Re: [linux-pm] Fundamental flaw in system suspend, exposed by freezer removal

[Alan Stern <stern@rowland.harvard.edu>]

On Wed, 27 Feb 2008, Rafael J. Wysocki wrote:

> > All right, we can set it to RESUME_RUNNING before calling the resume
> > method and then set it to 0 afterwards.  The point is that the value
> > shouldn't remain SUSPEND_DONE while resume runs, because it should be
> > legal for resume to register new children.
> 
> I'm not sure.  The core moves the device to dpm_active only after ->resume()
> has run.

Actually the move is done before the method is called.  So this isn't a
problem.

...
> > > > The one tricky thing to watch out for is when a suspend or resume 
> > > > method wants to unregister the device being suspended or resumed.
> > > 
> > > That can't happen, because dev->sem is taken by suspend_device() and
> > > device_del() would lock up attempting to acquire it once again.
> > 
> > We'll have to fix device_del() to prevent that from happening.  Your 
> > in_sleep_context() approach should work.
> 
> I'm not sure if we need to do it.  It's always been like this, so the current
> drivers' ->suspend() and ->resume() don't unregister the device they're called
> for.  I don't see any advantage from doing that for future drivers.

All right, if it doesn't happen now then we don't need to allow for it.  
That makes life a little simpler.

Alan Stern