Re: [PATCH 1/2] MMC: fix all hangs related to mmc/sd card insert/removal during suspend/resume.

[Rafael J. Wysocki]

On Wednesday, June 23, 2010, Andrew Morton wrote:
> On Tue, 22 Jun 2010 23:53:21 +0200
> "Rafael J. Wysocki" <rjw@xxxxxxx> wrote:
> 
> > On Tuesday, June 22, 2010, Maxim Levitsky wrote:
> > > On Tue, 2010-06-22 at 11:19 +0200, Rafael J. Wysocki wrote: 
> > > > On Tuesday, June 22, 2010, Maxim Levitsky wrote:
> > > > > On Mon, 2010-06-21 at 22:26 +0200, Rafael J. Wysocki wrote: 
> > ...
> > > > > I *do* use freezable  work-queue.
> > > > 
> > > > I overlooked that, sorry.
> > > > 
> > > > > However since this is pm notifier, it is called before userspace and the
> > > > > workqueue is frozen.
> > > > > Therefore I would like manually to freeze the workqueue from the pm
> > > > > notifier.
> > > > 
> > > > No, that won't work.  You need to find an alternative solution.  I guess you
> > > > may insert a work item that's going to sleep until a condition is
> > > > satisfied (analogous to a workqueue barrier) and wait for it to 
> > > This screams to be done in generic way.
> > > Something like suspend_workqueue() and resume_workqueue();
> > 
> > Well, there was no need for that until now. :-)
> > 
> > > In addition to that I just found that .suspend function sometimes can
> > > return -ENOSYS, which triggers card removal. I wrongly remove that chunk
> > > of code.
> > > 
> > > To make the thing picture perfect I would have to invest more time to
> > > it, I will do so as soon as I finish my exams.
> > > 
> > > Meanwhile the current patch already fixes all but corner cases or rather
> > > nasty hang on suspend with any MMC/SD card inserted.
> > 
> > OK
> > 
> > I think Andrew has already taken [2/2].
> 
> I took them both, but I need to come back to this discussion to work
> out what to do with them now.

I think they are worth merging.  At least they shouldn't break things and
Maxim has already promised to clean up that code in future.

> > Andrew, who's maintaining MMC now?
> 
> Pierre stopped doing it, so I'm now pretending to.
> 
> I actually pretend to maintain a huge number of subsystems and should
> sprinkle akpms all over MAINTAINERS so stuff doesn't get lost.

I see.

If you have anything PM-related, I can handle that too.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/