14 files changed, 144 insertions, 107 deletions

diff --git a/drivers/infiniband/hw/bnxt_re/qplib_fp.c b/drivers/infiniband/hw/bnxt_re/qplib_fp.c
index 439d0c7c5d..04258676d0 100644
--- a/drivers/infiniband/hw/bnxt_re/qplib_fp.c
+++ b/drivers/infiniband/hw/bnxt_re/qplib_fp.c
@@ -1013,7 +1013,8 @@ int bnxt_qplib_create_qp(struct bnxt_qplib_res *res, struct bnxt_qplib_qp *qp)
 	hwq_attr.stride = sizeof(struct sq_sge);
 	hwq_attr.depth = bnxt_qplib_get_depth(sq);
 	hwq_attr.aux_stride = psn_sz;
-	hwq_attr.aux_depth = bnxt_qplib_set_sq_size(sq, qp->wqe_mode);
+	hwq_attr.aux_depth = psn_sz ? bnxt_qplib_set_sq_size(sq, qp->wqe_mode)
+				    : 0;
 	/* Update msn tbl size */
 	if (BNXT_RE_HW_RETX(qp->dev_cap_flags) && psn_sz) {
 		hwq_attr.aux_depth = roundup_pow_of_two(bnxt_qplib_set_sq_size(sq, qp->wqe_mode));
diff --git a/drivers/infiniband/hw/hns/hns_roce_cq.c b/drivers/infiniband/hw/hns/hns_roce_cq.c
index 1b6d16af8c..2517c972c6 100644
--- a/drivers/infiniband/hw/hns/hns_roce_cq.c
+++ b/drivers/infiniband/hw/hns/hns_roce_cq.c
@@ -151,7 +151,7 @@ static int alloc_cqc(struct hns_roce_dev *hr_dev, struct hns_roce_cq *hr_cq)
 		return ret;
 	}
 
-	ret = xa_err(xa_store(&cq_table->array, hr_cq->cqn, hr_cq, GFP_KERNEL));
+	ret = xa_err(xa_store_irq(&cq_table->array, hr_cq->cqn, hr_cq, GFP_KERNEL));
 	if (ret) {
 		ibdev_err(ibdev, "failed to xa_store CQ, ret = %d.\n", ret);
 		goto err_put;
@@ -164,7 +164,7 @@ static int alloc_cqc(struct hns_roce_dev *hr_dev, struct hns_roce_cq *hr_cq)
 	return 0;
 
 err_xa:
-	xa_erase(&cq_table->array, hr_cq->cqn);
+	xa_erase_irq(&cq_table->array, hr_cq->cqn);
 err_put:
 	hns_roce_table_put(hr_dev, &cq_table->table, hr_cq->cqn);
 
@@ -183,7 +183,7 @@ static void free_cqc(struct hns_roce_dev *hr_dev, struct hns_roce_cq *hr_cq)
 		dev_err(dev, "DESTROY_CQ failed (%d) for CQN %06lx\n", ret,
 			hr_cq->cqn);
 
-	xa_erase(&cq_table->array, hr_cq->cqn);
+	xa_erase_irq(&cq_table->array, hr_cq->cqn);
 
 	/* Waiting interrupt process procedure carried out */
 	synchronize_irq(hr_dev->eq_table.eq[hr_cq->vector].irq);
@@ -477,13 +477,6 @@ void hns_roce_cq_event(struct hns_roce_dev *hr_dev, u32 cqn, int event_type)
 	struct ib_event event;
 	struct ib_cq *ibcq;
 
-	hr_cq = xa_load(&hr_dev->cq_table.array,
-			cqn & (hr_dev->caps.num_cqs - 1));
-	if (!hr_cq) {
-		dev_warn(dev, "async event for bogus CQ 0x%06x\n", cqn);
-		return;
-	}
-
 	if (event_type != HNS_ROCE_EVENT_TYPE_CQ_ID_INVALID &&
 	    event_type != HNS_ROCE_EVENT_TYPE_CQ_ACCESS_ERROR &&
 	    event_type != HNS_ROCE_EVENT_TYPE_CQ_OVERFLOW) {
@@ -492,7 +485,16 @@ void hns_roce_cq_event(struct hns_roce_dev *hr_dev, u32 cqn, int event_type)
 		return;
 	}
 
