Merge branch '5.18/scsi-queue' into 5.18/scsi-fixes

Pull the remaining commits from 5.18/scsi-queue into fixes.

Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

8 files changed, 247 insertions, 87 deletions

diff --git a/drivers/scsi/lpfc/lpfc.h b/drivers/scsi/lpfc/lpfc.h
index f0cf8ffdc5f3..0025760230e5 100644
--- a/drivers/scsi/lpfc/lpfc.h
+++ b/drivers/scsi/lpfc/lpfc.h
@@ -897,6 +897,11 @@ enum lpfc_irq_chann_mode {
 	NHT_MODE,
 };
 
+enum lpfc_hba_bit_flags {
+	FABRIC_COMANDS_BLOCKED,
+	HBA_PCI_ERR,
+};
+
 struct lpfc_hba {
 	/* SCSI interface function jump table entries */
 	struct lpfc_io_buf * (*lpfc_get_scsi_buf)
@@ -1043,7 +1048,6 @@ struct lpfc_hba {
 					 * Firmware supports Forced Link Speed
 					 * capability
 					 */
-#define HBA_PCI_ERR		0x80000 /* The PCI slot is offline */
 #define HBA_FLOGI_ISSUED	0x100000 /* FLOGI was issued */
 #define HBA_SHORT_CMF		0x200000 /* shorter CMF timer routine */
 #define HBA_CGN_DAY_WRAP	0x400000 /* HBA Congestion info day wraps */
@@ -1350,7 +1354,6 @@ struct lpfc_hba {
 	atomic_t fabric_iocb_count;
 	struct timer_list fabric_block_timer;
 	unsigned long bit_flags;
-#define	FABRIC_COMANDS_BLOCKED	0
 	atomic_t num_rsrc_err;
 	atomic_t num_cmd_success;
 	unsigned long last_rsrc_error_time;
diff --git a/drivers/scsi/lpfc/lpfc_crtn.h b/drivers/scsi/lpfc/lpfc_crtn.h
index 96408cd6c4c8..9897a1aa387b 100644
--- a/drivers/scsi/lpfc/lpfc_crtn.h
+++ b/drivers/scsi/lpfc/lpfc_crtn.h
@@ -670,3 +670,6 @@ struct lpfc_vmid *lpfc_get_vmid_from_hashtable(struct lpfc_vport *vport,
 					      uint32_t hash, uint8_t *buf);
 void lpfc_vmid_vport_cleanup(struct lpfc_vport *vport);
 int lpfc_issue_els_qfpa(struct lpfc_vport *vport);
+
+void lpfc_sli_rpi_release(struct lpfc_vport *vport,
+			  struct lpfc_nodelist *ndlp);
diff --git a/drivers/scsi/lpfc/lpfc_hbadisc.c b/drivers/scsi/lpfc/lpfc_hbadisc.c
index 0144da30e3db..2b877dff5ed4 100644
--- a/drivers/scsi/lpfc/lpfc_hbadisc.c
+++ b/drivers/scsi/lpfc/lpfc_hbadisc.c
@@ -109,8 +109,8 @@ lpfc_rport_invalid(struct fc_rport *rport)
 
 	ndlp = rdata->pnode;
 	if (!rdata->pnode) {
-		pr_err("**** %s: NULL ndlp on rport x%px SID x%x\n",
-		       __func__, rport, rport->scsi_target_id);
+		pr_info("**** %s: NULL ndlp on rport x%px SID x%x\n",
+			__func__, rport, rport->scsi_target_id);
 		return -EINVAL;
 	}
 
@@ -169,9 +169,10 @@ lpfc_dev_loss_tmo_callbk(struct fc_rport *rport)
 
 	lpfc_printf_vlog(ndlp->vport, KERN_INFO, LOG_NODE,
 			 "3181 dev_loss_callbk x%06x, rport x%px flg x%x "
-			 "load_flag x%x refcnt %d\n",
+			 "load_flag x%x refcnt %d state %d xpt x%x\n",
 			 ndlp->nlp_DID, ndlp->rport, ndlp->nlp_flag,
-			 vport->load_flag, kref_read(&ndlp->kref));
+			 vport->load_flag, kref_read(&ndlp->kref),
+			 ndlp->nlp_state, ndlp->fc4_xpt_flags);
 
 	/* Don't schedule a worker thread event if the vport is going down.
 	 * The teardown process cleans up the node via lpfc_drop_node.
@@ -181,6 +182,11 @@ lpfc_dev_loss_tmo_callbk(struct fc_rport *rport)
 		ndlp->rport = NULL;
 
 		ndlp->fc4_xpt_flags &= ~SCSI_XPT_REGD;
+		/* clear the NLP_XPT_REGD if the node is not registered
+		 * with nvme-fc
+		 */
+		if (ndlp->fc4_xpt_flags == NLP_XPT_REGD)
+			ndlp->fc4_xpt_flags &= ~NLP_XPT_REGD;
 
 		/* Remove the node reference from remote_port_add now.
 		 * The driver will not call remote_port_delete.
@@ -225,18 +231,36 @@ lpfc_dev_loss_tmo_callbk(struct fc_rport *rport)
 	ndlp->rport = NULL;
 	spin_unlock_irqrestore(&ndlp->lock, iflags);
 
