Adding upstream version 6.6.15.upstream/6.6.15

Signed-off-by: Daniel Baumann <daniel.baumann@progress-linux.org>

4 files changed, 1640 insertions, 0 deletions

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/ipoib/ethtool.c b/drivers/net/ethernet/mellanox/mlx5/core/ipoib/ethtool.c
new file mode 100644
index 0000000000..779d92b762
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/ipoib/ethtool.c
@@ -0,0 +1,296 @@
+/*
+ * Copyright (c) 2017, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "en.h"
+#include "ipoib.h"
+#include "en/fs_ethtool.h"
+
+static void mlx5i_get_drvinfo(struct net_device *dev,
+			      struct ethtool_drvinfo *drvinfo)
+{
+	struct mlx5e_priv *priv = mlx5i_epriv(dev);
+
+	mlx5e_ethtool_get_drvinfo(priv, drvinfo);
+	strscpy(drvinfo->driver, KBUILD_MODNAME "[ib_ipoib]",
+		sizeof(drvinfo->driver));
+}
+
+static void mlx5i_get_strings(struct net_device *dev, u32 stringset, u8 *data)
+{
+	struct mlx5e_priv *priv  = mlx5i_epriv(dev);
+
+	mlx5e_ethtool_get_strings(priv, stringset, data);
+}
+
+static int mlx5i_get_sset_count(struct net_device *dev, int sset)
+{
+	struct mlx5e_priv *priv = mlx5i_epriv(dev);
+
+	return mlx5e_ethtool_get_sset_count(priv, sset);
+}
+
+static void mlx5i_get_ethtool_stats(struct net_device *dev,
+				    struct ethtool_stats *stats,
+				    u64 *data)
+{
+	struct mlx5e_priv *priv = mlx5i_epriv(dev);
+
+	mlx5e_ethtool_get_ethtool_stats(priv, stats, data);
+}
+
+static int mlx5i_set_ringparam(struct net_device *dev,
+			       struct ethtool_ringparam *param,
+			       struct kernel_ethtool_ringparam *kernel_param,
+			       struct netlink_ext_ack *extack)
+{
+	struct mlx5e_priv *priv = mlx5i_epriv(dev);
+
+	return mlx5e_ethtool_set_ringparam(priv, param);
+}
+
+static void mlx5i_get_ringparam(struct net_device *dev,
+				struct ethtool_ringparam *param,
+				struct kernel_ethtool_ringparam *kernel_param,
+				struct netlink_ext_ack *extack)
+{
+	struct mlx5e_priv *priv = mlx5i_epriv(dev);
+
+	mlx5e_ethtool_get_ringparam(priv, param, kernel_param);
+}
+
+static int mlx5i_set_channels(struct net_device *dev,
+			      struct ethtool_channels *ch)
+{
+	struct mlx5i_priv *ipriv = netdev_priv(dev);
+	struct mlx5e_priv *epriv = mlx5i_epriv(dev);
+
+	/* rtnl lock protects from race between this ethtool op and sub
+	 * interface ndo_init/uninit.
+	 */
+	ASSERT_RTNL();
+	if (ipriv->num_sub_interfaces > 0) {
+		mlx5_core_warn(epriv->mdev,
+			       "can't change number of channels for interfaces with sub interfaces (%u)\n",
+			       ipriv->num_sub_interfaces);
+		return -EINVAL;
+	}
+
+	return mlx5e_ethtool_set_channels(epriv, ch);
+}
+
+static void mlx5i_get_channels(struct net_device *dev,
+			       struct ethtool_channels *ch)
+{
+	struct mlx5e_priv *priv = mlx5i_epriv(dev);
+
+	mlx5e_ethtool_get_channels(priv, ch);
+}
+
+static int mlx5i_set_coalesce(struct net_device *netdev,
+			      struct ethtool_coalesce *coal,
+			      struct kernel_ethtool_coalesce *kernel_coal,
+			      struct netlink_ext_ack *extack)
+{
+	struct mlx5e_priv *priv = mlx5i_epriv(netdev);
+
+	return mlx5e_ethtool_set_coalesce(priv, coal, kernel_coal, extack);
+}
+
+static int mlx5i_get_coalesce(struct net_device *netdev,
+			      struct ethtool_coalesce *coal,
+			      struct kernel_ethtool_coalesce *kernel_coal,
+			      struct netlink_ext_ack *extack)
+{
+	struct mlx5e_priv *priv = mlx5i_epriv(netdev);
+
+	return mlx5e_ethtool_get_coalesce(priv, coal, kernel_coal);
+}
+
+static int mlx5i_get_ts_info(struct net_device *netdev,
+			     struct ethtool_ts_info *info)
+{
+	struct mlx5e_priv *priv = mlx5i_epriv(netdev);
+
+	return mlx5e_ethtool_get_ts_info(priv, info);
+}
+
+static int mlx5i_flash_device(struct net_device *netdev,
+			      struct ethtool_flash *flash)
+{
+	struct mlx5e_priv *priv = mlx5i_epriv(netdev);
+
+	return mlx5e_ethtool_flash_device(priv, flash);
+}
+
+static inline int mlx5_ptys_width_enum_to_int(enum mlx5_ptys_width width)
+{
+	switch (width) {
+	case MLX5_PTYS_WIDTH_1X:  return  1;
+	case MLX5_PTYS_WIDTH_2X:  return  2;
+	case MLX5_PTYS_WIDTH_4X:  return  4;
+	case MLX5_PTYS_WIDTH_8X:  return  8;
+	case MLX5_PTYS_WIDTH_12X: return 12;
+	default:		  return -1;
+	}
+}
+
+enum mlx5_ptys_rate {
+	MLX5_PTYS_RATE_SDR	= 1 << 0,
+	MLX5_PTYS_RATE_DDR	= 1 << 1,
+	MLX5_PTYS_RATE_QDR	= 1 << 2,
+	MLX5_PTYS_RATE_FDR10	= 1 << 3,
+	MLX5_PTYS_RATE_FDR	= 1 << 4,
+	MLX5_PTYS_RATE_EDR	= 1 << 5,
+	MLX5_PTYS_RATE_HDR	= 1 << 6,
+	MLX5_PTYS_RATE_NDR	= 1 << 7,
+	MLX5_PTYS_RATE_XDR	= 1 << 8,
+};
+
+static inline int mlx5_ptys_rate_enum_to_int(enum mlx5_ptys_rate rate)
+{
+	switch (rate) {
+	case MLX5_PTYS_RATE_SDR:   return 2500;
+	case MLX5_PTYS_RATE_DDR:   return 5000;
+	case MLX5_PTYS_RATE_QDR:
+	case MLX5_PTYS_RATE_FDR10: return 10000;
+	case MLX5_PTYS_RATE_FDR:   return 14000;
+	case MLX5_PTYS_RATE_EDR:   return 25000;
+	case MLX5_PTYS_RATE_HDR:   return 50000;
+	case MLX5_PTYS_RATE_NDR:   return 100000;
+	case MLX5_PTYS_RATE_XDR:   return 200000;
+	default:		   return -1;
+	}
+}
+
+static u32 mlx5i_get_speed_settings(u16 ib_link_width_oper, u16 ib_proto_oper)
+{
+	int rate, width;
+
+	rate = mlx5_ptys_rate_enum_to_int(ib_proto_oper);
+	if (rate < 0)
+		return SPEED_UNKNOWN;
+	width = mlx5_ptys_width_enum_to_int(ib_link_width_oper);
+	if (width < 0)
+		return SPEED_UNKNOWN;
+
+	return rate * width;
+}
+
+static int mlx5i_get_link_ksettings(struct net_device *netdev,
+				    struct ethtool_link_ksettings *link_ksettings)
+{
+	struct mlx5e_priv *priv = mlx5i_epriv(netdev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+	u16 ib_link_width_oper;
+	u16 ib_proto_oper;
+	int speed, ret;
+
+	ret = mlx5_query_ib_port_oper(mdev, &ib_link_width_oper, &ib_proto_oper,
+				      1);
+	if (ret)
+		return ret;
+
+	ethtool_link_ksettings_zero_link_mode(link_ksettings, supported);
+	ethtool_link_ksettings_zero_link_mode(link_ksettings, advertising);
+
+	speed = mlx5i_get_speed_settings(ib_link_width_oper, ib_proto_oper);
+	link_ksettings->base.speed = speed;
+	link_ksettings->base.duplex = speed == SPEED_UNKNOWN ? DUPLEX_UNKNOWN : DUPLEX_FULL;
+
+	link_ksettings->base.port = PORT_OTHER;
+
+	link_ksettings->base.autoneg = AUTONEG_DISABLE;
+
+	return 0;
+}
+
+static u32 mlx5i_flow_type_mask(u32 flow_type)
+{
+	return flow_type & ~(FLOW_EXT | FLOW_MAC_EXT | FLOW_RSS);
+}
+
+static int mlx5i_set_rxnfc(struct net_device *dev, struct ethtool_rxnfc *cmd)
+{
+	struct mlx5e_priv *priv = mlx5i_epriv(dev);
+	struct ethtool_rx_flow_spec *fs = &cmd->fs;
+
+	if (mlx5i_flow_type_mask(fs->flow_type) == ETHER_FLOW)
+		return -EINVAL;
+
+	return mlx5e_ethtool_set_rxnfc(priv, cmd);
+}
+
+static int mlx5i_get_rxnfc(struct net_device *dev, struct ethtool_rxnfc *info,
+			   u32 *rule_locs)
+{
+	struct mlx5e_priv *priv = mlx5i_epriv(dev);
+
+	/* ETHTOOL_GRXRINGS is needed by ethtool -x which is not part
+	 * of rxnfc. We keep this logic out of mlx5e_ethtool_get_rxnfc,
+	 * to avoid breaking "ethtool -x" when mlx5e_ethtool_get_rxnfc
+	 * is compiled out via CONFIG_MLX5_EN_RXNFC=n.
