scsi: lpfc: Support non-uniform allocation of MSIX vectors to hardware queues

So far MSIX vector allocation assumed it would be 1:1 with hardware
queues. However, there are several reasons why fewer MSIX vectors may be
allocated than hardware queues such as the platform being out of vectors or
adapter limits being less than cpu count.

This patch reworks the MSIX/EQ relationships with the per-cpu hardware
queues so they can function independently. MSIX vectors will be equitably
split been cpu sockets/cores and then the per-cpu hardware queues will be
mapped to the vectors most efficient for them.

Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

11 files changed, 831 insertions, 343 deletions

diff --git a/drivers/scsi/lpfc/lpfc.h b/drivers/scsi/lpfc/lpfc.h
index 0f8964fdfecf..9fd2811ffa8b 100644
--- a/drivers/scsi/lpfc/lpfc.h
+++ b/drivers/scsi/lpfc/lpfc.h
@@ -84,8 +84,6 @@ struct lpfc_sli2_slim;
 #define LPFC_HB_MBOX_INTERVAL   5	/* Heart beat interval in seconds. */
 #define LPFC_HB_MBOX_TIMEOUT    30	/* Heart beat timeout  in seconds. */
 
-#define LPFC_LOOK_AHEAD_OFF	0	/* Look ahead logic is turned off */
-
 /* Error Attention event polling interval */
 #define LPFC_ERATT_POLL_INTERVAL	5 /* EATT poll interval in seconds */
 
@@ -821,6 +819,7 @@ struct lpfc_hba {
 	uint32_t cfg_fcp_imax;
 	uint32_t cfg_fcp_cpu_map;
 	uint32_t cfg_hdw_queue;
+	uint32_t cfg_irq_chann;
 	uint32_t cfg_suppress_rsp;
 	uint32_t cfg_nvme_oas;
 	uint32_t cfg_nvme_embed_cmd;
@@ -1042,6 +1041,9 @@ struct lpfc_hba {
 	struct dentry *debug_nvmeio_trc;
 	struct lpfc_debugfs_nvmeio_trc *nvmeio_trc;
 	struct dentry *debug_hdwqinfo;
+#ifdef LPFC_HDWQ_LOCK_STAT
+	struct dentry *debug_lockstat;
+#endif
 	atomic_t nvmeio_trc_cnt;
 	uint32_t nvmeio_trc_size;
 	uint32_t nvmeio_trc_output_idx;
@@ -1161,6 +1163,7 @@ struct lpfc_hba {
 #define LPFC_CHECK_NVME_IO	1
 #define LPFC_CHECK_NVMET_RCV	2
 #define LPFC_CHECK_NVMET_IO	4
+#define LPFC_CHECK_SCSI_IO	8
 	uint16_t ktime_on;
 	uint64_t ktime_data_samples;
 	uint64_t ktime_status_samples;
diff --git a/drivers/scsi/lpfc/lpfc_attr.c b/drivers/scsi/lpfc/lpfc_attr.c
index 787812dd57a9..fc7f80d68638 100644
--- a/drivers/scsi/lpfc/lpfc_attr.c
+++ b/drivers/scsi/lpfc/lpfc_attr.c
@@ -4958,7 +4958,7 @@ lpfc_fcp_imax_store(struct device *dev, struct device_attribute *attr,
 	phba->cfg_fcp_imax = (uint32_t)val;
 	phba->initial_imax = phba->cfg_fcp_imax;
 
-	for (i = 0; i < phba->cfg_hdw_queue; i += LPFC_MAX_EQ_DELAY_EQID_CNT)
+	for (i = 0; i < phba->cfg_irq_chann; i += LPFC_MAX_EQ_DELAY_EQID_CNT)
 		lpfc_modify_hba_eq_delay(phba, i, LPFC_MAX_EQ_DELAY_EQID_CNT,
 					 val);
 
@@ -5059,13 +5059,6 @@ lpfc_fcp_cpu_map_show(struct device *dev, struct device_attribute *attr,
 				phba->cfg_fcp_cpu_map,
 				phba->sli4_hba.num_online_cpu);
 		break;
-	case 2:
-		len += snprintf(buf + len, PAGE_SIZE-len,
-				"fcp_cpu_map: Driver centric mapping (%d): "
-				"%d online CPUs\n",
-				phba->cfg_fcp_cpu_map,
-				phba->sli4_hba.num_online_cpu);
-		break;
 	}
 
 	while (phba->sli4_hba.curr_disp_cpu < phba->sli4_hba.num_present_cpu) {
@@ -5076,35 +5069,35 @@ lpfc_fcp_cpu_map_show(struct device *dev, struct device_attribute *attr,
 				len += snprintf(
 					buf + len, PAGE_SIZE - len,
 					"CPU %02d hdwq None "
-					"physid %d coreid %d\n",
+					"physid %d coreid %d ht %d\n",
 					phba->sli4_hba.curr_disp_cpu,
 					cpup->phys_id,
-					cpup->core_id);
+					cpup->core_id, cpup->hyper);
 			else
 				len += snprintf(
 					buf + len, PAGE_SIZE - len,
-					"CPU %02d hdwq %04d "
-					"physid %d coreid %d\n",
+					"CPU %02d EQ %04d hdwq %04d "
+					"physid %d coreid %d ht %d\n",
 					phba->sli4_hba.curr_disp_cpu,
-					cpup->hdwq, cpup->phys_id,
-					cpup->core_id);
+					cpup->eq, cpup->hdwq, cpup->phys_id,
+					cpup->core_id, cpup->hyper);
 		} else {
 			if (cpup->hdwq == LPFC_VECTOR_MAP_EMPTY)
 				len += snprintf(
 					buf + len, PAGE_SIZE - len,
 					"CPU %02d hdwq None "
-					"physid %d coreid %d IRQ %d\n",
+					"physid %d coreid %d ht %d IRQ %d\n",
 					phba->sli4_hba.curr_disp_cpu,
 					cpup->phys_id,
-					cpup->core_id, cpup->irq);
+					cpup->core_id, cpup->hyper, cpup->irq);
 			else
 				len += snprintf(
 					buf + len, PAGE_SIZE - len,
-					"CPU %02d hdwq %04d "
-					"physid %d coreid %d IRQ %d\n",
+					"CPU %02d EQ %04d hdwq %04d "
+					"physid %d coreid %d ht %d IRQ %d\n",
 					phba->sli4_hba.curr_disp_cpu,
-					cpup->hdwq, cpup->phys_id,
-					cpup->core_id, cpup->irq);
+					cpup->eq, cpup->hdwq, cpup->phys_id,
+					cpup->core_id, cpup->hyper, cpup->irq);
 		}
 
 		phba->sli4_hba.curr_disp_cpu++;
@@ -5146,14 +5139,13 @@ lpfc_fcp_cpu_map_store(struct device *dev, struct device_attribute *attr,
 # lpfc_fcp_cpu_map: Defines how to map CPUs to IRQ vectors
 # for the HBA.
 #
-# Value range is [0 to 2]. Default value is LPFC_DRIVER_CPU_MAP (2).
+# Value range is [0 to 1]. Default value is LPFC_HBA_CPU_MAP (1).
 #	0 - Do not affinitze IRQ vectors
 #	1 - Affintize HBA vectors with respect to each HBA
 #	    (start with CPU0 for each HBA)
-#	2 - Affintize HBA vectors with respect to the entire driver
-#	    (round robin thru all CPUs across all HBAs)
+# This also defines how Hardware Queues are mapped to specific CPUs.
 */
-static int lpfc_fcp_cpu_map = LPFC_DRIVER_CPU_MAP;
+static int lpfc_fcp_cpu_map = LPFC_HBA_CPU_MAP;
 module_param(lpfc_fcp_cpu_map, int, S_IRUGO|S_IWUSR);
 MODULE_PARM_DESC(lpfc_fcp_cpu_map,
 		 "Defines how to map CPUs to IRQ vectors per HBA");
@@ -5187,7 +5179,7 @@ lpfc_fcp_cpu_map_init(struct lpfc_hba *phba, int val)
 	lpfc_printf_log(phba, KERN_ERR, LOG_INIT,
 			"3326 lpfc_fcp_cpu_map: %d out of range, using "
 			"default\n", val);
-	phba->cfg_fcp_cpu_map = LPFC_DRIVER_CPU_MAP;
+	phba->cfg_fcp_cpu_map = LPFC_HBA_CPU_MAP;
 
 	return 0;
 }
@@ -5308,7 +5300,7 @@ LPFC_ATTR_R(xri_rebalancing, 1, 0, 1, "Enable/Disable XRI rebalancing");
  * CPU. Otherwise, the default 0 (Round Robin) scheduling of FCP/NVME I/Os
  * through WQs will be used.
  */
-LPFC_ATTR_RW(fcp_io_sched, LPFC_FCP_SCHED_BY_HDWQ,
+LPFC_ATTR_RW(fcp_io_sched, LPFC_FCP_SCHED_BY_CPU,
 	     LPFC_FCP_SCHED_BY_HDWQ,
 	     LPFC_FCP_SCHED_BY_CPU,
 	     "Determine scheduling algorithm for "
@@ -5474,18 +5466,18 @@ LPFC_ATTR_RW(nvme_embed_cmd, 1, 0, 2,
 	     "Embed NVME Command in WQE");
 
 /*
- * lpfc_hdw_queue: Set the number of IO channels the driver
+ * lpfc_hdw_queue: Set the number of Hardware Queues the driver
  * will advertise it supports to the NVME and  SCSI layers. This also
- * will map to the number of EQ/CQ/WQs the driver will create.
+ * will map to the number of CQ/WQ pairs the driver will create.
  *
  * The NVME Layer will try to create this many, plus 1 administrative
  * hardware queue. The administrative queue will always map to WQ 0
- * A hardware IO queue maps (qidx) to a specific driver WQ.
+ * A hardware IO queue maps (qidx) to a specific driver CQ/WQ.
  *
  *      0    = Configure the number of hdw queues to the number of active CPUs.
- *      1,64 = Manually specify how many hdw queues to use.
+ *      1,128 = Manually specify how many hdw queues to use.
  *
- * Value range is [0,64]. Default value is 0.
+ * Value range is [0,128]. Default value is 0.
  */
 LPFC_ATTR_R(hdw_queue,
 	    LPFC_HBA_HDWQ_DEF,
@@ -5493,6 +5485,22 @@ LPFC_ATTR_R(hdw_queue,
 	    "Set the number of I/O Hardware Queues");
 
 /*
+ * lpfc_irq_chann: Set the number of IRQ vectors that are available
+ * for Hardware Queues to utilize.  This also will map to the number
+ * of EQ / MSI-X vectors the driver will create. This should never be
+ * more than the number of Hardware Queues
+ *
+ *      0     = Configure number of IRQ Channels to the number of active CPUs.
+ *      1,128 = Manually specify how many IRQ Channels to use.
+ *
+ * Value range is [0,128]. Default value is 0.
+ */
+LPFC_ATTR_R(irq_chann,
+	    LPFC_HBA_HDWQ_DEF,
+	    LPFC_HBA_HDWQ_MIN, LPFC_HBA_HDWQ_MAX,
+	    "Set the number of I/O IRQ Channels");
+
+/*
 # lpfc_enable_hba_reset: Allow or prevent HBA resets to the hardware.
 #       0  = HBA resets disabled
 #       1  = HBA resets enabled (default)
@@ -5533,16 +5541,6 @@ LPFC_ATTR_RW(XLanePriority, 0, 0x0, 0x7f, "CS_CTL for Express Lane Feature.");
 LPFC_ATTR_R(enable_bg, 0, 0, 1, "Enable BlockGuard Support");
 
 /*
-# lpfc_fcp_look_ahead: Look ahead for completions in FCP start routine
-#       0  = disabled (default)
-#       1  = enabled
-# Value range is [0,1]. Default value is 0.
-#
-# This feature in under investigation and may be supported in the future.
-*/
-unsigned int lpfc_fcp_look_ahead = LPFC_LOOK_AHEAD_OFF;
-
-/*
 # lpfc_prot_mask: i
 #	- Bit mask of host protection capabilities used to register with the
 #	  SCSI mid-layer
@@ -5788,6 +5786,7 @@ struct device_attribute *lpfc_hba_attrs[] = {
 	&dev_attr_lpfc_fcp_imax,
 	&dev_attr_lpfc_fcp_cpu_map,
 	&dev_attr_lpfc_hdw_queue,
+	&dev_attr_lpfc_irq_chann,
 	&dev_attr_lpfc_suppress_rsp,
 	&dev_attr_lpfc_nvmet_mrq,
 	&dev_attr_lpfc_nvmet_mrq_post,
@@ -6867,6 +6866,7 @@ lpfc_get_cfgparam(struct lpfc_hba *phba)
 	lpfc_nvme_enable_fb_init(phba, lpfc_nvme_enable_fb);
 	lpfc_nvmet_fb_size_init(phba, lpfc_nvmet_fb_size);
 	lpfc_hdw_queue_init(phba, lpfc_hdw_queue);
+	lpfc_irq_chann_init(phba, lpfc_irq_chann);
 	lpfc_enable_bbcr_init(phba, lpfc_enable_bbcr);
 	lpfc_enable_dpp_init(phba, lpfc_enable_dpp);
 
@@ -6891,6 +6891,10 @@ lpfc_get_cfgparam(struct lpfc_hba *phba)
 	/* A value of 0 means use the number of CPUs found in the system */
 	if (phba->cfg_hdw_queue == 0)
 		phba->cfg_hdw_queue = phba->sli4_hba.num_present_cpu;
+	if (phba->cfg_irq_chann == 0)
+		phba->cfg_irq_chann = phba->sli4_hba.num_present_cpu;
+	if (phba->cfg_irq_chann > phba->cfg_hdw_queue)
+		phba->cfg_irq_chann = phba->cfg_hdw_queue;
 
 	phba->cfg_soft_wwnn = 0L;
 	phba->cfg_soft_wwpn = 0L;
@@ -6933,6 +6937,10 @@ lpfc_nvme_mod_param_dep(struct lpfc_hba *phba)
 {
 	if (phba->cfg_hdw_queue > phba->sli4_hba.num_present_cpu)
 		phba->cfg_hdw_queue = phba->sli4_hba.num_present_cpu;
+	if (phba->cfg_irq_chann > phba->sli4_hba.num_present_cpu)
+		phba->cfg_irq_chann = phba->sli4_hba.num_present_cpu;
+	if (phba->cfg_irq_chann > phba->cfg_hdw_queue)
+		phba->cfg_irq_chann = phba->cfg_hdw_queue;
 
 	if (phba->cfg_enable_fc4_type & LPFC_ENABLE_NVME &&
 	    phba->nvmet_support) {
@@ -6953,11 +6961,11 @@ lpfc_nvme_mod_param_dep(struct lpfc_hba *phba)
 		}
 
 		if (!phba->cfg_nvmet_mrq)
-			phba->cfg_nvmet_mrq = phba->cfg_hdw_queue;
+			phba->cfg_nvmet_mrq = phba->cfg_irq_chann;
 
