Re: 2.6.19-rc1-mm1

["Michal Piotrowski" <michal.k.k.piotrowski@gmail.com>]

On 11/10/06, Neil Brown <neilb@suse.de> wrote:
> On Tuesday October 10, michal.k.k.piotrowski@gmail.com wrote:
> > On 10/10/06, Michal Piotrowski <michal.k.k.piotrowski@gmail.com> wrote:
> > > Hi,
> > >
> > > On 10/10/06, Andrew Morton <akpm@osdl.org> wrote:
> > > >
> > > > ftp://ftp.kernel.org/pub/linux/kernel/people/akpm/patches/2.6/2.6.19-rc1/2.6.19-rc1-mm1/
> > > >
> > >
> > > Kernel 2.6.19-rc1-mm1 + Neil's avoid_lockdep_warning_in_md.patch
> > > (http://www.ussg.iu.edu/hypermail/linux/kernel/0610.1/0642.html)
> > >
> > > (I'll try to reproduce this without Neil's patch).
> >
> > I can't reproduce this without Neil's patch.
> >
>
> Despite this circumstantial evidence, I don't see how my patch could
> possible have an effect here....
>
> Looking at the code, starting at _cpu_down in the CONFIG_HOTPLUG_CPU
> case, the call notifier chain 'cpu_chain' contains
> workqueue_cpu_callback which does 'mutex_lock(&workqueue_mutex)' in
> the "DOWN_PREPARE" case and mutex_unlock(&workqueue_mutex) in the
> DOWN_FAILED and DEAD cases.
>
> blocking_notifier_call_chain is
>         down_read(&nh->rwsem);
>         ret = notifier_call_chain(&nh->head, val, v);
>         up_read(&nh->rwsem);
>
> and so holds ->rwsem while calling the callback.
> So the locking sequence ends up as:
>
>  down_read(&cpu_chain.rwsem);
>  mutex_lock(&workqueue_mutex);
>  up_read(&cpu_chain.rwsem);
>
>  down_read(&cpu_chain.rwsem);
>  mutex_unlock(&workqueue_mutex);
>  up_read(&workqueue_mutex);
>
> and lockdep doesn't seem to like this.  It sees workqueue_mutex
> claimed while cpu_chain.rwsem is held. and then it sees
> cpu_chain.rwsem claimed while workqueue_mutex is held, which looks a
> bit like a class ABBA deadlock.
> Of course because it is a 'down_read' rather than a 'down', it isn't
> really a dead lock.
>
> I don't know how to tell lockdep to do the right thing, but I'll leave
> that up to Ingo et al.
>
> Why it didn't trigger without my patch I cannot imagine.  Are you sure
> the config was identical (you didn't remove CONFIG_HOTPLUG_CPU or
> anything did you?).

No, I didn't remove CONFIG_HOTPLUG_CPU or anything else.

I didn't do enough testing - only a few hibernatins.

>
> NeilBrown
>

Regards,
Michal

-- 
Michal K. K. Piotrowski
LTG - Linux Testers Group
(http://www.stardust.webpages.pl/ltg/)