scsi: lpfc: Fix deadlock on host_lock during cable pulls

During cable pull testing a deadlock was seen between lpfc_nlp_counters()
vs lpfc_mbox_process_link_up() vs lpfc_work_list_done(). They are all
waiting on the shost->host_lock.

Issue is all of these cases raise irq when taking out the lock but use
spin_unlock_irq() when unlocking. The unlock path is will unconditionally
re-enable interrupts in cases where irq state should be preserved. The
re-enablement allowed the other paths to execute which then causes the
deadlock.

Fix by converting the lock/unlock to irqsave/irqrestore.

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>

diff --git a/drivers/scsi/lpfc/lpfc_hbadisc.c b/drivers/scsi/lpfc/lpfc_hbadisc.c
index 22b6fd08e7f5e8553dbc77e48f0b3c81dfe8aef4..c8d14709343fb9edc2684e3a4cd63c26e9686521 100644 (file)
--- a/drivers/scsi/lpfc/lpfc_hbadisc.c
+++ b/drivers/scsi/lpfc/lpfc_hbadisc.c
@@ -118,6 +118,7 @@ lpfc_dev_loss_tmo_callbk(struct fc_rport *rport)
        struct lpfc_work_evt *evtp;
        int  put_node;
        int  put_rport;
+       unsigned long iflags;
 
        rdata = rport->dd_data;
        ndlp = rdata->pnode;
@@ -170,22 +171,22 @@ lpfc_dev_loss_tmo_callbk(struct fc_rport *rport)
        }
 
        shost = lpfc_shost_from_vport(vport);
-       spin_lock_irq(shost->host_lock);
+       spin_lock_irqsave(shost->host_lock, iflags);
        ndlp->nlp_flag |= NLP_IN_DEV_LOSS;
-       spin_unlock_irq(shost->host_lock);
+       spin_unlock_irqrestore(shost->host_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);
 
-       spin_lock_irq(&phba->hbalock);
+       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_irq(&phba->hbalock);
+       spin_unlock_irqrestore(&phba->hbalock, iflags);
 
        return;
 }
@@ -212,14 +213,15 @@ lpfc_dev_loss_tmo_handler(struct lpfc_nodelist *ndlp)
        int  put_node;
        int warn_on = 0;
        int fcf_inuse = 0;
+       unsigned long iflags;
 
        rport = ndlp->rport;
        vport = ndlp->vport;
        shost = lpfc_shost_from_vport(vport);
 
-       spin_lock_irq(shost->host_lock);
+       spin_lock_irqsave(shost->host_lock, iflags);
        ndlp->nlp_flag &= ~NLP_IN_DEV_LOSS;
-       spin_unlock_irq(shost->host_lock);
+       spin_unlock_irqrestore(shost->host_lock, iflags);
 
        if (!rport)
                return fcf_inuse;
@@ -3115,8 +3117,9 @@ lpfc_mbx_process_link_up(struct lpfc_hba *phba, struct lpfc_mbx_read_top *la)
        int rc;
        struct fcf_record *fcf_record;
        uint32_t fc_flags = 0;
+       unsigned long iflags;
 
-       spin_lock_irq(&phba->hbalock);
+       spin_lock_irqsave(&phba->hbalock, iflags);
        phba->fc_linkspeed = bf_get(lpfc_mbx_read_top_link_spd, la);
 
        if (!(phba->hba_flag & HBA_FCOE_MODE)) {
@@ -3213,12 +3216,12 @@ lpfc_mbx_process_link_up(struct lpfc_hba *phba, struct lpfc_mbx_read_top *la)
                vport->fc_myDID = phba->fc_pref_DID;
                fc_flags |= FC_LBIT;
        }
-       spin_unlock_irq(&phba->hbalock);
+       spin_unlock_irqrestore(&phba->hbalock, iflags);
 
        if (fc_flags) {
-               spin_lock_irq(shost->host_lock);
+               spin_lock_irqsave(shost->host_lock, iflags);
                vport->fc_flag |= fc_flags;
-               spin_unlock_irq(shost->host_lock);
+               spin_unlock_irqrestore(shost->host_lock, iflags);
        }
 