-	refcount_inc(&hr_cq->refcount);
+	xa_lock(&hr_dev->cq_table.array);
+	hr_cq = xa_load(&hr_dev->cq_table.array,
+			cqn & (hr_dev->caps.num_cqs - 1));
+	if (hr_cq)
+		refcount_inc(&hr_cq->refcount);
+	xa_unlock(&hr_dev->cq_table.array);
+	if (!hr_cq) {
+		dev_warn(dev, "async event for bogus CQ 0x%06x\n", cqn);
+		return;
+	}
 
 	ibcq = &hr_cq->ib_cq;
 	if (ibcq->event_handler) {
diff --git a/drivers/infiniband/hw/hns/hns_roce_hem.h b/drivers/infiniband/hw/hns/hns_roce_hem.h
index 7d23d3c51d..fea6d7d508 100644
--- a/drivers/infiniband/hw/hns/hns_roce_hem.h
+++ b/drivers/infiniband/hw/hns/hns_roce_hem.h
@@ -61,16 +61,16 @@ enum {
 	 (sizeof(struct scatterlist) + sizeof(void *)))
 
 #define check_whether_bt_num_3(type, hop_num) \
-	(type < HEM_TYPE_MTT && hop_num == 2)
+	((type) < HEM_TYPE_MTT && (hop_num) == 2)
 
 #define check_whether_bt_num_2(type, hop_num) \
-	((type < HEM_TYPE_MTT && hop_num == 1) || \
-	(type >= HEM_TYPE_MTT && hop_num == 2))
+	(((type) < HEM_TYPE_MTT && (hop_num) == 1) || \
+	((type) >= HEM_TYPE_MTT && (hop_num) == 2))
 
 #define check_whether_bt_num_1(type, hop_num) \
-	((type < HEM_TYPE_MTT && hop_num == HNS_ROCE_HOP_NUM_0) || \
-	(type >= HEM_TYPE_MTT && hop_num == 1) || \
-	(type >= HEM_TYPE_MTT && hop_num == HNS_ROCE_HOP_NUM_0))
+	(((type) < HEM_TYPE_MTT && (hop_num) == HNS_ROCE_HOP_NUM_0) || \
+	((type) >= HEM_TYPE_MTT && (hop_num) == 1) || \
+	((type) >= HEM_TYPE_MTT && (hop_num) == HNS_ROCE_HOP_NUM_0))
 
 struct hns_roce_hem_chunk {
 	struct list_head	 list;
diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
index 34e58e09b1..f95ec4618f 100644
--- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
+++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
@@ -2105,7 +2105,7 @@ static void apply_func_caps(struct hns_roce_dev *hr_dev)
 					 caps->gmv_bt_num *
 					 (HNS_HW_PAGE_SIZE / caps->gmv_entry_sz));
 
-		caps->gmv_entry_num = caps->gmv_bt_num * (PAGE_SIZE /
+		caps->gmv_entry_num = caps->gmv_bt_num * (HNS_HW_PAGE_SIZE /
 							  caps->gmv_entry_sz);
 	} else {
 		u32 func_num = max_t(u32, 1, hr_dev->func_num);
@@ -3715,8 +3715,9 @@ static void get_cqe_status(struct hns_roce_dev *hr_dev, struct hns_roce_qp *qp,
 		   wc->status == IB_WC_WR_FLUSH_ERR))
 		return;
 
-	ibdev_err(&hr_dev->ib_dev, "error cqe status 0x%x:\n", cqe_status);
-	print_hex_dump(KERN_ERR, "", DUMP_PREFIX_NONE, 16, 4, cqe,
+	ibdev_err_ratelimited(&hr_dev->ib_dev, "error cqe status 0x%x:\n",
+			      cqe_status);
+	print_hex_dump(KERN_DEBUG, "", DUMP_PREFIX_NONE, 16, 4, cqe,
 		       cq->cqe_size, false);
 	wc->vendor_err = hr_reg_read(cqe, CQE_SUB_STATUS);
 