+	 */
+	if (info->cmd == ETHTOOL_GRXRINGS) {
+		info->data = priv->channels.params.num_channels;
+		return 0;
+	}
+
+	return mlx5e_ethtool_get_rxnfc(priv, info, rule_locs);
+}
+
+const struct ethtool_ops mlx5i_ethtool_ops = {
+	.supported_coalesce_params = ETHTOOL_COALESCE_USECS |
+				     ETHTOOL_COALESCE_MAX_FRAMES |
+				     ETHTOOL_COALESCE_USE_ADAPTIVE,
+	.get_drvinfo        = mlx5i_get_drvinfo,
+	.get_strings        = mlx5i_get_strings,
+	.get_sset_count     = mlx5i_get_sset_count,
+	.get_ethtool_stats  = mlx5i_get_ethtool_stats,
+	.get_ringparam      = mlx5i_get_ringparam,
+	.set_ringparam      = mlx5i_set_ringparam,
+	.flash_device       = mlx5i_flash_device,
+	.get_channels       = mlx5i_get_channels,
+	.set_channels       = mlx5i_set_channels,
+	.get_coalesce       = mlx5i_get_coalesce,
+	.set_coalesce       = mlx5i_set_coalesce,
+	.get_ts_info        = mlx5i_get_ts_info,
+	.get_rxnfc          = mlx5i_get_rxnfc,
+	.set_rxnfc          = mlx5i_set_rxnfc,
+	.get_link_ksettings = mlx5i_get_link_ksettings,
+	.get_link           = ethtool_op_get_link,
+};
+
+const struct ethtool_ops mlx5i_pkey_ethtool_ops = {
+	.get_drvinfo        = mlx5i_get_drvinfo,
+	.get_link           = ethtool_op_get_link,
+	.get_ts_info        = mlx5i_get_ts_info,
+};
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/ipoib/ipoib.c b/drivers/net/ethernet/mellanox/mlx5/core/ipoib/ipoib.c
new file mode 100644
index 0000000000..baa7ef8123
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/ipoib/ipoib.c
@@ -0,0 +1,848 @@
+/*
+ * Copyright (c) 2017, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <rdma/ib_verbs.h>
+#include <linux/mlx5/fs.h>
+#include "en.h"
+#include "en/params.h"
+#include "ipoib.h"
+#include "en/fs_ethtool.h"
+
+#define IB_DEFAULT_Q_KEY   0xb1b
+#define MLX5I_PARAMS_DEFAULT_LOG_RQ_SIZE 9
+
+static int mlx5i_open(struct net_device *netdev);
+static int mlx5i_close(struct net_device *netdev);
+static int mlx5i_change_mtu(struct net_device *netdev, int new_mtu);
+
+static const struct net_device_ops mlx5i_netdev_ops = {
+	.ndo_open                = mlx5i_open,
+	.ndo_stop                = mlx5i_close,
+	.ndo_get_stats64         = mlx5i_get_stats,
+	.ndo_init                = mlx5i_dev_init,
+	.ndo_uninit              = mlx5i_dev_cleanup,
+	.ndo_change_mtu          = mlx5i_change_mtu,
+	.ndo_eth_ioctl            = mlx5i_ioctl,
+};
+
+/* IPoIB mlx5 netdev profile */
+static void mlx5i_build_nic_params(struct mlx5_core_dev *mdev,
+				   struct mlx5e_params *params)
+{
+	/* Override RQ params as IPoIB supports only LINKED LIST RQ for now */
+	MLX5E_SET_PFLAG(params, MLX5E_PFLAG_RX_STRIDING_RQ, false);
+	mlx5e_set_rq_type(mdev, params);
+	mlx5e_init_rq_type_params(mdev, params);
+
+	/* RQ size in ipoib by default is 512 */
+	params->log_rq_mtu_frames = is_kdump_kernel() ?
