Re: BUG: sleeping function called from invalid context at kernel/mutex.c

["Rafael J. Wysocki" <rjw@sisk.pl>]

On Wednesday, April 06, 2011, Peter Zijlstra wrote:
> On Wed, 2011-04-06 at 06:43 +0200, Rafael J. Wysocki wrote:
> > > the $subject text sound like it triggered might_sleep(), and that had a
> > > system_state != SYSTEM_RUNNING bail condition, but then, I've no clue
> > > what resume looks like.
> > 
> > Early resume looks pretty much like the system startup, e.g. everything
> > called from syscore_ops should not be sleepable (although mutexes shouldn't
> > trigger, because that code is effectively single-threaded, unless somebody
> > holds the mutex in question when that code is being executed, but that would
> > deadlock anyway). 
> 
> Right, so system_state != SYSTEM_RUNNING should be true for resume?

It is not, although it probably should be.

First, some time ago there was opposition to adding more different possible
values of system_state and it wasn't clear which of the existing values should
be used instead of SYSTEM_RUNNING during suspend/resume.  We ended up sticking
to SYSTEM_RUNNING for that reason long enough for some code to develop the
expectation of system_state == SYSTEM_RUNNING during suspend/resume (IOW,
changing that right now would probably break stuff).

Second, even if we decide to switch from SYSTEM_RUNNING to something else
during suspend (and back during resume), it's not entirely clear what's the
right place to do so.

Thanks,
Rafael