRC_IP_96_127, 0x00, 4),
+	MLXSW_AFK_ELEMENT_INST_BUF(SRC_IP_64_95, 0x04, 4),
+};
+
+static struct mlxsw_afk_element_inst mlxsw_sp_afk_element_info_packet_type[] = {
+	MLXSW_AFK_ELEMENT_INST_U32(ETHERTYPE, 0x00, 0, 16),
+};
+
+static const struct mlxsw_afk_block mlxsw_sp1_afk_blocks[] = {
+	MLXSW_AFK_BLOCK(0x10, mlxsw_sp_afk_element_info_l2_dmac),
+	MLXSW_AFK_BLOCK(0x11, mlxsw_sp_afk_element_info_l2_smac),
+	MLXSW_AFK_BLOCK(0x12, mlxsw_sp_afk_element_info_l2_smac_ex),
+	MLXSW_AFK_BLOCK(0x30, mlxsw_sp_afk_element_info_ipv4_sip),
+	MLXSW_AFK_BLOCK(0x31, mlxsw_sp_afk_element_info_ipv4_dip),
+	MLXSW_AFK_BLOCK(0x32, mlxsw_sp_afk_element_info_ipv4),
+	MLXSW_AFK_BLOCK(0x33, mlxsw_sp_afk_element_info_ipv4_ex),
+	MLXSW_AFK_BLOCK(0x60, mlxsw_sp_afk_element_info_ipv6_dip),
+	MLXSW_AFK_BLOCK(0x65, mlxsw_sp_afk_element_info_ipv6_ex1),
+	MLXSW_AFK_BLOCK(0x62, mlxsw_sp_afk_element_info_ipv6_sip),
+	MLXSW_AFK_BLOCK(0x63, mlxsw_sp_afk_element_info_ipv6_sip_ex),
+	MLXSW_AFK_BLOCK(0xB0, mlxsw_sp_afk_element_info_packet_type),
+};
+
+#define MLXSW_SP1_AFK_KEY_BLOCK_SIZE 16
+
+static void mlxsw_sp1_afk_encode_block(char *output, int block_index,
+				       char *block)
+{
+	unsigned int offset = block_index * MLXSW_SP1_AFK_KEY_BLOCK_SIZE;
+	char *output_indexed = output + offset;
+
+	memcpy(output_indexed, block, MLXSW_SP1_AFK_KEY_BLOCK_SIZE);
+}
+
+static void mlxsw_sp1_afk_clear_block(char *output, int block_index)
+{
+	unsigned int offset = block_index * MLXSW_SP1_AFK_KEY_BLOCK_SIZE;
+	char *output_indexed = output + offset;
+
+	memset(output_indexed, 0, MLXSW_SP1_AFK_KEY_BLOCK_SIZE);
+}
+
+const struct mlxsw_afk_ops mlxsw_sp1_afk_ops = {
+	.blocks		= mlxsw_sp1_afk_blocks,
+	.blocks_count	= ARRAY_SIZE(mlxsw_sp1_afk_blocks),
+	.encode_block	= mlxsw_sp1_afk_encode_block,
+	.clear_block	= mlxsw_sp1_afk_clear_block,
+};
+
+static struct mlxsw_afk_element_inst mlxsw_sp_afk_element_info_mac_0[] = {
+	MLXSW_AFK_ELEMENT_INST_BUF(DMAC_0_31, 0x04, 4),
+};
+
+static struct mlxsw_afk_element_inst mlxsw_sp_afk_element_info_mac_1[] = {
+	MLXSW_AFK_ELEMENT_INST_BUF(SMAC_0_31, 0x04, 4),
+};
+
+static struct mlxsw_afk_element_inst mlxsw_sp_afk_element_info_mac_2[] = {
+	MLXSW_AFK_ELEMENT_INST_BUF(SMAC_32_47, 0x04, 2),
+	MLXSW_AFK_ELEMENT_INST_BUF(DMAC_32_47, 0x06, 2),
+};
+
+static struct mlxsw_afk_element_inst mlxsw_sp_afk_element_info_mac_3[] = {
+	MLXSW_AFK_ELEMENT_INST_U32(PCP, 0x00, 0, 3),
+	MLXSW_AFK_ELEMENT_INST_U32(VID, 0x04, 16, 12),
+	MLXSW_AFK_ELEMENT_INST_BUF(DMAC_32_47, 0x06, 2),
+};
+
+static struct mlxsw_afk_element_inst mlxsw_sp_afk_element_info_mac_4[] = {
+	MLXSW_AFK_ELEMENT_INST_U32(PCP, 0x00, 0, 3),
+	MLXSW_AFK_ELEMENT_INST_U32(VID, 0x04, 16, 12),
+	MLXSW_AFK_ELEMENT_INST_U32(ETHERTYPE, 0x04, 0, 16),
+};
+
+static struct mlxsw_afk_element_inst mlxsw_sp_afk_element_info_mac_5[] = {
+	MLXSW_AFK_ELEMENT_INST_U32(VID, 0x04, 16, 12),
+	MLXSW_AFK_ELEMENT_INST_EXT_U32(SRC_SYS_PORT, 0x04, 0, 8, -1, true), /* RX_ACL_SYSTEM_PORT */
+};
+
+static struct mlxsw_afk_element_inst mlxsw_sp_afk_element_info_ipv4_0[] = {
+	MLXSW_AFK_ELEMENT_INST_BUF(DST_IP_0_31, 0x04, 4),
+};
+
+static struct mlxsw_afk_element_inst mlxsw_sp_afk_element_info_ipv4_1[] = {
+	MLXSW_AFK_ELEMENT_INST_BUF(SRC_IP_0_31, 0x04, 4),
+};
+
+static struct mlxsw_afk_element_inst mlxsw_sp_afk_element_info_ipv4_2[] = {
+	MLXSW_AFK_ELEMENT_INST_U32(IP_DSCP, 0x04, 0, 6),
+	MLXSW_AFK_ELEMENT_INST_U32(IP_ECN, 0x04, 6, 2),
+	MLXSW_AFK_ELEMENT_INST_U32(IP_TTL_, 0x04, 8, 8),
+	MLXSW_AFK_ELEMENT_INST_U32(IP_PROTO, 0x04, 16, 8),
+};
+
+static struct mlxsw_afk_element_inst mlxsw_sp_afk_element_info_ipv4_4[] = {
+	MLXSW_AFK_ELEMENT_INST_U32(VIRT_ROUTER_LSB, 0x04, 24, 8),
+	MLXSW_AFK_ELEMENT_INST_EXT_U32(VIRT_ROUTER_MSB, 0x00, 0, 3, 0, true),
+};
+
+static struct mlxsw_afk_element_inst mlxsw_sp_afk_element_info_ipv6_0[] = {
+	MLXSW_AFK_ELEMENT_INST_BUF(DST_IP_32_63, 0x04, 4),
+};
+
+static struct mlxsw_afk_element_inst mlxsw_sp_afk_element_info_ipv6_1[] = {
+	MLXSW_AFK_ELEMENT_INST_BUF(DST_IP_64_95, 0x04, 4),
+};
+
+static struct mlxsw_afk_element_inst mlxsw_sp_afk_element_info_ipv6_2[] = {
+	MLXSW_AFK_ELEMENT_INST_BUF(DST_IP_96_127, 0x04, 4),
+};
+
+static struct mlxsw_afk_element_inst mlxsw_sp_afk_element_info_ipv6_3[] = {
+	MLXSW_AFK_ELEMENT_INST_BUF(SRC_IP_32_63, 0x04, 4),
+};
+
+static struct mlxsw_afk_element_inst mlxsw_sp_afk_element_info_ipv6_4[] = {
+	MLXSW_AFK_ELEMENT_INST_BUF(SRC_IP_64_95, 0x04, 4),
+};
+
+static struct mlxsw_afk_element_inst mlxsw_sp_afk_element_info_ipv6_5[] = {
+	MLXSW_AFK_ELEMENT_INST_BUF(SRC_IP_96_127, 0x04, 4),
+};
+
+static struct mlxsw_afk_element_inst mlxsw_sp_afk_element_info_l4_0[] = {
+	MLXSW_AFK_ELEMENT_INST_U32(SRC_L4_PORT, 0x04, 16, 16),
+	MLXSW_AFK_ELEMENT_INST_U32(DST_L4_PORT, 0x04, 0, 16),
+};
+
+static struct mlxsw_afk_element_inst mlxsw_sp_afk_element_info_l4_2[] = {
+	MLXSW_AFK_ELEMENT_INST_U32(TCP_FLAGS, 0x04, 16, 9), /* TCP_CONTROL + TCP_ECN */
+};
+
+static const struct mlxsw_afk_block mlxsw_sp2_afk_blocks[] = {
+	MLXSW_AFK_BLOCK(0x10, mlxsw_sp_afk_element_info_mac_0),
+	MLXSW_AFK_BLOCK(0x11, mlxsw_sp_afk_element_info_mac_1),
+	MLXSW_AFK_BLOCK(0x12, mlxsw_sp_afk_element_info_mac_2),
+	MLXSW_AFK_BLOCK(0x13, mlxsw_sp_afk_element_info_mac_3),
+	MLXSW_AFK_BLOCK(0x14, mlxsw_sp_afk_element_info_mac_4),
+	MLXSW_AFK_BLOCK(0x15, mlxsw_sp_afk_element_info_mac_5),
+	MLXSW_AFK_BLOCK(0x38, mlxsw_sp_afk_element_info_ipv4_0),
+	MLXSW_AFK_BLOCK(0x39, mlxsw_sp_afk_element_info_ipv4_1),
+	MLXSW_AFK_BLOCK(0x3A, mlxsw_sp_afk_element_info_ipv4_2),
+	MLXSW_AFK_BLOCK(0x3C, mlxsw_sp_afk_element_info_ipv4_4),
+	MLXSW_AFK_BLOCK(0x40, mlxsw_sp_afk_element_info_ipv6_0),
+	MLXSW_AFK_BLOCK(0x41, mlxsw_sp_afk_element_info_ipv6_1),
+	MLXSW_AFK_BLOCK(0x42, mlxsw_sp_afk_element_info_ipv6_2),
+	MLXSW_AFK_BLOCK(0x43, mlxsw_sp_afk_element_info_ipv6_3),
+	MLXSW_AFK_BLOCK(0x44, mlxsw_sp_afk_element_info_ipv6_4),
+	MLXSW_AFK_BLOCK(0x45, mlxsw_sp_afk_element_info_ipv6_5),
+	MLXSW_AFK_BLOCK(0x90, mlxsw_sp_afk_element_info_l4_0),
+	MLXSW_AFK_BLOCK(0x92, mlxsw_sp_afk_element_info_l4_2),
+};
+
+#define MLXSW_SP2_AFK_BITS_PER_BLOCK 36
+
+/* A block in Spectrum-2 is of the following form:
+ *
+ * +--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+
+ * |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |35|34|33|32|
+ * +-----------------------------------------------------------------------------------------------+
+ * |31|30|29|28|27|26|25|24|23|22|21|20|19|18|17|16|15|14|13|12|11|10| 9| 8| 7| 6| 5| 4| 3| 2| 1| 0|
+ * +--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+
+ */
+MLXSW_ITEM64(sp2_afk, block, value, 0x00, 0, MLXSW_SP2_AFK_BITS_PER_BLOCK);
+
+/* The key / mask block layout in Spectrum-2 is of the following form:
+ *
+ * +--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+
+ * |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |                block11_high                   |
+ * +-----------------------------------------------------------------------------------------------+
+ * |                    block11_low                               |         block10_high           |
+ * +--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+
+ * ...
+ */
+
+struct mlxsw_sp2_afk_block_layout {
+	unsigned short offset;
+	struct mlxsw_item item;
+};
+
+#define MLXSW_SP2_AFK_BLOCK_LAYOUT(_block, _offset, _shift)			\
+	{									\
+		.offset = _offset,						\
+		{								\
+			.shift = _shift,					\
+			.size = {.bits = MLXSW_SP2_AFK_BITS_PER_BLOCK},		\
+			.name = #_block,					\
+		}								\
+	}									\
+
+static const struct mlxsw_sp2_afk_block_layout mlxsw_sp2_afk_blocks_layout[] = {
+	MLXSW_SP2_AFK_BLOCK_LAYOUT(block0, 0x30, 0),
+	MLXSW_SP2_AFK_BLOCK_LAYOUT(block1, 0x2C, 4),
+	MLXSW_SP2_AFK_BLOCK_LAYOUT(block2, 0x28, 8),
+	MLXSW_SP2_AFK_BLOCK_LAYOUT(block3, 0x24, 12),
+	MLXSW_SP2_AFK_BLOCK_LAYOUT(block4, 0x20, 16),
+	MLXSW_SP2_AFK_BLOCK_LAYOUT(block5, 0x1C, 20),
+	MLXSW_SP2_AFK_BLOCK_LAYOUT(block6, 0x18, 24),
+	MLXSW_SP2_AFK_BLOCK_LAYOUT(block7, 0x14, 28),
+	MLXSW_SP2_AFK_BLOCK_LAYOUT(block8, 0x0C, 0),
+	MLXSW_SP2_AFK_BLOCK_LAYOUT(block9, 0x08, 4),
+	MLXSW_SP2_AFK_BLOCK_LAYOUT(block10, 0x04, 8),
+	MLXSW_SP2_AFK_BLOCK_LAYOUT(block11, 0x00, 12),
+};
+
+static void __mlxsw_sp2_afk_block_value_set(char *output, int block_index,
+					    u64 block_value)
+{
+	const struct mlxsw_sp2_afk_block_layout *block_layout;
+
+	if (WARN_ON(block_index < 0 ||
+		    block_index >= ARRAY_SIZE(mlxsw_sp2_afk_blocks_layout)))
+		return;
+
+	block_layout = &mlxsw_sp2_afk_blocks_layout[block_index];
+	__mlxsw_item_set64(output + block_layout->offset,
+			   &block_layout->item, 0, block_value);
+}
+
+static void mlxsw_sp2_afk_encode_block(char *output, int block_index,
+				       char *block)
+{
+	u64 block_value = mlxsw_sp2_afk_block_value_get(block);
+
+	__mlxsw_sp2_afk_block_value_set(output, block_index, block_value);
+}
+
+static void mlxsw_sp2_afk_clear_block(char *output, int block_index)
+{
+	__mlxsw_sp2_afk_block_value_set(output, block_index, 0);
+}
+
+const struct mlxsw_afk_ops mlxsw_sp2_afk_ops = {
+	.blocks		= mlxsw_sp2_afk_blocks,
+	.blocks_count	= ARRAY_SIZE(mlxsw_sp2_afk_blocks),
+	.encode_block	= mlxsw_sp2_afk_encode_block,
+	.clear_block	= mlxsw_sp2_afk_clear_block,
+};
+
+static struct mlxsw_afk_element_inst mlxsw_sp_afk_element_info_mac_5b[] = {
+	MLXSW_AFK_ELEMENT_INST_U32(VID, 0x04, 18, 12),
+	MLXSW_AFK_ELEMENT_INST_EXT_U32(SRC_SYS_PORT, 0x04, 0, 9, -1, true), /* RX_ACL_SYSTEM_PORT */
+};
+
+static struct mlxsw_afk_element_inst mlxsw_sp_afk_element_info_ipv4_4b[] = {
+	MLXSW_AFK_ELEMENT_INST_U32(VIRT_ROUTER_LSB, 0x04, 13, 8),
+	MLXSW_AFK_ELEMENT_INST_U32(VIRT_ROUTER_MSB, 0x04, 21, 4),
+};
+
+static struct mlxsw_afk_element_inst mlxsw_sp_afk_element_info_ipv6_2b[] = {
+	MLXSW_AFK_ELEMENT_INST_BUF(DST_IP_96_127, 0x04, 4),
+};
+
+static const struct mlxsw_afk_block mlxsw_sp4_afk_blocks[] = {
+	MLXSW_AFK_BLOCK(0x10, mlxsw_sp_afk_element_info_mac_0),
+	MLXSW_AFK_BLOCK(0x11, mlxsw_sp_afk_element_info_mac_1),
+	MLXSW_AFK_BLOCK(0x12, mlxsw_sp_afk_element_info_mac_2),
+	MLXSW_AFK_BLOCK(0x13, mlxsw_sp_afk_element_info_mac_3),
+	MLXSW_AFK_BLOCK(0x14, mlxsw_sp_afk_element_info_mac_4),
+	MLXSW_AFK_BLOCK(0x1A, mlxsw_sp_afk_element_info_mac_5b),
+	MLXSW_AFK_BLOCK(0x38, mlxsw_sp_afk_element_info_ipv4_0),
+	MLXSW_AFK_BLOCK(0x39, mlxsw_sp_afk_element_info_ipv4_1),
+	MLXSW_AFK_BLOCK(0x3A, mlxsw_sp_afk_element_info_ipv4_2),
+	MLXSW_AFK_BLOCK(0x35, mlxsw_sp_afk_element_info_ipv4_4b),
+	MLXSW_AFK_BLOCK(0x40, mlxsw_sp_afk_element_info_ipv6_0),
+	MLXSW_AFK_BLOCK(0x41, mlxsw_sp_afk_element_info_ipv6_1),
+	MLXSW_AFK_BLOCK(0x47, mlxsw_sp_afk_element_info_ipv6_2b),
+	MLXSW_AFK_BLOCK(0x43, mlxsw_sp_afk_element_info_ipv6_3),
+	MLXSW_AFK_BLOCK(0x44, mlxsw_sp_afk_element_info_ipv6_4),
+	MLXSW_AFK_BLOCK(0x45, mlxsw_sp_afk_element_info_ipv6_5),
+	MLXSW_AFK_BLOCK(0x90, mlxsw_sp_afk_element_info_l4_0),
+	MLXSW_AFK_BLOCK(0x92, mlxsw_sp_afk_element_info_l4_2),
+};
+
+const struct mlxsw_afk_ops mlxsw_sp4_afk_ops = {
+	.blocks		= mlxsw_sp4_afk_blocks,
+	.blocks_count	= ARRAY_SIZE(mlxsw_sp4_afk_blocks),
+	.encode_block	= mlxsw_sp2_afk_encode_block,
+	.clear_block	= mlxsw_sp2_afk_clear_block,
+};
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_acl_tcam.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_acl_tcam.c
new file mode 100644
index 000000000..3b9ba8fa2
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_acl_tcam.c
@@ -0,0 +1,1867 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2017-2018 Mellanox Technologies. All rights reserved */
+
+#include <linux/kernel.h>
+#include <linux/slab.h>
+#include <linux/errno.h>
+#include <linux/bitops.h>
+#include <linux/list.h>
+#include <linux/rhashtable.h>
+#include <linux/netdevice.h>
+#include <linux/mutex.h>
+#include <trace/events/mlxsw.h>
+
+#include "reg.h"
+#include "core.h"
+#include "resources.h"
+#include "spectrum.h"
+#include "spectrum_acl_tcam.h"
+#include "core_acl_flex_keys.h"
+
+size_t mlxsw_sp_acl_tcam_priv_size(struct mlxsw_sp *mlxsw_sp)
+{
+	const struct mlxsw_sp_acl_tcam_ops *ops = mlxsw_sp->acl_tcam_ops;
+
+	return ops->priv_size;
+}
+
+#define MLXSW_SP_ACL_TCAM_VREGION_REHASH_INTRVL_DFLT 5000 /* ms */
+#define MLXSW_SP_ACL_TCAM_VREGION_REHASH_INTRVL_MIN 3000 /* ms */
+#define MLXSW_SP_ACL_TCAM_VREGION_REHASH_CREDITS 100 /* number of entries */
+
+int mlxsw_sp_acl_tcam_init(struct mlxsw_sp *mlxsw_sp,
+			   struct mlxsw_sp_acl_tcam *tcam)
+{
+	const struct mlxsw_sp_acl_tcam_ops *ops = mlxsw_sp->acl_tcam_ops;
+	u64 max_tcam_regions;
+	u64 max_regions;
+	u64 max_groups;
+	int err;
+
+	mutex_init(&tcam->lock);
+	tcam->vregion_rehash_intrvl =
+			MLXSW_SP_ACL_TCAM_VREGION_REHASH_INTRVL_DFLT;
+	INIT_LIST_HEAD(&tcam->vregion_list);
+
+	max_tcam_regions = MLXSW_CORE_RES_GET(mlxsw_sp->core,
+					      ACL_MAX_TCAM_REGIONS);
+	max_regions = MLXSW_CORE_RES_GET(mlxsw_sp->core, ACL_MAX_REGIONS);
+
+	/* Use 1:1 mapping between ACL region and TCAM region */
+	if (max_tcam_regions < max_regions)
+		max_regions = max_tcam_regions;
+
+	tcam->used_regions = bitmap_zalloc(max_regions, GFP_KERNEL);
+	if (!tcam->used_regions)
+		return -ENOMEM;
+	tcam->max_regions = max_regions;
+
+	max_groups = MLXSW_CORE_RES_GET(mlxsw_sp->core, ACL_MAX_GROUPS);
+	tcam->used_groups = bitmap_zalloc(max_groups, GFP_KERNEL);
+	if (!tcam->used_groups) {
+		err = -ENOMEM;
+		goto err_alloc_used_groups;
+	}
+	tcam->max_groups = max_groups;
+	tcam->max_group_size = MLXSW_CORE_RES_GET(mlxsw_sp->core,
+						 ACL_MAX_GROUP_SIZE);
+
+	err = ops->init(mlxsw_sp, tcam->priv, tcam);
+	if (err)
+		goto err_tcam_init;
+
+	return 0;
+
+err_tcam_init:
+	bitmap_free(tcam->used_groups);
+err_alloc_used_groups:
+	bitmap_free(tcam->used_regions);
+	return err;
+}
+
+void mlxsw_sp_acl_tcam_fini(struct mlxsw_sp *mlxsw_sp,
+			    struct mlxsw_sp_acl_tcam *tcam)
+{
+	const struct mlxsw_sp_acl_tcam_ops *ops = mlxsw_sp->acl_tcam_ops;
+
+	mutex_destroy(&tcam->lock);
+	ops->fini(mlxsw_sp, tcam->priv);
+	bitmap_free(tcam->used_groups);
+	bitmap_free(tcam->used_regions);
+}
+
+int mlxsw_sp_acl_tcam_priority_get(struct mlxsw_sp *mlxsw_sp,
+				   struct mlxsw_sp_acl_rule_info *rulei,
+				   u32 *priority, bool fillup_priority)
+{
+	u64 max_priority;
+
+	if (!fillup_priority) {
+		*priority = 0;
+		return 0;
+	}
+
+	if (!MLXSW_CORE_RES_VALID(mlxsw_sp->core, KVD_SIZE))
+		return -EIO;
+
+	/* Priority range is 1..cap_kvd_size-1. */
+	max_priority = MLXSW_CORE_RES_GET(mlxsw_sp->core, KVD_SIZE) - 1;
+	if (rulei->priority >= max_priority)
+		return -EINVAL;
+
+	/* Unlike in TC, in HW, higher number means higher priority. */
+	*priority = max_priority - rulei->priority;
+	return 0;
+}
+
+static int mlxsw_sp_acl_tcam_region_id_get(struct mlxsw_sp_acl_tcam *tcam,
+					   u16 *p_id)
+{
+	u16 id;
+
+	id = find_first_zero_bit(tcam->used_regions, tcam->max_regions);
+	if (id < tcam->max_regions) {
+		__set_bit(id, tcam->used_regions);
+		*p_id = id;
+		return 0;
+	}
+	return -ENOBUFS;
+}
+
+static void mlxsw_sp_acl_tcam_region_id_put(struct mlxsw_sp_acl_tcam *tcam,
+					    u16 id)
+{
+	__clear_bit(id, tcam->used_regions);
+}
+
+static int mlxsw_sp_acl_tcam_group_id_get(struct mlxsw_sp_acl_tcam *tcam,
+					  u16 *p_id)
+{
+	u16 id;
+
+	id = find_first_zero_bit(tcam->used_groups, tcam->max_groups);
+	if (id < tcam->max_groups) {
+		__set_bit(id, tcam->used_groups);
+		*p_id = id;
+		return 0;
+	}
+	return -ENOBUFS;
+}
+
+static void mlxsw_sp_acl_tcam_group_id_put(struct mlxsw_sp_acl_tcam *tcam,
+					   u16 id)
+{
+	__clear_bit(id, tcam->used_groups);
+}
+
+struct mlxsw_sp_acl_tcam_pattern {
+	const enum mlxsw_afk_element *elements;
+	unsigned int elements_count;
+};
+
+struct mlxsw_sp_acl_tcam_group {
+	struct mlxsw_sp_acl_tcam *tcam;
+	u16 id;
+	struct mutex lock; /* guards region list updates */
+	struct list_head region_list;
+	unsigned int region_count;
+};
+
+struct mlxsw_sp_acl_tcam_vgroup {
+	struct mlxsw_sp_acl_tcam_group group;
+	struct list_head vregion_list;
+	struct rhashtable vchunk_ht;
+	const struct mlxsw_sp_acl_tcam_pattern *patterns;
+	unsigned int patterns_count;
+	bool tmplt_elusage_set;
+	struct mlxsw_afk_element_usage tmplt_elusage;
+	bool vregion_rehash_enabled;
+	unsigned int *p_min_prio;
+	unsigned int *p_max_prio;
+};
+
+struct mlxsw_sp_acl_tcam_rehash_ctx {
+	void *hints_priv;
+	bool this_is_rollback;
+	struct mlxsw_sp_acl_tcam_vchunk *current_vchunk; /* vchunk being
+							  * currently migrated.
+							  */
+	struct mlxsw_sp_acl_tcam_ventry *start_ventry; /* ventry to start
+							* migration from in
+							* a vchunk being
+							* currently migrated.
+							*/
+	struct mlxsw_sp_acl_tcam_ventry *stop_ventry; /* ventry to stop
+						       * migration at
+						       * a vchunk being
+						       * currently migrated.
+						       */
+};
+
+struct mlxsw_sp_acl_tcam_vregion {
+	struct mutex lock; /* Protects consistency of region, region2 pointers
+			    * and vchunk_list.
+			    */
+	struct mlxsw_sp_acl_tcam_region *region;
+	struct mlxsw_sp_acl_tcam_region *region2; /* Used during migration */
+	struct list_head list; /* Member of a TCAM group */
+	struct list_head tlist; /* Member of a TCAM */
+	struct list_head vchunk_list; /* List of vchunks under this vregion */
+	struct mlxsw_afk_key_info *key_info;
+	struct mlxsw_sp_acl_tcam *tcam;
+	struct mlxsw_sp_acl_tcam_vgroup *vgroup;
+	struct {
+		struct delayed_work dw;
+		struct mlxsw_sp_acl_tcam_rehash_ctx ctx;
+	} rehash;
+	struct mlxsw_sp *mlxsw_sp;
+	unsigned int ref_count;
+};
+
+struct mlxsw_sp_acl_tcam_vchunk;
+
+struct mlxsw_sp_acl_tcam_chunk {
+	struct mlxsw_sp_acl_tcam_vchunk *vchunk;
+	struct mlxsw_sp_acl_tcam_region *region;
+	unsigned long priv[];
+	/* priv has to be always the last item */
+};
+
+struct mlxsw_sp_acl_tcam_vchunk {
+	struct mlxsw_sp_acl_tcam_chunk *chunk;
+	struct mlxsw_sp_acl_tcam_chunk *chunk2; /* Used during migration */
+	struct list_head list; /* Member of a TCAM vregion */
+	struct rhash_head ht_node; /* Member of a chunk HT */
+	struct list_head ventry_list;
+	unsigned int priority; /* Priority within the vregion and group */
+	struct mlxsw_sp_acl_tcam_vgroup *vgroup;
+	struct mlxsw_sp_acl_tcam_vregion *vregion;
+	unsigned int ref_count;
+};
+
+struct mlxsw_sp_acl_tcam_entry {
+	struct mlxsw_sp_acl_tcam_ventry *ventry;
+	struct mlxsw_sp_acl_tcam_chunk *chunk;
+	unsigned long priv[];
+	/* priv has to be always the last item */
+};
+
+struct mlxsw_sp_acl_tcam_ventry {
+	struct mlxsw_sp_acl_tcam_entry *entry;
+	struct list_head list; /* Member of a TCAM vchunk */
+	struct mlxsw_sp_acl_tcam_vchunk *vchunk;
+	struct mlxsw_sp_acl_rule_info *rulei;
+};
+
+static const struct rhashtable_params mlxsw_sp_acl_tcam_vchunk_ht_params = {
+	.key_len = sizeof(unsigned int),
+	.key_offset = offsetof(struct mlxsw_sp_acl_tcam_vchunk, priority),
+	.head_offset = offsetof(struct mlxsw_sp_acl_tcam_vchunk, ht_node),
+	.automatic_shrinking = true,
+};
+
+static int mlxsw_sp_acl_tcam_group_update(struct mlxsw_sp *mlxsw_sp,
+					  struct mlxsw_sp_acl_tcam_group *group)
+{
+	struct mlxsw_sp_acl_tcam_region *region;
+	char pagt_pl[MLXSW_REG_PAGT_LEN];
+	int acl_index = 0;
+
+	mlxsw_reg_pagt_pack(pagt_pl, group->id);
+	list_for_each_entry(region, &group->region_list, list) {
+		bool multi = false;
+
+		/* Check if the next entry in the list has the same vregion. */
+		if (region->list.next != &group->region_list &&
+		    list_next_entry(region, list)->vregion == region->vregion)
+			multi = true;
+		mlxsw_reg_pagt_acl_id_pack(pagt_pl, acl_index++,
+					   region->id, multi);
+	}
+	mlxsw_reg_pagt_size_set(pagt_pl, acl_index);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(pagt), pagt_pl);
+}
+
+static int
+mlxsw_sp_acl_tcam_group_add(struct mlxsw_sp_acl_tcam *tcam,
+			    struct mlxsw_sp_acl_tcam_group *group)
+{
+	int err;
+
+	group->tcam = tcam;
+	INIT_LIST_HEAD(&group->region_list);
+
+	err = mlxsw_sp_acl_tcam_group_id_get(tcam, &group->id);
+	if (err)
+		return err;
+
+	mutex_init(&group->lock);
+
+	return 0;
+}
+
+static void mlxsw_sp_acl_tcam_group_del(struct mlxsw_sp_acl_tcam_group *group)
+{
+	struct mlxsw_sp_acl_tcam *tcam = group->tcam;
+
+	mutex_destroy(&group->lock);
+	mlxsw_sp_acl_tcam_group_id_put(tcam, group->id);
+	WARN_ON(!list_empty(&group->region_list));
+}
+
+static int
+mlxsw_sp_acl_tcam_vgroup_add(struct mlxsw_sp *mlxsw_sp,
+			     struct mlxsw_sp_acl_tcam *tcam,
+			     struct mlxsw_sp_acl_tcam_vgroup *vgroup,
+			     const struct mlxsw_sp_acl_tcam_pattern *patterns,
+			     unsigned int patterns_count,
+			     struct mlxsw_afk_element_usage *tmplt_elusage,
+			     bool vregion_rehash_enabled,
+			     unsigned int *p_min_prio,
+			     unsigned int *p_max_prio)
+{
+	int err;
+
+	vgroup->patterns = patterns;
+	vgroup->patterns_count = patterns_count;
+	vgroup->vregion_rehash_enabled = vregion_rehash_enabled;
+	vgroup->p_min_prio = p_min_prio;
+	vgroup->p_max_prio = p_max_prio;
+
+	if (tmplt_elusage) {
+		vgroup->tmplt_elusage_set = true;
+		memcpy(&vgroup->tmplt_elusage, tmplt_elusage,
+		       sizeof(vgroup->tmplt_elusage));
+	}
+	INIT_LIST_HEAD(&vgroup->vregion_list);
+
+	err = mlxsw_sp_acl_tcam_group_add(tcam, &vgroup->group);
+	if (err)
+		return err;
+
+	err = rhashtable_init(&vgroup->vchunk_ht,
+			      &mlxsw_sp_acl_tcam_vchunk_ht_params);
+	if (err)
+		goto err_rhashtable_init;
+
+	return 0;
+
+err_rhashtable_init:
+	mlxsw_sp_acl_tcam_group_del(&vgroup->group);
+	return err;
+}
+
+static void
+mlxsw_sp_acl_tcam_vgroup_del(struct mlxsw_sp_acl_tcam_vgroup *vgroup)
+{
+	rhashtable_destroy(&vgroup->vchunk_ht);
+	mlxsw_sp_acl_tcam_group_del(&vgroup->group);
+	WARN_ON(!list_empty(&vgroup->vregion_list));
+}
+
+static int
+mlxsw_sp_acl_tcam_group_bind(struct mlxsw_sp *mlxsw_sp,
+			     struct mlxsw_sp_acl_tcam_group *group,
+			     struct mlxsw_sp_port *mlxsw_sp_port,
+			     bool ingress)
+{
+	char ppbt_pl[MLXSW_REG_PPBT_LEN];
+
+	mlxsw_reg_ppbt_pack(ppbt_pl, ingress ? MLXSW_REG_PXBT_E_IACL :
+					       MLXSW_REG_PXBT_E_EACL,
+			    MLXSW_REG_PXBT_OP_BIND, mlxsw_sp_port->local_port,
+			    group->id);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(ppbt), ppbt_pl);
+}
+
+static void
+mlxsw_sp_acl_tcam_group_unbind(struct mlxsw_sp *mlxsw_sp,
+			       struct mlxsw_sp_acl_tcam_group *group,
+			       struct mlxsw_sp_port *mlxsw_sp_port,
+			       bool ingress)
+{
+	char ppbt_pl[MLXSW_REG_PPBT_LEN];
+
+	mlxsw_reg_ppbt_pack(ppbt_pl, ingress ? MLXSW_REG_PXBT_E_IACL :
+					       MLXSW_REG_PXBT_E_EACL,
+			    MLXSW_REG_PXBT_OP_UNBIND, mlxsw_sp_port->local_port,
+			    group->id);
+	mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(ppbt), ppbt_pl);
+}
+
+static u16
+mlxsw_sp_acl_tcam_group_id(struct mlxsw_sp_acl_tcam_group *group)
+{
+	return group->id;
+}
+
+static unsigned int
+mlxsw_sp_acl_tcam_vregion_prio(struct mlxsw_sp_acl_tcam_vregion *vregion)
+{
+	struct mlxsw_sp_acl_tcam_vchunk *vchunk;
+
+	if (list_empty(&vregion->vchunk_list))
+		return 0;
+	/* As a priority of a vregion, return priority of the first vchunk */
+	vchunk = list_first_entry(&vregion->vchunk_list,
+				  typeof(*vchunk), list);
+	return vchunk->priority;
+}
+
+static unsigned int
+mlxsw_sp_acl_tcam_vregion_max_prio(struct mlxsw_sp_acl_tcam_vregion *vregion)
+{
+	struct mlxsw_sp_acl_tcam_vchunk *vchunk;
+
+	if (list_empty(&vregion->vchunk_list))
+		return 0;
+	vchunk = list_last_entry(&vregion->vchunk_list,
+				 typeof(*vchunk), list);
+	return vchunk->priority;
+}
+
+static void
+mlxsw_sp_acl_tcam_vgroup_prio_update(struct mlxsw_sp_acl_tcam_vgroup *vgroup)
+{
+	struct mlxsw_sp_acl_tcam_vregion *vregion;
+
+	if (list_empty(&vgroup->vregion_list))
+		return;
+	vregion = list_first_entry(&vgroup->vregion_list,
+				   typeof(*vregion), list);
+	*vgroup->p_min_prio = mlxsw_sp_acl_tcam_vregion_prio(vregion);
+	vregion = list_last_entry(&vgroup->vregion_list,
+				  typeof(*vregion), list);
+	*vgroup->p_max_prio = mlxsw_sp_acl_tcam_vregion_max_prio(vregion);
+}
+
+static int
+mlxsw_sp_acl_tcam_group_region_attach(struct mlxsw_sp *mlxsw_sp,
+				      struct mlxsw_sp_acl_tcam_group *group,
+				      struct mlxsw_sp_acl_tcam_region *region,
+				      unsigned int priority,
+				      struct mlxsw_sp_acl_tcam_region *next_region)
+{
+	struct mlxsw_sp_acl_tcam_region *region2;
+	struct list_head *pos;
+	int err;
+
+	mutex_lock(&group->lock);
+	if (group->region_count == group->tcam->max_group_size) {
+		err = -ENOBUFS;
+		goto err_region_count_check;
+	}
+
+	if (next_region) {
+		/* If the next region is defined, place the new one
+		 * before it. The next one is a sibling.
+		 */
+		pos = &next_region->list;
+	} else {
+		/* Position the region inside the list according to priority */
+		list_for_each(pos, &group->region_list) {
+			region2 = list_entry(pos, typeof(*region2), list);
+			if (mlxsw_sp_acl_tcam_vregion_prio(region2->vregion) >
+			    priority)
+				break;
+		}
+	}
+	list_add_tail(&region->list, pos);
+	region->group = group;
+
+	err = mlxsw_sp_acl_tcam_group_update(mlxsw_sp, group);
+	if (err)
+		goto err_group_update;
+
+	group->region_count++;
+	mutex_unlock(&group->lock);
+	return 0;
+
+err_group_update:
+	list_del(&region->list);
+err_region_count_check:
+	mutex_unlock(&group->lock);
+	return err;
+}
+
+static void
+mlxsw_sp_acl_tcam_group_region_detach(struct mlxsw_sp *mlxsw_sp,
+				      struct mlxsw_sp_acl_tcam_region *region)
+{
+	struct mlxsw_sp_acl_tcam_group *group = region->group;
+
+	mutex_lock(&group->lock);
+	list_del(&region->list);
+	group->region_count--;
+	mlxsw_sp_acl_tcam_group_update(mlxsw_sp, group);
+	mutex_unlock(&group->lock);
+}
+
+static int
+mlxsw_sp_acl_tcam_vgroup_vregion_attach(struct mlxsw_sp *mlxsw_sp,
+					struct mlxsw_sp_acl_tcam_vgroup *vgroup,
+					struct mlxsw_sp_acl_tcam_vregion *vregion,
+					unsigned int priority)
+{
+	struct mlxsw_sp_acl_tcam_vregion *vregion2;
+	struct list_head *pos;
+	int err;
+
+	/* Position the vregion inside the list according to priority */
+	list_for_each(pos, &vgroup->vregion_list) {
+		vregion2 = list_entry(pos, typeof(*vregion2), list);
+		if (mlxsw_sp_acl_tcam_vregion_prio(vregion2) > priority)
+			break;
+	}
+	list_add_tail(&vregion->list, pos);
+
+	err = mlxsw_sp_acl_tcam_group_region_attach(mlxsw_sp, &vgroup->group,
+						    vregion->region,
+						    priority, NULL);
+	if (err)
+		goto err_region_attach;
+
+	return 0;
+
+err_region_attach:
+	list_del(&vregion->list);
+	return err;
+}
+
+static void
+mlxsw_sp_acl_tcam_vgroup_vregion_detach(struct mlxsw_sp *mlxsw_sp,
+					struct mlxsw_sp_acl_tcam_vregion *vregion)
+{
+	list_del(&vregion->list);
+	if (vregion->region2)
+		mlxsw_sp_acl_tcam_group_region_detach(mlxsw_sp,
+						      vregion->region2);
+	mlxsw_sp_acl_tcam_group_region_detach(mlxsw_sp, vregion->region);
+}
+
+static struct mlxsw_sp_acl_tcam_vregion *
+mlxsw_sp_acl_tcam_vgroup_vregion_find(struct mlxsw_sp_acl_tcam_vgroup *vgroup,
+				      unsigned int priority,
+				      struct mlxsw_afk_element_usage *elusage,
+				      bool *p_need_split)
+{
+	struct mlxsw_sp_acl_tcam_vregion *vregion, *vregion2;
+	struct list_head *pos;
+	bool issubset;
+
+	list_for_each(pos, &vgroup->vregion_list) {
+		vregion = list_entry(pos, typeof(*vregion), list);
+
+		/* First, check if the requested priority does not rather belong
+		 * under some of the next vregions.
+		 */
+		if (pos->next != &vgroup->vregion_list) { /* not last */
+			vregion2 = list_entry(pos->next, typeof(*vregion2),
+					      list);
+			if (priority >=
+			    mlxsw_sp_acl_tcam_vregion_prio(vregion2))
+				continue;
+		}
+
+		issubset = mlxsw_afk_key_info_subset(vregion->key_info,
+						     elusage);
+
+		/* If requested element usage would not fit and the priority
+		 * is lower than the currently inspected vregion we cannot
+		 * use this region, so return NULL to indicate new vregion has
+		 * to be created.
+		 */
+		if (!issubset &&
+		    priority < mlxsw_sp_acl_tcam_vregion_prio(vregion))
+			return NULL;
+
+		/* If requested element usage would not fit and the priority
+		 * is higher than the currently inspected vregion we cannot
+		 * use this vregion. There is still some hope that the next
+		 * vregion would be the fit. So let it be processed and
+		 * eventually break at the check right above this.
+		 */
+		if (!issubset &&
+		    priority > mlxsw_sp_acl_tcam_vregion_max_prio(vregion))
+			continue;
+
+		/* Indicate if the vregion needs to be split in order to add
+		 * the requested priority. Split is needed when requested
+		 * element usage won't fit into the found vregion.
+		 */
+		*p_need_split = !issubset;
+		return vregion;
+	}
+	return NULL; /* New vregion has to be created. */
+}
+
+static void
+mlxsw_sp_acl_tcam_vgroup_use_patterns(struct mlxsw_sp_acl_tcam_vgroup *vgroup,
+				      struct mlxsw_afk_element_usage *elusage,
+				      struct mlxsw_afk_element_usage *out)
+{
+	const struct mlxsw_sp_acl_tcam_pattern *pattern;
+	int i;
+
+	/* In case the template is set, we don't have to look up the pattern
+	 * and just use the template.
+	 */
+	if (vgroup->tmplt_elusage_set) {
+		memcpy(out, &vgroup->tmplt_elusage, sizeof(*out));
+		WARN_ON(!mlxsw_afk_element_usage_subset(elusage, out));
+		return;
+	}
+
+	for (i = 0; i < vgroup->patterns_count; i++) {
+		pattern = &vgroup->patterns[i];
+		mlxsw_afk_element_usage_fill(out, pattern->elements,
+					     pattern->elements_count);
+		if (mlxsw_afk_element_usage_subset(elusage, out))
+			return;
+	}
+	memcpy(out, elusage, sizeof(*out));
+}
+
+static int
+mlxsw_sp_acl_tcam_region_alloc(struct mlxsw_sp *mlxsw_sp,
+			       struct mlxsw_sp_acl_tcam_region *region)
+{
+	struct mlxsw_afk_key_info *key_info = region->key_info;
+	char ptar_pl[MLXSW_REG_PTAR_LEN];
+	unsigned int encodings_count;
+	int i;
+	int err;
+
+	mlxsw_reg_ptar_pack(ptar_pl, MLXSW_REG_PTAR_OP_ALLOC,
+			    region->key_type,
+			    MLXSW_SP_ACL_TCAM_REGION_BASE_COUNT,
+			    region->id, region->tcam_region_info);
+	encodings_count = mlxsw_afk_key_info_blocks_count_get(key_info);
+	for (i = 0; i < encodings_count; i++) {
+		u16 encoding;
+
+		encoding = mlxsw_afk_key_info_block_encoding_get(key_info, i);
+		mlxsw_reg_ptar_key_id_pack(ptar_pl, i, encoding);
+	}
+	err = mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(ptar), ptar_pl);
+	if (err)
+		return err;
+	mlxsw_reg_ptar_unpack(ptar_pl, region->tcam_region_info);
+	return 0;
+}
+
+static void
+mlxsw_sp_acl_tcam_region_free(struct mlxsw_sp *mlxsw_sp,
+			      struct mlxsw_sp_acl_tcam_region *region)
+{
+	char ptar_pl[MLXSW_REG_PTAR_LEN];
+
+	mlxsw_reg_ptar_pack(ptar_pl, MLXSW_REG_PTAR_OP_FREE,
+			    region->key_type, 0, region->id,
+			    region->tcam_region_info);
+	mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(ptar), ptar_pl);
+}
+
+static int
+mlxsw_sp_acl_tcam_region_enable(struct mlxsw_sp *mlxsw_sp,
+				struct mlxsw_sp_acl_tcam_region *region)
+{
+	char pacl_pl[MLXSW_REG_PACL_LEN];
+
+	mlxsw_reg_pacl_pack(pacl_pl, region->id, true,
+			    region->tcam_region_info);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(pacl), pacl_pl);
+}
+
+static void
+mlxsw_sp_acl_tcam_region_disable(struct mlxsw_sp *mlxsw_sp,
+				 struct mlxsw_sp_acl_tcam_region *region)
+{
+	char pacl_pl[MLXSW_REG_PACL_LEN];
+
+	mlxsw_reg_pacl_pack(pacl_pl, region->id, false,
+			    region->tcam_region_info);
+	mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(pacl), pacl_pl);
+}
+
+static struct mlxsw_sp_acl_tcam_region *
+mlxsw_sp_acl_tcam_region_create(struct mlxsw_sp *mlxsw_sp,
+				struct mlxsw_sp_acl_tcam *tcam,
+				struct mlxsw_sp_acl_tcam_vregion *vregion,
+				void *hints_priv)
+{
+	const struct mlxsw_sp_acl_tcam_ops *ops = mlxsw_sp->acl_tcam_ops;
+	struct mlxsw_sp_acl_tcam_region *region;
+	int err;
+
+	region = kzalloc(sizeof(*region) + ops->region_priv_size, GFP_KERNEL);
+	if (!region)
+		return ERR_PTR(-ENOMEM);
+	region->mlxsw_sp = mlxsw_sp;
+	region->vregion = vregion;
+	region->key_info = vregion->key_info;
+
+	err = mlxsw_sp_acl_tcam_region_id_get(tcam, &region->id);
+	if (err)
+		goto err_region_id_get;
+
+	err = ops->region_associate(mlxsw_sp, region);
+	if (err)
+		goto err_tcam_region_associate;
+
+	region->key_type = ops->key_type;
+	err = mlxsw_sp_acl_tcam_region_alloc(mlxsw_sp, region);
+	if (err)
+		goto err_tcam_region_alloc;
+
+	err = mlxsw_sp_acl_tcam_region_enable(mlxsw_sp, region);
+	if (err)
+		goto err_tcam_region_enable;
+
+	err = ops->region_init(mlxsw_sp, region->priv, tcam->priv,
+			       region, hints_priv);
+	if (err)
+		goto err_tcam_region_init;
+
+	return region;
+
+err_tcam_region_init:
+	mlxsw_sp_acl_tcam_region_disable(mlxsw_sp, region);
+err_tcam_region_enable:
+	mlxsw_sp_acl_tcam_region_free(mlxsw_sp, region);
+err_tcam_region_alloc:
+err_tcam_region_associate:
+	mlxsw_sp_acl_tcam_region_id_put(tcam, region->id);
+err_region_id_get:
+	kfree(region);
+	return ERR_PTR(err);
+}
+
+static void
+mlxsw_sp_acl_tcam_region_destroy(struct mlxsw_sp *mlxsw_sp,
+				 struct mlxsw_sp_acl_tcam_region *region)
+{
+	const struct mlxsw_sp_acl_tcam_ops *ops = mlxsw_sp->acl_tcam_ops;
+
+	ops->region_fini(mlxsw_sp, region->priv);
+	mlxsw_sp_acl_tcam_region_disable(mlxsw_sp, region);
+	mlxsw_sp_acl_tcam_region_free(mlxsw_sp, region);
+	mlxsw_sp_acl_tcam_region_id_put(region->group->tcam,
+					region->id);
+	kfree(region);
+}
+
+static void
+mlxsw_sp_acl_tcam_vregion_rehash_work_schedule(struct mlxsw_sp_acl_tcam_vregion *vregion)
+{
+	unsigned long interval = vregion->tcam->vregion_rehash_intrvl;
+
+	if (!interval)
+		return;
+	mlxsw_core_schedule_dw(&vregion->rehash.dw,
+			       msecs_to_jiffies(interval));
+}
+
+static void
+mlxsw_sp_acl_tcam_vregion_rehash(struct mlxsw_sp *mlxsw_sp,
+				 struct mlxsw_sp_acl_tcam_vregion *vregion,
+				 int *credits);
+
+static void mlxsw_sp_acl_tcam_vregion_rehash_work(struct work_struct *work)
+{
+	struct mlxsw_sp_acl_tcam_vregion *vregion =
+		container_of(work, struct mlxsw_sp_acl_tcam_vregion,
+			     rehash.dw.work);
+	int credits = MLXSW_SP_ACL_TCAM_VREGION_REHASH_CREDITS;
+
+	mlxsw_sp_acl_tcam_vregion_rehash(vregion->mlxsw_sp, vregion, &credits);
+	if (credits < 0)
+		/* Rehash gone out of credits so it was interrupted.
+		 * Schedule the work as soon as possible to continue.
+		 */
+		mlxsw_core_schedule_dw(&vregion->rehash.dw, 0);
+	else
+		mlxsw_sp_acl_tcam_vregion_rehash_work_schedule(vregion);
+}
+
+static void
+mlxsw_sp_acl_tcam_rehash_ctx_vchunk_changed(struct mlxsw_sp_acl_tcam_vchunk *vchunk)
+{
+	struct mlxsw_sp_acl_tcam_vregion *vregion = vchunk->vregion;
+
+	/* If a rule was added or deleted from vchunk which is currently
+	 * under rehash migration, we have to reset the ventry pointers
+	 * to make sure all rules are properly migrated.
+	 */
+	if (vregion->rehash.ctx.current_vchunk == vchunk) {
+		vregion->rehash.ctx.start_ventry = NULL;
+		vregion->rehash.ctx.stop_ventry = NULL;
+	}
+}
+
+static void
+mlxsw_sp_acl_tcam_rehash_ctx_vregion_changed(struct mlxsw_sp_acl_tcam_vregion *vregion)
+{
+	/* If a chunk was added or deleted from vregion we have to reset
+	 * the current chunk pointer to make sure all chunks
+	 * are properly migrated.
+	 */
+	vregion->rehash.ctx.current_vchunk = NULL;
+}
+
+static struct mlxsw_sp_acl_tcam_vregion *
+mlxsw_sp_acl_tcam_vregion_create(struct mlxsw_sp *mlxsw_sp,
+				 struct mlxsw_sp_acl_tcam_vgroup *vgroup,
+				 unsigned int priority,
+				 struct mlxsw_afk_element_usage *elusage)
+{
+	const struct mlxsw_sp_acl_tcam_ops *ops = mlxsw_sp->acl_tcam_ops;
+	struct mlxsw_afk *afk = mlxsw_sp_acl_afk(mlxsw_sp->acl);
+	struct mlxsw_sp_acl_tcam *tcam = vgroup->group.tcam;
+	struct mlxsw_sp_acl_tcam_vregion *vregion;
+	int err;
+
+	vregion = kzalloc(sizeof(*vregion), GFP_KERNEL);
+	if (!vregion)
+		return ERR_PTR(-ENOMEM);
+	INIT_LIST_HEAD(&vregion->vchunk_list);
+	mutex_init(&vregion->lock);
+	vregion->tcam = tcam;
+	vregion->mlxsw_sp = mlxsw_sp;
+	vregion->vgroup = vgroup;
+	vregion->ref_count = 1;
+
+	vregion->key_info = mlxsw_afk_key_info_get(afk, elusage);
+	if (IS_ERR(vregion->key_info)) {
+		err = PTR_ERR(vregion->key_info);
+		goto err_key_info_get;
+	}
+
+	vregion->region = mlxsw_sp_acl_tcam_region_create(mlxsw_sp, tcam,
+							  vregion, NULL);
+	if (IS_ERR(vregion->region)) {
+		err = PTR_ERR(vregion->region);
+		goto err_region_create;
+	}
+
+	err = mlxsw_sp_acl_tcam_vgroup_vregion_attach(mlxsw_sp, vgroup, vregion,
+						      priority);
+	if (err)
+		goto err_vgroup_vregion_attach;
+
+	if (vgroup->vregion_rehash_enabled && ops->region_rehash_hints_get) {
+		/* Create the delayed work for vregion periodic rehash */
+		INIT_DELAYED_WORK(&vregion->rehash.dw,
+				  mlxsw_sp_acl_tcam_vregion_rehash_work);
+		mlxsw_sp_acl_tcam_vregion_rehash_work_schedule(vregion);
+		mutex_lock(&tcam->lock);
+		list_add_tail(&vregion->tlist, &tcam->vregion_list);
+		mutex_unlock(&tcam->lock);
+	}
+
+	return vregion;
+
+err_vgroup_vregion_attach:
+	mlxsw_sp_acl_tcam_region_destroy(mlxsw_sp, vregion->region);
+err_region_create:
+	mlxsw_afk_key_info_put(vregion->key_info);
+err_key_info_get:
+	kfree(vregion);
+	return ERR_PTR(err);
+}
+
+static void
+mlxsw_sp_acl_tcam_vregion_destroy(struct mlxsw_sp *mlxsw_sp,
+				  struct mlxsw_sp_acl_tcam_vregion *vregion)
+{
+	const struct mlxsw_sp_acl_tcam_ops *ops = mlxsw_sp->acl_tcam_ops;
+	struct mlxsw_sp_acl_tcam_vgroup *vgroup = vregion->vgroup;
+	struct mlxsw_sp_acl_tcam *tcam = vregion->tcam;
+
+	if (vgroup->vregion_rehash_enabled && ops->region_rehash_hints_get) {
+		mutex_lock(&tcam->lock);
+		list_del(&vregion->tlist);
+		mutex_unlock(&tcam->lock);
+		cancel_delayed_work_sync(&vregion->rehash.dw);
+	}
+	mlxsw_sp_acl_tcam_vgroup_vregion_detach(mlxsw_sp, vregion);
+	if (vregion->region2)
+		mlxsw_sp_acl_tcam_region_destroy(mlxsw_sp, vregion->region2);
+	mlxsw_sp_acl_tcam_region_destroy(mlxsw_sp, vregion->region);
+	mlxsw_afk_key_info_put(vregion->key_info);
+	mutex_destroy(&vregion->lock);
+	kfree(vregion);
+}
+
+u32 mlxsw_sp_acl_tcam_vregion_rehash_intrvl_get(struct mlxsw_sp *mlxsw_sp,
+						struct mlxsw_sp_acl_tcam *tcam)
+{
+	const struct mlxsw_sp_acl_tcam_ops *ops = mlxsw_sp->acl_tcam_ops;
+	u32 vregion_rehash_intrvl;
+
+	if (WARN_ON(!ops->region_rehash_hints_get))
+		return 0;
+	vregion_rehash_intrvl = tcam->vregion_rehash_intrvl;
+	return vregion_rehash_intrvl;
+}
+
+int mlxsw_sp_acl_tcam_vregion_rehash_intrvl_set(struct mlxsw_sp *mlxsw_sp,
+						struct mlxsw_sp_acl_tcam *tcam,
+						u32 val)
+{
+	const struct mlxsw_sp_acl_tcam_ops *ops = mlxsw_sp->acl_tcam_ops;
+	struct mlxsw_sp_acl_tcam_vregion *vregion;
+
+	if (val < MLXSW_SP_ACL_TCAM_VREGION_REHASH_INTRVL_MIN && val)
+		return -EINVAL;
+	if (WARN_ON(!ops->region_rehash_hints_get))
+		return -EOPNOTSUPP;
+	tcam->vregion_rehash_intrvl = val;
+	mutex_lock(&tcam->lock);
+	list_for_each_entry(vregion, &tcam->vregion_list, tlist) {
+		if (val)
+			mlxsw_core_schedule_dw(&vregion->rehash.dw, 0);
+		else
+			cancel_delayed_work_sync(&vregion->rehash.dw);
+	}
+	mutex_unlock(&tcam->lock);
+	return 0;
+}
+
+static struct mlxsw_sp_acl_tcam_vregion *
+mlxsw_sp_acl_tcam_vregion_get(struct mlxsw_sp *mlxsw_sp,
+			      struct mlxsw_sp_acl_tcam_vgroup *vgroup,
+			      unsigned int priority,
+			      struct mlxsw_afk_element_usage *elusage)
+{
+	struct mlxsw_afk_element_usage vregion_elusage;
+	struct mlxsw_sp_acl_tcam_vregion *vregion;
+	bool need_split;
+
+	vregion = mlxsw_sp_acl_tcam_vgroup_vregion_find(vgroup, priority,
+							elusage, &need_split);
+	if (vregion) {
+		if (need_split) {
+			/* According to priority, new vchunk should belong to
+			 * an existing vregion. However, this vchunk needs
+			 * elements that vregion does not contain. We need
+			 * to split the existing vregion into two and create
+			 * a new vregion for the new vchunk in between.
+			 * This is not supported now.
+			 */
+			return ERR_PTR(-EOPNOTSUPP);
+		}
+		vregion->ref_count++;
+		return vregion;
+	}
+
+	mlxsw_sp_acl_tcam_vgroup_use_patterns(vgroup, elusage,
+					      &vregion_elusage);
+
+	return mlxsw_sp_acl_tcam_vregion_create(mlxsw_sp, vgroup, priority,
+						&vregion_elusage);
+}
+
+static void
+mlxsw_sp_acl_tcam_vregion_put(struct mlxsw_sp *mlxsw_sp,
+			      struct mlxsw_sp_acl_tcam_vregion *vregion)
+{
+	if (--vregion->ref_count)
+		return;
+	mlxsw_sp_acl_tcam_vregion_destroy(mlxsw_sp, vregion);
+}
+
+static struct mlxsw_sp_acl_tcam_chunk *
+mlxsw_sp_acl_tcam_chunk_create(struct mlxsw_sp *mlxsw_sp,
+			       struct mlxsw_sp_acl_tcam_vchunk *vchunk,
+			       struct mlxsw_sp_acl_tcam_region *region)
+{
+	const struct mlxsw_sp_acl_tcam_ops *ops = mlxsw_sp->acl_tcam_ops;
+	struct mlxsw_sp_acl_tcam_chunk *chunk;
+
+	chunk = kzalloc(sizeof(*chunk) + ops->chunk_priv_size, GFP_KERNEL);
+	if (!chunk)
+		return ERR_PTR(-ENOMEM);
+	chunk->vchunk = vchunk;
+	chunk->region = region;
+
+	ops->chunk_init(region->priv, chunk->priv, vchunk->priority);
+	return chunk;
+}
+
+static void
+mlxsw_sp_acl_tcam_chunk_destroy(struct mlxsw_sp *mlxsw_sp,
+				struct mlxsw_sp_acl_tcam_chunk *chunk)
+{
+	const struct mlxsw_sp_acl_tcam_ops *ops = mlxsw_sp->acl_tcam_ops;
+
+	ops->chunk_fini(chunk->priv);
+	kfree(chunk);
+}
+
+static struct mlxsw_sp_acl_tcam_vchunk *
+mlxsw_sp_acl_tcam_vchunk_create(struct mlxsw_sp *mlxsw_sp,
+				struct mlxsw_sp_acl_tcam_vgroup *vgroup,
+				unsigned int priority,
+				struct mlxsw_afk_element_usage *elusage)
+{
+	struct mlxsw_sp_acl_tcam_vchunk *vchunk, *vchunk2;
+	struct mlxsw_sp_acl_tcam_vregion *vregion;
+	struct list_head *pos;
+	int err;
+
+	if (priority == MLXSW_SP_ACL_TCAM_CATCHALL_PRIO)
+		return ERR_PTR(-EINVAL);
+
+	vchunk = kzalloc(sizeof(*vchunk), GFP_KERNEL);
+	if (!vchunk)
+		return ERR_PTR(-ENOMEM);
+	INIT_LIST_HEAD(&vchunk->ventry_list);
+	vchunk->priority = priority;
+	vchunk->vgroup = vgroup;
+	vchunk->ref_count = 1;
+
+	vregion = mlxsw_sp_acl_tcam_vregion_get(mlxsw_sp, vgroup,
+						priority, elusage);
+	if (IS_ERR(vregion)) {
+		err = PTR_ERR(vregion);
+		goto err_vregion_get;
+	}
+
+	vchunk->vregion = vregion;
+
+	err = rhashtable_insert_fast(&vgroup->vchunk_ht, &vchunk->ht_node,
+				     mlxsw_sp_acl_tcam_vchunk_ht_params);
+	if (err)
+		goto err_rhashtable_insert;
+
+	mutex_lock(&vregion->lock);
+	vchunk->chunk = mlxsw_sp_acl_tcam_chunk_create(mlxsw_sp, vchunk,
+						       vchunk->vregion->region);
+	if (IS_ERR(vchunk->chunk)) {
+		mutex_unlock(&vregion->lock);
+		err = PTR_ERR(vchunk->chunk);
+		goto err_chunk_create;
+	}
+
+	mlxsw_sp_acl_tcam_rehash_ctx_vregion_changed(vregion);
+
+	/* Position the vchunk inside the list according to priority */
+	list_for_each(pos, &vregion->vchunk_list) {
+		vchunk2 = list_entry(pos, typeof(*vchunk2), list);
+		if (vchunk2->priority > priority)
+			break;
+	}
+	list_add_tail(&vchunk->list, pos);
+	mutex_unlock(&vregion->lock);
+	mlxsw_sp_acl_tcam_vgroup_prio_update(vgroup);
+
+	return vchunk;
+
+err_chunk_create:
+	rhashtable_remove_fast(&vgroup->vchunk_ht, &vchunk->ht_node,
+			       mlxsw_sp_acl_tcam_vchunk_ht_params);
+err_rhashtable_insert:
+	mlxsw_sp_acl_tcam_vregion_put(mlxsw_sp, vregion);
+err_vregion_get:
+	kfree(vchunk);
+	return ERR_PTR(err);
+}
+
+static void
+mlxsw_sp_acl_tcam_vchunk_destroy(struct mlxsw_sp *mlxsw_sp,
+				 struct mlxsw_sp_acl_tcam_vchunk *vchunk)
+{
+	struct mlxsw_sp_acl_tcam_vregion *vregion = vchunk->vregion;
+	struct mlxsw_sp_acl_tcam_vgroup *vgroup = vchunk->vgroup;
+
+	mutex_lock(&vregion->lock);
+	mlxsw_sp_acl_tcam_rehash_ctx_vregion_changed(vregion);
+	list_del(&vchunk->list);
+	if (vchunk->chunk2)
+		mlxsw_sp_acl_tcam_chunk_destroy(mlxsw_sp, vchunk->chunk2);
+	mlxsw_sp_acl_tcam_chunk_destroy(mlxsw_sp, vchunk->chunk);
+	mutex_unlock(&vregion->lock);
+	rhashtable_remove_fast(&vgroup->vchunk_ht, &vchunk->ht_node,
+			       mlxsw_sp_acl_tcam_vchunk_ht_params);
+	mlxsw_sp_acl_tcam_vregion_put(mlxsw_sp, vchunk->vregion);
+	kfree(vchunk);
+	mlxsw_sp_acl_tcam_vgroup_prio_update(vgroup);
+}
+
+static struct mlxsw_sp_acl_tcam_vchunk *
+mlxsw_sp_acl_tcam_vchunk_get(struct mlxsw_sp *mlxsw_sp,
+			     struct mlxsw_sp_acl_tcam_vgroup *vgroup,
+			     unsigned int priority,
+			     struct mlxsw_afk_element_usage *elusage)
+{
+	struct mlxsw_sp_acl_tcam_vchunk *vchunk;
+
+	vchunk = rhashtable_lookup_fast(&vgroup->vchunk_ht, &priority,
+					mlxsw_sp_acl_tcam_vchunk_ht_params);
+	if (vchunk) {
+		if (WARN_ON(!mlxsw_afk_key_info_subset(vchunk->vregion->key_info,
+						       elusage)))
+			return ERR_PTR(-EINVAL);
+		vchunk->ref_count++;
+		return vchunk;
+	}
+	return mlxsw_sp_acl_tcam_vchunk_create(mlxsw_sp, vgroup,
+					       priority, elusage);
+}
+
+static void
+mlxsw_sp_acl_tcam_vchunk_put(struct mlxsw_sp *mlxsw_sp,
+			     struct mlxsw_sp_acl_tcam_vchunk *vchunk)
+{
+	if (--vchunk->ref_count)
+		return;
+	mlxsw_sp_acl_tcam_vchunk_destroy(mlxsw_sp, vchunk);
+}
+
+static struct mlxsw_sp_acl_tcam_entry *
+mlxsw_sp_acl_tcam_entry_create(struct mlxsw_sp *mlxsw_sp,
+			       struct mlxsw_sp_acl_tcam_ventry *ventry,
+			       struct mlxsw_sp_acl_tcam_chunk *chunk)
+{
+	const struct mlxsw_sp_acl_tcam_ops *ops = mlxsw_sp->acl_tcam_ops;
+	struct mlxsw_sp_acl_tcam_entry *entry;
+	int err;
+
+	entry = kzalloc(sizeof(*entry) + ops->entry_priv_size, GFP_KERNEL);
+	if (!entry)
+		return ERR_PTR(-ENOMEM);
+	entry->ventry = ventry;
+	entry->chunk = chunk;
+
+	err = ops->entry_add(mlxsw_sp, chunk->region->priv, chunk->priv,
+			     entry->priv, ventry->rulei);
+	if (err)
+		goto err_entry_add;
+
+	return entry;
+
+err_entry_add:
+	kfree(entry);
+	return ERR_PTR(err);
+}
+
+static void mlxsw_sp_acl_tcam_entry_destroy(struct mlxsw_sp *mlxsw_sp,
+					    struct mlxsw_sp_acl_tcam_entry *entry)
+{
+	const struct mlxsw_sp_acl_tcam_ops *ops = mlxsw_sp->acl_tcam_ops;
+
+	ops->entry_del(mlxsw_sp, entry->chunk->region->priv,
+		       entry->chunk->priv, entry->priv);
+	kfree(entry);
+}
+
+static int
+mlxsw_sp_acl_tcam_entry_action_replace(struct mlxsw_sp *mlxsw_sp,
+				       struct mlxsw_sp_acl_tcam_region *region,
+				       struct mlxsw_sp_acl_tcam_entry *entry,
+				       struct mlxsw_sp_acl_rule_info *rulei)
+{
+	const struct mlxsw_sp_acl_tcam_ops *ops = mlxsw_sp->acl_tcam_ops;
+
+	return ops->entry_action_replace(mlxsw_sp, region->priv,
+					 entry->priv, rulei);
+}
+
+static int
+mlxsw_sp_acl_tcam_entry_activity_get(struct mlxsw_sp *mlxsw_sp,
+				     struct mlxsw_sp_acl_tcam_entry *entry,
+				     bool *activity)
+{
+	const struct mlxsw_sp_acl_tcam_ops *ops = mlxsw_sp->acl_tcam_ops;
+
+	return ops->entry_activity_get(mlxsw_sp, entry->chunk->region->priv,
+				       entry->priv, activity);
+}
+
+static int mlxsw_sp_acl_tcam_ventry_add(struct mlxsw_sp *mlxsw_sp,
+					struct mlxsw_sp_acl_tcam_vgroup *vgroup,
+					struct mlxsw_sp_acl_tcam_ventry *ventry,
+					struct mlxsw_sp_acl_rule_info *rulei)
+{
+	struct mlxsw_sp_acl_tcam_vregion *vregion;
+	struct mlxsw_sp_acl_tcam_vchunk *vchunk;
+	int err;
+
+	vchunk = mlxsw_sp_acl_tcam_vchunk_get(mlxsw_sp, vgroup, rulei->priority,
+					      &rulei->values.elusage);
+	if (IS_ERR(vchunk))
+		return PTR_ERR(vchunk);
+
+	ventry->vchunk = vchunk;
+	ventry->rulei = rulei;
+	vregion = vchunk->vregion;
+
+	mutex_lock(&vregion->lock);
+	ventry->entry = mlxsw_sp_acl_tcam_entry_create(mlxsw_sp, ventry,
+						       vchunk->chunk);
+	if (IS_ERR(ventry->entry)) {
+		mutex_unlock(&vregion->lock);
+		err = PTR_ERR(ventry->entry);
+		goto err_entry_create;
+	}
+
+	list_add_tail(&ventry->list, &vchunk->ventry_list);
+	mlxsw_sp_acl_tcam_rehash_ctx_vchunk_changed(vchunk);
+	mutex_unlock(&vregion->lock);
+
+	return 0;
+
+err_entry_create:
+	mlxsw_sp_acl_tcam_vchunk_put(mlxsw_sp, vchunk);
+	return err;
+}
+
+static void mlxsw_sp_acl_tcam_ventry_del(struct mlxsw_sp *mlxsw_sp,
+					 struct mlxsw_sp_acl_tcam_ventry *ventry)
+{
+	struct mlxsw_sp_acl_tcam_vchunk *vchunk = ventry->vchunk;
+	struct mlxsw_sp_acl_tcam_vregion *vregion = vchunk->vregion;
+
+	mutex_lock(&vregion->lock);
+	mlxsw_sp_acl_tcam_rehash_ctx_vchunk_changed(vchunk);
+	list_del(&ventry->list);
+	mlxsw_sp_acl_tcam_entry_destroy(mlxsw_sp, ventry->entry);
+	mutex_unlock(&vregion->lock);
+	mlxsw_sp_acl_tcam_vchunk_put(mlxsw_sp, vchunk);
+}
+
+static int
+mlxsw_sp_acl_tcam_ventry_action_replace(struct mlxsw_sp *mlxsw_sp,
+					struct mlxsw_sp_acl_tcam_ventry *ventry,
+					struct mlxsw_sp_acl_rule_info *rulei)
+{
+	struct mlxsw_sp_acl_tcam_vchunk *vchunk = ventry->vchunk;
+
+	return mlxsw_sp_acl_tcam_entry_action_replace(mlxsw_sp,
+						      vchunk->vregion->region,
+						      ventry->entry, rulei);
+}
+
+static int
+mlxsw_sp_acl_tcam_ventry_activity_get(struct mlxsw_sp *mlxsw_sp,
+				      struct mlxsw_sp_acl_tcam_ventry *ventry,
+				      bool *activity)
+{
+	return mlxsw_sp_acl_tcam_entry_activity_get(mlxsw_sp,
+						    ventry->entry, activity);
+}
+
+static int
+mlxsw_sp_acl_tcam_ventry_migrate(struct mlxsw_sp *mlxsw_sp,
+				 struct mlxsw_sp_acl_tcam_ventry *ventry,
+				 struct mlxsw_sp_acl_tcam_chunk *chunk,
+				 int *credits)
+{
+	struct mlxsw_sp_acl_tcam_entry *new_entry;
+
+	/* First check if the entry is not already where we want it to be. */
+	if (ventry->entry->chunk == chunk)
+		return 0;
+
+	if (--(*credits) < 0)
+		return 0;
+
+	new_entry = mlxsw_sp_acl_tcam_entry_create(mlxsw_sp, ventry, chunk);
+	if (IS_ERR(new_entry))
+		return PTR_ERR(new_entry);
+	mlxsw_sp_acl_tcam_entry_destroy(mlxsw_sp, ventry->entry);
+	ventry->entry = new_entry;
+	return 0;
+}
+
+static int
+mlxsw_sp_acl_tcam_vchunk_migrate_start(struct mlxsw_sp *mlxsw_sp,
+				       struct mlxsw_sp_acl_tcam_vchunk *vchunk,
+				       struct mlxsw_sp_acl_tcam_region *region,
+				       struct mlxsw_sp_acl_tcam_rehash_ctx *ctx)
+{
+	struct mlxsw_sp_acl_tcam_chunk *new_chunk;
+
+	new_chunk = mlxsw_sp_acl_tcam_chunk_create(mlxsw_sp, vchunk, region);
+	if (IS_ERR(new_chunk))
+		return PTR_ERR(new_chunk);
+	vchunk->chunk2 = vchunk->chunk;
+	vchunk->chunk = new_chunk;
+	ctx->current_vchunk = vchunk;
+	ctx->start_ventry = NULL;
+	ctx->stop_ventry = NULL;
+	return 0;
+}
+
+static void
+mlxsw_sp_acl_tcam_vchunk_migrate_end(struct mlxsw_sp *mlxsw_sp,
+				     struct mlxsw_sp_acl_tcam_vchunk *vchunk,
+				     struct mlxsw_sp_acl_tcam_rehash_ctx *ctx)
+{
+	mlxsw_sp_acl_tcam_chunk_destroy(mlxsw_sp, vchunk->chunk2);
+	vchunk->chunk2 = NULL;
+	ctx->current_vchunk = NULL;
+}
+
+static int
+mlxsw_sp_acl_tcam_vchunk_migrate_one(struct mlxsw_sp *mlxsw_sp,
+				     struct mlxsw_sp_acl_tcam_vchunk *vchunk,
+				     struct mlxsw_sp_acl_tcam_region *region,
+				     struct mlxsw_sp_acl_tcam_rehash_ctx *ctx,
+				     int *credits)
+{
+	struct mlxsw_sp_acl_tcam_ventry *ventry;
+	int err;
+
+	if (vchunk->chunk->region != region) {
+		err = mlxsw_sp_acl_tcam_vchunk_migrate_start(mlxsw_sp, vchunk,
+							     region, ctx);
+		if (err)
+			return err;
+	} else if (!vchunk->chunk2) {
+		/* The chunk is already as it should be, nothing to do. */
+		return 0;
+	}
+
+	/* If the migration got interrupted, we have the ventry to start from
+	 * stored in context.
+	 */
+	if (ctx->start_ventry)
+		ventry = ctx->start_ventry;
+	else
+		ventry = list_first_entry(&vchunk->ventry_list,
+					  typeof(*ventry), list);
+
+	list_for_each_entry_from(ventry, &vchunk->ventry_list, list) {
+		/* During rollback, once we reach the ventry that failed
+		 * to migrate, we are done.
+		 */
+		if (ventry == ctx->stop_ventry)
+			break;
+
+		err = mlxsw_sp_acl_tcam_ventry_migrate(mlxsw_sp, ventry,
+						       vchunk->chunk, credits);
+		if (err) {
+			if (ctx->this_is_rollback) {
+				/* Save the ventry which we ended with and try
+				 * to continue later on.
+				 */
+				ctx->start_ventry = ventry;
+				return err;
+			}
+			/* Swap the chunk and chunk2 pointers so the follow-up
+			 * rollback call will see the original chunk pointer
+			 * in vchunk->chunk.
+			 */
+			swap(vchunk->chunk, vchunk->chunk2);
+			/* The rollback has to be done from beginning of the
+			 * chunk, that is why we have to null the start_ventry.
+			 * However, we know where to stop the rollback,
+			 * at the current ventry.
+			 */
+			ctx->start_ventry = NULL;
+			ctx->stop_ventry = ventry;
+			return err;
+		} else if (*credits < 0) {
+			/* We are out of credits, the rest of the ventries
+			 * will be migrated later. Save the ventry
+			 * which we ended with.
+			 */
+			ctx->start_ventry = ventry;
+			return 0;
+		}
+	}
+
+	mlxsw_sp_acl_tcam_vchunk_migrate_end(mlxsw_sp, vchunk, ctx);
+	return 0;
+}
+
+static int
+mlxsw_sp_acl_tcam_vchunk_migrate_all(struct mlxsw_sp *mlxsw_sp,
+				     struct mlxsw_sp_acl_tcam_vregion *vregion,
+				     struct mlxsw_sp_acl_tcam_rehash_ctx *ctx,
+				     int *credits)
+{
+	struct mlxsw_sp_acl_tcam_vchunk *vchunk;
+	int err;
+
+	/* If the migration got interrupted, we have the vchunk
+	 * we are working on stored in context.
+	 */
+	if (ctx->current_vchunk)
+		vchunk = ctx->current_vchunk;
+	else
+		vchunk = list_first_entry(&vregion->vchunk_list,
+					  typeof(*vchunk), list);
+
+	list_for_each_entry_from(vchunk, &vregion->vchunk_list, list) {
+		err = mlxsw_sp_acl_tcam_vchunk_migrate_one(mlxsw_sp, vchunk,
+							   vregion->region,
+							   ctx, credits);
+		if (err || *credits < 0)
+			return err;
+	}
+	return 0;
+}
+
+static int
+mlxsw_sp_acl_tcam_vregion_migrate(struct mlxsw_sp *mlxsw_sp,
+				  struct mlxsw_sp_acl_tcam_vregion *vregion,
+				  struct mlxsw_sp_acl_tcam_rehash_ctx *ctx,
+				  int *credits)
+{
+	int err, err2;
+
+	trace_mlxsw_sp_acl_tcam_vregion_migrate(mlxsw_sp, vregion);
+	mutex_lock(&vregion->lock);
+	err = mlxsw_sp_acl_tcam_vchunk_migrate_all(mlxsw_sp, vregion,
+						   ctx, credits);
+	if (err) {
+		/* In case migration was not successful, we need to swap
+		 * so the original region pointer is assigned again
+		 * to vregion->region.
+		 */
+		swap(vregion->region, vregion->region2);
+		ctx->current_vchunk = NULL;
+		ctx->this_is_rollback = true;
+		err2 = mlxsw_sp_acl_tcam_vchunk_migrate_all(mlxsw_sp, vregion,
+							    ctx, credits);
+		if (err2) {
+			trace_mlxsw_sp_acl_tcam_vregion_rehash_rollback_failed(mlxsw_sp,
+									       vregion);
+			dev_err(mlxsw_sp->bus_info->dev, "Failed to rollback during vregion migration fail\n");
+			/* Let the rollback to be continued later on. */
+		}
+	}
+	mutex_unlock(&vregion->lock);
+	trace_mlxsw_sp_acl_tcam_vregion_migrate_end(mlxsw_sp, vregion);
+	return err;
+}
+
+static bool
+mlxsw_sp_acl_tcam_vregion_rehash_in_progress(const struct mlxsw_sp_acl_tcam_rehash_ctx *ctx)
+{
+	return ctx->hints_priv;
+}
+
+static int
+mlxsw_sp_acl_tcam_vregion_rehash_start(struct mlxsw_sp *mlxsw_sp,
+				       struct mlxsw_sp_acl_tcam_vregion *vregion,
+				       struct mlxsw_sp_acl_tcam_rehash_ctx *ctx)
+{
+	const struct mlxsw_sp_acl_tcam_ops *ops = mlxsw_sp->acl_tcam_ops;
+	unsigned int priority = mlxsw_sp_acl_tcam_vregion_prio(vregion);
+	struct mlxsw_sp_acl_tcam_region *new_region;
+	void *hints_priv;
+	int err;
+
+	trace_mlxsw_sp_acl_tcam_vregion_rehash(mlxsw_sp, vregion);
+
+	hints_priv = ops->region_rehash_hints_get(vregion->region->priv);
+	if (IS_ERR(hints_priv))
+		return PTR_ERR(hints_priv);
+
+	new_region = mlxsw_sp_acl_tcam_region_create(mlxsw_sp, vregion->tcam,
+						     vregion, hints_priv);
+	if (IS_ERR(new_region)) {
+		err = PTR_ERR(new_region);
+		goto err_region_create;
+	}
+
+	/* vregion->region contains the pointer to the new region
+	 * we are going to migrate to.
+	 */
+	vregion->region2 = vregion->region;
+	vregion->region = new_region;
+	err = mlxsw_sp_acl_tcam_group_region_attach(mlxsw_sp,
+						    vregion->region2->group,
+						    new_region, priority,
+						    vregion->region2);
+	if (err)
+		goto err_group_region_attach;
+
+	ctx->hints_priv = hints_priv;
+	ctx->this_is_rollback = false;
+
+	return 0;
+
+err_group_region_attach:
+	vregion->region = vregion->region2;
+	vregion->region2 = NULL;
+	mlxsw_sp_acl_tcam_region_destroy(mlxsw_sp, new_region);
+err_region_create:
+	ops->region_rehash_hints_put(hints_priv);
+	return err;
+}
+
+static void
+mlxsw_sp_acl_tcam_vregion_rehash_end(struct mlxsw_sp *mlxsw_sp,
+				     struct mlxsw_sp_acl_tcam_vregion *vregion,
+				     struct mlxsw_sp_acl_tcam_rehash_ctx *ctx)
+{
+	struct mlxsw_sp_acl_tcam_region *unused_region = vregion->region2;
+	const struct mlxsw_sp_acl_tcam_ops *ops = mlxsw_sp->acl_tcam_ops;
+
+	vregion->region2 = NULL;
+	mlxsw_sp_acl_tcam_group_region_detach(mlxsw_sp, unused_region);
+	mlxsw_sp_acl_tcam_region_destroy(mlxsw_sp, unused_region);
+	ops->region_rehash_hints_put(ctx->hints_priv);
+	ctx->hints_priv = NULL;
+}
+
+static void
+mlxsw_sp_acl_tcam_vregion_rehash(struct mlxsw_sp *mlxsw_sp,
+				 struct mlxsw_sp_acl_tcam_vregion *vregion,
+				 int *credits)
+{
+	struct mlxsw_sp_acl_tcam_rehash_ctx *ctx = &vregion->rehash.ctx;
+	int err;
+
+	/* Check if the previous rehash work was interrupted
+	 * which means we have to continue it now.
+	 * If not, start a new rehash.
+	 */
+	if (!mlxsw_sp_acl_tcam_vregion_rehash_in_progress(ctx)) {
+		err = mlxsw_sp_acl_tcam_vregion_rehash_start(mlxsw_sp,
+							     vregion, ctx);
+		if (err) {
+			if (err != -EAGAIN)
+				dev_err(mlxsw_sp->bus_info->dev, "Failed get rehash hints\n");
+			return;
+		}
+	}
+
+	err = mlxsw_sp_acl_tcam_vregion_migrate(mlxsw_sp, vregion,
+						ctx, credits);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Failed to migrate vregion\n");
+	}
+
+	if (*credits >= 0)
+		mlxsw_sp_acl_tcam_vregion_rehash_end(mlxsw_sp, vregion, ctx);
+}
+
+static const enum mlxsw_afk_element mlxsw_sp_acl_tcam_pattern_ipv4[] = {
+	MLXSW_AFK_ELEMENT_SRC_SYS_PORT,
+	MLXSW_AFK_ELEMENT_DMAC_32_47,
+	MLXSW_AFK_ELEMENT_DMAC_0_31,
+	MLXSW_AFK_ELEMENT_SMAC_32_47,
+	MLXSW_AFK_ELEMENT_SMAC_0_31,
+	MLXSW_AFK_ELEMENT_ETHERTYPE,
+	MLXSW_AFK_ELEMENT_IP_PROTO,
+	MLXSW_AFK_ELEMENT_SRC_IP_0_31,
+	MLXSW_AFK_ELEMENT_DST_IP_0_31,
+	MLXSW_AFK_ELEMENT_DST_L4_PORT,
+	MLXSW_AFK_ELEMENT_SRC_L4_PORT,
+	MLXSW_AFK_ELEMENT_VID,
+	MLXSW_AFK_ELEMENT_PCP,
+	MLXSW_AFK_ELEMENT_TCP_FLAGS,
+	MLXSW_AFK_ELEMENT_IP_TTL_,
+	MLXSW_AFK_ELEMENT_IP_ECN,
+	MLXSW_AFK_ELEMENT_IP_DSCP,
+};
+
+static const enum mlxsw_afk_element mlxsw_sp_acl_tcam_pattern_ipv6[] = {
+	MLXSW_AFK_ELEMENT_ETHERTYPE,
+	MLXSW_AFK_ELEMENT_IP_PROTO,
+	MLXSW_AFK_ELEMENT_SRC_IP_96_127,
+	MLXSW_AFK_ELEMENT_SRC_IP_64_95,
+	MLXSW_AFK_ELEMENT_SRC_IP_32_63,
+	MLXSW_AFK_ELEMENT_SRC_IP_0_31,
+	MLXSW_AFK_ELEMENT_DST_IP_96_127,
+	MLXSW_AFK_ELEMENT_DST_IP_64_95,
+	MLXSW_AFK_ELEMENT_DST_IP_32_63,
+	MLXSW_AFK_ELEMENT_DST_IP_0_31,
+	MLXSW_AFK_ELEMENT_DST_L4_PORT,
+	MLXSW_AFK_ELEMENT_SRC_L4_PORT,
+};
+
+static const struct mlxsw_sp_acl_tcam_pattern mlxsw_sp_acl_tcam_patterns[] = {
+	{
+		.elements = mlxsw_sp_acl_tcam_pattern_ipv4,
+		.elements_count = ARRAY_SIZE(mlxsw_sp_acl_tcam_pattern_ipv4),
+	},
+	{
+		.elements = mlxsw_sp_acl_tcam_pattern_ipv6,
+		.elements_count = ARRAY_SIZE(mlxsw_sp_acl_tcam_pattern_ipv6),
+	},
+};
+
+#define MLXSW_SP_ACL_TCAM_PATTERNS_COUNT \
+	ARRAY_SIZE(mlxsw_sp_acl_tcam_patterns)
+
+struct mlxsw_sp_acl_tcam_flower_ruleset {
+	struct mlxsw_sp_acl_tcam_vgroup vgroup;
+};
+
+struct mlxsw_sp_acl_tcam_flower_rule {
+	struct mlxsw_sp_acl_tcam_ventry ventry;
+};
+
+static int
+mlxsw_sp_acl_tcam_flower_ruleset_add(struct mlxsw_sp *mlxsw_sp,
+				     struct mlxsw_sp_acl_tcam *tcam,
+				     void *ruleset_priv,
+				     struct mlxsw_afk_element_usage *tmplt_elusage,
+				     unsigned int *p_min_prio,
+				     unsigned int *p_max_prio)
+{
+	struct mlxsw_sp_acl_tcam_flower_ruleset *ruleset = ruleset_priv;
+
+	return mlxsw_sp_acl_tcam_vgroup_add(mlxsw_sp, tcam, &ruleset->vgroup,
+					    mlxsw_sp_acl_tcam_patterns,
+					    MLXSW_SP_ACL_TCAM_PATTERNS_COUNT,
+					    tmplt_elusage, true,
+					    p_min_prio, p_max_prio);
+}
+
+static void
+mlxsw_sp_acl_tcam_flower_ruleset_del(struct mlxsw_sp *mlxsw_sp,
+				     void *ruleset_priv)
+{
+	struct mlxsw_sp_acl_tcam_flower_ruleset *ruleset = ruleset_priv;
+
+	mlxsw_sp_acl_tcam_vgroup_del(&ruleset->vgroup);
+}
+
+static int
+mlxsw_sp_acl_tcam_flower_ruleset_bind(struct mlxsw_sp *mlxsw_sp,
+				      void *ruleset_priv,
+				      struct mlxsw_sp_port *mlxsw_sp_port,
+				      bool ingress)
+{
+	struct mlxsw_sp_acl_tcam_flower_ruleset *ruleset = ruleset_priv;
+
+	return mlxsw_sp_acl_tcam_group_bind(mlxsw_sp, &ruleset->vgroup.group,
+					    mlxsw_sp_port, ingress);
+}
+
+static void
+mlxsw_sp_acl_tcam_flower_ruleset_unbind(struct mlxsw_sp *mlxsw_sp,
+					void *ruleset_priv,
+					struct mlxsw_sp_port *mlxsw_sp_port,
+					bool ingress)
+{
+	struct mlxsw_sp_acl_tcam_flower_ruleset *ruleset = ruleset_priv;
+
+	mlxsw_sp_acl_tcam_group_unbind(mlxsw_sp, &ruleset->vgroup.group,
+				       mlxsw_sp_port, ingress);
+}
+
+static u16
+mlxsw_sp_acl_tcam_flower_ruleset_group_id(void *ruleset_priv)
+{
+	struct mlxsw_sp_acl_tcam_flower_ruleset *ruleset = ruleset_priv;
+
+	return mlxsw_sp_acl_tcam_group_id(&ruleset->vgroup.group);
+}
+
+static int
+mlxsw_sp_acl_tcam_flower_rule_add(struct mlxsw_sp *mlxsw_sp,
+				  void *ruleset_priv, void *rule_priv,
+				  struct mlxsw_sp_acl_rule_info *rulei)
+{
+	struct mlxsw_sp_acl_tcam_flower_ruleset *ruleset = ruleset_priv;
+	struct mlxsw_sp_acl_tcam_flower_rule *rule = rule_priv;
+
+	return mlxsw_sp_acl_tcam_ventry_add(mlxsw_sp, &ruleset->vgroup,
+					    &rule->ventry, rulei);
+}
+
+static void
+mlxsw_sp_acl_tcam_flower_rule_del(struct mlxsw_sp *mlxsw_sp, void *rule_priv)
+{
+	struct mlxsw_sp_acl_tcam_flower_rule *rule = rule_priv;
+
+	mlxsw_sp_acl_tcam_ventry_del(mlxsw_sp, &rule->ventry);
+}
+
+static int
+mlxsw_sp_acl_tcam_flower_rule_action_replace(struct mlxsw_sp *mlxsw_sp,
+					     void *rule_priv,
+					     struct mlxsw_sp_acl_rule_info *rulei)
+{
+	return -EOPNOTSUPP;
+}
+
+static int
+mlxsw_sp_acl_tcam_flower_rule_activity_get(struct mlxsw_sp *mlxsw_sp,
+					   void *rule_priv, bool *activity)
+{
+	struct mlxsw_sp_acl_tcam_flower_rule *rule = rule_priv;
+
+	return mlxsw_sp_acl_tcam_ventry_activity_get(mlxsw_sp, &rule->ventry,
+						     activity);
+}
+
+static const struct mlxsw_sp_acl_profile_ops mlxsw_sp_acl_tcam_flower_ops = {
+	.ruleset_priv_size	= sizeof(struct mlxsw_sp_acl_tcam_flower_ruleset),
+	.ruleset_add		= mlxsw_sp_acl_tcam_flower_ruleset_add,
+	.ruleset_del		= mlxsw_sp_acl_tcam_flower_ruleset_del,
+	.ruleset_bind		= mlxsw_sp_acl_tcam_flower_ruleset_bind,
+	.ruleset_unbind		= mlxsw_sp_acl_tcam_flower_ruleset_unbind,
+	.ruleset_group_id	= mlxsw_sp_acl_tcam_flower_ruleset_group_id,
+	.rule_priv_size		= sizeof(struct mlxsw_sp_acl_tcam_flower_rule),
+	.rule_add		= mlxsw_sp_acl_tcam_flower_rule_add,
+	.rule_del		= mlxsw_sp_acl_tcam_flower_rule_del,
+	.rule_action_replace	= mlxsw_sp_acl_tcam_flower_rule_action_replace,
+	.rule_activity_get	= mlxsw_sp_acl_tcam_flower_rule_activity_get,
+};
+
+struct mlxsw_sp_acl_tcam_mr_ruleset {
+	struct mlxsw_sp_acl_tcam_vchunk *vchunk;
+	struct mlxsw_sp_acl_tcam_vgroup vgroup;
+};
+
+struct mlxsw_sp_acl_tcam_mr_rule {
+	struct mlxsw_sp_acl_tcam_ventry ventry;
+};
+
+static int
+mlxsw_sp_acl_tcam_mr_ruleset_add(struct mlxsw_sp *mlxsw_sp,
+				 struct mlxsw_sp_acl_tcam *tcam,
+				 void *ruleset_priv,
+				 struct mlxsw_afk_element_usage *tmplt_elusage,
+				 unsigned int *p_min_prio,
+				 unsigned int *p_max_prio)
+{
+	struct mlxsw_sp_acl_tcam_mr_ruleset *ruleset = ruleset_priv;
+	int err;
+
+	err = mlxsw_sp_acl_tcam_vgroup_add(mlxsw_sp, tcam, &ruleset->vgroup,
+					   mlxsw_sp_acl_tcam_patterns,
+					   MLXSW_SP_ACL_TCAM_PATTERNS_COUNT,
+					   tmplt_elusage, false,
+					   p_min_prio, p_max_prio);
+	if (err)
+		return err;
+
+	/* For most of the TCAM clients it would make sense to take a tcam chunk
+	 * only when the first rule is written. This is not the case for
+	 * multicast router as it is required to bind the multicast router to a
+	 * specific ACL Group ID which must exist in HW before multicast router
+	 * is initialized.
+	 */
+	ruleset->vchunk = mlxsw_sp_acl_tcam_vchunk_get(mlxsw_sp,
+						       &ruleset->vgroup, 1,
+						       tmplt_elusage);
+	if (IS_ERR(ruleset->vchunk)) {
+		err = PTR_ERR(ruleset->vchunk);
+		goto err_chunk_get;
+	}
+
+	return 0;
+
+err_chunk_get:
+	mlxsw_sp_acl_tcam_vgroup_del(&ruleset->vgroup);
+	return err;
+}
+
+static void
+mlxsw_sp_acl_tcam_mr_ruleset_del(struct mlxsw_sp *mlxsw_sp, void *ruleset_priv)
+{
+	struct mlxsw_sp_acl_tcam_mr_ruleset *ruleset = ruleset_priv;
+
+	mlxsw_sp_acl_tcam_vchunk_put(mlxsw_sp, ruleset->vchunk);
+	mlxsw_sp_acl_tcam_vgroup_del(&ruleset->vgroup);
+}
+
+static int
+mlxsw_sp_acl_tcam_mr_ruleset_bind(struct mlxsw_sp *mlxsw_sp, void *ruleset_priv,
+				  struct mlxsw_sp_port *mlxsw_sp_port,
+				  bool ingress)
+{
+	/* Binding is done when initializing multicast router */
+	return 0;
+}
+
+static void
+mlxsw_sp_acl_tcam_mr_ruleset_unbind(struct mlxsw_sp *mlxsw_sp,
+				    void *ruleset_priv,
+				    struct mlxsw_sp_port *mlxsw_sp_port,
+				    bool ingress)
+{
+}
+
+static u16
+mlxsw_sp_acl_tcam_mr_ruleset_group_id(void *ruleset_priv)
+{
+	struct mlxsw_sp_acl_tcam_mr_ruleset *ruleset = ruleset_priv;
+
+	return mlxsw_sp_acl_tcam_group_id(&ruleset->vgroup.group);
+}
+
+static int
+mlxsw_sp_acl_tcam_mr_rule_add(struct mlxsw_sp *mlxsw_sp, void *ruleset_priv,
+			      void *rule_priv,
+			      struct mlxsw_sp_acl_rule_info *rulei)
+{
+	struct mlxsw_sp_acl_tcam_mr_ruleset *ruleset = ruleset_priv;
+	struct mlxsw_sp_acl_tcam_mr_rule *rule = rule_priv;
+
+	return mlxsw_sp_acl_tcam_ventry_add(mlxsw_sp, &ruleset->vgroup,
+					   &rule->ventry, rulei);
+}
+
+static void
+mlxsw_sp_acl_tcam_mr_rule_del(struct mlxsw_sp *mlxsw_sp, void *rule_priv)
+{
+	struct mlxsw_sp_acl_tcam_mr_rule *rule = rule_priv;
+
+	mlxsw_sp_acl_tcam_ventry_del(mlxsw_sp, &rule->ventry);
+}
+
+static int
+mlxsw_sp_acl_tcam_mr_rule_action_replace(struct mlxsw_sp *mlxsw_sp,
+					 void *rule_priv,
+					 struct mlxsw_sp_acl_rule_info *rulei)
+{
+	struct mlxsw_sp_acl_tcam_mr_rule *rule = rule_priv;
+
+	return mlxsw_sp_acl_tcam_ventry_action_replace(mlxsw_sp, &rule->ventry,
+						       rulei);
+}
+
+static int
+mlxsw_sp_acl_tcam_mr_rule_activity_get(struct mlxsw_sp *mlxsw_sp,
+				       void *rule_priv, bool *activity)
+{
+	*activity = false;
+
+	return 0;
+}
+
+static const struct mlxsw_sp_acl_profile_ops mlxsw_sp_acl_tcam_mr_ops = {
+	.ruleset_priv_size	= sizeof(struct mlxsw_sp_acl_tcam_mr_ruleset),
+	.ruleset_add		= mlxsw_sp_acl_tcam_mr_ruleset_add,
+	.ruleset_del		= mlxsw_sp_acl_tcam_mr_ruleset_del,
+	.ruleset_bind		= mlxsw_sp_acl_tcam_mr_ruleset_bind,
+	.ruleset_unbind		= mlxsw_sp_acl_tcam_mr_ruleset_unbind,
+	.ruleset_group_id	= mlxsw_sp_acl_tcam_mr_ruleset_group_id,
+	.rule_priv_size		= sizeof(struct mlxsw_sp_acl_tcam_mr_rule),
+	.rule_add		= mlxsw_sp_acl_tcam_mr_rule_add,
+	.rule_del		= mlxsw_sp_acl_tcam_mr_rule_del,
+	.rule_action_replace	= mlxsw_sp_acl_tcam_mr_rule_action_replace,
+	.rule_activity_get	= mlxsw_sp_acl_tcam_mr_rule_activity_get,
+};
+
+static const struct mlxsw_sp_acl_profile_ops *
+mlxsw_sp_acl_tcam_profile_ops_arr[] = {
+	[MLXSW_SP_ACL_PROFILE_FLOWER] = &mlxsw_sp_acl_tcam_flower_ops,
+	[MLXSW_SP_ACL_PROFILE_MR] = &mlxsw_sp_acl_tcam_mr_ops,
+};
+
+const struct mlxsw_sp_acl_profile_ops *
+mlxsw_sp_acl_tcam_profile_ops(struct mlxsw_sp *mlxsw_sp,
+			      enum mlxsw_sp_acl_profile profile)
+{
+	const struct mlxsw_sp_acl_profile_ops *ops;
+
+	if (WARN_ON(profile >= ARRAY_SIZE(mlxsw_sp_acl_tcam_profile_ops_arr)))
+		return NULL;
+	ops = mlxsw_sp_acl_tcam_profile_ops_arr[profile];
+	if (WARN_ON(!ops))
+		return NULL;
+	return ops;
+}
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_acl_tcam.h b/drivers/net/ethernet/mellanox/mlxsw/spectrum_acl_tcam.h
new file mode 100644
index 000000000..edbbc89e7
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_acl_tcam.h
@@ -0,0 +1,312 @@
+/* SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0 */
+/* Copyright (c) 2017-2018 Mellanox Technologies. All rights reserved */
+
+#ifndef _MLXSW_SPECTRUM_ACL_TCAM_H
+#define _MLXSW_SPECTRUM_ACL_TCAM_H
+
+#include <linux/list.h>
+#include <linux/parman.h>
+
+#include "reg.h"
+#include "spectrum.h"
+#include "core_acl_flex_keys.h"
+
+struct mlxsw_sp_acl_tcam {
+	unsigned long *used_regions; /* bit array */
+	unsigned int max_regions;
+	unsigned long *used_groups;  /* bit array */
+	unsigned int max_groups;
+	unsigned int max_group_size;
+	struct mutex lock; /* guards vregion list */
+	struct list_head vregion_list;
+	u32 vregion_rehash_intrvl;   /* ms */
+	unsigned long priv[];
+	/* priv has to be always the last item */
+};
+
+size_t mlxsw_sp_acl_tcam_priv_size(struct mlxsw_sp *mlxsw_sp);
+int mlxsw_sp_acl_tcam_init(struct mlxsw_sp *mlxsw_sp,
+			   struct mlxsw_sp_acl_tcam *tcam);
+void mlxsw_sp_acl_tcam_fini(struct mlxsw_sp *mlxsw_sp,
+			    struct mlxsw_sp_acl_tcam *tcam);
+u32 mlxsw_sp_acl_tcam_vregion_rehash_intrvl_get(struct mlxsw_sp *mlxsw_sp,
+						struct mlxsw_sp_acl_tcam *tcam);
+int mlxsw_sp_acl_tcam_vregion_rehash_intrvl_set(struct mlxsw_sp *mlxsw_sp,
+						struct mlxsw_sp_acl_tcam *tcam,
+						u32 val);
+int mlxsw_sp_acl_tcam_priority_get(struct mlxsw_sp *mlxsw_sp,
+				   struct mlxsw_sp_acl_rule_info *rulei,
+				   u32 *priority, bool fillup_priority);
+
+struct mlxsw_sp_acl_profile_ops {
+	size_t ruleset_priv_size;
+	int (*ruleset_add)(struct mlxsw_sp *mlxsw_sp,
+			   struct mlxsw_sp_acl_tcam *tcam, void *ruleset_priv,
+			   struct mlxsw_afk_element_usage *tmplt_elusage,
+			   unsigned int *p_min_prio, unsigned int *p_max_prio);
+	void (*ruleset_del)(struct mlxsw_sp *mlxsw_sp, void *ruleset_priv);
+	int (*ruleset_bind)(struct mlxsw_sp *mlxsw_sp, void *ruleset_priv,
+			    struct mlxsw_sp_port *mlxsw_sp_port,
+			    bool ingress);
+	void (*ruleset_unbind)(struct mlxsw_sp *mlxsw_sp, void *ruleset_priv,
+			       struct mlxsw_sp_port *mlxsw_sp_port,
+			       bool ingress);
+	u16 (*ruleset_group_id)(void *ruleset_priv);
+	size_t rule_priv_size;
+	int (*rule_add)(struct mlxsw_sp *mlxsw_sp,
+			void *ruleset_priv, void *rule_priv,
+			struct mlxsw_sp_acl_rule_info *rulei);
+	void (*rule_del)(struct mlxsw_sp *mlxsw_sp, void *rule_priv);
+	int (*rule_action_replace)(struct mlxsw_sp *mlxsw_sp, void *rule_priv,
+				   struct mlxsw_sp_acl_rule_info *rulei);
+	int (*rule_activity_get)(struct mlxsw_sp *mlxsw_sp, void *rule_priv,
+				 bool *activity);
+};
+
+const struct mlxsw_sp_acl_profile_ops *
+mlxsw_sp_acl_tcam_profile_ops(struct mlxsw_sp *mlxsw_sp,
+			      enum mlxsw_sp_acl_profile profile);
+
+#define MLXSW_SP_ACL_TCAM_REGION_BASE_COUNT 16
+#define MLXSW_SP_ACL_TCAM_REGION_RESIZE_STEP 16
+
+#define MLXSW_SP_ACL_TCAM_CATCHALL_PRIO (~0U)
+
+#define MLXSW_SP_ACL_TCAM_MASK_LEN \
+	(MLXSW_REG_PTCEX_FLEX_KEY_BLOCKS_LEN * BITS_PER_BYTE)
+
+struct mlxsw_sp_acl_tcam_group;
+struct mlxsw_sp_acl_tcam_vregion;
+
+struct mlxsw_sp_acl_tcam_region {
+	struct mlxsw_sp_acl_tcam_vregion *vregion;
+	struct mlxsw_sp_acl_tcam_group *group;
+	struct list_head list; /* Member of a TCAM group */
+	enum mlxsw_reg_ptar_key_type key_type;
+	u16 id; /* ACL ID and region ID - they are same */
+	char tcam_region_info[MLXSW_REG_PXXX_TCAM_REGION_INFO_LEN];
+	struct mlxsw_afk_key_info *key_info;
+	struct mlxsw_sp *mlxsw_sp;
+	unsigned long priv[];
+	/* priv has to be always the last item */
+};
+
+struct mlxsw_sp_acl_ctcam_region {
+	struct parman *parman;
+	const struct mlxsw_sp_acl_ctcam_region_ops *ops;
+	struct mlxsw_sp_acl_tcam_region *region;
+};
+
+struct mlxsw_sp_acl_ctcam_chunk {
+	struct parman_prio parman_prio;
+};
+
+struct mlxsw_sp_acl_ctcam_entry {
+	struct parman_item parman_item;
+};
+
+struct mlxsw_sp_acl_ctcam_region_ops {
+	int (*entry_insert)(struct mlxsw_sp_acl_ctcam_region *cregion,
+			    struct mlxsw_sp_acl_ctcam_entry *centry,
+			    const char *mask);
+	void (*entry_remove)(struct mlxsw_sp_acl_ctcam_region *cregion,
+			     struct mlxsw_sp_acl_ctcam_entry *centry);
+};
+
+int
+mlxsw_sp_acl_ctcam_region_init(struct mlxsw_sp *mlxsw_sp,
+			       struct mlxsw_sp_acl_ctcam_region *cregion,
+			       struct mlxsw_sp_acl_tcam_region *region,
+			       const struct mlxsw_sp_acl_ctcam_region_ops *ops);
+void mlxsw_sp_acl_ctcam_region_fini(struct mlxsw_sp_acl_ctcam_region *cregion);
+void mlxsw_sp_acl_ctcam_chunk_init(struct mlxsw_sp_acl_ctcam_region *cregion,
+				   struct mlxsw_sp_acl_ctcam_chunk *cchunk,
+				   unsigned int priority);
+void mlxsw_sp_acl_ctcam_chunk_fini(struct mlxsw_sp_acl_ctcam_chunk *cchunk);
+int mlxsw_sp_acl_ctcam_entry_add(struct mlxsw_sp *mlxsw_sp,
+				 struct mlxsw_sp_acl_ctcam_region *cregion,
+				 struct mlxsw_sp_acl_ctcam_chunk *cchunk,
+				 struct mlxsw_sp_acl_ctcam_entry *centry,
+				 struct mlxsw_sp_acl_rule_info *rulei,
+				 bool fillup_priority);
+void mlxsw_sp_acl_ctcam_entry_del(struct mlxsw_sp *mlxsw_sp,
+				  struct mlxsw_sp_acl_ctcam_region *cregion,
+				  struct mlxsw_sp_acl_ctcam_chunk *cchunk,
+				  struct mlxsw_sp_acl_ctcam_entry *centry);
+int mlxsw_sp_acl_ctcam_entry_action_replace(struct mlxsw_sp *mlxsw_sp,
+					    struct mlxsw_sp_acl_ctcam_region *cregion,
+					    struct mlxsw_sp_acl_ctcam_entry *centry,
+					    struct mlxsw_sp_acl_rule_info *rulei);
+static inline unsigned int
+mlxsw_sp_acl_ctcam_entry_offset(struct mlxsw_sp_acl_ctcam_entry *centry)
+{
+	return centry->parman_item.index;
+}
+
+enum mlxsw_sp_acl_atcam_region_type {
+	MLXSW_SP_ACL_ATCAM_REGION_TYPE_2KB,
+	MLXSW_SP_ACL_ATCAM_REGION_TYPE_4KB,
+	MLXSW_SP_ACL_ATCAM_REGION_TYPE_8KB,
+	MLXSW_SP_ACL_ATCAM_REGION_TYPE_12KB,
+	__MLXSW_SP_ACL_ATCAM_REGION_TYPE_MAX,
+};
+
+#define MLXSW_SP_ACL_ATCAM_REGION_TYPE_MAX \
+	(__MLXSW_SP_ACL_ATCAM_REGION_TYPE_MAX - 1)
+
+struct mlxsw_sp_acl_atcam {
+	struct mlxsw_sp_acl_erp_core *erp_core;
+};
+
+struct mlxsw_sp_acl_atcam_region {
+	struct rhashtable entries_ht; /* A-TCAM only */
+	struct list_head entries_list; /* A-TCAM only */
+	struct mlxsw_sp_acl_ctcam_region cregion;
+	const struct mlxsw_sp_acl_atcam_region_ops *ops;
+	struct mlxsw_sp_acl_tcam_region *region;
+	struct mlxsw_sp_acl_atcam *atcam;
+	enum mlxsw_sp_acl_atcam_region_type type;
+	struct mlxsw_sp_acl_erp_table *erp_table;
+	void *priv;
+};
+
+struct mlxsw_sp_acl_atcam_entry_ht_key {
+	char full_enc_key[MLXSW_REG_PTCEX_FLEX_KEY_BLOCKS_LEN]; /* Encoded
+								 * key.
+								 */
+	u8 erp_id;
+};
+
+struct mlxsw_sp_acl_atcam_chunk {
+	struct mlxsw_sp_acl_ctcam_chunk cchunk;
+};
+
+struct mlxsw_sp_acl_atcam_entry {
+	struct rhash_head ht_node;
+	struct list_head list; /* Member in entries_list */
+	struct mlxsw_sp_acl_atcam_entry_ht_key ht_key;
+	char enc_key[MLXSW_REG_PTCEX_FLEX_KEY_BLOCKS_LEN]; /* Encoded key,
+							    * minus delta bits.
+							    */
+	struct {
+		u16 start;
+		u8 mask;
+		u8 value;
+	} delta_info;
+	struct mlxsw_sp_acl_ctcam_entry centry;
+	struct mlxsw_sp_acl_atcam_lkey_id *lkey_id;
+	struct mlxsw_sp_acl_erp_mask *erp_mask;
+};
+
+static inline struct mlxsw_sp_acl_atcam_region *
+mlxsw_sp_acl_tcam_cregion_aregion(struct mlxsw_sp_acl_ctcam_region *cregion)
+{
+	return container_of(cregion, struct mlxsw_sp_acl_atcam_region, cregion);
+}
+
+static inline struct mlxsw_sp_acl_atcam_entry *
+mlxsw_sp_acl_tcam_centry_aentry(struct mlxsw_sp_acl_ctcam_entry *centry)
+{
+	return container_of(centry, struct mlxsw_sp_acl_atcam_entry, centry);
+}
+
+int mlxsw_sp_acl_atcam_region_associate(struct mlxsw_sp *mlxsw_sp,
+					u16 region_id);
+int
+mlxsw_sp_acl_atcam_region_init(struct mlxsw_sp *mlxsw_sp,
+			       struct mlxsw_sp_acl_atcam *atcam,
+			       struct mlxsw_sp_acl_atcam_region *aregion,
+			       struct mlxsw_sp_acl_tcam_region *region,
+			       void *hints_priv,
+			       const struct mlxsw_sp_acl_ctcam_region_ops *ops);
+void mlxsw_sp_acl_atcam_region_fini(struct mlxsw_sp_acl_atcam_region *aregion);
+void mlxsw_sp_acl_atcam_chunk_init(struct mlxsw_sp_acl_atcam_region *aregion,
+				   struct mlxsw_sp_acl_atcam_chunk *achunk,
+				   unsigned int priority);
+void mlxsw_sp_acl_atcam_chunk_fini(struct mlxsw_sp_acl_atcam_chunk *achunk);
+int mlxsw_sp_acl_atcam_entry_add(struct mlxsw_sp *mlxsw_sp,
+				 struct mlxsw_sp_acl_atcam_region *aregion,
+				 struct mlxsw_sp_acl_atcam_chunk *achunk,
+				 struct mlxsw_sp_acl_atcam_entry *aentry,
+				 struct mlxsw_sp_acl_rule_info *rulei);
+void mlxsw_sp_acl_atcam_entry_del(struct mlxsw_sp *mlxsw_sp,
+				  struct mlxsw_sp_acl_atcam_region *aregion,
+				  struct mlxsw_sp_acl_atcam_chunk *achunk,
+				  struct mlxsw_sp_acl_atcam_entry *aentry);
+int mlxsw_sp_acl_atcam_entry_action_replace(struct mlxsw_sp *mlxsw_sp,
+					    struct mlxsw_sp_acl_atcam_region *aregion,
+					    struct mlxsw_sp_acl_atcam_entry *aentry,
+					    struct mlxsw_sp_acl_rule_info *rulei);
+int mlxsw_sp_acl_atcam_init(struct mlxsw_sp *mlxsw_sp,
+			    struct mlxsw_sp_acl_atcam *atcam);
+void mlxsw_sp_acl_atcam_fini(struct mlxsw_sp *mlxsw_sp,
+			     struct mlxsw_sp_acl_atcam *atcam);
+void *
+mlxsw_sp_acl_atcam_rehash_hints_get(struct mlxsw_sp_acl_atcam_region *aregion);
+void mlxsw_sp_acl_atcam_rehash_hints_put(void *hints_priv);
+
+struct mlxsw_sp_acl_erp_delta;
+
+u16 mlxsw_sp_acl_erp_delta_start(const struct mlxsw_sp_acl_erp_delta *delta);
+u8 mlxsw_sp_acl_erp_delta_mask(const struct mlxsw_sp_acl_erp_delta *delta);
+u8 mlxsw_sp_acl_erp_delta_value(const struct mlxsw_sp_acl_erp_delta *delta,
+				const char *enc_key);
+void mlxsw_sp_acl_erp_delta_clear(const struct mlxsw_sp_acl_erp_delta *delta,
+				  const char *enc_key);
+
+struct mlxsw_sp_acl_erp_mask;
+
+bool
+mlxsw_sp_acl_erp_mask_is_ctcam(const struct mlxsw_sp_acl_erp_mask *erp_mask);
+u8 mlxsw_sp_acl_erp_mask_erp_id(const struct mlxsw_sp_acl_erp_mask *erp_mask);
+const struct mlxsw_sp_acl_erp_delta *
+mlxsw_sp_acl_erp_delta(const struct mlxsw_sp_acl_erp_mask *erp_mask);
+struct mlxsw_sp_acl_erp_mask *
+mlxsw_sp_acl_erp_mask_get(struct mlxsw_sp_acl_atcam_region *aregion,
+			  const char *mask, bool ctcam);
+void mlxsw_sp_acl_erp_mask_put(struct mlxsw_sp_acl_atcam_region *aregion,
+			       struct mlxsw_sp_acl_erp_mask *erp_mask);
+int mlxsw_sp_acl_erp_bf_insert(struct mlxsw_sp *mlxsw_sp,
+			       struct mlxsw_sp_acl_atcam_region *aregion,
+			       struct mlxsw_sp_acl_erp_mask *erp_mask,
+			       struct mlxsw_sp_acl_atcam_entry *aentry);
+void mlxsw_sp_acl_erp_bf_remove(struct mlxsw_sp *mlxsw_sp,
+				struct mlxsw_sp_acl_atcam_region *aregion,
+				struct mlxsw_sp_acl_erp_mask *erp_mask,
+				struct mlxsw_sp_acl_atcam_entry *aentry);
+void *
+mlxsw_sp_acl_erp_rehash_hints_get(struct mlxsw_sp_acl_atcam_region *aregion);
+void mlxsw_sp_acl_erp_rehash_hints_put(void *hints_priv);
+int mlxsw_sp_acl_erp_region_init(struct mlxsw_sp_acl_atcam_region *aregion,
+				 void *hints_priv);
+void mlxsw_sp_acl_erp_region_fini(struct mlxsw_sp_acl_atcam_region *aregion);
+int mlxsw_sp_acl_erps_init(struct mlxsw_sp *mlxsw_sp,
+			   struct mlxsw_sp_acl_atcam *atcam);
+void mlxsw_sp_acl_erps_fini(struct mlxsw_sp *mlxsw_sp,
+			    struct mlxsw_sp_acl_atcam *atcam);
+
+struct mlxsw_sp_acl_bf;
+
+struct mlxsw_sp_acl_bf_ops {
+	unsigned int (*index_get)(struct mlxsw_sp_acl_bf *bf,
+				  struct mlxsw_sp_acl_atcam_region *aregion,
+				  struct mlxsw_sp_acl_atcam_entry *aentry);
+};
+
+int
+mlxsw_sp_acl_bf_entry_add(struct mlxsw_sp *mlxsw_sp,
+			  struct mlxsw_sp_acl_bf *bf,
+			  struct mlxsw_sp_acl_atcam_region *aregion,
+			  unsigned int erp_bank,
+			  struct mlxsw_sp_acl_atcam_entry *aentry);
+void
+mlxsw_sp_acl_bf_entry_del(struct mlxsw_sp *mlxsw_sp,
+			  struct mlxsw_sp_acl_bf *bf,
+			  struct mlxsw_sp_acl_atcam_region *aregion,
+			  unsigned int erp_bank,
+			  struct mlxsw_sp_acl_atcam_entry *aentry);
+struct mlxsw_sp_acl_bf *
+mlxsw_sp_acl_bf_init(struct mlxsw_sp *mlxsw_sp, unsigned int num_erp_banks);
+void mlxsw_sp_acl_bf_fini(struct mlxsw_sp_acl_bf *bf);
+
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_buffers.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_buffers.c
new file mode 100644
index 000000000..c9f1c79f3
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_buffers.c
@@ -0,0 +1,1786 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2015-2018 Mellanox Technologies. All rights reserved */
+
+#include <linux/kernel.h>
+#include <linux/types.h>
+#include <linux/dcbnl.h>
+#include <linux/if_ether.h>
+#include <linux/list.h>
+#include <linux/netlink.h>
+
+#include "spectrum.h"
+#include "core.h"
+#include "port.h"
+#include "reg.h"
+
+struct mlxsw_sp_sb_pr {
+	enum mlxsw_reg_sbpr_mode mode;
+	u32 size;
+	u8 freeze_mode:1,
+	   freeze_size:1;
+};
+
+struct mlxsw_cp_sb_occ {
+	u32 cur;
+	u32 max;
+};
+
+struct mlxsw_sp_sb_cm {
+	u32 min_buff;
+	u32 max_buff;
+	u16 pool_index;
+	struct mlxsw_cp_sb_occ occ;
+	u8 freeze_pool:1,
+	   freeze_thresh:1;
+};
+
+#define MLXSW_SP_SB_INFI -1U
+#define MLXSW_SP_SB_REST -2U
+
+struct mlxsw_sp_sb_pm {
+	u32 min_buff;
+	u32 max_buff;
+	struct mlxsw_cp_sb_occ occ;
+};
+
+struct mlxsw_sp_sb_mm {
+	u32 min_buff;
+	u32 max_buff;
+	u16 pool_index;
+};
+
+struct mlxsw_sp_sb_pool_des {
+	enum mlxsw_reg_sbxx_dir dir;
+	u8 pool;
+};
+
+#define MLXSW_SP_SB_POOL_ING		0
+#define MLXSW_SP_SB_POOL_EGR		4
+#define MLXSW_SP_SB_POOL_EGR_MC		8
+#define MLXSW_SP_SB_POOL_ING_CPU	9
+#define MLXSW_SP_SB_POOL_EGR_CPU	10
+
+static const struct mlxsw_sp_sb_pool_des mlxsw_sp1_sb_pool_dess[] = {
+	{MLXSW_REG_SBXX_DIR_INGRESS, 0},
+	{MLXSW_REG_SBXX_DIR_INGRESS, 1},
+	{MLXSW_REG_SBXX_DIR_INGRESS, 2},
+	{MLXSW_REG_SBXX_DIR_INGRESS, 3},
+	{MLXSW_REG_SBXX_DIR_EGRESS, 0},
+	{MLXSW_REG_SBXX_DIR_EGRESS, 1},
+	{MLXSW_REG_SBXX_DIR_EGRESS, 2},
+	{MLXSW_REG_SBXX_DIR_EGRESS, 3},
+	{MLXSW_REG_SBXX_DIR_EGRESS, 15},
+	{MLXSW_REG_SBXX_DIR_INGRESS, 4},
+	{MLXSW_REG_SBXX_DIR_EGRESS, 4},
+};
+
+static const struct mlxsw_sp_sb_pool_des mlxsw_sp2_sb_pool_dess[] = {
+	{MLXSW_REG_SBXX_DIR_INGRESS, 0},
+	{MLXSW_REG_SBXX_DIR_INGRESS, 1},
+	{MLXSW_REG_SBXX_DIR_INGRESS, 2},
+	{MLXSW_REG_SBXX_DIR_INGRESS, 3},
+	{MLXSW_REG_SBXX_DIR_EGRESS, 0},
+	{MLXSW_REG_SBXX_DIR_EGRESS, 1},
+	{MLXSW_REG_SBXX_DIR_EGRESS, 2},
+	{MLXSW_REG_SBXX_DIR_EGRESS, 3},
+	{MLXSW_REG_SBXX_DIR_EGRESS, 15},
+	{MLXSW_REG_SBXX_DIR_INGRESS, 4},
+	{MLXSW_REG_SBXX_DIR_EGRESS, 4},
+};
+
+#define MLXSW_SP_SB_ING_TC_COUNT 8
+#define MLXSW_SP_SB_EG_TC_COUNT 16
+
+struct mlxsw_sp_sb_port {
+	struct mlxsw_sp_sb_cm ing_cms[MLXSW_SP_SB_ING_TC_COUNT];
+	struct mlxsw_sp_sb_cm eg_cms[MLXSW_SP_SB_EG_TC_COUNT];
+	struct mlxsw_sp_sb_pm *pms;
+};
+
+struct mlxsw_sp_sb {
+	struct mlxsw_sp_sb_pr *prs;
+	struct mlxsw_sp_sb_port *ports;
+	u32 cell_size;
+	u32 max_headroom_cells;
+	u64 sb_size;
+};
+
+struct mlxsw_sp_sb_vals {
+	unsigned int pool_count;
+	const struct mlxsw_sp_sb_pool_des *pool_dess;
+	const struct mlxsw_sp_sb_pm *pms;
+	const struct mlxsw_sp_sb_pm *pms_cpu;
+	const struct mlxsw_sp_sb_pr *prs;
+	const struct mlxsw_sp_sb_mm *mms;
+	const struct mlxsw_sp_sb_cm *cms_ingress;
+	const struct mlxsw_sp_sb_cm *cms_egress;
+	const struct mlxsw_sp_sb_cm *cms_cpu;
+	unsigned int mms_count;
+	unsigned int cms_ingress_count;
+	unsigned int cms_egress_count;
+	unsigned int cms_cpu_count;
+};
+
+struct mlxsw_sp_sb_ops {
+	u32 (*int_buf_size_get)(int mtu, u32 speed);
+};
+
+u32 mlxsw_sp_cells_bytes(const struct mlxsw_sp *mlxsw_sp, u32 cells)
+{
+	return mlxsw_sp->sb->cell_size * cells;
+}
+
+u32 mlxsw_sp_bytes_cells(const struct mlxsw_sp *mlxsw_sp, u32 bytes)
+{
+	return DIV_ROUND_UP(bytes, mlxsw_sp->sb->cell_size);
+}
+
+static u32 mlxsw_sp_port_headroom_8x_adjust(const struct mlxsw_sp_port *mlxsw_sp_port,
+					    u32 size_cells)
+{
+	/* Ports with eight lanes use two headroom buffers between which the
+	 * configured headroom size is split. Therefore, multiply the calculated
+	 * headroom size by two.
+	 */
+	return mlxsw_sp_port->mapping.width == 8 ? 2 * size_cells : size_cells;
+}
+
+static struct mlxsw_sp_sb_pr *mlxsw_sp_sb_pr_get(struct mlxsw_sp *mlxsw_sp,
+						 u16 pool_index)
+{
+	return &mlxsw_sp->sb->prs[pool_index];
+}
+
+static bool mlxsw_sp_sb_cm_exists(u8 pg_buff, enum mlxsw_reg_sbxx_dir dir)
+{
+	if (dir == MLXSW_REG_SBXX_DIR_INGRESS)
+		return pg_buff < MLXSW_SP_SB_ING_TC_COUNT;
+	else
+		return pg_buff < MLXSW_SP_SB_EG_TC_COUNT;
+}
+
+static struct mlxsw_sp_sb_cm *mlxsw_sp_sb_cm_get(struct mlxsw_sp *mlxsw_sp,
+						 u16 local_port, u8 pg_buff,
+						 enum mlxsw_reg_sbxx_dir dir)
+{
+	struct mlxsw_sp_sb_port *sb_port = &mlxsw_sp->sb->ports[local_port];
+
+	WARN_ON(!mlxsw_sp_sb_cm_exists(pg_buff, dir));
+	if (dir == MLXSW_REG_SBXX_DIR_INGRESS)
+		return &sb_port->ing_cms[pg_buff];
+	else
+		return &sb_port->eg_cms[pg_buff];
+}
+
+static struct mlxsw_sp_sb_pm *mlxsw_sp_sb_pm_get(struct mlxsw_sp *mlxsw_sp,
+						 u16 local_port, u16 pool_index)
+{
+	return &mlxsw_sp->sb->ports[local_port].pms[pool_index];
+}
+
+static int mlxsw_sp_sb_pr_write(struct mlxsw_sp *mlxsw_sp, u16 pool_index,
+				enum mlxsw_reg_sbpr_mode mode,
+				u32 size, bool infi_size)
+{
+	const struct mlxsw_sp_sb_pool_des *des =
+		&mlxsw_sp->sb_vals->pool_dess[pool_index];
+	char sbpr_pl[MLXSW_REG_SBPR_LEN];
+	struct mlxsw_sp_sb_pr *pr;
+	int err;
+
+	mlxsw_reg_sbpr_pack(sbpr_pl, des->pool, des->dir, mode,
+			    size, infi_size);
+	err = mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(sbpr), sbpr_pl);
+	if (err)
+		return err;
+
+	if (infi_size)
+		size = mlxsw_sp_bytes_cells(mlxsw_sp, mlxsw_sp->sb->sb_size);
+	pr = mlxsw_sp_sb_pr_get(mlxsw_sp, pool_index);
+	pr->mode = mode;
+	pr->size = size;
+	return 0;
+}
+
+static int mlxsw_sp_sb_pr_desc_write(struct mlxsw_sp *mlxsw_sp,
+				     enum mlxsw_reg_sbxx_dir dir,
+				     enum mlxsw_reg_sbpr_mode mode,
+				     u32 size, bool infi_size)
+{
+	char sbpr_pl[MLXSW_REG_SBPR_LEN];
+
+	/* The FW default descriptor buffer configuration uses only pool 14 for
+	 * descriptors.
+	 */
+	mlxsw_reg_sbpr_pack(sbpr_pl, 14, dir, mode, size, infi_size);
+	mlxsw_reg_sbpr_desc_set(sbpr_pl, true);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(sbpr), sbpr_pl);
+}
+
+static int mlxsw_sp_sb_cm_write(struct mlxsw_sp *mlxsw_sp, u16 local_port,
+				u8 pg_buff, u32 min_buff, u32 max_buff,
+				bool infi_max, u16 pool_index)
+{
+	const struct mlxsw_sp_sb_pool_des *des =
+		&mlxsw_sp->sb_vals->pool_dess[pool_index];
+	char sbcm_pl[MLXSW_REG_SBCM_LEN];
+	struct mlxsw_sp_sb_cm *cm;
+	int err;
+
+	mlxsw_reg_sbcm_pack(sbcm_pl, local_port, pg_buff, des->dir,
+			    min_buff, max_buff, infi_max, des->pool);
+	err = mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(sbcm), sbcm_pl);
+	if (err)
+		return err;
+
+	if (mlxsw_sp_sb_cm_exists(pg_buff, des->dir)) {
+		if (infi_max)
+			max_buff = mlxsw_sp_bytes_cells(mlxsw_sp,
+							mlxsw_sp->sb->sb_size);
+
+		cm = mlxsw_sp_sb_cm_get(mlxsw_sp, local_port, pg_buff,
+					des->dir);
+		cm->min_buff = min_buff;
+		cm->max_buff = max_buff;
+		cm->pool_index = pool_index;
+	}
+	return 0;
+}
+
+static int mlxsw_sp_sb_pm_write(struct mlxsw_sp *mlxsw_sp, u16 local_port,
+				u16 pool_index, u32 min_buff, u32 max_buff)
+{
+	const struct mlxsw_sp_sb_pool_des *des =
+		&mlxsw_sp->sb_vals->pool_dess[pool_index];
+	char sbpm_pl[MLXSW_REG_SBPM_LEN];
+	struct mlxsw_sp_sb_pm *pm;
+	int err;
+
+	mlxsw_reg_sbpm_pack(sbpm_pl, local_port, des->pool, des->dir, false,
+			    min_buff, max_buff);
+	err = mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(sbpm), sbpm_pl);
+	if (err)
+		return err;
+
+	pm = mlxsw_sp_sb_pm_get(mlxsw_sp, local_port, pool_index);
+	pm->min_buff = min_buff;
+	pm->max_buff = max_buff;
+	return 0;
+}
+
+static int mlxsw_sp_sb_pm_occ_clear(struct mlxsw_sp *mlxsw_sp, u16 local_port,
+				    u16 pool_index, struct list_head *bulk_list)
+{
+	const struct mlxsw_sp_sb_pool_des *des =
+		&mlxsw_sp->sb_vals->pool_dess[pool_index];
+	char sbpm_pl[MLXSW_REG_SBPM_LEN];
+
+	if (local_port == MLXSW_PORT_CPU_PORT &&
+	    des->dir == MLXSW_REG_SBXX_DIR_INGRESS)
+		return 0;
+
+	mlxsw_reg_sbpm_pack(sbpm_pl, local_port, des->pool, des->dir,
+			    true, 0, 0);
+	return mlxsw_reg_trans_query(mlxsw_sp->core, MLXSW_REG(sbpm), sbpm_pl,
+				     bulk_list, NULL, 0);
+}
+
+static void mlxsw_sp_sb_pm_occ_query_cb(struct mlxsw_core *mlxsw_core,
+					char *sbpm_pl, size_t sbpm_pl_len,
+					unsigned long cb_priv)
+{
+	struct mlxsw_sp_sb_pm *pm = (struct mlxsw_sp_sb_pm *) cb_priv;
+
+	mlxsw_reg_sbpm_unpack(sbpm_pl, &pm->occ.cur, &pm->occ.max);
+}
+
+static int mlxsw_sp_sb_pm_occ_query(struct mlxsw_sp *mlxsw_sp, u16 local_port,
+				    u16 pool_index, struct list_head *bulk_list)
+{
+	const struct mlxsw_sp_sb_pool_des *des =
+		&mlxsw_sp->sb_vals->pool_dess[pool_index];
+	char sbpm_pl[MLXSW_REG_SBPM_LEN];
+	struct mlxsw_sp_sb_pm *pm;
+
+	if (local_port == MLXSW_PORT_CPU_PORT &&
+	    des->dir == MLXSW_REG_SBXX_DIR_INGRESS)
+		return 0;
+
+	pm = mlxsw_sp_sb_pm_get(mlxsw_sp, local_port, pool_index);
+	mlxsw_reg_sbpm_pack(sbpm_pl, local_port, des->pool, des->dir,
+			    false, 0, 0);
+	return mlxsw_reg_trans_query(mlxsw_sp->core, MLXSW_REG(sbpm), sbpm_pl,
+				     bulk_list,
+				     mlxsw_sp_sb_pm_occ_query_cb,
+				     (unsigned long) pm);
+}
+
+void mlxsw_sp_hdroom_prios_reset_buf_idx(struct mlxsw_sp_hdroom *hdroom)
+{
+	int prio;
+
+	for (prio = 0; prio < IEEE_8021QAZ_MAX_TCS; prio++) {
+		switch (hdroom->mode) {
+		case MLXSW_SP_HDROOM_MODE_DCB:
+			hdroom->prios.prio[prio].buf_idx = hdroom->prios.prio[prio].ets_buf_idx;
+			break;
+		case MLXSW_SP_HDROOM_MODE_TC:
+			hdroom->prios.prio[prio].buf_idx = hdroom->prios.prio[prio].set_buf_idx;
+			break;
+		}
+	}
+}
+
+void mlxsw_sp_hdroom_bufs_reset_lossiness(struct mlxsw_sp_hdroom *hdroom)
+{
+	int prio;
+	int i;
+
+	for (i = 0; i < DCBX_MAX_BUFFERS; i++)
+		hdroom->bufs.buf[i].lossy = true;
+
+	for (prio = 0; prio < IEEE_8021Q_MAX_PRIORITIES; prio++) {
+		if (!hdroom->prios.prio[prio].lossy)
+			hdroom->bufs.buf[hdroom->prios.prio[prio].buf_idx].lossy = false;
+	}
+}
+
+static u16 mlxsw_sp_hdroom_buf_threshold_get(const struct mlxsw_sp *mlxsw_sp, int mtu)
+{
+	return 2 * mlxsw_sp_bytes_cells(mlxsw_sp, mtu);
+}
+
+static void mlxsw_sp_hdroom_buf_pack(char *pbmc_pl, int index, u16 size, u16 thres, bool lossy)
+{
+	if (lossy)
+		mlxsw_reg_pbmc_lossy_buffer_pack(pbmc_pl, index, size);
+	else
+		mlxsw_reg_pbmc_lossless_buffer_pack(pbmc_pl, index, size,
+						    thres);
+}
+
+static u16 mlxsw_sp_hdroom_buf_delay_get(const struct mlxsw_sp *mlxsw_sp,
+					 const struct mlxsw_sp_hdroom *hdroom)
+{
+	u16 delay_cells;
+
+	delay_cells = mlxsw_sp_bytes_cells(mlxsw_sp, hdroom->delay_bytes);
+
+	/* In the worst case scenario the delay will be made up of packets that
+	 * are all of size CELL_SIZE + 1, which means each packet will require
+	 * almost twice its true size when buffered in the switch. We therefore
+	 * multiply this value by the "cell factor", which is close to 2.
+	 *
+	 * Another MTU is added in case the transmitting host already started
+	 * transmitting a maximum length frame when the PFC packet was received.
+	 */
+	return 2 * delay_cells + mlxsw_sp_bytes_cells(mlxsw_sp, hdroom->mtu);
+}
+
+static u32 mlxsw_sp_hdroom_int_buf_size_get(struct mlxsw_sp *mlxsw_sp, int mtu, u32 speed)
+{
+	u32 buffsize = mlxsw_sp->sb_ops->int_buf_size_get(mtu, speed);
+
+	return mlxsw_sp_bytes_cells(mlxsw_sp, buffsize) + 1;
+}
+
+static bool mlxsw_sp_hdroom_buf_is_used(const struct mlxsw_sp_hdroom *hdroom, int buf)
+{
+	int prio;
+
+	for (prio = 0; prio < IEEE_8021QAZ_MAX_TCS; prio++) {
+		if (hdroom->prios.prio[prio].buf_idx == buf)
+			return true;
+	}
+	return false;
+}
+
+void mlxsw_sp_hdroom_bufs_reset_sizes(struct mlxsw_sp_port *mlxsw_sp_port,
+				      struct mlxsw_sp_hdroom *hdroom)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	u16 reserve_cells;
+	int i;
+
+	/* Internal buffer. */
+	reserve_cells = mlxsw_sp_hdroom_int_buf_size_get(mlxsw_sp, mlxsw_sp_port->max_mtu,
+							 mlxsw_sp_port->max_speed);
+	reserve_cells = mlxsw_sp_port_headroom_8x_adjust(mlxsw_sp_port, reserve_cells);
+	hdroom->int_buf.reserve_cells = reserve_cells;
+
+	if (hdroom->int_buf.enable)
+		hdroom->int_buf.size_cells = reserve_cells;
+	else
+		hdroom->int_buf.size_cells = 0;
+
+	/* PG buffers. */
+	for (i = 0; i < DCBX_MAX_BUFFERS; i++) {
+		struct mlxsw_sp_hdroom_buf *buf = &hdroom->bufs.buf[i];
+		u16 thres_cells;
+		u16 delay_cells;
+
+		if (!mlxsw_sp_hdroom_buf_is_used(hdroom, i)) {
+			thres_cells = 0;
+			delay_cells = 0;
+		} else if (buf->lossy) {
+			thres_cells = mlxsw_sp_hdroom_buf_threshold_get(mlxsw_sp, hdroom->mtu);
+			delay_cells = 0;
+		} else {
+			thres_cells = mlxsw_sp_hdroom_buf_threshold_get(mlxsw_sp, hdroom->mtu);
+			delay_cells = mlxsw_sp_hdroom_buf_delay_get(mlxsw_sp, hdroom);
+		}
+
+		thres_cells = mlxsw_sp_port_headroom_8x_adjust(mlxsw_sp_port, thres_cells);
+		delay_cells = mlxsw_sp_port_headroom_8x_adjust(mlxsw_sp_port, delay_cells);
+
+		buf->thres_cells = thres_cells;
+		if (hdroom->mode == MLXSW_SP_HDROOM_MODE_DCB) {
+			buf->size_cells = thres_cells + delay_cells;
+		} else {
+			/* Do not allow going below the minimum size, even if
+			 * the user requested it.
+			 */
+			buf->size_cells = max(buf->set_size_cells, buf->thres_cells);
+		}
+	}
+}
+
+#define MLXSW_SP_PB_UNUSED 8
+
+static int mlxsw_sp_hdroom_configure_buffers(struct mlxsw_sp_port *mlxsw_sp_port,
+					     const struct mlxsw_sp_hdroom *hdroom, bool force)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	char pbmc_pl[MLXSW_REG_PBMC_LEN];
+	bool dirty;
+	int err;
+	int i;
+
+	dirty = memcmp(&mlxsw_sp_port->hdroom->bufs, &hdroom->bufs, sizeof(hdroom->bufs));
+	if (!dirty && !force)
+		return 0;
+
+	mlxsw_reg_pbmc_pack(pbmc_pl, mlxsw_sp_port->local_port, 0xffff, 0xffff / 2);
+	for (i = 0; i < MLXSW_SP_PB_COUNT; i++) {
+		const struct mlxsw_sp_hdroom_buf *buf = &hdroom->bufs.buf[i];
+
+		if (i == MLXSW_SP_PB_UNUSED)
+			continue;
+
+		mlxsw_sp_hdroom_buf_pack(pbmc_pl, i, buf->size_cells, buf->thres_cells, buf->lossy);
+	}
+
+	mlxsw_reg_pbmc_lossy_buffer_pack(pbmc_pl, MLXSW_REG_PBMC_PORT_SHARED_BUF_IDX, 0);
+	err = mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(pbmc), pbmc_pl);
+	if (err)
+		return err;
+
+	mlxsw_sp_port->hdroom->bufs = hdroom->bufs;
+	return 0;
+}
+
+static int mlxsw_sp_hdroom_configure_priomap(struct mlxsw_sp_port *mlxsw_sp_port,
+					     const struct mlxsw_sp_hdroom *hdroom, bool force)
+{
+	char pptb_pl[MLXSW_REG_PPTB_LEN];
+	bool dirty;
+	int prio;
+	int err;
+
+	dirty = memcmp(&mlxsw_sp_port->hdroom->prios, &hdroom->prios, sizeof(hdroom->prios));
+	if (!dirty && !force)
+		return 0;
+
+	mlxsw_reg_pptb_pack(pptb_pl, mlxsw_sp_port->local_port);
+	for (prio = 0; prio < IEEE_8021QAZ_MAX_TCS; prio++)
+		mlxsw_reg_pptb_prio_to_buff_pack(pptb_pl, prio, hdroom->prios.prio[prio].buf_idx);
+
+	err = mlxsw_reg_write(mlxsw_sp_port->mlxsw_sp->core, MLXSW_REG(pptb), pptb_pl);
+	if (err)
+		return err;
+
+	mlxsw_sp_port->hdroom->prios = hdroom->prios;
+	return 0;
+}
+
+static int mlxsw_sp_hdroom_configure_int_buf(struct mlxsw_sp_port *mlxsw_sp_port,
+					     const struct mlxsw_sp_hdroom *hdroom, bool force)
+{
+	char sbib_pl[MLXSW_REG_SBIB_LEN];
+	bool dirty;
+	int err;
+
+	dirty = memcmp(&mlxsw_sp_port->hdroom->int_buf, &hdroom->int_buf, sizeof(hdroom->int_buf));
+	if (!dirty && !force)
+		return 0;
+
+	mlxsw_reg_sbib_pack(sbib_pl, mlxsw_sp_port->local_port, hdroom->int_buf.size_cells);
+	err = mlxsw_reg_write(mlxsw_sp_port->mlxsw_sp->core, MLXSW_REG(sbib), sbib_pl);
+	if (err)
+		return err;
+
+	mlxsw_sp_port->hdroom->int_buf = hdroom->int_buf;
+	return 0;
+}
+
+static bool mlxsw_sp_hdroom_bufs_fit(struct mlxsw_sp *mlxsw_sp,
+				     const struct mlxsw_sp_hdroom *hdroom)
+{
+	u32 taken_headroom_cells = 0;
+	int i;
+
+	for (i = 0; i < MLXSW_SP_PB_COUNT; i++)
+		taken_headroom_cells += hdroom->bufs.buf[i].size_cells;
+
+	taken_headroom_cells += hdroom->int_buf.reserve_cells;
+	return taken_headroom_cells <= mlxsw_sp->sb->max_headroom_cells;
+}
+
+static int __mlxsw_sp_hdroom_configure(struct mlxsw_sp_port *mlxsw_sp_port,
+				       const struct mlxsw_sp_hdroom *hdroom, bool force)
+{
+	struct mlxsw_sp_hdroom orig_hdroom;
+	struct mlxsw_sp_hdroom tmp_hdroom;
+	int err;
+	int i;
+
+	/* Port buffers need to be configured in three steps. First, all buffers
+	 * with non-zero size are configured. Then, prio-to-buffer map is
+	 * updated, allowing traffic to flow to the now non-zero buffers.
+	 * Finally, zero-sized buffers are configured, because now no traffic
+	 * should be directed to them anymore. This way, in a non-congested
+	 * system, no packet drops are introduced by the reconfiguration.
+	 */
+
+	orig_hdroom = *mlxsw_sp_port->hdroom;
+	tmp_hdroom = orig_hdroom;
+	for (i = 0; i < MLXSW_SP_PB_COUNT; i++) {
+		if (hdroom->bufs.buf[i].size_cells)
+			tmp_hdroom.bufs.buf[i] = hdroom->bufs.buf[i];
+	}
+
+	if (!mlxsw_sp_hdroom_bufs_fit(mlxsw_sp_port->mlxsw_sp, &tmp_hdroom) ||
+	    !mlxsw_sp_hdroom_bufs_fit(mlxsw_sp_port->mlxsw_sp, hdroom))
+		return -ENOBUFS;
+
+	err = mlxsw_sp_hdroom_configure_buffers(mlxsw_sp_port, &tmp_hdroom, force);
+	if (err)
+		return err;
+
+	err = mlxsw_sp_hdroom_configure_priomap(mlxsw_sp_port, hdroom, force);
+	if (err)
+		goto err_configure_priomap;
+
+	err = mlxsw_sp_hdroom_configure_buffers(mlxsw_sp_port, hdroom, false);
+	if (err)
+		goto err_configure_buffers;
+
+	err = mlxsw_sp_hdroom_configure_int_buf(mlxsw_sp_port, hdroom, false);
+	if (err)
+		goto err_configure_int_buf;
+
+	*mlxsw_sp_port->hdroom = *hdroom;
+	return 0;
+
+err_configure_int_buf:
+	mlxsw_sp_hdroom_configure_buffers(mlxsw_sp_port, &tmp_hdroom, false);
+err_configure_buffers:
+	mlxsw_sp_hdroom_configure_priomap(mlxsw_sp_port, &tmp_hdroom, false);
+err_configure_priomap:
+	mlxsw_sp_hdroom_configure_buffers(mlxsw_sp_port, &orig_hdroom, false);
+	return err;
+}
+
+int mlxsw_sp_hdroom_configure(struct mlxsw_sp_port *mlxsw_sp_port,
+			      const struct mlxsw_sp_hdroom *hdroom)
+{
+	return __mlxsw_sp_hdroom_configure(mlxsw_sp_port, hdroom, false);
+}
+
+static int mlxsw_sp_port_headroom_init(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	struct mlxsw_sp_hdroom hdroom = {};
+	u32 size9;
+	int prio;
+
+	hdroom.mtu = mlxsw_sp_port->dev->mtu;
+	hdroom.mode = MLXSW_SP_HDROOM_MODE_DCB;
+	for (prio = 0; prio < IEEE_8021QAZ_MAX_TCS; prio++)
+		hdroom.prios.prio[prio].lossy = true;
+
+	mlxsw_sp_hdroom_bufs_reset_lossiness(&hdroom);
+	mlxsw_sp_hdroom_bufs_reset_sizes(mlxsw_sp_port, &hdroom);
+
+	/* Buffer 9 is used for control traffic. */
+	size9 = mlxsw_sp_port_headroom_8x_adjust(mlxsw_sp_port, mlxsw_sp_port->max_mtu);
+	hdroom.bufs.buf[9].size_cells = mlxsw_sp_bytes_cells(mlxsw_sp, size9);
+
+	return __mlxsw_sp_hdroom_configure(mlxsw_sp_port, &hdroom, true);
+}
+
+static int mlxsw_sp_sb_port_init(struct mlxsw_sp *mlxsw_sp,
+				 struct mlxsw_sp_sb_port *sb_port)
+{
+	struct mlxsw_sp_sb_pm *pms;
+
+	pms = kcalloc(mlxsw_sp->sb_vals->pool_count, sizeof(*pms),
+		      GFP_KERNEL);
+	if (!pms)
+		return -ENOMEM;
+	sb_port->pms = pms;
+	return 0;
+}
+
+static void mlxsw_sp_sb_port_fini(struct mlxsw_sp_sb_port *sb_port)
+{
+	kfree(sb_port->pms);
+}
+
+static int mlxsw_sp_sb_ports_init(struct mlxsw_sp *mlxsw_sp)
+{
+	unsigned int max_ports = mlxsw_core_max_ports(mlxsw_sp->core);
+	struct mlxsw_sp_sb_pr *prs;
+	int i;
+	int err;
+
+	mlxsw_sp->sb->ports = kcalloc(max_ports,
+				      sizeof(struct mlxsw_sp_sb_port),
+				      GFP_KERNEL);
+	if (!mlxsw_sp->sb->ports)
+		return -ENOMEM;
+
+	prs = kcalloc(mlxsw_sp->sb_vals->pool_count, sizeof(*prs),
+		      GFP_KERNEL);
+	if (!prs) {
+		err = -ENOMEM;
+		goto err_alloc_prs;
+	}
+	mlxsw_sp->sb->prs = prs;
+
+	for (i = 0; i < max_ports; i++) {
+		err = mlxsw_sp_sb_port_init(mlxsw_sp, &mlxsw_sp->sb->ports[i]);
+		if (err)
+			goto err_sb_port_init;
+	}
+
+	return 0;
+
+err_sb_port_init:
+	for (i--; i >= 0; i--)
+		mlxsw_sp_sb_port_fini(&mlxsw_sp->sb->ports[i]);
+	kfree(mlxsw_sp->sb->prs);
+err_alloc_prs:
+	kfree(mlxsw_sp->sb->ports);
+	return err;
+}
+
+static void mlxsw_sp_sb_ports_fini(struct mlxsw_sp *mlxsw_sp)
+{
+	int max_ports = mlxsw_core_max_ports(mlxsw_sp->core);
+	int i;
+
+	for (i = max_ports - 1; i >= 0; i--)
+		mlxsw_sp_sb_port_fini(&mlxsw_sp->sb->ports[i]);
+	kfree(mlxsw_sp->sb->prs);
+	kfree(mlxsw_sp->sb->ports);
+}
+
+#define MLXSW_SP_SB_PR(_mode, _size)	\
+	{				\
+		.mode = _mode,		\
+		.size = _size,		\
+	}
+
+#define MLXSW_SP_SB_PR_EXT(_mode, _size, _freeze_mode, _freeze_size)	\
+	{								\
+		.mode = _mode,						\
+		.size = _size,						\
+		.freeze_mode = _freeze_mode,				\
+		.freeze_size = _freeze_size,				\
+	}
+
+#define MLXSW_SP1_SB_PR_CPU_SIZE	(256 * 1000)
+
+/* Order according to mlxsw_sp1_sb_pool_dess */
+static const struct mlxsw_sp_sb_pr mlxsw_sp1_sb_prs[] = {
+	MLXSW_SP_SB_PR(MLXSW_REG_SBPR_MODE_DYNAMIC, MLXSW_SP_SB_REST),
+	MLXSW_SP_SB_PR(MLXSW_REG_SBPR_MODE_DYNAMIC, 0),
+	MLXSW_SP_SB_PR(MLXSW_REG_SBPR_MODE_DYNAMIC, 0),
+	MLXSW_SP_SB_PR(MLXSW_REG_SBPR_MODE_DYNAMIC, 0),
+	MLXSW_SP_SB_PR_EXT(MLXSW_REG_SBPR_MODE_DYNAMIC, MLXSW_SP_SB_REST,
+			   true, false),
+	MLXSW_SP_SB_PR(MLXSW_REG_SBPR_MODE_DYNAMIC, 0),
+	MLXSW_SP_SB_PR(MLXSW_REG_SBPR_MODE_DYNAMIC, 0),
+	MLXSW_SP_SB_PR(MLXSW_REG_SBPR_MODE_DYNAMIC, 0),
+	MLXSW_SP_SB_PR_EXT(MLXSW_REG_SBPR_MODE_STATIC, MLXSW_SP_SB_INFI,
+			   true, true),
+	MLXSW_SP_SB_PR_EXT(MLXSW_REG_SBPR_MODE_DYNAMIC,
+			   MLXSW_SP1_SB_PR_CPU_SIZE, true, false),
+	MLXSW_SP_SB_PR_EXT(MLXSW_REG_SBPR_MODE_DYNAMIC,
+			   MLXSW_SP1_SB_PR_CPU_SIZE, true, false),
+};
+
+#define MLXSW_SP2_SB_PR_CPU_SIZE	(256 * 1000)
+
+/* Order according to mlxsw_sp2_sb_pool_dess */
+static const struct mlxsw_sp_sb_pr mlxsw_sp2_sb_prs[] = {
+	MLXSW_SP_SB_PR(MLXSW_REG_SBPR_MODE_DYNAMIC, MLXSW_SP_SB_REST),
+	MLXSW_SP_SB_PR(MLXSW_REG_SBPR_MODE_STATIC, 0),
+	MLXSW_SP_SB_PR(MLXSW_REG_SBPR_MODE_STATIC, 0),
+	MLXSW_SP_SB_PR(MLXSW_REG_SBPR_MODE_STATIC, 0),
+	MLXSW_SP_SB_PR_EXT(MLXSW_REG_SBPR_MODE_DYNAMIC, MLXSW_SP_SB_REST,
+			   true, false),
+	MLXSW_SP_SB_PR(MLXSW_REG_SBPR_MODE_STATIC, 0),
+	MLXSW_SP_SB_PR(MLXSW_REG_SBPR_MODE_STATIC, 0),
+	MLXSW_SP_SB_PR(MLXSW_REG_SBPR_MODE_STATIC, 0),
+	MLXSW_SP_SB_PR_EXT(MLXSW_REG_SBPR_MODE_STATIC, MLXSW_SP_SB_INFI,
+			   true, true),
+	MLXSW_SP_SB_PR_EXT(MLXSW_REG_SBPR_MODE_DYNAMIC,
+			   MLXSW_SP2_SB_PR_CPU_SIZE, true, false),
+	MLXSW_SP_SB_PR_EXT(MLXSW_REG_SBPR_MODE_DYNAMIC,
+			   MLXSW_SP2_SB_PR_CPU_SIZE, true, false),
+};
+
+static int mlxsw_sp_sb_prs_init(struct mlxsw_sp *mlxsw_sp,
+				const struct mlxsw_sp_sb_pr *prs,
+				const struct mlxsw_sp_sb_pool_des *pool_dess,
+				size_t prs_len)
+{
+	/* Round down, unlike mlxsw_sp_bytes_cells(). */
+	u32 sb_cells = div_u64(mlxsw_sp->sb->sb_size, mlxsw_sp->sb->cell_size);
+	u32 rest_cells[2] = {sb_cells, sb_cells};
+	int i;
+	int err;
+
+	/* Calculate how much space to give to the "REST" pools in either
+	 * direction.
+	 */
+	for (i = 0; i < prs_len; i++) {
+		enum mlxsw_reg_sbxx_dir dir = pool_dess[i].dir;
+		u32 size = prs[i].size;
+		u32 size_cells;
+
+		if (size == MLXSW_SP_SB_INFI || size == MLXSW_SP_SB_REST)
+			continue;
+
+		size_cells = mlxsw_sp_bytes_cells(mlxsw_sp, size);
+		if (WARN_ON_ONCE(size_cells > rest_cells[dir]))
+			continue;
+
+		rest_cells[dir] -= size_cells;
+	}
+
+	for (i = 0; i < prs_len; i++) {
+		u32 size = prs[i].size;
+		u32 size_cells;
+
+		if (size == MLXSW_SP_SB_INFI) {
+			err = mlxsw_sp_sb_pr_write(mlxsw_sp, i, prs[i].mode,
+						   0, true);
+		} else if (size == MLXSW_SP_SB_REST) {
+			size_cells = rest_cells[pool_dess[i].dir];
+			err = mlxsw_sp_sb_pr_write(mlxsw_sp, i, prs[i].mode,
+						   size_cells, false);
+		} else {
+			size_cells = mlxsw_sp_bytes_cells(mlxsw_sp, size);
+			err = mlxsw_sp_sb_pr_write(mlxsw_sp, i, prs[i].mode,
+						   size_cells, false);
+		}
+		if (err)
+			return err;
+	}
+
+	err = mlxsw_sp_sb_pr_desc_write(mlxsw_sp, MLXSW_REG_SBXX_DIR_INGRESS,
+					MLXSW_REG_SBPR_MODE_DYNAMIC, 0, true);
+	if (err)
+		return err;
+
+	err = mlxsw_sp_sb_pr_desc_write(mlxsw_sp, MLXSW_REG_SBXX_DIR_EGRESS,
+					MLXSW_REG_SBPR_MODE_DYNAMIC, 0, true);
+	if (err)
+		return err;
+
+	return 0;
+}
+
+#define MLXSW_SP_SB_CM(_min_buff, _max_buff, _pool)	\
+	{						\
+		.min_buff = _min_buff,			\
+		.max_buff = _max_buff,			\
+		.pool_index = _pool,			\
+	}
+
+#define MLXSW_SP_SB_CM_ING(_min_buff, _max_buff)	\
+	{						\
+		.min_buff = _min_buff,			\
+		.max_buff = _max_buff,			\
+		.pool_index = MLXSW_SP_SB_POOL_ING,	\
+	}
+
+#define MLXSW_SP_SB_CM_EGR(_min_buff, _max_buff)	\
+	{						\
+		.min_buff = _min_buff,			\
+		.max_buff = _max_buff,			\
+		.pool_index = MLXSW_SP_SB_POOL_EGR,	\
+	}
+
+#define MLXSW_SP_SB_CM_EGR_MC(_min_buff, _max_buff)	\
+	{						\
+		.min_buff = _min_buff,			\
+		.max_buff = _max_buff,			\
+		.pool_index = MLXSW_SP_SB_POOL_EGR_MC,	\
+		.freeze_pool = true,			\
+		.freeze_thresh = true,			\
+	}
+
+static const struct mlxsw_sp_sb_cm mlxsw_sp1_sb_cms_ingress[] = {
+	MLXSW_SP_SB_CM_ING(10000, 8),
+	MLXSW_SP_SB_CM_ING(0, MLXSW_REG_SBXX_DYN_MAX_BUFF_MIN),
+	MLXSW_SP_SB_CM_ING(0, MLXSW_REG_SBXX_DYN_MAX_BUFF_MIN),
+	MLXSW_SP_SB_CM_ING(0, MLXSW_REG_SBXX_DYN_MAX_BUFF_MIN),
+	MLXSW_SP_SB_CM_ING(0, MLXSW_REG_SBXX_DYN_MAX_BUFF_MIN),
+	MLXSW_SP_SB_CM_ING(0, MLXSW_REG_SBXX_DYN_MAX_BUFF_MIN),
+	MLXSW_SP_SB_CM_ING(0, MLXSW_REG_SBXX_DYN_MAX_BUFF_MIN),
+	MLXSW_SP_SB_CM_ING(0, MLXSW_REG_SBXX_DYN_MAX_BUFF_MIN),
+	MLXSW_SP_SB_CM_ING(0, 0), /* dummy, this PG does not exist */
+	MLXSW_SP_SB_CM(10000, 8, MLXSW_SP_SB_POOL_ING_CPU),
+};
+
+static const struct mlxsw_sp_sb_cm mlxsw_sp2_sb_cms_ingress[] = {
+	MLXSW_SP_SB_CM_ING(0, 7),
+	MLXSW_SP_SB_CM_ING(0, MLXSW_REG_SBXX_DYN_MAX_BUFF_MIN),
+	MLXSW_SP_SB_CM_ING(0, MLXSW_REG_SBXX_DYN_MAX_BUFF_MIN),
+	MLXSW_SP_SB_CM_ING(0, MLXSW_REG_SBXX_DYN_MAX_BUFF_MIN),
+	MLXSW_SP_SB_CM_ING(0, MLXSW_REG_SBXX_DYN_MAX_BUFF_MIN),
+	MLXSW_SP_SB_CM_ING(0, MLXSW_REG_SBXX_DYN_MAX_BUFF_MIN),
+	MLXSW_SP_SB_CM_ING(0, MLXSW_REG_SBXX_DYN_MAX_BUFF_MIN),
+	MLXSW_SP_SB_CM_ING(0, MLXSW_REG_SBXX_DYN_MAX_BUFF_MIN),
+	MLXSW_SP_SB_CM_ING(0, 0), /* dummy, this PG does not exist */
+	MLXSW_SP_SB_CM(10000, 8, MLXSW_SP_SB_POOL_ING_CPU),
+};
+
+static const struct mlxsw_sp_sb_cm mlxsw_sp1_sb_cms_egress[] = {
+	MLXSW_SP_SB_CM_EGR(1500, 9),
+	MLXSW_SP_SB_CM_EGR(1500, 9),
+	MLXSW_SP_SB_CM_EGR(1500, 9),
+	MLXSW_SP_SB_CM_EGR(1500, 9),
+	MLXSW_SP_SB_CM_EGR(1500, 9),
+	MLXSW_SP_SB_CM_EGR(1500, 9),
+	MLXSW_SP_SB_CM_EGR(1500, 9),
+	MLXSW_SP_SB_CM_EGR(1500, 9),
+	MLXSW_SP_SB_CM_EGR_MC(0, MLXSW_SP_SB_INFI),
+	MLXSW_SP_SB_CM_EGR_MC(0, MLXSW_SP_SB_INFI),
+	MLXSW_SP_SB_CM_EGR_MC(0, MLXSW_SP_SB_INFI),
+	MLXSW_SP_SB_CM_EGR_MC(0, MLXSW_SP_SB_INFI),
+	MLXSW_SP_SB_CM_EGR_MC(0, MLXSW_SP_SB_INFI),
+	MLXSW_SP_SB_CM_EGR_MC(0, MLXSW_SP_SB_INFI),
+	MLXSW_SP_SB_CM_EGR_MC(0, MLXSW_SP_SB_INFI),
+	MLXSW_SP_SB_CM_EGR_MC(0, MLXSW_SP_SB_INFI),
+	MLXSW_SP_SB_CM_EGR(1, 0xff),
+};
+
+static const struct mlxsw_sp_sb_cm mlxsw_sp2_sb_cms_egress[] = {
+	MLXSW_SP_SB_CM_EGR(0, 7),
+	MLXSW_SP_SB_CM_EGR(0, 7),
+	MLXSW_SP_SB_CM_EGR(0, 7),
+	MLXSW_SP_SB_CM_EGR(0, 7),
+	MLXSW_SP_SB_CM_EGR(0, 7),
+	MLXSW_SP_SB_CM_EGR(0, 7),
+	MLXSW_SP_SB_CM_EGR(0, 7),
+	MLXSW_SP_SB_CM_EGR(0, 7),
+	MLXSW_SP_SB_CM_EGR_MC(0, MLXSW_SP_SB_INFI),
+	MLXSW_SP_SB_CM_EGR_MC(0, MLXSW_SP_SB_INFI),
+	MLXSW_SP_SB_CM_EGR_MC(0, MLXSW_SP_SB_INFI),
+	MLXSW_SP_SB_CM_EGR_MC(0, MLXSW_SP_SB_INFI),
+	MLXSW_SP_SB_CM_EGR_MC(0, MLXSW_SP_SB_INFI),
+	MLXSW_SP_SB_CM_EGR_MC(0, MLXSW_SP_SB_INFI),
+	MLXSW_SP_SB_CM_EGR_MC(0, MLXSW_SP_SB_INFI),
+	MLXSW_SP_SB_CM_EGR_MC(0, MLXSW_SP_SB_INFI),
+	MLXSW_SP_SB_CM_EGR(1, 0xff),
+};
+
+#define MLXSW_SP_CPU_PORT_SB_CM MLXSW_SP_SB_CM(0, 0, MLXSW_SP_SB_POOL_EGR_CPU)
+
+static const struct mlxsw_sp_sb_cm mlxsw_sp_cpu_port_sb_cms[] = {
+	MLXSW_SP_SB_CM(1000, 8, MLXSW_SP_SB_POOL_EGR_CPU),
+	MLXSW_SP_SB_CM(1000, 8, MLXSW_SP_SB_POOL_EGR_CPU),
+	MLXSW_SP_SB_CM(1000, 8, MLXSW_SP_SB_POOL_EGR_CPU),
+	MLXSW_SP_SB_CM(1000, 8, MLXSW_SP_SB_POOL_EGR_CPU),
+	MLXSW_SP_SB_CM(1000, 8, MLXSW_SP_SB_POOL_EGR_CPU),
+	MLXSW_SP_SB_CM(1000, 8, MLXSW_SP_SB_POOL_EGR_CPU),
+	MLXSW_SP_CPU_PORT_SB_CM,
+	MLXSW_SP_SB_CM(1000, 8, MLXSW_SP_SB_POOL_EGR_CPU),
+	MLXSW_SP_CPU_PORT_SB_CM,
+	MLXSW_SP_CPU_PORT_SB_CM,
+	MLXSW_SP_CPU_PORT_SB_CM,
+	MLXSW_SP_CPU_PORT_SB_CM,
+	MLXSW_SP_CPU_PORT_SB_CM,
+	MLXSW_SP_CPU_PORT_SB_CM,
+	MLXSW_SP_CPU_PORT_SB_CM,
+	MLXSW_SP_CPU_PORT_SB_CM,
+	MLXSW_SP_CPU_PORT_SB_CM,
+	MLXSW_SP_CPU_PORT_SB_CM,
+	MLXSW_SP_CPU_PORT_SB_CM,
+	MLXSW_SP_CPU_PORT_SB_CM,
+	MLXSW_SP_CPU_PORT_SB_CM,
+	MLXSW_SP_CPU_PORT_SB_CM,
+	MLXSW_SP_CPU_PORT_SB_CM,
+	MLXSW_SP_CPU_PORT_SB_CM,
+	MLXSW_SP_CPU_PORT_SB_CM,
+	MLXSW_SP_CPU_PORT_SB_CM,
+	MLXSW_SP_CPU_PORT_SB_CM,
+	MLXSW_SP_CPU_PORT_SB_CM,
+	MLXSW_SP_CPU_PORT_SB_CM,
+	MLXSW_SP_CPU_PORT_SB_CM,
+	MLXSW_SP_CPU_PORT_SB_CM,
+	MLXSW_SP_CPU_PORT_SB_CM,
+};
+
+static bool
+mlxsw_sp_sb_pool_is_static(struct mlxsw_sp *mlxsw_sp, u16 pool_index)
+{
+	struct mlxsw_sp_sb_pr *pr = mlxsw_sp_sb_pr_get(mlxsw_sp, pool_index);
+
+	return pr->mode == MLXSW_REG_SBPR_MODE_STATIC;
+}
+
+static int __mlxsw_sp_sb_cms_init(struct mlxsw_sp *mlxsw_sp, u16 local_port,
+				  enum mlxsw_reg_sbxx_dir dir,
+				  const struct mlxsw_sp_sb_cm *cms,
+				  size_t cms_len)
+{
+	const struct mlxsw_sp_sb_vals *sb_vals = mlxsw_sp->sb_vals;
+	int i;
+	int err;
+
+	for (i = 0; i < cms_len; i++) {
+		const struct mlxsw_sp_sb_cm *cm;
+		u32 min_buff;
+		u32 max_buff;
+
+		if (i == 8 && dir == MLXSW_REG_SBXX_DIR_INGRESS)
+			continue; /* PG number 8 does not exist, skip it */
+		cm = &cms[i];
+		if (WARN_ON(sb_vals->pool_dess[cm->pool_index].dir != dir))
+			continue;
+
+		min_buff = mlxsw_sp_bytes_cells(mlxsw_sp, cm->min_buff);
+		max_buff = cm->max_buff;
+		if (max_buff == MLXSW_SP_SB_INFI) {
+			err = mlxsw_sp_sb_cm_write(mlxsw_sp, local_port, i,
+						   min_buff, 0,
+						   true, cm->pool_index);
+		} else {
+			if (mlxsw_sp_sb_pool_is_static(mlxsw_sp,
+						       cm->pool_index))
+				max_buff = mlxsw_sp_bytes_cells(mlxsw_sp,
+								max_buff);
+			err = mlxsw_sp_sb_cm_write(mlxsw_sp, local_port, i,
+						   min_buff, max_buff,
+						   false, cm->pool_index);
+		}
+		if (err)
+			return err;
+	}
+	return 0;
+}
+
+static int mlxsw_sp_port_sb_cms_init(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	int err;
+
+	err = __mlxsw_sp_sb_cms_init(mlxsw_sp,
+				     mlxsw_sp_port->local_port,
+				     MLXSW_REG_SBXX_DIR_INGRESS,
+				     mlxsw_sp->sb_vals->cms_ingress,
+				     mlxsw_sp->sb_vals->cms_ingress_count);
+	if (err)
+		return err;
+	return __mlxsw_sp_sb_cms_init(mlxsw_sp_port->mlxsw_sp,
+				      mlxsw_sp_port->local_port,
+				      MLXSW_REG_SBXX_DIR_EGRESS,
+				      mlxsw_sp->sb_vals->cms_egress,
+				      mlxsw_sp->sb_vals->cms_egress_count);
+}
+
+static int mlxsw_sp_cpu_port_sb_cms_init(struct mlxsw_sp *mlxsw_sp)
+{
+	return __mlxsw_sp_sb_cms_init(mlxsw_sp, 0, MLXSW_REG_SBXX_DIR_EGRESS,
+				      mlxsw_sp->sb_vals->cms_cpu,
+				      mlxsw_sp->sb_vals->cms_cpu_count);
+}
+
+#define MLXSW_SP_SB_PM(_min_buff, _max_buff)	\
+	{					\
+		.min_buff = _min_buff,		\
+		.max_buff = _max_buff,		\
+	}
+
+/* Order according to mlxsw_sp1_sb_pool_dess */
+static const struct mlxsw_sp_sb_pm mlxsw_sp1_sb_pms[] = {
+	MLXSW_SP_SB_PM(0, MLXSW_REG_SBXX_DYN_MAX_BUFF_MAX),
+	MLXSW_SP_SB_PM(0, MLXSW_REG_SBXX_DYN_MAX_BUFF_MIN),
+	MLXSW_SP_SB_PM(0, MLXSW_REG_SBXX_DYN_MAX_BUFF_MIN),
+	MLXSW_SP_SB_PM(0, MLXSW_REG_SBXX_DYN_MAX_BUFF_MIN),
+	MLXSW_SP_SB_PM(0, 7),
+	MLXSW_SP_SB_PM(0, MLXSW_REG_SBXX_DYN_MAX_BUFF_MIN),
+	MLXSW_SP_SB_PM(0, MLXSW_REG_SBXX_DYN_MAX_BUFF_MIN),
+	MLXSW_SP_SB_PM(0, MLXSW_REG_SBXX_DYN_MAX_BUFF_MIN),
+	MLXSW_SP_SB_PM(10000, 90000),
+	MLXSW_SP_SB_PM(0, 8),	/* 50% occupancy */
+	MLXSW_SP_SB_PM(0, MLXSW_REG_SBXX_DYN_MAX_BUFF_MIN),
+};
+
+/* Order according to mlxsw_sp2_sb_pool_dess */
+static const struct mlxsw_sp_sb_pm mlxsw_sp2_sb_pms[] = {
+	MLXSW_SP_SB_PM(0, 7),
+	MLXSW_SP_SB_PM(0, 0),
+	MLXSW_SP_SB_PM(0, 0),
+	MLXSW_SP_SB_PM(0, 0),
+	MLXSW_SP_SB_PM(0, 7),
+	MLXSW_SP_SB_PM(0, 0),
+	MLXSW_SP_SB_PM(0, 0),
+	MLXSW_SP_SB_PM(0, 0),
+	MLXSW_SP_SB_PM(10000, 90000),
+	MLXSW_SP_SB_PM(0, 8),	/* 50% occupancy */
+	MLXSW_SP_SB_PM(0, MLXSW_REG_SBXX_DYN_MAX_BUFF_MIN),
+};
+
+/* Order according to mlxsw_sp*_sb_pool_dess */
+static const struct mlxsw_sp_sb_pm mlxsw_sp_cpu_port_sb_pms[] = {
+	MLXSW_SP_SB_PM(0, 0),
+	MLXSW_SP_SB_PM(0, 0),
+	MLXSW_SP_SB_PM(0, 0),
+	MLXSW_SP_SB_PM(0, 0),
+	MLXSW_SP_SB_PM(0, 0),
+	MLXSW_SP_SB_PM(0, 0),
+	MLXSW_SP_SB_PM(0, 0),
+	MLXSW_SP_SB_PM(0, 0),
+	MLXSW_SP_SB_PM(0, 90000),
+	MLXSW_SP_SB_PM(0, 0),
+	MLXSW_SP_SB_PM(0, MLXSW_REG_SBXX_DYN_MAX_BUFF_MAX),
+};
+
+static int mlxsw_sp_sb_pms_init(struct mlxsw_sp *mlxsw_sp, u16 local_port,
+				const struct mlxsw_sp_sb_pm *pms,
+				bool skip_ingress)
+{
+	int i, err;
+
+	for (i = 0; i < mlxsw_sp->sb_vals->pool_count; i++) {
+		const struct mlxsw_sp_sb_pm *pm = &pms[i];
+		const struct mlxsw_sp_sb_pool_des *des;
+		u32 max_buff;
+		u32 min_buff;
+
+		des = &mlxsw_sp->sb_vals->pool_dess[i];
+		if (skip_ingress && des->dir == MLXSW_REG_SBXX_DIR_INGRESS)
+			continue;
+
+		min_buff = mlxsw_sp_bytes_cells(mlxsw_sp, pm->min_buff);
+		max_buff = pm->max_buff;
+		if (mlxsw_sp_sb_pool_is_static(mlxsw_sp, i))
+			max_buff = mlxsw_sp_bytes_cells(mlxsw_sp, max_buff);
+		err = mlxsw_sp_sb_pm_write(mlxsw_sp, local_port, i, min_buff,
+					   max_buff);
+		if (err)
+			return err;
+	}
+	return 0;
+}
+
+static int mlxsw_sp_port_sb_pms_init(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+
+	return mlxsw_sp_sb_pms_init(mlxsw_sp, mlxsw_sp_port->local_port,
+				    mlxsw_sp->sb_vals->pms, false);
+}
+
+static int mlxsw_sp_cpu_port_sb_pms_init(struct mlxsw_sp *mlxsw_sp)
+{
+	return mlxsw_sp_sb_pms_init(mlxsw_sp, 0, mlxsw_sp->sb_vals->pms_cpu,
+				    true);
+}
+
+#define MLXSW_SP_SB_MM(_min_buff, _max_buff)		\
+	{						\
+		.min_buff = _min_buff,			\
+		.max_buff = _max_buff,			\
+		.pool_index = MLXSW_SP_SB_POOL_EGR,	\
+	}
+
+static const struct mlxsw_sp_sb_mm mlxsw_sp_sb_mms[] = {
+	MLXSW_SP_SB_MM(0, 6),
+	MLXSW_SP_SB_MM(0, 6),
+	MLXSW_SP_SB_MM(0, 6),
+	MLXSW_SP_SB_MM(0, 6),
+	MLXSW_SP_SB_MM(0, 6),
+	MLXSW_SP_SB_MM(0, 6),
+	MLXSW_SP_SB_MM(0, 6),
+	MLXSW_SP_SB_MM(0, 6),
+	MLXSW_SP_SB_MM(0, 6),
+	MLXSW_SP_SB_MM(0, 6),
+	MLXSW_SP_SB_MM(0, 6),
+	MLXSW_SP_SB_MM(0, 6),
+	MLXSW_SP_SB_MM(0, 6),
+	MLXSW_SP_SB_MM(0, 6),
+	MLXSW_SP_SB_MM(0, 6),
+};
+
+static int mlxsw_sp_sb_mms_init(struct mlxsw_sp *mlxsw_sp)
+{
+	char sbmm_pl[MLXSW_REG_SBMM_LEN];
+	int i;
+	int err;
+
+	for (i = 0; i < mlxsw_sp->sb_vals->mms_count; i++) {
+		const struct mlxsw_sp_sb_pool_des *des;
+		const struct mlxsw_sp_sb_mm *mc;
+		u32 min_buff;
+
+		mc = &mlxsw_sp->sb_vals->mms[i];
+		des = &mlxsw_sp->sb_vals->pool_dess[mc->pool_index];
+		/* All pools used by sb_mm's are initialized using dynamic
+		 * thresholds, therefore 'max_buff' isn't specified in cells.
+		 */
+		min_buff = mlxsw_sp_bytes_cells(mlxsw_sp, mc->min_buff);
+		mlxsw_reg_sbmm_pack(sbmm_pl, i, min_buff, mc->max_buff,
+				    des->pool);
+		err = mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(sbmm), sbmm_pl);
+		if (err)
+			return err;
+	}
+	return 0;
+}
+
+static void mlxsw_sp_pool_count(struct mlxsw_sp *mlxsw_sp,
+				u16 *p_ingress_len, u16 *p_egress_len)
+{
+	int i;
+
+	for (i = 0; i < mlxsw_sp->sb_vals->pool_count; ++i) {
+		if (mlxsw_sp->sb_vals->pool_dess[i].dir ==
+		    MLXSW_REG_SBXX_DIR_INGRESS)
+			(*p_ingress_len)++;
+		else
+			(*p_egress_len)++;
+	}
+
+	WARN(*p_egress_len == 0, "No egress pools\n");
+}
+
+const struct mlxsw_sp_sb_vals mlxsw_sp1_sb_vals = {
+	.pool_count = ARRAY_SIZE(mlxsw_sp1_sb_pool_dess),
+	.pool_dess = mlxsw_sp1_sb_pool_dess,
+	.pms = mlxsw_sp1_sb_pms,
+	.pms_cpu = mlxsw_sp_cpu_port_sb_pms,
+	.prs = mlxsw_sp1_sb_prs,
+	.mms = mlxsw_sp_sb_mms,
+	.cms_ingress = mlxsw_sp1_sb_cms_ingress,
+	.cms_egress = mlxsw_sp1_sb_cms_egress,
+	.cms_cpu = mlxsw_sp_cpu_port_sb_cms,
+	.mms_count = ARRAY_SIZE(mlxsw_sp_sb_mms),
+	.cms_ingress_count = ARRAY_SIZE(mlxsw_sp1_sb_cms_ingress),
+	.cms_egress_count = ARRAY_SIZE(mlxsw_sp1_sb_cms_egress),
+	.cms_cpu_count = ARRAY_SIZE(mlxsw_sp_cpu_port_sb_cms),
+};
+
+const struct mlxsw_sp_sb_vals mlxsw_sp2_sb_vals = {
+	.pool_count = ARRAY_SIZE(mlxsw_sp2_sb_pool_dess),
+	.pool_dess = mlxsw_sp2_sb_pool_dess,
+	.pms = mlxsw_sp2_sb_pms,
+	.pms_cpu = mlxsw_sp_cpu_port_sb_pms,
+	.prs = mlxsw_sp2_sb_prs,
+	.mms = mlxsw_sp_sb_mms,
+	.cms_ingress = mlxsw_sp2_sb_cms_ingress,
+	.cms_egress = mlxsw_sp2_sb_cms_egress,
+	.cms_cpu = mlxsw_sp_cpu_port_sb_cms,
+	.mms_count = ARRAY_SIZE(mlxsw_sp_sb_mms),
+	.cms_ingress_count = ARRAY_SIZE(mlxsw_sp2_sb_cms_ingress),
+	.cms_egress_count = ARRAY_SIZE(mlxsw_sp2_sb_cms_egress),
+	.cms_cpu_count = ARRAY_SIZE(mlxsw_sp_cpu_port_sb_cms),
+};
+
+static u32 mlxsw_sp1_pb_int_buf_size_get(int mtu, u32 speed)
+{
+	return mtu * 5 / 2;
+}
+
+static u32 __mlxsw_sp_pb_int_buf_size_get(int mtu, u32 speed, u32 buffer_factor)
+{
+	return 3 * mtu + buffer_factor * speed / 1000;
+}
+
+#define MLXSW_SP2_SPAN_EG_MIRROR_BUFFER_FACTOR 38
+
+static u32 mlxsw_sp2_pb_int_buf_size_get(int mtu, u32 speed)
+{
+	int factor = MLXSW_SP2_SPAN_EG_MIRROR_BUFFER_FACTOR;
+
+	return __mlxsw_sp_pb_int_buf_size_get(mtu, speed, factor);
+}
+
+#define MLXSW_SP3_SPAN_EG_MIRROR_BUFFER_FACTOR 50
+
+static u32 mlxsw_sp3_pb_int_buf_size_get(int mtu, u32 speed)
+{
+	int factor = MLXSW_SP3_SPAN_EG_MIRROR_BUFFER_FACTOR;
+
+	return __mlxsw_sp_pb_int_buf_size_get(mtu, speed, factor);
+}
+
+const struct mlxsw_sp_sb_ops mlxsw_sp1_sb_ops = {
+	.int_buf_size_get = mlxsw_sp1_pb_int_buf_size_get,
+};
+
+const struct mlxsw_sp_sb_ops mlxsw_sp2_sb_ops = {
+	.int_buf_size_get = mlxsw_sp2_pb_int_buf_size_get,
+};
+
+const struct mlxsw_sp_sb_ops mlxsw_sp3_sb_ops = {
+	.int_buf_size_get = mlxsw_sp3_pb_int_buf_size_get,
+};
+
+int mlxsw_sp_buffers_init(struct mlxsw_sp *mlxsw_sp)
+{
+	u32 max_headroom_size;
+	u16 ing_pool_count = 0;
+	u16 eg_pool_count = 0;
+	int err;
+
+	if (!MLXSW_CORE_RES_VALID(mlxsw_sp->core, CELL_SIZE))
+		return -EIO;
+
+	if (!MLXSW_CORE_RES_VALID(mlxsw_sp->core, GUARANTEED_SHARED_BUFFER))
+		return -EIO;
+
+	if (!MLXSW_CORE_RES_VALID(mlxsw_sp->core, MAX_HEADROOM_SIZE))
+		return -EIO;
+
+	mlxsw_sp->sb = kzalloc(sizeof(*mlxsw_sp->sb), GFP_KERNEL);
+	if (!mlxsw_sp->sb)
+		return -ENOMEM;
+	mlxsw_sp->sb->cell_size = MLXSW_CORE_RES_GET(mlxsw_sp->core, CELL_SIZE);
+	mlxsw_sp->sb->sb_size = MLXSW_CORE_RES_GET(mlxsw_sp->core,
+						   GUARANTEED_SHARED_BUFFER);
+	max_headroom_size = MLXSW_CORE_RES_GET(mlxsw_sp->core,
+					       MAX_HEADROOM_SIZE);
+	/* Round down, because this limit must not be overstepped. */
+	mlxsw_sp->sb->max_headroom_cells = max_headroom_size /
+						mlxsw_sp->sb->cell_size;
+
+	err = mlxsw_sp_sb_ports_init(mlxsw_sp);
+	if (err)
+		goto err_sb_ports_init;
+	err = mlxsw_sp_sb_prs_init(mlxsw_sp, mlxsw_sp->sb_vals->prs,
+				   mlxsw_sp->sb_vals->pool_dess,
+				   mlxsw_sp->sb_vals->pool_count);
+	if (err)
+		goto err_sb_prs_init;
+	err = mlxsw_sp_cpu_port_sb_cms_init(mlxsw_sp);
+	if (err)
+		goto err_sb_cpu_port_sb_cms_init;
+	err = mlxsw_sp_cpu_port_sb_pms_init(mlxsw_sp);
+	if (err)
+		goto err_sb_cpu_port_pms_init;
+	err = mlxsw_sp_sb_mms_init(mlxsw_sp);
+	if (err)
+		goto err_sb_mms_init;
+	mlxsw_sp_pool_count(mlxsw_sp, &ing_pool_count, &eg_pool_count);
+	err = devl_sb_register(priv_to_devlink(mlxsw_sp->core), 0,
+			       mlxsw_sp->sb->sb_size,
+			       ing_pool_count,
+			       eg_pool_count,
+			       MLXSW_SP_SB_ING_TC_COUNT,
+			       MLXSW_SP_SB_EG_TC_COUNT);
+	if (err)
+		goto err_devlink_sb_register;
+
+	return 0;
+
+err_devlink_sb_register:
+err_sb_mms_init:
+err_sb_cpu_port_pms_init:
+err_sb_cpu_port_sb_cms_init:
+err_sb_prs_init:
+	mlxsw_sp_sb_ports_fini(mlxsw_sp);
+err_sb_ports_init:
+	kfree(mlxsw_sp->sb);
+	return err;
+}
+
+void mlxsw_sp_buffers_fini(struct mlxsw_sp *mlxsw_sp)
+{
+	devl_sb_unregister(priv_to_devlink(mlxsw_sp->core), 0);
+	mlxsw_sp_sb_ports_fini(mlxsw_sp);
+	kfree(mlxsw_sp->sb);
+}
+
+int mlxsw_sp_port_buffers_init(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	int err;
+
+	mlxsw_sp_port->hdroom = kzalloc(sizeof(*mlxsw_sp_port->hdroom), GFP_KERNEL);
+	if (!mlxsw_sp_port->hdroom)
+		return -ENOMEM;
+	mlxsw_sp_port->hdroom->mtu = mlxsw_sp_port->dev->mtu;
+
+	err = mlxsw_sp_port_headroom_init(mlxsw_sp_port);
+	if (err)
+		goto err_headroom_init;
+	err = mlxsw_sp_port_sb_cms_init(mlxsw_sp_port);
+	if (err)
+		goto err_port_sb_cms_init;
+	err = mlxsw_sp_port_sb_pms_init(mlxsw_sp_port);
+	if (err)
+		goto err_port_sb_pms_init;
+	return 0;
+
+err_port_sb_pms_init:
+err_port_sb_cms_init:
+err_headroom_init:
+	kfree(mlxsw_sp_port->hdroom);
+	return err;
+}
+
+void mlxsw_sp_port_buffers_fini(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	kfree(mlxsw_sp_port->hdroom);
+}
+
+int mlxsw_sp_sb_pool_get(struct mlxsw_core *mlxsw_core,
+			 unsigned int sb_index, u16 pool_index,
+			 struct devlink_sb_pool_info *pool_info)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_core_driver_priv(mlxsw_core);
+	enum mlxsw_reg_sbxx_dir dir;
+	struct mlxsw_sp_sb_pr *pr;
+
+	dir = mlxsw_sp->sb_vals->pool_dess[pool_index].dir;
+	pr = mlxsw_sp_sb_pr_get(mlxsw_sp, pool_index);
+	pool_info->pool_type = (enum devlink_sb_pool_type) dir;
+	pool_info->size = mlxsw_sp_cells_bytes(mlxsw_sp, pr->size);
+	pool_info->threshold_type = (enum devlink_sb_threshold_type) pr->mode;
+	pool_info->cell_size = mlxsw_sp->sb->cell_size;
+	return 0;
+}
+
+int mlxsw_sp_sb_pool_set(struct mlxsw_core *mlxsw_core,
+			 unsigned int sb_index, u16 pool_index, u32 size,
+			 enum devlink_sb_threshold_type threshold_type,
+			 struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_core_driver_priv(mlxsw_core);
+	u32 pool_size = mlxsw_sp_bytes_cells(mlxsw_sp, size);
+	const struct mlxsw_sp_sb_pr *pr;
+	enum mlxsw_reg_sbpr_mode mode;
+
+	mode = (enum mlxsw_reg_sbpr_mode) threshold_type;
+	pr = &mlxsw_sp->sb_vals->prs[pool_index];
+
+	if (size > MLXSW_CORE_RES_GET(mlxsw_sp->core,
+				      GUARANTEED_SHARED_BUFFER)) {
+		NL_SET_ERR_MSG_MOD(extack, "Exceeded shared buffer size");
+		return -EINVAL;
+	}
+
+	if (pr->freeze_mode && pr->mode != mode) {
+		NL_SET_ERR_MSG_MOD(extack, "Changing this pool's threshold type is forbidden");
+		return -EINVAL;
+	}
+
+	if (pr->freeze_size && pr->size != size) {
+		NL_SET_ERR_MSG_MOD(extack, "Changing this pool's size is forbidden");
+		return -EINVAL;
+	}
+
+	return mlxsw_sp_sb_pr_write(mlxsw_sp, pool_index, mode,
+				    pool_size, false);
+}
+
+#define MLXSW_SP_SB_THRESHOLD_TO_ALPHA_OFFSET (-2) /* 3->1, 16->14 */
+
+static u32 mlxsw_sp_sb_threshold_out(struct mlxsw_sp *mlxsw_sp, u16 pool_index,
+				     u32 max_buff)
+{
+	struct mlxsw_sp_sb_pr *pr = mlxsw_sp_sb_pr_get(mlxsw_sp, pool_index);
+
+	if (pr->mode == MLXSW_REG_SBPR_MODE_DYNAMIC)
+		return max_buff - MLXSW_SP_SB_THRESHOLD_TO_ALPHA_OFFSET;
+	return mlxsw_sp_cells_bytes(mlxsw_sp, max_buff);
+}
+
+static int mlxsw_sp_sb_threshold_in(struct mlxsw_sp *mlxsw_sp, u16 pool_index,
+				    u32 threshold, u32 *p_max_buff,
+				    struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp_sb_pr *pr = mlxsw_sp_sb_pr_get(mlxsw_sp, pool_index);
+
+	if (pr->mode == MLXSW_REG_SBPR_MODE_DYNAMIC) {
+		int val;
+
+		val = threshold + MLXSW_SP_SB_THRESHOLD_TO_ALPHA_OFFSET;
+		if (val < MLXSW_REG_SBXX_DYN_MAX_BUFF_MIN ||
+		    val > MLXSW_REG_SBXX_DYN_MAX_BUFF_MAX) {
+			NL_SET_ERR_MSG_MOD(extack, "Invalid dynamic threshold value");
+			return -EINVAL;
+		}
+		*p_max_buff = val;
+	} else {
+		*p_max_buff = mlxsw_sp_bytes_cells(mlxsw_sp, threshold);
+	}
+	return 0;
+}
+
+int mlxsw_sp_sb_port_pool_get(struct mlxsw_core_port *mlxsw_core_port,
+			      unsigned int sb_index, u16 pool_index,
+			      u32 *p_threshold)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port =
+			mlxsw_core_port_driver_priv(mlxsw_core_port);
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	u16 local_port = mlxsw_sp_port->local_port;
+	struct mlxsw_sp_sb_pm *pm = mlxsw_sp_sb_pm_get(mlxsw_sp, local_port,
+						       pool_index);
+
+	*p_threshold = mlxsw_sp_sb_threshold_out(mlxsw_sp, pool_index,
+						 pm->max_buff);
+	return 0;
+}
+
+int mlxsw_sp_sb_port_pool_set(struct mlxsw_core_port *mlxsw_core_port,
+			      unsigned int sb_index, u16 pool_index,
+			      u32 threshold, struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port =
+			mlxsw_core_port_driver_priv(mlxsw_core_port);
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	u16 local_port = mlxsw_sp_port->local_port;
+	u32 max_buff;
+	int err;
+
+	if (local_port == MLXSW_PORT_CPU_PORT) {
+		NL_SET_ERR_MSG_MOD(extack, "Changing CPU port's threshold is forbidden");
+		return -EINVAL;
+	}
+
+	err = mlxsw_sp_sb_threshold_in(mlxsw_sp, pool_index,
+				       threshold, &max_buff, extack);
+	if (err)
+		return err;
+
+	return mlxsw_sp_sb_pm_write(mlxsw_sp, local_port, pool_index,
+				    0, max_buff);
+}
+
+int mlxsw_sp_sb_tc_pool_bind_get(struct mlxsw_core_port *mlxsw_core_port,
+				 unsigned int sb_index, u16 tc_index,
+				 enum devlink_sb_pool_type pool_type,
+				 u16 *p_pool_index, u32 *p_threshold)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port =
+			mlxsw_core_port_driver_priv(mlxsw_core_port);
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	u16 local_port = mlxsw_sp_port->local_port;
+	u8 pg_buff = tc_index;
+	enum mlxsw_reg_sbxx_dir dir = (enum mlxsw_reg_sbxx_dir) pool_type;
+	struct mlxsw_sp_sb_cm *cm = mlxsw_sp_sb_cm_get(mlxsw_sp, local_port,
+						       pg_buff, dir);
+
+	*p_threshold = mlxsw_sp_sb_threshold_out(mlxsw_sp, cm->pool_index,
+						 cm->max_buff);
+	*p_pool_index = cm->pool_index;
+	return 0;
+}
+
+int mlxsw_sp_sb_tc_pool_bind_set(struct mlxsw_core_port *mlxsw_core_port,
+				 unsigned int sb_index, u16 tc_index,
+				 enum devlink_sb_pool_type pool_type,
+				 u16 pool_index, u32 threshold,
+				 struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port =
+			mlxsw_core_port_driver_priv(mlxsw_core_port);
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	u16 local_port = mlxsw_sp_port->local_port;
+	const struct mlxsw_sp_sb_cm *cm;
+	u8 pg_buff = tc_index;
+	enum mlxsw_reg_sbxx_dir dir = (enum mlxsw_reg_sbxx_dir) pool_type;
+	u32 max_buff;
+	int err;
+
+	if (local_port == MLXSW_PORT_CPU_PORT) {
+		NL_SET_ERR_MSG_MOD(extack, "Changing CPU port's binding is forbidden");
+		return -EINVAL;
+	}
+
+	if (dir != mlxsw_sp->sb_vals->pool_dess[pool_index].dir) {
+		NL_SET_ERR_MSG_MOD(extack, "Binding egress TC to ingress pool and vice versa is forbidden");
+		return -EINVAL;
+	}
+
+	if (dir == MLXSW_REG_SBXX_DIR_INGRESS)
+		cm = &mlxsw_sp->sb_vals->cms_ingress[tc_index];
+	else
+		cm = &mlxsw_sp->sb_vals->cms_egress[tc_index];
+
+	if (cm->freeze_pool && cm->pool_index != pool_index) {
+		NL_SET_ERR_MSG_MOD(extack, "Binding this TC to a different pool is forbidden");
+		return -EINVAL;
+	}
+
+	if (cm->freeze_thresh && cm->max_buff != threshold) {
+		NL_SET_ERR_MSG_MOD(extack, "Changing this TC's threshold is forbidden");
+		return -EINVAL;
+	}
+
+	err = mlxsw_sp_sb_threshold_in(mlxsw_sp, pool_index,
+				       threshold, &max_buff, extack);
+	if (err)
+		return err;
+
+	return mlxsw_sp_sb_cm_write(mlxsw_sp, local_port, pg_buff,
+				    0, max_buff, false, pool_index);
+}
+
+#define MASKED_COUNT_MAX \
+	(MLXSW_REG_SBSR_REC_MAX_COUNT / \
+	 (MLXSW_SP_SB_ING_TC_COUNT + MLXSW_SP_SB_EG_TC_COUNT))
+
+struct mlxsw_sp_sb_sr_occ_query_cb_ctx {
+	u8 masked_count;
+	u16 local_port_1;
+};
+
+static void mlxsw_sp_sb_sr_occ_query_cb(struct mlxsw_core *mlxsw_core,
+					char *sbsr_pl, size_t sbsr_pl_len,
+					unsigned long cb_priv)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_core_driver_priv(mlxsw_core);
+	struct mlxsw_sp_sb_sr_occ_query_cb_ctx cb_ctx;
+	u8 masked_count;
+	u16 local_port;
+	int rec_index = 0;
+	struct mlxsw_sp_sb_cm *cm;
+	int i;
+
+	memcpy(&cb_ctx, &cb_priv, sizeof(cb_ctx));
+
+	masked_count = 0;
+	for (local_port = cb_ctx.local_port_1;
+	     local_port < mlxsw_core_max_ports(mlxsw_core); local_port++) {
+		if (!mlxsw_sp->ports[local_port])
+			continue;
+		if (local_port == MLXSW_PORT_CPU_PORT) {
+			/* Ingress quotas are not supported for the CPU port */
+			masked_count++;
+			continue;
+		}
+		for (i = 0; i < MLXSW_SP_SB_ING_TC_COUNT; i++) {
+			cm = mlxsw_sp_sb_cm_get(mlxsw_sp, local_port, i,
+						MLXSW_REG_SBXX_DIR_INGRESS);
+			mlxsw_reg_sbsr_rec_unpack(sbsr_pl, rec_index++,
+						  &cm->occ.cur, &cm->occ.max);
+		}
+		if (++masked_count == cb_ctx.masked_count)
+			break;
+	}
+	masked_count = 0;
+	for (local_port = cb_ctx.local_port_1;
+	     local_port < mlxsw_core_max_ports(mlxsw_core); local_port++) {
+		if (!mlxsw_sp->ports[local_port])
+			continue;
+		for (i = 0; i < MLXSW_SP_SB_EG_TC_COUNT; i++) {
+			cm = mlxsw_sp_sb_cm_get(mlxsw_sp, local_port, i,
+						MLXSW_REG_SBXX_DIR_EGRESS);
+			mlxsw_reg_sbsr_rec_unpack(sbsr_pl, rec_index++,
+						  &cm->occ.cur, &cm->occ.max);
+		}
+		if (++masked_count == cb_ctx.masked_count)
+			break;
+	}
+}
+
+int mlxsw_sp_sb_occ_snapshot(struct mlxsw_core *mlxsw_core,
+			     unsigned int sb_index)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_core_driver_priv(mlxsw_core);
+	u16 local_port, local_port_1, last_local_port;
+	struct mlxsw_sp_sb_sr_occ_query_cb_ctx cb_ctx;
+	u8 masked_count, current_page = 0;
+	unsigned long cb_priv = 0;
+	LIST_HEAD(bulk_list);
+	char *sbsr_pl;
+	int i;
+	int err;
+	int err2;
+
+	sbsr_pl = kmalloc(MLXSW_REG_SBSR_LEN, GFP_KERNEL);
+	if (!sbsr_pl)
+		return -ENOMEM;
+
+	local_port = MLXSW_PORT_CPU_PORT;
+next_batch:
+	local_port_1 = local_port;
+	masked_count = 0;
+	mlxsw_reg_sbsr_pack(sbsr_pl, false);
+	mlxsw_reg_sbsr_port_page_set(sbsr_pl, current_page);
+	last_local_port = current_page * MLXSW_REG_SBSR_NUM_PORTS_IN_PAGE +
+			  MLXSW_REG_SBSR_NUM_PORTS_IN_PAGE - 1;
+
+	for (i = 0; i < MLXSW_SP_SB_ING_TC_COUNT; i++)
+		mlxsw_reg_sbsr_pg_buff_mask_set(sbsr_pl, i, 1);
+	for (i = 0; i < MLXSW_SP_SB_EG_TC_COUNT; i++)
+		mlxsw_reg_sbsr_tclass_mask_set(sbsr_pl, i, 1);
+	for (; local_port < mlxsw_core_max_ports(mlxsw_core); local_port++) {
+		if (!mlxsw_sp->ports[local_port])
+			continue;
+		if (local_port > last_local_port) {
+			current_page++;
+			goto do_query;
+		}
+		if (local_port != MLXSW_PORT_CPU_PORT) {
+			/* Ingress quotas are not supported for the CPU port */
+			mlxsw_reg_sbsr_ingress_port_mask_set(sbsr_pl,
+							     local_port, 1);
+		}
+		mlxsw_reg_sbsr_egress_port_mask_set(sbsr_pl, local_port, 1);
+		for (i = 0; i < mlxsw_sp->sb_vals->pool_count; i++) {
+			err = mlxsw_sp_sb_pm_occ_query(mlxsw_sp, local_port, i,
+						       &bulk_list);
+			if (err)
+				goto out;
+		}
+		if (++masked_count == MASKED_COUNT_MAX)
+			goto do_query;
+	}
+
+do_query:
+	cb_ctx.masked_count = masked_count;
+	cb_ctx.local_port_1 = local_port_1;
+	memcpy(&cb_priv, &cb_ctx, sizeof(cb_ctx));
+	err = mlxsw_reg_trans_query(mlxsw_core, MLXSW_REG(sbsr), sbsr_pl,
+				    &bulk_list, mlxsw_sp_sb_sr_occ_query_cb,
+				    cb_priv);
+	if (err)
+		goto out;
+	if (local_port < mlxsw_core_max_ports(mlxsw_core)) {
+		local_port++;
+		goto next_batch;
+	}
+
+out:
+	err2 = mlxsw_reg_trans_bulk_wait(&bulk_list);
+	if (!err)
+		err = err2;
+	kfree(sbsr_pl);
+	return err;
+}
+
+int mlxsw_sp_sb_occ_max_clear(struct mlxsw_core *mlxsw_core,
+			      unsigned int sb_index)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_core_driver_priv(mlxsw_core);
+	u16 local_port, last_local_port;
+	LIST_HEAD(bulk_list);
+	unsigned int masked_count;
+	u8 current_page = 0;
+	char *sbsr_pl;
+	int i;
+	int err;
+	int err2;
+
+	sbsr_pl = kmalloc(MLXSW_REG_SBSR_LEN, GFP_KERNEL);
+	if (!sbsr_pl)
+		return -ENOMEM;
+
+	local_port = MLXSW_PORT_CPU_PORT;
+next_batch:
+	masked_count = 0;
+	mlxsw_reg_sbsr_pack(sbsr_pl, true);
+	mlxsw_reg_sbsr_port_page_set(sbsr_pl, current_page);
+	last_local_port = current_page * MLXSW_REG_SBSR_NUM_PORTS_IN_PAGE +
+			  MLXSW_REG_SBSR_NUM_PORTS_IN_PAGE - 1;
+
+	for (i = 0; i < MLXSW_SP_SB_ING_TC_COUNT; i++)
+		mlxsw_reg_sbsr_pg_buff_mask_set(sbsr_pl, i, 1);
+	for (i = 0; i < MLXSW_SP_SB_EG_TC_COUNT; i++)
+		mlxsw_reg_sbsr_tclass_mask_set(sbsr_pl, i, 1);
+	for (; local_port < mlxsw_core_max_ports(mlxsw_core); local_port++) {
+		if (!mlxsw_sp->ports[local_port])
+			continue;
+		if (local_port > last_local_port) {
+			current_page++;
+			goto do_query;
+		}
+		if (local_port != MLXSW_PORT_CPU_PORT) {
+			/* Ingress quotas are not supported for the CPU port */
+			mlxsw_reg_sbsr_ingress_port_mask_set(sbsr_pl,
+							     local_port, 1);
+		}
+		mlxsw_reg_sbsr_egress_port_mask_set(sbsr_pl, local_port, 1);
+		for (i = 0; i < mlxsw_sp->sb_vals->pool_count; i++) {
+			err = mlxsw_sp_sb_pm_occ_clear(mlxsw_sp, local_port, i,
+						       &bulk_list);
+			if (err)
+				goto out;
+		}
+		if (++masked_count == MASKED_COUNT_MAX)
+			goto do_query;
+	}
+
+do_query:
+	err = mlxsw_reg_trans_query(mlxsw_core, MLXSW_REG(sbsr), sbsr_pl,
+				    &bulk_list, NULL, 0);
+	if (err)
+		goto out;
+	if (local_port < mlxsw_core_max_ports(mlxsw_core)) {
+		local_port++;
+		goto next_batch;
+	}
+
+out:
+	err2 = mlxsw_reg_trans_bulk_wait(&bulk_list);
+	if (!err)
+		err = err2;
+	kfree(sbsr_pl);
+	return err;
+}
+
+int mlxsw_sp_sb_occ_port_pool_get(struct mlxsw_core_port *mlxsw_core_port,
+				  unsigned int sb_index, u16 pool_index,
+				  u32 *p_cur, u32 *p_max)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port =
+			mlxsw_core_port_driver_priv(mlxsw_core_port);
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	u16 local_port = mlxsw_sp_port->local_port;
+	struct mlxsw_sp_sb_pm *pm = mlxsw_sp_sb_pm_get(mlxsw_sp, local_port,
+						       pool_index);
+
+	*p_cur = mlxsw_sp_cells_bytes(mlxsw_sp, pm->occ.cur);
+	*p_max = mlxsw_sp_cells_bytes(mlxsw_sp, pm->occ.max);
+	return 0;
+}
+
+int mlxsw_sp_sb_occ_tc_port_bind_get(struct mlxsw_core_port *mlxsw_core_port,
+				     unsigned int sb_index, u16 tc_index,
+				     enum devlink_sb_pool_type pool_type,
+				     u32 *p_cur, u32 *p_max)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port =
+			mlxsw_core_port_driver_priv(mlxsw_core_port);
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	u16 local_port = mlxsw_sp_port->local_port;
+	u8 pg_buff = tc_index;
+	enum mlxsw_reg_sbxx_dir dir = (enum mlxsw_reg_sbxx_dir) pool_type;
+	struct mlxsw_sp_sb_cm *cm = mlxsw_sp_sb_cm_get(mlxsw_sp, local_port,
+						       pg_buff, dir);
+
+	*p_cur = mlxsw_sp_cells_bytes(mlxsw_sp, cm->occ.cur);
+	*p_max = mlxsw_sp_cells_bytes(mlxsw_sp, cm->occ.max);
+	return 0;
+}
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_cnt.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_cnt.c
new file mode 100644
index 000000000..ee59c7915
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_cnt.c
@@ -0,0 +1,306 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2017-2018 Mellanox Technologies. All rights reserved */
+
+#include <linux/kernel.h>
+#include <linux/bitops.h>
+#include <linux/spinlock.h>
+
+#include "spectrum_cnt.h"
+
+struct mlxsw_sp_counter_sub_pool {
+	u64 size;
+	unsigned int base_index;
+	enum mlxsw_res_id entry_size_res_id;
+	const char *resource_name; /* devlink resource name */
+	u64 resource_id; /* devlink resource id */
+	unsigned int entry_size;
+	unsigned int bank_count;
+	atomic_t active_entries_count;
+};
+
+struct mlxsw_sp_counter_pool {
+	u64 pool_size;
+	unsigned long *usage; /* Usage bitmap */
+	spinlock_t counter_pool_lock; /* Protects counter pool allocations */
+	atomic_t active_entries_count;
+	unsigned int sub_pools_count;
+	struct mlxsw_sp_counter_sub_pool sub_pools[];
+};
+
+static const struct mlxsw_sp_counter_sub_pool mlxsw_sp_counter_sub_pools[] = {
+	[MLXSW_SP_COUNTER_SUB_POOL_FLOW] = {
+		.entry_size_res_id = MLXSW_RES_ID_COUNTER_SIZE_PACKETS_BYTES,
+		.resource_name = MLXSW_SP_RESOURCE_NAME_COUNTERS_FLOW,
+		.resource_id = MLXSW_SP_RESOURCE_COUNTERS_FLOW,
+		.bank_count = 6,
+	},
+	[MLXSW_SP_COUNTER_SUB_POOL_RIF] = {
+		.entry_size_res_id = MLXSW_RES_ID_COUNTER_SIZE_ROUTER_BASIC,
+		.resource_name = MLXSW_SP_RESOURCE_NAME_COUNTERS_RIF,
+		.resource_id = MLXSW_SP_RESOURCE_COUNTERS_RIF,
+		.bank_count = 2,
+	}
+};
+
+static u64 mlxsw_sp_counter_sub_pool_occ_get(void *priv)
+{
+	const struct mlxsw_sp_counter_sub_pool *sub_pool = priv;
+
+	return atomic_read(&sub_pool->active_entries_count);
+}
+
+static int mlxsw_sp_counter_sub_pools_init(struct mlxsw_sp *mlxsw_sp)
+{
+	struct mlxsw_sp_counter_pool *pool = mlxsw_sp->counter_pool;
+	struct devlink *devlink = priv_to_devlink(mlxsw_sp->core);
+	struct mlxsw_sp_counter_sub_pool *sub_pool;
+	unsigned int base_index = 0;
+	enum mlxsw_res_id res_id;
+	int err;
+	int i;
+
+	for (i = 0; i < pool->sub_pools_count; i++) {
+		sub_pool = &pool->sub_pools[i];
+		res_id = sub_pool->entry_size_res_id;
+
+		if (!mlxsw_core_res_valid(mlxsw_sp->core, res_id))
+			return -EIO;
+		sub_pool->entry_size = mlxsw_core_res_get(mlxsw_sp->core,
+							  res_id);
+		err = devl_resource_size_get(devlink,
+					     sub_pool->resource_id,
+					     &sub_pool->size);
+		if (err)
+			goto err_resource_size_get;
+
+		devl_resource_occ_get_register(devlink,
+					       sub_pool->resource_id,
+					       mlxsw_sp_counter_sub_pool_occ_get,
+					       sub_pool);
+
+		sub_pool->base_index = base_index;
+		base_index += sub_pool->size;
+		atomic_set(&sub_pool->active_entries_count, 0);
+	}
+	return 0;
+
+err_resource_size_get:
+	for (i--; i >= 0; i--) {
+		sub_pool = &pool->sub_pools[i];
+
+		devl_resource_occ_get_unregister(devlink,
+						 sub_pool->resource_id);
+	}
+	return err;
+}
+
+static void mlxsw_sp_counter_sub_pools_fini(struct mlxsw_sp *mlxsw_sp)
+{
+	struct mlxsw_sp_counter_pool *pool = mlxsw_sp->counter_pool;
+	struct devlink *devlink = priv_to_devlink(mlxsw_sp->core);
+	struct mlxsw_sp_counter_sub_pool *sub_pool;
+	int i;
+
+	for (i = 0; i < pool->sub_pools_count; i++) {
+		sub_pool = &pool->sub_pools[i];
+
+		WARN_ON(atomic_read(&sub_pool->active_entries_count));
+		devl_resource_occ_get_unregister(devlink,
+						 sub_pool->resource_id);
+	}
+}
+
+static u64 mlxsw_sp_counter_pool_occ_get(void *priv)
+{
+	const struct mlxsw_sp_counter_pool *pool = priv;
+
+	return atomic_read(&pool->active_entries_count);
+}
+
+int mlxsw_sp_counter_pool_init(struct mlxsw_sp *mlxsw_sp)
+{
+	unsigned int sub_pools_count = ARRAY_SIZE(mlxsw_sp_counter_sub_pools);
+	struct devlink *devlink = priv_to_devlink(mlxsw_sp->core);
+	struct mlxsw_sp_counter_pool *pool;
+	int err;
+
+	pool = kzalloc(struct_size(pool, sub_pools, sub_pools_count),
+		       GFP_KERNEL);
+	if (!pool)
+		return -ENOMEM;
+	mlxsw_sp->counter_pool = pool;
+	pool->sub_pools_count = sub_pools_count;
+	memcpy(pool->sub_pools, mlxsw_sp_counter_sub_pools,
+	       flex_array_size(pool, sub_pools, pool->sub_pools_count));
+	spin_lock_init(&pool->counter_pool_lock);
+	atomic_set(&pool->active_entries_count, 0);
+
+	err = devl_resource_size_get(devlink, MLXSW_SP_RESOURCE_COUNTERS,
+				     &pool->pool_size);
+	if (err)
+		goto err_pool_resource_size_get;
+	devl_resource_occ_get_register(devlink, MLXSW_SP_RESOURCE_COUNTERS,
+				       mlxsw_sp_counter_pool_occ_get, pool);
+
+	pool->usage = bitmap_zalloc(pool->pool_size, GFP_KERNEL);
+	if (!pool->usage) {
+		err = -ENOMEM;
+		goto err_usage_alloc;
+	}
+
+	err = mlxsw_sp_counter_sub_pools_init(mlxsw_sp);
+	if (err)
+		goto err_sub_pools_init;
+
+	return 0;
+
+err_sub_pools_init:
+	bitmap_free(pool->usage);
+err_usage_alloc:
+	devl_resource_occ_get_unregister(devlink,
+					 MLXSW_SP_RESOURCE_COUNTERS);
+err_pool_resource_size_get:
+	kfree(pool);
+	return err;
+}
+
+void mlxsw_sp_counter_pool_fini(struct mlxsw_sp *mlxsw_sp)
+{
+	struct mlxsw_sp_counter_pool *pool = mlxsw_sp->counter_pool;
+	struct devlink *devlink = priv_to_devlink(mlxsw_sp->core);
+
+	mlxsw_sp_counter_sub_pools_fini(mlxsw_sp);
+	WARN_ON(find_first_bit(pool->usage, pool->pool_size) !=
+			       pool->pool_size);
+	WARN_ON(atomic_read(&pool->active_entries_count));
+	bitmap_free(pool->usage);
+	devl_resource_occ_get_unregister(devlink,
+					 MLXSW_SP_RESOURCE_COUNTERS);
+	kfree(pool);
+}
+
+int mlxsw_sp_counter_alloc(struct mlxsw_sp *mlxsw_sp,
+			   enum mlxsw_sp_counter_sub_pool_id sub_pool_id,
+			   unsigned int *p_counter_index)
+{
+	struct mlxsw_sp_counter_pool *pool = mlxsw_sp->counter_pool;
+	struct mlxsw_sp_counter_sub_pool *sub_pool;
+	unsigned int entry_index;
+	unsigned int stop_index;
+	int i, err;
+
+	sub_pool = &pool->sub_pools[sub_pool_id];
+	stop_index = sub_pool->base_index + sub_pool->size;
+	entry_index = sub_pool->base_index;
+
+	spin_lock(&pool->counter_pool_lock);
+	entry_index = find_next_zero_bit(pool->usage, stop_index, entry_index);
+	if (entry_index == stop_index) {
+		err = -ENOBUFS;
+		goto err_alloc;
+	}
+	/* The sub-pools can contain non-integer number of entries
+	 * so we must check for overflow
+	 */
+	if (entry_index + sub_pool->entry_size > stop_index) {
+		err = -ENOBUFS;
+		goto err_alloc;
+	}
+	for (i = 0; i < sub_pool->entry_size; i++)
+		__set_bit(entry_index + i, pool->usage);
+	spin_unlock(&pool->counter_pool_lock);
+
+	*p_counter_index = entry_index;
+	atomic_add(sub_pool->entry_size, &sub_pool->active_entries_count);
+	atomic_add(sub_pool->entry_size, &pool->active_entries_count);
+	return 0;
+
+err_alloc:
+	spin_unlock(&pool->counter_pool_lock);
+	return err;
+}
+
+void mlxsw_sp_counter_free(struct mlxsw_sp *mlxsw_sp,
+			   enum mlxsw_sp_counter_sub_pool_id sub_pool_id,
+			   unsigned int counter_index)
+{
+	struct mlxsw_sp_counter_pool *pool = mlxsw_sp->counter_pool;
+	struct mlxsw_sp_counter_sub_pool *sub_pool;
+	int i;
+
+	if (WARN_ON(counter_index >= pool->pool_size))
+		return;
+	sub_pool = &pool->sub_pools[sub_pool_id];
+	spin_lock(&pool->counter_pool_lock);
+	for (i = 0; i < sub_pool->entry_size; i++)
+		__clear_bit(counter_index + i, pool->usage);
+	spin_unlock(&pool->counter_pool_lock);
+	atomic_sub(sub_pool->entry_size, &sub_pool->active_entries_count);
+	atomic_sub(sub_pool->entry_size, &pool->active_entries_count);
+}
+
+int mlxsw_sp_counter_resources_register(struct mlxsw_core *mlxsw_core)
+{
+	static struct devlink_resource_size_params size_params;
+	struct devlink *devlink = priv_to_devlink(mlxsw_core);
+	const struct mlxsw_sp_counter_sub_pool *sub_pool;
+	unsigned int total_bank_config;
+	u64 sub_pool_size;
+	u64 base_index;
+	u64 pool_size;
+	u64 bank_size;
+	int err;
+	int i;
+
+	if (!MLXSW_CORE_RES_VALID(mlxsw_core, COUNTER_POOL_SIZE) ||
+	    !MLXSW_CORE_RES_VALID(mlxsw_core, COUNTER_BANK_SIZE))
+		return -EIO;
+
+	pool_size = MLXSW_CORE_RES_GET(mlxsw_core, COUNTER_POOL_SIZE);
+	bank_size = MLXSW_CORE_RES_GET(mlxsw_core, COUNTER_BANK_SIZE);
+
+	devlink_resource_size_params_init(&size_params, pool_size,
+					  pool_size, bank_size,
+					  DEVLINK_RESOURCE_UNIT_ENTRY);
+	err = devl_resource_register(devlink,
+				     MLXSW_SP_RESOURCE_NAME_COUNTERS,
+				     pool_size,
+				     MLXSW_SP_RESOURCE_COUNTERS,
+				     DEVLINK_RESOURCE_ID_PARENT_TOP,
+				     &size_params);
+	if (err)
+		return err;
+
+	/* Allocation is based on bank count which should be
+	 * specified for each sub pool statically.