diff --git a/drivers/infiniband/hw/hns/hns_roce_main.c b/drivers/infiniband/hw/hns/hns_roce_main.c
index b55fe6911f..a33d3cedbc 100644
--- a/drivers/infiniband/hw/hns/hns_roce_main.c
+++ b/drivers/infiniband/hw/hns/hns_roce_main.c
@@ -37,6 +37,7 @@
 #include <rdma/ib_smi.h>
 #include <rdma/ib_user_verbs.h>
 #include <rdma/ib_cache.h>
+#include "hnae3.h"
 #include "hns_roce_common.h"
 #include "hns_roce_device.h"
 #include "hns_roce_hem.h"
diff --git a/drivers/infiniband/hw/hns/hns_roce_mr.c b/drivers/infiniband/hw/hns/hns_roce_mr.c
index 91cd580480..0d42fd197c 100644
--- a/drivers/infiniband/hw/hns/hns_roce_mr.c
+++ b/drivers/infiniband/hw/hns/hns_roce_mr.c
@@ -433,18 +433,18 @@ int hns_roce_map_mr_sg(struct ib_mr *ibmr, struct scatterlist *sg, int sg_nents,
 	struct ib_device *ibdev = &hr_dev->ib_dev;
 	struct hns_roce_mr *mr = to_hr_mr(ibmr);
 	struct hns_roce_mtr *mtr = &mr->pbl_mtr;
-	int ret = 0;
+	int ret, sg_num = 0;
 
 	mr->npages = 0;
 	mr->page_list = kvcalloc(mr->pbl_mtr.hem_cfg.buf_pg_count,
 				 sizeof(dma_addr_t), GFP_KERNEL);
 	if (!mr->page_list)
-		return ret;
+		return sg_num;
 
-	ret = ib_sg_to_pages(ibmr, sg, sg_nents, sg_offset, hns_roce_set_page);
-	if (ret < 1) {
+	sg_num = ib_sg_to_pages(ibmr, sg, sg_nents, sg_offset, hns_roce_set_page);
+	if (sg_num < 1) {
 		ibdev_err(ibdev, "failed to store sg pages %u %u, cnt = %d.\n",
-			  mr->npages, mr->pbl_mtr.hem_cfg.buf_pg_count, ret);
+			  mr->npages, mr->pbl_mtr.hem_cfg.buf_pg_count, sg_num);
 		goto err_page_list;
 	}
 
@@ -455,17 +455,16 @@ int hns_roce_map_mr_sg(struct ib_mr *ibmr, struct scatterlist *sg, int sg_nents,
 	ret = hns_roce_mtr_map(hr_dev, mtr, mr->page_list, mr->npages);
 	if (ret) {
 		ibdev_err(ibdev, "failed to map sg mtr, ret = %d.\n", ret);
-		ret = 0;
+		sg_num = 0;
 	} else {
 		mr->pbl_mtr.hem_cfg.buf_pg_shift = (u32)ilog2(ibmr->page_size);
-		ret = mr->npages;
 	}
 
 err_page_list:
 	kvfree(mr->page_list);
 	mr->page_list = NULL;
 
-	return ret;
+	return sg_num;
 }
 
 static void hns_roce_mw_free(struct hns_roce_dev *hr_dev,
diff --git a/drivers/infiniband/hw/hns/hns_roce_srq.c b/drivers/infiniband/hw/hns/hns_roce_srq.c
index 4abae94778..8f48c6723e 100644
--- a/drivers/infiniband/hw/hns/hns_roce_srq.c
+++ b/drivers/infiniband/hw/hns/hns_roce_srq.c
@@ -123,7 +123,7 @@ static int alloc_srqc(struct hns_roce_dev *hr_dev, struct hns_roce_srq *srq)
 		return ret;
 	}
 
-	ret = xa_err(xa_store(&srq_table->xa, srq->srqn, srq, GFP_KERNEL));
+	ret = xa_err(xa_store_irq(&srq_table->xa, srq->srqn, srq, GFP_KERNEL));
 	if (ret) {
 		ibdev_err(ibdev, "failed to store SRQC, ret = %d.\n", ret);
 		goto err_put;
@@ -136,7 +136,7 @@ static int alloc_srqc(struct hns_roce_dev *hr_dev, struct hns_roce_srq *srq)
 	return 0;
 
 err_xa:
-	xa_erase(&srq_table->xa, srq->srqn);
+	xa_erase_irq(&srq_table->xa, srq->srqn);
 err_put:
 	hns_roce_table_put(hr_dev, &srq_table->table, srq->srqn);
 
