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:

   - enable dual mode (initiator + target) qla2xxx operation. (Quinn +
     Himanshu)

   - add a framework for qla2xxx async fabric discovery. (Quinn +
     Himanshu)

   - enable iscsi PDU DDP completion offload in cxgbit/T6 NICs. (Varun)

   - fix target-core handling of aborted failed commands. (Bart)

   - fix a long standing target-core issue NULL pointer dereference with
     active I/O LUN shutdown. (Rob Millner + Bryant + nab)"

* 'for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/nab/target-pending: (44 commits)
  target: Add counters for ABORT_TASK success + failure
  iscsi-target: Fix early login failure statistics misses
  target: Fix NULL dereference during LUN lookup + active I/O shutdown
  target: Delete tmr from list before processing
  target: Fix handling of aborted failed commands
  uapi: fix linux/target_core_user.h userspace compilation errors
  target: export protocol identifier
  qla2xxx: Fix a warning reported by the "smatch" static checker
  target/iscsi: Fix unsolicited data seq_end_offset calculation
  target/cxgbit: add T6 iSCSI DDP completion feature
  target/cxgbit: Enable DDP for T6 only if data sequence and pdu are in order
  target/cxgbit: Use T6 specific macros to get ETH/IP hdr len
  target/cxgbit: use cxgb4_tp_smt_idx() to get smt idx
  target/iscsi: split iscsit_check_dataout_hdr()
  target: Remove command flag CMD_T_DEV_ACTIVE
  target: Remove command flag CMD_T_BUSY
  target: Move session check from target_put_sess_cmd() into target_release_cmd_kref()
  target: Inline transport_cmd_check_stop()
  target: Remove an overly chatty debug message
  target: Stop execution if CMD_T_STOP has been set
  ...

19 files changed, 436 insertions, 289 deletions

diff --git a/drivers/target/iscsi/cxgbit/cxgbit_cm.c b/drivers/target/iscsi/cxgbit/cxgbit_cm.c
index 2fb1bf1a26c5..37a05185dcbe 100644
--- a/drivers/target/iscsi/cxgbit/cxgbit_cm.c
+++ b/drivers/target/iscsi/cxgbit/cxgbit_cm.c
@@ -872,7 +872,8 @@ cxgbit_offload_init(struct cxgbit_sock *csk, int iptype, __u8 *peer_ip,
 			goto out;
 		csk->mtu = ndev->mtu;
 		csk->tx_chan = cxgb4_port_chan(ndev);
-		csk->smac_idx = (cxgb4_port_viid(ndev) & 0x7F) << 1;
+		csk->smac_idx = cxgb4_tp_smt_idx(cdev->lldi.adapter_type,
+						 cxgb4_port_viid(ndev));
 		step = cdev->lldi.ntxq /
 			cdev->lldi.nchan;
 		csk->txq_idx = cxgb4_port_idx(ndev) * step;
@@ -907,7 +908,8 @@ cxgbit_offload_init(struct cxgbit_sock *csk, int iptype, __u8 *peer_ip,
 		port_id = cxgb4_port_idx(ndev);
 		csk->mtu = dst_mtu(dst);
 		csk->tx_chan = cxgb4_port_chan(ndev);
-		csk->smac_idx = (cxgb4_port_viid(ndev) & 0x7F) << 1;
+		csk->smac_idx = cxgb4_tp_smt_idx(cdev->lldi.adapter_type,
+						 cxgb4_port_viid(ndev));
 		step = cdev->lldi.ntxq /
 			cdev->lldi.nports;
 		csk->txq_idx = (port_id * step) +
@@ -1066,6 +1068,7 @@ cxgbit_pass_accept_rpl(struct cxgbit_sock *csk, struct cpl_pass_accept_req *req)
 	struct sk_buff *skb;
 	const struct tcphdr *tcph;
 	struct cpl_t5_pass_accept_rpl *rpl5;
+	struct cxgb4_lld_info *lldi = &csk->com.cdev->lldi;
 	unsigned int len = roundup(sizeof(*rpl5), 16);
 	unsigned int mtu_idx;
 	u64 opt0;
@@ -1111,6 +1114,9 @@ cxgbit_pass_accept_rpl(struct cxgbit_sock *csk, struct cpl_pass_accept_req *req)
 	opt2 = RX_CHANNEL_V(0) |
 		RSS_QUEUE_VALID_F | RSS_QUEUE_V(csk->rss_qid);
 
+	if (!is_t5(lldi->adapter_type))
+		opt2 |= RX_FC_DISABLE_F;
+
 	if (req->tcpopt.tstamp)
 		opt2 |= TSTAMPS_EN_F;
 	if (req->tcpopt.sack)
@@ -1119,8 +1125,13 @@ cxgbit_pass_accept_rpl(struct cxgbit_sock *csk, struct cpl_pass_accept_req *req)
 		opt2 |= WND_SCALE_EN_F;
 
 	hlen = ntohl(req->hdr_len);
-	tcph = (const void *)(req + 1) + ETH_HDR_LEN_G(hlen) +
-		IP_HDR_LEN_G(hlen);
+
+	if (is_t5(lldi->adapter_type))
+		tcph = (struct tcphdr *)((u8 *)(req + 1) +
+		       ETH_HDR_LEN_G(hlen) + IP_HDR_LEN_G(hlen));
+	else
+		tcph = (struct tcphdr *)((u8 *)(req + 1) +
+		       T6_ETH_HDR_LEN_G(hlen) + T6_IP_HDR_LEN_G(hlen));
 
 	if (tcph->ece && tcph->cwr)
 		opt2 |= CCTRL_ECN_V(1);
@@ -1726,7 +1737,7 @@ static bool cxgbit_credit_err(const struct cxgbit_sock *csk)
 	}
 
 	while (skb) {
-		credit += skb->csum;
+		credit += (__force u32)skb->csum;
 		skb = cxgbit_skcb_tx_wr_next(skb);
 	}
 
@@ -1753,6 +1764,7 @@ static void cxgbit_fw4_ack(struct cxgbit_sock *csk, struct sk_buff *skb)
 
 	while (credits) {
 		struct sk_buff *p = cxgbit_sock_peek_wr(csk);
+		const u32 csum = (__force u32)p->csum;
 
 		if (unlikely(!p)) {
 			pr_err("csk 0x%p,%u, cr %u,%u+%u, empty.\n",
@@ -1761,17 +1773,17 @@ static void cxgbit_fw4_ack(struct cxgbit_sock *csk, struct sk_buff *skb)
 			break;
 		}
 
-		if (unlikely(credits < p->csum)) {
+		if (unlikely(credits < csum)) {
 			pr_warn("csk 0x%p,%u, cr %u,%u+%u, < %u.\n",
 				csk,  csk->tid,
 				credits, csk->wr_cred, csk->wr_una_cred,
-				p->csum);
-			p->csum -= credits;
+				csum);
+			p->csum = (__force __wsum)(csum - credits);
 			break;
 		}
 
 		cxgbit_sock_dequeue_wr(csk);
-		credits -= p->csum;
+		credits -= csum;
 		kfree_skb(p);
 	}
 
diff --git a/drivers/target/iscsi/cxgbit/cxgbit_lro.h b/drivers/target/iscsi/cxgbit/cxgbit_lro.h
index 28c11bd1b930..dcaed3a1d23f 100644
--- a/drivers/target/iscsi/cxgbit/cxgbit_lro.h
+++ b/drivers/target/iscsi/cxgbit/cxgbit_lro.h
@@ -31,8 +31,9 @@ enum cxgbit_pducb_flags {
 	PDUCBF_RX_DATA		= (1 << 1), /* received pdu payload */
 	PDUCBF_RX_STATUS	= (1 << 2), /* received ddp status */
 	PDUCBF_RX_DATA_DDPD	= (1 << 3), /* pdu payload ddp'd */
-	PDUCBF_RX_HCRC_ERR	= (1 << 4), /* header digest error */
-	PDUCBF_RX_DCRC_ERR	= (1 << 5), /* data digest error */
+	PDUCBF_RX_DDP_CMP	= (1 << 4), /* ddp completion */
+	PDUCBF_RX_HCRC_ERR	= (1 << 5), /* header digest error */
+	PDUCBF_RX_DCRC_ERR	= (1 << 6), /* data digest error */
 };
 
 struct cxgbit_lro_pdu_cb {
diff --git a/drivers/target/iscsi/cxgbit/cxgbit_main.c b/drivers/target/iscsi/cxgbit/cxgbit_main.c
index 96eedfc49c94..4fd775ace541 100644
--- a/drivers/target/iscsi/cxgbit/cxgbit_main.c
+++ b/drivers/target/iscsi/cxgbit/cxgbit_main.c
@@ -165,29 +165,24 @@ static int cxgbit_uld_state_change(void *handle, enum cxgb4_state state)
 }
 
 static void
-cxgbit_proc_ddp_status(unsigned int tid, struct cpl_rx_data_ddp *cpl,
-		       struct cxgbit_lro_pdu_cb *pdu_cb)
+cxgbit_process_ddpvld(struct cxgbit_sock *csk, struct cxgbit_lro_pdu_cb *pdu_cb,
+		      u32 ddpvld)
 {
-	unsigned int status = ntohl(cpl->ddpvld);
 
