1 files changed, 529 insertions, 166 deletions

diff --git a/drivers/infiniband/hw/efa/efa_verbs.c b/drivers/infiniband/hw/efa/efa_verbs.c
index ec5545870554..31454643f8c5 100644
--- a/drivers/infiniband/hw/efa/efa_verbs.c
+++ b/drivers/infiniband/hw/efa/efa_verbs.c
@@ -1,9 +1,12 @@
 // SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
 /*
- * Copyright 2018-2019 Amazon.com, Inc. or its affiliates. All rights reserved.
+ * Copyright 2018-2022 Amazon.com, Inc. or its affiliates. All rights reserved.
  */
 
+#include <linux/dma-buf.h>
+#include <linux/dma-resv.h>
 #include <linux/vmalloc.h>
+#include <linux/log2.h>
 
 #include <rdma/ib_addr.h>
 #include <rdma/ib_umem.h>
@@ -12,6 +15,7 @@
 #include <rdma/uverbs_ioctl.h>
 
 #include "efa.h"
+#include "efa_io_defs.h"
 
 enum {
 	EFA_MMAP_DMA_PAGE = 0,
@@ -29,31 +33,53 @@ struct efa_user_mmap_entry {
 	u8 mmap_flag;
 };
 
-#define EFA_DEFINE_STATS(op) \
-	op(EFA_TX_BYTES, "tx_bytes") \
-	op(EFA_TX_PKTS, "tx_pkts") \
-	op(EFA_RX_BYTES, "rx_bytes") \
-	op(EFA_RX_PKTS, "rx_pkts") \
-	op(EFA_RX_DROPS, "rx_drops") \
+#define EFA_DEFINE_DEVICE_STATS(op) \
 	op(EFA_SUBMITTED_CMDS, "submitted_cmds") \
 	op(EFA_COMPLETED_CMDS, "completed_cmds") \
+	op(EFA_CMDS_ERR, "cmds_err") \
 	op(EFA_NO_COMPLETION_CMDS, "no_completion_cmds") \
 	op(EFA_KEEP_ALIVE_RCVD, "keep_alive_rcvd") \
 	op(EFA_ALLOC_PD_ERR, "alloc_pd_err") \
 	op(EFA_CREATE_QP_ERR, "create_qp_err") \
+	op(EFA_CREATE_CQ_ERR, "create_cq_err") \
 	op(EFA_REG_MR_ERR, "reg_mr_err") \
 	op(EFA_ALLOC_UCONTEXT_ERR, "alloc_ucontext_err") \
-	op(EFA_CREATE_AH_ERR, "create_ah_err")
+	op(EFA_CREATE_AH_ERR, "create_ah_err") \
+	op(EFA_MMAP_ERR, "mmap_err")
+
+#define EFA_DEFINE_PORT_STATS(op) \
+	op(EFA_TX_BYTES, "tx_bytes") \
+	op(EFA_TX_PKTS, "tx_pkts") \
+	op(EFA_RX_BYTES, "rx_bytes") \
+	op(EFA_RX_PKTS, "rx_pkts") \
+	op(EFA_RX_DROPS, "rx_drops") \
+	op(EFA_SEND_BYTES, "send_bytes") \
+	op(EFA_SEND_WRS, "send_wrs") \
+	op(EFA_RECV_BYTES, "recv_bytes") \
+	op(EFA_RECV_WRS, "recv_wrs") \
+	op(EFA_RDMA_READ_WRS, "rdma_read_wrs") \
+	op(EFA_RDMA_READ_BYTES, "rdma_read_bytes") \
+	op(EFA_RDMA_READ_WR_ERR, "rdma_read_wr_err") \
+	op(EFA_RDMA_READ_RESP_BYTES, "rdma_read_resp_bytes") \
 
 #define EFA_STATS_ENUM(ename, name) ename,
-#define EFA_STATS_STR(ename, name) [ename] = name,
+#define EFA_STATS_STR(ename, nam) \
+	[ename].name = nam,
+
+enum efa_hw_device_stats {
+	EFA_DEFINE_DEVICE_STATS(EFA_STATS_ENUM)
+};
+
+static const struct rdma_stat_desc efa_device_stats_descs[] = {
+	EFA_DEFINE_DEVICE_STATS(EFA_STATS_STR)
+};
 
-enum efa_hw_stats {
-	EFA_DEFINE_STATS(EFA_STATS_ENUM)
+enum efa_hw_port_stats {
+	EFA_DEFINE_PORT_STATS(EFA_STATS_ENUM)
 };
 
-static const char *const efa_stats_names[] = {
-	EFA_DEFINE_STATS(EFA_STATS_STR)
+static const struct rdma_stat_desc efa_port_stats_descs[] = {
+	EFA_DEFINE_PORT_STATS(EFA_STATS_STR)
 };
 
 #define EFA_CHUNK_PAYLOAD_SHIFT       12
@@ -139,13 +165,9 @@ to_emmap(struct rdma_user_mmap_entry *rdma_entry)
 	return container_of(rdma_entry, struct efa_user_mmap_entry, rdma_entry);
 }
 
-static inline bool is_rdma_read_cap(struct efa_dev *dev)
-{
-	return dev->dev_attr.device_caps & EFA_ADMIN_FEATURE_DEVICE_ATTR_DESC_RDMA_READ_MASK;
-}
-
-#define field_avail(x, fld, sz) (offsetof(typeof(x), fld) + \
-				 sizeof_field(typeof(x), fld) <= (sz))
+#define EFA_DEV_CAP(dev, cap) \
+	((dev)->dev_attr.device_caps & \
+	 EFA_ADMIN_FEATURE_DEVICE_ATTR_DESC_##cap##_MASK)
 
 #define is_reserved_cleared(reserved) \
 	!memchr_inv(reserved, 0, sizeof(reserved))
@@ -169,6 +191,14 @@ static void *efa_zalloc_mapped(struct efa_dev *dev, dma_addr_t *dma_addr,
 	return addr;
 }
 
+static void efa_free_mapped(struct efa_dev *dev, void *cpu_addr,
+			    dma_addr_t dma_addr,
+			    size_t size, enum dma_data_direction dir)
+{
+	dma_unmap_single(&dev->pdev->dev, dma_addr, size, dir);
+	free_pages_exact(cpu_addr, size);
+}
+
 int efa_query_device(struct ib_device *ibdev,
 		     struct ib_device_attr *props,
 		     struct ib_udata *udata)
@@ -204,6 +234,7 @@ int efa_query_device(struct ib_device *ibdev,
 	props->max_send_sge = dev_attr->max_sq_sge;
 	props->max_recv_sge = dev_attr->max_rq_sge;
 	props->max_sge_rd = dev_attr->max_wr_rdma_sge;
+	props->max_pkeys = 1;
 
 	if (udata && udata->outlen) {
 		resp.max_sq_sge = dev_attr->max_sq_sge;
@@ -212,9 +243,16 @@ int efa_query_device(struct ib_device *ibdev,
 		resp.max_rq_wr = dev_attr->max_rq_depth;
 		resp.max_rdma_size = dev_attr->max_rdma_size;
 
-		if (is_rdma_read_cap(dev))
+		resp.device_caps |= EFA_QUERY_DEVICE_CAPS_CQ_WITH_SGID;
+		if (EFA_DEV_CAP(dev, RDMA_READ))
 			resp.device_caps |= EFA_QUERY_DEVICE_CAPS_RDMA_READ;
 