+		MLX5E_PARAMS_MINIMUM_LOG_RQ_SIZE :
+		MLX5I_PARAMS_DEFAULT_LOG_RQ_SIZE;
+
+	params->packet_merge.type = MLX5E_PACKET_MERGE_NONE;
+	params->hard_mtu = MLX5_IB_GRH_BYTES + MLX5_IPOIB_HARD_LEN;
+
+	/* CQE compression is not supported for IPoIB */
+	params->rx_cqe_compress_def = false;
+	MLX5E_SET_PFLAG(params, MLX5E_PFLAG_RX_CQE_COMPRESS, params->rx_cqe_compress_def);
+}
+
+/* Called directly after IPoIB netdevice was created to initialize SW structs */
+int mlx5i_init(struct mlx5_core_dev *mdev, struct net_device *netdev)
+{
+	struct mlx5e_priv *priv  = mlx5i_epriv(netdev);
+
+	netif_carrier_off(netdev);
+	mlx5e_set_netdev_mtu_boundaries(priv);
+	netdev->mtu = netdev->max_mtu;
+
+	mlx5e_build_nic_params(priv, NULL, netdev->mtu);
+	mlx5i_build_nic_params(mdev, &priv->channels.params);
+
+	mlx5e_timestamp_init(priv);
+
+	/* netdev init */
+	netdev->hw_features    |= NETIF_F_SG;
+	netdev->hw_features    |= NETIF_F_IP_CSUM;
+	netdev->hw_features    |= NETIF_F_IPV6_CSUM;
+	netdev->hw_features    |= NETIF_F_GRO;
+	netdev->hw_features    |= NETIF_F_TSO;
+	netdev->hw_features    |= NETIF_F_TSO6;
+	netdev->hw_features    |= NETIF_F_RXCSUM;
+	netdev->hw_features    |= NETIF_F_RXHASH;
+
+	netdev->netdev_ops = &mlx5i_netdev_ops;
+	netdev->ethtool_ops = &mlx5i_ethtool_ops;
+
+	return 0;
+}
+
+/* Called directly before IPoIB netdevice is destroyed to cleanup SW structs */
+void mlx5i_cleanup(struct mlx5e_priv *priv)
+{
+	mlx5e_priv_cleanup(priv);
+}
+
+static void mlx5i_grp_sw_update_stats(struct mlx5e_priv *priv)
+{
+	struct rtnl_link_stats64 s = {};
+	int i, j;
+
+	for (i = 0; i < priv->stats_nch; i++) {
+		struct mlx5e_channel_stats *channel_stats;
+		struct mlx5e_rq_stats *rq_stats;
+
+		channel_stats = priv->channel_stats[i];
+		rq_stats = &channel_stats->rq;
+
+		s.rx_packets += rq_stats->packets;
+		s.rx_bytes   += rq_stats->bytes;
+
+		for (j = 0; j < priv->max_opened_tc; j++) {
+			struct mlx5e_sq_stats *sq_stats = &channel_stats->sq[j];
+
+			s.tx_packets           += sq_stats->packets;
+			s.tx_bytes             += sq_stats->bytes;
+			s.tx_dropped           += sq_stats->dropped;
+		}
+	}
+
+	memset(&priv->stats.sw, 0, sizeof(s));
+
+	priv->stats.sw.rx_packets = s.rx_packets;
+	priv->stats.sw.rx_bytes = s.rx_bytes;
+	priv->stats.sw.tx_packets = s.tx_packets;
+	priv->stats.sw.tx_bytes = s.tx_bytes;
+	priv->stats.sw.tx_queue_dropped = s.tx_dropped;
+}
+
+void mlx5i_get_stats(struct net_device *dev, struct rtnl_link_stats64 *stats)
+{
+	struct mlx5e_priv     *priv   = mlx5i_epriv(dev);
+	struct mlx5e_sw_stats *sstats = &priv->stats.sw;
+
+	mlx5i_grp_sw_update_stats(priv);
+
+	stats->rx_packets = sstats->rx_packets;
+	stats->rx_bytes   = sstats->rx_bytes;
+	stats->tx_packets = sstats->tx_packets;
+	stats->tx_bytes   = sstats->tx_bytes;
+	stats->tx_dropped = sstats->tx_queue_dropped;
+}
+
+struct net_device *mlx5i_parent_get(struct net_device *netdev)
+{
+	struct mlx5e_priv *priv = mlx5i_epriv(netdev);
+	struct mlx5i_priv *ipriv, *parent_ipriv;
+	struct net_device *parent_dev;
+	int parent_ifindex;
+
+	ipriv = priv->ppriv;
+
+	parent_ifindex = netdev->netdev_ops->ndo_get_iflink(netdev);
+	parent_dev = dev_get_by_index(dev_net(netdev), parent_ifindex);
+	if (!parent_dev)
+		return NULL;
+
+	parent_ipriv = netdev_priv(parent_dev);
+
+	ASSERT_RTNL();
+	parent_ipriv->num_sub_interfaces++;
+
+	ipriv->parent_dev = parent_dev;
+
+	return parent_dev;
+}
+
+void mlx5i_parent_put(struct net_device *netdev)
+{
+	struct mlx5e_priv *priv = mlx5i_epriv(netdev);
+	struct mlx5i_priv *ipriv, *parent_ipriv;
+
+	ipriv = priv->ppriv;
+	parent_ipriv = netdev_priv(ipriv->parent_dev);
+
+	ASSERT_RTNL();
+	parent_ipriv->num_sub_interfaces--;
+
+	dev_put(ipriv->parent_dev);
+}
+
+int mlx5i_init_underlay_qp(struct mlx5e_priv *priv)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+	struct mlx5i_priv *ipriv = priv->ppriv;
+	int ret;
+
+	{
+		u32 in[MLX5_ST_SZ_DW(rst2init_qp_in)] = {};
+		u32 *qpc;
+
+		qpc = MLX5_ADDR_OF(rst2init_qp_in, in, qpc);
+
+		MLX5_SET(qpc, qpc, pm_state, MLX5_QP_PM_MIGRATED);
+		MLX5_SET(qpc, qpc, primary_address_path.pkey_index,
+			 ipriv->pkey_index);
+		MLX5_SET(qpc, qpc, primary_address_path.vhca_port_num, 1);
+		MLX5_SET(qpc, qpc, q_key, IB_DEFAULT_Q_KEY);
+
+		MLX5_SET(rst2init_qp_in, in, opcode, MLX5_CMD_OP_RST2INIT_QP);
+		MLX5_SET(rst2init_qp_in, in, qpn, ipriv->qpn);
+		ret = mlx5_cmd_exec_in(mdev, rst2init_qp, in);
+		if (ret)
+			goto err_qp_modify_to_err;
+	}
+	{
+		u32 in[MLX5_ST_SZ_DW(init2rtr_qp_in)] = {};
+
+		MLX5_SET(init2rtr_qp_in, in, opcode, MLX5_CMD_OP_INIT2RTR_QP);
+		MLX5_SET(init2rtr_qp_in, in, qpn, ipriv->qpn);
+		ret = mlx5_cmd_exec_in(mdev, init2rtr_qp, in);
+		if (ret)
+			goto err_qp_modify_to_err;
+	}
+	{
+		u32 in[MLX5_ST_SZ_DW(rtr2rts_qp_in)] = {};
+
+		MLX5_SET(rtr2rts_qp_in, in, opcode, MLX5_CMD_OP_RTR2RTS_QP);
+		MLX5_SET(rtr2rts_qp_in, in, qpn, ipriv->qpn);
+		ret = mlx5_cmd_exec_in(mdev, rtr2rts_qp, in);
+		if (ret)
+			goto err_qp_modify_to_err;
+	}
+	return 0;
+
+err_qp_modify_to_err:
+	{
+		u32 in[MLX5_ST_SZ_DW(qp_2err_in)] = {};
+
+		MLX5_SET(qp_2err_in, in, opcode, MLX5_CMD_OP_2ERR_QP);
+		MLX5_SET(qp_2err_in, in, qpn, ipriv->qpn);
+		mlx5_cmd_exec_in(mdev, qp_2err, in);
+	}
+	return ret;
+}
+
+void mlx5i_uninit_underlay_qp(struct mlx5e_priv *priv)
+{
+	struct mlx5i_priv *ipriv = priv->ppriv;
+	struct mlx5_core_dev *mdev = priv->mdev;
+	u32 in[MLX5_ST_SZ_DW(qp_2rst_in)] = {};
+
+	MLX5_SET(qp_2rst_in, in, opcode, MLX5_CMD_OP_2RST_QP);
+	MLX5_SET(qp_2rst_in, in, qpn, ipriv->qpn);
+	mlx5_cmd_exec_in(mdev, qp_2rst, in);
+}
+
+#define MLX5_QP_ENHANCED_ULP_STATELESS_MODE 2
+
