Merge branch 'for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/nab/target-pending

Pull SCSI target updates from Nicholas Bellinger:
 "The highlights this round include:

   - Add target_alloc_session() w/ callback helper for doing se_session
     allocation + tag + se_node_acl lookup.  (HCH + nab)

   - Tree-wide fabric driver conversion to use target_alloc_session()

   - Convert sbp-target to use percpu_ida tag pre-allocation, and
     TARGET_SCF_ACK_KREF I/O krefs (Chris Boot + nab)

   - Convert usb-gadget to use percpu_ida tag pre-allocation, and
     TARGET_SCF_ACK_KREF I/O krefs (Andrzej Pietrasiewicz + nab)

   - Convert xen-scsiback to use percpu_ida tag pre-allocation, and
     TARGET_SCF_ACK_KREF I/O krefs (Juergen Gross + nab)

   - Convert tcm_fc to use TARGET_SCF_ACK_KREF I/O + TMR krefs

   - Convert ib_srpt to use percpu_ida tag pre-allocation

   - Add DebugFS node for qla2xxx target sess list (Quinn)

   - Rework iser-target connection termination (Jenny + Sagi)

   - Convert iser-target to new CQ API (HCH)

   - Add pass-through WRITE_SAME support for IBLOCK (Mike Christie)

   - Introduce data_bitmap for asynchronous access of data area (Sheng
     Yang + Andy)

   - Fix target_release_cmd_kref shutdown comp leak (Himanshu Madhani)

  Also, there is a separate PULL request coming for cxgb4 NIC driver
  prerequisites for supporting hw iscsi segmentation offload (ISO), that
  will be the base for a number of v4.7 developments involving
  iscsi-target hw offloads"

* 'for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/nab/target-pending: (36 commits)
  target: Fix target_release_cmd_kref shutdown comp leak
  target: Avoid DataIN transfers for non-GOOD SAM status
  target/user: Report capability of handling out-of-order completions to userspace
  target/user: Fix size_t format-spec build warning
  target/user: Don't free expired command when time out
  target/user: Introduce data_bitmap, replace data_length/data_head/data_tail
  target/user: Free data ring in unified function
  target/user: Use iovec[] to describe continuous area
  target: Remove enum transport_lunflags_table
  target/iblock: pass WRITE_SAME to device if possible
  iser-target: Kill the ->isert_cmd back pointer in struct iser_tx_desc
  iser-target: Kill struct isert_rdma_wr
  iser-target: Convert to new CQ API
  iser-target: Split and properly type the login buffer
  iser-target: Remove ISER_RECV_DATA_SEG_LEN
  iser-target: Remove impossible condition from isert_wait_conn
  iser-target: Remove redundant wait in release_conn
  iser-target: Rework connection termination
  iser-target: Separate flows for np listeners and connections cma events
  iser-target: Add new state ISER_CONN_BOUND to isert_conn
  ...

29 files changed, 1240 insertions, 1195 deletions

diff --git a/Documentation/target/tcmu-design.txt b/Documentation/target/tcmu-design.txt
index bef81e42788f..4cebc1ebf99a 100644
--- a/Documentation/target/tcmu-design.txt
+++ b/Documentation/target/tcmu-design.txt
@@ -117,7 +117,9 @@ userspace (respectively) to put commands on the ring, and indicate
 when the commands are completed.
 
 version - 1 (userspace should abort if otherwise)
-flags - none yet defined.
+flags:
+- TCMU_MAILBOX_FLAG_CAP_OOOC: indicates out-of-order completion is
+  supported.  See "The Command Ring" for details.
 cmdr_off - The offset of the start of the command ring from the start
 of the memory region, to account for the mailbox size.
 cmdr_size - The size of the command ring. This does *not* need to be a
@@ -162,6 +164,13 @@ rsp.sense_buffer if necessary. Userspace then increments
 mailbox.cmd_tail by entry.hdr.length (mod cmdr_size) and signals the
 kernel via the UIO method, a 4-byte write to the file descriptor.
 