-	pdu_cb->flags |= PDUCBF_RX_STATUS;
-	pdu_cb->ddigest = ntohl(cpl->ulp_crc);
-	pdu_cb->pdulen = ntohs(cpl->len);
-
-	if (status & (1 << CPL_RX_ISCSI_DDP_STATUS_HCRC_SHIFT)) {
-		pr_info("tid 0x%x, status 0x%x, hcrc bad.\n", tid, status);
+	if (ddpvld & (1 << CPL_RX_ISCSI_DDP_STATUS_HCRC_SHIFT)) {
+		pr_info("tid 0x%x, status 0x%x, hcrc bad.\n", csk->tid, ddpvld);
 		pdu_cb->flags |= PDUCBF_RX_HCRC_ERR;
 	}
 
-	if (status & (1 << CPL_RX_ISCSI_DDP_STATUS_DCRC_SHIFT)) {
-		pr_info("tid 0x%x, status 0x%x, dcrc bad.\n", tid, status);
+	if (ddpvld & (1 << CPL_RX_ISCSI_DDP_STATUS_DCRC_SHIFT)) {
+		pr_info("tid 0x%x, status 0x%x, dcrc bad.\n", csk->tid, ddpvld);
 		pdu_cb->flags |= PDUCBF_RX_DCRC_ERR;
 	}
 
-	if (status & (1 << CPL_RX_ISCSI_DDP_STATUS_PAD_SHIFT))
-		pr_info("tid 0x%x, status 0x%x, pad bad.\n", tid, status);
+	if (ddpvld & (1 << CPL_RX_ISCSI_DDP_STATUS_PAD_SHIFT))
+		pr_info("tid 0x%x, status 0x%x, pad bad.\n", csk->tid, ddpvld);
 
-	if ((status & (1 << CPL_RX_ISCSI_DDP_STATUS_DDP_SHIFT)) &&
+	if ((ddpvld & (1 << CPL_RX_ISCSI_DDP_STATUS_DDP_SHIFT)) &&
 	    (!(pdu_cb->flags & PDUCBF_RX_DATA))) {
 		pdu_cb->flags |= PDUCBF_RX_DATA_DDPD;
 	}
@@ -201,13 +196,17 @@ cxgbit_lro_add_packet_rsp(struct sk_buff *skb, u8 op, const __be64 *rsp)
 						lro_cb->pdu_idx);
 	struct cpl_rx_iscsi_ddp *cpl = (struct cpl_rx_iscsi_ddp *)(rsp + 1);
 
-	cxgbit_proc_ddp_status(lro_cb->csk->tid, cpl, pdu_cb);
+	cxgbit_process_ddpvld(lro_cb->csk, pdu_cb, be32_to_cpu(cpl->ddpvld));
+
+	pdu_cb->flags |= PDUCBF_RX_STATUS;
+	pdu_cb->ddigest = ntohl(cpl->ulp_crc);
+	pdu_cb->pdulen = ntohs(cpl->len);
 
 	if (pdu_cb->flags & PDUCBF_RX_HDR)
 		pdu_cb->complete = true;
 
-	lro_cb->complete = true;
 	lro_cb->pdu_totallen += pdu_cb->pdulen;
+	lro_cb->complete = true;
 	lro_cb->pdu_idx++;
 }
 
@@ -257,7 +256,7 @@ cxgbit_lro_add_packet_gl(struct sk_buff *skb, u8 op, const struct pkt_gl *gl)
 			cxgbit_skcb_flags(skb) = 0;
 
 		lro_cb->complete = false;
-	} else {
+	} else if (op == CPL_ISCSI_DATA) {
 		struct cpl_iscsi_data *cpl = (struct cpl_iscsi_data *)gl->va;
 
 		offset = sizeof(struct cpl_iscsi_data);
@@ -267,6 +266,36 @@ cxgbit_lro_add_packet_gl(struct sk_buff *skb, u8 op, const struct pkt_gl *gl)
 		pdu_cb->doffset = lro_cb->offset;
 		pdu_cb->nr_dfrags = gl->nfrags;
 		pdu_cb->dfrag_idx = skb_shinfo(skb)->nr_frags;
+		lro_cb->complete = false;
+	} else {
+		struct cpl_rx_iscsi_cmp *cpl;
+
+		cpl = (struct cpl_rx_iscsi_cmp *)gl->va;
+		offset = sizeof(struct cpl_rx_iscsi_cmp);
+		pdu_cb->flags |= (PDUCBF_RX_HDR | PDUCBF_RX_STATUS);
+		len = be16_to_cpu(cpl->len);
+		pdu_cb->hdr = gl->va + offset;
+		pdu_cb->hlen = len;
+		pdu_cb->hfrag_idx = skb_shinfo(skb)->nr_frags;
+		pdu_cb->ddigest = be32_to_cpu(cpl->ulp_crc);
+		pdu_cb->pdulen = ntohs(cpl->len);
+
+		if (unlikely(gl->nfrags > 1))
+			cxgbit_skcb_flags(skb) = 0;
+
+		cxgbit_process_ddpvld(lro_cb->csk, pdu_cb,
+				      be32_to_cpu(cpl->ddpvld));
+
+		if (pdu_cb->flags & PDUCBF_RX_DATA_DDPD) {
+			pdu_cb->flags |= PDUCBF_RX_DDP_CMP;
+			pdu_cb->complete = true;
+		} else if (pdu_cb->flags & PDUCBF_RX_DATA) {
+			pdu_cb->complete = true;
+		}
+
+		lro_cb->pdu_totallen += pdu_cb->hlen + pdu_cb->dlen;
+		lro_cb->complete = true;
+		lro_cb->pdu_idx++;
 	}
 
 	cxgbit_copy_frags(skb, gl, offset);
@@ -413,6 +442,7 @@ cxgbit_uld_lro_rx_handler(void *hndl, const __be64 *rsp,
 	switch (op) {
 	case CPL_ISCSI_HDR:
 	case CPL_ISCSI_DATA:
+	case CPL_RX_ISCSI_CMP:
 	case CPL_RX_ISCSI_DDP:
 	case CPL_FW4_ACK:
 		lro_flush = false;
@@ -454,12 +484,13 @@ cxgbit_uld_lro_rx_handler(void *hndl, const __be64 *rsp,
 		if (unlikely(op != *(u8 *)gl->va)) {
 			pr_info("? FL 0x%p,RSS%#llx,FL %#llx,len %u.\n",
 				gl->va, be64_to_cpu(*rsp),
-				be64_to_cpu(*(u64 *)gl->va),
+				get_unaligned_be64(gl->va),
 				gl->tot_len);
 			return 0;
 		}
 
-		if (op == CPL_ISCSI_HDR || op == CPL_ISCSI_DATA) {
+		if ((op == CPL_ISCSI_HDR) || (op == CPL_ISCSI_DATA) ||
+		    (op == CPL_RX_ISCSI_CMP)) {
 			if (!cxgbit_lro_receive(csk, op, rsp, gl, lro_mgr,
 						napi))
 				return 0;
diff --git a/drivers/target/iscsi/cxgbit/cxgbit_target.c b/drivers/target/iscsi/cxgbit/cxgbit_target.c
index 8bcb9b71f764..2714e5901d18 100644
--- a/drivers/target/iscsi/cxgbit/cxgbit_target.c
+++ b/drivers/target/iscsi/cxgbit/cxgbit_target.c
@@ -162,12 +162,14 @@ cxgbit_tx_data_wr(struct cxgbit_sock *csk, struct sk_buff *skb, u32 dlen,
 		  u32 len, u32 credits, u32 compl)
 {
 	struct fw_ofld_tx_data_wr *req;
+	const struct cxgb4_lld_info *lldi = &csk->com.cdev->lldi;
 	u32 submode = cxgbit_skcb_submode(skb);
 	u32 wr_ulp_mode = 0;
 	u32 hdr_size = sizeof(*req);
 	u32 opcode = FW_OFLD_TX_DATA_WR;
 	u32 immlen = 0;
-	u32 force = TX_FORCE_V(!submode);
+	u32 force = is_t5(lldi->adapter_type) ? TX_FORCE_V(!submode) :
+		    T6_TX_FORCE_F;
 
 	if (cxgbit_skcb_flags(skb) & SKCBF_TX_ISO) {
 		opcode = FW_ISCSI_TX_DATA_WR;
@@ -243,7 +245,7 @@ void cxgbit_push_tx_frames(struct cxgbit_sock *csk)
 		}
 		__skb_unlink(skb, &csk->txq);
 		set_wr_txq(skb, CPL_PRIORITY_DATA, csk->txq_idx);
-		skb->csum = credits_needed + flowclen16;
+		skb->csum = (__force __wsum)(credits_needed + flowclen16);
 		csk->wr_cred -= credits_needed;
 		csk->wr_una_cred += credits_needed;
 
