From: James Smart <jsmart2021@gmail.com>
Date: Mon, 4 Jan 2021 10:02:35 -0800
Subject: scsi: lpfc: Fix NVMe recovery after mailbox timeout
Patch-mainline: Queued in subsystem maintainer repository
Git-repo: git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi.git
Git-commit: 9ec58ec7d41a08d79d996407b03b23da69990814
References: bsc#1180891

If a mailbox command times out, the SLI port is deemed in error and the
port is reset.  The HBA cleanup is not returning I/Os to the NVMe layer
before the port is unregistered. This is due to the HBA being marked
offline (!SLI_ACTIVE) and cleanup being done by the mailbox timeout handler
rather than an general adapter reset routine.  The mailbox timeout handler
mailbox handler only cleaned up SCSI I/Os.

Fix by reworking the mailbox handler to:

 - After handling the mailbox error, detect the board is already in
   failure (may be due to another error), and leave cleanup to the
   other handler.

 - If the mailbox command timeout is initial detector of the port error,
   continue with the board cleanup and marking the adapter offline
   (!SLI_ACTIVE). Remove the SCSI-only I/O cleanup routine. The generic
   reset adapter routine that is subsequently invoked, will clean up the
   I/Os.

 - Have the reset adapter routine flush all NVMe and SCSI I/Os if the
   adapter has been marked failed (!SLI_ACTIVE).

 - Rework the NVMe I/O terminate routine to take a status code to fail the
   I/O with and update so that cleaned up I/O calls the wqe completion
   routine. Currently it is bypassing the wqe cleanup and calling the NVMe
   I/O completion directly. The wqe completion routine will take care of
   data structure and node cleanup then call the NVMe I/O completion
   handler.

Link: https://lore.kernel.org/r/20210104180240.46824-11-jsmart2021@gmail.com
Co-developed-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Acked-by: Daniel Wagner <dwagner@suse.de>
---
 drivers/scsi/lpfc/lpfc_crtn.h |    4 ++--
 drivers/scsi/lpfc/lpfc_init.c |    8 ++++++--
 drivers/scsi/lpfc/lpfc_nvme.c |   33 +++++++++++++++++----------------
 drivers/scsi/lpfc/lpfc_sli.c  |   20 ++++++++++++--------
 4 files changed, 37 insertions(+), 28 deletions(-)

--- a/drivers/scsi/lpfc/lpfc_crtn.h
+++ b/drivers/scsi/lpfc/lpfc_crtn.h
@@ -256,7 +256,6 @@ void lpfc_nvmet_ctxbuf_post(struct lpfc_
 int lpfc_nvmet_rcv_unsol_abort(struct lpfc_vport *vport,
 			       struct fc_frame_header *fc_hdr);
 void lpfc_nvmet_wqfull_process(struct lpfc_hba *phba, struct lpfc_queue *wq);
-void lpfc_sli_flush_nvme_rings(struct lpfc_hba *phba);
 void lpfc_nvme_wait_for_io_drain(struct lpfc_hba *phba);
 void lpfc_sli4_build_dflt_fcf_record(struct lpfc_hba *, struct fcf_record *,
 			uint16_t);
@@ -595,7 +594,8 @@ void lpfc_release_io_buf(struct lpfc_hba
 void lpfc_io_ktime(struct lpfc_hba *phba, struct lpfc_io_buf *ncmd);
 void lpfc_wqe_cmd_template(void);
 void lpfc_nvmet_cmd_template(void);
-void lpfc_nvme_cancel_iocb(struct lpfc_hba *phba, struct lpfc_iocbq *pwqeIn);
+void lpfc_nvme_cancel_iocb(struct lpfc_hba *phba, struct lpfc_iocbq *pwqeIn,
+			   uint32_t stat, uint32_t param);
 extern int lpfc_enable_nvmet_cnt;
 extern unsigned long long lpfc_enable_nvmet[];
 extern int lpfc_no_hba_reset_cnt;
--- a/drivers/scsi/lpfc/lpfc_init.c
+++ b/drivers/scsi/lpfc/lpfc_init.c
@@ -6195,10 +6195,14 @@ lpfc_reset_hba(struct lpfc_hba *phba)
 		phba->link_state = LPFC_HBA_ERROR;
 		return;
 	}
-	if (phba->sli.sli_flag & LPFC_SLI_ACTIVE)
+
+	/* If not LPFC_SLI_ACTIVE, force all IO to be flushed */
+	if (phba->sli.sli_flag & LPFC_SLI_ACTIVE) {
 		lpfc_offline_prep(phba, LPFC_MBX_WAIT);
-	else
+	} else {
 		lpfc_offline_prep(phba, LPFC_MBX_NO_WAIT);
+		lpfc_sli_flush_io_rings(phba);
+	}
 	lpfc_offline(phba);
 	lpfc_sli_brdrestart(phba);
 	lpfc_online(phba);
--- a/drivers/scsi/lpfc/lpfc_nvme.c
+++ b/drivers/scsi/lpfc/lpfc_nvme.c
@@ -2447,14 +2447,17 @@ lpfc_nvme_wait_for_io_drain(struct lpfc_
 }
 
 void
-lpfc_nvme_cancel_iocb(struct lpfc_hba *phba, struct lpfc_iocbq *pwqeIn)
+lpfc_nvme_cancel_iocb(struct lpfc_hba *phba, struct lpfc_iocbq *pwqeIn,
+		      uint32_t stat, uint32_t param)
 {
 #if (IS_ENABLED(CONFIG_NVME_FC))
 	struct lpfc_io_buf *lpfc_ncmd;
 	struct nvmefc_fcp_req *nCmd;
-	struct lpfc_nvme_fcpreq_priv *freqpriv;
+	struct lpfc_wcqe_complete wcqe;
+	struct lpfc_wcqe_complete *wcqep = &wcqe;
 