@@ -154,7 +154,7 @@ static void free_srqc(struct hns_roce_dev *hr_dev, struct hns_roce_srq *srq)
 		dev_err(hr_dev->dev, "DESTROY_SRQ failed (%d) for SRQN %06lx\n",
 			ret, srq->srqn);
 
-	xa_erase(&srq_table->xa, srq->srqn);
+	xa_erase_irq(&srq_table->xa, srq->srqn);
 
 	if (refcount_dec_and_test(&srq->refcount))
 		complete(&srq->free);
diff --git a/drivers/infiniband/hw/mana/cq.c b/drivers/infiniband/hw/mana/cq.c
index 4a71e678d0..89fcc09ded 100644
--- a/drivers/infiniband/hw/mana/cq.c
+++ b/drivers/infiniband/hw/mana/cq.c
@@ -39,37 +39,13 @@ int mana_ib_create_cq(struct ib_cq *ibcq, const struct ib_cq_init_attr *attr,
 	}
 
 	cq->cqe = attr->cqe;
-	cq->umem = ib_umem_get(ibdev, ucmd.buf_addr, cq->cqe * COMP_ENTRY_SIZE,
-			       IB_ACCESS_LOCAL_WRITE);
-	if (IS_ERR(cq->umem)) {
-		err = PTR_ERR(cq->umem);
-		ibdev_dbg(ibdev, "Failed to get umem for create cq, err %d\n",
-			  err);
-		return err;
-	}
-
-	err = mana_ib_create_zero_offset_dma_region(mdev, cq->umem, &cq->gdma_region);
+	err = mana_ib_create_queue(mdev, ucmd.buf_addr, cq->cqe * COMP_ENTRY_SIZE, &cq->queue);
 	if (err) {
-		ibdev_dbg(ibdev,
-			  "Failed to create dma region for create cq, %d\n",
-			  err);
-		goto err_release_umem;
+		ibdev_dbg(ibdev, "Failed to create queue for create cq, %d\n", err);
+		return err;
 	}
 
-	ibdev_dbg(ibdev,
-		  "create_dma_region ret %d gdma_region 0x%llx\n",
-		  err, cq->gdma_region);
-
-	/*
-	 * The CQ ID is not known at this time. The ID is generated at create_qp
-	 */
-	cq->id = INVALID_QUEUE_ID;
-
 	return 0;
-
-err_release_umem:
-	ib_umem_release(cq->umem);
-	return err;
 }
 
 int mana_ib_destroy_cq(struct ib_cq *ibcq, struct ib_udata *udata)
@@ -78,24 +54,16 @@ int mana_ib_destroy_cq(struct ib_cq *ibcq, struct ib_udata *udata)
 	struct ib_device *ibdev = ibcq->device;
 	struct mana_ib_dev *mdev;
 	struct gdma_context *gc;
-	int err;
 
 	mdev = container_of(ibdev, struct mana_ib_dev, ib_dev);
 	gc = mdev_to_gc(mdev);
 
-	err = mana_ib_gd_destroy_dma_region(mdev, cq->gdma_region);
-	if (err) {
-		ibdev_dbg(ibdev,
-			  "Failed to destroy dma region, %d\n", err);
-		return err;
+	if (cq->queue.id != INVALID_QUEUE_ID) {
+		kfree(gc->cq_table[cq->queue.id]);
+		gc->cq_table[cq->queue.id] = NULL;
 	}
 
-	if (cq->id != INVALID_QUEUE_ID) {
-		kfree(gc->cq_table[cq->id]);
-		gc->cq_table[cq->id] = NULL;
-	}
-
-	ib_umem_release(cq->umem);
+	mana_ib_destroy_queue(mdev, &cq->queue);
 
 	return 0;
 }
@@ -113,8 +81,10 @@ int mana_ib_install_cq_cb(struct mana_ib_dev *mdev, struct mana_ib_cq *cq)
 	struct gdma_context *gc = mdev_to_gc(mdev);
 	struct gdma_queue *gdma_cq;
 
