scsi: lpfc: Correct irq handling via locks when taking adapter offline
authorJames Smart <jsmart2021@gmail.com>
Mon, 10 Sep 2018 17:30:46 +0000 (10:30 -0700)
committerMartin K. Petersen <martin.petersen@oracle.com>
Wed, 12 Sep 2018 00:37:33 +0000 (20:37 -0400)
commit523128e53b1e82a7eb422168eddd0c566973520d
tree919c6490bfb70ba9f8b91db59827c215a3691249
parent0ef01a2d95fd62bb4f536e7ce4d5e8e74b97a244
scsi: lpfc: Correct irq handling via locks when taking adapter offline

When taking the board offline while performing i/o, unsafe locking errors
occurred and irq level isn't properly managed.

In lpfc_sli_hba_down, spin_lock_irqsave(&phba->hbalock, flags) does not
disable softirqs raised from timer expiry.  It is possible that a softirq is
raised from the lpfc_els_retry_delay routine and recursively requests the same
phba->hbalock spinlock causing deadlock.

Address the deadlocks by creating a new port_list lock. The softirq behavior
can then be managed a level deeper into the calling sequences.

Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <james.smart@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
drivers/scsi/lpfc/lpfc.h
drivers/scsi/lpfc/lpfc_ct.c
drivers/scsi/lpfc/lpfc_els.c
drivers/scsi/lpfc/lpfc_hbadisc.c
drivers/scsi/lpfc/lpfc_init.c
drivers/scsi/lpfc/lpfc_sli.c
drivers/scsi/lpfc/lpfc_vport.c