fnic: Fix to cleanup aborted IO to avoid device being offlined by mid-layer
authorSatish Kharat <satishkh@cisco.com>
Fri, 18 Mar 2016 18:22:48 +0000 (11:22 -0700)
committerMartin K. Petersen <martin.petersen@oracle.com>
Mon, 11 Apr 2016 20:57:09 +0000 (16:57 -0400)
If an I/O times out and an abort issued by host, if the abort is
successful we need to set scsi status as DID_ABORT. Or else the
mid-layer error handler which looks for this error code, will offline
the device. Also if the original I/O is not found in fnic firmware, we
will consider the abort as successful.  The start_time assignment is
moved because of the new goto.  Fnic driver version changed from
1.6.0.17a to 1.6.0.19, version 1.6.0.18 has been skipped

[mkp: Fixed checkpatch warning]

Signed-off-by: Satish Kharat <satishkh@cisco.com>
Signed-off-by: Sesidhar Baddela <sebaddel@cisco.com>
Reviewed-by: Ewan D. Milne <emilne@redhat.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
drivers/scsi/fnic/fnic.h
drivers/scsi/fnic/fnic_scsi.c

index ce129e5..52a53f8 100644 (file)
@@ -39,7 +39,7 @@
 
 #define DRV_NAME               "fnic"
 #define DRV_DESCRIPTION                "Cisco FCoE HBA Driver"
-#define DRV_VERSION            "1.6.0.17a"
+#define DRV_VERSION            "1.6.0.19"
 #define PFX                    DRV_NAME ": "
 #define DFX                     DRV_NAME "%d: "
 
index f3032ca..01b480d 100644 (file)
@@ -1091,6 +1091,11 @@ static void fnic_fcpio_itmf_cmpl_handler(struct fnic *fnic,
                                atomic64_inc(
                                        &term_stats->terminate_fw_timeouts);
                        break;
+               case FCPIO_ITMF_REJECTED:
+                       FNIC_SCSI_DBG(KERN_INFO, fnic->lport->host,
+                               "abort reject recd. id %d\n",
+                               (int)(id & FNIC_TAG_MASK));
+                       break;
                case FCPIO_IO_NOT_FOUND:
                        if (CMD_FLAGS(sc) & FNIC_IO_ABTS_ISSUED)
                                atomic64_inc(&abts_stats->abort_io_not_found);
@@ -1111,9 +1116,15 @@ static void fnic_fcpio_itmf_cmpl_handler(struct fnic *fnic,
                        spin_unlock_irqrestore(io_lock, flags);
                        return;
                }
-               CMD_ABTS_STATUS(sc) = hdr_status;
+
                CMD_FLAGS(sc) |= FNIC_IO_ABT_TERM_DONE;
 
+               /* If the status is IO not found consider it as success */
+               if (hdr_status == FCPIO_IO_NOT_FOUND)
+                       CMD_ABTS_STATUS(sc) = FCPIO_SUCCESS;
+               else
+                       CMD_ABTS_STATUS(sc) = hdr_status;
+
                atomic64_dec(&fnic_stats->io_stats.active_ios);
                if (atomic64_read(&fnic->io_cmpl_skip))
                        atomic64_dec(&fnic->io_cmpl_skip);
@@ -1926,21 +1937,31 @@ int fnic_abort_cmd(struct scsi_cmnd *sc)
 
        CMD_STATE(sc) = FNIC_IOREQ_ABTS_COMPLETE;
 
+       start_time = io_req->start_time;
        /*
         * firmware completed the abort, check the status,
-        * free the io_req irrespective of failure or success
+        * free the io_req if successful. If abort fails,
+        * Device reset will clean the I/O.
         */
-       if (CMD_ABTS_STATUS(sc) != FCPIO_SUCCESS)
+       if (CMD_ABTS_STATUS(sc) == FCPIO_SUCCESS)
+               CMD_SP(sc) = NULL;
+       else {
                ret = FAILED;
-
-       CMD_SP(sc) = NULL;
+               spin_unlock_irqrestore(io_lock, flags);
+               goto fnic_abort_cmd_end;
+       }
 
        spin_unlock_irqrestore(io_lock, flags);
 
-       start_time = io_req->start_time;
        fnic_release_ioreq_buf(fnic, io_req, sc);
        mempool_free(io_req, fnic->io_req_pool);
 
+       if (sc->scsi_done) {
+       /* Call SCSI completion function to complete the IO */
+               sc->result = (DID_ABORT << 16);
+               sc->scsi_done(sc);
+       }
+
 fnic_abort_cmd_end:
        FNIC_TRACE(fnic_abort_cmd, sc->device->host->host_no,
                  sc->request->tag, sc,