@@ -651,26 +653,6 @@ static int cxgbit_set_iso_npdu(struct cxgbit_sock *csk)
 	u32 max_npdu, max_iso_npdu;
 
 	if (conn->login->leading_connection) {
-		param = iscsi_find_param_from_key(DATASEQUENCEINORDER,
-						  conn->param_list);
-		if (!param) {
-			pr_err("param not found key %s\n", DATASEQUENCEINORDER);
-			return -1;
-		}
-
-		if (strcmp(param->value, YES))
-			return 0;
-
-		param = iscsi_find_param_from_key(DATAPDUINORDER,
-						  conn->param_list);
-		if (!param) {
-			pr_err("param not found key %s\n", DATAPDUINORDER);
-			return -1;
-		}
-
-		if (strcmp(param->value, YES))
-			return 0;
-
 		param = iscsi_find_param_from_key(MAXBURSTLENGTH,
 						  conn->param_list);
 		if (!param) {
@@ -681,11 +663,6 @@ static int cxgbit_set_iso_npdu(struct cxgbit_sock *csk)
 		if (kstrtou32(param->value, 0, &mbl) < 0)
 			return -1;
 	} else {
-		if (!conn->sess->sess_ops->DataSequenceInOrder)
-			return 0;
-		if (!conn->sess->sess_ops->DataPDUInOrder)
-			return 0;
-
 		mbl = conn->sess->sess_ops->MaxBurstLength;
 	}
 
@@ -704,6 +681,53 @@ static int cxgbit_set_iso_npdu(struct cxgbit_sock *csk)
 	return 0;
 }
 
+/*
+ * cxgbit_seq_pdu_inorder()
+ * @csk: pointer to cxgbit socket structure
+ *
+ * This function checks whether data sequence and data
+ * pdu are in order.
+ *
+ * Return: returns -1 on error, 0 if data sequence and
+ * data pdu are in order, 1 if data sequence or data pdu
+ * is not in order.
+ */
+static int cxgbit_seq_pdu_inorder(struct cxgbit_sock *csk)
+{
+	struct iscsi_conn *conn = csk->conn;
+	struct iscsi_param *param;
+
+	if (conn->login->leading_connection) {
+		param = iscsi_find_param_from_key(DATASEQUENCEINORDER,
+						  conn->param_list);
+		if (!param) {
+			pr_err("param not found key %s\n", DATASEQUENCEINORDER);
+			return -1;
+		}
+
+		if (strcmp(param->value, YES))
+			return 1;
+
+		param = iscsi_find_param_from_key(DATAPDUINORDER,
+						  conn->param_list);
+		if (!param) {
+			pr_err("param not found key %s\n", DATAPDUINORDER);
+			return -1;
+		}
+
+		if (strcmp(param->value, YES))
+			return 1;
+
+	} else {
+		if (!conn->sess->sess_ops->DataSequenceInOrder)
+			return 1;
+		if (!conn->sess->sess_ops->DataPDUInOrder)
+			return 1;
+	}
+
+	return 0;
+}
+
 static int cxgbit_set_params(struct iscsi_conn *conn)
 {
 	struct cxgbit_sock *csk = conn->context;
@@ -730,11 +754,24 @@ static int cxgbit_set_params(struct iscsi_conn *conn)
 	}
 
 	if (!erl) {
+		int ret;
+
+		ret = cxgbit_seq_pdu_inorder(csk);
+		if (ret < 0) {
+			return -1;
+		} else if (ret > 0) {
+			if (is_t5(cdev->lldi.adapter_type))
+				goto enable_ddp;
+			else
+				goto enable_digest;
+		}
+
 		if (test_bit(CDEV_ISO_ENABLE, &cdev->flags)) {
 			if (cxgbit_set_iso_npdu(csk))
 				return -1;
 		}
 
+enable_ddp:
 		if (test_bit(CDEV_DDP_ENABLE, &cdev->flags)) {
 			if (cxgbit_setup_conn_pgidx(csk,
 						    ppm->tformat.pgsz_idx_dflt))
@@ -743,6 +780,7 @@ static int cxgbit_set_params(struct iscsi_conn *conn)
 		}
 	}
 
+enable_digest:
 	if (cxgbit_set_digest(csk))
 		return -1;
 
@@ -983,11 +1021,36 @@ static int cxgbit_handle_iscsi_dataout(struct cxgbit_sock *csk)
 	int rc, sg_nents, sg_off;
 	bool dcrc_err = false;
 
-	rc = iscsit_check_dataout_hdr(conn, (unsigned char *)hdr, &cmd);
-	if (rc < 0)
-		return rc;
-	else if (!cmd)
-		return 0;
+	if (pdu_cb->flags & PDUCBF_RX_DDP_CMP) {
+		u32 offset = be32_to_cpu(hdr->offset);
+		u32 ddp_data_len;
+		u32 payload_length = ntoh24(hdr->dlength);
+		bool success = false;
+
+		cmd = iscsit_find_cmd_from_itt_or_dump(conn, hdr->itt, 0);
+		if (!cmd)
+			return 0;
+
+		ddp_data_len = offset - cmd->write_data_done;
+		atomic_long_add(ddp_data_len, &conn->sess->rx_data_octets);
+
+		cmd->write_data_done = offset;
+		cmd->next_burst_len = ddp_data_len;
+		cmd->data_sn = be32_to_cpu(hdr->datasn);
+
+		rc = __iscsit_check_dataout_hdr(conn, (unsigned char *)hdr,
+						cmd, payload_length, &success);
+		if (rc < 0)
+			return rc;
+		else if (!success)
+			return 0;
+	} else {
+		rc = iscsit_check_dataout_hdr(conn, (unsigned char *)hdr, &cmd);
+		if (rc < 0)
+			return rc;
+		else if (!cmd)
+			return 0;
+	}
 
 	if (pdu_cb->flags & PDUCBF_RX_DCRC_ERR) {
 		pr_err("ITT: 0x%08x, Offset: %u, Length: %u,"
@@ -1351,6 +1414,9 @@ static void cxgbit_lro_hskb_reset(struct cxgbit_sock *csk)
 	for (i = 0; i < ssi->nr_frags; i++)
 		put_page(skb_frag_page(&ssi->frags[i]));
 	ssi->nr_frags = 0;
+	skb->data_len = 0;
+	skb->truesize -= skb->len;
+	skb->len = 0;
 }
 
 static void
@@ -1364,39 +1430,42 @@ cxgbit_lro_skb_merge(struct cxgbit_sock *csk, struct sk_buff *skb, u8 pdu_idx)
 	unsigned int len = 0;
 
 	if (pdu_cb->flags & PDUCBF_RX_HDR) {
-		hpdu_cb->flags = pdu_cb->flags;
+		u8 hfrag_idx = hssi->nr_frags;
+
+		hpdu_cb->flags |= pdu_cb->flags;
 		hpdu_cb->seq = pdu_cb->seq;
 		hpdu_cb->hdr = pdu_cb->hdr;
 		hpdu_cb->hlen = pdu_cb->hlen;
 
-		memcpy(&hssi->frags[0], &ssi->frags[pdu_cb->hfrag_idx],
+		memcpy(&hssi->frags[hfrag_idx], &ssi->frags[pdu_cb->hfrag_idx],
 		       sizeof(skb_frag_t));
 
-		get_page(skb_frag_page(&hssi->frags[0]));
-		hssi->nr_frags = 1;
-		hpdu_cb->frags = 1;
-		hpdu_cb->hfrag_idx = 0;
+		get_page(skb_frag_page(&hssi->frags[hfrag_idx]));
+		hssi->nr_frags++;
+		hpdu_cb->frags++;
+		hpdu_cb->hfrag_idx = hfrag_idx;
 
-		len = hssi->frags[0].size;
-		hskb->len = len;
-		hskb->data_len = len;
-		hskb->truesize = len;
+		len = hssi->frags[hfrag_idx].size;
+		hskb->len += len;
+		hskb->data_len += len;
+		hskb->truesize += len;
 	}
 
 	if (pdu_cb->flags & PDUCBF_RX_DATA) {
-		u8 hfrag_idx = 1, i;
+		u8 dfrag_idx = hssi->nr_frags, i;
 
 		hpdu_cb->flags |= pdu_cb->flags;
+		hpdu_cb->dfrag_idx = dfrag_idx;
 
 		len = 0;
-		for (i = 0; i < pdu_cb->nr_dfrags; hfrag_idx++, i++) {
-			memcpy(&hssi->frags[hfrag_idx],
+		for (i = 0; i < pdu_cb->nr_dfrags; dfrag_idx++, i++) {
+			memcpy(&hssi->frags[dfrag_idx],
 			       &ssi->frags[pdu_cb->dfrag_idx + i],
 			       sizeof(skb_frag_t));
 
-			get_page(skb_frag_page(&hssi->frags[hfrag_idx]));
+			get_page(skb_frag_page(&hssi->frags[dfrag_idx]));
 
-			len += hssi->frags[hfrag_idx].size;
+			len += hssi->frags[dfrag_idx].size;
 
 			hssi->nr_frags++;
 			hpdu_cb->frags++;
@@ -1405,7 +1474,6 @@ cxgbit_lro_skb_merge(struct cxgbit_sock *csk, struct sk_buff *skb, u8 pdu_idx)
 		hpdu_cb->dlen = pdu_cb->dlen;
 		hpdu_cb->doffset = hpdu_cb->hlen;
 		hpdu_cb->nr_dfrags = pdu_cb->nr_dfrags;
-		hpdu_cb->dfrag_idx = 1;
 		hskb->len += len;
 		hskb->data_len += len;
 		hskb->truesize += len;
@@ -1490,10 +1558,15 @@ static int cxgbit_rx_lro_skb(struct cxgbit_sock *csk, struct sk_buff *skb)
 
 static int cxgbit_rx_skb(struct cxgbit_sock *csk, struct sk_buff *skb)
 {
+	struct cxgb4_lld_info *lldi = &csk->com.cdev->lldi;
 	int ret = -1;
 