+		if (EFA_DEV_CAP(dev, RNR_RETRY))
+			resp.device_caps |= EFA_QUERY_DEVICE_CAPS_RNR_RETRY;
+
+		if (dev->neqs)
+			resp.device_caps |= EFA_QUERY_DEVICE_CAPS_CQ_NOTIFICATIONS;
+
 		err = ib_copy_to_udata(udata, &resp,
 				       min(sizeof(resp), udata->outlen));
 		if (err) {
@@ -227,7 +265,7 @@ int efa_query_device(struct ib_device *ibdev,
 	return 0;
 }
 
-int efa_query_port(struct ib_device *ibdev, u8 port,
+int efa_query_port(struct ib_device *ibdev, u32 port,
 		   struct ib_port_attr *props)
 {
 	struct efa_dev *dev = to_edev(ibdev);
@@ -260,7 +298,7 @@ int efa_query_qp(struct ib_qp *ibqp, struct ib_qp_attr *qp_attr,
 
 #define EFA_QUERY_QP_SUPP_MASK \
 	(IB_QP_STATE | IB_QP_PKEY_INDEX | IB_QP_PORT | \
-	 IB_QP_QKEY | IB_QP_SQ_PSN | IB_QP_CAP)
+	 IB_QP_QKEY | IB_QP_SQ_PSN | IB_QP_CAP | IB_QP_RNR_RETRY)
 
 	if (qp_attr_mask & ~EFA_QUERY_QP_SUPP_MASK) {
 		ibdev_dbg(&dev->ibdev,
@@ -282,6 +320,7 @@ int efa_query_qp(struct ib_qp *ibqp, struct ib_qp_attr *qp_attr,
 	qp_attr->sq_psn = result.sq_psn;
 	qp_attr->sq_draining = result.sq_draining;
 	qp_attr->port_num = 1;
+	qp_attr->rnr_retry = result.rnr_retry;
 
 	qp_attr->cap.max_send_wr = qp->max_send_wr;
 	qp_attr->cap.max_recv_wr = qp->max_recv_wr;
@@ -298,7 +337,7 @@ int efa_query_qp(struct ib_qp *ibqp, struct ib_qp_attr *qp_attr,
 	return 0;
 }
 
-int efa_query_gid(struct ib_device *ibdev, u8 port, int index,
+int efa_query_gid(struct ib_device *ibdev, u32 port, int index,
 		  union ib_gid *gid)
 {
 	struct efa_dev *dev = to_edev(ibdev);
@@ -308,7 +347,7 @@ int efa_query_gid(struct ib_device *ibdev, u8 port, int index,
 	return 0;
 }
 
-int efa_query_pkey(struct ib_device *ibdev, u8 port, u16 index,
+int efa_query_pkey(struct ib_device *ibdev, u32 port, u16 index,
 		   u16 *pkey)
 {
 	if (index > 0)
@@ -367,17 +406,18 @@ int efa_alloc_pd(struct ib_pd *ibpd, struct ib_udata *udata)
 err_dealloc_pd:
 	efa_pd_dealloc(dev, result.pdn);
 err_out:
-	atomic64_inc(&dev->stats.sw_stats.alloc_pd_err);
+	atomic64_inc(&dev->stats.alloc_pd_err);
 	return err;
 }
 
-void efa_dealloc_pd(struct ib_pd *ibpd, struct ib_udata *udata)
+int efa_dealloc_pd(struct ib_pd *ibpd, struct ib_udata *udata)
 {
 	struct efa_dev *dev = to_edev(ibpd->device);
 	struct efa_pd *pd = to_epd(ibpd);
 
 	ibdev_dbg(&dev->ibdev, "Dealloc pd[%d]\n", pd->pdn);
 	efa_pd_dealloc(dev, pd->pdn);
+	return 0;
 }
 
 static int efa_destroy_qp_handle(struct efa_dev *dev, u32 qp_handle)
@@ -402,6 +442,9 @@ int efa_destroy_qp(struct ib_qp *ibqp, struct ib_udata *udata)
 	int err;
 
 	ibdev_dbg(&dev->ibdev, "Destroy qp[%u]\n", ibqp->qp_num);
+
+	efa_qp_user_mmap_entries_remove(qp);
+
 	err = efa_destroy_qp_handle(dev, qp->qp_handle);
 	if (err)
 		return err;
@@ -411,12 +454,10 @@ int efa_destroy_qp(struct ib_qp *ibqp, struct ib_udata *udata)
 			  "qp->cpu_addr[0x%p] freed: size[%lu], dma[%pad]\n",
 			  qp->rq_cpu_addr, qp->rq_size,
 			  &qp->rq_dma_addr);
-		dma_unmap_single(&dev->pdev->dev, qp->rq_dma_addr, qp->rq_size,
-				 DMA_TO_DEVICE);
+		efa_free_mapped(dev, qp->rq_cpu_addr, qp->rq_dma_addr,
+				qp->rq_size, DMA_TO_DEVICE);
 	}
 
-	efa_qp_user_mmap_entries_remove(qp);
-	kfree(qp);
 	return 0;
 }
 
@@ -575,17 +616,16 @@ static int efa_qp_validate_attr(struct efa_dev *dev,
 	return 0;
 }
 
-struct ib_qp *efa_create_qp(struct ib_pd *ibpd,
-			    struct ib_qp_init_attr *init_attr,
-			    struct ib_udata *udata)
+int efa_create_qp(struct ib_qp *ibqp, struct ib_qp_init_attr *init_attr,
+		  struct ib_udata *udata)
 {
 	struct efa_com_create_qp_params create_qp_params = {};
 	struct efa_com_create_qp_result create_qp_resp;
-	struct efa_dev *dev = to_edev(ibpd->device);
+	struct efa_dev *dev = to_edev(ibqp->device);
 	struct efa_ibv_create_qp_resp resp = {};
 	struct efa_ibv_create_qp cmd = {};
+	struct efa_qp *qp = to_eqp(ibqp);
 	struct efa_ucontext *ucontext;
-	struct efa_qp *qp;
 	int err;
 
 	ucontext = rdma_udata_to_drv_context(udata, struct efa_ucontext,
@@ -599,7 +639,7 @@ struct ib_qp *efa_create_qp(struct ib_pd *ibpd,
 	if (err)
 		goto err_out;
 
-	if (!field_avail(cmd, driver_qp_type, udata->inlen)) {
+	if (offsetofend(typeof(cmd), driver_qp_type) > udata->inlen) {
 		ibdev_dbg(&dev->ibdev,
 			  "Incompatible ABI params, no input udata\n");
 		err = -EINVAL;
@@ -630,14 +670,8 @@ struct ib_qp *efa_create_qp(struct ib_pd *ibpd,
 		goto err_out;
 	}
 
