Message-ID: <CAGyP=7cr1S462e+ZNQY_s3ygmHbYpRo6OLrx7RCLKX8h6F=OnQ@mail.gmail.com>
Date:   Sat, 15 May 2021 13:14:27 +0530
From:   Palash Oswal <oswalpalash@...il.com>
To:     Pavel Begunkov <asml.silence@...il.com>
Cc:     Jens Axboe <axboe@...nel.dk>, io-uring@...r.kernel.org,
        LKML <linux-kernel@...r.kernel.org>,
        syzbot+11bf59db879676f59e52@...kaller.appspotmail.com
Subject: Re: INFO: task hung in io_uring_cancel_sqpoll

On Fri, May 14, 2021 at 3:01 AM Pavel Begunkov <asml.silence@...il.com> wrote:
>
> On 5/13/21 10:28 PM, Pavel Begunkov wrote:
> > On 5/10/21 5:47 AM, Palash Oswal wrote:
> >> On Mon, May 3, 2021 at 4:15 PM Pavel Begunkov <asml.silence@...il.com>
> >> wrote:
> >>
> >>> On 5/3/21 5:41 AM, Palash Oswal wrote:
> >>>> On Mon, May 3, 2021 at 3:42 AM Pavel Begunkov <asml.silence@...il.com>
> >>> wrote:
> >>>>> The test might be very useful. Would you send a patch to
> >>>>> liburing? Or mind the repro being taken as a test?
> >>>>
> >>>> Pavel,
> >>>>
> >>>> I'm working on a PR for liburing with this test. Do you know how I can
> >>>
> >>> Perfect, thanks
> >>>
> >>>> address this behavior?
> >>>
> >>> As the hang is sqpoll cancellations, it's most probably
> >>> fixed in 5.13 + again awaits to be taken for stable.
> >>>
> >>> Don't know about segfaults, but it was so for long, and
> >>> syz reproducers are ofthen faults for me, and exit with 0
> >>> in the end. So, I wouldn't worry about it.
> >>>
> >>>
> >> Hey Pavel,
> >> The bug actually fails to reproduce on 5.12 when the fork() call is made by
> >> the runtests.sh script. This causes the program to end correctly, and the
> >> hang does not occur. I verified this on 5.12 where the bug isn't patched.
> >> Just running the `sqpoll-cancel-hang` triggers the bug; whereas running it
> >> after being forked from runtests.sh does not trigger the bug.
> >
> > I see. fyi, it's always good to wait for 5 minutes, because some useful
> > logs are not generated immediately but do timeout based hang detection.
> >
> > I'd think that may be due CLONE_IO and how to whom it binds workers,
> > but can you try first:
> >
> > 1) timeout -s INT -k $TIMEOUT $TIMEOUT ./sqpoll-cancel-hang
>
> edit:
>
> timeout -s INT -k 60 60 ./sqpoll-cancel-hang
>
> And privileged, root/sudo
>
> >
> > 2) remove timeout from <liburing>/tests/Makefile and run
> > "./runtests.sh sqpoll-cancel-hang" again looking for faults?
> >
> > diff --git a/test/runtests.sh b/test/runtests.sh
> > index e8f4ae5..2b51dca 100755
> > --- a/test/runtests.sh
> > +++ b/test/runtests.sh
> > @@ -91,7 +91,8 @@ run_test()
> >       fi
> >
> >       # Run the test
> > -     timeout -s INT -k $TIMEOUT $TIMEOUT ./$test_name $dev
> > +     # timeout -s INT -k $TIMEOUT $TIMEOUT ./$test_name $dev
> > +     ./$test_name $dev
> >       local status=$?
> >
> >       # Check test status

root@...kaller:~/liburing/test# timeout -s INT -k 60 60 ./sqpoll-cancel-hang
[   19.381358] sqpoll-cancel-h[300]: segfault at 0 ip 0000556f7fa325e3
sp 00007ffee497d980 error 6 in s]
[   19.387323] Code: 89 d8 8d 34 90 8b 45 04 ba 03 00 00 00 c1 e0 04
03 45 64 39 c6 48 0f 42 f0 45 31 c6
root@...kaller:~/liburing/test# [  243.511620] INFO: task
iou-sqp-300:301 blocked for more than 120 sec.
[  243.514146]       Not tainted 5.12.0 #142
[  243.515301] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs"
disables this message.
[  243.517629] task:iou-sqp-300     state:D stack:    0 pid:  301
ppid:     1 flags:0x00004004
[  243.520155] Call Trace:
[  243.520956]  __schedule+0xb1d/0x1130
[  243.522102]  ? __sched_text_start+0x8/0x8
[  243.523195]  ? io_wq_worker_sleeping+0x145/0x500
[  243.524588]  schedule+0x131/0x1c0
[  243.525892]  io_uring_cancel_sqpoll+0x288/0x350
[  243.527610]  ? io_sq_thread_unpark+0xd0/0xd0
[  243.529084]  ? mutex_lock+0xbb/0x130
[  243.530327]  ? init_wait_entry+0xe0/0xe0
[  243.532805]  ? wait_for_completion_killable_timeout+0x20/0x20
[  243.535411]  io_sq_thread+0x174c/0x18c0
[  243.536520]  ? io_rsrc_put_work+0x380/0x380
[  243.537904]  ? init_wait_entry+0xe0/0xe0
[  243.538935]  ? _raw_spin_lock_irq+0xa5/0x180
[  243.540203]  ? _raw_spin_lock_irqsave+0x190/0x190
[  243.542398]  ? calculate_sigpending+0x6b/0xa0
[  243.543868]  ? io_rsrc_put_work+0x380/0x380
[  243.545377]  ret_from_fork+0x22/0x30
^C
root@...kaller:~/liburing/test# ps
  PID TTY          TIME CMD
  269 ttyS0    00:00:00 login
  294 ttyS0    00:00:00 bash
  300 ttyS0    00:00:00 sqpoll-cancel-h
  305 ttyS0    00:00:00 ps