+	if (cq->queue.id >= gc->max_num_cqs)
+		return -EINVAL;
 	/* Create CQ table entry */
-	WARN_ON(gc->cq_table[cq->id]);
+	WARN_ON(gc->cq_table[cq->queue.id]);
 	gdma_cq = kzalloc(sizeof(*gdma_cq), GFP_KERNEL);
 	if (!gdma_cq)
 		return -ENOMEM;
@@ -122,7 +92,7 @@ int mana_ib_install_cq_cb(struct mana_ib_dev *mdev, struct mana_ib_cq *cq)
 	gdma_cq->cq.context = cq;
 	gdma_cq->type = GDMA_CQ;
 	gdma_cq->cq.callback = mana_ib_cq_handler;
-	gdma_cq->id = cq->id;
-	gc->cq_table[cq->id] = gdma_cq;
+	gdma_cq->id = cq->queue.id;
+	gc->cq_table[cq->queue.id] = gdma_cq;
 	return 0;
 }
diff --git a/drivers/infiniband/hw/mana/main.c b/drivers/infiniband/hw/mana/main.c
index 71e33feee6..4524c6b807 100644
--- a/drivers/infiniband/hw/mana/main.c
+++ b/drivers/infiniband/hw/mana/main.c
@@ -237,6 +237,49 @@ void mana_ib_dealloc_ucontext(struct ib_ucontext *ibcontext)
 		ibdev_dbg(ibdev, "Failed to destroy doorbell page %d\n", ret);
 }
 
+int mana_ib_create_queue(struct mana_ib_dev *mdev, u64 addr, u32 size,
+			 struct mana_ib_queue *queue)
+{
+	struct ib_umem *umem;
+	int err;
+
+	queue->umem = NULL;
+	queue->id = INVALID_QUEUE_ID;
+	queue->gdma_region = GDMA_INVALID_DMA_REGION;
+
+	umem = ib_umem_get(&mdev->ib_dev, addr, size, IB_ACCESS_LOCAL_WRITE);
+	if (IS_ERR(umem)) {
+		err = PTR_ERR(umem);
+		ibdev_dbg(&mdev->ib_dev, "Failed to get umem, %d\n", err);
+		return err;
+	}
+
+	err = mana_ib_create_zero_offset_dma_region(mdev, umem, &queue->gdma_region);
+	if (err) {
+		ibdev_dbg(&mdev->ib_dev, "Failed to create dma region, %d\n", err);
+		goto free_umem;
+	}
+	queue->umem = umem;
+
+	ibdev_dbg(&mdev->ib_dev,
+		  "create_dma_region ret %d gdma_region 0x%llx\n",
+		  err, queue->gdma_region);
+
+	return 0;
+free_umem:
+	ib_umem_release(umem);
+	return err;
+}
+
+void mana_ib_destroy_queue(struct mana_ib_dev *mdev, struct mana_ib_queue *queue)
+{
+	/* Ignore return code as there is not much we can do about it.
+	 * The error message is printed inside.
+	 */
+	mana_ib_gd_destroy_dma_region(mdev, queue->gdma_region);
+	ib_umem_release(queue->umem);
+}
+
 static int
 mana_ib_gd_first_dma_region(struct mana_ib_dev *dev,
 			    struct gdma_context *gc,
diff --git a/drivers/infiniband/hw/mana/mana_ib.h b/drivers/infiniband/hw/mana/mana_ib.h
index f83390eebb..6acb5c281c 100644
--- a/drivers/infiniband/hw/mana/mana_ib.h
+++ b/drivers/infiniband/hw/mana/mana_ib.h
@@ -45,6 +45,12 @@ struct mana_ib_adapter_caps {
 	u32 max_inline_data_size;
 };
 
+struct mana_ib_queue {
+	struct ib_umem *umem;
+	u64 gdma_region;
+	u64 id;
+};
+
 struct mana_ib_dev {
 	struct ib_device ib_dev;
 	struct gdma_dev *gdma_dev;
@@ -82,10 +88,8 @@ struct mana_ib_mr {
 
 struct mana_ib_cq {
 	struct ib_cq ibcq;
-	struct ib_umem *umem;
+	struct mana_ib_queue queue;
 	int cqe;
-	u64 gdma_region;
-	u64 id;
 	u32 comp_vector;
 };
 
