Re: workqueue lockup debug

[Dave Chinner <david@fromorbit.com>]

On Fri, Nov 08, 2024 at 09:54:19AM +0100, Peter Zijlstra wrote:
> On Fri, Nov 08, 2024 at 09:57:38AM +1100, Dave Chinner wrote:
> > On Thu, Nov 07, 2024 at 01:39:39PM +0100, Thorsten Leemhuis wrote:
> > > On 24.10.24 17:49, John Garry wrote:
> > > > Hi workqueue and scheduler maintainers,
> > > > 
> > > > As reported in https://lore.kernel.org/linux-fsdevel/df9db1ce-17d9-49f1-
> > > > ab6d-7ed9a4f1f9c0@oracle.com/T/
> > > > #m506b9edb1340cdddd87c6d14d20222ca8d7e8796, I am experiencing a
> > > > workqueue lockup for v6.12-rcX.
> > > 
> > > John, what this resolved in between? This and the other thread[1] look
> > > stalled, but I might be missing something. Asking, because I have this
> > > on my list of tracked regressions and wonder if this is something that
> > > better should be solved one way or another before 6.12.
> > > 
> > > [1]
> > > https://lore.kernel.org/lkml/63d6ceeb-a22f-4dee-bc9d-8687ce4c7355@oracle.com/
> > 
> > I'm still seeing the scheduler bug in -rc6.
> 
> But that WARN you reported earlier isn't there anymore. So what exactly
> are you seeing now?

I just went back and checked the logs. It does appear I was running
an -rc5 base kernel, but that just makes me really confused because
I was running mount path debug trace_printk patches at the time and
that kernel was clearly built on a 6.12-rc6 base.

Yet, apparently, despite getting the tracing output I expected to
see, the kernel still reported as 6.12.0-rc5-dgc+. I suspect some
kind of weird delta rebuild issue (possibly some of my machines have
dropped time sync), but that's my problem to sort out.

IOWs, ignore this, I'll do a full rebuild on -rc7 and test again.
If you don't hear from me again, I haven't hit new problems.

-Dave.
-- 
Dave Chinner
david@fromorbit.com