Merge patch series "qla2xxx driver misc features"

Nilesh Javali <njavali@marvell.com> says:

Martin,

Please apply the qla2xxx driver miscellaneous features and bug fixes
to the scsi tree at your earliest convenience.

Link: https://lore.kernel.org/r/20230821130045.34850-1-njavali@marvell.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

16 files changed, 1085 insertions, 93 deletions

diff --git a/drivers/scsi/qla2xxx/qla_attr.c b/drivers/scsi/qla2xxx/qla_attr.c
index b00222459607..44449c70a375 100644
--- a/drivers/scsi/qla2xxx/qla_attr.c
+++ b/drivers/scsi/qla2xxx/qla_attr.c
@@ -3093,8 +3093,6 @@ qla24xx_vport_create(struct fc_vport *fc_vport, bool disable)
 			vha->flags.difdix_supported = 1;
 			ql_dbg(ql_dbg_user, vha, 0x7082,
 			    "Registered for DIF/DIX type 1 and 3 protection.\n");
-			if (ql2xenabledif == 1)
-				prot = SHOST_DIX_TYPE0_PROTECTION;
 			scsi_host_set_prot(vha->host,
 			    prot | SHOST_DIF_TYPE1_PROTECTION
 			    | SHOST_DIF_TYPE2_PROTECTION
diff --git a/drivers/scsi/qla2xxx/qla_dbg.c b/drivers/scsi/qla2xxx/qla_dbg.c
index d7e8454304ce..691ef827a5ab 100644
--- a/drivers/scsi/qla2xxx/qla_dbg.c
+++ b/drivers/scsi/qla2xxx/qla_dbg.c
@@ -12,13 +12,12 @@
  * ----------------------------------------------------------------------
  * | Module Init and Probe        |       0x0199       |                |
  * | Mailbox commands             |       0x1206       | 0x11a5-0x11ff	|
- * | Device Discovery             |       0x2134       | 0x210e-0x2115  |
- * |                              |                    | 0x211c-0x2128  |
- * |                              |                    | 0x212c-0x2134  |
+ * | Device Discovery             |       0x2134       | 0x2112-0x2115  |
+ * |                              |                    | 0x2127-0x2128  |
  * | Queue Command and IO tracing |       0x3074       | 0x300b         |
  * |                              |                    | 0x3027-0x3028  |
  * |                              |                    | 0x303d-0x3041  |
- * |                              |                    | 0x302d,0x3033  |
+ * |                              |                    | 0x302e,0x3033  |
  * |                              |                    | 0x3036,0x3038  |
  * |                              |                    | 0x303a		|
  * | DPC Thread                   |       0x4023       | 0x4002,0x4013  |
diff --git a/drivers/scsi/qla2xxx/qla_dbg.h b/drivers/scsi/qla2xxx/qla_dbg.h
index 70482b55d240..54f0a412226f 100644
--- a/drivers/scsi/qla2xxx/qla_dbg.h
+++ b/drivers/scsi/qla2xxx/qla_dbg.h
@@ -368,6 +368,7 @@ ql_log_qp(uint32_t, struct qla_qpair *, int32_t, const char *fmt, ...);
 #define ql_dbg_tgt_tmr	0x00001000 /* Target mode task management */
 #define ql_dbg_tgt_dif  0x00000800 /* Target mode dif */
 #define ql_dbg_edif	0x00000400 /* edif and purex debug */
+#define ql_dbg_unsol	0x00000100 /* Unsolicited path debug */
 
 extern int qla27xx_dump_mpi_ram(struct qla_hw_data *, uint32_t, uint32_t *,
 	uint32_t, void **);
diff --git a/drivers/scsi/qla2xxx/qla_def.h b/drivers/scsi/qla2xxx/qla_def.h
index b3b1df34afd3..9ec39bcd41b5 100644
--- a/drivers/scsi/qla2xxx/qla_def.h
+++ b/drivers/scsi/qla2xxx/qla_def.h
@@ -346,6 +346,12 @@ struct name_list_extended {
 	u8			sent;
 };
 
+struct qla_nvme_fc_rjt {
+	struct fcnvme_ls_rjt *c;
+	dma_addr_t  cdma;
+	u16 size;
+};
+
 struct els_reject {
 	struct fc_els_ls_rjt *c;
 	dma_addr_t  cdma;
@@ -503,6 +509,20 @@ struct ct_arg {
 	port_id_t	id;
 };
 
+struct qla_nvme_lsrjt_pt_arg {
+	struct fc_port *fcport;
+	u8 opcode;
+	u8 vp_idx;
+	u8 reason;
+	u8 explanation;
+	__le16 nport_handle;
+	u16 control_flags;
+	__le16 ox_id;
+	__le32 xchg_address;
+	u32 tx_byte_count, rx_byte_count;
+	dma_addr_t tx_addr, rx_addr;
+};
+
 /*
  * SRB extensions.
  */
@@ -611,13 +631,16 @@ struct srb_iocb {
 			void *desc;
 
 			/* These are only used with ls4 requests */
-			int cmd_len;
-			int rsp_len;
+			__le32 cmd_len;
+			__le32 rsp_len;
 			dma_addr_t cmd_dma;
 			dma_addr_t rsp_dma;
 			enum nvmefc_fcp_datadir dir;
 			uint32_t dl;
 			uint32_t timeout_sec;
+			__le32 exchange_address;
+			__le16 nport_handle;
+			__le16 ox_id;
 			struct	list_head   entry;
 		} nvme;
 		struct {
@@ -707,6 +730,10 @@ typedef struct srb {
 	struct fc_port *fcport;
 	struct scsi_qla_host *vha;
 	unsigned int start_timer:1;
+	unsigned int abort:1;
+	unsigned int aborted:1;
+	unsigned int completed:1;
+	unsigned int unsol_rsp:1;
 
 	uint32_t handle;
 	uint16_t flags;
@@ -2542,6 +2569,7 @@ enum rscn_addr_format {
 typedef struct fc_port {
 	struct list_head list;
 	struct scsi_qla_host *vha;
+	struct list_head unsol_ctx_head;
 
 	unsigned int conf_compl_supported:1;
 	unsigned int deleted:2;
@@ -3742,6 +3770,16 @@ struct qla_fw_resources {
 	u16 pad;
 };
 
+struct qla_fw_res {
+	u16      iocb_total;
+	u16      iocb_limit;
+	atomic_t iocb_used;
+
+	u16      exch_total;
+	u16      exch_limit;
+	atomic_t exch_used;
+};
+
 #define QLA_IOCB_PCT_LIMIT 95
 
 struct  qla_buf_pool {
@@ -3787,6 +3825,12 @@ struct qla_qpair {
 
 	uint16_t id;			/* qp number used with FW */
 	uint16_t vp_idx;		/* vport ID */
+
+	uint16_t dsd_inuse;
+	uint16_t dsd_avail;
+	struct list_head dsd_list;
+#define NUM_DSD_CHAIN 4096
+
 	mempool_t *srb_mempool;
 
 	struct pci_dev  *pdev;
@@ -4384,7 +4428,6 @@ struct qla_hw_data {
 	uint8_t		aen_mbx_count;
 	atomic_t	num_pend_mbx_stage1;
 	atomic_t	num_pend_mbx_stage2;
-	atomic_t	num_pend_mbx_stage3;
 	uint16_t	frame_payload_size;
 
 	uint32_t	login_retry_count;
@@ -4715,11 +4758,6 @@ struct qla_hw_data {
 	struct fw_blob	*hablob;
 	struct qla82xx_legacy_intr_set nx_legacy_intr;
 
-	uint16_t	gbl_dsd_inuse;
-	uint16_t	gbl_dsd_avail;
-	struct list_head gbl_dsd_list;
-#define NUM_DSD_CHAIN 4096
-
 	uint8_t fw_type;
 	uint32_t file_prd_off;	/* File firmware product offset */
 
@@ -4801,6 +4839,8 @@ struct qla_hw_data {
 	struct els_reject elsrej;
 	u8 edif_post_stop_cnt_down;
 	struct qla_vp_map *vp_map;
+	struct qla_nvme_fc_rjt lsrjt;
+	struct qla_fw_res fwres ____cacheline_aligned;
 };
 
 #define RX_ELS_SIZE (roundup(sizeof(struct enode) + ELS_MAX_PAYLOAD, SMP_CACHE_BYTES))
@@ -4833,6 +4873,7 @@ struct active_regions {
  * is variable) starting at "iocb".
  */
 struct purex_item {
+	void *purls_context;
 	struct list_head list;
 	struct scsi_qla_host *vha;
 	void (*process_item)(struct scsi_qla_host *vha,
diff --git a/drivers/scsi/qla2xxx/qla_dfs.c b/drivers/scsi/qla2xxx/qla_dfs.c
index 1925cc6897b6..f060e593685d 100644
--- a/drivers/scsi/qla2xxx/qla_dfs.c
+++ b/drivers/scsi/qla2xxx/qla_dfs.c
@@ -276,6 +276,16 @@ qla_dfs_fw_resource_cnt_show(struct seq_file *s, void *unused)
 
 		seq_printf(s, "estimate exchange used[%d] high water limit [%d] n",
 			   exch_used, ha->base_qpair->fwres.exch_limit);
+
+		if (ql2xenforce_iocb_limit == 2) {
+			iocbs_used = atomic_read(&ha->fwres.iocb_used);
+			exch_used  = atomic_read(&ha->fwres.exch_used);
+			seq_printf(s, "        estimate iocb2 used [%d] high water limit [%d]\n",
+					iocbs_used, ha->fwres.iocb_limit);
+
+			seq_printf(s, "        estimate exchange2 used[%d] high water limit [%d] \n",
+					exch_used, ha->fwres.exch_limit);
+		}
 	}
 
 	return 0;
diff --git a/drivers/scsi/qla2xxx/qla_gbl.h b/drivers/scsi/qla2xxx/qla_gbl.h
index 816c0b9ecd0e..09cb9413670a 100644
--- a/drivers/scsi/qla2xxx/qla_gbl.h
+++ b/drivers/scsi/qla2xxx/qla_gbl.h
@@ -603,7 +603,11 @@ qla2xxx_msix_rsp_q_hs(int irq, void *dev_id);
 fc_port_t *qla2x00_find_fcport_by_loopid(scsi_qla_host_t *, uint16_t);
 fc_port_t *qla2x00_find_fcport_by_wwpn(scsi_qla_host_t *, u8 *, u8);
 fc_port_t *qla2x00_find_fcport_by_nportid(scsi_qla_host_t *, port_id_t *, u8);
-void __qla_consume_iocb(struct scsi_qla_host *vha, void **pkt, struct rsp_que **rsp);
+void qla24xx_queue_purex_item(scsi_qla_host_t *, struct purex_item *,
+			      void (*process_item)(struct scsi_qla_host *,
+			      struct purex_item *));
+void __qla_consume_iocb(struct scsi_qla_host *, void **, struct rsp_que **);
+void qla2xxx_process_purls_iocb(void **pkt, struct rsp_que **rsp);
 
