blk-mq: fix a hung issue when fsync
diff mbox series

Message ID 1548838916-25051-1-git-send-email-jianchao.w.wang@oracle.com
State New
Headers show
Series
  • blk-mq: fix a hung issue when fsync
Related show

Commit Message

jianchao.wang Jan. 30, 2019, 9:01 a.m. UTC
Florian reported a io hung issue when fsync(). It should be
triggered by following race condition.

data + post flush         a flush

blk_flush_complete_seq
  case REQ_FSEQ_DATA
    blk_flush_queue_rq
    issued to driver      blk_mq_dispatch_rq_list
                            try to issue a flush req
                            failed due to NON-NCQ command
                            .queue_rq return BLK_STS_DEV_RESOURCE

request completion
  req->end_io // doesn't check RESTART
  mq_flush_data_end_io
    case REQ_FSEQ_POSTFLUSH
      blk_kick_flush
        do nothing because previous flush
        has not been completed
     blk_mq_run_hw_queue
                              insert rq to hctx->dispatch
                              due to RESTART is still set, do nothing

To fix this, replace the blk_mq_run_hw_queue in mq_flush_data_end_io
with blk_mq_sched_restart to check and clear the RESTART flag.

Fixes: bd166ef1 (blk-mq-sched: add framework for MQ capable IO schedulers)
Reported-by: Florian Stecker <m19@florianstecker.de>
Tested-by: Florian Stecker <m19@florianstecker.de>
Signed-off-by: Jianchao Wang <jianchao.w.wang@oracle.com>
---
 block/blk-flush.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

Comments

Jens Axboe Jan. 30, 2019, 3:54 p.m. UTC | #1
On 1/30/19 2:01 AM, Jianchao Wang wrote:
> Florian reported a io hung issue when fsync(). It should be
> triggered by following race condition.
> 
> data + post flush         a flush
> 
> blk_flush_complete_seq
>   case REQ_FSEQ_DATA
>     blk_flush_queue_rq
>     issued to driver      blk_mq_dispatch_rq_list
>                             try to issue a flush req
>                             failed due to NON-NCQ command
>                             .queue_rq return BLK_STS_DEV_RESOURCE
> 
> request completion
>   req->end_io // doesn't check RESTART
>   mq_flush_data_end_io
>     case REQ_FSEQ_POSTFLUSH
>       blk_kick_flush
>         do nothing because previous flush
>         has not been completed
>      blk_mq_run_hw_queue
>                               insert rq to hctx->dispatch
>                               due to RESTART is still set, do nothing
> 
> To fix this, replace the blk_mq_run_hw_queue in mq_flush_data_end_io
> with blk_mq_sched_restart to check and clear the RESTART flag.

Applied, thanks.
Thibaut Sautereau Feb. 17, 2019, 3:37 p.m. UTC | #2
On Wed, Jan 30, 2019 at 08:54:09AM -0700, Jens Axboe wrote:
> On 1/30/19 2:01 AM, Jianchao Wang wrote:
> > Florian reported a io hung issue when fsync(). It should be
> > triggered by following race condition.
> > 
> > data + post flush         a flush
> > 
> > blk_flush_complete_seq
> >   case REQ_FSEQ_DATA
> >     blk_flush_queue_rq
> >     issued to driver      blk_mq_dispatch_rq_list
> >                             try to issue a flush req
> >                             failed due to NON-NCQ command
> >                             .queue_rq return BLK_STS_DEV_RESOURCE
> > 
> > request completion
> >   req->end_io // doesn't check RESTART
> >   mq_flush_data_end_io
> >     case REQ_FSEQ_POSTFLUSH
> >       blk_kick_flush
> >         do nothing because previous flush
> >         has not been completed
> >      blk_mq_run_hw_queue
> >                               insert rq to hctx->dispatch
> >                               due to RESTART is still set, do nothing
> > 
> > To fix this, replace the blk_mq_run_hw_queue in mq_flush_data_end_io
> > with blk_mq_sched_restart to check and clear the RESTART flag.
> 
> Applied, thanks.
> 
> -- 
> Jens Axboe

Can this be applied to stable kernels please?

It's commit 85bd6e61f34dffa8ec2dc75ff3c02ee7b2f1cbce upstream.

Thanks,
Sasha Levin Feb. 17, 2019, 5:28 p.m. UTC | #3
On Sun, Feb 17, 2019 at 04:37:29PM +0100, Thibaut Sautereau wrote:
>On Wed, Jan 30, 2019 at 08:54:09AM -0700, Jens Axboe wrote:
>> On 1/30/19 2:01 AM, Jianchao Wang wrote:
>> > Florian reported a io hung issue when fsync(). It should be
>> > triggered by following race condition.
>> >
>> > data + post flush         a flush
>> >
>> > blk_flush_complete_seq
>> >   case REQ_FSEQ_DATA
>> >     blk_flush_queue_rq
>> >     issued to driver      blk_mq_dispatch_rq_list
>> >                             try to issue a flush req
>> >                             failed due to NON-NCQ command
>> >                             .queue_rq return BLK_STS_DEV_RESOURCE
>> >
>> > request completion
>> >   req->end_io // doesn't check RESTART
>> >   mq_flush_data_end_io
>> >     case REQ_FSEQ_POSTFLUSH
>> >       blk_kick_flush
>> >         do nothing because previous flush
>> >         has not been completed
>> >      blk_mq_run_hw_queue
>> >                               insert rq to hctx->dispatch
>> >                               due to RESTART is still set, do nothing
>> >
>> > To fix this, replace the blk_mq_run_hw_queue in mq_flush_data_end_io
>> > with blk_mq_sched_restart to check and clear the RESTART flag.
>>
>> Applied, thanks.
>>
>> --
>> Jens Axboe
>
>Can this be applied to stable kernels please?
>
>It's commit 85bd6e61f34dffa8ec2dc75ff3c02ee7b2f1cbce upstream.

I've queued it for 4.20, 4.19 and 4.14.

--
Thanks,
Sasha

Patch
diff mbox series

diff --git a/block/blk-flush.c b/block/blk-flush.c
index a3fc7191..6e0f2d9 100644
--- a/block/blk-flush.c
+++ b/block/blk-flush.c
@@ -335,7 +335,7 @@  static void mq_flush_data_end_io(struct request *rq, blk_status_t error)
 	blk_flush_complete_seq(rq, fq, REQ_FSEQ_DATA, error);
 	spin_unlock_irqrestore(&fq->mq_flush_lock, flags);
 
-	blk_mq_run_hw_queue(hctx, true);
+	blk_mq_sched_restart(hctx);
 }
 
 /**