+If TCMU_MAILBOX_FLAG_CAP_OOOC is set for mailbox->flags, kernel is
+capable of handling out-of-order completions. In this case, userspace can
+handle command in different order other than original. Since kernel would
+still process the commands in the same order it appeared in the command
+ring, userspace need to update the cmd->id when completing the
+command(a.k.a steal the original command's entry).
+
 When the opcode is PAD, userspace only updates cmd_tail as above --
 it's a no-op. (The kernel inserts PAD entries to ensure each CMD entry
 is contiguous within the command ring.)
diff --git a/drivers/infiniband/ulp/isert/ib_isert.c b/drivers/infiniband/ulp/isert/ib_isert.c
index f121e6129339..60b30d338a81 100644
--- a/drivers/infiniband/ulp/isert/ib_isert.c
+++ b/drivers/infiniband/ulp/isert/ib_isert.c
@@ -49,22 +49,25 @@ static struct workqueue_struct *isert_release_wq;
 static void
 isert_unmap_cmd(struct isert_cmd *isert_cmd, struct isert_conn *isert_conn);
 static int
-isert_map_rdma(struct iscsi_conn *conn, struct iscsi_cmd *cmd,
-	       struct isert_rdma_wr *wr);
+isert_map_rdma(struct isert_cmd *isert_cmd, struct iscsi_conn *conn);
 static void
 isert_unreg_rdma(struct isert_cmd *isert_cmd, struct isert_conn *isert_conn);
 static int
-isert_reg_rdma(struct iscsi_conn *conn, struct iscsi_cmd *cmd,
-	       struct isert_rdma_wr *wr);
+isert_reg_rdma(struct isert_cmd *isert_cmd, struct iscsi_conn *conn);
 static int
 isert_put_response(struct iscsi_conn *conn, struct iscsi_cmd *cmd);
 static int
-isert_rdma_post_recvl(struct isert_conn *isert_conn);
+isert_login_post_recv(struct isert_conn *isert_conn);
 static int
 isert_rdma_accept(struct isert_conn *isert_conn);
 struct rdma_cm_id *isert_setup_id(struct isert_np *isert_np);
 
 static void isert_release_work(struct work_struct *work);
+static void isert_wait4flush(struct isert_conn *isert_conn);
+static void isert_recv_done(struct ib_cq *cq, struct ib_wc *wc);
+static void isert_send_done(struct ib_cq *cq, struct ib_wc *wc);
+static void isert_login_recv_done(struct ib_cq *cq, struct ib_wc *wc);
+static void isert_login_send_done(struct ib_cq *cq, struct ib_wc *wc);
 
 static inline bool
 isert_prot_cmd(struct isert_conn *conn, struct se_cmd *cmd)
@@ -177,12 +180,6 @@ err:
 	return ret;
 }
 
-static void
-isert_cq_event_callback(struct ib_event *e, void *context)
-{
-	isert_dbg("event: %d\n", e->event);
-}
-
 static int
 isert_alloc_rx_descriptors(struct isert_conn *isert_conn)
 {
@@ -212,6 +209,7 @@ isert_alloc_rx_descriptors(struct isert_conn *isert_conn)
 		rx_sg->addr = rx_desc->dma_addr;
 		rx_sg->length = ISER_RX_PAYLOAD_SIZE;
 		rx_sg->lkey = device->pd->local_dma_lkey;
+		rx_desc->rx_cqe.done = isert_recv_done;
 	}
 
 	return 0;
@@ -250,9 +248,6 @@ isert_free_rx_descriptors(struct isert_conn *isert_conn)
 	isert_conn->rx_descs = NULL;
 }
 
-static void isert_cq_work(struct work_struct *);
-static void isert_cq_callback(struct ib_cq *, void *);
-
 static void
 isert_free_comps(struct isert_device *device)
 {
@@ -261,10 +256,8 @@ isert_free_comps(struct isert_device *device)
 	for (i = 0; i < device->comps_used; i++) {
 		struct isert_comp *comp = &device->comps[i];
 
-		if (comp->cq) {
-			cancel_work_sync(&comp->work);
-			ib_destroy_cq(comp->cq);
-		}
+		if (comp->cq)
+			ib_free_cq(comp->cq);
 	}
 	kfree(device->comps);
 }
@@ -293,28 +286,17 @@ isert_alloc_comps(struct isert_device *device)
 	max_cqe = min(ISER_MAX_CQ_LEN, device->ib_device->attrs.max_cqe);
 
 	for (i = 0; i < device->comps_used; i++) {
-		struct ib_cq_init_attr cq_attr = {};
 		struct isert_comp *comp = &device->comps[i];
 
 		comp->device = device;
-		INIT_WORK(&comp->work, isert_cq_work);
-		cq_attr.cqe = max_cqe;
-		cq_attr.comp_vector = i;
-		comp->cq = ib_create_cq(device->ib_device,
-					isert_cq_callback,
-					isert_cq_event_callback,
-					(void *)comp,
-					&cq_attr);
+		comp->cq = ib_alloc_cq(device->ib_device, comp, max_cqe, i,
+				IB_POLL_WORKQUEUE);
 		if (IS_ERR(comp->cq)) {
 			isert_err("Unable to allocate cq\n");
 			ret = PTR_ERR(comp->cq);
 			comp->cq = NULL;
 			goto out_cq;
 		}
-
-		ret = ib_req_notify_cq(comp->cq, IB_CQ_NEXT_COMP);
-		if (ret)
-			goto out_cq;
 	}
 