 		/* Adjust lpfc_nvmet_mrq to avoid running out of WQE slots */
-		if (phba->cfg_nvmet_mrq > phba->cfg_hdw_queue) {
-			phba->cfg_nvmet_mrq = phba->cfg_hdw_queue;
+		if (phba->cfg_nvmet_mrq > phba->cfg_irq_chann) {
+			phba->cfg_nvmet_mrq = phba->cfg_irq_chann;
 			lpfc_printf_log(phba, KERN_ERR, LOG_NVME_DISC,
 					"6018 Adjust lpfc_nvmet_mrq to %d\n",
 					phba->cfg_nvmet_mrq);
diff --git a/drivers/scsi/lpfc/lpfc_crtn.h b/drivers/scsi/lpfc/lpfc_crtn.h
index 726cd6a7c452..982401c31c12 100644
--- a/drivers/scsi/lpfc/lpfc_crtn.h
+++ b/drivers/scsi/lpfc/lpfc_crtn.h
@@ -440,7 +440,6 @@ extern spinlock_t _dump_buf_lock;
 extern int _dump_buf_done;
 extern spinlock_t pgcnt_lock;
 extern unsigned int pgcnt;
-extern unsigned int lpfc_fcp_look_ahead;
 
 /* Interface exported by fabric iocb scheduler */
 void lpfc_fabric_abort_nport(struct lpfc_nodelist *);
diff --git a/drivers/scsi/lpfc/lpfc_debugfs.c b/drivers/scsi/lpfc/lpfc_debugfs.c
index 2a2c46766eb6..72076b2cd4ff 100644
--- a/drivers/scsi/lpfc/lpfc_debugfs.c
+++ b/drivers/scsi/lpfc/lpfc_debugfs.c
@@ -378,6 +378,67 @@ skipit:
 	return len;
 }
 
+static int lpfc_debugfs_last_xripool;
+
+/**
+ * lpfc_debugfs_common_xri_data - Dump Hardware Queue info to a buffer
+ * @phba: The HBA to gather host buffer info from.
+ * @buf: The buffer to dump log into.
+ * @size: The maximum amount of data to process.
+ *
+ * Description:
+ * This routine dumps the Hardware Queue info from the @phba to @buf up to
+ * @size number of bytes. A header that describes the current hdwq state will be
+ * dumped to @buf first and then info on each hdwq entry will be dumped to @buf
+ * until @size bytes have been dumped or all the hdwq info has been dumped.
+ *
+ * Notes:
+ * This routine will rotate through each configured Hardware Queue each
+ * time called.
+ *
+ * Return Value:
+ * This routine returns the amount of bytes that were dumped into @buf and will
+ * not exceed @size.
+ **/
+static int
+lpfc_debugfs_commonxripools_data(struct lpfc_hba *phba, char *buf, int size)
+{
+	struct lpfc_sli4_hdw_queue *qp;
+	int len = 0;
+	int i, out;
+	unsigned long iflag;
+
+	for (i = 0; i < phba->cfg_hdw_queue; i++) {
+		if (len > (LPFC_DUMP_MULTIXRIPOOL_SIZE - 80))
+			break;
+		qp = &phba->sli4_hba.hdwq[lpfc_debugfs_last_xripool];
+
+		len +=  snprintf(buf + len, size - len, "HdwQ %d Info ", i);
+		spin_lock_irqsave(&qp->abts_scsi_buf_list_lock, iflag);
+		spin_lock(&qp->abts_nvme_buf_list_lock);
+		spin_lock(&qp->io_buf_list_get_lock);
+		spin_lock(&qp->io_buf_list_put_lock);
+		out = qp->total_io_bufs - (qp->get_io_bufs + qp->put_io_bufs +
+			qp->abts_scsi_io_bufs + qp->abts_nvme_io_bufs);
+		len +=  snprintf(buf + len, size - len,
+				 "tot:%d get:%d put:%d mt:%d "
+				 "ABTS scsi:%d nvme:%d Out:%d\n",
+			qp->total_io_bufs, qp->get_io_bufs, qp->put_io_bufs,
+			qp->empty_io_bufs, qp->abts_scsi_io_bufs,
+			qp->abts_nvme_io_bufs, out);
+		spin_unlock(&qp->io_buf_list_put_lock);
+		spin_unlock(&qp->io_buf_list_get_lock);
+		spin_unlock(&qp->abts_nvme_buf_list_lock);
+		spin_unlock_irqrestore(&qp->abts_scsi_buf_list_lock, iflag);
+
+		lpfc_debugfs_last_xripool++;
+		if (lpfc_debugfs_last_xripool >= phba->cfg_hdw_queue)
+			lpfc_debugfs_last_xripool = 0;
+	}
+
+	return len;
+}
+
 /**
  * lpfc_debugfs_multixripools_data - Display multi-XRI pools information
  * @phba: The HBA to gather host buffer info from.
@@ -405,6 +466,17 @@ lpfc_debugfs_multixripools_data(struct lpfc_hba *phba, char *buf, int size)
 	u32 txcmplq_cnt;
 	char tmp[LPFC_DEBUG_OUT_LINE_SZ] = {0};
 
+	if (phba->sli_rev != LPFC_SLI_REV4)
+		return 0;
+
+	if (!phba->sli4_hba.hdwq)
+		return 0;
+
+	if (!phba->cfg_xri_rebalancing) {
+		i = lpfc_debugfs_commonxripools_data(phba, buf, size);
+		return i;
+	}
+
 	/*
 	 * Pbl: Current number of free XRIs in public pool
 	 * Pvt: Current number of free XRIs in private pool
@@ -498,10 +570,12 @@ lpfc_debugfs_multixripools_data(struct lpfc_hba *phba, char *buf, int size)
 	return strnlen(buf, size);
 }
 
-static int lpfc_debugfs_last_hdwq;
+
+#ifdef LPFC_HDWQ_LOCK_STAT
+static int lpfc_debugfs_last_lock;
 
 /**
- * lpfc_debugfs_hdwqinfo_data - Dump Hardware Queue info to a buffer
+ * lpfc_debugfs_lockstat_data - Dump Hardware Queue info to a buffer
  * @phba: The HBA to gather host buffer info from.
  * @buf: The buffer to dump log into.
  * @size: The maximum amount of data to process.
@@ -521,12 +595,11 @@ static int lpfc_debugfs_last_hdwq;
  * not exceed @size.
  **/
 static int
-lpfc_debugfs_hdwqinfo_data(struct lpfc_hba *phba, char *buf, int size)
+lpfc_debugfs_lockstat_data(struct lpfc_hba *phba, char *buf, int size)
 {
 	struct lpfc_sli4_hdw_queue *qp;
 	int len = 0;
-	int i, out;
-	unsigned long iflag;
+	int i;
 
 	if (phba->sli_rev != LPFC_SLI_REV4)
 		return 0;
@@ -535,35 +608,40 @@ lpfc_debugfs_hdwqinfo_data(struct lpfc_hba *phba, char *buf, int size)
 		return 0;
 
 	for (i = 0; i < phba->cfg_hdw_queue; i++) {
-		if (len > (LPFC_HDWQINFO_SIZE - 80))
+		if (len > (LPFC_HDWQINFO_SIZE - 100))
 			break;
-		qp = &phba->sli4_hba.hdwq[lpfc_debugfs_last_hdwq];
+		qp = &phba->sli4_hba.hdwq[lpfc_debugfs_last_lock];
 
-		len +=  snprintf(buf + len, size - len, "HdwQ %d Info ", i);
-		spin_lock_irqsave(&qp->abts_scsi_buf_list_lock, iflag);
-		spin_lock(&qp->abts_nvme_buf_list_lock);
-		spin_lock(&qp->io_buf_list_get_lock);
-		spin_lock(&qp->io_buf_list_put_lock);
-		out = qp->total_io_bufs - (qp->get_io_bufs + qp->put_io_bufs +
-			qp->abts_scsi_io_bufs + qp->abts_nvme_io_bufs);
-		len +=  snprintf(buf + len, size - len,
-				 "tot:%d get:%d put:%d mt:%d "
-				 "ABTS scsi:%d nvme:%d Out:%d\n",
-			qp->total_io_bufs, qp->get_io_bufs, qp->put_io_bufs,
-			qp->empty_io_bufs, qp->abts_scsi_io_bufs,
-			qp->abts_nvme_io_bufs, out);
-		spin_unlock(&qp->io_buf_list_put_lock);
-		spin_unlock(&qp->io_buf_list_get_lock);
-		spin_unlock(&qp->abts_nvme_buf_list_lock);
-		spin_unlock_irqrestore(&qp->abts_scsi_buf_list_lock, iflag);
+		len +=  snprintf(buf + len, size - len, "HdwQ %03d Lock ", i);
+		if (phba->cfg_xri_rebalancing) {
+			len +=  snprintf(buf + len, size - len,
+					 "get_pvt:%d mv_pvt:%d "
+					 "mv2pub:%d mv2pvt:%d "
+					 "put_pvt:%d put_pub:%d wq:%d\n",
+					 qp->lock_conflict.alloc_pvt_pool,
+					 qp->lock_conflict.mv_from_pvt_pool,
+					 qp->lock_conflict.mv_to_pub_pool,
+					 qp->lock_conflict.mv_to_pvt_pool,
+					 qp->lock_conflict.free_pvt_pool,
+					 qp->lock_conflict.free_pub_pool,
+					 qp->lock_conflict.wq_access);
+		} else {
+			len +=  snprintf(buf + len, size - len,
+					 "get:%d put:%d free:%d wq:%d\n",
+					 qp->lock_conflict.alloc_xri_get,
+					 qp->lock_conflict.alloc_xri_put,
+					 qp->lock_conflict.free_xri,
+					 qp->lock_conflict.wq_access);
+		}
 
-		lpfc_debugfs_last_hdwq++;
-		if (lpfc_debugfs_last_hdwq >= phba->cfg_hdw_queue)
-			lpfc_debugfs_last_hdwq = 0;
+		lpfc_debugfs_last_lock++;
+		if (lpfc_debugfs_last_lock >= phba->cfg_hdw_queue)
+			lpfc_debugfs_last_lock = 0;
 	}
 
 	return len;
 }
+#endif
 
 static int lpfc_debugfs_last_hba_slim_off;
 
@@ -964,7 +1042,7 @@ lpfc_debugfs_nvmestat_data(struct lpfc_vport *vport, char *buf, int size)
 	struct lpfc_nvme_lport *lport;
 	uint64_t data1, data2, data3;
 	uint64_t tot, totin, totout;
-	int cnt, i, maxch;
+	int cnt, i;
 	int len = 0;
 
 	if (phba->nvmet_support) {
@@ -1106,10 +1184,6 @@ lpfc_debugfs_nvmestat_data(struct lpfc_vport *vport, char *buf, int size)
 				atomic_read(&lport->fc4NvmeLsRequests),
 				atomic_read(&lport->fc4NvmeLsCmpls));
 
-		if (phba->cfg_hdw_queue < LPFC_HBA_HDWQ_MAX)
-			maxch = phba->cfg_hdw_queue;
-		else
-			maxch = LPFC_HBA_HDWQ_MAX;
 		totin = 0;
 		totout = 0;
 		for (i = 0; i < phba->cfg_hdw_queue; i++) {
@@ -1547,7 +1621,7 @@ lpfc_debugfs_cpucheck_data(struct lpfc_vport *vport, char *buf, int size)
 {
 	struct lpfc_hba   *phba = vport->phba;
 	struct lpfc_sli4_hdw_queue *qp;
-	int i, j;
+	int i, j, max_cnt;
 	int len = 0;
 	uint32_t tot_xmt;
 	uint32_t tot_rcv;
@@ -1565,6 +1639,7 @@ lpfc_debugfs_cpucheck_data(struct lpfc_vport *vport, char *buf, int size)
 	} else {
 		len += snprintf(buf + len, PAGE_SIZE - len, "\n");
 	}
+	max_cnt = size - LPFC_DEBUG_OUT_LINE_SZ;
 
 	for (i = 0; i < phba->cfg_hdw_queue; i++) {
 		qp = &phba->sli4_hba.hdwq[i];
@@ -1606,6 +1681,11 @@ lpfc_debugfs_cpucheck_data(struct lpfc_vport *vport, char *buf, int size)
 		}
 		len += snprintf(buf + len, PAGE_SIZE - len,
 				"Total: %x\n", tot_xmt);
+		if (len >= max_cnt) {
+			len += snprintf(buf + len, PAGE_SIZE - len,
+					"Truncated ...\n");
+			return len;
+		}
 	}
 	return len;
 }
@@ -1904,11 +1984,8 @@ lpfc_debugfs_multixripools_open(struct inode *inode, struct file *file)
 		goto out;
 	}
 
-	if (phba->cfg_xri_rebalancing)
-		debug->len = lpfc_debugfs_multixripools_data(
-			phba, debug->buffer, LPFC_DUMP_MULTIXRIPOOL_SIZE);
-	else
-		debug->len = 0;
+	debug->len = lpfc_debugfs_multixripools_data(
+		phba, debug->buffer, LPFC_DUMP_MULTIXRIPOOL_SIZE);
 
 	debug->i_private = inode->i_private;
 	file->private_data = debug;
@@ -1918,8 +1995,9 @@ out:
 	return rc;
 }
 
+#ifdef LPFC_HDWQ_LOCK_STAT
 /**
- * lpfc_debugfs_hdwqinfo_open - Open the hdwqinfo debugfs buffer
+ * lpfc_debugfs_lockstat_open - Open the lockstat debugfs buffer
  * @inode: The inode pointer that contains a vport pointer.
  * @file: The file pointer to attach the log output.
  *
@@ -1934,7 +2012,7 @@ out:
  * error value.
  **/
 static int
-lpfc_debugfs_hdwqinfo_open(struct inode *inode, struct file *file)
+lpfc_debugfs_lockstat_open(struct inode *inode, struct file *file)
 {
 	struct lpfc_hba *phba = inode->i_private;
 	struct lpfc_debug *debug;
@@ -1951,7 +2029,7 @@ lpfc_debugfs_hdwqinfo_open(struct inode *inode, struct file *file)
 		goto out;
 	}
 
-	debug->len = lpfc_debugfs_hdwqinfo_data(phba, debug->buffer,
+	debug->len = lpfc_debugfs_lockstat_data(phba, debug->buffer,
 		LPFC_HBQINFO_SIZE);
 	file->private_data = debug;
 