 /*
  * Global Function Prototypes in qla_sup.c source file.
@@ -666,9 +670,11 @@ extern int qla2xxx_get_vpd_field(scsi_qla_host_t *, char *, char *, size_t);
 extern void qla2xxx_flash_npiv_conf(scsi_qla_host_t *);
 extern int qla24xx_read_fcp_prio_cfg(scsi_qla_host_t *);
 extern int qla2x00_mailbox_passthru(struct bsg_job *bsg_job);
-int __qla_copy_purex_to_buffer(struct scsi_qla_host *vha, void **pkt,
-	struct rsp_que **rsp, u8 *buf, u32 buf_len);
-
+int qla2x00_sys_ld_info(struct bsg_job *bsg_job);
+int __qla_copy_purex_to_buffer(struct scsi_qla_host *, void **,
+	struct rsp_que **, u8 *, u32);
+struct purex_item *qla27xx_copy_multiple_pkt(struct scsi_qla_host *vha,
+	void **pkt, struct rsp_que **rsp, bool is_purls, bool byte_order);
 int qla_mailbox_passthru(scsi_qla_host_t *vha, uint16_t *mbx_in,
 			 uint16_t *mbx_out);
 
diff --git a/drivers/scsi/qla2xxx/qla_init.c b/drivers/scsi/qla2xxx/qla_init.c
index 059175f2c8f5..82077edfda1f 100644
--- a/drivers/scsi/qla2xxx/qla_init.c
+++ b/drivers/scsi/qla2xxx/qla_init.c
@@ -2223,6 +2223,8 @@ __qla2x00_async_tm_cmd(struct tmf_arg *arg)
 			rval = QLA_FUNCTION_FAILED;
 		}
 	}
+	if (tm_iocb->u.tmf.data)
+		rval = tm_iocb->u.tmf.data;
 
 done_free_sp:
 	/* ref: INIT */
@@ -4203,7 +4205,7 @@ void qla_init_iocb_limit(scsi_qla_host_t *vha)
 	u8 i;
 	struct qla_hw_data *ha = vha->hw;
 
-	 __qla_adjust_iocb_limit(ha->base_qpair);
+	__qla_adjust_iocb_limit(ha->base_qpair);
 	ha->base_qpair->fwres.iocbs_used = 0;
 	ha->base_qpair->fwres.exch_used  = 0;
 
@@ -4214,6 +4216,14 @@ void qla_init_iocb_limit(scsi_qla_host_t *vha)
 			ha->queue_pair_map[i]->fwres.exch_used = 0;
 		}
 	}
+
+	ha->fwres.iocb_total = ha->orig_fw_iocb_count;
+	ha->fwres.iocb_limit = (ha->orig_fw_iocb_count * QLA_IOCB_PCT_LIMIT) / 100;
+	ha->fwres.exch_total = ha->orig_fw_xcb_count;
+	ha->fwres.exch_limit = (ha->orig_fw_xcb_count * QLA_IOCB_PCT_LIMIT) / 100;
+
+	atomic_set(&ha->fwres.iocb_used, 0);
+	atomic_set(&ha->fwres.exch_used, 0);
 }
 
 void qla_adjust_iocb_limit(scsi_qla_host_t *vha)
@@ -5554,6 +5564,7 @@ qla2x00_alloc_fcport(scsi_qla_host_t *vha, gfp_t flags)
 	INIT_WORK(&fcport->reg_work, qla_register_fcport_fn);
 	INIT_LIST_HEAD(&fcport->gnl_entry);
 	INIT_LIST_HEAD(&fcport->list);
+	INIT_LIST_HEAD(&fcport->unsol_ctx_head);
 
 	INIT_LIST_HEAD(&fcport->sess_cmd_list);
 	spin_lock_init(&fcport->sess_cmd_lock);
@@ -5596,7 +5607,7 @@ static void qla_get_login_template(scsi_qla_host_t *vha)
 	__be32 *q;
 
 	memset(ha->init_cb, 0, ha->init_cb_size);
-	sz = min_t(int, sizeof(struct fc_els_csp), ha->init_cb_size);
+	sz = min_t(int, sizeof(struct fc_els_flogi), ha->init_cb_size);
 	rval = qla24xx_get_port_login_templ(vha, ha->init_cb_dma,
 					    ha->init_cb, sz);
 	if (rval != QLA_SUCCESS) {
@@ -7390,14 +7401,15 @@ qla2x00_abort_isp_cleanup(scsi_qla_host_t *vha)
 	}
 
 	/* purge MBox commands */
-	if (atomic_read(&ha->num_pend_mbx_stage3)) {
+	spin_lock_irqsave(&ha->hardware_lock, flags);
+	if (test_bit(MBX_INTR_WAIT, &ha->mbx_cmd_flags)) {
 		clear_bit(MBX_INTR_WAIT, &ha->mbx_cmd_flags);
 		complete(&ha->mbx_intr_comp);
 	}
+	spin_unlock_irqrestore(&ha->hardware_lock, flags);
 
 	i = 0;
-	while (atomic_read(&ha->num_pend_mbx_stage3) ||
-	    atomic_read(&ha->num_pend_mbx_stage2) ||
+	while (atomic_read(&ha->num_pend_mbx_stage2) ||
 	    atomic_read(&ha->num_pend_mbx_stage1)) {
 		msleep(20);
 		i++;
@@ -9643,6 +9655,7 @@ struct qla_qpair *qla2xxx_create_qpair(struct scsi_qla_host *vha, int qos,
 		qpair->vp_idx = vp_idx;
 		qpair->fw_started = ha->flags.fw_started;
 		INIT_LIST_HEAD(&qpair->hints_list);
+		INIT_LIST_HEAD(&qpair->dsd_list);
 		qpair->chip_reset = ha->base_qpair->chip_reset;
 		qpair->enable_class_2 = ha->base_qpair->enable_class_2;
 		qpair->enable_explicit_conf =
@@ -9771,6 +9784,19 @@ int qla2xxx_delete_qpair(struct scsi_qla_host *vha, struct qla_qpair *qpair)
 	if (ret != QLA_SUCCESS)
 		goto fail;
 
+	if (!list_empty(&qpair->dsd_list)) {
+		struct dsd_dma *dsd_ptr, *tdsd_ptr;
+
+		/* clean up allocated prev pool */
+		list_for_each_entry_safe(dsd_ptr, tdsd_ptr,
+					 &qpair->dsd_list, list) {
+			dma_pool_free(ha->dl_dma_pool, dsd_ptr->dsd_addr,
+				      dsd_ptr->dsd_list_dma);
+			list_del(&dsd_ptr->list);
+			kfree(dsd_ptr);
+		}
+	}
+
 	mutex_lock(&ha->mq_lock);
 	ha->queue_pair_map[qpair->id] = NULL;
 	clear_bit(qpair->id, ha->qpair_qid_map);
diff --git a/drivers/scsi/qla2xxx/qla_inline.h b/drivers/scsi/qla2xxx/qla_inline.h
index 0167e85ba058..0556969f6dc1 100644
--- a/drivers/scsi/qla2xxx/qla_inline.h
+++ b/drivers/scsi/qla2xxx/qla_inline.h
@@ -386,6 +386,7 @@ enum {
 	RESOURCE_IOCB = BIT_0,
 	RESOURCE_EXCH = BIT_1,  /* exchange */
 	RESOURCE_FORCE = BIT_2,
+	RESOURCE_HA = BIT_3,
 };
 
 static inline int
@@ -393,7 +394,7 @@ qla_get_fw_resources(struct qla_qpair *qp, struct iocb_resource *iores)
 {
 	u16 iocbs_used, i;
 	u16 exch_used;
-	struct qla_hw_data *ha = qp->vha->hw;
+	struct qla_hw_data *ha = qp->hw;
 
 	if (!ql2xenforce_iocb_limit) {
 		iores->res_type = RESOURCE_NONE;
@@ -428,15 +429,69 @@ qla_get_fw_resources(struct qla_qpair *qp, struct iocb_resource *iores)
 			return -ENOSPC;
 		}
 	}
+
+	if (ql2xenforce_iocb_limit == 2) {
+		if ((iores->iocb_cnt + atomic_read(&ha->fwres.iocb_used)) >=
+		    ha->fwres.iocb_limit) {
+			iores->res_type = RESOURCE_NONE;
+			return -ENOSPC;
+		}
+
+		if (iores->res_type & RESOURCE_EXCH) {
+			if ((iores->exch_cnt + atomic_read(&ha->fwres.exch_used)) >=
+			    ha->fwres.exch_limit) {
+				iores->res_type = RESOURCE_NONE;
+				return -ENOSPC;
+			}
+		}
+	}
+
 force:
 	qp->fwres.iocbs_used += iores->iocb_cnt;
 	qp->fwres.exch_used += iores->exch_cnt;
+	if (ql2xenforce_iocb_limit == 2) {
+		atomic_add(iores->iocb_cnt, &ha->fwres.iocb_used);
+		atomic_add(iores->exch_cnt, &ha->fwres.exch_used);
+		iores->res_type |= RESOURCE_HA;
+	}
 	return 0;
 }
 
+/*
+ * decrement to zero.  This routine will not decrement below zero
+ * @v:  pointer of type atomic_t
+ * @amount: amount to decrement from v
+ */
+static void qla_atomic_dtz(atomic_t *v, int amount)
+{
+	int c, old, dec;
+
+	c = atomic_read(v);
+	for (;;) {
+		dec = c - amount;
+		if (unlikely(dec < 0))
+			dec = 0;
+
+		old = atomic_cmpxchg((v), c, dec);
+		if (likely(old == c))
+			break;
+		c = old;
+	}
+}
+
 static inline void
 qla_put_fw_resources(struct qla_qpair *qp, struct iocb_resource *iores)
 {
+	struct qla_hw_data *ha = qp->hw;
+
+	if (iores->res_type & RESOURCE_HA) {
+		if (iores->res_type & RESOURCE_IOCB)
+			qla_atomic_dtz(&ha->fwres.iocb_used, iores->iocb_cnt);
+
+		if (iores->res_type & RESOURCE_EXCH)
+			qla_atomic_dtz(&ha->fwres.exch_used, iores->exch_cnt);
+	}
+
 	if (iores->res_type & RESOURCE_IOCB) {
 		if (qp->fwres.iocbs_used >= iores->iocb_cnt) {
 			qp->fwres.iocbs_used -= iores->iocb_cnt;
diff --git a/drivers/scsi/qla2xxx/qla_iocb.c b/drivers/scsi/qla2xxx/qla_iocb.c
index 1c6e300ed3ab..4450c0d000cc 100644
--- a/drivers/scsi/qla2xxx/qla_iocb.c
+++ b/drivers/scsi/qla2xxx/qla_iocb.c
@@ -11,6 +11,7 @@
 
