linux-kernel - Re: [PATCH v1] workqueue: annotate data-races around pwq->stats

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <a89746d8897c81ed6e55c99d8ba71223ef3e318c@linux.dev>
Date: Wed, 23 Apr 2025 15:48:14 +0000
From: "Jiayuan Chen" <jiayuan.chen@...ux.dev>
To: "Tejun Heo" <tj@...nel.org>
Cc: linux-kernel@...r.kernel.org, mrpre@....com, mkoutny@...e.com,
 syzbot+01affb1491750534256d@...kaller.appspotmail.com, "Lai Jiangshan"
 <jiangshanlai@...il.com>
Subject: Re: [PATCH v1] workqueue: annotate data-races around pwq->stats

April 23, 2025 at 23:30, "Tejun Heo" <tj@...nel.org> wrote:



> 
> On Wed, Apr 23, 2025 at 08:53:41PM +0800, Jiayuan Chen wrote:
> 
> > 
> > Suppress warning by annotating these accesses using
> > 
> >  READ_ONCE() / WRITE_ONCE().
> > 
> >  
> > 
> >  Reported-by: syzbot+01affb1491750534256d@...kaller.appspotmail.com
> > 
> >  Signed-off-by: Jiayuan Chen <jiayuan.chen@...ux.dev>
> > 
> >  ---
> > 
> >  kernel/workqueue.c | 2 +-
> > 
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> > 
> >  
> > 
> >  diff --git a/kernel/workqueue.c b/kernel/workqueue.c
> > 
> >  index cf6203282737..d78640b5d188 100644
> > 
> >  --- a/kernel/workqueue.c
> > 
> >  +++ b/kernel/workqueue.c
> > 
> >  @@ -3241,7 +3241,7 @@ __acquires(&pool->lock)
> > 
> >  * point will only record its address.
> > 
> >  */
> > 
> >  trace_workqueue_execute_end(work, worker->current_func);
> > 
> >  - pwq->stats[PWQ_STAT_COMPLETED]++;
> > 
> >  + WRITE_ONCE(pwq->stats[PWQ_STAT_COMPLETED], READ_ONCE(pwq->stats[PWQ_STAT_COMPLETED]) + 1);
> > 
> 
> The function acquires pool->lock down below. Can you move it down inside the
> 
> locked region instead of adding READ/WRITE_ONCE()?
> 
> Thanks.
> 
> -- 
> 
> tejun
>
Thanks, I can do that.
Previously, I thought it was for performance considerations that the
metrics calculation is out of the lock's scope...