-	qp = kzalloc(sizeof(*qp), GFP_KERNEL);
-	if (!qp) {
-		err = -ENOMEM;
-		goto err_out;
-	}
-
 	create_qp_params.uarn = ucontext->uarn;
-	create_qp_params.pd = to_epd(ibpd)->pdn;
+	create_qp_params.pd = to_epd(ibqp->pd)->pdn;
 
 	if (init_attr->qp_type == IB_QPT_UD) {
 		create_qp_params.qp_type = EFA_ADMIN_QP_TYPE_UD;
@@ -648,7 +682,7 @@ struct ib_qp *efa_create_qp(struct ib_pd *ibpd,
 			  "Unsupported qp type %d driver qp type %d\n",
 			  init_attr->qp_type, cmd.driver_qp_type);
 		err = -EOPNOTSUPP;
-		goto err_free_qp;
+		goto err_out;
 	}
 
 	ibdev_dbg(&dev->ibdev, "Create QP: qp type %d driver qp type %#x\n",
@@ -666,7 +700,7 @@ struct ib_qp *efa_create_qp(struct ib_pd *ibpd,
 						    qp->rq_size, DMA_TO_DEVICE);
 		if (!qp->rq_cpu_addr) {
 			err = -ENOMEM;
-			goto err_free_qp;
+			goto err_out;
 		}
 
 		ibdev_dbg(&dev->ibdev,
@@ -693,7 +727,6 @@ struct ib_qp *efa_create_qp(struct ib_pd *ibpd,
 
 	qp->qp_handle = create_qp_resp.qp_handle;
 	qp->ibqp.qp_num = create_qp_resp.qp_num;
-	qp->ibqp.qp_type = init_attr->qp_type;
 	qp->max_send_wr = init_attr->cap.max_send_wr;
 	qp->max_recv_wr = init_attr->cap.max_recv_wr;
 	qp->max_send_sge = init_attr->cap.max_send_sge;
@@ -713,25 +746,128 @@ struct ib_qp *efa_create_qp(struct ib_pd *ibpd,
 
 	ibdev_dbg(&dev->ibdev, "Created qp[%d]\n", qp->ibqp.qp_num);
 
-	return &qp->ibqp;
+	return 0;
 
 err_remove_mmap_entries:
 	efa_qp_user_mmap_entries_remove(qp);
 err_destroy_qp:
 	efa_destroy_qp_handle(dev, create_qp_resp.qp_handle);
 err_free_mapped:
-	if (qp->rq_size) {
-		dma_unmap_single(&dev->pdev->dev, qp->rq_dma_addr, qp->rq_size,
-				 DMA_TO_DEVICE);
+	if (qp->rq_size)
+		efa_free_mapped(dev, qp->rq_cpu_addr, qp->rq_dma_addr,
+				qp->rq_size, DMA_TO_DEVICE);
+err_out:
+	atomic64_inc(&dev->stats.create_qp_err);
+	return err;
+}
 
-		if (!qp->rq_mmap_entry)
-			free_pages_exact(qp->rq_cpu_addr, qp->rq_size);
+static const struct {
+	int			valid;
+	enum ib_qp_attr_mask	req_param;
+	enum ib_qp_attr_mask	opt_param;
+} srd_qp_state_table[IB_QPS_ERR + 1][IB_QPS_ERR + 1] = {
+	[IB_QPS_RESET] = {
+		[IB_QPS_RESET] = { .valid = 1 },
+		[IB_QPS_INIT]  = {
+			.valid = 1,
+			.req_param = IB_QP_PKEY_INDEX |
+				     IB_QP_PORT |
+				     IB_QP_QKEY,
+		},
+	},
+	[IB_QPS_INIT] = {
+		[IB_QPS_RESET] = { .valid = 1 },
+		[IB_QPS_ERR]   = { .valid = 1 },
+		[IB_QPS_INIT]  = {
+			.valid = 1,
+			.opt_param = IB_QP_PKEY_INDEX |
+				     IB_QP_PORT |
+				     IB_QP_QKEY,
+		},
+		[IB_QPS_RTR]   = {
+			.valid = 1,
+			.opt_param = IB_QP_PKEY_INDEX |
+				     IB_QP_QKEY,
+		},
+	},
+	[IB_QPS_RTR] = {
+		[IB_QPS_RESET] = { .valid = 1 },
+		[IB_QPS_ERR]   = { .valid = 1 },
+		[IB_QPS_RTS]   = {
+			.valid = 1,
+			.req_param = IB_QP_SQ_PSN,
+			.opt_param = IB_QP_CUR_STATE |
+				     IB_QP_QKEY |
+				     IB_QP_RNR_RETRY,
+
+		}
+	},
+	[IB_QPS_RTS] = {
+		[IB_QPS_RESET] = { .valid = 1 },
+		[IB_QPS_ERR]   = { .valid = 1 },
+		[IB_QPS_RTS]   = {
+			.valid = 1,
+			.opt_param = IB_QP_CUR_STATE |
+				     IB_QP_QKEY,
+		},
+		[IB_QPS_SQD] = {
+			.valid = 1,
+			.opt_param = IB_QP_EN_SQD_ASYNC_NOTIFY,
+		},
+	},
+	[IB_QPS_SQD] = {
+		[IB_QPS_RESET] = { .valid = 1 },
+		[IB_QPS_ERR]   = { .valid = 1 },
+		[IB_QPS_RTS]   = {
+			.valid = 1,
+			.opt_param = IB_QP_CUR_STATE |
+				     IB_QP_QKEY,
+		},
+		[IB_QPS_SQD] = {
+			.valid = 1,
+			.opt_param = IB_QP_PKEY_INDEX |
+				     IB_QP_QKEY,
+		}
+	},
+	[IB_QPS_SQE] = {
+		[IB_QPS_RESET] = { .valid = 1 },
+		[IB_QPS_ERR]   = { .valid = 1 },
+		[IB_QPS_RTS]   = {
+			.valid = 1,
+			.opt_param = IB_QP_CUR_STATE |
+				     IB_QP_QKEY,
+		}
+	},
+	[IB_QPS_ERR] = {
+		[IB_QPS_RESET] = { .valid = 1 },
+		[IB_QPS_ERR]   = { .valid = 1 },
 	}
-err_free_qp:
-	kfree(qp);
-err_out:
-	atomic64_inc(&dev->stats.sw_stats.create_qp_err);
-	return ERR_PTR(err);
+};
+
+static bool efa_modify_srd_qp_is_ok(enum ib_qp_state cur_state,
+				    enum ib_qp_state next_state,
+				    enum ib_qp_attr_mask mask)
+{
+	enum ib_qp_attr_mask req_param, opt_param;
+
+	if (mask & IB_QP_CUR_STATE  &&
+	    cur_state != IB_QPS_RTR && cur_state != IB_QPS_RTS &&
+	    cur_state != IB_QPS_SQD && cur_state != IB_QPS_SQE)
+		return false;
+
+	if (!srd_qp_state_table[cur_state][next_state].valid)
+		return false;
+
+	req_param = srd_qp_state_table[cur_state][next_state].req_param;
+	opt_param = srd_qp_state_table[cur_state][next_state].opt_param;
+
+	if ((mask & req_param) != req_param)
+		return false;
+
+	if (mask & ~(req_param | opt_param | IB_QP_STATE))
+		return false;
+
+	return true;
 }
 
 static int efa_modify_qp_validate(struct efa_dev *dev, struct efa_qp *qp,
@@ -739,9 +875,12 @@ static int efa_modify_qp_validate(struct efa_dev *dev, struct efa_qp *qp,
 				  enum ib_qp_state cur_state,
 				  enum ib_qp_state new_state)
 {
+	int err;
+
 #define EFA_MODIFY_QP_SUPP_MASK \
 	(IB_QP_STATE | IB_QP_CUR_STATE | IB_QP_EN_SQD_ASYNC_NOTIFY | \
-	 IB_QP_PKEY_INDEX | IB_QP_PORT | IB_QP_QKEY | IB_QP_SQ_PSN)
+	 IB_QP_PKEY_INDEX | IB_QP_PORT | IB_QP_QKEY | IB_QP_SQ_PSN | \
+	 IB_QP_RNR_RETRY)
 
 	if (qp_attr_mask & ~EFA_MODIFY_QP_SUPP_MASK) {
 		ibdev_dbg(&dev->ibdev,
@@ -750,8 +889,14 @@ static int efa_modify_qp_validate(struct efa_dev *dev, struct efa_qp *qp,
 		return -EOPNOTSUPP;
 	}
 