-	if (likely(cxgbit_skcb_flags(skb) & SKCBF_RX_LRO))
-		ret = cxgbit_rx_lro_skb(csk, skb);
+	if (likely(cxgbit_skcb_flags(skb) & SKCBF_RX_LRO)) {
+		if (is_t5(lldi->adapter_type))
+			ret = cxgbit_rx_lro_skb(csk, skb);
+		else
+			ret = cxgbit_process_lro_skb(csk, skb);
+	}
 
 	__kfree_skb(skb);
 	return ret;
diff --git a/drivers/target/iscsi/iscsi_target.c b/drivers/target/iscsi/iscsi_target.c
index da2c73a255de..2285988c209b 100644
--- a/drivers/target/iscsi/iscsi_target.c
+++ b/drivers/target/iscsi/iscsi_target.c
@@ -1431,36 +1431,17 @@ static void iscsit_do_crypto_hash_buf(
 }
 
 int
-iscsit_check_dataout_hdr(struct iscsi_conn *conn, unsigned char *buf,
-			  struct iscsi_cmd **out_cmd)
+__iscsit_check_dataout_hdr(struct iscsi_conn *conn, void *buf,
+			   struct iscsi_cmd *cmd, u32 payload_length,
+			   bool *success)
 {
-	struct iscsi_data *hdr = (struct iscsi_data *)buf;
-	struct iscsi_cmd *cmd = NULL;
+	struct iscsi_data *hdr = buf;
 	struct se_cmd *se_cmd;
-	u32 payload_length = ntoh24(hdr->dlength);
 	int rc;
 
-	if (!payload_length) {
-		pr_warn("DataOUT payload is ZERO, ignoring.\n");
-		return 0;
-	}
-
 	/* iSCSI write */
 	atomic_long_add(payload_length, &conn->sess->rx_data_octets);
 
-	if (payload_length > conn->conn_ops->MaxXmitDataSegmentLength) {
-		pr_err("DataSegmentLength: %u is greater than"
-			" MaxXmitDataSegmentLength: %u\n", payload_length,
-			conn->conn_ops->MaxXmitDataSegmentLength);
-		return iscsit_add_reject(conn, ISCSI_REASON_PROTOCOL_ERROR,
-					 buf);
-	}
-
-	cmd = iscsit_find_cmd_from_itt_or_dump(conn, hdr->itt,
-			payload_length);
-	if (!cmd)
-		return 0;
-
 	pr_debug("Got DataOut ITT: 0x%08x, TTT: 0x%08x,"
 		" DataSN: 0x%08x, Offset: %u, Length: %u, CID: %hu\n",
 		hdr->itt, hdr->ttt, hdr->datasn, ntohl(hdr->offset),
@@ -1545,7 +1526,7 @@ iscsit_check_dataout_hdr(struct iscsi_conn *conn, unsigned char *buf,
 		}
 	}
 	/*
-	 * Preform DataSN, DataSequenceInOrder, DataPDUInOrder, and
+	 * Perform DataSN, DataSequenceInOrder, DataPDUInOrder, and
 	 * within-command recovery checks before receiving the payload.
 	 */
 	rc = iscsit_check_pre_dataout(cmd, buf);
@@ -1553,10 +1534,44 @@ iscsit_check_dataout_hdr(struct iscsi_conn *conn, unsigned char *buf,
 		return 0;
 	else if (rc == DATAOUT_CANNOT_RECOVER)
 		return -1;
-
-	*out_cmd = cmd;
+	*success = true;
 	return 0;
 }
+EXPORT_SYMBOL(__iscsit_check_dataout_hdr);
+
+int
+iscsit_check_dataout_hdr(struct iscsi_conn *conn, void *buf,
+			 struct iscsi_cmd **out_cmd)
+{
+	struct iscsi_data *hdr = buf;
+	struct iscsi_cmd *cmd;
+	u32 payload_length = ntoh24(hdr->dlength);
+	int rc;
+	bool success = false;
+
+	if (!payload_length) {
+		pr_warn_ratelimited("DataOUT payload is ZERO, ignoring.\n");
+		return 0;
+	}
+
+	if (payload_length > conn->conn_ops->MaxXmitDataSegmentLength) {
+		pr_err_ratelimited("DataSegmentLength: %u is greater than"
+			" MaxXmitDataSegmentLength: %u\n", payload_length,
+			conn->conn_ops->MaxXmitDataSegmentLength);
+		return iscsit_add_reject(conn, ISCSI_REASON_PROTOCOL_ERROR, buf);
+	}
+
+	cmd = iscsit_find_cmd_from_itt_or_dump(conn, hdr->itt, payload_length);
+	if (!cmd)
+		return 0;
+
+	rc = __iscsit_check_dataout_hdr(conn, buf, cmd, payload_length, &success);
+
+	if (success)
+		*out_cmd = cmd;
+
+	return rc;
+}
 EXPORT_SYMBOL(iscsit_check_dataout_hdr);
 
 static int
@@ -1920,6 +1935,28 @@ out:
 	return ret;
 }
 
+static enum tcm_tmreq_table iscsit_convert_tmf(u8 iscsi_tmf)
+{
+	switch (iscsi_tmf) {
+	case ISCSI_TM_FUNC_ABORT_TASK:
+		return TMR_ABORT_TASK;
+	case ISCSI_TM_FUNC_ABORT_TASK_SET:
+		return TMR_ABORT_TASK_SET;
+	case ISCSI_TM_FUNC_CLEAR_ACA:
+		return TMR_CLEAR_ACA;
+	case ISCSI_TM_FUNC_CLEAR_TASK_SET:
+		return TMR_CLEAR_TASK_SET;
+	case ISCSI_TM_FUNC_LOGICAL_UNIT_RESET:
+		return TMR_LUN_RESET;
+	case ISCSI_TM_FUNC_TARGET_WARM_RESET:
+		return TMR_TARGET_WARM_RESET;
+	case ISCSI_TM_FUNC_TARGET_COLD_RESET:
+		return TMR_TARGET_COLD_RESET;
+	default:
+		return TMR_UNKNOWN;
+	}
+}
+
 int
 iscsit_handle_task_mgt_cmd(struct iscsi_conn *conn, struct iscsi_cmd *cmd,
 			   unsigned char *buf)
@@ -1929,7 +1966,7 @@ iscsit_handle_task_mgt_cmd(struct iscsi_conn *conn, struct iscsi_cmd *cmd,
 	struct iscsi_tm *hdr;
 	int out_of_order_cmdsn = 0, ret;
 	bool sess_ref = false;
-	u8 function;
+	u8 function, tcm_function = TMR_UNKNOWN;
 
 	hdr			= (struct iscsi_tm *) buf;
 	hdr->flags &= ~ISCSI_FLAG_CMD_FINAL;
@@ -1975,54 +2012,27 @@ iscsit_handle_task_mgt_cmd(struct iscsi_conn *conn, struct iscsi_cmd *cmd,
 	 * LIO-Target $FABRIC_MOD
 	 */
 	if (function != ISCSI_TM_FUNC_TASK_REASSIGN) {
-
-		u8 tcm_function;
-		int ret;
-
 		transport_init_se_cmd(&cmd->se_cmd, &iscsi_ops,
 				      conn->sess->se_sess, 0, DMA_NONE,
 				      TCM_SIMPLE_TAG, cmd->sense_buffer + 2);
 
 		target_get_sess_cmd(&cmd->se_cmd, true);
 		sess_ref = true;
-
-		switch (function) {
-		case ISCSI_TM_FUNC_ABORT_TASK:
-			tcm_function = TMR_ABORT_TASK;
-			break;
-		case ISCSI_TM_FUNC_ABORT_TASK_SET:
-			tcm_function = TMR_ABORT_TASK_SET;
-			break;
-		case ISCSI_TM_FUNC_CLEAR_ACA:
-			tcm_function = TMR_CLEAR_ACA;
-			break;
-		case ISCSI_TM_FUNC_CLEAR_TASK_SET:
-			tcm_function = TMR_CLEAR_TASK_SET;
-			break;
-		case ISCSI_TM_FUNC_LOGICAL_UNIT_RESET:
-			tcm_function = TMR_LUN_RESET;
-			break;
-		case ISCSI_TM_FUNC_TARGET_WARM_RESET:
-			tcm_function = TMR_TARGET_WARM_RESET;
-			break;
-		case ISCSI_TM_FUNC_TARGET_COLD_RESET:
-			tcm_function = TMR_TARGET_COLD_RESET;
-			break;
-		default:
+		tcm_function = iscsit_convert_tmf(function);
+		if (tcm_function == TMR_UNKNOWN) {
 			pr_err("Unknown iSCSI TMR Function:"
 			       " 0x%02x\n", function);
 			return iscsit_add_reject_cmd(cmd,
 				ISCSI_REASON_BOOKMARK_NO_RESOURCES, buf);
 		}
-
-		ret = core_tmr_alloc_req(&cmd->se_cmd, cmd->tmr_req,
-					 tcm_function, GFP_KERNEL);
-		if (ret < 0)
-			return iscsit_add_reject_cmd(cmd,
+	}
+	ret = core_tmr_alloc_req(&cmd->se_cmd, cmd->tmr_req, tcm_function,
+				 GFP_KERNEL);
+	if (ret < 0)
+		return iscsit_add_reject_cmd(cmd,
 				ISCSI_REASON_BOOKMARK_NO_RESOURCES, buf);
 