 	return 0;
@@ -582,7 +564,6 @@ isert_init_conn(struct isert_conn *isert_conn)
 	INIT_LIST_HEAD(&isert_conn->node);
 	init_completion(&isert_conn->login_comp);
 	init_completion(&isert_conn->login_req_comp);
-	init_completion(&isert_conn->wait);
 	kref_init(&isert_conn->kref);
 	mutex_init(&isert_conn->mutex);
 	spin_lock_init(&isert_conn->pool_lock);
@@ -596,11 +577,13 @@ isert_free_login_buf(struct isert_conn *isert_conn)
 	struct ib_device *ib_dev = isert_conn->device->ib_device;
 
 	ib_dma_unmap_single(ib_dev, isert_conn->login_rsp_dma,
-			    ISER_RX_LOGIN_SIZE, DMA_TO_DEVICE);
+			    ISER_RX_PAYLOAD_SIZE, DMA_TO_DEVICE);
+	kfree(isert_conn->login_rsp_buf);
+
 	ib_dma_unmap_single(ib_dev, isert_conn->login_req_dma,
-			    ISCSI_DEF_MAX_RECV_SEG_LEN,
+			    ISER_RX_PAYLOAD_SIZE,
 			    DMA_FROM_DEVICE);
-	kfree(isert_conn->login_buf);
+	kfree(isert_conn->login_req_buf);
 }
 
 static int
@@ -609,50 +592,48 @@ isert_alloc_login_buf(struct isert_conn *isert_conn,
 {
 	int ret;
 
-	isert_conn->login_buf = kzalloc(ISCSI_DEF_MAX_RECV_SEG_LEN +
-					ISER_RX_LOGIN_SIZE, GFP_KERNEL);
-	if (!isert_conn->login_buf) {
+	isert_conn->login_req_buf = kzalloc(sizeof(*isert_conn->login_req_buf),
+			GFP_KERNEL);
+	if (!isert_conn->login_req_buf) {
 		isert_err("Unable to allocate isert_conn->login_buf\n");
 		return -ENOMEM;
 	}
 
-	isert_conn->login_req_buf = isert_conn->login_buf;
-	isert_conn->login_rsp_buf = isert_conn->login_buf +
-				    ISCSI_DEF_MAX_RECV_SEG_LEN;
-
-	isert_dbg("Set login_buf: %p login_req_buf: %p login_rsp_buf: %p\n",
-		 isert_conn->login_buf, isert_conn->login_req_buf,
-		 isert_conn->login_rsp_buf);
-
 	isert_conn->login_req_dma = ib_dma_map_single(ib_dev,
-				(void *)isert_conn->login_req_buf,
-				ISCSI_DEF_MAX_RECV_SEG_LEN, DMA_FROM_DEVICE);
-
+				isert_conn->login_req_buf,
+				ISER_RX_PAYLOAD_SIZE, DMA_FROM_DEVICE);
 	ret = ib_dma_mapping_error(ib_dev, isert_conn->login_req_dma);
 	if (ret) {
 		isert_err("login_req_dma mapping error: %d\n", ret);
 		isert_conn->login_req_dma = 0;
-		goto out_login_buf;
+		goto out_free_login_req_buf;
 	}
 
-	isert_conn->login_rsp_dma = ib_dma_map_single(ib_dev,
-					(void *)isert_conn->login_rsp_buf,
-					ISER_RX_LOGIN_SIZE, DMA_TO_DEVICE);
+	isert_conn->login_rsp_buf = kzalloc(ISER_RX_PAYLOAD_SIZE, GFP_KERNEL);
+	if (!isert_conn->login_rsp_buf) {
+		isert_err("Unable to allocate isert_conn->login_rspbuf\n");
+		goto out_unmap_login_req_buf;
+	}
 
+	isert_conn->login_rsp_dma = ib_dma_map_single(ib_dev,
+					isert_conn->login_rsp_buf,
+					ISER_RX_PAYLOAD_SIZE, DMA_TO_DEVICE);
 	ret = ib_dma_mapping_error(ib_dev, isert_conn->login_rsp_dma);
 	if (ret) {
 		isert_err("login_rsp_dma mapping error: %d\n", ret);
 		isert_conn->login_rsp_dma = 0;
-		goto out_req_dma_map;
+		goto out_free_login_rsp_buf;
 	}
 
 	return 0;
 
-out_req_dma_map:
+out_free_login_rsp_buf:
+	kfree(isert_conn->login_rsp_buf);
+out_unmap_login_req_buf:
 	ib_dma_unmap_single(ib_dev, isert_conn->login_req_dma,
-			    ISCSI_DEF_MAX_RECV_SEG_LEN, DMA_FROM_DEVICE);
-out_login_buf:
-	kfree(isert_conn->login_buf);
+			    ISER_RX_PAYLOAD_SIZE, DMA_FROM_DEVICE);
+out_free_login_req_buf:
+	kfree(isert_conn->login_req_buf);
 	return ret;
 }
 
