From: Keith Busch <kbusch@kernel.org>
To: Sagi Grimberg <sagi@grimberg.me>
Cc: Christoph Hellwig <hch@lst.de>, Jens Axboe <axboe@kernel.dk>,
Jeffle Xu <jefflexu@linux.alibaba.com>,
Ming Lei <ming.lei@redhat.com>,
Damien Le Moal <Damien.LeMoal@wdc.com>,
"Wunderlich, Mark" <mark.wunderlich@intel.com>,
"Vasudevan, Anil" <anil.vasudevan@intel.com>,
linux-block@vger.kernel.org, linux-fsdevel@vger.kernel.org,
linux-nvme@lists.infradead.org
Subject: Re: [PATCH 12/15] block: switch polling to be bio based
Date: Fri, 14 May 2021 09:26:12 -0700 [thread overview]
Message-ID: <20210514162612.GA2706199@dhcp-10-100-145-180.wdc.com> (raw)
In-Reply-To: <20210514025026.GA2447336@dhcp-10-100-145-180.wdc.com>
On Thu, May 13, 2021 at 07:50:26PM -0700, Keith Busch wrote:
> On Wed, May 12, 2021 at 03:12:37PM -0700, Keith Busch wrote:
> > On Wed, May 12, 2021 at 03:03:40PM -0700, Sagi Grimberg wrote:
> > > On 5/12/21 6:15 AM, Christoph Hellwig wrote:
> > > > + *
> > > > + * Note: the caller must either be the context that submitted @bio, or
> > > > + * be in a RCU critical section to prevent freeing of @bio.
> > > > + */
> > > > +int bio_poll(struct bio *bio, unsigned int flags)
> > > > +{
> > > > + struct request_queue *q = bio->bi_bdev->bd_disk->queue;
> > > > + blk_qc_t cookie = READ_ONCE(bio->bi_cookie);
> > > > +
> > > > + if (cookie == BLK_QC_T_NONE ||
> > > > + !test_bit(QUEUE_FLAG_POLL, &q->queue_flags))
> > > > + return 0;
> > > > +
> > > > + if (current->plug)
> > > > + blk_flush_plug_list(current->plug, false);
> > > > +
> > > > + /* not yet implemented, so this should not happen */
> > > > + if (WARN_ON_ONCE(!queue_is_mq(q)))
> > >
> > > What happens if the I/O wasn't (yet) queued to the bottom device
> > > (i.e. no available path in nvme-mpath)?
> > >
> > > In this case the disk would be the mpath device node (which is
> > > not mq...)
> >
> > The bi_cookie should remain BLK_QC_T_NONE in that case, so we wouldn't
> > get to the warning. But if that does happen, it doesn't appear that
> > anyone is going to wake up thread that needs to poll for this bio's
> > completion when a path becomes available for dispatch. I think it would
> > make sense for nvme-mpath to just clear the REQ_POLLED flag if it
> > doesn't immediately have viable path.
>
> It looks like there is a way to hit this warning. If you induce a path
> error to create a failover, the nvme-mpath will reset the bio's bdev to
> the mpath one, which is not MQ. I can occaisionally hit it with a fault
> injection test.
Oh, and it gets a bit worse. Using pvsync2 will block forever once this
warning is hit, or if nvme-mpath had to requeue the IO for lack of a
path. io_uring doesn't have that hang problem though.
Christoph, I think patch 15 needs something like this:
---
diff --git a/drivers/nvme/host/multipath.c b/drivers/nvme/host/multipath.c
index 7febdb57f690..d40a9331daf7 100644
--- a/drivers/nvme/host/multipath.c
+++ b/drivers/nvme/host/multipath.c
@@ -419,6 +419,11 @@ static void nvme_requeue_work(struct work_struct *work)
* path.
*/
bio_set_dev(bio, head->disk->part0);
+
+ if (bio->bi_opf & REQ_POLLED) {
+ bio->bi_opf &= ~REQ_POLLED;
+ bio->bi_cookie = BLK_QC_T_NONE;
+ }
submit_bio_noacct(bio);
}
}
--
This should fix the hang since requeued bio's will use an interrupt
driven queue, but it doesn't fix the warning. The recent commit
"nvme-multipath: reset bdev to ns head when failover" looks like it
makes preventing the polling thread from using the non-MQ head disk
not possible.