        lpfc_linkup(phba);
@@ -3292,22 +3295,22 @@ lpfc_mbx_process_link_up(struct lpfc_hba *phba, struct lpfc_mbx_read_top *la)
                 * The driver is expected to do FIP/FCF. Call the port
                 * and get the FCF Table.
                 */
-               spin_lock_irq(&phba->hbalock);
+               spin_lock_irqsave(&phba->hbalock, iflags);
                if (phba->hba_flag & FCF_TS_INPROG) {
-                       spin_unlock_irq(&phba->hbalock);
+                       spin_unlock_irqrestore(&phba->hbalock, iflags);
                        return;
                }
                /* This is the initial FCF discovery scan */
                phba->fcf.fcf_flag |= FCF_INIT_DISC;
-               spin_unlock_irq(&phba->hbalock);
+               spin_unlock_irqrestore(&phba->hbalock, iflags);
                lpfc_printf_log(phba, KERN_INFO, LOG_FIP | LOG_DISCOVERY,
                                "2778 Start FCF table scan at linkup\n");
                rc = lpfc_sli4_fcf_scan_read_fcf_rec(phba,
                                                     LPFC_FCOE_FCF_GET_FIRST);
                if (rc) {
-                       spin_lock_irq(&phba->hbalock);
+                       spin_lock_irqsave(&phba->hbalock, iflags);
                        phba->fcf.fcf_flag &= ~FCF_INIT_DISC;
-                       spin_unlock_irq(&phba->hbalock);
+                       spin_unlock_irqrestore(&phba->hbalock, iflags);
                        goto out;
                }
                /* Reset FCF roundrobin bmask for new discovery */
@@ -3366,6 +3369,7 @@ lpfc_mbx_cmpl_read_topology(struct lpfc_hba *phba, LPFC_MBOXQ_t *pmb)
        MAILBOX_t *mb = &pmb->u.mb;
        struct lpfc_dmabuf *mp = (struct lpfc_dmabuf *)(pmb->ctx_buf);
        uint8_t attn_type;
+       unsigned long iflags;
 
        /* Unblock ELS traffic */
        pring = lpfc_phba_elsring(phba);
@@ -3387,12 +3391,12 @@ lpfc_mbx_cmpl_read_topology(struct lpfc_hba *phba, LPFC_MBOXQ_t *pmb)
 
        memcpy(&phba->alpa_map[0], mp->virt, 128);
 
-       spin_lock_irq(shost->host_lock);
+       spin_lock_irqsave(shost->host_lock, iflags);
        if (bf_get(lpfc_mbx_read_top_pb, la))
                vport->fc_flag |= FC_BYPASSED_MODE;
        else
                vport->fc_flag &= ~FC_BYPASSED_MODE;
-       spin_unlock_irq(shost->host_lock);
+       spin_unlock_irqrestore(shost->host_lock, iflags);
 
        if (phba->fc_eventTag <= la->eventTag) {
                phba->fc_stat.LinkMultiEvent++;
@@ -3403,12 +3407,12 @@ lpfc_mbx_cmpl_read_topology(struct lpfc_hba *phba, LPFC_MBOXQ_t *pmb)
 
        phba->fc_eventTag = la->eventTag;
        if (phba->sli_rev < LPFC_SLI_REV4) {
-               spin_lock_irq(&phba->hbalock);
+               spin_lock_irqsave(&phba->hbalock, iflags);
                if (bf_get(lpfc_mbx_read_top_mm, la))
                        phba->sli.sli_flag |= LPFC_MENLO_MAINT;
                else
                        phba->sli.sli_flag &= ~LPFC_MENLO_MAINT;
-               spin_unlock_irq(&phba->hbalock);
+               spin_unlock_irqrestore(&phba->hbalock, iflags);
        }
 
        phba->link_events++;
@@ -4196,8 +4200,9 @@ static void
 lpfc_nlp_counters(struct lpfc_vport *vport, int state, int count)
 {
        struct Scsi_Host *shost = lpfc_shost_from_vport(vport);
+       unsigned long iflags;
 
-       spin_lock_irq(shost->host_lock);
+       spin_lock_irqsave(shost->host_lock, iflags);
        switch (state) {
        case NLP_STE_UNUSED_NODE:
                vport->fc_unused_cnt += count;
@@ -4227,7 +4232,7 @@ lpfc_nlp_counters(struct lpfc_vport *vport, int state, int count)
                        vport->fc_npr_cnt += count;
                break;
        }
-       spin_unlock_irq(shost->host_lock);
+       spin_unlock_irqrestore(shost->host_lock, iflags);
 }
 
 static void