 #include <scsi/scsi_tcq.h>
 
+static int qla_start_scsi_type6(srb_t *sp);
 /**
  * qla2x00_get_cmd_direction() - Determine control_flag data direction.
  * @sp: SCSI command
@@ -590,8 +591,6 @@ qla24xx_build_scsi_type_6_iocbs(srb_t *sp, struct cmd_type_6 *cmd_pkt,
 	uint16_t tot_dsds)
 {
 	struct dsd64 *cur_dsd = NULL, *next_dsd;
-	scsi_qla_host_t	*vha;
-	struct qla_hw_data *ha;
 	struct scsi_cmnd *cmd;
 	struct	scatterlist *cur_seg;
 	uint8_t avail_dsds;
@@ -613,9 +612,6 @@ qla24xx_build_scsi_type_6_iocbs(srb_t *sp, struct cmd_type_6 *cmd_pkt,
 		return 0;
 	}
 
-	vha = sp->vha;
-	ha = vha->hw;
-
 	/* Set transfer direction */
 	if (cmd->sc_data_direction == DMA_TO_DEVICE) {
 		cmd_pkt->control_flags = cpu_to_le16(CF_WRITE_DATA);
@@ -636,14 +632,13 @@ qla24xx_build_scsi_type_6_iocbs(srb_t *sp, struct cmd_type_6 *cmd_pkt,
 		tot_dsds -= avail_dsds;
 		dsd_list_len = (avail_dsds + 1) * QLA_DSD_SIZE;
 
-		dsd_ptr = list_first_entry(&ha->gbl_dsd_list,
-		    struct dsd_dma, list);
+		dsd_ptr = list_first_entry(&qpair->dsd_list, struct dsd_dma, list);
 		next_dsd = dsd_ptr->dsd_addr;
 		list_del(&dsd_ptr->list);
-		ha->gbl_dsd_avail--;
+		qpair->dsd_avail--;
 		list_add_tail(&dsd_ptr->list, &ctx->dsd_list);
 		ctx->dsd_use_cnt++;
-		ha->gbl_dsd_inuse++;
+		qpair->dsd_inuse++;
 
 		if (first_iocb) {
 			first_iocb = 0;
@@ -1722,6 +1717,8 @@ qla24xx_dif_start_scsi(srb_t *sp)
 	if (scsi_get_prot_op(cmd) == SCSI_PROT_NORMAL) {
 		if (cmd->cmd_len <= 16)
 			return qla24xx_start_scsi(sp);
+		else
+			return qla_start_scsi_type6(sp);
 	}
 
 	/* Setup device pointers. */
@@ -2101,6 +2098,8 @@ qla2xxx_dif_start_scsi_mq(srb_t *sp)
 	if (scsi_get_prot_op(cmd) == SCSI_PROT_NORMAL) {
 		if (cmd->cmd_len <= 16)
 			return qla2xxx_start_scsi_mq(sp);
+		else
+			return qla_start_scsi_type6(sp);
 	}
 
 	spin_lock_irqsave(&qpair->qp_lock, flags);
@@ -3367,6 +3366,7 @@ qla82xx_start_scsi(srb_t *sp)
 	struct qla_hw_data *ha = vha->hw;
 	struct req_que *req = NULL;
 	struct rsp_que *rsp = NULL;
+	struct qla_qpair *qpair = sp->qpair;
 
 	/* Setup device pointers. */
 	reg = &ha->iobase->isp82;
@@ -3415,18 +3415,18 @@ qla82xx_start_scsi(srb_t *sp)
 		uint16_t i;
 
 		more_dsd_lists = qla24xx_calc_dsd_lists(tot_dsds);
-		if ((more_dsd_lists + ha->gbl_dsd_inuse) >= NUM_DSD_CHAIN) {
+		if ((more_dsd_lists + qpair->dsd_inuse) >= NUM_DSD_CHAIN) {
 			ql_dbg(ql_dbg_io, vha, 0x300d,
 			    "Num of DSD list %d is than %d for cmd=%p.\n",
-			    more_dsd_lists + ha->gbl_dsd_inuse, NUM_DSD_CHAIN,
+			    more_dsd_lists + qpair->dsd_inuse, NUM_DSD_CHAIN,
 			    cmd);
 			goto queuing_error;
 		}
 
-		if (more_dsd_lists <= ha->gbl_dsd_avail)
+		if (more_dsd_lists <= qpair->dsd_avail)
 			goto sufficient_dsds;
 		else
-			more_dsd_lists -= ha->gbl_dsd_avail;
+			more_dsd_lists -= qpair->dsd_avail;
 
 		for (i = 0; i < more_dsd_lists; i++) {
 			dsd_ptr = kzalloc(sizeof(struct dsd_dma), GFP_ATOMIC);
@@ -3446,8 +3446,8 @@ qla82xx_start_scsi(srb_t *sp)
 				    "for cmd=%p.\n", cmd);
 				goto queuing_error;
 			}
-			list_add_tail(&dsd_ptr->list, &ha->gbl_dsd_list);
-			ha->gbl_dsd_avail++;
+			list_add_tail(&dsd_ptr->list, &qpair->dsd_list);
+			qpair->dsd_avail++;
 		}
 
 sufficient_dsds:
@@ -3766,21 +3766,28 @@ qla_nvme_ls(srb_t *sp, struct pt_ls4_request *cmd_pkt)
 	nvme = &sp->u.iocb_cmd;
 	cmd_pkt->entry_type = PT_LS4_REQUEST;
 	cmd_pkt->entry_count = 1;
-	cmd_pkt->control_flags = cpu_to_le16(CF_LS4_ORIGINATOR << CF_LS4_SHIFT);
-
 	cmd_pkt->timeout = cpu_to_le16(nvme->u.nvme.timeout_sec);
-	cmd_pkt->nport_handle = cpu_to_le16(sp->fcport->loop_id);
 	cmd_pkt->vp_index = sp->fcport->vha->vp_idx;
 
+	if (sp->unsol_rsp) {
+		cmd_pkt->control_flags =
+				cpu_to_le16(CF_LS4_RESPONDER << CF_LS4_SHIFT);
+		cmd_pkt->nport_handle = nvme->u.nvme.nport_handle;
+		cmd_pkt->exchange_address = nvme->u.nvme.exchange_address;
+	} else {
+		cmd_pkt->control_flags =
+				cpu_to_le16(CF_LS4_ORIGINATOR << CF_LS4_SHIFT);
+		cmd_pkt->nport_handle = cpu_to_le16(sp->fcport->loop_id);
+		cmd_pkt->rx_dseg_count = cpu_to_le16(1);
+		cmd_pkt->rx_byte_count = nvme->u.nvme.rsp_len;
+		cmd_pkt->dsd[1].length  = nvme->u.nvme.rsp_len;
+		put_unaligned_le64(nvme->u.nvme.rsp_dma, &cmd_pkt->dsd[1].address);
+	}
+
 	cmd_pkt->tx_dseg_count = cpu_to_le16(1);
-	cmd_pkt->tx_byte_count = cpu_to_le32(nvme->u.nvme.cmd_len);
-	cmd_pkt->dsd[0].length = cpu_to_le32(nvme->u.nvme.cmd_len);
+	cmd_pkt->tx_byte_count = nvme->u.nvme.cmd_len;
+	cmd_pkt->dsd[0].length = nvme->u.nvme.cmd_len;
 	put_unaligned_le64(nvme->u.nvme.cmd_dma, &cmd_pkt->dsd[0].address);
-
-	cmd_pkt->rx_dseg_count = cpu_to_le16(1);
-	cmd_pkt->rx_byte_count = cpu_to_le32(nvme->u.nvme.rsp_len);
-	cmd_pkt->dsd[1].length = cpu_to_le32(nvme->u.nvme.rsp_len);
-	put_unaligned_le64(nvme->u.nvme.rsp_dma, &cmd_pkt->dsd[1].address);
 }
 
 static void
@@ -4197,3 +4204,267 @@ queuing_error:
 