-	/* We need to hold the node by incrementing the reference
-	 * count until this queued work is done
-	 */
-	evtp->evt_arg1 = lpfc_nlp_get(ndlp);
+	if (phba->worker_thread) {
+		/* We need to hold the node by incrementing the reference
+		 * count until this queued work is done
+		 */
+		evtp->evt_arg1 = lpfc_nlp_get(ndlp);
+
+		spin_lock_irqsave(&phba->hbalock, iflags);
+		if (evtp->evt_arg1) {
+			evtp->evt = LPFC_EVT_DEV_LOSS;
+			list_add_tail(&evtp->evt_listp, &phba->work_list);
+			lpfc_worker_wake_up(phba);
+		}
+		spin_unlock_irqrestore(&phba->hbalock, iflags);
+	} else {
+		lpfc_printf_vlog(ndlp->vport, KERN_INFO, LOG_NODE,
+				 "3188 worker thread is stopped %s x%06x, "
+				 " rport x%px flg x%x load_flag x%x refcnt "
+				 "%d\n", __func__, ndlp->nlp_DID,
+				 ndlp->rport, ndlp->nlp_flag,
+				 vport->load_flag, kref_read(&ndlp->kref));
+		if (!(ndlp->fc4_xpt_flags & NVME_XPT_REGD)) {
+			spin_lock_irqsave(&ndlp->lock, iflags);
+			/* Node is in dev loss.  No further transaction. */
+			ndlp->nlp_flag &= ~NLP_IN_DEV_LOSS;
+			spin_unlock_irqrestore(&ndlp->lock, iflags);
+			lpfc_disc_state_machine(vport, ndlp, NULL,
+						NLP_EVT_DEVICE_RM);
+		}
 
-	spin_lock_irqsave(&phba->hbalock, iflags);
-	if (evtp->evt_arg1) {
-		evtp->evt = LPFC_EVT_DEV_LOSS;
-		list_add_tail(&evtp->evt_listp, &phba->work_list);
-		lpfc_worker_wake_up(phba);
 	}
-	spin_unlock_irqrestore(&phba->hbalock, iflags);
 
 	return;
 }
@@ -503,11 +527,12 @@ lpfc_dev_loss_tmo_handler(struct lpfc_nodelist *ndlp)
 		lpfc_printf_vlog(vport, KERN_ERR, LOG_TRACE_EVENT,
 				 "0203 Devloss timeout on "
 				 "WWPN %02x:%02x:%02x:%02x:%02x:%02x:%02x:%02x "
-				 "NPort x%06x Data: x%x x%x x%x\n",
+				 "NPort x%06x Data: x%x x%x x%x refcnt %d\n",
 				 *name, *(name+1), *(name+2), *(name+3),
 				 *(name+4), *(name+5), *(name+6), *(name+7),
 				 ndlp->nlp_DID, ndlp->nlp_flag,
-				 ndlp->nlp_state, ndlp->nlp_rpi);
+				 ndlp->nlp_state, ndlp->nlp_rpi,
+				 kref_read(&ndlp->kref));
 	} else {
 		lpfc_printf_vlog(vport, KERN_INFO, LOG_TRACE_EVENT,
 				 "0204 Devloss timeout on "
@@ -755,18 +780,22 @@ lpfc_work_list_done(struct lpfc_hba *phba)
 	int free_evt;
 	int fcf_inuse;
 	uint32_t nlp_did;
+	bool hba_pci_err;
 
 	spin_lock_irq(&phba->hbalock);
 	while (!list_empty(&phba->work_list)) {
 		list_remove_head((&phba->work_list), evtp, typeof(*evtp),
 				 evt_listp);
 		spin_unlock_irq(&phba->hbalock);
+		hba_pci_err = test_bit(HBA_PCI_ERR, &phba->bit_flags);
 		free_evt = 1;
 		switch (evtp->evt) {
 		case LPFC_EVT_ELS_RETRY:
 			ndlp = (struct lpfc_nodelist *) (evtp->evt_arg1);
-			lpfc_els_retry_delay_handler(ndlp);
-			free_evt = 0; /* evt is part of ndlp */
+			if (!hba_pci_err) {
+				lpfc_els_retry_delay_handler(ndlp);
+				free_evt = 0; /* evt is part of ndlp */
+			}
 			/* decrement the node reference count held
 			 * for this queued work
 			 */
@@ -788,8 +817,10 @@ lpfc_work_list_done(struct lpfc_hba *phba)
 			break;
 		case LPFC_EVT_RECOVER_PORT:
 			ndlp = (struct lpfc_nodelist *)(evtp->evt_arg1);
-			lpfc_sli_abts_recover_port(ndlp->vport, ndlp);
-			free_evt = 0;
+			if (!hba_pci_err) {
+				lpfc_sli_abts_recover_port(ndlp->vport, ndlp);
+				free_evt = 0;
+			}
 			/* decrement the node reference count held for
 			 * this queued work
 			 */
@@ -859,14 +890,18 @@ lpfc_work_done(struct lpfc_hba *phba)
 	struct lpfc_vport **vports;
 	struct lpfc_vport *vport;
 	int i;
+	bool hba_pci_err;
 
+	hba_pci_err = test_bit(HBA_PCI_ERR, &phba->bit_flags);
 	spin_lock_irq(&phba->hbalock);
 	ha_copy = phba->work_ha;
 	phba->work_ha = 0;
 	spin_unlock_irq(&phba->hbalock);
+	if (hba_pci_err)
+		ha_copy = 0;
 