@@ -726,7 +707,7 @@ isert_connect_request(struct rdma_cm_id *cma_id, struct rdma_cm_event *event)
 	if (ret)
 		goto out_conn_dev;
 
-	ret = isert_rdma_post_recvl(isert_conn);
+	ret = isert_login_post_recv(isert_conn);
 	if (ret)
 		goto out_conn_dev;
 
@@ -773,7 +754,7 @@ isert_connect_release(struct isert_conn *isert_conn)
 		ib_destroy_qp(isert_conn->qp);
 	}
 
-	if (isert_conn->login_buf)
+	if (isert_conn->login_req_buf)
 		isert_free_login_buf(isert_conn);
 
 	isert_device_put(device);
@@ -820,12 +801,30 @@ isert_put_conn(struct isert_conn *isert_conn)
 	kref_put(&isert_conn->kref, isert_release_kref);
 }
 
+static void
+isert_handle_unbound_conn(struct isert_conn *isert_conn)
+{
+	struct isert_np *isert_np = isert_conn->cm_id->context;
+
+	mutex_lock(&isert_np->mutex);
+	if (!list_empty(&isert_conn->node)) {
+		/*
+		 * This means iscsi doesn't know this connection
+		 * so schedule a cleanup ourselves
+		 */
+		list_del_init(&isert_conn->node);
+		isert_put_conn(isert_conn);
+		queue_work(isert_release_wq, &isert_conn->release_work);
+	}
+	mutex_unlock(&isert_np->mutex);
+}
+
 /**
  * isert_conn_terminate() - Initiate connection termination
  * @isert_conn: isert connection struct
  *
  * Notes:
- * In case the connection state is FULL_FEATURE, move state
+ * In case the connection state is BOUND, move state
  * to TEMINATING and start teardown sequence (rdma_disconnect).
  * In case the connection state is UP, complete flush as well.
  *
@@ -837,23 +836,16 @@ isert_conn_terminate(struct isert_conn *isert_conn)
 {
 	int err;
 
-	switch (isert_conn->state) {
-	case ISER_CONN_TERMINATING:
-		break;
-	case ISER_CONN_UP:
-	case ISER_CONN_FULL_FEATURE: /* FALLTHRU */
-		isert_info("Terminating conn %p state %d\n",
-			   isert_conn, isert_conn->state);
-		isert_conn->state = ISER_CONN_TERMINATING;
-		err = rdma_disconnect(isert_conn->cm_id);
-		if (err)
-			isert_warn("Failed rdma_disconnect isert_conn %p\n",
-				   isert_conn);
-		break;
-	default:
-		isert_warn("conn %p teminating in state %d\n",
-			   isert_conn, isert_conn->state);
-	}
+	if (isert_conn->state >= ISER_CONN_TERMINATING)
+		return;
+
+	isert_info("Terminating conn %p state %d\n",
+		   isert_conn, isert_conn->state);
+	isert_conn->state = ISER_CONN_TERMINATING;
+	err = rdma_disconnect(isert_conn->cm_id);
+	if (err)
+		isert_warn("Failed rdma_disconnect isert_conn %p\n",
+			   isert_conn);
 }
 
 static int
@@ -887,35 +879,27 @@ static int
 isert_disconnected_handler(struct rdma_cm_id *cma_id,
 			   enum rdma_cm_event_type event)
 {
-	struct isert_np *isert_np = cma_id->context;
-	struct isert_conn *isert_conn;
-	bool terminating = false;
-
-	if (isert_np->cm_id == cma_id)
-		return isert_np_cma_handler(cma_id->context, event);
-
-	isert_conn = cma_id->qp->qp_context;
+	struct isert_conn *isert_conn = cma_id->qp->qp_context;
 
 	mutex_lock(&isert_conn->mutex);
-	terminating = (isert_conn->state == ISER_CONN_TERMINATING);
-	isert_conn_terminate(isert_conn);
-	mutex_unlock(&isert_conn->mutex);
-
-	isert_info("conn %p completing wait\n", isert_conn);
-	complete(&isert_conn->wait);
-
-	if (terminating)
-		goto out;
-
-	mutex_lock(&isert_np->mutex);
-	if (!list_empty(&isert_conn->node)) {
-		list_del_init(&isert_conn->node);
-		isert_put_conn(isert_conn);
-		queue_work(isert_release_wq, &isert_conn->release_work);
+	switch (isert_conn->state) {
+	case ISER_CONN_TERMINATING:
+		break;
+	case ISER_CONN_UP:
+		isert_conn_terminate(isert_conn);
+		isert_wait4flush(isert_conn);
+		isert_handle_unbound_conn(isert_conn);
+		break;
+	case ISER_CONN_BOUND:
+	case ISER_CONN_FULL_FEATURE: /* FALLTHRU */
+		iscsit_cause_connection_reinstatement(isert_conn->conn, 0);
+		break;
+	default:
+		isert_warn("conn %p teminating in state %d\n",
+			   isert_conn, isert_conn->state);
 	}
-	mutex_unlock(&isert_np->mutex);
+	mutex_unlock(&isert_conn->mutex);
 
