Re: Failure to release lock after CPU hot-unplug canceled

[Heiko Carstens <heiko.carstens@de.ibm.com>]

On Tue, Jan 09, 2007 at 05:57:40PM +0530, Srivatsa Vaddagiri wrote:
> On Tue, Jan 09, 2007 at 01:17:38PM +0100, Heiko Carstens wrote:
> > missing in kernel cpu.c in _cpu_down() in case CPU_DOWN_PREPARE
> > returned with NOTIFY_BAD. However... this reveals that there is just a
> > more fundamental problem.
> >
> > The workqueue code grabs a lock on CPU_[UP|DOWN]_PREPARE and releases it
> > again on CPU_DOWN_FAILED/CPU_UP_CANCELED. If something in the callchain
> > returns NOTIFY_BAD the rest of the entries in the callchain won't be
> > called anymore. But DOWN_FAILED/UP_CANCELED will be called for every
> > entry.
> > So we might even end up with a mutex_unlock(&workqueue_mutex) even if
> > mutex_lock(&workqueue_mutex) hasn't been called...
> 
> This is a known problem. Gautham had sent out patches to address them
> 
> http://lkml.org/lkml/2006/11/14/93
> 
> Looks like they are in latest mm tree. Perhaps the testcase should be
> retried against latest mm.

Ah, nice! Wasn't aware of that. But I still think we should have a
CPU_DOWN_FAILED in case CPU_DOWN_PREPARED failed.
Also the slab cache code hasn't been changed to make use of the of the
new CPU_LOCK_[ACQUIRE|RELEASE] stuff. I'm going to send patches in reply
to this mail.