 	/* First, try to post the next mailbox command to SLI4 device */
-	if (phba->pci_dev_grp == LPFC_PCI_DEV_OC)
+	if (phba->pci_dev_grp == LPFC_PCI_DEV_OC && !hba_pci_err)
 		lpfc_sli4_post_async_mbox(phba);
 
 	if (ha_copy & HA_ERATT) {
@@ -886,7 +921,7 @@ lpfc_work_done(struct lpfc_hba *phba)
 		lpfc_handle_latt(phba);
 
 	/* Handle VMID Events */
-	if (lpfc_is_vmid_enabled(phba)) {
+	if (lpfc_is_vmid_enabled(phba) && !hba_pci_err) {
 		if (phba->pport->work_port_events &
 		    WORKER_CHECK_VMID_ISSUE_QFPA) {
 			lpfc_check_vmid_qfpa_issue(phba);
@@ -936,6 +971,8 @@ lpfc_work_done(struct lpfc_hba *phba)
 			work_port_events = vport->work_port_events;
 			vport->work_port_events &= ~work_port_events;
 			spin_unlock_irq(&vport->work_port_lock);
+			if (hba_pci_err)
+				continue;
 			if (work_port_events & WORKER_DISC_TMO)
 				lpfc_disc_timeout_handler(vport);
 			if (work_port_events & WORKER_ELS_TMO)
@@ -1173,12 +1210,14 @@ lpfc_linkdown(struct lpfc_hba *phba)
 	struct lpfc_vport **vports;
 	LPFC_MBOXQ_t          *mb;
 	int i;
+	int offline;
 
 	if (phba->link_state == LPFC_LINK_DOWN)
 		return 0;
 
 	/* Block all SCSI stack I/Os */
 	lpfc_scsi_dev_block(phba);
+	offline = pci_channel_offline(phba->pcidev);
 
 	phba->defer_flogi_acc_flag = false;
 
@@ -1219,7 +1258,7 @@ lpfc_linkdown(struct lpfc_hba *phba)
 	lpfc_destroy_vport_work_array(phba, vports);
 
 	/* Clean up any SLI3 firmware default rpi's */
-	if (phba->sli_rev > LPFC_SLI_REV3)
+	if (phba->sli_rev > LPFC_SLI_REV3 || offline)
 		goto skip_unreg_did;
 
 	mb = mempool_alloc(phba->mbox_mem_pool, GFP_KERNEL);
@@ -4712,6 +4751,11 @@ lpfc_nlp_unreg_node(struct lpfc_vport *vport, struct lpfc_nodelist *ndlp)
 	spin_lock_irqsave(&ndlp->lock, iflags);
 	if (!(ndlp->fc4_xpt_flags & NLP_XPT_REGD)) {
 		spin_unlock_irqrestore(&ndlp->lock, iflags);
+		lpfc_printf_vlog(vport, KERN_INFO, LOG_SLI,
+				 "0999 %s Not regd: ndlp x%px rport x%px DID "
+				 "x%x FLG x%x XPT x%x\n",
+				  __func__, ndlp, ndlp->rport, ndlp->nlp_DID,
+				  ndlp->nlp_flag, ndlp->fc4_xpt_flags);
 		return;
 	}
 
@@ -4722,6 +4766,13 @@ lpfc_nlp_unreg_node(struct lpfc_vport *vport, struct lpfc_nodelist *ndlp)
 	    ndlp->fc4_xpt_flags & SCSI_XPT_REGD) {
 		vport->phba->nport_event_cnt++;
 		lpfc_unregister_remote_port(ndlp);
+	} else if (!ndlp->rport) {
+		lpfc_printf_vlog(vport, KERN_INFO, LOG_SLI,
+				 "1999 %s NDLP in devloss x%px DID x%x FLG x%x"
+				 " XPT x%x refcnt %d\n",
+				 __func__, ndlp, ndlp->nlp_DID, ndlp->nlp_flag,
+				 ndlp->fc4_xpt_flags,
+				 kref_read(&ndlp->kref));
 	}
 
 	if (ndlp->fc4_xpt_flags & NVME_XPT_REGD) {
@@ -5371,6 +5422,7 @@ lpfc_unreg_rpi(struct lpfc_vport *vport, struct lpfc_nodelist *ndlp)
 				ndlp->nlp_flag &= ~NLP_UNREG_INP;
 				mempool_free(mbox, phba->mbox_mem_pool);
 				acc_plogi = 1;
+				lpfc_nlp_put(ndlp);
 			}
 		} else {
 			lpfc_printf_vlog(vport, KERN_INFO,
@@ -6097,12 +6149,34 @@ lpfc_disc_flush_list(struct lpfc_vport *vport)
 	}
 }
 
+/*
+ * lpfc_notify_xport_npr - notifies xport of node disappearance
+ * @vport: Pointer to Virtual Port object.
+ *
+ * Transitions all ndlps to NPR state.  When lpfc_nlp_set_state
+ * calls lpfc_nlp_state_cleanup, the ndlp->rport is unregistered
+ * and transport notified that the node is gone.
+ * Return Code:
+ *	none
+ */
+static void
+lpfc_notify_xport_npr(struct lpfc_vport *vport)
+{
+	struct lpfc_nodelist *ndlp, *next_ndlp;
+
+	list_for_each_entry_safe(ndlp, next_ndlp, &vport->fc_nodes,
+				 nlp_listp) {
+		lpfc_nlp_set_state(vport, ndlp, NLP_STE_NPR_NODE);
+	}
+}
 void
 lpfc_cleanup_discovery_resources(struct lpfc_vport *vport)
 {
 	lpfc_els_flush_rscn(vport);
 	lpfc_els_flush_cmd(vport);
 	lpfc_disc_flush_list(vport);
+	if (pci_channel_offline(vport->phba->pcidev))
+		lpfc_notify_xport_npr(vport);
 }
 