@@ -169,6 +173,10 @@ int mana_ib_create_dma_region(struct mana_ib_dev *dev, struct ib_umem *umem,
 int mana_ib_gd_destroy_dma_region(struct mana_ib_dev *dev,
 				  mana_handle_t gdma_region);
 
+int mana_ib_create_queue(struct mana_ib_dev *mdev, u64 addr, u32 size,
+			 struct mana_ib_queue *queue);
+void mana_ib_destroy_queue(struct mana_ib_dev *mdev, struct mana_ib_queue *queue);
+
 struct ib_wq *mana_ib_create_wq(struct ib_pd *pd,
 				struct ib_wq_init_attr *init_attr,
 				struct ib_udata *udata);
diff --git a/drivers/infiniband/hw/mana/qp.c b/drivers/infiniband/hw/mana/qp.c
index 6e7627745c..d7485ee6a6 100644
--- a/drivers/infiniband/hw/mana/qp.c
+++ b/drivers/infiniband/hw/mana/qp.c
@@ -197,7 +197,7 @@ static int mana_ib_create_qp_rss(struct ib_qp *ibqp, struct ib_pd *pd,
 		wq_spec.gdma_region = wq->gdma_region;
 		wq_spec.queue_size = wq->wq_buf_size;
 
-		cq_spec.gdma_region = cq->gdma_region;
+		cq_spec.gdma_region = cq->queue.gdma_region;
 		cq_spec.queue_size = cq->cqe * COMP_ENTRY_SIZE;
 		cq_spec.modr_ctx_id = 0;
 		eq = &mpc->ac->eqs[cq->comp_vector % gc->max_num_queues];
@@ -213,16 +213,16 @@ static int mana_ib_create_qp_rss(struct ib_qp *ibqp, struct ib_pd *pd,
 
 		/* The GDMA regions are now owned by the WQ object */
 		wq->gdma_region = GDMA_INVALID_DMA_REGION;
-		cq->gdma_region = GDMA_INVALID_DMA_REGION;
+		cq->queue.gdma_region = GDMA_INVALID_DMA_REGION;
 
 		wq->id = wq_spec.queue_index;
-		cq->id = cq_spec.queue_index;
+		cq->queue.id = cq_spec.queue_index;
 
 		ibdev_dbg(&mdev->ib_dev,
 			  "ret %d rx_object 0x%llx wq id %llu cq id %llu\n",
-			  ret, wq->rx_object, wq->id, cq->id);
+			  ret, wq->rx_object, wq->id, cq->queue.id);
 
-		resp.entries[i].cqid = cq->id;
+		resp.entries[i].cqid = cq->queue.id;
 		resp.entries[i].wqid = wq->id;
 
 		mana_ind_table[i] = wq->rx_object;
@@ -232,7 +232,7 @@ static int mana_ib_create_qp_rss(struct ib_qp *ibqp, struct ib_pd *pd,
 		if (ret)
 			goto fail;
 
-		gdma_cq_allocated[i] = gc->cq_table[cq->id];
+		gdma_cq_allocated[i] = gc->cq_table[cq->queue.id];
 	}
 	resp.num_entries = i;
 
@@ -264,7 +264,7 @@ fail:
 		wq = container_of(ibwq, struct mana_ib_wq, ibwq);
 		cq = container_of(ibcq, struct mana_ib_cq, ibcq);
 
-		gc->cq_table[cq->id] = NULL;
+		gc->cq_table[cq->queue.id] = NULL;
 		kfree(gdma_cq_allocated[i]);
 
 		mana_destroy_wq_obj(mpc, GDMA_RQ, wq->rx_object);
@@ -374,7 +374,7 @@ static int mana_ib_create_qp_raw(struct ib_qp *ibqp, struct ib_pd *ibpd,
 	wq_spec.gdma_region = qp->sq_gdma_region;
 	wq_spec.queue_size = ucmd.sq_buf_size;
 