+	 */
+	total_bank_config = 0;
+	base_index = 0;
+	for (i = 0; i < ARRAY_SIZE(mlxsw_sp_counter_sub_pools); i++) {
+		sub_pool = &mlxsw_sp_counter_sub_pools[i];
+		sub_pool_size = sub_pool->bank_count * bank_size;
+		/* The last bank can't be fully used */
+		if (base_index + sub_pool_size > pool_size)
+			sub_pool_size = pool_size - base_index;
+		base_index += sub_pool_size;
+
+		devlink_resource_size_params_init(&size_params, sub_pool_size,
+						  sub_pool_size, bank_size,
+						  DEVLINK_RESOURCE_UNIT_ENTRY);
+		err = devl_resource_register(devlink,
+					     sub_pool->resource_name,
+					     sub_pool_size,
+					     sub_pool->resource_id,
+					     MLXSW_SP_RESOURCE_COUNTERS,
+					     &size_params);
+		if (err)
+			return err;
+		total_bank_config += sub_pool->bank_count;
+	}
+
+	/* Check config is valid, no bank over subscription */
+	if (WARN_ON(total_bank_config > div64_u64(pool_size, bank_size) + 1))
+		return -EINVAL;
+
+	return 0;
+}
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_cnt.h b/drivers/net/ethernet/mellanox/mlxsw/spectrum_cnt.h
new file mode 100644
index 000000000..15c8d4de8
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_cnt.h
@@ -0,0 +1,25 @@
+/* SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0 */
+/* Copyright (c) 2017-2018 Mellanox Technologies. All rights reserved */
+
+#ifndef _MLXSW_SPECTRUM_CNT_H
+#define _MLXSW_SPECTRUM_CNT_H
+
+#include "core.h"
+#include "spectrum.h"
+
+enum mlxsw_sp_counter_sub_pool_id {
+	MLXSW_SP_COUNTER_SUB_POOL_RIF,
+	MLXSW_SP_COUNTER_SUB_POOL_FLOW,
+};
+
+int mlxsw_sp_counter_alloc(struct mlxsw_sp *mlxsw_sp,
+			   enum mlxsw_sp_counter_sub_pool_id sub_pool_id,
+			   unsigned int *p_counter_index);
+void mlxsw_sp_counter_free(struct mlxsw_sp *mlxsw_sp,
+			   enum mlxsw_sp_counter_sub_pool_id sub_pool_id,
+			   unsigned int counter_index);
+int mlxsw_sp_counter_pool_init(struct mlxsw_sp *mlxsw_sp);
+void mlxsw_sp_counter_pool_fini(struct mlxsw_sp *mlxsw_sp);
+int mlxsw_sp_counter_resources_register(struct mlxsw_core *mlxsw_core);
+
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_dcb.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_dcb.c
new file mode 100644
index 000000000..aff6d4f35
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_dcb.c
@@ -0,0 +1,740 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2016-2018 Mellanox Technologies. All rights reserved */
+
+#include <linux/netdevice.h>
+#include <linux/string.h>
+#include <linux/bitops.h>
+#include <net/dcbnl.h>
+
+#include "spectrum.h"
+#include "reg.h"
+
+static u8 mlxsw_sp_dcbnl_getdcbx(struct net_device __always_unused *dev)
+{
+	return DCB_CAP_DCBX_HOST | DCB_CAP_DCBX_VER_IEEE;
+}
+
+static u8 mlxsw_sp_dcbnl_setdcbx(struct net_device __always_unused *dev,
+				 u8 mode)
+{
+	return (mode != (DCB_CAP_DCBX_HOST | DCB_CAP_DCBX_VER_IEEE)) ? 1 : 0;
+}
+
+static int mlxsw_sp_dcbnl_ieee_getets(struct net_device *dev,
+				      struct ieee_ets *ets)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+
+	memcpy(ets, mlxsw_sp_port->dcb.ets, sizeof(*ets));
+
+	return 0;
+}
+
+static int mlxsw_sp_port_ets_validate(struct mlxsw_sp_port *mlxsw_sp_port,
+				      struct ieee_ets *ets)
+{
+	struct net_device *dev = mlxsw_sp_port->dev;
+	bool has_ets_tc = false;
+	int i, tx_bw_sum = 0;
+
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) {
+		switch (ets->tc_tsa[i]) {
+		case IEEE_8021QAZ_TSA_STRICT:
+			break;
+		case IEEE_8021QAZ_TSA_ETS:
+			has_ets_tc = true;
+			tx_bw_sum += ets->tc_tx_bw[i];
+			break;
+		default:
+			netdev_err(dev, "Only strict priority and ETS are supported\n");
+			return -EINVAL;
+		}
+
+		if (ets->prio_tc[i] >= IEEE_8021QAZ_MAX_TCS) {
+			netdev_err(dev, "Invalid TC\n");
+			return -EINVAL;
+		}
+	}
+
+	if (has_ets_tc && tx_bw_sum != 100) {
+		netdev_err(dev, "Total ETS bandwidth should equal 100\n");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int mlxsw_sp_port_headroom_ets_set(struct mlxsw_sp_port *mlxsw_sp_port,
+					  struct ieee_ets *ets)
+{
+	struct net_device *dev = mlxsw_sp_port->dev;
+	struct mlxsw_sp_hdroom hdroom;
+	int prio;
+	int err;
+
+	hdroom = *mlxsw_sp_port->hdroom;
+	for (prio = 0; prio < IEEE_8021QAZ_MAX_TCS; prio++)
+		hdroom.prios.prio[prio].ets_buf_idx = ets->prio_tc[prio];
+	mlxsw_sp_hdroom_prios_reset_buf_idx(&hdroom);
+	mlxsw_sp_hdroom_bufs_reset_lossiness(&hdroom);
+	mlxsw_sp_hdroom_bufs_reset_sizes(mlxsw_sp_port, &hdroom);
+
+	err = mlxsw_sp_hdroom_configure(mlxsw_sp_port, &hdroom);
+	if (err) {
+		netdev_err(dev, "Failed to configure port's headroom\n");
+		return err;
+	}
+
+	return 0;
+}
+
+static int __mlxsw_sp_dcbnl_ieee_setets(struct mlxsw_sp_port *mlxsw_sp_port,
+					struct ieee_ets *ets)
+{
+	struct ieee_ets *my_ets = mlxsw_sp_port->dcb.ets;
+	struct net_device *dev = mlxsw_sp_port->dev;
+	int i, err;
+
+	/* Egress configuration. */
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) {
+		bool dwrr = ets->tc_tsa[i] == IEEE_8021QAZ_TSA_ETS;
+		u8 weight = ets->tc_tx_bw[i];
+
+		err = mlxsw_sp_port_ets_set(mlxsw_sp_port,
+					    MLXSW_REG_QEEC_HR_SUBGROUP, i,
+					    0, dwrr, weight);
+		if (err) {
+			netdev_err(dev, "Failed to link subgroup ETS element %d to group\n",
+				   i);
+			goto err_port_ets_set;
+		}
+	}
+
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) {
+		err = mlxsw_sp_port_prio_tc_set(mlxsw_sp_port, i,
+						ets->prio_tc[i]);
+		if (err) {
+			netdev_err(dev, "Failed to map prio %d to TC %d\n", i,
+				   ets->prio_tc[i]);
+			goto err_port_prio_tc_set;
+		}
+	}
+
+	/* Ingress configuration. */
+	err = mlxsw_sp_port_headroom_ets_set(mlxsw_sp_port, ets);
+	if (err)
+		goto err_port_headroom_set;
+
+	return 0;
+
+err_port_headroom_set:
+	i = IEEE_8021QAZ_MAX_TCS;
+err_port_prio_tc_set:
+	for (i--; i >= 0; i--)
+		mlxsw_sp_port_prio_tc_set(mlxsw_sp_port, i, my_ets->prio_tc[i]);
+	i = IEEE_8021QAZ_MAX_TCS;
+err_port_ets_set:
+	for (i--; i >= 0; i--) {
+		bool dwrr = my_ets->tc_tsa[i] == IEEE_8021QAZ_TSA_ETS;
+		u8 weight = my_ets->tc_tx_bw[i];
+
+		err = mlxsw_sp_port_ets_set(mlxsw_sp_port,
+					    MLXSW_REG_QEEC_HR_SUBGROUP, i,
+					    0, dwrr, weight);
+	}
+	return err;
+}
+
+static int mlxsw_sp_dcbnl_ieee_setets(struct net_device *dev,
+				      struct ieee_ets *ets)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+	int err;
+
+	err = mlxsw_sp_port_ets_validate(mlxsw_sp_port, ets);
+	if (err)
+		return err;
+
+	err = __mlxsw_sp_dcbnl_ieee_setets(mlxsw_sp_port, ets);
+	if (err)
+		return err;
+
+	memcpy(mlxsw_sp_port->dcb.ets, ets, sizeof(*ets));
+	mlxsw_sp_port->dcb.ets->ets_cap = IEEE_8021QAZ_MAX_TCS;
+
+	return 0;
+}
+
+static int mlxsw_sp_dcbnl_app_validate(struct net_device *dev,
+				       struct dcb_app *app)
+{
+	if (app->priority >= IEEE_8021QAZ_MAX_TCS) {
+		netdev_err(dev, "APP entry with priority value %u is invalid\n",
+			   app->priority);
+		return -EINVAL;
+	}
+
+	switch (app->selector) {
+	case IEEE_8021QAZ_APP_SEL_DSCP:
+		if (app->protocol >= 64) {
+			netdev_err(dev, "DSCP APP entry with protocol value %u is invalid\n",
+				   app->protocol);
+			return -EINVAL;
+		}
+		break;
+
+	case IEEE_8021QAZ_APP_SEL_ETHERTYPE:
+		if (app->protocol) {
+			netdev_err(dev, "EtherType APP entries with protocol value != 0 not supported\n");
+			return -EINVAL;
+		}
+		break;
+
+	default:
+		netdev_err(dev, "APP entries with selector %u not supported\n",
+			   app->selector);
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static u8
+mlxsw_sp_port_dcb_app_default_prio(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	u8 prio_mask;
+
+	prio_mask = dcb_ieee_getapp_default_prio_mask(mlxsw_sp_port->dev);
+	if (prio_mask)
+		/* Take the highest configured priority. */
+		return fls(prio_mask) - 1;
+
+	return 0;
+}
+
+static void
+mlxsw_sp_port_dcb_app_dscp_prio_map(struct mlxsw_sp_port *mlxsw_sp_port,
+				    u8 default_prio,
+				    struct dcb_ieee_app_dscp_map *map)
+{
+	int i;
+
+	dcb_ieee_getapp_dscp_prio_mask_map(mlxsw_sp_port->dev, map);
+	for (i = 0; i < ARRAY_SIZE(map->map); ++i) {
+		if (map->map[i])
+			map->map[i] = fls(map->map[i]) - 1;
+		else
+			map->map[i] = default_prio;
+	}
+}
+
+static bool
+mlxsw_sp_port_dcb_app_prio_dscp_map(struct mlxsw_sp_port *mlxsw_sp_port,
+				    struct dcb_ieee_app_prio_map *map)
+{
+	bool have_dscp = false;
+	int i;
+
+	dcb_ieee_getapp_prio_dscp_mask_map(mlxsw_sp_port->dev, map);
+	for (i = 0; i < ARRAY_SIZE(map->map); ++i) {
+		if (map->map[i]) {
+			map->map[i] = fls64(map->map[i]) - 1;
+			have_dscp = true;
+		}
+	}
+
+	return have_dscp;
+}
+
+static int
+mlxsw_sp_port_dcb_app_update_qpts(struct mlxsw_sp_port *mlxsw_sp_port,
+				  enum mlxsw_reg_qpts_trust_state ts)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	char qpts_pl[MLXSW_REG_QPTS_LEN];
+
+	mlxsw_reg_qpts_pack(qpts_pl, mlxsw_sp_port->local_port, ts);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(qpts), qpts_pl);
+}
+
+static int
+mlxsw_sp_port_dcb_app_update_qrwe(struct mlxsw_sp_port *mlxsw_sp_port,
+				  bool rewrite_dscp)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	char qrwe_pl[MLXSW_REG_QRWE_LEN];
+
+	mlxsw_reg_qrwe_pack(qrwe_pl, mlxsw_sp_port->local_port,
+			    false, rewrite_dscp);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(qrwe), qrwe_pl);
+}
+
+static int
+mlxsw_sp_port_dcb_toggle_trust(struct mlxsw_sp_port *mlxsw_sp_port,
+			       enum mlxsw_reg_qpts_trust_state ts)
+{
+	bool rewrite_dscp = ts == MLXSW_REG_QPTS_TRUST_STATE_DSCP;
+	int err;
+
+	if (mlxsw_sp_port->dcb.trust_state == ts)
+		return 0;
+
+	err = mlxsw_sp_port_dcb_app_update_qpts(mlxsw_sp_port, ts);
+	if (err)
+		return err;
+
+	err = mlxsw_sp_port_dcb_app_update_qrwe(mlxsw_sp_port, rewrite_dscp);
+	if (err)
+		goto err_update_qrwe;
+
+	mlxsw_sp_port->dcb.trust_state = ts;
+	return 0;
+
+err_update_qrwe:
+	mlxsw_sp_port_dcb_app_update_qpts(mlxsw_sp_port,
+					  mlxsw_sp_port->dcb.trust_state);
+	return err;
+}
+
+static int
+mlxsw_sp_port_dcb_app_update_qpdp(struct mlxsw_sp_port *mlxsw_sp_port,
+				  u8 default_prio)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	char qpdp_pl[MLXSW_REG_QPDP_LEN];
+
+	mlxsw_reg_qpdp_pack(qpdp_pl, mlxsw_sp_port->local_port, default_prio);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(qpdp), qpdp_pl);
+}
+
+static int
+mlxsw_sp_port_dcb_app_update_qpdpm(struct mlxsw_sp_port *mlxsw_sp_port,
+				   struct dcb_ieee_app_dscp_map *map)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	char qpdpm_pl[MLXSW_REG_QPDPM_LEN];
+	short int i;
+
+	mlxsw_reg_qpdpm_pack(qpdpm_pl, mlxsw_sp_port->local_port);
+	for (i = 0; i < ARRAY_SIZE(map->map); ++i)
+		mlxsw_reg_qpdpm_dscp_pack(qpdpm_pl, i, map->map[i]);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(qpdpm), qpdpm_pl);
+}
+
+static int
+mlxsw_sp_port_dcb_app_update_qpdsm(struct mlxsw_sp_port *mlxsw_sp_port,
+				   struct dcb_ieee_app_prio_map *map)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	char qpdsm_pl[MLXSW_REG_QPDSM_LEN];
+	short int i;
+
+	mlxsw_reg_qpdsm_pack(qpdsm_pl, mlxsw_sp_port->local_port);
+	for (i = 0; i < ARRAY_SIZE(map->map); ++i)
+		mlxsw_reg_qpdsm_prio_pack(qpdsm_pl, i, map->map[i]);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(qpdsm), qpdsm_pl);
+}
+
+static int mlxsw_sp_port_dcb_app_update(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	struct dcb_ieee_app_prio_map prio_map;
+	struct dcb_ieee_app_dscp_map dscp_map;
+	u8 default_prio;
+	bool have_dscp;
+	int err;
+
+	default_prio = mlxsw_sp_port_dcb_app_default_prio(mlxsw_sp_port);
+	err = mlxsw_sp_port_dcb_app_update_qpdp(mlxsw_sp_port, default_prio);
+	if (err) {
+		netdev_err(mlxsw_sp_port->dev, "Couldn't configure port default priority\n");
+		return err;
+	}
+
+	have_dscp = mlxsw_sp_port_dcb_app_prio_dscp_map(mlxsw_sp_port,
+							&prio_map);
+
+	mlxsw_sp_port_dcb_app_dscp_prio_map(mlxsw_sp_port, default_prio,
+					    &dscp_map);
+	err = mlxsw_sp_port_dcb_app_update_qpdpm(mlxsw_sp_port,
+						 &dscp_map);
+	if (err) {
+		netdev_err(mlxsw_sp_port->dev, "Couldn't configure priority map\n");
+		return err;
+	}
+
+	err = mlxsw_sp_port_dcb_app_update_qpdsm(mlxsw_sp_port,
+						 &prio_map);
+	if (err) {
+		netdev_err(mlxsw_sp_port->dev, "Couldn't configure DSCP rewrite map\n");
+		return err;
+	}
+
+	if (!have_dscp) {
+		err = mlxsw_sp_port_dcb_toggle_trust(mlxsw_sp_port,
+					MLXSW_REG_QPTS_TRUST_STATE_PCP);
+		if (err)
+			netdev_err(mlxsw_sp_port->dev, "Couldn't switch to trust L2\n");
+		return err;
+	}
+
+	err = mlxsw_sp_port_dcb_toggle_trust(mlxsw_sp_port,
+					     MLXSW_REG_QPTS_TRUST_STATE_DSCP);
+	if (err) {
+		/* A failure to set trust DSCP means that the QPDPM and QPDSM
+		 * maps installed above are not in effect. And since we are here
+		 * attempting to set trust DSCP, we couldn't have attempted to
+		 * switch trust to PCP. Thus no cleanup is necessary.
+		 */
+		netdev_err(mlxsw_sp_port->dev, "Couldn't switch to trust L3\n");
+		return err;
+	}
+
+	return 0;
+}
+
+static int mlxsw_sp_dcbnl_ieee_setapp(struct net_device *dev,
+				      struct dcb_app *app)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+	int err;
+
+	err = mlxsw_sp_dcbnl_app_validate(dev, app);
+	if (err)
+		return err;
+
+	err = dcb_ieee_setapp(dev, app);
+	if (err)
+		return err;
+
+	err = mlxsw_sp_port_dcb_app_update(mlxsw_sp_port);
+	if (err)
+		goto err_update;
+
+	return 0;
+
+err_update:
+	dcb_ieee_delapp(dev, app);
+	return err;
+}
+
+static int mlxsw_sp_dcbnl_ieee_delapp(struct net_device *dev,
+				      struct dcb_app *app)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+	int err;
+
+	err = dcb_ieee_delapp(dev, app);
+	if (err)
+		return err;
+
+	err = mlxsw_sp_port_dcb_app_update(mlxsw_sp_port);
+	if (err)
+		netdev_err(dev, "Failed to update DCB APP configuration\n");
+	return 0;
+}
+
+static int mlxsw_sp_dcbnl_ieee_getmaxrate(struct net_device *dev,
+					  struct ieee_maxrate *maxrate)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+
+	memcpy(maxrate, mlxsw_sp_port->dcb.maxrate, sizeof(*maxrate));
+
+	return 0;
+}
+
+static int mlxsw_sp_dcbnl_ieee_setmaxrate(struct net_device *dev,
+					  struct ieee_maxrate *maxrate)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+	struct ieee_maxrate *my_maxrate = mlxsw_sp_port->dcb.maxrate;
+	int err, i;
+
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) {
+		err = mlxsw_sp_port_ets_maxrate_set(mlxsw_sp_port,
+						    MLXSW_REG_QEEC_HR_SUBGROUP,
+						    i, 0,
+						    maxrate->tc_maxrate[i], 0);
+		if (err) {
+			netdev_err(dev, "Failed to set maxrate for TC %d\n", i);
+			goto err_port_ets_maxrate_set;
+		}
+	}
+
+	memcpy(mlxsw_sp_port->dcb.maxrate, maxrate, sizeof(*maxrate));
+
+	return 0;
+
+err_port_ets_maxrate_set:
+	for (i--; i >= 0; i--)
+		mlxsw_sp_port_ets_maxrate_set(mlxsw_sp_port,
+					      MLXSW_REG_QEEC_HR_SUBGROUP,
+					      i, 0,
+					      my_maxrate->tc_maxrate[i], 0);
+	return err;
+}
+
+static int mlxsw_sp_port_pfc_cnt_get(struct mlxsw_sp_port *mlxsw_sp_port,
+				     u8 prio)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	struct ieee_pfc *my_pfc = mlxsw_sp_port->dcb.pfc;
+	char ppcnt_pl[MLXSW_REG_PPCNT_LEN];
+	int err;
+
+	mlxsw_reg_ppcnt_pack(ppcnt_pl, mlxsw_sp_port->local_port,
+			     MLXSW_REG_PPCNT_PRIO_CNT, prio);
+	err = mlxsw_reg_query(mlxsw_sp->core, MLXSW_REG(ppcnt), ppcnt_pl);
+	if (err)
+		return err;
+
+	my_pfc->requests[prio] = mlxsw_reg_ppcnt_tx_pause_get(ppcnt_pl);
+	my_pfc->indications[prio] = mlxsw_reg_ppcnt_rx_pause_get(ppcnt_pl);
+
+	return 0;
+}
+
+static int mlxsw_sp_dcbnl_ieee_getpfc(struct net_device *dev,
+				      struct ieee_pfc *pfc)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+	int err, i;
+
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) {
+		err = mlxsw_sp_port_pfc_cnt_get(mlxsw_sp_port, i);
+		if (err) {
+			netdev_err(dev, "Failed to get PFC count for priority %d\n",
+				   i);
+			return err;
+		}
+	}
+
+	memcpy(pfc, mlxsw_sp_port->dcb.pfc, sizeof(*pfc));
+
+	return 0;
+}
+
+static int mlxsw_sp_port_pfc_set(struct mlxsw_sp_port *mlxsw_sp_port,
+				 struct ieee_pfc *pfc)
+{
+	char pfcc_pl[MLXSW_REG_PFCC_LEN];
+
+	mlxsw_reg_pfcc_pack(pfcc_pl, mlxsw_sp_port->local_port);
+	mlxsw_reg_pfcc_pprx_set(pfcc_pl, mlxsw_sp_port->link.rx_pause);
+	mlxsw_reg_pfcc_pptx_set(pfcc_pl, mlxsw_sp_port->link.tx_pause);
+	mlxsw_reg_pfcc_prio_pack(pfcc_pl, pfc->pfc_en);
+
+	return mlxsw_reg_write(mlxsw_sp_port->mlxsw_sp->core, MLXSW_REG(pfcc),
+			       pfcc_pl);
+}
+
+static int mlxsw_sp_dcbnl_ieee_setpfc(struct net_device *dev,
+				      struct ieee_pfc *pfc)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+	bool pause_en = mlxsw_sp_port_is_pause_en(mlxsw_sp_port);
+	struct mlxsw_sp_hdroom orig_hdroom;
+	struct mlxsw_sp_hdroom hdroom;
+	int prio;
+	int err;
+
+	if (pause_en && pfc->pfc_en) {
+		netdev_err(dev, "PAUSE frames already enabled on port\n");
+		return -EINVAL;
+	}
+
+	orig_hdroom = *mlxsw_sp_port->hdroom;
+
+	hdroom = orig_hdroom;
+	if (pfc->pfc_en)
+		hdroom.delay_bytes = DIV_ROUND_UP(pfc->delay, BITS_PER_BYTE);
+	else
+		hdroom.delay_bytes = 0;
+
+	for (prio = 0; prio < IEEE_8021QAZ_MAX_TCS; prio++)
+		hdroom.prios.prio[prio].lossy = !(pfc->pfc_en & BIT(prio));
+
+	mlxsw_sp_hdroom_bufs_reset_lossiness(&hdroom);
+	mlxsw_sp_hdroom_bufs_reset_sizes(mlxsw_sp_port, &hdroom);
+
+	err = mlxsw_sp_hdroom_configure(mlxsw_sp_port, &hdroom);
+	if (err) {
+		netdev_err(dev, "Failed to configure port's headroom for PFC\n");
+		return err;
+	}
+
+	err = mlxsw_sp_port_pfc_set(mlxsw_sp_port, pfc);
+	if (err) {
+		netdev_err(dev, "Failed to configure PFC\n");
+		goto err_port_pfc_set;
+	}
+
+	memcpy(mlxsw_sp_port->dcb.pfc, pfc, sizeof(*pfc));
+	mlxsw_sp_port->dcb.pfc->pfc_cap = IEEE_8021QAZ_MAX_TCS;
+
+	return 0;
+
+err_port_pfc_set:
+	mlxsw_sp_hdroom_configure(mlxsw_sp_port, &orig_hdroom);
+	return err;
+}
+
+static int mlxsw_sp_dcbnl_getbuffer(struct net_device *dev, struct dcbnl_buffer *buf)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+	struct mlxsw_sp_hdroom *hdroom = mlxsw_sp_port->hdroom;
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	int prio;
+	int i;
+
+	buf->total_size = 0;
+
+	BUILD_BUG_ON(DCBX_MAX_BUFFERS > MLXSW_SP_PB_COUNT);
+	for (i = 0; i < MLXSW_SP_PB_COUNT; i++) {
+		u32 bytes = mlxsw_sp_cells_bytes(mlxsw_sp, hdroom->bufs.buf[i].size_cells);
+
+		if (i < DCBX_MAX_BUFFERS)
+			buf->buffer_size[i] = bytes;
+		buf->total_size += bytes;
+	}
+
+	buf->total_size += mlxsw_sp_cells_bytes(mlxsw_sp, hdroom->int_buf.size_cells);
+
+	for (prio = 0; prio < IEEE_8021Q_MAX_PRIORITIES; prio++)
+		buf->prio2buffer[prio] = hdroom->prios.prio[prio].buf_idx;
+
+	return 0;
+}
+
+static int mlxsw_sp_dcbnl_setbuffer(struct net_device *dev, struct dcbnl_buffer *buf)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	struct mlxsw_sp_hdroom hdroom;
+	int prio;
+	int i;
+
+	hdroom = *mlxsw_sp_port->hdroom;
+
+	if (hdroom.mode != MLXSW_SP_HDROOM_MODE_TC) {
+		netdev_err(dev, "The use of dcbnl_setbuffer is only allowed if egress is configured using TC\n");
+		return -EINVAL;
+	}
+
+	for (prio = 0; prio < IEEE_8021Q_MAX_PRIORITIES; prio++)
+		hdroom.prios.prio[prio].set_buf_idx = buf->prio2buffer[prio];
+
+	BUILD_BUG_ON(DCBX_MAX_BUFFERS > MLXSW_SP_PB_COUNT);
+	for (i = 0; i < DCBX_MAX_BUFFERS; i++)
+		hdroom.bufs.buf[i].set_size_cells = mlxsw_sp_bytes_cells(mlxsw_sp,
+									 buf->buffer_size[i]);
+
+	mlxsw_sp_hdroom_prios_reset_buf_idx(&hdroom);
+	mlxsw_sp_hdroom_bufs_reset_lossiness(&hdroom);
+	mlxsw_sp_hdroom_bufs_reset_sizes(mlxsw_sp_port, &hdroom);
+	return mlxsw_sp_hdroom_configure(mlxsw_sp_port, &hdroom);
+}
+
+static const struct dcbnl_rtnl_ops mlxsw_sp_dcbnl_ops = {
+	.ieee_getets		= mlxsw_sp_dcbnl_ieee_getets,
+	.ieee_setets		= mlxsw_sp_dcbnl_ieee_setets,
+	.ieee_getmaxrate	= mlxsw_sp_dcbnl_ieee_getmaxrate,
+	.ieee_setmaxrate	= mlxsw_sp_dcbnl_ieee_setmaxrate,
+	.ieee_getpfc		= mlxsw_sp_dcbnl_ieee_getpfc,
+	.ieee_setpfc		= mlxsw_sp_dcbnl_ieee_setpfc,
+	.ieee_setapp		= mlxsw_sp_dcbnl_ieee_setapp,
+	.ieee_delapp		= mlxsw_sp_dcbnl_ieee_delapp,
+
+	.getdcbx		= mlxsw_sp_dcbnl_getdcbx,
+	.setdcbx		= mlxsw_sp_dcbnl_setdcbx,
+
+	.dcbnl_getbuffer	= mlxsw_sp_dcbnl_getbuffer,
+	.dcbnl_setbuffer	= mlxsw_sp_dcbnl_setbuffer,
+};
+
+static int mlxsw_sp_port_ets_init(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	mlxsw_sp_port->dcb.ets = kzalloc(sizeof(*mlxsw_sp_port->dcb.ets),
+					 GFP_KERNEL);
+	if (!mlxsw_sp_port->dcb.ets)
+		return -ENOMEM;
+
+	mlxsw_sp_port->dcb.ets->ets_cap = IEEE_8021QAZ_MAX_TCS;
+
+	return 0;
+}
+
+static void mlxsw_sp_port_ets_fini(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	kfree(mlxsw_sp_port->dcb.ets);
+}
+
+static int mlxsw_sp_port_maxrate_init(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	int i;
+
+	mlxsw_sp_port->dcb.maxrate = kmalloc(sizeof(*mlxsw_sp_port->dcb.maxrate),
+					     GFP_KERNEL);
+	if (!mlxsw_sp_port->dcb.maxrate)
+		return -ENOMEM;
+
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++)
+		mlxsw_sp_port->dcb.maxrate->tc_maxrate[i] = MLXSW_REG_QEEC_MAS_DIS;
+
+	return 0;
+}
+
+static void mlxsw_sp_port_maxrate_fini(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	kfree(mlxsw_sp_port->dcb.maxrate);
+}
+
+static int mlxsw_sp_port_pfc_init(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	mlxsw_sp_port->dcb.pfc = kzalloc(sizeof(*mlxsw_sp_port->dcb.pfc),
+					 GFP_KERNEL);
+	if (!mlxsw_sp_port->dcb.pfc)
+		return -ENOMEM;
+
+	mlxsw_sp_port->dcb.pfc->pfc_cap = IEEE_8021QAZ_MAX_TCS;
+
+	return 0;
+}
+
+static void mlxsw_sp_port_pfc_fini(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	kfree(mlxsw_sp_port->dcb.pfc);
+}
+
+int mlxsw_sp_port_dcb_init(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	int err;
+
+	err = mlxsw_sp_port_ets_init(mlxsw_sp_port);
+	if (err)
+		return err;
+	err = mlxsw_sp_port_maxrate_init(mlxsw_sp_port);
+	if (err)
+		goto err_port_maxrate_init;
+	err = mlxsw_sp_port_pfc_init(mlxsw_sp_port);
+	if (err)
+		goto err_port_pfc_init;
+
+	mlxsw_sp_port->dcb.trust_state = MLXSW_REG_QPTS_TRUST_STATE_PCP;
+	mlxsw_sp_port->dev->dcbnl_ops = &mlxsw_sp_dcbnl_ops;
+
+	return 0;
+
+err_port_pfc_init:
+	mlxsw_sp_port_maxrate_fini(mlxsw_sp_port);
+err_port_maxrate_init:
+	mlxsw_sp_port_ets_fini(mlxsw_sp_port);
+	return err;
+}
+
+void mlxsw_sp_port_dcb_fini(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	mlxsw_sp_port_pfc_fini(mlxsw_sp_port);
+	mlxsw_sp_port_maxrate_fini(mlxsw_sp_port);
+	mlxsw_sp_port_ets_fini(mlxsw_sp_port);
+}
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_dpipe.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_dpipe.c
new file mode 100644
index 000000000..5416093c0
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_dpipe.c
@@ -0,0 +1,1308 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2017-2018 Mellanox Technologies. All rights reserved */
+
+#include <linux/kernel.h>
+#include <linux/mutex.h>
+#include <net/devlink.h>
+
+#include "spectrum.h"
+#include "spectrum_dpipe.h"
+#include "spectrum_router.h"
+
+enum mlxsw_sp_field_metadata_id {
+	MLXSW_SP_DPIPE_FIELD_METADATA_ERIF_PORT,
+	MLXSW_SP_DPIPE_FIELD_METADATA_L3_FORWARD,
+	MLXSW_SP_DPIPE_FIELD_METADATA_L3_DROP,
+	MLXSW_SP_DPIPE_FIELD_METADATA_ADJ_INDEX,
+	MLXSW_SP_DPIPE_FIELD_METADATA_ADJ_SIZE,
+	MLXSW_SP_DPIPE_FIELD_METADATA_ADJ_HASH_INDEX,
+};
+
+static struct devlink_dpipe_field mlxsw_sp_dpipe_fields_metadata[] = {
+	{
+		.name = "erif_port",
+		.id = MLXSW_SP_DPIPE_FIELD_METADATA_ERIF_PORT,
+		.bitwidth = 32,
+		.mapping_type = DEVLINK_DPIPE_FIELD_MAPPING_TYPE_IFINDEX,
+	},
+	{
+		.name = "l3_forward",
+		.id = MLXSW_SP_DPIPE_FIELD_METADATA_L3_FORWARD,
+		.bitwidth = 1,
+	},
+	{
+		.name = "l3_drop",
+		.id = MLXSW_SP_DPIPE_FIELD_METADATA_L3_DROP,
+		.bitwidth = 1,
+	},
+	{
+		.name = "adj_index",
+		.id = MLXSW_SP_DPIPE_FIELD_METADATA_ADJ_INDEX,
+		.bitwidth = 32,
+	},
+	{
+		.name = "adj_size",
+		.id = MLXSW_SP_DPIPE_FIELD_METADATA_ADJ_SIZE,
+		.bitwidth = 32,
+	},
+	{
+		.name = "adj_hash_index",
+		.id = MLXSW_SP_DPIPE_FIELD_METADATA_ADJ_HASH_INDEX,
+		.bitwidth = 32,
+	},
+};
+
+enum mlxsw_sp_dpipe_header_id {
+	MLXSW_SP_DPIPE_HEADER_METADATA,
+};
+
+static struct devlink_dpipe_header mlxsw_sp_dpipe_header_metadata = {
+	.name = "mlxsw_meta",
+	.id = MLXSW_SP_DPIPE_HEADER_METADATA,
+	.fields = mlxsw_sp_dpipe_fields_metadata,
+	.fields_count = ARRAY_SIZE(mlxsw_sp_dpipe_fields_metadata),
+};
+
+static struct devlink_dpipe_header *mlxsw_dpipe_headers[] = {
+	&mlxsw_sp_dpipe_header_metadata,
+	&devlink_dpipe_header_ethernet,
+	&devlink_dpipe_header_ipv4,
+	&devlink_dpipe_header_ipv6,
+};
+
+static struct devlink_dpipe_headers mlxsw_sp_dpipe_headers = {
+	.headers = mlxsw_dpipe_headers,
+	.headers_count = ARRAY_SIZE(mlxsw_dpipe_headers),
+};
+
+static int mlxsw_sp_dpipe_table_erif_actions_dump(void *priv,
+						  struct sk_buff *skb)
+{
+	struct devlink_dpipe_action action = {0};
+	int err;
+
+	action.type = DEVLINK_DPIPE_ACTION_TYPE_FIELD_MODIFY;
+	action.header = &mlxsw_sp_dpipe_header_metadata;
+	action.field_id = MLXSW_SP_DPIPE_FIELD_METADATA_L3_FORWARD;
+
+	err = devlink_dpipe_action_put(skb, &action);
+	if (err)
+		return err;
+
+	action.type = DEVLINK_DPIPE_ACTION_TYPE_FIELD_MODIFY;
+	action.header = &mlxsw_sp_dpipe_header_metadata;
+	action.field_id = MLXSW_SP_DPIPE_FIELD_METADATA_L3_DROP;
+
+	return devlink_dpipe_action_put(skb, &action);
+}
+
+static int mlxsw_sp_dpipe_table_erif_matches_dump(void *priv,
+						  struct sk_buff *skb)
+{
+	struct devlink_dpipe_match match = {0};
+
+	match.type = DEVLINK_DPIPE_MATCH_TYPE_FIELD_EXACT;
+	match.header = &mlxsw_sp_dpipe_header_metadata;
+	match.field_id = MLXSW_SP_DPIPE_FIELD_METADATA_ERIF_PORT;
+
+	return devlink_dpipe_match_put(skb, &match);
+}
+
+static void
+mlxsw_sp_erif_match_action_prepare(struct devlink_dpipe_match *match,
+				   struct devlink_dpipe_action *action)
+{
+	action->type = DEVLINK_DPIPE_ACTION_TYPE_FIELD_MODIFY;
+	action->header = &mlxsw_sp_dpipe_header_metadata;
+	action->field_id = MLXSW_SP_DPIPE_FIELD_METADATA_L3_FORWARD;
+
+	match->type = DEVLINK_DPIPE_MATCH_TYPE_FIELD_EXACT;
+	match->header = &mlxsw_sp_dpipe_header_metadata;
+	match->field_id = MLXSW_SP_DPIPE_FIELD_METADATA_ERIF_PORT;
+}
+
+static int mlxsw_sp_erif_entry_prepare(struct devlink_dpipe_entry *entry,
+				       struct devlink_dpipe_value *match_value,
+				       struct devlink_dpipe_match *match,
+				       struct devlink_dpipe_value *action_value,
+				       struct devlink_dpipe_action *action)
+{
+	entry->match_values = match_value;
+	entry->match_values_count = 1;
+
+	entry->action_values = action_value;
+	entry->action_values_count = 1;
+
+	match_value->match = match;
+	match_value->value_size = sizeof(u32);
+	match_value->value = kmalloc(match_value->value_size, GFP_KERNEL);
+	if (!match_value->value)
+		return -ENOMEM;
+
+	action_value->action = action;
+	action_value->value_size = sizeof(u32);
+	action_value->value = kmalloc(action_value->value_size, GFP_KERNEL);
+	if (!action_value->value)
+		goto err_action_alloc;
+	return 0;
+
+err_action_alloc:
+	kfree(match_value->value);
+	return -ENOMEM;
+}
+
+static int mlxsw_sp_erif_entry_get(struct mlxsw_sp *mlxsw_sp,
+				   struct devlink_dpipe_entry *entry,
+				   struct mlxsw_sp_rif *rif,
+				   bool counters_enabled)
+{
+	u32 *action_value;
+	u32 *rif_value;
+	u64 cnt;
+	int err;
+
+	/* Set Match RIF index */
+	rif_value = entry->match_values->value;
+	*rif_value = mlxsw_sp_rif_index(rif);
+	entry->match_values->mapping_value = mlxsw_sp_rif_dev_ifindex(rif);
+	entry->match_values->mapping_valid = true;
+
+	/* Set Action Forwarding */
+	action_value = entry->action_values->value;
+	*action_value = 1;
+
+	entry->counter_valid = false;
+	entry->counter = 0;
+	entry->index = mlxsw_sp_rif_index(rif);
+
+	if (!counters_enabled)
+		return 0;
+
+	err = mlxsw_sp_rif_counter_value_get(mlxsw_sp, rif,
+					     MLXSW_SP_RIF_COUNTER_EGRESS,
+					     &cnt);
+	if (!err) {
+		entry->counter = cnt;
+		entry->counter_valid = true;
+	}
+	return 0;
+}
+
+static int
+mlxsw_sp_dpipe_table_erif_entries_dump(void *priv, bool counters_enabled,
+				       struct devlink_dpipe_dump_ctx *dump_ctx)
+{
+	struct devlink_dpipe_value match_value, action_value;
+	struct devlink_dpipe_action action = {0};
+	struct devlink_dpipe_match match = {0};
+	struct devlink_dpipe_entry entry = {0};
+	struct mlxsw_sp *mlxsw_sp = priv;
+	unsigned int rif_count;
+	int i, j;
+	int err;
+
+	memset(&match_value, 0, sizeof(match_value));
+	memset(&action_value, 0, sizeof(action_value));
+
+	mlxsw_sp_erif_match_action_prepare(&match, &action);
+	err = mlxsw_sp_erif_entry_prepare(&entry, &match_value, &match,
+					  &action_value, &action);
+	if (err)
+		return err;
+
+	rif_count = MLXSW_CORE_RES_GET(mlxsw_sp->core, MAX_RIFS);
+	mutex_lock(&mlxsw_sp->router->lock);
+	i = 0;
+start_again:
+	err = devlink_dpipe_entry_ctx_prepare(dump_ctx);
+	if (err)
+		goto err_ctx_prepare;
+	j = 0;
+	for (; i < rif_count; i++) {
+		struct mlxsw_sp_rif *rif = mlxsw_sp_rif_by_index(mlxsw_sp, i);
+
+		if (!rif || !mlxsw_sp_rif_dev(rif))
+			continue;
+		err = mlxsw_sp_erif_entry_get(mlxsw_sp, &entry, rif,
+					      counters_enabled);
+		if (err)
+			goto err_entry_get;
+		err = devlink_dpipe_entry_ctx_append(dump_ctx, &entry);
+		if (err) {
+			if (err == -EMSGSIZE) {
+				if (!j)
+					goto err_entry_append;
+				break;
+			}
+			goto err_entry_append;
+		}
+		j++;
+	}
+
+	devlink_dpipe_entry_ctx_close(dump_ctx);
+	if (i != rif_count)
+		goto start_again;
+	mutex_unlock(&mlxsw_sp->router->lock);
+
+	devlink_dpipe_entry_clear(&entry);
+	return 0;
+err_entry_append:
+err_entry_get:
+err_ctx_prepare:
+	mutex_unlock(&mlxsw_sp->router->lock);
+	devlink_dpipe_entry_clear(&entry);
+	return err;
+}
+
+static int mlxsw_sp_dpipe_table_erif_counters_update(void *priv, bool enable)
+{
+	struct mlxsw_sp *mlxsw_sp = priv;
+	int i;
+
+	mutex_lock(&mlxsw_sp->router->lock);
+	for (i = 0; i < MLXSW_CORE_RES_GET(mlxsw_sp->core, MAX_RIFS); i++) {
+		struct mlxsw_sp_rif *rif = mlxsw_sp_rif_by_index(mlxsw_sp, i);
+
+		if (!rif)
+			continue;
+		if (enable)
+			mlxsw_sp_rif_counter_alloc(rif,
+						   MLXSW_SP_RIF_COUNTER_EGRESS);
+		else
+			mlxsw_sp_rif_counter_free(rif,
+						  MLXSW_SP_RIF_COUNTER_EGRESS);
+	}
+	mutex_unlock(&mlxsw_sp->router->lock);
+	return 0;
+}
+
+static u64 mlxsw_sp_dpipe_table_erif_size_get(void *priv)
+{
+	struct mlxsw_sp *mlxsw_sp = priv;
+
+	return MLXSW_CORE_RES_GET(mlxsw_sp->core, MAX_RIFS);
+}
+
+static struct devlink_dpipe_table_ops mlxsw_sp_erif_ops = {
+	.matches_dump = mlxsw_sp_dpipe_table_erif_matches_dump,
+	.actions_dump = mlxsw_sp_dpipe_table_erif_actions_dump,
+	.entries_dump = mlxsw_sp_dpipe_table_erif_entries_dump,
+	.counters_set_update = mlxsw_sp_dpipe_table_erif_counters_update,
+	.size_get = mlxsw_sp_dpipe_table_erif_size_get,
+};
+
+static int mlxsw_sp_dpipe_erif_table_init(struct mlxsw_sp *mlxsw_sp)
+{
+	struct devlink *devlink = priv_to_devlink(mlxsw_sp->core);
+
+	return devl_dpipe_table_register(devlink,
+					 MLXSW_SP_DPIPE_TABLE_NAME_ERIF,
+					 &mlxsw_sp_erif_ops,
+					 mlxsw_sp, false);
+}
+
+static void mlxsw_sp_dpipe_erif_table_fini(struct mlxsw_sp *mlxsw_sp)
+{
+	struct devlink *devlink = priv_to_devlink(mlxsw_sp->core);
+
+	devl_dpipe_table_unregister(devlink, MLXSW_SP_DPIPE_TABLE_NAME_ERIF);
+}
+
+static int mlxsw_sp_dpipe_table_host_matches_dump(struct sk_buff *skb, int type)
+{
+	struct devlink_dpipe_match match = {0};
+	int err;
+
+	match.type = DEVLINK_DPIPE_MATCH_TYPE_FIELD_EXACT;
+	match.header = &mlxsw_sp_dpipe_header_metadata;
+	match.field_id = MLXSW_SP_DPIPE_FIELD_METADATA_ERIF_PORT;
+
+	err = devlink_dpipe_match_put(skb, &match);
+	if (err)
+		return err;
+
+	switch (type) {
+	case AF_INET:
+		match.type = DEVLINK_DPIPE_MATCH_TYPE_FIELD_EXACT;
+		match.header = &devlink_dpipe_header_ipv4;
+		match.field_id = DEVLINK_DPIPE_FIELD_IPV4_DST_IP;
+		break;
+	case AF_INET6:
+		match.type = DEVLINK_DPIPE_MATCH_TYPE_FIELD_EXACT;
+		match.header = &devlink_dpipe_header_ipv6;
+		match.field_id = DEVLINK_DPIPE_FIELD_IPV6_DST_IP;
+		break;
+	default:
+		WARN_ON(1);
+		return -EINVAL;
+	}
+
+	return devlink_dpipe_match_put(skb, &match);
+}
+
+static int
+mlxsw_sp_dpipe_table_host4_matches_dump(void *priv, struct sk_buff *skb)
+{
+	return mlxsw_sp_dpipe_table_host_matches_dump(skb, AF_INET);
+}
+
+static int
+mlxsw_sp_dpipe_table_host_actions_dump(void *priv, struct sk_buff *skb)
+{
+	struct devlink_dpipe_action action = {0};
+
+	action.type = DEVLINK_DPIPE_ACTION_TYPE_FIELD_MODIFY;
+	action.header = &devlink_dpipe_header_ethernet;
+	action.field_id = DEVLINK_DPIPE_FIELD_ETHERNET_DST_MAC;
+
+	return devlink_dpipe_action_put(skb, &action);
+}
+
+enum mlxsw_sp_dpipe_table_host_match {
+	MLXSW_SP_DPIPE_TABLE_HOST_MATCH_RIF,
+	MLXSW_SP_DPIPE_TABLE_HOST_MATCH_DIP,
+	MLXSW_SP_DPIPE_TABLE_HOST_MATCH_COUNT,
+};
+
+static void
+mlxsw_sp_dpipe_table_host_match_action_prepare(struct devlink_dpipe_match *matches,
+					       struct devlink_dpipe_action *action,
+					       int type)
+{
+	struct devlink_dpipe_match *match;
+
+	match = &matches[MLXSW_SP_DPIPE_TABLE_HOST_MATCH_RIF];
+	match->type = DEVLINK_DPIPE_MATCH_TYPE_FIELD_EXACT;
+	match->header = &mlxsw_sp_dpipe_header_metadata;
+	match->field_id = MLXSW_SP_DPIPE_FIELD_METADATA_ERIF_PORT;
+
+	match = &matches[MLXSW_SP_DPIPE_TABLE_HOST_MATCH_DIP];
+	match->type = DEVLINK_DPIPE_MATCH_TYPE_FIELD_EXACT;
+	switch (type) {
+	case AF_INET:
+		match->header = &devlink_dpipe_header_ipv4;
+		match->field_id = DEVLINK_DPIPE_FIELD_IPV4_DST_IP;
+		break;
+	case AF_INET6:
+		match->header = &devlink_dpipe_header_ipv6;
+		match->field_id = DEVLINK_DPIPE_FIELD_IPV6_DST_IP;
+		break;
+	default:
+		WARN_ON(1);
+		return;
+	}
+
+	action->type = DEVLINK_DPIPE_ACTION_TYPE_FIELD_MODIFY;
+	action->header = &devlink_dpipe_header_ethernet;
+	action->field_id = DEVLINK_DPIPE_FIELD_ETHERNET_DST_MAC;
+}
+
+static int
+mlxsw_sp_dpipe_table_host_entry_prepare(struct devlink_dpipe_entry *entry,
+					struct devlink_dpipe_value *match_values,
+					struct devlink_dpipe_match *matches,
+					struct devlink_dpipe_value *action_value,
+					struct devlink_dpipe_action *action,
+					int type)
+{
+	struct devlink_dpipe_value *match_value;
+	struct devlink_dpipe_match *match;
+
+	entry->match_values = match_values;
+	entry->match_values_count = MLXSW_SP_DPIPE_TABLE_HOST_MATCH_COUNT;
+
+	entry->action_values = action_value;
+	entry->action_values_count = 1;
+
+	match = &matches[MLXSW_SP_DPIPE_TABLE_HOST_MATCH_RIF];
+	match_value = &match_values[MLXSW_SP_DPIPE_TABLE_HOST_MATCH_RIF];
+
+	match_value->match = match;
+	match_value->value_size = sizeof(u32);
+	match_value->value = kmalloc(match_value->value_size, GFP_KERNEL);
+	if (!match_value->value)
+		return -ENOMEM;
+
+	match = &matches[MLXSW_SP_DPIPE_TABLE_HOST_MATCH_DIP];
+	match_value = &match_values[MLXSW_SP_DPIPE_TABLE_HOST_MATCH_DIP];
+
+	match_value->match = match;
+	switch (type) {
+	case AF_INET:
+		match_value->value_size = sizeof(u32);
+		break;
+	case AF_INET6:
+		match_value->value_size = sizeof(struct in6_addr);
+		break;
+	default:
+		WARN_ON(1);
+		return -EINVAL;
+	}
+
+	match_value->value = kmalloc(match_value->value_size, GFP_KERNEL);
+	if (!match_value->value)
+		return -ENOMEM;
+
+	action_value->action = action;
+	action_value->value_size = sizeof(u64);
+	action_value->value = kmalloc(action_value->value_size, GFP_KERNEL);
+	if (!action_value->value)
+		return -ENOMEM;
+
+	return 0;
+}
+
+static void
+__mlxsw_sp_dpipe_table_host_entry_fill(struct devlink_dpipe_entry *entry,
+				       struct mlxsw_sp_rif *rif,
+				       unsigned char *ha, void *dip)
+{
+	struct devlink_dpipe_value *value;
+	u32 *rif_value;
+	u8 *ha_value;
+
+	/* Set Match RIF index */
+	value = &entry->match_values[MLXSW_SP_DPIPE_TABLE_HOST_MATCH_RIF];
+
+	rif_value = value->value;
+	*rif_value = mlxsw_sp_rif_index(rif);
+	value->mapping_value = mlxsw_sp_rif_dev_ifindex(rif);
+	value->mapping_valid = true;
+
+	/* Set Match DIP */
+	value = &entry->match_values[MLXSW_SP_DPIPE_TABLE_HOST_MATCH_DIP];
+	memcpy(value->value, dip, value->value_size);
+
+	/* Set Action DMAC */
+	value = entry->action_values;
+	ha_value = value->value;
+	ether_addr_copy(ha_value, ha);
+}
+
+static void
+mlxsw_sp_dpipe_table_host4_entry_fill(struct devlink_dpipe_entry *entry,
+				      struct mlxsw_sp_neigh_entry *neigh_entry,
+				      struct mlxsw_sp_rif *rif)
+{
+	unsigned char *ha;
+	u32 dip;
+
+	ha = mlxsw_sp_neigh_entry_ha(neigh_entry);
+	dip = mlxsw_sp_neigh4_entry_dip(neigh_entry);
+	__mlxsw_sp_dpipe_table_host_entry_fill(entry, rif, ha, &dip);
+}
+
+static void
+mlxsw_sp_dpipe_table_host6_entry_fill(struct devlink_dpipe_entry *entry,
+				      struct mlxsw_sp_neigh_entry *neigh_entry,
+				      struct mlxsw_sp_rif *rif)
+{
+	struct in6_addr *dip;
+	unsigned char *ha;
+
+	ha = mlxsw_sp_neigh_entry_ha(neigh_entry);
+	dip = mlxsw_sp_neigh6_entry_dip(neigh_entry);
+
+	__mlxsw_sp_dpipe_table_host_entry_fill(entry, rif, ha, dip);
+}
+
+static void
+mlxsw_sp_dpipe_table_host_entry_fill(struct mlxsw_sp *mlxsw_sp,
+				     struct devlink_dpipe_entry *entry,
+				     struct mlxsw_sp_neigh_entry *neigh_entry,
+				     struct mlxsw_sp_rif *rif,
+				     int type)
+{
+	int err;
+
+	switch (type) {
+	case AF_INET:
+		mlxsw_sp_dpipe_table_host4_entry_fill(entry, neigh_entry, rif);
+		break;
+	case AF_INET6:
+		mlxsw_sp_dpipe_table_host6_entry_fill(entry, neigh_entry, rif);
+		break;
+	default:
+		WARN_ON(1);
+		return;
+	}
+
+	err = mlxsw_sp_neigh_counter_get(mlxsw_sp, neigh_entry,
+					 &entry->counter);
+	if (!err)
+		entry->counter_valid = true;
+}
+
+static int
+mlxsw_sp_dpipe_table_host_entries_get(struct mlxsw_sp *mlxsw_sp,
+				      struct devlink_dpipe_entry *entry,
+				      bool counters_enabled,
+				      struct devlink_dpipe_dump_ctx *dump_ctx,
+				      int type)
+{
+	int rif_neigh_count = 0;
+	int rif_neigh_skip = 0;
+	int neigh_count = 0;
+	int rif_count;
+	int i, j;
+	int err;
+
+	mutex_lock(&mlxsw_sp->router->lock);
+	i = 0;
+	rif_count = MLXSW_CORE_RES_GET(mlxsw_sp->core, MAX_RIFS);
+start_again:
+	err = devlink_dpipe_entry_ctx_prepare(dump_ctx);
+	if (err)
+		goto err_ctx_prepare;
+	j = 0;
+	rif_neigh_skip = rif_neigh_count;
+	for (; i < MLXSW_CORE_RES_GET(mlxsw_sp->core, MAX_RIFS); i++) {
+		struct mlxsw_sp_rif *rif = mlxsw_sp_rif_by_index(mlxsw_sp, i);
+		struct mlxsw_sp_neigh_entry *neigh_entry;
+
+		if (!rif)
+			continue;
+
+		rif_neigh_count = 0;
+		mlxsw_sp_rif_neigh_for_each(neigh_entry, rif) {
+			int neigh_type = mlxsw_sp_neigh_entry_type(neigh_entry);
+
+			if (neigh_type != type)
+				continue;
+
+			if (neigh_type == AF_INET6 &&
+			    mlxsw_sp_neigh_ipv6_ignore(neigh_entry))
+				continue;
+
+			if (rif_neigh_count < rif_neigh_skip)
+				goto skip;
+
+			mlxsw_sp_dpipe_table_host_entry_fill(mlxsw_sp, entry,
+							     neigh_entry, rif,
+							     type);
+			entry->index = neigh_count;
+			err = devlink_dpipe_entry_ctx_append(dump_ctx, entry);
+			if (err) {
+				if (err == -EMSGSIZE) {
+					if (!j)
+						goto err_entry_append;
+					else
+						goto out;
+				}
+				goto err_entry_append;
+			}
+			neigh_count++;
+			j++;
+skip:
+			rif_neigh_count++;
+		}
+		rif_neigh_skip = 0;
+	}
+out:
+	devlink_dpipe_entry_ctx_close(dump_ctx);
+	if (i != rif_count)
+		goto start_again;
+
+	mutex_unlock(&mlxsw_sp->router->lock);
+	return 0;
+
+err_ctx_prepare:
+err_entry_append:
+	mutex_unlock(&mlxsw_sp->router->lock);
+	return err;
+}
+
+static int
+mlxsw_sp_dpipe_table_host_entries_dump(struct mlxsw_sp *mlxsw_sp,
+				       bool counters_enabled,
+				       struct devlink_dpipe_dump_ctx *dump_ctx,
+				       int type)
+{
+	struct devlink_dpipe_value match_values[MLXSW_SP_DPIPE_TABLE_HOST_MATCH_COUNT];
+	struct devlink_dpipe_match matches[MLXSW_SP_DPIPE_TABLE_HOST_MATCH_COUNT];
+	struct devlink_dpipe_value action_value;
+	struct devlink_dpipe_action action = {0};
+	struct devlink_dpipe_entry entry = {0};
+	int err;
+
+	memset(matches, 0, MLXSW_SP_DPIPE_TABLE_HOST_MATCH_COUNT *
+			   sizeof(matches[0]));
+	memset(match_values, 0, MLXSW_SP_DPIPE_TABLE_HOST_MATCH_COUNT *
+				sizeof(match_values[0]));
+	memset(&action_value, 0, sizeof(action_value));
+
+	mlxsw_sp_dpipe_table_host_match_action_prepare(matches, &action, type);
+	err = mlxsw_sp_dpipe_table_host_entry_prepare(&entry, match_values,
+						      matches, &action_value,
+						      &action, type);
+	if (err)
+		goto out;
+
+	err = mlxsw_sp_dpipe_table_host_entries_get(mlxsw_sp, &entry,
+						    counters_enabled, dump_ctx,
+						    type);
+out:
+	devlink_dpipe_entry_clear(&entry);
+	return err;
+}
+
+static int
+mlxsw_sp_dpipe_table_host4_entries_dump(void *priv, bool counters_enabled,
+					struct devlink_dpipe_dump_ctx *dump_ctx)
+{
+	struct mlxsw_sp *mlxsw_sp = priv;
+
+	return mlxsw_sp_dpipe_table_host_entries_dump(mlxsw_sp,
+						      counters_enabled,
+						      dump_ctx, AF_INET);
+}
+
+static void
+mlxsw_sp_dpipe_table_host_counters_update(struct mlxsw_sp *mlxsw_sp,
+					  bool enable, int type)
+{
+	int i;
+
+	mutex_lock(&mlxsw_sp->router->lock);
+	for (i = 0; i < MLXSW_CORE_RES_GET(mlxsw_sp->core, MAX_RIFS); i++) {
+		struct mlxsw_sp_rif *rif = mlxsw_sp_rif_by_index(mlxsw_sp, i);
+		struct mlxsw_sp_neigh_entry *neigh_entry;
+
+		if (!rif)
+			continue;
+		mlxsw_sp_rif_neigh_for_each(neigh_entry, rif) {
+			int neigh_type = mlxsw_sp_neigh_entry_type(neigh_entry);
+
+			if (neigh_type != type)
+				continue;
+
+			if (neigh_type == AF_INET6 &&
+			    mlxsw_sp_neigh_ipv6_ignore(neigh_entry))
+				continue;
+
+			mlxsw_sp_neigh_entry_counter_update(mlxsw_sp,
+							    neigh_entry,
+							    enable);
+		}
+	}
+	mutex_unlock(&mlxsw_sp->router->lock);
+}
+
+static int mlxsw_sp_dpipe_table_host4_counters_update(void *priv, bool enable)
+{
+	struct mlxsw_sp *mlxsw_sp = priv;
+
+	mlxsw_sp_dpipe_table_host_counters_update(mlxsw_sp, enable, AF_INET);
+	return 0;
+}
+
+static u64
+mlxsw_sp_dpipe_table_host_size_get(struct mlxsw_sp *mlxsw_sp, int type)
+{
+	u64 size = 0;
+	int i;
+
+	mutex_lock(&mlxsw_sp->router->lock);
+	for (i = 0; i < MLXSW_CORE_RES_GET(mlxsw_sp->core, MAX_RIFS); i++) {
+		struct mlxsw_sp_rif *rif = mlxsw_sp_rif_by_index(mlxsw_sp, i);
+		struct mlxsw_sp_neigh_entry *neigh_entry;
+
+		if (!rif)
+			continue;
+		mlxsw_sp_rif_neigh_for_each(neigh_entry, rif) {
+			int neigh_type = mlxsw_sp_neigh_entry_type(neigh_entry);
+
+			if (neigh_type != type)
+				continue;
+
+			if (neigh_type == AF_INET6 &&
+			    mlxsw_sp_neigh_ipv6_ignore(neigh_entry))
+				continue;
+
+			size++;
+		}
+	}
+	mutex_unlock(&mlxsw_sp->router->lock);
+
+	return size;
+}
+
+static u64 mlxsw_sp_dpipe_table_host4_size_get(void *priv)
+{
+	struct mlxsw_sp *mlxsw_sp = priv;
+
+	return mlxsw_sp_dpipe_table_host_size_get(mlxsw_sp, AF_INET);
+}
+
+static struct devlink_dpipe_table_ops mlxsw_sp_host4_ops = {
+	.matches_dump = mlxsw_sp_dpipe_table_host4_matches_dump,
+	.actions_dump = mlxsw_sp_dpipe_table_host_actions_dump,
+	.entries_dump = mlxsw_sp_dpipe_table_host4_entries_dump,
+	.counters_set_update = mlxsw_sp_dpipe_table_host4_counters_update,
+	.size_get = mlxsw_sp_dpipe_table_host4_size_get,
+};
+
+#define MLXSW_SP_DPIPE_TABLE_RESOURCE_UNIT_HOST4 1
+
+static int mlxsw_sp_dpipe_host4_table_init(struct mlxsw_sp *mlxsw_sp)
+{
+	struct devlink *devlink = priv_to_devlink(mlxsw_sp->core);
+	int err;
+
+	err = devl_dpipe_table_register(devlink,
+					MLXSW_SP_DPIPE_TABLE_NAME_HOST4,
+					&mlxsw_sp_host4_ops,
+					mlxsw_sp, false);
+	if (err)
+		return err;
+
+	err = devl_dpipe_table_resource_set(devlink,
+					    MLXSW_SP_DPIPE_TABLE_NAME_HOST4,
+					    MLXSW_SP_RESOURCE_KVD_HASH_SINGLE,
+					    MLXSW_SP_DPIPE_TABLE_RESOURCE_UNIT_HOST4);
+	if (err)
+		goto err_resource_set;
+
+	return 0;
+
+err_resource_set:
+	devl_dpipe_table_unregister(devlink,
+				    MLXSW_SP_DPIPE_TABLE_NAME_HOST4);
+	return err;
+}
+
+static void mlxsw_sp_dpipe_host4_table_fini(struct mlxsw_sp *mlxsw_sp)
+{
+	struct devlink *devlink = priv_to_devlink(mlxsw_sp->core);
+
+	devl_dpipe_table_unregister(devlink,
+				    MLXSW_SP_DPIPE_TABLE_NAME_HOST4);
+}
+
+static int
+mlxsw_sp_dpipe_table_host6_matches_dump(void *priv, struct sk_buff *skb)
+{
+	return mlxsw_sp_dpipe_table_host_matches_dump(skb, AF_INET6);
+}
+
+static int
+mlxsw_sp_dpipe_table_host6_entries_dump(void *priv, bool counters_enabled,
+					struct devlink_dpipe_dump_ctx *dump_ctx)
+{
+	struct mlxsw_sp *mlxsw_sp = priv;
+
+	return mlxsw_sp_dpipe_table_host_entries_dump(mlxsw_sp,
+						      counters_enabled,
+						      dump_ctx, AF_INET6);
+}
+
+static int mlxsw_sp_dpipe_table_host6_counters_update(void *priv, bool enable)
+{
+	struct mlxsw_sp *mlxsw_sp = priv;
+
+	mlxsw_sp_dpipe_table_host_counters_update(mlxsw_sp, enable, AF_INET6);
+	return 0;
+}
+
+static u64 mlxsw_sp_dpipe_table_host6_size_get(void *priv)
+{
+	struct mlxsw_sp *mlxsw_sp = priv;
+
+	return mlxsw_sp_dpipe_table_host_size_get(mlxsw_sp, AF_INET6);
+}
+
+static struct devlink_dpipe_table_ops mlxsw_sp_host6_ops = {
+	.matches_dump = mlxsw_sp_dpipe_table_host6_matches_dump,
+	.actions_dump = mlxsw_sp_dpipe_table_host_actions_dump,
+	.entries_dump = mlxsw_sp_dpipe_table_host6_entries_dump,
+	.counters_set_update = mlxsw_sp_dpipe_table_host6_counters_update,
+	.size_get = mlxsw_sp_dpipe_table_host6_size_get,
+};
+
+#define MLXSW_SP_DPIPE_TABLE_RESOURCE_UNIT_HOST6 2
+
+static int mlxsw_sp_dpipe_host6_table_init(struct mlxsw_sp *mlxsw_sp)
+{
+	struct devlink *devlink = priv_to_devlink(mlxsw_sp->core);
+	int err;
+
+	err = devl_dpipe_table_register(devlink,
+					MLXSW_SP_DPIPE_TABLE_NAME_HOST6,
+					&mlxsw_sp_host6_ops,
+					mlxsw_sp, false);
+	if (err)
+		return err;
+
+	err = devl_dpipe_table_resource_set(devlink,
+					    MLXSW_SP_DPIPE_TABLE_NAME_HOST6,
+					    MLXSW_SP_RESOURCE_KVD_HASH_DOUBLE,
+					    MLXSW_SP_DPIPE_TABLE_RESOURCE_UNIT_HOST6);
+	if (err)
+		goto err_resource_set;
+
+	return 0;
+
+err_resource_set:
+	devl_dpipe_table_unregister(devlink,
+				    MLXSW_SP_DPIPE_TABLE_NAME_HOST6);
+	return err;
+}
+
+static void mlxsw_sp_dpipe_host6_table_fini(struct mlxsw_sp *mlxsw_sp)
+{
+	struct devlink *devlink = priv_to_devlink(mlxsw_sp->core);
+
+	devl_dpipe_table_unregister(devlink,
+				    MLXSW_SP_DPIPE_TABLE_NAME_HOST6);
+}
+
+static int mlxsw_sp_dpipe_table_adj_matches_dump(void *priv,
+						 struct sk_buff *skb)
+{
+	struct devlink_dpipe_match match = {0};
+	int err;
+
+	match.type = DEVLINK_DPIPE_MATCH_TYPE_FIELD_EXACT;
+	match.header = &mlxsw_sp_dpipe_header_metadata;
+	match.field_id = MLXSW_SP_DPIPE_FIELD_METADATA_ADJ_INDEX;
+
+	err = devlink_dpipe_match_put(skb, &match);
+	if (err)
+		return err;
+
+	match.type = DEVLINK_DPIPE_MATCH_TYPE_FIELD_EXACT;
+	match.header = &mlxsw_sp_dpipe_header_metadata;
+	match.field_id = MLXSW_SP_DPIPE_FIELD_METADATA_ADJ_SIZE;
+
+	err = devlink_dpipe_match_put(skb, &match);
+	if (err)
+		return err;
+
+	match.type = DEVLINK_DPIPE_MATCH_TYPE_FIELD_EXACT;
+	match.header = &mlxsw_sp_dpipe_header_metadata;
+	match.field_id = MLXSW_SP_DPIPE_FIELD_METADATA_ADJ_HASH_INDEX;
+
+	return devlink_dpipe_match_put(skb, &match);
+}
+
+static int mlxsw_sp_dpipe_table_adj_actions_dump(void *priv,
+						 struct sk_buff *skb)
+{
+	struct devlink_dpipe_action action = {0};
+	int err;
+
+	action.type = DEVLINK_DPIPE_ACTION_TYPE_FIELD_MODIFY;
+	action.header = &devlink_dpipe_header_ethernet;
+	action.field_id = DEVLINK_DPIPE_FIELD_ETHERNET_DST_MAC;
+
+	err = devlink_dpipe_action_put(skb, &action);
+	if (err)
+		return err;
+
+	action.type = DEVLINK_DPIPE_ACTION_TYPE_FIELD_MODIFY;
+	action.header = &mlxsw_sp_dpipe_header_metadata;
+	action.field_id = MLXSW_SP_DPIPE_FIELD_METADATA_ERIF_PORT;
+
+	return devlink_dpipe_action_put(skb, &action);
+}
+
+static u64 mlxsw_sp_dpipe_table_adj_size(struct mlxsw_sp *mlxsw_sp)
+{
+	struct mlxsw_sp_nexthop *nh;
+	u64 size = 0;
+
+	mlxsw_sp_nexthop_for_each(nh, mlxsw_sp->router)
+		if (mlxsw_sp_nexthop_is_forward(nh) &&
+		    !mlxsw_sp_nexthop_group_has_ipip(nh))
+			size++;
+	return size;
+}
+
+enum mlxsw_sp_dpipe_table_adj_match {
+	MLXSW_SP_DPIPE_TABLE_ADJ_MATCH_INDEX,
+	MLXSW_SP_DPIPE_TABLE_ADJ_MATCH_SIZE,
+	MLXSW_SP_DPIPE_TABLE_ADJ_MATCH_HASH_INDEX,
+	MLXSW_SP_DPIPE_TABLE_ADJ_MATCH_COUNT,
+};
+
+enum mlxsw_sp_dpipe_table_adj_action {
+	MLXSW_SP_DPIPE_TABLE_ADJ_ACTION_DST_MAC,
+	MLXSW_SP_DPIPE_TABLE_ADJ_ACTION_ERIF_PORT,
+	MLXSW_SP_DPIPE_TABLE_ADJ_ACTION_COUNT,
+};
+
+static void
+mlxsw_sp_dpipe_table_adj_match_action_prepare(struct devlink_dpipe_match *matches,
+					      struct devlink_dpipe_action *actions)
+{
+	struct devlink_dpipe_action *action;
+	struct devlink_dpipe_match *match;
+
+	match = &matches[MLXSW_SP_DPIPE_TABLE_ADJ_MATCH_INDEX];
+	match->type = DEVLINK_DPIPE_MATCH_TYPE_FIELD_EXACT;
+	match->header = &mlxsw_sp_dpipe_header_metadata;
+	match->field_id = MLXSW_SP_DPIPE_FIELD_METADATA_ADJ_INDEX;
+
+	match = &matches[MLXSW_SP_DPIPE_TABLE_ADJ_MATCH_SIZE];
+	match->type = DEVLINK_DPIPE_MATCH_TYPE_FIELD_EXACT;
+	match->header = &mlxsw_sp_dpipe_header_metadata;
+	match->field_id = MLXSW_SP_DPIPE_FIELD_METADATA_ADJ_SIZE;
+
+	match = &matches[MLXSW_SP_DPIPE_TABLE_ADJ_MATCH_HASH_INDEX];
+	match->type = DEVLINK_DPIPE_MATCH_TYPE_FIELD_EXACT;
+	match->header = &mlxsw_sp_dpipe_header_metadata;
+	match->field_id = MLXSW_SP_DPIPE_FIELD_METADATA_ADJ_HASH_INDEX;
+
+	action = &actions[MLXSW_SP_DPIPE_TABLE_ADJ_ACTION_DST_MAC];
+	action->type = DEVLINK_DPIPE_ACTION_TYPE_FIELD_MODIFY;
+	action->header = &devlink_dpipe_header_ethernet;
+	action->field_id = DEVLINK_DPIPE_FIELD_ETHERNET_DST_MAC;
+
+	action = &actions[MLXSW_SP_DPIPE_TABLE_ADJ_ACTION_ERIF_PORT];
+	action->type = DEVLINK_DPIPE_ACTION_TYPE_FIELD_MODIFY;
+	action->header = &mlxsw_sp_dpipe_header_metadata;
+	action->field_id = MLXSW_SP_DPIPE_FIELD_METADATA_ERIF_PORT;
+}
+
+static int
+mlxsw_sp_dpipe_table_adj_entry_prepare(struct devlink_dpipe_entry *entry,
+				       struct devlink_dpipe_value *match_values,
+				       struct devlink_dpipe_match *matches,
+				       struct devlink_dpipe_value *action_values,
+				       struct devlink_dpipe_action *actions)
+{	struct devlink_dpipe_value *action_value;
+	struct devlink_dpipe_value *match_value;
+	struct devlink_dpipe_action *action;
+	struct devlink_dpipe_match *match;
+
+	entry->match_values = match_values;
+	entry->match_values_count = MLXSW_SP_DPIPE_TABLE_ADJ_MATCH_COUNT;
+
+	entry->action_values = action_values;
+	entry->action_values_count = MLXSW_SP_DPIPE_TABLE_ADJ_ACTION_COUNT;
+
+	match = &matches[MLXSW_SP_DPIPE_TABLE_ADJ_MATCH_INDEX];
+	match_value = &match_values[MLXSW_SP_DPIPE_TABLE_ADJ_MATCH_INDEX];
+
+	match_value->match = match;
+	match_value->value_size = sizeof(u32);
+	match_value->value = kmalloc(match_value->value_size, GFP_KERNEL);
+	if (!match_value->value)
+		return -ENOMEM;
+
+	match = &matches[MLXSW_SP_DPIPE_TABLE_ADJ_MATCH_SIZE];
+	match_value = &match_values[MLXSW_SP_DPIPE_TABLE_ADJ_MATCH_SIZE];
+
+	match_value->match = match;
+	match_value->value_size = sizeof(u32);
+	match_value->value = kmalloc(match_value->value_size, GFP_KERNEL);
+	if (!match_value->value)
+		return -ENOMEM;
+
+	match = &matches[MLXSW_SP_DPIPE_TABLE_ADJ_MATCH_HASH_INDEX];
+	match_value = &match_values[MLXSW_SP_DPIPE_TABLE_ADJ_MATCH_HASH_INDEX];
+
+	match_value->match = match;
+	match_value->value_size = sizeof(u32);
+	match_value->value = kmalloc(match_value->value_size, GFP_KERNEL);
+	if (!match_value->value)
+		return -ENOMEM;
+
+	action = &actions[MLXSW_SP_DPIPE_TABLE_ADJ_ACTION_DST_MAC];
+	action_value = &action_values[MLXSW_SP_DPIPE_TABLE_ADJ_ACTION_DST_MAC];
+
+	action_value->action = action;
+	action_value->value_size = sizeof(u64);
+	action_value->value = kmalloc(action_value->value_size, GFP_KERNEL);
+	if (!action_value->value)
+		return -ENOMEM;
+
+	action = &actions[MLXSW_SP_DPIPE_TABLE_ADJ_ACTION_ERIF_PORT];
+	action_value = &action_values[MLXSW_SP_DPIPE_TABLE_ADJ_ACTION_ERIF_PORT];
+
+	action_value->action = action;
+	action_value->value_size = sizeof(u32);
+	action_value->value = kmalloc(action_value->value_size, GFP_KERNEL);
+	if (!action_value->value)
+		return -ENOMEM;
+
+	return 0;
+}
+
+static void
+__mlxsw_sp_dpipe_table_adj_entry_fill(struct devlink_dpipe_entry *entry,
+				      u32 adj_index, u32 adj_size,
+				      u32 adj_hash_index, unsigned char *ha,
+				      struct mlxsw_sp_rif *rif)
+{
+	struct devlink_dpipe_value *value;
+	u32 *p_rif_value;
+	u32 *p_index;
+
+	value = &entry->match_values[MLXSW_SP_DPIPE_TABLE_ADJ_MATCH_INDEX];
+	p_index = value->value;
+	*p_index = adj_index;
+
+	value = &entry->match_values[MLXSW_SP_DPIPE_TABLE_ADJ_MATCH_SIZE];
+	p_index = value->value;
+	*p_index = adj_size;
+
+	value = &entry->match_values[MLXSW_SP_DPIPE_TABLE_ADJ_MATCH_HASH_INDEX];
+	p_index = value->value;
+	*p_index = adj_hash_index;
+
+	value = &entry->action_values[MLXSW_SP_DPIPE_TABLE_ADJ_ACTION_DST_MAC];
+	ether_addr_copy(value->value, ha);
+
+	value = &entry->action_values[MLXSW_SP_DPIPE_TABLE_ADJ_ACTION_ERIF_PORT];
+	p_rif_value = value->value;
+	*p_rif_value = mlxsw_sp_rif_index(rif);
+	value->mapping_value = mlxsw_sp_rif_dev_ifindex(rif);
+	value->mapping_valid = true;
+}
+
+static void mlxsw_sp_dpipe_table_adj_entry_fill(struct mlxsw_sp *mlxsw_sp,
+						struct mlxsw_sp_nexthop *nh,
+						struct devlink_dpipe_entry *entry)
+{
+	struct mlxsw_sp_rif *rif = mlxsw_sp_nexthop_rif(nh);
+	unsigned char *ha = mlxsw_sp_nexthop_ha(nh);
+	u32 adj_hash_index = 0;
+	u32 adj_index = 0;
+	u32 adj_size = 0;
+	int err;
+
+	mlxsw_sp_nexthop_indexes(nh, &adj_index, &adj_size, &adj_hash_index);
+	__mlxsw_sp_dpipe_table_adj_entry_fill(entry, adj_index, adj_size,
+					      adj_hash_index, ha, rif);
+	err = mlxsw_sp_nexthop_counter_get(mlxsw_sp, nh, &entry->counter);
+	if (!err)
+		entry->counter_valid = true;
+}
+
+static int
+mlxsw_sp_dpipe_table_adj_entries_get(struct mlxsw_sp *mlxsw_sp,
+				     struct devlink_dpipe_entry *entry,
+				     bool counters_enabled,
+				     struct devlink_dpipe_dump_ctx *dump_ctx)
+{
+	struct mlxsw_sp_nexthop *nh;
+	int entry_index = 0;
+	int nh_count_max;
+	int nh_count = 0;
+	int nh_skip;
+	int j;
+	int err;
+
+	mutex_lock(&mlxsw_sp->router->lock);
+	nh_count_max = mlxsw_sp_dpipe_table_adj_size(mlxsw_sp);
+start_again:
+	err = devlink_dpipe_entry_ctx_prepare(dump_ctx);
+	if (err)
+		goto err_ctx_prepare;
+	j = 0;
+	nh_skip = nh_count;
+	nh_count = 0;
+	mlxsw_sp_nexthop_for_each(nh, mlxsw_sp->router) {
+		if (!mlxsw_sp_nexthop_is_forward(nh) ||
+		    mlxsw_sp_nexthop_group_has_ipip(nh))
+			continue;
+
+		if (nh_count < nh_skip)
+			goto skip;
+
+		mlxsw_sp_dpipe_table_adj_entry_fill(mlxsw_sp, nh, entry);
+		entry->index = entry_index;
+		err = devlink_dpipe_entry_ctx_append(dump_ctx, entry);
+		if (err) {
+			if (err == -EMSGSIZE) {
+				if (!j)
+					goto err_entry_append;
+				break;
+			}
+			goto err_entry_append;
+		}
+		entry_index++;
+		j++;
+skip:
+		nh_count++;
+	}
+
+	devlink_dpipe_entry_ctx_close(dump_ctx);
+	if (nh_count != nh_count_max)
+		goto start_again;
+	mutex_unlock(&mlxsw_sp->router->lock);
+
+	return 0;
+
+err_ctx_prepare:
+err_entry_append:
+	mutex_unlock(&mlxsw_sp->router->lock);
+	return err;
+}
+
+static int
+mlxsw_sp_dpipe_table_adj_entries_dump(void *priv, bool counters_enabled,
+				      struct devlink_dpipe_dump_ctx *dump_ctx)
+{
+	struct devlink_dpipe_value action_values[MLXSW_SP_DPIPE_TABLE_ADJ_ACTION_COUNT];
+	struct devlink_dpipe_value match_values[MLXSW_SP_DPIPE_TABLE_ADJ_MATCH_COUNT];
+	struct devlink_dpipe_action actions[MLXSW_SP_DPIPE_TABLE_ADJ_ACTION_COUNT];
+	struct devlink_dpipe_match matches[MLXSW_SP_DPIPE_TABLE_ADJ_MATCH_COUNT];
+	struct devlink_dpipe_entry entry = {0};
+	struct mlxsw_sp *mlxsw_sp = priv;
+	int err;
+
+	memset(matches, 0, MLXSW_SP_DPIPE_TABLE_ADJ_MATCH_COUNT *
+			   sizeof(matches[0]));
+	memset(match_values, 0, MLXSW_SP_DPIPE_TABLE_ADJ_MATCH_COUNT *
+				sizeof(match_values[0]));
+	memset(actions, 0, MLXSW_SP_DPIPE_TABLE_ADJ_ACTION_COUNT *
+			   sizeof(actions[0]));
+	memset(action_values, 0, MLXSW_SP_DPIPE_TABLE_ADJ_ACTION_COUNT *
+				 sizeof(action_values[0]));
+
+	mlxsw_sp_dpipe_table_adj_match_action_prepare(matches, actions);
+	err = mlxsw_sp_dpipe_table_adj_entry_prepare(&entry,
+						     match_values, matches,
+						     action_values, actions);
+	if (err)
+		goto out;
+
+	err = mlxsw_sp_dpipe_table_adj_entries_get(mlxsw_sp, &entry,
+						   counters_enabled, dump_ctx);
+out:
+	devlink_dpipe_entry_clear(&entry);
+	return err;
+}
+
+static int mlxsw_sp_dpipe_table_adj_counters_update(void *priv, bool enable)
+{
+	char ratr_pl[MLXSW_REG_RATR_LEN];
+	struct mlxsw_sp *mlxsw_sp = priv;
+	struct mlxsw_sp_nexthop *nh;
+	u32 adj_hash_index = 0;
+	u32 adj_index = 0;
+	u32 adj_size = 0;
+
+	mlxsw_sp_nexthop_for_each(nh, mlxsw_sp->router) {
+		if (!mlxsw_sp_nexthop_is_forward(nh) ||
+		    mlxsw_sp_nexthop_group_has_ipip(nh))
+			continue;
+
+		mlxsw_sp_nexthop_indexes(nh, &adj_index, &adj_size,
+					 &adj_hash_index);
+		if (enable)
+			mlxsw_sp_nexthop_counter_alloc(mlxsw_sp, nh);
+		else
+			mlxsw_sp_nexthop_counter_free(mlxsw_sp, nh);
+		mlxsw_sp_nexthop_eth_update(mlxsw_sp,
+					    adj_index + adj_hash_index, nh,
+					    true, ratr_pl);
+	}
+	return 0;
+}
+
+static u64
+mlxsw_sp_dpipe_table_adj_size_get(void *priv)
+{
+	struct mlxsw_sp *mlxsw_sp = priv;
+	u64 size;
+
+	mutex_lock(&mlxsw_sp->router->lock);
+	size = mlxsw_sp_dpipe_table_adj_size(mlxsw_sp);
+	mutex_unlock(&mlxsw_sp->router->lock);
+
+	return size;
+}
+
+static struct devlink_dpipe_table_ops mlxsw_sp_dpipe_table_adj_ops = {
+	.matches_dump = mlxsw_sp_dpipe_table_adj_matches_dump,
+	.actions_dump = mlxsw_sp_dpipe_table_adj_actions_dump,
+	.entries_dump = mlxsw_sp_dpipe_table_adj_entries_dump,
+	.counters_set_update = mlxsw_sp_dpipe_table_adj_counters_update,
+	.size_get = mlxsw_sp_dpipe_table_adj_size_get,
+};
+
+#define MLXSW_SP_DPIPE_TABLE_RESOURCE_UNIT_ADJ 1
+
+static int mlxsw_sp_dpipe_adj_table_init(struct mlxsw_sp *mlxsw_sp)
+{
+	struct devlink *devlink = priv_to_devlink(mlxsw_sp->core);
+	int err;
+
+	err = devl_dpipe_table_register(devlink,
+					MLXSW_SP_DPIPE_TABLE_NAME_ADJ,
+					&mlxsw_sp_dpipe_table_adj_ops,
+					mlxsw_sp, false);
+	if (err)
+		return err;
+
+	err = devl_dpipe_table_resource_set(devlink,
+					    MLXSW_SP_DPIPE_TABLE_NAME_ADJ,
+					    MLXSW_SP_RESOURCE_KVD_LINEAR,
+					    MLXSW_SP_DPIPE_TABLE_RESOURCE_UNIT_ADJ);
+	if (err)
+		goto err_resource_set;
+
+	return 0;
+
+err_resource_set:
+	devl_dpipe_table_unregister(devlink,
+				    MLXSW_SP_DPIPE_TABLE_NAME_ADJ);
+	return err;
+}
+
+static void mlxsw_sp_dpipe_adj_table_fini(struct mlxsw_sp *mlxsw_sp)
+{
+	struct devlink *devlink = priv_to_devlink(mlxsw_sp->core);
+
+	devl_dpipe_table_unregister(devlink,
+				    MLXSW_SP_DPIPE_TABLE_NAME_ADJ);
+}
+
+int mlxsw_sp_dpipe_init(struct mlxsw_sp *mlxsw_sp)
+{
+	struct devlink *devlink = priv_to_devlink(mlxsw_sp->core);
+	int err;
+
+	devl_dpipe_headers_register(devlink, &mlxsw_sp_dpipe_headers);
+
+	err = mlxsw_sp_dpipe_erif_table_init(mlxsw_sp);
+	if (err)
+		goto err_erif_table_init;
+
+	err = mlxsw_sp_dpipe_host4_table_init(mlxsw_sp);
+	if (err)
+		goto err_host4_table_init;
+
+	err = mlxsw_sp_dpipe_host6_table_init(mlxsw_sp);
+	if (err)
+		goto err_host6_table_init;
+
+	err = mlxsw_sp_dpipe_adj_table_init(mlxsw_sp);
+	if (err)
+		goto err_adj_table_init;
+
+	return 0;
+err_adj_table_init:
+	mlxsw_sp_dpipe_host6_table_fini(mlxsw_sp);
+err_host6_table_init:
+	mlxsw_sp_dpipe_host4_table_fini(mlxsw_sp);
+err_host4_table_init:
+	mlxsw_sp_dpipe_erif_table_fini(mlxsw_sp);
+err_erif_table_init:
+	devl_dpipe_headers_unregister(priv_to_devlink(mlxsw_sp->core));
+	return err;
+}
+
+void mlxsw_sp_dpipe_fini(struct mlxsw_sp *mlxsw_sp)
+{
+	struct devlink *devlink = priv_to_devlink(mlxsw_sp->core);
+
+	mlxsw_sp_dpipe_adj_table_fini(mlxsw_sp);
+	mlxsw_sp_dpipe_host6_table_fini(mlxsw_sp);
+	mlxsw_sp_dpipe_host4_table_fini(mlxsw_sp);
+	mlxsw_sp_dpipe_erif_table_fini(mlxsw_sp);
+	devl_dpipe_headers_unregister(devlink);
+}
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_dpipe.h b/drivers/net/ethernet/mellanox/mlxsw/spectrum_dpipe.h
new file mode 100644
index 000000000..246dbb3c0
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_dpipe.h
@@ -0,0 +1,15 @@
+/* SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0 */
+/* Copyright (c) 2017-2018 Mellanox Technologies. All rights reserved */
+
+#ifndef _MLXSW_PIPELINE_H_
+#define _MLXSW_PIPELINE_H_
+
+int mlxsw_sp_dpipe_init(struct mlxsw_sp *mlxsw_sp);
+void mlxsw_sp_dpipe_fini(struct mlxsw_sp *mlxsw_sp);
+
+#define MLXSW_SP_DPIPE_TABLE_NAME_ERIF "mlxsw_erif"
+#define MLXSW_SP_DPIPE_TABLE_NAME_HOST4 "mlxsw_host4"
+#define MLXSW_SP_DPIPE_TABLE_NAME_HOST6 "mlxsw_host6"
+#define MLXSW_SP_DPIPE_TABLE_NAME_ADJ "mlxsw_adj"
+
+#endif /* _MLXSW_PIPELINE_H_*/
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_ethtool.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_ethtool.c
new file mode 100644
index 000000000..dcd79d7e2
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_ethtool.c
@@ -0,0 +1,2023 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2020 Mellanox Technologies. All rights reserved */
+
+#include "reg.h"
+#include "core.h"
+#include "spectrum.h"
+#include "core_env.h"
+
+static const char mlxsw_sp_driver_version[] = "1.0";
+
+static void mlxsw_sp_port_get_drvinfo(struct net_device *dev,
+				      struct ethtool_drvinfo *drvinfo)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+
+	strscpy(drvinfo->driver, mlxsw_sp->bus_info->device_kind,
+		sizeof(drvinfo->driver));
+	strscpy(drvinfo->version, mlxsw_sp_driver_version,
+		sizeof(drvinfo->version));
+	snprintf(drvinfo->fw_version, sizeof(drvinfo->fw_version),
+		 "%d.%d.%d",
+		 mlxsw_sp->bus_info->fw_rev.major,
+		 mlxsw_sp->bus_info->fw_rev.minor,
+		 mlxsw_sp->bus_info->fw_rev.subminor);
+	strscpy(drvinfo->bus_info, mlxsw_sp->bus_info->device_name,
+		sizeof(drvinfo->bus_info));
+}
+
+struct mlxsw_sp_ethtool_link_ext_state_opcode_mapping {
+	u32 status_opcode;
+	enum ethtool_link_ext_state link_ext_state;
+	u8 link_ext_substate;
+};
+
+static const struct mlxsw_sp_ethtool_link_ext_state_opcode_mapping
+mlxsw_sp_link_ext_state_opcode_map[] = {
+	{2, ETHTOOL_LINK_EXT_STATE_AUTONEG,
+		ETHTOOL_LINK_EXT_SUBSTATE_AN_NO_PARTNER_DETECTED},
+	{3, ETHTOOL_LINK_EXT_STATE_AUTONEG,
+		ETHTOOL_LINK_EXT_SUBSTATE_AN_ACK_NOT_RECEIVED},
+	{4, ETHTOOL_LINK_EXT_STATE_AUTONEG,
+		ETHTOOL_LINK_EXT_SUBSTATE_AN_NEXT_PAGE_EXCHANGE_FAILED},
+	{36, ETHTOOL_LINK_EXT_STATE_AUTONEG,
+		ETHTOOL_LINK_EXT_SUBSTATE_AN_NO_PARTNER_DETECTED_FORCE_MODE},
+	{38, ETHTOOL_LINK_EXT_STATE_AUTONEG,
+		ETHTOOL_LINK_EXT_SUBSTATE_AN_FEC_MISMATCH_DURING_OVERRIDE},
+	{39, ETHTOOL_LINK_EXT_STATE_AUTONEG,
+		ETHTOOL_LINK_EXT_SUBSTATE_AN_NO_HCD},
+
+	{5, ETHTOOL_LINK_EXT_STATE_LINK_TRAINING_FAILURE,
+		ETHTOOL_LINK_EXT_SUBSTATE_LT_KR_FRAME_LOCK_NOT_ACQUIRED},
+	{6, ETHTOOL_LINK_EXT_STATE_LINK_TRAINING_FAILURE,
+		ETHTOOL_LINK_EXT_SUBSTATE_LT_KR_LINK_INHIBIT_TIMEOUT},
+	{7, ETHTOOL_LINK_EXT_STATE_LINK_TRAINING_FAILURE,
+		ETHTOOL_LINK_EXT_SUBSTATE_LT_KR_LINK_PARTNER_DID_NOT_SET_RECEIVER_READY},
+	{8, ETHTOOL_LINK_EXT_STATE_LINK_TRAINING_FAILURE, 0},
+	{14, ETHTOOL_LINK_EXT_STATE_LINK_TRAINING_FAILURE,
+		ETHTOOL_LINK_EXT_SUBSTATE_LT_REMOTE_FAULT},
+
+	{9, ETHTOOL_LINK_EXT_STATE_LINK_LOGICAL_MISMATCH,
+		ETHTOOL_LINK_EXT_SUBSTATE_LLM_PCS_DID_NOT_ACQUIRE_BLOCK_LOCK},
+	{10, ETHTOOL_LINK_EXT_STATE_LINK_LOGICAL_MISMATCH,
+		ETHTOOL_LINK_EXT_SUBSTATE_LLM_PCS_DID_NOT_ACQUIRE_AM_LOCK},
+	{11, ETHTOOL_LINK_EXT_STATE_LINK_LOGICAL_MISMATCH,
+		ETHTOOL_LINK_EXT_SUBSTATE_LLM_PCS_DID_NOT_GET_ALIGN_STATUS},
+	{12, ETHTOOL_LINK_EXT_STATE_LINK_LOGICAL_MISMATCH,
+		ETHTOOL_LINK_EXT_SUBSTATE_LLM_FC_FEC_IS_NOT_LOCKED},
+	{13, ETHTOOL_LINK_EXT_STATE_LINK_LOGICAL_MISMATCH,
+		ETHTOOL_LINK_EXT_SUBSTATE_LLM_RS_FEC_IS_NOT_LOCKED},
+
+	{15, ETHTOOL_LINK_EXT_STATE_BAD_SIGNAL_INTEGRITY, 0},
+	{17, ETHTOOL_LINK_EXT_STATE_BAD_SIGNAL_INTEGRITY,
+		ETHTOOL_LINK_EXT_SUBSTATE_BSI_LARGE_NUMBER_OF_PHYSICAL_ERRORS},
+	{42, ETHTOOL_LINK_EXT_STATE_BAD_SIGNAL_INTEGRITY,
+		ETHTOOL_LINK_EXT_SUBSTATE_BSI_UNSUPPORTED_RATE},
+
+	{1024, ETHTOOL_LINK_EXT_STATE_NO_CABLE, 0},
+
+	{16, ETHTOOL_LINK_EXT_STATE_CABLE_ISSUE,
+		ETHTOOL_LINK_EXT_SUBSTATE_CI_UNSUPPORTED_CABLE},
+	{20, ETHTOOL_LINK_EXT_STATE_CABLE_ISSUE,
+		ETHTOOL_LINK_EXT_SUBSTATE_CI_UNSUPPORTED_CABLE},
+	{29, ETHTOOL_LINK_EXT_STATE_CABLE_ISSUE,
+		ETHTOOL_LINK_EXT_SUBSTATE_CI_UNSUPPORTED_CABLE},
+	{1025, ETHTOOL_LINK_EXT_STATE_CABLE_ISSUE,
+		ETHTOOL_LINK_EXT_SUBSTATE_CI_UNSUPPORTED_CABLE},
+	{1029, ETHTOOL_LINK_EXT_STATE_CABLE_ISSUE,
+		ETHTOOL_LINK_EXT_SUBSTATE_CI_UNSUPPORTED_CABLE},
+	{1031, ETHTOOL_LINK_EXT_STATE_CABLE_ISSUE, 0},
+
+	{1027, ETHTOOL_LINK_EXT_STATE_EEPROM_ISSUE, 0},
+
+	{23, ETHTOOL_LINK_EXT_STATE_CALIBRATION_FAILURE, 0},
+
+	{1032, ETHTOOL_LINK_EXT_STATE_POWER_BUDGET_EXCEEDED, 0},
+
+	{1030, ETHTOOL_LINK_EXT_STATE_OVERHEAT, 0},
+
+	{1042, ETHTOOL_LINK_EXT_STATE_MODULE,
+	 ETHTOOL_LINK_EXT_SUBSTATE_MODULE_CMIS_NOT_READY},
+};
+
+static void
+mlxsw_sp_port_set_link_ext_state(struct mlxsw_sp_ethtool_link_ext_state_opcode_mapping
+				 link_ext_state_mapping,
+				 struct ethtool_link_ext_state_info *link_ext_state_info)
+{
+	switch (link_ext_state_mapping.link_ext_state) {
+	case ETHTOOL_LINK_EXT_STATE_AUTONEG:
+		link_ext_state_info->autoneg =
+			link_ext_state_mapping.link_ext_substate;
+		break;
+	case ETHTOOL_LINK_EXT_STATE_LINK_TRAINING_FAILURE:
+		link_ext_state_info->link_training =
+			link_ext_state_mapping.link_ext_substate;
+		break;
+	case ETHTOOL_LINK_EXT_STATE_LINK_LOGICAL_MISMATCH:
+		link_ext_state_info->link_logical_mismatch =
+			link_ext_state_mapping.link_ext_substate;
+		break;
+	case ETHTOOL_LINK_EXT_STATE_BAD_SIGNAL_INTEGRITY:
+		link_ext_state_info->bad_signal_integrity =
+			link_ext_state_mapping.link_ext_substate;
+		break;
+	case ETHTOOL_LINK_EXT_STATE_CABLE_ISSUE:
+		link_ext_state_info->cable_issue =
+			link_ext_state_mapping.link_ext_substate;
+		break;
+	case ETHTOOL_LINK_EXT_STATE_MODULE:
+		link_ext_state_info->module =
+			link_ext_state_mapping.link_ext_substate;
+		break;
+	default:
+		break;
+	}
+
+	link_ext_state_info->link_ext_state = link_ext_state_mapping.link_ext_state;
+}
+
+static int
+mlxsw_sp_port_get_link_ext_state(struct net_device *dev,
+				 struct ethtool_link_ext_state_info *link_ext_state_info)
+{
+	struct mlxsw_sp_ethtool_link_ext_state_opcode_mapping link_ext_state_mapping;
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+	char pddr_pl[MLXSW_REG_PDDR_LEN];
+	int opcode, err, i;
+	u32 status_opcode;
+
+	if (netif_carrier_ok(dev))
+		return -ENODATA;
+
+	mlxsw_reg_pddr_pack(pddr_pl, mlxsw_sp_port->local_port,
+			    MLXSW_REG_PDDR_PAGE_SELECT_TROUBLESHOOTING_INFO);
+
+	opcode = MLXSW_REG_PDDR_TRBLSH_GROUP_OPCODE_MONITOR;
+	mlxsw_reg_pddr_trblsh_group_opcode_set(pddr_pl, opcode);
+
+	err = mlxsw_reg_query(mlxsw_sp_port->mlxsw_sp->core, MLXSW_REG(pddr),
+			      pddr_pl);
+	if (err)
+		return err;
+
+	status_opcode = mlxsw_reg_pddr_trblsh_status_opcode_get(pddr_pl);
+	if (!status_opcode)
+		return -ENODATA;
+
+	for (i = 0; i < ARRAY_SIZE(mlxsw_sp_link_ext_state_opcode_map); i++) {
+		link_ext_state_mapping = mlxsw_sp_link_ext_state_opcode_map[i];
+		if (link_ext_state_mapping.status_opcode == status_opcode) {
+			mlxsw_sp_port_set_link_ext_state(link_ext_state_mapping,
+							 link_ext_state_info);
+			return 0;
+		}
+	}
+
+	return -ENODATA;
+}
+
+static void mlxsw_sp_port_get_pauseparam(struct net_device *dev,
+					 struct ethtool_pauseparam *pause)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+
+	pause->rx_pause = mlxsw_sp_port->link.rx_pause;
+	pause->tx_pause = mlxsw_sp_port->link.tx_pause;
+}
+
+static int mlxsw_sp_port_pause_set(struct mlxsw_sp_port *mlxsw_sp_port,
+				   struct ethtool_pauseparam *pause)
+{
+	char pfcc_pl[MLXSW_REG_PFCC_LEN];
+
+	mlxsw_reg_pfcc_pack(pfcc_pl, mlxsw_sp_port->local_port);
+	mlxsw_reg_pfcc_pprx_set(pfcc_pl, pause->rx_pause);
+	mlxsw_reg_pfcc_pptx_set(pfcc_pl, pause->tx_pause);
+
+	return mlxsw_reg_write(mlxsw_sp_port->mlxsw_sp->core, MLXSW_REG(pfcc),
+			       pfcc_pl);
+}
+
+/* Maximum delay buffer needed in case of PAUSE frames. Similar to PFC delay, but is
+ * measured in bytes. Assumes 100m cable and does not take into account MTU.
+ */
+#define MLXSW_SP_PAUSE_DELAY_BYTES 19476
+
+static int mlxsw_sp_port_set_pauseparam(struct net_device *dev,
+					struct ethtool_pauseparam *pause)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+	bool pause_en = pause->tx_pause || pause->rx_pause;
+	struct mlxsw_sp_hdroom orig_hdroom;
+	struct mlxsw_sp_hdroom hdroom;
+	int prio;
+	int err;
+
+	if (mlxsw_sp_port->dcb.pfc && mlxsw_sp_port->dcb.pfc->pfc_en) {
+		netdev_err(dev, "PFC already enabled on port\n");
+		return -EINVAL;
+	}
+
+	if (pause->autoneg) {
+		netdev_err(dev, "PAUSE frames autonegotiation isn't supported\n");
+		return -EINVAL;
+	}
+
+	orig_hdroom = *mlxsw_sp_port->hdroom;
+
+	hdroom = orig_hdroom;
+	if (pause_en)
+		hdroom.delay_bytes = MLXSW_SP_PAUSE_DELAY_BYTES;
+	else
+		hdroom.delay_bytes = 0;
+
+	for (prio = 0; prio < IEEE_8021QAZ_MAX_TCS; prio++)
+		hdroom.prios.prio[prio].lossy = !pause_en;
+
+	mlxsw_sp_hdroom_bufs_reset_lossiness(&hdroom);
+	mlxsw_sp_hdroom_bufs_reset_sizes(mlxsw_sp_port, &hdroom);
+
+	err = mlxsw_sp_hdroom_configure(mlxsw_sp_port, &hdroom);
+	if (err) {
+		netdev_err(dev, "Failed to configure port's headroom\n");
+		return err;
+	}
+
+	err = mlxsw_sp_port_pause_set(mlxsw_sp_port, pause);
+	if (err) {
+		netdev_err(dev, "Failed to set PAUSE parameters\n");
+		goto err_port_pause_configure;
+	}
+
+	mlxsw_sp_port->link.rx_pause = pause->rx_pause;
+	mlxsw_sp_port->link.tx_pause = pause->tx_pause;
+
+	return 0;
+
+err_port_pause_configure:
+	mlxsw_sp_hdroom_configure(mlxsw_sp_port, &orig_hdroom);
+	return err;
+}
+
+struct mlxsw_sp_port_hw_stats {
+	char str[ETH_GSTRING_LEN];
+	u64 (*getter)(const char *payload);
+	bool cells_bytes;
+};
+
+static struct mlxsw_sp_port_hw_stats mlxsw_sp_port_hw_stats[] = {
+	{
+		.str = "a_frames_transmitted_ok",
+		.getter = mlxsw_reg_ppcnt_a_frames_transmitted_ok_get,
+	},
+	{
+		.str = "a_frames_received_ok",
+		.getter = mlxsw_reg_ppcnt_a_frames_received_ok_get,
+	},
+	{
+		.str = "a_frame_check_sequence_errors",
+		.getter = mlxsw_reg_ppcnt_a_frame_check_sequence_errors_get,
+	},
+	{
+		.str = "a_alignment_errors",
+		.getter = mlxsw_reg_ppcnt_a_alignment_errors_get,
+	},
+	{
+		.str = "a_octets_transmitted_ok",
+		.getter = mlxsw_reg_ppcnt_a_octets_transmitted_ok_get,
+	},
+	{
+		.str = "a_octets_received_ok",
+		.getter = mlxsw_reg_ppcnt_a_octets_received_ok_get,
+	},
+	{
+		.str = "a_multicast_frames_xmitted_ok",
+		.getter = mlxsw_reg_ppcnt_a_multicast_frames_xmitted_ok_get,
+	},
+	{
+		.str = "a_broadcast_frames_xmitted_ok",
+		.getter = mlxsw_reg_ppcnt_a_broadcast_frames_xmitted_ok_get,
+	},
+	{
+		.str = "a_multicast_frames_received_ok",
+		.getter = mlxsw_reg_ppcnt_a_multicast_frames_received_ok_get,
+	},
+	{
+		.str = "a_broadcast_frames_received_ok",
+		.getter = mlxsw_reg_ppcnt_a_broadcast_frames_received_ok_get,
+	},
+	{
+		.str = "a_in_range_length_errors",
+		.getter = mlxsw_reg_ppcnt_a_in_range_length_errors_get,
+	},
+	{
+		.str = "a_out_of_range_length_field",
+		.getter = mlxsw_reg_ppcnt_a_out_of_range_length_field_get,
+	},
+	{
+		.str = "a_frame_too_long_errors",
+		.getter = mlxsw_reg_ppcnt_a_frame_too_long_errors_get,
+	},
+	{
+		.str = "a_symbol_error_during_carrier",
+		.getter = mlxsw_reg_ppcnt_a_symbol_error_during_carrier_get,
+	},
+	{
+		.str = "a_mac_control_frames_transmitted",
+		.getter = mlxsw_reg_ppcnt_a_mac_control_frames_transmitted_get,
+	},
+	{
+		.str = "a_mac_control_frames_received",
+		.getter = mlxsw_reg_ppcnt_a_mac_control_frames_received_get,
+	},
+	{
+		.str = "a_unsupported_opcodes_received",
+		.getter = mlxsw_reg_ppcnt_a_unsupported_opcodes_received_get,
+	},
+	{
+		.str = "a_pause_mac_ctrl_frames_received",
+		.getter = mlxsw_reg_ppcnt_a_pause_mac_ctrl_frames_received_get,
+	},
+	{
+		.str = "a_pause_mac_ctrl_frames_xmitted",
+		.getter = mlxsw_reg_ppcnt_a_pause_mac_ctrl_frames_transmitted_get,
+	},
+};
+
+#define MLXSW_SP_PORT_HW_STATS_LEN ARRAY_SIZE(mlxsw_sp_port_hw_stats)
+
+static struct mlxsw_sp_port_hw_stats mlxsw_sp_port_hw_rfc_2863_stats[] = {
+	{
+		.str = "if_in_discards",
+		.getter = mlxsw_reg_ppcnt_if_in_discards_get,
+	},
+	{
+		.str = "if_out_discards",
+		.getter = mlxsw_reg_ppcnt_if_out_discards_get,
+	},
+	{
+		.str = "if_out_errors",
+		.getter = mlxsw_reg_ppcnt_if_out_errors_get,
+	},
+};
+
+#define MLXSW_SP_PORT_HW_RFC_2863_STATS_LEN \
+	ARRAY_SIZE(mlxsw_sp_port_hw_rfc_2863_stats)
+
+static struct mlxsw_sp_port_hw_stats mlxsw_sp_port_hw_rfc_2819_stats[] = {
+	{
+		.str = "ether_stats_undersize_pkts",
+		.getter = mlxsw_reg_ppcnt_ether_stats_undersize_pkts_get,
+	},
+	{
+		.str = "ether_stats_oversize_pkts",
+		.getter = mlxsw_reg_ppcnt_ether_stats_oversize_pkts_get,
+	},
+	{
+		.str = "ether_stats_fragments",
+		.getter = mlxsw_reg_ppcnt_ether_stats_fragments_get,
+	},
+	{
+		.str = "ether_pkts64octets",
+		.getter = mlxsw_reg_ppcnt_ether_stats_pkts64octets_get,
+	},
+	{
+		.str = "ether_pkts65to127octets",
+		.getter = mlxsw_reg_ppcnt_ether_stats_pkts65to127octets_get,
+	},
+	{
+		.str = "ether_pkts128to255octets",
+		.getter = mlxsw_reg_ppcnt_ether_stats_pkts128to255octets_get,
+	},
+	{
+		.str = "ether_pkts256to511octets",
+		.getter = mlxsw_reg_ppcnt_ether_stats_pkts256to511octets_get,
+	},
+	{
+		.str = "ether_pkts512to1023octets",
+		.getter = mlxsw_reg_ppcnt_ether_stats_pkts512to1023octets_get,
+	},
+	{
+		.str = "ether_pkts1024to1518octets",
+		.getter = mlxsw_reg_ppcnt_ether_stats_pkts1024to1518octets_get,
+	},
+	{
+		.str = "ether_pkts1519to2047octets",
+		.getter = mlxsw_reg_ppcnt_ether_stats_pkts1519to2047octets_get,
+	},
+	{
+		.str = "ether_pkts2048to4095octets",
+		.getter = mlxsw_reg_ppcnt_ether_stats_pkts2048to4095octets_get,
+	},
+	{
+		.str = "ether_pkts4096to8191octets",
+		.getter = mlxsw_reg_ppcnt_ether_stats_pkts4096to8191octets_get,
+	},
+	{
+		.str = "ether_pkts8192to10239octets",
+		.getter = mlxsw_reg_ppcnt_ether_stats_pkts8192to10239octets_get,
+	},
+};
+
+#define MLXSW_SP_PORT_HW_RFC_2819_STATS_LEN \
+	ARRAY_SIZE(mlxsw_sp_port_hw_rfc_2819_stats)
+
+static struct mlxsw_sp_port_hw_stats mlxsw_sp_port_hw_rfc_3635_stats[] = {
+	{
+		.str = "dot3stats_fcs_errors",
+		.getter = mlxsw_reg_ppcnt_dot3stats_fcs_errors_get,
+	},
+	{
+		.str = "dot3stats_symbol_errors",
+		.getter = mlxsw_reg_ppcnt_dot3stats_symbol_errors_get,
+	},
+	{
+		.str = "dot3control_in_unknown_opcodes",
+		.getter = mlxsw_reg_ppcnt_dot3control_in_unknown_opcodes_get,
+	},
+	{
+		.str = "dot3in_pause_frames",
+		.getter = mlxsw_reg_ppcnt_dot3in_pause_frames_get,
+	},
+};
+
+#define MLXSW_SP_PORT_HW_RFC_3635_STATS_LEN \
+	ARRAY_SIZE(mlxsw_sp_port_hw_rfc_3635_stats)
+
+static struct mlxsw_sp_port_hw_stats mlxsw_sp_port_hw_ext_stats[] = {
+	{
+		.str = "ecn_marked",
+		.getter = mlxsw_reg_ppcnt_ecn_marked_get,
+	},
+};
+
+#define MLXSW_SP_PORT_HW_EXT_STATS_LEN ARRAY_SIZE(mlxsw_sp_port_hw_ext_stats)
+
+static struct mlxsw_sp_port_hw_stats mlxsw_sp_port_hw_discard_stats[] = {
+	{
+		.str = "discard_ingress_general",
+		.getter = mlxsw_reg_ppcnt_ingress_general_get,
+	},
+	{
+		.str = "discard_ingress_policy_engine",
+		.getter = mlxsw_reg_ppcnt_ingress_policy_engine_get,
+	},
+	{
+		.str = "discard_ingress_vlan_membership",
+		.getter = mlxsw_reg_ppcnt_ingress_vlan_membership_get,
+	},
+	{
+		.str = "discard_ingress_tag_frame_type",
+		.getter = mlxsw_reg_ppcnt_ingress_tag_frame_type_get,
+	},
+	{
+		.str = "discard_egress_vlan_membership",
+		.getter = mlxsw_reg_ppcnt_egress_vlan_membership_get,
+	},
+	{
+		.str = "discard_loopback_filter",
+		.getter = mlxsw_reg_ppcnt_loopback_filter_get,
+	},
+	{
+		.str = "discard_egress_general",
+		.getter = mlxsw_reg_ppcnt_egress_general_get,
+	},
+	{
+		.str = "discard_egress_hoq",
+		.getter = mlxsw_reg_ppcnt_egress_hoq_get,
+	},
+	{
+		.str = "discard_egress_policy_engine",
+		.getter = mlxsw_reg_ppcnt_egress_policy_engine_get,
+	},
+	{
+		.str = "discard_ingress_tx_link_down",
+		.getter = mlxsw_reg_ppcnt_ingress_tx_link_down_get,
+	},
+	{
+		.str = "discard_egress_stp_filter",
+		.getter = mlxsw_reg_ppcnt_egress_stp_filter_get,
+	},
+	{
+		.str = "discard_egress_sll",
+		.getter = mlxsw_reg_ppcnt_egress_sll_get,
+	},
+};
+
+#define MLXSW_SP_PORT_HW_DISCARD_STATS_LEN \
+	ARRAY_SIZE(mlxsw_sp_port_hw_discard_stats)
+
+static struct mlxsw_sp_port_hw_stats mlxsw_sp_port_hw_prio_stats[] = {
+	{
+		.str = "rx_octets_prio",
+		.getter = mlxsw_reg_ppcnt_rx_octets_get,
+	},
+	{
+		.str = "rx_frames_prio",
+		.getter = mlxsw_reg_ppcnt_rx_frames_get,
+	},
+	{
+		.str = "tx_octets_prio",
+		.getter = mlxsw_reg_ppcnt_tx_octets_get,
+	},
+	{
+		.str = "tx_frames_prio",
+		.getter = mlxsw_reg_ppcnt_tx_frames_get,
+	},
+	{
+		.str = "rx_pause_prio",
+		.getter = mlxsw_reg_ppcnt_rx_pause_get,
+	},
+	{
+		.str = "rx_pause_duration_prio",
+		.getter = mlxsw_reg_ppcnt_rx_pause_duration_get,
+	},
+	{
+		.str = "tx_pause_prio",
+		.getter = mlxsw_reg_ppcnt_tx_pause_get,
+	},
+	{
+		.str = "tx_pause_duration_prio",
+		.getter = mlxsw_reg_ppcnt_tx_pause_duration_get,
+	},
+};
+
+#define MLXSW_SP_PORT_HW_PRIO_STATS_LEN ARRAY_SIZE(mlxsw_sp_port_hw_prio_stats)
+
+static struct mlxsw_sp_port_hw_stats mlxsw_sp_port_hw_tc_stats[] = {
+	{
+		.str = "tc_transmit_queue_tc",
+		.getter = mlxsw_reg_ppcnt_tc_transmit_queue_get,
+		.cells_bytes = true,
+	},
+	{
+		.str = "tc_no_buffer_discard_uc_tc",
+		.getter = mlxsw_reg_ppcnt_tc_no_buffer_discard_uc_get,
+	},
+};
+
+#define MLXSW_SP_PORT_HW_TC_STATS_LEN ARRAY_SIZE(mlxsw_sp_port_hw_tc_stats)
+
+struct mlxsw_sp_port_stats {
+	char str[ETH_GSTRING_LEN];
+	u64 (*getter)(struct mlxsw_sp_port *mlxsw_sp_port);
+};
+
+static u64
+mlxsw_sp_port_get_transceiver_overheat_stats(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	struct mlxsw_core *mlxsw_core = mlxsw_sp_port->mlxsw_sp->core;
+	u8 slot_index = mlxsw_sp_port->mapping.slot_index;
+	u8 module = mlxsw_sp_port->mapping.module;
+	u64 stats;
+	int err;
+
+	err = mlxsw_env_module_overheat_counter_get(mlxsw_core, slot_index,
+						    module, &stats);
+	if (err)
+		return mlxsw_sp_port->module_overheat_initial_val;
+
+	return stats - mlxsw_sp_port->module_overheat_initial_val;
+}
+
+static struct mlxsw_sp_port_stats mlxsw_sp_port_transceiver_stats[] = {
+	{
+		.str = "transceiver_overheat",
+		.getter = mlxsw_sp_port_get_transceiver_overheat_stats,
+	},
+};
+
+#define MLXSW_SP_PORT_HW_TRANSCEIVER_STATS_LEN ARRAY_SIZE(mlxsw_sp_port_transceiver_stats)
+
+#define MLXSW_SP_PORT_ETHTOOL_STATS_LEN (MLXSW_SP_PORT_HW_STATS_LEN + \
+					 MLXSW_SP_PORT_HW_RFC_2863_STATS_LEN + \
+					 MLXSW_SP_PORT_HW_RFC_2819_STATS_LEN + \
+					 MLXSW_SP_PORT_HW_RFC_3635_STATS_LEN + \
+					 MLXSW_SP_PORT_HW_EXT_STATS_LEN + \
+					 MLXSW_SP_PORT_HW_DISCARD_STATS_LEN + \
+					 (MLXSW_SP_PORT_HW_PRIO_STATS_LEN * \
+					  IEEE_8021QAZ_MAX_TCS) + \
+					 (MLXSW_SP_PORT_HW_TC_STATS_LEN * \
+					  TC_MAX_QUEUE) + \
+					  MLXSW_SP_PORT_HW_TRANSCEIVER_STATS_LEN)
+
+static void mlxsw_sp_port_get_prio_strings(u8 **p, int prio)
+{
+	int i;
+
+	for (i = 0; i < MLXSW_SP_PORT_HW_PRIO_STATS_LEN; i++) {
+		snprintf(*p, ETH_GSTRING_LEN, "%.29s_%.1d",
+			 mlxsw_sp_port_hw_prio_stats[i].str, prio);
+		*p += ETH_GSTRING_LEN;
+	}
+}
+
+static void mlxsw_sp_port_get_tc_strings(u8 **p, int tc)
+{
+	int i;
+
+	for (i = 0; i < MLXSW_SP_PORT_HW_TC_STATS_LEN; i++) {
+		snprintf(*p, ETH_GSTRING_LEN, "%.29s_%.1d",
+			 mlxsw_sp_port_hw_tc_stats[i].str, tc);
+		*p += ETH_GSTRING_LEN;
+	}
+}
+
+static void mlxsw_sp_port_get_strings(struct net_device *dev,
+				      u32 stringset, u8 *data)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+	u8 *p = data;
+	int i;
+
+	switch (stringset) {
+	case ETH_SS_STATS:
+		for (i = 0; i < MLXSW_SP_PORT_HW_STATS_LEN; i++) {
+			memcpy(p, mlxsw_sp_port_hw_stats[i].str,
+			       ETH_GSTRING_LEN);
+			p += ETH_GSTRING_LEN;
+		}
+
+		for (i = 0; i < MLXSW_SP_PORT_HW_RFC_2863_STATS_LEN; i++) {
+			memcpy(p, mlxsw_sp_port_hw_rfc_2863_stats[i].str,
+			       ETH_GSTRING_LEN);
+			p += ETH_GSTRING_LEN;
+		}
+
+		for (i = 0; i < MLXSW_SP_PORT_HW_RFC_2819_STATS_LEN; i++) {
+			memcpy(p, mlxsw_sp_port_hw_rfc_2819_stats[i].str,
+			       ETH_GSTRING_LEN);
+			p += ETH_GSTRING_LEN;
+		}
+
+		for (i = 0; i < MLXSW_SP_PORT_HW_RFC_3635_STATS_LEN; i++) {
+			memcpy(p, mlxsw_sp_port_hw_rfc_3635_stats[i].str,
+			       ETH_GSTRING_LEN);
+			p += ETH_GSTRING_LEN;
+		}
+
+		for (i = 0; i < MLXSW_SP_PORT_HW_EXT_STATS_LEN; i++) {
+			memcpy(p, mlxsw_sp_port_hw_ext_stats[i].str,
+			       ETH_GSTRING_LEN);
+			p += ETH_GSTRING_LEN;
+		}
+
+		for (i = 0; i < MLXSW_SP_PORT_HW_DISCARD_STATS_LEN; i++) {
+			memcpy(p, mlxsw_sp_port_hw_discard_stats[i].str,
+			       ETH_GSTRING_LEN);
+			p += ETH_GSTRING_LEN;
+		}
+
+		for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++)
+			mlxsw_sp_port_get_prio_strings(&p, i);
+
+		for (i = 0; i < TC_MAX_QUEUE; i++)
+			mlxsw_sp_port_get_tc_strings(&p, i);
+
+		mlxsw_sp_port->mlxsw_sp->ptp_ops->get_stats_strings(&p);
+
+		for (i = 0; i < MLXSW_SP_PORT_HW_TRANSCEIVER_STATS_LEN; i++) {
+			memcpy(p, mlxsw_sp_port_transceiver_stats[i].str,
+			       ETH_GSTRING_LEN);
+			p += ETH_GSTRING_LEN;
+		}
+		break;
+	}
+}
+
+static int mlxsw_sp_port_set_phys_id(struct net_device *dev,
+				     enum ethtool_phys_id_state state)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	char mlcr_pl[MLXSW_REG_MLCR_LEN];
+	bool active;
+
+	switch (state) {
+	case ETHTOOL_ID_ACTIVE:
+		active = true;
+		break;
+	case ETHTOOL_ID_INACTIVE:
+		active = false;
+		break;
+	default:
+		return -EOPNOTSUPP;
+	}
+
+	mlxsw_reg_mlcr_pack(mlcr_pl, mlxsw_sp_port->local_port, active);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(mlcr), mlcr_pl);
+}
+
+static int
+mlxsw_sp_get_hw_stats_by_group(struct mlxsw_sp_port_hw_stats **p_hw_stats,
+			       int *p_len, enum mlxsw_reg_ppcnt_grp grp)
+{
+	switch (grp) {
+	case MLXSW_REG_PPCNT_IEEE_8023_CNT:
+		*p_hw_stats = mlxsw_sp_port_hw_stats;
+		*p_len = MLXSW_SP_PORT_HW_STATS_LEN;
+		break;
+	case MLXSW_REG_PPCNT_RFC_2863_CNT:
+		*p_hw_stats = mlxsw_sp_port_hw_rfc_2863_stats;
+		*p_len = MLXSW_SP_PORT_HW_RFC_2863_STATS_LEN;
+		break;
+	case MLXSW_REG_PPCNT_RFC_2819_CNT:
+		*p_hw_stats = mlxsw_sp_port_hw_rfc_2819_stats;
+		*p_len = MLXSW_SP_PORT_HW_RFC_2819_STATS_LEN;
+		break;
+	case MLXSW_REG_PPCNT_RFC_3635_CNT:
+		*p_hw_stats = mlxsw_sp_port_hw_rfc_3635_stats;
+		*p_len = MLXSW_SP_PORT_HW_RFC_3635_STATS_LEN;
+		break;
+	case MLXSW_REG_PPCNT_EXT_CNT:
+		*p_hw_stats = mlxsw_sp_port_hw_ext_stats;
+		*p_len = MLXSW_SP_PORT_HW_EXT_STATS_LEN;
+		break;
+	case MLXSW_REG_PPCNT_DISCARD_CNT:
+		*p_hw_stats = mlxsw_sp_port_hw_discard_stats;
+		*p_len = MLXSW_SP_PORT_HW_DISCARD_STATS_LEN;
+		break;
+	case MLXSW_REG_PPCNT_PRIO_CNT:
+		*p_hw_stats = mlxsw_sp_port_hw_prio_stats;
+		*p_len = MLXSW_SP_PORT_HW_PRIO_STATS_LEN;
+		break;
+	case MLXSW_REG_PPCNT_TC_CNT:
+		*p_hw_stats = mlxsw_sp_port_hw_tc_stats;
+		*p_len = MLXSW_SP_PORT_HW_TC_STATS_LEN;
+		break;
+	default:
+		WARN_ON(1);
+		return -EOPNOTSUPP;
+	}
+	return 0;
+}
+
+static void __mlxsw_sp_port_get_stats(struct net_device *dev,
+				      enum mlxsw_reg_ppcnt_grp grp, int prio,
+				      u64 *data, int data_index)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	struct mlxsw_sp_port_hw_stats *hw_stats;
+	char ppcnt_pl[MLXSW_REG_PPCNT_LEN];
+	int i, len;
+	int err;
+
+	err = mlxsw_sp_get_hw_stats_by_group(&hw_stats, &len, grp);
+	if (err)
+		return;
+	mlxsw_sp_port_get_stats_raw(dev, grp, prio, ppcnt_pl);
+	for (i = 0; i < len; i++) {
+		data[data_index + i] = hw_stats[i].getter(ppcnt_pl);
+		if (!hw_stats[i].cells_bytes)
+			continue;
+		data[data_index + i] = mlxsw_sp_cells_bytes(mlxsw_sp,
+							    data[data_index + i]);
+	}
+}
+
+static void __mlxsw_sp_port_get_env_stats(struct net_device *dev, u64 *data, int data_index,
+					  struct mlxsw_sp_port_stats *port_stats,
+					  int len)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+	int i;
+
+	for (i = 0; i < len; i++)
+		data[data_index + i] = port_stats[i].getter(mlxsw_sp_port);
+}
+
+static void mlxsw_sp_port_get_stats(struct net_device *dev,
+				    struct ethtool_stats *stats, u64 *data)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+	int i, data_index = 0;
+
+	/* IEEE 802.3 Counters */
+	__mlxsw_sp_port_get_stats(dev, MLXSW_REG_PPCNT_IEEE_8023_CNT, 0,
+				  data, data_index);
+	data_index = MLXSW_SP_PORT_HW_STATS_LEN;
+
+	/* RFC 2863 Counters */
+	__mlxsw_sp_port_get_stats(dev, MLXSW_REG_PPCNT_RFC_2863_CNT, 0,
+				  data, data_index);
+	data_index += MLXSW_SP_PORT_HW_RFC_2863_STATS_LEN;
+
+	/* RFC 2819 Counters */
+	__mlxsw_sp_port_get_stats(dev, MLXSW_REG_PPCNT_RFC_2819_CNT, 0,
+				  data, data_index);
+	data_index += MLXSW_SP_PORT_HW_RFC_2819_STATS_LEN;
+
+	/* RFC 3635 Counters */
+	__mlxsw_sp_port_get_stats(dev, MLXSW_REG_PPCNT_RFC_3635_CNT, 0,
+				  data, data_index);
+	data_index += MLXSW_SP_PORT_HW_RFC_3635_STATS_LEN;
+
+	/* Extended Counters */
+	__mlxsw_sp_port_get_stats(dev, MLXSW_REG_PPCNT_EXT_CNT, 0,
+				  data, data_index);
+	data_index += MLXSW_SP_PORT_HW_EXT_STATS_LEN;
+
+	/* Discard Counters */
+	__mlxsw_sp_port_get_stats(dev, MLXSW_REG_PPCNT_DISCARD_CNT, 0,
+				  data, data_index);
+	data_index += MLXSW_SP_PORT_HW_DISCARD_STATS_LEN;
+
+	/* Per-Priority Counters */
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) {
+		__mlxsw_sp_port_get_stats(dev, MLXSW_REG_PPCNT_PRIO_CNT, i,
+					  data, data_index);
+		data_index += MLXSW_SP_PORT_HW_PRIO_STATS_LEN;
+	}
+
+	/* Per-TC Counters */
+	for (i = 0; i < TC_MAX_QUEUE; i++) {
+		__mlxsw_sp_port_get_stats(dev, MLXSW_REG_PPCNT_TC_CNT, i,
+					  data, data_index);
+		data_index += MLXSW_SP_PORT_HW_TC_STATS_LEN;
+	}
+
+	/* PTP counters */
+	mlxsw_sp_port->mlxsw_sp->ptp_ops->get_stats(mlxsw_sp_port,
+						    data, data_index);
+	data_index += mlxsw_sp_port->mlxsw_sp->ptp_ops->get_stats_count();
+
+	/* Transceiver counters */
+	__mlxsw_sp_port_get_env_stats(dev, data, data_index, mlxsw_sp_port_transceiver_stats,
+				      MLXSW_SP_PORT_HW_TRANSCEIVER_STATS_LEN);
+	data_index += MLXSW_SP_PORT_HW_TRANSCEIVER_STATS_LEN;
+}
+
+static int mlxsw_sp_port_get_sset_count(struct net_device *dev, int sset)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+
+	switch (sset) {
+	case ETH_SS_STATS:
+		return MLXSW_SP_PORT_ETHTOOL_STATS_LEN +
+			mlxsw_sp_port->mlxsw_sp->ptp_ops->get_stats_count();
+	default:
+		return -EOPNOTSUPP;
+	}
+}
+
+static void
+mlxsw_sp_port_get_link_supported(struct mlxsw_sp *mlxsw_sp, u32 eth_proto_cap,
+				 struct ethtool_link_ksettings *cmd)
+{
+	const struct mlxsw_sp_port_type_speed_ops *ops;
+
+	ops = mlxsw_sp->port_type_speed_ops;
+
+	ethtool_link_ksettings_add_link_mode(cmd, supported, Asym_Pause);
+	ethtool_link_ksettings_add_link_mode(cmd, supported, Autoneg);
+	ethtool_link_ksettings_add_link_mode(cmd, supported, Pause);
+
+	ops->from_ptys_supported_port(mlxsw_sp, eth_proto_cap, cmd);
+	ops->from_ptys_link(mlxsw_sp, eth_proto_cap,
+			    cmd->link_modes.supported);
+}
+
+static void
+mlxsw_sp_port_get_link_advertise(struct mlxsw_sp *mlxsw_sp,
+				 u32 eth_proto_admin, bool autoneg,
+				 struct ethtool_link_ksettings *cmd)
+{
+	const struct mlxsw_sp_port_type_speed_ops *ops;
+
+	ops = mlxsw_sp->port_type_speed_ops;
+
+	if (!autoneg)
+		return;
+
+	ethtool_link_ksettings_add_link_mode(cmd, advertising, Autoneg);
+	ops->from_ptys_link(mlxsw_sp, eth_proto_admin,
+			    cmd->link_modes.advertising);
+}
+
+static u8
+mlxsw_sp_port_connector_port(enum mlxsw_reg_ptys_connector_type connector_type)
+{
+	switch (connector_type) {
+	case MLXSW_REG_PTYS_CONNECTOR_TYPE_UNKNOWN_OR_NO_CONNECTOR:
+		return PORT_OTHER;
+	case MLXSW_REG_PTYS_CONNECTOR_TYPE_PORT_NONE:
+		return PORT_NONE;
+	case MLXSW_REG_PTYS_CONNECTOR_TYPE_PORT_TP:
+		return PORT_TP;
+	case MLXSW_REG_PTYS_CONNECTOR_TYPE_PORT_AUI:
+		return PORT_AUI;
+	case MLXSW_REG_PTYS_CONNECTOR_TYPE_PORT_BNC:
+		return PORT_BNC;
+	case MLXSW_REG_PTYS_CONNECTOR_TYPE_PORT_MII:
+		return PORT_MII;
+	case MLXSW_REG_PTYS_CONNECTOR_TYPE_PORT_FIBRE:
+		return PORT_FIBRE;
+	case MLXSW_REG_PTYS_CONNECTOR_TYPE_PORT_DA:
+		return PORT_DA;
+	case MLXSW_REG_PTYS_CONNECTOR_TYPE_PORT_OTHER:
+		return PORT_OTHER;
+	default:
+		WARN_ON_ONCE(1);
+		return PORT_OTHER;
+	}
+}
+
+static int mlxsw_sp_port_ptys_query(struct mlxsw_sp_port *mlxsw_sp_port,
+				    u32 *p_eth_proto_cap, u32 *p_eth_proto_admin,
+				    u32 *p_eth_proto_oper, u8 *p_connector_type)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	const struct mlxsw_sp_port_type_speed_ops *ops;
+	char ptys_pl[MLXSW_REG_PTYS_LEN];
+	int err;
+
+	ops = mlxsw_sp->port_type_speed_ops;
+
+	ops->reg_ptys_eth_pack(mlxsw_sp, ptys_pl, mlxsw_sp_port->local_port, 0, false);
+	err = mlxsw_reg_query(mlxsw_sp->core, MLXSW_REG(ptys), ptys_pl);
+	if (err)
+		return err;
+
+	ops->reg_ptys_eth_unpack(mlxsw_sp, ptys_pl, p_eth_proto_cap, p_eth_proto_admin,
+				 p_eth_proto_oper);
+	if (p_connector_type)
+		*p_connector_type = mlxsw_reg_ptys_connector_type_get(ptys_pl);
+	return 0;
+}
+
+static int mlxsw_sp_port_get_link_ksettings(struct net_device *dev,
+					    struct ethtool_link_ksettings *cmd)
+{
+	u32 eth_proto_cap, eth_proto_admin, eth_proto_oper;
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	const struct mlxsw_sp_port_type_speed_ops *ops;
+	u8 connector_type;
+	bool autoneg;
+	int err;
+
+	err = mlxsw_sp_port_ptys_query(mlxsw_sp_port, &eth_proto_cap, &eth_proto_admin,
+				       &eth_proto_oper, &connector_type);
+	if (err)
+		return err;
+
+	ops = mlxsw_sp->port_type_speed_ops;
+	autoneg = mlxsw_sp_port->link.autoneg;
+
+	mlxsw_sp_port_get_link_supported(mlxsw_sp, eth_proto_cap, cmd);
+
+	mlxsw_sp_port_get_link_advertise(mlxsw_sp, eth_proto_admin, autoneg, cmd);
+
+	cmd->base.autoneg = autoneg ? AUTONEG_ENABLE : AUTONEG_DISABLE;
+	cmd->base.port = mlxsw_sp_port_connector_port(connector_type);
+	ops->from_ptys_link_mode(mlxsw_sp, netif_carrier_ok(dev),
+				 eth_proto_oper, cmd);
+
+	return 0;
+}
+
+static int
+mlxsw_sp_port_set_link_ksettings(struct net_device *dev,
+				 const struct ethtool_link_ksettings *cmd)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	const struct mlxsw_sp_port_type_speed_ops *ops;
+	char ptys_pl[MLXSW_REG_PTYS_LEN];
+	u32 eth_proto_cap, eth_proto_new;
+	bool autoneg;
+	int err;
+
+	ops = mlxsw_sp->port_type_speed_ops;
+
+	ops->reg_ptys_eth_pack(mlxsw_sp, ptys_pl, mlxsw_sp_port->local_port,
+			       0, false);
+	err = mlxsw_reg_query(mlxsw_sp->core, MLXSW_REG(ptys), ptys_pl);
+	if (err)
+		return err;
+	ops->reg_ptys_eth_unpack(mlxsw_sp, ptys_pl, &eth_proto_cap, NULL, NULL);
+
+	autoneg = cmd->base.autoneg == AUTONEG_ENABLE;
+	eth_proto_new = autoneg ?
+		ops->to_ptys_advert_link(mlxsw_sp, cmd) :
+		ops->to_ptys_speed_lanes(mlxsw_sp, mlxsw_sp_port->mapping.width,
+					 cmd);
+
+	eth_proto_new = eth_proto_new & eth_proto_cap;
+	if (!eth_proto_new) {
+		netdev_err(dev, "No supported speed or lanes requested\n");
+		return -EINVAL;
+	}
+
+	ops->reg_ptys_eth_pack(mlxsw_sp, ptys_pl, mlxsw_sp_port->local_port,
+			       eth_proto_new, autoneg);
+	err = mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(ptys), ptys_pl);
+	if (err)
+		return err;
+
+	mlxsw_sp_port->link.autoneg = autoneg;
+
+	if (!netif_running(dev))
+		return 0;
+
+	mlxsw_sp_port_admin_status_set(mlxsw_sp_port, false);
+	mlxsw_sp_port_admin_status_set(mlxsw_sp_port, true);
+
+	return 0;
+}
+
+static int mlxsw_sp_get_module_info(struct net_device *netdev,
+				    struct ethtool_modinfo *modinfo)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(netdev);
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+
+	return mlxsw_env_get_module_info(netdev, mlxsw_sp->core,
+					 mlxsw_sp_port->mapping.slot_index,
+					 mlxsw_sp_port->mapping.module,
+					 modinfo);
+}
+
+static int mlxsw_sp_get_module_eeprom(struct net_device *netdev,
+				      struct ethtool_eeprom *ee, u8 *data)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(netdev);
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	u8 slot_index = mlxsw_sp_port->mapping.slot_index;
+	u8 module = mlxsw_sp_port->mapping.module;
+
+	return mlxsw_env_get_module_eeprom(netdev, mlxsw_sp->core, slot_index,
+					   module, ee, data);
+}
+
+static int
+mlxsw_sp_get_module_eeprom_by_page(struct net_device *dev,
+				   const struct ethtool_module_eeprom *page,
+				   struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	u8 slot_index = mlxsw_sp_port->mapping.slot_index;
+	u8 module = mlxsw_sp_port->mapping.module;
+
+	return mlxsw_env_get_module_eeprom_by_page(mlxsw_sp->core, slot_index,
+						   module, page, extack);
+}
+
+static int
+mlxsw_sp_get_ts_info(struct net_device *netdev, struct ethtool_ts_info *info)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(netdev);
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+
+	return mlxsw_sp->ptp_ops->get_ts_info(mlxsw_sp, info);
+}
+
+static void
+mlxsw_sp_get_eth_phy_stats(struct net_device *dev,
+			   struct ethtool_eth_phy_stats *phy_stats)
+{
+	char ppcnt_pl[MLXSW_REG_PPCNT_LEN];
+
+	if (mlxsw_sp_port_get_stats_raw(dev, MLXSW_REG_PPCNT_IEEE_8023_CNT,
+					0, ppcnt_pl))
+		return;
+
+	phy_stats->SymbolErrorDuringCarrier =
+		mlxsw_reg_ppcnt_a_symbol_error_during_carrier_get(ppcnt_pl);
+}
+
+static void
+mlxsw_sp_get_eth_mac_stats(struct net_device *dev,
+			   struct ethtool_eth_mac_stats *mac_stats)
+{
+	char ppcnt_pl[MLXSW_REG_PPCNT_LEN];
+
+	if (mlxsw_sp_port_get_stats_raw(dev, MLXSW_REG_PPCNT_IEEE_8023_CNT,
+					0, ppcnt_pl))
+		return;
+
+	mac_stats->FramesTransmittedOK =
+		mlxsw_reg_ppcnt_a_frames_transmitted_ok_get(ppcnt_pl);
+	mac_stats->FramesReceivedOK =
+		mlxsw_reg_ppcnt_a_frames_received_ok_get(ppcnt_pl);
+	mac_stats->FrameCheckSequenceErrors =
+		mlxsw_reg_ppcnt_a_frame_check_sequence_errors_get(ppcnt_pl);
+	mac_stats->AlignmentErrors =
+		mlxsw_reg_ppcnt_a_alignment_errors_get(ppcnt_pl);
+	mac_stats->OctetsTransmittedOK =
+		mlxsw_reg_ppcnt_a_octets_transmitted_ok_get(ppcnt_pl);
+	mac_stats->OctetsReceivedOK =
+		mlxsw_reg_ppcnt_a_octets_received_ok_get(ppcnt_pl);
+	mac_stats->MulticastFramesXmittedOK =
+		mlxsw_reg_ppcnt_a_multicast_frames_xmitted_ok_get(ppcnt_pl);
+	mac_stats->BroadcastFramesXmittedOK =
+		mlxsw_reg_ppcnt_a_broadcast_frames_xmitted_ok_get(ppcnt_pl);
+	mac_stats->MulticastFramesReceivedOK =
+		mlxsw_reg_ppcnt_a_multicast_frames_received_ok_get(ppcnt_pl);
+	mac_stats->BroadcastFramesReceivedOK =
+		mlxsw_reg_ppcnt_a_broadcast_frames_received_ok_get(ppcnt_pl);
+	mac_stats->InRangeLengthErrors =
+		mlxsw_reg_ppcnt_a_in_range_length_errors_get(ppcnt_pl);
+	mac_stats->OutOfRangeLengthField =
+		mlxsw_reg_ppcnt_a_out_of_range_length_field_get(ppcnt_pl);
+	mac_stats->FrameTooLongErrors =
+		mlxsw_reg_ppcnt_a_frame_too_long_errors_get(ppcnt_pl);
+}
+
+static void
+mlxsw_sp_get_eth_ctrl_stats(struct net_device *dev,
+			    struct ethtool_eth_ctrl_stats *ctrl_stats)
+{
+	char ppcnt_pl[MLXSW_REG_PPCNT_LEN];
+
+	if (mlxsw_sp_port_get_stats_raw(dev, MLXSW_REG_PPCNT_IEEE_8023_CNT,
+					0, ppcnt_pl))
+		return;
+
+	ctrl_stats->MACControlFramesTransmitted =
+		mlxsw_reg_ppcnt_a_mac_control_frames_transmitted_get(ppcnt_pl);
+	ctrl_stats->MACControlFramesReceived =
+		mlxsw_reg_ppcnt_a_mac_control_frames_received_get(ppcnt_pl);
+	ctrl_stats->UnsupportedOpcodesReceived =
+		mlxsw_reg_ppcnt_a_unsupported_opcodes_received_get(ppcnt_pl);
+}
+
+static const struct ethtool_rmon_hist_range mlxsw_rmon_ranges[] = {
+	{    0,    64 },
+	{   65,   127 },
+	{  128,   255 },
+	{  256,   511 },
+	{  512,  1023 },
+	{ 1024,  1518 },
+	{ 1519,  2047 },
+	{ 2048,  4095 },
+	{ 4096,  8191 },
+	{ 8192, 10239 },
+	{}
+};
+
+static void
+mlxsw_sp_get_rmon_stats(struct net_device *dev,
+			struct ethtool_rmon_stats *rmon,
+			const struct ethtool_rmon_hist_range **ranges)
+{
+	char ppcnt_pl[MLXSW_REG_PPCNT_LEN];
+
+	if (mlxsw_sp_port_get_stats_raw(dev, MLXSW_REG_PPCNT_RFC_2819_CNT,
+					0, ppcnt_pl))
+		return;
+
+	rmon->undersize_pkts =
+		mlxsw_reg_ppcnt_ether_stats_undersize_pkts_get(ppcnt_pl);
+	rmon->oversize_pkts =
+		mlxsw_reg_ppcnt_ether_stats_oversize_pkts_get(ppcnt_pl);
+	rmon->fragments =
+		mlxsw_reg_ppcnt_ether_stats_fragments_get(ppcnt_pl);
+
+	rmon->hist[0] = mlxsw_reg_ppcnt_ether_stats_pkts64octets_get(ppcnt_pl);
+	rmon->hist[1] =
+		mlxsw_reg_ppcnt_ether_stats_pkts65to127octets_get(ppcnt_pl);
+	rmon->hist[2] =
+		mlxsw_reg_ppcnt_ether_stats_pkts128to255octets_get(ppcnt_pl);
+	rmon->hist[3] =
+		mlxsw_reg_ppcnt_ether_stats_pkts256to511octets_get(ppcnt_pl);
+	rmon->hist[4] =
+		mlxsw_reg_ppcnt_ether_stats_pkts512to1023octets_get(ppcnt_pl);
+	rmon->hist[5] =
+		mlxsw_reg_ppcnt_ether_stats_pkts1024to1518octets_get(ppcnt_pl);
+	rmon->hist[6] =
+		mlxsw_reg_ppcnt_ether_stats_pkts1519to2047octets_get(ppcnt_pl);
+	rmon->hist[7] =
+		mlxsw_reg_ppcnt_ether_stats_pkts2048to4095octets_get(ppcnt_pl);
+	rmon->hist[8] =
+		mlxsw_reg_ppcnt_ether_stats_pkts4096to8191octets_get(ppcnt_pl);
+	rmon->hist[9] =
+		mlxsw_reg_ppcnt_ether_stats_pkts8192to10239octets_get(ppcnt_pl);
+
+	*ranges = mlxsw_rmon_ranges;
+}
+
+static int mlxsw_sp_reset(struct net_device *dev, u32 *flags)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	u8 slot_index = mlxsw_sp_port->mapping.slot_index;
+	u8 module = mlxsw_sp_port->mapping.module;
+
+	return mlxsw_env_reset_module(dev, mlxsw_sp->core, slot_index,
+				      module, flags);
+}
+
+static int
+mlxsw_sp_get_module_power_mode(struct net_device *dev,
+			       struct ethtool_module_power_mode_params *params,
+			       struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	u8 slot_index = mlxsw_sp_port->mapping.slot_index;
+	u8 module = mlxsw_sp_port->mapping.module;
+
+	return mlxsw_env_get_module_power_mode(mlxsw_sp->core, slot_index,
+					       module, params, extack);
+}
+
+static int
+mlxsw_sp_set_module_power_mode(struct net_device *dev,
+			       const struct ethtool_module_power_mode_params *params,
+			       struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	u8 slot_index = mlxsw_sp_port->mapping.slot_index;
+	u8 module = mlxsw_sp_port->mapping.module;
+
+	return mlxsw_env_set_module_power_mode(mlxsw_sp->core, slot_index,
+					       module, params->policy, extack);
+}
+
+const struct ethtool_ops mlxsw_sp_port_ethtool_ops = {
+	.cap_link_lanes_supported	= true,
+	.get_drvinfo			= mlxsw_sp_port_get_drvinfo,
+	.get_link			= ethtool_op_get_link,
+	.get_link_ext_state		= mlxsw_sp_port_get_link_ext_state,
+	.get_pauseparam			= mlxsw_sp_port_get_pauseparam,
+	.set_pauseparam			= mlxsw_sp_port_set_pauseparam,
+	.get_strings			= mlxsw_sp_port_get_strings,
+	.set_phys_id			= mlxsw_sp_port_set_phys_id,
+	.get_ethtool_stats		= mlxsw_sp_port_get_stats,
+	.get_sset_count			= mlxsw_sp_port_get_sset_count,
+	.get_link_ksettings		= mlxsw_sp_port_get_link_ksettings,
+	.set_link_ksettings		= mlxsw_sp_port_set_link_ksettings,
+	.get_module_info		= mlxsw_sp_get_module_info,
+	.get_module_eeprom		= mlxsw_sp_get_module_eeprom,
+	.get_module_eeprom_by_page	= mlxsw_sp_get_module_eeprom_by_page,
+	.get_ts_info			= mlxsw_sp_get_ts_info,
+	.get_eth_phy_stats		= mlxsw_sp_get_eth_phy_stats,
+	.get_eth_mac_stats		= mlxsw_sp_get_eth_mac_stats,
+	.get_eth_ctrl_stats		= mlxsw_sp_get_eth_ctrl_stats,
+	.get_rmon_stats			= mlxsw_sp_get_rmon_stats,
+	.reset				= mlxsw_sp_reset,
+	.get_module_power_mode		= mlxsw_sp_get_module_power_mode,
+	.set_module_power_mode		= mlxsw_sp_set_module_power_mode,
+};
+
+struct mlxsw_sp1_port_link_mode {
+	enum ethtool_link_mode_bit_indices mask_ethtool;
+	u32 mask;
+	u32 speed;
+};
+
+static const struct mlxsw_sp1_port_link_mode mlxsw_sp1_port_link_mode[] = {
+	{
+		.mask		= MLXSW_REG_PTYS_ETH_SPEED_100BASE_T,
+		.mask_ethtool	= ETHTOOL_LINK_MODE_100baseT_Full_BIT,
+		.speed		= SPEED_100,
+	},
+	{
+		.mask		= MLXSW_REG_PTYS_ETH_SPEED_SGMII |
+				  MLXSW_REG_PTYS_ETH_SPEED_1000BASE_KX,
+		.mask_ethtool	= ETHTOOL_LINK_MODE_1000baseKX_Full_BIT,
+		.speed		= SPEED_1000,
+	},
+	{
+		.mask		= MLXSW_REG_PTYS_ETH_SPEED_1000BASE_T,
+		.mask_ethtool   = ETHTOOL_LINK_MODE_1000baseT_Full_BIT,
+		.speed          = SPEED_1000,
+	},
+	{
+		.mask		= MLXSW_REG_PTYS_ETH_SPEED_10GBASE_CX4 |
+				  MLXSW_REG_PTYS_ETH_SPEED_10GBASE_KX4,
+		.mask_ethtool	= ETHTOOL_LINK_MODE_10000baseKX4_Full_BIT,
+		.speed		= SPEED_10000,
+	},
+	{
+		.mask		= MLXSW_REG_PTYS_ETH_SPEED_10GBASE_KR |
+				  MLXSW_REG_PTYS_ETH_SPEED_10GBASE_CR |
+				  MLXSW_REG_PTYS_ETH_SPEED_10GBASE_SR |
+				  MLXSW_REG_PTYS_ETH_SPEED_10GBASE_ER_LR,
+		.mask_ethtool	= ETHTOOL_LINK_MODE_10000baseKR_Full_BIT,
+		.speed		= SPEED_10000,
+	},
+	{
+		.mask		= MLXSW_REG_PTYS_ETH_SPEED_40GBASE_CR4,
+		.mask_ethtool	= ETHTOOL_LINK_MODE_40000baseCR4_Full_BIT,
+		.speed		= SPEED_40000,
+	},
+	{
+		.mask		= MLXSW_REG_PTYS_ETH_SPEED_40GBASE_KR4,
+		.mask_ethtool	= ETHTOOL_LINK_MODE_40000baseKR4_Full_BIT,
+		.speed		= SPEED_40000,
+	},
+	{
+		.mask		= MLXSW_REG_PTYS_ETH_SPEED_40GBASE_SR4,
+		.mask_ethtool	= ETHTOOL_LINK_MODE_40000baseSR4_Full_BIT,
+		.speed		= SPEED_40000,
+	},
+	{
+		.mask		= MLXSW_REG_PTYS_ETH_SPEED_40GBASE_LR4_ER4,
+		.mask_ethtool	= ETHTOOL_LINK_MODE_40000baseLR4_Full_BIT,
+		.speed		= SPEED_40000,
+	},
+	{
+		.mask		= MLXSW_REG_PTYS_ETH_SPEED_25GBASE_CR,
+		.mask_ethtool	= ETHTOOL_LINK_MODE_25000baseCR_Full_BIT,
+		.speed		= SPEED_25000,
+	},
+	{
+		.mask		= MLXSW_REG_PTYS_ETH_SPEED_25GBASE_KR,
+		.mask_ethtool	= ETHTOOL_LINK_MODE_25000baseKR_Full_BIT,
+		.speed		= SPEED_25000,
+	},
+	{
+		.mask		= MLXSW_REG_PTYS_ETH_SPEED_25GBASE_SR,
+		.mask_ethtool	= ETHTOOL_LINK_MODE_25000baseSR_Full_BIT,
+		.speed		= SPEED_25000,
+	},
+	{
+		.mask		= MLXSW_REG_PTYS_ETH_SPEED_50GBASE_CR2,
+		.mask_ethtool	= ETHTOOL_LINK_MODE_50000baseCR2_Full_BIT,
+		.speed		= SPEED_50000,
+	},
+	{
+		.mask		= MLXSW_REG_PTYS_ETH_SPEED_50GBASE_KR2,
+		.mask_ethtool	= ETHTOOL_LINK_MODE_50000baseKR2_Full_BIT,
+		.speed		= SPEED_50000,
+	},
+	{
+		.mask		= MLXSW_REG_PTYS_ETH_SPEED_50GBASE_SR2,
+		.mask_ethtool	= ETHTOOL_LINK_MODE_50000baseSR2_Full_BIT,
+		.speed		= SPEED_50000,
+	},
+	{
+		.mask		= MLXSW_REG_PTYS_ETH_SPEED_100GBASE_CR4,
+		.mask_ethtool	= ETHTOOL_LINK_MODE_100000baseCR4_Full_BIT,
+		.speed		= SPEED_100000,
+	},
+	{
+		.mask		= MLXSW_REG_PTYS_ETH_SPEED_100GBASE_SR4,
+		.mask_ethtool	= ETHTOOL_LINK_MODE_100000baseSR4_Full_BIT,
+		.speed		= SPEED_100000,
+	},
+	{
+		.mask		= MLXSW_REG_PTYS_ETH_SPEED_100GBASE_KR4,
+		.mask_ethtool	= ETHTOOL_LINK_MODE_100000baseKR4_Full_BIT,
+		.speed		= SPEED_100000,
+	},
+	{
+		.mask		= MLXSW_REG_PTYS_ETH_SPEED_100GBASE_LR4_ER4,
+		.mask_ethtool	= ETHTOOL_LINK_MODE_100000baseLR4_ER4_Full_BIT,
+		.speed		= SPEED_100000,
+	},
+};
+
+#define MLXSW_SP1_PORT_LINK_MODE_LEN ARRAY_SIZE(mlxsw_sp1_port_link_mode)
+
+static void
+mlxsw_sp1_from_ptys_supported_port(struct mlxsw_sp *mlxsw_sp,
+				   u32 ptys_eth_proto,
+				   struct ethtool_link_ksettings *cmd)
+{
+	if (ptys_eth_proto & (MLXSW_REG_PTYS_ETH_SPEED_10GBASE_CR |
+			      MLXSW_REG_PTYS_ETH_SPEED_10GBASE_SR |
+			      MLXSW_REG_PTYS_ETH_SPEED_40GBASE_CR4 |
+			      MLXSW_REG_PTYS_ETH_SPEED_40GBASE_SR4 |
+			      MLXSW_REG_PTYS_ETH_SPEED_100GBASE_SR4 |
+			      MLXSW_REG_PTYS_ETH_SPEED_SGMII))
+		ethtool_link_ksettings_add_link_mode(cmd, supported, FIBRE);
+
+	if (ptys_eth_proto & (MLXSW_REG_PTYS_ETH_SPEED_10GBASE_KR |
+			      MLXSW_REG_PTYS_ETH_SPEED_10GBASE_KX4 |
+			      MLXSW_REG_PTYS_ETH_SPEED_40GBASE_KR4 |
+			      MLXSW_REG_PTYS_ETH_SPEED_100GBASE_KR4 |
+			      MLXSW_REG_PTYS_ETH_SPEED_1000BASE_KX))
+		ethtool_link_ksettings_add_link_mode(cmd, supported, Backplane);
+}
+
+static void
+mlxsw_sp1_from_ptys_link(struct mlxsw_sp *mlxsw_sp, u32 ptys_eth_proto,
+			 unsigned long *mode)
+{
+	int i;
+
+	for (i = 0; i < MLXSW_SP1_PORT_LINK_MODE_LEN; i++) {
+		if (ptys_eth_proto & mlxsw_sp1_port_link_mode[i].mask)
+			__set_bit(mlxsw_sp1_port_link_mode[i].mask_ethtool,
+				  mode);
+	}
+}
+
+static u32
+mlxsw_sp1_from_ptys_speed(struct mlxsw_sp *mlxsw_sp, u32 ptys_eth_proto)
+{
+	int i;
+
+	for (i = 0; i < MLXSW_SP1_PORT_LINK_MODE_LEN; i++) {
+		if (ptys_eth_proto & mlxsw_sp1_port_link_mode[i].mask)
+			return mlxsw_sp1_port_link_mode[i].speed;
+	}
+
+	return SPEED_UNKNOWN;
+}
+
+static void
+mlxsw_sp1_from_ptys_link_mode(struct mlxsw_sp *mlxsw_sp, bool carrier_ok,
+			      u32 ptys_eth_proto,
+			      struct ethtool_link_ksettings *cmd)
+{
+	struct mlxsw_sp1_port_link_mode link;
+	int i;
+
+	cmd->base.speed = SPEED_UNKNOWN;
+	cmd->base.duplex = DUPLEX_UNKNOWN;
+	cmd->lanes = 0;
+
+	if (!carrier_ok)
+		return;
+
+	for (i = 0; i < MLXSW_SP1_PORT_LINK_MODE_LEN; i++) {
+		if (ptys_eth_proto & mlxsw_sp1_port_link_mode[i].mask) {
+			link = mlxsw_sp1_port_link_mode[i];
+			ethtool_params_from_link_mode(cmd,
+						      link.mask_ethtool);
+		}
+	}
+}
+
+static int mlxsw_sp1_ptys_max_speed(struct mlxsw_sp_port *mlxsw_sp_port, u32 *p_max_speed)
+{
+	u32 eth_proto_cap;
+	u32 max_speed = 0;
+	int err;
+	int i;
+
+	err = mlxsw_sp_port_ptys_query(mlxsw_sp_port, &eth_proto_cap, NULL, NULL, NULL);
+	if (err)
+		return err;
+
+	for (i = 0; i < MLXSW_SP1_PORT_LINK_MODE_LEN; i++) {
+		if ((eth_proto_cap & mlxsw_sp1_port_link_mode[i].mask) &&
+		    mlxsw_sp1_port_link_mode[i].speed > max_speed)
+			max_speed = mlxsw_sp1_port_link_mode[i].speed;
+	}
+
+	*p_max_speed = max_speed;
+	return 0;
+}
+
+static u32
+mlxsw_sp1_to_ptys_advert_link(struct mlxsw_sp *mlxsw_sp,
+			      const struct ethtool_link_ksettings *cmd)
+{
+	u32 ptys_proto = 0;
+	int i;
+
+	for (i = 0; i < MLXSW_SP1_PORT_LINK_MODE_LEN; i++) {
+		if (test_bit(mlxsw_sp1_port_link_mode[i].mask_ethtool,
+			     cmd->link_modes.advertising))
+			ptys_proto |= mlxsw_sp1_port_link_mode[i].mask;
+	}
+	return ptys_proto;
+}
+
+static u32 mlxsw_sp1_to_ptys_speed_lanes(struct mlxsw_sp *mlxsw_sp, u8 width,
+					 const struct ethtool_link_ksettings *cmd)
+{
+	u32 ptys_proto = 0;
+	int i;
+
+	if (cmd->lanes > width)
+		return ptys_proto;
+
+	for (i = 0; i < MLXSW_SP1_PORT_LINK_MODE_LEN; i++) {
+		if (cmd->base.speed == mlxsw_sp1_port_link_mode[i].speed)
+			ptys_proto |= mlxsw_sp1_port_link_mode[i].mask;
+	}
+	return ptys_proto;
+}
+
+static void
+mlxsw_sp1_reg_ptys_eth_pack(struct mlxsw_sp *mlxsw_sp, char *payload,
+			    u16 local_port, u32 proto_admin, bool autoneg)
+{
+	mlxsw_reg_ptys_eth_pack(payload, local_port, proto_admin, autoneg);
+}
+
+static void
+mlxsw_sp1_reg_ptys_eth_unpack(struct mlxsw_sp *mlxsw_sp, char *payload,
+			      u32 *p_eth_proto_cap, u32 *p_eth_proto_admin,
+			      u32 *p_eth_proto_oper)
+{
+	mlxsw_reg_ptys_eth_unpack(payload, p_eth_proto_cap, p_eth_proto_admin,
+				  p_eth_proto_oper);
+}
+
+static u32 mlxsw_sp1_ptys_proto_cap_masked_get(u32 eth_proto_cap)
+{
+	u32 ptys_proto_cap_masked = 0;
+	int i;
+
+	for (i = 0; i < MLXSW_SP1_PORT_LINK_MODE_LEN; i++) {
+		if (mlxsw_sp1_port_link_mode[i].mask & eth_proto_cap)
+			ptys_proto_cap_masked |=
+				mlxsw_sp1_port_link_mode[i].mask;
+	}
+
+	return ptys_proto_cap_masked;
+}
+
+const struct mlxsw_sp_port_type_speed_ops mlxsw_sp1_port_type_speed_ops = {
+	.from_ptys_supported_port	= mlxsw_sp1_from_ptys_supported_port,
+	.from_ptys_link			= mlxsw_sp1_from_ptys_link,
+	.from_ptys_speed		= mlxsw_sp1_from_ptys_speed,
+	.from_ptys_link_mode		= mlxsw_sp1_from_ptys_link_mode,
+	.ptys_max_speed			= mlxsw_sp1_ptys_max_speed,
+	.to_ptys_advert_link		= mlxsw_sp1_to_ptys_advert_link,
+	.to_ptys_speed_lanes		= mlxsw_sp1_to_ptys_speed_lanes,
+	.reg_ptys_eth_pack		= mlxsw_sp1_reg_ptys_eth_pack,
+	.reg_ptys_eth_unpack		= mlxsw_sp1_reg_ptys_eth_unpack,
+	.ptys_proto_cap_masked_get	= mlxsw_sp1_ptys_proto_cap_masked_get,
+};
+
+static const enum ethtool_link_mode_bit_indices
+mlxsw_sp2_mask_ethtool_sgmii_100m[] = {
+	ETHTOOL_LINK_MODE_100baseT_Full_BIT,
+};
+
+#define MLXSW_SP2_MASK_ETHTOOL_SGMII_100M_LEN \
+	ARRAY_SIZE(mlxsw_sp2_mask_ethtool_sgmii_100m)
+
+static const enum ethtool_link_mode_bit_indices
+mlxsw_sp2_mask_ethtool_1000base_x_sgmii[] = {
+	ETHTOOL_LINK_MODE_1000baseT_Full_BIT,
+	ETHTOOL_LINK_MODE_1000baseKX_Full_BIT,
+};
+
+#define MLXSW_SP2_MASK_ETHTOOL_1000BASE_X_SGMII_LEN \
+	ARRAY_SIZE(mlxsw_sp2_mask_ethtool_1000base_x_sgmii)
+
+static const enum ethtool_link_mode_bit_indices
+mlxsw_sp2_mask_ethtool_5gbase_r[] = {
+	ETHTOOL_LINK_MODE_5000baseT_Full_BIT,
+};
+
+#define MLXSW_SP2_MASK_ETHTOOL_5GBASE_R_LEN \
+	ARRAY_SIZE(mlxsw_sp2_mask_ethtool_5gbase_r)
+
+static const enum ethtool_link_mode_bit_indices
+mlxsw_sp2_mask_ethtool_xfi_xaui_1_10g[] = {
+	ETHTOOL_LINK_MODE_10000baseT_Full_BIT,
+	ETHTOOL_LINK_MODE_10000baseKR_Full_BIT,
+	ETHTOOL_LINK_MODE_10000baseR_FEC_BIT,
+	ETHTOOL_LINK_MODE_10000baseCR_Full_BIT,
+	ETHTOOL_LINK_MODE_10000baseSR_Full_BIT,
+	ETHTOOL_LINK_MODE_10000baseLR_Full_BIT,
+	ETHTOOL_LINK_MODE_10000baseER_Full_BIT,
+};
+
+#define MLXSW_SP2_MASK_ETHTOOL_XFI_XAUI_1_10G_LEN \
+	ARRAY_SIZE(mlxsw_sp2_mask_ethtool_xfi_xaui_1_10g)
+
+static const enum ethtool_link_mode_bit_indices
+mlxsw_sp2_mask_ethtool_xlaui_4_xlppi_4_40g[] = {
+	ETHTOOL_LINK_MODE_40000baseKR4_Full_BIT,
+	ETHTOOL_LINK_MODE_40000baseCR4_Full_BIT,
+	ETHTOOL_LINK_MODE_40000baseSR4_Full_BIT,
+	ETHTOOL_LINK_MODE_40000baseLR4_Full_BIT,
+};
+
+#define MLXSW_SP2_MASK_ETHTOOL_XLAUI_4_XLPPI_4_40G_LEN \
+	ARRAY_SIZE(mlxsw_sp2_mask_ethtool_xlaui_4_xlppi_4_40g)
+
+static const enum ethtool_link_mode_bit_indices
+mlxsw_sp2_mask_ethtool_25gaui_1_25gbase_cr_kr[] = {
+	ETHTOOL_LINK_MODE_25000baseCR_Full_BIT,
+	ETHTOOL_LINK_MODE_25000baseKR_Full_BIT,
+	ETHTOOL_LINK_MODE_25000baseSR_Full_BIT,
+};
+
+#define MLXSW_SP2_MASK_ETHTOOL_25GAUI_1_25GBASE_CR_KR_LEN \
+	ARRAY_SIZE(mlxsw_sp2_mask_ethtool_25gaui_1_25gbase_cr_kr)
+
+static const enum ethtool_link_mode_bit_indices
+mlxsw_sp2_mask_ethtool_50gaui_2_laui_2_50gbase_cr2_kr2[] = {
+	ETHTOOL_LINK_MODE_50000baseCR2_Full_BIT,
+	ETHTOOL_LINK_MODE_50000baseKR2_Full_BIT,
+	ETHTOOL_LINK_MODE_50000baseSR2_Full_BIT,
+};
+
+#define MLXSW_SP2_MASK_ETHTOOL_50GAUI_2_LAUI_2_50GBASE_CR2_KR2_LEN \
+	ARRAY_SIZE(mlxsw_sp2_mask_ethtool_50gaui_2_laui_2_50gbase_cr2_kr2)
+
+static const enum ethtool_link_mode_bit_indices
+mlxsw_sp2_mask_ethtool_50gaui_1_laui_1_50gbase_cr_kr[] = {
+	ETHTOOL_LINK_MODE_50000baseKR_Full_BIT,
+	ETHTOOL_LINK_MODE_50000baseSR_Full_BIT,
+	ETHTOOL_LINK_MODE_50000baseCR_Full_BIT,
+	ETHTOOL_LINK_MODE_50000baseLR_ER_FR_Full_BIT,
+	ETHTOOL_LINK_MODE_50000baseDR_Full_BIT,
+};
+
+#define MLXSW_SP2_MASK_ETHTOOL_50GAUI_1_LAUI_1_50GBASE_CR_KR_LEN \
+	ARRAY_SIZE(mlxsw_sp2_mask_ethtool_50gaui_1_laui_1_50gbase_cr_kr)
+
+static const enum ethtool_link_mode_bit_indices
+mlxsw_sp2_mask_ethtool_caui_4_100gbase_cr4_kr4[] = {
+	ETHTOOL_LINK_MODE_100000baseKR4_Full_BIT,
+	ETHTOOL_LINK_MODE_100000baseSR4_Full_BIT,
+	ETHTOOL_LINK_MODE_100000baseCR4_Full_BIT,
+	ETHTOOL_LINK_MODE_100000baseLR4_ER4_Full_BIT,
+};
+
+#define MLXSW_SP2_MASK_ETHTOOL_CAUI_4_100GBASE_CR4_KR4_LEN \
+	ARRAY_SIZE(mlxsw_sp2_mask_ethtool_caui_4_100gbase_cr4_kr4)
+
+static const enum ethtool_link_mode_bit_indices
+mlxsw_sp2_mask_ethtool_100gaui_2_100gbase_cr2_kr2[] = {
+	ETHTOOL_LINK_MODE_100000baseKR2_Full_BIT,
+	ETHTOOL_LINK_MODE_100000baseSR2_Full_BIT,
+	ETHTOOL_LINK_MODE_100000baseCR2_Full_BIT,
+	ETHTOOL_LINK_MODE_100000baseLR2_ER2_FR2_Full_BIT,
+	ETHTOOL_LINK_MODE_100000baseDR2_Full_BIT,
+};
+
+#define MLXSW_SP2_MASK_ETHTOOL_100GAUI_2_100GBASE_CR2_KR2_LEN \
+	ARRAY_SIZE(mlxsw_sp2_mask_ethtool_100gaui_2_100gbase_cr2_kr2)
+
+static const enum ethtool_link_mode_bit_indices
+mlxsw_sp2_mask_ethtool_200gaui_4_200gbase_cr4_kr4[] = {
+	ETHTOOL_LINK_MODE_200000baseKR4_Full_BIT,
+	ETHTOOL_LINK_MODE_200000baseSR4_Full_BIT,
+	ETHTOOL_LINK_MODE_200000baseLR4_ER4_FR4_Full_BIT,
+	ETHTOOL_LINK_MODE_200000baseDR4_Full_BIT,
+	ETHTOOL_LINK_MODE_200000baseCR4_Full_BIT,
+};
+
+#define MLXSW_SP2_MASK_ETHTOOL_200GAUI_4_200GBASE_CR4_KR4_LEN \
+	ARRAY_SIZE(mlxsw_sp2_mask_ethtool_200gaui_4_200gbase_cr4_kr4)
+
+static const enum ethtool_link_mode_bit_indices
+mlxsw_sp2_mask_ethtool_400gaui_8[] = {
+	ETHTOOL_LINK_MODE_400000baseKR8_Full_BIT,
+	ETHTOOL_LINK_MODE_400000baseSR8_Full_BIT,
+	ETHTOOL_LINK_MODE_400000baseLR8_ER8_FR8_Full_BIT,
+	ETHTOOL_LINK_MODE_400000baseDR8_Full_BIT,
+	ETHTOOL_LINK_MODE_400000baseCR8_Full_BIT,
+};
+
+#define MLXSW_SP2_MASK_ETHTOOL_400GAUI_8_LEN \
+	ARRAY_SIZE(mlxsw_sp2_mask_ethtool_400gaui_8)
+
+#define MLXSW_SP_PORT_MASK_WIDTH_1X	BIT(0)
+#define MLXSW_SP_PORT_MASK_WIDTH_2X	BIT(1)
+#define MLXSW_SP_PORT_MASK_WIDTH_4X	BIT(2)
+#define MLXSW_SP_PORT_MASK_WIDTH_8X	BIT(3)
+
+static u8 mlxsw_sp_port_mask_width_get(u8 width)
+{
+	switch (width) {
+	case 1:
+		return MLXSW_SP_PORT_MASK_WIDTH_1X;
+	case 2:
+		return MLXSW_SP_PORT_MASK_WIDTH_2X;
+	case 4:
+		return MLXSW_SP_PORT_MASK_WIDTH_4X;
+	case 8:
+		return MLXSW_SP_PORT_MASK_WIDTH_8X;
+	default:
+		WARN_ON_ONCE(1);
+		return 0;
+	}
+}
+
+struct mlxsw_sp2_port_link_mode {
+	const enum ethtool_link_mode_bit_indices *mask_ethtool;
+	int m_ethtool_len;
+	u32 mask;
+	u32 speed;
+	u32 width;
+	u8 mask_sup_width;
+};
+
+static const struct mlxsw_sp2_port_link_mode mlxsw_sp2_port_link_mode[] = {
+	{
+		.mask		= MLXSW_REG_PTYS_EXT_ETH_SPEED_SGMII_100M,
+		.mask_ethtool	= mlxsw_sp2_mask_ethtool_sgmii_100m,
+		.m_ethtool_len	= MLXSW_SP2_MASK_ETHTOOL_SGMII_100M_LEN,
+		.mask_sup_width	= MLXSW_SP_PORT_MASK_WIDTH_1X |
+				  MLXSW_SP_PORT_MASK_WIDTH_2X |
+				  MLXSW_SP_PORT_MASK_WIDTH_4X |
+				  MLXSW_SP_PORT_MASK_WIDTH_8X,
+		.speed		= SPEED_100,
+		.width		= 1,
+	},
+	{
+		.mask		= MLXSW_REG_PTYS_EXT_ETH_SPEED_1000BASE_X_SGMII,
+		.mask_ethtool	= mlxsw_sp2_mask_ethtool_1000base_x_sgmii,
+		.m_ethtool_len	= MLXSW_SP2_MASK_ETHTOOL_1000BASE_X_SGMII_LEN,
+		.mask_sup_width	= MLXSW_SP_PORT_MASK_WIDTH_1X |
+				  MLXSW_SP_PORT_MASK_WIDTH_2X |
+				  MLXSW_SP_PORT_MASK_WIDTH_4X |
+				  MLXSW_SP_PORT_MASK_WIDTH_8X,
+		.speed		= SPEED_1000,
+		.width		= 1,
+	},
+	{
+		.mask		= MLXSW_REG_PTYS_EXT_ETH_SPEED_5GBASE_R,
+		.mask_ethtool	= mlxsw_sp2_mask_ethtool_5gbase_r,
+		.m_ethtool_len	= MLXSW_SP2_MASK_ETHTOOL_5GBASE_R_LEN,
+		.mask_sup_width	= MLXSW_SP_PORT_MASK_WIDTH_1X |
+				  MLXSW_SP_PORT_MASK_WIDTH_2X |
+				  MLXSW_SP_PORT_MASK_WIDTH_4X |
+				  MLXSW_SP_PORT_MASK_WIDTH_8X,
+		.speed		= SPEED_5000,
+		.width		= 1,
+	},
+	{
+		.mask		= MLXSW_REG_PTYS_EXT_ETH_SPEED_XFI_XAUI_1_10G,
+		.mask_ethtool	= mlxsw_sp2_mask_ethtool_xfi_xaui_1_10g,
+		.m_ethtool_len	= MLXSW_SP2_MASK_ETHTOOL_XFI_XAUI_1_10G_LEN,
+		.mask_sup_width	= MLXSW_SP_PORT_MASK_WIDTH_1X |
+				  MLXSW_SP_PORT_MASK_WIDTH_2X |
+				  MLXSW_SP_PORT_MASK_WIDTH_4X |
+				  MLXSW_SP_PORT_MASK_WIDTH_8X,
+		.speed		= SPEED_10000,
+		.width		= 1,
+	},
+	{
+		.mask		= MLXSW_REG_PTYS_EXT_ETH_SPEED_XLAUI_4_XLPPI_4_40G,
+		.mask_ethtool	= mlxsw_sp2_mask_ethtool_xlaui_4_xlppi_4_40g,
+		.m_ethtool_len	= MLXSW_SP2_MASK_ETHTOOL_XLAUI_4_XLPPI_4_40G_LEN,
+		.mask_sup_width	= MLXSW_SP_PORT_MASK_WIDTH_4X |
+				  MLXSW_SP_PORT_MASK_WIDTH_8X,
+		.speed		= SPEED_40000,
+		.width		= 4,
+	},
+	{
+		.mask		= MLXSW_REG_PTYS_EXT_ETH_SPEED_25GAUI_1_25GBASE_CR_KR,
+		.mask_ethtool	= mlxsw_sp2_mask_ethtool_25gaui_1_25gbase_cr_kr,
+		.m_ethtool_len	= MLXSW_SP2_MASK_ETHTOOL_25GAUI_1_25GBASE_CR_KR_LEN,
+		.mask_sup_width	= MLXSW_SP_PORT_MASK_WIDTH_1X |
+				  MLXSW_SP_PORT_MASK_WIDTH_2X |
+				  MLXSW_SP_PORT_MASK_WIDTH_4X |
+				  MLXSW_SP_PORT_MASK_WIDTH_8X,
+		.speed		= SPEED_25000,
+		.width		= 1,
+	},
+	{
+		.mask		= MLXSW_REG_PTYS_EXT_ETH_SPEED_50GAUI_2_LAUI_2_50GBASE_CR2_KR2,
+		.mask_ethtool	= mlxsw_sp2_mask_ethtool_50gaui_2_laui_2_50gbase_cr2_kr2,
+		.m_ethtool_len	= MLXSW_SP2_MASK_ETHTOOL_50GAUI_2_LAUI_2_50GBASE_CR2_KR2_LEN,
+		.mask_sup_width	= MLXSW_SP_PORT_MASK_WIDTH_2X |
+				  MLXSW_SP_PORT_MASK_WIDTH_4X |
+				  MLXSW_SP_PORT_MASK_WIDTH_8X,
+		.speed		= SPEED_50000,
+		.width		= 2,
+	},
+	{
+		.mask		= MLXSW_REG_PTYS_EXT_ETH_SPEED_50GAUI_1_LAUI_1_50GBASE_CR_KR,
+		.mask_ethtool	= mlxsw_sp2_mask_ethtool_50gaui_1_laui_1_50gbase_cr_kr,
+		.m_ethtool_len	= MLXSW_SP2_MASK_ETHTOOL_50GAUI_1_LAUI_1_50GBASE_CR_KR_LEN,
+		.mask_sup_width	= MLXSW_SP_PORT_MASK_WIDTH_1X,
+		.speed		= SPEED_50000,
+		.width		= 1,
+	},
+	{
+		.mask		= MLXSW_REG_PTYS_EXT_ETH_SPEED_CAUI_4_100GBASE_CR4_KR4,
+		.mask_ethtool	= mlxsw_sp2_mask_ethtool_caui_4_100gbase_cr4_kr4,
+		.m_ethtool_len	= MLXSW_SP2_MASK_ETHTOOL_CAUI_4_100GBASE_CR4_KR4_LEN,
+		.mask_sup_width	= MLXSW_SP_PORT_MASK_WIDTH_4X |
+				  MLXSW_SP_PORT_MASK_WIDTH_8X,
+		.speed		= SPEED_100000,
+		.width		= 4,
+	},
+	{
+		.mask		= MLXSW_REG_PTYS_EXT_ETH_SPEED_100GAUI_2_100GBASE_CR2_KR2,
+		.mask_ethtool	= mlxsw_sp2_mask_ethtool_100gaui_2_100gbase_cr2_kr2,
+		.m_ethtool_len	= MLXSW_SP2_MASK_ETHTOOL_100GAUI_2_100GBASE_CR2_KR2_LEN,
+		.mask_sup_width	= MLXSW_SP_PORT_MASK_WIDTH_2X,
+		.speed		= SPEED_100000,
+		.width		= 2,
+	},
+	{
+		.mask		= MLXSW_REG_PTYS_EXT_ETH_SPEED_200GAUI_4_200GBASE_CR4_KR4,
+		.mask_ethtool	= mlxsw_sp2_mask_ethtool_200gaui_4_200gbase_cr4_kr4,
+		.m_ethtool_len	= MLXSW_SP2_MASK_ETHTOOL_200GAUI_4_200GBASE_CR4_KR4_LEN,
+		.mask_sup_width	= MLXSW_SP_PORT_MASK_WIDTH_4X |
+				  MLXSW_SP_PORT_MASK_WIDTH_8X,
+		.speed		= SPEED_200000,
+		.width		= 4,
+	},
+	{
+		.mask		= MLXSW_REG_PTYS_EXT_ETH_SPEED_400GAUI_8,
+		.mask_ethtool	= mlxsw_sp2_mask_ethtool_400gaui_8,
+		.m_ethtool_len	= MLXSW_SP2_MASK_ETHTOOL_400GAUI_8_LEN,
+		.mask_sup_width	= MLXSW_SP_PORT_MASK_WIDTH_8X,
+		.speed		= SPEED_400000,
+		.width		= 8,
+	},
+};
+
+#define MLXSW_SP2_PORT_LINK_MODE_LEN ARRAY_SIZE(mlxsw_sp2_port_link_mode)
+
+static void
+mlxsw_sp2_from_ptys_supported_port(struct mlxsw_sp *mlxsw_sp,
+				   u32 ptys_eth_proto,
+				   struct ethtool_link_ksettings *cmd)
+{
+	ethtool_link_ksettings_add_link_mode(cmd, supported, FIBRE);
+	ethtool_link_ksettings_add_link_mode(cmd, supported, Backplane);
+}
+
+static void
+mlxsw_sp2_set_bit_ethtool(const struct mlxsw_sp2_port_link_mode *link_mode,
+			  unsigned long *mode)
+{
+	int i;
+
+	for (i = 0; i < link_mode->m_ethtool_len; i++)
+		__set_bit(link_mode->mask_ethtool[i], mode);
+}
+
+static void
+mlxsw_sp2_from_ptys_link(struct mlxsw_sp *mlxsw_sp, u32 ptys_eth_proto,
+			 unsigned long *mode)
+{
+	int i;
+
+	for (i = 0; i < MLXSW_SP2_PORT_LINK_MODE_LEN; i++) {
+		if (ptys_eth_proto & mlxsw_sp2_port_link_mode[i].mask)
+			mlxsw_sp2_set_bit_ethtool(&mlxsw_sp2_port_link_mode[i],
+						  mode);
+	}
+}
+
+static u32
+mlxsw_sp2_from_ptys_speed(struct mlxsw_sp *mlxsw_sp, u32 ptys_eth_proto)
+{
+	int i;
+
+	for (i = 0; i < MLXSW_SP2_PORT_LINK_MODE_LEN; i++) {
+		if (ptys_eth_proto & mlxsw_sp2_port_link_mode[i].mask)
+			return mlxsw_sp2_port_link_mode[i].speed;
+	}
+
+	return SPEED_UNKNOWN;
+}
+
+static void
+mlxsw_sp2_from_ptys_link_mode(struct mlxsw_sp *mlxsw_sp, bool carrier_ok,
+			      u32 ptys_eth_proto,
+			      struct ethtool_link_ksettings *cmd)
+{
+	struct mlxsw_sp2_port_link_mode link;
+	int i;
+
+	cmd->base.speed = SPEED_UNKNOWN;
+	cmd->base.duplex = DUPLEX_UNKNOWN;
+	cmd->lanes = 0;
+
+	if (!carrier_ok)
+		return;
+
+	for (i = 0; i < MLXSW_SP2_PORT_LINK_MODE_LEN; i++) {
+		if (ptys_eth_proto & mlxsw_sp2_port_link_mode[i].mask) {
+			link = mlxsw_sp2_port_link_mode[i];
+			ethtool_params_from_link_mode(cmd,
+						      link.mask_ethtool[1]);
+		}
+	}
+}
+
+static int mlxsw_sp2_ptys_max_speed(struct mlxsw_sp_port *mlxsw_sp_port, u32 *p_max_speed)
+{
+	u32 eth_proto_cap;
+	u32 max_speed = 0;
+	int err;
+	int i;
+
+	err = mlxsw_sp_port_ptys_query(mlxsw_sp_port, &eth_proto_cap, NULL, NULL, NULL);
+	if (err)
+		return err;
+
+	for (i = 0; i < MLXSW_SP2_PORT_LINK_MODE_LEN; i++) {
+		if ((eth_proto_cap & mlxsw_sp2_port_link_mode[i].mask) &&
+		    mlxsw_sp2_port_link_mode[i].speed > max_speed)
+			max_speed = mlxsw_sp2_port_link_mode[i].speed;
+	}
+
+	*p_max_speed = max_speed;
+	return 0;
+}
+
+static bool
+mlxsw_sp2_test_bit_ethtool(const struct mlxsw_sp2_port_link_mode *link_mode,
+			   const unsigned long *mode)
+{
+	int cnt = 0;
+	int i;
+
+	for (i = 0; i < link_mode->m_ethtool_len; i++) {
+		if (test_bit(link_mode->mask_ethtool[i], mode))
+			cnt++;
+	}
+
+	return cnt == link_mode->m_ethtool_len;
+}
+
+static u32
+mlxsw_sp2_to_ptys_advert_link(struct mlxsw_sp *mlxsw_sp,
+			      const struct ethtool_link_ksettings *cmd)
+{
+	u32 ptys_proto = 0;
+	int i;
+
+	for (i = 0; i < MLXSW_SP2_PORT_LINK_MODE_LEN; i++) {
+		if (mlxsw_sp2_test_bit_ethtool(&mlxsw_sp2_port_link_mode[i],
+					       cmd->link_modes.advertising))
+			ptys_proto |= mlxsw_sp2_port_link_mode[i].mask;
+	}
+	return ptys_proto;
+}
+
+static u32 mlxsw_sp2_to_ptys_speed_lanes(struct mlxsw_sp *mlxsw_sp, u8 width,
+					 const struct ethtool_link_ksettings *cmd)
+{
+	u8 mask_width = mlxsw_sp_port_mask_width_get(width);
+	struct mlxsw_sp2_port_link_mode link_mode;
+	u32 ptys_proto = 0;
+	int i;
+
+	if (cmd->lanes > width)
+		return ptys_proto;
+
+	for (i = 0; i < MLXSW_SP2_PORT_LINK_MODE_LEN; i++) {
+		if (cmd->base.speed == mlxsw_sp2_port_link_mode[i].speed) {
+			link_mode = mlxsw_sp2_port_link_mode[i];
+
+			if (!cmd->lanes) {
+				/* If number of lanes was not set by user space,
+				 * choose the link mode that supports the width
+				 * of the port.
+				 */
+				if (mask_width & link_mode.mask_sup_width)
+					ptys_proto |= link_mode.mask;
+			} else if (cmd->lanes == link_mode.width) {
+				/* Else if the number of lanes was set, choose
+				 * the link mode that its actual width equals to
+				 * it.
+				 */
+				ptys_proto |= link_mode.mask;
+			}
+		}
+	}
+	return ptys_proto;
+}
+
+static void
+mlxsw_sp2_reg_ptys_eth_pack(struct mlxsw_sp *mlxsw_sp, char *payload,
+			    u16 local_port, u32 proto_admin,
+			    bool autoneg)
+{
+	mlxsw_reg_ptys_ext_eth_pack(payload, local_port, proto_admin, autoneg);
+}
+
+static void
+mlxsw_sp2_reg_ptys_eth_unpack(struct mlxsw_sp *mlxsw_sp, char *payload,
+			      u32 *p_eth_proto_cap, u32 *p_eth_proto_admin,
+			      u32 *p_eth_proto_oper)
+{
+	mlxsw_reg_ptys_ext_eth_unpack(payload, p_eth_proto_cap,
+				      p_eth_proto_admin, p_eth_proto_oper);
+}
+
+static u32 mlxsw_sp2_ptys_proto_cap_masked_get(u32 eth_proto_cap)
+{
+	u32 ptys_proto_cap_masked = 0;
+	int i;
+
+	for (i = 0; i < MLXSW_SP2_PORT_LINK_MODE_LEN; i++) {
+		if (mlxsw_sp2_port_link_mode[i].mask & eth_proto_cap)
+			ptys_proto_cap_masked |=
+				mlxsw_sp2_port_link_mode[i].mask;
+	}
+
+	return ptys_proto_cap_masked;
+}
+
+const struct mlxsw_sp_port_type_speed_ops mlxsw_sp2_port_type_speed_ops = {
+	.from_ptys_supported_port	= mlxsw_sp2_from_ptys_supported_port,
+	.from_ptys_link			= mlxsw_sp2_from_ptys_link,
+	.from_ptys_speed		= mlxsw_sp2_from_ptys_speed,
+	.from_ptys_link_mode		= mlxsw_sp2_from_ptys_link_mode,
+	.ptys_max_speed			= mlxsw_sp2_ptys_max_speed,
+	.to_ptys_advert_link		= mlxsw_sp2_to_ptys_advert_link,
+	.to_ptys_speed_lanes		= mlxsw_sp2_to_ptys_speed_lanes,
+	.reg_ptys_eth_pack		= mlxsw_sp2_reg_ptys_eth_pack,
+	.reg_ptys_eth_unpack		= mlxsw_sp2_reg_ptys_eth_unpack,
+	.ptys_proto_cap_masked_get	= mlxsw_sp2_ptys_proto_cap_masked_get,
+};
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_fid.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_fid.c
new file mode 100644
index 000000000..b6ee2d658
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_fid.c
@@ -0,0 +1,1876 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2017-2018 Mellanox Technologies. All rights reserved */
+
+#include <linux/kernel.h>
+#include <linux/bitops.h>
+#include <linux/if_vlan.h>
+#include <linux/if_bridge.h>
+#include <linux/netdevice.h>
+#include <linux/rhashtable.h>
+#include <linux/rtnetlink.h>
+#include <linux/refcount.h>
+
+#include "spectrum.h"
+#include "reg.h"
+
+struct mlxsw_sp_fid_family;
+
+struct mlxsw_sp_fid_core {
+	struct rhashtable fid_ht;
+	struct rhashtable vni_ht;
+	struct mlxsw_sp_fid_family *fid_family_arr[MLXSW_SP_FID_TYPE_MAX];
+	unsigned int *port_fid_mappings;
+};
+
+struct mlxsw_sp_fid_port_vid {
+	struct list_head list;
+	u16 local_port;
+	u16 vid;
+};
+
+struct mlxsw_sp_fid {
+	struct list_head list;
+	struct mlxsw_sp_rif *rif;
+	refcount_t ref_count;
+	u16 fid_index;
+	u16 fid_offset;
+	struct mlxsw_sp_fid_family *fid_family;
+	struct rhash_head ht_node;
+
+	struct rhash_head vni_ht_node;
+	enum mlxsw_sp_nve_type nve_type;
+	__be32 vni;
+	u32 nve_flood_index;
+	int nve_ifindex;
+	u8 vni_valid:1,
+	   nve_flood_index_valid:1;
+	struct list_head port_vid_list; /* Ordered by local port. */
+};
+
+struct mlxsw_sp_fid_8021q {
+	struct mlxsw_sp_fid common;
+	u16 vid;
+};
+
+struct mlxsw_sp_fid_8021d {
+	struct mlxsw_sp_fid common;
+	int br_ifindex;
+};
+
+static const struct rhashtable_params mlxsw_sp_fid_ht_params = {
+	.key_len = sizeof_field(struct mlxsw_sp_fid, fid_index),
+	.key_offset = offsetof(struct mlxsw_sp_fid, fid_index),
+	.head_offset = offsetof(struct mlxsw_sp_fid, ht_node),
+};
+
+static const struct rhashtable_params mlxsw_sp_fid_vni_ht_params = {
+	.key_len = sizeof_field(struct mlxsw_sp_fid, vni),
+	.key_offset = offsetof(struct mlxsw_sp_fid, vni),
+	.head_offset = offsetof(struct mlxsw_sp_fid, vni_ht_node),
+};
+
+struct mlxsw_sp_flood_table {
+	enum mlxsw_sp_flood_type packet_type;
+	enum mlxsw_flood_table_type table_type;
+	int table_index;
+};
+
+struct mlxsw_sp_fid_ops {
+	void (*setup)(struct mlxsw_sp_fid *fid, const void *arg);
+	int (*configure)(struct mlxsw_sp_fid *fid);
+	void (*deconfigure)(struct mlxsw_sp_fid *fid);
+	int (*index_alloc)(struct mlxsw_sp_fid *fid, const void *arg,
+			   u16 *p_fid_index);
+	bool (*compare)(const struct mlxsw_sp_fid *fid,
+			const void *arg);
+	int (*port_vid_map)(struct mlxsw_sp_fid *fid,
+			    struct mlxsw_sp_port *port, u16 vid);
+	void (*port_vid_unmap)(struct mlxsw_sp_fid *fid,
+			       struct mlxsw_sp_port *port, u16 vid);
+	int (*vni_set)(struct mlxsw_sp_fid *fid);
+	void (*vni_clear)(struct mlxsw_sp_fid *fid);
+	int (*nve_flood_index_set)(struct mlxsw_sp_fid *fid);
+	void (*nve_flood_index_clear)(struct mlxsw_sp_fid *fid);
+	void (*fdb_clear_offload)(const struct mlxsw_sp_fid *fid,
+				  const struct net_device *nve_dev);
+	int (*vid_to_fid_rif_update)(const struct mlxsw_sp_fid *fid,
+				     const struct mlxsw_sp_rif *rif);
+};
+
+struct mlxsw_sp_fid_family {
+	enum mlxsw_sp_fid_type type;
+	size_t fid_size;
+	u16 start_index;
+	u16 end_index;
+	struct list_head fids_list;
+	unsigned long *fids_bitmap;
+	const struct mlxsw_sp_flood_table *flood_tables;
+	int nr_flood_tables;
+	enum mlxsw_sp_rif_type rif_type;
+	const struct mlxsw_sp_fid_ops *ops;
+	struct mlxsw_sp *mlxsw_sp;
+	bool flood_rsp;
+	enum mlxsw_reg_bridge_type bridge_type;
+	u16 pgt_base;
+	bool smpe_index_valid;
+};
+
+static const int mlxsw_sp_sfgc_uc_packet_types[MLXSW_REG_SFGC_TYPE_MAX] = {
+	[MLXSW_REG_SFGC_TYPE_UNKNOWN_UNICAST]			= 1,
+};
+
+static const int mlxsw_sp_sfgc_bc_packet_types[MLXSW_REG_SFGC_TYPE_MAX] = {
+	[MLXSW_REG_SFGC_TYPE_BROADCAST]				= 1,
+	[MLXSW_REG_SFGC_TYPE_UNREGISTERED_MULTICAST_NON_IP]	= 1,
+	[MLXSW_REG_SFGC_TYPE_IPV4_LINK_LOCAL]			= 1,
+	[MLXSW_REG_SFGC_TYPE_IPV6_ALL_HOST]			= 1,
+	[MLXSW_REG_SFGC_TYPE_UNREGISTERED_MULTICAST_IPV6]	= 1,
+};
+
+static const int mlxsw_sp_sfgc_mc_packet_types[MLXSW_REG_SFGC_TYPE_MAX] = {
+	[MLXSW_REG_SFGC_TYPE_UNREGISTERED_MULTICAST_IPV4]	= 1,
+};
+
+static const int *mlxsw_sp_packet_type_sfgc_types[] = {
+	[MLXSW_SP_FLOOD_TYPE_UC]	= mlxsw_sp_sfgc_uc_packet_types,
+	[MLXSW_SP_FLOOD_TYPE_BC]	= mlxsw_sp_sfgc_bc_packet_types,
+	[MLXSW_SP_FLOOD_TYPE_MC]	= mlxsw_sp_sfgc_mc_packet_types,
+};
+
+bool mlxsw_sp_fid_is_dummy(struct mlxsw_sp *mlxsw_sp, u16 fid_index)
+{
+	enum mlxsw_sp_fid_type fid_type = MLXSW_SP_FID_TYPE_DUMMY;
+	struct mlxsw_sp_fid_family *fid_family;
+
+	fid_family = mlxsw_sp->fid_core->fid_family_arr[fid_type];
+
+	return fid_family->start_index == fid_index;
+}
+
+struct mlxsw_sp_fid *mlxsw_sp_fid_lookup_by_index(struct mlxsw_sp *mlxsw_sp,
+						  u16 fid_index)
+{
+	struct mlxsw_sp_fid *fid;
+
+	fid = rhashtable_lookup_fast(&mlxsw_sp->fid_core->fid_ht, &fid_index,
+				     mlxsw_sp_fid_ht_params);
+	if (fid)
+		refcount_inc(&fid->ref_count);
+
+	return fid;
+}
+
+int mlxsw_sp_fid_nve_ifindex(const struct mlxsw_sp_fid *fid, int *nve_ifindex)
+{
+	if (!fid->vni_valid)
+		return -EINVAL;
+
+	*nve_ifindex = fid->nve_ifindex;
+
+	return 0;
+}
+
+int mlxsw_sp_fid_nve_type(const struct mlxsw_sp_fid *fid,
+			  enum mlxsw_sp_nve_type *p_type)
+{
+	if (!fid->vni_valid)
+		return -EINVAL;
+
+	*p_type = fid->nve_type;
+
+	return 0;
+}
+
+struct mlxsw_sp_fid *mlxsw_sp_fid_lookup_by_vni(struct mlxsw_sp *mlxsw_sp,
+						__be32 vni)
+{
+	struct mlxsw_sp_fid *fid;
+
+	fid = rhashtable_lookup_fast(&mlxsw_sp->fid_core->vni_ht, &vni,
+				     mlxsw_sp_fid_vni_ht_params);
+	if (fid)
+		refcount_inc(&fid->ref_count);
+
+	return fid;
+}
+
+int mlxsw_sp_fid_vni(const struct mlxsw_sp_fid *fid, __be32 *vni)
+{
+	if (!fid->vni_valid)
+		return -EINVAL;
+
+	*vni = fid->vni;
+
+	return 0;
+}
+
+int mlxsw_sp_fid_nve_flood_index_set(struct mlxsw_sp_fid *fid,
+				     u32 nve_flood_index)
+{
+	struct mlxsw_sp_fid_family *fid_family = fid->fid_family;
+	const struct mlxsw_sp_fid_ops *ops = fid_family->ops;
+	int err;
+
+	if (WARN_ON(fid->nve_flood_index_valid))
+		return -EINVAL;
+
+	fid->nve_flood_index = nve_flood_index;
+	fid->nve_flood_index_valid = true;
+	err = ops->nve_flood_index_set(fid);
+	if (err)
+		goto err_nve_flood_index_set;
+
+	return 0;
+
+err_nve_flood_index_set:
+	fid->nve_flood_index_valid = false;
+	return err;
+}
+
+void mlxsw_sp_fid_nve_flood_index_clear(struct mlxsw_sp_fid *fid)
+{
+	struct mlxsw_sp_fid_family *fid_family = fid->fid_family;
+	const struct mlxsw_sp_fid_ops *ops = fid_family->ops;
+
+	if (WARN_ON(!fid->nve_flood_index_valid))
+		return;
+
+	fid->nve_flood_index_valid = false;
+	ops->nve_flood_index_clear(fid);
+}
+
+bool mlxsw_sp_fid_nve_flood_index_is_set(const struct mlxsw_sp_fid *fid)
+{
+	return fid->nve_flood_index_valid;
+}
+
+int mlxsw_sp_fid_vni_set(struct mlxsw_sp_fid *fid, enum mlxsw_sp_nve_type type,
+			 __be32 vni, int nve_ifindex)
+{
+	struct mlxsw_sp_fid_family *fid_family = fid->fid_family;
+	const struct mlxsw_sp_fid_ops *ops = fid_family->ops;
+	struct mlxsw_sp *mlxsw_sp = fid_family->mlxsw_sp;
+	int err;
+
+	if (WARN_ON(fid->vni_valid))
+		return -EINVAL;
+
+	fid->nve_type = type;
+	fid->nve_ifindex = nve_ifindex;
+	fid->vni = vni;
+	err = rhashtable_lookup_insert_fast(&mlxsw_sp->fid_core->vni_ht,
+					    &fid->vni_ht_node,
+					    mlxsw_sp_fid_vni_ht_params);
+	if (err)
+		return err;
+
+	fid->vni_valid = true;
+	err = ops->vni_set(fid);
+	if (err)
+		goto err_vni_set;
+
+	return 0;
+
+err_vni_set:
+	fid->vni_valid = false;
+	rhashtable_remove_fast(&mlxsw_sp->fid_core->vni_ht, &fid->vni_ht_node,
+			       mlxsw_sp_fid_vni_ht_params);
+	return err;
+}
+
+void mlxsw_sp_fid_vni_clear(struct mlxsw_sp_fid *fid)
+{
+	struct mlxsw_sp_fid_family *fid_family = fid->fid_family;
+	const struct mlxsw_sp_fid_ops *ops = fid_family->ops;
+	struct mlxsw_sp *mlxsw_sp = fid_family->mlxsw_sp;
+
+	if (WARN_ON(!fid->vni_valid))
+		return;
+
+	fid->vni_valid = false;
+	ops->vni_clear(fid);
+	rhashtable_remove_fast(&mlxsw_sp->fid_core->vni_ht, &fid->vni_ht_node,
+			       mlxsw_sp_fid_vni_ht_params);
+}
+
+bool mlxsw_sp_fid_vni_is_set(const struct mlxsw_sp_fid *fid)
+{
+	return fid->vni_valid;
+}
+
+void mlxsw_sp_fid_fdb_clear_offload(const struct mlxsw_sp_fid *fid,
+				    const struct net_device *nve_dev)
+{
+	struct mlxsw_sp_fid_family *fid_family = fid->fid_family;
+	const struct mlxsw_sp_fid_ops *ops = fid_family->ops;
+
+	if (ops->fdb_clear_offload)
+		ops->fdb_clear_offload(fid, nve_dev);
+}
+
+static const struct mlxsw_sp_flood_table *
+mlxsw_sp_fid_flood_table_lookup(const struct mlxsw_sp_fid *fid,
+				enum mlxsw_sp_flood_type packet_type)
+{
+	struct mlxsw_sp_fid_family *fid_family = fid->fid_family;
+	int i;
+
+	for (i = 0; i < fid_family->nr_flood_tables; i++) {
+		if (fid_family->flood_tables[i].packet_type != packet_type)
+			continue;
+		return &fid_family->flood_tables[i];
+	}
+
+	return NULL;
+}
+
+static u16
+mlxsw_sp_fid_family_num_fids(const struct mlxsw_sp_fid_family *fid_family)
+{
+	return fid_family->end_index - fid_family->start_index + 1;
+}
+
+static u16
+mlxsw_sp_fid_flood_table_mid(const struct mlxsw_sp_fid_family *fid_family,
+			     const struct mlxsw_sp_flood_table *flood_table,
+			     u16 fid_offset)
+{
+	u16 num_fids;
+
+	num_fids = mlxsw_sp_fid_family_num_fids(fid_family);
+	return fid_family->pgt_base + num_fids * flood_table->table_index +
+	       fid_offset;
+}
+
+int mlxsw_sp_fid_flood_set(struct mlxsw_sp_fid *fid,
+			   enum mlxsw_sp_flood_type packet_type, u16 local_port,
+			   bool member)
+{
+	struct mlxsw_sp_fid_family *fid_family = fid->fid_family;
+	const struct mlxsw_sp_flood_table *flood_table;
+	u16 mid_index;
+
+	if (WARN_ON(!fid_family->flood_tables))
+		return -EINVAL;
+
+	flood_table = mlxsw_sp_fid_flood_table_lookup(fid, packet_type);
+	if (!flood_table)
+		return -ESRCH;
+
+	mid_index = mlxsw_sp_fid_flood_table_mid(fid_family, flood_table,
+						 fid->fid_offset);
+	return mlxsw_sp_pgt_entry_port_set(fid_family->mlxsw_sp, mid_index,
+					   fid->fid_index, local_port, member);
+}
+
+int mlxsw_sp_fid_port_vid_map(struct mlxsw_sp_fid *fid,
+			      struct mlxsw_sp_port *mlxsw_sp_port, u16 vid)
+{
+	if (WARN_ON(!fid->fid_family->ops->port_vid_map))
+		return -EINVAL;
+	return fid->fid_family->ops->port_vid_map(fid, mlxsw_sp_port, vid);
+}
+
+void mlxsw_sp_fid_port_vid_unmap(struct mlxsw_sp_fid *fid,
+				 struct mlxsw_sp_port *mlxsw_sp_port, u16 vid)
+{
+	fid->fid_family->ops->port_vid_unmap(fid, mlxsw_sp_port, vid);
+}
+
+u16 mlxsw_sp_fid_index(const struct mlxsw_sp_fid *fid)
+{
+	return fid->fid_index;
+}
+
+enum mlxsw_sp_fid_type mlxsw_sp_fid_type(const struct mlxsw_sp_fid *fid)
+{
+	return fid->fid_family->type;
+}
+
+struct mlxsw_sp_rif *mlxsw_sp_fid_rif(const struct mlxsw_sp_fid *fid)
+{
+	return fid->rif;
+}
+
+enum mlxsw_sp_rif_type
+mlxsw_sp_fid_type_rif_type(const struct mlxsw_sp *mlxsw_sp,
+			   enum mlxsw_sp_fid_type type)
+{
+	struct mlxsw_sp_fid_core *fid_core = mlxsw_sp->fid_core;
+
+	return fid_core->fid_family_arr[type]->rif_type;
+}
+
+static struct mlxsw_sp_fid_8021q *
+mlxsw_sp_fid_8021q_fid(const struct mlxsw_sp_fid *fid)
+{
+	return container_of(fid, struct mlxsw_sp_fid_8021q, common);
+}
+
+u16 mlxsw_sp_fid_8021q_vid(const struct mlxsw_sp_fid *fid)
+{
+	return mlxsw_sp_fid_8021q_fid(fid)->vid;
+}
+
+static void mlxsw_sp_fid_8021q_setup(struct mlxsw_sp_fid *fid, const void *arg)
+{
+	u16 vid = *(u16 *) arg;
+
+	mlxsw_sp_fid_8021q_fid(fid)->vid = vid;
+	fid->fid_offset = fid->fid_index - fid->fid_family->start_index;
+}
+
+static enum mlxsw_reg_sfmr_op mlxsw_sp_sfmr_op(bool valid)
+{
+	return valid ? MLXSW_REG_SFMR_OP_CREATE_FID :
+		       MLXSW_REG_SFMR_OP_DESTROY_FID;
+}
+
+static int mlxsw_sp_fid_op(const struct mlxsw_sp_fid *fid, bool valid)
+{
+	struct mlxsw_sp *mlxsw_sp = fid->fid_family->mlxsw_sp;
+	char sfmr_pl[MLXSW_REG_SFMR_LEN];
+	u16 smpe;
+
+	smpe = fid->fid_family->smpe_index_valid ? fid->fid_index : 0;
+
+	mlxsw_reg_sfmr_pack(sfmr_pl, mlxsw_sp_sfmr_op(valid), fid->fid_index,
+			    fid->fid_offset, fid->fid_family->flood_rsp,
+			    fid->fid_family->bridge_type,
+			    fid->fid_family->smpe_index_valid, smpe);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(sfmr), sfmr_pl);
+}
+
+static int mlxsw_sp_fid_edit_op(const struct mlxsw_sp_fid *fid,
+				const struct mlxsw_sp_rif *rif)
+{
+	struct mlxsw_sp *mlxsw_sp = fid->fid_family->mlxsw_sp;
+	char sfmr_pl[MLXSW_REG_SFMR_LEN];
+	u16 smpe;
+
+	smpe = fid->fid_family->smpe_index_valid ? fid->fid_index : 0;
+
+	mlxsw_reg_sfmr_pack(sfmr_pl, MLXSW_REG_SFMR_OP_CREATE_FID,
+			    fid->fid_index, fid->fid_offset,
+			    fid->fid_family->flood_rsp,
+			    fid->fid_family->bridge_type,
+			    fid->fid_family->smpe_index_valid, smpe);
+	mlxsw_reg_sfmr_vv_set(sfmr_pl, fid->vni_valid);
+	mlxsw_reg_sfmr_vni_set(sfmr_pl, be32_to_cpu(fid->vni));
+	mlxsw_reg_sfmr_vtfp_set(sfmr_pl, fid->nve_flood_index_valid);
+	mlxsw_reg_sfmr_nve_tunnel_flood_ptr_set(sfmr_pl, fid->nve_flood_index);
+
+	if (rif) {
+		mlxsw_reg_sfmr_irif_v_set(sfmr_pl, true);
+		mlxsw_reg_sfmr_irif_set(sfmr_pl, mlxsw_sp_rif_index(rif));
+	}
+
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(sfmr), sfmr_pl);
+}
+
+static int mlxsw_sp_fid_vni_to_fid_map(const struct mlxsw_sp_fid *fid,
+				       const struct mlxsw_sp_rif *rif,
+				       bool valid)
+{
+	struct mlxsw_sp *mlxsw_sp = fid->fid_family->mlxsw_sp;
+	char svfa_pl[MLXSW_REG_SVFA_LEN];
+	bool irif_valid;
+	u16 irif_index;
+
+	irif_valid = !!rif;
+	irif_index = rif ? mlxsw_sp_rif_index(rif) : 0;
+
+	mlxsw_reg_svfa_vni_pack(svfa_pl, valid, fid->fid_index,
+				be32_to_cpu(fid->vni), irif_valid, irif_index);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(svfa), svfa_pl);
+}
+
+static int mlxsw_sp_fid_to_fid_rif_update(const struct mlxsw_sp_fid *fid,
+					  const struct mlxsw_sp_rif *rif)
+{
+	return mlxsw_sp_fid_edit_op(fid, rif);
+}
+
+static int mlxsw_sp_fid_vni_to_fid_rif_update(const struct mlxsw_sp_fid *fid,
+					      const struct mlxsw_sp_rif *rif)
+{
+	if (!fid->vni_valid)
+		return 0;
+
+	return mlxsw_sp_fid_vni_to_fid_map(fid, rif, fid->vni_valid);
+}
+
+static int
+mlxsw_sp_fid_vid_to_fid_map(const struct mlxsw_sp_fid *fid, u16 vid, bool valid,
+			    const struct mlxsw_sp_rif *rif)
+{
+	struct mlxsw_sp *mlxsw_sp = fid->fid_family->mlxsw_sp;
+	char svfa_pl[MLXSW_REG_SVFA_LEN];
+	bool irif_valid;
+	u16 irif_index;
+
+	irif_valid = !!rif;
+	irif_index = rif ? mlxsw_sp_rif_index(rif) : 0;
+
+	mlxsw_reg_svfa_vid_pack(svfa_pl, valid, fid->fid_index, vid, irif_valid,
+				irif_index);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(svfa), svfa_pl);
+}
+
+static int
+mlxsw_sp_fid_8021q_vid_to_fid_rif_update(const struct mlxsw_sp_fid *fid,
+					 const struct mlxsw_sp_rif *rif)
+{
+	struct mlxsw_sp_fid_8021q *fid_8021q = mlxsw_sp_fid_8021q_fid(fid);
+
+	/* Update the global VID => FID mapping we created when the FID was
+	 * configured.
+	 */
+	return mlxsw_sp_fid_vid_to_fid_map(fid, fid_8021q->vid, true, rif);
+}
+
+static int
+mlxsw_sp_fid_port_vid_to_fid_rif_update_one(const struct mlxsw_sp_fid *fid,
+					    struct mlxsw_sp_fid_port_vid *pv,
+					    bool irif_valid, u16 irif_index)
+{
+	struct mlxsw_sp *mlxsw_sp = fid->fid_family->mlxsw_sp;
+	char svfa_pl[MLXSW_REG_SVFA_LEN];
+
+	mlxsw_reg_svfa_port_vid_pack(svfa_pl, pv->local_port, true,
+				     fid->fid_index, pv->vid, irif_valid,
+				     irif_index);
+
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(svfa), svfa_pl);
+}
+
+static int mlxsw_sp_fid_vid_to_fid_rif_set(const struct mlxsw_sp_fid *fid,
+					   const struct mlxsw_sp_rif *rif)
+{
+	struct mlxsw_sp *mlxsw_sp = fid->fid_family->mlxsw_sp;
+	struct mlxsw_sp_fid_port_vid *pv;
+	u16 irif_index;
+	int err;
+
+	err = fid->fid_family->ops->vid_to_fid_rif_update(fid, rif);
+	if (err)
+		return err;
+
+	irif_index = mlxsw_sp_rif_index(rif);
+
+	list_for_each_entry(pv, &fid->port_vid_list, list) {
+		/* If port is not in virtual mode, then it does not have any
+		 * {Port, VID}->FID mappings that need to be updated with the
+		 * ingress RIF.
+		 */
+		if (!mlxsw_sp->fid_core->port_fid_mappings[pv->local_port])
+			continue;
+
+		err = mlxsw_sp_fid_port_vid_to_fid_rif_update_one(fid, pv,
+								  true,
+								  irif_index);
+		if (err)
+			goto err_port_vid_to_fid_rif_update_one;
+	}
+
+	return 0;
+
+err_port_vid_to_fid_rif_update_one:
+	list_for_each_entry_continue_reverse(pv, &fid->port_vid_list, list) {
+		if (!mlxsw_sp->fid_core->port_fid_mappings[pv->local_port])
+			continue;
+
+		mlxsw_sp_fid_port_vid_to_fid_rif_update_one(fid, pv, false, 0);
+	}
+
+	fid->fid_family->ops->vid_to_fid_rif_update(fid, NULL);
+	return err;
+}
+
+static void mlxsw_sp_fid_vid_to_fid_rif_unset(const struct mlxsw_sp_fid *fid)
+{
+	struct mlxsw_sp *mlxsw_sp = fid->fid_family->mlxsw_sp;
+	struct mlxsw_sp_fid_port_vid *pv;
+
+	list_for_each_entry(pv, &fid->port_vid_list, list) {
+		/* If port is not in virtual mode, then it does not have any
+		 * {Port, VID}->FID mappings that need to be updated.
+		 */
+		if (!mlxsw_sp->fid_core->port_fid_mappings[pv->local_port])
+			continue;
+
+		mlxsw_sp_fid_port_vid_to_fid_rif_update_one(fid, pv, false, 0);
+	}
+
+	fid->fid_family->ops->vid_to_fid_rif_update(fid, NULL);
+}
+
+static int mlxsw_sp_fid_reiv_handle(struct mlxsw_sp_fid *fid, u16 rif_index,
+				    bool valid, u8 port_page)
+{
+	u16 local_port_end = (port_page + 1) * MLXSW_REG_REIV_REC_MAX_COUNT - 1;
+	u16 local_port_start = port_page * MLXSW_REG_REIV_REC_MAX_COUNT;
+	struct mlxsw_sp *mlxsw_sp = fid->fid_family->mlxsw_sp;
+	struct mlxsw_sp_fid_port_vid *port_vid;
+	u8 rec_num, entries_num = 0;
+	char *reiv_pl;
+	int err;
+
+	reiv_pl = kmalloc(MLXSW_REG_REIV_LEN, GFP_KERNEL);
+	if (!reiv_pl)
+		return -ENOMEM;
+
+	mlxsw_reg_reiv_pack(reiv_pl, port_page, rif_index);
+
+	list_for_each_entry(port_vid, &fid->port_vid_list, list) {
+		/* port_vid_list is sorted by local_port. */
+		if (port_vid->local_port < local_port_start)
+			continue;
+
+		if (port_vid->local_port > local_port_end)
+			break;
+
+		rec_num = port_vid->local_port % MLXSW_REG_REIV_REC_MAX_COUNT;
+		mlxsw_reg_reiv_rec_update_set(reiv_pl, rec_num, true);
+		mlxsw_reg_reiv_rec_evid_set(reiv_pl, rec_num,
+					    valid ? port_vid->vid : 0);
+		entries_num++;
+	}
+
+	if (!entries_num) {
+		kfree(reiv_pl);
+		return 0;
+	}
+
+	err = mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(reiv), reiv_pl);
+	if (err)
+		goto err_reg_write;
+
+	kfree(reiv_pl);
+	return 0;
+
+err_reg_write:
+	kfree(reiv_pl);
+	return err;
+}
+
+static int mlxsw_sp_fid_erif_eport_to_vid_map(struct mlxsw_sp_fid *fid,
+					      u16 rif_index, bool valid)
+{
+	struct mlxsw_sp *mlxsw_sp = fid->fid_family->mlxsw_sp;
+	u8 num_port_pages;
+	int err, i;
+
+	num_port_pages = mlxsw_core_max_ports(mlxsw_sp->core) /
+			 MLXSW_REG_REIV_REC_MAX_COUNT + 1;
+
+	for (i = 0; i < num_port_pages; i++) {
+		err = mlxsw_sp_fid_reiv_handle(fid, rif_index, valid, i);
+		if (err)
+			goto err_reiv_handle;
+	}
+
+	return 0;
+
+err_reiv_handle:
+	for (; i >= 0; i--)
+		mlxsw_sp_fid_reiv_handle(fid, rif_index, !valid, i);
+	return err;
+}
+
+int mlxsw_sp_fid_rif_set(struct mlxsw_sp_fid *fid, struct mlxsw_sp_rif *rif)
+{
+	u16 rif_index = mlxsw_sp_rif_index(rif);
+	int err;
+
+	err = mlxsw_sp_fid_to_fid_rif_update(fid, rif);
+	if (err)
+		return err;
+
+	err = mlxsw_sp_fid_vni_to_fid_rif_update(fid, rif);
+	if (err)
+		goto err_vni_to_fid_rif_update;
+
+	err = mlxsw_sp_fid_vid_to_fid_rif_set(fid, rif);
+	if (err)
+		goto err_vid_to_fid_rif_set;
+
+	err = mlxsw_sp_fid_erif_eport_to_vid_map(fid, rif_index, true);
+	if (err)
+		goto err_erif_eport_to_vid_map;
+
+	fid->rif = rif;
+	return 0;
+
+err_erif_eport_to_vid_map:
+	mlxsw_sp_fid_vid_to_fid_rif_unset(fid);
+err_vid_to_fid_rif_set:
+	mlxsw_sp_fid_vni_to_fid_rif_update(fid, NULL);
+err_vni_to_fid_rif_update:
+	mlxsw_sp_fid_to_fid_rif_update(fid, NULL);
+	return err;
+}
+
+void mlxsw_sp_fid_rif_unset(struct mlxsw_sp_fid *fid)
+{
+	u16 rif_index;
+
+	if (!fid->rif)
+		return;
+
+	rif_index = mlxsw_sp_rif_index(fid->rif);
+	fid->rif = NULL;
+
+	mlxsw_sp_fid_erif_eport_to_vid_map(fid, rif_index, false);
+	mlxsw_sp_fid_vid_to_fid_rif_unset(fid);
+	mlxsw_sp_fid_vni_to_fid_rif_update(fid, NULL);
+	mlxsw_sp_fid_to_fid_rif_update(fid, NULL);
+}
+
+static int mlxsw_sp_fid_vni_op(const struct mlxsw_sp_fid *fid)
+{
+	int err;
+
+	err = mlxsw_sp_fid_vni_to_fid_map(fid, fid->rif, fid->vni_valid);
+	if (err)
+		return err;
+
+	err = mlxsw_sp_fid_edit_op(fid, fid->rif);
+	if (err)
+		goto err_fid_edit_op;
+
+	return 0;
+
+err_fid_edit_op:
+	mlxsw_sp_fid_vni_to_fid_map(fid, fid->rif, !fid->vni_valid);
+	return err;
+}
+
+static int __mlxsw_sp_fid_port_vid_map(const struct mlxsw_sp_fid *fid,
+				       u16 local_port, u16 vid, bool valid)
+{
+	struct mlxsw_sp *mlxsw_sp = fid->fid_family->mlxsw_sp;
+	char svfa_pl[MLXSW_REG_SVFA_LEN];
+	bool irif_valid = false;
+	u16 irif_index = 0;
+
+	if (fid->rif) {
+		irif_valid = true;
+		irif_index = mlxsw_sp_rif_index(fid->rif);
+	}
+
+	mlxsw_reg_svfa_port_vid_pack(svfa_pl, local_port, valid, fid->fid_index,
+				     vid, irif_valid, irif_index);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(svfa), svfa_pl);
+}
+
+static struct mlxsw_sp_fid_8021d *
+mlxsw_sp_fid_8021d_fid(const struct mlxsw_sp_fid *fid)
+{
+	return container_of(fid, struct mlxsw_sp_fid_8021d, common);
+}
+
+static void mlxsw_sp_fid_8021d_setup(struct mlxsw_sp_fid *fid, const void *arg)
+{
+	int br_ifindex = *(int *) arg;
+
+	mlxsw_sp_fid_8021d_fid(fid)->br_ifindex = br_ifindex;
+	fid->fid_offset = fid->fid_index - fid->fid_family->start_index;
+}
+
+static int mlxsw_sp_fid_8021d_configure(struct mlxsw_sp_fid *fid)
+{
+	return mlxsw_sp_fid_op(fid, true);
+}
+
+static void mlxsw_sp_fid_8021d_deconfigure(struct mlxsw_sp_fid *fid)
+{
+	if (fid->vni_valid)
+		mlxsw_sp_nve_fid_disable(fid->fid_family->mlxsw_sp, fid);
+	mlxsw_sp_fid_op(fid, false);
+}
+
+static int mlxsw_sp_fid_8021d_index_alloc(struct mlxsw_sp_fid *fid,
+					  const void *arg, u16 *p_fid_index)
+{
+	struct mlxsw_sp_fid_family *fid_family = fid->fid_family;
+	u16 nr_fids, fid_index;
+
+	nr_fids = fid_family->end_index - fid_family->start_index + 1;
+	fid_index = find_first_zero_bit(fid_family->fids_bitmap, nr_fids);
+	if (fid_index == nr_fids)
+		return -ENOBUFS;
+	*p_fid_index = fid_family->start_index + fid_index;
+
+	return 0;
+}
+
+static bool
+mlxsw_sp_fid_8021d_compare(const struct mlxsw_sp_fid *fid, const void *arg)
+{
+	int br_ifindex = *(int *) arg;
+
+	return mlxsw_sp_fid_8021d_fid(fid)->br_ifindex == br_ifindex;
+}
+
+static int mlxsw_sp_port_vp_mode_trans(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	struct mlxsw_sp_port_vlan *mlxsw_sp_port_vlan;
+	int err;
+
+	list_for_each_entry(mlxsw_sp_port_vlan, &mlxsw_sp_port->vlans_list,
+			    list) {
+		struct mlxsw_sp_fid *fid = mlxsw_sp_port_vlan->fid;
+		u16 vid = mlxsw_sp_port_vlan->vid;
+
+		if (!fid)
+			continue;
+
+		err = __mlxsw_sp_fid_port_vid_map(fid,
+						  mlxsw_sp_port->local_port,
+						  vid, true);
+		if (err)
+			goto err_fid_port_vid_map;
+	}
+
+	err = mlxsw_sp_port_vp_mode_set(mlxsw_sp_port, true);
+	if (err)
+		goto err_port_vp_mode_set;
+
+	return 0;
+
+err_port_vp_mode_set:
+err_fid_port_vid_map:
+	list_for_each_entry_continue_reverse(mlxsw_sp_port_vlan,
+					     &mlxsw_sp_port->vlans_list, list) {
+		struct mlxsw_sp_fid *fid = mlxsw_sp_port_vlan->fid;
+		u16 vid = mlxsw_sp_port_vlan->vid;
+
+		if (!fid)
+			continue;
+
+		__mlxsw_sp_fid_port_vid_map(fid, mlxsw_sp_port->local_port, vid,
+					    false);
+	}
+	return err;
+}
+
+static void mlxsw_sp_port_vlan_mode_trans(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	struct mlxsw_sp_port_vlan *mlxsw_sp_port_vlan;
+
+	mlxsw_sp_port_vp_mode_set(mlxsw_sp_port, false);
+
+	list_for_each_entry_reverse(mlxsw_sp_port_vlan,
+				    &mlxsw_sp_port->vlans_list, list) {
+		struct mlxsw_sp_fid *fid = mlxsw_sp_port_vlan->fid;
+		u16 vid = mlxsw_sp_port_vlan->vid;
+
+		if (!fid)
+			continue;
+
+		__mlxsw_sp_fid_port_vid_map(fid, mlxsw_sp_port->local_port, vid,
+					    false);
+	}
+}
+
+static int
+mlxsw_sp_fid_port_vid_list_add(struct mlxsw_sp_fid *fid, u16 local_port,
+			       u16 vid)
+{
+	struct mlxsw_sp_fid_port_vid *port_vid, *tmp_port_vid;
+
+	port_vid = kzalloc(sizeof(*port_vid), GFP_KERNEL);
+	if (!port_vid)
+		return -ENOMEM;
+
+	port_vid->local_port = local_port;
+	port_vid->vid = vid;
+
+	list_for_each_entry(tmp_port_vid, &fid->port_vid_list, list) {
+		if (tmp_port_vid->local_port > local_port)
+			break;
+	}
+
+	list_add_tail(&port_vid->list, &tmp_port_vid->list);
+	return 0;
+}
+
+static void
+mlxsw_sp_fid_port_vid_list_del(struct mlxsw_sp_fid *fid, u16 local_port,
+			       u16 vid)
+{
+	struct mlxsw_sp_fid_port_vid *port_vid, *tmp;
+
+	list_for_each_entry_safe(port_vid, tmp, &fid->port_vid_list, list) {
+		if (port_vid->local_port != local_port || port_vid->vid != vid)
+			continue;
+
+		list_del(&port_vid->list);
+		kfree(port_vid);
+		return;
+	}
+}
+
+static int
+mlxsw_sp_fid_mpe_table_map(const struct mlxsw_sp_fid *fid, u16 local_port,
+			   u16 vid, bool valid)
+{
+	struct mlxsw_sp *mlxsw_sp = fid->fid_family->mlxsw_sp;
+	char smpe_pl[MLXSW_REG_SMPE_LEN];
+
+	mlxsw_reg_smpe_pack(smpe_pl, local_port, fid->fid_index,
+			    valid ? vid : 0);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(smpe), smpe_pl);
+}
+
+static int
+mlxsw_sp_fid_erif_eport_to_vid_map_one(const struct mlxsw_sp_fid *fid,
+				       u16 local_port, u16 vid, bool valid)
+{
+	u8 port_page = local_port / MLXSW_REG_REIV_REC_MAX_COUNT;
+	u8 rec_num = local_port % MLXSW_REG_REIV_REC_MAX_COUNT;
+	struct mlxsw_sp *mlxsw_sp = fid->fid_family->mlxsw_sp;
+	u16 rif_index = mlxsw_sp_rif_index(fid->rif);
+	char *reiv_pl;
+	int err;
+
+	reiv_pl = kmalloc(MLXSW_REG_REIV_LEN, GFP_KERNEL);
+	if (!reiv_pl)
+		return -ENOMEM;
+
+	mlxsw_reg_reiv_pack(reiv_pl, port_page, rif_index);
+	mlxsw_reg_reiv_rec_update_set(reiv_pl, rec_num, true);
+	mlxsw_reg_reiv_rec_evid_set(reiv_pl, rec_num, valid ? vid : 0);
+	err = mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(reiv), reiv_pl);
+	kfree(reiv_pl);
+	return err;
+}
+
+static int mlxsw_sp_fid_evid_map(const struct mlxsw_sp_fid *fid, u16 local_port,
+				 u16 vid, bool valid)
+{
+	int err;
+
+	err = mlxsw_sp_fid_mpe_table_map(fid, local_port, vid, valid);
+	if (err)
+		return err;
+
+	if (!fid->rif)
+		return 0;
+
+	err = mlxsw_sp_fid_erif_eport_to_vid_map_one(fid, local_port, vid,
+						     valid);
+	if (err)
+		goto err_erif_eport_to_vid_map_one;
+
+	return 0;
+
+err_erif_eport_to_vid_map_one:
+	mlxsw_sp_fid_mpe_table_map(fid, local_port, vid, !valid);
+	return err;
+}
+
+static int mlxsw_sp_fid_8021d_port_vid_map(struct mlxsw_sp_fid *fid,
+					   struct mlxsw_sp_port *mlxsw_sp_port,
+					   u16 vid)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	u16 local_port = mlxsw_sp_port->local_port;
+	int err;
+
+	err = __mlxsw_sp_fid_port_vid_map(fid, mlxsw_sp_port->local_port, vid,
+					  true);
+	if (err)
+		return err;
+
+	err = mlxsw_sp_fid_evid_map(fid, local_port, vid, true);
+	if (err)
+		goto err_fid_evid_map;
+
+	err = mlxsw_sp_fid_port_vid_list_add(fid, mlxsw_sp_port->local_port,
+					     vid);
+	if (err)
+		goto err_port_vid_list_add;
+
+	if (mlxsw_sp->fid_core->port_fid_mappings[local_port]++ == 0) {
+		err = mlxsw_sp_port_vp_mode_trans(mlxsw_sp_port);
+		if (err)
+			goto err_port_vp_mode_trans;
+	}
+
+	return 0;
+
+err_port_vp_mode_trans:
+	mlxsw_sp->fid_core->port_fid_mappings[local_port]--;
+	mlxsw_sp_fid_port_vid_list_del(fid, mlxsw_sp_port->local_port, vid);
+err_port_vid_list_add:
+	mlxsw_sp_fid_evid_map(fid, local_port, vid, false);
+err_fid_evid_map:
+	__mlxsw_sp_fid_port_vid_map(fid, mlxsw_sp_port->local_port, vid, false);
+	return err;
+}
+
+static void
+mlxsw_sp_fid_8021d_port_vid_unmap(struct mlxsw_sp_fid *fid,
+				  struct mlxsw_sp_port *mlxsw_sp_port, u16 vid)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	u16 local_port = mlxsw_sp_port->local_port;
+
+	if (mlxsw_sp->fid_core->port_fid_mappings[local_port] == 1)
+		mlxsw_sp_port_vlan_mode_trans(mlxsw_sp_port);
+	mlxsw_sp->fid_core->port_fid_mappings[local_port]--;
+	mlxsw_sp_fid_port_vid_list_del(fid, mlxsw_sp_port->local_port, vid);
+	mlxsw_sp_fid_evid_map(fid, local_port, vid, false);
+	__mlxsw_sp_fid_port_vid_map(fid, mlxsw_sp_port->local_port, vid, false);
+}
+
+static int mlxsw_sp_fid_8021d_vni_set(struct mlxsw_sp_fid *fid)
+{
+	return mlxsw_sp_fid_vni_op(fid);
+}
+
+static void mlxsw_sp_fid_8021d_vni_clear(struct mlxsw_sp_fid *fid)
+{
+	mlxsw_sp_fid_vni_op(fid);
+}
+
+static int mlxsw_sp_fid_8021d_nve_flood_index_set(struct mlxsw_sp_fid *fid)
+{
+	return mlxsw_sp_fid_edit_op(fid, fid->rif);
+}
+
+static void mlxsw_sp_fid_8021d_nve_flood_index_clear(struct mlxsw_sp_fid *fid)
+{
+	mlxsw_sp_fid_edit_op(fid, fid->rif);
+}
+
+static void
+mlxsw_sp_fid_8021d_fdb_clear_offload(const struct mlxsw_sp_fid *fid,
+				     const struct net_device *nve_dev)
+{
+	br_fdb_clear_offload(nve_dev, 0);
+}
+
+static int
+mlxsw_sp_fid_8021d_vid_to_fid_rif_update(const struct mlxsw_sp_fid *fid,
+					 const struct mlxsw_sp_rif *rif)
+{
+	return 0;
+}
+
+static const struct mlxsw_sp_fid_ops mlxsw_sp_fid_8021d_ops = {
+	.setup			= mlxsw_sp_fid_8021d_setup,
+	.configure		= mlxsw_sp_fid_8021d_configure,
+	.deconfigure		= mlxsw_sp_fid_8021d_deconfigure,
+	.index_alloc		= mlxsw_sp_fid_8021d_index_alloc,
+	.compare		= mlxsw_sp_fid_8021d_compare,
+	.port_vid_map		= mlxsw_sp_fid_8021d_port_vid_map,
+	.port_vid_unmap		= mlxsw_sp_fid_8021d_port_vid_unmap,
+	.vni_set		= mlxsw_sp_fid_8021d_vni_set,
+	.vni_clear		= mlxsw_sp_fid_8021d_vni_clear,
+	.nve_flood_index_set	= mlxsw_sp_fid_8021d_nve_flood_index_set,
+	.nve_flood_index_clear	= mlxsw_sp_fid_8021d_nve_flood_index_clear,
+	.fdb_clear_offload	= mlxsw_sp_fid_8021d_fdb_clear_offload,
+	.vid_to_fid_rif_update  = mlxsw_sp_fid_8021d_vid_to_fid_rif_update,
+};
+
+#define MLXSW_SP_FID_8021Q_MAX (VLAN_N_VID - 2)
+#define MLXSW_SP_FID_RFID_MAX (11 * 1024)
+#define MLXSW_SP_FID_8021Q_PGT_BASE 0
+#define MLXSW_SP_FID_8021D_PGT_BASE (3 * MLXSW_SP_FID_8021Q_MAX)
+
+static const struct mlxsw_sp_flood_table mlxsw_sp_fid_8021d_flood_tables[] = {
+	{
+		.packet_type	= MLXSW_SP_FLOOD_TYPE_UC,
+		.table_type	= MLXSW_REG_SFGC_TABLE_TYPE_FID_OFFSET,
+		.table_index	= 0,
+	},
+	{
+		.packet_type	= MLXSW_SP_FLOOD_TYPE_MC,
+		.table_type	= MLXSW_REG_SFGC_TABLE_TYPE_FID_OFFSET,
+		.table_index	= 1,
+	},
+	{
+		.packet_type	= MLXSW_SP_FLOOD_TYPE_BC,
+		.table_type	= MLXSW_REG_SFGC_TABLE_TYPE_FID_OFFSET,
+		.table_index	= 2,
+	},
+};
+
+static bool
+mlxsw_sp_fid_8021q_compare(const struct mlxsw_sp_fid *fid, const void *arg)
+{
+	u16 vid = *(u16 *) arg;
+
+	return mlxsw_sp_fid_8021q_fid(fid)->vid == vid;
+}
+
+static void
+mlxsw_sp_fid_8021q_fdb_clear_offload(const struct mlxsw_sp_fid *fid,
+				     const struct net_device *nve_dev)
+{
+	br_fdb_clear_offload(nve_dev, mlxsw_sp_fid_8021q_vid(fid));
+}
+
+static void mlxsw_sp_fid_rfid_setup(struct mlxsw_sp_fid *fid, const void *arg)
+{
+	fid->fid_offset = 0;
+}
+
+static int mlxsw_sp_fid_rfid_configure(struct mlxsw_sp_fid *fid)
+{
+	return mlxsw_sp_fid_op(fid, true);
+}
+
+static void mlxsw_sp_fid_rfid_deconfigure(struct mlxsw_sp_fid *fid)
+{
+	mlxsw_sp_fid_op(fid, false);
+}
+
+static int mlxsw_sp_fid_rfid_index_alloc(struct mlxsw_sp_fid *fid,
+					 const void *arg, u16 *p_fid_index)
+{
+	u16 rif_index = *(u16 *) arg;
+
+	*p_fid_index = fid->fid_family->start_index + rif_index;
+
+	return 0;
+}
+
+static bool mlxsw_sp_fid_rfid_compare(const struct mlxsw_sp_fid *fid,
+				      const void *arg)
+{
+	u16 rif_index = *(u16 *) arg;
+
+	return fid->fid_index == rif_index + fid->fid_family->start_index;
+}
+
+static int mlxsw_sp_fid_rfid_port_vid_map(struct mlxsw_sp_fid *fid,
+					  struct mlxsw_sp_port *mlxsw_sp_port,
+					  u16 vid)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	u16 local_port = mlxsw_sp_port->local_port;
+	int err;
+
+	err = mlxsw_sp_fid_port_vid_list_add(fid, mlxsw_sp_port->local_port,
+					     vid);
+	if (err)
+		return err;
+
+	/* Using legacy bridge model, we only need to transition the port to
+	 * virtual mode since {Port, VID} => FID is done by the firmware upon
+	 * RIF creation. Using unified bridge model, we need to map
+	 * {Port, VID} => FID and map egress VID.