-	if (!ib_modify_qp_is_ok(cur_state, new_state, IB_QPT_UD,
-				qp_attr_mask)) {
+	if (qp->ibqp.qp_type == IB_QPT_DRIVER)
+		err = !efa_modify_srd_qp_is_ok(cur_state, new_state,
+					       qp_attr_mask);
+	else
+		err = !ib_modify_qp_is_ok(cur_state, new_state, IB_QPT_UD,
+					  qp_attr_mask);
+
+	if (err) {
 		ibdev_dbg(&dev->ibdev, "Invalid modify QP parameters\n");
 		return -EINVAL;
 	}
@@ -779,6 +924,9 @@ int efa_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *qp_attr,
 	enum ib_qp_state new_state;
 	int err;
 
+	if (qp_attr_mask & ~IB_QP_ATTR_STANDARD_BITS)
+		return -EOPNOTSUPP;
+
 	if (udata->inlen &&
 	    !ib_is_udata_cleared(udata, 0, udata->inlen)) {
 		ibdev_dbg(&dev->ibdev,
@@ -798,28 +946,36 @@ int efa_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *qp_attr,
 	params.qp_handle = qp->qp_handle;
 
 	if (qp_attr_mask & IB_QP_STATE) {
-		params.modify_mask |= BIT(EFA_ADMIN_QP_STATE_BIT) |
-				      BIT(EFA_ADMIN_CUR_QP_STATE_BIT);
-		params.cur_qp_state = qp_attr->cur_qp_state;
-		params.qp_state = qp_attr->qp_state;
+		EFA_SET(&params.modify_mask, EFA_ADMIN_MODIFY_QP_CMD_QP_STATE,
+			1);
+		EFA_SET(&params.modify_mask,
+			EFA_ADMIN_MODIFY_QP_CMD_CUR_QP_STATE, 1);
+		params.cur_qp_state = cur_state;
+		params.qp_state = new_state;
 	}
 
 	if (qp_attr_mask & IB_QP_EN_SQD_ASYNC_NOTIFY) {
-		params.modify_mask |=
-			BIT(EFA_ADMIN_SQ_DRAINED_ASYNC_NOTIFY_BIT);
+		EFA_SET(&params.modify_mask,
+			EFA_ADMIN_MODIFY_QP_CMD_SQ_DRAINED_ASYNC_NOTIFY, 1);
 		params.sq_drained_async_notify = qp_attr->en_sqd_async_notify;
 	}
 
 	if (qp_attr_mask & IB_QP_QKEY) {
-		params.modify_mask |= BIT(EFA_ADMIN_QKEY_BIT);
+		EFA_SET(&params.modify_mask, EFA_ADMIN_MODIFY_QP_CMD_QKEY, 1);
 		params.qkey = qp_attr->qkey;
 	}
 
 	if (qp_attr_mask & IB_QP_SQ_PSN) {
-		params.modify_mask |= BIT(EFA_ADMIN_SQ_PSN_BIT);
+		EFA_SET(&params.modify_mask, EFA_ADMIN_MODIFY_QP_CMD_SQ_PSN, 1);
 		params.sq_psn = qp_attr->sq_psn;
 	}
 
+	if (qp_attr_mask & IB_QP_RNR_RETRY) {
+		EFA_SET(&params.modify_mask, EFA_ADMIN_MODIFY_QP_CMD_RNR_RETRY,
+			1);
+		params.rnr_retry = qp_attr->rnr_retry;
+	}
+
 	err = efa_com_modify_qp(&dev->edev, &params);
 	if (err)
 		return err;
@@ -836,7 +992,13 @@ static int efa_destroy_cq_idx(struct efa_dev *dev, int cq_idx)
 	return efa_com_destroy_cq(&dev->edev, &params);
 }
 
-void efa_destroy_cq(struct ib_cq *ibcq, struct ib_udata *udata)
+static void efa_cq_user_mmap_entries_remove(struct efa_cq *cq)
+{
+	rdma_user_mmap_entry_remove(cq->db_mmap_entry);
+	rdma_user_mmap_entry_remove(cq->mmap_entry);
+}
+
+int efa_destroy_cq(struct ib_cq *ibcq, struct ib_udata *udata)
 {
 	struct efa_dev *dev = to_edev(ibcq->device);
 	struct efa_cq *cq = to_ecq(ibcq);
@@ -845,14 +1007,25 @@ void efa_destroy_cq(struct ib_cq *ibcq, struct ib_udata *udata)
 		  "Destroy cq[%d] virt[0x%p] freed: size[%lu], dma[%pad]\n",
 		  cq->cq_idx, cq->cpu_addr, cq->size, &cq->dma_addr);
 
+	efa_cq_user_mmap_entries_remove(cq);
 	efa_destroy_cq_idx(dev, cq->cq_idx);
-	dma_unmap_single(&dev->pdev->dev, cq->dma_addr, cq->size,
-			 DMA_FROM_DEVICE);
-	rdma_user_mmap_entry_remove(cq->mmap_entry);
+	if (cq->eq) {
+		xa_erase(&dev->cqs_xa, cq->cq_idx);
+		synchronize_irq(cq->eq->irq.irqn);
+	}
+	efa_free_mapped(dev, cq->cpu_addr, cq->dma_addr, cq->size,
+			DMA_FROM_DEVICE);
+	return 0;
+}
+
+static struct efa_eq *efa_vec2eq(struct efa_dev *dev, int vec)
+{
+	return &dev->eqs[vec];
 }
 
 static int cq_mmap_entries_setup(struct efa_dev *dev, struct efa_cq *cq,
-				 struct efa_ibv_create_cq_resp *resp)
+				 struct efa_ibv_create_cq_resp *resp,
+				 bool db_valid)
 {
 	resp->q_mmap_size = cq->size;
 	cq->mmap_entry = efa_user_mmap_entry_insert(&cq->ucontext->ibucontext,
@@ -862,6 +1035,21 @@ static int cq_mmap_entries_setup(struct efa_dev *dev, struct efa_cq *cq,
 	if (!cq->mmap_entry)
 		return -ENOMEM;
 
+	if (db_valid) {
+		cq->db_mmap_entry =
+			efa_user_mmap_entry_insert(&cq->ucontext->ibucontext,
+						   dev->db_bar_addr + resp->db_off,
+						   PAGE_SIZE, EFA_MMAP_IO_NC,
+						   &resp->db_mmap_key);
+		if (!cq->db_mmap_entry) {
+			rdma_user_mmap_entry_remove(cq->mmap_entry);
+			return -ENOMEM;
+		}
+
+		resp->db_off &= ~PAGE_MASK;
+		resp->comp_mask |= EFA_CREATE_CQ_RESP_DB_OFF;
+	}
+
 	return 0;
 }
 
