[PATCH 01/12] ARM: clean up idle handlers

[Russell King - ARM Linux]

On Tue, Oct 25, 2011 at 09:56:20AM +0200, Tony Lindgren wrote:
> * Nicolas Pitre <nico at fluxnic.net> [111024 12:16]:
> > Let's factor out the need_resched() check instead of having it duplicated
> > in every pm_idle implementations to avoid inconsistencies (omap2_pm_idle
> > was missing it already).
> > 
> > The forceful re-enablement of IRQs after pm_idle has returned can go.
> > The warning certainly doesn't trigger for existing users.  Similar for
> > the redundant  local_irq_disable() call in the OMAP implementations.
> 
> Looks good to me.
>  
> > And finally move the comment explaining the reason for the turning off
> > of IRQs to a more proper location.
> 
> One minor comment on that comment below.
>  
> > --- a/arch/arm/kernel/process.c
> > +++ b/arch/arm/kernel/process.c
> ...
> 
> > -				 * This will eventually be removed - pm_idle
> > -				 * functions should always return with IRQs
> > -				 * enabled.
> > +				 * pm_idle functions should always
> > +				 * return with IRQs enabled.
> >  				 */
> >  				WARN_ON(irqs_disabled());
> > +			} else
> >  				local_irq_enable();
> > -			}
> 
> Any reason to use "should always return" instead of "must always return
> with IRQs enabled"?

I think it's now a 'must' rather than a 'should'.  I'm not aware of any
cases recently where IRQs were disabled at this point.