 /*****************************************************************************/
diff --git a/drivers/scsi/lpfc/lpfc_init.c b/drivers/scsi/lpfc/lpfc_init.c
index eed6464bd880..461d333b1b3a 100644
--- a/drivers/scsi/lpfc/lpfc_init.c
+++ b/drivers/scsi/lpfc/lpfc_init.c
@@ -95,6 +95,7 @@ static void lpfc_sli4_oas_verify(struct lpfc_hba *phba);
 static uint16_t lpfc_find_cpu_handle(struct lpfc_hba *, uint16_t, int);
 static void lpfc_setup_bg(struct lpfc_hba *, struct Scsi_Host *);
 static int lpfc_sli4_cgn_parm_chg_evt(struct lpfc_hba *);
+static void lpfc_sli4_prep_dev_for_reset(struct lpfc_hba *phba);
 
 static struct scsi_transport_template *lpfc_transport_template = NULL;
 static struct scsi_transport_template *lpfc_vport_transport_template = NULL;
@@ -1642,7 +1643,7 @@ lpfc_sli4_offline_eratt(struct lpfc_hba *phba)
 {
 	spin_lock_irq(&phba->hbalock);
 	if (phba->link_state == LPFC_HBA_ERROR &&
-	    phba->hba_flag & HBA_PCI_ERR) {
+		test_bit(HBA_PCI_ERR, &phba->bit_flags)) {
 		spin_unlock_irq(&phba->hbalock);
 		return;
 	}
@@ -1985,6 +1986,7 @@ lpfc_handle_eratt_s4(struct lpfc_hba *phba)
 	if (pci_channel_offline(phba->pcidev)) {
 		lpfc_printf_log(phba, KERN_ERR, LOG_TRACE_EVENT,
 				"3166 pci channel is offline\n");
+		lpfc_sli_flush_io_rings(phba);
 		return;
 	}
 
@@ -2973,6 +2975,22 @@ lpfc_cleanup(struct lpfc_vport *vport)
 					NLP_EVT_DEVICE_RM);
 	}
 
+	/* This is a special case flush to return all
+	 * IOs before entering this loop. There are
+	 * two points in the code where a flush is
+	 * avoided if the FC_UNLOADING flag is set.
+	 * one is in the multipool destroy,
+	 * (this prevents a crash) and the other is
+	 * in the nvme abort handler, ( also prevents
+	 * a crash). Both of these exceptions are
+	 * cases where the slot is still accessible.
+	 * The flush here is only when the pci slot
+	 * is offline.
+	 */
+	if (vport->load_flag & FC_UNLOADING &&
+	    pci_channel_offline(phba->pcidev))
+		lpfc_sli_flush_io_rings(vport->phba);
+
 	/* At this point, ALL ndlp's should be gone
 	 * because of the previous NLP_EVT_DEVICE_RM.
 	 * Lets wait for this to happen, if needed.
@@ -2985,7 +3003,7 @@ lpfc_cleanup(struct lpfc_vport *vport)
 			list_for_each_entry_safe(ndlp, next_ndlp,
 						&vport->fc_nodes, nlp_listp) {
 				lpfc_printf_vlog(ndlp->vport, KERN_ERR,
-						 LOG_TRACE_EVENT,
+						 LOG_DISCOVERY,
 						 "0282 did:x%x ndlp:x%px "
 						 "refcnt:%d xflags x%x nflag x%x\n",
 						 ndlp->nlp_DID, (void *)ndlp,
@@ -3682,7 +3700,8 @@ lpfc_offline_prep(struct lpfc_hba *phba, int mbx_action)
 	struct lpfc_vport **vports;
 	struct Scsi_Host *shost;
 	int i;
-	int offline = 0;
+	int offline;
+	bool hba_pci_err;
 
 	if (vport->fc_flag & FC_OFFLINE_MODE)
 		return;
@@ -3692,6 +3711,7 @@ lpfc_offline_prep(struct lpfc_hba *phba, int mbx_action)
 	lpfc_linkdown(phba);
 
 	offline =  pci_channel_offline(phba->pcidev);
+	hba_pci_err = test_bit(HBA_PCI_ERR, &phba->bit_flags);
 
 	/* Issue an unreg_login to all nodes on all vports */
 	vports = lpfc_create_vport_work_array(phba);
@@ -3715,11 +3735,14 @@ lpfc_offline_prep(struct lpfc_hba *phba, int mbx_action)
 				ndlp->nlp_flag &= ~NLP_NPR_ADISC;
 				spin_unlock_irq(&ndlp->lock);
 
-				if (offline) {
+				if (offline || hba_pci_err) {
 					spin_lock_irq(&ndlp->lock);
 					ndlp->nlp_flag &= ~(NLP_UNREG_INP |
 							    NLP_RPI_REGISTERED);
 					spin_unlock_irq(&ndlp->lock);
+					if (phba->sli_rev == LPFC_SLI_REV4)
+						lpfc_sli_rpi_release(vports[i],
+								     ndlp);
 				} else {
 					lpfc_unreg_rpi(vports[i], ndlp);
 				}
@@ -13354,8 +13377,9 @@ lpfc_sli4_hba_unset(struct lpfc_hba *phba)
 	/* Abort all iocbs associated with the hba */
 	lpfc_sli_hba_iocb_abort(phba);
 