+int mlx5i_create_underlay_qp(struct mlx5e_priv *priv)
+{
+	const unsigned char *dev_addr = priv->netdev->dev_addr;
+	u32 out[MLX5_ST_SZ_DW(create_qp_out)] = {};
+	u32 in[MLX5_ST_SZ_DW(create_qp_in)] = {};
+	struct mlx5i_priv *ipriv = priv->ppriv;
+	void *addr_path;
+	int qpn = 0;
+	int ret = 0;
+	void *qpc;
+
+	if (MLX5_CAP_GEN(priv->mdev, mkey_by_name)) {
+		qpn = (dev_addr[1] << 16) + (dev_addr[2] << 8) + dev_addr[3];
+		MLX5_SET(create_qp_in, in, input_qpn, qpn);
+	}
+
+	qpc = MLX5_ADDR_OF(create_qp_in, in, qpc);
+	MLX5_SET(qpc, qpc, ts_format, mlx5_get_qp_default_ts(priv->mdev));
+	MLX5_SET(qpc, qpc, st, MLX5_QP_ST_UD);
+	MLX5_SET(qpc, qpc, pm_state, MLX5_QP_PM_MIGRATED);
+	MLX5_SET(qpc, qpc, ulp_stateless_offload_mode,
+		 MLX5_QP_ENHANCED_ULP_STATELESS_MODE);
+
+	addr_path = MLX5_ADDR_OF(qpc, qpc, primary_address_path);
+	MLX5_SET(ads, addr_path, vhca_port_num, 1);
+	MLX5_SET(ads, addr_path, grh, 1);
+
+	MLX5_SET(create_qp_in, in, opcode, MLX5_CMD_OP_CREATE_QP);
+	ret = mlx5_cmd_exec_inout(priv->mdev, create_qp, in, out);
+	if (ret)
+		return ret;
+
+	ipriv->qpn = MLX5_GET(create_qp_out, out, qpn);
+
+	return 0;
+}
+
+void mlx5i_destroy_underlay_qp(struct mlx5_core_dev *mdev, u32 qpn)
+{
+	u32 in[MLX5_ST_SZ_DW(destroy_qp_in)] = {};
+
+	MLX5_SET(destroy_qp_in, in, opcode, MLX5_CMD_OP_DESTROY_QP);
+	MLX5_SET(destroy_qp_in, in, qpn, qpn);
+	mlx5_cmd_exec_in(mdev, destroy_qp, in);
+}
+
+int mlx5i_update_nic_rx(struct mlx5e_priv *priv)
+{
+	return mlx5e_refresh_tirs(priv, true, true);
+}
+
+int mlx5i_create_tis(struct mlx5_core_dev *mdev, u32 underlay_qpn, u32 *tisn)
+{
+	u32 in[MLX5_ST_SZ_DW(create_tis_in)] = {};
+	void *tisc;
+
+	tisc = MLX5_ADDR_OF(create_tis_in, in, ctx);
+
+	MLX5_SET(tisc, tisc, underlay_qpn, underlay_qpn);
+
+	return mlx5e_create_tis(mdev, in, tisn);
+}
+
+static int mlx5i_init_tx(struct mlx5e_priv *priv)
+{
+	struct mlx5i_priv *ipriv = priv->ppriv;
+	int err;
+
+	err = mlx5i_create_underlay_qp(priv);
+	if (err) {
+		mlx5_core_warn(priv->mdev, "create underlay QP failed, %d\n", err);
+		return err;
+	}
+
+	err = mlx5i_create_tis(priv->mdev, ipriv->qpn, &priv->tisn[0][0]);
+	if (err) {
+		mlx5_core_warn(priv->mdev, "create tis failed, %d\n", err);
+		goto err_destroy_underlay_qp;
+	}
+
+	return 0;
+
+err_destroy_underlay_qp:
+	mlx5i_destroy_underlay_qp(priv->mdev, ipriv->qpn);
+	return err;
+}
+
+static void mlx5i_cleanup_tx(struct mlx5e_priv *priv)
+{
+	struct mlx5i_priv *ipriv = priv->ppriv;
+
+	mlx5e_destroy_tis(priv->mdev, priv->tisn[0][0]);
+	mlx5i_destroy_underlay_qp(priv->mdev, ipriv->qpn);
+}
+
+static int mlx5i_create_flow_steering(struct mlx5e_priv *priv)
+{
+	struct mlx5_flow_namespace *ns =
+		mlx5_get_flow_namespace(priv->mdev, MLX5_FLOW_NAMESPACE_KERNEL);
+	int err;
+
+
+	if (!ns)
+		return -EINVAL;
+
+	mlx5e_fs_set_ns(priv->fs, ns, false);
+	err = mlx5e_arfs_create_tables(priv->fs, priv->rx_res,
+				       !!(priv->netdev->hw_features & NETIF_F_NTUPLE));
+	if (err) {
+		netdev_err(priv->netdev, "Failed to create arfs tables, err=%d\n",
+			   err);
+		priv->netdev->hw_features &= ~NETIF_F_NTUPLE;
+	}
+
+	err = mlx5e_create_ttc_table(priv->fs, priv->rx_res);
+	if (err) {
+		netdev_err(priv->netdev, "Failed to create ttc table, err=%d\n",
+			   err);
+		goto err_destroy_arfs_tables;
+	}
+
+	mlx5e_ethtool_init_steering(priv->fs);
+
+	return 0;
+
+err_destroy_arfs_tables:
+	mlx5e_arfs_destroy_tables(priv->fs,
+				  !!(priv->netdev->hw_features & NETIF_F_NTUPLE));
+
+	return err;
+}
+
+static void mlx5i_destroy_flow_steering(struct mlx5e_priv *priv)
+{
+	mlx5e_destroy_ttc_table(priv->fs);
+	mlx5e_arfs_destroy_tables(priv->fs,
+				  !!(priv->netdev->hw_features & NETIF_F_NTUPLE));
+	mlx5e_ethtool_cleanup_steering(priv->fs);
+}
+
+static int mlx5i_init_rx(struct mlx5e_priv *priv)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+	int err;
+
+	priv->fs = mlx5e_fs_init(priv->profile, mdev,
+				 !test_bit(MLX5E_STATE_DESTROYING, &priv->state),
+				 priv->dfs_root);
+	if (!priv->fs) {
+		netdev_err(priv->netdev, "FS allocation failed\n");
+		return -ENOMEM;
+	}
+
+	priv->rx_res = mlx5e_rx_res_alloc();
+	if (!priv->rx_res) {
+		err = -ENOMEM;
+		goto err_free_fs;
+	}
+
+	mlx5e_create_q_counters(priv);
+
+	err = mlx5e_open_drop_rq(priv, &priv->drop_rq);
+	if (err) {
+		mlx5_core_err(mdev, "open drop rq failed, %d\n", err);
+		goto err_destroy_q_counters;
+	}
+
+	err = mlx5e_rx_res_init(priv->rx_res, priv->mdev, 0,
+				priv->max_nch, priv->drop_rq.rqn,
+				&priv->channels.params.packet_merge,
+				priv->channels.params.num_channels);
+	if (err)
+		goto err_close_drop_rq;
+
+	err = mlx5i_create_flow_steering(priv);
+	if (err)
+		goto err_destroy_rx_res;
+
+	return 0;
+
+err_destroy_rx_res:
+	mlx5e_rx_res_destroy(priv->rx_res);
+err_close_drop_rq:
+	mlx5e_close_drop_rq(&priv->drop_rq);
+err_destroy_q_counters:
+	mlx5e_destroy_q_counters(priv);
+	mlx5e_rx_res_free(priv->rx_res);
+	priv->rx_res = NULL;
+err_free_fs:
+	mlx5e_fs_cleanup(priv->fs);
+	return err;
+}
+
+static void mlx5i_cleanup_rx(struct mlx5e_priv *priv)
+{
+	mlx5i_destroy_flow_steering(priv);
+	mlx5e_rx_res_destroy(priv->rx_res);
+	mlx5e_close_drop_rq(&priv->drop_rq);
+	mlx5e_destroy_q_counters(priv);
+	mlx5e_rx_res_free(priv->rx_res);
+	priv->rx_res = NULL;
+	mlx5e_fs_cleanup(priv->fs);
+}
+
+/* The stats groups order is opposite to the update_stats() order calls */
+static mlx5e_stats_grp_t mlx5i_stats_grps[] = {
+	&MLX5E_STATS_GRP(sw),
+	&MLX5E_STATS_GRP(qcnt),
+	&MLX5E_STATS_GRP(vnic_env),
+	&MLX5E_STATS_GRP(vport),
+	&MLX5E_STATS_GRP(802_3),
+	&MLX5E_STATS_GRP(2863),
+	&MLX5E_STATS_GRP(2819),
+	&MLX5E_STATS_GRP(phy),
+	&MLX5E_STATS_GRP(pcie),
+	&MLX5E_STATS_GRP(per_prio),