-out:
 	return 0;
 }
 
@@ -934,12 +918,16 @@ isert_connect_error(struct rdma_cm_id *cma_id)
 static int
 isert_cma_handler(struct rdma_cm_id *cma_id, struct rdma_cm_event *event)
 {
+	struct isert_np *isert_np = cma_id->context;
 	int ret = 0;
 
 	isert_info("%s (%d): status %d id %p np %p\n",
 		   rdma_event_msg(event->event), event->event,
 		   event->status, cma_id, cma_id->context);
 
+	if (isert_np->cm_id == cma_id)
+		return isert_np_cma_handler(cma_id->context, event->event);
+
 	switch (event->event) {
 	case RDMA_CM_EVENT_CONNECT_REQUEST:
 		ret = isert_connect_request(cma_id, event);
@@ -977,7 +965,8 @@ isert_post_recvm(struct isert_conn *isert_conn, u32 count)
 
 	for (rx_wr = isert_conn->rx_wr, i = 0; i < count; i++, rx_wr++) {
 		rx_desc = &isert_conn->rx_descs[i];
-		rx_wr->wr_id = (uintptr_t)rx_desc;
+
+		rx_wr->wr_cqe = &rx_desc->rx_cqe;
 		rx_wr->sg_list = &rx_desc->rx_sg;
 		rx_wr->num_sge = 1;
 		rx_wr->next = rx_wr + 1;
@@ -985,13 +974,10 @@ isert_post_recvm(struct isert_conn *isert_conn, u32 count)
 	rx_wr--;
 	rx_wr->next = NULL; /* mark end of work requests list */
 
-	isert_conn->post_recv_buf_count += count;
 	ret = ib_post_recv(isert_conn->qp, isert_conn->rx_wr,
 			   &rx_wr_failed);
-	if (ret) {
+	if (ret)
 		isert_err("ib_post_recv() failed with ret: %d\n", ret);
-		isert_conn->post_recv_buf_count -= count;
-	}
 
 	return ret;
 }
@@ -1002,23 +988,20 @@ isert_post_recv(struct isert_conn *isert_conn, struct iser_rx_desc *rx_desc)
 	struct ib_recv_wr *rx_wr_failed, rx_wr;
 	int ret;
 
-	rx_wr.wr_id = (uintptr_t)rx_desc;
+	rx_wr.wr_cqe = &rx_desc->rx_cqe;
 	rx_wr.sg_list = &rx_desc->rx_sg;
 	rx_wr.num_sge = 1;
 	rx_wr.next = NULL;
 
-	isert_conn->post_recv_buf_count++;
 	ret = ib_post_recv(isert_conn->qp, &rx_wr, &rx_wr_failed);
-	if (ret) {
+	if (ret)
 		isert_err("ib_post_recv() failed with ret: %d\n", ret);
-		isert_conn->post_recv_buf_count--;
-	}
 
 	return ret;
 }
 
 static int
-isert_post_send(struct isert_conn *isert_conn, struct iser_tx_desc *tx_desc)
+isert_login_post_send(struct isert_conn *isert_conn, struct iser_tx_desc *tx_desc)
 {
 	struct ib_device *ib_dev = isert_conn->cm_id->device;
 	struct ib_send_wr send_wr, *send_wr_failed;
@@ -1027,8 +1010,10 @@ isert_post_send(struct isert_conn *isert_conn, struct iser_tx_desc *tx_desc)
 	ib_dma_sync_single_for_device(ib_dev, tx_desc->dma_addr,
 				      ISER_HEADERS_LEN, DMA_TO_DEVICE);
 
+	tx_desc->tx_cqe.done = isert_login_send_done;
+
 	send_wr.next	= NULL;
-	send_wr.wr_id	= (uintptr_t)tx_desc;
+	send_wr.wr_cqe	= &tx_desc->tx_cqe;
 	send_wr.sg_list	= tx_desc->tx_sg;
 	send_wr.num_sge	= tx_desc->num_sge;
 	send_wr.opcode	= IB_WR_SEND;
@@ -1056,7 +1041,6 @@ isert_create_send_desc(struct isert_conn *isert_conn,
 	tx_desc->iser_header.flags = ISCSI_CTRL;
 
 	tx_desc->num_sge = 1;
-	tx_desc->isert_cmd = isert_cmd;
 
 	if (tx_desc->tx_sg[0].lkey != device->pd->local_dma_lkey) {
 		tx_desc->tx_sg[0].lkey = device->pd->local_dma_lkey;
@@ -1097,8 +1081,9 @@ isert_init_send_wr(struct isert_conn *isert_conn, struct isert_cmd *isert_cmd,
 {
 	struct iser_tx_desc *tx_desc = &isert_cmd->tx_desc;
 