@@ -1960,6 +2038,48 @@ out:
 	return rc;
 }
 
+static ssize_t
+lpfc_debugfs_lockstat_write(struct file *file, const char __user *buf,
+			    size_t nbytes, loff_t *ppos)
+{
+	struct lpfc_debug *debug = file->private_data;
+	struct lpfc_hba *phba = (struct lpfc_hba *)debug->i_private;
+	struct lpfc_sli4_hdw_queue *qp;
+	char mybuf[64];
+	char *pbuf;
+	int i;
+
+	/* Protect copy from user */
+	if (!access_ok(buf, nbytes))
+		return -EFAULT;
+
+	memset(mybuf, 0, sizeof(mybuf));
+
+	if (copy_from_user(mybuf, buf, nbytes))
+		return -EFAULT;
+	pbuf = &mybuf[0];
+
+	if ((strncmp(pbuf, "reset", strlen("reset")) == 0) ||
+	    (strncmp(pbuf, "zero", strlen("zero")) == 0)) {
+		for (i = 0; i < phba->cfg_hdw_queue; i++) {
+			qp = &phba->sli4_hba.hdwq[i];
+			qp->lock_conflict.alloc_xri_get = 0;
+			qp->lock_conflict.alloc_xri_put = 0;
+			qp->lock_conflict.free_xri = 0;
+			qp->lock_conflict.wq_access = 0;
+			qp->lock_conflict.alloc_pvt_pool = 0;
+			qp->lock_conflict.mv_from_pvt_pool = 0;
+			qp->lock_conflict.mv_to_pub_pool = 0;
+			qp->lock_conflict.mv_to_pvt_pool = 0;
+			qp->lock_conflict.free_pvt_pool = 0;
+			qp->lock_conflict.free_pub_pool = 0;
+			qp->lock_conflict.wq_access = 0;
+		}
+	}
+	return nbytes;
+}
+#endif
+
 /**
  * lpfc_debugfs_dumpHBASlim_open - Open the Dump HBA SLIM debugfs buffer
  * @inode: The inode pointer that contains a vport pointer.
@@ -2816,7 +2936,7 @@ lpfc_debugfs_cpucheck_open(struct inode *inode, struct file *file)
 	}
 
 	debug->len = lpfc_debugfs_cpucheck_data(vport, debug->buffer,
-		LPFC_NVMEKTIME_SIZE);
+		LPFC_CPUCHECK_SIZE);
 
 	debug->i_private = inode->i_private;
 	file->private_data = debug;
@@ -2851,8 +2971,18 @@ lpfc_debugfs_cpucheck_write(struct file *file, const char __user *buf,
 		if (phba->nvmet_support)
 			phba->cpucheck_on |= LPFC_CHECK_NVMET_IO;
 		else
+			phba->cpucheck_on |= (LPFC_CHECK_NVME_IO |
+				LPFC_CHECK_SCSI_IO);
+		return strlen(pbuf);
+	} else if ((strncmp(pbuf, "nvme_on", sizeof("nvme_on") - 1) == 0)) {
+		if (phba->nvmet_support)
+			phba->cpucheck_on |= LPFC_CHECK_NVMET_IO;
+		else
 			phba->cpucheck_on |= LPFC_CHECK_NVME_IO;
 		return strlen(pbuf);
+	} else if ((strncmp(pbuf, "scsi_on", sizeof("scsi_on") - 1) == 0)) {
+		phba->cpucheck_on |= LPFC_CHECK_SCSI_IO;
+		return strlen(pbuf);
 	} else if ((strncmp(pbuf, "rcv",
 		   sizeof("rcv") - 1) == 0)) {
 		if (phba->nvmet_support)
@@ -3732,46 +3862,38 @@ lpfc_idiag_cqs_for_eq(struct lpfc_hba *phba, char *pbuffer,
 		int *len, int max_cnt, int eqidx, int eq_id)
 {
 	struct lpfc_queue *qp;
-	int qidx, rc;
+	int rc;
 
-	for (qidx = 0; qidx < phba->cfg_hdw_queue; qidx++) {
-		qp = phba->sli4_hba.hdwq[qidx].fcp_cq;
-		if (qp->assoc_qid != eq_id)
-			continue;
+	qp = phba->sli4_hba.hdwq[eqidx].fcp_cq;
 
-		*len = __lpfc_idiag_print_cq(qp, "FCP", pbuffer, *len);
+	*len = __lpfc_idiag_print_cq(qp, "FCP", pbuffer, *len);
 
-		/* Reset max counter */
-		qp->CQ_max_cqe = 0;
+	/* Reset max counter */
+	qp->CQ_max_cqe = 0;
 
-		if (*len >= max_cnt)
-			return 1;
+	if (*len >= max_cnt)
+		return 1;
 
-		rc = lpfc_idiag_wqs_for_cq(phba, "FCP", pbuffer, len,
-				max_cnt, qp->queue_id);
-		if (rc)
-			return 1;
-	}
+	rc = lpfc_idiag_wqs_for_cq(phba, "FCP", pbuffer, len,
+				   max_cnt, qp->queue_id);
+	if (rc)
+		return 1;
 
 	if (phba->cfg_enable_fc4_type & LPFC_ENABLE_NVME) {
-		for (qidx = 0; qidx < phba->cfg_hdw_queue; qidx++) {
-			qp = phba->sli4_hba.hdwq[qidx].nvme_cq;
-			if (qp->assoc_qid != eq_id)
-				continue;
+		qp = phba->sli4_hba.hdwq[eqidx].nvme_cq;
 
-			*len = __lpfc_idiag_print_cq(qp, "NVME", pbuffer, *len);
+		*len = __lpfc_idiag_print_cq(qp, "NVME", pbuffer, *len);
 
-			/* Reset max counter */
-			qp->CQ_max_cqe = 0;
+		/* Reset max counter */
+		qp->CQ_max_cqe = 0;
 
-			if (*len >= max_cnt)
-				return 1;
+		if (*len >= max_cnt)
+			return 1;
 
-			rc = lpfc_idiag_wqs_for_cq(phba, "NVME", pbuffer, len,
-						   max_cnt, qp->queue_id);
-			if (rc)
-				return 1;
-		}
+		rc = lpfc_idiag_wqs_for_cq(phba, "NVME", pbuffer, len,
+					   max_cnt, qp->queue_id);
+		if (rc)
+			return 1;
 	}
 
 	if ((eqidx < phba->cfg_nvmet_mrq) && phba->nvmet_support) {
@@ -3812,9 +3934,10 @@ __lpfc_idiag_print_eq(struct lpfc_queue *qp, char *eqtype,
 			(unsigned long long)qp->q_cnt_4, qp->q_mode);
 	len += snprintf(pbuffer + len, LPFC_QUE_INFO_GET_BUF_SIZE - len,
 			"EQID[%02d], QE-CNT[%04d], QE-SZ[%04d], "
-			"HST-IDX[%04d], PRT-IDX[%04d], PST[%03d]",
+			"HST-IDX[%04d], PRT-IDX[%04d], PST[%03d] AFFIN[%03d]",
 			qp->queue_id, qp->entry_count, qp->entry_size,
-			qp->host_index, qp->hba_index, qp->entry_repost);
+			qp->host_index, qp->hba_index, qp->entry_repost,
+			qp->chann);
 	len +=  snprintf(pbuffer + len, LPFC_QUE_INFO_GET_BUF_SIZE - len, "\n");
 
 	return len;
@@ -3869,7 +3992,7 @@ lpfc_idiag_queinfo_read(struct file *file, char __user *buf, size_t nbytes,
 			phba->lpfc_idiag_last_eq = 0;
 
 		len += snprintf(pbuffer + len, LPFC_QUE_INFO_GET_BUF_SIZE - len,
-					"EQ %d out of %d HBA EQs\n",
+					"HDWQ %d out of %d HBA HDWQs\n",
 					x, phba->cfg_hdw_queue);
 
 		/* Fast-path EQ */
@@ -5299,14 +5422,17 @@ static const struct file_operations lpfc_debugfs_op_hbqinfo = {
 	.release =      lpfc_debugfs_release,
 };
 
-#undef lpfc_debugfs_op_hdwqinfo
-static const struct file_operations lpfc_debugfs_op_hdwqinfo = {
+#ifdef LPFC_HDWQ_LOCK_STAT
+#undef lpfc_debugfs_op_lockstat
+static const struct file_operations lpfc_debugfs_op_lockstat = {
 	.owner =        THIS_MODULE,
-	.open =         lpfc_debugfs_hdwqinfo_open,
+	.open =         lpfc_debugfs_lockstat_open,
 	.llseek =       lpfc_debugfs_lseek,
 	.read =         lpfc_debugfs_read,
+	.write =        lpfc_debugfs_lockstat_write,
 	.release =      lpfc_debugfs_release,
 };
+#endif
 
 #undef lpfc_debugfs_op_dumpHBASlim
 static const struct file_operations lpfc_debugfs_op_dumpHBASlim = {
@@ -5756,17 +5882,19 @@ lpfc_debugfs_initialize(struct lpfc_vport *vport)
 					    phba->hba_debugfs_root,
 					    phba, &lpfc_debugfs_op_hbqinfo);
 
-		/* Setup hdwqinfo */
-		snprintf(name, sizeof(name), "hdwqinfo");
-		phba->debug_hdwqinfo =
+#ifdef LPFC_HDWQ_LOCK_STAT
+		/* Setup lockstat */
+		snprintf(name, sizeof(name), "lockstat");
+		phba->debug_lockstat =
 			debugfs_create_file(name, S_IFREG | 0644,
 					    phba->hba_debugfs_root,
-					    phba, &lpfc_debugfs_op_hdwqinfo);
-		if (!phba->debug_hdwqinfo) {
+					    phba, &lpfc_debugfs_op_lockstat);
+		if (!phba->debug_lockstat) {
 			lpfc_printf_vlog(vport, KERN_ERR, LOG_INIT,
-					 "0511 Cant create debugfs hdwqinfo\n");
+					 "0913 Cant create debugfs lockstat\n");
 			goto debug_failed;
 		}
+#endif
 
 		/* Setup dumpHBASlim */
 		if (phba->sli_rev < LPFC_SLI_REV4) {
@@ -6006,7 +6134,7 @@ nvmeio_off:
 				    vport, &lpfc_debugfs_op_scsistat);
 	if (!vport->debug_scsistat) {
 		lpfc_printf_vlog(vport, KERN_ERR, LOG_INIT,
-				 "0811 Cannot create debugfs scsistat\n");
+				 "0914 Cannot create debugfs scsistat\n");
 		goto debug_failed;
 	}
 
@@ -6171,9 +6299,10 @@ lpfc_debugfs_terminate(struct lpfc_vport *vport)
 		debugfs_remove(phba->debug_hbqinfo); /* hbqinfo */
 		phba->debug_hbqinfo = NULL;
 
-		debugfs_remove(phba->debug_hdwqinfo); /* hdwqinfo */
-		phba->debug_hdwqinfo = NULL;
-
+#ifdef LPFC_HDWQ_LOCK_STAT
+		debugfs_remove(phba->debug_lockstat); /* lockstat */
+		phba->debug_lockstat = NULL;
+#endif
 		debugfs_remove(phba->debug_dumpHBASlim); /* HBASlim */
 		phba->debug_dumpHBASlim = NULL;
 
diff --git a/drivers/scsi/lpfc/lpfc_debugfs.h b/drivers/scsi/lpfc/lpfc_debugfs.h
index cf256a6dca42..1fbee6496f85 100644
--- a/drivers/scsi/lpfc/lpfc_debugfs.h
+++ b/drivers/scsi/lpfc/lpfc_debugfs.h
@@ -290,9 +290,6 @@ struct lpfc_idiag {
 /* multixripool output buffer size */
 #define LPFC_DUMP_MULTIXRIPOOL_SIZE 8192
 
-/* hdwqinfo output buffer size */
-#define LPFC_HDWQINFO_SIZE 8192
-
 enum {
 	DUMP_FCP,
 	DUMP_NVME,
diff --git a/drivers/scsi/lpfc/lpfc_hw4.h b/drivers/scsi/lpfc/lpfc_hw4.h
index cd39845c909f..665852291a4f 100644
--- a/drivers/scsi/lpfc/lpfc_hw4.h
+++ b/drivers/scsi/lpfc/lpfc_hw4.h
@@ -211,9 +211,8 @@ struct lpfc_sli_intf {
 #define LPFC_DEF_IMAX          150000
 
 #define LPFC_MIN_CPU_MAP       0
-#define LPFC_MAX_CPU_MAP       2
+#define LPFC_MAX_CPU_MAP       1
 #define LPFC_HBA_CPU_MAP       1
-#define LPFC_DRIVER_CPU_MAP    2  /* Default */
 
 /* PORT_CAPABILITIES constants. */
 #define LPFC_MAX_SUPPORTED_PAGES	8
diff --git a/drivers/scsi/lpfc/lpfc_init.c b/drivers/scsi/lpfc/lpfc_init.c
index d9db29817f6b..145c08f112a3 100644
--- a/drivers/scsi/lpfc/lpfc_init.c
+++ b/drivers/scsi/lpfc/lpfc_init.c
@@ -37,6 +37,7 @@
 #include <linux/miscdevice.h>
 #include <linux/percpu.h>
 #include <linux/msi.h>
+#include <linux/irq.h>
 #include <linux/bitops.h>
 
 #include <scsi/scsi.h>
@@ -92,6 +93,8 @@ static void lpfc_sli4_cq_event_release_all(struct lpfc_hba *);
 static void lpfc_sli4_disable_intr(struct lpfc_hba *);
 static uint32_t lpfc_sli4_enable_intr(struct lpfc_hba *, uint32_t);
 static void lpfc_sli4_oas_verify(struct lpfc_hba *phba);
+static uint16_t lpfc_find_eq_handle(struct lpfc_hba *, uint16_t);
+static uint16_t lpfc_find_cpu_handle(struct lpfc_hba *, uint16_t, int);
 
 static struct scsi_transport_template *lpfc_transport_template = NULL;
 static struct scsi_transport_template *lpfc_vport_transport_template = NULL;
@@ -1367,13 +1370,13 @@ lpfc_hb_timeout_handler(struct lpfc_hba *phba)
 		}
 
 		/* Interrupts per sec per EQ */
-		val = phba->cfg_fcp_imax / phba->cfg_hdw_queue;
+		val = phba->cfg_fcp_imax / phba->cfg_irq_chann;
 		tick_cqe = val / CONFIG_HZ; /* Per tick per EQ */
 
 		/* Assume 1 CQE/ISR, calc max CQEs allowed for time duration */
 		max_cqe = time_elapsed * tick_cqe;
 