@@ -870,18 +1058,22 @@ int efa_create_cq(struct ib_cq *ibcq, const struct ib_cq_init_attr *attr,
 {
 	struct efa_ucontext *ucontext = rdma_udata_to_drv_context(
 		udata, struct efa_ucontext, ibucontext);
+	struct efa_com_create_cq_params params = {};
 	struct efa_ibv_create_cq_resp resp = {};
-	struct efa_com_create_cq_params params;
 	struct efa_com_create_cq_result result;
 	struct ib_device *ibdev = ibcq->device;
 	struct efa_dev *dev = to_edev(ibdev);
 	struct efa_ibv_create_cq cmd = {};
 	struct efa_cq *cq = to_ecq(ibcq);
 	int entries = attr->cqe;
+	bool set_src_addr;
 	int err;
 
 	ibdev_dbg(ibdev, "create_cq entries %d\n", entries);
 
+	if (attr->flags)
+		return -EOPNOTSUPP;
+
 	if (entries < 1 || entries > dev->dev_attr.max_cq_depth) {
 		ibdev_dbg(ibdev,
 			  "cq: requested entries[%u] non-positive or greater than max[%u]\n",
@@ -890,7 +1082,7 @@ int efa_create_cq(struct ib_cq *ibcq, const struct ib_cq_init_attr *attr,
 		goto err_out;
 	}
 
-	if (!field_avail(cmd, num_sub_cqs, udata->inlen)) {
+	if (offsetofend(typeof(cmd), num_sub_cqs) > udata->inlen) {
 		ibdev_dbg(ibdev,
 			  "Incompatible ABI params, no input udata\n");
 		err = -EINVAL;
@@ -913,14 +1105,17 @@ int efa_create_cq(struct ib_cq *ibcq, const struct ib_cq_init_attr *attr,
 		goto err_out;
 	}
 
-	if (cmd.comp_mask || !is_reserved_cleared(cmd.reserved_50)) {
+	if (cmd.comp_mask || !is_reserved_cleared(cmd.reserved_58)) {
 		ibdev_dbg(ibdev,
 			  "Incompatible ABI params, unknown fields in udata\n");
 		err = -EINVAL;
 		goto err_out;
 	}
 
-	if (!cmd.cq_entry_size) {
+	set_src_addr = !!(cmd.flags & EFA_CREATE_CQ_WITH_SGID);
+	if ((cmd.cq_entry_size != sizeof(struct efa_io_rx_cdesc_ex)) &&
+	    (set_src_addr ||
+	     cmd.cq_entry_size != sizeof(struct efa_io_rx_cdesc))) {
 		ibdev_dbg(ibdev,
 			  "Invalid entry size [%u]\n", cmd.cq_entry_size);
 		err = -EINVAL;
@@ -949,29 +1144,46 @@ int efa_create_cq(struct ib_cq *ibcq, const struct ib_cq_init_attr *attr,
 	params.dma_addr = cq->dma_addr;
 	params.entry_size_in_bytes = cmd.cq_entry_size;
 	params.num_sub_cqs = cmd.num_sub_cqs;
+	params.set_src_addr = set_src_addr;
+	if (cmd.flags & EFA_CREATE_CQ_WITH_COMPLETION_CHANNEL) {
+		cq->eq = efa_vec2eq(dev, attr->comp_vector);
+		params.eqn = cq->eq->eeq.eqn;
+		params.interrupt_mode_enabled = true;
+	}
+
 	err = efa_com_create_cq(&dev->edev, &params, &result);
 	if (err)
 		goto err_free_mapped;
 
+	resp.db_off = result.db_off;
 	resp.cq_idx = result.cq_idx;
 	cq->cq_idx = result.cq_idx;
 	cq->ibcq.cqe = result.actual_depth;
 	WARN_ON_ONCE(entries != result.actual_depth);
 
-	err = cq_mmap_entries_setup(dev, cq, &resp);
+	err = cq_mmap_entries_setup(dev, cq, &resp, result.db_valid);
 	if (err) {
 		ibdev_dbg(ibdev, "Could not setup cq[%u] mmap entries\n",
 			  cq->cq_idx);
 		goto err_destroy_cq;
 	}
 
+	if (cq->eq) {
+		err = xa_err(xa_store(&dev->cqs_xa, cq->cq_idx, cq, GFP_KERNEL));
+		if (err) {
+			ibdev_dbg(ibdev, "Failed to store cq[%u] in xarray\n",
+				  cq->cq_idx);
+			goto err_remove_mmap;
+		}
+	}
+
 	if (udata->outlen) {
 		err = ib_copy_to_udata(udata, &resp,
 				       min(sizeof(resp), udata->outlen));
 		if (err) {
 			ibdev_dbg(ibdev,
 				  "Failed to copy udata for create_cq\n");
-			goto err_remove_mmap;
+			goto err_xa_erase;
 		}
 	}
 
@@ -980,18 +1192,19 @@ int efa_create_cq(struct ib_cq *ibcq, const struct ib_cq_init_attr *attr,
 
 	return 0;
 
+err_xa_erase:
+	if (cq->eq)
+		xa_erase(&dev->cqs_xa, cq->cq_idx);
 err_remove_mmap:
-	rdma_user_mmap_entry_remove(cq->mmap_entry);
+	efa_cq_user_mmap_entries_remove(cq);
 err_destroy_cq:
 	efa_destroy_cq_idx(dev, cq->cq_idx);
 err_free_mapped:
-	dma_unmap_single(&dev->pdev->dev, cq->dma_addr, cq->size,
-			 DMA_FROM_DEVICE);
-	if (!cq->mmap_entry)
-		free_pages_exact(cq->cpu_addr, cq->size);
+	efa_free_mapped(dev, cq->cpu_addr, cq->dma_addr, cq->size,
+			DMA_FROM_DEVICE);
 
 err_out:
-	atomic64_inc(&dev->stats.sw_stats.create_cq_err);
+	atomic64_inc(&dev->stats.create_cq_err);
 	return err;
 }
 
@@ -1008,8 +1221,7 @@ static int umem_to_page_list(struct efa_dev *dev,
 	ibdev_dbg(&dev->ibdev, "hp_cnt[%u], pages_in_hp[%u]\n",
 		  hp_cnt, pages_in_hp);
 
-	rdma_for_each_block(umem->sg_head.sgl, &biter, umem->nmap,
-			    BIT(hp_shift))
+	rdma_umem_for_each_dma_block(umem, &biter, BIT(hp_shift))
 		page_list[hp_idx++] = rdma_block_iter_dma_address(&biter);
 
 	return 0;
@@ -1021,7 +1233,7 @@ static struct scatterlist *efa_vmalloc_buf_to_sg(u64 *buf, int page_cnt)
 	struct page *pg;
 	int i;
 
-	sglist = kcalloc(page_cnt, sizeof(*sglist), GFP_KERNEL);
+	sglist = kmalloc_array(page_cnt, sizeof(*sglist), GFP_KERNEL);
 	if (!sglist)
 		return NULL;
 	sg_init_table(sglist, page_cnt);
@@ -1341,54 +1553,49 @@ static int efa_create_pbl(struct efa_dev *dev,
 	return 0;
 }
 
