[PATCH 03/12] powerpc: Rework runlatch code

[Benjamin Herrenschmidt]

On Thu, 2012-04-05 at 23:38 +0200, Andreas Schwab wrote:
> Benjamin Herrenschmidt <benh at kernel.crashing.org> writes:
> 
> > This moves the inlines into system.h and changes the runlatch
> > code to use the thread local flags (non-atomic) rather than
> > the TIF flags (atomic) to keep track of the latch state.
> >
> > The code to turn it back on in an asynchronous interrupt is
> > now simplified and partially inlined.
> 
> This breaks the X server in that it no longer receives SIGIO signals.

Interesting. I have not observed that on my machine here, however I have
seen on my dual G5 (and not on the quad) some occasional loss of
interrupts for seconds which might be related.

Have you verified that you can "fix" it with reverting just the above
patch ? I would have more easily blamed the rest of my series than this
specific one... Anyways, I'll have a closer look but it might have to
wait for tuesday.

Cheers,
Ben.