-	isert_cmd->rdma_wr.iser_ib_op = ISER_IB_SEND;
-	send_wr->wr_id = (uintptr_t)&isert_cmd->tx_desc;
+	isert_cmd->iser_ib_op = ISER_IB_SEND;
+	tx_desc->tx_cqe.done = isert_send_done;
+	send_wr->wr_cqe = &tx_desc->tx_cqe;
 
 	if (isert_conn->snd_w_inv && isert_cmd->inv_rkey) {
 		send_wr->opcode  = IB_WR_SEND_WITH_INV;
@@ -1113,7 +1098,7 @@ isert_init_send_wr(struct isert_conn *isert_conn, struct isert_cmd *isert_cmd,
 }
 
 static int
-isert_rdma_post_recvl(struct isert_conn *isert_conn)
+isert_login_post_recv(struct isert_conn *isert_conn)
 {
 	struct ib_recv_wr rx_wr, *rx_wr_fail;
 	struct ib_sge sge;
@@ -1121,23 +1106,22 @@ isert_rdma_post_recvl(struct isert_conn *isert_conn)
 
 	memset(&sge, 0, sizeof(struct ib_sge));
 	sge.addr = isert_conn->login_req_dma;
-	sge.length = ISER_RX_LOGIN_SIZE;
+	sge.length = ISER_RX_PAYLOAD_SIZE;
 	sge.lkey = isert_conn->device->pd->local_dma_lkey;
 
 	isert_dbg("Setup sge: addr: %llx length: %d 0x%08x\n",
 		sge.addr, sge.length, sge.lkey);
 
+	isert_conn->login_req_buf->rx_cqe.done = isert_login_recv_done;
+
 	memset(&rx_wr, 0, sizeof(struct ib_recv_wr));
-	rx_wr.wr_id = (uintptr_t)isert_conn->login_req_buf;
+	rx_wr.wr_cqe = &isert_conn->login_req_buf->rx_cqe;
 	rx_wr.sg_list = &sge;
 	rx_wr.num_sge = 1;
 
-	isert_conn->post_recv_buf_count++;
 	ret = ib_post_recv(isert_conn->qp, &rx_wr, &rx_wr_fail);
-	if (ret) {
+	if (ret)
 		isert_err("ib_post_recv() failed: %d\n", ret);
-		isert_conn->post_recv_buf_count--;
-	}
 
 	return ret;
 }
@@ -1203,12 +1187,12 @@ isert_put_login_tx(struct iscsi_conn *conn, struct iscsi_login *login,
 			goto post_send;
 		}
 
-		ret = isert_rdma_post_recvl(isert_conn);
+		ret = isert_login_post_recv(isert_conn);
 		if (ret)
 			return ret;
 	}
 post_send:
-	ret = isert_post_send(isert_conn, tx_desc);
+	ret = isert_login_post_send(isert_conn, tx_desc);
 	if (ret)
 		return ret;
 
@@ -1218,7 +1202,7 @@ post_send:
 static void
 isert_rx_login_req(struct isert_conn *isert_conn)
 {
-	struct iser_rx_desc *rx_desc = (void *)isert_conn->login_req_buf;
+	struct iser_rx_desc *rx_desc = isert_conn->login_req_buf;
 	int rx_buflen = isert_conn->login_req_len;
 	struct iscsi_conn *conn = isert_conn->conn;
 	struct iscsi_login *login = conn->conn_login;
@@ -1551,12 +1535,42 @@ isert_rx_opcode(struct isert_conn *isert_conn, struct iser_rx_desc *rx_desc,
 }
 
 static void
-isert_rx_do_work(struct iser_rx_desc *rx_desc, struct isert_conn *isert_conn)
+isert_print_wc(struct ib_wc *wc, const char *type)
+{
+	if (wc->status != IB_WC_WR_FLUSH_ERR)
+		isert_err("%s failure: %s (%d) vend_err %x\n", type,
+			  ib_wc_status_msg(wc->status), wc->status,
+			  wc->vendor_err);
+	else
+		isert_dbg("%s failure: %s (%d)\n", type,
+			  ib_wc_status_msg(wc->status), wc->status);
+}
+
+static void
+isert_recv_done(struct ib_cq *cq, struct ib_wc *wc)
 {
+	struct isert_conn *isert_conn = wc->qp->qp_context;
+	struct ib_device *ib_dev = isert_conn->cm_id->device;
+	struct iser_rx_desc *rx_desc = cqe_to_rx_desc(wc->wr_cqe);
+	struct iscsi_hdr *hdr = &rx_desc->iscsi_header;
 	struct iser_ctrl *iser_ctrl = &rx_desc->iser_header;
 	uint64_t read_va = 0, write_va = 0;
 	uint32_t read_stag = 0, write_stag = 0;
 
