Re: 2.6.12-rc6-mm1

[Richard Purdie <rpurdie@rpsys.net>]

On Sun, 2005-06-19 at 10:11 +0100, Russell King wrote:
> On Sun, Jun 19, 2005 at 10:02:26AM +0100, Russell King wrote:
> > On Sun, Jun 19, 2005 at 02:20:48AM +0100, Richard Purdie wrote:
> > > On Sun, 2005-06-19 at 00:18 +0100, Russell King wrote: 
> > > > Thinking about what's probably happening, I suspect all the ARM suspend
> > > > and resume code needs to be reworked to save more state.  I'll try to
> > > > cook up a patch tomorrow to fix it, but I'll need you to provide
> > > > feedback.
> > > 
> > > Ok, thanks. I'm happy to test any fixes/patches.
> > 
> > This should resolve the problem - we now rely on the stack pointer for
> > each CPU mode to remain constant throughout the running time of the
> > kernel, which includes across suspend/resume cycles.
> 
> Actually, this patch is probably an all-round better solution.

This patch (the simpler of the two using cpu_init()) allows the pxa to
suspend/resume happily with the git-arm-smp.patch applied.

Richard