-	/* Wait for completion of device XRI exchange busy */
-	lpfc_sli4_xri_exchange_busy_wait(phba);
+	if (!pci_channel_offline(phba->pcidev))
+		/* Wait for completion of device XRI exchange busy */
+		lpfc_sli4_xri_exchange_busy_wait(phba);
 
 	/* per-phba callback de-registration for hotplug event */
 	if (phba->pport)
@@ -13374,15 +13398,12 @@ lpfc_sli4_hba_unset(struct lpfc_hba *phba)
 	/* Disable FW logging to host memory */
 	lpfc_ras_stop_fwlog(phba);
 
-	/* Unset the queues shared with the hardware then release all
-	 * allocated resources.
-	 */
-	lpfc_sli4_queue_unset(phba);
-	lpfc_sli4_queue_destroy(phba);
-
 	/* Reset SLI4 HBA FCoE function */
 	lpfc_pci_function_reset(phba);
 
+	/* release all queue allocated resources. */
+	lpfc_sli4_queue_destroy(phba);
+
 	/* Free RAS DMA memory */
 	if (phba->ras_fwlog.ras_enabled)
 		lpfc_sli4_ras_dma_free(phba);
@@ -14262,6 +14283,7 @@ lpfc_sli_prep_dev_for_perm_failure(struct lpfc_hba *phba)
 			"2711 PCI channel permanent disable for failure\n");
 	/* Block all SCSI devices' I/Os on the host */
 	lpfc_scsi_dev_block(phba);
+	lpfc_sli4_prep_dev_for_reset(phba);
 
 	/* stop all timers */
 	lpfc_stop_hba_timers(phba);
@@ -15057,24 +15079,28 @@ lpfc_sli4_prep_dev_for_recover(struct lpfc_hba *phba)
 static void
 lpfc_sli4_prep_dev_for_reset(struct lpfc_hba *phba)
 {
-	lpfc_printf_log(phba, KERN_ERR, LOG_TRACE_EVENT,
-			"2826 PCI channel disable preparing for reset\n");
+	int offline =  pci_channel_offline(phba->pcidev);
+
+	lpfc_printf_log(phba, KERN_ERR, LOG_INIT,
+			"2826 PCI channel disable preparing for reset offline"
+			" %d\n", offline);
 
 	/* Block any management I/Os to the device */
 	lpfc_block_mgmt_io(phba, LPFC_MBX_NO_WAIT);
 
-	/* Block all SCSI devices' I/Os on the host */
-	lpfc_scsi_dev_block(phba);
 
+	/* HBA_PCI_ERR was set in io_error_detect */
+	lpfc_offline_prep(phba, LPFC_MBX_NO_WAIT);
 	/* Flush all driver's outstanding I/Os as we are to reset */
 	lpfc_sli_flush_io_rings(phba);
+	lpfc_offline(phba);
 
 	/* stop all timers */
 	lpfc_stop_hba_timers(phba);
 
+	lpfc_sli4_queue_destroy(phba);
 	/* Disable interrupt and pci device */
 	lpfc_sli4_disable_intr(phba);
-	lpfc_sli4_queue_destroy(phba);
 	pci_disable_device(phba->pcidev);
 }
 
@@ -15123,6 +15149,7 @@ lpfc_io_error_detected_s4(struct pci_dev *pdev, pci_channel_state_t state)
 {
 	struct Scsi_Host *shost = pci_get_drvdata(pdev);
 	struct lpfc_hba *phba = ((struct lpfc_vport *)shost->hostdata)->phba;
+	bool hba_pci_err;
 
 	switch (state) {
 	case pci_channel_io_normal:
@@ -15130,17 +15157,24 @@ lpfc_io_error_detected_s4(struct pci_dev *pdev, pci_channel_state_t state)
 		lpfc_sli4_prep_dev_for_recover(phba);
 		return PCI_ERS_RESULT_CAN_RECOVER;
 	case pci_channel_io_frozen:
-		phba->hba_flag |= HBA_PCI_ERR;
+		hba_pci_err = test_and_set_bit(HBA_PCI_ERR, &phba->bit_flags);
 		/* Fatal error, prepare for slot reset */
-		lpfc_sli4_prep_dev_for_reset(phba);
+		if (!hba_pci_err)
+			lpfc_sli4_prep_dev_for_reset(phba);
+		else
+			lpfc_printf_log(phba, KERN_ERR, LOG_INIT,
+					"2832  Already handling PCI error "
+					"state: x%x\n", state);
 		return PCI_ERS_RESULT_NEED_RESET;
 	case pci_channel_io_perm_failure:
-		phba->hba_flag |= HBA_PCI_ERR;
+		set_bit(HBA_PCI_ERR, &phba->bit_flags);
 		/* Permanent failure, prepare for device down */
 		lpfc_sli4_prep_dev_for_perm_failure(phba);
 		return PCI_ERS_RESULT_DISCONNECT;
 	default:
-		phba->hba_flag |= HBA_PCI_ERR;
+		hba_pci_err = test_and_set_bit(HBA_PCI_ERR, &phba->bit_flags);
+		if (!hba_pci_err)
+			lpfc_sli4_prep_dev_for_reset(phba);
 		/* Unknown state, prepare and request slot reset */
 		lpfc_printf_log(phba, KERN_ERR, LOG_TRACE_EVENT,
 				"2825 Unknown PCI error state: x%x\n", state);
@@ -15174,17 +15208,21 @@ lpfc_io_slot_reset_s4(struct pci_dev *pdev)
 	struct lpfc_hba *phba = ((struct lpfc_vport *)shost->hostdata)->phba;
 	struct lpfc_sli *psli = &phba->sli;
 	uint32_t intr_mode;
+	bool hba_pci_err;
 