+	 */
+	err = __mlxsw_sp_fid_port_vid_map(fid, mlxsw_sp_port->local_port, vid,
+					  true);
+	if (err)
+		goto err_port_vid_map;
+
+	if (fid->rif) {
+		err = mlxsw_sp_fid_erif_eport_to_vid_map_one(fid, local_port,
+							     vid, true);
+		if (err)
+			goto err_erif_eport_to_vid_map_one;
+	}
+
+	if (mlxsw_sp->fid_core->port_fid_mappings[local_port]++ == 0) {
+		err = mlxsw_sp_port_vp_mode_trans(mlxsw_sp_port);
+		if (err)
+			goto err_port_vp_mode_trans;
+	}
+
+	return 0;
+
+err_port_vp_mode_trans:
+	mlxsw_sp->fid_core->port_fid_mappings[local_port]--;
+	if (fid->rif)
+		mlxsw_sp_fid_erif_eport_to_vid_map_one(fid, local_port, vid,
+						       false);
+err_erif_eport_to_vid_map_one:
+	__mlxsw_sp_fid_port_vid_map(fid, mlxsw_sp_port->local_port, vid, false);
+err_port_vid_map:
+	mlxsw_sp_fid_port_vid_list_del(fid, mlxsw_sp_port->local_port, vid);
+	return err;
+}
+
+static void
+mlxsw_sp_fid_rfid_port_vid_unmap(struct mlxsw_sp_fid *fid,
+				 struct mlxsw_sp_port *mlxsw_sp_port, u16 vid)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	u16 local_port = mlxsw_sp_port->local_port;
+
+	if (mlxsw_sp->fid_core->port_fid_mappings[local_port] == 1)
+		mlxsw_sp_port_vlan_mode_trans(mlxsw_sp_port);
+	mlxsw_sp->fid_core->port_fid_mappings[local_port]--;
+
+	if (fid->rif)
+		mlxsw_sp_fid_erif_eport_to_vid_map_one(fid, local_port, vid,
+						       false);
+	__mlxsw_sp_fid_port_vid_map(fid, mlxsw_sp_port->local_port, vid, false);
+	mlxsw_sp_fid_port_vid_list_del(fid, mlxsw_sp_port->local_port, vid);
+}
+
+static int mlxsw_sp_fid_rfid_vni_set(struct mlxsw_sp_fid *fid)
+{
+	return -EOPNOTSUPP;
+}
+
+static void mlxsw_sp_fid_rfid_vni_clear(struct mlxsw_sp_fid *fid)
+{
+	WARN_ON_ONCE(1);
+}
+
+static int mlxsw_sp_fid_rfid_nve_flood_index_set(struct mlxsw_sp_fid *fid)
+{
+	return -EOPNOTSUPP;
+}
+
+static void mlxsw_sp_fid_rfid_nve_flood_index_clear(struct mlxsw_sp_fid *fid)
+{
+	WARN_ON_ONCE(1);
+}
+
+static int
+mlxsw_sp_fid_rfid_vid_to_fid_rif_update(const struct mlxsw_sp_fid *fid,
+					const struct mlxsw_sp_rif *rif)
+{
+	return 0;
+}
+
+static const struct mlxsw_sp_fid_ops mlxsw_sp_fid_rfid_ops = {
+	.setup			= mlxsw_sp_fid_rfid_setup,
+	.configure		= mlxsw_sp_fid_rfid_configure,
+	.deconfigure		= mlxsw_sp_fid_rfid_deconfigure,
+	.index_alloc		= mlxsw_sp_fid_rfid_index_alloc,
+	.compare		= mlxsw_sp_fid_rfid_compare,
+	.port_vid_map		= mlxsw_sp_fid_rfid_port_vid_map,
+	.port_vid_unmap		= mlxsw_sp_fid_rfid_port_vid_unmap,
+	.vni_set                = mlxsw_sp_fid_rfid_vni_set,
+	.vni_clear		= mlxsw_sp_fid_rfid_vni_clear,
+	.nve_flood_index_set	= mlxsw_sp_fid_rfid_nve_flood_index_set,
+	.nve_flood_index_clear	= mlxsw_sp_fid_rfid_nve_flood_index_clear,
+	.vid_to_fid_rif_update  = mlxsw_sp_fid_rfid_vid_to_fid_rif_update,
+};
+
+static void mlxsw_sp_fid_dummy_setup(struct mlxsw_sp_fid *fid, const void *arg)
+{
+	fid->fid_offset = 0;
+}
+
+static int mlxsw_sp_fid_dummy_configure(struct mlxsw_sp_fid *fid)
+{
+	return mlxsw_sp_fid_op(fid, true);
+}
+
+static void mlxsw_sp_fid_dummy_deconfigure(struct mlxsw_sp_fid *fid)
+{
+	mlxsw_sp_fid_op(fid, false);
+}
+
+static int mlxsw_sp_fid_dummy_index_alloc(struct mlxsw_sp_fid *fid,
+					  const void *arg, u16 *p_fid_index)
+{
+	*p_fid_index = fid->fid_family->start_index;
+
+	return 0;
+}
+
+static bool mlxsw_sp_fid_dummy_compare(const struct mlxsw_sp_fid *fid,
+				       const void *arg)
+{
+	return true;
+}
+
+static int mlxsw_sp_fid_dummy_vni_set(struct mlxsw_sp_fid *fid)
+{
+	return -EOPNOTSUPP;
+}
+
+static void mlxsw_sp_fid_dummy_vni_clear(struct mlxsw_sp_fid *fid)
+{
+	WARN_ON_ONCE(1);
+}
+
+static int mlxsw_sp_fid_dummy_nve_flood_index_set(struct mlxsw_sp_fid *fid)
+{
+	return -EOPNOTSUPP;
+}
+
+static void mlxsw_sp_fid_dummy_nve_flood_index_clear(struct mlxsw_sp_fid *fid)
+{
+	WARN_ON_ONCE(1);
+}
+
+static const struct mlxsw_sp_fid_ops mlxsw_sp_fid_dummy_ops = {
+	.setup			= mlxsw_sp_fid_dummy_setup,
+	.configure		= mlxsw_sp_fid_dummy_configure,
+	.deconfigure		= mlxsw_sp_fid_dummy_deconfigure,
+	.index_alloc		= mlxsw_sp_fid_dummy_index_alloc,
+	.compare		= mlxsw_sp_fid_dummy_compare,
+	.vni_set                = mlxsw_sp_fid_dummy_vni_set,
+	.vni_clear		= mlxsw_sp_fid_dummy_vni_clear,
+	.nve_flood_index_set	= mlxsw_sp_fid_dummy_nve_flood_index_set,
+	.nve_flood_index_clear	= mlxsw_sp_fid_dummy_nve_flood_index_clear,
+};
+
+static int mlxsw_sp_fid_8021q_configure(struct mlxsw_sp_fid *fid)
+{
+	struct mlxsw_sp_fid_8021q *fid_8021q = mlxsw_sp_fid_8021q_fid(fid);
+	int err;
+
+	err = mlxsw_sp_fid_op(fid, true);
+	if (err)
+		return err;
+
+	err = mlxsw_sp_fid_vid_to_fid_map(fid, fid_8021q->vid, true, fid->rif);
+	if (err)
+		goto err_vid_to_fid_map;
+
+	return 0;
+
+err_vid_to_fid_map:
+	mlxsw_sp_fid_op(fid, false);
+	return err;
+}
+
+static void mlxsw_sp_fid_8021q_deconfigure(struct mlxsw_sp_fid *fid)
+{
+	struct mlxsw_sp_fid_8021q *fid_8021q = mlxsw_sp_fid_8021q_fid(fid);
+
+	if (fid->vni_valid)
+		mlxsw_sp_nve_fid_disable(fid->fid_family->mlxsw_sp, fid);
+
+	mlxsw_sp_fid_vid_to_fid_map(fid, fid_8021q->vid, false, NULL);
+	mlxsw_sp_fid_op(fid, false);
+}
+
+static int mlxsw_sp_fid_8021q_port_vid_map(struct mlxsw_sp_fid *fid,
+					   struct mlxsw_sp_port *mlxsw_sp_port,
+					   u16 vid)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	u16 local_port = mlxsw_sp_port->local_port;
+	int err;
+
+	/* In case there are no {Port, VID} => FID mappings on the port,
+	 * we can use the global VID => FID mapping we created when the
+	 * FID was configured, otherwise, configure new mapping.
+	 */
+	if (mlxsw_sp->fid_core->port_fid_mappings[local_port]) {
+		err =  __mlxsw_sp_fid_port_vid_map(fid, local_port, vid, true);
+		if (err)
+			return err;
+	}
+
+	err = mlxsw_sp_fid_evid_map(fid, local_port, vid, true);
+	if (err)
+		goto err_fid_evid_map;
+
+	err = mlxsw_sp_fid_port_vid_list_add(fid, mlxsw_sp_port->local_port,
+					     vid);
+	if (err)
+		goto err_port_vid_list_add;
+
+	return 0;
+
+err_port_vid_list_add:
+	 mlxsw_sp_fid_evid_map(fid, local_port, vid, false);
+err_fid_evid_map:
+	if (mlxsw_sp->fid_core->port_fid_mappings[local_port])
+		__mlxsw_sp_fid_port_vid_map(fid, local_port, vid, false);
+	return err;
+}
+
+static void
+mlxsw_sp_fid_8021q_port_vid_unmap(struct mlxsw_sp_fid *fid,
+				  struct mlxsw_sp_port *mlxsw_sp_port, u16 vid)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	u16 local_port = mlxsw_sp_port->local_port;
+
+	mlxsw_sp_fid_port_vid_list_del(fid, mlxsw_sp_port->local_port, vid);
+	mlxsw_sp_fid_evid_map(fid, local_port, vid, false);
+	if (mlxsw_sp->fid_core->port_fid_mappings[local_port])
+		__mlxsw_sp_fid_port_vid_map(fid, local_port, vid, false);
+}
+
+static const struct mlxsw_sp_fid_ops mlxsw_sp_fid_8021q_ops = {
+	.setup			= mlxsw_sp_fid_8021q_setup,
+	.configure		= mlxsw_sp_fid_8021q_configure,
+	.deconfigure		= mlxsw_sp_fid_8021q_deconfigure,
+	.index_alloc		= mlxsw_sp_fid_8021d_index_alloc,
+	.compare		= mlxsw_sp_fid_8021q_compare,
+	.port_vid_map		= mlxsw_sp_fid_8021q_port_vid_map,
+	.port_vid_unmap		= mlxsw_sp_fid_8021q_port_vid_unmap,
+	.vni_set		= mlxsw_sp_fid_8021d_vni_set,
+	.vni_clear		= mlxsw_sp_fid_8021d_vni_clear,
+	.nve_flood_index_set	= mlxsw_sp_fid_8021d_nve_flood_index_set,
+	.nve_flood_index_clear	= mlxsw_sp_fid_8021d_nve_flood_index_clear,
+	.fdb_clear_offload	= mlxsw_sp_fid_8021q_fdb_clear_offload,
+	.vid_to_fid_rif_update  = mlxsw_sp_fid_8021q_vid_to_fid_rif_update,
+};
+
+/* There are 4K-2 802.1Q FIDs */
+#define MLXSW_SP_FID_8021Q_START	1 /* FID 0 is reserved. */
+#define MLXSW_SP_FID_8021Q_END		(MLXSW_SP_FID_8021Q_START + \
+					 MLXSW_SP_FID_8021Q_MAX - 1)
+
+/* There are 1K 802.1D FIDs */
+#define MLXSW_SP_FID_8021D_START	(MLXSW_SP_FID_8021Q_END + 1)
+#define MLXSW_SP_FID_8021D_END		(MLXSW_SP_FID_8021D_START + \
+					 MLXSW_SP_FID_8021D_MAX - 1)
+
+/* There is one dummy FID */
+#define MLXSW_SP_FID_DUMMY		(MLXSW_SP_FID_8021D_END + 1)
+
+/* There are 11K rFIDs */
+#define MLXSW_SP_RFID_START		(MLXSW_SP_FID_DUMMY + 1)
+#define MLXSW_SP_RFID_END		(MLXSW_SP_RFID_START + \
+					 MLXSW_SP_FID_RFID_MAX - 1)
+
+static const struct mlxsw_sp_fid_family mlxsw_sp1_fid_8021q_family = {
+	.type			= MLXSW_SP_FID_TYPE_8021Q,
+	.fid_size		= sizeof(struct mlxsw_sp_fid_8021q),
+	.start_index		= MLXSW_SP_FID_8021Q_START,
+	.end_index		= MLXSW_SP_FID_8021Q_END,
+	.flood_tables		= mlxsw_sp_fid_8021d_flood_tables,
+	.nr_flood_tables	= ARRAY_SIZE(mlxsw_sp_fid_8021d_flood_tables),
+	.rif_type		= MLXSW_SP_RIF_TYPE_VLAN,
+	.ops			= &mlxsw_sp_fid_8021q_ops,
+	.flood_rsp              = false,
+	.bridge_type            = MLXSW_REG_BRIDGE_TYPE_0,
+	.pgt_base		= MLXSW_SP_FID_8021Q_PGT_BASE,
+	.smpe_index_valid	= false,
+};
+
+static const struct mlxsw_sp_fid_family mlxsw_sp1_fid_8021d_family = {
+	.type			= MLXSW_SP_FID_TYPE_8021D,
+	.fid_size		= sizeof(struct mlxsw_sp_fid_8021d),
+	.start_index		= MLXSW_SP_FID_8021D_START,
+	.end_index		= MLXSW_SP_FID_8021D_END,
+	.flood_tables		= mlxsw_sp_fid_8021d_flood_tables,
+	.nr_flood_tables	= ARRAY_SIZE(mlxsw_sp_fid_8021d_flood_tables),
+	.rif_type		= MLXSW_SP_RIF_TYPE_FID,
+	.ops			= &mlxsw_sp_fid_8021d_ops,
+	.bridge_type            = MLXSW_REG_BRIDGE_TYPE_1,
+	.pgt_base		= MLXSW_SP_FID_8021D_PGT_BASE,
+	.smpe_index_valid       = false,
+};
+
+static const struct mlxsw_sp_fid_family mlxsw_sp1_fid_dummy_family = {
+	.type			= MLXSW_SP_FID_TYPE_DUMMY,
+	.fid_size		= sizeof(struct mlxsw_sp_fid),
+	.start_index		= MLXSW_SP_FID_DUMMY,
+	.end_index		= MLXSW_SP_FID_DUMMY,
+	.ops			= &mlxsw_sp_fid_dummy_ops,
+	.smpe_index_valid       = false,
+};
+
+static const struct mlxsw_sp_fid_family mlxsw_sp_fid_rfid_family = {
+	.type			= MLXSW_SP_FID_TYPE_RFID,
+	.fid_size		= sizeof(struct mlxsw_sp_fid),
+	.start_index		= MLXSW_SP_RFID_START,
+	.end_index		= MLXSW_SP_RFID_END,
+	.rif_type		= MLXSW_SP_RIF_TYPE_SUBPORT,
+	.ops			= &mlxsw_sp_fid_rfid_ops,
+	.flood_rsp              = true,
+	.smpe_index_valid       = false,
+};
+
+const struct mlxsw_sp_fid_family *mlxsw_sp1_fid_family_arr[] = {
+	[MLXSW_SP_FID_TYPE_8021Q]	= &mlxsw_sp1_fid_8021q_family,
+	[MLXSW_SP_FID_TYPE_8021D]	= &mlxsw_sp1_fid_8021d_family,
+	[MLXSW_SP_FID_TYPE_DUMMY]	= &mlxsw_sp1_fid_dummy_family,
+	[MLXSW_SP_FID_TYPE_RFID]	= &mlxsw_sp_fid_rfid_family,
+};
+
+static const struct mlxsw_sp_fid_family mlxsw_sp2_fid_8021q_family = {
+	.type			= MLXSW_SP_FID_TYPE_8021Q,
+	.fid_size		= sizeof(struct mlxsw_sp_fid_8021q),
+	.start_index		= MLXSW_SP_FID_8021Q_START,
+	.end_index		= MLXSW_SP_FID_8021Q_END,
+	.flood_tables		= mlxsw_sp_fid_8021d_flood_tables,
+	.nr_flood_tables	= ARRAY_SIZE(mlxsw_sp_fid_8021d_flood_tables),
+	.rif_type		= MLXSW_SP_RIF_TYPE_VLAN,
+	.ops			= &mlxsw_sp_fid_8021q_ops,
+	.flood_rsp              = false,
+	.bridge_type            = MLXSW_REG_BRIDGE_TYPE_0,
+	.pgt_base		= MLXSW_SP_FID_8021Q_PGT_BASE,
+	.smpe_index_valid	= true,
+};
+
+static const struct mlxsw_sp_fid_family mlxsw_sp2_fid_8021d_family = {
+	.type			= MLXSW_SP_FID_TYPE_8021D,
+	.fid_size		= sizeof(struct mlxsw_sp_fid_8021d),
+	.start_index		= MLXSW_SP_FID_8021D_START,
+	.end_index		= MLXSW_SP_FID_8021D_END,
+	.flood_tables		= mlxsw_sp_fid_8021d_flood_tables,
+	.nr_flood_tables	= ARRAY_SIZE(mlxsw_sp_fid_8021d_flood_tables),
+	.rif_type		= MLXSW_SP_RIF_TYPE_FID,
+	.ops			= &mlxsw_sp_fid_8021d_ops,
+	.bridge_type            = MLXSW_REG_BRIDGE_TYPE_1,
+	.pgt_base		= MLXSW_SP_FID_8021D_PGT_BASE,
+	.smpe_index_valid       = true,
+};
+
+static const struct mlxsw_sp_fid_family mlxsw_sp2_fid_dummy_family = {
+	.type			= MLXSW_SP_FID_TYPE_DUMMY,
+	.fid_size		= sizeof(struct mlxsw_sp_fid),
+	.start_index		= MLXSW_SP_FID_DUMMY,
+	.end_index		= MLXSW_SP_FID_DUMMY,
+	.ops			= &mlxsw_sp_fid_dummy_ops,
+	.smpe_index_valid       = false,
+};
+
+const struct mlxsw_sp_fid_family *mlxsw_sp2_fid_family_arr[] = {
+	[MLXSW_SP_FID_TYPE_8021Q]	= &mlxsw_sp2_fid_8021q_family,
+	[MLXSW_SP_FID_TYPE_8021D]	= &mlxsw_sp2_fid_8021d_family,
+	[MLXSW_SP_FID_TYPE_DUMMY]	= &mlxsw_sp2_fid_dummy_family,
+	[MLXSW_SP_FID_TYPE_RFID]	= &mlxsw_sp_fid_rfid_family,
+};
+
+static struct mlxsw_sp_fid *mlxsw_sp_fid_lookup(struct mlxsw_sp *mlxsw_sp,
+						enum mlxsw_sp_fid_type type,
+						const void *arg)
+{
+	struct mlxsw_sp_fid_family *fid_family;
+	struct mlxsw_sp_fid *fid;
+
+	fid_family = mlxsw_sp->fid_core->fid_family_arr[type];
+	list_for_each_entry(fid, &fid_family->fids_list, list) {
+		if (!fid->fid_family->ops->compare(fid, arg))
+			continue;
+		refcount_inc(&fid->ref_count);
+		return fid;
+	}
+
+	return NULL;
+}
+
+static struct mlxsw_sp_fid *mlxsw_sp_fid_get(struct mlxsw_sp *mlxsw_sp,
+					     enum mlxsw_sp_fid_type type,
+					     const void *arg)
+{
+	struct mlxsw_sp_fid_family *fid_family;
+	struct mlxsw_sp_fid *fid;
+	u16 fid_index;
+	int err;
+
+	fid = mlxsw_sp_fid_lookup(mlxsw_sp, type, arg);
+	if (fid)
+		return fid;
+
+	fid_family = mlxsw_sp->fid_core->fid_family_arr[type];
+	fid = kzalloc(fid_family->fid_size, GFP_KERNEL);
+	if (!fid)
+		return ERR_PTR(-ENOMEM);
+
+	INIT_LIST_HEAD(&fid->port_vid_list);
+	fid->fid_family = fid_family;
+
+	err = fid->fid_family->ops->index_alloc(fid, arg, &fid_index);
+	if (err)
+		goto err_index_alloc;
+	fid->fid_index = fid_index;
+	__set_bit(fid_index - fid_family->start_index, fid_family->fids_bitmap);
+
+	fid->fid_family->ops->setup(fid, arg);
+
+	err = fid->fid_family->ops->configure(fid);
+	if (err)
+		goto err_configure;
+
+	err = rhashtable_insert_fast(&mlxsw_sp->fid_core->fid_ht, &fid->ht_node,
+				     mlxsw_sp_fid_ht_params);
+	if (err)
+		goto err_rhashtable_insert;
+
+	list_add(&fid->list, &fid_family->fids_list);
+	refcount_set(&fid->ref_count, 1);
+	return fid;
+
+err_rhashtable_insert:
+	fid->fid_family->ops->deconfigure(fid);
+err_configure:
+	__clear_bit(fid_index - fid_family->start_index,
+		    fid_family->fids_bitmap);
+err_index_alloc:
+	kfree(fid);
+	return ERR_PTR(err);
+}
+
+void mlxsw_sp_fid_put(struct mlxsw_sp_fid *fid)
+{
+	struct mlxsw_sp_fid_family *fid_family = fid->fid_family;
+	struct mlxsw_sp *mlxsw_sp = fid_family->mlxsw_sp;
+
+	if (!refcount_dec_and_test(&fid->ref_count))
+		return;
+
+	list_del(&fid->list);
+	rhashtable_remove_fast(&mlxsw_sp->fid_core->fid_ht,
+			       &fid->ht_node, mlxsw_sp_fid_ht_params);
+	fid->fid_family->ops->deconfigure(fid);
+	__clear_bit(fid->fid_index - fid_family->start_index,
+		    fid_family->fids_bitmap);
+	WARN_ON_ONCE(!list_empty(&fid->port_vid_list));
+	kfree(fid);
+}
+
+struct mlxsw_sp_fid *mlxsw_sp_fid_8021q_get(struct mlxsw_sp *mlxsw_sp, u16 vid)
+{
+	return mlxsw_sp_fid_get(mlxsw_sp, MLXSW_SP_FID_TYPE_8021Q, &vid);
+}
+
+struct mlxsw_sp_fid *mlxsw_sp_fid_8021d_get(struct mlxsw_sp *mlxsw_sp,
+					    int br_ifindex)
+{
+	return mlxsw_sp_fid_get(mlxsw_sp, MLXSW_SP_FID_TYPE_8021D, &br_ifindex);
+}
+
+struct mlxsw_sp_fid *mlxsw_sp_fid_8021q_lookup(struct mlxsw_sp *mlxsw_sp,
+					       u16 vid)
+{
+	return mlxsw_sp_fid_lookup(mlxsw_sp, MLXSW_SP_FID_TYPE_8021Q, &vid);
+}
+
+struct mlxsw_sp_fid *mlxsw_sp_fid_8021d_lookup(struct mlxsw_sp *mlxsw_sp,
+					       int br_ifindex)
+{
+	return mlxsw_sp_fid_lookup(mlxsw_sp, MLXSW_SP_FID_TYPE_8021D,
+				   &br_ifindex);
+}
+
+struct mlxsw_sp_fid *mlxsw_sp_fid_rfid_get(struct mlxsw_sp *mlxsw_sp,
+					   u16 rif_index)
+{
+	return mlxsw_sp_fid_get(mlxsw_sp, MLXSW_SP_FID_TYPE_RFID, &rif_index);
+}
+
+struct mlxsw_sp_fid *mlxsw_sp_fid_dummy_get(struct mlxsw_sp *mlxsw_sp)
+{
+	return mlxsw_sp_fid_get(mlxsw_sp, MLXSW_SP_FID_TYPE_DUMMY, NULL);
+}
+
+static int
+mlxsw_sp_fid_flood_table_init(struct mlxsw_sp_fid_family *fid_family,
+			      const struct mlxsw_sp_flood_table *flood_table)
+{
+	enum mlxsw_sp_flood_type packet_type = flood_table->packet_type;
+	struct mlxsw_sp *mlxsw_sp = fid_family->mlxsw_sp;
+	const int *sfgc_packet_types;
+	u16 num_fids, mid_base;
+	int err, i;
+
+	mid_base = mlxsw_sp_fid_flood_table_mid(fid_family, flood_table, 0);
+	num_fids = mlxsw_sp_fid_family_num_fids(fid_family);
+	err = mlxsw_sp_pgt_mid_alloc_range(mlxsw_sp, mid_base, num_fids);
+	if (err)
+		return err;
+
+	sfgc_packet_types = mlxsw_sp_packet_type_sfgc_types[packet_type];
+	for (i = 0; i < MLXSW_REG_SFGC_TYPE_MAX; i++) {
+		char sfgc_pl[MLXSW_REG_SFGC_LEN];
+
+		if (!sfgc_packet_types[i])
+			continue;
+
+		mlxsw_reg_sfgc_pack(sfgc_pl, i, fid_family->bridge_type,
+				    flood_table->table_type, 0, mid_base);
+
+		err = mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(sfgc), sfgc_pl);
+		if (err)
+			goto err_reg_write;
+	}
+
+	return 0;
+
+err_reg_write:
+	mlxsw_sp_pgt_mid_free_range(mlxsw_sp, mid_base, num_fids);
+	return err;
+}
+
+static void
+mlxsw_sp_fid_flood_table_fini(struct mlxsw_sp_fid_family *fid_family,
+			      const struct mlxsw_sp_flood_table *flood_table)
+{
+	struct mlxsw_sp *mlxsw_sp = fid_family->mlxsw_sp;
+	u16 num_fids, mid_base;
+
+	mid_base = mlxsw_sp_fid_flood_table_mid(fid_family, flood_table, 0);
+	num_fids = mlxsw_sp_fid_family_num_fids(fid_family);
+	mlxsw_sp_pgt_mid_free_range(mlxsw_sp, mid_base, num_fids);
+}
+
+static int
+mlxsw_sp_fid_flood_tables_init(struct mlxsw_sp_fid_family *fid_family)
+{
+	int i;
+
+	for (i = 0; i < fid_family->nr_flood_tables; i++) {
+		const struct mlxsw_sp_flood_table *flood_table;
+		int err;
+
+		flood_table = &fid_family->flood_tables[i];
+		err = mlxsw_sp_fid_flood_table_init(fid_family, flood_table);
+		if (err)
+			return err;
+	}
+
+	return 0;
+}
+
+static void
+mlxsw_sp_fid_flood_tables_fini(struct mlxsw_sp_fid_family *fid_family)
+{
+	int i;
+
+	for (i = 0; i < fid_family->nr_flood_tables; i++) {
+		const struct mlxsw_sp_flood_table *flood_table;
+
+		flood_table = &fid_family->flood_tables[i];
+		mlxsw_sp_fid_flood_table_fini(fid_family, flood_table);
+	}
+}
+
+static int mlxsw_sp_fid_family_register(struct mlxsw_sp *mlxsw_sp,
+					const struct mlxsw_sp_fid_family *tmpl)
+{
+	u16 nr_fids = tmpl->end_index - tmpl->start_index + 1;
+	struct mlxsw_sp_fid_family *fid_family;
+	int err;
+
+	fid_family = kmemdup(tmpl, sizeof(*fid_family), GFP_KERNEL);
+	if (!fid_family)
+		return -ENOMEM;
+
+	fid_family->mlxsw_sp = mlxsw_sp;
+	INIT_LIST_HEAD(&fid_family->fids_list);
+	fid_family->fids_bitmap = bitmap_zalloc(nr_fids, GFP_KERNEL);
+	if (!fid_family->fids_bitmap) {
+		err = -ENOMEM;
+		goto err_alloc_fids_bitmap;
+	}
+
+	if (fid_family->flood_tables) {
+		err = mlxsw_sp_fid_flood_tables_init(fid_family);
+		if (err)
+			goto err_fid_flood_tables_init;
+	}
+
+	mlxsw_sp->fid_core->fid_family_arr[tmpl->type] = fid_family;
+
+	return 0;
+
+err_fid_flood_tables_init:
+	bitmap_free(fid_family->fids_bitmap);
+err_alloc_fids_bitmap:
+	kfree(fid_family);
+	return err;
+}
+
+static void
+mlxsw_sp_fid_family_unregister(struct mlxsw_sp *mlxsw_sp,
+			       struct mlxsw_sp_fid_family *fid_family)
+{
+	mlxsw_sp->fid_core->fid_family_arr[fid_family->type] = NULL;
+
+	if (fid_family->flood_tables)
+		mlxsw_sp_fid_flood_tables_fini(fid_family);
+
+	bitmap_free(fid_family->fids_bitmap);
+	WARN_ON_ONCE(!list_empty(&fid_family->fids_list));
+	kfree(fid_family);
+}
+
+int mlxsw_sp_port_fids_init(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+
+	/* Track number of FIDs configured on the port with mapping type
+	 * PORT_VID_TO_FID, so that we know when to transition the port
+	 * back to non-virtual (VLAN) mode.
+	 */
+	mlxsw_sp->fid_core->port_fid_mappings[mlxsw_sp_port->local_port] = 0;
+
+	return mlxsw_sp_port_vp_mode_set(mlxsw_sp_port, false);
+}
+
+void mlxsw_sp_port_fids_fini(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+
+	mlxsw_sp->fid_core->port_fid_mappings[mlxsw_sp_port->local_port] = 0;
+}
+
+int mlxsw_sp_fids_init(struct mlxsw_sp *mlxsw_sp)
+{
+	unsigned int max_ports = mlxsw_core_max_ports(mlxsw_sp->core);
+	struct mlxsw_sp_fid_core *fid_core;
+	int err, i;
+
+	fid_core = kzalloc(sizeof(*mlxsw_sp->fid_core), GFP_KERNEL);
+	if (!fid_core)
+		return -ENOMEM;
+	mlxsw_sp->fid_core = fid_core;
+
+	err = rhashtable_init(&fid_core->fid_ht, &mlxsw_sp_fid_ht_params);
+	if (err)
+		goto err_rhashtable_fid_init;
+
+	err = rhashtable_init(&fid_core->vni_ht, &mlxsw_sp_fid_vni_ht_params);
+	if (err)
+		goto err_rhashtable_vni_init;
+
+	fid_core->port_fid_mappings = kcalloc(max_ports, sizeof(unsigned int),
+					      GFP_KERNEL);
+	if (!fid_core->port_fid_mappings) {
+		err = -ENOMEM;
+		goto err_alloc_port_fid_mappings;
+	}
+
+	for (i = 0; i < MLXSW_SP_FID_TYPE_MAX; i++) {
+		err = mlxsw_sp_fid_family_register(mlxsw_sp,
+						   mlxsw_sp->fid_family_arr[i]);
+
+		if (err)
+			goto err_fid_ops_register;
+	}
+
+	return 0;
+
+err_fid_ops_register:
+	for (i--; i >= 0; i--) {
+		struct mlxsw_sp_fid_family *fid_family;
+
+		fid_family = fid_core->fid_family_arr[i];
+		mlxsw_sp_fid_family_unregister(mlxsw_sp, fid_family);
+	}
+	kfree(fid_core->port_fid_mappings);
+err_alloc_port_fid_mappings:
+	rhashtable_destroy(&fid_core->vni_ht);
+err_rhashtable_vni_init:
+	rhashtable_destroy(&fid_core->fid_ht);
+err_rhashtable_fid_init:
+	kfree(fid_core);
+	return err;
+}
+
+void mlxsw_sp_fids_fini(struct mlxsw_sp *mlxsw_sp)
+{
+	struct mlxsw_sp_fid_core *fid_core = mlxsw_sp->fid_core;
+	int i;
+
+	for (i = 0; i < MLXSW_SP_FID_TYPE_MAX; i++)
+		mlxsw_sp_fid_family_unregister(mlxsw_sp,
+					       fid_core->fid_family_arr[i]);
+	kfree(fid_core->port_fid_mappings);
+	rhashtable_destroy(&fid_core->vni_ht);
+	rhashtable_destroy(&fid_core->fid_ht);
+	kfree(fid_core);
+}
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_flow.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_flow.c
new file mode 100644
index 000000000..9e50c823a
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_flow.c
@@ -0,0 +1,295 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2017-2020 Mellanox Technologies. All rights reserved */
+
+#include <linux/kernel.h>
+#include <linux/slab.h>
+#include <linux/errno.h>
+#include <linux/list.h>
+#include <net/net_namespace.h>
+
+#include "spectrum.h"
+
+struct mlxsw_sp_flow_block *
+mlxsw_sp_flow_block_create(struct mlxsw_sp *mlxsw_sp, struct net *net)
+{
+	struct mlxsw_sp_flow_block *block;
+
+	block = kzalloc(sizeof(*block), GFP_KERNEL);
+	if (!block)
+		return NULL;
+	INIT_LIST_HEAD(&block->binding_list);
+	INIT_LIST_HEAD(&block->mall.list);
+	block->mlxsw_sp = mlxsw_sp;
+	block->net = net;
+	return block;
+}
+
+void mlxsw_sp_flow_block_destroy(struct mlxsw_sp_flow_block *block)
+{
+	WARN_ON(!list_empty(&block->binding_list));
+	kfree(block);
+}
+
+static struct mlxsw_sp_flow_block_binding *
+mlxsw_sp_flow_block_lookup(struct mlxsw_sp_flow_block *block,
+			   struct mlxsw_sp_port *mlxsw_sp_port, bool ingress)
+{
+	struct mlxsw_sp_flow_block_binding *binding;
+
+	list_for_each_entry(binding, &block->binding_list, list)
+		if (binding->mlxsw_sp_port == mlxsw_sp_port &&
+		    binding->ingress == ingress)
+			return binding;
+	return NULL;
+}
+
+static bool
+mlxsw_sp_flow_block_ruleset_bound(const struct mlxsw_sp_flow_block *block)
+{
+	return block->ruleset_zero;
+}
+
+static int mlxsw_sp_flow_block_bind(struct mlxsw_sp *mlxsw_sp,
+				    struct mlxsw_sp_flow_block *block,
+				    struct mlxsw_sp_port *mlxsw_sp_port,
+				    bool ingress,
+				    struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp_flow_block_binding *binding;
+	int err;
+
+	if (WARN_ON(mlxsw_sp_flow_block_lookup(block, mlxsw_sp_port, ingress)))
+		return -EEXIST;
+
+	if (ingress && block->ingress_blocker_rule_count) {
+		NL_SET_ERR_MSG_MOD(extack, "Block cannot be bound to ingress because it contains unsupported rules");
+		return -EOPNOTSUPP;
+	}
+
+	if (!ingress && block->egress_blocker_rule_count) {
+		NL_SET_ERR_MSG_MOD(extack, "Block cannot be bound to egress because it contains unsupported rules");
+		return -EOPNOTSUPP;
+	}
+
+	err = mlxsw_sp_mall_port_bind(block, mlxsw_sp_port, extack);
+	if (err)
+		return err;
+
+	binding = kzalloc(sizeof(*binding), GFP_KERNEL);
+	if (!binding) {
+		err = -ENOMEM;
+		goto err_binding_alloc;
+	}
+	binding->mlxsw_sp_port = mlxsw_sp_port;
+	binding->ingress = ingress;
+
+	if (mlxsw_sp_flow_block_ruleset_bound(block)) {
+		err = mlxsw_sp_acl_ruleset_bind(mlxsw_sp, block, binding);
+		if (err)
+			goto err_ruleset_bind;
+	}
+
+	if (ingress)
+		block->ingress_binding_count++;
+	else
+		block->egress_binding_count++;
+	list_add(&binding->list, &block->binding_list);
+	return 0;
+
+err_ruleset_bind:
+	kfree(binding);
+err_binding_alloc:
+	mlxsw_sp_mall_port_unbind(block, mlxsw_sp_port);
+
+	return err;
+}
+
+static int mlxsw_sp_flow_block_unbind(struct mlxsw_sp *mlxsw_sp,
+				      struct mlxsw_sp_flow_block *block,
+				      struct mlxsw_sp_port *mlxsw_sp_port,
+				      bool ingress)
+{
+	struct mlxsw_sp_flow_block_binding *binding;
+
+	binding = mlxsw_sp_flow_block_lookup(block, mlxsw_sp_port, ingress);
+	if (!binding)
+		return -ENOENT;
+
+	list_del(&binding->list);
+
+	if (ingress)
+		block->ingress_binding_count--;
+	else
+		block->egress_binding_count--;
+
+	if (mlxsw_sp_flow_block_ruleset_bound(block))
+		mlxsw_sp_acl_ruleset_unbind(mlxsw_sp, block, binding);
+
+	kfree(binding);
+
+	mlxsw_sp_mall_port_unbind(block, mlxsw_sp_port);
+
+	return 0;
+}
+
+static int mlxsw_sp_flow_block_mall_cb(struct mlxsw_sp_flow_block *flow_block,
+				       struct tc_cls_matchall_offload *f)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_flow_block_mlxsw_sp(flow_block);
+
+	switch (f->command) {
+	case TC_CLSMATCHALL_REPLACE:
+		return mlxsw_sp_mall_replace(mlxsw_sp, flow_block, f);
+	case TC_CLSMATCHALL_DESTROY:
+		mlxsw_sp_mall_destroy(flow_block, f);
+		return 0;
+	default:
+		return -EOPNOTSUPP;
+	}
+}
+
+static int mlxsw_sp_flow_block_flower_cb(struct mlxsw_sp_flow_block *flow_block,
+					 struct flow_cls_offload *f)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_flow_block_mlxsw_sp(flow_block);
+
+	switch (f->command) {
+	case FLOW_CLS_REPLACE:
+		return mlxsw_sp_flower_replace(mlxsw_sp, flow_block, f);
+	case FLOW_CLS_DESTROY:
+		mlxsw_sp_flower_destroy(mlxsw_sp, flow_block, f);
+		return 0;
+	case FLOW_CLS_STATS:
+		return mlxsw_sp_flower_stats(mlxsw_sp, flow_block, f);
+	case FLOW_CLS_TMPLT_CREATE:
+		return mlxsw_sp_flower_tmplt_create(mlxsw_sp, flow_block, f);
+	case FLOW_CLS_TMPLT_DESTROY:
+		mlxsw_sp_flower_tmplt_destroy(mlxsw_sp, flow_block, f);
+		return 0;
+	default:
+		return -EOPNOTSUPP;
+	}
+}
+
+static int mlxsw_sp_flow_block_cb(enum tc_setup_type type,
+				  void *type_data, void *cb_priv)
+{
+	struct mlxsw_sp_flow_block *flow_block = cb_priv;
+
+	if (mlxsw_sp_flow_block_disabled(flow_block))
+		return -EOPNOTSUPP;
+
+	switch (type) {
+	case TC_SETUP_CLSMATCHALL:
+		return mlxsw_sp_flow_block_mall_cb(flow_block, type_data);
+	case TC_SETUP_CLSFLOWER:
+		return mlxsw_sp_flow_block_flower_cb(flow_block, type_data);
+	default:
+		return -EOPNOTSUPP;
+	}
+}
+
+static void mlxsw_sp_tc_block_release(void *cb_priv)
+{
+	struct mlxsw_sp_flow_block *flow_block = cb_priv;
+
+	mlxsw_sp_flow_block_destroy(flow_block);
+}
+
+static LIST_HEAD(mlxsw_sp_block_cb_list);
+
+static int mlxsw_sp_setup_tc_block_bind(struct mlxsw_sp_port *mlxsw_sp_port,
+					struct flow_block_offload *f,
+					bool ingress)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	struct mlxsw_sp_flow_block *flow_block;
+	struct flow_block_cb *block_cb;
+	bool register_block = false;
+	int err;
+
+	block_cb = flow_block_cb_lookup(f->block, mlxsw_sp_flow_block_cb,
+					mlxsw_sp);
+	if (!block_cb) {
+		flow_block = mlxsw_sp_flow_block_create(mlxsw_sp, f->net);
+		if (!flow_block)
+			return -ENOMEM;
+		block_cb = flow_block_cb_alloc(mlxsw_sp_flow_block_cb,
+					       mlxsw_sp, flow_block,
+					       mlxsw_sp_tc_block_release);
+		if (IS_ERR(block_cb)) {
+			mlxsw_sp_flow_block_destroy(flow_block);
+			return PTR_ERR(block_cb);
+		}
+		register_block = true;
+	} else {
+		flow_block = flow_block_cb_priv(block_cb);
+	}
+	flow_block_cb_incref(block_cb);
+	err = mlxsw_sp_flow_block_bind(mlxsw_sp, flow_block,
+				       mlxsw_sp_port, ingress, f->extack);
+	if (err)
+		goto err_block_bind;
+
+	if (ingress)
+		mlxsw_sp_port->ing_flow_block = flow_block;
+	else
+		mlxsw_sp_port->eg_flow_block = flow_block;
+
+	if (register_block) {
+		flow_block_cb_add(block_cb, f);
+		list_add_tail(&block_cb->driver_list, &mlxsw_sp_block_cb_list);
+	}
+
+	return 0;
+
+err_block_bind:
+	if (!flow_block_cb_decref(block_cb))
+		flow_block_cb_free(block_cb);
+	return err;
+}
+
+static void mlxsw_sp_setup_tc_block_unbind(struct mlxsw_sp_port *mlxsw_sp_port,
+					   struct flow_block_offload *f,
+					   bool ingress)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	struct mlxsw_sp_flow_block *flow_block;
+	struct flow_block_cb *block_cb;
+	int err;
+
+	block_cb = flow_block_cb_lookup(f->block, mlxsw_sp_flow_block_cb,
+					mlxsw_sp);
+	if (!block_cb)
+		return;
+
+	if (ingress)
+		mlxsw_sp_port->ing_flow_block = NULL;
+	else
+		mlxsw_sp_port->eg_flow_block = NULL;
+
+	flow_block = flow_block_cb_priv(block_cb);
+	err = mlxsw_sp_flow_block_unbind(mlxsw_sp, flow_block,
+					 mlxsw_sp_port, ingress);
+	if (!err && !flow_block_cb_decref(block_cb)) {
+		flow_block_cb_remove(block_cb, f);
+		list_del(&block_cb->driver_list);
+	}
+}
+
+int mlxsw_sp_setup_tc_block_clsact(struct mlxsw_sp_port *mlxsw_sp_port,
+				   struct flow_block_offload *f,
+				   bool ingress)
+{
+	f->driver_block_list = &mlxsw_sp_block_cb_list;
+
+	switch (f->command) {
+	case FLOW_BLOCK_BIND:
+		return mlxsw_sp_setup_tc_block_bind(mlxsw_sp_port, f, ingress);
+	case FLOW_BLOCK_UNBIND:
+		mlxsw_sp_setup_tc_block_unbind(mlxsw_sp_port, f, ingress);
+		return 0;
+	default:
+		return -EOPNOTSUPP;
+	}
+}
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_flower.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_flower.c
new file mode 100644
index 000000000..e91fb205e
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_flower.c
@@ -0,0 +1,796 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2017-2018 Mellanox Technologies. All rights reserved */
+
+#include <linux/kernel.h>
+#include <linux/errno.h>
+#include <linux/netdevice.h>
+#include <linux/log2.h>
+#include <net/net_namespace.h>
+#include <net/flow_dissector.h>
+#include <net/pkt_cls.h>
+#include <net/tc_act/tc_gact.h>
+#include <net/tc_act/tc_mirred.h>
+#include <net/tc_act/tc_vlan.h>
+
+#include "spectrum.h"
+#include "core_acl_flex_keys.h"
+
+static int mlxsw_sp_policer_validate(const struct flow_action *action,
+				     const struct flow_action_entry *act,
+				     struct netlink_ext_ack *extack)
+{
+	if (act->police.exceed.act_id != FLOW_ACTION_DROP) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "Offload not supported when exceed action is not drop");
+		return -EOPNOTSUPP;
+	}
+
+	if (act->police.notexceed.act_id != FLOW_ACTION_PIPE &&
+	    act->police.notexceed.act_id != FLOW_ACTION_ACCEPT) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "Offload not supported when conform action is not pipe or ok");
+		return -EOPNOTSUPP;
+	}
+
+	if (act->police.notexceed.act_id == FLOW_ACTION_ACCEPT &&
+	    !flow_action_is_last_entry(action, act)) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "Offload not supported when conform action is ok, but action is not last");
+		return -EOPNOTSUPP;
+	}
+
+	if (act->police.peakrate_bytes_ps ||
+	    act->police.avrate || act->police.overhead) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "Offload not supported when peakrate/avrate/overhead is configured");
+		return -EOPNOTSUPP;
+	}
+
+	if (act->police.rate_pkt_ps) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "QoS offload not support packets per second");
+		return -EOPNOTSUPP;
+	}
+
+	return 0;
+}
+
+static int mlxsw_sp_flower_parse_actions(struct mlxsw_sp *mlxsw_sp,
+					 struct mlxsw_sp_flow_block *block,
+					 struct mlxsw_sp_acl_rule_info *rulei,
+					 struct flow_action *flow_action,
+					 struct netlink_ext_ack *extack)
+{
+	const struct flow_action_entry *act;
+	int mirror_act_count = 0;
+	int police_act_count = 0;
+	int sample_act_count = 0;
+	int err, i;
+
+	if (!flow_action_has_entries(flow_action))
+		return 0;
+	if (!flow_action_mixed_hw_stats_check(flow_action, extack))
+		return -EOPNOTSUPP;
+
+	act = flow_action_first_entry_get(flow_action);
+	if (act->hw_stats & FLOW_ACTION_HW_STATS_DISABLED) {
+		/* Nothing to do */
+	} else if (act->hw_stats & FLOW_ACTION_HW_STATS_IMMEDIATE) {
+		/* Count action is inserted first */
+		err = mlxsw_sp_acl_rulei_act_count(mlxsw_sp, rulei, extack);
+		if (err)
+			return err;
+	} else {
+		NL_SET_ERR_MSG_MOD(extack, "Unsupported action HW stats type");
+		return -EOPNOTSUPP;
+	}
+
+	flow_action_for_each(i, act, flow_action) {
+		switch (act->id) {
+		case FLOW_ACTION_ACCEPT:
+			err = mlxsw_sp_acl_rulei_act_terminate(rulei);
+			if (err) {
+				NL_SET_ERR_MSG_MOD(extack, "Cannot append terminate action");
+				return err;
+			}
+			break;
+		case FLOW_ACTION_DROP: {
+			bool ingress;
+
+			if (mlxsw_sp_flow_block_is_mixed_bound(block)) {
+				NL_SET_ERR_MSG_MOD(extack, "Drop action is not supported when block is bound to ingress and egress");
+				return -EOPNOTSUPP;
+			}
+			ingress = mlxsw_sp_flow_block_is_ingress_bound(block);
+			err = mlxsw_sp_acl_rulei_act_drop(rulei, ingress,
+							  act->cookie, extack);
+			if (err) {
+				NL_SET_ERR_MSG_MOD(extack, "Cannot append drop action");
+				return err;
+			}
+
+			/* Forbid block with this rulei to be bound
+			 * to ingress/egress in future. Ingress rule is
+			 * a blocker for egress and vice versa.
+			 */
+			if (ingress)
+				rulei->egress_bind_blocker = 1;
+			else
+				rulei->ingress_bind_blocker = 1;
+			}
+			break;
+		case FLOW_ACTION_TRAP:
+			err = mlxsw_sp_acl_rulei_act_trap(rulei);
+			if (err) {
+				NL_SET_ERR_MSG_MOD(extack, "Cannot append trap action");
+				return err;
+			}
+			break;
+		case FLOW_ACTION_GOTO: {
+			u32 chain_index = act->chain_index;
+			struct mlxsw_sp_acl_ruleset *ruleset;
+			u16 group_id;
+
+			ruleset = mlxsw_sp_acl_ruleset_lookup(mlxsw_sp, block,
+							      chain_index,
+							      MLXSW_SP_ACL_PROFILE_FLOWER);
+			if (IS_ERR(ruleset))
+				return PTR_ERR(ruleset);
+
+			group_id = mlxsw_sp_acl_ruleset_group_id(ruleset);
+			err = mlxsw_sp_acl_rulei_act_jump(rulei, group_id);
+			if (err) {
+				NL_SET_ERR_MSG_MOD(extack, "Cannot append jump action");
+				return err;
+			}
+			}
+			break;
+		case FLOW_ACTION_REDIRECT: {
+			struct net_device *out_dev;
+			struct mlxsw_sp_fid *fid;
+			u16 fid_index;
+
+			if (mlxsw_sp_flow_block_is_egress_bound(block)) {
+				NL_SET_ERR_MSG_MOD(extack, "Redirect action is not supported on egress");
+				return -EOPNOTSUPP;
+			}
+
+			/* Forbid block with this rulei to be bound
+			 * to egress in future.
+			 */
+			rulei->egress_bind_blocker = 1;
+
+			fid = mlxsw_sp_acl_dummy_fid(mlxsw_sp);
+			fid_index = mlxsw_sp_fid_index(fid);
+			err = mlxsw_sp_acl_rulei_act_fid_set(mlxsw_sp, rulei,
+							     fid_index, extack);
+			if (err)
+				return err;
+
+			out_dev = act->dev;
+			err = mlxsw_sp_acl_rulei_act_fwd(mlxsw_sp, rulei,
+							 out_dev, extack);
+			if (err)
+				return err;
+			}
+			break;
+		case FLOW_ACTION_MIRRED: {
+			struct net_device *out_dev = act->dev;
+
+			if (mirror_act_count++) {
+				NL_SET_ERR_MSG_MOD(extack, "Multiple mirror actions per rule are not supported");
+				return -EOPNOTSUPP;
+			}
+
+			err = mlxsw_sp_acl_rulei_act_mirror(mlxsw_sp, rulei,
+							    block, out_dev,
+							    extack);
+			if (err)
+				return err;
+			}
+			break;
+		case FLOW_ACTION_VLAN_MANGLE: {
+			u16 proto = be16_to_cpu(act->vlan.proto);
+			u8 prio = act->vlan.prio;
+			u16 vid = act->vlan.vid;
+
+			err = mlxsw_sp_acl_rulei_act_vlan(mlxsw_sp, rulei,
+							  act->id, vid,
+							  proto, prio, extack);
+			if (err)
+				return err;
+			break;
+			}
+		case FLOW_ACTION_PRIORITY:
+			err = mlxsw_sp_acl_rulei_act_priority(mlxsw_sp, rulei,
+							      act->priority,
+							      extack);
+			if (err)
+				return err;
+			break;
+		case FLOW_ACTION_MANGLE: {
+			enum flow_action_mangle_base htype = act->mangle.htype;
+			__be32 be_mask = (__force __be32) act->mangle.mask;
+			__be32 be_val = (__force __be32) act->mangle.val;
+			u32 offset = act->mangle.offset;
+			u32 mask = be32_to_cpu(be_mask);
+			u32 val = be32_to_cpu(be_val);
+
+			err = mlxsw_sp_acl_rulei_act_mangle(mlxsw_sp, rulei,
+							    htype, offset,
+							    mask, val, extack);
+			if (err)
+				return err;
+			break;
+			}
+		case FLOW_ACTION_POLICE: {
+			u32 burst;
+
+			if (police_act_count++) {
+				NL_SET_ERR_MSG_MOD(extack, "Multiple police actions per rule are not supported");
+				return -EOPNOTSUPP;
+			}
+
+			err = mlxsw_sp_policer_validate(flow_action, act, extack);
+			if (err)
+				return err;
+
+			/* The kernel might adjust the requested burst size so
+			 * that it is not exactly a power of two. Re-adjust it
+			 * here since the hardware only supports burst sizes
+			 * that are a power of two.
+			 */
+			burst = roundup_pow_of_two(act->police.burst);
+			err = mlxsw_sp_acl_rulei_act_police(mlxsw_sp, rulei,
+							    act->hw_index,
+							    act->police.rate_bytes_ps,
+							    burst, extack);
+			if (err)
+				return err;
+			break;
+			}
+		case FLOW_ACTION_SAMPLE: {
+			if (sample_act_count++) {
+				NL_SET_ERR_MSG_MOD(extack, "Multiple sample actions per rule are not supported");
+				return -EOPNOTSUPP;
+			}
+
+			err = mlxsw_sp_acl_rulei_act_sample(mlxsw_sp, rulei,
+							    block,
+							    act->sample.psample_group,
+							    act->sample.rate,
+							    act->sample.trunc_size,
+							    act->sample.truncate,
+							    extack);
+			if (err)
+				return err;
+			break;
+			}
+		default:
+			NL_SET_ERR_MSG_MOD(extack, "Unsupported action");
+			dev_err(mlxsw_sp->bus_info->dev, "Unsupported action\n");
+			return -EOPNOTSUPP;
+		}
+	}
+
+	if (rulei->ipv6_valid) {
+		NL_SET_ERR_MSG_MOD(extack, "Unsupported mangle field");
+		return -EOPNOTSUPP;
+	}
+
+	return 0;
+}
+
+static int mlxsw_sp_flower_parse_meta(struct mlxsw_sp_acl_rule_info *rulei,
+				      struct flow_cls_offload *f,
+				      struct mlxsw_sp_flow_block *block)
+{
+	struct flow_rule *rule = flow_cls_offload_flow_rule(f);
+	struct mlxsw_sp_port *mlxsw_sp_port;
+	struct net_device *ingress_dev;
+	struct flow_match_meta match;
+
+	if (!flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_META))
+		return 0;
+
+	flow_rule_match_meta(rule, &match);
+	if (match.mask->ingress_ifindex != 0xFFFFFFFF) {
+		NL_SET_ERR_MSG_MOD(f->common.extack, "Unsupported ingress ifindex mask");
+		return -EINVAL;
+	}
+
+	ingress_dev = __dev_get_by_index(block->net,
+					 match.key->ingress_ifindex);
+	if (!ingress_dev) {
+		NL_SET_ERR_MSG_MOD(f->common.extack, "Can't find specified ingress port to match on");
+		return -EINVAL;
+	}
+
+	if (!mlxsw_sp_port_dev_check(ingress_dev)) {
+		NL_SET_ERR_MSG_MOD(f->common.extack, "Can't match on non-mlxsw ingress port");
+		return -EINVAL;
+	}
+
+	mlxsw_sp_port = netdev_priv(ingress_dev);
+	if (mlxsw_sp_port->mlxsw_sp != block->mlxsw_sp) {
+		NL_SET_ERR_MSG_MOD(f->common.extack, "Can't match on a port from different device");
+		return -EINVAL;
+	}
+
+	mlxsw_sp_acl_rulei_keymask_u32(rulei,
+				       MLXSW_AFK_ELEMENT_SRC_SYS_PORT,
+				       mlxsw_sp_port->local_port,
+				       0xFFFFFFFF);
+	return 0;
+}
+
+static void mlxsw_sp_flower_parse_ipv4(struct mlxsw_sp_acl_rule_info *rulei,
+				       struct flow_cls_offload *f)
+{
+	struct flow_match_ipv4_addrs match;
+
+	flow_rule_match_ipv4_addrs(f->rule, &match);
+
+	mlxsw_sp_acl_rulei_keymask_buf(rulei, MLXSW_AFK_ELEMENT_SRC_IP_0_31,
+				       (char *) &match.key->src,
+				       (char *) &match.mask->src, 4);
+	mlxsw_sp_acl_rulei_keymask_buf(rulei, MLXSW_AFK_ELEMENT_DST_IP_0_31,
+				       (char *) &match.key->dst,
+				       (char *) &match.mask->dst, 4);
+}
+
+static void mlxsw_sp_flower_parse_ipv6(struct mlxsw_sp_acl_rule_info *rulei,
+				       struct flow_cls_offload *f)
+{
+	struct flow_match_ipv6_addrs match;
+
+	flow_rule_match_ipv6_addrs(f->rule, &match);
+
+	mlxsw_sp_acl_rulei_keymask_buf(rulei, MLXSW_AFK_ELEMENT_SRC_IP_96_127,
+				       &match.key->src.s6_addr[0x0],
+				       &match.mask->src.s6_addr[0x0], 4);
+	mlxsw_sp_acl_rulei_keymask_buf(rulei, MLXSW_AFK_ELEMENT_SRC_IP_64_95,
+				       &match.key->src.s6_addr[0x4],
+				       &match.mask->src.s6_addr[0x4], 4);
+	mlxsw_sp_acl_rulei_keymask_buf(rulei, MLXSW_AFK_ELEMENT_SRC_IP_32_63,
+				       &match.key->src.s6_addr[0x8],
+				       &match.mask->src.s6_addr[0x8], 4);
+	mlxsw_sp_acl_rulei_keymask_buf(rulei, MLXSW_AFK_ELEMENT_SRC_IP_0_31,
+				       &match.key->src.s6_addr[0xC],
+				       &match.mask->src.s6_addr[0xC], 4);
+	mlxsw_sp_acl_rulei_keymask_buf(rulei, MLXSW_AFK_ELEMENT_DST_IP_96_127,
+				       &match.key->dst.s6_addr[0x0],
+				       &match.mask->dst.s6_addr[0x0], 4);
+	mlxsw_sp_acl_rulei_keymask_buf(rulei, MLXSW_AFK_ELEMENT_DST_IP_64_95,
+				       &match.key->dst.s6_addr[0x4],
+				       &match.mask->dst.s6_addr[0x4], 4);
+	mlxsw_sp_acl_rulei_keymask_buf(rulei, MLXSW_AFK_ELEMENT_DST_IP_32_63,
+				       &match.key->dst.s6_addr[0x8],
+				       &match.mask->dst.s6_addr[0x8], 4);
+	mlxsw_sp_acl_rulei_keymask_buf(rulei, MLXSW_AFK_ELEMENT_DST_IP_0_31,
+				       &match.key->dst.s6_addr[0xC],
+				       &match.mask->dst.s6_addr[0xC], 4);
+}
+
+static int mlxsw_sp_flower_parse_ports(struct mlxsw_sp *mlxsw_sp,
+				       struct mlxsw_sp_acl_rule_info *rulei,
+				       struct flow_cls_offload *f,
+				       u8 ip_proto)
+{
+	const struct flow_rule *rule = flow_cls_offload_flow_rule(f);
+	struct flow_match_ports match;
+
+	if (!flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_PORTS))
+		return 0;
+
+	if (ip_proto != IPPROTO_TCP && ip_proto != IPPROTO_UDP) {
+		NL_SET_ERR_MSG_MOD(f->common.extack, "Only UDP and TCP keys are supported");
+		dev_err(mlxsw_sp->bus_info->dev, "Only UDP and TCP keys are supported\n");
+		return -EINVAL;
+	}
+
+	flow_rule_match_ports(rule, &match);
+	mlxsw_sp_acl_rulei_keymask_u32(rulei, MLXSW_AFK_ELEMENT_DST_L4_PORT,
+				       ntohs(match.key->dst),
+				       ntohs(match.mask->dst));
+	mlxsw_sp_acl_rulei_keymask_u32(rulei, MLXSW_AFK_ELEMENT_SRC_L4_PORT,
+				       ntohs(match.key->src),
+				       ntohs(match.mask->src));
+	return 0;
+}
+
+static int mlxsw_sp_flower_parse_tcp(struct mlxsw_sp *mlxsw_sp,
+				     struct mlxsw_sp_acl_rule_info *rulei,
+				     struct flow_cls_offload *f,
+				     u8 ip_proto)
+{
+	const struct flow_rule *rule = flow_cls_offload_flow_rule(f);
+	struct flow_match_tcp match;
+
+	if (!flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_TCP))
+		return 0;
+
+	if (ip_proto != IPPROTO_TCP) {
+		NL_SET_ERR_MSG_MOD(f->common.extack, "TCP keys supported only for TCP");
+		dev_err(mlxsw_sp->bus_info->dev, "TCP keys supported only for TCP\n");
+		return -EINVAL;
+	}
+
+	flow_rule_match_tcp(rule, &match);
+
+	if (match.mask->flags & htons(0x0E00)) {
+		NL_SET_ERR_MSG_MOD(f->common.extack, "TCP flags match not supported on reserved bits");
+		dev_err(mlxsw_sp->bus_info->dev, "TCP flags match not supported on reserved bits\n");
+		return -EINVAL;
+	}
+
+	mlxsw_sp_acl_rulei_keymask_u32(rulei, MLXSW_AFK_ELEMENT_TCP_FLAGS,
+				       ntohs(match.key->flags),
+				       ntohs(match.mask->flags));
+	return 0;
+}
+
+static int mlxsw_sp_flower_parse_ip(struct mlxsw_sp *mlxsw_sp,
+				    struct mlxsw_sp_acl_rule_info *rulei,
+				    struct flow_cls_offload *f,
+				    u16 n_proto)
+{
+	const struct flow_rule *rule = flow_cls_offload_flow_rule(f);
+	struct flow_match_ip match;
+
+	if (!flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_IP))
+		return 0;
+
+	if (n_proto != ETH_P_IP && n_proto != ETH_P_IPV6) {
+		NL_SET_ERR_MSG_MOD(f->common.extack, "IP keys supported only for IPv4/6");
+		dev_err(mlxsw_sp->bus_info->dev, "IP keys supported only for IPv4/6\n");
+		return -EINVAL;
+	}
+
+	flow_rule_match_ip(rule, &match);
+
+	mlxsw_sp_acl_rulei_keymask_u32(rulei, MLXSW_AFK_ELEMENT_IP_TTL_,
+				       match.key->ttl, match.mask->ttl);
+
+	mlxsw_sp_acl_rulei_keymask_u32(rulei, MLXSW_AFK_ELEMENT_IP_ECN,
+				       match.key->tos & 0x3,
+				       match.mask->tos & 0x3);
+
+	mlxsw_sp_acl_rulei_keymask_u32(rulei, MLXSW_AFK_ELEMENT_IP_DSCP,
+				       match.key->tos >> 2,
+				       match.mask->tos >> 2);
+
+	return 0;
+}
+
+static int mlxsw_sp_flower_parse(struct mlxsw_sp *mlxsw_sp,
+				 struct mlxsw_sp_flow_block *block,
+				 struct mlxsw_sp_acl_rule_info *rulei,
+				 struct flow_cls_offload *f)
+{
+	struct flow_rule *rule = flow_cls_offload_flow_rule(f);
+	struct flow_dissector *dissector = rule->match.dissector;
+	u16 n_proto_mask = 0;
+	u16 n_proto_key = 0;
+	u16 addr_type = 0;
+	u8 ip_proto = 0;
+	int err;
+
+	if (dissector->used_keys &
+	    ~(BIT(FLOW_DISSECTOR_KEY_META) |
+	      BIT(FLOW_DISSECTOR_KEY_CONTROL) |
+	      BIT(FLOW_DISSECTOR_KEY_BASIC) |
+	      BIT(FLOW_DISSECTOR_KEY_ETH_ADDRS) |
+	      BIT(FLOW_DISSECTOR_KEY_IPV4_ADDRS) |
+	      BIT(FLOW_DISSECTOR_KEY_IPV6_ADDRS) |
+	      BIT(FLOW_DISSECTOR_KEY_PORTS) |
+	      BIT(FLOW_DISSECTOR_KEY_TCP) |
+	      BIT(FLOW_DISSECTOR_KEY_IP) |
+	      BIT(FLOW_DISSECTOR_KEY_VLAN))) {
+		dev_err(mlxsw_sp->bus_info->dev, "Unsupported key\n");
+		NL_SET_ERR_MSG_MOD(f->common.extack, "Unsupported key");
+		return -EOPNOTSUPP;
+	}
+
+	mlxsw_sp_acl_rulei_priority(rulei, f->common.prio);
+
+	err = mlxsw_sp_flower_parse_meta(rulei, f, block);
+	if (err)
+		return err;
+
+	if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_CONTROL)) {
+		struct flow_match_control match;
+
+		flow_rule_match_control(rule, &match);
+		addr_type = match.key->addr_type;
+	}
+
+	if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_BASIC)) {
+		struct flow_match_basic match;
+
+		flow_rule_match_basic(rule, &match);
+		n_proto_key = ntohs(match.key->n_proto);
+		n_proto_mask = ntohs(match.mask->n_proto);
+
+		if (n_proto_key == ETH_P_ALL) {
+			n_proto_key = 0;
+			n_proto_mask = 0;
+		}
+		mlxsw_sp_acl_rulei_keymask_u32(rulei,
+					       MLXSW_AFK_ELEMENT_ETHERTYPE,
+					       n_proto_key, n_proto_mask);
+
+		ip_proto = match.key->ip_proto;
+		mlxsw_sp_acl_rulei_keymask_u32(rulei,
+					       MLXSW_AFK_ELEMENT_IP_PROTO,
+					       match.key->ip_proto,
+					       match.mask->ip_proto);
+	}
+
+	if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_ETH_ADDRS)) {
+		struct flow_match_eth_addrs match;
+
+		flow_rule_match_eth_addrs(rule, &match);
+		mlxsw_sp_acl_rulei_keymask_buf(rulei,
+					       MLXSW_AFK_ELEMENT_DMAC_32_47,
+					       match.key->dst,
+					       match.mask->dst, 2);
+		mlxsw_sp_acl_rulei_keymask_buf(rulei,
+					       MLXSW_AFK_ELEMENT_DMAC_0_31,
+					       match.key->dst + 2,
+					       match.mask->dst + 2, 4);
+		mlxsw_sp_acl_rulei_keymask_buf(rulei,
+					       MLXSW_AFK_ELEMENT_SMAC_32_47,
+					       match.key->src,
+					       match.mask->src, 2);
+		mlxsw_sp_acl_rulei_keymask_buf(rulei,
+					       MLXSW_AFK_ELEMENT_SMAC_0_31,
+					       match.key->src + 2,
+					       match.mask->src + 2, 4);
+	}
+
+	if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_VLAN)) {
+		struct flow_match_vlan match;
+
+		flow_rule_match_vlan(rule, &match);
+		if (mlxsw_sp_flow_block_is_egress_bound(block) &&
+		    match.mask->vlan_id) {
+			NL_SET_ERR_MSG_MOD(f->common.extack, "vlan_id key is not supported on egress");
+			return -EOPNOTSUPP;
+		}
+
+		/* Forbid block with this rulei to be bound
+		 * to egress in future.
+		 */
+		rulei->egress_bind_blocker = 1;
+
+		if (match.mask->vlan_id != 0)
+			mlxsw_sp_acl_rulei_keymask_u32(rulei,
+						       MLXSW_AFK_ELEMENT_VID,
+						       match.key->vlan_id,
+						       match.mask->vlan_id);
+		if (match.mask->vlan_priority != 0)
+			mlxsw_sp_acl_rulei_keymask_u32(rulei,
+						       MLXSW_AFK_ELEMENT_PCP,
+						       match.key->vlan_priority,
+						       match.mask->vlan_priority);
+	}
+
+	if (addr_type == FLOW_DISSECTOR_KEY_IPV4_ADDRS)
+		mlxsw_sp_flower_parse_ipv4(rulei, f);
+
+	if (addr_type == FLOW_DISSECTOR_KEY_IPV6_ADDRS)
+		mlxsw_sp_flower_parse_ipv6(rulei, f);
+
+	err = mlxsw_sp_flower_parse_ports(mlxsw_sp, rulei, f, ip_proto);
+	if (err)
+		return err;
+	err = mlxsw_sp_flower_parse_tcp(mlxsw_sp, rulei, f, ip_proto);
+	if (err)
+		return err;
+
+	err = mlxsw_sp_flower_parse_ip(mlxsw_sp, rulei, f, n_proto_key & n_proto_mask);
+	if (err)
+		return err;
+
+	return mlxsw_sp_flower_parse_actions(mlxsw_sp, block, rulei,
+					     &f->rule->action,
+					     f->common.extack);
+}
+
+static int mlxsw_sp_flower_mall_prio_check(struct mlxsw_sp_flow_block *block,
+					   struct flow_cls_offload *f)
+{
+	bool ingress = mlxsw_sp_flow_block_is_ingress_bound(block);
+	unsigned int mall_min_prio;
+	unsigned int mall_max_prio;
+	int err;
+
+	err = mlxsw_sp_mall_prio_get(block, f->common.chain_index,
+				     &mall_min_prio, &mall_max_prio);
+	if (err) {
+		if (err == -ENOENT)
+			/* No matchall filters installed on this chain. */
+			return 0;
+		NL_SET_ERR_MSG(f->common.extack, "Failed to get matchall priorities");
+		return err;
+	}
+	if (ingress && f->common.prio <= mall_min_prio) {
+		NL_SET_ERR_MSG(f->common.extack, "Failed to add in front of existing matchall rules");
+		return -EOPNOTSUPP;
+	}
+	if (!ingress && f->common.prio >= mall_max_prio) {
+		NL_SET_ERR_MSG(f->common.extack, "Failed to add behind of existing matchall rules");
+		return -EOPNOTSUPP;
+	}
+	return 0;
+}
+
+int mlxsw_sp_flower_replace(struct mlxsw_sp *mlxsw_sp,
+			    struct mlxsw_sp_flow_block *block,
+			    struct flow_cls_offload *f)
+{
+	struct mlxsw_sp_acl_rule_info *rulei;
+	struct mlxsw_sp_acl_ruleset *ruleset;
+	struct mlxsw_sp_acl_rule *rule;
+	int err;
+
+	err = mlxsw_sp_flower_mall_prio_check(block, f);
+	if (err)
+		return err;
+
+	ruleset = mlxsw_sp_acl_ruleset_get(mlxsw_sp, block,
+					   f->common.chain_index,
+					   MLXSW_SP_ACL_PROFILE_FLOWER, NULL);
+	if (IS_ERR(ruleset))
+		return PTR_ERR(ruleset);
+
+	rule = mlxsw_sp_acl_rule_create(mlxsw_sp, ruleset, f->cookie, NULL,
+					f->common.extack);
+	if (IS_ERR(rule)) {
+		err = PTR_ERR(rule);
+		goto err_rule_create;
+	}
+
+	rulei = mlxsw_sp_acl_rule_rulei(rule);
+	err = mlxsw_sp_flower_parse(mlxsw_sp, block, rulei, f);
+	if (err)
+		goto err_flower_parse;
+
+	err = mlxsw_sp_acl_rulei_commit(rulei);
+	if (err)
+		goto err_rulei_commit;
+
+	err = mlxsw_sp_acl_rule_add(mlxsw_sp, rule);
+	if (err)
+		goto err_rule_add;
+
+	mlxsw_sp_acl_ruleset_put(mlxsw_sp, ruleset);
+	return 0;
+
+err_rule_add:
+err_rulei_commit:
+err_flower_parse:
+	mlxsw_sp_acl_rule_destroy(mlxsw_sp, rule);
+err_rule_create:
+	mlxsw_sp_acl_ruleset_put(mlxsw_sp, ruleset);
+	return err;
+}
+
+void mlxsw_sp_flower_destroy(struct mlxsw_sp *mlxsw_sp,
+			     struct mlxsw_sp_flow_block *block,
+			     struct flow_cls_offload *f)
+{
+	struct mlxsw_sp_acl_ruleset *ruleset;
+	struct mlxsw_sp_acl_rule *rule;
+
+	ruleset = mlxsw_sp_acl_ruleset_get(mlxsw_sp, block,
+					   f->common.chain_index,
+					   MLXSW_SP_ACL_PROFILE_FLOWER, NULL);
+	if (IS_ERR(ruleset))
+		return;
+
+	rule = mlxsw_sp_acl_rule_lookup(mlxsw_sp, ruleset, f->cookie);
+	if (rule) {
+		mlxsw_sp_acl_rule_del(mlxsw_sp, rule);
+		mlxsw_sp_acl_rule_destroy(mlxsw_sp, rule);
+	}
+
+	mlxsw_sp_acl_ruleset_put(mlxsw_sp, ruleset);
+}
+
+int mlxsw_sp_flower_stats(struct mlxsw_sp *mlxsw_sp,
+			  struct mlxsw_sp_flow_block *block,
+			  struct flow_cls_offload *f)
+{
+	enum flow_action_hw_stats used_hw_stats = FLOW_ACTION_HW_STATS_DISABLED;
+	struct mlxsw_sp_acl_ruleset *ruleset;
+	struct mlxsw_sp_acl_rule *rule;
+	u64 packets;
+	u64 lastuse;
+	u64 bytes;
+	u64 drops;
+	int err;
+
+	ruleset = mlxsw_sp_acl_ruleset_get(mlxsw_sp, block,
+					   f->common.chain_index,
+					   MLXSW_SP_ACL_PROFILE_FLOWER, NULL);
+	if (WARN_ON(IS_ERR(ruleset)))
+		return -EINVAL;
+
+	rule = mlxsw_sp_acl_rule_lookup(mlxsw_sp, ruleset, f->cookie);
+	if (!rule)
+		return -EINVAL;
+
+	err = mlxsw_sp_acl_rule_get_stats(mlxsw_sp, rule, &packets, &bytes,
+					  &drops, &lastuse, &used_hw_stats);
+	if (err)
+		goto err_rule_get_stats;
+
+	flow_stats_update(&f->stats, bytes, packets, drops, lastuse,
+			  used_hw_stats);
+
+	mlxsw_sp_acl_ruleset_put(mlxsw_sp, ruleset);
+	return 0;
+
+err_rule_get_stats:
+	mlxsw_sp_acl_ruleset_put(mlxsw_sp, ruleset);
+	return err;
+}
+
+int mlxsw_sp_flower_tmplt_create(struct mlxsw_sp *mlxsw_sp,
+				 struct mlxsw_sp_flow_block *block,
+				 struct flow_cls_offload *f)
+{
+	struct mlxsw_sp_acl_ruleset *ruleset;
+	struct mlxsw_sp_acl_rule_info rulei;
+	int err;
+
+	memset(&rulei, 0, sizeof(rulei));
+	err = mlxsw_sp_flower_parse(mlxsw_sp, block, &rulei, f);
+	if (err)
+		return err;
+	ruleset = mlxsw_sp_acl_ruleset_get(mlxsw_sp, block,
+					   f->common.chain_index,
+					   MLXSW_SP_ACL_PROFILE_FLOWER,
+					   &rulei.values.elusage);
+
+	/* keep the reference to the ruleset */
+	return PTR_ERR_OR_ZERO(ruleset);
+}
+
+void mlxsw_sp_flower_tmplt_destroy(struct mlxsw_sp *mlxsw_sp,
+				   struct mlxsw_sp_flow_block *block,
+				   struct flow_cls_offload *f)
+{
+	struct mlxsw_sp_acl_ruleset *ruleset;
+
+	ruleset = mlxsw_sp_acl_ruleset_get(mlxsw_sp, block,
+					   f->common.chain_index,
+					   MLXSW_SP_ACL_PROFILE_FLOWER, NULL);
+	if (IS_ERR(ruleset))
+		return;
+	/* put the reference to the ruleset kept in create */
+	mlxsw_sp_acl_ruleset_put(mlxsw_sp, ruleset);
+	mlxsw_sp_acl_ruleset_put(mlxsw_sp, ruleset);
+}
+
+int mlxsw_sp_flower_prio_get(struct mlxsw_sp *mlxsw_sp,
+			     struct mlxsw_sp_flow_block *block,
+			     u32 chain_index, unsigned int *p_min_prio,
+			     unsigned int *p_max_prio)
+{
+	struct mlxsw_sp_acl_ruleset *ruleset;
+
+	ruleset = mlxsw_sp_acl_ruleset_lookup(mlxsw_sp, block,
+					      chain_index,
+					      MLXSW_SP_ACL_PROFILE_FLOWER);
+	if (IS_ERR(ruleset))
+		/* In case there are no flower rules, the caller
+		 * receives -ENOENT to indicate there is no need
+		 * to check the priorities.
+		 */
+		return PTR_ERR(ruleset);
+	mlxsw_sp_acl_ruleset_prio_get(ruleset, p_min_prio, p_max_prio);
+	return 0;
+}
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_ipip.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_ipip.c
new file mode 100644
index 000000000..a2ee695a3
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_ipip.c
@@ -0,0 +1,676 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2017-2018 Mellanox Technologies. All rights reserved */
+
+#include <net/ip_tunnels.h>
+#include <net/ip6_tunnel.h>
+#include <net/inet_ecn.h>
+
+#include "spectrum_ipip.h"
+#include "reg.h"
+
+struct ip_tunnel_parm
+mlxsw_sp_ipip_netdev_parms4(const struct net_device *ol_dev)
+{
+	struct ip_tunnel *tun = netdev_priv(ol_dev);
+
+	return tun->parms;
+}
+
+struct __ip6_tnl_parm
+mlxsw_sp_ipip_netdev_parms6(const struct net_device *ol_dev)
+{
+	struct ip6_tnl *tun = netdev_priv(ol_dev);
+
+	return tun->parms;
+}
+
+static bool mlxsw_sp_ipip_parms4_has_ikey(const struct ip_tunnel_parm *parms)
+{
+	return !!(parms->i_flags & TUNNEL_KEY);
+}
+
+static bool mlxsw_sp_ipip_parms6_has_ikey(const struct __ip6_tnl_parm *parms)
+{
+	return !!(parms->i_flags & TUNNEL_KEY);
+}
+
+static bool mlxsw_sp_ipip_parms4_has_okey(const struct ip_tunnel_parm *parms)
+{
+	return !!(parms->o_flags & TUNNEL_KEY);
+}
+
+static bool mlxsw_sp_ipip_parms6_has_okey(const struct __ip6_tnl_parm *parms)
+{
+	return !!(parms->o_flags & TUNNEL_KEY);
+}
+
+static u32 mlxsw_sp_ipip_parms4_ikey(const struct ip_tunnel_parm *parms)
+{
+	return mlxsw_sp_ipip_parms4_has_ikey(parms) ?
+		be32_to_cpu(parms->i_key) : 0;
+}
+
+static u32 mlxsw_sp_ipip_parms6_ikey(const struct __ip6_tnl_parm *parms)
+{
+	return mlxsw_sp_ipip_parms6_has_ikey(parms) ?
+		be32_to_cpu(parms->i_key) : 0;
+}
+
+static u32 mlxsw_sp_ipip_parms4_okey(const struct ip_tunnel_parm *parms)
+{
+	return mlxsw_sp_ipip_parms4_has_okey(parms) ?
+		be32_to_cpu(parms->o_key) : 0;
+}
+
+static u32 mlxsw_sp_ipip_parms6_okey(const struct __ip6_tnl_parm *parms)
+{
+	return mlxsw_sp_ipip_parms6_has_okey(parms) ?
+		be32_to_cpu(parms->o_key) : 0;
+}
+
+static union mlxsw_sp_l3addr
+mlxsw_sp_ipip_parms4_saddr(const struct ip_tunnel_parm *parms)
+{
+	return (union mlxsw_sp_l3addr) { .addr4 = parms->iph.saddr };
+}
+
+static union mlxsw_sp_l3addr
+mlxsw_sp_ipip_parms6_saddr(const struct __ip6_tnl_parm *parms)
+{
+	return (union mlxsw_sp_l3addr) { .addr6 = parms->laddr };
+}
+
+static union mlxsw_sp_l3addr
+mlxsw_sp_ipip_parms4_daddr(const struct ip_tunnel_parm *parms)
+{
+	return (union mlxsw_sp_l3addr) { .addr4 = parms->iph.daddr };
+}
+
+static union mlxsw_sp_l3addr
+mlxsw_sp_ipip_parms6_daddr(const struct __ip6_tnl_parm *parms)
+{
+	return (union mlxsw_sp_l3addr) { .addr6 = parms->raddr };
+}
+
+union mlxsw_sp_l3addr
+mlxsw_sp_ipip_netdev_saddr(enum mlxsw_sp_l3proto proto,
+			   const struct net_device *ol_dev)
+{
+	struct ip_tunnel_parm parms4;
+	struct __ip6_tnl_parm parms6;
+
+	switch (proto) {
+	case MLXSW_SP_L3_PROTO_IPV4:
+		parms4 = mlxsw_sp_ipip_netdev_parms4(ol_dev);
+		return mlxsw_sp_ipip_parms4_saddr(&parms4);
+	case MLXSW_SP_L3_PROTO_IPV6:
+		parms6 = mlxsw_sp_ipip_netdev_parms6(ol_dev);
+		return mlxsw_sp_ipip_parms6_saddr(&parms6);
+	}
+
+	WARN_ON(1);
+	return (union mlxsw_sp_l3addr) {0};
+}
+
+static __be32 mlxsw_sp_ipip_netdev_daddr4(const struct net_device *ol_dev)
+{
+
+	struct ip_tunnel_parm parms4 = mlxsw_sp_ipip_netdev_parms4(ol_dev);
+
+	return mlxsw_sp_ipip_parms4_daddr(&parms4).addr4;
+}
+
+static union mlxsw_sp_l3addr
+mlxsw_sp_ipip_netdev_daddr(enum mlxsw_sp_l3proto proto,
+			   const struct net_device *ol_dev)
+{
+	struct ip_tunnel_parm parms4;
+	struct __ip6_tnl_parm parms6;
+
+	switch (proto) {
+	case MLXSW_SP_L3_PROTO_IPV4:
+		parms4 = mlxsw_sp_ipip_netdev_parms4(ol_dev);
+		return mlxsw_sp_ipip_parms4_daddr(&parms4);
+	case MLXSW_SP_L3_PROTO_IPV6:
+		parms6 = mlxsw_sp_ipip_netdev_parms6(ol_dev);
+		return mlxsw_sp_ipip_parms6_daddr(&parms6);
+	}
+
+	WARN_ON(1);
+	return (union mlxsw_sp_l3addr) {0};
+}
+
+bool mlxsw_sp_l3addr_is_zero(union mlxsw_sp_l3addr addr)
+{
+	union mlxsw_sp_l3addr naddr = {0};
+
+	return !memcmp(&addr, &naddr, sizeof(naddr));
+}
+
+static struct mlxsw_sp_ipip_parms
+mlxsw_sp_ipip_netdev_parms_init_gre4(const struct net_device *ol_dev)
+{
+	struct ip_tunnel_parm parms = mlxsw_sp_ipip_netdev_parms4(ol_dev);
+
+	return (struct mlxsw_sp_ipip_parms) {
+		.proto = MLXSW_SP_L3_PROTO_IPV4,
+		.saddr = mlxsw_sp_ipip_parms4_saddr(&parms),
+		.daddr = mlxsw_sp_ipip_parms4_daddr(&parms),
+		.link = parms.link,
+		.ikey = mlxsw_sp_ipip_parms4_ikey(&parms),
+		.okey = mlxsw_sp_ipip_parms4_okey(&parms),
+	};
+}
+
+static int
+mlxsw_sp_ipip_nexthop_update_gre4(struct mlxsw_sp *mlxsw_sp, u32 adj_index,
+				  struct mlxsw_sp_ipip_entry *ipip_entry,
+				  bool force, char *ratr_pl)
+{
+	u16 rif_index = mlxsw_sp_ipip_lb_rif_index(ipip_entry->ol_lb);
+	__be32 daddr4 = mlxsw_sp_ipip_netdev_daddr4(ipip_entry->ol_dev);
+	enum mlxsw_reg_ratr_op op;
+
+	op = force ? MLXSW_REG_RATR_OP_WRITE_WRITE_ENTRY :
+		     MLXSW_REG_RATR_OP_WRITE_WRITE_ENTRY_ON_ACTIVITY;
+	mlxsw_reg_ratr_pack(ratr_pl, op, true, MLXSW_REG_RATR_TYPE_IPIP,
+			    adj_index, rif_index);
+	mlxsw_reg_ratr_ipip4_entry_pack(ratr_pl, be32_to_cpu(daddr4));
+
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(ratr), ratr_pl);
+}
+
+static int
+mlxsw_sp_ipip_decap_config_gre4(struct mlxsw_sp *mlxsw_sp,
+				struct mlxsw_sp_ipip_entry *ipip_entry,
+				u32 tunnel_index)
+{
+	u16 rif_index = mlxsw_sp_ipip_lb_rif_index(ipip_entry->ol_lb);
+	u16 ul_rif_id = mlxsw_sp_ipip_lb_ul_rif_id(ipip_entry->ol_lb);
+	char rtdp_pl[MLXSW_REG_RTDP_LEN];
+	struct ip_tunnel_parm parms;
+	unsigned int type_check;
+	bool has_ikey;
+	u32 daddr4;
+	u32 ikey;
+
+	parms = mlxsw_sp_ipip_netdev_parms4(ipip_entry->ol_dev);
+	has_ikey = mlxsw_sp_ipip_parms4_has_ikey(&parms);
+	ikey = mlxsw_sp_ipip_parms4_ikey(&parms);
+
+	mlxsw_reg_rtdp_pack(rtdp_pl, MLXSW_REG_RTDP_TYPE_IPIP, tunnel_index);
+	mlxsw_reg_rtdp_egress_router_interface_set(rtdp_pl, ul_rif_id);
+
+	type_check = has_ikey ?
+		MLXSW_REG_RTDP_IPIP_TYPE_CHECK_ALLOW_GRE_KEY :
+		MLXSW_REG_RTDP_IPIP_TYPE_CHECK_ALLOW_GRE;
+
+	/* Linux demuxes tunnels based on packet SIP (which must match tunnel
+	 * remote IP). Thus configure decap so that it filters out packets that
+	 * are not IPv4 or have the wrong SIP. IPIP_DECAP_ERROR trap is
+	 * generated for packets that fail this criterion. Linux then handles
+	 * such packets in slow path and generates ICMP destination unreachable.
+	 */
+	daddr4 = be32_to_cpu(mlxsw_sp_ipip_netdev_daddr4(ipip_entry->ol_dev));
+	mlxsw_reg_rtdp_ipip4_pack(rtdp_pl, rif_index,
+				  MLXSW_REG_RTDP_IPIP_SIP_CHECK_FILTER_IPV4,
+				  type_check, has_ikey, daddr4, ikey);
+
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(rtdp), rtdp_pl);
+}
+
+static bool mlxsw_sp_ipip_tunnel_complete(enum mlxsw_sp_l3proto proto,
+					  const struct net_device *ol_dev)
+{
+	union mlxsw_sp_l3addr saddr = mlxsw_sp_ipip_netdev_saddr(proto, ol_dev);
+	union mlxsw_sp_l3addr daddr = mlxsw_sp_ipip_netdev_daddr(proto, ol_dev);
+
+	/* Tunnels with unset local or remote address are valid in Linux and
+	 * used for lightweight tunnels (LWT) and Non-Broadcast Multi-Access
+	 * (NBMA) tunnels. In principle these can be offloaded, but the driver
+	 * currently doesn't support this. So punt.
+	 */
+	return !mlxsw_sp_l3addr_is_zero(saddr) &&
+	       !mlxsw_sp_l3addr_is_zero(daddr);
+}
+
+static bool mlxsw_sp_ipip_can_offload_gre4(const struct mlxsw_sp *mlxsw_sp,
+					   const struct net_device *ol_dev)
+{
+	struct ip_tunnel *tunnel = netdev_priv(ol_dev);
+	__be16 okflags = TUNNEL_KEY; /* We can't offload any other features. */
+	bool inherit_ttl = tunnel->parms.iph.ttl == 0;
+	bool inherit_tos = tunnel->parms.iph.tos & 0x1;
+
+	return (tunnel->parms.i_flags & ~okflags) == 0 &&
+	       (tunnel->parms.o_flags & ~okflags) == 0 &&
+	       inherit_ttl && inherit_tos &&
+	       mlxsw_sp_ipip_tunnel_complete(MLXSW_SP_L3_PROTO_IPV4, ol_dev);
+}
+
+static struct mlxsw_sp_rif_ipip_lb_config
+mlxsw_sp_ipip_ol_loopback_config_gre4(struct mlxsw_sp *mlxsw_sp,
+				      const struct net_device *ol_dev)
+{
+	struct ip_tunnel_parm parms = mlxsw_sp_ipip_netdev_parms4(ol_dev);
+	enum mlxsw_reg_ritr_loopback_ipip_type lb_ipipt;
+
+	lb_ipipt = mlxsw_sp_ipip_parms4_has_okey(&parms) ?
+		MLXSW_REG_RITR_LOOPBACK_IPIP_TYPE_IP_IN_GRE_KEY_IN_IP :
+		MLXSW_REG_RITR_LOOPBACK_IPIP_TYPE_IP_IN_GRE_IN_IP;
+	return (struct mlxsw_sp_rif_ipip_lb_config){
+		.lb_ipipt = lb_ipipt,
+		.okey = mlxsw_sp_ipip_parms4_okey(&parms),
+		.ul_protocol = MLXSW_SP_L3_PROTO_IPV4,
+		.saddr = mlxsw_sp_ipip_netdev_saddr(MLXSW_SP_L3_PROTO_IPV4,
+						    ol_dev),
+	};
+}
+
+static int
+mlxsw_sp_ipip_ol_netdev_change_gre(struct mlxsw_sp *mlxsw_sp,
+				   struct mlxsw_sp_ipip_entry *ipip_entry,
+				   const struct mlxsw_sp_ipip_parms *new_parms,
+				   struct netlink_ext_ack *extack)
+{
+	const struct mlxsw_sp_ipip_parms *old_parms = &ipip_entry->parms;
+	bool update_tunnel = false;
+	bool update_decap = false;
+	bool update_nhs = false;
+	int err = 0;
+
+	if (!mlxsw_sp_l3addr_eq(&new_parms->saddr, &old_parms->saddr)) {
+		u16 ul_tb_id = mlxsw_sp_ipip_dev_ul_tb_id(ipip_entry->ol_dev);
+
+		/* Since the local address has changed, if there is another
+		 * tunnel with a matching saddr, both need to be demoted.
+		 */
+		if (mlxsw_sp_ipip_demote_tunnel_by_saddr(mlxsw_sp,
+							 new_parms->proto,
+							 new_parms->saddr,
+							 ul_tb_id,
+							 ipip_entry)) {
+			mlxsw_sp_ipip_entry_demote_tunnel(mlxsw_sp, ipip_entry);
+			return 0;
+		}
+
+		update_tunnel = true;
+	} else if (old_parms->okey != new_parms->okey ||
+		   old_parms->link != new_parms->link) {
+		update_tunnel = true;
+	} else if (!mlxsw_sp_l3addr_eq(&new_parms->daddr, &old_parms->daddr)) {
+		update_nhs = true;
+	} else if (old_parms->ikey != new_parms->ikey) {
+		update_decap = true;
+	}
+
+	if (update_tunnel)
+		err = __mlxsw_sp_ipip_entry_update_tunnel(mlxsw_sp, ipip_entry,
+							  true, true, true,
+							  extack);
+	else if (update_nhs)
+		err = __mlxsw_sp_ipip_entry_update_tunnel(mlxsw_sp, ipip_entry,
+							  false, false, true,
+							  extack);
+	else if (update_decap)
+		err = __mlxsw_sp_ipip_entry_update_tunnel(mlxsw_sp, ipip_entry,
+							  false, false, false,
+							  extack);
+	if (err)
+		return err;
+
+	ipip_entry->parms = *new_parms;
+	return 0;
+}
+
+static int
+mlxsw_sp_ipip_ol_netdev_change_gre4(struct mlxsw_sp *mlxsw_sp,
+				    struct mlxsw_sp_ipip_entry *ipip_entry,
+				    struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp_ipip_parms new_parms;
+
+	new_parms = mlxsw_sp_ipip_netdev_parms_init_gre4(ipip_entry->ol_dev);
+	return mlxsw_sp_ipip_ol_netdev_change_gre(mlxsw_sp, ipip_entry,
+						  &new_parms, extack);
+}
+
+static int
+mlxsw_sp_ipip_rem_addr_set_gre4(struct mlxsw_sp *mlxsw_sp,
+				struct mlxsw_sp_ipip_entry *ipip_entry)
+{
+	return 0;
+}
+
+static void
+mlxsw_sp_ipip_rem_addr_unset_gre4(struct mlxsw_sp *mlxsw_sp,
+				  const struct mlxsw_sp_ipip_entry *ipip_entry)
+{
+}
+
+static const struct mlxsw_sp_ipip_ops mlxsw_sp_ipip_gre4_ops = {
+	.dev_type = ARPHRD_IPGRE,
+	.ul_proto = MLXSW_SP_L3_PROTO_IPV4,
+	.inc_parsing_depth = false,
+	.parms_init = mlxsw_sp_ipip_netdev_parms_init_gre4,
+	.nexthop_update = mlxsw_sp_ipip_nexthop_update_gre4,
+	.decap_config = mlxsw_sp_ipip_decap_config_gre4,
+	.can_offload = mlxsw_sp_ipip_can_offload_gre4,
+	.ol_loopback_config = mlxsw_sp_ipip_ol_loopback_config_gre4,
+	.ol_netdev_change = mlxsw_sp_ipip_ol_netdev_change_gre4,
+	.rem_ip_addr_set = mlxsw_sp_ipip_rem_addr_set_gre4,
+	.rem_ip_addr_unset = mlxsw_sp_ipip_rem_addr_unset_gre4,
+};
+
+static struct mlxsw_sp_ipip_parms
+mlxsw_sp1_ipip_netdev_parms_init_gre6(const struct net_device *ol_dev)
+{
+	struct mlxsw_sp_ipip_parms parms = {0};
+
+	WARN_ON_ONCE(1);
+	return parms;
+}
+
+static int
+mlxsw_sp1_ipip_nexthop_update_gre6(struct mlxsw_sp *mlxsw_sp, u32 adj_index,
+				   struct mlxsw_sp_ipip_entry *ipip_entry,
+				   bool force, char *ratr_pl)
+{
+	WARN_ON_ONCE(1);
+	return -EINVAL;
+}
+
+static int
+mlxsw_sp1_ipip_decap_config_gre6(struct mlxsw_sp *mlxsw_sp,
+				 struct mlxsw_sp_ipip_entry *ipip_entry,
+				 u32 tunnel_index)
+{
+	WARN_ON_ONCE(1);
+	return -EINVAL;
+}
+
+static bool mlxsw_sp1_ipip_can_offload_gre6(const struct mlxsw_sp *mlxsw_sp,
+					    const struct net_device *ol_dev)
+{
+	return false;
+}
+
+static struct mlxsw_sp_rif_ipip_lb_config
+mlxsw_sp1_ipip_ol_loopback_config_gre6(struct mlxsw_sp *mlxsw_sp,
+				       const struct net_device *ol_dev)
+{
+	struct mlxsw_sp_rif_ipip_lb_config config = {0};
+
+	WARN_ON_ONCE(1);
+	return config;
+}
+
+static int
+mlxsw_sp1_ipip_ol_netdev_change_gre6(struct mlxsw_sp *mlxsw_sp,
+				     struct mlxsw_sp_ipip_entry *ipip_entry,
+				     struct netlink_ext_ack *extack)
+{
+	WARN_ON_ONCE(1);
+	return -EINVAL;
+}
+
+static int
+mlxsw_sp1_ipip_rem_addr_set_gre6(struct mlxsw_sp *mlxsw_sp,
+				 struct mlxsw_sp_ipip_entry *ipip_entry)
+{
+	WARN_ON_ONCE(1);
+	return -EINVAL;
+}
+
+static void
+mlxsw_sp1_ipip_rem_addr_unset_gre6(struct mlxsw_sp *mlxsw_sp,
+				   const struct mlxsw_sp_ipip_entry *ipip_entry)
+{
+	WARN_ON_ONCE(1);
+}
+
+static const struct mlxsw_sp_ipip_ops mlxsw_sp1_ipip_gre6_ops = {
+	.dev_type = ARPHRD_IP6GRE,
+	.ul_proto = MLXSW_SP_L3_PROTO_IPV6,
+	.inc_parsing_depth = true,
+	.parms_init = mlxsw_sp1_ipip_netdev_parms_init_gre6,
+	.nexthop_update = mlxsw_sp1_ipip_nexthop_update_gre6,
+	.decap_config = mlxsw_sp1_ipip_decap_config_gre6,
+	.can_offload = mlxsw_sp1_ipip_can_offload_gre6,
+	.ol_loopback_config = mlxsw_sp1_ipip_ol_loopback_config_gre6,
+	.ol_netdev_change = mlxsw_sp1_ipip_ol_netdev_change_gre6,
+	.rem_ip_addr_set = mlxsw_sp1_ipip_rem_addr_set_gre6,
+	.rem_ip_addr_unset = mlxsw_sp1_ipip_rem_addr_unset_gre6,
+};
+
+const struct mlxsw_sp_ipip_ops *mlxsw_sp1_ipip_ops_arr[] = {
+	[MLXSW_SP_IPIP_TYPE_GRE4] = &mlxsw_sp_ipip_gre4_ops,
+	[MLXSW_SP_IPIP_TYPE_GRE6] = &mlxsw_sp1_ipip_gre6_ops,
+};
+
+static struct mlxsw_sp_ipip_parms
+mlxsw_sp2_ipip_netdev_parms_init_gre6(const struct net_device *ol_dev)
+{
+	struct __ip6_tnl_parm parms = mlxsw_sp_ipip_netdev_parms6(ol_dev);
+
+	return (struct mlxsw_sp_ipip_parms) {
+		.proto = MLXSW_SP_L3_PROTO_IPV6,
+		.saddr = mlxsw_sp_ipip_parms6_saddr(&parms),
+		.daddr = mlxsw_sp_ipip_parms6_daddr(&parms),
+		.link = parms.link,
+		.ikey = mlxsw_sp_ipip_parms6_ikey(&parms),
+		.okey = mlxsw_sp_ipip_parms6_okey(&parms),
+	};
+}
+
+static int
+mlxsw_sp2_ipip_nexthop_update_gre6(struct mlxsw_sp *mlxsw_sp, u32 adj_index,
+				   struct mlxsw_sp_ipip_entry *ipip_entry,
+				   bool force, char *ratr_pl)
+{
+	u16 rif_index = mlxsw_sp_ipip_lb_rif_index(ipip_entry->ol_lb);
+	enum mlxsw_reg_ratr_op op;
+
+	op = force ? MLXSW_REG_RATR_OP_WRITE_WRITE_ENTRY :
+		     MLXSW_REG_RATR_OP_WRITE_WRITE_ENTRY_ON_ACTIVITY;
+	mlxsw_reg_ratr_pack(ratr_pl, op, true, MLXSW_REG_RATR_TYPE_IPIP,
+			    adj_index, rif_index);
+	mlxsw_reg_ratr_ipip6_entry_pack(ratr_pl,
+					ipip_entry->dip_kvdl_index);
+
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(ratr), ratr_pl);
+}
+
+static int
+mlxsw_sp2_ipip_decap_config_gre6(struct mlxsw_sp *mlxsw_sp,
+				 struct mlxsw_sp_ipip_entry *ipip_entry,
+				 u32 tunnel_index)
+{
+	u16 rif_index = mlxsw_sp_ipip_lb_rif_index(ipip_entry->ol_lb);
+	u16 ul_rif_id = mlxsw_sp_ipip_lb_ul_rif_id(ipip_entry->ol_lb);
+	char rtdp_pl[MLXSW_REG_RTDP_LEN];
+	struct __ip6_tnl_parm parms;
+	unsigned int type_check;
+	bool has_ikey;
+	u32 ikey;
+
+	parms = mlxsw_sp_ipip_netdev_parms6(ipip_entry->ol_dev);
+	has_ikey = mlxsw_sp_ipip_parms6_has_ikey(&parms);
+	ikey = mlxsw_sp_ipip_parms6_ikey(&parms);
+
+	mlxsw_reg_rtdp_pack(rtdp_pl, MLXSW_REG_RTDP_TYPE_IPIP, tunnel_index);
+	mlxsw_reg_rtdp_egress_router_interface_set(rtdp_pl, ul_rif_id);
+
+	type_check = has_ikey ?
+		MLXSW_REG_RTDP_IPIP_TYPE_CHECK_ALLOW_GRE_KEY :
+		MLXSW_REG_RTDP_IPIP_TYPE_CHECK_ALLOW_GRE;
+
+	/* Linux demuxes tunnels based on packet SIP (which must match tunnel
+	 * remote IP). Thus configure decap so that it filters out packets that
+	 * are not IPv6 or have the wrong SIP. IPIP_DECAP_ERROR trap is
+	 * generated for packets that fail this criterion. Linux then handles
+	 * such packets in slow path and generates ICMP destination unreachable.
+	 */
+	mlxsw_reg_rtdp_ipip6_pack(rtdp_pl, rif_index,
+				  MLXSW_REG_RTDP_IPIP_SIP_CHECK_FILTER_IPV6,
+				  type_check, has_ikey,
+				  ipip_entry->dip_kvdl_index, ikey);
+
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(rtdp), rtdp_pl);
+}
+
+static bool mlxsw_sp2_ipip_can_offload_gre6(const struct mlxsw_sp *mlxsw_sp,
+					    const struct net_device *ol_dev)
+{
+	struct __ip6_tnl_parm tparm = mlxsw_sp_ipip_netdev_parms6(ol_dev);
+	bool inherit_tos = tparm.flags & IP6_TNL_F_USE_ORIG_TCLASS;
+	bool inherit_ttl = tparm.hop_limit == 0;
+	__be16 okflags = TUNNEL_KEY; /* We can't offload any other features. */
+
+	return (tparm.i_flags & ~okflags) == 0 &&
+	       (tparm.o_flags & ~okflags) == 0 &&
+	       inherit_ttl && inherit_tos &&
+	       mlxsw_sp_ipip_tunnel_complete(MLXSW_SP_L3_PROTO_IPV6, ol_dev);
+}
+
+static struct mlxsw_sp_rif_ipip_lb_config
+mlxsw_sp2_ipip_ol_loopback_config_gre6(struct mlxsw_sp *mlxsw_sp,
+				       const struct net_device *ol_dev)
+{
+	struct __ip6_tnl_parm parms = mlxsw_sp_ipip_netdev_parms6(ol_dev);
+	enum mlxsw_reg_ritr_loopback_ipip_type lb_ipipt;
+
+	lb_ipipt = mlxsw_sp_ipip_parms6_has_okey(&parms) ?
+		MLXSW_REG_RITR_LOOPBACK_IPIP_TYPE_IP_IN_GRE_KEY_IN_IP :
+		MLXSW_REG_RITR_LOOPBACK_IPIP_TYPE_IP_IN_GRE_IN_IP;
+	return (struct mlxsw_sp_rif_ipip_lb_config){
+		.lb_ipipt = lb_ipipt,
+		.okey = mlxsw_sp_ipip_parms6_okey(&parms),
+		.ul_protocol = MLXSW_SP_L3_PROTO_IPV6,
+		.saddr = mlxsw_sp_ipip_netdev_saddr(MLXSW_SP_L3_PROTO_IPV6,
+						    ol_dev),
+	};
+}
+
+static int
+mlxsw_sp2_ipip_ol_netdev_change_gre6(struct mlxsw_sp *mlxsw_sp,
+				     struct mlxsw_sp_ipip_entry *ipip_entry,
+				     struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp_ipip_parms new_parms;
+
+	new_parms = mlxsw_sp2_ipip_netdev_parms_init_gre6(ipip_entry->ol_dev);
+	return mlxsw_sp_ipip_ol_netdev_change_gre(mlxsw_sp, ipip_entry,
+						  &new_parms, extack);
+}
+
+static int
+mlxsw_sp2_ipip_rem_addr_set_gre6(struct mlxsw_sp *mlxsw_sp,
+				 struct mlxsw_sp_ipip_entry *ipip_entry)
+{
+	return mlxsw_sp_ipv6_addr_kvdl_index_get(mlxsw_sp,
+						 &ipip_entry->parms.daddr.addr6,
+						 &ipip_entry->dip_kvdl_index);
+}
+
+static void
+mlxsw_sp2_ipip_rem_addr_unset_gre6(struct mlxsw_sp *mlxsw_sp,
+				   const struct mlxsw_sp_ipip_entry *ipip_entry)
+{
+	mlxsw_sp_ipv6_addr_put(mlxsw_sp, &ipip_entry->parms.daddr.addr6);
+}
+
+static const struct mlxsw_sp_ipip_ops mlxsw_sp2_ipip_gre6_ops = {
+	.dev_type = ARPHRD_IP6GRE,
+	.ul_proto = MLXSW_SP_L3_PROTO_IPV6,
+	.inc_parsing_depth = true,
+	.parms_init = mlxsw_sp2_ipip_netdev_parms_init_gre6,
+	.nexthop_update = mlxsw_sp2_ipip_nexthop_update_gre6,
+	.decap_config = mlxsw_sp2_ipip_decap_config_gre6,
+	.can_offload = mlxsw_sp2_ipip_can_offload_gre6,
+	.ol_loopback_config = mlxsw_sp2_ipip_ol_loopback_config_gre6,
+	.ol_netdev_change = mlxsw_sp2_ipip_ol_netdev_change_gre6,
+	.rem_ip_addr_set = mlxsw_sp2_ipip_rem_addr_set_gre6,
+	.rem_ip_addr_unset = mlxsw_sp2_ipip_rem_addr_unset_gre6,
+};
+
+const struct mlxsw_sp_ipip_ops *mlxsw_sp2_ipip_ops_arr[] = {
+	[MLXSW_SP_IPIP_TYPE_GRE4] = &mlxsw_sp_ipip_gre4_ops,
+	[MLXSW_SP_IPIP_TYPE_GRE6] = &mlxsw_sp2_ipip_gre6_ops,
+};
+
+static int mlxsw_sp_ipip_ecn_encap_init_one(struct mlxsw_sp *mlxsw_sp,
+					    u8 inner_ecn, u8 outer_ecn)
+{
+	char tieem_pl[MLXSW_REG_TIEEM_LEN];
+
+	mlxsw_reg_tieem_pack(tieem_pl, inner_ecn, outer_ecn);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(tieem), tieem_pl);
+}
+
+int mlxsw_sp_ipip_ecn_encap_init(struct mlxsw_sp *mlxsw_sp)
+{
+	int i;
+
+	/* Iterate over inner ECN values */
+	for (i = INET_ECN_NOT_ECT; i <= INET_ECN_CE; i++) {
+		u8 outer_ecn = INET_ECN_encapsulate(0, i);
+		int err;
+
+		err = mlxsw_sp_ipip_ecn_encap_init_one(mlxsw_sp, i, outer_ecn);
+		if (err)
+			return err;
+	}
+
+	return 0;
+}
+
+static int mlxsw_sp_ipip_ecn_decap_init_one(struct mlxsw_sp *mlxsw_sp,
+					    u8 inner_ecn, u8 outer_ecn)
+{
+	char tidem_pl[MLXSW_REG_TIDEM_LEN];
+	u8 new_inner_ecn;
+	bool trap_en;
+
+	new_inner_ecn = mlxsw_sp_tunnel_ecn_decap(outer_ecn, inner_ecn,
+						  &trap_en);
+	mlxsw_reg_tidem_pack(tidem_pl, outer_ecn, inner_ecn, new_inner_ecn,
+			     trap_en, trap_en ? MLXSW_TRAP_ID_DECAP_ECN0 : 0);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(tidem), tidem_pl);
+}
+
+int mlxsw_sp_ipip_ecn_decap_init(struct mlxsw_sp *mlxsw_sp)
+{
+	int i, j, err;
+
+	/* Iterate over inner ECN values */
+	for (i = INET_ECN_NOT_ECT; i <= INET_ECN_CE; i++) {
+		/* Iterate over outer ECN values */
+		for (j = INET_ECN_NOT_ECT; j <= INET_ECN_CE; j++) {
+			err = mlxsw_sp_ipip_ecn_decap_init_one(mlxsw_sp, i, j);
+			if (err)
+				return err;
+		}
+	}
+
+	return 0;
+}
+
+struct net_device *
+mlxsw_sp_ipip_netdev_ul_dev_get(const struct net_device *ol_dev)
+{
+	struct net *net = dev_net(ol_dev);
+	struct ip_tunnel *tun4;
+	struct ip6_tnl *tun6;
+
+	switch (ol_dev->type) {
+	case ARPHRD_IPGRE:
+		tun4 = netdev_priv(ol_dev);
+		return dev_get_by_index_rcu(net, tun4->parms.link);
+	case ARPHRD_IP6GRE:
+		tun6 = netdev_priv(ol_dev);
+		return dev_get_by_index_rcu(net, tun6->parms.link);
+	default:
+		return NULL;
+	}
+}
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_ipip.h b/drivers/net/ethernet/mellanox/mlxsw/spectrum_ipip.h
new file mode 100644
index 000000000..8cc259dcc
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_ipip.h
@@ -0,0 +1,84 @@
+/* SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0 */
+/* Copyright (c) 2017-2018 Mellanox Technologies. All rights reserved */
+
+#ifndef _MLXSW_IPIP_H_
+#define _MLXSW_IPIP_H_
+
+#include "spectrum_router.h"
+#include <net/ip_fib.h>
+#include <linux/if_tunnel.h>
+#include <net/ip6_tunnel.h>
+
+struct ip_tunnel_parm
+mlxsw_sp_ipip_netdev_parms4(const struct net_device *ol_dev);
+struct __ip6_tnl_parm
+mlxsw_sp_ipip_netdev_parms6(const struct net_device *ol_dev);
+
+union mlxsw_sp_l3addr
+mlxsw_sp_ipip_netdev_saddr(enum mlxsw_sp_l3proto proto,
+			   const struct net_device *ol_dev);
+
+bool mlxsw_sp_l3addr_is_zero(union mlxsw_sp_l3addr addr);
+
+enum mlxsw_sp_ipip_type {
+	MLXSW_SP_IPIP_TYPE_GRE4,
+	MLXSW_SP_IPIP_TYPE_GRE6,
+	MLXSW_SP_IPIP_TYPE_MAX,
+};
+
+struct mlxsw_sp_ipip_parms {
+	enum mlxsw_sp_l3proto proto;
+	union mlxsw_sp_l3addr saddr;
+	union mlxsw_sp_l3addr daddr;
+	int link;
+	u32 ikey;
+	u32 okey;
+};
+
+struct mlxsw_sp_ipip_entry {
+	enum mlxsw_sp_ipip_type ipipt;
+	struct net_device *ol_dev; /* Overlay. */
+	struct mlxsw_sp_rif_ipip_lb *ol_lb;
+	struct mlxsw_sp_fib_entry *decap_fib_entry;
+	struct list_head ipip_list_node;
+	struct mlxsw_sp_ipip_parms parms;
+	u32 dip_kvdl_index;
+};
+
+struct mlxsw_sp_ipip_ops {
+	int dev_type;
+	enum mlxsw_sp_l3proto ul_proto; /* Underlay. */
+	bool inc_parsing_depth;
+
+	struct mlxsw_sp_ipip_parms
+	(*parms_init)(const struct net_device *ol_dev);
+
+	int (*nexthop_update)(struct mlxsw_sp *mlxsw_sp, u32 adj_index,
+			      struct mlxsw_sp_ipip_entry *ipip_entry,
+			      bool force, char *ratr_pl);
+
+	bool (*can_offload)(const struct mlxsw_sp *mlxsw_sp,
+			    const struct net_device *ol_dev);
+
+	/* Return a configuration for creating an overlay loopback RIF. */
+	struct mlxsw_sp_rif_ipip_lb_config
+	(*ol_loopback_config)(struct mlxsw_sp *mlxsw_sp,
+			      const struct net_device *ol_dev);
+
+	int (*decap_config)(struct mlxsw_sp *mlxsw_sp,
+			    struct mlxsw_sp_ipip_entry *ipip_entry,
+			    u32 tunnel_index);
+
+	int (*ol_netdev_change)(struct mlxsw_sp *mlxsw_sp,
+				struct mlxsw_sp_ipip_entry *ipip_entry,
+				struct netlink_ext_ack *extack);
+	int (*rem_ip_addr_set)(struct mlxsw_sp *mlxsw_sp,
+			       struct mlxsw_sp_ipip_entry *ipip_entry);
+	void (*rem_ip_addr_unset)(struct mlxsw_sp *mlxsw_sp,
+				  const struct mlxsw_sp_ipip_entry *ipip_entry);
+};
+
+extern const struct mlxsw_sp_ipip_ops *mlxsw_sp1_ipip_ops_arr[];
+extern const struct mlxsw_sp_ipip_ops *mlxsw_sp2_ipip_ops_arr[];
+
+#endif /* _MLXSW_IPIP_H_*/
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_kvdl.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_kvdl.c
new file mode 100644
index 000000000..20d72f1c0
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_kvdl.c
@@ -0,0 +1,87 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2016-2018 Mellanox Technologies. All rights reserved */
+
+#include <linux/kernel.h>
+#include <linux/mutex.h>
+#include <linux/slab.h>
+
+#include "spectrum.h"
+
+struct mlxsw_sp_kvdl {
+	const struct mlxsw_sp_kvdl_ops *kvdl_ops;
+	struct mutex kvdl_lock; /* Protects kvdl allocations */
+	unsigned long priv[];
+	/* priv has to be always the last item */
+};
+
+int mlxsw_sp_kvdl_init(struct mlxsw_sp *mlxsw_sp)
+{
+	const struct mlxsw_sp_kvdl_ops *kvdl_ops = mlxsw_sp->kvdl_ops;
+	struct mlxsw_sp_kvdl *kvdl;
+	int err;
+
+	kvdl = kzalloc(sizeof(*mlxsw_sp->kvdl) + kvdl_ops->priv_size,
+		       GFP_KERNEL);
+	if (!kvdl)
+		return -ENOMEM;
+	mutex_init(&kvdl->kvdl_lock);
+	kvdl->kvdl_ops = kvdl_ops;
+	mlxsw_sp->kvdl = kvdl;
+
+	err = kvdl_ops->init(mlxsw_sp, kvdl->priv);
+	if (err)
+		goto err_init;
+	return 0;
+
+err_init:
+	mutex_destroy(&kvdl->kvdl_lock);
+	kfree(kvdl);
+	return err;
+}
+
+void mlxsw_sp_kvdl_fini(struct mlxsw_sp *mlxsw_sp)
+{
+	struct mlxsw_sp_kvdl *kvdl = mlxsw_sp->kvdl;
+
+	kvdl->kvdl_ops->fini(mlxsw_sp, kvdl->priv);
+	mutex_destroy(&kvdl->kvdl_lock);
+	kfree(kvdl);
+}
+
+int mlxsw_sp_kvdl_alloc(struct mlxsw_sp *mlxsw_sp,
+			enum mlxsw_sp_kvdl_entry_type type,
+			unsigned int entry_count, u32 *p_entry_index)
+{
+	struct mlxsw_sp_kvdl *kvdl = mlxsw_sp->kvdl;
+	int err;
+
+	mutex_lock(&kvdl->kvdl_lock);
+	err = kvdl->kvdl_ops->alloc(mlxsw_sp, kvdl->priv, type,
+				    entry_count, p_entry_index);
+	mutex_unlock(&kvdl->kvdl_lock);
+
+	return err;
+}
+
+void mlxsw_sp_kvdl_free(struct mlxsw_sp *mlxsw_sp,
+			enum mlxsw_sp_kvdl_entry_type type,
+			unsigned int entry_count, int entry_index)
+{
+	struct mlxsw_sp_kvdl *kvdl = mlxsw_sp->kvdl;
+
+	mutex_lock(&kvdl->kvdl_lock);
+	kvdl->kvdl_ops->free(mlxsw_sp, kvdl->priv, type,
+			     entry_count, entry_index);
+	mutex_unlock(&kvdl->kvdl_lock);
+}
+
+int mlxsw_sp_kvdl_alloc_count_query(struct mlxsw_sp *mlxsw_sp,
+				    enum mlxsw_sp_kvdl_entry_type type,
+				    unsigned int entry_count,
+				    unsigned int *p_alloc_count)
+{
+	struct mlxsw_sp_kvdl *kvdl = mlxsw_sp->kvdl;
+
+	return kvdl->kvdl_ops->alloc_size_query(mlxsw_sp, kvdl->priv, type,
+						entry_count, p_alloc_count);
+}
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_matchall.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_matchall.c
new file mode 100644
index 000000000..07b371cd9
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_matchall.c
@@ -0,0 +1,478 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2017-2020 Mellanox Technologies. All rights reserved */
+
+#include <linux/kernel.h>
+#include <linux/errno.h>
+#include <linux/netdevice.h>
+#include <net/flow_offload.h>
+
+#include "spectrum.h"
+#include "spectrum_span.h"
+#include "reg.h"
+
+static struct mlxsw_sp_mall_entry *
+mlxsw_sp_mall_entry_find(struct mlxsw_sp_flow_block *block, unsigned long cookie)
+{
+	struct mlxsw_sp_mall_entry *mall_entry;
+
+	list_for_each_entry(mall_entry, &block->mall.list, list)
+		if (mall_entry->cookie == cookie)
+			return mall_entry;
+
+	return NULL;
+}
+
+static int
+mlxsw_sp_mall_port_mirror_add(struct mlxsw_sp_port *mlxsw_sp_port,
+			      struct mlxsw_sp_mall_entry *mall_entry,
+			      struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	struct mlxsw_sp_span_agent_parms agent_parms = {};
+	struct mlxsw_sp_span_trigger_parms parms;
+	enum mlxsw_sp_span_trigger trigger;
+	int err;
+
+	if (!mall_entry->mirror.to_dev) {
+		NL_SET_ERR_MSG(extack, "Could not find requested device");
+		return -EINVAL;
+	}
+
+	agent_parms.to_dev = mall_entry->mirror.to_dev;
+	err = mlxsw_sp_span_agent_get(mlxsw_sp, &mall_entry->mirror.span_id,
+				      &agent_parms);
+	if (err) {
+		NL_SET_ERR_MSG(extack, "Failed to get SPAN agent");
+		return err;
+	}
+
+	err = mlxsw_sp_span_analyzed_port_get(mlxsw_sp_port,
+					      mall_entry->ingress);
+	if (err) {
+		NL_SET_ERR_MSG(extack, "Failed to get analyzed port");
+		goto err_analyzed_port_get;
+	}
+
+	trigger = mall_entry->ingress ? MLXSW_SP_SPAN_TRIGGER_INGRESS :
+					MLXSW_SP_SPAN_TRIGGER_EGRESS;
+	parms.span_id = mall_entry->mirror.span_id;
+	parms.probability_rate = 1;
+	err = mlxsw_sp_span_agent_bind(mlxsw_sp, trigger, mlxsw_sp_port,
+				       &parms);
+	if (err) {
+		NL_SET_ERR_MSG(extack, "Failed to bind SPAN agent");
+		goto err_agent_bind;
+	}
+
+	return 0;
+
+err_agent_bind:
+	mlxsw_sp_span_analyzed_port_put(mlxsw_sp_port, mall_entry->ingress);
+err_analyzed_port_get:
+	mlxsw_sp_span_agent_put(mlxsw_sp, mall_entry->mirror.span_id);
+	return err;
+}
+
+static void
+mlxsw_sp_mall_port_mirror_del(struct mlxsw_sp_port *mlxsw_sp_port,
+			      struct mlxsw_sp_mall_entry *mall_entry)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	struct mlxsw_sp_span_trigger_parms parms;
+	enum mlxsw_sp_span_trigger trigger;
+
+	trigger = mall_entry->ingress ? MLXSW_SP_SPAN_TRIGGER_INGRESS :
+					MLXSW_SP_SPAN_TRIGGER_EGRESS;
+	parms.span_id = mall_entry->mirror.span_id;
+	mlxsw_sp_span_agent_unbind(mlxsw_sp, trigger, mlxsw_sp_port, &parms);
+	mlxsw_sp_span_analyzed_port_put(mlxsw_sp_port, mall_entry->ingress);
+	mlxsw_sp_span_agent_put(mlxsw_sp, mall_entry->mirror.span_id);
+}
+
+static int mlxsw_sp_mall_port_sample_set(struct mlxsw_sp_port *mlxsw_sp_port,
+					 bool enable, u32 rate)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	char mpsc_pl[MLXSW_REG_MPSC_LEN];
+
+	mlxsw_reg_mpsc_pack(mpsc_pl, mlxsw_sp_port->local_port, enable, rate);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(mpsc), mpsc_pl);
+}
+
+static int
+mlxsw_sp_mall_port_sample_add(struct mlxsw_sp_port *mlxsw_sp_port,
+			      struct mlxsw_sp_mall_entry *mall_entry,
+			      struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	struct mlxsw_sp_sample_trigger trigger;
+	int err;
+
+	if (mall_entry->ingress)
+		trigger.type = MLXSW_SP_SAMPLE_TRIGGER_TYPE_INGRESS;
+	else
+		trigger.type = MLXSW_SP_SAMPLE_TRIGGER_TYPE_EGRESS;
+	trigger.local_port = mlxsw_sp_port->local_port;
+	err = mlxsw_sp_sample_trigger_params_set(mlxsw_sp, &trigger,
+						 &mall_entry->sample.params,
+						 extack);
+	if (err)
+		return err;
+
+	err = mlxsw_sp->mall_ops->sample_add(mlxsw_sp, mlxsw_sp_port,
+					     mall_entry, extack);
+	if (err)
+		goto err_port_sample_set;
+	return 0;
+
+err_port_sample_set:
+	mlxsw_sp_sample_trigger_params_unset(mlxsw_sp, &trigger);
+	return err;
+}
+
+static void
+mlxsw_sp_mall_port_sample_del(struct mlxsw_sp_port *mlxsw_sp_port,
+			      struct mlxsw_sp_mall_entry *mall_entry)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	struct mlxsw_sp_sample_trigger trigger;
+
+	if (mall_entry->ingress)
+		trigger.type = MLXSW_SP_SAMPLE_TRIGGER_TYPE_INGRESS;
+	else
+		trigger.type = MLXSW_SP_SAMPLE_TRIGGER_TYPE_EGRESS;
+	trigger.local_port = mlxsw_sp_port->local_port;
+
+	mlxsw_sp->mall_ops->sample_del(mlxsw_sp, mlxsw_sp_port, mall_entry);
+	mlxsw_sp_sample_trigger_params_unset(mlxsw_sp, &trigger);
+}
+
+static int
+mlxsw_sp_mall_port_rule_add(struct mlxsw_sp_port *mlxsw_sp_port,
+			    struct mlxsw_sp_mall_entry *mall_entry,
+			    struct netlink_ext_ack *extack)
+{
+	switch (mall_entry->type) {
+	case MLXSW_SP_MALL_ACTION_TYPE_MIRROR:
+		return mlxsw_sp_mall_port_mirror_add(mlxsw_sp_port, mall_entry,
+						     extack);
+	case MLXSW_SP_MALL_ACTION_TYPE_SAMPLE:
+		return mlxsw_sp_mall_port_sample_add(mlxsw_sp_port, mall_entry,
+						     extack);
+	default:
+		WARN_ON(1);
+		return -EINVAL;
+	}
+}
+
+static void
+mlxsw_sp_mall_port_rule_del(struct mlxsw_sp_port *mlxsw_sp_port,
+			    struct mlxsw_sp_mall_entry *mall_entry)
+{
+	switch (mall_entry->type) {
+	case MLXSW_SP_MALL_ACTION_TYPE_MIRROR:
+		mlxsw_sp_mall_port_mirror_del(mlxsw_sp_port, mall_entry);
+		break;
+	case MLXSW_SP_MALL_ACTION_TYPE_SAMPLE:
+		mlxsw_sp_mall_port_sample_del(mlxsw_sp_port, mall_entry);
+		break;
+	default:
+		WARN_ON(1);
+	}
+}
+
+static void mlxsw_sp_mall_prio_update(struct mlxsw_sp_flow_block *block)
+{
+	struct mlxsw_sp_mall_entry *mall_entry;
+
+	if (list_empty(&block->mall.list))
+		return;
+	block->mall.min_prio = UINT_MAX;
+	block->mall.max_prio = 0;
+	list_for_each_entry(mall_entry, &block->mall.list, list) {
+		if (mall_entry->priority < block->mall.min_prio)
+			block->mall.min_prio = mall_entry->priority;
+		if (mall_entry->priority > block->mall.max_prio)
+			block->mall.max_prio = mall_entry->priority;
+	}
+}
+
+int mlxsw_sp_mall_replace(struct mlxsw_sp *mlxsw_sp,
+			  struct mlxsw_sp_flow_block *block,
+			  struct tc_cls_matchall_offload *f)
+{
+	struct mlxsw_sp_flow_block_binding *binding;
+	struct mlxsw_sp_mall_entry *mall_entry;
+	__be16 protocol = f->common.protocol;
+	struct flow_action_entry *act;
+	unsigned int flower_min_prio;
+	unsigned int flower_max_prio;
+	bool flower_prio_valid;
+	int err;
+
+	if (!flow_offload_has_one_action(&f->rule->action)) {
+		NL_SET_ERR_MSG(f->common.extack, "Only singular actions are supported");
+		return -EOPNOTSUPP;
+	}
+
+	if (f->common.chain_index) {
+		NL_SET_ERR_MSG(f->common.extack, "Only chain 0 is supported");
+		return -EOPNOTSUPP;
+	}
+
+	if (mlxsw_sp_flow_block_is_mixed_bound(block)) {
+		NL_SET_ERR_MSG(f->common.extack, "Only not mixed bound blocks are supported");
+		return -EOPNOTSUPP;
+	}
+
+	err = mlxsw_sp_flower_prio_get(mlxsw_sp, block, f->common.chain_index,
+				       &flower_min_prio, &flower_max_prio);
+	if (err) {
+		if (err != -ENOENT) {
+			NL_SET_ERR_MSG(f->common.extack, "Failed to get flower priorities");
+			return err;
+		}
+		flower_prio_valid = false;
+		/* No flower filters are installed in specified chain. */
+	} else {
+		flower_prio_valid = true;
+	}
+
+	if (protocol != htons(ETH_P_ALL)) {
+		NL_SET_ERR_MSG(f->common.extack, "matchall rules only supported with 'all' protocol");
+		return -EOPNOTSUPP;
+	}
+
+	mall_entry = kzalloc(sizeof(*mall_entry), GFP_KERNEL);
+	if (!mall_entry)
+		return -ENOMEM;
+	mall_entry->cookie = f->cookie;
+	mall_entry->priority = f->common.prio;
+	mall_entry->ingress = mlxsw_sp_flow_block_is_ingress_bound(block);
+
+	if (flower_prio_valid && mall_entry->ingress &&
+	    mall_entry->priority >= flower_min_prio) {
+		NL_SET_ERR_MSG(f->common.extack, "Failed to add behind existing flower rules");
+		err = -EOPNOTSUPP;
+		goto errout;
+	}
+	if (flower_prio_valid && !mall_entry->ingress &&
+	    mall_entry->priority <= flower_max_prio) {
+		NL_SET_ERR_MSG(f->common.extack, "Failed to add in front of existing flower rules");
+		err = -EOPNOTSUPP;
+		goto errout;
+	}
+
+	act = &f->rule->action.entries[0];
+
+	switch (act->id) {
+	case FLOW_ACTION_MIRRED:
+		mall_entry->type = MLXSW_SP_MALL_ACTION_TYPE_MIRROR;
+		mall_entry->mirror.to_dev = act->dev;
+		break;
+	case FLOW_ACTION_SAMPLE:
+		mall_entry->type = MLXSW_SP_MALL_ACTION_TYPE_SAMPLE;
+		mall_entry->sample.params.psample_group = act->sample.psample_group;
+		mall_entry->sample.params.truncate = act->sample.truncate;
+		mall_entry->sample.params.trunc_size = act->sample.trunc_size;
+		mall_entry->sample.params.rate = act->sample.rate;
+		break;
+	default:
+		err = -EOPNOTSUPP;
+		goto errout;
+	}
+
+	list_for_each_entry(binding, &block->binding_list, list) {
+		err = mlxsw_sp_mall_port_rule_add(binding->mlxsw_sp_port,
+						  mall_entry, f->common.extack);
+		if (err)
+			goto rollback;
+	}
+
+	block->rule_count++;
+	if (mall_entry->ingress)
+		block->egress_blocker_rule_count++;
+	else
+		block->ingress_blocker_rule_count++;
+	list_add_tail(&mall_entry->list, &block->mall.list);
+	mlxsw_sp_mall_prio_update(block);
+	return 0;
+
+rollback:
+	list_for_each_entry_continue_reverse(binding, &block->binding_list,
+					     list)
+		mlxsw_sp_mall_port_rule_del(binding->mlxsw_sp_port, mall_entry);
+errout:
+	kfree(mall_entry);
+	return err;
+}
+
+void mlxsw_sp_mall_destroy(struct mlxsw_sp_flow_block *block,
+			   struct tc_cls_matchall_offload *f)
+{
+	struct mlxsw_sp_flow_block_binding *binding;
+	struct mlxsw_sp_mall_entry *mall_entry;
+
+	mall_entry = mlxsw_sp_mall_entry_find(block, f->cookie);
+	if (!mall_entry) {
+		NL_SET_ERR_MSG(f->common.extack, "Entry not found");
+		return;
+	}
+
+	list_del(&mall_entry->list);
+	if (mall_entry->ingress)
+		block->egress_blocker_rule_count--;
+	else
+		block->ingress_blocker_rule_count--;
+	block->rule_count--;
+	list_for_each_entry(binding, &block->binding_list, list)
+		mlxsw_sp_mall_port_rule_del(binding->mlxsw_sp_port, mall_entry);
+	kfree_rcu(mall_entry, rcu); /* sample RX packets may be in-flight */
+	mlxsw_sp_mall_prio_update(block);
+}
+
+int mlxsw_sp_mall_port_bind(struct mlxsw_sp_flow_block *block,
+			    struct mlxsw_sp_port *mlxsw_sp_port,
+			    struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp_mall_entry *mall_entry;
+	int err;
+
+	list_for_each_entry(mall_entry, &block->mall.list, list) {
+		err = mlxsw_sp_mall_port_rule_add(mlxsw_sp_port, mall_entry,
+						  extack);
+		if (err)
+			goto rollback;
+	}
+	return 0;
+
+rollback:
+	list_for_each_entry_continue_reverse(mall_entry, &block->mall.list,
+					     list)
+		mlxsw_sp_mall_port_rule_del(mlxsw_sp_port, mall_entry);
+	return err;
+}
+
+void mlxsw_sp_mall_port_unbind(struct mlxsw_sp_flow_block *block,
+			       struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	struct mlxsw_sp_mall_entry *mall_entry;
+
+	list_for_each_entry(mall_entry, &block->mall.list, list)
+		mlxsw_sp_mall_port_rule_del(mlxsw_sp_port, mall_entry);
+}
+
+int mlxsw_sp_mall_prio_get(struct mlxsw_sp_flow_block *block, u32 chain_index,
+			   unsigned int *p_min_prio, unsigned int *p_max_prio)
+{
+	if (chain_index || list_empty(&block->mall.list))
+		/* In case there are no matchall rules, the caller
+		 * receives -ENOENT to indicate there is no need
+		 * to check the priorities.
+		 */
+		return -ENOENT;
+	*p_min_prio = block->mall.min_prio;
+	*p_max_prio = block->mall.max_prio;
+	return 0;
+}
+
+static int mlxsw_sp1_mall_sample_add(struct mlxsw_sp *mlxsw_sp,
+				     struct mlxsw_sp_port *mlxsw_sp_port,
+				     struct mlxsw_sp_mall_entry *mall_entry,
+				     struct netlink_ext_ack *extack)
+{
+	u32 rate = mall_entry->sample.params.rate;
+
+	if (!mall_entry->ingress) {
+		NL_SET_ERR_MSG(extack, "Sampling is not supported on egress");
+		return -EOPNOTSUPP;
+	}
+
+	if (rate > MLXSW_REG_MPSC_RATE_MAX) {
+		NL_SET_ERR_MSG(extack, "Unsupported sampling rate");
+		return -EOPNOTSUPP;
+	}
+
+	return mlxsw_sp_mall_port_sample_set(mlxsw_sp_port, true, rate);
+}
+
+static void mlxsw_sp1_mall_sample_del(struct mlxsw_sp *mlxsw_sp,
+				      struct mlxsw_sp_port *mlxsw_sp_port,
+				      struct mlxsw_sp_mall_entry *mall_entry)
+{
+	mlxsw_sp_mall_port_sample_set(mlxsw_sp_port, false, 1);
+}
+
+const struct mlxsw_sp_mall_ops mlxsw_sp1_mall_ops = {
+	.sample_add = mlxsw_sp1_mall_sample_add,
+	.sample_del = mlxsw_sp1_mall_sample_del,
+};
+
+static int mlxsw_sp2_mall_sample_add(struct mlxsw_sp *mlxsw_sp,
+				     struct mlxsw_sp_port *mlxsw_sp_port,
+				     struct mlxsw_sp_mall_entry *mall_entry,
+				     struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp_span_trigger_parms trigger_parms = {};
+	struct mlxsw_sp_span_agent_parms agent_parms = {
+		.to_dev = NULL,	/* Mirror to CPU. */
+		.session_id = MLXSW_SP_SPAN_SESSION_ID_SAMPLING,
+	};
+	u32 rate = mall_entry->sample.params.rate;
+	enum mlxsw_sp_span_trigger span_trigger;
+	int err;
+
+	err = mlxsw_sp_span_agent_get(mlxsw_sp, &mall_entry->sample.span_id,
+				      &agent_parms);
+	if (err) {
+		NL_SET_ERR_MSG(extack, "Failed to get SPAN agent");
+		return err;
+	}
+
+	err = mlxsw_sp_span_analyzed_port_get(mlxsw_sp_port,
+					      mall_entry->ingress);
+	if (err) {
+		NL_SET_ERR_MSG(extack, "Failed to get analyzed port");
+		goto err_analyzed_port_get;
+	}
+
+	span_trigger = mall_entry->ingress ? MLXSW_SP_SPAN_TRIGGER_INGRESS :
+					     MLXSW_SP_SPAN_TRIGGER_EGRESS;
+	trigger_parms.span_id = mall_entry->sample.span_id;
+	trigger_parms.probability_rate = rate;
+	err = mlxsw_sp_span_agent_bind(mlxsw_sp, span_trigger, mlxsw_sp_port,
+				       &trigger_parms);
+	if (err) {
+		NL_SET_ERR_MSG(extack, "Failed to bind SPAN agent");
+		goto err_agent_bind;
+	}
+
+	return 0;
+
+err_agent_bind:
+	mlxsw_sp_span_analyzed_port_put(mlxsw_sp_port, mall_entry->ingress);
+err_analyzed_port_get:
+	mlxsw_sp_span_agent_put(mlxsw_sp, mall_entry->sample.span_id);
+	return err;
+}
+
+static void mlxsw_sp2_mall_sample_del(struct mlxsw_sp *mlxsw_sp,
+				      struct mlxsw_sp_port *mlxsw_sp_port,
+				      struct mlxsw_sp_mall_entry *mall_entry)
+{
+	struct mlxsw_sp_span_trigger_parms trigger_parms = {};
+	enum mlxsw_sp_span_trigger span_trigger;
+
+	span_trigger = mall_entry->ingress ? MLXSW_SP_SPAN_TRIGGER_INGRESS :
+					     MLXSW_SP_SPAN_TRIGGER_EGRESS;
+	trigger_parms.span_id = mall_entry->sample.span_id;
+	mlxsw_sp_span_agent_unbind(mlxsw_sp, span_trigger, mlxsw_sp_port,
+				   &trigger_parms);
+	mlxsw_sp_span_analyzed_port_put(mlxsw_sp_port, mall_entry->ingress);
+	mlxsw_sp_span_agent_put(mlxsw_sp, mall_entry->sample.span_id);
+}
+
+const struct mlxsw_sp_mall_ops mlxsw_sp2_mall_ops = {
+	.sample_add = mlxsw_sp2_mall_sample_add,
+	.sample_del = mlxsw_sp2_mall_sample_del,
+};
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_mr.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_mr.c
new file mode 100644
index 000000000..1f6bc0c7e
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_mr.c
@@ -0,0 +1,1075 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2017-2018 Mellanox Technologies. All rights reserved */
+
+#include <linux/mutex.h>
+#include <linux/rhashtable.h>
+#include <net/ipv6.h>
+
+#include "spectrum_mr.h"
+#include "spectrum_router.h"
+
+struct mlxsw_sp_mr {
+	const struct mlxsw_sp_mr_ops *mr_ops;
+	void *catchall_route_priv;
+	struct delayed_work stats_update_dw;
+	struct list_head table_list;
+	struct mutex table_list_lock; /* Protects table_list */
+#define MLXSW_SP_MR_ROUTES_COUNTER_UPDATE_INTERVAL 5000 /* ms */
+	unsigned long priv[];
+	/* priv has to be always the last item */
+};
+
+struct mlxsw_sp_mr_vif;
+struct mlxsw_sp_mr_vif_ops {
+	bool (*is_regular)(const struct mlxsw_sp_mr_vif *vif);
+};
+
+struct mlxsw_sp_mr_vif {
+	struct net_device *dev;
+	const struct mlxsw_sp_rif *rif;
+	unsigned long vif_flags;
+
+	/* A list of route_vif_entry structs that point to routes that the VIF
+	 * instance is used as one of the egress VIFs
+	 */
+	struct list_head route_evif_list;
+
+	/* A list of route_vif_entry structs that point to routes that the VIF
+	 * instance is used as an ingress VIF
+	 */
+	struct list_head route_ivif_list;
+
+	/* Protocol specific operations for a VIF */
+	const struct mlxsw_sp_mr_vif_ops *ops;
+};
+
+struct mlxsw_sp_mr_route_vif_entry {
+	struct list_head vif_node;
+	struct list_head route_node;
+	struct mlxsw_sp_mr_vif *mr_vif;
+	struct mlxsw_sp_mr_route *mr_route;
+};
+
+struct mlxsw_sp_mr_table;
+struct mlxsw_sp_mr_table_ops {
+	bool (*is_route_valid)(const struct mlxsw_sp_mr_table *mr_table,
+			       const struct mr_mfc *mfc);
+	void (*key_create)(struct mlxsw_sp_mr_table *mr_table,
+			   struct mlxsw_sp_mr_route_key *key,
+			   struct mr_mfc *mfc);
+	bool (*is_route_starg)(const struct mlxsw_sp_mr_table *mr_table,
+			       const struct mlxsw_sp_mr_route *mr_route);
+};
+
+struct mlxsw_sp_mr_table {
+	struct list_head node;
+	enum mlxsw_sp_l3proto proto;
+	struct mlxsw_sp *mlxsw_sp;
+	u32 vr_id;
+	struct mlxsw_sp_mr_vif vifs[MAXVIFS];
+	struct list_head route_list;
+	struct mutex route_list_lock; /* Protects route_list */
+	struct rhashtable route_ht;
+	const struct mlxsw_sp_mr_table_ops *ops;
+	char catchall_route_priv[];
+	/* catchall_route_priv has to be always the last item */
+};
+
+struct mlxsw_sp_mr_route {
+	struct list_head node;
+	struct rhash_head ht_node;
+	struct mlxsw_sp_mr_route_key key;
+	enum mlxsw_sp_mr_route_action route_action;
+	u16 min_mtu;
+	struct mr_mfc *mfc;
+	void *route_priv;
+	const struct mlxsw_sp_mr_table *mr_table;
+	/* A list of route_vif_entry structs that point to the egress VIFs */
+	struct list_head evif_list;
+	/* A route_vif_entry struct that point to the ingress VIF */
+	struct mlxsw_sp_mr_route_vif_entry ivif;
+};
+
+static const struct rhashtable_params mlxsw_sp_mr_route_ht_params = {
+	.key_len = sizeof(struct mlxsw_sp_mr_route_key),
+	.key_offset = offsetof(struct mlxsw_sp_mr_route, key),
+	.head_offset = offsetof(struct mlxsw_sp_mr_route, ht_node),
+	.automatic_shrinking = true,
+};
+
+static bool mlxsw_sp_mr_vif_valid(const struct mlxsw_sp_mr_vif *vif)
+{
+	return vif->ops->is_regular(vif) && vif->dev && vif->rif;
+}
+
+static bool mlxsw_sp_mr_vif_exists(const struct mlxsw_sp_mr_vif *vif)
+{
+	return vif->dev;
+}
+
+static bool
+mlxsw_sp_mr_route_ivif_in_evifs(const struct mlxsw_sp_mr_route *mr_route)
+{
+	vifi_t ivif = mr_route->mfc->mfc_parent;
+
+	return mr_route->mfc->mfc_un.res.ttls[ivif] != 255;
+}
+
+static int
+mlxsw_sp_mr_route_valid_evifs_num(const struct mlxsw_sp_mr_route *mr_route)
+{
+	struct mlxsw_sp_mr_route_vif_entry *rve;
+	int valid_evifs;
+
+	valid_evifs = 0;
+	list_for_each_entry(rve, &mr_route->evif_list, route_node)
+		if (mlxsw_sp_mr_vif_valid(rve->mr_vif))
+			valid_evifs++;
+	return valid_evifs;
+}
+
+static enum mlxsw_sp_mr_route_action
+mlxsw_sp_mr_route_action(const struct mlxsw_sp_mr_route *mr_route)
+{
+	struct mlxsw_sp_mr_route_vif_entry *rve;
+
+	/* If the ingress port is not regular and resolved, trap the route */
+	if (!mlxsw_sp_mr_vif_valid(mr_route->ivif.mr_vif))
+		return MLXSW_SP_MR_ROUTE_ACTION_TRAP;
+
+	/* The kernel does not match a (*,G) route that the ingress interface is
+	 * not one of the egress interfaces, so trap these kind of routes.
+	 */
+	if (mr_route->mr_table->ops->is_route_starg(mr_route->mr_table,
+						    mr_route) &&
+	    !mlxsw_sp_mr_route_ivif_in_evifs(mr_route))
+		return MLXSW_SP_MR_ROUTE_ACTION_TRAP;
+
+	/* If the route has no valid eVIFs, trap it. */
+	if (!mlxsw_sp_mr_route_valid_evifs_num(mr_route))
+		return MLXSW_SP_MR_ROUTE_ACTION_TRAP;
+
+	/* If one of the eVIFs has no RIF, trap-and-forward the route as there
+	 * is some more routing to do in software too.
+	 */
+	list_for_each_entry(rve, &mr_route->evif_list, route_node)
+		if (mlxsw_sp_mr_vif_exists(rve->mr_vif) && !rve->mr_vif->rif)
+			return MLXSW_SP_MR_ROUTE_ACTION_TRAP_AND_FORWARD;
+
+	return MLXSW_SP_MR_ROUTE_ACTION_FORWARD;
+}
+
+static enum mlxsw_sp_mr_route_prio
+mlxsw_sp_mr_route_prio(const struct mlxsw_sp_mr_route *mr_route)
+{
+	return mr_route->mr_table->ops->is_route_starg(mr_route->mr_table,
+						       mr_route) ?
+		MLXSW_SP_MR_ROUTE_PRIO_STARG : MLXSW_SP_MR_ROUTE_PRIO_SG;
+}
+
+static int mlxsw_sp_mr_route_evif_link(struct mlxsw_sp_mr_route *mr_route,
+				       struct mlxsw_sp_mr_vif *mr_vif)
+{
+	struct mlxsw_sp_mr_route_vif_entry *rve;
+
+	rve = kzalloc(sizeof(*rve), GFP_KERNEL);
+	if (!rve)
+		return -ENOMEM;
+	rve->mr_route = mr_route;
+	rve->mr_vif = mr_vif;
+	list_add_tail(&rve->route_node, &mr_route->evif_list);
+	list_add_tail(&rve->vif_node, &mr_vif->route_evif_list);
+	return 0;
+}
+
+static void
+mlxsw_sp_mr_route_evif_unlink(struct mlxsw_sp_mr_route_vif_entry *rve)
+{
+	list_del(&rve->route_node);
+	list_del(&rve->vif_node);
+	kfree(rve);
+}
+
+static void mlxsw_sp_mr_route_ivif_link(struct mlxsw_sp_mr_route *mr_route,
+					struct mlxsw_sp_mr_vif *mr_vif)
+{
+	mr_route->ivif.mr_route = mr_route;
+	mr_route->ivif.mr_vif = mr_vif;
+	list_add_tail(&mr_route->ivif.vif_node, &mr_vif->route_ivif_list);
+}
+
+static void mlxsw_sp_mr_route_ivif_unlink(struct mlxsw_sp_mr_route *mr_route)
+{
+	list_del(&mr_route->ivif.vif_node);
+}
+
+static int
+mlxsw_sp_mr_route_info_create(struct mlxsw_sp_mr_table *mr_table,
+			      struct mlxsw_sp_mr_route *mr_route,
+			      struct mlxsw_sp_mr_route_info *route_info)
+{
+	struct mlxsw_sp_mr_route_vif_entry *rve;
+	u16 *erif_indices;
+	u16 irif_index;
+	u16 erif = 0;
+
+	erif_indices = kmalloc_array(MAXVIFS, sizeof(*erif_indices),
+				     GFP_KERNEL);
+	if (!erif_indices)
+		return -ENOMEM;
+
+	list_for_each_entry(rve, &mr_route->evif_list, route_node) {
+		if (mlxsw_sp_mr_vif_valid(rve->mr_vif)) {
+			u16 rifi = mlxsw_sp_rif_index(rve->mr_vif->rif);
+
+			erif_indices[erif++] = rifi;
+		}
+	}
+
+	if (mlxsw_sp_mr_vif_valid(mr_route->ivif.mr_vif))
+		irif_index = mlxsw_sp_rif_index(mr_route->ivif.mr_vif->rif);
+	else
+		irif_index = 0;
+
+	route_info->irif_index = irif_index;
+	route_info->erif_indices = erif_indices;
+	route_info->min_mtu = mr_route->min_mtu;
+	route_info->route_action = mr_route->route_action;
+	route_info->erif_num = erif;
+	return 0;
+}
+
+static void
+mlxsw_sp_mr_route_info_destroy(struct mlxsw_sp_mr_route_info *route_info)
+{
+	kfree(route_info->erif_indices);
+}
+
+static int mlxsw_sp_mr_route_write(struct mlxsw_sp_mr_table *mr_table,
+				   struct mlxsw_sp_mr_route *mr_route,
+				   bool replace)
+{
+	struct mlxsw_sp *mlxsw_sp = mr_table->mlxsw_sp;
+	struct mlxsw_sp_mr_route_info route_info;
+	struct mlxsw_sp_mr *mr = mlxsw_sp->mr;
+	int err;
+
+	err = mlxsw_sp_mr_route_info_create(mr_table, mr_route, &route_info);
+	if (err)
+		return err;
+
+	if (!replace) {
+		struct mlxsw_sp_mr_route_params route_params;
+
+		mr_route->route_priv = kzalloc(mr->mr_ops->route_priv_size,
+					       GFP_KERNEL);
+		if (!mr_route->route_priv) {
+			err = -ENOMEM;
+			goto out;
+		}
+
+		route_params.key = mr_route->key;
+		route_params.value = route_info;
+		route_params.prio = mlxsw_sp_mr_route_prio(mr_route);
+		err = mr->mr_ops->route_create(mlxsw_sp, mr->priv,
+					       mr_route->route_priv,
+					       &route_params);
+		if (err)
+			kfree(mr_route->route_priv);
+	} else {
+		err = mr->mr_ops->route_update(mlxsw_sp, mr_route->route_priv,
+					       &route_info);
+	}
+out:
+	mlxsw_sp_mr_route_info_destroy(&route_info);
+	return err;
+}
+
+static void mlxsw_sp_mr_route_erase(struct mlxsw_sp_mr_table *mr_table,
+				    struct mlxsw_sp_mr_route *mr_route)
+{
+	struct mlxsw_sp *mlxsw_sp = mr_table->mlxsw_sp;
+	struct mlxsw_sp_mr *mr = mlxsw_sp->mr;
+
+	mr->mr_ops->route_destroy(mlxsw_sp, mr->priv, mr_route->route_priv);
+	kfree(mr_route->route_priv);
+}
+
+static struct mlxsw_sp_mr_route *
+mlxsw_sp_mr_route_create(struct mlxsw_sp_mr_table *mr_table,
+			 struct mr_mfc *mfc)
+{
+	struct mlxsw_sp_mr_route_vif_entry *rve, *tmp;
+	struct mlxsw_sp_mr_route *mr_route;
+	int err = 0;
+	int i;
+
+	/* Allocate and init a new route and fill it with parameters */
+	mr_route = kzalloc(sizeof(*mr_route), GFP_KERNEL);
+	if (!mr_route)
+		return ERR_PTR(-ENOMEM);
+	INIT_LIST_HEAD(&mr_route->evif_list);
+
+	/* Find min_mtu and link iVIF and eVIFs */
+	mr_route->min_mtu = ETH_MAX_MTU;
+	mr_cache_hold(mfc);
+	mr_route->mfc = mfc;
+	mr_table->ops->key_create(mr_table, &mr_route->key, mr_route->mfc);
+
+	mr_route->mr_table = mr_table;
+	for (i = 0; i < MAXVIFS; i++) {
+		if (mfc->mfc_un.res.ttls[i] != 255) {
+			err = mlxsw_sp_mr_route_evif_link(mr_route,
+							  &mr_table->vifs[i]);
+			if (err)
+				goto err;
+			if (mr_table->vifs[i].dev &&
+			    mr_table->vifs[i].dev->mtu < mr_route->min_mtu)
+				mr_route->min_mtu = mr_table->vifs[i].dev->mtu;
+		}
+	}
+	mlxsw_sp_mr_route_ivif_link(mr_route,
+				    &mr_table->vifs[mfc->mfc_parent]);
+
+	mr_route->route_action = mlxsw_sp_mr_route_action(mr_route);
+	return mr_route;
+err:
+	mr_cache_put(mfc);
+	list_for_each_entry_safe(rve, tmp, &mr_route->evif_list, route_node)
+		mlxsw_sp_mr_route_evif_unlink(rve);
+	kfree(mr_route);
+	return ERR_PTR(err);
+}
+
+static void mlxsw_sp_mr_route_destroy(struct mlxsw_sp_mr_table *mr_table,
+				      struct mlxsw_sp_mr_route *mr_route)
+{
+	struct mlxsw_sp_mr_route_vif_entry *rve, *tmp;
+
+	mlxsw_sp_mr_route_ivif_unlink(mr_route);
+	mr_cache_put(mr_route->mfc);
+	list_for_each_entry_safe(rve, tmp, &mr_route->evif_list, route_node)
+		mlxsw_sp_mr_route_evif_unlink(rve);
+	kfree(mr_route);
+}
+
+static void mlxsw_sp_mr_mfc_offload_set(struct mlxsw_sp_mr_route *mr_route,
+					bool offload)
+{
+	if (offload)
+		mr_route->mfc->mfc_flags |= MFC_OFFLOAD;
+	else
+		mr_route->mfc->mfc_flags &= ~MFC_OFFLOAD;
+}
+
+static void mlxsw_sp_mr_mfc_offload_update(struct mlxsw_sp_mr_route *mr_route)
+{
+	bool offload;
+
+	offload = mr_route->route_action != MLXSW_SP_MR_ROUTE_ACTION_TRAP;
+	mlxsw_sp_mr_mfc_offload_set(mr_route, offload);
+}
+
+static void __mlxsw_sp_mr_route_del(struct mlxsw_sp_mr_table *mr_table,
+				    struct mlxsw_sp_mr_route *mr_route)
+{
+	WARN_ON_ONCE(!mutex_is_locked(&mr_table->route_list_lock));
+
+	mlxsw_sp_mr_mfc_offload_set(mr_route, false);
+	rhashtable_remove_fast(&mr_table->route_ht, &mr_route->ht_node,
+			       mlxsw_sp_mr_route_ht_params);
+	list_del(&mr_route->node);
+	mlxsw_sp_mr_route_erase(mr_table, mr_route);
+	mlxsw_sp_mr_route_destroy(mr_table, mr_route);
+}
+
+int mlxsw_sp_mr_route_add(struct mlxsw_sp_mr_table *mr_table,
+			  struct mr_mfc *mfc, bool replace)
+{
+	struct mlxsw_sp_mr_route *mr_orig_route = NULL;
+	struct mlxsw_sp_mr_route *mr_route;
+	int err;
+
+	if (!mr_table->ops->is_route_valid(mr_table, mfc))
+		return -EINVAL;
+
+	/* Create a new route */
+	mr_route = mlxsw_sp_mr_route_create(mr_table, mfc);
+	if (IS_ERR(mr_route))
+		return PTR_ERR(mr_route);
+
+	/* Find any route with a matching key */
+	mr_orig_route = rhashtable_lookup_fast(&mr_table->route_ht,
+					       &mr_route->key,
+					       mlxsw_sp_mr_route_ht_params);
+	if (replace) {
+		/* On replace case, make the route point to the new route_priv.
+		 */
+		if (WARN_ON(!mr_orig_route)) {
+			err = -ENOENT;
+			goto err_no_orig_route;
+		}
+		mr_route->route_priv = mr_orig_route->route_priv;
+	} else if (mr_orig_route) {
+		/* On non replace case, if another route with the same key was
+		 * found, abort, as duplicate routes are used for proxy routes.
+		 */
+		dev_warn(mr_table->mlxsw_sp->bus_info->dev,
+			 "Offloading proxy routes is not supported.\n");
+		err = -EINVAL;
+		goto err_duplicate_route;
+	}
+
+	/* Write the route to the hardware */
+	err = mlxsw_sp_mr_route_write(mr_table, mr_route, replace);
+	if (err)
+		goto err_mr_route_write;
+
+	/* Put it in the table data-structures */
+	mutex_lock(&mr_table->route_list_lock);
+	list_add_tail(&mr_route->node, &mr_table->route_list);
+	mutex_unlock(&mr_table->route_list_lock);
+	err = rhashtable_insert_fast(&mr_table->route_ht,
+				     &mr_route->ht_node,
+				     mlxsw_sp_mr_route_ht_params);
+	if (err)
+		goto err_rhashtable_insert;
+
+	/* Destroy the original route */
+	if (replace) {
+		rhashtable_remove_fast(&mr_table->route_ht,
+				       &mr_orig_route->ht_node,
+				       mlxsw_sp_mr_route_ht_params);
+		list_del(&mr_orig_route->node);
+		mlxsw_sp_mr_route_destroy(mr_table, mr_orig_route);
+	}
+
+	mlxsw_sp_mr_mfc_offload_update(mr_route);
+	return 0;
+
+err_rhashtable_insert:
+	mutex_lock(&mr_table->route_list_lock);
+	list_del(&mr_route->node);
+	mutex_unlock(&mr_table->route_list_lock);
+	mlxsw_sp_mr_route_erase(mr_table, mr_route);
+err_mr_route_write:
+err_no_orig_route:
+err_duplicate_route:
+	mlxsw_sp_mr_route_destroy(mr_table, mr_route);
+	return err;
+}
+
+void mlxsw_sp_mr_route_del(struct mlxsw_sp_mr_table *mr_table,
+			   struct mr_mfc *mfc)
+{
+	struct mlxsw_sp_mr_route *mr_route;
+	struct mlxsw_sp_mr_route_key key;
+
+	mr_table->ops->key_create(mr_table, &key, mfc);
+	mr_route = rhashtable_lookup_fast(&mr_table->route_ht, &key,
+					  mlxsw_sp_mr_route_ht_params);
+	if (mr_route) {
+		mutex_lock(&mr_table->route_list_lock);
+		__mlxsw_sp_mr_route_del(mr_table, mr_route);
+		mutex_unlock(&mr_table->route_list_lock);
+	}
+}
+
+/* Should be called after the VIF struct is updated */
+static int
+mlxsw_sp_mr_route_ivif_resolve(struct mlxsw_sp_mr_table *mr_table,
+			       struct mlxsw_sp_mr_route_vif_entry *rve)
+{
+	struct mlxsw_sp *mlxsw_sp = mr_table->mlxsw_sp;
+	enum mlxsw_sp_mr_route_action route_action;
+	struct mlxsw_sp_mr *mr = mlxsw_sp->mr;
+	u16 irif_index;
+	int err;
+
+	route_action = mlxsw_sp_mr_route_action(rve->mr_route);
+	if (route_action == MLXSW_SP_MR_ROUTE_ACTION_TRAP)
+		return 0;
+
+	/* rve->mr_vif->rif is guaranteed to be valid at this stage */
+	irif_index = mlxsw_sp_rif_index(rve->mr_vif->rif);
+	err = mr->mr_ops->route_irif_update(mlxsw_sp, rve->mr_route->route_priv,
+					    irif_index);
+	if (err)
+		return err;
+
+	err = mr->mr_ops->route_action_update(mlxsw_sp,
+					      rve->mr_route->route_priv,
+					      route_action);
+	if (err)
+		/* No need to rollback here because the iRIF change only takes
+		 * place after the action has been updated.
+		 */
+		return err;
+
+	rve->mr_route->route_action = route_action;
+	mlxsw_sp_mr_mfc_offload_update(rve->mr_route);
+	return 0;
+}
+
+static void
+mlxsw_sp_mr_route_ivif_unresolve(struct mlxsw_sp_mr_table *mr_table,
+				 struct mlxsw_sp_mr_route_vif_entry *rve)
+{
+	struct mlxsw_sp *mlxsw_sp = mr_table->mlxsw_sp;
+	struct mlxsw_sp_mr *mr = mlxsw_sp->mr;
+
+	mr->mr_ops->route_action_update(mlxsw_sp, rve->mr_route->route_priv,
+					MLXSW_SP_MR_ROUTE_ACTION_TRAP);
+	rve->mr_route->route_action = MLXSW_SP_MR_ROUTE_ACTION_TRAP;
+	mlxsw_sp_mr_mfc_offload_update(rve->mr_route);
+}
+
+/* Should be called after the RIF struct is updated */
+static int
+mlxsw_sp_mr_route_evif_resolve(struct mlxsw_sp_mr_table *mr_table,
+			       struct mlxsw_sp_mr_route_vif_entry *rve)
+{
+	struct mlxsw_sp *mlxsw_sp = mr_table->mlxsw_sp;
+	enum mlxsw_sp_mr_route_action route_action;
+	struct mlxsw_sp_mr *mr = mlxsw_sp->mr;
+	u16 erif_index = 0;
+	int err;
+
+	/* Add the eRIF */
+	if (mlxsw_sp_mr_vif_valid(rve->mr_vif)) {
+		erif_index = mlxsw_sp_rif_index(rve->mr_vif->rif);
+		err = mr->mr_ops->route_erif_add(mlxsw_sp,
+						 rve->mr_route->route_priv,
+						 erif_index);
+		if (err)
+			return err;
+	}
+
+	/* Update the route action, as the new eVIF can be a tunnel or a pimreg
+	 * device which will require updating the action.
+	 */
+	route_action = mlxsw_sp_mr_route_action(rve->mr_route);
+	if (route_action != rve->mr_route->route_action) {
+		err = mr->mr_ops->route_action_update(mlxsw_sp,
+						      rve->mr_route->route_priv,
+						      route_action);
+		if (err)
+			goto err_route_action_update;
+	}
+
+	/* Update the minimum MTU */
+	if (rve->mr_vif->dev->mtu < rve->mr_route->min_mtu) {
+		rve->mr_route->min_mtu = rve->mr_vif->dev->mtu;
+		err = mr->mr_ops->route_min_mtu_update(mlxsw_sp,
+						       rve->mr_route->route_priv,
+						       rve->mr_route->min_mtu);
+		if (err)
+			goto err_route_min_mtu_update;
+	}
+
+	rve->mr_route->route_action = route_action;
+	mlxsw_sp_mr_mfc_offload_update(rve->mr_route);
+	return 0;
+
+err_route_min_mtu_update:
+	if (route_action != rve->mr_route->route_action)
+		mr->mr_ops->route_action_update(mlxsw_sp,
+						rve->mr_route->route_priv,
+						rve->mr_route->route_action);
+err_route_action_update:
+	if (mlxsw_sp_mr_vif_valid(rve->mr_vif))
+		mr->mr_ops->route_erif_del(mlxsw_sp, rve->mr_route->route_priv,
+					   erif_index);
+	return err;
+}
+
+/* Should be called before the RIF struct is updated */
+static void
+mlxsw_sp_mr_route_evif_unresolve(struct mlxsw_sp_mr_table *mr_table,
+				 struct mlxsw_sp_mr_route_vif_entry *rve)
+{
+	struct mlxsw_sp *mlxsw_sp = mr_table->mlxsw_sp;
+	enum mlxsw_sp_mr_route_action route_action;
+	struct mlxsw_sp_mr *mr = mlxsw_sp->mr;
+	u16 rifi;
+
+	/* If the unresolved RIF was not valid, no need to delete it */
+	if (!mlxsw_sp_mr_vif_valid(rve->mr_vif))
+		return;
+
+	/* Update the route action: if there is only one valid eVIF in the
+	 * route, set the action to trap as the VIF deletion will lead to zero
+	 * valid eVIFs. On any other case, use the mlxsw_sp_mr_route_action to
+	 * determine the route action.
+	 */
+	if (mlxsw_sp_mr_route_valid_evifs_num(rve->mr_route) == 1)
+		route_action = MLXSW_SP_MR_ROUTE_ACTION_TRAP;
+	else
+		route_action = mlxsw_sp_mr_route_action(rve->mr_route);
+	if (route_action != rve->mr_route->route_action)
+		mr->mr_ops->route_action_update(mlxsw_sp,
+						rve->mr_route->route_priv,
+						route_action);
+
+	/* Delete the erif from the route */
+	rifi = mlxsw_sp_rif_index(rve->mr_vif->rif);
+	mr->mr_ops->route_erif_del(mlxsw_sp, rve->mr_route->route_priv, rifi);
+	rve->mr_route->route_action = route_action;
+	mlxsw_sp_mr_mfc_offload_update(rve->mr_route);
+}
+
+static int mlxsw_sp_mr_vif_resolve(struct mlxsw_sp_mr_table *mr_table,
+				   struct net_device *dev,
+				   struct mlxsw_sp_mr_vif *mr_vif,
+				   unsigned long vif_flags,
+				   const struct mlxsw_sp_rif *rif)
+{
+	struct mlxsw_sp_mr_route_vif_entry *irve, *erve;
+	int err;
+
+	/* Update the VIF */
+	mr_vif->dev = dev;
+	mr_vif->rif = rif;
+	mr_vif->vif_flags = vif_flags;
+
+	/* Update all routes where this VIF is used as an unresolved iRIF */
+	list_for_each_entry(irve, &mr_vif->route_ivif_list, vif_node) {
+		err = mlxsw_sp_mr_route_ivif_resolve(mr_table, irve);
+		if (err)
+			goto err_irif_unresolve;
+	}
+
+	/* Update all routes where this VIF is used as an unresolved eRIF */
+	list_for_each_entry(erve, &mr_vif->route_evif_list, vif_node) {
+		err = mlxsw_sp_mr_route_evif_resolve(mr_table, erve);
+		if (err)
+			goto err_erif_unresolve;
+	}
+	return 0;
+
+err_erif_unresolve:
+	list_for_each_entry_continue_reverse(erve, &mr_vif->route_evif_list,
+					     vif_node)
+		mlxsw_sp_mr_route_evif_unresolve(mr_table, erve);
+err_irif_unresolve:
+	list_for_each_entry_continue_reverse(irve, &mr_vif->route_ivif_list,
+					     vif_node)
+		mlxsw_sp_mr_route_ivif_unresolve(mr_table, irve);
+	mr_vif->rif = NULL;
+	return err;
+}
+
+static void mlxsw_sp_mr_vif_unresolve(struct mlxsw_sp_mr_table *mr_table,
+				      struct net_device *dev,
+				      struct mlxsw_sp_mr_vif *mr_vif)
+{
+	struct mlxsw_sp_mr_route_vif_entry *rve;
+
+	/* Update all routes where this VIF is used as an unresolved eRIF */
+	list_for_each_entry(rve, &mr_vif->route_evif_list, vif_node)
+		mlxsw_sp_mr_route_evif_unresolve(mr_table, rve);
+
+	/* Update all routes where this VIF is used as an unresolved iRIF */
+	list_for_each_entry(rve, &mr_vif->route_ivif_list, vif_node)
+		mlxsw_sp_mr_route_ivif_unresolve(mr_table, rve);
+
+	/* Update the VIF */
+	mr_vif->dev = dev;
+	mr_vif->rif = NULL;
+}
+
+int mlxsw_sp_mr_vif_add(struct mlxsw_sp_mr_table *mr_table,
+			struct net_device *dev, vifi_t vif_index,
+			unsigned long vif_flags, const struct mlxsw_sp_rif *rif)
+{
+	struct mlxsw_sp_mr_vif *mr_vif = &mr_table->vifs[vif_index];
+
+	if (WARN_ON(vif_index >= MAXVIFS))
+		return -EINVAL;
+	if (mr_vif->dev)
+		return -EEXIST;
+	return mlxsw_sp_mr_vif_resolve(mr_table, dev, mr_vif, vif_flags, rif);
+}
+
+void mlxsw_sp_mr_vif_del(struct mlxsw_sp_mr_table *mr_table, vifi_t vif_index)
+{
+	struct mlxsw_sp_mr_vif *mr_vif = &mr_table->vifs[vif_index];
+
+	if (WARN_ON(vif_index >= MAXVIFS))
+		return;
+	if (WARN_ON(!mr_vif->dev))
+		return;
+	mlxsw_sp_mr_vif_unresolve(mr_table, NULL, mr_vif);
+}
+
+static struct mlxsw_sp_mr_vif *
+mlxsw_sp_mr_dev_vif_lookup(struct mlxsw_sp_mr_table *mr_table,
+			   const struct net_device *dev)
+{
+	vifi_t vif_index;
+
+	for (vif_index = 0; vif_index < MAXVIFS; vif_index++)
+		if (mr_table->vifs[vif_index].dev == dev)
+			return &mr_table->vifs[vif_index];
+	return NULL;
+}
+
+int mlxsw_sp_mr_rif_add(struct mlxsw_sp_mr_table *mr_table,
+			const struct mlxsw_sp_rif *rif)
+{
+	const struct net_device *rif_dev = mlxsw_sp_rif_dev(rif);
+	struct mlxsw_sp_mr_vif *mr_vif;
+
+	if (!rif_dev)
+		return 0;
+
+	mr_vif = mlxsw_sp_mr_dev_vif_lookup(mr_table, rif_dev);
+	if (!mr_vif)
+		return 0;
+	return mlxsw_sp_mr_vif_resolve(mr_table, mr_vif->dev, mr_vif,
+				       mr_vif->vif_flags, rif);
+}
+
+void mlxsw_sp_mr_rif_del(struct mlxsw_sp_mr_table *mr_table,
+			 const struct mlxsw_sp_rif *rif)
+{
+	const struct net_device *rif_dev = mlxsw_sp_rif_dev(rif);
+	struct mlxsw_sp_mr_vif *mr_vif;
+
+	if (!rif_dev)
+		return;
+
+	mr_vif = mlxsw_sp_mr_dev_vif_lookup(mr_table, rif_dev);
+	if (!mr_vif)
+		return;
+	mlxsw_sp_mr_vif_unresolve(mr_table, mr_vif->dev, mr_vif);
+}
+
+void mlxsw_sp_mr_rif_mtu_update(struct mlxsw_sp_mr_table *mr_table,
+				const struct mlxsw_sp_rif *rif, int mtu)
+{
+	const struct net_device *rif_dev = mlxsw_sp_rif_dev(rif);
+	struct mlxsw_sp *mlxsw_sp = mr_table->mlxsw_sp;
+	struct mlxsw_sp_mr_route_vif_entry *rve;
+	struct mlxsw_sp_mr *mr = mlxsw_sp->mr;
+	struct mlxsw_sp_mr_vif *mr_vif;
+
+	if (!rif_dev)
+		return;
+
+	/* Search for a VIF that use that RIF */
+	mr_vif = mlxsw_sp_mr_dev_vif_lookup(mr_table, rif_dev);
+	if (!mr_vif)
+		return;
+
+	/* Update all the routes that uses that VIF as eVIF */
+	list_for_each_entry(rve, &mr_vif->route_evif_list, vif_node) {
+		if (mtu < rve->mr_route->min_mtu) {
+			rve->mr_route->min_mtu = mtu;
+			mr->mr_ops->route_min_mtu_update(mlxsw_sp,
+							 rve->mr_route->route_priv,
+							 mtu);
+		}
+	}
+}
+
+/* Protocol specific functions */
+static bool
+mlxsw_sp_mr_route4_validate(const struct mlxsw_sp_mr_table *mr_table,
+			    const struct mr_mfc *c)
+{
+	struct mfc_cache *mfc = (struct mfc_cache *) c;
+
+	/* If the route is a (*,*) route, abort, as these kind of routes are
+	 * used for proxy routes.
+	 */
+	if (mfc->mfc_origin == htonl(INADDR_ANY) &&
+	    mfc->mfc_mcastgrp == htonl(INADDR_ANY)) {
+		dev_warn(mr_table->mlxsw_sp->bus_info->dev,
+			 "Offloading proxy routes is not supported.\n");
+		return false;
+	}
+	return true;
+}
+
+static void mlxsw_sp_mr_route4_key(struct mlxsw_sp_mr_table *mr_table,
+				   struct mlxsw_sp_mr_route_key *key,
+				   struct mr_mfc *c)
+{
+	const struct mfc_cache *mfc = (struct mfc_cache *) c;
+	bool starg;
+
+	starg = (mfc->mfc_origin == htonl(INADDR_ANY));
+
+	memset(key, 0, sizeof(*key));
+	key->vrid = mr_table->vr_id;
+	key->proto = MLXSW_SP_L3_PROTO_IPV4;
+	key->group.addr4 = mfc->mfc_mcastgrp;
+	key->group_mask.addr4 = htonl(0xffffffff);
+	key->source.addr4 = mfc->mfc_origin;
+	key->source_mask.addr4 = htonl(starg ? 0 : 0xffffffff);
+}
+
+static bool mlxsw_sp_mr_route4_starg(const struct mlxsw_sp_mr_table *mr_table,
+				     const struct mlxsw_sp_mr_route *mr_route)
+{
+	return mr_route->key.source_mask.addr4 == htonl(INADDR_ANY);
+}
+
+static bool mlxsw_sp_mr_vif4_is_regular(const struct mlxsw_sp_mr_vif *vif)
+{
+	return !(vif->vif_flags & (VIFF_TUNNEL | VIFF_REGISTER));
+}
+
+static bool
+mlxsw_sp_mr_route6_validate(const struct mlxsw_sp_mr_table *mr_table,
+			    const struct mr_mfc *c)
+{
+	struct mfc6_cache *mfc = (struct mfc6_cache *) c;
+
+	/* If the route is a (*,*) route, abort, as these kind of routes are
+	 * used for proxy routes.
+	 */
+	if (ipv6_addr_any(&mfc->mf6c_origin) &&
+	    ipv6_addr_any(&mfc->mf6c_mcastgrp)) {
+		dev_warn(mr_table->mlxsw_sp->bus_info->dev,
+			 "Offloading proxy routes is not supported.\n");
+		return false;
+	}
+	return true;
+}
+
+static void mlxsw_sp_mr_route6_key(struct mlxsw_sp_mr_table *mr_table,
+				   struct mlxsw_sp_mr_route_key *key,
+				   struct mr_mfc *c)
+{
+	const struct mfc6_cache *mfc = (struct mfc6_cache *) c;
+
+	memset(key, 0, sizeof(*key));
+	key->vrid = mr_table->vr_id;
+	key->proto = MLXSW_SP_L3_PROTO_IPV6;
+	key->group.addr6 = mfc->mf6c_mcastgrp;
+	memset(&key->group_mask.addr6, 0xff, sizeof(key->group_mask.addr6));
+	key->source.addr6 = mfc->mf6c_origin;
+	if (!ipv6_addr_any(&mfc->mf6c_origin))
+		memset(&key->source_mask.addr6, 0xff,
+		       sizeof(key->source_mask.addr6));
+}
+
+static bool mlxsw_sp_mr_route6_starg(const struct mlxsw_sp_mr_table *mr_table,
+				     const struct mlxsw_sp_mr_route *mr_route)
+{
+	return ipv6_addr_any(&mr_route->key.source_mask.addr6);
+}
+
+static bool mlxsw_sp_mr_vif6_is_regular(const struct mlxsw_sp_mr_vif *vif)
+{
+	return !(vif->vif_flags & MIFF_REGISTER);
+}
+
+static struct
+mlxsw_sp_mr_vif_ops mlxsw_sp_mr_vif_ops_arr[] = {
+	{
+		.is_regular = mlxsw_sp_mr_vif4_is_regular,
+	},
+	{
+		.is_regular = mlxsw_sp_mr_vif6_is_regular,
+	},
+};
+
+static struct
+mlxsw_sp_mr_table_ops mlxsw_sp_mr_table_ops_arr[] = {
+	{
+		.is_route_valid = mlxsw_sp_mr_route4_validate,
+		.key_create = mlxsw_sp_mr_route4_key,
+		.is_route_starg = mlxsw_sp_mr_route4_starg,
+	},
+	{
+		.is_route_valid = mlxsw_sp_mr_route6_validate,
+		.key_create = mlxsw_sp_mr_route6_key,
+		.is_route_starg = mlxsw_sp_mr_route6_starg,
+	},
+
+};
+
+struct mlxsw_sp_mr_table *mlxsw_sp_mr_table_create(struct mlxsw_sp *mlxsw_sp,
+						   u32 vr_id,
+						   enum mlxsw_sp_l3proto proto)
+{
+	struct mlxsw_sp_mr_route_params catchall_route_params = {
+		.prio = MLXSW_SP_MR_ROUTE_PRIO_CATCHALL,
+		.key = {
+			.vrid = vr_id,
+			.proto = proto,
+		},
+		.value = {
+			.route_action = MLXSW_SP_MR_ROUTE_ACTION_TRAP,
+		}
+	};
+	struct mlxsw_sp_mr *mr = mlxsw_sp->mr;
+	struct mlxsw_sp_mr_table *mr_table;
+	int err;
+	int i;
+
+	mr_table = kzalloc(sizeof(*mr_table) + mr->mr_ops->route_priv_size,
+			   GFP_KERNEL);
+	if (!mr_table)
+		return ERR_PTR(-ENOMEM);
+
+	mr_table->vr_id = vr_id;
+	mr_table->mlxsw_sp = mlxsw_sp;
+	mr_table->proto = proto;
+	mr_table->ops = &mlxsw_sp_mr_table_ops_arr[proto];
+	INIT_LIST_HEAD(&mr_table->route_list);
+	mutex_init(&mr_table->route_list_lock);
+
+	err = rhashtable_init(&mr_table->route_ht,
+			      &mlxsw_sp_mr_route_ht_params);
+	if (err)
+		goto err_route_rhashtable_init;
+
+	for (i = 0; i < MAXVIFS; i++) {
+		INIT_LIST_HEAD(&mr_table->vifs[i].route_evif_list);
+		INIT_LIST_HEAD(&mr_table->vifs[i].route_ivif_list);
+		mr_table->vifs[i].ops = &mlxsw_sp_mr_vif_ops_arr[proto];
+	}
+
+	err = mr->mr_ops->route_create(mlxsw_sp, mr->priv,
+				       mr_table->catchall_route_priv,
+				       &catchall_route_params);
+	if (err)
+		goto err_ops_route_create;
+	mutex_lock(&mr->table_list_lock);
+	list_add_tail(&mr_table->node, &mr->table_list);
+	mutex_unlock(&mr->table_list_lock);
+	return mr_table;
+
+err_ops_route_create:
+	rhashtable_destroy(&mr_table->route_ht);
+err_route_rhashtable_init:
+	mutex_destroy(&mr_table->route_list_lock);
+	kfree(mr_table);
+	return ERR_PTR(err);
+}
+
+void mlxsw_sp_mr_table_destroy(struct mlxsw_sp_mr_table *mr_table)
+{
+	struct mlxsw_sp *mlxsw_sp = mr_table->mlxsw_sp;
+	struct mlxsw_sp_mr *mr = mlxsw_sp->mr;
+
+	WARN_ON(!mlxsw_sp_mr_table_empty(mr_table));
+	mutex_lock(&mr->table_list_lock);
+	list_del(&mr_table->node);
+	mutex_unlock(&mr->table_list_lock);
+	mr->mr_ops->route_destroy(mlxsw_sp, mr->priv,
+				  &mr_table->catchall_route_priv);
+	rhashtable_destroy(&mr_table->route_ht);
+	mutex_destroy(&mr_table->route_list_lock);
+	kfree(mr_table);
+}
+
+void mlxsw_sp_mr_table_flush(struct mlxsw_sp_mr_table *mr_table)
+{
+	struct mlxsw_sp_mr_route *mr_route, *tmp;
+	int i;
+
+	mutex_lock(&mr_table->route_list_lock);
+	list_for_each_entry_safe(mr_route, tmp, &mr_table->route_list, node)
+		__mlxsw_sp_mr_route_del(mr_table, mr_route);
+	mutex_unlock(&mr_table->route_list_lock);
+
+	for (i = 0; i < MAXVIFS; i++) {
+		mr_table->vifs[i].dev = NULL;
+		mr_table->vifs[i].rif = NULL;
+	}
+}
+
+bool mlxsw_sp_mr_table_empty(const struct mlxsw_sp_mr_table *mr_table)
+{
+	int i;
+
+	for (i = 0; i < MAXVIFS; i++)
+		if (mr_table->vifs[i].dev)
+			return false;
+	return list_empty(&mr_table->route_list);
+}
+
+static void mlxsw_sp_mr_route_stats_update(struct mlxsw_sp *mlxsw_sp,
+					   struct mlxsw_sp_mr_route *mr_route)
+{
+	struct mlxsw_sp_mr *mr = mlxsw_sp->mr;
+	u64 packets, bytes;
+
+	if (mr_route->route_action == MLXSW_SP_MR_ROUTE_ACTION_TRAP)
+		return;
+
+	mr->mr_ops->route_stats(mlxsw_sp, mr_route->route_priv, &packets,
+				&bytes);
+
+	if (mr_route->mfc->mfc_un.res.pkt != packets)
+		mr_route->mfc->mfc_un.res.lastuse = jiffies;
+	mr_route->mfc->mfc_un.res.pkt = packets;
+	mr_route->mfc->mfc_un.res.bytes = bytes;
+}
+
+static void mlxsw_sp_mr_stats_update(struct work_struct *work)
+{
+	struct mlxsw_sp_mr *mr = container_of(work, struct mlxsw_sp_mr,
+					      stats_update_dw.work);
+	struct mlxsw_sp_mr_table *mr_table;
+	struct mlxsw_sp_mr_route *mr_route;
+	unsigned long interval;
+
+	mutex_lock(&mr->table_list_lock);
+	list_for_each_entry(mr_table, &mr->table_list, node) {
+		mutex_lock(&mr_table->route_list_lock);
+		list_for_each_entry(mr_route, &mr_table->route_list, node)
+			mlxsw_sp_mr_route_stats_update(mr_table->mlxsw_sp,
+						       mr_route);
+		mutex_unlock(&mr_table->route_list_lock);
+	}
+	mutex_unlock(&mr->table_list_lock);
+
+	interval = msecs_to_jiffies(MLXSW_SP_MR_ROUTES_COUNTER_UPDATE_INTERVAL);
+	mlxsw_core_schedule_dw(&mr->stats_update_dw, interval);
+}
+
+int mlxsw_sp_mr_init(struct mlxsw_sp *mlxsw_sp,
+		     const struct mlxsw_sp_mr_ops *mr_ops)
+{
+	struct mlxsw_sp_mr *mr;
+	unsigned long interval;
+	int err;
+
+	mr = kzalloc(sizeof(*mr) + mr_ops->priv_size, GFP_KERNEL);
+	if (!mr)
+		return -ENOMEM;
+	mr->mr_ops = mr_ops;
+	mlxsw_sp->mr = mr;
+	INIT_LIST_HEAD(&mr->table_list);
+	mutex_init(&mr->table_list_lock);
+
+	err = mr_ops->init(mlxsw_sp, mr->priv);
+	if (err)
+		goto err;
+
+	/* Create the delayed work for counter updates */
+	INIT_DELAYED_WORK(&mr->stats_update_dw, mlxsw_sp_mr_stats_update);
+	interval = msecs_to_jiffies(MLXSW_SP_MR_ROUTES_COUNTER_UPDATE_INTERVAL);
+	mlxsw_core_schedule_dw(&mr->stats_update_dw, interval);
+	return 0;
+err:
+	mutex_destroy(&mr->table_list_lock);
+	kfree(mr);
+	return err;
+}
+
+void mlxsw_sp_mr_fini(struct mlxsw_sp *mlxsw_sp)
+{
+	struct mlxsw_sp_mr *mr = mlxsw_sp->mr;
+
+	cancel_delayed_work_sync(&mr->stats_update_dw);
+	mr->mr_ops->fini(mlxsw_sp, mr->priv);
+	mutex_destroy(&mr->table_list_lock);
+	kfree(mr);
+}
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_mr.h b/drivers/net/ethernet/mellanox/mlxsw/spectrum_mr.h
new file mode 100644
index 000000000..3cde3671f
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_mr.h
@@ -0,0 +1,95 @@
+/* SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0 */
+/* Copyright (c) 2017-2018 Mellanox Technologies. All rights reserved */
+
+#ifndef _MLXSW_SPECTRUM_MCROUTER_H
+#define _MLXSW_SPECTRUM_MCROUTER_H
+
+#include <linux/mroute.h>
+#include <linux/mroute6.h>
+#include "spectrum_router.h"
+#include "spectrum.h"
+
+enum mlxsw_sp_mr_route_action {
+	MLXSW_SP_MR_ROUTE_ACTION_FORWARD,
+	MLXSW_SP_MR_ROUTE_ACTION_TRAP,
+	MLXSW_SP_MR_ROUTE_ACTION_TRAP_AND_FORWARD,
+};
+
+struct mlxsw_sp_mr_route_key {
+	int vrid;
+	enum mlxsw_sp_l3proto proto;
+	union mlxsw_sp_l3addr group;
+	union mlxsw_sp_l3addr group_mask;
+	union mlxsw_sp_l3addr source;
+	union mlxsw_sp_l3addr source_mask;
+};
+
+struct mlxsw_sp_mr_route_info {
+	enum mlxsw_sp_mr_route_action route_action;
+	u16 irif_index;
+	u16 *erif_indices;
+	size_t erif_num;
+	u16 min_mtu;
+};
+
+struct mlxsw_sp_mr_route_params {
+	struct mlxsw_sp_mr_route_key key;
+	struct mlxsw_sp_mr_route_info value;
+	enum mlxsw_sp_mr_route_prio prio;
+};
+
+struct mlxsw_sp_mr_ops {
+	int priv_size;
+	int route_priv_size;
+	int (*init)(struct mlxsw_sp *mlxsw_sp, void *priv);
+	int (*route_create)(struct mlxsw_sp *mlxsw_sp, void *priv,
+			    void *route_priv,
+			    struct mlxsw_sp_mr_route_params *route_params);
+	int (*route_update)(struct mlxsw_sp *mlxsw_sp, void *route_priv,
+			    struct mlxsw_sp_mr_route_info *route_info);
+	int (*route_stats)(struct mlxsw_sp *mlxsw_sp, void *route_priv,
+			   u64 *packets, u64 *bytes);
+	int (*route_action_update)(struct mlxsw_sp *mlxsw_sp, void *route_priv,
+				   enum mlxsw_sp_mr_route_action route_action);
+	int (*route_min_mtu_update)(struct mlxsw_sp *mlxsw_sp, void *route_priv,
+				    u16 min_mtu);
+	int (*route_irif_update)(struct mlxsw_sp *mlxsw_sp, void *route_priv,
+				 u16 irif_index);
+	int (*route_erif_add)(struct mlxsw_sp *mlxsw_sp, void *route_priv,
+			      u16 erif_index);
+	int (*route_erif_del)(struct mlxsw_sp *mlxsw_sp, void *route_priv,
+			      u16 erif_index);
+	void (*route_destroy)(struct mlxsw_sp *mlxsw_sp, void *priv,
+			      void *route_priv);
+	void (*fini)(struct mlxsw_sp *mlxsw_sp, void *priv);
+};
+
+struct mlxsw_sp_mr;
+struct mlxsw_sp_mr_table;
+
+int mlxsw_sp_mr_init(struct mlxsw_sp *mlxsw_sp,
+		     const struct mlxsw_sp_mr_ops *mr_ops);
+void mlxsw_sp_mr_fini(struct mlxsw_sp *mlxsw_sp);
+int mlxsw_sp_mr_route_add(struct mlxsw_sp_mr_table *mr_table,
+			  struct mr_mfc *mfc, bool replace);
+void mlxsw_sp_mr_route_del(struct mlxsw_sp_mr_table *mr_table,
+			   struct mr_mfc *mfc);
+int mlxsw_sp_mr_vif_add(struct mlxsw_sp_mr_table *mr_table,
+			struct net_device *dev, vifi_t vif_index,
+			unsigned long vif_flags,
+			const struct mlxsw_sp_rif *rif);
+void mlxsw_sp_mr_vif_del(struct mlxsw_sp_mr_table *mr_table, vifi_t vif_index);
+int mlxsw_sp_mr_rif_add(struct mlxsw_sp_mr_table *mr_table,
+			const struct mlxsw_sp_rif *rif);
+void mlxsw_sp_mr_rif_del(struct mlxsw_sp_mr_table *mr_table,
+			 const struct mlxsw_sp_rif *rif);
+void mlxsw_sp_mr_rif_mtu_update(struct mlxsw_sp_mr_table *mr_table,
+				const struct mlxsw_sp_rif *rif, int mtu);
+struct mlxsw_sp_mr_table *mlxsw_sp_mr_table_create(struct mlxsw_sp *mlxsw_sp,
+						   u32 tb_id,
+						   enum mlxsw_sp_l3proto proto);
+void mlxsw_sp_mr_table_destroy(struct mlxsw_sp_mr_table *mr_table);
+void mlxsw_sp_mr_table_flush(struct mlxsw_sp_mr_table *mr_table);
+bool mlxsw_sp_mr_table_empty(const struct mlxsw_sp_mr_table *mr_table);
+
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_mr_tcam.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_mr_tcam.c
new file mode 100644
index 000000000..221aa6a47
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_mr_tcam.c
@@ -0,0 +1,615 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2017-2018 Mellanox Technologies. All rights reserved */
+
+#include <linux/kernel.h>
+#include <linux/list.h>
+#include <linux/netdevice.h>
+
+#include "spectrum_mr_tcam.h"
+#include "reg.h"
+#include "spectrum.h"
+#include "core_acl_flex_actions.h"
+#include "spectrum_mr.h"
+
+struct mlxsw_sp_mr_tcam {
+	void *priv;
+};
+
+/* This struct maps to one RIGR2 register entry */
+struct mlxsw_sp_mr_erif_sublist {
+	struct list_head list;
+	u32 rigr2_kvdl_index;
+	int num_erifs;
+	u16 erif_indices[MLXSW_REG_RIGR2_MAX_ERIFS];
+	bool synced;
+};
+
+struct mlxsw_sp_mr_tcam_erif_list {
+	struct list_head erif_sublists;
+	u32 kvdl_index;
+};
+
+static bool
+mlxsw_sp_mr_erif_sublist_full(struct mlxsw_sp *mlxsw_sp,
+			      struct mlxsw_sp_mr_erif_sublist *erif_sublist)
+{
+	int erif_list_entries = MLXSW_CORE_RES_GET(mlxsw_sp->core,
+						   MC_ERIF_LIST_ENTRIES);
+
+	return erif_sublist->num_erifs == erif_list_entries;
+}
+
+static void
+mlxsw_sp_mr_erif_list_init(struct mlxsw_sp_mr_tcam_erif_list *erif_list)
+{
+	INIT_LIST_HEAD(&erif_list->erif_sublists);
+}
+
+static struct mlxsw_sp_mr_erif_sublist *
+mlxsw_sp_mr_erif_sublist_create(struct mlxsw_sp *mlxsw_sp,
+				struct mlxsw_sp_mr_tcam_erif_list *erif_list)
+{
+	struct mlxsw_sp_mr_erif_sublist *erif_sublist;
+	int err;
+
+	erif_sublist = kzalloc(sizeof(*erif_sublist), GFP_KERNEL);
+	if (!erif_sublist)
+		return ERR_PTR(-ENOMEM);
+	err = mlxsw_sp_kvdl_alloc(mlxsw_sp, MLXSW_SP_KVDL_ENTRY_TYPE_MCRIGR,
+				  1, &erif_sublist->rigr2_kvdl_index);
+	if (err) {
+		kfree(erif_sublist);
+		return ERR_PTR(err);
+	}
+
+	list_add_tail(&erif_sublist->list, &erif_list->erif_sublists);
+	return erif_sublist;
+}
+
+static void
+mlxsw_sp_mr_erif_sublist_destroy(struct mlxsw_sp *mlxsw_sp,
+				 struct mlxsw_sp_mr_erif_sublist *erif_sublist)
+{
+	list_del(&erif_sublist->list);
+	mlxsw_sp_kvdl_free(mlxsw_sp, MLXSW_SP_KVDL_ENTRY_TYPE_MCRIGR,
+			   1, erif_sublist->rigr2_kvdl_index);
+	kfree(erif_sublist);
+}
+
+static int
+mlxsw_sp_mr_erif_list_add(struct mlxsw_sp *mlxsw_sp,
+			  struct mlxsw_sp_mr_tcam_erif_list *erif_list,
+			  u16 erif_index)
+{
+	struct mlxsw_sp_mr_erif_sublist *sublist;
+
+	/* If either there is no erif_entry or the last one is full, allocate a
+	 * new one.
+	 */
+	if (list_empty(&erif_list->erif_sublists)) {
+		sublist = mlxsw_sp_mr_erif_sublist_create(mlxsw_sp, erif_list);
+		if (IS_ERR(sublist))
+			return PTR_ERR(sublist);
+		erif_list->kvdl_index = sublist->rigr2_kvdl_index;
+	} else {
+		sublist = list_last_entry(&erif_list->erif_sublists,
+					  struct mlxsw_sp_mr_erif_sublist,
+					  list);
+		sublist->synced = false;
+		if (mlxsw_sp_mr_erif_sublist_full(mlxsw_sp, sublist)) {
+			sublist = mlxsw_sp_mr_erif_sublist_create(mlxsw_sp,
+								  erif_list);
+			if (IS_ERR(sublist))
+				return PTR_ERR(sublist);
+		}
+	}
+
+	/* Add the eRIF to the last entry's last index */
+	sublist->erif_indices[sublist->num_erifs++] = erif_index;
+	return 0;
+}
+
+static void
+mlxsw_sp_mr_erif_list_flush(struct mlxsw_sp *mlxsw_sp,
+			    struct mlxsw_sp_mr_tcam_erif_list *erif_list)
+{
+	struct mlxsw_sp_mr_erif_sublist *erif_sublist, *tmp;
+
+	list_for_each_entry_safe(erif_sublist, tmp, &erif_list->erif_sublists,
+				 list)
+		mlxsw_sp_mr_erif_sublist_destroy(mlxsw_sp, erif_sublist);
+}
+
+static int
+mlxsw_sp_mr_erif_list_commit(struct mlxsw_sp *mlxsw_sp,
+			     struct mlxsw_sp_mr_tcam_erif_list *erif_list)
+{
+	struct mlxsw_sp_mr_erif_sublist *curr_sublist;
+	char rigr2_pl[MLXSW_REG_RIGR2_LEN];
+	int err;
+	int i;
+
+	list_for_each_entry(curr_sublist, &erif_list->erif_sublists, list) {
+		if (curr_sublist->synced)
+			continue;
+
+		/* If the sublist is not the last one, pack the next index */
+		if (list_is_last(&curr_sublist->list,
+				 &erif_list->erif_sublists)) {
+			mlxsw_reg_rigr2_pack(rigr2_pl,
+					     curr_sublist->rigr2_kvdl_index,
+					     false, 0);
+		} else {
+			struct mlxsw_sp_mr_erif_sublist *next_sublist;
+
+			next_sublist = list_next_entry(curr_sublist, list);
+			mlxsw_reg_rigr2_pack(rigr2_pl,
+					     curr_sublist->rigr2_kvdl_index,
+					     true,
+					     next_sublist->rigr2_kvdl_index);
+		}
+
+		/* Pack all the erifs */
+		for (i = 0; i < curr_sublist->num_erifs; i++) {
+			u16 erif_index = curr_sublist->erif_indices[i];
+
+			mlxsw_reg_rigr2_erif_entry_pack(rigr2_pl, i, true,
+							erif_index);
+		}
+
+		/* Write the entry */
+		err = mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(rigr2),
+				      rigr2_pl);
+		if (err)
+			/* No need of a rollback here because this
+			 * hardware entry should not be pointed yet.
+			 */
+			return err;
+		curr_sublist->synced = true;
+	}
+	return 0;
+}
+
+static void mlxsw_sp_mr_erif_list_move(struct mlxsw_sp_mr_tcam_erif_list *to,
+				       struct mlxsw_sp_mr_tcam_erif_list *from)
+{
+	list_splice(&from->erif_sublists, &to->erif_sublists);
+	to->kvdl_index = from->kvdl_index;
+}
+
+struct mlxsw_sp_mr_tcam_route {
+	struct mlxsw_sp_mr_tcam_erif_list erif_list;
+	struct mlxsw_afa_block *afa_block;
+	u32 counter_index;
+	enum mlxsw_sp_mr_route_action action;
+	struct mlxsw_sp_mr_route_key key;
+	u16 irif_index;
+	u16 min_mtu;
+	void *priv;
+};
+
+static struct mlxsw_afa_block *
+mlxsw_sp_mr_tcam_afa_block_create(struct mlxsw_sp *mlxsw_sp,
+				  enum mlxsw_sp_mr_route_action route_action,
+				  u16 irif_index, u32 counter_index,
+				  u16 min_mtu,
+				  struct mlxsw_sp_mr_tcam_erif_list *erif_list)
+{
+	struct mlxsw_afa_block *afa_block;
+	int err;
+
+	afa_block = mlxsw_afa_block_create(mlxsw_sp->afa);
+	if (IS_ERR(afa_block))
+		return afa_block;
+
+	err = mlxsw_afa_block_append_allocated_counter(afa_block,
+						       counter_index);
+	if (err)
+		goto err;
+
+	switch (route_action) {
+	case MLXSW_SP_MR_ROUTE_ACTION_TRAP:
+		err = mlxsw_afa_block_append_trap(afa_block,
+						  MLXSW_TRAP_ID_ACL1);
+		if (err)
+			goto err;
+		break;
+	case MLXSW_SP_MR_ROUTE_ACTION_TRAP_AND_FORWARD:
+	case MLXSW_SP_MR_ROUTE_ACTION_FORWARD:
+		/* If we are about to append a multicast router action, commit
+		 * the erif_list.