-		cmd->tmr_req->se_tmr_req = cmd->se_cmd.se_tmr_req;
-	}
+	cmd->tmr_req->se_tmr_req = cmd->se_cmd.se_tmr_req;
 
 	cmd->iscsi_opcode	= ISCSI_OP_SCSI_TMFUNC;
 	cmd->i_state		= ISTATE_SEND_TASKMGTRSP;
@@ -4136,7 +4146,7 @@ int iscsit_close_connection(
 	/*
 	 * During Connection recovery drop unacknowledged out of order
 	 * commands for this connection, and prepare the other commands
-	 * for realligence.
+	 * for reallegiance.
 	 *
 	 * During normal operation clear the out of order commands (but
 	 * do not free the struct iscsi_ooo_cmdsn's) and release all
@@ -4144,7 +4154,7 @@ int iscsit_close_connection(
 	 */
 	if (atomic_read(&conn->connection_recovery)) {
 		iscsit_discard_unacknowledged_ooo_cmdsns_for_conn(conn);
-		iscsit_prepare_cmds_for_realligance(conn);
+		iscsit_prepare_cmds_for_reallegiance(conn);
 	} else {
 		iscsit_clear_ooo_cmdsns_for_conn(conn);
 		iscsit_release_commands_from_conn(conn);
diff --git a/drivers/target/iscsi/iscsi_target_erl0.c b/drivers/target/iscsi/iscsi_target_erl0.c
index b54e72c7ab0f..a8bcbc43b047 100644
--- a/drivers/target/iscsi/iscsi_target_erl0.c
+++ b/drivers/target/iscsi/iscsi_target_erl0.c
@@ -44,10 +44,8 @@ void iscsit_set_dataout_sequence_values(
 	 */
 	if (cmd->unsolicited_data) {
 		cmd->seq_start_offset = cmd->write_data_done;
-		cmd->seq_end_offset = (cmd->write_data_done +
-			((cmd->se_cmd.data_length >
-			  conn->sess->sess_ops->FirstBurstLength) ?
-			 conn->sess->sess_ops->FirstBurstLength : cmd->se_cmd.data_length));
+		cmd->seq_end_offset = min(cmd->se_cmd.data_length,
+					conn->sess->sess_ops->FirstBurstLength);
 		return;
 	}
 
diff --git a/drivers/target/iscsi/iscsi_target_erl2.c b/drivers/target/iscsi/iscsi_target_erl2.c
index faf9ae014b30..8df9c90f3db3 100644
--- a/drivers/target/iscsi/iscsi_target_erl2.c
+++ b/drivers/target/iscsi/iscsi_target_erl2.c
@@ -312,7 +312,7 @@ int iscsit_discard_unacknowledged_ooo_cmdsns_for_conn(struct iscsi_conn *conn)
 	return 0;
 }
 
-int iscsit_prepare_cmds_for_realligance(struct iscsi_conn *conn)
+int iscsit_prepare_cmds_for_reallegiance(struct iscsi_conn *conn)
 {
 	u32 cmd_count = 0;
 	struct iscsi_cmd *cmd, *cmd_tmp;
@@ -347,7 +347,7 @@ int iscsit_prepare_cmds_for_realligance(struct iscsi_conn *conn)
 
 		if ((cmd->iscsi_opcode != ISCSI_OP_SCSI_CMD) &&
 		    (cmd->iscsi_opcode != ISCSI_OP_NOOP_OUT)) {
-			pr_debug("Not performing realligence on"
+			pr_debug("Not performing reallegiance on"
 				" Opcode: 0x%02x, ITT: 0x%08x, CmdSN: 0x%08x,"
 				" CID: %hu\n", cmd->iscsi_opcode,
 				cmd->init_task_tag, cmd->cmd_sn, conn->cid);
@@ -382,7 +382,7 @@ int iscsit_prepare_cmds_for_realligance(struct iscsi_conn *conn)
 		cmd_count++;
 		pr_debug("Preparing Opcode: 0x%02x, ITT: 0x%08x,"
 			" CmdSN: 0x%08x, StatSN: 0x%08x, CID: %hu for"
-			" realligence.\n", cmd->iscsi_opcode,
+			" reallegiance.\n", cmd->iscsi_opcode,
 			cmd->init_task_tag, cmd->cmd_sn, cmd->stat_sn,
 			conn->cid);
 
diff --git a/drivers/target/iscsi/iscsi_target_erl2.h b/drivers/target/iscsi/iscsi_target_erl2.h
index 7965f1e86506..634d01e13652 100644
--- a/drivers/target/iscsi/iscsi_target_erl2.h
+++ b/drivers/target/iscsi/iscsi_target_erl2.h
@@ -19,7 +19,7 @@ extern int iscsit_remove_cmd_from_connection_recovery(struct iscsi_cmd *,
 			struct iscsi_session *);
 extern void iscsit_discard_cr_cmds_by_expstatsn(struct iscsi_conn_recovery *, u32);
 extern int iscsit_discard_unacknowledged_ooo_cmdsns_for_conn(struct iscsi_conn *);
-extern int iscsit_prepare_cmds_for_realligance(struct iscsi_conn *);
+extern int iscsit_prepare_cmds_for_reallegiance(struct iscsi_conn *);
 extern int iscsit_connection_recovery_transport_reset(struct iscsi_conn *);
 
 #endif /*** ISCSI_TARGET_ERL2_H ***/
diff --git a/drivers/target/iscsi/iscsi_target_login.c b/drivers/target/iscsi/iscsi_target_login.c
index eab274d17b5c..746b97f8e4f7 100644
--- a/drivers/target/iscsi/iscsi_target_login.c
+++ b/drivers/target/iscsi/iscsi_target_login.c
@@ -223,7 +223,7 @@ int iscsi_check_for_session_reinstatement(struct iscsi_conn *conn)
 		return 0;
 
 	pr_debug("%s iSCSI Session SID %u is still active for %s,"
-		" preforming session reinstatement.\n", (sessiontype) ?
+		" performing session reinstatement.\n", (sessiontype) ?
 		"Discovery" : "Normal", sess->sid,
 		sess->sess_ops->InitiatorName);
 
diff --git a/drivers/target/iscsi/iscsi_target_nego.c b/drivers/target/iscsi/iscsi_target_nego.c
index 46388c9e08da..5269e9ef031c 100644
--- a/drivers/target/iscsi/iscsi_target_nego.c
+++ b/drivers/target/iscsi/iscsi_target_nego.c
@@ -1249,16 +1249,16 @@ int iscsi_target_start_negotiation(
 {
 	int ret;
 
-       if (conn->sock) {
-               struct sock *sk = conn->sock->sk;
+	if (conn->sock) {
+		struct sock *sk = conn->sock->sk;
 
-               write_lock_bh(&sk->sk_callback_lock);
-               set_bit(LOGIN_FLAGS_READY, &conn->login_flags);
-               write_unlock_bh(&sk->sk_callback_lock);
-       }
+		write_lock_bh(&sk->sk_callback_lock);
+		set_bit(LOGIN_FLAGS_READY, &conn->login_flags);
+		write_unlock_bh(&sk->sk_callback_lock);
+	}
 
-       ret = iscsi_target_do_login(conn, login);
-       if (ret < 0) {
+	ret = iscsi_target_do_login(conn, login);
+	if (ret < 0) {
 		cancel_delayed_work_sync(&conn->login_work);
 		cancel_delayed_work_sync(&conn->login_cleanup_work);
 		iscsi_target_restore_sock_callbacks(conn);
diff --git a/drivers/target/iscsi/iscsi_target_tmr.c b/drivers/target/iscsi/iscsi_target_tmr.c
index 3d637055c36f..cb231c907d51 100644
--- a/drivers/target/iscsi/iscsi_target_tmr.c
+++ b/drivers/target/iscsi/iscsi_target_tmr.c
@@ -440,14 +440,14 @@ static int iscsit_task_reassign_complete(
 		break;
 	default:
 		 pr_err("Illegal iSCSI Opcode 0x%02x during"
-			" command realligence\n", cmd->iscsi_opcode);
+			" command reallegiance\n", cmd->iscsi_opcode);
 		return -1;
 	}
 
 	if (ret != 0)
 		return ret;
 
-	pr_debug("Completed connection realligence for Opcode: 0x%02x,"
+	pr_debug("Completed connection reallegiance for Opcode: 0x%02x,"
 		" ITT: 0x%08x to CID: %hu.\n", cmd->iscsi_opcode,
 			cmd->init_task_tag, conn->cid);
 
diff --git a/drivers/target/iscsi/iscsi_target_util.c b/drivers/target/iscsi/iscsi_target_util.c
index b5a1b4ccba12..cc5958882431 100644
--- a/drivers/target/iscsi/iscsi_target_util.c
+++ b/drivers/target/iscsi/iscsi_target_util.c
@@ -417,6 +417,7 @@ struct iscsi_cmd *iscsit_find_cmd_from_itt_or_dump(
 
 	return NULL;
 }
+EXPORT_SYMBOL(iscsit_find_cmd_from_itt_or_dump);
 
 struct iscsi_cmd *iscsit_find_cmd_from_ttt(
 	struct iscsi_conn *conn,
@@ -1377,33 +1378,6 @@ int tx_data(
 	return iscsit_do_tx_data(conn, &c);
 }
 