+	&MLX5E_STATS_GRP(pme),
+	&MLX5E_STATS_GRP(channels),
+	&MLX5E_STATS_GRP(per_port_buff_congest),
+};
+
+static unsigned int mlx5i_stats_grps_num(struct mlx5e_priv *priv)
+{
+	return ARRAY_SIZE(mlx5i_stats_grps);
+}
+
+static const struct mlx5e_profile mlx5i_nic_profile = {
+	.init		   = mlx5i_init,
+	.cleanup	   = mlx5i_cleanup,
+	.init_tx	   = mlx5i_init_tx,
+	.cleanup_tx	   = mlx5i_cleanup_tx,
+	.init_rx	   = mlx5i_init_rx,
+	.cleanup_rx	   = mlx5i_cleanup_rx,
+	.enable		   = NULL, /* mlx5i_enable */
+	.disable	   = NULL, /* mlx5i_disable */
+	.update_rx	   = mlx5i_update_nic_rx,
+	.update_stats	   = NULL, /* mlx5i_update_stats */
+	.update_carrier    = NULL, /* no HW update in IB link */
+	.rx_handlers       = &mlx5i_rx_handlers,
+	.max_tc		   = MLX5I_MAX_NUM_TC,
+	.stats_grps        = mlx5i_stats_grps,
+	.stats_grps_num    = mlx5i_stats_grps_num,
+};
+
+/* mlx5i netdev NDos */
+
+static int mlx5i_change_mtu(struct net_device *netdev, int new_mtu)
+{
+	struct mlx5e_priv *priv = mlx5i_epriv(netdev);
+	struct mlx5e_params new_params;
+	int err = 0;
+
+	mutex_lock(&priv->state_lock);
+
+	new_params = priv->channels.params;
+	new_params.sw_mtu = new_mtu;
+
+	err = mlx5e_safe_switch_params(priv, &new_params, NULL, NULL, true);
+	if (err)
+		goto out;
+
+	netdev->mtu = new_params.sw_mtu;
+
+out:
+	mutex_unlock(&priv->state_lock);
+	return err;
+}
+
+int mlx5i_dev_init(struct net_device *dev)
+{
+	struct mlx5e_priv    *priv   = mlx5i_epriv(dev);
+	struct mlx5i_priv    *ipriv  = priv->ppriv;
+	u8 addr_mod[3];
+
+	/* Set dev address using underlay QP */
+	addr_mod[0] = (ipriv->qpn >> 16) & 0xff;
+	addr_mod[1] = (ipriv->qpn >>  8) & 0xff;
+	addr_mod[2] = (ipriv->qpn) & 0xff;
+	dev_addr_mod(dev, 1, addr_mod, sizeof(addr_mod));
+
+	/* Add QPN to net-device mapping to HT */
+	mlx5i_pkey_add_qpn(dev, ipriv->qpn);
+
+	return 0;
+}
+
+int mlx5i_ioctl(struct net_device *dev, struct ifreq *ifr, int cmd)
+{
+	struct mlx5e_priv *priv = mlx5i_epriv(dev);
+
+	switch (cmd) {
+	case SIOCSHWTSTAMP:
+		return mlx5e_hwstamp_set(priv, ifr);
+	case SIOCGHWTSTAMP:
+		return mlx5e_hwstamp_get(priv, ifr);
+	default:
+		return -EOPNOTSUPP;
+	}
+}
+
+void mlx5i_dev_cleanup(struct net_device *dev)
+{
+	struct mlx5e_priv    *priv   = mlx5i_epriv(dev);
+	struct mlx5i_priv    *ipriv = priv->ppriv;
+
+	mlx5i_uninit_underlay_qp(priv);
+
+	/* Delete QPN to net-device mapping from HT */
+	mlx5i_pkey_del_qpn(dev, ipriv->qpn);
+}
+
+static int mlx5i_open(struct net_device *netdev)
+{
+	struct mlx5e_priv *epriv = mlx5i_epriv(netdev);
+	struct mlx5i_priv *ipriv = epriv->ppriv;
+	struct mlx5_core_dev *mdev = epriv->mdev;
+	int err;
+
+	mutex_lock(&epriv->state_lock);
+
+	set_bit(MLX5E_STATE_OPENED, &epriv->state);
+
+	err = mlx5i_init_underlay_qp(epriv);
+	if (err) {
+		mlx5_core_warn(mdev, "prepare underlay qp state failed, %d\n", err);
+		goto err_clear_state_opened_flag;
+	}
+
+	err = mlx5_fs_add_rx_underlay_qpn(mdev, ipriv->qpn);
+	if (err) {
+		mlx5_core_warn(mdev, "attach underlay qp to ft failed, %d\n", err);
+		goto err_reset_qp;
+	}
+
+	err = mlx5e_open_channels(epriv, &epriv->channels);
+	if (err)
+		goto err_remove_fs_underlay_qp;
+
+	err = epriv->profile->update_rx(epriv);
+	if (err)
+		goto err_close_channels;
+
+	mlx5e_activate_priv_channels(epriv);
+
+	mutex_unlock(&epriv->state_lock);
+	return 0;
+
+err_close_channels:
+	mlx5e_close_channels(&epriv->channels);
+err_remove_fs_underlay_qp:
+	mlx5_fs_remove_rx_underlay_qpn(mdev, ipriv->qpn);
+err_reset_qp:
+	mlx5i_uninit_underlay_qp(epriv);
+err_clear_state_opened_flag:
+	clear_bit(MLX5E_STATE_OPENED, &epriv->state);
+	mutex_unlock(&epriv->state_lock);
+	return err;
+}
+
+static int mlx5i_close(struct net_device *netdev)
+{
+	struct mlx5e_priv *epriv = mlx5i_epriv(netdev);
+	struct mlx5i_priv *ipriv = epriv->ppriv;
+	struct mlx5_core_dev *mdev = epriv->mdev;
+
+	/* May already be CLOSED in case a previous configuration operation
+	 * (e.g RX/TX queue size change) that involves close&open failed.
+	 */
+	mutex_lock(&epriv->state_lock);
+
+	if (!test_bit(MLX5E_STATE_OPENED, &epriv->state))
+		goto unlock;
+
+	clear_bit(MLX5E_STATE_OPENED, &epriv->state);
+
+	netif_carrier_off(epriv->netdev);
+	mlx5_fs_remove_rx_underlay_qpn(mdev, ipriv->qpn);
+	mlx5e_deactivate_priv_channels(epriv);
+	mlx5e_close_channels(&epriv->channels);
+	mlx5i_uninit_underlay_qp(epriv);
+unlock:
+	mutex_unlock(&epriv->state_lock);
+	return 0;
+}
+
+/* IPoIB RDMA netdev callbacks */
+static int mlx5i_attach_mcast(struct net_device *netdev, struct ib_device *hca,
+			      union ib_gid *gid, u16 lid, int set_qkey,
+			      u32 qkey)
+{
+	struct mlx5e_priv    *epriv = mlx5i_epriv(netdev);
+	struct mlx5_core_dev *mdev  = epriv->mdev;
+	struct mlx5i_priv    *ipriv = epriv->ppriv;
+	int err;
+
+	mlx5_core_dbg(mdev, "attaching QPN 0x%x, MGID %pI6\n", ipriv->qpn,
+		      gid->raw);
+	err = mlx5_core_attach_mcg(mdev, gid, ipriv->qpn);
+	if (err)
+		mlx5_core_warn(mdev, "failed attaching QPN 0x%x, MGID %pI6\n",
+			       ipriv->qpn, gid->raw);
+
+	if (set_qkey) {
+		mlx5_core_dbg(mdev, "%s setting qkey 0x%x\n",
+			      netdev->name, qkey);
+		ipriv->qkey = qkey;
+	}
+
+	return err;
+}
+
+static int mlx5i_detach_mcast(struct net_device *netdev, struct ib_device *hca,
+			      union ib_gid *gid, u16 lid)
+{
+	struct mlx5e_priv    *epriv = mlx5i_epriv(netdev);
+	struct mlx5_core_dev *mdev  = epriv->mdev;
+	struct mlx5i_priv    *ipriv = epriv->ppriv;
+	int err;
+
+	mlx5_core_dbg(mdev, "detaching QPN 0x%x, MGID %pI6\n", ipriv->qpn,
+		      gid->raw);
+
+	err = mlx5_core_detach_mcg(mdev, gid, ipriv->qpn);
+	if (err)