-		for (i = 0; i < phba->cfg_hdw_queue; i++) {
+		for (i = 0; i < phba->cfg_irq_chann; i++) {
 			/* Fast-path EQ */
 			qp = phba->sli4_hba.hdwq[i].hba_eq;
 			if (!qp)
@@ -1397,7 +1400,7 @@ lpfc_hb_timeout_handler(struct lpfc_hba *phba)
 				if (val) {
 					/* First, interrupts per sec per EQ */
 					val = phba->cfg_fcp_imax /
-						phba->cfg_hdw_queue;
+						phba->cfg_irq_chann;
 
 					/* us delay between each interrupt */
 					val = LPFC_SEC_TO_USEC / val;
@@ -4335,8 +4338,13 @@ lpfc_create_port(struct lpfc_hba *phba, int instance, struct device *dev)
 	shost->max_lun = vport->cfg_max_luns;
 	shost->this_id = -1;
 	shost->max_cmd_len = 16;
+
 	if (phba->sli_rev == LPFC_SLI_REV4) {
-		shost->nr_hw_queues = phba->cfg_hdw_queue;
+		if (phba->cfg_fcp_io_sched == LPFC_FCP_SCHED_BY_HDWQ)
+			shost->nr_hw_queues = phba->cfg_hdw_queue;
+		else
+			shost->nr_hw_queues = phba->sli4_hba.num_present_cpu;
+
 		shost->dma_boundary =
 			phba->sli4_hba.pc_sli4_params.sge_supp_len-1;
 		shost->sg_tablesize = phba->cfg_scsi_seg_cnt;
@@ -6819,7 +6827,7 @@ lpfc_sli4_driver_resource_setup(struct lpfc_hba *phba)
 		goto out_remove_rpi_hdrs;
 	}
 
-	phba->sli4_hba.hba_eq_hdl = kcalloc(phba->cfg_hdw_queue,
+	phba->sli4_hba.hba_eq_hdl = kcalloc(phba->cfg_irq_chann,
 					    sizeof(struct lpfc_hba_eq_hdl),
 					    GFP_KERNEL);
 	if (!phba->sli4_hba.hba_eq_hdl) {
@@ -8257,7 +8265,7 @@ lpfc_sli4_read_config(struct lpfc_hba *phba)
 	struct lpfc_rsrc_desc_fcfcoe *desc;
 	char *pdesc_0;
 	uint16_t forced_link_speed;
-	uint32_t if_type;
+	uint32_t if_type, qmin;
 	int length, i, rc = 0, rc2;
 
 	pmb = (LPFC_MBOXQ_t *) mempool_alloc(phba->mbox_mem_pool, GFP_KERNEL);
@@ -8362,40 +8370,44 @@ lpfc_sli4_read_config(struct lpfc_hba *phba)
 				phba->sli4_hba.max_cfg_param.max_rq);
 
 		/*
-		 * Calculate NVME queue resources based on how
-		 * many WQ/CQs are available.
+		 * Calculate queue resources based on how
+		 * many WQ/CQ/EQs are available.
 		 */
-		if (phba->cfg_enable_fc4_type & LPFC_ENABLE_NVME) {
-			length = phba->sli4_hba.max_cfg_param.max_wq;
-			if (phba->sli4_hba.max_cfg_param.max_cq <
-			    phba->sli4_hba.max_cfg_param.max_wq)
-				length = phba->sli4_hba.max_cfg_param.max_cq;
+		qmin = phba->sli4_hba.max_cfg_param.max_wq;
+		if (phba->sli4_hba.max_cfg_param.max_cq < qmin)
+			qmin = phba->sli4_hba.max_cfg_param.max_cq;
+		if (phba->sli4_hba.max_cfg_param.max_eq < qmin)
+			qmin = phba->sli4_hba.max_cfg_param.max_eq;
+		/*
+		 * Whats left after this can go toward NVME / FCP.
+		 * The minus 4 accounts for ELS, NVME LS, MBOX
+		 * plus one extra. When configured for
+		 * NVMET, FCP io channel WQs are not created.
+		 */
+		qmin -= 4;
 
-			/*
-			 * Whats left after this can go toward NVME.
-			 * The minus 6 accounts for ELS, NVME LS, MBOX
-			 * plus a couple extra. When configured for
-			 * NVMET, FCP io channel WQs are not created.
-			 */
-			length -= 6;
-
-			/* Take off FCP queues */
-			if (!phba->nvmet_support)
-				length -= phba->cfg_hdw_queue;
-
-			/* Check to see if there is enough for NVME */
-			if (phba->cfg_hdw_queue > length) {
-				lpfc_printf_log(
-					phba, KERN_ERR, LOG_SLI,
-					"2005 Reducing NVME IO channel to %d: "
-					"WQ %d CQ %d CommonIO %d\n",
-					length,
+		/* If NVME is configured, double the number of CQ/WQs needed */
+		if ((phba->cfg_enable_fc4_type & LPFC_ENABLE_NVME) &&
+		    !phba->nvmet_support)
+			qmin /= 2;
+
+		/* Check to see if there is enough for NVME */
+		if ((phba->cfg_irq_chann > qmin) ||
+		    (phba->cfg_hdw_queue > qmin)) {
+			lpfc_printf_log(phba, KERN_ERR, LOG_SLI,
+					"2005 Reducing Queues: "
+					"WQ %d CQ %d EQ %d: min %d: "
+					"IRQ %d HDWQ %d\n",
 					phba->sli4_hba.max_cfg_param.max_wq,
 					phba->sli4_hba.max_cfg_param.max_cq,
+					phba->sli4_hba.max_cfg_param.max_eq,
+					qmin, phba->cfg_irq_chann,
 					phba->cfg_hdw_queue);
 
-				phba->cfg_hdw_queue = length;
-			}
+			if (phba->cfg_irq_chann > qmin)
+				phba->cfg_irq_chann = qmin;
+			if (phba->cfg_hdw_queue > qmin)
+				phba->cfg_hdw_queue = qmin;
 		}
 	}
 
@@ -8612,25 +8624,17 @@ lpfc_sli4_queue_verify(struct lpfc_hba *phba)
 	 * device parameters
 	 */
 
-	if (phba->cfg_hdw_queue > phba->sli4_hba.max_cfg_param.max_eq) {
-		lpfc_printf_log(phba, KERN_ERR, LOG_INIT,
-				"2575 Reducing IO channels to match number of "
-				"available EQs: from %d to %d\n",
-				phba->cfg_hdw_queue,
-				phba->sli4_hba.max_cfg_param.max_eq);
-		phba->cfg_hdw_queue = phba->sli4_hba.max_cfg_param.max_eq;
-	}
-
 	if (phba->nvmet_support) {
-		if (phba->cfg_hdw_queue < phba->cfg_nvmet_mrq)
-			phba->cfg_nvmet_mrq = phba->cfg_hdw_queue;
+		if (phba->cfg_irq_chann < phba->cfg_nvmet_mrq)
+			phba->cfg_nvmet_mrq = phba->cfg_irq_chann;
 	}
 	if (phba->cfg_nvmet_mrq > LPFC_NVMET_MRQ_MAX)
 		phba->cfg_nvmet_mrq = LPFC_NVMET_MRQ_MAX;
 
 	lpfc_printf_log(phba, KERN_ERR, LOG_INIT,
-			"2574 IO channels: hdwQ %d MRQ: %d\n",
-			phba->cfg_hdw_queue, phba->cfg_nvmet_mrq);
+			"2574 IO channels: hdwQ %d IRQ %d MRQ: %d\n",
+			phba->cfg_hdw_queue, phba->cfg_irq_chann,
+			phba->cfg_nvmet_mrq);
 
 	/* Get EQ depth from module parameter, fake the default for now */
 	phba->sli4_hba.eq_esize = LPFC_EQE_SIZE_4B;
@@ -8658,6 +8662,7 @@ lpfc_alloc_nvme_wq_cq(struct lpfc_hba *phba, int wqidx)
 	}
 	qdesc->qe_valid = 1;
 	qdesc->hdwq = wqidx;
+	qdesc->chann = lpfc_find_cpu_handle(phba, wqidx, LPFC_FIND_BY_HDWQ);
 	phba->sli4_hba.hdwq[wqidx].nvme_cq = qdesc;
 
 	qdesc = lpfc_sli4_queue_alloc(phba, LPFC_EXPANDED_PAGE_SIZE,
@@ -8669,6 +8674,7 @@ lpfc_alloc_nvme_wq_cq(struct lpfc_hba *phba, int wqidx)
 		return 1;
 	}
 	qdesc->hdwq = wqidx;
+	qdesc->chann = wqidx;
 	phba->sli4_hba.hdwq[wqidx].nvme_wq = qdesc;
 	list_add_tail(&qdesc->wq_list, &phba->sli4_hba.lpfc_wq_list);
 	return 0;
@@ -8698,6 +8704,7 @@ lpfc_alloc_fcp_wq_cq(struct lpfc_hba *phba, int wqidx)
 	}
 	qdesc->qe_valid = 1;
 	qdesc->hdwq = wqidx;
+	qdesc->chann = lpfc_find_cpu_handle(phba, wqidx, LPFC_FIND_BY_HDWQ);
 	phba->sli4_hba.hdwq[wqidx].fcp_cq = qdesc;
 
 	/* Create Fast Path FCP WQs */
@@ -8720,6 +8727,7 @@ lpfc_alloc_fcp_wq_cq(struct lpfc_hba *phba, int wqidx)
 		return 1;
 	}
 	qdesc->hdwq = wqidx;
+	qdesc->chann = wqidx;
 	phba->sli4_hba.hdwq[wqidx].fcp_wq = qdesc;
 	list_add_tail(&qdesc->wq_list, &phba->sli4_hba.lpfc_wq_list);
 	return 0;
@@ -8743,7 +8751,7 @@ int
 lpfc_sli4_queue_create(struct lpfc_hba *phba)
 {
 	struct lpfc_queue *qdesc;
-	int idx;
+	int idx, eqidx;
 	struct lpfc_sli4_hdw_queue *qp;
 
 	/*
@@ -8829,7 +8837,18 @@ lpfc_sli4_queue_create(struct lpfc_hba *phba)
 
 	/* Create HBA Event Queues (EQs) */
 	for (idx = 0; idx < phba->cfg_hdw_queue; idx++) {
-		/* Create EQs */
+		/*
+		 * If there are more Hardware Queues than available
+		 * CQs, multiple Hardware Queues may share a common EQ.
+		 */
+		if (idx >= phba->cfg_irq_chann) {
+			/* Share an existing EQ */
+			eqidx = lpfc_find_eq_handle(phba, idx);
+			phba->sli4_hba.hdwq[idx].hba_eq =
+				phba->sli4_hba.hdwq[eqidx].hba_eq;
+			continue;
+		}
+		/* Create an EQ */
 		qdesc = lpfc_sli4_queue_alloc(phba, LPFC_DEFAULT_PAGE_SIZE,
 					      phba->sli4_hba.eq_esize,
 					      phba->sli4_hba.eq_ecount);
@@ -8840,20 +8859,27 @@ lpfc_sli4_queue_create(struct lpfc_hba *phba)
 		}
 		qdesc->qe_valid = 1;
 		qdesc->hdwq = idx;
+
+		/* Save the CPU this EQ is affinitised to */
+		eqidx = lpfc_find_eq_handle(phba, idx);
+		qdesc->chann = lpfc_find_cpu_handle(phba, eqidx,
+						    LPFC_FIND_BY_EQ);
 		phba->sli4_hba.hdwq[idx].hba_eq = qdesc;
 	}
 
 
 	/* Allocate SCSI SLI4 CQ/WQs */
-	for (idx = 0; idx < phba->cfg_hdw_queue; idx++)
+	for (idx = 0; idx < phba->cfg_hdw_queue; idx++) {
 		if (lpfc_alloc_fcp_wq_cq(phba, idx))
 			goto out_error;
+	}
 
 	/* Allocate NVME SLI4 CQ/WQs */
 	if (phba->cfg_enable_fc4_type & LPFC_ENABLE_NVME) {
-		for (idx = 0; idx < phba->cfg_hdw_queue; idx++)
+		for (idx = 0; idx < phba->cfg_hdw_queue; idx++) {
 			if (lpfc_alloc_nvme_wq_cq(phba, idx))
 				goto out_error;
+		}
 
 		if (phba->nvmet_support) {
 			for (idx = 0; idx < phba->cfg_nvmet_mrq; idx++) {
@@ -8871,6 +8897,7 @@ lpfc_sli4_queue_create(struct lpfc_hba *phba)
 				}
 				qdesc->qe_valid = 1;
 				qdesc->hdwq = idx;
+				qdesc->chann = idx;
 				phba->sli4_hba.nvmet_cqset[idx] = qdesc;
 			}
 		}
@@ -8902,6 +8929,7 @@ lpfc_sli4_queue_create(struct lpfc_hba *phba)
 		goto out_error;
 	}
 	qdesc->qe_valid = 1;
+	qdesc->chann = 0;
 	phba->sli4_hba.els_cq = qdesc;
 
 
@@ -8919,6 +8947,7 @@ lpfc_sli4_queue_create(struct lpfc_hba *phba)
 				"0505 Failed allocate slow-path MQ\n");
 		goto out_error;
 	}
+	qdesc->chann = 0;
 	phba->sli4_hba.mbx_wq = qdesc;
 
 	/*
@@ -8934,6 +8963,7 @@ lpfc_sli4_queue_create(struct lpfc_hba *phba)
 				"0504 Failed allocate slow-path ELS WQ\n");
 		goto out_error;
 	}
+	qdesc->chann = 0;
 	phba->sli4_hba.els_wq = qdesc;
 	list_add_tail(&qdesc->wq_list, &phba->sli4_hba.lpfc_wq_list);
 
@@ -8947,6 +8977,7 @@ lpfc_sli4_queue_create(struct lpfc_hba *phba)
 					"6079 Failed allocate NVME LS CQ\n");
 			goto out_error;
 		}
+		qdesc->chann = 0;
 		qdesc->qe_valid = 1;
 		phba->sli4_hba.nvmels_cq = qdesc;
 
@@ -8959,6 +8990,7 @@ lpfc_sli4_queue_create(struct lpfc_hba *phba)
 					"6080 Failed allocate NVME LS WQ\n");
 			goto out_error;
 		}
+		qdesc->chann = 0;
 		phba->sli4_hba.nvmels_wq = qdesc;
 		list_add_tail(&qdesc->wq_list, &phba->sli4_hba.lpfc_wq_list);
 	}
@@ -9085,17 +9117,21 @@ lpfc_sli4_release_queues(struct lpfc_queue ***qs, int max)
 }
 
 static inline void
-lpfc_sli4_release_hdwq(struct lpfc_sli4_hdw_queue *hdwq, int max)
+lpfc_sli4_release_hdwq(struct lpfc_hba *phba)
 {
+	struct lpfc_sli4_hdw_queue *hdwq;
 	uint32_t idx;
 
