[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <CA+G9fYs3LT9OaKVZw9C3v3YxGPBk6ZWCdqur=hL-XoLM5rgtww@mail.gmail.com>
Date: Fri, 26 May 2023 11:09:39 +0530
From: Naresh Kamboju <naresh.kamboju@...aro.org>
To: Joel Fernandes <joel@...lfernandes.org>
Cc: open list <linux-kernel@...r.kernel.org>,
linux-stable <stable@...r.kernel.org>,
linux-perf-users@...r.kernel.org, rcu <rcu@...r.kernel.org>,
lkft-triage@...ts.linaro.org,
Peter Zijlstra <peterz@...radead.org>,
"Paul E. McKenney" <paulmck@...nel.org>,
Arnaldo Carvalho de Melo <acme@...nel.org>,
Ingo Molnar <mingo@...hat.com>,
Vincent Guittot <vincent.guittot@...aro.org>,
Juri Lelli <juri.lelli@...hat.com>,
Dan Carpenter <dan.carpenter@...aro.org>,
Arnaldo Carvalho de Melo <acme@...hat.com>
Subject: Re: Perf: RIP: 0010:__schedule on qemu-x86_64
HI Joel,
On Thu, 25 May 2023 at 20:51, Joel Fernandes <joel@...lfernandes.org> wrote:
>
> Hello Naresh,
>
> On Wed, May 24, 2023 at 5:02 AM Naresh Kamboju
> <naresh.kamboju@...aro.org> wrote:
> >
> > Following kernel crash noticed while running perf testing on qemu-x86_64
> > with stable-rc 6.3.4-rc2 + kselftest merge configs.
> >
> > Reported-by: Linux Kernel Functional Testing <lkft@...aro.org>
> >
> > logs:
> > =====
> > INFO: Performing perf record test...
> >
> > Parse sched tracepoints fields: Ok
> > 16: syscalls:sys_enter_openat event fields :
> > --- start ---
> > test child forked, pid 255
> > <4>[ 152.221288] int3: 0000 [#1] PREEMPT SMP PTI
> > <4>[ 152.221564] CPU: 0 PID: 255 Comm: perf Not tainted 6.3.4-rc2 #1
> > <4>[ 152.221607] Hardware name: QEMU Standard PC (Q35 + ICH9, 2009),
> > BIOS 1.14.0-2 04/01/2014
> > <4>[ 152.221733] RIP: 0010:__schedule+0x29e/0xaa0
<trim>
> I wonder if this is a deadlock related to runqueue lock since another
> CPU call stack are similarly in schedule.
>
> I tried to run gdb on your vmlinux but it does not have symbols, would
> it be possible for you to provide a copy of vmlinux with symbols?
DEBUG_INFO not enabled on this build.
>
> Were you able to narrow down the commit/release using git bisect to
> see when it started happening?
I will bisect this problem and get back to you.
>
> Thank you,
>
> - Joel
- Naresh
Powered by blists - more mailing lists