+		mlx5_core_dbg(mdev, "failed detaching QPN 0x%x, MGID %pI6\n",
+			      ipriv->qpn, gid->raw);
+
+	return err;
+}
+
+static int mlx5i_xmit(struct net_device *dev, struct sk_buff *skb,
+		      struct ib_ah *address, u32 dqpn)
+{
+	struct mlx5e_priv *epriv = mlx5i_epriv(dev);
+	struct mlx5e_txqsq *sq   = epriv->txq2sq[skb_get_queue_mapping(skb)];
+	struct mlx5_ib_ah *mah   = to_mah(address);
+	struct mlx5i_priv *ipriv = epriv->ppriv;
+
+	mlx5i_sq_xmit(sq, skb, &mah->av, dqpn, ipriv->qkey, netdev_xmit_more());
+
+	return NETDEV_TX_OK;
+}
+
+static void mlx5i_set_pkey_index(struct net_device *netdev, int id)
+{
+	struct mlx5i_priv *ipriv = netdev_priv(netdev);
+
+	ipriv->pkey_index = (u16)id;
+}
+
+static int mlx5i_check_required_hca_cap(struct mlx5_core_dev *mdev)
+{
+	if (MLX5_CAP_GEN(mdev, port_type) != MLX5_CAP_PORT_TYPE_IB)
+		return -EOPNOTSUPP;
+
+	if (!MLX5_CAP_GEN(mdev, ipoib_enhanced_offloads)) {
+		mlx5_core_warn(mdev, "IPoIB enhanced offloads are not supported\n");
+		return -EOPNOTSUPP;
+	}
+
+	return 0;
+}
+
+static void mlx5_rdma_netdev_free(struct net_device *netdev)
+{
+	struct mlx5e_priv *priv = mlx5i_epriv(netdev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+	struct mlx5i_priv *ipriv = priv->ppriv;
+	const struct mlx5e_profile *profile = priv->profile;
+
+	mlx5e_detach_netdev(priv);
+	profile->cleanup(priv);
+
+	if (!ipriv->sub_interface) {
+		mlx5i_pkey_qpn_ht_cleanup(netdev);
+		mlx5e_destroy_mdev_resources(mdev);
+	}
+}
+
+static bool mlx5_is_sub_interface(struct mlx5_core_dev *mdev)
+{
+	return mdev->mlx5e_res.hw_objs.pdn != 0;
+}
+
+static const struct mlx5e_profile *mlx5_get_profile(struct mlx5_core_dev *mdev)
+{
+	if (mlx5_is_sub_interface(mdev))
+		return mlx5i_pkey_get_profile();
+	return &mlx5i_nic_profile;
+}
+
+static int mlx5_rdma_setup_rn(struct ib_device *ibdev, u32 port_num,
+			      struct net_device *netdev, void *param)
+{
+	struct mlx5_core_dev *mdev = (struct mlx5_core_dev *)param;
+	const struct mlx5e_profile *prof = mlx5_get_profile(mdev);
+	struct mlx5i_priv *ipriv;
+	struct mlx5e_priv *epriv;
+	struct rdma_netdev *rn;
+	int err;
+
+	ipriv = netdev_priv(netdev);
+	epriv = mlx5i_epriv(netdev);
+
+	ipriv->sub_interface = mlx5_is_sub_interface(mdev);
+	if (!ipriv->sub_interface) {
+		err = mlx5i_pkey_qpn_ht_init(netdev);
+		if (err) {
+			mlx5_core_warn(mdev, "allocate qpn_to_netdev ht failed\n");
+			return err;
+		}
+
+		/* This should only be called once per mdev */
+		err = mlx5e_create_mdev_resources(mdev);
+		if (err)
+			goto destroy_ht;
+	}
+
+	err = mlx5e_priv_init(epriv, prof, netdev, mdev);
+	if (err)
+		goto destroy_mdev_resources;
+
+	epriv->profile = prof;
+	epriv->ppriv = ipriv;
+
+	prof->init(mdev, netdev);
+
+	err = mlx5e_attach_netdev(epriv);
+	if (err)
+		goto detach;
+	netif_carrier_off(netdev);
+
+	/* set rdma_netdev func pointers */
+	rn = &ipriv->rn;
+	rn->hca  = ibdev;
+	rn->send = mlx5i_xmit;
+	rn->attach_mcast = mlx5i_attach_mcast;
+	rn->detach_mcast = mlx5i_detach_mcast;
+	rn->set_id = mlx5i_set_pkey_index;
+
+	netdev->priv_destructor = mlx5_rdma_netdev_free;
+	netdev->needs_free_netdev = 1;
+
+	return 0;
+
+detach:
+	prof->cleanup(epriv);
+	if (ipriv->sub_interface)
+		return err;
+destroy_mdev_resources:
+	mlx5e_destroy_mdev_resources(mdev);
+destroy_ht:
+	mlx5i_pkey_qpn_ht_cleanup(netdev);
+	return err;
+}
+
+int mlx5_rdma_rn_get_params(struct mlx5_core_dev *mdev,
+			    struct ib_device *device,
+			    struct rdma_netdev_alloc_params *params)
+{
+	int nch;
+	int rc;
+
+	rc = mlx5i_check_required_hca_cap(mdev);
+	if (rc)
+		return rc;
+
+	nch = mlx5e_get_max_num_channels(mdev);
+
+	*params = (struct rdma_netdev_alloc_params){
+		.sizeof_priv = sizeof(struct mlx5i_priv) +
+			       sizeof(struct mlx5e_priv),
+		.txqs = nch * MLX5E_MAX_NUM_TC,
+		.rxqs = nch,
+		.param = mdev,
+		.initialize_rdma_netdev = mlx5_rdma_setup_rn,
+	};
+
+	return 0;
+}
+EXPORT_SYMBOL(mlx5_rdma_rn_get_params);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/ipoib/ipoib.h b/drivers/net/ethernet/mellanox/mlx5/core/ipoib/ipoib.h
new file mode 100644
index 0000000000..f3f2af9720
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/ipoib/ipoib.h
@@ -0,0 +1,127 @@
+/*
+ * Copyright (c) 2017, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef __MLX5E_IPOB_H__
+#define __MLX5E_IPOB_H__
+
+#ifdef CONFIG_MLX5_CORE_IPOIB
+
+#include <linux/mlx5/fs.h>
+#include "en.h"
+
+#define MLX5I_MAX_NUM_TC 1
+
+extern const struct ethtool_ops mlx5i_ethtool_ops;
+extern const struct ethtool_ops mlx5i_pkey_ethtool_ops;
+extern const struct mlx5e_rx_handlers mlx5i_rx_handlers;
+
+#define MLX5_IB_GRH_BYTES       40
+#define MLX5_IPOIB_ENCAP_LEN    4
+#define MLX5_IPOIB_PSEUDO_LEN   20
+#define MLX5_IPOIB_HARD_LEN     (MLX5_IPOIB_PSEUDO_LEN + MLX5_IPOIB_ENCAP_LEN)
+
+/* ipoib rdma netdev's private data structure */
+struct mlx5i_priv {
+	struct rdma_netdev rn; /* keep this first */
+	u32 qpn;
+	bool   sub_interface;
+	u32    num_sub_interfaces;
+	u32    qkey;
+	u16    pkey_index;
+	struct mlx5i_pkey_qpn_ht *qpn_htbl;
+	struct net_device *parent_dev;
+	char  *mlx5e_priv[];
+};
+
+int mlx5i_create_tis(struct mlx5_core_dev *mdev, u32 underlay_qpn, u32 *tisn);
+
+/* Underlay QP create/destroy functions */
+int mlx5i_create_underlay_qp(struct mlx5e_priv *priv);
+void mlx5i_destroy_underlay_qp(struct mlx5_core_dev *mdev, u32 qpn);
+
+/* Underlay QP state modification init/uninit functions */
+int mlx5i_init_underlay_qp(struct mlx5e_priv *priv);
+void mlx5i_uninit_underlay_qp(struct mlx5e_priv *priv);
+
+/* Allocate/Free underlay QPN to net-device hash table */
+int mlx5i_pkey_qpn_ht_init(struct net_device *netdev);
