Enable retries for SYNCRONIZE_CACHE commands
PATCH 1/1 Enable retries for SYNCRONIZE_CACHE commands
sd is using SYNCHRONIZE_CACHE to simulate barrier requests; however no error recovery is taking place as it's being injected from the block layer. So we should be enable retries here to clear any transient error.
--- drivers/scsi/sd.c | 1 + 1 files changed, 1 insertions(+), 0 deletions(-)
drivers/scsi/sd.c +1 -0
@@ -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 ; }
Signature · 1 line
1.6.0.2
Quoted 16 lines · Hannes Reinecke
sd is using
SYNCHRONIZE_CACHEto simulate barrier requests; however no error recovery is taking place as it's being injected from the block layer. So we should be enable retries here to clear any transient error.Signed-off-by Hannes Reinecke <[email protected]>--- drivers/scsi/sd.c | 1 + 1 files changed, 1 insertions(+), 0 deletions(-)drivers/scsi/sd.c +0
-0
drivers/scsi/sd.c static void sd_prepare_flush(struct request_queue *q, · line 1046
Quoted 7 lines
drivers/scsi/sd.c +1
-0@@ -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 ; }
My patch already does that and also does it in sd_sync_cache().
http://kerneltrap.org/mailarchive/linux-scsi/2010/4/20/6884656
Please wait till Friday. I will try to find some time to update it to include another sysfs-device timeout value.
Sorry that I didn't manage to that so far, but I'm trying very hard to finish my thesis and the remaining two days per week for DDN, Lustre and Linux are then quickly taken by emergency tasks...
Thanks,
Bernd