[SCSI] Enable retries for SYNCRONIZE_CACHE commands to fix I/O error

hreinecke · James Bottomley · commit c213e1407be6 · 2010-05-05T12:13:26.000-04:00
Some arrays are giving I/O errors with ext3 filesystems when
SYNCHRONIZE_CACHE gets a UNIT_ATTENTION.  What is happening is that
these commands have no retries, so the UNIT_ATTENTION causes the
barrier to fail.  We should be enable retries here to clear any
transient error and allow the barrier to succeed.

Signed-off-by: Hannes Reinecke &lt;hare@suse.de&gt;
Cc: Stable Tree &lt;stable@kernel.org&gt;
Signed-off-by: James Bottomley &lt;James.Bottomley@suse.de&gt;
diff --git a/drivers/scsi/sd.c b/drivers/scsi/sd.c
@@ -1040,6 +1040,7 @@ static void sd_prepare_flush(struct request_queue *q, struct request *rq)
 {
 	rq->cmd_type = REQ_TYPE_BLOCK_PC;
 	rq->timeout = SD_TIMEOUT;
+	rq->retries = SD_MAX_RETRIES;
 	rq->cmd[0] = SYNCHRONIZE_CACHE;
 	rq->cmd_len = 10;
 }

Original file line number	Diff line number	Diff line change
`@@ -1040,6 +1040,7 @@ static void sd_prepare_flush(struct request_queue q, struct request rq)`
`1040`	`1040`	`{`
`1041`	`1041`	`rq->cmd_type = REQ_TYPE_BLOCK_PC;`
`1042`	`1042`	`rq->timeout = SD_TIMEOUT;`
	`1043`	`+ rq->retries = SD_MAX_RETRIES;`
`1043`	`1044`	`rq->cmd[0] = SYNCHRONIZE_CACHE;`
`1044`	`1045`	`rq->cmd_len = 10;`
`1045`	`1046`	`}`