[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <ZsYTZTEbtMuVyp1o@slm.duckdns.org>
Date: Wed, 21 Aug 2024 06:18:45 -1000
From: Tejun Heo <tj@...nel.org>
To: "Paul E. McKenney" <paulmck@...nel.org>
Cc: matthew.brost@...el.com, jiangshanlai@...il.com, sfr@...b.auug.org.au,
linux-kernel@...r.kernel.org, linux-next@...r.kernel.org
Subject: Re: [BUG bisected] sysfs: cannot create duplicate filename
'/devices/virtual/workqueue/scsi_tmf_1073822488'
On Tue, Aug 20, 2024 at 07:16:00PM -0700, Paul E. McKenney wrote:
> Running rcutorture scenario TREE01 on next-20240820 on a dual-socket x86
> system (and for that matter on my laptop) got me this warning during boot:
>
> sysfs: cannot create duplicate filename '/devices/virtual/workqueue/scsi_tmf_1073822488'
>
> The number at the end varies from run to run.
>
> Bisection landed here:
>
> b188c57af2b5 ("workqueue: Split alloc_workqueue into internal function and lockdep init")
>
> My reproducer is to run this at the top of the Linux-kernel file tree:
>
> tools/testing/selftests/rcutorture/bin/kvm.sh --allcpus --duration 1m --configs "2*TREE01" --trust-make
>
> Reverting this commit on top of next-20240820 gets me clean runs.
>
> I cannot see how the patch would make this happen, but then again,
> I don't know this code at all well.
>
> Thoughts?
Hmmm... I have a hard time imagining that change causing that error. My bet
is unreliable reproducer. I'll try the repro.
Thanks.
--
tejun
Powered by blists - more mailing lists