After reboot, and the runtests.sh diff applied ( to remove timeout )
root@...kaller:~/liburing/test# ./runtests.sh sqpoll-cancel-hang
Running test sqp[   45.332140] Running test sqpoll-cancel-hang:
oll-cancel-hang:
[   45.352524] sqpoll-cancel-h[314]: segfault at 0 ip 000056025bd085e3
sp 00007fffb08e20b0 error 6 in s]
[   45.356601] Code: 89 d8 8d 34 90 8b 45 04 ba 03 00 00 00 c1 e0 04
03 45 64 39 c6 48 0f 42 f0 45 31 c6
[  243.019384] INFO: task iou-sqp-314:315 blocked for more than 120 seconds.
[  243.021483]       Not tainted 5.12.0 #142
[  243.022633] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs"
disables this message.
[  243.024651] task:iou-sqp-314     state:D stack:    0 pid:  315
ppid:   313 flags:0x00004004
[  243.026822] Call Trace:
[  243.027478]  __schedule+0xb1d/0x1130
[  243.028382]  ? __sched_text_start+0x8/0x8
[  243.029536]  ? io_wq_worker_sleeping+0x145/0x500
[  243.030932]  schedule+0x131/0x1c0
[  243.031920]  io_uring_cancel_sqpoll+0x288/0x350
[  243.033393]  ? io_sq_thread_unpark+0xd0/0xd0
[  243.034713]  ? mutex_lock+0xbb/0x130
[  243.035775]  ? init_wait_entry+0xe0/0xe0
[  243.037036]  ? wait_for_completion_killable_timeout+0x20/0x20
[  243.039492]  io_sq_thread+0x174c/0x18c0
[  243.040894]  ? io_rsrc_put_work+0x380/0x380
[  243.042463]  ? init_wait_entry+0xe0/0xe0
[  243.043990]  ? _raw_spin_lock_irq+0xa5/0x180
[  243.045581]  ? _raw_spin_lock_irqsave+0x190/0x190
[  243.047545]  ? calculate_sigpending+0x6b/0xa0
[  243.049262]  ? io_rsrc_put_work+0x380/0x380
[  243.050861]  ret_from_fork+0x22/0x30
^C
root@...kaller:~/liburing/test# ps
  PID TTY          TIME CMD
  285 ttyS0    00:00:00 login
  300 ttyS0    00:00:00 bash
  314 ttyS0    00:00:00 sqpoll-cancel-h
  318 ttyS0    00:00:00 ps