-static bool sockaddr_equal(struct sockaddr_storage *x, struct sockaddr_storage *y)
-{
-	switch (x->ss_family) {
-	case AF_INET: {
-		struct sockaddr_in *sinx = (struct sockaddr_in *)x;
-		struct sockaddr_in *siny = (struct sockaddr_in *)y;
-		if (sinx->sin_addr.s_addr != siny->sin_addr.s_addr)
-			return false;
-		if (sinx->sin_port != siny->sin_port)
-			return false;
-		break;
-	}
-	case AF_INET6: {
-		struct sockaddr_in6 *sinx = (struct sockaddr_in6 *)x;
-		struct sockaddr_in6 *siny = (struct sockaddr_in6 *)y;
-		if (!ipv6_addr_equal(&sinx->sin6_addr, &siny->sin6_addr))
-			return false;
-		if (sinx->sin6_port != siny->sin6_port)
-			return false;
-		break;
-	}
-	default:
-		return false;
-	}
-	return true;
-}
-
 void iscsit_collect_login_stats(
 	struct iscsi_conn *conn,
 	u8 status_class,
@@ -1420,13 +1394,6 @@ void iscsit_collect_login_stats(
 	ls = &tiqn->login_stats;
 
 	spin_lock(&ls->lock);
-	if (sockaddr_equal(&conn->login_sockaddr, &ls->last_intr_fail_sockaddr) &&
-	    ((get_jiffies_64() - ls->last_fail_time) < 10)) {
-		/* We already have the failure info for this login */
-		spin_unlock(&ls->lock);
-		return;
-	}
-
 	if (status_class == ISCSI_STATUS_CLS_SUCCESS)
 		ls->accepts++;
 	else if (status_class == ISCSI_STATUS_CLS_REDIRECT) {
@@ -1471,10 +1438,10 @@ struct iscsi_tiqn *iscsit_snmp_get_tiqn(struct iscsi_conn *conn)
 {
 	struct iscsi_portal_group *tpg;
 
-	if (!conn || !conn->sess)
+	if (!conn)
 		return NULL;
 
-	tpg = conn->sess->tpg;
+	tpg = conn->tpg;
 	if (!tpg)
 		return NULL;
 
diff --git a/drivers/target/target_core_device.c b/drivers/target/target_core_device.c
index 26929c44d703..c754ae33bf7b 100644
--- a/drivers/target/target_core_device.c
+++ b/drivers/target/target_core_device.c
@@ -78,12 +78,16 @@ transport_lookup_cmd_lun(struct se_cmd *se_cmd, u64 unpacked_lun)
 					&deve->read_bytes);
 
 		se_lun = rcu_dereference(deve->se_lun);
+
+		if (!percpu_ref_tryget_live(&se_lun->lun_ref)) {
+			se_lun = NULL;
+			goto out_unlock;
+		}
+
 		se_cmd->se_lun = rcu_dereference(deve->se_lun);
 		se_cmd->pr_res_key = deve->pr_res_key;
 		se_cmd->orig_fe_lun = unpacked_lun;
 		se_cmd->se_cmd_flags |= SCF_SE_LUN_CMD;
-
-		percpu_ref_get(&se_lun->lun_ref);
 		se_cmd->lun_ref_active = true;
 
 		if ((se_cmd->data_direction == DMA_TO_DEVICE) &&
@@ -97,6 +101,7 @@ transport_lookup_cmd_lun(struct se_cmd *se_cmd, u64 unpacked_lun)
 			goto ref_dev;
 		}
 	}
+out_unlock:
 	rcu_read_unlock();
 
 	if (!se_lun) {
@@ -163,7 +168,6 @@ int transport_lookup_tmr_lun(struct se_cmd *se_cmd, u64 unpacked_lun)
 	rcu_read_lock();
 	deve = target_nacl_find_deve(nacl, unpacked_lun);
 	if (deve) {
-		se_tmr->tmr_lun = rcu_dereference(deve->se_lun);
 		se_cmd->se_lun = rcu_dereference(deve->se_lun);
 		se_lun = rcu_dereference(deve->se_lun);
 		se_cmd->pr_res_key = deve->pr_res_key;
@@ -816,6 +820,7 @@ struct se_device *target_alloc_device(struct se_hba *hba, const char *name)
 	xcopy_lun = &dev->xcopy_lun;
 	rcu_assign_pointer(xcopy_lun->lun_se_dev, dev);
 	init_completion(&xcopy_lun->lun_ref_comp);
+	init_completion(&xcopy_lun->lun_shutdown_comp);
 	INIT_LIST_HEAD(&xcopy_lun->lun_deve_list);
 	INIT_LIST_HEAD(&xcopy_lun->lun_dev_link);
 	mutex_init(&xcopy_lun->lun_tg_pt_md_mutex);
diff --git a/drivers/target/target_core_sbc.c b/drivers/target/target_core_sbc.c
index df7b6e95c019..68d8aef7ab78 100644
--- a/drivers/target/target_core_sbc.c
+++ b/drivers/target/target_core_sbc.c
@@ -604,7 +604,7 @@ static sense_reason_t compare_and_write_callback(struct se_cmd *cmd, bool succes
 
 	spin_lock_irq(&cmd->t_state_lock);
 	cmd->t_state = TRANSPORT_PROCESSING;
-	cmd->transport_state |= CMD_T_ACTIVE|CMD_T_BUSY|CMD_T_SENT;
+	cmd->transport_state |= CMD_T_ACTIVE | CMD_T_SENT;
 	spin_unlock_irq(&cmd->t_state_lock);
 
 	__target_execute_cmd(cmd, false);
diff --git a/drivers/target/target_core_stat.c b/drivers/target/target_core_stat.c
index 1a39033d2bff..8038255b21e8 100644
--- a/drivers/target/target_core_stat.c
+++ b/drivers/target/target_core_stat.c
@@ -158,12 +158,28 @@ static ssize_t target_stat_tgt_resets_show(struct config_item *item,
 			atomic_long_read(&to_stat_tgt_dev(item)->num_resets));
 }
 
+static ssize_t target_stat_tgt_aborts_complete_show(struct config_item *item,
+		char *page)
+{
+	return snprintf(page, PAGE_SIZE, "%lu\n",
+			atomic_long_read(&to_stat_tgt_dev(item)->aborts_complete));
+}
+
+static ssize_t target_stat_tgt_aborts_no_task_show(struct config_item *item,
+		char *page)
+{
+	return snprintf(page, PAGE_SIZE, "%lu\n",
+			atomic_long_read(&to_stat_tgt_dev(item)->aborts_no_task));
+}
+
 CONFIGFS_ATTR_RO(target_stat_tgt_, inst);
 CONFIGFS_ATTR_RO(target_stat_tgt_, indx);
 CONFIGFS_ATTR_RO(target_stat_tgt_, num_lus);
 CONFIGFS_ATTR_RO(target_stat_tgt_, status);
 CONFIGFS_ATTR_RO(target_stat_tgt_, non_access_lus);
 CONFIGFS_ATTR_RO(target_stat_tgt_, resets);
+CONFIGFS_ATTR_RO(target_stat_tgt_, aborts_complete);
+CONFIGFS_ATTR_RO(target_stat_tgt_, aborts_no_task);
 
 static struct configfs_attribute *target_stat_scsi_tgt_dev_attrs[] = {
 	&target_stat_tgt_attr_inst,
@@ -172,6 +188,8 @@ static struct configfs_attribute *target_stat_scsi_tgt_dev_attrs[] = {
 	&target_stat_tgt_attr_status,
 	&target_stat_tgt_attr_non_access_lus,
 	&target_stat_tgt_attr_resets,
+	&target_stat_tgt_attr_aborts_complete,
+	&target_stat_tgt_attr_aborts_no_task,
 	NULL,
 };
 
@@ -795,16 +813,34 @@ static ssize_t target_stat_transport_dev_name_show(struct config_item *item,
 	return ret;
 }
 
+static ssize_t target_stat_transport_proto_id_show(struct config_item *item,
+		char *page)
+{
+	struct se_lun *lun = to_transport_stat(item);
+	struct se_device *dev;
+	struct se_portal_group *tpg = lun->lun_tpg;
+	ssize_t ret = -ENODEV;
+
+	rcu_read_lock();
+	dev = rcu_dereference(lun->lun_se_dev);
+	if (dev)
+		ret = snprintf(page, PAGE_SIZE, "%u\n", tpg->proto_id);
+	rcu_read_unlock();
+	return ret;
+}
+
 CONFIGFS_ATTR_RO(target_stat_transport_, inst);
 CONFIGFS_ATTR_RO(target_stat_transport_, device);
 CONFIGFS_ATTR_RO(target_stat_transport_, indx);
 CONFIGFS_ATTR_RO(target_stat_transport_, dev_name);
+CONFIGFS_ATTR_RO(target_stat_transport_, proto_id);
 
 static struct configfs_attribute *target_stat_scsi_transport_attrs[] = {
 	&target_stat_transport_attr_inst,
 	&target_stat_transport_attr_device,
 	&target_stat_transport_attr_indx,
 	&target_stat_transport_attr_dev_name,
+	&target_stat_transport_attr_proto_id,
 	NULL,
 };
 
diff --git a/drivers/target/target_core_tmr.c b/drivers/target/target_core_tmr.c
index 4f229e711e1c..dce1e1b47316 100644
--- a/drivers/target/target_core_tmr.c
+++ b/drivers/target/target_core_tmr.c
@@ -175,10 +175,9 @@ void core_tmr_abort_task(
 		printk("ABORT_TASK: Found referenced %s task_tag: %llu\n",
 			se_cmd->se_tfo->get_fabric_name(), ref_tag);
 