-	for (idx = 0; idx < max; idx++) {
-		lpfc_sli4_queue_free(hdwq[idx].hba_eq);
+	hdwq = phba->sli4_hba.hdwq;
+	for (idx = 0; idx < phba->cfg_hdw_queue; idx++) {
+		if (idx < phba->cfg_irq_chann)
+			lpfc_sli4_queue_free(hdwq[idx].hba_eq);
+		hdwq[idx].hba_eq = NULL;
+
 		lpfc_sli4_queue_free(hdwq[idx].fcp_cq);
 		lpfc_sli4_queue_free(hdwq[idx].nvme_cq);
 		lpfc_sli4_queue_free(hdwq[idx].fcp_wq);
 		lpfc_sli4_queue_free(hdwq[idx].nvme_wq);
-		hdwq[idx].hba_eq = NULL;
 		hdwq[idx].fcp_cq = NULL;
 		hdwq[idx].nvme_cq = NULL;
 		hdwq[idx].fcp_wq = NULL;
@@ -9120,8 +9156,7 @@ lpfc_sli4_queue_destroy(struct lpfc_hba *phba)
 {
 	/* Release HBA eqs */
 	if (phba->sli4_hba.hdwq)
-		lpfc_sli4_release_hdwq(phba->sli4_hba.hdwq,
-				       phba->cfg_hdw_queue);
+		lpfc_sli4_release_hdwq(phba);
 
 	if (phba->nvmet_support) {
 		lpfc_sli4_release_queues(&phba->sli4_hba.nvmet_cqset,
@@ -9202,7 +9237,6 @@ lpfc_create_wq_cq(struct lpfc_hba *phba, struct lpfc_queue *eq,
 			qidx, (uint32_t)rc);
 		return rc;
 	}
-	cq->chann = qidx;
 
 	if (qtype != LPFC_MBOX) {
 		/* Setup cq_map for fast lookup */
@@ -9222,7 +9256,6 @@ lpfc_create_wq_cq(struct lpfc_hba *phba, struct lpfc_queue *eq,
 			/* no need to tear down cq - caller will do so */
 			return rc;
 		}
-		wq->chann = qidx;
 
 		/* Bind this CQ/WQ to the NVME ring */
 		pring = wq->pring;
@@ -9252,6 +9285,38 @@ lpfc_create_wq_cq(struct lpfc_hba *phba, struct lpfc_queue *eq,
 }
 
 /**
+ * lpfc_setup_cq_lookup - Setup the CQ lookup table
+ * @phba: pointer to lpfc hba data structure.
+ *
+ * This routine will populate the cq_lookup table by all
+ * available CQ queue_id's.
+ **/
+void
+lpfc_setup_cq_lookup(struct lpfc_hba *phba)
+{
+	struct lpfc_queue *eq, *childq;
+	struct lpfc_sli4_hdw_queue *qp;
+	int qidx;
+
+	qp = phba->sli4_hba.hdwq;
+	memset(phba->sli4_hba.cq_lookup, 0,
+	       (sizeof(struct lpfc_queue *) * (phba->sli4_hba.cq_max + 1)));
+	for (qidx = 0; qidx < phba->cfg_irq_chann; qidx++) {
+		eq = qp[qidx].hba_eq;
+		if (!eq)
+			continue;
+		list_for_each_entry(childq, &eq->child_list, list) {
+			if (childq->queue_id > phba->sli4_hba.cq_max)
+				continue;
+			if ((childq->subtype == LPFC_FCP) ||
+			    (childq->subtype == LPFC_NVME))
+				phba->sli4_hba.cq_lookup[childq->queue_id] =
+					childq;
+		}
+	}
+}
+
+/**
  * lpfc_sli4_queue_setup - Set up all the SLI4 queues
  * @phba: pointer to lpfc hba data structure.
  *
@@ -9331,7 +9396,7 @@ lpfc_sli4_queue_setup(struct lpfc_hba *phba)
 		rc = -ENOMEM;
 		goto out_error;
 	}
-	for (qidx = 0; qidx < phba->cfg_hdw_queue; qidx++) {
+	for (qidx = 0; qidx < phba->cfg_irq_chann; qidx++) {
 		if (!qp[qidx].hba_eq) {
 			lpfc_printf_log(phba, KERN_ERR, LOG_INIT,
 					"0522 Fast-path EQ (%d) not "
@@ -9578,11 +9643,23 @@ lpfc_sli4_queue_setup(struct lpfc_hba *phba)
 			phba->sli4_hba.dat_rq->queue_id,
 			phba->sli4_hba.els_cq->queue_id);
 
-	for (qidx = 0; qidx < phba->cfg_hdw_queue;
+	for (qidx = 0; qidx < phba->cfg_irq_chann;
 	     qidx += LPFC_MAX_EQ_DELAY_EQID_CNT)
 		lpfc_modify_hba_eq_delay(phba, qidx, LPFC_MAX_EQ_DELAY_EQID_CNT,
 					 phba->cfg_fcp_imax);
 
+	if (phba->sli4_hba.cq_max) {
+		kfree(phba->sli4_hba.cq_lookup);
+		phba->sli4_hba.cq_lookup = kcalloc((phba->sli4_hba.cq_max + 1),
+			sizeof(struct lpfc_queue *), GFP_KERNEL);
+		if (!phba->sli4_hba.cq_lookup) {
+			lpfc_printf_log(phba, KERN_ERR, LOG_INIT,
+					"0549 Failed setup of CQ Lookup table: "
+					"size 0x%x\n", phba->sli4_hba.cq_max);
+			goto out_destroy;
+		}
+		lpfc_setup_cq_lookup(phba);
+	}
 	return 0;
 
 out_destroy:
@@ -9664,9 +9741,14 @@ lpfc_sli4_queue_unset(struct lpfc_hba *phba)
 			lpfc_wq_destroy(phba, qp->nvme_wq);
 			lpfc_cq_destroy(phba, qp->fcp_cq);
 			lpfc_cq_destroy(phba, qp->nvme_cq);
-			lpfc_eq_destroy(phba, qp->hba_eq);
+			if (qidx < phba->cfg_irq_chann)
+				lpfc_eq_destroy(phba, qp->hba_eq);
 		}
 	}
+
+	kfree(phba->sli4_hba.cq_lookup);
+	phba->sli4_hba.cq_lookup = NULL;
+	phba->sli4_hba.cq_max = 0;
 }
 
 /**
@@ -10446,22 +10528,198 @@ lpfc_sli_disable_intr(struct lpfc_hba *phba)
 }
 
 /**
+ * lpfc_find_cpu_handle - Find the CPU that corresponds to the specified EQ
+ * @phba: pointer to lpfc hba data structure.
+ * @id: EQ vector index or Hardware Queue index
+ * @match: LPFC_FIND_BY_EQ = match by EQ
+ *         LPFC_FIND_BY_HDWQ = match by Hardware Queue
+ */
+static uint16_t
+lpfc_find_cpu_handle(struct lpfc_hba *phba, uint16_t id, int match)
+{
+	struct lpfc_vector_map_info *cpup;
+	int cpu;
+
+	/* Find the desired phys_id for the specified EQ */
+	cpup = phba->sli4_hba.cpu_map;
+	for (cpu = 0; cpu < phba->sli4_hba.num_present_cpu; cpu++) {
+		if ((match == LPFC_FIND_BY_EQ) &&
+		    (cpup->irq != LPFC_VECTOR_MAP_EMPTY) &&
+		    (cpup->eq == id))
+			return cpu;
+		if ((match == LPFC_FIND_BY_HDWQ) && (cpup->hdwq == id))
+			return cpu;
+		cpup++;
+	}
+	return 0;
+}
+
+/**
+ * lpfc_find_eq_handle - Find the EQ that corresponds to the specified
+ *                       Hardware Queue
+ * @phba: pointer to lpfc hba data structure.
+ * @hdwq: Hardware Queue index
+ */
+static uint16_t
+lpfc_find_eq_handle(struct lpfc_hba *phba, uint16_t hdwq)
+{
+	struct lpfc_vector_map_info *cpup;
+	int cpu;
+
+	/* Find the desired phys_id for the specified EQ */
+	cpup = phba->sli4_hba.cpu_map;
+	for (cpu = 0; cpu < phba->sli4_hba.num_present_cpu; cpu++) {
+		if (cpup->hdwq == hdwq)
+			return cpup->eq;
+		cpup++;
+	}
+	return 0;
+}
+
+/**
+ * lpfc_find_phys_id_eq - Find the next EQ that corresponds to the specified
+ *                        Physical Id.
+ * @phba: pointer to lpfc hba data structure.
+ * @eqidx: EQ index
+ * @phys_id: CPU package physical id
+ */
+static uint16_t
+lpfc_find_phys_id_eq(struct lpfc_hba *phba, uint16_t eqidx, uint16_t phys_id)
+{
+	struct lpfc_vector_map_info *cpup;
+	int cpu, desired_phys_id;
+
+	desired_phys_id = LPFC_VECTOR_MAP_EMPTY;
+
+	/* Find the desired phys_id for the specified EQ */
+	cpup = phba->sli4_hba.cpu_map;
+	for (cpu = 0; cpu < phba->sli4_hba.num_present_cpu; cpu++) {
+		if ((cpup->irq != LPFC_VECTOR_MAP_EMPTY) &&
+		    (cpup->eq == eqidx)) {
+			desired_phys_id = cpup->phys_id;
+			break;
+		}
+		cpup++;
+	}
+	if (phys_id == desired_phys_id)
+		return eqidx;
+
+	/* Find a EQ thats on the specified phys_id */
+	cpup = phba->sli4_hba.cpu_map;
+	for (cpu = 0; cpu < phba->sli4_hba.num_present_cpu; cpu++) {
+		if ((cpup->irq != LPFC_VECTOR_MAP_EMPTY) &&
+		    (cpup->phys_id == phys_id))
+			return cpup->eq;
+		cpup++;
+	}
+	return 0;
+}
+
+/**
+ * lpfc_find_cpu_map - Find next available CPU map entry that matches the
+ *                     phys_id and core_id.
+ * @phba: pointer to lpfc hba data structure.
+ * @phys_id: CPU package physical id
+ * @core_id: CPU core id
+ * @hdwqidx: Hardware Queue index
+ * @eqidx: EQ index
+ * @isr_avail: Should an IRQ be associated with this entry
+ */
+static struct lpfc_vector_map_info *
+lpfc_find_cpu_map(struct lpfc_hba *phba, uint16_t phys_id, uint16_t core_id,
+		  uint16_t hdwqidx, uint16_t eqidx, int isr_avail)
+{
+	struct lpfc_vector_map_info *cpup;
+	int cpu;
+
+	cpup = phba->sli4_hba.cpu_map;
+	for (cpu = 0; cpu < phba->sli4_hba.num_present_cpu; cpu++) {
+		/* Does the cpup match the one we are looking for */
+		if ((cpup->phys_id == phys_id) &&
+		    (cpup->core_id == core_id)) {
+			/* If it has been already assigned, then skip it */
+			if (cpup->hdwq != LPFC_VECTOR_MAP_EMPTY) {
+				cpup++;
+				continue;
+			}
+			/* Ensure we are on the same phys_id as the first one */
+			if (!isr_avail)
+				cpup->eq = lpfc_find_phys_id_eq(phba, eqidx,
+								phys_id);
+			else
+				cpup->eq = eqidx;
+
+			cpup->hdwq = hdwqidx;
+			if (isr_avail) {
+				cpup->irq =
+				    pci_irq_vector(phba->pcidev, eqidx);
+
+				/* Now affinitize to the selected CPU */
+				irq_set_affinity_hint(cpup->irq,
+						      get_cpu_mask(cpu));
+				irq_set_status_flags(cpup->irq,
+						     IRQ_NO_BALANCING);
+
+				lpfc_printf_log(phba, KERN_INFO, LOG_INIT,
+						"3330 Set Affinity: CPU %d "
+						"EQ %d irq %d (HDWQ %x)\n",
+						cpu, cpup->eq,
+						cpup->irq, cpup->hdwq);
+			}
+			return cpup;
+		}
+		cpup++;
+	}
+	return 0;
+}
+
+#ifdef CONFIG_X86
+/**
+ * lpfc_find_hyper - Determine if the CPU map entry is hyper-threaded
+ * @phba: pointer to lpfc hba data structure.
+ * @cpu: CPU map index
+ * @phys_id: CPU package physical id
+ * @core_id: CPU core id
+ */
+static int
+lpfc_find_hyper(struct lpfc_hba *phba, int cpu,
+		uint16_t phys_id, uint16_t core_id)
+{
+	struct lpfc_vector_map_info *cpup;
+	int idx;
+
+	cpup = phba->sli4_hba.cpu_map;
+	for (idx = 0; idx < phba->sli4_hba.num_present_cpu; idx++) {
+		/* Does the cpup match the one we are looking for */
+		if ((cpup->phys_id == phys_id) &&
+		    (cpup->core_id == core_id) &&
+		    (cpu != idx)) {
+			return 1;
+		}
+		cpup++;
+	}
+	return 0;
+}
+#endif
+
+/**
  * lpfc_cpu_affinity_check - Check vector CPU affinity mappings
  * @phba: pointer to lpfc hba data structure.
+ * @vectors: number of msix vectors allocated.
  *
  * The routine will figure out the CPU affinity assignment for every
- * MSI-X vector allocated for the HBA.  The hba_eq_hdl will be updated
- * with a pointer to the CPU mask that defines ALL the CPUs this vector
- * can be associated with. If the vector can be unquely associated with
- * a single CPU, that CPU will be recorded in hba_eq_hdl[index].cpu.
+ * MSI-X vector allocated for the HBA.
  * In addition, the CPU to IO channel mapping will be calculated
  * and the phba->sli4_hba.cpu_map array will reflect this.
  */
 static void
-lpfc_cpu_affinity_check(struct lpfc_hba *phba)
+lpfc_cpu_affinity_check(struct lpfc_hba *phba, int vectors)
 {
+	int i, j, idx, phys_id;
+	int max_phys_id, min_phys_id;
+	int max_core_id, min_core_id;
 	struct lpfc_vector_map_info *cpup;
-	int cpu, idx;
+	int cpu, eqidx, hdwqidx, isr_avail;
 #ifdef CONFIG_X86
 	struct cpuinfo_x86 *cpuinfo;
 #endif
@@ -10471,6 +10729,12 @@ lpfc_cpu_affinity_check(struct lpfc_hba *phba)
 	       (sizeof(struct lpfc_vector_map_info) *
 	       phba->sli4_hba.num_present_cpu));
 