+		 */
+		err = mlxsw_sp_mr_erif_list_commit(mlxsw_sp, erif_list);
+		if (err)
+			goto err;
+
+		err = mlxsw_afa_block_append_mcrouter(afa_block, irif_index,
+						      min_mtu, false,
+						      erif_list->kvdl_index);
+		if (err)
+			goto err;
+
+		if (route_action == MLXSW_SP_MR_ROUTE_ACTION_TRAP_AND_FORWARD) {
+			err = mlxsw_afa_block_append_trap_and_forward(afa_block,
+								      MLXSW_TRAP_ID_ACL2);
+			if (err)
+				goto err;
+		}
+		break;
+	default:
+		err = -EINVAL;
+		goto err;
+	}
+
+	err = mlxsw_afa_block_commit(afa_block);
+	if (err)
+		goto err;
+	return afa_block;
+err:
+	mlxsw_afa_block_destroy(afa_block);
+	return ERR_PTR(err);
+}
+
+static void
+mlxsw_sp_mr_tcam_afa_block_destroy(struct mlxsw_afa_block *afa_block)
+{
+	mlxsw_afa_block_destroy(afa_block);
+}
+
+static int
+mlxsw_sp_mr_tcam_erif_populate(struct mlxsw_sp *mlxsw_sp,
+			       struct mlxsw_sp_mr_tcam_erif_list *erif_list,
+			       struct mlxsw_sp_mr_route_info *route_info)
+{
+	int err;
+	int i;
+
+	for (i = 0; i < route_info->erif_num; i++) {
+		u16 erif_index = route_info->erif_indices[i];
+
+		err = mlxsw_sp_mr_erif_list_add(mlxsw_sp, erif_list,
+						erif_index);
+		if (err)
+			return err;
+	}
+	return 0;
+}
+
+static int
+mlxsw_sp_mr_tcam_route_create(struct mlxsw_sp *mlxsw_sp, void *priv,
+			      void *route_priv,
+			      struct mlxsw_sp_mr_route_params *route_params)
+{
+	const struct mlxsw_sp_mr_tcam_ops *ops = mlxsw_sp->mr_tcam_ops;
+	struct mlxsw_sp_mr_tcam_route *route = route_priv;
+	struct mlxsw_sp_mr_tcam *mr_tcam = priv;
+	int err;
+
+	route->key = route_params->key;
+	route->irif_index = route_params->value.irif_index;
+	route->min_mtu = route_params->value.min_mtu;
+	route->action = route_params->value.route_action;
+
+	/* Create the egress RIFs list */
+	mlxsw_sp_mr_erif_list_init(&route->erif_list);
+	err = mlxsw_sp_mr_tcam_erif_populate(mlxsw_sp, &route->erif_list,
+					     &route_params->value);
+	if (err)
+		goto err_erif_populate;
+
+	/* Create the flow counter */
+	err = mlxsw_sp_flow_counter_alloc(mlxsw_sp, &route->counter_index);
+	if (err)
+		goto err_counter_alloc;
+
+	/* Create the flexible action block */
+	route->afa_block = mlxsw_sp_mr_tcam_afa_block_create(mlxsw_sp,
+							     route->action,
+							     route->irif_index,
+							     route->counter_index,
+							     route->min_mtu,
+							     &route->erif_list);
+	if (IS_ERR(route->afa_block)) {
+		err = PTR_ERR(route->afa_block);
+		goto err_afa_block_create;
+	}
+
+	route->priv = kzalloc(ops->route_priv_size, GFP_KERNEL);
+	if (!route->priv) {
+		err = -ENOMEM;
+		goto err_route_priv_alloc;
+	}
+
+	/* Write the route to the TCAM */
+	err = ops->route_create(mlxsw_sp, mr_tcam->priv, route->priv,
+				&route->key, route->afa_block,
+				route_params->prio);
+	if (err)
+		goto err_route_create;
+	return 0;
+
+err_route_create:
+	kfree(route->priv);
+err_route_priv_alloc:
+	mlxsw_sp_mr_tcam_afa_block_destroy(route->afa_block);
+err_afa_block_create:
+	mlxsw_sp_flow_counter_free(mlxsw_sp, route->counter_index);
+err_erif_populate:
+err_counter_alloc:
+	mlxsw_sp_mr_erif_list_flush(mlxsw_sp, &route->erif_list);
+	return err;
+}
+
+static void mlxsw_sp_mr_tcam_route_destroy(struct mlxsw_sp *mlxsw_sp,
+					   void *priv, void *route_priv)
+{
+	const struct mlxsw_sp_mr_tcam_ops *ops = mlxsw_sp->mr_tcam_ops;
+	struct mlxsw_sp_mr_tcam_route *route = route_priv;
+	struct mlxsw_sp_mr_tcam *mr_tcam = priv;
+
+	ops->route_destroy(mlxsw_sp, mr_tcam->priv, route->priv, &route->key);
+	kfree(route->priv);
+	mlxsw_sp_mr_tcam_afa_block_destroy(route->afa_block);
+	mlxsw_sp_flow_counter_free(mlxsw_sp, route->counter_index);
+	mlxsw_sp_mr_erif_list_flush(mlxsw_sp, &route->erif_list);
+}
+
+static int mlxsw_sp_mr_tcam_route_stats(struct mlxsw_sp *mlxsw_sp,
+					void *route_priv, u64 *packets,
+					u64 *bytes)
+{
+	struct mlxsw_sp_mr_tcam_route *route = route_priv;
+
+	return mlxsw_sp_flow_counter_get(mlxsw_sp, route->counter_index,
+					 packets, bytes);
+}
+
+static int
+mlxsw_sp_mr_tcam_route_action_update(struct mlxsw_sp *mlxsw_sp,
+				     void *route_priv,
+				     enum mlxsw_sp_mr_route_action route_action)
+{
+	const struct mlxsw_sp_mr_tcam_ops *ops = mlxsw_sp->mr_tcam_ops;
+	struct mlxsw_sp_mr_tcam_route *route = route_priv;
+	struct mlxsw_afa_block *afa_block;
+	int err;
+
+	/* Create a new flexible action block */
+	afa_block = mlxsw_sp_mr_tcam_afa_block_create(mlxsw_sp, route_action,
+						      route->irif_index,
+						      route->counter_index,
+						      route->min_mtu,
+						      &route->erif_list);
+	if (IS_ERR(afa_block))
+		return PTR_ERR(afa_block);
+
+	/* Update the TCAM route entry */
+	err = ops->route_update(mlxsw_sp, route->priv, &route->key, afa_block);
+	if (err)
+		goto err;
+
+	/* Delete the old one */
+	mlxsw_sp_mr_tcam_afa_block_destroy(route->afa_block);
+	route->afa_block = afa_block;
+	route->action = route_action;
+	return 0;
+err:
+	mlxsw_sp_mr_tcam_afa_block_destroy(afa_block);
+	return err;
+}
+
+static int mlxsw_sp_mr_tcam_route_min_mtu_update(struct mlxsw_sp *mlxsw_sp,
+						 void *route_priv, u16 min_mtu)
+{
+	const struct mlxsw_sp_mr_tcam_ops *ops = mlxsw_sp->mr_tcam_ops;
+	struct mlxsw_sp_mr_tcam_route *route = route_priv;
+	struct mlxsw_afa_block *afa_block;
+	int err;
+
+	/* Create a new flexible action block */
+	afa_block = mlxsw_sp_mr_tcam_afa_block_create(mlxsw_sp,
+						      route->action,
+						      route->irif_index,
+						      route->counter_index,
+						      min_mtu,
+						      &route->erif_list);
+	if (IS_ERR(afa_block))
+		return PTR_ERR(afa_block);
+
+	/* Update the TCAM route entry */
+	err = ops->route_update(mlxsw_sp, route->priv, &route->key, afa_block);
+	if (err)
+		goto err;
+
+	/* Delete the old one */
+	mlxsw_sp_mr_tcam_afa_block_destroy(route->afa_block);
+	route->afa_block = afa_block;
+	route->min_mtu = min_mtu;
+	return 0;
+err:
+	mlxsw_sp_mr_tcam_afa_block_destroy(afa_block);
+	return err;
+}
+
+static int mlxsw_sp_mr_tcam_route_irif_update(struct mlxsw_sp *mlxsw_sp,
+					      void *route_priv, u16 irif_index)
+{
+	struct mlxsw_sp_mr_tcam_route *route = route_priv;
+
+	if (route->action != MLXSW_SP_MR_ROUTE_ACTION_TRAP)
+		return -EINVAL;
+	route->irif_index = irif_index;
+	return 0;
+}
+
+static int mlxsw_sp_mr_tcam_route_erif_add(struct mlxsw_sp *mlxsw_sp,
+					   void *route_priv, u16 erif_index)
+{
+	struct mlxsw_sp_mr_tcam_route *route = route_priv;
+	int err;
+
+	err = mlxsw_sp_mr_erif_list_add(mlxsw_sp, &route->erif_list,
+					erif_index);
+	if (err)
+		return err;
+
+	/* Commit the action only if the route action is not TRAP */
+	if (route->action != MLXSW_SP_MR_ROUTE_ACTION_TRAP)
+		return mlxsw_sp_mr_erif_list_commit(mlxsw_sp,
+						    &route->erif_list);
+	return 0;
+}
+
+static int mlxsw_sp_mr_tcam_route_erif_del(struct mlxsw_sp *mlxsw_sp,
+					   void *route_priv, u16 erif_index)
+{
+	const struct mlxsw_sp_mr_tcam_ops *ops = mlxsw_sp->mr_tcam_ops;
+	struct mlxsw_sp_mr_tcam_route *route = route_priv;
+	struct mlxsw_sp_mr_erif_sublist *erif_sublist;
+	struct mlxsw_sp_mr_tcam_erif_list erif_list;
+	struct mlxsw_afa_block *afa_block;
+	int err;
+	int i;
+
+	/* Create a copy of the original erif_list without the deleted entry */
+	mlxsw_sp_mr_erif_list_init(&erif_list);
+	list_for_each_entry(erif_sublist, &route->erif_list.erif_sublists, list) {
+		for (i = 0; i < erif_sublist->num_erifs; i++) {
+			u16 curr_erif = erif_sublist->erif_indices[i];
+
+			if (curr_erif == erif_index)
+				continue;
+			err = mlxsw_sp_mr_erif_list_add(mlxsw_sp, &erif_list,
+							curr_erif);
+			if (err)
+				goto err_erif_list_add;
+		}
+	}
+
+	/* Create the flexible action block pointing to the new erif_list */
+	afa_block = mlxsw_sp_mr_tcam_afa_block_create(mlxsw_sp, route->action,
+						      route->irif_index,
+						      route->counter_index,
+						      route->min_mtu,
+						      &erif_list);
+	if (IS_ERR(afa_block)) {
+		err = PTR_ERR(afa_block);
+		goto err_afa_block_create;
+	}
+
+	/* Update the TCAM route entry */
+	err = ops->route_update(mlxsw_sp, route->priv, &route->key, afa_block);
+	if (err)
+		goto err_route_write;
+
+	mlxsw_sp_mr_tcam_afa_block_destroy(route->afa_block);
+	mlxsw_sp_mr_erif_list_flush(mlxsw_sp, &route->erif_list);
+	route->afa_block = afa_block;
+	mlxsw_sp_mr_erif_list_move(&route->erif_list, &erif_list);
+	return 0;
+
+err_route_write:
+	mlxsw_sp_mr_tcam_afa_block_destroy(afa_block);
+err_afa_block_create:
+err_erif_list_add:
+	mlxsw_sp_mr_erif_list_flush(mlxsw_sp, &erif_list);
+	return err;
+}
+
+static int
+mlxsw_sp_mr_tcam_route_update(struct mlxsw_sp *mlxsw_sp, void *route_priv,
+			      struct mlxsw_sp_mr_route_info *route_info)
+{
+	const struct mlxsw_sp_mr_tcam_ops *ops = mlxsw_sp->mr_tcam_ops;
+	struct mlxsw_sp_mr_tcam_route *route = route_priv;
+	struct mlxsw_sp_mr_tcam_erif_list erif_list;
+	struct mlxsw_afa_block *afa_block;
+	int err;
+
+	/* Create a new erif_list */
+	mlxsw_sp_mr_erif_list_init(&erif_list);
+	err = mlxsw_sp_mr_tcam_erif_populate(mlxsw_sp, &erif_list, route_info);
+	if (err)
+		goto err_erif_populate;
+
+	/* Create the flexible action block pointing to the new erif_list */
+	afa_block = mlxsw_sp_mr_tcam_afa_block_create(mlxsw_sp,
+						      route_info->route_action,
+						      route_info->irif_index,
+						      route->counter_index,
+						      route_info->min_mtu,
+						      &erif_list);
+	if (IS_ERR(afa_block)) {
+		err = PTR_ERR(afa_block);
+		goto err_afa_block_create;
+	}
+
+	/* Update the TCAM route entry */
+	err = ops->route_update(mlxsw_sp, route->priv, &route->key, afa_block);
+	if (err)
+		goto err_route_write;
+
+	mlxsw_sp_mr_tcam_afa_block_destroy(route->afa_block);
+	mlxsw_sp_mr_erif_list_flush(mlxsw_sp, &route->erif_list);
+	route->afa_block = afa_block;
+	mlxsw_sp_mr_erif_list_move(&route->erif_list, &erif_list);
+	route->action = route_info->route_action;
+	route->irif_index = route_info->irif_index;
+	route->min_mtu = route_info->min_mtu;
+	return 0;
+
+err_route_write:
+	mlxsw_sp_mr_tcam_afa_block_destroy(afa_block);
+err_afa_block_create:
+err_erif_populate:
+	mlxsw_sp_mr_erif_list_flush(mlxsw_sp, &erif_list);
+	return err;
+}
+
+static int mlxsw_sp_mr_tcam_init(struct mlxsw_sp *mlxsw_sp, void *priv)
+{
+	const struct mlxsw_sp_mr_tcam_ops *ops = mlxsw_sp->mr_tcam_ops;
+	struct mlxsw_sp_mr_tcam *mr_tcam = priv;
+	int err;
+
+	if (!MLXSW_CORE_RES_VALID(mlxsw_sp->core, MC_ERIF_LIST_ENTRIES))
+		return -EIO;
+
+	mr_tcam->priv = kzalloc(ops->priv_size, GFP_KERNEL);
+	if (!mr_tcam->priv)
+		return -ENOMEM;
+
+	err = ops->init(mlxsw_sp, mr_tcam->priv);
+	if (err)
+		goto err_init;
+	return 0;
+
+err_init:
+	kfree(mr_tcam->priv);
+	return err;
+}
+
+static void mlxsw_sp_mr_tcam_fini(struct mlxsw_sp *mlxsw_sp, void *priv)
+{
+	const struct mlxsw_sp_mr_tcam_ops *ops = mlxsw_sp->mr_tcam_ops;
+	struct mlxsw_sp_mr_tcam *mr_tcam = priv;
+
+	ops->fini(mr_tcam->priv);
+	kfree(mr_tcam->priv);
+}
+
+const struct mlxsw_sp_mr_ops mlxsw_sp_mr_tcam_ops = {
+	.priv_size = sizeof(struct mlxsw_sp_mr_tcam),
+	.route_priv_size = sizeof(struct mlxsw_sp_mr_tcam_route),
+	.init = mlxsw_sp_mr_tcam_init,
+	.route_create = mlxsw_sp_mr_tcam_route_create,
+	.route_update = mlxsw_sp_mr_tcam_route_update,
+	.route_stats = mlxsw_sp_mr_tcam_route_stats,
+	.route_action_update = mlxsw_sp_mr_tcam_route_action_update,
+	.route_min_mtu_update = mlxsw_sp_mr_tcam_route_min_mtu_update,
+	.route_irif_update = mlxsw_sp_mr_tcam_route_irif_update,
+	.route_erif_add = mlxsw_sp_mr_tcam_route_erif_add,
+	.route_erif_del = mlxsw_sp_mr_tcam_route_erif_del,
+	.route_destroy = mlxsw_sp_mr_tcam_route_destroy,
+	.fini = mlxsw_sp_mr_tcam_fini,
+};
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_mr_tcam.h b/drivers/net/ethernet/mellanox/mlxsw/spectrum_mr_tcam.h
new file mode 100644
index 000000000..3c84151b4
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_mr_tcam.h
@@ -0,0 +1,12 @@
+/* SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0 */
+/* Copyright (c) 2017-2018 Mellanox Technologies. All rights reserved */
+
+#ifndef _MLXSW_SPECTRUM_MCROUTER_TCAM_H
+#define _MLXSW_SPECTRUM_MCROUTER_TCAM_H
+
+#include "spectrum.h"
+#include "spectrum_mr.h"
+
+extern const struct mlxsw_sp_mr_ops mlxsw_sp_mr_tcam_ops;
+
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_nve.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_nve.c
new file mode 100644
index 000000000..d2b57a045
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_nve.c
@@ -0,0 +1,1173 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2018 Mellanox Technologies. All rights reserved */
+
+#include <linux/err.h>
+#include <linux/gfp.h>
+#include <linux/kernel.h>
+#include <linux/list.h>
+#include <linux/netlink.h>
+#include <linux/rtnetlink.h>
+#include <linux/slab.h>
+#include <net/inet_ecn.h>
+#include <net/ipv6.h>
+
+#include "reg.h"
+#include "spectrum.h"
+#include "spectrum_nve.h"
+
+const struct mlxsw_sp_nve_ops *mlxsw_sp1_nve_ops_arr[] = {
+	[MLXSW_SP_NVE_TYPE_VXLAN]	= &mlxsw_sp1_nve_vxlan_ops,
+};
+
+const struct mlxsw_sp_nve_ops *mlxsw_sp2_nve_ops_arr[] = {
+	[MLXSW_SP_NVE_TYPE_VXLAN]	= &mlxsw_sp2_nve_vxlan_ops,
+};
+
+struct mlxsw_sp_nve_mc_entry;
+struct mlxsw_sp_nve_mc_record;
+struct mlxsw_sp_nve_mc_list;
+
+struct mlxsw_sp_nve_mc_record_ops {
+	enum mlxsw_reg_tnumt_record_type type;
+	int (*entry_add)(struct mlxsw_sp_nve_mc_record *mc_record,
+			 struct mlxsw_sp_nve_mc_entry *mc_entry,
+			 const union mlxsw_sp_l3addr *addr);
+	void (*entry_del)(const struct mlxsw_sp_nve_mc_record *mc_record,
+			  const struct mlxsw_sp_nve_mc_entry *mc_entry);
+	void (*entry_set)(const struct mlxsw_sp_nve_mc_record *mc_record,
+			  const struct mlxsw_sp_nve_mc_entry *mc_entry,
+			  char *tnumt_pl, unsigned int entry_index);
+	bool (*entry_compare)(const struct mlxsw_sp_nve_mc_record *mc_record,
+			      const struct mlxsw_sp_nve_mc_entry *mc_entry,
+			      const union mlxsw_sp_l3addr *addr);
+};
+
+struct mlxsw_sp_nve_mc_list_key {
+	u16 fid_index;
+};
+
+struct mlxsw_sp_nve_mc_ipv6_entry {
+	struct in6_addr addr6;
+	u32 addr6_kvdl_index;
+};
+
+struct mlxsw_sp_nve_mc_entry {
+	union {
+		__be32 addr4;
+		struct mlxsw_sp_nve_mc_ipv6_entry ipv6_entry;
+	};
+	u8 valid:1;
+};
+
+struct mlxsw_sp_nve_mc_record {
+	struct list_head list;
+	enum mlxsw_sp_l3proto proto;
+	unsigned int num_entries;
+	struct mlxsw_sp *mlxsw_sp;
+	struct mlxsw_sp_nve_mc_list *mc_list;
+	const struct mlxsw_sp_nve_mc_record_ops *ops;
+	u32 kvdl_index;
+	struct mlxsw_sp_nve_mc_entry entries[];
+};
+
+struct mlxsw_sp_nve_mc_list {
+	struct list_head records_list;
+	struct rhash_head ht_node;
+	struct mlxsw_sp_nve_mc_list_key key;
+};
+
+static const struct rhashtable_params mlxsw_sp_nve_mc_list_ht_params = {
+	.key_len = sizeof(struct mlxsw_sp_nve_mc_list_key),
+	.key_offset = offsetof(struct mlxsw_sp_nve_mc_list, key),
+	.head_offset = offsetof(struct mlxsw_sp_nve_mc_list, ht_node),
+};
+
+static int
+mlxsw_sp_nve_mc_record_ipv4_entry_add(struct mlxsw_sp_nve_mc_record *mc_record,
+				      struct mlxsw_sp_nve_mc_entry *mc_entry,
+				      const union mlxsw_sp_l3addr *addr)
+{
+	mc_entry->addr4 = addr->addr4;
+
+	return 0;
+}
+
+static void
+mlxsw_sp_nve_mc_record_ipv4_entry_del(const struct mlxsw_sp_nve_mc_record *mc_record,
+				      const struct mlxsw_sp_nve_mc_entry *mc_entry)
+{
+}
+
+static void
+mlxsw_sp_nve_mc_record_ipv4_entry_set(const struct mlxsw_sp_nve_mc_record *mc_record,
+				      const struct mlxsw_sp_nve_mc_entry *mc_entry,
+				      char *tnumt_pl, unsigned int entry_index)
+{
+	u32 udip = be32_to_cpu(mc_entry->addr4);
+
+	mlxsw_reg_tnumt_udip_set(tnumt_pl, entry_index, udip);
+}
+
+static bool
+mlxsw_sp_nve_mc_record_ipv4_entry_compare(const struct mlxsw_sp_nve_mc_record *mc_record,
+					  const struct mlxsw_sp_nve_mc_entry *mc_entry,
+					  const union mlxsw_sp_l3addr *addr)
+{
+	return mc_entry->addr4 == addr->addr4;
+}
+
+static const struct mlxsw_sp_nve_mc_record_ops
+mlxsw_sp_nve_mc_record_ipv4_ops = {
+	.type		= MLXSW_REG_TNUMT_RECORD_TYPE_IPV4,
+	.entry_add	= &mlxsw_sp_nve_mc_record_ipv4_entry_add,
+	.entry_del	= &mlxsw_sp_nve_mc_record_ipv4_entry_del,
+	.entry_set	= &mlxsw_sp_nve_mc_record_ipv4_entry_set,
+	.entry_compare	= &mlxsw_sp_nve_mc_record_ipv4_entry_compare,
+};
+
+static int
+mlxsw_sp_nve_mc_record_ipv6_entry_add(struct mlxsw_sp_nve_mc_record *mc_record,
+				      struct mlxsw_sp_nve_mc_entry *mc_entry,
+				      const union mlxsw_sp_l3addr *addr)
+{
+	u32 kvdl_index;
+	int err;
+
+	err = mlxsw_sp_ipv6_addr_kvdl_index_get(mc_record->mlxsw_sp,
+						&addr->addr6, &kvdl_index);
+	if (err)
+		return err;
+
+	mc_entry->ipv6_entry.addr6 = addr->addr6;
+	mc_entry->ipv6_entry.addr6_kvdl_index = kvdl_index;
+	return 0;
+}
+
+static void
+mlxsw_sp_nve_mc_record_ipv6_entry_del(const struct mlxsw_sp_nve_mc_record *mc_record,
+				      const struct mlxsw_sp_nve_mc_entry *mc_entry)
+{
+	mlxsw_sp_ipv6_addr_put(mc_record->mlxsw_sp,
+			       &mc_entry->ipv6_entry.addr6);
+}
+
+static void
+mlxsw_sp_nve_mc_record_ipv6_entry_set(const struct mlxsw_sp_nve_mc_record *mc_record,
+				      const struct mlxsw_sp_nve_mc_entry *mc_entry,
+				      char *tnumt_pl, unsigned int entry_index)
+{
+	u32 udip_ptr = mc_entry->ipv6_entry.addr6_kvdl_index;
+
+	mlxsw_reg_tnumt_udip_ptr_set(tnumt_pl, entry_index, udip_ptr);
+}
+
+static bool
+mlxsw_sp_nve_mc_record_ipv6_entry_compare(const struct mlxsw_sp_nve_mc_record *mc_record,
+					  const struct mlxsw_sp_nve_mc_entry *mc_entry,
+					  const union mlxsw_sp_l3addr *addr)
+{
+	return ipv6_addr_equal(&mc_entry->ipv6_entry.addr6, &addr->addr6);
+}
+
+static const struct mlxsw_sp_nve_mc_record_ops
+mlxsw_sp_nve_mc_record_ipv6_ops = {
+	.type		= MLXSW_REG_TNUMT_RECORD_TYPE_IPV6,
+	.entry_add	= &mlxsw_sp_nve_mc_record_ipv6_entry_add,
+	.entry_del	= &mlxsw_sp_nve_mc_record_ipv6_entry_del,
+	.entry_set	= &mlxsw_sp_nve_mc_record_ipv6_entry_set,
+	.entry_compare	= &mlxsw_sp_nve_mc_record_ipv6_entry_compare,
+};
+
+static const struct mlxsw_sp_nve_mc_record_ops *
+mlxsw_sp_nve_mc_record_ops_arr[] = {
+	[MLXSW_SP_L3_PROTO_IPV4] = &mlxsw_sp_nve_mc_record_ipv4_ops,
+	[MLXSW_SP_L3_PROTO_IPV6] = &mlxsw_sp_nve_mc_record_ipv6_ops,
+};
+
+int mlxsw_sp_nve_learned_ip_resolve(struct mlxsw_sp *mlxsw_sp, u32 uip,
+				    enum mlxsw_sp_l3proto proto,
+				    union mlxsw_sp_l3addr *addr)
+{
+	switch (proto) {
+	case MLXSW_SP_L3_PROTO_IPV4:
+		addr->addr4 = cpu_to_be32(uip);
+		return 0;
+	default:
+		WARN_ON(1);
+		return -EINVAL;
+	}
+}
+
+static struct mlxsw_sp_nve_mc_list *
+mlxsw_sp_nve_mc_list_find(struct mlxsw_sp *mlxsw_sp,
+			  const struct mlxsw_sp_nve_mc_list_key *key)
+{
+	struct mlxsw_sp_nve *nve = mlxsw_sp->nve;
+
+	return rhashtable_lookup_fast(&nve->mc_list_ht, key,
+				      mlxsw_sp_nve_mc_list_ht_params);
+}
+
+static struct mlxsw_sp_nve_mc_list *
+mlxsw_sp_nve_mc_list_create(struct mlxsw_sp *mlxsw_sp,
+			    const struct mlxsw_sp_nve_mc_list_key *key)
+{
+	struct mlxsw_sp_nve *nve = mlxsw_sp->nve;
+	struct mlxsw_sp_nve_mc_list *mc_list;
+	int err;
+
+	mc_list = kmalloc(sizeof(*mc_list), GFP_KERNEL);
+	if (!mc_list)
+		return ERR_PTR(-ENOMEM);
+
+	INIT_LIST_HEAD(&mc_list->records_list);
+	mc_list->key = *key;
+
+	err = rhashtable_insert_fast(&nve->mc_list_ht, &mc_list->ht_node,
+				     mlxsw_sp_nve_mc_list_ht_params);
+	if (err)
+		goto err_rhashtable_insert;
+
+	return mc_list;
+
+err_rhashtable_insert:
+	kfree(mc_list);
+	return ERR_PTR(err);
+}
+
+static void mlxsw_sp_nve_mc_list_destroy(struct mlxsw_sp *mlxsw_sp,
+					 struct mlxsw_sp_nve_mc_list *mc_list)
+{
+	struct mlxsw_sp_nve *nve = mlxsw_sp->nve;
+
+	rhashtable_remove_fast(&nve->mc_list_ht, &mc_list->ht_node,
+			       mlxsw_sp_nve_mc_list_ht_params);
+	WARN_ON(!list_empty(&mc_list->records_list));
+	kfree(mc_list);
+}
+
+static struct mlxsw_sp_nve_mc_list *
+mlxsw_sp_nve_mc_list_get(struct mlxsw_sp *mlxsw_sp,
+			 const struct mlxsw_sp_nve_mc_list_key *key)
+{
+	struct mlxsw_sp_nve_mc_list *mc_list;
+
+	mc_list = mlxsw_sp_nve_mc_list_find(mlxsw_sp, key);
+	if (mc_list)
+		return mc_list;
+
+	return mlxsw_sp_nve_mc_list_create(mlxsw_sp, key);
+}
+
+static void
+mlxsw_sp_nve_mc_list_put(struct mlxsw_sp *mlxsw_sp,
+			 struct mlxsw_sp_nve_mc_list *mc_list)
+{
+	if (!list_empty(&mc_list->records_list))
+		return;
+	mlxsw_sp_nve_mc_list_destroy(mlxsw_sp, mc_list);
+}
+
+static struct mlxsw_sp_nve_mc_record *
+mlxsw_sp_nve_mc_record_create(struct mlxsw_sp *mlxsw_sp,
+			      struct mlxsw_sp_nve_mc_list *mc_list,
+			      enum mlxsw_sp_l3proto proto)
+{
+	unsigned int num_max_entries = mlxsw_sp->nve->num_max_mc_entries[proto];
+	struct mlxsw_sp_nve_mc_record *mc_record;
+	int err;
+
+	mc_record = kzalloc(struct_size(mc_record, entries, num_max_entries),
+			    GFP_KERNEL);
+	if (!mc_record)
+		return ERR_PTR(-ENOMEM);
+
+	err = mlxsw_sp_kvdl_alloc(mlxsw_sp, MLXSW_SP_KVDL_ENTRY_TYPE_TNUMT, 1,
+				  &mc_record->kvdl_index);
+	if (err)
+		goto err_kvdl_alloc;
+
+	mc_record->ops = mlxsw_sp_nve_mc_record_ops_arr[proto];
+	mc_record->mlxsw_sp = mlxsw_sp;
+	mc_record->mc_list = mc_list;
+	mc_record->proto = proto;
+	list_add_tail(&mc_record->list, &mc_list->records_list);
+
+	return mc_record;
+
+err_kvdl_alloc:
+	kfree(mc_record);
+	return ERR_PTR(err);
+}
+
+static void
+mlxsw_sp_nve_mc_record_destroy(struct mlxsw_sp_nve_mc_record *mc_record)
+{
+	struct mlxsw_sp *mlxsw_sp = mc_record->mlxsw_sp;
+
+	list_del(&mc_record->list);
+	mlxsw_sp_kvdl_free(mlxsw_sp, MLXSW_SP_KVDL_ENTRY_TYPE_TNUMT, 1,
+			   mc_record->kvdl_index);
+	WARN_ON(mc_record->num_entries);
+	kfree(mc_record);
+}
+
+static struct mlxsw_sp_nve_mc_record *
+mlxsw_sp_nve_mc_record_get(struct mlxsw_sp *mlxsw_sp,
+			   struct mlxsw_sp_nve_mc_list *mc_list,
+			   enum mlxsw_sp_l3proto proto)
+{
+	struct mlxsw_sp_nve_mc_record *mc_record;
+
+	list_for_each_entry_reverse(mc_record, &mc_list->records_list, list) {
+		unsigned int num_entries = mc_record->num_entries;
+		struct mlxsw_sp_nve *nve = mlxsw_sp->nve;
+
+		if (mc_record->proto == proto &&
+		    num_entries < nve->num_max_mc_entries[proto])
+			return mc_record;
+	}
+
+	return mlxsw_sp_nve_mc_record_create(mlxsw_sp, mc_list, proto);
+}
+
+static void
+mlxsw_sp_nve_mc_record_put(struct mlxsw_sp_nve_mc_record *mc_record)
+{
+	if (mc_record->num_entries != 0)
+		return;
+
+	mlxsw_sp_nve_mc_record_destroy(mc_record);
+}
+
+static struct mlxsw_sp_nve_mc_entry *
+mlxsw_sp_nve_mc_free_entry_find(struct mlxsw_sp_nve_mc_record *mc_record)
+{
+	struct mlxsw_sp_nve *nve = mc_record->mlxsw_sp->nve;
+	unsigned int num_max_entries;
+	int i;
+
+	num_max_entries = nve->num_max_mc_entries[mc_record->proto];
+	for (i = 0; i < num_max_entries; i++) {
+		if (mc_record->entries[i].valid)
+			continue;
+		return &mc_record->entries[i];
+	}
+
+	return NULL;
+}
+
+static int
+mlxsw_sp_nve_mc_record_refresh(struct mlxsw_sp_nve_mc_record *mc_record)
+{
+	enum mlxsw_reg_tnumt_record_type type = mc_record->ops->type;
+	struct mlxsw_sp_nve_mc_list *mc_list = mc_record->mc_list;
+	struct mlxsw_sp *mlxsw_sp = mc_record->mlxsw_sp;
+	char tnumt_pl[MLXSW_REG_TNUMT_LEN];
+	unsigned int num_max_entries;
+	unsigned int num_entries = 0;
+	u32 next_kvdl_index = 0;
+	bool next_valid = false;
+	int i;
+
+	if (!list_is_last(&mc_record->list, &mc_list->records_list)) {
+		struct mlxsw_sp_nve_mc_record *next_record;
+
+		next_record = list_next_entry(mc_record, list);
+		next_kvdl_index = next_record->kvdl_index;
+		next_valid = true;
+	}
+
+	mlxsw_reg_tnumt_pack(tnumt_pl, type, MLXSW_REG_TUNNEL_PORT_NVE,
+			     mc_record->kvdl_index, next_valid,
+			     next_kvdl_index, mc_record->num_entries);
+
+	num_max_entries = mlxsw_sp->nve->num_max_mc_entries[mc_record->proto];
+	for (i = 0; i < num_max_entries; i++) {
+		struct mlxsw_sp_nve_mc_entry *mc_entry;
+
+		mc_entry = &mc_record->entries[i];
+		if (!mc_entry->valid)
+			continue;
+		mc_record->ops->entry_set(mc_record, mc_entry, tnumt_pl,
+					  num_entries++);
+	}
+
+	WARN_ON(num_entries != mc_record->num_entries);
+
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(tnumt), tnumt_pl);
+}
+
+static bool
+mlxsw_sp_nve_mc_record_is_first(struct mlxsw_sp_nve_mc_record *mc_record)
+{
+	struct mlxsw_sp_nve_mc_list *mc_list = mc_record->mc_list;
+	struct mlxsw_sp_nve_mc_record *first_record;
+
+	first_record = list_first_entry(&mc_list->records_list,
+					struct mlxsw_sp_nve_mc_record, list);
+
+	return mc_record == first_record;
+}
+
+static struct mlxsw_sp_nve_mc_entry *
+mlxsw_sp_nve_mc_entry_find(struct mlxsw_sp_nve_mc_record *mc_record,
+			   union mlxsw_sp_l3addr *addr)
+{
+	struct mlxsw_sp_nve *nve = mc_record->mlxsw_sp->nve;
+	unsigned int num_max_entries;
+	int i;
+
+	num_max_entries = nve->num_max_mc_entries[mc_record->proto];
+	for (i = 0; i < num_max_entries; i++) {
+		struct mlxsw_sp_nve_mc_entry *mc_entry;
+
+		mc_entry = &mc_record->entries[i];
+		if (!mc_entry->valid)
+			continue;
+		if (mc_record->ops->entry_compare(mc_record, mc_entry, addr))
+			return mc_entry;
+	}
+
+	return NULL;
+}
+
+static int
+mlxsw_sp_nve_mc_record_ip_add(struct mlxsw_sp_nve_mc_record *mc_record,
+			      union mlxsw_sp_l3addr *addr)
+{
+	struct mlxsw_sp_nve_mc_entry *mc_entry = NULL;
+	int err;
+
+	mc_entry = mlxsw_sp_nve_mc_free_entry_find(mc_record);
+	if (WARN_ON(!mc_entry))
+		return -EINVAL;
+
+	err = mc_record->ops->entry_add(mc_record, mc_entry, addr);
+	if (err)
+		return err;
+	mc_record->num_entries++;
+	mc_entry->valid = true;
+
+	err = mlxsw_sp_nve_mc_record_refresh(mc_record);
+	if (err)
+		goto err_record_refresh;
+
+	/* If this is a new record and not the first one, then we need to
+	 * update the next pointer of the previous entry
+	 */
+	if (mc_record->num_entries != 1 ||
+	    mlxsw_sp_nve_mc_record_is_first(mc_record))
+		return 0;
+
+	err = mlxsw_sp_nve_mc_record_refresh(list_prev_entry(mc_record, list));
+	if (err)
+		goto err_prev_record_refresh;
+
+	return 0;
+
+err_prev_record_refresh:
+err_record_refresh:
+	mc_entry->valid = false;
+	mc_record->num_entries--;
+	mc_record->ops->entry_del(mc_record, mc_entry);
+	return err;
+}
+
+static void
+mlxsw_sp_nve_mc_record_entry_del(struct mlxsw_sp_nve_mc_record *mc_record,
+				 struct mlxsw_sp_nve_mc_entry *mc_entry)
+{
+	struct mlxsw_sp_nve_mc_list *mc_list = mc_record->mc_list;
+
+	mc_entry->valid = false;
+	mc_record->num_entries--;
+
+	/* When the record continues to exist we only need to invalidate
+	 * the requested entry
+	 */
+	if (mc_record->num_entries != 0) {
+		mlxsw_sp_nve_mc_record_refresh(mc_record);
+		mc_record->ops->entry_del(mc_record, mc_entry);
+		return;
+	}
+
+	/* If the record needs to be deleted, but it is not the first,
+	 * then we need to make sure that the previous record no longer
+	 * points to it. Remove deleted record from the list to reflect
+	 * that and then re-add it at the end, so that it could be
+	 * properly removed by the record destruction code
+	 */
+	if (!mlxsw_sp_nve_mc_record_is_first(mc_record)) {
+		struct mlxsw_sp_nve_mc_record *prev_record;
+
+		prev_record = list_prev_entry(mc_record, list);
+		list_del(&mc_record->list);
+		mlxsw_sp_nve_mc_record_refresh(prev_record);
+		list_add_tail(&mc_record->list, &mc_list->records_list);
+		mc_record->ops->entry_del(mc_record, mc_entry);
+		return;
+	}
+
+	/* If the first record needs to be deleted, but the list is not
+	 * singular, then the second record needs to be written in the
+	 * first record's address, as this address is stored as a property
+	 * of the FID
+	 */
+	if (mlxsw_sp_nve_mc_record_is_first(mc_record) &&
+	    !list_is_singular(&mc_list->records_list)) {
+		struct mlxsw_sp_nve_mc_record *next_record;
+
+		next_record = list_next_entry(mc_record, list);
+		swap(mc_record->kvdl_index, next_record->kvdl_index);
+		mlxsw_sp_nve_mc_record_refresh(next_record);
+		mc_record->ops->entry_del(mc_record, mc_entry);
+		return;
+	}
+
+	/* This is the last case where the last remaining record needs to
+	 * be deleted. Simply delete the entry
+	 */
+	mc_record->ops->entry_del(mc_record, mc_entry);
+}
+
+static struct mlxsw_sp_nve_mc_record *
+mlxsw_sp_nve_mc_record_find(struct mlxsw_sp_nve_mc_list *mc_list,
+			    enum mlxsw_sp_l3proto proto,
+			    union mlxsw_sp_l3addr *addr,
+			    struct mlxsw_sp_nve_mc_entry **mc_entry)
+{
+	struct mlxsw_sp_nve_mc_record *mc_record;
+
+	list_for_each_entry(mc_record, &mc_list->records_list, list) {
+		if (mc_record->proto != proto)
+			continue;
+
+		*mc_entry = mlxsw_sp_nve_mc_entry_find(mc_record, addr);
+		if (*mc_entry)
+			return mc_record;
+	}
+
+	return NULL;
+}
+
+static int mlxsw_sp_nve_mc_list_ip_add(struct mlxsw_sp *mlxsw_sp,
+				       struct mlxsw_sp_nve_mc_list *mc_list,
+				       enum mlxsw_sp_l3proto proto,
+				       union mlxsw_sp_l3addr *addr)
+{
+	struct mlxsw_sp_nve_mc_record *mc_record;
+	int err;
+
+	mc_record = mlxsw_sp_nve_mc_record_get(mlxsw_sp, mc_list, proto);
+	if (IS_ERR(mc_record))
+		return PTR_ERR(mc_record);
+
+	err = mlxsw_sp_nve_mc_record_ip_add(mc_record, addr);
+	if (err)
+		goto err_ip_add;
+
+	return 0;
+
+err_ip_add:
+	mlxsw_sp_nve_mc_record_put(mc_record);
+	return err;
+}
+
+static void mlxsw_sp_nve_mc_list_ip_del(struct mlxsw_sp *mlxsw_sp,
+					struct mlxsw_sp_nve_mc_list *mc_list,
+					enum mlxsw_sp_l3proto proto,
+					union mlxsw_sp_l3addr *addr)
+{
+	struct mlxsw_sp_nve_mc_record *mc_record;
+	struct mlxsw_sp_nve_mc_entry *mc_entry;
+
+	mc_record = mlxsw_sp_nve_mc_record_find(mc_list, proto, addr,
+						&mc_entry);
+	if (!mc_record)
+		return;
+
+	mlxsw_sp_nve_mc_record_entry_del(mc_record, mc_entry);
+	mlxsw_sp_nve_mc_record_put(mc_record);
+}
+
+static int
+mlxsw_sp_nve_fid_flood_index_set(struct mlxsw_sp_fid *fid,
+				 struct mlxsw_sp_nve_mc_list *mc_list)
+{
+	struct mlxsw_sp_nve_mc_record *mc_record;
+
+	/* The address of the first record in the list is a property of
+	 * the FID and we never change it. It only needs to be set when
+	 * a new list is created
+	 */
+	if (mlxsw_sp_fid_nve_flood_index_is_set(fid))
+		return 0;
+
+	mc_record = list_first_entry(&mc_list->records_list,
+				     struct mlxsw_sp_nve_mc_record, list);
+
+	return mlxsw_sp_fid_nve_flood_index_set(fid, mc_record->kvdl_index);
+}
+
+static void
+mlxsw_sp_nve_fid_flood_index_clear(struct mlxsw_sp_fid *fid,
+				   struct mlxsw_sp_nve_mc_list *mc_list)
+{
+	struct mlxsw_sp_nve_mc_record *mc_record;
+
+	/* The address of the first record needs to be invalidated only when
+	 * the last record is about to be removed
+	 */
+	if (!list_is_singular(&mc_list->records_list))
+		return;
+
+	mc_record = list_first_entry(&mc_list->records_list,
+				     struct mlxsw_sp_nve_mc_record, list);
+	if (mc_record->num_entries != 1)
+		return;
+
+	return mlxsw_sp_fid_nve_flood_index_clear(fid);
+}
+
+int mlxsw_sp_nve_flood_ip_add(struct mlxsw_sp *mlxsw_sp,
+			      struct mlxsw_sp_fid *fid,
+			      enum mlxsw_sp_l3proto proto,
+			      union mlxsw_sp_l3addr *addr)
+{
+	struct mlxsw_sp_nve_mc_list_key key = { 0 };
+	struct mlxsw_sp_nve_mc_list *mc_list;
+	int err;
+
+	key.fid_index = mlxsw_sp_fid_index(fid);
+	mc_list = mlxsw_sp_nve_mc_list_get(mlxsw_sp, &key);
+	if (IS_ERR(mc_list))
+		return PTR_ERR(mc_list);
+
+	err = mlxsw_sp_nve_mc_list_ip_add(mlxsw_sp, mc_list, proto, addr);
+	if (err)
+		goto err_add_ip;
+
+	err = mlxsw_sp_nve_fid_flood_index_set(fid, mc_list);
+	if (err)
+		goto err_fid_flood_index_set;
+
+	return 0;
+
+err_fid_flood_index_set:
+	mlxsw_sp_nve_mc_list_ip_del(mlxsw_sp, mc_list, proto, addr);
+err_add_ip:
+	mlxsw_sp_nve_mc_list_put(mlxsw_sp, mc_list);
+	return err;
+}
+
+void mlxsw_sp_nve_flood_ip_del(struct mlxsw_sp *mlxsw_sp,
+			       struct mlxsw_sp_fid *fid,
+			       enum mlxsw_sp_l3proto proto,
+			       union mlxsw_sp_l3addr *addr)
+{
+	struct mlxsw_sp_nve_mc_list_key key = { 0 };
+	struct mlxsw_sp_nve_mc_list *mc_list;
+
+	key.fid_index = mlxsw_sp_fid_index(fid);
+	mc_list = mlxsw_sp_nve_mc_list_find(mlxsw_sp, &key);
+	if (!mc_list)
+		return;
+
+	mlxsw_sp_nve_fid_flood_index_clear(fid, mc_list);
+	mlxsw_sp_nve_mc_list_ip_del(mlxsw_sp, mc_list, proto, addr);
+	mlxsw_sp_nve_mc_list_put(mlxsw_sp, mc_list);
+}
+
+static void
+mlxsw_sp_nve_mc_record_delete(struct mlxsw_sp_nve_mc_record *mc_record)
+{
+	struct mlxsw_sp_nve *nve = mc_record->mlxsw_sp->nve;
+	unsigned int num_max_entries;
+	int i;
+
+	num_max_entries = nve->num_max_mc_entries[mc_record->proto];
+	for (i = 0; i < num_max_entries; i++) {
+		struct mlxsw_sp_nve_mc_entry *mc_entry = &mc_record->entries[i];
+
+		if (!mc_entry->valid)
+			continue;
+		mlxsw_sp_nve_mc_record_entry_del(mc_record, mc_entry);
+	}
+
+	WARN_ON(mc_record->num_entries);
+	mlxsw_sp_nve_mc_record_put(mc_record);
+}
+
+static void mlxsw_sp_nve_flood_ip_flush(struct mlxsw_sp *mlxsw_sp,
+					struct mlxsw_sp_fid *fid)
+{
+	struct mlxsw_sp_nve_mc_record *mc_record, *tmp;
+	struct mlxsw_sp_nve_mc_list_key key = { 0 };
+	struct mlxsw_sp_nve_mc_list *mc_list;
+
+	if (!mlxsw_sp_fid_nve_flood_index_is_set(fid))
+		return;
+
+	mlxsw_sp_fid_nve_flood_index_clear(fid);
+
+	key.fid_index = mlxsw_sp_fid_index(fid);
+	mc_list = mlxsw_sp_nve_mc_list_find(mlxsw_sp, &key);
+	if (WARN_ON(!mc_list))
+		return;
+
+	list_for_each_entry_safe(mc_record, tmp, &mc_list->records_list, list)
+		mlxsw_sp_nve_mc_record_delete(mc_record);
+
+	WARN_ON(!list_empty(&mc_list->records_list));
+	mlxsw_sp_nve_mc_list_put(mlxsw_sp, mc_list);
+}
+
+static int mlxsw_sp_nve_tunnel_init(struct mlxsw_sp *mlxsw_sp,
+				    struct mlxsw_sp_nve_config *config)
+{
+	struct mlxsw_sp_nve *nve = mlxsw_sp->nve;
+	const struct mlxsw_sp_nve_ops *ops;
+	int err;
+
+	if (nve->num_nve_tunnels++ != 0)
+		return 0;
+
+	nve->config = *config;
+
+	err = mlxsw_sp_kvdl_alloc(mlxsw_sp, MLXSW_SP_KVDL_ENTRY_TYPE_ADJ, 1,
+				  &nve->tunnel_index);
+	if (err)
+		goto err_kvdl_alloc;
+
+	ops = nve->nve_ops_arr[config->type];
+	err = ops->init(nve, config);
+	if (err)
+		goto err_ops_init;
+
+	return 0;
+
+err_ops_init:
+	mlxsw_sp_kvdl_free(mlxsw_sp, MLXSW_SP_KVDL_ENTRY_TYPE_ADJ, 1,
+			   nve->tunnel_index);
+err_kvdl_alloc:
+	memset(&nve->config, 0, sizeof(nve->config));
+	nve->num_nve_tunnels--;
+	return err;
+}
+
+static void mlxsw_sp_nve_tunnel_fini(struct mlxsw_sp *mlxsw_sp)
+{
+	struct mlxsw_sp_nve *nve = mlxsw_sp->nve;
+	const struct mlxsw_sp_nve_ops *ops;
+
+	ops = nve->nve_ops_arr[nve->config.type];
+
+	if (mlxsw_sp->nve->num_nve_tunnels == 1) {
+		ops->fini(nve);
+		mlxsw_sp_kvdl_free(mlxsw_sp, MLXSW_SP_KVDL_ENTRY_TYPE_ADJ, 1,
+				   nve->tunnel_index);
+		memset(&nve->config, 0, sizeof(nve->config));
+	}
+	nve->num_nve_tunnels--;
+}
+
+static void mlxsw_sp_nve_fdb_flush_by_fid(struct mlxsw_sp *mlxsw_sp,
+					  u16 fid_index)
+{
+	char sfdf_pl[MLXSW_REG_SFDF_LEN];
+
+	mlxsw_reg_sfdf_pack(sfdf_pl, MLXSW_REG_SFDF_FLUSH_PER_NVE_AND_FID);
+	mlxsw_reg_sfdf_fid_set(sfdf_pl, fid_index);
+	mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(sfdf), sfdf_pl);
+}
+
+static void mlxsw_sp_nve_fdb_clear_offload(struct mlxsw_sp *mlxsw_sp,
+					   const struct mlxsw_sp_fid *fid,
+					   const struct net_device *nve_dev,
+					   __be32 vni)
+{
+	const struct mlxsw_sp_nve_ops *ops;
+	enum mlxsw_sp_nve_type type;
+
+	if (WARN_ON(mlxsw_sp_fid_nve_type(fid, &type)))
+		return;
+
+	ops = mlxsw_sp->nve->nve_ops_arr[type];
+	ops->fdb_clear_offload(nve_dev, vni);
+}
+
+struct mlxsw_sp_nve_ipv6_ht_key {
+	u8 mac[ETH_ALEN];
+	u16 fid_index;
+};
+
+struct mlxsw_sp_nve_ipv6_ht_node {
+	struct rhash_head ht_node;
+	struct list_head list;
+	struct mlxsw_sp_nve_ipv6_ht_key key;
+	struct in6_addr addr6;
+};
+
+static const struct rhashtable_params mlxsw_sp_nve_ipv6_ht_params = {
+	.key_len = sizeof(struct mlxsw_sp_nve_ipv6_ht_key),
+	.key_offset = offsetof(struct mlxsw_sp_nve_ipv6_ht_node, key),
+	.head_offset = offsetof(struct mlxsw_sp_nve_ipv6_ht_node, ht_node),
+};
+
+int mlxsw_sp_nve_ipv6_addr_kvdl_set(struct mlxsw_sp *mlxsw_sp,
+				    const struct in6_addr *addr6,
+				    u32 *p_kvdl_index)
+{
+	return mlxsw_sp_ipv6_addr_kvdl_index_get(mlxsw_sp, addr6, p_kvdl_index);
+}
+
+void mlxsw_sp_nve_ipv6_addr_kvdl_unset(struct mlxsw_sp *mlxsw_sp,
+				       const struct in6_addr *addr6)
+{
+	mlxsw_sp_ipv6_addr_put(mlxsw_sp, addr6);
+}
+
+static struct mlxsw_sp_nve_ipv6_ht_node *
+mlxsw_sp_nve_ipv6_ht_node_lookup(struct mlxsw_sp *mlxsw_sp, const char *mac,
+				 u16 fid_index)
+{
+	struct mlxsw_sp_nve_ipv6_ht_key key = {};
+
+	ether_addr_copy(key.mac, mac);
+	key.fid_index = fid_index;
+	return rhashtable_lookup_fast(&mlxsw_sp->nve->ipv6_ht, &key,
+				      mlxsw_sp_nve_ipv6_ht_params);
+}
+
+static int mlxsw_sp_nve_ipv6_ht_insert(struct mlxsw_sp *mlxsw_sp,
+				       const char *mac, u16 fid_index,
+				       const struct in6_addr *addr6)
+{
+	struct mlxsw_sp_nve_ipv6_ht_node *ipv6_ht_node;
+	struct mlxsw_sp_nve *nve = mlxsw_sp->nve;
+	int err;
+
+	ipv6_ht_node = kzalloc(sizeof(*ipv6_ht_node), GFP_KERNEL);
+	if (!ipv6_ht_node)
+		return -ENOMEM;
+
+	ether_addr_copy(ipv6_ht_node->key.mac, mac);
+	ipv6_ht_node->key.fid_index = fid_index;
+	ipv6_ht_node->addr6 = *addr6;
+
+	err = rhashtable_insert_fast(&nve->ipv6_ht, &ipv6_ht_node->ht_node,
+				     mlxsw_sp_nve_ipv6_ht_params);
+	if (err)
+		goto err_rhashtable_insert;
+
+	list_add(&ipv6_ht_node->list, &nve->ipv6_addr_list);
+
+	return 0;
+
+err_rhashtable_insert:
+	kfree(ipv6_ht_node);
+	return err;
+}
+
+static void
+mlxsw_sp_nve_ipv6_ht_remove(struct mlxsw_sp *mlxsw_sp,
+			    struct mlxsw_sp_nve_ipv6_ht_node *ipv6_ht_node)
+{
+	struct mlxsw_sp_nve *nve = mlxsw_sp->nve;
+
+	list_del(&ipv6_ht_node->list);
+	rhashtable_remove_fast(&nve->ipv6_ht, &ipv6_ht_node->ht_node,
+			       mlxsw_sp_nve_ipv6_ht_params);
+	kfree(ipv6_ht_node);
+}
+
+int
+mlxsw_sp_nve_ipv6_addr_map_replace(struct mlxsw_sp *mlxsw_sp, const char *mac,
+				   u16 fid_index,
+				   const struct in6_addr *new_addr6)
+{
+	struct mlxsw_sp_nve_ipv6_ht_node *ipv6_ht_node;
+
+	ASSERT_RTNL();
+
+	ipv6_ht_node = mlxsw_sp_nve_ipv6_ht_node_lookup(mlxsw_sp, mac,
+							fid_index);
+	if (!ipv6_ht_node)
+		return mlxsw_sp_nve_ipv6_ht_insert(mlxsw_sp, mac, fid_index,
+						   new_addr6);
+
+	mlxsw_sp_ipv6_addr_put(mlxsw_sp, &ipv6_ht_node->addr6);
+	ipv6_ht_node->addr6 = *new_addr6;
+	return 0;
+}
+
+void mlxsw_sp_nve_ipv6_addr_map_del(struct mlxsw_sp *mlxsw_sp, const char *mac,
+				    u16 fid_index)
+{
+	struct mlxsw_sp_nve_ipv6_ht_node *ipv6_ht_node;
+
+	ASSERT_RTNL();
+
+	ipv6_ht_node = mlxsw_sp_nve_ipv6_ht_node_lookup(mlxsw_sp, mac,
+							fid_index);
+	if (WARN_ON(!ipv6_ht_node))
+		return;
+
+	mlxsw_sp_nve_ipv6_ht_remove(mlxsw_sp, ipv6_ht_node);
+}
+
+static void mlxsw_sp_nve_ipv6_addr_flush_by_fid(struct mlxsw_sp *mlxsw_sp,
+						u16 fid_index)
+{
+	struct mlxsw_sp_nve_ipv6_ht_node *ipv6_ht_node, *tmp;
+	struct mlxsw_sp_nve *nve = mlxsw_sp->nve;
+
+	list_for_each_entry_safe(ipv6_ht_node, tmp, &nve->ipv6_addr_list,
+				 list) {
+		if (ipv6_ht_node->key.fid_index != fid_index)
+			continue;
+
+		mlxsw_sp_ipv6_addr_put(mlxsw_sp, &ipv6_ht_node->addr6);
+		mlxsw_sp_nve_ipv6_ht_remove(mlxsw_sp, ipv6_ht_node);
+	}
+}
+
+int mlxsw_sp_nve_fid_enable(struct mlxsw_sp *mlxsw_sp, struct mlxsw_sp_fid *fid,
+			    struct mlxsw_sp_nve_params *params,
+			    struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp_nve *nve = mlxsw_sp->nve;
+	const struct mlxsw_sp_nve_ops *ops;
+	struct mlxsw_sp_nve_config config;
+	int err;
+
+	ops = nve->nve_ops_arr[params->type];
+
+	if (!ops->can_offload(nve, params, extack))
+		return -EINVAL;
+
+	memset(&config, 0, sizeof(config));
+	ops->nve_config(nve, params, &config);
+	if (nve->num_nve_tunnels &&
+	    memcmp(&config, &nve->config, sizeof(config))) {
+		NL_SET_ERR_MSG_MOD(extack, "Conflicting NVE tunnels configuration");
+		return -EINVAL;
+	}
+
+	err = mlxsw_sp_nve_tunnel_init(mlxsw_sp, &config);
+	if (err) {
+		NL_SET_ERR_MSG_MOD(extack, "Failed to initialize NVE tunnel");
+		return err;
+	}
+
+	err = mlxsw_sp_fid_vni_set(fid, params->type, params->vni,
+				   params->dev->ifindex);
+	if (err) {
+		NL_SET_ERR_MSG_MOD(extack, "Failed to set VNI on FID");
+		goto err_fid_vni_set;
+	}
+
+	err = ops->fdb_replay(params->dev, params->vni, extack);
+	if (err)
+		goto err_fdb_replay;
+
+	return 0;
+
+err_fdb_replay:
+	mlxsw_sp_fid_vni_clear(fid);
+err_fid_vni_set:
+	mlxsw_sp_nve_tunnel_fini(mlxsw_sp);
+	return err;
+}
+
+void mlxsw_sp_nve_fid_disable(struct mlxsw_sp *mlxsw_sp,
+			      struct mlxsw_sp_fid *fid)
+{
+	u16 fid_index = mlxsw_sp_fid_index(fid);
+	struct net_device *nve_dev;
+	int nve_ifindex;
+	__be32 vni;
+
+	mlxsw_sp_nve_flood_ip_flush(mlxsw_sp, fid);
+	mlxsw_sp_nve_fdb_flush_by_fid(mlxsw_sp, fid_index);
+	mlxsw_sp_nve_ipv6_addr_flush_by_fid(mlxsw_sp, fid_index);
+
+	if (WARN_ON(mlxsw_sp_fid_nve_ifindex(fid, &nve_ifindex) ||
+		    mlxsw_sp_fid_vni(fid, &vni)))
+		goto out;
+
+	nve_dev = dev_get_by_index(mlxsw_sp_net(mlxsw_sp), nve_ifindex);
+	if (!nve_dev)
+		goto out;
+
+	mlxsw_sp_nve_fdb_clear_offload(mlxsw_sp, fid, nve_dev, vni);
+	mlxsw_sp_fid_fdb_clear_offload(fid, nve_dev);
+
+	dev_put(nve_dev);
+
+out:
+	mlxsw_sp_fid_vni_clear(fid);
+	mlxsw_sp_nve_tunnel_fini(mlxsw_sp);
+}
+
+int mlxsw_sp_port_nve_init(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	char tnqdr_pl[MLXSW_REG_TNQDR_LEN];
+
+	mlxsw_reg_tnqdr_pack(tnqdr_pl, mlxsw_sp_port->local_port);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(tnqdr), tnqdr_pl);
+}
+
+void mlxsw_sp_port_nve_fini(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+}
+
+static int mlxsw_sp_nve_qos_init(struct mlxsw_sp *mlxsw_sp)
+{
+	char tnqcr_pl[MLXSW_REG_TNQCR_LEN];
+
+	mlxsw_reg_tnqcr_pack(tnqcr_pl);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(tnqcr), tnqcr_pl);
+}
+
+static int mlxsw_sp_nve_ecn_encap_init(struct mlxsw_sp *mlxsw_sp)
+{
+	int i;
+
+	/* Iterate over inner ECN values */
+	for (i = INET_ECN_NOT_ECT; i <= INET_ECN_CE; i++) {
+		u8 outer_ecn = INET_ECN_encapsulate(0, i);
+		char tneem_pl[MLXSW_REG_TNEEM_LEN];
+		int err;
+
+		mlxsw_reg_tneem_pack(tneem_pl, i, outer_ecn);
+		err = mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(tneem),
+				      tneem_pl);
+		if (err)
+			return err;
+	}
+
+	return 0;
+}
+
+static int __mlxsw_sp_nve_ecn_decap_init(struct mlxsw_sp *mlxsw_sp,
+					 u8 inner_ecn, u8 outer_ecn)
+{
+	char tndem_pl[MLXSW_REG_TNDEM_LEN];
+	u8 new_inner_ecn;
+	bool trap_en;
+
+	new_inner_ecn = mlxsw_sp_tunnel_ecn_decap(outer_ecn, inner_ecn,
+						  &trap_en);
+	mlxsw_reg_tndem_pack(tndem_pl, outer_ecn, inner_ecn, new_inner_ecn,
+			     trap_en, trap_en ? MLXSW_TRAP_ID_DECAP_ECN0 : 0);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(tndem), tndem_pl);
+}
+
+static int mlxsw_sp_nve_ecn_decap_init(struct mlxsw_sp *mlxsw_sp)
+{
+	int i;
+
+	/* Iterate over inner ECN values */
+	for (i = INET_ECN_NOT_ECT; i <= INET_ECN_CE; i++) {
+		int j;
+
+		/* Iterate over outer ECN values */
+		for (j = INET_ECN_NOT_ECT; j <= INET_ECN_CE; j++) {
+			int err;
+
+			err = __mlxsw_sp_nve_ecn_decap_init(mlxsw_sp, i, j);
+			if (err)
+				return err;
+		}
+	}
+
+	return 0;
+}
+
+static int mlxsw_sp_nve_ecn_init(struct mlxsw_sp *mlxsw_sp)
+{
+	int err;
+
+	err = mlxsw_sp_nve_ecn_encap_init(mlxsw_sp);
+	if (err)
+		return err;
+
+	return mlxsw_sp_nve_ecn_decap_init(mlxsw_sp);
+}
+
+static int mlxsw_sp_nve_resources_query(struct mlxsw_sp *mlxsw_sp)
+{
+	unsigned int max;
+
+	if (!MLXSW_CORE_RES_VALID(mlxsw_sp->core, MAX_NVE_MC_ENTRIES_IPV4) ||
+	    !MLXSW_CORE_RES_VALID(mlxsw_sp->core, MAX_NVE_MC_ENTRIES_IPV6))
+		return -EIO;
+	max = MLXSW_CORE_RES_GET(mlxsw_sp->core, MAX_NVE_MC_ENTRIES_IPV4);
+	mlxsw_sp->nve->num_max_mc_entries[MLXSW_SP_L3_PROTO_IPV4] = max;
+	max = MLXSW_CORE_RES_GET(mlxsw_sp->core, MAX_NVE_MC_ENTRIES_IPV6);
+	mlxsw_sp->nve->num_max_mc_entries[MLXSW_SP_L3_PROTO_IPV6] = max;
+
+	return 0;
+}
+
+int mlxsw_sp_nve_init(struct mlxsw_sp *mlxsw_sp)
+{
+	struct mlxsw_sp_nve *nve;
+	int err;
+
+	nve = kzalloc(sizeof(*mlxsw_sp->nve), GFP_KERNEL);
+	if (!nve)
+		return -ENOMEM;
+	mlxsw_sp->nve = nve;
+	nve->mlxsw_sp = mlxsw_sp;
+	nve->nve_ops_arr = mlxsw_sp->nve_ops_arr;
+
+	err = rhashtable_init(&nve->mc_list_ht,
+			      &mlxsw_sp_nve_mc_list_ht_params);
+	if (err)
+		goto err_mc_rhashtable_init;
+
+	err = rhashtable_init(&nve->ipv6_ht, &mlxsw_sp_nve_ipv6_ht_params);
+	if (err)
+		goto err_ipv6_rhashtable_init;
+
+	INIT_LIST_HEAD(&nve->ipv6_addr_list);
+
+	err = mlxsw_sp_nve_qos_init(mlxsw_sp);
+	if (err)
+		goto err_nve_qos_init;
+
+	err = mlxsw_sp_nve_ecn_init(mlxsw_sp);
+	if (err)
+		goto err_nve_ecn_init;
+
+	err = mlxsw_sp_nve_resources_query(mlxsw_sp);
+	if (err)
+		goto err_nve_resources_query;
+
+	return 0;
+
+err_nve_resources_query:
+err_nve_ecn_init:
+err_nve_qos_init:
+	rhashtable_destroy(&nve->ipv6_ht);
+err_ipv6_rhashtable_init:
+	rhashtable_destroy(&nve->mc_list_ht);
+err_mc_rhashtable_init:
+	mlxsw_sp->nve = NULL;
+	kfree(nve);
+	return err;
+}
+
+void mlxsw_sp_nve_fini(struct mlxsw_sp *mlxsw_sp)
+{
+	WARN_ON(mlxsw_sp->nve->num_nve_tunnels);
+	WARN_ON(!list_empty(&mlxsw_sp->nve->ipv6_addr_list));
+	rhashtable_destroy(&mlxsw_sp->nve->ipv6_ht);
+	rhashtable_destroy(&mlxsw_sp->nve->mc_list_ht);
+	kfree(mlxsw_sp->nve);
+	mlxsw_sp->nve = NULL;
+}
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_nve.h b/drivers/net/ethernet/mellanox/mlxsw/spectrum_nve.h
new file mode 100644
index 000000000..0d21de1d0
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_nve.h
@@ -0,0 +1,55 @@
+/* SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0 */
+/* Copyright (c) 2018 Mellanox Technologies. All rights reserved */
+
+#ifndef _MLXSW_SPECTRUM_NVE_H
+#define _MLXSW_SPECTRUM_NVE_H
+
+#include <linux/netlink.h>
+#include <linux/rhashtable.h>
+
+#include "spectrum.h"
+
+struct mlxsw_sp_nve_config {
+	enum mlxsw_sp_nve_type type;
+	u8 ttl;
+	u8 learning_en:1;
+	__be16 udp_dport;
+	__be32 flowlabel;
+	u32 ul_tb_id;
+	enum mlxsw_sp_l3proto ul_proto;
+	union mlxsw_sp_l3addr ul_sip;
+};
+
+struct mlxsw_sp_nve {
+	struct mlxsw_sp_nve_config config;
+	struct rhashtable mc_list_ht;
+	struct rhashtable ipv6_ht;
+	struct list_head ipv6_addr_list; /* Saves hash table nodes. */
+	struct mlxsw_sp *mlxsw_sp;
+	const struct mlxsw_sp_nve_ops **nve_ops_arr;
+	unsigned int num_nve_tunnels;	/* Protected by RTNL */
+	unsigned int num_max_mc_entries[MLXSW_SP_L3_PROTO_MAX];
+	u32 tunnel_index;
+	u16 ul_rif_index;	/* Reserved for Spectrum */
+};
+
+struct mlxsw_sp_nve_ops {
+	enum mlxsw_sp_nve_type type;
+	bool (*can_offload)(const struct mlxsw_sp_nve *nve,
+			    const struct mlxsw_sp_nve_params *params,
+			    struct netlink_ext_ack *extack);
+	void (*nve_config)(const struct mlxsw_sp_nve *nve,
+			   const struct mlxsw_sp_nve_params *params,
+			   struct mlxsw_sp_nve_config *config);
+	int (*init)(struct mlxsw_sp_nve *nve,
+		    const struct mlxsw_sp_nve_config *config);
+	void (*fini)(struct mlxsw_sp_nve *nve);
+	int (*fdb_replay)(const struct net_device *nve_dev, __be32 vni,
+			  struct netlink_ext_ack *extack);
+	void (*fdb_clear_offload)(const struct net_device *nve_dev, __be32 vni);
+};
+
+extern const struct mlxsw_sp_nve_ops mlxsw_sp1_nve_vxlan_ops;
+extern const struct mlxsw_sp_nve_ops mlxsw_sp2_nve_vxlan_ops;
+
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_nve_vxlan.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_nve_vxlan.c
new file mode 100644
index 000000000..cdd8818b4
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_nve_vxlan.c
@@ -0,0 +1,473 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2018 Mellanox Technologies. All rights reserved */
+
+#include <linux/netdevice.h>
+#include <linux/netlink.h>
+#include <linux/random.h>
+#include <net/vxlan.h>
+
+#include "reg.h"
+#include "spectrum.h"
+#include "spectrum_nve.h"
+
+#define MLXSW_SP_NVE_VXLAN_IPV4_SUPPORTED_FLAGS (VXLAN_F_UDP_ZERO_CSUM_TX | \
+						 VXLAN_F_LEARN)
+#define MLXSW_SP_NVE_VXLAN_IPV6_SUPPORTED_FLAGS (VXLAN_F_IPV6 | \
+						 VXLAN_F_UDP_ZERO_CSUM6_TX | \
+						 VXLAN_F_UDP_ZERO_CSUM6_RX)
+
+static bool mlxsw_sp_nve_vxlan_ipv4_flags_check(const struct vxlan_config *cfg,
+						struct netlink_ext_ack *extack)
+{
+	if (!(cfg->flags & VXLAN_F_UDP_ZERO_CSUM_TX)) {
+		NL_SET_ERR_MSG_MOD(extack, "VxLAN: Zero UDP checksum must be allowed for TX");
+		return false;
+	}
+
+	if (cfg->flags & ~MLXSW_SP_NVE_VXLAN_IPV4_SUPPORTED_FLAGS) {
+		NL_SET_ERR_MSG_MOD(extack, "VxLAN: Unsupported flag");
+		return false;
+	}
+
+	return true;
+}
+
+static bool mlxsw_sp_nve_vxlan_ipv6_flags_check(const struct vxlan_config *cfg,
+						struct netlink_ext_ack *extack)
+{
+	if (!(cfg->flags & VXLAN_F_UDP_ZERO_CSUM6_TX)) {
+		NL_SET_ERR_MSG_MOD(extack, "VxLAN: Zero UDP checksum must be allowed for TX");
+		return false;
+	}
+
+	if (!(cfg->flags & VXLAN_F_UDP_ZERO_CSUM6_RX)) {
+		NL_SET_ERR_MSG_MOD(extack, "VxLAN: Zero UDP checksum must be allowed for RX");
+		return false;
+	}
+
+	if (cfg->flags & ~MLXSW_SP_NVE_VXLAN_IPV6_SUPPORTED_FLAGS) {
+		NL_SET_ERR_MSG_MOD(extack, "VxLAN: Unsupported flag");
+		return false;
+	}
+
+	return true;
+}
+
+static bool mlxsw_sp_nve_vxlan_can_offload(const struct mlxsw_sp_nve *nve,
+					   const struct mlxsw_sp_nve_params *params,
+					   struct netlink_ext_ack *extack)
+{
+	struct vxlan_dev *vxlan = netdev_priv(params->dev);
+	struct vxlan_config *cfg = &vxlan->cfg;
+
+	if (vxlan_addr_multicast(&cfg->remote_ip)) {
+		NL_SET_ERR_MSG_MOD(extack, "VxLAN: Multicast destination IP is not supported");
+		return false;
+	}
+
+	if (vxlan_addr_any(&cfg->saddr)) {
+		NL_SET_ERR_MSG_MOD(extack, "VxLAN: Source address must be specified");
+		return false;
+	}
+
+	if (cfg->remote_ifindex) {
+		NL_SET_ERR_MSG_MOD(extack, "VxLAN: Local interface is not supported");
+		return false;
+	}
+
+	if (cfg->port_min || cfg->port_max) {
+		NL_SET_ERR_MSG_MOD(extack, "VxLAN: Only default UDP source port range is supported");
+		return false;
+	}
+
+	if (cfg->tos != 1) {
+		NL_SET_ERR_MSG_MOD(extack, "VxLAN: TOS must be configured to inherit");
+		return false;
+	}
+
+	if (cfg->flags & VXLAN_F_TTL_INHERIT) {
+		NL_SET_ERR_MSG_MOD(extack, "VxLAN: TTL must not be configured to inherit");
+		return false;
+	}
+
+	switch (cfg->saddr.sa.sa_family) {
+	case AF_INET:
+		if (!mlxsw_sp_nve_vxlan_ipv4_flags_check(cfg, extack))
+			return false;
+		break;
+	case AF_INET6:
+		if (!mlxsw_sp_nve_vxlan_ipv6_flags_check(cfg, extack))
+			return false;
+		break;
+	}
+
+	if (cfg->ttl == 0) {
+		NL_SET_ERR_MSG_MOD(extack, "VxLAN: TTL must not be configured to 0");
+		return false;
+	}
+
+	if (cfg->label != 0) {
+		NL_SET_ERR_MSG_MOD(extack, "VxLAN: Flow label must be configured to 0");
+		return false;
+	}
+
+	return true;
+}
+
+static bool mlxsw_sp1_nve_vxlan_can_offload(const struct mlxsw_sp_nve *nve,
+					    const struct mlxsw_sp_nve_params *params,
+					    struct netlink_ext_ack *extack)
+{
+	if (params->ethertype == ETH_P_8021AD) {
+		NL_SET_ERR_MSG_MOD(extack, "VxLAN: 802.1ad bridge is not supported with VxLAN");
+		return false;
+	}
+
+	return mlxsw_sp_nve_vxlan_can_offload(nve, params, extack);
+}
+
+static void
+mlxsw_sp_nve_vxlan_ul_proto_sip_config(const struct vxlan_config *cfg,
+				       struct mlxsw_sp_nve_config *config)
+{
+	switch (cfg->saddr.sa.sa_family) {
+	case AF_INET:
+		config->ul_proto = MLXSW_SP_L3_PROTO_IPV4;
+		config->ul_sip.addr4 = cfg->saddr.sin.sin_addr.s_addr;
+		break;
+	case AF_INET6:
+		config->ul_proto = MLXSW_SP_L3_PROTO_IPV6;
+		config->ul_sip.addr6 = cfg->saddr.sin6.sin6_addr;
+		break;
+	}
+}
+
+static void mlxsw_sp_nve_vxlan_config(const struct mlxsw_sp_nve *nve,
+				      const struct mlxsw_sp_nve_params *params,
+				      struct mlxsw_sp_nve_config *config)
+{
+	struct vxlan_dev *vxlan = netdev_priv(params->dev);
+	struct vxlan_config *cfg = &vxlan->cfg;
+
+	config->type = MLXSW_SP_NVE_TYPE_VXLAN;
+	config->ttl = cfg->ttl;
+	config->flowlabel = cfg->label;
+	config->learning_en = cfg->flags & VXLAN_F_LEARN ? 1 : 0;
+	config->ul_tb_id = RT_TABLE_MAIN;
+	mlxsw_sp_nve_vxlan_ul_proto_sip_config(cfg, config);
+	config->udp_dport = cfg->dst_port;
+}
+
+static void
+mlxsw_sp_nve_vxlan_config_prepare(char *tngcr_pl,
+				  const struct mlxsw_sp_nve_config *config)
+{
+	struct in6_addr addr6;
+	u8 udp_sport;
+
+	mlxsw_reg_tngcr_pack(tngcr_pl, MLXSW_REG_TNGCR_TYPE_VXLAN, true,
+			     config->ttl);
+	/* VxLAN driver's default UDP source port range is 32768 (0x8000)
+	 * to 60999 (0xee47). Set the upper 8 bits of the UDP source port
+	 * to a random number between 0x80 and 0xee
+	 */
+	get_random_bytes(&udp_sport, sizeof(udp_sport));
+	udp_sport = (udp_sport % (0xee - 0x80 + 1)) + 0x80;
+	mlxsw_reg_tngcr_nve_udp_sport_prefix_set(tngcr_pl, udp_sport);
+
+	switch (config->ul_proto) {
+	case MLXSW_SP_L3_PROTO_IPV4:
+		mlxsw_reg_tngcr_usipv4_set(tngcr_pl,
+					   be32_to_cpu(config->ul_sip.addr4));
+		break;
+	case MLXSW_SP_L3_PROTO_IPV6:
+		addr6 = config->ul_sip.addr6;
+		mlxsw_reg_tngcr_usipv6_memcpy_to(tngcr_pl,
+						 (const char *)&addr6);
+		break;
+	}
+}
+
+static int
+mlxsw_sp1_nve_vxlan_config_set(struct mlxsw_sp *mlxsw_sp,
+			       const struct mlxsw_sp_nve_config *config)
+{
+	char tngcr_pl[MLXSW_REG_TNGCR_LEN];
+	u16 ul_vr_id;
+	int err;
+
+	err = mlxsw_sp_router_tb_id_vr_id(mlxsw_sp, config->ul_tb_id,
+					  &ul_vr_id);
+	if (err)
+		return err;
+
+	mlxsw_sp_nve_vxlan_config_prepare(tngcr_pl, config);
+	mlxsw_reg_tngcr_learn_enable_set(tngcr_pl, config->learning_en);
+	mlxsw_reg_tngcr_underlay_virtual_router_set(tngcr_pl, ul_vr_id);
+
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(tngcr), tngcr_pl);
+}
+
+static void mlxsw_sp1_nve_vxlan_config_clear(struct mlxsw_sp *mlxsw_sp)
+{
+	char tngcr_pl[MLXSW_REG_TNGCR_LEN];
+
+	mlxsw_reg_tngcr_pack(tngcr_pl, MLXSW_REG_TNGCR_TYPE_VXLAN, false, 0);
+
+	mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(tngcr), tngcr_pl);
+}
+
+static int mlxsw_sp1_nve_vxlan_rtdp_set(struct mlxsw_sp *mlxsw_sp,
+					unsigned int tunnel_index)
+{
+	char rtdp_pl[MLXSW_REG_RTDP_LEN];
+
+	mlxsw_reg_rtdp_pack(rtdp_pl, MLXSW_REG_RTDP_TYPE_NVE, tunnel_index);
+
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(rtdp), rtdp_pl);
+}
+
+static int mlxsw_sp1_nve_vxlan_init(struct mlxsw_sp_nve *nve,
+				    const struct mlxsw_sp_nve_config *config)
+{
+	struct mlxsw_sp *mlxsw_sp = nve->mlxsw_sp;
+	int err;
+
+	err = mlxsw_sp_parsing_vxlan_udp_dport_set(mlxsw_sp, config->udp_dport);
+	if (err)
+		return err;
+
+	err = mlxsw_sp_parsing_depth_inc(mlxsw_sp);
+	if (err)
+		goto err_parsing_depth_inc;
+
+	err = mlxsw_sp1_nve_vxlan_config_set(mlxsw_sp, config);
+	if (err)
+		goto err_config_set;
+
+	err = mlxsw_sp1_nve_vxlan_rtdp_set(mlxsw_sp, nve->tunnel_index);
+	if (err)
+		goto err_rtdp_set;
+
+	err = mlxsw_sp_router_nve_promote_decap(mlxsw_sp, config->ul_tb_id,
+						config->ul_proto,
+						&config->ul_sip,
+						nve->tunnel_index);
+	if (err)
+		goto err_promote_decap;
+
+	return 0;
+
+err_promote_decap:
+err_rtdp_set:
+	mlxsw_sp1_nve_vxlan_config_clear(mlxsw_sp);
+err_config_set:
+	mlxsw_sp_parsing_depth_dec(mlxsw_sp);
+err_parsing_depth_inc:
+	mlxsw_sp_parsing_vxlan_udp_dport_set(mlxsw_sp, 0);
+	return err;
+}
+
+static void mlxsw_sp1_nve_vxlan_fini(struct mlxsw_sp_nve *nve)
+{
+	struct mlxsw_sp_nve_config *config = &nve->config;
+	struct mlxsw_sp *mlxsw_sp = nve->mlxsw_sp;
+
+	mlxsw_sp_router_nve_demote_decap(mlxsw_sp, config->ul_tb_id,
+					 config->ul_proto, &config->ul_sip);
+	mlxsw_sp1_nve_vxlan_config_clear(mlxsw_sp);
+	mlxsw_sp_parsing_depth_dec(mlxsw_sp);
+	mlxsw_sp_parsing_vxlan_udp_dport_set(mlxsw_sp, 0);
+}
+
+static int
+mlxsw_sp_nve_vxlan_fdb_replay(const struct net_device *nve_dev, __be32 vni,
+			      struct netlink_ext_ack *extack)
+{
+	if (WARN_ON(!netif_is_vxlan(nve_dev)))
+		return -EINVAL;
+	return vxlan_fdb_replay(nve_dev, vni, &mlxsw_sp_switchdev_notifier,
+				extack);
+}
+
+static void
+mlxsw_sp_nve_vxlan_clear_offload(const struct net_device *nve_dev, __be32 vni)
+{
+	if (WARN_ON(!netif_is_vxlan(nve_dev)))
+		return;
+	vxlan_fdb_clear_offload(nve_dev, vni);
+}
+
+const struct mlxsw_sp_nve_ops mlxsw_sp1_nve_vxlan_ops = {
+	.type		= MLXSW_SP_NVE_TYPE_VXLAN,
+	.can_offload	= mlxsw_sp1_nve_vxlan_can_offload,
+	.nve_config	= mlxsw_sp_nve_vxlan_config,
+	.init		= mlxsw_sp1_nve_vxlan_init,
+	.fini		= mlxsw_sp1_nve_vxlan_fini,
+	.fdb_replay	= mlxsw_sp_nve_vxlan_fdb_replay,
+	.fdb_clear_offload = mlxsw_sp_nve_vxlan_clear_offload,
+};
+
+static int mlxsw_sp2_nve_vxlan_learning_set(struct mlxsw_sp *mlxsw_sp,
+					    bool learning_en)
+{
+	char tnpc_pl[MLXSW_REG_TNPC_LEN];
+
+	mlxsw_reg_tnpc_pack(tnpc_pl, MLXSW_REG_TUNNEL_PORT_NVE,
+			    learning_en);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(tnpc), tnpc_pl);
+}
+
+static int
+mlxsw_sp2_nve_decap_ethertype_set(struct mlxsw_sp *mlxsw_sp)
+{
+	char spvid_pl[MLXSW_REG_SPVID_LEN] = {};
+
+	mlxsw_reg_spvid_tport_set(spvid_pl, true);
+	mlxsw_reg_spvid_local_port_set(spvid_pl,
+				       MLXSW_REG_TUNNEL_PORT_NVE);
+	mlxsw_reg_spvid_egr_et_set_set(spvid_pl, true);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(spvid), spvid_pl);
+}
+
+static int
+mlxsw_sp2_nve_vxlan_config_set(struct mlxsw_sp *mlxsw_sp,
+			       const struct mlxsw_sp_nve_config *config)
+{
+	char tngcr_pl[MLXSW_REG_TNGCR_LEN];
+	char spvtr_pl[MLXSW_REG_SPVTR_LEN];
+	u16 ul_rif_index;
+	int err;
+
+	err = mlxsw_sp_router_ul_rif_get(mlxsw_sp, config->ul_tb_id,
+					 &ul_rif_index);
+	if (err)
+		return err;
+	mlxsw_sp->nve->ul_rif_index = ul_rif_index;
+
+	err = mlxsw_sp2_nve_vxlan_learning_set(mlxsw_sp, config->learning_en);
+	if (err)
+		goto err_vxlan_learning_set;
+
+	mlxsw_sp_nve_vxlan_config_prepare(tngcr_pl, config);
+	mlxsw_reg_tngcr_underlay_rif_set(tngcr_pl, ul_rif_index);
+
+	err = mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(tngcr), tngcr_pl);
+	if (err)
+		goto err_tngcr_write;
+
+	mlxsw_reg_spvtr_pack(spvtr_pl, true, MLXSW_REG_TUNNEL_PORT_NVE,
+			     MLXSW_REG_SPVTR_IPVID_MODE_ALWAYS_PUSH_VLAN);
+	err = mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(spvtr), spvtr_pl);
+	if (err)
+		goto err_spvtr_write;
+
+	err = mlxsw_sp2_nve_decap_ethertype_set(mlxsw_sp);
+	if (err)
+		goto err_decap_ethertype_set;
+
+	return 0;
+
+err_decap_ethertype_set:
+	mlxsw_reg_spvtr_pack(spvtr_pl, true, MLXSW_REG_TUNNEL_PORT_NVE,
+			     MLXSW_REG_SPVTR_IPVID_MODE_IEEE_COMPLIANT_PVID);
+	mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(spvtr), spvtr_pl);
+err_spvtr_write:
+	mlxsw_reg_tngcr_pack(tngcr_pl, MLXSW_REG_TNGCR_TYPE_VXLAN, false, 0);
+	mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(tngcr), tngcr_pl);
+err_tngcr_write:
+	mlxsw_sp2_nve_vxlan_learning_set(mlxsw_sp, false);
+err_vxlan_learning_set:
+	mlxsw_sp_router_ul_rif_put(mlxsw_sp, ul_rif_index);
+	return err;
+}
+
+static void mlxsw_sp2_nve_vxlan_config_clear(struct mlxsw_sp *mlxsw_sp)
+{
+	char spvtr_pl[MLXSW_REG_SPVTR_LEN];
+	char tngcr_pl[MLXSW_REG_TNGCR_LEN];
+
+	mlxsw_reg_spvtr_pack(spvtr_pl, true, MLXSW_REG_TUNNEL_PORT_NVE,
+			     MLXSW_REG_SPVTR_IPVID_MODE_IEEE_COMPLIANT_PVID);
+	mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(spvtr), spvtr_pl);
+	mlxsw_reg_tngcr_pack(tngcr_pl, MLXSW_REG_TNGCR_TYPE_VXLAN, false, 0);
+	mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(tngcr), tngcr_pl);
+	mlxsw_sp2_nve_vxlan_learning_set(mlxsw_sp, false);
+	mlxsw_sp_router_ul_rif_put(mlxsw_sp, mlxsw_sp->nve->ul_rif_index);
+}
+
+static int mlxsw_sp2_nve_vxlan_rtdp_set(struct mlxsw_sp *mlxsw_sp,
+					unsigned int tunnel_index,
+					u16 ul_rif_index)
+{
+	char rtdp_pl[MLXSW_REG_RTDP_LEN];
+
+	mlxsw_reg_rtdp_pack(rtdp_pl, MLXSW_REG_RTDP_TYPE_NVE, tunnel_index);
+	mlxsw_reg_rtdp_egress_router_interface_set(rtdp_pl, ul_rif_index);
+
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(rtdp), rtdp_pl);
+}
+
+static int mlxsw_sp2_nve_vxlan_init(struct mlxsw_sp_nve *nve,
+				    const struct mlxsw_sp_nve_config *config)
+{
+	struct mlxsw_sp *mlxsw_sp = nve->mlxsw_sp;
+	int err;
+
+	err = mlxsw_sp_parsing_vxlan_udp_dport_set(mlxsw_sp, config->udp_dport);
+	if (err)
+		return err;
+
+	err = mlxsw_sp_parsing_depth_inc(mlxsw_sp);
+	if (err)
+		goto err_parsing_depth_inc;
+
+	err = mlxsw_sp2_nve_vxlan_config_set(mlxsw_sp, config);
+	if (err)
+		goto err_config_set;
+
+	err = mlxsw_sp2_nve_vxlan_rtdp_set(mlxsw_sp, nve->tunnel_index,
+					   nve->ul_rif_index);
+	if (err)
+		goto err_rtdp_set;
+
+	err = mlxsw_sp_router_nve_promote_decap(mlxsw_sp, config->ul_tb_id,
+						config->ul_proto,
+						&config->ul_sip,
+						nve->tunnel_index);
+	if (err)
+		goto err_promote_decap;
+
+	return 0;
+
+err_promote_decap:
+err_rtdp_set:
+	mlxsw_sp2_nve_vxlan_config_clear(mlxsw_sp);
+err_config_set:
+	mlxsw_sp_parsing_depth_dec(mlxsw_sp);
+err_parsing_depth_inc:
+	mlxsw_sp_parsing_vxlan_udp_dport_set(mlxsw_sp, 0);
+	return err;
+}
+
+static void mlxsw_sp2_nve_vxlan_fini(struct mlxsw_sp_nve *nve)
+{
+	struct mlxsw_sp_nve_config *config = &nve->config;
+	struct mlxsw_sp *mlxsw_sp = nve->mlxsw_sp;
+
+	mlxsw_sp_router_nve_demote_decap(mlxsw_sp, config->ul_tb_id,
+					 config->ul_proto, &config->ul_sip);
+	mlxsw_sp2_nve_vxlan_config_clear(mlxsw_sp);
+	mlxsw_sp_parsing_depth_dec(mlxsw_sp);
+	mlxsw_sp_parsing_vxlan_udp_dport_set(mlxsw_sp, 0);
+}
+
+const struct mlxsw_sp_nve_ops mlxsw_sp2_nve_vxlan_ops = {
+	.type		= MLXSW_SP_NVE_TYPE_VXLAN,
+	.can_offload	= mlxsw_sp_nve_vxlan_can_offload,
+	.nve_config	= mlxsw_sp_nve_vxlan_config,
+	.init		= mlxsw_sp2_nve_vxlan_init,
+	.fini		= mlxsw_sp2_nve_vxlan_fini,
+	.fdb_replay	= mlxsw_sp_nve_vxlan_fdb_replay,
+	.fdb_clear_offload = mlxsw_sp_nve_vxlan_clear_offload,
+};
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_pgt.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_pgt.c
new file mode 100644
index 000000000..7dd3dba0f
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_pgt.c
@@ -0,0 +1,346 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2022, NVIDIA CORPORATION & AFFILIATES. All rights reserved. */
+
+#include <linux/refcount.h>
+#include <linux/idr.h>
+
+#include "spectrum.h"
+#include "reg.h"
+
+struct mlxsw_sp_pgt {
+	struct idr pgt_idr;
+	u16 end_index; /* Exclusive. */
+	struct mutex lock; /* Protects PGT. */
+	bool smpe_index_valid;
+};
+
+struct mlxsw_sp_pgt_entry {
+	struct list_head ports_list;
+	u16 index;
+	u16 smpe_index;
+};
+
+struct mlxsw_sp_pgt_entry_port {
+	struct list_head list; /* Member of 'ports_list'. */
+	u16 local_port;
+};
+
+int mlxsw_sp_pgt_mid_alloc(struct mlxsw_sp *mlxsw_sp, u16 *p_mid)
+{
+	int index, err = 0;
+
+	mutex_lock(&mlxsw_sp->pgt->lock);
+	index = idr_alloc(&mlxsw_sp->pgt->pgt_idr, NULL, 0,
+			  mlxsw_sp->pgt->end_index, GFP_KERNEL);
+
+	if (index < 0) {
+		err = index;
+		goto err_idr_alloc;
+	}
+
+	*p_mid = index;
+	mutex_unlock(&mlxsw_sp->pgt->lock);
+	return 0;
+
+err_idr_alloc:
+	mutex_unlock(&mlxsw_sp->pgt->lock);
+	return err;
+}
+
+void mlxsw_sp_pgt_mid_free(struct mlxsw_sp *mlxsw_sp, u16 mid_base)
+{
+	mutex_lock(&mlxsw_sp->pgt->lock);
+	WARN_ON(idr_remove(&mlxsw_sp->pgt->pgt_idr, mid_base));
+	mutex_unlock(&mlxsw_sp->pgt->lock);
+}
+
+int
+mlxsw_sp_pgt_mid_alloc_range(struct mlxsw_sp *mlxsw_sp, u16 mid_base, u16 count)
+{
+	unsigned int idr_cursor;
+	int i, err;
+
+	mutex_lock(&mlxsw_sp->pgt->lock);
+
+	/* This function is supposed to be called several times as part of
+	 * driver init, in specific order. Verify that the mid_index is the
+	 * first free index in the idr, to be able to free the indexes in case
+	 * of error.
+	 */
+	idr_cursor = idr_get_cursor(&mlxsw_sp->pgt->pgt_idr);
+	if (WARN_ON(idr_cursor != mid_base)) {
+		err = -EINVAL;
+		goto err_idr_cursor;
+	}
+
+	for (i = 0; i < count; i++) {
+		err = idr_alloc_cyclic(&mlxsw_sp->pgt->pgt_idr, NULL,
+				       mid_base, mid_base + count, GFP_KERNEL);
+		if (err < 0)
+			goto err_idr_alloc_cyclic;
+	}
+
+	mutex_unlock(&mlxsw_sp->pgt->lock);
+	return 0;
+
+err_idr_alloc_cyclic:
+	for (i--; i >= 0; i--)
+		idr_remove(&mlxsw_sp->pgt->pgt_idr, mid_base + i);
+err_idr_cursor:
+	mutex_unlock(&mlxsw_sp->pgt->lock);
+	return err;
+}
+
+void
+mlxsw_sp_pgt_mid_free_range(struct mlxsw_sp *mlxsw_sp, u16 mid_base, u16 count)
+{
+	struct idr *pgt_idr = &mlxsw_sp->pgt->pgt_idr;
+	int i;
+
+	mutex_lock(&mlxsw_sp->pgt->lock);
+
+	for (i = 0; i < count; i++)
+		WARN_ON_ONCE(idr_remove(pgt_idr, mid_base + i));
+
+	mutex_unlock(&mlxsw_sp->pgt->lock);
+}
+
+static struct mlxsw_sp_pgt_entry_port *
+mlxsw_sp_pgt_entry_port_lookup(struct mlxsw_sp_pgt_entry *pgt_entry,
+			       u16 local_port)
+{
+	struct mlxsw_sp_pgt_entry_port *pgt_entry_port;
+
+	list_for_each_entry(pgt_entry_port, &pgt_entry->ports_list, list) {
+		if (pgt_entry_port->local_port == local_port)
+			return pgt_entry_port;
+	}
+
+	return NULL;
+}
+
+static struct mlxsw_sp_pgt_entry *
+mlxsw_sp_pgt_entry_create(struct mlxsw_sp_pgt *pgt, u16 mid, u16 smpe)
+{
+	struct mlxsw_sp_pgt_entry *pgt_entry;
+	void *ret;
+	int err;
+
+	pgt_entry = kzalloc(sizeof(*pgt_entry), GFP_KERNEL);
+	if (!pgt_entry)
+		return ERR_PTR(-ENOMEM);
+
+	ret = idr_replace(&pgt->pgt_idr, pgt_entry, mid);
+	if (IS_ERR(ret)) {
+		err = PTR_ERR(ret);
+		goto err_idr_replace;
+	}
+
+	INIT_LIST_HEAD(&pgt_entry->ports_list);
+	pgt_entry->index = mid;
+	pgt_entry->smpe_index = smpe;
+	return pgt_entry;
+
+err_idr_replace:
+	kfree(pgt_entry);
+	return ERR_PTR(err);
+}
+
+static void mlxsw_sp_pgt_entry_destroy(struct mlxsw_sp_pgt *pgt,
+				       struct mlxsw_sp_pgt_entry *pgt_entry)
+{
+	WARN_ON(!list_empty(&pgt_entry->ports_list));
+
+	pgt_entry = idr_replace(&pgt->pgt_idr, NULL, pgt_entry->index);
+	if (WARN_ON(IS_ERR(pgt_entry)))
+		return;
+
+	kfree(pgt_entry);
+}
+
+static struct mlxsw_sp_pgt_entry *
+mlxsw_sp_pgt_entry_get(struct mlxsw_sp_pgt *pgt, u16 mid, u16 smpe)
+{
+	struct mlxsw_sp_pgt_entry *pgt_entry;
+
+	pgt_entry = idr_find(&pgt->pgt_idr, mid);
+	if (pgt_entry)
+		return pgt_entry;
+
+	return mlxsw_sp_pgt_entry_create(pgt, mid, smpe);
+}
+
+static void mlxsw_sp_pgt_entry_put(struct mlxsw_sp_pgt *pgt, u16 mid)
+{
+	struct mlxsw_sp_pgt_entry *pgt_entry;
+
+	pgt_entry = idr_find(&pgt->pgt_idr, mid);
+	if (WARN_ON(!pgt_entry))
+		return;
+
+	if (list_empty(&pgt_entry->ports_list))
+		mlxsw_sp_pgt_entry_destroy(pgt, pgt_entry);
+}
+
+static void mlxsw_sp_pgt_smid2_port_set(char *smid2_pl, u16 local_port,
+					bool member)
+{
+	mlxsw_reg_smid2_port_set(smid2_pl, local_port, member);
+	mlxsw_reg_smid2_port_mask_set(smid2_pl, local_port, 1);
+}
+
+static int
+mlxsw_sp_pgt_entry_port_write(struct mlxsw_sp *mlxsw_sp,
+			      const struct mlxsw_sp_pgt_entry *pgt_entry,
+			      u16 local_port, bool member)
+{
+	char *smid2_pl;
+	int err;
+
+	smid2_pl = kmalloc(MLXSW_REG_SMID2_LEN, GFP_KERNEL);
+	if (!smid2_pl)
+		return -ENOMEM;
+
+	mlxsw_reg_smid2_pack(smid2_pl, pgt_entry->index, 0, 0,
+			     mlxsw_sp->pgt->smpe_index_valid,
+			     pgt_entry->smpe_index);
+
+	mlxsw_sp_pgt_smid2_port_set(smid2_pl, local_port, member);
+	err = mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(smid2), smid2_pl);
+
+	kfree(smid2_pl);
+
+	return err;
+}
+
+static struct mlxsw_sp_pgt_entry_port *
+mlxsw_sp_pgt_entry_port_create(struct mlxsw_sp *mlxsw_sp,
+			       struct mlxsw_sp_pgt_entry *pgt_entry,
+			       u16 local_port)
+{
+	struct mlxsw_sp_pgt_entry_port *pgt_entry_port;
+	int err;
+
+	pgt_entry_port = kzalloc(sizeof(*pgt_entry_port), GFP_KERNEL);
+	if (!pgt_entry_port)
+		return ERR_PTR(-ENOMEM);
+
+	err = mlxsw_sp_pgt_entry_port_write(mlxsw_sp, pgt_entry, local_port,
+					    true);
+	if (err)
+		goto err_pgt_entry_port_write;
+
+	pgt_entry_port->local_port = local_port;
+	list_add(&pgt_entry_port->list, &pgt_entry->ports_list);
+
+	return pgt_entry_port;
+
+err_pgt_entry_port_write:
+	kfree(pgt_entry_port);
+	return ERR_PTR(err);
+}
+
+static void
+mlxsw_sp_pgt_entry_port_destroy(struct mlxsw_sp *mlxsw_sp,
+				struct mlxsw_sp_pgt_entry *pgt_entry,
+				struct mlxsw_sp_pgt_entry_port *pgt_entry_port)
+
+{
+	list_del(&pgt_entry_port->list);
+	mlxsw_sp_pgt_entry_port_write(mlxsw_sp, pgt_entry,
+				      pgt_entry_port->local_port, false);
+	kfree(pgt_entry_port);
+}
+
+static int mlxsw_sp_pgt_entry_port_add(struct mlxsw_sp *mlxsw_sp, u16 mid,
+				       u16 smpe, u16 local_port)
+{
+	struct mlxsw_sp_pgt_entry_port *pgt_entry_port;
+	struct mlxsw_sp_pgt_entry *pgt_entry;
+	int err;
+
+	mutex_lock(&mlxsw_sp->pgt->lock);
+
+	pgt_entry = mlxsw_sp_pgt_entry_get(mlxsw_sp->pgt, mid, smpe);
+	if (IS_ERR(pgt_entry)) {
+		err = PTR_ERR(pgt_entry);
+		goto err_pgt_entry_get;
+	}
+
+	pgt_entry_port = mlxsw_sp_pgt_entry_port_create(mlxsw_sp, pgt_entry,
+							local_port);
+	if (IS_ERR(pgt_entry_port)) {
+		err = PTR_ERR(pgt_entry_port);
+		goto err_pgt_entry_port_get;
+	}
+
+	mutex_unlock(&mlxsw_sp->pgt->lock);
+	return 0;
+
+err_pgt_entry_port_get:
+	mlxsw_sp_pgt_entry_put(mlxsw_sp->pgt, mid);
+err_pgt_entry_get:
+	mutex_unlock(&mlxsw_sp->pgt->lock);
+	return err;
+}
+
+static void mlxsw_sp_pgt_entry_port_del(struct mlxsw_sp *mlxsw_sp,
+					u16 mid, u16 smpe, u16 local_port)
+{
+	struct mlxsw_sp_pgt_entry_port *pgt_entry_port;
+	struct mlxsw_sp_pgt_entry *pgt_entry;
+
+	mutex_lock(&mlxsw_sp->pgt->lock);
+
+	pgt_entry = idr_find(&mlxsw_sp->pgt->pgt_idr, mid);
+	if (!pgt_entry)
+		goto out;
+
+	pgt_entry_port = mlxsw_sp_pgt_entry_port_lookup(pgt_entry, local_port);
+	if (!pgt_entry_port)
+		goto out;
+
+	mlxsw_sp_pgt_entry_port_destroy(mlxsw_sp, pgt_entry, pgt_entry_port);
+	mlxsw_sp_pgt_entry_put(mlxsw_sp->pgt, mid);
+
+out:
+	mutex_unlock(&mlxsw_sp->pgt->lock);
+}
+
+int mlxsw_sp_pgt_entry_port_set(struct mlxsw_sp *mlxsw_sp, u16 mid,
+				u16 smpe, u16 local_port, bool member)
+{
+	if (member)
+		return mlxsw_sp_pgt_entry_port_add(mlxsw_sp, mid, smpe,
+						   local_port);
+
+	mlxsw_sp_pgt_entry_port_del(mlxsw_sp, mid, smpe, local_port);
+	return 0;
+}
+
+int mlxsw_sp_pgt_init(struct mlxsw_sp *mlxsw_sp)
+{
+	struct mlxsw_sp_pgt *pgt;
+
+	if (!MLXSW_CORE_RES_VALID(mlxsw_sp->core, PGT_SIZE))
+		return -EIO;
+
+	pgt = kzalloc(sizeof(*mlxsw_sp->pgt), GFP_KERNEL);
+	if (!pgt)
+		return -ENOMEM;
+
+	idr_init(&pgt->pgt_idr);
+	pgt->end_index = MLXSW_CORE_RES_GET(mlxsw_sp->core, PGT_SIZE);
+	mutex_init(&pgt->lock);
+	pgt->smpe_index_valid = mlxsw_sp->pgt_smpe_index_valid;
+	mlxsw_sp->pgt = pgt;
+	return 0;
+}
+
+void mlxsw_sp_pgt_fini(struct mlxsw_sp *mlxsw_sp)
+{
+	mutex_destroy(&mlxsw_sp->pgt->lock);
+	WARN_ON(!idr_is_empty(&mlxsw_sp->pgt->pgt_idr));
+	idr_destroy(&mlxsw_sp->pgt->pgt_idr);
+	kfree(mlxsw_sp->pgt);
+}
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_policer.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_policer.c
new file mode 100644
index 000000000..22ebb207c
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_policer.c
@@ -0,0 +1,468 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2020 Mellanox Technologies. All rights reserved */
+
+#include <linux/idr.h>
+#include <linux/log2.h>
+#include <linux/mutex.h>
+#include <linux/netlink.h>
+#include <net/devlink.h>
+
+#include "spectrum.h"
+
+struct mlxsw_sp_policer_family {
+	enum mlxsw_sp_policer_type type;
+	enum mlxsw_reg_qpcr_g qpcr_type;
+	struct mlxsw_sp *mlxsw_sp;
+	u16 start_index; /* Inclusive */
+	u16 end_index; /* Exclusive */
+	struct idr policer_idr;
+	struct mutex lock; /* Protects policer_idr */
+	atomic_t policers_count;
+	const struct mlxsw_sp_policer_family_ops *ops;
+};
+
+struct mlxsw_sp_policer {
+	struct mlxsw_sp_policer_params params;
+	u16 index;
+};
+
+struct mlxsw_sp_policer_family_ops {
+	int (*init)(struct mlxsw_sp_policer_family *family);
+	void (*fini)(struct mlxsw_sp_policer_family *family);
+	int (*policer_index_alloc)(struct mlxsw_sp_policer_family *family,
+				   struct mlxsw_sp_policer *policer);
+	struct mlxsw_sp_policer * (*policer_index_free)(struct mlxsw_sp_policer_family *family,
+							u16 policer_index);
+	int (*policer_init)(struct mlxsw_sp_policer_family *family,
+			    const struct mlxsw_sp_policer *policer);
+	int (*policer_params_check)(const struct mlxsw_sp_policer_family *family,
+				    const struct mlxsw_sp_policer_params *params,
+				    struct netlink_ext_ack *extack);
+};
+
+struct mlxsw_sp_policer_core {
+	struct mlxsw_sp_policer_family *family_arr[MLXSW_SP_POLICER_TYPE_MAX + 1];
+	const struct mlxsw_sp_policer_core_ops *ops;
+	u8 lowest_bs_bits;
+	u8 highest_bs_bits;
+};
+
+struct mlxsw_sp_policer_core_ops {
+	int (*init)(struct mlxsw_sp_policer_core *policer_core);
+};
+
+static u64 mlxsw_sp_policer_rate_bytes_ps_kbps(u64 rate_bytes_ps)
+{
+	return div_u64(rate_bytes_ps, 1000) * BITS_PER_BYTE;
+}
+
+static u8 mlxsw_sp_policer_burst_bytes_hw_units(u64 burst_bytes)
+{
+	/* Provided burst size is in bytes. The ASIC burst size value is
+	 * (2 ^ bs) * 512 bits. Convert the provided size to 512-bit units.
+	 */
+	u64 bs512 = div_u64(burst_bytes, 64);
+
+	if (!bs512)
+		return 0;
+
+	return fls64(bs512) - 1;
+}
+
+static u64 mlxsw_sp_policer_single_rate_occ_get(void *priv)
+{
+	struct mlxsw_sp_policer_family *family = priv;
+
+	return atomic_read(&family->policers_count);
+}
+
+static int
+mlxsw_sp_policer_single_rate_family_init(struct mlxsw_sp_policer_family *family)
+{
+	struct mlxsw_core *core = family->mlxsw_sp->core;
+	struct devlink *devlink;
+
+	/* CPU policers are allocated from the first N policers in the global
+	 * range, so skip them.
+	 */
+	if (!MLXSW_CORE_RES_VALID(core, MAX_GLOBAL_POLICERS) ||
+	    !MLXSW_CORE_RES_VALID(core, MAX_CPU_POLICERS))
+		return -EIO;
+
+	family->start_index = MLXSW_CORE_RES_GET(core, MAX_CPU_POLICERS);
+	family->end_index = MLXSW_CORE_RES_GET(core, MAX_GLOBAL_POLICERS);
+
+	atomic_set(&family->policers_count, 0);
+	devlink = priv_to_devlink(core);
+	devl_resource_occ_get_register(devlink,
+				       MLXSW_SP_RESOURCE_SINGLE_RATE_POLICERS,
+				       mlxsw_sp_policer_single_rate_occ_get,
+				       family);
+
+	return 0;
+}
+
+static void
+mlxsw_sp_policer_single_rate_family_fini(struct mlxsw_sp_policer_family *family)
+{
+	struct devlink *devlink = priv_to_devlink(family->mlxsw_sp->core);
+
+	devl_resource_occ_get_unregister(devlink,
+					 MLXSW_SP_RESOURCE_SINGLE_RATE_POLICERS);
+	WARN_ON(atomic_read(&family->policers_count) != 0);
+}
+
+static int
+mlxsw_sp_policer_single_rate_index_alloc(struct mlxsw_sp_policer_family *family,
+					 struct mlxsw_sp_policer *policer)
+{
+	int id;
+
+	mutex_lock(&family->lock);
+	id = idr_alloc(&family->policer_idr, policer, family->start_index,
+		       family->end_index, GFP_KERNEL);
+	mutex_unlock(&family->lock);
+
+	if (id < 0)
+		return id;
+
+	atomic_inc(&family->policers_count);
+	policer->index = id;
+
+	return 0;
+}
+
+static struct mlxsw_sp_policer *
+mlxsw_sp_policer_single_rate_index_free(struct mlxsw_sp_policer_family *family,
+					u16 policer_index)
+{
+	struct mlxsw_sp_policer *policer;
+
+	atomic_dec(&family->policers_count);
+
+	mutex_lock(&family->lock);
+	policer = idr_remove(&family->policer_idr, policer_index);
+	mutex_unlock(&family->lock);
+
+	WARN_ON(!policer);
+
+	return policer;
+}
+
+static int
+mlxsw_sp_policer_single_rate_init(struct mlxsw_sp_policer_family *family,
+				  const struct mlxsw_sp_policer *policer)
+{
+	u64 rate_kbps = mlxsw_sp_policer_rate_bytes_ps_kbps(policer->params.rate);
+	u8 bs = mlxsw_sp_policer_burst_bytes_hw_units(policer->params.burst);
+	struct mlxsw_sp *mlxsw_sp = family->mlxsw_sp;
+	char qpcr_pl[MLXSW_REG_QPCR_LEN];
+
+	mlxsw_reg_qpcr_pack(qpcr_pl, policer->index, MLXSW_REG_QPCR_IR_UNITS_K,
+			    true, rate_kbps, bs);
+	mlxsw_reg_qpcr_clear_counter_set(qpcr_pl, true);
+
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(qpcr), qpcr_pl);
+}
+
+static int
+mlxsw_sp_policer_single_rate_params_check(const struct mlxsw_sp_policer_family *family,
+					  const struct mlxsw_sp_policer_params *params,
+					  struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp_policer_core *policer_core = family->mlxsw_sp->policer_core;
+	u64 rate_bps = params->rate * BITS_PER_BYTE;
+	u8 bs;
+
+	if (!params->bytes) {
+		NL_SET_ERR_MSG_MOD(extack, "Only bandwidth policing is currently supported by single rate policers");
+		return -EINVAL;
+	}
+
+	if (!is_power_of_2(params->burst)) {
+		NL_SET_ERR_MSG_MOD(extack, "Policer burst size is not power of two");
+		return -EINVAL;
+	}
+
+	bs = mlxsw_sp_policer_burst_bytes_hw_units(params->burst);
+
+	if (bs < policer_core->lowest_bs_bits) {
+		NL_SET_ERR_MSG_MOD(extack, "Policer burst size lower than limit");
+		return -EINVAL;
+	}
+
+	if (bs > policer_core->highest_bs_bits) {
+		NL_SET_ERR_MSG_MOD(extack, "Policer burst size higher than limit");
+		return -EINVAL;
+	}
+
+	if (rate_bps < MLXSW_REG_QPCR_LOWEST_CIR_BITS) {
+		NL_SET_ERR_MSG_MOD(extack, "Policer rate lower than limit");
+		return -EINVAL;
+	}
+
+	if (rate_bps > MLXSW_REG_QPCR_HIGHEST_CIR_BITS) {
+		NL_SET_ERR_MSG_MOD(extack, "Policer rate higher than limit");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static const struct mlxsw_sp_policer_family_ops mlxsw_sp_policer_single_rate_ops = {
+	.init			= mlxsw_sp_policer_single_rate_family_init,
+	.fini			= mlxsw_sp_policer_single_rate_family_fini,
+	.policer_index_alloc	= mlxsw_sp_policer_single_rate_index_alloc,
+	.policer_index_free	= mlxsw_sp_policer_single_rate_index_free,
+	.policer_init		= mlxsw_sp_policer_single_rate_init,
+	.policer_params_check	= mlxsw_sp_policer_single_rate_params_check,
+};
+
+static const struct mlxsw_sp_policer_family mlxsw_sp_policer_single_rate_family = {
+	.type		= MLXSW_SP_POLICER_TYPE_SINGLE_RATE,
+	.qpcr_type	= MLXSW_REG_QPCR_G_GLOBAL,
+	.ops		= &mlxsw_sp_policer_single_rate_ops,
+};
+
+static const struct mlxsw_sp_policer_family *mlxsw_sp_policer_family_arr[] = {
+	[MLXSW_SP_POLICER_TYPE_SINGLE_RATE]	= &mlxsw_sp_policer_single_rate_family,
+};
+
+int mlxsw_sp_policer_add(struct mlxsw_sp *mlxsw_sp,
+			 enum mlxsw_sp_policer_type type,
+			 const struct mlxsw_sp_policer_params *params,
+			 struct netlink_ext_ack *extack, u16 *p_policer_index)
+{
+	struct mlxsw_sp_policer_family *family;
+	struct mlxsw_sp_policer *policer;
+	int err;
+
+	family = mlxsw_sp->policer_core->family_arr[type];
+
+	err = family->ops->policer_params_check(family, params, extack);
+	if (err)
+		return err;
+
+	policer = kmalloc(sizeof(*policer), GFP_KERNEL);
+	if (!policer)
+		return -ENOMEM;
+	policer->params = *params;
+
+	err = family->ops->policer_index_alloc(family, policer);
+	if (err) {
+		NL_SET_ERR_MSG_MOD(extack, "Failed to allocate policer index");
+		goto err_policer_index_alloc;
+	}
+
+	err = family->ops->policer_init(family, policer);
+	if (err) {
+		NL_SET_ERR_MSG_MOD(extack, "Failed to initialize policer");
+		goto err_policer_init;
+	}
+
+	*p_policer_index = policer->index;
+
+	return 0;
+
+err_policer_init:
+	family->ops->policer_index_free(family, policer->index);
+err_policer_index_alloc:
+	kfree(policer);
+	return err;
+}
+
+void mlxsw_sp_policer_del(struct mlxsw_sp *mlxsw_sp,
+			  enum mlxsw_sp_policer_type type, u16 policer_index)
+{
+	struct mlxsw_sp_policer_family *family;
+	struct mlxsw_sp_policer *policer;
+
+	family = mlxsw_sp->policer_core->family_arr[type];
+	policer = family->ops->policer_index_free(family, policer_index);
+	kfree(policer);
+}
+
+int mlxsw_sp_policer_drops_counter_get(struct mlxsw_sp *mlxsw_sp,
+				       enum mlxsw_sp_policer_type type,
+				       u16 policer_index, u64 *p_drops)
+{
+	struct mlxsw_sp_policer_family *family;
+	char qpcr_pl[MLXSW_REG_QPCR_LEN];
+	int err;
+
+	family = mlxsw_sp->policer_core->family_arr[type];
+
+	MLXSW_REG_ZERO(qpcr, qpcr_pl);
+	mlxsw_reg_qpcr_pid_set(qpcr_pl, policer_index);
+	mlxsw_reg_qpcr_g_set(qpcr_pl, family->qpcr_type);
+	err = mlxsw_reg_query(mlxsw_sp->core, MLXSW_REG(qpcr), qpcr_pl);
+	if (err)
+		return err;
+
+	*p_drops = mlxsw_reg_qpcr_violate_count_get(qpcr_pl);
+
+	return 0;
+}
+
+static int
+mlxsw_sp_policer_family_register(struct mlxsw_sp *mlxsw_sp,
+				 const struct mlxsw_sp_policer_family *tmpl)
+{
+	struct mlxsw_sp_policer_family *family;
+	int err;
+
+	family = kmemdup(tmpl, sizeof(*family), GFP_KERNEL);
+	if (!family)
+		return -ENOMEM;
+
+	family->mlxsw_sp = mlxsw_sp;
+	idr_init(&family->policer_idr);
+	mutex_init(&family->lock);
+
+	err = family->ops->init(family);
+	if (err)
+		goto err_family_init;
+
+	if (WARN_ON(family->start_index >= family->end_index)) {
+		err = -EINVAL;
+		goto err_index_check;
+	}
+
+	mlxsw_sp->policer_core->family_arr[tmpl->type] = family;
+
+	return 0;
+
+err_index_check:
+	family->ops->fini(family);
+err_family_init:
+	mutex_destroy(&family->lock);
+	idr_destroy(&family->policer_idr);
+	kfree(family);
+	return err;
+}
+
+static void
+mlxsw_sp_policer_family_unregister(struct mlxsw_sp *mlxsw_sp,
+				   struct mlxsw_sp_policer_family *family)
+{
+	family->ops->fini(family);
+	mutex_destroy(&family->lock);
+	WARN_ON(!idr_is_empty(&family->policer_idr));
+	idr_destroy(&family->policer_idr);
+	kfree(family);
+}
+
+int mlxsw_sp_policers_init(struct mlxsw_sp *mlxsw_sp)
+{
+	struct mlxsw_sp_policer_core *policer_core;
+	int i, err;
+
+	policer_core = kzalloc(sizeof(*policer_core), GFP_KERNEL);
+	if (!policer_core)
+		return -ENOMEM;
+	mlxsw_sp->policer_core = policer_core;
+	policer_core->ops = mlxsw_sp->policer_core_ops;
+
+	err = policer_core->ops->init(policer_core);
+	if (err)
+		goto err_init;
+
+	for (i = 0; i < MLXSW_SP_POLICER_TYPE_MAX + 1; i++) {
+		err = mlxsw_sp_policer_family_register(mlxsw_sp, mlxsw_sp_policer_family_arr[i]);
+		if (err)
+			goto err_family_register;
+	}
+
+	return 0;
+
+err_family_register:
+	for (i--; i >= 0; i--) {
+		struct mlxsw_sp_policer_family *family;
+
+		family = mlxsw_sp->policer_core->family_arr[i];
+		mlxsw_sp_policer_family_unregister(mlxsw_sp, family);
+	}
+err_init:
+	kfree(mlxsw_sp->policer_core);
+	return err;
+}
+
+void mlxsw_sp_policers_fini(struct mlxsw_sp *mlxsw_sp)
+{
+	int i;
+
+	for (i = MLXSW_SP_POLICER_TYPE_MAX; i >= 0; i--) {
+		struct mlxsw_sp_policer_family *family;
+
+		family = mlxsw_sp->policer_core->family_arr[i];
+		mlxsw_sp_policer_family_unregister(mlxsw_sp, family);
+	}
+
+	kfree(mlxsw_sp->policer_core);
+}
+
+int mlxsw_sp_policer_resources_register(struct mlxsw_core *mlxsw_core)
+{
+	u64 global_policers, cpu_policers, single_rate_policers;
+	struct devlink *devlink = priv_to_devlink(mlxsw_core);
+	struct devlink_resource_size_params size_params;
+	int err;
+
+	if (!MLXSW_CORE_RES_VALID(mlxsw_core, MAX_GLOBAL_POLICERS) ||
+	    !MLXSW_CORE_RES_VALID(mlxsw_core, MAX_CPU_POLICERS))
+		return -EIO;
+
+	global_policers = MLXSW_CORE_RES_GET(mlxsw_core, MAX_GLOBAL_POLICERS);
+	cpu_policers = MLXSW_CORE_RES_GET(mlxsw_core, MAX_CPU_POLICERS);
+	single_rate_policers = global_policers - cpu_policers;
+
+	devlink_resource_size_params_init(&size_params, global_policers,
+					  global_policers, 1,
+					  DEVLINK_RESOURCE_UNIT_ENTRY);
+	err = devl_resource_register(devlink, "global_policers",
+				     global_policers,
+				     MLXSW_SP_RESOURCE_GLOBAL_POLICERS,
+				     DEVLINK_RESOURCE_ID_PARENT_TOP,
+				     &size_params);
+	if (err)
+		return err;
+
+	devlink_resource_size_params_init(&size_params, single_rate_policers,
+					  single_rate_policers, 1,
+					  DEVLINK_RESOURCE_UNIT_ENTRY);
+	err = devl_resource_register(devlink, "single_rate_policers",
+				     single_rate_policers,
+				     MLXSW_SP_RESOURCE_SINGLE_RATE_POLICERS,
+				     MLXSW_SP_RESOURCE_GLOBAL_POLICERS,
+				     &size_params);
+	if (err)
+		return err;
+
+	return 0;
+}
+
+static int
+mlxsw_sp1_policer_core_init(struct mlxsw_sp_policer_core *policer_core)
+{
+	policer_core->lowest_bs_bits = MLXSW_REG_QPCR_LOWEST_CBS_BITS_SP1;
+	policer_core->highest_bs_bits = MLXSW_REG_QPCR_HIGHEST_CBS_BITS_SP1;
+
+	return 0;
+}
+
+const struct mlxsw_sp_policer_core_ops mlxsw_sp1_policer_core_ops = {
+	.init = mlxsw_sp1_policer_core_init,
+};
+
+static int
+mlxsw_sp2_policer_core_init(struct mlxsw_sp_policer_core *policer_core)
+{
+	policer_core->lowest_bs_bits = MLXSW_REG_QPCR_LOWEST_CBS_BITS_SP2;
+	policer_core->highest_bs_bits = MLXSW_REG_QPCR_HIGHEST_CBS_BITS_SP2;
+
+	return 0;
+}
+
+const struct mlxsw_sp_policer_core_ops mlxsw_sp2_policer_core_ops = {
+	.init = mlxsw_sp2_policer_core_init,
+};
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_ptp.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_ptp.c
new file mode 100644
index 000000000..7b01b9c20
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_ptp.c
@@ -0,0 +1,1726 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2019 Mellanox Technologies. All rights reserved */
+
+#include <linux/ptp_clock_kernel.h>
+#include <linux/clocksource.h>
+#include <linux/timecounter.h>
+#include <linux/spinlock.h>
+#include <linux/device.h>
+#include <linux/rhashtable.h>
+#include <linux/ptp_classify.h>
+#include <linux/if_ether.h>
+#include <linux/if_vlan.h>
+#include <linux/net_tstamp.h>
+#include <linux/refcount.h>
+
+#include "spectrum.h"
+#include "spectrum_ptp.h"
+#include "core.h"
+
+#define MLXSW_SP1_PTP_CLOCK_CYCLES_SHIFT	29
+#define MLXSW_SP1_PTP_CLOCK_FREQ_KHZ		156257 /* 6.4nSec */
+#define MLXSW_SP1_PTP_CLOCK_MASK		64
+
+#define MLXSW_SP1_PTP_HT_GC_INTERVAL		500 /* ms */
+
+/* How long, approximately, should the unmatched entries stay in the hash table
+ * before they are collected. Should be evenly divisible by the GC interval.
+ */
+#define MLXSW_SP1_PTP_HT_GC_TIMEOUT		1000 /* ms */
+
+struct mlxsw_sp_ptp_state {
+	struct mlxsw_sp *mlxsw_sp;
+};
+
+struct mlxsw_sp1_ptp_state {
+	struct mlxsw_sp_ptp_state common;
+	struct rhltable unmatched_ht;
+	spinlock_t unmatched_lock; /* protects the HT */
+	struct delayed_work ht_gc_dw;
+	u32 gc_cycle;
+};
+
+struct mlxsw_sp2_ptp_state {
+	struct mlxsw_sp_ptp_state common;
+	refcount_t ptp_port_enabled_ref; /* Number of ports with time stamping
+					  * enabled.
+					  */
+	struct hwtstamp_config config;
+	struct mutex lock; /* Protects 'config' and HW configuration. */
+};
+
+struct mlxsw_sp1_ptp_key {
+	u16 local_port;
+	u8 message_type;
+	u16 sequence_id;
+	u8 domain_number;
+	bool ingress;
+};
+
+struct mlxsw_sp1_ptp_unmatched {
+	struct mlxsw_sp1_ptp_key key;
+	struct rhlist_head ht_node;
+	struct rcu_head rcu;
+	struct sk_buff *skb;
+	u64 timestamp;
+	u32 gc_cycle;
+};
+
+static const struct rhashtable_params mlxsw_sp1_ptp_unmatched_ht_params = {
+	.key_len = sizeof_field(struct mlxsw_sp1_ptp_unmatched, key),
+	.key_offset = offsetof(struct mlxsw_sp1_ptp_unmatched, key),
+	.head_offset = offsetof(struct mlxsw_sp1_ptp_unmatched, ht_node),
+};
+
+struct mlxsw_sp_ptp_clock {
+	struct mlxsw_core *core;
+	struct ptp_clock *ptp;
+	struct ptp_clock_info ptp_info;
+};
+
+struct mlxsw_sp1_ptp_clock {
+	struct mlxsw_sp_ptp_clock common;
+	spinlock_t lock; /* protect this structure */
+	struct cyclecounter cycles;
+	struct timecounter tc;
+	u32 nominal_c_mult;
+	unsigned long overflow_period;
+	struct delayed_work overflow_work;
+};
+
+static struct mlxsw_sp1_ptp_state *
+mlxsw_sp1_ptp_state(struct mlxsw_sp *mlxsw_sp)
+{
+	return container_of(mlxsw_sp->ptp_state, struct mlxsw_sp1_ptp_state,
+			    common);
+}
+
+static struct mlxsw_sp2_ptp_state *
+mlxsw_sp2_ptp_state(struct mlxsw_sp *mlxsw_sp)
+{
+	return container_of(mlxsw_sp->ptp_state, struct mlxsw_sp2_ptp_state,
+			    common);
+}
+
+static struct mlxsw_sp1_ptp_clock *
+mlxsw_sp1_ptp_clock(struct ptp_clock_info *ptp)
+{
+	return container_of(ptp, struct mlxsw_sp1_ptp_clock, common.ptp_info);
+}
+
+static u64 __mlxsw_sp1_ptp_read_frc(struct mlxsw_sp1_ptp_clock *clock,
+				    struct ptp_system_timestamp *sts)
+{
+	struct mlxsw_core *mlxsw_core = clock->common.core;
+	u32 frc_h1, frc_h2, frc_l;
+
+	frc_h1 = mlxsw_core_read_frc_h(mlxsw_core);
+	ptp_read_system_prets(sts);
+	frc_l = mlxsw_core_read_frc_l(mlxsw_core);
+	ptp_read_system_postts(sts);
+	frc_h2 = mlxsw_core_read_frc_h(mlxsw_core);
+
+	if (frc_h1 != frc_h2) {
+		/* wrap around */
+		ptp_read_system_prets(sts);
+		frc_l = mlxsw_core_read_frc_l(mlxsw_core);
+		ptp_read_system_postts(sts);
+	}
+
+	return (u64) frc_l | (u64) frc_h2 << 32;
+}
+
+static u64 mlxsw_sp1_ptp_read_frc(const struct cyclecounter *cc)
+{
+	struct mlxsw_sp1_ptp_clock *clock =
+		container_of(cc, struct mlxsw_sp1_ptp_clock, cycles);
+
+	return __mlxsw_sp1_ptp_read_frc(clock, NULL) & cc->mask;
+}
+
+static int
+mlxsw_sp_ptp_phc_adjfreq(struct mlxsw_sp_ptp_clock *clock, int freq_adj)
+{
+	struct mlxsw_core *mlxsw_core = clock->core;
+	char mtutc_pl[MLXSW_REG_MTUTC_LEN];
+
+	mlxsw_reg_mtutc_pack(mtutc_pl, MLXSW_REG_MTUTC_OPERATION_ADJUST_FREQ,
+			     freq_adj, 0, 0, 0);
+	return mlxsw_reg_write(mlxsw_core, MLXSW_REG(mtutc), mtutc_pl);
+}
+
+static u64 mlxsw_sp1_ptp_ns2cycles(const struct timecounter *tc, u64 nsec)
+{
+	u64 cycles = (u64) nsec;
+
+	cycles <<= tc->cc->shift;
+	cycles = div_u64(cycles, tc->cc->mult);
+
+	return cycles;
+}
+
+static int
+mlxsw_sp1_ptp_phc_settime(struct mlxsw_sp1_ptp_clock *clock, u64 nsec)
+{
+	struct mlxsw_core *mlxsw_core = clock->common.core;
+	u64 next_sec, next_sec_in_nsec, cycles;
+	char mtutc_pl[MLXSW_REG_MTUTC_LEN];
+	char mtpps_pl[MLXSW_REG_MTPPS_LEN];
+	int err;
+
+	next_sec = div_u64(nsec, NSEC_PER_SEC) + 1;
+	next_sec_in_nsec = next_sec * NSEC_PER_SEC;
+
+	spin_lock_bh(&clock->lock);
+	cycles = mlxsw_sp1_ptp_ns2cycles(&clock->tc, next_sec_in_nsec);
+	spin_unlock_bh(&clock->lock);
+
+	mlxsw_reg_mtpps_vpin_pack(mtpps_pl, cycles);
+	err = mlxsw_reg_write(mlxsw_core, MLXSW_REG(mtpps), mtpps_pl);
+	if (err)
+		return err;
+
+	mlxsw_reg_mtutc_pack(mtutc_pl,
+			     MLXSW_REG_MTUTC_OPERATION_SET_TIME_AT_NEXT_SEC,
+			     0, next_sec, 0, 0);
+	return mlxsw_reg_write(mlxsw_core, MLXSW_REG(mtutc), mtutc_pl);
+}
+
+static int mlxsw_sp1_ptp_adjfine(struct ptp_clock_info *ptp, long scaled_ppm)
+{
+	struct mlxsw_sp1_ptp_clock *clock = mlxsw_sp1_ptp_clock(ptp);
+	int neg_adj = 0;
+	u32 diff;
+	u64 adj;
+	s32 ppb;
+
+	ppb = scaled_ppm_to_ppb(scaled_ppm);
+
+	if (ppb < 0) {
+		neg_adj = 1;
+		ppb = -ppb;
+	}
+
+	adj = clock->nominal_c_mult;
+	adj *= ppb;
+	diff = div_u64(adj, NSEC_PER_SEC);
+
+	spin_lock_bh(&clock->lock);
+	timecounter_read(&clock->tc);
+	clock->cycles.mult = neg_adj ? clock->nominal_c_mult - diff :
+				       clock->nominal_c_mult + diff;
+	spin_unlock_bh(&clock->lock);
+
+	return mlxsw_sp_ptp_phc_adjfreq(&clock->common, neg_adj ? -ppb : ppb);
+}
+
+static int mlxsw_sp1_ptp_adjtime(struct ptp_clock_info *ptp, s64 delta)
+{
+	struct mlxsw_sp1_ptp_clock *clock = mlxsw_sp1_ptp_clock(ptp);
+	u64 nsec;
+
+	spin_lock_bh(&clock->lock);
+	timecounter_adjtime(&clock->tc, delta);
+	nsec = timecounter_read(&clock->tc);
+	spin_unlock_bh(&clock->lock);
+
+	return mlxsw_sp1_ptp_phc_settime(clock, nsec);
+}
+
+static int mlxsw_sp1_ptp_gettimex(struct ptp_clock_info *ptp,
+				  struct timespec64 *ts,
+				  struct ptp_system_timestamp *sts)
+{
+	struct mlxsw_sp1_ptp_clock *clock = mlxsw_sp1_ptp_clock(ptp);
+	u64 cycles, nsec;
+
+	spin_lock_bh(&clock->lock);
+	cycles = __mlxsw_sp1_ptp_read_frc(clock, sts);
+	nsec = timecounter_cyc2time(&clock->tc, cycles);
+	spin_unlock_bh(&clock->lock);
+
+	*ts = ns_to_timespec64(nsec);
+
+	return 0;
+}
+
+static int mlxsw_sp1_ptp_settime(struct ptp_clock_info *ptp,
+				 const struct timespec64 *ts)
+{
+	struct mlxsw_sp1_ptp_clock *clock = mlxsw_sp1_ptp_clock(ptp);
+	u64 nsec = timespec64_to_ns(ts);
+
+	spin_lock_bh(&clock->lock);
+	timecounter_init(&clock->tc, &clock->cycles, nsec);
+	nsec = timecounter_read(&clock->tc);
+	spin_unlock_bh(&clock->lock);
+
+	return mlxsw_sp1_ptp_phc_settime(clock, nsec);
+}
+
+static const struct ptp_clock_info mlxsw_sp1_ptp_clock_info = {
+	.owner		= THIS_MODULE,
+	.name		= "mlxsw_sp_clock",
+	.max_adj	= 100000000,
+	.adjfine	= mlxsw_sp1_ptp_adjfine,
+	.adjtime	= mlxsw_sp1_ptp_adjtime,
+	.gettimex64	= mlxsw_sp1_ptp_gettimex,
+	.settime64	= mlxsw_sp1_ptp_settime,
+};
+
+static void mlxsw_sp1_ptp_clock_overflow(struct work_struct *work)
+{
+	struct delayed_work *dwork = to_delayed_work(work);
+	struct mlxsw_sp1_ptp_clock *clock;
+
+	clock = container_of(dwork, struct mlxsw_sp1_ptp_clock, overflow_work);
+
+	spin_lock_bh(&clock->lock);
+	timecounter_read(&clock->tc);
+	spin_unlock_bh(&clock->lock);
+	mlxsw_core_schedule_dw(&clock->overflow_work, clock->overflow_period);
+}
+
+struct mlxsw_sp_ptp_clock *
+mlxsw_sp1_ptp_clock_init(struct mlxsw_sp *mlxsw_sp, struct device *dev)
+{
+	u64 overflow_cycles, nsec, frac = 0;
+	struct mlxsw_sp1_ptp_clock *clock;
+	int err;
+
+	clock = kzalloc(sizeof(*clock), GFP_KERNEL);
+	if (!clock)
+		return ERR_PTR(-ENOMEM);
+
+	spin_lock_init(&clock->lock);
+	clock->cycles.read = mlxsw_sp1_ptp_read_frc;
+	clock->cycles.shift = MLXSW_SP1_PTP_CLOCK_CYCLES_SHIFT;
+	clock->cycles.mult = clocksource_khz2mult(MLXSW_SP1_PTP_CLOCK_FREQ_KHZ,
+						  clock->cycles.shift);
+	clock->nominal_c_mult = clock->cycles.mult;
+	clock->cycles.mask = CLOCKSOURCE_MASK(MLXSW_SP1_PTP_CLOCK_MASK);
+	clock->common.core = mlxsw_sp->core;
+
+	timecounter_init(&clock->tc, &clock->cycles, 0);
+
+	/* Calculate period in seconds to call the overflow watchdog - to make
+	 * sure counter is checked at least twice every wrap around.
+	 * The period is calculated as the minimum between max HW cycles count
+	 * (The clock source mask) and max amount of cycles that can be
+	 * multiplied by clock multiplier where the result doesn't exceed
+	 * 64bits.
+	 */
+	overflow_cycles = div64_u64(~0ULL >> 1, clock->cycles.mult);
+	overflow_cycles = min(overflow_cycles, div_u64(clock->cycles.mask, 3));
+
+	nsec = cyclecounter_cyc2ns(&clock->cycles, overflow_cycles, 0, &frac);
+	clock->overflow_period = nsecs_to_jiffies(nsec);
+
+	INIT_DELAYED_WORK(&clock->overflow_work, mlxsw_sp1_ptp_clock_overflow);
+	mlxsw_core_schedule_dw(&clock->overflow_work, 0);
+
+	clock->common.ptp_info = mlxsw_sp1_ptp_clock_info;
+	clock->common.ptp = ptp_clock_register(&clock->common.ptp_info, dev);
+	if (IS_ERR(clock->common.ptp)) {
+		err = PTR_ERR(clock->common.ptp);
+		dev_err(dev, "ptp_clock_register failed %d\n", err);
+		goto err_ptp_clock_register;
+	}
+
+	return &clock->common;
+
+err_ptp_clock_register:
+	cancel_delayed_work_sync(&clock->overflow_work);
+	kfree(clock);
+	return ERR_PTR(err);
+}
+
+void mlxsw_sp1_ptp_clock_fini(struct mlxsw_sp_ptp_clock *clock_common)
+{
+	struct mlxsw_sp1_ptp_clock *clock =
+		container_of(clock_common, struct mlxsw_sp1_ptp_clock, common);
+
+	ptp_clock_unregister(clock_common->ptp);
+	cancel_delayed_work_sync(&clock->overflow_work);
+	kfree(clock);
+}
+
+static u64 mlxsw_sp2_ptp_read_utc(struct mlxsw_sp_ptp_clock *clock,
+				  struct ptp_system_timestamp *sts)
+{
+	struct mlxsw_core *mlxsw_core = clock->core;
+	u32 utc_sec1, utc_sec2, utc_nsec;
+
+	utc_sec1 = mlxsw_core_read_utc_sec(mlxsw_core);
+	ptp_read_system_prets(sts);
+	utc_nsec = mlxsw_core_read_utc_nsec(mlxsw_core);
+	ptp_read_system_postts(sts);
+	utc_sec2 = mlxsw_core_read_utc_sec(mlxsw_core);
+
+	if (utc_sec1 != utc_sec2) {
+		/* Wrap around. */
+		ptp_read_system_prets(sts);
+		utc_nsec = mlxsw_core_read_utc_nsec(mlxsw_core);
+		ptp_read_system_postts(sts);
+	}
+
+	return (u64)utc_sec2 * NSEC_PER_SEC + utc_nsec;
+}
+
+static int
+mlxsw_sp2_ptp_phc_settime(struct mlxsw_sp_ptp_clock *clock, u64 nsec)
+{
+	struct mlxsw_core *mlxsw_core = clock->core;
+	char mtutc_pl[MLXSW_REG_MTUTC_LEN];
+	u32 sec, nsec_rem;
+
+	sec = div_u64_rem(nsec, NSEC_PER_SEC, &nsec_rem);
+	mlxsw_reg_mtutc_pack(mtutc_pl,
+			     MLXSW_REG_MTUTC_OPERATION_SET_TIME_IMMEDIATE,
+			     0, sec, nsec_rem, 0);
+	return mlxsw_reg_write(mlxsw_core, MLXSW_REG(mtutc), mtutc_pl);
+}
+
+static int mlxsw_sp2_ptp_adjfine(struct ptp_clock_info *ptp, long scaled_ppm)
+{
+	struct mlxsw_sp_ptp_clock *clock =
+		container_of(ptp, struct mlxsw_sp_ptp_clock, ptp_info);
+	s32 ppb = scaled_ppm_to_ppb(scaled_ppm);
+
+	/* In Spectrum-2 and newer ASICs, the frequency adjustment in MTUTC is
+	 * reversed, positive values mean to decrease the frequency. Adjust the
+	 * sign of PPB to this behavior.
+	 */
+	return mlxsw_sp_ptp_phc_adjfreq(clock, -ppb);
+}
+
+static int mlxsw_sp2_ptp_adjtime(struct ptp_clock_info *ptp, s64 delta)
+{
+	struct mlxsw_sp_ptp_clock *clock =
+		container_of(ptp, struct mlxsw_sp_ptp_clock, ptp_info);
+	struct mlxsw_core *mlxsw_core = clock->core;
+	char mtutc_pl[MLXSW_REG_MTUTC_LEN];
+
+	/* HW time adjustment range is s16. If out of range, set time instead. */
+	if (delta < S16_MIN || delta > S16_MAX) {
+		u64 nsec;
+
+		nsec = mlxsw_sp2_ptp_read_utc(clock, NULL);
+		nsec += delta;
+
+		return mlxsw_sp2_ptp_phc_settime(clock, nsec);
+	}
+
+	mlxsw_reg_mtutc_pack(mtutc_pl,
+			     MLXSW_REG_MTUTC_OPERATION_ADJUST_TIME,
+			     0, 0, 0, delta);
+	return mlxsw_reg_write(mlxsw_core, MLXSW_REG(mtutc), mtutc_pl);
+}
+
+static int mlxsw_sp2_ptp_gettimex(struct ptp_clock_info *ptp,
+				  struct timespec64 *ts,
+				  struct ptp_system_timestamp *sts)
+{
+	struct mlxsw_sp_ptp_clock *clock =
+		container_of(ptp, struct mlxsw_sp_ptp_clock, ptp_info);
+	u64 nsec;
+
+	nsec = mlxsw_sp2_ptp_read_utc(clock, sts);
+	*ts = ns_to_timespec64(nsec);
+
+	return 0;
+}
+
+static int mlxsw_sp2_ptp_settime(struct ptp_clock_info *ptp,
+				 const struct timespec64 *ts)
+{
+	struct mlxsw_sp_ptp_clock *clock =
+		container_of(ptp, struct mlxsw_sp_ptp_clock, ptp_info);
+	u64 nsec = timespec64_to_ns(ts);
+
+	return mlxsw_sp2_ptp_phc_settime(clock, nsec);
+}
+
+static const struct ptp_clock_info mlxsw_sp2_ptp_clock_info = {
+	.owner		= THIS_MODULE,
+	.name		= "mlxsw_sp_clock",
+	.max_adj	= MLXSW_REG_MTUTC_MAX_FREQ_ADJ,
+	.adjfine	= mlxsw_sp2_ptp_adjfine,
+	.adjtime	= mlxsw_sp2_ptp_adjtime,
+	.gettimex64	= mlxsw_sp2_ptp_gettimex,
+	.settime64	= mlxsw_sp2_ptp_settime,
+};
+
+struct mlxsw_sp_ptp_clock *
+mlxsw_sp2_ptp_clock_init(struct mlxsw_sp *mlxsw_sp, struct device *dev)
+{
+	struct mlxsw_sp_ptp_clock *clock;
+	int err;
+
+	clock = kzalloc(sizeof(*clock), GFP_KERNEL);
+	if (!clock)
+		return ERR_PTR(-ENOMEM);
+
+	clock->core = mlxsw_sp->core;
+
+	clock->ptp_info = mlxsw_sp2_ptp_clock_info;
+
+	err = mlxsw_sp2_ptp_phc_settime(clock, 0);
+	if (err) {
+		dev_err(dev, "setting UTC time failed %d\n", err);
+		goto err_ptp_phc_settime;
+	}
+
+	clock->ptp = ptp_clock_register(&clock->ptp_info, dev);
+	if (IS_ERR(clock->ptp)) {
+		err = PTR_ERR(clock->ptp);
+		dev_err(dev, "ptp_clock_register failed %d\n", err);
+		goto err_ptp_clock_register;
+	}
+
+	return clock;
+
+err_ptp_clock_register:
+err_ptp_phc_settime:
+	kfree(clock);
+	return ERR_PTR(err);
+}
+
+void mlxsw_sp2_ptp_clock_fini(struct mlxsw_sp_ptp_clock *clock)
+{
+	ptp_clock_unregister(clock->ptp);
+	kfree(clock);
+}
+
+static int mlxsw_sp_ptp_parse(struct sk_buff *skb,
+			      u8 *p_domain_number,
+			      u8 *p_message_type,
+			      u16 *p_sequence_id)
+{
+	unsigned int ptp_class;
+	struct ptp_header *hdr;
+
+	ptp_class = ptp_classify_raw(skb);
+
+	switch (ptp_class & PTP_CLASS_VMASK) {
+	case PTP_CLASS_V1:
+	case PTP_CLASS_V2:
+		break;
+	default:
+		return -ERANGE;
+	}
+
+	hdr = ptp_parse_header(skb, ptp_class);
+	if (!hdr)
+		return -EINVAL;
+
+	*p_message_type	 = ptp_get_msgtype(hdr, ptp_class);
+	*p_domain_number = hdr->domain_number;
+	*p_sequence_id	 = be16_to_cpu(hdr->sequence_id);
+
+	return 0;
+}
+
+/* Returns NULL on successful insertion, a pointer on conflict, or an ERR_PTR on
+ * error.
+ */
+static int
+mlxsw_sp1_ptp_unmatched_save(struct mlxsw_sp *mlxsw_sp,
+			     struct mlxsw_sp1_ptp_key key,
+			     struct sk_buff *skb,
+			     u64 timestamp)
+{
+	int cycles = MLXSW_SP1_PTP_HT_GC_TIMEOUT / MLXSW_SP1_PTP_HT_GC_INTERVAL;
+	struct mlxsw_sp1_ptp_state *ptp_state = mlxsw_sp1_ptp_state(mlxsw_sp);
+	struct mlxsw_sp1_ptp_unmatched *unmatched;
+	int err;
+
+	unmatched = kzalloc(sizeof(*unmatched), GFP_ATOMIC);
+	if (!unmatched)
+		return -ENOMEM;
+
+	unmatched->key = key;
+	unmatched->skb = skb;
+	unmatched->timestamp = timestamp;
+	unmatched->gc_cycle = ptp_state->gc_cycle + cycles;
+
+	err = rhltable_insert(&ptp_state->unmatched_ht, &unmatched->ht_node,
+			      mlxsw_sp1_ptp_unmatched_ht_params);
+	if (err)
+		kfree(unmatched);
+
+	return err;
+}
+
+static struct mlxsw_sp1_ptp_unmatched *
+mlxsw_sp1_ptp_unmatched_lookup(struct mlxsw_sp *mlxsw_sp,
+			       struct mlxsw_sp1_ptp_key key, int *p_length)
+{
+	struct mlxsw_sp1_ptp_state *ptp_state = mlxsw_sp1_ptp_state(mlxsw_sp);
+	struct mlxsw_sp1_ptp_unmatched *unmatched, *last = NULL;
+	struct rhlist_head *tmp, *list;
+	int length = 0;
+
+	list = rhltable_lookup(&ptp_state->unmatched_ht, &key,
+			       mlxsw_sp1_ptp_unmatched_ht_params);
+	rhl_for_each_entry_rcu(unmatched, tmp, list, ht_node) {
+		last = unmatched;
+		length++;
+	}
+
+	*p_length = length;
+	return last;
+}
+
+static int
+mlxsw_sp1_ptp_unmatched_remove(struct mlxsw_sp *mlxsw_sp,
+			       struct mlxsw_sp1_ptp_unmatched *unmatched)
+{
+	struct mlxsw_sp1_ptp_state *ptp_state = mlxsw_sp1_ptp_state(mlxsw_sp);
+
+	return rhltable_remove(&ptp_state->unmatched_ht,
+			       &unmatched->ht_node,
+			       mlxsw_sp1_ptp_unmatched_ht_params);
+}
+
+/* This function is called in the following scenarios:
+ *
+ * 1) When a packet is matched with its timestamp.
+ * 2) In several situation when it is necessary to immediately pass on
+ *    an SKB without a timestamp.
+ * 3) From GC indirectly through mlxsw_sp1_ptp_unmatched_finish().
+ *    This case is similar to 2) above.
+ */
+static void mlxsw_sp1_ptp_packet_finish(struct mlxsw_sp *mlxsw_sp,
+					struct sk_buff *skb, u16 local_port,
+					bool ingress,
+					struct skb_shared_hwtstamps *hwtstamps)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port;
+
+	/* Between capturing the packet and finishing it, there is a window of
+	 * opportunity for the originating port to go away (e.g. due to a
+	 * split). Also make sure the SKB device reference is still valid.
+	 */
+	mlxsw_sp_port = mlxsw_sp->ports[local_port];
+	if (!(mlxsw_sp_port && (!skb->dev || skb->dev == mlxsw_sp_port->dev))) {
+		dev_kfree_skb_any(skb);
+		return;
+	}
+
+	if (ingress) {
+		if (hwtstamps)
+			*skb_hwtstamps(skb) = *hwtstamps;
+		mlxsw_sp_rx_listener_no_mark_func(skb, local_port, mlxsw_sp);
+	} else {
+		/* skb_tstamp_tx() allows hwtstamps to be NULL. */
+		skb_tstamp_tx(skb, hwtstamps);
+		dev_kfree_skb_any(skb);
+	}
+}
+
+static void mlxsw_sp1_packet_timestamp(struct mlxsw_sp *mlxsw_sp,
+				       struct mlxsw_sp1_ptp_key key,
+				       struct sk_buff *skb,
+				       u64 timestamp)
+{
+	struct mlxsw_sp_ptp_clock *clock_common = mlxsw_sp->clock;
+	struct mlxsw_sp1_ptp_clock *clock =
+		container_of(clock_common, struct mlxsw_sp1_ptp_clock, common);
+
+	struct skb_shared_hwtstamps hwtstamps;
+	u64 nsec;
+
+	spin_lock_bh(&clock->lock);
+	nsec = timecounter_cyc2time(&clock->tc, timestamp);
+	spin_unlock_bh(&clock->lock);
+
+	hwtstamps.hwtstamp = ns_to_ktime(nsec);
+	mlxsw_sp1_ptp_packet_finish(mlxsw_sp, skb,
+				    key.local_port, key.ingress, &hwtstamps);
+}
+
+static void
+mlxsw_sp1_ptp_unmatched_finish(struct mlxsw_sp *mlxsw_sp,
+			       struct mlxsw_sp1_ptp_unmatched *unmatched)
+{
+	if (unmatched->skb && unmatched->timestamp)
+		mlxsw_sp1_packet_timestamp(mlxsw_sp, unmatched->key,
+					   unmatched->skb,
+					   unmatched->timestamp);
+	else if (unmatched->skb)
+		mlxsw_sp1_ptp_packet_finish(mlxsw_sp, unmatched->skb,
+					    unmatched->key.local_port,
+					    unmatched->key.ingress, NULL);
+	kfree_rcu(unmatched, rcu);
+}
+
+static void mlxsw_sp1_ptp_unmatched_free_fn(void *ptr, void *arg)
+{
+	struct mlxsw_sp1_ptp_unmatched *unmatched = ptr;
+
+	/* This is invoked at a point where the ports are gone already. Nothing
+	 * to do with whatever is left in the HT but to free it.
+	 */
+	if (unmatched->skb)
+		dev_kfree_skb_any(unmatched->skb);
+	kfree_rcu(unmatched, rcu);
+}
+
+static void mlxsw_sp1_ptp_got_piece(struct mlxsw_sp *mlxsw_sp,
+				    struct mlxsw_sp1_ptp_key key,
+				    struct sk_buff *skb, u64 timestamp)
+{
+	struct mlxsw_sp1_ptp_state *ptp_state = mlxsw_sp1_ptp_state(mlxsw_sp);
+	struct mlxsw_sp1_ptp_unmatched *unmatched;
+	int length;
+	int err;
+
+	rcu_read_lock();
+
+	spin_lock(&ptp_state->unmatched_lock);
+
+	unmatched = mlxsw_sp1_ptp_unmatched_lookup(mlxsw_sp, key, &length);
+	if (skb && unmatched && unmatched->timestamp) {
+		unmatched->skb = skb;
+	} else if (timestamp && unmatched && unmatched->skb) {
+		unmatched->timestamp = timestamp;
+	} else {
+		/* Either there is no entry to match, or one that is there is
+		 * incompatible.
+		 */
+		if (length < 100)
+			err = mlxsw_sp1_ptp_unmatched_save(mlxsw_sp, key,
+							   skb, timestamp);
+		else
+			err = -E2BIG;
+		if (err && skb)
+			mlxsw_sp1_ptp_packet_finish(mlxsw_sp, skb,
+						    key.local_port,
+						    key.ingress, NULL);
+		unmatched = NULL;
+	}
+
+	if (unmatched) {
+		err = mlxsw_sp1_ptp_unmatched_remove(mlxsw_sp, unmatched);
+		WARN_ON_ONCE(err);
+	}
+
+	spin_unlock(&ptp_state->unmatched_lock);
+
+	if (unmatched)
+		mlxsw_sp1_ptp_unmatched_finish(mlxsw_sp, unmatched);
+
+	rcu_read_unlock();
+}
+
+static void mlxsw_sp1_ptp_got_packet(struct mlxsw_sp *mlxsw_sp,
+				     struct sk_buff *skb, u16 local_port,
+				     bool ingress)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port;
+	struct mlxsw_sp1_ptp_key key;
+	u8 types;
+	int err;
+
+	mlxsw_sp_port = mlxsw_sp->ports[local_port];
+	if (!mlxsw_sp_port)
+		goto immediate;
+
+	types = ingress ? mlxsw_sp_port->ptp.ing_types :
+			  mlxsw_sp_port->ptp.egr_types;
+	if (!types)
+		goto immediate;
+
+	memset(&key, 0, sizeof(key));
+	key.local_port = local_port;
+	key.ingress = ingress;
+
+	err = mlxsw_sp_ptp_parse(skb, &key.domain_number, &key.message_type,
+				 &key.sequence_id);
+	if (err)
+		goto immediate;
+
+	/* For packets whose timestamping was not enabled on this port, don't
+	 * bother trying to match the timestamp.
+	 */
+	if (!((1 << key.message_type) & types))
+		goto immediate;
+
+	mlxsw_sp1_ptp_got_piece(mlxsw_sp, key, skb, 0);
+	return;
+
+immediate:
+	mlxsw_sp1_ptp_packet_finish(mlxsw_sp, skb, local_port, ingress, NULL);
+}
+
+void mlxsw_sp1_ptp_got_timestamp(struct mlxsw_sp *mlxsw_sp, bool ingress,
+				 u16 local_port, u8 message_type,
+				 u8 domain_number, u16 sequence_id,
+				 u64 timestamp)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port;
+	struct mlxsw_sp1_ptp_key key;
+	u8 types;
+
+	if (WARN_ON_ONCE(!mlxsw_sp_local_port_is_valid(mlxsw_sp, local_port)))
+		return;
+	mlxsw_sp_port = mlxsw_sp->ports[local_port];
+	if (!mlxsw_sp_port)
+		return;
+
+	types = ingress ? mlxsw_sp_port->ptp.ing_types :
+			  mlxsw_sp_port->ptp.egr_types;
+
+	/* For message types whose timestamping was not enabled on this port,
+	 * don't bother with the timestamp.
+	 */
+	if (!((1 << message_type) & types))
+		return;
+
+	memset(&key, 0, sizeof(key));
+	key.local_port = local_port;
+	key.domain_number = domain_number;
+	key.message_type = message_type;
+	key.sequence_id = sequence_id;
+	key.ingress = ingress;
+
+	mlxsw_sp1_ptp_got_piece(mlxsw_sp, key, NULL, timestamp);
+}
+
+void mlxsw_sp1_ptp_receive(struct mlxsw_sp *mlxsw_sp, struct sk_buff *skb,
+			   u16 local_port)
+{
+	skb_reset_mac_header(skb);
+	mlxsw_sp1_ptp_got_packet(mlxsw_sp, skb, local_port, true);
+}
+
+void mlxsw_sp1_ptp_transmitted(struct mlxsw_sp *mlxsw_sp,
+			       struct sk_buff *skb, u16 local_port)
+{
+	mlxsw_sp1_ptp_got_packet(mlxsw_sp, skb, local_port, false);
+}
+
+static void
+mlxsw_sp1_ptp_ht_gc_collect(struct mlxsw_sp1_ptp_state *ptp_state,
+			    struct mlxsw_sp1_ptp_unmatched *unmatched)
+{
+	struct mlxsw_sp *mlxsw_sp = ptp_state->common.mlxsw_sp;
+	struct mlxsw_sp_ptp_port_dir_stats *stats;
+	struct mlxsw_sp_port *mlxsw_sp_port;
+	int err;
+
+	/* If an unmatched entry has an SKB, it has to be handed over to the
+	 * networking stack. This is usually done from a trap handler, which is
+	 * invoked in a softirq context. Here we are going to do it in process
+	 * context. If that were to be interrupted by a softirq, it could cause
+	 * a deadlock when an attempt is made to take an already-taken lock
+	 * somewhere along the sending path. Disable softirqs to prevent this.
+	 */
+	local_bh_disable();
+
+	spin_lock(&ptp_state->unmatched_lock);
+	err = rhltable_remove(&ptp_state->unmatched_ht, &unmatched->ht_node,
+			      mlxsw_sp1_ptp_unmatched_ht_params);
+	spin_unlock(&ptp_state->unmatched_lock);
+
+	if (err)
+		/* The packet was matched with timestamp during the walk. */
+		goto out;
+
+	mlxsw_sp_port = mlxsw_sp->ports[unmatched->key.local_port];
+	if (mlxsw_sp_port) {
+		stats = unmatched->key.ingress ?
+			&mlxsw_sp_port->ptp.stats.rx_gcd :
+			&mlxsw_sp_port->ptp.stats.tx_gcd;
+		if (unmatched->skb)
+			stats->packets++;
+		else
+			stats->timestamps++;
+	}
+
+	/* mlxsw_sp1_ptp_unmatched_finish() invokes netif_receive_skb(). While
+	 * the comment at that function states that it can only be called in
+	 * soft IRQ context, this pattern of local_bh_disable() +
+	 * netif_receive_skb(), in process context, is seen elsewhere in the
+	 * kernel, notably in pktgen.
+	 */
+	mlxsw_sp1_ptp_unmatched_finish(mlxsw_sp, unmatched);
+
+out:
+	local_bh_enable();
+}
+
+static void mlxsw_sp1_ptp_ht_gc(struct work_struct *work)
+{
+	struct delayed_work *dwork = to_delayed_work(work);
+	struct mlxsw_sp1_ptp_unmatched *unmatched;
+	struct mlxsw_sp1_ptp_state *ptp_state;
+	struct rhashtable_iter iter;
+	u32 gc_cycle;
+	void *obj;
+
+	ptp_state = container_of(dwork, struct mlxsw_sp1_ptp_state, ht_gc_dw);
+	gc_cycle = ptp_state->gc_cycle++;
+
+	rhltable_walk_enter(&ptp_state->unmatched_ht, &iter);
+	rhashtable_walk_start(&iter);
+	while ((obj = rhashtable_walk_next(&iter))) {
+		if (IS_ERR(obj))
+			continue;
+
+		unmatched = obj;
+		if (unmatched->gc_cycle <= gc_cycle)
+			mlxsw_sp1_ptp_ht_gc_collect(ptp_state, unmatched);
+	}
+	rhashtable_walk_stop(&iter);
+	rhashtable_walk_exit(&iter);
+
+	mlxsw_core_schedule_dw(&ptp_state->ht_gc_dw,
+			       MLXSW_SP1_PTP_HT_GC_INTERVAL);
+}
+
+static int mlxsw_sp_ptp_mtptpt_set(struct mlxsw_sp *mlxsw_sp,
+				   enum mlxsw_reg_mtptpt_trap_id trap_id,
+				   u16 message_type)
+{
+	char mtptpt_pl[MLXSW_REG_MTPTPT_LEN];
+
+	mlxsw_reg_mtptpt_pack(mtptpt_pl, trap_id, message_type);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(mtptpt), mtptpt_pl);
+}
+
+static int mlxsw_sp1_ptp_set_fifo_clr_on_trap(struct mlxsw_sp *mlxsw_sp,
+					      bool clr)
+{
+	char mogcr_pl[MLXSW_REG_MOGCR_LEN] = {0};
+	int err;
+
+	err = mlxsw_reg_query(mlxsw_sp->core, MLXSW_REG(mogcr), mogcr_pl);
+	if (err)
+		return err;
+
+	mlxsw_reg_mogcr_ptp_iftc_set(mogcr_pl, clr);
+	mlxsw_reg_mogcr_ptp_eftc_set(mogcr_pl, clr);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(mogcr), mogcr_pl);
+}
+
+static int mlxsw_sp1_ptp_mtpppc_set(struct mlxsw_sp *mlxsw_sp,
+				    u16 ing_types, u16 egr_types)
+{
+	char mtpppc_pl[MLXSW_REG_MTPPPC_LEN];
+
+	mlxsw_reg_mtpppc_pack(mtpppc_pl, ing_types, egr_types);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(mtpppc), mtpppc_pl);
+}
+
+struct mlxsw_sp1_ptp_shaper_params {
+	u32 ethtool_speed;
+	enum mlxsw_reg_qpsc_port_speed port_speed;
+	u8 shaper_time_exp;
+	u8 shaper_time_mantissa;
+	u8 shaper_inc;
+	u8 shaper_bs;
+	u8 port_to_shaper_credits;
+	int ing_timestamp_inc;
+	int egr_timestamp_inc;
+};
+
+static const struct mlxsw_sp1_ptp_shaper_params
+mlxsw_sp1_ptp_shaper_params[] = {
+	{
+		.ethtool_speed		= SPEED_100,
+		.port_speed		= MLXSW_REG_QPSC_PORT_SPEED_100M,
+		.shaper_time_exp	= 4,
+		.shaper_time_mantissa	= 12,
+		.shaper_inc		= 9,
+		.shaper_bs		= 1,
+		.port_to_shaper_credits	= 1,
+		.ing_timestamp_inc	= -313,
+		.egr_timestamp_inc	= 313,
+	},
+	{
+		.ethtool_speed		= SPEED_1000,
+		.port_speed		= MLXSW_REG_QPSC_PORT_SPEED_1G,
+		.shaper_time_exp	= 0,
+		.shaper_time_mantissa	= 12,
+		.shaper_inc		= 6,
+		.shaper_bs		= 0,
+		.port_to_shaper_credits	= 1,
+		.ing_timestamp_inc	= -35,
+		.egr_timestamp_inc	= 35,
+	},
+	{
+		.ethtool_speed		= SPEED_10000,
+		.port_speed		= MLXSW_REG_QPSC_PORT_SPEED_10G,
+		.shaper_time_exp	= 0,
+		.shaper_time_mantissa	= 2,
+		.shaper_inc		= 14,
+		.shaper_bs		= 1,
+		.port_to_shaper_credits	= 1,
+		.ing_timestamp_inc	= -11,
+		.egr_timestamp_inc	= 11,
+	},
+	{
+		.ethtool_speed		= SPEED_25000,
+		.port_speed		= MLXSW_REG_QPSC_PORT_SPEED_25G,
+		.shaper_time_exp	= 0,
+		.shaper_time_mantissa	= 0,
+		.shaper_inc		= 11,
+		.shaper_bs		= 1,
+		.port_to_shaper_credits	= 1,
+		.ing_timestamp_inc	= -14,
+		.egr_timestamp_inc	= 14,
+	},
+};
+
+#define MLXSW_SP1_PTP_SHAPER_PARAMS_LEN ARRAY_SIZE(mlxsw_sp1_ptp_shaper_params)
+
+static int mlxsw_sp1_ptp_shaper_params_set(struct mlxsw_sp *mlxsw_sp)
+{
+	const struct mlxsw_sp1_ptp_shaper_params *params;
+	char qpsc_pl[MLXSW_REG_QPSC_LEN];
+	int i, err;
+
+	for (i = 0; i < MLXSW_SP1_PTP_SHAPER_PARAMS_LEN; i++) {
+		params = &mlxsw_sp1_ptp_shaper_params[i];
+		mlxsw_reg_qpsc_pack(qpsc_pl, params->port_speed,
+				    params->shaper_time_exp,
+				    params->shaper_time_mantissa,
+				    params->shaper_inc, params->shaper_bs,
+				    params->port_to_shaper_credits,
+				    params->ing_timestamp_inc,
+				    params->egr_timestamp_inc);
+		err = mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(qpsc), qpsc_pl);
+		if (err)
+			return err;
+	}
+
+	return 0;
+}
+
+static int mlxsw_sp_ptp_traps_set(struct mlxsw_sp *mlxsw_sp)
+{
+	u16 event_message_type;
+	int err;
+
+	/* Deliver these message types as PTP0. */
+	event_message_type = BIT(PTP_MSGTYPE_SYNC) |
+			     BIT(PTP_MSGTYPE_DELAY_REQ) |
+			     BIT(PTP_MSGTYPE_PDELAY_REQ) |
+			     BIT(PTP_MSGTYPE_PDELAY_RESP);
+
+	err = mlxsw_sp_ptp_mtptpt_set(mlxsw_sp, MLXSW_REG_MTPTPT_TRAP_ID_PTP0,
+				      event_message_type);
+	if (err)
+		return err;
+
+	/* Everything else is PTP1. */
+	err = mlxsw_sp_ptp_mtptpt_set(mlxsw_sp, MLXSW_REG_MTPTPT_TRAP_ID_PTP1,
+				      ~event_message_type);
+	if (err)
+		goto err_mtptpt1_set;
+
+	return 0;
+
+err_mtptpt1_set:
+	mlxsw_sp_ptp_mtptpt_set(mlxsw_sp, MLXSW_REG_MTPTPT_TRAP_ID_PTP0, 0);
+	return err;
+}
+
+static void mlxsw_sp_ptp_traps_unset(struct mlxsw_sp *mlxsw_sp)
+{
+	mlxsw_sp_ptp_mtptpt_set(mlxsw_sp, MLXSW_REG_MTPTPT_TRAP_ID_PTP1, 0);
+	mlxsw_sp_ptp_mtptpt_set(mlxsw_sp, MLXSW_REG_MTPTPT_TRAP_ID_PTP0, 0);
+}
+
+struct mlxsw_sp_ptp_state *mlxsw_sp1_ptp_init(struct mlxsw_sp *mlxsw_sp)
+{
+	struct mlxsw_sp1_ptp_state *ptp_state;
+	int err;
+
+	err = mlxsw_sp1_ptp_shaper_params_set(mlxsw_sp);
+	if (err)
+		return ERR_PTR(err);
+
+	ptp_state = kzalloc(sizeof(*ptp_state), GFP_KERNEL);
+	if (!ptp_state)
+		return ERR_PTR(-ENOMEM);
+	ptp_state->common.mlxsw_sp = mlxsw_sp;
+
+	spin_lock_init(&ptp_state->unmatched_lock);
+
+	err = rhltable_init(&ptp_state->unmatched_ht,
+			    &mlxsw_sp1_ptp_unmatched_ht_params);
+	if (err)
+		goto err_hashtable_init;
+
+	err = mlxsw_sp_ptp_traps_set(mlxsw_sp);
+	if (err)
+		goto err_ptp_traps_set;
+
+	err = mlxsw_sp1_ptp_set_fifo_clr_on_trap(mlxsw_sp, true);
+	if (err)
+		goto err_fifo_clr;
+
+	INIT_DELAYED_WORK(&ptp_state->ht_gc_dw, mlxsw_sp1_ptp_ht_gc);
+	mlxsw_core_schedule_dw(&ptp_state->ht_gc_dw,
+			       MLXSW_SP1_PTP_HT_GC_INTERVAL);
+	return &ptp_state->common;
+
+err_fifo_clr:
+	mlxsw_sp_ptp_traps_unset(mlxsw_sp);
+err_ptp_traps_set:
+	rhltable_destroy(&ptp_state->unmatched_ht);
+err_hashtable_init:
+	kfree(ptp_state);
+	return ERR_PTR(err);
+}
+
+void mlxsw_sp1_ptp_fini(struct mlxsw_sp_ptp_state *ptp_state_common)
+{
+	struct mlxsw_sp *mlxsw_sp = ptp_state_common->mlxsw_sp;
+	struct mlxsw_sp1_ptp_state *ptp_state;
+
+	ptp_state = mlxsw_sp1_ptp_state(mlxsw_sp);
+
+	cancel_delayed_work_sync(&ptp_state->ht_gc_dw);
+	mlxsw_sp1_ptp_mtpppc_set(mlxsw_sp, 0, 0);
+	mlxsw_sp1_ptp_set_fifo_clr_on_trap(mlxsw_sp, false);
+	mlxsw_sp_ptp_traps_unset(mlxsw_sp);
+	rhltable_free_and_destroy(&ptp_state->unmatched_ht,
+				  &mlxsw_sp1_ptp_unmatched_free_fn, NULL);
+	kfree(ptp_state);
+}
+
+int mlxsw_sp1_ptp_hwtstamp_get(struct mlxsw_sp_port *mlxsw_sp_port,
+			       struct hwtstamp_config *config)
+{
+	*config = mlxsw_sp_port->ptp.hwtstamp_config;
+	return 0;
+}
+
+static int
+mlxsw_sp1_ptp_get_message_types(const struct hwtstamp_config *config,
+				u16 *p_ing_types, u16 *p_egr_types,
+				enum hwtstamp_rx_filters *p_rx_filter)
+{
+	enum hwtstamp_rx_filters rx_filter = config->rx_filter;
+	enum hwtstamp_tx_types tx_type = config->tx_type;
+	u16 ing_types = 0x00;
+	u16 egr_types = 0x00;
+
+	switch (tx_type) {
+	case HWTSTAMP_TX_OFF:
+		egr_types = 0x00;
+		break;
+	case HWTSTAMP_TX_ON:
+		egr_types = 0xff;
+		break;
+	case HWTSTAMP_TX_ONESTEP_SYNC:
+	case HWTSTAMP_TX_ONESTEP_P2P:
+		return -ERANGE;
+	default:
+		return -EINVAL;
+	}
+
+	switch (rx_filter) {
+	case HWTSTAMP_FILTER_NONE:
+		ing_types = 0x00;
+		break;
+	case HWTSTAMP_FILTER_PTP_V1_L4_SYNC:
+	case HWTSTAMP_FILTER_PTP_V2_L4_SYNC:
+	case HWTSTAMP_FILTER_PTP_V2_L2_SYNC:
+	case HWTSTAMP_FILTER_PTP_V2_SYNC:
+		ing_types = 0x01;
+		break;
+	case HWTSTAMP_FILTER_PTP_V1_L4_DELAY_REQ:
+	case HWTSTAMP_FILTER_PTP_V2_L4_DELAY_REQ:
+	case HWTSTAMP_FILTER_PTP_V2_L2_DELAY_REQ:
+	case HWTSTAMP_FILTER_PTP_V2_DELAY_REQ:
+		ing_types = 0x02;
+		break;
+	case HWTSTAMP_FILTER_PTP_V1_L4_EVENT:
+	case HWTSTAMP_FILTER_PTP_V2_L4_EVENT:
+	case HWTSTAMP_FILTER_PTP_V2_L2_EVENT:
+	case HWTSTAMP_FILTER_PTP_V2_EVENT:
+		ing_types = 0x0f;
+		break;
+	case HWTSTAMP_FILTER_ALL:
+		ing_types = 0xff;
+		break;
+	case HWTSTAMP_FILTER_SOME:
+	case HWTSTAMP_FILTER_NTP_ALL:
+		return -ERANGE;
+	default:
+		return -EINVAL;
+	}
+
+	*p_ing_types = ing_types;
+	*p_egr_types = egr_types;
+	*p_rx_filter = rx_filter;
+	return 0;
+}
+
+static int mlxsw_sp1_ptp_mtpppc_update(struct mlxsw_sp_port *mlxsw_sp_port,
+				       u16 ing_types, u16 egr_types)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	struct mlxsw_sp_port *tmp;
+	u16 orig_ing_types = 0;
+	u16 orig_egr_types = 0;
+	int err;
+	int i;
+
+	/* MTPPPC configures timestamping globally, not per port. Find the
+	 * configuration that contains all configured timestamping requests.
+	 */
+	for (i = 1; i < mlxsw_core_max_ports(mlxsw_sp->core); i++) {
+		tmp = mlxsw_sp->ports[i];
+		if (tmp) {
+			orig_ing_types |= tmp->ptp.ing_types;
+			orig_egr_types |= tmp->ptp.egr_types;
+		}
+		if (tmp && tmp != mlxsw_sp_port) {
+			ing_types |= tmp->ptp.ing_types;
+			egr_types |= tmp->ptp.egr_types;
+		}
+	}
+
+	if ((ing_types || egr_types) && !(orig_ing_types || orig_egr_types)) {
+		err = mlxsw_sp_parsing_depth_inc(mlxsw_sp);
+		if (err) {
+			netdev_err(mlxsw_sp_port->dev, "Failed to increase parsing depth");
+			return err;
+		}
+	}
+	if (!(ing_types || egr_types) && (orig_ing_types || orig_egr_types))
+		mlxsw_sp_parsing_depth_dec(mlxsw_sp);
+
+	return mlxsw_sp1_ptp_mtpppc_set(mlxsw_sp_port->mlxsw_sp,
+				       ing_types, egr_types);
+}
+
+static bool mlxsw_sp1_ptp_hwtstamp_enabled(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	return mlxsw_sp_port->ptp.ing_types || mlxsw_sp_port->ptp.egr_types;
+}
+
+static int
+mlxsw_sp1_ptp_port_shaper_set(struct mlxsw_sp_port *mlxsw_sp_port, bool enable)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	char qeec_pl[MLXSW_REG_QEEC_LEN];
+
+	mlxsw_reg_qeec_ptps_pack(qeec_pl, mlxsw_sp_port->local_port, enable);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(qeec), qeec_pl);
+}
+
+static int mlxsw_sp1_ptp_port_shaper_check(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	bool ptps = false;
+	int err, i;
+	u32 speed;
+
+	if (!mlxsw_sp1_ptp_hwtstamp_enabled(mlxsw_sp_port))
+		return mlxsw_sp1_ptp_port_shaper_set(mlxsw_sp_port, false);
+
+	err = mlxsw_sp_port_speed_get(mlxsw_sp_port, &speed);
+	if (err)
+		return err;
+
+	for (i = 0; i < MLXSW_SP1_PTP_SHAPER_PARAMS_LEN; i++) {
+		if (mlxsw_sp1_ptp_shaper_params[i].ethtool_speed == speed) {
+			ptps = true;
+			break;
+		}
+	}
+
+	return mlxsw_sp1_ptp_port_shaper_set(mlxsw_sp_port, ptps);
+}
+
+void mlxsw_sp1_ptp_shaper_work(struct work_struct *work)
+{
+	struct delayed_work *dwork = to_delayed_work(work);
+	struct mlxsw_sp_port *mlxsw_sp_port;
+	int err;
+
+	mlxsw_sp_port = container_of(dwork, struct mlxsw_sp_port,
+				     ptp.shaper_dw);
+
+	if (!mlxsw_sp1_ptp_hwtstamp_enabled(mlxsw_sp_port))
+		return;
+
+	err = mlxsw_sp1_ptp_port_shaper_check(mlxsw_sp_port);
+	if (err)
+		netdev_err(mlxsw_sp_port->dev, "Failed to set up PTP shaper\n");
+}
+
+int mlxsw_sp1_ptp_hwtstamp_set(struct mlxsw_sp_port *mlxsw_sp_port,
+			       struct hwtstamp_config *config)
+{
+	enum hwtstamp_rx_filters rx_filter;
+	u16 ing_types;
+	u16 egr_types;
+	int err;
+
+	err = mlxsw_sp1_ptp_get_message_types(config, &ing_types, &egr_types,
+					      &rx_filter);
+	if (err)
+		return err;
+
+	err = mlxsw_sp1_ptp_mtpppc_update(mlxsw_sp_port, ing_types, egr_types);
+	if (err)
+		return err;
+
+	mlxsw_sp_port->ptp.hwtstamp_config = *config;
+	mlxsw_sp_port->ptp.ing_types = ing_types;
+	mlxsw_sp_port->ptp.egr_types = egr_types;
+
+	err = mlxsw_sp1_ptp_port_shaper_check(mlxsw_sp_port);
+	if (err)
+		return err;
+
+	/* Notify the ioctl caller what we are actually timestamping. */
+	config->rx_filter = rx_filter;
+
+	return 0;
+}
+
+int mlxsw_sp1_ptp_get_ts_info(struct mlxsw_sp *mlxsw_sp,
+			      struct ethtool_ts_info *info)
+{
+	info->phc_index = ptp_clock_index(mlxsw_sp->clock->ptp);
+
+	info->so_timestamping = SOF_TIMESTAMPING_TX_HARDWARE |
+				SOF_TIMESTAMPING_RX_HARDWARE |
+				SOF_TIMESTAMPING_RAW_HARDWARE;
+
+	info->tx_types = BIT(HWTSTAMP_TX_OFF) |
+			 BIT(HWTSTAMP_TX_ON);
+
+	info->rx_filters = BIT(HWTSTAMP_FILTER_NONE) |
+			   BIT(HWTSTAMP_FILTER_ALL);
+
+	return 0;
+}
+
+struct mlxsw_sp_ptp_port_stat {
+	char str[ETH_GSTRING_LEN];
+	ptrdiff_t offset;
+};
+
+#define MLXSW_SP_PTP_PORT_STAT(NAME, FIELD)				\
+	{								\
+		.str = NAME,						\
+		.offset = offsetof(struct mlxsw_sp_ptp_port_stats,	\
+				    FIELD),				\
+	}
+
+static const struct mlxsw_sp_ptp_port_stat mlxsw_sp_ptp_port_stats[] = {
+	MLXSW_SP_PTP_PORT_STAT("ptp_rx_gcd_packets",    rx_gcd.packets),
+	MLXSW_SP_PTP_PORT_STAT("ptp_rx_gcd_timestamps", rx_gcd.timestamps),
+	MLXSW_SP_PTP_PORT_STAT("ptp_tx_gcd_packets",    tx_gcd.packets),
+	MLXSW_SP_PTP_PORT_STAT("ptp_tx_gcd_timestamps", tx_gcd.timestamps),
+};
+
+#undef MLXSW_SP_PTP_PORT_STAT
+
+#define MLXSW_SP_PTP_PORT_STATS_LEN \
+	ARRAY_SIZE(mlxsw_sp_ptp_port_stats)
+
+int mlxsw_sp1_get_stats_count(void)
+{
+	return MLXSW_SP_PTP_PORT_STATS_LEN;
+}
+
+void mlxsw_sp1_get_stats_strings(u8 **p)
+{
+	int i;
+
+	for (i = 0; i < MLXSW_SP_PTP_PORT_STATS_LEN; i++) {
+		memcpy(*p, mlxsw_sp_ptp_port_stats[i].str,
+		       ETH_GSTRING_LEN);
+		*p += ETH_GSTRING_LEN;
+	}
+}
+
+void mlxsw_sp1_get_stats(struct mlxsw_sp_port *mlxsw_sp_port,
+			 u64 *data, int data_index)
+{
+	void *stats = &mlxsw_sp_port->ptp.stats;
+	ptrdiff_t offset;
+	int i;
+
+	data += data_index;
+	for (i = 0; i < MLXSW_SP_PTP_PORT_STATS_LEN; i++) {
+		offset = mlxsw_sp_ptp_port_stats[i].offset;
+		*data++ = *(u64 *)(stats + offset);
+	}
+}
+
+struct mlxsw_sp_ptp_state *mlxsw_sp2_ptp_init(struct mlxsw_sp *mlxsw_sp)
+{
+	struct mlxsw_sp2_ptp_state *ptp_state;
+	int err;
+
+	ptp_state = kzalloc(sizeof(*ptp_state), GFP_KERNEL);
+	if (!ptp_state)
+		return ERR_PTR(-ENOMEM);
+
+	ptp_state->common.mlxsw_sp = mlxsw_sp;
+
+	err = mlxsw_sp_ptp_traps_set(mlxsw_sp);
+	if (err)
+		goto err_ptp_traps_set;
+
+	refcount_set(&ptp_state->ptp_port_enabled_ref, 0);
+	mutex_init(&ptp_state->lock);
+	return &ptp_state->common;
+
+err_ptp_traps_set:
+	kfree(ptp_state);
+	return ERR_PTR(err);
+}
+
+void mlxsw_sp2_ptp_fini(struct mlxsw_sp_ptp_state *ptp_state_common)
+{
+	struct mlxsw_sp *mlxsw_sp = ptp_state_common->mlxsw_sp;
+	struct mlxsw_sp2_ptp_state *ptp_state;
+
+	ptp_state = mlxsw_sp2_ptp_state(mlxsw_sp);
+
+	mutex_destroy(&ptp_state->lock);
+	mlxsw_sp_ptp_traps_unset(mlxsw_sp);
+	kfree(ptp_state);
+}
+
+static u32 mlxsw_ptp_utc_time_stamp_sec_get(struct mlxsw_core *mlxsw_core,
+					    u8 cqe_ts_sec)
+{
+	u32 utc_sec = mlxsw_core_read_utc_sec(mlxsw_core);
+
+	if (cqe_ts_sec > (utc_sec & 0xff))
+		/* Time stamp above the last bits of UTC (UTC & 0xff) means the
+		 * latter has wrapped after the time stamp was collected.
+		 */
+		utc_sec -= 256;
+
+	utc_sec &= ~0xff;
+	utc_sec |= cqe_ts_sec;
+
+	return utc_sec;
+}
+
+static void mlxsw_sp2_ptp_hwtstamp_fill(struct mlxsw_core *mlxsw_core,
+					const struct mlxsw_skb_cb *cb,
+					struct skb_shared_hwtstamps *hwtstamps)
+{
+	u64 ts_sec, ts_nsec, nsec;
+
+	WARN_ON_ONCE(!cb->cqe_ts.sec && !cb->cqe_ts.nsec);
+
+	/* The time stamp in the CQE is represented by 38 bits, which is a short
+	 * representation of UTC time. Software should create the full time
+	 * stamp using the global UTC clock. The seconds have only 8 bits in the
+	 * CQE, to create the full time stamp, use the current UTC time and fix
+	 * the seconds according to the relation between UTC seconds and CQE
+	 * seconds.
+	 */
+	ts_sec = mlxsw_ptp_utc_time_stamp_sec_get(mlxsw_core, cb->cqe_ts.sec);
+	ts_nsec = cb->cqe_ts.nsec;
+
+	nsec = ts_sec * NSEC_PER_SEC + ts_nsec;
+
+	hwtstamps->hwtstamp = ns_to_ktime(nsec);
+}
+
+void mlxsw_sp2_ptp_receive(struct mlxsw_sp *mlxsw_sp, struct sk_buff *skb,
+			   u16 local_port)
+{
+	struct skb_shared_hwtstamps hwtstamps;
+
+	mlxsw_sp2_ptp_hwtstamp_fill(mlxsw_sp->core, mlxsw_skb_cb(skb),
+				    &hwtstamps);
+	*skb_hwtstamps(skb) = hwtstamps;
+	mlxsw_sp_rx_listener_no_mark_func(skb, local_port, mlxsw_sp);
+}
+
+void mlxsw_sp2_ptp_transmitted(struct mlxsw_sp *mlxsw_sp,
+			       struct sk_buff *skb, u16 local_port)
+{
+	struct skb_shared_hwtstamps hwtstamps;
+
+	mlxsw_sp2_ptp_hwtstamp_fill(mlxsw_sp->core, mlxsw_skb_cb(skb),
+				    &hwtstamps);
+	skb_tstamp_tx(skb, &hwtstamps);
+	dev_kfree_skb_any(skb);
+}
+
+int mlxsw_sp2_ptp_hwtstamp_get(struct mlxsw_sp_port *mlxsw_sp_port,
+			       struct hwtstamp_config *config)
+{
+	struct mlxsw_sp2_ptp_state *ptp_state;
+
+	ptp_state = mlxsw_sp2_ptp_state(mlxsw_sp_port->mlxsw_sp);
+
+	mutex_lock(&ptp_state->lock);
+	*config = ptp_state->config;
+	mutex_unlock(&ptp_state->lock);
+
+	return 0;
+}
+
+static int
+mlxsw_sp2_ptp_get_message_types(const struct hwtstamp_config *config,
+				u16 *p_ing_types, u16 *p_egr_types,
+				enum hwtstamp_rx_filters *p_rx_filter)
+{
+	enum hwtstamp_rx_filters rx_filter = config->rx_filter;
+	enum hwtstamp_tx_types tx_type = config->tx_type;
+	u16 ing_types = 0x00;
+	u16 egr_types = 0x00;
+
+	*p_rx_filter = rx_filter;
+
+	switch (rx_filter) {
+	case HWTSTAMP_FILTER_NONE:
+		ing_types = 0x00;
+		break;
+	case HWTSTAMP_FILTER_PTP_V1_L4_SYNC:
+	case HWTSTAMP_FILTER_PTP_V2_L4_SYNC:
+	case HWTSTAMP_FILTER_PTP_V2_L2_SYNC:
+	case HWTSTAMP_FILTER_PTP_V2_SYNC:
+	case HWTSTAMP_FILTER_PTP_V1_L4_DELAY_REQ:
+	case HWTSTAMP_FILTER_PTP_V2_L4_DELAY_REQ:
+	case HWTSTAMP_FILTER_PTP_V2_L2_DELAY_REQ:
+	case HWTSTAMP_FILTER_PTP_V2_DELAY_REQ:
+	case HWTSTAMP_FILTER_PTP_V1_L4_EVENT:
+	case HWTSTAMP_FILTER_PTP_V2_L4_EVENT:
+	case HWTSTAMP_FILTER_PTP_V2_L2_EVENT:
+	case HWTSTAMP_FILTER_PTP_V2_EVENT:
+		/* In Spectrum-2 and above, all packets get time stamp by
+		 * default and the driver fill the time stamp only for event
+		 * packets. Return all event types even if only specific types
+		 * were required.
+		 */
+		ing_types = 0x0f;
+		*p_rx_filter = HWTSTAMP_FILTER_SOME;
+		break;
+	case HWTSTAMP_FILTER_ALL:
+	case HWTSTAMP_FILTER_SOME:
+	case HWTSTAMP_FILTER_NTP_ALL:
+		return -ERANGE;
+	default:
+		return -EINVAL;
+	}
+
+	switch (tx_type) {
+	case HWTSTAMP_TX_OFF:
+		egr_types = 0x00;
+		break;
+	case HWTSTAMP_TX_ON:
+		egr_types = 0x0f;
+		break;
+	case HWTSTAMP_TX_ONESTEP_SYNC:
+	case HWTSTAMP_TX_ONESTEP_P2P:
+		return -ERANGE;
+	default:
+		return -EINVAL;
+	}
+
+	if ((ing_types && !egr_types) || (!ing_types && egr_types))
+		return -EINVAL;
+
+	*p_ing_types = ing_types;
+	*p_egr_types = egr_types;
+	return 0;
+}
+
+static int mlxsw_sp2_ptp_mtpcpc_set(struct mlxsw_sp *mlxsw_sp, bool ptp_trap_en,
+				    u16 ing_types, u16 egr_types)
+{
+	char mtpcpc_pl[MLXSW_REG_MTPCPC_LEN];
+
+	mlxsw_reg_mtpcpc_pack(mtpcpc_pl, false, 0, ptp_trap_en, ing_types,
+			      egr_types);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(mtpcpc), mtpcpc_pl);
+}
+
+static int mlxsw_sp2_ptp_enable(struct mlxsw_sp *mlxsw_sp, u16 ing_types,
+				u16 egr_types,
+				struct hwtstamp_config new_config)
+{
+	struct mlxsw_sp2_ptp_state *ptp_state = mlxsw_sp2_ptp_state(mlxsw_sp);
+	int err;
+
+	err = mlxsw_sp2_ptp_mtpcpc_set(mlxsw_sp, true, ing_types, egr_types);
+	if (err)
+		return err;
+
+	ptp_state->config = new_config;
+	return 0;
+}
+
+static int mlxsw_sp2_ptp_disable(struct mlxsw_sp *mlxsw_sp,
+				 struct hwtstamp_config new_config)
+{
+	struct mlxsw_sp2_ptp_state *ptp_state = mlxsw_sp2_ptp_state(mlxsw_sp);
+	int err;
+
+	err = mlxsw_sp2_ptp_mtpcpc_set(mlxsw_sp, false, 0, 0);
+	if (err)
+		return err;
+
+	ptp_state->config = new_config;
+	return 0;
+}
+
+static int mlxsw_sp2_ptp_configure_port(struct mlxsw_sp_port *mlxsw_sp_port,
+					u16 ing_types, u16 egr_types,
+					struct hwtstamp_config new_config)
+{
+	struct mlxsw_sp2_ptp_state *ptp_state;
+	int err;
+
+	ptp_state = mlxsw_sp2_ptp_state(mlxsw_sp_port->mlxsw_sp);
+
+	if (refcount_inc_not_zero(&ptp_state->ptp_port_enabled_ref))
+		return 0;
+
+	err = mlxsw_sp2_ptp_enable(mlxsw_sp_port->mlxsw_sp, ing_types,
+				   egr_types, new_config);
+	if (err)
+		return err;
+
+	refcount_set(&ptp_state->ptp_port_enabled_ref, 1);
+
+	return 0;
+}
+
+static int mlxsw_sp2_ptp_deconfigure_port(struct mlxsw_sp_port *mlxsw_sp_port,
+					  struct hwtstamp_config new_config)
+{
+	struct mlxsw_sp2_ptp_state *ptp_state;
+	int err;
+
+	ptp_state = mlxsw_sp2_ptp_state(mlxsw_sp_port->mlxsw_sp);
+
+	if (!refcount_dec_and_test(&ptp_state->ptp_port_enabled_ref))
+		return 0;
+
+	err = mlxsw_sp2_ptp_disable(mlxsw_sp_port->mlxsw_sp, new_config);
+	if (err)
+		goto err_ptp_disable;
+
+	return 0;
+
+err_ptp_disable:
+	refcount_set(&ptp_state->ptp_port_enabled_ref, 1);
+	return err;
+}
+
+int mlxsw_sp2_ptp_hwtstamp_set(struct mlxsw_sp_port *mlxsw_sp_port,
+			       struct hwtstamp_config *config)
+{
+	struct mlxsw_sp2_ptp_state *ptp_state;
+	enum hwtstamp_rx_filters rx_filter;
+	struct hwtstamp_config new_config;
+	u16 new_ing_types, new_egr_types;
+	bool ptp_enabled;
+	int err;
+
+	ptp_state = mlxsw_sp2_ptp_state(mlxsw_sp_port->mlxsw_sp);
+	mutex_lock(&ptp_state->lock);
+
+	err = mlxsw_sp2_ptp_get_message_types(config, &new_ing_types,
+					      &new_egr_types, &rx_filter);
+	if (err)
+		goto err_get_message_types;
+
+	new_config.flags = config->flags;
+	new_config.tx_type = config->tx_type;
+	new_config.rx_filter = rx_filter;
+
+	ptp_enabled = mlxsw_sp_port->ptp.ing_types ||
+		      mlxsw_sp_port->ptp.egr_types;
+
+	if ((new_ing_types || new_egr_types) && !ptp_enabled) {
+		err = mlxsw_sp2_ptp_configure_port(mlxsw_sp_port, new_ing_types,
+						   new_egr_types, new_config);
+		if (err)
+			goto err_configure_port;
+	} else if (!new_ing_types && !new_egr_types && ptp_enabled) {
+		err = mlxsw_sp2_ptp_deconfigure_port(mlxsw_sp_port, new_config);
+		if (err)
+			goto err_deconfigure_port;
+	}
+
+	mlxsw_sp_port->ptp.ing_types = new_ing_types;
+	mlxsw_sp_port->ptp.egr_types = new_egr_types;
+
+	/* Notify the ioctl caller what we are actually timestamping. */
+	config->rx_filter = rx_filter;
+	mutex_unlock(&ptp_state->lock);
+
+	return 0;
+
+err_deconfigure_port:
+err_configure_port:
+err_get_message_types:
+	mutex_unlock(&ptp_state->lock);
+	return err;
+}
+
+int mlxsw_sp2_ptp_get_ts_info(struct mlxsw_sp *mlxsw_sp,
+			      struct ethtool_ts_info *info)
+{
+	info->phc_index = ptp_clock_index(mlxsw_sp->clock->ptp);
+
+	info->so_timestamping = SOF_TIMESTAMPING_TX_HARDWARE |
+				SOF_TIMESTAMPING_RX_HARDWARE |
+				SOF_TIMESTAMPING_RAW_HARDWARE;
+
+	info->tx_types = BIT(HWTSTAMP_TX_OFF) |
+			 BIT(HWTSTAMP_TX_ON);
+
+	info->rx_filters = BIT(HWTSTAMP_FILTER_NONE) |
+			   BIT(HWTSTAMP_FILTER_PTP_V1_L4_EVENT) |
+			   BIT(HWTSTAMP_FILTER_PTP_V2_EVENT);
+
+	return 0;
+}
+
+int mlxsw_sp_ptp_txhdr_construct(struct mlxsw_core *mlxsw_core,
+				 struct mlxsw_sp_port *mlxsw_sp_port,
+				 struct sk_buff *skb,
+				 const struct mlxsw_tx_info *tx_info)
+{
+	mlxsw_sp_txhdr_construct(skb, tx_info);
+	return 0;
+}
+
+int mlxsw_sp2_ptp_txhdr_construct(struct mlxsw_core *mlxsw_core,
+				  struct mlxsw_sp_port *mlxsw_sp_port,
+				  struct sk_buff *skb,
+				  const struct mlxsw_tx_info *tx_info)
+{
+	/* In Spectrum-2 and Spectrum-3, in order for PTP event packets to have
+	 * their correction field correctly set on the egress port they must be
+	 * transmitted as data packets. Such packets ingress the ASIC via the
+	 * CPU port and must have a VLAN tag, as the CPU port is not configured
+	 * with a PVID. Push the default VLAN (4095), which is configured as
+	 * egress untagged on all the ports.
+	 */
+	if (!skb_vlan_tagged(skb)) {
+		skb = vlan_insert_tag_set_proto(skb, htons(ETH_P_8021Q),
+						MLXSW_SP_DEFAULT_VID);
+		if (!skb) {
+			this_cpu_inc(mlxsw_sp_port->pcpu_stats->tx_dropped);
+			return -ENOMEM;
+		}
+	}
+
+	return mlxsw_sp_txhdr_ptp_data_construct(mlxsw_core, mlxsw_sp_port, skb,
+						 tx_info);
+}
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_ptp.h b/drivers/net/ethernet/mellanox/mlxsw/spectrum_ptp.h
new file mode 100644
index 000000000..a8b882309
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_ptp.h
@@ -0,0 +1,263 @@
+/* SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0 */
+/* Copyright (c) 2019 Mellanox Technologies. All rights reserved */
+
+#ifndef _MLXSW_SPECTRUM_PTP_H
+#define _MLXSW_SPECTRUM_PTP_H
+
+#include <linux/device.h>
+#include <linux/rhashtable.h>
+
+struct mlxsw_sp;
+struct mlxsw_sp_port;
+struct mlxsw_sp_ptp_clock;
+
+static inline int mlxsw_sp_ptp_get_ts_info_noptp(struct ethtool_ts_info *info)
+{
+	info->so_timestamping = SOF_TIMESTAMPING_RX_SOFTWARE |
+				SOF_TIMESTAMPING_SOFTWARE;
+	info->phc_index = -1;
+	return 0;
+}
+
+#if IS_REACHABLE(CONFIG_PTP_1588_CLOCK)
+
+struct mlxsw_sp_ptp_clock *
+mlxsw_sp1_ptp_clock_init(struct mlxsw_sp *mlxsw_sp, struct device *dev);
+
+void mlxsw_sp1_ptp_clock_fini(struct mlxsw_sp_ptp_clock *clock);
+
+struct mlxsw_sp_ptp_state *mlxsw_sp1_ptp_init(struct mlxsw_sp *mlxsw_sp);
+
+void mlxsw_sp1_ptp_fini(struct mlxsw_sp_ptp_state *ptp_state);
+
+void mlxsw_sp1_ptp_receive(struct mlxsw_sp *mlxsw_sp, struct sk_buff *skb,
+			   u16 local_port);
+
+void mlxsw_sp1_ptp_transmitted(struct mlxsw_sp *mlxsw_sp,
+			       struct sk_buff *skb, u16 local_port);
+
+void mlxsw_sp1_ptp_got_timestamp(struct mlxsw_sp *mlxsw_sp, bool ingress,
+				 u16 local_port, u8 message_type,
+				 u8 domain_number, u16 sequence_id,
+				 u64 timestamp);
+
+int mlxsw_sp1_ptp_hwtstamp_get(struct mlxsw_sp_port *mlxsw_sp_port,
+			       struct hwtstamp_config *config);
+
+int mlxsw_sp1_ptp_hwtstamp_set(struct mlxsw_sp_port *mlxsw_sp_port,
+			       struct hwtstamp_config *config);
+
+void mlxsw_sp1_ptp_shaper_work(struct work_struct *work);
+
+int mlxsw_sp1_ptp_get_ts_info(struct mlxsw_sp *mlxsw_sp,
+			      struct ethtool_ts_info *info);
+
+int mlxsw_sp1_get_stats_count(void);
+void mlxsw_sp1_get_stats_strings(u8 **p);
+void mlxsw_sp1_get_stats(struct mlxsw_sp_port *mlxsw_sp_port,
+			 u64 *data, int data_index);
+
+int mlxsw_sp_ptp_txhdr_construct(struct mlxsw_core *mlxsw_core,
+				 struct mlxsw_sp_port *mlxsw_sp_port,
+				 struct sk_buff *skb,
+				 const struct mlxsw_tx_info *tx_info);
+
+struct mlxsw_sp_ptp_clock *
+mlxsw_sp2_ptp_clock_init(struct mlxsw_sp *mlxsw_sp, struct device *dev);
+
+void mlxsw_sp2_ptp_clock_fini(struct mlxsw_sp_ptp_clock *clock);
+
+struct mlxsw_sp_ptp_state *mlxsw_sp2_ptp_init(struct mlxsw_sp *mlxsw_sp);
+
+void mlxsw_sp2_ptp_fini(struct mlxsw_sp_ptp_state *ptp_state);
+
+void mlxsw_sp2_ptp_receive(struct mlxsw_sp *mlxsw_sp, struct sk_buff *skb,
+			   u16 local_port);
+
+void mlxsw_sp2_ptp_transmitted(struct mlxsw_sp *mlxsw_sp,
+			       struct sk_buff *skb, u16 local_port);
+
+int mlxsw_sp2_ptp_hwtstamp_get(struct mlxsw_sp_port *mlxsw_sp_port,
+			       struct hwtstamp_config *config);
+
+int mlxsw_sp2_ptp_hwtstamp_set(struct mlxsw_sp_port *mlxsw_sp_port,
+			       struct hwtstamp_config *config);
+
+int mlxsw_sp2_ptp_get_ts_info(struct mlxsw_sp *mlxsw_sp,
+			      struct ethtool_ts_info *info);
+
+int mlxsw_sp2_ptp_txhdr_construct(struct mlxsw_core *mlxsw_core,
+				  struct mlxsw_sp_port *mlxsw_sp_port,
+				  struct sk_buff *skb,
+				  const struct mlxsw_tx_info *tx_info);
+
+#else
+
+static inline struct mlxsw_sp_ptp_clock *
+mlxsw_sp1_ptp_clock_init(struct mlxsw_sp *mlxsw_sp, struct device *dev)
+{
+	return NULL;
+}
+
+static inline void mlxsw_sp1_ptp_clock_fini(struct mlxsw_sp_ptp_clock *clock)
+{
+}
+
+static inline struct mlxsw_sp_ptp_state *
+mlxsw_sp1_ptp_init(struct mlxsw_sp *mlxsw_sp)
+{
+	return NULL;
+}
+
+static inline void mlxsw_sp1_ptp_fini(struct mlxsw_sp_ptp_state *ptp_state)
+{
+}
+
+static inline void mlxsw_sp1_ptp_receive(struct mlxsw_sp *mlxsw_sp,
+					 struct sk_buff *skb, u16 local_port)
+{
+	mlxsw_sp_rx_listener_no_mark_func(skb, local_port, mlxsw_sp);
+}
+
+static inline void mlxsw_sp1_ptp_transmitted(struct mlxsw_sp *mlxsw_sp,
+					     struct sk_buff *skb, u16 local_port)
+{
+	dev_kfree_skb_any(skb);
+}
+
+static inline void
+mlxsw_sp1_ptp_got_timestamp(struct mlxsw_sp *mlxsw_sp, bool ingress,
+			    u16 local_port, u8 message_type,
+			    u8 domain_number,
+			    u16 sequence_id, u64 timestamp)
+{
+}
+
+static inline int
+mlxsw_sp1_ptp_hwtstamp_get(struct mlxsw_sp_port *mlxsw_sp_port,
+			   struct hwtstamp_config *config)
+{
+	return -EOPNOTSUPP;
+}
+
+static inline int
+mlxsw_sp1_ptp_hwtstamp_set(struct mlxsw_sp_port *mlxsw_sp_port,
+			   struct hwtstamp_config *config)
+{
+	return -EOPNOTSUPP;
+}
+
+static inline void mlxsw_sp1_ptp_shaper_work(struct work_struct *work)
+{
+}
+
+static inline int mlxsw_sp1_ptp_get_ts_info(struct mlxsw_sp *mlxsw_sp,
+					    struct ethtool_ts_info *info)
+{
+	return mlxsw_sp_ptp_get_ts_info_noptp(info);
+}
+
+static inline int mlxsw_sp1_get_stats_count(void)
+{
+	return 0;
+}
+
+static inline void mlxsw_sp1_get_stats_strings(u8 **p)
+{
+}
+
+static inline void mlxsw_sp1_get_stats(struct mlxsw_sp_port *mlxsw_sp_port,
+				       u64 *data, int data_index)
+{
+}
+
+static inline int
+mlxsw_sp_ptp_txhdr_construct(struct mlxsw_core *mlxsw_core,
+			     struct mlxsw_sp_port *mlxsw_sp_port,
+			     struct sk_buff *skb,
+			     const struct mlxsw_tx_info *tx_info)
+{
+	return -EOPNOTSUPP;
+}
+
+static inline struct mlxsw_sp_ptp_clock *
+mlxsw_sp2_ptp_clock_init(struct mlxsw_sp *mlxsw_sp, struct device *dev)
+{
+	return NULL;
+}
+
+static inline void mlxsw_sp2_ptp_clock_fini(struct mlxsw_sp_ptp_clock *clock)
+{
+}
+
+static inline struct mlxsw_sp_ptp_state *
+mlxsw_sp2_ptp_init(struct mlxsw_sp *mlxsw_sp)
+{
+	return NULL;
+}
+
+static inline void mlxsw_sp2_ptp_fini(struct mlxsw_sp_ptp_state *ptp_state)
+{
+}
+
+static inline void mlxsw_sp2_ptp_receive(struct mlxsw_sp *mlxsw_sp,
+					 struct sk_buff *skb, u16 local_port)
+{
+	mlxsw_sp_rx_listener_no_mark_func(skb, local_port, mlxsw_sp);
+}
+
+static inline void mlxsw_sp2_ptp_transmitted(struct mlxsw_sp *mlxsw_sp,
+					     struct sk_buff *skb, u16 local_port)
+{
+	dev_kfree_skb_any(skb);
+}
+
+static inline int
+mlxsw_sp2_ptp_hwtstamp_get(struct mlxsw_sp_port *mlxsw_sp_port,
+			   struct hwtstamp_config *config)
+{
+	return -EOPNOTSUPP;
+}
+
+static inline int
+mlxsw_sp2_ptp_hwtstamp_set(struct mlxsw_sp_port *mlxsw_sp_port,
+			   struct hwtstamp_config *config)
+{
+	return -EOPNOTSUPP;
+}
+
+static inline int mlxsw_sp2_ptp_get_ts_info(struct mlxsw_sp *mlxsw_sp,
+					    struct ethtool_ts_info *info)
+{
+	return mlxsw_sp_ptp_get_ts_info_noptp(info);
+}
+
+static inline int
+mlxsw_sp2_ptp_txhdr_construct(struct mlxsw_core *mlxsw_core,
+			      struct mlxsw_sp_port *mlxsw_sp_port,
+			      struct sk_buff *skb,
+			      const struct mlxsw_tx_info *tx_info)
+{
+	return -EOPNOTSUPP;
+}
+#endif
+
+static inline void mlxsw_sp2_ptp_shaper_work(struct work_struct *work)
+{
+}
+
+static inline int mlxsw_sp2_get_stats_count(void)
+{
+	return 0;
+}
+
+static inline void mlxsw_sp2_get_stats_strings(u8 **p)
+{
+}
+
+static inline void mlxsw_sp2_get_stats(struct mlxsw_sp_port *mlxsw_sp_port,
+				       u64 *data, int data_index)
+{
+}
+
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c
new file mode 100644
index 000000000..4243d3b88
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c
@@ -0,0 +1,2337 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2017-2018 Mellanox Technologies. All rights reserved */
+
+#include <linux/kernel.h>
+#include <linux/errno.h>
+#include <linux/netdevice.h>
+#include <net/pkt_cls.h>
+#include <net/red.h>
+
+#include "spectrum.h"
+#include "spectrum_span.h"
+#include "reg.h"
+
+#define MLXSW_SP_PRIO_BAND_TO_TCLASS(band) (IEEE_8021QAZ_MAX_TCS - band - 1)
+#define MLXSW_SP_PRIO_CHILD_TO_TCLASS(child) \
+	MLXSW_SP_PRIO_BAND_TO_TCLASS((child - 1))
+
+enum mlxsw_sp_qdisc_type {
+	MLXSW_SP_QDISC_NO_QDISC,
+	MLXSW_SP_QDISC_RED,
+	MLXSW_SP_QDISC_PRIO,
+	MLXSW_SP_QDISC_ETS,
+	MLXSW_SP_QDISC_TBF,
+	MLXSW_SP_QDISC_FIFO,
+};
+
+struct mlxsw_sp_qdisc;
+
+struct mlxsw_sp_qdisc_ops {
+	enum mlxsw_sp_qdisc_type type;
+	int (*check_params)(struct mlxsw_sp_port *mlxsw_sp_port,
+			    void *params);
+	int (*replace)(struct mlxsw_sp_port *mlxsw_sp_port, u32 handle,
+		       struct mlxsw_sp_qdisc *mlxsw_sp_qdisc, void *params);
+	int (*destroy)(struct mlxsw_sp_port *mlxsw_sp_port,
+		       struct mlxsw_sp_qdisc *mlxsw_sp_qdisc);
+	int (*get_stats)(struct mlxsw_sp_port *mlxsw_sp_port,
+			 struct mlxsw_sp_qdisc *mlxsw_sp_qdisc,
+			 struct tc_qopt_offload_stats *stats_ptr);
+	int (*get_xstats)(struct mlxsw_sp_port *mlxsw_sp_port,
+			  struct mlxsw_sp_qdisc *mlxsw_sp_qdisc,
+			  void *xstats_ptr);
+	void (*clean_stats)(struct mlxsw_sp_port *mlxsw_sp_port,
+			    struct mlxsw_sp_qdisc *mlxsw_sp_qdisc);
+	/* unoffload - to be used for a qdisc that stops being offloaded without
+	 * being destroyed.
+	 */
+	void (*unoffload)(struct mlxsw_sp_port *mlxsw_sp_port,
+			  struct mlxsw_sp_qdisc *mlxsw_sp_qdisc, void *params);
+	struct mlxsw_sp_qdisc *(*find_class)(struct mlxsw_sp_qdisc *mlxsw_sp_qdisc,
+					     u32 parent);
+	unsigned int num_classes;
+
+	u8 (*get_prio_bitmap)(struct mlxsw_sp_qdisc *mlxsw_sp_qdisc,
+			      struct mlxsw_sp_qdisc *child);
+	int (*get_tclass_num)(struct mlxsw_sp_qdisc *mlxsw_sp_qdisc,
+			      struct mlxsw_sp_qdisc *child);
+};
+
+struct mlxsw_sp_qdisc_ets_band {
+	u8 prio_bitmap;
+	int tclass_num;
+};
+
+struct mlxsw_sp_qdisc_ets_data {
+	struct mlxsw_sp_qdisc_ets_band bands[IEEE_8021QAZ_MAX_TCS];
+};
+
+struct mlxsw_sp_qdisc {
+	u32 handle;
+	union {
+		struct red_stats red;
+	} xstats_base;
+	struct mlxsw_sp_qdisc_stats {
+		u64 tx_bytes;
+		u64 tx_packets;
+		u64 drops;
+		u64 overlimits;
+		u64 backlog;
+	} stats_base;
+
+	union {
+		struct mlxsw_sp_qdisc_ets_data *ets_data;
+	};
+
+	struct mlxsw_sp_qdisc_ops *ops;
+	struct mlxsw_sp_qdisc *parent;
+	struct mlxsw_sp_qdisc *qdiscs;
+	unsigned int num_classes;
+};
+
+struct mlxsw_sp_qdisc_state {
+	struct mlxsw_sp_qdisc root_qdisc;
+
+	/* When a PRIO or ETS are added, the invisible FIFOs in their bands are
+	 * created first. When notifications for these FIFOs arrive, it is not
+	 * known what qdisc their parent handle refers to. It could be a
+	 * newly-created PRIO that will replace the currently-offloaded one, or
+	 * it could be e.g. a RED that will be attached below it.
+	 *
+	 * As the notifications start to arrive, use them to note what the
+	 * future parent handle is, and keep track of which child FIFOs were
+	 * seen. Then when the parent is known, retroactively offload those
+	 * FIFOs.
+	 */
+	u32 future_handle;
+	bool future_fifos[IEEE_8021QAZ_MAX_TCS];
+	struct mutex lock; /* Protects qdisc state. */
+};
+
+static bool
+mlxsw_sp_qdisc_compare(struct mlxsw_sp_qdisc *mlxsw_sp_qdisc, u32 handle)
+{
+	return mlxsw_sp_qdisc->ops && mlxsw_sp_qdisc->handle == handle;
+}
+
+static struct mlxsw_sp_qdisc *
+mlxsw_sp_qdisc_walk(struct mlxsw_sp_qdisc *qdisc,
+		    struct mlxsw_sp_qdisc *(*pre)(struct mlxsw_sp_qdisc *,
+						  void *),
+		    void *data)
+{
+	struct mlxsw_sp_qdisc *tmp;
+	unsigned int i;
+
+	if (pre) {
+		tmp = pre(qdisc, data);
+		if (tmp)
+			return tmp;
+	}
+
+	if (qdisc->ops) {
+		for (i = 0; i < qdisc->num_classes; i++) {
+			tmp = &qdisc->qdiscs[i];
+			if (qdisc->ops) {
+				tmp = mlxsw_sp_qdisc_walk(tmp, pre, data);
+				if (tmp)
+					return tmp;
+			}
+		}
+	}
+
+	return NULL;
+}
+
+static struct mlxsw_sp_qdisc *
+mlxsw_sp_qdisc_walk_cb_find(struct mlxsw_sp_qdisc *qdisc, void *data)
+{
+	u32 parent = *(u32 *)data;
+
+	if (qdisc->ops && TC_H_MAJ(qdisc->handle) == TC_H_MAJ(parent)) {
+		if (qdisc->ops->find_class)
+			return qdisc->ops->find_class(qdisc, parent);
+	}
+
+	return NULL;
+}
+
+static struct mlxsw_sp_qdisc *
+mlxsw_sp_qdisc_find(struct mlxsw_sp_port *mlxsw_sp_port, u32 parent)
+{
+	struct mlxsw_sp_qdisc_state *qdisc_state = mlxsw_sp_port->qdisc;
+
+	if (!qdisc_state)
+		return NULL;
+	if (parent == TC_H_ROOT)
+		return &qdisc_state->root_qdisc;
+	return mlxsw_sp_qdisc_walk(&qdisc_state->root_qdisc,
+				   mlxsw_sp_qdisc_walk_cb_find, &parent);
+}
+
+static struct mlxsw_sp_qdisc *
+mlxsw_sp_qdisc_walk_cb_find_by_handle(struct mlxsw_sp_qdisc *qdisc, void *data)
+{
+	u32 handle = *(u32 *)data;
+
+	if (qdisc->ops && qdisc->handle == handle)
+		return qdisc;
+	return NULL;
+}
+
+static struct mlxsw_sp_qdisc *
+mlxsw_sp_qdisc_find_by_handle(struct mlxsw_sp_port *mlxsw_sp_port, u32 handle)
+{
+	struct mlxsw_sp_qdisc_state *qdisc_state = mlxsw_sp_port->qdisc;
+
+	if (!qdisc_state)
+		return NULL;
+	return mlxsw_sp_qdisc_walk(&qdisc_state->root_qdisc,
+				   mlxsw_sp_qdisc_walk_cb_find_by_handle,
+				   &handle);
+}
+
+static void
+mlxsw_sp_qdisc_reduce_parent_backlog(struct mlxsw_sp_qdisc *mlxsw_sp_qdisc)
+{
+	struct mlxsw_sp_qdisc *tmp;
+
+	for (tmp = mlxsw_sp_qdisc->parent; tmp; tmp = tmp->parent)
+		tmp->stats_base.backlog -= mlxsw_sp_qdisc->stats_base.backlog;
+}
+
+static u8 mlxsw_sp_qdisc_get_prio_bitmap(struct mlxsw_sp_port *mlxsw_sp_port,
+					 struct mlxsw_sp_qdisc *mlxsw_sp_qdisc)
+{
+	struct mlxsw_sp_qdisc *parent = mlxsw_sp_qdisc->parent;
+
+	if (!parent)
+		return 0xff;
+	if (!parent->ops->get_prio_bitmap)
+		return mlxsw_sp_qdisc_get_prio_bitmap(mlxsw_sp_port, parent);
+	return parent->ops->get_prio_bitmap(parent, mlxsw_sp_qdisc);
+}
+
+#define MLXSW_SP_PORT_DEFAULT_TCLASS 0
+
+static int mlxsw_sp_qdisc_get_tclass_num(struct mlxsw_sp_port *mlxsw_sp_port,
+					 struct mlxsw_sp_qdisc *mlxsw_sp_qdisc)
+{
+	struct mlxsw_sp_qdisc *parent = mlxsw_sp_qdisc->parent;
+
+	if (!parent)
+		return MLXSW_SP_PORT_DEFAULT_TCLASS;
+	if (!parent->ops->get_tclass_num)
+		return mlxsw_sp_qdisc_get_tclass_num(mlxsw_sp_port, parent);
+	return parent->ops->get_tclass_num(parent, mlxsw_sp_qdisc);
+}
+
+static int
+mlxsw_sp_qdisc_destroy(struct mlxsw_sp_port *mlxsw_sp_port,
+		       struct mlxsw_sp_qdisc *mlxsw_sp_qdisc)
+{
+	struct mlxsw_sp_qdisc *root_qdisc = &mlxsw_sp_port->qdisc->root_qdisc;
+	int err_hdroom = 0;
+	int err = 0;
+	int i;
+
+	if (!mlxsw_sp_qdisc)
+		return 0;
+
+	if (root_qdisc == mlxsw_sp_qdisc) {
+		struct mlxsw_sp_hdroom hdroom = *mlxsw_sp_port->hdroom;
+
+		hdroom.mode = MLXSW_SP_HDROOM_MODE_DCB;
+		mlxsw_sp_hdroom_prios_reset_buf_idx(&hdroom);
+		mlxsw_sp_hdroom_bufs_reset_lossiness(&hdroom);
+		mlxsw_sp_hdroom_bufs_reset_sizes(mlxsw_sp_port, &hdroom);
+		err_hdroom = mlxsw_sp_hdroom_configure(mlxsw_sp_port, &hdroom);
+	}
+
+	if (!mlxsw_sp_qdisc->ops)
+		return 0;
+
+	for (i = 0; i < mlxsw_sp_qdisc->num_classes; i++)
+		mlxsw_sp_qdisc_destroy(mlxsw_sp_port,
+				       &mlxsw_sp_qdisc->qdiscs[i]);
+	mlxsw_sp_qdisc_reduce_parent_backlog(mlxsw_sp_qdisc);
+	if (mlxsw_sp_qdisc->ops->destroy)
+		err = mlxsw_sp_qdisc->ops->destroy(mlxsw_sp_port,
+						   mlxsw_sp_qdisc);
+	if (mlxsw_sp_qdisc->ops->clean_stats)
+		mlxsw_sp_qdisc->ops->clean_stats(mlxsw_sp_port, mlxsw_sp_qdisc);
+
+	mlxsw_sp_qdisc->handle = TC_H_UNSPEC;
+	mlxsw_sp_qdisc->ops = NULL;
+	mlxsw_sp_qdisc->num_classes = 0;
+	kfree(mlxsw_sp_qdisc->qdiscs);
+	mlxsw_sp_qdisc->qdiscs = NULL;
+	return err_hdroom ?: err;
+}
+
+struct mlxsw_sp_qdisc_tree_validate {
+	bool forbid_ets;
+	bool forbid_root_tbf;
+	bool forbid_tbf;
+	bool forbid_red;
+};
+
+static int
+__mlxsw_sp_qdisc_tree_validate(struct mlxsw_sp_qdisc *mlxsw_sp_qdisc,
+			       struct mlxsw_sp_qdisc_tree_validate validate);
+
+static int
+mlxsw_sp_qdisc_tree_validate_children(struct mlxsw_sp_qdisc *mlxsw_sp_qdisc,
+				      struct mlxsw_sp_qdisc_tree_validate validate)
+{
+	unsigned int i;
+	int err;
+
+	for (i = 0; i < mlxsw_sp_qdisc->num_classes; i++) {
+		err = __mlxsw_sp_qdisc_tree_validate(&mlxsw_sp_qdisc->qdiscs[i],
+						     validate);
+		if (err)
+			return err;
+	}
+
+	return 0;
+}
+
+static int
+__mlxsw_sp_qdisc_tree_validate(struct mlxsw_sp_qdisc *mlxsw_sp_qdisc,
+			       struct mlxsw_sp_qdisc_tree_validate validate)
+{
+	if (!mlxsw_sp_qdisc->ops)
+		return 0;
+
+	switch (mlxsw_sp_qdisc->ops->type) {
+	case MLXSW_SP_QDISC_FIFO:
+		break;
+	case MLXSW_SP_QDISC_RED:
+		if (validate.forbid_red)
+			return -EINVAL;
+		validate.forbid_red = true;
+		validate.forbid_root_tbf = true;
+		validate.forbid_ets = true;
+		break;
+	case MLXSW_SP_QDISC_TBF:
+		if (validate.forbid_root_tbf) {
+			if (validate.forbid_tbf)
+				return -EINVAL;
+			/* This is a TC TBF. */
+			validate.forbid_tbf = true;
+			validate.forbid_ets = true;
+		} else {
+			/* This is root TBF. */
+			validate.forbid_root_tbf = true;
+		}
+		break;
+	case MLXSW_SP_QDISC_PRIO:
+	case MLXSW_SP_QDISC_ETS:
+		if (validate.forbid_ets)
+			return -EINVAL;
+		validate.forbid_root_tbf = true;
+		validate.forbid_ets = true;
+		break;
+	default:
+		WARN_ON(1);
+		return -EINVAL;
+	}
+
+	return mlxsw_sp_qdisc_tree_validate_children(mlxsw_sp_qdisc, validate);
+}
+
+static int mlxsw_sp_qdisc_tree_validate(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	struct mlxsw_sp_qdisc_tree_validate validate = {};
+	struct mlxsw_sp_qdisc *mlxsw_sp_qdisc;
+
+	mlxsw_sp_qdisc = &mlxsw_sp_port->qdisc->root_qdisc;
+	return __mlxsw_sp_qdisc_tree_validate(mlxsw_sp_qdisc, validate);
+}
+
+static int mlxsw_sp_qdisc_create(struct mlxsw_sp_port *mlxsw_sp_port,
+				 u32 handle,
+				 struct mlxsw_sp_qdisc *mlxsw_sp_qdisc,
+				 struct mlxsw_sp_qdisc_ops *ops, void *params)
+{
+	struct mlxsw_sp_qdisc *root_qdisc = &mlxsw_sp_port->qdisc->root_qdisc;
+	struct mlxsw_sp_hdroom orig_hdroom;
+	unsigned int i;
+	int err;
+
+	err = ops->check_params(mlxsw_sp_port, params);
+	if (err)
+		return err;
+
+	if (ops->num_classes) {
+		mlxsw_sp_qdisc->qdiscs = kcalloc(ops->num_classes,
+						 sizeof(*mlxsw_sp_qdisc->qdiscs),
+						 GFP_KERNEL);
+		if (!mlxsw_sp_qdisc->qdiscs)
+			return -ENOMEM;
+
+		for (i = 0; i < ops->num_classes; i++)
+			mlxsw_sp_qdisc->qdiscs[i].parent = mlxsw_sp_qdisc;
+	}
+
+	orig_hdroom = *mlxsw_sp_port->hdroom;
+	if (root_qdisc == mlxsw_sp_qdisc) {
+		struct mlxsw_sp_hdroom hdroom = orig_hdroom;
+
+		hdroom.mode = MLXSW_SP_HDROOM_MODE_TC;
+		mlxsw_sp_hdroom_prios_reset_buf_idx(&hdroom);
+		mlxsw_sp_hdroom_bufs_reset_lossiness(&hdroom);
+		mlxsw_sp_hdroom_bufs_reset_sizes(mlxsw_sp_port, &hdroom);
+
+		err = mlxsw_sp_hdroom_configure(mlxsw_sp_port, &hdroom);
+		if (err)
+			goto err_hdroom_configure;
+	}
+
+	mlxsw_sp_qdisc->num_classes = ops->num_classes;
+	mlxsw_sp_qdisc->ops = ops;
+	mlxsw_sp_qdisc->handle = handle;
+	err = mlxsw_sp_qdisc_tree_validate(mlxsw_sp_port);
+	if (err)
+		goto err_replace;
+
+	err = ops->replace(mlxsw_sp_port, handle, mlxsw_sp_qdisc, params);
+	if (err)
+		goto err_replace;
+
+	return 0;
+
+err_replace:
+	mlxsw_sp_qdisc->handle = TC_H_UNSPEC;
+	mlxsw_sp_qdisc->ops = NULL;
+	mlxsw_sp_qdisc->num_classes = 0;
+	mlxsw_sp_hdroom_configure(mlxsw_sp_port, &orig_hdroom);
+err_hdroom_configure:
+	kfree(mlxsw_sp_qdisc->qdiscs);
+	mlxsw_sp_qdisc->qdiscs = NULL;
+	return err;
+}
+
+static int
+mlxsw_sp_qdisc_change(struct mlxsw_sp_port *mlxsw_sp_port, u32 handle,
+		      struct mlxsw_sp_qdisc *mlxsw_sp_qdisc, void *params)
+{
+	struct mlxsw_sp_qdisc_ops *ops = mlxsw_sp_qdisc->ops;
+	int err;
+
+	err = ops->check_params(mlxsw_sp_port, params);
+	if (err)
+		goto unoffload;
+
+	err = ops->replace(mlxsw_sp_port, handle, mlxsw_sp_qdisc, params);
+	if (err)
+		goto unoffload;
+
+	/* Check if the Qdisc changed. That includes a situation where an
+	 * invisible Qdisc replaces another one, or is being added for the
+	 * first time.
+	 */
+	if (mlxsw_sp_qdisc->handle != handle) {
+		if (ops->clean_stats)
+			ops->clean_stats(mlxsw_sp_port, mlxsw_sp_qdisc);
+	}
+
+	mlxsw_sp_qdisc->handle = handle;
+	return 0;
+
+unoffload:
+	if (ops->unoffload)
+		ops->unoffload(mlxsw_sp_port, mlxsw_sp_qdisc, params);
+
+	mlxsw_sp_qdisc_destroy(mlxsw_sp_port, mlxsw_sp_qdisc);
+	return err;
+}
+
+static int
+mlxsw_sp_qdisc_replace(struct mlxsw_sp_port *mlxsw_sp_port, u32 handle,
+		       struct mlxsw_sp_qdisc *mlxsw_sp_qdisc,
+		       struct mlxsw_sp_qdisc_ops *ops, void *params)
+{
+	if (mlxsw_sp_qdisc->ops && mlxsw_sp_qdisc->ops->type != ops->type)
+		/* In case this location contained a different qdisc of the
+		 * same type we can override the old qdisc configuration.
+		 * Otherwise, we need to remove the old qdisc before setting the
+		 * new one.
+		 */
+		mlxsw_sp_qdisc_destroy(mlxsw_sp_port, mlxsw_sp_qdisc);
+
+	if (!mlxsw_sp_qdisc->ops)
+		return mlxsw_sp_qdisc_create(mlxsw_sp_port, handle,
+					     mlxsw_sp_qdisc, ops, params);
+	else
+		return mlxsw_sp_qdisc_change(mlxsw_sp_port, handle,
+					     mlxsw_sp_qdisc, params);
+}
+
+static int
+mlxsw_sp_qdisc_get_stats(struct mlxsw_sp_port *mlxsw_sp_port,
+			 struct mlxsw_sp_qdisc *mlxsw_sp_qdisc,
+			 struct tc_qopt_offload_stats *stats_ptr)
+{
+	if (mlxsw_sp_qdisc && mlxsw_sp_qdisc->ops &&
+	    mlxsw_sp_qdisc->ops->get_stats)
+		return mlxsw_sp_qdisc->ops->get_stats(mlxsw_sp_port,
+						      mlxsw_sp_qdisc,
+						      stats_ptr);
+
+	return -EOPNOTSUPP;
+}
+
+static int
+mlxsw_sp_qdisc_get_xstats(struct mlxsw_sp_port *mlxsw_sp_port,
+			  struct mlxsw_sp_qdisc *mlxsw_sp_qdisc,
+			  void *xstats_ptr)
+{
+	if (mlxsw_sp_qdisc && mlxsw_sp_qdisc->ops &&
+	    mlxsw_sp_qdisc->ops->get_xstats)
+		return mlxsw_sp_qdisc->ops->get_xstats(mlxsw_sp_port,
+						      mlxsw_sp_qdisc,
+						      xstats_ptr);
+
+	return -EOPNOTSUPP;
+}
+
+static u64
+mlxsw_sp_xstats_backlog(struct mlxsw_sp_port_xstats *xstats, int tclass_num)
+{
+	return xstats->backlog[tclass_num] +
+	       xstats->backlog[tclass_num + 8];
+}
+
+static u64
+mlxsw_sp_xstats_tail_drop(struct mlxsw_sp_port_xstats *xstats, int tclass_num)
+{
+	return xstats->tail_drop[tclass_num] +
+	       xstats->tail_drop[tclass_num + 8];
+}
+
+static void
+mlxsw_sp_qdisc_bstats_per_priority_get(struct mlxsw_sp_port_xstats *xstats,
+				       u8 prio_bitmap, u64 *tx_packets,
+				       u64 *tx_bytes)
+{
+	int i;
+
+	*tx_packets = 0;
+	*tx_bytes = 0;
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) {
+		if (prio_bitmap & BIT(i)) {
+			*tx_packets += xstats->tx_packets[i];
+			*tx_bytes += xstats->tx_bytes[i];
+		}
+	}
+}
+
+static void
+mlxsw_sp_qdisc_collect_tc_stats(struct mlxsw_sp_port *mlxsw_sp_port,
+				struct mlxsw_sp_qdisc *mlxsw_sp_qdisc,
+				u64 *p_tx_bytes, u64 *p_tx_packets,
+				u64 *p_drops, u64 *p_backlog)
+{
+	struct mlxsw_sp_port_xstats *xstats;
+	u64 tx_bytes, tx_packets;
+	u8 prio_bitmap;
+	int tclass_num;
+
+	prio_bitmap = mlxsw_sp_qdisc_get_prio_bitmap(mlxsw_sp_port,
+						     mlxsw_sp_qdisc);
+	tclass_num = mlxsw_sp_qdisc_get_tclass_num(mlxsw_sp_port,
+						   mlxsw_sp_qdisc);
+	xstats = &mlxsw_sp_port->periodic_hw_stats.xstats;
+	mlxsw_sp_qdisc_bstats_per_priority_get(xstats, prio_bitmap,
+					       &tx_packets, &tx_bytes);
+
+	*p_tx_packets += tx_packets;
+	*p_tx_bytes += tx_bytes;
+	*p_drops += xstats->wred_drop[tclass_num] +
+		    mlxsw_sp_xstats_tail_drop(xstats, tclass_num);
+	*p_backlog += mlxsw_sp_xstats_backlog(xstats, tclass_num);
+}
+
+static void
+mlxsw_sp_qdisc_update_stats(struct mlxsw_sp *mlxsw_sp,
+			    struct mlxsw_sp_qdisc *mlxsw_sp_qdisc,
+			    u64 tx_bytes, u64 tx_packets,
+			    u64 drops, u64 backlog,
+			    struct tc_qopt_offload_stats *stats_ptr)
+{
+	struct mlxsw_sp_qdisc_stats *stats_base = &mlxsw_sp_qdisc->stats_base;
+
+	tx_bytes -= stats_base->tx_bytes;
+	tx_packets -= stats_base->tx_packets;
+	drops -= stats_base->drops;
+	backlog -= stats_base->backlog;
+
+	_bstats_update(stats_ptr->bstats, tx_bytes, tx_packets);
+	stats_ptr->qstats->drops += drops;
+	stats_ptr->qstats->backlog += mlxsw_sp_cells_bytes(mlxsw_sp, backlog);
+
+	stats_base->backlog += backlog;
+	stats_base->drops += drops;
+	stats_base->tx_bytes += tx_bytes;
+	stats_base->tx_packets += tx_packets;
+}
+
+static void
+mlxsw_sp_qdisc_get_tc_stats(struct mlxsw_sp_port *mlxsw_sp_port,
+			    struct mlxsw_sp_qdisc *mlxsw_sp_qdisc,
+			    struct tc_qopt_offload_stats *stats_ptr)
+{
+	u64 tx_packets = 0;
+	u64 tx_bytes = 0;
+	u64 backlog = 0;
+	u64 drops = 0;
+
+	mlxsw_sp_qdisc_collect_tc_stats(mlxsw_sp_port, mlxsw_sp_qdisc,
+					&tx_bytes, &tx_packets,
+					&drops, &backlog);
+	mlxsw_sp_qdisc_update_stats(mlxsw_sp_port->mlxsw_sp, mlxsw_sp_qdisc,
+				    tx_bytes, tx_packets, drops, backlog,
+				    stats_ptr);
+}
+
+static int
+mlxsw_sp_tclass_congestion_enable(struct mlxsw_sp_port *mlxsw_sp_port,
+				  int tclass_num, u32 min, u32 max,
+				  u32 probability, bool is_wred, bool is_ecn)
+{
+	char cwtpm_cmd[MLXSW_REG_CWTPM_LEN];
+	char cwtp_cmd[MLXSW_REG_CWTP_LEN];
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	int err;
+
+	mlxsw_reg_cwtp_pack(cwtp_cmd, mlxsw_sp_port->local_port, tclass_num);
+	mlxsw_reg_cwtp_profile_pack(cwtp_cmd, MLXSW_REG_CWTP_DEFAULT_PROFILE,
+				    roundup(min, MLXSW_REG_CWTP_MIN_VALUE),
+				    roundup(max, MLXSW_REG_CWTP_MIN_VALUE),
+				    probability);
+
+	err = mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(cwtp), cwtp_cmd);
+	if (err)
+		return err;
+
+	mlxsw_reg_cwtpm_pack(cwtpm_cmd, mlxsw_sp_port->local_port, tclass_num,
+			     MLXSW_REG_CWTP_DEFAULT_PROFILE, is_wred, is_ecn);
+
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(cwtpm), cwtpm_cmd);
+}
+
+static int
+mlxsw_sp_tclass_congestion_disable(struct mlxsw_sp_port *mlxsw_sp_port,
+				   int tclass_num)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	char cwtpm_cmd[MLXSW_REG_CWTPM_LEN];
+
+	mlxsw_reg_cwtpm_pack(cwtpm_cmd, mlxsw_sp_port->local_port, tclass_num,
+			     MLXSW_REG_CWTPM_RESET_PROFILE, false, false);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(cwtpm), cwtpm_cmd);
+}
+
+static void
+mlxsw_sp_setup_tc_qdisc_red_clean_stats(struct mlxsw_sp_port *mlxsw_sp_port,
+					struct mlxsw_sp_qdisc *mlxsw_sp_qdisc)
+{
+	struct mlxsw_sp_qdisc_stats *stats_base;
+	struct mlxsw_sp_port_xstats *xstats;
+	struct red_stats *red_base;
+	u8 prio_bitmap;
+	int tclass_num;
+
+	prio_bitmap = mlxsw_sp_qdisc_get_prio_bitmap(mlxsw_sp_port,
+						     mlxsw_sp_qdisc);
+	tclass_num = mlxsw_sp_qdisc_get_tclass_num(mlxsw_sp_port,
+						   mlxsw_sp_qdisc);
+	xstats = &mlxsw_sp_port->periodic_hw_stats.xstats;
+	stats_base = &mlxsw_sp_qdisc->stats_base;
+	red_base = &mlxsw_sp_qdisc->xstats_base.red;
+
+	mlxsw_sp_qdisc_bstats_per_priority_get(xstats, prio_bitmap,
+					       &stats_base->tx_packets,
+					       &stats_base->tx_bytes);
+	red_base->prob_mark = xstats->tc_ecn[tclass_num];
+	red_base->prob_drop = xstats->wred_drop[tclass_num];
+	red_base->pdrop = mlxsw_sp_xstats_tail_drop(xstats, tclass_num);
+
+	stats_base->overlimits = red_base->prob_drop + red_base->prob_mark;
+	stats_base->drops = red_base->prob_drop + red_base->pdrop;
+
+	stats_base->backlog = 0;
+}
+
+static int
+mlxsw_sp_qdisc_red_destroy(struct mlxsw_sp_port *mlxsw_sp_port,
+			   struct mlxsw_sp_qdisc *mlxsw_sp_qdisc)
+{
+	int tclass_num = mlxsw_sp_qdisc_get_tclass_num(mlxsw_sp_port,
+						       mlxsw_sp_qdisc);
+
+	return mlxsw_sp_tclass_congestion_disable(mlxsw_sp_port, tclass_num);
+}
+
+static int
+mlxsw_sp_qdisc_red_check_params(struct mlxsw_sp_port *mlxsw_sp_port,
+				void *params)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	struct tc_red_qopt_offload_params *p = params;
+
+	if (p->min > p->max) {
+		dev_err(mlxsw_sp->bus_info->dev,
+			"spectrum: RED: min %u is bigger then max %u\n", p->min,
+			p->max);
+		return -EINVAL;
+	}
+	if (p->max > MLXSW_CORE_RES_GET(mlxsw_sp->core,
+					GUARANTEED_SHARED_BUFFER)) {
+		dev_err(mlxsw_sp->bus_info->dev,
+			"spectrum: RED: max value %u is too big\n", p->max);
+		return -EINVAL;
+	}
+	if (p->min == 0 || p->max == 0) {
+		dev_err(mlxsw_sp->bus_info->dev,
+			"spectrum: RED: 0 value is illegal for min and max\n");
+		return -EINVAL;
+	}
+	return 0;
+}
+
+static int
+mlxsw_sp_qdisc_future_fifo_replace(struct mlxsw_sp_port *mlxsw_sp_port,
+				   u32 handle, unsigned int band,
+				   struct mlxsw_sp_qdisc *child_qdisc);
+static void
+mlxsw_sp_qdisc_future_fifos_init(struct mlxsw_sp_port *mlxsw_sp_port,
+				 u32 handle);
+
+static int
+mlxsw_sp_qdisc_red_replace(struct mlxsw_sp_port *mlxsw_sp_port, u32 handle,
+			   struct mlxsw_sp_qdisc *mlxsw_sp_qdisc,
+			   void *params)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	struct tc_red_qopt_offload_params *p = params;
+	int tclass_num;
+	u32 min, max;
+	u64 prob;
+	int err;
+
+	err = mlxsw_sp_qdisc_future_fifo_replace(mlxsw_sp_port, handle, 0,
+						 &mlxsw_sp_qdisc->qdiscs[0]);
+	if (err)
+		return err;
+	mlxsw_sp_qdisc_future_fifos_init(mlxsw_sp_port, TC_H_UNSPEC);
+
+	tclass_num = mlxsw_sp_qdisc_get_tclass_num(mlxsw_sp_port,
+						   mlxsw_sp_qdisc);
+
+	/* calculate probability in percentage */
+	prob = p->probability;
+	prob *= 100;
+	prob = DIV_ROUND_UP(prob, 1 << 16);
+	prob = DIV_ROUND_UP(prob, 1 << 16);
+	min = mlxsw_sp_bytes_cells(mlxsw_sp, p->min);
+	max = mlxsw_sp_bytes_cells(mlxsw_sp, p->max);
+	return mlxsw_sp_tclass_congestion_enable(mlxsw_sp_port, tclass_num,
+						 min, max, prob,
+						 !p->is_nodrop, p->is_ecn);
+}
+
+static void
+mlxsw_sp_qdisc_leaf_unoffload(struct mlxsw_sp_port *mlxsw_sp_port,
+			      struct mlxsw_sp_qdisc *mlxsw_sp_qdisc,
+			      struct gnet_stats_queue *qstats)
+{
+	u64 backlog;
+
+	backlog = mlxsw_sp_cells_bytes(mlxsw_sp_port->mlxsw_sp,
+				       mlxsw_sp_qdisc->stats_base.backlog);
+	qstats->backlog -= backlog;
+	mlxsw_sp_qdisc->stats_base.backlog = 0;
+}
+
+static void
+mlxsw_sp_qdisc_red_unoffload(struct mlxsw_sp_port *mlxsw_sp_port,
+			     struct mlxsw_sp_qdisc *mlxsw_sp_qdisc,
+			     void *params)
+{
+	struct tc_red_qopt_offload_params *p = params;
+
+	mlxsw_sp_qdisc_leaf_unoffload(mlxsw_sp_port, mlxsw_sp_qdisc, p->qstats);
+}
+
+static int
+mlxsw_sp_qdisc_get_red_xstats(struct mlxsw_sp_port *mlxsw_sp_port,
+			      struct mlxsw_sp_qdisc *mlxsw_sp_qdisc,
+			      void *xstats_ptr)
+{
+	struct red_stats *xstats_base = &mlxsw_sp_qdisc->xstats_base.red;
+	struct mlxsw_sp_port_xstats *xstats;
+	struct red_stats *res = xstats_ptr;
+	int early_drops, marks, pdrops;
+	int tclass_num;
+
+	tclass_num = mlxsw_sp_qdisc_get_tclass_num(mlxsw_sp_port,
+						   mlxsw_sp_qdisc);
+	xstats = &mlxsw_sp_port->periodic_hw_stats.xstats;
+
+	early_drops = xstats->wred_drop[tclass_num] - xstats_base->prob_drop;
+	marks = xstats->tc_ecn[tclass_num] - xstats_base->prob_mark;
+	pdrops = mlxsw_sp_xstats_tail_drop(xstats, tclass_num) -
+		 xstats_base->pdrop;
+
+	res->pdrop += pdrops;
+	res->prob_drop += early_drops;
+	res->prob_mark += marks;
+
+	xstats_base->pdrop += pdrops;
+	xstats_base->prob_drop += early_drops;
+	xstats_base->prob_mark += marks;
+	return 0;
+}
+
+static int
+mlxsw_sp_qdisc_get_red_stats(struct mlxsw_sp_port *mlxsw_sp_port,
+			     struct mlxsw_sp_qdisc *mlxsw_sp_qdisc,
+			     struct tc_qopt_offload_stats *stats_ptr)
+{
+	struct mlxsw_sp_qdisc_stats *stats_base;
+	struct mlxsw_sp_port_xstats *xstats;
+	u64 overlimits;
+	int tclass_num;
+
+	tclass_num = mlxsw_sp_qdisc_get_tclass_num(mlxsw_sp_port,
+						   mlxsw_sp_qdisc);
+	xstats = &mlxsw_sp_port->periodic_hw_stats.xstats;
+	stats_base = &mlxsw_sp_qdisc->stats_base;
+
+	mlxsw_sp_qdisc_get_tc_stats(mlxsw_sp_port, mlxsw_sp_qdisc, stats_ptr);
+	overlimits = xstats->wred_drop[tclass_num] +
+		     xstats->tc_ecn[tclass_num] - stats_base->overlimits;
+
+	stats_ptr->qstats->overlimits += overlimits;
+	stats_base->overlimits += overlimits;
+
+	return 0;
+}
+
+static struct mlxsw_sp_qdisc *
+mlxsw_sp_qdisc_leaf_find_class(struct mlxsw_sp_qdisc *mlxsw_sp_qdisc,
+			       u32 parent)
+{
+	/* RED and TBF are formally classful qdiscs, but all class references,
+	 * including X:0, just refer to the same one class.