 	dev_printk(KERN_INFO, &pdev->dev, "recovering from a slot reset.\n");
 	if (pci_enable_device_mem(pdev)) {
 		printk(KERN_ERR "lpfc: Cannot re-enable "
-			"PCI device after reset.\n");
+		       "PCI device after reset.\n");
 		return PCI_ERS_RESULT_DISCONNECT;
 	}
 
 	pci_restore_state(pdev);
 
-	phba->hba_flag &= ~HBA_PCI_ERR;
+	hba_pci_err = test_and_clear_bit(HBA_PCI_ERR, &phba->bit_flags);
+	if (!hba_pci_err)
+		dev_info(&pdev->dev,
+			 "hba_pci_err was not set, recovering slot reset.\n");
 	/*
 	 * As the new kernel behavior of pci_restore_state() API call clears
 	 * device saved_state flag, need to save the restored state again.
@@ -15198,6 +15236,8 @@ lpfc_io_slot_reset_s4(struct pci_dev *pdev)
 	psli->sli_flag &= ~LPFC_SLI_ACTIVE;
 	spin_unlock_irq(&phba->hbalock);
 
+	/* Init cpu_map array */
+	lpfc_cpu_map_array_init(phba);
 	/* Configure and enable interrupt */
 	intr_mode = lpfc_sli4_enable_intr(phba, phba->intr_mode);
 	if (intr_mode == LPFC_INTR_ERROR) {
@@ -15239,8 +15279,6 @@ lpfc_io_resume_s4(struct pci_dev *pdev)
 	 */
 	if (!(phba->sli.sli_flag & LPFC_SLI_ACTIVE)) {
 		/* Perform device reset */
-		lpfc_offline_prep(phba, LPFC_MBX_WAIT);
-		lpfc_offline(phba);
 		lpfc_sli_brdrestart(phba);
 		/* Bring the device back online */
 		lpfc_online(phba);
diff --git a/drivers/scsi/lpfc/lpfc_nvme.c b/drivers/scsi/lpfc/lpfc_nvme.c
index 1213a299f9aa..8d26f207ebd2 100644
--- a/drivers/scsi/lpfc/lpfc_nvme.c
+++ b/drivers/scsi/lpfc/lpfc_nvme.c
@@ -93,6 +93,11 @@ lpfc_nvme_create_queue(struct nvme_fc_local_port *pnvme_lport,
 
 	lport = (struct lpfc_nvme_lport *)pnvme_lport->private;
 	vport = lport->vport;
+
+	if (!vport || vport->load_flag & FC_UNLOADING ||
+	    vport->phba->hba_flag & HBA_IOQ_FLUSH)
+		return -ENODEV;
+
 	qhandle = kzalloc(sizeof(struct lpfc_nvme_qhandle), GFP_KERNEL);
 	if (qhandle == NULL)
 		return -ENOMEM;
@@ -267,7 +272,8 @@ lpfc_nvme_handle_lsreq(struct lpfc_hba *phba,
 		return -EINVAL;
 
 	remoteport = lpfc_rport->remoteport;
-	if (!vport->localport)
+	if (!vport->localport ||
+	    vport->phba->hba_flag & HBA_IOQ_FLUSH)
 		return -EINVAL;
 
 	lport = vport->localport->private;
@@ -559,6 +565,8 @@ __lpfc_nvme_ls_req(struct lpfc_vport *vport, struct lpfc_nodelist *ndlp,
 				 ndlp->nlp_DID, ntype, nstate);
 		return -ENODEV;
 	}
+	if (vport->phba->hba_flag & HBA_IOQ_FLUSH)
+		return -ENODEV;
 
 	if (!vport->phba->sli4_hba.nvmels_wq)
 		return -ENOMEM;
@@ -662,7 +670,8 @@ lpfc_nvme_ls_req(struct nvme_fc_local_port *pnvme_lport,
 		return -EINVAL;
 
 	vport = lport->vport;
-	if (vport->load_flag & FC_UNLOADING)
+	if (vport->load_flag & FC_UNLOADING ||
+	    vport->phba->hba_flag & HBA_IOQ_FLUSH)
 		return -ENODEV;
 
 	atomic_inc(&lport->fc4NvmeLsRequests);
@@ -1516,7 +1525,8 @@ lpfc_nvme_fcp_io_submit(struct nvme_fc_local_port *pnvme_lport,
 
 	phba = vport->phba;
 