-struct ib_mr *efa_reg_mr(struct ib_pd *ibpd, u64 start, u64 length,
-			 u64 virt_addr, int access_flags,
-			 struct ib_udata *udata)
+static struct efa_mr *efa_alloc_mr(struct ib_pd *ibpd, int access_flags,
+				   struct ib_udata *udata)
 {
 	struct efa_dev *dev = to_edev(ibpd->device);
-	struct efa_com_reg_mr_params params = {};
-	struct efa_com_reg_mr_result result = {};
-	struct pbl_context pbl;
 	int supp_access_flags;
-	unsigned int pg_sz;
 	struct efa_mr *mr;
-	int inline_size;
-	int err;
 
 	if (udata && udata->inlen &&
 	    !ib_is_udata_cleared(udata, 0, sizeof(udata->inlen))) {
 		ibdev_dbg(&dev->ibdev,
 			  "Incompatible ABI params, udata not cleared\n");
-		err = -EINVAL;
-		goto err_out;
+		return ERR_PTR(-EINVAL);
 	}
 
 	supp_access_flags =
 		IB_ACCESS_LOCAL_WRITE |
-		(is_rdma_read_cap(dev) ? IB_ACCESS_REMOTE_READ : 0);
+		(EFA_DEV_CAP(dev, RDMA_READ) ? IB_ACCESS_REMOTE_READ : 0);
 
 	access_flags &= ~IB_ACCESS_OPTIONAL;
 	if (access_flags & ~supp_access_flags) {
 		ibdev_dbg(&dev->ibdev,
 			  "Unsupported access flags[%#x], supported[%#x]\n",
 			  access_flags, supp_access_flags);
-		err = -EOPNOTSUPP;
-		goto err_out;
+		return ERR_PTR(-EOPNOTSUPP);
 	}
 
 	mr = kzalloc(sizeof(*mr), GFP_KERNEL);
-	if (!mr) {
-		err = -ENOMEM;
-		goto err_out;
-	}
+	if (!mr)
+		return ERR_PTR(-ENOMEM);
 
-	mr->umem = ib_umem_get(ibpd->device, start, length, access_flags);
-	if (IS_ERR(mr->umem)) {
-		err = PTR_ERR(mr->umem);
-		ibdev_dbg(&dev->ibdev,
-			  "Failed to pin and map user space memory[%d]\n", err);
-		goto err_free;
-	}
+	return mr;
+}
+
+static int efa_register_mr(struct ib_pd *ibpd, struct efa_mr *mr, u64 start,
+			   u64 length, u64 virt_addr, int access_flags)
+{
+	struct efa_dev *dev = to_edev(ibpd->device);
+	struct efa_com_reg_mr_params params = {};
+	struct efa_com_reg_mr_result result = {};
+	struct pbl_context pbl;
+	unsigned int pg_sz;
+	int inline_size;
+	int err;
 
 	params.pd = to_epd(ibpd)->pdn;
 	params.iova = virt_addr;
@@ -1399,15 +1606,13 @@ struct ib_mr *efa_reg_mr(struct ib_pd *ibpd, u64 start, u64 length,
 				       dev->dev_attr.page_size_cap,
 				       virt_addr);
 	if (!pg_sz) {
-		err = -EOPNOTSUPP;
 		ibdev_dbg(&dev->ibdev, "Failed to find a suitable page size in page_size_cap %#llx\n",
 			  dev->dev_attr.page_size_cap);
-		goto err_unmap;
+		return -EOPNOTSUPP;
 	}
 
-	params.page_shift = __ffs(pg_sz);
-	params.page_num = DIV_ROUND_UP(length + (start & (pg_sz - 1)),
-				       pg_sz);
+	params.page_shift = order_base_2(pg_sz);
+	params.page_num = ib_umem_num_dma_blocks(mr->umem, pg_sz);
 
 	ibdev_dbg(&dev->ibdev,
 		  "start %#llx length %#llx params.page_shift %u params.page_num %u\n",
@@ -1417,21 +1622,21 @@ struct ib_mr *efa_reg_mr(struct ib_pd *ibpd, u64 start, u64 length,
 	if (params.page_num <= inline_size) {
 		err = efa_create_inline_pbl(dev, mr, &params);
 		if (err)
-			goto err_unmap;
+			return err;
 
 		err = efa_com_register_mr(&dev->edev, &params, &result);
 		if (err)
-			goto err_unmap;
+			return err;
 	} else {
 		err = efa_create_pbl(dev, &pbl, mr, &params);
 		if (err)
-			goto err_unmap;
+			return err;
 
 		err = efa_com_register_mr(&dev->edev, &params, &result);
 		pbl_destroy(dev, &pbl);
 
 		if (err)
-			goto err_unmap;
+			return err;
 	}
 
 	mr->ibmr.lkey = result.l_key;
@@ -1439,14 +1644,83 @@ struct ib_mr *efa_reg_mr(struct ib_pd *ibpd, u64 start, u64 length,
 	mr->ibmr.length = length;
 	ibdev_dbg(&dev->ibdev, "Registered mr[%d]\n", mr->ibmr.lkey);
 
+	return 0;
+}
+
+struct ib_mr *efa_reg_user_mr_dmabuf(struct ib_pd *ibpd, u64 start,
+				     u64 length, u64 virt_addr,
+				     int fd, int access_flags,
+				     struct ib_udata *udata)
+{
+	struct efa_dev *dev = to_edev(ibpd->device);
+	struct ib_umem_dmabuf *umem_dmabuf;
+	struct efa_mr *mr;
+	int err;
+
+	mr = efa_alloc_mr(ibpd, access_flags, udata);
+	if (IS_ERR(mr)) {
+		err = PTR_ERR(mr);
+		goto err_out;
+	}
+
+	umem_dmabuf = ib_umem_dmabuf_get_pinned(ibpd->device, start, length, fd,
+						access_flags);
+	if (IS_ERR(umem_dmabuf)) {
+		err = PTR_ERR(umem_dmabuf);
+		ibdev_dbg(&dev->ibdev, "Failed to get dmabuf umem[%d]\n", err);
+		goto err_free;
+	}
+
+	mr->umem = &umem_dmabuf->umem;
+	err = efa_register_mr(ibpd, mr, start, length, virt_addr, access_flags);
+	if (err)
+		goto err_release;
+
 	return &mr->ibmr;
 
-err_unmap:
+err_release:
 	ib_umem_release(mr->umem);
 err_free:
 	kfree(mr);
 err_out:
-	atomic64_inc(&dev->stats.sw_stats.reg_mr_err);
+	atomic64_inc(&dev->stats.reg_mr_err);
+	return ERR_PTR(err);
+}
+
+struct ib_mr *efa_reg_mr(struct ib_pd *ibpd, u64 start, u64 length,
+			 u64 virt_addr, int access_flags,
+			 struct ib_udata *udata)
+{
+	struct efa_dev *dev = to_edev(ibpd->device);
+	struct efa_mr *mr;
+	int err;
+
+	mr = efa_alloc_mr(ibpd, access_flags, udata);
+	if (IS_ERR(mr)) {
+		err = PTR_ERR(mr);
+		goto err_out;
+	}
+
+	mr->umem = ib_umem_get(ibpd->device, start, length, access_flags);
+	if (IS_ERR(mr->umem)) {
+		err = PTR_ERR(mr->umem);
+		ibdev_dbg(&dev->ibdev,
+			  "Failed to pin and map user space memory[%d]\n", err);
+		goto err_free;
+	}
+
+	err = efa_register_mr(ibpd, mr, start, length, virt_addr, access_flags);
+	if (err)
+		goto err_release;
+
+	return &mr->ibmr;
+
+err_release:
+	ib_umem_release(mr->umem);
+err_free:
+	kfree(mr);
+err_out:
+	atomic64_inc(&dev->stats.reg_mr_err);
 	return ERR_PTR(err);
 }
 
@@ -1470,7 +1744,7 @@ int efa_dereg_mr(struct ib_mr *ibmr, struct ib_udata *udata)
 	return 0;
 }
 
