scsi: handle zone resources errors
authorDamien Le Moal <damien.lemoal@wdc.com>
Thu, 24 Sep 2020 20:53:30 +0000 (13:53 -0700)
committerJens Axboe <axboe@kernel.dk>
Tue, 13 Oct 2020 21:05:05 +0000 (15:05 -0600)
ZBC or ZAC disks that have a limit on the number of open zones may fail
a zone open command or a write to a zone that is not already implicitly
or explicitly open if the total number of open zones is already at the
maximum allowed.

For these operations, instead of returning the generic BLK_STS_IOERR,
return BLK_STS_ZONE_OPEN_RESOURCE which is returned as -ETOOMANYREFS to
the I/O issuer, allowing the device user to act appropriately on these
relatively benign zone resource errors.

Acked-by: Martin K. Petersen <martin.petersen@oracle.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com>
Signed-off-by: Keith Busch <kbusch@kernel.org>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
drivers/scsi/scsi_lib.c

index a89478a..72b1210 100644 (file)
@@ -758,6 +758,15 @@ static void scsi_io_completion_action(struct scsi_cmnd *cmd, int result)
                        /* See SSC3rXX or current. */
                        action = ACTION_FAIL;
                        break;
+               case DATA_PROTECT:
+                       action = ACTION_FAIL;
+                       if ((sshdr.asc == 0x0C && sshdr.ascq == 0x12) ||
+                           (sshdr.asc == 0x55 &&
+                            (sshdr.ascq == 0x0E || sshdr.ascq == 0x0F))) {
+                               /* Insufficient zone resources */
+                               blk_stat = BLK_STS_ZONE_OPEN_RESOURCE;
+                       }
+                       break;
                default:
                        action = ACTION_FAIL;
                        break;