-	cq_spec.gdma_region = send_cq->gdma_region;
+	cq_spec.gdma_region = send_cq->queue.gdma_region;
 	cq_spec.queue_size = send_cq->cqe * COMP_ENTRY_SIZE;
 	cq_spec.modr_ctx_id = 0;
 	eq_vec = send_cq->comp_vector % gc->max_num_queues;
@@ -392,10 +392,10 @@ static int mana_ib_create_qp_raw(struct ib_qp *ibqp, struct ib_pd *ibpd,
 
 	/* The GDMA regions are now owned by the WQ object */
 	qp->sq_gdma_region = GDMA_INVALID_DMA_REGION;
-	send_cq->gdma_region = GDMA_INVALID_DMA_REGION;
+	send_cq->queue.gdma_region = GDMA_INVALID_DMA_REGION;
 
 	qp->sq_id = wq_spec.queue_index;
-	send_cq->id = cq_spec.queue_index;
+	send_cq->queue.id = cq_spec.queue_index;
 
 	/* Create CQ table entry */
 	err = mana_ib_install_cq_cb(mdev, send_cq);
@@ -404,10 +404,10 @@ static int mana_ib_create_qp_raw(struct ib_qp *ibqp, struct ib_pd *ibpd,
 
 	ibdev_dbg(&mdev->ib_dev,
 		  "ret %d qp->tx_object 0x%llx sq id %llu cq id %llu\n", err,
-		  qp->tx_object, qp->sq_id, send_cq->id);
+		  qp->tx_object, qp->sq_id, send_cq->queue.id);
 
 	resp.sqid = qp->sq_id;
-	resp.cqid = send_cq->id;
+	resp.cqid = send_cq->queue.id;
 	resp.tx_vp_offset = pd->tx_vp_offset;
 
 	err = ib_copy_to_udata(udata, &resp, sizeof(resp));
@@ -422,7 +422,7 @@ static int mana_ib_create_qp_raw(struct ib_qp *ibqp, struct ib_pd *ibpd,
 
 err_release_gdma_cq:
 	kfree(gdma_cq);
-	gc->cq_table[send_cq->id] = NULL;
+	gc->cq_table[send_cq->queue.id] = NULL;
 
 err_destroy_wq_obj:
 	mana_destroy_wq_obj(mpc, GDMA_SQ, qp->tx_object);
diff --git a/drivers/infiniband/hw/mlx5/mem.c b/drivers/infiniband/hw/mlx5/mem.c
index 96ffbbaf0a..5a22be14d9 100644
--- a/drivers/infiniband/hw/mlx5/mem.c
+++ b/drivers/infiniband/hw/mlx5/mem.c
@@ -30,6 +30,7 @@
  * SOFTWARE.
  */
 
+#include <linux/io.h>
 #include <rdma/ib_umem_odp.h>
 #include "mlx5_ib.h"
 #include <linux/jiffies.h>
@@ -108,7 +109,6 @@ static int post_send_nop(struct mlx5_ib_dev *dev, struct ib_qp *ibqp, u64 wr_id,
 	__be32 mmio_wqe[16] = {};
 	unsigned long flags;
 	unsigned int idx;
-	int i;
 
 	if (unlikely(dev->mdev->state == MLX5_DEVICE_STATE_INTERNAL_ERROR))
 		return -EIO;
@@ -148,10 +148,8 @@ static int post_send_nop(struct mlx5_ib_dev *dev, struct ib_qp *ibqp, u64 wr_id,
 	 * we hit doorbell
 	 */
 	wmb();
-	for (i = 0; i < 8; i++)
-		mlx5_write64(&mmio_wqe[i * 2],
-			     bf->bfreg->map + bf->offset + i * 8);
-	io_stop_wc();
+	__iowrite64_copy(bf->bfreg->map + bf->offset, mmio_wqe,
+			 sizeof(mmio_wqe) / 8);
 
 	bf->offset ^= bf->buf_size;
 
diff --git a/drivers/infiniband/hw/mlx5/mlx5_ib.h b/drivers/infiniband/hw/mlx5/mlx5_ib.h
index bbe79b86c7..79ebafecca 100644
--- a/drivers/infiniband/hw/mlx5/mlx5_ib.h
+++ b/drivers/infiniband/hw/mlx5/mlx5_ib.h
@@ -643,9 +643,10 @@ struct mlx5_ib_mkey {
 	unsigned int ndescs;
 	struct wait_queue_head wait;
 	refcount_t usecount;
-	/* User Mkey must hold either a rb_key or a cache_ent. */
+	/* Cacheable user Mkey must hold either a rb_key or a cache_ent. */
 	struct mlx5r_cache_rb_key rb_key;
 	struct mlx5_cache_ent *cache_ent;
+	u8 cacheable : 1;
 };
 