+	if (unlikely(wc->status != IB_WC_SUCCESS)) {
+		isert_print_wc(wc, "recv");
+		if (wc->status != IB_WC_WR_FLUSH_ERR)
+			iscsit_cause_connection_reinstatement(isert_conn->conn, 0);
+		return;
+	}
+
+	ib_dma_sync_single_for_cpu(ib_dev, rx_desc->dma_addr,
+			ISER_RX_PAYLOAD_SIZE, DMA_FROM_DEVICE);
+
+	isert_dbg("DMA: 0x%llx, iSCSI opcode: 0x%02x, ITT: 0x%08x, flags: 0x%02x dlen: %d\n",
+		 rx_desc->dma_addr, hdr->opcode, hdr->itt, hdr->flags,
+		 (int)(wc->byte_len - ISER_HEADERS_LEN));
+
 	switch (iser_ctrl->flags & 0xF0) {
 	case ISCSI_CTRL:
 		if (iser_ctrl->flags & ISER_RSV) {
@@ -1584,56 +1598,40 @@ isert_rx_do_work(struct iser_rx_desc *rx_desc, struct isert_conn *isert_conn)
 
 	isert_rx_opcode(isert_conn, rx_desc,
 			read_stag, read_va, write_stag, write_va);
+
+	ib_dma_sync_single_for_device(ib_dev, rx_desc->dma_addr,
+			ISER_RX_PAYLOAD_SIZE, DMA_FROM_DEVICE);
 }
 
 static void
-isert_rcv_completion(struct iser_rx_desc *desc,
-		     struct isert_conn *isert_conn,
-		     u32 xfer_len)
+isert_login_recv_done(struct ib_cq *cq, struct ib_wc *wc)
 {
+	struct isert_conn *isert_conn = wc->qp->qp_context;
 	struct ib_device *ib_dev = isert_conn->cm_id->device;
-	struct iscsi_hdr *hdr;
-	u64 rx_dma;
-	int rx_buflen;
-
-	if ((char *)desc == isert_conn->login_req_buf) {
-		rx_dma = isert_conn->login_req_dma;
-		rx_buflen = ISER_RX_LOGIN_SIZE;
-		isert_dbg("login_buf: Using rx_dma: 0x%llx, rx_buflen: %d\n",
-			 rx_dma, rx_buflen);
-	} else {
-		rx_dma = desc->dma_addr;
-		rx_buflen = ISER_RX_PAYLOAD_SIZE;
-		isert_dbg("req_buf: Using rx_dma: 0x%llx, rx_buflen: %d\n",
-			 rx_dma, rx_buflen);
+
+	if (unlikely(wc->status != IB_WC_SUCCESS)) {
+		isert_print_wc(wc, "login recv");
+		return;
 	}
 
-	ib_dma_sync_single_for_cpu(ib_dev, rx_dma, rx_buflen, DMA_FROM_DEVICE);
+	ib_dma_sync_single_for_cpu(ib_dev, isert_conn->login_req_dma,
+			ISER_RX_PAYLOAD_SIZE, DMA_FROM_DEVICE);
 
-	hdr = &desc->iscsi_header;
-	isert_dbg("iSCSI opcode: 0x%02x, ITT: 0x%08x, flags: 0x%02x dlen: %d\n",
-		 hdr->opcode, hdr->itt, hdr->flags,
-		 (int)(xfer_len - ISER_HEADERS_LEN));
+	isert_conn->login_req_len = wc->byte_len - ISER_HEADERS_LEN;
 
-	if ((char *)desc == isert_conn->login_req_buf) {
-		isert_conn->login_req_len = xfer_len - ISER_HEADERS_LEN;
-		if (isert_conn->conn) {
-			struct iscsi_login *login = isert_conn->conn->conn_login;
+	if (isert_conn->conn) {
+		struct iscsi_login *login = isert_conn->conn->conn_login;
 
-			if (login && !login->first_request)
-				isert_rx_login_req(isert_conn);
-		}
-		mutex_lock(&isert_conn->mutex);
-		complete(&isert_conn->login_req_comp);
-		mutex_unlock(&isert_conn->mutex);
-	} else {
-		isert_rx_do_work(desc, isert_conn);
+		if (login && !login->first_request)
+			isert_rx_login_req(isert_conn);
 	}
 
-	ib_dma_sync_single_for_device(ib_dev, rx_dma, rx_buflen,
-				      DMA_FROM_DEVICE);
+	mutex_lock(&isert_conn->mutex);
+	complete(&isert_conn->login_req_comp);
+	mutex_unlock(&isert_conn->mutex);
 