+	 */
+	return &mlxsw_sp_qdisc->qdiscs[0];
+}
+
+static struct mlxsw_sp_qdisc_ops mlxsw_sp_qdisc_ops_red = {
+	.type = MLXSW_SP_QDISC_RED,
+	.check_params = mlxsw_sp_qdisc_red_check_params,
+	.replace = mlxsw_sp_qdisc_red_replace,
+	.unoffload = mlxsw_sp_qdisc_red_unoffload,
+	.destroy = mlxsw_sp_qdisc_red_destroy,
+	.get_stats = mlxsw_sp_qdisc_get_red_stats,
+	.get_xstats = mlxsw_sp_qdisc_get_red_xstats,
+	.clean_stats = mlxsw_sp_setup_tc_qdisc_red_clean_stats,
+	.find_class = mlxsw_sp_qdisc_leaf_find_class,
+	.num_classes = 1,
+};
+
+static int mlxsw_sp_qdisc_graft(struct mlxsw_sp_port *mlxsw_sp_port,
+				struct mlxsw_sp_qdisc *mlxsw_sp_qdisc,
+				u8 band, u32 child_handle);
+
+static int __mlxsw_sp_setup_tc_red(struct mlxsw_sp_port *mlxsw_sp_port,
+				   struct tc_red_qopt_offload *p)
+{
+	struct mlxsw_sp_qdisc *mlxsw_sp_qdisc;
+
+	mlxsw_sp_qdisc = mlxsw_sp_qdisc_find(mlxsw_sp_port, p->parent);
+	if (!mlxsw_sp_qdisc)
+		return -EOPNOTSUPP;
+
+	if (p->command == TC_RED_REPLACE)
+		return mlxsw_sp_qdisc_replace(mlxsw_sp_port, p->handle,
+					      mlxsw_sp_qdisc,
+					      &mlxsw_sp_qdisc_ops_red,
+					      &p->set);
+
+	if (!mlxsw_sp_qdisc_compare(mlxsw_sp_qdisc, p->handle))
+		return -EOPNOTSUPP;
+
+	switch (p->command) {
+	case TC_RED_DESTROY:
+		return mlxsw_sp_qdisc_destroy(mlxsw_sp_port, mlxsw_sp_qdisc);
+	case TC_RED_XSTATS:
+		return mlxsw_sp_qdisc_get_xstats(mlxsw_sp_port, mlxsw_sp_qdisc,
+						 p->xstats);
+	case TC_RED_STATS:
+		return mlxsw_sp_qdisc_get_stats(mlxsw_sp_port, mlxsw_sp_qdisc,
+						&p->stats);
+	case TC_RED_GRAFT:
+		return mlxsw_sp_qdisc_graft(mlxsw_sp_port, mlxsw_sp_qdisc, 0,
+					    p->child_handle);
+	default:
+		return -EOPNOTSUPP;
+	}
+}
+
+int mlxsw_sp_setup_tc_red(struct mlxsw_sp_port *mlxsw_sp_port,
+			  struct tc_red_qopt_offload *p)
+{
+	int err;
+
+	mutex_lock(&mlxsw_sp_port->qdisc->lock);
+	err = __mlxsw_sp_setup_tc_red(mlxsw_sp_port, p);
+	mutex_unlock(&mlxsw_sp_port->qdisc->lock);
+
+	return err;
+}
+
+static void
+mlxsw_sp_setup_tc_qdisc_leaf_clean_stats(struct mlxsw_sp_port *mlxsw_sp_port,
+					 struct mlxsw_sp_qdisc *mlxsw_sp_qdisc)
+{
+	u64 backlog_cells = 0;
+	u64 tx_packets = 0;
+	u64 tx_bytes = 0;
+	u64 drops = 0;
+
+	mlxsw_sp_qdisc_collect_tc_stats(mlxsw_sp_port, mlxsw_sp_qdisc,
+					&tx_bytes, &tx_packets,
+					&drops, &backlog_cells);
+
+	mlxsw_sp_qdisc->stats_base.tx_packets = tx_packets;
+	mlxsw_sp_qdisc->stats_base.tx_bytes = tx_bytes;
+	mlxsw_sp_qdisc->stats_base.drops = drops;
+	mlxsw_sp_qdisc->stats_base.backlog = 0;
+}
+
+static enum mlxsw_reg_qeec_hr
+mlxsw_sp_qdisc_tbf_hr(struct mlxsw_sp_port *mlxsw_sp_port,
+		      struct mlxsw_sp_qdisc *mlxsw_sp_qdisc)
+{
+	if (mlxsw_sp_qdisc == &mlxsw_sp_port->qdisc->root_qdisc)
+		return MLXSW_REG_QEEC_HR_PORT;
+
+	/* Configure subgroup shaper, so that both UC and MC traffic is subject
+	 * to shaping. That is unlike RED, however UC queue lengths are going to
+	 * be different than MC ones due to different pool and quota
+	 * configurations, so the configuration is not applicable. For shaper on
+	 * the other hand, subjecting the overall stream to the configured
+	 * shaper makes sense. Also note that that is what we do for
+	 * ieee_setmaxrate().
+	 */
+	return MLXSW_REG_QEEC_HR_SUBGROUP;
+}
+
+static int
+mlxsw_sp_qdisc_tbf_destroy(struct mlxsw_sp_port *mlxsw_sp_port,
+			   struct mlxsw_sp_qdisc *mlxsw_sp_qdisc)
+{
+	enum mlxsw_reg_qeec_hr hr = mlxsw_sp_qdisc_tbf_hr(mlxsw_sp_port,
+							  mlxsw_sp_qdisc);
+	int tclass_num = mlxsw_sp_qdisc_get_tclass_num(mlxsw_sp_port,
+						       mlxsw_sp_qdisc);
+
+	return mlxsw_sp_port_ets_maxrate_set(mlxsw_sp_port, hr, tclass_num, 0,
+					     MLXSW_REG_QEEC_MAS_DIS, 0);
+}
+
+static int
+mlxsw_sp_qdisc_tbf_bs(struct mlxsw_sp_port *mlxsw_sp_port,
+		      u32 max_size, u8 *p_burst_size)
+{
+	/* TBF burst size is configured in bytes. The ASIC burst size value is
+	 * ((2 ^ bs) * 512 bits. Convert the TBF bytes to 512-bit units.
+	 */
+	u32 bs512 = max_size / 64;
+	u8 bs = fls(bs512);
+
+	if (!bs)
+		return -EINVAL;
+	--bs;
+
+	/* Demand a power of two. */
+	if ((1 << bs) != bs512)
+		return -EINVAL;
+
+	if (bs < mlxsw_sp_port->mlxsw_sp->lowest_shaper_bs ||
+	    bs > MLXSW_REG_QEEC_HIGHEST_SHAPER_BS)
+		return -EINVAL;
+
+	*p_burst_size = bs;
+	return 0;
+}
+
+static u32
+mlxsw_sp_qdisc_tbf_max_size(u8 bs)
+{
+	return (1U << bs) * 64;
+}
+
+static u64
+mlxsw_sp_qdisc_tbf_rate_kbps(struct tc_tbf_qopt_offload_replace_params *p)
+{
+	/* TBF interface is in bytes/s, whereas Spectrum ASIC is configured in
+	 * Kbits/s.
+	 */
+	return div_u64(p->rate.rate_bytes_ps, 1000) * 8;
+}
+
+static int
+mlxsw_sp_qdisc_tbf_check_params(struct mlxsw_sp_port *mlxsw_sp_port,
+				void *params)
+{
+	struct tc_tbf_qopt_offload_replace_params *p = params;
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	u64 rate_kbps = mlxsw_sp_qdisc_tbf_rate_kbps(p);
+	u8 burst_size;
+	int err;
+
+	if (rate_kbps >= MLXSW_REG_QEEC_MAS_DIS) {
+		dev_err(mlxsw_sp_port->mlxsw_sp->bus_info->dev,
+			"spectrum: TBF: rate of %lluKbps must be below %u\n",
+			rate_kbps, MLXSW_REG_QEEC_MAS_DIS);
+		return -EINVAL;
+	}
+
+	err = mlxsw_sp_qdisc_tbf_bs(mlxsw_sp_port, p->max_size, &burst_size);
+	if (err) {
+		u8 highest_shaper_bs = MLXSW_REG_QEEC_HIGHEST_SHAPER_BS;
+
+		dev_err(mlxsw_sp->bus_info->dev,
+			"spectrum: TBF: invalid burst size of %u, must be a power of two between %u and %u",
+			p->max_size,
+			mlxsw_sp_qdisc_tbf_max_size(mlxsw_sp->lowest_shaper_bs),
+			mlxsw_sp_qdisc_tbf_max_size(highest_shaper_bs));
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int
+mlxsw_sp_qdisc_tbf_replace(struct mlxsw_sp_port *mlxsw_sp_port, u32 handle,
+			   struct mlxsw_sp_qdisc *mlxsw_sp_qdisc,
+			   void *params)
+{
+	enum mlxsw_reg_qeec_hr hr = mlxsw_sp_qdisc_tbf_hr(mlxsw_sp_port,
+							  mlxsw_sp_qdisc);
+	struct tc_tbf_qopt_offload_replace_params *p = params;
+	u64 rate_kbps = mlxsw_sp_qdisc_tbf_rate_kbps(p);
+	int tclass_num;
+	u8 burst_size;
+	int err;
+
+	err = mlxsw_sp_qdisc_future_fifo_replace(mlxsw_sp_port, handle, 0,
+						 &mlxsw_sp_qdisc->qdiscs[0]);
+	if (err)
+		return err;
+	mlxsw_sp_qdisc_future_fifos_init(mlxsw_sp_port, TC_H_UNSPEC);
+
+	tclass_num = mlxsw_sp_qdisc_get_tclass_num(mlxsw_sp_port,
+						   mlxsw_sp_qdisc);
+
+	err = mlxsw_sp_qdisc_tbf_bs(mlxsw_sp_port, p->max_size, &burst_size);
+	if (WARN_ON_ONCE(err))
+		/* check_params above was supposed to reject this value. */
+		return -EINVAL;
+
+	return mlxsw_sp_port_ets_maxrate_set(mlxsw_sp_port, hr, tclass_num, 0,
+					     rate_kbps, burst_size);
+}
+
+static void
+mlxsw_sp_qdisc_tbf_unoffload(struct mlxsw_sp_port *mlxsw_sp_port,
+			     struct mlxsw_sp_qdisc *mlxsw_sp_qdisc,
+			     void *params)
+{
+	struct tc_tbf_qopt_offload_replace_params *p = params;
+
+	mlxsw_sp_qdisc_leaf_unoffload(mlxsw_sp_port, mlxsw_sp_qdisc, p->qstats);
+}
+
+static int
+mlxsw_sp_qdisc_get_tbf_stats(struct mlxsw_sp_port *mlxsw_sp_port,
+			     struct mlxsw_sp_qdisc *mlxsw_sp_qdisc,
+			     struct tc_qopt_offload_stats *stats_ptr)
+{
+	mlxsw_sp_qdisc_get_tc_stats(mlxsw_sp_port, mlxsw_sp_qdisc,
+				    stats_ptr);
+	return 0;
+}
+
+static struct mlxsw_sp_qdisc_ops mlxsw_sp_qdisc_ops_tbf = {
+	.type = MLXSW_SP_QDISC_TBF,
+	.check_params = mlxsw_sp_qdisc_tbf_check_params,
+	.replace = mlxsw_sp_qdisc_tbf_replace,
+	.unoffload = mlxsw_sp_qdisc_tbf_unoffload,
+	.destroy = mlxsw_sp_qdisc_tbf_destroy,
+	.get_stats = mlxsw_sp_qdisc_get_tbf_stats,
+	.clean_stats = mlxsw_sp_setup_tc_qdisc_leaf_clean_stats,
+	.find_class = mlxsw_sp_qdisc_leaf_find_class,
+	.num_classes = 1,
+};
+
+static int __mlxsw_sp_setup_tc_tbf(struct mlxsw_sp_port *mlxsw_sp_port,
+				   struct tc_tbf_qopt_offload *p)
+{
+	struct mlxsw_sp_qdisc *mlxsw_sp_qdisc;
+
+	mlxsw_sp_qdisc = mlxsw_sp_qdisc_find(mlxsw_sp_port, p->parent);
+	if (!mlxsw_sp_qdisc)
+		return -EOPNOTSUPP;
+
+	if (p->command == TC_TBF_REPLACE)
+		return mlxsw_sp_qdisc_replace(mlxsw_sp_port, p->handle,
+					      mlxsw_sp_qdisc,
+					      &mlxsw_sp_qdisc_ops_tbf,
+					      &p->replace_params);
+
+	if (!mlxsw_sp_qdisc_compare(mlxsw_sp_qdisc, p->handle))
+		return -EOPNOTSUPP;
+
+	switch (p->command) {
+	case TC_TBF_DESTROY:
+		return mlxsw_sp_qdisc_destroy(mlxsw_sp_port, mlxsw_sp_qdisc);
+	case TC_TBF_STATS:
+		return mlxsw_sp_qdisc_get_stats(mlxsw_sp_port, mlxsw_sp_qdisc,
+						&p->stats);
+	case TC_TBF_GRAFT:
+		return mlxsw_sp_qdisc_graft(mlxsw_sp_port, mlxsw_sp_qdisc, 0,
+					    p->child_handle);
+	default:
+		return -EOPNOTSUPP;
+	}
+}
+
+int mlxsw_sp_setup_tc_tbf(struct mlxsw_sp_port *mlxsw_sp_port,
+			  struct tc_tbf_qopt_offload *p)
+{
+	int err;
+
+	mutex_lock(&mlxsw_sp_port->qdisc->lock);
+	err = __mlxsw_sp_setup_tc_tbf(mlxsw_sp_port, p);
+	mutex_unlock(&mlxsw_sp_port->qdisc->lock);
+
+	return err;
+}
+
+static int
+mlxsw_sp_qdisc_fifo_check_params(struct mlxsw_sp_port *mlxsw_sp_port,
+				 void *params)
+{
+	return 0;
+}
+
+static int
+mlxsw_sp_qdisc_fifo_replace(struct mlxsw_sp_port *mlxsw_sp_port, u32 handle,
+			    struct mlxsw_sp_qdisc *mlxsw_sp_qdisc,
+			    void *params)
+{
+	return 0;
+}
+
+static int
+mlxsw_sp_qdisc_get_fifo_stats(struct mlxsw_sp_port *mlxsw_sp_port,
+			      struct mlxsw_sp_qdisc *mlxsw_sp_qdisc,
+			      struct tc_qopt_offload_stats *stats_ptr)
+{
+	mlxsw_sp_qdisc_get_tc_stats(mlxsw_sp_port, mlxsw_sp_qdisc,
+				    stats_ptr);
+	return 0;
+}
+
+static struct mlxsw_sp_qdisc_ops mlxsw_sp_qdisc_ops_fifo = {
+	.type = MLXSW_SP_QDISC_FIFO,
+	.check_params = mlxsw_sp_qdisc_fifo_check_params,
+	.replace = mlxsw_sp_qdisc_fifo_replace,
+	.get_stats = mlxsw_sp_qdisc_get_fifo_stats,
+	.clean_stats = mlxsw_sp_setup_tc_qdisc_leaf_clean_stats,
+};
+
+static int
+mlxsw_sp_qdisc_future_fifo_replace(struct mlxsw_sp_port *mlxsw_sp_port,
+				   u32 handle, unsigned int band,
+				   struct mlxsw_sp_qdisc *child_qdisc)
+{
+	struct mlxsw_sp_qdisc_state *qdisc_state = mlxsw_sp_port->qdisc;
+
+	if (handle == qdisc_state->future_handle &&
+	    qdisc_state->future_fifos[band])
+		return mlxsw_sp_qdisc_replace(mlxsw_sp_port, TC_H_UNSPEC,
+					      child_qdisc,
+					      &mlxsw_sp_qdisc_ops_fifo,
+					      NULL);
+	return 0;
+}
+
+static void
+mlxsw_sp_qdisc_future_fifos_init(struct mlxsw_sp_port *mlxsw_sp_port,
+				 u32 handle)
+{
+	struct mlxsw_sp_qdisc_state *qdisc_state = mlxsw_sp_port->qdisc;
+
+	qdisc_state->future_handle = handle;
+	memset(qdisc_state->future_fifos, 0, sizeof(qdisc_state->future_fifos));
+}
+
+static int __mlxsw_sp_setup_tc_fifo(struct mlxsw_sp_port *mlxsw_sp_port,
+				    struct tc_fifo_qopt_offload *p)
+{
+	struct mlxsw_sp_qdisc_state *qdisc_state = mlxsw_sp_port->qdisc;
+	struct mlxsw_sp_qdisc *mlxsw_sp_qdisc;
+	unsigned int band;
+	u32 parent_handle;
+
+	mlxsw_sp_qdisc = mlxsw_sp_qdisc_find(mlxsw_sp_port, p->parent);
+	if (!mlxsw_sp_qdisc && p->handle == TC_H_UNSPEC) {
+		parent_handle = TC_H_MAJ(p->parent);
+		if (parent_handle != qdisc_state->future_handle) {
+			/* This notifications is for a different Qdisc than
+			 * previously. Wipe the future cache.
+			 */
+			mlxsw_sp_qdisc_future_fifos_init(mlxsw_sp_port,
+							 parent_handle);
+		}
+
+		band = TC_H_MIN(p->parent) - 1;
+		if (band < IEEE_8021QAZ_MAX_TCS) {
+			if (p->command == TC_FIFO_REPLACE)
+				qdisc_state->future_fifos[band] = true;
+			else if (p->command == TC_FIFO_DESTROY)
+				qdisc_state->future_fifos[band] = false;
+		}
+	}
+	if (!mlxsw_sp_qdisc)
+		return -EOPNOTSUPP;
+
+	if (p->command == TC_FIFO_REPLACE) {
+		return mlxsw_sp_qdisc_replace(mlxsw_sp_port, p->handle,
+					      mlxsw_sp_qdisc,
+					      &mlxsw_sp_qdisc_ops_fifo, NULL);
+	}
+
+	if (!mlxsw_sp_qdisc_compare(mlxsw_sp_qdisc, p->handle))
+		return -EOPNOTSUPP;
+
+	switch (p->command) {
+	case TC_FIFO_DESTROY:
+		return mlxsw_sp_qdisc_destroy(mlxsw_sp_port, mlxsw_sp_qdisc);
+	case TC_FIFO_STATS:
+		return mlxsw_sp_qdisc_get_stats(mlxsw_sp_port, mlxsw_sp_qdisc,
+						&p->stats);
+	case TC_FIFO_REPLACE: /* Handled above. */
+		break;
+	}
+
+	return -EOPNOTSUPP;
+}
+
+int mlxsw_sp_setup_tc_fifo(struct mlxsw_sp_port *mlxsw_sp_port,
+			   struct tc_fifo_qopt_offload *p)
+{
+	int err;
+
+	mutex_lock(&mlxsw_sp_port->qdisc->lock);
+	err = __mlxsw_sp_setup_tc_fifo(mlxsw_sp_port, p);
+	mutex_unlock(&mlxsw_sp_port->qdisc->lock);
+
+	return err;
+}
+
+static int __mlxsw_sp_qdisc_ets_destroy(struct mlxsw_sp_port *mlxsw_sp_port,
+					struct mlxsw_sp_qdisc *mlxsw_sp_qdisc)
+{
+	int i;
+
+	for (i = 0; i < mlxsw_sp_qdisc->num_classes; i++) {
+		mlxsw_sp_port_prio_tc_set(mlxsw_sp_port, i,
+					  MLXSW_SP_PORT_DEFAULT_TCLASS);
+		mlxsw_sp_port_ets_set(mlxsw_sp_port,
+				      MLXSW_REG_QEEC_HR_SUBGROUP,
+				      i, 0, false, 0);
+	}
+
+	kfree(mlxsw_sp_qdisc->ets_data);
+	mlxsw_sp_qdisc->ets_data = NULL;
+	return 0;
+}
+
+static int
+mlxsw_sp_qdisc_prio_destroy(struct mlxsw_sp_port *mlxsw_sp_port,
+			    struct mlxsw_sp_qdisc *mlxsw_sp_qdisc)
+{
+	return __mlxsw_sp_qdisc_ets_destroy(mlxsw_sp_port, mlxsw_sp_qdisc);
+}
+
+static int
+__mlxsw_sp_qdisc_ets_check_params(unsigned int nbands)
+{
+	if (nbands > IEEE_8021QAZ_MAX_TCS)
+		return -EOPNOTSUPP;
+
+	return 0;
+}
+
+static int
+mlxsw_sp_qdisc_prio_check_params(struct mlxsw_sp_port *mlxsw_sp_port,
+				 void *params)
+{
+	struct tc_prio_qopt_offload_params *p = params;
+
+	return __mlxsw_sp_qdisc_ets_check_params(p->bands);
+}
+
+static struct mlxsw_sp_qdisc *
+mlxsw_sp_qdisc_walk_cb_clean_stats(struct mlxsw_sp_qdisc *mlxsw_sp_qdisc,
+				   void *mlxsw_sp_port)
+{
+	u64 backlog;
+
+	if (mlxsw_sp_qdisc->ops) {
+		backlog = mlxsw_sp_qdisc->stats_base.backlog;
+		if (mlxsw_sp_qdisc->ops->clean_stats)
+			mlxsw_sp_qdisc->ops->clean_stats(mlxsw_sp_port,
+							 mlxsw_sp_qdisc);
+		mlxsw_sp_qdisc->stats_base.backlog = backlog;
+	}
+
+	return NULL;
+}
+
+static void
+mlxsw_sp_qdisc_tree_clean_stats(struct mlxsw_sp_port *mlxsw_sp_port,
+				struct mlxsw_sp_qdisc *mlxsw_sp_qdisc)
+{
+	mlxsw_sp_qdisc_walk(mlxsw_sp_qdisc, mlxsw_sp_qdisc_walk_cb_clean_stats,
+			    mlxsw_sp_port);
+}
+
+static int
+__mlxsw_sp_qdisc_ets_replace(struct mlxsw_sp_port *mlxsw_sp_port,
+			     struct mlxsw_sp_qdisc *mlxsw_sp_qdisc,
+			     u32 handle, unsigned int nbands,
+			     const unsigned int *quanta,
+			     const unsigned int *weights,
+			     const u8 *priomap)
+{
+	struct mlxsw_sp_qdisc_ets_data *ets_data = mlxsw_sp_qdisc->ets_data;
+	struct mlxsw_sp_qdisc_ets_band *ets_band;
+	struct mlxsw_sp_qdisc *child_qdisc;
+	u8 old_priomap, new_priomap;
+	int i, band;
+	int err;
+
+	if (!ets_data) {
+		ets_data = kzalloc(sizeof(*ets_data), GFP_KERNEL);
+		if (!ets_data)
+			return -ENOMEM;
+		mlxsw_sp_qdisc->ets_data = ets_data;
+
+		for (band = 0; band < mlxsw_sp_qdisc->num_classes; band++) {
+			int tclass_num = MLXSW_SP_PRIO_BAND_TO_TCLASS(band);
+
+			ets_band = &ets_data->bands[band];
+			ets_band->tclass_num = tclass_num;
+		}
+	}
+
+	for (band = 0; band < nbands; band++) {
+		int tclass_num;
+
+		child_qdisc = &mlxsw_sp_qdisc->qdiscs[band];
+		ets_band = &ets_data->bands[band];
+
+		tclass_num = ets_band->tclass_num;
+		old_priomap = ets_band->prio_bitmap;
+		new_priomap = 0;
+
+		err = mlxsw_sp_port_ets_set(mlxsw_sp_port,
+					    MLXSW_REG_QEEC_HR_SUBGROUP,
+					    tclass_num, 0, !!quanta[band],
+					    weights[band]);
+		if (err)
+			return err;
+
+		for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) {
+			if (priomap[i] == band) {
+				new_priomap |= BIT(i);
+				if (BIT(i) & old_priomap)
+					continue;
+				err = mlxsw_sp_port_prio_tc_set(mlxsw_sp_port,
+								i, tclass_num);
+				if (err)
+					return err;
+			}
+		}
+
+		ets_band->prio_bitmap = new_priomap;
+
+		if (old_priomap != new_priomap)
+			mlxsw_sp_qdisc_tree_clean_stats(mlxsw_sp_port,
+							child_qdisc);
+
+		err = mlxsw_sp_qdisc_future_fifo_replace(mlxsw_sp_port, handle,
+							 band, child_qdisc);
+		if (err)
+			return err;
+	}
+	for (; band < IEEE_8021QAZ_MAX_TCS; band++) {
+		ets_band = &ets_data->bands[band];
+		ets_band->prio_bitmap = 0;
+
+		child_qdisc = &mlxsw_sp_qdisc->qdiscs[band];
+		mlxsw_sp_qdisc_destroy(mlxsw_sp_port, child_qdisc);
+
+		mlxsw_sp_port_ets_set(mlxsw_sp_port,
+				      MLXSW_REG_QEEC_HR_SUBGROUP,
+				      ets_band->tclass_num, 0, false, 0);
+	}
+
+	mlxsw_sp_qdisc_future_fifos_init(mlxsw_sp_port, TC_H_UNSPEC);
+	return 0;
+}
+
+static int
+mlxsw_sp_qdisc_prio_replace(struct mlxsw_sp_port *mlxsw_sp_port, u32 handle,
+			    struct mlxsw_sp_qdisc *mlxsw_sp_qdisc,
+			    void *params)
+{
+	struct tc_prio_qopt_offload_params *p = params;
+	unsigned int zeroes[TCQ_ETS_MAX_BANDS] = {0};
+
+	return __mlxsw_sp_qdisc_ets_replace(mlxsw_sp_port, mlxsw_sp_qdisc,
+					    handle, p->bands, zeroes,
+					    zeroes, p->priomap);
+}
+
+static void
+__mlxsw_sp_qdisc_ets_unoffload(struct mlxsw_sp_port *mlxsw_sp_port,
+			       struct mlxsw_sp_qdisc *mlxsw_sp_qdisc,
+			       struct gnet_stats_queue *qstats)
+{
+	u64 backlog;
+
+	backlog = mlxsw_sp_cells_bytes(mlxsw_sp_port->mlxsw_sp,
+				       mlxsw_sp_qdisc->stats_base.backlog);
+	qstats->backlog -= backlog;
+}
+
+static void
+mlxsw_sp_qdisc_prio_unoffload(struct mlxsw_sp_port *mlxsw_sp_port,
+			      struct mlxsw_sp_qdisc *mlxsw_sp_qdisc,
+			      void *params)
+{
+	struct tc_prio_qopt_offload_params *p = params;
+
+	__mlxsw_sp_qdisc_ets_unoffload(mlxsw_sp_port, mlxsw_sp_qdisc,
+				       p->qstats);
+}
+
+static int
+mlxsw_sp_qdisc_get_prio_stats(struct mlxsw_sp_port *mlxsw_sp_port,
+			      struct mlxsw_sp_qdisc *mlxsw_sp_qdisc,
+			      struct tc_qopt_offload_stats *stats_ptr)
+{
+	struct mlxsw_sp_qdisc *tc_qdisc;
+	u64 tx_packets = 0;
+	u64 tx_bytes = 0;
+	u64 backlog = 0;
+	u64 drops = 0;
+	int i;
+
+	for (i = 0; i < mlxsw_sp_qdisc->num_classes; i++) {
+		tc_qdisc = &mlxsw_sp_qdisc->qdiscs[i];
+		mlxsw_sp_qdisc_collect_tc_stats(mlxsw_sp_port, tc_qdisc,
+						&tx_bytes, &tx_packets,
+						&drops, &backlog);
+	}
+
+	mlxsw_sp_qdisc_update_stats(mlxsw_sp_port->mlxsw_sp, mlxsw_sp_qdisc,
+				    tx_bytes, tx_packets, drops, backlog,
+				    stats_ptr);
+	return 0;
+}
+
+static void
+mlxsw_sp_setup_tc_qdisc_prio_clean_stats(struct mlxsw_sp_port *mlxsw_sp_port,
+					 struct mlxsw_sp_qdisc *mlxsw_sp_qdisc)
+{
+	struct mlxsw_sp_qdisc_stats *stats_base;
+	struct mlxsw_sp_port_xstats *xstats;
+	struct rtnl_link_stats64 *stats;
+	int i;
+
+	xstats = &mlxsw_sp_port->periodic_hw_stats.xstats;
+	stats = &mlxsw_sp_port->periodic_hw_stats.stats;
+	stats_base = &mlxsw_sp_qdisc->stats_base;
+
+	stats_base->tx_packets = stats->tx_packets;
+	stats_base->tx_bytes = stats->tx_bytes;
+
+	stats_base->drops = 0;
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) {
+		stats_base->drops += mlxsw_sp_xstats_tail_drop(xstats, i);
+		stats_base->drops += xstats->wred_drop[i];
+	}
+
+	mlxsw_sp_qdisc->stats_base.backlog = 0;
+}
+
+static struct mlxsw_sp_qdisc *
+mlxsw_sp_qdisc_prio_find_class(struct mlxsw_sp_qdisc *mlxsw_sp_qdisc,
+			       u32 parent)
+{
+	int child_index = TC_H_MIN(parent);
+	int band = child_index - 1;
+
+	if (band < 0 || band >= mlxsw_sp_qdisc->num_classes)
+		return NULL;
+	return &mlxsw_sp_qdisc->qdiscs[band];
+}
+
+static struct mlxsw_sp_qdisc_ets_band *
+mlxsw_sp_qdisc_ets_get_band(struct mlxsw_sp_qdisc *mlxsw_sp_qdisc,
+			    struct mlxsw_sp_qdisc *child)
+{
+	unsigned int band = child - mlxsw_sp_qdisc->qdiscs;
+
+	if (WARN_ON(band >= IEEE_8021QAZ_MAX_TCS))
+		band = 0;
+	return &mlxsw_sp_qdisc->ets_data->bands[band];
+}
+
+static u8
+mlxsw_sp_qdisc_ets_get_prio_bitmap(struct mlxsw_sp_qdisc *mlxsw_sp_qdisc,
+				   struct mlxsw_sp_qdisc *child)
+{
+	return mlxsw_sp_qdisc_ets_get_band(mlxsw_sp_qdisc, child)->prio_bitmap;
+}
+
+static int
+mlxsw_sp_qdisc_ets_get_tclass_num(struct mlxsw_sp_qdisc *mlxsw_sp_qdisc,
+				  struct mlxsw_sp_qdisc *child)
+{
+	return mlxsw_sp_qdisc_ets_get_band(mlxsw_sp_qdisc, child)->tclass_num;
+}
+
+static struct mlxsw_sp_qdisc_ops mlxsw_sp_qdisc_ops_prio = {
+	.type = MLXSW_SP_QDISC_PRIO,
+	.check_params = mlxsw_sp_qdisc_prio_check_params,
+	.replace = mlxsw_sp_qdisc_prio_replace,
+	.unoffload = mlxsw_sp_qdisc_prio_unoffload,
+	.destroy = mlxsw_sp_qdisc_prio_destroy,
+	.get_stats = mlxsw_sp_qdisc_get_prio_stats,
+	.clean_stats = mlxsw_sp_setup_tc_qdisc_prio_clean_stats,
+	.find_class = mlxsw_sp_qdisc_prio_find_class,
+	.num_classes = IEEE_8021QAZ_MAX_TCS,
+	.get_prio_bitmap = mlxsw_sp_qdisc_ets_get_prio_bitmap,
+	.get_tclass_num = mlxsw_sp_qdisc_ets_get_tclass_num,
+};
+
+static int
+mlxsw_sp_qdisc_ets_check_params(struct mlxsw_sp_port *mlxsw_sp_port,
+				void *params)
+{
+	struct tc_ets_qopt_offload_replace_params *p = params;
+
+	return __mlxsw_sp_qdisc_ets_check_params(p->bands);
+}
+
+static int
+mlxsw_sp_qdisc_ets_replace(struct mlxsw_sp_port *mlxsw_sp_port, u32 handle,
+			   struct mlxsw_sp_qdisc *mlxsw_sp_qdisc,
+			   void *params)
+{
+	struct tc_ets_qopt_offload_replace_params *p = params;
+
+	return __mlxsw_sp_qdisc_ets_replace(mlxsw_sp_port, mlxsw_sp_qdisc,
+					    handle, p->bands, p->quanta,
+					    p->weights, p->priomap);
+}
+
+static void
+mlxsw_sp_qdisc_ets_unoffload(struct mlxsw_sp_port *mlxsw_sp_port,
+			     struct mlxsw_sp_qdisc *mlxsw_sp_qdisc,
+			     void *params)
+{
+	struct tc_ets_qopt_offload_replace_params *p = params;
+
+	__mlxsw_sp_qdisc_ets_unoffload(mlxsw_sp_port, mlxsw_sp_qdisc,
+				       p->qstats);
+}
+
+static int
+mlxsw_sp_qdisc_ets_destroy(struct mlxsw_sp_port *mlxsw_sp_port,
+			   struct mlxsw_sp_qdisc *mlxsw_sp_qdisc)
+{
+	return __mlxsw_sp_qdisc_ets_destroy(mlxsw_sp_port, mlxsw_sp_qdisc);
+}
+
+static struct mlxsw_sp_qdisc_ops mlxsw_sp_qdisc_ops_ets = {
+	.type = MLXSW_SP_QDISC_ETS,
+	.check_params = mlxsw_sp_qdisc_ets_check_params,
+	.replace = mlxsw_sp_qdisc_ets_replace,
+	.unoffload = mlxsw_sp_qdisc_ets_unoffload,
+	.destroy = mlxsw_sp_qdisc_ets_destroy,
+	.get_stats = mlxsw_sp_qdisc_get_prio_stats,
+	.clean_stats = mlxsw_sp_setup_tc_qdisc_prio_clean_stats,
+	.find_class = mlxsw_sp_qdisc_prio_find_class,
+	.num_classes = IEEE_8021QAZ_MAX_TCS,
+	.get_prio_bitmap = mlxsw_sp_qdisc_ets_get_prio_bitmap,
+	.get_tclass_num = mlxsw_sp_qdisc_ets_get_tclass_num,
+};
+
+/* Linux allows linking of Qdiscs to arbitrary classes (so long as the resulting
+ * graph is free of cycles). These operations do not change the parent handle
+ * though, which means it can be incomplete (if there is more than one class
+ * where the Qdisc in question is grafted) or outright wrong (if the Qdisc was
+ * linked to a different class and then removed from the original class).
+ *
+ * E.g. consider this sequence of operations:
+ *
+ *  # tc qdisc add dev swp1 root handle 1: prio
+ *  # tc qdisc add dev swp1 parent 1:3 handle 13: red limit 1000000 avpkt 10000
+ *  RED: set bandwidth to 10Mbit
+ *  # tc qdisc link dev swp1 handle 13: parent 1:2
+ *
+ * At this point, both 1:2 and 1:3 have the same RED Qdisc instance as their
+ * child. But RED will still only claim that 1:3 is its parent. If it's removed
+ * from that band, its only parent will be 1:2, but it will continue to claim
+ * that it is in fact 1:3.
+ *
+ * The notification for child Qdisc replace (e.g. TC_RED_REPLACE) comes before
+ * the notification for parent graft (e.g. TC_PRIO_GRAFT). We take the replace
+ * notification to offload the child Qdisc, based on its parent handle, and use
+ * the graft operation to validate that the class where the child is actually
+ * grafted corresponds to the parent handle. If the two don't match, we
+ * unoffload the child.
+ */
+static int mlxsw_sp_qdisc_graft(struct mlxsw_sp_port *mlxsw_sp_port,
+				struct mlxsw_sp_qdisc *mlxsw_sp_qdisc,
+				u8 band, u32 child_handle)
+{
+	struct mlxsw_sp_qdisc *old_qdisc;
+	u32 parent;
+
+	if (band < mlxsw_sp_qdisc->num_classes &&
+	    mlxsw_sp_qdisc->qdiscs[band].handle == child_handle)
+		return 0;
+
+	if (!child_handle) {
+		/* This is an invisible FIFO replacing the original Qdisc.
+		 * Ignore it--the original Qdisc's destroy will follow.
+		 */
+		return 0;
+	}
+
+	/* See if the grafted qdisc is already offloaded on any tclass. If so,
+	 * unoffload it.
+	 */
+	old_qdisc = mlxsw_sp_qdisc_find_by_handle(mlxsw_sp_port,
+						  child_handle);
+	if (old_qdisc)
+		mlxsw_sp_qdisc_destroy(mlxsw_sp_port, old_qdisc);
+
+	parent = TC_H_MAKE(mlxsw_sp_qdisc->handle, band + 1);
+	mlxsw_sp_qdisc = mlxsw_sp_qdisc->ops->find_class(mlxsw_sp_qdisc,
+							 parent);
+	if (!WARN_ON(!mlxsw_sp_qdisc))
+		mlxsw_sp_qdisc_destroy(mlxsw_sp_port, mlxsw_sp_qdisc);
+
+	return -EOPNOTSUPP;
+}
+
+static int __mlxsw_sp_setup_tc_prio(struct mlxsw_sp_port *mlxsw_sp_port,
+				    struct tc_prio_qopt_offload *p)
+{
+	struct mlxsw_sp_qdisc *mlxsw_sp_qdisc;
+
+	mlxsw_sp_qdisc = mlxsw_sp_qdisc_find(mlxsw_sp_port, p->parent);
+	if (!mlxsw_sp_qdisc)
+		return -EOPNOTSUPP;
+
+	if (p->command == TC_PRIO_REPLACE)
+		return mlxsw_sp_qdisc_replace(mlxsw_sp_port, p->handle,
+					      mlxsw_sp_qdisc,
+					      &mlxsw_sp_qdisc_ops_prio,
+					      &p->replace_params);
+
+	if (!mlxsw_sp_qdisc_compare(mlxsw_sp_qdisc, p->handle))
+		return -EOPNOTSUPP;
+
+	switch (p->command) {
+	case TC_PRIO_DESTROY:
+		return mlxsw_sp_qdisc_destroy(mlxsw_sp_port, mlxsw_sp_qdisc);
+	case TC_PRIO_STATS:
+		return mlxsw_sp_qdisc_get_stats(mlxsw_sp_port, mlxsw_sp_qdisc,
+						&p->stats);
+	case TC_PRIO_GRAFT:
+		return mlxsw_sp_qdisc_graft(mlxsw_sp_port, mlxsw_sp_qdisc,
+					    p->graft_params.band,
+					    p->graft_params.child_handle);
+	default:
+		return -EOPNOTSUPP;
+	}
+}
+
+int mlxsw_sp_setup_tc_prio(struct mlxsw_sp_port *mlxsw_sp_port,
+			   struct tc_prio_qopt_offload *p)
+{
+	int err;
+
+	mutex_lock(&mlxsw_sp_port->qdisc->lock);
+	err = __mlxsw_sp_setup_tc_prio(mlxsw_sp_port, p);
+	mutex_unlock(&mlxsw_sp_port->qdisc->lock);
+
+	return err;
+}
+
+static int __mlxsw_sp_setup_tc_ets(struct mlxsw_sp_port *mlxsw_sp_port,
+				   struct tc_ets_qopt_offload *p)
+{
+	struct mlxsw_sp_qdisc *mlxsw_sp_qdisc;
+
+	mlxsw_sp_qdisc = mlxsw_sp_qdisc_find(mlxsw_sp_port, p->parent);
+	if (!mlxsw_sp_qdisc)
+		return -EOPNOTSUPP;
+
+	if (p->command == TC_ETS_REPLACE)
+		return mlxsw_sp_qdisc_replace(mlxsw_sp_port, p->handle,
+					      mlxsw_sp_qdisc,
+					      &mlxsw_sp_qdisc_ops_ets,
+					      &p->replace_params);
+
+	if (!mlxsw_sp_qdisc_compare(mlxsw_sp_qdisc, p->handle))
+		return -EOPNOTSUPP;
+
+	switch (p->command) {
+	case TC_ETS_DESTROY:
+		return mlxsw_sp_qdisc_destroy(mlxsw_sp_port, mlxsw_sp_qdisc);
+	case TC_ETS_STATS:
+		return mlxsw_sp_qdisc_get_stats(mlxsw_sp_port, mlxsw_sp_qdisc,
+						&p->stats);
+	case TC_ETS_GRAFT:
+		return mlxsw_sp_qdisc_graft(mlxsw_sp_port, mlxsw_sp_qdisc,
+					    p->graft_params.band,
+					    p->graft_params.child_handle);
+	default:
+		return -EOPNOTSUPP;
+	}
+}
+
+int mlxsw_sp_setup_tc_ets(struct mlxsw_sp_port *mlxsw_sp_port,
+			  struct tc_ets_qopt_offload *p)
+{
+	int err;
+
+	mutex_lock(&mlxsw_sp_port->qdisc->lock);
+	err = __mlxsw_sp_setup_tc_ets(mlxsw_sp_port, p);
+	mutex_unlock(&mlxsw_sp_port->qdisc->lock);
+
+	return err;
+}
+
+struct mlxsw_sp_qevent_block {
+	struct list_head binding_list;
+	struct list_head mall_entry_list;
+	struct mlxsw_sp *mlxsw_sp;
+};
+
+struct mlxsw_sp_qevent_binding {
+	struct list_head list;
+	struct mlxsw_sp_port *mlxsw_sp_port;
+	u32 handle;
+	int tclass_num;
+	enum mlxsw_sp_span_trigger span_trigger;
+	unsigned int action_mask;
+};
+
+static LIST_HEAD(mlxsw_sp_qevent_block_cb_list);
+
+static int mlxsw_sp_qevent_span_configure(struct mlxsw_sp *mlxsw_sp,
+					  struct mlxsw_sp_mall_entry *mall_entry,
+					  struct mlxsw_sp_qevent_binding *qevent_binding,
+					  const struct mlxsw_sp_span_agent_parms *agent_parms,
+					  int *p_span_id)
+{
+	enum mlxsw_sp_span_trigger span_trigger = qevent_binding->span_trigger;
+	struct mlxsw_sp_port *mlxsw_sp_port = qevent_binding->mlxsw_sp_port;
+	struct mlxsw_sp_span_trigger_parms trigger_parms = {};
+	bool ingress;
+	int span_id;
+	int err;
+
+	err = mlxsw_sp_span_agent_get(mlxsw_sp, &span_id, agent_parms);
+	if (err)
+		return err;
+
+	ingress = mlxsw_sp_span_trigger_is_ingress(span_trigger);
+	err = mlxsw_sp_span_analyzed_port_get(mlxsw_sp_port, ingress);
+	if (err)
+		goto err_analyzed_port_get;
+
+	trigger_parms.span_id = span_id;
+	trigger_parms.probability_rate = 1;
+	err = mlxsw_sp_span_agent_bind(mlxsw_sp, span_trigger, mlxsw_sp_port,
+				       &trigger_parms);
+	if (err)
+		goto err_agent_bind;
+
+	err = mlxsw_sp_span_trigger_enable(mlxsw_sp_port, span_trigger,
+					   qevent_binding->tclass_num);
+	if (err)
+		goto err_trigger_enable;
+
+	*p_span_id = span_id;
+	return 0;
+
+err_trigger_enable:
+	mlxsw_sp_span_agent_unbind(mlxsw_sp, span_trigger, mlxsw_sp_port,
+				   &trigger_parms);
+err_agent_bind:
+	mlxsw_sp_span_analyzed_port_put(mlxsw_sp_port, ingress);
+err_analyzed_port_get:
+	mlxsw_sp_span_agent_put(mlxsw_sp, span_id);
+	return err;
+}
+
+static void mlxsw_sp_qevent_span_deconfigure(struct mlxsw_sp *mlxsw_sp,
+					     struct mlxsw_sp_qevent_binding *qevent_binding,
+					     int span_id)
+{
+	enum mlxsw_sp_span_trigger span_trigger = qevent_binding->span_trigger;
+	struct mlxsw_sp_port *mlxsw_sp_port = qevent_binding->mlxsw_sp_port;
+	struct mlxsw_sp_span_trigger_parms trigger_parms = {
+		.span_id = span_id,
+	};
+	bool ingress;
+
+	ingress = mlxsw_sp_span_trigger_is_ingress(span_trigger);
+
+	mlxsw_sp_span_trigger_disable(mlxsw_sp_port, span_trigger,
+				      qevent_binding->tclass_num);
+	mlxsw_sp_span_agent_unbind(mlxsw_sp, span_trigger, mlxsw_sp_port,
+				   &trigger_parms);
+	mlxsw_sp_span_analyzed_port_put(mlxsw_sp_port, ingress);
+	mlxsw_sp_span_agent_put(mlxsw_sp, span_id);
+}
+
+static int mlxsw_sp_qevent_mirror_configure(struct mlxsw_sp *mlxsw_sp,
+					    struct mlxsw_sp_mall_entry *mall_entry,
+					    struct mlxsw_sp_qevent_binding *qevent_binding)
+{
+	struct mlxsw_sp_span_agent_parms agent_parms = {
+		.to_dev = mall_entry->mirror.to_dev,
+	};
+
+	return mlxsw_sp_qevent_span_configure(mlxsw_sp, mall_entry, qevent_binding,
+					      &agent_parms, &mall_entry->mirror.span_id);
+}
+
+static void mlxsw_sp_qevent_mirror_deconfigure(struct mlxsw_sp *mlxsw_sp,
+					       struct mlxsw_sp_mall_entry *mall_entry,
+					       struct mlxsw_sp_qevent_binding *qevent_binding)
+{
+	mlxsw_sp_qevent_span_deconfigure(mlxsw_sp, qevent_binding, mall_entry->mirror.span_id);
+}
+
+static int mlxsw_sp_qevent_trap_configure(struct mlxsw_sp *mlxsw_sp,
+					  struct mlxsw_sp_mall_entry *mall_entry,
+					  struct mlxsw_sp_qevent_binding *qevent_binding)
+{
+	struct mlxsw_sp_span_agent_parms agent_parms = {
+		.session_id = MLXSW_SP_SPAN_SESSION_ID_BUFFER,
+	};
+	int err;
+
+	err = mlxsw_sp_trap_group_policer_hw_id_get(mlxsw_sp,
+						    DEVLINK_TRAP_GROUP_GENERIC_ID_BUFFER_DROPS,
+						    &agent_parms.policer_enable,
+						    &agent_parms.policer_id);
+	if (err)
+		return err;
+
+	return mlxsw_sp_qevent_span_configure(mlxsw_sp, mall_entry, qevent_binding,
+					      &agent_parms, &mall_entry->trap.span_id);
+}
+
+static void mlxsw_sp_qevent_trap_deconfigure(struct mlxsw_sp *mlxsw_sp,
+					     struct mlxsw_sp_mall_entry *mall_entry,
+					     struct mlxsw_sp_qevent_binding *qevent_binding)
+{
+	mlxsw_sp_qevent_span_deconfigure(mlxsw_sp, qevent_binding, mall_entry->trap.span_id);
+}
+
+static int
+mlxsw_sp_qevent_entry_configure(struct mlxsw_sp *mlxsw_sp,
+				struct mlxsw_sp_mall_entry *mall_entry,
+				struct mlxsw_sp_qevent_binding *qevent_binding,
+				struct netlink_ext_ack *extack)
+{
+	if (!(BIT(mall_entry->type) & qevent_binding->action_mask)) {
+		NL_SET_ERR_MSG(extack, "Action not supported at this qevent");
+		return -EOPNOTSUPP;
+	}
+
+	switch (mall_entry->type) {
+	case MLXSW_SP_MALL_ACTION_TYPE_MIRROR:
+		return mlxsw_sp_qevent_mirror_configure(mlxsw_sp, mall_entry, qevent_binding);
+	case MLXSW_SP_MALL_ACTION_TYPE_TRAP:
+		return mlxsw_sp_qevent_trap_configure(mlxsw_sp, mall_entry, qevent_binding);
+	default:
+		/* This should have been validated away. */
+		WARN_ON(1);
+		return -EOPNOTSUPP;
+	}
+}
+
+static void mlxsw_sp_qevent_entry_deconfigure(struct mlxsw_sp *mlxsw_sp,
+					      struct mlxsw_sp_mall_entry *mall_entry,
+					      struct mlxsw_sp_qevent_binding *qevent_binding)
+{
+	switch (mall_entry->type) {
+	case MLXSW_SP_MALL_ACTION_TYPE_MIRROR:
+		return mlxsw_sp_qevent_mirror_deconfigure(mlxsw_sp, mall_entry, qevent_binding);
+	case MLXSW_SP_MALL_ACTION_TYPE_TRAP:
+		return mlxsw_sp_qevent_trap_deconfigure(mlxsw_sp, mall_entry, qevent_binding);
+	default:
+		WARN_ON(1);
+		return;
+	}
+}
+
+static int
+mlxsw_sp_qevent_binding_configure(struct mlxsw_sp_qevent_block *qevent_block,
+				  struct mlxsw_sp_qevent_binding *qevent_binding,
+				  struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp_mall_entry *mall_entry;
+	int err;
+
+	list_for_each_entry(mall_entry, &qevent_block->mall_entry_list, list) {
+		err = mlxsw_sp_qevent_entry_configure(qevent_block->mlxsw_sp, mall_entry,
+						      qevent_binding, extack);
+		if (err)
+			goto err_entry_configure;
+	}
+
+	return 0;
+
+err_entry_configure:
+	list_for_each_entry_continue_reverse(mall_entry, &qevent_block->mall_entry_list, list)
+		mlxsw_sp_qevent_entry_deconfigure(qevent_block->mlxsw_sp, mall_entry,
+						  qevent_binding);
+	return err;
+}
+
+static void mlxsw_sp_qevent_binding_deconfigure(struct mlxsw_sp_qevent_block *qevent_block,
+						struct mlxsw_sp_qevent_binding *qevent_binding)
+{
+	struct mlxsw_sp_mall_entry *mall_entry;
+
+	list_for_each_entry(mall_entry, &qevent_block->mall_entry_list, list)
+		mlxsw_sp_qevent_entry_deconfigure(qevent_block->mlxsw_sp, mall_entry,
+						  qevent_binding);
+}
+
+static int
+mlxsw_sp_qevent_block_configure(struct mlxsw_sp_qevent_block *qevent_block,
+				struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp_qevent_binding *qevent_binding;
+	int err;
+
+	list_for_each_entry(qevent_binding, &qevent_block->binding_list, list) {
+		err = mlxsw_sp_qevent_binding_configure(qevent_block,
+							qevent_binding,
+							extack);
+		if (err)
+			goto err_binding_configure;
+	}
+
+	return 0;
+
+err_binding_configure:
+	list_for_each_entry_continue_reverse(qevent_binding, &qevent_block->binding_list, list)
+		mlxsw_sp_qevent_binding_deconfigure(qevent_block, qevent_binding);
+	return err;
+}
+
+static void mlxsw_sp_qevent_block_deconfigure(struct mlxsw_sp_qevent_block *qevent_block)
+{
+	struct mlxsw_sp_qevent_binding *qevent_binding;
+
+	list_for_each_entry(qevent_binding, &qevent_block->binding_list, list)
+		mlxsw_sp_qevent_binding_deconfigure(qevent_block, qevent_binding);
+}
+
+static struct mlxsw_sp_mall_entry *
+mlxsw_sp_qevent_mall_entry_find(struct mlxsw_sp_qevent_block *block, unsigned long cookie)
+{
+	struct mlxsw_sp_mall_entry *mall_entry;
+
+	list_for_each_entry(mall_entry, &block->mall_entry_list, list)
+		if (mall_entry->cookie == cookie)
+			return mall_entry;
+
+	return NULL;
+}
+
+static int mlxsw_sp_qevent_mall_replace(struct mlxsw_sp *mlxsw_sp,
+					struct mlxsw_sp_qevent_block *qevent_block,
+					struct tc_cls_matchall_offload *f)
+{
+	struct mlxsw_sp_mall_entry *mall_entry;
+	struct flow_action_entry *act;
+	int err;
+
+	/* It should not currently be possible to replace a matchall rule. So
+	 * this must be a new rule.
+	 */
+	if (!list_empty(&qevent_block->mall_entry_list)) {
+		NL_SET_ERR_MSG(f->common.extack, "At most one filter supported");
+		return -EOPNOTSUPP;
+	}
+	if (f->rule->action.num_entries != 1) {
+		NL_SET_ERR_MSG(f->common.extack, "Only singular actions supported");
+		return -EOPNOTSUPP;
+	}
+	if (f->common.chain_index) {
+		NL_SET_ERR_MSG(f->common.extack, "Only chain 0 is supported");
+		return -EOPNOTSUPP;
+	}
+	if (f->common.protocol != htons(ETH_P_ALL)) {
+		NL_SET_ERR_MSG(f->common.extack, "Protocol matching not supported");
+		return -EOPNOTSUPP;
+	}
+
+	act = &f->rule->action.entries[0];
+	if (!(act->hw_stats & FLOW_ACTION_HW_STATS_DISABLED)) {
+		NL_SET_ERR_MSG(f->common.extack, "HW counters not supported on qevents");
+		return -EOPNOTSUPP;
+	}
+
+	mall_entry = kzalloc(sizeof(*mall_entry), GFP_KERNEL);
+	if (!mall_entry)
+		return -ENOMEM;
+	mall_entry->cookie = f->cookie;
+
+	if (act->id == FLOW_ACTION_MIRRED) {
+		mall_entry->type = MLXSW_SP_MALL_ACTION_TYPE_MIRROR;
+		mall_entry->mirror.to_dev = act->dev;
+	} else if (act->id == FLOW_ACTION_TRAP) {
+		mall_entry->type = MLXSW_SP_MALL_ACTION_TYPE_TRAP;
+	} else {
+		NL_SET_ERR_MSG(f->common.extack, "Unsupported action");
+		err = -EOPNOTSUPP;
+		goto err_unsupported_action;
+	}
+
+	list_add_tail(&mall_entry->list, &qevent_block->mall_entry_list);
+
+	err = mlxsw_sp_qevent_block_configure(qevent_block, f->common.extack);
+	if (err)
+		goto err_block_configure;
+
+	return 0;
+
+err_block_configure:
+	list_del(&mall_entry->list);
+err_unsupported_action:
+	kfree(mall_entry);
+	return err;
+}
+
+static void mlxsw_sp_qevent_mall_destroy(struct mlxsw_sp_qevent_block *qevent_block,
+					 struct tc_cls_matchall_offload *f)
+{
+	struct mlxsw_sp_mall_entry *mall_entry;
+
+	mall_entry = mlxsw_sp_qevent_mall_entry_find(qevent_block, f->cookie);
+	if (!mall_entry)
+		return;
+
+	mlxsw_sp_qevent_block_deconfigure(qevent_block);
+
+	list_del(&mall_entry->list);
+	kfree(mall_entry);
+}
+
+static int mlxsw_sp_qevent_block_mall_cb(struct mlxsw_sp_qevent_block *qevent_block,
+					 struct tc_cls_matchall_offload *f)
+{
+	struct mlxsw_sp *mlxsw_sp = qevent_block->mlxsw_sp;
+
+	switch (f->command) {
+	case TC_CLSMATCHALL_REPLACE:
+		return mlxsw_sp_qevent_mall_replace(mlxsw_sp, qevent_block, f);
+	case TC_CLSMATCHALL_DESTROY:
+		mlxsw_sp_qevent_mall_destroy(qevent_block, f);
+		return 0;
+	default:
+		return -EOPNOTSUPP;
+	}
+}
+
+static int mlxsw_sp_qevent_block_cb(enum tc_setup_type type, void *type_data, void *cb_priv)
+{
+	struct mlxsw_sp_qevent_block *qevent_block = cb_priv;
+
+	switch (type) {
+	case TC_SETUP_CLSMATCHALL:
+		return mlxsw_sp_qevent_block_mall_cb(qevent_block, type_data);
+	default:
+		return -EOPNOTSUPP;
+	}
+}
+
+static struct mlxsw_sp_qevent_block *mlxsw_sp_qevent_block_create(struct mlxsw_sp *mlxsw_sp,
+								  struct net *net)
+{
+	struct mlxsw_sp_qevent_block *qevent_block;
+
+	qevent_block = kzalloc(sizeof(*qevent_block), GFP_KERNEL);
+	if (!qevent_block)
+		return NULL;
+
+	INIT_LIST_HEAD(&qevent_block->binding_list);
+	INIT_LIST_HEAD(&qevent_block->mall_entry_list);
+	qevent_block->mlxsw_sp = mlxsw_sp;
+	return qevent_block;
+}
+
+static void
+mlxsw_sp_qevent_block_destroy(struct mlxsw_sp_qevent_block *qevent_block)
+{
+	WARN_ON(!list_empty(&qevent_block->binding_list));
+	WARN_ON(!list_empty(&qevent_block->mall_entry_list));
+	kfree(qevent_block);
+}
+
+static void mlxsw_sp_qevent_block_release(void *cb_priv)
+{
+	struct mlxsw_sp_qevent_block *qevent_block = cb_priv;
+
+	mlxsw_sp_qevent_block_destroy(qevent_block);
+}
+
+static struct mlxsw_sp_qevent_binding *
+mlxsw_sp_qevent_binding_create(struct mlxsw_sp_port *mlxsw_sp_port, u32 handle, int tclass_num,
+			       enum mlxsw_sp_span_trigger span_trigger,
+			       unsigned int action_mask)
+{
+	struct mlxsw_sp_qevent_binding *binding;
+
+	binding = kzalloc(sizeof(*binding), GFP_KERNEL);
+	if (!binding)
+		return ERR_PTR(-ENOMEM);
+
+	binding->mlxsw_sp_port = mlxsw_sp_port;
+	binding->handle = handle;
+	binding->tclass_num = tclass_num;
+	binding->span_trigger = span_trigger;
+	binding->action_mask = action_mask;
+	return binding;
+}
+
+static void
+mlxsw_sp_qevent_binding_destroy(struct mlxsw_sp_qevent_binding *binding)
+{
+	kfree(binding);
+}
+
+static struct mlxsw_sp_qevent_binding *
+mlxsw_sp_qevent_binding_lookup(struct mlxsw_sp_qevent_block *block,
+			       struct mlxsw_sp_port *mlxsw_sp_port,
+			       u32 handle,
+			       enum mlxsw_sp_span_trigger span_trigger)
+{
+	struct mlxsw_sp_qevent_binding *qevent_binding;
+
+	list_for_each_entry(qevent_binding, &block->binding_list, list)
+		if (qevent_binding->mlxsw_sp_port == mlxsw_sp_port &&
+		    qevent_binding->handle == handle &&
+		    qevent_binding->span_trigger == span_trigger)
+			return qevent_binding;
+	return NULL;
+}
+
+static int
+mlxsw_sp_setup_tc_block_qevent_bind(struct mlxsw_sp_port *mlxsw_sp_port,
+				    struct flow_block_offload *f,
+				    enum mlxsw_sp_span_trigger span_trigger,
+				    unsigned int action_mask)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	struct mlxsw_sp_qevent_binding *qevent_binding;
+	struct mlxsw_sp_qevent_block *qevent_block;
+	struct flow_block_cb *block_cb;
+	struct mlxsw_sp_qdisc *qdisc;
+	bool register_block = false;
+	int tclass_num;
+	int err;
+
+	block_cb = flow_block_cb_lookup(f->block, mlxsw_sp_qevent_block_cb, mlxsw_sp);
+	if (!block_cb) {
+		qevent_block = mlxsw_sp_qevent_block_create(mlxsw_sp, f->net);
+		if (!qevent_block)
+			return -ENOMEM;
+		block_cb = flow_block_cb_alloc(mlxsw_sp_qevent_block_cb, mlxsw_sp, qevent_block,
+					       mlxsw_sp_qevent_block_release);
+		if (IS_ERR(block_cb)) {
+			mlxsw_sp_qevent_block_destroy(qevent_block);
+			return PTR_ERR(block_cb);
+		}
+		register_block = true;
+	} else {
+		qevent_block = flow_block_cb_priv(block_cb);
+	}
+	flow_block_cb_incref(block_cb);
+
+	qdisc = mlxsw_sp_qdisc_find_by_handle(mlxsw_sp_port, f->sch->handle);
+	if (!qdisc) {
+		NL_SET_ERR_MSG(f->extack, "Qdisc not offloaded");
+		err = -ENOENT;
+		goto err_find_qdisc;
+	}
+
+	if (WARN_ON(mlxsw_sp_qevent_binding_lookup(qevent_block, mlxsw_sp_port, f->sch->handle,
+						   span_trigger))) {
+		err = -EEXIST;
+		goto err_binding_exists;
+	}
+
+	tclass_num = mlxsw_sp_qdisc_get_tclass_num(mlxsw_sp_port, qdisc);
+	qevent_binding = mlxsw_sp_qevent_binding_create(mlxsw_sp_port,
+							f->sch->handle,
+							tclass_num,
+							span_trigger,
+							action_mask);
+	if (IS_ERR(qevent_binding)) {
+		err = PTR_ERR(qevent_binding);
+		goto err_binding_create;
+	}
+
+	err = mlxsw_sp_qevent_binding_configure(qevent_block, qevent_binding,
+						f->extack);
+	if (err)
+		goto err_binding_configure;
+
+	list_add(&qevent_binding->list, &qevent_block->binding_list);
+
+	if (register_block) {
+		flow_block_cb_add(block_cb, f);
+		list_add_tail(&block_cb->driver_list, &mlxsw_sp_qevent_block_cb_list);
+	}
+
+	return 0;
+
+err_binding_configure:
+	mlxsw_sp_qevent_binding_destroy(qevent_binding);
+err_binding_create:
+err_binding_exists:
+err_find_qdisc:
+	if (!flow_block_cb_decref(block_cb))
+		flow_block_cb_free(block_cb);
+	return err;
+}
+
+static void mlxsw_sp_setup_tc_block_qevent_unbind(struct mlxsw_sp_port *mlxsw_sp_port,
+						  struct flow_block_offload *f,
+						  enum mlxsw_sp_span_trigger span_trigger)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	struct mlxsw_sp_qevent_binding *qevent_binding;
+	struct mlxsw_sp_qevent_block *qevent_block;
+	struct flow_block_cb *block_cb;
+
+	block_cb = flow_block_cb_lookup(f->block, mlxsw_sp_qevent_block_cb, mlxsw_sp);
+	if (!block_cb)
+		return;
+	qevent_block = flow_block_cb_priv(block_cb);
+
+	qevent_binding = mlxsw_sp_qevent_binding_lookup(qevent_block, mlxsw_sp_port, f->sch->handle,
+							span_trigger);
+	if (!qevent_binding)
+		return;
+
+	list_del(&qevent_binding->list);
+	mlxsw_sp_qevent_binding_deconfigure(qevent_block, qevent_binding);
+	mlxsw_sp_qevent_binding_destroy(qevent_binding);
+
+	if (!flow_block_cb_decref(block_cb)) {
+		flow_block_cb_remove(block_cb, f);
+		list_del(&block_cb->driver_list);
+	}
+}
+
+static int
+mlxsw_sp_setup_tc_block_qevent(struct mlxsw_sp_port *mlxsw_sp_port,
+			       struct flow_block_offload *f,
+			       enum mlxsw_sp_span_trigger span_trigger,
+			       unsigned int action_mask)
+{
+	f->driver_block_list = &mlxsw_sp_qevent_block_cb_list;
+
+	switch (f->command) {
+	case FLOW_BLOCK_BIND:
+		return mlxsw_sp_setup_tc_block_qevent_bind(mlxsw_sp_port, f,
+							   span_trigger,
+							   action_mask);
+	case FLOW_BLOCK_UNBIND:
+		mlxsw_sp_setup_tc_block_qevent_unbind(mlxsw_sp_port, f, span_trigger);
+		return 0;
+	default:
+		return -EOPNOTSUPP;
+	}
+}
+
+int mlxsw_sp_setup_tc_block_qevent_early_drop(struct mlxsw_sp_port *mlxsw_sp_port,
+					      struct flow_block_offload *f)
+{
+	unsigned int action_mask = BIT(MLXSW_SP_MALL_ACTION_TYPE_MIRROR) |
+				   BIT(MLXSW_SP_MALL_ACTION_TYPE_TRAP);
+
+	return mlxsw_sp_setup_tc_block_qevent(mlxsw_sp_port, f,
+					      MLXSW_SP_SPAN_TRIGGER_EARLY_DROP,
+					      action_mask);
+}
+
+int mlxsw_sp_setup_tc_block_qevent_mark(struct mlxsw_sp_port *mlxsw_sp_port,
+					struct flow_block_offload *f)
+{
+	unsigned int action_mask = BIT(MLXSW_SP_MALL_ACTION_TYPE_MIRROR);
+
+	return mlxsw_sp_setup_tc_block_qevent(mlxsw_sp_port, f,
+					      MLXSW_SP_SPAN_TRIGGER_ECN,
+					      action_mask);
+}
+
+int mlxsw_sp_tc_qdisc_init(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	struct mlxsw_sp_qdisc_state *qdisc_state;
+
+	qdisc_state = kzalloc(sizeof(*qdisc_state), GFP_KERNEL);
+	if (!qdisc_state)
+		return -ENOMEM;
+
+	mutex_init(&qdisc_state->lock);
+	mlxsw_sp_port->qdisc = qdisc_state;
+	return 0;
+}
+
+void mlxsw_sp_tc_qdisc_fini(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	mutex_destroy(&mlxsw_sp_port->qdisc->lock);
+	kfree(mlxsw_sp_port->qdisc);
+}
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c
new file mode 100644
index 000000000..ab0aa1a61
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c
@@ -0,0 +1,10617 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2016-2018 Mellanox Technologies. All rights reserved */
+
+#include <linux/kernel.h>
+#include <linux/types.h>
+#include <linux/rhashtable.h>
+#include <linux/bitops.h>
+#include <linux/in6.h>
+#include <linux/notifier.h>
+#include <linux/inetdevice.h>
+#include <linux/netdevice.h>
+#include <linux/if_bridge.h>
+#include <linux/socket.h>
+#include <linux/route.h>
+#include <linux/gcd.h>
+#include <linux/if_macvlan.h>
+#include <linux/refcount.h>
+#include <linux/jhash.h>
+#include <linux/net_namespace.h>
+#include <linux/mutex.h>
+#include <net/netevent.h>
+#include <net/neighbour.h>
+#include <net/arp.h>
+#include <net/inet_dscp.h>
+#include <net/ip_fib.h>
+#include <net/ip6_fib.h>
+#include <net/nexthop.h>
+#include <net/fib_rules.h>
+#include <net/ip_tunnels.h>
+#include <net/l3mdev.h>
+#include <net/addrconf.h>
+#include <net/ndisc.h>
+#include <net/ipv6.h>
+#include <net/fib_notifier.h>
+#include <net/switchdev.h>
+
+#include "spectrum.h"
+#include "core.h"
+#include "reg.h"
+#include "spectrum_cnt.h"
+#include "spectrum_dpipe.h"
+#include "spectrum_ipip.h"
+#include "spectrum_mr.h"
+#include "spectrum_mr_tcam.h"
+#include "spectrum_router.h"
+#include "spectrum_span.h"
+
+struct mlxsw_sp_fib;
+struct mlxsw_sp_vr;
+struct mlxsw_sp_lpm_tree;
+struct mlxsw_sp_rif_ops;
+
+struct mlxsw_sp_rif {
+	struct list_head nexthop_list;
+	struct list_head neigh_list;
+	struct net_device *dev; /* NULL for underlay RIF */
+	struct mlxsw_sp_fid *fid;
+	unsigned char addr[ETH_ALEN];
+	int mtu;
+	u16 rif_index;
+	u8 mac_profile_id;
+	u16 vr_id;
+	const struct mlxsw_sp_rif_ops *ops;
+	struct mlxsw_sp *mlxsw_sp;
+
+	unsigned int counter_ingress;
+	bool counter_ingress_valid;
+	unsigned int counter_egress;
+	bool counter_egress_valid;
+};
+
+struct mlxsw_sp_rif_params {
+	struct net_device *dev;
+	union {
+		u16 system_port;
+		u16 lag_id;
+	};
+	u16 vid;
+	bool lag;
+};
+
+struct mlxsw_sp_rif_subport {
+	struct mlxsw_sp_rif common;
+	refcount_t ref_count;
+	union {
+		u16 system_port;
+		u16 lag_id;
+	};
+	u16 vid;
+	bool lag;
+};
+
+struct mlxsw_sp_rif_ipip_lb {
+	struct mlxsw_sp_rif common;
+	struct mlxsw_sp_rif_ipip_lb_config lb_config;
+	u16 ul_vr_id; /* Reserved for Spectrum-2. */
+	u16 ul_rif_id; /* Reserved for Spectrum. */
+};
+
+struct mlxsw_sp_rif_params_ipip_lb {
+	struct mlxsw_sp_rif_params common;
+	struct mlxsw_sp_rif_ipip_lb_config lb_config;
+};
+
+struct mlxsw_sp_rif_ops {
+	enum mlxsw_sp_rif_type type;
+	size_t rif_size;
+
+	void (*setup)(struct mlxsw_sp_rif *rif,
+		      const struct mlxsw_sp_rif_params *params);
+	int (*configure)(struct mlxsw_sp_rif *rif,
+			 struct netlink_ext_ack *extack);
+	void (*deconfigure)(struct mlxsw_sp_rif *rif);
+	struct mlxsw_sp_fid * (*fid_get)(struct mlxsw_sp_rif *rif,
+					 struct netlink_ext_ack *extack);
+	void (*fdb_del)(struct mlxsw_sp_rif *rif, const char *mac);
+};
+
+struct mlxsw_sp_rif_mac_profile {
+	unsigned char mac_prefix[ETH_ALEN];
+	refcount_t ref_count;
+	u8 id;
+};
+
+struct mlxsw_sp_router_ops {
+	int (*init)(struct mlxsw_sp *mlxsw_sp);
+	int (*ipips_init)(struct mlxsw_sp *mlxsw_sp);
+};
+
+static struct mlxsw_sp_rif *
+mlxsw_sp_rif_find_by_dev(const struct mlxsw_sp *mlxsw_sp,
+			 const struct net_device *dev);
+static void mlxsw_sp_rif_destroy(struct mlxsw_sp_rif *rif);
+static void mlxsw_sp_lpm_tree_hold(struct mlxsw_sp_lpm_tree *lpm_tree);
+static void mlxsw_sp_lpm_tree_put(struct mlxsw_sp *mlxsw_sp,
+				  struct mlxsw_sp_lpm_tree *lpm_tree);
+static int mlxsw_sp_vr_lpm_tree_bind(struct mlxsw_sp *mlxsw_sp,
+				     const struct mlxsw_sp_fib *fib,
+				     u8 tree_id);
+static int mlxsw_sp_vr_lpm_tree_unbind(struct mlxsw_sp *mlxsw_sp,
+				       const struct mlxsw_sp_fib *fib);
+
+static unsigned int *
+mlxsw_sp_rif_p_counter_get(struct mlxsw_sp_rif *rif,
+			   enum mlxsw_sp_rif_counter_dir dir)
+{
+	switch (dir) {
+	case MLXSW_SP_RIF_COUNTER_EGRESS:
+		return &rif->counter_egress;
+	case MLXSW_SP_RIF_COUNTER_INGRESS:
+		return &rif->counter_ingress;
+	}
+	return NULL;
+}
+
+static bool
+mlxsw_sp_rif_counter_valid_get(struct mlxsw_sp_rif *rif,
+			       enum mlxsw_sp_rif_counter_dir dir)
+{
+	switch (dir) {
+	case MLXSW_SP_RIF_COUNTER_EGRESS:
+		return rif->counter_egress_valid;
+	case MLXSW_SP_RIF_COUNTER_INGRESS:
+		return rif->counter_ingress_valid;
+	}
+	return false;
+}
+
+static void
+mlxsw_sp_rif_counter_valid_set(struct mlxsw_sp_rif *rif,
+			       enum mlxsw_sp_rif_counter_dir dir,
+			       bool valid)
+{
+	switch (dir) {
+	case MLXSW_SP_RIF_COUNTER_EGRESS:
+		rif->counter_egress_valid = valid;
+		break;
+	case MLXSW_SP_RIF_COUNTER_INGRESS:
+		rif->counter_ingress_valid = valid;
+		break;
+	}
+}
+
+static int mlxsw_sp_rif_counter_edit(struct mlxsw_sp *mlxsw_sp, u16 rif_index,
+				     unsigned int counter_index, bool enable,
+				     enum mlxsw_sp_rif_counter_dir dir)
+{
+	char ritr_pl[MLXSW_REG_RITR_LEN];
+	bool is_egress = false;
+	int err;
+
+	if (dir == MLXSW_SP_RIF_COUNTER_EGRESS)
+		is_egress = true;
+	mlxsw_reg_ritr_rif_pack(ritr_pl, rif_index);
+	err = mlxsw_reg_query(mlxsw_sp->core, MLXSW_REG(ritr), ritr_pl);
+	if (err)
+		return err;
+
+	mlxsw_reg_ritr_counter_pack(ritr_pl, counter_index, enable,
+				    is_egress);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(ritr), ritr_pl);
+}
+
+int mlxsw_sp_rif_counter_value_get(struct mlxsw_sp *mlxsw_sp,
+				   struct mlxsw_sp_rif *rif,
+				   enum mlxsw_sp_rif_counter_dir dir, u64 *cnt)
+{
+	char ricnt_pl[MLXSW_REG_RICNT_LEN];
+	unsigned int *p_counter_index;
+	bool valid;
+	int err;
+
+	valid = mlxsw_sp_rif_counter_valid_get(rif, dir);
+	if (!valid)
+		return -EINVAL;
+
+	p_counter_index = mlxsw_sp_rif_p_counter_get(rif, dir);
+	if (!p_counter_index)
+		return -EINVAL;
+	mlxsw_reg_ricnt_pack(ricnt_pl, *p_counter_index,
+			     MLXSW_REG_RICNT_OPCODE_NOP);
+	err = mlxsw_reg_query(mlxsw_sp->core, MLXSW_REG(ricnt), ricnt_pl);
+	if (err)
+		return err;
+	*cnt = mlxsw_reg_ricnt_good_unicast_packets_get(ricnt_pl);
+	return 0;
+}
+
+struct mlxsw_sp_rif_counter_set_basic {
+	u64 good_unicast_packets;
+	u64 good_multicast_packets;
+	u64 good_broadcast_packets;
+	u64 good_unicast_bytes;
+	u64 good_multicast_bytes;
+	u64 good_broadcast_bytes;
+	u64 error_packets;
+	u64 discard_packets;
+	u64 error_bytes;
+	u64 discard_bytes;
+};
+
+static int
+mlxsw_sp_rif_counter_fetch_clear(struct mlxsw_sp_rif *rif,
+				 enum mlxsw_sp_rif_counter_dir dir,
+				 struct mlxsw_sp_rif_counter_set_basic *set)
+{
+	struct mlxsw_sp *mlxsw_sp = rif->mlxsw_sp;
+	char ricnt_pl[MLXSW_REG_RICNT_LEN];
+	unsigned int *p_counter_index;
+	int err;
+
+	if (!mlxsw_sp_rif_counter_valid_get(rif, dir))
+		return -EINVAL;
+
+	p_counter_index = mlxsw_sp_rif_p_counter_get(rif, dir);
+	if (!p_counter_index)
+		return -EINVAL;
+
+	mlxsw_reg_ricnt_pack(ricnt_pl, *p_counter_index,
+			     MLXSW_REG_RICNT_OPCODE_CLEAR);
+	err = mlxsw_reg_query(mlxsw_sp->core, MLXSW_REG(ricnt), ricnt_pl);
+	if (err)
+		return err;
+
+	if (!set)
+		return 0;
+
+#define MLXSW_SP_RIF_COUNTER_EXTRACT(NAME)				\
+		(set->NAME = mlxsw_reg_ricnt_ ## NAME ## _get(ricnt_pl))
+
+	MLXSW_SP_RIF_COUNTER_EXTRACT(good_unicast_packets);
+	MLXSW_SP_RIF_COUNTER_EXTRACT(good_multicast_packets);
+	MLXSW_SP_RIF_COUNTER_EXTRACT(good_broadcast_packets);
+	MLXSW_SP_RIF_COUNTER_EXTRACT(good_unicast_bytes);
+	MLXSW_SP_RIF_COUNTER_EXTRACT(good_multicast_bytes);
+	MLXSW_SP_RIF_COUNTER_EXTRACT(good_broadcast_bytes);
+	MLXSW_SP_RIF_COUNTER_EXTRACT(error_packets);
+	MLXSW_SP_RIF_COUNTER_EXTRACT(discard_packets);
+	MLXSW_SP_RIF_COUNTER_EXTRACT(error_bytes);
+	MLXSW_SP_RIF_COUNTER_EXTRACT(discard_bytes);
+
+#undef MLXSW_SP_RIF_COUNTER_EXTRACT
+
+	return 0;
+}
+
+static int mlxsw_sp_rif_counter_clear(struct mlxsw_sp *mlxsw_sp,
+				      unsigned int counter_index)
+{
+	char ricnt_pl[MLXSW_REG_RICNT_LEN];
+
+	mlxsw_reg_ricnt_pack(ricnt_pl, counter_index,
+			     MLXSW_REG_RICNT_OPCODE_CLEAR);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(ricnt), ricnt_pl);
+}
+
+int mlxsw_sp_rif_counter_alloc(struct mlxsw_sp_rif *rif,
+			       enum mlxsw_sp_rif_counter_dir dir)
+{
+	struct mlxsw_sp *mlxsw_sp = rif->mlxsw_sp;
+	unsigned int *p_counter_index;
+	int err;
+
+	if (mlxsw_sp_rif_counter_valid_get(rif, dir))
+		return 0;
+
+	p_counter_index = mlxsw_sp_rif_p_counter_get(rif, dir);
+	if (!p_counter_index)
+		return -EINVAL;
+
+	err = mlxsw_sp_counter_alloc(mlxsw_sp, MLXSW_SP_COUNTER_SUB_POOL_RIF,
+				     p_counter_index);
+	if (err)
+		return err;
+
+	err = mlxsw_sp_rif_counter_clear(mlxsw_sp, *p_counter_index);
+	if (err)
+		goto err_counter_clear;
+
+	err = mlxsw_sp_rif_counter_edit(mlxsw_sp, rif->rif_index,
+					*p_counter_index, true, dir);
+	if (err)
+		goto err_counter_edit;
+	mlxsw_sp_rif_counter_valid_set(rif, dir, true);
+	return 0;
+
+err_counter_edit:
+err_counter_clear:
+	mlxsw_sp_counter_free(mlxsw_sp, MLXSW_SP_COUNTER_SUB_POOL_RIF,
+			      *p_counter_index);
+	return err;
+}
+
+void mlxsw_sp_rif_counter_free(struct mlxsw_sp_rif *rif,
+			       enum mlxsw_sp_rif_counter_dir dir)
+{
+	struct mlxsw_sp *mlxsw_sp = rif->mlxsw_sp;
+	unsigned int *p_counter_index;
+
+	if (!mlxsw_sp_rif_counter_valid_get(rif, dir))
+		return;
+
+	p_counter_index = mlxsw_sp_rif_p_counter_get(rif, dir);
+	if (WARN_ON(!p_counter_index))
+		return;
+	mlxsw_sp_rif_counter_edit(mlxsw_sp, rif->rif_index,
+				  *p_counter_index, false, dir);
+	mlxsw_sp_counter_free(mlxsw_sp, MLXSW_SP_COUNTER_SUB_POOL_RIF,
+			      *p_counter_index);
+	mlxsw_sp_rif_counter_valid_set(rif, dir, false);
+}
+
+static void mlxsw_sp_rif_counters_alloc(struct mlxsw_sp_rif *rif)
+{
+	struct mlxsw_sp *mlxsw_sp = rif->mlxsw_sp;
+	struct devlink *devlink;
+
+	devlink = priv_to_devlink(mlxsw_sp->core);
+	if (!devlink_dpipe_table_counter_enabled(devlink,
+						 MLXSW_SP_DPIPE_TABLE_NAME_ERIF))
+		return;
+	mlxsw_sp_rif_counter_alloc(rif, MLXSW_SP_RIF_COUNTER_EGRESS);
+}
+
+static void mlxsw_sp_rif_counters_free(struct mlxsw_sp_rif *rif)
+{
+	mlxsw_sp_rif_counter_free(rif, MLXSW_SP_RIF_COUNTER_EGRESS);
+}
+
+#define MLXSW_SP_PREFIX_COUNT (sizeof(struct in6_addr) * BITS_PER_BYTE + 1)
+
+struct mlxsw_sp_prefix_usage {
+	DECLARE_BITMAP(b, MLXSW_SP_PREFIX_COUNT);
+};
+
+#define mlxsw_sp_prefix_usage_for_each(prefix, prefix_usage) \
+	for_each_set_bit(prefix, (prefix_usage)->b, MLXSW_SP_PREFIX_COUNT)
+
+static bool
+mlxsw_sp_prefix_usage_eq(struct mlxsw_sp_prefix_usage *prefix_usage1,
+			 struct mlxsw_sp_prefix_usage *prefix_usage2)
+{
+	return !memcmp(prefix_usage1, prefix_usage2, sizeof(*prefix_usage1));
+}
+
+static void
+mlxsw_sp_prefix_usage_cpy(struct mlxsw_sp_prefix_usage *prefix_usage1,
+			  struct mlxsw_sp_prefix_usage *prefix_usage2)
+{
+	memcpy(prefix_usage1, prefix_usage2, sizeof(*prefix_usage1));
+}
+
+static void
+mlxsw_sp_prefix_usage_set(struct mlxsw_sp_prefix_usage *prefix_usage,
+			  unsigned char prefix_len)
+{
+	set_bit(prefix_len, prefix_usage->b);
+}
+
+static void
+mlxsw_sp_prefix_usage_clear(struct mlxsw_sp_prefix_usage *prefix_usage,
+			    unsigned char prefix_len)
+{
+	clear_bit(prefix_len, prefix_usage->b);
+}
+
+struct mlxsw_sp_fib_key {
+	unsigned char addr[sizeof(struct in6_addr)];
+	unsigned char prefix_len;
+};
+
+enum mlxsw_sp_fib_entry_type {
+	MLXSW_SP_FIB_ENTRY_TYPE_REMOTE,
+	MLXSW_SP_FIB_ENTRY_TYPE_LOCAL,
+	MLXSW_SP_FIB_ENTRY_TYPE_TRAP,
+	MLXSW_SP_FIB_ENTRY_TYPE_BLACKHOLE,
+	MLXSW_SP_FIB_ENTRY_TYPE_UNREACHABLE,
+
+	/* This is a special case of local delivery, where a packet should be
+	 * decapsulated on reception. Note that there is no corresponding ENCAP,
+	 * because that's a type of next hop, not of FIB entry. (There can be
+	 * several next hops in a REMOTE entry, and some of them may be
+	 * encapsulating entries.)
+	 */
+	MLXSW_SP_FIB_ENTRY_TYPE_IPIP_DECAP,
+	MLXSW_SP_FIB_ENTRY_TYPE_NVE_DECAP,
+};
+
+struct mlxsw_sp_nexthop_group_info;
+struct mlxsw_sp_nexthop_group;
+struct mlxsw_sp_fib_entry;
+
+struct mlxsw_sp_fib_node {
+	struct mlxsw_sp_fib_entry *fib_entry;
+	struct list_head list;
+	struct rhash_head ht_node;
+	struct mlxsw_sp_fib *fib;
+	struct mlxsw_sp_fib_key key;
+};
+
+struct mlxsw_sp_fib_entry_decap {
+	struct mlxsw_sp_ipip_entry *ipip_entry;
+	u32 tunnel_index;
+};
+
+struct mlxsw_sp_fib_entry {
+	struct mlxsw_sp_fib_node *fib_node;
+	enum mlxsw_sp_fib_entry_type type;
+	struct list_head nexthop_group_node;
+	struct mlxsw_sp_nexthop_group *nh_group;
+	struct mlxsw_sp_fib_entry_decap decap; /* Valid for decap entries. */
+};
+
+struct mlxsw_sp_fib4_entry {
+	struct mlxsw_sp_fib_entry common;
+	struct fib_info *fi;
+	u32 tb_id;
+	dscp_t dscp;
+	u8 type;
+};
+
+struct mlxsw_sp_fib6_entry {
+	struct mlxsw_sp_fib_entry common;
+	struct list_head rt6_list;
+	unsigned int nrt6;
+};
+
+struct mlxsw_sp_rt6 {
+	struct list_head list;
+	struct fib6_info *rt;
+};
+
+struct mlxsw_sp_lpm_tree {
+	u8 id; /* tree ID */
+	unsigned int ref_count;
+	enum mlxsw_sp_l3proto proto;
+	unsigned long prefix_ref_count[MLXSW_SP_PREFIX_COUNT];
+	struct mlxsw_sp_prefix_usage prefix_usage;
+};
+
+struct mlxsw_sp_fib {
+	struct rhashtable ht;
+	struct list_head node_list;
+	struct mlxsw_sp_vr *vr;
+	struct mlxsw_sp_lpm_tree *lpm_tree;
+	enum mlxsw_sp_l3proto proto;
+};
+
+struct mlxsw_sp_vr {
+	u16 id; /* virtual router ID */
+	u32 tb_id; /* kernel fib table id */
+	unsigned int rif_count;
+	struct mlxsw_sp_fib *fib4;
+	struct mlxsw_sp_fib *fib6;
+	struct mlxsw_sp_mr_table *mr_table[MLXSW_SP_L3_PROTO_MAX];
+	struct mlxsw_sp_rif *ul_rif;
+	refcount_t ul_rif_refcnt;
+};
+
+static const struct rhashtable_params mlxsw_sp_fib_ht_params;
+
+static struct mlxsw_sp_fib *mlxsw_sp_fib_create(struct mlxsw_sp *mlxsw_sp,
+						struct mlxsw_sp_vr *vr,
+						enum mlxsw_sp_l3proto proto)
+{
+	struct mlxsw_sp_lpm_tree *lpm_tree;
+	struct mlxsw_sp_fib *fib;
+	int err;
+
+	lpm_tree = mlxsw_sp->router->lpm.proto_trees[proto];
+	fib = kzalloc(sizeof(*fib), GFP_KERNEL);
+	if (!fib)
+		return ERR_PTR(-ENOMEM);
+	err = rhashtable_init(&fib->ht, &mlxsw_sp_fib_ht_params);
+	if (err)
+		goto err_rhashtable_init;
+	INIT_LIST_HEAD(&fib->node_list);
+	fib->proto = proto;
+	fib->vr = vr;
+	fib->lpm_tree = lpm_tree;
+	mlxsw_sp_lpm_tree_hold(lpm_tree);
+	err = mlxsw_sp_vr_lpm_tree_bind(mlxsw_sp, fib, lpm_tree->id);
+	if (err)
+		goto err_lpm_tree_bind;
+	return fib;
+
+err_lpm_tree_bind:
+	mlxsw_sp_lpm_tree_put(mlxsw_sp, lpm_tree);
+err_rhashtable_init:
+	kfree(fib);
+	return ERR_PTR(err);
+}
+
+static void mlxsw_sp_fib_destroy(struct mlxsw_sp *mlxsw_sp,
+				 struct mlxsw_sp_fib *fib)
+{
+	mlxsw_sp_vr_lpm_tree_unbind(mlxsw_sp, fib);
+	mlxsw_sp_lpm_tree_put(mlxsw_sp, fib->lpm_tree);
+	WARN_ON(!list_empty(&fib->node_list));
+	rhashtable_destroy(&fib->ht);
+	kfree(fib);
+}
+
+static struct mlxsw_sp_lpm_tree *
+mlxsw_sp_lpm_tree_find_unused(struct mlxsw_sp *mlxsw_sp)
+{
+	static struct mlxsw_sp_lpm_tree *lpm_tree;
+	int i;
+
+	for (i = 0; i < mlxsw_sp->router->lpm.tree_count; i++) {
+		lpm_tree = &mlxsw_sp->router->lpm.trees[i];
+		if (lpm_tree->ref_count == 0)
+			return lpm_tree;
+	}
+	return NULL;
+}
+
+static int mlxsw_sp_lpm_tree_alloc(struct mlxsw_sp *mlxsw_sp,
+				   struct mlxsw_sp_lpm_tree *lpm_tree)
+{
+	char ralta_pl[MLXSW_REG_RALTA_LEN];
+
+	mlxsw_reg_ralta_pack(ralta_pl, true,
+			     (enum mlxsw_reg_ralxx_protocol) lpm_tree->proto,
+			     lpm_tree->id);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(ralta), ralta_pl);
+}
+
+static void mlxsw_sp_lpm_tree_free(struct mlxsw_sp *mlxsw_sp,
+				   struct mlxsw_sp_lpm_tree *lpm_tree)
+{
+	char ralta_pl[MLXSW_REG_RALTA_LEN];
+
+	mlxsw_reg_ralta_pack(ralta_pl, false,
+			     (enum mlxsw_reg_ralxx_protocol) lpm_tree->proto,
+			     lpm_tree->id);
+	mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(ralta), ralta_pl);
+}
+
+static int
+mlxsw_sp_lpm_tree_left_struct_set(struct mlxsw_sp *mlxsw_sp,
+				  struct mlxsw_sp_prefix_usage *prefix_usage,
+				  struct mlxsw_sp_lpm_tree *lpm_tree)
+{
+	char ralst_pl[MLXSW_REG_RALST_LEN];
+	u8 root_bin = 0;
+	u8 prefix;
+	u8 last_prefix = MLXSW_REG_RALST_BIN_NO_CHILD;
+
+	mlxsw_sp_prefix_usage_for_each(prefix, prefix_usage)
+		root_bin = prefix;
+
+	mlxsw_reg_ralst_pack(ralst_pl, root_bin, lpm_tree->id);
+	mlxsw_sp_prefix_usage_for_each(prefix, prefix_usage) {
+		if (prefix == 0)
+			continue;
+		mlxsw_reg_ralst_bin_pack(ralst_pl, prefix, last_prefix,
+					 MLXSW_REG_RALST_BIN_NO_CHILD);
+		last_prefix = prefix;
+	}
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(ralst), ralst_pl);
+}
+
+static struct mlxsw_sp_lpm_tree *
+mlxsw_sp_lpm_tree_create(struct mlxsw_sp *mlxsw_sp,
+			 struct mlxsw_sp_prefix_usage *prefix_usage,
+			 enum mlxsw_sp_l3proto proto)
+{
+	struct mlxsw_sp_lpm_tree *lpm_tree;
+	int err;
+
+	lpm_tree = mlxsw_sp_lpm_tree_find_unused(mlxsw_sp);
+	if (!lpm_tree)
+		return ERR_PTR(-EBUSY);
+	lpm_tree->proto = proto;
+	err = mlxsw_sp_lpm_tree_alloc(mlxsw_sp, lpm_tree);
+	if (err)
+		return ERR_PTR(err);
+
+	err = mlxsw_sp_lpm_tree_left_struct_set(mlxsw_sp, prefix_usage,
+						lpm_tree);
+	if (err)
+		goto err_left_struct_set;
+	memcpy(&lpm_tree->prefix_usage, prefix_usage,
+	       sizeof(lpm_tree->prefix_usage));
+	memset(&lpm_tree->prefix_ref_count, 0,
+	       sizeof(lpm_tree->prefix_ref_count));
+	lpm_tree->ref_count = 1;
+	return lpm_tree;
+
+err_left_struct_set:
+	mlxsw_sp_lpm_tree_free(mlxsw_sp, lpm_tree);
+	return ERR_PTR(err);
+}
+
+static void mlxsw_sp_lpm_tree_destroy(struct mlxsw_sp *mlxsw_sp,
+				      struct mlxsw_sp_lpm_tree *lpm_tree)
+{
+	mlxsw_sp_lpm_tree_free(mlxsw_sp, lpm_tree);
+}
+
+static struct mlxsw_sp_lpm_tree *
+mlxsw_sp_lpm_tree_get(struct mlxsw_sp *mlxsw_sp,
+		      struct mlxsw_sp_prefix_usage *prefix_usage,
+		      enum mlxsw_sp_l3proto proto)
+{
+	struct mlxsw_sp_lpm_tree *lpm_tree;
+	int i;
+
+	for (i = 0; i < mlxsw_sp->router->lpm.tree_count; i++) {
+		lpm_tree = &mlxsw_sp->router->lpm.trees[i];
+		if (lpm_tree->ref_count != 0 &&
+		    lpm_tree->proto == proto &&
+		    mlxsw_sp_prefix_usage_eq(&lpm_tree->prefix_usage,
+					     prefix_usage)) {
+			mlxsw_sp_lpm_tree_hold(lpm_tree);
+			return lpm_tree;
+		}
+	}
+	return mlxsw_sp_lpm_tree_create(mlxsw_sp, prefix_usage, proto);
+}
+
+static void mlxsw_sp_lpm_tree_hold(struct mlxsw_sp_lpm_tree *lpm_tree)
+{
+	lpm_tree->ref_count++;
+}
+
+static void mlxsw_sp_lpm_tree_put(struct mlxsw_sp *mlxsw_sp,
+				  struct mlxsw_sp_lpm_tree *lpm_tree)
+{
+	if (--lpm_tree->ref_count == 0)
+		mlxsw_sp_lpm_tree_destroy(mlxsw_sp, lpm_tree);
+}
+
+#define MLXSW_SP_LPM_TREE_MIN 1 /* tree 0 is reserved */
+
+static int mlxsw_sp_lpm_init(struct mlxsw_sp *mlxsw_sp)
+{
+	struct mlxsw_sp_prefix_usage req_prefix_usage = {{ 0 } };
+	struct mlxsw_sp_lpm_tree *lpm_tree;
+	u64 max_trees;
+	int err, i;
+
+	if (!MLXSW_CORE_RES_VALID(mlxsw_sp->core, MAX_LPM_TREES))
+		return -EIO;
+
+	max_trees = MLXSW_CORE_RES_GET(mlxsw_sp->core, MAX_LPM_TREES);
+	mlxsw_sp->router->lpm.tree_count = max_trees - MLXSW_SP_LPM_TREE_MIN;
+	mlxsw_sp->router->lpm.trees = kcalloc(mlxsw_sp->router->lpm.tree_count,
+					     sizeof(struct mlxsw_sp_lpm_tree),
+					     GFP_KERNEL);
+	if (!mlxsw_sp->router->lpm.trees)
+		return -ENOMEM;
+
+	for (i = 0; i < mlxsw_sp->router->lpm.tree_count; i++) {
+		lpm_tree = &mlxsw_sp->router->lpm.trees[i];
+		lpm_tree->id = i + MLXSW_SP_LPM_TREE_MIN;
+	}
+
+	lpm_tree = mlxsw_sp_lpm_tree_get(mlxsw_sp, &req_prefix_usage,
+					 MLXSW_SP_L3_PROTO_IPV4);
+	if (IS_ERR(lpm_tree)) {
+		err = PTR_ERR(lpm_tree);
+		goto err_ipv4_tree_get;
+	}
+	mlxsw_sp->router->lpm.proto_trees[MLXSW_SP_L3_PROTO_IPV4] = lpm_tree;
+
+	lpm_tree = mlxsw_sp_lpm_tree_get(mlxsw_sp, &req_prefix_usage,
+					 MLXSW_SP_L3_PROTO_IPV6);
+	if (IS_ERR(lpm_tree)) {
+		err = PTR_ERR(lpm_tree);
+		goto err_ipv6_tree_get;
+	}
+	mlxsw_sp->router->lpm.proto_trees[MLXSW_SP_L3_PROTO_IPV6] = lpm_tree;
+
+	return 0;
+
+err_ipv6_tree_get:
+	lpm_tree = mlxsw_sp->router->lpm.proto_trees[MLXSW_SP_L3_PROTO_IPV4];
+	mlxsw_sp_lpm_tree_put(mlxsw_sp, lpm_tree);
+err_ipv4_tree_get:
+	kfree(mlxsw_sp->router->lpm.trees);
+	return err;
+}
+
+static void mlxsw_sp_lpm_fini(struct mlxsw_sp *mlxsw_sp)
+{
+	struct mlxsw_sp_lpm_tree *lpm_tree;
+
+	lpm_tree = mlxsw_sp->router->lpm.proto_trees[MLXSW_SP_L3_PROTO_IPV6];
+	mlxsw_sp_lpm_tree_put(mlxsw_sp, lpm_tree);
+
+	lpm_tree = mlxsw_sp->router->lpm.proto_trees[MLXSW_SP_L3_PROTO_IPV4];
+	mlxsw_sp_lpm_tree_put(mlxsw_sp, lpm_tree);
+
+	kfree(mlxsw_sp->router->lpm.trees);
+}
+
+static bool mlxsw_sp_vr_is_used(const struct mlxsw_sp_vr *vr)
+{
+	return !!vr->fib4 || !!vr->fib6 ||
+	       !!vr->mr_table[MLXSW_SP_L3_PROTO_IPV4] ||
+	       !!vr->mr_table[MLXSW_SP_L3_PROTO_IPV6];
+}
+
+static struct mlxsw_sp_vr *mlxsw_sp_vr_find_unused(struct mlxsw_sp *mlxsw_sp)
+{
+	struct mlxsw_sp_vr *vr;
+	int i;
+
+	for (i = 0; i < MLXSW_CORE_RES_GET(mlxsw_sp->core, MAX_VRS); i++) {
+		vr = &mlxsw_sp->router->vrs[i];
+		if (!mlxsw_sp_vr_is_used(vr))
+			return vr;
+	}
+	return NULL;
+}
+
+static int mlxsw_sp_vr_lpm_tree_bind(struct mlxsw_sp *mlxsw_sp,
+				     const struct mlxsw_sp_fib *fib, u8 tree_id)
+{
+	char raltb_pl[MLXSW_REG_RALTB_LEN];
+
+	mlxsw_reg_raltb_pack(raltb_pl, fib->vr->id,
+			     (enum mlxsw_reg_ralxx_protocol) fib->proto,
+			     tree_id);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(raltb), raltb_pl);
+}
+
+static int mlxsw_sp_vr_lpm_tree_unbind(struct mlxsw_sp *mlxsw_sp,
+				       const struct mlxsw_sp_fib *fib)
+{
+	char raltb_pl[MLXSW_REG_RALTB_LEN];
+
+	/* Bind to tree 0 which is default */
+	mlxsw_reg_raltb_pack(raltb_pl, fib->vr->id,
+			     (enum mlxsw_reg_ralxx_protocol) fib->proto, 0);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(raltb), raltb_pl);
+}
+
+static u32 mlxsw_sp_fix_tb_id(u32 tb_id)
+{
+	/* For our purpose, squash main, default and local tables into one */
+	if (tb_id == RT_TABLE_LOCAL || tb_id == RT_TABLE_DEFAULT)
+		tb_id = RT_TABLE_MAIN;
+	return tb_id;
+}
+
+static struct mlxsw_sp_vr *mlxsw_sp_vr_find(struct mlxsw_sp *mlxsw_sp,
+					    u32 tb_id)
+{
+	struct mlxsw_sp_vr *vr;
+	int i;
+
+	tb_id = mlxsw_sp_fix_tb_id(tb_id);
+
+	for (i = 0; i < MLXSW_CORE_RES_GET(mlxsw_sp->core, MAX_VRS); i++) {
+		vr = &mlxsw_sp->router->vrs[i];
+		if (mlxsw_sp_vr_is_used(vr) && vr->tb_id == tb_id)
+			return vr;
+	}
+	return NULL;
+}
+
+int mlxsw_sp_router_tb_id_vr_id(struct mlxsw_sp *mlxsw_sp, u32 tb_id,
+				u16 *vr_id)
+{
+	struct mlxsw_sp_vr *vr;
+	int err = 0;
+
+	mutex_lock(&mlxsw_sp->router->lock);
+	vr = mlxsw_sp_vr_find(mlxsw_sp, tb_id);
+	if (!vr) {
+		err = -ESRCH;
+		goto out;
+	}
+	*vr_id = vr->id;
+out:
+	mutex_unlock(&mlxsw_sp->router->lock);
+	return err;
+}
+
+static struct mlxsw_sp_fib *mlxsw_sp_vr_fib(const struct mlxsw_sp_vr *vr,
+					    enum mlxsw_sp_l3proto proto)
+{
+	switch (proto) {
+	case MLXSW_SP_L3_PROTO_IPV4:
+		return vr->fib4;
+	case MLXSW_SP_L3_PROTO_IPV6:
+		return vr->fib6;
+	}
+	return NULL;
+}
+
+static struct mlxsw_sp_vr *mlxsw_sp_vr_create(struct mlxsw_sp *mlxsw_sp,
+					      u32 tb_id,
+					      struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp_mr_table *mr4_table, *mr6_table;
+	struct mlxsw_sp_fib *fib4;
+	struct mlxsw_sp_fib *fib6;
+	struct mlxsw_sp_vr *vr;
+	int err;
+
+	vr = mlxsw_sp_vr_find_unused(mlxsw_sp);
+	if (!vr) {
+		NL_SET_ERR_MSG_MOD(extack, "Exceeded number of supported virtual routers");
+		return ERR_PTR(-EBUSY);
+	}
+	fib4 = mlxsw_sp_fib_create(mlxsw_sp, vr, MLXSW_SP_L3_PROTO_IPV4);
+	if (IS_ERR(fib4))
+		return ERR_CAST(fib4);
+	fib6 = mlxsw_sp_fib_create(mlxsw_sp, vr, MLXSW_SP_L3_PROTO_IPV6);
+	if (IS_ERR(fib6)) {
+		err = PTR_ERR(fib6);
+		goto err_fib6_create;
+	}
+	mr4_table = mlxsw_sp_mr_table_create(mlxsw_sp, vr->id,
+					     MLXSW_SP_L3_PROTO_IPV4);
+	if (IS_ERR(mr4_table)) {
+		err = PTR_ERR(mr4_table);
+		goto err_mr4_table_create;
+	}
+	mr6_table = mlxsw_sp_mr_table_create(mlxsw_sp, vr->id,
+					     MLXSW_SP_L3_PROTO_IPV6);
+	if (IS_ERR(mr6_table)) {
+		err = PTR_ERR(mr6_table);
+		goto err_mr6_table_create;
+	}
+
+	vr->fib4 = fib4;
+	vr->fib6 = fib6;
+	vr->mr_table[MLXSW_SP_L3_PROTO_IPV4] = mr4_table;
+	vr->mr_table[MLXSW_SP_L3_PROTO_IPV6] = mr6_table;
+	vr->tb_id = tb_id;
+	return vr;
+
+err_mr6_table_create:
+	mlxsw_sp_mr_table_destroy(mr4_table);
+err_mr4_table_create:
+	mlxsw_sp_fib_destroy(mlxsw_sp, fib6);
+err_fib6_create:
+	mlxsw_sp_fib_destroy(mlxsw_sp, fib4);
+	return ERR_PTR(err);
+}
+
+static void mlxsw_sp_vr_destroy(struct mlxsw_sp *mlxsw_sp,
+				struct mlxsw_sp_vr *vr)
+{
+	mlxsw_sp_mr_table_destroy(vr->mr_table[MLXSW_SP_L3_PROTO_IPV6]);
+	vr->mr_table[MLXSW_SP_L3_PROTO_IPV6] = NULL;
+	mlxsw_sp_mr_table_destroy(vr->mr_table[MLXSW_SP_L3_PROTO_IPV4]);
+	vr->mr_table[MLXSW_SP_L3_PROTO_IPV4] = NULL;
+	mlxsw_sp_fib_destroy(mlxsw_sp, vr->fib6);
+	vr->fib6 = NULL;
+	mlxsw_sp_fib_destroy(mlxsw_sp, vr->fib4);
+	vr->fib4 = NULL;
+}
+
+static struct mlxsw_sp_vr *mlxsw_sp_vr_get(struct mlxsw_sp *mlxsw_sp, u32 tb_id,
+					   struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp_vr *vr;
+
+	tb_id = mlxsw_sp_fix_tb_id(tb_id);
+	vr = mlxsw_sp_vr_find(mlxsw_sp, tb_id);
+	if (!vr)
+		vr = mlxsw_sp_vr_create(mlxsw_sp, tb_id, extack);
+	return vr;
+}
+
+static void mlxsw_sp_vr_put(struct mlxsw_sp *mlxsw_sp, struct mlxsw_sp_vr *vr)
+{
+	if (!vr->rif_count && list_empty(&vr->fib4->node_list) &&
+	    list_empty(&vr->fib6->node_list) &&
+	    mlxsw_sp_mr_table_empty(vr->mr_table[MLXSW_SP_L3_PROTO_IPV4]) &&
+	    mlxsw_sp_mr_table_empty(vr->mr_table[MLXSW_SP_L3_PROTO_IPV6]))
+		mlxsw_sp_vr_destroy(mlxsw_sp, vr);
+}
+
+static bool
+mlxsw_sp_vr_lpm_tree_should_replace(struct mlxsw_sp_vr *vr,
+				    enum mlxsw_sp_l3proto proto, u8 tree_id)
+{
+	struct mlxsw_sp_fib *fib = mlxsw_sp_vr_fib(vr, proto);
+
+	if (!mlxsw_sp_vr_is_used(vr))
+		return false;
+	if (fib->lpm_tree->id == tree_id)
+		return true;
+	return false;
+}
+
+static int mlxsw_sp_vr_lpm_tree_replace(struct mlxsw_sp *mlxsw_sp,
+					struct mlxsw_sp_fib *fib,
+					struct mlxsw_sp_lpm_tree *new_tree)
+{
+	struct mlxsw_sp_lpm_tree *old_tree = fib->lpm_tree;
+	int err;
+
+	fib->lpm_tree = new_tree;
+	mlxsw_sp_lpm_tree_hold(new_tree);
+	err = mlxsw_sp_vr_lpm_tree_bind(mlxsw_sp, fib, new_tree->id);
+	if (err)
+		goto err_tree_bind;
+	mlxsw_sp_lpm_tree_put(mlxsw_sp, old_tree);
+	return 0;
+
+err_tree_bind:
+	mlxsw_sp_lpm_tree_put(mlxsw_sp, new_tree);
+	fib->lpm_tree = old_tree;
+	return err;
+}
+
+static int mlxsw_sp_vrs_lpm_tree_replace(struct mlxsw_sp *mlxsw_sp,
+					 struct mlxsw_sp_fib *fib,
+					 struct mlxsw_sp_lpm_tree *new_tree)
+{
+	enum mlxsw_sp_l3proto proto = fib->proto;
+	struct mlxsw_sp_lpm_tree *old_tree;
+	u8 old_id, new_id = new_tree->id;
+	struct mlxsw_sp_vr *vr;
+	int i, err;
+
+	old_tree = mlxsw_sp->router->lpm.proto_trees[proto];
+	old_id = old_tree->id;
+
+	for (i = 0; i < MLXSW_CORE_RES_GET(mlxsw_sp->core, MAX_VRS); i++) {
+		vr = &mlxsw_sp->router->vrs[i];
+		if (!mlxsw_sp_vr_lpm_tree_should_replace(vr, proto, old_id))
+			continue;
+		err = mlxsw_sp_vr_lpm_tree_replace(mlxsw_sp,
+						   mlxsw_sp_vr_fib(vr, proto),
+						   new_tree);
+		if (err)
+			goto err_tree_replace;
+	}
+
+	memcpy(new_tree->prefix_ref_count, old_tree->prefix_ref_count,
+	       sizeof(new_tree->prefix_ref_count));
+	mlxsw_sp->router->lpm.proto_trees[proto] = new_tree;
+	mlxsw_sp_lpm_tree_put(mlxsw_sp, old_tree);
+
+	return 0;
+
+err_tree_replace:
+	for (i--; i >= 0; i--) {
+		if (!mlxsw_sp_vr_lpm_tree_should_replace(vr, proto, new_id))
+			continue;
+		mlxsw_sp_vr_lpm_tree_replace(mlxsw_sp,
+					     mlxsw_sp_vr_fib(vr, proto),
+					     old_tree);
+	}
+	return err;
+}
+
+static int mlxsw_sp_vrs_init(struct mlxsw_sp *mlxsw_sp)
+{
+	struct mlxsw_sp_vr *vr;
+	u64 max_vrs;
+	int i;
+
+	if (!MLXSW_CORE_RES_VALID(mlxsw_sp->core, MAX_VRS))
+		return -EIO;
+
+	max_vrs = MLXSW_CORE_RES_GET(mlxsw_sp->core, MAX_VRS);
+	mlxsw_sp->router->vrs = kcalloc(max_vrs, sizeof(struct mlxsw_sp_vr),
+					GFP_KERNEL);
+	if (!mlxsw_sp->router->vrs)
+		return -ENOMEM;
+
+	for (i = 0; i < max_vrs; i++) {
+		vr = &mlxsw_sp->router->vrs[i];
+		vr->id = i;
+	}
+
+	return 0;
+}
+
+static void mlxsw_sp_router_fib_flush(struct mlxsw_sp *mlxsw_sp);
+
+static void mlxsw_sp_vrs_fini(struct mlxsw_sp *mlxsw_sp)
+{
+	/* At this stage we're guaranteed not to have new incoming
+	 * FIB notifications and the work queue is free from FIBs
+	 * sitting on top of mlxsw netdevs. However, we can still
+	 * have other FIBs queued. Flush the queue before flushing
+	 * the device's tables. No need for locks, as we're the only
+	 * writer.
+	 */
+	mlxsw_core_flush_owq();
+	mlxsw_sp_router_fib_flush(mlxsw_sp);
+	kfree(mlxsw_sp->router->vrs);
+}
+
+u32 mlxsw_sp_ipip_dev_ul_tb_id(const struct net_device *ol_dev)
+{
+	struct net_device *d;
+	u32 tb_id;
+
+	rcu_read_lock();
+	d = mlxsw_sp_ipip_netdev_ul_dev_get(ol_dev);
+	if (d)
+		tb_id = l3mdev_fib_table(d) ? : RT_TABLE_MAIN;
+	else
+		tb_id = RT_TABLE_MAIN;
+	rcu_read_unlock();
+
+	return tb_id;
+}
+
+static struct mlxsw_sp_rif *
+mlxsw_sp_rif_create(struct mlxsw_sp *mlxsw_sp,
+		    const struct mlxsw_sp_rif_params *params,
+		    struct netlink_ext_ack *extack);
+
+static struct mlxsw_sp_rif_ipip_lb *
+mlxsw_sp_ipip_ol_ipip_lb_create(struct mlxsw_sp *mlxsw_sp,
+				enum mlxsw_sp_ipip_type ipipt,
+				struct net_device *ol_dev,
+				struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp_rif_params_ipip_lb lb_params;
+	const struct mlxsw_sp_ipip_ops *ipip_ops;
+	struct mlxsw_sp_rif *rif;
+
+	ipip_ops = mlxsw_sp->router->ipip_ops_arr[ipipt];
+	lb_params = (struct mlxsw_sp_rif_params_ipip_lb) {
+		.common.dev = ol_dev,
+		.common.lag = false,
+		.lb_config = ipip_ops->ol_loopback_config(mlxsw_sp, ol_dev),
+	};
+
+	rif = mlxsw_sp_rif_create(mlxsw_sp, &lb_params.common, extack);
+	if (IS_ERR(rif))
+		return ERR_CAST(rif);
+	return container_of(rif, struct mlxsw_sp_rif_ipip_lb, common);
+}
+
+static struct mlxsw_sp_ipip_entry *
+mlxsw_sp_ipip_entry_alloc(struct mlxsw_sp *mlxsw_sp,
+			  enum mlxsw_sp_ipip_type ipipt,
+			  struct net_device *ol_dev)
+{
+	const struct mlxsw_sp_ipip_ops *ipip_ops;
+	struct mlxsw_sp_ipip_entry *ipip_entry;
+	struct mlxsw_sp_ipip_entry *ret = NULL;
+	int err;
+
+	ipip_ops = mlxsw_sp->router->ipip_ops_arr[ipipt];
+	ipip_entry = kzalloc(sizeof(*ipip_entry), GFP_KERNEL);
+	if (!ipip_entry)
+		return ERR_PTR(-ENOMEM);
+
+	ipip_entry->ol_lb = mlxsw_sp_ipip_ol_ipip_lb_create(mlxsw_sp, ipipt,
+							    ol_dev, NULL);
+	if (IS_ERR(ipip_entry->ol_lb)) {
+		ret = ERR_CAST(ipip_entry->ol_lb);
+		goto err_ol_ipip_lb_create;
+	}
+
+	ipip_entry->ipipt = ipipt;
+	ipip_entry->ol_dev = ol_dev;
+	ipip_entry->parms = ipip_ops->parms_init(ol_dev);
+
+	err = ipip_ops->rem_ip_addr_set(mlxsw_sp, ipip_entry);
+	if (err) {
+		ret = ERR_PTR(err);
+		goto err_rem_ip_addr_set;
+	}
+
+	return ipip_entry;
+
+err_rem_ip_addr_set:
+	mlxsw_sp_rif_destroy(&ipip_entry->ol_lb->common);
+err_ol_ipip_lb_create:
+	kfree(ipip_entry);
+	return ret;
+}
+
+static void mlxsw_sp_ipip_entry_dealloc(struct mlxsw_sp *mlxsw_sp,
+					struct mlxsw_sp_ipip_entry *ipip_entry)
+{
+	const struct mlxsw_sp_ipip_ops *ipip_ops =
+		mlxsw_sp->router->ipip_ops_arr[ipip_entry->ipipt];
+
+	ipip_ops->rem_ip_addr_unset(mlxsw_sp, ipip_entry);
+	mlxsw_sp_rif_destroy(&ipip_entry->ol_lb->common);
+	kfree(ipip_entry);
+}
+
+static bool
+mlxsw_sp_ipip_entry_saddr_matches(struct mlxsw_sp *mlxsw_sp,
+				  const enum mlxsw_sp_l3proto ul_proto,
+				  union mlxsw_sp_l3addr saddr,
+				  u32 ul_tb_id,
+				  struct mlxsw_sp_ipip_entry *ipip_entry)
+{
+	u32 tun_ul_tb_id = mlxsw_sp_ipip_dev_ul_tb_id(ipip_entry->ol_dev);
+	enum mlxsw_sp_ipip_type ipipt = ipip_entry->ipipt;
+	union mlxsw_sp_l3addr tun_saddr;
+
+	if (mlxsw_sp->router->ipip_ops_arr[ipipt]->ul_proto != ul_proto)
+		return false;
+
+	tun_saddr = mlxsw_sp_ipip_netdev_saddr(ul_proto, ipip_entry->ol_dev);
+	return tun_ul_tb_id == ul_tb_id &&
+	       mlxsw_sp_l3addr_eq(&tun_saddr, &saddr);
+}
+
+static int mlxsw_sp_ipip_decap_parsing_depth_inc(struct mlxsw_sp *mlxsw_sp,
+						 enum mlxsw_sp_ipip_type ipipt)
+{
+	const struct mlxsw_sp_ipip_ops *ipip_ops;
+
+	ipip_ops = mlxsw_sp->router->ipip_ops_arr[ipipt];
+
+	/* Not all tunnels require to increase the default pasing depth
+	 * (96 bytes).
+	 */
+	if (ipip_ops->inc_parsing_depth)
+		return mlxsw_sp_parsing_depth_inc(mlxsw_sp);
+
+	return 0;
+}
+
+static void mlxsw_sp_ipip_decap_parsing_depth_dec(struct mlxsw_sp *mlxsw_sp,
+						  enum mlxsw_sp_ipip_type ipipt)
+{
+	const struct mlxsw_sp_ipip_ops *ipip_ops =
+		mlxsw_sp->router->ipip_ops_arr[ipipt];
+
+	if (ipip_ops->inc_parsing_depth)
+		mlxsw_sp_parsing_depth_dec(mlxsw_sp);
+}
+
+static int
+mlxsw_sp_fib_entry_decap_init(struct mlxsw_sp *mlxsw_sp,
+			      struct mlxsw_sp_fib_entry *fib_entry,
+			      struct mlxsw_sp_ipip_entry *ipip_entry)
+{
+	u32 tunnel_index;
+	int err;
+
+	err = mlxsw_sp_kvdl_alloc(mlxsw_sp, MLXSW_SP_KVDL_ENTRY_TYPE_ADJ,
+				  1, &tunnel_index);
+	if (err)
+		return err;
+
+	err = mlxsw_sp_ipip_decap_parsing_depth_inc(mlxsw_sp,
+						    ipip_entry->ipipt);
+	if (err)
+		goto err_parsing_depth_inc;
+
+	ipip_entry->decap_fib_entry = fib_entry;
+	fib_entry->decap.ipip_entry = ipip_entry;
+	fib_entry->decap.tunnel_index = tunnel_index;
+
+	return 0;
+
+err_parsing_depth_inc:
+	mlxsw_sp_kvdl_free(mlxsw_sp, MLXSW_SP_KVDL_ENTRY_TYPE_ADJ, 1,
+			   fib_entry->decap.tunnel_index);
+	return err;
+}
+
+static void mlxsw_sp_fib_entry_decap_fini(struct mlxsw_sp *mlxsw_sp,
+					  struct mlxsw_sp_fib_entry *fib_entry)
+{
+	enum mlxsw_sp_ipip_type ipipt = fib_entry->decap.ipip_entry->ipipt;
+
+	/* Unlink this node from the IPIP entry that it's the decap entry of. */
+	fib_entry->decap.ipip_entry->decap_fib_entry = NULL;
+	fib_entry->decap.ipip_entry = NULL;
+	mlxsw_sp_ipip_decap_parsing_depth_dec(mlxsw_sp, ipipt);
+	mlxsw_sp_kvdl_free(mlxsw_sp, MLXSW_SP_KVDL_ENTRY_TYPE_ADJ,
+			   1, fib_entry->decap.tunnel_index);
+}
+
+static struct mlxsw_sp_fib_node *
+mlxsw_sp_fib_node_lookup(struct mlxsw_sp_fib *fib, const void *addr,
+			 size_t addr_len, unsigned char prefix_len);
+static int mlxsw_sp_fib_entry_update(struct mlxsw_sp *mlxsw_sp,
+				     struct mlxsw_sp_fib_entry *fib_entry);
+
+static void
+mlxsw_sp_ipip_entry_demote_decap(struct mlxsw_sp *mlxsw_sp,
+				 struct mlxsw_sp_ipip_entry *ipip_entry)
+{
+	struct mlxsw_sp_fib_entry *fib_entry = ipip_entry->decap_fib_entry;
+
+	mlxsw_sp_fib_entry_decap_fini(mlxsw_sp, fib_entry);
+	fib_entry->type = MLXSW_SP_FIB_ENTRY_TYPE_TRAP;
+
+	mlxsw_sp_fib_entry_update(mlxsw_sp, fib_entry);
+}
+
+static void
+mlxsw_sp_ipip_entry_promote_decap(struct mlxsw_sp *mlxsw_sp,
+				  struct mlxsw_sp_ipip_entry *ipip_entry,
+				  struct mlxsw_sp_fib_entry *decap_fib_entry)
+{
+	if (mlxsw_sp_fib_entry_decap_init(mlxsw_sp, decap_fib_entry,
+					  ipip_entry))
+		return;
+	decap_fib_entry->type = MLXSW_SP_FIB_ENTRY_TYPE_IPIP_DECAP;
+
+	if (mlxsw_sp_fib_entry_update(mlxsw_sp, decap_fib_entry))
+		mlxsw_sp_ipip_entry_demote_decap(mlxsw_sp, ipip_entry);
+}
+
+static struct mlxsw_sp_fib_entry *
+mlxsw_sp_router_ip2me_fib_entry_find(struct mlxsw_sp *mlxsw_sp, u32 tb_id,
+				     enum mlxsw_sp_l3proto proto,
+				     const union mlxsw_sp_l3addr *addr,
+				     enum mlxsw_sp_fib_entry_type type)
+{
+	struct mlxsw_sp_fib_node *fib_node;
+	unsigned char addr_prefix_len;
+	struct mlxsw_sp_fib *fib;
+	struct mlxsw_sp_vr *vr;
+	const void *addrp;
+	size_t addr_len;
+	u32 addr4;
+
+	vr = mlxsw_sp_vr_find(mlxsw_sp, tb_id);
+	if (!vr)
+		return NULL;
+	fib = mlxsw_sp_vr_fib(vr, proto);
+
+	switch (proto) {
+	case MLXSW_SP_L3_PROTO_IPV4:
+		addr4 = be32_to_cpu(addr->addr4);
+		addrp = &addr4;
+		addr_len = 4;
+		addr_prefix_len = 32;
+		break;
+	case MLXSW_SP_L3_PROTO_IPV6:
+		addrp = &addr->addr6;
+		addr_len = 16;
+		addr_prefix_len = 128;
+		break;
+	default:
+		WARN_ON(1);
+		return NULL;
+	}
+
+	fib_node = mlxsw_sp_fib_node_lookup(fib, addrp, addr_len,
+					    addr_prefix_len);
+	if (!fib_node || fib_node->fib_entry->type != type)
+		return NULL;
+
+	return fib_node->fib_entry;
+}
+
+/* Given an IPIP entry, find the corresponding decap route. */
+static struct mlxsw_sp_fib_entry *
+mlxsw_sp_ipip_entry_find_decap(struct mlxsw_sp *mlxsw_sp,
+			       struct mlxsw_sp_ipip_entry *ipip_entry)
+{
+	static struct mlxsw_sp_fib_node *fib_node;
+	const struct mlxsw_sp_ipip_ops *ipip_ops;
+	unsigned char saddr_prefix_len;
+	union mlxsw_sp_l3addr saddr;
+	struct mlxsw_sp_fib *ul_fib;
+	struct mlxsw_sp_vr *ul_vr;
+	const void *saddrp;
+	size_t saddr_len;
+	u32 ul_tb_id;
+	u32 saddr4;
+
+	ipip_ops = mlxsw_sp->router->ipip_ops_arr[ipip_entry->ipipt];
+
+	ul_tb_id = mlxsw_sp_ipip_dev_ul_tb_id(ipip_entry->ol_dev);
+	ul_vr = mlxsw_sp_vr_find(mlxsw_sp, ul_tb_id);
+	if (!ul_vr)
+		return NULL;
+
+	ul_fib = mlxsw_sp_vr_fib(ul_vr, ipip_ops->ul_proto);
+	saddr = mlxsw_sp_ipip_netdev_saddr(ipip_ops->ul_proto,
+					   ipip_entry->ol_dev);
+
+	switch (ipip_ops->ul_proto) {
+	case MLXSW_SP_L3_PROTO_IPV4:
+		saddr4 = be32_to_cpu(saddr.addr4);
+		saddrp = &saddr4;
+		saddr_len = 4;
+		saddr_prefix_len = 32;
+		break;
+	case MLXSW_SP_L3_PROTO_IPV6:
+		saddrp = &saddr.addr6;
+		saddr_len = 16;
+		saddr_prefix_len = 128;
+		break;
+	default:
+		WARN_ON(1);
+		return NULL;
+	}
+
+	fib_node = mlxsw_sp_fib_node_lookup(ul_fib, saddrp, saddr_len,
+					    saddr_prefix_len);
+	if (!fib_node ||
+	    fib_node->fib_entry->type != MLXSW_SP_FIB_ENTRY_TYPE_TRAP)
+		return NULL;
+
+	return fib_node->fib_entry;
+}
+
+static struct mlxsw_sp_ipip_entry *
+mlxsw_sp_ipip_entry_create(struct mlxsw_sp *mlxsw_sp,
+			   enum mlxsw_sp_ipip_type ipipt,
+			   struct net_device *ol_dev)
+{
+	struct mlxsw_sp_ipip_entry *ipip_entry;
+
+	ipip_entry = mlxsw_sp_ipip_entry_alloc(mlxsw_sp, ipipt, ol_dev);
+	if (IS_ERR(ipip_entry))
+		return ipip_entry;
+
+	list_add_tail(&ipip_entry->ipip_list_node,
+		      &mlxsw_sp->router->ipip_list);
+
+	return ipip_entry;
+}
+
+static void
+mlxsw_sp_ipip_entry_destroy(struct mlxsw_sp *mlxsw_sp,
+			    struct mlxsw_sp_ipip_entry *ipip_entry)
+{
+	list_del(&ipip_entry->ipip_list_node);
+	mlxsw_sp_ipip_entry_dealloc(mlxsw_sp, ipip_entry);
+}
+
+static bool
+mlxsw_sp_ipip_entry_matches_decap(struct mlxsw_sp *mlxsw_sp,
+				  const struct net_device *ul_dev,
+				  enum mlxsw_sp_l3proto ul_proto,
+				  union mlxsw_sp_l3addr ul_dip,
+				  struct mlxsw_sp_ipip_entry *ipip_entry)
+{
+	u32 ul_tb_id = l3mdev_fib_table(ul_dev) ? : RT_TABLE_MAIN;
+	enum mlxsw_sp_ipip_type ipipt = ipip_entry->ipipt;
+
+	if (mlxsw_sp->router->ipip_ops_arr[ipipt]->ul_proto != ul_proto)
+		return false;
+
+	return mlxsw_sp_ipip_entry_saddr_matches(mlxsw_sp, ul_proto, ul_dip,
+						 ul_tb_id, ipip_entry);
+}
+
+/* Given decap parameters, find the corresponding IPIP entry. */
+static struct mlxsw_sp_ipip_entry *
+mlxsw_sp_ipip_entry_find_by_decap(struct mlxsw_sp *mlxsw_sp, int ul_dev_ifindex,
+				  enum mlxsw_sp_l3proto ul_proto,
+				  union mlxsw_sp_l3addr ul_dip)
+{
+	struct mlxsw_sp_ipip_entry *ipip_entry = NULL;
+	struct net_device *ul_dev;
+
+	rcu_read_lock();
+
+	ul_dev = dev_get_by_index_rcu(mlxsw_sp_net(mlxsw_sp), ul_dev_ifindex);
+	if (!ul_dev)
+		goto out_unlock;
+
+	list_for_each_entry(ipip_entry, &mlxsw_sp->router->ipip_list,
+			    ipip_list_node)
+		if (mlxsw_sp_ipip_entry_matches_decap(mlxsw_sp, ul_dev,
+						      ul_proto, ul_dip,
+						      ipip_entry))
+			goto out_unlock;
+
+	rcu_read_unlock();
+
+	return NULL;
+
+out_unlock:
+	rcu_read_unlock();
+	return ipip_entry;
+}
+
+static bool mlxsw_sp_netdev_ipip_type(const struct mlxsw_sp *mlxsw_sp,
+				      const struct net_device *dev,
+				      enum mlxsw_sp_ipip_type *p_type)
+{
+	struct mlxsw_sp_router *router = mlxsw_sp->router;
+	const struct mlxsw_sp_ipip_ops *ipip_ops;
+	enum mlxsw_sp_ipip_type ipipt;
+
+	for (ipipt = 0; ipipt < MLXSW_SP_IPIP_TYPE_MAX; ++ipipt) {
+		ipip_ops = router->ipip_ops_arr[ipipt];
+		if (dev->type == ipip_ops->dev_type) {
+			if (p_type)
+				*p_type = ipipt;
+			return true;
+		}
+	}
+	return false;
+}
+
+static bool mlxsw_sp_netdev_is_ipip_ol(const struct mlxsw_sp *mlxsw_sp,
+				       const struct net_device *dev)
+{
+	return mlxsw_sp_netdev_ipip_type(mlxsw_sp, dev, NULL);
+}
+
+static struct mlxsw_sp_ipip_entry *
+mlxsw_sp_ipip_entry_find_by_ol_dev(struct mlxsw_sp *mlxsw_sp,
+				   const struct net_device *ol_dev)
+{
+	struct mlxsw_sp_ipip_entry *ipip_entry;
+
+	list_for_each_entry(ipip_entry, &mlxsw_sp->router->ipip_list,
+			    ipip_list_node)
+		if (ipip_entry->ol_dev == ol_dev)
+			return ipip_entry;
+
+	return NULL;
+}
+
+static struct mlxsw_sp_ipip_entry *
+mlxsw_sp_ipip_entry_find_by_ul_dev(const struct mlxsw_sp *mlxsw_sp,
+				   const struct net_device *ul_dev,
+				   struct mlxsw_sp_ipip_entry *start)
+{
+	struct mlxsw_sp_ipip_entry *ipip_entry;
+
+	ipip_entry = list_prepare_entry(start, &mlxsw_sp->router->ipip_list,
+					ipip_list_node);
+	list_for_each_entry_continue(ipip_entry, &mlxsw_sp->router->ipip_list,
+				     ipip_list_node) {
+		struct net_device *ol_dev = ipip_entry->ol_dev;
+		struct net_device *ipip_ul_dev;
+
+		rcu_read_lock();
+		ipip_ul_dev = mlxsw_sp_ipip_netdev_ul_dev_get(ol_dev);
+		rcu_read_unlock();
+
+		if (ipip_ul_dev == ul_dev)
+			return ipip_entry;
+	}
+
+	return NULL;
+}
+
+static bool mlxsw_sp_netdev_is_ipip_ul(struct mlxsw_sp *mlxsw_sp,
+				       const struct net_device *dev)
+{
+	return mlxsw_sp_ipip_entry_find_by_ul_dev(mlxsw_sp, dev, NULL);
+}
+
+static bool mlxsw_sp_netdevice_ipip_can_offload(struct mlxsw_sp *mlxsw_sp,
+						const struct net_device *ol_dev,
+						enum mlxsw_sp_ipip_type ipipt)
+{
+	const struct mlxsw_sp_ipip_ops *ops
+		= mlxsw_sp->router->ipip_ops_arr[ipipt];
+
+	return ops->can_offload(mlxsw_sp, ol_dev);
+}
+
+static int mlxsw_sp_netdevice_ipip_ol_reg_event(struct mlxsw_sp *mlxsw_sp,
+						struct net_device *ol_dev)
+{
+	enum mlxsw_sp_ipip_type ipipt = MLXSW_SP_IPIP_TYPE_MAX;
+	struct mlxsw_sp_ipip_entry *ipip_entry;
+	enum mlxsw_sp_l3proto ul_proto;
+	union mlxsw_sp_l3addr saddr;
+	u32 ul_tb_id;
+
+	mlxsw_sp_netdev_ipip_type(mlxsw_sp, ol_dev, &ipipt);
+	if (mlxsw_sp_netdevice_ipip_can_offload(mlxsw_sp, ol_dev, ipipt)) {
+		ul_tb_id = mlxsw_sp_ipip_dev_ul_tb_id(ol_dev);
+		ul_proto = mlxsw_sp->router->ipip_ops_arr[ipipt]->ul_proto;
+		saddr = mlxsw_sp_ipip_netdev_saddr(ul_proto, ol_dev);
+		if (!mlxsw_sp_ipip_demote_tunnel_by_saddr(mlxsw_sp, ul_proto,
+							  saddr, ul_tb_id,
+							  NULL)) {
+			ipip_entry = mlxsw_sp_ipip_entry_create(mlxsw_sp, ipipt,
+								ol_dev);
+			if (IS_ERR(ipip_entry))
+				return PTR_ERR(ipip_entry);
+		}
+	}
+
+	return 0;
+}
+
+static void mlxsw_sp_netdevice_ipip_ol_unreg_event(struct mlxsw_sp *mlxsw_sp,
+						   struct net_device *ol_dev)
+{
+	struct mlxsw_sp_ipip_entry *ipip_entry;
+
+	ipip_entry = mlxsw_sp_ipip_entry_find_by_ol_dev(mlxsw_sp, ol_dev);
+	if (ipip_entry)
+		mlxsw_sp_ipip_entry_destroy(mlxsw_sp, ipip_entry);
+}
+
+static void
+mlxsw_sp_ipip_entry_ol_up_event(struct mlxsw_sp *mlxsw_sp,
+				struct mlxsw_sp_ipip_entry *ipip_entry)
+{
+	struct mlxsw_sp_fib_entry *decap_fib_entry;
+
+	decap_fib_entry = mlxsw_sp_ipip_entry_find_decap(mlxsw_sp, ipip_entry);
+	if (decap_fib_entry)
+		mlxsw_sp_ipip_entry_promote_decap(mlxsw_sp, ipip_entry,
+						  decap_fib_entry);
+}
+
+static int
+mlxsw_sp_rif_ipip_lb_op(struct mlxsw_sp_rif_ipip_lb *lb_rif, u16 ul_vr_id,
+			u16 ul_rif_id, bool enable)
+{
+	struct mlxsw_sp_rif_ipip_lb_config lb_cf = lb_rif->lb_config;
+	enum mlxsw_reg_ritr_loopback_ipip_options ipip_options;
+	struct mlxsw_sp_rif *rif = &lb_rif->common;
+	struct mlxsw_sp *mlxsw_sp = rif->mlxsw_sp;
+	char ritr_pl[MLXSW_REG_RITR_LEN];
+	struct in6_addr *saddr6;
+	u32 saddr4;
+
+	ipip_options = MLXSW_REG_RITR_LOOPBACK_IPIP_OPTIONS_GRE_KEY_PRESET;
+	switch (lb_cf.ul_protocol) {
+	case MLXSW_SP_L3_PROTO_IPV4:
+		saddr4 = be32_to_cpu(lb_cf.saddr.addr4);
+		mlxsw_reg_ritr_pack(ritr_pl, enable, MLXSW_REG_RITR_LOOPBACK_IF,
+				    rif->rif_index, rif->vr_id, rif->dev->mtu);
+		mlxsw_reg_ritr_loopback_ipip4_pack(ritr_pl, lb_cf.lb_ipipt,
+						   ipip_options, ul_vr_id,
+						   ul_rif_id, saddr4,
+						   lb_cf.okey);
+		break;
+
+	case MLXSW_SP_L3_PROTO_IPV6:
+		saddr6 = &lb_cf.saddr.addr6;
+		mlxsw_reg_ritr_pack(ritr_pl, enable, MLXSW_REG_RITR_LOOPBACK_IF,
+				    rif->rif_index, rif->vr_id, rif->dev->mtu);
+		mlxsw_reg_ritr_loopback_ipip6_pack(ritr_pl, lb_cf.lb_ipipt,
+						   ipip_options, ul_vr_id,
+						   ul_rif_id, saddr6,
+						   lb_cf.okey);
+		break;
+	}
+
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(ritr), ritr_pl);
+}
+
+static int mlxsw_sp_netdevice_ipip_ol_update_mtu(struct mlxsw_sp *mlxsw_sp,
+						 struct net_device *ol_dev)
+{
+	struct mlxsw_sp_ipip_entry *ipip_entry;
+	struct mlxsw_sp_rif_ipip_lb *lb_rif;
+	int err = 0;
+
+	ipip_entry = mlxsw_sp_ipip_entry_find_by_ol_dev(mlxsw_sp, ol_dev);
+	if (ipip_entry) {
+		lb_rif = ipip_entry->ol_lb;
+		err = mlxsw_sp_rif_ipip_lb_op(lb_rif, lb_rif->ul_vr_id,
+					      lb_rif->ul_rif_id, true);
+		if (err)
+			goto out;
+		lb_rif->common.mtu = ol_dev->mtu;
+	}
+
+out:
+	return err;
+}
+
+static void mlxsw_sp_netdevice_ipip_ol_up_event(struct mlxsw_sp *mlxsw_sp,
+						struct net_device *ol_dev)
+{
+	struct mlxsw_sp_ipip_entry *ipip_entry;
+
+	ipip_entry = mlxsw_sp_ipip_entry_find_by_ol_dev(mlxsw_sp, ol_dev);
+	if (ipip_entry)
+		mlxsw_sp_ipip_entry_ol_up_event(mlxsw_sp, ipip_entry);
+}
+
+static void
+mlxsw_sp_ipip_entry_ol_down_event(struct mlxsw_sp *mlxsw_sp,
+				  struct mlxsw_sp_ipip_entry *ipip_entry)
+{
+	if (ipip_entry->decap_fib_entry)
+		mlxsw_sp_ipip_entry_demote_decap(mlxsw_sp, ipip_entry);
+}
+
+static void mlxsw_sp_netdevice_ipip_ol_down_event(struct mlxsw_sp *mlxsw_sp,
+						  struct net_device *ol_dev)
+{
+	struct mlxsw_sp_ipip_entry *ipip_entry;
+
+	ipip_entry = mlxsw_sp_ipip_entry_find_by_ol_dev(mlxsw_sp, ol_dev);
+	if (ipip_entry)
+		mlxsw_sp_ipip_entry_ol_down_event(mlxsw_sp, ipip_entry);
+}
+
+static void mlxsw_sp_nexthop_rif_migrate(struct mlxsw_sp *mlxsw_sp,
+					 struct mlxsw_sp_rif *old_rif,
+					 struct mlxsw_sp_rif *new_rif);
+static int
+mlxsw_sp_ipip_entry_ol_lb_update(struct mlxsw_sp *mlxsw_sp,
+				 struct mlxsw_sp_ipip_entry *ipip_entry,
+				 bool keep_encap,
+				 struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp_rif_ipip_lb *old_lb_rif = ipip_entry->ol_lb;
+	struct mlxsw_sp_rif_ipip_lb *new_lb_rif;
+
+	new_lb_rif = mlxsw_sp_ipip_ol_ipip_lb_create(mlxsw_sp,
+						     ipip_entry->ipipt,
+						     ipip_entry->ol_dev,
+						     extack);
+	if (IS_ERR(new_lb_rif))
+		return PTR_ERR(new_lb_rif);
+	ipip_entry->ol_lb = new_lb_rif;
+
+	if (keep_encap)
+		mlxsw_sp_nexthop_rif_migrate(mlxsw_sp, &old_lb_rif->common,
+					     &new_lb_rif->common);
+
+	mlxsw_sp_rif_destroy(&old_lb_rif->common);
+
+	return 0;
+}
+
+static void mlxsw_sp_nexthop_rif_update(struct mlxsw_sp *mlxsw_sp,
+					struct mlxsw_sp_rif *rif);
+
+/**
+ * __mlxsw_sp_ipip_entry_update_tunnel - Update offload related to IPIP entry.
+ * @mlxsw_sp: mlxsw_sp.
+ * @ipip_entry: IPIP entry.
+ * @recreate_loopback: Recreates the associated loopback RIF.
+ * @keep_encap: Updates next hops that use the tunnel netdevice. This is only
+ *              relevant when recreate_loopback is true.
+ * @update_nexthops: Updates next hops, keeping the current loopback RIF. This
+ *                   is only relevant when recreate_loopback is false.
+ * @extack: extack.
+ *
+ * Return: Non-zero value on failure.
+ */
+int __mlxsw_sp_ipip_entry_update_tunnel(struct mlxsw_sp *mlxsw_sp,
+					struct mlxsw_sp_ipip_entry *ipip_entry,
+					bool recreate_loopback,
+					bool keep_encap,
+					bool update_nexthops,
+					struct netlink_ext_ack *extack)
+{
+	int err;
+
+	/* RIFs can't be edited, so to update loopback, we need to destroy and
+	 * recreate it. That creates a window of opportunity where RALUE and
+	 * RATR registers end up referencing a RIF that's already gone. RATRs
+	 * are handled in mlxsw_sp_ipip_entry_ol_lb_update(), and to take care
+	 * of RALUE, demote the decap route back.
+	 */
+	if (ipip_entry->decap_fib_entry)
+		mlxsw_sp_ipip_entry_demote_decap(mlxsw_sp, ipip_entry);
+
+	if (recreate_loopback) {
+		err = mlxsw_sp_ipip_entry_ol_lb_update(mlxsw_sp, ipip_entry,
+						       keep_encap, extack);
+		if (err)
+			return err;
+	} else if (update_nexthops) {
+		mlxsw_sp_nexthop_rif_update(mlxsw_sp,
+					    &ipip_entry->ol_lb->common);
+	}
+
+	if (ipip_entry->ol_dev->flags & IFF_UP)
+		mlxsw_sp_ipip_entry_ol_up_event(mlxsw_sp, ipip_entry);
+
+	return 0;
+}
+
+static int mlxsw_sp_netdevice_ipip_ol_vrf_event(struct mlxsw_sp *mlxsw_sp,
+						struct net_device *ol_dev,
+						struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp_ipip_entry *ipip_entry =
+		mlxsw_sp_ipip_entry_find_by_ol_dev(mlxsw_sp, ol_dev);
+
+	if (!ipip_entry)
+		return 0;
+
+	return __mlxsw_sp_ipip_entry_update_tunnel(mlxsw_sp, ipip_entry,
+						   true, false, false, extack);
+}
+
+static int
+mlxsw_sp_netdevice_ipip_ul_vrf_event(struct mlxsw_sp *mlxsw_sp,
+				     struct mlxsw_sp_ipip_entry *ipip_entry,
+				     struct net_device *ul_dev,
+				     bool *demote_this,
+				     struct netlink_ext_ack *extack)
+{
+	u32 ul_tb_id = l3mdev_fib_table(ul_dev) ? : RT_TABLE_MAIN;
+	enum mlxsw_sp_l3proto ul_proto;
+	union mlxsw_sp_l3addr saddr;
+
+	/* Moving underlay to a different VRF might cause local address
+	 * conflict, and the conflicting tunnels need to be demoted.
+	 */
+	ul_proto = mlxsw_sp->router->ipip_ops_arr[ipip_entry->ipipt]->ul_proto;
+	saddr = mlxsw_sp_ipip_netdev_saddr(ul_proto, ipip_entry->ol_dev);
+	if (mlxsw_sp_ipip_demote_tunnel_by_saddr(mlxsw_sp, ul_proto,
+						 saddr, ul_tb_id,
+						 ipip_entry)) {
+		*demote_this = true;
+		return 0;
+	}
+
+	return __mlxsw_sp_ipip_entry_update_tunnel(mlxsw_sp, ipip_entry,
+						   true, true, false, extack);
+}
+
+static int
+mlxsw_sp_netdevice_ipip_ul_up_event(struct mlxsw_sp *mlxsw_sp,
+				    struct mlxsw_sp_ipip_entry *ipip_entry,
+				    struct net_device *ul_dev)
+{
+	return __mlxsw_sp_ipip_entry_update_tunnel(mlxsw_sp, ipip_entry,
+						   false, false, true, NULL);
+}
+
+static int
+mlxsw_sp_netdevice_ipip_ul_down_event(struct mlxsw_sp *mlxsw_sp,
+				      struct mlxsw_sp_ipip_entry *ipip_entry,
+				      struct net_device *ul_dev)
+{
+	/* A down underlay device causes encapsulated packets to not be
+	 * forwarded, but decap still works. So refresh next hops without
+	 * touching anything else.
+	 */
+	return __mlxsw_sp_ipip_entry_update_tunnel(mlxsw_sp, ipip_entry,
+						   false, false, true, NULL);
+}
+
+static int
+mlxsw_sp_netdevice_ipip_ol_change_event(struct mlxsw_sp *mlxsw_sp,
+					struct net_device *ol_dev,
+					struct netlink_ext_ack *extack)
+{
+	const struct mlxsw_sp_ipip_ops *ipip_ops;
+	struct mlxsw_sp_ipip_entry *ipip_entry;
+	int err;
+
+	ipip_entry = mlxsw_sp_ipip_entry_find_by_ol_dev(mlxsw_sp, ol_dev);
+	if (!ipip_entry)
+		/* A change might make a tunnel eligible for offloading, but
+		 * that is currently not implemented. What falls to slow path
+		 * stays there.
+		 */
+		return 0;
+
+	/* A change might make a tunnel not eligible for offloading. */
+	if (!mlxsw_sp_netdevice_ipip_can_offload(mlxsw_sp, ol_dev,
+						 ipip_entry->ipipt)) {
+		mlxsw_sp_ipip_entry_demote_tunnel(mlxsw_sp, ipip_entry);
+		return 0;
+	}
+
+	ipip_ops = mlxsw_sp->router->ipip_ops_arr[ipip_entry->ipipt];
+	err = ipip_ops->ol_netdev_change(mlxsw_sp, ipip_entry, extack);
+	return err;
+}
+
+void mlxsw_sp_ipip_entry_demote_tunnel(struct mlxsw_sp *mlxsw_sp,
+				       struct mlxsw_sp_ipip_entry *ipip_entry)
+{
+	struct net_device *ol_dev = ipip_entry->ol_dev;
+
+	if (ol_dev->flags & IFF_UP)
+		mlxsw_sp_ipip_entry_ol_down_event(mlxsw_sp, ipip_entry);
+	mlxsw_sp_ipip_entry_destroy(mlxsw_sp, ipip_entry);
+}
+
+/* The configuration where several tunnels have the same local address in the
+ * same underlay table needs special treatment in the HW. That is currently not
+ * implemented in the driver. This function finds and demotes the first tunnel
+ * with a given source address, except the one passed in the argument
+ * `except'.
+ */
+bool
+mlxsw_sp_ipip_demote_tunnel_by_saddr(struct mlxsw_sp *mlxsw_sp,
+				     enum mlxsw_sp_l3proto ul_proto,
+				     union mlxsw_sp_l3addr saddr,
+				     u32 ul_tb_id,
+				     const struct mlxsw_sp_ipip_entry *except)
+{
+	struct mlxsw_sp_ipip_entry *ipip_entry, *tmp;
+
+	list_for_each_entry_safe(ipip_entry, tmp, &mlxsw_sp->router->ipip_list,
+				 ipip_list_node) {
+		if (ipip_entry != except &&
+		    mlxsw_sp_ipip_entry_saddr_matches(mlxsw_sp, ul_proto, saddr,
+						      ul_tb_id, ipip_entry)) {
+			mlxsw_sp_ipip_entry_demote_tunnel(mlxsw_sp, ipip_entry);
+			return true;
+		}
+	}
+
+	return false;
+}
+
+static void mlxsw_sp_ipip_demote_tunnel_by_ul_netdev(struct mlxsw_sp *mlxsw_sp,
+						     struct net_device *ul_dev)
+{
+	struct mlxsw_sp_ipip_entry *ipip_entry, *tmp;
+
+	list_for_each_entry_safe(ipip_entry, tmp, &mlxsw_sp->router->ipip_list,
+				 ipip_list_node) {
+		struct net_device *ol_dev = ipip_entry->ol_dev;
+		struct net_device *ipip_ul_dev;
+
+		rcu_read_lock();
+		ipip_ul_dev = mlxsw_sp_ipip_netdev_ul_dev_get(ol_dev);
+		rcu_read_unlock();
+		if (ipip_ul_dev == ul_dev)
+			mlxsw_sp_ipip_entry_demote_tunnel(mlxsw_sp, ipip_entry);
+	}
+}
+
+static int mlxsw_sp_netdevice_ipip_ol_event(struct mlxsw_sp *mlxsw_sp,
+					    struct net_device *ol_dev,
+					    unsigned long event,
+					    struct netdev_notifier_info *info)
+{
+	struct netdev_notifier_changeupper_info *chup;
+	struct netlink_ext_ack *extack;
+	int err = 0;
+
+	switch (event) {
+	case NETDEV_REGISTER:
+		err = mlxsw_sp_netdevice_ipip_ol_reg_event(mlxsw_sp, ol_dev);
+		break;
+	case NETDEV_UNREGISTER:
+		mlxsw_sp_netdevice_ipip_ol_unreg_event(mlxsw_sp, ol_dev);
+		break;
+	case NETDEV_UP:
+		mlxsw_sp_netdevice_ipip_ol_up_event(mlxsw_sp, ol_dev);
+		break;
+	case NETDEV_DOWN:
+		mlxsw_sp_netdevice_ipip_ol_down_event(mlxsw_sp, ol_dev);
+		break;
+	case NETDEV_CHANGEUPPER:
+		chup = container_of(info, typeof(*chup), info);
+		extack = info->extack;
+		if (netif_is_l3_master(chup->upper_dev))
+			err = mlxsw_sp_netdevice_ipip_ol_vrf_event(mlxsw_sp,
+								   ol_dev,
+								   extack);
+		break;
+	case NETDEV_CHANGE:
+		extack = info->extack;
+		err = mlxsw_sp_netdevice_ipip_ol_change_event(mlxsw_sp,
+							      ol_dev, extack);
+		break;
+	case NETDEV_CHANGEMTU:
+		err = mlxsw_sp_netdevice_ipip_ol_update_mtu(mlxsw_sp, ol_dev);
+		break;
+	}
+	return err;
+}
+
+static int
+__mlxsw_sp_netdevice_ipip_ul_event(struct mlxsw_sp *mlxsw_sp,
+				   struct mlxsw_sp_ipip_entry *ipip_entry,
+				   struct net_device *ul_dev,
+				   bool *demote_this,
+				   unsigned long event,
+				   struct netdev_notifier_info *info)
+{
+	struct netdev_notifier_changeupper_info *chup;
+	struct netlink_ext_ack *extack;
+
+	switch (event) {
+	case NETDEV_CHANGEUPPER:
+		chup = container_of(info, typeof(*chup), info);
+		extack = info->extack;
+		if (netif_is_l3_master(chup->upper_dev))
+			return mlxsw_sp_netdevice_ipip_ul_vrf_event(mlxsw_sp,
+								    ipip_entry,
+								    ul_dev,
+								    demote_this,
+								    extack);
+		break;
+
+	case NETDEV_UP:
+		return mlxsw_sp_netdevice_ipip_ul_up_event(mlxsw_sp, ipip_entry,
+							   ul_dev);
+	case NETDEV_DOWN:
+		return mlxsw_sp_netdevice_ipip_ul_down_event(mlxsw_sp,
+							     ipip_entry,
+							     ul_dev);
+	}
+	return 0;
+}
+
+static int
+mlxsw_sp_netdevice_ipip_ul_event(struct mlxsw_sp *mlxsw_sp,
+				 struct net_device *ul_dev,
+				 unsigned long event,
+				 struct netdev_notifier_info *info)
+{
+	struct mlxsw_sp_ipip_entry *ipip_entry = NULL;
+	int err;
+
+	while ((ipip_entry = mlxsw_sp_ipip_entry_find_by_ul_dev(mlxsw_sp,
+								ul_dev,
+								ipip_entry))) {
+		struct mlxsw_sp_ipip_entry *prev;
+		bool demote_this = false;
+
+		err = __mlxsw_sp_netdevice_ipip_ul_event(mlxsw_sp, ipip_entry,
+							 ul_dev, &demote_this,
+							 event, info);
+		if (err) {
+			mlxsw_sp_ipip_demote_tunnel_by_ul_netdev(mlxsw_sp,
+								 ul_dev);
+			return err;
+		}
+
+		if (demote_this) {
+			if (list_is_first(&ipip_entry->ipip_list_node,
+					  &mlxsw_sp->router->ipip_list))
+				prev = NULL;
+			else
+				/* This can't be cached from previous iteration,
+				 * because that entry could be gone now.
+				 */
+				prev = list_prev_entry(ipip_entry,
+						       ipip_list_node);
+			mlxsw_sp_ipip_entry_demote_tunnel(mlxsw_sp, ipip_entry);
+			ipip_entry = prev;
+		}
+	}
+
+	return 0;
+}
+
+int mlxsw_sp_router_nve_promote_decap(struct mlxsw_sp *mlxsw_sp, u32 ul_tb_id,
+				      enum mlxsw_sp_l3proto ul_proto,
+				      const union mlxsw_sp_l3addr *ul_sip,
+				      u32 tunnel_index)
+{
+	enum mlxsw_sp_fib_entry_type type = MLXSW_SP_FIB_ENTRY_TYPE_TRAP;
+	struct mlxsw_sp_router *router = mlxsw_sp->router;
+	struct mlxsw_sp_fib_entry *fib_entry;
+	int err = 0;
+
+	mutex_lock(&mlxsw_sp->router->lock);
+
+	if (WARN_ON_ONCE(router->nve_decap_config.valid)) {
+		err = -EINVAL;
+		goto out;
+	}
+
+	router->nve_decap_config.ul_tb_id = ul_tb_id;
+	router->nve_decap_config.tunnel_index = tunnel_index;
+	router->nve_decap_config.ul_proto = ul_proto;
+	router->nve_decap_config.ul_sip = *ul_sip;
+	router->nve_decap_config.valid = true;
+
+	/* It is valid to create a tunnel with a local IP and only later
+	 * assign this IP address to a local interface
+	 */
+	fib_entry = mlxsw_sp_router_ip2me_fib_entry_find(mlxsw_sp, ul_tb_id,
+							 ul_proto, ul_sip,
+							 type);
+	if (!fib_entry)
+		goto out;
+
+	fib_entry->decap.tunnel_index = tunnel_index;
+	fib_entry->type = MLXSW_SP_FIB_ENTRY_TYPE_NVE_DECAP;
+
+	err = mlxsw_sp_fib_entry_update(mlxsw_sp, fib_entry);
+	if (err)
+		goto err_fib_entry_update;
+
+	goto out;
+
+err_fib_entry_update:
+	fib_entry->type = MLXSW_SP_FIB_ENTRY_TYPE_TRAP;
+	mlxsw_sp_fib_entry_update(mlxsw_sp, fib_entry);
+out:
+	mutex_unlock(&mlxsw_sp->router->lock);
+	return err;
+}
+
+void mlxsw_sp_router_nve_demote_decap(struct mlxsw_sp *mlxsw_sp, u32 ul_tb_id,
+				      enum mlxsw_sp_l3proto ul_proto,
+				      const union mlxsw_sp_l3addr *ul_sip)
+{
+	enum mlxsw_sp_fib_entry_type type = MLXSW_SP_FIB_ENTRY_TYPE_NVE_DECAP;
+	struct mlxsw_sp_router *router = mlxsw_sp->router;
+	struct mlxsw_sp_fib_entry *fib_entry;
+
+	mutex_lock(&mlxsw_sp->router->lock);
+
+	if (WARN_ON_ONCE(!router->nve_decap_config.valid))
+		goto out;
+
+	router->nve_decap_config.valid = false;
+
+	fib_entry = mlxsw_sp_router_ip2me_fib_entry_find(mlxsw_sp, ul_tb_id,
+							 ul_proto, ul_sip,
+							 type);
+	if (!fib_entry)
+		goto out;
+
+	fib_entry->type = MLXSW_SP_FIB_ENTRY_TYPE_TRAP;
+	mlxsw_sp_fib_entry_update(mlxsw_sp, fib_entry);
+out:
+	mutex_unlock(&mlxsw_sp->router->lock);
+}
+
+static bool mlxsw_sp_router_nve_is_decap(struct mlxsw_sp *mlxsw_sp,
+					 u32 ul_tb_id,
+					 enum mlxsw_sp_l3proto ul_proto,
+					 const union mlxsw_sp_l3addr *ul_sip)
+{
+	struct mlxsw_sp_router *router = mlxsw_sp->router;
+
+	return router->nve_decap_config.valid &&
+	       router->nve_decap_config.ul_tb_id == ul_tb_id &&
+	       router->nve_decap_config.ul_proto == ul_proto &&
+	       !memcmp(&router->nve_decap_config.ul_sip, ul_sip,
+		       sizeof(*ul_sip));
+}
+
+struct mlxsw_sp_neigh_key {
+	struct neighbour *n;
+};
+
+struct mlxsw_sp_neigh_entry {
+	struct list_head rif_list_node;
+	struct rhash_head ht_node;
+	struct mlxsw_sp_neigh_key key;
+	u16 rif;
+	bool connected;
+	unsigned char ha[ETH_ALEN];
+	struct list_head nexthop_list; /* list of nexthops using
+					* this neigh entry
+					*/
+	struct list_head nexthop_neighs_list_node;
+	unsigned int counter_index;
+	bool counter_valid;
+};
+
+static const struct rhashtable_params mlxsw_sp_neigh_ht_params = {
+	.key_offset = offsetof(struct mlxsw_sp_neigh_entry, key),
+	.head_offset = offsetof(struct mlxsw_sp_neigh_entry, ht_node),
+	.key_len = sizeof(struct mlxsw_sp_neigh_key),
+};
+
+struct mlxsw_sp_neigh_entry *
+mlxsw_sp_rif_neigh_next(struct mlxsw_sp_rif *rif,
+			struct mlxsw_sp_neigh_entry *neigh_entry)
+{
+	if (!neigh_entry) {
+		if (list_empty(&rif->neigh_list))
+			return NULL;
+		else
+			return list_first_entry(&rif->neigh_list,
+						typeof(*neigh_entry),
+						rif_list_node);
+	}
+	if (list_is_last(&neigh_entry->rif_list_node, &rif->neigh_list))
+		return NULL;
+	return list_next_entry(neigh_entry, rif_list_node);
+}
+
+int mlxsw_sp_neigh_entry_type(struct mlxsw_sp_neigh_entry *neigh_entry)
+{
+	return neigh_entry->key.n->tbl->family;
+}
+
+unsigned char *
+mlxsw_sp_neigh_entry_ha(struct mlxsw_sp_neigh_entry *neigh_entry)
+{
+	return neigh_entry->ha;
+}
+
+u32 mlxsw_sp_neigh4_entry_dip(struct mlxsw_sp_neigh_entry *neigh_entry)
+{
+	struct neighbour *n;
+
+	n = neigh_entry->key.n;
+	return ntohl(*((__be32 *) n->primary_key));
+}
+
+struct in6_addr *
+mlxsw_sp_neigh6_entry_dip(struct mlxsw_sp_neigh_entry *neigh_entry)
+{
+	struct neighbour *n;
+
+	n = neigh_entry->key.n;
+	return (struct in6_addr *) &n->primary_key;
+}
+
+int mlxsw_sp_neigh_counter_get(struct mlxsw_sp *mlxsw_sp,
+			       struct mlxsw_sp_neigh_entry *neigh_entry,
+			       u64 *p_counter)
+{
+	if (!neigh_entry->counter_valid)
+		return -EINVAL;
+
+	return mlxsw_sp_flow_counter_get(mlxsw_sp, neigh_entry->counter_index,
+					 p_counter, NULL);
+}
+
+static struct mlxsw_sp_neigh_entry *
+mlxsw_sp_neigh_entry_alloc(struct mlxsw_sp *mlxsw_sp, struct neighbour *n,
+			   u16 rif)
+{
+	struct mlxsw_sp_neigh_entry *neigh_entry;
+
+	neigh_entry = kzalloc(sizeof(*neigh_entry), GFP_KERNEL);
+	if (!neigh_entry)
+		return NULL;
+
+	neigh_entry->key.n = n;
+	neigh_entry->rif = rif;
+	INIT_LIST_HEAD(&neigh_entry->nexthop_list);
+
+	return neigh_entry;
+}
+
+static void mlxsw_sp_neigh_entry_free(struct mlxsw_sp_neigh_entry *neigh_entry)
+{
+	kfree(neigh_entry);
+}
+
+static int
+mlxsw_sp_neigh_entry_insert(struct mlxsw_sp *mlxsw_sp,
+			    struct mlxsw_sp_neigh_entry *neigh_entry)
+{
+	return rhashtable_insert_fast(&mlxsw_sp->router->neigh_ht,
+				      &neigh_entry->ht_node,
+				      mlxsw_sp_neigh_ht_params);
+}
+
+static void
+mlxsw_sp_neigh_entry_remove(struct mlxsw_sp *mlxsw_sp,
+			    struct mlxsw_sp_neigh_entry *neigh_entry)
+{
+	rhashtable_remove_fast(&mlxsw_sp->router->neigh_ht,
+			       &neigh_entry->ht_node,
+			       mlxsw_sp_neigh_ht_params);
+}
+
+static bool
+mlxsw_sp_neigh_counter_should_alloc(struct mlxsw_sp *mlxsw_sp,
+				    struct mlxsw_sp_neigh_entry *neigh_entry)
+{
+	struct devlink *devlink;
+	const char *table_name;
+
+	switch (mlxsw_sp_neigh_entry_type(neigh_entry)) {
+	case AF_INET:
+		table_name = MLXSW_SP_DPIPE_TABLE_NAME_HOST4;
+		break;
+	case AF_INET6:
+		table_name = MLXSW_SP_DPIPE_TABLE_NAME_HOST6;
+		break;
+	default:
+		WARN_ON(1);
+		return false;
+	}
+
+	devlink = priv_to_devlink(mlxsw_sp->core);
+	return devlink_dpipe_table_counter_enabled(devlink, table_name);
+}
+
+static void
+mlxsw_sp_neigh_counter_alloc(struct mlxsw_sp *mlxsw_sp,
+			     struct mlxsw_sp_neigh_entry *neigh_entry)
+{
+	if (!mlxsw_sp_neigh_counter_should_alloc(mlxsw_sp, neigh_entry))
+		return;
+
+	if (mlxsw_sp_flow_counter_alloc(mlxsw_sp, &neigh_entry->counter_index))
+		return;
+
+	neigh_entry->counter_valid = true;
+}
+
+static void
+mlxsw_sp_neigh_counter_free(struct mlxsw_sp *mlxsw_sp,
+			    struct mlxsw_sp_neigh_entry *neigh_entry)
+{
+	if (!neigh_entry->counter_valid)
+		return;
+	mlxsw_sp_flow_counter_free(mlxsw_sp,
+				   neigh_entry->counter_index);
+	neigh_entry->counter_valid = false;
+}
+
+static struct mlxsw_sp_neigh_entry *
+mlxsw_sp_neigh_entry_create(struct mlxsw_sp *mlxsw_sp, struct neighbour *n)
+{
+	struct mlxsw_sp_neigh_entry *neigh_entry;
+	struct mlxsw_sp_rif *rif;
+	int err;
+
+	rif = mlxsw_sp_rif_find_by_dev(mlxsw_sp, n->dev);
+	if (!rif)
+		return ERR_PTR(-EINVAL);
+
+	neigh_entry = mlxsw_sp_neigh_entry_alloc(mlxsw_sp, n, rif->rif_index);
+	if (!neigh_entry)
+		return ERR_PTR(-ENOMEM);
+
+	err = mlxsw_sp_neigh_entry_insert(mlxsw_sp, neigh_entry);
+	if (err)
+		goto err_neigh_entry_insert;
+
+	mlxsw_sp_neigh_counter_alloc(mlxsw_sp, neigh_entry);
+	atomic_inc(&mlxsw_sp->router->neighs_update.neigh_count);
+	list_add(&neigh_entry->rif_list_node, &rif->neigh_list);
+
+	return neigh_entry;
+
+err_neigh_entry_insert:
+	mlxsw_sp_neigh_entry_free(neigh_entry);
+	return ERR_PTR(err);
+}
+
+static void
+mlxsw_sp_neigh_entry_destroy(struct mlxsw_sp *mlxsw_sp,
+			     struct mlxsw_sp_neigh_entry *neigh_entry)
+{
+	list_del(&neigh_entry->rif_list_node);
+	atomic_dec(&mlxsw_sp->router->neighs_update.neigh_count);
+	mlxsw_sp_neigh_counter_free(mlxsw_sp, neigh_entry);
+	mlxsw_sp_neigh_entry_remove(mlxsw_sp, neigh_entry);
+	mlxsw_sp_neigh_entry_free(neigh_entry);
+}
+
+static struct mlxsw_sp_neigh_entry *
+mlxsw_sp_neigh_entry_lookup(struct mlxsw_sp *mlxsw_sp, struct neighbour *n)
+{
+	struct mlxsw_sp_neigh_key key;
+
+	key.n = n;
+	return rhashtable_lookup_fast(&mlxsw_sp->router->neigh_ht,
+				      &key, mlxsw_sp_neigh_ht_params);
+}
+
+static void
+mlxsw_sp_router_neighs_update_interval_init(struct mlxsw_sp *mlxsw_sp)
+{
+	unsigned long interval;
+
+#if IS_ENABLED(CONFIG_IPV6)
+	interval = min_t(unsigned long,
+			 NEIGH_VAR(&arp_tbl.parms, DELAY_PROBE_TIME),
+			 NEIGH_VAR(&nd_tbl.parms, DELAY_PROBE_TIME));
+#else
+	interval = NEIGH_VAR(&arp_tbl.parms, DELAY_PROBE_TIME);
+#endif
+	mlxsw_sp->router->neighs_update.interval = jiffies_to_msecs(interval);
+}
+
+static void mlxsw_sp_router_neigh_ent_ipv4_process(struct mlxsw_sp *mlxsw_sp,
+						   char *rauhtd_pl,
+						   int ent_index)
+{
+	u64 max_rifs = MLXSW_CORE_RES_GET(mlxsw_sp->core, MAX_RIFS);
+	struct net_device *dev;
+	struct neighbour *n;
+	__be32 dipn;
+	u32 dip;
+	u16 rif;
+
+	mlxsw_reg_rauhtd_ent_ipv4_unpack(rauhtd_pl, ent_index, &rif, &dip);
+
+	if (WARN_ON_ONCE(rif >= max_rifs))
+		return;
+	if (!mlxsw_sp->router->rifs[rif]) {
+		dev_err_ratelimited(mlxsw_sp->bus_info->dev, "Incorrect RIF in neighbour entry\n");
+		return;
+	}
+
+	dipn = htonl(dip);
+	dev = mlxsw_sp->router->rifs[rif]->dev;
+	n = neigh_lookup(&arp_tbl, &dipn, dev);
+	if (!n)
+		return;
+
+	netdev_dbg(dev, "Updating neighbour with IP=%pI4h\n", &dip);
+	neigh_event_send(n, NULL);
+	neigh_release(n);
+}
+
+#if IS_ENABLED(CONFIG_IPV6)
+static void mlxsw_sp_router_neigh_ent_ipv6_process(struct mlxsw_sp *mlxsw_sp,
+						   char *rauhtd_pl,
+						   int rec_index)
+{
+	struct net_device *dev;
+	struct neighbour *n;
+	struct in6_addr dip;
+	u16 rif;
+
+	mlxsw_reg_rauhtd_ent_ipv6_unpack(rauhtd_pl, rec_index, &rif,
+					 (char *) &dip);
+
+	if (!mlxsw_sp->router->rifs[rif]) {
+		dev_err_ratelimited(mlxsw_sp->bus_info->dev, "Incorrect RIF in neighbour entry\n");
+		return;
+	}
+
+	dev = mlxsw_sp->router->rifs[rif]->dev;
+	n = neigh_lookup(&nd_tbl, &dip, dev);
+	if (!n)
+		return;
+
+	netdev_dbg(dev, "Updating neighbour with IP=%pI6c\n", &dip);
+	neigh_event_send(n, NULL);
+	neigh_release(n);
+}
+#else
+static void mlxsw_sp_router_neigh_ent_ipv6_process(struct mlxsw_sp *mlxsw_sp,
+						   char *rauhtd_pl,
+						   int rec_index)
+{
+}
+#endif
+
+static void mlxsw_sp_router_neigh_rec_ipv4_process(struct mlxsw_sp *mlxsw_sp,
+						   char *rauhtd_pl,
+						   int rec_index)
+{
+	u8 num_entries;
+	int i;
+
+	num_entries = mlxsw_reg_rauhtd_ipv4_rec_num_entries_get(rauhtd_pl,
+								rec_index);
+	/* Hardware starts counting at 0, so add 1. */
+	num_entries++;
+
+	/* Each record consists of several neighbour entries. */
+	for (i = 0; i < num_entries; i++) {
+		int ent_index;
+
+		ent_index = rec_index * MLXSW_REG_RAUHTD_IPV4_ENT_PER_REC + i;
+		mlxsw_sp_router_neigh_ent_ipv4_process(mlxsw_sp, rauhtd_pl,
+						       ent_index);
+	}
+
+}
+
+static void mlxsw_sp_router_neigh_rec_ipv6_process(struct mlxsw_sp *mlxsw_sp,
+						   char *rauhtd_pl,
+						   int rec_index)
+{
+	/* One record contains one entry. */
+	mlxsw_sp_router_neigh_ent_ipv6_process(mlxsw_sp, rauhtd_pl,
+					       rec_index);
+}
+
+static void mlxsw_sp_router_neigh_rec_process(struct mlxsw_sp *mlxsw_sp,
+					      char *rauhtd_pl, int rec_index)
+{
+	switch (mlxsw_reg_rauhtd_rec_type_get(rauhtd_pl, rec_index)) {
+	case MLXSW_REG_RAUHTD_TYPE_IPV4:
+		mlxsw_sp_router_neigh_rec_ipv4_process(mlxsw_sp, rauhtd_pl,
+						       rec_index);
+		break;
+	case MLXSW_REG_RAUHTD_TYPE_IPV6:
+		mlxsw_sp_router_neigh_rec_ipv6_process(mlxsw_sp, rauhtd_pl,
+						       rec_index);
+		break;
+	}
+}
+
+static bool mlxsw_sp_router_rauhtd_is_full(char *rauhtd_pl)
+{
+	u8 num_rec, last_rec_index, num_entries;
+
+	num_rec = mlxsw_reg_rauhtd_num_rec_get(rauhtd_pl);
+	last_rec_index = num_rec - 1;
+
+	if (num_rec < MLXSW_REG_RAUHTD_REC_MAX_NUM)
+		return false;
+	if (mlxsw_reg_rauhtd_rec_type_get(rauhtd_pl, last_rec_index) ==
+	    MLXSW_REG_RAUHTD_TYPE_IPV6)
+		return true;
+
+	num_entries = mlxsw_reg_rauhtd_ipv4_rec_num_entries_get(rauhtd_pl,
+								last_rec_index);
+	if (++num_entries == MLXSW_REG_RAUHTD_IPV4_ENT_PER_REC)
+		return true;
+	return false;
+}
+
+static int
+__mlxsw_sp_router_neighs_update_rauhtd(struct mlxsw_sp *mlxsw_sp,
+				       char *rauhtd_pl,
+				       enum mlxsw_reg_rauhtd_type type)
+{
+	int i, num_rec;
+	int err;
+
+	/* Ensure the RIF we read from the device does not change mid-dump. */
+	mutex_lock(&mlxsw_sp->router->lock);
+	do {
+		mlxsw_reg_rauhtd_pack(rauhtd_pl, type);
+		err = mlxsw_reg_query(mlxsw_sp->core, MLXSW_REG(rauhtd),
+				      rauhtd_pl);
+		if (err) {
+			dev_err_ratelimited(mlxsw_sp->bus_info->dev, "Failed to dump neighbour table\n");
+			break;
+		}
+		num_rec = mlxsw_reg_rauhtd_num_rec_get(rauhtd_pl);
+		for (i = 0; i < num_rec; i++)
+			mlxsw_sp_router_neigh_rec_process(mlxsw_sp, rauhtd_pl,
+							  i);
+	} while (mlxsw_sp_router_rauhtd_is_full(rauhtd_pl));
+	mutex_unlock(&mlxsw_sp->router->lock);
+
+	return err;
+}
+
+static int mlxsw_sp_router_neighs_update_rauhtd(struct mlxsw_sp *mlxsw_sp)
+{
+	enum mlxsw_reg_rauhtd_type type;
+	char *rauhtd_pl;
+	int err;
+
+	if (!atomic_read(&mlxsw_sp->router->neighs_update.neigh_count))
+		return 0;
+
+	rauhtd_pl = kmalloc(MLXSW_REG_RAUHTD_LEN, GFP_KERNEL);
+	if (!rauhtd_pl)
+		return -ENOMEM;
+
+	type = MLXSW_REG_RAUHTD_TYPE_IPV4;
+	err = __mlxsw_sp_router_neighs_update_rauhtd(mlxsw_sp, rauhtd_pl, type);
+	if (err)
+		goto out;
+
+	type = MLXSW_REG_RAUHTD_TYPE_IPV6;
+	err = __mlxsw_sp_router_neighs_update_rauhtd(mlxsw_sp, rauhtd_pl, type);
+out:
+	kfree(rauhtd_pl);
+	return err;
+}
+
+static void mlxsw_sp_router_neighs_update_nh(struct mlxsw_sp *mlxsw_sp)
+{
+	struct mlxsw_sp_neigh_entry *neigh_entry;
+
+	mutex_lock(&mlxsw_sp->router->lock);
+	list_for_each_entry(neigh_entry, &mlxsw_sp->router->nexthop_neighs_list,
+			    nexthop_neighs_list_node)
+		/* If this neigh have nexthops, make the kernel think this neigh
+		 * is active regardless of the traffic.
+		 */
+		neigh_event_send(neigh_entry->key.n, NULL);
+	mutex_unlock(&mlxsw_sp->router->lock);
+}
+
+static void
+mlxsw_sp_router_neighs_update_work_schedule(struct mlxsw_sp *mlxsw_sp)
+{
+	unsigned long interval = mlxsw_sp->router->neighs_update.interval;
+
+	mlxsw_core_schedule_dw(&mlxsw_sp->router->neighs_update.dw,
+			       msecs_to_jiffies(interval));
+}
+
+static void mlxsw_sp_router_neighs_update_work(struct work_struct *work)
+{
+	struct mlxsw_sp_router *router;
+	int err;
+
+	router = container_of(work, struct mlxsw_sp_router,
+			      neighs_update.dw.work);
+	err = mlxsw_sp_router_neighs_update_rauhtd(router->mlxsw_sp);
+	if (err)
+		dev_err(router->mlxsw_sp->bus_info->dev, "Could not update kernel for neigh activity");
+
+	mlxsw_sp_router_neighs_update_nh(router->mlxsw_sp);
+
+	mlxsw_sp_router_neighs_update_work_schedule(router->mlxsw_sp);
+}
+
+static void mlxsw_sp_router_probe_unresolved_nexthops(struct work_struct *work)
+{
+	struct mlxsw_sp_neigh_entry *neigh_entry;
+	struct mlxsw_sp_router *router;
+
+	router = container_of(work, struct mlxsw_sp_router,
+			      nexthop_probe_dw.work);
+	/* Iterate over nexthop neighbours, find those who are unresolved and
+	 * send arp on them. This solves the chicken-egg problem when
+	 * the nexthop wouldn't get offloaded until the neighbor is resolved
+	 * but it wouldn't get resolved ever in case traffic is flowing in HW
+	 * using different nexthop.
+	 */
+	mutex_lock(&router->lock);
+	list_for_each_entry(neigh_entry, &router->nexthop_neighs_list,
+			    nexthop_neighs_list_node)
+		if (!neigh_entry->connected)
+			neigh_event_send(neigh_entry->key.n, NULL);
+	mutex_unlock(&router->lock);
+
+	mlxsw_core_schedule_dw(&router->nexthop_probe_dw,
+			       MLXSW_SP_UNRESOLVED_NH_PROBE_INTERVAL);
+}
+
+static void
+mlxsw_sp_nexthop_neigh_update(struct mlxsw_sp *mlxsw_sp,
+			      struct mlxsw_sp_neigh_entry *neigh_entry,
+			      bool removing, bool dead);
+
+static enum mlxsw_reg_rauht_op mlxsw_sp_rauht_op(bool adding)
+{
+	return adding ? MLXSW_REG_RAUHT_OP_WRITE_ADD :
+			MLXSW_REG_RAUHT_OP_WRITE_DELETE;
+}
+
+static int
+mlxsw_sp_router_neigh_entry_op4(struct mlxsw_sp *mlxsw_sp,
+				struct mlxsw_sp_neigh_entry *neigh_entry,
+				enum mlxsw_reg_rauht_op op)
+{
+	struct neighbour *n = neigh_entry->key.n;
+	u32 dip = ntohl(*((__be32 *) n->primary_key));
+	char rauht_pl[MLXSW_REG_RAUHT_LEN];
+
+	mlxsw_reg_rauht_pack4(rauht_pl, op, neigh_entry->rif, neigh_entry->ha,
+			      dip);
+	if (neigh_entry->counter_valid)
+		mlxsw_reg_rauht_pack_counter(rauht_pl,
+					     neigh_entry->counter_index);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(rauht), rauht_pl);
+}
+
+static int
+mlxsw_sp_router_neigh_entry_op6(struct mlxsw_sp *mlxsw_sp,
+				struct mlxsw_sp_neigh_entry *neigh_entry,
+				enum mlxsw_reg_rauht_op op)
+{
+	struct neighbour *n = neigh_entry->key.n;
+	char rauht_pl[MLXSW_REG_RAUHT_LEN];
+	const char *dip = n->primary_key;
+
+	mlxsw_reg_rauht_pack6(rauht_pl, op, neigh_entry->rif, neigh_entry->ha,
+			      dip);
+	if (neigh_entry->counter_valid)
+		mlxsw_reg_rauht_pack_counter(rauht_pl,
+					     neigh_entry->counter_index);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(rauht), rauht_pl);
+}
+
+bool mlxsw_sp_neigh_ipv6_ignore(struct mlxsw_sp_neigh_entry *neigh_entry)
+{
+	struct neighbour *n = neigh_entry->key.n;
+
+	/* Packets with a link-local destination address are trapped
+	 * after LPM lookup and never reach the neighbour table, so
+	 * there is no need to program such neighbours to the device.
+	 */
+	if (ipv6_addr_type((struct in6_addr *) &n->primary_key) &
+	    IPV6_ADDR_LINKLOCAL)
+		return true;
+	return false;
+}
+
+static void
+mlxsw_sp_neigh_entry_update(struct mlxsw_sp *mlxsw_sp,
+			    struct mlxsw_sp_neigh_entry *neigh_entry,
+			    bool adding)
+{
+	enum mlxsw_reg_rauht_op op = mlxsw_sp_rauht_op(adding);
+	int err;
+
+	if (!adding && !neigh_entry->connected)
+		return;
+	neigh_entry->connected = adding;
+	if (neigh_entry->key.n->tbl->family == AF_INET) {
+		err = mlxsw_sp_router_neigh_entry_op4(mlxsw_sp, neigh_entry,
+						      op);
+		if (err)
+			return;
+	} else if (neigh_entry->key.n->tbl->family == AF_INET6) {
+		if (mlxsw_sp_neigh_ipv6_ignore(neigh_entry))
+			return;
+		err = mlxsw_sp_router_neigh_entry_op6(mlxsw_sp, neigh_entry,
+						      op);
+		if (err)
+			return;
+	} else {
+		WARN_ON_ONCE(1);
+		return;
+	}
+
+	if (adding)
+		neigh_entry->key.n->flags |= NTF_OFFLOADED;
+	else
+		neigh_entry->key.n->flags &= ~NTF_OFFLOADED;
+}
+
+void
+mlxsw_sp_neigh_entry_counter_update(struct mlxsw_sp *mlxsw_sp,
+				    struct mlxsw_sp_neigh_entry *neigh_entry,
+				    bool adding)
+{
+	if (adding)
+		mlxsw_sp_neigh_counter_alloc(mlxsw_sp, neigh_entry);
+	else
+		mlxsw_sp_neigh_counter_free(mlxsw_sp, neigh_entry);
+	mlxsw_sp_neigh_entry_update(mlxsw_sp, neigh_entry, true);
+}
+
+struct mlxsw_sp_netevent_work {
+	struct work_struct work;
+	struct mlxsw_sp *mlxsw_sp;
+	struct neighbour *n;
+};
+
+static void mlxsw_sp_router_neigh_event_work(struct work_struct *work)
+{
+	struct mlxsw_sp_netevent_work *net_work =
+		container_of(work, struct mlxsw_sp_netevent_work, work);
+	struct mlxsw_sp *mlxsw_sp = net_work->mlxsw_sp;
+	struct mlxsw_sp_neigh_entry *neigh_entry;
+	struct neighbour *n = net_work->n;
+	unsigned char ha[ETH_ALEN];
+	bool entry_connected;
+	u8 nud_state, dead;
+
+	/* If these parameters are changed after we release the lock,
+	 * then we are guaranteed to receive another event letting us
+	 * know about it.
+	 */
+	read_lock_bh(&n->lock);
+	memcpy(ha, n->ha, ETH_ALEN);
+	nud_state = n->nud_state;
+	dead = n->dead;
+	read_unlock_bh(&n->lock);
+
+	mutex_lock(&mlxsw_sp->router->lock);
+	mlxsw_sp_span_respin(mlxsw_sp);
+
+	entry_connected = nud_state & NUD_VALID && !dead;
+	neigh_entry = mlxsw_sp_neigh_entry_lookup(mlxsw_sp, n);
+	if (!entry_connected && !neigh_entry)
+		goto out;
+	if (!neigh_entry) {
+		neigh_entry = mlxsw_sp_neigh_entry_create(mlxsw_sp, n);
+		if (IS_ERR(neigh_entry))
+			goto out;
+	}
+
+	if (neigh_entry->connected && entry_connected &&
+	    !memcmp(neigh_entry->ha, ha, ETH_ALEN))
+		goto out;
+
+	memcpy(neigh_entry->ha, ha, ETH_ALEN);
+	mlxsw_sp_neigh_entry_update(mlxsw_sp, neigh_entry, entry_connected);
+	mlxsw_sp_nexthop_neigh_update(mlxsw_sp, neigh_entry, !entry_connected,
+				      dead);
+
+	if (!neigh_entry->connected && list_empty(&neigh_entry->nexthop_list))
+		mlxsw_sp_neigh_entry_destroy(mlxsw_sp, neigh_entry);
+
+out:
+	mutex_unlock(&mlxsw_sp->router->lock);
+	neigh_release(n);
+	kfree(net_work);
+}
+
+static int mlxsw_sp_mp_hash_init(struct mlxsw_sp *mlxsw_sp);
+
+static void mlxsw_sp_router_mp_hash_event_work(struct work_struct *work)
+{
+	struct mlxsw_sp_netevent_work *net_work =
+		container_of(work, struct mlxsw_sp_netevent_work, work);
+	struct mlxsw_sp *mlxsw_sp = net_work->mlxsw_sp;
+
+	mlxsw_sp_mp_hash_init(mlxsw_sp);
+	kfree(net_work);
+}
+
+static int __mlxsw_sp_router_init(struct mlxsw_sp *mlxsw_sp);
+
+static void mlxsw_sp_router_update_priority_work(struct work_struct *work)
+{
+	struct mlxsw_sp_netevent_work *net_work =
+		container_of(work, struct mlxsw_sp_netevent_work, work);
+	struct mlxsw_sp *mlxsw_sp = net_work->mlxsw_sp;
+
+	__mlxsw_sp_router_init(mlxsw_sp);
+	kfree(net_work);
+}
+
+static int mlxsw_sp_router_schedule_work(struct net *net,
+					 struct notifier_block *nb,
+					 void (*cb)(struct work_struct *))
+{
+	struct mlxsw_sp_netevent_work *net_work;
+	struct mlxsw_sp_router *router;
+
+	router = container_of(nb, struct mlxsw_sp_router, netevent_nb);
+	if (!net_eq(net, mlxsw_sp_net(router->mlxsw_sp)))
+		return NOTIFY_DONE;
+
+	net_work = kzalloc(sizeof(*net_work), GFP_ATOMIC);
+	if (!net_work)
+		return NOTIFY_BAD;
+
+	INIT_WORK(&net_work->work, cb);
+	net_work->mlxsw_sp = router->mlxsw_sp;
+	mlxsw_core_schedule_work(&net_work->work);
+	return NOTIFY_DONE;
+}
+
+static int mlxsw_sp_router_netevent_event(struct notifier_block *nb,
+					  unsigned long event, void *ptr)
+{
+	struct mlxsw_sp_netevent_work *net_work;
+	struct mlxsw_sp_port *mlxsw_sp_port;
+	struct mlxsw_sp *mlxsw_sp;
+	unsigned long interval;
+	struct neigh_parms *p;
+	struct neighbour *n;
+
+	switch (event) {
+	case NETEVENT_DELAY_PROBE_TIME_UPDATE:
+		p = ptr;
+
+		/* We don't care about changes in the default table. */
+		if (!p->dev || (p->tbl->family != AF_INET &&
+				p->tbl->family != AF_INET6))
+			return NOTIFY_DONE;
+
+		/* We are in atomic context and can't take RTNL mutex,
+		 * so use RCU variant to walk the device chain.
+		 */
+		mlxsw_sp_port = mlxsw_sp_port_lower_dev_hold(p->dev);
+		if (!mlxsw_sp_port)
+			return NOTIFY_DONE;
+
+		mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+		interval = jiffies_to_msecs(NEIGH_VAR(p, DELAY_PROBE_TIME));
+		mlxsw_sp->router->neighs_update.interval = interval;
+
+		mlxsw_sp_port_dev_put(mlxsw_sp_port);
+		break;
+	case NETEVENT_NEIGH_UPDATE:
+		n = ptr;
+
+		if (n->tbl->family != AF_INET && n->tbl->family != AF_INET6)
+			return NOTIFY_DONE;
+
+		mlxsw_sp_port = mlxsw_sp_port_lower_dev_hold(n->dev);
+		if (!mlxsw_sp_port)
+			return NOTIFY_DONE;
+
+		net_work = kzalloc(sizeof(*net_work), GFP_ATOMIC);
+		if (!net_work) {
+			mlxsw_sp_port_dev_put(mlxsw_sp_port);
+			return NOTIFY_BAD;
+		}
+
+		INIT_WORK(&net_work->work, mlxsw_sp_router_neigh_event_work);
+		net_work->mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+		net_work->n = n;
+
+		/* Take a reference to ensure the neighbour won't be
+		 * destructed until we drop the reference in delayed
+		 * work.
+		 */
+		neigh_clone(n);
+		mlxsw_core_schedule_work(&net_work->work);
+		mlxsw_sp_port_dev_put(mlxsw_sp_port);
+		break;
+	case NETEVENT_IPV4_MPATH_HASH_UPDATE:
+	case NETEVENT_IPV6_MPATH_HASH_UPDATE:
+		return mlxsw_sp_router_schedule_work(ptr, nb,
+				mlxsw_sp_router_mp_hash_event_work);
+
+	case NETEVENT_IPV4_FWD_UPDATE_PRIORITY_UPDATE:
+		return mlxsw_sp_router_schedule_work(ptr, nb,
+				mlxsw_sp_router_update_priority_work);
+	}
+
+	return NOTIFY_DONE;
+}
+
+static int mlxsw_sp_neigh_init(struct mlxsw_sp *mlxsw_sp)
+{
+	int err;
+
+	err = rhashtable_init(&mlxsw_sp->router->neigh_ht,
+			      &mlxsw_sp_neigh_ht_params);
+	if (err)
+		return err;
+
+	/* Initialize the polling interval according to the default
+	 * table.
+	 */
+	mlxsw_sp_router_neighs_update_interval_init(mlxsw_sp);
+
+	/* Create the delayed works for the activity_update */
+	INIT_DELAYED_WORK(&mlxsw_sp->router->neighs_update.dw,
+			  mlxsw_sp_router_neighs_update_work);
+	INIT_DELAYED_WORK(&mlxsw_sp->router->nexthop_probe_dw,
+			  mlxsw_sp_router_probe_unresolved_nexthops);
+	atomic_set(&mlxsw_sp->router->neighs_update.neigh_count, 0);
+	mlxsw_core_schedule_dw(&mlxsw_sp->router->neighs_update.dw, 0);
+	mlxsw_core_schedule_dw(&mlxsw_sp->router->nexthop_probe_dw, 0);
+	return 0;
+}
+
+static void mlxsw_sp_neigh_fini(struct mlxsw_sp *mlxsw_sp)
+{
+	cancel_delayed_work_sync(&mlxsw_sp->router->neighs_update.dw);
+	cancel_delayed_work_sync(&mlxsw_sp->router->nexthop_probe_dw);
+	rhashtable_destroy(&mlxsw_sp->router->neigh_ht);
+}
+
+static void mlxsw_sp_neigh_rif_gone_sync(struct mlxsw_sp *mlxsw_sp,
+					 struct mlxsw_sp_rif *rif)
+{
+	struct mlxsw_sp_neigh_entry *neigh_entry, *tmp;
+
+	list_for_each_entry_safe(neigh_entry, tmp, &rif->neigh_list,
+				 rif_list_node) {
+		mlxsw_sp_neigh_entry_update(mlxsw_sp, neigh_entry, false);
+		mlxsw_sp_neigh_entry_destroy(mlxsw_sp, neigh_entry);
+	}
+}
+
+enum mlxsw_sp_nexthop_type {
+	MLXSW_SP_NEXTHOP_TYPE_ETH,
+	MLXSW_SP_NEXTHOP_TYPE_IPIP,
+};
+
+enum mlxsw_sp_nexthop_action {
+	/* Nexthop forwards packets to an egress RIF */
+	MLXSW_SP_NEXTHOP_ACTION_FORWARD,
+	/* Nexthop discards packets */
+	MLXSW_SP_NEXTHOP_ACTION_DISCARD,
+	/* Nexthop traps packets */
+	MLXSW_SP_NEXTHOP_ACTION_TRAP,
+};
+
+struct mlxsw_sp_nexthop_key {
+	struct fib_nh *fib_nh;
+};
+
+struct mlxsw_sp_nexthop {
+	struct list_head neigh_list_node; /* member of neigh entry list */
+	struct list_head rif_list_node;
+	struct list_head router_list_node;
+	struct mlxsw_sp_nexthop_group_info *nhgi; /* pointer back to the group
+						   * this nexthop belongs to
+						   */
+	struct rhash_head ht_node;
+	struct neigh_table *neigh_tbl;
+	struct mlxsw_sp_nexthop_key key;
+	unsigned char gw_addr[sizeof(struct in6_addr)];
+	int ifindex;
+	int nh_weight;
+	int norm_nh_weight;
+	int num_adj_entries;
+	struct mlxsw_sp_rif *rif;
+	u8 should_offload:1, /* set indicates this nexthop should be written
+			      * to the adjacency table.
+			      */
+	   offloaded:1, /* set indicates this nexthop was written to the
+			 * adjacency table.
+			 */
+	   update:1; /* set indicates this nexthop should be updated in the
+		      * adjacency table (f.e., its MAC changed).
+		      */
+	enum mlxsw_sp_nexthop_action action;
+	enum mlxsw_sp_nexthop_type type;
+	union {
+		struct mlxsw_sp_neigh_entry *neigh_entry;
+		struct mlxsw_sp_ipip_entry *ipip_entry;
+	};
+	unsigned int counter_index;
+	bool counter_valid;
+};
+
+enum mlxsw_sp_nexthop_group_type {
+	MLXSW_SP_NEXTHOP_GROUP_TYPE_IPV4,
+	MLXSW_SP_NEXTHOP_GROUP_TYPE_IPV6,
+	MLXSW_SP_NEXTHOP_GROUP_TYPE_OBJ,
+};
+
+struct mlxsw_sp_nexthop_group_info {
+	struct mlxsw_sp_nexthop_group *nh_grp;
+	u32 adj_index;
+	u16 ecmp_size;
+	u16 count;
+	int sum_norm_weight;
+	u8 adj_index_valid:1,
+	   gateway:1, /* routes using the group use a gateway */
+	   is_resilient:1;
+	struct list_head list; /* member in nh_res_grp_list */
+	struct mlxsw_sp_nexthop nexthops[0];
+#define nh_rif	nexthops[0].rif
+};
+
+struct mlxsw_sp_nexthop_group_vr_key {
+	u16 vr_id;
+	enum mlxsw_sp_l3proto proto;
+};
+
+struct mlxsw_sp_nexthop_group_vr_entry {
+	struct list_head list; /* member in vr_list */
+	struct rhash_head ht_node; /* member in vr_ht */
+	refcount_t ref_count;
+	struct mlxsw_sp_nexthop_group_vr_key key;
+};
+
+struct mlxsw_sp_nexthop_group {
+	struct rhash_head ht_node;
+	struct list_head fib_list; /* list of fib entries that use this group */
+	union {
+		struct {
+			struct fib_info *fi;
+		} ipv4;
+		struct {
+			u32 id;
+		} obj;
+	};
+	struct mlxsw_sp_nexthop_group_info *nhgi;
+	struct list_head vr_list;
+	struct rhashtable vr_ht;
+	enum mlxsw_sp_nexthop_group_type type;
+	bool can_destroy;
+};
+
+void mlxsw_sp_nexthop_counter_alloc(struct mlxsw_sp *mlxsw_sp,
+				    struct mlxsw_sp_nexthop *nh)
+{
+	struct devlink *devlink;
+
+	devlink = priv_to_devlink(mlxsw_sp->core);
+	if (!devlink_dpipe_table_counter_enabled(devlink,
+						 MLXSW_SP_DPIPE_TABLE_NAME_ADJ))
+		return;
+
+	if (mlxsw_sp_flow_counter_alloc(mlxsw_sp, &nh->counter_index))
+		return;
+
+	nh->counter_valid = true;
+}
+
+void mlxsw_sp_nexthop_counter_free(struct mlxsw_sp *mlxsw_sp,
+				   struct mlxsw_sp_nexthop *nh)
+{
+	if (!nh->counter_valid)
+		return;
+	mlxsw_sp_flow_counter_free(mlxsw_sp, nh->counter_index);
+	nh->counter_valid = false;
+}
+
+int mlxsw_sp_nexthop_counter_get(struct mlxsw_sp *mlxsw_sp,
+				 struct mlxsw_sp_nexthop *nh, u64 *p_counter)
+{
+	if (!nh->counter_valid)
+		return -EINVAL;
+
+	return mlxsw_sp_flow_counter_get(mlxsw_sp, nh->counter_index,
+					 p_counter, NULL);
+}
+
+struct mlxsw_sp_nexthop *mlxsw_sp_nexthop_next(struct mlxsw_sp_router *router,
+					       struct mlxsw_sp_nexthop *nh)
+{
+	if (!nh) {
+		if (list_empty(&router->nexthop_list))
+			return NULL;
+		else
+			return list_first_entry(&router->nexthop_list,
+						typeof(*nh), router_list_node);
+	}
+	if (list_is_last(&nh->router_list_node, &router->nexthop_list))
+		return NULL;
+	return list_next_entry(nh, router_list_node);
+}
+
+bool mlxsw_sp_nexthop_is_forward(const struct mlxsw_sp_nexthop *nh)
+{
+	return nh->offloaded && nh->action == MLXSW_SP_NEXTHOP_ACTION_FORWARD;
+}
+
+unsigned char *mlxsw_sp_nexthop_ha(struct mlxsw_sp_nexthop *nh)
+{
+	if (nh->type != MLXSW_SP_NEXTHOP_TYPE_ETH ||
+	    !mlxsw_sp_nexthop_is_forward(nh))
+		return NULL;
+	return nh->neigh_entry->ha;
+}
+
+int mlxsw_sp_nexthop_indexes(struct mlxsw_sp_nexthop *nh, u32 *p_adj_index,
+			     u32 *p_adj_size, u32 *p_adj_hash_index)
+{
+	struct mlxsw_sp_nexthop_group_info *nhgi = nh->nhgi;
+	u32 adj_hash_index = 0;
+	int i;
+
+	if (!nh->offloaded || !nhgi->adj_index_valid)
+		return -EINVAL;
+
+	*p_adj_index = nhgi->adj_index;
+	*p_adj_size = nhgi->ecmp_size;
+
+	for (i = 0; i < nhgi->count; i++) {
+		struct mlxsw_sp_nexthop *nh_iter = &nhgi->nexthops[i];
+
+		if (nh_iter == nh)
+			break;
+		if (nh_iter->offloaded)
+			adj_hash_index += nh_iter->num_adj_entries;
+	}
+
+	*p_adj_hash_index = adj_hash_index;
+	return 0;
+}
+
+struct mlxsw_sp_rif *mlxsw_sp_nexthop_rif(struct mlxsw_sp_nexthop *nh)
+{
+	return nh->rif;
+}
+
+bool mlxsw_sp_nexthop_group_has_ipip(struct mlxsw_sp_nexthop *nh)
+{
+	struct mlxsw_sp_nexthop_group_info *nhgi = nh->nhgi;
+	int i;
+
+	for (i = 0; i < nhgi->count; i++) {
+		struct mlxsw_sp_nexthop *nh_iter = &nhgi->nexthops[i];
+
+		if (nh_iter->type == MLXSW_SP_NEXTHOP_TYPE_IPIP)
+			return true;
+	}
+	return false;
+}
+
+static const struct rhashtable_params mlxsw_sp_nexthop_group_vr_ht_params = {
+	.key_offset = offsetof(struct mlxsw_sp_nexthop_group_vr_entry, key),
+	.head_offset = offsetof(struct mlxsw_sp_nexthop_group_vr_entry, ht_node),
+	.key_len = sizeof(struct mlxsw_sp_nexthop_group_vr_key),
+	.automatic_shrinking = true,
+};
+
+static struct mlxsw_sp_nexthop_group_vr_entry *
+mlxsw_sp_nexthop_group_vr_entry_lookup(struct mlxsw_sp_nexthop_group *nh_grp,
+				       const struct mlxsw_sp_fib *fib)
+{
+	struct mlxsw_sp_nexthop_group_vr_key key;
+
+	memset(&key, 0, sizeof(key));
+	key.vr_id = fib->vr->id;
+	key.proto = fib->proto;
+	return rhashtable_lookup_fast(&nh_grp->vr_ht, &key,
+				      mlxsw_sp_nexthop_group_vr_ht_params);
+}
+
+static int
+mlxsw_sp_nexthop_group_vr_entry_create(struct mlxsw_sp_nexthop_group *nh_grp,
+				       const struct mlxsw_sp_fib *fib)
+{
+	struct mlxsw_sp_nexthop_group_vr_entry *vr_entry;
+	int err;
+
+	vr_entry = kzalloc(sizeof(*vr_entry), GFP_KERNEL);
+	if (!vr_entry)
+		return -ENOMEM;
+
+	vr_entry->key.vr_id = fib->vr->id;
+	vr_entry->key.proto = fib->proto;
+	refcount_set(&vr_entry->ref_count, 1);
+
+	err = rhashtable_insert_fast(&nh_grp->vr_ht, &vr_entry->ht_node,
+				     mlxsw_sp_nexthop_group_vr_ht_params);
+	if (err)
+		goto err_hashtable_insert;
+
+	list_add(&vr_entry->list, &nh_grp->vr_list);
+
+	return 0;
+
+err_hashtable_insert:
+	kfree(vr_entry);
+	return err;
+}
+
+static void
+mlxsw_sp_nexthop_group_vr_entry_destroy(struct mlxsw_sp_nexthop_group *nh_grp,
+					struct mlxsw_sp_nexthop_group_vr_entry *vr_entry)
+{
+	list_del(&vr_entry->list);
+	rhashtable_remove_fast(&nh_grp->vr_ht, &vr_entry->ht_node,
+			       mlxsw_sp_nexthop_group_vr_ht_params);
+	kfree(vr_entry);
+}
+
+static int
+mlxsw_sp_nexthop_group_vr_link(struct mlxsw_sp_nexthop_group *nh_grp,
+			       const struct mlxsw_sp_fib *fib)
+{
+	struct mlxsw_sp_nexthop_group_vr_entry *vr_entry;
+
+	vr_entry = mlxsw_sp_nexthop_group_vr_entry_lookup(nh_grp, fib);
+	if (vr_entry) {
+		refcount_inc(&vr_entry->ref_count);
+		return 0;
+	}
+
+	return mlxsw_sp_nexthop_group_vr_entry_create(nh_grp, fib);
+}
+
+static void
+mlxsw_sp_nexthop_group_vr_unlink(struct mlxsw_sp_nexthop_group *nh_grp,
+				 const struct mlxsw_sp_fib *fib)
+{
+	struct mlxsw_sp_nexthop_group_vr_entry *vr_entry;
+
+	vr_entry = mlxsw_sp_nexthop_group_vr_entry_lookup(nh_grp, fib);
+	if (WARN_ON_ONCE(!vr_entry))
+		return;
+
+	if (!refcount_dec_and_test(&vr_entry->ref_count))
+		return;
+
+	mlxsw_sp_nexthop_group_vr_entry_destroy(nh_grp, vr_entry);
+}
+
+struct mlxsw_sp_nexthop_group_cmp_arg {
+	enum mlxsw_sp_nexthop_group_type type;
+	union {
+		struct fib_info *fi;
+		struct mlxsw_sp_fib6_entry *fib6_entry;
+		u32 id;
+	};
+};
+
+static bool
+mlxsw_sp_nexthop6_group_has_nexthop(const struct mlxsw_sp_nexthop_group *nh_grp,
+				    const struct in6_addr *gw, int ifindex,
+				    int weight)
+{
+	int i;
+
+	for (i = 0; i < nh_grp->nhgi->count; i++) {
+		const struct mlxsw_sp_nexthop *nh;
+
+		nh = &nh_grp->nhgi->nexthops[i];
+		if (nh->ifindex == ifindex && nh->nh_weight == weight &&
+		    ipv6_addr_equal(gw, (struct in6_addr *) nh->gw_addr))
+			return true;
+	}
+
+	return false;
+}
+
+static bool
+mlxsw_sp_nexthop6_group_cmp(const struct mlxsw_sp_nexthop_group *nh_grp,
+			    const struct mlxsw_sp_fib6_entry *fib6_entry)
+{
+	struct mlxsw_sp_rt6 *mlxsw_sp_rt6;
+
+	if (nh_grp->nhgi->count != fib6_entry->nrt6)
+		return false;
+
+	list_for_each_entry(mlxsw_sp_rt6, &fib6_entry->rt6_list, list) {
+		struct fib6_nh *fib6_nh = mlxsw_sp_rt6->rt->fib6_nh;
+		struct in6_addr *gw;
+		int ifindex, weight;
+
+		ifindex = fib6_nh->fib_nh_dev->ifindex;
+		weight = fib6_nh->fib_nh_weight;
+		gw = &fib6_nh->fib_nh_gw6;
+		if (!mlxsw_sp_nexthop6_group_has_nexthop(nh_grp, gw, ifindex,
+							 weight))
+			return false;
+	}
+
+	return true;
+}
+
+static int
+mlxsw_sp_nexthop_group_cmp(struct rhashtable_compare_arg *arg, const void *ptr)
+{
+	const struct mlxsw_sp_nexthop_group_cmp_arg *cmp_arg = arg->key;
+	const struct mlxsw_sp_nexthop_group *nh_grp = ptr;
+
+	if (nh_grp->type != cmp_arg->type)
+		return 1;
+
+	switch (cmp_arg->type) {
+	case MLXSW_SP_NEXTHOP_GROUP_TYPE_IPV4:
+		return cmp_arg->fi != nh_grp->ipv4.fi;
+	case MLXSW_SP_NEXTHOP_GROUP_TYPE_IPV6:
+		return !mlxsw_sp_nexthop6_group_cmp(nh_grp,
+						    cmp_arg->fib6_entry);
+	case MLXSW_SP_NEXTHOP_GROUP_TYPE_OBJ:
+		return cmp_arg->id != nh_grp->obj.id;
+	default:
+		WARN_ON(1);
+		return 1;
+	}
+}
+
+static u32 mlxsw_sp_nexthop_group_hash_obj(const void *data, u32 len, u32 seed)
+{
+	const struct mlxsw_sp_nexthop_group *nh_grp = data;
+	const struct mlxsw_sp_nexthop *nh;
+	struct fib_info *fi;
+	unsigned int val;
+	int i;
+
+	switch (nh_grp->type) {
+	case MLXSW_SP_NEXTHOP_GROUP_TYPE_IPV4:
+		fi = nh_grp->ipv4.fi;
+		return jhash(&fi, sizeof(fi), seed);
+	case MLXSW_SP_NEXTHOP_GROUP_TYPE_IPV6:
+		val = nh_grp->nhgi->count;
+		for (i = 0; i < nh_grp->nhgi->count; i++) {
+			nh = &nh_grp->nhgi->nexthops[i];
+			val ^= jhash(&nh->ifindex, sizeof(nh->ifindex), seed);
+			val ^= jhash(&nh->gw_addr, sizeof(nh->gw_addr), seed);
+		}
+		return jhash(&val, sizeof(val), seed);
+	case MLXSW_SP_NEXTHOP_GROUP_TYPE_OBJ:
+		return jhash(&nh_grp->obj.id, sizeof(nh_grp->obj.id), seed);
+	default:
+		WARN_ON(1);
+		return 0;
+	}
+}
+
+static u32
+mlxsw_sp_nexthop6_group_hash(struct mlxsw_sp_fib6_entry *fib6_entry, u32 seed)
+{
+	unsigned int val = fib6_entry->nrt6;
+	struct mlxsw_sp_rt6 *mlxsw_sp_rt6;
+
+	list_for_each_entry(mlxsw_sp_rt6, &fib6_entry->rt6_list, list) {
+		struct fib6_nh *fib6_nh = mlxsw_sp_rt6->rt->fib6_nh;
+		struct net_device *dev = fib6_nh->fib_nh_dev;
+		struct in6_addr *gw = &fib6_nh->fib_nh_gw6;
+
+		val ^= jhash(&dev->ifindex, sizeof(dev->ifindex), seed);
+		val ^= jhash(gw, sizeof(*gw), seed);
+	}
+
+	return jhash(&val, sizeof(val), seed);
+}
+
+static u32
+mlxsw_sp_nexthop_group_hash(const void *data, u32 len, u32 seed)
+{
+	const struct mlxsw_sp_nexthop_group_cmp_arg *cmp_arg = data;
+
+	switch (cmp_arg->type) {
+	case MLXSW_SP_NEXTHOP_GROUP_TYPE_IPV4:
+		return jhash(&cmp_arg->fi, sizeof(cmp_arg->fi), seed);
+	case MLXSW_SP_NEXTHOP_GROUP_TYPE_IPV6:
+		return mlxsw_sp_nexthop6_group_hash(cmp_arg->fib6_entry, seed);
+	case MLXSW_SP_NEXTHOP_GROUP_TYPE_OBJ:
+		return jhash(&cmp_arg->id, sizeof(cmp_arg->id), seed);
+	default:
+		WARN_ON(1);
+		return 0;
+	}
+}
+
+static const struct rhashtable_params mlxsw_sp_nexthop_group_ht_params = {
+	.head_offset = offsetof(struct mlxsw_sp_nexthop_group, ht_node),
+	.hashfn	     = mlxsw_sp_nexthop_group_hash,
+	.obj_hashfn  = mlxsw_sp_nexthop_group_hash_obj,
+	.obj_cmpfn   = mlxsw_sp_nexthop_group_cmp,
+};
+
+static int mlxsw_sp_nexthop_group_insert(struct mlxsw_sp *mlxsw_sp,
+					 struct mlxsw_sp_nexthop_group *nh_grp)
+{
+	if (nh_grp->type == MLXSW_SP_NEXTHOP_GROUP_TYPE_IPV6 &&
+	    !nh_grp->nhgi->gateway)
+		return 0;
+
+	return rhashtable_insert_fast(&mlxsw_sp->router->nexthop_group_ht,
+				      &nh_grp->ht_node,
+				      mlxsw_sp_nexthop_group_ht_params);
+}
+
+static void mlxsw_sp_nexthop_group_remove(struct mlxsw_sp *mlxsw_sp,
+					  struct mlxsw_sp_nexthop_group *nh_grp)
+{
+	if (nh_grp->type == MLXSW_SP_NEXTHOP_GROUP_TYPE_IPV6 &&
+	    !nh_grp->nhgi->gateway)
+		return;
+
+	rhashtable_remove_fast(&mlxsw_sp->router->nexthop_group_ht,
+			       &nh_grp->ht_node,
+			       mlxsw_sp_nexthop_group_ht_params);
+}
+
+static struct mlxsw_sp_nexthop_group *
+mlxsw_sp_nexthop4_group_lookup(struct mlxsw_sp *mlxsw_sp,
+			       struct fib_info *fi)
+{
+	struct mlxsw_sp_nexthop_group_cmp_arg cmp_arg;
+
+	cmp_arg.type = MLXSW_SP_NEXTHOP_GROUP_TYPE_IPV4;
+	cmp_arg.fi = fi;
+	return rhashtable_lookup_fast(&mlxsw_sp->router->nexthop_group_ht,
+				      &cmp_arg,
+				      mlxsw_sp_nexthop_group_ht_params);
+}
+
+static struct mlxsw_sp_nexthop_group *
+mlxsw_sp_nexthop6_group_lookup(struct mlxsw_sp *mlxsw_sp,
+			       struct mlxsw_sp_fib6_entry *fib6_entry)
+{
+	struct mlxsw_sp_nexthop_group_cmp_arg cmp_arg;
+
+	cmp_arg.type = MLXSW_SP_NEXTHOP_GROUP_TYPE_IPV6;
+	cmp_arg.fib6_entry = fib6_entry;
+	return rhashtable_lookup_fast(&mlxsw_sp->router->nexthop_group_ht,
+				      &cmp_arg,
+				      mlxsw_sp_nexthop_group_ht_params);
+}
+
+static const struct rhashtable_params mlxsw_sp_nexthop_ht_params = {
+	.key_offset = offsetof(struct mlxsw_sp_nexthop, key),
+	.head_offset = offsetof(struct mlxsw_sp_nexthop, ht_node),
+	.key_len = sizeof(struct mlxsw_sp_nexthop_key),
+};
+
+static int mlxsw_sp_nexthop_insert(struct mlxsw_sp *mlxsw_sp,
+				   struct mlxsw_sp_nexthop *nh)
+{
+	return rhashtable_insert_fast(&mlxsw_sp->router->nexthop_ht,
+				      &nh->ht_node, mlxsw_sp_nexthop_ht_params);
+}
+
+static void mlxsw_sp_nexthop_remove(struct mlxsw_sp *mlxsw_sp,
+				    struct mlxsw_sp_nexthop *nh)
+{
+	rhashtable_remove_fast(&mlxsw_sp->router->nexthop_ht, &nh->ht_node,
+			       mlxsw_sp_nexthop_ht_params);
+}
+
+static struct mlxsw_sp_nexthop *
+mlxsw_sp_nexthop_lookup(struct mlxsw_sp *mlxsw_sp,
+			struct mlxsw_sp_nexthop_key key)
+{
+	return rhashtable_lookup_fast(&mlxsw_sp->router->nexthop_ht, &key,
+				      mlxsw_sp_nexthop_ht_params);
+}
+
+static int mlxsw_sp_adj_index_mass_update_vr(struct mlxsw_sp *mlxsw_sp,
+					     enum mlxsw_sp_l3proto proto,
+					     u16 vr_id,
+					     u32 adj_index, u16 ecmp_size,
+					     u32 new_adj_index,
+					     u16 new_ecmp_size)
+{
+	char raleu_pl[MLXSW_REG_RALEU_LEN];
+
+	mlxsw_reg_raleu_pack(raleu_pl,
+			     (enum mlxsw_reg_ralxx_protocol) proto, vr_id,
+			     adj_index, ecmp_size, new_adj_index,
+			     new_ecmp_size);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(raleu), raleu_pl);
+}
+
+static int mlxsw_sp_adj_index_mass_update(struct mlxsw_sp *mlxsw_sp,
+					  struct mlxsw_sp_nexthop_group *nh_grp,
+					  u32 old_adj_index, u16 old_ecmp_size)
+{
+	struct mlxsw_sp_nexthop_group_info *nhgi = nh_grp->nhgi;
+	struct mlxsw_sp_nexthop_group_vr_entry *vr_entry;
+	int err;
+
+	list_for_each_entry(vr_entry, &nh_grp->vr_list, list) {
+		err = mlxsw_sp_adj_index_mass_update_vr(mlxsw_sp,
+							vr_entry->key.proto,
+							vr_entry->key.vr_id,
+							old_adj_index,
+							old_ecmp_size,
+							nhgi->adj_index,
+							nhgi->ecmp_size);
+		if (err)
+			goto err_mass_update_vr;
+	}
+	return 0;
+
+err_mass_update_vr:
+	list_for_each_entry_continue_reverse(vr_entry, &nh_grp->vr_list, list)
+		mlxsw_sp_adj_index_mass_update_vr(mlxsw_sp, vr_entry->key.proto,
+						  vr_entry->key.vr_id,
+						  nhgi->adj_index,
+						  nhgi->ecmp_size,
+						  old_adj_index, old_ecmp_size);
+	return err;
+}
+
+static int __mlxsw_sp_nexthop_eth_update(struct mlxsw_sp *mlxsw_sp,
+					 u32 adj_index,
+					 struct mlxsw_sp_nexthop *nh,
+					 bool force, char *ratr_pl)
+{
+	struct mlxsw_sp_neigh_entry *neigh_entry = nh->neigh_entry;
+	enum mlxsw_reg_ratr_op op;
+	u16 rif_index;
+
+	rif_index = nh->rif ? nh->rif->rif_index :
+			      mlxsw_sp->router->lb_rif_index;
+	op = force ? MLXSW_REG_RATR_OP_WRITE_WRITE_ENTRY :
+		     MLXSW_REG_RATR_OP_WRITE_WRITE_ENTRY_ON_ACTIVITY;
+	mlxsw_reg_ratr_pack(ratr_pl, op, true, MLXSW_REG_RATR_TYPE_ETHERNET,
+			    adj_index, rif_index);
+	switch (nh->action) {
+	case MLXSW_SP_NEXTHOP_ACTION_FORWARD:
+		mlxsw_reg_ratr_eth_entry_pack(ratr_pl, neigh_entry->ha);
+		break;
+	case MLXSW_SP_NEXTHOP_ACTION_DISCARD:
+		mlxsw_reg_ratr_trap_action_set(ratr_pl,
+					       MLXSW_REG_RATR_TRAP_ACTION_DISCARD_ERRORS);
+		break;
+	case MLXSW_SP_NEXTHOP_ACTION_TRAP:
+		mlxsw_reg_ratr_trap_action_set(ratr_pl,
+					       MLXSW_REG_RATR_TRAP_ACTION_TRAP);
+		mlxsw_reg_ratr_trap_id_set(ratr_pl, MLXSW_TRAP_ID_RTR_EGRESS0);
+		break;
+	default:
+		WARN_ON_ONCE(1);
+		return -EINVAL;
+	}
+	if (nh->counter_valid)
+		mlxsw_reg_ratr_counter_pack(ratr_pl, nh->counter_index, true);
+	else
+		mlxsw_reg_ratr_counter_pack(ratr_pl, 0, false);
+
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(ratr), ratr_pl);
+}
+
+int mlxsw_sp_nexthop_eth_update(struct mlxsw_sp *mlxsw_sp, u32 adj_index,
+				struct mlxsw_sp_nexthop *nh, bool force,
+				char *ratr_pl)
+{
+	int i;
+
+	for (i = 0; i < nh->num_adj_entries; i++) {
+		int err;
+
+		err = __mlxsw_sp_nexthop_eth_update(mlxsw_sp, adj_index + i,
+						    nh, force, ratr_pl);
+		if (err)
+			return err;
+	}
+
+	return 0;
+}
+
+static int __mlxsw_sp_nexthop_ipip_update(struct mlxsw_sp *mlxsw_sp,
+					  u32 adj_index,
+					  struct mlxsw_sp_nexthop *nh,
+					  bool force, char *ratr_pl)
+{
+	const struct mlxsw_sp_ipip_ops *ipip_ops;
+
+	ipip_ops = mlxsw_sp->router->ipip_ops_arr[nh->ipip_entry->ipipt];
+	return ipip_ops->nexthop_update(mlxsw_sp, adj_index, nh->ipip_entry,
+					force, ratr_pl);
+}
+
+static int mlxsw_sp_nexthop_ipip_update(struct mlxsw_sp *mlxsw_sp,
+					u32 adj_index,
+					struct mlxsw_sp_nexthop *nh, bool force,
+					char *ratr_pl)
+{
+	int i;
+
+	for (i = 0; i < nh->num_adj_entries; i++) {
+		int err;
+
+		err = __mlxsw_sp_nexthop_ipip_update(mlxsw_sp, adj_index + i,
+						     nh, force, ratr_pl);
+		if (err)
+			return err;
+	}
+
+	return 0;
+}
+
+static int mlxsw_sp_nexthop_update(struct mlxsw_sp *mlxsw_sp, u32 adj_index,
+				   struct mlxsw_sp_nexthop *nh, bool force,
+				   char *ratr_pl)
+{
+	/* When action is discard or trap, the nexthop must be
+	 * programmed as an Ethernet nexthop.