 #define MLX5_IB_MTT_PRESENT (MLX5_IB_MTT_READ | MLX5_IB_MTT_WRITE)
diff --git a/drivers/infiniband/hw/mlx5/mr.c b/drivers/infiniband/hw/mlx5/mr.c
index a8ee2ca1f4..ecc111ed5d 100644
--- a/drivers/infiniband/hw/mlx5/mr.c
+++ b/drivers/infiniband/hw/mlx5/mr.c
@@ -1158,6 +1158,7 @@ static struct mlx5_ib_mr *alloc_cacheable_mr(struct ib_pd *pd,
 		if (IS_ERR(mr))
 			return mr;
 		mr->mmkey.rb_key = rb_key;
+		mr->mmkey.cacheable = true;
 		return mr;
 	}
 
@@ -1168,6 +1169,7 @@ static struct mlx5_ib_mr *alloc_cacheable_mr(struct ib_pd *pd,
 	mr->ibmr.pd = pd;
 	mr->umem = umem;
 	mr->page_shift = order_base_2(page_size);
+	mr->mmkey.cacheable = true;
 	set_mr_fields(dev, mr, umem->length, access_flags, iova);
 
 	return mr;
@@ -1570,7 +1572,8 @@ static bool can_use_umr_rereg_access(struct mlx5_ib_dev *dev,
 	unsigned int diffs = current_access_flags ^ target_access_flags;
 
 	if (diffs & ~(IB_ACCESS_LOCAL_WRITE | IB_ACCESS_REMOTE_WRITE |
-		      IB_ACCESS_REMOTE_READ | IB_ACCESS_RELAXED_ORDERING))
+		      IB_ACCESS_REMOTE_READ | IB_ACCESS_RELAXED_ORDERING |
+		      IB_ACCESS_REMOTE_ATOMIC))
 		return false;
 	return mlx5r_umr_can_reconfig(dev, current_access_flags,
 				      target_access_flags);
@@ -1835,6 +1838,23 @@ end:
 	return ret;
 }
 
+static int mlx5_revoke_mr(struct mlx5_ib_mr *mr)
+{
+	struct mlx5_ib_dev *dev = to_mdev(mr->ibmr.device);
+	struct mlx5_cache_ent *ent = mr->mmkey.cache_ent;
+
+	if (mr->mmkey.cacheable && !mlx5r_umr_revoke_mr(mr) && !cache_ent_find_and_store(dev, mr))
+		return 0;
+
+	if (ent) {
+		spin_lock_irq(&ent->mkeys_queue.lock);
+		ent->in_use--;
+		mr->mmkey.cache_ent = NULL;
+		spin_unlock_irq(&ent->mkeys_queue.lock);
+	}
+	return destroy_mkey(dev, mr);
+}
+
 int mlx5_ib_dereg_mr(struct ib_mr *ibmr, struct ib_udata *udata)
 {
 	struct mlx5_ib_mr *mr = to_mmr(ibmr);
@@ -1880,16 +1900,9 @@ int mlx5_ib_dereg_mr(struct ib_mr *ibmr, struct ib_udata *udata)
 	}
 
 	/* Stop DMA */
-	if (mr->umem && mlx5r_umr_can_load_pas(dev, mr->umem->length))
-		if (mlx5r_umr_revoke_mr(mr) ||
-		    cache_ent_find_and_store(dev, mr))
-			mr->mmkey.cache_ent = NULL;
-
-	if (!mr->mmkey.cache_ent) {
-		rc = destroy_mkey(to_mdev(mr->ibmr.device), mr);
-		if (rc)
-			return rc;
-	}
+	rc = mlx5_revoke_mr(mr);
+	if (rc)
+		return rc;
 
 	if (mr->umem) {
 		bool is_odp = is_odp_mr(mr);