-	if (unlikely(vport->load_flag & FC_UNLOADING)) {
+	if ((unlikely(vport->load_flag & FC_UNLOADING)) ||
+	    phba->hba_flag & HBA_IOQ_FLUSH) {
 		lpfc_printf_vlog(vport, KERN_INFO, LOG_NVME_IOERR,
 				 "6124 Fail IO, Driver unload\n");
 		atomic_inc(&lport->xmt_fcp_err);
@@ -2169,8 +2179,7 @@ lpfc_nvme_lport_unreg_wait(struct lpfc_vport *vport,
 			abts_nvme = 0;
 			for (i = 0; i < phba->cfg_hdw_queue; i++) {
 				qp = &phba->sli4_hba.hdwq[i];
-				if (!vport || !vport->localport ||
-				    !qp || !qp->io_wq)
+				if (!vport->localport || !qp || !qp->io_wq)
 					return;
 
 				pring = qp->io_wq->pring;
@@ -2180,8 +2189,9 @@ lpfc_nvme_lport_unreg_wait(struct lpfc_vport *vport,
 				abts_scsi += qp->abts_scsi_io_bufs;
 				abts_nvme += qp->abts_nvme_io_bufs;
 			}
-			if (!vport || !vport->localport ||
-			    vport->phba->hba_flag & HBA_PCI_ERR)
+			if (!vport->localport ||
+			    test_bit(HBA_PCI_ERR, &vport->phba->bit_flags) ||
+			    vport->load_flag & FC_UNLOADING)
 				return;
 
 			lpfc_printf_vlog(vport, KERN_ERR, LOG_TRACE_EVENT,
@@ -2541,8 +2551,7 @@ lpfc_nvme_unregister_port(struct lpfc_vport *vport, struct lpfc_nodelist *ndlp)
 		 * return values is ignored.  The upcall is a courtesy to the
 		 * transport.
 		 */
-		if (vport->load_flag & FC_UNLOADING ||
-		    unlikely(vport->phba->hba_flag & HBA_PCI_ERR))
+		if (vport->load_flag & FC_UNLOADING)
 			(void)nvme_fc_set_remoteport_devloss(remoteport, 0);
 
 		ret = nvme_fc_unregister_remoteport(remoteport);
diff --git a/drivers/scsi/lpfc/lpfc_scsi.c b/drivers/scsi/lpfc/lpfc_scsi.c
index 3c132604fd91..ba9dbb51b75f 100644
--- a/drivers/scsi/lpfc/lpfc_scsi.c
+++ b/drivers/scsi/lpfc/lpfc_scsi.c
@@ -5929,13 +5929,15 @@ lpfc_abort_handler(struct scsi_cmnd *cmnd)
 	}
 
 	lpfc_cmd->waitq = &waitq;
-	if (phba->sli_rev == LPFC_SLI_REV4)
+	if (phba->sli_rev == LPFC_SLI_REV4) {
 		spin_unlock(&pring_s4->ring_lock);
-	else
+		ret_val = lpfc_sli4_issue_abort_iotag(phba, iocb,
+						      lpfc_sli_abort_fcp_cmpl);
+	} else {
 		pring = &phba->sli.sli3_ring[LPFC_FCP_RING];
-
-	ret_val = lpfc_sli_issue_abort_iotag(phba, pring, iocb,
-					     lpfc_sli_abort_fcp_cmpl);
+		ret_val = lpfc_sli_issue_abort_iotag(phba, pring, iocb,
+						     lpfc_sli_abort_fcp_cmpl);
+	}
 
 	/* Make sure HBA is alive */
 	lpfc_issue_hb_tmo(phba);
diff --git a/drivers/scsi/lpfc/lpfc_sli.c b/drivers/scsi/lpfc/lpfc_sli.c
index 20d40957a385..bda2a7ba4e77 100644
--- a/drivers/scsi/lpfc/lpfc_sli.c
+++ b/drivers/scsi/lpfc/lpfc_sli.c
@@ -2828,6 +2828,12 @@ __lpfc_sli_rpi_release(struct lpfc_vport *vport, struct lpfc_nodelist *ndlp)
 	ndlp->nlp_flag &= ~NLP_UNREG_INP;
 }
 
+void
+lpfc_sli_rpi_release(struct lpfc_vport *vport, struct lpfc_nodelist *ndlp)
+{
+	__lpfc_sli_rpi_release(vport, ndlp);
+}
+
 /**
  * lpfc_sli_def_mbox_cmpl - Default mailbox completion handler
  * @phba: Pointer to HBA context object.
@@ -3715,7 +3721,15 @@ lpfc_sli_process_sol_iocb(struct lpfc_hba *phba, struct lpfc_sli_ring *pring,
 	unsigned long iflag;
 	u32 ulp_command, ulp_status, ulp_word4, ulp_context, iotag;
 
+	if (phba->sli_rev == LPFC_SLI_REV4)
+		spin_lock_irqsave(&pring->ring_lock, iflag);
+	else
+		spin_lock_irqsave(&phba->hbalock, iflag);
 	cmdiocbp = lpfc_sli_iocbq_lookup(phba, pring, saveq);
+	if (phba->sli_rev == LPFC_SLI_REV4)
+		spin_unlock_irqrestore(&pring->ring_lock, iflag);
+	else
+		spin_unlock_irqrestore(&phba->hbalock, iflag);
 
 	ulp_command = get_job_cmnd(phba, saveq);
 	ulp_status = get_job_ulpstatus(phba, saveq);
@@ -4052,10 +4066,8 @@ lpfc_sli_handle_fast_ring_event(struct lpfc_hba *phba,
 				break;
 			}
 
