scsi: ibmvfc: Fix I/O hang when port is not mapped
authorBrian King <brking@linux.vnet.ibm.com>
Mon, 19 Sep 2016 13:59:19 +0000 (08:59 -0500)
committerMartin K. Petersen <martin.petersen@oracle.com>
Mon, 19 Sep 2016 23:50:33 +0000 (19:50 -0400)
If a VFC port gets unmapped in the VIOS, it may not respond with a CRQ
init complete following H_REG_CRQ. If this occurs, we can end up having
called scsi_block_requests and not a resulting unblock until the init
complete happens, which may never occur, and we end up hanging I/O
requests.  This patch ensures the host action stay set to
IBMVFC_HOST_ACTION_TGT_DEL so we move all rports into devloss state and
unblock unless we receive an init complete.

Cc: <stable@vger.kernel.org>
Signed-off-by: Brian King <brking@linux.vnet.ibm.com>
Acked-by: Tyrel Datwyler <tyreld@linux.vnet.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
drivers/scsi/ibmvscsi/ibmvfc.c

index 56b9f452425b21960b803046f59c8d176fffc63e..7e487c78279cbd76d5fee217394d9b431c0ada7b 100644 (file)
@@ -721,7 +721,6 @@ static int ibmvfc_reset_crq(struct ibmvfc_host *vhost)
        spin_lock_irqsave(vhost->host->host_lock, flags);
        vhost->state = IBMVFC_NO_CRQ;
        vhost->logged_in = 0;
-       ibmvfc_set_host_action(vhost, IBMVFC_HOST_ACTION_NONE);
 
        /* Clean out the queue */
        memset(crq->msgs, 0, PAGE_SIZE);