Re: [PATCH -next V5] blk-mq: fix tag_get wait task can't be awakened

[Guenter Roeck]

On 1/27/22 09:28, Jens Axboe wrote:
> On 1/26/22 6:32 PM, Guenter Roeck wrote:
> > Hi,
> >
> > On Thu, Jan 13, 2022 at 10:55:36AM +0800, Laibin Qiu wrote:
> > > In case of shared tags, there might be more than one hctx which
> > > allocates from the same tags, and each hctx is limited to allocate at
> > > most:
> > >          hctx_max_depth = max((bt->sb.depth + users - 1) / users, 4U);
> > >
> > > tag idle detection is lazy, and may be delayed for 30sec, so there
> > > could be just one real active hctx(queue) but all others are actually
> > > idle and still accounted as active because of the lazy idle detection.
> > > Then if wake_batch is > hctx_max_depth, driver tag allocation may wait
> > > forever on this real active hctx.
> > >
> > > Fix this by recalculating wake_batch when inc or dec active_queues.
> > >
> > > Fixes: 0d2602ca30e41 ("blk-mq: improve support for shared tags maps")
> > > Suggested-by: Ming Lei <ming.lei@xxxxxxxxxx>
> > > Suggested-by: John Garry <john.garry@xxxxxxxxxx>
> > > Signed-off-by: Laibin Qiu <qiulaibin@xxxxxxxxxx>
> >
> > I understand this problem has been reported already, but still:
> >
> > This patch causes a hang in several of my qemu emulations when
> > trying to boot from usb. Reverting it fixes the problem. Bisect log
> > is attached.
> >
> > Boot logs are available at
> > https://kerneltests.org/builders/qemu-arm-aspeed-master/builds/230/steps/qemubuildcommand/logs/stdio
> > but don't really show much: the affected tests simply hang until they
> > are aborted.
>
> This one got reported a few days ago, can you check if applying:
>
> https://git.kernel.dk/cgit/linux-block/commit/?h=block-5.17&id=10825410b956dc1ed8c5fbc8bbedaffdadde7f20
>
> fixes it for you?
Yes, it does.

Thanks,
Guenter