-	isert_conn->post_recv_buf_count--;
+	ib_dma_sync_single_for_device(ib_dev, isert_conn->login_req_dma,
+				ISER_RX_PAYLOAD_SIZE, DMA_FROM_DEVICE);
 }
 
 static int
@@ -1683,54 +1681,50 @@ isert_unmap_data_buf(struct isert_conn *isert_conn, struct isert_data_buf *data)
 static void
 isert_unmap_cmd(struct isert_cmd *isert_cmd, struct isert_conn *isert_conn)
 {
-	struct isert_rdma_wr *wr = &isert_cmd->rdma_wr;
-
 	isert_dbg("Cmd %p\n", isert_cmd);
 
-	if (wr->data.sg) {
+	if (isert_cmd->data.sg) {
 		isert_dbg("Cmd %p unmap_sg op\n", isert_cmd);
-		isert_unmap_data_buf(isert_conn, &wr->data);
+		isert_unmap_data_buf(isert_conn, &isert_cmd->data);
 	}
 
-	if (wr->rdma_wr) {
+	if (isert_cmd->rdma_wr) {
 		isert_dbg("Cmd %p free send_wr\n", isert_cmd);
-		kfree(wr->rdma_wr);
-		wr->rdma_wr = NULL;
+		kfree(isert_cmd->rdma_wr);
+		isert_cmd->rdma_wr = NULL;
 	}
 
-	if (wr->ib_sge) {
+	if (isert_cmd->ib_sge) {
 		isert_dbg("Cmd %p free ib_sge\n", isert_cmd);
-		kfree(wr->ib_sge);
-		wr->ib_sge = NULL;
+		kfree(isert_cmd->ib_sge);
+		isert_cmd->ib_sge = NULL;
 	}
 }
 
 static void
 isert_unreg_rdma(struct isert_cmd *isert_cmd, struct isert_conn *isert_conn)
 {
-	struct isert_rdma_wr *wr = &isert_cmd->rdma_wr;
-
 	isert_dbg("Cmd %p\n", isert_cmd);
 
-	if (wr->fr_desc) {
-		isert_dbg("Cmd %p free fr_desc %p\n", isert_cmd, wr->fr_desc);
-		if (wr->fr_desc->ind & ISERT_PROTECTED) {
-			isert_unmap_data_buf(isert_conn, &wr->prot);
-			wr->fr_desc->ind &= ~ISERT_PROTECTED;
+	if (isert_cmd->fr_desc) {
+		isert_dbg("Cmd %p free fr_desc %p\n", isert_cmd, isert_cmd->fr_desc);
+		if (isert_cmd->fr_desc->ind & ISERT_PROTECTED) {
+			isert_unmap_data_buf(isert_conn, &isert_cmd->prot);
+			isert_cmd->fr_desc->ind &= ~ISERT_PROTECTED;
 		}
 		spin_lock_bh(&isert_conn->pool_lock);
-		list_add_tail(&wr->fr_desc->list, &isert_conn->fr_pool);
+		list_add_tail(&isert_cmd->fr_desc->list, &isert_conn->fr_pool);
 		spin_unlock_bh(&isert_conn->pool_lock);
-		wr->fr_desc = NULL;
+		isert_cmd->fr_desc = NULL;
 	}
 
-	if (wr->data.sg) {
+	if (isert_cmd->data.sg) {
 		isert_dbg("Cmd %p unmap_sg op\n", isert_cmd);
-		isert_unmap_data_buf(isert_conn, &wr->data);
+		isert_unmap_data_buf(isert_conn, &isert_cmd->data);
 	}
 
-	wr->ib_sge = NULL;
-	wr->rdma_wr = NULL;
+	isert_cmd->ib_sge = NULL;
+	isert_cmd->rdma_wr = NULL;
 }
 
 static void
@@ -1882,52 +1876,70 @@ fail_mr_status:
 }
 
 static void
-isert_completion_rdma_write(struct iser_tx_desc *tx_desc,
-			    struct isert_cmd *isert_cmd)
+isert_rdma_write_done(struct ib_cq *cq, struct ib_wc *wc)
 {
-	struct isert_rdma_wr *wr = &isert_cmd->rdma_wr;
-	struct iscsi_cmd *cmd = isert_cmd->iscsi_cmd;
-	struct se_cmd *se_cmd = &cmd->se_cmd;
-	struct isert_conn *isert_conn = isert_cmd->conn;
+	struct isert_conn *isert_conn = wc->qp->qp_context;
 	struct isert_device *device = isert_conn->device;
+	struct iser_tx_desc *desc = cqe_to_tx_desc(wc->wr_cqe);
+	struct isert_cmd *isert_cmd = tx_desc_to_cmd(desc);
+	struct se_cmd *cmd = &isert_cmd->iscsi_cmd->se_cmd;
 	int ret = 0;
 
-	if (wr->fr_desc && wr->fr_desc->ind & ISERT_PROTECTED) {
-		ret = isert_check_pi_status(se_cmd,
-