 	return rval;
 }
+
+/**
+ * qla_start_scsi_type6() - Send a SCSI command to the ISP
+ * @sp: command to send to the ISP
+ *
+ * Returns non-zero if a failure occurred, else zero.
+ */
+static int
+qla_start_scsi_type6(srb_t *sp)
+{
+	int		nseg;
+	unsigned long   flags;
+	uint32_t	*clr_ptr;
+	uint32_t	handle;
+	struct cmd_type_6 *cmd_pkt;
+	uint16_t	cnt;
+	uint16_t	req_cnt;
+	uint16_t	tot_dsds;
+	struct req_que *req = NULL;
+	struct rsp_que *rsp;
+	struct scsi_cmnd *cmd = GET_CMD_SP(sp);
+	struct scsi_qla_host *vha = sp->fcport->vha;
+	struct qla_hw_data *ha = vha->hw;
+	struct qla_qpair *qpair = sp->qpair;
+	uint16_t more_dsd_lists = 0;
+	struct dsd_dma *dsd_ptr;
+	uint16_t i;
+	__be32 *fcp_dl;
+	uint8_t additional_cdb_len;
+	struct ct6_dsd *ctx;
+
+	/* Acquire qpair specific lock */
+	spin_lock_irqsave(&qpair->qp_lock, flags);
+
+	/* Setup qpair pointers */
+	req = qpair->req;
+	rsp = qpair->rsp;
+
+	/* So we know we haven't pci_map'ed anything yet */
+	tot_dsds = 0;
+
+	/* Send marker if required */
+	if (vha->marker_needed != 0) {
+		if (__qla2x00_marker(vha, qpair, 0, 0, MK_SYNC_ALL) != QLA_SUCCESS) {
+			spin_unlock_irqrestore(&qpair->qp_lock, flags);
+			return QLA_FUNCTION_FAILED;
+		}
+		vha->marker_needed = 0;
+	}
+
+	handle = qla2xxx_get_next_handle(req);
+	if (handle == 0)
+		goto queuing_error;
+
+	/* Map the sg table so we have an accurate count of sg entries needed */
+	if (scsi_sg_count(cmd)) {
+		nseg = dma_map_sg(&ha->pdev->dev, scsi_sglist(cmd),
+				  scsi_sg_count(cmd), cmd->sc_data_direction);
+		if (unlikely(!nseg))
+			goto queuing_error;
+	} else {
+		nseg = 0;
+	}
+
+	tot_dsds = nseg;
+
+	/* eventhough driver only need 1 T6 IOCB, FW still convert DSD to Continueation IOCB */
+	req_cnt = qla24xx_calc_iocbs(vha, tot_dsds);
+
+	sp->iores.res_type = RESOURCE_IOCB | RESOURCE_EXCH;
+	sp->iores.exch_cnt = 1;
+	sp->iores.iocb_cnt = req_cnt;
+
+	if (qla_get_fw_resources(sp->qpair, &sp->iores))
+		goto queuing_error;
+
+	more_dsd_lists = qla24xx_calc_dsd_lists(tot_dsds);
+	if ((more_dsd_lists + qpair->dsd_inuse) >= NUM_DSD_CHAIN) {
+		ql_dbg(ql_dbg_io, vha, 0x3028,
+		       "Num of DSD list %d is than %d for cmd=%p.\n",
+		       more_dsd_lists + qpair->dsd_inuse, NUM_DSD_CHAIN, cmd);
+		goto queuing_error;
+	}
+
+	if (more_dsd_lists <= qpair->dsd_avail)
+		goto sufficient_dsds;
+	else
+		more_dsd_lists -= qpair->dsd_avail;
+
+	for (i = 0; i < more_dsd_lists; i++) {
+		dsd_ptr = kzalloc(sizeof(*dsd_ptr), GFP_ATOMIC);
+		if (!dsd_ptr) {
+			ql_log(ql_log_fatal, vha, 0x3029,
+			    "Failed to allocate memory for dsd_dma for cmd=%p.\n", cmd);
+			goto queuing_error;
+		}
+		INIT_LIST_HEAD(&dsd_ptr->list);
+
+		dsd_ptr->dsd_addr = dma_pool_alloc(ha->dl_dma_pool,
+			GFP_ATOMIC, &dsd_ptr->dsd_list_dma);
+		if (!dsd_ptr->dsd_addr) {
+			kfree(dsd_ptr);
+			ql_log(ql_log_fatal, vha, 0x302a,
+			    "Failed to allocate memory for dsd_addr for cmd=%p.\n", cmd);
+			goto queuing_error;
+		}
+		list_add_tail(&dsd_ptr->list, &qpair->dsd_list);
+		qpair->dsd_avail++;
+	}
+
+sufficient_dsds:
+	req_cnt = 1;
+
+	if (req->cnt < (req_cnt + 2)) {
+		if (IS_SHADOW_REG_CAPABLE(ha)) {
+			cnt = *req->out_ptr;
+		} else {
+			cnt = (uint16_t)rd_reg_dword_relaxed(req->req_q_out);
+			if (qla2x00_check_reg16_for_disconnect(vha, cnt))
+				goto queuing_error;
+		}
+
+		if (req->ring_index < cnt)
+			req->cnt = cnt - req->ring_index;
+		else
+			req->cnt = req->length - (req->ring_index - cnt);
+		if (req->cnt < (req_cnt + 2))
+			goto queuing_error;
+	}
+
+	ctx = &sp->u.scmd.ct6_ctx;
+
+	memset(ctx, 0, sizeof(struct ct6_dsd));
+	ctx->fcp_cmnd = dma_pool_zalloc(ha->fcp_cmnd_dma_pool,
+		GFP_ATOMIC, &ctx->fcp_cmnd_dma);
+	if (!ctx->fcp_cmnd) {
+		ql_log(ql_log_fatal, vha, 0x3031,
+		    "Failed to allocate fcp_cmnd for cmd=%p.\n", cmd);
+		goto queuing_error;
+	}
+
+	/* Initialize the DSD list and dma handle */
+	INIT_LIST_HEAD(&ctx->dsd_list);
+	ctx->dsd_use_cnt = 0;
+
+	if (cmd->cmd_len > 16) {
+		additional_cdb_len = cmd->cmd_len - 16;
+		if (cmd->cmd_len % 4 ||
+		    cmd->cmd_len > QLA_CDB_BUF_SIZE) {
+			/*
+			 * SCSI command bigger than 16 bytes must be
+			 * multiple of 4 or too big.
+			 */
+			ql_log(ql_log_warn, vha, 0x3033,
+			    "scsi cmd len %d not multiple of 4 for cmd=%p.\n",
+			    cmd->cmd_len, cmd);
+			goto queuing_error_fcp_cmnd;
+		}
+		ctx->fcp_cmnd_len = 12 + cmd->cmd_len + 4;
+	} else {
+		additional_cdb_len = 0;
+		ctx->fcp_cmnd_len = 12 + 16 + 4;
+	}
+
+	/* Build command packet. */
+	req->current_outstanding_cmd = handle;
+	req->outstanding_cmds[handle] = sp;
+	sp->handle = handle;
+	cmd->host_scribble = (unsigned char *)(unsigned long)handle;
+	req->cnt -= req_cnt;
+
+	cmd_pkt = (struct cmd_type_6 *)req->ring_ptr;
+	cmd_pkt->handle = make_handle(req->id, handle);
+
+	/* tagged queuing modifier -- default is TSK_SIMPLE (0). */
+	clr_ptr = (uint32_t *)cmd_pkt + 2;
+	memset(clr_ptr, 0, REQUEST_ENTRY_SIZE - 8);
+	cmd_pkt->dseg_count = cpu_to_le16(tot_dsds);
+
+	/* Set NPORT-ID and LUN number */
+	cmd_pkt->nport_handle = cpu_to_le16(sp->fcport->loop_id);
+	cmd_pkt->port_id[0] = sp->fcport->d_id.b.al_pa;
+	cmd_pkt->port_id[1] = sp->fcport->d_id.b.area;
+	cmd_pkt->port_id[2] = sp->fcport->d_id.b.domain;
+	cmd_pkt->vp_index = sp->vha->vp_idx;
+
+	/* Build IOCB segments */
+	qla24xx_build_scsi_type_6_iocbs(sp, cmd_pkt, tot_dsds);
+
+	int_to_scsilun(cmd->device->lun, &cmd_pkt->lun);
+	host_to_fcp_swap((uint8_t *)&cmd_pkt->lun, sizeof(cmd_pkt->lun));
+
+	/* build FCP_CMND IU */
+	int_to_scsilun(cmd->device->lun, &ctx->fcp_cmnd->lun);
+	ctx->fcp_cmnd->additional_cdb_len = additional_cdb_len;
+
+	if (cmd->sc_data_direction == DMA_TO_DEVICE)
+		ctx->fcp_cmnd->additional_cdb_len |= 1;
+	else if (cmd->sc_data_direction == DMA_FROM_DEVICE)
+		ctx->fcp_cmnd->additional_cdb_len |= 2;
+
+	/* Populate the FCP_PRIO. */
+	if (ha->flags.fcp_prio_enabled)
+		ctx->fcp_cmnd->task_attribute |=
+		    sp->fcport->fcp_prio << 3;
+
+	memcpy(ctx->fcp_cmnd->cdb, cmd->cmnd, cmd->cmd_len);
+
+	fcp_dl = (__be32 *)(ctx->fcp_cmnd->cdb + 16 +
+	    additional_cdb_len);
+	*fcp_dl = htonl((uint32_t)scsi_bufflen(cmd));
+
+	cmd_pkt->fcp_cmnd_dseg_len = cpu_to_le16(ctx->fcp_cmnd_len);
+	put_unaligned_le64(ctx->fcp_cmnd_dma,
+			   &cmd_pkt->fcp_cmnd_dseg_address);
+
+	sp->flags |= SRB_FCP_CMND_DMA_VALID;
+	cmd_pkt->byte_count = cpu_to_le32((uint32_t)scsi_bufflen(cmd));
+	/* Set total data segment count. */
+	cmd_pkt->entry_count = (uint8_t)req_cnt;
+
+	wmb();
+	/* Adjust ring index. */
+	req->ring_index++;
+	if (req->ring_index == req->length) {
+		req->ring_index = 0;
+		req->ring_ptr = req->ring;
+	} else {
+		req->ring_ptr++;
+	}
+
+	sp->qpair->cmd_cnt++;
+	sp->flags |= SRB_DMA_VALID;
+
+	/* Set chip new ring index. */
+	wrt_reg_dword(req->req_q_in, req->ring_index);
+
+	/* Manage unprocessed RIO/ZIO commands in response queue. */
+	if (vha->flags.process_response_queue &&
+	    rsp->ring_ptr->signature != RESPONSE_PROCESSED)
+		qla24xx_process_response_queue(vha, rsp);
+
+	spin_unlock_irqrestore(&qpair->qp_lock, flags);
+
+	return QLA_SUCCESS;
+
+queuing_error_fcp_cmnd:
+	dma_pool_free(ha->fcp_cmnd_dma_pool, ctx->fcp_cmnd, ctx->fcp_cmnd_dma);
+
+queuing_error:
+	if (tot_dsds)
+		scsi_dma_unmap(cmd);
+
+	qla_put_fw_resources(sp->qpair, &sp->iores);
+
+	if (sp->u.scmd.crc_ctx) {
+		mempool_free(sp->u.scmd.crc_ctx, ha->ctx_mempool);
+		sp->u.scmd.crc_ctx = NULL;
+	}
+
+	spin_unlock_irqrestore(&qpair->qp_lock, flags);
+
+	return QLA_FUNCTION_FAILED;
+}
diff --git a/drivers/scsi/qla2xxx/qla_isr.c b/drivers/scsi/qla2xxx/qla_isr.c
index 1f42a413b598..e98788191897 100644
--- a/drivers/scsi/qla2xxx/qla_isr.c
+++ b/drivers/scsi/qla2xxx/qla_isr.c
@@ -56,6 +56,22 @@ const char *const port_state_str[] = {
 	[FCS_ONLINE]		= "ONLINE"
 };
 
+#define SFP_DISABLE_LASER_INITIATED    0x15  /* Sub code of 8070 AEN */
+#define SFP_ENABLE_LASER_INITIATED     0x16  /* Sub code of 8070 AEN */
+
+static inline void display_Laser_info(scsi_qla_host_t *vha,
+				      u16 mb1, u16 mb2, u16 mb3) {
+
+	if (mb1 == SFP_DISABLE_LASER_INITIATED)
+		ql_log(ql_log_warn, vha, 0xf0a2,
+		       "SFP temperature (%d C) reached/exceeded the threshold (%d C). Laser is disabled.\n",
+		       mb3, mb2);
+	if (mb1 == SFP_ENABLE_LASER_INITIATED)
+		ql_log(ql_log_warn, vha, 0xf0a3,
+		       "SFP temperature (%d C) reached normal operating level. Laser is enabled.\n",
+		       mb3);
+}
+
 static void
 qla24xx_process_abts(struct scsi_qla_host *vha, struct purex_item *pkt)
 {
@@ -823,6 +839,135 @@ qla83xx_handle_8200_aen(scsi_qla_host_t *vha, uint16_t *mb)
 	}
 }
 