-			spin_unlock_irqrestore(&phba->hbalock, iflag);
 			cmdiocbq = lpfc_sli_iocbq_lookup(phba, pring,
 							 &rspiocbq);
-			spin_lock_irqsave(&phba->hbalock, iflag);
 			if (unlikely(!cmdiocbq))
 				break;
 			if (cmdiocbq->cmd_flag & LPFC_DRIVER_ABORTED)
@@ -4536,42 +4548,62 @@ lpfc_sli_handle_slow_ring_event_s4(struct lpfc_hba *phba,
 void
 lpfc_sli_abort_iocb_ring(struct lpfc_hba *phba, struct lpfc_sli_ring *pring)
 {
-	LIST_HEAD(completions);
+	LIST_HEAD(tx_completions);
+	LIST_HEAD(txcmplq_completions);
 	struct lpfc_iocbq *iocb, *next_iocb;
+	int offline;
 
 	if (pring->ringno == LPFC_ELS_RING) {
 		lpfc_fabric_abort_hba(phba);
 	}
+	offline = pci_channel_offline(phba->pcidev);
 
 	/* Error everything on txq and txcmplq
 	 * First do the txq.
 	 */
 	if (phba->sli_rev >= LPFC_SLI_REV4) {
 		spin_lock_irq(&pring->ring_lock);
-		list_splice_init(&pring->txq, &completions);
+		list_splice_init(&pring->txq, &tx_completions);
 		pring->txq_cnt = 0;
-		spin_unlock_irq(&pring->ring_lock);
 
-		spin_lock_irq(&phba->hbalock);
-		/* Next issue ABTS for everything on the txcmplq */
-		list_for_each_entry_safe(iocb, next_iocb, &pring->txcmplq, list)
-			lpfc_sli_issue_abort_iotag(phba, pring, iocb, NULL);
-		spin_unlock_irq(&phba->hbalock);
+		if (offline) {
+			list_splice_init(&pring->txcmplq,
+					 &txcmplq_completions);
+		} else {
+			/* Next issue ABTS for everything on the txcmplq */
+			list_for_each_entry_safe(iocb, next_iocb,
+						 &pring->txcmplq, list)
+				lpfc_sli_issue_abort_iotag(phba, pring,
+							   iocb, NULL);
+		}
+		spin_unlock_irq(&pring->ring_lock);
 	} else {
 		spin_lock_irq(&phba->hbalock);
-		list_splice_init(&pring->txq, &completions);
+		list_splice_init(&pring->txq, &tx_completions);
 		pring->txq_cnt = 0;
 
-		/* Next issue ABTS for everything on the txcmplq */
-		list_for_each_entry_safe(iocb, next_iocb, &pring->txcmplq, list)
-			lpfc_sli_issue_abort_iotag(phba, pring, iocb, NULL);
+		if (offline) {
+			list_splice_init(&pring->txcmplq, &txcmplq_completions);
+		} else {
+			/* Next issue ABTS for everything on the txcmplq */
+			list_for_each_entry_safe(iocb, next_iocb,
+						 &pring->txcmplq, list)
+				lpfc_sli_issue_abort_iotag(phba, pring,
+							   iocb, NULL);
+		}
 		spin_unlock_irq(&phba->hbalock);
 	}
-	/* Make sure HBA is alive */
-	lpfc_issue_hb_tmo(phba);
 
+	if (offline) {
+		/* Cancel all the IOCBs from the completions list */
+		lpfc_sli_cancel_iocbs(phba, &txcmplq_completions,
+				      IOSTAT_LOCAL_REJECT, IOERR_SLI_ABORTED);
+	} else {
+		/* Make sure HBA is alive */
+		lpfc_issue_hb_tmo(phba);
+	}
 	/* Cancel all the IOCBs from the completions list */
-	lpfc_sli_cancel_iocbs(phba, &completions, IOSTAT_LOCAL_REJECT,
+	lpfc_sli_cancel_iocbs(phba, &tx_completions, IOSTAT_LOCAL_REJECT,
 			      IOERR_SLI_ABORTED);
 }
 
@@ -4624,11 +4656,6 @@ lpfc_sli_flush_io_rings(struct lpfc_hba *phba)
 	struct lpfc_iocbq *piocb, *next_iocb;
 
 	spin_lock_irq(&phba->hbalock);
-	if (phba->hba_flag & HBA_IOQ_FLUSH ||
-	    !phba->sli4_hba.hdwq) {
-		spin_unlock_irq(&phba->hbalock);
-		return;
-	}
 	/* Indicate the I/O queues are flushed */
 	phba->hba_flag |= HBA_IOQ_FLUSH;
 	spin_unlock_irq(&phba->hbalock);
@@ -10997,6 +11024,10 @@ lpfc_sli_issue_iocb(struct lpfc_hba *phba, uint32_t ring_number,
 	unsigned long iflags;
 	int rc;
 
+	/* If the PCI channel is in offline state, do not post iocbs. */
+	if (unlikely(pci_channel_offline(phba->pcidev)))
+		return IOCB_ERROR;
+
 	if (phba->sli_rev == LPFC_SLI_REV4) {
 		lpfc_sli_prep_wqe(phba, piocb);
 
diff --git a/drivers/scsi/lpfc/lpfc_version.h b/drivers/scsi/lpfc/lpfc_version.h
index e52f37e5d896..a4d3259b8c52 100644
--- a/drivers/scsi/lpfc/lpfc_version.h
+++ b/drivers/scsi/lpfc/lpfc_version.h
@@ -20,7 +20,7 @@
  * included with this package.                                     *
  *******************************************************************/
 
-#define LPFC_DRIVER_VERSION "14.2.0.0"
+#define LPFC_DRIVER_VERSION "14.2.0.1"
 #define LPFC_DRIVER_NAME		"lpfc"
 
 /* Used for SLI 2/3 */