runtests.sh without any changes
root@...kaller:~/liburing/test# ./runtests.sh sqpoll-cancel-hang
[   49.634886] Running test sqpoll-cancel-hang:
Running test sqpoll-cancel-hang:
[   49.658365] sqpoll-cancel-h[302]: segfault at 0 ip 000055a76e99c5e3
sp 00007ffdc255d1a0 error 6 in s]
[   49.661703] Code: 89 d8 8d 34 90 8b 45 04 ba 03 00 00 00 c1 e0 04
03 45 64 39 c6 48 0f 42 f0 45 31 c6
Test sqpoll-cancel-hang timed out (may not be a failure)
All tests passed
root@...kaller:~/liburing/test# ps
  PID TTY          TIME CMD
  269 ttyS0    00:00:00 login
  287 ttyS0    00:00:00 bash
  302 ttyS0    00:00:00 sqpoll-cancel-h
  309 ttyS0    00:00:00 ps
root@...kaller:~/liburing/test# [  243.324831] INFO: task
iou-sqp-302:303 blocked for more than 120 sec.
[  243.328320]       Not tainted 5.12.0 #142
[  243.330361] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs"
disables this message.
[  243.333930] task:iou-sqp-302     state:D stack:    0 pid:  303
ppid:     1 flags:0x00004004
[  243.337758] Call Trace:
[  243.338926]  __schedule+0xb1d/0x1130
[  243.340801]  ? __sched_text_start+0x8/0x8
[  243.342690]  ? io_wq_worker_sleeping+0x145/0x500
[  243.344903]  schedule+0x131/0x1c0
[  243.346626]  io_uring_cancel_sqpoll+0x288/0x350
[  243.348762]  ? io_sq_thread_unpark+0xd0/0xd0
[  243.351036]  ? mutex_lock+0xbb/0x130
[  243.352737]  ? init_wait_entry+0xe0/0xe0
[  243.354673]  ? wait_for_completion_killable_timeout+0x20/0x20
[  243.356989]  io_sq_thread+0x174c/0x18c0
[  243.358559]  ? io_rsrc_put_work+0x380/0x380
[  243.359981]  ? init_wait_entry+0xe0/0xe0
[  243.361185]  ? _raw_spin_lock_irq+0x110/0x180
[  243.362958]  ? _raw_spin_lock_irqsave+0x190/0x190
[  243.364260]  ? calculate_sigpending+0x6b/0xa0
[  243.365763]  ? io_rsrc_put_work+0x380/0x380
[  243.367041]  ret_from_fork+0x22/0x30




> >
> >
> >>
> >> The segfaults are benign, but notice the "All tests passed" in the previous
> >> mail. It should not have passed, as the run was on 5.12. Therefore I wanted
> >> to ask your input on how to address this odd behaviour, where the
> >> involvement of runtests.sh actually mitigated the bug.
> >>
> >>
> >>
> >>>> root@...kaller:~/liburing/test# ./runtests.sh sqpoll-cancel-hang
> >>>> Running test sqp[   15.310997] Running test sqpoll-cancel-hang:
> >>>> oll-cancel-hang:
> >>>> [   15.333348] sqpoll-cancel-h[305]: segfault at 0 ip 000055ad00e265e3
> >>> sp]
> >>>> [   15.334940] Code: 89 d8 8d 34 90 8b 45 04 ba 03 00 00 00 c1 e0 04 03
> >>> 46
> >>>> All tests passed
> >>>>
> >>>> root@...kaller:~/liburing/test# ./sqpoll-cancel-hang
> >>>> [   13.572639] sqpoll-cancel-h[298]: segfault at 0 ip 00005634c4a455e3
> >>> sp]
> >>>> [   13.576506] Code: 89 d8 8d 34 90 8b 45 04 ba 03 00 00 00 c1 e0 04 03
> >>> 46
> >>>> [   23.350459] random: crng init done
> >>>> [   23.352837] random: 7 urandom warning(s) missed due to ratelimiting
> >>>> [  243.090865] INFO: task iou-sqp-298:299 blocked for more than 120
> >>> secon.
> >>>> [  243.095187]       Not tainted 5.12.0 #142
> >>>> [  243.099800] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs"
> >>> disable.
> >>>> [  243.105928] task:iou-sqp-298     state:D stack:    0 pid:  299 ppid:
> >>> 4
> >>>> [  243.111044] Call Trace:
> >>>> [  243.112855]  __schedule+0xb1d/0x1130
> >>>> [  243.115549]  ? __sched_text_start+0x8/0x8
> >>>> [  243.118328]  ? io_wq_worker_sleeping+0x145/0x500
> >>>> [  243.121790]  schedule+0x131/0x1c0
> >>>> [  243.123698]  io_uring_cancel_sqpoll+0x288/0x350
> >>>> [  243.125977]  ? io_sq_thread_unpark+0xd0/0xd0
> >>>> [  243.128966]  ? mutex_lock+0xbb/0x130
> >>>> [  243.132572]  ? init_wait_entry+0xe0/0xe0
> >>>> [  243.135429]  ? wait_for_completion_killable_timeout+0x20/0x20
> >>>> [  243.138303]  io_sq_thread+0x174c/0x18c0
> >>>> [  243.140162]  ? io_rsrc_put_work+0x380/0x380
> >>>> [  243.141613]  ? init_wait_entry+0xe0/0xe0
> >>>> [  243.143686]  ? _raw_spin_lock_irq+0xa5/0x180
> >>>> [  243.145619]  ? _raw_spin_lock_irqsave+0x190/0x190
> >>>> [  243.147671]  ? calculate_sigpending+0x6b/0xa0
> >>>> [  243.149036]  ? io_rsrc_put_work+0x380/0x380
> >>>> [  243.150694]  ret_from_fork+0x22/0x30
> >>>>
> >>>> Palash
> >>>>
> >>>
> >>> --
> >>> Pavel Begunkov
> >>>
> >>
> >
>
> --
> Pavel Begunkov

Powered by blists - more mailing lists