+	 */
+	if (nh->type == MLXSW_SP_NEXTHOP_TYPE_ETH ||
+	    nh->action == MLXSW_SP_NEXTHOP_ACTION_DISCARD ||
+	    nh->action == MLXSW_SP_NEXTHOP_ACTION_TRAP)
+		return mlxsw_sp_nexthop_eth_update(mlxsw_sp, adj_index, nh,
+						   force, ratr_pl);
+	else
+		return mlxsw_sp_nexthop_ipip_update(mlxsw_sp, adj_index, nh,
+						    force, ratr_pl);
+}
+
+static int
+mlxsw_sp_nexthop_group_update(struct mlxsw_sp *mlxsw_sp,
+			      struct mlxsw_sp_nexthop_group_info *nhgi,
+			      bool reallocate)
+{
+	char ratr_pl[MLXSW_REG_RATR_LEN];
+	u32 adj_index = nhgi->adj_index; /* base */
+	struct mlxsw_sp_nexthop *nh;
+	int i;
+
+	for (i = 0; i < nhgi->count; i++) {
+		nh = &nhgi->nexthops[i];
+
+		if (!nh->should_offload) {
+			nh->offloaded = 0;
+			continue;
+		}
+
+		if (nh->update || reallocate) {
+			int err = 0;
+
+			err = mlxsw_sp_nexthop_update(mlxsw_sp, adj_index, nh,
+						      true, ratr_pl);
+			if (err)
+				return err;
+			nh->update = 0;
+			nh->offloaded = 1;
+		}
+		adj_index += nh->num_adj_entries;
+	}
+	return 0;
+}
+
+static int
+mlxsw_sp_nexthop_fib_entries_update(struct mlxsw_sp *mlxsw_sp,
+				    struct mlxsw_sp_nexthop_group *nh_grp)
+{
+	struct mlxsw_sp_fib_entry *fib_entry;
+	int err;
+
+	list_for_each_entry(fib_entry, &nh_grp->fib_list, nexthop_group_node) {
+		err = mlxsw_sp_fib_entry_update(mlxsw_sp, fib_entry);
+		if (err)
+			return err;
+	}
+	return 0;
+}
+
+struct mlxsw_sp_adj_grp_size_range {
+	u16 start; /* Inclusive */
+	u16 end; /* Inclusive */
+};
+
+/* Ordered by range start value */
+static const struct mlxsw_sp_adj_grp_size_range
+mlxsw_sp1_adj_grp_size_ranges[] = {
+	{ .start = 1, .end = 64 },
+	{ .start = 512, .end = 512 },
+	{ .start = 1024, .end = 1024 },
+	{ .start = 2048, .end = 2048 },
+	{ .start = 4096, .end = 4096 },
+};
+
+/* Ordered by range start value */
+static const struct mlxsw_sp_adj_grp_size_range
+mlxsw_sp2_adj_grp_size_ranges[] = {
+	{ .start = 1, .end = 128 },
+	{ .start = 256, .end = 256 },
+	{ .start = 512, .end = 512 },
+	{ .start = 1024, .end = 1024 },
+	{ .start = 2048, .end = 2048 },
+	{ .start = 4096, .end = 4096 },
+};
+
+static void mlxsw_sp_adj_grp_size_round_up(const struct mlxsw_sp *mlxsw_sp,
+					   u16 *p_adj_grp_size)
+{
+	int i;
+
+	for (i = 0; i < mlxsw_sp->router->adj_grp_size_ranges_count; i++) {
+		const struct mlxsw_sp_adj_grp_size_range *size_range;
+
+		size_range = &mlxsw_sp->router->adj_grp_size_ranges[i];
+
+		if (*p_adj_grp_size >= size_range->start &&
+		    *p_adj_grp_size <= size_range->end)
+			return;
+
+		if (*p_adj_grp_size <= size_range->end) {
+			*p_adj_grp_size = size_range->end;
+			return;
+		}
+	}
+}
+
+static void mlxsw_sp_adj_grp_size_round_down(const struct mlxsw_sp *mlxsw_sp,
+					     u16 *p_adj_grp_size,
+					     unsigned int alloc_size)
+{
+	int i;
+
+	for (i = mlxsw_sp->router->adj_grp_size_ranges_count - 1; i >= 0; i--) {
+		const struct mlxsw_sp_adj_grp_size_range *size_range;
+
+		size_range = &mlxsw_sp->router->adj_grp_size_ranges[i];
+
+		if (alloc_size >= size_range->end) {
+			*p_adj_grp_size = size_range->end;
+			return;
+		}
+	}
+}
+
+static int mlxsw_sp_fix_adj_grp_size(struct mlxsw_sp *mlxsw_sp,
+				     u16 *p_adj_grp_size)
+{
+	unsigned int alloc_size;
+	int err;
+
+	/* Round up the requested group size to the next size supported
+	 * by the device and make sure the request can be satisfied.
+	 */
+	mlxsw_sp_adj_grp_size_round_up(mlxsw_sp, p_adj_grp_size);
+	err = mlxsw_sp_kvdl_alloc_count_query(mlxsw_sp,
+					      MLXSW_SP_KVDL_ENTRY_TYPE_ADJ,
+					      *p_adj_grp_size, &alloc_size);
+	if (err)
+		return err;
+	/* It is possible the allocation results in more allocated
+	 * entries than requested. Try to use as much of them as
+	 * possible.
+	 */
+	mlxsw_sp_adj_grp_size_round_down(mlxsw_sp, p_adj_grp_size, alloc_size);
+
+	return 0;
+}
+
+static void
+mlxsw_sp_nexthop_group_normalize(struct mlxsw_sp_nexthop_group_info *nhgi)
+{
+	int i, g = 0, sum_norm_weight = 0;
+	struct mlxsw_sp_nexthop *nh;
+
+	for (i = 0; i < nhgi->count; i++) {
+		nh = &nhgi->nexthops[i];
+
+		if (!nh->should_offload)
+			continue;
+		if (g > 0)
+			g = gcd(nh->nh_weight, g);
+		else
+			g = nh->nh_weight;
+	}
+
+	for (i = 0; i < nhgi->count; i++) {
+		nh = &nhgi->nexthops[i];
+
+		if (!nh->should_offload)
+			continue;
+		nh->norm_nh_weight = nh->nh_weight / g;
+		sum_norm_weight += nh->norm_nh_weight;
+	}
+
+	nhgi->sum_norm_weight = sum_norm_weight;
+}
+
+static void
+mlxsw_sp_nexthop_group_rebalance(struct mlxsw_sp_nexthop_group_info *nhgi)
+{
+	int i, weight = 0, lower_bound = 0;
+	int total = nhgi->sum_norm_weight;
+	u16 ecmp_size = nhgi->ecmp_size;
+
+	for (i = 0; i < nhgi->count; i++) {
+		struct mlxsw_sp_nexthop *nh = &nhgi->nexthops[i];
+		int upper_bound;
+
+		if (!nh->should_offload)
+			continue;
+		weight += nh->norm_nh_weight;
+		upper_bound = DIV_ROUND_CLOSEST(ecmp_size * weight, total);
+		nh->num_adj_entries = upper_bound - lower_bound;
+		lower_bound = upper_bound;
+	}
+}
+
+static struct mlxsw_sp_nexthop *
+mlxsw_sp_rt6_nexthop(struct mlxsw_sp_nexthop_group *nh_grp,
+		     const struct mlxsw_sp_rt6 *mlxsw_sp_rt6);
+
+static void
+mlxsw_sp_nexthop4_group_offload_refresh(struct mlxsw_sp *mlxsw_sp,
+					struct mlxsw_sp_nexthop_group *nh_grp)
+{
+	int i;
+
+	for (i = 0; i < nh_grp->nhgi->count; i++) {
+		struct mlxsw_sp_nexthop *nh = &nh_grp->nhgi->nexthops[i];
+
+		if (nh->offloaded)
+			nh->key.fib_nh->fib_nh_flags |= RTNH_F_OFFLOAD;
+		else
+			nh->key.fib_nh->fib_nh_flags &= ~RTNH_F_OFFLOAD;
+	}
+}
+
+static void
+__mlxsw_sp_nexthop6_group_offload_refresh(struct mlxsw_sp_nexthop_group *nh_grp,
+					  struct mlxsw_sp_fib6_entry *fib6_entry)
+{
+	struct mlxsw_sp_rt6 *mlxsw_sp_rt6;
+
+	list_for_each_entry(mlxsw_sp_rt6, &fib6_entry->rt6_list, list) {
+		struct fib6_nh *fib6_nh = mlxsw_sp_rt6->rt->fib6_nh;
+		struct mlxsw_sp_nexthop *nh;
+
+		nh = mlxsw_sp_rt6_nexthop(nh_grp, mlxsw_sp_rt6);
+		if (nh && nh->offloaded)
+			fib6_nh->fib_nh_flags |= RTNH_F_OFFLOAD;
+		else
+			fib6_nh->fib_nh_flags &= ~RTNH_F_OFFLOAD;
+	}
+}
+
+static void
+mlxsw_sp_nexthop6_group_offload_refresh(struct mlxsw_sp *mlxsw_sp,
+					struct mlxsw_sp_nexthop_group *nh_grp)
+{
+	struct mlxsw_sp_fib6_entry *fib6_entry;
+
+	/* Unfortunately, in IPv6 the route and the nexthop are described by
+	 * the same struct, so we need to iterate over all the routes using the
+	 * nexthop group and set / clear the offload indication for them.
+	 */
+	list_for_each_entry(fib6_entry, &nh_grp->fib_list,
+			    common.nexthop_group_node)
+		__mlxsw_sp_nexthop6_group_offload_refresh(nh_grp, fib6_entry);
+}
+
+static void
+mlxsw_sp_nexthop_bucket_offload_refresh(struct mlxsw_sp *mlxsw_sp,
+					const struct mlxsw_sp_nexthop *nh,
+					u16 bucket_index)
+{
+	struct mlxsw_sp_nexthop_group *nh_grp = nh->nhgi->nh_grp;
+	bool offload = false, trap = false;
+
+	if (nh->offloaded) {
+		if (nh->action == MLXSW_SP_NEXTHOP_ACTION_TRAP)
+			trap = true;
+		else
+			offload = true;
+	}
+	nexthop_bucket_set_hw_flags(mlxsw_sp_net(mlxsw_sp), nh_grp->obj.id,
+				    bucket_index, offload, trap);
+}
+
+static void
+mlxsw_sp_nexthop_obj_group_offload_refresh(struct mlxsw_sp *mlxsw_sp,
+					   struct mlxsw_sp_nexthop_group *nh_grp)
+{
+	int i;
+
+	/* Do not update the flags if the nexthop group is being destroyed
+	 * since:
+	 * 1. The nexthop objects is being deleted, in which case the flags are
+	 * irrelevant.
+	 * 2. The nexthop group was replaced by a newer group, in which case
+	 * the flags of the nexthop object were already updated based on the
+	 * new group.
+	 */
+	if (nh_grp->can_destroy)
+		return;
+
+	nexthop_set_hw_flags(mlxsw_sp_net(mlxsw_sp), nh_grp->obj.id,
+			     nh_grp->nhgi->adj_index_valid, false);
+
+	/* Update flags of individual nexthop buckets in case of a resilient
+	 * nexthop group.
+	 */
+	if (!nh_grp->nhgi->is_resilient)
+		return;
+
+	for (i = 0; i < nh_grp->nhgi->count; i++) {
+		struct mlxsw_sp_nexthop *nh = &nh_grp->nhgi->nexthops[i];
+
+		mlxsw_sp_nexthop_bucket_offload_refresh(mlxsw_sp, nh, i);
+	}
+}
+
+static void
+mlxsw_sp_nexthop_group_offload_refresh(struct mlxsw_sp *mlxsw_sp,
+				       struct mlxsw_sp_nexthop_group *nh_grp)
+{
+	switch (nh_grp->type) {
+	case MLXSW_SP_NEXTHOP_GROUP_TYPE_IPV4:
+		mlxsw_sp_nexthop4_group_offload_refresh(mlxsw_sp, nh_grp);
+		break;
+	case MLXSW_SP_NEXTHOP_GROUP_TYPE_IPV6:
+		mlxsw_sp_nexthop6_group_offload_refresh(mlxsw_sp, nh_grp);
+		break;
+	case MLXSW_SP_NEXTHOP_GROUP_TYPE_OBJ:
+		mlxsw_sp_nexthop_obj_group_offload_refresh(mlxsw_sp, nh_grp);
+		break;
+	}
+}
+
+static int
+mlxsw_sp_nexthop_group_refresh(struct mlxsw_sp *mlxsw_sp,
+			       struct mlxsw_sp_nexthop_group *nh_grp)
+{
+	struct mlxsw_sp_nexthop_group_info *nhgi = nh_grp->nhgi;
+	u16 ecmp_size, old_ecmp_size;
+	struct mlxsw_sp_nexthop *nh;
+	bool offload_change = false;
+	u32 adj_index;
+	bool old_adj_index_valid;
+	u32 old_adj_index;
+	int i, err2, err;
+
+	if (!nhgi->gateway)
+		return mlxsw_sp_nexthop_fib_entries_update(mlxsw_sp, nh_grp);
+
+	for (i = 0; i < nhgi->count; i++) {
+		nh = &nhgi->nexthops[i];
+
+		if (nh->should_offload != nh->offloaded) {
+			offload_change = true;
+			if (nh->should_offload)
+				nh->update = 1;
+		}
+	}
+	if (!offload_change) {
+		/* Nothing was added or removed, so no need to reallocate. Just
+		 * update MAC on existing adjacency indexes.
+		 */
+		err = mlxsw_sp_nexthop_group_update(mlxsw_sp, nhgi, false);
+		if (err) {
+			dev_warn(mlxsw_sp->bus_info->dev, "Failed to update neigh MAC in adjacency table.\n");
+			goto set_trap;
+		}
+		/* Flags of individual nexthop buckets might need to be
+		 * updated.
+		 */
+		mlxsw_sp_nexthop_group_offload_refresh(mlxsw_sp, nh_grp);
+		return 0;
+	}
+	mlxsw_sp_nexthop_group_normalize(nhgi);
+	if (!nhgi->sum_norm_weight) {
+		/* No neigh of this group is connected so we just set
+		 * the trap and let everthing flow through kernel.
+		 */
+		err = 0;
+		goto set_trap;
+	}
+
+	ecmp_size = nhgi->sum_norm_weight;
+	err = mlxsw_sp_fix_adj_grp_size(mlxsw_sp, &ecmp_size);
+	if (err)
+		/* No valid allocation size available. */
+		goto set_trap;
+
+	err = mlxsw_sp_kvdl_alloc(mlxsw_sp, MLXSW_SP_KVDL_ENTRY_TYPE_ADJ,
+				  ecmp_size, &adj_index);
+	if (err) {
+		/* We ran out of KVD linear space, just set the
+		 * trap and let everything flow through kernel.
+		 */
+		dev_warn(mlxsw_sp->bus_info->dev, "Failed to allocate KVD linear area for nexthop group.\n");
+		goto set_trap;
+	}
+	old_adj_index_valid = nhgi->adj_index_valid;
+	old_adj_index = nhgi->adj_index;
+	old_ecmp_size = nhgi->ecmp_size;
+	nhgi->adj_index_valid = 1;
+	nhgi->adj_index = adj_index;
+	nhgi->ecmp_size = ecmp_size;
+	mlxsw_sp_nexthop_group_rebalance(nhgi);
+	err = mlxsw_sp_nexthop_group_update(mlxsw_sp, nhgi, true);
+	if (err) {
+		dev_warn(mlxsw_sp->bus_info->dev, "Failed to update neigh MAC in adjacency table.\n");
+		goto set_trap;
+	}
+
+	mlxsw_sp_nexthop_group_offload_refresh(mlxsw_sp, nh_grp);
+
+	if (!old_adj_index_valid) {
+		/* The trap was set for fib entries, so we have to call
+		 * fib entry update to unset it and use adjacency index.
+		 */
+		err = mlxsw_sp_nexthop_fib_entries_update(mlxsw_sp, nh_grp);
+		if (err) {
+			dev_warn(mlxsw_sp->bus_info->dev, "Failed to add adjacency index to fib entries.\n");
+			goto set_trap;
+		}
+		return 0;
+	}
+
+	err = mlxsw_sp_adj_index_mass_update(mlxsw_sp, nh_grp,
+					     old_adj_index, old_ecmp_size);
+	mlxsw_sp_kvdl_free(mlxsw_sp, MLXSW_SP_KVDL_ENTRY_TYPE_ADJ,
+			   old_ecmp_size, old_adj_index);
+	if (err) {
+		dev_warn(mlxsw_sp->bus_info->dev, "Failed to mass-update adjacency index for nexthop group.\n");
+		goto set_trap;
+	}
+
+	return 0;
+
+set_trap:
+	old_adj_index_valid = nhgi->adj_index_valid;
+	nhgi->adj_index_valid = 0;
+	for (i = 0; i < nhgi->count; i++) {
+		nh = &nhgi->nexthops[i];
+		nh->offloaded = 0;
+	}
+	err2 = mlxsw_sp_nexthop_fib_entries_update(mlxsw_sp, nh_grp);
+	if (err2)
+		dev_warn(mlxsw_sp->bus_info->dev, "Failed to set traps for fib entries.\n");
+	mlxsw_sp_nexthop_group_offload_refresh(mlxsw_sp, nh_grp);
+	if (old_adj_index_valid)
+		mlxsw_sp_kvdl_free(mlxsw_sp, MLXSW_SP_KVDL_ENTRY_TYPE_ADJ,
+				   nhgi->ecmp_size, nhgi->adj_index);
+	return err;
+}
+
+static void __mlxsw_sp_nexthop_neigh_update(struct mlxsw_sp_nexthop *nh,
+					    bool removing)
+{
+	if (!removing) {
+		nh->action = MLXSW_SP_NEXTHOP_ACTION_FORWARD;
+		nh->should_offload = 1;
+	} else if (nh->nhgi->is_resilient) {
+		nh->action = MLXSW_SP_NEXTHOP_ACTION_TRAP;
+		nh->should_offload = 1;
+	} else {
+		nh->should_offload = 0;
+	}
+	nh->update = 1;
+}
+
+static int
+mlxsw_sp_nexthop_dead_neigh_replace(struct mlxsw_sp *mlxsw_sp,
+				    struct mlxsw_sp_neigh_entry *neigh_entry)
+{
+	struct neighbour *n, *old_n = neigh_entry->key.n;
+	struct mlxsw_sp_nexthop *nh;
+	bool entry_connected;
+	u8 nud_state, dead;
+	int err;
+
+	nh = list_first_entry(&neigh_entry->nexthop_list,
+			      struct mlxsw_sp_nexthop, neigh_list_node);
+
+	n = neigh_lookup(nh->neigh_tbl, &nh->gw_addr, nh->rif->dev);
+	if (!n) {
+		n = neigh_create(nh->neigh_tbl, &nh->gw_addr, nh->rif->dev);
+		if (IS_ERR(n))
+			return PTR_ERR(n);
+		neigh_event_send(n, NULL);
+	}
+
+	mlxsw_sp_neigh_entry_remove(mlxsw_sp, neigh_entry);
+	neigh_entry->key.n = n;
+	err = mlxsw_sp_neigh_entry_insert(mlxsw_sp, neigh_entry);
+	if (err)
+		goto err_neigh_entry_insert;
+
+	read_lock_bh(&n->lock);
+	nud_state = n->nud_state;
+	dead = n->dead;
+	read_unlock_bh(&n->lock);
+	entry_connected = nud_state & NUD_VALID && !dead;
+
+	list_for_each_entry(nh, &neigh_entry->nexthop_list,
+			    neigh_list_node) {
+		neigh_release(old_n);
+		neigh_clone(n);
+		__mlxsw_sp_nexthop_neigh_update(nh, !entry_connected);
+		mlxsw_sp_nexthop_group_refresh(mlxsw_sp, nh->nhgi->nh_grp);
+	}
+
+	neigh_release(n);
+
+	return 0;
+
+err_neigh_entry_insert:
+	neigh_entry->key.n = old_n;
+	mlxsw_sp_neigh_entry_insert(mlxsw_sp, neigh_entry);
+	neigh_release(n);
+	return err;
+}
+
+static void
+mlxsw_sp_nexthop_neigh_update(struct mlxsw_sp *mlxsw_sp,
+			      struct mlxsw_sp_neigh_entry *neigh_entry,
+			      bool removing, bool dead)
+{
+	struct mlxsw_sp_nexthop *nh;
+
+	if (list_empty(&neigh_entry->nexthop_list))
+		return;
+
+	if (dead) {
+		int err;
+
+		err = mlxsw_sp_nexthop_dead_neigh_replace(mlxsw_sp,
+							  neigh_entry);
+		if (err)
+			dev_err(mlxsw_sp->bus_info->dev, "Failed to replace dead neigh\n");
+		return;
+	}
+
+	list_for_each_entry(nh, &neigh_entry->nexthop_list,
+			    neigh_list_node) {
+		__mlxsw_sp_nexthop_neigh_update(nh, removing);
+		mlxsw_sp_nexthop_group_refresh(mlxsw_sp, nh->nhgi->nh_grp);
+	}
+}
+
+static void mlxsw_sp_nexthop_rif_init(struct mlxsw_sp_nexthop *nh,
+				      struct mlxsw_sp_rif *rif)
+{
+	if (nh->rif)
+		return;
+
+	nh->rif = rif;
+	list_add(&nh->rif_list_node, &rif->nexthop_list);
+}
+
+static void mlxsw_sp_nexthop_rif_fini(struct mlxsw_sp_nexthop *nh)
+{
+	if (!nh->rif)
+		return;
+
+	list_del(&nh->rif_list_node);
+	nh->rif = NULL;
+}
+
+static int mlxsw_sp_nexthop_neigh_init(struct mlxsw_sp *mlxsw_sp,
+				       struct mlxsw_sp_nexthop *nh)
+{
+	struct mlxsw_sp_neigh_entry *neigh_entry;
+	struct neighbour *n;
+	u8 nud_state, dead;
+	int err;
+
+	if (!nh->nhgi->gateway || nh->neigh_entry)
+		return 0;
+
+	/* Take a reference of neigh here ensuring that neigh would
+	 * not be destructed before the nexthop entry is finished.
+	 * The reference is taken either in neigh_lookup() or
+	 * in neigh_create() in case n is not found.
+	 */
+	n = neigh_lookup(nh->neigh_tbl, &nh->gw_addr, nh->rif->dev);
+	if (!n) {
+		n = neigh_create(nh->neigh_tbl, &nh->gw_addr, nh->rif->dev);
+		if (IS_ERR(n))
+			return PTR_ERR(n);
+		neigh_event_send(n, NULL);
+	}
+	neigh_entry = mlxsw_sp_neigh_entry_lookup(mlxsw_sp, n);
+	if (!neigh_entry) {
+		neigh_entry = mlxsw_sp_neigh_entry_create(mlxsw_sp, n);
+		if (IS_ERR(neigh_entry)) {
+			err = -EINVAL;
+			goto err_neigh_entry_create;
+		}
+	}
+
+	/* If that is the first nexthop connected to that neigh, add to
+	 * nexthop_neighs_list
+	 */
+	if (list_empty(&neigh_entry->nexthop_list))
+		list_add_tail(&neigh_entry->nexthop_neighs_list_node,
+			      &mlxsw_sp->router->nexthop_neighs_list);
+
+	nh->neigh_entry = neigh_entry;
+	list_add_tail(&nh->neigh_list_node, &neigh_entry->nexthop_list);
+	read_lock_bh(&n->lock);
+	nud_state = n->nud_state;
+	dead = n->dead;
+	read_unlock_bh(&n->lock);
+	__mlxsw_sp_nexthop_neigh_update(nh, !(nud_state & NUD_VALID && !dead));
+
+	return 0;
+
+err_neigh_entry_create:
+	neigh_release(n);
+	return err;
+}
+
+static void mlxsw_sp_nexthop_neigh_fini(struct mlxsw_sp *mlxsw_sp,
+					struct mlxsw_sp_nexthop *nh)
+{
+	struct mlxsw_sp_neigh_entry *neigh_entry = nh->neigh_entry;
+	struct neighbour *n;
+
+	if (!neigh_entry)
+		return;
+	n = neigh_entry->key.n;
+
+	__mlxsw_sp_nexthop_neigh_update(nh, true);
+	list_del(&nh->neigh_list_node);
+	nh->neigh_entry = NULL;
+
+	/* If that is the last nexthop connected to that neigh, remove from
+	 * nexthop_neighs_list
+	 */
+	if (list_empty(&neigh_entry->nexthop_list))
+		list_del(&neigh_entry->nexthop_neighs_list_node);
+
+	if (!neigh_entry->connected && list_empty(&neigh_entry->nexthop_list))
+		mlxsw_sp_neigh_entry_destroy(mlxsw_sp, neigh_entry);
+
+	neigh_release(n);
+}
+
+static bool mlxsw_sp_ipip_netdev_ul_up(struct net_device *ol_dev)
+{
+	struct net_device *ul_dev;
+	bool is_up;
+
+	rcu_read_lock();
+	ul_dev = mlxsw_sp_ipip_netdev_ul_dev_get(ol_dev);
+	is_up = ul_dev ? (ul_dev->flags & IFF_UP) : true;
+	rcu_read_unlock();
+
+	return is_up;
+}
+
+static void mlxsw_sp_nexthop_ipip_init(struct mlxsw_sp *mlxsw_sp,
+				       struct mlxsw_sp_nexthop *nh,
+				       struct mlxsw_sp_ipip_entry *ipip_entry)
+{
+	bool removing;
+
+	if (!nh->nhgi->gateway || nh->ipip_entry)
+		return;
+
+	nh->ipip_entry = ipip_entry;
+	removing = !mlxsw_sp_ipip_netdev_ul_up(ipip_entry->ol_dev);
+	__mlxsw_sp_nexthop_neigh_update(nh, removing);
+	mlxsw_sp_nexthop_rif_init(nh, &ipip_entry->ol_lb->common);
+}
+
+static void mlxsw_sp_nexthop_ipip_fini(struct mlxsw_sp *mlxsw_sp,
+				       struct mlxsw_sp_nexthop *nh)
+{
+	struct mlxsw_sp_ipip_entry *ipip_entry = nh->ipip_entry;
+
+	if (!ipip_entry)
+		return;
+
+	__mlxsw_sp_nexthop_neigh_update(nh, true);
+	nh->ipip_entry = NULL;
+}
+
+static bool mlxsw_sp_nexthop4_ipip_type(const struct mlxsw_sp *mlxsw_sp,
+					const struct fib_nh *fib_nh,
+					enum mlxsw_sp_ipip_type *p_ipipt)
+{
+	struct net_device *dev = fib_nh->fib_nh_dev;
+
+	return dev &&
+	       fib_nh->nh_parent->fib_type == RTN_UNICAST &&
+	       mlxsw_sp_netdev_ipip_type(mlxsw_sp, dev, p_ipipt);
+}
+
+static int mlxsw_sp_nexthop_type_init(struct mlxsw_sp *mlxsw_sp,
+				      struct mlxsw_sp_nexthop *nh,
+				      const struct net_device *dev)
+{
+	const struct mlxsw_sp_ipip_ops *ipip_ops;
+	struct mlxsw_sp_ipip_entry *ipip_entry;
+	struct mlxsw_sp_rif *rif;
+	int err;
+
+	ipip_entry = mlxsw_sp_ipip_entry_find_by_ol_dev(mlxsw_sp, dev);
+	if (ipip_entry) {
+		ipip_ops = mlxsw_sp->router->ipip_ops_arr[ipip_entry->ipipt];
+		if (ipip_ops->can_offload(mlxsw_sp, dev)) {
+			nh->type = MLXSW_SP_NEXTHOP_TYPE_IPIP;
+			mlxsw_sp_nexthop_ipip_init(mlxsw_sp, nh, ipip_entry);
+			return 0;
+		}
+	}
+
+	nh->type = MLXSW_SP_NEXTHOP_TYPE_ETH;
+	rif = mlxsw_sp_rif_find_by_dev(mlxsw_sp, dev);
+	if (!rif)
+		return 0;
+
+	mlxsw_sp_nexthop_rif_init(nh, rif);
+	err = mlxsw_sp_nexthop_neigh_init(mlxsw_sp, nh);
+	if (err)
+		goto err_neigh_init;
+
+	return 0;
+
+err_neigh_init:
+	mlxsw_sp_nexthop_rif_fini(nh);
+	return err;
+}
+
+static void mlxsw_sp_nexthop_type_fini(struct mlxsw_sp *mlxsw_sp,
+				       struct mlxsw_sp_nexthop *nh)
+{
+	switch (nh->type) {
+	case MLXSW_SP_NEXTHOP_TYPE_ETH:
+		mlxsw_sp_nexthop_neigh_fini(mlxsw_sp, nh);
+		mlxsw_sp_nexthop_rif_fini(nh);
+		break;
+	case MLXSW_SP_NEXTHOP_TYPE_IPIP:
+		mlxsw_sp_nexthop_rif_fini(nh);
+		mlxsw_sp_nexthop_ipip_fini(mlxsw_sp, nh);
+		break;
+	}
+}
+
+static int mlxsw_sp_nexthop4_init(struct mlxsw_sp *mlxsw_sp,
+				  struct mlxsw_sp_nexthop_group *nh_grp,
+				  struct mlxsw_sp_nexthop *nh,
+				  struct fib_nh *fib_nh)
+{
+	struct net_device *dev = fib_nh->fib_nh_dev;
+	struct in_device *in_dev;
+	int err;
+
+	nh->nhgi = nh_grp->nhgi;
+	nh->key.fib_nh = fib_nh;
+#ifdef CONFIG_IP_ROUTE_MULTIPATH
+	nh->nh_weight = fib_nh->fib_nh_weight;
+#else
+	nh->nh_weight = 1;
+#endif
+	memcpy(&nh->gw_addr, &fib_nh->fib_nh_gw4, sizeof(fib_nh->fib_nh_gw4));
+	nh->neigh_tbl = &arp_tbl;
+	err = mlxsw_sp_nexthop_insert(mlxsw_sp, nh);
+	if (err)
+		return err;
+
+	mlxsw_sp_nexthop_counter_alloc(mlxsw_sp, nh);
+	list_add_tail(&nh->router_list_node, &mlxsw_sp->router->nexthop_list);
+
+	if (!dev)
+		return 0;
+	nh->ifindex = dev->ifindex;
+
+	rcu_read_lock();
+	in_dev = __in_dev_get_rcu(dev);
+	if (in_dev && IN_DEV_IGNORE_ROUTES_WITH_LINKDOWN(in_dev) &&
+	    fib_nh->fib_nh_flags & RTNH_F_LINKDOWN) {
+		rcu_read_unlock();
+		return 0;
+	}
+	rcu_read_unlock();
+
+	err = mlxsw_sp_nexthop_type_init(mlxsw_sp, nh, dev);
+	if (err)
+		goto err_nexthop_neigh_init;
+
+	return 0;
+
+err_nexthop_neigh_init:
+	list_del(&nh->router_list_node);
+	mlxsw_sp_nexthop_counter_free(mlxsw_sp, nh);
+	mlxsw_sp_nexthop_remove(mlxsw_sp, nh);
+	return err;
+}
+
+static void mlxsw_sp_nexthop4_fini(struct mlxsw_sp *mlxsw_sp,
+				   struct mlxsw_sp_nexthop *nh)
+{
+	mlxsw_sp_nexthop_type_fini(mlxsw_sp, nh);
+	list_del(&nh->router_list_node);
+	mlxsw_sp_nexthop_counter_free(mlxsw_sp, nh);
+	mlxsw_sp_nexthop_remove(mlxsw_sp, nh);
+}
+
+static void mlxsw_sp_nexthop4_event(struct mlxsw_sp *mlxsw_sp,
+				    unsigned long event, struct fib_nh *fib_nh)
+{
+	struct mlxsw_sp_nexthop_key key;
+	struct mlxsw_sp_nexthop *nh;
+
+	key.fib_nh = fib_nh;
+	nh = mlxsw_sp_nexthop_lookup(mlxsw_sp, key);
+	if (!nh)
+		return;
+
+	switch (event) {
+	case FIB_EVENT_NH_ADD:
+		mlxsw_sp_nexthop_type_init(mlxsw_sp, nh, fib_nh->fib_nh_dev);
+		break;
+	case FIB_EVENT_NH_DEL:
+		mlxsw_sp_nexthop_type_fini(mlxsw_sp, nh);
+		break;
+	}
+
+	mlxsw_sp_nexthop_group_refresh(mlxsw_sp, nh->nhgi->nh_grp);
+}
+
+static void mlxsw_sp_nexthop_rif_update(struct mlxsw_sp *mlxsw_sp,
+					struct mlxsw_sp_rif *rif)
+{
+	struct mlxsw_sp_nexthop *nh;
+	bool removing;
+
+	list_for_each_entry(nh, &rif->nexthop_list, rif_list_node) {
+		switch (nh->type) {
+		case MLXSW_SP_NEXTHOP_TYPE_ETH:
+			removing = false;
+			break;
+		case MLXSW_SP_NEXTHOP_TYPE_IPIP:
+			removing = !mlxsw_sp_ipip_netdev_ul_up(rif->dev);
+			break;
+		default:
+			WARN_ON(1);
+			continue;
+		}
+
+		__mlxsw_sp_nexthop_neigh_update(nh, removing);
+		mlxsw_sp_nexthop_group_refresh(mlxsw_sp, nh->nhgi->nh_grp);
+	}
+}
+
+static void mlxsw_sp_nexthop_rif_migrate(struct mlxsw_sp *mlxsw_sp,
+					 struct mlxsw_sp_rif *old_rif,
+					 struct mlxsw_sp_rif *new_rif)
+{
+	struct mlxsw_sp_nexthop *nh;
+
+	list_splice_init(&old_rif->nexthop_list, &new_rif->nexthop_list);
+	list_for_each_entry(nh, &new_rif->nexthop_list, rif_list_node)
+		nh->rif = new_rif;
+	mlxsw_sp_nexthop_rif_update(mlxsw_sp, new_rif);
+}
+
+static void mlxsw_sp_nexthop_rif_gone_sync(struct mlxsw_sp *mlxsw_sp,
+					   struct mlxsw_sp_rif *rif)
+{
+	struct mlxsw_sp_nexthop *nh, *tmp;
+
+	list_for_each_entry_safe(nh, tmp, &rif->nexthop_list, rif_list_node) {
+		mlxsw_sp_nexthop_type_fini(mlxsw_sp, nh);
+		mlxsw_sp_nexthop_group_refresh(mlxsw_sp, nh->nhgi->nh_grp);
+	}
+}
+
+static int mlxsw_sp_adj_trap_entry_init(struct mlxsw_sp *mlxsw_sp)
+{
+	enum mlxsw_reg_ratr_trap_action trap_action;
+	char ratr_pl[MLXSW_REG_RATR_LEN];
+	int err;
+
+	err = mlxsw_sp_kvdl_alloc(mlxsw_sp, MLXSW_SP_KVDL_ENTRY_TYPE_ADJ, 1,
+				  &mlxsw_sp->router->adj_trap_index);
+	if (err)
+		return err;
+
+	trap_action = MLXSW_REG_RATR_TRAP_ACTION_TRAP;
+	mlxsw_reg_ratr_pack(ratr_pl, MLXSW_REG_RATR_OP_WRITE_WRITE_ENTRY, true,
+			    MLXSW_REG_RATR_TYPE_ETHERNET,
+			    mlxsw_sp->router->adj_trap_index,
+			    mlxsw_sp->router->lb_rif_index);
+	mlxsw_reg_ratr_trap_action_set(ratr_pl, trap_action);
+	mlxsw_reg_ratr_trap_id_set(ratr_pl, MLXSW_TRAP_ID_RTR_EGRESS0);
+	err = mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(ratr), ratr_pl);
+	if (err)
+		goto err_ratr_write;
+
+	return 0;
+
+err_ratr_write:
+	mlxsw_sp_kvdl_free(mlxsw_sp, MLXSW_SP_KVDL_ENTRY_TYPE_ADJ, 1,
+			   mlxsw_sp->router->adj_trap_index);
+	return err;
+}
+
+static void mlxsw_sp_adj_trap_entry_fini(struct mlxsw_sp *mlxsw_sp)
+{
+	mlxsw_sp_kvdl_free(mlxsw_sp, MLXSW_SP_KVDL_ENTRY_TYPE_ADJ, 1,
+			   mlxsw_sp->router->adj_trap_index);
+}
+
+static int mlxsw_sp_nexthop_group_inc(struct mlxsw_sp *mlxsw_sp)
+{
+	int err;
+
+	if (refcount_inc_not_zero(&mlxsw_sp->router->num_groups))
+		return 0;
+
+	err = mlxsw_sp_adj_trap_entry_init(mlxsw_sp);
+	if (err)
+		return err;
+
+	refcount_set(&mlxsw_sp->router->num_groups, 1);
+
+	return 0;
+}
+
+static void mlxsw_sp_nexthop_group_dec(struct mlxsw_sp *mlxsw_sp)
+{
+	if (!refcount_dec_and_test(&mlxsw_sp->router->num_groups))
+		return;
+
+	mlxsw_sp_adj_trap_entry_fini(mlxsw_sp);
+}
+
+static void
+mlxsw_sp_nh_grp_activity_get(struct mlxsw_sp *mlxsw_sp,
+			     const struct mlxsw_sp_nexthop_group *nh_grp,
+			     unsigned long *activity)
+{
+	char *ratrad_pl;
+	int i, err;
+
+	ratrad_pl = kmalloc(MLXSW_REG_RATRAD_LEN, GFP_KERNEL);
+	if (!ratrad_pl)
+		return;
+
+	mlxsw_reg_ratrad_pack(ratrad_pl, nh_grp->nhgi->adj_index,
+			      nh_grp->nhgi->count);
+	err = mlxsw_reg_query(mlxsw_sp->core, MLXSW_REG(ratrad), ratrad_pl);
+	if (err)
+		goto out;
+
+	for (i = 0; i < nh_grp->nhgi->count; i++) {
+		if (!mlxsw_reg_ratrad_activity_vector_get(ratrad_pl, i))
+			continue;
+		bitmap_set(activity, i, 1);
+	}
+
+out:
+	kfree(ratrad_pl);
+}
+
+#define MLXSW_SP_NH_GRP_ACTIVITY_UPDATE_INTERVAL 1000 /* ms */
+
+static void
+mlxsw_sp_nh_grp_activity_update(struct mlxsw_sp *mlxsw_sp,
+				const struct mlxsw_sp_nexthop_group *nh_grp)
+{
+	unsigned long *activity;
+
+	activity = bitmap_zalloc(nh_grp->nhgi->count, GFP_KERNEL);
+	if (!activity)
+		return;
+
+	mlxsw_sp_nh_grp_activity_get(mlxsw_sp, nh_grp, activity);
+	nexthop_res_grp_activity_update(mlxsw_sp_net(mlxsw_sp), nh_grp->obj.id,
+					nh_grp->nhgi->count, activity);
+
+	bitmap_free(activity);
+}
+
+static void
+mlxsw_sp_nh_grp_activity_work_schedule(struct mlxsw_sp *mlxsw_sp)
+{
+	unsigned int interval = MLXSW_SP_NH_GRP_ACTIVITY_UPDATE_INTERVAL;
+
+	mlxsw_core_schedule_dw(&mlxsw_sp->router->nh_grp_activity_dw,
+			       msecs_to_jiffies(interval));
+}
+
+static void mlxsw_sp_nh_grp_activity_work(struct work_struct *work)
+{
+	struct mlxsw_sp_nexthop_group_info *nhgi;
+	struct mlxsw_sp_router *router;
+	bool reschedule = false;
+
+	router = container_of(work, struct mlxsw_sp_router,
+			      nh_grp_activity_dw.work);
+
+	mutex_lock(&router->lock);
+
+	list_for_each_entry(nhgi, &router->nh_res_grp_list, list) {
+		mlxsw_sp_nh_grp_activity_update(router->mlxsw_sp, nhgi->nh_grp);
+		reschedule = true;
+	}
+
+	mutex_unlock(&router->lock);
+
+	if (!reschedule)
+		return;
+	mlxsw_sp_nh_grp_activity_work_schedule(router->mlxsw_sp);
+}
+
+static int
+mlxsw_sp_nexthop_obj_single_validate(struct mlxsw_sp *mlxsw_sp,
+				     const struct nh_notifier_single_info *nh,
+				     struct netlink_ext_ack *extack)
+{
+	int err = -EINVAL;
+
+	if (nh->is_fdb)
+		NL_SET_ERR_MSG_MOD(extack, "FDB nexthops are not supported");
+	else if (nh->has_encap)
+		NL_SET_ERR_MSG_MOD(extack, "Encapsulating nexthops are not supported");
+	else
+		err = 0;
+
+	return err;
+}
+
+static int
+mlxsw_sp_nexthop_obj_group_entry_validate(struct mlxsw_sp *mlxsw_sp,
+					  const struct nh_notifier_single_info *nh,
+					  struct netlink_ext_ack *extack)
+{
+	int err;
+
+	err = mlxsw_sp_nexthop_obj_single_validate(mlxsw_sp, nh, extack);
+	if (err)
+		return err;
+
+	/* Device only nexthops with an IPIP device are programmed as
+	 * encapsulating adjacency entries.
+	 */
+	if (!nh->gw_family && !nh->is_reject &&
+	    !mlxsw_sp_netdev_ipip_type(mlxsw_sp, nh->dev, NULL)) {
+		NL_SET_ERR_MSG_MOD(extack, "Nexthop group entry does not have a gateway");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int
+mlxsw_sp_nexthop_obj_group_validate(struct mlxsw_sp *mlxsw_sp,
+				    const struct nh_notifier_grp_info *nh_grp,
+				    struct netlink_ext_ack *extack)
+{
+	int i;
+
+	if (nh_grp->is_fdb) {
+		NL_SET_ERR_MSG_MOD(extack, "FDB nexthop groups are not supported");
+		return -EINVAL;
+	}
+
+	for (i = 0; i < nh_grp->num_nh; i++) {
+		const struct nh_notifier_single_info *nh;
+		int err;
+
+		nh = &nh_grp->nh_entries[i].nh;
+		err = mlxsw_sp_nexthop_obj_group_entry_validate(mlxsw_sp, nh,
+								extack);
+		if (err)
+			return err;
+	}
+
+	return 0;
+}
+
+static int
+mlxsw_sp_nexthop_obj_res_group_size_validate(struct mlxsw_sp *mlxsw_sp,
+					     const struct nh_notifier_res_table_info *nh_res_table,
+					     struct netlink_ext_ack *extack)
+{
+	unsigned int alloc_size;
+	bool valid_size = false;
+	int err, i;
+
+	if (nh_res_table->num_nh_buckets < 32) {
+		NL_SET_ERR_MSG_MOD(extack, "Minimum number of buckets is 32");
+		return -EINVAL;
+	}
+
+	for (i = 0; i < mlxsw_sp->router->adj_grp_size_ranges_count; i++) {
+		const struct mlxsw_sp_adj_grp_size_range *size_range;
+
+		size_range = &mlxsw_sp->router->adj_grp_size_ranges[i];
+
+		if (nh_res_table->num_nh_buckets >= size_range->start &&
+		    nh_res_table->num_nh_buckets <= size_range->end) {
+			valid_size = true;
+			break;
+		}
+	}
+
+	if (!valid_size) {
+		NL_SET_ERR_MSG_MOD(extack, "Invalid number of buckets");
+		return -EINVAL;
+	}
+
+	err = mlxsw_sp_kvdl_alloc_count_query(mlxsw_sp,
+					      MLXSW_SP_KVDL_ENTRY_TYPE_ADJ,
+					      nh_res_table->num_nh_buckets,
+					      &alloc_size);
+	if (err || nh_res_table->num_nh_buckets != alloc_size) {
+		NL_SET_ERR_MSG_MOD(extack, "Number of buckets does not fit allocation size of any KVDL partition");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int
+mlxsw_sp_nexthop_obj_res_group_validate(struct mlxsw_sp *mlxsw_sp,
+					const struct nh_notifier_res_table_info *nh_res_table,
+					struct netlink_ext_ack *extack)
+{
+	int err;
+	u16 i;
+
+	err = mlxsw_sp_nexthop_obj_res_group_size_validate(mlxsw_sp,
+							   nh_res_table,
+							   extack);
+	if (err)
+		return err;
+
+	for (i = 0; i < nh_res_table->num_nh_buckets; i++) {
+		const struct nh_notifier_single_info *nh;
+		int err;
+
+		nh = &nh_res_table->nhs[i];
+		err = mlxsw_sp_nexthop_obj_group_entry_validate(mlxsw_sp, nh,
+								extack);
+		if (err)
+			return err;
+	}
+
+	return 0;
+}
+
+static int mlxsw_sp_nexthop_obj_validate(struct mlxsw_sp *mlxsw_sp,
+					 unsigned long event,
+					 struct nh_notifier_info *info)
+{
+	struct nh_notifier_single_info *nh;
+
+	if (event != NEXTHOP_EVENT_REPLACE &&
+	    event != NEXTHOP_EVENT_RES_TABLE_PRE_REPLACE &&
+	    event != NEXTHOP_EVENT_BUCKET_REPLACE)
+		return 0;
+
+	switch (info->type) {
+	case NH_NOTIFIER_INFO_TYPE_SINGLE:
+		return mlxsw_sp_nexthop_obj_single_validate(mlxsw_sp, info->nh,
+							    info->extack);
+	case NH_NOTIFIER_INFO_TYPE_GRP:
+		return mlxsw_sp_nexthop_obj_group_validate(mlxsw_sp,
+							   info->nh_grp,
+							   info->extack);
+	case NH_NOTIFIER_INFO_TYPE_RES_TABLE:
+		return mlxsw_sp_nexthop_obj_res_group_validate(mlxsw_sp,
+							       info->nh_res_table,
+							       info->extack);
+	case NH_NOTIFIER_INFO_TYPE_RES_BUCKET:
+		nh = &info->nh_res_bucket->new_nh;
+		return mlxsw_sp_nexthop_obj_group_entry_validate(mlxsw_sp, nh,
+								 info->extack);
+	default:
+		NL_SET_ERR_MSG_MOD(info->extack, "Unsupported nexthop type");
+		return -EOPNOTSUPP;
+	}
+}
+
+static bool mlxsw_sp_nexthop_obj_is_gateway(struct mlxsw_sp *mlxsw_sp,
+					    const struct nh_notifier_info *info)
+{
+	const struct net_device *dev;
+
+	switch (info->type) {
+	case NH_NOTIFIER_INFO_TYPE_SINGLE:
+		dev = info->nh->dev;
+		return info->nh->gw_family || info->nh->is_reject ||
+		       mlxsw_sp_netdev_ipip_type(mlxsw_sp, dev, NULL);
+	case NH_NOTIFIER_INFO_TYPE_GRP:
+	case NH_NOTIFIER_INFO_TYPE_RES_TABLE:
+		/* Already validated earlier. */
+		return true;
+	default:
+		return false;
+	}
+}
+
+static void mlxsw_sp_nexthop_obj_blackhole_init(struct mlxsw_sp *mlxsw_sp,
+						struct mlxsw_sp_nexthop *nh)
+{
+	u16 lb_rif_index = mlxsw_sp->router->lb_rif_index;
+
+	nh->action = MLXSW_SP_NEXTHOP_ACTION_DISCARD;
+	nh->should_offload = 1;
+	/* While nexthops that discard packets do not forward packets
+	 * via an egress RIF, they still need to be programmed using a
+	 * valid RIF, so use the loopback RIF created during init.
+	 */
+	nh->rif = mlxsw_sp->router->rifs[lb_rif_index];
+}
+
+static void mlxsw_sp_nexthop_obj_blackhole_fini(struct mlxsw_sp *mlxsw_sp,
+						struct mlxsw_sp_nexthop *nh)
+{
+	nh->rif = NULL;
+	nh->should_offload = 0;
+}
+
+static int
+mlxsw_sp_nexthop_obj_init(struct mlxsw_sp *mlxsw_sp,
+			  struct mlxsw_sp_nexthop_group *nh_grp,
+			  struct mlxsw_sp_nexthop *nh,
+			  struct nh_notifier_single_info *nh_obj, int weight)
+{
+	struct net_device *dev = nh_obj->dev;
+	int err;
+
+	nh->nhgi = nh_grp->nhgi;
+	nh->nh_weight = weight;
+
+	switch (nh_obj->gw_family) {
+	case AF_INET:
+		memcpy(&nh->gw_addr, &nh_obj->ipv4, sizeof(nh_obj->ipv4));
+		nh->neigh_tbl = &arp_tbl;
+		break;
+	case AF_INET6:
+		memcpy(&nh->gw_addr, &nh_obj->ipv6, sizeof(nh_obj->ipv6));
+#if IS_ENABLED(CONFIG_IPV6)
+		nh->neigh_tbl = &nd_tbl;
+#endif
+		break;
+	}
+
+	mlxsw_sp_nexthop_counter_alloc(mlxsw_sp, nh);
+	list_add_tail(&nh->router_list_node, &mlxsw_sp->router->nexthop_list);
+	nh->ifindex = dev->ifindex;
+
+	err = mlxsw_sp_nexthop_type_init(mlxsw_sp, nh, dev);
+	if (err)
+		goto err_type_init;
+
+	if (nh_obj->is_reject)
+		mlxsw_sp_nexthop_obj_blackhole_init(mlxsw_sp, nh);
+
+	/* In a resilient nexthop group, all the nexthops must be written to
+	 * the adjacency table. Even if they do not have a valid neighbour or
+	 * RIF.
+	 */
+	if (nh_grp->nhgi->is_resilient && !nh->should_offload) {
+		nh->action = MLXSW_SP_NEXTHOP_ACTION_TRAP;
+		nh->should_offload = 1;
+	}
+
+	return 0;
+
+err_type_init:
+	list_del(&nh->router_list_node);
+	mlxsw_sp_nexthop_counter_free(mlxsw_sp, nh);
+	return err;
+}
+
+static void mlxsw_sp_nexthop_obj_fini(struct mlxsw_sp *mlxsw_sp,
+				      struct mlxsw_sp_nexthop *nh)
+{
+	if (nh->action == MLXSW_SP_NEXTHOP_ACTION_DISCARD)
+		mlxsw_sp_nexthop_obj_blackhole_fini(mlxsw_sp, nh);
+	mlxsw_sp_nexthop_type_fini(mlxsw_sp, nh);
+	list_del(&nh->router_list_node);
+	mlxsw_sp_nexthop_counter_free(mlxsw_sp, nh);
+	nh->should_offload = 0;
+}
+
+static int
+mlxsw_sp_nexthop_obj_group_info_init(struct mlxsw_sp *mlxsw_sp,
+				     struct mlxsw_sp_nexthop_group *nh_grp,
+				     struct nh_notifier_info *info)
+{
+	struct mlxsw_sp_nexthop_group_info *nhgi;
+	struct mlxsw_sp_nexthop *nh;
+	bool is_resilient = false;
+	unsigned int nhs;
+	int err, i;
+
+	switch (info->type) {
+	case NH_NOTIFIER_INFO_TYPE_SINGLE:
+		nhs = 1;
+		break;
+	case NH_NOTIFIER_INFO_TYPE_GRP:
+		nhs = info->nh_grp->num_nh;
+		break;
+	case NH_NOTIFIER_INFO_TYPE_RES_TABLE:
+		nhs = info->nh_res_table->num_nh_buckets;
+		is_resilient = true;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	nhgi = kzalloc(struct_size(nhgi, nexthops, nhs), GFP_KERNEL);
+	if (!nhgi)
+		return -ENOMEM;
+	nh_grp->nhgi = nhgi;
+	nhgi->nh_grp = nh_grp;
+	nhgi->gateway = mlxsw_sp_nexthop_obj_is_gateway(mlxsw_sp, info);
+	nhgi->is_resilient = is_resilient;
+	nhgi->count = nhs;
+	for (i = 0; i < nhgi->count; i++) {
+		struct nh_notifier_single_info *nh_obj;
+		int weight;
+
+		nh = &nhgi->nexthops[i];
+		switch (info->type) {
+		case NH_NOTIFIER_INFO_TYPE_SINGLE:
+			nh_obj = info->nh;
+			weight = 1;
+			break;
+		case NH_NOTIFIER_INFO_TYPE_GRP:
+			nh_obj = &info->nh_grp->nh_entries[i].nh;
+			weight = info->nh_grp->nh_entries[i].weight;
+			break;
+		case NH_NOTIFIER_INFO_TYPE_RES_TABLE:
+			nh_obj = &info->nh_res_table->nhs[i];
+			weight = 1;
+			break;
+		default:
+			err = -EINVAL;
+			goto err_nexthop_obj_init;
+		}
+		err = mlxsw_sp_nexthop_obj_init(mlxsw_sp, nh_grp, nh, nh_obj,
+						weight);
+		if (err)
+			goto err_nexthop_obj_init;
+	}
+	err = mlxsw_sp_nexthop_group_inc(mlxsw_sp);
+	if (err)
+		goto err_group_inc;
+	err = mlxsw_sp_nexthop_group_refresh(mlxsw_sp, nh_grp);
+	if (err) {
+		NL_SET_ERR_MSG_MOD(info->extack, "Failed to write adjacency entries to the device");
+		goto err_group_refresh;
+	}
+
+	/* Add resilient nexthop groups to a list so that the activity of their
+	 * nexthop buckets will be periodically queried and cleared.
+	 */
+	if (nhgi->is_resilient) {
+		if (list_empty(&mlxsw_sp->router->nh_res_grp_list))
+			mlxsw_sp_nh_grp_activity_work_schedule(mlxsw_sp);
+		list_add(&nhgi->list, &mlxsw_sp->router->nh_res_grp_list);
+	}
+
+	return 0;
+
+err_group_refresh:
+	mlxsw_sp_nexthop_group_dec(mlxsw_sp);
+err_group_inc:
+	i = nhgi->count;
+err_nexthop_obj_init:
+	for (i--; i >= 0; i--) {
+		nh = &nhgi->nexthops[i];
+		mlxsw_sp_nexthop_obj_fini(mlxsw_sp, nh);
+	}
+	kfree(nhgi);
+	return err;
+}
+
+static void
+mlxsw_sp_nexthop_obj_group_info_fini(struct mlxsw_sp *mlxsw_sp,
+				     struct mlxsw_sp_nexthop_group *nh_grp)
+{
+	struct mlxsw_sp_nexthop_group_info *nhgi = nh_grp->nhgi;
+	struct mlxsw_sp_router *router = mlxsw_sp->router;
+	int i;
+
+	if (nhgi->is_resilient) {
+		list_del(&nhgi->list);
+		if (list_empty(&mlxsw_sp->router->nh_res_grp_list))
+			cancel_delayed_work(&router->nh_grp_activity_dw);
+	}
+
+	mlxsw_sp_nexthop_group_dec(mlxsw_sp);
+	for (i = nhgi->count - 1; i >= 0; i--) {
+		struct mlxsw_sp_nexthop *nh = &nhgi->nexthops[i];
+
+		mlxsw_sp_nexthop_obj_fini(mlxsw_sp, nh);
+	}
+	mlxsw_sp_nexthop_group_refresh(mlxsw_sp, nh_grp);
+	WARN_ON_ONCE(nhgi->adj_index_valid);
+	kfree(nhgi);
+}
+
+static struct mlxsw_sp_nexthop_group *
+mlxsw_sp_nexthop_obj_group_create(struct mlxsw_sp *mlxsw_sp,
+				  struct nh_notifier_info *info)
+{
+	struct mlxsw_sp_nexthop_group *nh_grp;
+	int err;
+
+	nh_grp = kzalloc(sizeof(*nh_grp), GFP_KERNEL);
+	if (!nh_grp)
+		return ERR_PTR(-ENOMEM);
+	INIT_LIST_HEAD(&nh_grp->vr_list);
+	err = rhashtable_init(&nh_grp->vr_ht,
+			      &mlxsw_sp_nexthop_group_vr_ht_params);
+	if (err)
+		goto err_nexthop_group_vr_ht_init;
+	INIT_LIST_HEAD(&nh_grp->fib_list);
+	nh_grp->type = MLXSW_SP_NEXTHOP_GROUP_TYPE_OBJ;
+	nh_grp->obj.id = info->id;
+
+	err = mlxsw_sp_nexthop_obj_group_info_init(mlxsw_sp, nh_grp, info);
+	if (err)
+		goto err_nexthop_group_info_init;
+
+	nh_grp->can_destroy = false;
+
+	return nh_grp;
+
+err_nexthop_group_info_init:
+	rhashtable_destroy(&nh_grp->vr_ht);
+err_nexthop_group_vr_ht_init:
+	kfree(nh_grp);
+	return ERR_PTR(err);
+}
+
+static void
+mlxsw_sp_nexthop_obj_group_destroy(struct mlxsw_sp *mlxsw_sp,
+				   struct mlxsw_sp_nexthop_group *nh_grp)
+{
+	if (!nh_grp->can_destroy)
+		return;
+	mlxsw_sp_nexthop_obj_group_info_fini(mlxsw_sp, nh_grp);
+	WARN_ON_ONCE(!list_empty(&nh_grp->fib_list));
+	WARN_ON_ONCE(!list_empty(&nh_grp->vr_list));
+	rhashtable_destroy(&nh_grp->vr_ht);
+	kfree(nh_grp);
+}
+
+static struct mlxsw_sp_nexthop_group *
+mlxsw_sp_nexthop_obj_group_lookup(struct mlxsw_sp *mlxsw_sp, u32 id)
+{
+	struct mlxsw_sp_nexthop_group_cmp_arg cmp_arg;
+
+	cmp_arg.type = MLXSW_SP_NEXTHOP_GROUP_TYPE_OBJ;
+	cmp_arg.id = id;
+	return rhashtable_lookup_fast(&mlxsw_sp->router->nexthop_group_ht,
+				      &cmp_arg,
+				      mlxsw_sp_nexthop_group_ht_params);
+}
+
+static int mlxsw_sp_nexthop_obj_group_add(struct mlxsw_sp *mlxsw_sp,
+					  struct mlxsw_sp_nexthop_group *nh_grp)
+{
+	return mlxsw_sp_nexthop_group_insert(mlxsw_sp, nh_grp);
+}
+
+static int
+mlxsw_sp_nexthop_obj_group_replace(struct mlxsw_sp *mlxsw_sp,
+				   struct mlxsw_sp_nexthop_group *nh_grp,
+				   struct mlxsw_sp_nexthop_group *old_nh_grp,
+				   struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp_nexthop_group_info *old_nhgi = old_nh_grp->nhgi;
+	struct mlxsw_sp_nexthop_group_info *new_nhgi = nh_grp->nhgi;
+	int err;
+
+	old_nh_grp->nhgi = new_nhgi;
+	new_nhgi->nh_grp = old_nh_grp;
+	nh_grp->nhgi = old_nhgi;
+	old_nhgi->nh_grp = nh_grp;
+
+	if (old_nhgi->adj_index_valid && new_nhgi->adj_index_valid) {
+		/* Both the old adjacency index and the new one are valid.
+		 * Routes are currently using the old one. Tell the device to
+		 * replace the old adjacency index with the new one.
+		 */
+		err = mlxsw_sp_adj_index_mass_update(mlxsw_sp, old_nh_grp,
+						     old_nhgi->adj_index,
+						     old_nhgi->ecmp_size);
+		if (err) {
+			NL_SET_ERR_MSG_MOD(extack, "Failed to replace old adjacency index with new one");
+			goto err_out;
+		}
+	} else if (old_nhgi->adj_index_valid && !new_nhgi->adj_index_valid) {
+		/* The old adjacency index is valid, while the new one is not.
+		 * Iterate over all the routes using the group and change them
+		 * to trap packets to the CPU.
+		 */
+		err = mlxsw_sp_nexthop_fib_entries_update(mlxsw_sp, old_nh_grp);
+		if (err) {
+			NL_SET_ERR_MSG_MOD(extack, "Failed to update routes to trap packets");
+			goto err_out;
+		}
+	} else if (!old_nhgi->adj_index_valid && new_nhgi->adj_index_valid) {
+		/* The old adjacency index is invalid, while the new one is.
+		 * Iterate over all the routes using the group and change them
+		 * to forward packets using the new valid index.
+		 */
+		err = mlxsw_sp_nexthop_fib_entries_update(mlxsw_sp, old_nh_grp);
+		if (err) {
+			NL_SET_ERR_MSG_MOD(extack, "Failed to update routes to forward packets");
+			goto err_out;
+		}
+	}
+
+	/* Make sure the flags are set / cleared based on the new nexthop group
+	 * information.
+	 */
+	mlxsw_sp_nexthop_obj_group_offload_refresh(mlxsw_sp, old_nh_grp);
+
+	/* At this point 'nh_grp' is just a shell that is not used by anyone
+	 * and its nexthop group info is the old info that was just replaced
+	 * with the new one. Remove it.
+	 */
+	nh_grp->can_destroy = true;
+	mlxsw_sp_nexthop_obj_group_destroy(mlxsw_sp, nh_grp);
+
+	return 0;
+
+err_out:
+	old_nhgi->nh_grp = old_nh_grp;
+	nh_grp->nhgi = new_nhgi;
+	new_nhgi->nh_grp = nh_grp;
+	old_nh_grp->nhgi = old_nhgi;
+	return err;
+}
+
+static int mlxsw_sp_nexthop_obj_new(struct mlxsw_sp *mlxsw_sp,
+				    struct nh_notifier_info *info)
+{
+	struct mlxsw_sp_nexthop_group *nh_grp, *old_nh_grp;
+	struct netlink_ext_ack *extack = info->extack;
+	int err;
+
+	nh_grp = mlxsw_sp_nexthop_obj_group_create(mlxsw_sp, info);
+	if (IS_ERR(nh_grp))
+		return PTR_ERR(nh_grp);
+
+	old_nh_grp = mlxsw_sp_nexthop_obj_group_lookup(mlxsw_sp, info->id);
+	if (!old_nh_grp)
+		err = mlxsw_sp_nexthop_obj_group_add(mlxsw_sp, nh_grp);
+	else
+		err = mlxsw_sp_nexthop_obj_group_replace(mlxsw_sp, nh_grp,
+							 old_nh_grp, extack);
+
+	if (err) {
+		nh_grp->can_destroy = true;
+		mlxsw_sp_nexthop_obj_group_destroy(mlxsw_sp, nh_grp);
+	}
+
+	return err;
+}
+
+static void mlxsw_sp_nexthop_obj_del(struct mlxsw_sp *mlxsw_sp,
+				     struct nh_notifier_info *info)
+{
+	struct mlxsw_sp_nexthop_group *nh_grp;
+
+	nh_grp = mlxsw_sp_nexthop_obj_group_lookup(mlxsw_sp, info->id);
+	if (!nh_grp)
+		return;
+
+	nh_grp->can_destroy = true;
+	mlxsw_sp_nexthop_group_remove(mlxsw_sp, nh_grp);
+
+	/* If the group still has routes using it, then defer the delete
+	 * operation until the last route using it is deleted.
+	 */
+	if (!list_empty(&nh_grp->fib_list))
+		return;
+	mlxsw_sp_nexthop_obj_group_destroy(mlxsw_sp, nh_grp);
+}
+
+static int mlxsw_sp_nexthop_obj_bucket_query(struct mlxsw_sp *mlxsw_sp,
+					     u32 adj_index, char *ratr_pl)
+{
+	MLXSW_REG_ZERO(ratr, ratr_pl);
+	mlxsw_reg_ratr_op_set(ratr_pl, MLXSW_REG_RATR_OP_QUERY_READ);
+	mlxsw_reg_ratr_adjacency_index_low_set(ratr_pl, adj_index);
+	mlxsw_reg_ratr_adjacency_index_high_set(ratr_pl, adj_index >> 16);
+
+	return mlxsw_reg_query(mlxsw_sp->core, MLXSW_REG(ratr), ratr_pl);
+}
+
+static int mlxsw_sp_nexthop_obj_bucket_compare(char *ratr_pl, char *ratr_pl_new)
+{
+	/* Clear the opcode and activity on both the old and new payload as
+	 * they are irrelevant for the comparison.
+	 */
+	mlxsw_reg_ratr_op_set(ratr_pl, MLXSW_REG_RATR_OP_QUERY_READ);
+	mlxsw_reg_ratr_a_set(ratr_pl, 0);
+	mlxsw_reg_ratr_op_set(ratr_pl_new, MLXSW_REG_RATR_OP_QUERY_READ);
+	mlxsw_reg_ratr_a_set(ratr_pl_new, 0);
+
+	/* If the contents of the adjacency entry are consistent with the
+	 * replacement request, then replacement was successful.
+	 */
+	if (!memcmp(ratr_pl, ratr_pl_new, MLXSW_REG_RATR_LEN))
+		return 0;
+
+	return -EINVAL;
+}
+
+static int
+mlxsw_sp_nexthop_obj_bucket_adj_update(struct mlxsw_sp *mlxsw_sp,
+				       struct mlxsw_sp_nexthop *nh,
+				       struct nh_notifier_info *info)
+{
+	u16 bucket_index = info->nh_res_bucket->bucket_index;
+	struct netlink_ext_ack *extack = info->extack;
+	bool force = info->nh_res_bucket->force;
+	char ratr_pl_new[MLXSW_REG_RATR_LEN];
+	char ratr_pl[MLXSW_REG_RATR_LEN];
+	u32 adj_index;
+	int err;
+
+	/* No point in trying an atomic replacement if the idle timer interval
+	 * is smaller than the interval in which we query and clear activity.
+	 */
+	if (!force && info->nh_res_bucket->idle_timer_ms <
+	    MLXSW_SP_NH_GRP_ACTIVITY_UPDATE_INTERVAL)
+		force = true;
+
+	adj_index = nh->nhgi->adj_index + bucket_index;
+	err = mlxsw_sp_nexthop_update(mlxsw_sp, adj_index, nh, force, ratr_pl);
+	if (err) {
+		NL_SET_ERR_MSG_MOD(extack, "Failed to overwrite nexthop bucket");
+		return err;
+	}
+
+	if (!force) {
+		err = mlxsw_sp_nexthop_obj_bucket_query(mlxsw_sp, adj_index,
+							ratr_pl_new);
+		if (err) {
+			NL_SET_ERR_MSG_MOD(extack, "Failed to query nexthop bucket state after replacement. State might be inconsistent");
+			return err;
+		}
+
+		err = mlxsw_sp_nexthop_obj_bucket_compare(ratr_pl, ratr_pl_new);
+		if (err) {
+			NL_SET_ERR_MSG_MOD(extack, "Nexthop bucket was not replaced because it was active during replacement");
+			return err;
+		}
+	}
+
+	nh->update = 0;
+	nh->offloaded = 1;
+	mlxsw_sp_nexthop_bucket_offload_refresh(mlxsw_sp, nh, bucket_index);
+
+	return 0;
+}
+
+static int mlxsw_sp_nexthop_obj_bucket_replace(struct mlxsw_sp *mlxsw_sp,
+					       struct nh_notifier_info *info)
+{
+	u16 bucket_index = info->nh_res_bucket->bucket_index;
+	struct netlink_ext_ack *extack = info->extack;
+	struct mlxsw_sp_nexthop_group_info *nhgi;
+	struct nh_notifier_single_info *nh_obj;
+	struct mlxsw_sp_nexthop_group *nh_grp;
+	struct mlxsw_sp_nexthop *nh;
+	int err;
+
+	nh_grp = mlxsw_sp_nexthop_obj_group_lookup(mlxsw_sp, info->id);
+	if (!nh_grp) {
+		NL_SET_ERR_MSG_MOD(extack, "Nexthop group was not found");
+		return -EINVAL;
+	}
+
+	nhgi = nh_grp->nhgi;
+
+	if (bucket_index >= nhgi->count) {
+		NL_SET_ERR_MSG_MOD(extack, "Nexthop bucket index out of range");
+		return -EINVAL;
+	}
+
+	nh = &nhgi->nexthops[bucket_index];
+	mlxsw_sp_nexthop_obj_fini(mlxsw_sp, nh);
+
+	nh_obj = &info->nh_res_bucket->new_nh;
+	err = mlxsw_sp_nexthop_obj_init(mlxsw_sp, nh_grp, nh, nh_obj, 1);
+	if (err) {
+		NL_SET_ERR_MSG_MOD(extack, "Failed to initialize nexthop object for nexthop bucket replacement");
+		goto err_nexthop_obj_init;
+	}
+
+	err = mlxsw_sp_nexthop_obj_bucket_adj_update(mlxsw_sp, nh, info);
+	if (err)
+		goto err_nexthop_obj_bucket_adj_update;
+
+	return 0;
+
+err_nexthop_obj_bucket_adj_update:
+	mlxsw_sp_nexthop_obj_fini(mlxsw_sp, nh);
+err_nexthop_obj_init:
+	nh_obj = &info->nh_res_bucket->old_nh;
+	mlxsw_sp_nexthop_obj_init(mlxsw_sp, nh_grp, nh, nh_obj, 1);
+	/* The old adjacency entry was not overwritten */
+	nh->update = 0;
+	nh->offloaded = 1;
+	return err;
+}
+
+static int mlxsw_sp_nexthop_obj_event(struct notifier_block *nb,
+				      unsigned long event, void *ptr)
+{
+	struct nh_notifier_info *info = ptr;
+	struct mlxsw_sp_router *router;
+	int err = 0;
+
+	router = container_of(nb, struct mlxsw_sp_router, nexthop_nb);
+	err = mlxsw_sp_nexthop_obj_validate(router->mlxsw_sp, event, info);
+	if (err)
+		goto out;
+
+	mutex_lock(&router->lock);
+
+	switch (event) {
+	case NEXTHOP_EVENT_REPLACE:
+		err = mlxsw_sp_nexthop_obj_new(router->mlxsw_sp, info);
+		break;
+	case NEXTHOP_EVENT_DEL:
+		mlxsw_sp_nexthop_obj_del(router->mlxsw_sp, info);
+		break;
+	case NEXTHOP_EVENT_BUCKET_REPLACE:
+		err = mlxsw_sp_nexthop_obj_bucket_replace(router->mlxsw_sp,
+							  info);
+		break;
+	default:
+		break;
+	}
+
+	mutex_unlock(&router->lock);
+
+out:
+	return notifier_from_errno(err);
+}
+
+static bool mlxsw_sp_fi_is_gateway(const struct mlxsw_sp *mlxsw_sp,
+				   struct fib_info *fi)
+{
+	const struct fib_nh *nh = fib_info_nh(fi, 0);
+
+	return nh->fib_nh_gw_family ||
+	       mlxsw_sp_nexthop4_ipip_type(mlxsw_sp, nh, NULL);
+}
+
+static int
+mlxsw_sp_nexthop4_group_info_init(struct mlxsw_sp *mlxsw_sp,
+				  struct mlxsw_sp_nexthop_group *nh_grp)
+{
+	unsigned int nhs = fib_info_num_path(nh_grp->ipv4.fi);
+	struct mlxsw_sp_nexthop_group_info *nhgi;
+	struct mlxsw_sp_nexthop *nh;
+	int err, i;
+
+	nhgi = kzalloc(struct_size(nhgi, nexthops, nhs), GFP_KERNEL);
+	if (!nhgi)
+		return -ENOMEM;
+	nh_grp->nhgi = nhgi;
+	nhgi->nh_grp = nh_grp;
+	nhgi->gateway = mlxsw_sp_fi_is_gateway(mlxsw_sp, nh_grp->ipv4.fi);
+	nhgi->count = nhs;
+	for (i = 0; i < nhgi->count; i++) {
+		struct fib_nh *fib_nh;
+
+		nh = &nhgi->nexthops[i];
+		fib_nh = fib_info_nh(nh_grp->ipv4.fi, i);
+		err = mlxsw_sp_nexthop4_init(mlxsw_sp, nh_grp, nh, fib_nh);
+		if (err)
+			goto err_nexthop4_init;
+	}
+	err = mlxsw_sp_nexthop_group_inc(mlxsw_sp);
+	if (err)
+		goto err_group_inc;
+	err = mlxsw_sp_nexthop_group_refresh(mlxsw_sp, nh_grp);
+	if (err)
+		goto err_group_refresh;
+
+	return 0;
+
+err_group_refresh:
+	mlxsw_sp_nexthop_group_dec(mlxsw_sp);
+err_group_inc:
+	i = nhgi->count;
+err_nexthop4_init:
+	for (i--; i >= 0; i--) {
+		nh = &nhgi->nexthops[i];
+		mlxsw_sp_nexthop4_fini(mlxsw_sp, nh);
+	}
+	kfree(nhgi);
+	return err;
+}
+
+static void
+mlxsw_sp_nexthop4_group_info_fini(struct mlxsw_sp *mlxsw_sp,
+				  struct mlxsw_sp_nexthop_group *nh_grp)
+{
+	struct mlxsw_sp_nexthop_group_info *nhgi = nh_grp->nhgi;
+	int i;
+
+	mlxsw_sp_nexthop_group_dec(mlxsw_sp);
+	for (i = nhgi->count - 1; i >= 0; i--) {
+		struct mlxsw_sp_nexthop *nh = &nhgi->nexthops[i];
+
+		mlxsw_sp_nexthop4_fini(mlxsw_sp, nh);
+	}
+	mlxsw_sp_nexthop_group_refresh(mlxsw_sp, nh_grp);
+	WARN_ON_ONCE(nhgi->adj_index_valid);
+	kfree(nhgi);
+}
+
+static struct mlxsw_sp_nexthop_group *
+mlxsw_sp_nexthop4_group_create(struct mlxsw_sp *mlxsw_sp, struct fib_info *fi)
+{
+	struct mlxsw_sp_nexthop_group *nh_grp;
+	int err;
+
+	nh_grp = kzalloc(sizeof(*nh_grp), GFP_KERNEL);
+	if (!nh_grp)
+		return ERR_PTR(-ENOMEM);
+	INIT_LIST_HEAD(&nh_grp->vr_list);
+	err = rhashtable_init(&nh_grp->vr_ht,
+			      &mlxsw_sp_nexthop_group_vr_ht_params);
+	if (err)
+		goto err_nexthop_group_vr_ht_init;
+	INIT_LIST_HEAD(&nh_grp->fib_list);
+	nh_grp->type = MLXSW_SP_NEXTHOP_GROUP_TYPE_IPV4;
+	nh_grp->ipv4.fi = fi;
+	fib_info_hold(fi);
+
+	err = mlxsw_sp_nexthop4_group_info_init(mlxsw_sp, nh_grp);
+	if (err)
+		goto err_nexthop_group_info_init;
+
+	err = mlxsw_sp_nexthop_group_insert(mlxsw_sp, nh_grp);
+	if (err)
+		goto err_nexthop_group_insert;
+
+	nh_grp->can_destroy = true;
+
+	return nh_grp;
+
+err_nexthop_group_insert:
+	mlxsw_sp_nexthop4_group_info_fini(mlxsw_sp, nh_grp);
+err_nexthop_group_info_init:
+	fib_info_put(fi);
+	rhashtable_destroy(&nh_grp->vr_ht);
+err_nexthop_group_vr_ht_init:
+	kfree(nh_grp);
+	return ERR_PTR(err);
+}
+
+static void
+mlxsw_sp_nexthop4_group_destroy(struct mlxsw_sp *mlxsw_sp,
+				struct mlxsw_sp_nexthop_group *nh_grp)
+{
+	if (!nh_grp->can_destroy)
+		return;
+	mlxsw_sp_nexthop_group_remove(mlxsw_sp, nh_grp);
+	mlxsw_sp_nexthop4_group_info_fini(mlxsw_sp, nh_grp);
+	fib_info_put(nh_grp->ipv4.fi);
+	WARN_ON_ONCE(!list_empty(&nh_grp->vr_list));
+	rhashtable_destroy(&nh_grp->vr_ht);
+	kfree(nh_grp);
+}
+
+static int mlxsw_sp_nexthop4_group_get(struct mlxsw_sp *mlxsw_sp,
+				       struct mlxsw_sp_fib_entry *fib_entry,
+				       struct fib_info *fi)
+{
+	struct mlxsw_sp_nexthop_group *nh_grp;
+
+	if (fi->nh) {
+		nh_grp = mlxsw_sp_nexthop_obj_group_lookup(mlxsw_sp,
+							   fi->nh->id);
+		if (WARN_ON_ONCE(!nh_grp))
+			return -EINVAL;
+		goto out;
+	}
+
+	nh_grp = mlxsw_sp_nexthop4_group_lookup(mlxsw_sp, fi);
+	if (!nh_grp) {
+		nh_grp = mlxsw_sp_nexthop4_group_create(mlxsw_sp, fi);
+		if (IS_ERR(nh_grp))
+			return PTR_ERR(nh_grp);
+	}
+out:
+	list_add_tail(&fib_entry->nexthop_group_node, &nh_grp->fib_list);
+	fib_entry->nh_group = nh_grp;
+	return 0;
+}
+
+static void mlxsw_sp_nexthop4_group_put(struct mlxsw_sp *mlxsw_sp,
+					struct mlxsw_sp_fib_entry *fib_entry)
+{
+	struct mlxsw_sp_nexthop_group *nh_grp = fib_entry->nh_group;
+
+	list_del(&fib_entry->nexthop_group_node);
+	if (!list_empty(&nh_grp->fib_list))
+		return;
+
+	if (nh_grp->type == MLXSW_SP_NEXTHOP_GROUP_TYPE_OBJ) {
+		mlxsw_sp_nexthop_obj_group_destroy(mlxsw_sp, nh_grp);
+		return;
+	}
+
+	mlxsw_sp_nexthop4_group_destroy(mlxsw_sp, nh_grp);
+}
+
+static bool
+mlxsw_sp_fib4_entry_should_offload(const struct mlxsw_sp_fib_entry *fib_entry)
+{
+	struct mlxsw_sp_fib4_entry *fib4_entry;
+
+	fib4_entry = container_of(fib_entry, struct mlxsw_sp_fib4_entry,
+				  common);
+	return !fib4_entry->dscp;
+}
+
+static bool
+mlxsw_sp_fib_entry_should_offload(const struct mlxsw_sp_fib_entry *fib_entry)
+{
+	struct mlxsw_sp_nexthop_group *nh_group = fib_entry->nh_group;
+
+	switch (fib_entry->fib_node->fib->proto) {
+	case MLXSW_SP_L3_PROTO_IPV4:
+		if (!mlxsw_sp_fib4_entry_should_offload(fib_entry))
+			return false;
+		break;
+	case MLXSW_SP_L3_PROTO_IPV6:
+		break;
+	}
+
+	switch (fib_entry->type) {
+	case MLXSW_SP_FIB_ENTRY_TYPE_REMOTE:
+		return !!nh_group->nhgi->adj_index_valid;
+	case MLXSW_SP_FIB_ENTRY_TYPE_LOCAL:
+		return !!nh_group->nhgi->nh_rif;
+	case MLXSW_SP_FIB_ENTRY_TYPE_BLACKHOLE:
+	case MLXSW_SP_FIB_ENTRY_TYPE_IPIP_DECAP:
+	case MLXSW_SP_FIB_ENTRY_TYPE_NVE_DECAP:
+		return true;
+	default:
+		return false;
+	}
+}
+
+static struct mlxsw_sp_nexthop *
+mlxsw_sp_rt6_nexthop(struct mlxsw_sp_nexthop_group *nh_grp,
+		     const struct mlxsw_sp_rt6 *mlxsw_sp_rt6)
+{
+	int i;
+
+	for (i = 0; i < nh_grp->nhgi->count; i++) {
+		struct mlxsw_sp_nexthop *nh = &nh_grp->nhgi->nexthops[i];
+		struct fib6_info *rt = mlxsw_sp_rt6->rt;
+
+		if (nh->rif && nh->rif->dev == rt->fib6_nh->fib_nh_dev &&
+		    ipv6_addr_equal((const struct in6_addr *) &nh->gw_addr,
+				    &rt->fib6_nh->fib_nh_gw6))
+			return nh;
+	}
+
+	return NULL;
+}
+
+static void
+mlxsw_sp_fib4_offload_failed_flag_set(struct mlxsw_sp *mlxsw_sp,
+				      struct fib_entry_notifier_info *fen_info)
+{
+	u32 *p_dst = (u32 *) &fen_info->dst;
+	struct fib_rt_info fri;
+
+	fri.fi = fen_info->fi;
+	fri.tb_id = fen_info->tb_id;
+	fri.dst = cpu_to_be32(*p_dst);
+	fri.dst_len = fen_info->dst_len;
+	fri.dscp = fen_info->dscp;
+	fri.type = fen_info->type;
+	fri.offload = false;
+	fri.trap = false;
+	fri.offload_failed = true;
+	fib_alias_hw_flags_set(mlxsw_sp_net(mlxsw_sp), &fri);
+}
+
+static void
+mlxsw_sp_fib4_entry_hw_flags_set(struct mlxsw_sp *mlxsw_sp,
+				 struct mlxsw_sp_fib_entry *fib_entry)
+{
+	u32 *p_dst = (u32 *) fib_entry->fib_node->key.addr;
+	int dst_len = fib_entry->fib_node->key.prefix_len;
+	struct mlxsw_sp_fib4_entry *fib4_entry;
+	struct fib_rt_info fri;
+	bool should_offload;
+
+	should_offload = mlxsw_sp_fib_entry_should_offload(fib_entry);
+	fib4_entry = container_of(fib_entry, struct mlxsw_sp_fib4_entry,
+				  common);
+	fri.fi = fib4_entry->fi;
+	fri.tb_id = fib4_entry->tb_id;
+	fri.dst = cpu_to_be32(*p_dst);
+	fri.dst_len = dst_len;
+	fri.dscp = fib4_entry->dscp;
+	fri.type = fib4_entry->type;
+	fri.offload = should_offload;
+	fri.trap = !should_offload;
+	fri.offload_failed = false;
+	fib_alias_hw_flags_set(mlxsw_sp_net(mlxsw_sp), &fri);
+}
+
+static void
+mlxsw_sp_fib4_entry_hw_flags_clear(struct mlxsw_sp *mlxsw_sp,
+				   struct mlxsw_sp_fib_entry *fib_entry)
+{
+	u32 *p_dst = (u32 *) fib_entry->fib_node->key.addr;
+	int dst_len = fib_entry->fib_node->key.prefix_len;
+	struct mlxsw_sp_fib4_entry *fib4_entry;
+	struct fib_rt_info fri;
+
+	fib4_entry = container_of(fib_entry, struct mlxsw_sp_fib4_entry,
+				  common);
+	fri.fi = fib4_entry->fi;
+	fri.tb_id = fib4_entry->tb_id;
+	fri.dst = cpu_to_be32(*p_dst);
+	fri.dst_len = dst_len;
+	fri.dscp = fib4_entry->dscp;
+	fri.type = fib4_entry->type;
+	fri.offload = false;
+	fri.trap = false;
+	fri.offload_failed = false;
+	fib_alias_hw_flags_set(mlxsw_sp_net(mlxsw_sp), &fri);
+}
+
+#if IS_ENABLED(CONFIG_IPV6)
+static void
+mlxsw_sp_fib6_offload_failed_flag_set(struct mlxsw_sp *mlxsw_sp,
+				      struct fib6_info **rt_arr,
+				      unsigned int nrt6)
+{
+	int i;
+
+	/* In IPv6 a multipath route is represented using multiple routes, so
+	 * we need to set the flags on all of them.
+	 */
+	for (i = 0; i < nrt6; i++)
+		fib6_info_hw_flags_set(mlxsw_sp_net(mlxsw_sp), rt_arr[i],
+				       false, false, true);
+}
+#else
+static void
+mlxsw_sp_fib6_offload_failed_flag_set(struct mlxsw_sp *mlxsw_sp,
+				      struct fib6_info **rt_arr,
+				      unsigned int nrt6)
+{
+}
+#endif
+
+#if IS_ENABLED(CONFIG_IPV6)
+static void
+mlxsw_sp_fib6_entry_hw_flags_set(struct mlxsw_sp *mlxsw_sp,
+				 struct mlxsw_sp_fib_entry *fib_entry)
+{
+	struct mlxsw_sp_fib6_entry *fib6_entry;
+	struct mlxsw_sp_rt6 *mlxsw_sp_rt6;
+	bool should_offload;
+
+	should_offload = mlxsw_sp_fib_entry_should_offload(fib_entry);
+
+	/* In IPv6 a multipath route is represented using multiple routes, so
+	 * we need to set the flags on all of them.
+	 */
+	fib6_entry = container_of(fib_entry, struct mlxsw_sp_fib6_entry,
+				  common);
+	list_for_each_entry(mlxsw_sp_rt6, &fib6_entry->rt6_list, list)
+		fib6_info_hw_flags_set(mlxsw_sp_net(mlxsw_sp), mlxsw_sp_rt6->rt,
+				       should_offload, !should_offload, false);
+}
+#else
+static void
+mlxsw_sp_fib6_entry_hw_flags_set(struct mlxsw_sp *mlxsw_sp,
+				 struct mlxsw_sp_fib_entry *fib_entry)
+{
+}
+#endif
+
+#if IS_ENABLED(CONFIG_IPV6)
+static void
+mlxsw_sp_fib6_entry_hw_flags_clear(struct mlxsw_sp *mlxsw_sp,
+				   struct mlxsw_sp_fib_entry *fib_entry)
+{
+	struct mlxsw_sp_fib6_entry *fib6_entry;
+	struct mlxsw_sp_rt6 *mlxsw_sp_rt6;
+
+	fib6_entry = container_of(fib_entry, struct mlxsw_sp_fib6_entry,
+				  common);
+	list_for_each_entry(mlxsw_sp_rt6, &fib6_entry->rt6_list, list)
+		fib6_info_hw_flags_set(mlxsw_sp_net(mlxsw_sp), mlxsw_sp_rt6->rt,
+				       false, false, false);
+}
+#else
+static void
+mlxsw_sp_fib6_entry_hw_flags_clear(struct mlxsw_sp *mlxsw_sp,
+				   struct mlxsw_sp_fib_entry *fib_entry)
+{
+}
+#endif
+
+static void
+mlxsw_sp_fib_entry_hw_flags_set(struct mlxsw_sp *mlxsw_sp,
+				struct mlxsw_sp_fib_entry *fib_entry)
+{
+	switch (fib_entry->fib_node->fib->proto) {
+	case MLXSW_SP_L3_PROTO_IPV4:
+		mlxsw_sp_fib4_entry_hw_flags_set(mlxsw_sp, fib_entry);
+		break;
+	case MLXSW_SP_L3_PROTO_IPV6:
+		mlxsw_sp_fib6_entry_hw_flags_set(mlxsw_sp, fib_entry);
+		break;
+	}
+}
+
+static void
+mlxsw_sp_fib_entry_hw_flags_clear(struct mlxsw_sp *mlxsw_sp,
+				  struct mlxsw_sp_fib_entry *fib_entry)
+{
+	switch (fib_entry->fib_node->fib->proto) {
+	case MLXSW_SP_L3_PROTO_IPV4:
+		mlxsw_sp_fib4_entry_hw_flags_clear(mlxsw_sp, fib_entry);
+		break;
+	case MLXSW_SP_L3_PROTO_IPV6:
+		mlxsw_sp_fib6_entry_hw_flags_clear(mlxsw_sp, fib_entry);
+		break;
+	}
+}
+
+static void
+mlxsw_sp_fib_entry_hw_flags_refresh(struct mlxsw_sp *mlxsw_sp,
+				    struct mlxsw_sp_fib_entry *fib_entry,
+				    enum mlxsw_reg_ralue_op op)
+{
+	switch (op) {
+	case MLXSW_REG_RALUE_OP_WRITE_WRITE:
+		mlxsw_sp_fib_entry_hw_flags_set(mlxsw_sp, fib_entry);
+		break;
+	case MLXSW_REG_RALUE_OP_WRITE_DELETE:
+		mlxsw_sp_fib_entry_hw_flags_clear(mlxsw_sp, fib_entry);
+		break;
+	default:
+		break;
+	}
+}
+
+static void
+mlxsw_sp_fib_entry_ralue_pack(char *ralue_pl,
+			      const struct mlxsw_sp_fib_entry *fib_entry,
+			      enum mlxsw_reg_ralue_op op)
+{
+	struct mlxsw_sp_fib *fib = fib_entry->fib_node->fib;
+	enum mlxsw_reg_ralxx_protocol proto;
+	u32 *p_dip;
+
+	proto = (enum mlxsw_reg_ralxx_protocol) fib->proto;
+
+	switch (fib->proto) {
+	case MLXSW_SP_L3_PROTO_IPV4:
+		p_dip = (u32 *) fib_entry->fib_node->key.addr;
+		mlxsw_reg_ralue_pack4(ralue_pl, proto, op, fib->vr->id,
+				      fib_entry->fib_node->key.prefix_len,
+				      *p_dip);
+		break;
+	case MLXSW_SP_L3_PROTO_IPV6:
+		mlxsw_reg_ralue_pack6(ralue_pl, proto, op, fib->vr->id,
+				      fib_entry->fib_node->key.prefix_len,
+				      fib_entry->fib_node->key.addr);
+		break;
+	}
+}
+
+static int mlxsw_sp_fib_entry_op_remote(struct mlxsw_sp *mlxsw_sp,
+					struct mlxsw_sp_fib_entry *fib_entry,
+					enum mlxsw_reg_ralue_op op)
+{
+	struct mlxsw_sp_nexthop_group *nh_group = fib_entry->nh_group;
+	struct mlxsw_sp_nexthop_group_info *nhgi = nh_group->nhgi;
+	char ralue_pl[MLXSW_REG_RALUE_LEN];
+	enum mlxsw_reg_ralue_trap_action trap_action;
+	u16 trap_id = 0;
+	u32 adjacency_index = 0;
+	u16 ecmp_size = 0;
+
+	/* In case the nexthop group adjacency index is valid, use it
+	 * with provided ECMP size. Otherwise, setup trap and pass
+	 * traffic to kernel.
+	 */
+	if (mlxsw_sp_fib_entry_should_offload(fib_entry)) {
+		trap_action = MLXSW_REG_RALUE_TRAP_ACTION_NOP;
+		adjacency_index = nhgi->adj_index;
+		ecmp_size = nhgi->ecmp_size;
+	} else if (!nhgi->adj_index_valid && nhgi->count && nhgi->nh_rif) {
+		trap_action = MLXSW_REG_RALUE_TRAP_ACTION_NOP;
+		adjacency_index = mlxsw_sp->router->adj_trap_index;
+		ecmp_size = 1;
+	} else {
+		trap_action = MLXSW_REG_RALUE_TRAP_ACTION_TRAP;
+		trap_id = MLXSW_TRAP_ID_RTR_INGRESS0;
+	}
+
+	mlxsw_sp_fib_entry_ralue_pack(ralue_pl, fib_entry, op);
+	mlxsw_reg_ralue_act_remote_pack(ralue_pl, trap_action, trap_id,
+					adjacency_index, ecmp_size);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(ralue), ralue_pl);
+}
+
+static int mlxsw_sp_fib_entry_op_local(struct mlxsw_sp *mlxsw_sp,
+				       struct mlxsw_sp_fib_entry *fib_entry,
+				       enum mlxsw_reg_ralue_op op)
+{
+	struct mlxsw_sp_rif *rif = fib_entry->nh_group->nhgi->nh_rif;
+	enum mlxsw_reg_ralue_trap_action trap_action;
+	char ralue_pl[MLXSW_REG_RALUE_LEN];
+	u16 trap_id = 0;
+	u16 rif_index = 0;
+
+	if (mlxsw_sp_fib_entry_should_offload(fib_entry)) {
+		trap_action = MLXSW_REG_RALUE_TRAP_ACTION_NOP;
+		rif_index = rif->rif_index;
+	} else {
+		trap_action = MLXSW_REG_RALUE_TRAP_ACTION_TRAP;
+		trap_id = MLXSW_TRAP_ID_RTR_INGRESS0;
+	}
+
+	mlxsw_sp_fib_entry_ralue_pack(ralue_pl, fib_entry, op);
+	mlxsw_reg_ralue_act_local_pack(ralue_pl, trap_action, trap_id,
+				       rif_index);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(ralue), ralue_pl);
+}
+
+static int mlxsw_sp_fib_entry_op_trap(struct mlxsw_sp *mlxsw_sp,
+				      struct mlxsw_sp_fib_entry *fib_entry,
+				      enum mlxsw_reg_ralue_op op)
+{
+	char ralue_pl[MLXSW_REG_RALUE_LEN];
+
+	mlxsw_sp_fib_entry_ralue_pack(ralue_pl, fib_entry, op);
+	mlxsw_reg_ralue_act_ip2me_pack(ralue_pl);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(ralue), ralue_pl);
+}
+
+static int mlxsw_sp_fib_entry_op_blackhole(struct mlxsw_sp *mlxsw_sp,
+					   struct mlxsw_sp_fib_entry *fib_entry,
+					   enum mlxsw_reg_ralue_op op)
+{
+	enum mlxsw_reg_ralue_trap_action trap_action;
+	char ralue_pl[MLXSW_REG_RALUE_LEN];
+
+	trap_action = MLXSW_REG_RALUE_TRAP_ACTION_DISCARD_ERROR;
+	mlxsw_sp_fib_entry_ralue_pack(ralue_pl, fib_entry, op);
+	mlxsw_reg_ralue_act_local_pack(ralue_pl, trap_action, 0, 0);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(ralue), ralue_pl);
+}
+
+static int
+mlxsw_sp_fib_entry_op_unreachable(struct mlxsw_sp *mlxsw_sp,
+				  struct mlxsw_sp_fib_entry *fib_entry,
+				  enum mlxsw_reg_ralue_op op)
+{
+	enum mlxsw_reg_ralue_trap_action trap_action;
+	char ralue_pl[MLXSW_REG_RALUE_LEN];
+	u16 trap_id;
+
+	trap_action = MLXSW_REG_RALUE_TRAP_ACTION_TRAP;
+	trap_id = MLXSW_TRAP_ID_RTR_INGRESS1;
+
+	mlxsw_sp_fib_entry_ralue_pack(ralue_pl, fib_entry, op);
+	mlxsw_reg_ralue_act_local_pack(ralue_pl, trap_action, trap_id, 0);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(ralue), ralue_pl);
+}
+
+static int
+mlxsw_sp_fib_entry_op_ipip_decap(struct mlxsw_sp *mlxsw_sp,
+				 struct mlxsw_sp_fib_entry *fib_entry,
+				 enum mlxsw_reg_ralue_op op)
+{
+	struct mlxsw_sp_ipip_entry *ipip_entry = fib_entry->decap.ipip_entry;
+	const struct mlxsw_sp_ipip_ops *ipip_ops;
+	char ralue_pl[MLXSW_REG_RALUE_LEN];
+	int err;
+
+	if (WARN_ON(!ipip_entry))
+		return -EINVAL;
+
+	ipip_ops = mlxsw_sp->router->ipip_ops_arr[ipip_entry->ipipt];
+	err = ipip_ops->decap_config(mlxsw_sp, ipip_entry,
+				     fib_entry->decap.tunnel_index);
+	if (err)
+		return err;
+
+	mlxsw_sp_fib_entry_ralue_pack(ralue_pl, fib_entry, op);
+	mlxsw_reg_ralue_act_ip2me_tun_pack(ralue_pl,
+					   fib_entry->decap.tunnel_index);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(ralue), ralue_pl);
+}
+
+static int mlxsw_sp_fib_entry_op_nve_decap(struct mlxsw_sp *mlxsw_sp,
+					   struct mlxsw_sp_fib_entry *fib_entry,
+					   enum mlxsw_reg_ralue_op op)
+{
+	char ralue_pl[MLXSW_REG_RALUE_LEN];
+
+	mlxsw_sp_fib_entry_ralue_pack(ralue_pl, fib_entry, op);
+	mlxsw_reg_ralue_act_ip2me_tun_pack(ralue_pl,
+					   fib_entry->decap.tunnel_index);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(ralue), ralue_pl);
+}
+
+static int __mlxsw_sp_fib_entry_op(struct mlxsw_sp *mlxsw_sp,
+				   struct mlxsw_sp_fib_entry *fib_entry,
+				   enum mlxsw_reg_ralue_op op)
+{
+	switch (fib_entry->type) {
+	case MLXSW_SP_FIB_ENTRY_TYPE_REMOTE:
+		return mlxsw_sp_fib_entry_op_remote(mlxsw_sp, fib_entry, op);
+	case MLXSW_SP_FIB_ENTRY_TYPE_LOCAL:
+		return mlxsw_sp_fib_entry_op_local(mlxsw_sp, fib_entry, op);
+	case MLXSW_SP_FIB_ENTRY_TYPE_TRAP:
+		return mlxsw_sp_fib_entry_op_trap(mlxsw_sp, fib_entry, op);
+	case MLXSW_SP_FIB_ENTRY_TYPE_BLACKHOLE:
+		return mlxsw_sp_fib_entry_op_blackhole(mlxsw_sp, fib_entry, op);
+	case MLXSW_SP_FIB_ENTRY_TYPE_UNREACHABLE:
+		return mlxsw_sp_fib_entry_op_unreachable(mlxsw_sp, fib_entry,
+							 op);
+	case MLXSW_SP_FIB_ENTRY_TYPE_IPIP_DECAP:
+		return mlxsw_sp_fib_entry_op_ipip_decap(mlxsw_sp,
+							fib_entry, op);
+	case MLXSW_SP_FIB_ENTRY_TYPE_NVE_DECAP:
+		return mlxsw_sp_fib_entry_op_nve_decap(mlxsw_sp, fib_entry, op);
+	}
+	return -EINVAL;
+}
+
+static int mlxsw_sp_fib_entry_op(struct mlxsw_sp *mlxsw_sp,
+				 struct mlxsw_sp_fib_entry *fib_entry,
+				 enum mlxsw_reg_ralue_op op)
+{
+	int err = __mlxsw_sp_fib_entry_op(mlxsw_sp, fib_entry, op);
+
+	if (err)
+		return err;
+
+	mlxsw_sp_fib_entry_hw_flags_refresh(mlxsw_sp, fib_entry, op);
+
+	return err;
+}
+
+static int mlxsw_sp_fib_entry_update(struct mlxsw_sp *mlxsw_sp,
+				     struct mlxsw_sp_fib_entry *fib_entry)
+{
+	return mlxsw_sp_fib_entry_op(mlxsw_sp, fib_entry,
+				     MLXSW_REG_RALUE_OP_WRITE_WRITE);
+}
+
+static int mlxsw_sp_fib_entry_del(struct mlxsw_sp *mlxsw_sp,
+				  struct mlxsw_sp_fib_entry *fib_entry)
+{
+	return mlxsw_sp_fib_entry_op(mlxsw_sp, fib_entry,
+				     MLXSW_REG_RALUE_OP_WRITE_DELETE);
+}
+
+static int
+mlxsw_sp_fib4_entry_type_set(struct mlxsw_sp *mlxsw_sp,
+			     const struct fib_entry_notifier_info *fen_info,
+			     struct mlxsw_sp_fib_entry *fib_entry)
+{
+	struct mlxsw_sp_nexthop_group_info *nhgi = fib_entry->nh_group->nhgi;
+	union mlxsw_sp_l3addr dip = { .addr4 = htonl(fen_info->dst) };
+	struct mlxsw_sp_router *router = mlxsw_sp->router;
+	u32 tb_id = mlxsw_sp_fix_tb_id(fen_info->tb_id);
+	int ifindex = nhgi->nexthops[0].ifindex;
+	struct mlxsw_sp_ipip_entry *ipip_entry;
+
+	switch (fen_info->type) {
+	case RTN_LOCAL:
+		ipip_entry = mlxsw_sp_ipip_entry_find_by_decap(mlxsw_sp, ifindex,
+							       MLXSW_SP_L3_PROTO_IPV4, dip);
+		if (ipip_entry && ipip_entry->ol_dev->flags & IFF_UP) {
+			fib_entry->type = MLXSW_SP_FIB_ENTRY_TYPE_IPIP_DECAP;
+			return mlxsw_sp_fib_entry_decap_init(mlxsw_sp,
+							     fib_entry,
+							     ipip_entry);
+		}
+		if (mlxsw_sp_router_nve_is_decap(mlxsw_sp, tb_id,
+						 MLXSW_SP_L3_PROTO_IPV4,
+						 &dip)) {
+			u32 tunnel_index;
+
+			tunnel_index = router->nve_decap_config.tunnel_index;
+			fib_entry->decap.tunnel_index = tunnel_index;
+			fib_entry->type = MLXSW_SP_FIB_ENTRY_TYPE_NVE_DECAP;
+			return 0;
+		}
+		fallthrough;
+	case RTN_BROADCAST:
+		fib_entry->type = MLXSW_SP_FIB_ENTRY_TYPE_TRAP;
+		return 0;
+	case RTN_BLACKHOLE:
+		fib_entry->type = MLXSW_SP_FIB_ENTRY_TYPE_BLACKHOLE;
+		return 0;
+	case RTN_UNREACHABLE:
+	case RTN_PROHIBIT:
+		/* Packets hitting these routes need to be trapped, but
+		 * can do so with a lower priority than packets directed
+		 * at the host, so use action type local instead of trap.
+		 */
+		fib_entry->type = MLXSW_SP_FIB_ENTRY_TYPE_UNREACHABLE;
+		return 0;
+	case RTN_UNICAST:
+		if (nhgi->gateway)
+			fib_entry->type = MLXSW_SP_FIB_ENTRY_TYPE_REMOTE;
+		else
+			fib_entry->type = MLXSW_SP_FIB_ENTRY_TYPE_LOCAL;
+		return 0;
+	default:
+		return -EINVAL;
+	}
+}
+
+static void
+mlxsw_sp_fib_entry_type_unset(struct mlxsw_sp *mlxsw_sp,
+			      struct mlxsw_sp_fib_entry *fib_entry)
+{
+	switch (fib_entry->type) {
+	case MLXSW_SP_FIB_ENTRY_TYPE_IPIP_DECAP:
+		mlxsw_sp_fib_entry_decap_fini(mlxsw_sp, fib_entry);
+		break;
+	default:
+		break;
+	}
+}
+
+static void
+mlxsw_sp_fib4_entry_type_unset(struct mlxsw_sp *mlxsw_sp,
+			       struct mlxsw_sp_fib4_entry *fib4_entry)
+{
+	mlxsw_sp_fib_entry_type_unset(mlxsw_sp, &fib4_entry->common);
+}
+
+static struct mlxsw_sp_fib4_entry *
+mlxsw_sp_fib4_entry_create(struct mlxsw_sp *mlxsw_sp,
+			   struct mlxsw_sp_fib_node *fib_node,
+			   const struct fib_entry_notifier_info *fen_info)
+{
+	struct mlxsw_sp_fib4_entry *fib4_entry;
+	struct mlxsw_sp_fib_entry *fib_entry;
+	int err;
+
+	fib4_entry = kzalloc(sizeof(*fib4_entry), GFP_KERNEL);
+	if (!fib4_entry)
+		return ERR_PTR(-ENOMEM);
+	fib_entry = &fib4_entry->common;
+
+	err = mlxsw_sp_nexthop4_group_get(mlxsw_sp, fib_entry, fen_info->fi);
+	if (err)
+		goto err_nexthop4_group_get;
+
+	err = mlxsw_sp_nexthop_group_vr_link(fib_entry->nh_group,
+					     fib_node->fib);
+	if (err)
+		goto err_nexthop_group_vr_link;
+
+	err = mlxsw_sp_fib4_entry_type_set(mlxsw_sp, fen_info, fib_entry);
+	if (err)
+		goto err_fib4_entry_type_set;
+
+	fib4_entry->fi = fen_info->fi;
+	fib_info_hold(fib4_entry->fi);
+	fib4_entry->tb_id = fen_info->tb_id;
+	fib4_entry->type = fen_info->type;
+	fib4_entry->dscp = fen_info->dscp;
+
+	fib_entry->fib_node = fib_node;
+
+	return fib4_entry;
+
+err_fib4_entry_type_set:
+	mlxsw_sp_nexthop_group_vr_unlink(fib_entry->nh_group, fib_node->fib);
+err_nexthop_group_vr_link:
+	mlxsw_sp_nexthop4_group_put(mlxsw_sp, &fib4_entry->common);
+err_nexthop4_group_get:
+	kfree(fib4_entry);
+	return ERR_PTR(err);
+}
+
+static void mlxsw_sp_fib4_entry_destroy(struct mlxsw_sp *mlxsw_sp,
+					struct mlxsw_sp_fib4_entry *fib4_entry)
+{
+	struct mlxsw_sp_fib_node *fib_node = fib4_entry->common.fib_node;
+
+	fib_info_put(fib4_entry->fi);
+	mlxsw_sp_fib4_entry_type_unset(mlxsw_sp, fib4_entry);
+	mlxsw_sp_nexthop_group_vr_unlink(fib4_entry->common.nh_group,
+					 fib_node->fib);
+	mlxsw_sp_nexthop4_group_put(mlxsw_sp, &fib4_entry->common);
+	kfree(fib4_entry);
+}
+
+static struct mlxsw_sp_fib4_entry *
+mlxsw_sp_fib4_entry_lookup(struct mlxsw_sp *mlxsw_sp,
+			   const struct fib_entry_notifier_info *fen_info)
+{
+	struct mlxsw_sp_fib4_entry *fib4_entry;
+	struct mlxsw_sp_fib_node *fib_node;
+	struct mlxsw_sp_fib *fib;
+	struct mlxsw_sp_vr *vr;
+
+	vr = mlxsw_sp_vr_find(mlxsw_sp, fen_info->tb_id);
+	if (!vr)
+		return NULL;
+	fib = mlxsw_sp_vr_fib(vr, MLXSW_SP_L3_PROTO_IPV4);
+
+	fib_node = mlxsw_sp_fib_node_lookup(fib, &fen_info->dst,
+					    sizeof(fen_info->dst),
+					    fen_info->dst_len);
+	if (!fib_node)
+		return NULL;
+
+	fib4_entry = container_of(fib_node->fib_entry,
+				  struct mlxsw_sp_fib4_entry, common);
+	if (fib4_entry->tb_id == fen_info->tb_id &&
+	    fib4_entry->dscp == fen_info->dscp &&
+	    fib4_entry->type == fen_info->type &&
+	    fib4_entry->fi == fen_info->fi)
+		return fib4_entry;
+
+	return NULL;
+}
+
+static const struct rhashtable_params mlxsw_sp_fib_ht_params = {
+	.key_offset = offsetof(struct mlxsw_sp_fib_node, key),
+	.head_offset = offsetof(struct mlxsw_sp_fib_node, ht_node),
+	.key_len = sizeof(struct mlxsw_sp_fib_key),
+	.automatic_shrinking = true,
+};
+
+static int mlxsw_sp_fib_node_insert(struct mlxsw_sp_fib *fib,
+				    struct mlxsw_sp_fib_node *fib_node)
+{
+	return rhashtable_insert_fast(&fib->ht, &fib_node->ht_node,
+				      mlxsw_sp_fib_ht_params);
+}
+
+static void mlxsw_sp_fib_node_remove(struct mlxsw_sp_fib *fib,
+				     struct mlxsw_sp_fib_node *fib_node)
+{
+	rhashtable_remove_fast(&fib->ht, &fib_node->ht_node,
+			       mlxsw_sp_fib_ht_params);
+}
+
+static struct mlxsw_sp_fib_node *
+mlxsw_sp_fib_node_lookup(struct mlxsw_sp_fib *fib, const void *addr,
+			 size_t addr_len, unsigned char prefix_len)
+{
+	struct mlxsw_sp_fib_key key;
+
+	memset(&key, 0, sizeof(key));
+	memcpy(key.addr, addr, addr_len);
+	key.prefix_len = prefix_len;
+	return rhashtable_lookup_fast(&fib->ht, &key, mlxsw_sp_fib_ht_params);
+}
+
+static struct mlxsw_sp_fib_node *
+mlxsw_sp_fib_node_create(struct mlxsw_sp_fib *fib, const void *addr,
+			 size_t addr_len, unsigned char prefix_len)
+{
+	struct mlxsw_sp_fib_node *fib_node;
+
+	fib_node = kzalloc(sizeof(*fib_node), GFP_KERNEL);
+	if (!fib_node)
+		return NULL;
+
+	list_add(&fib_node->list, &fib->node_list);
+	memcpy(fib_node->key.addr, addr, addr_len);
+	fib_node->key.prefix_len = prefix_len;
+
+	return fib_node;
+}
+
+static void mlxsw_sp_fib_node_destroy(struct mlxsw_sp_fib_node *fib_node)
+{
+	list_del(&fib_node->list);
+	kfree(fib_node);
+}
+
+static int mlxsw_sp_fib_lpm_tree_link(struct mlxsw_sp *mlxsw_sp,
+				      struct mlxsw_sp_fib_node *fib_node)
+{
+	struct mlxsw_sp_prefix_usage req_prefix_usage;
+	struct mlxsw_sp_fib *fib = fib_node->fib;
+	struct mlxsw_sp_lpm_tree *lpm_tree;
+	int err;
+
+	lpm_tree = mlxsw_sp->router->lpm.proto_trees[fib->proto];
+	if (lpm_tree->prefix_ref_count[fib_node->key.prefix_len] != 0)
+		goto out;
+
+	mlxsw_sp_prefix_usage_cpy(&req_prefix_usage, &lpm_tree->prefix_usage);
+	mlxsw_sp_prefix_usage_set(&req_prefix_usage, fib_node->key.prefix_len);
+	lpm_tree = mlxsw_sp_lpm_tree_get(mlxsw_sp, &req_prefix_usage,
+					 fib->proto);
+	if (IS_ERR(lpm_tree))
+		return PTR_ERR(lpm_tree);
+
+	err = mlxsw_sp_vrs_lpm_tree_replace(mlxsw_sp, fib, lpm_tree);
+	if (err)
+		goto err_lpm_tree_replace;
+
+out:
+	lpm_tree->prefix_ref_count[fib_node->key.prefix_len]++;
+	return 0;
+
+err_lpm_tree_replace:
+	mlxsw_sp_lpm_tree_put(mlxsw_sp, lpm_tree);
+	return err;
+}
+
+static void mlxsw_sp_fib_lpm_tree_unlink(struct mlxsw_sp *mlxsw_sp,
+					 struct mlxsw_sp_fib_node *fib_node)
+{
+	struct mlxsw_sp_lpm_tree *lpm_tree = fib_node->fib->lpm_tree;
+	struct mlxsw_sp_prefix_usage req_prefix_usage;
+	struct mlxsw_sp_fib *fib = fib_node->fib;
+	int err;
+
+	if (--lpm_tree->prefix_ref_count[fib_node->key.prefix_len] != 0)
+		return;
+	/* Try to construct a new LPM tree from the current prefix usage
+	 * minus the unused one. If we fail, continue using the old one.
+	 */
+	mlxsw_sp_prefix_usage_cpy(&req_prefix_usage, &lpm_tree->prefix_usage);
+	mlxsw_sp_prefix_usage_clear(&req_prefix_usage,
+				    fib_node->key.prefix_len);
+	lpm_tree = mlxsw_sp_lpm_tree_get(mlxsw_sp, &req_prefix_usage,
+					 fib->proto);
+	if (IS_ERR(lpm_tree))
+		return;
+
+	err = mlxsw_sp_vrs_lpm_tree_replace(mlxsw_sp, fib, lpm_tree);
+	if (err)
+		goto err_lpm_tree_replace;
+
+	return;
+
+err_lpm_tree_replace:
+	mlxsw_sp_lpm_tree_put(mlxsw_sp, lpm_tree);
+}
+
+static int mlxsw_sp_fib_node_init(struct mlxsw_sp *mlxsw_sp,
+				  struct mlxsw_sp_fib_node *fib_node,
+				  struct mlxsw_sp_fib *fib)
+{
+	int err;
+
+	err = mlxsw_sp_fib_node_insert(fib, fib_node);
+	if (err)
+		return err;
+	fib_node->fib = fib;
+
+	err = mlxsw_sp_fib_lpm_tree_link(mlxsw_sp, fib_node);
+	if (err)
+		goto err_fib_lpm_tree_link;
+
+	return 0;
+
+err_fib_lpm_tree_link:
+	fib_node->fib = NULL;
+	mlxsw_sp_fib_node_remove(fib, fib_node);
+	return err;
+}
+
+static void mlxsw_sp_fib_node_fini(struct mlxsw_sp *mlxsw_sp,
+				   struct mlxsw_sp_fib_node *fib_node)
+{
+	struct mlxsw_sp_fib *fib = fib_node->fib;
+
+	mlxsw_sp_fib_lpm_tree_unlink(mlxsw_sp, fib_node);
+	fib_node->fib = NULL;
+	mlxsw_sp_fib_node_remove(fib, fib_node);
+}
+
+static struct mlxsw_sp_fib_node *
+mlxsw_sp_fib_node_get(struct mlxsw_sp *mlxsw_sp, u32 tb_id, const void *addr,
+		      size_t addr_len, unsigned char prefix_len,
+		      enum mlxsw_sp_l3proto proto)
+{
+	struct mlxsw_sp_fib_node *fib_node;
+	struct mlxsw_sp_fib *fib;
+	struct mlxsw_sp_vr *vr;
+	int err;
+
+	vr = mlxsw_sp_vr_get(mlxsw_sp, tb_id, NULL);
+	if (IS_ERR(vr))
+		return ERR_CAST(vr);
+	fib = mlxsw_sp_vr_fib(vr, proto);
+
+	fib_node = mlxsw_sp_fib_node_lookup(fib, addr, addr_len, prefix_len);
+	if (fib_node)
+		return fib_node;
+
+	fib_node = mlxsw_sp_fib_node_create(fib, addr, addr_len, prefix_len);
+	if (!fib_node) {
+		err = -ENOMEM;
+		goto err_fib_node_create;
+	}
+
+	err = mlxsw_sp_fib_node_init(mlxsw_sp, fib_node, fib);
+	if (err)
+		goto err_fib_node_init;
+
+	return fib_node;
+
+err_fib_node_init:
+	mlxsw_sp_fib_node_destroy(fib_node);
+err_fib_node_create:
+	mlxsw_sp_vr_put(mlxsw_sp, vr);
+	return ERR_PTR(err);
+}
+
+static void mlxsw_sp_fib_node_put(struct mlxsw_sp *mlxsw_sp,
+				  struct mlxsw_sp_fib_node *fib_node)
+{
+	struct mlxsw_sp_vr *vr = fib_node->fib->vr;
+
+	if (fib_node->fib_entry)
+		return;
+	mlxsw_sp_fib_node_fini(mlxsw_sp, fib_node);
+	mlxsw_sp_fib_node_destroy(fib_node);
+	mlxsw_sp_vr_put(mlxsw_sp, vr);
+}
+
+static int mlxsw_sp_fib_node_entry_link(struct mlxsw_sp *mlxsw_sp,
+					struct mlxsw_sp_fib_entry *fib_entry)
+{
+	struct mlxsw_sp_fib_node *fib_node = fib_entry->fib_node;
+	int err;
+
+	fib_node->fib_entry = fib_entry;
+
+	err = mlxsw_sp_fib_entry_update(mlxsw_sp, fib_entry);
+	if (err)
+		goto err_fib_entry_update;
+
+	return 0;
+
+err_fib_entry_update:
+	fib_node->fib_entry = NULL;
+	return err;
+}
+
+static void
+mlxsw_sp_fib_node_entry_unlink(struct mlxsw_sp *mlxsw_sp,
+			       struct mlxsw_sp_fib_entry *fib_entry)
+{
+	struct mlxsw_sp_fib_node *fib_node = fib_entry->fib_node;
+
+	mlxsw_sp_fib_entry_del(mlxsw_sp, fib_entry);
+	fib_node->fib_entry = NULL;
+}
+
+static bool mlxsw_sp_fib4_allow_replace(struct mlxsw_sp_fib4_entry *fib4_entry)
+{
+	struct mlxsw_sp_fib_node *fib_node = fib4_entry->common.fib_node;
+	struct mlxsw_sp_fib4_entry *fib4_replaced;
+
+	if (!fib_node->fib_entry)
+		return true;
+
+	fib4_replaced = container_of(fib_node->fib_entry,
+				     struct mlxsw_sp_fib4_entry, common);
+	if (fib4_entry->tb_id == RT_TABLE_MAIN &&
+	    fib4_replaced->tb_id == RT_TABLE_LOCAL)
+		return false;
+
+	return true;
+}
+
+static int
+mlxsw_sp_router_fib4_replace(struct mlxsw_sp *mlxsw_sp,
+			     const struct fib_entry_notifier_info *fen_info)
+{
+	struct mlxsw_sp_fib4_entry *fib4_entry, *fib4_replaced;
+	struct mlxsw_sp_fib_entry *replaced;
+	struct mlxsw_sp_fib_node *fib_node;
+	int err;
+
+	if (fen_info->fi->nh &&
+	    !mlxsw_sp_nexthop_obj_group_lookup(mlxsw_sp, fen_info->fi->nh->id))
+		return 0;
+
+	fib_node = mlxsw_sp_fib_node_get(mlxsw_sp, fen_info->tb_id,
+					 &fen_info->dst, sizeof(fen_info->dst),
+					 fen_info->dst_len,
+					 MLXSW_SP_L3_PROTO_IPV4);
+	if (IS_ERR(fib_node)) {
+		dev_warn(mlxsw_sp->bus_info->dev, "Failed to get FIB node\n");
+		return PTR_ERR(fib_node);
+	}
+
+	fib4_entry = mlxsw_sp_fib4_entry_create(mlxsw_sp, fib_node, fen_info);
+	if (IS_ERR(fib4_entry)) {
+		dev_warn(mlxsw_sp->bus_info->dev, "Failed to create FIB entry\n");
+		err = PTR_ERR(fib4_entry);
+		goto err_fib4_entry_create;
+	}
+
+	if (!mlxsw_sp_fib4_allow_replace(fib4_entry)) {
+		mlxsw_sp_fib4_entry_destroy(mlxsw_sp, fib4_entry);
+		mlxsw_sp_fib_node_put(mlxsw_sp, fib_node);
+		return 0;
+	}
+
+	replaced = fib_node->fib_entry;
+	err = mlxsw_sp_fib_node_entry_link(mlxsw_sp, &fib4_entry->common);
+	if (err) {
+		dev_warn(mlxsw_sp->bus_info->dev, "Failed to link FIB entry to node\n");
+		goto err_fib_node_entry_link;
+	}
+
+	/* Nothing to replace */
+	if (!replaced)
+		return 0;
+
+	mlxsw_sp_fib_entry_hw_flags_clear(mlxsw_sp, replaced);
+	fib4_replaced = container_of(replaced, struct mlxsw_sp_fib4_entry,
+				     common);
+	mlxsw_sp_fib4_entry_destroy(mlxsw_sp, fib4_replaced);
+
+	return 0;
+
+err_fib_node_entry_link:
+	fib_node->fib_entry = replaced;
+	mlxsw_sp_fib4_entry_destroy(mlxsw_sp, fib4_entry);
+err_fib4_entry_create:
+	mlxsw_sp_fib_node_put(mlxsw_sp, fib_node);
+	return err;
+}
+
+static void mlxsw_sp_router_fib4_del(struct mlxsw_sp *mlxsw_sp,
+				     struct fib_entry_notifier_info *fen_info)
+{
+	struct mlxsw_sp_fib4_entry *fib4_entry;
+	struct mlxsw_sp_fib_node *fib_node;
+
+	fib4_entry = mlxsw_sp_fib4_entry_lookup(mlxsw_sp, fen_info);
+	if (!fib4_entry)
+		return;
+	fib_node = fib4_entry->common.fib_node;
+
+	mlxsw_sp_fib_node_entry_unlink(mlxsw_sp, &fib4_entry->common);
+	mlxsw_sp_fib4_entry_destroy(mlxsw_sp, fib4_entry);
+	mlxsw_sp_fib_node_put(mlxsw_sp, fib_node);
+}
+
+static bool mlxsw_sp_fib6_rt_should_ignore(const struct fib6_info *rt)
+{
+	/* Multicast routes aren't supported, so ignore them. Neighbour
+	 * Discovery packets are specifically trapped.
+	 */
+	if (ipv6_addr_type(&rt->fib6_dst.addr) & IPV6_ADDR_MULTICAST)
+		return true;
+
+	/* Cloned routes are irrelevant in the forwarding path. */
+	if (rt->fib6_flags & RTF_CACHE)
+		return true;
+
+	return false;
+}
+
+static struct mlxsw_sp_rt6 *mlxsw_sp_rt6_create(struct fib6_info *rt)
+{
+	struct mlxsw_sp_rt6 *mlxsw_sp_rt6;
+
+	mlxsw_sp_rt6 = kzalloc(sizeof(*mlxsw_sp_rt6), GFP_KERNEL);
+	if (!mlxsw_sp_rt6)
+		return ERR_PTR(-ENOMEM);
+
+	/* In case of route replace, replaced route is deleted with
+	 * no notification. Take reference to prevent accessing freed
+	 * memory.
+	 */
+	mlxsw_sp_rt6->rt = rt;
+	fib6_info_hold(rt);
+
+	return mlxsw_sp_rt6;
+}
+
+#if IS_ENABLED(CONFIG_IPV6)
+static void mlxsw_sp_rt6_release(struct fib6_info *rt)
+{
+	fib6_info_release(rt);
+}
+#else
+static void mlxsw_sp_rt6_release(struct fib6_info *rt)
+{
+}
+#endif
+
+static void mlxsw_sp_rt6_destroy(struct mlxsw_sp_rt6 *mlxsw_sp_rt6)
+{
+	struct fib6_nh *fib6_nh = mlxsw_sp_rt6->rt->fib6_nh;
+
+	if (!mlxsw_sp_rt6->rt->nh)
+		fib6_nh->fib_nh_flags &= ~RTNH_F_OFFLOAD;
+	mlxsw_sp_rt6_release(mlxsw_sp_rt6->rt);
+	kfree(mlxsw_sp_rt6);
+}
+
+static struct fib6_info *
+mlxsw_sp_fib6_entry_rt(const struct mlxsw_sp_fib6_entry *fib6_entry)
+{
+	return list_first_entry(&fib6_entry->rt6_list, struct mlxsw_sp_rt6,
+				list)->rt;
+}
+
+static struct mlxsw_sp_rt6 *
+mlxsw_sp_fib6_entry_rt_find(const struct mlxsw_sp_fib6_entry *fib6_entry,
+			    const struct fib6_info *rt)
+{
+	struct mlxsw_sp_rt6 *mlxsw_sp_rt6;
+
+	list_for_each_entry(mlxsw_sp_rt6, &fib6_entry->rt6_list, list) {
+		if (mlxsw_sp_rt6->rt == rt)
+			return mlxsw_sp_rt6;
+	}
+
+	return NULL;
+}
+
+static bool mlxsw_sp_nexthop6_ipip_type(const struct mlxsw_sp *mlxsw_sp,
+					const struct fib6_info *rt,
+					enum mlxsw_sp_ipip_type *ret)
+{
+	return rt->fib6_nh->fib_nh_dev &&
+	       mlxsw_sp_netdev_ipip_type(mlxsw_sp, rt->fib6_nh->fib_nh_dev, ret);
+}
+
+static int mlxsw_sp_nexthop6_init(struct mlxsw_sp *mlxsw_sp,
+				  struct mlxsw_sp_nexthop_group *nh_grp,
+				  struct mlxsw_sp_nexthop *nh,
+				  const struct fib6_info *rt)
+{
+	struct net_device *dev = rt->fib6_nh->fib_nh_dev;
+	int err;
+
+	nh->nhgi = nh_grp->nhgi;
+	nh->nh_weight = rt->fib6_nh->fib_nh_weight;
+	memcpy(&nh->gw_addr, &rt->fib6_nh->fib_nh_gw6, sizeof(nh->gw_addr));
+#if IS_ENABLED(CONFIG_IPV6)
+	nh->neigh_tbl = &nd_tbl;
+#endif
+	mlxsw_sp_nexthop_counter_alloc(mlxsw_sp, nh);
+
+	list_add_tail(&nh->router_list_node, &mlxsw_sp->router->nexthop_list);
+
+	if (!dev)
+		return 0;
+	nh->ifindex = dev->ifindex;
+
+	err = mlxsw_sp_nexthop_type_init(mlxsw_sp, nh, dev);
+	if (err)
+		goto err_nexthop_type_init;
+
+	return 0;
+
+err_nexthop_type_init:
+	list_del(&nh->router_list_node);
+	mlxsw_sp_nexthop_counter_free(mlxsw_sp, nh);
+	return err;
+}
+
+static void mlxsw_sp_nexthop6_fini(struct mlxsw_sp *mlxsw_sp,
+				   struct mlxsw_sp_nexthop *nh)
+{
+	mlxsw_sp_nexthop_type_fini(mlxsw_sp, nh);
+	list_del(&nh->router_list_node);
+	mlxsw_sp_nexthop_counter_free(mlxsw_sp, nh);
+}
+
+static bool mlxsw_sp_rt6_is_gateway(const struct mlxsw_sp *mlxsw_sp,
+				    const struct fib6_info *rt)
+{
+	return rt->fib6_nh->fib_nh_gw_family ||
+	       mlxsw_sp_nexthop6_ipip_type(mlxsw_sp, rt, NULL);
+}
+
+static int
+mlxsw_sp_nexthop6_group_info_init(struct mlxsw_sp *mlxsw_sp,
+				  struct mlxsw_sp_nexthop_group *nh_grp,
+				  struct mlxsw_sp_fib6_entry *fib6_entry)
+{
+	struct mlxsw_sp_nexthop_group_info *nhgi;
+	struct mlxsw_sp_rt6 *mlxsw_sp_rt6;
+	struct mlxsw_sp_nexthop *nh;
+	int err, i;
+
+	nhgi = kzalloc(struct_size(nhgi, nexthops, fib6_entry->nrt6),
+		       GFP_KERNEL);
+	if (!nhgi)
+		return -ENOMEM;
+	nh_grp->nhgi = nhgi;
+	nhgi->nh_grp = nh_grp;
+	mlxsw_sp_rt6 = list_first_entry(&fib6_entry->rt6_list,
+					struct mlxsw_sp_rt6, list);
+	nhgi->gateway = mlxsw_sp_rt6_is_gateway(mlxsw_sp, mlxsw_sp_rt6->rt);
+	nhgi->count = fib6_entry->nrt6;
+	for (i = 0; i < nhgi->count; i++) {
+		struct fib6_info *rt = mlxsw_sp_rt6->rt;
+
+		nh = &nhgi->nexthops[i];
+		err = mlxsw_sp_nexthop6_init(mlxsw_sp, nh_grp, nh, rt);
+		if (err)
+			goto err_nexthop6_init;
+		mlxsw_sp_rt6 = list_next_entry(mlxsw_sp_rt6, list);
+	}
+	nh_grp->nhgi = nhgi;
+	err = mlxsw_sp_nexthop_group_inc(mlxsw_sp);
+	if (err)
+		goto err_group_inc;
+	err = mlxsw_sp_nexthop_group_refresh(mlxsw_sp, nh_grp);
+	if (err)
+		goto err_group_refresh;
+
+	return 0;
+
+err_group_refresh:
+	mlxsw_sp_nexthop_group_dec(mlxsw_sp);
+err_group_inc:
+	i = nhgi->count;
+err_nexthop6_init:
+	for (i--; i >= 0; i--) {
+		nh = &nhgi->nexthops[i];
+		mlxsw_sp_nexthop6_fini(mlxsw_sp, nh);
+	}
+	kfree(nhgi);
+	return err;
+}
+
+static void
+mlxsw_sp_nexthop6_group_info_fini(struct mlxsw_sp *mlxsw_sp,
+				  struct mlxsw_sp_nexthop_group *nh_grp)
+{
+	struct mlxsw_sp_nexthop_group_info *nhgi = nh_grp->nhgi;
+	int i;
+
+	mlxsw_sp_nexthop_group_dec(mlxsw_sp);
+	for (i = nhgi->count - 1; i >= 0; i--) {
+		struct mlxsw_sp_nexthop *nh = &nhgi->nexthops[i];
+
+		mlxsw_sp_nexthop6_fini(mlxsw_sp, nh);
+	}
+	mlxsw_sp_nexthop_group_refresh(mlxsw_sp, nh_grp);
+	WARN_ON_ONCE(nhgi->adj_index_valid);
+	kfree(nhgi);
+}
+
+static struct mlxsw_sp_nexthop_group *
+mlxsw_sp_nexthop6_group_create(struct mlxsw_sp *mlxsw_sp,
+			       struct mlxsw_sp_fib6_entry *fib6_entry)
+{
+	struct mlxsw_sp_nexthop_group *nh_grp;
+	int err;
+
+	nh_grp = kzalloc(sizeof(*nh_grp), GFP_KERNEL);
+	if (!nh_grp)
+		return ERR_PTR(-ENOMEM);
+	INIT_LIST_HEAD(&nh_grp->vr_list);
+	err = rhashtable_init(&nh_grp->vr_ht,
+			      &mlxsw_sp_nexthop_group_vr_ht_params);
+	if (err)
+		goto err_nexthop_group_vr_ht_init;
+	INIT_LIST_HEAD(&nh_grp->fib_list);
+	nh_grp->type = MLXSW_SP_NEXTHOP_GROUP_TYPE_IPV6;
+
+	err = mlxsw_sp_nexthop6_group_info_init(mlxsw_sp, nh_grp, fib6_entry);
+	if (err)
+		goto err_nexthop_group_info_init;
+
+	err = mlxsw_sp_nexthop_group_insert(mlxsw_sp, nh_grp);
+	if (err)
+		goto err_nexthop_group_insert;
+
+	nh_grp->can_destroy = true;
+
+	return nh_grp;
+
+err_nexthop_group_insert:
+	mlxsw_sp_nexthop6_group_info_fini(mlxsw_sp, nh_grp);
+err_nexthop_group_info_init:
+	rhashtable_destroy(&nh_grp->vr_ht);
+err_nexthop_group_vr_ht_init:
+	kfree(nh_grp);
+	return ERR_PTR(err);
+}
+
+static void
+mlxsw_sp_nexthop6_group_destroy(struct mlxsw_sp *mlxsw_sp,
+				struct mlxsw_sp_nexthop_group *nh_grp)
+{
+	if (!nh_grp->can_destroy)
+		return;
+	mlxsw_sp_nexthop_group_remove(mlxsw_sp, nh_grp);
+	mlxsw_sp_nexthop6_group_info_fini(mlxsw_sp, nh_grp);
+	WARN_ON_ONCE(!list_empty(&nh_grp->vr_list));
+	rhashtable_destroy(&nh_grp->vr_ht);
+	kfree(nh_grp);
+}
+
+static int mlxsw_sp_nexthop6_group_get(struct mlxsw_sp *mlxsw_sp,
+				       struct mlxsw_sp_fib6_entry *fib6_entry)
+{
+	struct fib6_info *rt = mlxsw_sp_fib6_entry_rt(fib6_entry);
+	struct mlxsw_sp_nexthop_group *nh_grp;
+
+	if (rt->nh) {
+		nh_grp = mlxsw_sp_nexthop_obj_group_lookup(mlxsw_sp,
+							   rt->nh->id);
+		if (WARN_ON_ONCE(!nh_grp))
+			return -EINVAL;
+		goto out;
+	}
+
+	nh_grp = mlxsw_sp_nexthop6_group_lookup(mlxsw_sp, fib6_entry);
+	if (!nh_grp) {
+		nh_grp = mlxsw_sp_nexthop6_group_create(mlxsw_sp, fib6_entry);
+		if (IS_ERR(nh_grp))
+			return PTR_ERR(nh_grp);
+	}
+
+	/* The route and the nexthop are described by the same struct, so we
+	 * need to the update the nexthop offload indication for the new route.
+	 */
+	__mlxsw_sp_nexthop6_group_offload_refresh(nh_grp, fib6_entry);
+
+out:
+	list_add_tail(&fib6_entry->common.nexthop_group_node,
+		      &nh_grp->fib_list);
+	fib6_entry->common.nh_group = nh_grp;
+
+	return 0;
+}
+
+static void mlxsw_sp_nexthop6_group_put(struct mlxsw_sp *mlxsw_sp,
+					struct mlxsw_sp_fib_entry *fib_entry)
+{
+	struct mlxsw_sp_nexthop_group *nh_grp = fib_entry->nh_group;
+
+	list_del(&fib_entry->nexthop_group_node);
+	if (!list_empty(&nh_grp->fib_list))
+		return;
+
+	if (nh_grp->type == MLXSW_SP_NEXTHOP_GROUP_TYPE_OBJ) {
+		mlxsw_sp_nexthop_obj_group_destroy(mlxsw_sp, nh_grp);
+		return;
+	}
+
+	mlxsw_sp_nexthop6_group_destroy(mlxsw_sp, nh_grp);
+}
+
+static int
+mlxsw_sp_nexthop6_group_update(struct mlxsw_sp *mlxsw_sp,
+			       struct mlxsw_sp_fib6_entry *fib6_entry)
+{
+	struct mlxsw_sp_nexthop_group *old_nh_grp = fib6_entry->common.nh_group;
+	struct mlxsw_sp_fib_node *fib_node = fib6_entry->common.fib_node;
+	int err;
+
+	mlxsw_sp_nexthop_group_vr_unlink(old_nh_grp, fib_node->fib);
+	fib6_entry->common.nh_group = NULL;
+	list_del(&fib6_entry->common.nexthop_group_node);
+
+	err = mlxsw_sp_nexthop6_group_get(mlxsw_sp, fib6_entry);
+	if (err)
+		goto err_nexthop6_group_get;
+
+	err = mlxsw_sp_nexthop_group_vr_link(fib6_entry->common.nh_group,
+					     fib_node->fib);
+	if (err)
+		goto err_nexthop_group_vr_link;
+
+	/* In case this entry is offloaded, then the adjacency index
+	 * currently associated with it in the device's table is that
+	 * of the old group. Start using the new one instead.
+	 */
+	err = mlxsw_sp_fib_entry_update(mlxsw_sp, &fib6_entry->common);
+	if (err)
+		goto err_fib_entry_update;
+
+	if (list_empty(&old_nh_grp->fib_list))
+		mlxsw_sp_nexthop6_group_destroy(mlxsw_sp, old_nh_grp);
+
+	return 0;
+
+err_fib_entry_update:
+	mlxsw_sp_nexthop_group_vr_unlink(fib6_entry->common.nh_group,
+					 fib_node->fib);
+err_nexthop_group_vr_link:
+	mlxsw_sp_nexthop6_group_put(mlxsw_sp, &fib6_entry->common);
+err_nexthop6_group_get:
+	list_add_tail(&fib6_entry->common.nexthop_group_node,
+		      &old_nh_grp->fib_list);
+	fib6_entry->common.nh_group = old_nh_grp;
+	mlxsw_sp_nexthop_group_vr_link(old_nh_grp, fib_node->fib);
+	return err;
+}
+
+static int
+mlxsw_sp_fib6_entry_nexthop_add(struct mlxsw_sp *mlxsw_sp,
+				struct mlxsw_sp_fib6_entry *fib6_entry,
+				struct fib6_info **rt_arr, unsigned int nrt6)
+{
+	struct mlxsw_sp_rt6 *mlxsw_sp_rt6;
+	int err, i;
+
+	for (i = 0; i < nrt6; i++) {
+		mlxsw_sp_rt6 = mlxsw_sp_rt6_create(rt_arr[i]);
+		if (IS_ERR(mlxsw_sp_rt6)) {
+			err = PTR_ERR(mlxsw_sp_rt6);
+			goto err_rt6_unwind;
+		}
+
+		list_add_tail(&mlxsw_sp_rt6->list, &fib6_entry->rt6_list);
+		fib6_entry->nrt6++;
+	}
+
+	err = mlxsw_sp_nexthop6_group_update(mlxsw_sp, fib6_entry);
+	if (err)
+		goto err_rt6_unwind;
+
+	return 0;
+
+err_rt6_unwind:
+	for (; i > 0; i--) {
+		fib6_entry->nrt6--;
+		mlxsw_sp_rt6 = list_last_entry(&fib6_entry->rt6_list,
+					       struct mlxsw_sp_rt6, list);
+		list_del(&mlxsw_sp_rt6->list);
+		mlxsw_sp_rt6_destroy(mlxsw_sp_rt6);
+	}
+	return err;
+}
+
+static void
+mlxsw_sp_fib6_entry_nexthop_del(struct mlxsw_sp *mlxsw_sp,
+				struct mlxsw_sp_fib6_entry *fib6_entry,
+				struct fib6_info **rt_arr, unsigned int nrt6)
+{
+	struct mlxsw_sp_rt6 *mlxsw_sp_rt6;
+	int i;
+
+	for (i = 0; i < nrt6; i++) {
+		mlxsw_sp_rt6 = mlxsw_sp_fib6_entry_rt_find(fib6_entry,
+							   rt_arr[i]);
+		if (WARN_ON_ONCE(!mlxsw_sp_rt6))
+			continue;
+
+		fib6_entry->nrt6--;
+		list_del(&mlxsw_sp_rt6->list);
+		mlxsw_sp_rt6_destroy(mlxsw_sp_rt6);
+	}
+
+	mlxsw_sp_nexthop6_group_update(mlxsw_sp, fib6_entry);
+}
+
+static int
+mlxsw_sp_fib6_entry_type_set_local(struct mlxsw_sp *mlxsw_sp,
+				   struct mlxsw_sp_fib_entry *fib_entry,
+				   const struct fib6_info *rt)
+{
+	struct mlxsw_sp_nexthop_group_info *nhgi = fib_entry->nh_group->nhgi;
+	union mlxsw_sp_l3addr dip = { .addr6 = rt->fib6_dst.addr };
+	u32 tb_id = mlxsw_sp_fix_tb_id(rt->fib6_table->tb6_id);
+	struct mlxsw_sp_router *router = mlxsw_sp->router;
+	int ifindex = nhgi->nexthops[0].ifindex;
+	struct mlxsw_sp_ipip_entry *ipip_entry;
+
+	fib_entry->type = MLXSW_SP_FIB_ENTRY_TYPE_TRAP;
+	ipip_entry = mlxsw_sp_ipip_entry_find_by_decap(mlxsw_sp, ifindex,
+						       MLXSW_SP_L3_PROTO_IPV6,
+						       dip);
+
+	if (ipip_entry && ipip_entry->ol_dev->flags & IFF_UP) {
+		fib_entry->type = MLXSW_SP_FIB_ENTRY_TYPE_IPIP_DECAP;
+		return mlxsw_sp_fib_entry_decap_init(mlxsw_sp, fib_entry,
+						     ipip_entry);
+	}
+	if (mlxsw_sp_router_nve_is_decap(mlxsw_sp, tb_id,
+					 MLXSW_SP_L3_PROTO_IPV6, &dip)) {
+		u32 tunnel_index;
+
+		tunnel_index = router->nve_decap_config.tunnel_index;
+		fib_entry->decap.tunnel_index = tunnel_index;
+		fib_entry->type = MLXSW_SP_FIB_ENTRY_TYPE_NVE_DECAP;
+	}
+
+	return 0;
+}
+
+static int mlxsw_sp_fib6_entry_type_set(struct mlxsw_sp *mlxsw_sp,
+					struct mlxsw_sp_fib_entry *fib_entry,
+					const struct fib6_info *rt)
+{
+	if (rt->fib6_flags & RTF_LOCAL)
+		return mlxsw_sp_fib6_entry_type_set_local(mlxsw_sp, fib_entry,
+							  rt);
+	if (rt->fib6_flags & RTF_ANYCAST)
+		fib_entry->type = MLXSW_SP_FIB_ENTRY_TYPE_TRAP;
+	else if (rt->fib6_type == RTN_BLACKHOLE)
+		fib_entry->type = MLXSW_SP_FIB_ENTRY_TYPE_BLACKHOLE;
+	else if (rt->fib6_flags & RTF_REJECT)
+		fib_entry->type = MLXSW_SP_FIB_ENTRY_TYPE_UNREACHABLE;
+	else if (fib_entry->nh_group->nhgi->gateway)
+		fib_entry->type = MLXSW_SP_FIB_ENTRY_TYPE_REMOTE;
+	else
+		fib_entry->type = MLXSW_SP_FIB_ENTRY_TYPE_LOCAL;
+
+	return 0;
+}
+
+static void
+mlxsw_sp_fib6_entry_rt_destroy_all(struct mlxsw_sp_fib6_entry *fib6_entry)
+{
+	struct mlxsw_sp_rt6 *mlxsw_sp_rt6, *tmp;
+
+	list_for_each_entry_safe(mlxsw_sp_rt6, tmp, &fib6_entry->rt6_list,
+				 list) {
+		fib6_entry->nrt6--;
+		list_del(&mlxsw_sp_rt6->list);
+		mlxsw_sp_rt6_destroy(mlxsw_sp_rt6);
+	}
+}
+
+static struct mlxsw_sp_fib6_entry *
+mlxsw_sp_fib6_entry_create(struct mlxsw_sp *mlxsw_sp,
+			   struct mlxsw_sp_fib_node *fib_node,
+			   struct fib6_info **rt_arr, unsigned int nrt6)
+{
+	struct mlxsw_sp_fib6_entry *fib6_entry;
+	struct mlxsw_sp_fib_entry *fib_entry;
+	struct mlxsw_sp_rt6 *mlxsw_sp_rt6;
+	int err, i;
+
+	fib6_entry = kzalloc(sizeof(*fib6_entry), GFP_KERNEL);
+	if (!fib6_entry)
+		return ERR_PTR(-ENOMEM);
+	fib_entry = &fib6_entry->common;
+
+	INIT_LIST_HEAD(&fib6_entry->rt6_list);
+
+	for (i = 0; i < nrt6; i++) {
+		mlxsw_sp_rt6 = mlxsw_sp_rt6_create(rt_arr[i]);
+		if (IS_ERR(mlxsw_sp_rt6)) {
+			err = PTR_ERR(mlxsw_sp_rt6);
+			goto err_rt6_unwind;
+		}
+		list_add_tail(&mlxsw_sp_rt6->list, &fib6_entry->rt6_list);
+		fib6_entry->nrt6++;
+	}
+
+	err = mlxsw_sp_nexthop6_group_get(mlxsw_sp, fib6_entry);
+	if (err)
+		goto err_rt6_unwind;
+
+	err = mlxsw_sp_nexthop_group_vr_link(fib_entry->nh_group,
+					     fib_node->fib);
+	if (err)
+		goto err_nexthop_group_vr_link;
+
+	err = mlxsw_sp_fib6_entry_type_set(mlxsw_sp, fib_entry, rt_arr[0]);
+	if (err)
+		goto err_fib6_entry_type_set;
+
+	fib_entry->fib_node = fib_node;
+
+	return fib6_entry;
+
+err_fib6_entry_type_set:
+	mlxsw_sp_nexthop_group_vr_unlink(fib_entry->nh_group, fib_node->fib);
+err_nexthop_group_vr_link:
+	mlxsw_sp_nexthop6_group_put(mlxsw_sp, fib_entry);
+err_rt6_unwind:
+	for (; i > 0; i--) {
+		fib6_entry->nrt6--;
+		mlxsw_sp_rt6 = list_last_entry(&fib6_entry->rt6_list,
+					       struct mlxsw_sp_rt6, list);
+		list_del(&mlxsw_sp_rt6->list);
+		mlxsw_sp_rt6_destroy(mlxsw_sp_rt6);
+	}
+	kfree(fib6_entry);
+	return ERR_PTR(err);
+}
+
+static void
+mlxsw_sp_fib6_entry_type_unset(struct mlxsw_sp *mlxsw_sp,
+			       struct mlxsw_sp_fib6_entry *fib6_entry)
+{
+	mlxsw_sp_fib_entry_type_unset(mlxsw_sp, &fib6_entry->common);
+}
+
+static void mlxsw_sp_fib6_entry_destroy(struct mlxsw_sp *mlxsw_sp,
+					struct mlxsw_sp_fib6_entry *fib6_entry)
+{
+	struct mlxsw_sp_fib_node *fib_node = fib6_entry->common.fib_node;
+
+	mlxsw_sp_fib6_entry_type_unset(mlxsw_sp, fib6_entry);
+	mlxsw_sp_nexthop_group_vr_unlink(fib6_entry->common.nh_group,
+					 fib_node->fib);
+	mlxsw_sp_nexthop6_group_put(mlxsw_sp, &fib6_entry->common);
+	mlxsw_sp_fib6_entry_rt_destroy_all(fib6_entry);
+	WARN_ON(fib6_entry->nrt6);
+	kfree(fib6_entry);
+}
+
+static struct mlxsw_sp_fib6_entry *
+mlxsw_sp_fib6_entry_lookup(struct mlxsw_sp *mlxsw_sp,
+			   const struct fib6_info *rt)
+{
+	struct mlxsw_sp_fib6_entry *fib6_entry;
+	struct mlxsw_sp_fib_node *fib_node;
+	struct mlxsw_sp_fib *fib;
+	struct fib6_info *cmp_rt;
+	struct mlxsw_sp_vr *vr;
+
+	vr = mlxsw_sp_vr_find(mlxsw_sp, rt->fib6_table->tb6_id);
+	if (!vr)
+		return NULL;
+	fib = mlxsw_sp_vr_fib(vr, MLXSW_SP_L3_PROTO_IPV6);
+
+	fib_node = mlxsw_sp_fib_node_lookup(fib, &rt->fib6_dst.addr,
+					    sizeof(rt->fib6_dst.addr),
+					    rt->fib6_dst.plen);
+	if (!fib_node)
+		return NULL;
+
+	fib6_entry = container_of(fib_node->fib_entry,
+				  struct mlxsw_sp_fib6_entry, common);
+	cmp_rt = mlxsw_sp_fib6_entry_rt(fib6_entry);
+	if (rt->fib6_table->tb6_id == cmp_rt->fib6_table->tb6_id &&
+	    rt->fib6_metric == cmp_rt->fib6_metric &&
+	    mlxsw_sp_fib6_entry_rt_find(fib6_entry, rt))
+		return fib6_entry;
+
+	return NULL;
+}
+
+static bool mlxsw_sp_fib6_allow_replace(struct mlxsw_sp_fib6_entry *fib6_entry)
+{
+	struct mlxsw_sp_fib_node *fib_node = fib6_entry->common.fib_node;
+	struct mlxsw_sp_fib6_entry *fib6_replaced;
+	struct fib6_info *rt, *rt_replaced;
+
+	if (!fib_node->fib_entry)
+		return true;
+
+	fib6_replaced = container_of(fib_node->fib_entry,
+				     struct mlxsw_sp_fib6_entry,
+				     common);
+	rt = mlxsw_sp_fib6_entry_rt(fib6_entry);
+	rt_replaced = mlxsw_sp_fib6_entry_rt(fib6_replaced);
+	if (rt->fib6_table->tb6_id == RT_TABLE_MAIN &&
+	    rt_replaced->fib6_table->tb6_id == RT_TABLE_LOCAL)
+		return false;
+
+	return true;
+}
+
+static int mlxsw_sp_router_fib6_replace(struct mlxsw_sp *mlxsw_sp,
+					struct fib6_info **rt_arr,
+					unsigned int nrt6)
+{
+	struct mlxsw_sp_fib6_entry *fib6_entry, *fib6_replaced;
+	struct mlxsw_sp_fib_entry *replaced;
+	struct mlxsw_sp_fib_node *fib_node;
+	struct fib6_info *rt = rt_arr[0];
+	int err;
+
+	if (rt->fib6_src.plen)
+		return -EINVAL;
+
+	if (mlxsw_sp_fib6_rt_should_ignore(rt))
+		return 0;
+
+	if (rt->nh && !mlxsw_sp_nexthop_obj_group_lookup(mlxsw_sp, rt->nh->id))
+		return 0;
+
+	fib_node = mlxsw_sp_fib_node_get(mlxsw_sp, rt->fib6_table->tb6_id,
+					 &rt->fib6_dst.addr,
+					 sizeof(rt->fib6_dst.addr),
+					 rt->fib6_dst.plen,
+					 MLXSW_SP_L3_PROTO_IPV6);
+	if (IS_ERR(fib_node))
+		return PTR_ERR(fib_node);
+
+	fib6_entry = mlxsw_sp_fib6_entry_create(mlxsw_sp, fib_node, rt_arr,
+						nrt6);
+	if (IS_ERR(fib6_entry)) {
+		err = PTR_ERR(fib6_entry);
+		goto err_fib6_entry_create;
+	}
+
+	if (!mlxsw_sp_fib6_allow_replace(fib6_entry)) {
+		mlxsw_sp_fib6_entry_destroy(mlxsw_sp, fib6_entry);
+		mlxsw_sp_fib_node_put(mlxsw_sp, fib_node);
+		return 0;
+	}
+
+	replaced = fib_node->fib_entry;
+	err = mlxsw_sp_fib_node_entry_link(mlxsw_sp, &fib6_entry->common);
+	if (err)
+		goto err_fib_node_entry_link;
+
+	/* Nothing to replace */
+	if (!replaced)
+		return 0;
+
+	mlxsw_sp_fib_entry_hw_flags_clear(mlxsw_sp, replaced);
+	fib6_replaced = container_of(replaced, struct mlxsw_sp_fib6_entry,
+				     common);
+	mlxsw_sp_fib6_entry_destroy(mlxsw_sp, fib6_replaced);
+
+	return 0;
+
+err_fib_node_entry_link:
+	fib_node->fib_entry = replaced;
+	mlxsw_sp_fib6_entry_destroy(mlxsw_sp, fib6_entry);
+err_fib6_entry_create:
+	mlxsw_sp_fib_node_put(mlxsw_sp, fib_node);
+	return err;
+}
+
+static int mlxsw_sp_router_fib6_append(struct mlxsw_sp *mlxsw_sp,
+				       struct fib6_info **rt_arr,
+				       unsigned int nrt6)
+{
+	struct mlxsw_sp_fib6_entry *fib6_entry;
+	struct mlxsw_sp_fib_node *fib_node;
+	struct fib6_info *rt = rt_arr[0];
+	int err;
+
+	if (rt->fib6_src.plen)
+		return -EINVAL;
+
+	if (mlxsw_sp_fib6_rt_should_ignore(rt))
+		return 0;
+
+	fib_node = mlxsw_sp_fib_node_get(mlxsw_sp, rt->fib6_table->tb6_id,
+					 &rt->fib6_dst.addr,
+					 sizeof(rt->fib6_dst.addr),
+					 rt->fib6_dst.plen,
+					 MLXSW_SP_L3_PROTO_IPV6);
+	if (IS_ERR(fib_node))
+		return PTR_ERR(fib_node);
+
+	if (WARN_ON_ONCE(!fib_node->fib_entry)) {
+		mlxsw_sp_fib_node_put(mlxsw_sp, fib_node);
+		return -EINVAL;
+	}
+
+	fib6_entry = container_of(fib_node->fib_entry,
+				  struct mlxsw_sp_fib6_entry, common);
+	err = mlxsw_sp_fib6_entry_nexthop_add(mlxsw_sp, fib6_entry, rt_arr,
+					      nrt6);
+	if (err)
+		goto err_fib6_entry_nexthop_add;
+
+	return 0;
+
+err_fib6_entry_nexthop_add:
+	mlxsw_sp_fib_node_put(mlxsw_sp, fib_node);
+	return err;
+}
+
+static void mlxsw_sp_router_fib6_del(struct mlxsw_sp *mlxsw_sp,
+				     struct fib6_info **rt_arr,
+				     unsigned int nrt6)
+{
+	struct mlxsw_sp_fib6_entry *fib6_entry;
+	struct mlxsw_sp_fib_node *fib_node;
+	struct fib6_info *rt = rt_arr[0];
+
+	if (mlxsw_sp_fib6_rt_should_ignore(rt))
+		return;
+
+	/* Multipath routes are first added to the FIB trie and only then
+	 * notified. If we vetoed the addition, we will get a delete
+	 * notification for a route we do not have. Therefore, do not warn if
+	 * route was not found.
+	 */
+	fib6_entry = mlxsw_sp_fib6_entry_lookup(mlxsw_sp, rt);
+	if (!fib6_entry)
+		return;
+
+	/* If not all the nexthops are deleted, then only reduce the nexthop
+	 * group.
+	 */
+	if (nrt6 != fib6_entry->nrt6) {
+		mlxsw_sp_fib6_entry_nexthop_del(mlxsw_sp, fib6_entry, rt_arr,
+						nrt6);
+		return;
+	}
+
+	fib_node = fib6_entry->common.fib_node;
+
+	mlxsw_sp_fib_node_entry_unlink(mlxsw_sp, &fib6_entry->common);
+	mlxsw_sp_fib6_entry_destroy(mlxsw_sp, fib6_entry);
+	mlxsw_sp_fib_node_put(mlxsw_sp, fib_node);
+}
+
+static struct mlxsw_sp_mr_table *
+mlxsw_sp_router_fibmr_family_to_table(struct mlxsw_sp_vr *vr, int family)
+{
+	if (family == RTNL_FAMILY_IPMR)
+		return vr->mr_table[MLXSW_SP_L3_PROTO_IPV4];
+	else
+		return vr->mr_table[MLXSW_SP_L3_PROTO_IPV6];
+}
+
+static int mlxsw_sp_router_fibmr_add(struct mlxsw_sp *mlxsw_sp,
+				     struct mfc_entry_notifier_info *men_info,
+				     bool replace)
+{
+	struct mlxsw_sp_mr_table *mrt;
+	struct mlxsw_sp_vr *vr;
+
+	vr = mlxsw_sp_vr_get(mlxsw_sp, men_info->tb_id, NULL);
+	if (IS_ERR(vr))
+		return PTR_ERR(vr);
+
+	mrt = mlxsw_sp_router_fibmr_family_to_table(vr, men_info->info.family);
+	return mlxsw_sp_mr_route_add(mrt, men_info->mfc, replace);
+}
+
+static void mlxsw_sp_router_fibmr_del(struct mlxsw_sp *mlxsw_sp,
+				      struct mfc_entry_notifier_info *men_info)
+{
+	struct mlxsw_sp_mr_table *mrt;
+	struct mlxsw_sp_vr *vr;
+
+	vr = mlxsw_sp_vr_find(mlxsw_sp, men_info->tb_id);
+	if (WARN_ON(!vr))
+		return;
+
+	mrt = mlxsw_sp_router_fibmr_family_to_table(vr, men_info->info.family);
+	mlxsw_sp_mr_route_del(mrt, men_info->mfc);
+	mlxsw_sp_vr_put(mlxsw_sp, vr);
+}
+
+static int
+mlxsw_sp_router_fibmr_vif_add(struct mlxsw_sp *mlxsw_sp,
+			      struct vif_entry_notifier_info *ven_info)
+{
+	struct mlxsw_sp_mr_table *mrt;
+	struct mlxsw_sp_rif *rif;
+	struct mlxsw_sp_vr *vr;
+
+	vr = mlxsw_sp_vr_get(mlxsw_sp, ven_info->tb_id, NULL);
+	if (IS_ERR(vr))
+		return PTR_ERR(vr);
+
+	mrt = mlxsw_sp_router_fibmr_family_to_table(vr, ven_info->info.family);
+	rif = mlxsw_sp_rif_find_by_dev(mlxsw_sp, ven_info->dev);
+	return mlxsw_sp_mr_vif_add(mrt, ven_info->dev,
+				   ven_info->vif_index,
+				   ven_info->vif_flags, rif);
+}
+
+static void
+mlxsw_sp_router_fibmr_vif_del(struct mlxsw_sp *mlxsw_sp,
+			      struct vif_entry_notifier_info *ven_info)
+{
+	struct mlxsw_sp_mr_table *mrt;
+	struct mlxsw_sp_vr *vr;
+
+	vr = mlxsw_sp_vr_find(mlxsw_sp, ven_info->tb_id);
+	if (WARN_ON(!vr))
+		return;
+
+	mrt = mlxsw_sp_router_fibmr_family_to_table(vr, ven_info->info.family);
+	mlxsw_sp_mr_vif_del(mrt, ven_info->vif_index);
+	mlxsw_sp_vr_put(mlxsw_sp, vr);
+}
+
+static void mlxsw_sp_fib4_node_flush(struct mlxsw_sp *mlxsw_sp,
+				     struct mlxsw_sp_fib_node *fib_node)
+{
+	struct mlxsw_sp_fib4_entry *fib4_entry;
+
+	fib4_entry = container_of(fib_node->fib_entry,
+				  struct mlxsw_sp_fib4_entry, common);
+	mlxsw_sp_fib_node_entry_unlink(mlxsw_sp, fib_node->fib_entry);
+	mlxsw_sp_fib4_entry_destroy(mlxsw_sp, fib4_entry);
+	mlxsw_sp_fib_node_put(mlxsw_sp, fib_node);
+}
+
+static void mlxsw_sp_fib6_node_flush(struct mlxsw_sp *mlxsw_sp,
+				     struct mlxsw_sp_fib_node *fib_node)
+{
+	struct mlxsw_sp_fib6_entry *fib6_entry;
+
+	fib6_entry = container_of(fib_node->fib_entry,
+				  struct mlxsw_sp_fib6_entry, common);
+	mlxsw_sp_fib_node_entry_unlink(mlxsw_sp, fib_node->fib_entry);
+	mlxsw_sp_fib6_entry_destroy(mlxsw_sp, fib6_entry);
+	mlxsw_sp_fib_node_put(mlxsw_sp, fib_node);
+}
+
+static void mlxsw_sp_fib_node_flush(struct mlxsw_sp *mlxsw_sp,
+				    struct mlxsw_sp_fib_node *fib_node)
+{
+	switch (fib_node->fib->proto) {
+	case MLXSW_SP_L3_PROTO_IPV4:
+		mlxsw_sp_fib4_node_flush(mlxsw_sp, fib_node);
+		break;
+	case MLXSW_SP_L3_PROTO_IPV6:
+		mlxsw_sp_fib6_node_flush(mlxsw_sp, fib_node);
+		break;
+	}
+}
+
+static void mlxsw_sp_vr_fib_flush(struct mlxsw_sp *mlxsw_sp,
+				  struct mlxsw_sp_vr *vr,
+				  enum mlxsw_sp_l3proto proto)
+{
+	struct mlxsw_sp_fib *fib = mlxsw_sp_vr_fib(vr, proto);
+	struct mlxsw_sp_fib_node *fib_node, *tmp;
+
+	list_for_each_entry_safe(fib_node, tmp, &fib->node_list, list) {
+		bool do_break = &tmp->list == &fib->node_list;
+
+		mlxsw_sp_fib_node_flush(mlxsw_sp, fib_node);
+		if (do_break)
+			break;
+	}
+}
+
+static void mlxsw_sp_router_fib_flush(struct mlxsw_sp *mlxsw_sp)
+{
+	int i, j;
+
+	for (i = 0; i < MLXSW_CORE_RES_GET(mlxsw_sp->core, MAX_VRS); i++) {
+		struct mlxsw_sp_vr *vr = &mlxsw_sp->router->vrs[i];
+
+		if (!mlxsw_sp_vr_is_used(vr))
+			continue;
+
+		for (j = 0; j < MLXSW_SP_L3_PROTO_MAX; j++)
+			mlxsw_sp_mr_table_flush(vr->mr_table[j]);
+		mlxsw_sp_vr_fib_flush(mlxsw_sp, vr, MLXSW_SP_L3_PROTO_IPV4);
+
+		/* If virtual router was only used for IPv4, then it's no
+		 * longer used.
+		 */
+		if (!mlxsw_sp_vr_is_used(vr))
+			continue;
+		mlxsw_sp_vr_fib_flush(mlxsw_sp, vr, MLXSW_SP_L3_PROTO_IPV6);
+	}
+}
+
+struct mlxsw_sp_fib6_event_work {
+	struct fib6_info **rt_arr;
+	unsigned int nrt6;
+};
+
+struct mlxsw_sp_fib_event_work {
+	struct work_struct work;
+	union {
+		struct mlxsw_sp_fib6_event_work fib6_work;
+		struct fib_entry_notifier_info fen_info;
+		struct fib_rule_notifier_info fr_info;
+		struct fib_nh_notifier_info fnh_info;
+		struct mfc_entry_notifier_info men_info;
+		struct vif_entry_notifier_info ven_info;
+	};
+	struct mlxsw_sp *mlxsw_sp;
+	unsigned long event;
+};
+
+static int
+mlxsw_sp_router_fib6_work_init(struct mlxsw_sp_fib6_event_work *fib6_work,
+			       struct fib6_entry_notifier_info *fen6_info)
+{
+	struct fib6_info *rt = fen6_info->rt;
+	struct fib6_info **rt_arr;
+	struct fib6_info *iter;
+	unsigned int nrt6;
+	int i = 0;
+
+	nrt6 = fen6_info->nsiblings + 1;
+
+	rt_arr = kcalloc(nrt6, sizeof(struct fib6_info *), GFP_ATOMIC);
+	if (!rt_arr)
+		return -ENOMEM;
+
+	fib6_work->rt_arr = rt_arr;
+	fib6_work->nrt6 = nrt6;
+
+	rt_arr[0] = rt;
+	fib6_info_hold(rt);
+
+	if (!fen6_info->nsiblings)
+		return 0;
+
+	list_for_each_entry(iter, &rt->fib6_siblings, fib6_siblings) {
+		if (i == fen6_info->nsiblings)
+			break;
+
+		rt_arr[i + 1] = iter;
+		fib6_info_hold(iter);
+		i++;
+	}
+	WARN_ON_ONCE(i != fen6_info->nsiblings);
+
+	return 0;
+}
+
+static void
+mlxsw_sp_router_fib6_work_fini(struct mlxsw_sp_fib6_event_work *fib6_work)
+{
+	int i;
+
+	for (i = 0; i < fib6_work->nrt6; i++)
+		mlxsw_sp_rt6_release(fib6_work->rt_arr[i]);
+	kfree(fib6_work->rt_arr);
+}
+
+static void mlxsw_sp_router_fib4_event_work(struct work_struct *work)
+{
+	struct mlxsw_sp_fib_event_work *fib_work =
+		container_of(work, struct mlxsw_sp_fib_event_work, work);
+	struct mlxsw_sp *mlxsw_sp = fib_work->mlxsw_sp;
+	int err;
+
+	mutex_lock(&mlxsw_sp->router->lock);
+	mlxsw_sp_span_respin(mlxsw_sp);
+
+	switch (fib_work->event) {
+	case FIB_EVENT_ENTRY_REPLACE:
+		err = mlxsw_sp_router_fib4_replace(mlxsw_sp,
+						   &fib_work->fen_info);
+		if (err) {
+			dev_warn(mlxsw_sp->bus_info->dev, "FIB replace failed.\n");
+			mlxsw_sp_fib4_offload_failed_flag_set(mlxsw_sp,
+							      &fib_work->fen_info);
+		}
+		fib_info_put(fib_work->fen_info.fi);
+		break;
+	case FIB_EVENT_ENTRY_DEL:
+		mlxsw_sp_router_fib4_del(mlxsw_sp, &fib_work->fen_info);
+		fib_info_put(fib_work->fen_info.fi);
+		break;
+	case FIB_EVENT_NH_ADD:
+	case FIB_EVENT_NH_DEL:
+		mlxsw_sp_nexthop4_event(mlxsw_sp, fib_work->event,
+					fib_work->fnh_info.fib_nh);
+		fib_info_put(fib_work->fnh_info.fib_nh->nh_parent);
+		break;
+	}
+	mutex_unlock(&mlxsw_sp->router->lock);
+	kfree(fib_work);
+}
+
+static void mlxsw_sp_router_fib6_event_work(struct work_struct *work)
+{
+	struct mlxsw_sp_fib_event_work *fib_work =
+		    container_of(work, struct mlxsw_sp_fib_event_work, work);
+	struct mlxsw_sp_fib6_event_work *fib6_work = &fib_work->fib6_work;
+	struct mlxsw_sp *mlxsw_sp = fib_work->mlxsw_sp;
+	int err;
+
+	mutex_lock(&mlxsw_sp->router->lock);
+	mlxsw_sp_span_respin(mlxsw_sp);
+
+	switch (fib_work->event) {
+	case FIB_EVENT_ENTRY_REPLACE:
+		err = mlxsw_sp_router_fib6_replace(mlxsw_sp,
+						   fib6_work->rt_arr,
+						   fib6_work->nrt6);
+		if (err) {
+			dev_warn(mlxsw_sp->bus_info->dev, "FIB replace failed.\n");
+			mlxsw_sp_fib6_offload_failed_flag_set(mlxsw_sp,
+							      fib6_work->rt_arr,
+							      fib6_work->nrt6);
+		}
+		mlxsw_sp_router_fib6_work_fini(fib6_work);
+		break;
+	case FIB_EVENT_ENTRY_APPEND:
+		err = mlxsw_sp_router_fib6_append(mlxsw_sp,
+						  fib6_work->rt_arr,
+						  fib6_work->nrt6);
+		if (err) {
+			dev_warn(mlxsw_sp->bus_info->dev, "FIB append failed.\n");
+			mlxsw_sp_fib6_offload_failed_flag_set(mlxsw_sp,
+							      fib6_work->rt_arr,
+							      fib6_work->nrt6);
+		}
+		mlxsw_sp_router_fib6_work_fini(fib6_work);
+		break;
+	case FIB_EVENT_ENTRY_DEL:
+		mlxsw_sp_router_fib6_del(mlxsw_sp,
+					 fib6_work->rt_arr,
+					 fib6_work->nrt6);
+		mlxsw_sp_router_fib6_work_fini(fib6_work);
+		break;
+	}
+	mutex_unlock(&mlxsw_sp->router->lock);
+	kfree(fib_work);
+}
+
+static void mlxsw_sp_router_fibmr_event_work(struct work_struct *work)
+{
+	struct mlxsw_sp_fib_event_work *fib_work =
+		container_of(work, struct mlxsw_sp_fib_event_work, work);
+	struct mlxsw_sp *mlxsw_sp = fib_work->mlxsw_sp;
+	bool replace;
+	int err;
+
+	rtnl_lock();
+	mutex_lock(&mlxsw_sp->router->lock);
+	switch (fib_work->event) {
+	case FIB_EVENT_ENTRY_REPLACE:
+	case FIB_EVENT_ENTRY_ADD:
+		replace = fib_work->event == FIB_EVENT_ENTRY_REPLACE;
+
+		err = mlxsw_sp_router_fibmr_add(mlxsw_sp, &fib_work->men_info,
+						replace);
+		if (err)
+			dev_warn(mlxsw_sp->bus_info->dev, "MR entry add failed.\n");
+		mr_cache_put(fib_work->men_info.mfc);
+		break;
+	case FIB_EVENT_ENTRY_DEL:
+		mlxsw_sp_router_fibmr_del(mlxsw_sp, &fib_work->men_info);
+		mr_cache_put(fib_work->men_info.mfc);
+		break;
+	case FIB_EVENT_VIF_ADD:
+		err = mlxsw_sp_router_fibmr_vif_add(mlxsw_sp,
+						    &fib_work->ven_info);
+		if (err)
+			dev_warn(mlxsw_sp->bus_info->dev, "MR VIF add failed.\n");
+		dev_put(fib_work->ven_info.dev);
+		break;
+	case FIB_EVENT_VIF_DEL:
+		mlxsw_sp_router_fibmr_vif_del(mlxsw_sp,
+					      &fib_work->ven_info);
+		dev_put(fib_work->ven_info.dev);
+		break;
+	}
+	mutex_unlock(&mlxsw_sp->router->lock);
+	rtnl_unlock();
+	kfree(fib_work);
+}
+
+static void mlxsw_sp_router_fib4_event(struct mlxsw_sp_fib_event_work *fib_work,
+				       struct fib_notifier_info *info)
+{
+	struct fib_entry_notifier_info *fen_info;
+	struct fib_nh_notifier_info *fnh_info;
+
+	switch (fib_work->event) {
+	case FIB_EVENT_ENTRY_REPLACE:
+	case FIB_EVENT_ENTRY_DEL:
+		fen_info = container_of(info, struct fib_entry_notifier_info,
+					info);
+		fib_work->fen_info = *fen_info;
+		/* Take reference on fib_info to prevent it from being
+		 * freed while work is queued. Release it afterwards.
+		 */
+		fib_info_hold(fib_work->fen_info.fi);
+		break;
+	case FIB_EVENT_NH_ADD:
+	case FIB_EVENT_NH_DEL:
+		fnh_info = container_of(info, struct fib_nh_notifier_info,
+					info);
+		fib_work->fnh_info = *fnh_info;
+		fib_info_hold(fib_work->fnh_info.fib_nh->nh_parent);
+		break;
+	}
+}
+
+static int mlxsw_sp_router_fib6_event(struct mlxsw_sp_fib_event_work *fib_work,
+				      struct fib_notifier_info *info)
+{
+	struct fib6_entry_notifier_info *fen6_info;
+	int err;
+
+	switch (fib_work->event) {
+	case FIB_EVENT_ENTRY_REPLACE:
+	case FIB_EVENT_ENTRY_APPEND:
+	case FIB_EVENT_ENTRY_DEL:
+		fen6_info = container_of(info, struct fib6_entry_notifier_info,
+					 info);
+		err = mlxsw_sp_router_fib6_work_init(&fib_work->fib6_work,
+						     fen6_info);
+		if (err)
+			return err;
+		break;
+	}
+
+	return 0;
+}
+
+static void
+mlxsw_sp_router_fibmr_event(struct mlxsw_sp_fib_event_work *fib_work,
+			    struct fib_notifier_info *info)
+{
+	switch (fib_work->event) {
+	case FIB_EVENT_ENTRY_REPLACE:
+	case FIB_EVENT_ENTRY_ADD:
+	case FIB_EVENT_ENTRY_DEL:
+		memcpy(&fib_work->men_info, info, sizeof(fib_work->men_info));
+		mr_cache_hold(fib_work->men_info.mfc);
+		break;
+	case FIB_EVENT_VIF_ADD:
+	case FIB_EVENT_VIF_DEL:
+		memcpy(&fib_work->ven_info, info, sizeof(fib_work->ven_info));
+		dev_hold(fib_work->ven_info.dev);
+		break;
+	}
+}
+
+static int mlxsw_sp_router_fib_rule_event(unsigned long event,
+					  struct fib_notifier_info *info,
+					  struct mlxsw_sp *mlxsw_sp)
+{
+	struct netlink_ext_ack *extack = info->extack;
+	struct fib_rule_notifier_info *fr_info;
+	struct fib_rule *rule;
+	int err = 0;
+
+	/* nothing to do at the moment */
+	if (event == FIB_EVENT_RULE_DEL)
+		return 0;
+
+	fr_info = container_of(info, struct fib_rule_notifier_info, info);
+	rule = fr_info->rule;
+
+	/* Rule only affects locally generated traffic */
+	if (rule->iifindex == mlxsw_sp_net(mlxsw_sp)->loopback_dev->ifindex)
+		return 0;
+
+	switch (info->family) {
+	case AF_INET:
+		if (!fib4_rule_default(rule) && !rule->l3mdev)
+			err = -EOPNOTSUPP;
+		break;
+	case AF_INET6:
+		if (!fib6_rule_default(rule) && !rule->l3mdev)
+			err = -EOPNOTSUPP;
+		break;
+	case RTNL_FAMILY_IPMR:
+		if (!ipmr_rule_default(rule) && !rule->l3mdev)
+			err = -EOPNOTSUPP;
+		break;
+	case RTNL_FAMILY_IP6MR:
+		if (!ip6mr_rule_default(rule) && !rule->l3mdev)
+			err = -EOPNOTSUPP;
+		break;
+	}
+
+	if (err < 0)
+		NL_SET_ERR_MSG_MOD(extack, "FIB rules not supported");
+
+	return err;
+}
+
+/* Called with rcu_read_lock() */
+static int mlxsw_sp_router_fib_event(struct notifier_block *nb,
+				     unsigned long event, void *ptr)
+{
+	struct mlxsw_sp_fib_event_work *fib_work;
+	struct fib_notifier_info *info = ptr;
+	struct mlxsw_sp_router *router;
+	int err;
+
+	if ((info->family != AF_INET && info->family != AF_INET6 &&
+	     info->family != RTNL_FAMILY_IPMR &&
+	     info->family != RTNL_FAMILY_IP6MR))
+		return NOTIFY_DONE;
+
+	router = container_of(nb, struct mlxsw_sp_router, fib_nb);
+
+	switch (event) {
+	case FIB_EVENT_RULE_ADD:
+	case FIB_EVENT_RULE_DEL:
+		err = mlxsw_sp_router_fib_rule_event(event, info,
+						     router->mlxsw_sp);
+		return notifier_from_errno(err);
+	case FIB_EVENT_ENTRY_ADD:
+	case FIB_EVENT_ENTRY_REPLACE:
+	case FIB_EVENT_ENTRY_APPEND:
+		if (info->family == AF_INET) {
+			struct fib_entry_notifier_info *fen_info = ptr;
+
+			if (fen_info->fi->fib_nh_is_v6) {
+				NL_SET_ERR_MSG_MOD(info->extack, "IPv6 gateway with IPv4 route is not supported");
+				return notifier_from_errno(-EINVAL);
+			}
+		}
+		break;
+	}
+
+	fib_work = kzalloc(sizeof(*fib_work), GFP_ATOMIC);
+	if (!fib_work)
+		return NOTIFY_BAD;
+
+	fib_work->mlxsw_sp = router->mlxsw_sp;
+	fib_work->event = event;
+
+	switch (info->family) {
+	case AF_INET:
+		INIT_WORK(&fib_work->work, mlxsw_sp_router_fib4_event_work);
+		mlxsw_sp_router_fib4_event(fib_work, info);
+		break;
+	case AF_INET6:
+		INIT_WORK(&fib_work->work, mlxsw_sp_router_fib6_event_work);
+		err = mlxsw_sp_router_fib6_event(fib_work, info);
+		if (err)
+			goto err_fib_event;
+		break;
+	case RTNL_FAMILY_IP6MR:
+	case RTNL_FAMILY_IPMR:
+		INIT_WORK(&fib_work->work, mlxsw_sp_router_fibmr_event_work);
+		mlxsw_sp_router_fibmr_event(fib_work, info);
+		break;
+	}
+
+	mlxsw_core_schedule_work(&fib_work->work);
+
+	return NOTIFY_DONE;
+
+err_fib_event:
+	kfree(fib_work);
+	return NOTIFY_BAD;
+}
+
+static struct mlxsw_sp_rif *
+mlxsw_sp_rif_find_by_dev(const struct mlxsw_sp *mlxsw_sp,
+			 const struct net_device *dev)
+{
+	int i;
+
+	for (i = 0; i < MLXSW_CORE_RES_GET(mlxsw_sp->core, MAX_RIFS); i++)
+		if (mlxsw_sp->router->rifs[i] &&
+		    mlxsw_sp->router->rifs[i]->dev == dev)
+			return mlxsw_sp->router->rifs[i];
+
+	return NULL;
+}
+
+bool mlxsw_sp_rif_exists(struct mlxsw_sp *mlxsw_sp,
+			 const struct net_device *dev)
+{
+	struct mlxsw_sp_rif *rif;
+
+	mutex_lock(&mlxsw_sp->router->lock);
+	rif = mlxsw_sp_rif_find_by_dev(mlxsw_sp, dev);
+	mutex_unlock(&mlxsw_sp->router->lock);
+
+	return rif;
+}
+
+u16 mlxsw_sp_rif_vid(struct mlxsw_sp *mlxsw_sp, const struct net_device *dev)
+{
+	struct mlxsw_sp_rif *rif;
+	u16 vid = 0;
+
+	mutex_lock(&mlxsw_sp->router->lock);
+	rif = mlxsw_sp_rif_find_by_dev(mlxsw_sp, dev);
+	if (!rif)
+		goto out;
+
+	/* We only return the VID for VLAN RIFs. Otherwise we return an
+	 * invalid value (0).
+	 */
+	if (rif->ops->type != MLXSW_SP_RIF_TYPE_VLAN)
+		goto out;
+
+	vid = mlxsw_sp_fid_8021q_vid(rif->fid);
+
+out:
+	mutex_unlock(&mlxsw_sp->router->lock);
+	return vid;
+}
+
+static int mlxsw_sp_router_rif_disable(struct mlxsw_sp *mlxsw_sp, u16 rif)
+{
+	char ritr_pl[MLXSW_REG_RITR_LEN];
+	int err;
+
+	mlxsw_reg_ritr_rif_pack(ritr_pl, rif);
+	err = mlxsw_reg_query(mlxsw_sp->core, MLXSW_REG(ritr), ritr_pl);
+	if (err)
+		return err;
+
+	mlxsw_reg_ritr_enable_set(ritr_pl, false);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(ritr), ritr_pl);
+}
+
+static void mlxsw_sp_router_rif_gone_sync(struct mlxsw_sp *mlxsw_sp,
+					  struct mlxsw_sp_rif *rif)
+{
+	mlxsw_sp_router_rif_disable(mlxsw_sp, rif->rif_index);
+	mlxsw_sp_nexthop_rif_gone_sync(mlxsw_sp, rif);
+	mlxsw_sp_neigh_rif_gone_sync(mlxsw_sp, rif);
+}
+
+static bool
+mlxsw_sp_rif_should_config(struct mlxsw_sp_rif *rif, struct net_device *dev,
+			   unsigned long event)
+{
+	struct inet6_dev *inet6_dev;
+	bool addr_list_empty = true;
+	struct in_device *idev;
+
+	switch (event) {
+	case NETDEV_UP:
+		return rif == NULL;
+	case NETDEV_DOWN:
+		rcu_read_lock();
+		idev = __in_dev_get_rcu(dev);
+		if (idev && idev->ifa_list)
+			addr_list_empty = false;
+
+		inet6_dev = __in6_dev_get(dev);
+		if (addr_list_empty && inet6_dev &&
+		    !list_empty(&inet6_dev->addr_list))
+			addr_list_empty = false;
+		rcu_read_unlock();
+
+		/* macvlans do not have a RIF, but rather piggy back on the
+		 * RIF of their lower device.
+		 */
+		if (netif_is_macvlan(dev) && addr_list_empty)
+			return true;
+
+		if (rif && addr_list_empty &&
+		    !netif_is_l3_slave(rif->dev))
+			return true;
+		/* It is possible we already removed the RIF ourselves
+		 * if it was assigned to a netdev that is now a bridge
+		 * or LAG slave.
+		 */
+		return false;
+	}
+
+	return false;
+}
+
+static enum mlxsw_sp_rif_type
+mlxsw_sp_dev_rif_type(const struct mlxsw_sp *mlxsw_sp,
+		      const struct net_device *dev)
+{
+	enum mlxsw_sp_fid_type type;
+
+	if (mlxsw_sp_netdev_ipip_type(mlxsw_sp, dev, NULL))
+		return MLXSW_SP_RIF_TYPE_IPIP_LB;
+
+	/* Otherwise RIF type is derived from the type of the underlying FID. */
+	if (is_vlan_dev(dev) && netif_is_bridge_master(vlan_dev_real_dev(dev)))
+		type = MLXSW_SP_FID_TYPE_8021Q;
+	else if (netif_is_bridge_master(dev) && br_vlan_enabled(dev))
+		type = MLXSW_SP_FID_TYPE_8021Q;
+	else if (netif_is_bridge_master(dev))
+		type = MLXSW_SP_FID_TYPE_8021D;
+	else
+		type = MLXSW_SP_FID_TYPE_RFID;
+
+	return mlxsw_sp_fid_type_rif_type(mlxsw_sp, type);
+}
+
+static int mlxsw_sp_rif_index_alloc(struct mlxsw_sp *mlxsw_sp, u16 *p_rif_index)
+{
+	int i;
+
+	for (i = 0; i < MLXSW_CORE_RES_GET(mlxsw_sp->core, MAX_RIFS); i++) {
+		if (!mlxsw_sp->router->rifs[i]) {
+			*p_rif_index = i;
+			return 0;
+		}
+	}
+
+	return -ENOBUFS;
+}
+
+static struct mlxsw_sp_rif *mlxsw_sp_rif_alloc(size_t rif_size, u16 rif_index,
+					       u16 vr_id,
+					       struct net_device *l3_dev)
+{
+	struct mlxsw_sp_rif *rif;
+
+	rif = kzalloc(rif_size, GFP_KERNEL);
+	if (!rif)
+		return NULL;
+
+	INIT_LIST_HEAD(&rif->nexthop_list);
+	INIT_LIST_HEAD(&rif->neigh_list);
+	if (l3_dev) {
+		ether_addr_copy(rif->addr, l3_dev->dev_addr);
+		rif->mtu = l3_dev->mtu;
+		rif->dev = l3_dev;
+	}
+	rif->vr_id = vr_id;
+	rif->rif_index = rif_index;
+
+	return rif;
+}
+
+struct mlxsw_sp_rif *mlxsw_sp_rif_by_index(const struct mlxsw_sp *mlxsw_sp,
+					   u16 rif_index)
+{
+	return mlxsw_sp->router->rifs[rif_index];
+}
+
+u16 mlxsw_sp_rif_index(const struct mlxsw_sp_rif *rif)
+{
+	return rif->rif_index;
+}
+
+u16 mlxsw_sp_ipip_lb_rif_index(const struct mlxsw_sp_rif_ipip_lb *lb_rif)
+{
+	return lb_rif->common.rif_index;
+}
+
+u16 mlxsw_sp_ipip_lb_ul_vr_id(const struct mlxsw_sp_rif_ipip_lb *lb_rif)
+{
+	u32 ul_tb_id = mlxsw_sp_ipip_dev_ul_tb_id(lb_rif->common.dev);
+	struct mlxsw_sp_vr *ul_vr;
+
+	ul_vr = mlxsw_sp_vr_get(lb_rif->common.mlxsw_sp, ul_tb_id, NULL);
+	if (WARN_ON(IS_ERR(ul_vr)))
+		return 0;
+
+	return ul_vr->id;
+}
+
+u16 mlxsw_sp_ipip_lb_ul_rif_id(const struct mlxsw_sp_rif_ipip_lb *lb_rif)
+{
+	return lb_rif->ul_rif_id;
+}
+
+static bool
+mlxsw_sp_router_port_l3_stats_enabled(struct mlxsw_sp_rif *rif)
+{
+	return mlxsw_sp_rif_counter_valid_get(rif,
+					      MLXSW_SP_RIF_COUNTER_EGRESS) &&
+	       mlxsw_sp_rif_counter_valid_get(rif,
+					      MLXSW_SP_RIF_COUNTER_INGRESS);
+}
+
+static int
+mlxsw_sp_router_port_l3_stats_enable(struct mlxsw_sp_rif *rif)
+{
+	int err;
+
+	err = mlxsw_sp_rif_counter_alloc(rif, MLXSW_SP_RIF_COUNTER_INGRESS);
+	if (err)
+		return err;
+
+	/* Clear stale data. */
+	err = mlxsw_sp_rif_counter_fetch_clear(rif,
+					       MLXSW_SP_RIF_COUNTER_INGRESS,
+					       NULL);
+	if (err)
+		goto err_clear_ingress;
+
+	err = mlxsw_sp_rif_counter_alloc(rif, MLXSW_SP_RIF_COUNTER_EGRESS);
+	if (err)
+		goto err_alloc_egress;
+
+	/* Clear stale data. */
+	err = mlxsw_sp_rif_counter_fetch_clear(rif,
+					       MLXSW_SP_RIF_COUNTER_EGRESS,
+					       NULL);
+	if (err)
+		goto err_clear_egress;
+
+	return 0;
+
+err_clear_egress:
+	mlxsw_sp_rif_counter_free(rif, MLXSW_SP_RIF_COUNTER_EGRESS);
+err_alloc_egress:
+err_clear_ingress:
+	mlxsw_sp_rif_counter_free(rif, MLXSW_SP_RIF_COUNTER_INGRESS);
+	return err;
+}
+
+static void
+mlxsw_sp_router_port_l3_stats_disable(struct mlxsw_sp_rif *rif)
+{
+	mlxsw_sp_rif_counter_free(rif, MLXSW_SP_RIF_COUNTER_EGRESS);
+	mlxsw_sp_rif_counter_free(rif, MLXSW_SP_RIF_COUNTER_INGRESS);
+}
+
+static void
+mlxsw_sp_router_port_l3_stats_report_used(struct mlxsw_sp_rif *rif,
+					  struct netdev_notifier_offload_xstats_info *info)
+{
+	if (!mlxsw_sp_router_port_l3_stats_enabled(rif))
+		return;
+	netdev_offload_xstats_report_used(info->report_used);
+}
+
+static int
+mlxsw_sp_router_port_l3_stats_fetch(struct mlxsw_sp_rif *rif,
+				    struct rtnl_hw_stats64 *p_stats)
+{
+	struct mlxsw_sp_rif_counter_set_basic ingress;
+	struct mlxsw_sp_rif_counter_set_basic egress;
+	int err;
+
+	err = mlxsw_sp_rif_counter_fetch_clear(rif,
+					       MLXSW_SP_RIF_COUNTER_INGRESS,
+					       &ingress);
+	if (err)
+		return err;
+
+	err = mlxsw_sp_rif_counter_fetch_clear(rif,
+					       MLXSW_SP_RIF_COUNTER_EGRESS,
+					       &egress);
+	if (err)
+		return err;
+
+#define MLXSW_SP_ROUTER_ALL_GOOD(SET, SFX)		\
+		((SET.good_unicast_ ## SFX) +		\
+		 (SET.good_multicast_ ## SFX) +		\
+		 (SET.good_broadcast_ ## SFX))
+
+	p_stats->rx_packets = MLXSW_SP_ROUTER_ALL_GOOD(ingress, packets);
+	p_stats->tx_packets = MLXSW_SP_ROUTER_ALL_GOOD(egress, packets);
+	p_stats->rx_bytes = MLXSW_SP_ROUTER_ALL_GOOD(ingress, bytes);
+	p_stats->tx_bytes = MLXSW_SP_ROUTER_ALL_GOOD(egress, bytes);
+	p_stats->rx_errors = ingress.error_packets;
+	p_stats->tx_errors = egress.error_packets;
+	p_stats->rx_dropped = ingress.discard_packets;
+	p_stats->tx_dropped = egress.discard_packets;
+	p_stats->multicast = ingress.good_multicast_packets +
+			     ingress.good_broadcast_packets;
+
+#undef MLXSW_SP_ROUTER_ALL_GOOD
+
+	return 0;
+}
+
+static int
+mlxsw_sp_router_port_l3_stats_report_delta(struct mlxsw_sp_rif *rif,
+					   struct netdev_notifier_offload_xstats_info *info)
+{
+	struct rtnl_hw_stats64 stats = {};
+	int err;
+
+	if (!mlxsw_sp_router_port_l3_stats_enabled(rif))
+		return 0;
+
+	err = mlxsw_sp_router_port_l3_stats_fetch(rif, &stats);
+	if (err)
+		return err;
+
+	netdev_offload_xstats_report_delta(info->report_delta, &stats);
+	return 0;
+}
+
+struct mlxsw_sp_router_hwstats_notify_work {
+	struct work_struct work;
+	struct net_device *dev;
+};
+
+static void mlxsw_sp_router_hwstats_notify_work(struct work_struct *work)
+{
+	struct mlxsw_sp_router_hwstats_notify_work *hws_work =
+		container_of(work, struct mlxsw_sp_router_hwstats_notify_work,
+			     work);
+
+	rtnl_lock();
+	rtnl_offload_xstats_notify(hws_work->dev);
+	rtnl_unlock();
+	dev_put(hws_work->dev);
+	kfree(hws_work);
+}
+
+static void
+mlxsw_sp_router_hwstats_notify_schedule(struct net_device *dev)
+{
+	struct mlxsw_sp_router_hwstats_notify_work *hws_work;
+
+	/* To collect notification payload, the core ends up sending another
+	 * notifier block message, which would deadlock on the attempt to
+	 * acquire the router lock again. Just postpone the notification until
+	 * later.
+	 */
+
+	hws_work = kzalloc(sizeof(*hws_work), GFP_KERNEL);
+	if (!hws_work)
+		return;
+
+	INIT_WORK(&hws_work->work, mlxsw_sp_router_hwstats_notify_work);
+	dev_hold(dev);
+	hws_work->dev = dev;
+	mlxsw_core_schedule_work(&hws_work->work);
+}
+
+int mlxsw_sp_rif_dev_ifindex(const struct mlxsw_sp_rif *rif)
+{
+	return rif->dev->ifindex;
+}
+
+const struct net_device *mlxsw_sp_rif_dev(const struct mlxsw_sp_rif *rif)
+{
+	return rif->dev;
+}
+
+static void mlxsw_sp_rif_push_l3_stats(struct mlxsw_sp_rif *rif)
+{
+	struct rtnl_hw_stats64 stats = {};
+
+	if (!mlxsw_sp_router_port_l3_stats_fetch(rif, &stats))
+		netdev_offload_xstats_push_delta(rif->dev,
+						 NETDEV_OFFLOAD_XSTATS_TYPE_L3,
+						 &stats);
+}
+
+static struct mlxsw_sp_rif *
+mlxsw_sp_rif_create(struct mlxsw_sp *mlxsw_sp,
+		    const struct mlxsw_sp_rif_params *params,
+		    struct netlink_ext_ack *extack)
+{
+	u32 tb_id = l3mdev_fib_table(params->dev);
+	const struct mlxsw_sp_rif_ops *ops;
+	struct mlxsw_sp_fid *fid = NULL;
+	enum mlxsw_sp_rif_type type;
+	struct mlxsw_sp_rif *rif;
+	struct mlxsw_sp_vr *vr;
+	u16 rif_index;
+	int i, err;
+
+	type = mlxsw_sp_dev_rif_type(mlxsw_sp, params->dev);
+	ops = mlxsw_sp->router->rif_ops_arr[type];
+
+	vr = mlxsw_sp_vr_get(mlxsw_sp, tb_id ? : RT_TABLE_MAIN, extack);
+	if (IS_ERR(vr))
+		return ERR_CAST(vr);
+	vr->rif_count++;
+
+	err = mlxsw_sp_rif_index_alloc(mlxsw_sp, &rif_index);
+	if (err) {
+		NL_SET_ERR_MSG_MOD(extack, "Exceeded number of supported router interfaces");
+		goto err_rif_index_alloc;
+	}
+
+	rif = mlxsw_sp_rif_alloc(ops->rif_size, rif_index, vr->id, params->dev);
+	if (!rif) {
+		err = -ENOMEM;
+		goto err_rif_alloc;
+	}
+	dev_hold(rif->dev);
+	mlxsw_sp->router->rifs[rif_index] = rif;
+	rif->mlxsw_sp = mlxsw_sp;
+	rif->ops = ops;
+
+	if (ops->fid_get) {
+		fid = ops->fid_get(rif, extack);
+		if (IS_ERR(fid)) {
+			err = PTR_ERR(fid);
+			goto err_fid_get;
+		}
+		rif->fid = fid;
+	}
+
+	if (ops->setup)
+		ops->setup(rif, params);
+
+	err = ops->configure(rif, extack);
+	if (err)
+		goto err_configure;
+
+	for (i = 0; i < MLXSW_SP_L3_PROTO_MAX; i++) {
+		err = mlxsw_sp_mr_rif_add(vr->mr_table[i], rif);
+		if (err)
+			goto err_mr_rif_add;
+	}
+
+	if (netdev_offload_xstats_enabled(rif->dev,
+					  NETDEV_OFFLOAD_XSTATS_TYPE_L3)) {
+		err = mlxsw_sp_router_port_l3_stats_enable(rif);
+		if (err)
+			goto err_stats_enable;
+		mlxsw_sp_router_hwstats_notify_schedule(rif->dev);
+	} else {
+		mlxsw_sp_rif_counters_alloc(rif);
+	}
+
+	atomic_inc(&mlxsw_sp->router->rifs_count);
+	return rif;
+
+err_stats_enable:
+err_mr_rif_add:
+	for (i--; i >= 0; i--)
+		mlxsw_sp_mr_rif_del(vr->mr_table[i], rif);
+	ops->deconfigure(rif);
+err_configure:
+	if (fid)
+		mlxsw_sp_fid_put(fid);
+err_fid_get:
+	mlxsw_sp->router->rifs[rif_index] = NULL;
+	dev_put(rif->dev);
+	kfree(rif);
+err_rif_alloc:
+err_rif_index_alloc:
+	vr->rif_count--;
+	mlxsw_sp_vr_put(mlxsw_sp, vr);
+	return ERR_PTR(err);
+}
+
+static void mlxsw_sp_rif_destroy(struct mlxsw_sp_rif *rif)
+{
+	const struct mlxsw_sp_rif_ops *ops = rif->ops;
+	struct mlxsw_sp *mlxsw_sp = rif->mlxsw_sp;
+	struct mlxsw_sp_fid *fid = rif->fid;
+	struct mlxsw_sp_vr *vr;
+	int i;
+
+	atomic_dec(&mlxsw_sp->router->rifs_count);
+	mlxsw_sp_router_rif_gone_sync(mlxsw_sp, rif);
+	vr = &mlxsw_sp->router->vrs[rif->vr_id];
+
+	if (netdev_offload_xstats_enabled(rif->dev,
+					  NETDEV_OFFLOAD_XSTATS_TYPE_L3)) {
+		mlxsw_sp_rif_push_l3_stats(rif);
+		mlxsw_sp_router_port_l3_stats_disable(rif);
+		mlxsw_sp_router_hwstats_notify_schedule(rif->dev);
+	} else {
+		mlxsw_sp_rif_counters_free(rif);
+	}
+
+	for (i = 0; i < MLXSW_SP_L3_PROTO_MAX; i++)
+		mlxsw_sp_mr_rif_del(vr->mr_table[i], rif);
+	ops->deconfigure(rif);
+	if (fid)
+		/* Loopback RIFs are not associated with a FID. */
+		mlxsw_sp_fid_put(fid);
+	mlxsw_sp->router->rifs[rif->rif_index] = NULL;
+	dev_put(rif->dev);
+	kfree(rif);
+	vr->rif_count--;
+	mlxsw_sp_vr_put(mlxsw_sp, vr);
+}
+
+void mlxsw_sp_rif_destroy_by_dev(struct mlxsw_sp *mlxsw_sp,
+				 struct net_device *dev)
+{
+	struct mlxsw_sp_rif *rif;
+
+	mutex_lock(&mlxsw_sp->router->lock);
+	rif = mlxsw_sp_rif_find_by_dev(mlxsw_sp, dev);
+	if (!rif)
+		goto out;
+	mlxsw_sp_rif_destroy(rif);
+out:
+	mutex_unlock(&mlxsw_sp->router->lock);
+}
+
+static void
+mlxsw_sp_rif_subport_params_init(struct mlxsw_sp_rif_params *params,
+				 struct mlxsw_sp_port_vlan *mlxsw_sp_port_vlan)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = mlxsw_sp_port_vlan->mlxsw_sp_port;
+
+	params->vid = mlxsw_sp_port_vlan->vid;
+	params->lag = mlxsw_sp_port->lagged;
+	if (params->lag)
+		params->lag_id = mlxsw_sp_port->lag_id;
+	else
+		params->system_port = mlxsw_sp_port->local_port;
+}
+
+static struct mlxsw_sp_rif_subport *
+mlxsw_sp_rif_subport_rif(const struct mlxsw_sp_rif *rif)
+{
+	return container_of(rif, struct mlxsw_sp_rif_subport, common);
+}
+
+static struct mlxsw_sp_rif *
+mlxsw_sp_rif_subport_get(struct mlxsw_sp *mlxsw_sp,
+			 const struct mlxsw_sp_rif_params *params,
+			 struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp_rif_subport *rif_subport;
+	struct mlxsw_sp_rif *rif;
+
+	rif = mlxsw_sp_rif_find_by_dev(mlxsw_sp, params->dev);
+	if (!rif)
+		return mlxsw_sp_rif_create(mlxsw_sp, params, extack);
+
+	rif_subport = mlxsw_sp_rif_subport_rif(rif);
+	refcount_inc(&rif_subport->ref_count);
+	return rif;
+}
+
+static void mlxsw_sp_rif_subport_put(struct mlxsw_sp_rif *rif)
+{
+	struct mlxsw_sp_rif_subport *rif_subport;
+
+	rif_subport = mlxsw_sp_rif_subport_rif(rif);
+	if (!refcount_dec_and_test(&rif_subport->ref_count))
+		return;
+
+	mlxsw_sp_rif_destroy(rif);
+}
+
+static int mlxsw_sp_rif_mac_profile_index_alloc(struct mlxsw_sp *mlxsw_sp,
+						struct mlxsw_sp_rif_mac_profile *profile,
+						struct netlink_ext_ack *extack)
+{
+	u8 max_rif_mac_profiles = mlxsw_sp->router->max_rif_mac_profile;
+	struct mlxsw_sp_router *router = mlxsw_sp->router;
+	int id;
+
+	id = idr_alloc(&router->rif_mac_profiles_idr, profile, 0,
+		       max_rif_mac_profiles, GFP_KERNEL);
+
+	if (id >= 0) {
+		profile->id = id;
+		return 0;
+	}
+
+	if (id == -ENOSPC)
+		NL_SET_ERR_MSG_MOD(extack,
+				   "Exceeded number of supported router interface MAC profiles");
+
+	return id;
+}
+
+static struct mlxsw_sp_rif_mac_profile *
+mlxsw_sp_rif_mac_profile_index_free(struct mlxsw_sp *mlxsw_sp, u8 mac_profile)
+{
+	struct mlxsw_sp_rif_mac_profile *profile;
+
+	profile = idr_remove(&mlxsw_sp->router->rif_mac_profiles_idr,
+			     mac_profile);
+	WARN_ON(!profile);
+	return profile;
+}
+
+static struct mlxsw_sp_rif_mac_profile *
+mlxsw_sp_rif_mac_profile_alloc(const char *mac)
+{
+	struct mlxsw_sp_rif_mac_profile *profile;
+
+	profile = kzalloc(sizeof(*profile), GFP_KERNEL);
+	if (!profile)
+		return NULL;
+
+	ether_addr_copy(profile->mac_prefix, mac);
+	refcount_set(&profile->ref_count, 1);
+	return profile;
+}
+
+static struct mlxsw_sp_rif_mac_profile *
+mlxsw_sp_rif_mac_profile_find(const struct mlxsw_sp *mlxsw_sp, const char *mac)
+{
+	struct mlxsw_sp_router *router = mlxsw_sp->router;
+	struct mlxsw_sp_rif_mac_profile *profile;
+	int id;
+
+	idr_for_each_entry(&router->rif_mac_profiles_idr, profile, id) {
+		if (ether_addr_equal_masked(profile->mac_prefix, mac,
+					    mlxsw_sp->mac_mask))
+			return profile;
+	}
+
+	return NULL;
+}
+
+static u64 mlxsw_sp_rif_mac_profiles_occ_get(void *priv)
+{
+	const struct mlxsw_sp *mlxsw_sp = priv;
+
+	return atomic_read(&mlxsw_sp->router->rif_mac_profiles_count);
+}
+
+static u64 mlxsw_sp_rifs_occ_get(void *priv)
+{
+	const struct mlxsw_sp *mlxsw_sp = priv;
+
+	return atomic_read(&mlxsw_sp->router->rifs_count);
+}
+
+static struct mlxsw_sp_rif_mac_profile *
+mlxsw_sp_rif_mac_profile_create(struct mlxsw_sp *mlxsw_sp, const char *mac,
+				struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp_rif_mac_profile *profile;
+	int err;
+
+	profile = mlxsw_sp_rif_mac_profile_alloc(mac);
+	if (!profile)
+		return ERR_PTR(-ENOMEM);
+
+	err = mlxsw_sp_rif_mac_profile_index_alloc(mlxsw_sp, profile, extack);
+	if (err)
+		goto profile_index_alloc_err;
+
+	atomic_inc(&mlxsw_sp->router->rif_mac_profiles_count);
+	return profile;
+
+profile_index_alloc_err:
+	kfree(profile);
+	return ERR_PTR(err);
+}
+
+static void mlxsw_sp_rif_mac_profile_destroy(struct mlxsw_sp *mlxsw_sp,
+					     u8 mac_profile)
+{
+	struct mlxsw_sp_rif_mac_profile *profile;
+
+	atomic_dec(&mlxsw_sp->router->rif_mac_profiles_count);
+	profile = mlxsw_sp_rif_mac_profile_index_free(mlxsw_sp, mac_profile);
+	kfree(profile);
+}
+
+static int mlxsw_sp_rif_mac_profile_get(struct mlxsw_sp *mlxsw_sp,
+					const char *mac, u8 *p_mac_profile,
+					struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp_rif_mac_profile *profile;
+
+	profile = mlxsw_sp_rif_mac_profile_find(mlxsw_sp, mac);
+	if (profile) {
+		refcount_inc(&profile->ref_count);
+		goto out;
+	}
+
+	profile = mlxsw_sp_rif_mac_profile_create(mlxsw_sp, mac, extack);
+	if (IS_ERR(profile))
+		return PTR_ERR(profile);
+
+out:
+	*p_mac_profile = profile->id;
+	return 0;
+}
+
+static void mlxsw_sp_rif_mac_profile_put(struct mlxsw_sp *mlxsw_sp,
+					 u8 mac_profile)
+{
+	struct mlxsw_sp_rif_mac_profile *profile;
+
+	profile = idr_find(&mlxsw_sp->router->rif_mac_profiles_idr,
+			   mac_profile);
+	if (WARN_ON(!profile))
+		return;
+
+	if (!refcount_dec_and_test(&profile->ref_count))
+		return;
+
+	mlxsw_sp_rif_mac_profile_destroy(mlxsw_sp, mac_profile);
+}
+
+static bool mlxsw_sp_rif_mac_profile_is_shared(const struct mlxsw_sp_rif *rif)
+{
+	struct mlxsw_sp *mlxsw_sp = rif->mlxsw_sp;
+	struct mlxsw_sp_rif_mac_profile *profile;
+
+	profile = idr_find(&mlxsw_sp->router->rif_mac_profiles_idr,
+			   rif->mac_profile_id);
+	if (WARN_ON(!profile))
+		return false;
+
+	return refcount_read(&profile->ref_count) > 1;
+}
+
+static int mlxsw_sp_rif_mac_profile_edit(struct mlxsw_sp_rif *rif,
+					 const char *new_mac)
+{
+	struct mlxsw_sp *mlxsw_sp = rif->mlxsw_sp;
+	struct mlxsw_sp_rif_mac_profile *profile;
+
+	profile = idr_find(&mlxsw_sp->router->rif_mac_profiles_idr,
+			   rif->mac_profile_id);
+	if (WARN_ON(!profile))
+		return -EINVAL;
+
+	ether_addr_copy(profile->mac_prefix, new_mac);
+	return 0;
+}
+
+static int
+mlxsw_sp_rif_mac_profile_replace(struct mlxsw_sp *mlxsw_sp,
+				 struct mlxsw_sp_rif *rif,
+				 const char *new_mac,
+				 struct netlink_ext_ack *extack)
+{
+	u8 mac_profile;
+	int err;
+
+	if (!mlxsw_sp_rif_mac_profile_is_shared(rif) &&
+	    !mlxsw_sp_rif_mac_profile_find(mlxsw_sp, new_mac))
+		return mlxsw_sp_rif_mac_profile_edit(rif, new_mac);
+
+	err = mlxsw_sp_rif_mac_profile_get(mlxsw_sp, new_mac,
+					   &mac_profile, extack);
+	if (err)
+		return err;
+
+	mlxsw_sp_rif_mac_profile_put(mlxsw_sp, rif->mac_profile_id);
+	rif->mac_profile_id = mac_profile;
+	return 0;
+}
+
+static int
+__mlxsw_sp_port_vlan_router_join(struct mlxsw_sp_port_vlan *mlxsw_sp_port_vlan,
+				 struct net_device *l3_dev,
+				 struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = mlxsw_sp_port_vlan->mlxsw_sp_port;
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	struct mlxsw_sp_rif_params params = {
+		.dev = l3_dev,
+	};
+	u16 vid = mlxsw_sp_port_vlan->vid;
+	struct mlxsw_sp_rif *rif;
+	struct mlxsw_sp_fid *fid;
+	int err;
+
+	mlxsw_sp_rif_subport_params_init(&params, mlxsw_sp_port_vlan);
+	rif = mlxsw_sp_rif_subport_get(mlxsw_sp, &params, extack);
+	if (IS_ERR(rif))
+		return PTR_ERR(rif);
+
+	/* FID was already created, just take a reference */
+	fid = rif->ops->fid_get(rif, extack);
+	err = mlxsw_sp_fid_port_vid_map(fid, mlxsw_sp_port, vid);
+	if (err)
+		goto err_fid_port_vid_map;
+
+	err = mlxsw_sp_port_vid_learning_set(mlxsw_sp_port, vid, false);
+	if (err)
+		goto err_port_vid_learning_set;
+
+	err = mlxsw_sp_port_vid_stp_set(mlxsw_sp_port, vid,
+					BR_STATE_FORWARDING);
+	if (err)
+		goto err_port_vid_stp_set;
+
+	mlxsw_sp_port_vlan->fid = fid;
+
+	return 0;
+
+err_port_vid_stp_set:
+	mlxsw_sp_port_vid_learning_set(mlxsw_sp_port, vid, true);
+err_port_vid_learning_set:
+	mlxsw_sp_fid_port_vid_unmap(fid, mlxsw_sp_port, vid);
+err_fid_port_vid_map:
+	mlxsw_sp_fid_put(fid);
+	mlxsw_sp_rif_subport_put(rif);
+	return err;
+}
+
+static void
+__mlxsw_sp_port_vlan_router_leave(struct mlxsw_sp_port_vlan *mlxsw_sp_port_vlan)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = mlxsw_sp_port_vlan->mlxsw_sp_port;
+	struct mlxsw_sp_fid *fid = mlxsw_sp_port_vlan->fid;
+	struct mlxsw_sp_rif *rif = mlxsw_sp_fid_rif(fid);
+	u16 vid = mlxsw_sp_port_vlan->vid;
+
+	if (WARN_ON(mlxsw_sp_fid_type(fid) != MLXSW_SP_FID_TYPE_RFID))
+		return;
+
+	mlxsw_sp_port_vlan->fid = NULL;
+	mlxsw_sp_port_vid_stp_set(mlxsw_sp_port, vid, BR_STATE_BLOCKING);
+	mlxsw_sp_port_vid_learning_set(mlxsw_sp_port, vid, true);
+	mlxsw_sp_fid_port_vid_unmap(fid, mlxsw_sp_port, vid);
+	mlxsw_sp_fid_put(fid);
+	mlxsw_sp_rif_subport_put(rif);
+}
+
+int
+mlxsw_sp_port_vlan_router_join(struct mlxsw_sp_port_vlan *mlxsw_sp_port_vlan,
+			       struct net_device *l3_dev,
+			       struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port_vlan->mlxsw_sp_port->mlxsw_sp;
+	struct mlxsw_sp_rif *rif;
+	int err = 0;
+
+	mutex_lock(&mlxsw_sp->router->lock);
+	rif = mlxsw_sp_rif_find_by_dev(mlxsw_sp, l3_dev);
+	if (!rif)
+		goto out;
+
+	err = __mlxsw_sp_port_vlan_router_join(mlxsw_sp_port_vlan, l3_dev,
+					       extack);
+out:
+	mutex_unlock(&mlxsw_sp->router->lock);
+	return err;
+}
+
+void
+mlxsw_sp_port_vlan_router_leave(struct mlxsw_sp_port_vlan *mlxsw_sp_port_vlan)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port_vlan->mlxsw_sp_port->mlxsw_sp;
+
+	mutex_lock(&mlxsw_sp->router->lock);
+	__mlxsw_sp_port_vlan_router_leave(mlxsw_sp_port_vlan);
+	mutex_unlock(&mlxsw_sp->router->lock);
+}
+
+static int mlxsw_sp_inetaddr_port_vlan_event(struct net_device *l3_dev,
+					     struct net_device *port_dev,
+					     unsigned long event, u16 vid,
+					     struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(port_dev);
+	struct mlxsw_sp_port_vlan *mlxsw_sp_port_vlan;
+
+	mlxsw_sp_port_vlan = mlxsw_sp_port_vlan_find_by_vid(mlxsw_sp_port, vid);
+	if (WARN_ON(!mlxsw_sp_port_vlan))
+		return -EINVAL;
+
+	switch (event) {
+	case NETDEV_UP:
+		return __mlxsw_sp_port_vlan_router_join(mlxsw_sp_port_vlan,
+							l3_dev, extack);
+	case NETDEV_DOWN:
+		__mlxsw_sp_port_vlan_router_leave(mlxsw_sp_port_vlan);
+		break;
+	}
+
+	return 0;
+}
+
+static int mlxsw_sp_inetaddr_port_event(struct net_device *port_dev,
+					unsigned long event,
+					struct netlink_ext_ack *extack)
+{
+	if (netif_is_any_bridge_port(port_dev) || netif_is_lag_port(port_dev))
+		return 0;
+
+	return mlxsw_sp_inetaddr_port_vlan_event(port_dev, port_dev, event,
+						 MLXSW_SP_DEFAULT_VID, extack);
+}
+
+static int __mlxsw_sp_inetaddr_lag_event(struct net_device *l3_dev,
+					 struct net_device *lag_dev,
+					 unsigned long event, u16 vid,
+					 struct netlink_ext_ack *extack)
+{
+	struct net_device *port_dev;
+	struct list_head *iter;
+	int err;
+
+	netdev_for_each_lower_dev(lag_dev, port_dev, iter) {
+		if (mlxsw_sp_port_dev_check(port_dev)) {
+			err = mlxsw_sp_inetaddr_port_vlan_event(l3_dev,
+								port_dev,
+								event, vid,
+								extack);
+			if (err)
+				return err;
+		}
+	}
+
+	return 0;
+}
+
+static int mlxsw_sp_inetaddr_lag_event(struct net_device *lag_dev,
+				       unsigned long event,
+				       struct netlink_ext_ack *extack)
+{
+	if (netif_is_bridge_port(lag_dev))
+		return 0;
+
+	return __mlxsw_sp_inetaddr_lag_event(lag_dev, lag_dev, event,
+					     MLXSW_SP_DEFAULT_VID, extack);
+}
+
+static int mlxsw_sp_inetaddr_bridge_event(struct mlxsw_sp *mlxsw_sp,
+					  struct net_device *l3_dev,
+					  unsigned long event,
+					  struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp_rif_params params = {
+		.dev = l3_dev,
+	};
+	struct mlxsw_sp_rif *rif;
+
+	switch (event) {
+	case NETDEV_UP:
+		if (netif_is_bridge_master(l3_dev) && br_vlan_enabled(l3_dev)) {
+			u16 proto;
+
+			br_vlan_get_proto(l3_dev, &proto);
+			if (proto == ETH_P_8021AD) {
+				NL_SET_ERR_MSG_MOD(extack, "Adding an IP address to 802.1ad bridge is not supported");
+				return -EOPNOTSUPP;
+			}
+		}
+		rif = mlxsw_sp_rif_create(mlxsw_sp, &params, extack);
+		if (IS_ERR(rif))
+			return PTR_ERR(rif);
+		break;
+	case NETDEV_DOWN:
+		rif = mlxsw_sp_rif_find_by_dev(mlxsw_sp, l3_dev);
+		mlxsw_sp_rif_destroy(rif);
+		break;
+	}
+
+	return 0;
+}
+
+static int mlxsw_sp_inetaddr_vlan_event(struct mlxsw_sp *mlxsw_sp,
+					struct net_device *vlan_dev,
+					unsigned long event,
+					struct netlink_ext_ack *extack)
+{
+	struct net_device *real_dev = vlan_dev_real_dev(vlan_dev);
+	u16 vid = vlan_dev_vlan_id(vlan_dev);
+
+	if (netif_is_bridge_port(vlan_dev))
+		return 0;
+
+	if (mlxsw_sp_port_dev_check(real_dev))
+		return mlxsw_sp_inetaddr_port_vlan_event(vlan_dev, real_dev,
+							 event, vid, extack);
+	else if (netif_is_lag_master(real_dev))
+		return __mlxsw_sp_inetaddr_lag_event(vlan_dev, real_dev, event,
+						     vid, extack);
+	else if (netif_is_bridge_master(real_dev) && br_vlan_enabled(real_dev))
+		return mlxsw_sp_inetaddr_bridge_event(mlxsw_sp, vlan_dev, event,
+						      extack);
+
+	return 0;
+}
+
+static bool mlxsw_sp_rif_macvlan_is_vrrp4(const u8 *mac)
+{
+	u8 vrrp4[ETH_ALEN] = { 0x00, 0x00, 0x5e, 0x00, 0x01, 0x00 };
+	u8 mask[ETH_ALEN] = { 0xff, 0xff, 0xff, 0xff, 0xff, 0x00 };
+
+	return ether_addr_equal_masked(mac, vrrp4, mask);
+}
+
+static bool mlxsw_sp_rif_macvlan_is_vrrp6(const u8 *mac)
+{
+	u8 vrrp6[ETH_ALEN] = { 0x00, 0x00, 0x5e, 0x00, 0x02, 0x00 };
+	u8 mask[ETH_ALEN] = { 0xff, 0xff, 0xff, 0xff, 0xff, 0x00 };
+
+	return ether_addr_equal_masked(mac, vrrp6, mask);
+}
+
+static int mlxsw_sp_rif_vrrp_op(struct mlxsw_sp *mlxsw_sp, u16 rif_index,
+				const u8 *mac, bool adding)
+{
+	char ritr_pl[MLXSW_REG_RITR_LEN];
+	u8 vrrp_id = adding ? mac[5] : 0;
+	int err;
+
+	if (!mlxsw_sp_rif_macvlan_is_vrrp4(mac) &&
+	    !mlxsw_sp_rif_macvlan_is_vrrp6(mac))
+		return 0;
+
+	mlxsw_reg_ritr_rif_pack(ritr_pl, rif_index);
+	err = mlxsw_reg_query(mlxsw_sp->core, MLXSW_REG(ritr), ritr_pl);
+	if (err)
+		return err;
+
+	if (mlxsw_sp_rif_macvlan_is_vrrp4(mac))
+		mlxsw_reg_ritr_if_vrrp_id_ipv4_set(ritr_pl, vrrp_id);
+	else
+		mlxsw_reg_ritr_if_vrrp_id_ipv6_set(ritr_pl, vrrp_id);
+
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(ritr), ritr_pl);
+}
+
+static int mlxsw_sp_rif_macvlan_add(struct mlxsw_sp *mlxsw_sp,
+				    const struct net_device *macvlan_dev,
+				    struct netlink_ext_ack *extack)
+{
+	struct macvlan_dev *vlan = netdev_priv(macvlan_dev);
+	struct mlxsw_sp_rif *rif;
+	int err;
+
+	rif = mlxsw_sp_rif_find_by_dev(mlxsw_sp, vlan->lowerdev);
+	if (!rif) {
+		NL_SET_ERR_MSG_MOD(extack, "macvlan is only supported on top of router interfaces");
+		return -EOPNOTSUPP;
+	}
+
+	err = mlxsw_sp_rif_fdb_op(mlxsw_sp, macvlan_dev->dev_addr,
+				  mlxsw_sp_fid_index(rif->fid), true);
+	if (err)
+		return err;
+
+	err = mlxsw_sp_rif_vrrp_op(mlxsw_sp, rif->rif_index,
+				   macvlan_dev->dev_addr, true);
+	if (err)
+		goto err_rif_vrrp_add;
+
+	/* Make sure the bridge driver does not have this MAC pointing at
+	 * some other port.