-int efa_get_port_immutable(struct ib_device *ibdev, u8 port_num,
+int efa_get_port_immutable(struct ib_device *ibdev, u32 port_num,
 			   struct ib_port_immutable *immutable)
 {
 	struct ib_port_attr attr;
@@ -1497,11 +1771,39 @@ static int efa_dealloc_uar(struct efa_dev *dev, u16 uarn)
 	return efa_com_dealloc_uar(&dev->edev, &params);
 }
 
+#define EFA_CHECK_USER_COMP(_dev, _comp_mask, _attr, _mask, _attr_str) \
+	(_attr_str = (!(_dev)->dev_attr._attr || ((_comp_mask) & (_mask))) ? \
+		     NULL : #_attr)
+
+static int efa_user_comp_handshake(const struct ib_ucontext *ibucontext,
+				   const struct efa_ibv_alloc_ucontext_cmd *cmd)
+{
+	struct efa_dev *dev = to_edev(ibucontext->device);
+	char *attr_str;
+
+	if (EFA_CHECK_USER_COMP(dev, cmd->comp_mask, max_tx_batch,
+				EFA_ALLOC_UCONTEXT_CMD_COMP_TX_BATCH, attr_str))
+		goto err;
+
+	if (EFA_CHECK_USER_COMP(dev, cmd->comp_mask, min_sq_depth,
+				EFA_ALLOC_UCONTEXT_CMD_COMP_MIN_SQ_WR,
+				attr_str))
+		goto err;
+
+	return 0;
+
+err:
+	ibdev_dbg(&dev->ibdev, "Userspace handshake failed for %s attribute\n",
+		  attr_str);
+	return -EOPNOTSUPP;
+}
+
 int efa_alloc_ucontext(struct ib_ucontext *ibucontext, struct ib_udata *udata)
 {
 	struct efa_ucontext *ucontext = to_eucontext(ibucontext);
 	struct efa_dev *dev = to_edev(ibucontext->device);
 	struct efa_ibv_alloc_ucontext_resp resp = {};
+	struct efa_ibv_alloc_ucontext_cmd cmd = {};
 	struct efa_com_alloc_uar_result result;
 	int err;
 
@@ -1510,6 +1812,18 @@ int efa_alloc_ucontext(struct ib_ucontext *ibucontext, struct ib_udata *udata)
 	 * we will ack input fields in our response.
 	 */
 
+	err = ib_copy_from_udata(&cmd, udata,
+				 min(sizeof(cmd), udata->inlen));
+	if (err) {
+		ibdev_dbg(&dev->ibdev,
+			  "Cannot copy udata for alloc_ucontext\n");
+		goto err_out;
+	}
+
+	err = efa_user_comp_handshake(ibucontext, &cmd);
+	if (err)
+		goto err_out;
+
 	err = efa_com_alloc_uar(&dev->edev, &result);
 	if (err)
 		goto err_out;
@@ -1521,20 +1835,20 @@ int efa_alloc_ucontext(struct ib_ucontext *ibucontext, struct ib_udata *udata)
 	resp.sub_cqs_per_cq = dev->dev_attr.sub_cqs_per_cq;
 	resp.inline_buf_size = dev->dev_attr.inline_buf_size;
 	resp.max_llq_size = dev->dev_attr.max_llq_size;
+	resp.max_tx_batch = dev->dev_attr.max_tx_batch;
+	resp.min_sq_wr = dev->dev_attr.min_sq_depth;
 
-	if (udata && udata->outlen) {
-		err = ib_copy_to_udata(udata, &resp,
-				       min(sizeof(resp), udata->outlen));
-		if (err)
-			goto err_dealloc_uar;
-	}
+	err = ib_copy_to_udata(udata, &resp,
+			       min(sizeof(resp), udata->outlen));
+	if (err)
+		goto err_dealloc_uar;
 
 	return 0;
 
 err_dealloc_uar:
 	efa_dealloc_uar(dev, result.uarn);
 err_out:
-	atomic64_inc(&dev->stats.sw_stats.alloc_ucontext_err);
+	atomic64_inc(&dev->stats.alloc_ucontext_err);
 	return err;
 }
 
@@ -1550,10 +1864,6 @@ void efa_mmap_free(struct rdma_user_mmap_entry *rdma_entry)
 {
 	struct efa_user_mmap_entry *entry = to_emmap(rdma_entry);
 
-	/* DMA mapping is already gone, now free the pages */
-	if (entry->mmap_flag == EFA_MMAP_DMA_PAGE)
-		free_pages_exact(phys_to_virt(entry->address),
-				 entry->rdma_entry.npages * PAGE_SIZE);
 	kfree(entry);
 }
 
@@ -1571,6 +1881,7 @@ static int __efa_mmap(struct efa_dev *dev, struct efa_ucontext *ucontext,
 		ibdev_dbg(&dev->ibdev,
 			  "pgoff[%#lx] does not have valid entry\n",
 			  vma->vm_pgoff);
+		atomic64_inc(&dev->stats.mmap_err);
 		return -EINVAL;
 	}
 	entry = to_emmap(rdma_entry);
@@ -1606,12 +1917,14 @@ static int __efa_mmap(struct efa_dev *dev, struct efa_ucontext *ucontext,
 		err = -EINVAL;
 	}
 
-	if (err)
+	if (err) {
 		ibdev_dbg(
 			&dev->ibdev,
 			"Couldn't mmap address[%#llx] length[%#zx] mmap_flag[%d] err[%d]\n",
 			entry->address, rdma_entry->npages * PAGE_SIZE,
 			entry->mmap_flag, err);
+		atomic64_inc(&dev->stats.mmap_err);
+	}
 
 	rdma_user_mmap_entry_put(rdma_entry);
 	return err;
@@ -1642,10 +1955,10 @@ static int efa_ah_destroy(struct efa_dev *dev, struct efa_ah *ah)
 }
 
 int efa_create_ah(struct ib_ah *ibah,
-		  struct rdma_ah_attr *ah_attr,
-		  u32 flags,
+		  struct rdma_ah_init_attr *init_attr,
 		  struct ib_udata *udata)
 {
+	struct rdma_ah_attr *ah_attr = init_attr->ah_attr;
 	struct efa_dev *dev = to_edev(ibah->device);
 	struct efa_com_create_ah_params params = {};
 	struct efa_ibv_create_ah_resp resp = {};
@@ -1653,7 +1966,7 @@ int efa_create_ah(struct ib_ah *ibah,
 	struct efa_ah *ah = to_eah(ibah);
 	int err;
 
-	if (!(flags & RDMA_CREATE_AH_SLEEPABLE)) {
+	if (!(init_attr->flags & RDMA_CREATE_AH_SLEEPABLE)) {
 		ibdev_dbg(&dev->ibdev,
 			  "Create address handle is not supported in atomic context\n");
 		err = -EOPNOTSUPP;
@@ -1695,11 +2008,11 @@ int efa_create_ah(struct ib_ah *ibah,
 err_destroy_ah:
 	efa_ah_destroy(dev, ah);
 err_out:
-	atomic64_inc(&dev->stats.sw_stats.create_ah_err);
+	atomic64_inc(&dev->stats.create_ah_err);
 	return err;
 }
 