+void mlx5i_pkey_qpn_ht_cleanup(struct net_device *netdev);
+
+/* Add/Remove an underlay QPN to net-device mapping to/from the hash table */
+int mlx5i_pkey_add_qpn(struct net_device *netdev, u32 qpn);
+int mlx5i_pkey_del_qpn(struct net_device *netdev, u32 qpn);
+
+/* Get the net-device corresponding to the given underlay QPN */
+struct net_device *mlx5i_pkey_get_netdev(struct net_device *netdev, u32 qpn);
+
+/* Shared ndo functions */
+int mlx5i_dev_init(struct net_device *dev);
+void mlx5i_dev_cleanup(struct net_device *dev);
+int mlx5i_ioctl(struct net_device *dev, struct ifreq *ifr, int cmd);
+
+/* Parent profile functions */
+int mlx5i_init(struct mlx5_core_dev *mdev, struct net_device *netdev);
+void mlx5i_cleanup(struct mlx5e_priv *priv);
+
+int mlx5i_update_nic_rx(struct mlx5e_priv *priv);
+
+/* Get child interface nic profile */
+const struct mlx5e_profile *mlx5i_pkey_get_profile(void);
+
+/* Extract mlx5e_priv from IPoIB netdev */
+#define mlx5i_epriv(netdev) ((void *)(((struct mlx5i_priv *)netdev_priv(netdev))->mlx5e_priv))
+
+struct mlx5_wqe_eth_pad {
+	u8 rsvd0[16];
+};
+
+struct mlx5i_tx_wqe {
+	struct mlx5_wqe_ctrl_seg     ctrl;
+	struct mlx5_wqe_datagram_seg datagram;
+	struct mlx5_wqe_eth_pad      pad;
+	struct mlx5_wqe_eth_seg      eth;
+	struct mlx5_wqe_data_seg     data[];
+};
+
+#define MLX5I_SQ_FETCH_WQE(sq, pi) \
+	((struct mlx5i_tx_wqe *)mlx5e_fetch_wqe(&(sq)->wq, pi, sizeof(struct mlx5i_tx_wqe)))
+
+void mlx5i_sq_xmit(struct mlx5e_txqsq *sq, struct sk_buff *skb,
+		   struct mlx5_av *av, u32 dqpn, u32 dqkey, bool xmit_more);
+void mlx5i_get_stats(struct net_device *dev, struct rtnl_link_stats64 *stats);
+
+/* Reference management for child to parent interfaces. */
+struct net_device *mlx5i_parent_get(struct net_device *netdev);
+void mlx5i_parent_put(struct net_device *netdev);
+
+#endif /* CONFIG_MLX5_CORE_IPOIB */
+#endif /* __MLX5E_IPOB_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/ipoib/ipoib_vlan.c b/drivers/net/ethernet/mellanox/mlx5/core/ipoib/ipoib_vlan.c
new file mode 100644
index 0000000000..03e6812979
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/ipoib/ipoib_vlan.c
@@ -0,0 +1,369 @@
+/*
+ * Copyright (c) 2017, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/hash.h>
+#include "ipoib.h"
+
+#define MLX5I_MAX_LOG_PKEY_SUP 7
+
+struct qpn_to_netdev {
+	struct net_device *netdev;
+	struct hlist_node hlist;
+	u32 underlay_qpn;
+};
+
+struct mlx5i_pkey_qpn_ht {
+	struct hlist_head buckets[1 << MLX5I_MAX_LOG_PKEY_SUP];
+	spinlock_t ht_lock; /* Synchronise with NAPI */
+};
+
+int mlx5i_pkey_qpn_ht_init(struct net_device *netdev)
+{
+	struct mlx5i_priv *ipriv = netdev_priv(netdev);
+	struct mlx5i_pkey_qpn_ht *qpn_htbl;
+
+	qpn_htbl = kzalloc(sizeof(*qpn_htbl), GFP_KERNEL);
+	if (!qpn_htbl)
+		return -ENOMEM;
+
+	ipriv->qpn_htbl = qpn_htbl;
+	spin_lock_init(&qpn_htbl->ht_lock);
+
+	return 0;
+}
+
+void mlx5i_pkey_qpn_ht_cleanup(struct net_device *netdev)
+{
+	struct mlx5i_priv *ipriv = netdev_priv(netdev);
+
+	kfree(ipriv->qpn_htbl);
+}
+
+static struct qpn_to_netdev *mlx5i_find_qpn_to_netdev_node(struct hlist_head *buckets,
+							   u32 qpn)
+{
+	struct hlist_head *h = &buckets[hash_32(qpn, MLX5I_MAX_LOG_PKEY_SUP)];
+	struct qpn_to_netdev *node;
+
+	hlist_for_each_entry(node, h, hlist) {
+		if (node->underlay_qpn == qpn)
+			return node;
+	}
+
+	return NULL;
+}
+
+int mlx5i_pkey_add_qpn(struct net_device *netdev, u32 qpn)
+{
+	struct mlx5i_priv *ipriv = netdev_priv(netdev);
+	struct mlx5i_pkey_qpn_ht *ht = ipriv->qpn_htbl;
+	u8 key = hash_32(qpn, MLX5I_MAX_LOG_PKEY_SUP);
+	struct qpn_to_netdev *new_node;
+
+	new_node = kzalloc(sizeof(*new_node), GFP_KERNEL);
+	if (!new_node)
+		return -ENOMEM;
+
+	new_node->netdev = netdev;
+	new_node->underlay_qpn = qpn;
+	spin_lock_bh(&ht->ht_lock);
+	hlist_add_head(&new_node->hlist, &ht->buckets[key]);
+	spin_unlock_bh(&ht->ht_lock);
+
+	return 0;
+}
+
+int mlx5i_pkey_del_qpn(struct net_device *netdev, u32 qpn)
+{
+	struct mlx5e_priv *epriv = mlx5i_epriv(netdev);
+	struct mlx5i_priv *ipriv = epriv->ppriv;
+	struct mlx5i_pkey_qpn_ht *ht = ipriv->qpn_htbl;
+	struct qpn_to_netdev *node;
+
+	node = mlx5i_find_qpn_to_netdev_node(ht->buckets, qpn);
+	if (!node) {
+		mlx5_core_warn(epriv->mdev, "QPN to netdev delete from HT failed\n");
+		return -EINVAL;
+	}
+
+	spin_lock_bh(&ht->ht_lock);
+	hlist_del_init(&node->hlist);
+	spin_unlock_bh(&ht->ht_lock);
+	kfree(node);
+
+	return 0;
+}
+
+struct net_device *mlx5i_pkey_get_netdev(struct net_device *netdev, u32 qpn)
+{
+	struct mlx5i_priv *ipriv = netdev_priv(netdev);
+	struct qpn_to_netdev *node;
+
+	node = mlx5i_find_qpn_to_netdev_node(ipriv->qpn_htbl->buckets, qpn);
+	if (!node)
+		return NULL;
+
+	return node->netdev;
+}
+
+static int mlx5i_pkey_open(struct net_device *netdev);
+static int mlx5i_pkey_close(struct net_device *netdev);
+static int mlx5i_pkey_dev_init(struct net_device *dev);
+static void mlx5i_pkey_dev_cleanup(struct net_device *netdev);
+static int mlx5i_pkey_change_mtu(struct net_device *netdev, int new_mtu);
+static int mlx5i_pkey_ioctl(struct net_device *dev, struct ifreq *ifr, int cmd);
+
+static const struct net_device_ops mlx5i_pkey_netdev_ops = {
+	.ndo_open                = mlx5i_pkey_open,
+	.ndo_stop                = mlx5i_pkey_close,
+	.ndo_init                = mlx5i_pkey_dev_init,
+	.ndo_get_stats64         = mlx5i_get_stats,
+	.ndo_uninit              = mlx5i_pkey_dev_cleanup,
+	.ndo_change_mtu          = mlx5i_pkey_change_mtu,
+	.ndo_eth_ioctl            = mlx5i_pkey_ioctl,
+};
+
+/* Child NDOs */
+static int mlx5i_pkey_dev_init(struct net_device *dev)
+{
+	struct mlx5e_priv *priv = mlx5i_epriv(dev);