+	 */
+	if (rif->ops->fdb_del)
+		rif->ops->fdb_del(rif, macvlan_dev->dev_addr);
+
+	return 0;
+
+err_rif_vrrp_add:
+	mlxsw_sp_rif_fdb_op(mlxsw_sp, macvlan_dev->dev_addr,
+			    mlxsw_sp_fid_index(rif->fid), false);
+	return err;
+}
+
+static void __mlxsw_sp_rif_macvlan_del(struct mlxsw_sp *mlxsw_sp,
+				       const struct net_device *macvlan_dev)
+{
+	struct macvlan_dev *vlan = netdev_priv(macvlan_dev);
+	struct mlxsw_sp_rif *rif;
+
+	rif = mlxsw_sp_rif_find_by_dev(mlxsw_sp, vlan->lowerdev);
+	/* If we do not have a RIF, then we already took care of
+	 * removing the macvlan's MAC during RIF deletion.
+	 */
+	if (!rif)
+		return;
+	mlxsw_sp_rif_vrrp_op(mlxsw_sp, rif->rif_index, macvlan_dev->dev_addr,
+			     false);
+	mlxsw_sp_rif_fdb_op(mlxsw_sp, macvlan_dev->dev_addr,
+			    mlxsw_sp_fid_index(rif->fid), false);
+}
+
+void mlxsw_sp_rif_macvlan_del(struct mlxsw_sp *mlxsw_sp,
+			      const struct net_device *macvlan_dev)
+{
+	mutex_lock(&mlxsw_sp->router->lock);
+	__mlxsw_sp_rif_macvlan_del(mlxsw_sp, macvlan_dev);
+	mutex_unlock(&mlxsw_sp->router->lock);
+}
+
+static int mlxsw_sp_inetaddr_macvlan_event(struct mlxsw_sp *mlxsw_sp,
+					   struct net_device *macvlan_dev,
+					   unsigned long event,
+					   struct netlink_ext_ack *extack)
+{
+	switch (event) {
+	case NETDEV_UP:
+		return mlxsw_sp_rif_macvlan_add(mlxsw_sp, macvlan_dev, extack);
+	case NETDEV_DOWN:
+		__mlxsw_sp_rif_macvlan_del(mlxsw_sp, macvlan_dev);
+		break;
+	}
+
+	return 0;
+}
+
+static int __mlxsw_sp_inetaddr_event(struct mlxsw_sp *mlxsw_sp,
+				     struct net_device *dev,
+				     unsigned long event,
+				     struct netlink_ext_ack *extack)
+{
+	if (mlxsw_sp_port_dev_check(dev))
+		return mlxsw_sp_inetaddr_port_event(dev, event, extack);
+	else if (netif_is_lag_master(dev))
+		return mlxsw_sp_inetaddr_lag_event(dev, event, extack);
+	else if (netif_is_bridge_master(dev))
+		return mlxsw_sp_inetaddr_bridge_event(mlxsw_sp, dev, event,
+						      extack);
+	else if (is_vlan_dev(dev))
+		return mlxsw_sp_inetaddr_vlan_event(mlxsw_sp, dev, event,
+						    extack);
+	else if (netif_is_macvlan(dev))
+		return mlxsw_sp_inetaddr_macvlan_event(mlxsw_sp, dev, event,
+						       extack);
+	else
+		return 0;
+}
+
+static int mlxsw_sp_inetaddr_event(struct notifier_block *nb,
+				   unsigned long event, void *ptr)
+{
+	struct in_ifaddr *ifa = (struct in_ifaddr *) ptr;
+	struct net_device *dev = ifa->ifa_dev->dev;
+	struct mlxsw_sp_router *router;
+	struct mlxsw_sp_rif *rif;
+	int err = 0;
+
+	/* NETDEV_UP event is handled by mlxsw_sp_inetaddr_valid_event */
+	if (event == NETDEV_UP)
+		return NOTIFY_DONE;
+
+	router = container_of(nb, struct mlxsw_sp_router, inetaddr_nb);
+	mutex_lock(&router->lock);
+	rif = mlxsw_sp_rif_find_by_dev(router->mlxsw_sp, dev);
+	if (!mlxsw_sp_rif_should_config(rif, dev, event))
+		goto out;
+
+	err = __mlxsw_sp_inetaddr_event(router->mlxsw_sp, dev, event, NULL);
+out:
+	mutex_unlock(&router->lock);
+	return notifier_from_errno(err);
+}
+
+int mlxsw_sp_inetaddr_valid_event(struct notifier_block *unused,
+				  unsigned long event, void *ptr)
+{
+	struct in_validator_info *ivi = (struct in_validator_info *) ptr;
+	struct net_device *dev = ivi->ivi_dev->dev;
+	struct mlxsw_sp *mlxsw_sp;
+	struct mlxsw_sp_rif *rif;
+	int err = 0;
+
+	mlxsw_sp = mlxsw_sp_lower_get(dev);
+	if (!mlxsw_sp)
+		return NOTIFY_DONE;
+
+	mutex_lock(&mlxsw_sp->router->lock);
+	rif = mlxsw_sp_rif_find_by_dev(mlxsw_sp, dev);
+	if (!mlxsw_sp_rif_should_config(rif, dev, event))
+		goto out;
+
+	err = __mlxsw_sp_inetaddr_event(mlxsw_sp, dev, event, ivi->extack);
+out:
+	mutex_unlock(&mlxsw_sp->router->lock);
+	return notifier_from_errno(err);
+}
+
+struct mlxsw_sp_inet6addr_event_work {
+	struct work_struct work;
+	struct mlxsw_sp *mlxsw_sp;
+	struct net_device *dev;
+	unsigned long event;
+};
+
+static void mlxsw_sp_inet6addr_event_work(struct work_struct *work)
+{
+	struct mlxsw_sp_inet6addr_event_work *inet6addr_work =
+		container_of(work, struct mlxsw_sp_inet6addr_event_work, work);
+	struct mlxsw_sp *mlxsw_sp = inet6addr_work->mlxsw_sp;
+	struct net_device *dev = inet6addr_work->dev;
+	unsigned long event = inet6addr_work->event;
+	struct mlxsw_sp_rif *rif;
+
+	rtnl_lock();
+	mutex_lock(&mlxsw_sp->router->lock);
+
+	rif = mlxsw_sp_rif_find_by_dev(mlxsw_sp, dev);
+	if (!mlxsw_sp_rif_should_config(rif, dev, event))
+		goto out;
+
+	__mlxsw_sp_inetaddr_event(mlxsw_sp, dev, event, NULL);
+out:
+	mutex_unlock(&mlxsw_sp->router->lock);
+	rtnl_unlock();
+	dev_put(dev);
+	kfree(inet6addr_work);
+}
+
+/* Called with rcu_read_lock() */
+static int mlxsw_sp_inet6addr_event(struct notifier_block *nb,
+				    unsigned long event, void *ptr)
+{
+	struct inet6_ifaddr *if6 = (struct inet6_ifaddr *) ptr;
+	struct mlxsw_sp_inet6addr_event_work *inet6addr_work;
+	struct net_device *dev = if6->idev->dev;
+	struct mlxsw_sp_router *router;
+
+	/* NETDEV_UP event is handled by mlxsw_sp_inet6addr_valid_event */
+	if (event == NETDEV_UP)
+		return NOTIFY_DONE;
+
+	inet6addr_work = kzalloc(sizeof(*inet6addr_work), GFP_ATOMIC);
+	if (!inet6addr_work)
+		return NOTIFY_BAD;
+
+	router = container_of(nb, struct mlxsw_sp_router, inet6addr_nb);
+	INIT_WORK(&inet6addr_work->work, mlxsw_sp_inet6addr_event_work);
+	inet6addr_work->mlxsw_sp = router->mlxsw_sp;
+	inet6addr_work->dev = dev;
+	inet6addr_work->event = event;
+	dev_hold(dev);
+	mlxsw_core_schedule_work(&inet6addr_work->work);
+
+	return NOTIFY_DONE;
+}
+
+int mlxsw_sp_inet6addr_valid_event(struct notifier_block *unused,
+				   unsigned long event, void *ptr)
+{
+	struct in6_validator_info *i6vi = (struct in6_validator_info *) ptr;
+	struct net_device *dev = i6vi->i6vi_dev->dev;
+	struct mlxsw_sp *mlxsw_sp;
+	struct mlxsw_sp_rif *rif;
+	int err = 0;
+
+	mlxsw_sp = mlxsw_sp_lower_get(dev);
+	if (!mlxsw_sp)
+		return NOTIFY_DONE;
+
+	mutex_lock(&mlxsw_sp->router->lock);
+	rif = mlxsw_sp_rif_find_by_dev(mlxsw_sp, dev);
+	if (!mlxsw_sp_rif_should_config(rif, dev, event))
+		goto out;
+
+	err = __mlxsw_sp_inetaddr_event(mlxsw_sp, dev, event, i6vi->extack);
+out:
+	mutex_unlock(&mlxsw_sp->router->lock);
+	return notifier_from_errno(err);
+}
+
+static int mlxsw_sp_rif_edit(struct mlxsw_sp *mlxsw_sp, u16 rif_index,
+			     const char *mac, int mtu, u8 mac_profile)
+{
+	char ritr_pl[MLXSW_REG_RITR_LEN];
+	int err;
+
+	mlxsw_reg_ritr_rif_pack(ritr_pl, rif_index);
+	err = mlxsw_reg_query(mlxsw_sp->core, MLXSW_REG(ritr), ritr_pl);
+	if (err)
+		return err;
+
+	mlxsw_reg_ritr_mtu_set(ritr_pl, mtu);
+	mlxsw_reg_ritr_if_mac_memcpy_to(ritr_pl, mac);
+	mlxsw_reg_ritr_if_mac_profile_id_set(ritr_pl, mac_profile);
+	mlxsw_reg_ritr_op_set(ritr_pl, MLXSW_REG_RITR_RIF_CREATE);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(ritr), ritr_pl);
+}
+
+static int
+mlxsw_sp_router_port_change_event(struct mlxsw_sp *mlxsw_sp,
+				  struct mlxsw_sp_rif *rif,
+				  struct netlink_ext_ack *extack)
+{
+	struct net_device *dev = rif->dev;
+	u8 old_mac_profile;
+	u16 fid_index;
+	int err;
+
+	fid_index = mlxsw_sp_fid_index(rif->fid);
+
+	err = mlxsw_sp_rif_fdb_op(mlxsw_sp, rif->addr, fid_index, false);
+	if (err)
+		return err;
+
+	old_mac_profile = rif->mac_profile_id;
+	err = mlxsw_sp_rif_mac_profile_replace(mlxsw_sp, rif, dev->dev_addr,
+					       extack);
+	if (err)
+		goto err_rif_mac_profile_replace;
+
+	err = mlxsw_sp_rif_edit(mlxsw_sp, rif->rif_index, dev->dev_addr,
+				dev->mtu, rif->mac_profile_id);
+	if (err)
+		goto err_rif_edit;
+
+	err = mlxsw_sp_rif_fdb_op(mlxsw_sp, dev->dev_addr, fid_index, true);
+	if (err)
+		goto err_rif_fdb_op;
+
+	if (rif->mtu != dev->mtu) {
+		struct mlxsw_sp_vr *vr;
+		int i;
+
+		/* The RIF is relevant only to its mr_table instance, as unlike
+		 * unicast routing, in multicast routing a RIF cannot be shared
+		 * between several multicast routing tables.
+		 */
+		vr = &mlxsw_sp->router->vrs[rif->vr_id];
+		for (i = 0; i < MLXSW_SP_L3_PROTO_MAX; i++)
+			mlxsw_sp_mr_rif_mtu_update(vr->mr_table[i],
+						   rif, dev->mtu);
+	}
+
+	ether_addr_copy(rif->addr, dev->dev_addr);
+	rif->mtu = dev->mtu;
+
+	netdev_dbg(dev, "Updated RIF=%d\n", rif->rif_index);
+
+	return 0;
+
+err_rif_fdb_op:
+	mlxsw_sp_rif_edit(mlxsw_sp, rif->rif_index, rif->addr, rif->mtu,
+			  old_mac_profile);
+err_rif_edit:
+	mlxsw_sp_rif_mac_profile_replace(mlxsw_sp, rif, rif->addr, extack);
+err_rif_mac_profile_replace:
+	mlxsw_sp_rif_fdb_op(mlxsw_sp, rif->addr, fid_index, true);
+	return err;
+}
+
+static int mlxsw_sp_router_port_pre_changeaddr_event(struct mlxsw_sp_rif *rif,
+			    struct netdev_notifier_pre_changeaddr_info *info)
+{
+	struct mlxsw_sp *mlxsw_sp = rif->mlxsw_sp;
+	struct mlxsw_sp_rif_mac_profile *profile;
+	struct netlink_ext_ack *extack;
+	u8 max_rif_mac_profiles;
+	u64 occ;
+
+	extack = netdev_notifier_info_to_extack(&info->info);
+
+	profile = mlxsw_sp_rif_mac_profile_find(mlxsw_sp, info->dev_addr);
+	if (profile)
+		return 0;
+
+	max_rif_mac_profiles = mlxsw_sp->router->max_rif_mac_profile;
+	occ = mlxsw_sp_rif_mac_profiles_occ_get(mlxsw_sp);
+	if (occ < max_rif_mac_profiles)
+		return 0;
+
+	if (!mlxsw_sp_rif_mac_profile_is_shared(rif))
+		return 0;
+
+	NL_SET_ERR_MSG_MOD(extack, "Exceeded number of supported router interface MAC profiles");
+	return -ENOBUFS;
+}
+
+static bool mlxsw_sp_is_offload_xstats_event(unsigned long event)
+{
+	switch (event) {
+	case NETDEV_OFFLOAD_XSTATS_ENABLE:
+	case NETDEV_OFFLOAD_XSTATS_DISABLE:
+	case NETDEV_OFFLOAD_XSTATS_REPORT_USED:
+	case NETDEV_OFFLOAD_XSTATS_REPORT_DELTA:
+		return true;
+	}
+
+	return false;
+}
+
+static int
+mlxsw_sp_router_port_offload_xstats_cmd(struct mlxsw_sp_rif *rif,
+					unsigned long event,
+					struct netdev_notifier_offload_xstats_info *info)
+{
+	switch (info->type) {
+	case NETDEV_OFFLOAD_XSTATS_TYPE_L3:
+		break;
+	default:
+		return 0;
+	}
+
+	switch (event) {
+	case NETDEV_OFFLOAD_XSTATS_ENABLE:
+		return mlxsw_sp_router_port_l3_stats_enable(rif);
+	case NETDEV_OFFLOAD_XSTATS_DISABLE:
+		mlxsw_sp_router_port_l3_stats_disable(rif);
+		return 0;
+	case NETDEV_OFFLOAD_XSTATS_REPORT_USED:
+		mlxsw_sp_router_port_l3_stats_report_used(rif, info);
+		return 0;
+	case NETDEV_OFFLOAD_XSTATS_REPORT_DELTA:
+		return mlxsw_sp_router_port_l3_stats_report_delta(rif, info);
+	}
+
+	WARN_ON_ONCE(1);
+	return 0;
+}
+
+static int
+mlxsw_sp_netdevice_offload_xstats_cmd(struct mlxsw_sp *mlxsw_sp,
+				      struct net_device *dev,
+				      unsigned long event,
+				      struct netdev_notifier_offload_xstats_info *info)
+{
+	struct mlxsw_sp_rif *rif;
+
+	rif = mlxsw_sp_rif_find_by_dev(mlxsw_sp, dev);
+	if (!rif)
+		return 0;
+
+	return mlxsw_sp_router_port_offload_xstats_cmd(rif, event, info);
+}
+
+static bool mlxsw_sp_is_router_event(unsigned long event)
+{
+	switch (event) {
+	case NETDEV_PRE_CHANGEADDR:
+	case NETDEV_CHANGEADDR:
+	case NETDEV_CHANGEMTU:
+		return true;
+	default:
+		return false;
+	}
+}
+
+static int mlxsw_sp_netdevice_router_port_event(struct net_device *dev,
+						unsigned long event, void *ptr)
+{
+	struct netlink_ext_ack *extack = netdev_notifier_info_to_extack(ptr);
+	struct mlxsw_sp *mlxsw_sp;
+	struct mlxsw_sp_rif *rif;
+
+	mlxsw_sp = mlxsw_sp_lower_get(dev);
+	if (!mlxsw_sp)
+		return 0;
+
+	rif = mlxsw_sp_rif_find_by_dev(mlxsw_sp, dev);
+	if (!rif)
+		return 0;
+
+	switch (event) {
+	case NETDEV_CHANGEMTU:
+	case NETDEV_CHANGEADDR:
+		return mlxsw_sp_router_port_change_event(mlxsw_sp, rif, extack);
+	case NETDEV_PRE_CHANGEADDR:
+		return mlxsw_sp_router_port_pre_changeaddr_event(rif, ptr);
+	default:
+		WARN_ON_ONCE(1);
+		break;
+	}
+
+	return 0;
+}
+
+static int mlxsw_sp_port_vrf_join(struct mlxsw_sp *mlxsw_sp,
+				  struct net_device *l3_dev,
+				  struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp_rif *rif;
+
+	/* If netdev is already associated with a RIF, then we need to
+	 * destroy it and create a new one with the new virtual router ID.
+	 */
+	rif = mlxsw_sp_rif_find_by_dev(mlxsw_sp, l3_dev);
+	if (rif)
+		__mlxsw_sp_inetaddr_event(mlxsw_sp, l3_dev, NETDEV_DOWN,
+					  extack);
+
+	return __mlxsw_sp_inetaddr_event(mlxsw_sp, l3_dev, NETDEV_UP, extack);
+}
+
+static void mlxsw_sp_port_vrf_leave(struct mlxsw_sp *mlxsw_sp,
+				    struct net_device *l3_dev)
+{
+	struct mlxsw_sp_rif *rif;
+
+	rif = mlxsw_sp_rif_find_by_dev(mlxsw_sp, l3_dev);
+	if (!rif)
+		return;
+	__mlxsw_sp_inetaddr_event(mlxsw_sp, l3_dev, NETDEV_DOWN, NULL);
+}
+
+static bool mlxsw_sp_is_vrf_event(unsigned long event, void *ptr)
+{
+	struct netdev_notifier_changeupper_info *info = ptr;
+
+	if (event != NETDEV_PRECHANGEUPPER && event != NETDEV_CHANGEUPPER)
+		return false;
+	return netif_is_l3_master(info->upper_dev);
+}
+
+static int
+mlxsw_sp_netdevice_vrf_event(struct net_device *l3_dev, unsigned long event,
+			     struct netdev_notifier_changeupper_info *info)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_lower_get(l3_dev);
+	int err = 0;
+
+	/* We do not create a RIF for a macvlan, but only use it to
+	 * direct more MAC addresses to the router.
+	 */
+	if (!mlxsw_sp || netif_is_macvlan(l3_dev))
+		return 0;
+
+	switch (event) {
+	case NETDEV_PRECHANGEUPPER:
+		break;
+	case NETDEV_CHANGEUPPER:
+		if (info->linking) {
+			struct netlink_ext_ack *extack;
+
+			extack = netdev_notifier_info_to_extack(&info->info);
+			err = mlxsw_sp_port_vrf_join(mlxsw_sp, l3_dev, extack);
+		} else {
+			mlxsw_sp_port_vrf_leave(mlxsw_sp, l3_dev);
+		}
+		break;
+	}
+
+	return err;
+}
+
+static int mlxsw_sp_router_netdevice_event(struct notifier_block *nb,
+					   unsigned long event, void *ptr)
+{
+	struct net_device *dev = netdev_notifier_info_to_dev(ptr);
+	struct mlxsw_sp_router *router;
+	struct mlxsw_sp *mlxsw_sp;
+	int err = 0;
+
+	router = container_of(nb, struct mlxsw_sp_router, netdevice_nb);
+	mlxsw_sp = router->mlxsw_sp;
+
+	mutex_lock(&mlxsw_sp->router->lock);
+
+	if (mlxsw_sp_is_offload_xstats_event(event))
+		err = mlxsw_sp_netdevice_offload_xstats_cmd(mlxsw_sp, dev,
+							    event, ptr);
+	else if (mlxsw_sp_netdev_is_ipip_ol(mlxsw_sp, dev))
+		err = mlxsw_sp_netdevice_ipip_ol_event(mlxsw_sp, dev,
+						       event, ptr);
+	else if (mlxsw_sp_netdev_is_ipip_ul(mlxsw_sp, dev))
+		err = mlxsw_sp_netdevice_ipip_ul_event(mlxsw_sp, dev,
+						       event, ptr);
+	else if (mlxsw_sp_is_router_event(event))
+		err = mlxsw_sp_netdevice_router_port_event(dev, event, ptr);
+	else if (mlxsw_sp_is_vrf_event(event, ptr))
+		err = mlxsw_sp_netdevice_vrf_event(dev, event, ptr);
+
+	mutex_unlock(&mlxsw_sp->router->lock);
+
+	return notifier_from_errno(err);
+}
+
+static int __mlxsw_sp_rif_macvlan_flush(struct net_device *dev,
+					struct netdev_nested_priv *priv)
+{
+	struct mlxsw_sp_rif *rif = (struct mlxsw_sp_rif *)priv->data;
+
+	if (!netif_is_macvlan(dev))
+		return 0;
+
+	return mlxsw_sp_rif_fdb_op(rif->mlxsw_sp, dev->dev_addr,
+				   mlxsw_sp_fid_index(rif->fid), false);
+}
+
+static int mlxsw_sp_rif_macvlan_flush(struct mlxsw_sp_rif *rif)
+{
+	struct netdev_nested_priv priv = {
+		.data = (void *)rif,
+	};
+
+	if (!netif_is_macvlan_port(rif->dev))
+		return 0;
+
+	netdev_warn(rif->dev, "Router interface is deleted. Upper macvlans will not work\n");
+	return netdev_walk_all_upper_dev_rcu(rif->dev,
+					     __mlxsw_sp_rif_macvlan_flush, &priv);
+}
+
+static void mlxsw_sp_rif_subport_setup(struct mlxsw_sp_rif *rif,
+				       const struct mlxsw_sp_rif_params *params)
+{
+	struct mlxsw_sp_rif_subport *rif_subport;
+
+	rif_subport = mlxsw_sp_rif_subport_rif(rif);
+	refcount_set(&rif_subport->ref_count, 1);
+	rif_subport->vid = params->vid;
+	rif_subport->lag = params->lag;
+	if (params->lag)
+		rif_subport->lag_id = params->lag_id;
+	else
+		rif_subport->system_port = params->system_port;
+}
+
+static int mlxsw_sp_rif_subport_op(struct mlxsw_sp_rif *rif, bool enable)
+{
+	struct mlxsw_sp *mlxsw_sp = rif->mlxsw_sp;
+	struct mlxsw_sp_rif_subport *rif_subport;
+	char ritr_pl[MLXSW_REG_RITR_LEN];
+	u16 efid;
+
+	rif_subport = mlxsw_sp_rif_subport_rif(rif);
+	mlxsw_reg_ritr_pack(ritr_pl, enable, MLXSW_REG_RITR_SP_IF,
+			    rif->rif_index, rif->vr_id, rif->dev->mtu);
+	mlxsw_reg_ritr_mac_pack(ritr_pl, rif->dev->dev_addr);
+	mlxsw_reg_ritr_if_mac_profile_id_set(ritr_pl, rif->mac_profile_id);
+	efid = mlxsw_sp_fid_index(rif->fid);
+	mlxsw_reg_ritr_sp_if_pack(ritr_pl, rif_subport->lag,
+				  rif_subport->lag ? rif_subport->lag_id :
+						     rif_subport->system_port,
+				  efid, 0);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(ritr), ritr_pl);
+}
+
+static int mlxsw_sp_rif_subport_configure(struct mlxsw_sp_rif *rif,
+					  struct netlink_ext_ack *extack)
+{
+	u8 mac_profile;
+	int err;
+
+	err = mlxsw_sp_rif_mac_profile_get(rif->mlxsw_sp, rif->addr,
+					   &mac_profile, extack);
+	if (err)
+		return err;
+	rif->mac_profile_id = mac_profile;
+
+	err = mlxsw_sp_rif_subport_op(rif, true);
+	if (err)
+		goto err_rif_subport_op;
+
+	err = mlxsw_sp_rif_fdb_op(rif->mlxsw_sp, rif->dev->dev_addr,
+				  mlxsw_sp_fid_index(rif->fid), true);
+	if (err)
+		goto err_rif_fdb_op;
+
+	err = mlxsw_sp_fid_rif_set(rif->fid, rif);
+	if (err)
+		goto err_fid_rif_set;
+
+	return 0;
+
+err_fid_rif_set:
+	mlxsw_sp_rif_fdb_op(rif->mlxsw_sp, rif->dev->dev_addr,
+			    mlxsw_sp_fid_index(rif->fid), false);
+err_rif_fdb_op:
+	mlxsw_sp_rif_subport_op(rif, false);
+err_rif_subport_op:
+	mlxsw_sp_rif_mac_profile_put(rif->mlxsw_sp, mac_profile);
+	return err;
+}
+
+static void mlxsw_sp_rif_subport_deconfigure(struct mlxsw_sp_rif *rif)
+{
+	struct mlxsw_sp_fid *fid = rif->fid;
+
+	mlxsw_sp_fid_rif_unset(fid);
+	mlxsw_sp_rif_fdb_op(rif->mlxsw_sp, rif->dev->dev_addr,
+			    mlxsw_sp_fid_index(fid), false);
+	mlxsw_sp_rif_macvlan_flush(rif);
+	mlxsw_sp_rif_subport_op(rif, false);
+	mlxsw_sp_rif_mac_profile_put(rif->mlxsw_sp, rif->mac_profile_id);
+}
+
+static struct mlxsw_sp_fid *
+mlxsw_sp_rif_subport_fid_get(struct mlxsw_sp_rif *rif,
+			     struct netlink_ext_ack *extack)
+{
+	return mlxsw_sp_fid_rfid_get(rif->mlxsw_sp, rif->rif_index);
+}
+
+static const struct mlxsw_sp_rif_ops mlxsw_sp_rif_subport_ops = {
+	.type			= MLXSW_SP_RIF_TYPE_SUBPORT,
+	.rif_size		= sizeof(struct mlxsw_sp_rif_subport),
+	.setup			= mlxsw_sp_rif_subport_setup,
+	.configure		= mlxsw_sp_rif_subport_configure,
+	.deconfigure		= mlxsw_sp_rif_subport_deconfigure,
+	.fid_get		= mlxsw_sp_rif_subport_fid_get,
+};
+
+static int mlxsw_sp_rif_fid_op(struct mlxsw_sp_rif *rif, u16 fid, bool enable)
+{
+	enum mlxsw_reg_ritr_if_type type = MLXSW_REG_RITR_FID_IF;
+	struct mlxsw_sp *mlxsw_sp = rif->mlxsw_sp;
+	char ritr_pl[MLXSW_REG_RITR_LEN];
+
+	mlxsw_reg_ritr_pack(ritr_pl, enable, type, rif->rif_index, rif->vr_id,
+			    rif->dev->mtu);
+	mlxsw_reg_ritr_mac_pack(ritr_pl, rif->dev->dev_addr);
+	mlxsw_reg_ritr_if_mac_profile_id_set(ritr_pl, rif->mac_profile_id);
+	mlxsw_reg_ritr_fid_if_fid_set(ritr_pl, fid);
+
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(ritr), ritr_pl);
+}
+
+u16 mlxsw_sp_router_port(const struct mlxsw_sp *mlxsw_sp)
+{
+	return mlxsw_core_max_ports(mlxsw_sp->core) + 1;
+}
+
+static int mlxsw_sp_rif_fid_configure(struct mlxsw_sp_rif *rif,
+				      struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp *mlxsw_sp = rif->mlxsw_sp;
+	u16 fid_index = mlxsw_sp_fid_index(rif->fid);
+	u8 mac_profile;
+	int err;
+
+	err = mlxsw_sp_rif_mac_profile_get(mlxsw_sp, rif->addr,
+					   &mac_profile, extack);
+	if (err)
+		return err;
+	rif->mac_profile_id = mac_profile;
+
+	err = mlxsw_sp_rif_fid_op(rif, fid_index, true);
+	if (err)
+		goto err_rif_fid_op;
+
+	err = mlxsw_sp_fid_flood_set(rif->fid, MLXSW_SP_FLOOD_TYPE_MC,
+				     mlxsw_sp_router_port(mlxsw_sp), true);
+	if (err)
+		goto err_fid_mc_flood_set;
+
+	err = mlxsw_sp_fid_flood_set(rif->fid, MLXSW_SP_FLOOD_TYPE_BC,
+				     mlxsw_sp_router_port(mlxsw_sp), true);
+	if (err)
+		goto err_fid_bc_flood_set;
+
+	err = mlxsw_sp_rif_fdb_op(rif->mlxsw_sp, rif->dev->dev_addr,
+				  mlxsw_sp_fid_index(rif->fid), true);
+	if (err)
+		goto err_rif_fdb_op;
+
+	err = mlxsw_sp_fid_rif_set(rif->fid, rif);
+	if (err)
+		goto err_fid_rif_set;
+
+	return 0;
+
+err_fid_rif_set:
+	mlxsw_sp_rif_fdb_op(rif->mlxsw_sp, rif->dev->dev_addr,
+			    mlxsw_sp_fid_index(rif->fid), false);
+err_rif_fdb_op:
+	mlxsw_sp_fid_flood_set(rif->fid, MLXSW_SP_FLOOD_TYPE_BC,
+			       mlxsw_sp_router_port(mlxsw_sp), false);
+err_fid_bc_flood_set:
+	mlxsw_sp_fid_flood_set(rif->fid, MLXSW_SP_FLOOD_TYPE_MC,
+			       mlxsw_sp_router_port(mlxsw_sp), false);
+err_fid_mc_flood_set:
+	mlxsw_sp_rif_fid_op(rif, fid_index, false);
+err_rif_fid_op:
+	mlxsw_sp_rif_mac_profile_put(mlxsw_sp, mac_profile);
+	return err;
+}
+
+static void mlxsw_sp_rif_fid_deconfigure(struct mlxsw_sp_rif *rif)
+{
+	u16 fid_index = mlxsw_sp_fid_index(rif->fid);
+	struct mlxsw_sp *mlxsw_sp = rif->mlxsw_sp;
+	struct mlxsw_sp_fid *fid = rif->fid;
+
+	mlxsw_sp_fid_rif_unset(fid);
+	mlxsw_sp_rif_fdb_op(rif->mlxsw_sp, rif->dev->dev_addr,
+			    mlxsw_sp_fid_index(fid), false);
+	mlxsw_sp_rif_macvlan_flush(rif);
+	mlxsw_sp_fid_flood_set(rif->fid, MLXSW_SP_FLOOD_TYPE_BC,
+			       mlxsw_sp_router_port(mlxsw_sp), false);
+	mlxsw_sp_fid_flood_set(rif->fid, MLXSW_SP_FLOOD_TYPE_MC,
+			       mlxsw_sp_router_port(mlxsw_sp), false);
+	mlxsw_sp_rif_fid_op(rif, fid_index, false);
+	mlxsw_sp_rif_mac_profile_put(rif->mlxsw_sp, rif->mac_profile_id);
+}
+
+static struct mlxsw_sp_fid *
+mlxsw_sp_rif_fid_fid_get(struct mlxsw_sp_rif *rif,
+			 struct netlink_ext_ack *extack)
+{
+	return mlxsw_sp_fid_8021d_get(rif->mlxsw_sp, rif->dev->ifindex);
+}
+
+static void mlxsw_sp_rif_fid_fdb_del(struct mlxsw_sp_rif *rif, const char *mac)
+{
+	struct switchdev_notifier_fdb_info info = {};
+	struct net_device *dev;
+
+	dev = br_fdb_find_port(rif->dev, mac, 0);
+	if (!dev)
+		return;
+
+	info.addr = mac;
+	info.vid = 0;
+	call_switchdev_notifiers(SWITCHDEV_FDB_DEL_TO_BRIDGE, dev, &info.info,
+				 NULL);
+}
+
+static const struct mlxsw_sp_rif_ops mlxsw_sp_rif_fid_ops = {
+	.type			= MLXSW_SP_RIF_TYPE_FID,
+	.rif_size		= sizeof(struct mlxsw_sp_rif),
+	.configure		= mlxsw_sp_rif_fid_configure,
+	.deconfigure		= mlxsw_sp_rif_fid_deconfigure,
+	.fid_get		= mlxsw_sp_rif_fid_fid_get,
+	.fdb_del		= mlxsw_sp_rif_fid_fdb_del,
+};
+
+static struct mlxsw_sp_fid *
+mlxsw_sp_rif_vlan_fid_get(struct mlxsw_sp_rif *rif,
+			  struct netlink_ext_ack *extack)
+{
+	struct net_device *br_dev;
+	u16 vid;
+	int err;
+
+	if (is_vlan_dev(rif->dev)) {
+		vid = vlan_dev_vlan_id(rif->dev);
+		br_dev = vlan_dev_real_dev(rif->dev);
+		if (WARN_ON(!netif_is_bridge_master(br_dev)))
+			return ERR_PTR(-EINVAL);
+	} else {
+		err = br_vlan_get_pvid(rif->dev, &vid);
+		if (err < 0 || !vid) {
+			NL_SET_ERR_MSG_MOD(extack, "Couldn't determine bridge PVID");
+			return ERR_PTR(-EINVAL);
+		}
+	}
+
+	return mlxsw_sp_fid_8021q_get(rif->mlxsw_sp, vid);
+}
+
+static void mlxsw_sp_rif_vlan_fdb_del(struct mlxsw_sp_rif *rif, const char *mac)
+{
+	struct switchdev_notifier_fdb_info info = {};
+	u16 vid = mlxsw_sp_fid_8021q_vid(rif->fid);
+	struct net_device *br_dev;
+	struct net_device *dev;
+
+	br_dev = is_vlan_dev(rif->dev) ? vlan_dev_real_dev(rif->dev) : rif->dev;
+	dev = br_fdb_find_port(br_dev, mac, vid);
+	if (!dev)
+		return;
+
+	info.addr = mac;
+	info.vid = vid;
+	call_switchdev_notifiers(SWITCHDEV_FDB_DEL_TO_BRIDGE, dev, &info.info,
+				 NULL);
+}
+
+static int mlxsw_sp_rif_vlan_op(struct mlxsw_sp_rif *rif, u16 vid, u16 efid,
+				bool enable)
+{
+	struct mlxsw_sp *mlxsw_sp = rif->mlxsw_sp;
+	char ritr_pl[MLXSW_REG_RITR_LEN];
+
+	mlxsw_reg_ritr_vlan_if_pack(ritr_pl, enable, rif->rif_index, rif->vr_id,
+				    rif->dev->mtu, rif->dev->dev_addr,
+				    rif->mac_profile_id, vid, efid);
+
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(ritr), ritr_pl);
+}
+
+static int mlxsw_sp_rif_vlan_configure(struct mlxsw_sp_rif *rif, u16 efid,
+				       struct netlink_ext_ack *extack)
+{
+	u16 vid = mlxsw_sp_fid_8021q_vid(rif->fid);
+	struct mlxsw_sp *mlxsw_sp = rif->mlxsw_sp;
+	u8 mac_profile;
+	int err;
+
+	err = mlxsw_sp_rif_mac_profile_get(mlxsw_sp, rif->addr,
+					   &mac_profile, extack);
+	if (err)
+		return err;
+	rif->mac_profile_id = mac_profile;
+
+	err = mlxsw_sp_rif_vlan_op(rif, vid, efid, true);
+	if (err)
+		goto err_rif_vlan_fid_op;
+
+	err = mlxsw_sp_fid_flood_set(rif->fid, MLXSW_SP_FLOOD_TYPE_MC,
+				     mlxsw_sp_router_port(mlxsw_sp), true);
+	if (err)
+		goto err_fid_mc_flood_set;
+
+	err = mlxsw_sp_fid_flood_set(rif->fid, MLXSW_SP_FLOOD_TYPE_BC,
+				     mlxsw_sp_router_port(mlxsw_sp), true);
+	if (err)
+		goto err_fid_bc_flood_set;
+
+	err = mlxsw_sp_rif_fdb_op(rif->mlxsw_sp, rif->dev->dev_addr,
+				  mlxsw_sp_fid_index(rif->fid), true);
+	if (err)
+		goto err_rif_fdb_op;
+
+	err = mlxsw_sp_fid_rif_set(rif->fid, rif);
+	if (err)
+		goto err_fid_rif_set;
+
+	return 0;
+
+err_fid_rif_set:
+	mlxsw_sp_rif_fdb_op(rif->mlxsw_sp, rif->dev->dev_addr,
+			    mlxsw_sp_fid_index(rif->fid), false);
+err_rif_fdb_op:
+	mlxsw_sp_fid_flood_set(rif->fid, MLXSW_SP_FLOOD_TYPE_BC,
+			       mlxsw_sp_router_port(mlxsw_sp), false);
+err_fid_bc_flood_set:
+	mlxsw_sp_fid_flood_set(rif->fid, MLXSW_SP_FLOOD_TYPE_MC,
+			       mlxsw_sp_router_port(mlxsw_sp), false);
+err_fid_mc_flood_set:
+	mlxsw_sp_rif_vlan_op(rif, vid, 0, false);
+err_rif_vlan_fid_op:
+	mlxsw_sp_rif_mac_profile_put(mlxsw_sp, mac_profile);
+	return err;
+}
+
+static void mlxsw_sp_rif_vlan_deconfigure(struct mlxsw_sp_rif *rif)
+{
+	u16 vid = mlxsw_sp_fid_8021q_vid(rif->fid);
+	struct mlxsw_sp *mlxsw_sp = rif->mlxsw_sp;
+
+	mlxsw_sp_fid_rif_unset(rif->fid);
+	mlxsw_sp_rif_fdb_op(rif->mlxsw_sp, rif->dev->dev_addr,
+			    mlxsw_sp_fid_index(rif->fid), false);
+	mlxsw_sp_rif_macvlan_flush(rif);
+	mlxsw_sp_fid_flood_set(rif->fid, MLXSW_SP_FLOOD_TYPE_BC,
+			       mlxsw_sp_router_port(mlxsw_sp), false);
+	mlxsw_sp_fid_flood_set(rif->fid, MLXSW_SP_FLOOD_TYPE_MC,
+			       mlxsw_sp_router_port(mlxsw_sp), false);
+	mlxsw_sp_rif_vlan_op(rif, vid, 0, false);
+	mlxsw_sp_rif_mac_profile_put(rif->mlxsw_sp, rif->mac_profile_id);
+}
+
+static int mlxsw_sp1_rif_vlan_configure(struct mlxsw_sp_rif *rif,
+					struct netlink_ext_ack *extack)
+{
+	return mlxsw_sp_rif_vlan_configure(rif, 0, extack);
+}
+
+static const struct mlxsw_sp_rif_ops mlxsw_sp1_rif_vlan_ops = {
+	.type			= MLXSW_SP_RIF_TYPE_VLAN,
+	.rif_size		= sizeof(struct mlxsw_sp_rif),
+	.configure		= mlxsw_sp1_rif_vlan_configure,
+	.deconfigure		= mlxsw_sp_rif_vlan_deconfigure,
+	.fid_get		= mlxsw_sp_rif_vlan_fid_get,
+	.fdb_del		= mlxsw_sp_rif_vlan_fdb_del,
+};
+
+static int mlxsw_sp2_rif_vlan_configure(struct mlxsw_sp_rif *rif,
+					struct netlink_ext_ack *extack)
+{
+	u16 efid = mlxsw_sp_fid_index(rif->fid);
+
+	return mlxsw_sp_rif_vlan_configure(rif, efid, extack);
+}
+
+static const struct mlxsw_sp_rif_ops mlxsw_sp2_rif_vlan_ops = {
+	.type			= MLXSW_SP_RIF_TYPE_VLAN,
+	.rif_size		= sizeof(struct mlxsw_sp_rif),
+	.configure		= mlxsw_sp2_rif_vlan_configure,
+	.deconfigure		= mlxsw_sp_rif_vlan_deconfigure,
+	.fid_get		= mlxsw_sp_rif_vlan_fid_get,
+	.fdb_del		= mlxsw_sp_rif_vlan_fdb_del,
+};
+
+static struct mlxsw_sp_rif_ipip_lb *
+mlxsw_sp_rif_ipip_lb_rif(struct mlxsw_sp_rif *rif)
+{
+	return container_of(rif, struct mlxsw_sp_rif_ipip_lb, common);
+}
+
+static void
+mlxsw_sp_rif_ipip_lb_setup(struct mlxsw_sp_rif *rif,
+			   const struct mlxsw_sp_rif_params *params)
+{
+	struct mlxsw_sp_rif_params_ipip_lb *params_lb;
+	struct mlxsw_sp_rif_ipip_lb *rif_lb;
+
+	params_lb = container_of(params, struct mlxsw_sp_rif_params_ipip_lb,
+				 common);
+	rif_lb = mlxsw_sp_rif_ipip_lb_rif(rif);
+	rif_lb->lb_config = params_lb->lb_config;
+}
+
+static int
+mlxsw_sp1_rif_ipip_lb_configure(struct mlxsw_sp_rif *rif,
+				struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp_rif_ipip_lb *lb_rif = mlxsw_sp_rif_ipip_lb_rif(rif);
+	u32 ul_tb_id = mlxsw_sp_ipip_dev_ul_tb_id(rif->dev);
+	struct mlxsw_sp *mlxsw_sp = rif->mlxsw_sp;
+	struct mlxsw_sp_vr *ul_vr;
+	int err;
+
+	ul_vr = mlxsw_sp_vr_get(mlxsw_sp, ul_tb_id, NULL);
+	if (IS_ERR(ul_vr))
+		return PTR_ERR(ul_vr);
+
+	err = mlxsw_sp_rif_ipip_lb_op(lb_rif, ul_vr->id, 0, true);
+	if (err)
+		goto err_loopback_op;
+
+	lb_rif->ul_vr_id = ul_vr->id;
+	lb_rif->ul_rif_id = 0;
+	++ul_vr->rif_count;
+	return 0;
+
+err_loopback_op:
+	mlxsw_sp_vr_put(mlxsw_sp, ul_vr);
+	return err;
+}
+
+static void mlxsw_sp1_rif_ipip_lb_deconfigure(struct mlxsw_sp_rif *rif)
+{
+	struct mlxsw_sp_rif_ipip_lb *lb_rif = mlxsw_sp_rif_ipip_lb_rif(rif);
+	struct mlxsw_sp *mlxsw_sp = rif->mlxsw_sp;
+	struct mlxsw_sp_vr *ul_vr;
+
+	ul_vr = &mlxsw_sp->router->vrs[lb_rif->ul_vr_id];
+	mlxsw_sp_rif_ipip_lb_op(lb_rif, ul_vr->id, 0, false);
+
+	--ul_vr->rif_count;
+	mlxsw_sp_vr_put(mlxsw_sp, ul_vr);
+}
+
+static const struct mlxsw_sp_rif_ops mlxsw_sp1_rif_ipip_lb_ops = {
+	.type			= MLXSW_SP_RIF_TYPE_IPIP_LB,
+	.rif_size		= sizeof(struct mlxsw_sp_rif_ipip_lb),
+	.setup                  = mlxsw_sp_rif_ipip_lb_setup,
+	.configure		= mlxsw_sp1_rif_ipip_lb_configure,
+	.deconfigure		= mlxsw_sp1_rif_ipip_lb_deconfigure,
+};
+
+static const struct mlxsw_sp_rif_ops *mlxsw_sp1_rif_ops_arr[] = {
+	[MLXSW_SP_RIF_TYPE_SUBPORT]	= &mlxsw_sp_rif_subport_ops,
+	[MLXSW_SP_RIF_TYPE_VLAN]	= &mlxsw_sp1_rif_vlan_ops,
+	[MLXSW_SP_RIF_TYPE_FID]		= &mlxsw_sp_rif_fid_ops,
+	[MLXSW_SP_RIF_TYPE_IPIP_LB]	= &mlxsw_sp1_rif_ipip_lb_ops,
+};
+
+static int
+mlxsw_sp_rif_ipip_lb_ul_rif_op(struct mlxsw_sp_rif *ul_rif, bool enable)
+{
+	struct mlxsw_sp *mlxsw_sp = ul_rif->mlxsw_sp;
+	char ritr_pl[MLXSW_REG_RITR_LEN];
+
+	mlxsw_reg_ritr_pack(ritr_pl, enable, MLXSW_REG_RITR_LOOPBACK_IF,
+			    ul_rif->rif_index, ul_rif->vr_id, IP_MAX_MTU);
+	mlxsw_reg_ritr_loopback_protocol_set(ritr_pl,
+					     MLXSW_REG_RITR_LOOPBACK_GENERIC);
+
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(ritr), ritr_pl);
+}
+
+static struct mlxsw_sp_rif *
+mlxsw_sp_ul_rif_create(struct mlxsw_sp *mlxsw_sp, struct mlxsw_sp_vr *vr,
+		       struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp_rif *ul_rif;
+	u16 rif_index;
+	int err;
+
+	err = mlxsw_sp_rif_index_alloc(mlxsw_sp, &rif_index);
+	if (err) {
+		NL_SET_ERR_MSG_MOD(extack, "Exceeded number of supported router interfaces");
+		return ERR_PTR(err);
+	}
+
+	ul_rif = mlxsw_sp_rif_alloc(sizeof(*ul_rif), rif_index, vr->id, NULL);
+	if (!ul_rif)
+		return ERR_PTR(-ENOMEM);
+
+	mlxsw_sp->router->rifs[rif_index] = ul_rif;
+	ul_rif->mlxsw_sp = mlxsw_sp;
+	err = mlxsw_sp_rif_ipip_lb_ul_rif_op(ul_rif, true);
+	if (err)
+		goto ul_rif_op_err;
+
+	atomic_inc(&mlxsw_sp->router->rifs_count);
+	return ul_rif;
+
+ul_rif_op_err:
+	mlxsw_sp->router->rifs[rif_index] = NULL;
+	kfree(ul_rif);
+	return ERR_PTR(err);
+}
+
+static void mlxsw_sp_ul_rif_destroy(struct mlxsw_sp_rif *ul_rif)
+{
+	struct mlxsw_sp *mlxsw_sp = ul_rif->mlxsw_sp;
+
+	atomic_dec(&mlxsw_sp->router->rifs_count);
+	mlxsw_sp_rif_ipip_lb_ul_rif_op(ul_rif, false);
+	mlxsw_sp->router->rifs[ul_rif->rif_index] = NULL;
+	kfree(ul_rif);
+}
+
+static struct mlxsw_sp_rif *
+mlxsw_sp_ul_rif_get(struct mlxsw_sp *mlxsw_sp, u32 tb_id,
+		    struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp_vr *vr;
+	int err;
+
+	vr = mlxsw_sp_vr_get(mlxsw_sp, tb_id, extack);
+	if (IS_ERR(vr))
+		return ERR_CAST(vr);
+
+	if (refcount_inc_not_zero(&vr->ul_rif_refcnt))
+		return vr->ul_rif;
+
+	vr->ul_rif = mlxsw_sp_ul_rif_create(mlxsw_sp, vr, extack);
+	if (IS_ERR(vr->ul_rif)) {
+		err = PTR_ERR(vr->ul_rif);
+		goto err_ul_rif_create;
+	}
+
+	vr->rif_count++;
+	refcount_set(&vr->ul_rif_refcnt, 1);
+
+	return vr->ul_rif;
+
+err_ul_rif_create:
+	mlxsw_sp_vr_put(mlxsw_sp, vr);
+	return ERR_PTR(err);
+}
+
+static void mlxsw_sp_ul_rif_put(struct mlxsw_sp_rif *ul_rif)
+{
+	struct mlxsw_sp *mlxsw_sp = ul_rif->mlxsw_sp;
+	struct mlxsw_sp_vr *vr;
+
+	vr = &mlxsw_sp->router->vrs[ul_rif->vr_id];
+
+	if (!refcount_dec_and_test(&vr->ul_rif_refcnt))
+		return;
+
+	vr->rif_count--;
+	mlxsw_sp_ul_rif_destroy(ul_rif);
+	mlxsw_sp_vr_put(mlxsw_sp, vr);
+}
+
+int mlxsw_sp_router_ul_rif_get(struct mlxsw_sp *mlxsw_sp, u32 ul_tb_id,
+			       u16 *ul_rif_index)
+{
+	struct mlxsw_sp_rif *ul_rif;
+	int err = 0;
+
+	mutex_lock(&mlxsw_sp->router->lock);
+	ul_rif = mlxsw_sp_ul_rif_get(mlxsw_sp, ul_tb_id, NULL);
+	if (IS_ERR(ul_rif)) {
+		err = PTR_ERR(ul_rif);
+		goto out;
+	}
+	*ul_rif_index = ul_rif->rif_index;
+out:
+	mutex_unlock(&mlxsw_sp->router->lock);
+	return err;
+}
+
+void mlxsw_sp_router_ul_rif_put(struct mlxsw_sp *mlxsw_sp, u16 ul_rif_index)
+{
+	struct mlxsw_sp_rif *ul_rif;
+
+	mutex_lock(&mlxsw_sp->router->lock);
+	ul_rif = mlxsw_sp->router->rifs[ul_rif_index];
+	if (WARN_ON(!ul_rif))
+		goto out;
+
+	mlxsw_sp_ul_rif_put(ul_rif);
+out:
+	mutex_unlock(&mlxsw_sp->router->lock);
+}
+
+static int
+mlxsw_sp2_rif_ipip_lb_configure(struct mlxsw_sp_rif *rif,
+				struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp_rif_ipip_lb *lb_rif = mlxsw_sp_rif_ipip_lb_rif(rif);
+	u32 ul_tb_id = mlxsw_sp_ipip_dev_ul_tb_id(rif->dev);
+	struct mlxsw_sp *mlxsw_sp = rif->mlxsw_sp;
+	struct mlxsw_sp_rif *ul_rif;
+	int err;
+
+	ul_rif = mlxsw_sp_ul_rif_get(mlxsw_sp, ul_tb_id, NULL);
+	if (IS_ERR(ul_rif))
+		return PTR_ERR(ul_rif);
+
+	err = mlxsw_sp_rif_ipip_lb_op(lb_rif, 0, ul_rif->rif_index, true);
+	if (err)
+		goto err_loopback_op;
+
+	lb_rif->ul_vr_id = 0;
+	lb_rif->ul_rif_id = ul_rif->rif_index;
+
+	return 0;
+
+err_loopback_op:
+	mlxsw_sp_ul_rif_put(ul_rif);
+	return err;
+}
+
+static void mlxsw_sp2_rif_ipip_lb_deconfigure(struct mlxsw_sp_rif *rif)
+{
+	struct mlxsw_sp_rif_ipip_lb *lb_rif = mlxsw_sp_rif_ipip_lb_rif(rif);
+	struct mlxsw_sp *mlxsw_sp = rif->mlxsw_sp;
+	struct mlxsw_sp_rif *ul_rif;
+
+	ul_rif = mlxsw_sp_rif_by_index(mlxsw_sp, lb_rif->ul_rif_id);
+	mlxsw_sp_rif_ipip_lb_op(lb_rif, 0, lb_rif->ul_rif_id, false);
+	mlxsw_sp_ul_rif_put(ul_rif);
+}
+
+static const struct mlxsw_sp_rif_ops mlxsw_sp2_rif_ipip_lb_ops = {
+	.type			= MLXSW_SP_RIF_TYPE_IPIP_LB,
+	.rif_size		= sizeof(struct mlxsw_sp_rif_ipip_lb),
+	.setup                  = mlxsw_sp_rif_ipip_lb_setup,
+	.configure		= mlxsw_sp2_rif_ipip_lb_configure,
+	.deconfigure		= mlxsw_sp2_rif_ipip_lb_deconfigure,
+};
+
+static const struct mlxsw_sp_rif_ops *mlxsw_sp2_rif_ops_arr[] = {
+	[MLXSW_SP_RIF_TYPE_SUBPORT]	= &mlxsw_sp_rif_subport_ops,
+	[MLXSW_SP_RIF_TYPE_VLAN]	= &mlxsw_sp2_rif_vlan_ops,
+	[MLXSW_SP_RIF_TYPE_FID]		= &mlxsw_sp_rif_fid_ops,
+	[MLXSW_SP_RIF_TYPE_IPIP_LB]	= &mlxsw_sp2_rif_ipip_lb_ops,
+};
+
+static int mlxsw_sp_rifs_init(struct mlxsw_sp *mlxsw_sp)
+{
+	u64 max_rifs = MLXSW_CORE_RES_GET(mlxsw_sp->core, MAX_RIFS);
+	struct devlink *devlink = priv_to_devlink(mlxsw_sp->core);
+	struct mlxsw_core *core = mlxsw_sp->core;
+
+	if (!MLXSW_CORE_RES_VALID(core, MAX_RIF_MAC_PROFILES))
+		return -EIO;
+	mlxsw_sp->router->max_rif_mac_profile =
+		MLXSW_CORE_RES_GET(core, MAX_RIF_MAC_PROFILES);
+
+	mlxsw_sp->router->rifs = kcalloc(max_rifs,
+					 sizeof(struct mlxsw_sp_rif *),
+					 GFP_KERNEL);
+	if (!mlxsw_sp->router->rifs)
+		return -ENOMEM;
+
+	idr_init(&mlxsw_sp->router->rif_mac_profiles_idr);
+	atomic_set(&mlxsw_sp->router->rif_mac_profiles_count, 0);
+	atomic_set(&mlxsw_sp->router->rifs_count, 0);
+	devl_resource_occ_get_register(devlink,
+				       MLXSW_SP_RESOURCE_RIF_MAC_PROFILES,
+				       mlxsw_sp_rif_mac_profiles_occ_get,
+				       mlxsw_sp);
+	devl_resource_occ_get_register(devlink,
+				       MLXSW_SP_RESOURCE_RIFS,
+				       mlxsw_sp_rifs_occ_get,
+				       mlxsw_sp);
+
+	return 0;
+}
+
+static void mlxsw_sp_rifs_fini(struct mlxsw_sp *mlxsw_sp)
+{
+	struct devlink *devlink = priv_to_devlink(mlxsw_sp->core);
+	int i;
+
+	WARN_ON_ONCE(atomic_read(&mlxsw_sp->router->rifs_count));
+	for (i = 0; i < MLXSW_CORE_RES_GET(mlxsw_sp->core, MAX_RIFS); i++)
+		WARN_ON_ONCE(mlxsw_sp->router->rifs[i]);
+
+	devl_resource_occ_get_unregister(devlink, MLXSW_SP_RESOURCE_RIFS);
+	devl_resource_occ_get_unregister(devlink,
+					 MLXSW_SP_RESOURCE_RIF_MAC_PROFILES);
+	WARN_ON(!idr_is_empty(&mlxsw_sp->router->rif_mac_profiles_idr));
+	idr_destroy(&mlxsw_sp->router->rif_mac_profiles_idr);
+	kfree(mlxsw_sp->router->rifs);
+}
+
+static int
+mlxsw_sp_ipip_config_tigcr(struct mlxsw_sp *mlxsw_sp)
+{
+	char tigcr_pl[MLXSW_REG_TIGCR_LEN];
+
+	mlxsw_reg_tigcr_pack(tigcr_pl, true, 0);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(tigcr), tigcr_pl);
+}
+
+static int mlxsw_sp_ipips_init(struct mlxsw_sp *mlxsw_sp)
+{
+	int err;
+
+	INIT_LIST_HEAD(&mlxsw_sp->router->ipip_list);
+
+	err = mlxsw_sp_ipip_ecn_encap_init(mlxsw_sp);
+	if (err)
+		return err;
+	err = mlxsw_sp_ipip_ecn_decap_init(mlxsw_sp);
+	if (err)
+		return err;
+
+	return mlxsw_sp_ipip_config_tigcr(mlxsw_sp);
+}
+
+static int mlxsw_sp1_ipips_init(struct mlxsw_sp *mlxsw_sp)
+{
+	mlxsw_sp->router->ipip_ops_arr = mlxsw_sp1_ipip_ops_arr;
+	return mlxsw_sp_ipips_init(mlxsw_sp);
+}
+
+static int mlxsw_sp2_ipips_init(struct mlxsw_sp *mlxsw_sp)
+{
+	mlxsw_sp->router->ipip_ops_arr = mlxsw_sp2_ipip_ops_arr;
+	return mlxsw_sp_ipips_init(mlxsw_sp);
+}
+
+static void mlxsw_sp_ipips_fini(struct mlxsw_sp *mlxsw_sp)
+{
+	WARN_ON(!list_empty(&mlxsw_sp->router->ipip_list));
+}
+
+static void mlxsw_sp_router_fib_dump_flush(struct notifier_block *nb)
+{
+	struct mlxsw_sp_router *router;
+
+	/* Flush pending FIB notifications and then flush the device's
+	 * table before requesting another dump. The FIB notification
+	 * block is unregistered, so no need to take RTNL.
+	 */
+	mlxsw_core_flush_owq();
+	router = container_of(nb, struct mlxsw_sp_router, fib_nb);
+	mlxsw_sp_router_fib_flush(router->mlxsw_sp);
+}
+
+#ifdef CONFIG_IP_ROUTE_MULTIPATH
+struct mlxsw_sp_mp_hash_config {
+	DECLARE_BITMAP(headers, __MLXSW_REG_RECR2_HEADER_CNT);
+	DECLARE_BITMAP(fields, __MLXSW_REG_RECR2_FIELD_CNT);
+	DECLARE_BITMAP(inner_headers, __MLXSW_REG_RECR2_HEADER_CNT);
+	DECLARE_BITMAP(inner_fields, __MLXSW_REG_RECR2_INNER_FIELD_CNT);
+	bool inc_parsing_depth;
+};
+
+#define MLXSW_SP_MP_HASH_HEADER_SET(_headers, _header) \
+	bitmap_set(_headers, MLXSW_REG_RECR2_##_header, 1)
+
+#define MLXSW_SP_MP_HASH_FIELD_SET(_fields, _field) \
+	bitmap_set(_fields, MLXSW_REG_RECR2_##_field, 1)
+
+#define MLXSW_SP_MP_HASH_FIELD_RANGE_SET(_fields, _field, _nr) \
+	bitmap_set(_fields, MLXSW_REG_RECR2_##_field, _nr)
+
+static void mlxsw_sp_mp_hash_inner_l3(struct mlxsw_sp_mp_hash_config *config)
+{
+	unsigned long *inner_headers = config->inner_headers;
+	unsigned long *inner_fields = config->inner_fields;
+
+	/* IPv4 inner */
+	MLXSW_SP_MP_HASH_HEADER_SET(inner_headers, IPV4_EN_NOT_TCP_NOT_UDP);
+	MLXSW_SP_MP_HASH_HEADER_SET(inner_headers, IPV4_EN_TCP_UDP);
+	MLXSW_SP_MP_HASH_FIELD_RANGE_SET(inner_fields, INNER_IPV4_SIP0, 4);
+	MLXSW_SP_MP_HASH_FIELD_RANGE_SET(inner_fields, INNER_IPV4_DIP0, 4);
+	/* IPv6 inner */
+	MLXSW_SP_MP_HASH_HEADER_SET(inner_headers, IPV6_EN_NOT_TCP_NOT_UDP);
+	MLXSW_SP_MP_HASH_HEADER_SET(inner_headers, IPV6_EN_TCP_UDP);
+	MLXSW_SP_MP_HASH_FIELD_SET(inner_fields, INNER_IPV6_SIP0_7);
+	MLXSW_SP_MP_HASH_FIELD_RANGE_SET(inner_fields, INNER_IPV6_SIP8, 8);
+	MLXSW_SP_MP_HASH_FIELD_SET(inner_fields, INNER_IPV6_DIP0_7);
+	MLXSW_SP_MP_HASH_FIELD_RANGE_SET(inner_fields, INNER_IPV6_DIP8, 8);
+	MLXSW_SP_MP_HASH_FIELD_SET(inner_fields, INNER_IPV6_NEXT_HEADER);
+	MLXSW_SP_MP_HASH_FIELD_SET(inner_fields, INNER_IPV6_FLOW_LABEL);
+}
+
+static void mlxsw_sp_mp4_hash_outer_addr(struct mlxsw_sp_mp_hash_config *config)
+{
+	unsigned long *headers = config->headers;
+	unsigned long *fields = config->fields;
+
+	MLXSW_SP_MP_HASH_HEADER_SET(headers, IPV4_EN_NOT_TCP_NOT_UDP);
+	MLXSW_SP_MP_HASH_HEADER_SET(headers, IPV4_EN_TCP_UDP);
+	MLXSW_SP_MP_HASH_FIELD_RANGE_SET(fields, IPV4_SIP0, 4);
+	MLXSW_SP_MP_HASH_FIELD_RANGE_SET(fields, IPV4_DIP0, 4);
+}
+
+static void
+mlxsw_sp_mp_hash_inner_custom(struct mlxsw_sp_mp_hash_config *config,
+			      u32 hash_fields)
+{
+	unsigned long *inner_headers = config->inner_headers;
+	unsigned long *inner_fields = config->inner_fields;
+
+	/* IPv4 Inner */
+	MLXSW_SP_MP_HASH_HEADER_SET(inner_headers, IPV4_EN_NOT_TCP_NOT_UDP);
+	MLXSW_SP_MP_HASH_HEADER_SET(inner_headers, IPV4_EN_TCP_UDP);
+	if (hash_fields & FIB_MULTIPATH_HASH_FIELD_INNER_SRC_IP)
+		MLXSW_SP_MP_HASH_FIELD_RANGE_SET(inner_fields, INNER_IPV4_SIP0, 4);
+	if (hash_fields & FIB_MULTIPATH_HASH_FIELD_INNER_DST_IP)
+		MLXSW_SP_MP_HASH_FIELD_RANGE_SET(inner_fields, INNER_IPV4_DIP0, 4);
+	if (hash_fields & FIB_MULTIPATH_HASH_FIELD_INNER_IP_PROTO)
+		MLXSW_SP_MP_HASH_FIELD_SET(inner_fields, INNER_IPV4_PROTOCOL);
+	/* IPv6 inner */
+	MLXSW_SP_MP_HASH_HEADER_SET(inner_headers, IPV6_EN_NOT_TCP_NOT_UDP);
+	MLXSW_SP_MP_HASH_HEADER_SET(inner_headers, IPV6_EN_TCP_UDP);
+	if (hash_fields & FIB_MULTIPATH_HASH_FIELD_INNER_SRC_IP) {
+		MLXSW_SP_MP_HASH_FIELD_SET(inner_fields, INNER_IPV6_SIP0_7);
+		MLXSW_SP_MP_HASH_FIELD_RANGE_SET(inner_fields, INNER_IPV6_SIP8, 8);
+	}
+	if (hash_fields & FIB_MULTIPATH_HASH_FIELD_INNER_DST_IP) {
+		MLXSW_SP_MP_HASH_FIELD_SET(inner_fields, INNER_IPV6_DIP0_7);
+		MLXSW_SP_MP_HASH_FIELD_RANGE_SET(inner_fields, INNER_IPV6_DIP8, 8);
+	}
+	if (hash_fields & FIB_MULTIPATH_HASH_FIELD_INNER_IP_PROTO)
+		MLXSW_SP_MP_HASH_FIELD_SET(inner_fields, INNER_IPV6_NEXT_HEADER);
+	if (hash_fields & FIB_MULTIPATH_HASH_FIELD_INNER_FLOWLABEL)
+		MLXSW_SP_MP_HASH_FIELD_SET(inner_fields, INNER_IPV6_FLOW_LABEL);
+	/* L4 inner */
+	MLXSW_SP_MP_HASH_HEADER_SET(inner_headers, TCP_UDP_EN_IPV4);
+	MLXSW_SP_MP_HASH_HEADER_SET(inner_headers, TCP_UDP_EN_IPV6);
+	if (hash_fields & FIB_MULTIPATH_HASH_FIELD_INNER_SRC_PORT)
+		MLXSW_SP_MP_HASH_FIELD_SET(inner_fields, INNER_TCP_UDP_SPORT);
+	if (hash_fields & FIB_MULTIPATH_HASH_FIELD_INNER_DST_PORT)
+		MLXSW_SP_MP_HASH_FIELD_SET(inner_fields, INNER_TCP_UDP_DPORT);
+}
+
+static void mlxsw_sp_mp4_hash_init(struct mlxsw_sp *mlxsw_sp,
+				   struct mlxsw_sp_mp_hash_config *config)
+{
+	struct net *net = mlxsw_sp_net(mlxsw_sp);
+	unsigned long *headers = config->headers;
+	unsigned long *fields = config->fields;
+	u32 hash_fields;
+
+	switch (READ_ONCE(net->ipv4.sysctl_fib_multipath_hash_policy)) {
+	case 0:
+		mlxsw_sp_mp4_hash_outer_addr(config);
+		break;
+	case 1:
+		mlxsw_sp_mp4_hash_outer_addr(config);
+		MLXSW_SP_MP_HASH_HEADER_SET(headers, TCP_UDP_EN_IPV4);
+		MLXSW_SP_MP_HASH_FIELD_SET(fields, IPV4_PROTOCOL);
+		MLXSW_SP_MP_HASH_FIELD_SET(fields, TCP_UDP_SPORT);
+		MLXSW_SP_MP_HASH_FIELD_SET(fields, TCP_UDP_DPORT);
+		break;
+	case 2:
+		/* Outer */
+		mlxsw_sp_mp4_hash_outer_addr(config);
+		/* Inner */
+		mlxsw_sp_mp_hash_inner_l3(config);
+		break;
+	case 3:
+		hash_fields = READ_ONCE(net->ipv4.sysctl_fib_multipath_hash_fields);
+		/* Outer */
+		MLXSW_SP_MP_HASH_HEADER_SET(headers, IPV4_EN_NOT_TCP_NOT_UDP);
+		MLXSW_SP_MP_HASH_HEADER_SET(headers, IPV4_EN_TCP_UDP);
+		MLXSW_SP_MP_HASH_HEADER_SET(headers, TCP_UDP_EN_IPV4);
+		if (hash_fields & FIB_MULTIPATH_HASH_FIELD_SRC_IP)
+			MLXSW_SP_MP_HASH_FIELD_RANGE_SET(fields, IPV4_SIP0, 4);
+		if (hash_fields & FIB_MULTIPATH_HASH_FIELD_DST_IP)
+			MLXSW_SP_MP_HASH_FIELD_RANGE_SET(fields, IPV4_DIP0, 4);
+		if (hash_fields & FIB_MULTIPATH_HASH_FIELD_IP_PROTO)
+			MLXSW_SP_MP_HASH_FIELD_SET(fields, IPV4_PROTOCOL);
+		if (hash_fields & FIB_MULTIPATH_HASH_FIELD_SRC_PORT)
+			MLXSW_SP_MP_HASH_FIELD_SET(fields, TCP_UDP_SPORT);
+		if (hash_fields & FIB_MULTIPATH_HASH_FIELD_DST_PORT)
+			MLXSW_SP_MP_HASH_FIELD_SET(fields, TCP_UDP_DPORT);
+		/* Inner */
+		mlxsw_sp_mp_hash_inner_custom(config, hash_fields);
+		break;
+	}
+}
+
+static void mlxsw_sp_mp6_hash_outer_addr(struct mlxsw_sp_mp_hash_config *config)
+{
+	unsigned long *headers = config->headers;
+	unsigned long *fields = config->fields;
+
+	MLXSW_SP_MP_HASH_HEADER_SET(headers, IPV6_EN_NOT_TCP_NOT_UDP);
+	MLXSW_SP_MP_HASH_HEADER_SET(headers, IPV6_EN_TCP_UDP);
+	MLXSW_SP_MP_HASH_FIELD_SET(fields, IPV6_SIP0_7);
+	MLXSW_SP_MP_HASH_FIELD_RANGE_SET(fields, IPV6_SIP8, 8);
+	MLXSW_SP_MP_HASH_FIELD_SET(fields, IPV6_DIP0_7);
+	MLXSW_SP_MP_HASH_FIELD_RANGE_SET(fields, IPV6_DIP8, 8);
+}
+
+static void mlxsw_sp_mp6_hash_init(struct mlxsw_sp *mlxsw_sp,
+				   struct mlxsw_sp_mp_hash_config *config)
+{
+	u32 hash_fields = ip6_multipath_hash_fields(mlxsw_sp_net(mlxsw_sp));
+	unsigned long *headers = config->headers;
+	unsigned long *fields = config->fields;
+
+	switch (ip6_multipath_hash_policy(mlxsw_sp_net(mlxsw_sp))) {
+	case 0:
+		mlxsw_sp_mp6_hash_outer_addr(config);
+		MLXSW_SP_MP_HASH_FIELD_SET(fields, IPV6_NEXT_HEADER);
+		MLXSW_SP_MP_HASH_FIELD_SET(fields, IPV6_FLOW_LABEL);
+		break;
+	case 1:
+		mlxsw_sp_mp6_hash_outer_addr(config);
+		MLXSW_SP_MP_HASH_HEADER_SET(headers, TCP_UDP_EN_IPV6);
+		MLXSW_SP_MP_HASH_FIELD_SET(fields, IPV6_NEXT_HEADER);
+		MLXSW_SP_MP_HASH_FIELD_SET(fields, TCP_UDP_SPORT);
+		MLXSW_SP_MP_HASH_FIELD_SET(fields, TCP_UDP_DPORT);
+		break;
+	case 2:
+		/* Outer */
+		mlxsw_sp_mp6_hash_outer_addr(config);
+		MLXSW_SP_MP_HASH_FIELD_SET(fields, IPV6_NEXT_HEADER);
+		MLXSW_SP_MP_HASH_FIELD_SET(fields, IPV6_FLOW_LABEL);
+		/* Inner */
+		mlxsw_sp_mp_hash_inner_l3(config);
+		config->inc_parsing_depth = true;
+		break;
+	case 3:
+		/* Outer */
+		MLXSW_SP_MP_HASH_HEADER_SET(headers, IPV6_EN_NOT_TCP_NOT_UDP);
+		MLXSW_SP_MP_HASH_HEADER_SET(headers, IPV6_EN_TCP_UDP);
+		MLXSW_SP_MP_HASH_HEADER_SET(headers, TCP_UDP_EN_IPV6);
+		if (hash_fields & FIB_MULTIPATH_HASH_FIELD_SRC_IP) {
+			MLXSW_SP_MP_HASH_FIELD_SET(fields, IPV6_SIP0_7);
+			MLXSW_SP_MP_HASH_FIELD_RANGE_SET(fields, IPV6_SIP8, 8);
+		}
+		if (hash_fields & FIB_MULTIPATH_HASH_FIELD_DST_IP) {
+			MLXSW_SP_MP_HASH_FIELD_SET(fields, IPV6_DIP0_7);
+			MLXSW_SP_MP_HASH_FIELD_RANGE_SET(fields, IPV6_DIP8, 8);
+		}
+		if (hash_fields & FIB_MULTIPATH_HASH_FIELD_IP_PROTO)
+			MLXSW_SP_MP_HASH_FIELD_SET(fields, IPV6_NEXT_HEADER);
+		if (hash_fields & FIB_MULTIPATH_HASH_FIELD_FLOWLABEL)
+			MLXSW_SP_MP_HASH_FIELD_SET(fields, IPV6_FLOW_LABEL);
+		if (hash_fields & FIB_MULTIPATH_HASH_FIELD_SRC_PORT)
+			MLXSW_SP_MP_HASH_FIELD_SET(fields, TCP_UDP_SPORT);
+		if (hash_fields & FIB_MULTIPATH_HASH_FIELD_DST_PORT)
+			MLXSW_SP_MP_HASH_FIELD_SET(fields, TCP_UDP_DPORT);
+		/* Inner */
+		mlxsw_sp_mp_hash_inner_custom(config, hash_fields);
+		if (hash_fields & FIB_MULTIPATH_HASH_FIELD_INNER_MASK)
+			config->inc_parsing_depth = true;
+		break;
+	}
+}
+
+static int mlxsw_sp_mp_hash_parsing_depth_adjust(struct mlxsw_sp *mlxsw_sp,
+						 bool old_inc_parsing_depth,
+						 bool new_inc_parsing_depth)
+{
+	int err;
+
+	if (!old_inc_parsing_depth && new_inc_parsing_depth) {
+		err = mlxsw_sp_parsing_depth_inc(mlxsw_sp);
+		if (err)
+			return err;
+		mlxsw_sp->router->inc_parsing_depth = true;
+	} else if (old_inc_parsing_depth && !new_inc_parsing_depth) {
+		mlxsw_sp_parsing_depth_dec(mlxsw_sp);
+		mlxsw_sp->router->inc_parsing_depth = false;
+	}
+
+	return 0;
+}
+
+static int mlxsw_sp_mp_hash_init(struct mlxsw_sp *mlxsw_sp)
+{
+	bool old_inc_parsing_depth, new_inc_parsing_depth;
+	struct mlxsw_sp_mp_hash_config config = {};
+	char recr2_pl[MLXSW_REG_RECR2_LEN];
+	unsigned long bit;
+	u32 seed;
+	int err;
+
+	seed = jhash(mlxsw_sp->base_mac, sizeof(mlxsw_sp->base_mac), 0);
+	mlxsw_reg_recr2_pack(recr2_pl, seed);
+	mlxsw_sp_mp4_hash_init(mlxsw_sp, &config);
+	mlxsw_sp_mp6_hash_init(mlxsw_sp, &config);
+
+	old_inc_parsing_depth = mlxsw_sp->router->inc_parsing_depth;
+	new_inc_parsing_depth = config.inc_parsing_depth;
+	err = mlxsw_sp_mp_hash_parsing_depth_adjust(mlxsw_sp,
+						    old_inc_parsing_depth,
+						    new_inc_parsing_depth);
+	if (err)
+		return err;
+
+	for_each_set_bit(bit, config.headers, __MLXSW_REG_RECR2_HEADER_CNT)
+		mlxsw_reg_recr2_outer_header_enables_set(recr2_pl, bit, 1);
+	for_each_set_bit(bit, config.fields, __MLXSW_REG_RECR2_FIELD_CNT)
+		mlxsw_reg_recr2_outer_header_fields_enable_set(recr2_pl, bit, 1);
+	for_each_set_bit(bit, config.inner_headers, __MLXSW_REG_RECR2_HEADER_CNT)
+		mlxsw_reg_recr2_inner_header_enables_set(recr2_pl, bit, 1);
+	for_each_set_bit(bit, config.inner_fields, __MLXSW_REG_RECR2_INNER_FIELD_CNT)
+		mlxsw_reg_recr2_inner_header_fields_enable_set(recr2_pl, bit, 1);
+
+	err = mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(recr2), recr2_pl);
+	if (err)
+		goto err_reg_write;
+
+	return 0;
+
+err_reg_write:
+	mlxsw_sp_mp_hash_parsing_depth_adjust(mlxsw_sp, new_inc_parsing_depth,
+					      old_inc_parsing_depth);
+	return err;
+}
+
+static void mlxsw_sp_mp_hash_fini(struct mlxsw_sp *mlxsw_sp)
+{
+	bool old_inc_parsing_depth = mlxsw_sp->router->inc_parsing_depth;
+
+	mlxsw_sp_mp_hash_parsing_depth_adjust(mlxsw_sp, old_inc_parsing_depth,
+					      false);
+}
+#else
+static int mlxsw_sp_mp_hash_init(struct mlxsw_sp *mlxsw_sp)
+{
+	return 0;
+}
+
+static void mlxsw_sp_mp_hash_fini(struct mlxsw_sp *mlxsw_sp)
+{
+}
+#endif
+
+static int mlxsw_sp_dscp_init(struct mlxsw_sp *mlxsw_sp)
+{
+	char rdpm_pl[MLXSW_REG_RDPM_LEN];
+	unsigned int i;
+
+	MLXSW_REG_ZERO(rdpm, rdpm_pl);
+
+	/* HW is determining switch priority based on DSCP-bits, but the
+	 * kernel is still doing that based on the ToS. Since there's a
+	 * mismatch in bits we need to make sure to translate the right
+	 * value ToS would observe, skipping the 2 least-significant ECN bits.
+	 */
+	for (i = 0; i < MLXSW_REG_RDPM_DSCP_ENTRY_REC_MAX_COUNT; i++)
+		mlxsw_reg_rdpm_pack(rdpm_pl, i, rt_tos2priority(i << 2));
+
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(rdpm), rdpm_pl);
+}
+
+static int __mlxsw_sp_router_init(struct mlxsw_sp *mlxsw_sp)
+{
+	struct net *net = mlxsw_sp_net(mlxsw_sp);
+	char rgcr_pl[MLXSW_REG_RGCR_LEN];
+	u64 max_rifs;
+	bool usp;
+
+	if (!MLXSW_CORE_RES_VALID(mlxsw_sp->core, MAX_RIFS))
+		return -EIO;
+	max_rifs = MLXSW_CORE_RES_GET(mlxsw_sp->core, MAX_RIFS);
+	usp = READ_ONCE(net->ipv4.sysctl_ip_fwd_update_priority);
+
+	mlxsw_reg_rgcr_pack(rgcr_pl, true, true);
+	mlxsw_reg_rgcr_max_router_interfaces_set(rgcr_pl, max_rifs);
+	mlxsw_reg_rgcr_usp_set(rgcr_pl, usp);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(rgcr), rgcr_pl);
+}
+
+static void __mlxsw_sp_router_fini(struct mlxsw_sp *mlxsw_sp)
+{
+	char rgcr_pl[MLXSW_REG_RGCR_LEN];
+
+	mlxsw_reg_rgcr_pack(rgcr_pl, false, false);
+	mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(rgcr), rgcr_pl);
+}
+
+static int mlxsw_sp_lb_rif_init(struct mlxsw_sp *mlxsw_sp)
+{
+	u16 lb_rif_index;
+	int err;
+
+	/* Create a generic loopback RIF associated with the main table
+	 * (default VRF). Any table can be used, but the main table exists
+	 * anyway, so we do not waste resources.
+	 */
+	err = mlxsw_sp_router_ul_rif_get(mlxsw_sp, RT_TABLE_MAIN,
+					 &lb_rif_index);
+	if (err)
+		return err;
+
+	mlxsw_sp->router->lb_rif_index = lb_rif_index;
+
+	return 0;
+}
+
+static void mlxsw_sp_lb_rif_fini(struct mlxsw_sp *mlxsw_sp)
+{
+	mlxsw_sp_router_ul_rif_put(mlxsw_sp, mlxsw_sp->router->lb_rif_index);
+}
+
+static int mlxsw_sp1_router_init(struct mlxsw_sp *mlxsw_sp)
+{
+	size_t size_ranges_count = ARRAY_SIZE(mlxsw_sp1_adj_grp_size_ranges);
+
+	mlxsw_sp->router->rif_ops_arr = mlxsw_sp1_rif_ops_arr;
+	mlxsw_sp->router->adj_grp_size_ranges = mlxsw_sp1_adj_grp_size_ranges;
+	mlxsw_sp->router->adj_grp_size_ranges_count = size_ranges_count;
+
+	return 0;
+}
+
+const struct mlxsw_sp_router_ops mlxsw_sp1_router_ops = {
+	.init = mlxsw_sp1_router_init,
+	.ipips_init = mlxsw_sp1_ipips_init,
+};
+
+static int mlxsw_sp2_router_init(struct mlxsw_sp *mlxsw_sp)
+{
+	size_t size_ranges_count = ARRAY_SIZE(mlxsw_sp2_adj_grp_size_ranges);
+
+	mlxsw_sp->router->rif_ops_arr = mlxsw_sp2_rif_ops_arr;
+	mlxsw_sp->router->adj_grp_size_ranges = mlxsw_sp2_adj_grp_size_ranges;
+	mlxsw_sp->router->adj_grp_size_ranges_count = size_ranges_count;
+
+	return 0;
+}
+
+const struct mlxsw_sp_router_ops mlxsw_sp2_router_ops = {
+	.init = mlxsw_sp2_router_init,
+	.ipips_init = mlxsw_sp2_ipips_init,
+};
+
+int mlxsw_sp_router_init(struct mlxsw_sp *mlxsw_sp,
+			 struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp_router *router;
+	int err;
+
+	router = kzalloc(sizeof(*mlxsw_sp->router), GFP_KERNEL);
+	if (!router)
+		return -ENOMEM;
+	mutex_init(&router->lock);
+	mlxsw_sp->router = router;
+	router->mlxsw_sp = mlxsw_sp;
+
+	err = mlxsw_sp->router_ops->init(mlxsw_sp);
+	if (err)
+		goto err_router_ops_init;
+
+	INIT_LIST_HEAD(&mlxsw_sp->router->nh_res_grp_list);
+	INIT_DELAYED_WORK(&mlxsw_sp->router->nh_grp_activity_dw,
+			  mlxsw_sp_nh_grp_activity_work);
+	INIT_LIST_HEAD(&mlxsw_sp->router->nexthop_neighs_list);
+	err = __mlxsw_sp_router_init(mlxsw_sp);
+	if (err)
+		goto err_router_init;
+
+	err = mlxsw_sp_rifs_init(mlxsw_sp);
+	if (err)
+		goto err_rifs_init;
+
+	err = mlxsw_sp->router_ops->ipips_init(mlxsw_sp);
+	if (err)
+		goto err_ipips_init;
+
+	err = rhashtable_init(&mlxsw_sp->router->nexthop_ht,
+			      &mlxsw_sp_nexthop_ht_params);
+	if (err)
+		goto err_nexthop_ht_init;
+
+	err = rhashtable_init(&mlxsw_sp->router->nexthop_group_ht,
+			      &mlxsw_sp_nexthop_group_ht_params);
+	if (err)
+		goto err_nexthop_group_ht_init;
+
+	INIT_LIST_HEAD(&mlxsw_sp->router->nexthop_list);
+	err = mlxsw_sp_lpm_init(mlxsw_sp);
+	if (err)
+		goto err_lpm_init;
+
+	err = mlxsw_sp_mr_init(mlxsw_sp, &mlxsw_sp_mr_tcam_ops);
+	if (err)
+		goto err_mr_init;
+
+	err = mlxsw_sp_vrs_init(mlxsw_sp);
+	if (err)
+		goto err_vrs_init;
+
+	err = mlxsw_sp_lb_rif_init(mlxsw_sp);
+	if (err)
+		goto err_lb_rif_init;
+
+	err = mlxsw_sp_neigh_init(mlxsw_sp);
+	if (err)
+		goto err_neigh_init;
+
+	err = mlxsw_sp_mp_hash_init(mlxsw_sp);
+	if (err)
+		goto err_mp_hash_init;
+
+	err = mlxsw_sp_dscp_init(mlxsw_sp);
+	if (err)
+		goto err_dscp_init;
+
+	router->inetaddr_nb.notifier_call = mlxsw_sp_inetaddr_event;
+	err = register_inetaddr_notifier(&router->inetaddr_nb);
+	if (err)
+		goto err_register_inetaddr_notifier;
+
+	router->inet6addr_nb.notifier_call = mlxsw_sp_inet6addr_event;
+	err = register_inet6addr_notifier(&router->inet6addr_nb);
+	if (err)
+		goto err_register_inet6addr_notifier;
+
+	mlxsw_sp->router->netevent_nb.notifier_call =
+		mlxsw_sp_router_netevent_event;
+	err = register_netevent_notifier(&mlxsw_sp->router->netevent_nb);
+	if (err)
+		goto err_register_netevent_notifier;
+
+	mlxsw_sp->router->nexthop_nb.notifier_call =
+		mlxsw_sp_nexthop_obj_event;
+	err = register_nexthop_notifier(mlxsw_sp_net(mlxsw_sp),
+					&mlxsw_sp->router->nexthop_nb,
+					extack);
+	if (err)
+		goto err_register_nexthop_notifier;
+
+	mlxsw_sp->router->fib_nb.notifier_call = mlxsw_sp_router_fib_event;
+	err = register_fib_notifier(mlxsw_sp_net(mlxsw_sp),
+				    &mlxsw_sp->router->fib_nb,
+				    mlxsw_sp_router_fib_dump_flush, extack);
+	if (err)
+		goto err_register_fib_notifier;
+
+	mlxsw_sp->router->netdevice_nb.notifier_call =
+		mlxsw_sp_router_netdevice_event;
+	err = register_netdevice_notifier_net(mlxsw_sp_net(mlxsw_sp),
+					      &mlxsw_sp->router->netdevice_nb);
+	if (err)
+		goto err_register_netdev_notifier;
+
+	return 0;
+
+err_register_netdev_notifier:
+	unregister_fib_notifier(mlxsw_sp_net(mlxsw_sp),
+				&mlxsw_sp->router->fib_nb);
+err_register_fib_notifier:
+	unregister_nexthop_notifier(mlxsw_sp_net(mlxsw_sp),
+				    &mlxsw_sp->router->nexthop_nb);
+err_register_nexthop_notifier:
+	unregister_netevent_notifier(&mlxsw_sp->router->netevent_nb);
+err_register_netevent_notifier:
+	unregister_inet6addr_notifier(&router->inet6addr_nb);
+err_register_inet6addr_notifier:
+	unregister_inetaddr_notifier(&router->inetaddr_nb);
+err_register_inetaddr_notifier:
+	mlxsw_core_flush_owq();
+err_dscp_init:
+	mlxsw_sp_mp_hash_fini(mlxsw_sp);
+err_mp_hash_init:
+	mlxsw_sp_neigh_fini(mlxsw_sp);
+err_neigh_init:
+	mlxsw_sp_lb_rif_fini(mlxsw_sp);
+err_lb_rif_init:
+	mlxsw_sp_vrs_fini(mlxsw_sp);
+err_vrs_init:
+	mlxsw_sp_mr_fini(mlxsw_sp);
+err_mr_init:
+	mlxsw_sp_lpm_fini(mlxsw_sp);
+err_lpm_init:
+	rhashtable_destroy(&mlxsw_sp->router->nexthop_group_ht);
+err_nexthop_group_ht_init:
+	rhashtable_destroy(&mlxsw_sp->router->nexthop_ht);
+err_nexthop_ht_init:
+	mlxsw_sp_ipips_fini(mlxsw_sp);
+err_ipips_init:
+	mlxsw_sp_rifs_fini(mlxsw_sp);
+err_rifs_init:
+	__mlxsw_sp_router_fini(mlxsw_sp);
+err_router_init:
+	cancel_delayed_work_sync(&mlxsw_sp->router->nh_grp_activity_dw);
+err_router_ops_init:
+	mutex_destroy(&mlxsw_sp->router->lock);
+	kfree(mlxsw_sp->router);
+	return err;
+}
+
+void mlxsw_sp_router_fini(struct mlxsw_sp *mlxsw_sp)
+{
+	unregister_netdevice_notifier_net(mlxsw_sp_net(mlxsw_sp),
+					  &mlxsw_sp->router->netdevice_nb);
+	unregister_fib_notifier(mlxsw_sp_net(mlxsw_sp),
+				&mlxsw_sp->router->fib_nb);
+	unregister_nexthop_notifier(mlxsw_sp_net(mlxsw_sp),
+				    &mlxsw_sp->router->nexthop_nb);
+	unregister_netevent_notifier(&mlxsw_sp->router->netevent_nb);
+	unregister_inet6addr_notifier(&mlxsw_sp->router->inet6addr_nb);
+	unregister_inetaddr_notifier(&mlxsw_sp->router->inetaddr_nb);
+	mlxsw_core_flush_owq();
+	mlxsw_sp_mp_hash_fini(mlxsw_sp);
+	mlxsw_sp_neigh_fini(mlxsw_sp);
+	mlxsw_sp_lb_rif_fini(mlxsw_sp);
+	mlxsw_sp_vrs_fini(mlxsw_sp);
+	mlxsw_sp_mr_fini(mlxsw_sp);
+	mlxsw_sp_lpm_fini(mlxsw_sp);
+	rhashtable_destroy(&mlxsw_sp->router->nexthop_group_ht);
+	rhashtable_destroy(&mlxsw_sp->router->nexthop_ht);
+	mlxsw_sp_ipips_fini(mlxsw_sp);
+	mlxsw_sp_rifs_fini(mlxsw_sp);
+	__mlxsw_sp_router_fini(mlxsw_sp);
+	cancel_delayed_work_sync(&mlxsw_sp->router->nh_grp_activity_dw);
+	mutex_destroy(&mlxsw_sp->router->lock);
+	kfree(mlxsw_sp->router);
+}
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.h b/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.h
new file mode 100644
index 000000000..c5dfb972b
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.h
@@ -0,0 +1,166 @@
+/* SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0 */
+/* Copyright (c) 2017-2018 Mellanox Technologies. All rights reserved */
+
+#ifndef _MLXSW_ROUTER_H_
+#define _MLXSW_ROUTER_H_
+
+#include "spectrum.h"
+#include "reg.h"
+
+struct mlxsw_sp_router_nve_decap {
+	u32 ul_tb_id;
+	u32 tunnel_index;
+	enum mlxsw_sp_l3proto ul_proto;
+	union mlxsw_sp_l3addr ul_sip;
+	u8 valid:1;
+};
+
+struct mlxsw_sp_router {
+	struct mlxsw_sp *mlxsw_sp;
+	struct mlxsw_sp_rif **rifs;
+	struct idr rif_mac_profiles_idr;
+	atomic_t rif_mac_profiles_count;
+	atomic_t rifs_count;
+	u8 max_rif_mac_profile;
+	struct mlxsw_sp_vr *vrs;
+	struct rhashtable neigh_ht;
+	struct rhashtable nexthop_group_ht;
+	struct rhashtable nexthop_ht;
+	struct list_head nexthop_list;
+	struct {
+		/* One tree for each protocol: IPv4 and IPv6 */
+		struct mlxsw_sp_lpm_tree *proto_trees[2];
+		struct mlxsw_sp_lpm_tree *trees;
+		unsigned int tree_count;
+	} lpm;
+	struct {
+		struct delayed_work dw;
+		unsigned long interval;	/* ms */
+		atomic_t neigh_count;
+	} neighs_update;
+	struct delayed_work nexthop_probe_dw;
+#define MLXSW_SP_UNRESOLVED_NH_PROBE_INTERVAL 5000 /* ms */
+	struct list_head nexthop_neighs_list;
+	struct list_head ipip_list;
+	struct notifier_block nexthop_nb;
+	struct notifier_block fib_nb;
+	struct notifier_block netevent_nb;
+	struct notifier_block inetaddr_nb;
+	struct notifier_block inet6addr_nb;
+	struct notifier_block netdevice_nb;
+	const struct mlxsw_sp_rif_ops **rif_ops_arr;
+	const struct mlxsw_sp_ipip_ops **ipip_ops_arr;
+	struct mlxsw_sp_router_nve_decap nve_decap_config;
+	struct mutex lock; /* Protects shared router resources */
+	struct mlxsw_sp_fib_entry_op_ctx *ll_op_ctx;
+	u16 lb_rif_index;
+	const struct mlxsw_sp_adj_grp_size_range *adj_grp_size_ranges;
+	size_t adj_grp_size_ranges_count;
+	struct delayed_work nh_grp_activity_dw;
+	struct list_head nh_res_grp_list;
+	bool inc_parsing_depth;
+	refcount_t num_groups;
+	u32 adj_trap_index;
+};
+
+struct mlxsw_sp_rif_ipip_lb;
+struct mlxsw_sp_rif_ipip_lb_config {
+	enum mlxsw_reg_ritr_loopback_ipip_type lb_ipipt;
+	u32 okey;
+	enum mlxsw_sp_l3proto ul_protocol; /* Underlay. */
+	union mlxsw_sp_l3addr saddr;
+};
+
+enum mlxsw_sp_rif_counter_dir {
+	MLXSW_SP_RIF_COUNTER_INGRESS,
+	MLXSW_SP_RIF_COUNTER_EGRESS,
+};
+
+struct mlxsw_sp_neigh_entry;
+struct mlxsw_sp_nexthop;
+struct mlxsw_sp_ipip_entry;
+
+struct mlxsw_sp_rif *mlxsw_sp_rif_by_index(const struct mlxsw_sp *mlxsw_sp,
+					   u16 rif_index);
+u16 mlxsw_sp_ipip_lb_rif_index(const struct mlxsw_sp_rif_ipip_lb *rif);
+u16 mlxsw_sp_ipip_lb_ul_vr_id(const struct mlxsw_sp_rif_ipip_lb *rif);
+u16 mlxsw_sp_ipip_lb_ul_rif_id(const struct mlxsw_sp_rif_ipip_lb *lb_rif);
+u32 mlxsw_sp_ipip_dev_ul_tb_id(const struct net_device *ol_dev);
+int mlxsw_sp_rif_dev_ifindex(const struct mlxsw_sp_rif *rif);
+const struct net_device *mlxsw_sp_rif_dev(const struct mlxsw_sp_rif *rif);
+int mlxsw_sp_rif_counter_value_get(struct mlxsw_sp *mlxsw_sp,
+				   struct mlxsw_sp_rif *rif,
+				   enum mlxsw_sp_rif_counter_dir dir,
+				   u64 *cnt);
+void mlxsw_sp_rif_counter_free(struct mlxsw_sp_rif *rif,
+			       enum mlxsw_sp_rif_counter_dir dir);
+int mlxsw_sp_rif_counter_alloc(struct mlxsw_sp_rif *rif,
+			       enum mlxsw_sp_rif_counter_dir dir);
+struct mlxsw_sp_neigh_entry *
+mlxsw_sp_rif_neigh_next(struct mlxsw_sp_rif *rif,
+			struct mlxsw_sp_neigh_entry *neigh_entry);
+int mlxsw_sp_neigh_entry_type(struct mlxsw_sp_neigh_entry *neigh_entry);
+unsigned char *
+mlxsw_sp_neigh_entry_ha(struct mlxsw_sp_neigh_entry *neigh_entry);
+u32 mlxsw_sp_neigh4_entry_dip(struct mlxsw_sp_neigh_entry *neigh_entry);
+struct in6_addr *
+mlxsw_sp_neigh6_entry_dip(struct mlxsw_sp_neigh_entry *neigh_entry);
+
+#define mlxsw_sp_rif_neigh_for_each(neigh_entry, rif)				\
+	for (neigh_entry = mlxsw_sp_rif_neigh_next(rif, NULL); neigh_entry;	\
+	     neigh_entry = mlxsw_sp_rif_neigh_next(rif, neigh_entry))
+int mlxsw_sp_neigh_counter_get(struct mlxsw_sp *mlxsw_sp,
+			       struct mlxsw_sp_neigh_entry *neigh_entry,
+			       u64 *p_counter);
+void
+mlxsw_sp_neigh_entry_counter_update(struct mlxsw_sp *mlxsw_sp,
+				    struct mlxsw_sp_neigh_entry *neigh_entry,
+				    bool adding);
+bool mlxsw_sp_neigh_ipv6_ignore(struct mlxsw_sp_neigh_entry *neigh_entry);
+int __mlxsw_sp_ipip_entry_update_tunnel(struct mlxsw_sp *mlxsw_sp,
+					struct mlxsw_sp_ipip_entry *ipip_entry,
+					bool recreate_loopback,
+					bool keep_encap,
+					bool update_nexthops,
+					struct netlink_ext_ack *extack);
+void mlxsw_sp_ipip_entry_demote_tunnel(struct mlxsw_sp *mlxsw_sp,
+				       struct mlxsw_sp_ipip_entry *ipip_entry);
+bool
+mlxsw_sp_ipip_demote_tunnel_by_saddr(struct mlxsw_sp *mlxsw_sp,
+				     enum mlxsw_sp_l3proto ul_proto,
+				     union mlxsw_sp_l3addr saddr,
+				     u32 ul_tb_id,
+				     const struct mlxsw_sp_ipip_entry *except);
+struct mlxsw_sp_nexthop *mlxsw_sp_nexthop_next(struct mlxsw_sp_router *router,
+					       struct mlxsw_sp_nexthop *nh);
+bool mlxsw_sp_nexthop_is_forward(const struct mlxsw_sp_nexthop *nh);
+unsigned char *mlxsw_sp_nexthop_ha(struct mlxsw_sp_nexthop *nh);
+int mlxsw_sp_nexthop_indexes(struct mlxsw_sp_nexthop *nh, u32 *p_adj_index,
+			     u32 *p_adj_size, u32 *p_adj_hash_index);
+struct mlxsw_sp_rif *mlxsw_sp_nexthop_rif(struct mlxsw_sp_nexthop *nh);
+bool mlxsw_sp_nexthop_group_has_ipip(struct mlxsw_sp_nexthop *nh);
+#define mlxsw_sp_nexthop_for_each(nh, router)				\
+	for (nh = mlxsw_sp_nexthop_next(router, NULL); nh;		\
+	     nh = mlxsw_sp_nexthop_next(router, nh))
+int mlxsw_sp_nexthop_counter_get(struct mlxsw_sp *mlxsw_sp,
+				 struct mlxsw_sp_nexthop *nh, u64 *p_counter);
+int mlxsw_sp_nexthop_eth_update(struct mlxsw_sp *mlxsw_sp, u32 adj_index,
+				struct mlxsw_sp_nexthop *nh, bool force,
+				char *ratr_pl);
+void mlxsw_sp_nexthop_counter_alloc(struct mlxsw_sp *mlxsw_sp,
+				    struct mlxsw_sp_nexthop *nh);
+void mlxsw_sp_nexthop_counter_free(struct mlxsw_sp *mlxsw_sp,
+				   struct mlxsw_sp_nexthop *nh);
+
+static inline bool mlxsw_sp_l3addr_eq(const union mlxsw_sp_l3addr *addr1,
+				      const union mlxsw_sp_l3addr *addr2)
+{
+	return !memcmp(addr1, addr2, sizeof(*addr1));
+}
+
+int mlxsw_sp_ipip_ecn_encap_init(struct mlxsw_sp *mlxsw_sp);
+int mlxsw_sp_ipip_ecn_decap_init(struct mlxsw_sp *mlxsw_sp);
+struct net_device *
+mlxsw_sp_ipip_netdev_ul_dev_get(const struct net_device *ol_dev);
+
+#endif /* _MLXSW_ROUTER_H_*/
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_span.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_span.c
new file mode 100644
index 000000000..b3472fb94
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_span.c
@@ -0,0 +1,1741 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2018 Mellanox Technologies. All rights reserved */
+
+#include <linux/if_bridge.h>
+#include <linux/list.h>
+#include <linux/mutex.h>
+#include <linux/refcount.h>
+#include <linux/rtnetlink.h>
+#include <linux/workqueue.h>
+#include <net/arp.h>
+#include <net/gre.h>
+#include <net/lag.h>
+#include <net/ndisc.h>
+#include <net/ip6_tunnel.h>
+
+#include "spectrum.h"
+#include "spectrum_ipip.h"
+#include "spectrum_span.h"
+#include "spectrum_switchdev.h"
+
+struct mlxsw_sp_span {
+	struct work_struct work;
+	struct mlxsw_sp *mlxsw_sp;
+	const struct mlxsw_sp_span_trigger_ops **span_trigger_ops_arr;
+	const struct mlxsw_sp_span_entry_ops **span_entry_ops_arr;
+	size_t span_entry_ops_arr_size;
+	struct list_head analyzed_ports_list;
+	struct mutex analyzed_ports_lock; /* Protects analyzed_ports_list */
+	struct list_head trigger_entries_list;
+	u16 policer_id_base;
+	refcount_t policer_id_base_ref_count;
+	atomic_t active_entries_count;
+	int entries_count;
+	struct mlxsw_sp_span_entry entries[];
+};
+
+struct mlxsw_sp_span_analyzed_port {
+	struct list_head list; /* Member of analyzed_ports_list */
+	refcount_t ref_count;
+	u16 local_port;
+	bool ingress;
+};
+
+struct mlxsw_sp_span_trigger_entry {
+	struct list_head list; /* Member of trigger_entries_list */
+	struct mlxsw_sp_span *span;
+	const struct mlxsw_sp_span_trigger_ops *ops;
+	refcount_t ref_count;
+	u16 local_port;
+	enum mlxsw_sp_span_trigger trigger;
+	struct mlxsw_sp_span_trigger_parms parms;
+};
+
+enum mlxsw_sp_span_trigger_type {
+	MLXSW_SP_SPAN_TRIGGER_TYPE_PORT,
+	MLXSW_SP_SPAN_TRIGGER_TYPE_GLOBAL,
+};
+
+struct mlxsw_sp_span_trigger_ops {
+	int (*bind)(struct mlxsw_sp_span_trigger_entry *trigger_entry);
+	void (*unbind)(struct mlxsw_sp_span_trigger_entry *trigger_entry);
+	bool (*matches)(struct mlxsw_sp_span_trigger_entry *trigger_entry,
+			enum mlxsw_sp_span_trigger trigger,
+			struct mlxsw_sp_port *mlxsw_sp_port);
+	int (*enable)(struct mlxsw_sp_span_trigger_entry *trigger_entry,
+		      struct mlxsw_sp_port *mlxsw_sp_port, u8 tc);
+	void (*disable)(struct mlxsw_sp_span_trigger_entry *trigger_entry,
+			struct mlxsw_sp_port *mlxsw_sp_port, u8 tc);
+};
+
+static void mlxsw_sp_span_respin_work(struct work_struct *work);
+
+static u64 mlxsw_sp_span_occ_get(void *priv)
+{
+	const struct mlxsw_sp *mlxsw_sp = priv;
+
+	return atomic_read(&mlxsw_sp->span->active_entries_count);
+}
+
+int mlxsw_sp_span_init(struct mlxsw_sp *mlxsw_sp)
+{
+	struct devlink *devlink = priv_to_devlink(mlxsw_sp->core);
+	struct mlxsw_sp_span *span;
+	int i, entries_count, err;
+
+	if (!MLXSW_CORE_RES_VALID(mlxsw_sp->core, MAX_SPAN))
+		return -EIO;
+
+	entries_count = MLXSW_CORE_RES_GET(mlxsw_sp->core, MAX_SPAN);
+	span = kzalloc(struct_size(span, entries, entries_count), GFP_KERNEL);
+	if (!span)
+		return -ENOMEM;
+	refcount_set(&span->policer_id_base_ref_count, 0);
+	span->entries_count = entries_count;
+	atomic_set(&span->active_entries_count, 0);
+	mutex_init(&span->analyzed_ports_lock);
+	INIT_LIST_HEAD(&span->analyzed_ports_list);
+	INIT_LIST_HEAD(&span->trigger_entries_list);
+	span->mlxsw_sp = mlxsw_sp;
+	mlxsw_sp->span = span;
+
+	for (i = 0; i < mlxsw_sp->span->entries_count; i++)
+		mlxsw_sp->span->entries[i].id = i;
+
+	err = mlxsw_sp->span_ops->init(mlxsw_sp);
+	if (err)
+		goto err_init;
+
+	devl_resource_occ_get_register(devlink, MLXSW_SP_RESOURCE_SPAN,
+				       mlxsw_sp_span_occ_get, mlxsw_sp);
+	INIT_WORK(&span->work, mlxsw_sp_span_respin_work);
+
+	return 0;
+
+err_init:
+	mutex_destroy(&mlxsw_sp->span->analyzed_ports_lock);
+	kfree(mlxsw_sp->span);
+	return err;
+}
+
+void mlxsw_sp_span_fini(struct mlxsw_sp *mlxsw_sp)
+{
+	struct devlink *devlink = priv_to_devlink(mlxsw_sp->core);
+
+	cancel_work_sync(&mlxsw_sp->span->work);
+	devl_resource_occ_get_unregister(devlink, MLXSW_SP_RESOURCE_SPAN);
+
+	WARN_ON_ONCE(!list_empty(&mlxsw_sp->span->trigger_entries_list));
+	WARN_ON_ONCE(!list_empty(&mlxsw_sp->span->analyzed_ports_list));
+	mutex_destroy(&mlxsw_sp->span->analyzed_ports_lock);
+	kfree(mlxsw_sp->span);
+}
+
+static bool mlxsw_sp1_span_cpu_can_handle(const struct net_device *dev)
+{
+	return !dev;
+}
+
+static int mlxsw_sp1_span_entry_cpu_parms(struct mlxsw_sp *mlxsw_sp,
+					  const struct net_device *to_dev,
+					  struct mlxsw_sp_span_parms *sparmsp)
+{
+	return -EOPNOTSUPP;
+}
+
+static int
+mlxsw_sp1_span_entry_cpu_configure(struct mlxsw_sp_span_entry *span_entry,
+				   struct mlxsw_sp_span_parms sparms)
+{
+	return -EOPNOTSUPP;
+}
+
+static void
+mlxsw_sp1_span_entry_cpu_deconfigure(struct mlxsw_sp_span_entry *span_entry)
+{
+}
+
+static const
+struct mlxsw_sp_span_entry_ops mlxsw_sp1_span_entry_ops_cpu = {
+	.is_static = true,
+	.can_handle = mlxsw_sp1_span_cpu_can_handle,
+	.parms_set = mlxsw_sp1_span_entry_cpu_parms,
+	.configure = mlxsw_sp1_span_entry_cpu_configure,
+	.deconfigure = mlxsw_sp1_span_entry_cpu_deconfigure,
+};
+
+static int
+mlxsw_sp_span_entry_phys_parms(struct mlxsw_sp *mlxsw_sp,
+			       const struct net_device *to_dev,
+			       struct mlxsw_sp_span_parms *sparmsp)
+{
+	sparmsp->dest_port = netdev_priv(to_dev);
+	return 0;
+}
+
+static int
+mlxsw_sp_span_entry_phys_configure(struct mlxsw_sp_span_entry *span_entry,
+				   struct mlxsw_sp_span_parms sparms)
+{
+	struct mlxsw_sp_port *dest_port = sparms.dest_port;
+	struct mlxsw_sp *mlxsw_sp = dest_port->mlxsw_sp;
+	u16 local_port = dest_port->local_port;
+	char mpat_pl[MLXSW_REG_MPAT_LEN];
+	int pa_id = span_entry->id;
+
+	/* Create a new port analayzer entry for local_port. */
+	mlxsw_reg_mpat_pack(mpat_pl, pa_id, local_port, true,
+			    MLXSW_REG_MPAT_SPAN_TYPE_LOCAL_ETH);
+	mlxsw_reg_mpat_session_id_set(mpat_pl, sparms.session_id);
+	mlxsw_reg_mpat_pide_set(mpat_pl, sparms.policer_enable);
+	mlxsw_reg_mpat_pid_set(mpat_pl, sparms.policer_id);
+
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(mpat), mpat_pl);
+}
+
+static void
+mlxsw_sp_span_entry_deconfigure_common(struct mlxsw_sp_span_entry *span_entry,
+				       enum mlxsw_reg_mpat_span_type span_type)
+{
+	struct mlxsw_sp_port *dest_port = span_entry->parms.dest_port;
+	struct mlxsw_sp *mlxsw_sp = dest_port->mlxsw_sp;
+	u16 local_port = dest_port->local_port;
+	char mpat_pl[MLXSW_REG_MPAT_LEN];
+	int pa_id = span_entry->id;
+
+	mlxsw_reg_mpat_pack(mpat_pl, pa_id, local_port, false, span_type);
+	mlxsw_reg_mpat_session_id_set(mpat_pl, span_entry->parms.session_id);
+	mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(mpat), mpat_pl);
+}
+
+static void
+mlxsw_sp_span_entry_phys_deconfigure(struct mlxsw_sp_span_entry *span_entry)
+{
+	mlxsw_sp_span_entry_deconfigure_common(span_entry,
+					    MLXSW_REG_MPAT_SPAN_TYPE_LOCAL_ETH);
+}
+
+static const
+struct mlxsw_sp_span_entry_ops mlxsw_sp_span_entry_ops_phys = {
+	.is_static = true,
+	.can_handle = mlxsw_sp_port_dev_check,
+	.parms_set = mlxsw_sp_span_entry_phys_parms,
+	.configure = mlxsw_sp_span_entry_phys_configure,
+	.deconfigure = mlxsw_sp_span_entry_phys_deconfigure,
+};
+
+static int mlxsw_sp_span_dmac(struct neigh_table *tbl,
+			      const void *pkey,
+			      struct net_device *dev,
+			      unsigned char dmac[ETH_ALEN])
+{
+	struct neighbour *neigh = neigh_lookup(tbl, pkey, dev);
+	int err = 0;
+
+	if (!neigh) {
+		neigh = neigh_create(tbl, pkey, dev);
+		if (IS_ERR(neigh))
+			return PTR_ERR(neigh);
+	}
+
+	neigh_event_send(neigh, NULL);
+
+	read_lock_bh(&neigh->lock);
+	if ((neigh->nud_state & NUD_VALID) && !neigh->dead)
+		memcpy(dmac, neigh->ha, ETH_ALEN);
+	else
+		err = -ENOENT;
+	read_unlock_bh(&neigh->lock);
+
+	neigh_release(neigh);
+	return err;
+}
+
+static int
+mlxsw_sp_span_entry_unoffloadable(struct mlxsw_sp_span_parms *sparmsp)
+{
+	sparmsp->dest_port = NULL;
+	return 0;
+}
+
+static struct net_device *
+mlxsw_sp_span_entry_bridge_8021q(const struct net_device *br_dev,
+				 unsigned char *dmac,
+				 u16 *p_vid)
+{
+	struct bridge_vlan_info vinfo;
+	struct net_device *edev;
+	u16 vid = *p_vid;
+
+	if (!vid && WARN_ON(br_vlan_get_pvid(br_dev, &vid)))
+		return NULL;
+	if (!vid || br_vlan_get_info(br_dev, vid, &vinfo) ||
+	    !(vinfo.flags & BRIDGE_VLAN_INFO_BRENTRY))
+		return NULL;
+
+	edev = br_fdb_find_port(br_dev, dmac, vid);
+	if (!edev)
+		return NULL;
+
+	if (br_vlan_get_info(edev, vid, &vinfo))
+		return NULL;
+	if (vinfo.flags & BRIDGE_VLAN_INFO_UNTAGGED)
+		*p_vid = 0;
+	else
+		*p_vid = vid;
+	return edev;
+}
+
+static struct net_device *
+mlxsw_sp_span_entry_bridge_8021d(const struct net_device *br_dev,
+				 unsigned char *dmac)
+{
+	return br_fdb_find_port(br_dev, dmac, 0);
+}
+
+static struct net_device *
+mlxsw_sp_span_entry_bridge(const struct net_device *br_dev,
+			   unsigned char dmac[ETH_ALEN],
+			   u16 *p_vid)
+{
+	struct mlxsw_sp_bridge_port *bridge_port;
+	enum mlxsw_reg_spms_state spms_state;
+	struct net_device *dev = NULL;
+	struct mlxsw_sp_port *port;
+	u8 stp_state;
+
+	if (br_vlan_enabled(br_dev))
+		dev = mlxsw_sp_span_entry_bridge_8021q(br_dev, dmac, p_vid);
+	else if (!*p_vid)
+		dev = mlxsw_sp_span_entry_bridge_8021d(br_dev, dmac);
+	if (!dev)
+		return NULL;
+
+	port = mlxsw_sp_port_dev_lower_find(dev);
+	if (!port)
+		return NULL;
+
+	bridge_port = mlxsw_sp_bridge_port_find(port->mlxsw_sp->bridge, dev);
+	if (!bridge_port)
+		return NULL;
+
+	stp_state = mlxsw_sp_bridge_port_stp_state(bridge_port);
+	spms_state = mlxsw_sp_stp_spms_state(stp_state);
+	if (spms_state != MLXSW_REG_SPMS_STATE_FORWARDING)
+		return NULL;
+
+	return dev;
+}
+
+static struct net_device *
+mlxsw_sp_span_entry_vlan(const struct net_device *vlan_dev,
+			 u16 *p_vid)
+{
+	*p_vid = vlan_dev_vlan_id(vlan_dev);
+	return vlan_dev_real_dev(vlan_dev);
+}
+
+static struct net_device *
+mlxsw_sp_span_entry_lag(struct net_device *lag_dev)
+{
+	struct net_device *dev;
+	struct list_head *iter;
+
+	netdev_for_each_lower_dev(lag_dev, dev, iter)
+		if (netif_carrier_ok(dev) &&
+		    net_lag_port_dev_txable(dev) &&
+		    mlxsw_sp_port_dev_check(dev))
+			return dev;
+
+	return NULL;
+}
+
+static __maybe_unused int
+mlxsw_sp_span_entry_tunnel_parms_common(struct net_device *edev,
+					union mlxsw_sp_l3addr saddr,
+					union mlxsw_sp_l3addr daddr,
+					union mlxsw_sp_l3addr gw,
+					__u8 ttl,
+					struct neigh_table *tbl,
+					struct mlxsw_sp_span_parms *sparmsp)
+{
+	unsigned char dmac[ETH_ALEN];
+	u16 vid = 0;
+
+	if (mlxsw_sp_l3addr_is_zero(gw))
+		gw = daddr;
+
+	if (!edev || mlxsw_sp_span_dmac(tbl, &gw, edev, dmac))
+		goto unoffloadable;
+
+	if (is_vlan_dev(edev))
+		edev = mlxsw_sp_span_entry_vlan(edev, &vid);
+
+	if (netif_is_bridge_master(edev)) {
+		edev = mlxsw_sp_span_entry_bridge(edev, dmac, &vid);
+		if (!edev)
+			goto unoffloadable;
+	}
+
+	if (is_vlan_dev(edev)) {
+		if (vid || !(edev->flags & IFF_UP))
+			goto unoffloadable;
+		edev = mlxsw_sp_span_entry_vlan(edev, &vid);
+	}
+
+	if (netif_is_lag_master(edev)) {
+		if (!(edev->flags & IFF_UP))
+			goto unoffloadable;
+		edev = mlxsw_sp_span_entry_lag(edev);
+		if (!edev)
+			goto unoffloadable;
+	}
+
+	if (!mlxsw_sp_port_dev_check(edev))
+		goto unoffloadable;
+
+	sparmsp->dest_port = netdev_priv(edev);
+	sparmsp->ttl = ttl;
+	memcpy(sparmsp->dmac, dmac, ETH_ALEN);
+	memcpy(sparmsp->smac, edev->dev_addr, ETH_ALEN);
+	sparmsp->saddr = saddr;
+	sparmsp->daddr = daddr;
+	sparmsp->vid = vid;
+	return 0;
+
+unoffloadable:
+	return mlxsw_sp_span_entry_unoffloadable(sparmsp);
+}
+
+#if IS_ENABLED(CONFIG_NET_IPGRE)
+static struct net_device *
+mlxsw_sp_span_gretap4_route(const struct net_device *to_dev,
+			    __be32 *saddrp, __be32 *daddrp)
+{
+	struct ip_tunnel *tun = netdev_priv(to_dev);
+	struct net_device *dev = NULL;
+	struct ip_tunnel_parm parms;
+	struct rtable *rt = NULL;
+	struct flowi4 fl4;
+
+	/* We assume "dev" stays valid after rt is put. */
+	ASSERT_RTNL();
+
+	parms = mlxsw_sp_ipip_netdev_parms4(to_dev);
+	ip_tunnel_init_flow(&fl4, parms.iph.protocol, *daddrp, *saddrp,
+			    0, 0, dev_net(to_dev), parms.link, tun->fwmark, 0,
+			    0);
+
+	rt = ip_route_output_key(tun->net, &fl4);
+	if (IS_ERR(rt))
+		return NULL;
+
+	if (rt->rt_type != RTN_UNICAST)
+		goto out;
+
+	dev = rt->dst.dev;
+	*saddrp = fl4.saddr;
+	if (rt->rt_gw_family == AF_INET)
+		*daddrp = rt->rt_gw4;
+	/* can not offload if route has an IPv6 gateway */
+	else if (rt->rt_gw_family == AF_INET6)
+		dev = NULL;
+
+out:
+	ip_rt_put(rt);
+	return dev;
+}
+
+static int
+mlxsw_sp_span_entry_gretap4_parms(struct mlxsw_sp *mlxsw_sp,
+				  const struct net_device *to_dev,
+				  struct mlxsw_sp_span_parms *sparmsp)
+{
+	struct ip_tunnel_parm tparm = mlxsw_sp_ipip_netdev_parms4(to_dev);
+	union mlxsw_sp_l3addr saddr = { .addr4 = tparm.iph.saddr };
+	union mlxsw_sp_l3addr daddr = { .addr4 = tparm.iph.daddr };
+	bool inherit_tos = tparm.iph.tos & 0x1;
+	bool inherit_ttl = !tparm.iph.ttl;
+	union mlxsw_sp_l3addr gw = daddr;
+	struct net_device *l3edev;
+
+	if (!(to_dev->flags & IFF_UP) ||
+	    /* Reject tunnels with GRE keys, checksums, etc. */
+	    tparm.i_flags || tparm.o_flags ||
+	    /* Require a fixed TTL and a TOS copied from the mirrored packet. */
+	    inherit_ttl || !inherit_tos ||
+	    /* A destination address may not be "any". */
+	    mlxsw_sp_l3addr_is_zero(daddr))
+		return mlxsw_sp_span_entry_unoffloadable(sparmsp);
+
+	l3edev = mlxsw_sp_span_gretap4_route(to_dev, &saddr.addr4, &gw.addr4);
+	return mlxsw_sp_span_entry_tunnel_parms_common(l3edev, saddr, daddr, gw,
+						       tparm.iph.ttl,
+						       &arp_tbl, sparmsp);
+}
+
+static int
+mlxsw_sp_span_entry_gretap4_configure(struct mlxsw_sp_span_entry *span_entry,
+				      struct mlxsw_sp_span_parms sparms)
+{
+	struct mlxsw_sp_port *dest_port = sparms.dest_port;
+	struct mlxsw_sp *mlxsw_sp = dest_port->mlxsw_sp;
+	u16 local_port = dest_port->local_port;
+	char mpat_pl[MLXSW_REG_MPAT_LEN];
+	int pa_id = span_entry->id;
+
+	/* Create a new port analayzer entry for local_port. */
+	mlxsw_reg_mpat_pack(mpat_pl, pa_id, local_port, true,
+			    MLXSW_REG_MPAT_SPAN_TYPE_REMOTE_ETH_L3);
+	mlxsw_reg_mpat_pide_set(mpat_pl, sparms.policer_enable);
+	mlxsw_reg_mpat_pid_set(mpat_pl, sparms.policer_id);
+	mlxsw_reg_mpat_eth_rspan_pack(mpat_pl, sparms.vid);
+	mlxsw_reg_mpat_eth_rspan_l2_pack(mpat_pl,
+				    MLXSW_REG_MPAT_ETH_RSPAN_VERSION_NO_HEADER,
+				    sparms.dmac, !!sparms.vid);
+	mlxsw_reg_mpat_eth_rspan_l3_ipv4_pack(mpat_pl,
+					      sparms.ttl, sparms.smac,
+					      be32_to_cpu(sparms.saddr.addr4),
+					      be32_to_cpu(sparms.daddr.addr4));
+
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(mpat), mpat_pl);
+}
+
+static void
+mlxsw_sp_span_entry_gretap4_deconfigure(struct mlxsw_sp_span_entry *span_entry)
+{
+	mlxsw_sp_span_entry_deconfigure_common(span_entry,
+					MLXSW_REG_MPAT_SPAN_TYPE_REMOTE_ETH_L3);
+}
+
+static const struct mlxsw_sp_span_entry_ops mlxsw_sp_span_entry_ops_gretap4 = {
+	.can_handle = netif_is_gretap,
+	.parms_set = mlxsw_sp_span_entry_gretap4_parms,
+	.configure = mlxsw_sp_span_entry_gretap4_configure,
+	.deconfigure = mlxsw_sp_span_entry_gretap4_deconfigure,
+};
+#endif
+
+#if IS_ENABLED(CONFIG_IPV6_GRE)
+static struct net_device *
+mlxsw_sp_span_gretap6_route(const struct net_device *to_dev,
+			    struct in6_addr *saddrp,
+			    struct in6_addr *daddrp)
+{
+	struct ip6_tnl *t = netdev_priv(to_dev);
+	struct flowi6 fl6 = t->fl.u.ip6;
+	struct net_device *dev = NULL;
+	struct dst_entry *dst;
+	struct rt6_info *rt6;
+
+	/* We assume "dev" stays valid after dst is released. */
+	ASSERT_RTNL();
+
+	fl6.flowi6_mark = t->parms.fwmark;
+	if (!ip6_tnl_xmit_ctl(t, &fl6.saddr, &fl6.daddr))
+		return NULL;
+
+	dst = ip6_route_output(t->net, NULL, &fl6);
+	if (!dst || dst->error)
+		goto out;
+
+	rt6 = container_of(dst, struct rt6_info, dst);
+
+	dev = dst->dev;
+	*saddrp = fl6.saddr;
+	*daddrp = rt6->rt6i_gateway;
+
+out:
+	dst_release(dst);
+	return dev;
+}
+
+static int
+mlxsw_sp_span_entry_gretap6_parms(struct mlxsw_sp *mlxsw_sp,
+				  const struct net_device *to_dev,
+				  struct mlxsw_sp_span_parms *sparmsp)
+{
+	struct __ip6_tnl_parm tparm = mlxsw_sp_ipip_netdev_parms6(to_dev);
+	bool inherit_tos = tparm.flags & IP6_TNL_F_USE_ORIG_TCLASS;
+	union mlxsw_sp_l3addr saddr = { .addr6 = tparm.laddr };
+	union mlxsw_sp_l3addr daddr = { .addr6 = tparm.raddr };
+	bool inherit_ttl = !tparm.hop_limit;
+	union mlxsw_sp_l3addr gw = daddr;
+	struct net_device *l3edev;
+
+	if (!(to_dev->flags & IFF_UP) ||
+	    /* Reject tunnels with GRE keys, checksums, etc. */
+	    tparm.i_flags || tparm.o_flags ||
+	    /* Require a fixed TTL and a TOS copied from the mirrored packet. */
+	    inherit_ttl || !inherit_tos ||
+	    /* A destination address may not be "any". */
+	    mlxsw_sp_l3addr_is_zero(daddr))
+		return mlxsw_sp_span_entry_unoffloadable(sparmsp);
+
+	l3edev = mlxsw_sp_span_gretap6_route(to_dev, &saddr.addr6, &gw.addr6);
+	return mlxsw_sp_span_entry_tunnel_parms_common(l3edev, saddr, daddr, gw,
+						       tparm.hop_limit,
+						       &nd_tbl, sparmsp);
+}
+
+static int
+mlxsw_sp_span_entry_gretap6_configure(struct mlxsw_sp_span_entry *span_entry,
+				      struct mlxsw_sp_span_parms sparms)
+{
+	struct mlxsw_sp_port *dest_port = sparms.dest_port;
+	struct mlxsw_sp *mlxsw_sp = dest_port->mlxsw_sp;
+	u16 local_port = dest_port->local_port;
+	char mpat_pl[MLXSW_REG_MPAT_LEN];
+	int pa_id = span_entry->id;
+
+	/* Create a new port analayzer entry for local_port. */
+	mlxsw_reg_mpat_pack(mpat_pl, pa_id, local_port, true,
+			    MLXSW_REG_MPAT_SPAN_TYPE_REMOTE_ETH_L3);
+	mlxsw_reg_mpat_pide_set(mpat_pl, sparms.policer_enable);
+	mlxsw_reg_mpat_pid_set(mpat_pl, sparms.policer_id);
+	mlxsw_reg_mpat_eth_rspan_pack(mpat_pl, sparms.vid);
+	mlxsw_reg_mpat_eth_rspan_l2_pack(mpat_pl,
+				    MLXSW_REG_MPAT_ETH_RSPAN_VERSION_NO_HEADER,
+				    sparms.dmac, !!sparms.vid);
+	mlxsw_reg_mpat_eth_rspan_l3_ipv6_pack(mpat_pl, sparms.ttl, sparms.smac,
+					      sparms.saddr.addr6,
+					      sparms.daddr.addr6);
+
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(mpat), mpat_pl);
+}
+
+static void
+mlxsw_sp_span_entry_gretap6_deconfigure(struct mlxsw_sp_span_entry *span_entry)
+{
+	mlxsw_sp_span_entry_deconfigure_common(span_entry,
+					MLXSW_REG_MPAT_SPAN_TYPE_REMOTE_ETH_L3);
+}
+
+static const
+struct mlxsw_sp_span_entry_ops mlxsw_sp_span_entry_ops_gretap6 = {
+	.can_handle = netif_is_ip6gretap,
+	.parms_set = mlxsw_sp_span_entry_gretap6_parms,
+	.configure = mlxsw_sp_span_entry_gretap6_configure,
+	.deconfigure = mlxsw_sp_span_entry_gretap6_deconfigure,
+};
+#endif
+
+static bool
+mlxsw_sp_span_vlan_can_handle(const struct net_device *dev)
+{
+	return is_vlan_dev(dev) &&
+	       mlxsw_sp_port_dev_check(vlan_dev_real_dev(dev));
+}
+
+static int
+mlxsw_sp_span_entry_vlan_parms(struct mlxsw_sp *mlxsw_sp,
+			       const struct net_device *to_dev,
+			       struct mlxsw_sp_span_parms *sparmsp)
+{
+	struct net_device *real_dev;
+	u16 vid;
+
+	if (!(to_dev->flags & IFF_UP))
+		return mlxsw_sp_span_entry_unoffloadable(sparmsp);
+
+	real_dev = mlxsw_sp_span_entry_vlan(to_dev, &vid);
+	sparmsp->dest_port = netdev_priv(real_dev);
+	sparmsp->vid = vid;
+	return 0;
+}
+
+static int
+mlxsw_sp_span_entry_vlan_configure(struct mlxsw_sp_span_entry *span_entry,
+				   struct mlxsw_sp_span_parms sparms)
+{
+	struct mlxsw_sp_port *dest_port = sparms.dest_port;
+	struct mlxsw_sp *mlxsw_sp = dest_port->mlxsw_sp;
+	u16 local_port = dest_port->local_port;
+	char mpat_pl[MLXSW_REG_MPAT_LEN];
+	int pa_id = span_entry->id;
+
+	mlxsw_reg_mpat_pack(mpat_pl, pa_id, local_port, true,
+			    MLXSW_REG_MPAT_SPAN_TYPE_REMOTE_ETH);
+	mlxsw_reg_mpat_pide_set(mpat_pl, sparms.policer_enable);
+	mlxsw_reg_mpat_pid_set(mpat_pl, sparms.policer_id);
+	mlxsw_reg_mpat_eth_rspan_pack(mpat_pl, sparms.vid);
+
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(mpat), mpat_pl);
+}
+
+static void
+mlxsw_sp_span_entry_vlan_deconfigure(struct mlxsw_sp_span_entry *span_entry)
+{
+	mlxsw_sp_span_entry_deconfigure_common(span_entry,
+					MLXSW_REG_MPAT_SPAN_TYPE_REMOTE_ETH);
+}
+
+static const
+struct mlxsw_sp_span_entry_ops mlxsw_sp_span_entry_ops_vlan = {
+	.can_handle = mlxsw_sp_span_vlan_can_handle,
+	.parms_set = mlxsw_sp_span_entry_vlan_parms,
+	.configure = mlxsw_sp_span_entry_vlan_configure,
+	.deconfigure = mlxsw_sp_span_entry_vlan_deconfigure,
+};
+
+static const
+struct mlxsw_sp_span_entry_ops *mlxsw_sp1_span_entry_ops_arr[] = {
+	&mlxsw_sp1_span_entry_ops_cpu,
+	&mlxsw_sp_span_entry_ops_phys,
+#if IS_ENABLED(CONFIG_NET_IPGRE)
+	&mlxsw_sp_span_entry_ops_gretap4,
+#endif
+#if IS_ENABLED(CONFIG_IPV6_GRE)
+	&mlxsw_sp_span_entry_ops_gretap6,
+#endif
+	&mlxsw_sp_span_entry_ops_vlan,
+};
+
+static bool mlxsw_sp2_span_cpu_can_handle(const struct net_device *dev)
+{
+	return !dev;
+}
+
+static int mlxsw_sp2_span_entry_cpu_parms(struct mlxsw_sp *mlxsw_sp,
+					  const struct net_device *to_dev,
+					  struct mlxsw_sp_span_parms *sparmsp)
+{
+	sparmsp->dest_port = mlxsw_sp->ports[MLXSW_PORT_CPU_PORT];
+	return 0;
+}
+
+static int
+mlxsw_sp2_span_entry_cpu_configure(struct mlxsw_sp_span_entry *span_entry,
+				   struct mlxsw_sp_span_parms sparms)
+{
+	/* Mirroring to the CPU port is like mirroring to any other physical
+	 * port. Its local port is used instead of that of the physical port.
+	 */
+	return mlxsw_sp_span_entry_phys_configure(span_entry, sparms);
+}
+
+static void
+mlxsw_sp2_span_entry_cpu_deconfigure(struct mlxsw_sp_span_entry *span_entry)
+{
+	enum mlxsw_reg_mpat_span_type span_type;
+
+	span_type = MLXSW_REG_MPAT_SPAN_TYPE_LOCAL_ETH;
+	mlxsw_sp_span_entry_deconfigure_common(span_entry, span_type);
+}
+
+static const
+struct mlxsw_sp_span_entry_ops mlxsw_sp2_span_entry_ops_cpu = {
+	.is_static = true,
+	.can_handle = mlxsw_sp2_span_cpu_can_handle,
+	.parms_set = mlxsw_sp2_span_entry_cpu_parms,
+	.configure = mlxsw_sp2_span_entry_cpu_configure,
+	.deconfigure = mlxsw_sp2_span_entry_cpu_deconfigure,
+};
+
+static const
+struct mlxsw_sp_span_entry_ops *mlxsw_sp2_span_entry_ops_arr[] = {
+	&mlxsw_sp2_span_entry_ops_cpu,
+	&mlxsw_sp_span_entry_ops_phys,
+#if IS_ENABLED(CONFIG_NET_IPGRE)
+	&mlxsw_sp_span_entry_ops_gretap4,
+#endif
+#if IS_ENABLED(CONFIG_IPV6_GRE)
+	&mlxsw_sp_span_entry_ops_gretap6,
+#endif
+	&mlxsw_sp_span_entry_ops_vlan,
+};
+
+static int
+mlxsw_sp_span_entry_nop_parms(struct mlxsw_sp *mlxsw_sp,
+			      const struct net_device *to_dev,
+			      struct mlxsw_sp_span_parms *sparmsp)
+{
+	return mlxsw_sp_span_entry_unoffloadable(sparmsp);
+}
+
+static int
+mlxsw_sp_span_entry_nop_configure(struct mlxsw_sp_span_entry *span_entry,
+				  struct mlxsw_sp_span_parms sparms)
+{
+	return 0;
+}
+
+static void
+mlxsw_sp_span_entry_nop_deconfigure(struct mlxsw_sp_span_entry *span_entry)
+{
+}
+
+static const struct mlxsw_sp_span_entry_ops mlxsw_sp_span_entry_ops_nop = {
+	.parms_set = mlxsw_sp_span_entry_nop_parms,
+	.configure = mlxsw_sp_span_entry_nop_configure,
+	.deconfigure = mlxsw_sp_span_entry_nop_deconfigure,
+};
+
+static void
+mlxsw_sp_span_entry_configure(struct mlxsw_sp *mlxsw_sp,
+			      struct mlxsw_sp_span_entry *span_entry,
+			      struct mlxsw_sp_span_parms sparms)
+{
+	int err;
+
+	if (!sparms.dest_port)
+		goto set_parms;
+
+	if (sparms.dest_port->mlxsw_sp != mlxsw_sp) {
+		dev_err(mlxsw_sp->bus_info->dev,
+			"Cannot mirror to a port which belongs to a different mlxsw instance\n");
+		sparms.dest_port = NULL;
+		goto set_parms;
+	}
+
+	err = span_entry->ops->configure(span_entry, sparms);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Failed to offload mirror\n");
+		sparms.dest_port = NULL;
+		goto set_parms;
+	}
+
+set_parms:
+	span_entry->parms = sparms;
+}
+
+static void
+mlxsw_sp_span_entry_deconfigure(struct mlxsw_sp_span_entry *span_entry)
+{
+	if (span_entry->parms.dest_port)
+		span_entry->ops->deconfigure(span_entry);
+}
+
+static int mlxsw_sp_span_policer_id_base_set(struct mlxsw_sp_span *span,
+					     u16 policer_id)
+{
+	struct mlxsw_sp *mlxsw_sp = span->mlxsw_sp;
+	u16 policer_id_base;
+	int err;
+
+	/* Policers set on SPAN agents must be in the range of
+	 * `policer_id_base .. policer_id_base + max_span_agents - 1`. If the
+	 * base is set and the new policer is not within the range, then we
+	 * must error out.
+	 */
+	if (refcount_read(&span->policer_id_base_ref_count)) {
+		if (policer_id < span->policer_id_base ||
+		    policer_id >= span->policer_id_base + span->entries_count)
+			return -EINVAL;
+
+		refcount_inc(&span->policer_id_base_ref_count);
+		return 0;
+	}
+
+	/* Base must be even. */
+	policer_id_base = policer_id % 2 == 0 ? policer_id : policer_id - 1;
+	err = mlxsw_sp->span_ops->policer_id_base_set(mlxsw_sp,
+						      policer_id_base);
+	if (err)
+		return err;
+
+	span->policer_id_base = policer_id_base;
+	refcount_set(&span->policer_id_base_ref_count, 1);
+
+	return 0;
+}
+
+static void mlxsw_sp_span_policer_id_base_unset(struct mlxsw_sp_span *span)
+{
+	if (refcount_dec_and_test(&span->policer_id_base_ref_count))
+		span->policer_id_base = 0;
+}
+
+static struct mlxsw_sp_span_entry *
+mlxsw_sp_span_entry_create(struct mlxsw_sp *mlxsw_sp,
+			   const struct net_device *to_dev,
+			   const struct mlxsw_sp_span_entry_ops *ops,
+			   struct mlxsw_sp_span_parms sparms)
+{
+	struct mlxsw_sp_span_entry *span_entry = NULL;
+	int i;
+
+	/* find a free entry to use */
+	for (i = 0; i < mlxsw_sp->span->entries_count; i++) {
+		if (!refcount_read(&mlxsw_sp->span->entries[i].ref_count)) {
+			span_entry = &mlxsw_sp->span->entries[i];
+			break;
+		}
+	}
+	if (!span_entry)
+		return NULL;
+
+	if (sparms.policer_enable) {
+		int err;
+
+		err = mlxsw_sp_span_policer_id_base_set(mlxsw_sp->span,
+							sparms.policer_id);
+		if (err)
+			return NULL;
+	}
+
+	atomic_inc(&mlxsw_sp->span->active_entries_count);
+	span_entry->ops = ops;
+	refcount_set(&span_entry->ref_count, 1);
+	span_entry->to_dev = to_dev;
+	mlxsw_sp_span_entry_configure(mlxsw_sp, span_entry, sparms);
+
+	return span_entry;
+}
+
+static void mlxsw_sp_span_entry_destroy(struct mlxsw_sp *mlxsw_sp,
+					struct mlxsw_sp_span_entry *span_entry)
+{
+	mlxsw_sp_span_entry_deconfigure(span_entry);
+	atomic_dec(&mlxsw_sp->span->active_entries_count);
+	if (span_entry->parms.policer_enable)
+		mlxsw_sp_span_policer_id_base_unset(mlxsw_sp->span);
+}
+
+struct mlxsw_sp_span_entry *
+mlxsw_sp_span_entry_find_by_port(struct mlxsw_sp *mlxsw_sp,
+				 const struct net_device *to_dev)
+{
+	int i;
+
+	for (i = 0; i < mlxsw_sp->span->entries_count; i++) {
+		struct mlxsw_sp_span_entry *curr = &mlxsw_sp->span->entries[i];
+
+		if (refcount_read(&curr->ref_count) && curr->to_dev == to_dev)
+			return curr;
+	}
+	return NULL;
+}
+
+void mlxsw_sp_span_entry_invalidate(struct mlxsw_sp *mlxsw_sp,
+				    struct mlxsw_sp_span_entry *span_entry)
+{
+	mlxsw_sp_span_entry_deconfigure(span_entry);
+	span_entry->ops = &mlxsw_sp_span_entry_ops_nop;
+}
+
+static struct mlxsw_sp_span_entry *
+mlxsw_sp_span_entry_find_by_id(struct mlxsw_sp *mlxsw_sp, int span_id)
+{
+	int i;
+
+	for (i = 0; i < mlxsw_sp->span->entries_count; i++) {
+		struct mlxsw_sp_span_entry *curr = &mlxsw_sp->span->entries[i];
+
+		if (refcount_read(&curr->ref_count) && curr->id == span_id)
+			return curr;
+	}
+	return NULL;
+}
+
+static struct mlxsw_sp_span_entry *
+mlxsw_sp_span_entry_find_by_parms(struct mlxsw_sp *mlxsw_sp,
+				  const struct net_device *to_dev,
+				  const struct mlxsw_sp_span_parms *sparms)
+{
+	int i;
+
+	for (i = 0; i < mlxsw_sp->span->entries_count; i++) {
+		struct mlxsw_sp_span_entry *curr = &mlxsw_sp->span->entries[i];
+
+		if (refcount_read(&curr->ref_count) && curr->to_dev == to_dev &&
+		    curr->parms.policer_enable == sparms->policer_enable &&
+		    curr->parms.policer_id == sparms->policer_id &&
+		    curr->parms.session_id == sparms->session_id)
+			return curr;
+	}
+	return NULL;
+}
+
+static struct mlxsw_sp_span_entry *
+mlxsw_sp_span_entry_get(struct mlxsw_sp *mlxsw_sp,
+			const struct net_device *to_dev,
+			const struct mlxsw_sp_span_entry_ops *ops,
+			struct mlxsw_sp_span_parms sparms)
+{
+	struct mlxsw_sp_span_entry *span_entry;
+
+	span_entry = mlxsw_sp_span_entry_find_by_parms(mlxsw_sp, to_dev,
+						       &sparms);
+	if (span_entry) {
+		/* Already exists, just take a reference */
+		refcount_inc(&span_entry->ref_count);
+		return span_entry;
+	}
+
+	return mlxsw_sp_span_entry_create(mlxsw_sp, to_dev, ops, sparms);
+}
+
+static int mlxsw_sp_span_entry_put(struct mlxsw_sp *mlxsw_sp,
+				   struct mlxsw_sp_span_entry *span_entry)
+{
+	if (refcount_dec_and_test(&span_entry->ref_count))
+		mlxsw_sp_span_entry_destroy(mlxsw_sp, span_entry);
+	return 0;
+}
+
+static int mlxsw_sp_span_port_buffer_update(struct mlxsw_sp_port *mlxsw_sp_port, bool enable)
+{
+	struct mlxsw_sp_hdroom hdroom;
+
+	hdroom = *mlxsw_sp_port->hdroom;
+	hdroom.int_buf.enable = enable;
+	mlxsw_sp_hdroom_bufs_reset_sizes(mlxsw_sp_port, &hdroom);
+
+	return mlxsw_sp_hdroom_configure(mlxsw_sp_port, &hdroom);
+}
+
+static int
+mlxsw_sp_span_port_buffer_enable(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	return mlxsw_sp_span_port_buffer_update(mlxsw_sp_port, true);
+}
+
+static void mlxsw_sp_span_port_buffer_disable(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	mlxsw_sp_span_port_buffer_update(mlxsw_sp_port, false);
+}
+
+static struct mlxsw_sp_span_analyzed_port *
+mlxsw_sp_span_analyzed_port_find(struct mlxsw_sp_span *span, u16 local_port,
+				 bool ingress)
+{
+	struct mlxsw_sp_span_analyzed_port *analyzed_port;
+
+	list_for_each_entry(analyzed_port, &span->analyzed_ports_list, list) {
+		if (analyzed_port->local_port == local_port &&
+		    analyzed_port->ingress == ingress)
+			return analyzed_port;
+	}
+
+	return NULL;
+}
+
+static const struct mlxsw_sp_span_entry_ops *
+mlxsw_sp_span_entry_ops(struct mlxsw_sp *mlxsw_sp,
+			const struct net_device *to_dev)
+{
+	struct mlxsw_sp_span *span = mlxsw_sp->span;
+	size_t i;
+
+	for (i = 0; i < span->span_entry_ops_arr_size; ++i)
+		if (span->span_entry_ops_arr[i]->can_handle(to_dev))
+			return span->span_entry_ops_arr[i];
+
+	return NULL;
+}
+
+static void mlxsw_sp_span_respin_work(struct work_struct *work)
+{
+	struct mlxsw_sp_span *span;
+	struct mlxsw_sp *mlxsw_sp;
+	int i, err;
+
+	span = container_of(work, struct mlxsw_sp_span, work);
+	mlxsw_sp = span->mlxsw_sp;
+
+	rtnl_lock();
+	for (i = 0; i < mlxsw_sp->span->entries_count; i++) {
+		struct mlxsw_sp_span_entry *curr = &mlxsw_sp->span->entries[i];
+		struct mlxsw_sp_span_parms sparms = {NULL};
+
+		if (!refcount_read(&curr->ref_count))
+			continue;
+
+		if (curr->ops->is_static)
+			continue;
+
+		err = curr->ops->parms_set(mlxsw_sp, curr->to_dev, &sparms);
+		if (err)
+			continue;
+
+		if (memcmp(&sparms, &curr->parms, sizeof(sparms))) {
+			mlxsw_sp_span_entry_deconfigure(curr);
+			mlxsw_sp_span_entry_configure(mlxsw_sp, curr, sparms);
+		}
+	}
+	rtnl_unlock();
+}
+
+void mlxsw_sp_span_respin(struct mlxsw_sp *mlxsw_sp)
+{
+	if (atomic_read(&mlxsw_sp->span->active_entries_count) == 0)
+		return;
+	mlxsw_core_schedule_work(&mlxsw_sp->span->work);
+}
+
+int mlxsw_sp_span_agent_get(struct mlxsw_sp *mlxsw_sp, int *p_span_id,
+			    const struct mlxsw_sp_span_agent_parms *parms)
+{
+	const struct net_device *to_dev = parms->to_dev;
+	const struct mlxsw_sp_span_entry_ops *ops;
+	struct mlxsw_sp_span_entry *span_entry;
+	struct mlxsw_sp_span_parms sparms;
+	int err;
+
+	ASSERT_RTNL();
+
+	ops = mlxsw_sp_span_entry_ops(mlxsw_sp, to_dev);
+	if (!ops) {
+		dev_err(mlxsw_sp->bus_info->dev, "Cannot mirror to requested destination\n");
+		return -EOPNOTSUPP;
+	}
+
+	memset(&sparms, 0, sizeof(sparms));
+	err = ops->parms_set(mlxsw_sp, to_dev, &sparms);
+	if (err)
+		return err;
+
+	sparms.policer_id = parms->policer_id;
+	sparms.policer_enable = parms->policer_enable;
+	sparms.session_id = parms->session_id;
+	span_entry = mlxsw_sp_span_entry_get(mlxsw_sp, to_dev, ops, sparms);
+	if (!span_entry)
+		return -ENOBUFS;
+
+	*p_span_id = span_entry->id;
+
+	return 0;
+}
+
+void mlxsw_sp_span_agent_put(struct mlxsw_sp *mlxsw_sp, int span_id)
+{
+	struct mlxsw_sp_span_entry *span_entry;
+
+	ASSERT_RTNL();
+
+	span_entry = mlxsw_sp_span_entry_find_by_id(mlxsw_sp, span_id);
+	if (WARN_ON_ONCE(!span_entry))
+		return;
+
+	mlxsw_sp_span_entry_put(mlxsw_sp, span_entry);
+}
+
+static struct mlxsw_sp_span_analyzed_port *
+mlxsw_sp_span_analyzed_port_create(struct mlxsw_sp_span *span,
+				   struct mlxsw_sp_port *mlxsw_sp_port,
+				   bool ingress)
+{
+	struct mlxsw_sp_span_analyzed_port *analyzed_port;
+	int err;
+
+	analyzed_port = kzalloc(sizeof(*analyzed_port), GFP_KERNEL);
+	if (!analyzed_port)
+		return ERR_PTR(-ENOMEM);
+
+	refcount_set(&analyzed_port->ref_count, 1);
+	analyzed_port->local_port = mlxsw_sp_port->local_port;
+	analyzed_port->ingress = ingress;
+	list_add_tail(&analyzed_port->list, &span->analyzed_ports_list);
+
+	/* An egress mirror buffer should be allocated on the egress port which
+	 * does the mirroring.
+	 */
+	if (!ingress) {
+		err = mlxsw_sp_span_port_buffer_enable(mlxsw_sp_port);
+		if (err)
+			goto err_buffer_update;
+	}
+
+	return analyzed_port;
+
+err_buffer_update:
+	list_del(&analyzed_port->list);
+	kfree(analyzed_port);
+	return ERR_PTR(err);
+}
+
+static void
+mlxsw_sp_span_analyzed_port_destroy(struct mlxsw_sp_port *mlxsw_sp_port,
+				    struct mlxsw_sp_span_analyzed_port *
+				    analyzed_port)
+{
+	/* Remove egress mirror buffer now that port is no longer analyzed
+	 * at egress.
+	 */
+	if (!analyzed_port->ingress)
+		mlxsw_sp_span_port_buffer_disable(mlxsw_sp_port);
+
+	list_del(&analyzed_port->list);
+	kfree(analyzed_port);
+}
+
+int mlxsw_sp_span_analyzed_port_get(struct mlxsw_sp_port *mlxsw_sp_port,
+				    bool ingress)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	struct mlxsw_sp_span_analyzed_port *analyzed_port;
+	u16 local_port = mlxsw_sp_port->local_port;
+	int err = 0;
+
+	mutex_lock(&mlxsw_sp->span->analyzed_ports_lock);
+
+	analyzed_port = mlxsw_sp_span_analyzed_port_find(mlxsw_sp->span,
+							 local_port, ingress);
+	if (analyzed_port) {
+		refcount_inc(&analyzed_port->ref_count);
+		goto out_unlock;
+	}
+
+	analyzed_port = mlxsw_sp_span_analyzed_port_create(mlxsw_sp->span,
+							   mlxsw_sp_port,
+							   ingress);
+	if (IS_ERR(analyzed_port))
+		err = PTR_ERR(analyzed_port);
+
+out_unlock:
+	mutex_unlock(&mlxsw_sp->span->analyzed_ports_lock);
+	return err;
+}
+
+void mlxsw_sp_span_analyzed_port_put(struct mlxsw_sp_port *mlxsw_sp_port,
+				     bool ingress)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	struct mlxsw_sp_span_analyzed_port *analyzed_port;
+	u16 local_port = mlxsw_sp_port->local_port;
+
+	mutex_lock(&mlxsw_sp->span->analyzed_ports_lock);
+
+	analyzed_port = mlxsw_sp_span_analyzed_port_find(mlxsw_sp->span,
+							 local_port, ingress);
+	if (WARN_ON_ONCE(!analyzed_port))
+		goto out_unlock;
+
+	if (!refcount_dec_and_test(&analyzed_port->ref_count))
+		goto out_unlock;
+
+	mlxsw_sp_span_analyzed_port_destroy(mlxsw_sp_port, analyzed_port);
+
+out_unlock:
+	mutex_unlock(&mlxsw_sp->span->analyzed_ports_lock);
+}
+
+static int
+__mlxsw_sp_span_trigger_port_bind(struct mlxsw_sp_span *span,
+				  struct mlxsw_sp_span_trigger_entry *
+				  trigger_entry, bool enable)
+{
+	char mpar_pl[MLXSW_REG_MPAR_LEN];
+	enum mlxsw_reg_mpar_i_e i_e;
+
+	switch (trigger_entry->trigger) {
+	case MLXSW_SP_SPAN_TRIGGER_INGRESS:
+		i_e = MLXSW_REG_MPAR_TYPE_INGRESS;
+		break;
+	case MLXSW_SP_SPAN_TRIGGER_EGRESS:
+		i_e = MLXSW_REG_MPAR_TYPE_EGRESS;
+		break;
+	default:
+		WARN_ON_ONCE(1);
+		return -EINVAL;
+	}
+
+	if (trigger_entry->parms.probability_rate > MLXSW_REG_MPAR_RATE_MAX)
+		return -EINVAL;
+
+	mlxsw_reg_mpar_pack(mpar_pl, trigger_entry->local_port, i_e, enable,
+			    trigger_entry->parms.span_id,
+			    trigger_entry->parms.probability_rate);
+	return mlxsw_reg_write(span->mlxsw_sp->core, MLXSW_REG(mpar), mpar_pl);
+}
+
+static int
+mlxsw_sp_span_trigger_port_bind(struct mlxsw_sp_span_trigger_entry *
+				trigger_entry)
+{
+	return __mlxsw_sp_span_trigger_port_bind(trigger_entry->span,
+						 trigger_entry, true);
+}
+
+static void
+mlxsw_sp_span_trigger_port_unbind(struct mlxsw_sp_span_trigger_entry *
+				  trigger_entry)
+{
+	__mlxsw_sp_span_trigger_port_bind(trigger_entry->span, trigger_entry,
+					  false);
+}
+
+static bool
+mlxsw_sp_span_trigger_port_matches(struct mlxsw_sp_span_trigger_entry *
+				   trigger_entry,
+				   enum mlxsw_sp_span_trigger trigger,
+				   struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	return trigger_entry->trigger == trigger &&
+	       trigger_entry->local_port == mlxsw_sp_port->local_port;
+}
+
+static int
+mlxsw_sp_span_trigger_port_enable(struct mlxsw_sp_span_trigger_entry *
+				  trigger_entry,
+				  struct mlxsw_sp_port *mlxsw_sp_port, u8 tc)
+{
+	/* Port trigger are enabled during binding. */
+	return 0;
+}
+
+static void
+mlxsw_sp_span_trigger_port_disable(struct mlxsw_sp_span_trigger_entry *
+				   trigger_entry,
+				   struct mlxsw_sp_port *mlxsw_sp_port, u8 tc)
+{
+}
+
+static const struct mlxsw_sp_span_trigger_ops
+mlxsw_sp_span_trigger_port_ops = {
+	.bind = mlxsw_sp_span_trigger_port_bind,
+	.unbind = mlxsw_sp_span_trigger_port_unbind,
+	.matches = mlxsw_sp_span_trigger_port_matches,
+	.enable = mlxsw_sp_span_trigger_port_enable,
+	.disable = mlxsw_sp_span_trigger_port_disable,
+};
+
+static int
+mlxsw_sp1_span_trigger_global_bind(struct mlxsw_sp_span_trigger_entry *
+				   trigger_entry)
+{
+	return -EOPNOTSUPP;
+}
+
+static void
+mlxsw_sp1_span_trigger_global_unbind(struct mlxsw_sp_span_trigger_entry *
+				     trigger_entry)
+{
+}
+
+static bool
+mlxsw_sp1_span_trigger_global_matches(struct mlxsw_sp_span_trigger_entry *
+				      trigger_entry,
+				      enum mlxsw_sp_span_trigger trigger,
+				      struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	WARN_ON_ONCE(1);
+	return false;
+}
+
+static int
+mlxsw_sp1_span_trigger_global_enable(struct mlxsw_sp_span_trigger_entry *
+				     trigger_entry,
+				     struct mlxsw_sp_port *mlxsw_sp_port,
+				     u8 tc)
+{
+	return -EOPNOTSUPP;
+}
+
+static void
+mlxsw_sp1_span_trigger_global_disable(struct mlxsw_sp_span_trigger_entry *
+				      trigger_entry,
+				      struct mlxsw_sp_port *mlxsw_sp_port,
+				      u8 tc)
+{
+}
+
+static const struct mlxsw_sp_span_trigger_ops
+mlxsw_sp1_span_trigger_global_ops = {
+	.bind = mlxsw_sp1_span_trigger_global_bind,
+	.unbind = mlxsw_sp1_span_trigger_global_unbind,
+	.matches = mlxsw_sp1_span_trigger_global_matches,
+	.enable = mlxsw_sp1_span_trigger_global_enable,
+	.disable = mlxsw_sp1_span_trigger_global_disable,
+};
+
+static const struct mlxsw_sp_span_trigger_ops *
+mlxsw_sp1_span_trigger_ops_arr[] = {
+	[MLXSW_SP_SPAN_TRIGGER_TYPE_PORT] = &mlxsw_sp_span_trigger_port_ops,
+	[MLXSW_SP_SPAN_TRIGGER_TYPE_GLOBAL] =
+		&mlxsw_sp1_span_trigger_global_ops,
+};
+
+static int
+mlxsw_sp2_span_trigger_global_bind(struct mlxsw_sp_span_trigger_entry *
+				   trigger_entry)
+{
+	struct mlxsw_sp *mlxsw_sp = trigger_entry->span->mlxsw_sp;
+	enum mlxsw_reg_mpagr_trigger trigger;
+	char mpagr_pl[MLXSW_REG_MPAGR_LEN];
+
+	switch (trigger_entry->trigger) {
+	case MLXSW_SP_SPAN_TRIGGER_TAIL_DROP:
+		trigger = MLXSW_REG_MPAGR_TRIGGER_INGRESS_SHARED_BUFFER;
+		break;
+	case MLXSW_SP_SPAN_TRIGGER_EARLY_DROP:
+		trigger = MLXSW_REG_MPAGR_TRIGGER_INGRESS_WRED;
+		break;
+	case MLXSW_SP_SPAN_TRIGGER_ECN:
+		trigger = MLXSW_REG_MPAGR_TRIGGER_EGRESS_ECN;
+		break;
+	default:
+		WARN_ON_ONCE(1);
+		return -EINVAL;
+	}
+
+	if (trigger_entry->parms.probability_rate > MLXSW_REG_MPAGR_RATE_MAX)
+		return -EINVAL;
+
+	mlxsw_reg_mpagr_pack(mpagr_pl, trigger, trigger_entry->parms.span_id,
+			     trigger_entry->parms.probability_rate);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(mpagr), mpagr_pl);
+}
+
+static void
+mlxsw_sp2_span_trigger_global_unbind(struct mlxsw_sp_span_trigger_entry *
+				     trigger_entry)
+{
+	/* There is no unbinding for global triggers. The trigger should be
+	 * disabled on all ports by now.
+	 */
+}
+
+static bool
+mlxsw_sp2_span_trigger_global_matches(struct mlxsw_sp_span_trigger_entry *
+				      trigger_entry,
+				      enum mlxsw_sp_span_trigger trigger,
+				      struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	return trigger_entry->trigger == trigger;
+}
+
+static int
+__mlxsw_sp2_span_trigger_global_enable(struct mlxsw_sp_span_trigger_entry *
+				       trigger_entry,
+				       struct mlxsw_sp_port *mlxsw_sp_port,
+				       u8 tc, bool enable)
+{
+	struct mlxsw_sp *mlxsw_sp = trigger_entry->span->mlxsw_sp;
+	char momte_pl[MLXSW_REG_MOMTE_LEN];
+	enum mlxsw_reg_momte_type type;
+	int err;
+
+	switch (trigger_entry->trigger) {
+	case MLXSW_SP_SPAN_TRIGGER_TAIL_DROP:
+		type = MLXSW_REG_MOMTE_TYPE_SHARED_BUFFER_TCLASS;
+		break;
+	case MLXSW_SP_SPAN_TRIGGER_EARLY_DROP:
+		type = MLXSW_REG_MOMTE_TYPE_WRED;
+		break;
+	case MLXSW_SP_SPAN_TRIGGER_ECN:
+		type = MLXSW_REG_MOMTE_TYPE_ECN;
+		break;
+	default:
+		WARN_ON_ONCE(1);
+		return -EINVAL;
+	}
+
+	/* Query existing configuration in order to only change the state of
+	 * the specified traffic class.
+	 */
+	mlxsw_reg_momte_pack(momte_pl, mlxsw_sp_port->local_port, type);
+	err = mlxsw_reg_query(mlxsw_sp->core, MLXSW_REG(momte), momte_pl);
+	if (err)
+		return err;
+
+	mlxsw_reg_momte_tclass_en_set(momte_pl, tc, enable);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(momte), momte_pl);
+}
+
+static int
+mlxsw_sp2_span_trigger_global_enable(struct mlxsw_sp_span_trigger_entry *
+				     trigger_entry,
+				     struct mlxsw_sp_port *mlxsw_sp_port,
+				     u8 tc)
+{
+	return __mlxsw_sp2_span_trigger_global_enable(trigger_entry,
+						      mlxsw_sp_port, tc, true);
+}
+
+static void
+mlxsw_sp2_span_trigger_global_disable(struct mlxsw_sp_span_trigger_entry *
+				      trigger_entry,
+				      struct mlxsw_sp_port *mlxsw_sp_port,
+				      u8 tc)
+{
+	__mlxsw_sp2_span_trigger_global_enable(trigger_entry, mlxsw_sp_port, tc,
+					       false);
+}
+
+static const struct mlxsw_sp_span_trigger_ops
+mlxsw_sp2_span_trigger_global_ops = {
+	.bind = mlxsw_sp2_span_trigger_global_bind,
+	.unbind = mlxsw_sp2_span_trigger_global_unbind,
+	.matches = mlxsw_sp2_span_trigger_global_matches,
+	.enable = mlxsw_sp2_span_trigger_global_enable,
+	.disable = mlxsw_sp2_span_trigger_global_disable,
+};
+
+static const struct mlxsw_sp_span_trigger_ops *
+mlxsw_sp2_span_trigger_ops_arr[] = {
+	[MLXSW_SP_SPAN_TRIGGER_TYPE_PORT] = &mlxsw_sp_span_trigger_port_ops,
+	[MLXSW_SP_SPAN_TRIGGER_TYPE_GLOBAL] =
+		&mlxsw_sp2_span_trigger_global_ops,
+};
+
+static void
+mlxsw_sp_span_trigger_ops_set(struct mlxsw_sp_span_trigger_entry *trigger_entry)
+{
+	struct mlxsw_sp_span *span = trigger_entry->span;
+	enum mlxsw_sp_span_trigger_type type;
+
+	switch (trigger_entry->trigger) {
+	case MLXSW_SP_SPAN_TRIGGER_INGRESS:
+	case MLXSW_SP_SPAN_TRIGGER_EGRESS:
+		type = MLXSW_SP_SPAN_TRIGGER_TYPE_PORT;
+		break;
+	case MLXSW_SP_SPAN_TRIGGER_TAIL_DROP:
+	case MLXSW_SP_SPAN_TRIGGER_EARLY_DROP:
+	case MLXSW_SP_SPAN_TRIGGER_ECN:
+		type = MLXSW_SP_SPAN_TRIGGER_TYPE_GLOBAL;
+		break;
+	default:
+		WARN_ON_ONCE(1);
+		return;
+	}
+
+	trigger_entry->ops = span->span_trigger_ops_arr[type];
+}
+
+static struct mlxsw_sp_span_trigger_entry *
+mlxsw_sp_span_trigger_entry_create(struct mlxsw_sp_span *span,
+				   enum mlxsw_sp_span_trigger trigger,
+				   struct mlxsw_sp_port *mlxsw_sp_port,
+				   const struct mlxsw_sp_span_trigger_parms
+				   *parms)
+{
+	struct mlxsw_sp_span_trigger_entry *trigger_entry;
+	int err;
+
+	trigger_entry = kzalloc(sizeof(*trigger_entry), GFP_KERNEL);
+	if (!trigger_entry)
+		return ERR_PTR(-ENOMEM);
+
+	refcount_set(&trigger_entry->ref_count, 1);
+	trigger_entry->local_port = mlxsw_sp_port ? mlxsw_sp_port->local_port :
+						    0;
+	trigger_entry->trigger = trigger;
+	memcpy(&trigger_entry->parms, parms, sizeof(trigger_entry->parms));
+	trigger_entry->span = span;
+	mlxsw_sp_span_trigger_ops_set(trigger_entry);
+	list_add_tail(&trigger_entry->list, &span->trigger_entries_list);
+
+	err = trigger_entry->ops->bind(trigger_entry);
+	if (err)
+		goto err_trigger_entry_bind;
+
+	return trigger_entry;
+
+err_trigger_entry_bind:
+	list_del(&trigger_entry->list);
+	kfree(trigger_entry);
+	return ERR_PTR(err);
+}
+
+static void
+mlxsw_sp_span_trigger_entry_destroy(struct mlxsw_sp_span *span,
+				    struct mlxsw_sp_span_trigger_entry *
+				    trigger_entry)
+{
+	trigger_entry->ops->unbind(trigger_entry);
+	list_del(&trigger_entry->list);
+	kfree(trigger_entry);
+}
+
+static struct mlxsw_sp_span_trigger_entry *
+mlxsw_sp_span_trigger_entry_find(struct mlxsw_sp_span *span,
+				 enum mlxsw_sp_span_trigger trigger,
+				 struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	struct mlxsw_sp_span_trigger_entry *trigger_entry;
+
+	list_for_each_entry(trigger_entry, &span->trigger_entries_list, list) {
+		if (trigger_entry->ops->matches(trigger_entry, trigger,
+						mlxsw_sp_port))
+			return trigger_entry;
+	}
+
+	return NULL;
+}
+
+int mlxsw_sp_span_agent_bind(struct mlxsw_sp *mlxsw_sp,
+			     enum mlxsw_sp_span_trigger trigger,
+			     struct mlxsw_sp_port *mlxsw_sp_port,
+			     const struct mlxsw_sp_span_trigger_parms *parms)
+{
+	struct mlxsw_sp_span_trigger_entry *trigger_entry;
+	int err = 0;
+
+	ASSERT_RTNL();
+
+	if (!mlxsw_sp_span_entry_find_by_id(mlxsw_sp, parms->span_id))
+		return -EINVAL;
+
+	trigger_entry = mlxsw_sp_span_trigger_entry_find(mlxsw_sp->span,
+							 trigger,
+							 mlxsw_sp_port);
+	if (trigger_entry) {
+		if (trigger_entry->parms.span_id != parms->span_id ||
+		    trigger_entry->parms.probability_rate !=
+		    parms->probability_rate)
+			return -EINVAL;
+		refcount_inc(&trigger_entry->ref_count);
+		goto out;
+	}
+
+	trigger_entry = mlxsw_sp_span_trigger_entry_create(mlxsw_sp->span,
+							   trigger,
+							   mlxsw_sp_port,
+							   parms);
+	if (IS_ERR(trigger_entry))
+		err = PTR_ERR(trigger_entry);
+
+out:
+	return err;
+}
+
+void mlxsw_sp_span_agent_unbind(struct mlxsw_sp *mlxsw_sp,
+				enum mlxsw_sp_span_trigger trigger,
+				struct mlxsw_sp_port *mlxsw_sp_port,
+				const struct mlxsw_sp_span_trigger_parms *parms)
+{
+	struct mlxsw_sp_span_trigger_entry *trigger_entry;
+
+	ASSERT_RTNL();
+
+	if (WARN_ON_ONCE(!mlxsw_sp_span_entry_find_by_id(mlxsw_sp,
+							 parms->span_id)))
+		return;
+
+	trigger_entry = mlxsw_sp_span_trigger_entry_find(mlxsw_sp->span,
+							 trigger,
+							 mlxsw_sp_port);
+	if (WARN_ON_ONCE(!trigger_entry))
+		return;
+
+	if (!refcount_dec_and_test(&trigger_entry->ref_count))
+		return;
+
+	mlxsw_sp_span_trigger_entry_destroy(mlxsw_sp->span, trigger_entry);
+}
+
+int mlxsw_sp_span_trigger_enable(struct mlxsw_sp_port *mlxsw_sp_port,
+				 enum mlxsw_sp_span_trigger trigger, u8 tc)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	struct mlxsw_sp_span_trigger_entry *trigger_entry;
+
+	ASSERT_RTNL();
+
+	trigger_entry = mlxsw_sp_span_trigger_entry_find(mlxsw_sp->span,
+							 trigger,
+							 mlxsw_sp_port);
+	if (WARN_ON_ONCE(!trigger_entry))
+		return -EINVAL;
+
+	return trigger_entry->ops->enable(trigger_entry, mlxsw_sp_port, tc);
+}
+
+void mlxsw_sp_span_trigger_disable(struct mlxsw_sp_port *mlxsw_sp_port,
+				   enum mlxsw_sp_span_trigger trigger, u8 tc)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	struct mlxsw_sp_span_trigger_entry *trigger_entry;
+
+	ASSERT_RTNL();
+
+	trigger_entry = mlxsw_sp_span_trigger_entry_find(mlxsw_sp->span,
+							 trigger,
+							 mlxsw_sp_port);
+	if (WARN_ON_ONCE(!trigger_entry))
+		return;
+
+	return trigger_entry->ops->disable(trigger_entry, mlxsw_sp_port, tc);
+}
+
+bool mlxsw_sp_span_trigger_is_ingress(enum mlxsw_sp_span_trigger trigger)
+{
+	switch (trigger) {
+	case MLXSW_SP_SPAN_TRIGGER_INGRESS:
+	case MLXSW_SP_SPAN_TRIGGER_EARLY_DROP:
+	case MLXSW_SP_SPAN_TRIGGER_TAIL_DROP:
+		return true;
+	case MLXSW_SP_SPAN_TRIGGER_EGRESS:
+	case MLXSW_SP_SPAN_TRIGGER_ECN:
+		return false;
+	}
+
+	WARN_ON_ONCE(1);
+	return false;
+}
+
+static int mlxsw_sp1_span_init(struct mlxsw_sp *mlxsw_sp)
+{
+	size_t arr_size = ARRAY_SIZE(mlxsw_sp1_span_entry_ops_arr);
+
+	/* Must be first to avoid NULL pointer dereference by subsequent
+	 * can_handle() callbacks.
+	 */
+	if (WARN_ON(mlxsw_sp1_span_entry_ops_arr[0] !=
+		    &mlxsw_sp1_span_entry_ops_cpu))
+		return -EINVAL;
+
+	mlxsw_sp->span->span_trigger_ops_arr = mlxsw_sp1_span_trigger_ops_arr;
+	mlxsw_sp->span->span_entry_ops_arr = mlxsw_sp1_span_entry_ops_arr;
+	mlxsw_sp->span->span_entry_ops_arr_size = arr_size;
+
+	return 0;
+}
+
+static int mlxsw_sp1_span_policer_id_base_set(struct mlxsw_sp *mlxsw_sp,
+					      u16 policer_id_base)
+{
+	return -EOPNOTSUPP;
+}
+
+const struct mlxsw_sp_span_ops mlxsw_sp1_span_ops = {
+	.init = mlxsw_sp1_span_init,
+	.policer_id_base_set = mlxsw_sp1_span_policer_id_base_set,
+};
+
+static int mlxsw_sp2_span_init(struct mlxsw_sp *mlxsw_sp)
+{
+	size_t arr_size = ARRAY_SIZE(mlxsw_sp2_span_entry_ops_arr);
+
+	/* Must be first to avoid NULL pointer dereference by subsequent
+	 * can_handle() callbacks.
+	 */
+	if (WARN_ON(mlxsw_sp2_span_entry_ops_arr[0] !=
+		    &mlxsw_sp2_span_entry_ops_cpu))
+		return -EINVAL;
+
+	mlxsw_sp->span->span_trigger_ops_arr = mlxsw_sp2_span_trigger_ops_arr;
+	mlxsw_sp->span->span_entry_ops_arr = mlxsw_sp2_span_entry_ops_arr;
+	mlxsw_sp->span->span_entry_ops_arr_size = arr_size;
+
+	return 0;
+}
+
+#define MLXSW_SP2_SPAN_EG_MIRROR_BUFFER_FACTOR 38
+#define MLXSW_SP3_SPAN_EG_MIRROR_BUFFER_FACTOR 50
+
+static int mlxsw_sp2_span_policer_id_base_set(struct mlxsw_sp *mlxsw_sp,
+					      u16 policer_id_base)
+{
+	char mogcr_pl[MLXSW_REG_MOGCR_LEN];
+	int err;
+
+	err = mlxsw_reg_query(mlxsw_sp->core, MLXSW_REG(mogcr), mogcr_pl);
+	if (err)
+		return err;
+
+	mlxsw_reg_mogcr_mirroring_pid_base_set(mogcr_pl, policer_id_base);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(mogcr), mogcr_pl);
+}
+
+const struct mlxsw_sp_span_ops mlxsw_sp2_span_ops = {
+	.init = mlxsw_sp2_span_init,
+	.policer_id_base_set = mlxsw_sp2_span_policer_id_base_set,
+};
+
+const struct mlxsw_sp_span_ops mlxsw_sp3_span_ops = {
+	.init = mlxsw_sp2_span_init,
+	.policer_id_base_set = mlxsw_sp2_span_policer_id_base_set,
+};
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_span.h b/drivers/net/ethernet/mellanox/mlxsw/spectrum_span.h
new file mode 100644
index 000000000..82e711afb
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_span.h
@@ -0,0 +1,129 @@
+/* SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0 */
+/* Copyright (c) 2018 Mellanox Technologies. All rights reserved */
+
+#ifndef _MLXSW_SPECTRUM_SPAN_H
+#define _MLXSW_SPECTRUM_SPAN_H
+
+#include <linux/types.h>
+#include <linux/if_ether.h>
+#include <linux/refcount.h>
+
+#include "spectrum_router.h"
+
+struct mlxsw_sp;
+struct mlxsw_sp_port;
+
+/* SPAN session identifiers that correspond to MLXSW_TRAP_ID_MIRROR_SESSION<i>
+ * trap identifiers. The session identifier is an attribute of the SPAN agent,
+ * which determines the trap identifier of packets that are mirrored to the
+ * CPU. Packets that are trapped to the CPU for the same logical reason (e.g.,
+ * buffer drops) should use the same session identifier.
+ */
+enum mlxsw_sp_span_session_id {
+	MLXSW_SP_SPAN_SESSION_ID_BUFFER,
+	MLXSW_SP_SPAN_SESSION_ID_SAMPLING,
+
+	__MLXSW_SP_SPAN_SESSION_ID_MAX = 8,
+};
+
+struct mlxsw_sp_span_parms {
+	struct mlxsw_sp_port *dest_port; /* NULL for unoffloaded SPAN. */
+	unsigned int ttl;
+	unsigned char dmac[ETH_ALEN];
+	unsigned char smac[ETH_ALEN];
+	union mlxsw_sp_l3addr daddr;
+	union mlxsw_sp_l3addr saddr;
+	u16 vid;
+	u16 policer_id;
+	bool policer_enable;
+	enum mlxsw_sp_span_session_id session_id;
+};
+
+enum mlxsw_sp_span_trigger {
+	MLXSW_SP_SPAN_TRIGGER_INGRESS,
+	MLXSW_SP_SPAN_TRIGGER_EGRESS,
+	MLXSW_SP_SPAN_TRIGGER_TAIL_DROP,
+	MLXSW_SP_SPAN_TRIGGER_EARLY_DROP,
+	MLXSW_SP_SPAN_TRIGGER_ECN,
+};
+
+struct mlxsw_sp_span_trigger_parms {
+	int span_id;
+	u32 probability_rate;
+};
+
+struct mlxsw_sp_span_agent_parms {
+	const struct net_device *to_dev;
+	u16 policer_id;
+	bool policer_enable;
+	enum mlxsw_sp_span_session_id session_id;
+};
+
+struct mlxsw_sp_span_entry_ops;
+
+struct mlxsw_sp_span_ops {
+	int (*init)(struct mlxsw_sp *mlxsw_sp);
+	int (*policer_id_base_set)(struct mlxsw_sp *mlxsw_sp,
+				   u16 policer_id_base);
+};
+
+struct mlxsw_sp_span_entry {
+	const struct net_device *to_dev;
+	const struct mlxsw_sp_span_entry_ops *ops;
+	struct mlxsw_sp_span_parms parms;
+	refcount_t ref_count;
+	int id;
+};
+
+struct mlxsw_sp_span_entry_ops {
+	bool is_static;
+	bool (*can_handle)(const struct net_device *to_dev);
+	int (*parms_set)(struct mlxsw_sp *mlxsw_sp,
+			 const struct net_device *to_dev,
+			 struct mlxsw_sp_span_parms *sparmsp);
+	int (*configure)(struct mlxsw_sp_span_entry *span_entry,
+			 struct mlxsw_sp_span_parms sparms);
+	void (*deconfigure)(struct mlxsw_sp_span_entry *span_entry);
+};
+
+int mlxsw_sp_span_init(struct mlxsw_sp *mlxsw_sp);
+void mlxsw_sp_span_fini(struct mlxsw_sp *mlxsw_sp);
+void mlxsw_sp_span_respin(struct mlxsw_sp *mlxsw_sp);
+
+struct mlxsw_sp_span_entry *
+mlxsw_sp_span_entry_find_by_port(struct mlxsw_sp *mlxsw_sp,
+				 const struct net_device *to_dev);
+
+void mlxsw_sp_span_entry_invalidate(struct mlxsw_sp *mlxsw_sp,
+				    struct mlxsw_sp_span_entry *span_entry);
+
+int mlxsw_sp_span_port_mtu_update(struct mlxsw_sp_port *port, u16 mtu);
+void mlxsw_sp_span_speed_update_work(struct work_struct *work);
+
+int mlxsw_sp_span_agent_get(struct mlxsw_sp *mlxsw_sp, int *p_span_id,
+			    const struct mlxsw_sp_span_agent_parms *parms);
+void mlxsw_sp_span_agent_put(struct mlxsw_sp *mlxsw_sp, int span_id);
+int mlxsw_sp_span_analyzed_port_get(struct mlxsw_sp_port *mlxsw_sp_port,
+				    bool ingress);
+void mlxsw_sp_span_analyzed_port_put(struct mlxsw_sp_port *mlxsw_sp_port,
+				     bool ingress);
+int mlxsw_sp_span_agent_bind(struct mlxsw_sp *mlxsw_sp,
+			     enum mlxsw_sp_span_trigger trigger,
+			     struct mlxsw_sp_port *mlxsw_sp_port,
+			     const struct mlxsw_sp_span_trigger_parms *parms);
+void
+mlxsw_sp_span_agent_unbind(struct mlxsw_sp *mlxsw_sp,
+			   enum mlxsw_sp_span_trigger trigger,
+			   struct mlxsw_sp_port *mlxsw_sp_port,
+			   const struct mlxsw_sp_span_trigger_parms *parms);
+int mlxsw_sp_span_trigger_enable(struct mlxsw_sp_port *mlxsw_sp_port,
+				 enum mlxsw_sp_span_trigger trigger, u8 tc);
+void mlxsw_sp_span_trigger_disable(struct mlxsw_sp_port *mlxsw_sp_port,
+				   enum mlxsw_sp_span_trigger trigger, u8 tc);
+bool mlxsw_sp_span_trigger_is_ingress(enum mlxsw_sp_span_trigger trigger);
+
+extern const struct mlxsw_sp_span_ops mlxsw_sp1_span_ops;
+extern const struct mlxsw_sp_span_ops mlxsw_sp2_span_ops;
+extern const struct mlxsw_sp_span_ops mlxsw_sp3_span_ops;
+
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_switchdev.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_switchdev.c
new file mode 100644
index 000000000..1290b2d3e
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_switchdev.c
@@ -0,0 +1,4015 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2015-2018 Mellanox Technologies. All rights reserved */
+
+#include <linux/kernel.h>
+#include <linux/types.h>
+#include <linux/netdevice.h>
+#include <linux/etherdevice.h>
+#include <linux/slab.h>
+#include <linux/device.h>
+#include <linux/skbuff.h>
+#include <linux/if_vlan.h>
+#include <linux/if_bridge.h>
+#include <linux/workqueue.h>
+#include <linux/jiffies.h>
+#include <linux/rtnetlink.h>
+#include <linux/netlink.h>
+#include <net/switchdev.h>
+#include <net/vxlan.h>
+
+#include "spectrum_span.h"
+#include "spectrum_switchdev.h"
+#include "spectrum.h"
+#include "core.h"
+#include "reg.h"
+
+struct mlxsw_sp_bridge_ops;
+
+struct mlxsw_sp_bridge {
+	struct mlxsw_sp *mlxsw_sp;
+	struct {
+		struct delayed_work dw;
+#define MLXSW_SP_DEFAULT_LEARNING_INTERVAL 100
+		unsigned int interval; /* ms */
+	} fdb_notify;
+#define MLXSW_SP_MIN_AGEING_TIME 10
+#define MLXSW_SP_MAX_AGEING_TIME 1000000
+#define MLXSW_SP_DEFAULT_AGEING_TIME 300
+	u32 ageing_time;
+	bool vlan_enabled_exists;
+	struct list_head bridges_list;
+	DECLARE_BITMAP(mids_bitmap, MLXSW_SP_MID_MAX);
+	const struct mlxsw_sp_bridge_ops *bridge_8021q_ops;
+	const struct mlxsw_sp_bridge_ops *bridge_8021d_ops;
+	const struct mlxsw_sp_bridge_ops *bridge_8021ad_ops;
+};
+
+struct mlxsw_sp_bridge_device {
+	struct net_device *dev;
+	struct list_head list;
+	struct list_head ports_list;
+	struct list_head mdb_list;
+	struct rhashtable mdb_ht;
+	u8 vlan_enabled:1,
+	   multicast_enabled:1,
+	   mrouter:1;
+	const struct mlxsw_sp_bridge_ops *ops;
+};
+
+struct mlxsw_sp_bridge_port {
+	struct net_device *dev;
+	struct mlxsw_sp_bridge_device *bridge_device;
+	struct list_head list;
+	struct list_head vlans_list;
+	unsigned int ref_count;
+	u8 stp_state;
+	unsigned long flags;
+	bool mrouter;
+	bool lagged;
+	union {
+		u16 lag_id;
+		u16 system_port;
+	};
+};
+
+struct mlxsw_sp_bridge_vlan {
+	struct list_head list;
+	struct list_head port_vlan_list;
+	u16 vid;
+};
+
+struct mlxsw_sp_bridge_ops {
+	int (*port_join)(struct mlxsw_sp_bridge_device *bridge_device,
+			 struct mlxsw_sp_bridge_port *bridge_port,
+			 struct mlxsw_sp_port *mlxsw_sp_port,
+			 struct netlink_ext_ack *extack);
+	void (*port_leave)(struct mlxsw_sp_bridge_device *bridge_device,
+			   struct mlxsw_sp_bridge_port *bridge_port,
+			   struct mlxsw_sp_port *mlxsw_sp_port);
+	int (*vxlan_join)(struct mlxsw_sp_bridge_device *bridge_device,
+			  const struct net_device *vxlan_dev, u16 vid,
+			  struct netlink_ext_ack *extack);
+	struct mlxsw_sp_fid *
+		(*fid_get)(struct mlxsw_sp_bridge_device *bridge_device,
+			   u16 vid, struct netlink_ext_ack *extack);
+	struct mlxsw_sp_fid *
+		(*fid_lookup)(struct mlxsw_sp_bridge_device *bridge_device,
+			      u16 vid);
+	u16 (*fid_vid)(struct mlxsw_sp_bridge_device *bridge_device,
+		       const struct mlxsw_sp_fid *fid);
+};
+
+struct mlxsw_sp_switchdev_ops {
+	void (*init)(struct mlxsw_sp *mlxsw_sp);
+};
+
+struct mlxsw_sp_mdb_entry_key {
+	unsigned char addr[ETH_ALEN];
+	u16 fid;
+};
+
+struct mlxsw_sp_mdb_entry {
+	struct list_head list;
+	struct rhash_head ht_node;
+	struct mlxsw_sp_mdb_entry_key key;
+	u16 mid;
+	struct list_head ports_list;
+	u16 ports_count;
+};
+
+struct mlxsw_sp_mdb_entry_port {
+	struct list_head list; /* Member of 'ports_list'. */
+	u16 local_port;
+	refcount_t refcount;
+	bool mrouter;
+};
+
+static const struct rhashtable_params mlxsw_sp_mdb_ht_params = {
+	.key_offset = offsetof(struct mlxsw_sp_mdb_entry, key),
+	.head_offset = offsetof(struct mlxsw_sp_mdb_entry, ht_node),
+	.key_len = sizeof(struct mlxsw_sp_mdb_entry_key),
+};
+
+static int
+mlxsw_sp_bridge_port_fdb_flush(struct mlxsw_sp *mlxsw_sp,
+			       struct mlxsw_sp_bridge_port *bridge_port,
+			       u16 fid_index);
+
+static void
+mlxsw_sp_bridge_port_mdb_flush(struct mlxsw_sp_port *mlxsw_sp_port,
+			       struct mlxsw_sp_bridge_port *bridge_port,
+			       u16 fid_index);
+
+static int
+mlxsw_sp_bridge_mdb_mc_enable_sync(struct mlxsw_sp *mlxsw_sp,
+				   struct mlxsw_sp_bridge_device
+				   *bridge_device, bool mc_enabled);
+
+static void
+mlxsw_sp_port_mrouter_update_mdb(struct mlxsw_sp_port *mlxsw_sp_port,
+				 struct mlxsw_sp_bridge_port *bridge_port,
+				 bool add);
+
+static struct mlxsw_sp_bridge_device *
+mlxsw_sp_bridge_device_find(const struct mlxsw_sp_bridge *bridge,
+			    const struct net_device *br_dev)
+{
+	struct mlxsw_sp_bridge_device *bridge_device;
+
+	list_for_each_entry(bridge_device, &bridge->bridges_list, list)
+		if (bridge_device->dev == br_dev)
+			return bridge_device;
+
+	return NULL;
+}
+
+bool mlxsw_sp_bridge_device_is_offloaded(const struct mlxsw_sp *mlxsw_sp,
+					 const struct net_device *br_dev)
+{
+	return !!mlxsw_sp_bridge_device_find(mlxsw_sp->bridge, br_dev);
+}
+
+static int mlxsw_sp_bridge_device_upper_rif_destroy(struct net_device *dev,
+						    struct netdev_nested_priv *priv)
+{
+	struct mlxsw_sp *mlxsw_sp = priv->data;
+
+	mlxsw_sp_rif_destroy_by_dev(mlxsw_sp, dev);
+	return 0;
+}
+
+static void mlxsw_sp_bridge_device_rifs_destroy(struct mlxsw_sp *mlxsw_sp,
+						struct net_device *dev)
+{
+	struct netdev_nested_priv priv = {
+		.data = (void *)mlxsw_sp,
+	};
+
+	mlxsw_sp_rif_destroy_by_dev(mlxsw_sp, dev);
+	netdev_walk_all_upper_dev_rcu(dev,
+				      mlxsw_sp_bridge_device_upper_rif_destroy,
+				      &priv);
+}
+
+static int mlxsw_sp_bridge_device_vxlan_init(struct mlxsw_sp_bridge *bridge,
+					     struct net_device *br_dev,
+					     struct netlink_ext_ack *extack)
+{
+	struct net_device *dev, *stop_dev;
+	struct list_head *iter;
+	int err;
+
+	netdev_for_each_lower_dev(br_dev, dev, iter) {
+		if (netif_is_vxlan(dev) && netif_running(dev)) {
+			err = mlxsw_sp_bridge_vxlan_join(bridge->mlxsw_sp,
+							 br_dev, dev, 0,
+							 extack);
+			if (err) {
+				stop_dev = dev;
+				goto err_vxlan_join;
+			}
+		}
+	}
+
+	return 0;
+
+err_vxlan_join:
+	netdev_for_each_lower_dev(br_dev, dev, iter) {
+		if (netif_is_vxlan(dev) && netif_running(dev)) {
+			if (stop_dev == dev)
+				break;
+			mlxsw_sp_bridge_vxlan_leave(bridge->mlxsw_sp, dev);
+		}
+	}
+	return err;
+}
+
+static void mlxsw_sp_bridge_device_vxlan_fini(struct mlxsw_sp_bridge *bridge,
+					      struct net_device *br_dev)
+{
+	struct net_device *dev;
+	struct list_head *iter;
+
+	netdev_for_each_lower_dev(br_dev, dev, iter) {
+		if (netif_is_vxlan(dev) && netif_running(dev))
+			mlxsw_sp_bridge_vxlan_leave(bridge->mlxsw_sp, dev);
+	}
+}
+
+static void mlxsw_sp_fdb_notify_work_schedule(struct mlxsw_sp *mlxsw_sp,
+					      bool no_delay)
+{
+	struct mlxsw_sp_bridge *bridge = mlxsw_sp->bridge;
+	unsigned int interval = no_delay ? 0 : bridge->fdb_notify.interval;
+
+	mlxsw_core_schedule_dw(&bridge->fdb_notify.dw,
+			       msecs_to_jiffies(interval));
+}
+
+static struct mlxsw_sp_bridge_device *
+mlxsw_sp_bridge_device_create(struct mlxsw_sp_bridge *bridge,
+			      struct net_device *br_dev,
+			      struct netlink_ext_ack *extack)
+{
+	struct device *dev = bridge->mlxsw_sp->bus_info->dev;
+	struct mlxsw_sp_bridge_device *bridge_device;
+	bool vlan_enabled = br_vlan_enabled(br_dev);
+	int err;
+
+	if (vlan_enabled && bridge->vlan_enabled_exists) {
+		dev_err(dev, "Only one VLAN-aware bridge is supported\n");
+		NL_SET_ERR_MSG_MOD(extack, "Only one VLAN-aware bridge is supported");
+		return ERR_PTR(-EINVAL);
+	}
+
+	bridge_device = kzalloc(sizeof(*bridge_device), GFP_KERNEL);
+	if (!bridge_device)
+		return ERR_PTR(-ENOMEM);
+
+	err = rhashtable_init(&bridge_device->mdb_ht, &mlxsw_sp_mdb_ht_params);
+	if (err)
+		goto err_mdb_rhashtable_init;
+
+	bridge_device->dev = br_dev;
+	bridge_device->vlan_enabled = vlan_enabled;
+	bridge_device->multicast_enabled = br_multicast_enabled(br_dev);
+	bridge_device->mrouter = br_multicast_router(br_dev);
+	INIT_LIST_HEAD(&bridge_device->ports_list);
+	if (vlan_enabled) {
+		u16 proto;
+
+		bridge->vlan_enabled_exists = true;
+		br_vlan_get_proto(br_dev, &proto);
+		if (proto == ETH_P_8021AD)
+			bridge_device->ops = bridge->bridge_8021ad_ops;
+		else
+			bridge_device->ops = bridge->bridge_8021q_ops;
+	} else {
+		bridge_device->ops = bridge->bridge_8021d_ops;
+	}
+	INIT_LIST_HEAD(&bridge_device->mdb_list);
+
+	if (list_empty(&bridge->bridges_list))
+		mlxsw_sp_fdb_notify_work_schedule(bridge->mlxsw_sp, false);
+	list_add(&bridge_device->list, &bridge->bridges_list);
+
+	/* It is possible we already have VXLAN devices enslaved to the bridge.
+	 * In which case, we need to replay their configuration as if they were
+	 * just now enslaved to the bridge.
+	 */
+	err = mlxsw_sp_bridge_device_vxlan_init(bridge, br_dev, extack);
+	if (err)
+		goto err_vxlan_init;
+
+	return bridge_device;
+
+err_vxlan_init:
+	list_del(&bridge_device->list);
+	if (bridge_device->vlan_enabled)
+		bridge->vlan_enabled_exists = false;
+	rhashtable_destroy(&bridge_device->mdb_ht);
+err_mdb_rhashtable_init:
+	kfree(bridge_device);
+	return ERR_PTR(err);
+}
+
+static void
+mlxsw_sp_bridge_device_destroy(struct mlxsw_sp_bridge *bridge,
+			       struct mlxsw_sp_bridge_device *bridge_device)
+{
+	mlxsw_sp_bridge_device_vxlan_fini(bridge, bridge_device->dev);
+	mlxsw_sp_bridge_device_rifs_destroy(bridge->mlxsw_sp,
+					    bridge_device->dev);
+	list_del(&bridge_device->list);
+	if (list_empty(&bridge->bridges_list))
+		cancel_delayed_work(&bridge->fdb_notify.dw);
+	if (bridge_device->vlan_enabled)
+		bridge->vlan_enabled_exists = false;
+	WARN_ON(!list_empty(&bridge_device->ports_list));
+	WARN_ON(!list_empty(&bridge_device->mdb_list));
+	rhashtable_destroy(&bridge_device->mdb_ht);
+	kfree(bridge_device);
+}
+
+static struct mlxsw_sp_bridge_device *
+mlxsw_sp_bridge_device_get(struct mlxsw_sp_bridge *bridge,
+			   struct net_device *br_dev,
+			   struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp_bridge_device *bridge_device;
+
+	bridge_device = mlxsw_sp_bridge_device_find(bridge, br_dev);
+	if (bridge_device)
+		return bridge_device;
+
+	return mlxsw_sp_bridge_device_create(bridge, br_dev, extack);
+}
+
+static void
+mlxsw_sp_bridge_device_put(struct mlxsw_sp_bridge *bridge,
+			   struct mlxsw_sp_bridge_device *bridge_device)
+{
+	if (list_empty(&bridge_device->ports_list))
+		mlxsw_sp_bridge_device_destroy(bridge, bridge_device);
+}
+
+static struct mlxsw_sp_bridge_port *
+__mlxsw_sp_bridge_port_find(const struct mlxsw_sp_bridge_device *bridge_device,
+			    const struct net_device *brport_dev)
+{
+	struct mlxsw_sp_bridge_port *bridge_port;
+
+	list_for_each_entry(bridge_port, &bridge_device->ports_list, list) {
+		if (bridge_port->dev == brport_dev)
+			return bridge_port;
+	}
+
+	return NULL;
+}
+
+struct mlxsw_sp_bridge_port *
+mlxsw_sp_bridge_port_find(struct mlxsw_sp_bridge *bridge,
+			  struct net_device *brport_dev)
+{
+	struct net_device *br_dev = netdev_master_upper_dev_get(brport_dev);
+	struct mlxsw_sp_bridge_device *bridge_device;
+
+	if (!br_dev)
+		return NULL;
+
+	bridge_device = mlxsw_sp_bridge_device_find(bridge, br_dev);
+	if (!bridge_device)
+		return NULL;
+
+	return __mlxsw_sp_bridge_port_find(bridge_device, brport_dev);
+}
+
+static struct mlxsw_sp_bridge_port *
+mlxsw_sp_bridge_port_create(struct mlxsw_sp_bridge_device *bridge_device,
+			    struct net_device *brport_dev,
+			    struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp_bridge_port *bridge_port;
+	struct mlxsw_sp_port *mlxsw_sp_port;
+	int err;
+
+	bridge_port = kzalloc(sizeof(*bridge_port), GFP_KERNEL);
+	if (!bridge_port)
+		return ERR_PTR(-ENOMEM);
+
+	mlxsw_sp_port = mlxsw_sp_port_dev_lower_find(brport_dev);
+	bridge_port->lagged = mlxsw_sp_port->lagged;
+	if (bridge_port->lagged)
+		bridge_port->lag_id = mlxsw_sp_port->lag_id;
+	else
+		bridge_port->system_port = mlxsw_sp_port->local_port;
+	bridge_port->dev = brport_dev;
+	bridge_port->bridge_device = bridge_device;
+	bridge_port->stp_state = BR_STATE_DISABLED;
+	bridge_port->flags = BR_LEARNING | BR_FLOOD | BR_LEARNING_SYNC |
+			     BR_MCAST_FLOOD;
+	INIT_LIST_HEAD(&bridge_port->vlans_list);
+	list_add(&bridge_port->list, &bridge_device->ports_list);
+	bridge_port->ref_count = 1;
+
+	err = switchdev_bridge_port_offload(brport_dev, mlxsw_sp_port->dev,
+					    NULL, NULL, NULL, false, extack);
+	if (err)
+		goto err_switchdev_offload;
+
+	return bridge_port;
+
+err_switchdev_offload:
+	list_del(&bridge_port->list);
+	kfree(bridge_port);
+	return ERR_PTR(err);
+}
+
+static void
+mlxsw_sp_bridge_port_destroy(struct mlxsw_sp_bridge_port *bridge_port)
+{
+	switchdev_bridge_port_unoffload(bridge_port->dev, NULL, NULL, NULL);
+	list_del(&bridge_port->list);
+	WARN_ON(!list_empty(&bridge_port->vlans_list));
+	kfree(bridge_port);
+}
+
+static struct mlxsw_sp_bridge_port *
+mlxsw_sp_bridge_port_get(struct mlxsw_sp_bridge *bridge,
+			 struct net_device *brport_dev,
+			 struct netlink_ext_ack *extack)
+{
+	struct net_device *br_dev = netdev_master_upper_dev_get(brport_dev);
+	struct mlxsw_sp_bridge_device *bridge_device;
+	struct mlxsw_sp_bridge_port *bridge_port;
+	int err;
+
+	bridge_port = mlxsw_sp_bridge_port_find(bridge, brport_dev);
+	if (bridge_port) {
+		bridge_port->ref_count++;
+		return bridge_port;
+	}
+
+	bridge_device = mlxsw_sp_bridge_device_get(bridge, br_dev, extack);
+	if (IS_ERR(bridge_device))
+		return ERR_CAST(bridge_device);
+
+	bridge_port = mlxsw_sp_bridge_port_create(bridge_device, brport_dev,
+						  extack);
+	if (IS_ERR(bridge_port)) {
+		err = PTR_ERR(bridge_port);
+		goto err_bridge_port_create;
+	}
+
+	return bridge_port;
+
+err_bridge_port_create:
+	mlxsw_sp_bridge_device_put(bridge, bridge_device);
+	return ERR_PTR(err);
+}
+
+static void mlxsw_sp_bridge_port_put(struct mlxsw_sp_bridge *bridge,
+				     struct mlxsw_sp_bridge_port *bridge_port)
+{
+	struct mlxsw_sp_bridge_device *bridge_device;
+
+	if (--bridge_port->ref_count != 0)
+		return;
+	bridge_device = bridge_port->bridge_device;
+	mlxsw_sp_bridge_port_destroy(bridge_port);
+	mlxsw_sp_bridge_device_put(bridge, bridge_device);
+}
+
+static struct mlxsw_sp_port_vlan *
+mlxsw_sp_port_vlan_find_by_bridge(struct mlxsw_sp_port *mlxsw_sp_port,
+				  const struct mlxsw_sp_bridge_device *
+				  bridge_device,
+				  u16 vid)
+{
+	struct mlxsw_sp_port_vlan *mlxsw_sp_port_vlan;
+
+	list_for_each_entry(mlxsw_sp_port_vlan, &mlxsw_sp_port->vlans_list,
+			    list) {
+		if (!mlxsw_sp_port_vlan->bridge_port)
+			continue;
+		if (mlxsw_sp_port_vlan->bridge_port->bridge_device !=
+		    bridge_device)
+			continue;
+		if (bridge_device->vlan_enabled &&
+		    mlxsw_sp_port_vlan->vid != vid)
+			continue;
+		return mlxsw_sp_port_vlan;
+	}
+
+	return NULL;
+}
+
+static struct mlxsw_sp_port_vlan*
+mlxsw_sp_port_vlan_find_by_fid(struct mlxsw_sp_port *mlxsw_sp_port,
+			       u16 fid_index)
+{
+	struct mlxsw_sp_port_vlan *mlxsw_sp_port_vlan;
+
+	list_for_each_entry(mlxsw_sp_port_vlan, &mlxsw_sp_port->vlans_list,
+			    list) {
+		struct mlxsw_sp_fid *fid = mlxsw_sp_port_vlan->fid;
+
+		if (fid && mlxsw_sp_fid_index(fid) == fid_index)
+			return mlxsw_sp_port_vlan;
+	}
+
+	return NULL;
+}
+
+static struct mlxsw_sp_bridge_vlan *
+mlxsw_sp_bridge_vlan_find(const struct mlxsw_sp_bridge_port *bridge_port,
+			  u16 vid)
+{
+	struct mlxsw_sp_bridge_vlan *bridge_vlan;
+
+	list_for_each_entry(bridge_vlan, &bridge_port->vlans_list, list) {
+		if (bridge_vlan->vid == vid)
+			return bridge_vlan;
+	}
+
+	return NULL;
+}
+
+static struct mlxsw_sp_bridge_vlan *
+mlxsw_sp_bridge_vlan_create(struct mlxsw_sp_bridge_port *bridge_port, u16 vid)
+{
+	struct mlxsw_sp_bridge_vlan *bridge_vlan;
+
+	bridge_vlan = kzalloc(sizeof(*bridge_vlan), GFP_KERNEL);
+	if (!bridge_vlan)
+		return NULL;
+
+	INIT_LIST_HEAD(&bridge_vlan->port_vlan_list);
+	bridge_vlan->vid = vid;
+	list_add(&bridge_vlan->list, &bridge_port->vlans_list);
+
+	return bridge_vlan;
+}
+
+static void
+mlxsw_sp_bridge_vlan_destroy(struct mlxsw_sp_bridge_vlan *bridge_vlan)
+{
+	list_del(&bridge_vlan->list);
+	WARN_ON(!list_empty(&bridge_vlan->port_vlan_list));
+	kfree(bridge_vlan);
+}
+
+static struct mlxsw_sp_bridge_vlan *
+mlxsw_sp_bridge_vlan_get(struct mlxsw_sp_bridge_port *bridge_port, u16 vid)
+{
+	struct mlxsw_sp_bridge_vlan *bridge_vlan;
+
+	bridge_vlan = mlxsw_sp_bridge_vlan_find(bridge_port, vid);
+	if (bridge_vlan)
+		return bridge_vlan;
+
+	return mlxsw_sp_bridge_vlan_create(bridge_port, vid);
+}
+
+static void mlxsw_sp_bridge_vlan_put(struct mlxsw_sp_bridge_vlan *bridge_vlan)
+{
+	if (list_empty(&bridge_vlan->port_vlan_list))
+		mlxsw_sp_bridge_vlan_destroy(bridge_vlan);
+}
+
+static int
+mlxsw_sp_port_bridge_vlan_stp_set(struct mlxsw_sp_port *mlxsw_sp_port,
+				  struct mlxsw_sp_bridge_vlan *bridge_vlan,
+				  u8 state)
+{
+	struct mlxsw_sp_port_vlan *mlxsw_sp_port_vlan;
+
+	list_for_each_entry(mlxsw_sp_port_vlan, &bridge_vlan->port_vlan_list,
+			    bridge_vlan_node) {
+		if (mlxsw_sp_port_vlan->mlxsw_sp_port != mlxsw_sp_port)
+			continue;
+		return mlxsw_sp_port_vid_stp_set(mlxsw_sp_port,
+						 bridge_vlan->vid, state);
+	}
+
+	return 0;
+}
+
+static int mlxsw_sp_port_attr_stp_state_set(struct mlxsw_sp_port *mlxsw_sp_port,
+					    struct net_device *orig_dev,
+					    u8 state)
+{
+	struct mlxsw_sp_bridge_port *bridge_port;
+	struct mlxsw_sp_bridge_vlan *bridge_vlan;
+	int err;
+
+	/* It's possible we failed to enslave the port, yet this
+	 * operation is executed due to it being deferred.
+	 */
+	bridge_port = mlxsw_sp_bridge_port_find(mlxsw_sp_port->mlxsw_sp->bridge,
+						orig_dev);
+	if (!bridge_port)
+		return 0;
+
+	list_for_each_entry(bridge_vlan, &bridge_port->vlans_list, list) {
+		err = mlxsw_sp_port_bridge_vlan_stp_set(mlxsw_sp_port,
+							bridge_vlan, state);
+		if (err)
+			goto err_port_bridge_vlan_stp_set;
+	}
+
+	bridge_port->stp_state = state;
+
+	return 0;
+
+err_port_bridge_vlan_stp_set:
+	list_for_each_entry_continue_reverse(bridge_vlan,
+					     &bridge_port->vlans_list, list)
+		mlxsw_sp_port_bridge_vlan_stp_set(mlxsw_sp_port, bridge_vlan,
+						  bridge_port->stp_state);
+	return err;
+}
+
+static int
+mlxsw_sp_port_bridge_vlan_flood_set(struct mlxsw_sp_port *mlxsw_sp_port,
+				    struct mlxsw_sp_bridge_vlan *bridge_vlan,
+				    enum mlxsw_sp_flood_type packet_type,
+				    bool member)
+{
+	struct mlxsw_sp_port_vlan *mlxsw_sp_port_vlan;
+
+	list_for_each_entry(mlxsw_sp_port_vlan, &bridge_vlan->port_vlan_list,
+			    bridge_vlan_node) {
+		if (mlxsw_sp_port_vlan->mlxsw_sp_port != mlxsw_sp_port)
+			continue;
+		return mlxsw_sp_fid_flood_set(mlxsw_sp_port_vlan->fid,
+					      packet_type,
+					      mlxsw_sp_port->local_port,
+					      member);
+	}
+
+	return 0;
+}
+
+static int
+mlxsw_sp_bridge_port_flood_table_set(struct mlxsw_sp_port *mlxsw_sp_port,
+				     struct mlxsw_sp_bridge_port *bridge_port,
+				     enum mlxsw_sp_flood_type packet_type,
+				     bool member)
+{
+	struct mlxsw_sp_bridge_vlan *bridge_vlan;
+	int err;
+
+	list_for_each_entry(bridge_vlan, &bridge_port->vlans_list, list) {
+		err = mlxsw_sp_port_bridge_vlan_flood_set(mlxsw_sp_port,
+							  bridge_vlan,
+							  packet_type,
+							  member);
+		if (err)
+			goto err_port_bridge_vlan_flood_set;
+	}
+
+	return 0;
+
+err_port_bridge_vlan_flood_set:
+	list_for_each_entry_continue_reverse(bridge_vlan,
+					     &bridge_port->vlans_list, list)
+		mlxsw_sp_port_bridge_vlan_flood_set(mlxsw_sp_port, bridge_vlan,
+						    packet_type, !member);
+	return err;
+}
+
+static int
+mlxsw_sp_bridge_vlans_flood_set(struct mlxsw_sp_bridge_vlan *bridge_vlan,
+				enum mlxsw_sp_flood_type packet_type,
+				bool member)
+{
+	struct mlxsw_sp_port_vlan *mlxsw_sp_port_vlan;
+	int err;
+
+	list_for_each_entry(mlxsw_sp_port_vlan, &bridge_vlan->port_vlan_list,
+			    bridge_vlan_node) {
+		u16 local_port = mlxsw_sp_port_vlan->mlxsw_sp_port->local_port;
+
+		err = mlxsw_sp_fid_flood_set(mlxsw_sp_port_vlan->fid,
+					     packet_type, local_port, member);
+		if (err)
+			goto err_fid_flood_set;
+	}
+
+	return 0;
+
+err_fid_flood_set:
+	list_for_each_entry_continue_reverse(mlxsw_sp_port_vlan,
+					     &bridge_vlan->port_vlan_list,
+					     list) {
+		u16 local_port = mlxsw_sp_port_vlan->mlxsw_sp_port->local_port;
+
+		mlxsw_sp_fid_flood_set(mlxsw_sp_port_vlan->fid, packet_type,
+				       local_port, !member);
+	}
+
+	return err;
+}
+
+static int
+mlxsw_sp_bridge_ports_flood_table_set(struct mlxsw_sp_bridge_port *bridge_port,
+				      enum mlxsw_sp_flood_type packet_type,
+				      bool member)
+{
+	struct mlxsw_sp_bridge_vlan *bridge_vlan;
+	int err;
+
+	list_for_each_entry(bridge_vlan, &bridge_port->vlans_list, list) {
+		err = mlxsw_sp_bridge_vlans_flood_set(bridge_vlan, packet_type,
+						      member);
+		if (err)
+			goto err_bridge_vlans_flood_set;
+	}
+
+	return 0;
+
+err_bridge_vlans_flood_set:
+	list_for_each_entry_continue_reverse(bridge_vlan,
+					     &bridge_port->vlans_list, list)
+		mlxsw_sp_bridge_vlans_flood_set(bridge_vlan, packet_type,
+						!member);
+	return err;
+}
+
+static int
+mlxsw_sp_port_bridge_vlan_learning_set(struct mlxsw_sp_port *mlxsw_sp_port,
+				       struct mlxsw_sp_bridge_vlan *bridge_vlan,
+				       bool set)
+{
+	struct mlxsw_sp_port_vlan *mlxsw_sp_port_vlan;
+	u16 vid = bridge_vlan->vid;
+
+	list_for_each_entry(mlxsw_sp_port_vlan, &bridge_vlan->port_vlan_list,
+			    bridge_vlan_node) {
+		if (mlxsw_sp_port_vlan->mlxsw_sp_port != mlxsw_sp_port)
+			continue;
+		return mlxsw_sp_port_vid_learning_set(mlxsw_sp_port, vid, set);
+	}
+
+	return 0;
+}
+
+static int
+mlxsw_sp_bridge_port_learning_set(struct mlxsw_sp_port *mlxsw_sp_port,
+				  struct mlxsw_sp_bridge_port *bridge_port,
+				  bool set)
+{
+	struct mlxsw_sp_bridge_vlan *bridge_vlan;
+	int err;
+
+	list_for_each_entry(bridge_vlan, &bridge_port->vlans_list, list) {
+		err = mlxsw_sp_port_bridge_vlan_learning_set(mlxsw_sp_port,
+							     bridge_vlan, set);
+		if (err)
+			goto err_port_bridge_vlan_learning_set;
+	}
+
+	return 0;
+
+err_port_bridge_vlan_learning_set:
+	list_for_each_entry_continue_reverse(bridge_vlan,
+					     &bridge_port->vlans_list, list)
+		mlxsw_sp_port_bridge_vlan_learning_set(mlxsw_sp_port,
+						       bridge_vlan, !set);
+	return err;
+}
+
+static int
+mlxsw_sp_port_attr_br_pre_flags_set(struct mlxsw_sp_port *mlxsw_sp_port,
+				    struct switchdev_brport_flags flags)
+{
+	if (flags.mask & ~(BR_LEARNING | BR_FLOOD | BR_MCAST_FLOOD))
+		return -EINVAL;
+
+	return 0;
+}
+
+static int mlxsw_sp_port_attr_br_flags_set(struct mlxsw_sp_port *mlxsw_sp_port,
+					   struct net_device *orig_dev,
+					   struct switchdev_brport_flags flags)
+{
+	struct mlxsw_sp_bridge_port *bridge_port;
+	int err;
+
+	bridge_port = mlxsw_sp_bridge_port_find(mlxsw_sp_port->mlxsw_sp->bridge,
+						orig_dev);
+	if (!bridge_port)
+		return 0;
+
+	if (flags.mask & BR_FLOOD) {
+		err = mlxsw_sp_bridge_port_flood_table_set(mlxsw_sp_port,
+							   bridge_port,
+							   MLXSW_SP_FLOOD_TYPE_UC,
+							   flags.val & BR_FLOOD);
+		if (err)
+			return err;
+	}
+
+	if (flags.mask & BR_LEARNING) {
+		err = mlxsw_sp_bridge_port_learning_set(mlxsw_sp_port,
+							bridge_port,
+							flags.val & BR_LEARNING);
+		if (err)
+			return err;
+	}
+
+	if (bridge_port->bridge_device->multicast_enabled)
+		goto out;
+
+	if (flags.mask & BR_MCAST_FLOOD) {
+		err = mlxsw_sp_bridge_port_flood_table_set(mlxsw_sp_port,
+							   bridge_port,
+							   MLXSW_SP_FLOOD_TYPE_MC,
+							   flags.val & BR_MCAST_FLOOD);
+		if (err)
+			return err;
+	}
+
+out:
+	memcpy(&bridge_port->flags, &flags.val, sizeof(flags.val));
+	return 0;
+}
+
+static int mlxsw_sp_ageing_set(struct mlxsw_sp *mlxsw_sp, u32 ageing_time)
+{
+	char sfdat_pl[MLXSW_REG_SFDAT_LEN];
+	int err;
+
+	mlxsw_reg_sfdat_pack(sfdat_pl, ageing_time);
+	err = mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(sfdat), sfdat_pl);
+	if (err)
+		return err;
+	mlxsw_sp->bridge->ageing_time = ageing_time;
+	return 0;
+}
+
+static int mlxsw_sp_port_attr_br_ageing_set(struct mlxsw_sp_port *mlxsw_sp_port,
+					    unsigned long ageing_clock_t)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	unsigned long ageing_jiffies = clock_t_to_jiffies(ageing_clock_t);
+	u32 ageing_time = jiffies_to_msecs(ageing_jiffies) / 1000;
+
+	if (ageing_time < MLXSW_SP_MIN_AGEING_TIME ||
+	    ageing_time > MLXSW_SP_MAX_AGEING_TIME)
+		return -ERANGE;
+
+	return mlxsw_sp_ageing_set(mlxsw_sp, ageing_time);
+}
+
+static int mlxsw_sp_port_attr_br_vlan_set(struct mlxsw_sp_port *mlxsw_sp_port,
+					  struct net_device *orig_dev,
+					  bool vlan_enabled)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	struct mlxsw_sp_bridge_device *bridge_device;
+
+	bridge_device = mlxsw_sp_bridge_device_find(mlxsw_sp->bridge, orig_dev);
+	if (WARN_ON(!bridge_device))
+		return -EINVAL;
+
+	if (bridge_device->vlan_enabled == vlan_enabled)
+		return 0;
+
+	netdev_err(bridge_device->dev, "VLAN filtering can't be changed for existing bridge\n");
+	return -EINVAL;
+}
+
+static int mlxsw_sp_port_attr_br_vlan_proto_set(struct mlxsw_sp_port *mlxsw_sp_port,
+						struct net_device *orig_dev,
+						u16 vlan_proto)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	struct mlxsw_sp_bridge_device *bridge_device;
+
+	bridge_device = mlxsw_sp_bridge_device_find(mlxsw_sp->bridge, orig_dev);
+	if (WARN_ON(!bridge_device))
+		return -EINVAL;
+
+	netdev_err(bridge_device->dev, "VLAN protocol can't be changed on existing bridge\n");
+	return -EINVAL;
+}
+
+static int mlxsw_sp_port_attr_mrouter_set(struct mlxsw_sp_port *mlxsw_sp_port,
+					  struct net_device *orig_dev,
+					  bool is_port_mrouter)
+{
+	struct mlxsw_sp_bridge_port *bridge_port;
+	int err;
+
+	bridge_port = mlxsw_sp_bridge_port_find(mlxsw_sp_port->mlxsw_sp->bridge,
+						orig_dev);
+	if (!bridge_port)
+		return 0;
+
+	mlxsw_sp_port_mrouter_update_mdb(mlxsw_sp_port, bridge_port,
+					 is_port_mrouter);
+
+	if (!bridge_port->bridge_device->multicast_enabled)
+		goto out;
+
+	err = mlxsw_sp_bridge_port_flood_table_set(mlxsw_sp_port, bridge_port,
+						   MLXSW_SP_FLOOD_TYPE_MC,
+						   is_port_mrouter);
+	if (err)
+		return err;
+
+out:
+	bridge_port->mrouter = is_port_mrouter;
+	return 0;
+}
+
+static bool mlxsw_sp_mc_flood(const struct mlxsw_sp_bridge_port *bridge_port)
+{
+	const struct mlxsw_sp_bridge_device *bridge_device;
+
+	bridge_device = bridge_port->bridge_device;
+	return bridge_device->multicast_enabled ? bridge_port->mrouter :
+					bridge_port->flags & BR_MCAST_FLOOD;
+}
+
+static int mlxsw_sp_port_mc_disabled_set(struct mlxsw_sp_port *mlxsw_sp_port,
+					 struct net_device *orig_dev,
+					 bool mc_disabled)
+{
+	enum mlxsw_sp_flood_type packet_type = MLXSW_SP_FLOOD_TYPE_MC;
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	struct mlxsw_sp_bridge_device *bridge_device;
+	struct mlxsw_sp_bridge_port *bridge_port;
+	int err;
+
+	/* It's possible we failed to enslave the port, yet this
+	 * operation is executed due to it being deferred.
+	 */
+	bridge_device = mlxsw_sp_bridge_device_find(mlxsw_sp->bridge, orig_dev);
+	if (!bridge_device)
+		return 0;
+
+	if (bridge_device->multicast_enabled == !mc_disabled)
+		return 0;
+
+	bridge_device->multicast_enabled = !mc_disabled;
+	err = mlxsw_sp_bridge_mdb_mc_enable_sync(mlxsw_sp, bridge_device,
+						 !mc_disabled);
+	if (err)
+		goto err_mc_enable_sync;
+
+	list_for_each_entry(bridge_port, &bridge_device->ports_list, list) {
+		bool member = mlxsw_sp_mc_flood(bridge_port);
+
+		err = mlxsw_sp_bridge_ports_flood_table_set(bridge_port,
+							    packet_type,
+							    member);
+		if (err)
+			goto err_flood_table_set;
+	}
+
+	return 0;
+
+err_flood_table_set:
+	list_for_each_entry_continue_reverse(bridge_port,
+					     &bridge_device->ports_list, list) {
+		bool member = mlxsw_sp_mc_flood(bridge_port);
+
+		mlxsw_sp_bridge_ports_flood_table_set(bridge_port, packet_type,
+						      !member);
+	}
+	mlxsw_sp_bridge_mdb_mc_enable_sync(mlxsw_sp, bridge_device,
+					   mc_disabled);
+err_mc_enable_sync:
+	bridge_device->multicast_enabled = mc_disabled;
+	return err;
+}
+
+static struct mlxsw_sp_mdb_entry_port *
+mlxsw_sp_mdb_entry_port_lookup(struct mlxsw_sp_mdb_entry *mdb_entry,
+			       u16 local_port)
+{
+	struct mlxsw_sp_mdb_entry_port *mdb_entry_port;
+
+	list_for_each_entry(mdb_entry_port, &mdb_entry->ports_list, list) {
+		if (mdb_entry_port->local_port == local_port)
+			return mdb_entry_port;
+	}
+
+	return NULL;
+}
+
+static struct mlxsw_sp_mdb_entry_port *
+mlxsw_sp_mdb_entry_port_get(struct mlxsw_sp *mlxsw_sp,
+			    struct mlxsw_sp_mdb_entry *mdb_entry,
+			    u16 local_port)
+{
+	struct mlxsw_sp_mdb_entry_port *mdb_entry_port;
+	int err;
+
+	mdb_entry_port = mlxsw_sp_mdb_entry_port_lookup(mdb_entry, local_port);
+	if (mdb_entry_port) {
+		if (mdb_entry_port->mrouter &&
+		    refcount_read(&mdb_entry_port->refcount) == 1)
+			mdb_entry->ports_count++;
+
+		refcount_inc(&mdb_entry_port->refcount);
+		return mdb_entry_port;
+	}
+
+	err = mlxsw_sp_pgt_entry_port_set(mlxsw_sp, mdb_entry->mid,
+					  mdb_entry->key.fid, local_port, true);
+	if (err)
+		return ERR_PTR(err);
+
+	mdb_entry_port = kzalloc(sizeof(*mdb_entry_port), GFP_KERNEL);
+	if (!mdb_entry_port) {
+		err = -ENOMEM;
+		goto err_mdb_entry_port_alloc;
+	}
+
+	mdb_entry_port->local_port = local_port;
+	refcount_set(&mdb_entry_port->refcount, 1);
+	list_add(&mdb_entry_port->list, &mdb_entry->ports_list);
+	mdb_entry->ports_count++;
+
+	return mdb_entry_port;
+
+err_mdb_entry_port_alloc:
+	mlxsw_sp_pgt_entry_port_set(mlxsw_sp, mdb_entry->mid,
+				    mdb_entry->key.fid, local_port, false);
+	return ERR_PTR(err);
+}
+
+static void
+mlxsw_sp_mdb_entry_port_put(struct mlxsw_sp *mlxsw_sp,
+			    struct mlxsw_sp_mdb_entry *mdb_entry,
+			    u16 local_port, bool force)
+{
+	struct mlxsw_sp_mdb_entry_port *mdb_entry_port;
+
+	mdb_entry_port = mlxsw_sp_mdb_entry_port_lookup(mdb_entry, local_port);
+	if (!mdb_entry_port)
+		return;
+
+	if (!force && !refcount_dec_and_test(&mdb_entry_port->refcount)) {
+		if (mdb_entry_port->mrouter &&
+		    refcount_read(&mdb_entry_port->refcount) == 1)
+			mdb_entry->ports_count--;
+		return;
+	}
+
+	mdb_entry->ports_count--;
+	list_del(&mdb_entry_port->list);
+	kfree(mdb_entry_port);
+	mlxsw_sp_pgt_entry_port_set(mlxsw_sp, mdb_entry->mid,
+				    mdb_entry->key.fid, local_port, false);
+}
+
+static __always_unused struct mlxsw_sp_mdb_entry_port *
+mlxsw_sp_mdb_entry_mrouter_port_get(struct mlxsw_sp *mlxsw_sp,
+				    struct mlxsw_sp_mdb_entry *mdb_entry,
+				    u16 local_port)
+{
+	struct mlxsw_sp_mdb_entry_port *mdb_entry_port;
+	int err;
+
+	mdb_entry_port = mlxsw_sp_mdb_entry_port_lookup(mdb_entry, local_port);
+	if (mdb_entry_port) {
+		if (!mdb_entry_port->mrouter)
+			refcount_inc(&mdb_entry_port->refcount);
+		return mdb_entry_port;
+	}
+
+	err = mlxsw_sp_pgt_entry_port_set(mlxsw_sp, mdb_entry->mid,
+					  mdb_entry->key.fid, local_port, true);
+	if (err)
+		return ERR_PTR(err);
+
+	mdb_entry_port = kzalloc(sizeof(*mdb_entry_port), GFP_KERNEL);
+	if (!mdb_entry_port) {
+		err = -ENOMEM;
+		goto err_mdb_entry_port_alloc;
+	}
+
+	mdb_entry_port->local_port = local_port;
+	refcount_set(&mdb_entry_port->refcount, 1);
+	mdb_entry_port->mrouter = true;
+	list_add(&mdb_entry_port->list, &mdb_entry->ports_list);
+
+	return mdb_entry_port;
+
+err_mdb_entry_port_alloc:
+	mlxsw_sp_pgt_entry_port_set(mlxsw_sp, mdb_entry->mid,
+				    mdb_entry->key.fid, local_port, false);
+	return ERR_PTR(err);
+}
+
+static __always_unused void
+mlxsw_sp_mdb_entry_mrouter_port_put(struct mlxsw_sp *mlxsw_sp,
+				    struct mlxsw_sp_mdb_entry *mdb_entry,
+				    u16 local_port)
+{
+	struct mlxsw_sp_mdb_entry_port *mdb_entry_port;
+
+	mdb_entry_port = mlxsw_sp_mdb_entry_port_lookup(mdb_entry, local_port);
+	if (!mdb_entry_port)
+		return;
+
+	if (!mdb_entry_port->mrouter)
+		return;
+
+	mdb_entry_port->mrouter = false;
+	if (!refcount_dec_and_test(&mdb_entry_port->refcount))
+		return;
+
+	list_del(&mdb_entry_port->list);
+	kfree(mdb_entry_port);
+	mlxsw_sp_pgt_entry_port_set(mlxsw_sp, mdb_entry->mid,
+				    mdb_entry->key.fid, local_port, false);
+}
+
+static void
+mlxsw_sp_bridge_mrouter_update_mdb(struct mlxsw_sp *mlxsw_sp,
+				   struct mlxsw_sp_bridge_device *bridge_device,
+				   bool add)
+{
+	u16 local_port = mlxsw_sp_router_port(mlxsw_sp);
+	struct mlxsw_sp_mdb_entry *mdb_entry;
+
+	list_for_each_entry(mdb_entry, &bridge_device->mdb_list, list) {
+		if (add)
+			mlxsw_sp_mdb_entry_mrouter_port_get(mlxsw_sp, mdb_entry,
+							    local_port);
+		else
+			mlxsw_sp_mdb_entry_mrouter_port_put(mlxsw_sp, mdb_entry,
+							    local_port);
+	}
+}
+
+static int
+mlxsw_sp_port_attr_br_mrouter_set(struct mlxsw_sp_port *mlxsw_sp_port,
+				  struct net_device *orig_dev,
+				  bool is_mrouter)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	struct mlxsw_sp_bridge_device *bridge_device;
+
+	/* It's possible we failed to enslave the port, yet this
+	 * operation is executed due to it being deferred.