+/**
+ * qla27xx_copy_multiple_pkt() - Copy over purex/purls packets that can
+ * span over multiple IOCBs.
+ * @vha: SCSI driver HA context
+ * @pkt: ELS packet
+ * @rsp: Response queue
+ * @is_purls: True, for Unsolicited Received FC-NVMe LS rsp IOCB
+ *            false, for Unsolicited Received ELS IOCB
+ * @byte_order: True, to change the byte ordering of iocb payload
+ */
+struct purex_item *
+qla27xx_copy_multiple_pkt(struct scsi_qla_host *vha, void **pkt,
+			  struct rsp_que **rsp, bool is_purls,
+			  bool byte_order)
+{
+	struct purex_entry_24xx *purex = NULL;
+	struct pt_ls4_rx_unsol *purls = NULL;
+	struct rsp_que *rsp_q = *rsp;
+	sts_cont_entry_t *new_pkt;
+	uint16_t no_bytes = 0, total_bytes = 0, pending_bytes = 0;
+	uint16_t buffer_copy_offset = 0, payload_size = 0;
+	uint16_t entry_count, entry_count_remaining;
+	struct purex_item *item;
+	void *iocb_pkt = NULL;
+
+	if (is_purls) {
+		purls = *pkt;
+		total_bytes = (le16_to_cpu(purls->frame_size) & 0x0FFF) -
+			      PURX_ELS_HEADER_SIZE;
+		entry_count = entry_count_remaining = purls->entry_count;
+		payload_size = sizeof(purls->payload);
+	} else {
+		purex = *pkt;
+		total_bytes = (le16_to_cpu(purex->frame_size) & 0x0FFF) -
+			      PURX_ELS_HEADER_SIZE;
+		entry_count = entry_count_remaining = purex->entry_count;
+		payload_size = sizeof(purex->els_frame_payload);
+	}
+
+	pending_bytes = total_bytes;
+	no_bytes = (pending_bytes > payload_size) ? payload_size :
+		   pending_bytes;
+	ql_dbg(ql_dbg_async, vha, 0x509a,
+	       "%s LS, frame_size 0x%x, entry count %d\n",
+	       (is_purls ? "PURLS" : "FPIN"), total_bytes, entry_count);
+
+	item = qla24xx_alloc_purex_item(vha, total_bytes);
+	if (!item)
+		return item;
+
+	iocb_pkt = &item->iocb;
+
+	if (is_purls)
+		memcpy(iocb_pkt, &purls->payload[0], no_bytes);
+	else
+		memcpy(iocb_pkt, &purex->els_frame_payload[0], no_bytes);
+	buffer_copy_offset += no_bytes;
+	pending_bytes -= no_bytes;
+	--entry_count_remaining;
+
+	if (is_purls)
+		((response_t *)purls)->signature = RESPONSE_PROCESSED;
+	else
+		((response_t *)purex)->signature = RESPONSE_PROCESSED;
+	wmb();
+
+	do {
+		while ((total_bytes > 0) && (entry_count_remaining > 0)) {
+			if (rsp_q->ring_ptr->signature == RESPONSE_PROCESSED) {
+				ql_dbg(ql_dbg_async, vha, 0x5084,
+				       "Ran out of IOCBs, partial data 0x%x\n",
+				       buffer_copy_offset);
+				cpu_relax();
+				continue;
+			}
+
+			new_pkt = (sts_cont_entry_t *)rsp_q->ring_ptr;
+			*pkt = new_pkt;
+
+			if (new_pkt->entry_type != STATUS_CONT_TYPE) {
+				ql_log(ql_log_warn, vha, 0x507a,
+				       "Unexpected IOCB type, partial data 0x%x\n",
+				       buffer_copy_offset);
+				break;
+			}
+
+			rsp_q->ring_index++;
+			if (rsp_q->ring_index == rsp_q->length) {
+				rsp_q->ring_index = 0;
+				rsp_q->ring_ptr = rsp_q->ring;
+			} else {
+				rsp_q->ring_ptr++;
+			}
+			no_bytes = (pending_bytes > sizeof(new_pkt->data)) ?
+				sizeof(new_pkt->data) : pending_bytes;
+			if ((buffer_copy_offset + no_bytes) <= total_bytes) {
+				memcpy(((uint8_t *)iocb_pkt + buffer_copy_offset),
+				       new_pkt->data, no_bytes);
+				buffer_copy_offset += no_bytes;
+				pending_bytes -= no_bytes;
+				--entry_count_remaining;
+			} else {
+				ql_log(ql_log_warn, vha, 0x5044,
+				       "Attempt to copy more that we got, optimizing..%x\n",
+				       buffer_copy_offset);
+				memcpy(((uint8_t *)iocb_pkt + buffer_copy_offset),
+				       new_pkt->data,
+				       total_bytes - buffer_copy_offset);
+			}
+
+			((response_t *)new_pkt)->signature = RESPONSE_PROCESSED;
+			wmb();
+		}
+
+		if (pending_bytes != 0 || entry_count_remaining != 0) {
+			ql_log(ql_log_fatal, vha, 0x508b,
+			       "Dropping partial FPIN, underrun bytes = 0x%x, entry cnts 0x%x\n",
+			       total_bytes, entry_count_remaining);
+			qla24xx_free_purex_item(item);
+			return NULL;
+		}
+	} while (entry_count_remaining > 0);
+
+	if (byte_order)
+		host_to_fcp_swap((uint8_t *)&item->iocb, total_bytes);
+
+	return item;
+}
+
 int
 qla2x00_is_a_vp_did(scsi_qla_host_t *vha, uint32_t rscn_entry)
 {
@@ -958,7 +1103,7 @@ initialize_purex_header:
 	return item;
 }
 
-static void
+void
 qla24xx_queue_purex_item(scsi_qla_host_t *vha, struct purex_item *pkt,
 			 void (*process_item)(struct scsi_qla_host *vha,
 					      struct purex_item *pkt))
@@ -1798,6 +1943,8 @@ global_port_update:
 		break;
 
 	case MBA_TEMPERATURE_ALERT:
+		if (IS_QLA27XX(ha) || IS_QLA28XX(ha))
+			display_Laser_info(vha, mb[1], mb[2], mb[3]);
 		ql_dbg(ql_dbg_async, vha, 0x505e,
 		    "TEMPERATURE ALERT: %04x %04x %04x\n", mb[1], mb[2], mb[3]);
 		break;
@@ -3811,6 +3958,7 @@ void qla24xx_process_response_queue(struct scsi_qla_host *vha,
 	struct qla_hw_data *ha = vha->hw;
 	struct purex_entry_24xx *purex_entry;
 	struct purex_item *pure_item;
+	struct pt_ls4_rx_unsol *p;
 	u16 rsp_in = 0, cur_ring_index;
 	int is_shadow_hba;
 
@@ -3983,7 +4131,19 @@ process_err:
 			qla28xx_sa_update_iocb_entry(vha, rsp->req,
 				(struct sa_update_28xx *)pkt);
 			break;
-
+		case PT_LS4_UNSOL:
+			p = (void *)pkt;
+			if (qla_chk_cont_iocb_avail(vha, rsp, (response_t *)pkt, rsp_in)) {
+				rsp->ring_ptr = (response_t *)pkt;
+				rsp->ring_index = cur_ring_index;
+
+				ql_dbg(ql_dbg_init, vha, 0x2124,
+				       "Defer processing UNSOL LS req opcode %#x...\n",
+				       p->payload[0]);
+				return;
+			}
+			qla2xxx_process_purls_iocb((void **)&pkt, &rsp);
+			break;
 		default:
 			/* Type Not Supported. */
 			ql_dbg(ql_dbg_async, vha, 0x5042,
diff --git a/drivers/scsi/qla2xxx/qla_mbx.c b/drivers/scsi/qla2xxx/qla_mbx.c
index b05f93037875..21ec32b4fb28 100644
--- a/drivers/scsi/qla2xxx/qla_mbx.c
+++ b/drivers/scsi/qla2xxx/qla_mbx.c
@@ -273,7 +273,6 @@ qla2x00_mailbox_command(scsi_qla_host_t *vha, mbx_cmd_t *mcp)
 		spin_unlock_irqrestore(&ha->hardware_lock, flags);
 
 		wait_time = jiffies;
-		atomic_inc(&ha->num_pend_mbx_stage3);
 		if (!wait_for_completion_timeout(&ha->mbx_intr_comp,
 		    mcp->tov * HZ)) {
 			ql_dbg(ql_dbg_mbx, vha, 0x117a,
@@ -290,7 +289,6 @@ qla2x00_mailbox_command(scsi_qla_host_t *vha, mbx_cmd_t *mcp)
 				spin_unlock_irqrestore(&ha->hardware_lock,
 				    flags);
 				atomic_dec(&ha->num_pend_mbx_stage2);
-				atomic_dec(&ha->num_pend_mbx_stage3);
 				rval = QLA_ABORTED;
 				goto premature_exit;
 			}
@@ -302,11 +300,9 @@ qla2x00_mailbox_command(scsi_qla_host_t *vha, mbx_cmd_t *mcp)
 			ha->flags.mbox_busy = 0;
 			spin_unlock_irqrestore(&ha->hardware_lock, flags);
 			atomic_dec(&ha->num_pend_mbx_stage2);
-			atomic_dec(&ha->num_pend_mbx_stage3);
 			rval = QLA_ABORTED;
 			goto premature_exit;
 		}
-		atomic_dec(&ha->num_pend_mbx_stage3);
 
 		if (time_after(jiffies, wait_time + 5 * HZ))
 			ql_log(ql_log_warn, vha, 0x1015, "cmd=0x%x, waited %d msecs\n",
diff --git a/drivers/scsi/qla2xxx/qla_nvme.c b/drivers/scsi/qla2xxx/qla_nvme.c
index 9941b38eac93..62a67662cbf3 100644
--- a/drivers/scsi/qla2xxx/qla_nvme.c
+++ b/drivers/scsi/qla2xxx/qla_nvme.c
@@ -12,6 +12,26 @@
 #include <linux/blk-mq.h>
 
 static struct nvme_fc_port_template qla_nvme_fc_transport;
+static int qla_nvme_ls_reject_iocb(struct scsi_qla_host *vha,
+				   struct qla_qpair *qp,
+				   struct qla_nvme_lsrjt_pt_arg *a,
+				   bool is_xchg_terminate);
+
+struct qla_nvme_unsol_ctx {
+	struct list_head elem;
+	struct scsi_qla_host *vha;
+	struct fc_port *fcport;
+	struct srb *sp;
+	struct nvmefc_ls_rsp lsrsp;
+	struct nvmefc_ls_rsp *fd_rsp;
+	struct work_struct lsrsp_work;
+	struct work_struct abort_work;
+	__le32 exchange_address;
+	__le16 nport_handle;
+	__le16 ox_id;
+	int comp_status;
+	spinlock_t cmd_lock;
+};
 
 int qla_nvme_register_remote(struct scsi_qla_host *vha, struct fc_port *fcport)
 {
@@ -216,6 +236,55 @@ static void qla_nvme_sp_ls_done(srb_t *sp, int res)
 	schedule_work(&priv->ls_work);
 }
 