-	if (!pwqeIn->context1) {
+	lpfc_ncmd = (struct lpfc_io_buf *)pwqeIn->context1;
+	if (!lpfc_ncmd) {
 		lpfc_sli_release_iocbq(phba, pwqeIn);
 		return;
 	}
@@ -2464,31 +2467,29 @@ lpfc_nvme_cancel_iocb(struct lpfc_hba *p
 		lpfc_sli_release_iocbq(phba, pwqeIn);
 		return;
 	}
-	lpfc_ncmd = (struct lpfc_io_buf *)pwqeIn->context1;
 
 	spin_lock(&lpfc_ncmd->buf_lock);
-	if (!lpfc_ncmd->nvmeCmd) {
+	nCmd = lpfc_ncmd->nvmeCmd;
+	if (!nCmd) {
 		spin_unlock(&lpfc_ncmd->buf_lock);
 		lpfc_release_nvme_buf(phba, lpfc_ncmd);
 		return;
 	}
+	spin_unlock(&lpfc_ncmd->buf_lock);
 
-	nCmd = lpfc_ncmd->nvmeCmd;
 	lpfc_printf_log(phba, KERN_INFO, LOG_NVME_IOERR,
 			"6194 NVME Cancel xri %x\n",
 			lpfc_ncmd->cur_iocbq.sli4_xritag);
 
-	nCmd->transferred_length = 0;
-	nCmd->rcv_rsplen = 0;
-	nCmd->status = NVME_SC_INTERNAL;
-	freqpriv = nCmd->private;
-	freqpriv->nvme_buf = NULL;
-	lpfc_ncmd->nvmeCmd = NULL;
-
-	spin_unlock(&lpfc_ncmd->buf_lock);
-	nCmd->done(nCmd);
+	wcqep->word0 = 0;
+	bf_set(lpfc_wcqe_c_status, wcqep, stat);
+	wcqep->parameter = param;
+	wcqep->word3 = 0; /* xb is 0 */
 
 	/* Call release with XB=1 to queue the IO into the abort list. */
-	lpfc_release_nvme_buf(phba, lpfc_ncmd);
+	if (phba->sli.sli_flag & LPFC_SLI_ACTIVE)
+		bf_set(lpfc_wcqe_c_xb, wcqep, 1);
+
+	(pwqeIn->wqe_cmpl)(phba, pwqeIn, wcqep);
 #endif
 }
--- a/drivers/scsi/lpfc/lpfc_sli.c
+++ b/drivers/scsi/lpfc/lpfc_sli.c
@@ -1535,15 +1535,19 @@ lpfc_sli_cancel_iocbs(struct lpfc_hba *p
 
 	while (!list_empty(iocblist)) {
 		list_remove_head(iocblist, piocb, struct lpfc_iocbq, list);
-		if (!piocb->iocb_cmpl) {
+		if (piocb->wqe_cmpl) {
 			if (piocb->iocb_flag & LPFC_IO_NVME)
-				lpfc_nvme_cancel_iocb(phba, piocb);
+				lpfc_nvme_cancel_iocb(phba, piocb,
+						      ulpstatus, ulpWord4);
 			else
 				lpfc_sli_release_iocbq(phba, piocb);
-		} else {
+
+		} else if (piocb->iocb_cmpl) {
 			piocb->iocb.ulpStatus = ulpstatus;
 			piocb->iocb.un.ulpWord[4] = ulpWord4;
 			(piocb->iocb_cmpl) (phba, piocb, piocb);
+		} else {
+			lpfc_sli_release_iocbq(phba, piocb);
 		}
 	}
 	return;
@@ -8162,8 +8166,10 @@ lpfc_mbox_timeout_handler(struct lpfc_hb
 
 	struct lpfc_sli *psli = &phba->sli;
 
-	/* If the mailbox completed, process the completion and return */
-	if (lpfc_sli4_process_missed_mbox_completions(phba))
+	/* If the mailbox completed, process the completion */
+	lpfc_sli4_process_missed_mbox_completions(phba);
+
+	if (!(psli->sli_flag & LPFC_SLI_ACTIVE))
 		return;
 
 	if (pmbox != NULL)
@@ -8204,8 +8210,6 @@ lpfc_mbox_timeout_handler(struct lpfc_hb
 	psli->sli_flag &= ~LPFC_SLI_ACTIVE;
 	spin_unlock_irq(&phba->hbalock);
 
-	lpfc_sli_abort_fcp_rings(phba);
-
 	lpfc_printf_log(phba, KERN_ERR, LOG_TRACE_EVENT,
 			"0345 Resetting board due to mailbox timeout\n");
 
@@ -11676,7 +11680,7 @@ lpfc_sli_validate_fcp_iocb(struct lpfc_i
 	struct lpfc_io_buf *lpfc_cmd;
 	int rc = 1;
 
-	if (iocbq->vport != vport)
+	if (!iocbq || iocbq->vport != vport)
 		return rc;
 
 	if (!(iocbq->iocb_flag &  LPFC_IO_FCP) ||