+	 */
+	bridge_device = mlxsw_sp_bridge_device_find(mlxsw_sp->bridge, orig_dev);
+	if (!bridge_device)
+		return 0;
+
+	if (bridge_device->mrouter != is_mrouter)
+		mlxsw_sp_bridge_mrouter_update_mdb(mlxsw_sp, bridge_device,
+						   is_mrouter);
+	bridge_device->mrouter = is_mrouter;
+	return 0;
+}
+
+static int mlxsw_sp_port_attr_set(struct net_device *dev, const void *ctx,
+				  const struct switchdev_attr *attr,
+				  struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+	int err;
+
+	switch (attr->id) {
+	case SWITCHDEV_ATTR_ID_PORT_STP_STATE:
+		err = mlxsw_sp_port_attr_stp_state_set(mlxsw_sp_port,
+						       attr->orig_dev,
+						       attr->u.stp_state);
+		break;
+	case SWITCHDEV_ATTR_ID_PORT_PRE_BRIDGE_FLAGS:
+		err = mlxsw_sp_port_attr_br_pre_flags_set(mlxsw_sp_port,
+							  attr->u.brport_flags);
+		break;
+	case SWITCHDEV_ATTR_ID_PORT_BRIDGE_FLAGS:
+		err = mlxsw_sp_port_attr_br_flags_set(mlxsw_sp_port,
+						      attr->orig_dev,
+						      attr->u.brport_flags);
+		break;
+	case SWITCHDEV_ATTR_ID_BRIDGE_AGEING_TIME:
+		err = mlxsw_sp_port_attr_br_ageing_set(mlxsw_sp_port,
+						       attr->u.ageing_time);
+		break;
+	case SWITCHDEV_ATTR_ID_BRIDGE_VLAN_FILTERING:
+		err = mlxsw_sp_port_attr_br_vlan_set(mlxsw_sp_port,
+						     attr->orig_dev,
+						     attr->u.vlan_filtering);
+		break;
+	case SWITCHDEV_ATTR_ID_BRIDGE_VLAN_PROTOCOL:
+		err = mlxsw_sp_port_attr_br_vlan_proto_set(mlxsw_sp_port,
+							   attr->orig_dev,
+							   attr->u.vlan_protocol);
+		break;
+	case SWITCHDEV_ATTR_ID_PORT_MROUTER:
+		err = mlxsw_sp_port_attr_mrouter_set(mlxsw_sp_port,
+						     attr->orig_dev,
+						     attr->u.mrouter);
+		break;
+	case SWITCHDEV_ATTR_ID_BRIDGE_MC_DISABLED:
+		err = mlxsw_sp_port_mc_disabled_set(mlxsw_sp_port,
+						    attr->orig_dev,
+						    attr->u.mc_disabled);
+		break;
+	case SWITCHDEV_ATTR_ID_BRIDGE_MROUTER:
+		err = mlxsw_sp_port_attr_br_mrouter_set(mlxsw_sp_port,
+							attr->orig_dev,
+							attr->u.mrouter);
+		break;
+	default:
+		err = -EOPNOTSUPP;
+		break;
+	}
+
+	mlxsw_sp_span_respin(mlxsw_sp_port->mlxsw_sp);
+
+	return err;
+}
+
+static int
+mlxsw_sp_port_vlan_fid_join(struct mlxsw_sp_port_vlan *mlxsw_sp_port_vlan,
+			    struct mlxsw_sp_bridge_port *bridge_port,
+			    struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = mlxsw_sp_port_vlan->mlxsw_sp_port;
+	struct mlxsw_sp_bridge_device *bridge_device;
+	u16 local_port = mlxsw_sp_port->local_port;
+	u16 vid = mlxsw_sp_port_vlan->vid;
+	struct mlxsw_sp_fid *fid;
+	int err;
+
+	bridge_device = bridge_port->bridge_device;
+	fid = bridge_device->ops->fid_get(bridge_device, vid, extack);
+	if (IS_ERR(fid))
+		return PTR_ERR(fid);
+
+	err = mlxsw_sp_fid_flood_set(fid, MLXSW_SP_FLOOD_TYPE_UC, local_port,
+				     bridge_port->flags & BR_FLOOD);
+	if (err)
+		goto err_fid_uc_flood_set;
+
+	err = mlxsw_sp_fid_flood_set(fid, MLXSW_SP_FLOOD_TYPE_MC, local_port,
+				     mlxsw_sp_mc_flood(bridge_port));
+	if (err)
+		goto err_fid_mc_flood_set;
+
+	err = mlxsw_sp_fid_flood_set(fid, MLXSW_SP_FLOOD_TYPE_BC, local_port,
+				     true);
+	if (err)
+		goto err_fid_bc_flood_set;
+
+	err = mlxsw_sp_fid_port_vid_map(fid, mlxsw_sp_port, vid);
+	if (err)
+		goto err_fid_port_vid_map;
+
+	mlxsw_sp_port_vlan->fid = fid;
+
+	return 0;
+
+err_fid_port_vid_map:
+	mlxsw_sp_fid_flood_set(fid, MLXSW_SP_FLOOD_TYPE_BC, local_port, false);
+err_fid_bc_flood_set:
+	mlxsw_sp_fid_flood_set(fid, MLXSW_SP_FLOOD_TYPE_MC, local_port, false);
+err_fid_mc_flood_set:
+	mlxsw_sp_fid_flood_set(fid, MLXSW_SP_FLOOD_TYPE_UC, local_port, false);
+err_fid_uc_flood_set:
+	mlxsw_sp_fid_put(fid);
+	return err;
+}
+
+static void
+mlxsw_sp_port_vlan_fid_leave(struct mlxsw_sp_port_vlan *mlxsw_sp_port_vlan)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = mlxsw_sp_port_vlan->mlxsw_sp_port;
+	struct mlxsw_sp_fid *fid = mlxsw_sp_port_vlan->fid;
+	u16 local_port = mlxsw_sp_port->local_port;
+	u16 vid = mlxsw_sp_port_vlan->vid;
+
+	mlxsw_sp_port_vlan->fid = NULL;
+	mlxsw_sp_fid_port_vid_unmap(fid, mlxsw_sp_port, vid);
+	mlxsw_sp_fid_flood_set(fid, MLXSW_SP_FLOOD_TYPE_BC, local_port, false);
+	mlxsw_sp_fid_flood_set(fid, MLXSW_SP_FLOOD_TYPE_MC, local_port, false);
+	mlxsw_sp_fid_flood_set(fid, MLXSW_SP_FLOOD_TYPE_UC, local_port, false);
+	mlxsw_sp_fid_put(fid);
+}
+
+static u16
+mlxsw_sp_port_pvid_determine(const struct mlxsw_sp_port *mlxsw_sp_port,
+			     u16 vid, bool is_pvid)
+{
+	if (is_pvid)
+		return vid;
+	else if (mlxsw_sp_port->pvid == vid)
+		return 0;	/* Dis-allow untagged packets */
+	else
+		return mlxsw_sp_port->pvid;
+}
+
+static int
+mlxsw_sp_port_vlan_bridge_join(struct mlxsw_sp_port_vlan *mlxsw_sp_port_vlan,
+			       struct mlxsw_sp_bridge_port *bridge_port,
+			       struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = mlxsw_sp_port_vlan->mlxsw_sp_port;
+	struct mlxsw_sp_bridge_vlan *bridge_vlan;
+	u16 vid = mlxsw_sp_port_vlan->vid;
+	int err;
+
+	/* No need to continue if only VLAN flags were changed */
+	if (mlxsw_sp_port_vlan->bridge_port)
+		return 0;
+
+	err = mlxsw_sp_port_vlan_fid_join(mlxsw_sp_port_vlan, bridge_port,
+					  extack);
+	if (err)
+		return err;
+
+	err = mlxsw_sp_port_vid_learning_set(mlxsw_sp_port, vid,
+					     bridge_port->flags & BR_LEARNING);
+	if (err)
+		goto err_port_vid_learning_set;
+
+	err = mlxsw_sp_port_vid_stp_set(mlxsw_sp_port, vid,
+					bridge_port->stp_state);
+	if (err)
+		goto err_port_vid_stp_set;
+
+	bridge_vlan = mlxsw_sp_bridge_vlan_get(bridge_port, vid);
+	if (!bridge_vlan) {
+		err = -ENOMEM;
+		goto err_bridge_vlan_get;
+	}
+
+	list_add(&mlxsw_sp_port_vlan->bridge_vlan_node,
+		 &bridge_vlan->port_vlan_list);
+
+	mlxsw_sp_bridge_port_get(mlxsw_sp_port->mlxsw_sp->bridge,
+				 bridge_port->dev, extack);
+	mlxsw_sp_port_vlan->bridge_port = bridge_port;
+
+	return 0;
+
+err_bridge_vlan_get:
+	mlxsw_sp_port_vid_stp_set(mlxsw_sp_port, vid, BR_STATE_DISABLED);
+err_port_vid_stp_set:
+	mlxsw_sp_port_vid_learning_set(mlxsw_sp_port, vid, false);
+err_port_vid_learning_set:
+	mlxsw_sp_port_vlan_fid_leave(mlxsw_sp_port_vlan);
+	return err;
+}
+
+void
+mlxsw_sp_port_vlan_bridge_leave(struct mlxsw_sp_port_vlan *mlxsw_sp_port_vlan)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = mlxsw_sp_port_vlan->mlxsw_sp_port;
+	struct mlxsw_sp_fid *fid = mlxsw_sp_port_vlan->fid;
+	struct mlxsw_sp_bridge_vlan *bridge_vlan;
+	struct mlxsw_sp_bridge_port *bridge_port;
+	u16 vid = mlxsw_sp_port_vlan->vid;
+	bool last_port;
+
+	if (WARN_ON(mlxsw_sp_fid_type(fid) != MLXSW_SP_FID_TYPE_8021Q &&
+		    mlxsw_sp_fid_type(fid) != MLXSW_SP_FID_TYPE_8021D))
+		return;
+
+	bridge_port = mlxsw_sp_port_vlan->bridge_port;
+	bridge_vlan = mlxsw_sp_bridge_vlan_find(bridge_port, vid);
+	last_port = list_is_singular(&bridge_vlan->port_vlan_list);
+
+	list_del(&mlxsw_sp_port_vlan->bridge_vlan_node);
+	mlxsw_sp_bridge_vlan_put(bridge_vlan);
+	mlxsw_sp_port_vid_stp_set(mlxsw_sp_port, vid, BR_STATE_DISABLED);
+	mlxsw_sp_port_vid_learning_set(mlxsw_sp_port, vid, false);
+	if (last_port)
+		mlxsw_sp_bridge_port_fdb_flush(mlxsw_sp_port->mlxsw_sp,
+					       bridge_port,
+					       mlxsw_sp_fid_index(fid));
+
+	mlxsw_sp_bridge_port_mdb_flush(mlxsw_sp_port, bridge_port,
+				       mlxsw_sp_fid_index(fid));
+
+	mlxsw_sp_port_vlan_fid_leave(mlxsw_sp_port_vlan);
+
+	mlxsw_sp_bridge_port_put(mlxsw_sp_port->mlxsw_sp->bridge, bridge_port);
+	mlxsw_sp_port_vlan->bridge_port = NULL;
+}
+
+static int
+mlxsw_sp_bridge_port_vlan_add(struct mlxsw_sp_port *mlxsw_sp_port,
+			      struct mlxsw_sp_bridge_port *bridge_port,
+			      u16 vid, bool is_untagged, bool is_pvid,
+			      struct netlink_ext_ack *extack)
+{
+	u16 pvid = mlxsw_sp_port_pvid_determine(mlxsw_sp_port, vid, is_pvid);
+	struct mlxsw_sp_port_vlan *mlxsw_sp_port_vlan;
+	u16 old_pvid = mlxsw_sp_port->pvid;
+	u16 proto;
+	int err;
+
+	/* The only valid scenario in which a port-vlan already exists, is if
+	 * the VLAN flags were changed and the port-vlan is associated with the
+	 * correct bridge port
+	 */
+	mlxsw_sp_port_vlan = mlxsw_sp_port_vlan_find_by_vid(mlxsw_sp_port, vid);
+	if (mlxsw_sp_port_vlan &&
+	    mlxsw_sp_port_vlan->bridge_port != bridge_port)
+		return -EEXIST;
+
+	if (!mlxsw_sp_port_vlan) {
+		mlxsw_sp_port_vlan = mlxsw_sp_port_vlan_create(mlxsw_sp_port,
+							       vid);
+		if (IS_ERR(mlxsw_sp_port_vlan))
+			return PTR_ERR(mlxsw_sp_port_vlan);
+	}
+
+	err = mlxsw_sp_port_vlan_set(mlxsw_sp_port, vid, vid, true,
+				     is_untagged);
+	if (err)
+		goto err_port_vlan_set;
+
+	br_vlan_get_proto(bridge_port->bridge_device->dev, &proto);
+	err = mlxsw_sp_port_pvid_set(mlxsw_sp_port, pvid, proto);
+	if (err)
+		goto err_port_pvid_set;
+
+	err = mlxsw_sp_port_vlan_bridge_join(mlxsw_sp_port_vlan, bridge_port,
+					     extack);
+	if (err)
+		goto err_port_vlan_bridge_join;
+
+	return 0;
+
+err_port_vlan_bridge_join:
+	mlxsw_sp_port_pvid_set(mlxsw_sp_port, old_pvid, proto);
+err_port_pvid_set:
+	mlxsw_sp_port_vlan_set(mlxsw_sp_port, vid, vid, false, false);
+err_port_vlan_set:
+	mlxsw_sp_port_vlan_destroy(mlxsw_sp_port_vlan);
+	return err;
+}
+
+static int
+mlxsw_sp_br_ban_rif_pvid_change(struct mlxsw_sp *mlxsw_sp,
+				const struct net_device *br_dev,
+				const struct switchdev_obj_port_vlan *vlan)
+{
+	u16 pvid;
+
+	pvid = mlxsw_sp_rif_vid(mlxsw_sp, br_dev);
+	if (!pvid)
+		return 0;
+
+	if (vlan->flags & BRIDGE_VLAN_INFO_PVID) {
+		if (vlan->vid != pvid) {
+			netdev_err(br_dev, "Can't change PVID, it's used by router interface\n");
+			return -EBUSY;
+		}
+	} else {
+		if (vlan->vid == pvid) {
+			netdev_err(br_dev, "Can't remove PVID, it's used by router interface\n");
+			return -EBUSY;
+		}
+	}
+
+	return 0;
+}
+
+static int mlxsw_sp_port_vlans_add(struct mlxsw_sp_port *mlxsw_sp_port,
+				   const struct switchdev_obj_port_vlan *vlan,
+				   struct netlink_ext_ack *extack)
+{
+	bool flag_untagged = vlan->flags & BRIDGE_VLAN_INFO_UNTAGGED;
+	bool flag_pvid = vlan->flags & BRIDGE_VLAN_INFO_PVID;
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	struct net_device *orig_dev = vlan->obj.orig_dev;
+	struct mlxsw_sp_bridge_port *bridge_port;
+
+	if (netif_is_bridge_master(orig_dev)) {
+		int err = 0;
+
+		if (br_vlan_enabled(orig_dev))
+			err = mlxsw_sp_br_ban_rif_pvid_change(mlxsw_sp,
+							      orig_dev, vlan);
+		if (!err)
+			err = -EOPNOTSUPP;
+		return err;
+	}
+
+	bridge_port = mlxsw_sp_bridge_port_find(mlxsw_sp->bridge, orig_dev);
+	if (WARN_ON(!bridge_port))
+		return -EINVAL;
+
+	if (!bridge_port->bridge_device->vlan_enabled)
+		return 0;
+
+	return mlxsw_sp_bridge_port_vlan_add(mlxsw_sp_port, bridge_port,
+					     vlan->vid, flag_untagged,
+					     flag_pvid, extack);
+}
+
+static enum mlxsw_reg_sfdf_flush_type mlxsw_sp_fdb_flush_type(bool lagged)
+{
+	return lagged ? MLXSW_REG_SFDF_FLUSH_PER_LAG_AND_FID :
+			MLXSW_REG_SFDF_FLUSH_PER_PORT_AND_FID;
+}
+
+static int
+mlxsw_sp_bridge_port_fdb_flush(struct mlxsw_sp *mlxsw_sp,
+			       struct mlxsw_sp_bridge_port *bridge_port,
+			       u16 fid_index)
+{
+	bool lagged = bridge_port->lagged;
+	char sfdf_pl[MLXSW_REG_SFDF_LEN];
+	u16 system_port;
+
+	system_port = lagged ? bridge_port->lag_id : bridge_port->system_port;
+	mlxsw_reg_sfdf_pack(sfdf_pl, mlxsw_sp_fdb_flush_type(lagged));
+	mlxsw_reg_sfdf_fid_set(sfdf_pl, fid_index);
+	mlxsw_reg_sfdf_port_fid_system_port_set(sfdf_pl, system_port);
+
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(sfdf), sfdf_pl);
+}
+
+static enum mlxsw_reg_sfd_rec_policy mlxsw_sp_sfd_rec_policy(bool dynamic)
+{
+	return dynamic ? MLXSW_REG_SFD_REC_POLICY_DYNAMIC_ENTRY_INGRESS :
+			 MLXSW_REG_SFD_REC_POLICY_DYNAMIC_ENTRY_MLAG;
+}
+
+static enum mlxsw_reg_sfd_op mlxsw_sp_sfd_op(bool adding)
+{
+	return adding ? MLXSW_REG_SFD_OP_WRITE_EDIT :
+			MLXSW_REG_SFD_OP_WRITE_REMOVE;
+}
+
+static int
+mlxsw_sp_port_fdb_tun_uc_op4(struct mlxsw_sp *mlxsw_sp, bool dynamic,
+			     const char *mac, u16 fid, __be32 addr, bool adding)
+{
+	char *sfd_pl;
+	u8 num_rec;
+	u32 uip;
+	int err;
+
+	sfd_pl = kmalloc(MLXSW_REG_SFD_LEN, GFP_KERNEL);
+	if (!sfd_pl)
+		return -ENOMEM;
+
+	uip = be32_to_cpu(addr);
+	mlxsw_reg_sfd_pack(sfd_pl, mlxsw_sp_sfd_op(adding), 0);
+	mlxsw_reg_sfd_uc_tunnel_pack4(sfd_pl, 0,
+				      mlxsw_sp_sfd_rec_policy(dynamic), mac,
+				      fid, MLXSW_REG_SFD_REC_ACTION_NOP, uip);
+	num_rec = mlxsw_reg_sfd_num_rec_get(sfd_pl);
+	err = mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(sfd), sfd_pl);
+	if (err)
+		goto out;
+
+	if (num_rec != mlxsw_reg_sfd_num_rec_get(sfd_pl))
+		err = -EBUSY;
+
+out:
+	kfree(sfd_pl);
+	return err;
+}
+
+static int mlxsw_sp_port_fdb_tun_uc_op6_sfd_write(struct mlxsw_sp *mlxsw_sp,
+						  const char *mac, u16 fid,
+						  u32 kvdl_index, bool adding)
+{
+	char *sfd_pl;
+	u8 num_rec;
+	int err;
+
+	sfd_pl = kmalloc(MLXSW_REG_SFD_LEN, GFP_KERNEL);
+	if (!sfd_pl)
+		return -ENOMEM;
+
+	mlxsw_reg_sfd_pack(sfd_pl, mlxsw_sp_sfd_op(adding), 0);
+	mlxsw_reg_sfd_uc_tunnel_pack6(sfd_pl, 0, mac, fid,
+				      MLXSW_REG_SFD_REC_ACTION_NOP, kvdl_index);
+	num_rec = mlxsw_reg_sfd_num_rec_get(sfd_pl);
+	err = mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(sfd), sfd_pl);
+	if (err)
+		goto out;
+
+	if (num_rec != mlxsw_reg_sfd_num_rec_get(sfd_pl))
+		err = -EBUSY;
+
+out:
+	kfree(sfd_pl);
+	return err;
+}
+
+static int mlxsw_sp_port_fdb_tun_uc_op6_add(struct mlxsw_sp *mlxsw_sp,
+					    const char *mac, u16 fid,
+					    const struct in6_addr *addr)
+{
+	u32 kvdl_index;
+	int err;
+
+	err = mlxsw_sp_nve_ipv6_addr_kvdl_set(mlxsw_sp, addr, &kvdl_index);
+	if (err)
+		return err;
+
+	err = mlxsw_sp_port_fdb_tun_uc_op6_sfd_write(mlxsw_sp, mac, fid,
+						     kvdl_index, true);
+	if (err)
+		goto err_sfd_write;
+
+	err = mlxsw_sp_nve_ipv6_addr_map_replace(mlxsw_sp, mac, fid, addr);
+	if (err)
+		/* Replace can fail only for creating new mapping, so removing
+		 * the FDB entry in the error path is OK.
+		 */
+		goto err_addr_replace;
+
+	return 0;
+
+err_addr_replace:
+	mlxsw_sp_port_fdb_tun_uc_op6_sfd_write(mlxsw_sp, mac, fid, kvdl_index,
+					       false);
+err_sfd_write:
+	mlxsw_sp_nve_ipv6_addr_kvdl_unset(mlxsw_sp, addr);
+	return err;
+}
+
+static void mlxsw_sp_port_fdb_tun_uc_op6_del(struct mlxsw_sp *mlxsw_sp,
+					     const char *mac, u16 fid,
+					     const struct in6_addr *addr)
+{
+	mlxsw_sp_nve_ipv6_addr_map_del(mlxsw_sp, mac, fid);
+	mlxsw_sp_port_fdb_tun_uc_op6_sfd_write(mlxsw_sp, mac, fid, 0, false);
+	mlxsw_sp_nve_ipv6_addr_kvdl_unset(mlxsw_sp, addr);
+}
+
+static int
+mlxsw_sp_port_fdb_tun_uc_op6(struct mlxsw_sp *mlxsw_sp, const char *mac,
+			     u16 fid, const struct in6_addr *addr, bool adding)
+{
+	if (adding)
+		return mlxsw_sp_port_fdb_tun_uc_op6_add(mlxsw_sp, mac, fid,
+							addr);
+
+	mlxsw_sp_port_fdb_tun_uc_op6_del(mlxsw_sp, mac, fid, addr);
+	return 0;
+}
+
+static int mlxsw_sp_port_fdb_tunnel_uc_op(struct mlxsw_sp *mlxsw_sp,
+					  const char *mac, u16 fid,
+					  enum mlxsw_sp_l3proto proto,
+					  const union mlxsw_sp_l3addr *addr,
+					  bool adding, bool dynamic)
+{
+	switch (proto) {
+	case MLXSW_SP_L3_PROTO_IPV4:
+		return mlxsw_sp_port_fdb_tun_uc_op4(mlxsw_sp, dynamic, mac, fid,
+						    addr->addr4, adding);
+	case MLXSW_SP_L3_PROTO_IPV6:
+		return mlxsw_sp_port_fdb_tun_uc_op6(mlxsw_sp, mac, fid,
+						    &addr->addr6, adding);
+	default:
+		WARN_ON(1);
+		return -EOPNOTSUPP;
+	}
+}
+
+static int __mlxsw_sp_port_fdb_uc_op(struct mlxsw_sp *mlxsw_sp, u16 local_port,
+				     const char *mac, u16 fid, u16 vid,
+				     bool adding,
+				     enum mlxsw_reg_sfd_rec_action action,
+				     enum mlxsw_reg_sfd_rec_policy policy)
+{
+	char *sfd_pl;
+	u8 num_rec;
+	int err;
+
+	sfd_pl = kmalloc(MLXSW_REG_SFD_LEN, GFP_KERNEL);
+	if (!sfd_pl)
+		return -ENOMEM;
+
+	mlxsw_reg_sfd_pack(sfd_pl, mlxsw_sp_sfd_op(adding), 0);
+	mlxsw_reg_sfd_uc_pack(sfd_pl, 0, policy, mac, fid, vid, action,
+			      local_port);
+	num_rec = mlxsw_reg_sfd_num_rec_get(sfd_pl);
+	err = mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(sfd), sfd_pl);
+	if (err)
+		goto out;
+
+	if (num_rec != mlxsw_reg_sfd_num_rec_get(sfd_pl))
+		err = -EBUSY;
+
+out:
+	kfree(sfd_pl);
+	return err;
+}
+
+static int mlxsw_sp_port_fdb_uc_op(struct mlxsw_sp *mlxsw_sp, u16 local_port,
+				   const char *mac, u16 fid, u16 vid,
+				   bool adding, bool dynamic)
+{
+	return __mlxsw_sp_port_fdb_uc_op(mlxsw_sp, local_port, mac, fid, vid,
+					 adding, MLXSW_REG_SFD_REC_ACTION_NOP,
+					 mlxsw_sp_sfd_rec_policy(dynamic));
+}
+
+int mlxsw_sp_rif_fdb_op(struct mlxsw_sp *mlxsw_sp, const char *mac, u16 fid,
+			bool adding)
+{
+	return __mlxsw_sp_port_fdb_uc_op(mlxsw_sp, 0, mac, fid, 0, adding,
+					 MLXSW_REG_SFD_REC_ACTION_FORWARD_IP_ROUTER,
+					 MLXSW_REG_SFD_REC_POLICY_STATIC_ENTRY);
+}
+
+static int mlxsw_sp_port_fdb_uc_lag_op(struct mlxsw_sp *mlxsw_sp, u16 lag_id,
+				       const char *mac, u16 fid, u16 lag_vid,
+				       bool adding, bool dynamic)
+{
+	char *sfd_pl;
+	u8 num_rec;
+	int err;
+
+	sfd_pl = kmalloc(MLXSW_REG_SFD_LEN, GFP_KERNEL);
+	if (!sfd_pl)
+		return -ENOMEM;
+
+	mlxsw_reg_sfd_pack(sfd_pl, mlxsw_sp_sfd_op(adding), 0);
+	mlxsw_reg_sfd_uc_lag_pack(sfd_pl, 0, mlxsw_sp_sfd_rec_policy(dynamic),
+				  mac, fid, MLXSW_REG_SFD_REC_ACTION_NOP,
+				  lag_vid, lag_id);
+	num_rec = mlxsw_reg_sfd_num_rec_get(sfd_pl);
+	err = mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(sfd), sfd_pl);
+	if (err)
+		goto out;
+
+	if (num_rec != mlxsw_reg_sfd_num_rec_get(sfd_pl))
+		err = -EBUSY;
+
+out:
+	kfree(sfd_pl);
+	return err;
+}
+
+static int
+mlxsw_sp_port_fdb_set(struct mlxsw_sp_port *mlxsw_sp_port,
+		      struct switchdev_notifier_fdb_info *fdb_info, bool adding)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	struct net_device *orig_dev = fdb_info->info.dev;
+	struct mlxsw_sp_port_vlan *mlxsw_sp_port_vlan;
+	struct mlxsw_sp_bridge_device *bridge_device;
+	struct mlxsw_sp_bridge_port *bridge_port;
+	u16 fid_index, vid;
+
+	bridge_port = mlxsw_sp_bridge_port_find(mlxsw_sp->bridge, orig_dev);
+	if (!bridge_port)
+		return -EINVAL;
+
+	bridge_device = bridge_port->bridge_device;
+	mlxsw_sp_port_vlan = mlxsw_sp_port_vlan_find_by_bridge(mlxsw_sp_port,
+							       bridge_device,
+							       fdb_info->vid);
+	if (!mlxsw_sp_port_vlan)
+		return 0;
+
+	fid_index = mlxsw_sp_fid_index(mlxsw_sp_port_vlan->fid);
+	vid = mlxsw_sp_port_vlan->vid;
+
+	if (!bridge_port->lagged)
+		return mlxsw_sp_port_fdb_uc_op(mlxsw_sp,
+					       bridge_port->system_port,
+					       fdb_info->addr, fid_index, vid,
+					       adding, false);
+	else
+		return mlxsw_sp_port_fdb_uc_lag_op(mlxsw_sp,
+						   bridge_port->lag_id,
+						   fdb_info->addr, fid_index,
+						   vid, adding, false);
+}
+
+static int mlxsw_sp_mdb_entry_write(struct mlxsw_sp *mlxsw_sp,
+				    const struct mlxsw_sp_mdb_entry *mdb_entry,
+				    bool adding)
+{
+	char *sfd_pl;
+	u8 num_rec;
+	int err;
+
+	sfd_pl = kmalloc(MLXSW_REG_SFD_LEN, GFP_KERNEL);
+	if (!sfd_pl)
+		return -ENOMEM;
+
+	mlxsw_reg_sfd_pack(sfd_pl, mlxsw_sp_sfd_op(adding), 0);
+	mlxsw_reg_sfd_mc_pack(sfd_pl, 0, mdb_entry->key.addr,
+			      mdb_entry->key.fid, MLXSW_REG_SFD_REC_ACTION_NOP,
+			      mdb_entry->mid);
+	num_rec = mlxsw_reg_sfd_num_rec_get(sfd_pl);
+	err = mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(sfd), sfd_pl);
+	if (err)
+		goto out;
+
+	if (num_rec != mlxsw_reg_sfd_num_rec_get(sfd_pl))
+		err = -EBUSY;
+
+out:
+	kfree(sfd_pl);
+	return err;
+}
+
+static void
+mlxsw_sp_bridge_port_get_ports_bitmap(struct mlxsw_sp *mlxsw_sp,
+				      struct mlxsw_sp_bridge_port *bridge_port,
+				      struct mlxsw_sp_ports_bitmap *ports_bm)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port;
+	u64 max_lag_members, i;
+	int lag_id;
+
+	if (!bridge_port->lagged) {
+		set_bit(bridge_port->system_port, ports_bm->bitmap);
+	} else {
+		max_lag_members = MLXSW_CORE_RES_GET(mlxsw_sp->core,
+						     MAX_LAG_MEMBERS);
+		lag_id = bridge_port->lag_id;
+		for (i = 0; i < max_lag_members; i++) {
+			mlxsw_sp_port = mlxsw_sp_port_lagged_get(mlxsw_sp,
+								 lag_id, i);
+			if (mlxsw_sp_port)
+				set_bit(mlxsw_sp_port->local_port,
+					ports_bm->bitmap);
+		}
+	}
+}
+
+static void
+mlxsw_sp_mc_get_mrouters_bitmap(struct mlxsw_sp_ports_bitmap *flood_bm,
+				struct mlxsw_sp_bridge_device *bridge_device,
+				struct mlxsw_sp *mlxsw_sp)
+{
+	struct mlxsw_sp_bridge_port *bridge_port;
+
+	list_for_each_entry(bridge_port, &bridge_device->ports_list, list) {
+		if (bridge_port->mrouter) {
+			mlxsw_sp_bridge_port_get_ports_bitmap(mlxsw_sp,
+							      bridge_port,
+							      flood_bm);
+		}
+	}
+}
+
+static int mlxsw_sp_mc_mdb_mrouters_add(struct mlxsw_sp *mlxsw_sp,
+					struct mlxsw_sp_ports_bitmap *ports_bm,
+					struct mlxsw_sp_mdb_entry *mdb_entry)
+{
+	struct mlxsw_sp_mdb_entry_port *mdb_entry_port;
+	unsigned int nbits = ports_bm->nbits;
+	int i;
+
+	for_each_set_bit(i, ports_bm->bitmap, nbits) {
+		mdb_entry_port = mlxsw_sp_mdb_entry_mrouter_port_get(mlxsw_sp,
+								     mdb_entry,
+								     i);
+		if (IS_ERR(mdb_entry_port)) {
+			nbits = i;
+			goto err_mrouter_port_get;
+		}
+	}
+
+	return 0;
+
+err_mrouter_port_get:
+	for_each_set_bit(i, ports_bm->bitmap, nbits)
+		mlxsw_sp_mdb_entry_mrouter_port_put(mlxsw_sp, mdb_entry, i);
+	return PTR_ERR(mdb_entry_port);
+}
+
+static void mlxsw_sp_mc_mdb_mrouters_del(struct mlxsw_sp *mlxsw_sp,
+					 struct mlxsw_sp_ports_bitmap *ports_bm,
+					 struct mlxsw_sp_mdb_entry *mdb_entry)
+{
+	int i;
+
+	for_each_set_bit(i, ports_bm->bitmap, ports_bm->nbits)
+		mlxsw_sp_mdb_entry_mrouter_port_put(mlxsw_sp, mdb_entry, i);
+}
+
+static int
+mlxsw_sp_mc_mdb_mrouters_set(struct mlxsw_sp *mlxsw_sp,
+			     struct mlxsw_sp_bridge_device *bridge_device,
+			     struct mlxsw_sp_mdb_entry *mdb_entry, bool add)
+{
+	struct mlxsw_sp_ports_bitmap ports_bm;
+	int err;
+
+	err = mlxsw_sp_port_bitmap_init(mlxsw_sp, &ports_bm);
+	if (err)
+		return err;
+
+	mlxsw_sp_mc_get_mrouters_bitmap(&ports_bm, bridge_device, mlxsw_sp);
+
+	if (add)
+		err = mlxsw_sp_mc_mdb_mrouters_add(mlxsw_sp, &ports_bm,
+						   mdb_entry);
+	else
+		mlxsw_sp_mc_mdb_mrouters_del(mlxsw_sp, &ports_bm, mdb_entry);
+
+	mlxsw_sp_port_bitmap_fini(&ports_bm);
+	return err;
+}
+
+static struct mlxsw_sp_mdb_entry *
+mlxsw_sp_mc_mdb_entry_init(struct mlxsw_sp *mlxsw_sp,
+			   struct mlxsw_sp_bridge_device *bridge_device,
+			   const unsigned char *addr, u16 fid, u16 local_port)
+{
+	struct mlxsw_sp_mdb_entry_port *mdb_entry_port;
+	struct mlxsw_sp_mdb_entry *mdb_entry;
+	int err;
+
+	mdb_entry = kzalloc(sizeof(*mdb_entry), GFP_KERNEL);
+	if (!mdb_entry)
+		return ERR_PTR(-ENOMEM);
+
+	ether_addr_copy(mdb_entry->key.addr, addr);
+	mdb_entry->key.fid = fid;
+	err = mlxsw_sp_pgt_mid_alloc(mlxsw_sp, &mdb_entry->mid);
+	if (err)
+		goto err_pgt_mid_alloc;
+
+	INIT_LIST_HEAD(&mdb_entry->ports_list);
+
+	err = mlxsw_sp_mc_mdb_mrouters_set(mlxsw_sp, bridge_device, mdb_entry,
+					   true);
+	if (err)
+		goto err_mdb_mrouters_set;
+
+	mdb_entry_port = mlxsw_sp_mdb_entry_port_get(mlxsw_sp, mdb_entry,
+						     local_port);
+	if (IS_ERR(mdb_entry_port)) {
+		err = PTR_ERR(mdb_entry_port);
+		goto err_mdb_entry_port_get;
+	}
+
+	if (bridge_device->multicast_enabled) {
+		err = mlxsw_sp_mdb_entry_write(mlxsw_sp, mdb_entry, true);
+		if (err)
+			goto err_mdb_entry_write;
+	}
+
+	err = rhashtable_insert_fast(&bridge_device->mdb_ht,
+				     &mdb_entry->ht_node,
+				     mlxsw_sp_mdb_ht_params);
+	if (err)
+		goto err_rhashtable_insert;
+
+	list_add_tail(&mdb_entry->list, &bridge_device->mdb_list);
+
+	return mdb_entry;
+
+err_rhashtable_insert:
+	if (bridge_device->multicast_enabled)
+		mlxsw_sp_mdb_entry_write(mlxsw_sp, mdb_entry, false);
+err_mdb_entry_write:
+	mlxsw_sp_mdb_entry_port_put(mlxsw_sp, mdb_entry, local_port, false);
+err_mdb_entry_port_get:
+	mlxsw_sp_mc_mdb_mrouters_set(mlxsw_sp, bridge_device, mdb_entry, false);
+err_mdb_mrouters_set:
+	mlxsw_sp_pgt_mid_free(mlxsw_sp, mdb_entry->mid);
+err_pgt_mid_alloc:
+	kfree(mdb_entry);
+	return ERR_PTR(err);
+}
+
+static void
+mlxsw_sp_mc_mdb_entry_fini(struct mlxsw_sp *mlxsw_sp,
+			   struct mlxsw_sp_mdb_entry *mdb_entry,
+			   struct mlxsw_sp_bridge_device *bridge_device,
+			   u16 local_port, bool force)
+{
+	list_del(&mdb_entry->list);
+	rhashtable_remove_fast(&bridge_device->mdb_ht, &mdb_entry->ht_node,
+			       mlxsw_sp_mdb_ht_params);
+	if (bridge_device->multicast_enabled)
+		mlxsw_sp_mdb_entry_write(mlxsw_sp, mdb_entry, false);
+	mlxsw_sp_mdb_entry_port_put(mlxsw_sp, mdb_entry, local_port, force);
+	mlxsw_sp_mc_mdb_mrouters_set(mlxsw_sp, bridge_device, mdb_entry, false);
+	WARN_ON(!list_empty(&mdb_entry->ports_list));
+	mlxsw_sp_pgt_mid_free(mlxsw_sp, mdb_entry->mid);
+	kfree(mdb_entry);
+}
+
+static struct mlxsw_sp_mdb_entry *
+mlxsw_sp_mc_mdb_entry_get(struct mlxsw_sp *mlxsw_sp,
+			  struct mlxsw_sp_bridge_device *bridge_device,
+			  const unsigned char *addr, u16 fid, u16 local_port)
+{
+	struct mlxsw_sp_mdb_entry_key key = {};
+	struct mlxsw_sp_mdb_entry *mdb_entry;
+
+	ether_addr_copy(key.addr, addr);
+	key.fid = fid;
+	mdb_entry = rhashtable_lookup_fast(&bridge_device->mdb_ht, &key,
+					   mlxsw_sp_mdb_ht_params);
+	if (mdb_entry) {
+		struct mlxsw_sp_mdb_entry_port *mdb_entry_port;
+
+		mdb_entry_port = mlxsw_sp_mdb_entry_port_get(mlxsw_sp,
+							     mdb_entry,
+							     local_port);
+		if (IS_ERR(mdb_entry_port))
+			return ERR_CAST(mdb_entry_port);
+
+		return mdb_entry;
+	}
+
+	return mlxsw_sp_mc_mdb_entry_init(mlxsw_sp, bridge_device, addr, fid,
+					  local_port);
+}
+
+static bool
+mlxsw_sp_mc_mdb_entry_remove(struct mlxsw_sp_mdb_entry *mdb_entry,
+			     struct mlxsw_sp_mdb_entry_port *removed_entry_port,
+			     bool force)
+{
+	if (mdb_entry->ports_count > 1)
+		return false;
+
+	if (force)
+		return true;
+
+	if (!removed_entry_port->mrouter &&
+	    refcount_read(&removed_entry_port->refcount) > 1)
+		return false;
+
+	if (removed_entry_port->mrouter &&
+	    refcount_read(&removed_entry_port->refcount) > 2)
+		return false;
+
+	return true;
+}
+
+static void
+mlxsw_sp_mc_mdb_entry_put(struct mlxsw_sp *mlxsw_sp,
+			  struct mlxsw_sp_bridge_device *bridge_device,
+			  struct mlxsw_sp_mdb_entry *mdb_entry, u16 local_port,
+			  bool force)
+{
+	struct mlxsw_sp_mdb_entry_port *mdb_entry_port;
+
+	mdb_entry_port = mlxsw_sp_mdb_entry_port_lookup(mdb_entry, local_port);
+	if (!mdb_entry_port)
+		return;
+
+	/* Avoid a temporary situation in which the MDB entry points to an empty
+	 * PGT entry, as otherwise packets will be temporarily dropped instead
+	 * of being flooded. Instead, in this situation, call
+	 * mlxsw_sp_mc_mdb_entry_fini(), which first deletes the MDB entry and
+	 * then releases the PGT entry.
+	 */
+	if (mlxsw_sp_mc_mdb_entry_remove(mdb_entry, mdb_entry_port, force))
+		mlxsw_sp_mc_mdb_entry_fini(mlxsw_sp, mdb_entry, bridge_device,
+					   local_port, force);
+	else
+		mlxsw_sp_mdb_entry_port_put(mlxsw_sp, mdb_entry, local_port,
+					    force);
+}
+
+static int mlxsw_sp_port_mdb_add(struct mlxsw_sp_port *mlxsw_sp_port,
+				 const struct switchdev_obj_port_mdb *mdb)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	struct net_device *orig_dev = mdb->obj.orig_dev;
+	struct mlxsw_sp_port_vlan *mlxsw_sp_port_vlan;
+	struct mlxsw_sp_bridge_device *bridge_device;
+	struct mlxsw_sp_bridge_port *bridge_port;
+	struct mlxsw_sp_mdb_entry *mdb_entry;
+	u16 fid_index;
+
+	bridge_port = mlxsw_sp_bridge_port_find(mlxsw_sp->bridge, orig_dev);
+	if (!bridge_port)
+		return 0;
+
+	bridge_device = bridge_port->bridge_device;
+	mlxsw_sp_port_vlan = mlxsw_sp_port_vlan_find_by_bridge(mlxsw_sp_port,
+							       bridge_device,
+							       mdb->vid);
+	if (!mlxsw_sp_port_vlan)
+		return 0;
+
+	fid_index = mlxsw_sp_fid_index(mlxsw_sp_port_vlan->fid);
+
+	mdb_entry = mlxsw_sp_mc_mdb_entry_get(mlxsw_sp, bridge_device,
+					      mdb->addr, fid_index,
+					      mlxsw_sp_port->local_port);
+	if (IS_ERR(mdb_entry))
+		return PTR_ERR(mdb_entry);
+
+	return 0;
+}
+
+static int
+mlxsw_sp_bridge_mdb_mc_enable_sync(struct mlxsw_sp *mlxsw_sp,
+				   struct mlxsw_sp_bridge_device *bridge_device,
+				   bool mc_enabled)
+{
+	struct mlxsw_sp_mdb_entry *mdb_entry;
+	int err;
+
+	list_for_each_entry(mdb_entry, &bridge_device->mdb_list, list) {
+		err = mlxsw_sp_mdb_entry_write(mlxsw_sp, mdb_entry, mc_enabled);
+		if (err)
+			goto err_mdb_entry_write;
+	}
+	return 0;
+
+err_mdb_entry_write:
+	list_for_each_entry_continue_reverse(mdb_entry,
+					     &bridge_device->mdb_list, list)
+		mlxsw_sp_mdb_entry_write(mlxsw_sp, mdb_entry, !mc_enabled);
+	return err;
+}
+
+static void
+mlxsw_sp_port_mrouter_update_mdb(struct mlxsw_sp_port *mlxsw_sp_port,
+				 struct mlxsw_sp_bridge_port *bridge_port,
+				 bool add)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	struct mlxsw_sp_bridge_device *bridge_device;
+	u16 local_port = mlxsw_sp_port->local_port;
+	struct mlxsw_sp_mdb_entry *mdb_entry;
+
+	bridge_device = bridge_port->bridge_device;
+
+	list_for_each_entry(mdb_entry, &bridge_device->mdb_list, list) {
+		if (add)
+			mlxsw_sp_mdb_entry_mrouter_port_get(mlxsw_sp, mdb_entry,
+							    local_port);
+		else
+			mlxsw_sp_mdb_entry_mrouter_port_put(mlxsw_sp, mdb_entry,
+							    local_port);
+	}
+}
+
+static int mlxsw_sp_port_obj_add(struct net_device *dev, const void *ctx,
+				 const struct switchdev_obj *obj,
+				 struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+	const struct switchdev_obj_port_vlan *vlan;
+	int err = 0;
+
+	switch (obj->id) {
+	case SWITCHDEV_OBJ_ID_PORT_VLAN:
+		vlan = SWITCHDEV_OBJ_PORT_VLAN(obj);
+
+		err = mlxsw_sp_port_vlans_add(mlxsw_sp_port, vlan, extack);
+
+		/* The event is emitted before the changes are actually
+		 * applied to the bridge. Therefore schedule the respin
+		 * call for later, so that the respin logic sees the
+		 * updated bridge state.
+		 */
+		mlxsw_sp_span_respin(mlxsw_sp_port->mlxsw_sp);
+		break;
+	case SWITCHDEV_OBJ_ID_PORT_MDB:
+		err = mlxsw_sp_port_mdb_add(mlxsw_sp_port,
+					    SWITCHDEV_OBJ_PORT_MDB(obj));
+		break;
+	default:
+		err = -EOPNOTSUPP;
+		break;
+	}
+
+	return err;
+}
+
+static void
+mlxsw_sp_bridge_port_vlan_del(struct mlxsw_sp_port *mlxsw_sp_port,
+			      struct mlxsw_sp_bridge_port *bridge_port, u16 vid)
+{
+	u16 pvid = mlxsw_sp_port->pvid == vid ? 0 : mlxsw_sp_port->pvid;
+	struct mlxsw_sp_port_vlan *mlxsw_sp_port_vlan;
+	u16 proto;
+
+	mlxsw_sp_port_vlan = mlxsw_sp_port_vlan_find_by_vid(mlxsw_sp_port, vid);
+	if (WARN_ON(!mlxsw_sp_port_vlan))
+		return;
+
+	mlxsw_sp_port_vlan_bridge_leave(mlxsw_sp_port_vlan);
+	br_vlan_get_proto(bridge_port->bridge_device->dev, &proto);
+	mlxsw_sp_port_pvid_set(mlxsw_sp_port, pvid, proto);
+	mlxsw_sp_port_vlan_set(mlxsw_sp_port, vid, vid, false, false);
+	mlxsw_sp_port_vlan_destroy(mlxsw_sp_port_vlan);
+}
+
+static int mlxsw_sp_port_vlans_del(struct mlxsw_sp_port *mlxsw_sp_port,
+				   const struct switchdev_obj_port_vlan *vlan)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	struct net_device *orig_dev = vlan->obj.orig_dev;
+	struct mlxsw_sp_bridge_port *bridge_port;
+
+	if (netif_is_bridge_master(orig_dev))
+		return -EOPNOTSUPP;
+
+	bridge_port = mlxsw_sp_bridge_port_find(mlxsw_sp->bridge, orig_dev);
+	if (WARN_ON(!bridge_port))
+		return -EINVAL;
+
+	if (!bridge_port->bridge_device->vlan_enabled)
+		return 0;
+
+	mlxsw_sp_bridge_port_vlan_del(mlxsw_sp_port, bridge_port, vlan->vid);
+
+	return 0;
+}
+
+static int mlxsw_sp_port_mdb_del(struct mlxsw_sp_port *mlxsw_sp_port,
+				 const struct switchdev_obj_port_mdb *mdb)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	struct net_device *orig_dev = mdb->obj.orig_dev;
+	struct mlxsw_sp_port_vlan *mlxsw_sp_port_vlan;
+	struct mlxsw_sp_bridge_device *bridge_device;
+	struct net_device *dev = mlxsw_sp_port->dev;
+	struct mlxsw_sp_bridge_port *bridge_port;
+	struct mlxsw_sp_mdb_entry_key key = {};
+	struct mlxsw_sp_mdb_entry *mdb_entry;
+	u16 fid_index;
+
+	bridge_port = mlxsw_sp_bridge_port_find(mlxsw_sp->bridge, orig_dev);
+	if (!bridge_port)
+		return 0;
+
+	bridge_device = bridge_port->bridge_device;
+	mlxsw_sp_port_vlan = mlxsw_sp_port_vlan_find_by_bridge(mlxsw_sp_port,
+							       bridge_device,
+							       mdb->vid);
+	if (!mlxsw_sp_port_vlan)
+		return 0;
+
+	fid_index = mlxsw_sp_fid_index(mlxsw_sp_port_vlan->fid);
+
+	ether_addr_copy(key.addr, mdb->addr);
+	key.fid = fid_index;
+	mdb_entry = rhashtable_lookup_fast(&bridge_device->mdb_ht, &key,
+					   mlxsw_sp_mdb_ht_params);
+	if (!mdb_entry) {
+		netdev_err(dev, "Unable to remove port from MC DB\n");
+		return -EINVAL;
+	}
+
+	mlxsw_sp_mc_mdb_entry_put(mlxsw_sp, bridge_device, mdb_entry,
+				  mlxsw_sp_port->local_port, false);
+	return 0;
+}
+
+static void
+mlxsw_sp_bridge_port_mdb_flush(struct mlxsw_sp_port *mlxsw_sp_port,
+			       struct mlxsw_sp_bridge_port *bridge_port,
+			       u16 fid_index)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	struct mlxsw_sp_bridge_device *bridge_device;
+	struct mlxsw_sp_mdb_entry *mdb_entry, *tmp;
+	u16 local_port = mlxsw_sp_port->local_port;
+
+	bridge_device = bridge_port->bridge_device;
+
+	list_for_each_entry_safe(mdb_entry, tmp, &bridge_device->mdb_list,
+				 list) {
+		if (mdb_entry->key.fid != fid_index)
+			continue;
+
+		if (bridge_port->mrouter)
+			mlxsw_sp_mdb_entry_mrouter_port_put(mlxsw_sp,
+							    mdb_entry,
+							    local_port);
+
+		mlxsw_sp_mc_mdb_entry_put(mlxsw_sp, bridge_device, mdb_entry,
+					  local_port, true);
+	}
+}
+
+static int mlxsw_sp_port_obj_del(struct net_device *dev, const void *ctx,
+				 const struct switchdev_obj *obj)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+	int err = 0;
+
+	switch (obj->id) {
+	case SWITCHDEV_OBJ_ID_PORT_VLAN:
+		err = mlxsw_sp_port_vlans_del(mlxsw_sp_port,
+					      SWITCHDEV_OBJ_PORT_VLAN(obj));
+		break;
+	case SWITCHDEV_OBJ_ID_PORT_MDB:
+		err = mlxsw_sp_port_mdb_del(mlxsw_sp_port,
+					    SWITCHDEV_OBJ_PORT_MDB(obj));
+		break;
+	default:
+		err = -EOPNOTSUPP;
+		break;
+	}
+
+	mlxsw_sp_span_respin(mlxsw_sp_port->mlxsw_sp);
+
+	return err;
+}
+
+static struct mlxsw_sp_port *mlxsw_sp_lag_rep_port(struct mlxsw_sp *mlxsw_sp,
+						   u16 lag_id)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port;
+	u64 max_lag_members;
+	int i;
+
+	max_lag_members = MLXSW_CORE_RES_GET(mlxsw_sp->core,
+					     MAX_LAG_MEMBERS);
+	for (i = 0; i < max_lag_members; i++) {
+		mlxsw_sp_port = mlxsw_sp_port_lagged_get(mlxsw_sp, lag_id, i);
+		if (mlxsw_sp_port)
+			return mlxsw_sp_port;
+	}
+	return NULL;
+}
+
+static int
+mlxsw_sp_bridge_vlan_aware_port_join(struct mlxsw_sp_bridge_port *bridge_port,
+				     struct mlxsw_sp_port *mlxsw_sp_port,
+				     struct netlink_ext_ack *extack)
+{
+	if (is_vlan_dev(bridge_port->dev)) {
+		NL_SET_ERR_MSG_MOD(extack, "Can not enslave a VLAN device to a VLAN-aware bridge");
+		return -EINVAL;
+	}
+
+	/* Port is no longer usable as a router interface */
+	if (mlxsw_sp_port->default_vlan->fid)
+		mlxsw_sp_port_vlan_router_leave(mlxsw_sp_port->default_vlan);
+
+	return 0;
+}
+
+static int
+mlxsw_sp_bridge_8021q_port_join(struct mlxsw_sp_bridge_device *bridge_device,
+				struct mlxsw_sp_bridge_port *bridge_port,
+				struct mlxsw_sp_port *mlxsw_sp_port,
+				struct netlink_ext_ack *extack)
+{
+	return mlxsw_sp_bridge_vlan_aware_port_join(bridge_port, mlxsw_sp_port,
+						    extack);
+}
+
+static void
+mlxsw_sp_bridge_vlan_aware_port_leave(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	/* Make sure untagged frames are allowed to ingress */
+	mlxsw_sp_port_pvid_set(mlxsw_sp_port, MLXSW_SP_DEFAULT_VID,
+			       ETH_P_8021Q);
+}
+
+static void
+mlxsw_sp_bridge_8021q_port_leave(struct mlxsw_sp_bridge_device *bridge_device,
+				 struct mlxsw_sp_bridge_port *bridge_port,
+				 struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	mlxsw_sp_bridge_vlan_aware_port_leave(mlxsw_sp_port);
+}
+
+static int
+mlxsw_sp_bridge_vlan_aware_vxlan_join(struct mlxsw_sp_bridge_device *bridge_device,
+				      const struct net_device *vxlan_dev,
+				      u16 vid, u16 ethertype,
+				      struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_lower_get(bridge_device->dev);
+	struct vxlan_dev *vxlan = netdev_priv(vxlan_dev);
+	struct mlxsw_sp_nve_params params = {
+		.type = MLXSW_SP_NVE_TYPE_VXLAN,
+		.vni = vxlan->cfg.vni,
+		.dev = vxlan_dev,
+		.ethertype = ethertype,
+	};
+	struct mlxsw_sp_fid *fid;
+	int err;
+
+	/* If the VLAN is 0, we need to find the VLAN that is configured as
+	 * PVID and egress untagged on the bridge port of the VxLAN device.
+	 * It is possible no such VLAN exists
+	 */
+	if (!vid) {
+		err = mlxsw_sp_vxlan_mapped_vid(vxlan_dev, &vid);
+		if (err || !vid)
+			return err;
+	}
+
+	fid = mlxsw_sp_fid_8021q_get(mlxsw_sp, vid);
+	if (IS_ERR(fid)) {
+		NL_SET_ERR_MSG_MOD(extack, "Failed to create 802.1Q FID");
+		return PTR_ERR(fid);
+	}
+
+	if (mlxsw_sp_fid_vni_is_set(fid)) {
+		NL_SET_ERR_MSG_MOD(extack, "VNI is already set on FID");
+		err = -EINVAL;
+		goto err_vni_exists;
+	}
+
+	err = mlxsw_sp_nve_fid_enable(mlxsw_sp, fid, &params, extack);
+	if (err)
+		goto err_nve_fid_enable;
+
+	return 0;
+
+err_nve_fid_enable:
+err_vni_exists:
+	mlxsw_sp_fid_put(fid);
+	return err;
+}
+
+static int
+mlxsw_sp_bridge_8021q_vxlan_join(struct mlxsw_sp_bridge_device *bridge_device,
+				 const struct net_device *vxlan_dev, u16 vid,
+				 struct netlink_ext_ack *extack)
+{
+	return mlxsw_sp_bridge_vlan_aware_vxlan_join(bridge_device, vxlan_dev,
+						     vid, ETH_P_8021Q, extack);
+}
+
+static struct net_device *
+mlxsw_sp_bridge_8021q_vxlan_dev_find(struct net_device *br_dev, u16 vid)
+{
+	struct net_device *dev;
+	struct list_head *iter;
+
+	netdev_for_each_lower_dev(br_dev, dev, iter) {
+		u16 pvid;
+		int err;
+
+		if (!netif_is_vxlan(dev))
+			continue;
+
+		err = mlxsw_sp_vxlan_mapped_vid(dev, &pvid);
+		if (err || pvid != vid)
+			continue;
+
+		return dev;
+	}
+
+	return NULL;
+}
+
+static struct mlxsw_sp_fid *
+mlxsw_sp_bridge_8021q_fid_get(struct mlxsw_sp_bridge_device *bridge_device,
+			      u16 vid, struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_lower_get(bridge_device->dev);
+
+	return mlxsw_sp_fid_8021q_get(mlxsw_sp, vid);
+}
+
+static struct mlxsw_sp_fid *
+mlxsw_sp_bridge_8021q_fid_lookup(struct mlxsw_sp_bridge_device *bridge_device,
+				 u16 vid)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_lower_get(bridge_device->dev);
+
+	return mlxsw_sp_fid_8021q_lookup(mlxsw_sp, vid);
+}
+
+static u16
+mlxsw_sp_bridge_8021q_fid_vid(struct mlxsw_sp_bridge_device *bridge_device,
+			      const struct mlxsw_sp_fid *fid)
+{
+	return mlxsw_sp_fid_8021q_vid(fid);
+}
+
+static const struct mlxsw_sp_bridge_ops mlxsw_sp_bridge_8021q_ops = {
+	.port_join	= mlxsw_sp_bridge_8021q_port_join,
+	.port_leave	= mlxsw_sp_bridge_8021q_port_leave,
+	.vxlan_join	= mlxsw_sp_bridge_8021q_vxlan_join,
+	.fid_get	= mlxsw_sp_bridge_8021q_fid_get,
+	.fid_lookup	= mlxsw_sp_bridge_8021q_fid_lookup,
+	.fid_vid	= mlxsw_sp_bridge_8021q_fid_vid,
+};
+
+static bool
+mlxsw_sp_port_is_br_member(const struct mlxsw_sp_port *mlxsw_sp_port,
+			   const struct net_device *br_dev)
+{
+	struct mlxsw_sp_port_vlan *mlxsw_sp_port_vlan;
+
+	list_for_each_entry(mlxsw_sp_port_vlan, &mlxsw_sp_port->vlans_list,
+			    list) {
+		if (mlxsw_sp_port_vlan->bridge_port &&
+		    mlxsw_sp_port_vlan->bridge_port->bridge_device->dev ==
+		    br_dev)
+			return true;
+	}
+
+	return false;
+}
+
+static int
+mlxsw_sp_bridge_8021d_port_join(struct mlxsw_sp_bridge_device *bridge_device,
+				struct mlxsw_sp_bridge_port *bridge_port,
+				struct mlxsw_sp_port *mlxsw_sp_port,
+				struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp_port_vlan *mlxsw_sp_port_vlan;
+	struct net_device *dev = bridge_port->dev;
+	u16 vid;
+
+	vid = is_vlan_dev(dev) ? vlan_dev_vlan_id(dev) : MLXSW_SP_DEFAULT_VID;
+	mlxsw_sp_port_vlan = mlxsw_sp_port_vlan_find_by_vid(mlxsw_sp_port, vid);
+	if (WARN_ON(!mlxsw_sp_port_vlan))
+		return -EINVAL;
+
+	if (mlxsw_sp_port_is_br_member(mlxsw_sp_port, bridge_device->dev)) {
+		NL_SET_ERR_MSG_MOD(extack, "Can not bridge VLAN uppers of the same port");
+		return -EINVAL;
+	}
+
+	/* Port is no longer usable as a router interface */
+	if (mlxsw_sp_port_vlan->fid)
+		mlxsw_sp_port_vlan_router_leave(mlxsw_sp_port_vlan);
+
+	return mlxsw_sp_port_vlan_bridge_join(mlxsw_sp_port_vlan, bridge_port,
+					      extack);
+}
+
+static void
+mlxsw_sp_bridge_8021d_port_leave(struct mlxsw_sp_bridge_device *bridge_device,
+				 struct mlxsw_sp_bridge_port *bridge_port,
+				 struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	struct mlxsw_sp_port_vlan *mlxsw_sp_port_vlan;
+	struct net_device *dev = bridge_port->dev;
+	u16 vid;
+
+	vid = is_vlan_dev(dev) ? vlan_dev_vlan_id(dev) : MLXSW_SP_DEFAULT_VID;
+	mlxsw_sp_port_vlan = mlxsw_sp_port_vlan_find_by_vid(mlxsw_sp_port, vid);
+	if (!mlxsw_sp_port_vlan || !mlxsw_sp_port_vlan->bridge_port)
+		return;
+
+	mlxsw_sp_port_vlan_bridge_leave(mlxsw_sp_port_vlan);
+}
+
+static int
+mlxsw_sp_bridge_8021d_vxlan_join(struct mlxsw_sp_bridge_device *bridge_device,
+				 const struct net_device *vxlan_dev, u16 vid,
+				 struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_lower_get(bridge_device->dev);
+	struct vxlan_dev *vxlan = netdev_priv(vxlan_dev);
+	struct mlxsw_sp_nve_params params = {
+		.type = MLXSW_SP_NVE_TYPE_VXLAN,
+		.vni = vxlan->cfg.vni,
+		.dev = vxlan_dev,
+		.ethertype = ETH_P_8021Q,
+	};
+	struct mlxsw_sp_fid *fid;
+	int err;
+
+	fid = mlxsw_sp_fid_8021d_get(mlxsw_sp, bridge_device->dev->ifindex);
+	if (IS_ERR(fid)) {
+		NL_SET_ERR_MSG_MOD(extack, "Failed to create 802.1D FID");
+		return -EINVAL;
+	}
+
+	if (mlxsw_sp_fid_vni_is_set(fid)) {
+		NL_SET_ERR_MSG_MOD(extack, "VNI is already set on FID");
+		err = -EINVAL;
+		goto err_vni_exists;
+	}
+
+	err = mlxsw_sp_nve_fid_enable(mlxsw_sp, fid, &params, extack);
+	if (err)
+		goto err_nve_fid_enable;
+
+	return 0;
+
+err_nve_fid_enable:
+err_vni_exists:
+	mlxsw_sp_fid_put(fid);
+	return err;
+}
+
+static struct mlxsw_sp_fid *
+mlxsw_sp_bridge_8021d_fid_get(struct mlxsw_sp_bridge_device *bridge_device,
+			      u16 vid, struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_lower_get(bridge_device->dev);
+
+	return mlxsw_sp_fid_8021d_get(mlxsw_sp, bridge_device->dev->ifindex);
+}
+
+static struct mlxsw_sp_fid *
+mlxsw_sp_bridge_8021d_fid_lookup(struct mlxsw_sp_bridge_device *bridge_device,
+				 u16 vid)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_lower_get(bridge_device->dev);
+
+	/* The only valid VLAN for a VLAN-unaware bridge is 0 */
+	if (vid)
+		return NULL;
+
+	return mlxsw_sp_fid_8021d_lookup(mlxsw_sp, bridge_device->dev->ifindex);
+}
+
+static u16
+mlxsw_sp_bridge_8021d_fid_vid(struct mlxsw_sp_bridge_device *bridge_device,
+			      const struct mlxsw_sp_fid *fid)
+{
+	return 0;
+}
+
+static const struct mlxsw_sp_bridge_ops mlxsw_sp_bridge_8021d_ops = {
+	.port_join	= mlxsw_sp_bridge_8021d_port_join,
+	.port_leave	= mlxsw_sp_bridge_8021d_port_leave,
+	.vxlan_join	= mlxsw_sp_bridge_8021d_vxlan_join,
+	.fid_get	= mlxsw_sp_bridge_8021d_fid_get,
+	.fid_lookup	= mlxsw_sp_bridge_8021d_fid_lookup,
+	.fid_vid	= mlxsw_sp_bridge_8021d_fid_vid,
+};
+
+static int
+mlxsw_sp_bridge_8021ad_port_join(struct mlxsw_sp_bridge_device *bridge_device,
+				 struct mlxsw_sp_bridge_port *bridge_port,
+				 struct mlxsw_sp_port *mlxsw_sp_port,
+				 struct netlink_ext_ack *extack)
+{
+	int err;
+
+	err = mlxsw_sp_port_vlan_classification_set(mlxsw_sp_port, true, false);
+	if (err)
+		return err;
+
+	err = mlxsw_sp_bridge_vlan_aware_port_join(bridge_port, mlxsw_sp_port,
+						   extack);
+	if (err)
+		goto err_bridge_vlan_aware_port_join;
+
+	return 0;
+
+err_bridge_vlan_aware_port_join:
+	mlxsw_sp_port_vlan_classification_set(mlxsw_sp_port, false, true);
+	return err;
+}
+
+static void
+mlxsw_sp_bridge_8021ad_port_leave(struct mlxsw_sp_bridge_device *bridge_device,
+				  struct mlxsw_sp_bridge_port *bridge_port,
+				  struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	mlxsw_sp_bridge_vlan_aware_port_leave(mlxsw_sp_port);
+	mlxsw_sp_port_vlan_classification_set(mlxsw_sp_port, false, true);
+}
+
+static int
+mlxsw_sp_bridge_8021ad_vxlan_join(struct mlxsw_sp_bridge_device *bridge_device,
+				  const struct net_device *vxlan_dev, u16 vid,
+				  struct netlink_ext_ack *extack)
+{
+	return mlxsw_sp_bridge_vlan_aware_vxlan_join(bridge_device, vxlan_dev,
+						     vid, ETH_P_8021AD, extack);
+}
+
+static const struct mlxsw_sp_bridge_ops mlxsw_sp1_bridge_8021ad_ops = {
+	.port_join	= mlxsw_sp_bridge_8021ad_port_join,
+	.port_leave	= mlxsw_sp_bridge_8021ad_port_leave,
+	.vxlan_join	= mlxsw_sp_bridge_8021ad_vxlan_join,
+	.fid_get	= mlxsw_sp_bridge_8021q_fid_get,
+	.fid_lookup	= mlxsw_sp_bridge_8021q_fid_lookup,
+	.fid_vid	= mlxsw_sp_bridge_8021q_fid_vid,
+};
+
+static int
+mlxsw_sp2_bridge_8021ad_port_join(struct mlxsw_sp_bridge_device *bridge_device,
+				  struct mlxsw_sp_bridge_port *bridge_port,
+				  struct mlxsw_sp_port *mlxsw_sp_port,
+				  struct netlink_ext_ack *extack)
+{
+	int err;
+
+	/* The EtherType of decapsulated packets is determined at the egress
+	 * port to allow 802.1d and 802.1ad bridges with VXLAN devices to
+	 * co-exist.
+	 */
+	err = mlxsw_sp_port_egress_ethtype_set(mlxsw_sp_port, ETH_P_8021AD);
+	if (err)
+		return err;
+
+	err = mlxsw_sp_bridge_8021ad_port_join(bridge_device, bridge_port,
+					       mlxsw_sp_port, extack);
+	if (err)
+		goto err_bridge_8021ad_port_join;
+
+	return 0;
+
+err_bridge_8021ad_port_join:
+	mlxsw_sp_port_egress_ethtype_set(mlxsw_sp_port, ETH_P_8021Q);
+	return err;
+}
+
+static void
+mlxsw_sp2_bridge_8021ad_port_leave(struct mlxsw_sp_bridge_device *bridge_device,
+				   struct mlxsw_sp_bridge_port *bridge_port,
+				   struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	mlxsw_sp_bridge_8021ad_port_leave(bridge_device, bridge_port,
+					  mlxsw_sp_port);
+	mlxsw_sp_port_egress_ethtype_set(mlxsw_sp_port, ETH_P_8021Q);
+}
+
+static const struct mlxsw_sp_bridge_ops mlxsw_sp2_bridge_8021ad_ops = {
+	.port_join	= mlxsw_sp2_bridge_8021ad_port_join,
+	.port_leave	= mlxsw_sp2_bridge_8021ad_port_leave,
+	.vxlan_join	= mlxsw_sp_bridge_8021ad_vxlan_join,
+	.fid_get	= mlxsw_sp_bridge_8021q_fid_get,
+	.fid_lookup	= mlxsw_sp_bridge_8021q_fid_lookup,
+	.fid_vid	= mlxsw_sp_bridge_8021q_fid_vid,
+};
+
+int mlxsw_sp_port_bridge_join(struct mlxsw_sp_port *mlxsw_sp_port,
+			      struct net_device *brport_dev,
+			      struct net_device *br_dev,
+			      struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	struct mlxsw_sp_bridge_device *bridge_device;
+	struct mlxsw_sp_bridge_port *bridge_port;
+	int err;
+
+	bridge_port = mlxsw_sp_bridge_port_get(mlxsw_sp->bridge, brport_dev,
+					       extack);
+	if (IS_ERR(bridge_port))
+		return PTR_ERR(bridge_port);
+	bridge_device = bridge_port->bridge_device;
+
+	err = bridge_device->ops->port_join(bridge_device, bridge_port,
+					    mlxsw_sp_port, extack);
+	if (err)
+		goto err_port_join;
+
+	return 0;
+
+err_port_join:
+	mlxsw_sp_bridge_port_put(mlxsw_sp->bridge, bridge_port);
+	return err;
+}
+
+void mlxsw_sp_port_bridge_leave(struct mlxsw_sp_port *mlxsw_sp_port,
+				struct net_device *brport_dev,
+				struct net_device *br_dev)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	struct mlxsw_sp_bridge_device *bridge_device;
+	struct mlxsw_sp_bridge_port *bridge_port;
+
+	bridge_device = mlxsw_sp_bridge_device_find(mlxsw_sp->bridge, br_dev);
+	if (!bridge_device)
+		return;
+	bridge_port = __mlxsw_sp_bridge_port_find(bridge_device, brport_dev);
+	if (!bridge_port)
+		return;
+
+	bridge_device->ops->port_leave(bridge_device, bridge_port,
+				       mlxsw_sp_port);
+	mlxsw_sp_bridge_port_put(mlxsw_sp->bridge, bridge_port);
+}
+
+int mlxsw_sp_bridge_vxlan_join(struct mlxsw_sp *mlxsw_sp,
+			       const struct net_device *br_dev,
+			       const struct net_device *vxlan_dev, u16 vid,
+			       struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp_bridge_device *bridge_device;
+
+	bridge_device = mlxsw_sp_bridge_device_find(mlxsw_sp->bridge, br_dev);
+	if (WARN_ON(!bridge_device))
+		return -EINVAL;
+
+	return bridge_device->ops->vxlan_join(bridge_device, vxlan_dev, vid,
+					      extack);
+}
+
+void mlxsw_sp_bridge_vxlan_leave(struct mlxsw_sp *mlxsw_sp,
+				 const struct net_device *vxlan_dev)
+{
+	struct vxlan_dev *vxlan = netdev_priv(vxlan_dev);
+	struct mlxsw_sp_fid *fid;
+
+	/* If the VxLAN device is down, then the FID does not have a VNI */
+	fid = mlxsw_sp_fid_lookup_by_vni(mlxsw_sp, vxlan->cfg.vni);
+	if (!fid)
+		return;
+
+	mlxsw_sp_nve_fid_disable(mlxsw_sp, fid);
+	/* Drop both the reference we just took during lookup and the reference
+	 * the VXLAN device took.
+	 */
+	mlxsw_sp_fid_put(fid);
+	mlxsw_sp_fid_put(fid);
+}
+
+static void
+mlxsw_sp_switchdev_vxlan_addr_convert(const union vxlan_addr *vxlan_addr,
+				      enum mlxsw_sp_l3proto *proto,
+				      union mlxsw_sp_l3addr *addr)
+{
+	if (vxlan_addr->sa.sa_family == AF_INET) {
+		addr->addr4 = vxlan_addr->sin.sin_addr.s_addr;
+		*proto = MLXSW_SP_L3_PROTO_IPV4;
+	} else {
+		addr->addr6 = vxlan_addr->sin6.sin6_addr;
+		*proto = MLXSW_SP_L3_PROTO_IPV6;
+	}
+}
+
+static void
+mlxsw_sp_switchdev_addr_vxlan_convert(enum mlxsw_sp_l3proto proto,
+				      const union mlxsw_sp_l3addr *addr,
+				      union vxlan_addr *vxlan_addr)
+{
+	switch (proto) {
+	case MLXSW_SP_L3_PROTO_IPV4:
+		vxlan_addr->sa.sa_family = AF_INET;
+		vxlan_addr->sin.sin_addr.s_addr = addr->addr4;
+		break;
+	case MLXSW_SP_L3_PROTO_IPV6:
+		vxlan_addr->sa.sa_family = AF_INET6;
+		vxlan_addr->sin6.sin6_addr = addr->addr6;
+		break;
+	}
+}
+
+static void mlxsw_sp_fdb_vxlan_call_notifiers(struct net_device *dev,
+					      const char *mac,
+					      enum mlxsw_sp_l3proto proto,
+					      union mlxsw_sp_l3addr *addr,
+					      __be32 vni, bool adding)
+{
+	struct switchdev_notifier_vxlan_fdb_info info;
+	struct vxlan_dev *vxlan = netdev_priv(dev);
+	enum switchdev_notifier_type type;
+
+	type = adding ? SWITCHDEV_VXLAN_FDB_ADD_TO_BRIDGE :
+			SWITCHDEV_VXLAN_FDB_DEL_TO_BRIDGE;
+	mlxsw_sp_switchdev_addr_vxlan_convert(proto, addr, &info.remote_ip);
+	info.remote_port = vxlan->cfg.dst_port;
+	info.remote_vni = vni;
+	info.remote_ifindex = 0;
+	ether_addr_copy(info.eth_addr, mac);
+	info.vni = vni;
+	info.offloaded = adding;
+	call_switchdev_notifiers(type, dev, &info.info, NULL);
+}
+
+static void mlxsw_sp_fdb_nve_call_notifiers(struct net_device *dev,
+					    const char *mac,
+					    enum mlxsw_sp_l3proto proto,
+					    union mlxsw_sp_l3addr *addr,
+					    __be32 vni,
+					    bool adding)
+{
+	if (netif_is_vxlan(dev))
+		mlxsw_sp_fdb_vxlan_call_notifiers(dev, mac, proto, addr, vni,
+						  adding);
+}
+
+static void
+mlxsw_sp_fdb_call_notifiers(enum switchdev_notifier_type type,
+			    const char *mac, u16 vid,
+			    struct net_device *dev, bool offloaded)
+{
+	struct switchdev_notifier_fdb_info info = {};
+
+	info.addr = mac;
+	info.vid = vid;
+	info.offloaded = offloaded;
+	call_switchdev_notifiers(type, dev, &info.info, NULL);
+}
+
+static void mlxsw_sp_fdb_notify_mac_process(struct mlxsw_sp *mlxsw_sp,
+					    char *sfn_pl, int rec_index,
+					    bool adding)
+{
+	struct mlxsw_sp_port_vlan *mlxsw_sp_port_vlan;
+	struct mlxsw_sp_bridge_device *bridge_device;
+	struct mlxsw_sp_bridge_port *bridge_port;
+	struct mlxsw_sp_port *mlxsw_sp_port;
+	u16 local_port, vid, fid, evid = 0;
+	enum switchdev_notifier_type type;
+	char mac[ETH_ALEN];
+	bool do_notification = true;
+	int err;
+
+	mlxsw_reg_sfn_mac_unpack(sfn_pl, rec_index, mac, &fid, &local_port);
+
+	if (WARN_ON_ONCE(!mlxsw_sp_local_port_is_valid(mlxsw_sp, local_port)))
+		return;
+	mlxsw_sp_port = mlxsw_sp->ports[local_port];
+	if (!mlxsw_sp_port) {
+		dev_err_ratelimited(mlxsw_sp->bus_info->dev, "Incorrect local port in FDB notification\n");
+		goto just_remove;
+	}
+
+	if (mlxsw_sp_fid_is_dummy(mlxsw_sp, fid))
+		goto just_remove;
+
+	mlxsw_sp_port_vlan = mlxsw_sp_port_vlan_find_by_fid(mlxsw_sp_port, fid);
+	if (!mlxsw_sp_port_vlan) {
+		netdev_err(mlxsw_sp_port->dev, "Failed to find a matching {Port, VID} following FDB notification\n");
+		goto just_remove;
+	}
+
+	bridge_port = mlxsw_sp_port_vlan->bridge_port;
+	if (!bridge_port) {
+		netdev_err(mlxsw_sp_port->dev, "{Port, VID} not associated with a bridge\n");
+		goto just_remove;
+	}
+
+	bridge_device = bridge_port->bridge_device;
+	vid = bridge_device->vlan_enabled ? mlxsw_sp_port_vlan->vid : 0;
+	evid = mlxsw_sp_port_vlan->vid;
+
+do_fdb_op:
+	err = mlxsw_sp_port_fdb_uc_op(mlxsw_sp, local_port, mac, fid, evid,
+				      adding, true);
+	if (err) {
+		dev_err_ratelimited(mlxsw_sp->bus_info->dev, "Failed to set FDB entry\n");
+		return;
+	}
+
+	if (!do_notification)
+		return;
+	type = adding ? SWITCHDEV_FDB_ADD_TO_BRIDGE : SWITCHDEV_FDB_DEL_TO_BRIDGE;
+	mlxsw_sp_fdb_call_notifiers(type, mac, vid, bridge_port->dev, adding);
+
+	return;
+
+just_remove:
+	adding = false;
+	do_notification = false;
+	goto do_fdb_op;
+}
+
+static void mlxsw_sp_fdb_notify_mac_lag_process(struct mlxsw_sp *mlxsw_sp,
+						char *sfn_pl, int rec_index,
+						bool adding)
+{
+	struct mlxsw_sp_port_vlan *mlxsw_sp_port_vlan;
+	struct mlxsw_sp_bridge_device *bridge_device;
+	struct mlxsw_sp_bridge_port *bridge_port;
+	struct mlxsw_sp_port *mlxsw_sp_port;
+	enum switchdev_notifier_type type;
+	char mac[ETH_ALEN];
+	u16 lag_vid = 0;
+	u16 lag_id;
+	u16 vid, fid;
+	bool do_notification = true;
+	int err;
+
+	mlxsw_reg_sfn_mac_lag_unpack(sfn_pl, rec_index, mac, &fid, &lag_id);
+	mlxsw_sp_port = mlxsw_sp_lag_rep_port(mlxsw_sp, lag_id);
+	if (!mlxsw_sp_port) {
+		dev_err_ratelimited(mlxsw_sp->bus_info->dev, "Cannot find port representor for LAG\n");
+		goto just_remove;
+	}
+
+	if (mlxsw_sp_fid_is_dummy(mlxsw_sp, fid))
+		goto just_remove;
+
+	mlxsw_sp_port_vlan = mlxsw_sp_port_vlan_find_by_fid(mlxsw_sp_port, fid);
+	if (!mlxsw_sp_port_vlan) {
+		netdev_err(mlxsw_sp_port->dev, "Failed to find a matching {Port, VID} following FDB notification\n");
+		goto just_remove;
+	}
+
+	bridge_port = mlxsw_sp_port_vlan->bridge_port;
+	if (!bridge_port) {
+		netdev_err(mlxsw_sp_port->dev, "{Port, VID} not associated with a bridge\n");
+		goto just_remove;
+	}
+
+	bridge_device = bridge_port->bridge_device;
+	vid = bridge_device->vlan_enabled ? mlxsw_sp_port_vlan->vid : 0;
+	lag_vid = mlxsw_sp_port_vlan->vid;
+
+do_fdb_op:
+	err = mlxsw_sp_port_fdb_uc_lag_op(mlxsw_sp, lag_id, mac, fid, lag_vid,
+					  adding, true);
+	if (err) {
+		dev_err_ratelimited(mlxsw_sp->bus_info->dev, "Failed to set FDB entry\n");
+		return;
+	}
+
+	if (!do_notification)
+		return;
+	type = adding ? SWITCHDEV_FDB_ADD_TO_BRIDGE : SWITCHDEV_FDB_DEL_TO_BRIDGE;
+	mlxsw_sp_fdb_call_notifiers(type, mac, vid, bridge_port->dev, adding);
+
+	return;
+
+just_remove:
+	adding = false;
+	do_notification = false;
+	goto do_fdb_op;
+}
+
+static int
+__mlxsw_sp_fdb_notify_mac_uc_tunnel_process(struct mlxsw_sp *mlxsw_sp,
+					    const struct mlxsw_sp_fid *fid,
+					    bool adding,
+					    struct net_device **nve_dev,
+					    u16 *p_vid, __be32 *p_vni)
+{
+	struct mlxsw_sp_bridge_device *bridge_device;
+	struct net_device *br_dev, *dev;
+	int nve_ifindex;
+	int err;
+
+	err = mlxsw_sp_fid_nve_ifindex(fid, &nve_ifindex);
+	if (err)
+		return err;
+
+	err = mlxsw_sp_fid_vni(fid, p_vni);
+	if (err)
+		return err;
+
+	dev = __dev_get_by_index(mlxsw_sp_net(mlxsw_sp), nve_ifindex);
+	if (!dev)
+		return -EINVAL;
+	*nve_dev = dev;
+
+	if (!netif_running(dev))
+		return -EINVAL;
+
+	if (adding && !br_port_flag_is_set(dev, BR_LEARNING))
+		return -EINVAL;
+
+	if (adding && netif_is_vxlan(dev)) {
+		struct vxlan_dev *vxlan = netdev_priv(dev);
+
+		if (!(vxlan->cfg.flags & VXLAN_F_LEARN))
+			return -EINVAL;
+	}
+
+	br_dev = netdev_master_upper_dev_get(dev);
+	if (!br_dev)
+		return -EINVAL;
+
+	bridge_device = mlxsw_sp_bridge_device_find(mlxsw_sp->bridge, br_dev);
+	if (!bridge_device)
+		return -EINVAL;
+
+	*p_vid = bridge_device->ops->fid_vid(bridge_device, fid);
+
+	return 0;
+}
+
+static void mlxsw_sp_fdb_notify_mac_uc_tunnel_process(struct mlxsw_sp *mlxsw_sp,
+						      char *sfn_pl,
+						      int rec_index,
+						      bool adding)
+{
+	enum mlxsw_reg_sfn_uc_tunnel_protocol sfn_proto;
+	enum switchdev_notifier_type type;
+	struct net_device *nve_dev;
+	union mlxsw_sp_l3addr addr;
+	struct mlxsw_sp_fid *fid;
+	char mac[ETH_ALEN];
+	u16 fid_index, vid;
+	__be32 vni;
+	u32 uip;
+	int err;
+
+	mlxsw_reg_sfn_uc_tunnel_unpack(sfn_pl, rec_index, mac, &fid_index,
+				       &uip, &sfn_proto);
+
+	fid = mlxsw_sp_fid_lookup_by_index(mlxsw_sp, fid_index);
+	if (!fid)
+		goto err_fid_lookup;
+
+	err = mlxsw_sp_nve_learned_ip_resolve(mlxsw_sp, uip,
+					      (enum mlxsw_sp_l3proto) sfn_proto,
+					      &addr);
+	if (err)
+		goto err_ip_resolve;
+
+	err = __mlxsw_sp_fdb_notify_mac_uc_tunnel_process(mlxsw_sp, fid, adding,
+							  &nve_dev, &vid, &vni);
+	if (err)
+		goto err_fdb_process;
+
+	err = mlxsw_sp_port_fdb_tunnel_uc_op(mlxsw_sp, mac, fid_index,
+					     (enum mlxsw_sp_l3proto) sfn_proto,
+					     &addr, adding, true);
+	if (err)
+		goto err_fdb_op;
+
+	mlxsw_sp_fdb_nve_call_notifiers(nve_dev, mac,
+					(enum mlxsw_sp_l3proto) sfn_proto,
+					&addr, vni, adding);
+
+	type = adding ? SWITCHDEV_FDB_ADD_TO_BRIDGE :
+			SWITCHDEV_FDB_DEL_TO_BRIDGE;
+	mlxsw_sp_fdb_call_notifiers(type, mac, vid, nve_dev, adding);
+
+	mlxsw_sp_fid_put(fid);
+
+	return;
+
+err_fdb_op:
+err_fdb_process:
+err_ip_resolve:
+	mlxsw_sp_fid_put(fid);
+err_fid_lookup:
+	/* Remove an FDB entry in case we cannot process it. Otherwise the
+	 * device will keep sending the same notification over and over again.
+	 */
+	mlxsw_sp_port_fdb_tunnel_uc_op(mlxsw_sp, mac, fid_index,
+				       (enum mlxsw_sp_l3proto) sfn_proto, &addr,
+				       false, true);
+}
+
+static void mlxsw_sp_fdb_notify_rec_process(struct mlxsw_sp *mlxsw_sp,
+					    char *sfn_pl, int rec_index)
+{
+	switch (mlxsw_reg_sfn_rec_type_get(sfn_pl, rec_index)) {
+	case MLXSW_REG_SFN_REC_TYPE_LEARNED_MAC:
+		mlxsw_sp_fdb_notify_mac_process(mlxsw_sp, sfn_pl,
+						rec_index, true);
+		break;
+	case MLXSW_REG_SFN_REC_TYPE_AGED_OUT_MAC:
+		mlxsw_sp_fdb_notify_mac_process(mlxsw_sp, sfn_pl,
+						rec_index, false);
+		break;
+	case MLXSW_REG_SFN_REC_TYPE_LEARNED_MAC_LAG:
+		mlxsw_sp_fdb_notify_mac_lag_process(mlxsw_sp, sfn_pl,
+						    rec_index, true);
+		break;
+	case MLXSW_REG_SFN_REC_TYPE_AGED_OUT_MAC_LAG:
+		mlxsw_sp_fdb_notify_mac_lag_process(mlxsw_sp, sfn_pl,
+						    rec_index, false);
+		break;
+	case MLXSW_REG_SFN_REC_TYPE_LEARNED_UNICAST_TUNNEL:
+		mlxsw_sp_fdb_notify_mac_uc_tunnel_process(mlxsw_sp, sfn_pl,
+							  rec_index, true);
+		break;
+	case MLXSW_REG_SFN_REC_TYPE_AGED_OUT_UNICAST_TUNNEL:
+		mlxsw_sp_fdb_notify_mac_uc_tunnel_process(mlxsw_sp, sfn_pl,
+							  rec_index, false);
+		break;
+	}
+}
+
+#define MLXSW_SP_FDB_SFN_QUERIES_PER_SESSION 10
+
+static void mlxsw_sp_fdb_notify_work(struct work_struct *work)
+{
+	struct mlxsw_sp_bridge *bridge;
+	struct mlxsw_sp *mlxsw_sp;
+	bool reschedule = false;
+	char *sfn_pl;
+	int queries;
+	u8 num_rec;
+	int i;
+	int err;
+
+	sfn_pl = kmalloc(MLXSW_REG_SFN_LEN, GFP_KERNEL);
+	if (!sfn_pl)
+		return;
+
+	bridge = container_of(work, struct mlxsw_sp_bridge, fdb_notify.dw.work);
+	mlxsw_sp = bridge->mlxsw_sp;
+
+	rtnl_lock();
+	if (list_empty(&bridge->bridges_list))
+		goto out;
+	reschedule = true;
+	queries = MLXSW_SP_FDB_SFN_QUERIES_PER_SESSION;
+	while (queries > 0) {
+		mlxsw_reg_sfn_pack(sfn_pl);
+		err = mlxsw_reg_query(mlxsw_sp->core, MLXSW_REG(sfn), sfn_pl);
+		if (err) {
+			dev_err_ratelimited(mlxsw_sp->bus_info->dev, "Failed to get FDB notifications\n");
+			goto out;
+		}
+		num_rec = mlxsw_reg_sfn_num_rec_get(sfn_pl);
+		for (i = 0; i < num_rec; i++)
+			mlxsw_sp_fdb_notify_rec_process(mlxsw_sp, sfn_pl, i);
+		if (num_rec != MLXSW_REG_SFN_REC_MAX_COUNT)
+			goto out;
+		queries--;
+	}
+
+out:
+	rtnl_unlock();
+	kfree(sfn_pl);
+	if (!reschedule)
+		return;
+	mlxsw_sp_fdb_notify_work_schedule(mlxsw_sp, !queries);
+}
+
+struct mlxsw_sp_switchdev_event_work {
+	struct work_struct work;
+	union {
+		struct switchdev_notifier_fdb_info fdb_info;
+		struct switchdev_notifier_vxlan_fdb_info vxlan_fdb_info;
+	};
+	struct net_device *dev;
+	unsigned long event;
+};
+
+static void
+mlxsw_sp_switchdev_bridge_vxlan_fdb_event(struct mlxsw_sp *mlxsw_sp,
+					  struct mlxsw_sp_switchdev_event_work *
+					  switchdev_work,
+					  struct mlxsw_sp_fid *fid, __be32 vni)
+{
+	struct switchdev_notifier_vxlan_fdb_info vxlan_fdb_info;
+	struct switchdev_notifier_fdb_info *fdb_info;
+	struct net_device *dev = switchdev_work->dev;
+	enum mlxsw_sp_l3proto proto;
+	union mlxsw_sp_l3addr addr;
+	int err;
+
+	fdb_info = &switchdev_work->fdb_info;
+	err = vxlan_fdb_find_uc(dev, fdb_info->addr, vni, &vxlan_fdb_info);
+	if (err)
+		return;
+
+	mlxsw_sp_switchdev_vxlan_addr_convert(&vxlan_fdb_info.remote_ip,
+					      &proto, &addr);
+
+	switch (switchdev_work->event) {
+	case SWITCHDEV_FDB_ADD_TO_DEVICE:
+		err = mlxsw_sp_port_fdb_tunnel_uc_op(mlxsw_sp,
+						     vxlan_fdb_info.eth_addr,
+						     mlxsw_sp_fid_index(fid),
+						     proto, &addr, true, false);
+		if (err)
+			return;
+		vxlan_fdb_info.offloaded = true;
+		call_switchdev_notifiers(SWITCHDEV_VXLAN_FDB_OFFLOADED, dev,
+					 &vxlan_fdb_info.info, NULL);
+		mlxsw_sp_fdb_call_notifiers(SWITCHDEV_FDB_OFFLOADED,
+					    vxlan_fdb_info.eth_addr,
+					    fdb_info->vid, dev, true);
+		break;
+	case SWITCHDEV_FDB_DEL_TO_DEVICE:
+		err = mlxsw_sp_port_fdb_tunnel_uc_op(mlxsw_sp,
+						     vxlan_fdb_info.eth_addr,
+						     mlxsw_sp_fid_index(fid),
+						     proto, &addr, false,
+						     false);
+		vxlan_fdb_info.offloaded = false;
+		call_switchdev_notifiers(SWITCHDEV_VXLAN_FDB_OFFLOADED, dev,
+					 &vxlan_fdb_info.info, NULL);
+		break;
+	}
+}
+
+static void
+mlxsw_sp_switchdev_bridge_nve_fdb_event(struct mlxsw_sp_switchdev_event_work *
+					switchdev_work)
+{
+	struct mlxsw_sp_bridge_device *bridge_device;
+	struct net_device *dev = switchdev_work->dev;
+	struct net_device *br_dev;
+	struct mlxsw_sp *mlxsw_sp;
+	struct mlxsw_sp_fid *fid;
+	__be32 vni;
+	int err;
+
+	if (switchdev_work->event != SWITCHDEV_FDB_ADD_TO_DEVICE &&
+	    switchdev_work->event != SWITCHDEV_FDB_DEL_TO_DEVICE)
+		return;
+
+	if (switchdev_work->event == SWITCHDEV_FDB_ADD_TO_DEVICE &&
+	    (!switchdev_work->fdb_info.added_by_user ||
+	     switchdev_work->fdb_info.is_local))
+		return;
+
+	if (!netif_running(dev))
+		return;
+	br_dev = netdev_master_upper_dev_get(dev);
+	if (!br_dev)
+		return;
+	if (!netif_is_bridge_master(br_dev))
+		return;
+	mlxsw_sp = mlxsw_sp_lower_get(br_dev);
+	if (!mlxsw_sp)
+		return;
+	bridge_device = mlxsw_sp_bridge_device_find(mlxsw_sp->bridge, br_dev);
+	if (!bridge_device)
+		return;
+
+	fid = bridge_device->ops->fid_lookup(bridge_device,
+					     switchdev_work->fdb_info.vid);
+	if (!fid)
+		return;
+
+	err = mlxsw_sp_fid_vni(fid, &vni);
+	if (err)
+		goto out;
+
+	mlxsw_sp_switchdev_bridge_vxlan_fdb_event(mlxsw_sp, switchdev_work, fid,
+						  vni);
+
+out:
+	mlxsw_sp_fid_put(fid);
+}
+
+static void mlxsw_sp_switchdev_bridge_fdb_event_work(struct work_struct *work)
+{
+	struct mlxsw_sp_switchdev_event_work *switchdev_work =
+		container_of(work, struct mlxsw_sp_switchdev_event_work, work);
+	struct net_device *dev = switchdev_work->dev;
+	struct switchdev_notifier_fdb_info *fdb_info;
+	struct mlxsw_sp_port *mlxsw_sp_port;
+	int err;
+
+	rtnl_lock();
+	if (netif_is_vxlan(dev)) {
+		mlxsw_sp_switchdev_bridge_nve_fdb_event(switchdev_work);
+		goto out;
+	}
+
+	mlxsw_sp_port = mlxsw_sp_port_dev_lower_find(dev);
+	if (!mlxsw_sp_port)
+		goto out;
+
+	switch (switchdev_work->event) {
+	case SWITCHDEV_FDB_ADD_TO_DEVICE:
+		fdb_info = &switchdev_work->fdb_info;
+		if (!fdb_info->added_by_user || fdb_info->is_local)
+			break;
+		err = mlxsw_sp_port_fdb_set(mlxsw_sp_port, fdb_info, true);
+		if (err)
+			break;
+		mlxsw_sp_fdb_call_notifiers(SWITCHDEV_FDB_OFFLOADED,
+					    fdb_info->addr,
+					    fdb_info->vid, dev, true);
+		break;
+	case SWITCHDEV_FDB_DEL_TO_DEVICE:
+		fdb_info = &switchdev_work->fdb_info;
+		mlxsw_sp_port_fdb_set(mlxsw_sp_port, fdb_info, false);
+		break;
+	case SWITCHDEV_FDB_ADD_TO_BRIDGE:
+	case SWITCHDEV_FDB_DEL_TO_BRIDGE:
+		/* These events are only used to potentially update an existing
+		 * SPAN mirror.
+		 */
+		break;
+	}
+
+	mlxsw_sp_span_respin(mlxsw_sp_port->mlxsw_sp);
+
+out:
+	rtnl_unlock();
+	kfree(switchdev_work->fdb_info.addr);
+	kfree(switchdev_work);
+	dev_put(dev);
+}
+
+static void
+mlxsw_sp_switchdev_vxlan_fdb_add(struct mlxsw_sp *mlxsw_sp,
+				 struct mlxsw_sp_switchdev_event_work *
+				 switchdev_work)
+{
+	struct switchdev_notifier_vxlan_fdb_info *vxlan_fdb_info;
+	struct mlxsw_sp_bridge_device *bridge_device;
+	struct net_device *dev = switchdev_work->dev;
+	u8 all_zeros_mac[ETH_ALEN] = { 0 };
+	enum mlxsw_sp_l3proto proto;
+	union mlxsw_sp_l3addr addr;
+	struct net_device *br_dev;
+	struct mlxsw_sp_fid *fid;
+	u16 vid;
+	int err;
+
+	vxlan_fdb_info = &switchdev_work->vxlan_fdb_info;
+	br_dev = netdev_master_upper_dev_get(dev);
+
+	bridge_device = mlxsw_sp_bridge_device_find(mlxsw_sp->bridge, br_dev);
+	if (!bridge_device)
+		return;
+
+	fid = mlxsw_sp_fid_lookup_by_vni(mlxsw_sp, vxlan_fdb_info->vni);
+	if (!fid)
+		return;
+
+	mlxsw_sp_switchdev_vxlan_addr_convert(&vxlan_fdb_info->remote_ip,
+					      &proto, &addr);
+
+	if (ether_addr_equal(vxlan_fdb_info->eth_addr, all_zeros_mac)) {
+		err = mlxsw_sp_nve_flood_ip_add(mlxsw_sp, fid, proto, &addr);
+		if (err) {
+			mlxsw_sp_fid_put(fid);
+			return;
+		}
+		vxlan_fdb_info->offloaded = true;
+		call_switchdev_notifiers(SWITCHDEV_VXLAN_FDB_OFFLOADED, dev,
+					 &vxlan_fdb_info->info, NULL);
+		mlxsw_sp_fid_put(fid);
+		return;
+	}
+
+	/* The device has a single FDB table, whereas Linux has two - one
+	 * in the bridge driver and another in the VxLAN driver. We only
+	 * program an entry to the device if the MAC points to the VxLAN
+	 * device in the bridge's FDB table
+	 */
+	vid = bridge_device->ops->fid_vid(bridge_device, fid);
+	if (br_fdb_find_port(br_dev, vxlan_fdb_info->eth_addr, vid) != dev)
+		goto err_br_fdb_find;
+
+	err = mlxsw_sp_port_fdb_tunnel_uc_op(mlxsw_sp, vxlan_fdb_info->eth_addr,
+					     mlxsw_sp_fid_index(fid), proto,
+					     &addr, true, false);
+	if (err)
+		goto err_fdb_tunnel_uc_op;
+	vxlan_fdb_info->offloaded = true;
+	call_switchdev_notifiers(SWITCHDEV_VXLAN_FDB_OFFLOADED, dev,
+				 &vxlan_fdb_info->info, NULL);
+	mlxsw_sp_fdb_call_notifiers(SWITCHDEV_FDB_OFFLOADED,
+				    vxlan_fdb_info->eth_addr, vid, dev, true);
+
+	mlxsw_sp_fid_put(fid);
+
+	return;
+
+err_fdb_tunnel_uc_op:
+err_br_fdb_find:
+	mlxsw_sp_fid_put(fid);
+}
+
+static void
+mlxsw_sp_switchdev_vxlan_fdb_del(struct mlxsw_sp *mlxsw_sp,
+				 struct mlxsw_sp_switchdev_event_work *
+				 switchdev_work)
+{
+	struct switchdev_notifier_vxlan_fdb_info *vxlan_fdb_info;
+	struct mlxsw_sp_bridge_device *bridge_device;
+	struct net_device *dev = switchdev_work->dev;
+	struct net_device *br_dev = netdev_master_upper_dev_get(dev);
+	u8 all_zeros_mac[ETH_ALEN] = { 0 };
+	enum mlxsw_sp_l3proto proto;
+	union mlxsw_sp_l3addr addr;
+	struct mlxsw_sp_fid *fid;
+	u16 vid;
+
+	vxlan_fdb_info = &switchdev_work->vxlan_fdb_info;
+	if (!vxlan_fdb_info->offloaded)
+		return;
+
+	bridge_device = mlxsw_sp_bridge_device_find(mlxsw_sp->bridge, br_dev);
+	if (!bridge_device)
+		return;
+
+	fid = mlxsw_sp_fid_lookup_by_vni(mlxsw_sp, vxlan_fdb_info->vni);
+	if (!fid)
+		return;
+
+	mlxsw_sp_switchdev_vxlan_addr_convert(&vxlan_fdb_info->remote_ip,
+					      &proto, &addr);
+
+	if (ether_addr_equal(vxlan_fdb_info->eth_addr, all_zeros_mac)) {
+		mlxsw_sp_nve_flood_ip_del(mlxsw_sp, fid, proto, &addr);
+		mlxsw_sp_fid_put(fid);
+		return;
+	}
+
+	mlxsw_sp_port_fdb_tunnel_uc_op(mlxsw_sp, vxlan_fdb_info->eth_addr,
+				       mlxsw_sp_fid_index(fid), proto, &addr,
+				       false, false);
+	vid = bridge_device->ops->fid_vid(bridge_device, fid);
+	mlxsw_sp_fdb_call_notifiers(SWITCHDEV_FDB_OFFLOADED,
+				    vxlan_fdb_info->eth_addr, vid, dev, false);
+
+	mlxsw_sp_fid_put(fid);
+}
+
+static void mlxsw_sp_switchdev_vxlan_fdb_event_work(struct work_struct *work)
+{
+	struct mlxsw_sp_switchdev_event_work *switchdev_work =
+		container_of(work, struct mlxsw_sp_switchdev_event_work, work);
+	struct net_device *dev = switchdev_work->dev;
+	struct mlxsw_sp *mlxsw_sp;
+	struct net_device *br_dev;
+
+	rtnl_lock();
+
+	if (!netif_running(dev))
+		goto out;
+	br_dev = netdev_master_upper_dev_get(dev);
+	if (!br_dev)
+		goto out;
+	if (!netif_is_bridge_master(br_dev))
+		goto out;
+	mlxsw_sp = mlxsw_sp_lower_get(br_dev);
+	if (!mlxsw_sp)
+		goto out;
+
+	switch (switchdev_work->event) {
+	case SWITCHDEV_VXLAN_FDB_ADD_TO_DEVICE:
+		mlxsw_sp_switchdev_vxlan_fdb_add(mlxsw_sp, switchdev_work);
+		break;
+	case SWITCHDEV_VXLAN_FDB_DEL_TO_DEVICE:
+		mlxsw_sp_switchdev_vxlan_fdb_del(mlxsw_sp, switchdev_work);
+		break;
+	}
+
+out:
+	rtnl_unlock();
+	kfree(switchdev_work);
+	dev_put(dev);
+}
+
+static int
+mlxsw_sp_switchdev_vxlan_work_prepare(struct mlxsw_sp_switchdev_event_work *
+				      switchdev_work,
+				      struct switchdev_notifier_info *info)
+{
+	struct vxlan_dev *vxlan = netdev_priv(switchdev_work->dev);
+	struct switchdev_notifier_vxlan_fdb_info *vxlan_fdb_info;
+	struct vxlan_config *cfg = &vxlan->cfg;
+	struct netlink_ext_ack *extack;
+
+	extack = switchdev_notifier_info_to_extack(info);
+	vxlan_fdb_info = container_of(info,
+				      struct switchdev_notifier_vxlan_fdb_info,
+				      info);
+
+	if (vxlan_fdb_info->remote_port != cfg->dst_port) {
+		NL_SET_ERR_MSG_MOD(extack, "VxLAN: FDB: Non-default remote port is not supported");
+		return -EOPNOTSUPP;
+	}
+	if (vxlan_fdb_info->remote_vni != cfg->vni ||
+	    vxlan_fdb_info->vni != cfg->vni) {
+		NL_SET_ERR_MSG_MOD(extack, "VxLAN: FDB: Non-default VNI is not supported");
+		return -EOPNOTSUPP;
+	}
+	if (vxlan_fdb_info->remote_ifindex) {
+		NL_SET_ERR_MSG_MOD(extack, "VxLAN: FDB: Local interface is not supported");
+		return -EOPNOTSUPP;
+	}
+	if (is_multicast_ether_addr(vxlan_fdb_info->eth_addr)) {
+		NL_SET_ERR_MSG_MOD(extack, "VxLAN: FDB: Multicast MAC addresses not supported");
+		return -EOPNOTSUPP;
+	}
+	if (vxlan_addr_multicast(&vxlan_fdb_info->remote_ip)) {
+		NL_SET_ERR_MSG_MOD(extack, "VxLAN: FDB: Multicast destination IP is not supported");
+		return -EOPNOTSUPP;
+	}
+
+	switchdev_work->vxlan_fdb_info = *vxlan_fdb_info;
+
+	return 0;
+}
+
+/* Called under rcu_read_lock() */
+static int mlxsw_sp_switchdev_event(struct notifier_block *unused,
+				    unsigned long event, void *ptr)
+{
+	struct net_device *dev = switchdev_notifier_info_to_dev(ptr);
+	struct mlxsw_sp_switchdev_event_work *switchdev_work;
+	struct switchdev_notifier_fdb_info *fdb_info;
+	struct switchdev_notifier_info *info = ptr;
+	struct net_device *br_dev;
+	int err;
+
+	if (event == SWITCHDEV_PORT_ATTR_SET) {
+		err = switchdev_handle_port_attr_set(dev, ptr,
+						     mlxsw_sp_port_dev_check,
+						     mlxsw_sp_port_attr_set);
+		return notifier_from_errno(err);
+	}
+
+	/* Tunnel devices are not our uppers, so check their master instead */
+	br_dev = netdev_master_upper_dev_get_rcu(dev);
+	if (!br_dev)
+		return NOTIFY_DONE;
+	if (!netif_is_bridge_master(br_dev))
+		return NOTIFY_DONE;
+	if (!mlxsw_sp_port_dev_lower_find_rcu(br_dev))
+		return NOTIFY_DONE;
+
+	switchdev_work = kzalloc(sizeof(*switchdev_work), GFP_ATOMIC);
+	if (!switchdev_work)
+		return NOTIFY_BAD;
+
+	switchdev_work->dev = dev;
+	switchdev_work->event = event;
+
+	switch (event) {
+	case SWITCHDEV_FDB_ADD_TO_DEVICE:
+	case SWITCHDEV_FDB_DEL_TO_DEVICE:
+	case SWITCHDEV_FDB_ADD_TO_BRIDGE:
+	case SWITCHDEV_FDB_DEL_TO_BRIDGE:
+		fdb_info = container_of(info,
+					struct switchdev_notifier_fdb_info,
+					info);
+		INIT_WORK(&switchdev_work->work,
+			  mlxsw_sp_switchdev_bridge_fdb_event_work);
+		memcpy(&switchdev_work->fdb_info, ptr,
+		       sizeof(switchdev_work->fdb_info));
+		switchdev_work->fdb_info.addr = kzalloc(ETH_ALEN, GFP_ATOMIC);
+		if (!switchdev_work->fdb_info.addr)
+			goto err_addr_alloc;
+		ether_addr_copy((u8 *)switchdev_work->fdb_info.addr,
+				fdb_info->addr);
+		/* Take a reference on the device. This can be either
+		 * upper device containig mlxsw_sp_port or just a
+		 * mlxsw_sp_port
+		 */
+		dev_hold(dev);
+		break;
+	case SWITCHDEV_VXLAN_FDB_ADD_TO_DEVICE:
+	case SWITCHDEV_VXLAN_FDB_DEL_TO_DEVICE:
+		INIT_WORK(&switchdev_work->work,
+			  mlxsw_sp_switchdev_vxlan_fdb_event_work);
+		err = mlxsw_sp_switchdev_vxlan_work_prepare(switchdev_work,
+							    info);
+		if (err)
+			goto err_vxlan_work_prepare;
+		dev_hold(dev);
+		break;
+	default:
+		kfree(switchdev_work);
+		return NOTIFY_DONE;
+	}
+
+	mlxsw_core_schedule_work(&switchdev_work->work);
+
+	return NOTIFY_DONE;
+
+err_vxlan_work_prepare:
+err_addr_alloc:
+	kfree(switchdev_work);
+	return NOTIFY_BAD;
+}
+
+struct notifier_block mlxsw_sp_switchdev_notifier = {
+	.notifier_call = mlxsw_sp_switchdev_event,
+};
+
+static int
+mlxsw_sp_switchdev_vxlan_vlan_add(struct mlxsw_sp *mlxsw_sp,
+				  struct mlxsw_sp_bridge_device *bridge_device,
+				  const struct net_device *vxlan_dev, u16 vid,
+				  bool flag_untagged, bool flag_pvid,
+				  struct netlink_ext_ack *extack)
+{
+	struct vxlan_dev *vxlan = netdev_priv(vxlan_dev);
+	__be32 vni = vxlan->cfg.vni;
+	struct mlxsw_sp_fid *fid;
+	u16 old_vid;
+	int err;
+
+	/* We cannot have the same VLAN as PVID and egress untagged on multiple
+	 * VxLAN devices. Note that we get this notification before the VLAN is
+	 * actually added to the bridge's database, so it is not possible for
+	 * the lookup function to return 'vxlan_dev'
+	 */
+	if (flag_untagged && flag_pvid &&
+	    mlxsw_sp_bridge_8021q_vxlan_dev_find(bridge_device->dev, vid)) {
+		NL_SET_ERR_MSG_MOD(extack, "VLAN already mapped to a different VNI");
+		return -EINVAL;
+	}
+
+	if (!netif_running(vxlan_dev))
+		return 0;
+
+	/* First case: FID is not associated with this VNI, but the new VLAN
+	 * is both PVID and egress untagged. Need to enable NVE on the FID, if
+	 * it exists
+	 */
+	fid = mlxsw_sp_fid_lookup_by_vni(mlxsw_sp, vni);
+	if (!fid) {
+		if (!flag_untagged || !flag_pvid)
+			return 0;
+		return bridge_device->ops->vxlan_join(bridge_device, vxlan_dev,
+						      vid, extack);
+	}
+
+	/* Second case: FID is associated with the VNI and the VLAN associated
+	 * with the FID is the same as the notified VLAN. This means the flags
+	 * (PVID / egress untagged) were toggled and that NVE should be
+	 * disabled on the FID
+	 */
+	old_vid = mlxsw_sp_fid_8021q_vid(fid);
+	if (vid == old_vid) {
+		if (WARN_ON(flag_untagged && flag_pvid)) {
+			mlxsw_sp_fid_put(fid);
+			return -EINVAL;
+		}
+		mlxsw_sp_bridge_vxlan_leave(mlxsw_sp, vxlan_dev);
+		mlxsw_sp_fid_put(fid);
+		return 0;
+	}
+
+	/* Third case: A new VLAN was configured on the VxLAN device, but this
+	 * VLAN is not PVID, so there is nothing to do.
+	 */
+	if (!flag_pvid) {
+		mlxsw_sp_fid_put(fid);
+		return 0;
+	}
+
+	/* Fourth case: Thew new VLAN is PVID, which means the VLAN currently
+	 * mapped to the VNI should be unmapped
+	 */
+	mlxsw_sp_bridge_vxlan_leave(mlxsw_sp, vxlan_dev);
+	mlxsw_sp_fid_put(fid);
+
+	/* Fifth case: The new VLAN is also egress untagged, which means the
+	 * VLAN needs to be mapped to the VNI
+	 */
+	if (!flag_untagged)
+		return 0;
+
+	err = bridge_device->ops->vxlan_join(bridge_device, vxlan_dev, vid, extack);
+	if (err)
+		goto err_vxlan_join;
+
+	return 0;
+
+err_vxlan_join:
+	bridge_device->ops->vxlan_join(bridge_device, vxlan_dev, old_vid, NULL);
+	return err;
+}
+
+static void
+mlxsw_sp_switchdev_vxlan_vlan_del(struct mlxsw_sp *mlxsw_sp,
+				  struct mlxsw_sp_bridge_device *bridge_device,
+				  const struct net_device *vxlan_dev, u16 vid)
+{
+	struct vxlan_dev *vxlan = netdev_priv(vxlan_dev);
+	__be32 vni = vxlan->cfg.vni;
+	struct mlxsw_sp_fid *fid;
+
+	if (!netif_running(vxlan_dev))
+		return;
+
+	fid = mlxsw_sp_fid_lookup_by_vni(mlxsw_sp, vni);
+	if (!fid)
+		return;
+
+	/* A different VLAN than the one mapped to the VNI is deleted */
+	if (mlxsw_sp_fid_8021q_vid(fid) != vid)
+		goto out;
+
+	mlxsw_sp_bridge_vxlan_leave(mlxsw_sp, vxlan_dev);
+
+out:
+	mlxsw_sp_fid_put(fid);
+}
+
+static int
+mlxsw_sp_switchdev_vxlan_vlans_add(struct net_device *vxlan_dev,
+				   struct switchdev_notifier_port_obj_info *
+				   port_obj_info)
+{
+	struct switchdev_obj_port_vlan *vlan =
+		SWITCHDEV_OBJ_PORT_VLAN(port_obj_info->obj);
+	bool flag_untagged = vlan->flags & BRIDGE_VLAN_INFO_UNTAGGED;
+	bool flag_pvid = vlan->flags & BRIDGE_VLAN_INFO_PVID;
+	struct mlxsw_sp_bridge_device *bridge_device;
+	struct netlink_ext_ack *extack;
+	struct mlxsw_sp *mlxsw_sp;
+	struct net_device *br_dev;
+
+	extack = switchdev_notifier_info_to_extack(&port_obj_info->info);
+	br_dev = netdev_master_upper_dev_get(vxlan_dev);
+	if (!br_dev)
+		return 0;
+
+	mlxsw_sp = mlxsw_sp_lower_get(br_dev);
+	if (!mlxsw_sp)
+		return 0;
+
+	port_obj_info->handled = true;
+
+	bridge_device = mlxsw_sp_bridge_device_find(mlxsw_sp->bridge, br_dev);
+	if (!bridge_device)
+		return -EINVAL;
+
+	if (!bridge_device->vlan_enabled)
+		return 0;
+
+	return mlxsw_sp_switchdev_vxlan_vlan_add(mlxsw_sp, bridge_device,
+						 vxlan_dev, vlan->vid,
+						 flag_untagged,
+						 flag_pvid, extack);
+}
+
+static void
+mlxsw_sp_switchdev_vxlan_vlans_del(struct net_device *vxlan_dev,
+				   struct switchdev_notifier_port_obj_info *
+				   port_obj_info)
+{
+	struct switchdev_obj_port_vlan *vlan =
+		SWITCHDEV_OBJ_PORT_VLAN(port_obj_info->obj);
+	struct mlxsw_sp_bridge_device *bridge_device;
+	struct mlxsw_sp *mlxsw_sp;
+	struct net_device *br_dev;
+
+	br_dev = netdev_master_upper_dev_get(vxlan_dev);
+	if (!br_dev)
+		return;
+
+	mlxsw_sp = mlxsw_sp_lower_get(br_dev);
+	if (!mlxsw_sp)
+		return;
+
+	port_obj_info->handled = true;
+
+	bridge_device = mlxsw_sp_bridge_device_find(mlxsw_sp->bridge, br_dev);
+	if (!bridge_device)
+		return;
+
+	if (!bridge_device->vlan_enabled)
+		return;
+
+	mlxsw_sp_switchdev_vxlan_vlan_del(mlxsw_sp, bridge_device, vxlan_dev,
+					  vlan->vid);
+}
+
+static int
+mlxsw_sp_switchdev_handle_vxlan_obj_add(struct net_device *vxlan_dev,
+					struct switchdev_notifier_port_obj_info *
+					port_obj_info)
+{
+	int err = 0;
+
+	switch (port_obj_info->obj->id) {
+	case SWITCHDEV_OBJ_ID_PORT_VLAN:
+		err = mlxsw_sp_switchdev_vxlan_vlans_add(vxlan_dev,
+							 port_obj_info);
+		break;
+	default:
+		break;
+	}
+
+	return err;
+}
+
+static void
+mlxsw_sp_switchdev_handle_vxlan_obj_del(struct net_device *vxlan_dev,
+					struct switchdev_notifier_port_obj_info *
+					port_obj_info)
+{
+	switch (port_obj_info->obj->id) {
+	case SWITCHDEV_OBJ_ID_PORT_VLAN:
+		mlxsw_sp_switchdev_vxlan_vlans_del(vxlan_dev, port_obj_info);
+		break;
+	default:
+		break;
+	}
+}
+
+static int mlxsw_sp_switchdev_blocking_event(struct notifier_block *unused,
+					     unsigned long event, void *ptr)
+{
+	struct net_device *dev = switchdev_notifier_info_to_dev(ptr);
+	int err = 0;
+
+	switch (event) {
+	case SWITCHDEV_PORT_OBJ_ADD:
+		if (netif_is_vxlan(dev))
+			err = mlxsw_sp_switchdev_handle_vxlan_obj_add(dev, ptr);
+		else
+			err = switchdev_handle_port_obj_add(dev, ptr,
+							mlxsw_sp_port_dev_check,
+							mlxsw_sp_port_obj_add);
+		return notifier_from_errno(err);
+	case SWITCHDEV_PORT_OBJ_DEL:
+		if (netif_is_vxlan(dev))
+			mlxsw_sp_switchdev_handle_vxlan_obj_del(dev, ptr);
+		else
+			err = switchdev_handle_port_obj_del(dev, ptr,
+							mlxsw_sp_port_dev_check,
+							mlxsw_sp_port_obj_del);
+		return notifier_from_errno(err);
+	case SWITCHDEV_PORT_ATTR_SET:
+		err = switchdev_handle_port_attr_set(dev, ptr,
+						     mlxsw_sp_port_dev_check,
+						     mlxsw_sp_port_attr_set);
+		return notifier_from_errno(err);
+	}
+
+	return NOTIFY_DONE;
+}
+
+static struct notifier_block mlxsw_sp_switchdev_blocking_notifier = {
+	.notifier_call = mlxsw_sp_switchdev_blocking_event,
+};
+
+u8
+mlxsw_sp_bridge_port_stp_state(struct mlxsw_sp_bridge_port *bridge_port)
+{
+	return bridge_port->stp_state;
+}
+
+static int mlxsw_sp_fdb_init(struct mlxsw_sp *mlxsw_sp)
+{
+	struct mlxsw_sp_bridge *bridge = mlxsw_sp->bridge;
+	struct notifier_block *nb;
+	int err;
+
+	err = mlxsw_sp_ageing_set(mlxsw_sp, MLXSW_SP_DEFAULT_AGEING_TIME);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Failed to set default ageing time\n");
+		return err;
+	}
+
+	err = register_switchdev_notifier(&mlxsw_sp_switchdev_notifier);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Failed to register switchdev notifier\n");
+		return err;
+	}
+
+	nb = &mlxsw_sp_switchdev_blocking_notifier;
+	err = register_switchdev_blocking_notifier(nb);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Failed to register switchdev blocking notifier\n");
+		goto err_register_switchdev_blocking_notifier;
+	}
+
+	INIT_DELAYED_WORK(&bridge->fdb_notify.dw, mlxsw_sp_fdb_notify_work);
+	bridge->fdb_notify.interval = MLXSW_SP_DEFAULT_LEARNING_INTERVAL;
+	return 0;
+
+err_register_switchdev_blocking_notifier:
+	unregister_switchdev_notifier(&mlxsw_sp_switchdev_notifier);
+	return err;
+}
+
+static void mlxsw_sp_fdb_fini(struct mlxsw_sp *mlxsw_sp)
+{
+	struct notifier_block *nb;
+
+	cancel_delayed_work_sync(&mlxsw_sp->bridge->fdb_notify.dw);
+
+	nb = &mlxsw_sp_switchdev_blocking_notifier;
+	unregister_switchdev_blocking_notifier(nb);
+
+	unregister_switchdev_notifier(&mlxsw_sp_switchdev_notifier);
+}
+
+static void mlxsw_sp1_switchdev_init(struct mlxsw_sp *mlxsw_sp)
+{
+	mlxsw_sp->bridge->bridge_8021ad_ops = &mlxsw_sp1_bridge_8021ad_ops;
+}
+
+const struct mlxsw_sp_switchdev_ops mlxsw_sp1_switchdev_ops = {
+	.init	= mlxsw_sp1_switchdev_init,
+};
+
+static void mlxsw_sp2_switchdev_init(struct mlxsw_sp *mlxsw_sp)
+{
+	mlxsw_sp->bridge->bridge_8021ad_ops = &mlxsw_sp2_bridge_8021ad_ops;
+}
+
+const struct mlxsw_sp_switchdev_ops mlxsw_sp2_switchdev_ops = {
+	.init	= mlxsw_sp2_switchdev_init,
+};
+
+int mlxsw_sp_switchdev_init(struct mlxsw_sp *mlxsw_sp)
+{
+	struct mlxsw_sp_bridge *bridge;
+
+	bridge = kzalloc(sizeof(*mlxsw_sp->bridge), GFP_KERNEL);
+	if (!bridge)
+		return -ENOMEM;
+	mlxsw_sp->bridge = bridge;
+	bridge->mlxsw_sp = mlxsw_sp;
+
+	INIT_LIST_HEAD(&mlxsw_sp->bridge->bridges_list);
+
+	bridge->bridge_8021q_ops = &mlxsw_sp_bridge_8021q_ops;
+	bridge->bridge_8021d_ops = &mlxsw_sp_bridge_8021d_ops;
+
+	mlxsw_sp->switchdev_ops->init(mlxsw_sp);
+
+	return mlxsw_sp_fdb_init(mlxsw_sp);
+}
+
+void mlxsw_sp_switchdev_fini(struct mlxsw_sp *mlxsw_sp)
+{
+	mlxsw_sp_fdb_fini(mlxsw_sp);
+	WARN_ON(!list_empty(&mlxsw_sp->bridge->bridges_list));
+	kfree(mlxsw_sp->bridge);
+}
+
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_switchdev.h b/drivers/net/ethernet/mellanox/mlxsw/spectrum_switchdev.h
new file mode 100644
index 000000000..c218e10bd
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_switchdev.h
@@ -0,0 +1,13 @@
+/* SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0 */
+/* Copyright (c) 2018 Mellanox Technologies. All rights reserved */
+
+#include <linux/netdevice.h>
+
+struct mlxsw_sp_bridge;
+struct mlxsw_sp_bridge_port;
+
+struct mlxsw_sp_bridge_port *
+mlxsw_sp_bridge_port_find(struct mlxsw_sp_bridge *bridge,
+			  struct net_device *brport_dev);
+
+u8 mlxsw_sp_bridge_port_stp_state(struct mlxsw_sp_bridge_port *bridge_port);
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_trap.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_trap.c
new file mode 100644
index 000000000..f4bfdb6da
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_trap.c
@@ -0,0 +1,1969 @@
+// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
+/* Copyright (c) 2019 Mellanox Technologies. All rights reserved */
+
+#include <linux/bitops.h>
+#include <linux/kernel.h>
+#include <linux/netlink.h>
+#include <net/devlink.h>
+#include <uapi/linux/devlink.h>
+
+#include "core.h"
+#include "reg.h"
+#include "spectrum.h"
+#include "spectrum_trap.h"
+
+struct mlxsw_sp_trap_policer_item {
+	struct devlink_trap_policer policer;
+	u16 hw_id;
+};
+
+struct mlxsw_sp_trap_group_item {
+	struct devlink_trap_group group;
+	u16 hw_group_id;
+	u8 priority;
+	u8 fixed_policer:1; /* Whether policer binding can change */
+};
+
+#define MLXSW_SP_TRAP_LISTENERS_MAX 3
+
+struct mlxsw_sp_trap_item {
+	struct devlink_trap trap;
+	struct mlxsw_listener listeners_arr[MLXSW_SP_TRAP_LISTENERS_MAX];
+	u8 is_source:1;
+};
+
+/* All driver-specific traps must be documented in
+ * Documentation/networking/devlink/mlxsw.rst
+ */
+enum {
+	DEVLINK_MLXSW_TRAP_ID_BASE = DEVLINK_TRAP_GENERIC_ID_MAX,
+	DEVLINK_MLXSW_TRAP_ID_IRIF_DISABLED,
+	DEVLINK_MLXSW_TRAP_ID_ERIF_DISABLED,
+};
+
+#define DEVLINK_MLXSW_TRAP_NAME_IRIF_DISABLED \
+	"irif_disabled"
+#define DEVLINK_MLXSW_TRAP_NAME_ERIF_DISABLED \
+	"erif_disabled"
+
+#define MLXSW_SP_TRAP_METADATA DEVLINK_TRAP_METADATA_TYPE_F_IN_PORT
+
+enum {
+	/* Packet was mirrored from ingress. */
+	MLXSW_SP_MIRROR_REASON_INGRESS = 1,
+	/* Packet was mirrored from policy engine. */
+	MLXSW_SP_MIRROR_REASON_POLICY_ENGINE = 2,
+	/* Packet was early dropped. */
+	MLXSW_SP_MIRROR_REASON_INGRESS_WRED = 9,
+	/* Packet was mirrored from egress. */
+	MLXSW_SP_MIRROR_REASON_EGRESS = 14,
+};
+
+static int mlxsw_sp_rx_listener(struct mlxsw_sp *mlxsw_sp, struct sk_buff *skb,
+				u16 local_port,
+				struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	struct mlxsw_sp_port_pcpu_stats *pcpu_stats;
+
+	if (unlikely(!mlxsw_sp_port)) {
+		dev_warn_ratelimited(mlxsw_sp->bus_info->dev, "Port %d: skb received for non-existent port\n",
+				     local_port);
+		kfree_skb(skb);
+		return -EINVAL;
+	}
+
+	skb->dev = mlxsw_sp_port->dev;
+
+	pcpu_stats = this_cpu_ptr(mlxsw_sp_port->pcpu_stats);
+	u64_stats_update_begin(&pcpu_stats->syncp);
+	pcpu_stats->rx_packets++;
+	pcpu_stats->rx_bytes += skb->len;
+	u64_stats_update_end(&pcpu_stats->syncp);
+
+	skb->protocol = eth_type_trans(skb, skb->dev);
+
+	return 0;
+}
+
+static void mlxsw_sp_rx_drop_listener(struct sk_buff *skb, u16 local_port,
+				      void *trap_ctx)
+{
+	struct devlink_port *in_devlink_port;
+	struct mlxsw_sp_port *mlxsw_sp_port;
+	struct mlxsw_sp *mlxsw_sp;
+	struct devlink *devlink;
+	int err;
+
+	mlxsw_sp = devlink_trap_ctx_priv(trap_ctx);
+	mlxsw_sp_port = mlxsw_sp->ports[local_port];
+
+	err = mlxsw_sp_rx_listener(mlxsw_sp, skb, local_port, mlxsw_sp_port);
+	if (err)
+		return;
+
+	devlink = priv_to_devlink(mlxsw_sp->core);
+	in_devlink_port = mlxsw_core_port_devlink_port_get(mlxsw_sp->core,
+							   local_port);
+	skb_push(skb, ETH_HLEN);
+	devlink_trap_report(devlink, skb, trap_ctx, in_devlink_port, NULL);
+	consume_skb(skb);
+}
+
+static void mlxsw_sp_rx_acl_drop_listener(struct sk_buff *skb, u16 local_port,
+					  void *trap_ctx)
+{
+	u32 cookie_index = mlxsw_skb_cb(skb)->rx_md_info.cookie_index;
+	const struct flow_action_cookie *fa_cookie;
+	struct devlink_port *in_devlink_port;
+	struct mlxsw_sp_port *mlxsw_sp_port;
+	struct mlxsw_sp *mlxsw_sp;
+	struct devlink *devlink;
+	int err;
+
+	mlxsw_sp = devlink_trap_ctx_priv(trap_ctx);
+	mlxsw_sp_port = mlxsw_sp->ports[local_port];
+
+	err = mlxsw_sp_rx_listener(mlxsw_sp, skb, local_port, mlxsw_sp_port);
+	if (err)
+		return;
+
+	devlink = priv_to_devlink(mlxsw_sp->core);
+	in_devlink_port = mlxsw_core_port_devlink_port_get(mlxsw_sp->core,
+							   local_port);
+	skb_push(skb, ETH_HLEN);
+	rcu_read_lock();
+	fa_cookie = mlxsw_sp_acl_act_cookie_lookup(mlxsw_sp, cookie_index);
+	devlink_trap_report(devlink, skb, trap_ctx, in_devlink_port, fa_cookie);
+	rcu_read_unlock();
+	consume_skb(skb);
+}
+
+static int __mlxsw_sp_rx_no_mark_listener(struct sk_buff *skb, u16 local_port,
+					  void *trap_ctx)
+{
+	struct devlink_port *in_devlink_port;
+	struct mlxsw_sp_port *mlxsw_sp_port;
+	struct mlxsw_sp *mlxsw_sp;
+	struct devlink *devlink;
+	int err;
+
+	mlxsw_sp = devlink_trap_ctx_priv(trap_ctx);
+	mlxsw_sp_port = mlxsw_sp->ports[local_port];
+
+	err = mlxsw_sp_rx_listener(mlxsw_sp, skb, local_port, mlxsw_sp_port);
+	if (err)
+		return err;
+
+	devlink = priv_to_devlink(mlxsw_sp->core);
+	in_devlink_port = mlxsw_core_port_devlink_port_get(mlxsw_sp->core,
+							   local_port);
+	skb_push(skb, ETH_HLEN);
+	devlink_trap_report(devlink, skb, trap_ctx, in_devlink_port, NULL);
+	skb_pull(skb, ETH_HLEN);
+
+	return 0;
+}
+
+static void mlxsw_sp_rx_no_mark_listener(struct sk_buff *skb, u16 local_port,
+					 void *trap_ctx)
+{
+	int err;
+
+	err = __mlxsw_sp_rx_no_mark_listener(skb, local_port, trap_ctx);
+	if (err)
+		return;
+
+	netif_receive_skb(skb);
+}
+
+static void mlxsw_sp_rx_mark_listener(struct sk_buff *skb, u16 local_port,
+				      void *trap_ctx)
+{
+	skb->offload_fwd_mark = 1;
+	mlxsw_sp_rx_no_mark_listener(skb, local_port, trap_ctx);
+}
+
+static void mlxsw_sp_rx_l3_mark_listener(struct sk_buff *skb, u16 local_port,
+					 void *trap_ctx)
+{
+	skb->offload_l3_fwd_mark = 1;
+	skb->offload_fwd_mark = 1;
+	mlxsw_sp_rx_no_mark_listener(skb, local_port, trap_ctx);
+}
+
+static void mlxsw_sp_rx_ptp_listener(struct sk_buff *skb, u16 local_port,
+				     void *trap_ctx)
+{
+	struct mlxsw_sp *mlxsw_sp = devlink_trap_ctx_priv(trap_ctx);
+	int err;
+
+	err = __mlxsw_sp_rx_no_mark_listener(skb, local_port, trap_ctx);
+	if (err)
+		return;
+
+	/* The PTP handler expects skb->data to point to the start of the
+	 * Ethernet header.
+	 */
+	skb_push(skb, ETH_HLEN);
+	mlxsw_sp_ptp_receive(mlxsw_sp, skb, local_port);
+}
+
+static struct mlxsw_sp_port *
+mlxsw_sp_sample_tx_port_get(struct mlxsw_sp *mlxsw_sp,
+			    const struct mlxsw_rx_md_info *rx_md_info)
+{
+	u16 local_port;
+
+	if (!rx_md_info->tx_port_valid)
+		return NULL;
+
+	if (rx_md_info->tx_port_is_lag)
+		local_port = mlxsw_core_lag_mapping_get(mlxsw_sp->core,
+							rx_md_info->tx_lag_id,
+							rx_md_info->tx_lag_port_index);
+	else
+		local_port = rx_md_info->tx_sys_port;
+
+	if (local_port >= mlxsw_core_max_ports(mlxsw_sp->core))
+		return NULL;
+
+	return mlxsw_sp->ports[local_port];
+}
+
+/* The latency units are determined according to MOGCR.mirror_latency_units. It
+ * defaults to 64 nanoseconds.
+ */
+#define MLXSW_SP_MIRROR_LATENCY_SHIFT	6
+
+static void mlxsw_sp_psample_md_init(struct mlxsw_sp *mlxsw_sp,
+				     struct psample_metadata *md,
+				     struct sk_buff *skb, int in_ifindex,
+				     bool truncate, u32 trunc_size)
+{
+	struct mlxsw_rx_md_info *rx_md_info = &mlxsw_skb_cb(skb)->rx_md_info;
+	struct mlxsw_sp_port *mlxsw_sp_port;
+
+	md->trunc_size = truncate ? trunc_size : skb->len;
+	md->in_ifindex = in_ifindex;
+	mlxsw_sp_port = mlxsw_sp_sample_tx_port_get(mlxsw_sp, rx_md_info);
+	md->out_ifindex = mlxsw_sp_port && mlxsw_sp_port->dev ?
+			  mlxsw_sp_port->dev->ifindex : 0;
+	md->out_tc_valid = rx_md_info->tx_tc_valid;
+	md->out_tc = rx_md_info->tx_tc;
+	md->out_tc_occ_valid = rx_md_info->tx_congestion_valid;
+	md->out_tc_occ = rx_md_info->tx_congestion;
+	md->latency_valid = rx_md_info->latency_valid;
+	md->latency = rx_md_info->latency;
+	md->latency <<= MLXSW_SP_MIRROR_LATENCY_SHIFT;
+}
+
+static void mlxsw_sp_rx_sample_listener(struct sk_buff *skb, u16 local_port,
+					void *trap_ctx)
+{
+	struct mlxsw_sp *mlxsw_sp = devlink_trap_ctx_priv(trap_ctx);
+	struct mlxsw_sp_sample_trigger trigger;
+	struct mlxsw_sp_sample_params *params;
+	struct mlxsw_sp_port *mlxsw_sp_port;
+	struct psample_metadata md = {};
+	int err;
+
+	err = __mlxsw_sp_rx_no_mark_listener(skb, local_port, trap_ctx);
+	if (err)
+		return;
+
+	mlxsw_sp_port = mlxsw_sp->ports[local_port];
+	if (!mlxsw_sp_port)
+		goto out;
+
+	trigger.type = MLXSW_SP_SAMPLE_TRIGGER_TYPE_INGRESS;
+	trigger.local_port = local_port;
+	params = mlxsw_sp_sample_trigger_params_lookup(mlxsw_sp, &trigger);
+	if (!params)
+		goto out;
+
+	/* The psample module expects skb->data to point to the start of the
+	 * Ethernet header.
+	 */
+	skb_push(skb, ETH_HLEN);
+	mlxsw_sp_psample_md_init(mlxsw_sp, &md, skb,
+				 mlxsw_sp_port->dev->ifindex, params->truncate,
+				 params->trunc_size);
+	psample_sample_packet(params->psample_group, skb, params->rate, &md);
+out:
+	consume_skb(skb);
+}
+
+static void mlxsw_sp_rx_sample_tx_listener(struct sk_buff *skb, u16 local_port,
+					   void *trap_ctx)
+{
+	struct mlxsw_rx_md_info *rx_md_info = &mlxsw_skb_cb(skb)->rx_md_info;
+	struct mlxsw_sp *mlxsw_sp = devlink_trap_ctx_priv(trap_ctx);
+	struct mlxsw_sp_port *mlxsw_sp_port, *mlxsw_sp_port_tx;
+	struct mlxsw_sp_sample_trigger trigger;
+	struct mlxsw_sp_sample_params *params;
+	struct psample_metadata md = {};
+	int err;
+
+	/* Locally generated packets are not reported from the policy engine
+	 * trigger, so do not report them from the egress trigger as well.
+	 */
+	if (local_port == MLXSW_PORT_CPU_PORT)
+		goto out;
+
+	err = __mlxsw_sp_rx_no_mark_listener(skb, local_port, trap_ctx);
+	if (err)
+		return;
+
+	mlxsw_sp_port = mlxsw_sp->ports[local_port];
+	if (!mlxsw_sp_port)
+		goto out;
+
+	/* Packet was sampled from Tx, so we need to retrieve the sample
+	 * parameters based on the Tx port and not the Rx port.
+	 */
+	mlxsw_sp_port_tx = mlxsw_sp_sample_tx_port_get(mlxsw_sp, rx_md_info);
+	if (!mlxsw_sp_port_tx)
+		goto out;
+
+	trigger.type = MLXSW_SP_SAMPLE_TRIGGER_TYPE_EGRESS;
+	trigger.local_port = mlxsw_sp_port_tx->local_port;
+	params = mlxsw_sp_sample_trigger_params_lookup(mlxsw_sp, &trigger);
+	if (!params)
+		goto out;
+
+	/* The psample module expects skb->data to point to the start of the
+	 * Ethernet header.
+	 */
+	skb_push(skb, ETH_HLEN);
+	mlxsw_sp_psample_md_init(mlxsw_sp, &md, skb,
+				 mlxsw_sp_port->dev->ifindex, params->truncate,
+				 params->trunc_size);
+	psample_sample_packet(params->psample_group, skb, params->rate, &md);
+out:
+	consume_skb(skb);
+}
+
+static void mlxsw_sp_rx_sample_acl_listener(struct sk_buff *skb, u16 local_port,
+					    void *trap_ctx)
+{
+	struct mlxsw_sp *mlxsw_sp = devlink_trap_ctx_priv(trap_ctx);
+	struct mlxsw_sp_sample_trigger trigger = {
+		.type = MLXSW_SP_SAMPLE_TRIGGER_TYPE_POLICY_ENGINE,
+	};
+	struct mlxsw_sp_sample_params *params;
+	struct mlxsw_sp_port *mlxsw_sp_port;
+	struct psample_metadata md = {};
+	int err;
+
+	err = __mlxsw_sp_rx_no_mark_listener(skb, local_port, trap_ctx);
+	if (err)
+		return;
+
+	mlxsw_sp_port = mlxsw_sp->ports[local_port];
+	if (!mlxsw_sp_port)
+		goto out;
+
+	params = mlxsw_sp_sample_trigger_params_lookup(mlxsw_sp, &trigger);
+	if (!params)
+		goto out;
+
+	/* The psample module expects skb->data to point to the start of the
+	 * Ethernet header.
+	 */
+	skb_push(skb, ETH_HLEN);
+	mlxsw_sp_psample_md_init(mlxsw_sp, &md, skb,
+				 mlxsw_sp_port->dev->ifindex, params->truncate,
+				 params->trunc_size);
+	psample_sample_packet(params->psample_group, skb, params->rate, &md);
+out:
+	consume_skb(skb);
+}
+
+#define MLXSW_SP_TRAP_DROP(_id, _group_id)				      \
+	DEVLINK_TRAP_GENERIC(DROP, DROP, _id,				      \
+			     DEVLINK_TRAP_GROUP_GENERIC_ID_##_group_id,	      \
+			     MLXSW_SP_TRAP_METADATA)
+
+#define MLXSW_SP_TRAP_DROP_EXT(_id, _group_id, _metadata)		      \
+	DEVLINK_TRAP_GENERIC(DROP, DROP, _id,				      \
+			     DEVLINK_TRAP_GROUP_GENERIC_ID_##_group_id,	      \
+			     MLXSW_SP_TRAP_METADATA | (_metadata))
+
+#define MLXSW_SP_TRAP_BUFFER_DROP(_id)					      \
+	DEVLINK_TRAP_GENERIC(DROP, TRAP, _id,				      \
+			     DEVLINK_TRAP_GROUP_GENERIC_ID_BUFFER_DROPS,      \
+			     MLXSW_SP_TRAP_METADATA)
+
+#define MLXSW_SP_TRAP_DRIVER_DROP(_id, _group_id)			      \
+	DEVLINK_TRAP_DRIVER(DROP, DROP, DEVLINK_MLXSW_TRAP_ID_##_id,	      \
+			    DEVLINK_MLXSW_TRAP_NAME_##_id,		      \
+			    DEVLINK_TRAP_GROUP_GENERIC_ID_##_group_id,	      \
+			    MLXSW_SP_TRAP_METADATA)
+
+#define MLXSW_SP_TRAP_EXCEPTION(_id, _group_id)		      \
+	DEVLINK_TRAP_GENERIC(EXCEPTION, TRAP, _id,			      \
+			     DEVLINK_TRAP_GROUP_GENERIC_ID_##_group_id,	      \
+			     MLXSW_SP_TRAP_METADATA)
+
+#define MLXSW_SP_TRAP_CONTROL(_id, _group_id, _action)			      \
+	DEVLINK_TRAP_GENERIC(CONTROL, _action, _id,			      \
+			     DEVLINK_TRAP_GROUP_GENERIC_ID_##_group_id,	      \
+			     MLXSW_SP_TRAP_METADATA)
+
+#define MLXSW_SP_RXL_DISCARD(_id, _group_id)				      \
+	MLXSW_RXL_DIS(mlxsw_sp_rx_drop_listener, DISCARD_##_id,		      \
+		      TRAP_EXCEPTION_TO_CPU, false, SP_##_group_id,	      \
+		      SET_FW_DEFAULT, SP_##_group_id)
+
+#define MLXSW_SP_RXL_ACL_DISCARD(_id, _en_group_id, _dis_group_id)	      \
+	MLXSW_RXL_DIS(mlxsw_sp_rx_acl_drop_listener, DISCARD_##_id,	      \
+		      TRAP_EXCEPTION_TO_CPU, false, SP_##_en_group_id,	      \
+		      SET_FW_DEFAULT, SP_##_dis_group_id)
+
+#define MLXSW_SP_RXL_BUFFER_DISCARD(_mirror_reason)			      \
+	MLXSW_RXL_MIRROR(mlxsw_sp_rx_drop_listener, 0, SP_BUFFER_DISCARDS,    \
+			 MLXSW_SP_MIRROR_REASON_##_mirror_reason)
+
+#define MLXSW_SP_RXL_EXCEPTION(_id, _group_id, _action)			      \
+	MLXSW_RXL(mlxsw_sp_rx_mark_listener, _id,			      \
+		   _action, false, SP_##_group_id, SET_FW_DEFAULT)
+
+#define MLXSW_SP_RXL_NO_MARK(_id, _group_id, _action, _is_ctrl)		      \
+	MLXSW_RXL(mlxsw_sp_rx_no_mark_listener, _id, _action,		      \
+		  _is_ctrl, SP_##_group_id, DISCARD)
+
+#define MLXSW_SP_RXL_MARK(_id, _group_id, _action, _is_ctrl)		      \
+	MLXSW_RXL(mlxsw_sp_rx_mark_listener, _id, _action, _is_ctrl,	      \
+		  SP_##_group_id, DISCARD)
+
+#define MLXSW_SP_RXL_L3_MARK(_id, _group_id, _action, _is_ctrl)		      \
+	MLXSW_RXL(mlxsw_sp_rx_l3_mark_listener, _id, _action, _is_ctrl,	      \
+		  SP_##_group_id, DISCARD)
+
+#define MLXSW_SP_TRAP_POLICER(_id, _rate, _burst)			      \
+	DEVLINK_TRAP_POLICER(_id, _rate, _burst,			      \
+			     MLXSW_REG_QPCR_HIGHEST_CIR,		      \
+			     MLXSW_REG_QPCR_LOWEST_CIR,			      \
+			     1 << MLXSW_REG_QPCR_HIGHEST_CBS,		      \
+			     1 << MLXSW_REG_QPCR_LOWEST_CBS)
+
+/* Ordered by policer identifier */
+static const struct mlxsw_sp_trap_policer_item
+mlxsw_sp_trap_policer_items_arr[] = {
+	{
+		.policer = MLXSW_SP_TRAP_POLICER(1, 10 * 1024, 4096),
+	},
+	{
+		.policer = MLXSW_SP_TRAP_POLICER(2, 128, 128),
+	},
+	{
+		.policer = MLXSW_SP_TRAP_POLICER(3, 128, 128),
+	},
+	{
+		.policer = MLXSW_SP_TRAP_POLICER(4, 128, 128),
+	},
+	{
+		.policer = MLXSW_SP_TRAP_POLICER(5, 16 * 1024, 8192),
+	},
+	{
+		.policer = MLXSW_SP_TRAP_POLICER(6, 128, 128),
+	},
+	{
+		.policer = MLXSW_SP_TRAP_POLICER(7, 1024, 512),
+	},
+	{
+		.policer = MLXSW_SP_TRAP_POLICER(8, 20 * 1024, 8192),
+	},
+	{
+		.policer = MLXSW_SP_TRAP_POLICER(9, 128, 128),
+	},
+	{
+		.policer = MLXSW_SP_TRAP_POLICER(10, 1024, 512),
+	},
+	{
+		.policer = MLXSW_SP_TRAP_POLICER(11, 256, 128),
+	},
+	{
+		.policer = MLXSW_SP_TRAP_POLICER(12, 128, 128),
+	},
+	{
+		.policer = MLXSW_SP_TRAP_POLICER(13, 128, 128),
+	},
+	{
+		.policer = MLXSW_SP_TRAP_POLICER(14, 1024, 512),
+	},
+	{
+		.policer = MLXSW_SP_TRAP_POLICER(15, 1024, 512),
+	},
+	{
+		.policer = MLXSW_SP_TRAP_POLICER(16, 24 * 1024, 16384),
+	},
+	{
+		.policer = MLXSW_SP_TRAP_POLICER(17, 19 * 1024, 8192),
+	},
+	{
+		.policer = MLXSW_SP_TRAP_POLICER(18, 1024, 512),
+	},
+	{
+		.policer = MLXSW_SP_TRAP_POLICER(19, 1024, 512),
+	},
+	{
+		.policer = MLXSW_SP_TRAP_POLICER(20, 10240, 4096),
+	},
+};
+
+static const struct mlxsw_sp_trap_group_item mlxsw_sp_trap_group_items_arr[] = {
+	{
+		.group = DEVLINK_TRAP_GROUP_GENERIC(L2_DROPS, 1),
+		.hw_group_id = MLXSW_REG_HTGT_TRAP_GROUP_SP_L2_DISCARDS,
+		.priority = 0,
+	},
+	{
+		.group = DEVLINK_TRAP_GROUP_GENERIC(L3_DROPS, 1),
+		.hw_group_id = MLXSW_REG_HTGT_TRAP_GROUP_SP_L3_DISCARDS,
+		.priority = 0,
+	},
+	{
+		.group = DEVLINK_TRAP_GROUP_GENERIC(L3_EXCEPTIONS, 1),
+		.hw_group_id = MLXSW_REG_HTGT_TRAP_GROUP_SP_L3_EXCEPTIONS,
+		.priority = 2,
+	},
+	{
+		.group = DEVLINK_TRAP_GROUP_GENERIC(TUNNEL_DROPS, 1),
+		.hw_group_id = MLXSW_REG_HTGT_TRAP_GROUP_SP_TUNNEL_DISCARDS,
+		.priority = 0,
+	},
+	{
+		.group = DEVLINK_TRAP_GROUP_GENERIC(ACL_DROPS, 1),
+		.hw_group_id = MLXSW_REG_HTGT_TRAP_GROUP_SP_ACL_DISCARDS,
+		.priority = 0,
+	},
+	{
+		.group = DEVLINK_TRAP_GROUP_GENERIC(STP, 2),
+		.hw_group_id = MLXSW_REG_HTGT_TRAP_GROUP_SP_STP,
+		.priority = 5,
+	},
+	{
+		.group = DEVLINK_TRAP_GROUP_GENERIC(LACP, 3),
+		.hw_group_id = MLXSW_REG_HTGT_TRAP_GROUP_SP_LACP,
+		.priority = 5,
+	},
+	{
+		.group = DEVLINK_TRAP_GROUP_GENERIC(LLDP, 4),
+		.hw_group_id = MLXSW_REG_HTGT_TRAP_GROUP_SP_LLDP,
+		.priority = 5,
+	},
+	{
+		.group = DEVLINK_TRAP_GROUP_GENERIC(MC_SNOOPING, 5),
+		.hw_group_id = MLXSW_REG_HTGT_TRAP_GROUP_SP_MC_SNOOPING,
+		.priority = 3,
+	},
+	{
+		.group = DEVLINK_TRAP_GROUP_GENERIC(DHCP, 6),
+		.hw_group_id = MLXSW_REG_HTGT_TRAP_GROUP_SP_DHCP,
+		.priority = 2,
+	},
+	{
+		.group = DEVLINK_TRAP_GROUP_GENERIC(NEIGH_DISCOVERY, 7),
+		.hw_group_id = MLXSW_REG_HTGT_TRAP_GROUP_SP_NEIGH_DISCOVERY,
+		.priority = 2,
+	},
+	{
+		.group = DEVLINK_TRAP_GROUP_GENERIC(BFD, 8),
+		.hw_group_id = MLXSW_REG_HTGT_TRAP_GROUP_SP_BFD,
+		.priority = 5,
+	},
+	{
+		.group = DEVLINK_TRAP_GROUP_GENERIC(OSPF, 9),
+		.hw_group_id = MLXSW_REG_HTGT_TRAP_GROUP_SP_OSPF,
+		.priority = 5,
+	},
+	{
+		.group = DEVLINK_TRAP_GROUP_GENERIC(BGP, 10),
+		.hw_group_id = MLXSW_REG_HTGT_TRAP_GROUP_SP_BGP,
+		.priority = 4,
+	},
+	{
+		.group = DEVLINK_TRAP_GROUP_GENERIC(VRRP, 11),
+		.hw_group_id = MLXSW_REG_HTGT_TRAP_GROUP_SP_VRRP,
+		.priority = 5,
+	},
+	{
+		.group = DEVLINK_TRAP_GROUP_GENERIC(PIM, 12),
+		.hw_group_id = MLXSW_REG_HTGT_TRAP_GROUP_SP_PIM,
+		.priority = 5,
+	},
+	{
+		.group = DEVLINK_TRAP_GROUP_GENERIC(UC_LB, 13),
+		.hw_group_id = MLXSW_REG_HTGT_TRAP_GROUP_SP_LBERROR,
+		.priority = 0,
+	},
+	{
+		.group = DEVLINK_TRAP_GROUP_GENERIC(LOCAL_DELIVERY, 14),
+		.hw_group_id = MLXSW_REG_HTGT_TRAP_GROUP_SP_IP2ME,
+		.priority = 2,
+	},
+	{
+		.group = DEVLINK_TRAP_GROUP_GENERIC(EXTERNAL_DELIVERY, 19),
+		.hw_group_id = MLXSW_REG_HTGT_TRAP_GROUP_SP_EXTERNAL_ROUTE,
+		.priority = 1,
+	},
+	{
+		.group = DEVLINK_TRAP_GROUP_GENERIC(IPV6, 15),
+		.hw_group_id = MLXSW_REG_HTGT_TRAP_GROUP_SP_IPV6,
+		.priority = 2,
+	},
+	{
+		.group = DEVLINK_TRAP_GROUP_GENERIC(PTP_EVENT, 16),
+		.hw_group_id = MLXSW_REG_HTGT_TRAP_GROUP_SP_PTP0,
+		.priority = 5,
+	},
+	{
+		.group = DEVLINK_TRAP_GROUP_GENERIC(PTP_GENERAL, 17),
+		.hw_group_id = MLXSW_REG_HTGT_TRAP_GROUP_SP_PTP1,
+		.priority = 2,
+	},
+	{
+		.group = DEVLINK_TRAP_GROUP_GENERIC(ACL_TRAP, 18),
+		.hw_group_id = MLXSW_REG_HTGT_TRAP_GROUP_SP_FLOW_LOGGING,
+		.priority = 4,
+	},
+};
+
+static const struct mlxsw_sp_trap_item mlxsw_sp_trap_items_arr[] = {
+	{
+		.trap = MLXSW_SP_TRAP_DROP(SMAC_MC, L2_DROPS),
+		.listeners_arr = {
+			MLXSW_SP_RXL_DISCARD(ING_PACKET_SMAC_MC, L2_DISCARDS),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_DROP(VLAN_TAG_MISMATCH, L2_DROPS),
+		.listeners_arr = {
+			MLXSW_SP_RXL_DISCARD(ING_SWITCH_VTAG_ALLOW,
+					     L2_DISCARDS),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_DROP(INGRESS_VLAN_FILTER, L2_DROPS),
+		.listeners_arr = {
+			MLXSW_SP_RXL_DISCARD(ING_SWITCH_VLAN, L2_DISCARDS),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_DROP(INGRESS_STP_FILTER, L2_DROPS),
+		.listeners_arr = {
+			MLXSW_SP_RXL_DISCARD(ING_SWITCH_STP, L2_DISCARDS),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_DROP(EMPTY_TX_LIST, L2_DROPS),
+		.listeners_arr = {
+			MLXSW_SP_RXL_DISCARD(LOOKUP_SWITCH_UC, L2_DISCARDS),
+			MLXSW_SP_RXL_DISCARD(LOOKUP_SWITCH_MC_NULL, L2_DISCARDS),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_DROP(PORT_LOOPBACK_FILTER, L2_DROPS),
+		.listeners_arr = {
+			MLXSW_SP_RXL_DISCARD(LOOKUP_SWITCH_LB, L2_DISCARDS),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_DROP(BLACKHOLE_ROUTE, L3_DROPS),
+		.listeners_arr = {
+			MLXSW_SP_RXL_DISCARD(ROUTER2, L3_DISCARDS),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_DROP(NON_IP_PACKET, L3_DROPS),
+		.listeners_arr = {
+			MLXSW_SP_RXL_DISCARD(ING_ROUTER_NON_IP_PACKET,
+					     L3_DISCARDS),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_DROP(UC_DIP_MC_DMAC, L3_DROPS),
+		.listeners_arr = {
+			MLXSW_SP_RXL_DISCARD(ING_ROUTER_UC_DIP_MC_DMAC,
+					     L3_DISCARDS),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_DROP(DIP_LB, L3_DROPS),
+		.listeners_arr = {
+			MLXSW_SP_RXL_DISCARD(ING_ROUTER_DIP_LB, L3_DISCARDS),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_DROP(SIP_MC, L3_DROPS),
+		.listeners_arr = {
+			MLXSW_SP_RXL_DISCARD(ING_ROUTER_SIP_MC, L3_DISCARDS),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_DROP(SIP_LB, L3_DROPS),
+		.listeners_arr = {
+			MLXSW_SP_RXL_DISCARD(ING_ROUTER_SIP_LB, L3_DISCARDS),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_DROP(CORRUPTED_IP_HDR, L3_DROPS),
+		.listeners_arr = {
+			MLXSW_SP_RXL_DISCARD(ING_ROUTER_CORRUPTED_IP_HDR,
+					     L3_DISCARDS),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_DROP(IPV4_SIP_BC, L3_DROPS),
+		.listeners_arr = {
+			MLXSW_SP_RXL_DISCARD(ING_ROUTER_IPV4_SIP_BC,
+					     L3_DISCARDS),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_DROP(IPV6_MC_DIP_RESERVED_SCOPE,
+					   L3_DROPS),
+		.listeners_arr = {
+			MLXSW_SP_RXL_DISCARD(IPV6_MC_DIP_RESERVED_SCOPE,
+					     L3_DISCARDS),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_DROP(IPV6_MC_DIP_INTERFACE_LOCAL_SCOPE,
+					   L3_DROPS),
+		.listeners_arr = {
+			MLXSW_SP_RXL_DISCARD(IPV6_MC_DIP_INTERFACE_LOCAL_SCOPE,
+					     L3_DISCARDS),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_EXCEPTION(MTU_ERROR, L3_EXCEPTIONS),
+		.listeners_arr = {
+			MLXSW_SP_RXL_EXCEPTION(MTUERROR, L3_EXCEPTIONS,
+					       TRAP_TO_CPU),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_EXCEPTION(TTL_ERROR, L3_EXCEPTIONS),
+		.listeners_arr = {
+			MLXSW_SP_RXL_EXCEPTION(TTLERROR, L3_EXCEPTIONS,
+					       TRAP_TO_CPU),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_EXCEPTION(RPF, L3_EXCEPTIONS),
+		.listeners_arr = {
+			MLXSW_SP_RXL_EXCEPTION(RPF, L3_EXCEPTIONS, TRAP_TO_CPU),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_EXCEPTION(REJECT_ROUTE, L3_EXCEPTIONS),
+		.listeners_arr = {
+			MLXSW_SP_RXL_EXCEPTION(RTR_INGRESS1, L3_EXCEPTIONS,
+					       TRAP_TO_CPU),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_EXCEPTION(UNRESOLVED_NEIGH,
+						L3_EXCEPTIONS),
+		.listeners_arr = {
+			MLXSW_SP_RXL_EXCEPTION(HOST_MISS_IPV4, L3_EXCEPTIONS,
+					       TRAP_TO_CPU),
+			MLXSW_SP_RXL_EXCEPTION(HOST_MISS_IPV6, L3_EXCEPTIONS,
+					       TRAP_TO_CPU),
+			MLXSW_SP_RXL_EXCEPTION(RTR_EGRESS0, L3_EXCEPTIONS,
+					       TRAP_EXCEPTION_TO_CPU),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_EXCEPTION(IPV4_LPM_UNICAST_MISS,
+						L3_EXCEPTIONS),
+		.listeners_arr = {
+			MLXSW_SP_RXL_EXCEPTION(DISCARD_ROUTER_LPM4,
+					       L3_EXCEPTIONS,
+					       TRAP_EXCEPTION_TO_CPU),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_EXCEPTION(IPV6_LPM_UNICAST_MISS,
+						L3_EXCEPTIONS),
+		.listeners_arr = {
+			MLXSW_SP_RXL_EXCEPTION(DISCARD_ROUTER_LPM6,
+					       L3_EXCEPTIONS,
+					       TRAP_EXCEPTION_TO_CPU),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_DRIVER_DROP(IRIF_DISABLED, L3_DROPS),
+		.listeners_arr = {
+			MLXSW_SP_RXL_DISCARD(ROUTER_IRIF_EN, L3_DISCARDS),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_DRIVER_DROP(ERIF_DISABLED, L3_DROPS),
+		.listeners_arr = {
+			MLXSW_SP_RXL_DISCARD(ROUTER_ERIF_EN, L3_DISCARDS),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_DROP(NON_ROUTABLE, L3_DROPS),
+		.listeners_arr = {
+			MLXSW_SP_RXL_DISCARD(NON_ROUTABLE, L3_DISCARDS),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_EXCEPTION(DECAP_ERROR, TUNNEL_DROPS),
+		.listeners_arr = {
+			MLXSW_SP_RXL_EXCEPTION(DECAP_ECN0, TUNNEL_DISCARDS,
+					       TRAP_EXCEPTION_TO_CPU),
+			MLXSW_SP_RXL_EXCEPTION(IPIP_DECAP_ERROR,
+					       TUNNEL_DISCARDS,
+					       TRAP_EXCEPTION_TO_CPU),
+			MLXSW_SP_RXL_EXCEPTION(DISCARD_DEC_PKT, TUNNEL_DISCARDS,
+					       TRAP_EXCEPTION_TO_CPU),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_DROP(OVERLAY_SMAC_MC, TUNNEL_DROPS),
+		.listeners_arr = {
+			MLXSW_SP_RXL_DISCARD(OVERLAY_SMAC_MC, TUNNEL_DISCARDS),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_DROP_EXT(INGRESS_FLOW_ACTION_DROP,
+					       ACL_DROPS,
+					       DEVLINK_TRAP_METADATA_TYPE_F_FA_COOKIE),
+		.listeners_arr = {
+			MLXSW_SP_RXL_ACL_DISCARD(INGRESS_ACL, ACL_DISCARDS,
+						 DUMMY),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_DROP_EXT(EGRESS_FLOW_ACTION_DROP,
+					       ACL_DROPS,
+					       DEVLINK_TRAP_METADATA_TYPE_F_FA_COOKIE),
+		.listeners_arr = {
+			MLXSW_SP_RXL_ACL_DISCARD(EGRESS_ACL, ACL_DISCARDS,
+						 DUMMY),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_CONTROL(STP, STP, TRAP),
+		.listeners_arr = {
+			MLXSW_SP_RXL_NO_MARK(STP, STP, TRAP_TO_CPU, true),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_CONTROL(LACP, LACP, TRAP),
+		.listeners_arr = {
+			MLXSW_SP_RXL_NO_MARK(LACP, LACP, TRAP_TO_CPU, true),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_CONTROL(LLDP, LLDP, TRAP),
+		.listeners_arr = {
+			MLXSW_RXL(mlxsw_sp_rx_ptp_listener, LLDP, TRAP_TO_CPU,
+				  true, SP_LLDP, DISCARD),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_CONTROL(IGMP_QUERY, MC_SNOOPING, MIRROR),
+		.listeners_arr = {
+			MLXSW_SP_RXL_MARK(IGMP_QUERY, MC_SNOOPING,
+					  MIRROR_TO_CPU, false),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_CONTROL(IGMP_V1_REPORT, MC_SNOOPING,
+					      TRAP),
+		.listeners_arr = {
+			MLXSW_SP_RXL_NO_MARK(IGMP_V1_REPORT, MC_SNOOPING,
+					     TRAP_TO_CPU, false),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_CONTROL(IGMP_V2_REPORT, MC_SNOOPING,
+					      TRAP),
+		.listeners_arr = {
+			MLXSW_SP_RXL_NO_MARK(IGMP_V2_REPORT, MC_SNOOPING,
+					     TRAP_TO_CPU, false),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_CONTROL(IGMP_V3_REPORT, MC_SNOOPING,
+					      TRAP),
+		.listeners_arr = {
+			MLXSW_SP_RXL_NO_MARK(IGMP_V3_REPORT, MC_SNOOPING,
+					     TRAP_TO_CPU, false),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_CONTROL(IGMP_V2_LEAVE, MC_SNOOPING,
+					      TRAP),
+		.listeners_arr = {
+			MLXSW_SP_RXL_NO_MARK(IGMP_V2_LEAVE, MC_SNOOPING,
+					     TRAP_TO_CPU, false),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_CONTROL(MLD_QUERY, MC_SNOOPING, MIRROR),
+		.listeners_arr = {
+			MLXSW_SP_RXL_MARK(IPV6_MLDV12_LISTENER_QUERY,
+					  MC_SNOOPING, MIRROR_TO_CPU, false),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_CONTROL(MLD_V1_REPORT, MC_SNOOPING,
+					      TRAP),
+		.listeners_arr = {
+			MLXSW_SP_RXL_NO_MARK(IPV6_MLDV1_LISTENER_REPORT,
+					     MC_SNOOPING, TRAP_TO_CPU, false),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_CONTROL(MLD_V2_REPORT, MC_SNOOPING,
+					      TRAP),
+		.listeners_arr = {
+			MLXSW_SP_RXL_NO_MARK(IPV6_MLDV2_LISTENER_REPORT,
+					     MC_SNOOPING, TRAP_TO_CPU, false),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_CONTROL(MLD_V1_DONE, MC_SNOOPING,
+					      TRAP),
+		.listeners_arr = {
+			MLXSW_SP_RXL_NO_MARK(IPV6_MLDV1_LISTENER_DONE,
+					     MC_SNOOPING, TRAP_TO_CPU, false),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_CONTROL(IPV4_DHCP, DHCP, TRAP),
+		.listeners_arr = {
+			MLXSW_SP_RXL_MARK(IPV4_DHCP, DHCP, TRAP_TO_CPU, false),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_CONTROL(IPV6_DHCP, DHCP, TRAP),
+		.listeners_arr = {
+			MLXSW_SP_RXL_MARK(IPV6_DHCP, DHCP, TRAP_TO_CPU, false),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_CONTROL(ARP_REQUEST, NEIGH_DISCOVERY,
+					      MIRROR),
+		.listeners_arr = {
+			MLXSW_SP_RXL_MARK(ROUTER_ARPBC, NEIGH_DISCOVERY,
+					  TRAP_TO_CPU, false),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_CONTROL(ARP_RESPONSE, NEIGH_DISCOVERY,
+					      MIRROR),
+		.listeners_arr = {
+			MLXSW_SP_RXL_MARK(ROUTER_ARPUC, NEIGH_DISCOVERY,
+					  TRAP_TO_CPU, false),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_CONTROL(ARP_OVERLAY, NEIGH_DISCOVERY,
+					      TRAP),
+		.listeners_arr = {
+			MLXSW_SP_RXL_NO_MARK(NVE_DECAP_ARP, NEIGH_DISCOVERY,
+					     TRAP_TO_CPU, false),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_CONTROL(IPV6_NEIGH_SOLICIT,
+					      NEIGH_DISCOVERY, TRAP),
+		.listeners_arr = {
+			MLXSW_SP_RXL_MARK(L3_IPV6_NEIGHBOR_SOLICITATION,
+					  NEIGH_DISCOVERY, TRAP_TO_CPU, false),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_CONTROL(IPV6_NEIGH_ADVERT,
+					      NEIGH_DISCOVERY, TRAP),
+		.listeners_arr = {
+			MLXSW_SP_RXL_MARK(L3_IPV6_NEIGHBOR_ADVERTISEMENT,
+					  NEIGH_DISCOVERY, TRAP_TO_CPU, false),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_CONTROL(IPV4_BFD, BFD, TRAP),
+		.listeners_arr = {
+			MLXSW_SP_RXL_MARK(IPV4_BFD, BFD, TRAP_TO_CPU, false),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_CONTROL(IPV6_BFD, BFD, TRAP),
+		.listeners_arr = {
+			MLXSW_SP_RXL_MARK(IPV6_BFD, BFD, TRAP_TO_CPU, false),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_CONTROL(IPV4_OSPF, OSPF, TRAP),
+		.listeners_arr = {
+			MLXSW_SP_RXL_MARK(IPV4_OSPF, OSPF, TRAP_TO_CPU, false),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_CONTROL(IPV6_OSPF, OSPF, TRAP),
+		.listeners_arr = {
+			MLXSW_SP_RXL_MARK(IPV6_OSPF, OSPF, TRAP_TO_CPU, false),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_CONTROL(IPV4_BGP, BGP, TRAP),
+		.listeners_arr = {
+			MLXSW_SP_RXL_MARK(IPV4_BGP, BGP, TRAP_TO_CPU, false),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_CONTROL(IPV6_BGP, BGP, TRAP),
+		.listeners_arr = {
+			MLXSW_SP_RXL_MARK(IPV6_BGP, BGP, TRAP_TO_CPU, false),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_CONTROL(IPV4_VRRP, VRRP, TRAP),
+		.listeners_arr = {
+			MLXSW_SP_RXL_MARK(IPV4_VRRP, VRRP, TRAP_TO_CPU, false),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_CONTROL(IPV6_VRRP, VRRP, TRAP),
+		.listeners_arr = {
+			MLXSW_SP_RXL_MARK(IPV6_VRRP, VRRP, TRAP_TO_CPU, false),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_CONTROL(IPV4_PIM, PIM, TRAP),
+		.listeners_arr = {
+			MLXSW_SP_RXL_MARK(IPV4_PIM, PIM, TRAP_TO_CPU, false),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_CONTROL(IPV6_PIM, PIM, TRAP),
+		.listeners_arr = {
+			MLXSW_SP_RXL_MARK(IPV6_PIM, PIM, TRAP_TO_CPU, false),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_CONTROL(UC_LB, UC_LB, MIRROR),
+		.listeners_arr = {
+			MLXSW_SP_RXL_L3_MARK(LBERROR, LBERROR, MIRROR_TO_CPU,
+					     false),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_CONTROL(LOCAL_ROUTE, LOCAL_DELIVERY,
+					      TRAP),
+		.listeners_arr = {
+			MLXSW_SP_RXL_MARK(IP2ME, IP2ME, TRAP_TO_CPU, false),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_CONTROL(EXTERNAL_ROUTE, EXTERNAL_DELIVERY,
+					      TRAP),
+		.listeners_arr = {
+			MLXSW_SP_RXL_MARK(RTR_INGRESS0, EXTERNAL_ROUTE,
+					  TRAP_TO_CPU, false),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_CONTROL(IPV6_UC_DIP_LINK_LOCAL_SCOPE,
+					      LOCAL_DELIVERY, TRAP),
+		.listeners_arr = {
+			MLXSW_SP_RXL_MARK(IPV6_LINK_LOCAL_DEST, IP2ME,
+					  TRAP_TO_CPU, false),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_CONTROL(IPV4_ROUTER_ALERT, LOCAL_DELIVERY,
+					      TRAP),
+		.listeners_arr = {
+			MLXSW_SP_RXL_MARK(ROUTER_ALERT_IPV4, IP2ME, TRAP_TO_CPU,
+					  false),
+		},
+	},
+	{
+		/* IPV6_ROUTER_ALERT is defined in uAPI as 22, but it is not
+		 * used in this file, so undefine it.
+		 */
+		#undef IPV6_ROUTER_ALERT
+		.trap = MLXSW_SP_TRAP_CONTROL(IPV6_ROUTER_ALERT, LOCAL_DELIVERY,
+					      TRAP),
+		.listeners_arr = {
+			MLXSW_SP_RXL_MARK(ROUTER_ALERT_IPV6, IP2ME, TRAP_TO_CPU,
+					  false),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_CONTROL(IPV6_DIP_ALL_NODES, IPV6, TRAP),
+		.listeners_arr = {
+			MLXSW_SP_RXL_MARK(IPV6_ALL_NODES_LINK, IPV6,
+					  TRAP_TO_CPU, false),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_CONTROL(IPV6_DIP_ALL_ROUTERS, IPV6, TRAP),
+		.listeners_arr = {
+			MLXSW_SP_RXL_MARK(IPV6_ALL_ROUTERS_LINK, IPV6,
+					  TRAP_TO_CPU, false),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_CONTROL(IPV6_ROUTER_SOLICIT, IPV6, TRAP),
+		.listeners_arr = {
+			MLXSW_SP_RXL_MARK(L3_IPV6_ROUTER_SOLICITATION, IPV6,
+					  TRAP_TO_CPU, false),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_CONTROL(IPV6_ROUTER_ADVERT, IPV6, TRAP),
+		.listeners_arr = {
+			MLXSW_SP_RXL_MARK(L3_IPV6_ROUTER_ADVERTISEMENT, IPV6,
+					  TRAP_TO_CPU, false),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_CONTROL(IPV6_REDIRECT, IPV6, TRAP),
+		.listeners_arr = {
+			MLXSW_SP_RXL_MARK(L3_IPV6_REDIRECTION, IPV6,
+					  TRAP_TO_CPU, false),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_CONTROL(PTP_EVENT, PTP_EVENT, TRAP),
+		.listeners_arr = {
+			MLXSW_RXL(mlxsw_sp_rx_ptp_listener, PTP0, TRAP_TO_CPU,
+				  false, SP_PTP0, DISCARD),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_CONTROL(PTP_GENERAL, PTP_GENERAL, TRAP),
+		.listeners_arr = {
+			MLXSW_SP_RXL_NO_MARK(PTP1, PTP1, TRAP_TO_CPU, false),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_CONTROL(FLOW_ACTION_TRAP, ACL_TRAP, TRAP),
+		.listeners_arr = {
+			MLXSW_SP_RXL_NO_MARK(ACL0, FLOW_LOGGING, TRAP_TO_CPU,
+					     false),
+		},
+	},
+	{
+		.trap = MLXSW_SP_TRAP_DROP(BLACKHOLE_NEXTHOP, L3_DROPS),
+		.listeners_arr = {
+			MLXSW_SP_RXL_DISCARD(ROUTER3, L3_DISCARDS),
+		},
+	},
+};
+
+static struct mlxsw_sp_trap_policer_item *
+mlxsw_sp_trap_policer_item_lookup(struct mlxsw_sp *mlxsw_sp, u32 id)
+{
+	struct mlxsw_sp_trap *trap = mlxsw_sp->trap;
+	int i;
+
+	for (i = 0; i < trap->policers_count; i++) {
+		if (trap->policer_items_arr[i].policer.id == id)
+			return &trap->policer_items_arr[i];
+	}
+
+	return NULL;
+}
+
+static struct mlxsw_sp_trap_group_item *
+mlxsw_sp_trap_group_item_lookup(struct mlxsw_sp *mlxsw_sp, u16 id)
+{
+	struct mlxsw_sp_trap *trap = mlxsw_sp->trap;
+	int i;
+
+	for (i = 0; i < trap->groups_count; i++) {
+		if (trap->group_items_arr[i].group.id == id)
+			return &trap->group_items_arr[i];
+	}
+
+	return NULL;
+}
+
+static struct mlxsw_sp_trap_item *
+mlxsw_sp_trap_item_lookup(struct mlxsw_sp *mlxsw_sp, u16 id)
+{
+	struct mlxsw_sp_trap *trap = mlxsw_sp->trap;
+	int i;
+
+	for (i = 0; i < trap->traps_count; i++) {
+		if (trap->trap_items_arr[i].trap.id == id)
+			return &trap->trap_items_arr[i];
+	}
+
+	return NULL;
+}
+
+static int mlxsw_sp_trap_cpu_policers_set(struct mlxsw_sp *mlxsw_sp)
+{
+	struct mlxsw_sp_trap *trap = mlxsw_sp->trap;
+	char qpcr_pl[MLXSW_REG_QPCR_LEN];
+	u16 hw_id;
+
+	/* The purpose of "thin" policer is to drop as many packets
+	 * as possible. The dummy group is using it.
+	 */
+	hw_id = find_first_zero_bit(trap->policers_usage, trap->max_policers);
+	if (WARN_ON(hw_id == trap->max_policers))
+		return -ENOBUFS;
+
+	__set_bit(hw_id, trap->policers_usage);
+	trap->thin_policer_hw_id = hw_id;
+	mlxsw_reg_qpcr_pack(qpcr_pl, hw_id, MLXSW_REG_QPCR_IR_UNITS_M,
+			    false, 1, 4);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(qpcr), qpcr_pl);
+}
+
+static int mlxsw_sp_trap_dummy_group_init(struct mlxsw_sp *mlxsw_sp)
+{
+	char htgt_pl[MLXSW_REG_HTGT_LEN];
+
+	mlxsw_reg_htgt_pack(htgt_pl, MLXSW_REG_HTGT_TRAP_GROUP_SP_DUMMY,
+			    mlxsw_sp->trap->thin_policer_hw_id, 0, 1);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(htgt), htgt_pl);
+}
+
+static int mlxsw_sp_trap_policer_items_arr_init(struct mlxsw_sp *mlxsw_sp)
+{
+	size_t arr_size = ARRAY_SIZE(mlxsw_sp_trap_policer_items_arr);
+	size_t elem_size = sizeof(struct mlxsw_sp_trap_policer_item);
+	struct mlxsw_sp_trap *trap = mlxsw_sp->trap;
+	size_t free_policers = 0;
+	u32 last_id;
+	int i;
+
+	for_each_clear_bit(i, trap->policers_usage, trap->max_policers)
+		free_policers++;
+
+	if (arr_size > free_policers) {
+		dev_err(mlxsw_sp->bus_info->dev, "Exceeded number of supported packet trap policers\n");
+		return -ENOBUFS;
+	}
+
+	trap->policer_items_arr = kcalloc(free_policers, elem_size, GFP_KERNEL);
+	if (!trap->policer_items_arr)
+		return -ENOMEM;
+
+	trap->policers_count = free_policers;
+
+	/* Initialize policer items array with pre-defined policers. */
+	memcpy(trap->policer_items_arr, mlxsw_sp_trap_policer_items_arr,
+	       elem_size * arr_size);
+
+	/* Initialize policer items array with the rest of the available
+	 * policers.
+	 */
+	last_id = mlxsw_sp_trap_policer_items_arr[arr_size - 1].policer.id;
+	for (i = arr_size; i < trap->policers_count; i++) {
+		const struct mlxsw_sp_trap_policer_item *policer_item;
+
+		/* Use parameters set for first policer and override
+		 * relevant ones.
+		 */
+		policer_item = &mlxsw_sp_trap_policer_items_arr[0];
+		trap->policer_items_arr[i] = *policer_item;
+		trap->policer_items_arr[i].policer.id = ++last_id;
+		trap->policer_items_arr[i].policer.init_rate = 1;
+		trap->policer_items_arr[i].policer.init_burst = 16;
+	}
+
+	return 0;
+}
+
+static void mlxsw_sp_trap_policer_items_arr_fini(struct mlxsw_sp *mlxsw_sp)
+{
+	kfree(mlxsw_sp->trap->policer_items_arr);
+}
+
+static int mlxsw_sp_trap_policers_init(struct mlxsw_sp *mlxsw_sp)
+{
+	struct devlink *devlink = priv_to_devlink(mlxsw_sp->core);
+	const struct mlxsw_sp_trap_policer_item *policer_item;
+	struct mlxsw_sp_trap *trap = mlxsw_sp->trap;
+	int err, i;
+
+	err = mlxsw_sp_trap_policer_items_arr_init(mlxsw_sp);
+	if (err)
+		return err;
+
+	for (i = 0; i < trap->policers_count; i++) {
+		policer_item = &trap->policer_items_arr[i];
+		err = devl_trap_policers_register(devlink,
+						  &policer_item->policer, 1);
+		if (err)
+			goto err_trap_policer_register;
+	}
+
+	return 0;
+
+err_trap_policer_register:
+	for (i--; i >= 0; i--) {
+		policer_item = &trap->policer_items_arr[i];
+		devl_trap_policers_unregister(devlink,
+					      &policer_item->policer, 1);
+	}
+	mlxsw_sp_trap_policer_items_arr_fini(mlxsw_sp);
+	return err;
+}
+
+static void mlxsw_sp_trap_policers_fini(struct mlxsw_sp *mlxsw_sp)
+{
+	struct devlink *devlink = priv_to_devlink(mlxsw_sp->core);
+	const struct mlxsw_sp_trap_policer_item *policer_item;
+	struct mlxsw_sp_trap *trap = mlxsw_sp->trap;
+	int i;
+
+	for (i = trap->policers_count - 1; i >= 0; i--) {
+		policer_item = &trap->policer_items_arr[i];
+		devl_trap_policers_unregister(devlink,
+					      &policer_item->policer, 1);
+	}
+	mlxsw_sp_trap_policer_items_arr_fini(mlxsw_sp);
+}
+
+static int mlxsw_sp_trap_group_items_arr_init(struct mlxsw_sp *mlxsw_sp)
+{
+	size_t common_groups_count = ARRAY_SIZE(mlxsw_sp_trap_group_items_arr);
+	const struct mlxsw_sp_trap_group_item *spec_group_items_arr;
+	size_t elem_size = sizeof(struct mlxsw_sp_trap_group_item);
+	struct mlxsw_sp_trap *trap = mlxsw_sp->trap;
+	size_t groups_count, spec_groups_count;
+	int err;
+
+	err = mlxsw_sp->trap_ops->groups_init(mlxsw_sp, &spec_group_items_arr,
+					      &spec_groups_count);
+	if (err)
+		return err;
+
+	/* The group items array is created by concatenating the common trap
+	 * group items and the ASIC-specific trap group items.
+	 */
+	groups_count = common_groups_count + spec_groups_count;
+	trap->group_items_arr = kcalloc(groups_count, elem_size, GFP_KERNEL);
+	if (!trap->group_items_arr)
+		return -ENOMEM;
+
+	memcpy(trap->group_items_arr, mlxsw_sp_trap_group_items_arr,
+	       elem_size * common_groups_count);
+	memcpy(trap->group_items_arr + common_groups_count,
+	       spec_group_items_arr, elem_size * spec_groups_count);
+
+	trap->groups_count = groups_count;
+
+	return 0;
+}
+
+static void mlxsw_sp_trap_group_items_arr_fini(struct mlxsw_sp *mlxsw_sp)
+{
+	kfree(mlxsw_sp->trap->group_items_arr);
+}
+
+static int mlxsw_sp_trap_groups_init(struct mlxsw_sp *mlxsw_sp)
+{
+	struct devlink *devlink = priv_to_devlink(mlxsw_sp->core);
+	const struct mlxsw_sp_trap_group_item *group_item;
+	struct mlxsw_sp_trap *trap = mlxsw_sp->trap;
+	int err, i;
+
+	err = mlxsw_sp_trap_group_items_arr_init(mlxsw_sp);
+	if (err)
+		return err;
+
+	for (i = 0; i < trap->groups_count; i++) {
+		group_item = &trap->group_items_arr[i];
+		err = devl_trap_groups_register(devlink, &group_item->group, 1);
+		if (err)
+			goto err_trap_group_register;
+	}
+
+	return 0;
+
+err_trap_group_register:
+	for (i--; i >= 0; i--) {
+		group_item = &trap->group_items_arr[i];
+		devl_trap_groups_unregister(devlink, &group_item->group, 1);
+	}
+	mlxsw_sp_trap_group_items_arr_fini(mlxsw_sp);
+	return err;
+}
+
+static void mlxsw_sp_trap_groups_fini(struct mlxsw_sp *mlxsw_sp)
+{
+	struct devlink *devlink = priv_to_devlink(mlxsw_sp->core);
+	struct mlxsw_sp_trap *trap = mlxsw_sp->trap;
+	int i;
+
+	for (i = trap->groups_count - 1; i >= 0; i--) {
+		const struct mlxsw_sp_trap_group_item *group_item;
+
+		group_item = &trap->group_items_arr[i];
+		devl_trap_groups_unregister(devlink, &group_item->group, 1);
+	}
+	mlxsw_sp_trap_group_items_arr_fini(mlxsw_sp);
+}
+
+static bool
+mlxsw_sp_trap_listener_is_valid(const struct mlxsw_listener *listener)
+{
+	return listener->trap_id != 0;
+}
+
+static int mlxsw_sp_trap_items_arr_init(struct mlxsw_sp *mlxsw_sp)
+{
+	size_t common_traps_count = ARRAY_SIZE(mlxsw_sp_trap_items_arr);
+	const struct mlxsw_sp_trap_item *spec_trap_items_arr;
+	size_t elem_size = sizeof(struct mlxsw_sp_trap_item);
+	struct mlxsw_sp_trap *trap = mlxsw_sp->trap;
+	size_t traps_count, spec_traps_count;
+	int err;
+
+	err = mlxsw_sp->trap_ops->traps_init(mlxsw_sp, &spec_trap_items_arr,
+					     &spec_traps_count);
+	if (err)
+		return err;
+
+	/* The trap items array is created by concatenating the common trap
+	 * items and the ASIC-specific trap items.
+	 */
+	traps_count = common_traps_count + spec_traps_count;
+	trap->trap_items_arr = kcalloc(traps_count, elem_size, GFP_KERNEL);
+	if (!trap->trap_items_arr)
+		return -ENOMEM;
+
+	memcpy(trap->trap_items_arr, mlxsw_sp_trap_items_arr,
+	       elem_size * common_traps_count);
+	memcpy(trap->trap_items_arr + common_traps_count,
+	       spec_trap_items_arr, elem_size * spec_traps_count);
+
+	trap->traps_count = traps_count;
+
+	return 0;
+}
+
+static void mlxsw_sp_trap_items_arr_fini(struct mlxsw_sp *mlxsw_sp)
+{
+	kfree(mlxsw_sp->trap->trap_items_arr);
+}
+
+static int mlxsw_sp_traps_init(struct mlxsw_sp *mlxsw_sp)
+{
+	struct devlink *devlink = priv_to_devlink(mlxsw_sp->core);
+	struct mlxsw_sp_trap *trap = mlxsw_sp->trap;
+	const struct mlxsw_sp_trap_item *trap_item;
+	int err, i;
+
+	err = mlxsw_sp_trap_items_arr_init(mlxsw_sp);
+	if (err)
+		return err;
+
+	for (i = 0; i < trap->traps_count; i++) {
+		trap_item = &trap->trap_items_arr[i];
+		err = devl_traps_register(devlink, &trap_item->trap, 1,
+					  mlxsw_sp);
+		if (err)
+			goto err_trap_register;
+	}
+
+	return 0;
+
+err_trap_register:
+	for (i--; i >= 0; i--) {
+		trap_item = &trap->trap_items_arr[i];
+		devl_traps_unregister(devlink, &trap_item->trap, 1);
+	}
+	mlxsw_sp_trap_items_arr_fini(mlxsw_sp);
+	return err;
+}
+
+static void mlxsw_sp_traps_fini(struct mlxsw_sp *mlxsw_sp)
+{
+	struct devlink *devlink = priv_to_devlink(mlxsw_sp->core);
+	struct mlxsw_sp_trap *trap = mlxsw_sp->trap;
+	int i;
+
+	for (i = trap->traps_count - 1; i >= 0; i--) {
+		const struct mlxsw_sp_trap_item *trap_item;
+
+		trap_item = &trap->trap_items_arr[i];
+		devl_traps_unregister(devlink, &trap_item->trap, 1);
+	}
+	mlxsw_sp_trap_items_arr_fini(mlxsw_sp);
+}
+
+int mlxsw_sp_devlink_traps_init(struct mlxsw_sp *mlxsw_sp)
+{
+	int err;
+
+	err = mlxsw_sp_trap_cpu_policers_set(mlxsw_sp);
+	if (err)
+		return err;
+
+	err = mlxsw_sp_trap_dummy_group_init(mlxsw_sp);
+	if (err)
+		return err;
+
+	err = mlxsw_sp_trap_policers_init(mlxsw_sp);
+	if (err)
+		return err;
+
+	err = mlxsw_sp_trap_groups_init(mlxsw_sp);
+	if (err)
+		goto err_trap_groups_init;
+
+	err = mlxsw_sp_traps_init(mlxsw_sp);
+	if (err)
+		goto err_traps_init;
+
+	return 0;
+
+err_traps_init:
+	mlxsw_sp_trap_groups_fini(mlxsw_sp);
+err_trap_groups_init:
+	mlxsw_sp_trap_policers_fini(mlxsw_sp);
+	return err;
+}
+
+void mlxsw_sp_devlink_traps_fini(struct mlxsw_sp *mlxsw_sp)
+{
+	mlxsw_sp_traps_fini(mlxsw_sp);
+	mlxsw_sp_trap_groups_fini(mlxsw_sp);
+	mlxsw_sp_trap_policers_fini(mlxsw_sp);
+}
+
+int mlxsw_sp_trap_init(struct mlxsw_core *mlxsw_core,
+		       const struct devlink_trap *trap, void *trap_ctx)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_core_driver_priv(mlxsw_core);
+	const struct mlxsw_sp_trap_item *trap_item;
+	int i;
+
+	trap_item = mlxsw_sp_trap_item_lookup(mlxsw_sp, trap->id);
+	if (WARN_ON(!trap_item))
+		return -EINVAL;
+
+	for (i = 0; i < MLXSW_SP_TRAP_LISTENERS_MAX; i++) {
+		const struct mlxsw_listener *listener;
+		int err;
+
+		listener = &trap_item->listeners_arr[i];
+		if (!mlxsw_sp_trap_listener_is_valid(listener))
+			continue;
+		err = mlxsw_core_trap_register(mlxsw_core, listener, trap_ctx);
+		if (err)
+			return err;
+	}
+
+	return 0;
+}
+
+void mlxsw_sp_trap_fini(struct mlxsw_core *mlxsw_core,
+			const struct devlink_trap *trap, void *trap_ctx)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_core_driver_priv(mlxsw_core);
+	const struct mlxsw_sp_trap_item *trap_item;
+	int i;
+
+	trap_item = mlxsw_sp_trap_item_lookup(mlxsw_sp, trap->id);
+	if (WARN_ON(!trap_item))
+		return;
+
+	for (i = MLXSW_SP_TRAP_LISTENERS_MAX - 1; i >= 0; i--) {
+		const struct mlxsw_listener *listener;
+
+		listener = &trap_item->listeners_arr[i];
+		if (!mlxsw_sp_trap_listener_is_valid(listener))
+			continue;
+		mlxsw_core_trap_unregister(mlxsw_core, listener, trap_ctx);
+	}
+}
+
+int mlxsw_sp_trap_action_set(struct mlxsw_core *mlxsw_core,
+			     const struct devlink_trap *trap,
+			     enum devlink_trap_action action,
+			     struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_core_driver_priv(mlxsw_core);
+	const struct mlxsw_sp_trap_item *trap_item;
+	int i;
+
+	trap_item = mlxsw_sp_trap_item_lookup(mlxsw_sp, trap->id);
+	if (WARN_ON(!trap_item))
+		return -EINVAL;
+
+	if (trap_item->is_source) {
+		NL_SET_ERR_MSG_MOD(extack, "Changing the action of source traps is not supported");
+		return -EOPNOTSUPP;
+	}
+
+	for (i = 0; i < MLXSW_SP_TRAP_LISTENERS_MAX; i++) {
+		const struct mlxsw_listener *listener;
+		bool enabled;
+		int err;
+
+		listener = &trap_item->listeners_arr[i];
+		if (!mlxsw_sp_trap_listener_is_valid(listener))
+			continue;
+
+		switch (action) {
+		case DEVLINK_TRAP_ACTION_DROP:
+			enabled = false;
+			break;
+		case DEVLINK_TRAP_ACTION_TRAP:
+			enabled = true;
+			break;
+		default:
+			return -EINVAL;
+		}
+		err = mlxsw_core_trap_state_set(mlxsw_core, listener, enabled);
+		if (err)
+			return err;
+	}
+
+	return 0;
+}
+
+static int
+__mlxsw_sp_trap_group_init(struct mlxsw_core *mlxsw_core,
+			   const struct devlink_trap_group *group,
+			   u32 policer_id, struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_core_driver_priv(mlxsw_core);
+	u16 hw_policer_id = MLXSW_REG_HTGT_INVALID_POLICER;
+	const struct mlxsw_sp_trap_group_item *group_item;
+	char htgt_pl[MLXSW_REG_HTGT_LEN];
+
+	group_item = mlxsw_sp_trap_group_item_lookup(mlxsw_sp, group->id);
+	if (WARN_ON(!group_item))
+		return -EINVAL;
+
+	if (group_item->fixed_policer && policer_id != group->init_policer_id) {
+		NL_SET_ERR_MSG_MOD(extack, "Changing the policer binding of this group is not supported");
+		return -EOPNOTSUPP;
+	}
+
+	if (policer_id) {
+		struct mlxsw_sp_trap_policer_item *policer_item;
+
+		policer_item = mlxsw_sp_trap_policer_item_lookup(mlxsw_sp,
+								 policer_id);
+		if (WARN_ON(!policer_item))
+			return -EINVAL;
+		hw_policer_id = policer_item->hw_id;
+	}
+
+	mlxsw_reg_htgt_pack(htgt_pl, group_item->hw_group_id, hw_policer_id,
+			    group_item->priority, group_item->priority);
+	return mlxsw_reg_write(mlxsw_core, MLXSW_REG(htgt), htgt_pl);
+}
+
+int mlxsw_sp_trap_group_init(struct mlxsw_core *mlxsw_core,
+			     const struct devlink_trap_group *group)
+{
+	return __mlxsw_sp_trap_group_init(mlxsw_core, group,
+					  group->init_policer_id, NULL);
+}
+
+int mlxsw_sp_trap_group_set(struct mlxsw_core *mlxsw_core,
+			    const struct devlink_trap_group *group,
+			    const struct devlink_trap_policer *policer,
+			    struct netlink_ext_ack *extack)
+{
+	u32 policer_id = policer ? policer->id : 0;
+
+	return __mlxsw_sp_trap_group_init(mlxsw_core, group, policer_id,
+					  extack);
+}
+
+static int
+mlxsw_sp_trap_policer_item_init(struct mlxsw_sp *mlxsw_sp,
+				struct mlxsw_sp_trap_policer_item *policer_item)
+{
+	struct mlxsw_sp_trap *trap = mlxsw_sp->trap;
+	u16 hw_id;
+
+	/* We should be able to allocate a policer because the number of
+	 * policers we registered with devlink is in according with the number
+	 * of available policers.
+	 */
+	hw_id = find_first_zero_bit(trap->policers_usage, trap->max_policers);
+	if (WARN_ON(hw_id == trap->max_policers))
+		return -ENOBUFS;
+
+	__set_bit(hw_id, trap->policers_usage);
+	policer_item->hw_id = hw_id;
+
+	return 0;
+}
+
+static void
+mlxsw_sp_trap_policer_item_fini(struct mlxsw_sp *mlxsw_sp,
+				struct mlxsw_sp_trap_policer_item *policer_item)
+{
+	__clear_bit(policer_item->hw_id, mlxsw_sp->trap->policers_usage);
+}
+
+static int mlxsw_sp_trap_policer_bs(u64 burst, u8 *p_burst_size,
+				    struct netlink_ext_ack *extack)
+{
+	int bs = fls64(burst) - 1;
+
+	if (burst != (BIT_ULL(bs))) {
+		NL_SET_ERR_MSG_MOD(extack, "Policer burst size is not power of two");
+		return -EINVAL;
+	}
+
+	*p_burst_size = bs;
+
+	return 0;
+}
+
+static int __mlxsw_sp_trap_policer_set(struct mlxsw_sp *mlxsw_sp, u16 hw_id,
+				       u64 rate, u64 burst, bool clear_counter,
+				       struct netlink_ext_ack *extack)
+{
+	char qpcr_pl[MLXSW_REG_QPCR_LEN];
+	u8 burst_size;
+	int err;
+
+	err = mlxsw_sp_trap_policer_bs(burst, &burst_size, extack);
+	if (err)
+		return err;
+
+	mlxsw_reg_qpcr_pack(qpcr_pl, hw_id, MLXSW_REG_QPCR_IR_UNITS_M, false,
+			    rate, burst_size);
+	mlxsw_reg_qpcr_clear_counter_set(qpcr_pl, clear_counter);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(qpcr), qpcr_pl);
+}
+
+int mlxsw_sp_trap_policer_init(struct mlxsw_core *mlxsw_core,
+			       const struct devlink_trap_policer *policer)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_core_driver_priv(mlxsw_core);
+	struct mlxsw_sp_trap_policer_item *policer_item;
+	int err;
+
+	policer_item = mlxsw_sp_trap_policer_item_lookup(mlxsw_sp, policer->id);
+	if (WARN_ON(!policer_item))
+		return -EINVAL;
+
+	err = mlxsw_sp_trap_policer_item_init(mlxsw_sp, policer_item);
+	if (err)
+		return err;
+
+	err = __mlxsw_sp_trap_policer_set(mlxsw_sp, policer_item->hw_id,
+					  policer->init_rate,
+					  policer->init_burst, true, NULL);
+	if (err)
+		goto err_trap_policer_set;
+
+	return 0;
+
+err_trap_policer_set:
+	mlxsw_sp_trap_policer_item_fini(mlxsw_sp, policer_item);
+	return err;
+}
+
+void mlxsw_sp_trap_policer_fini(struct mlxsw_core *mlxsw_core,
+				const struct devlink_trap_policer *policer)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_core_driver_priv(mlxsw_core);
+	struct mlxsw_sp_trap_policer_item *policer_item;
+
+	policer_item = mlxsw_sp_trap_policer_item_lookup(mlxsw_sp, policer->id);
+	if (WARN_ON(!policer_item))
+		return;
+
+	mlxsw_sp_trap_policer_item_fini(mlxsw_sp, policer_item);
+}
+
+int mlxsw_sp_trap_policer_set(struct mlxsw_core *mlxsw_core,
+			      const struct devlink_trap_policer *policer,
+			      u64 rate, u64 burst,
+			      struct netlink_ext_ack *extack)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_core_driver_priv(mlxsw_core);
+	struct mlxsw_sp_trap_policer_item *policer_item;
+
+	policer_item = mlxsw_sp_trap_policer_item_lookup(mlxsw_sp, policer->id);
+	if (WARN_ON(!policer_item))
+		return -EINVAL;
+
+	return __mlxsw_sp_trap_policer_set(mlxsw_sp, policer_item->hw_id,
+					   rate, burst, false, extack);
+}
+
+int
+mlxsw_sp_trap_policer_counter_get(struct mlxsw_core *mlxsw_core,
+				  const struct devlink_trap_policer *policer,
+				  u64 *p_drops)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_core_driver_priv(mlxsw_core);
+	struct mlxsw_sp_trap_policer_item *policer_item;
+	char qpcr_pl[MLXSW_REG_QPCR_LEN];
+	int err;
+
+	policer_item = mlxsw_sp_trap_policer_item_lookup(mlxsw_sp, policer->id);
+	if (WARN_ON(!policer_item))
+		return -EINVAL;
+
+	mlxsw_reg_qpcr_pack(qpcr_pl, policer_item->hw_id,
+			    MLXSW_REG_QPCR_IR_UNITS_M, false, 0, 0);
+	err = mlxsw_reg_query(mlxsw_sp->core, MLXSW_REG(qpcr), qpcr_pl);
+	if (err)
+		return err;
+
+	*p_drops = mlxsw_reg_qpcr_violate_count_get(qpcr_pl);
+
+	return 0;
+}
+
+int mlxsw_sp_trap_group_policer_hw_id_get(struct mlxsw_sp *mlxsw_sp, u16 id,
+					  bool *p_enabled, u16 *p_hw_id)
+{
+	struct mlxsw_sp_trap_policer_item *pol_item;
+	struct mlxsw_sp_trap_group_item *gr_item;
+	u32 pol_id;
+
+	gr_item = mlxsw_sp_trap_group_item_lookup(mlxsw_sp, id);
+	if (!gr_item)
+		return -ENOENT;
+
+	pol_id = gr_item->group.init_policer_id;
+	if (!pol_id) {
+		*p_enabled = false;
+		return 0;
+	}
+
+	pol_item = mlxsw_sp_trap_policer_item_lookup(mlxsw_sp, pol_id);
+	if (WARN_ON(!pol_item))
+		return -ENOENT;
+
+	*p_enabled = true;
+	*p_hw_id = pol_item->hw_id;
+	return 0;
+}
+
+static const struct mlxsw_sp_trap_group_item
+mlxsw_sp1_trap_group_items_arr[] = {
+	{
+		.group = DEVLINK_TRAP_GROUP_GENERIC(ACL_SAMPLE, 0),
+		.hw_group_id = MLXSW_REG_HTGT_TRAP_GROUP_SP_PKT_SAMPLE,
+		.priority = 0,
+	},
+};
+
+static const struct mlxsw_sp_trap_item
+mlxsw_sp1_trap_items_arr[] = {
+	{
+		.trap = MLXSW_SP_TRAP_CONTROL(FLOW_ACTION_SAMPLE, ACL_SAMPLE,
+					      MIRROR),
+		.listeners_arr = {
+			MLXSW_RXL(mlxsw_sp_rx_sample_listener, PKT_SAMPLE,
+				  MIRROR_TO_CPU, false, SP_PKT_SAMPLE, DISCARD),
+		},
+	},
+};
+
+static int
+mlxsw_sp1_trap_groups_init(struct mlxsw_sp *mlxsw_sp,
+			   const struct mlxsw_sp_trap_group_item **arr,
+			   size_t *p_groups_count)
+{
+	*arr = mlxsw_sp1_trap_group_items_arr;
+	*p_groups_count = ARRAY_SIZE(mlxsw_sp1_trap_group_items_arr);
+
+	return 0;
+}
+
+static int mlxsw_sp1_traps_init(struct mlxsw_sp *mlxsw_sp,
+				const struct mlxsw_sp_trap_item **arr,
+				size_t *p_traps_count)
+{
+	*arr = mlxsw_sp1_trap_items_arr;
+	*p_traps_count = ARRAY_SIZE(mlxsw_sp1_trap_items_arr);
+
+	return 0;
+}
+
+const struct mlxsw_sp_trap_ops mlxsw_sp1_trap_ops = {
+	.groups_init = mlxsw_sp1_trap_groups_init,
+	.traps_init = mlxsw_sp1_traps_init,
+};
+
+static const struct mlxsw_sp_trap_group_item
+mlxsw_sp2_trap_group_items_arr[] = {
+	{
+		.group = DEVLINK_TRAP_GROUP_GENERIC(BUFFER_DROPS, 20),
+		.hw_group_id = MLXSW_REG_HTGT_TRAP_GROUP_SP_BUFFER_DISCARDS,
+		.priority = 0,
+		.fixed_policer = true,
+	},
+	{
+		.group = DEVLINK_TRAP_GROUP_GENERIC(ACL_SAMPLE, 0),
+		.hw_group_id = MLXSW_REG_HTGT_TRAP_GROUP_SP_PKT_SAMPLE,
+		.priority = 0,
+		.fixed_policer = true,
+	},
+};
+
+static const struct mlxsw_sp_trap_item
+mlxsw_sp2_trap_items_arr[] = {
+	{
+		.trap = MLXSW_SP_TRAP_BUFFER_DROP(EARLY_DROP),
+		.listeners_arr = {
+			MLXSW_SP_RXL_BUFFER_DISCARD(INGRESS_WRED),
+		},
+		.is_source = true,
+	},
+	{
+		.trap = MLXSW_SP_TRAP_CONTROL(FLOW_ACTION_SAMPLE, ACL_SAMPLE,
+					      MIRROR),
+		.listeners_arr = {
+			MLXSW_RXL_MIRROR(mlxsw_sp_rx_sample_listener, 1,
+					 SP_PKT_SAMPLE,
+					 MLXSW_SP_MIRROR_REASON_INGRESS),
+			MLXSW_RXL_MIRROR(mlxsw_sp_rx_sample_tx_listener, 1,
+					 SP_PKT_SAMPLE,
+					 MLXSW_SP_MIRROR_REASON_EGRESS),
+			MLXSW_RXL_MIRROR(mlxsw_sp_rx_sample_acl_listener, 1,
+					 SP_PKT_SAMPLE,
+					 MLXSW_SP_MIRROR_REASON_POLICY_ENGINE),
+		},
+	},
+};
+
+static int
+mlxsw_sp2_trap_groups_init(struct mlxsw_sp *mlxsw_sp,
+			   const struct mlxsw_sp_trap_group_item **arr,
+			   size_t *p_groups_count)
+{
+	*arr = mlxsw_sp2_trap_group_items_arr;
+	*p_groups_count = ARRAY_SIZE(mlxsw_sp2_trap_group_items_arr);
+
+	return 0;
+}
+
+static int mlxsw_sp2_traps_init(struct mlxsw_sp *mlxsw_sp,
+				const struct mlxsw_sp_trap_item **arr,
+				size_t *p_traps_count)
+{
+	*arr = mlxsw_sp2_trap_items_arr;
+	*p_traps_count = ARRAY_SIZE(mlxsw_sp2_trap_items_arr);
+
+	return 0;
+}
+
+const struct mlxsw_sp_trap_ops mlxsw_sp2_trap_ops = {
+	.groups_init = mlxsw_sp2_trap_groups_init,
+	.traps_init = mlxsw_sp2_traps_init,
+};
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_trap.h b/drivers/net/ethernet/mellanox/mlxsw/spectrum_trap.h
new file mode 100644
index 000000000..b8df684be
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_trap.h
@@ -0,0 +1,38 @@
+/* SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0 */
+/* Copyright (c) 2020 Mellanox Technologies. All rights reserved */
+
+#ifndef _MLXSW_SPECTRUM_TRAP_H
+#define _MLXSW_SPECTRUM_TRAP_H
+
+#include <linux/list.h>
+#include <net/devlink.h>
+
+struct mlxsw_sp_trap {
+	struct mlxsw_sp_trap_policer_item *policer_items_arr;
+	size_t policers_count; /* Number of registered policers */
+
+	struct mlxsw_sp_trap_group_item *group_items_arr;
+	size_t groups_count; /* Number of registered groups */
+
+	struct mlxsw_sp_trap_item *trap_items_arr;
+	size_t traps_count; /* Number of registered traps */
+
+	u16 thin_policer_hw_id;
+
+	u64 max_policers;
+	unsigned long policers_usage[]; /* Usage bitmap */
+};
+
+struct mlxsw_sp_trap_ops {
+	int (*groups_init)(struct mlxsw_sp *mlxsw_sp,
+			   const struct mlxsw_sp_trap_group_item **arr,
+			   size_t *p_groups_count);
+	int (*traps_init)(struct mlxsw_sp *mlxsw_sp,
+			  const struct mlxsw_sp_trap_item **arr,
+			  size_t *p_traps_count);
+};
+
+extern const struct mlxsw_sp_trap_ops mlxsw_sp1_trap_ops;
+extern const struct mlxsw_sp_trap_ops mlxsw_sp2_trap_ops;
+
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlxsw/trap.h b/drivers/net/ethernet/mellanox/mlxsw/trap.h
new file mode 100644
index 000000000..8da169663
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/trap.h
@@ -0,0 +1,144 @@
+/* SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0 */
+/* Copyright (c) 2015-2018 Mellanox Technologies. All rights reserved */
+
+#ifndef _MLXSW_TRAP_H
+#define _MLXSW_TRAP_H
+
+enum {
+	/* Ethernet EMAD and FDB miss */
+	MLXSW_TRAP_ID_FDB_MC = 0x01,
+	MLXSW_TRAP_ID_ETHEMAD = 0x05,
+	/* L2 traps for specific packet types */
+	MLXSW_TRAP_ID_STP = 0x10,
+	MLXSW_TRAP_ID_LACP = 0x11,
+	MLXSW_TRAP_ID_EAPOL = 0x12,
+	MLXSW_TRAP_ID_LLDP = 0x13,
+	MLXSW_TRAP_ID_MMRP = 0x14,
+	MLXSW_TRAP_ID_MVRP = 0x15,
+	MLXSW_TRAP_ID_RPVST = 0x16,
+	MLXSW_TRAP_ID_DHCP = 0x19,
+	MLXSW_TRAP_ID_PTP0 = 0x28,
+	MLXSW_TRAP_ID_PTP1 = 0x29,
+	MLXSW_TRAP_ID_IGMP_QUERY = 0x30,
+	MLXSW_TRAP_ID_IGMP_V1_REPORT = 0x31,
+	MLXSW_TRAP_ID_IGMP_V2_REPORT = 0x32,
+	MLXSW_TRAP_ID_IGMP_V2_LEAVE = 0x33,
+	MLXSW_TRAP_ID_IGMP_V3_REPORT = 0x34,
+	MLXSW_TRAP_ID_PKT_SAMPLE = 0x38,
+	MLXSW_TRAP_ID_FID_MISS = 0x3D,
+	MLXSW_TRAP_ID_DECAP_ECN0 = 0x40,
+	MLXSW_TRAP_ID_MTUERROR = 0x52,
+	MLXSW_TRAP_ID_TTLERROR = 0x53,
+	MLXSW_TRAP_ID_LBERROR = 0x54,
+	MLXSW_TRAP_ID_IPV4_OSPF = 0x55,
+	MLXSW_TRAP_ID_IPV4_PIM = 0x58,
+	MLXSW_TRAP_ID_IPV4_VRRP = 0x59,
+	MLXSW_TRAP_ID_RPF = 0x5C,
+	MLXSW_TRAP_ID_IP2ME = 0x5F,
+	MLXSW_TRAP_ID_IPV6_UNSPECIFIED_ADDRESS = 0x60,
+	MLXSW_TRAP_ID_IPV6_LINK_LOCAL_DEST = 0x61,
+	MLXSW_TRAP_ID_IPV6_LINK_LOCAL_SRC = 0x62,
+	MLXSW_TRAP_ID_IPV6_ALL_NODES_LINK = 0x63,
+	MLXSW_TRAP_ID_IPV6_OSPF = 0x64,
+	MLXSW_TRAP_ID_IPV6_MLDV12_LISTENER_QUERY = 0x65,
+	MLXSW_TRAP_ID_IPV6_MLDV1_LISTENER_REPORT = 0x66,
+	MLXSW_TRAP_ID_IPV6_MLDV1_LISTENER_DONE = 0x67,
+	MLXSW_TRAP_ID_IPV6_MLDV2_LISTENER_REPORT = 0x68,
+	MLXSW_TRAP_ID_IPV6_DHCP = 0x69,
+	MLXSW_TRAP_ID_IPV6_ALL_ROUTERS_LINK = 0x6F,
+	MLXSW_TRAP_ID_RTR_INGRESS0 = 0x70,
+	MLXSW_TRAP_ID_RTR_INGRESS1 = 0x71,
+	MLXSW_TRAP_ID_IPV6_PIM = 0x79,
+	MLXSW_TRAP_ID_IPV6_VRRP = 0x7A,
+	MLXSW_TRAP_ID_RTR_EGRESS0 = 0x80,
+	MLXSW_TRAP_ID_IPV4_BGP = 0x88,
+	MLXSW_TRAP_ID_IPV6_BGP = 0x89,
+	MLXSW_TRAP_ID_L3_IPV6_ROUTER_SOLICITATION = 0x8A,
+	MLXSW_TRAP_ID_L3_IPV6_ROUTER_ADVERTISEMENT = 0x8B,
+	MLXSW_TRAP_ID_L3_IPV6_NEIGHBOR_SOLICITATION = 0x8C,
+	MLXSW_TRAP_ID_L3_IPV6_NEIGHBOR_ADVERTISEMENT = 0x8D,
+	MLXSW_TRAP_ID_L3_IPV6_REDIRECTION = 0x8E,
+	MLXSW_TRAP_ID_IPV4_DHCP = 0x8F,
+	MLXSW_TRAP_ID_HOST_MISS_IPV4 = 0x90,
+	MLXSW_TRAP_ID_IPV6_MC_LINK_LOCAL_DEST = 0x91,
+	MLXSW_TRAP_ID_HOST_MISS_IPV6 = 0x92,
+	MLXSW_TRAP_ID_IPIP_DECAP_ERROR = 0xB1,
+	MLXSW_TRAP_ID_NVE_DECAP_ARP = 0xB8,
+	MLXSW_TRAP_ID_NVE_ENCAP_ARP = 0xBD,
+	MLXSW_TRAP_ID_IPV4_BFD = 0xD0,
+	MLXSW_TRAP_ID_IPV6_BFD = 0xD1,
+	MLXSW_TRAP_ID_ROUTER_ALERT_IPV4 = 0xD6,
+	MLXSW_TRAP_ID_ROUTER_ALERT_IPV6 = 0xD7,
+	MLXSW_TRAP_ID_ROUTER_ARPBC = 0xE0,
+	MLXSW_TRAP_ID_ROUTER_ARPUC = 0xE1,
+	MLXSW_TRAP_ID_DISCARD_NON_ROUTABLE = 0x11A,
+	MLXSW_TRAP_ID_DISCARD_ROUTER2 = 0x130,
+	MLXSW_TRAP_ID_DISCARD_ROUTER3 = 0x131,
+	MLXSW_TRAP_ID_DISCARD_ING_PACKET_SMAC_MC = 0x140,
+	MLXSW_TRAP_ID_DISCARD_ING_SWITCH_VTAG_ALLOW = 0x148,
+	MLXSW_TRAP_ID_DISCARD_ING_SWITCH_VLAN = 0x149,
+	MLXSW_TRAP_ID_DISCARD_ING_SWITCH_STP = 0x14A,
+	MLXSW_TRAP_ID_DISCARD_LOOKUP_SWITCH_UC = 0x150,
+	MLXSW_TRAP_ID_DISCARD_LOOKUP_SWITCH_MC_NULL = 0x151,
+	MLXSW_TRAP_ID_DISCARD_LOOKUP_SWITCH_LB = 0x152,
+	MLXSW_TRAP_ID_DISCARD_ING_ROUTER_NON_IP_PACKET = 0x160,
+	MLXSW_TRAP_ID_DISCARD_ING_ROUTER_UC_DIP_MC_DMAC = 0x161,
+	MLXSW_TRAP_ID_DISCARD_ING_ROUTER_DIP_LB = 0x162,
+	MLXSW_TRAP_ID_DISCARD_ING_ROUTER_SIP_MC = 0x163,
+	MLXSW_TRAP_ID_DISCARD_ING_ROUTER_SIP_CLASS_E = 0x164,
+	MLXSW_TRAP_ID_DISCARD_ING_ROUTER_SIP_LB = 0x165,
+	MLXSW_TRAP_ID_DISCARD_ING_ROUTER_CORRUPTED_IP_HDR = 0x167,
+	MLXSW_TRAP_ID_DISCARD_ING_ROUTER_MC_DMAC = 0x168,
+	MLXSW_TRAP_ID_DISCARD_ING_ROUTER_SIP_DIP = 0x169,
+	MLXSW_TRAP_ID_DISCARD_ING_ROUTER_IPV4_SIP_BC = 0x16A,
+	MLXSW_TRAP_ID_DISCARD_ING_ROUTER_IPV4_DIP_LOCAL_NET = 0x16B,
+	MLXSW_TRAP_ID_DISCARD_ING_ROUTER_DIP_LINK_LOCAL = 0x16C,
+	MLXSW_TRAP_ID_DISCARD_ROUTER_IRIF_EN = 0x178,
+	MLXSW_TRAP_ID_DISCARD_ROUTER_ERIF_EN = 0x179,
+	MLXSW_TRAP_ID_DISCARD_ROUTER_LPM4 = 0x17B,
+	MLXSW_TRAP_ID_DISCARD_ROUTER_LPM6 = 0x17C,
+	MLXSW_TRAP_ID_DISCARD_DEC_PKT = 0x188,
+	MLXSW_TRAP_ID_DISCARD_OVERLAY_SMAC_MC = 0x190,
+	MLXSW_TRAP_ID_DISCARD_IPV6_MC_DIP_RESERVED_SCOPE = 0x1B0,
+	MLXSW_TRAP_ID_DISCARD_IPV6_MC_DIP_INTERFACE_LOCAL_SCOPE = 0x1B1,
+	MLXSW_TRAP_ID_ACL0 = 0x1C0,
+	/* Multicast trap used for routes with trap action */
+	MLXSW_TRAP_ID_ACL1 = 0x1C1,
+	/* Multicast trap used for routes with trap-and-forward action */
+	MLXSW_TRAP_ID_ACL2 = 0x1C2,
+	MLXSW_TRAP_ID_DISCARD_INGRESS_ACL = 0x1C3,
+	MLXSW_TRAP_ID_DISCARD_EGRESS_ACL = 0x1C4,
+	MLXSW_TRAP_ID_MIRROR_SESSION0 = 0x220,
+	MLXSW_TRAP_ID_MIRROR_SESSION1 = 0x221,
+	MLXSW_TRAP_ID_MIRROR_SESSION2 = 0x222,
+	MLXSW_TRAP_ID_MIRROR_SESSION3 = 0x223,
+	MLXSW_TRAP_ID_MIRROR_SESSION4 = 0x224,
+	MLXSW_TRAP_ID_MIRROR_SESSION5 = 0x225,
+	MLXSW_TRAP_ID_MIRROR_SESSION6 = 0x226,
+	MLXSW_TRAP_ID_MIRROR_SESSION7 = 0x227,
+
+	MLXSW_TRAP_ID_MAX = 0x3FF,
+};
+
+enum mlxsw_event_trap_id {
+	/* Fatal Event generated by FW */
+	MLXSW_TRAP_ID_MFDE = 0x3,
+	/* Port Up/Down event generated by hardware */
+	MLXSW_TRAP_ID_PUDE = 0x8,
+	/* Port Module Plug/Unplug Event generated by hardware */
+	MLXSW_TRAP_ID_PMPE = 0x9,
+	/* Temperature Warning event generated by hardware */
+	MLXSW_TRAP_ID_MTWE = 0xC,
+	/* PTP Ingress FIFO has a new entry */
+	MLXSW_TRAP_ID_PTP_ING_FIFO = 0x2D,
+	/* PTP Egress FIFO has a new entry */
+	MLXSW_TRAP_ID_PTP_EGR_FIFO = 0x2E,
+	/* Downstream Device Status Change */
+	MLXSW_TRAP_ID_DSDSC = 0x321,
+	/* Binary Code Transfer Operation Executed Event */
+	MLXSW_TRAP_ID_BCTOE = 0x322,
+	/* Port mapping change */
+	MLXSW_TRAP_ID_PMLPE = 0x32E,
+};
+
+#endif /* _MLXSW_TRAP_H */
diff --git a/drivers/net/ethernet/mellanox/mlxsw/txheader.h b/drivers/net/ethernet/mellanox/mlxsw/txheader.h
new file mode 100644
index 000000000..da51dd9d5
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/txheader.h
@@ -0,0 +1,49 @@
+/* SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0 */
+/* Copyright (c) 2015-2018 Mellanox Technologies. All rights reserved */
+
+#ifndef _MLXSW_TXHEADER_H
+#define _MLXSW_TXHEADER_H
+
+#define MLXSW_TXHDR_LEN 0x10
+#define MLXSW_TXHDR_VERSION_0 0
+#define MLXSW_TXHDR_VERSION_1 1
+
+enum {
+	MLXSW_TXHDR_ETH_CTL,
+	MLXSW_TXHDR_ETH_DATA,
+};
+
+#define MLXSW_TXHDR_PROTO_ETH 1
+
+enum {
+	MLXSW_TXHDR_ETCLASS_0,
+	MLXSW_TXHDR_ETCLASS_1,
+	MLXSW_TXHDR_ETCLASS_2,
+	MLXSW_TXHDR_ETCLASS_3,
+	MLXSW_TXHDR_ETCLASS_4,
+	MLXSW_TXHDR_ETCLASS_5,
+	MLXSW_TXHDR_ETCLASS_6,
+	MLXSW_TXHDR_ETCLASS_7,
+};
+
+enum {
+	MLXSW_TXHDR_RDQ_OTHER,
+	MLXSW_TXHDR_RDQ_EMAD = 0x1f,
+};
+
+#define MLXSW_TXHDR_CTCLASS3 0
+#define MLXSW_TXHDR_CPU_SIG 0
+#define MLXSW_TXHDR_SIG 0xE0E0
+#define MLXSW_TXHDR_STCLASS_NONE 0
+
+enum {
+	MLXSW_TXHDR_NOT_EMAD,
+	MLXSW_TXHDR_EMAD,
+};
+
+enum {
+	MLXSW_TXHDR_TYPE_DATA,
+	MLXSW_TXHDR_TYPE_CONTROL = 6,
+};
+
+#endif