-		if (!__target_check_io_state(se_cmd, se_sess, 0)) {
-			spin_unlock_irqrestore(&se_sess->sess_cmd_lock, flags);
-			goto out;
-		}
+		if (!__target_check_io_state(se_cmd, se_sess, 0))
+			continue;
+
 		list_del_init(&se_cmd->se_cmd_list);
 		spin_unlock_irqrestore(&se_sess->sess_cmd_lock, flags);
 
@@ -191,14 +190,15 @@ void core_tmr_abort_task(
 		printk("ABORT_TASK: Sending TMR_FUNCTION_COMPLETE for"
 				" ref_tag: %llu\n", ref_tag);
 		tmr->response = TMR_FUNCTION_COMPLETE;
+		atomic_long_inc(&dev->aborts_complete);
 		return;
 	}
 	spin_unlock_irqrestore(&se_sess->sess_cmd_lock, flags);
 
-out:
 	printk("ABORT_TASK: Sending TMR_TASK_DOES_NOT_EXIST for ref_tag: %lld\n",
 			tmr->ref_task_tag);
 	tmr->response = TMR_TASK_DOES_NOT_EXIST;
+	atomic_long_inc(&dev->aborts_no_task);
 }
 
 static void core_tmr_drain_tmr_list(
@@ -217,13 +217,8 @@ static void core_tmr_drain_tmr_list(
 	 * LUN_RESET tmr..
 	 */
 	spin_lock_irqsave(&dev->se_tmr_lock, flags);
+	list_del_init(&tmr->tmr_list);
 	list_for_each_entry_safe(tmr_p, tmr_pp, &dev->dev_tmr_list, tmr_list) {
-		/*
-		 * Allow the received TMR to return with FUNCTION_COMPLETE.
-		 */
-		if (tmr_p == tmr)
-			continue;
-
 		cmd = tmr_p->task_cmd;
 		if (!cmd) {
 			pr_err("Unable to locate struct se_cmd for TMR\n");
diff --git a/drivers/target/target_core_tpg.c b/drivers/target/target_core_tpg.c
index 4a8b180c478b..c0dbfa016575 100644
--- a/drivers/target/target_core_tpg.c
+++ b/drivers/target/target_core_tpg.c
@@ -445,7 +445,7 @@ static void core_tpg_lun_ref_release(struct percpu_ref *ref)
 {
 	struct se_lun *lun = container_of(ref, struct se_lun, lun_ref);
 
-	complete(&lun->lun_ref_comp);
+	complete(&lun->lun_shutdown_comp);
 }
 
 /* Does not change se_wwn->priv. */
@@ -572,6 +572,7 @@ struct se_lun *core_tpg_alloc_lun(
 	lun->lun_link_magic = SE_LUN_LINK_MAGIC;
 	atomic_set(&lun->lun_acl_count, 0);
 	init_completion(&lun->lun_ref_comp);
+	init_completion(&lun->lun_shutdown_comp);
 	INIT_LIST_HEAD(&lun->lun_deve_list);
 	INIT_LIST_HEAD(&lun->lun_dev_link);
 	atomic_set(&lun->lun_tg_pt_secondary_offline, 0);
diff --git a/drivers/target/target_core_transport.c b/drivers/target/target_core_transport.c
index 437591bc7c08..434d9d693989 100644
--- a/drivers/target/target_core_transport.c
+++ b/drivers/target/target_core_transport.c
@@ -593,9 +593,6 @@ static void target_remove_from_state_list(struct se_cmd *cmd)
 	if (!dev)
 		return;
 
-	if (cmd->transport_state & CMD_T_BUSY)
-		return;
-
 	spin_lock_irqsave(&dev->execute_task_lock, flags);
 	if (cmd->state_active) {
 		list_del(&cmd->state_list);
@@ -604,24 +601,18 @@ static void target_remove_from_state_list(struct se_cmd *cmd)
 	spin_unlock_irqrestore(&dev->execute_task_lock, flags);
 }
 
-static int transport_cmd_check_stop(struct se_cmd *cmd, bool remove_from_lists,
-				    bool write_pending)
+static int transport_cmd_check_stop_to_fabric(struct se_cmd *cmd)
 {
 	unsigned long flags;
 
-	if (remove_from_lists) {
-		target_remove_from_state_list(cmd);
+	target_remove_from_state_list(cmd);
 
-		/*
-		 * Clear struct se_cmd->se_lun before the handoff to FE.
-		 */
-		cmd->se_lun = NULL;
-	}
+	/*
+	 * Clear struct se_cmd->se_lun before the handoff to FE.
+	 */
+	cmd->se_lun = NULL;
 
 	spin_lock_irqsave(&cmd->t_state_lock, flags);
-	if (write_pending)
-		cmd->t_state = TRANSPORT_WRITE_PENDING;
-
 	/*
 	 * Determine if frontend context caller is requesting the stopping of
 	 * this command for frontend exceptions.
@@ -635,31 +626,18 @@ static int transport_cmd_check_stop(struct se_cmd *cmd, bool remove_from_lists,
 		complete_all(&cmd->t_transport_stop_comp);
 		return 1;
 	}
-
 	cmd->transport_state &= ~CMD_T_ACTIVE;
-	if (remove_from_lists) {
-		/*
-		 * Some fabric modules like tcm_loop can release
-		 * their internally allocated I/O reference now and
-		 * struct se_cmd now.
-		 *
-		 * Fabric modules are expected to return '1' here if the
-		 * se_cmd being passed is released at this point,
-		 * or zero if not being released.
-		 */
-		if (cmd->se_tfo->check_stop_free != NULL) {
-			spin_unlock_irqrestore(&cmd->t_state_lock, flags);
-			return cmd->se_tfo->check_stop_free(cmd);
-		}
-	}
-
 	spin_unlock_irqrestore(&cmd->t_state_lock, flags);
-	return 0;
-}
 
-static int transport_cmd_check_stop_to_fabric(struct se_cmd *cmd)
-{
-	return transport_cmd_check_stop(cmd, true, false);
+	/*
+	 * Some fabric modules like tcm_loop can release their internally
+	 * allocated I/O reference and struct se_cmd now.
+	 *
+	 * Fabric modules are expected to return '1' here if the se_cmd being
+	 * passed is released at this point, or zero if not being released.
+	 */
+	return cmd->se_tfo->check_stop_free ? cmd->se_tfo->check_stop_free(cmd)
+		: 0;
 }
 
 static void transport_lun_remove_cmd(struct se_cmd *cmd)
@@ -733,7 +711,6 @@ void target_complete_cmd(struct se_cmd *cmd, u8 scsi_status)
 
 
 	spin_lock_irqsave(&cmd->t_state_lock, flags);
-	cmd->transport_state &= ~CMD_T_BUSY;
 
 	if (dev && dev->transport->transport_complete) {
 		dev->transport->transport_complete(cmd,
@@ -1246,7 +1223,6 @@ void transport_init_se_cmd(
 	init_completion(&cmd->cmd_wait_comp);
 	spin_lock_init(&cmd->t_state_lock);
 	kref_init(&cmd->cmd_kref);
-	cmd->transport_state = CMD_T_DEV_ACTIVE;
 
 	cmd->se_tfo = tfo;
 	cmd->se_sess = se_sess;
@@ -1671,6 +1647,9 @@ void transport_generic_request_failure(struct se_cmd *cmd,
 {
 	int ret = 0, post_ret = 0;
 