+static void qla_nvme_release_lsrsp_cmd_kref(struct kref *kref)
+{
+	struct srb *sp = container_of(kref, struct srb, cmd_kref);
+	struct qla_nvme_unsol_ctx *uctx = sp->priv;
+	struct nvmefc_ls_rsp *fd_rsp;
+	unsigned long flags;
+
+	if (!uctx) {
+		qla2x00_rel_sp(sp);
+		return;
+	}
+
+	spin_lock_irqsave(&uctx->cmd_lock, flags);
+	uctx->sp = NULL;
+	sp->priv = NULL;
+	spin_unlock_irqrestore(&uctx->cmd_lock, flags);
+
+	fd_rsp = uctx->fd_rsp;
+
+	list_del(&uctx->elem);
+
+	fd_rsp->done(fd_rsp);
+	kfree(uctx);
+	qla2x00_rel_sp(sp);
+}
+
+static void qla_nvme_lsrsp_complete(struct work_struct *work)
+{
+	struct qla_nvme_unsol_ctx *uctx =
+		container_of(work, struct qla_nvme_unsol_ctx, lsrsp_work);
+
+	kref_put(&uctx->sp->cmd_kref, qla_nvme_release_lsrsp_cmd_kref);
+}
+
+static void qla_nvme_sp_lsrsp_done(srb_t *sp, int res)
+{
+	struct qla_nvme_unsol_ctx *uctx = sp->priv;
+
+	if (WARN_ON_ONCE(kref_read(&sp->cmd_kref) == 0))
+		return;
+
+	if (res)
+		res = -EINVAL;
+
+	uctx->comp_status = res;
+	INIT_WORK(&uctx->lsrsp_work, qla_nvme_lsrsp_complete);
+	schedule_work(&uctx->lsrsp_work);
+}
+
 /* it assumed that QPair lock is held. */
 static void qla_nvme_sp_done(srb_t *sp, int res)
 {
@@ -288,6 +357,92 @@ out:
 	kref_put(&sp->cmd_kref, sp->put_fn);
 }
 
+static int qla_nvme_xmt_ls_rsp(struct nvme_fc_local_port *lport,
+			       struct nvme_fc_remote_port *rport,
+			       struct nvmefc_ls_rsp *fd_resp)
+{
+	struct qla_nvme_unsol_ctx *uctx = container_of(fd_resp,
+				struct qla_nvme_unsol_ctx, lsrsp);
+	struct qla_nvme_rport *qla_rport = rport->private;
+	fc_port_t *fcport = qla_rport->fcport;
+	struct scsi_qla_host *vha = uctx->vha;
+	struct qla_hw_data *ha = vha->hw;
+	struct qla_nvme_lsrjt_pt_arg a;
+	struct srb_iocb *nvme;
+	srb_t *sp;
+	int rval = QLA_FUNCTION_FAILED;
+	uint8_t cnt = 0;
+
+	if (!fcport || fcport->deleted)
+		goto out;
+
+	if (!ha->flags.fw_started)
+		goto out;
+
+	/* Alloc SRB structure */
+	sp = qla2x00_get_sp(vha, fcport, GFP_ATOMIC);
+	if (!sp)
+		goto out;
+
+	sp->type = SRB_NVME_LS;
+	sp->name = "nvme_ls";
+	sp->done = qla_nvme_sp_lsrsp_done;
+	sp->put_fn = qla_nvme_release_lsrsp_cmd_kref;
+	sp->priv = (void *)uctx;
+	sp->unsol_rsp = 1;
+	uctx->sp = sp;
+	spin_lock_init(&uctx->cmd_lock);
+	nvme = &sp->u.iocb_cmd;
+	uctx->fd_rsp = fd_resp;
+	nvme->u.nvme.desc = fd_resp;
+	nvme->u.nvme.dir = 0;
+	nvme->u.nvme.dl = 0;
+	nvme->u.nvme.timeout_sec = 0;
+	nvme->u.nvme.cmd_dma = fd_resp->rspdma;
+	nvme->u.nvme.cmd_len = fd_resp->rsplen;
+	nvme->u.nvme.rsp_len = 0;
+	nvme->u.nvme.rsp_dma = 0;
+	nvme->u.nvme.exchange_address = uctx->exchange_address;
+	nvme->u.nvme.nport_handle = uctx->nport_handle;
+	nvme->u.nvme.ox_id = uctx->ox_id;
+	dma_sync_single_for_device(&ha->pdev->dev, nvme->u.nvme.cmd_dma,
+				   le32_to_cpu(fd_resp->rsplen), DMA_TO_DEVICE);
+
+	ql_dbg(ql_dbg_unsol, vha, 0x2122,
+	       "Unsol lsreq portid=%06x %8phC exchange_address 0x%x ox_id 0x%x hdl 0x%x\n",
+	       fcport->d_id.b24, fcport->port_name, uctx->exchange_address,
+	       uctx->ox_id, uctx->nport_handle);
+retry:
+	rval = qla2x00_start_sp(sp);
+	switch (rval) {
+	case QLA_SUCCESS:
+		break;
+	case EAGAIN:
+		msleep(PURLS_MSLEEP_INTERVAL);
+		cnt++;
+		if (cnt < PURLS_RETRY_COUNT)
+			goto retry;
+
+		fallthrough;
+	default:
+		ql_dbg(ql_log_warn, vha, 0x2123,
+		       "Failed to xmit Unsol ls response = %d\n", rval);
+		rval = -EIO;
+		qla2x00_rel_sp(sp);
+		goto out;
+	}
+
+	return 0;
+out:
+	memset((void *)&a, 0, sizeof(a));
+	a.vp_idx = vha->vp_idx;
+	a.nport_handle = uctx->nport_handle;
+	a.xchg_address = uctx->exchange_address;
+	qla_nvme_ls_reject_iocb(vha, ha->base_qpair, &a, true);
+	kfree(uctx);
+	return rval;
+}
+
 static void qla_nvme_ls_abort(struct nvme_fc_local_port *lport,
     struct nvme_fc_remote_port *rport, struct nvmefc_ls_req *fd)
 {
@@ -355,7 +510,7 @@ static int qla_nvme_ls_req(struct nvme_fc_local_port *lport,
 	nvme->u.nvme.timeout_sec = fd->timeout;
 	nvme->u.nvme.cmd_dma = fd->rqstdma;
 	dma_sync_single_for_device(&ha->pdev->dev, nvme->u.nvme.cmd_dma,
-	    fd->rqstlen, DMA_TO_DEVICE);
+	    le32_to_cpu(fd->rqstlen), DMA_TO_DEVICE);
 
