From: Ming Lei <ming.lei@redhat.com>
To: Alan Adamson <alan.adamson@oracle.com>
Cc: axboe@kernel.dk, linux-block@vger.kernel.org,
martin.petersen@oracle.com, himanshu.madhani@oracle.com
Subject: Re: [PATCH 0/1] block: don't inject fake timeouts on quiesced queues
Date: Sat, 9 May 2020 06:05:26 +0800 [thread overview]
Message-ID: <20200508220526.GC1389136@T590> (raw)
In-Reply-To: <1588974394-15430-1-git-send-email-alan.adamson@oracle.com>
On Fri, May 08, 2020 at 02:46:34PM -0700, Alan Adamson wrote:
> While using the block fake timeout injector to reproduce a nvme error handling hang, a hang was
> observed when the following script was run:
>
> echo 100 > /sys/kernel/debug/fail_io_timeout/probability
> echo 1000 > /sys/kernel/debug/fail_io_timeout/times
> echo 1 > /sys/block/nvme0n1/io-timeout-fail
> dd if=/dev/nvme0n1 of=/dev/null bs=512 count=1
>
>
> dmesg:
>
> [ 370.018164] INFO: task kworker/u113:9:1191 blocked for more than 122 seconds.
> [ 370.018849] Not tainted 5.7.0-rc4 #1
> [ 370.019251] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
> [ 370.019653] kworker/u113:9 D 0 1191 2 0x80004000
> [ 370.019660] Workqueue: nvme-reset-wq nvme_reset_work [nvme]
> [ 370.019661] Call Trace:
> [ 370.019667] __schedule+0x2dc/0x710
> [ 370.019668] schedule+0x44/0xb0
> [ 370.019671] blk_mq_freeze_queue_wait+0x4b/0xb0
> [ 370.019675] ? finish_wait+0x80/0x80
> [ 370.019681] nvme_wait_freeze+0x36/0x50 [nvme_core]
> [ 370.019683] nvme_reset_work+0xb65/0xf2b [nvme]
> [ 370.019688] process_one_work+0x1ab/0x380
> [ 370.019689] worker_thread+0x37/0x3b0
> [ 370.019691] kthread+0x120/0x140
> [ 370.019692] ? create_worker+0x1b0/0x1b0
> [ 370.019693] ? kthread_park+0x90/0x90
> [ 370.019696] ret_from_fork+0x35/0x40
>
> This occurs when a fake timeout is scheduled on a request that is in he process
> of being cancelled due to the previous fake timeout.
Not sure root cause is timeout injection, and in theory request queue shouldn't
have been quiesced before freezing, otherwise it is easy to cause deadlock.
BTW, could you share us how you conclude it is related with queue quiesce?
Thanks,
Ming
next prev parent reply other threads:[~2020-05-08 22:05 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-05-08 21:46 [PATCH 0/1] block: don't inject fake timeouts on quiesced queues Alan Adamson
2020-05-08 22:05 ` Ming Lei [this message]
2020-05-08 22:23 ` Alan Adamson
2020-05-09 2:28 ` 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=20200508220526.GC1389136@T590 \
--to=ming.lei@redhat.com \
--cc=alan.adamson@oracle.com \
--cc=axboe@kernel.dk \
--cc=himanshu.madhani@oracle.com \
--cc=linux-block@vger.kernel.org \
--cc=martin.petersen@oracle.com \
/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).