next prev parent reply other threads:[~2021-05-14 16:26 UTC|newest]
Thread overview: 45+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-05-12 13:15 switch block layer polling to a bio based model v3 Christoph Hellwig
2021-05-12 13:15 ` [PATCH 01/15] direct-io: remove blk_poll support Christoph Hellwig
2021-05-12 13:15 ` [PATCH 02/15] block: don't try to poll multi-bio I/Os in __blkdev_direct_IO Christoph Hellwig
2021-05-12 20:54 ` Sagi Grimberg
2021-05-16 18:01 ` Kanchan Joshi
2021-05-12 13:15 ` [PATCH 03/15] iomap: don't try to poll multi-bio I/Os in __iomap_dio_rw Christoph Hellwig
2021-05-12 21:32 ` Darrick J. Wong
2021-05-12 13:15 ` [PATCH 04/15] blk-mq: factor out a blk_qc_to_hctx helper Christoph Hellwig
2021-05-12 20:58 ` Sagi Grimberg
2021-05-12 13:15 ` [PATCH 05/15] blk-mq: factor out a "classic" poll helper Christoph Hellwig
2021-05-12 20:59 ` Sagi Grimberg
2021-05-12 13:15 ` [PATCH 06/15] blk-mq: remove blk_qc_t_to_tag and blk_qc_t_is_internal Christoph Hellwig
2021-05-12 21:42 ` Sagi Grimberg
2021-05-12 13:15 ` [PATCH 07/15] blk-mq: remove blk_qc_t_valid Christoph Hellwig
2021-05-12 21:43 ` Sagi Grimberg
2021-05-12 13:15 ` [PATCH 08/15] io_uring: don't sleep when polling for I/O Christoph Hellwig
2021-05-12 21:55 ` Sagi Grimberg
2021-05-19 13:02 ` Christoph Hellwig
2021-05-12 13:15 ` [PATCH 09/15] block: rename REQ_HIPRI to REQ_POLLED Christoph Hellwig
2021-05-12 21:56 ` Sagi Grimberg
2021-05-12 22:43 ` Chaitanya Kulkarni
2021-05-12 13:15 ` [PATCH 10/15] block: use SLAB_TYPESAFE_BY_RCU for the bio slab Christoph Hellwig
2021-05-12 13:15 ` [PATCH 11/15] block: define 'struct bvec_iter' as packed Christoph Hellwig
2021-05-12 21:58 ` Sagi Grimberg
2021-05-12 13:15 ` [PATCH 12/15] block: switch polling to be bio based Christoph Hellwig
2021-05-12 22:03 ` Sagi Grimberg
2021-05-12 22:12 ` Keith Busch
2021-05-14 2:50 ` Keith Busch
2021-05-14 16:26 ` Keith Busch [this message]
2021-05-14 16:28 ` Christoph Hellwig
2021-05-13 1:26 ` Ming Lei
2021-05-13 1:23 ` Ming Lei
2021-05-12 13:15 ` [PATCH 13/15] block: don't allow writing to the poll queue attribute Christoph Hellwig
2021-05-12 22:04 ` Sagi Grimberg
2021-05-12 13:15 ` [PATCH 14/15] nvme-multipath: set QUEUE_FLAG_NOWAIT Christoph Hellwig
2021-05-12 13:15 ` [PATCH 15/15] nvme-multipath: enable polled I/O Christoph Hellwig
2021-05-12 22:10 ` Sagi Grimberg
2021-05-12 22:16 ` Keith Busch
2021-05-13 23:33 ` switch block layer polling to a bio based model v3 Wunderlich, Mark
-- strict thread matches above, loose matches on Subject: below --
2021-04-27 16:16 switch block layer polling to a bio based model v2 Christoph Hellwig
2021-04-27 16:16 ` [PATCH 12/15] block: switch polling to be bio based Christoph Hellwig
2021-04-28 2:28 ` Ming Lei
2021-04-29 7:20 ` Christoph Hellwig
2021-04-29 7:36 ` Ming Lei
2021-04-29 9:50 ` Christoph Hellwig
2021-04-29 14:42 ` Ming Lei
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20210514162612.GA2706199@dhcp-10-100-145-180.wdc.com \
--to=kbusch@kernel.org \
--cc=Damien.LeMoal@wdc.com \
--cc=anil.vasudevan@intel.com \
--cc=axboe@kernel.dk \
--cc=hch@lst.de \
--cc=jefflexu@linux.alibaba.com \
--cc=linux-block@vger.kernel.org \
--cc=linux-fsdevel@vger.kernel.org \
--cc=linux-nvme@lists.infradead.org \
--cc=mark.wunderlich@intel.com \
--cc=ming.lei@redhat.com \
--cc=sagi@grimberg.me \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).