 	rval = qla2x00_start_sp(sp);
 	if (rval != QLA_SUCCESS) {
@@ -720,6 +875,7 @@ static struct nvme_fc_port_template qla_nvme_fc_transport = {
 	.ls_abort	= qla_nvme_ls_abort,
 	.fcp_io		= qla_nvme_post_cmd,
 	.fcp_abort	= qla_nvme_fcp_abort,
+	.xmt_ls_rsp	= qla_nvme_xmt_ls_rsp,
 	.map_queues	= qla_nvme_map_queues,
 	.max_hw_queues  = DEF_NVME_HW_QUEUES,
 	.max_sgl_segments = 1024,
@@ -924,3 +1080,248 @@ inline void qla_wait_nvme_release_cmd_kref(srb_t *orig_sp)
 		return;
 	kref_put(&orig_sp->cmd_kref, orig_sp->put_fn);
 }
+
+static void qla_nvme_fc_format_rjt(void *buf, u8 ls_cmd, u8 reason,
+				   u8 explanation, u8 vendor)
+{
+	struct fcnvme_ls_rjt *rjt = buf;
+
+	rjt->w0.ls_cmd = FCNVME_LSDESC_RQST;
+	rjt->desc_list_len = fcnvme_lsdesc_len(sizeof(struct fcnvme_ls_rjt));
+	rjt->rqst.desc_tag = cpu_to_be32(FCNVME_LSDESC_RQST);
+	rjt->rqst.desc_len =
+		fcnvme_lsdesc_len(sizeof(struct fcnvme_lsdesc_rqst));
+	rjt->rqst.w0.ls_cmd = ls_cmd;
+	rjt->rjt.desc_tag = cpu_to_be32(FCNVME_LSDESC_RJT);
+	rjt->rjt.desc_len = fcnvme_lsdesc_len(sizeof(struct fcnvme_lsdesc_rjt));
+	rjt->rjt.reason_code = reason;
+	rjt->rjt.reason_explanation = explanation;
+	rjt->rjt.vendor = vendor;
+}
+
+static void qla_nvme_lsrjt_pt_iocb(struct scsi_qla_host *vha,
+				   struct pt_ls4_request *lsrjt_iocb,
+				   struct qla_nvme_lsrjt_pt_arg *a)
+{
+	lsrjt_iocb->entry_type = PT_LS4_REQUEST;
+	lsrjt_iocb->entry_count = 1;
+	lsrjt_iocb->sys_define = 0;
+	lsrjt_iocb->entry_status = 0;
+	lsrjt_iocb->handle = QLA_SKIP_HANDLE;
+	lsrjt_iocb->nport_handle = a->nport_handle;
+	lsrjt_iocb->exchange_address = a->xchg_address;
+	lsrjt_iocb->vp_index = a->vp_idx;
+
+	lsrjt_iocb->control_flags = cpu_to_le16(a->control_flags);
+
+	put_unaligned_le64(a->tx_addr, &lsrjt_iocb->dsd[0].address);
+	lsrjt_iocb->dsd[0].length = cpu_to_le32(a->tx_byte_count);
+	lsrjt_iocb->tx_dseg_count = cpu_to_le16(1);
+	lsrjt_iocb->tx_byte_count = cpu_to_le32(a->tx_byte_count);
+
+	put_unaligned_le64(a->rx_addr, &lsrjt_iocb->dsd[1].address);
+	lsrjt_iocb->dsd[1].length = 0;
+	lsrjt_iocb->rx_dseg_count = 0;
+	lsrjt_iocb->rx_byte_count = 0;
+}
+
+static int
+qla_nvme_ls_reject_iocb(struct scsi_qla_host *vha, struct qla_qpair *qp,
+			struct qla_nvme_lsrjt_pt_arg *a, bool is_xchg_terminate)
+{
+	struct pt_ls4_request *lsrjt_iocb;
+
+	lsrjt_iocb = __qla2x00_alloc_iocbs(qp, NULL);
+	if (!lsrjt_iocb) {
+		ql_log(ql_log_warn, vha, 0x210e,
+		       "qla2x00_alloc_iocbs failed.\n");
+		return QLA_FUNCTION_FAILED;
+	}
+
+	if (!is_xchg_terminate) {
+		qla_nvme_fc_format_rjt((void *)vha->hw->lsrjt.c, a->opcode,
+				       a->reason, a->explanation, 0);
+
+		a->tx_byte_count = sizeof(struct fcnvme_ls_rjt);
+		a->tx_addr = vha->hw->lsrjt.cdma;
+		a->control_flags = CF_LS4_RESPONDER << CF_LS4_SHIFT;
+
+		ql_dbg(ql_dbg_unsol, vha, 0x211f,
+		       "Sending nvme fc ls reject ox_id %04x op %04x\n",
+		       a->ox_id, a->opcode);
+		ql_dump_buffer(ql_dbg_unsol + ql_dbg_verbose, vha, 0x210f,
+			       vha->hw->lsrjt.c, sizeof(*vha->hw->lsrjt.c));
+	} else {
+		a->tx_byte_count = 0;
+		a->control_flags = CF_LS4_RESPONDER_TERM << CF_LS4_SHIFT;
+		ql_dbg(ql_dbg_unsol, vha, 0x2110,
+		       "Terminate nvme ls xchg 0x%x\n", a->xchg_address);
+	}
+
+	qla_nvme_lsrjt_pt_iocb(vha, lsrjt_iocb, a);
+	/* flush iocb to mem before notifying hw doorbell */
+	wmb();
+	qla2x00_start_iocbs(vha, qp->req);
+	return 0;
+}
+
+/*
+ * qla2xxx_process_purls_pkt() - Pass-up Unsolicited
+ * Received FC-NVMe Link Service pkt to nvme_fc_rcv_ls_req().
+ * LLDD need to provide memory for response buffer, which
+ * will be used to reference the exchange corresponding
+ * to the LS when issuing an ls response. LLDD will have to free
+ * response buffer in lport->ops->xmt_ls_rsp().
+ *
+ * @vha: SCSI qla host
+ * @item: ptr to purex_item
+ */
+static void
+qla2xxx_process_purls_pkt(struct scsi_qla_host *vha, struct purex_item *item)
+{
+	struct qla_nvme_unsol_ctx *uctx = item->purls_context;
+	fc_port_t *fcport = uctx->fcport;
+	struct qla_nvme_lsrjt_pt_arg a;
+	int ret = 1;
+
+#if (IS_ENABLED(CONFIG_NVME_FC))
+	ret = nvme_fc_rcv_ls_req(fcport->nvme_remote_port, &uctx->lsrsp,
+				 &item->iocb, item->size);
+#endif
+	if (ret) {
+		ql_dbg(ql_dbg_unsol, vha, 0x2125, "NVMe tranport ls_req failed\n");
+		memset((void *)&a, 0, sizeof(a));
+		a.vp_idx = vha->vp_idx;
+		a.nport_handle = uctx->nport_handle;
+		a.xchg_address = uctx->exchange_address;
+		qla_nvme_ls_reject_iocb(vha, vha->hw->base_qpair, &a, true);
+		list_del(&uctx->elem);
+		kfree(uctx);
+	}
+}
+
+static scsi_qla_host_t *
+qla2xxx_get_vha_from_vp_idx(struct qla_hw_data *ha, uint16_t vp_index)
+{
+	scsi_qla_host_t *base_vha, *vha, *tvp;
+	unsigned long flags;
+
+	base_vha = pci_get_drvdata(ha->pdev);
+
+	if (!vp_index && !ha->num_vhosts)
+		return base_vha;
+
+	spin_lock_irqsave(&ha->vport_slock, flags);
+	list_for_each_entry_safe(vha, tvp, &ha->vp_list, list) {
+		if (vha->vp_idx == vp_index) {
+			spin_unlock_irqrestore(&ha->vport_slock, flags);
+			return vha;
+		}
+	}
+	spin_unlock_irqrestore(&ha->vport_slock, flags);
+
+	return NULL;
+}
+
+void qla2xxx_process_purls_iocb(void **pkt, struct rsp_que **rsp)
+{
+	struct nvme_fc_remote_port *rport;
+	struct qla_nvme_rport *qla_rport;
+	struct qla_nvme_lsrjt_pt_arg a;
+	struct pt_ls4_rx_unsol *p = *pkt;
+	struct qla_nvme_unsol_ctx *uctx;
+	struct rsp_que *rsp_q = *rsp;
+	struct qla_hw_data *ha;
+	scsi_qla_host_t	*vha;
+	fc_port_t *fcport = NULL;
+	struct purex_item *item;
+	port_id_t d_id = {0};
+	port_id_t id = {0};
+	u8 *opcode;
+	bool xmt_reject = false;
+
+	ha = rsp_q->hw;
+
+	vha = qla2xxx_get_vha_from_vp_idx(ha, p->vp_index);
+	if (!vha) {
+		ql_log(ql_log_warn, NULL, 0x2110, "Invalid vp index %d\n", p->vp_index);
+		WARN_ON_ONCE(1);
+		return;
+	}
+
+	memset((void *)&a, 0, sizeof(a));
+	opcode = (u8 *)&p->payload[0];
+	a.opcode = opcode[3];
+	a.vp_idx = p->vp_index;
+	a.nport_handle = p->nport_handle;
+	a.ox_id = p->ox_id;
+	a.xchg_address = p->exchange_address;
+
+	id.b.domain = p->s_id.domain;
+	id.b.area   = p->s_id.area;
+	id.b.al_pa  = p->s_id.al_pa;
+	d_id.b.domain = p->d_id[2];
+	d_id.b.area   = p->d_id[1];
+	d_id.b.al_pa  = p->d_id[0];
+
+	fcport = qla2x00_find_fcport_by_nportid(vha, &id, 0);
+	if (!fcport) {
+		ql_dbg(ql_dbg_unsol, vha, 0x211e,
+		       "Failed to find sid=%06x did=%06x\n",
+		       id.b24, d_id.b24);
+		a.reason = FCNVME_RJT_RC_INV_ASSOC;
+		a.explanation = FCNVME_RJT_EXP_NONE;
+		xmt_reject = true;
+		goto out;
+	}
+	rport = fcport->nvme_remote_port;
+	qla_rport = rport->private;
+
+	item = qla27xx_copy_multiple_pkt(vha, pkt, rsp, true, false);
+	if (!item) {
+		a.reason = FCNVME_RJT_RC_LOGIC;
+		a.explanation = FCNVME_RJT_EXP_NONE;
+		xmt_reject = true;
+		goto out;
+	}
+
+	uctx = kzalloc(sizeof(*uctx), GFP_ATOMIC);
+	if (!uctx) {
+		ql_log(ql_log_info, vha, 0x2126, "Failed allocate memory\n");
+		a.reason = FCNVME_RJT_RC_LOGIC;
+		a.explanation = FCNVME_RJT_EXP_NONE;
+		xmt_reject = true;
+		kfree(item);
+		goto out;
+	}
+
+	uctx->vha = vha;
+	uctx->fcport = fcport;
+	uctx->exchange_address = p->exchange_address;
+	uctx->nport_handle = p->nport_handle;
+	uctx->ox_id = p->ox_id;
+	qla_rport->uctx = uctx;
+	INIT_LIST_HEAD(&uctx->elem);
+	list_add_tail(&uctx->elem, &fcport->unsol_ctx_head);
+	item->purls_context = (void *)uctx;
+
+	ql_dbg(ql_dbg_unsol, vha, 0x2121,
+	       "PURLS OP[%01x] size %d xchg addr 0x%x portid %06x\n",
+	       item->iocb.iocb[3], item->size, uctx->exchange_address,
+	       fcport->d_id.b24);
+	/* +48    0  1  2  3  4  5  6  7  8  9  A  B  C  D  E  F
+	 * ----- -----------------------------------------------
+	 * 0000: 00 00 00 05 28 00 00 00 07 00 00 00 08 00 00 00
+	 * 0010: ab ec 0f cc 00 00 8d 7d 05 00 00 00 10 00 00 00
+	 * 0020: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
+	 */
+	ql_dump_buffer(ql_dbg_unsol + ql_dbg_verbose, vha, 0x2120,
+		       &item->iocb, item->size);
+
+	qla24xx_queue_purex_item(vha, item, qla2xxx_process_purls_pkt);
+out:
+	if (xmt_reject) {
+		qla_nvme_ls_reject_iocb(vha, (*rsp)->qpair, &a, false);
+		__qla_consume_iocb(vha, pkt, rsp);
+	}
+}
diff --git a/drivers/scsi/qla2xxx/qla_nvme.h b/drivers/scsi/qla2xxx/qla_nvme.h
index d299478371b2..a253ac55171b 100644
--- a/drivers/scsi/qla2xxx/qla_nvme.h
+++ b/drivers/scsi/qla2xxx/qla_nvme.h
@@ -21,6 +21,7 @@
 #define Q2T_NVME_NUM_TAGS 2048
 #define QLA_MAX_FC_SEGMENTS 64
 
+struct qla_nvme_unsol_ctx;
 struct scsi_qla_host;
 struct qla_hw_data;
 struct req_que;
@@ -37,6 +38,7 @@ struct nvme_private {
 
 struct qla_nvme_rport {
 	struct fc_port *fcport;
+	struct qla_nvme_unsol_ctx *uctx;
 };
 
 #define COMMAND_NVME    0x88            /* Command Type FC-NVMe IOCB */
@@ -75,6 +77,9 @@ struct cmd_nvme {
 	struct dsd64 nvme_dsd;
 };
 
+#define PURLS_MSLEEP_INTERVAL	1
+#define PURLS_RETRY_COUNT	5
+
 #define PT_LS4_REQUEST 0x89	/* Link Service pass-through IOCB (request) */
 struct pt_ls4_request {
 	uint8_t entry_type;
@@ -118,21 +123,19 @@ struct pt_ls4_rx_unsol {
 	__le32	exchange_address;
 	uint8_t d_id[3];
 	uint8_t r_ctl;
-	be_id_t s_id;
+	le_id_t s_id;
 	uint8_t cs_ctl;
 	uint8_t f_ctl[3];
 	uint8_t type;
 	__le16	seq_cnt;
 	uint8_t df_ctl;
 	uint8_t seq_id;
-	__le16	rx_id;
-	__le16	ox_id;
-	__le32	param;
-	__le32	desc0;
+	__le16 rx_id;
+	__le16 ox_id;
+	__le32  desc0;
 #define PT_LS4_PAYLOAD_OFFSET 0x2c
 #define PT_LS4_FIRST_PACKET_LEN 20
-	__le32	desc_len;
-	__le32	payload[3];
+	__le32 payload[5];
 };
 