+	struct mlx5i_priv *ipriv, *parent_ipriv;
+	struct net_device *parent_dev;
+
+	ipriv = priv->ppriv;
+
+	/* Link to parent */
+	parent_dev = mlx5i_parent_get(dev);
+	if (!parent_dev) {
+		mlx5_core_warn(priv->mdev, "failed to get parent device\n");
+		return -EINVAL;
+	}
+
+	if (dev->num_rx_queues < parent_dev->real_num_rx_queues) {
+		mlx5_core_warn(priv->mdev,
+			       "failed to create child device with rx queues [%d] less than parent's [%d]\n",
+			       dev->num_rx_queues,
+			       parent_dev->real_num_rx_queues);
+		mlx5i_parent_put(dev);
+		return -EINVAL;
+	}
+
+	/* Get QPN to netdevice hash table from parent */
+	parent_ipriv = netdev_priv(parent_dev);
+	ipriv->qpn_htbl = parent_ipriv->qpn_htbl;
+
+	return mlx5i_dev_init(dev);
+}
+
+static int mlx5i_pkey_ioctl(struct net_device *dev, struct ifreq *ifr, int cmd)
+{
+	return mlx5i_ioctl(dev, ifr, cmd);
+}
+
+static void mlx5i_pkey_dev_cleanup(struct net_device *netdev)
+{
+	mlx5i_parent_put(netdev);
+	return mlx5i_dev_cleanup(netdev);
+}
+
+static int mlx5i_pkey_open(struct net_device *netdev)
+{
+	struct mlx5e_priv *epriv = mlx5i_epriv(netdev);
+	struct mlx5i_priv *ipriv = epriv->ppriv;
+	struct mlx5_core_dev *mdev = epriv->mdev;
+	int err;
+
+	mutex_lock(&epriv->state_lock);
+
+	set_bit(MLX5E_STATE_OPENED, &epriv->state);
+
+	err = mlx5i_init_underlay_qp(epriv);
+	if (err) {
+		mlx5_core_warn(mdev, "prepare child underlay qp state failed, %d\n", err);
+		goto err_release_lock;
+	}
+
+	err = mlx5_fs_add_rx_underlay_qpn(mdev, ipriv->qpn);
+	if (err) {
+		mlx5_core_warn(mdev, "attach child underlay qp to ft failed, %d\n", err);
+		goto err_unint_underlay_qp;
+	}
+
+	err = mlx5i_create_tis(mdev, ipriv->qpn, &epriv->tisn[0][0]);
+	if (err) {
+		mlx5_core_warn(mdev, "create child tis failed, %d\n", err);
+		goto err_remove_rx_uderlay_qp;
+	}
+
+	err = mlx5e_open_channels(epriv, &epriv->channels);
+	if (err) {
+		mlx5_core_warn(mdev, "opening child channels failed, %d\n", err);
+		goto err_clear_state_opened_flag;
+	}
+	err = epriv->profile->update_rx(epriv);
+	if (err)
+		goto err_close_channels;
+	mlx5e_activate_priv_channels(epriv);
+	mutex_unlock(&epriv->state_lock);
+
+	return 0;
+
+err_close_channels:
+	mlx5e_close_channels(&epriv->channels);
+err_clear_state_opened_flag:
+	mlx5e_destroy_tis(mdev, epriv->tisn[0][0]);
+err_remove_rx_uderlay_qp:
+	mlx5_fs_remove_rx_underlay_qpn(mdev, ipriv->qpn);
+err_unint_underlay_qp:
+	mlx5i_uninit_underlay_qp(epriv);
+err_release_lock:
+	clear_bit(MLX5E_STATE_OPENED, &epriv->state);
+	mutex_unlock(&epriv->state_lock);
+	return err;
+}
+
+static int mlx5i_pkey_close(struct net_device *netdev)
+{
+	struct mlx5e_priv *priv = mlx5i_epriv(netdev);
+	struct mlx5i_priv *ipriv = priv->ppriv;
+	struct mlx5_core_dev *mdev = priv->mdev;
+
+	mutex_lock(&priv->state_lock);
+
+	if (!test_bit(MLX5E_STATE_OPENED, &priv->state))
+		goto unlock;
+
+	clear_bit(MLX5E_STATE_OPENED, &priv->state);
+
+	netif_carrier_off(priv->netdev);
+	mlx5_fs_remove_rx_underlay_qpn(mdev, ipriv->qpn);
+	mlx5i_uninit_underlay_qp(priv);
+	mlx5e_deactivate_priv_channels(priv);
+	mlx5e_close_channels(&priv->channels);
+	mlx5e_destroy_tis(mdev, priv->tisn[0][0]);
+unlock:
+	mutex_unlock(&priv->state_lock);
+	return 0;
+}
+
+static int mlx5i_pkey_change_mtu(struct net_device *netdev, int new_mtu)
+{
+	struct mlx5e_priv *priv = mlx5i_epriv(netdev);
+
+	mutex_lock(&priv->state_lock);
+	netdev->mtu = new_mtu;
+	mutex_unlock(&priv->state_lock);
+
+	return 0;
+}
+
+/* Called directly after IPoIB netdevice was created to initialize SW structs */
+static int mlx5i_pkey_init(struct mlx5_core_dev *mdev,
+			   struct net_device *netdev)
+{
+	struct mlx5e_priv *priv  = mlx5i_epriv(netdev);
+	int err;
+
+	err = mlx5i_init(mdev, netdev);
+	if (err)
+		return err;
+
+	/* Override parent ndo */
+	netdev->netdev_ops = &mlx5i_pkey_netdev_ops;
+
+	/* Set child limited ethtool support */
+	netdev->ethtool_ops = &mlx5i_pkey_ethtool_ops;
+
+	/* Use dummy rqs */
+	priv->channels.params.log_rq_mtu_frames = MLX5E_PARAMS_MINIMUM_LOG_RQ_SIZE;
+
+	return 0;
+}
+
+/* Called directly before IPoIB netdevice is destroyed to cleanup SW structs */
+static void mlx5i_pkey_cleanup(struct mlx5e_priv *priv)
+{
+	mlx5i_cleanup(priv);
+}
+
+static int mlx5i_pkey_init_tx(struct mlx5e_priv *priv)
+{
+	int err;
+
+	err = mlx5i_create_underlay_qp(priv);
+	if (err)
+		mlx5_core_warn(priv->mdev, "create child underlay QP failed, %d\n", err);
+
+	return err;
+}
+
+static void mlx5i_pkey_cleanup_tx(struct mlx5e_priv *priv)
+{
+	struct mlx5i_priv *ipriv = priv->ppriv;
+
+	mlx5i_destroy_underlay_qp(priv->mdev, ipriv->qpn);
+}
+
+static int mlx5i_pkey_init_rx(struct mlx5e_priv *priv)
+{
+	/* Since the rx resources are shared between child and parent, the
+	 * parent interface is taking care of rx resource allocation and init
+	 */
+	return 0;
+}
+
+static void mlx5i_pkey_cleanup_rx(struct mlx5e_priv *priv)
+{
+	/* Since the rx resources are shared between child and parent, the
+	 * parent interface is taking care of rx resource free and de-init
+	 */
+}
+
+static const struct mlx5e_profile mlx5i_pkey_nic_profile = {
+	.init		   = mlx5i_pkey_init,
+	.cleanup	   = mlx5i_pkey_cleanup,
+	.init_tx	   = mlx5i_pkey_init_tx,
+	.cleanup_tx	   = mlx5i_pkey_cleanup_tx,
+	.init_rx	   = mlx5i_pkey_init_rx,
+	.cleanup_rx	   = mlx5i_pkey_cleanup_rx,
+	.enable		   = NULL,
+	.disable	   = NULL,
+	.update_rx	   = mlx5i_update_nic_rx,
+	.update_stats	   = NULL,
+	.rx_handlers       = &mlx5i_rx_handlers,
+	.max_tc		   = MLX5I_MAX_NUM_TC,
+};
+
+const struct mlx5e_profile *mlx5i_pkey_get_profile(void)
+{
+	return &mlx5i_pkey_nic_profile;
+}