scsi: smartpqi: Shorten drive visibility after removal

Check the response code returned from the LUN reset task management
function and if it indicates the LUN is not valid, do not retry.

Reduce rescan worker delay to 5 seconds for the event handler only.

The removal of a drive from the OS could have been delayed up to 30 seconds
after being physically pulled.

The driver was retrying a LUN reset 3 times even though the return code
indiciated the LUN was no longer valid. There was a 10 second delay between
each retry. Additionally, the rescan worker was scheduled to run 10 seconds
after the driver received the event.

Link: https://lore.kernel.org/r/165730601025.177165.9416869335174437006.stgit@brunhilda
Reviewed-by: Scott Teel <scott.teel@microchip.com>
Reviewed-by: Kevin Barnett <kevin.barnett@microchip.com>
Signed-off-by: Mike McGowen <Mike.McGowen@microchip.com>
Signed-off-by: Don Brace <don.brace@microchip.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

diff --git a/drivers/scsi/smartpqi/smartpqi.h b/drivers/scsi/smartpqi/smartpqi.h
index 2e40320129c08b15d08c44d0560005c4258fdb42..49895c6ca64c7565ef63fb535b5a427a8f4b21aa 100644 (file)
--- a/drivers/scsi/smartpqi/smartpqi.h
+++ b/drivers/scsi/smartpqi/smartpqi.h
@@ -708,6 +708,7 @@ typedef u32 pqi_index_t;
 #define SOP_TMF_COMPLETE               0x0
 #define SOP_TMF_REJECTED               0x4
 #define SOP_TMF_FUNCTION_SUCCEEDED     0x8
+#define SOP_RC_INCORRECT_LOGICAL_UNIT  0x9
 
 /* additional CDB bytes usage field codes */
 #define SOP_ADDITIONAL_CDB_BYTES_0     0       /* 16-byte CDB */

diff --git a/drivers/scsi/smartpqi/smartpqi_init.c b/drivers/scsi/smartpqi/smartpqi_init.c
index 7c0d069a315831fd68c7a8219318818be37a08d1..fa5cd2dfa3ad916ca3fc9ac30f2d75520625b84e 100644 (file)
--- a/drivers/scsi/smartpqi/smartpqi_init.c
+++ b/drivers/scsi/smartpqi/smartpqi_init.c
@@ -3322,6 +3322,9 @@ static int pqi_interpret_task_management_response(struct pqi_ctrl_info *ctrl_inf
        case SOP_TMF_REJECTED:
                rc = -EAGAIN;
                break;
+       case SOP_RC_INCORRECT_LOGICAL_UNIT:
+               rc = -ENODEV;
+               break;
        default:
                rc = -EIO;
                break;
@@ -3697,8 +3700,11 @@ static void pqi_event_worker(struct work_struct *work)
                event++;
        }
 
+#define PQI_RESCAN_WORK_FOR_EVENT_DELAY                (5 * HZ)
+
        if (rescan_needed)
-               pqi_schedule_rescan_worker_delayed(ctrl_info);
+               pqi_schedule_rescan_worker_with_delay(ctrl_info,
+                       PQI_RESCAN_WORK_FOR_EVENT_DELAY);
 
 out:
        pqi_ctrl_unbusy(ctrl_info);
@@ -6256,7 +6262,7 @@ static int pqi_lun_reset_with_retries(struct pqi_ctrl_info *ctrl_info, struct pq
 
        for (retries = 0;;) {
                reset_rc = pqi_lun_reset(ctrl_info, device);
-               if (reset_rc == 0 || ++retries > PQI_LUN_RESET_RETRIES)
+               if (reset_rc == 0 || reset_rc == -ENODEV || ++retries > PQI_LUN_RESET_RETRIES)
                        break;
                msleep(PQI_LUN_RESET_RETRY_INTERVAL_MSECS);
        }