 /*
diff --git a/drivers/scsi/qla2xxx/qla_nx.h b/drivers/scsi/qla2xxx/qla_nx.h
index 6dc80c8ddf79..5d1bdc15b75c 100644
--- a/drivers/scsi/qla2xxx/qla_nx.h
+++ b/drivers/scsi/qla2xxx/qla_nx.h
@@ -857,7 +857,9 @@ struct fcp_cmnd {
 	uint8_t task_attribute;
 	uint8_t task_management;
 	uint8_t additional_cdb_len;
-	uint8_t cdb[260]; /* 256 for CDB len and 4 for FCP_DL */
+#define QLA_CDB_BUF_SIZE  256
+#define QLA_FCP_DL_SIZE   4
+	uint8_t cdb[QLA_CDB_BUF_SIZE + QLA_FCP_DL_SIZE]; /* 256 for CDB len and 4 for FCP_DL */
 };
 
 struct dsd_dma {
diff --git a/drivers/scsi/qla2xxx/qla_os.c b/drivers/scsi/qla2xxx/qla_os.c
index 03bc3a0b45b6..50db08265c51 100644
--- a/drivers/scsi/qla2xxx/qla_os.c
+++ b/drivers/scsi/qla2xxx/qla_os.c
@@ -44,10 +44,11 @@ module_param(ql2xfulldump_on_mpifail, int, S_IRUGO | S_IWUSR);
 MODULE_PARM_DESC(ql2xfulldump_on_mpifail,
 		 "Set this to take full dump on MPI hang.");
 
-int ql2xenforce_iocb_limit = 1;
+int ql2xenforce_iocb_limit = 2;
 module_param(ql2xenforce_iocb_limit, int, S_IRUGO | S_IWUSR);
 MODULE_PARM_DESC(ql2xenforce_iocb_limit,
-		 "Enforce IOCB throttling, to avoid FW congestion. (default: 1)");
+		 "Enforce IOCB throttling, to avoid FW congestion. (default: 2) "
+		 "1: track usage per queue, 2: track usage per adapter");
 
 /*
  * CT6 CTX allocation cache
@@ -432,6 +433,7 @@ static void qla_init_base_qpair(struct scsi_qla_host *vha, struct req_que *req,
 	ha->base_qpair->msix = &ha->msix_entries[QLA_MSIX_RSP_Q];
 	ha->base_qpair->srb_mempool = ha->srb_mempool;
 	INIT_LIST_HEAD(&ha->base_qpair->hints_list);
+	INIT_LIST_HEAD(&ha->base_qpair->dsd_list);
 	ha->base_qpair->enable_class_2 = ql2xenableclass2;
 	/* init qpair to this cpu. Will adjust at run time. */
 	qla_cpu_update(rsp->qpair, raw_smp_processor_id());
@@ -750,9 +752,9 @@ void qla2x00_sp_free_dma(srb_t *sp)
 
 		dma_pool_free(ha->fcp_cmnd_dma_pool, ctx1->fcp_cmnd,
 		    ctx1->fcp_cmnd_dma);
-		list_splice(&ctx1->dsd_list, &ha->gbl_dsd_list);
-		ha->gbl_dsd_inuse -= ctx1->dsd_use_cnt;
-		ha->gbl_dsd_avail += ctx1->dsd_use_cnt;
+		list_splice(&ctx1->dsd_list, &sp->qpair->dsd_list);
+		sp->qpair->dsd_inuse -= ctx1->dsd_use_cnt;
+		sp->qpair->dsd_avail += ctx1->dsd_use_cnt;
 	}
 
 	if (sp->flags & SRB_GOT_BUF)
@@ -836,9 +838,9 @@ void qla2xxx_qpair_sp_free_dma(srb_t *sp)
 
 		dma_pool_free(ha->fcp_cmnd_dma_pool, ctx1->fcp_cmnd,
 		    ctx1->fcp_cmnd_dma);
-		list_splice(&ctx1->dsd_list, &ha->gbl_dsd_list);
-		ha->gbl_dsd_inuse -= ctx1->dsd_use_cnt;
-		ha->gbl_dsd_avail += ctx1->dsd_use_cnt;
+		list_splice(&ctx1->dsd_list, &sp->qpair->dsd_list);
+		sp->qpair->dsd_inuse -= ctx1->dsd_use_cnt;
+		sp->qpair->dsd_avail += ctx1->dsd_use_cnt;
 		sp->flags &= ~SRB_FCP_CMND_DMA_VALID;
 	}
 
@@ -3007,7 +3009,6 @@ qla2x00_probe_one(struct pci_dev *pdev, const struct pci_device_id *id)
 	ha->max_exchg = FW_MAX_EXCHANGES_CNT;
 	atomic_set(&ha->num_pend_mbx_stage1, 0);
 	atomic_set(&ha->num_pend_mbx_stage2, 0);
-	atomic_set(&ha->num_pend_mbx_stage3, 0);
 	atomic_set(&ha->zio_threshold, DEFAULT_ZIO_THRESHOLD);
 	ha->last_zio_threshold = DEFAULT_ZIO_THRESHOLD;
 	INIT_LIST_HEAD(&ha->tmf_pending);
@@ -3288,6 +3289,13 @@ qla2x00_probe_one(struct pci_dev *pdev, const struct pci_device_id *id)
 	host->max_id = ha->max_fibre_devices;
 	host->cmd_per_lun = 3;
 	host->unique_id = host->host_no;
+
+	if (ql2xenabledif && ql2xenabledif != 2) {
+		ql_log(ql_log_warn, base_vha, 0x302d,
+		       "Invalid value for ql2xenabledif, resetting it to default (2)\n");
+		ql2xenabledif = 2;
+	}
+
 	if (IS_T10_PI_CAPABLE(ha) && ql2xenabledif)
 		host->max_cmd_len = 32;
 	else
@@ -3524,8 +3532,6 @@ skip_dpc:
 			base_vha->flags.difdix_supported = 1;
 			ql_dbg(ql_dbg_init, base_vha, 0x00f1,
 			    "Registering for DIF/DIX type 1 and 3 protection.\n");
-			if (ql2xenabledif == 1)
-				prot = SHOST_DIX_TYPE0_PROTECTION;
 			if (ql2xprotmask)
 				scsi_host_set_prot(host, ql2xprotmask);
 			else
@@ -4402,7 +4408,6 @@ qla2x00_mem_alloc(struct qla_hw_data *ha, uint16_t req_len, uint16_t rsp_len,
 			   "sf_init_cb=%p.\n", ha->sf_init_cb);
 	}
 
-	INIT_LIST_HEAD(&ha->gbl_dsd_list);
 
 	/* Get consistent memory allocated for Async Port-Database. */
 	if (!IS_FWI2_CAPABLE(ha)) {
@@ -4457,8 +4462,9 @@ qla2x00_mem_alloc(struct qla_hw_data *ha, uint16_t req_len, uint16_t rsp_len,
 
 	ha->elsrej.size = sizeof(struct fc_els_ls_rjt) + 16;
 	ha->elsrej.c = dma_alloc_coherent(&ha->pdev->dev,
-	    ha->elsrej.size, &ha->elsrej.cdma, GFP_KERNEL);
-
+					  ha->elsrej.size,
+					  &ha->elsrej.cdma,
+					  GFP_KERNEL);
 	if (!ha->elsrej.c) {
 		ql_dbg_pci(ql_dbg_init, ha->pdev, 0xffff,
 		    "Alloc failed for els reject cmd.\n");
@@ -4467,8 +4473,21 @@ qla2x00_mem_alloc(struct qla_hw_data *ha, uint16_t req_len, uint16_t rsp_len,
 	ha->elsrej.c->er_cmd = ELS_LS_RJT;
 	ha->elsrej.c->er_reason = ELS_RJT_LOGIC;
 	ha->elsrej.c->er_explan = ELS_EXPL_UNAB_DATA;
+
+	ha->lsrjt.size = sizeof(struct fcnvme_ls_rjt);
+	ha->lsrjt.c = dma_alloc_coherent(&ha->pdev->dev, ha->lsrjt.size,
+			&ha->lsrjt.cdma, GFP_KERNEL);
+	if (!ha->lsrjt.c) {
+		ql_dbg_pci(ql_dbg_init, ha->pdev, 0xffff,
+			   "Alloc failed for nvme fc reject cmd.\n");
+		goto fail_lsrjt;
+	}
+
 	return 0;
 
+fail_lsrjt:
+	dma_free_coherent(&ha->pdev->dev, ha->elsrej.size,
+			  ha->elsrej.c, ha->elsrej.cdma);
 fail_elsrej:
 	dma_pool_destroy(ha->purex_dma_pool);
 fail_flt:
@@ -4934,18 +4953,16 @@ qla2x00_mem_free(struct qla_hw_data *ha)
 	ha->gid_list = NULL;
 	ha->gid_list_dma = 0;
 
-	if (IS_QLA82XX(ha)) {
-		if (!list_empty(&ha->gbl_dsd_list)) {
-			struct dsd_dma *dsd_ptr, *tdsd_ptr;
-
-			/* clean up allocated prev pool */
-			list_for_each_entry_safe(dsd_ptr,
-				tdsd_ptr, &ha->gbl_dsd_list, list) {
-				dma_pool_free(ha->dl_dma_pool,
-				dsd_ptr->dsd_addr, dsd_ptr->dsd_list_dma);
-				list_del(&dsd_ptr->list);
-				kfree(dsd_ptr);
-			}
+	if (!list_empty(&ha->base_qpair->dsd_list)) {
+		struct dsd_dma *dsd_ptr, *tdsd_ptr;
+
+		/* clean up allocated prev pool */
+		list_for_each_entry_safe(dsd_ptr, tdsd_ptr,
+					 &ha->base_qpair->dsd_list, list) {
+			dma_pool_free(ha->dl_dma_pool, dsd_ptr->dsd_addr,
+				      dsd_ptr->dsd_list_dma);
+			list_del(&dsd_ptr->list);
+			kfree(dsd_ptr);
 		}
 	}
 
@@ -5000,6 +5017,12 @@ qla2x00_mem_free(struct qla_hw_data *ha)
 		ha->elsrej.c = NULL;
 	}
 
+	if (ha->lsrjt.c) {
+		dma_free_coherent(&ha->pdev->dev, ha->lsrjt.size, ha->lsrjt.c,
+				  ha->lsrjt.cdma);
+		ha->lsrjt.c = NULL;
+	}
+
 	ha->init_cb = NULL;
 	ha->init_cb_dma = 0;
 
diff --git a/drivers/scsi/qla2xxx/qla_version.h b/drivers/scsi/qla2xxx/qla_version.h
index 81bdf6b03241..d903563e969e 100644
--- a/drivers/scsi/qla2xxx/qla_version.h
+++ b/drivers/scsi/qla2xxx/qla_version.h
@@ -6,9 +6,9 @@
 /*
  * Driver version
  */
-#define QLA2XXX_VERSION      "10.02.08.500-k"
+#define QLA2XXX_VERSION      "10.02.09.100-k"
 
 #define QLA_DRIVER_MAJOR_VER	10
 #define QLA_DRIVER_MINOR_VER	2
-#define QLA_DRIVER_PATCH_VER	8
-#define QLA_DRIVER_BETA_VER	500
+#define QLA_DRIVER_PATCH_VER	9
+#define QLA_DRIVER_BETA_VER	100