+	if (transport_check_aborted_status(cmd, 1))
+		return;
+
 	pr_debug("-----[ Storage Engine Exception for cmd: %p ITT: 0x%08llx"
 		" CDB: 0x%02x\n", cmd, cmd->tag, cmd->t_task_cdb[0]);
 	pr_debug("-----[ i_state: %d t_state: %d sense_reason: %d\n",
@@ -1801,7 +1780,7 @@ void __target_execute_cmd(struct se_cmd *cmd, bool do_checks)
 		return;
 err:
 	spin_lock_irq(&cmd->t_state_lock);
-	cmd->transport_state &= ~(CMD_T_BUSY|CMD_T_SENT);
+	cmd->transport_state &= ~CMD_T_SENT;
 	spin_unlock_irq(&cmd->t_state_lock);
 
 	transport_generic_request_failure(cmd, ret);
@@ -1829,7 +1808,7 @@ static int target_write_prot_action(struct se_cmd *cmd)
 					     sectors, 0, cmd->t_prot_sg, 0);
 		if (unlikely(cmd->pi_err)) {
 			spin_lock_irq(&cmd->t_state_lock);
-			cmd->transport_state &= ~(CMD_T_BUSY|CMD_T_SENT);
+			cmd->transport_state &= ~CMD_T_SENT;
 			spin_unlock_irq(&cmd->t_state_lock);
 			transport_generic_request_failure(cmd, cmd->pi_err);
 			return -1;
@@ -1918,7 +1897,7 @@ void target_execute_cmd(struct se_cmd *cmd)
 	}
 
 	cmd->t_state = TRANSPORT_PROCESSING;
-	cmd->transport_state |= CMD_T_ACTIVE|CMD_T_BUSY|CMD_T_SENT;
+	cmd->transport_state |= CMD_T_ACTIVE | CMD_T_SENT;
 	spin_unlock_irq(&cmd->t_state_lock);
 
 	if (target_write_prot_action(cmd))
@@ -1926,7 +1905,7 @@ void target_execute_cmd(struct se_cmd *cmd)
 
 	if (target_handle_task_attr(cmd)) {
 		spin_lock_irq(&cmd->t_state_lock);
-		cmd->transport_state &= ~(CMD_T_BUSY | CMD_T_SENT);
+		cmd->transport_state &= ~CMD_T_SENT;
 		spin_unlock_irq(&cmd->t_state_lock);
 		return;
 	}
@@ -1979,8 +1958,6 @@ static void transport_complete_task_attr(struct se_cmd *cmd)
 	if (cmd->sam_task_attr == TCM_SIMPLE_TAG) {
 		atomic_dec_mb(&dev->simple_cmds);
 		dev->dev_cur_ordered_id++;
-		pr_debug("Incremented dev->dev_cur_ordered_id: %u for SIMPLE\n",
-			 dev->dev_cur_ordered_id);
 	} else if (cmd->sam_task_attr == TCM_HEAD_TAG) {
 		dev->dev_cur_ordered_id++;
 		pr_debug("Incremented dev_cur_ordered_id: %u for HEAD_OF_QUEUE\n",
@@ -2387,6 +2364,7 @@ EXPORT_SYMBOL(target_alloc_sgl);
 sense_reason_t
 transport_generic_new_cmd(struct se_cmd *cmd)
 {
+	unsigned long flags;
 	int ret = 0;
 	bool zero_flag = !(cmd->se_cmd_flags & SCF_SCSI_DATA_CDB);
 
@@ -2452,7 +2430,24 @@ transport_generic_new_cmd(struct se_cmd *cmd)
 		target_execute_cmd(cmd);
 		return 0;
 	}
-	transport_cmd_check_stop(cmd, false, true);
+
+	spin_lock_irqsave(&cmd->t_state_lock, flags);
+	cmd->t_state = TRANSPORT_WRITE_PENDING;
+	/*
+	 * Determine if frontend context caller is requesting the stopping of
+	 * this command for frontend exceptions.
+	 */
+	if (cmd->transport_state & CMD_T_STOP) {
+		pr_debug("%s:%d CMD_T_STOP for ITT: 0x%08llx\n",
+			 __func__, __LINE__, cmd->tag);
+
+		spin_unlock_irqrestore(&cmd->t_state_lock, flags);
+
+		complete_all(&cmd->t_transport_stop_comp);
+		return 0;
+	}
+	cmd->transport_state &= ~CMD_T_ACTIVE;
+	spin_unlock_irqrestore(&cmd->t_state_lock, flags);
 
 	ret = cmd->se_tfo->write_pending(cmd);
 	if (ret == -EAGAIN || ret == -ENOMEM)
@@ -2595,39 +2590,38 @@ static void target_release_cmd_kref(struct kref *kref)
 	unsigned long flags;
 	bool fabric_stop;
 
-	spin_lock_irqsave(&se_sess->sess_cmd_lock, flags);
+	if (se_sess) {
+		spin_lock_irqsave(&se_sess->sess_cmd_lock, flags);
 
-	spin_lock(&se_cmd->t_state_lock);
-	fabric_stop = (se_cmd->transport_state & CMD_T_FABRIC_STOP) &&
-		      (se_cmd->transport_state & CMD_T_ABORTED);
-	spin_unlock(&se_cmd->t_state_lock);
+		spin_lock(&se_cmd->t_state_lock);
+		fabric_stop = (se_cmd->transport_state & CMD_T_FABRIC_STOP) &&
+			      (se_cmd->transport_state & CMD_T_ABORTED);
+		spin_unlock(&se_cmd->t_state_lock);
 
-	if (se_cmd->cmd_wait_set || fabric_stop) {
+		if (se_cmd->cmd_wait_set || fabric_stop) {
+			list_del_init(&se_cmd->se_cmd_list);
+			spin_unlock_irqrestore(&se_sess->sess_cmd_lock, flags);
+			target_free_cmd_mem(se_cmd);
+			complete(&se_cmd->cmd_wait_comp);
+			return;
+		}
 		list_del_init(&se_cmd->se_cmd_list);
 		spin_unlock_irqrestore(&se_sess->sess_cmd_lock, flags);
-		target_free_cmd_mem(se_cmd);
-		complete(&se_cmd->cmd_wait_comp);
-		return;
 	}
-	list_del_init(&se_cmd->se_cmd_list);
-	spin_unlock_irqrestore(&se_sess->sess_cmd_lock, flags);
 
 	target_free_cmd_mem(se_cmd);
 	se_cmd->se_tfo->release_cmd(se_cmd);
 }
 
-/* target_put_sess_cmd - Check for active I/O shutdown via kref_put
- * @se_cmd:	command descriptor to drop
+/**
+ * target_put_sess_cmd - decrease the command reference count
+ * @se_cmd:	command to drop a reference from
+ *
+ * Returns 1 if and only if this target_put_sess_cmd() call caused the
+ * refcount to drop to zero. Returns zero otherwise.
  */
 int target_put_sess_cmd(struct se_cmd *se_cmd)
 {
-	struct se_session *se_sess = se_cmd->se_sess;
-
-	if (!se_sess) {
-		target_free_cmd_mem(se_cmd);
-		se_cmd->se_tfo->release_cmd(se_cmd);
-		return 1;
-	}
 	return kref_put(&se_cmd->cmd_kref, target_release_cmd_kref);
 }
 EXPORT_SYMBOL(target_put_sess_cmd);
@@ -2706,10 +2700,39 @@ void target_wait_for_sess_cmds(struct se_session *se_sess)
 }
 EXPORT_SYMBOL(target_wait_for_sess_cmds);
 
+static void target_lun_confirm(struct percpu_ref *ref)
+{
+	struct se_lun *lun = container_of(ref, struct se_lun, lun_ref);
+
+	complete(&lun->lun_ref_comp);
+}
+
 void transport_clear_lun_ref(struct se_lun *lun)
 {
-	percpu_ref_kill(&lun->lun_ref);
+	/*
+	 * Mark the percpu-ref as DEAD, switch to atomic_t mode, drop
+	 * the initial reference and schedule confirm kill to be
+	 * executed after one full RCU grace period has completed.
+	 */
+	percpu_ref_kill_and_confirm(&lun->lun_ref, target_lun_confirm);
+	/*
+	 * The first completion waits for percpu_ref_switch_to_atomic_rcu()
+	 * to call target_lun_confirm after lun->lun_ref has been marked
+	 * as __PERCPU_REF_DEAD on all CPUs, and switches to atomic_t
+	 * mode so that percpu_ref_tryget_live() lookup of lun->lun_ref
+	 * fails for all new incoming I/O.
+	 */
 	wait_for_completion(&lun->lun_ref_comp);
+	/*
+	 * The second completion waits for percpu_ref_put_many() to
+	 * invoke ->release() after lun->lun_ref has switched to
+	 * atomic_t mode, and lun->lun_ref.count has reached zero.
+	 *
+	 * At this point all target-core lun->lun_ref references have
+	 * been dropped via transport_lun_remove_cmd(), and it's safe
+	 * to proceed with the remaining LUN shutdown.
+	 */
+	wait_for_completion(&lun->lun_shutdown_comp);
 }
 
 static bool
@@ -2765,11 +2788,8 @@ __transport_wait_for_tasks(struct se_cmd *cmd, bool fabric_stop,
 }
 
 /**
- * transport_wait_for_tasks - wait for completion to occur
- * @cmd:	command to wait
- *
- * Called from frontend fabric context to wait for storage engine
- * to pause and/or release frontend generated struct se_cmd.
+ * transport_wait_for_tasks - set CMD_T_STOP and wait for t_transport_stop_comp
+ * @cmd: command to wait on
  */
 bool transport_wait_for_tasks(struct se_cmd *cmd)
 {
diff --git a/drivers/target/tcm_fc/tfc_cmd.c b/drivers/target/tcm_fc/tfc_cmd.c
index 9af7842b8178..ec372860106f 100644
--- a/drivers/target/tcm_fc/tfc_cmd.c
+++ b/drivers/target/tcm_fc/tfc_cmd.c
@@ -83,14 +83,12 @@ void ft_dump_cmd(struct ft_cmd *cmd, const char *caller)
 static void ft_free_cmd(struct ft_cmd *cmd)
 {
 	struct fc_frame *fp;
-	struct fc_lport *lport;
 	struct ft_sess *sess;
 
 	if (!cmd)
 		return;
 	sess = cmd->sess;
 	fp = cmd->req_frame;
-	lport = fr_dev(fp);
 	if (fr_seq(fp))
 		fc_seq_release(fr_seq(fp));
 	fc_frame_free(fp);