-void efa_destroy_ah(struct ib_ah *ibah, u32 flags)
+int efa_destroy_ah(struct ib_ah *ibah, u32 flags)
 {
 	struct efa_dev *dev = to_edev(ibah->pd->device);
 	struct efa_ah *ah = to_eah(ibah);
@@ -1709,32 +2022,64 @@ void efa_destroy_ah(struct ib_ah *ibah, u32 flags)
 	if (!(flags & RDMA_DESTROY_AH_SLEEPABLE)) {
 		ibdev_dbg(&dev->ibdev,
 			  "Destroy address handle is not supported in atomic context\n");
-		return;
+		return -EOPNOTSUPP;
 	}
 
 	efa_ah_destroy(dev, ah);
+	return 0;
 }
 
-struct rdma_hw_stats *efa_alloc_hw_stats(struct ib_device *ibdev, u8 port_num)
+struct rdma_hw_stats *efa_alloc_hw_port_stats(struct ib_device *ibdev,
+					      u32 port_num)
 {
-	return rdma_alloc_hw_stats_struct(efa_stats_names,
-					  ARRAY_SIZE(efa_stats_names),
+	return rdma_alloc_hw_stats_struct(efa_port_stats_descs,
+					  ARRAY_SIZE(efa_port_stats_descs),
 					  RDMA_HW_STATS_DEFAULT_LIFESPAN);
 }
 
-int efa_get_hw_stats(struct ib_device *ibdev, struct rdma_hw_stats *stats,
-		     u8 port_num, int index)
+struct rdma_hw_stats *efa_alloc_hw_device_stats(struct ib_device *ibdev)
+{
+	return rdma_alloc_hw_stats_struct(efa_device_stats_descs,
+					  ARRAY_SIZE(efa_device_stats_descs),
+					  RDMA_HW_STATS_DEFAULT_LIFESPAN);
+}
+
+static int efa_fill_device_stats(struct efa_dev *dev,
+				 struct rdma_hw_stats *stats)
+{
+	struct efa_com_stats_admin *as = &dev->edev.aq.stats;
+	struct efa_stats *s = &dev->stats;
+
+	stats->value[EFA_SUBMITTED_CMDS] = atomic64_read(&as->submitted_cmd);
+	stats->value[EFA_COMPLETED_CMDS] = atomic64_read(&as->completed_cmd);
+	stats->value[EFA_CMDS_ERR] = atomic64_read(&as->cmd_err);
+	stats->value[EFA_NO_COMPLETION_CMDS] = atomic64_read(&as->no_completion);
+
+	stats->value[EFA_KEEP_ALIVE_RCVD] = atomic64_read(&s->keep_alive_rcvd);
+	stats->value[EFA_ALLOC_PD_ERR] = atomic64_read(&s->alloc_pd_err);
+	stats->value[EFA_CREATE_QP_ERR] = atomic64_read(&s->create_qp_err);
+	stats->value[EFA_CREATE_CQ_ERR] = atomic64_read(&s->create_cq_err);
+	stats->value[EFA_REG_MR_ERR] = atomic64_read(&s->reg_mr_err);
+	stats->value[EFA_ALLOC_UCONTEXT_ERR] =
+		atomic64_read(&s->alloc_ucontext_err);
+	stats->value[EFA_CREATE_AH_ERR] = atomic64_read(&s->create_ah_err);
+	stats->value[EFA_MMAP_ERR] = atomic64_read(&s->mmap_err);
+
+	return ARRAY_SIZE(efa_device_stats_descs);
+}
+
+static int efa_fill_port_stats(struct efa_dev *dev, struct rdma_hw_stats *stats,
+			       u32 port_num)
 {
 	struct efa_com_get_stats_params params = {};
 	union efa_com_get_stats_result result;
-	struct efa_dev *dev = to_edev(ibdev);
+	struct efa_com_rdma_read_stats *rrs;
+	struct efa_com_messages_stats *ms;
 	struct efa_com_basic_stats *bs;
-	struct efa_com_stats_admin *as;
-	struct efa_stats *s;
 	int err;
 
-	params.type = EFA_ADMIN_GET_STATS_TYPE_BASIC;
 	params.scope = EFA_ADMIN_GET_STATS_SCOPE_ALL;
+	params.type = EFA_ADMIN_GET_STATS_TYPE_BASIC;
 
 	err = efa_com_get_stats(&dev->edev, &params, &result);
 	if (err)
@@ -1747,24 +2092,42 @@ int efa_get_hw_stats(struct ib_device *ibdev, struct rdma_hw_stats *stats,
 	stats->value[EFA_RX_PKTS] = bs->rx_pkts;
 	stats->value[EFA_RX_DROPS] = bs->rx_drops;
 
-	as = &dev->edev.aq.stats;
-	stats->value[EFA_SUBMITTED_CMDS] = atomic64_read(&as->submitted_cmd);
-	stats->value[EFA_COMPLETED_CMDS] = atomic64_read(&as->completed_cmd);
-	stats->value[EFA_NO_COMPLETION_CMDS] = atomic64_read(&as->no_completion);
+	params.type = EFA_ADMIN_GET_STATS_TYPE_MESSAGES;
+	err = efa_com_get_stats(&dev->edev, &params, &result);
+	if (err)
+		return err;
 
-	s = &dev->stats;
-	stats->value[EFA_KEEP_ALIVE_RCVD] = atomic64_read(&s->keep_alive_rcvd);
-	stats->value[EFA_ALLOC_PD_ERR] = atomic64_read(&s->sw_stats.alloc_pd_err);
-	stats->value[EFA_CREATE_QP_ERR] = atomic64_read(&s->sw_stats.create_qp_err);
-	stats->value[EFA_REG_MR_ERR] = atomic64_read(&s->sw_stats.reg_mr_err);
-	stats->value[EFA_ALLOC_UCONTEXT_ERR] = atomic64_read(&s->sw_stats.alloc_ucontext_err);
-	stats->value[EFA_CREATE_AH_ERR] = atomic64_read(&s->sw_stats.create_ah_err);
+	ms = &result.messages_stats;
+	stats->value[EFA_SEND_BYTES] = ms->send_bytes;
+	stats->value[EFA_SEND_WRS] = ms->send_wrs;
+	stats->value[EFA_RECV_BYTES] = ms->recv_bytes;
+	stats->value[EFA_RECV_WRS] = ms->recv_wrs;
 
-	return ARRAY_SIZE(efa_stats_names);
+	params.type = EFA_ADMIN_GET_STATS_TYPE_RDMA_READ;
+	err = efa_com_get_stats(&dev->edev, &params, &result);
+	if (err)
+		return err;
+
+	rrs = &result.rdma_read_stats;
+	stats->value[EFA_RDMA_READ_WRS] = rrs->read_wrs;
+	stats->value[EFA_RDMA_READ_BYTES] = rrs->read_bytes;
+	stats->value[EFA_RDMA_READ_WR_ERR] = rrs->read_wr_err;
+	stats->value[EFA_RDMA_READ_RESP_BYTES] = rrs->read_resp_bytes;
+
+	return ARRAY_SIZE(efa_port_stats_descs);
+}
+
+int efa_get_hw_stats(struct ib_device *ibdev, struct rdma_hw_stats *stats,
+		     u32 port_num, int index)
+{
+	if (port_num)
+		return efa_fill_port_stats(to_edev(ibdev), stats, port_num);
+	else
+		return efa_fill_device_stats(to_edev(ibdev), stats);
 }
 
 enum rdma_link_layer efa_port_link_layer(struct ib_device *ibdev,
-					 u8 port_num)
+					 u32 port_num)
 {
 	return IB_LINK_LAYER_UNSPECIFIED;
 }