+	max_phys_id = 0;
+	min_phys_id = 0xffff;
+	max_core_id = 0;
+	min_core_id = 0xffff;
+	phys_id = 0;
+
 	/* Update CPU map with physical id and core id of each CPU */
 	cpup = phba->sli4_hba.cpu_map;
 	for (cpu = 0; cpu < phba->sli4_hba.num_present_cpu; cpu++) {
@@ -10478,34 +10742,91 @@ lpfc_cpu_affinity_check(struct lpfc_hba *phba)
 		cpuinfo = &cpu_data(cpu);
 		cpup->phys_id = cpuinfo->phys_proc_id;
 		cpup->core_id = cpuinfo->cpu_core_id;
+		cpup->hyper = lpfc_find_hyper(phba, cpu,
+					      cpup->phys_id, cpup->core_id);
 #else
 		/* No distinction between CPUs for other platforms */
 		cpup->phys_id = 0;
-		cpup->core_id = 0;
+		cpup->core_id = cpu;
+		cpup->hyper = 0;
 #endif
+
 		lpfc_printf_log(phba, KERN_INFO, LOG_INIT,
 				"3328 CPU physid %d coreid %d\n",
 				cpup->phys_id, cpup->core_id);
+
+		if (cpup->phys_id > max_phys_id)
+			max_phys_id = cpup->phys_id;
+		if (cpup->phys_id < min_phys_id)
+			min_phys_id = cpup->phys_id;
+
+		if (cpup->core_id > max_core_id)
+			max_core_id = cpup->core_id;
+		if (cpup->core_id < min_core_id)
+			min_core_id = cpup->core_id;
+
 		cpup++;
 	}
 
-	for (idx = 0; idx <  phba->cfg_hdw_queue; idx++) {
-		cpup = &phba->sli4_hba.cpu_map[idx];
-		cpup->irq = pci_irq_vector(phba->pcidev, idx);
+	/*
+	 * If the number of IRQ vectors == number of CPUs,
+	 * mapping is pretty simple: 1 to 1.
+	 * This is the desired path if NVME is enabled.
+	 */
+	if (vectors == phba->sli4_hba.num_present_cpu) {
+		cpup = phba->sli4_hba.cpu_map;
+		for (idx = 0; idx < vectors; idx++) {
+			cpup->eq = idx;
+			cpup->hdwq = idx;
+			cpup->irq = pci_irq_vector(phba->pcidev, idx);
+
+			/* Now affinitize to the selected CPU */
+			irq_set_affinity_hint(
+				pci_irq_vector(phba->pcidev, idx),
+				get_cpu_mask(idx));
+			irq_set_status_flags(cpup->irq, IRQ_NO_BALANCING);
 
-		/* For now assume vector N maps to CPU N */
-		irq_set_affinity_hint(cpup->irq, get_cpu_mask(idx));
-		cpup->hdwq = idx;
+			lpfc_printf_log(phba, KERN_INFO, LOG_INIT,
+					"3336 Set Affinity: CPU %d "
+					"EQ %d irq %d\n",
+					idx, cpup->eq,
+					pci_irq_vector(phba->pcidev, idx));
+			cpup++;
+		}
+		return;
+	}
 
-		lpfc_printf_log(phba, KERN_INFO, LOG_INIT,
-			"3336 Set Affinity: CPU %d "
-			"hdwq %d irq %d\n",
-			cpu, cpup->hdwq, cpup->irq);
+	idx = 0;
+	isr_avail = 1;
+	eqidx = 0;
+	hdwqidx = 0;
+
+	/* Mapping is more complicated for this case. Hardware Queues are
+	 * assigned in a "ping pong" fashion, ping pong-ing between the
+	 * available phys_id's.
+	 */
+	while (idx < phba->sli4_hba.num_present_cpu) {
+		for (i = min_core_id; i <= max_core_id; i++) {
+			for (j = min_phys_id; j <= max_phys_id; j++) {
+				cpup = lpfc_find_cpu_map(phba, j, i, hdwqidx,
+							 eqidx, isr_avail);
+				if (!cpup)
+					continue;
+				idx++;
+				hdwqidx++;
+				if (hdwqidx >= phba->cfg_hdw_queue)
+					hdwqidx = 0;
+				eqidx++;
+				if (eqidx >= phba->cfg_irq_chann) {
+					isr_avail = 0;
+					eqidx = 0;
+				}
+			}
+		}
 	}
 	return;
 }
 
-
 /**
  * lpfc_sli4_enable_msix - Enable MSI-X interrupt mode to SLI-4 device
  * @phba: pointer to lpfc hba data structure.
@@ -10524,7 +10845,7 @@ lpfc_sli4_enable_msix(struct lpfc_hba *phba)
 	char *name;
 
 	/* Set up MSI-X multi-message vectors */
-	vectors = phba->cfg_hdw_queue;
+	vectors = phba->cfg_irq_chann;
 
 	rc = pci_alloc_irq_vectors(phba->pcidev,
 				(phba->nvmet_support) ? 1 : 2,
@@ -10545,7 +10866,6 @@ lpfc_sli4_enable_msix(struct lpfc_hba *phba)
 
 		phba->sli4_hba.hba_eq_hdl[index].idx = index;
 		phba->sli4_hba.hba_eq_hdl[index].phba = phba;
-		atomic_set(&phba->sli4_hba.hba_eq_hdl[index].hba_eq_in_use, 1);
 		rc = request_irq(pci_irq_vector(phba->pcidev, index),
 			 &lpfc_sli4_hba_intr_handler, 0,
 			 name,
@@ -10558,17 +10878,16 @@ lpfc_sli4_enable_msix(struct lpfc_hba *phba)
 		}
 	}
 
-	if (vectors != phba->cfg_hdw_queue) {
+	if (vectors != phba->cfg_irq_chann) {
 		lpfc_printf_log(phba, KERN_ERR, LOG_INIT,
 				"3238 Reducing IO channels to match number of "
 				"MSI-X vectors, requested %d got %d\n",
-				phba->cfg_hdw_queue, vectors);
-		if (phba->cfg_hdw_queue > vectors)
-			phba->cfg_hdw_queue = vectors;
+				phba->cfg_irq_chann, vectors);
+		if (phba->cfg_irq_chann > vectors)
+			phba->cfg_irq_chann = vectors;
 		if (phba->cfg_nvmet_mrq > vectors)
 			phba->cfg_nvmet_mrq = vectors;
 	}
-	lpfc_cpu_affinity_check(phba);
 
 	return rc;
 
@@ -10623,7 +10942,7 @@ lpfc_sli4_enable_msi(struct lpfc_hba *phba)
 		return rc;
 	}
 
-	for (index = 0; index < phba->cfg_hdw_queue; index++) {
+	for (index = 0; index < phba->cfg_irq_chann; index++) {
 		phba->sli4_hba.hba_eq_hdl[index].idx = index;
 		phba->sli4_hba.hba_eq_hdl[index].phba = phba;
 	}
@@ -10688,11 +11007,10 @@ lpfc_sli4_enable_intr(struct lpfc_hba *phba, uint32_t cfg_mode)
 			phba->intr_type = INTx;
 			intr_mode = 0;
 
-			for (idx = 0; idx < phba->cfg_hdw_queue; idx++) {
+			for (idx = 0; idx < phba->cfg_irq_chann; idx++) {
 				eqhdl = &phba->sli4_hba.hba_eq_hdl[idx];
 				eqhdl->idx = idx;
 				eqhdl->phba = phba;
-				atomic_set(&eqhdl->hba_eq_in_use, 1);
 			}
 		}
 	}
@@ -10716,7 +11034,7 @@ lpfc_sli4_disable_intr(struct lpfc_hba *phba)
 		int index;
 
 		/* Free up MSI-X multi-message vectors */
-		for (index = 0; index < phba->cfg_hdw_queue; index++) {
+		for (index = 0; index < phba->cfg_irq_chann; index++) {
 			irq_set_affinity_hint(
 				pci_irq_vector(phba->pcidev, index),
 				NULL);
@@ -12092,12 +12410,13 @@ lpfc_pci_probe_one_s4(struct pci_dev *pdev, const struct pci_device_id *pid)
 	}
 	/* Default to single EQ for non-MSI-X */
 	if (phba->intr_type != MSIX) {
-		phba->cfg_hdw_queue = 1;
+		phba->cfg_irq_chann = 1;
 		if (phba->cfg_enable_fc4_type & LPFC_ENABLE_NVME) {
 			if (phba->nvmet_support)
 				phba->cfg_nvmet_mrq = 1;
 		}
 	}
+	lpfc_cpu_affinity_check(phba, phba->cfg_irq_chann);
 
 	/* Create SCSI host to the physical port */
 	error = lpfc_create_shost(phba);
diff --git a/drivers/scsi/lpfc/lpfc_nvme.c b/drivers/scsi/lpfc/lpfc_nvme.c
index c9aacd56a449..9480257c5143 100644
--- a/drivers/scsi/lpfc/lpfc_nvme.c
+++ b/drivers/scsi/lpfc/lpfc_nvme.c
@@ -239,7 +239,7 @@ lpfc_nvme_create_queue(struct nvme_fc_local_port *pnvme_lport,
 	if (qidx) {
 		str = "IO ";  /* IO queue */
 		qhandle->index = ((qidx - 1) %
-			vport->phba->cfg_hdw_queue);
+			lpfc_nvme_template.max_hw_queues);
 	} else {
 		str = "ADM";  /* Admin queue */
 		qhandle->index = qidx;
@@ -1546,14 +1546,12 @@ lpfc_nvme_fcp_io_submit(struct nvme_fc_local_port *pnvme_lport,
 		}
 	}
 
+	/* Lookup Hardware Queue index based on fcp_io_sched module parameter */
 	if (phba->cfg_fcp_io_sched == LPFC_FCP_SCHED_BY_HDWQ) {
 		idx = lpfc_queue_info->index;
 	} else {
 		cpu = smp_processor_id();
-		if (cpu < phba->cfg_hdw_queue)
-			idx = cpu;
-		else
-			idx = cpu % phba->cfg_hdw_queue;
+		idx = phba->sli4_hba.cpu_map[cpu].hdwq;
 	}
 
 	lpfc_ncmd = lpfc_get_nvme_buf(phba, ndlp, idx, expedite);
@@ -2060,7 +2058,13 @@ lpfc_nvme_create_localport(struct lpfc_vport *vport)
 	 * allocate + 3, one for cmd, one for rsp and one for this alignment
 	 */
 	lpfc_nvme_template.max_sgl_segments = phba->cfg_nvme_seg_cnt + 1;
-	lpfc_nvme_template.max_hw_queues = phba->cfg_hdw_queue;
+
+	/* Advertise how many hw queues we support based on fcp_io_sched */
+	if (phba->cfg_fcp_io_sched == LPFC_FCP_SCHED_BY_HDWQ)
+		lpfc_nvme_template.max_hw_queues = phba->cfg_hdw_queue;
+	else
+		lpfc_nvme_template.max_hw_queues =
+			phba->sli4_hba.num_present_cpu;
 
 	/* localport is allocated from the stack, but the registration
 	 * call allocates heap memory as well as the private area.
@@ -2554,6 +2558,8 @@ lpfc_nvme_wait_for_io_drain(struct lpfc_hba *phba)
 	 * WQEs have been removed from the txcmplqs.
 	 */
 	for (i = 0; i < phba->cfg_hdw_queue; i++) {
+		if (!phba->sli4_hba.hdwq[i].nvme_wq)
+			continue;
 		pring = phba->sli4_hba.hdwq[i].nvme_wq->pring;
 
 		if (!pring)
diff --git a/drivers/scsi/lpfc/lpfc_scsi.c b/drivers/scsi/lpfc/lpfc_scsi.c
index 7b22cc995d7f..a827520789f1 100644
--- a/drivers/scsi/lpfc/lpfc_scsi.c
+++ b/drivers/scsi/lpfc/lpfc_scsi.c
@@ -692,10 +692,7 @@ lpfc_get_scsi_buf_s4(struct lpfc_hba *phba, struct lpfc_nodelist *ndlp,
 		tag = blk_mq_unique_tag(cmnd->request);
 		idx = blk_mq_unique_tag_to_hwq(tag);
 	} else {
-		if (cpu < phba->cfg_hdw_queue)
-			idx = cpu;
-		else
-			idx = cpu % phba->cfg_hdw_queue;
+		idx = phba->sli4_hba.cpu_map[cpu].hdwq;
 	}
 
 	lpfc_cmd = lpfc_get_io_buf(phba, ndlp, idx,
@@ -3650,6 +3647,9 @@ lpfc_scsi_cmd_iocb_cmpl(struct lpfc_hba *phba, struct lpfc_iocbq *pIocbIn,
 	struct Scsi_Host *shost;
 	int idx;
 	uint32_t logit = LOG_FCP;
+#ifdef CONFIG_SCSI_LPFC_DEBUG_FS
+	int cpu;
+#endif
 
 	/* Sanity check on return of outstanding command */
 	cmd = lpfc_cmd->pCmd;
@@ -3660,6 +3660,13 @@ lpfc_scsi_cmd_iocb_cmpl(struct lpfc_hba *phba, struct lpfc_iocbq *pIocbIn,
 	if (phba->sli4_hba.hdwq)
 		phba->sli4_hba.hdwq[idx].scsi_cstat.io_cmpls++;
 
+#ifdef CONFIG_SCSI_LPFC_DEBUG_FS
+	if (phba->cpucheck_on & LPFC_CHECK_SCSI_IO) {
+		cpu = smp_processor_id();
+		if (cpu < LPFC_CHECK_CPU_CNT)
+			phba->sli4_hba.hdwq[idx].cpucheck_cmpl_io[cpu]++;
+	}
+#endif
 	shost = cmd->device->host;
 
 	lpfc_cmd->result = (pIocbOut->iocb.un.ulpWord[4] & IOERR_PARAM_MASK);
@@ -4336,6 +4343,9 @@ lpfc_queuecommand(struct Scsi_Host *shost, struct scsi_cmnd *cmnd)
 	struct lpfc_io_buf *lpfc_cmd;
 	struct fc_rport *rport = starget_to_rport(scsi_target(cmnd->device));
 	int err, idx;
+#ifdef CONFIG_SCSI_LPFC_DEBUG_FS
+	int cpu;
+#endif
 
 	rdata = lpfc_rport_data_from_scsi_device(cmnd->device);
 
@@ -4450,6 +4460,16 @@ lpfc_queuecommand(struct Scsi_Host *shost, struct scsi_cmnd *cmnd)
 
 	lpfc_scsi_prep_cmnd(vport, lpfc_cmd, ndlp);
 
