[SCSI] Lpfc 8.3.28: FC and SCSI Discovery Fixes
FC and SCSI Discovery Fixes:
- Clear the virtual fabrics bit (word 1 bit 30) when sending the FLOGI
and FDISC. (CR 124339)
- Return a MLQUEUE_DEVICE_BUSY if the driver detects that an I/O is being
retried too quickly (CR 124668)
- Remove NDLP reference put in lpfc_cmpl_els_logo_acc for all but fabric
nodes (CR 123924)
- Only retry FDISCs every second and stop retrying after devloss number
of retries (CR 13939)
- Check to see if vports are unloading before adding them to the vport
work array. (CR 124996)
- Fixed illegal state transition during driver unload (CR 124191)
- Added missing protection on setting/clearing of vport->fc_flag bit (CR 126002)
- Set NPIV flag in lpfc_mbx_process_link_up for all ports sli3 and
above. (CR 126094)
- Clear FCP command bytes that are not used. (CR 126209)
Signed-off-by: Alex Iannicelli <alex.iannicelli@emulex.com>
Signed-off-by: James Smart <james.smart@emulex.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
diff --git a/drivers/scsi/lpfc/lpfc_scsi.c b/drivers/scsi/lpfc/lpfc_scsi.c
index f1af3f9..c60f5d0 100644
--- a/drivers/scsi/lpfc/lpfc_scsi.c
+++ b/drivers/scsi/lpfc/lpfc_scsi.c
@@ -2913,8 +2913,8 @@
int_to_scsilun(lpfc_cmd->pCmd->device->lun,
&lpfc_cmd->fcp_cmnd->fcp_lun);
- memcpy(&fcp_cmnd->fcpCdb[0], scsi_cmnd->cmnd, 16);
-
+ memset(&fcp_cmnd->fcpCdb[0], 0, LPFC_FCP_CDB_LEN);
+ memcpy(&fcp_cmnd->fcpCdb[0], scsi_cmnd->cmnd, scsi_cmnd->cmd_len);
if (scsi_populate_tag_msg(scsi_cmnd, tag)) {
switch (tag[0]) {
case HEAD_OF_QUEUE_TAG:
@@ -3238,6 +3238,15 @@
cmnd->result = err;
goto out_fail_command;
}
+ /*
+ * Do not let the mid-layer retry I/O too fast. If an I/O is retried
+ * without waiting a bit then indicate that the device is busy.
+ */
+ if (cmnd->retries &&
+ time_before(jiffies, (cmnd->jiffies_at_alloc +
+ msecs_to_jiffies(LPFC_RETRY_PAUSE *
+ cmnd->retries))))
+ return SCSI_MLQUEUE_DEVICE_BUSY;
ndlp = rdata->pnode;
if ((scsi_get_prot_op(cmnd) != SCSI_PROT_NORMAL) &&