+#ifdef CONFIG_SCSI_LPFC_DEBUG_FS
+	if (phba->cpucheck_on & LPFC_CHECK_SCSI_IO) {
+		cpu = smp_processor_id();
+		if (cpu < LPFC_CHECK_CPU_CNT) {
+			struct lpfc_sli4_hdw_queue *hdwq =
+					&phba->sli4_hba.hdwq[lpfc_cmd->hdwq_no];
+			hdwq->cpucheck_xmt_io[cpu]++;
+		}
+	}
+#endif
 	err = lpfc_sli_issue_iocb(phba, LPFC_FCP_RING,
 				  &lpfc_cmd->cur_iocbq, SLI_IOCB_RET_IOCB);
 	if (err) {
diff --git a/drivers/scsi/lpfc/lpfc_sli.c b/drivers/scsi/lpfc/lpfc_sli.c
index c0f0adccdea7..0cc81321643d 100644
--- a/drivers/scsi/lpfc/lpfc_sli.c
+++ b/drivers/scsi/lpfc/lpfc_sli.c
@@ -5586,7 +5586,7 @@ lpfc_sli4_arm_cqeq_intr(struct lpfc_hba *phba)
 						LPFC_QUEUE_REARM);
 		}
 
-		for (qidx = 0; qidx < phba->cfg_hdw_queue; qidx++)
+		for (qidx = 0; qidx < phba->cfg_irq_chann; qidx++)
 			sli4_hba->sli4_eq_release(qp[qidx].hba_eq,
 						LPFC_QUEUE_REARM);
 	}
@@ -7878,7 +7878,7 @@ lpfc_sli4_process_missed_mbox_completions(struct lpfc_hba *phba)
 	/* Find the eq associated with the mcq */
 
 	if (sli4_hba->hdwq)
-		for (eqidx = 0; eqidx < phba->cfg_hdw_queue; eqidx++)
+		for (eqidx = 0; eqidx < phba->cfg_irq_chann; eqidx++)
 			if (sli4_hba->hdwq[eqidx].hba_eq->queue_id ==
 			    sli4_hba->mbx_cq->assoc_qid) {
 				fpeq = sli4_hba->hdwq[eqidx].hba_eq;
@@ -10058,12 +10058,9 @@ int
 lpfc_sli_issue_iocb(struct lpfc_hba *phba, uint32_t ring_number,
 		    struct lpfc_iocbq *piocb, uint32_t flag)
 {
-	struct lpfc_hba_eq_hdl *hba_eq_hdl;
 	struct lpfc_sli_ring *pring;
-	struct lpfc_queue *fpeq;
-	struct lpfc_eqe *eqe;
 	unsigned long iflags;
-	int rc, idx;
+	int rc;
 
 	if (phba->sli_rev == LPFC_SLI_REV4) {
 		pring = lpfc_sli4_calc_ring(phba, piocb);
@@ -10073,34 +10070,6 @@ lpfc_sli_issue_iocb(struct lpfc_hba *phba, uint32_t ring_number,
 		spin_lock_irqsave(&pring->ring_lock, iflags);
 		rc = __lpfc_sli_issue_iocb(phba, ring_number, piocb, flag);
 		spin_unlock_irqrestore(&pring->ring_lock, iflags);
-
-		if (lpfc_fcp_look_ahead && (piocb->iocb_flag &  LPFC_IO_FCP)) {
-			idx = piocb->hba_wqidx;
-			hba_eq_hdl = &phba->sli4_hba.hba_eq_hdl[idx];
-
-			if (atomic_dec_and_test(&hba_eq_hdl->hba_eq_in_use)) {
-
-				/* Get associated EQ with this index */
-				fpeq = phba->sli4_hba.hdwq[idx].hba_eq;
-
-				/* Turn off interrupts from this EQ */
-				phba->sli4_hba.sli4_eq_clr_intr(fpeq);
-
-				/*
-				 * Process all the events on FCP EQ
-				 */
-				while ((eqe = lpfc_sli4_eq_get(fpeq))) {
-					lpfc_sli4_hba_handle_eqe(phba,
-						eqe, idx);
-					fpeq->EQ_processed++;
-				}
-
-				/* Always clear and re-arm the EQ */
-				phba->sli4_hba.sli4_eq_release(fpeq,
-					LPFC_QUEUE_REARM);
-			}
-			atomic_inc(&hba_eq_hdl->hba_eq_in_use);
-		}
 	} else {
 		/* For now, SLI2/3 will still use hbalock */
 		spin_lock_irqsave(&phba->hbalock, iflags);
@@ -13651,7 +13620,7 @@ lpfc_sli4_sp_handle_eqe(struct lpfc_hba *phba, struct lpfc_eqe *eqe,
 	/* Save EQ associated with this CQ */
 	cq->assoc_qp = speq;
 
-	if (!queue_work(phba->wq, &cq->spwork))
+	if (!queue_work_on(cq->chann, phba->wq, &cq->spwork))
 		lpfc_printf_log(phba, KERN_ERR, LOG_SLI,
 				"0390 Cannot schedule soft IRQ "
 				"for CQ eqcqid=%d, cqid=%d on CPU %d\n",
@@ -14057,18 +14026,11 @@ lpfc_sli4_hba_handle_eqe(struct lpfc_hba *phba, struct lpfc_eqe *eqe,
 	/* Get the reference to the corresponding CQ */
 	cqid = bf_get_le32(lpfc_eqe_resource_id, eqe);
 
-	/* First check for NVME/SCSI completion */
-	if ((phba->cfg_enable_fc4_type & LPFC_ENABLE_NVME) &&
-	    (cqid == phba->sli4_hba.hdwq[qidx].nvme_cq_map)) {
-		/* Process NVME / NVMET command completion */
-		cq = phba->sli4_hba.hdwq[qidx].nvme_cq;
-		goto  process_cq;
-	}
-
-	if (cqid == phba->sli4_hba.hdwq[qidx].fcp_cq_map) {
-		/* Process FCP command completion */
-		cq = phba->sli4_hba.hdwq[qidx].fcp_cq;
-		goto  process_cq;
+	/* Use the fast lookup method first */
+	if (cqid <= phba->sli4_hba.cq_max) {
+		cq = phba->sli4_hba.cq_lookup[cqid];
+		if (cq)
+			goto  work_cq;
 	}
 
 	/* Next check for NVMET completion */
@@ -14103,9 +14065,7 @@ process_cq:
 		return;
 	}
 
-	/* Save EQ associated with this CQ */
-	cq->assoc_qp = phba->sli4_hba.hdwq[qidx].hba_eq;
-
+work_cq:
 	if (!queue_work_on(cq->chann, phba->wq, &cq->irqwork))
 		lpfc_printf_log(phba, KERN_ERR, LOG_SLI,
 				"0363 Cannot schedule soft IRQ "
@@ -14233,15 +14193,6 @@ lpfc_sli4_hba_intr_handler(int irq, void *dev_id)
 	if (unlikely(!fpeq))
 		return IRQ_NONE;
 
-	if (lpfc_fcp_look_ahead) {
-		if (atomic_dec_and_test(&hba_eq_hdl->hba_eq_in_use))
-			phba->sli4_hba.sli4_eq_clr_intr(fpeq);
-		else {
-			atomic_inc(&hba_eq_hdl->hba_eq_in_use);
-			return IRQ_NONE;
-		}
-	}
-
 	/* Check device state for handling interrupt */
 	if (unlikely(lpfc_intr_state_check(phba))) {
 		/* Check again for link_state with lock held */
@@ -14250,8 +14201,6 @@ lpfc_sli4_hba_intr_handler(int irq, void *dev_id)
 			/* Flush, clear interrupt, and rearm the EQ */
 			lpfc_sli4_eq_flush(phba, fpeq);
 		spin_unlock_irqrestore(&phba->hbalock, iflag);
-		if (lpfc_fcp_look_ahead)
-			atomic_inc(&hba_eq_hdl->hba_eq_in_use);
 		return IRQ_NONE;
 	}
 
@@ -14274,12 +14223,6 @@ lpfc_sli4_hba_intr_handler(int irq, void *dev_id)
 
 	if (unlikely(ecount == 0)) {
 		fpeq->EQ_no_entry++;
-
-		if (lpfc_fcp_look_ahead) {
-			atomic_inc(&hba_eq_hdl->hba_eq_in_use);
-			return IRQ_NONE;
-		}
-
 		if (phba->intr_type == MSIX)
 			/* MSI-X treated interrupt served as no EQ share INT */
 			lpfc_printf_log(phba, KERN_WARNING, LOG_SLI,
@@ -14289,9 +14232,6 @@ lpfc_sli4_hba_intr_handler(int irq, void *dev_id)
 			return IRQ_NONE;
 	}
 
-	if (lpfc_fcp_look_ahead)
-		atomic_inc(&hba_eq_hdl->hba_eq_in_use);
-
 	return IRQ_HANDLED;
 } /* lpfc_sli4_fp_intr_handler */
 
@@ -14329,7 +14269,7 @@ lpfc_sli4_intr_handler(int irq, void *dev_id)
 	/*
 	 * Invoke fast-path host attention interrupt handling as appropriate.
 	 */
-	for (qidx = 0; qidx < phba->cfg_hdw_queue; qidx++) {
+	for (qidx = 0; qidx < phba->cfg_irq_chann; qidx++) {
 		hba_irq_rc = lpfc_sli4_hba_intr_handler(irq,
 					&phba->sli4_hba.hba_eq_hdl[qidx]);
 		if (hba_irq_rc == IRQ_HANDLED)
@@ -14516,7 +14456,7 @@ lpfc_modify_hba_eq_delay(struct lpfc_hba *phba, uint32_t startq,
 	union lpfc_sli4_cfg_shdr *shdr;
 	uint16_t dmult;
 
-	if (startq >= phba->cfg_hdw_queue)
+	if (startq >= phba->cfg_irq_chann)
 		return 0;
 
 	mbox = mempool_alloc(phba->mbox_mem_pool, GFP_KERNEL);
@@ -14530,7 +14470,7 @@ lpfc_modify_hba_eq_delay(struct lpfc_hba *phba, uint32_t startq,
 	eq_delay = &mbox->u.mqe.un.eq_delay;
 
 	/* Calculate delay multiper from maximum interrupt per second */
-	result = imax / phba->cfg_hdw_queue;
+	result = imax / phba->cfg_irq_chann;
 	if (result > LPFC_DMULT_CONST || result == 0)
 		dmult = 0;
 	else
@@ -14539,7 +14479,7 @@ lpfc_modify_hba_eq_delay(struct lpfc_hba *phba, uint32_t startq,
 		dmult = LPFC_DMULT_MAX;
 
 	cnt = 0;
-	for (qidx = startq; qidx < phba->cfg_hdw_queue; qidx++) {
+	for (qidx = startq; qidx < phba->cfg_irq_chann; qidx++) {
 		eq = phba->sli4_hba.hdwq[qidx].hba_eq;
 		if (!eq)
 			continue;
@@ -14557,7 +14497,7 @@ lpfc_modify_hba_eq_delay(struct lpfc_hba *phba, uint32_t startq,
 			val =  phba->cfg_fcp_imax;
 			if (val) {
 				/* First, interrupts per sec per EQ */
-				val = phba->cfg_fcp_imax / phba->cfg_hdw_queue;
+				val = phba->cfg_fcp_imax / phba->cfg_irq_chann;
 
 				/* us delay between each interrupt */
 				val = LPFC_SEC_TO_USEC / val;
@@ -14852,10 +14792,13 @@ lpfc_cq_create(struct lpfc_hba *phba, struct lpfc_queue *cq,
 	cq->subtype = subtype;
 	cq->queue_id = bf_get(lpfc_mbx_cq_create_q_id, &cq_create->u.response);
 	cq->assoc_qid = eq->queue_id;
+	cq->assoc_qp = eq;
 	cq->host_index = 0;
 	cq->hba_index = 0;
 	cq->entry_repost = LPFC_CQ_REPOST;
 
+	if (cq->queue_id > phba->sli4_hba.cq_max)
+		phba->sli4_hba.cq_max = cq->queue_id;
 out:
 	mempool_free(mbox, phba->mbox_mem_pool);
 	return status;
@@ -15061,6 +15004,7 @@ lpfc_cq_create_set(struct lpfc_hba *phba, struct lpfc_queue **cqp,
 		cq->type = type;
 		cq->subtype = subtype;
 		cq->assoc_qid = eq->queue_id;
+		cq->assoc_qp = eq;
 		cq->host_index = 0;
 		cq->hba_index = 0;
 		cq->entry_repost = LPFC_CQ_REPOST;
@@ -15101,6 +15045,8 @@ lpfc_cq_create_set(struct lpfc_hba *phba, struct lpfc_queue **cqp,
 	for (idx = 0; idx < numcq; idx++) {
 		cq = cqp[idx];
 		cq->queue_id = rc + idx;
+		if (cq->queue_id > phba->sli4_hba.cq_max)
+			phba->sli4_hba.cq_max = cq->queue_id;
 	}
 
 out:
@@ -19664,7 +19610,8 @@ lpfc_sli4_issue_wqe(struct lpfc_hba *phba, struct lpfc_sli4_hdw_queue *qp,
 	/* NVME_LS and NVME_LS ABTS requests. */
 	if (pwqe->iocb_flag & LPFC_IO_NVME_LS) {
 		pring =  phba->sli4_hba.nvmels_wq->pring;
-		spin_lock_irqsave(&pring->ring_lock, iflags);
+		lpfc_qp_spin_lock_irqsave(&pring->ring_lock, iflags,
+					  qp, wq_access);
 		sglq = __lpfc_sli_get_els_sglq(phba, pwqe);
 		if (!sglq) {
 			spin_unlock_irqrestore(&pring->ring_lock, iflags);
@@ -19697,7 +19644,8 @@ lpfc_sli4_issue_wqe(struct lpfc_hba *phba, struct lpfc_sli4_hdw_queue *qp,
 
 		bf_set(wqe_cqid, &wqe->generic.wqe_com, qp->nvme_cq_map);
 
-		spin_lock_irqsave(&pring->ring_lock, iflags);
+		lpfc_qp_spin_lock_irqsave(&pring->ring_lock, iflags,
+					  qp, wq_access);
 		ret = lpfc_sli4_wq_put(wq, wqe);
 		if (ret) {
 			spin_unlock_irqrestore(&pring->ring_lock, iflags);
@@ -19724,7 +19672,8 @@ lpfc_sli4_issue_wqe(struct lpfc_hba *phba, struct lpfc_sli4_hdw_queue *qp,
 		       pwqe->sli4_xritag);
 		bf_set(wqe_cqid, &wqe->generic.wqe_com, qp->nvme_cq_map);
 
-		spin_lock_irqsave(&pring->ring_lock, iflags);
+		lpfc_qp_spin_lock_irqsave(&pring->ring_lock, iflags,
+					  qp, wq_access);
 		ret = lpfc_sli4_wq_put(wq, wqe);
 		if (ret) {
 			spin_unlock_irqrestore(&pring->ring_lock, iflags);
@@ -19872,18 +19821,20 @@ void lpfc_move_xri_pvt_to_pbl(struct lpfc_hba *phba, u32 hwqid)
 {
 	struct lpfc_pbl_pool *pbl_pool;
 	struct lpfc_pvt_pool *pvt_pool;
+	struct lpfc_sli4_hdw_queue *qp;
 	struct lpfc_io_buf *lpfc_ncmd;
 	struct lpfc_io_buf *lpfc_ncmd_next;
 	unsigned long iflag;
 	struct list_head tmp_list;
 	u32 tmp_count;
 
-	pbl_pool = &phba->sli4_hba.hdwq[hwqid].p_multixri_pool->pbl_pool;
-	pvt_pool = &phba->sli4_hba.hdwq[hwqid].p_multixri_pool->pvt_pool;
+	qp = &phba->sli4_hba.hdwq[hwqid];
+	pbl_pool = &qp->p_multixri_pool->pbl_pool;
+	pvt_pool = &qp->p_multixri_pool->pvt_pool;
 	tmp_count = 0;
 
-	spin_lock_irqsave(&pbl_pool->lock, iflag);
-	spin_lock(&pvt_pool->lock);
+	lpfc_qp_spin_lock_irqsave(&pbl_pool->lock, iflag, qp, mv_to_pub_pool);
+	lpfc_qp_spin_lock(&pvt_pool->lock, qp, mv_from_pvt_pool);
 
 	if (pvt_pool->count > pvt_pool->low_watermark) {
 		/* Step 1: move (all - low_watermark) from pvt_pool
@@ -19936,7 +19887,8 @@ void lpfc_move_xri_pvt_to_pbl(struct lpfc_hba *phba, u32 hwqid)
  *   false - if the specified pbl_pool is empty or locked by someone else
  **/
 static bool
-_lpfc_move_xri_pbl_to_pvt(struct lpfc_hba *phba, struct lpfc_pbl_pool *pbl_pool,
+_lpfc_move_xri_pbl_to_pvt(struct lpfc_hba *phba, struct lpfc_sli4_hdw_queue *qp,
+			  struct lpfc_pbl_pool *pbl_pool,
 			  struct lpfc_pvt_pool *pvt_pool, u32 count)
 {
 	struct lpfc_io_buf *lpfc_ncmd;
@@ -19948,7 +19900,7 @@ _lpfc_move_xri_pbl_to_pvt(struct lpfc_hba *phba, struct lpfc_pbl_pool *pbl_pool,
 	if (ret) {
 		if (pbl_pool->count) {
 			/* Move a batch of XRIs from public to private pool */
-			spin_lock(&pvt_pool->lock);
+			lpfc_qp_spin_lock(&pvt_pool->lock, qp, mv_to_pvt_pool);
 			list_for_each_entry_safe(lpfc_ncmd,
 						 lpfc_ncmd_next,
 						 &pbl_pool->list,
@@ -19990,16 +19942,18 @@ void lpfc_move_xri_pbl_to_pvt(struct lpfc_hba *phba, u32 hwqid, u32 count)
 	struct lpfc_multixri_pool *next_multixri_pool;
 	struct lpfc_pvt_pool *pvt_pool;
 	struct lpfc_pbl_pool *pbl_pool;
+	struct lpfc_sli4_hdw_queue *qp;
 	u32 next_hwqid;
 	u32 hwq_count;
 	int ret;
 
-	multixri_pool = phba->sli4_hba.hdwq[hwqid].p_multixri_pool;
+	qp = &phba->sli4_hba.hdwq[hwqid];
+	multixri_pool = qp->p_multixri_pool;
 	pvt_pool = &multixri_pool->pvt_pool;
 	pbl_pool = &multixri_pool->pbl_pool;
 
 	/* Check if local pbl_pool is available */
-	ret = _lpfc_move_xri_pbl_to_pvt(phba, pbl_pool, pvt_pool, count);
+	ret = _lpfc_move_xri_pbl_to_pvt(phba, qp, pbl_pool, pvt_pool, count);
 	if (ret) {
 #ifdef LPFC_MXP_STAT
 		multixri_pool->local_pbl_hit_count++;
@@ -20022,7 +19976,7 @@ void lpfc_move_xri_pbl_to_pvt(struct lpfc_hba *phba, u32 hwqid, u32 count)
 
 		/* Check if the public free xri pool is available */
 		ret = _lpfc_move_xri_pbl_to_pvt(
-			phba, pbl_pool, pvt_pool, count);
+			phba, qp, pbl_pool, pvt_pool, count);
 
 		/* Exit while-loop if success or all hwqid are checked */
 	} while (!ret && next_hwqid != multixri_pool->rrb_next_hwqid);
@@ -20138,20 +20092,23 @@ void lpfc_release_io_buf(struct lpfc_hba *phba, struct lpfc_io_buf *lpfc_ncmd,
 		if ((pvt_pool->count < pvt_pool->low_watermark) ||
 		    (xri_owned < xri_limit &&
 		     pvt_pool->count < pvt_pool->high_watermark)) {
-			spin_lock_irqsave(&pvt_pool->lock, iflag);
+			lpfc_qp_spin_lock_irqsave(&pvt_pool->lock, iflag,
+						  qp, free_pvt_pool);
 			list_add_tail(&lpfc_ncmd->list,
 				      &pvt_pool->list);
 			pvt_pool->count++;
 			spin_unlock_irqrestore(&pvt_pool->lock, iflag);
 		} else {
-			spin_lock_irqsave(&pbl_pool->lock, iflag);
+			lpfc_qp_spin_lock_irqsave(&pbl_pool->lock, iflag,
+						  qp, free_pub_pool);
 			list_add_tail(&lpfc_ncmd->list,
 				      &pbl_pool->list);
 			pbl_pool->count++;
 			spin_unlock_irqrestore(&pbl_pool->lock, iflag);
 		}
 	} else {
-		spin_lock_irqsave(&qp->io_buf_list_put_lock, iflag);
+		lpfc_qp_spin_lock_irqsave(&qp->io_buf_list_put_lock, iflag,
+					  qp, free_xri);
 		list_add_tail(&lpfc_ncmd->list,
 			      &qp->lpfc_io_buf_list_put);
 		qp->put_io_bufs++;
@@ -20174,6 +20131,7 @@ void lpfc_release_io_buf(struct lpfc_hba *phba, struct lpfc_io_buf *lpfc_ncmd,
  **/
 static struct lpfc_io_buf *
 lpfc_get_io_buf_from_private_pool(struct lpfc_hba *phba,
+				  struct lpfc_sli4_hdw_queue *qp,
 				  struct lpfc_pvt_pool *pvt_pool,
 				  struct lpfc_nodelist *ndlp)
 {
@@ -20181,7 +20139,7 @@ lpfc_get_io_buf_from_private_pool(struct lpfc_hba *phba,
 	struct lpfc_io_buf *lpfc_ncmd_next;
 	unsigned long iflag;
 
-	spin_lock_irqsave(&pvt_pool->lock, iflag);
+	lpfc_qp_spin_lock_irqsave(&pvt_pool->lock, iflag, qp, alloc_pvt_pool);
 	list_for_each_entry_safe(lpfc_ncmd, lpfc_ncmd_next,
 				 &pvt_pool->list, list) {
 		if (lpfc_test_rrq_active(
@@ -20276,7 +20234,7 @@ lpfc_get_io_buf_from_multixri_pools(struct lpfc_hba *phba,
 		lpfc_move_xri_pbl_to_pvt(phba, hwqid, XRI_BATCH);
 
 	/* Get one XRI from private free xri pool */
-	lpfc_ncmd = lpfc_get_io_buf_from_private_pool(phba, pvt_pool, ndlp);
+	lpfc_ncmd = lpfc_get_io_buf_from_private_pool(phba, qp, pvt_pool, ndlp);
 
 	if (lpfc_ncmd) {
 		lpfc_ncmd->hdwq = qp;
@@ -20349,11 +20307,13 @@ struct lpfc_io_buf *lpfc_get_io_buf(struct lpfc_hba *phba,
 		lpfc_cmd = lpfc_get_io_buf_from_multixri_pools(
 			phba, ndlp, hwqid, expedite);
 	else {
-		spin_lock_irqsave(&qp->io_buf_list_get_lock, iflag);
+		lpfc_qp_spin_lock_irqsave(&qp->io_buf_list_get_lock, iflag,
+					  qp, alloc_xri_get);
 		if (qp->get_io_bufs > LPFC_NVME_EXPEDITE_XRICNT || expedite)
 			lpfc_cmd = lpfc_io_buf(phba, ndlp, hwqid);
 		if (!lpfc_cmd) {
-			spin_lock(&qp->io_buf_list_put_lock);
+			lpfc_qp_spin_lock(&qp->io_buf_list_put_lock,
+					  qp, alloc_xri_put);
 			list_splice(&qp->lpfc_io_buf_list_put,
 				    &qp->lpfc_io_buf_list_get);
 			qp->get_io_bufs += qp->put_io_bufs;
diff --git a/drivers/scsi/lpfc/lpfc_sli4.h b/drivers/scsi/lpfc/lpfc_sli4.h
index f5e58cd4c6ac..c381f2cb4909 100644
--- a/drivers/scsi/lpfc/lpfc_sli4.h
+++ b/drivers/scsi/lpfc/lpfc_sli4.h
@@ -41,7 +41,7 @@
 
 /* Multi-queue arrangement for FCP EQ/CQ/WQ tuples */
 #define LPFC_HBA_HDWQ_MIN	0
-#define LPFC_HBA_HDWQ_MAX	64
+#define LPFC_HBA_HDWQ_MAX	128
 #define LPFC_HBA_HDWQ_DEF	0
 
 /* Common buffer size to accomidate SCSI and NVME IO buffers */
@@ -166,16 +166,19 @@ struct lpfc_queue {
 	uint32_t assoc_qid;     /* Queue ID associated with, for CQ/WQ/MQ */
 	uint32_t host_index;	/* The host's index for putting or getting */
 	uint32_t hba_index;	/* The last known hba index for get or put */
+	uint32_t q_mode;
 
 	struct lpfc_sli_ring *pring; /* ptr to io ring associated with q */
 	struct lpfc_rqb *rqbp;	/* ptr to RQ buffers */
 
-	uint32_t q_mode;
 	uint16_t page_count;	/* Number of pages allocated for this queue */
 	uint16_t page_size;	/* size of page allocated for this queue */
 #define LPFC_EXPANDED_PAGE_SIZE	16384
 #define LPFC_DEFAULT_PAGE_SIZE	4096
-	uint16_t chann;		/* IO channel this queue is associated with */
+	uint16_t chann;		/* Hardware Queue association WQ/CQ */
+				/* CPU affinity for EQ */
+#define LPFC_FIND_BY_EQ		0
+#define LPFC_FIND_BY_HDWQ	1
 	uint8_t db_format;
 #define LPFC_DB_RING_FORMAT	0x01
 #define LPFC_DB_LIST_FORMAT	0x02
@@ -431,11 +434,6 @@ struct lpfc_hba_eq_hdl {
 	uint32_t idx;
 	char handler_name[LPFC_SLI4_HANDLER_NAME_SZ];
 	struct lpfc_hba *phba;
-	atomic_t hba_eq_in_use;
-	struct cpumask *cpumask;
-	/* CPU affinitsed to or 0xffffffff if multiple */
-	uint32_t cpu;
-#define LPFC_MULTI_CPU_AFFINITY 0xffffffff
 };
 
 /*BB Credit recovery value*/
@@ -529,7 +527,9 @@ struct lpfc_vector_map_info {
 	uint16_t	phys_id;
 	uint16_t	core_id;
 	uint16_t	irq;
+	uint16_t	eq;
 	uint16_t	hdwq;
+	uint16_t	hyper;
 };
 #define LPFC_VECTOR_MAP_EMPTY	0xffff
 
@@ -593,6 +593,21 @@ struct lpfc_fc4_ctrl_stat {
 	u32 io_cmpls;
 };
 
+#ifdef LPFC_HDWQ_LOCK_STAT
+struct lpfc_lock_stat {
+	uint32_t alloc_xri_get;
+	uint32_t alloc_xri_put;
+	uint32_t free_xri;
+	uint32_t wq_access;
+	uint32_t alloc_pvt_pool;
+	uint32_t mv_from_pvt_pool;
+	uint32_t mv_to_pub_pool;
+	uint32_t mv_to_pvt_pool;
+	uint32_t free_pub_pool;
+	uint32_t free_pvt_pool;
+};
+#endif
+
 /* SLI4 HBA data structure entries */
 struct lpfc_sli4_hdw_queue {
 	/* Pointers to the constructed SLI4 queues */
@@ -626,6 +641,9 @@ struct lpfc_sli4_hdw_queue {
 	/* FC-4 Stats counters */
 	struct lpfc_fc4_ctrl_stat nvme_cstat;
 	struct lpfc_fc4_ctrl_stat scsi_cstat;
+#ifdef LPFC_HDWQ_LOCK_STAT
+	struct lpfc_lock_stat lock_conflict;
+#endif
 
 #ifdef CONFIG_SCSI_LPFC_DEBUG_FS
 #define LPFC_CHECK_CPU_CNT    128
@@ -635,6 +653,34 @@ struct lpfc_sli4_hdw_queue {
 #endif
 };
 
+#ifdef LPFC_HDWQ_LOCK_STAT
+/* compile time trylock stats */
+#define lpfc_qp_spin_lock_irqsave(lock, flag, qp, lstat) \
+	{ \
+	int only_once = 1; \
+	while (spin_trylock_irqsave(lock, flag) == 0) { \
+		if (only_once) { \
+			only_once = 0; \
+			qp->lock_conflict.lstat++; \
+		} \
+	} \
+	}
+#define lpfc_qp_spin_lock(lock, qp, lstat) \
+	{ \
+	int only_once = 1; \
+	while (spin_trylock(lock) == 0) { \
+		if (only_once) { \
+			only_once = 0; \
+			qp->lock_conflict.lstat++; \
+		} \
+	} \
+	}
+#else
+#define lpfc_qp_spin_lock_irqsave(lock, flag, qp, lstat) \
+	spin_lock_irqsave(lock, flag)
+#define lpfc_qp_spin_lock(lock, qp, lstat) spin_lock(lock)
+#endif
+
 struct lpfc_sli4_hba {
 	void __iomem *conf_regs_memmap_p; /* Kernel memory mapped address for
 					   * config space registers
@@ -764,6 +810,8 @@ struct lpfc_sli4_hba {
 	uint16_t nvmet_xri_cnt;
 	uint16_t nvmet_io_wait_cnt;
 	uint16_t nvmet_io_wait_total;
+	uint16_t cq_max;
+	struct lpfc_queue **cq_lookup;
 	struct list_head lpfc_els_sgl_list;
 	struct list_head lpfc_abts_els_sgl_list;
 	spinlock_t abts_scsi_buf_list_lock; /* list of aborted SCSI IOs */